diff --git a/thirdparty/README.md b/thirdparty/README.md
index 84e56db86ed..186f183ea26 100644
--- a/thirdparty/README.md
+++ b/thirdparty/README.md
@@ -19,7 +19,7 @@ comments.
 ## bullet
 
 - Upstream: https://github.com/bulletphysics/bullet3
-- Version: git (12409f1118a7c7a266f9071350c70789dfe73bb9, Commits on Sep 6, 2018 )
+- Version: git (126b676, 2018-12-31)
 - License: zlib
 
 Files extracted from upstream source:
diff --git a/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3BroadphaseCallback.h b/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3BroadphaseCallback.h
index 1bc56cf80a5..bec0800a6ff 100644
--- a/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3BroadphaseCallback.h
+++ b/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3BroadphaseCallback.h
@@ -19,22 +19,20 @@ subject to the following restrictions:
 #include "Bullet3Common/b3Vector3.h"
 struct b3BroadphaseProxy;
 
-
-struct	b3BroadphaseAabbCallback
+struct b3BroadphaseAabbCallback
 {
 	virtual ~b3BroadphaseAabbCallback() {}
-	virtual bool	process(const b3BroadphaseProxy* proxy) = 0;
+	virtual bool process(const b3BroadphaseProxy* proxy) = 0;
 };
 
-
-struct	b3BroadphaseRayCallback : public b3BroadphaseAabbCallback
+struct b3BroadphaseRayCallback : public b3BroadphaseAabbCallback
 {
 	///added some cached data to accelerate ray-AABB tests
-	b3Vector3		m_rayDirectionInverse;
-	unsigned int	m_signs[3];
-	b3Scalar		m_lambda_max;
+	b3Vector3 m_rayDirectionInverse;
+	unsigned int m_signs[3];
+	b3Scalar m_lambda_max;
 
 	virtual ~b3BroadphaseRayCallback() {}
 };
 
-#endif //B3_BROADPHASE_CALLBACK_H
+#endif  //B3_BROADPHASE_CALLBACK_H
diff --git a/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3DynamicBvh.cpp b/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3DynamicBvh.cpp
index 0f04efe3311..a0dc1da95d8 100644
--- a/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3DynamicBvh.cpp
+++ b/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3DynamicBvh.cpp
@@ -17,209 +17,224 @@ subject to the following restrictions:
 #include "b3DynamicBvh.h"
 
 //
-typedef b3AlignedObjectArray<b3DbvtNode*>			b3NodeArray;
-typedef b3AlignedObjectArray<const b3DbvtNode*>	b3ConstNodeArray;
+typedef b3AlignedObjectArray<b3DbvtNode*> b3NodeArray;
+typedef b3AlignedObjectArray<const b3DbvtNode*> b3ConstNodeArray;
 
 //
 struct b3DbvtNodeEnumerator : b3DynamicBvh::ICollide
 {
-	b3ConstNodeArray	nodes;
+	b3ConstNodeArray nodes;
 	void Process(const b3DbvtNode* n) { nodes.push_back(n); }
 };
 
 //
-static B3_DBVT_INLINE int			b3IndexOf(const b3DbvtNode* node)
+static B3_DBVT_INLINE int b3IndexOf(const b3DbvtNode* node)
 {
-	return(node->parent->childs[1]==node);
+	return (node->parent->childs[1] == node);
 }
 
 //
-static B3_DBVT_INLINE b3DbvtVolume	b3Merge(	const b3DbvtVolume& a,
-									  const b3DbvtVolume& b)
+static B3_DBVT_INLINE b3DbvtVolume b3Merge(const b3DbvtVolume& a,
+										   const b3DbvtVolume& b)
 {
-#if (B3_DBVT_MERGE_IMPL==B3_DBVT_IMPL_SSE)
+#if (B3_DBVT_MERGE_IMPL == B3_DBVT_IMPL_SSE)
 	B3_ATTRIBUTE_ALIGNED16(char locals[sizeof(b3DbvtAabbMm)]);
-	b3DbvtVolume&	res=*(b3DbvtVolume*)locals;
+	b3DbvtVolume& res = *(b3DbvtVolume*)locals;
 #else
-		b3DbvtVolume	res;
+	b3DbvtVolume res;
 #endif
-	b3Merge(a,b,res);
-	return(res);
+	b3Merge(a, b, res);
+	return (res);
 }
 
 // volume+edge lengths
-static B3_DBVT_INLINE b3Scalar		b3Size(const b3DbvtVolume& a)
+static B3_DBVT_INLINE b3Scalar b3Size(const b3DbvtVolume& a)
 {
-	const b3Vector3	edges=a.Lengths();
-	return(	edges.x*edges.y*edges.z+
-		edges.x+edges.y+edges.z);
+	const b3Vector3 edges = a.Lengths();
+	return (edges.x * edges.y * edges.z +
+			edges.x + edges.y + edges.z);
 }
 
 //
-static void						b3GetMaxDepth(const b3DbvtNode* node,int depth,int& maxdepth)
+static void b3GetMaxDepth(const b3DbvtNode* node, int depth, int& maxdepth)
 {
-	if(node->isinternal())
+	if (node->isinternal())
 	{
-		b3GetMaxDepth(node->childs[0],depth+1,maxdepth);
-		b3GetMaxDepth(node->childs[1],depth+1,maxdepth);
-	} else maxdepth=b3Max(maxdepth,depth);
+		b3GetMaxDepth(node->childs[0], depth + 1, maxdepth);
+		b3GetMaxDepth(node->childs[1], depth + 1, maxdepth);
+	}
+	else
+		maxdepth = b3Max(maxdepth, depth);
 }
 
 //
-static B3_DBVT_INLINE void			b3DeleteNode(	b3DynamicBvh* pdbvt,
-										   b3DbvtNode* node)
+static B3_DBVT_INLINE void b3DeleteNode(b3DynamicBvh* pdbvt,
+										b3DbvtNode* node)
 {
 	b3AlignedFree(pdbvt->m_free);
-	pdbvt->m_free=node;
+	pdbvt->m_free = node;
 }
 
 //
-static void						b3RecurseDeleteNode(	b3DynamicBvh* pdbvt,
-												  b3DbvtNode* node)
+static void b3RecurseDeleteNode(b3DynamicBvh* pdbvt,
+								b3DbvtNode* node)
 {
-	if(!node->isleaf())
+	if (!node->isleaf())
 	{
-		b3RecurseDeleteNode(pdbvt,node->childs[0]);
-		b3RecurseDeleteNode(pdbvt,node->childs[1]);
+		b3RecurseDeleteNode(pdbvt, node->childs[0]);
+		b3RecurseDeleteNode(pdbvt, node->childs[1]);
 	}
-	if(node==pdbvt->m_root) pdbvt->m_root=0;
-	b3DeleteNode(pdbvt,node);
+	if (node == pdbvt->m_root) pdbvt->m_root = 0;
+	b3DeleteNode(pdbvt, node);
 }
 
 //
-static B3_DBVT_INLINE b3DbvtNode*	b3CreateNode(	b3DynamicBvh* pdbvt,
-										   b3DbvtNode* parent,
-										   void* data)
+static B3_DBVT_INLINE b3DbvtNode* b3CreateNode(b3DynamicBvh* pdbvt,
+											   b3DbvtNode* parent,
+											   void* data)
 {
-	b3DbvtNode*	node;
-	if(pdbvt->m_free)
-	{ node=pdbvt->m_free;pdbvt->m_free=0; }
-	else
-	{ node=new(b3AlignedAlloc(sizeof(b3DbvtNode),16)) b3DbvtNode(); }
-	node->parent	=	parent;
-	node->data		=	data;
-	node->childs[1]	=	0;
-	return(node);
-}
-
-//
-static B3_DBVT_INLINE b3DbvtNode*	b3CreateNode(	b3DynamicBvh* pdbvt,
-										   b3DbvtNode* parent,
-										   const b3DbvtVolume& volume,
-										   void* data)
-{
-	b3DbvtNode*	node=b3CreateNode(pdbvt,parent,data);
-	node->volume=volume;
-	return(node);
-}
-
-//
-static B3_DBVT_INLINE b3DbvtNode*	b3CreateNode(	b3DynamicBvh* pdbvt,
-										   b3DbvtNode* parent,
-										   const b3DbvtVolume& volume0,
-										   const b3DbvtVolume& volume1,
-										   void* data)
-{
-	b3DbvtNode*	node=b3CreateNode(pdbvt,parent,data);
-	b3Merge(volume0,volume1,node->volume);
-	return(node);
-}
-
-//
-static void						b3InsertLeaf(	b3DynamicBvh* pdbvt,
-										   b3DbvtNode* root,
-										   b3DbvtNode* leaf)
-{
-	if(!pdbvt->m_root)
+	b3DbvtNode* node;
+	if (pdbvt->m_free)
 	{
-		pdbvt->m_root	=	leaf;
-		leaf->parent	=	0;
+		node = pdbvt->m_free;
+		pdbvt->m_free = 0;
 	}
 	else
 	{
-		if(!root->isleaf())
+		node = new (b3AlignedAlloc(sizeof(b3DbvtNode), 16)) b3DbvtNode();
+	}
+	node->parent = parent;
+	node->data = data;
+	node->childs[1] = 0;
+	return (node);
+}
+
+//
+static B3_DBVT_INLINE b3DbvtNode* b3CreateNode(b3DynamicBvh* pdbvt,
+											   b3DbvtNode* parent,
+											   const b3DbvtVolume& volume,
+											   void* data)
+{
+	b3DbvtNode* node = b3CreateNode(pdbvt, parent, data);
+	node->volume = volume;
+	return (node);
+}
+
+//
+static B3_DBVT_INLINE b3DbvtNode* b3CreateNode(b3DynamicBvh* pdbvt,
+											   b3DbvtNode* parent,
+											   const b3DbvtVolume& volume0,
+											   const b3DbvtVolume& volume1,
+											   void* data)
+{
+	b3DbvtNode* node = b3CreateNode(pdbvt, parent, data);
+	b3Merge(volume0, volume1, node->volume);
+	return (node);
+}
+
+//
+static void b3InsertLeaf(b3DynamicBvh* pdbvt,
+						 b3DbvtNode* root,
+						 b3DbvtNode* leaf)
+{
+	if (!pdbvt->m_root)
+	{
+		pdbvt->m_root = leaf;
+		leaf->parent = 0;
+	}
+	else
+	{
+		if (!root->isleaf())
 		{
-			do	{
-				root=root->childs[b3Select(	leaf->volume,
-					root->childs[0]->volume,
-					root->childs[1]->volume)];
-			} while(!root->isleaf());
+			do
+			{
+				root = root->childs[b3Select(leaf->volume,
+											 root->childs[0]->volume,
+											 root->childs[1]->volume)];
+			} while (!root->isleaf());
 		}
-		b3DbvtNode*	prev=root->parent;
-		b3DbvtNode*	node=b3CreateNode(pdbvt,prev,leaf->volume,root->volume,0);
-		if(prev)
+		b3DbvtNode* prev = root->parent;
+		b3DbvtNode* node = b3CreateNode(pdbvt, prev, leaf->volume, root->volume, 0);
+		if (prev)
 		{
-			prev->childs[b3IndexOf(root)]	=	node;
-			node->childs[0]				=	root;root->parent=node;
-			node->childs[1]				=	leaf;leaf->parent=node;
-			do	{
-				if(!prev->volume.Contain(node->volume))
-					b3Merge(prev->childs[0]->volume,prev->childs[1]->volume,prev->volume);
+			prev->childs[b3IndexOf(root)] = node;
+			node->childs[0] = root;
+			root->parent = node;
+			node->childs[1] = leaf;
+			leaf->parent = node;
+			do
+			{
+				if (!prev->volume.Contain(node->volume))
+					b3Merge(prev->childs[0]->volume, prev->childs[1]->volume, prev->volume);
 				else
 					break;
-				node=prev;
-			} while(0!=(prev=node->parent));
+				node = prev;
+			} while (0 != (prev = node->parent));
 		}
 		else
 		{
-			node->childs[0]	=	root;root->parent=node;
-			node->childs[1]	=	leaf;leaf->parent=node;
-			pdbvt->m_root	=	node;
+			node->childs[0] = root;
+			root->parent = node;
+			node->childs[1] = leaf;
+			leaf->parent = node;
+			pdbvt->m_root = node;
 		}
 	}
 }
 
 //
-static b3DbvtNode*				b3RemoveLeaf(	b3DynamicBvh* pdbvt,
-										   b3DbvtNode* leaf)
+static b3DbvtNode* b3RemoveLeaf(b3DynamicBvh* pdbvt,
+								b3DbvtNode* leaf)
 {
-	if(leaf==pdbvt->m_root)
+	if (leaf == pdbvt->m_root)
 	{
-		pdbvt->m_root=0;
-		return(0);
+		pdbvt->m_root = 0;
+		return (0);
 	}
 	else
 	{
-		b3DbvtNode*	parent=leaf->parent;
-		b3DbvtNode*	prev=parent->parent;
-		b3DbvtNode*	sibling=parent->childs[1-b3IndexOf(leaf)];			
-		if(prev)
+		b3DbvtNode* parent = leaf->parent;
+		b3DbvtNode* prev = parent->parent;
+		b3DbvtNode* sibling = parent->childs[1 - b3IndexOf(leaf)];
+		if (prev)
 		{
-			prev->childs[b3IndexOf(parent)]=sibling;
-			sibling->parent=prev;
-			b3DeleteNode(pdbvt,parent);
-			while(prev)
+			prev->childs[b3IndexOf(parent)] = sibling;
+			sibling->parent = prev;
+			b3DeleteNode(pdbvt, parent);
+			while (prev)
 			{
-				const b3DbvtVolume	pb=prev->volume;
-				b3Merge(prev->childs[0]->volume,prev->childs[1]->volume,prev->volume);
-				if(b3NotEqual(pb,prev->volume))
+				const b3DbvtVolume pb = prev->volume;
+				b3Merge(prev->childs[0]->volume, prev->childs[1]->volume, prev->volume);
+				if (b3NotEqual(pb, prev->volume))
 				{
-					prev=prev->parent;
-				} else break;
+					prev = prev->parent;
+				}
+				else
+					break;
 			}
-			return(prev?prev:pdbvt->m_root);
+			return (prev ? prev : pdbvt->m_root);
 		}
 		else
-		{								
-			pdbvt->m_root=sibling;
-			sibling->parent=0;
-			b3DeleteNode(pdbvt,parent);
-			return(pdbvt->m_root);
-		}			
+		{
+			pdbvt->m_root = sibling;
+			sibling->parent = 0;
+			b3DeleteNode(pdbvt, parent);
+			return (pdbvt->m_root);
+		}
 	}
 }
 
 //
-static void						b3FetchLeaves(b3DynamicBvh* pdbvt,
-											b3DbvtNode* root,
-											b3NodeArray& leaves,
-											int depth=-1)
+static void b3FetchLeaves(b3DynamicBvh* pdbvt,
+						  b3DbvtNode* root,
+						  b3NodeArray& leaves,
+						  int depth = -1)
 {
-	if(root->isinternal()&&depth)
+	if (root->isinternal() && depth)
 	{
-		b3FetchLeaves(pdbvt,root->childs[0],leaves,depth-1);
-		b3FetchLeaves(pdbvt,root->childs[1],leaves,depth-1);
-		b3DeleteNode(pdbvt,root);
+		b3FetchLeaves(pdbvt, root->childs[0], leaves, depth - 1);
+		b3FetchLeaves(pdbvt, root->childs[1], leaves, depth - 1);
+		b3DeleteNode(pdbvt, root);
 	}
 	else
 	{
@@ -227,50 +242,50 @@ static void						b3FetchLeaves(b3DynamicBvh* pdbvt,
 	}
 }
 
-static bool						b3LeftOfAxis(	const b3DbvtNode* node,
-										   const b3Vector3& org,
-										   const b3Vector3& axis)
+static bool b3LeftOfAxis(const b3DbvtNode* node,
+						 const b3Vector3& org,
+						 const b3Vector3& axis)
 {
-	return b3Dot(axis,node->volume.Center()-org) <= 0;
+	return b3Dot(axis, node->volume.Center() - org) <= 0;
 }
 
 // Partitions leaves such that leaves[0, n) are on the
 // left of axis, and leaves[n, count) are on the right
 // of axis. returns N.
-static int						b3Split(	b3DbvtNode** leaves,
-									  int count,
-									  const b3Vector3& org,
-									  const b3Vector3& axis)
+static int b3Split(b3DbvtNode** leaves,
+				   int count,
+				   const b3Vector3& org,
+				   const b3Vector3& axis)
 {
-	int begin=0;
-	int end=count;
-	for(;;)
+	int begin = 0;
+	int end = count;
+	for (;;)
 	{
-		while(begin!=end && b3LeftOfAxis(leaves[begin],org,axis))
+		while (begin != end && b3LeftOfAxis(leaves[begin], org, axis))
 		{
 			++begin;
 		}
 
-		if(begin==end)
+		if (begin == end)
 		{
 			break;
 		}
 
-		while(begin!=end && !b3LeftOfAxis(leaves[end-1],org,axis))
+		while (begin != end && !b3LeftOfAxis(leaves[end - 1], org, axis))
 		{
 			--end;
 		}
 
-		if(begin==end)
+		if (begin == end)
 		{
 			break;
 		}
 
 		// swap out of place nodes
 		--end;
-		b3DbvtNode* temp=leaves[begin];
-		leaves[begin]=leaves[end];
-		leaves[end]=temp;
+		b3DbvtNode* temp = leaves[begin];
+		leaves[begin] = leaves[end];
+		leaves[end] = temp;
 		++begin;
 	}
 
@@ -278,149 +293,152 @@ static int						b3Split(	b3DbvtNode** leaves,
 }
 
 //
-static b3DbvtVolume				b3Bounds(	b3DbvtNode** leaves,
-										 int count)
+static b3DbvtVolume b3Bounds(b3DbvtNode** leaves,
+							 int count)
 {
-#if B3_DBVT_MERGE_IMPL==B3_DBVT_IMPL_SSE
-	B3_ATTRIBUTE_ALIGNED16(char	locals[sizeof(b3DbvtVolume)]);
-	b3DbvtVolume&	volume=*(b3DbvtVolume*)locals;
-	volume=leaves[0]->volume;
+#if B3_DBVT_MERGE_IMPL == B3_DBVT_IMPL_SSE
+	B3_ATTRIBUTE_ALIGNED16(char locals[sizeof(b3DbvtVolume)]);
+	b3DbvtVolume& volume = *(b3DbvtVolume*)locals;
+	volume = leaves[0]->volume;
 #else
-	b3DbvtVolume volume=leaves[0]->volume;
+	b3DbvtVolume volume = leaves[0]->volume;
 #endif
-	for(int i=1,ni=count;i<ni;++i)
+	for (int i = 1, ni = count; i < ni; ++i)
 	{
-		b3Merge(volume,leaves[i]->volume,volume);
+		b3Merge(volume, leaves[i]->volume, volume);
 	}
-	return(volume);
+	return (volume);
 }
 
 //
-static void						b3BottomUp(	b3DynamicBvh* pdbvt,
-										 b3DbvtNode** leaves,
-										 int count)
+static void b3BottomUp(b3DynamicBvh* pdbvt,
+					   b3DbvtNode** leaves,
+					   int count)
 {
-	while(count>1)
+	while (count > 1)
 	{
-		b3Scalar	minsize=B3_INFINITY;
-		int			minidx[2]={-1,-1};
-		for(int i=0;i<count;++i)
+		b3Scalar minsize = B3_INFINITY;
+		int minidx[2] = {-1, -1};
+		for (int i = 0; i < count; ++i)
 		{
-			for(int j=i+1;j<count;++j)
+			for (int j = i + 1; j < count; ++j)
 			{
-				const b3Scalar	sz=b3Size(b3Merge(leaves[i]->volume,leaves[j]->volume));
-				if(sz<minsize)
+				const b3Scalar sz = b3Size(b3Merge(leaves[i]->volume, leaves[j]->volume));
+				if (sz < minsize)
 				{
-					minsize		=	sz;
-					minidx[0]	=	i;
-					minidx[1]	=	j;
+					minsize = sz;
+					minidx[0] = i;
+					minidx[1] = j;
 				}
 			}
 		}
-		b3DbvtNode*	n[]	=	{leaves[minidx[0]],leaves[minidx[1]]};
-		b3DbvtNode*	p	=	b3CreateNode(pdbvt,0,n[0]->volume,n[1]->volume,0);
-		p->childs[0]		=	n[0];
-		p->childs[1]		=	n[1];
-		n[0]->parent		=	p;
-		n[1]->parent		=	p;
-		leaves[minidx[0]]	=	p;
-		leaves[minidx[1]]	=	leaves[count-1];
+		b3DbvtNode* n[] = {leaves[minidx[0]], leaves[minidx[1]]};
+		b3DbvtNode* p = b3CreateNode(pdbvt, 0, n[0]->volume, n[1]->volume, 0);
+		p->childs[0] = n[0];
+		p->childs[1] = n[1];
+		n[0]->parent = p;
+		n[1]->parent = p;
+		leaves[minidx[0]] = p;
+		leaves[minidx[1]] = leaves[count - 1];
 		--count;
 	}
 }
 
 //
-static b3DbvtNode*			b3TopDown(b3DynamicBvh* pdbvt,
-									b3DbvtNode** leaves,
-									int count,
-									int bu_treshold)
+static b3DbvtNode* b3TopDown(b3DynamicBvh* pdbvt,
+							 b3DbvtNode** leaves,
+							 int count,
+							 int bu_treshold)
 {
-	static const b3Vector3	axis[]={b3MakeVector3(1,0,0),
-		b3MakeVector3(0,1,0),
-		b3MakeVector3(0,0,1)};
-	b3Assert(bu_treshold>1);
-	if(count>1)
+	static const b3Vector3 axis[] = {b3MakeVector3(1, 0, 0),
+									 b3MakeVector3(0, 1, 0),
+									 b3MakeVector3(0, 0, 1)};
+	b3Assert(bu_treshold > 1);
+	if (count > 1)
 	{
-		if(count>bu_treshold)
+		if (count > bu_treshold)
 		{
-			const b3DbvtVolume	vol=b3Bounds(leaves,count);
-			const b3Vector3			org=vol.Center();
-			int						partition;
-			int						bestaxis=-1;
-			int						bestmidp=count;
-			int						splitcount[3][2]={{0,0},{0,0},{0,0}};
+			const b3DbvtVolume vol = b3Bounds(leaves, count);
+			const b3Vector3 org = vol.Center();
+			int partition;
+			int bestaxis = -1;
+			int bestmidp = count;
+			int splitcount[3][2] = {{0, 0}, {0, 0}, {0, 0}};
 			int i;
-			for( i=0;i<count;++i)
+			for (i = 0; i < count; ++i)
 			{
-				const b3Vector3	x=leaves[i]->volume.Center()-org;
-				for(int j=0;j<3;++j)
+				const b3Vector3 x = leaves[i]->volume.Center() - org;
+				for (int j = 0; j < 3; ++j)
 				{
-					++splitcount[j][b3Dot(x,axis[j])>0?1:0];
+					++splitcount[j][b3Dot(x, axis[j]) > 0 ? 1 : 0];
 				}
 			}
-			for( i=0;i<3;++i)
+			for (i = 0; i < 3; ++i)
 			{
-				if((splitcount[i][0]>0)&&(splitcount[i][1]>0))
+				if ((splitcount[i][0] > 0) && (splitcount[i][1] > 0))
 				{
-					const int	midp=(int)b3Fabs(b3Scalar(splitcount[i][0]-splitcount[i][1]));
-					if(midp<bestmidp)
+					const int midp = (int)b3Fabs(b3Scalar(splitcount[i][0] - splitcount[i][1]));
+					if (midp < bestmidp)
 					{
-						bestaxis=i;
-						bestmidp=midp;
+						bestaxis = i;
+						bestmidp = midp;
 					}
 				}
 			}
-			if(bestaxis>=0)
+			if (bestaxis >= 0)
 			{
-				partition=b3Split(leaves,count,org,axis[bestaxis]);
-				b3Assert(partition!=0 && partition!=count);
+				partition = b3Split(leaves, count, org, axis[bestaxis]);
+				b3Assert(partition != 0 && partition != count);
 			}
 			else
 			{
-				partition=count/2+1;
+				partition = count / 2 + 1;
 			}
-			b3DbvtNode*	node=b3CreateNode(pdbvt,0,vol,0);
-			node->childs[0]=b3TopDown(pdbvt,&leaves[0],partition,bu_treshold);
-			node->childs[1]=b3TopDown(pdbvt,&leaves[partition],count-partition,bu_treshold);
-			node->childs[0]->parent=node;
-			node->childs[1]->parent=node;
-			return(node);
+			b3DbvtNode* node = b3CreateNode(pdbvt, 0, vol, 0);
+			node->childs[0] = b3TopDown(pdbvt, &leaves[0], partition, bu_treshold);
+			node->childs[1] = b3TopDown(pdbvt, &leaves[partition], count - partition, bu_treshold);
+			node->childs[0]->parent = node;
+			node->childs[1]->parent = node;
+			return (node);
 		}
 		else
 		{
-			b3BottomUp(pdbvt,leaves,count);
-			return(leaves[0]);
+			b3BottomUp(pdbvt, leaves, count);
+			return (leaves[0]);
 		}
 	}
-	return(leaves[0]);
+	return (leaves[0]);
 }
 
 //
-static B3_DBVT_INLINE b3DbvtNode*	b3Sort(b3DbvtNode* n,b3DbvtNode*& r)
+static B3_DBVT_INLINE b3DbvtNode* b3Sort(b3DbvtNode* n, b3DbvtNode*& r)
 {
-	b3DbvtNode*	p=n->parent;
+	b3DbvtNode* p = n->parent;
 	b3Assert(n->isinternal());
-	if(p>n)
+	if (p > n)
 	{
-		const int		i=b3IndexOf(n);
-		const int		j=1-i;
-		b3DbvtNode*	s=p->childs[j];
-		b3DbvtNode*	q=p->parent;
-		b3Assert(n==p->childs[i]);
-		if(q) q->childs[b3IndexOf(p)]=n; else r=n;
-		s->parent=n;
-		p->parent=n;
-		n->parent=q;
-		p->childs[0]=n->childs[0];
-		p->childs[1]=n->childs[1];
-		n->childs[0]->parent=p;
-		n->childs[1]->parent=p;
-		n->childs[i]=p;
-		n->childs[j]=s;
-		b3Swap(p->volume,n->volume);
-		return(p);
+		const int i = b3IndexOf(n);
+		const int j = 1 - i;
+		b3DbvtNode* s = p->childs[j];
+		b3DbvtNode* q = p->parent;
+		b3Assert(n == p->childs[i]);
+		if (q)
+			q->childs[b3IndexOf(p)] = n;
+		else
+			r = n;
+		s->parent = n;
+		p->parent = n;
+		n->parent = q;
+		p->childs[0] = n->childs[0];
+		p->childs[1] = n->childs[1];
+		n->childs[0]->parent = p;
+		n->childs[1]->parent = p;
+		n->childs[i] = p;
+		n->childs[j] = s;
+		b3Swap(p->volume, n->volume);
+		return (p);
 	}
-	return(n);
+	return (n);
 }
 
 #if 0
@@ -438,11 +456,11 @@ static B3_DBVT_INLINE b3DbvtNode*	walkup(b3DbvtNode* n,int count)
 //
 b3DynamicBvh::b3DynamicBvh()
 {
-	m_root		=	0;
-	m_free		=	0;
-	m_lkhd		=	-1;
-	m_leaves	=	0;
-	m_opath		=	0;
+	m_root = 0;
+	m_free = 0;
+	m_lkhd = -1;
+	m_leaves = 0;
+	m_opath = 0;
 }
 
 //
@@ -452,228 +470,233 @@ b3DynamicBvh::~b3DynamicBvh()
 }
 
 //
-void			b3DynamicBvh::clear()
+void b3DynamicBvh::clear()
 {
-	if(m_root)	
-		b3RecurseDeleteNode(this,m_root);
+	if (m_root)
+		b3RecurseDeleteNode(this, m_root);
 	b3AlignedFree(m_free);
-	m_free=0;
-	m_lkhd		=	-1;
+	m_free = 0;
+	m_lkhd = -1;
 	m_stkStack.clear();
-	m_opath		=	0;
-	
+	m_opath = 0;
 }
 
 //
-void			b3DynamicBvh::optimizeBottomUp()
+void b3DynamicBvh::optimizeBottomUp()
 {
-	if(m_root)
+	if (m_root)
 	{
 		b3NodeArray leaves;
 		leaves.reserve(m_leaves);
-		b3FetchLeaves(this,m_root,leaves);
-		b3BottomUp(this,&leaves[0],leaves.size());
-		m_root=leaves[0];
+		b3FetchLeaves(this, m_root, leaves);
+		b3BottomUp(this, &leaves[0], leaves.size());
+		m_root = leaves[0];
 	}
 }
 
 //
-void			b3DynamicBvh::optimizeTopDown(int bu_treshold)
+void b3DynamicBvh::optimizeTopDown(int bu_treshold)
 {
-	if(m_root)
+	if (m_root)
 	{
-		b3NodeArray	leaves;
+		b3NodeArray leaves;
 		leaves.reserve(m_leaves);
-		b3FetchLeaves(this,m_root,leaves);
-		m_root=b3TopDown(this,&leaves[0],leaves.size(),bu_treshold);
+		b3FetchLeaves(this, m_root, leaves);
+		m_root = b3TopDown(this, &leaves[0], leaves.size(), bu_treshold);
 	}
 }
 
 //
-void			b3DynamicBvh::optimizeIncremental(int passes)
+void b3DynamicBvh::optimizeIncremental(int passes)
 {
-	if(passes<0) passes=m_leaves;
-	if(m_root&&(passes>0))
+	if (passes < 0) passes = m_leaves;
+	if (m_root && (passes > 0))
 	{
-		do	{
-			b3DbvtNode*		node=m_root;
-			unsigned	bit=0;
-			while(node->isinternal())
+		do
+		{
+			b3DbvtNode* node = m_root;
+			unsigned bit = 0;
+			while (node->isinternal())
 			{
-				node=b3Sort(node,m_root)->childs[(m_opath>>bit)&1];
-				bit=(bit+1)&(sizeof(unsigned)*8-1);
+				node = b3Sort(node, m_root)->childs[(m_opath >> bit) & 1];
+				bit = (bit + 1) & (sizeof(unsigned) * 8 - 1);
 			}
 			update(node);
 			++m_opath;
-		} while(--passes);
+		} while (--passes);
 	}
 }
 
 //
-b3DbvtNode*	b3DynamicBvh::insert(const b3DbvtVolume& volume,void* data)
+b3DbvtNode* b3DynamicBvh::insert(const b3DbvtVolume& volume, void* data)
 {
-	b3DbvtNode*	leaf=b3CreateNode(this,0,volume,data);
-	b3InsertLeaf(this,m_root,leaf);
+	b3DbvtNode* leaf = b3CreateNode(this, 0, volume, data);
+	b3InsertLeaf(this, m_root, leaf);
 	++m_leaves;
-	return(leaf);
+	return (leaf);
 }
 
 //
-void			b3DynamicBvh::update(b3DbvtNode* leaf,int lookahead)
+void b3DynamicBvh::update(b3DbvtNode* leaf, int lookahead)
 {
-	b3DbvtNode*	root=b3RemoveLeaf(this,leaf);
-	if(root)
+	b3DbvtNode* root = b3RemoveLeaf(this, leaf);
+	if (root)
 	{
-		if(lookahead>=0)
+		if (lookahead >= 0)
 		{
-			for(int i=0;(i<lookahead)&&root->parent;++i)
+			for (int i = 0; (i < lookahead) && root->parent; ++i)
 			{
-				root=root->parent;
+				root = root->parent;
 			}
-		} else root=m_root;
+		}
+		else
+			root = m_root;
 	}
-	b3InsertLeaf(this,root,leaf);
+	b3InsertLeaf(this, root, leaf);
 }
 
 //
-void			b3DynamicBvh::update(b3DbvtNode* leaf,b3DbvtVolume& volume)
+void b3DynamicBvh::update(b3DbvtNode* leaf, b3DbvtVolume& volume)
 {
-	b3DbvtNode*	root=b3RemoveLeaf(this,leaf);
-	if(root)
+	b3DbvtNode* root = b3RemoveLeaf(this, leaf);
+	if (root)
 	{
-		if(m_lkhd>=0)
+		if (m_lkhd >= 0)
 		{
-			for(int i=0;(i<m_lkhd)&&root->parent;++i)
+			for (int i = 0; (i < m_lkhd) && root->parent; ++i)
 			{
-				root=root->parent;
+				root = root->parent;
 			}
-		} else root=m_root;
+		}
+		else
+			root = m_root;
 	}
-	leaf->volume=volume;
-	b3InsertLeaf(this,root,leaf);
+	leaf->volume = volume;
+	b3InsertLeaf(this, root, leaf);
 }
 
 //
-bool			b3DynamicBvh::update(b3DbvtNode* leaf,b3DbvtVolume& volume,const b3Vector3& velocity,b3Scalar margin)
+bool b3DynamicBvh::update(b3DbvtNode* leaf, b3DbvtVolume& volume, const b3Vector3& velocity, b3Scalar margin)
 {
-	if(leaf->volume.Contain(volume)) return(false);
-	volume.Expand(b3MakeVector3(margin,margin,margin));
+	if (leaf->volume.Contain(volume)) return (false);
+	volume.Expand(b3MakeVector3(margin, margin, margin));
 	volume.SignedExpand(velocity);
-	update(leaf,volume);
-	return(true);
+	update(leaf, volume);
+	return (true);
 }
 
 //
-bool			b3DynamicBvh::update(b3DbvtNode* leaf,b3DbvtVolume& volume,const b3Vector3& velocity)
+bool b3DynamicBvh::update(b3DbvtNode* leaf, b3DbvtVolume& volume, const b3Vector3& velocity)
 {
-	if(leaf->volume.Contain(volume)) return(false);
+	if (leaf->volume.Contain(volume)) return (false);
 	volume.SignedExpand(velocity);
-	update(leaf,volume);
-	return(true);
+	update(leaf, volume);
+	return (true);
 }
 
 //
-bool			b3DynamicBvh::update(b3DbvtNode* leaf,b3DbvtVolume& volume,b3Scalar margin)
+bool b3DynamicBvh::update(b3DbvtNode* leaf, b3DbvtVolume& volume, b3Scalar margin)
 {
-	if(leaf->volume.Contain(volume)) return(false);
-	volume.Expand(b3MakeVector3(margin,margin,margin));
-	update(leaf,volume);
-	return(true);
+	if (leaf->volume.Contain(volume)) return (false);
+	volume.Expand(b3MakeVector3(margin, margin, margin));
+	update(leaf, volume);
+	return (true);
 }
 
 //
-void			b3DynamicBvh::remove(b3DbvtNode* leaf)
+void b3DynamicBvh::remove(b3DbvtNode* leaf)
 {
-	b3RemoveLeaf(this,leaf);
-	b3DeleteNode(this,leaf);
+	b3RemoveLeaf(this, leaf);
+	b3DeleteNode(this, leaf);
 	--m_leaves;
 }
 
 //
-void			b3DynamicBvh::write(IWriter* iwriter) const
+void b3DynamicBvh::write(IWriter* iwriter) const
 {
-	b3DbvtNodeEnumerator	nodes;
-	nodes.nodes.reserve(m_leaves*2);
-	enumNodes(m_root,nodes);
-	iwriter->Prepare(m_root,nodes.nodes.size());
-	for(int i=0;i<nodes.nodes.size();++i)
+	b3DbvtNodeEnumerator nodes;
+	nodes.nodes.reserve(m_leaves * 2);
+	enumNodes(m_root, nodes);
+	iwriter->Prepare(m_root, nodes.nodes.size());
+	for (int i = 0; i < nodes.nodes.size(); ++i)
 	{
-		const b3DbvtNode* n=nodes.nodes[i];
-		int			p=-1;
-		if(n->parent) p=nodes.nodes.findLinearSearch(n->parent);
-		if(n->isinternal())
+		const b3DbvtNode* n = nodes.nodes[i];
+		int p = -1;
+		if (n->parent) p = nodes.nodes.findLinearSearch(n->parent);
+		if (n->isinternal())
 		{
-			const int	c0=nodes.nodes.findLinearSearch(n->childs[0]);
-			const int	c1=nodes.nodes.findLinearSearch(n->childs[1]);
-			iwriter->WriteNode(n,i,p,c0,c1);
+			const int c0 = nodes.nodes.findLinearSearch(n->childs[0]);
+			const int c1 = nodes.nodes.findLinearSearch(n->childs[1]);
+			iwriter->WriteNode(n, i, p, c0, c1);
 		}
 		else
 		{
-			iwriter->WriteLeaf(n,i,p);
-		}	
+			iwriter->WriteLeaf(n, i, p);
+		}
 	}
 }
 
 //
-void			b3DynamicBvh::clone(b3DynamicBvh& dest,IClone* iclone) const
+void b3DynamicBvh::clone(b3DynamicBvh& dest, IClone* iclone) const
 {
 	dest.clear();
-	if(m_root!=0)
-	{	
-		b3AlignedObjectArray<sStkCLN>	stack;
+	if (m_root != 0)
+	{
+		b3AlignedObjectArray<sStkCLN> stack;
 		stack.reserve(m_leaves);
-		stack.push_back(sStkCLN(m_root,0));
-		do	{
-			const int		i=stack.size()-1;
-			const sStkCLN	e=stack[i];
-			b3DbvtNode*			n=b3CreateNode(&dest,e.parent,e.node->volume,e.node->data);
+		stack.push_back(sStkCLN(m_root, 0));
+		do
+		{
+			const int i = stack.size() - 1;
+			const sStkCLN e = stack[i];
+			b3DbvtNode* n = b3CreateNode(&dest, e.parent, e.node->volume, e.node->data);
 			stack.pop_back();
-			if(e.parent!=0)
-				e.parent->childs[i&1]=n;
+			if (e.parent != 0)
+				e.parent->childs[i & 1] = n;
 			else
-				dest.m_root=n;
-			if(e.node->isinternal())
+				dest.m_root = n;
+			if (e.node->isinternal())
 			{
-				stack.push_back(sStkCLN(e.node->childs[0],n));
-				stack.push_back(sStkCLN(e.node->childs[1],n));
+				stack.push_back(sStkCLN(e.node->childs[0], n));
+				stack.push_back(sStkCLN(e.node->childs[1], n));
 			}
 			else
 			{
 				iclone->CloneLeaf(n);
 			}
-		} while(stack.size()>0);
+		} while (stack.size() > 0);
 	}
 }
 
 //
-int				b3DynamicBvh::maxdepth(const b3DbvtNode* node)
+int b3DynamicBvh::maxdepth(const b3DbvtNode* node)
 {
-	int	depth=0;
-	if(node) b3GetMaxDepth(node,1,depth);
-	return(depth);
+	int depth = 0;
+	if (node) b3GetMaxDepth(node, 1, depth);
+	return (depth);
 }
 
 //
-int				b3DynamicBvh::countLeaves(const b3DbvtNode* node)
+int b3DynamicBvh::countLeaves(const b3DbvtNode* node)
 {
-	if(node->isinternal())
-		return(countLeaves(node->childs[0])+countLeaves(node->childs[1]));
+	if (node->isinternal())
+		return (countLeaves(node->childs[0]) + countLeaves(node->childs[1]));
 	else
-		return(1);
+		return (1);
 }
 
 //
-void			b3DynamicBvh::extractLeaves(const b3DbvtNode* node,b3AlignedObjectArray<const b3DbvtNode*>& leaves)
+void b3DynamicBvh::extractLeaves(const b3DbvtNode* node, b3AlignedObjectArray<const b3DbvtNode*>& leaves)
 {
-	if(node->isinternal())
+	if (node->isinternal())
 	{
-		extractLeaves(node->childs[0],leaves);
-		extractLeaves(node->childs[1],leaves);
+		extractLeaves(node->childs[0], leaves);
+		extractLeaves(node->childs[1], leaves);
 	}
 	else
 	{
 		leaves.push_back(node);
-	}	
+	}
 }
 
 //
@@ -682,7 +705,6 @@ void			b3DynamicBvh::extractLeaves(const b3DbvtNode* node,b3AlignedObjectArray<c
 #include <stdio.h>
 #include <stdlib.h>
 
-
 /*
 q6600,2.4ghz
 
@@ -722,603 +744,608 @@ struct b3DbvtBenchmark
 {
 	struct NilPolicy : b3DynamicBvh::ICollide
 	{
-		NilPolicy() : m_pcount(0),m_depth(-B3_INFINITY),m_checksort(true)		{}
-		void	Process(const b3DbvtNode*,const b3DbvtNode*)				{ ++m_pcount; }
-		void	Process(const b3DbvtNode*)									{ ++m_pcount; }
-		void	Process(const b3DbvtNode*,b3Scalar depth)
+		NilPolicy() : m_pcount(0), m_depth(-B3_INFINITY), m_checksort(true) {}
+		void Process(const b3DbvtNode*, const b3DbvtNode*) { ++m_pcount; }
+		void Process(const b3DbvtNode*) { ++m_pcount; }
+		void Process(const b3DbvtNode*, b3Scalar depth)
 		{
 			++m_pcount;
-			if(m_checksort)
-			{ if(depth>=m_depth) m_depth=depth; else printf("wrong depth: %f (should be >= %f)\r\n",depth,m_depth); }
+			if (m_checksort)
+			{
+				if (depth >= m_depth)
+					m_depth = depth;
+				else
+					printf("wrong depth: %f (should be >= %f)\r\n", depth, m_depth);
+			}
 		}
-		int			m_pcount;
-		b3Scalar	m_depth;
-		bool		m_checksort;
+		int m_pcount;
+		b3Scalar m_depth;
+		bool m_checksort;
 	};
 	struct P14 : b3DynamicBvh::ICollide
 	{
 		struct Node
 		{
-			const b3DbvtNode*	leaf;
-			b3Scalar			depth;
+			const b3DbvtNode* leaf;
+			b3Scalar depth;
 		};
-		void Process(const b3DbvtNode* leaf,b3Scalar depth)
+		void Process(const b3DbvtNode* leaf, b3Scalar depth)
 		{
-			Node	n;
-			n.leaf	=	leaf;
-			n.depth	=	depth;
+			Node n;
+			n.leaf = leaf;
+			n.depth = depth;
 		}
-		static int sortfnc(const Node& a,const Node& b)
+		static int sortfnc(const Node& a, const Node& b)
 		{
-			if(a.depth<b.depth) return(+1);
-			if(a.depth>b.depth) return(-1);
-			return(0);
+			if (a.depth < b.depth) return (+1);
+			if (a.depth > b.depth) return (-1);
+			return (0);
 		}
-		b3AlignedObjectArray<Node>		m_nodes;
+		b3AlignedObjectArray<Node> m_nodes;
 	};
 	struct P15 : b3DynamicBvh::ICollide
 	{
 		struct Node
 		{
-			const b3DbvtNode*	leaf;
-			b3Scalar			depth;
+			const b3DbvtNode* leaf;
+			b3Scalar depth;
 		};
 		void Process(const b3DbvtNode* leaf)
 		{
-			Node	n;
-			n.leaf	=	leaf;
-			n.depth	=	dot(leaf->volume.Center(),m_axis);
+			Node n;
+			n.leaf = leaf;
+			n.depth = dot(leaf->volume.Center(), m_axis);
 		}
-		static int sortfnc(const Node& a,const Node& b)
+		static int sortfnc(const Node& a, const Node& b)
 		{
-			if(a.depth<b.depth) return(+1);
-			if(a.depth>b.depth) return(-1);
-			return(0);
+			if (a.depth < b.depth) return (+1);
+			if (a.depth > b.depth) return (-1);
+			return (0);
 		}
-		b3AlignedObjectArray<Node>		m_nodes;
-		b3Vector3						m_axis;
+		b3AlignedObjectArray<Node> m_nodes;
+		b3Vector3 m_axis;
 	};
-	static b3Scalar			RandUnit()
+	static b3Scalar RandUnit()
 	{
-		return(rand()/(b3Scalar)RAND_MAX);
+		return (rand() / (b3Scalar)RAND_MAX);
 	}
-	static b3Vector3		RandVector3()
+	static b3Vector3 RandVector3()
 	{
-		return(b3Vector3(RandUnit(),RandUnit(),RandUnit()));
+		return (b3Vector3(RandUnit(), RandUnit(), RandUnit()));
 	}
-	static b3Vector3		RandVector3(b3Scalar cs)
+	static b3Vector3 RandVector3(b3Scalar cs)
 	{
-		return(RandVector3()*cs-b3Vector3(cs,cs,cs)/2);
+		return (RandVector3() * cs - b3Vector3(cs, cs, cs) / 2);
 	}
-	static b3DbvtVolume	RandVolume(b3Scalar cs,b3Scalar eb,b3Scalar es)
+	static b3DbvtVolume RandVolume(b3Scalar cs, b3Scalar eb, b3Scalar es)
 	{
-		return(b3DbvtVolume::FromCE(RandVector3(cs),b3Vector3(eb,eb,eb)+RandVector3()*es));
+		return (b3DbvtVolume::FromCE(RandVector3(cs), b3Vector3(eb, eb, eb) + RandVector3() * es));
 	}
-	static b3Transform		RandTransform(b3Scalar cs)
+	static b3Transform RandTransform(b3Scalar cs)
 	{
-		b3Transform	t;
+		b3Transform t;
 		t.setOrigin(RandVector3(cs));
-		t.setRotation(b3Quaternion(RandUnit()*B3_PI*2,RandUnit()*B3_PI*2,RandUnit()*B3_PI*2).normalized());
-		return(t);
+		t.setRotation(b3Quaternion(RandUnit() * B3_PI * 2, RandUnit() * B3_PI * 2, RandUnit() * B3_PI * 2).normalized());
+		return (t);
 	}
-	static void				RandTree(b3Scalar cs,b3Scalar eb,b3Scalar es,int leaves,b3DynamicBvh& dbvt)
+	static void RandTree(b3Scalar cs, b3Scalar eb, b3Scalar es, int leaves, b3DynamicBvh& dbvt)
 	{
 		dbvt.clear();
-		for(int i=0;i<leaves;++i)
+		for (int i = 0; i < leaves; ++i)
 		{
-			dbvt.insert(RandVolume(cs,eb,es),0);
+			dbvt.insert(RandVolume(cs, eb, es), 0);
 		}
 	}
 };
 
-void			b3DynamicBvh::benchmark()
+void b3DynamicBvh::benchmark()
 {
-	static const b3Scalar	cfgVolumeCenterScale		=	100;
-	static const b3Scalar	cfgVolumeExentsBase			=	1;
-	static const b3Scalar	cfgVolumeExentsScale		=	4;
-	static const int		cfgLeaves					=	8192;
-	static const bool		cfgEnable					=	true;
+	static const b3Scalar cfgVolumeCenterScale = 100;
+	static const b3Scalar cfgVolumeExentsBase = 1;
+	static const b3Scalar cfgVolumeExentsScale = 4;
+	static const int cfgLeaves = 8192;
+	static const bool cfgEnable = true;
 
 	//[1] b3DbvtVolume intersections
-	bool					cfgBenchmark1_Enable		=	cfgEnable;
-	static const int		cfgBenchmark1_Iterations	=	8;
-	static const int		cfgBenchmark1_Reference		=	3499;
+	bool cfgBenchmark1_Enable = cfgEnable;
+	static const int cfgBenchmark1_Iterations = 8;
+	static const int cfgBenchmark1_Reference = 3499;
 	//[2] b3DbvtVolume merges
-	bool					cfgBenchmark2_Enable		=	cfgEnable;
-	static const int		cfgBenchmark2_Iterations	=	4;
-	static const int		cfgBenchmark2_Reference		=	1945;
+	bool cfgBenchmark2_Enable = cfgEnable;
+	static const int cfgBenchmark2_Iterations = 4;
+	static const int cfgBenchmark2_Reference = 1945;
 	//[3] b3DynamicBvh::collideTT
-	bool					cfgBenchmark3_Enable		=	cfgEnable;
-	static const int		cfgBenchmark3_Iterations	=	512;
-	static const int		cfgBenchmark3_Reference		=	5485;
+	bool cfgBenchmark3_Enable = cfgEnable;
+	static const int cfgBenchmark3_Iterations = 512;
+	static const int cfgBenchmark3_Reference = 5485;
 	//[4] b3DynamicBvh::collideTT self
-	bool					cfgBenchmark4_Enable		=	cfgEnable;
-	static const int		cfgBenchmark4_Iterations	=	512;
-	static const int		cfgBenchmark4_Reference		=	2814;
+	bool cfgBenchmark4_Enable = cfgEnable;
+	static const int cfgBenchmark4_Iterations = 512;
+	static const int cfgBenchmark4_Reference = 2814;
 	//[5] b3DynamicBvh::collideTT xform
-	bool					cfgBenchmark5_Enable		=	cfgEnable;
-	static const int		cfgBenchmark5_Iterations	=	512;
-	static const b3Scalar	cfgBenchmark5_OffsetScale	=	2;
-	static const int		cfgBenchmark5_Reference		=	7379;
+	bool cfgBenchmark5_Enable = cfgEnable;
+	static const int cfgBenchmark5_Iterations = 512;
+	static const b3Scalar cfgBenchmark5_OffsetScale = 2;
+	static const int cfgBenchmark5_Reference = 7379;
 	//[6] b3DynamicBvh::collideTT xform,self
-	bool					cfgBenchmark6_Enable		=	cfgEnable;
-	static const int		cfgBenchmark6_Iterations	=	512;
-	static const b3Scalar	cfgBenchmark6_OffsetScale	=	2;
-	static const int		cfgBenchmark6_Reference		=	7270;
+	bool cfgBenchmark6_Enable = cfgEnable;
+	static const int cfgBenchmark6_Iterations = 512;
+	static const b3Scalar cfgBenchmark6_OffsetScale = 2;
+	static const int cfgBenchmark6_Reference = 7270;
 	//[7] b3DynamicBvh::rayTest
-	bool					cfgBenchmark7_Enable		=	cfgEnable;
-	static const int		cfgBenchmark7_Passes		=	32;
-	static const int		cfgBenchmark7_Iterations	=	65536;
-	static const int		cfgBenchmark7_Reference		=	6307;
+	bool cfgBenchmark7_Enable = cfgEnable;
+	static const int cfgBenchmark7_Passes = 32;
+	static const int cfgBenchmark7_Iterations = 65536;
+	static const int cfgBenchmark7_Reference = 6307;
 	//[8] insert/remove
-	bool					cfgBenchmark8_Enable		=	cfgEnable;
-	static const int		cfgBenchmark8_Passes		=	32;
-	static const int		cfgBenchmark8_Iterations	=	65536;
-	static const int		cfgBenchmark8_Reference		=	2105;
+	bool cfgBenchmark8_Enable = cfgEnable;
+	static const int cfgBenchmark8_Passes = 32;
+	static const int cfgBenchmark8_Iterations = 65536;
+	static const int cfgBenchmark8_Reference = 2105;
 	//[9] updates (teleport)
-	bool					cfgBenchmark9_Enable		=	cfgEnable;
-	static const int		cfgBenchmark9_Passes		=	32;
-	static const int		cfgBenchmark9_Iterations	=	65536;
-	static const int		cfgBenchmark9_Reference		=	1879;
+	bool cfgBenchmark9_Enable = cfgEnable;
+	static const int cfgBenchmark9_Passes = 32;
+	static const int cfgBenchmark9_Iterations = 65536;
+	static const int cfgBenchmark9_Reference = 1879;
 	//[10] updates (jitter)
-	bool					cfgBenchmark10_Enable		=	cfgEnable;
-	static const b3Scalar	cfgBenchmark10_Scale		=	cfgVolumeCenterScale/10000;
-	static const int		cfgBenchmark10_Passes		=	32;
-	static const int		cfgBenchmark10_Iterations	=	65536;
-	static const int		cfgBenchmark10_Reference	=	1244;
+	bool cfgBenchmark10_Enable = cfgEnable;
+	static const b3Scalar cfgBenchmark10_Scale = cfgVolumeCenterScale / 10000;
+	static const int cfgBenchmark10_Passes = 32;
+	static const int cfgBenchmark10_Iterations = 65536;
+	static const int cfgBenchmark10_Reference = 1244;
 	//[11] optimize (incremental)
-	bool					cfgBenchmark11_Enable		=	cfgEnable;
-	static const int		cfgBenchmark11_Passes		=	64;
-	static const int		cfgBenchmark11_Iterations	=	65536;
-	static const int		cfgBenchmark11_Reference	=	2510;
+	bool cfgBenchmark11_Enable = cfgEnable;
+	static const int cfgBenchmark11_Passes = 64;
+	static const int cfgBenchmark11_Iterations = 65536;
+	static const int cfgBenchmark11_Reference = 2510;
 	//[12] b3DbvtVolume notequal
-	bool					cfgBenchmark12_Enable		=	cfgEnable;
-	static const int		cfgBenchmark12_Iterations	=	32;
-	static const int		cfgBenchmark12_Reference	=	3677;
+	bool cfgBenchmark12_Enable = cfgEnable;
+	static const int cfgBenchmark12_Iterations = 32;
+	static const int cfgBenchmark12_Reference = 3677;
 	//[13] culling(OCL+fullsort)
-	bool					cfgBenchmark13_Enable		=	cfgEnable;
-	static const int		cfgBenchmark13_Iterations	=	1024;
-	static const int		cfgBenchmark13_Reference	=	2231;
+	bool cfgBenchmark13_Enable = cfgEnable;
+	static const int cfgBenchmark13_Iterations = 1024;
+	static const int cfgBenchmark13_Reference = 2231;
 	//[14] culling(OCL+qsort)
-	bool					cfgBenchmark14_Enable		=	cfgEnable;
-	static const int		cfgBenchmark14_Iterations	=	8192;
-	static const int		cfgBenchmark14_Reference	=	3500;
+	bool cfgBenchmark14_Enable = cfgEnable;
+	static const int cfgBenchmark14_Iterations = 8192;
+	static const int cfgBenchmark14_Reference = 3500;
 	//[15] culling(KDOP+qsort)
-	bool					cfgBenchmark15_Enable		=	cfgEnable;
-	static const int		cfgBenchmark15_Iterations	=	8192;
-	static const int		cfgBenchmark15_Reference	=	1151;
+	bool cfgBenchmark15_Enable = cfgEnable;
+	static const int cfgBenchmark15_Iterations = 8192;
+	static const int cfgBenchmark15_Reference = 1151;
 	//[16] insert/remove batch
-	bool					cfgBenchmark16_Enable		=	cfgEnable;
-	static const int		cfgBenchmark16_BatchCount	=	256;
-	static const int		cfgBenchmark16_Passes		=	16384;
-	static const int		cfgBenchmark16_Reference	=	5138;
+	bool cfgBenchmark16_Enable = cfgEnable;
+	static const int cfgBenchmark16_BatchCount = 256;
+	static const int cfgBenchmark16_Passes = 16384;
+	static const int cfgBenchmark16_Reference = 5138;
 	//[17] select
-	bool					cfgBenchmark17_Enable		=	cfgEnable;
-	static const int		cfgBenchmark17_Iterations	=	4;
-	static const int		cfgBenchmark17_Reference	=	3390;
+	bool cfgBenchmark17_Enable = cfgEnable;
+	static const int cfgBenchmark17_Iterations = 4;
+	static const int cfgBenchmark17_Reference = 3390;
 
-	b3Clock					wallclock;
+	b3Clock wallclock;
 	printf("Benchmarking dbvt...\r\n");
-	printf("\tWorld scale: %f\r\n",cfgVolumeCenterScale);
-	printf("\tExtents base: %f\r\n",cfgVolumeExentsBase);
-	printf("\tExtents range: %f\r\n",cfgVolumeExentsScale);
-	printf("\tLeaves: %u\r\n",cfgLeaves);
-	printf("\tsizeof(b3DbvtVolume): %u bytes\r\n",sizeof(b3DbvtVolume));
-	printf("\tsizeof(b3DbvtNode):   %u bytes\r\n",sizeof(b3DbvtNode));
-	if(cfgBenchmark1_Enable)
-	{// Benchmark 1	
+	printf("\tWorld scale: %f\r\n", cfgVolumeCenterScale);
+	printf("\tExtents base: %f\r\n", cfgVolumeExentsBase);
+	printf("\tExtents range: %f\r\n", cfgVolumeExentsScale);
+	printf("\tLeaves: %u\r\n", cfgLeaves);
+	printf("\tsizeof(b3DbvtVolume): %u bytes\r\n", sizeof(b3DbvtVolume));
+	printf("\tsizeof(b3DbvtNode):   %u bytes\r\n", sizeof(b3DbvtNode));
+	if (cfgBenchmark1_Enable)
+	{  // Benchmark 1
 		srand(380843);
-		b3AlignedObjectArray<b3DbvtVolume>	volumes;
-		b3AlignedObjectArray<bool>			results;
+		b3AlignedObjectArray<b3DbvtVolume> volumes;
+		b3AlignedObjectArray<bool> results;
 		volumes.resize(cfgLeaves);
 		results.resize(cfgLeaves);
-		for(int i=0;i<cfgLeaves;++i)
+		for (int i = 0; i < cfgLeaves; ++i)
 		{
-			volumes[i]=b3DbvtBenchmark::RandVolume(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale);
+			volumes[i] = b3DbvtBenchmark::RandVolume(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale);
 		}
 		printf("[1] b3DbvtVolume intersections: ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark1_Iterations;++i)
+		for (int i = 0; i < cfgBenchmark1_Iterations; ++i)
 		{
-			for(int j=0;j<cfgLeaves;++j)
+			for (int j = 0; j < cfgLeaves; ++j)
 			{
-				for(int k=0;k<cfgLeaves;++k)
+				for (int k = 0; k < cfgLeaves; ++k)
 				{
-					results[k]=Intersect(volumes[j],volumes[k]);
+					results[k] = Intersect(volumes[j], volumes[k]);
 				}
 			}
 		}
-		const int time=(int)wallclock.getTimeMilliseconds();
-		printf("%u ms (%i%%)\r\n",time,(time-cfgBenchmark1_Reference)*100/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		printf("%u ms (%i%%)\r\n", time, (time - cfgBenchmark1_Reference) * 100 / time);
 	}
-	if(cfgBenchmark2_Enable)
-	{// Benchmark 2	
+	if (cfgBenchmark2_Enable)
+	{  // Benchmark 2
 		srand(380843);
-		b3AlignedObjectArray<b3DbvtVolume>	volumes;
-		b3AlignedObjectArray<b3DbvtVolume>	results;
+		b3AlignedObjectArray<b3DbvtVolume> volumes;
+		b3AlignedObjectArray<b3DbvtVolume> results;
 		volumes.resize(cfgLeaves);
 		results.resize(cfgLeaves);
-		for(int i=0;i<cfgLeaves;++i)
+		for (int i = 0; i < cfgLeaves; ++i)
 		{
-			volumes[i]=b3DbvtBenchmark::RandVolume(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale);
+			volumes[i] = b3DbvtBenchmark::RandVolume(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale);
 		}
 		printf("[2] b3DbvtVolume merges: ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark2_Iterations;++i)
+		for (int i = 0; i < cfgBenchmark2_Iterations; ++i)
 		{
-			for(int j=0;j<cfgLeaves;++j)
+			for (int j = 0; j < cfgLeaves; ++j)
 			{
-				for(int k=0;k<cfgLeaves;++k)
+				for (int k = 0; k < cfgLeaves; ++k)
 				{
-					Merge(volumes[j],volumes[k],results[k]);
+					Merge(volumes[j], volumes[k], results[k]);
 				}
 			}
 		}
-		const int time=(int)wallclock.getTimeMilliseconds();
-		printf("%u ms (%i%%)\r\n",time,(time-cfgBenchmark2_Reference)*100/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		printf("%u ms (%i%%)\r\n", time, (time - cfgBenchmark2_Reference) * 100 / time);
 	}
-	if(cfgBenchmark3_Enable)
-	{// Benchmark 3	
+	if (cfgBenchmark3_Enable)
+	{  // Benchmark 3
 		srand(380843);
-		b3DynamicBvh						dbvt[2];
-		b3DbvtBenchmark::NilPolicy	policy;
-		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt[0]);
-		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt[1]);
+		b3DynamicBvh dbvt[2];
+		b3DbvtBenchmark::NilPolicy policy;
+		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt[0]);
+		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt[1]);
 		dbvt[0].optimizeTopDown();
 		dbvt[1].optimizeTopDown();
 		printf("[3] b3DynamicBvh::collideTT: ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark3_Iterations;++i)
+		for (int i = 0; i < cfgBenchmark3_Iterations; ++i)
 		{
-			b3DynamicBvh::collideTT(dbvt[0].m_root,dbvt[1].m_root,policy);
+			b3DynamicBvh::collideTT(dbvt[0].m_root, dbvt[1].m_root, policy);
 		}
-		const int time=(int)wallclock.getTimeMilliseconds();
-		printf("%u ms (%i%%)\r\n",time,(time-cfgBenchmark3_Reference)*100/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		printf("%u ms (%i%%)\r\n", time, (time - cfgBenchmark3_Reference) * 100 / time);
 	}
-	if(cfgBenchmark4_Enable)
-	{// Benchmark 4
+	if (cfgBenchmark4_Enable)
+	{  // Benchmark 4
 		srand(380843);
-		b3DynamicBvh						dbvt;
-		b3DbvtBenchmark::NilPolicy	policy;
-		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		b3DynamicBvh dbvt;
+		b3DbvtBenchmark::NilPolicy policy;
+		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
 		printf("[4] b3DynamicBvh::collideTT self: ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark4_Iterations;++i)
+		for (int i = 0; i < cfgBenchmark4_Iterations; ++i)
 		{
-			b3DynamicBvh::collideTT(dbvt.m_root,dbvt.m_root,policy);
+			b3DynamicBvh::collideTT(dbvt.m_root, dbvt.m_root, policy);
 		}
-		const int time=(int)wallclock.getTimeMilliseconds();
-		printf("%u ms (%i%%)\r\n",time,(time-cfgBenchmark4_Reference)*100/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		printf("%u ms (%i%%)\r\n", time, (time - cfgBenchmark4_Reference) * 100 / time);
 	}
-	if(cfgBenchmark5_Enable)
-	{// Benchmark 5	
+	if (cfgBenchmark5_Enable)
+	{  // Benchmark 5
 		srand(380843);
-		b3DynamicBvh								dbvt[2];
-		b3AlignedObjectArray<b3Transform>	transforms;
-		b3DbvtBenchmark::NilPolicy			policy;
+		b3DynamicBvh dbvt[2];
+		b3AlignedObjectArray<b3Transform> transforms;
+		b3DbvtBenchmark::NilPolicy policy;
 		transforms.resize(cfgBenchmark5_Iterations);
-		for(int i=0;i<transforms.size();++i)
+		for (int i = 0; i < transforms.size(); ++i)
 		{
-			transforms[i]=b3DbvtBenchmark::RandTransform(cfgVolumeCenterScale*cfgBenchmark5_OffsetScale);
+			transforms[i] = b3DbvtBenchmark::RandTransform(cfgVolumeCenterScale * cfgBenchmark5_OffsetScale);
 		}
-		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt[0]);
-		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt[1]);
+		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt[0]);
+		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt[1]);
 		dbvt[0].optimizeTopDown();
 		dbvt[1].optimizeTopDown();
 		printf("[5] b3DynamicBvh::collideTT xform: ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark5_Iterations;++i)
+		for (int i = 0; i < cfgBenchmark5_Iterations; ++i)
 		{
-			b3DynamicBvh::collideTT(dbvt[0].m_root,dbvt[1].m_root,transforms[i],policy);
+			b3DynamicBvh::collideTT(dbvt[0].m_root, dbvt[1].m_root, transforms[i], policy);
 		}
-		const int time=(int)wallclock.getTimeMilliseconds();
-		printf("%u ms (%i%%)\r\n",time,(time-cfgBenchmark5_Reference)*100/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		printf("%u ms (%i%%)\r\n", time, (time - cfgBenchmark5_Reference) * 100 / time);
 	}
-	if(cfgBenchmark6_Enable)
-	{// Benchmark 6	
+	if (cfgBenchmark6_Enable)
+	{  // Benchmark 6
 		srand(380843);
-		b3DynamicBvh								dbvt;
-		b3AlignedObjectArray<b3Transform>	transforms;
-		b3DbvtBenchmark::NilPolicy			policy;
+		b3DynamicBvh dbvt;
+		b3AlignedObjectArray<b3Transform> transforms;
+		b3DbvtBenchmark::NilPolicy policy;
 		transforms.resize(cfgBenchmark6_Iterations);
-		for(int i=0;i<transforms.size();++i)
+		for (int i = 0; i < transforms.size(); ++i)
 		{
-			transforms[i]=b3DbvtBenchmark::RandTransform(cfgVolumeCenterScale*cfgBenchmark6_OffsetScale);
+			transforms[i] = b3DbvtBenchmark::RandTransform(cfgVolumeCenterScale * cfgBenchmark6_OffsetScale);
 		}
-		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
 		printf("[6] b3DynamicBvh::collideTT xform,self: ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark6_Iterations;++i)
+		for (int i = 0; i < cfgBenchmark6_Iterations; ++i)
 		{
-			b3DynamicBvh::collideTT(dbvt.m_root,dbvt.m_root,transforms[i],policy);		
+			b3DynamicBvh::collideTT(dbvt.m_root, dbvt.m_root, transforms[i], policy);
 		}
-		const int time=(int)wallclock.getTimeMilliseconds();
-		printf("%u ms (%i%%)\r\n",time,(time-cfgBenchmark6_Reference)*100/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		printf("%u ms (%i%%)\r\n", time, (time - cfgBenchmark6_Reference) * 100 / time);
 	}
-	if(cfgBenchmark7_Enable)
-	{// Benchmark 7	
+	if (cfgBenchmark7_Enable)
+	{  // Benchmark 7
 		srand(380843);
-		b3DynamicBvh								dbvt;
-		b3AlignedObjectArray<b3Vector3>		rayorg;
-		b3AlignedObjectArray<b3Vector3>		raydir;
-		b3DbvtBenchmark::NilPolicy			policy;
+		b3DynamicBvh dbvt;
+		b3AlignedObjectArray<b3Vector3> rayorg;
+		b3AlignedObjectArray<b3Vector3> raydir;
+		b3DbvtBenchmark::NilPolicy policy;
 		rayorg.resize(cfgBenchmark7_Iterations);
 		raydir.resize(cfgBenchmark7_Iterations);
-		for(int i=0;i<rayorg.size();++i)
+		for (int i = 0; i < rayorg.size(); ++i)
 		{
-			rayorg[i]=b3DbvtBenchmark::RandVector3(cfgVolumeCenterScale*2);
-			raydir[i]=b3DbvtBenchmark::RandVector3(cfgVolumeCenterScale*2);
+			rayorg[i] = b3DbvtBenchmark::RandVector3(cfgVolumeCenterScale * 2);
+			raydir[i] = b3DbvtBenchmark::RandVector3(cfgVolumeCenterScale * 2);
 		}
-		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
 		printf("[7] b3DynamicBvh::rayTest: ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark7_Passes;++i)
+		for (int i = 0; i < cfgBenchmark7_Passes; ++i)
 		{
-			for(int j=0;j<cfgBenchmark7_Iterations;++j)
+			for (int j = 0; j < cfgBenchmark7_Iterations; ++j)
 			{
-				b3DynamicBvh::rayTest(dbvt.m_root,rayorg[j],rayorg[j]+raydir[j],policy);
+				b3DynamicBvh::rayTest(dbvt.m_root, rayorg[j], rayorg[j] + raydir[j], policy);
 			}
 		}
-		const int	time=(int)wallclock.getTimeMilliseconds();
-		unsigned	rays=cfgBenchmark7_Passes*cfgBenchmark7_Iterations;
-		printf("%u ms (%i%%),(%u r/s)\r\n",time,(time-cfgBenchmark7_Reference)*100/time,(rays*1000)/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		unsigned rays = cfgBenchmark7_Passes * cfgBenchmark7_Iterations;
+		printf("%u ms (%i%%),(%u r/s)\r\n", time, (time - cfgBenchmark7_Reference) * 100 / time, (rays * 1000) / time);
 	}
-	if(cfgBenchmark8_Enable)
-	{// Benchmark 8	
+	if (cfgBenchmark8_Enable)
+	{  // Benchmark 8
 		srand(380843);
-		b3DynamicBvh								dbvt;
-		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		b3DynamicBvh dbvt;
+		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
 		printf("[8] insert/remove: ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark8_Passes;++i)
+		for (int i = 0; i < cfgBenchmark8_Passes; ++i)
 		{
-			for(int j=0;j<cfgBenchmark8_Iterations;++j)
+			for (int j = 0; j < cfgBenchmark8_Iterations; ++j)
 			{
-				dbvt.remove(dbvt.insert(b3DbvtBenchmark::RandVolume(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale),0));
+				dbvt.remove(dbvt.insert(b3DbvtBenchmark::RandVolume(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale), 0));
 			}
 		}
-		const int	time=(int)wallclock.getTimeMilliseconds();
-		const int	ir=cfgBenchmark8_Passes*cfgBenchmark8_Iterations;
-		printf("%u ms (%i%%),(%u ir/s)\r\n",time,(time-cfgBenchmark8_Reference)*100/time,ir*1000/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		const int ir = cfgBenchmark8_Passes * cfgBenchmark8_Iterations;
+		printf("%u ms (%i%%),(%u ir/s)\r\n", time, (time - cfgBenchmark8_Reference) * 100 / time, ir * 1000 / time);
 	}
-	if(cfgBenchmark9_Enable)
-	{// Benchmark 9	
+	if (cfgBenchmark9_Enable)
+	{  // Benchmark 9
 		srand(380843);
-		b3DynamicBvh										dbvt;
-		b3AlignedObjectArray<const b3DbvtNode*>	leaves;
-		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		b3DynamicBvh dbvt;
+		b3AlignedObjectArray<const b3DbvtNode*> leaves;
+		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
-		dbvt.extractLeaves(dbvt.m_root,leaves);
+		dbvt.extractLeaves(dbvt.m_root, leaves);
 		printf("[9] updates (teleport): ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark9_Passes;++i)
+		for (int i = 0; i < cfgBenchmark9_Passes; ++i)
 		{
-			for(int j=0;j<cfgBenchmark9_Iterations;++j)
+			for (int j = 0; j < cfgBenchmark9_Iterations; ++j)
 			{
-				dbvt.update(const_cast<b3DbvtNode*>(leaves[rand()%cfgLeaves]),
-					b3DbvtBenchmark::RandVolume(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale));
+				dbvt.update(const_cast<b3DbvtNode*>(leaves[rand() % cfgLeaves]),
+							b3DbvtBenchmark::RandVolume(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale));
 			}
 		}
-		const int	time=(int)wallclock.getTimeMilliseconds();
-		const int	up=cfgBenchmark9_Passes*cfgBenchmark9_Iterations;
-		printf("%u ms (%i%%),(%u u/s)\r\n",time,(time-cfgBenchmark9_Reference)*100/time,up*1000/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		const int up = cfgBenchmark9_Passes * cfgBenchmark9_Iterations;
+		printf("%u ms (%i%%),(%u u/s)\r\n", time, (time - cfgBenchmark9_Reference) * 100 / time, up * 1000 / time);
 	}
-	if(cfgBenchmark10_Enable)
-	{// Benchmark 10	
+	if (cfgBenchmark10_Enable)
+	{  // Benchmark 10
 		srand(380843);
-		b3DynamicBvh										dbvt;
-		b3AlignedObjectArray<const b3DbvtNode*>	leaves;
-		b3AlignedObjectArray<b3Vector3>				vectors;
+		b3DynamicBvh dbvt;
+		b3AlignedObjectArray<const b3DbvtNode*> leaves;
+		b3AlignedObjectArray<b3Vector3> vectors;
 		vectors.resize(cfgBenchmark10_Iterations);
-		for(int i=0;i<vectors.size();++i)
+		for (int i = 0; i < vectors.size(); ++i)
 		{
-			vectors[i]=(b3DbvtBenchmark::RandVector3()*2-b3Vector3(1,1,1))*cfgBenchmark10_Scale;
+			vectors[i] = (b3DbvtBenchmark::RandVector3() * 2 - b3Vector3(1, 1, 1)) * cfgBenchmark10_Scale;
 		}
-		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
-		dbvt.extractLeaves(dbvt.m_root,leaves);
+		dbvt.extractLeaves(dbvt.m_root, leaves);
 		printf("[10] updates (jitter): ");
 		wallclock.reset();
 
-		for(int i=0;i<cfgBenchmark10_Passes;++i)
+		for (int i = 0; i < cfgBenchmark10_Passes; ++i)
 		{
-			for(int j=0;j<cfgBenchmark10_Iterations;++j)
-			{			
-				const b3Vector3&	d=vectors[j];
-				b3DbvtNode*		l=const_cast<b3DbvtNode*>(leaves[rand()%cfgLeaves]);
-				b3DbvtVolume		v=b3DbvtVolume::FromMM(l->volume.Mins()+d,l->volume.Maxs()+d);
-				dbvt.update(l,v);
+			for (int j = 0; j < cfgBenchmark10_Iterations; ++j)
+			{
+				const b3Vector3& d = vectors[j];
+				b3DbvtNode* l = const_cast<b3DbvtNode*>(leaves[rand() % cfgLeaves]);
+				b3DbvtVolume v = b3DbvtVolume::FromMM(l->volume.Mins() + d, l->volume.Maxs() + d);
+				dbvt.update(l, v);
 			}
 		}
-		const int	time=(int)wallclock.getTimeMilliseconds();
-		const int	up=cfgBenchmark10_Passes*cfgBenchmark10_Iterations;
-		printf("%u ms (%i%%),(%u u/s)\r\n",time,(time-cfgBenchmark10_Reference)*100/time,up*1000/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		const int up = cfgBenchmark10_Passes * cfgBenchmark10_Iterations;
+		printf("%u ms (%i%%),(%u u/s)\r\n", time, (time - cfgBenchmark10_Reference) * 100 / time, up * 1000 / time);
 	}
-	if(cfgBenchmark11_Enable)
-	{// Benchmark 11	
+	if (cfgBenchmark11_Enable)
+	{  // Benchmark 11
 		srand(380843);
-		b3DynamicBvh										dbvt;
-		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		b3DynamicBvh dbvt;
+		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
 		printf("[11] optimize (incremental): ");
-		wallclock.reset();	
-		for(int i=0;i<cfgBenchmark11_Passes;++i)
+		wallclock.reset();
+		for (int i = 0; i < cfgBenchmark11_Passes; ++i)
 		{
 			dbvt.optimizeIncremental(cfgBenchmark11_Iterations);
 		}
-		const int	time=(int)wallclock.getTimeMilliseconds();
-		const int	op=cfgBenchmark11_Passes*cfgBenchmark11_Iterations;
-		printf("%u ms (%i%%),(%u o/s)\r\n",time,(time-cfgBenchmark11_Reference)*100/time,op/time*1000);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		const int op = cfgBenchmark11_Passes * cfgBenchmark11_Iterations;
+		printf("%u ms (%i%%),(%u o/s)\r\n", time, (time - cfgBenchmark11_Reference) * 100 / time, op / time * 1000);
 	}
-	if(cfgBenchmark12_Enable)
-	{// Benchmark 12	
+	if (cfgBenchmark12_Enable)
+	{  // Benchmark 12
 		srand(380843);
-		b3AlignedObjectArray<b3DbvtVolume>	volumes;
-		b3AlignedObjectArray<bool>				results;
+		b3AlignedObjectArray<b3DbvtVolume> volumes;
+		b3AlignedObjectArray<bool> results;
 		volumes.resize(cfgLeaves);
 		results.resize(cfgLeaves);
-		for(int i=0;i<cfgLeaves;++i)
+		for (int i = 0; i < cfgLeaves; ++i)
 		{
-			volumes[i]=b3DbvtBenchmark::RandVolume(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale);
+			volumes[i] = b3DbvtBenchmark::RandVolume(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale);
 		}
 		printf("[12] b3DbvtVolume notequal: ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark12_Iterations;++i)
+		for (int i = 0; i < cfgBenchmark12_Iterations; ++i)
 		{
-			for(int j=0;j<cfgLeaves;++j)
+			for (int j = 0; j < cfgLeaves; ++j)
 			{
-				for(int k=0;k<cfgLeaves;++k)
+				for (int k = 0; k < cfgLeaves; ++k)
 				{
-					results[k]=NotEqual(volumes[j],volumes[k]);
+					results[k] = NotEqual(volumes[j], volumes[k]);
 				}
 			}
 		}
-		const int time=(int)wallclock.getTimeMilliseconds();
-		printf("%u ms (%i%%)\r\n",time,(time-cfgBenchmark12_Reference)*100/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		printf("%u ms (%i%%)\r\n", time, (time - cfgBenchmark12_Reference) * 100 / time);
 	}
-	if(cfgBenchmark13_Enable)
-	{// Benchmark 13	
+	if (cfgBenchmark13_Enable)
+	{  // Benchmark 13
 		srand(380843);
-		b3DynamicBvh								dbvt;
-		b3AlignedObjectArray<b3Vector3>		vectors;
-		b3DbvtBenchmark::NilPolicy			policy;
+		b3DynamicBvh dbvt;
+		b3AlignedObjectArray<b3Vector3> vectors;
+		b3DbvtBenchmark::NilPolicy policy;
 		vectors.resize(cfgBenchmark13_Iterations);
-		for(int i=0;i<vectors.size();++i)
+		for (int i = 0; i < vectors.size(); ++i)
 		{
-			vectors[i]=(b3DbvtBenchmark::RandVector3()*2-b3Vector3(1,1,1)).normalized();
+			vectors[i] = (b3DbvtBenchmark::RandVector3() * 2 - b3Vector3(1, 1, 1)).normalized();
 		}
-		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
 		printf("[13] culling(OCL+fullsort): ");
-		wallclock.reset();	
-		for(int i=0;i<cfgBenchmark13_Iterations;++i)
+		wallclock.reset();
+		for (int i = 0; i < cfgBenchmark13_Iterations; ++i)
 		{
-			static const b3Scalar	offset=0;
-			policy.m_depth=-B3_INFINITY;
-			dbvt.collideOCL(dbvt.m_root,&vectors[i],&offset,vectors[i],1,policy);
+			static const b3Scalar offset = 0;
+			policy.m_depth = -B3_INFINITY;
+			dbvt.collideOCL(dbvt.m_root, &vectors[i], &offset, vectors[i], 1, policy);
 		}
-		const int	time=(int)wallclock.getTimeMilliseconds();
-		const int	t=cfgBenchmark13_Iterations;
-		printf("%u ms (%i%%),(%u t/s)\r\n",time,(time-cfgBenchmark13_Reference)*100/time,(t*1000)/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		const int t = cfgBenchmark13_Iterations;
+		printf("%u ms (%i%%),(%u t/s)\r\n", time, (time - cfgBenchmark13_Reference) * 100 / time, (t * 1000) / time);
 	}
-	if(cfgBenchmark14_Enable)
-	{// Benchmark 14	
+	if (cfgBenchmark14_Enable)
+	{  // Benchmark 14
 		srand(380843);
-		b3DynamicBvh								dbvt;
-		b3AlignedObjectArray<b3Vector3>		vectors;
-		b3DbvtBenchmark::P14				policy;
+		b3DynamicBvh dbvt;
+		b3AlignedObjectArray<b3Vector3> vectors;
+		b3DbvtBenchmark::P14 policy;
 		vectors.resize(cfgBenchmark14_Iterations);
-		for(int i=0;i<vectors.size();++i)
+		for (int i = 0; i < vectors.size(); ++i)
 		{
-			vectors[i]=(b3DbvtBenchmark::RandVector3()*2-b3Vector3(1,1,1)).normalized();
+			vectors[i] = (b3DbvtBenchmark::RandVector3() * 2 - b3Vector3(1, 1, 1)).normalized();
 		}
-		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
 		policy.m_nodes.reserve(cfgLeaves);
 		printf("[14] culling(OCL+qsort): ");
-		wallclock.reset();	
-		for(int i=0;i<cfgBenchmark14_Iterations;++i)
+		wallclock.reset();
+		for (int i = 0; i < cfgBenchmark14_Iterations; ++i)
 		{
-			static const b3Scalar	offset=0;
+			static const b3Scalar offset = 0;
 			policy.m_nodes.resize(0);
-			dbvt.collideOCL(dbvt.m_root,&vectors[i],&offset,vectors[i],1,policy,false);
+			dbvt.collideOCL(dbvt.m_root, &vectors[i], &offset, vectors[i], 1, policy, false);
 			policy.m_nodes.quickSort(b3DbvtBenchmark::P14::sortfnc);
 		}
-		const int	time=(int)wallclock.getTimeMilliseconds();
-		const int	t=cfgBenchmark14_Iterations;
-		printf("%u ms (%i%%),(%u t/s)\r\n",time,(time-cfgBenchmark14_Reference)*100/time,(t*1000)/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		const int t = cfgBenchmark14_Iterations;
+		printf("%u ms (%i%%),(%u t/s)\r\n", time, (time - cfgBenchmark14_Reference) * 100 / time, (t * 1000) / time);
 	}
-	if(cfgBenchmark15_Enable)
-	{// Benchmark 15	
+	if (cfgBenchmark15_Enable)
+	{  // Benchmark 15
 		srand(380843);
-		b3DynamicBvh								dbvt;
-		b3AlignedObjectArray<b3Vector3>		vectors;
-		b3DbvtBenchmark::P15				policy;
+		b3DynamicBvh dbvt;
+		b3AlignedObjectArray<b3Vector3> vectors;
+		b3DbvtBenchmark::P15 policy;
 		vectors.resize(cfgBenchmark15_Iterations);
-		for(int i=0;i<vectors.size();++i)
+		for (int i = 0; i < vectors.size(); ++i)
 		{
-			vectors[i]=(b3DbvtBenchmark::RandVector3()*2-b3Vector3(1,1,1)).normalized();
+			vectors[i] = (b3DbvtBenchmark::RandVector3() * 2 - b3Vector3(1, 1, 1)).normalized();
 		}
-		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
 		policy.m_nodes.reserve(cfgLeaves);
 		printf("[15] culling(KDOP+qsort): ");
-		wallclock.reset();	
-		for(int i=0;i<cfgBenchmark15_Iterations;++i)
+		wallclock.reset();
+		for (int i = 0; i < cfgBenchmark15_Iterations; ++i)
 		{
-			static const b3Scalar	offset=0;
+			static const b3Scalar offset = 0;
 			policy.m_nodes.resize(0);
-			policy.m_axis=vectors[i];
-			dbvt.collideKDOP(dbvt.m_root,&vectors[i],&offset,1,policy);
+			policy.m_axis = vectors[i];
+			dbvt.collideKDOP(dbvt.m_root, &vectors[i], &offset, 1, policy);
 			policy.m_nodes.quickSort(b3DbvtBenchmark::P15::sortfnc);
 		}
-		const int	time=(int)wallclock.getTimeMilliseconds();
-		const int	t=cfgBenchmark15_Iterations;
-		printf("%u ms (%i%%),(%u t/s)\r\n",time,(time-cfgBenchmark15_Reference)*100/time,(t*1000)/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		const int t = cfgBenchmark15_Iterations;
+		printf("%u ms (%i%%),(%u t/s)\r\n", time, (time - cfgBenchmark15_Reference) * 100 / time, (t * 1000) / time);
 	}
-	if(cfgBenchmark16_Enable)
-	{// Benchmark 16	
+	if (cfgBenchmark16_Enable)
+	{  // Benchmark 16
 		srand(380843);
-		b3DynamicBvh								dbvt;
-		b3AlignedObjectArray<b3DbvtNode*>	batch;
-		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		b3DynamicBvh dbvt;
+		b3AlignedObjectArray<b3DbvtNode*> batch;
+		b3DbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
 		batch.reserve(cfgBenchmark16_BatchCount);
-		printf("[16] insert/remove batch(%u): ",cfgBenchmark16_BatchCount);
+		printf("[16] insert/remove batch(%u): ", cfgBenchmark16_BatchCount);
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark16_Passes;++i)
+		for (int i = 0; i < cfgBenchmark16_Passes; ++i)
 		{
-			for(int j=0;j<cfgBenchmark16_BatchCount;++j)
+			for (int j = 0; j < cfgBenchmark16_BatchCount; ++j)
 			{
-				batch.push_back(dbvt.insert(b3DbvtBenchmark::RandVolume(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale),0));
+				batch.push_back(dbvt.insert(b3DbvtBenchmark::RandVolume(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale), 0));
 			}
-			for(int j=0;j<cfgBenchmark16_BatchCount;++j)
+			for (int j = 0; j < cfgBenchmark16_BatchCount; ++j)
 			{
 				dbvt.remove(batch[j]);
 			}
 			batch.resize(0);
 		}
-		const int	time=(int)wallclock.getTimeMilliseconds();
-		const int	ir=cfgBenchmark16_Passes*cfgBenchmark16_BatchCount;
-		printf("%u ms (%i%%),(%u bir/s)\r\n",time,(time-cfgBenchmark16_Reference)*100/time,int(ir*1000.0/time));
+		const int time = (int)wallclock.getTimeMilliseconds();
+		const int ir = cfgBenchmark16_Passes * cfgBenchmark16_BatchCount;
+		printf("%u ms (%i%%),(%u bir/s)\r\n", time, (time - cfgBenchmark16_Reference) * 100 / time, int(ir * 1000.0 / time));
 	}
-	if(cfgBenchmark17_Enable)
-	{// Benchmark 17
+	if (cfgBenchmark17_Enable)
+	{  // Benchmark 17
 		srand(380843);
-		b3AlignedObjectArray<b3DbvtVolume>	volumes;
-		b3AlignedObjectArray<int>			results;
-		b3AlignedObjectArray<int>			indices;
+		b3AlignedObjectArray<b3DbvtVolume> volumes;
+		b3AlignedObjectArray<int> results;
+		b3AlignedObjectArray<int> indices;
 		volumes.resize(cfgLeaves);
 		results.resize(cfgLeaves);
 		indices.resize(cfgLeaves);
-		for(int i=0;i<cfgLeaves;++i)
+		for (int i = 0; i < cfgLeaves; ++i)
 		{
-			indices[i]=i;
-			volumes[i]=b3DbvtBenchmark::RandVolume(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale);
+			indices[i] = i;
+			volumes[i] = b3DbvtBenchmark::RandVolume(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale);
 		}
-		for(int i=0;i<cfgLeaves;++i)
+		for (int i = 0; i < cfgLeaves; ++i)
 		{
-			b3Swap(indices[i],indices[rand()%cfgLeaves]);
+			b3Swap(indices[i], indices[rand() % cfgLeaves]);
 		}
 		printf("[17] b3DbvtVolume select: ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark17_Iterations;++i)
+		for (int i = 0; i < cfgBenchmark17_Iterations; ++i)
 		{
-			for(int j=0;j<cfgLeaves;++j)
+			for (int j = 0; j < cfgLeaves; ++j)
 			{
-				for(int k=0;k<cfgLeaves;++k)
+				for (int k = 0; k < cfgLeaves; ++k)
 				{
-					const int idx=indices[k];
-					results[idx]=Select(volumes[idx],volumes[j],volumes[k]);
+					const int idx = indices[k];
+					results[idx] = Select(volumes[idx], volumes[j], volumes[k]);
 				}
 			}
 		}
-		const int time=(int)wallclock.getTimeMilliseconds();
-		printf("%u ms (%i%%)\r\n",time,(time-cfgBenchmark17_Reference)*100/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		printf("%u ms (%i%%)\r\n", time, (time - cfgBenchmark17_Reference) * 100 / time);
 	}
 	printf("\r\n\r\n");
 }
diff --git a/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3DynamicBvh.h b/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3DynamicBvh.h
index c004b9130fd..f44e3377fe0 100644
--- a/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3DynamicBvh.h
+++ b/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3DynamicBvh.h
@@ -26,50 +26,49 @@ subject to the following restrictions:
 // Compile time configuration
 //
 
-
 // Implementation profiles
-#define B3_DBVT_IMPL_GENERIC		0	// Generic implementation	
-#define B3_DBVT_IMPL_SSE			1	// SSE
+#define B3_DBVT_IMPL_GENERIC 0  // Generic implementation
+#define B3_DBVT_IMPL_SSE 1      // SSE
 
 // Template implementation of ICollide
 #ifdef _WIN32
-#if (defined (_MSC_VER) && _MSC_VER >= 1400)
-#define	B3_DBVT_USE_TEMPLATE		1
+#if (defined(_MSC_VER) && _MSC_VER >= 1400)
+#define B3_DBVT_USE_TEMPLATE 1
 #else
-#define	B3_DBVT_USE_TEMPLATE		0
+#define B3_DBVT_USE_TEMPLATE 0
 #endif
 #else
-#define	B3_DBVT_USE_TEMPLATE		0
+#define B3_DBVT_USE_TEMPLATE 0
 #endif
 
 // Use only intrinsics instead of inline asm
-#define B3_DBVT_USE_INTRINSIC_SSE	1
+#define B3_DBVT_USE_INTRINSIC_SSE 1
 
 // Using memmov for collideOCL
-#define B3_DBVT_USE_MEMMOVE		1
+#define B3_DBVT_USE_MEMMOVE 1
 
 // Enable benchmarking code
-#define	B3_DBVT_ENABLE_BENCHMARK	0
+#define B3_DBVT_ENABLE_BENCHMARK 0
 
 // Inlining
-#define B3_DBVT_INLINE				B3_FORCE_INLINE
+#define B3_DBVT_INLINE B3_FORCE_INLINE
 
 // Specific methods implementation
 
 //SSE gives errors on a MSVC 7.1
-#if defined (B3_USE_SSE) //&& defined (_WIN32)
-#define B3_DBVT_SELECT_IMPL		B3_DBVT_IMPL_SSE
-#define B3_DBVT_MERGE_IMPL			B3_DBVT_IMPL_SSE
-#define B3_DBVT_INT0_IMPL			B3_DBVT_IMPL_SSE
+#if defined(B3_USE_SSE)  //&& defined (_WIN32)
+#define B3_DBVT_SELECT_IMPL B3_DBVT_IMPL_SSE
+#define B3_DBVT_MERGE_IMPL B3_DBVT_IMPL_SSE
+#define B3_DBVT_INT0_IMPL B3_DBVT_IMPL_SSE
 #else
-#define B3_DBVT_SELECT_IMPL		B3_DBVT_IMPL_GENERIC
-#define B3_DBVT_MERGE_IMPL			B3_DBVT_IMPL_GENERIC
-#define B3_DBVT_INT0_IMPL			B3_DBVT_IMPL_GENERIC
+#define B3_DBVT_SELECT_IMPL B3_DBVT_IMPL_GENERIC
+#define B3_DBVT_MERGE_IMPL B3_DBVT_IMPL_GENERIC
+#define B3_DBVT_INT0_IMPL B3_DBVT_IMPL_GENERIC
 #endif
 
-#if	(B3_DBVT_SELECT_IMPL==B3_DBVT_IMPL_SSE)||	\
-	(B3_DBVT_MERGE_IMPL==B3_DBVT_IMPL_SSE)||	\
-	(B3_DBVT_INT0_IMPL==B3_DBVT_IMPL_SSE)
+#if (B3_DBVT_SELECT_IMPL == B3_DBVT_IMPL_SSE) || \
+	(B3_DBVT_MERGE_IMPL == B3_DBVT_IMPL_SSE) ||  \
+	(B3_DBVT_INT0_IMPL == B3_DBVT_IMPL_SSE)
 #include <emmintrin.h>
 #endif
 
@@ -78,21 +77,24 @@ subject to the following restrictions:
 //
 
 #if B3_DBVT_USE_TEMPLATE
-#define	B3_DBVT_VIRTUAL
+#define B3_DBVT_VIRTUAL
 #define B3_DBVT_VIRTUAL_DTOR(a)
-#define B3_DBVT_PREFIX					template <typename T>
-#define B3_DBVT_IPOLICY				T& policy
-#define B3_DBVT_CHECKTYPE				static const ICollide&	typechecker=*(T*)1;(void)typechecker;
+#define B3_DBVT_PREFIX template <typename T>
+#define B3_DBVT_IPOLICY T& policy
+#define B3_DBVT_CHECKTYPE                        \
+	static const ICollide& typechecker = *(T*)1; \
+	(void)typechecker;
 #else
-#define	B3_DBVT_VIRTUAL_DTOR(a)		virtual ~a() {}
-#define B3_DBVT_VIRTUAL				virtual
+#define B3_DBVT_VIRTUAL_DTOR(a) \
+	virtual ~a() {}
+#define B3_DBVT_VIRTUAL virtual
 #define B3_DBVT_PREFIX
-#define B3_DBVT_IPOLICY				ICollide& policy
+#define B3_DBVT_IPOLICY ICollide& policy
 #define B3_DBVT_CHECKTYPE
 #endif
 
 #if B3_DBVT_USE_MEMMOVE
-#if !defined( __CELLOS_LV2__) && !defined(__MWERKS__)
+#if !defined(__CELLOS_LV2__) && !defined(__MWERKS__)
 #include <memory.h>
 #endif
 #include <string.h>
@@ -126,187 +128,188 @@ subject to the following restrictions:
 // Defaults volumes
 //
 
-/* b3DbvtAabbMm			*/ 
-struct	b3DbvtAabbMm
+/* b3DbvtAabbMm			*/
+struct b3DbvtAabbMm
 {
-	B3_DBVT_INLINE b3Vector3			Center() const	{ return((mi+mx)/2); }
-	B3_DBVT_INLINE b3Vector3			Lengths() const	{ return(mx-mi); }
-	B3_DBVT_INLINE b3Vector3			Extents() const	{ return((mx-mi)/2); }
-	B3_DBVT_INLINE const b3Vector3&	Mins() const	{ return(mi); }
-	B3_DBVT_INLINE const b3Vector3&	Maxs() const	{ return(mx); }
-	static inline b3DbvtAabbMm		FromCE(const b3Vector3& c,const b3Vector3& e);
-	static inline b3DbvtAabbMm		FromCR(const b3Vector3& c,b3Scalar r);
-	static inline b3DbvtAabbMm		FromMM(const b3Vector3& mi,const b3Vector3& mx);
-	static inline b3DbvtAabbMm		FromPoints(const b3Vector3* pts,int n);
-	static inline b3DbvtAabbMm		FromPoints(const b3Vector3** ppts,int n);
-	B3_DBVT_INLINE void				Expand(const b3Vector3& e);
-	B3_DBVT_INLINE void				SignedExpand(const b3Vector3& e);
-	B3_DBVT_INLINE bool				Contain(const b3DbvtAabbMm& a) const;
-	B3_DBVT_INLINE int					Classify(const b3Vector3& n,b3Scalar o,int s) const;
-	B3_DBVT_INLINE b3Scalar			ProjectMinimum(const b3Vector3& v,unsigned signs) const;
-	B3_DBVT_INLINE friend bool			b3Intersect(	const b3DbvtAabbMm& a,
-		const b3DbvtAabbMm& b);
-	
-	B3_DBVT_INLINE friend bool			b3Intersect(	const b3DbvtAabbMm& a,
-		const b3Vector3& b);
+	B3_DBVT_INLINE b3Vector3 Center() const { return ((mi + mx) / 2); }
+	B3_DBVT_INLINE b3Vector3 Lengths() const { return (mx - mi); }
+	B3_DBVT_INLINE b3Vector3 Extents() const { return ((mx - mi) / 2); }
+	B3_DBVT_INLINE const b3Vector3& Mins() const { return (mi); }
+	B3_DBVT_INLINE const b3Vector3& Maxs() const { return (mx); }
+	static inline b3DbvtAabbMm FromCE(const b3Vector3& c, const b3Vector3& e);
+	static inline b3DbvtAabbMm FromCR(const b3Vector3& c, b3Scalar r);
+	static inline b3DbvtAabbMm FromMM(const b3Vector3& mi, const b3Vector3& mx);
+	static inline b3DbvtAabbMm FromPoints(const b3Vector3* pts, int n);
+	static inline b3DbvtAabbMm FromPoints(const b3Vector3** ppts, int n);
+	B3_DBVT_INLINE void Expand(const b3Vector3& e);
+	B3_DBVT_INLINE void SignedExpand(const b3Vector3& e);
+	B3_DBVT_INLINE bool Contain(const b3DbvtAabbMm& a) const;
+	B3_DBVT_INLINE int Classify(const b3Vector3& n, b3Scalar o, int s) const;
+	B3_DBVT_INLINE b3Scalar ProjectMinimum(const b3Vector3& v, unsigned signs) const;
+	B3_DBVT_INLINE friend bool b3Intersect(const b3DbvtAabbMm& a,
+										   const b3DbvtAabbMm& b);
+
+	B3_DBVT_INLINE friend bool b3Intersect(const b3DbvtAabbMm& a,
+										   const b3Vector3& b);
+
+	B3_DBVT_INLINE friend b3Scalar b3Proximity(const b3DbvtAabbMm& a,
+											   const b3DbvtAabbMm& b);
+	B3_DBVT_INLINE friend int b3Select(const b3DbvtAabbMm& o,
+									   const b3DbvtAabbMm& a,
+									   const b3DbvtAabbMm& b);
+	B3_DBVT_INLINE friend void b3Merge(const b3DbvtAabbMm& a,
+									   const b3DbvtAabbMm& b,
+									   b3DbvtAabbMm& r);
+	B3_DBVT_INLINE friend bool b3NotEqual(const b3DbvtAabbMm& a,
+										  const b3DbvtAabbMm& b);
+
+	B3_DBVT_INLINE b3Vector3& tMins() { return (mi); }
+	B3_DBVT_INLINE b3Vector3& tMaxs() { return (mx); }
 
-	B3_DBVT_INLINE friend b3Scalar		b3Proximity(	const b3DbvtAabbMm& a,
-		const b3DbvtAabbMm& b);
-	B3_DBVT_INLINE friend int			b3Select(		const b3DbvtAabbMm& o,
-		const b3DbvtAabbMm& a,
-		const b3DbvtAabbMm& b);
-	B3_DBVT_INLINE friend void			b3Merge(		const b3DbvtAabbMm& a,
-		const b3DbvtAabbMm& b,
-		b3DbvtAabbMm& r);
-	B3_DBVT_INLINE friend bool			b3NotEqual(	const b3DbvtAabbMm& a,
-		const b3DbvtAabbMm& b);
-    
-    B3_DBVT_INLINE b3Vector3&	tMins()	{ return(mi); }
-	B3_DBVT_INLINE b3Vector3&	tMaxs()	{ return(mx); }
-    
 private:
-	B3_DBVT_INLINE void				AddSpan(const b3Vector3& d,b3Scalar& smi,b3Scalar& smx) const;
+	B3_DBVT_INLINE void AddSpan(const b3Vector3& d, b3Scalar& smi, b3Scalar& smx) const;
+
 private:
-	b3Vector3	mi,mx;
+	b3Vector3 mi, mx;
 };
 
-// Types	
-typedef	b3DbvtAabbMm	b3DbvtVolume;
+// Types
+typedef b3DbvtAabbMm b3DbvtVolume;
 
-/* b3DbvtNode				*/ 
-struct	b3DbvtNode
+/* b3DbvtNode				*/
+struct b3DbvtNode
 {
-	b3DbvtVolume	volume;
-	b3DbvtNode*		parent;
-	B3_DBVT_INLINE bool	isleaf() const		{ return(childs[1]==0); }
-	B3_DBVT_INLINE bool	isinternal() const	{ return(!isleaf()); }
-	union
-	{
-		b3DbvtNode*	childs[2];
-		void*	data;
-		int		dataAsInt;
+	b3DbvtVolume volume;
+	b3DbvtNode* parent;
+	B3_DBVT_INLINE bool isleaf() const { return (childs[1] == 0); }
+	B3_DBVT_INLINE bool isinternal() const { return (!isleaf()); }
+	union {
+		b3DbvtNode* childs[2];
+		void* data;
+		int dataAsInt;
 	};
 };
 
 ///The b3DynamicBvh class implements a fast dynamic bounding volume tree based on axis aligned bounding boxes (aabb tree).
 ///This b3DynamicBvh is used for soft body collision detection and for the b3DynamicBvhBroadphase. It has a fast insert, remove and update of nodes.
 ///Unlike the b3QuantizedBvh, nodes can be dynamically moved around, which allows for change in topology of the underlying data structure.
-struct	b3DynamicBvh
+struct b3DynamicBvh
 {
-	/* Stack element	*/ 
-	struct	sStkNN
+	/* Stack element	*/
+	struct sStkNN
 	{
-		const b3DbvtNode*	a;
-		const b3DbvtNode*	b;
+		const b3DbvtNode* a;
+		const b3DbvtNode* b;
 		sStkNN() {}
-		sStkNN(const b3DbvtNode* na,const b3DbvtNode* nb) : a(na),b(nb) {}
+		sStkNN(const b3DbvtNode* na, const b3DbvtNode* nb) : a(na), b(nb) {}
 	};
-	struct	sStkNP
+	struct sStkNP
 	{
-		const b3DbvtNode*	node;
-		int			mask;
-		sStkNP(const b3DbvtNode* n,unsigned m) : node(n),mask(m) {}
+		const b3DbvtNode* node;
+		int mask;
+		sStkNP(const b3DbvtNode* n, unsigned m) : node(n), mask(m) {}
 	};
-	struct	sStkNPS
+	struct sStkNPS
 	{
-		const b3DbvtNode*	node;
-		int			mask;
-		b3Scalar	value;
+		const b3DbvtNode* node;
+		int mask;
+		b3Scalar value;
 		sStkNPS() {}
-		sStkNPS(const b3DbvtNode* n,unsigned m,b3Scalar v) : node(n),mask(m),value(v) {}
+		sStkNPS(const b3DbvtNode* n, unsigned m, b3Scalar v) : node(n), mask(m), value(v) {}
 	};
-	struct	sStkCLN
+	struct sStkCLN
 	{
-		const b3DbvtNode*	node;
-		b3DbvtNode*		parent;
-		sStkCLN(const b3DbvtNode* n,b3DbvtNode* p) : node(n),parent(p) {}
+		const b3DbvtNode* node;
+		b3DbvtNode* parent;
+		sStkCLN(const b3DbvtNode* n, b3DbvtNode* p) : node(n), parent(p) {}
 	};
 	// Policies/Interfaces
 
-	/* ICollide	*/ 
-	struct	ICollide
-	{		
+	/* ICollide	*/
+	struct ICollide
+	{
 		B3_DBVT_VIRTUAL_DTOR(ICollide)
-			B3_DBVT_VIRTUAL void	Process(const b3DbvtNode*,const b3DbvtNode*)		{}
-		B3_DBVT_VIRTUAL void	Process(const b3DbvtNode*)					{}
-		B3_DBVT_VIRTUAL void	Process(const b3DbvtNode* n,b3Scalar)			{ Process(n); }
-		B3_DBVT_VIRTUAL bool	Descent(const b3DbvtNode*)					{ return(true); }
-		B3_DBVT_VIRTUAL bool	AllLeaves(const b3DbvtNode*)					{ return(true); }
+		B3_DBVT_VIRTUAL void Process(const b3DbvtNode*, const b3DbvtNode*) {}
+		B3_DBVT_VIRTUAL void Process(const b3DbvtNode*) {}
+		B3_DBVT_VIRTUAL void Process(const b3DbvtNode* n, b3Scalar) { Process(n); }
+		B3_DBVT_VIRTUAL bool Descent(const b3DbvtNode*) { return (true); }
+		B3_DBVT_VIRTUAL bool AllLeaves(const b3DbvtNode*) { return (true); }
 	};
-	/* IWriter	*/ 
-	struct	IWriter
+	/* IWriter	*/
+	struct IWriter
 	{
 		virtual ~IWriter() {}
-		virtual void		Prepare(const b3DbvtNode* root,int numnodes)=0;
-		virtual void		WriteNode(const b3DbvtNode*,int index,int parent,int child0,int child1)=0;
-		virtual void		WriteLeaf(const b3DbvtNode*,int index,int parent)=0;
+		virtual void Prepare(const b3DbvtNode* root, int numnodes) = 0;
+		virtual void WriteNode(const b3DbvtNode*, int index, int parent, int child0, int child1) = 0;
+		virtual void WriteLeaf(const b3DbvtNode*, int index, int parent) = 0;
 	};
-	/* IClone	*/ 
-	struct	IClone
+	/* IClone	*/
+	struct IClone
 	{
-		virtual ~IClone()	{}
-		virtual void		CloneLeaf(b3DbvtNode*) {}
+		virtual ~IClone() {}
+		virtual void CloneLeaf(b3DbvtNode*) {}
 	};
 
 	// Constants
-	enum	{
-		B3_SIMPLE_STACKSIZE	=	64,
-		B3_DOUBLE_STACKSIZE	=	B3_SIMPLE_STACKSIZE*2
+	enum
+	{
+		B3_SIMPLE_STACKSIZE = 64,
+		B3_DOUBLE_STACKSIZE = B3_SIMPLE_STACKSIZE * 2
 	};
 
 	// Fields
-	b3DbvtNode*		m_root;
-	b3DbvtNode*		m_free;
-	int				m_lkhd;
-	int				m_leaves;
-	unsigned		m_opath;
-
-	
-	b3AlignedObjectArray<sStkNN>	m_stkStack;
-	mutable b3AlignedObjectArray<const b3DbvtNode*>	m_rayTestStack;
+	b3DbvtNode* m_root;
+	b3DbvtNode* m_free;
+	int m_lkhd;
+	int m_leaves;
+	unsigned m_opath;
 
+	b3AlignedObjectArray<sStkNN> m_stkStack;
+	mutable b3AlignedObjectArray<const b3DbvtNode*> m_rayTestStack;
 
 	// Methods
 	b3DynamicBvh();
 	~b3DynamicBvh();
-	void			clear();
-	bool			empty() const { return(0==m_root); }
-	void			optimizeBottomUp();
-	void			optimizeTopDown(int bu_treshold=128);
-	void			optimizeIncremental(int passes);
-	b3DbvtNode*		insert(const b3DbvtVolume& box,void* data);
-	void			update(b3DbvtNode* leaf,int lookahead=-1);
-	void			update(b3DbvtNode* leaf,b3DbvtVolume& volume);
-	bool			update(b3DbvtNode* leaf,b3DbvtVolume& volume,const b3Vector3& velocity,b3Scalar margin);
-	bool			update(b3DbvtNode* leaf,b3DbvtVolume& volume,const b3Vector3& velocity);
-	bool			update(b3DbvtNode* leaf,b3DbvtVolume& volume,b3Scalar margin);	
-	void			remove(b3DbvtNode* leaf);
-	void			write(IWriter* iwriter) const;
-	void			clone(b3DynamicBvh& dest,IClone* iclone=0) const;
-	static int		maxdepth(const b3DbvtNode* node);
-	static int		countLeaves(const b3DbvtNode* node);
-	static void		extractLeaves(const b3DbvtNode* node,b3AlignedObjectArray<const b3DbvtNode*>& leaves);
+	void clear();
+	bool empty() const { return (0 == m_root); }
+	void optimizeBottomUp();
+	void optimizeTopDown(int bu_treshold = 128);
+	void optimizeIncremental(int passes);
+	b3DbvtNode* insert(const b3DbvtVolume& box, void* data);
+	void update(b3DbvtNode* leaf, int lookahead = -1);
+	void update(b3DbvtNode* leaf, b3DbvtVolume& volume);
+	bool update(b3DbvtNode* leaf, b3DbvtVolume& volume, const b3Vector3& velocity, b3Scalar margin);
+	bool update(b3DbvtNode* leaf, b3DbvtVolume& volume, const b3Vector3& velocity);
+	bool update(b3DbvtNode* leaf, b3DbvtVolume& volume, b3Scalar margin);
+	void remove(b3DbvtNode* leaf);
+	void write(IWriter* iwriter) const;
+	void clone(b3DynamicBvh& dest, IClone* iclone = 0) const;
+	static int maxdepth(const b3DbvtNode* node);
+	static int countLeaves(const b3DbvtNode* node);
+	static void extractLeaves(const b3DbvtNode* node, b3AlignedObjectArray<const b3DbvtNode*>& leaves);
 #if B3_DBVT_ENABLE_BENCHMARK
-	static void		benchmark();
+	static void benchmark();
 #else
-	static void		benchmark(){}
+	static void benchmark()
+	{
+	}
 #endif
 	// B3_DBVT_IPOLICY must support ICollide policy/interface
 	B3_DBVT_PREFIX
-		static void		enumNodes(	const b3DbvtNode* root,
-		B3_DBVT_IPOLICY);
+	static void enumNodes(const b3DbvtNode* root,
+						  B3_DBVT_IPOLICY);
 	B3_DBVT_PREFIX
-		static void		enumLeaves(	const b3DbvtNode* root,
-		B3_DBVT_IPOLICY);
+	static void enumLeaves(const b3DbvtNode* root,
+						   B3_DBVT_IPOLICY);
 	B3_DBVT_PREFIX
-		void		collideTT(	const b3DbvtNode* root0,
-		const b3DbvtNode* root1,
-		B3_DBVT_IPOLICY);
+	void collideTT(const b3DbvtNode* root0,
+				   const b3DbvtNode* root1,
+				   B3_DBVT_IPOLICY);
 
 	B3_DBVT_PREFIX
-		void		collideTTpersistentStack(	const b3DbvtNode* root0,
-		  const b3DbvtNode* root1,
-		  B3_DBVT_IPOLICY);
+	void collideTTpersistentStack(const b3DbvtNode* root0,
+								  const b3DbvtNode* root1,
+								  B3_DBVT_IPOLICY);
 #if 0
 	B3_DBVT_PREFIX
 		void		collideTT(	const b3DbvtNode* root0,
@@ -322,71 +325,81 @@ struct	b3DynamicBvh
 #endif
 
 	B3_DBVT_PREFIX
-		void		collideTV(	const b3DbvtNode* root,
-		const b3DbvtVolume& volume,
-		B3_DBVT_IPOLICY) const;
+	void collideTV(const b3DbvtNode* root,
+				   const b3DbvtVolume& volume,
+				   B3_DBVT_IPOLICY) const;
 	///rayTest is a re-entrant ray test, and can be called in parallel as long as the b3AlignedAlloc is thread-safe (uses locking etc)
 	///rayTest is slower than rayTestInternal, because it builds a local stack, using memory allocations, and it recomputes signs/rayDirectionInverses each time
 	B3_DBVT_PREFIX
-		static void		rayTest(	const b3DbvtNode* root,
-		const b3Vector3& rayFrom,
-		const b3Vector3& rayTo,
-		B3_DBVT_IPOLICY);
+	static void rayTest(const b3DbvtNode* root,
+						const b3Vector3& rayFrom,
+						const b3Vector3& rayTo,
+						B3_DBVT_IPOLICY);
 	///rayTestInternal is faster than rayTest, because it uses a persistent stack (to reduce dynamic memory allocations to a minimum) and it uses precomputed signs/rayInverseDirections
 	///rayTestInternal is used by b3DynamicBvhBroadphase to accelerate world ray casts
 	B3_DBVT_PREFIX
-		void		rayTestInternal(	const b3DbvtNode* root,
-								const b3Vector3& rayFrom,
-								const b3Vector3& rayTo,
-								const b3Vector3& rayDirectionInverse,
-								unsigned int signs[3],
-								b3Scalar lambda_max,
-								const b3Vector3& aabbMin,
-								const b3Vector3& aabbMax,
-								B3_DBVT_IPOLICY) const;
+	void rayTestInternal(const b3DbvtNode* root,
+						 const b3Vector3& rayFrom,
+						 const b3Vector3& rayTo,
+						 const b3Vector3& rayDirectionInverse,
+						 unsigned int signs[3],
+						 b3Scalar lambda_max,
+						 const b3Vector3& aabbMin,
+						 const b3Vector3& aabbMax,
+						 B3_DBVT_IPOLICY) const;
 
 	B3_DBVT_PREFIX
-		static void		collideKDOP(const b3DbvtNode* root,
-		const b3Vector3* normals,
-		const b3Scalar* offsets,
-		int count,
-		B3_DBVT_IPOLICY);
+	static void collideKDOP(const b3DbvtNode* root,
+							const b3Vector3* normals,
+							const b3Scalar* offsets,
+							int count,
+							B3_DBVT_IPOLICY);
 	B3_DBVT_PREFIX
-		static void		collideOCL(	const b3DbvtNode* root,
-		const b3Vector3* normals,
-		const b3Scalar* offsets,
-		const b3Vector3& sortaxis,
-		int count,								
-		B3_DBVT_IPOLICY,
-		bool fullsort=true);
+	static void collideOCL(const b3DbvtNode* root,
+						   const b3Vector3* normals,
+						   const b3Scalar* offsets,
+						   const b3Vector3& sortaxis,
+						   int count,
+						   B3_DBVT_IPOLICY,
+						   bool fullsort = true);
 	B3_DBVT_PREFIX
-		static void		collideTU(	const b3DbvtNode* root,
-		B3_DBVT_IPOLICY);
-	// Helpers	
-	static B3_DBVT_INLINE int	nearest(const int* i,const b3DynamicBvh::sStkNPS* a,b3Scalar v,int l,int h)
+	static void collideTU(const b3DbvtNode* root,
+						  B3_DBVT_IPOLICY);
+	// Helpers
+	static B3_DBVT_INLINE int nearest(const int* i, const b3DynamicBvh::sStkNPS* a, b3Scalar v, int l, int h)
 	{
-		int	m=0;
-		while(l<h)
+		int m = 0;
+		while (l < h)
 		{
-			m=(l+h)>>1;
-			if(a[i[m]].value>=v) l=m+1; else h=m;
+			m = (l + h) >> 1;
+			if (a[i[m]].value >= v)
+				l = m + 1;
+			else
+				h = m;
 		}
-		return(h);
+		return (h);
 	}
-	static B3_DBVT_INLINE int	allocate(	b3AlignedObjectArray<int>& ifree,
-		b3AlignedObjectArray<sStkNPS>& stock,
-		const sStkNPS& value)
+	static B3_DBVT_INLINE int allocate(b3AlignedObjectArray<int>& ifree,
+									   b3AlignedObjectArray<sStkNPS>& stock,
+									   const sStkNPS& value)
 	{
-		int	i;
-		if(ifree.size()>0)
-		{ i=ifree[ifree.size()-1];ifree.pop_back();stock[i]=value; }
+		int i;
+		if (ifree.size() > 0)
+		{
+			i = ifree[ifree.size() - 1];
+			ifree.pop_back();
+			stock[i] = value;
+		}
 		else
-		{ i=stock.size();stock.push_back(value); }
-		return(i); 
+		{
+			i = stock.size();
+			stock.push_back(value);
+		}
+		return (i);
 	}
 	//
 private:
-	b3DynamicBvh(const b3DynamicBvh&)	{}	
+	b3DynamicBvh(const b3DynamicBvh&) {}
 };
 
 //
@@ -394,227 +407,252 @@ private:
 //
 
 //
-inline b3DbvtAabbMm			b3DbvtAabbMm::FromCE(const b3Vector3& c,const b3Vector3& e)
+inline b3DbvtAabbMm b3DbvtAabbMm::FromCE(const b3Vector3& c, const b3Vector3& e)
 {
 	b3DbvtAabbMm box;
-	box.mi=c-e;box.mx=c+e;
-	return(box);
+	box.mi = c - e;
+	box.mx = c + e;
+	return (box);
 }
 
 //
-inline b3DbvtAabbMm			b3DbvtAabbMm::FromCR(const b3Vector3& c,b3Scalar r)
+inline b3DbvtAabbMm b3DbvtAabbMm::FromCR(const b3Vector3& c, b3Scalar r)
 {
-	return(FromCE(c,b3MakeVector3(r,r,r)));
+	return (FromCE(c, b3MakeVector3(r, r, r)));
 }
 
 //
-inline b3DbvtAabbMm			b3DbvtAabbMm::FromMM(const b3Vector3& mi,const b3Vector3& mx)
+inline b3DbvtAabbMm b3DbvtAabbMm::FromMM(const b3Vector3& mi, const b3Vector3& mx)
 {
 	b3DbvtAabbMm box;
-	box.mi=mi;box.mx=mx;
-	return(box);
+	box.mi = mi;
+	box.mx = mx;
+	return (box);
 }
 
 //
-inline b3DbvtAabbMm			b3DbvtAabbMm::FromPoints(const b3Vector3* pts,int n)
+inline b3DbvtAabbMm b3DbvtAabbMm::FromPoints(const b3Vector3* pts, int n)
 {
 	b3DbvtAabbMm box;
-	box.mi=box.mx=pts[0];
-	for(int i=1;i<n;++i)
+	box.mi = box.mx = pts[0];
+	for (int i = 1; i < n; ++i)
 	{
 		box.mi.setMin(pts[i]);
 		box.mx.setMax(pts[i]);
 	}
-	return(box);
+	return (box);
 }
 
 //
-inline b3DbvtAabbMm			b3DbvtAabbMm::FromPoints(const b3Vector3** ppts,int n)
+inline b3DbvtAabbMm b3DbvtAabbMm::FromPoints(const b3Vector3** ppts, int n)
 {
 	b3DbvtAabbMm box;
-	box.mi=box.mx=*ppts[0];
-	for(int i=1;i<n;++i)
+	box.mi = box.mx = *ppts[0];
+	for (int i = 1; i < n; ++i)
 	{
 		box.mi.setMin(*ppts[i]);
 		box.mx.setMax(*ppts[i]);
 	}
-	return(box);
+	return (box);
 }
 
 //
-B3_DBVT_INLINE void		b3DbvtAabbMm::Expand(const b3Vector3& e)
+B3_DBVT_INLINE void b3DbvtAabbMm::Expand(const b3Vector3& e)
 {
-	mi-=e;mx+=e;
+	mi -= e;
+	mx += e;
 }
 
 //
-B3_DBVT_INLINE void		b3DbvtAabbMm::SignedExpand(const b3Vector3& e)
+B3_DBVT_INLINE void b3DbvtAabbMm::SignedExpand(const b3Vector3& e)
 {
-	if(e.x>0) mx.setX(mx.x+e[0]); else mi.setX(mi.x+e[0]);
-	if(e.y>0) mx.setY(mx.y+e[1]); else mi.setY(mi.y+e[1]);
-	if(e.z>0) mx.setZ(mx.z+e[2]); else mi.setZ(mi.z+e[2]);
+	if (e.x > 0)
+		mx.setX(mx.x + e[0]);
+	else
+		mi.setX(mi.x + e[0]);
+	if (e.y > 0)
+		mx.setY(mx.y + e[1]);
+	else
+		mi.setY(mi.y + e[1]);
+	if (e.z > 0)
+		mx.setZ(mx.z + e[2]);
+	else
+		mi.setZ(mi.z + e[2]);
 }
 
 //
-B3_DBVT_INLINE bool		b3DbvtAabbMm::Contain(const b3DbvtAabbMm& a) const
+B3_DBVT_INLINE bool b3DbvtAabbMm::Contain(const b3DbvtAabbMm& a) const
 {
-	return(	(mi.x<=a.mi.x)&&
-		(mi.y<=a.mi.y)&&
-		(mi.z<=a.mi.z)&&
-		(mx.x>=a.mx.x)&&
-		(mx.y>=a.mx.y)&&
-		(mx.z>=a.mx.z));
+	return ((mi.x <= a.mi.x) &&
+			(mi.y <= a.mi.y) &&
+			(mi.z <= a.mi.z) &&
+			(mx.x >= a.mx.x) &&
+			(mx.y >= a.mx.y) &&
+			(mx.z >= a.mx.z));
 }
 
 //
-B3_DBVT_INLINE int		b3DbvtAabbMm::Classify(const b3Vector3& n,b3Scalar o,int s) const
+B3_DBVT_INLINE int b3DbvtAabbMm::Classify(const b3Vector3& n, b3Scalar o, int s) const
 {
-	b3Vector3			pi,px;
-	switch(s)
+	b3Vector3 pi, px;
+	switch (s)
 	{
-	case	(0+0+0):	px=b3MakeVector3(mi.x,mi.y,mi.z);
-		pi=b3MakeVector3(mx.x,mx.y,mx.z);break;
-	case	(1+0+0):	px=b3MakeVector3(mx.x,mi.y,mi.z);
-		pi=b3MakeVector3(mi.x,mx.y,mx.z);break;
-	case	(0+2+0):	px=b3MakeVector3(mi.x,mx.y,mi.z);
-		pi=b3MakeVector3(mx.x,mi.y,mx.z);break;
-	case	(1+2+0):	px=b3MakeVector3(mx.x,mx.y,mi.z);
-		pi=b3MakeVector3(mi.x,mi.y,mx.z);break;
-	case	(0+0+4):	px=b3MakeVector3(mi.x,mi.y,mx.z);
-		pi=b3MakeVector3(mx.x,mx.y,mi.z);break;
-	case	(1+0+4):	px=b3MakeVector3(mx.x,mi.y,mx.z);
-		pi=b3MakeVector3(mi.x,mx.y,mi.z);break;
-	case	(0+2+4):	px=b3MakeVector3(mi.x,mx.y,mx.z);
-		pi=b3MakeVector3(mx.x,mi.y,mi.z);break;
-	case	(1+2+4):	px=b3MakeVector3(mx.x,mx.y,mx.z);
-		pi=b3MakeVector3(mi.x,mi.y,mi.z);break;
+		case (0 + 0 + 0):
+			px = b3MakeVector3(mi.x, mi.y, mi.z);
+			pi = b3MakeVector3(mx.x, mx.y, mx.z);
+			break;
+		case (1 + 0 + 0):
+			px = b3MakeVector3(mx.x, mi.y, mi.z);
+			pi = b3MakeVector3(mi.x, mx.y, mx.z);
+			break;
+		case (0 + 2 + 0):
+			px = b3MakeVector3(mi.x, mx.y, mi.z);
+			pi = b3MakeVector3(mx.x, mi.y, mx.z);
+			break;
+		case (1 + 2 + 0):
+			px = b3MakeVector3(mx.x, mx.y, mi.z);
+			pi = b3MakeVector3(mi.x, mi.y, mx.z);
+			break;
+		case (0 + 0 + 4):
+			px = b3MakeVector3(mi.x, mi.y, mx.z);
+			pi = b3MakeVector3(mx.x, mx.y, mi.z);
+			break;
+		case (1 + 0 + 4):
+			px = b3MakeVector3(mx.x, mi.y, mx.z);
+			pi = b3MakeVector3(mi.x, mx.y, mi.z);
+			break;
+		case (0 + 2 + 4):
+			px = b3MakeVector3(mi.x, mx.y, mx.z);
+			pi = b3MakeVector3(mx.x, mi.y, mi.z);
+			break;
+		case (1 + 2 + 4):
+			px = b3MakeVector3(mx.x, mx.y, mx.z);
+			pi = b3MakeVector3(mi.x, mi.y, mi.z);
+			break;
 	}
-	if((b3Dot(n,px)+o)<0)		return(-1);
-	if((b3Dot(n,pi)+o)>=0)	return(+1);
-	return(0);
+	if ((b3Dot(n, px) + o) < 0) return (-1);
+	if ((b3Dot(n, pi) + o) >= 0) return (+1);
+	return (0);
 }
 
 //
-B3_DBVT_INLINE b3Scalar	b3DbvtAabbMm::ProjectMinimum(const b3Vector3& v,unsigned signs) const
+B3_DBVT_INLINE b3Scalar b3DbvtAabbMm::ProjectMinimum(const b3Vector3& v, unsigned signs) const
 {
-	const b3Vector3*	b[]={&mx,&mi};
-	const b3Vector3		p = b3MakeVector3(	b[(signs>>0)&1]->x,
-		b[(signs>>1)&1]->y,
-		b[(signs>>2)&1]->z);
-	return(b3Dot(p,v));
+	const b3Vector3* b[] = {&mx, &mi};
+	const b3Vector3 p = b3MakeVector3(b[(signs >> 0) & 1]->x,
+									  b[(signs >> 1) & 1]->y,
+									  b[(signs >> 2) & 1]->z);
+	return (b3Dot(p, v));
 }
 
 //
-B3_DBVT_INLINE void		b3DbvtAabbMm::AddSpan(const b3Vector3& d,b3Scalar& smi,b3Scalar& smx) const
+B3_DBVT_INLINE void b3DbvtAabbMm::AddSpan(const b3Vector3& d, b3Scalar& smi, b3Scalar& smx) const
 {
-	for(int i=0;i<3;++i)
+	for (int i = 0; i < 3; ++i)
 	{
-		if(d[i]<0)
-		{ smi+=mx[i]*d[i];smx+=mi[i]*d[i]; }
+		if (d[i] < 0)
+		{
+			smi += mx[i] * d[i];
+			smx += mi[i] * d[i];
+		}
 		else
-		{ smi+=mi[i]*d[i];smx+=mx[i]*d[i]; }
+		{
+			smi += mi[i] * d[i];
+			smx += mx[i] * d[i];
+		}
 	}
 }
 
 //
-B3_DBVT_INLINE bool		b3Intersect(	const b3DbvtAabbMm& a,
-								  const b3DbvtAabbMm& b)
+B3_DBVT_INLINE bool b3Intersect(const b3DbvtAabbMm& a,
+								const b3DbvtAabbMm& b)
 {
-#if	B3_DBVT_INT0_IMPL == B3_DBVT_IMPL_SSE
-	const __m128	rt(_mm_or_ps(	_mm_cmplt_ps(_mm_load_ps(b.mx),_mm_load_ps(a.mi)),
-		_mm_cmplt_ps(_mm_load_ps(a.mx),_mm_load_ps(b.mi))));
-#if defined (_WIN32)
-	const __int32*	pu((const __int32*)&rt);
+#if B3_DBVT_INT0_IMPL == B3_DBVT_IMPL_SSE
+	const __m128 rt(_mm_or_ps(_mm_cmplt_ps(_mm_load_ps(b.mx), _mm_load_ps(a.mi)),
+							  _mm_cmplt_ps(_mm_load_ps(a.mx), _mm_load_ps(b.mi))));
+#if defined(_WIN32)
+	const __int32* pu((const __int32*)&rt);
 #else
-    const int*	pu((const int*)&rt);
+	const int* pu((const int*)&rt);
 #endif
-	return((pu[0]|pu[1]|pu[2])==0);
+	return ((pu[0] | pu[1] | pu[2]) == 0);
 #else
-	return(	(a.mi.x<=b.mx.x)&&
-		(a.mx.x>=b.mi.x)&&
-		(a.mi.y<=b.mx.y)&&
-		(a.mx.y>=b.mi.y)&&
-		(a.mi.z<=b.mx.z)&&		
-		(a.mx.z>=b.mi.z));
+	return ((a.mi.x <= b.mx.x) &&
+			(a.mx.x >= b.mi.x) &&
+			(a.mi.y <= b.mx.y) &&
+			(a.mx.y >= b.mi.y) &&
+			(a.mi.z <= b.mx.z) &&
+			(a.mx.z >= b.mi.z));
 #endif
 }
 
-
-
 //
-B3_DBVT_INLINE bool		b3Intersect(	const b3DbvtAabbMm& a,
-								  const b3Vector3& b)
+B3_DBVT_INLINE bool b3Intersect(const b3DbvtAabbMm& a,
+								const b3Vector3& b)
 {
-	return(	(b.x>=a.mi.x)&&
-		(b.y>=a.mi.y)&&
-		(b.z>=a.mi.z)&&
-		(b.x<=a.mx.x)&&
-		(b.y<=a.mx.y)&&
-		(b.z<=a.mx.z));
+	return ((b.x >= a.mi.x) &&
+			(b.y >= a.mi.y) &&
+			(b.z >= a.mi.z) &&
+			(b.x <= a.mx.x) &&
+			(b.y <= a.mx.y) &&
+			(b.z <= a.mx.z));
 }
 
-
-
-
-
 //////////////////////////////////////
 
-
 //
-B3_DBVT_INLINE b3Scalar	b3Proximity(	const b3DbvtAabbMm& a,
-								  const b3DbvtAabbMm& b)
+B3_DBVT_INLINE b3Scalar b3Proximity(const b3DbvtAabbMm& a,
+									const b3DbvtAabbMm& b)
 {
-	const b3Vector3	d=(a.mi+a.mx)-(b.mi+b.mx);
-	return(b3Fabs(d.x)+b3Fabs(d.y)+b3Fabs(d.z));
+	const b3Vector3 d = (a.mi + a.mx) - (b.mi + b.mx);
+	return (b3Fabs(d.x) + b3Fabs(d.y) + b3Fabs(d.z));
 }
 
-
-
 //
-B3_DBVT_INLINE int			b3Select(	const b3DbvtAabbMm& o,
-							   const b3DbvtAabbMm& a,
-							   const b3DbvtAabbMm& b)
+B3_DBVT_INLINE int b3Select(const b3DbvtAabbMm& o,
+							const b3DbvtAabbMm& a,
+							const b3DbvtAabbMm& b)
 {
-#if	B3_DBVT_SELECT_IMPL == B3_DBVT_IMPL_SSE
-    
-#if defined (_WIN32)
-	static B3_ATTRIBUTE_ALIGNED16(const unsigned __int32)	mask[]={0x7fffffff,0x7fffffff,0x7fffffff,0x7fffffff};
+#if B3_DBVT_SELECT_IMPL == B3_DBVT_IMPL_SSE
+
+#if defined(_WIN32)
+	static B3_ATTRIBUTE_ALIGNED16(const unsigned __int32) mask[] = {0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff};
 #else
-    static B3_ATTRIBUTE_ALIGNED16(const unsigned int)	mask[]={0x7fffffff,0x7fffffff,0x7fffffff,0x00000000 /*0x7fffffff*/};
+	static B3_ATTRIBUTE_ALIGNED16(const unsigned int) mask[] = {0x7fffffff, 0x7fffffff, 0x7fffffff, 0x00000000 /*0x7fffffff*/};
 #endif
 	///@todo: the intrinsic version is 11% slower
 #if B3_DBVT_USE_INTRINSIC_SSE
 
-	union b3SSEUnion ///NOTE: if we use more intrinsics, move b3SSEUnion into the LinearMath directory
+	union b3SSEUnion  ///NOTE: if we use more intrinsics, move b3SSEUnion into the LinearMath directory
 	{
-	   __m128		ssereg;
-	   float		floats[4];
-	   int			ints[4];
+		__m128 ssereg;
+		float floats[4];
+		int ints[4];
 	};
 
-	__m128	omi(_mm_load_ps(o.mi));
-	omi=_mm_add_ps(omi,_mm_load_ps(o.mx));
-	__m128	ami(_mm_load_ps(a.mi));
-	ami=_mm_add_ps(ami,_mm_load_ps(a.mx));
-	ami=_mm_sub_ps(ami,omi);
-	ami=_mm_and_ps(ami,_mm_load_ps((const float*)mask));
-	__m128	bmi(_mm_load_ps(b.mi));
-	bmi=_mm_add_ps(bmi,_mm_load_ps(b.mx));
-	bmi=_mm_sub_ps(bmi,omi);
-	bmi=_mm_and_ps(bmi,_mm_load_ps((const float*)mask));
-	__m128	t0(_mm_movehl_ps(ami,ami));
-	ami=_mm_add_ps(ami,t0);
-	ami=_mm_add_ss(ami,_mm_shuffle_ps(ami,ami,1));
-	__m128 t1(_mm_movehl_ps(bmi,bmi));
-	bmi=_mm_add_ps(bmi,t1);
-	bmi=_mm_add_ss(bmi,_mm_shuffle_ps(bmi,bmi,1));
-	
+	__m128 omi(_mm_load_ps(o.mi));
+	omi = _mm_add_ps(omi, _mm_load_ps(o.mx));
+	__m128 ami(_mm_load_ps(a.mi));
+	ami = _mm_add_ps(ami, _mm_load_ps(a.mx));
+	ami = _mm_sub_ps(ami, omi);
+	ami = _mm_and_ps(ami, _mm_load_ps((const float*)mask));
+	__m128 bmi(_mm_load_ps(b.mi));
+	bmi = _mm_add_ps(bmi, _mm_load_ps(b.mx));
+	bmi = _mm_sub_ps(bmi, omi);
+	bmi = _mm_and_ps(bmi, _mm_load_ps((const float*)mask));
+	__m128 t0(_mm_movehl_ps(ami, ami));
+	ami = _mm_add_ps(ami, t0);
+	ami = _mm_add_ss(ami, _mm_shuffle_ps(ami, ami, 1));
+	__m128 t1(_mm_movehl_ps(bmi, bmi));
+	bmi = _mm_add_ps(bmi, t1);
+	bmi = _mm_add_ss(bmi, _mm_shuffle_ps(bmi, bmi, 1));
+
 	b3SSEUnion tmp;
-	tmp.ssereg = _mm_cmple_ss(bmi,ami);
-	return tmp.ints[0]&1;
+	tmp.ssereg = _mm_cmple_ss(bmi, ami);
+	return tmp.ints[0] & 1;
 
 #else
-	B3_ATTRIBUTE_ALIGNED16(__int32	r[1]);
+	B3_ATTRIBUTE_ALIGNED16(__int32 r[1]);
 	__asm
 	{
 		mov		eax,o
@@ -642,46 +680,52 @@ B3_DBVT_INLINE int			b3Select(	const b3DbvtAabbMm& o,
 			cmpless	xmm2,xmm1
 			movss	r,xmm2
 	}
-	return(r[0]&1);
+	return (r[0] & 1);
 #endif
 #else
-	return(b3Proximity(o,a)<b3Proximity(o,b)?0:1);
+	return (b3Proximity(o, a) < b3Proximity(o, b) ? 0 : 1);
 #endif
 }
 
 //
-B3_DBVT_INLINE void		b3Merge(	const b3DbvtAabbMm& a,
-							  const b3DbvtAabbMm& b,
-							  b3DbvtAabbMm& r)
+B3_DBVT_INLINE void b3Merge(const b3DbvtAabbMm& a,
+							const b3DbvtAabbMm& b,
+							b3DbvtAabbMm& r)
 {
-#if B3_DBVT_MERGE_IMPL==B3_DBVT_IMPL_SSE
-	__m128	ami(_mm_load_ps(a.mi));
-	__m128	amx(_mm_load_ps(a.mx));
-	__m128	bmi(_mm_load_ps(b.mi));
-	__m128	bmx(_mm_load_ps(b.mx));
-	ami=_mm_min_ps(ami,bmi);
-	amx=_mm_max_ps(amx,bmx);
-	_mm_store_ps(r.mi,ami);
-	_mm_store_ps(r.mx,amx);
+#if B3_DBVT_MERGE_IMPL == B3_DBVT_IMPL_SSE
+	__m128 ami(_mm_load_ps(a.mi));
+	__m128 amx(_mm_load_ps(a.mx));
+	__m128 bmi(_mm_load_ps(b.mi));
+	__m128 bmx(_mm_load_ps(b.mx));
+	ami = _mm_min_ps(ami, bmi);
+	amx = _mm_max_ps(amx, bmx);
+	_mm_store_ps(r.mi, ami);
+	_mm_store_ps(r.mx, amx);
 #else
-	for(int i=0;i<3;++i)
+	for (int i = 0; i < 3; ++i)
 	{
-		if(a.mi[i]<b.mi[i]) r.mi[i]=a.mi[i]; else r.mi[i]=b.mi[i];
-		if(a.mx[i]>b.mx[i]) r.mx[i]=a.mx[i]; else r.mx[i]=b.mx[i];
+		if (a.mi[i] < b.mi[i])
+			r.mi[i] = a.mi[i];
+		else
+			r.mi[i] = b.mi[i];
+		if (a.mx[i] > b.mx[i])
+			r.mx[i] = a.mx[i];
+		else
+			r.mx[i] = b.mx[i];
 	}
 #endif
 }
 
 //
-B3_DBVT_INLINE bool		b3NotEqual(	const b3DbvtAabbMm& a,
-								 const b3DbvtAabbMm& b)
+B3_DBVT_INLINE bool b3NotEqual(const b3DbvtAabbMm& a,
+							   const b3DbvtAabbMm& b)
 {
-	return(	(a.mi.x!=b.mi.x)||
-		(a.mi.y!=b.mi.y)||
-		(a.mi.z!=b.mi.z)||
-		(a.mx.x!=b.mx.x)||
-		(a.mx.y!=b.mx.y)||
-		(a.mx.z!=b.mx.z));
+	return ((a.mi.x != b.mi.x) ||
+			(a.mi.y != b.mi.y) ||
+			(a.mi.z != b.mi.z) ||
+			(a.mx.x != b.mx.x) ||
+			(a.mx.y != b.mx.y) ||
+			(a.mx.z != b.mx.z));
 }
 
 //
@@ -690,162 +734,162 @@ B3_DBVT_INLINE bool		b3NotEqual(	const b3DbvtAabbMm& a,
 
 //
 B3_DBVT_PREFIX
-inline void		b3DynamicBvh::enumNodes(	const b3DbvtNode* root,
-								  B3_DBVT_IPOLICY)
+inline void b3DynamicBvh::enumNodes(const b3DbvtNode* root,
+									B3_DBVT_IPOLICY)
 {
 	B3_DBVT_CHECKTYPE
-		policy.Process(root);
-	if(root->isinternal())
+	policy.Process(root);
+	if (root->isinternal())
 	{
-		enumNodes(root->childs[0],policy);
-		enumNodes(root->childs[1],policy);
+		enumNodes(root->childs[0], policy);
+		enumNodes(root->childs[1], policy);
 	}
 }
 
 //
 B3_DBVT_PREFIX
-inline void		b3DynamicBvh::enumLeaves(	const b3DbvtNode* root,
-								   B3_DBVT_IPOLICY)
+inline void b3DynamicBvh::enumLeaves(const b3DbvtNode* root,
+									 B3_DBVT_IPOLICY)
 {
 	B3_DBVT_CHECKTYPE
-		if(root->isinternal())
-		{
-			enumLeaves(root->childs[0],policy);
-			enumLeaves(root->childs[1],policy);
-		}
-		else
-		{
-			policy.Process(root);
-		}
+	if (root->isinternal())
+	{
+		enumLeaves(root->childs[0], policy);
+		enumLeaves(root->childs[1], policy);
+	}
+	else
+	{
+		policy.Process(root);
+	}
 }
 
 //
 B3_DBVT_PREFIX
-inline void		b3DynamicBvh::collideTT(	const b3DbvtNode* root0,
-								  const b3DbvtNode* root1,
-								  B3_DBVT_IPOLICY)
+inline void b3DynamicBvh::collideTT(const b3DbvtNode* root0,
+									const b3DbvtNode* root1,
+									B3_DBVT_IPOLICY)
 {
 	B3_DBVT_CHECKTYPE
-		if(root0&&root1)
+	if (root0 && root1)
+	{
+		int depth = 1;
+		int treshold = B3_DOUBLE_STACKSIZE - 4;
+		b3AlignedObjectArray<sStkNN> stkStack;
+		stkStack.resize(B3_DOUBLE_STACKSIZE);
+		stkStack[0] = sStkNN(root0, root1);
+		do
 		{
-			int								depth=1;
-			int								treshold=B3_DOUBLE_STACKSIZE-4;
-			b3AlignedObjectArray<sStkNN>	stkStack;
-			stkStack.resize(B3_DOUBLE_STACKSIZE);
-			stkStack[0]=sStkNN(root0,root1);
-			do	{		
-				sStkNN	p=stkStack[--depth];
-				if(depth>treshold)
+			sStkNN p = stkStack[--depth];
+			if (depth > treshold)
+			{
+				stkStack.resize(stkStack.size() * 2);
+				treshold = stkStack.size() - 4;
+			}
+			if (p.a == p.b)
+			{
+				if (p.a->isinternal())
 				{
-					stkStack.resize(stkStack.size()*2);
-					treshold=stkStack.size()-4;
+					stkStack[depth++] = sStkNN(p.a->childs[0], p.a->childs[0]);
+					stkStack[depth++] = sStkNN(p.a->childs[1], p.a->childs[1]);
+					stkStack[depth++] = sStkNN(p.a->childs[0], p.a->childs[1]);
 				}
-				if(p.a==p.b)
+			}
+			else if (b3Intersect(p.a->volume, p.b->volume))
+			{
+				if (p.a->isinternal())
 				{
-					if(p.a->isinternal())
+					if (p.b->isinternal())
 					{
-						stkStack[depth++]=sStkNN(p.a->childs[0],p.a->childs[0]);
-						stkStack[depth++]=sStkNN(p.a->childs[1],p.a->childs[1]);
-						stkStack[depth++]=sStkNN(p.a->childs[0],p.a->childs[1]);
-					}
-				}
-				else if(b3Intersect(p.a->volume,p.b->volume))
-				{
-					if(p.a->isinternal())
-					{
-						if(p.b->isinternal())
-						{
-							stkStack[depth++]=sStkNN(p.a->childs[0],p.b->childs[0]);
-							stkStack[depth++]=sStkNN(p.a->childs[1],p.b->childs[0]);
-							stkStack[depth++]=sStkNN(p.a->childs[0],p.b->childs[1]);
-							stkStack[depth++]=sStkNN(p.a->childs[1],p.b->childs[1]);
-						}
-						else
-						{
-							stkStack[depth++]=sStkNN(p.a->childs[0],p.b);
-							stkStack[depth++]=sStkNN(p.a->childs[1],p.b);
-						}
+						stkStack[depth++] = sStkNN(p.a->childs[0], p.b->childs[0]);
+						stkStack[depth++] = sStkNN(p.a->childs[1], p.b->childs[0]);
+						stkStack[depth++] = sStkNN(p.a->childs[0], p.b->childs[1]);
+						stkStack[depth++] = sStkNN(p.a->childs[1], p.b->childs[1]);
 					}
 					else
 					{
-						if(p.b->isinternal())
-						{
-							stkStack[depth++]=sStkNN(p.a,p.b->childs[0]);
-							stkStack[depth++]=sStkNN(p.a,p.b->childs[1]);
-						}
-						else
-						{
-							policy.Process(p.a,p.b);
-						}
+						stkStack[depth++] = sStkNN(p.a->childs[0], p.b);
+						stkStack[depth++] = sStkNN(p.a->childs[1], p.b);
 					}
 				}
-			} while(depth);
-		}
+				else
+				{
+					if (p.b->isinternal())
+					{
+						stkStack[depth++] = sStkNN(p.a, p.b->childs[0]);
+						stkStack[depth++] = sStkNN(p.a, p.b->childs[1]);
+					}
+					else
+					{
+						policy.Process(p.a, p.b);
+					}
+				}
+			}
+		} while (depth);
+	}
 }
 
-
-
 B3_DBVT_PREFIX
-inline void		b3DynamicBvh::collideTTpersistentStack(	const b3DbvtNode* root0,
-								  const b3DbvtNode* root1,
-								  B3_DBVT_IPOLICY)
+inline void b3DynamicBvh::collideTTpersistentStack(const b3DbvtNode* root0,
+												   const b3DbvtNode* root1,
+												   B3_DBVT_IPOLICY)
 {
 	B3_DBVT_CHECKTYPE
-		if(root0&&root1)
+	if (root0 && root1)
+	{
+		int depth = 1;
+		int treshold = B3_DOUBLE_STACKSIZE - 4;
+
+		m_stkStack.resize(B3_DOUBLE_STACKSIZE);
+		m_stkStack[0] = sStkNN(root0, root1);
+		do
 		{
-			int								depth=1;
-			int								treshold=B3_DOUBLE_STACKSIZE-4;
-			
-			m_stkStack.resize(B3_DOUBLE_STACKSIZE);
-			m_stkStack[0]=sStkNN(root0,root1);
-			do	{		
-				sStkNN	p=m_stkStack[--depth];
-				if(depth>treshold)
+			sStkNN p = m_stkStack[--depth];
+			if (depth > treshold)
+			{
+				m_stkStack.resize(m_stkStack.size() * 2);
+				treshold = m_stkStack.size() - 4;
+			}
+			if (p.a == p.b)
+			{
+				if (p.a->isinternal())
 				{
-					m_stkStack.resize(m_stkStack.size()*2);
-					treshold=m_stkStack.size()-4;
+					m_stkStack[depth++] = sStkNN(p.a->childs[0], p.a->childs[0]);
+					m_stkStack[depth++] = sStkNN(p.a->childs[1], p.a->childs[1]);
+					m_stkStack[depth++] = sStkNN(p.a->childs[0], p.a->childs[1]);
 				}
-				if(p.a==p.b)
+			}
+			else if (b3Intersect(p.a->volume, p.b->volume))
+			{
+				if (p.a->isinternal())
 				{
-					if(p.a->isinternal())
+					if (p.b->isinternal())
 					{
-						m_stkStack[depth++]=sStkNN(p.a->childs[0],p.a->childs[0]);
-						m_stkStack[depth++]=sStkNN(p.a->childs[1],p.a->childs[1]);
-						m_stkStack[depth++]=sStkNN(p.a->childs[0],p.a->childs[1]);
-					}
-				}
-				else if(b3Intersect(p.a->volume,p.b->volume))
-				{
-					if(p.a->isinternal())
-					{
-						if(p.b->isinternal())
-						{
-							m_stkStack[depth++]=sStkNN(p.a->childs[0],p.b->childs[0]);
-							m_stkStack[depth++]=sStkNN(p.a->childs[1],p.b->childs[0]);
-							m_stkStack[depth++]=sStkNN(p.a->childs[0],p.b->childs[1]);
-							m_stkStack[depth++]=sStkNN(p.a->childs[1],p.b->childs[1]);
-						}
-						else
-						{
-							m_stkStack[depth++]=sStkNN(p.a->childs[0],p.b);
-							m_stkStack[depth++]=sStkNN(p.a->childs[1],p.b);
-						}
+						m_stkStack[depth++] = sStkNN(p.a->childs[0], p.b->childs[0]);
+						m_stkStack[depth++] = sStkNN(p.a->childs[1], p.b->childs[0]);
+						m_stkStack[depth++] = sStkNN(p.a->childs[0], p.b->childs[1]);
+						m_stkStack[depth++] = sStkNN(p.a->childs[1], p.b->childs[1]);
 					}
 					else
 					{
-						if(p.b->isinternal())
-						{
-							m_stkStack[depth++]=sStkNN(p.a,p.b->childs[0]);
-							m_stkStack[depth++]=sStkNN(p.a,p.b->childs[1]);
-						}
-						else
-						{
-							policy.Process(p.a,p.b);
-						}
+						m_stkStack[depth++] = sStkNN(p.a->childs[0], p.b);
+						m_stkStack[depth++] = sStkNN(p.a->childs[1], p.b);
 					}
 				}
-			} while(depth);
-		}
+				else
+				{
+					if (p.b->isinternal())
+					{
+						m_stkStack[depth++] = sStkNN(p.a, p.b->childs[0]);
+						m_stkStack[depth++] = sStkNN(p.a, p.b->childs[1]);
+					}
+					else
+					{
+						policy.Process(p.a, p.b);
+					}
+				}
+			}
+		} while (depth);
+	}
 }
 
 #if 0
@@ -915,337 +959,356 @@ inline void		b3DynamicBvh::collideTT(	const b3DbvtNode* root0,
 	const b3Transform	xform=xform0.inverse()*xform1;
 	collideTT(root0,root1,xform,policy);
 }
-#endif 
+#endif
 
 //
 B3_DBVT_PREFIX
-inline void		b3DynamicBvh::collideTV(	const b3DbvtNode* root,
-								  const b3DbvtVolume& vol,
-								  B3_DBVT_IPOLICY) const
+inline void b3DynamicBvh::collideTV(const b3DbvtNode* root,
+									const b3DbvtVolume& vol,
+									B3_DBVT_IPOLICY) const
 {
 	B3_DBVT_CHECKTYPE
-		if(root)
+	if (root)
+	{
+		B3_ATTRIBUTE_ALIGNED16(b3DbvtVolume)
+		volume(vol);
+		b3AlignedObjectArray<const b3DbvtNode*> stack;
+		stack.resize(0);
+		stack.reserve(B3_SIMPLE_STACKSIZE);
+		stack.push_back(root);
+		do
 		{
-			B3_ATTRIBUTE_ALIGNED16(b3DbvtVolume)		volume(vol);
-			b3AlignedObjectArray<const b3DbvtNode*>	stack;
-			stack.resize(0);
-			stack.reserve(B3_SIMPLE_STACKSIZE);
-			stack.push_back(root);
-			do	{
-				const b3DbvtNode*	n=stack[stack.size()-1];
-				stack.pop_back();
-				if(b3Intersect(n->volume,volume))
+			const b3DbvtNode* n = stack[stack.size() - 1];
+			stack.pop_back();
+			if (b3Intersect(n->volume, volume))
+			{
+				if (n->isinternal())
 				{
-					if(n->isinternal())
-					{
-						stack.push_back(n->childs[0]);
-						stack.push_back(n->childs[1]);
-					}
-					else
-					{
-						policy.Process(n);
-					}
+					stack.push_back(n->childs[0]);
+					stack.push_back(n->childs[1]);
 				}
-			} while(stack.size()>0);
-		}
+				else
+				{
+					policy.Process(n);
+				}
+			}
+		} while (stack.size() > 0);
+	}
 }
 
 B3_DBVT_PREFIX
-inline void		b3DynamicBvh::rayTestInternal(	const b3DbvtNode* root,
-								const b3Vector3& rayFrom,
-								const b3Vector3& rayTo,
-								const b3Vector3& rayDirectionInverse,
-								unsigned int signs[3],
-								b3Scalar lambda_max,
-								const b3Vector3& aabbMin,
-								const b3Vector3& aabbMax,
-								B3_DBVT_IPOLICY) const
+inline void b3DynamicBvh::rayTestInternal(const b3DbvtNode* root,
+										  const b3Vector3& rayFrom,
+										  const b3Vector3& rayTo,
+										  const b3Vector3& rayDirectionInverse,
+										  unsigned int signs[3],
+										  b3Scalar lambda_max,
+										  const b3Vector3& aabbMin,
+										  const b3Vector3& aabbMax,
+										  B3_DBVT_IPOLICY) const
 {
-        (void) rayTo;
+	(void)rayTo;
 	B3_DBVT_CHECKTYPE
-	if(root)
+	if (root)
 	{
-		int								depth=1;
-		int								treshold=B3_DOUBLE_STACKSIZE-2;
-		b3AlignedObjectArray<const b3DbvtNode*>&	stack = m_rayTestStack;
+		int depth = 1;
+		int treshold = B3_DOUBLE_STACKSIZE - 2;
+		b3AlignedObjectArray<const b3DbvtNode*>& stack = m_rayTestStack;
 		stack.resize(B3_DOUBLE_STACKSIZE);
-		stack[0]=root;
+		stack[0] = root;
 		b3Vector3 bounds[2];
-		do	
+		do
 		{
-			const b3DbvtNode*	node=stack[--depth];
-			bounds[0] = node->volume.Mins()-aabbMax;
-			bounds[1] = node->volume.Maxs()-aabbMin;
-			b3Scalar tmin=1.f,lambda_min=0.f;
-			unsigned int result1=false;
-			result1 = b3RayAabb2(rayFrom,rayDirectionInverse,signs,bounds,tmin,lambda_min,lambda_max);
-			if(result1)
+			const b3DbvtNode* node = stack[--depth];
+			bounds[0] = node->volume.Mins() - aabbMax;
+			bounds[1] = node->volume.Maxs() - aabbMin;
+			b3Scalar tmin = 1.f, lambda_min = 0.f;
+			unsigned int result1 = false;
+			result1 = b3RayAabb2(rayFrom, rayDirectionInverse, signs, bounds, tmin, lambda_min, lambda_max);
+			if (result1)
 			{
-				if(node->isinternal())
+				if (node->isinternal())
 				{
-					if(depth>treshold)
+					if (depth > treshold)
 					{
-						stack.resize(stack.size()*2);
-						treshold=stack.size()-2;
+						stack.resize(stack.size() * 2);
+						treshold = stack.size() - 2;
 					}
-					stack[depth++]=node->childs[0];
-					stack[depth++]=node->childs[1];
+					stack[depth++] = node->childs[0];
+					stack[depth++] = node->childs[1];
 				}
 				else
 				{
 					policy.Process(node);
 				}
 			}
-		} while(depth);
+		} while (depth);
 	}
 }
 
 //
 B3_DBVT_PREFIX
-inline void		b3DynamicBvh::rayTest(	const b3DbvtNode* root,
-								const b3Vector3& rayFrom,
-								const b3Vector3& rayTo,
-								B3_DBVT_IPOLICY)
-{
-	B3_DBVT_CHECKTYPE
-		if(root)
-		{
-			b3Vector3 rayDir = (rayTo-rayFrom);
-			rayDir.normalize ();
-
-			///what about division by zero? --> just set rayDirection[i] to INF/B3_LARGE_FLOAT
-			b3Vector3 rayDirectionInverse;
-			rayDirectionInverse[0] = rayDir[0] == b3Scalar(0.0) ? b3Scalar(B3_LARGE_FLOAT) : b3Scalar(1.0) / rayDir[0];
-			rayDirectionInverse[1] = rayDir[1] == b3Scalar(0.0) ? b3Scalar(B3_LARGE_FLOAT) : b3Scalar(1.0) / rayDir[1];
-			rayDirectionInverse[2] = rayDir[2] == b3Scalar(0.0) ? b3Scalar(B3_LARGE_FLOAT) : b3Scalar(1.0) / rayDir[2];
-			unsigned int signs[3] = { rayDirectionInverse[0] < 0.0, rayDirectionInverse[1] < 0.0, rayDirectionInverse[2] < 0.0};
-
-			b3Scalar lambda_max = rayDir.dot(rayTo-rayFrom);
-#ifdef COMPARE_BTRAY_AABB2
-			b3Vector3 resultNormal;
-#endif//COMPARE_BTRAY_AABB2
-			
-			b3AlignedObjectArray<const b3DbvtNode*>	stack;
-
-			int								depth=1;
-			int								treshold=B3_DOUBLE_STACKSIZE-2;
-
-			stack.resize(B3_DOUBLE_STACKSIZE);
-			stack[0]=root;
-			b3Vector3 bounds[2];
-			do	{
-				const b3DbvtNode*	node=stack[--depth];
-
-				bounds[0] = node->volume.Mins();
-				bounds[1] = node->volume.Maxs();
-				
-				b3Scalar tmin=1.f,lambda_min=0.f;
-				unsigned int result1 = b3RayAabb2(rayFrom,rayDirectionInverse,signs,bounds,tmin,lambda_min,lambda_max);
-
-#ifdef COMPARE_BTRAY_AABB2
-				b3Scalar param=1.f;
-				bool result2 = b3RayAabb(rayFrom,rayTo,node->volume.Mins(),node->volume.Maxs(),param,resultNormal);
-				b3Assert(result1 == result2);
-#endif //TEST_BTRAY_AABB2
-
-				if(result1)
-				{
-					if(node->isinternal())
-					{
-						if(depth>treshold)
-						{
-							stack.resize(stack.size()*2);
-							treshold=stack.size()-2;
-						}
-						stack[depth++]=node->childs[0];
-						stack[depth++]=node->childs[1];
-					}
-					else
-					{
-						policy.Process(node);
-					}
-				}
-			} while(depth);
-
-		}
-}
-
-//
-B3_DBVT_PREFIX
-inline void		b3DynamicBvh::collideKDOP(const b3DbvtNode* root,
-									const b3Vector3* normals,
-									const b3Scalar* offsets,
-									int count,
-									B3_DBVT_IPOLICY)
-{
-	B3_DBVT_CHECKTYPE
-		if(root)
-		{
-			const int						inside=(1<<count)-1;
-			b3AlignedObjectArray<sStkNP>	stack;
-			int								signs[sizeof(unsigned)*8];
-			b3Assert(count<int (sizeof(signs)/sizeof(signs[0])));
-			for(int i=0;i<count;++i)
-			{
-				signs[i]=	((normals[i].x>=0)?1:0)+
-					((normals[i].y>=0)?2:0)+
-					((normals[i].z>=0)?4:0);
-			}
-			stack.reserve(B3_SIMPLE_STACKSIZE);
-			stack.push_back(sStkNP(root,0));
-			do	{
-				sStkNP	se=stack[stack.size()-1];
-				bool	out=false;
-				stack.pop_back();
-				for(int i=0,j=1;(!out)&&(i<count);++i,j<<=1)
-				{
-					if(0==(se.mask&j))
-					{
-						const int	side=se.node->volume.Classify(normals[i],offsets[i],signs[i]);
-						switch(side)
-						{
-						case	-1:	out=true;break;
-						case	+1:	se.mask|=j;break;
-						}
-					}
-				}
-				if(!out)
-				{
-					if((se.mask!=inside)&&(se.node->isinternal()))
-					{
-						stack.push_back(sStkNP(se.node->childs[0],se.mask));
-						stack.push_back(sStkNP(se.node->childs[1],se.mask));
-					}
-					else
-					{
-						if(policy.AllLeaves(se.node)) enumLeaves(se.node,policy);
-					}
-				}
-			} while(stack.size());
-		}
-}
-
-//
-B3_DBVT_PREFIX
-inline void		b3DynamicBvh::collideOCL(	const b3DbvtNode* root,
-								   const b3Vector3* normals,
-								   const b3Scalar* offsets,
-								   const b3Vector3& sortaxis,
-								   int count,
-								   B3_DBVT_IPOLICY,
-								   bool fsort)
-{
-	B3_DBVT_CHECKTYPE
-		if(root)
-		{
-			const unsigned					srtsgns=(sortaxis[0]>=0?1:0)+
-				(sortaxis[1]>=0?2:0)+
-				(sortaxis[2]>=0?4:0);
-			const int						inside=(1<<count)-1;
-			b3AlignedObjectArray<sStkNPS>	stock;
-			b3AlignedObjectArray<int>		ifree;
-			b3AlignedObjectArray<int>		stack;
-			int								signs[sizeof(unsigned)*8];
-			b3Assert(count<int (sizeof(signs)/sizeof(signs[0])));
-			for(int i=0;i<count;++i)
-			{
-				signs[i]=	((normals[i].x>=0)?1:0)+
-					((normals[i].y>=0)?2:0)+
-					((normals[i].z>=0)?4:0);
-			}
-			stock.reserve(B3_SIMPLE_STACKSIZE);
-			stack.reserve(B3_SIMPLE_STACKSIZE);
-			ifree.reserve(B3_SIMPLE_STACKSIZE);
-			stack.push_back(allocate(ifree,stock,sStkNPS(root,0,root->volume.ProjectMinimum(sortaxis,srtsgns))));
-			do	{
-				const int	id=stack[stack.size()-1];
-				sStkNPS		se=stock[id];
-				stack.pop_back();ifree.push_back(id);
-				if(se.mask!=inside)
-				{
-					bool	out=false;
-					for(int i=0,j=1;(!out)&&(i<count);++i,j<<=1)
-					{
-						if(0==(se.mask&j))
-						{
-							const int	side=se.node->volume.Classify(normals[i],offsets[i],signs[i]);
-							switch(side)
-							{
-							case	-1:	out=true;break;
-							case	+1:	se.mask|=j;break;
-							}
-						}
-					}
-					if(out) continue;
-				}
-				if(policy.Descent(se.node))
-				{
-					if(se.node->isinternal())
-					{
-						const b3DbvtNode* pns[]={	se.node->childs[0],se.node->childs[1]};
-						sStkNPS		nes[]={	sStkNPS(pns[0],se.mask,pns[0]->volume.ProjectMinimum(sortaxis,srtsgns)),
-							sStkNPS(pns[1],se.mask,pns[1]->volume.ProjectMinimum(sortaxis,srtsgns))};
-						const int	q=nes[0].value<nes[1].value?1:0;				
-						int			j=stack.size();
-						if(fsort&&(j>0))
-						{
-							/* Insert 0	*/ 
-							j=nearest(&stack[0],&stock[0],nes[q].value,0,stack.size());
-							stack.push_back(0);
-#if B3_DBVT_USE_MEMMOVE
-							memmove(&stack[j+1],&stack[j],sizeof(int)*(stack.size()-j-1));
-#else
-							for(int k=stack.size()-1;k>j;--k) stack[k]=stack[k-1];
-#endif
-							stack[j]=allocate(ifree,stock,nes[q]);
-							/* Insert 1	*/ 
-							j=nearest(&stack[0],&stock[0],nes[1-q].value,j,stack.size());
-							stack.push_back(0);
-#if B3_DBVT_USE_MEMMOVE
-							memmove(&stack[j+1],&stack[j],sizeof(int)*(stack.size()-j-1));
-#else
-							for(int k=stack.size()-1;k>j;--k) stack[k]=stack[k-1];
-#endif
-							stack[j]=allocate(ifree,stock,nes[1-q]);
-						}
-						else
-						{
-							stack.push_back(allocate(ifree,stock,nes[q]));
-							stack.push_back(allocate(ifree,stock,nes[1-q]));
-						}
-					}
-					else
-					{
-						policy.Process(se.node,se.value);
-					}
-				}
-			} while(stack.size());
-		}
-}
-
-//
-B3_DBVT_PREFIX
-inline void		b3DynamicBvh::collideTU(	const b3DbvtNode* root,
+inline void b3DynamicBvh::rayTest(const b3DbvtNode* root,
+								  const b3Vector3& rayFrom,
+								  const b3Vector3& rayTo,
 								  B3_DBVT_IPOLICY)
 {
 	B3_DBVT_CHECKTYPE
-		if(root)
+	if (root)
+	{
+		b3Vector3 rayDir = (rayTo - rayFrom);
+		rayDir.normalize();
+
+		///what about division by zero? --> just set rayDirection[i] to INF/B3_LARGE_FLOAT
+		b3Vector3 rayDirectionInverse;
+		rayDirectionInverse[0] = rayDir[0] == b3Scalar(0.0) ? b3Scalar(B3_LARGE_FLOAT) : b3Scalar(1.0) / rayDir[0];
+		rayDirectionInverse[1] = rayDir[1] == b3Scalar(0.0) ? b3Scalar(B3_LARGE_FLOAT) : b3Scalar(1.0) / rayDir[1];
+		rayDirectionInverse[2] = rayDir[2] == b3Scalar(0.0) ? b3Scalar(B3_LARGE_FLOAT) : b3Scalar(1.0) / rayDir[2];
+		unsigned int signs[3] = {rayDirectionInverse[0] < 0.0, rayDirectionInverse[1] < 0.0, rayDirectionInverse[2] < 0.0};
+
+		b3Scalar lambda_max = rayDir.dot(rayTo - rayFrom);
+#ifdef COMPARE_BTRAY_AABB2
+		b3Vector3 resultNormal;
+#endif  //COMPARE_BTRAY_AABB2
+
+		b3AlignedObjectArray<const b3DbvtNode*> stack;
+
+		int depth = 1;
+		int treshold = B3_DOUBLE_STACKSIZE - 2;
+
+		stack.resize(B3_DOUBLE_STACKSIZE);
+		stack[0] = root;
+		b3Vector3 bounds[2];
+		do
 		{
-			b3AlignedObjectArray<const b3DbvtNode*>	stack;
-			stack.reserve(B3_SIMPLE_STACKSIZE);
-			stack.push_back(root);
-			do	{
-				const b3DbvtNode*	n=stack[stack.size()-1];
-				stack.pop_back();
-				if(policy.Descent(n))
+			const b3DbvtNode* node = stack[--depth];
+
+			bounds[0] = node->volume.Mins();
+			bounds[1] = node->volume.Maxs();
+
+			b3Scalar tmin = 1.f, lambda_min = 0.f;
+			unsigned int result1 = b3RayAabb2(rayFrom, rayDirectionInverse, signs, bounds, tmin, lambda_min, lambda_max);
+
+#ifdef COMPARE_BTRAY_AABB2
+			b3Scalar param = 1.f;
+			bool result2 = b3RayAabb(rayFrom, rayTo, node->volume.Mins(), node->volume.Maxs(), param, resultNormal);
+			b3Assert(result1 == result2);
+#endif  //TEST_BTRAY_AABB2
+
+			if (result1)
+			{
+				if (node->isinternal())
 				{
-					if(n->isinternal())
-					{ stack.push_back(n->childs[0]);stack.push_back(n->childs[1]); }
-					else
-					{ policy.Process(n); }
+					if (depth > treshold)
+					{
+						stack.resize(stack.size() * 2);
+						treshold = stack.size() - 2;
+					}
+					stack[depth++] = node->childs[0];
+					stack[depth++] = node->childs[1];
 				}
-			} while(stack.size()>0);
+				else
+				{
+					policy.Process(node);
+				}
+			}
+		} while (depth);
+	}
+}
+
+//
+B3_DBVT_PREFIX
+inline void b3DynamicBvh::collideKDOP(const b3DbvtNode* root,
+									  const b3Vector3* normals,
+									  const b3Scalar* offsets,
+									  int count,
+									  B3_DBVT_IPOLICY)
+{
+	B3_DBVT_CHECKTYPE
+	if (root)
+	{
+		const int inside = (1 << count) - 1;
+		b3AlignedObjectArray<sStkNP> stack;
+		int signs[sizeof(unsigned) * 8];
+		b3Assert(count < int(sizeof(signs) / sizeof(signs[0])));
+		for (int i = 0; i < count; ++i)
+		{
+			signs[i] = ((normals[i].x >= 0) ? 1 : 0) +
+					   ((normals[i].y >= 0) ? 2 : 0) +
+					   ((normals[i].z >= 0) ? 4 : 0);
 		}
+		stack.reserve(B3_SIMPLE_STACKSIZE);
+		stack.push_back(sStkNP(root, 0));
+		do
+		{
+			sStkNP se = stack[stack.size() - 1];
+			bool out = false;
+			stack.pop_back();
+			for (int i = 0, j = 1; (!out) && (i < count); ++i, j <<= 1)
+			{
+				if (0 == (se.mask & j))
+				{
+					const int side = se.node->volume.Classify(normals[i], offsets[i], signs[i]);
+					switch (side)
+					{
+						case -1:
+							out = true;
+							break;
+						case +1:
+							se.mask |= j;
+							break;
+					}
+				}
+			}
+			if (!out)
+			{
+				if ((se.mask != inside) && (se.node->isinternal()))
+				{
+					stack.push_back(sStkNP(se.node->childs[0], se.mask));
+					stack.push_back(sStkNP(se.node->childs[1], se.mask));
+				}
+				else
+				{
+					if (policy.AllLeaves(se.node)) enumLeaves(se.node, policy);
+				}
+			}
+		} while (stack.size());
+	}
+}
+
+//
+B3_DBVT_PREFIX
+inline void b3DynamicBvh::collideOCL(const b3DbvtNode* root,
+									 const b3Vector3* normals,
+									 const b3Scalar* offsets,
+									 const b3Vector3& sortaxis,
+									 int count,
+									 B3_DBVT_IPOLICY,
+									 bool fsort)
+{
+	B3_DBVT_CHECKTYPE
+	if (root)
+	{
+		const unsigned srtsgns = (sortaxis[0] >= 0 ? 1 : 0) +
+								 (sortaxis[1] >= 0 ? 2 : 0) +
+								 (sortaxis[2] >= 0 ? 4 : 0);
+		const int inside = (1 << count) - 1;
+		b3AlignedObjectArray<sStkNPS> stock;
+		b3AlignedObjectArray<int> ifree;
+		b3AlignedObjectArray<int> stack;
+		int signs[sizeof(unsigned) * 8];
+		b3Assert(count < int(sizeof(signs) / sizeof(signs[0])));
+		for (int i = 0; i < count; ++i)
+		{
+			signs[i] = ((normals[i].x >= 0) ? 1 : 0) +
+					   ((normals[i].y >= 0) ? 2 : 0) +
+					   ((normals[i].z >= 0) ? 4 : 0);
+		}
+		stock.reserve(B3_SIMPLE_STACKSIZE);
+		stack.reserve(B3_SIMPLE_STACKSIZE);
+		ifree.reserve(B3_SIMPLE_STACKSIZE);
+		stack.push_back(allocate(ifree, stock, sStkNPS(root, 0, root->volume.ProjectMinimum(sortaxis, srtsgns))));
+		do
+		{
+			const int id = stack[stack.size() - 1];
+			sStkNPS se = stock[id];
+			stack.pop_back();
+			ifree.push_back(id);
+			if (se.mask != inside)
+			{
+				bool out = false;
+				for (int i = 0, j = 1; (!out) && (i < count); ++i, j <<= 1)
+				{
+					if (0 == (se.mask & j))
+					{
+						const int side = se.node->volume.Classify(normals[i], offsets[i], signs[i]);
+						switch (side)
+						{
+							case -1:
+								out = true;
+								break;
+							case +1:
+								se.mask |= j;
+								break;
+						}
+					}
+				}
+				if (out) continue;
+			}
+			if (policy.Descent(se.node))
+			{
+				if (se.node->isinternal())
+				{
+					const b3DbvtNode* pns[] = {se.node->childs[0], se.node->childs[1]};
+					sStkNPS nes[] = {sStkNPS(pns[0], se.mask, pns[0]->volume.ProjectMinimum(sortaxis, srtsgns)),
+									 sStkNPS(pns[1], se.mask, pns[1]->volume.ProjectMinimum(sortaxis, srtsgns))};
+					const int q = nes[0].value < nes[1].value ? 1 : 0;
+					int j = stack.size();
+					if (fsort && (j > 0))
+					{
+						/* Insert 0	*/
+						j = nearest(&stack[0], &stock[0], nes[q].value, 0, stack.size());
+						stack.push_back(0);
+#if B3_DBVT_USE_MEMMOVE
+						memmove(&stack[j + 1], &stack[j], sizeof(int) * (stack.size() - j - 1));
+#else
+						for (int k = stack.size() - 1; k > j; --k) stack[k] = stack[k - 1];
+#endif
+						stack[j] = allocate(ifree, stock, nes[q]);
+						/* Insert 1	*/
+						j = nearest(&stack[0], &stock[0], nes[1 - q].value, j, stack.size());
+						stack.push_back(0);
+#if B3_DBVT_USE_MEMMOVE
+						memmove(&stack[j + 1], &stack[j], sizeof(int) * (stack.size() - j - 1));
+#else
+						for (int k = stack.size() - 1; k > j; --k) stack[k] = stack[k - 1];
+#endif
+						stack[j] = allocate(ifree, stock, nes[1 - q]);
+					}
+					else
+					{
+						stack.push_back(allocate(ifree, stock, nes[q]));
+						stack.push_back(allocate(ifree, stock, nes[1 - q]));
+					}
+				}
+				else
+				{
+					policy.Process(se.node, se.value);
+				}
+			}
+		} while (stack.size());
+	}
+}
+
+//
+B3_DBVT_PREFIX
+inline void b3DynamicBvh::collideTU(const b3DbvtNode* root,
+									B3_DBVT_IPOLICY)
+{
+	B3_DBVT_CHECKTYPE
+	if (root)
+	{
+		b3AlignedObjectArray<const b3DbvtNode*> stack;
+		stack.reserve(B3_SIMPLE_STACKSIZE);
+		stack.push_back(root);
+		do
+		{
+			const b3DbvtNode* n = stack[stack.size() - 1];
+			stack.pop_back();
+			if (policy.Descent(n))
+			{
+				if (n->isinternal())
+				{
+					stack.push_back(n->childs[0]);
+					stack.push_back(n->childs[1]);
+				}
+				else
+				{
+					policy.Process(n);
+				}
+			}
+		} while (stack.size() > 0);
+	}
 }
 
 //
diff --git a/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3DynamicBvhBroadphase.cpp b/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3DynamicBvhBroadphase.cpp
index bc150955b83..dea2ddb0f2f 100644
--- a/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3DynamicBvhBroadphase.cpp
+++ b/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3DynamicBvhBroadphase.cpp
@@ -22,28 +22,27 @@ subject to the following restrictions:
 // Profiling
 //
 
-#if B3_DBVT_BP_PROFILE||B3_DBVT_BP_ENABLE_BENCHMARK
+#if B3_DBVT_BP_PROFILE || B3_DBVT_BP_ENABLE_BENCHMARK
 #include <stdio.h>
 #endif
 
 #if B3_DBVT_BP_PROFILE
-struct	b3ProfileScope
+struct b3ProfileScope
 {
-	__forceinline b3ProfileScope(b3Clock& clock,unsigned long& value) :
-	m_clock(&clock),m_value(&value),m_base(clock.getTimeMicroseconds())
+	__forceinline b3ProfileScope(b3Clock& clock, unsigned long& value) : m_clock(&clock), m_value(&value), m_base(clock.getTimeMicroseconds())
 	{
 	}
 	__forceinline ~b3ProfileScope()
 	{
-		(*m_value)+=m_clock->getTimeMicroseconds()-m_base;
+		(*m_value) += m_clock->getTimeMicroseconds() - m_base;
 	}
-	b3Clock*		m_clock;
-	unsigned long*	m_value;
-	unsigned long	m_base;
+	b3Clock* m_clock;
+	unsigned long* m_value;
+	unsigned long m_base;
 };
-#define	b3SPC(_value_)	b3ProfileScope	spc_scope(m_clock,_value_)
+#define b3SPC(_value_) b3ProfileScope spc_scope(m_clock, _value_)
 #else
-#define	b3SPC(_value_)
+#define b3SPC(_value_)
 #endif
 
 //
@@ -52,66 +51,75 @@ struct	b3ProfileScope
 
 //
 template <typename T>
-static inline void	b3ListAppend(T* item,T*& list)
+static inline void b3ListAppend(T* item, T*& list)
 {
-	item->links[0]=0;
-	item->links[1]=list;
-	if(list) list->links[0]=item;
-	list=item;
+	item->links[0] = 0;
+	item->links[1] = list;
+	if (list) list->links[0] = item;
+	list = item;
 }
 
 //
 template <typename T>
-static inline void	b3ListRemove(T* item,T*& list)
+static inline void b3ListRemove(T* item, T*& list)
 {
-	if(item->links[0]) item->links[0]->links[1]=item->links[1]; else list=item->links[1];
-	if(item->links[1]) item->links[1]->links[0]=item->links[0];
+	if (item->links[0])
+		item->links[0]->links[1] = item->links[1];
+	else
+		list = item->links[1];
+	if (item->links[1]) item->links[1]->links[0] = item->links[0];
 }
 
 //
 template <typename T>
-static inline int	b3ListCount(T* root)
+static inline int b3ListCount(T* root)
 {
-	int	n=0;
-	while(root) { ++n;root=root->links[1]; }
-	return(n);
+	int n = 0;
+	while (root)
+	{
+		++n;
+		root = root->links[1];
+	}
+	return (n);
 }
 
 //
 template <typename T>
-static inline void	b3Clear(T& value)
+static inline void b3Clear(T& value)
 {
-	static const struct ZeroDummy : T {} zerodummy;
-	value=zerodummy;
+	static const struct ZeroDummy : T
+	{
+	} zerodummy;
+	value = zerodummy;
 }
 
 //
 // Colliders
 //
 
-/* Tree collider	*/ 
-struct	b3DbvtTreeCollider : b3DynamicBvh::ICollide
+/* Tree collider	*/
+struct b3DbvtTreeCollider : b3DynamicBvh::ICollide
 {
-	b3DynamicBvhBroadphase*	pbp;
-	b3DbvtProxy*		proxy;
+	b3DynamicBvhBroadphase* pbp;
+	b3DbvtProxy* proxy;
 	b3DbvtTreeCollider(b3DynamicBvhBroadphase* p) : pbp(p) {}
-	void	Process(const b3DbvtNode* na,const b3DbvtNode* nb)
+	void Process(const b3DbvtNode* na, const b3DbvtNode* nb)
 	{
-		if(na!=nb)
+		if (na != nb)
 		{
-			b3DbvtProxy*	pa=(b3DbvtProxy*)na->data;
-			b3DbvtProxy*	pb=(b3DbvtProxy*)nb->data;
+			b3DbvtProxy* pa = (b3DbvtProxy*)na->data;
+			b3DbvtProxy* pb = (b3DbvtProxy*)nb->data;
 #if B3_DBVT_BP_SORTPAIRS
-			if(pa->m_uniqueId>pb->m_uniqueId) 
-				b3Swap(pa,pb);
+			if (pa->m_uniqueId > pb->m_uniqueId)
+				b3Swap(pa, pb);
 #endif
-			pbp->m_paircache->addOverlappingPair(pa->getUid(),pb->getUid());
+			pbp->m_paircache->addOverlappingPair(pa->getUid(), pb->getUid());
 			++pbp->m_newpairs;
 		}
 	}
-	void	Process(const b3DbvtNode* n)
+	void Process(const b3DbvtNode* n)
 	{
-		Process(n,proxy->leaf);
+		Process(n, proxy->leaf);
 	}
 };
 
@@ -122,26 +130,26 @@ struct	b3DbvtTreeCollider : b3DynamicBvh::ICollide
 //
 b3DynamicBvhBroadphase::b3DynamicBvhBroadphase(int proxyCapacity, b3OverlappingPairCache* paircache)
 {
-	m_deferedcollide	=	false;
-	m_needcleanup		=	true;
-	m_releasepaircache	=	(paircache!=0)?false:true;
-	m_prediction		=	0;
-	m_stageCurrent		=	0;
-	m_fixedleft			=	0;
-	m_fupdates			=	1;
-	m_dupdates			=	0;
-	m_cupdates			=	10;
-	m_newpairs			=	1;
-	m_updates_call		=	0;
-	m_updates_done		=	0;
-	m_updates_ratio		=	0;
-	m_paircache			=	paircache? paircache	: new(b3AlignedAlloc(sizeof(b3HashedOverlappingPairCache),16)) b3HashedOverlappingPairCache();
-	
-	m_pid				=	0;
-	m_cid				=	0;
-	for(int i=0;i<=STAGECOUNT;++i)
+	m_deferedcollide = false;
+	m_needcleanup = true;
+	m_releasepaircache = (paircache != 0) ? false : true;
+	m_prediction = 0;
+	m_stageCurrent = 0;
+	m_fixedleft = 0;
+	m_fupdates = 1;
+	m_dupdates = 0;
+	m_cupdates = 10;
+	m_newpairs = 1;
+	m_updates_call = 0;
+	m_updates_done = 0;
+	m_updates_ratio = 0;
+	m_paircache = paircache ? paircache : new (b3AlignedAlloc(sizeof(b3HashedOverlappingPairCache), 16)) b3HashedOverlappingPairCache();
+
+	m_pid = 0;
+	m_cid = 0;
+	for (int i = 0; i <= STAGECOUNT; ++i)
 	{
-		m_stageRoots[i]=0;
+		m_stageRoots[i] = 0;
 	}
 #if B3_DBVT_BP_PROFILE
 	b3Clear(m_profiling);
@@ -152,7 +160,7 @@ b3DynamicBvhBroadphase::b3DynamicBvhBroadphase(int proxyCapacity, b3OverlappingP
 //
 b3DynamicBvhBroadphase::~b3DynamicBvhBroadphase()
 {
-	if(m_releasepaircache) 
+	if (m_releasepaircache)
 	{
 		m_paircache->~b3OverlappingPairCache();
 		b3AlignedFree(m_paircache);
@@ -160,53 +168,53 @@ b3DynamicBvhBroadphase::~b3DynamicBvhBroadphase()
 }
 
 //
-b3BroadphaseProxy*				b3DynamicBvhBroadphase::createProxy(	const b3Vector3& aabbMin,
-															  const b3Vector3& aabbMax,
-															  int objectId,
-															  void* userPtr,
-															   int collisionFilterGroup,
-																int collisionFilterMask)
+b3BroadphaseProxy* b3DynamicBvhBroadphase::createProxy(const b3Vector3& aabbMin,
+													   const b3Vector3& aabbMax,
+													   int objectId,
+													   void* userPtr,
+													   int collisionFilterGroup,
+													   int collisionFilterMask)
 {
 	b3DbvtProxy* mem = &m_proxies[objectId];
-	b3DbvtProxy*		proxy=new(mem) b3DbvtProxy(	aabbMin,aabbMax,userPtr,
-		collisionFilterGroup,
-		collisionFilterMask);
+	b3DbvtProxy* proxy = new (mem) b3DbvtProxy(aabbMin, aabbMax, userPtr,
+											   collisionFilterGroup,
+											   collisionFilterMask);
 
-	b3DbvtAabbMm aabb = b3DbvtVolume::FromMM(aabbMin,aabbMax);
+	b3DbvtAabbMm aabb = b3DbvtVolume::FromMM(aabbMin, aabbMax);
 
 	//bproxy->aabb			=	b3DbvtVolume::FromMM(aabbMin,aabbMax);
-	proxy->stage		=	m_stageCurrent;
-	proxy->m_uniqueId	=	objectId;
-	proxy->leaf			=	m_sets[0].insert(aabb,proxy);
-	b3ListAppend(proxy,m_stageRoots[m_stageCurrent]);
-	if(!m_deferedcollide)
+	proxy->stage = m_stageCurrent;
+	proxy->m_uniqueId = objectId;
+	proxy->leaf = m_sets[0].insert(aabb, proxy);
+	b3ListAppend(proxy, m_stageRoots[m_stageCurrent]);
+	if (!m_deferedcollide)
 	{
-		b3DbvtTreeCollider	collider(this);
-		collider.proxy=proxy;
-		m_sets[0].collideTV(m_sets[0].m_root,aabb,collider);
-		m_sets[1].collideTV(m_sets[1].m_root,aabb,collider);
+		b3DbvtTreeCollider collider(this);
+		collider.proxy = proxy;
+		m_sets[0].collideTV(m_sets[0].m_root, aabb, collider);
+		m_sets[1].collideTV(m_sets[1].m_root, aabb, collider);
 	}
-	return(proxy);
+	return (proxy);
 }
 
 //
-void							b3DynamicBvhBroadphase::destroyProxy(	b3BroadphaseProxy* absproxy,
-															   b3Dispatcher* dispatcher)
+void b3DynamicBvhBroadphase::destroyProxy(b3BroadphaseProxy* absproxy,
+										  b3Dispatcher* dispatcher)
 {
-	b3DbvtProxy*	proxy=(b3DbvtProxy*)absproxy;
-	if(proxy->stage==STAGECOUNT)
+	b3DbvtProxy* proxy = (b3DbvtProxy*)absproxy;
+	if (proxy->stage == STAGECOUNT)
 		m_sets[1].remove(proxy->leaf);
 	else
 		m_sets[0].remove(proxy->leaf);
-	b3ListRemove(proxy,m_stageRoots[proxy->stage]);
-	m_paircache->removeOverlappingPairsContainingProxy(proxy->getUid(),dispatcher);
-	
-	m_needcleanup=true;
+	b3ListRemove(proxy, m_stageRoots[proxy->stage]);
+	m_paircache->removeOverlappingPairsContainingProxy(proxy->getUid(), dispatcher);
+
+	m_needcleanup = true;
 }
 
-void	b3DynamicBvhBroadphase::getAabb(int objectId,b3Vector3& aabbMin, b3Vector3& aabbMax ) const
+void b3DynamicBvhBroadphase::getAabb(int objectId, b3Vector3& aabbMin, b3Vector3& aabbMax) const
 {
-	const b3DbvtProxy*						proxy=&m_proxies[objectId];
+	const b3DbvtProxy* proxy = &m_proxies[objectId];
 	aabbMin = proxy->m_aabbMin;
 	aabbMax = proxy->m_aabbMax;
 }
@@ -219,235 +227,223 @@ void	b3DynamicBvhBroadphase::getAabb(b3BroadphaseProxy* absproxy,b3Vector3& aabb
 }
 */
 
-
-struct	BroadphaseRayTester : b3DynamicBvh::ICollide
+struct BroadphaseRayTester : b3DynamicBvh::ICollide
 {
 	b3BroadphaseRayCallback& m_rayCallback;
 	BroadphaseRayTester(b3BroadphaseRayCallback& orgCallback)
-		:m_rayCallback(orgCallback)
+		: m_rayCallback(orgCallback)
 	{
 	}
-	void					Process(const b3DbvtNode* leaf)
+	void Process(const b3DbvtNode* leaf)
 	{
-		b3DbvtProxy*	proxy=(b3DbvtProxy*)leaf->data;
+		b3DbvtProxy* proxy = (b3DbvtProxy*)leaf->data;
 		m_rayCallback.process(proxy);
 	}
-};	
+};
 
-void	b3DynamicBvhBroadphase::rayTest(const b3Vector3& rayFrom,const b3Vector3& rayTo, b3BroadphaseRayCallback& rayCallback,const b3Vector3& aabbMin,const b3Vector3& aabbMax)
+void b3DynamicBvhBroadphase::rayTest(const b3Vector3& rayFrom, const b3Vector3& rayTo, b3BroadphaseRayCallback& rayCallback, const b3Vector3& aabbMin, const b3Vector3& aabbMax)
 {
 	BroadphaseRayTester callback(rayCallback);
 
-	m_sets[0].rayTestInternal(	m_sets[0].m_root,
-		rayFrom,
-		rayTo,
-		rayCallback.m_rayDirectionInverse,
-		rayCallback.m_signs,
-		rayCallback.m_lambda_max,
-		aabbMin,
-		aabbMax,
-		callback);
-
-	m_sets[1].rayTestInternal(	m_sets[1].m_root,
-		rayFrom,
-		rayTo,
-		rayCallback.m_rayDirectionInverse,
-		rayCallback.m_signs,
-		rayCallback.m_lambda_max,
-		aabbMin,
-		aabbMax,
-		callback);
+	m_sets[0].rayTestInternal(m_sets[0].m_root,
+							  rayFrom,
+							  rayTo,
+							  rayCallback.m_rayDirectionInverse,
+							  rayCallback.m_signs,
+							  rayCallback.m_lambda_max,
+							  aabbMin,
+							  aabbMax,
+							  callback);
 
+	m_sets[1].rayTestInternal(m_sets[1].m_root,
+							  rayFrom,
+							  rayTo,
+							  rayCallback.m_rayDirectionInverse,
+							  rayCallback.m_signs,
+							  rayCallback.m_lambda_max,
+							  aabbMin,
+							  aabbMax,
+							  callback);
 }
 
-
-struct	BroadphaseAabbTester : b3DynamicBvh::ICollide
+struct BroadphaseAabbTester : b3DynamicBvh::ICollide
 {
 	b3BroadphaseAabbCallback& m_aabbCallback;
 	BroadphaseAabbTester(b3BroadphaseAabbCallback& orgCallback)
-		:m_aabbCallback(orgCallback)
+		: m_aabbCallback(orgCallback)
 	{
 	}
-	void					Process(const b3DbvtNode* leaf)
+	void Process(const b3DbvtNode* leaf)
 	{
-		b3DbvtProxy*	proxy=(b3DbvtProxy*)leaf->data;
+		b3DbvtProxy* proxy = (b3DbvtProxy*)leaf->data;
 		m_aabbCallback.process(proxy);
 	}
-};	
+};
 
-void	b3DynamicBvhBroadphase::aabbTest(const b3Vector3& aabbMin,const b3Vector3& aabbMax,b3BroadphaseAabbCallback& aabbCallback)
+void b3DynamicBvhBroadphase::aabbTest(const b3Vector3& aabbMin, const b3Vector3& aabbMax, b3BroadphaseAabbCallback& aabbCallback)
 {
 	BroadphaseAabbTester callback(aabbCallback);
 
-	const B3_ATTRIBUTE_ALIGNED16(b3DbvtVolume)	bounds=b3DbvtVolume::FromMM(aabbMin,aabbMax);
-		//process all children, that overlap with  the given AABB bounds
-	m_sets[0].collideTV(m_sets[0].m_root,bounds,callback);
-	m_sets[1].collideTV(m_sets[1].m_root,bounds,callback);
-
+	const B3_ATTRIBUTE_ALIGNED16(b3DbvtVolume) bounds = b3DbvtVolume::FromMM(aabbMin, aabbMax);
+	//process all children, that overlap with  the given AABB bounds
+	m_sets[0].collideTV(m_sets[0].m_root, bounds, callback);
+	m_sets[1].collideTV(m_sets[1].m_root, bounds, callback);
 }
 
-
-
 //
-void							b3DynamicBvhBroadphase::setAabb(int objectId,
-														  const b3Vector3& aabbMin,
-														  const b3Vector3& aabbMax,
-														  b3Dispatcher* /*dispatcher*/)
+void b3DynamicBvhBroadphase::setAabb(int objectId,
+									 const b3Vector3& aabbMin,
+									 const b3Vector3& aabbMax,
+									 b3Dispatcher* /*dispatcher*/)
 {
-	b3DbvtProxy*						proxy=&m_proxies[objectId];
-//	b3DbvtProxy*						proxy=(b3DbvtProxy*)absproxy;
-	B3_ATTRIBUTE_ALIGNED16(b3DbvtVolume)	aabb=b3DbvtVolume::FromMM(aabbMin,aabbMax);
+	b3DbvtProxy* proxy = &m_proxies[objectId];
+	//	b3DbvtProxy*						proxy=(b3DbvtProxy*)absproxy;
+	B3_ATTRIBUTE_ALIGNED16(b3DbvtVolume)
+	aabb = b3DbvtVolume::FromMM(aabbMin, aabbMax);
 #if B3_DBVT_BP_PREVENTFALSEUPDATE
-	if(b3NotEqual(aabb,proxy->leaf->volume))
+	if (b3NotEqual(aabb, proxy->leaf->volume))
 #endif
 	{
-		bool	docollide=false;
-		if(proxy->stage==STAGECOUNT)
-		{/* fixed -> dynamic set	*/ 
+		bool docollide = false;
+		if (proxy->stage == STAGECOUNT)
+		{ /* fixed -> dynamic set	*/
 			m_sets[1].remove(proxy->leaf);
-			proxy->leaf=m_sets[0].insert(aabb,proxy);
-			docollide=true;
+			proxy->leaf = m_sets[0].insert(aabb, proxy);
+			docollide = true;
 		}
 		else
-		{/* dynamic set				*/ 
+		{ /* dynamic set				*/
 			++m_updates_call;
-			if(b3Intersect(proxy->leaf->volume,aabb))
-			{/* Moving				*/ 
+			if (b3Intersect(proxy->leaf->volume, aabb))
+			{ /* Moving				*/
 
-				const b3Vector3	delta=aabbMin-proxy->m_aabbMin;
-				b3Vector3		velocity(((proxy->m_aabbMax-proxy->m_aabbMin)/2)*m_prediction);
-				if(delta[0]<0) velocity[0]=-velocity[0];
-				if(delta[1]<0) velocity[1]=-velocity[1];
-				if(delta[2]<0) velocity[2]=-velocity[2];
-				if	(
-#ifdef B3_DBVT_BP_MARGIN				
-					m_sets[0].update(proxy->leaf,aabb,velocity,B3_DBVT_BP_MARGIN)
+				const b3Vector3 delta = aabbMin - proxy->m_aabbMin;
+				b3Vector3 velocity(((proxy->m_aabbMax - proxy->m_aabbMin) / 2) * m_prediction);
+				if (delta[0] < 0) velocity[0] = -velocity[0];
+				if (delta[1] < 0) velocity[1] = -velocity[1];
+				if (delta[2] < 0) velocity[2] = -velocity[2];
+				if (
+#ifdef B3_DBVT_BP_MARGIN
+					m_sets[0].update(proxy->leaf, aabb, velocity, B3_DBVT_BP_MARGIN)
 #else
-					m_sets[0].update(proxy->leaf,aabb,velocity)
+					m_sets[0].update(proxy->leaf, aabb, velocity)
 #endif
-					)
+				)
 				{
 					++m_updates_done;
-					docollide=true;
+					docollide = true;
 				}
 			}
 			else
-			{/* Teleporting			*/ 
-				m_sets[0].update(proxy->leaf,aabb);
+			{ /* Teleporting			*/
+				m_sets[0].update(proxy->leaf, aabb);
 				++m_updates_done;
-				docollide=true;
-			}	
+				docollide = true;
+			}
 		}
-		b3ListRemove(proxy,m_stageRoots[proxy->stage]);
+		b3ListRemove(proxy, m_stageRoots[proxy->stage]);
 		proxy->m_aabbMin = aabbMin;
 		proxy->m_aabbMax = aabbMax;
-		proxy->stage	=	m_stageCurrent;
-		b3ListAppend(proxy,m_stageRoots[m_stageCurrent]);
-		if(docollide)
+		proxy->stage = m_stageCurrent;
+		b3ListAppend(proxy, m_stageRoots[m_stageCurrent]);
+		if (docollide)
 		{
-			m_needcleanup=true;
-			if(!m_deferedcollide)
+			m_needcleanup = true;
+			if (!m_deferedcollide)
 			{
-				b3DbvtTreeCollider	collider(this);
-				m_sets[1].collideTTpersistentStack(m_sets[1].m_root,proxy->leaf,collider);
-				m_sets[0].collideTTpersistentStack(m_sets[0].m_root,proxy->leaf,collider);
+				b3DbvtTreeCollider collider(this);
+				m_sets[1].collideTTpersistentStack(m_sets[1].m_root, proxy->leaf, collider);
+				m_sets[0].collideTTpersistentStack(m_sets[0].m_root, proxy->leaf, collider);
 			}
-		}	
+		}
 	}
 }
 
-
 //
-void							b3DynamicBvhBroadphase::setAabbForceUpdate(		b3BroadphaseProxy* absproxy,
-														  const b3Vector3& aabbMin,
-														  const b3Vector3& aabbMax,
-														  b3Dispatcher* /*dispatcher*/)
+void b3DynamicBvhBroadphase::setAabbForceUpdate(b3BroadphaseProxy* absproxy,
+												const b3Vector3& aabbMin,
+												const b3Vector3& aabbMax,
+												b3Dispatcher* /*dispatcher*/)
 {
-	b3DbvtProxy*						proxy=(b3DbvtProxy*)absproxy;
-	B3_ATTRIBUTE_ALIGNED16(b3DbvtVolume)	aabb=b3DbvtVolume::FromMM(aabbMin,aabbMax);
-	bool	docollide=false;
-	if(proxy->stage==STAGECOUNT)
-	{/* fixed -> dynamic set	*/ 
+	b3DbvtProxy* proxy = (b3DbvtProxy*)absproxy;
+	B3_ATTRIBUTE_ALIGNED16(b3DbvtVolume)
+	aabb = b3DbvtVolume::FromMM(aabbMin, aabbMax);
+	bool docollide = false;
+	if (proxy->stage == STAGECOUNT)
+	{ /* fixed -> dynamic set	*/
 		m_sets[1].remove(proxy->leaf);
-		proxy->leaf=m_sets[0].insert(aabb,proxy);
-		docollide=true;
+		proxy->leaf = m_sets[0].insert(aabb, proxy);
+		docollide = true;
 	}
 	else
-	{/* dynamic set				*/ 
+	{ /* dynamic set				*/
 		++m_updates_call;
-		/* Teleporting			*/ 
-		m_sets[0].update(proxy->leaf,aabb);
+		/* Teleporting			*/
+		m_sets[0].update(proxy->leaf, aabb);
 		++m_updates_done;
-		docollide=true;
+		docollide = true;
 	}
-	b3ListRemove(proxy,m_stageRoots[proxy->stage]);
+	b3ListRemove(proxy, m_stageRoots[proxy->stage]);
 	proxy->m_aabbMin = aabbMin;
 	proxy->m_aabbMax = aabbMax;
-	proxy->stage	=	m_stageCurrent;
-	b3ListAppend(proxy,m_stageRoots[m_stageCurrent]);
-	if(docollide)
+	proxy->stage = m_stageCurrent;
+	b3ListAppend(proxy, m_stageRoots[m_stageCurrent]);
+	if (docollide)
 	{
-		m_needcleanup=true;
-		if(!m_deferedcollide)
+		m_needcleanup = true;
+		if (!m_deferedcollide)
 		{
-			b3DbvtTreeCollider	collider(this);
-			m_sets[1].collideTTpersistentStack(m_sets[1].m_root,proxy->leaf,collider);
-			m_sets[0].collideTTpersistentStack(m_sets[0].m_root,proxy->leaf,collider);
+			b3DbvtTreeCollider collider(this);
+			m_sets[1].collideTTpersistentStack(m_sets[1].m_root, proxy->leaf, collider);
+			m_sets[0].collideTTpersistentStack(m_sets[0].m_root, proxy->leaf, collider);
 		}
-	}	
+	}
 }
 
 //
-void							b3DynamicBvhBroadphase::calculateOverlappingPairs(b3Dispatcher* dispatcher)
+void b3DynamicBvhBroadphase::calculateOverlappingPairs(b3Dispatcher* dispatcher)
 {
 	collide(dispatcher);
 #if B3_DBVT_BP_PROFILE
-	if(0==(m_pid%B3_DBVT_BP_PROFILING_RATE))
-	{	
-		printf("fixed(%u) dynamics(%u) pairs(%u)\r\n",m_sets[1].m_leaves,m_sets[0].m_leaves,m_paircache->getNumOverlappingPairs());
-		unsigned int	total=m_profiling.m_total;
-		if(total<=0) total=1;
-		printf("ddcollide: %u%% (%uus)\r\n",(50+m_profiling.m_ddcollide*100)/total,m_profiling.m_ddcollide/B3_DBVT_BP_PROFILING_RATE);
-		printf("fdcollide: %u%% (%uus)\r\n",(50+m_profiling.m_fdcollide*100)/total,m_profiling.m_fdcollide/B3_DBVT_BP_PROFILING_RATE);
-		printf("cleanup:   %u%% (%uus)\r\n",(50+m_profiling.m_cleanup*100)/total,m_profiling.m_cleanup/B3_DBVT_BP_PROFILING_RATE);
-		printf("total:     %uus\r\n",total/B3_DBVT_BP_PROFILING_RATE);
-		const unsigned long	sum=m_profiling.m_ddcollide+
-			m_profiling.m_fdcollide+
-			m_profiling.m_cleanup;
-		printf("leaked: %u%% (%uus)\r\n",100-((50+sum*100)/total),(total-sum)/B3_DBVT_BP_PROFILING_RATE);
-		printf("job counts: %u%%\r\n",(m_profiling.m_jobcount*100)/((m_sets[0].m_leaves+m_sets[1].m_leaves)*B3_DBVT_BP_PROFILING_RATE));
+	if (0 == (m_pid % B3_DBVT_BP_PROFILING_RATE))
+	{
+		printf("fixed(%u) dynamics(%u) pairs(%u)\r\n", m_sets[1].m_leaves, m_sets[0].m_leaves, m_paircache->getNumOverlappingPairs());
+		unsigned int total = m_profiling.m_total;
+		if (total <= 0) total = 1;
+		printf("ddcollide: %u%% (%uus)\r\n", (50 + m_profiling.m_ddcollide * 100) / total, m_profiling.m_ddcollide / B3_DBVT_BP_PROFILING_RATE);
+		printf("fdcollide: %u%% (%uus)\r\n", (50 + m_profiling.m_fdcollide * 100) / total, m_profiling.m_fdcollide / B3_DBVT_BP_PROFILING_RATE);
+		printf("cleanup:   %u%% (%uus)\r\n", (50 + m_profiling.m_cleanup * 100) / total, m_profiling.m_cleanup / B3_DBVT_BP_PROFILING_RATE);
+		printf("total:     %uus\r\n", total / B3_DBVT_BP_PROFILING_RATE);
+		const unsigned long sum = m_profiling.m_ddcollide +
+								  m_profiling.m_fdcollide +
+								  m_profiling.m_cleanup;
+		printf("leaked: %u%% (%uus)\r\n", 100 - ((50 + sum * 100) / total), (total - sum) / B3_DBVT_BP_PROFILING_RATE);
+		printf("job counts: %u%%\r\n", (m_profiling.m_jobcount * 100) / ((m_sets[0].m_leaves + m_sets[1].m_leaves) * B3_DBVT_BP_PROFILING_RATE));
 		b3Clear(m_profiling);
 		m_clock.reset();
 	}
 #endif
 
 	performDeferredRemoval(dispatcher);
-
 }
 
 void b3DynamicBvhBroadphase::performDeferredRemoval(b3Dispatcher* dispatcher)
 {
-
 	if (m_paircache->hasDeferredRemoval())
 	{
-
-		b3BroadphasePairArray&	overlappingPairArray = m_paircache->getOverlappingPairArray();
+		b3BroadphasePairArray& overlappingPairArray = m_paircache->getOverlappingPairArray();
 
 		//perform a sort, to find duplicates and to sort 'invalid' pairs to the end
 		overlappingPairArray.quickSort(b3BroadphasePairSortPredicate());
 
 		int invalidPair = 0;
 
-		
 		int i;
 
-		b3BroadphasePair previousPair = b3MakeBroadphasePair(-1,-1);
-		
-		
-		
-		for (i=0;i<overlappingPairArray.size();i++)
+		b3BroadphasePair previousPair = b3MakeBroadphasePair(-1, -1);
+
+		for (i = 0; i < overlappingPairArray.size(); i++)
 		{
-		
 			b3BroadphasePair& pair = overlappingPairArray[i];
 
 			bool isDuplicate = (pair == previousPair);
@@ -459,33 +455,34 @@ void b3DynamicBvhBroadphase::performDeferredRemoval(b3Dispatcher* dispatcher)
 			if (!isDuplicate)
 			{
 				//important to perform AABB check that is consistent with the broadphase
-				b3DbvtProxy*		pa=&m_proxies[pair.x];
-				b3DbvtProxy*		pb=&m_proxies[pair.y];
-				bool hasOverlap = b3Intersect(pa->leaf->volume,pb->leaf->volume);
+				b3DbvtProxy* pa = &m_proxies[pair.x];
+				b3DbvtProxy* pb = &m_proxies[pair.y];
+				bool hasOverlap = b3Intersect(pa->leaf->volume, pb->leaf->volume);
 
 				if (hasOverlap)
 				{
 					needsRemoval = false;
-				} else
+				}
+				else
 				{
 					needsRemoval = true;
 				}
-			} else
+			}
+			else
 			{
 				//remove duplicate
 				needsRemoval = true;
 				//should have no algorithm
 			}
-			
+
 			if (needsRemoval)
 			{
-				m_paircache->cleanOverlappingPair(pair,dispatcher);
+				m_paircache->cleanOverlappingPair(pair, dispatcher);
 
 				pair.x = -1;
 				pair.y = -1;
 				invalidPair++;
-			} 
-			
+			}
 		}
 
 		//perform a sort, to sort 'invalid' pairs to the end
@@ -495,7 +492,7 @@ void b3DynamicBvhBroadphase::performDeferredRemoval(b3Dispatcher* dispatcher)
 }
 
 //
-void							b3DynamicBvhBroadphase::collide(b3Dispatcher* dispatcher)
+void b3DynamicBvhBroadphase::collide(b3Dispatcher* dispatcher)
 {
 	/*printf("---------------------------------------------------------\n");
 	printf("m_sets[0].m_leaves=%d\n",m_sets[0].m_leaves);
@@ -512,285 +509,293 @@ void							b3DynamicBvhBroadphase::collide(b3Dispatcher* dispatcher)
 	}
 */
 
-
-
 	b3SPC(m_profiling.m_total);
-	/* optimize				*/ 
-	m_sets[0].optimizeIncremental(1+(m_sets[0].m_leaves*m_dupdates)/100);
-	if(m_fixedleft)
+	/* optimize				*/
+	m_sets[0].optimizeIncremental(1 + (m_sets[0].m_leaves * m_dupdates) / 100);
+	if (m_fixedleft)
 	{
-		const int count=1+(m_sets[1].m_leaves*m_fupdates)/100;
-		m_sets[1].optimizeIncremental(1+(m_sets[1].m_leaves*m_fupdates)/100);
-		m_fixedleft=b3Max<int>(0,m_fixedleft-count);
+		const int count = 1 + (m_sets[1].m_leaves * m_fupdates) / 100;
+		m_sets[1].optimizeIncremental(1 + (m_sets[1].m_leaves * m_fupdates) / 100);
+		m_fixedleft = b3Max<int>(0, m_fixedleft - count);
 	}
-	/* dynamic -> fixed set	*/ 
-	m_stageCurrent=(m_stageCurrent+1)%STAGECOUNT;
-	b3DbvtProxy*	current=m_stageRoots[m_stageCurrent];
-	if(current)
+	/* dynamic -> fixed set	*/
+	m_stageCurrent = (m_stageCurrent + 1) % STAGECOUNT;
+	b3DbvtProxy* current = m_stageRoots[m_stageCurrent];
+	if (current)
 	{
-		b3DbvtTreeCollider	collider(this);
-		do	{
-			b3DbvtProxy*	next=current->links[1];
-			b3ListRemove(current,m_stageRoots[current->stage]);
-			b3ListAppend(current,m_stageRoots[STAGECOUNT]);
+		b3DbvtTreeCollider collider(this);
+		do
+		{
+			b3DbvtProxy* next = current->links[1];
+			b3ListRemove(current, m_stageRoots[current->stage]);
+			b3ListAppend(current, m_stageRoots[STAGECOUNT]);
 #if B3_DBVT_BP_ACCURATESLEEPING
-			m_paircache->removeOverlappingPairsContainingProxy(current,dispatcher);
-			collider.proxy=current;
-			b3DynamicBvh::collideTV(m_sets[0].m_root,current->aabb,collider);
-			b3DynamicBvh::collideTV(m_sets[1].m_root,current->aabb,collider);
+			m_paircache->removeOverlappingPairsContainingProxy(current, dispatcher);
+			collider.proxy = current;
+			b3DynamicBvh::collideTV(m_sets[0].m_root, current->aabb, collider);
+			b3DynamicBvh::collideTV(m_sets[1].m_root, current->aabb, collider);
 #endif
 			m_sets[0].remove(current->leaf);
-			B3_ATTRIBUTE_ALIGNED16(b3DbvtVolume)	curAabb=b3DbvtVolume::FromMM(current->m_aabbMin,current->m_aabbMax);
-			current->leaf	=	m_sets[1].insert(curAabb,current);
-			current->stage	=	STAGECOUNT;	
-			current			=	next;
-		} while(current);
-		m_fixedleft=m_sets[1].m_leaves;
-		m_needcleanup=true;
+			B3_ATTRIBUTE_ALIGNED16(b3DbvtVolume)
+			curAabb = b3DbvtVolume::FromMM(current->m_aabbMin, current->m_aabbMax);
+			current->leaf = m_sets[1].insert(curAabb, current);
+			current->stage = STAGECOUNT;
+			current = next;
+		} while (current);
+		m_fixedleft = m_sets[1].m_leaves;
+		m_needcleanup = true;
 	}
-	/* collide dynamics		*/ 
+	/* collide dynamics		*/
 	{
-		b3DbvtTreeCollider	collider(this);
-		if(m_deferedcollide)
+		b3DbvtTreeCollider collider(this);
+		if (m_deferedcollide)
 		{
 			b3SPC(m_profiling.m_fdcollide);
-			m_sets[0].collideTTpersistentStack(m_sets[0].m_root,m_sets[1].m_root,collider);
+			m_sets[0].collideTTpersistentStack(m_sets[0].m_root, m_sets[1].m_root, collider);
 		}
-		if(m_deferedcollide)
+		if (m_deferedcollide)
 		{
 			b3SPC(m_profiling.m_ddcollide);
-			m_sets[0].collideTTpersistentStack(m_sets[0].m_root,m_sets[0].m_root,collider);
+			m_sets[0].collideTTpersistentStack(m_sets[0].m_root, m_sets[0].m_root, collider);
 		}
 	}
-	/* clean up				*/ 
-	if(m_needcleanup)
+	/* clean up				*/
+	if (m_needcleanup)
 	{
 		b3SPC(m_profiling.m_cleanup);
-		b3BroadphasePairArray&	pairs=m_paircache->getOverlappingPairArray();
-		if(pairs.size()>0)
+		b3BroadphasePairArray& pairs = m_paircache->getOverlappingPairArray();
+		if (pairs.size() > 0)
 		{
-
-			int			ni=b3Min(pairs.size(),b3Max<int>(m_newpairs,(pairs.size()*m_cupdates)/100));
-			for(int i=0;i<ni;++i)
+			int ni = b3Min(pairs.size(), b3Max<int>(m_newpairs, (pairs.size() * m_cupdates) / 100));
+			for (int i = 0; i < ni; ++i)
 			{
-				b3BroadphasePair&	p=pairs[(m_cid+i)%pairs.size()];
-				b3DbvtProxy*		pa=&m_proxies[p.x];
-				b3DbvtProxy*		pb=&m_proxies[p.y];
-				if(!b3Intersect(pa->leaf->volume,pb->leaf->volume))
+				b3BroadphasePair& p = pairs[(m_cid + i) % pairs.size()];
+				b3DbvtProxy* pa = &m_proxies[p.x];
+				b3DbvtProxy* pb = &m_proxies[p.y];
+				if (!b3Intersect(pa->leaf->volume, pb->leaf->volume))
 				{
 #if B3_DBVT_BP_SORTPAIRS
-					if(pa->m_uniqueId>pb->m_uniqueId) 
-						b3Swap(pa,pb);
+					if (pa->m_uniqueId > pb->m_uniqueId)
+						b3Swap(pa, pb);
 #endif
-					m_paircache->removeOverlappingPair(pa->getUid(),pb->getUid(),dispatcher);
-					--ni;--i;
+					m_paircache->removeOverlappingPair(pa->getUid(), pb->getUid(), dispatcher);
+					--ni;
+					--i;
 				}
 			}
-			if(pairs.size()>0) m_cid=(m_cid+ni)%pairs.size(); else m_cid=0;
+			if (pairs.size() > 0)
+				m_cid = (m_cid + ni) % pairs.size();
+			else
+				m_cid = 0;
 		}
 	}
 	++m_pid;
-	m_newpairs=1;
-	m_needcleanup=false;
-	if(m_updates_call>0)
-	{ m_updates_ratio=m_updates_done/(b3Scalar)m_updates_call; }
+	m_newpairs = 1;
+	m_needcleanup = false;
+	if (m_updates_call > 0)
+	{
+		m_updates_ratio = m_updates_done / (b3Scalar)m_updates_call;
+	}
 	else
-	{ m_updates_ratio=0; }
-	m_updates_done/=2;
-	m_updates_call/=2;
+	{
+		m_updates_ratio = 0;
+	}
+	m_updates_done /= 2;
+	m_updates_call /= 2;
 }
 
 //
-void							b3DynamicBvhBroadphase::optimize()
+void b3DynamicBvhBroadphase::optimize()
 {
 	m_sets[0].optimizeTopDown();
 	m_sets[1].optimizeTopDown();
 }
 
 //
-b3OverlappingPairCache*			b3DynamicBvhBroadphase::getOverlappingPairCache()
+b3OverlappingPairCache* b3DynamicBvhBroadphase::getOverlappingPairCache()
 {
-	return(m_paircache);
+	return (m_paircache);
 }
 
 //
-const b3OverlappingPairCache*	b3DynamicBvhBroadphase::getOverlappingPairCache() const
+const b3OverlappingPairCache* b3DynamicBvhBroadphase::getOverlappingPairCache() const
 {
-	return(m_paircache);
+	return (m_paircache);
 }
 
 //
-void							b3DynamicBvhBroadphase::getBroadphaseAabb(b3Vector3& aabbMin,b3Vector3& aabbMax) const
+void b3DynamicBvhBroadphase::getBroadphaseAabb(b3Vector3& aabbMin, b3Vector3& aabbMax) const
 {
+	B3_ATTRIBUTE_ALIGNED16(b3DbvtVolume)
+	bounds;
 
-	B3_ATTRIBUTE_ALIGNED16(b3DbvtVolume)	bounds;
-
-	if(!m_sets[0].empty())
-		if(!m_sets[1].empty())	b3Merge(	m_sets[0].m_root->volume,
-			m_sets[1].m_root->volume,bounds);
+	if (!m_sets[0].empty())
+		if (!m_sets[1].empty())
+			b3Merge(m_sets[0].m_root->volume,
+					m_sets[1].m_root->volume, bounds);
 		else
-			bounds=m_sets[0].m_root->volume;
-	else if(!m_sets[1].empty())	bounds=m_sets[1].m_root->volume;
+			bounds = m_sets[0].m_root->volume;
+	else if (!m_sets[1].empty())
+		bounds = m_sets[1].m_root->volume;
 	else
-		bounds=b3DbvtVolume::FromCR(b3MakeVector3(0,0,0),0);
-	aabbMin=bounds.Mins();
-	aabbMax=bounds.Maxs();
+		bounds = b3DbvtVolume::FromCR(b3MakeVector3(0, 0, 0), 0);
+	aabbMin = bounds.Mins();
+	aabbMax = bounds.Maxs();
 }
 
 void b3DynamicBvhBroadphase::resetPool(b3Dispatcher* dispatcher)
 {
-	
 	int totalObjects = m_sets[0].m_leaves + m_sets[1].m_leaves;
 	if (!totalObjects)
 	{
 		//reset internal dynamic tree data structures
 		m_sets[0].clear();
 		m_sets[1].clear();
-		
-		m_deferedcollide	=	false;
-		m_needcleanup		=	true;
-		m_stageCurrent		=	0;
-		m_fixedleft			=	0;
-		m_fupdates			=	1;
-		m_dupdates			=	0;
-		m_cupdates			=	10;
-		m_newpairs			=	1;
-		m_updates_call		=	0;
-		m_updates_done		=	0;
-		m_updates_ratio		=	0;
-		
-		m_pid				=	0;
-		m_cid				=	0;
-		for(int i=0;i<=STAGECOUNT;++i)
+
+		m_deferedcollide = false;
+		m_needcleanup = true;
+		m_stageCurrent = 0;
+		m_fixedleft = 0;
+		m_fupdates = 1;
+		m_dupdates = 0;
+		m_cupdates = 10;
+		m_newpairs = 1;
+		m_updates_call = 0;
+		m_updates_done = 0;
+		m_updates_ratio = 0;
+
+		m_pid = 0;
+		m_cid = 0;
+		for (int i = 0; i <= STAGECOUNT; ++i)
 		{
-			m_stageRoots[i]=0;
+			m_stageRoots[i] = 0;
 		}
 	}
 }
 
 //
-void							b3DynamicBvhBroadphase::printStats()
-{}
+void b3DynamicBvhBroadphase::printStats()
+{
+}
 
 //
 #if B3_DBVT_BP_ENABLE_BENCHMARK
 
-struct	b3BroadphaseBenchmark
+struct b3BroadphaseBenchmark
 {
-	struct	Experiment
+	struct Experiment
 	{
-		const char*			name;
-		int					object_count;
-		int					update_count;
-		int					spawn_count;
-		int					iterations;
-		b3Scalar			speed;
-		b3Scalar			amplitude;
+		const char* name;
+		int object_count;
+		int update_count;
+		int spawn_count;
+		int iterations;
+		b3Scalar speed;
+		b3Scalar amplitude;
 	};
-	struct	Object
+	struct Object
 	{
-		b3Vector3			center;
-		b3Vector3			extents;
-		b3BroadphaseProxy*	proxy;
-		b3Scalar			time;
-		void				update(b3Scalar speed,b3Scalar amplitude,b3BroadphaseInterface* pbi)
+		b3Vector3 center;
+		b3Vector3 extents;
+		b3BroadphaseProxy* proxy;
+		b3Scalar time;
+		void update(b3Scalar speed, b3Scalar amplitude, b3BroadphaseInterface* pbi)
 		{
-			time		+=	speed;
-			center[0]	=	b3Cos(time*(b3Scalar)2.17)*amplitude+
-				b3Sin(time)*amplitude/2;
-			center[1]	=	b3Cos(time*(b3Scalar)1.38)*amplitude+
-				b3Sin(time)*amplitude;
-			center[2]	=	b3Sin(time*(b3Scalar)0.777)*amplitude;
-			pbi->setAabb(proxy,center-extents,center+extents,0);
+			time += speed;
+			center[0] = b3Cos(time * (b3Scalar)2.17) * amplitude +
+						b3Sin(time) * amplitude / 2;
+			center[1] = b3Cos(time * (b3Scalar)1.38) * amplitude +
+						b3Sin(time) * amplitude;
+			center[2] = b3Sin(time * (b3Scalar)0.777) * amplitude;
+			pbi->setAabb(proxy, center - extents, center + extents, 0);
 		}
 	};
-	static int		UnsignedRand(int range=RAND_MAX-1)	{ return(rand()%(range+1)); }
-	static b3Scalar	UnitRand()							{ return(UnsignedRand(16384)/(b3Scalar)16384); }
-	static void		OutputTime(const char* name,b3Clock& c,unsigned count=0)
+	static int UnsignedRand(int range = RAND_MAX - 1) { return (rand() % (range + 1)); }
+	static b3Scalar UnitRand() { return (UnsignedRand(16384) / (b3Scalar)16384); }
+	static void OutputTime(const char* name, b3Clock& c, unsigned count = 0)
 	{
-		const unsigned long	us=c.getTimeMicroseconds();
-		const unsigned long	ms=(us+500)/1000;
-		const b3Scalar		sec=us/(b3Scalar)(1000*1000);
-		if(count>0)
-			printf("%s : %u us (%u ms), %.2f/s\r\n",name,us,ms,count/sec);
+		const unsigned long us = c.getTimeMicroseconds();
+		const unsigned long ms = (us + 500) / 1000;
+		const b3Scalar sec = us / (b3Scalar)(1000 * 1000);
+		if (count > 0)
+			printf("%s : %u us (%u ms), %.2f/s\r\n", name, us, ms, count / sec);
 		else
-			printf("%s : %u us (%u ms)\r\n",name,us,ms);
+			printf("%s : %u us (%u ms)\r\n", name, us, ms);
 	}
 };
 
-void							b3DynamicBvhBroadphase::benchmark(b3BroadphaseInterface* pbi)
+void b3DynamicBvhBroadphase::benchmark(b3BroadphaseInterface* pbi)
 {
-	static const b3BroadphaseBenchmark::Experiment		experiments[]=
-	{
-		{"1024o.10%",1024,10,0,8192,(b3Scalar)0.005,(b3Scalar)100},
-		/*{"4096o.10%",4096,10,0,8192,(b3Scalar)0.005,(b3Scalar)100},
+	static const b3BroadphaseBenchmark::Experiment experiments[] =
+		{
+			{"1024o.10%", 1024, 10, 0, 8192, (b3Scalar)0.005, (b3Scalar)100},
+			/*{"4096o.10%",4096,10,0,8192,(b3Scalar)0.005,(b3Scalar)100},
 		{"8192o.10%",8192,10,0,8192,(b3Scalar)0.005,(b3Scalar)100},*/
-	};
-	static const int										nexperiments=sizeof(experiments)/sizeof(experiments[0]);
-	b3AlignedObjectArray<b3BroadphaseBenchmark::Object*>	objects;
-	b3Clock													wallclock;
-	/* Begin			*/ 
-	for(int iexp=0;iexp<nexperiments;++iexp)
+		};
+	static const int nexperiments = sizeof(experiments) / sizeof(experiments[0]);
+	b3AlignedObjectArray<b3BroadphaseBenchmark::Object*> objects;
+	b3Clock wallclock;
+	/* Begin			*/
+	for (int iexp = 0; iexp < nexperiments; ++iexp)
 	{
-		const b3BroadphaseBenchmark::Experiment&	experiment=experiments[iexp];
-		const int									object_count=experiment.object_count;
-		const int									update_count=(object_count*experiment.update_count)/100;
-		const int									spawn_count=(object_count*experiment.spawn_count)/100;
-		const b3Scalar								speed=experiment.speed;	
-		const b3Scalar								amplitude=experiment.amplitude;
-		printf("Experiment #%u '%s':\r\n",iexp,experiment.name);
-		printf("\tObjects: %u\r\n",object_count);
-		printf("\tUpdate: %u\r\n",update_count);
-		printf("\tSpawn: %u\r\n",spawn_count);
-		printf("\tSpeed: %f\r\n",speed);
-		printf("\tAmplitude: %f\r\n",amplitude);
+		const b3BroadphaseBenchmark::Experiment& experiment = experiments[iexp];
+		const int object_count = experiment.object_count;
+		const int update_count = (object_count * experiment.update_count) / 100;
+		const int spawn_count = (object_count * experiment.spawn_count) / 100;
+		const b3Scalar speed = experiment.speed;
+		const b3Scalar amplitude = experiment.amplitude;
+		printf("Experiment #%u '%s':\r\n", iexp, experiment.name);
+		printf("\tObjects: %u\r\n", object_count);
+		printf("\tUpdate: %u\r\n", update_count);
+		printf("\tSpawn: %u\r\n", spawn_count);
+		printf("\tSpeed: %f\r\n", speed);
+		printf("\tAmplitude: %f\r\n", amplitude);
 		srand(180673);
-		/* Create objects	*/ 
+		/* Create objects	*/
 		wallclock.reset();
 		objects.reserve(object_count);
-		for(int i=0;i<object_count;++i)
+		for (int i = 0; i < object_count; ++i)
 		{
-			b3BroadphaseBenchmark::Object*	po=new b3BroadphaseBenchmark::Object();
-			po->center[0]=b3BroadphaseBenchmark::UnitRand()*50;
-			po->center[1]=b3BroadphaseBenchmark::UnitRand()*50;
-			po->center[2]=b3BroadphaseBenchmark::UnitRand()*50;
-			po->extents[0]=b3BroadphaseBenchmark::UnitRand()*2+2;
-			po->extents[1]=b3BroadphaseBenchmark::UnitRand()*2+2;
-			po->extents[2]=b3BroadphaseBenchmark::UnitRand()*2+2;
-			po->time=b3BroadphaseBenchmark::UnitRand()*2000;
-			po->proxy=pbi->createProxy(po->center-po->extents,po->center+po->extents,0,po,1,1,0,0);
+			b3BroadphaseBenchmark::Object* po = new b3BroadphaseBenchmark::Object();
+			po->center[0] = b3BroadphaseBenchmark::UnitRand() * 50;
+			po->center[1] = b3BroadphaseBenchmark::UnitRand() * 50;
+			po->center[2] = b3BroadphaseBenchmark::UnitRand() * 50;
+			po->extents[0] = b3BroadphaseBenchmark::UnitRand() * 2 + 2;
+			po->extents[1] = b3BroadphaseBenchmark::UnitRand() * 2 + 2;
+			po->extents[2] = b3BroadphaseBenchmark::UnitRand() * 2 + 2;
+			po->time = b3BroadphaseBenchmark::UnitRand() * 2000;
+			po->proxy = pbi->createProxy(po->center - po->extents, po->center + po->extents, 0, po, 1, 1, 0, 0);
 			objects.push_back(po);
 		}
-		b3BroadphaseBenchmark::OutputTime("\tInitialization",wallclock);
-		/* First update		*/ 
+		b3BroadphaseBenchmark::OutputTime("\tInitialization", wallclock);
+		/* First update		*/
 		wallclock.reset();
-		for(int i=0;i<objects.size();++i)
+		for (int i = 0; i < objects.size(); ++i)
 		{
-			objects[i]->update(speed,amplitude,pbi);
+			objects[i]->update(speed, amplitude, pbi);
 		}
-		b3BroadphaseBenchmark::OutputTime("\tFirst update",wallclock);
-		/* Updates			*/ 
+		b3BroadphaseBenchmark::OutputTime("\tFirst update", wallclock);
+		/* Updates			*/
 		wallclock.reset();
-		for(int i=0;i<experiment.iterations;++i)
+		for (int i = 0; i < experiment.iterations; ++i)
 		{
-			for(int j=0;j<update_count;++j)
-			{				
-				objects[j]->update(speed,amplitude,pbi);
+			for (int j = 0; j < update_count; ++j)
+			{
+				objects[j]->update(speed, amplitude, pbi);
 			}
 			pbi->calculateOverlappingPairs(0);
 		}
-		b3BroadphaseBenchmark::OutputTime("\tUpdate",wallclock,experiment.iterations);
-		/* Clean up			*/ 
+		b3BroadphaseBenchmark::OutputTime("\tUpdate", wallclock, experiment.iterations);
+		/* Clean up			*/
 		wallclock.reset();
-		for(int i=0;i<objects.size();++i)
+		for (int i = 0; i < objects.size(); ++i)
 		{
-			pbi->destroyProxy(objects[i]->proxy,0);
+			pbi->destroyProxy(objects[i]->proxy, 0);
 			delete objects[i];
 		}
 		objects.resize(0);
-		b3BroadphaseBenchmark::OutputTime("\tRelease",wallclock);
+		b3BroadphaseBenchmark::OutputTime("\tRelease", wallclock);
 	}
-
 }
 #else
 /*void							b3DynamicBvhBroadphase::benchmark(b3BroadphaseInterface*)
@@ -799,6 +804,5 @@ void							b3DynamicBvhBroadphase::benchmark(b3BroadphaseInterface* pbi)
 #endif
 
 #if B3_DBVT_BP_PROFILE
-#undef	b3SPC
+#undef b3SPC
 #endif
-
diff --git a/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3DynamicBvhBroadphase.h b/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3DynamicBvhBroadphase.h
index 7ac085d90cd..c235e40148c 100644
--- a/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3DynamicBvhBroadphase.h
+++ b/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3DynamicBvhBroadphase.h
@@ -27,46 +27,43 @@ subject to the following restrictions:
 // Compile time config
 //
 
-#define	B3_DBVT_BP_PROFILE					0
+#define B3_DBVT_BP_PROFILE 0
 //#define B3_DBVT_BP_SORTPAIRS				1
-#define B3_DBVT_BP_PREVENTFALSEUPDATE		0
-#define B3_DBVT_BP_ACCURATESLEEPING		0
-#define B3_DBVT_BP_ENABLE_BENCHMARK		0
-#define B3_DBVT_BP_MARGIN					(b3Scalar)0.05
+#define B3_DBVT_BP_PREVENTFALSEUPDATE 0
+#define B3_DBVT_BP_ACCURATESLEEPING 0
+#define B3_DBVT_BP_ENABLE_BENCHMARK 0
+#define B3_DBVT_BP_MARGIN (b3Scalar)0.05
 
 #if B3_DBVT_BP_PROFILE
-#define	B3_DBVT_BP_PROFILING_RATE	256
+#define B3_DBVT_BP_PROFILING_RATE 256
 
 #endif
 
-
-
-
-B3_ATTRIBUTE_ALIGNED16(struct) b3BroadphaseProxy
+B3_ATTRIBUTE_ALIGNED16(struct)
+b3BroadphaseProxy
 {
+	B3_DECLARE_ALIGNED_ALLOCATOR();
 
-B3_DECLARE_ALIGNED_ALLOCATOR();
-	
 	///optional filtering to cull potential collisions
 	enum CollisionFilterGroups
 	{
-	        DefaultFilter = 1,
-	        StaticFilter = 2,
-	        KinematicFilter = 4,
-	        DebrisFilter = 8,
-			SensorTrigger = 16,
-			CharacterFilter = 32,
-	        AllFilter = -1 //all bits sets: DefaultFilter | StaticFilter | KinematicFilter | DebrisFilter | SensorTrigger
+		DefaultFilter = 1,
+		StaticFilter = 2,
+		KinematicFilter = 4,
+		DebrisFilter = 8,
+		SensorTrigger = 16,
+		CharacterFilter = 32,
+		AllFilter = -1  //all bits sets: DefaultFilter | StaticFilter | KinematicFilter | DebrisFilter | SensorTrigger
 	};
 
 	//Usually the client b3CollisionObject or Rigidbody class
-	void*	m_clientObject;
+	void* m_clientObject;
 	int m_collisionFilterGroup;
 	int m_collisionFilterMask;
-	int			m_uniqueId;//m_uniqueId is introduced for paircache. could get rid of this, by calculating the address offset etc.
+	int m_uniqueId;  //m_uniqueId is introduced for paircache. could get rid of this, by calculating the address offset etc.
 
-	b3Vector3	m_aabbMin;
-	b3Vector3	m_aabbMax;
+	b3Vector3 m_aabbMin;
+	b3Vector3 m_aabbMax;
 
 	B3_FORCE_INLINE int getUid() const
 	{
@@ -74,116 +71,112 @@ B3_DECLARE_ALIGNED_ALLOCATOR();
 	}
 
 	//used for memory pools
-	b3BroadphaseProxy() :m_clientObject(0)
+	b3BroadphaseProxy() : m_clientObject(0)
 	{
 	}
 
-	b3BroadphaseProxy(const b3Vector3& aabbMin,const b3Vector3& aabbMax,void* userPtr, int collisionFilterGroup,  int collisionFilterMask)
-		:m_clientObject(userPtr),
-		m_collisionFilterGroup(collisionFilterGroup),
-		m_collisionFilterMask(collisionFilterMask),
-		m_aabbMin(aabbMin),
-		m_aabbMax(aabbMax)
+	b3BroadphaseProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, void* userPtr, int collisionFilterGroup, int collisionFilterMask)
+		: m_clientObject(userPtr),
+		  m_collisionFilterGroup(collisionFilterGroup),
+		  m_collisionFilterMask(collisionFilterMask),
+		  m_aabbMin(aabbMin),
+		  m_aabbMax(aabbMax)
 	{
 	}
 };
 
-
-
-
-
 //
 // b3DbvtProxy
 //
 struct b3DbvtProxy : b3BroadphaseProxy
 {
-	/* Fields		*/ 
+	/* Fields		*/
 	//b3DbvtAabbMm	aabb;
-	b3DbvtNode*		leaf;
-	b3DbvtProxy*	links[2];
-	int				stage;
-	/* ctor			*/ 
+	b3DbvtNode* leaf;
+	b3DbvtProxy* links[2];
+	int stage;
+	/* ctor			*/
 
 	explicit b3DbvtProxy() {}
-	b3DbvtProxy(const b3Vector3& aabbMin,const b3Vector3& aabbMax,void* userPtr, int collisionFilterGroup,  int collisionFilterMask) :
-	b3BroadphaseProxy(aabbMin,aabbMax,userPtr,collisionFilterGroup,collisionFilterMask)
+	b3DbvtProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, void* userPtr, int collisionFilterGroup, int collisionFilterMask) : b3BroadphaseProxy(aabbMin, aabbMax, userPtr, collisionFilterGroup, collisionFilterMask)
 	{
-		links[0]=links[1]=0;
+		links[0] = links[1] = 0;
 	}
 };
 
-typedef b3AlignedObjectArray<b3DbvtProxy*>	b3DbvtProxyArray;
+typedef b3AlignedObjectArray<b3DbvtProxy*> b3DbvtProxyArray;
 
 ///The b3DynamicBvhBroadphase implements a broadphase using two dynamic AABB bounding volume hierarchies/trees (see b3DynamicBvh).
 ///One tree is used for static/non-moving objects, and another tree is used for dynamic objects. Objects can move from one tree to the other.
 ///This is a very fast broadphase, especially for very dynamic worlds where many objects are moving. Its insert/add and remove of objects is generally faster than the sweep and prune broadphases b3AxisSweep3 and b332BitAxisSweep3.
-struct	b3DynamicBvhBroadphase 
+struct b3DynamicBvhBroadphase
 {
-	/* Config		*/ 
-	enum	{
-		DYNAMIC_SET			=	0,	/* Dynamic set index	*/ 
-		FIXED_SET			=	1,	/* Fixed set index		*/ 
-		STAGECOUNT			=	2	/* Number of stages		*/ 
+	/* Config		*/
+	enum
+	{
+		DYNAMIC_SET = 0, /* Dynamic set index	*/
+		FIXED_SET = 1,   /* Fixed set index		*/
+		STAGECOUNT = 2   /* Number of stages		*/
 	};
-	/* Fields		*/ 
-	b3DynamicBvh					m_sets[2];					// Dbvt sets
-	b3DbvtProxy*			m_stageRoots[STAGECOUNT+1];	// Stages list
+	/* Fields		*/
+	b3DynamicBvh m_sets[2];                     // Dbvt sets
+	b3DbvtProxy* m_stageRoots[STAGECOUNT + 1];  // Stages list
 
-	b3AlignedObjectArray<b3DbvtProxy>	m_proxies;
-	b3OverlappingPairCache*	m_paircache;				// Pair cache
-	b3Scalar				m_prediction;				// Velocity prediction
-	int						m_stageCurrent;				// Current stage
-	int						m_fupdates;					// % of fixed updates per frame
-	int						m_dupdates;					// % of dynamic updates per frame
-	int						m_cupdates;					// % of cleanup updates per frame
-	int						m_newpairs;					// Number of pairs created
-	int						m_fixedleft;				// Fixed optimization left
-	unsigned				m_updates_call;				// Number of updates call
-	unsigned				m_updates_done;				// Number of updates done
-	b3Scalar				m_updates_ratio;			// m_updates_done/m_updates_call
-	int						m_pid;						// Parse id
-	int						m_cid;						// Cleanup index
-	bool					m_releasepaircache;			// Release pair cache on delete
-	bool					m_deferedcollide;			// Defere dynamic/static collision to collide call
-	bool					m_needcleanup;				// Need to run cleanup?
+	b3AlignedObjectArray<b3DbvtProxy> m_proxies;
+	b3OverlappingPairCache* m_paircache;  // Pair cache
+	b3Scalar m_prediction;                // Velocity prediction
+	int m_stageCurrent;                   // Current stage
+	int m_fupdates;                       // % of fixed updates per frame
+	int m_dupdates;                       // % of dynamic updates per frame
+	int m_cupdates;                       // % of cleanup updates per frame
+	int m_newpairs;                       // Number of pairs created
+	int m_fixedleft;                      // Fixed optimization left
+	unsigned m_updates_call;              // Number of updates call
+	unsigned m_updates_done;              // Number of updates done
+	b3Scalar m_updates_ratio;             // m_updates_done/m_updates_call
+	int m_pid;                            // Parse id
+	int m_cid;                            // Cleanup index
+	bool m_releasepaircache;              // Release pair cache on delete
+	bool m_deferedcollide;                // Defere dynamic/static collision to collide call
+	bool m_needcleanup;                   // Need to run cleanup?
 #if B3_DBVT_BP_PROFILE
-	b3Clock					m_clock;
-	struct	{
-		unsigned long		m_total;
-		unsigned long		m_ddcollide;
-		unsigned long		m_fdcollide;
-		unsigned long		m_cleanup;
-		unsigned long		m_jobcount;
-	}				m_profiling;
+	b3Clock m_clock;
+	struct
+	{
+		unsigned long m_total;
+		unsigned long m_ddcollide;
+		unsigned long m_fdcollide;
+		unsigned long m_cleanup;
+		unsigned long m_jobcount;
+	} m_profiling;
 #endif
-	/* Methods		*/ 
-	b3DynamicBvhBroadphase(int proxyCapacity, b3OverlappingPairCache* paircache=0);
+	/* Methods		*/
+	b3DynamicBvhBroadphase(int proxyCapacity, b3OverlappingPairCache* paircache = 0);
 	virtual ~b3DynamicBvhBroadphase();
-	void							collide(b3Dispatcher* dispatcher);
-	void							optimize();
-	
+	void collide(b3Dispatcher* dispatcher);
+	void optimize();
+
 	/* b3BroadphaseInterface Implementation	*/
-	b3BroadphaseProxy*				createProxy(const b3Vector3& aabbMin,const b3Vector3& aabbMax,int objectIndex,void* userPtr, int collisionFilterGroup, int collisionFilterMask);
-	virtual void					destroyProxy(b3BroadphaseProxy* proxy,b3Dispatcher* dispatcher);
-	virtual void					setAabb(int objectId,const b3Vector3& aabbMin,const b3Vector3& aabbMax,b3Dispatcher* dispatcher);
-	virtual void					rayTest(const b3Vector3& rayFrom,const b3Vector3& rayTo, b3BroadphaseRayCallback& rayCallback, const b3Vector3& aabbMin=b3MakeVector3(0,0,0), const b3Vector3& aabbMax = b3MakeVector3(0,0,0));
-	virtual void					aabbTest(const b3Vector3& aabbMin, const b3Vector3& aabbMax, b3BroadphaseAabbCallback& callback);
+	b3BroadphaseProxy* createProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, int objectIndex, void* userPtr, int collisionFilterGroup, int collisionFilterMask);
+	virtual void destroyProxy(b3BroadphaseProxy* proxy, b3Dispatcher* dispatcher);
+	virtual void setAabb(int objectId, const b3Vector3& aabbMin, const b3Vector3& aabbMax, b3Dispatcher* dispatcher);
+	virtual void rayTest(const b3Vector3& rayFrom, const b3Vector3& rayTo, b3BroadphaseRayCallback& rayCallback, const b3Vector3& aabbMin = b3MakeVector3(0, 0, 0), const b3Vector3& aabbMax = b3MakeVector3(0, 0, 0));
+	virtual void aabbTest(const b3Vector3& aabbMin, const b3Vector3& aabbMax, b3BroadphaseAabbCallback& callback);
 
 	//virtual void					getAabb(b3BroadphaseProxy* proxy,b3Vector3& aabbMin, b3Vector3& aabbMax ) const;
-	virtual void					getAabb(int objectId,b3Vector3& aabbMin, b3Vector3& aabbMax ) const;
-	virtual	void					calculateOverlappingPairs(b3Dispatcher* dispatcher=0);
-	virtual	b3OverlappingPairCache*	getOverlappingPairCache();
-	virtual	const b3OverlappingPairCache*	getOverlappingPairCache() const;
-	virtual	void					getBroadphaseAabb(b3Vector3& aabbMin,b3Vector3& aabbMax) const;
-	virtual	void					printStats();
-
+	virtual void getAabb(int objectId, b3Vector3& aabbMin, b3Vector3& aabbMax) const;
+	virtual void calculateOverlappingPairs(b3Dispatcher* dispatcher = 0);
+	virtual b3OverlappingPairCache* getOverlappingPairCache();
+	virtual const b3OverlappingPairCache* getOverlappingPairCache() const;
+	virtual void getBroadphaseAabb(b3Vector3& aabbMin, b3Vector3& aabbMax) const;
+	virtual void printStats();
 
 	///reset broadphase internal structures, to ensure determinism/reproducability
 	virtual void resetPool(b3Dispatcher* dispatcher);
 
-	void	performDeferredRemoval(b3Dispatcher* dispatcher);
-	
-	void	setVelocityPrediction(b3Scalar prediction)
+	void performDeferredRemoval(b3Dispatcher* dispatcher);
+
+	void setVelocityPrediction(b3Scalar prediction)
 	{
 		m_prediction = prediction;
 	}
@@ -192,15 +185,13 @@ struct	b3DynamicBvhBroadphase
 		return m_prediction;
 	}
 
-	///this setAabbForceUpdate is similar to setAabb but always forces the aabb update. 
+	///this setAabbForceUpdate is similar to setAabb but always forces the aabb update.
 	///it is not part of the b3BroadphaseInterface but specific to b3DynamicBvhBroadphase.
 	///it bypasses certain optimizations that prevent aabb updates (when the aabb shrinks), see
 	///http://code.google.com/p/bullet/issues/detail?id=223
-	void							setAabbForceUpdate(		b3BroadphaseProxy* absproxy,const b3Vector3& aabbMin,const b3Vector3& aabbMax,b3Dispatcher* /*dispatcher*/);
+	void setAabbForceUpdate(b3BroadphaseProxy* absproxy, const b3Vector3& aabbMin, const b3Vector3& aabbMax, b3Dispatcher* /*dispatcher*/);
 
 	//static void						benchmark(b3BroadphaseInterface*);
-
-
 };
 
 #endif
diff --git a/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3OverlappingPair.h b/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3OverlappingPair.h
index 39bf27de3e3..4ff9ebae81f 100644
--- a/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3OverlappingPair.h
+++ b/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3OverlappingPair.h
@@ -23,20 +23,20 @@ subject to the following restrictions:
 
 typedef b3Int4 b3BroadphasePair;
 
-inline b3Int4 b3MakeBroadphasePair(int xx,int yy)
+inline b3Int4 b3MakeBroadphasePair(int xx, int yy)
 {
 	b3Int4 pair;
 
 	if (xx < yy)
-    { 
-        pair.x = xx; 
-        pair.y = yy;
-    }
-    else 
-    { 
+	{
+		pair.x = xx;
+		pair.y = yy;
+	}
+	else
+	{
 		pair.x = yy;
-        pair.y = xx;
-    }
+		pair.y = xx;
+	}
 	pair.z = B3_NEW_PAIR_MARKER;
 	pair.w = B3_NEW_PAIR_MARKER;
 	return pair;
@@ -51,22 +51,20 @@ inline b3Int4 b3MakeBroadphasePair(int xx,int yy)
 
 class b3BroadphasePairSortPredicate
 {
-	public:
-
-		bool operator() ( const b3BroadphasePair& a, const b3BroadphasePair& b ) const
-		{
-			const int uidA0 = a.x;
-			const int uidB0 = b.x;
-			const int uidA1 = a.y;
-			const int uidB1 = b.y;
-			return uidA0 > uidB0 || (uidA0 == uidB0 && uidA1 > uidB1); 
-		}
+public:
+	bool operator()(const b3BroadphasePair& a, const b3BroadphasePair& b) const
+	{
+		const int uidA0 = a.x;
+		const int uidB0 = b.x;
+		const int uidA1 = a.y;
+		const int uidB1 = b.y;
+		return uidA0 > uidB0 || (uidA0 == uidB0 && uidA1 > uidB1);
+	}
 };
 
-B3_FORCE_INLINE bool operator==(const b3BroadphasePair& a, const b3BroadphasePair& b) 
+B3_FORCE_INLINE bool operator==(const b3BroadphasePair& a, const b3BroadphasePair& b)
 {
-	 return (a.x == b.x ) && (a.y == b.y );
+	return (a.x == b.x) && (a.y == b.y);
 }
 
-#endif //B3_OVERLAPPING_PAIR_H
-
+#endif  //B3_OVERLAPPING_PAIR_H
diff --git a/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3OverlappingPairCache.cpp b/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3OverlappingPairCache.cpp
index e4bda61624a..19773244bee 100644
--- a/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3OverlappingPairCache.cpp
+++ b/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3OverlappingPairCache.cpp
@@ -13,8 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #include "b3OverlappingPairCache.h"
 
 //#include "b3Dispatcher.h"
@@ -23,35 +21,26 @@ subject to the following restrictions:
 
 #include <stdio.h>
 
-int	b3g_overlappingPairs = 0;
-int b3g_removePairs =0;
-int b3g_addedPairs =0;
-int b3g_findPairs =0;
+int b3g_overlappingPairs = 0;
+int b3g_removePairs = 0;
+int b3g_addedPairs = 0;
+int b3g_findPairs = 0;
 
-
-
-
-b3HashedOverlappingPairCache::b3HashedOverlappingPairCache():
-	m_overlapFilterCallback(0)
+b3HashedOverlappingPairCache::b3HashedOverlappingPairCache() : m_overlapFilterCallback(0)
 //,	m_blockedForChanges(false)
 {
-	int initialAllocatedSize= 2;
+	int initialAllocatedSize = 2;
 	m_overlappingPairArray.reserve(initialAllocatedSize);
 	growTables();
 }
 
-
-
-
 b3HashedOverlappingPairCache::~b3HashedOverlappingPairCache()
 {
 }
 
-
-
-void	b3HashedOverlappingPairCache::cleanOverlappingPair(b3BroadphasePair& pair,b3Dispatcher* dispatcher)
+void b3HashedOverlappingPairCache::cleanOverlappingPair(b3BroadphasePair& pair, b3Dispatcher* dispatcher)
 {
-/*	if (pair.m_algorithm)
+	/*	if (pair.m_algorithm)
 	{
 		{
 			pair.m_algorithm->~b3CollisionAlgorithm();
@@ -60,91 +49,74 @@ void	b3HashedOverlappingPairCache::cleanOverlappingPair(b3BroadphasePair& pair,b
 		}
 	}
 	*/
-
 }
 
-
-
-
-void	b3HashedOverlappingPairCache::cleanProxyFromPairs(int proxy,b3Dispatcher* dispatcher)
+void b3HashedOverlappingPairCache::cleanProxyFromPairs(int proxy, b3Dispatcher* dispatcher)
 {
-
-	class	CleanPairCallback : public b3OverlapCallback
+	class CleanPairCallback : public b3OverlapCallback
 	{
 		int m_cleanProxy;
-		b3OverlappingPairCache*	m_pairCache;
+		b3OverlappingPairCache* m_pairCache;
 		b3Dispatcher* m_dispatcher;
 
 	public:
-		CleanPairCallback(int cleanProxy,b3OverlappingPairCache* pairCache,b3Dispatcher* dispatcher)
-			:m_cleanProxy(cleanProxy),
-			m_pairCache(pairCache),
-			m_dispatcher(dispatcher)
+		CleanPairCallback(int cleanProxy, b3OverlappingPairCache* pairCache, b3Dispatcher* dispatcher)
+			: m_cleanProxy(cleanProxy),
+			  m_pairCache(pairCache),
+			  m_dispatcher(dispatcher)
 		{
 		}
-		virtual	bool	processOverlap(b3BroadphasePair& pair)
+		virtual bool processOverlap(b3BroadphasePair& pair)
 		{
 			if ((pair.x == m_cleanProxy) ||
 				(pair.y == m_cleanProxy))
 			{
-				m_pairCache->cleanOverlappingPair(pair,m_dispatcher);
+				m_pairCache->cleanOverlappingPair(pair, m_dispatcher);
 			}
 			return false;
 		}
-		
 	};
 
-	CleanPairCallback cleanPairs(proxy,this,dispatcher);
-
-	processAllOverlappingPairs(&cleanPairs,dispatcher);
+	CleanPairCallback cleanPairs(proxy, this, dispatcher);
 
+	processAllOverlappingPairs(&cleanPairs, dispatcher);
 }
 
-
-
-
-void	b3HashedOverlappingPairCache::removeOverlappingPairsContainingProxy(int proxy,b3Dispatcher* dispatcher)
+void b3HashedOverlappingPairCache::removeOverlappingPairsContainingProxy(int proxy, b3Dispatcher* dispatcher)
 {
-
-	class	RemovePairCallback : public b3OverlapCallback
+	class RemovePairCallback : public b3OverlapCallback
 	{
 		int m_obsoleteProxy;
 
 	public:
 		RemovePairCallback(int obsoleteProxy)
-			:m_obsoleteProxy(obsoleteProxy)
+			: m_obsoleteProxy(obsoleteProxy)
 		{
 		}
-		virtual	bool	processOverlap(b3BroadphasePair& pair)
+		virtual bool processOverlap(b3BroadphasePair& pair)
 		{
 			return ((pair.x == m_obsoleteProxy) ||
-				(pair.y == m_obsoleteProxy));
+					(pair.y == m_obsoleteProxy));
 		}
-		
 	};
 
-
 	RemovePairCallback removeCallback(proxy);
 
-	processAllOverlappingPairs(&removeCallback,dispatcher);
+	processAllOverlappingPairs(&removeCallback, dispatcher);
 }
 
-
-
-
-
 b3BroadphasePair* b3HashedOverlappingPairCache::findPair(int proxy0, int proxy1)
 {
 	b3g_findPairs++;
-	if(proxy0 >proxy1) 
-		b3Swap(proxy0,proxy1);
+	if (proxy0 > proxy1)
+		b3Swap(proxy0, proxy1);
 	int proxyId1 = proxy0;
 	int proxyId2 = proxy1;
 
 	/*if (proxyId1 > proxyId2) 
 		b3Swap(proxyId1, proxyId2);*/
 
-	int hash = static_cast<int>(getHash(static_cast<unsigned int>(proxyId1), static_cast<unsigned int>(proxyId2)) & (m_overlappingPairArray.capacity()-1));
+	int hash = static_cast<int>(getHash(static_cast<unsigned int>(proxyId1), static_cast<unsigned int>(proxyId2)) & (m_overlappingPairArray.capacity() - 1));
 
 	if (hash >= m_hashTable.size())
 	{
@@ -169,9 +141,8 @@ b3BroadphasePair* b3HashedOverlappingPairCache::findPair(int proxy0, int proxy1)
 
 //#include <stdio.h>
 
-void	b3HashedOverlappingPairCache::growTables()
+void b3HashedOverlappingPairCache::growTables()
 {
-
 	int newCapacity = m_overlappingPairArray.capacity();
 
 	if (m_hashTable.size() < newCapacity)
@@ -182,10 +153,9 @@ void	b3HashedOverlappingPairCache::growTables()
 		m_hashTable.resize(newCapacity);
 		m_next.resize(newCapacity);
 
-
 		int i;
 
-		for (i= 0; i < newCapacity; ++i)
+		for (i = 0; i < newCapacity; ++i)
 		{
 			m_hashTable[i] = B3_NULL_PAIR;
 		}
@@ -194,35 +164,31 @@ void	b3HashedOverlappingPairCache::growTables()
 			m_next[i] = B3_NULL_PAIR;
 		}
 
-		for(i=0;i<curHashtableSize;i++)
+		for (i = 0; i < curHashtableSize; i++)
 		{
-	
 			const b3BroadphasePair& pair = m_overlappingPairArray[i];
 			int proxyId1 = pair.x;
 			int proxyId2 = pair.y;
 			/*if (proxyId1 > proxyId2) 
 				b3Swap(proxyId1, proxyId2);*/
-			int	hashValue = static_cast<int>(getHash(static_cast<unsigned int>(proxyId1),static_cast<unsigned int>(proxyId2)) & (m_overlappingPairArray.capacity()-1));	// New hash value with new mask
+			int hashValue = static_cast<int>(getHash(static_cast<unsigned int>(proxyId1), static_cast<unsigned int>(proxyId2)) & (m_overlappingPairArray.capacity() - 1));  // New hash value with new mask
 			m_next[i] = m_hashTable[hashValue];
 			m_hashTable[hashValue] = i;
 		}
-
-
 	}
 }
 
 b3BroadphasePair* b3HashedOverlappingPairCache::internalAddPair(int proxy0, int proxy1)
 {
-	if(proxy0>proxy1) 
-		b3Swap(proxy0,proxy1);
+	if (proxy0 > proxy1)
+		b3Swap(proxy0, proxy1);
 	int proxyId1 = proxy0;
 	int proxyId2 = proxy1;
 
 	/*if (proxyId1 > proxyId2) 
 		b3Swap(proxyId1, proxyId2);*/
 
-	int	hash = static_cast<int>(getHash(static_cast<unsigned int>(proxyId1),static_cast<unsigned int>(proxyId2)) & (m_overlappingPairArray.capacity()-1));	// New hash value with new mask
-
+	int hash = static_cast<int>(getHash(static_cast<unsigned int>(proxyId1), static_cast<unsigned int>(proxyId2)) & (m_overlappingPairArray.capacity() - 1));  // New hash value with new mask
 
 	b3BroadphasePair* pair = internalFindPair(proxy0, proxy1, hash);
 	if (pair != NULL)
@@ -243,8 +209,8 @@ b3BroadphasePair* b3HashedOverlappingPairCache::internalAddPair(int proxy0, int
 	pair = &m_overlappingPairArray.expandNonInitializing();
 
 	//this is where we add an actual pair, so also call the 'ghost'
-//	if (m_ghostPairCallback)
-//		m_ghostPairCallback->addOverlappingPair(proxy0,proxy1);
+	//	if (m_ghostPairCallback)
+	//		m_ghostPairCallback->addOverlappingPair(proxy0,proxy1);
 
 	int newCapacity = m_overlappingPairArray.capacity();
 
@@ -252,16 +218,15 @@ b3BroadphasePair* b3HashedOverlappingPairCache::internalAddPair(int proxy0, int
 	{
 		growTables();
 		//hash with new capacity
-		hash = static_cast<int>(getHash(static_cast<unsigned int>(proxyId1),static_cast<unsigned int>(proxyId2)) & (m_overlappingPairArray.capacity()-1));
+		hash = static_cast<int>(getHash(static_cast<unsigned int>(proxyId1), static_cast<unsigned int>(proxyId2)) & (m_overlappingPairArray.capacity() - 1));
 	}
-	
-	*pair = b3MakeBroadphasePair(proxy0,proxy1);
-	
-//	pair->m_pProxy0 = proxy0;
-//	pair->m_pProxy1 = proxy1;
+
+	*pair = b3MakeBroadphasePair(proxy0, proxy1);
+
+	//	pair->m_pProxy0 = proxy0;
+	//	pair->m_pProxy1 = proxy1;
 	//pair->m_algorithm = 0;
 	//pair->m_internalTmpValue = 0;
-	
 
 	m_next[count] = m_hashTable[hash];
 	m_hashTable[hash] = count;
@@ -269,20 +234,18 @@ b3BroadphasePair* b3HashedOverlappingPairCache::internalAddPair(int proxy0, int
 	return pair;
 }
 
-
-
-void* b3HashedOverlappingPairCache::removeOverlappingPair(int proxy0, int proxy1,b3Dispatcher* dispatcher)
+void* b3HashedOverlappingPairCache::removeOverlappingPair(int proxy0, int proxy1, b3Dispatcher* dispatcher)
 {
 	b3g_removePairs++;
-	if(proxy0>proxy1) 
-		b3Swap(proxy0,proxy1);
+	if (proxy0 > proxy1)
+		b3Swap(proxy0, proxy1);
 	int proxyId1 = proxy0;
 	int proxyId2 = proxy1;
 
 	/*if (proxyId1 > proxyId2) 
 		b3Swap(proxyId1, proxyId2);*/
 
-	int	hash = static_cast<int>(getHash(static_cast<unsigned int>(proxyId1),static_cast<unsigned int>(proxyId2)) & (m_overlappingPairArray.capacity()-1));
+	int hash = static_cast<int>(getHash(static_cast<unsigned int>(proxyId1), static_cast<unsigned int>(proxyId2)) & (m_overlappingPairArray.capacity() - 1));
 
 	b3BroadphasePair* pair = internalFindPair(proxy0, proxy1, hash);
 	if (pair == NULL)
@@ -290,9 +253,7 @@ void* b3HashedOverlappingPairCache::removeOverlappingPair(int proxy0, int proxy1
 		return 0;
 	}
 
-	cleanOverlappingPair(*pair,dispatcher);
-
-	
+	cleanOverlappingPair(*pair, dispatcher);
 
 	int pairIndex = int(pair - &m_overlappingPairArray[0]);
 	b3Assert(pairIndex < m_overlappingPairArray.size());
@@ -336,8 +297,8 @@ void* b3HashedOverlappingPairCache::removeOverlappingPair(int proxy0, int proxy1
 
 	// Remove the last pair from the hash table.
 	const b3BroadphasePair* last = &m_overlappingPairArray[lastPairIndex];
-		/* missing swap here too, Nat. */ 
-	int lastHash = static_cast<int>(getHash(static_cast<unsigned int>(last->x), static_cast<unsigned int>(last->y)) & (m_overlappingPairArray.capacity()-1));
+	/* missing swap here too, Nat. */
+	int lastHash = static_cast<int>(getHash(static_cast<unsigned int>(last->x), static_cast<unsigned int>(last->y)) & (m_overlappingPairArray.capacity() - 1));
 
 	index = m_hashTable[lastHash];
 	b3Assert(index != B3_NULL_PAIR);
@@ -372,47 +333,42 @@ void* b3HashedOverlappingPairCache::removeOverlappingPair(int proxy0, int proxy1
 }
 //#include <stdio.h>
 
-void	b3HashedOverlappingPairCache::processAllOverlappingPairs(b3OverlapCallback* callback,b3Dispatcher* dispatcher)
+void b3HashedOverlappingPairCache::processAllOverlappingPairs(b3OverlapCallback* callback, b3Dispatcher* dispatcher)
 {
-
 	int i;
 
-//	printf("m_overlappingPairArray.size()=%d\n",m_overlappingPairArray.size());
-	for (i=0;i<m_overlappingPairArray.size();)
+	//	printf("m_overlappingPairArray.size()=%d\n",m_overlappingPairArray.size());
+	for (i = 0; i < m_overlappingPairArray.size();)
 	{
-	
 		b3BroadphasePair* pair = &m_overlappingPairArray[i];
 		if (callback->processOverlap(*pair))
 		{
-			removeOverlappingPair(pair->x,pair->y,dispatcher);
+			removeOverlappingPair(pair->x, pair->y, dispatcher);
 
 			b3g_overlappingPairs--;
-		} else
+		}
+		else
 		{
 			i++;
 		}
 	}
 }
 
-
-
-
-
-void	b3HashedOverlappingPairCache::sortOverlappingPairs(b3Dispatcher* dispatcher)
+void b3HashedOverlappingPairCache::sortOverlappingPairs(b3Dispatcher* dispatcher)
 {
 	///need to keep hashmap in sync with pair address, so rebuild all
 	b3BroadphasePairArray tmpPairs;
 	int i;
-	for (i=0;i<m_overlappingPairArray.size();i++)
+	for (i = 0; i < m_overlappingPairArray.size(); i++)
 	{
 		tmpPairs.push_back(m_overlappingPairArray[i]);
 	}
 
-	for (i=0;i<tmpPairs.size();i++)
+	for (i = 0; i < tmpPairs.size(); i++)
 	{
-		removeOverlappingPair(tmpPairs[i].x,tmpPairs[i].y,dispatcher);
+		removeOverlappingPair(tmpPairs[i].x, tmpPairs[i].y, dispatcher);
 	}
-	
+
 	for (i = 0; i < m_next.size(); i++)
 	{
 		m_next[i] = B3_NULL_PAIR;
@@ -420,33 +376,29 @@ void	b3HashedOverlappingPairCache::sortOverlappingPairs(b3Dispatcher* dispatcher
 
 	tmpPairs.quickSort(b3BroadphasePairSortPredicate());
 
-	for (i=0;i<tmpPairs.size();i++)
+	for (i = 0; i < tmpPairs.size(); i++)
 	{
-		addOverlappingPair(tmpPairs[i].x ,tmpPairs[i].y);
+		addOverlappingPair(tmpPairs[i].x, tmpPairs[i].y);
 	}
-
-	
 }
 
-
-void*	b3SortedOverlappingPairCache::removeOverlappingPair(int proxy0,int proxy1, b3Dispatcher* dispatcher )
+void* b3SortedOverlappingPairCache::removeOverlappingPair(int proxy0, int proxy1, b3Dispatcher* dispatcher)
 {
 	if (!hasDeferredRemoval())
 	{
-		b3BroadphasePair findPair = b3MakeBroadphasePair(proxy0,proxy1);
-		
+		b3BroadphasePair findPair = b3MakeBroadphasePair(proxy0, proxy1);
 
 		int findIndex = m_overlappingPairArray.findLinearSearch(findPair);
 		if (findIndex < m_overlappingPairArray.size())
 		{
 			b3g_overlappingPairs--;
 			b3BroadphasePair& pair = m_overlappingPairArray[findIndex];
-			
-			cleanOverlappingPair(pair,dispatcher);
+
+			cleanOverlappingPair(pair, dispatcher);
 			//if (m_ghostPairCallback)
 			//	m_ghostPairCallback->removeOverlappingPair(proxy0, proxy1,dispatcher);
-			
-			m_overlappingPairArray.swap(findIndex,m_overlappingPairArray.capacity()-1);
+
+			m_overlappingPairArray.swap(findIndex, m_overlappingPairArray.capacity() - 1);
 			m_overlappingPairArray.pop_back();
 			return 0;
 		}
@@ -455,100 +407,77 @@ void*	b3SortedOverlappingPairCache::removeOverlappingPair(int proxy0,int proxy1,
 	return 0;
 }
 
-
-
-
-
-
-
-
-b3BroadphasePair*	b3SortedOverlappingPairCache::addOverlappingPair(int proxy0,int proxy1)
+b3BroadphasePair* b3SortedOverlappingPairCache::addOverlappingPair(int proxy0, int proxy1)
 {
 	//don't add overlap with own
 	b3Assert(proxy0 != proxy1);
 
-	if (!needsBroadphaseCollision(proxy0,proxy1))
+	if (!needsBroadphaseCollision(proxy0, proxy1))
 		return 0;
-	
+
 	b3BroadphasePair* pair = &m_overlappingPairArray.expandNonInitializing();
-	*pair = b3MakeBroadphasePair(proxy0,proxy1);
-	
-	
+	*pair = b3MakeBroadphasePair(proxy0, proxy1);
+
 	b3g_overlappingPairs++;
 	b3g_addedPairs++;
-	
-//	if (m_ghostPairCallback)
-//		m_ghostPairCallback->addOverlappingPair(proxy0, proxy1);
+
+	//	if (m_ghostPairCallback)
+	//		m_ghostPairCallback->addOverlappingPair(proxy0, proxy1);
 	return pair;
-	
 }
 
 ///this findPair becomes really slow. Either sort the list to speedup the query, or
 ///use a different solution. It is mainly used for Removing overlapping pairs. Removal could be delayed.
 ///we could keep a linked list in each proxy, and store pair in one of the proxies (with lowest memory address)
 ///Also we can use a 2D bitmap, which can be useful for a future GPU implementation
- b3BroadphasePair*	b3SortedOverlappingPairCache::findPair(int proxy0,int proxy1)
+b3BroadphasePair* b3SortedOverlappingPairCache::findPair(int proxy0, int proxy1)
 {
-	if (!needsBroadphaseCollision(proxy0,proxy1))
+	if (!needsBroadphaseCollision(proxy0, proxy1))
 		return 0;
 
-	b3BroadphasePair tmpPair = b3MakeBroadphasePair(proxy0,proxy1);
+	b3BroadphasePair tmpPair = b3MakeBroadphasePair(proxy0, proxy1);
 	int findIndex = m_overlappingPairArray.findLinearSearch(tmpPair);
 
 	if (findIndex < m_overlappingPairArray.size())
 	{
 		//b3Assert(it != m_overlappingPairSet.end());
-		 b3BroadphasePair* pair = &m_overlappingPairArray[findIndex];
+		b3BroadphasePair* pair = &m_overlappingPairArray[findIndex];
 		return pair;
 	}
 	return 0;
 }
 
-
-
-
-
-
-
-
-
-
 //#include <stdio.h>
 
-void	b3SortedOverlappingPairCache::processAllOverlappingPairs(b3OverlapCallback* callback,b3Dispatcher* dispatcher)
+void b3SortedOverlappingPairCache::processAllOverlappingPairs(b3OverlapCallback* callback, b3Dispatcher* dispatcher)
 {
-
 	int i;
 
-	for (i=0;i<m_overlappingPairArray.size();)
+	for (i = 0; i < m_overlappingPairArray.size();)
 	{
-	
 		b3BroadphasePair* pair = &m_overlappingPairArray[i];
 		if (callback->processOverlap(*pair))
 		{
-			cleanOverlappingPair(*pair,dispatcher);
+			cleanOverlappingPair(*pair, dispatcher);
 			pair->x = -1;
 			pair->y = -1;
-			m_overlappingPairArray.swap(i,m_overlappingPairArray.size()-1);
+			m_overlappingPairArray.swap(i, m_overlappingPairArray.size() - 1);
 			m_overlappingPairArray.pop_back();
 			b3g_overlappingPairs--;
-		} else
+		}
+		else
 		{
 			i++;
 		}
 	}
 }
 
-
-
-
-b3SortedOverlappingPairCache::b3SortedOverlappingPairCache():
-	m_blockedForChanges(false),
-	m_hasDeferredRemoval(true),
-	m_overlapFilterCallback(0)
+b3SortedOverlappingPairCache::b3SortedOverlappingPairCache() : m_blockedForChanges(false),
+															   m_hasDeferredRemoval(true),
+															   m_overlapFilterCallback(0)
 
 {
-	int initialAllocatedSize= 2;
+	int initialAllocatedSize = 2;
 	m_overlappingPairArray.reserve(initialAllocatedSize);
 }
 
@@ -556,9 +485,9 @@ b3SortedOverlappingPairCache::~b3SortedOverlappingPairCache()
 {
 }
 
-void	b3SortedOverlappingPairCache::cleanOverlappingPair(b3BroadphasePair& pair,b3Dispatcher* dispatcher)
+void b3SortedOverlappingPairCache::cleanOverlappingPair(b3BroadphasePair& pair, b3Dispatcher* dispatcher)
 {
-/*	if (pair.m_algorithm)
+	/*	if (pair.m_algorithm)
 	{
 		{
 			pair.m_algorithm->~b3CollisionAlgorithm();
@@ -570,69 +499,61 @@ void	b3SortedOverlappingPairCache::cleanOverlappingPair(b3BroadphasePair& pair,b
 	*/
 }
 
-
-void	b3SortedOverlappingPairCache::cleanProxyFromPairs(int proxy,b3Dispatcher* dispatcher)
+void b3SortedOverlappingPairCache::cleanProxyFromPairs(int proxy, b3Dispatcher* dispatcher)
 {
-
-	class	CleanPairCallback : public b3OverlapCallback
+	class CleanPairCallback : public b3OverlapCallback
 	{
 		int m_cleanProxy;
-		b3OverlappingPairCache*	m_pairCache;
+		b3OverlappingPairCache* m_pairCache;
 		b3Dispatcher* m_dispatcher;
 
 	public:
-		CleanPairCallback(int cleanProxy,b3OverlappingPairCache* pairCache,b3Dispatcher* dispatcher)
-			:m_cleanProxy(cleanProxy),
-			m_pairCache(pairCache),
-			m_dispatcher(dispatcher)
+		CleanPairCallback(int cleanProxy, b3OverlappingPairCache* pairCache, b3Dispatcher* dispatcher)
+			: m_cleanProxy(cleanProxy),
+			  m_pairCache(pairCache),
+			  m_dispatcher(dispatcher)
 		{
 		}
-		virtual	bool	processOverlap(b3BroadphasePair& pair)
+		virtual bool processOverlap(b3BroadphasePair& pair)
 		{
 			if ((pair.x == m_cleanProxy) ||
 				(pair.y == m_cleanProxy))
 			{
-				m_pairCache->cleanOverlappingPair(pair,m_dispatcher);
+				m_pairCache->cleanOverlappingPair(pair, m_dispatcher);
 			}
 			return false;
 		}
-		
 	};
 
-	CleanPairCallback cleanPairs(proxy,this,dispatcher);
-
-	processAllOverlappingPairs(&cleanPairs,dispatcher);
+	CleanPairCallback cleanPairs(proxy, this, dispatcher);
 
+	processAllOverlappingPairs(&cleanPairs, dispatcher);
 }
 
-
-void	b3SortedOverlappingPairCache::removeOverlappingPairsContainingProxy(int proxy,b3Dispatcher* dispatcher)
+void b3SortedOverlappingPairCache::removeOverlappingPairsContainingProxy(int proxy, b3Dispatcher* dispatcher)
 {
-
-	class	RemovePairCallback : public b3OverlapCallback
+	class RemovePairCallback : public b3OverlapCallback
 	{
 		int m_obsoleteProxy;
 
 	public:
 		RemovePairCallback(int obsoleteProxy)
-			:m_obsoleteProxy(obsoleteProxy)
+			: m_obsoleteProxy(obsoleteProxy)
 		{
 		}
-		virtual	bool	processOverlap(b3BroadphasePair& pair)
+		virtual bool processOverlap(b3BroadphasePair& pair)
 		{
 			return ((pair.x == m_obsoleteProxy) ||
-				(pair.y == m_obsoleteProxy));
+					(pair.y == m_obsoleteProxy));
 		}
-		
 	};
 
 	RemovePairCallback removeCallback(proxy);
 
-	processAllOverlappingPairs(&removeCallback,dispatcher);
+	processAllOverlappingPairs(&removeCallback, dispatcher);
 }
 
-void	b3SortedOverlappingPairCache::sortOverlappingPairs(b3Dispatcher* dispatcher)
+void b3SortedOverlappingPairCache::sortOverlappingPairs(b3Dispatcher* dispatcher)
 {
 	//should already be sorted
 }
-
diff --git a/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3OverlappingPairCache.h b/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3OverlappingPairCache.h
index f67eb676f1d..f1de1d94eb9 100644
--- a/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3OverlappingPairCache.h
+++ b/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/b3OverlappingPairCache.h
@@ -22,152 +22,136 @@ subject to the following restrictions:
 class b3Dispatcher;
 #include "b3OverlappingPair.h"
 
+typedef b3AlignedObjectArray<b3BroadphasePair> b3BroadphasePairArray;
 
-
-typedef b3AlignedObjectArray<b3BroadphasePair>	b3BroadphasePairArray;
-
-struct	b3OverlapCallback
+struct b3OverlapCallback
 {
 	virtual ~b3OverlapCallback()
-	{}
+	{
+	}
 	//return true for deletion of the pair
-	virtual bool	processOverlap(b3BroadphasePair& pair) = 0;
-
+	virtual bool processOverlap(b3BroadphasePair& pair) = 0;
 };
 
 struct b3OverlapFilterCallback
 {
 	virtual ~b3OverlapFilterCallback()
-	{}
+	{
+	}
 	// return true when pairs need collision
-	virtual bool	needBroadphaseCollision(int proxy0,int proxy1) const = 0;
+	virtual bool needBroadphaseCollision(int proxy0, int proxy1) const = 0;
 };
 
-
-
-
-
-
-
 extern int b3g_removePairs;
 extern int b3g_addedPairs;
 extern int b3g_findPairs;
 
-const int B3_NULL_PAIR=0xffffffff;
+const int B3_NULL_PAIR = 0xffffffff;
 
 ///The b3OverlappingPairCache provides an interface for overlapping pair management (add, remove, storage), used by the b3BroadphaseInterface broadphases.
 ///The b3HashedOverlappingPairCache and b3SortedOverlappingPairCache classes are two implementations.
-class b3OverlappingPairCache 
+class b3OverlappingPairCache
 {
 public:
-	virtual ~b3OverlappingPairCache() {} // this is needed so we can get to the derived class destructor
+	virtual ~b3OverlappingPairCache() {}  // this is needed so we can get to the derived class destructor
 
-	virtual b3BroadphasePair*	getOverlappingPairArrayPtr() = 0;
-	
-	virtual const b3BroadphasePair*	getOverlappingPairArrayPtr() const = 0;
+	virtual b3BroadphasePair* getOverlappingPairArrayPtr() = 0;
 
-	virtual b3BroadphasePairArray&	getOverlappingPairArray() = 0;
+	virtual const b3BroadphasePair* getOverlappingPairArrayPtr() const = 0;
 
-	virtual	void	cleanOverlappingPair(b3BroadphasePair& pair,b3Dispatcher* dispatcher) = 0;
+	virtual b3BroadphasePairArray& getOverlappingPairArray() = 0;
+
+	virtual void cleanOverlappingPair(b3BroadphasePair& pair, b3Dispatcher* dispatcher) = 0;
 
 	virtual int getNumOverlappingPairs() const = 0;
 
-	virtual void	cleanProxyFromPairs(int proxy,b3Dispatcher* dispatcher) = 0;
+	virtual void cleanProxyFromPairs(int proxy, b3Dispatcher* dispatcher) = 0;
 
-	virtual	void setOverlapFilterCallback(b3OverlapFilterCallback* callback) = 0;
+	virtual void setOverlapFilterCallback(b3OverlapFilterCallback* callback) = 0;
 
-	virtual void	processAllOverlappingPairs(b3OverlapCallback*,b3Dispatcher* dispatcher) = 0;
+	virtual void processAllOverlappingPairs(b3OverlapCallback*, b3Dispatcher* dispatcher) = 0;
 
 	virtual b3BroadphasePair* findPair(int proxy0, int proxy1) = 0;
 
-	virtual bool	hasDeferredRemoval() = 0;
+	virtual bool hasDeferredRemoval() = 0;
 
 	//virtual	void	setInternalGhostPairCallback(b3OverlappingPairCallback* ghostPairCallback)=0;
 
-	virtual b3BroadphasePair* 	addOverlappingPair(int proxy0,int proxy1)=0;
-	virtual void*	removeOverlappingPair(int proxy0,int proxy1,b3Dispatcher* dispatcher)=0;
-	virtual void	removeOverlappingPairsContainingProxy(int /*proxy0*/,b3Dispatcher* /*dispatcher*/)=0;
-
-	virtual void	sortOverlappingPairs(b3Dispatcher* dispatcher) = 0;
-
+	virtual b3BroadphasePair* addOverlappingPair(int proxy0, int proxy1) = 0;
+	virtual void* removeOverlappingPair(int proxy0, int proxy1, b3Dispatcher* dispatcher) = 0;
+	virtual void removeOverlappingPairsContainingProxy(int /*proxy0*/, b3Dispatcher* /*dispatcher*/) = 0;
 
+	virtual void sortOverlappingPairs(b3Dispatcher* dispatcher) = 0;
 };
 
 /// Hash-space based Pair Cache, thanks to Erin Catto, Box2D, http://www.box2d.org, and Pierre Terdiman, Codercorner, http://codercorner.com
 class b3HashedOverlappingPairCache : public b3OverlappingPairCache
 {
-	b3BroadphasePairArray	m_overlappingPairArray;
+	b3BroadphasePairArray m_overlappingPairArray;
 	b3OverlapFilterCallback* m_overlapFilterCallback;
-//	bool		m_blockedForChanges;
-
+	//	bool		m_blockedForChanges;
 
 public:
 	b3HashedOverlappingPairCache();
 	virtual ~b3HashedOverlappingPairCache();
 
-	
-	virtual void	removeOverlappingPairsContainingProxy(int proxy,b3Dispatcher* dispatcher);
+	virtual void removeOverlappingPairsContainingProxy(int proxy, b3Dispatcher* dispatcher);
 
-	virtual void*	removeOverlappingPair(int proxy0,int proxy1,b3Dispatcher* dispatcher);
-	
-	B3_FORCE_INLINE bool needsBroadphaseCollision(int proxy0,int proxy1) const
+	virtual void* removeOverlappingPair(int proxy0, int proxy1, b3Dispatcher* dispatcher);
+
+	B3_FORCE_INLINE bool needsBroadphaseCollision(int proxy0, int proxy1) const
 	{
 		if (m_overlapFilterCallback)
-			return m_overlapFilterCallback->needBroadphaseCollision(proxy0,proxy1);
+			return m_overlapFilterCallback->needBroadphaseCollision(proxy0, proxy1);
 
-		bool collides = true;//(proxy0->m_collisionFilterGroup & proxy1->m_collisionFilterMask) != 0;
+		bool collides = true;  //(proxy0->m_collisionFilterGroup & proxy1->m_collisionFilterMask) != 0;
 		//collides = collides && (proxy1->m_collisionFilterGroup & proxy0->m_collisionFilterMask);
-		
+
 		return collides;
 	}
 
 	// Add a pair and return the new pair. If the pair already exists,
 	// no new pair is created and the old one is returned.
-	virtual b3BroadphasePair* 	addOverlappingPair(int proxy0,int proxy1)
+	virtual b3BroadphasePair* addOverlappingPair(int proxy0, int proxy1)
 	{
 		b3g_addedPairs++;
 
-		if (!needsBroadphaseCollision(proxy0,proxy1))
+		if (!needsBroadphaseCollision(proxy0, proxy1))
 			return 0;
 
-		return internalAddPair(proxy0,proxy1);
+		return internalAddPair(proxy0, proxy1);
 	}
 
-	
+	void cleanProxyFromPairs(int proxy, b3Dispatcher* dispatcher);
 
-	void	cleanProxyFromPairs(int proxy,b3Dispatcher* dispatcher);
+	virtual void processAllOverlappingPairs(b3OverlapCallback*, b3Dispatcher* dispatcher);
 
-	
-	virtual void	processAllOverlappingPairs(b3OverlapCallback*,b3Dispatcher* dispatcher);
-
-	virtual b3BroadphasePair*	getOverlappingPairArrayPtr()
+	virtual b3BroadphasePair* getOverlappingPairArrayPtr()
 	{
 		return &m_overlappingPairArray[0];
 	}
 
-	const b3BroadphasePair*	getOverlappingPairArrayPtr() const
+	const b3BroadphasePair* getOverlappingPairArrayPtr() const
 	{
 		return &m_overlappingPairArray[0];
 	}
 
-	b3BroadphasePairArray&	getOverlappingPairArray()
+	b3BroadphasePairArray& getOverlappingPairArray()
 	{
 		return m_overlappingPairArray;
 	}
 
-	const b3BroadphasePairArray&	getOverlappingPairArray() const
+	const b3BroadphasePairArray& getOverlappingPairArray() const
 	{
 		return m_overlappingPairArray;
 	}
 
-	void	cleanOverlappingPair(b3BroadphasePair& pair,b3Dispatcher* dispatcher);
-
-
+	void cleanOverlappingPair(b3BroadphasePair& pair, b3Dispatcher* dispatcher);
 
 	b3BroadphasePair* findPair(int proxy0, int proxy1);
 
 	int GetCount() const { return m_overlappingPairArray.size(); }
-//	b3BroadphasePair* GetPairs() { return m_pairs; }
+	//	b3BroadphasePair* GetPairs() { return m_pairs; }
 
 	b3OverlapFilterCallback* getOverlapFilterCallback()
 	{
@@ -179,19 +163,19 @@ public:
 		m_overlapFilterCallback = callback;
 	}
 
-	int	getNumOverlappingPairs() const
+	int getNumOverlappingPairs() const
 	{
 		return m_overlappingPairArray.size();
 	}
-private:
-	
-	b3BroadphasePair* 	internalAddPair(int proxy0,int proxy1);
 
-	void	growTables();
+private:
+	b3BroadphasePair* internalAddPair(int proxy0, int proxy1);
+
+	void growTables();
 
 	B3_FORCE_INLINE bool equalsPair(const b3BroadphasePair& pair, int proxyId1, int proxyId2)
-	{	
-		return pair.x == proxyId1 && pair.y  == proxyId2;
+	{
+		return pair.x == proxyId1 && pair.y == proxyId2;
 	}
 
 	/*
@@ -210,43 +194,37 @@ private:
 	}
 	*/
 
-
-	
-	B3_FORCE_INLINE	unsigned int getHash(unsigned int proxyId1, unsigned int proxyId2)
+	B3_FORCE_INLINE unsigned int getHash(unsigned int proxyId1, unsigned int proxyId2)
 	{
-		int key = static_cast<int>(((unsigned int)proxyId1) | (((unsigned int)proxyId2) <<16));
+		int key = static_cast<int>(((unsigned int)proxyId1) | (((unsigned int)proxyId2) << 16));
 		// Thomas Wang's hash
 
 		key += ~(key << 15);
-		key ^=  (key >> 10);
-		key +=  (key << 3);
-		key ^=  (key >> 6);
+		key ^= (key >> 10);
+		key += (key << 3);
+		key ^= (key >> 6);
 		key += ~(key << 11);
-		key ^=  (key >> 16);
+		key ^= (key >> 16);
 		return static_cast<unsigned int>(key);
 	}
-	
-
-
-
 
 	B3_FORCE_INLINE b3BroadphasePair* internalFindPair(int proxy0, int proxy1, int hash)
 	{
 		int proxyId1 = proxy0;
 		int proxyId2 = proxy1;
-		#if 0 // wrong, 'equalsPair' use unsorted uids, copy-past devil striked again. Nat.
+#if 0  // wrong, 'equalsPair' use unsorted uids, copy-past devil striked again. Nat.
 		if (proxyId1 > proxyId2) 
 			b3Swap(proxyId1, proxyId2);
-		#endif
+#endif
 
 		int index = m_hashTable[hash];
-		
-		while( index != B3_NULL_PAIR && equalsPair(m_overlappingPairArray[index], proxyId1, proxyId2) == false)
+
+		while (index != B3_NULL_PAIR && equalsPair(m_overlappingPairArray[index], proxyId1, proxyId2) == false)
 		{
 			index = m_next[index];
 		}
 
-		if ( index == B3_NULL_PAIR )
+		if (index == B3_NULL_PAIR)
 		{
 			return NULL;
 		}
@@ -256,161 +234,142 @@ private:
 		return &m_overlappingPairArray[index];
 	}
 
-	virtual bool	hasDeferredRemoval()
+	virtual bool hasDeferredRemoval()
 	{
 		return false;
 	}
 
-/*	virtual	void	setInternalGhostPairCallback(b3OverlappingPairCallback* ghostPairCallback)
+	/*	virtual	void	setInternalGhostPairCallback(b3OverlappingPairCallback* ghostPairCallback)
 	{
 		m_ghostPairCallback = ghostPairCallback;
 	}
 	*/
 
-	virtual void	sortOverlappingPairs(b3Dispatcher* dispatcher);
-	
+	virtual void sortOverlappingPairs(b3Dispatcher* dispatcher);
 
 protected:
-	
-	b3AlignedObjectArray<int>	m_hashTable;
-	b3AlignedObjectArray<int>	m_next;
-//	b3OverlappingPairCallback*	m_ghostPairCallback;
-	
+	b3AlignedObjectArray<int> m_hashTable;
+	b3AlignedObjectArray<int> m_next;
+	//	b3OverlappingPairCallback*	m_ghostPairCallback;
 };
 
-
-
-
 ///b3SortedOverlappingPairCache maintains the objects with overlapping AABB
 ///Typically managed by the Broadphase, Axis3Sweep or b3SimpleBroadphase
-class	b3SortedOverlappingPairCache : public b3OverlappingPairCache
+class b3SortedOverlappingPairCache : public b3OverlappingPairCache
 {
-	protected:
-		//avoid brute-force finding all the time
-		b3BroadphasePairArray	m_overlappingPairArray;
+protected:
+	//avoid brute-force finding all the time
+	b3BroadphasePairArray m_overlappingPairArray;
 
-		//during the dispatch, check that user doesn't destroy/create proxy
-		bool		m_blockedForChanges;
+	//during the dispatch, check that user doesn't destroy/create proxy
+	bool m_blockedForChanges;
 
-		///by default, do the removal during the pair traversal
-		bool		m_hasDeferredRemoval;
-		
-		//if set, use the callback instead of the built in filter in needBroadphaseCollision
-		b3OverlapFilterCallback* m_overlapFilterCallback;
+	///by default, do the removal during the pair traversal
+	bool m_hasDeferredRemoval;
 
-//		b3OverlappingPairCallback*	m_ghostPairCallback;
+	//if set, use the callback instead of the built in filter in needBroadphaseCollision
+	b3OverlapFilterCallback* m_overlapFilterCallback;
 
-	public:
-			
-		b3SortedOverlappingPairCache();	
-		virtual ~b3SortedOverlappingPairCache();
+	//		b3OverlappingPairCallback*	m_ghostPairCallback;
 
-		virtual void	processAllOverlappingPairs(b3OverlapCallback*,b3Dispatcher* dispatcher);
+public:
+	b3SortedOverlappingPairCache();
+	virtual ~b3SortedOverlappingPairCache();
 
-		void*	removeOverlappingPair(int proxy0,int proxy1,b3Dispatcher* dispatcher);
+	virtual void processAllOverlappingPairs(b3OverlapCallback*, b3Dispatcher* dispatcher);
 
-		void	cleanOverlappingPair(b3BroadphasePair& pair,b3Dispatcher* dispatcher);
-		
-		b3BroadphasePair*	addOverlappingPair(int proxy0,int proxy1);
+	void* removeOverlappingPair(int proxy0, int proxy1, b3Dispatcher* dispatcher);
 
-		b3BroadphasePair*	findPair(int proxy0,int proxy1);
-			
-		
-		void	cleanProxyFromPairs(int proxy,b3Dispatcher* dispatcher);
+	void cleanOverlappingPair(b3BroadphasePair& pair, b3Dispatcher* dispatcher);
 
-		virtual void	removeOverlappingPairsContainingProxy(int proxy,b3Dispatcher* dispatcher);
+	b3BroadphasePair* addOverlappingPair(int proxy0, int proxy1);
 
+	b3BroadphasePair* findPair(int proxy0, int proxy1);
 
-		inline bool needsBroadphaseCollision(int proxy0,int proxy1) const
-		{
-			if (m_overlapFilterCallback)
-				return m_overlapFilterCallback->needBroadphaseCollision(proxy0,proxy1);
+	void cleanProxyFromPairs(int proxy, b3Dispatcher* dispatcher);
 
-			bool collides = true;//(proxy0->m_collisionFilterGroup & proxy1->m_collisionFilterMask) != 0;
-			//collides = collides && (proxy1->m_collisionFilterGroup & proxy0->m_collisionFilterMask);
-			
-			return collides;
-		}
-		
-		b3BroadphasePairArray&	getOverlappingPairArray()
-		{
-			return m_overlappingPairArray;
-		}
+	virtual void removeOverlappingPairsContainingProxy(int proxy, b3Dispatcher* dispatcher);
 
-		const b3BroadphasePairArray&	getOverlappingPairArray() const
-		{
-			return m_overlappingPairArray;
-		}
+	inline bool needsBroadphaseCollision(int proxy0, int proxy1) const
+	{
+		if (m_overlapFilterCallback)
+			return m_overlapFilterCallback->needBroadphaseCollision(proxy0, proxy1);
 
-		
+		bool collides = true;  //(proxy0->m_collisionFilterGroup & proxy1->m_collisionFilterMask) != 0;
+		//collides = collides && (proxy1->m_collisionFilterGroup & proxy0->m_collisionFilterMask);
 
+		return collides;
+	}
 
-		b3BroadphasePair*	getOverlappingPairArrayPtr()
-		{
-			return &m_overlappingPairArray[0];
-		}
+	b3BroadphasePairArray& getOverlappingPairArray()
+	{
+		return m_overlappingPairArray;
+	}
 
-		const b3BroadphasePair*	getOverlappingPairArrayPtr() const
-		{
-			return &m_overlappingPairArray[0];
-		}
+	const b3BroadphasePairArray& getOverlappingPairArray() const
+	{
+		return m_overlappingPairArray;
+	}
 
-		int	getNumOverlappingPairs() const
-		{
-			return m_overlappingPairArray.size();
-		}
-		
-		b3OverlapFilterCallback* getOverlapFilterCallback()
-		{
-			return m_overlapFilterCallback;
-		}
+	b3BroadphasePair* getOverlappingPairArrayPtr()
+	{
+		return &m_overlappingPairArray[0];
+	}
 
-		void setOverlapFilterCallback(b3OverlapFilterCallback* callback)
-		{
-			m_overlapFilterCallback = callback;
-		}
+	const b3BroadphasePair* getOverlappingPairArrayPtr() const
+	{
+		return &m_overlappingPairArray[0];
+	}
 
-		virtual bool	hasDeferredRemoval()
-		{
-			return m_hasDeferredRemoval;
-		}
+	int getNumOverlappingPairs() const
+	{
+		return m_overlappingPairArray.size();
+	}
 
-/*		virtual	void	setInternalGhostPairCallback(b3OverlappingPairCallback* ghostPairCallback)
+	b3OverlapFilterCallback* getOverlapFilterCallback()
+	{
+		return m_overlapFilterCallback;
+	}
+
+	void setOverlapFilterCallback(b3OverlapFilterCallback* callback)
+	{
+		m_overlapFilterCallback = callback;
+	}
+
+	virtual bool hasDeferredRemoval()
+	{
+		return m_hasDeferredRemoval;
+	}
+
+	/*		virtual	void	setInternalGhostPairCallback(b3OverlappingPairCallback* ghostPairCallback)
 		{
 			m_ghostPairCallback = ghostPairCallback;
 		}
 		*/
-		virtual void	sortOverlappingPairs(b3Dispatcher* dispatcher);
-		
-
+	virtual void sortOverlappingPairs(b3Dispatcher* dispatcher);
 };
 
-
-
 ///b3NullPairCache skips add/removal of overlapping pairs. Userful for benchmarking and unit testing.
 class b3NullPairCache : public b3OverlappingPairCache
 {
-
-	b3BroadphasePairArray	m_overlappingPairArray;
+	b3BroadphasePairArray m_overlappingPairArray;
 
 public:
-
-	virtual b3BroadphasePair*	getOverlappingPairArrayPtr()
+	virtual b3BroadphasePair* getOverlappingPairArrayPtr()
 	{
 		return &m_overlappingPairArray[0];
 	}
-	const b3BroadphasePair*	getOverlappingPairArrayPtr() const
+	const b3BroadphasePair* getOverlappingPairArrayPtr() const
 	{
 		return &m_overlappingPairArray[0];
 	}
-	b3BroadphasePairArray&	getOverlappingPairArray()
+	b3BroadphasePairArray& getOverlappingPairArray()
 	{
 		return m_overlappingPairArray;
 	}
-	
-	virtual	void	cleanOverlappingPair(b3BroadphasePair& /*pair*/,b3Dispatcher* /*dispatcher*/)
-	{
 
+	virtual void cleanOverlappingPair(b3BroadphasePair& /*pair*/, b3Dispatcher* /*dispatcher*/)
+	{
 	}
 
 	virtual int getNumOverlappingPairs() const
@@ -418,16 +377,15 @@ public:
 		return 0;
 	}
 
-	virtual void	cleanProxyFromPairs(int /*proxy*/,b3Dispatcher* /*dispatcher*/)
-	{
-
-	}
-
-	virtual	void setOverlapFilterCallback(b3OverlapFilterCallback* /*callback*/)
+	virtual void cleanProxyFromPairs(int /*proxy*/, b3Dispatcher* /*dispatcher*/)
 	{
 	}
 
-	virtual void	processAllOverlappingPairs(b3OverlapCallback*,b3Dispatcher* /*dispatcher*/)
+	virtual void setOverlapFilterCallback(b3OverlapFilterCallback* /*callback*/)
+	{
+	}
+
+	virtual void processAllOverlappingPairs(b3OverlapCallback*, b3Dispatcher* /*dispatcher*/)
 	{
 	}
 
@@ -436,39 +394,34 @@ public:
 		return 0;
 	}
 
-	virtual bool	hasDeferredRemoval()
+	virtual bool hasDeferredRemoval()
 	{
 		return true;
 	}
 
-//	virtual	void	setInternalGhostPairCallback(b3OverlappingPairCallback* /* ghostPairCallback */)
-//	{
-//
-//	}
+	//	virtual	void	setInternalGhostPairCallback(b3OverlappingPairCallback* /* ghostPairCallback */)
+	//	{
+	//
+	//	}
 
-	virtual b3BroadphasePair*	addOverlappingPair(int /*proxy0*/,int /*proxy1*/)
+	virtual b3BroadphasePair* addOverlappingPair(int /*proxy0*/, int /*proxy1*/)
 	{
 		return 0;
 	}
 
-	virtual void*	removeOverlappingPair(int /*proxy0*/,int /*proxy1*/,b3Dispatcher* /*dispatcher*/)
+	virtual void* removeOverlappingPair(int /*proxy0*/, int /*proxy1*/, b3Dispatcher* /*dispatcher*/)
 	{
 		return 0;
 	}
 
-	virtual void	removeOverlappingPairsContainingProxy(int /*proxy0*/,b3Dispatcher* /*dispatcher*/)
+	virtual void removeOverlappingPairsContainingProxy(int /*proxy0*/, b3Dispatcher* /*dispatcher*/)
 	{
 	}
-	
-	virtual void	sortOverlappingPairs(b3Dispatcher* dispatcher)
+
+	virtual void sortOverlappingPairs(b3Dispatcher* dispatcher)
 	{
-        (void) dispatcher;
+		(void)dispatcher;
 	}
-
-
 };
 
-
-#endif //B3_OVERLAPPING_PAIR_CACHE_H
-
-
+#endif  //B3_OVERLAPPING_PAIR_CACHE_H
diff --git a/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/shared/b3Aabb.h b/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/shared/b3Aabb.h
index 7f9bf990bfc..343a2c0e215 100644
--- a/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/shared/b3Aabb.h
+++ b/thirdparty/bullet/Bullet3Collision/BroadPhaseCollision/shared/b3Aabb.h
@@ -2,7 +2,6 @@
 #ifndef B3_AABB_H
 #define B3_AABB_H
 
-
 #include "Bullet3Common/shared/b3Float4.h"
 #include "Bullet3Common/shared/b3Mat3x3.h"
 
@@ -10,44 +9,42 @@ typedef struct b3Aabb b3Aabb_t;
 
 struct b3Aabb
 {
-	union
-	{
+	union {
 		float m_min[4];
 		b3Float4 m_minVec;
 		int m_minIndices[4];
 	};
-	union
-	{
-		float	m_max[4];
+	union {
+		float m_max[4];
 		b3Float4 m_maxVec;
 		int m_signedMaxIndices[4];
 	};
 };
 
-inline void b3TransformAabb2(b3Float4ConstArg localAabbMin,b3Float4ConstArg localAabbMax, float margin,
-						b3Float4ConstArg pos,
-						b3QuatConstArg orn,
-						b3Float4* aabbMinOut,b3Float4* aabbMaxOut)
+inline void b3TransformAabb2(b3Float4ConstArg localAabbMin, b3Float4ConstArg localAabbMax, float margin,
+							 b3Float4ConstArg pos,
+							 b3QuatConstArg orn,
+							 b3Float4* aabbMinOut, b3Float4* aabbMaxOut)
 {
-		b3Float4 localHalfExtents = 0.5f*(localAabbMax-localAabbMin);
-		localHalfExtents+=b3MakeFloat4(margin,margin,margin,0.f);
-		b3Float4 localCenter = 0.5f*(localAabbMax+localAabbMin);
-		b3Mat3x3 m;
-		m = b3QuatGetRotationMatrix(orn);
-		b3Mat3x3 abs_b = b3AbsoluteMat3x3(m);
-		b3Float4 center = b3TransformPoint(localCenter,pos,orn);
-		
-		b3Float4 extent = b3MakeFloat4(b3Dot3F4(localHalfExtents,b3GetRow(abs_b,0)),
-										 b3Dot3F4(localHalfExtents,b3GetRow(abs_b,1)),
-										 b3Dot3F4(localHalfExtents,b3GetRow(abs_b,2)),
-										 0.f);
-		*aabbMinOut = center-extent;
-		*aabbMaxOut = center+extent;
+	b3Float4 localHalfExtents = 0.5f * (localAabbMax - localAabbMin);
+	localHalfExtents += b3MakeFloat4(margin, margin, margin, 0.f);
+	b3Float4 localCenter = 0.5f * (localAabbMax + localAabbMin);
+	b3Mat3x3 m;
+	m = b3QuatGetRotationMatrix(orn);
+	b3Mat3x3 abs_b = b3AbsoluteMat3x3(m);
+	b3Float4 center = b3TransformPoint(localCenter, pos, orn);
+
+	b3Float4 extent = b3MakeFloat4(b3Dot3F4(localHalfExtents, b3GetRow(abs_b, 0)),
+								   b3Dot3F4(localHalfExtents, b3GetRow(abs_b, 1)),
+								   b3Dot3F4(localHalfExtents, b3GetRow(abs_b, 2)),
+								   0.f);
+	*aabbMinOut = center - extent;
+	*aabbMaxOut = center + extent;
 }
 
 /// conservative test for overlap between two aabbs
-inline bool b3TestAabbAgainstAabb(b3Float4ConstArg aabbMin1,b3Float4ConstArg aabbMax1,
-								b3Float4ConstArg aabbMin2, b3Float4ConstArg aabbMax2)
+inline bool b3TestAabbAgainstAabb(b3Float4ConstArg aabbMin1, b3Float4ConstArg aabbMax1,
+								  b3Float4ConstArg aabbMin2, b3Float4ConstArg aabbMax2)
 {
 	bool overlap = true;
 	overlap = (aabbMin1.x > aabbMax2.x || aabbMax1.x < aabbMin2.x) ? false : overlap;
@@ -56,4 +53,4 @@ inline bool b3TestAabbAgainstAabb(b3Float4ConstArg aabbMin1,b3Float4ConstArg aab
 	return overlap;
 }
 
-#endif //B3_AABB_H
+#endif  //B3_AABB_H
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3Config.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3Config.h
index 65d4a21613a..518da89c541 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3Config.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3Config.h
@@ -1,41 +1,39 @@
 #ifndef B3_CONFIG_H
 #define B3_CONFIG_H
 
-struct	b3Config
+struct b3Config
 {
-	int	m_maxConvexBodies;
-	int	m_maxConvexShapes;
-	int	m_maxBroadphasePairs;
+	int m_maxConvexBodies;
+	int m_maxConvexShapes;
+	int m_maxBroadphasePairs;
 	int m_maxContactCapacity;
 	int m_compoundPairCapacity;
 
 	int m_maxVerticesPerFace;
 	int m_maxFacesPerShape;
-	int	m_maxConvexVertices;
+	int m_maxConvexVertices;
 	int m_maxConvexIndices;
 	int m_maxConvexUniqueEdges;
-	
-	int	m_maxCompoundChildShapes;
-	
+
+	int m_maxCompoundChildShapes;
+
 	int m_maxTriConvexPairCapacity;
 
 	b3Config()
-		:m_maxConvexBodies(128*1024),
-		m_maxVerticesPerFace(64),
-		m_maxFacesPerShape(12),
-		m_maxConvexVertices(8192),
-		m_maxConvexIndices(81920),
-		m_maxConvexUniqueEdges(8192),
-		m_maxCompoundChildShapes(8192),
-		m_maxTriConvexPairCapacity(256*1024)
+		: m_maxConvexBodies(128 * 1024),
+		  m_maxVerticesPerFace(64),
+		  m_maxFacesPerShape(12),
+		  m_maxConvexVertices(8192),
+		  m_maxConvexIndices(81920),
+		  m_maxConvexUniqueEdges(8192),
+		  m_maxCompoundChildShapes(8192),
+		  m_maxTriConvexPairCapacity(256 * 1024)
 	{
 		m_maxConvexShapes = m_maxConvexBodies;
-		m_maxBroadphasePairs = 16*m_maxConvexBodies;
+		m_maxBroadphasePairs = 16 * m_maxConvexBodies;
 		m_maxContactCapacity = m_maxBroadphasePairs;
-		m_compoundPairCapacity = 1024*1024;
+		m_compoundPairCapacity = 1024 * 1024;
 	}
 };
 
-
-#endif//B3_CONFIG_H
-
+#endif  //B3_CONFIG_H
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3Contact4.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3Contact4.h
index fb251656731..c2cd3c729b3 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3Contact4.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3Contact4.h
@@ -19,28 +19,37 @@ subject to the following restrictions:
 #include "Bullet3Common/b3Vector3.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3Contact4Data.h"
 
-B3_ATTRIBUTE_ALIGNED16(struct) b3Contact4 : public b3Contact4Data
+B3_ATTRIBUTE_ALIGNED16(struct)
+b3Contact4 : public b3Contact4Data
 {
 	B3_DECLARE_ALIGNED_ALLOCATOR();
 
-	int getBodyA()const {return abs(m_bodyAPtrAndSignBit);}
-	int getBodyB()const {return abs(m_bodyBPtrAndSignBit);}
-	bool isBodyAFixed()const { return m_bodyAPtrAndSignBit<0;}
-	bool isBodyBFixed()const { return m_bodyBPtrAndSignBit<0;}
+	int getBodyA() const { return abs(m_bodyAPtrAndSignBit); }
+	int getBodyB() const { return abs(m_bodyBPtrAndSignBit); }
+	bool isBodyAFixed() const { return m_bodyAPtrAndSignBit < 0; }
+	bool isBodyBFixed() const { return m_bodyBPtrAndSignBit < 0; }
 	//	todo. make it safer
 	int& getBatchIdx() { return m_batchIdx; }
 	const int& getBatchIdx() const { return m_batchIdx; }
-	float getRestituitionCoeff() const { return ((float)m_restituitionCoeffCmp/(float)0xffff); }
-	void setRestituitionCoeff( float c ) { b3Assert( c >= 0.f && c <= 1.f ); m_restituitionCoeffCmp = (unsigned short)(c*0xffff); }
-	float getFrictionCoeff() const { return ((float)m_frictionCoeffCmp/(float)0xffff); }
-	void setFrictionCoeff( float c ) { b3Assert( c >= 0.f && c <= 1.f ); m_frictionCoeffCmp = (unsigned short)(c*0xffff); }
+	float getRestituitionCoeff() const { return ((float)m_restituitionCoeffCmp / (float)0xffff); }
+	void setRestituitionCoeff(float c)
+	{
+		b3Assert(c >= 0.f && c <= 1.f);
+		m_restituitionCoeffCmp = (unsigned short)(c * 0xffff);
+	}
+	float getFrictionCoeff() const { return ((float)m_frictionCoeffCmp / (float)0xffff); }
+	void setFrictionCoeff(float c)
+	{
+		b3Assert(c >= 0.f && c <= 1.f);
+		m_frictionCoeffCmp = (unsigned short)(c * 0xffff);
+	}
 
 	//float& getNPoints() { return m_worldNormal[3]; }
-	int getNPoints() const { return (int) m_worldNormalOnB.w; }
+	int getNPoints() const { return (int)m_worldNormalOnB.w; }
 
 	float getPenetration(int idx) const { return m_worldPosB[idx].w; }
 
-	bool isInvalid() const { return (getBodyA()==0 || getBodyB()==0); }
+	bool isInvalid() const { return (getBodyA() == 0 || getBodyB() == 0); }
 };
 
-#endif //B3_CONTACT4_H
+#endif  //B3_CONTACT4_H
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3ConvexUtility.cpp b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3ConvexUtility.cpp
index 55706fa6313..a5dab74a344 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3ConvexUtility.cpp
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3ConvexUtility.cpp
@@ -13,52 +13,42 @@ subject to the following restrictions:
 */
 //Originally written by Erwin Coumans
 
-
 #include "b3ConvexUtility.h"
 #include "Bullet3Geometry/b3ConvexHullComputer.h"
 #include "Bullet3Geometry/b3GrahamScan2dConvexHull.h"
 #include "Bullet3Common/b3Quaternion.h"
 #include "Bullet3Common/b3HashMap.h"
 
-
-
-
-
 b3ConvexUtility::~b3ConvexUtility()
 {
 }
 
-bool	b3ConvexUtility::initializePolyhedralFeatures(const b3Vector3* orgVertices, int numPoints, bool mergeCoplanarTriangles)
+bool b3ConvexUtility::initializePolyhedralFeatures(const b3Vector3* orgVertices, int numPoints, bool mergeCoplanarTriangles)
 {
-	
-	
-
 	b3ConvexHullComputer conv;
-	conv.compute(&orgVertices[0].getX(), sizeof(b3Vector3),numPoints,0.f,0.f);
+	conv.compute(&orgVertices[0].getX(), sizeof(b3Vector3), numPoints, 0.f, 0.f);
 
 	b3AlignedObjectArray<b3Vector3> faceNormals;
 	int numFaces = conv.faces.size();
 	faceNormals.resize(numFaces);
 	b3ConvexHullComputer* convexUtil = &conv;
 
-	
-	b3AlignedObjectArray<b3MyFace>	tmpFaces;
+	b3AlignedObjectArray<b3MyFace> tmpFaces;
 	tmpFaces.resize(numFaces);
 
 	int numVertices = convexUtil->vertices.size();
 	m_vertices.resize(numVertices);
-	for (int p=0;p<numVertices;p++)
+	for (int p = 0; p < numVertices; p++)
 	{
 		m_vertices[p] = convexUtil->vertices[p];
 	}
 
-
-	for (int i=0;i<numFaces;i++)
+	for (int i = 0; i < numFaces; i++)
 	{
 		int face = convexUtil->faces[i];
 		//printf("face=%d\n",face);
-		const b3ConvexHullComputer::Edge*  firstEdge = &convexUtil->edges[face];
-		const b3ConvexHullComputer::Edge*  edge = firstEdge;
+		const b3ConvexHullComputer::Edge* firstEdge = &convexUtil->edges[face];
+		const b3ConvexHullComputer::Edge* edge = firstEdge;
 
 		b3Vector3 edges[3];
 		int numEdges = 0;
@@ -66,25 +56,23 @@ bool	b3ConvexUtility::initializePolyhedralFeatures(const b3Vector3* orgVertices,
 
 		do
 		{
-			
 			int src = edge->getSourceVertex();
 			tmpFaces[i].m_indices.push_back(src);
 			int targ = edge->getTargetVertex();
 			b3Vector3 wa = convexUtil->vertices[src];
 
 			b3Vector3 wb = convexUtil->vertices[targ];
-			b3Vector3 newEdge = wb-wa;
+			b3Vector3 newEdge = wb - wa;
 			newEdge.normalize();
-			if (numEdges<2)
+			if (numEdges < 2)
 				edges[numEdges++] = newEdge;
 
 			edge = edge->getNextEdgeOfFace();
-		} while (edge!=firstEdge);
+		} while (edge != firstEdge);
 
 		b3Scalar planeEq = 1e30f;
 
-		
-		if (numEdges==2)
+		if (numEdges == 2)
 		{
 			faceNormals[i] = edges[0].cross(edges[1]);
 			faceNormals[i].normalize();
@@ -92,20 +80,19 @@ bool	b3ConvexUtility::initializePolyhedralFeatures(const b3Vector3* orgVertices,
 			tmpFaces[i].m_plane[1] = faceNormals[i].getY();
 			tmpFaces[i].m_plane[2] = faceNormals[i].getZ();
 			tmpFaces[i].m_plane[3] = planeEq;
-
 		}
 		else
 		{
-			b3Assert(0);//degenerate?
+			b3Assert(0);  //degenerate?
 			faceNormals[i].setZero();
 		}
 
-		for (int v=0;v<tmpFaces[i].m_indices.size();v++)
+		for (int v = 0; v < tmpFaces[i].m_indices.size(); v++)
 		{
 			b3Scalar eq = m_vertices[tmpFaces[i].m_indices[v]].dot(faceNormals[i]);
-			if (planeEq>eq)
+			if (planeEq > eq)
 			{
-				planeEq=eq;
+				planeEq = eq;
 			}
 		}
 		tmpFaces[i].m_plane[3] = -planeEq;
@@ -113,89 +100,86 @@ bool	b3ConvexUtility::initializePolyhedralFeatures(const b3Vector3* orgVertices,
 
 	//merge coplanar faces and copy them to m_polyhedron
 
-	b3Scalar faceWeldThreshold= 0.999f;
+	b3Scalar faceWeldThreshold = 0.999f;
 	b3AlignedObjectArray<int> todoFaces;
-	for (int i=0;i<tmpFaces.size();i++)
+	for (int i = 0; i < tmpFaces.size(); i++)
 		todoFaces.push_back(i);
 
 	while (todoFaces.size())
 	{
 		b3AlignedObjectArray<int> coplanarFaceGroup;
-		int refFace = todoFaces[todoFaces.size()-1];
+		int refFace = todoFaces[todoFaces.size() - 1];
 
 		coplanarFaceGroup.push_back(refFace);
 		b3MyFace& faceA = tmpFaces[refFace];
 		todoFaces.pop_back();
 
-		b3Vector3 faceNormalA = b3MakeVector3(faceA.m_plane[0],faceA.m_plane[1],faceA.m_plane[2]);
-		for (int j=todoFaces.size()-1;j>=0;j--)
+		b3Vector3 faceNormalA = b3MakeVector3(faceA.m_plane[0], faceA.m_plane[1], faceA.m_plane[2]);
+		for (int j = todoFaces.size() - 1; j >= 0; j--)
 		{
 			int i = todoFaces[j];
 			b3MyFace& faceB = tmpFaces[i];
-			b3Vector3 faceNormalB = b3MakeVector3(faceB.m_plane[0],faceB.m_plane[1],faceB.m_plane[2]);
-			if (faceNormalA.dot(faceNormalB)>faceWeldThreshold)
+			b3Vector3 faceNormalB = b3MakeVector3(faceB.m_plane[0], faceB.m_plane[1], faceB.m_plane[2]);
+			if (faceNormalA.dot(faceNormalB) > faceWeldThreshold)
 			{
 				coplanarFaceGroup.push_back(i);
 				todoFaces.remove(i);
 			}
 		}
 
-
 		bool did_merge = false;
-		if (coplanarFaceGroup.size()>1)
+		if (coplanarFaceGroup.size() > 1)
 		{
 			//do the merge: use Graham Scan 2d convex hull
 
 			b3AlignedObjectArray<b3GrahamVector3> orgpoints;
-			b3Vector3 averageFaceNormal = b3MakeVector3(0,0,0);
+			b3Vector3 averageFaceNormal = b3MakeVector3(0, 0, 0);
 
-			for (int i=0;i<coplanarFaceGroup.size();i++)
+			for (int i = 0; i < coplanarFaceGroup.size(); i++)
 			{
-//				m_polyhedron->m_faces.push_back(tmpFaces[coplanarFaceGroup[i]]);
+				//				m_polyhedron->m_faces.push_back(tmpFaces[coplanarFaceGroup[i]]);
 
 				b3MyFace& face = tmpFaces[coplanarFaceGroup[i]];
-				b3Vector3 faceNormal = b3MakeVector3(face.m_plane[0],face.m_plane[1],face.m_plane[2]);
-				averageFaceNormal+=faceNormal;
-				for (int f=0;f<face.m_indices.size();f++)
+				b3Vector3 faceNormal = b3MakeVector3(face.m_plane[0], face.m_plane[1], face.m_plane[2]);
+				averageFaceNormal += faceNormal;
+				for (int f = 0; f < face.m_indices.size(); f++)
 				{
 					int orgIndex = face.m_indices[f];
 					b3Vector3 pt = m_vertices[orgIndex];
-					
+
 					bool found = false;
 
-					for (int i=0;i<orgpoints.size();i++)
+					for (int i = 0; i < orgpoints.size(); i++)
 					{
 						//if ((orgpoints[i].m_orgIndex == orgIndex) || ((rotatedPt-orgpoints[i]).length2()<0.0001))
 						if (orgpoints[i].m_orgIndex == orgIndex)
 						{
-							found=true;
+							found = true;
 							break;
 						}
 					}
 					if (!found)
-						orgpoints.push_back(b3GrahamVector3(pt,orgIndex));
+						orgpoints.push_back(b3GrahamVector3(pt, orgIndex));
 				}
 			}
 
-			
-
 			b3MyFace combinedFace;
-			for (int i=0;i<4;i++)
+			for (int i = 0; i < 4; i++)
 				combinedFace.m_plane[i] = tmpFaces[coplanarFaceGroup[0]].m_plane[i];
 
 			b3AlignedObjectArray<b3GrahamVector3> hull;
 
 			averageFaceNormal.normalize();
-			b3GrahamScanConvexHull2D(orgpoints,hull,averageFaceNormal);
+			b3GrahamScanConvexHull2D(orgpoints, hull, averageFaceNormal);
 
-			for (int i=0;i<hull.size();i++)
+			for (int i = 0; i < hull.size(); i++)
 			{
 				combinedFace.m_indices.push_back(hull[i].m_orgIndex);
-				for(int k = 0; k < orgpoints.size(); k++) 
+				for (int k = 0; k < orgpoints.size(); k++)
 				{
-					if(orgpoints[k].m_orgIndex == hull[i].m_orgIndex) 
+					if (orgpoints[k].m_orgIndex == hull[i].m_orgIndex)
 					{
-						orgpoints[k].m_orgIndex = -1; // invalidate...
+						orgpoints[k].m_orgIndex = -1;  // invalidate...
 						break;
 					}
 				}
@@ -203,38 +187,41 @@ bool	b3ConvexUtility::initializePolyhedralFeatures(const b3Vector3* orgVertices,
 
 			// are there rejected vertices?
 			bool reject_merge = false;
-			
 
-
-			for(int i = 0; i < orgpoints.size(); i++) {
-				if(orgpoints[i].m_orgIndex == -1)
-					continue; // this is in the hull...
+			for (int i = 0; i < orgpoints.size(); i++)
+			{
+				if (orgpoints[i].m_orgIndex == -1)
+					continue;  // this is in the hull...
 				// this vertex is rejected -- is anybody else using this vertex?
-				for(int j = 0; j < tmpFaces.size(); j++) {
-					
+				for (int j = 0; j < tmpFaces.size(); j++)
+				{
 					b3MyFace& face = tmpFaces[j];
 					// is this a face of the current coplanar group?
 					bool is_in_current_group = false;
-					for(int k = 0; k < coplanarFaceGroup.size(); k++) {
-						if(coplanarFaceGroup[k] == j) {
+					for (int k = 0; k < coplanarFaceGroup.size(); k++)
+					{
+						if (coplanarFaceGroup[k] == j)
+						{
 							is_in_current_group = true;
 							break;
 						}
 					}
-					if(is_in_current_group) // ignore this face...
+					if (is_in_current_group)  // ignore this face...
 						continue;
 					// does this face use this rejected vertex?
-					for(int v = 0; v < face.m_indices.size(); v++) {
-						if(face.m_indices[v] == orgpoints[i].m_orgIndex) {
+					for (int v = 0; v < face.m_indices.size(); v++)
+					{
+						if (face.m_indices[v] == orgpoints[i].m_orgIndex)
+						{
 							// this rejected vertex is used in another face -- reject merge
 							reject_merge = true;
 							break;
 						}
 					}
-					if(reject_merge)
+					if (reject_merge)
 						break;
 				}
-				if(reject_merge)
+				if (reject_merge)
 					break;
 			}
 
@@ -245,18 +232,14 @@ bool	b3ConvexUtility::initializePolyhedralFeatures(const b3Vector3* orgVertices,
 				m_faces.push_back(combinedFace);
 			}
 		}
-		if(!did_merge)
+		if (!did_merge)
 		{
-			for (int i=0;i<coplanarFaceGroup.size();i++)
+			for (int i = 0; i < coplanarFaceGroup.size(); i++)
 			{
 				b3MyFace face = tmpFaces[coplanarFaceGroup[i]];
 				m_faces.push_back(face);
 			}
-
-		} 
-
-
-
+		}
 	}
 
 	initialize();
@@ -264,43 +247,38 @@ bool	b3ConvexUtility::initializePolyhedralFeatures(const b3Vector3* orgVertices,
 	return true;
 }
 
-
-
-
-
-
 inline bool IsAlmostZero(const b3Vector3& v)
 {
-	if(fabsf(v.getX())>1e-6 || fabsf(v.getY())>1e-6 || fabsf(v.getZ())>1e-6)	return false;
+	if (fabsf(v.getX()) > 1e-6 || fabsf(v.getY()) > 1e-6 || fabsf(v.getZ()) > 1e-6) return false;
 	return true;
 }
 
 struct b3InternalVertexPair
 {
-	b3InternalVertexPair(short int v0,short int v1)
-		:m_v0(v0),
-		m_v1(v1)
+	b3InternalVertexPair(short int v0, short int v1)
+		: m_v0(v0),
+		  m_v1(v1)
 	{
-		if (m_v1>m_v0)
-			b3Swap(m_v0,m_v1);
+		if (m_v1 > m_v0)
+			b3Swap(m_v0, m_v1);
 	}
 	short int m_v0;
 	short int m_v1;
 	int getHash() const
 	{
-		return m_v0+(m_v1<<16);
+		return m_v0 + (m_v1 << 16);
 	}
 	bool equals(const b3InternalVertexPair& other) const
 	{
-		return m_v0==other.m_v0 && m_v1==other.m_v1;
+		return m_v0 == other.m_v0 && m_v1 == other.m_v1;
 	}
 };
 
 struct b3InternalEdge
 {
 	b3InternalEdge()
-		:m_face0(-1),
-		m_face1(-1)
+		: m_face0(-1),
+		  m_face1(-1)
 	{
 	}
 	short int m_face0;
@@ -312,23 +290,31 @@ struct b3InternalEdge
 #ifdef TEST_INTERNAL_OBJECTS
 bool b3ConvexUtility::testContainment() const
 {
-	for(int p=0;p<8;p++)
+	for (int p = 0; p < 8; p++)
 	{
 		b3Vector3 LocalPt;
-		if(p==0)		LocalPt = m_localCenter + b3Vector3(m_extents[0], m_extents[1], m_extents[2]);
-		else if(p==1)	LocalPt = m_localCenter + b3Vector3(m_extents[0], m_extents[1], -m_extents[2]);
-		else if(p==2)	LocalPt = m_localCenter + b3Vector3(m_extents[0], -m_extents[1], m_extents[2]);
-		else if(p==3)	LocalPt = m_localCenter + b3Vector3(m_extents[0], -m_extents[1], -m_extents[2]);
-		else if(p==4)	LocalPt = m_localCenter + b3Vector3(-m_extents[0], m_extents[1], m_extents[2]);
-		else if(p==5)	LocalPt = m_localCenter + b3Vector3(-m_extents[0], m_extents[1], -m_extents[2]);
-		else if(p==6)	LocalPt = m_localCenter + b3Vector3(-m_extents[0], -m_extents[1], m_extents[2]);
-		else if(p==7)	LocalPt = m_localCenter + b3Vector3(-m_extents[0], -m_extents[1], -m_extents[2]);
+		if (p == 0)
+			LocalPt = m_localCenter + b3Vector3(m_extents[0], m_extents[1], m_extents[2]);
+		else if (p == 1)
+			LocalPt = m_localCenter + b3Vector3(m_extents[0], m_extents[1], -m_extents[2]);
+		else if (p == 2)
+			LocalPt = m_localCenter + b3Vector3(m_extents[0], -m_extents[1], m_extents[2]);
+		else if (p == 3)
+			LocalPt = m_localCenter + b3Vector3(m_extents[0], -m_extents[1], -m_extents[2]);
+		else if (p == 4)
+			LocalPt = m_localCenter + b3Vector3(-m_extents[0], m_extents[1], m_extents[2]);
+		else if (p == 5)
+			LocalPt = m_localCenter + b3Vector3(-m_extents[0], m_extents[1], -m_extents[2]);
+		else if (p == 6)
+			LocalPt = m_localCenter + b3Vector3(-m_extents[0], -m_extents[1], m_extents[2]);
+		else if (p == 7)
+			LocalPt = m_localCenter + b3Vector3(-m_extents[0], -m_extents[1], -m_extents[2]);
 
-		for(int i=0;i<m_faces.size();i++)
+		for (int i = 0; i < m_faces.size(); i++)
 		{
 			const b3Vector3 Normal(m_faces[i].m_plane[0], m_faces[i].m_plane[1], m_faces[i].m_plane[2]);
 			const b3Scalar d = LocalPt.dot(Normal) + m_faces[i].m_plane[3];
-			if(d>0.0f)
+			if (d > 0.0f)
 				return false;
 		}
 	}
@@ -336,39 +322,38 @@ bool b3ConvexUtility::testContainment() const
 }
 #endif
 
-void	b3ConvexUtility::initialize()
+void b3ConvexUtility::initialize()
 {
-
-	b3HashMap<b3InternalVertexPair,b3InternalEdge> edges;
+	b3HashMap<b3InternalVertexPair, b3InternalEdge> edges;
 
 	b3Scalar TotalArea = 0.0f;
-	
+
 	m_localCenter.setValue(0, 0, 0);
-	for(int i=0;i<m_faces.size();i++)
+	for (int i = 0; i < m_faces.size(); i++)
 	{
 		int numVertices = m_faces[i].m_indices.size();
 		int NbTris = numVertices;
-		for(int j=0;j<NbTris;j++)
+		for (int j = 0; j < NbTris; j++)
 		{
-			int k = (j+1)%numVertices;
-			b3InternalVertexPair vp(m_faces[i].m_indices[j],m_faces[i].m_indices[k]);
+			int k = (j + 1) % numVertices;
+			b3InternalVertexPair vp(m_faces[i].m_indices[j], m_faces[i].m_indices[k]);
 			b3InternalEdge* edptr = edges.find(vp);
-			b3Vector3 edge = m_vertices[vp.m_v1]-m_vertices[vp.m_v0];
+			b3Vector3 edge = m_vertices[vp.m_v1] - m_vertices[vp.m_v0];
 			edge.normalize();
 
 			bool found = false;
-			b3Vector3 diff,diff2;
+			b3Vector3 diff, diff2;
 
-			for (int p=0;p<m_uniqueEdges.size();p++)
+			for (int p = 0; p < m_uniqueEdges.size(); p++)
 			{
-				diff = m_uniqueEdges[p]-edge;
-				diff2 = m_uniqueEdges[p]+edge;
+				diff = m_uniqueEdges[p] - edge;
+				diff2 = m_uniqueEdges[p] + edge;
 
-			//	if ((diff.length2()==0.f) || 
+				//	if ((diff.length2()==0.f) ||
 				//	(diff2.length2()==0.f))
 
-				if (IsAlmostZero(diff) || 
-				IsAlmostZero(diff2))
+				if (IsAlmostZero(diff) ||
+					IsAlmostZero(diff2))
 				{
 					found = true;
 					break;
@@ -382,106 +367,101 @@ void	b3ConvexUtility::initialize()
 
 			if (edptr)
 			{
-					//TBD: figure out why I added this assert
-//				b3Assert(edptr->m_face0>=0);
-	//			b3Assert(edptr->m_face1<0);
+				//TBD: figure out why I added this assert
+				//				b3Assert(edptr->m_face0>=0);
+				//			b3Assert(edptr->m_face1<0);
 				edptr->m_face1 = i;
-			} else
+			}
+			else
 			{
 				b3InternalEdge ed;
 				ed.m_face0 = i;
-				edges.insert(vp,ed);
+				edges.insert(vp, ed);
 			}
 		}
 	}
 
 #ifdef USE_CONNECTED_FACES
-	for(int i=0;i<m_faces.size();i++)
+	for (int i = 0; i < m_faces.size(); i++)
 	{
 		int numVertices = m_faces[i].m_indices.size();
 		m_faces[i].m_connectedFaces.resize(numVertices);
 
-		for(int j=0;j<numVertices;j++)
+		for (int j = 0; j < numVertices; j++)
 		{
-			int k = (j+1)%numVertices;
-			b3InternalVertexPair vp(m_faces[i].m_indices[j],m_faces[i].m_indices[k]);
+			int k = (j + 1) % numVertices;
+			b3InternalVertexPair vp(m_faces[i].m_indices[j], m_faces[i].m_indices[k]);
 			b3InternalEdge* edptr = edges.find(vp);
 			b3Assert(edptr);
-			b3Assert(edptr->m_face0>=0);
-			b3Assert(edptr->m_face1>=0);
+			b3Assert(edptr->m_face0 >= 0);
+			b3Assert(edptr->m_face1 >= 0);
 
-			int connectedFace = (edptr->m_face0==i)?edptr->m_face1:edptr->m_face0;
+			int connectedFace = (edptr->m_face0 == i) ? edptr->m_face1 : edptr->m_face0;
 			m_faces[i].m_connectedFaces[j] = connectedFace;
 		}
 	}
-#endif//USE_CONNECTED_FACES
+#endif  //USE_CONNECTED_FACES
 
-	for(int i=0;i<m_faces.size();i++)
+	for (int i = 0; i < m_faces.size(); i++)
 	{
 		int numVertices = m_faces[i].m_indices.size();
-		int NbTris = numVertices-2;
-		
+		int NbTris = numVertices - 2;
+
 		const b3Vector3& p0 = m_vertices[m_faces[i].m_indices[0]];
-		for(int j=1;j<=NbTris;j++)
+		for (int j = 1; j <= NbTris; j++)
 		{
-			int k = (j+1)%numVertices;
+			int k = (j + 1) % numVertices;
 			const b3Vector3& p1 = m_vertices[m_faces[i].m_indices[j]];
 			const b3Vector3& p2 = m_vertices[m_faces[i].m_indices[k]];
 			b3Scalar Area = ((p0 - p1).cross(p0 - p2)).length() * 0.5f;
-			b3Vector3 Center = (p0+p1+p2)/3.0f;
+			b3Vector3 Center = (p0 + p1 + p2) / 3.0f;
 			m_localCenter += Area * Center;
 			TotalArea += Area;
 		}
 	}
 	m_localCenter /= TotalArea;
 
-
-
-
 #ifdef TEST_INTERNAL_OBJECTS
-	if(1)
+	if (1)
 	{
 		m_radius = FLT_MAX;
-		for(int i=0;i<m_faces.size();i++)
+		for (int i = 0; i < m_faces.size(); i++)
 		{
 			const b3Vector3 Normal(m_faces[i].m_plane[0], m_faces[i].m_plane[1], m_faces[i].m_plane[2]);
 			const b3Scalar dist = b3Fabs(m_localCenter.dot(Normal) + m_faces[i].m_plane[3]);
-			if(dist<m_radius)
+			if (dist < m_radius)
 				m_radius = dist;
 		}
 
-	
 		b3Scalar MinX = FLT_MAX;
 		b3Scalar MinY = FLT_MAX;
 		b3Scalar MinZ = FLT_MAX;
 		b3Scalar MaxX = -FLT_MAX;
 		b3Scalar MaxY = -FLT_MAX;
 		b3Scalar MaxZ = -FLT_MAX;
-		for(int i=0; i<m_vertices.size(); i++)
+		for (int i = 0; i < m_vertices.size(); i++)
 		{
 			const b3Vector3& pt = m_vertices[i];
-			if(pt.getX()<MinX)	MinX = pt.getX();
-			if(pt.getX()>MaxX)	MaxX = pt.getX();
-			if(pt.getY()<MinY)	MinY = pt.getY();
-			if(pt.getY()>MaxY)	MaxY = pt.getY();
-			if(pt.getZ()<MinZ)	MinZ = pt.getZ();
-			if(pt.getZ()>MaxZ)	MaxZ = pt.getZ();
+			if (pt.getX() < MinX) MinX = pt.getX();
+			if (pt.getX() > MaxX) MaxX = pt.getX();
+			if (pt.getY() < MinY) MinY = pt.getY();
+			if (pt.getY() > MaxY) MaxY = pt.getY();
+			if (pt.getZ() < MinZ) MinZ = pt.getZ();
+			if (pt.getZ() > MaxZ) MaxZ = pt.getZ();
 		}
-		mC.setValue(MaxX+MinX, MaxY+MinY, MaxZ+MinZ);
-		mE.setValue(MaxX-MinX, MaxY-MinY, MaxZ-MinZ);
+		mC.setValue(MaxX + MinX, MaxY + MinY, MaxZ + MinZ);
+		mE.setValue(MaxX - MinX, MaxY - MinY, MaxZ - MinZ);
 
-
-
-//		const b3Scalar r = m_radius / sqrtf(2.0f);
+		//		const b3Scalar r = m_radius / sqrtf(2.0f);
 		const b3Scalar r = m_radius / sqrtf(3.0f);
 		const int LargestExtent = mE.maxAxis();
-		const b3Scalar Step = (mE[LargestExtent]*0.5f - r)/1024.0f;
+		const b3Scalar Step = (mE[LargestExtent] * 0.5f - r) / 1024.0f;
 		m_extents[0] = m_extents[1] = m_extents[2] = r;
-		m_extents[LargestExtent] = mE[LargestExtent]*0.5f;
+		m_extents[LargestExtent] = mE[LargestExtent] * 0.5f;
 		bool FoundBox = false;
-		for(int j=0;j<1024;j++)
+		for (int j = 0; j < 1024; j++)
 		{
-			if(testContainment())
+			if (testContainment())
 			{
 				FoundBox = true;
 				break;
@@ -489,25 +469,25 @@ void	b3ConvexUtility::initialize()
 
 			m_extents[LargestExtent] -= Step;
 		}
-		if(!FoundBox)
+		if (!FoundBox)
 		{
 			m_extents[0] = m_extents[1] = m_extents[2] = r;
 		}
 		else
 		{
 			// Refine the box
-			const b3Scalar Step = (m_radius - r)/1024.0f;
-			const int e0 = (1<<LargestExtent) & 3;
-			const int e1 = (1<<e0) & 3;
+			const b3Scalar Step = (m_radius - r) / 1024.0f;
+			const int e0 = (1 << LargestExtent) & 3;
+			const int e1 = (1 << e0) & 3;
 
-			for(int j=0;j<1024;j++)
+			for (int j = 0; j < 1024; j++)
 			{
 				const b3Scalar Saved0 = m_extents[e0];
 				const b3Scalar Saved1 = m_extents[e1];
 				m_extents[e0] += Step;
 				m_extents[e1] += Step;
 
-				if(!testContainment())
+				if (!testContainment())
 				{
 					m_extents[e0] = Saved0;
 					m_extents[e1] = Saved1;
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3ConvexUtility.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3ConvexUtility.h
index 86c4151f8c7..4c8a88cbda9 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3ConvexUtility.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3ConvexUtility.h
@@ -20,43 +20,36 @@ subject to the following restrictions:
 #include "Bullet3Common/b3AlignedObjectArray.h"
 #include "Bullet3Common/b3Transform.h"
 
-
-
-
 struct b3MyFace
 {
-	b3AlignedObjectArray<int>	m_indices;
-	b3Scalar	m_plane[4];
+	b3AlignedObjectArray<int> m_indices;
+	b3Scalar m_plane[4];
 };
 
-B3_ATTRIBUTE_ALIGNED16(class) b3ConvexUtility
+B3_ATTRIBUTE_ALIGNED16(class)
+b3ConvexUtility
 {
-	public:
+public:
 	B3_DECLARE_ALIGNED_ALLOCATOR();
 
-	b3Vector3		m_localCenter;
-	b3Vector3		m_extents;
-	b3Vector3		mC;
-	b3Vector3		mE;
-	b3Scalar		m_radius;
-	
-	b3AlignedObjectArray<b3Vector3>	m_vertices;
-	b3AlignedObjectArray<b3MyFace>	m_faces;
+	b3Vector3 m_localCenter;
+	b3Vector3 m_extents;
+	b3Vector3 mC;
+	b3Vector3 mE;
+	b3Scalar m_radius;
+
+	b3AlignedObjectArray<b3Vector3> m_vertices;
+	b3AlignedObjectArray<b3MyFace> m_faces;
 	b3AlignedObjectArray<b3Vector3> m_uniqueEdges;
 
-		
 	b3ConvexUtility()
 	{
 	}
 	virtual ~b3ConvexUtility();
 
-	bool	initializePolyhedralFeatures(const b3Vector3* orgVertices, int numVertices, bool mergeCoplanarTriangles=true);
-		
-	void	initialize();
+	bool initializePolyhedralFeatures(const b3Vector3* orgVertices, int numVertices, bool mergeCoplanarTriangles = true);
+
+	void initialize();
 	bool testContainment() const;
-
-
-
 };
 #endif
-	
\ No newline at end of file
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3CpuNarrowPhase.cpp b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3CpuNarrowPhase.cpp
index c3134b2c659..e0b2161100a 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3CpuNarrowPhase.cpp
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3CpuNarrowPhase.cpp
@@ -5,15 +5,13 @@
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3ConvexPolyhedronData.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3ContactConvexConvexSAT.h"
 
-
 struct b3CpuNarrowPhaseInternalData
 {
 	b3AlignedObjectArray<b3Aabb> m_localShapeAABBCPU;
-	b3AlignedObjectArray<b3Collidable>	m_collidablesCPU;
+	b3AlignedObjectArray<b3Collidable> m_collidablesCPU;
 	b3AlignedObjectArray<b3ConvexUtility*> m_convexData;
 	b3Config m_config;
 
-
 	b3AlignedObjectArray<b3ConvexPolyhedronData> m_convexPolyhedra;
 	b3AlignedObjectArray<b3Vector3> m_uniqueEdges;
 	b3AlignedObjectArray<b3Vector3> m_convexVertices;
@@ -22,10 +20,9 @@ struct b3CpuNarrowPhaseInternalData
 
 	b3AlignedObjectArray<b3Contact4Data> m_contacts;
 
-	int	m_numAcceleratedShapes;
+	int m_numAcceleratedShapes;
 };
 
-
 const b3AlignedObjectArray<b3Contact4Data>& b3CpuNarrowPhase::getContacts() const
 {
 	return m_data->m_contacts;
@@ -41,7 +38,6 @@ const b3Collidable& b3CpuNarrowPhase::getCollidableCpu(int collidableIndex) cons
 	return m_data->m_collidablesCPU[collidableIndex];
 }
 
-
 b3CpuNarrowPhase::b3CpuNarrowPhase(const struct b3Config& config)
 {
 	m_data = new b3CpuNarrowPhaseInternalData;
@@ -61,7 +57,7 @@ void b3CpuNarrowPhase::computeContacts(b3AlignedObjectArray<b3Int4>& pairs, b3Al
 	int maxContactCapacity = m_data->m_config.m_maxContactCapacity;
 	m_data->m_contacts.resize(maxContactCapacity);
 
-	for (int i=0;i<nPairs;i++)
+	for (int i = 0; i < nPairs; i++)
 	{
 		int bodyIndexA = pairs[i].x;
 		int bodyIndexB = pairs[i].y;
@@ -71,64 +67,57 @@ void b3CpuNarrowPhase::computeContacts(b3AlignedObjectArray<b3Int4>& pairs, b3Al
 		if (m_data->m_collidablesCPU[collidableIndexA].m_shapeType == SHAPE_SPHERE &&
 			m_data->m_collidablesCPU[collidableIndexB].m_shapeType == SHAPE_CONVEX_HULL)
 		{
-//			computeContactSphereConvex(i,bodyIndexA,bodyIndexB,collidableIndexA,collidableIndexB,&bodies[0],
-//				&m_data->m_collidablesCPU[0],&hostConvexData[0],&hostVertices[0],&hostIndices[0],&hostFaces[0],&hostContacts[0],nContacts,maxContactCapacity);
+			//			computeContactSphereConvex(i,bodyIndexA,bodyIndexB,collidableIndexA,collidableIndexB,&bodies[0],
+			//				&m_data->m_collidablesCPU[0],&hostConvexData[0],&hostVertices[0],&hostIndices[0],&hostFaces[0],&hostContacts[0],nContacts,maxContactCapacity);
 		}
 
 		if (m_data->m_collidablesCPU[collidableIndexA].m_shapeType == SHAPE_CONVEX_HULL &&
 			m_data->m_collidablesCPU[collidableIndexB].m_shapeType == SHAPE_SPHERE)
 		{
-//			computeContactSphereConvex(i,bodyIndexB,bodyIndexA,collidableIndexB,collidableIndexA,&bodies[0],
-//				&m_data->m_collidablesCPU[0],&hostConvexData[0],&hostVertices[0],&hostIndices[0],&hostFaces[0],&hostContacts[0],nContacts,maxContactCapacity);
+			//			computeContactSphereConvex(i,bodyIndexB,bodyIndexA,collidableIndexB,collidableIndexA,&bodies[0],
+			//				&m_data->m_collidablesCPU[0],&hostConvexData[0],&hostVertices[0],&hostIndices[0],&hostFaces[0],&hostContacts[0],nContacts,maxContactCapacity);
 			//printf("convex-sphere\n");
-			
 		}
 
 		if (m_data->m_collidablesCPU[collidableIndexA].m_shapeType == SHAPE_CONVEX_HULL &&
 			m_data->m_collidablesCPU[collidableIndexB].m_shapeType == SHAPE_PLANE)
 		{
-//			computeContactPlaneConvex(i,bodyIndexB,bodyIndexA,collidableIndexB,collidableIndexA,&bodies[0],
-//			&m_data->m_collidablesCPU[0],&hostConvexData[0],&hostVertices[0],&hostIndices[0],&hostFaces[0],&hostContacts[0],nContacts,maxContactCapacity);
-//			printf("convex-plane\n");
-			
+			//			computeContactPlaneConvex(i,bodyIndexB,bodyIndexA,collidableIndexB,collidableIndexA,&bodies[0],
+			//			&m_data->m_collidablesCPU[0],&hostConvexData[0],&hostVertices[0],&hostIndices[0],&hostFaces[0],&hostContacts[0],nContacts,maxContactCapacity);
+			//			printf("convex-plane\n");
 		}
 
 		if (m_data->m_collidablesCPU[collidableIndexA].m_shapeType == SHAPE_PLANE &&
 			m_data->m_collidablesCPU[collidableIndexB].m_shapeType == SHAPE_CONVEX_HULL)
 		{
-//			computeContactPlaneConvex(i,bodyIndexA,bodyIndexB,collidableIndexA,collidableIndexB,&bodies[0],
-//			&m_data->m_collidablesCPU[0],&hostConvexData[0],&hostVertices[0],&hostIndices[0],&hostFaces[0],&hostContacts[0],nContacts,maxContactCapacity);
-//			printf("plane-convex\n");
-			
+			//			computeContactPlaneConvex(i,bodyIndexA,bodyIndexB,collidableIndexA,collidableIndexB,&bodies[0],
+			//			&m_data->m_collidablesCPU[0],&hostConvexData[0],&hostVertices[0],&hostIndices[0],&hostFaces[0],&hostContacts[0],nContacts,maxContactCapacity);
+			//			printf("plane-convex\n");
 		}
 
-			if (m_data->m_collidablesCPU[collidableIndexA].m_shapeType == SHAPE_COMPOUND_OF_CONVEX_HULLS &&
+		if (m_data->m_collidablesCPU[collidableIndexA].m_shapeType == SHAPE_COMPOUND_OF_CONVEX_HULLS &&
 			m_data->m_collidablesCPU[collidableIndexB].m_shapeType == SHAPE_COMPOUND_OF_CONVEX_HULLS)
 		{
-//			computeContactCompoundCompound(i,bodyIndexB,bodyIndexA,collidableIndexB,collidableIndexA,&bodies[0],
-//			&m_data->m_collidablesCPU[0],&hostConvexData[0],&cpuChildShapes[0], hostAabbsWorldSpace,hostAabbsLocalSpace,hostVertices,hostUniqueEdges,hostIndices,hostFaces,&hostContacts[0],
-//			nContacts,maxContactCapacity,treeNodesCPU,subTreesCPU,bvhInfoCPU);	
-//			printf("convex-plane\n");
-			
+			//			computeContactCompoundCompound(i,bodyIndexB,bodyIndexA,collidableIndexB,collidableIndexA,&bodies[0],
+			//			&m_data->m_collidablesCPU[0],&hostConvexData[0],&cpuChildShapes[0], hostAabbsWorldSpace,hostAabbsLocalSpace,hostVertices,hostUniqueEdges,hostIndices,hostFaces,&hostContacts[0],
+			//			nContacts,maxContactCapacity,treeNodesCPU,subTreesCPU,bvhInfoCPU);
+			//			printf("convex-plane\n");
 		}
 
-
-				if (m_data->m_collidablesCPU[collidableIndexA].m_shapeType == SHAPE_COMPOUND_OF_CONVEX_HULLS &&
+		if (m_data->m_collidablesCPU[collidableIndexA].m_shapeType == SHAPE_COMPOUND_OF_CONVEX_HULLS &&
 			m_data->m_collidablesCPU[collidableIndexB].m_shapeType == SHAPE_PLANE)
 		{
-//			computeContactPlaneCompound(i,bodyIndexB,bodyIndexA,collidableIndexB,collidableIndexA,&bodies[0],
-//			&m_data->m_collidablesCPU[0],&hostConvexData[0],&cpuChildShapes[0], &hostVertices[0],&hostIndices[0],&hostFaces[0],&hostContacts[0],nContacts,maxContactCapacity);
-//			printf("convex-plane\n");
-			
+			//			computeContactPlaneCompound(i,bodyIndexB,bodyIndexA,collidableIndexB,collidableIndexA,&bodies[0],
+			//			&m_data->m_collidablesCPU[0],&hostConvexData[0],&cpuChildShapes[0], &hostVertices[0],&hostIndices[0],&hostFaces[0],&hostContacts[0],nContacts,maxContactCapacity);
+			//			printf("convex-plane\n");
 		}
 
 		if (m_data->m_collidablesCPU[collidableIndexA].m_shapeType == SHAPE_PLANE &&
 			m_data->m_collidablesCPU[collidableIndexB].m_shapeType == SHAPE_COMPOUND_OF_CONVEX_HULLS)
 		{
-//			computeContactPlaneCompound(i,bodyIndexA,bodyIndexB,collidableIndexA,collidableIndexB,&bodies[0],
-//			&m_data->m_collidablesCPU[0],&hostConvexData[0],&cpuChildShapes[0],&hostVertices[0],&hostIndices[0],&hostFaces[0],&hostContacts[0],nContacts,maxContactCapacity);
-//			printf("plane-convex\n");
-			
+			//			computeContactPlaneCompound(i,bodyIndexA,bodyIndexB,collidableIndexA,collidableIndexB,&bodies[0],
+			//			&m_data->m_collidablesCPU[0],&hostConvexData[0],&cpuChildShapes[0],&hostVertices[0],&hostIndices[0],&hostFaces[0],&hostContacts[0],nContacts,maxContactCapacity);
+			//			printf("plane-convex\n");
 		}
 
 		if (m_data->m_collidablesCPU[collidableIndexA].m_shapeType == SHAPE_CONVEX_HULL &&
@@ -137,54 +126,48 @@ void b3CpuNarrowPhase::computeContacts(b3AlignedObjectArray<b3Int4>& pairs, b3Al
 			//printf("pairs[i].z=%d\n",pairs[i].z);
 			//int contactIndex = computeContactConvexConvex2(i,bodyIndexA,bodyIndexB,collidableIndexA,collidableIndexB,bodies,
 			//		m_data->m_collidablesCPU,hostConvexData,hostVertices,hostUniqueEdges,hostIndices,hostFaces,hostContacts,nContacts,maxContactCapacity,oldHostContacts);
-			int contactIndex = b3ContactConvexConvexSAT(i,bodyIndexA,bodyIndexB,collidableIndexA,collidableIndexB,bodies,
-				m_data->m_collidablesCPU,m_data->m_convexPolyhedra,m_data->m_convexVertices,m_data->m_uniqueEdges,m_data->m_convexIndices,m_data->m_convexFaces,m_data->m_contacts,numContacts,maxContactCapacity);
+			int contactIndex = b3ContactConvexConvexSAT(i, bodyIndexA, bodyIndexB, collidableIndexA, collidableIndexB, bodies,
+														m_data->m_collidablesCPU, m_data->m_convexPolyhedra, m_data->m_convexVertices, m_data->m_uniqueEdges, m_data->m_convexIndices, m_data->m_convexFaces, m_data->m_contacts, numContacts, maxContactCapacity);
 
-
-			if (contactIndex>=0)
+			if (contactIndex >= 0)
 			{
 				pairs[i].z = contactIndex;
 			}
-//			printf("plane-convex\n");
-			
+			//			printf("plane-convex\n");
 		}
-
-
 	}
 
 	m_data->m_contacts.resize(numContacts);
 }
 
-int	b3CpuNarrowPhase::registerConvexHullShape(b3ConvexUtility* utilPtr)
+int b3CpuNarrowPhase::registerConvexHullShape(b3ConvexUtility* utilPtr)
 {
 	int collidableIndex = allocateCollidable();
-	if (collidableIndex<0)
+	if (collidableIndex < 0)
 		return collidableIndex;
 
-	
 	b3Collidable& col = m_data->m_collidablesCPU[collidableIndex];
 	col.m_shapeType = SHAPE_CONVEX_HULL;
 	col.m_shapeIndex = -1;
-	
-	
+
 	{
-		b3Vector3 localCenter=b3MakeVector3(0,0,0);
-		for (int i=0;i<utilPtr->m_vertices.size();i++)
-			localCenter+=utilPtr->m_vertices[i];
-		localCenter*= (1.f/utilPtr->m_vertices.size());
+		b3Vector3 localCenter = b3MakeVector3(0, 0, 0);
+		for (int i = 0; i < utilPtr->m_vertices.size(); i++)
+			localCenter += utilPtr->m_vertices[i];
+		localCenter *= (1.f / utilPtr->m_vertices.size());
 		utilPtr->m_localCenter = localCenter;
 
-		col.m_shapeIndex = registerConvexHullShapeInternal(utilPtr,col);
+		col.m_shapeIndex = registerConvexHullShapeInternal(utilPtr, col);
 	}
 
-	if (col.m_shapeIndex>=0)
+	if (col.m_shapeIndex >= 0)
 	{
 		b3Aabb aabb;
-		
-		b3Vector3 myAabbMin=b3MakeVector3(1e30f,1e30f,1e30f);
-		b3Vector3 myAabbMax=b3MakeVector3(-1e30f,-1e30f,-1e30f);
 
-		for (int i=0;i<utilPtr->m_vertices.size();i++)
+		b3Vector3 myAabbMin = b3MakeVector3(1e30f, 1e30f, 1e30f);
+		b3Vector3 myAabbMax = b3MakeVector3(-1e30f, -1e30f, -1e30f);
+
+		for (int i = 0; i < utilPtr->m_vertices.size(); i++)
 		{
 			myAabbMin.setMin(utilPtr->m_vertices[i]);
 			myAabbMax.setMax(utilPtr->m_vertices[i]);
@@ -200,44 +183,42 @@ int	b3CpuNarrowPhase::registerConvexHullShape(b3ConvexUtility* utilPtr)
 		aabb.m_signedMaxIndices[3] = 0;
 
 		m_data->m_localShapeAABBCPU.push_back(aabb);
-
 	}
-	
+
 	return collidableIndex;
 }
 
-int	b3CpuNarrowPhase::allocateCollidable()
+int b3CpuNarrowPhase::allocateCollidable()
 {
 	int curSize = m_data->m_collidablesCPU.size();
-	if (curSize<m_data->m_config.m_maxConvexShapes)
+	if (curSize < m_data->m_config.m_maxConvexShapes)
 	{
 		m_data->m_collidablesCPU.expand();
 		return curSize;
 	}
 	else
 	{
-		b3Error("allocateCollidable out-of-range %d\n",m_data->m_config.m_maxConvexShapes);
+		b3Error("allocateCollidable out-of-range %d\n", m_data->m_config.m_maxConvexShapes);
 	}
 	return -1;
-
 }
 
-int	b3CpuNarrowPhase::registerConvexHullShape(const float* vertices, int strideInBytes, int numVertices, const float* scaling)
+int b3CpuNarrowPhase::registerConvexHullShape(const float* vertices, int strideInBytes, int numVertices, const float* scaling)
 {
 	b3AlignedObjectArray<b3Vector3> verts;
 
-	unsigned char* vts = (unsigned char*) vertices;
-	for (int i=0;i<numVertices;i++)
+	unsigned char* vts = (unsigned char*)vertices;
+	for (int i = 0; i < numVertices; i++)
 	{
-		float* vertex = (float*) &vts[i*strideInBytes];
-		verts.push_back(b3MakeVector3(vertex[0]*scaling[0],vertex[1]*scaling[1],vertex[2]*scaling[2]));
+		float* vertex = (float*)&vts[i * strideInBytes];
+		verts.push_back(b3MakeVector3(vertex[0] * scaling[0], vertex[1] * scaling[1], vertex[2] * scaling[2]));
 	}
 
 	b3ConvexUtility* utilPtr = new b3ConvexUtility();
 	bool merge = true;
 	if (numVertices)
 	{
-		utilPtr->initializePolyhedralFeatures(&verts[0],verts.size(),merge);
+		utilPtr->initializePolyhedralFeatures(&verts[0], verts.size(), merge);
 	}
 
 	int collidableIndex = registerConvexHullShape(utilPtr);
@@ -246,74 +227,67 @@ int	b3CpuNarrowPhase::registerConvexHullShape(const float* vertices, int strideI
 	return collidableIndex;
 }
 
-
-int b3CpuNarrowPhase::registerConvexHullShapeInternal(b3ConvexUtility* convexPtr,b3Collidable& col)
+int b3CpuNarrowPhase::registerConvexHullShapeInternal(b3ConvexUtility* convexPtr, b3Collidable& col)
 {
+	m_data->m_convexData.resize(m_data->m_numAcceleratedShapes + 1);
+	m_data->m_convexPolyhedra.resize(m_data->m_numAcceleratedShapes + 1);
 
-	m_data->m_convexData.resize(m_data->m_numAcceleratedShapes+1);
-	m_data->m_convexPolyhedra.resize(m_data->m_numAcceleratedShapes+1);
-	
-    
-	b3ConvexPolyhedronData& convex = m_data->m_convexPolyhedra.at(m_data->m_convexPolyhedra.size()-1);
+	b3ConvexPolyhedronData& convex = m_data->m_convexPolyhedra.at(m_data->m_convexPolyhedra.size() - 1);
 	convex.mC = convexPtr->mC;
 	convex.mE = convexPtr->mE;
-	convex.m_extents= convexPtr->m_extents;
+	convex.m_extents = convexPtr->m_extents;
 	convex.m_localCenter = convexPtr->m_localCenter;
 	convex.m_radius = convexPtr->m_radius;
-	
+
 	convex.m_numUniqueEdges = convexPtr->m_uniqueEdges.size();
 	int edgeOffset = m_data->m_uniqueEdges.size();
 	convex.m_uniqueEdgesOffset = edgeOffset;
-	
-	m_data->m_uniqueEdges.resize(edgeOffset+convex.m_numUniqueEdges);
-    
+
+	m_data->m_uniqueEdges.resize(edgeOffset + convex.m_numUniqueEdges);
+
 	//convex data here
 	int i;
-	for ( i=0;i<convexPtr->m_uniqueEdges.size();i++)
+	for (i = 0; i < convexPtr->m_uniqueEdges.size(); i++)
 	{
-		m_data->m_uniqueEdges[edgeOffset+i] = convexPtr->m_uniqueEdges[i];
+		m_data->m_uniqueEdges[edgeOffset + i] = convexPtr->m_uniqueEdges[i];
 	}
-    
+
 	int faceOffset = m_data->m_convexFaces.size();
 	convex.m_faceOffset = faceOffset;
 	convex.m_numFaces = convexPtr->m_faces.size();
 
-	m_data->m_convexFaces.resize(faceOffset+convex.m_numFaces);
-	
+	m_data->m_convexFaces.resize(faceOffset + convex.m_numFaces);
 
-	for (i=0;i<convexPtr->m_faces.size();i++)
+	for (i = 0; i < convexPtr->m_faces.size(); i++)
 	{
-		m_data->m_convexFaces[convex.m_faceOffset+i].m_plane = b3MakeVector3(convexPtr->m_faces[i].m_plane[0],
-																			convexPtr->m_faces[i].m_plane[1],
-																			convexPtr->m_faces[i].m_plane[2],
-																			convexPtr->m_faces[i].m_plane[3]);
+		m_data->m_convexFaces[convex.m_faceOffset + i].m_plane = b3MakeVector3(convexPtr->m_faces[i].m_plane[0],
+																			   convexPtr->m_faces[i].m_plane[1],
+																			   convexPtr->m_faces[i].m_plane[2],
+																			   convexPtr->m_faces[i].m_plane[3]);
 
-		
 		int indexOffset = m_data->m_convexIndices.size();
 		int numIndices = convexPtr->m_faces[i].m_indices.size();
-		m_data->m_convexFaces[convex.m_faceOffset+i].m_numIndices = numIndices;
-		m_data->m_convexFaces[convex.m_faceOffset+i].m_indexOffset = indexOffset;
-		m_data->m_convexIndices.resize(indexOffset+numIndices);
-		for (int p=0;p<numIndices;p++)
+		m_data->m_convexFaces[convex.m_faceOffset + i].m_numIndices = numIndices;
+		m_data->m_convexFaces[convex.m_faceOffset + i].m_indexOffset = indexOffset;
+		m_data->m_convexIndices.resize(indexOffset + numIndices);
+		for (int p = 0; p < numIndices; p++)
 		{
-			m_data->m_convexIndices[indexOffset+p] = convexPtr->m_faces[i].m_indices[p];
+			m_data->m_convexIndices[indexOffset + p] = convexPtr->m_faces[i].m_indices[p];
 		}
 	}
-    
+
 	convex.m_numVertices = convexPtr->m_vertices.size();
 	int vertexOffset = m_data->m_convexVertices.size();
-	convex.m_vertexOffset =vertexOffset;
-	
-	m_data->m_convexVertices.resize(vertexOffset+convex.m_numVertices);
-	for (int i=0;i<convexPtr->m_vertices.size();i++)
+	convex.m_vertexOffset = vertexOffset;
+
+	m_data->m_convexVertices.resize(vertexOffset + convex.m_numVertices);
+	for (int i = 0; i < convexPtr->m_vertices.size(); i++)
 	{
-		m_data->m_convexVertices[vertexOffset+i] = convexPtr->m_vertices[i];
+		m_data->m_convexVertices[vertexOffset + i] = convexPtr->m_vertices[i];
 	}
 
 	(m_data->m_convexData)[m_data->m_numAcceleratedShapes] = convexPtr;
-	
-    
-    
+
 	return m_data->m_numAcceleratedShapes++;
 }
 
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3CpuNarrowPhase.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3CpuNarrowPhase.h
index 528be3346de..f02353c2657 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3CpuNarrowPhase.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3CpuNarrowPhase.h
@@ -12,66 +12,55 @@
 class b3CpuNarrowPhase
 {
 protected:
-
-	struct b3CpuNarrowPhaseInternalData*	m_data;
+	struct b3CpuNarrowPhaseInternalData* m_data;
 	int m_acceleratedCompanionShapeIndex;
 	int m_planeBodyIndex;
-	int	m_static0Index;
+	int m_static0Index;
 
-	int registerConvexHullShapeInternal(class b3ConvexUtility* convexPtr,b3Collidable& col);
+	int registerConvexHullShapeInternal(class b3ConvexUtility* convexPtr, b3Collidable& col);
 	int registerConcaveMeshShape(b3AlignedObjectArray<b3Vector3>* vertices, b3AlignedObjectArray<int>* indices, b3Collidable& col, const float* scaling);
 
 public:
-
-	
-
-
 	b3CpuNarrowPhase(const struct b3Config& config);
 
 	virtual ~b3CpuNarrowPhase(void);
 
-	int		registerSphereShape(float radius);
-	int		registerPlaneShape(const b3Vector3& planeNormal, float planeConstant);
+	int registerSphereShape(float radius);
+	int registerPlaneShape(const b3Vector3& planeNormal, float planeConstant);
 
 	int registerCompoundShape(b3AlignedObjectArray<b3GpuChildShape>* childShapes);
 	int registerFace(const b3Vector3& faceNormal, float faceConstant);
-	
-	int	registerConcaveMesh(b3AlignedObjectArray<b3Vector3>* vertices, b3AlignedObjectArray<int>* indices,const float* scaling);
-	
+
+	int registerConcaveMesh(b3AlignedObjectArray<b3Vector3>* vertices, b3AlignedObjectArray<int>* indices, const float* scaling);
+
 	//do they need to be merged?
-	
-	int	registerConvexHullShape(b3ConvexUtility* utilPtr);
-	int	registerConvexHullShape(const float* vertices, int strideInBytes, int numVertices, const float* scaling);
+
+	int registerConvexHullShape(b3ConvexUtility* utilPtr);
+	int registerConvexHullShape(const float* vertices, int strideInBytes, int numVertices, const float* scaling);
 
 	//int registerRigidBody(int collidableIndex, float mass, const float* position, const float* orientation, const float* aabbMin, const float* aabbMax,bool writeToGpu);
-	void setObjectTransform(const float* position, const float* orientation , int bodyIndex);
+	void setObjectTransform(const float* position, const float* orientation, int bodyIndex);
 
-	void	writeAllBodiesToGpu();
-	void  reset();
-	void	readbackAllBodiesToCpu();
-	bool	getObjectTransformFromCpu(float* position, float* orientation , int bodyIndex) const;
+	void writeAllBodiesToGpu();
+	void reset();
+	void readbackAllBodiesToCpu();
+	bool getObjectTransformFromCpu(float* position, float* orientation, int bodyIndex) const;
 
-	void setObjectTransformCpu(float* position, float* orientation , int bodyIndex);
+	void setObjectTransformCpu(float* position, float* orientation, int bodyIndex);
 	void setObjectVelocityCpu(float* linVel, float* angVel, int bodyIndex);
 
-	
 	//virtual void computeContacts(cl_mem broadphasePairs, int numBroadphasePairs, cl_mem aabbsWorldSpace, int numObjects);
 	virtual void computeContacts(b3AlignedObjectArray<b3Int4>& pairs, b3AlignedObjectArray<b3Aabb>& aabbsWorldSpace, b3AlignedObjectArray<b3RigidBodyData>& bodies);
 
-
-	
 	const struct b3RigidBodyData* getBodiesCpu() const;
 	//struct b3RigidBodyData* getBodiesCpu();
 
-	int	getNumBodiesGpu() const;
+	int getNumBodiesGpu() const;
 
-	
-	int	getNumBodyInertiasGpu() const;
+	int getNumBodyInertiasGpu() const;
 
-	
 	const struct b3Collidable* getCollidablesCpu() const;
-	int		getNumCollidablesGpu() const;
-
+	int getNumCollidablesGpu() const;
 
 	/*const struct b3Contact4* getContactsCPU() const;
 
@@ -80,8 +69,7 @@ public:
 	*/
 
 	const b3AlignedObjectArray<b3Contact4Data>& getContacts() const;
-	
-	
+
 	int getNumRigidBodies() const;
 
 	int allocateCollidable();
@@ -93,13 +81,12 @@ public:
 	b3Collidable& getCollidableCpu(int collidableIndex);
 	const b3Collidable& getCollidableCpu(int collidableIndex) const;
 
-	const b3CpuNarrowPhaseInternalData*	getInternalData() const
+	const b3CpuNarrowPhaseInternalData* getInternalData() const
 	{
-			return m_data;
+		return m_data;
 	}
 
 	const struct b3Aabb& getLocalSpaceAabb(int collidableIndex) const;
 };
 
-#endif //B3_CPU_NARROWPHASE_H
-
+#endif  //B3_CPU_NARROWPHASE_H
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3RaycastInfo.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3RaycastInfo.h
index fba8bd07a4f..b50c0eca4f9 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3RaycastInfo.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3RaycastInfo.h
@@ -4,21 +4,22 @@
 
 #include "Bullet3Common/b3Vector3.h"
 
-B3_ATTRIBUTE_ALIGNED16(struct) b3RayInfo
+B3_ATTRIBUTE_ALIGNED16(struct)
+b3RayInfo
 {
 	b3Vector3 m_from;
 	b3Vector3 m_to;
 };
 
-B3_ATTRIBUTE_ALIGNED16(struct) b3RayHit
+B3_ATTRIBUTE_ALIGNED16(struct)
+b3RayHit
 {
-		b3Scalar	m_hitFraction;
-		int	m_hitBody;
-		int	m_hitResult1;
-		int	m_hitResult2;
-		b3Vector3 m_hitPoint;
-		b3Vector3 m_hitNormal;
+	b3Scalar m_hitFraction;
+	int m_hitBody;
+	int m_hitResult1;
+	int m_hitResult2;
+	b3Vector3 m_hitPoint;
+	b3Vector3 m_hitNormal;
 };
 
-#endif //B3_RAYCAST_INFO_H
-
+#endif  //B3_RAYCAST_INFO_H
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3RigidBodyCL.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3RigidBodyCL.h
index d58f71802f7..be1be57f055 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3RigidBodyCL.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/b3RigidBodyCL.h
@@ -20,11 +20,9 @@ subject to the following restrictions:
 #include "Bullet3Common/b3Matrix3x3.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3RigidBodyData.h"
 
-
-inline float	b3GetInvMass(const b3RigidBodyData& body)
+inline float b3GetInvMass(const b3RigidBodyData& body)
 {
-		return body.m_invMass;
+	return body.m_invMass;
 }
 
-
-#endif//B3_RIGID_BODY_CL
+#endif  //B3_RIGID_BODY_CL
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3BvhSubtreeInfoData.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3BvhSubtreeInfoData.h
index 8788ccbb471..d6beb662b55 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3BvhSubtreeInfoData.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3BvhSubtreeInfoData.h
@@ -7,14 +7,13 @@ typedef struct b3BvhSubtreeInfoData b3BvhSubtreeInfoData_t;
 struct b3BvhSubtreeInfoData
 {
 	//12 bytes
-	unsigned short int	m_quantizedAabbMin[3];
-	unsigned short int	m_quantizedAabbMax[3];
+	unsigned short int m_quantizedAabbMin[3];
+	unsigned short int m_quantizedAabbMax[3];
 	//4 bytes, points to the root of the subtree
-	int			m_rootNodeIndex;
+	int m_rootNodeIndex;
 	//4 bytes
-	int			m_subtreeSize;
-	int			m_padding[3];
+	int m_subtreeSize;
+	int m_padding[3];
 };
 
-#endif //B3_BVH_SUBTREE_INFO_DATA_H
-
+#endif  //B3_BVH_SUBTREE_INFO_DATA_H
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3BvhTraversal.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3BvhTraversal.h
index 2618da24bca..7c2507cc98b 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3BvhTraversal.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3BvhTraversal.h
@@ -7,69 +7,64 @@
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3BvhSubtreeInfoData.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3QuantizedBvhNodeData.h"
 
-
-
 // work-in-progress
-void   b3BvhTraversal( __global const b3Int4* pairs, 
-									__global const b3RigidBodyData* rigidBodies, 
-									__global const b3Collidable* collidables,
-									__global b3Aabb* aabbs,
-									__global b3Int4* concavePairsOut,
-									__global volatile int* numConcavePairsOut,
-									__global const b3BvhSubtreeInfo* subtreeHeadersRoot,
-									__global const b3QuantizedBvhNode* quantizedNodesRoot,
-									__global const b3BvhInfo* bvhInfos,
-									int numPairs,
-									int maxNumConcavePairsCapacity,
-									int id)
+void b3BvhTraversal(__global const b3Int4* pairs,
+					__global const b3RigidBodyData* rigidBodies,
+					__global const b3Collidable* collidables,
+					__global b3Aabb* aabbs,
+					__global b3Int4* concavePairsOut,
+					__global volatile int* numConcavePairsOut,
+					__global const b3BvhSubtreeInfo* subtreeHeadersRoot,
+					__global const b3QuantizedBvhNode* quantizedNodesRoot,
+					__global const b3BvhInfo* bvhInfos,
+					int numPairs,
+					int maxNumConcavePairsCapacity,
+					int id)
 {
-	
 	int bodyIndexA = pairs[id].x;
 	int bodyIndexB = pairs[id].y;
 	int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;
 	int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;
-	
+
 	//once the broadphase avoids static-static pairs, we can remove this test
-	if ((rigidBodies[bodyIndexA].m_invMass==0) &&(rigidBodies[bodyIndexB].m_invMass==0))
+	if ((rigidBodies[bodyIndexA].m_invMass == 0) && (rigidBodies[bodyIndexB].m_invMass == 0))
 	{
 		return;
 	}
-		
-	if (collidables[collidableIndexA].m_shapeType!=SHAPE_CONCAVE_TRIMESH)
+
+	if (collidables[collidableIndexA].m_shapeType != SHAPE_CONCAVE_TRIMESH)
 		return;
 
 	int shapeTypeB = collidables[collidableIndexB].m_shapeType;
-		
-	if (shapeTypeB!=SHAPE_CONVEX_HULL &&
-		shapeTypeB!=SHAPE_SPHERE	&&
-		shapeTypeB!=SHAPE_COMPOUND_OF_CONVEX_HULLS
-		)
+
+	if (shapeTypeB != SHAPE_CONVEX_HULL &&
+		shapeTypeB != SHAPE_SPHERE &&
+		shapeTypeB != SHAPE_COMPOUND_OF_CONVEX_HULLS)
 		return;
 
 	b3BvhInfo bvhInfo = bvhInfos[collidables[collidableIndexA].m_numChildShapes];
 
-	b3Float4	bvhAabbMin = bvhInfo.m_aabbMin;
-	b3Float4	bvhAabbMax = bvhInfo.m_aabbMax;
-	b3Float4	bvhQuantization = bvhInfo.m_quantization;
+	b3Float4 bvhAabbMin = bvhInfo.m_aabbMin;
+	b3Float4 bvhAabbMax = bvhInfo.m_aabbMax;
+	b3Float4 bvhQuantization = bvhInfo.m_quantization;
 	int numSubtreeHeaders = bvhInfo.m_numSubTrees;
 	__global const b3BvhSubtreeInfoData* subtreeHeaders = &subtreeHeadersRoot[bvhInfo.m_subTreeOffset];
 	__global const b3QuantizedBvhNodeData* quantizedNodes = &quantizedNodesRoot[bvhInfo.m_nodeOffset];
-	
 
 	unsigned short int quantizedQueryAabbMin[3];
 	unsigned short int quantizedQueryAabbMax[3];
-	b3QuantizeWithClamp(quantizedQueryAabbMin,aabbs[bodyIndexB].m_minVec,false,bvhAabbMin, bvhAabbMax,bvhQuantization);
-	b3QuantizeWithClamp(quantizedQueryAabbMax,aabbs[bodyIndexB].m_maxVec,true ,bvhAabbMin, bvhAabbMax,bvhQuantization);
-	
-	for (int i=0;i<numSubtreeHeaders;i++)
+	b3QuantizeWithClamp(quantizedQueryAabbMin, aabbs[bodyIndexB].m_minVec, false, bvhAabbMin, bvhAabbMax, bvhQuantization);
+	b3QuantizeWithClamp(quantizedQueryAabbMax, aabbs[bodyIndexB].m_maxVec, true, bvhAabbMin, bvhAabbMax, bvhQuantization);
+
+	for (int i = 0; i < numSubtreeHeaders; i++)
 	{
 		b3BvhSubtreeInfoData subtree = subtreeHeaders[i];
-				
-		int overlap = b3TestQuantizedAabbAgainstQuantizedAabbSlow(quantizedQueryAabbMin,quantizedQueryAabbMax,subtree.m_quantizedAabbMin,subtree.m_quantizedAabbMax);
+
+		int overlap = b3TestQuantizedAabbAgainstQuantizedAabbSlow(quantizedQueryAabbMin, quantizedQueryAabbMax, subtree.m_quantizedAabbMin, subtree.m_quantizedAabbMax);
 		if (overlap != 0)
 		{
 			int startNodeIndex = subtree.m_rootNodeIndex;
-			int endNodeIndex = subtree.m_rootNodeIndex+subtree.m_subtreeSize;
+			int endNodeIndex = subtree.m_rootNodeIndex + subtree.m_subtreeSize;
 			int curIndex = startNodeIndex;
 			int escapeIndex;
 			int isLeafNode;
@@ -77,43 +72,46 @@ void   b3BvhTraversal( __global const b3Int4* pairs,
 			while (curIndex < endNodeIndex)
 			{
 				b3QuantizedBvhNodeData rootNode = quantizedNodes[curIndex];
-				aabbOverlap = b3TestQuantizedAabbAgainstQuantizedAabbSlow(quantizedQueryAabbMin,quantizedQueryAabbMax,rootNode.m_quantizedAabbMin,rootNode.m_quantizedAabbMax);
+				aabbOverlap = b3TestQuantizedAabbAgainstQuantizedAabbSlow(quantizedQueryAabbMin, quantizedQueryAabbMax, rootNode.m_quantizedAabbMin, rootNode.m_quantizedAabbMax);
 				isLeafNode = b3IsLeaf(&rootNode);
 				if (aabbOverlap)
 				{
 					if (isLeafNode)
 					{
 						int triangleIndex = b3GetTriangleIndex(&rootNode);
-						if (shapeTypeB==SHAPE_COMPOUND_OF_CONVEX_HULLS)
+						if (shapeTypeB == SHAPE_COMPOUND_OF_CONVEX_HULLS)
 						{
-								int numChildrenB = collidables[collidableIndexB].m_numChildShapes;
-								int pairIdx = b3AtomicAdd (numConcavePairsOut,numChildrenB);
-								for (int b=0;b<numChildrenB;b++)
+							int numChildrenB = collidables[collidableIndexB].m_numChildShapes;
+							int pairIdx = b3AtomicAdd(numConcavePairsOut, numChildrenB);
+							for (int b = 0; b < numChildrenB; b++)
+							{
+								if ((pairIdx + b) < maxNumConcavePairsCapacity)
 								{
-									if ((pairIdx+b)<maxNumConcavePairsCapacity)
-									{
-										int childShapeIndexB = collidables[collidableIndexB].m_shapeIndex+b;
-										b3Int4 newPair = b3MakeInt4(bodyIndexA,bodyIndexB,triangleIndex,childShapeIndexB);
-										concavePairsOut[pairIdx+b] = newPair;
-									}
+									int childShapeIndexB = collidables[collidableIndexB].m_shapeIndex + b;
+									b3Int4 newPair = b3MakeInt4(bodyIndexA, bodyIndexB, triangleIndex, childShapeIndexB);
+									concavePairsOut[pairIdx + b] = newPair;
 								}
-						} else
+							}
+						}
+						else
 						{
 							int pairIdx = b3AtomicInc(numConcavePairsOut);
-							if (pairIdx<maxNumConcavePairsCapacity)
+							if (pairIdx < maxNumConcavePairsCapacity)
 							{
-								b3Int4 newPair = b3MakeInt4(bodyIndexA,bodyIndexB,triangleIndex,0);
+								b3Int4 newPair = b3MakeInt4(bodyIndexA, bodyIndexB, triangleIndex, 0);
 								concavePairsOut[pairIdx] = newPair;
 							}
 						}
-					} 
+					}
 					curIndex++;
-				} else
+				}
+				else
 				{
 					if (isLeafNode)
 					{
 						curIndex++;
-					} else
+					}
+					else
 					{
 						escapeIndex = b3GetEscapeIndex(&rootNode);
 						curIndex += escapeIndex;
@@ -122,5 +120,4 @@ void   b3BvhTraversal( __global const b3Int4* pairs,
 			}
 		}
 	}
-
 }
\ No newline at end of file
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3ClipFaces.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3ClipFaces.h
index 8009e7d6e0b..0d9b13f1d69 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3ClipFaces.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3ClipFaces.h
@@ -1,7 +1,6 @@
 #ifndef B3_CLIP_FACES_H
 #define B3_CLIP_FACES_H
 
-
 #include "Bullet3Common/shared/b3Int4.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3RigidBodyData.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3Collidable.h"
@@ -10,38 +9,36 @@
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3QuantizedBvhNodeData.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3ConvexPolyhedronData.h"
 
-
-inline b3Float4 b3Lerp3(b3Float4ConstArg a,b3Float4ConstArg b, float  t)
+inline b3Float4 b3Lerp3(b3Float4ConstArg a, b3Float4ConstArg b, float t)
 {
-	return b3MakeFloat4(	a.x + (b.x - a.x) * t,
+	return b3MakeFloat4(a.x + (b.x - a.x) * t,
 						a.y + (b.y - a.y) * t,
 						a.z + (b.z - a.z) * t,
 						0.f);
 }
 
 // Clips a face to the back of a plane, return the number of vertices out, stored in ppVtxOut
-int clipFaceGlobal(__global const b3Float4* pVtxIn, int numVertsIn, b3Float4ConstArg planeNormalWS,float planeEqWS, __global b3Float4* ppVtxOut)
+int clipFaceGlobal(__global const b3Float4* pVtxIn, int numVertsIn, b3Float4ConstArg planeNormalWS, float planeEqWS, __global b3Float4* ppVtxOut)
 {
-	
 	int ve;
 	float ds, de;
 	int numVertsOut = 0;
-    //double-check next test
-    //	if (numVertsIn < 2)
-    //		return 0;
-    
-	b3Float4 firstVertex=pVtxIn[numVertsIn-1];
+	//double-check next test
+	//	if (numVertsIn < 2)
+	//		return 0;
+
+	b3Float4 firstVertex = pVtxIn[numVertsIn - 1];
 	b3Float4 endVertex = pVtxIn[0];
-	
-	ds = b3Dot(planeNormalWS,firstVertex)+planeEqWS;
-    
+
+	ds = b3Dot(planeNormalWS, firstVertex) + planeEqWS;
+
 	for (ve = 0; ve < numVertsIn; ve++)
 	{
-		endVertex=pVtxIn[ve];
-		de = b3Dot(planeNormalWS,endVertex)+planeEqWS;
-		if (ds<0)
+		endVertex = pVtxIn[ve];
+		de = b3Dot(planeNormalWS, endVertex) + planeEqWS;
+		if (ds < 0)
 		{
-			if (de<0)
+			if (de < 0)
 			{
 				// Start < 0, end < 0, so output endVertex
 				ppVtxOut[numVertsOut++] = endVertex;
@@ -49,15 +46,15 @@ int clipFaceGlobal(__global const b3Float4* pVtxIn, int numVertsIn, b3Float4Cons
 			else
 			{
 				// Start < 0, end >= 0, so output intersection
-				ppVtxOut[numVertsOut++] = b3Lerp3(firstVertex, endVertex,(ds * 1.f/(ds - de)) );
+				ppVtxOut[numVertsOut++] = b3Lerp3(firstVertex, endVertex, (ds * 1.f / (ds - de)));
 			}
 		}
 		else
 		{
-			if (de<0)
+			if (de < 0)
 			{
 				// Start >= 0, end < 0 so output intersection and end
-				ppVtxOut[numVertsOut++] = b3Lerp3(firstVertex, endVertex,(ds * 1.f/(ds - de)) );
+				ppVtxOut[numVertsOut++] = b3Lerp3(firstVertex, endVertex, (ds * 1.f / (ds - de)));
 				ppVtxOut[numVertsOut++] = endVertex;
 			}
 		}
@@ -67,90 +64,81 @@ int clipFaceGlobal(__global const b3Float4* pVtxIn, int numVertsIn, b3Float4Cons
 	return numVertsOut;
 }
 
-
-__kernel void   clipFacesAndFindContactsKernel(    __global const b3Float4* separatingNormals,
-                                                   __global const int* hasSeparatingAxis,
-                                                   __global b3Int4* clippingFacesOut,
-                                                   __global b3Float4* worldVertsA1,
-                                                   __global b3Float4* worldNormalsA1,
-                                                   __global b3Float4* worldVertsB1,
-                                                   __global b3Float4* worldVertsB2,
-                                                    int vertexFaceCapacity,
-															int pairIndex
-                                                   )
+__kernel void clipFacesAndFindContactsKernel(__global const b3Float4* separatingNormals,
+											 __global const int* hasSeparatingAxis,
+											 __global b3Int4* clippingFacesOut,
+											 __global b3Float4* worldVertsA1,
+											 __global b3Float4* worldNormalsA1,
+											 __global b3Float4* worldVertsB1,
+											 __global b3Float4* worldVertsB2,
+											 int vertexFaceCapacity,
+											 int pairIndex)
 {
-//    int i = get_global_id(0);
+	//    int i = get_global_id(0);
 	//int pairIndex = i;
 	int i = pairIndex;
-    
+
 	float minDist = -1e30f;
 	float maxDist = 0.02f;
-    
-//	if (i<numPairs)
+
+	//	if (i<numPairs)
 	{
-        
 		if (hasSeparatingAxis[i])
 		{
-            
-//			int bodyIndexA = pairs[i].x;
-	//		int bodyIndexB = pairs[i].y;
-		    
-            int numLocalContactsOut = 0;
+			//			int bodyIndexA = pairs[i].x;
+			//		int bodyIndexB = pairs[i].y;
 
-            int capacityWorldVertsB2 = vertexFaceCapacity;
-            
-            __global b3Float4* pVtxIn = &worldVertsB1[pairIndex*capacityWorldVertsB2];
-            __global b3Float4* pVtxOut = &worldVertsB2[pairIndex*capacityWorldVertsB2];
-            
+			int numLocalContactsOut = 0;
 
-            {
-                __global b3Int4* clippingFaces = clippingFacesOut;
-            
-                
-                int closestFaceA = clippingFaces[pairIndex].x;
-               // int closestFaceB = clippingFaces[pairIndex].y;
-                int numVertsInA = clippingFaces[pairIndex].z;
-                int numVertsInB = clippingFaces[pairIndex].w;
-                
-                int numVertsOut = 0;
-                
-                if (closestFaceA>=0)
-                {
-                    
-                    
-                    
-                    // clip polygon to back of planes of all faces of hull A that are adjacent to witness face
-                    
-                    for(int e0=0;e0<numVertsInA;e0++)
-                    {
-                        const b3Float4 aw = worldVertsA1[pairIndex*capacityWorldVertsB2+e0];
-                        const b3Float4 bw = worldVertsA1[pairIndex*capacityWorldVertsB2+((e0+1)%numVertsInA)];
-                        const b3Float4 WorldEdge0 = aw - bw;
-                        b3Float4 worldPlaneAnormal1 = worldNormalsA1[pairIndex];
-                        b3Float4 planeNormalWS1 = -b3Cross(WorldEdge0,worldPlaneAnormal1);
-                        b3Float4 worldA1 = aw;
-                        float planeEqWS1 = -b3Dot(worldA1,planeNormalWS1);
-                        b3Float4 planeNormalWS = planeNormalWS1;
-                        float planeEqWS=planeEqWS1;
-                        numVertsOut = clipFaceGlobal(pVtxIn, numVertsInB, planeNormalWS,planeEqWS, pVtxOut);
-                        __global b3Float4* tmp = pVtxOut;
-                        pVtxOut = pVtxIn;
-                        pVtxIn = tmp;
-                        numVertsInB = numVertsOut;
-                        numVertsOut = 0;
-                    }
-                    
-                    b3Float4 planeNormalWS = worldNormalsA1[pairIndex];
-                    float planeEqWS=-b3Dot(planeNormalWS,worldVertsA1[pairIndex*capacityWorldVertsB2]);
-                    
-                    for (int i=0;i<numVertsInB;i++)
-                    {
-                        float depth = b3Dot(planeNormalWS,pVtxIn[i])+planeEqWS;
-                        if (depth <=minDist)
-                        {
-                            depth = minDist;
-                        }
-/*
+			int capacityWorldVertsB2 = vertexFaceCapacity;
+
+			__global b3Float4* pVtxIn = &worldVertsB1[pairIndex * capacityWorldVertsB2];
+			__global b3Float4* pVtxOut = &worldVertsB2[pairIndex * capacityWorldVertsB2];
+
+			{
+				__global b3Int4* clippingFaces = clippingFacesOut;
+
+				int closestFaceA = clippingFaces[pairIndex].x;
+				// int closestFaceB = clippingFaces[pairIndex].y;
+				int numVertsInA = clippingFaces[pairIndex].z;
+				int numVertsInB = clippingFaces[pairIndex].w;
+
+				int numVertsOut = 0;
+
+				if (closestFaceA >= 0)
+				{
+					// clip polygon to back of planes of all faces of hull A that are adjacent to witness face
+
+					for (int e0 = 0; e0 < numVertsInA; e0++)
+					{
+						const b3Float4 aw = worldVertsA1[pairIndex * capacityWorldVertsB2 + e0];
+						const b3Float4 bw = worldVertsA1[pairIndex * capacityWorldVertsB2 + ((e0 + 1) % numVertsInA)];
+						const b3Float4 WorldEdge0 = aw - bw;
+						b3Float4 worldPlaneAnormal1 = worldNormalsA1[pairIndex];
+						b3Float4 planeNormalWS1 = -b3Cross(WorldEdge0, worldPlaneAnormal1);
+						b3Float4 worldA1 = aw;
+						float planeEqWS1 = -b3Dot(worldA1, planeNormalWS1);
+						b3Float4 planeNormalWS = planeNormalWS1;
+						float planeEqWS = planeEqWS1;
+						numVertsOut = clipFaceGlobal(pVtxIn, numVertsInB, planeNormalWS, planeEqWS, pVtxOut);
+						__global b3Float4* tmp = pVtxOut;
+						pVtxOut = pVtxIn;
+						pVtxIn = tmp;
+						numVertsInB = numVertsOut;
+						numVertsOut = 0;
+					}
+
+					b3Float4 planeNormalWS = worldNormalsA1[pairIndex];
+					float planeEqWS = -b3Dot(planeNormalWS, worldVertsA1[pairIndex * capacityWorldVertsB2]);
+
+					for (int i = 0; i < numVertsInB; i++)
+					{
+						float depth = b3Dot(planeNormalWS, pVtxIn[i]) + planeEqWS;
+						if (depth <= minDist)
+						{
+							depth = minDist;
+						}
+						/*
 						static float maxDepth = 0.f;
 						if (depth < maxDepth)
 						{
@@ -163,26 +151,21 @@ __kernel void   clipFacesAndFindContactsKernel(    __global const b3Float4* sepa
 
 						}
 */
-                        if (depth <=maxDist)
-                        {
-                            b3Float4 pointInWorld = pVtxIn[i];
-                            pVtxOut[numLocalContactsOut++] = b3MakeFloat4(pointInWorld.x,pointInWorld.y,pointInWorld.z,depth);
-                        }
-                    }
-                    
-                }
-                clippingFaces[pairIndex].w =numLocalContactsOut;
-                
+						if (depth <= maxDist)
+						{
+							b3Float4 pointInWorld = pVtxIn[i];
+							pVtxOut[numLocalContactsOut++] = b3MakeFloat4(pointInWorld.x, pointInWorld.y, pointInWorld.z, depth);
+						}
+					}
+				}
+				clippingFaces[pairIndex].w = numLocalContactsOut;
+			}
 
-            }
-            
-            for (int i=0;i<numLocalContactsOut;i++)
-                pVtxIn[i] = pVtxOut[i];
-                
-		}//		if (hasSeparatingAxis[i])
-	}//	if (i<numPairs)
-    
+			for (int i = 0; i < numLocalContactsOut; i++)
+				pVtxIn[i] = pVtxOut[i];
+
+		}  //		if (hasSeparatingAxis[i])
+	}      //	if (i<numPairs)
 }
 
-#endif //B3_CLIP_FACES_H
-
+#endif  //B3_CLIP_FACES_H
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3Collidable.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3Collidable.h
index 77cdc7b7a9d..9a8c668af2c 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3Collidable.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3Collidable.h
@@ -2,40 +2,36 @@
 #ifndef B3_COLLIDABLE_H
 #define B3_COLLIDABLE_H
 
-
 #include "Bullet3Common/shared/b3Float4.h"
 #include "Bullet3Common/shared/b3Quat.h"
 
 enum b3ShapeTypes
 {
-	SHAPE_HEIGHT_FIELD=1,
+	SHAPE_HEIGHT_FIELD = 1,
 
-	SHAPE_CONVEX_HULL=3,
-	SHAPE_PLANE=4,
-	SHAPE_CONCAVE_TRIMESH=5,
-	SHAPE_COMPOUND_OF_CONVEX_HULLS=6,
-	SHAPE_SPHERE=7,
+	SHAPE_CONVEX_HULL = 3,
+	SHAPE_PLANE = 4,
+	SHAPE_CONCAVE_TRIMESH = 5,
+	SHAPE_COMPOUND_OF_CONVEX_HULLS = 6,
+	SHAPE_SPHERE = 7,
 	MAX_NUM_SHAPE_TYPES,
 };
 
 typedef struct b3Collidable b3Collidable_t;
 
-
 struct b3Collidable
 {
 	union {
 		int m_numChildShapes;
 		int m_bvhIndex;
 	};
-	union
-	{
+	union {
 		float m_radius;
-		int	m_compoundBvhIndex;
+		int m_compoundBvhIndex;
 	};
 
 	int m_shapeType;
-	union
-	{
+	union {
 		int m_shapeIndex;
 		float m_height;
 	};
@@ -44,33 +40,30 @@ struct b3Collidable
 typedef struct b3GpuChildShape b3GpuChildShape_t;
 struct b3GpuChildShape
 {
-	b3Float4	m_childPosition;
-	b3Quat		m_childOrientation;
-	union
-	{
-		int			m_shapeIndex;//used for SHAPE_COMPOUND_OF_CONVEX_HULLS
-		int			m_capsuleAxis;
+	b3Float4 m_childPosition;
+	b3Quat m_childOrientation;
+	union {
+		int m_shapeIndex;  //used for SHAPE_COMPOUND_OF_CONVEX_HULLS
+		int m_capsuleAxis;
 	};
-	union 
-	{
-		float		m_radius;//used for childshape of SHAPE_COMPOUND_OF_SPHERES or SHAPE_COMPOUND_OF_CAPSULES
-		int			m_numChildShapes;//used for compound shape
+	union {
+		float m_radius;        //used for childshape of SHAPE_COMPOUND_OF_SPHERES or SHAPE_COMPOUND_OF_CAPSULES
+		int m_numChildShapes;  //used for compound shape
 	};
-	union 
-	{
-		float		m_height;//used for childshape of SHAPE_COMPOUND_OF_CAPSULES
-		int	m_collidableShapeIndex;
+	union {
+		float m_height;  //used for childshape of SHAPE_COMPOUND_OF_CAPSULES
+		int m_collidableShapeIndex;
 	};
-	int			m_shapeType;
+	int m_shapeType;
 };
 
 struct b3CompoundOverlappingPair
 {
 	int m_bodyIndexA;
 	int m_bodyIndexB;
-//	int	m_pairType;
+	//	int	m_pairType;
 	int m_childShapeIndexA;
 	int m_childShapeIndexB;
 };
 
-#endif //B3_COLLIDABLE_H
+#endif  //B3_COLLIDABLE_H
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3Contact4Data.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3Contact4Data.h
index dfd45cc566e..d5f6daa9939 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3Contact4Data.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3Contact4Data.h
@@ -3,26 +3,24 @@
 
 #include "Bullet3Common/shared/b3Float4.h"
 
-typedef  struct b3Contact4Data b3Contact4Data_t;
+typedef struct b3Contact4Data b3Contact4Data_t;
 
 struct b3Contact4Data
 {
-	b3Float4	m_worldPosB[4];
-//	b3Float4	m_localPosA[4];
-//	b3Float4	m_localPosB[4];
-	b3Float4	m_worldNormalOnB;	//	w: m_nPoints
-	unsigned short  m_restituitionCoeffCmp;
-	unsigned short  m_frictionCoeffCmp;
+	b3Float4 m_worldPosB[4];
+	//	b3Float4	m_localPosA[4];
+	//	b3Float4	m_localPosB[4];
+	b3Float4 m_worldNormalOnB;  //	w: m_nPoints
+	unsigned short m_restituitionCoeffCmp;
+	unsigned short m_frictionCoeffCmp;
 	int m_batchIdx;
-	int m_bodyAPtrAndSignBit;//x:m_bodyAPtr, y:m_bodyBPtr
+	int m_bodyAPtrAndSignBit;  //x:m_bodyAPtr, y:m_bodyBPtr
 	int m_bodyBPtrAndSignBit;
 
-	int	m_childIndexA;
-	int	m_childIndexB;
+	int m_childIndexA;
+	int m_childIndexB;
 	int m_unused1;
 	int m_unused2;
-
-
 };
 
 inline int b3Contact4Data_getNumPoints(const struct b3Contact4Data* contact)
@@ -35,6 +33,4 @@ inline void b3Contact4Data_setNumPoints(struct b3Contact4Data* contact, int numP
 	contact->m_worldNormalOnB.w = (float)numPoints;
 };
 
-
-
-#endif //B3_CONTACT4DATA_H
\ No newline at end of file
+#endif  //B3_CONTACT4DATA_H
\ No newline at end of file
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3ContactConvexConvexSAT.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3ContactConvexConvexSAT.h
index f295f01a6c3..ca68f4bc4eb 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3ContactConvexConvexSAT.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3ContactConvexConvexSAT.h
@@ -2,48 +2,43 @@
 #ifndef B3_CONTACT_CONVEX_CONVEX_SAT_H
 #define B3_CONTACT_CONVEX_CONVEX_SAT_H
 
-
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3Contact4Data.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3FindSeparatingAxis.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3ReduceContacts.h"
 
 #define B3_MAX_VERTS 1024
 
-
-
-inline b3Float4 b3Lerp3(const b3Float4& a,const b3Float4& b, float  t)
+inline b3Float4 b3Lerp3(const b3Float4& a, const b3Float4& b, float t)
 {
-	return b3MakeVector3(	a.x + (b.x - a.x) * t,
-						a.y + (b.y - a.y) * t,
-						a.z + (b.z - a.z) * t,
-						0.f);
+	return b3MakeVector3(a.x + (b.x - a.x) * t,
+						 a.y + (b.y - a.y) * t,
+						 a.z + (b.z - a.z) * t,
+						 0.f);
 }
 
-
 // Clips a face to the back of a plane, return the number of vertices out, stored in ppVtxOut
-inline int b3ClipFace(const b3Float4* pVtxIn, int numVertsIn, b3Float4& planeNormalWS,float planeEqWS, b3Float4* ppVtxOut)
+inline int b3ClipFace(const b3Float4* pVtxIn, int numVertsIn, b3Float4& planeNormalWS, float planeEqWS, b3Float4* ppVtxOut)
 {
-	
 	int ve;
 	float ds, de;
 	int numVertsOut = 0;
 	if (numVertsIn < 2)
 		return 0;
 
-	b3Float4 firstVertex=pVtxIn[numVertsIn-1];
+	b3Float4 firstVertex = pVtxIn[numVertsIn - 1];
 	b3Float4 endVertex = pVtxIn[0];
-	
-	ds = b3Dot3F4(planeNormalWS,firstVertex)+planeEqWS;
+
+	ds = b3Dot3F4(planeNormalWS, firstVertex) + planeEqWS;
 
 	for (ve = 0; ve < numVertsIn; ve++)
 	{
-		endVertex=pVtxIn[ve];
+		endVertex = pVtxIn[ve];
 
-		de = b3Dot3F4(planeNormalWS,endVertex)+planeEqWS;
+		de = b3Dot3F4(planeNormalWS, endVertex) + planeEqWS;
 
-		if (ds<0)
+		if (ds < 0)
 		{
-			if (de<0)
+			if (de < 0)
 			{
 				// Start < 0, end < 0, so output endVertex
 				ppVtxOut[numVertsOut++] = endVertex;
@@ -51,15 +46,15 @@ inline int b3ClipFace(const b3Float4* pVtxIn, int numVertsIn, b3Float4& planeNor
 			else
 			{
 				// Start < 0, end >= 0, so output intersection
-				ppVtxOut[numVertsOut++] = b3Lerp3(firstVertex, endVertex,(ds * 1.f/(ds - de)) );
+				ppVtxOut[numVertsOut++] = b3Lerp3(firstVertex, endVertex, (ds * 1.f / (ds - de)));
 			}
 		}
 		else
 		{
-			if (de<0)
+			if (de < 0)
 			{
 				// Start >= 0, end < 0 so output intersection and end
-				ppVtxOut[numVertsOut++] = b3Lerp3(firstVertex, endVertex,(ds * 1.f/(ds - de)) );
+				ppVtxOut[numVertsOut++] = b3Lerp3(firstVertex, endVertex, (ds * 1.f / (ds - de)));
 				ppVtxOut[numVertsOut++] = endVertex;
 			}
 		}
@@ -69,36 +64,35 @@ inline int b3ClipFace(const b3Float4* pVtxIn, int numVertsIn, b3Float4& planeNor
 	return numVertsOut;
 }
 
-
-inline int b3ClipFaceAgainstHull(const b3Float4& separatingNormal, const b3ConvexPolyhedronData* hullA,  
-	const b3Float4& posA, const b3Quaternion& ornA, b3Float4* worldVertsB1, int numWorldVertsB1,
-	b3Float4* worldVertsB2, int capacityWorldVertsB2,
-	const float minDist, float maxDist,
-	const b3AlignedObjectArray<b3Float4>& verticesA,	const b3AlignedObjectArray<b3GpuFace>& facesA,	const b3AlignedObjectArray<int>& indicesA,
-	//const b3Float4* verticesB,	const b3GpuFace* facesB,	const int* indicesB,
-	b3Float4* contactsOut,
-	int contactCapacity)
+inline int b3ClipFaceAgainstHull(const b3Float4& separatingNormal, const b3ConvexPolyhedronData* hullA,
+								 const b3Float4& posA, const b3Quaternion& ornA, b3Float4* worldVertsB1, int numWorldVertsB1,
+								 b3Float4* worldVertsB2, int capacityWorldVertsB2,
+								 const float minDist, float maxDist,
+								 const b3AlignedObjectArray<b3Float4>& verticesA, const b3AlignedObjectArray<b3GpuFace>& facesA, const b3AlignedObjectArray<int>& indicesA,
+								 //const b3Float4* verticesB,	const b3GpuFace* facesB,	const int* indicesB,
+								 b3Float4* contactsOut,
+								 int contactCapacity)
 {
 	int numContactsOut = 0;
 
 	b3Float4* pVtxIn = worldVertsB1;
 	b3Float4* pVtxOut = worldVertsB2;
-	
+
 	int numVertsIn = numWorldVertsB1;
 	int numVertsOut = 0;
 
-	int closestFaceA=-1;
+	int closestFaceA = -1;
 	{
 		float dmin = FLT_MAX;
-		for(int face=0;face<hullA->m_numFaces;face++)
+		for (int face = 0; face < hullA->m_numFaces; face++)
 		{
 			const b3Float4 Normal = b3MakeVector3(
-				facesA[hullA->m_faceOffset+face].m_plane.x, 
-				facesA[hullA->m_faceOffset+face].m_plane.y, 
-				facesA[hullA->m_faceOffset+face].m_plane.z,0.f);
-			const b3Float4 faceANormalWS = b3QuatRotate(ornA,Normal);
-		
-			float d = b3Dot3F4(faceANormalWS,separatingNormal);
+				facesA[hullA->m_faceOffset + face].m_plane.x,
+				facesA[hullA->m_faceOffset + face].m_plane.y,
+				facesA[hullA->m_faceOffset + face].m_plane.z, 0.f);
+			const b3Float4 faceANormalWS = b3QuatRotate(ornA, Normal);
+
+			float d = b3Dot3F4(faceANormalWS, separatingNormal);
 			if (d < dmin)
 			{
 				dmin = d;
@@ -106,33 +100,33 @@ inline int b3ClipFaceAgainstHull(const b3Float4& separatingNormal, const b3Conve
 			}
 		}
 	}
-	if (closestFaceA<0)
+	if (closestFaceA < 0)
 		return numContactsOut;
 
-	b3GpuFace polyA = facesA[hullA->m_faceOffset+closestFaceA];
+	b3GpuFace polyA = facesA[hullA->m_faceOffset + closestFaceA];
 
 	// clip polygon to back of planes of all faces of hull A that are adjacent to witness face
 	//int numContacts = numWorldVertsB1;
 	int numVerticesA = polyA.m_numIndices;
-	for(int e0=0;e0<numVerticesA;e0++)
+	for (int e0 = 0; e0 < numVerticesA; e0++)
 	{
-		const b3Float4 a = verticesA[hullA->m_vertexOffset+indicesA[polyA.m_indexOffset+e0]];
-		const b3Float4 b = verticesA[hullA->m_vertexOffset+indicesA[polyA.m_indexOffset+((e0+1)%numVerticesA)]];
+		const b3Float4 a = verticesA[hullA->m_vertexOffset + indicesA[polyA.m_indexOffset + e0]];
+		const b3Float4 b = verticesA[hullA->m_vertexOffset + indicesA[polyA.m_indexOffset + ((e0 + 1) % numVerticesA)]];
 		const b3Float4 edge0 = a - b;
-		const b3Float4 WorldEdge0 = b3QuatRotate(ornA,edge0);
-		b3Float4 planeNormalA = b3MakeFloat4(polyA.m_plane.x,polyA.m_plane.y,polyA.m_plane.z,0.f);
-		b3Float4 worldPlaneAnormal1 = b3QuatRotate(ornA,planeNormalA);
+		const b3Float4 WorldEdge0 = b3QuatRotate(ornA, edge0);
+		b3Float4 planeNormalA = b3MakeFloat4(polyA.m_plane.x, polyA.m_plane.y, polyA.m_plane.z, 0.f);
+		b3Float4 worldPlaneAnormal1 = b3QuatRotate(ornA, planeNormalA);
+
+		b3Float4 planeNormalWS1 = -b3Cross3(WorldEdge0, worldPlaneAnormal1);
+		b3Float4 worldA1 = b3TransformPoint(a, posA, ornA);
+		float planeEqWS1 = -b3Dot3F4(worldA1, planeNormalWS1);
 
-		b3Float4 planeNormalWS1 = -b3Cross3(WorldEdge0,worldPlaneAnormal1);
-		b3Float4 worldA1 = b3TransformPoint(a,posA,ornA);
-		float planeEqWS1 = -b3Dot3F4(worldA1,planeNormalWS1);
-		
 		b3Float4 planeNormalWS = planeNormalWS1;
-		float planeEqWS=planeEqWS1;
-		
+		float planeEqWS = planeEqWS1;
+
 		//clip face
 		//clipFace(*pVtxIn, *pVtxOut,planeNormalWS,planeEqWS);
-		numVertsOut = b3ClipFace(pVtxIn, numVertsIn, planeNormalWS,planeEqWS, pVtxOut);
+		numVertsOut = b3ClipFace(pVtxIn, numVertsIn, planeNormalWS, planeEqWS, pVtxOut);
 
 		//btSwap(pVtxIn,pVtxOut);
 		b3Float4* tmp = pVtxOut;
@@ -142,32 +136,32 @@ inline int b3ClipFaceAgainstHull(const b3Float4& separatingNormal, const b3Conve
 		numVertsOut = 0;
 	}
 
-	
 	// only keep points that are behind the witness face
 	{
-		b3Float4 localPlaneNormal  = b3MakeFloat4(polyA.m_plane.x,polyA.m_plane.y,polyA.m_plane.z,0.f);
+		b3Float4 localPlaneNormal = b3MakeFloat4(polyA.m_plane.x, polyA.m_plane.y, polyA.m_plane.z, 0.f);
 		float localPlaneEq = polyA.m_plane.w;
-		b3Float4 planeNormalWS = b3QuatRotate(ornA,localPlaneNormal);
-		float planeEqWS=localPlaneEq-b3Dot3F4(planeNormalWS,posA);
-		for (int i=0;i<numVertsIn;i++)
+		b3Float4 planeNormalWS = b3QuatRotate(ornA, localPlaneNormal);
+		float planeEqWS = localPlaneEq - b3Dot3F4(planeNormalWS, posA);
+		for (int i = 0; i < numVertsIn; i++)
 		{
-			float depth = b3Dot3F4(planeNormalWS,pVtxIn[i])+planeEqWS;
-			if (depth <=minDist)
+			float depth = b3Dot3F4(planeNormalWS, pVtxIn[i]) + planeEqWS;
+			if (depth <= minDist)
 			{
 				depth = minDist;
 			}
-			if (numContactsOut<contactCapacity)
+			if (numContactsOut < contactCapacity)
 			{
-				if (depth <=maxDist)
+				if (depth <= maxDist)
 				{
 					b3Float4 pointInWorld = pVtxIn[i];
 					//resultOut.addContactPoint(separatingNormal,point,depth);
-					contactsOut[numContactsOut++] = b3MakeVector3(pointInWorld.x,pointInWorld.y,pointInWorld.z,depth);
+					contactsOut[numContactsOut++] = b3MakeVector3(pointInWorld.x, pointInWorld.y, pointInWorld.z, depth);
 					//printf("depth=%f\n",depth);
 				}
-			} else
+			}
+			else
 			{
-				b3Error("exceeding contact capacity (%d,%df)\n", numContactsOut,contactCapacity);
+				b3Error("exceeding contact capacity (%d,%df)\n", numContactsOut, contactCapacity);
 			}
 		}
 	}
@@ -175,62 +169,60 @@ inline int b3ClipFaceAgainstHull(const b3Float4& separatingNormal, const b3Conve
 	return numContactsOut;
 }
 
+inline int b3ClipHullAgainstHull(const b3Float4& separatingNormal,
+								 const b3ConvexPolyhedronData& hullA, const b3ConvexPolyhedronData& hullB,
+								 const b3Float4& posA, const b3Quaternion& ornA, const b3Float4& posB, const b3Quaternion& ornB,
+								 b3Float4* worldVertsB1, b3Float4* worldVertsB2, int capacityWorldVerts,
+								 const float minDist, float maxDist,
+								 const b3AlignedObjectArray<b3Float4>& verticesA, const b3AlignedObjectArray<b3GpuFace>& facesA, const b3AlignedObjectArray<int>& indicesA,
+								 const b3AlignedObjectArray<b3Float4>& verticesB, const b3AlignedObjectArray<b3GpuFace>& facesB, const b3AlignedObjectArray<int>& indicesB,
 
-
-inline int	b3ClipHullAgainstHull(const b3Float4& separatingNormal, 
-	const b3ConvexPolyhedronData& hullA, const b3ConvexPolyhedronData& hullB, 
-	const b3Float4& posA, const b3Quaternion& ornA,const b3Float4& posB, const b3Quaternion& ornB, 
-	b3Float4* worldVertsB1, b3Float4* worldVertsB2, int capacityWorldVerts,
-	const float minDist, float maxDist,
-	const b3AlignedObjectArray<b3Float4>& verticesA,	const b3AlignedObjectArray<b3GpuFace>& facesA,	const b3AlignedObjectArray<int>& indicesA,
-	const b3AlignedObjectArray<b3Float4>& verticesB,	const b3AlignedObjectArray<b3GpuFace>& facesB,	const b3AlignedObjectArray<int>& indicesB,
-
-	b3Float4*	contactsOut,
-	int contactCapacity)
+								 b3Float4* contactsOut,
+								 int contactCapacity)
 {
 	int numContactsOut = 0;
-	int numWorldVertsB1= 0;
-	
+	int numWorldVertsB1 = 0;
+
 	B3_PROFILE("clipHullAgainstHull");
 
 	//float curMaxDist=maxDist;
-	int closestFaceB=-1;
+	int closestFaceB = -1;
 	float dmax = -FLT_MAX;
 
 	{
 		//B3_PROFILE("closestFaceB");
-		if (hullB.m_numFaces!=1)
+		if (hullB.m_numFaces != 1)
 		{
 			//printf("wtf\n");
 		}
 		static bool once = true;
 		//printf("separatingNormal=%f,%f,%f\n",separatingNormal.x,separatingNormal.y,separatingNormal.z);
-		
-		for(int face=0;face<hullB.m_numFaces;face++)
+
+		for (int face = 0; face < hullB.m_numFaces; face++)
 		{
 #ifdef BT_DEBUG_SAT_FACE
 			if (once)
-				printf("face %d\n",face);
-			const b3GpuFace* faceB = &facesB[hullB.m_faceOffset+face];
+				printf("face %d\n", face);
+			const b3GpuFace* faceB = &facesB[hullB.m_faceOffset + face];
 			if (once)
 			{
-				for (int i=0;i<faceB->m_numIndices;i++)
+				for (int i = 0; i < faceB->m_numIndices; i++)
 				{
-					b3Float4 vert = verticesB[hullB.m_vertexOffset+indicesB[faceB->m_indexOffset+i]];
-					printf("vert[%d] = %f,%f,%f\n",i,vert.x,vert.y,vert.z);
+					b3Float4 vert = verticesB[hullB.m_vertexOffset + indicesB[faceB->m_indexOffset + i]];
+					printf("vert[%d] = %f,%f,%f\n", i, vert.x, vert.y, vert.z);
 				}
 			}
-#endif //BT_DEBUG_SAT_FACE
-			//if (facesB[hullB.m_faceOffset+face].m_numIndices>2)
+#endif  //BT_DEBUG_SAT_FACE \
+	//if (facesB[hullB.m_faceOffset+face].m_numIndices>2)
 			{
-				const b3Float4 Normal = b3MakeVector3(facesB[hullB.m_faceOffset+face].m_plane.x, 
-					facesB[hullB.m_faceOffset+face].m_plane.y, facesB[hullB.m_faceOffset+face].m_plane.z,0.f);
+				const b3Float4 Normal = b3MakeVector3(facesB[hullB.m_faceOffset + face].m_plane.x,
+													  facesB[hullB.m_faceOffset + face].m_plane.y, facesB[hullB.m_faceOffset + face].m_plane.z, 0.f);
 				const b3Float4 WorldNormal = b3QuatRotate(ornB, Normal);
 #ifdef BT_DEBUG_SAT_FACE
 				if (once)
-					printf("faceNormal = %f,%f,%f\n",Normal.x,Normal.y,Normal.z);
+					printf("faceNormal = %f,%f,%f\n", Normal.x, Normal.y, Normal.z);
 #endif
-				float d = b3Dot3F4(WorldNormal,separatingNormal);
+				float d = b3Dot3F4(WorldNormal, separatingNormal);
 				if (d > dmax)
 				{
 					dmax = d;
@@ -241,79 +233,73 @@ inline int	b3ClipHullAgainstHull(const b3Float4& separatingNormal,
 		once = false;
 	}
 
-	
-	b3Assert(closestFaceB>=0);
+	b3Assert(closestFaceB >= 0);
 	{
 		//B3_PROFILE("worldVertsB1");
-		const b3GpuFace& polyB = facesB[hullB.m_faceOffset+closestFaceB];
+		const b3GpuFace& polyB = facesB[hullB.m_faceOffset + closestFaceB];
 		const int numVertices = polyB.m_numIndices;
-		for(int e0=0;e0<numVertices;e0++)
+		for (int e0 = 0; e0 < numVertices; e0++)
 		{
-			const b3Float4& b = verticesB[hullB.m_vertexOffset+indicesB[polyB.m_indexOffset+e0]];
-			worldVertsB1[numWorldVertsB1++] = b3TransformPoint(b,posB,ornB);
+			const b3Float4& b = verticesB[hullB.m_vertexOffset + indicesB[polyB.m_indexOffset + e0]];
+			worldVertsB1[numWorldVertsB1++] = b3TransformPoint(b, posB, ornB);
 		}
 	}
 
-	if (closestFaceB>=0)
+	if (closestFaceB >= 0)
 	{
 		//B3_PROFILE("clipFaceAgainstHull");
-		numContactsOut = b3ClipFaceAgainstHull((b3Float4&)separatingNormal, &hullA, 
-				posA,ornA,
-				worldVertsB1,numWorldVertsB1,worldVertsB2,capacityWorldVerts, minDist, maxDist,
-				verticesA,				facesA,				indicesA,
-				contactsOut,contactCapacity);
+		numContactsOut = b3ClipFaceAgainstHull((b3Float4&)separatingNormal, &hullA,
+											   posA, ornA,
+											   worldVertsB1, numWorldVertsB1, worldVertsB2, capacityWorldVerts, minDist, maxDist,
+											   verticesA, facesA, indicesA,
+											   contactsOut, contactCapacity);
 	}
 
 	return numContactsOut;
 }
 
-
-
-
 inline int b3ClipHullHullSingle(
-			int bodyIndexA, int bodyIndexB,
-										 const b3Float4& posA,
-										 const b3Quaternion& ornA,
-										 const b3Float4& posB,
-										 const b3Quaternion& ornB,
+	int bodyIndexA, int bodyIndexB,
+	const b3Float4& posA,
+	const b3Quaternion& ornA,
+	const b3Float4& posB,
+	const b3Quaternion& ornB,
 
-			int collidableIndexA, int collidableIndexB,
+	int collidableIndexA, int collidableIndexB,
 
-			const b3AlignedObjectArray<b3RigidBodyData>* bodyBuf, 
-			b3AlignedObjectArray<b3Contact4Data>* globalContactOut, 
-			int& nContacts,
-			
-			const b3AlignedObjectArray<b3ConvexPolyhedronData>& hostConvexDataA,
-			const b3AlignedObjectArray<b3ConvexPolyhedronData>& hostConvexDataB,
-	
-			const b3AlignedObjectArray<b3Vector3>& verticesA, 
-			const b3AlignedObjectArray<b3Vector3>& uniqueEdgesA, 
-			const b3AlignedObjectArray<b3GpuFace>& facesA,
-			const b3AlignedObjectArray<int>& indicesA,
-	
-			const b3AlignedObjectArray<b3Vector3>& verticesB,
-			const b3AlignedObjectArray<b3Vector3>& uniqueEdgesB,
-			const b3AlignedObjectArray<b3GpuFace>& facesB,
-			const b3AlignedObjectArray<int>& indicesB,
+	const b3AlignedObjectArray<b3RigidBodyData>* bodyBuf,
+	b3AlignedObjectArray<b3Contact4Data>* globalContactOut,
+	int& nContacts,
 
-			const b3AlignedObjectArray<b3Collidable>& hostCollidablesA,
-			const b3AlignedObjectArray<b3Collidable>& hostCollidablesB,
-			const b3Vector3& sepNormalWorldSpace,
-			int maxContactCapacity			)
+	const b3AlignedObjectArray<b3ConvexPolyhedronData>& hostConvexDataA,
+	const b3AlignedObjectArray<b3ConvexPolyhedronData>& hostConvexDataB,
+
+	const b3AlignedObjectArray<b3Vector3>& verticesA,
+	const b3AlignedObjectArray<b3Vector3>& uniqueEdgesA,
+	const b3AlignedObjectArray<b3GpuFace>& facesA,
+	const b3AlignedObjectArray<int>& indicesA,
+
+	const b3AlignedObjectArray<b3Vector3>& verticesB,
+	const b3AlignedObjectArray<b3Vector3>& uniqueEdgesB,
+	const b3AlignedObjectArray<b3GpuFace>& facesB,
+	const b3AlignedObjectArray<int>& indicesB,
+
+	const b3AlignedObjectArray<b3Collidable>& hostCollidablesA,
+	const b3AlignedObjectArray<b3Collidable>& hostCollidablesB,
+	const b3Vector3& sepNormalWorldSpace,
+	int maxContactCapacity)
 {
 	int contactIndex = -1;
 	b3ConvexPolyhedronData hullA, hullB;
-    
-    b3Collidable colA = hostCollidablesA[collidableIndexA];
-    hullA = hostConvexDataA[colA.m_shapeIndex];
-    //printf("numvertsA = %d\n",hullA.m_numVertices);
-    
-    
-    b3Collidable colB = hostCollidablesB[collidableIndexB];
-    hullB = hostConvexDataB[colB.m_shapeIndex];
-    //printf("numvertsB = %d\n",hullB.m_numVertices);
-    
-	
+
+	b3Collidable colA = hostCollidablesA[collidableIndexA];
+	hullA = hostConvexDataA[colA.m_shapeIndex];
+	//printf("numvertsA = %d\n",hullA.m_numVertices);
+
+	b3Collidable colB = hostCollidablesB[collidableIndexB];
+	hullB = hostConvexDataB[colB.m_shapeIndex];
+	//printf("numvertsB = %d\n",hullB.m_numVertices);
+
 	b3Float4 contactsOut[B3_MAX_VERTS];
 	int localContactCapacity = B3_MAX_VERTS;
 
@@ -321,187 +307,168 @@ inline int b3ClipHullHullSingle(
 	b3Assert(_finite(bodyBuf->at(bodyIndexA).m_pos.x));
 	b3Assert(_finite(bodyBuf->at(bodyIndexB).m_pos.x));
 #endif
-	
-	
+
 	{
-		
 		b3Float4 worldVertsB1[B3_MAX_VERTS];
 		b3Float4 worldVertsB2[B3_MAX_VERTS];
 		int capacityWorldVerts = B3_MAX_VERTS;
 
-		b3Float4 hostNormal = b3MakeFloat4(sepNormalWorldSpace.x,sepNormalWorldSpace.y,sepNormalWorldSpace.z,0.f);
+		b3Float4 hostNormal = b3MakeFloat4(sepNormalWorldSpace.x, sepNormalWorldSpace.y, sepNormalWorldSpace.z, 0.f);
 		int shapeA = hostCollidablesA[collidableIndexA].m_shapeIndex;
 		int shapeB = hostCollidablesB[collidableIndexB].m_shapeIndex;
 
 		b3Scalar minDist = -1;
 		b3Scalar maxDist = 0.;
 
-		        
-
-		b3Transform trA,trB;
+		b3Transform trA, trB;
 		{
-		//B3_PROFILE("b3TransformPoint computation");
-		//trA.setIdentity();
-		trA.setOrigin(b3MakeVector3(posA.x,posA.y,posA.z));
-		trA.setRotation(b3Quaternion(ornA.x,ornA.y,ornA.z,ornA.w));
-				
-		//trB.setIdentity();
-		trB.setOrigin(b3MakeVector3(posB.x,posB.y,posB.z));
-		trB.setRotation(b3Quaternion(ornB.x,ornB.y,ornB.z,ornB.w));
+			//B3_PROFILE("b3TransformPoint computation");
+			//trA.setIdentity();
+			trA.setOrigin(b3MakeVector3(posA.x, posA.y, posA.z));
+			trA.setRotation(b3Quaternion(ornA.x, ornA.y, ornA.z, ornA.w));
+
+			//trB.setIdentity();
+			trB.setOrigin(b3MakeVector3(posB.x, posB.y, posB.z));
+			trB.setRotation(b3Quaternion(ornB.x, ornB.y, ornB.z, ornB.w));
 		}
 
 		b3Quaternion trAorn = trA.getRotation();
-        b3Quaternion trBorn = trB.getRotation();
-        
-		int numContactsOut = b3ClipHullAgainstHull(hostNormal, 
-						hostConvexDataA.at(shapeA), 
-						hostConvexDataB.at(shapeB),
-								(b3Float4&)trA.getOrigin(), (b3Quaternion&)trAorn,
-								(b3Float4&)trB.getOrigin(), (b3Quaternion&)trBorn,
-								worldVertsB1,worldVertsB2,capacityWorldVerts,
-								minDist, maxDist,
-								verticesA,	facesA,indicesA,
-								verticesB,	facesB,indicesB,
-								
-								contactsOut,localContactCapacity);
+		b3Quaternion trBorn = trB.getRotation();
 
-		if (numContactsOut>0)
+		int numContactsOut = b3ClipHullAgainstHull(hostNormal,
+												   hostConvexDataA.at(shapeA),
+												   hostConvexDataB.at(shapeB),
+												   (b3Float4&)trA.getOrigin(), (b3Quaternion&)trAorn,
+												   (b3Float4&)trB.getOrigin(), (b3Quaternion&)trBorn,
+												   worldVertsB1, worldVertsB2, capacityWorldVerts,
+												   minDist, maxDist,
+												   verticesA, facesA, indicesA,
+												   verticesB, facesB, indicesB,
+
+												   contactsOut, localContactCapacity);
+
+		if (numContactsOut > 0)
 		{
 			B3_PROFILE("overlap");
 
 			b3Float4 normalOnSurfaceB = (b3Float4&)hostNormal;
-//			b3Float4 centerOut;
-			
+			//			b3Float4 centerOut;
+
 			b3Int4 contactIdx;
 			contactIdx.x = 0;
 			contactIdx.y = 1;
 			contactIdx.z = 2;
 			contactIdx.w = 3;
-			
+
 			int numPoints = 0;
-					
+
 			{
 				B3_PROFILE("extractManifold");
-				numPoints = b3ReduceContacts(contactsOut, numContactsOut, normalOnSurfaceB,  &contactIdx);
+				numPoints = b3ReduceContacts(contactsOut, numContactsOut, normalOnSurfaceB, &contactIdx);
 			}
-					
+
 			b3Assert(numPoints);
-					
-			if (nContacts<maxContactCapacity)
+
+			if (nContacts < maxContactCapacity)
 			{
 				contactIndex = nContacts;
 				globalContactOut->expand();
 				b3Contact4Data& contact = globalContactOut->at(nContacts);
-				contact.m_batchIdx = 0;//i;
-				contact.m_bodyAPtrAndSignBit = (bodyBuf->at(bodyIndexA).m_invMass==0)? -bodyIndexA:bodyIndexA;
-				contact.m_bodyBPtrAndSignBit = (bodyBuf->at(bodyIndexB).m_invMass==0)? -bodyIndexB:bodyIndexB;
+				contact.m_batchIdx = 0;  //i;
+				contact.m_bodyAPtrAndSignBit = (bodyBuf->at(bodyIndexA).m_invMass == 0) ? -bodyIndexA : bodyIndexA;
+				contact.m_bodyBPtrAndSignBit = (bodyBuf->at(bodyIndexB).m_invMass == 0) ? -bodyIndexB : bodyIndexB;
 
 				contact.m_frictionCoeffCmp = 45874;
 				contact.m_restituitionCoeffCmp = 0;
-					
-			//	float distance = 0.f;
-				for (int p=0;p<numPoints;p++)
+
+				//	float distance = 0.f;
+				for (int p = 0; p < numPoints; p++)
 				{
-					contact.m_worldPosB[p] = contactsOut[contactIdx.s[p]];//check if it is actually on B
-					contact.m_worldNormalOnB = normalOnSurfaceB; 
+					contact.m_worldPosB[p] = contactsOut[contactIdx.s[p]];  //check if it is actually on B
+					contact.m_worldNormalOnB = normalOnSurfaceB;
 				}
 				//printf("bodyIndexA %d,bodyIndexB %d,normal=%f,%f,%f numPoints %d\n",bodyIndexA,bodyIndexB,normalOnSurfaceB.x,normalOnSurfaceB.y,normalOnSurfaceB.z,numPoints);
 				contact.m_worldNormalOnB.w = (b3Scalar)numPoints;
 				nContacts++;
-			} else
+			}
+			else
 			{
-				b3Error("Error: exceeding contact capacity (%d/%d)\n", nContacts,maxContactCapacity);
+				b3Error("Error: exceeding contact capacity (%d/%d)\n", nContacts, maxContactCapacity);
 			}
 		}
 	}
 	return contactIndex;
 }
 
-	
-
-
-
 inline int b3ContactConvexConvexSAT(
-																int pairIndex,
-																int bodyIndexA, int bodyIndexB, 
-																int collidableIndexA, int collidableIndexB, 
-																const b3AlignedObjectArray<b3RigidBodyData>& rigidBodies, 
-																const b3AlignedObjectArray<b3Collidable>& collidables,
-																const b3AlignedObjectArray<b3ConvexPolyhedronData>& convexShapes,
-																const b3AlignedObjectArray<b3Float4>& convexVertices,
-																const b3AlignedObjectArray<b3Float4>& uniqueEdges,
-																const b3AlignedObjectArray<int>& convexIndices,
-																const b3AlignedObjectArray<b3GpuFace>& faces,
-																b3AlignedObjectArray<b3Contact4Data>& globalContactsOut,
-																int& nGlobalContactsOut,
-																int maxContactCapacity)
+	int pairIndex,
+	int bodyIndexA, int bodyIndexB,
+	int collidableIndexA, int collidableIndexB,
+	const b3AlignedObjectArray<b3RigidBodyData>& rigidBodies,
+	const b3AlignedObjectArray<b3Collidable>& collidables,
+	const b3AlignedObjectArray<b3ConvexPolyhedronData>& convexShapes,
+	const b3AlignedObjectArray<b3Float4>& convexVertices,
+	const b3AlignedObjectArray<b3Float4>& uniqueEdges,
+	const b3AlignedObjectArray<int>& convexIndices,
+	const b3AlignedObjectArray<b3GpuFace>& faces,
+	b3AlignedObjectArray<b3Contact4Data>& globalContactsOut,
+	int& nGlobalContactsOut,
+	int maxContactCapacity)
 {
 	int contactIndex = -1;
 
-
 	b3Float4 posA = rigidBodies[bodyIndexA].m_pos;
 	b3Quaternion ornA = rigidBodies[bodyIndexA].m_quat;
 	b3Float4 posB = rigidBodies[bodyIndexB].m_pos;
 	b3Quaternion ornB = rigidBodies[bodyIndexB].m_quat;
-	
 
 	b3ConvexPolyhedronData hullA, hullB;
-    
+
 	b3Float4 sepNormalWorldSpace;
 
-	
-
-    b3Collidable colA = collidables[collidableIndexA];
-    hullA = convexShapes[colA.m_shapeIndex];
-    //printf("numvertsA = %d\n",hullA.m_numVertices);
-    
-    
-    b3Collidable colB = collidables[collidableIndexB];
-    hullB = convexShapes[colB.m_shapeIndex];
-    //printf("numvertsB = %d\n",hullB.m_numVertices);
-    
-	
+	b3Collidable colA = collidables[collidableIndexA];
+	hullA = convexShapes[colA.m_shapeIndex];
+	//printf("numvertsA = %d\n",hullA.m_numVertices);
 
+	b3Collidable colB = collidables[collidableIndexB];
+	hullB = convexShapes[colB.m_shapeIndex];
+	//printf("numvertsB = %d\n",hullB.m_numVertices);
 
 #ifdef _WIN32
 	b3Assert(_finite(rigidBodies[bodyIndexA].m_pos.x));
 	b3Assert(_finite(rigidBodies[bodyIndexB].m_pos.x));
 #endif
-	
-		bool foundSepAxis = b3FindSeparatingAxis(hullA,hullB,
-							posA,
-							ornA,
-							posB,
-							ornB,
 
-							convexVertices,uniqueEdges,faces,convexIndices,
-							convexVertices,uniqueEdges,faces,convexIndices,
-							
-							sepNormalWorldSpace
-							);
+	bool foundSepAxis = b3FindSeparatingAxis(hullA, hullB,
+											 posA,
+											 ornA,
+											 posB,
+											 ornB,
+
+											 convexVertices, uniqueEdges, faces, convexIndices,
+											 convexVertices, uniqueEdges, faces, convexIndices,
+
+											 sepNormalWorldSpace);
 
-	
 	if (foundSepAxis)
 	{
-		
-		
 		contactIndex = b3ClipHullHullSingle(
 			bodyIndexA, bodyIndexB,
-						   posA,ornA,
-						   posB,ornB,
+			posA, ornA,
+			posB, ornB,
 			collidableIndexA, collidableIndexB,
-			&rigidBodies, 
+			&rigidBodies,
 			&globalContactsOut,
 			nGlobalContactsOut,
-			
+
 			convexShapes,
 			convexShapes,
-	
-			convexVertices, 
-			uniqueEdges, 
+
+			convexVertices,
+			uniqueEdges,
 			faces,
 			convexIndices,
-	
+
 			convexVertices,
 			uniqueEdges,
 			faces,
@@ -511,10 +478,9 @@ inline int b3ContactConvexConvexSAT(
 			collidables,
 			sepNormalWorldSpace,
 			maxContactCapacity);
-			
 	}
 
 	return contactIndex;
 }
 
-#endif //B3_CONTACT_CONVEX_CONVEX_SAT_H
+#endif  //B3_CONTACT_CONVEX_CONVEX_SAT_H
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3ContactSphereSphere.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3ContactSphereSphere.h
index a3fa82287b8..acf7c1b1802 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3ContactSphereSphere.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3ContactSphereSphere.h
@@ -2,32 +2,24 @@
 #ifndef B3_CONTACT_SPHERE_SPHERE_H
 #define B3_CONTACT_SPHERE_SPHERE_H
 
-
-
-
-
-void	computeContactSphereConvex(int pairIndex,
-																int bodyIndexA, int bodyIndexB, 
-																int collidableIndexA, int collidableIndexB, 
-																const b3RigidBodyData* rigidBodies, 
-																const b3Collidable* collidables,
-																const b3ConvexPolyhedronData* convexShapes,
-																const b3Vector3* convexVertices,
-																const int* convexIndices,
-																const b3GpuFace* faces,
-																b3Contact4* globalContactsOut,
-																int& nGlobalContactsOut,
-																int maxContactCapacity)
+void computeContactSphereConvex(int pairIndex,
+								int bodyIndexA, int bodyIndexB,
+								int collidableIndexA, int collidableIndexB,
+								const b3RigidBodyData* rigidBodies,
+								const b3Collidable* collidables,
+								const b3ConvexPolyhedronData* convexShapes,
+								const b3Vector3* convexVertices,
+								const int* convexIndices,
+								const b3GpuFace* faces,
+								b3Contact4* globalContactsOut,
+								int& nGlobalContactsOut,
+								int maxContactCapacity)
 {
-
 	float radius = collidables[collidableIndexA].m_radius;
 	float4 spherePos1 = rigidBodies[bodyIndexA].m_pos;
 	b3Quaternion sphereOrn = rigidBodies[bodyIndexA].m_quat;
 
-
-
 	float4 pos = rigidBodies[bodyIndexB].m_pos;
-	
 
 	b3Quaternion quat = rigidBodies[bodyIndexB].m_quat;
 
@@ -44,63 +36,64 @@ void	computeContactSphereConvex(int pairIndex,
 	int numFaces = convexShapes[shapeIndex].m_numFaces;
 	float4 closestPnt = b3MakeVector3(0, 0, 0, 0);
 	float4 hitNormalWorld = b3MakeVector3(0, 0, 0, 0);
-	float minDist = -1000000.f; // TODO: What is the largest/smallest float?
+	float minDist = -1000000.f;  // TODO: What is the largest/smallest float?
 	bool bCollide = true;
 	int region = -1;
 	float4 localHitNormal;
-	for ( int f = 0; f < numFaces; f++ )
+	for (int f = 0; f < numFaces; f++)
 	{
-		b3GpuFace face = faces[convexShapes[shapeIndex].m_faceOffset+f];
+		b3GpuFace face = faces[convexShapes[shapeIndex].m_faceOffset + f];
 		float4 planeEqn;
-		float4 localPlaneNormal = b3MakeVector3(face.m_plane.x,face.m_plane.y,face.m_plane.z,0.f);
-		float4 n1 = localPlaneNormal;//quatRotate(quat,localPlaneNormal);
+		float4 localPlaneNormal = b3MakeVector3(face.m_plane.x, face.m_plane.y, face.m_plane.z, 0.f);
+		float4 n1 = localPlaneNormal;  //quatRotate(quat,localPlaneNormal);
 		planeEqn = n1;
 		planeEqn[3] = face.m_plane.w;
 
 		float4 pntReturn;
 		float dist = signedDistanceFromPointToPlane(spherePos, planeEqn, &pntReturn);
 
-		if ( dist > radius)
+		if (dist > radius)
 		{
 			bCollide = false;
 			break;
 		}
 
-		if ( dist > 0 )
+		if (dist > 0)
 		{
 			//might hit an edge or vertex
 			b3Vector3 out;
 
 			bool isInPoly = IsPointInPolygon(spherePos,
-					&face,
-					&convexVertices[convexShapes[shapeIndex].m_vertexOffset],
-					convexIndices,
-                    &out);
+											 &face,
+											 &convexVertices[convexShapes[shapeIndex].m_vertexOffset],
+											 convexIndices,
+											 &out);
 			if (isInPoly)
 			{
-				if (dist>minDist)
+				if (dist > minDist)
 				{
 					minDist = dist;
 					closestPnt = pntReturn;
 					localHitNormal = planeEqn;
-					region=1;
+					region = 1;
 				}
-			} else
+			}
+			else
 			{
-				b3Vector3 tmp = spherePos-out;
+				b3Vector3 tmp = spherePos - out;
 				b3Scalar l2 = tmp.length2();
-				if (l2<radius*radius)
+				if (l2 < radius * radius)
 				{
-					dist  = b3Sqrt(l2);
-					if (dist>minDist)
+					dist = b3Sqrt(l2);
+					if (dist > minDist)
 					{
 						minDist = dist;
 						closestPnt = out;
-						localHitNormal = tmp/dist;
-						region=2;
+						localHitNormal = tmp / dist;
+						region = 2;
 					}
-					
-				} else
+				}
+				else
 				{
 					bCollide = false;
 					break;
@@ -109,12 +102,12 @@ void	computeContactSphereConvex(int pairIndex,
 		}
 		else
 		{
-			if ( dist > minDist )
+			if (dist > minDist)
 			{
 				minDist = dist;
 				closestPnt = pntReturn;
 				localHitNormal = planeEqn;
-				region=3;
+				region = 3;
 			}
 		}
 	}
@@ -123,40 +116,38 @@ void	computeContactSphereConvex(int pairIndex,
 
 	if (bCollide && minDist > -10000)
 	{
-		
-		float4 normalOnSurfaceB1 = tr.getBasis()*localHitNormal;//-hitNormalWorld;
+		float4 normalOnSurfaceB1 = tr.getBasis() * localHitNormal;  //-hitNormalWorld;
 		float4 pOnB1 = tr(closestPnt);
 		//printf("dist ,%f,",minDist);
-		float actualDepth = minDist-radius;
-		if (actualDepth<0)
+		float actualDepth = minDist - radius;
+		if (actualDepth < 0)
 		{
-		//printf("actualDepth = ,%f,", actualDepth);
-		//printf("normalOnSurfaceB1 = ,%f,%f,%f,", normalOnSurfaceB1.x,normalOnSurfaceB1.y,normalOnSurfaceB1.z);
-		//printf("region=,%d,\n", region);
-		pOnB1[3] = actualDepth;
+			//printf("actualDepth = ,%f,", actualDepth);
+			//printf("normalOnSurfaceB1 = ,%f,%f,%f,", normalOnSurfaceB1.x,normalOnSurfaceB1.y,normalOnSurfaceB1.z);
+			//printf("region=,%d,\n", region);
+			pOnB1[3] = actualDepth;
 
-		int dstIdx;
-//    dstIdx = nGlobalContactsOut++;//AppendInc( nGlobalContactsOut, dstIdx );
-		
-		if (nGlobalContactsOut < maxContactCapacity)
-		{
-			dstIdx=nGlobalContactsOut;
-			nGlobalContactsOut++;
+			int dstIdx;
+			//    dstIdx = nGlobalContactsOut++;//AppendInc( nGlobalContactsOut, dstIdx );
 
-			b3Contact4* c = &globalContactsOut[dstIdx];
-			c->m_worldNormalOnB = normalOnSurfaceB1;
-			c->setFrictionCoeff(0.7);
-			c->setRestituitionCoeff(0.f);
+			if (nGlobalContactsOut < maxContactCapacity)
+			{
+				dstIdx = nGlobalContactsOut;
+				nGlobalContactsOut++;
 
-			c->m_batchIdx = pairIndex;
-			c->m_bodyAPtrAndSignBit = rigidBodies[bodyIndexA].m_invMass==0?-bodyIndexA:bodyIndexA;
-			c->m_bodyBPtrAndSignBit = rigidBodies[bodyIndexB].m_invMass==0?-bodyIndexB:bodyIndexB;
-			c->m_worldPosB[0] = pOnB1;
-			int numPoints = 1;
-			c->m_worldNormalOnB.w = (b3Scalar)numPoints;
-		}//if (dstIdx < numPairs)
+				b3Contact4* c = &globalContactsOut[dstIdx];
+				c->m_worldNormalOnB = normalOnSurfaceB1;
+				c->setFrictionCoeff(0.7);
+				c->setRestituitionCoeff(0.f);
+
+				c->m_batchIdx = pairIndex;
+				c->m_bodyAPtrAndSignBit = rigidBodies[bodyIndexA].m_invMass == 0 ? -bodyIndexA : bodyIndexA;
+				c->m_bodyBPtrAndSignBit = rigidBodies[bodyIndexB].m_invMass == 0 ? -bodyIndexB : bodyIndexB;
+				c->m_worldPosB[0] = pOnB1;
+				int numPoints = 1;
+				c->m_worldNormalOnB.w = (b3Scalar)numPoints;
+			}  //if (dstIdx < numPairs)
 		}
-	}//if (hasCollision)
-	
+	}  //if (hasCollision)
 }
-#endif //B3_CONTACT_SPHERE_SPHERE_H
+#endif  //B3_CONTACT_SPHERE_SPHERE_H
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3ConvexPolyhedronData.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3ConvexPolyhedronData.h
index 5c5f4e297fc..d5a73bd4f57 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3ConvexPolyhedronData.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3ConvexPolyhedronData.h
@@ -2,8 +2,6 @@
 #ifndef B3_CONVEX_POLYHEDRON_DATA_H
 #define B3_CONVEX_POLYHEDRON_DATA_H
 
-
-
 #include "Bullet3Common/shared/b3Float4.h"
 #include "Bullet3Common/shared/b3Quat.h"
 
@@ -21,20 +19,20 @@ typedef struct b3ConvexPolyhedronData b3ConvexPolyhedronData_t;
 
 struct b3ConvexPolyhedronData
 {
-	b3Float4		m_localCenter;
-	b3Float4		m_extents;
-	b3Float4		mC;
-	b3Float4		mE;
+	b3Float4 m_localCenter;
+	b3Float4 m_extents;
+	b3Float4 mC;
+	b3Float4 mE;
 
-	float			m_radius;
-	int	m_faceOffset;
+	float m_radius;
+	int m_faceOffset;
 	int m_numFaces;
-	int	m_numVertices;
+	int m_numVertices;
 
 	int m_vertexOffset;
-	int	m_uniqueEdgesOffset;
-	int	m_numUniqueEdges;
+	int m_uniqueEdgesOffset;
+	int m_numUniqueEdges;
 	int m_unused;
 };
 
-#endif //B3_CONVEX_POLYHEDRON_DATA_H
+#endif  //B3_CONVEX_POLYHEDRON_DATA_H
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3FindConcaveSatAxis.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3FindConcaveSatAxis.h
index 89993f35654..983554eb2e5 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3FindConcaveSatAxis.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3FindConcaveSatAxis.h
@@ -3,7 +3,6 @@
 
 #define B3_TRIANGLE_NUM_CONVEX_FACES 5
 
-
 #include "Bullet3Common/shared/b3Int4.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3RigidBodyData.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3Collidable.h"
@@ -12,25 +11,24 @@
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3QuantizedBvhNodeData.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3ConvexPolyhedronData.h"
 
-
-inline void b3Project(__global const b3ConvexPolyhedronData* hull,  b3Float4ConstArg pos, b3QuatConstArg orn, 
-const b3Float4* dir, __global const b3Float4* vertices, float* min, float* max)
+inline void b3Project(__global const b3ConvexPolyhedronData* hull, b3Float4ConstArg pos, b3QuatConstArg orn,
+					  const b3Float4* dir, __global const b3Float4* vertices, float* min, float* max)
 {
 	min[0] = FLT_MAX;
 	max[0] = -FLT_MAX;
 	int numVerts = hull->m_numVertices;
 
-	const b3Float4 localDir = b3QuatRotate(b3QuatInverse(orn),*dir);
-	float offset = b3Dot(pos,*dir);
-	for(int i=0;i<numVerts;i++)
+	const b3Float4 localDir = b3QuatRotate(b3QuatInverse(orn), *dir);
+	float offset = b3Dot(pos, *dir);
+	for (int i = 0; i < numVerts; i++)
 	{
-		float dp = b3Dot(vertices[hull->m_vertexOffset+i],localDir);
-		if(dp < min[0])	
+		float dp = b3Dot(vertices[hull->m_vertexOffset + i], localDir);
+		if (dp < min[0])
 			min[0] = dp;
-		if(dp > max[0])	
+		if (dp > max[0])
 			max[0] = dp;
 	}
-	if(min[0]>max[0])
+	if (min[0] > max[0])
 	{
 		float tmp = min[0];
 		min[0] = max[0];
@@ -40,53 +38,49 @@ const b3Float4* dir, __global const b3Float4* vertices, float* min, float* max)
 	max[0] += offset;
 }
 
-
-inline bool b3TestSepAxis(const b3ConvexPolyhedronData* hullA, __global const b3ConvexPolyhedronData* hullB, 
-	b3Float4ConstArg posA,b3QuatConstArg ornA,
-	b3Float4ConstArg posB,b3QuatConstArg ornB,
-	b3Float4* sep_axis, const b3Float4* verticesA, __global const b3Float4* verticesB,float* depth)
+inline bool b3TestSepAxis(const b3ConvexPolyhedronData* hullA, __global const b3ConvexPolyhedronData* hullB,
+						  b3Float4ConstArg posA, b3QuatConstArg ornA,
+						  b3Float4ConstArg posB, b3QuatConstArg ornB,
+						  b3Float4* sep_axis, const b3Float4* verticesA, __global const b3Float4* verticesB, float* depth)
 {
-	float Min0,Max0;
-	float Min1,Max1;
-	b3Project(hullA,posA,ornA,sep_axis,verticesA, &Min0, &Max0);
-	b3Project(hullB,posB,ornB, sep_axis,verticesB, &Min1, &Max1);
+	float Min0, Max0;
+	float Min1, Max1;
+	b3Project(hullA, posA, ornA, sep_axis, verticesA, &Min0, &Max0);
+	b3Project(hullB, posB, ornB, sep_axis, verticesB, &Min1, &Max1);
 
-	if(Max0<Min1 || Max1<Min0)
+	if (Max0 < Min1 || Max1 < Min0)
 		return false;
 
 	float d0 = Max0 - Min1;
 	float d1 = Max1 - Min0;
-	*depth = d0<d1 ? d0:d1;
+	*depth = d0 < d1 ? d0 : d1;
 	return true;
 }
 
+bool b3FindSeparatingAxis(const b3ConvexPolyhedronData* hullA, __global const b3ConvexPolyhedronData* hullB,
+						  b3Float4ConstArg posA1,
+						  b3QuatConstArg ornA,
+						  b3Float4ConstArg posB1,
+						  b3QuatConstArg ornB,
+						  b3Float4ConstArg DeltaC2,
 
-bool b3FindSeparatingAxis(	const b3ConvexPolyhedronData* hullA, __global const b3ConvexPolyhedronData* hullB, 
-	b3Float4ConstArg posA1,
-	b3QuatConstArg ornA,
-	b3Float4ConstArg posB1,
-	b3QuatConstArg ornB,
-	b3Float4ConstArg DeltaC2,
-	
-	const b3Float4* verticesA, 
-	const b3Float4* uniqueEdgesA, 
-	const b3GpuFace* facesA,
-	const int*  indicesA,
+						  const b3Float4* verticesA,
+						  const b3Float4* uniqueEdgesA,
+						  const b3GpuFace* facesA,
+						  const int* indicesA,
 
-	__global const b3Float4* verticesB, 
-	__global const b3Float4* uniqueEdgesB, 
-	__global const b3GpuFace* facesB,
-	__global const int*  indicesB,
-	b3Float4* sep,
-	float* dmin)
+						  __global const b3Float4* verticesB,
+						  __global const b3Float4* uniqueEdgesB,
+						  __global const b3GpuFace* facesB,
+						  __global const int* indicesB,
+						  b3Float4* sep,
+						  float* dmin)
 {
-	
-
 	b3Float4 posA = posA1;
 	posA.w = 0.f;
 	b3Float4 posB = posB1;
 	posB.w = 0.f;
-/*
+	/*
 	static int maxFaceVertex = 0;
 
 	int curFaceVertexAB = hullA->m_numFaces*hullB->m_numVertices;
@@ -102,300 +96,289 @@ bool b3FindSeparatingAxis(	const b3ConvexPolyhedronData* hullA, __global const b
 	}
 */
 
-	int curPlaneTests=0;
+	int curPlaneTests = 0;
 	{
 		int numFacesA = hullA->m_numFaces;
 		// Test normals from hullA
-		for(int i=0;i<numFacesA;i++)
+		for (int i = 0; i < numFacesA; i++)
 		{
-			const b3Float4 normal = facesA[hullA->m_faceOffset+i].m_plane;
-			b3Float4 faceANormalWS = b3QuatRotate(ornA,normal);
-			if (b3Dot(DeltaC2,faceANormalWS)<0)
-				faceANormalWS*=-1.f;
+			const b3Float4 normal = facesA[hullA->m_faceOffset + i].m_plane;
+			b3Float4 faceANormalWS = b3QuatRotate(ornA, normal);
+			if (b3Dot(DeltaC2, faceANormalWS) < 0)
+				faceANormalWS *= -1.f;
 			curPlaneTests++;
 			float d;
-			if(!b3TestSepAxis( hullA, hullB, posA,ornA,posB,ornB,&faceANormalWS, verticesA, verticesB,&d))
+			if (!b3TestSepAxis(hullA, hullB, posA, ornA, posB, ornB, &faceANormalWS, verticesA, verticesB, &d))
 				return false;
-			if(d<*dmin)
+			if (d < *dmin)
 			{
 				*dmin = d;
 				*sep = faceANormalWS;
 			}
 		}
 	}
-	if((b3Dot(-DeltaC2,*sep))>0.0f)
+	if ((b3Dot(-DeltaC2, *sep)) > 0.0f)
 	{
 		*sep = -(*sep);
 	}
 	return true;
 }
 
+b3Vector3 unitSphere162[] =
+	{
+		b3MakeVector3(0.000000, -1.000000, 0.000000),
+		b3MakeVector3(0.203181, -0.967950, 0.147618),
+		b3MakeVector3(-0.077607, -0.967950, 0.238853),
+		b3MakeVector3(0.723607, -0.447220, 0.525725),
+		b3MakeVector3(0.609547, -0.657519, 0.442856),
+		b3MakeVector3(0.812729, -0.502301, 0.295238),
+		b3MakeVector3(-0.251147, -0.967949, 0.000000),
+		b3MakeVector3(-0.077607, -0.967950, -0.238853),
+		b3MakeVector3(0.203181, -0.967950, -0.147618),
+		b3MakeVector3(0.860698, -0.251151, 0.442858),
+		b3MakeVector3(-0.276388, -0.447220, 0.850649),
+		b3MakeVector3(-0.029639, -0.502302, 0.864184),
+		b3MakeVector3(-0.155215, -0.251152, 0.955422),
+		b3MakeVector3(-0.894426, -0.447216, 0.000000),
+		b3MakeVector3(-0.831051, -0.502299, 0.238853),
+		b3MakeVector3(-0.956626, -0.251149, 0.147618),
+		b3MakeVector3(-0.276388, -0.447220, -0.850649),
+		b3MakeVector3(-0.483971, -0.502302, -0.716565),
+		b3MakeVector3(-0.436007, -0.251152, -0.864188),
+		b3MakeVector3(0.723607, -0.447220, -0.525725),
+		b3MakeVector3(0.531941, -0.502302, -0.681712),
+		b3MakeVector3(0.687159, -0.251152, -0.681715),
+		b3MakeVector3(0.687159, -0.251152, 0.681715),
+		b3MakeVector3(-0.436007, -0.251152, 0.864188),
+		b3MakeVector3(-0.956626, -0.251149, -0.147618),
+		b3MakeVector3(-0.155215, -0.251152, -0.955422),
+		b3MakeVector3(0.860698, -0.251151, -0.442858),
+		b3MakeVector3(0.276388, 0.447220, 0.850649),
+		b3MakeVector3(0.483971, 0.502302, 0.716565),
+		b3MakeVector3(0.232822, 0.657519, 0.716563),
+		b3MakeVector3(-0.723607, 0.447220, 0.525725),
+		b3MakeVector3(-0.531941, 0.502302, 0.681712),
+		b3MakeVector3(-0.609547, 0.657519, 0.442856),
+		b3MakeVector3(-0.723607, 0.447220, -0.525725),
+		b3MakeVector3(-0.812729, 0.502301, -0.295238),
+		b3MakeVector3(-0.609547, 0.657519, -0.442856),
+		b3MakeVector3(0.276388, 0.447220, -0.850649),
+		b3MakeVector3(0.029639, 0.502302, -0.864184),
+		b3MakeVector3(0.232822, 0.657519, -0.716563),
+		b3MakeVector3(0.894426, 0.447216, 0.000000),
+		b3MakeVector3(0.831051, 0.502299, -0.238853),
+		b3MakeVector3(0.753442, 0.657515, 0.000000),
+		b3MakeVector3(-0.232822, -0.657519, 0.716563),
+		b3MakeVector3(-0.162456, -0.850654, 0.499995),
+		b3MakeVector3(0.052790, -0.723612, 0.688185),
+		b3MakeVector3(0.138199, -0.894429, 0.425321),
+		b3MakeVector3(0.262869, -0.525738, 0.809012),
+		b3MakeVector3(0.361805, -0.723611, 0.587779),
+		b3MakeVector3(0.531941, -0.502302, 0.681712),
+		b3MakeVector3(0.425323, -0.850654, 0.309011),
+		b3MakeVector3(0.812729, -0.502301, -0.295238),
+		b3MakeVector3(0.609547, -0.657519, -0.442856),
+		b3MakeVector3(0.850648, -0.525736, 0.000000),
+		b3MakeVector3(0.670817, -0.723611, -0.162457),
+		b3MakeVector3(0.670817, -0.723610, 0.162458),
+		b3MakeVector3(0.425323, -0.850654, -0.309011),
+		b3MakeVector3(0.447211, -0.894428, 0.000001),
+		b3MakeVector3(-0.753442, -0.657515, 0.000000),
+		b3MakeVector3(-0.525730, -0.850652, 0.000000),
+		b3MakeVector3(-0.638195, -0.723609, 0.262864),
+		b3MakeVector3(-0.361801, -0.894428, 0.262864),
+		b3MakeVector3(-0.688189, -0.525736, 0.499997),
+		b3MakeVector3(-0.447211, -0.723610, 0.525729),
+		b3MakeVector3(-0.483971, -0.502302, 0.716565),
+		b3MakeVector3(-0.232822, -0.657519, -0.716563),
+		b3MakeVector3(-0.162456, -0.850654, -0.499995),
+		b3MakeVector3(-0.447211, -0.723611, -0.525727),
+		b3MakeVector3(-0.361801, -0.894429, -0.262863),
+		b3MakeVector3(-0.688189, -0.525736, -0.499997),
+		b3MakeVector3(-0.638195, -0.723609, -0.262863),
+		b3MakeVector3(-0.831051, -0.502299, -0.238853),
+		b3MakeVector3(0.361804, -0.723612, -0.587779),
+		b3MakeVector3(0.138197, -0.894429, -0.425321),
+		b3MakeVector3(0.262869, -0.525738, -0.809012),
+		b3MakeVector3(0.052789, -0.723611, -0.688186),
+		b3MakeVector3(-0.029639, -0.502302, -0.864184),
+		b3MakeVector3(0.956626, 0.251149, 0.147618),
+		b3MakeVector3(0.956626, 0.251149, -0.147618),
+		b3MakeVector3(0.951058, -0.000000, 0.309013),
+		b3MakeVector3(1.000000, 0.000000, 0.000000),
+		b3MakeVector3(0.947213, -0.276396, 0.162458),
+		b3MakeVector3(0.951058, 0.000000, -0.309013),
+		b3MakeVector3(0.947213, -0.276396, -0.162458),
+		b3MakeVector3(0.155215, 0.251152, 0.955422),
+		b3MakeVector3(0.436007, 0.251152, 0.864188),
+		b3MakeVector3(-0.000000, -0.000000, 1.000000),
+		b3MakeVector3(0.309017, 0.000000, 0.951056),
+		b3MakeVector3(0.138199, -0.276398, 0.951055),
+		b3MakeVector3(0.587786, 0.000000, 0.809017),
+		b3MakeVector3(0.447216, -0.276398, 0.850648),
+		b3MakeVector3(-0.860698, 0.251151, 0.442858),
+		b3MakeVector3(-0.687159, 0.251152, 0.681715),
+		b3MakeVector3(-0.951058, -0.000000, 0.309013),
+		b3MakeVector3(-0.809018, 0.000000, 0.587783),
+		b3MakeVector3(-0.861803, -0.276396, 0.425324),
+		b3MakeVector3(-0.587786, 0.000000, 0.809017),
+		b3MakeVector3(-0.670819, -0.276397, 0.688191),
+		b3MakeVector3(-0.687159, 0.251152, -0.681715),
+		b3MakeVector3(-0.860698, 0.251151, -0.442858),
+		b3MakeVector3(-0.587786, -0.000000, -0.809017),
+		b3MakeVector3(-0.809018, -0.000000, -0.587783),
+		b3MakeVector3(-0.670819, -0.276397, -0.688191),
+		b3MakeVector3(-0.951058, 0.000000, -0.309013),
+		b3MakeVector3(-0.861803, -0.276396, -0.425324),
+		b3MakeVector3(0.436007, 0.251152, -0.864188),
+		b3MakeVector3(0.155215, 0.251152, -0.955422),
+		b3MakeVector3(0.587786, -0.000000, -0.809017),
+		b3MakeVector3(0.309017, -0.000000, -0.951056),
+		b3MakeVector3(0.447216, -0.276398, -0.850648),
+		b3MakeVector3(0.000000, 0.000000, -1.000000),
+		b3MakeVector3(0.138199, -0.276398, -0.951055),
+		b3MakeVector3(0.670820, 0.276396, 0.688190),
+		b3MakeVector3(0.809019, -0.000002, 0.587783),
+		b3MakeVector3(0.688189, 0.525736, 0.499997),
+		b3MakeVector3(0.861804, 0.276394, 0.425323),
+		b3MakeVector3(0.831051, 0.502299, 0.238853),
+		b3MakeVector3(-0.447216, 0.276397, 0.850649),
+		b3MakeVector3(-0.309017, -0.000001, 0.951056),
+		b3MakeVector3(-0.262869, 0.525738, 0.809012),
+		b3MakeVector3(-0.138199, 0.276397, 0.951055),
+		b3MakeVector3(0.029639, 0.502302, 0.864184),
+		b3MakeVector3(-0.947213, 0.276396, -0.162458),
+		b3MakeVector3(-1.000000, 0.000001, 0.000000),
+		b3MakeVector3(-0.850648, 0.525736, -0.000000),
+		b3MakeVector3(-0.947213, 0.276397, 0.162458),
+		b3MakeVector3(-0.812729, 0.502301, 0.295238),
+		b3MakeVector3(-0.138199, 0.276397, -0.951055),
+		b3MakeVector3(-0.309016, -0.000000, -0.951057),
+		b3MakeVector3(-0.262869, 0.525738, -0.809012),
+		b3MakeVector3(-0.447215, 0.276397, -0.850649),
+		b3MakeVector3(-0.531941, 0.502302, -0.681712),
+		b3MakeVector3(0.861804, 0.276396, -0.425322),
+		b3MakeVector3(0.809019, 0.000000, -0.587782),
+		b3MakeVector3(0.688189, 0.525736, -0.499997),
+		b3MakeVector3(0.670821, 0.276397, -0.688189),
+		b3MakeVector3(0.483971, 0.502302, -0.716565),
+		b3MakeVector3(0.077607, 0.967950, 0.238853),
+		b3MakeVector3(0.251147, 0.967949, 0.000000),
+		b3MakeVector3(0.000000, 1.000000, 0.000000),
+		b3MakeVector3(0.162456, 0.850654, 0.499995),
+		b3MakeVector3(0.361800, 0.894429, 0.262863),
+		b3MakeVector3(0.447209, 0.723612, 0.525728),
+		b3MakeVector3(0.525730, 0.850652, 0.000000),
+		b3MakeVector3(0.638194, 0.723610, 0.262864),
+		b3MakeVector3(-0.203181, 0.967950, 0.147618),
+		b3MakeVector3(-0.425323, 0.850654, 0.309011),
+		b3MakeVector3(-0.138197, 0.894430, 0.425320),
+		b3MakeVector3(-0.361804, 0.723612, 0.587778),
+		b3MakeVector3(-0.052790, 0.723612, 0.688185),
+		b3MakeVector3(-0.203181, 0.967950, -0.147618),
+		b3MakeVector3(-0.425323, 0.850654, -0.309011),
+		b3MakeVector3(-0.447210, 0.894429, 0.000000),
+		b3MakeVector3(-0.670817, 0.723611, -0.162457),
+		b3MakeVector3(-0.670817, 0.723611, 0.162457),
+		b3MakeVector3(0.077607, 0.967950, -0.238853),
+		b3MakeVector3(0.162456, 0.850654, -0.499995),
+		b3MakeVector3(-0.138197, 0.894430, -0.425320),
+		b3MakeVector3(-0.052790, 0.723612, -0.688185),
+		b3MakeVector3(-0.361804, 0.723612, -0.587778),
+		b3MakeVector3(0.361800, 0.894429, -0.262863),
+		b3MakeVector3(0.638194, 0.723610, -0.262864),
+		b3MakeVector3(0.447209, 0.723612, -0.525728)};
 
-b3Vector3 unitSphere162[]=
+bool b3FindSeparatingAxisEdgeEdge(const b3ConvexPolyhedronData* hullA, __global const b3ConvexPolyhedronData* hullB,
+								  b3Float4ConstArg posA1,
+								  b3QuatConstArg ornA,
+								  b3Float4ConstArg posB1,
+								  b3QuatConstArg ornB,
+								  b3Float4ConstArg DeltaC2,
+								  const b3Float4* verticesA,
+								  const b3Float4* uniqueEdgesA,
+								  const b3GpuFace* facesA,
+								  const int* indicesA,
+								  __global const b3Float4* verticesB,
+								  __global const b3Float4* uniqueEdgesB,
+								  __global const b3GpuFace* facesB,
+								  __global const int* indicesB,
+								  b3Float4* sep,
+								  float* dmin,
+								  bool searchAllEdgeEdge)
 {
-	b3MakeVector3(0.000000,-1.000000,0.000000),
-b3MakeVector3(0.203181,-0.967950,0.147618),
-b3MakeVector3(-0.077607,-0.967950,0.238853),
-b3MakeVector3(0.723607,-0.447220,0.525725),
-b3MakeVector3(0.609547,-0.657519,0.442856),
-b3MakeVector3(0.812729,-0.502301,0.295238),
-b3MakeVector3(-0.251147,-0.967949,0.000000),
-b3MakeVector3(-0.077607,-0.967950,-0.238853),
-b3MakeVector3(0.203181,-0.967950,-0.147618),
-b3MakeVector3(0.860698,-0.251151,0.442858),
-b3MakeVector3(-0.276388,-0.447220,0.850649),
-b3MakeVector3(-0.029639,-0.502302,0.864184),
-b3MakeVector3(-0.155215,-0.251152,0.955422),
-b3MakeVector3(-0.894426,-0.447216,0.000000),
-b3MakeVector3(-0.831051,-0.502299,0.238853),
-b3MakeVector3(-0.956626,-0.251149,0.147618),
-b3MakeVector3(-0.276388,-0.447220,-0.850649),
-b3MakeVector3(-0.483971,-0.502302,-0.716565),
-b3MakeVector3(-0.436007,-0.251152,-0.864188),
-b3MakeVector3(0.723607,-0.447220,-0.525725),
-b3MakeVector3(0.531941,-0.502302,-0.681712),
-b3MakeVector3(0.687159,-0.251152,-0.681715),
-b3MakeVector3(0.687159,-0.251152,0.681715),
-b3MakeVector3(-0.436007,-0.251152,0.864188),
-b3MakeVector3(-0.956626,-0.251149,-0.147618),
-b3MakeVector3(-0.155215,-0.251152,-0.955422),
-b3MakeVector3(0.860698,-0.251151,-0.442858),
-b3MakeVector3(0.276388,0.447220,0.850649),
-b3MakeVector3(0.483971,0.502302,0.716565),
-b3MakeVector3(0.232822,0.657519,0.716563),
-b3MakeVector3(-0.723607,0.447220,0.525725),
-b3MakeVector3(-0.531941,0.502302,0.681712),
-b3MakeVector3(-0.609547,0.657519,0.442856),
-b3MakeVector3(-0.723607,0.447220,-0.525725),
-b3MakeVector3(-0.812729,0.502301,-0.295238),
-b3MakeVector3(-0.609547,0.657519,-0.442856),
-b3MakeVector3(0.276388,0.447220,-0.850649),
-b3MakeVector3(0.029639,0.502302,-0.864184),
-b3MakeVector3(0.232822,0.657519,-0.716563),
-b3MakeVector3(0.894426,0.447216,0.000000),
-b3MakeVector3(0.831051,0.502299,-0.238853),
-b3MakeVector3(0.753442,0.657515,0.000000),
-b3MakeVector3(-0.232822,-0.657519,0.716563),
-b3MakeVector3(-0.162456,-0.850654,0.499995),
-b3MakeVector3(0.052790,-0.723612,0.688185),
-b3MakeVector3(0.138199,-0.894429,0.425321),
-b3MakeVector3(0.262869,-0.525738,0.809012),
-b3MakeVector3(0.361805,-0.723611,0.587779),
-b3MakeVector3(0.531941,-0.502302,0.681712),
-b3MakeVector3(0.425323,-0.850654,0.309011),
-b3MakeVector3(0.812729,-0.502301,-0.295238),
-b3MakeVector3(0.609547,-0.657519,-0.442856),
-b3MakeVector3(0.850648,-0.525736,0.000000),
-b3MakeVector3(0.670817,-0.723611,-0.162457),
-b3MakeVector3(0.670817,-0.723610,0.162458),
-b3MakeVector3(0.425323,-0.850654,-0.309011),
-b3MakeVector3(0.447211,-0.894428,0.000001),
-b3MakeVector3(-0.753442,-0.657515,0.000000),
-b3MakeVector3(-0.525730,-0.850652,0.000000),
-b3MakeVector3(-0.638195,-0.723609,0.262864),
-b3MakeVector3(-0.361801,-0.894428,0.262864),
-b3MakeVector3(-0.688189,-0.525736,0.499997),
-b3MakeVector3(-0.447211,-0.723610,0.525729),
-b3MakeVector3(-0.483971,-0.502302,0.716565),
-b3MakeVector3(-0.232822,-0.657519,-0.716563),
-b3MakeVector3(-0.162456,-0.850654,-0.499995),
-b3MakeVector3(-0.447211,-0.723611,-0.525727),
-b3MakeVector3(-0.361801,-0.894429,-0.262863),
-b3MakeVector3(-0.688189,-0.525736,-0.499997),
-b3MakeVector3(-0.638195,-0.723609,-0.262863),
-b3MakeVector3(-0.831051,-0.502299,-0.238853),
-b3MakeVector3(0.361804,-0.723612,-0.587779),
-b3MakeVector3(0.138197,-0.894429,-0.425321),
-b3MakeVector3(0.262869,-0.525738,-0.809012),
-b3MakeVector3(0.052789,-0.723611,-0.688186),
-b3MakeVector3(-0.029639,-0.502302,-0.864184),
-b3MakeVector3(0.956626,0.251149,0.147618),
-b3MakeVector3(0.956626,0.251149,-0.147618),
-b3MakeVector3(0.951058,-0.000000,0.309013),
-b3MakeVector3(1.000000,0.000000,0.000000),
-b3MakeVector3(0.947213,-0.276396,0.162458),
-b3MakeVector3(0.951058,0.000000,-0.309013),
-b3MakeVector3(0.947213,-0.276396,-0.162458),
-b3MakeVector3(0.155215,0.251152,0.955422),
-b3MakeVector3(0.436007,0.251152,0.864188),
-b3MakeVector3(-0.000000,-0.000000,1.000000),
-b3MakeVector3(0.309017,0.000000,0.951056),
-b3MakeVector3(0.138199,-0.276398,0.951055),
-b3MakeVector3(0.587786,0.000000,0.809017),
-b3MakeVector3(0.447216,-0.276398,0.850648),
-b3MakeVector3(-0.860698,0.251151,0.442858),
-b3MakeVector3(-0.687159,0.251152,0.681715),
-b3MakeVector3(-0.951058,-0.000000,0.309013),
-b3MakeVector3(-0.809018,0.000000,0.587783),
-b3MakeVector3(-0.861803,-0.276396,0.425324),
-b3MakeVector3(-0.587786,0.000000,0.809017),
-b3MakeVector3(-0.670819,-0.276397,0.688191),
-b3MakeVector3(-0.687159,0.251152,-0.681715),
-b3MakeVector3(-0.860698,0.251151,-0.442858),
-b3MakeVector3(-0.587786,-0.000000,-0.809017),
-b3MakeVector3(-0.809018,-0.000000,-0.587783),
-b3MakeVector3(-0.670819,-0.276397,-0.688191),
-b3MakeVector3(-0.951058,0.000000,-0.309013),
-b3MakeVector3(-0.861803,-0.276396,-0.425324),
-b3MakeVector3(0.436007,0.251152,-0.864188),
-b3MakeVector3(0.155215,0.251152,-0.955422),
-b3MakeVector3(0.587786,-0.000000,-0.809017),
-b3MakeVector3(0.309017,-0.000000,-0.951056),
-b3MakeVector3(0.447216,-0.276398,-0.850648),
-b3MakeVector3(0.000000,0.000000,-1.000000),
-b3MakeVector3(0.138199,-0.276398,-0.951055),
-b3MakeVector3(0.670820,0.276396,0.688190),
-b3MakeVector3(0.809019,-0.000002,0.587783),
-b3MakeVector3(0.688189,0.525736,0.499997),
-b3MakeVector3(0.861804,0.276394,0.425323),
-b3MakeVector3(0.831051,0.502299,0.238853),
-b3MakeVector3(-0.447216,0.276397,0.850649),
-b3MakeVector3(-0.309017,-0.000001,0.951056),
-b3MakeVector3(-0.262869,0.525738,0.809012),
-b3MakeVector3(-0.138199,0.276397,0.951055),
-b3MakeVector3(0.029639,0.502302,0.864184),
-b3MakeVector3(-0.947213,0.276396,-0.162458),
-b3MakeVector3(-1.000000,0.000001,0.000000),
-b3MakeVector3(-0.850648,0.525736,-0.000000),
-b3MakeVector3(-0.947213,0.276397,0.162458),
-b3MakeVector3(-0.812729,0.502301,0.295238),
-b3MakeVector3(-0.138199,0.276397,-0.951055),
-b3MakeVector3(-0.309016,-0.000000,-0.951057),
-b3MakeVector3(-0.262869,0.525738,-0.809012),
-b3MakeVector3(-0.447215,0.276397,-0.850649),
-b3MakeVector3(-0.531941,0.502302,-0.681712),
-b3MakeVector3(0.861804,0.276396,-0.425322),
-b3MakeVector3(0.809019,0.000000,-0.587782),
-b3MakeVector3(0.688189,0.525736,-0.499997),
-b3MakeVector3(0.670821,0.276397,-0.688189),
-b3MakeVector3(0.483971,0.502302,-0.716565),
-b3MakeVector3(0.077607,0.967950,0.238853),
-b3MakeVector3(0.251147,0.967949,0.000000),
-b3MakeVector3(0.000000,1.000000,0.000000),
-b3MakeVector3(0.162456,0.850654,0.499995),
-b3MakeVector3(0.361800,0.894429,0.262863),
-b3MakeVector3(0.447209,0.723612,0.525728),
-b3MakeVector3(0.525730,0.850652,0.000000),
-b3MakeVector3(0.638194,0.723610,0.262864),
-b3MakeVector3(-0.203181,0.967950,0.147618),
-b3MakeVector3(-0.425323,0.850654,0.309011),
-b3MakeVector3(-0.138197,0.894430,0.425320),
-b3MakeVector3(-0.361804,0.723612,0.587778),
-b3MakeVector3(-0.052790,0.723612,0.688185),
-b3MakeVector3(-0.203181,0.967950,-0.147618),
-b3MakeVector3(-0.425323,0.850654,-0.309011),
-b3MakeVector3(-0.447210,0.894429,0.000000),
-b3MakeVector3(-0.670817,0.723611,-0.162457),
-b3MakeVector3(-0.670817,0.723611,0.162457),
-b3MakeVector3(0.077607,0.967950,-0.238853),
-b3MakeVector3(0.162456,0.850654,-0.499995),
-b3MakeVector3(-0.138197,0.894430,-0.425320),
-b3MakeVector3(-0.052790,0.723612,-0.688185),
-b3MakeVector3(-0.361804,0.723612,-0.587778),
-b3MakeVector3(0.361800,0.894429,-0.262863),
-b3MakeVector3(0.638194,0.723610,-0.262864),
-b3MakeVector3(0.447209,0.723612,-0.525728)
-};
-
-
-bool b3FindSeparatingAxisEdgeEdge(	const b3ConvexPolyhedronData* hullA, __global const b3ConvexPolyhedronData* hullB, 
-	b3Float4ConstArg posA1,
-	b3QuatConstArg ornA,
-	b3Float4ConstArg posB1,
-	b3QuatConstArg ornB,
-	b3Float4ConstArg DeltaC2,
-	const b3Float4* verticesA, 
-	const b3Float4* uniqueEdgesA, 
-	const b3GpuFace* facesA,
-	const int*  indicesA,
-	__global const b3Float4* verticesB, 
-	__global const b3Float4* uniqueEdgesB, 
-	__global const b3GpuFace* facesB,
-	__global const int*  indicesB,
-		b3Float4* sep,
-	float* dmin,
-	bool searchAllEdgeEdge)
-{
-
-
 	b3Float4 posA = posA1;
 	posA.w = 0.f;
 	b3Float4 posB = posB1;
 	posB.w = 0.f;
 
-//	int curPlaneTests=0;
+	//	int curPlaneTests=0;
 
 	int curEdgeEdge = 0;
 	// Test edges
 	static int maxEdgeTests = 0;
 	int curEdgeTests = hullA->m_numUniqueEdges * hullB->m_numUniqueEdges;
-	if (curEdgeTests >maxEdgeTests )
+	if (curEdgeTests > maxEdgeTests)
 	{
-		maxEdgeTests  = curEdgeTests ;
-		printf("maxEdgeTests = %d\n",maxEdgeTests );
-		printf("hullA->m_numUniqueEdges = %d\n",hullA->m_numUniqueEdges);
-		printf("hullB->m_numUniqueEdges = %d\n",hullB->m_numUniqueEdges);
-
+		maxEdgeTests = curEdgeTests;
+		printf("maxEdgeTests = %d\n", maxEdgeTests);
+		printf("hullA->m_numUniqueEdges = %d\n", hullA->m_numUniqueEdges);
+		printf("hullB->m_numUniqueEdges = %d\n", hullB->m_numUniqueEdges);
 	}
 
-	
 	if (searchAllEdgeEdge)
 	{
-		for(int e0=0;e0<hullA->m_numUniqueEdges;e0++)
+		for (int e0 = 0; e0 < hullA->m_numUniqueEdges; e0++)
 		{
-			const b3Float4 edge0 = uniqueEdgesA[hullA->m_uniqueEdgesOffset+e0];
-			b3Float4 edge0World = b3QuatRotate(ornA,edge0);
+			const b3Float4 edge0 = uniqueEdgesA[hullA->m_uniqueEdgesOffset + e0];
+			b3Float4 edge0World = b3QuatRotate(ornA, edge0);
 
-			for(int e1=0;e1<hullB->m_numUniqueEdges;e1++)
+			for (int e1 = 0; e1 < hullB->m_numUniqueEdges; e1++)
 			{
-				const b3Float4 edge1 = uniqueEdgesB[hullB->m_uniqueEdgesOffset+e1];
-				b3Float4 edge1World = b3QuatRotate(ornB,edge1);
+				const b3Float4 edge1 = uniqueEdgesB[hullB->m_uniqueEdgesOffset + e1];
+				b3Float4 edge1World = b3QuatRotate(ornB, edge1);
 
-
-				b3Float4 crossje = b3Cross(edge0World,edge1World);
+				b3Float4 crossje = b3Cross(edge0World, edge1World);
 
 				curEdgeEdge++;
-				if(!b3IsAlmostZero(crossje))
+				if (!b3IsAlmostZero(crossje))
 				{
 					crossje = b3Normalized(crossje);
-					if (b3Dot(DeltaC2,crossje)<0)
+					if (b3Dot(DeltaC2, crossje) < 0)
 						crossje *= -1.f;
 
 					float dist;
 					bool result = true;
 					{
-						float Min0,Max0;
-						float Min1,Max1;
-						b3Project(hullA,posA,ornA,&crossje,verticesA, &Min0, &Max0);
-						b3Project(hullB,posB,ornB,&crossje,verticesB, &Min1, &Max1);
-				
-						if(Max0<Min1 || Max1<Min0)
+						float Min0, Max0;
+						float Min1, Max1;
+						b3Project(hullA, posA, ornA, &crossje, verticesA, &Min0, &Max0);
+						b3Project(hullB, posB, ornB, &crossje, verticesB, &Min1, &Max1);
+
+						if (Max0 < Min1 || Max1 < Min0)
 							return false;
-                    
+
 						float d0 = Max0 - Min1;
 						float d1 = Max1 - Min0;
-						dist = d0<d1 ? d0:d1;
+						dist = d0 < d1 ? d0 : d1;
 						result = true;
-
 					}
-				
 
-					if(dist<*dmin)
+					if (dist < *dmin)
 					{
 						*dmin = dist;
 						*sep = crossje;
 					}
 				}
 			}
-
 		}
-	} else
+	}
+	else
 	{
-		int numDirections = sizeof(unitSphere162)/sizeof(b3Vector3);
+		int numDirections = sizeof(unitSphere162) / sizeof(b3Vector3);
 		//printf("numDirections =%d\n",numDirections );
 
-
-		for(int i=0;i<numDirections;i++)
+		for (int i = 0; i < numDirections; i++)
 		{
 			b3Float4 crossje = unitSphere162[i];
 			{
@@ -404,23 +387,21 @@ bool b3FindSeparatingAxisEdgeEdge(	const b3ConvexPolyhedronData* hullA, __global
 					float dist;
 					bool result = true;
 					{
-						float Min0,Max0;
-						float Min1,Max1;
-						b3Project(hullA,posA,ornA,&crossje,verticesA, &Min0, &Max0);
-						b3Project(hullB,posB,ornB,&crossje,verticesB, &Min1, &Max1);
-				
-						if(Max0<Min1 || Max1<Min0)
+						float Min0, Max0;
+						float Min1, Max1;
+						b3Project(hullA, posA, ornA, &crossje, verticesA, &Min0, &Max0);
+						b3Project(hullB, posB, ornB, &crossje, verticesB, &Min1, &Max1);
+
+						if (Max0 < Min1 || Max1 < Min0)
 							return false;
-                    
+
 						float d0 = Max0 - Min1;
 						float d1 = Max1 - Min0;
-						dist = d0<d1 ? d0:d1;
+						dist = d0 < d1 ? d0 : d1;
 						result = true;
-
 					}
-				
 
-					if(dist<*dmin)
+					if (dist < *dmin)
 					{
 						*dmin = dist;
 						*sep = crossje;
@@ -428,50 +409,45 @@ bool b3FindSeparatingAxisEdgeEdge(	const b3ConvexPolyhedronData* hullA, __global
 				}
 			}
 		}
-
 	}
 
-	
-	if((b3Dot(-DeltaC2,*sep))>0.0f)
+	if ((b3Dot(-DeltaC2, *sep)) > 0.0f)
 	{
 		*sep = -(*sep);
 	}
 	return true;
 }
 
+inline int b3FindClippingFaces(b3Float4ConstArg separatingNormal,
+							   __global const b3ConvexPolyhedronData_t* hullA, __global const b3ConvexPolyhedronData_t* hullB,
+							   b3Float4ConstArg posA, b3QuatConstArg ornA, b3Float4ConstArg posB, b3QuatConstArg ornB,
+							   __global b3Float4* worldVertsA1,
+							   __global b3Float4* worldNormalsA1,
+							   __global b3Float4* worldVertsB1,
+							   int capacityWorldVerts,
+							   const float minDist, float maxDist,
+							   __global const b3Float4* verticesA,
+							   __global const b3GpuFace_t* facesA,
+							   __global const int* indicesA,
+							   __global const b3Float4* verticesB,
+							   __global const b3GpuFace_t* facesB,
+							   __global const int* indicesB,
 
-
-inline int	b3FindClippingFaces(b3Float4ConstArg separatingNormal,
-                      __global const b3ConvexPolyhedronData_t* hullA, __global const b3ConvexPolyhedronData_t* hullB,
-                      b3Float4ConstArg posA, b3QuatConstArg ornA,b3Float4ConstArg posB, b3QuatConstArg ornB,
-                       __global b3Float4* worldVertsA1,
-                      __global b3Float4* worldNormalsA1,
-                      __global b3Float4* worldVertsB1,
-                      int capacityWorldVerts,
-                      const float minDist, float maxDist,
-                      __global const b3Float4* verticesA,
-                      __global const b3GpuFace_t* facesA,
-                      __global const int* indicesA,
-						__global const b3Float4* verticesB,
-                      __global const b3GpuFace_t* facesB,
-                      __global const int* indicesB,
-
-                      __global b3Int4* clippingFaces, int pairIndex)
+							   __global b3Int4* clippingFaces, int pairIndex)
 {
 	int numContactsOut = 0;
-	int numWorldVertsB1= 0;
-    
-    
-	int closestFaceB=-1;
+	int numWorldVertsB1 = 0;
+
+	int closestFaceB = -1;
 	float dmax = -FLT_MAX;
-    
+
 	{
-		for(int face=0;face<hullB->m_numFaces;face++)
+		for (int face = 0; face < hullB->m_numFaces; face++)
 		{
-			const b3Float4 Normal = b3MakeFloat4(facesB[hullB->m_faceOffset+face].m_plane.x,
-                                              facesB[hullB->m_faceOffset+face].m_plane.y, facesB[hullB->m_faceOffset+face].m_plane.z,0.f);
+			const b3Float4 Normal = b3MakeFloat4(facesB[hullB->m_faceOffset + face].m_plane.x,
+												 facesB[hullB->m_faceOffset + face].m_plane.y, facesB[hullB->m_faceOffset + face].m_plane.z, 0.f);
 			const b3Float4 WorldNormal = b3QuatRotate(ornB, Normal);
-			float d = b3Dot(WorldNormal,separatingNormal);
+			float d = b3Dot(WorldNormal, separatingNormal);
 			if (d > dmax)
 			{
 				dmax = d;
@@ -479,81 +455,76 @@ inline int	b3FindClippingFaces(b3Float4ConstArg separatingNormal,
 			}
 		}
 	}
-    
+
 	{
-		const b3GpuFace_t polyB = facesB[hullB->m_faceOffset+closestFaceB];
+		const b3GpuFace_t polyB = facesB[hullB->m_faceOffset + closestFaceB];
 		const int numVertices = polyB.m_numIndices;
-		for(int e0=0;e0<numVertices;e0++)
+		for (int e0 = 0; e0 < numVertices; e0++)
 		{
-			const b3Float4 b = verticesB[hullB->m_vertexOffset+indicesB[polyB.m_indexOffset+e0]];
-			worldVertsB1[pairIndex*capacityWorldVerts+numWorldVertsB1++] = b3TransformPoint(b,posB,ornB);
+			const b3Float4 b = verticesB[hullB->m_vertexOffset + indicesB[polyB.m_indexOffset + e0]];
+			worldVertsB1[pairIndex * capacityWorldVerts + numWorldVertsB1++] = b3TransformPoint(b, posB, ornB);
 		}
 	}
-    
-    int closestFaceA=-1;
+
+	int closestFaceA = -1;
 	{
 		float dmin = FLT_MAX;
-		for(int face=0;face<hullA->m_numFaces;face++)
+		for (int face = 0; face < hullA->m_numFaces; face++)
 		{
 			const b3Float4 Normal = b3MakeFloat4(
-                                              facesA[hullA->m_faceOffset+face].m_plane.x,
-                                              facesA[hullA->m_faceOffset+face].m_plane.y,
-                                              facesA[hullA->m_faceOffset+face].m_plane.z,
-                                              0.f);
-			const b3Float4 faceANormalWS = b3QuatRotate(ornA,Normal);
-            
-			float d = b3Dot(faceANormalWS,separatingNormal);
+				facesA[hullA->m_faceOffset + face].m_plane.x,
+				facesA[hullA->m_faceOffset + face].m_plane.y,
+				facesA[hullA->m_faceOffset + face].m_plane.z,
+				0.f);
+			const b3Float4 faceANormalWS = b3QuatRotate(ornA, Normal);
+
+			float d = b3Dot(faceANormalWS, separatingNormal);
 			if (d < dmin)
 			{
 				dmin = d;
 				closestFaceA = face;
-                worldNormalsA1[pairIndex] = faceANormalWS;
+				worldNormalsA1[pairIndex] = faceANormalWS;
 			}
 		}
 	}
-    
-    int numVerticesA = facesA[hullA->m_faceOffset+closestFaceA].m_numIndices;
-	for(int e0=0;e0<numVerticesA;e0++)
+
+	int numVerticesA = facesA[hullA->m_faceOffset + closestFaceA].m_numIndices;
+	for (int e0 = 0; e0 < numVerticesA; e0++)
 	{
-        const b3Float4 a = verticesA[hullA->m_vertexOffset+indicesA[facesA[hullA->m_faceOffset+closestFaceA].m_indexOffset+e0]];
-        worldVertsA1[pairIndex*capacityWorldVerts+e0] = b3TransformPoint(a, posA,ornA);
-    }
-    
-    clippingFaces[pairIndex].x = closestFaceA;
-    clippingFaces[pairIndex].y = closestFaceB;
-    clippingFaces[pairIndex].z = numVerticesA;
-    clippingFaces[pairIndex].w = numWorldVertsB1;
-    
-    
+		const b3Float4 a = verticesA[hullA->m_vertexOffset + indicesA[facesA[hullA->m_faceOffset + closestFaceA].m_indexOffset + e0]];
+		worldVertsA1[pairIndex * capacityWorldVerts + e0] = b3TransformPoint(a, posA, ornA);
+	}
+
+	clippingFaces[pairIndex].x = closestFaceA;
+	clippingFaces[pairIndex].y = closestFaceB;
+	clippingFaces[pairIndex].z = numVerticesA;
+	clippingFaces[pairIndex].w = numWorldVertsB1;
+
 	return numContactsOut;
 }
 
-
-        
-
-__kernel void   b3FindConcaveSeparatingAxisKernel( __global b3Int4* concavePairs,
-																					__global const b3RigidBodyData* rigidBodies,
-																					__global const b3Collidable* collidables,
-																					__global const b3ConvexPolyhedronData* convexShapes, 
-																					__global const b3Float4* vertices,
-																					__global const b3Float4* uniqueEdges,
-																					__global const b3GpuFace* faces,
-																					__global const int* indices,
-																					__global const b3GpuChildShape* gpuChildShapes,
-																					__global b3Aabb* aabbs,
-																					__global b3Float4* concaveSeparatingNormalsOut,
-																					__global b3Int4* clippingFacesOut,
-																					__global b3Vector3* worldVertsA1Out,
-																					__global b3Vector3* worldNormalsA1Out,
-																					__global b3Vector3* worldVertsB1Out,
-																					__global int* hasSeparatingNormals,
-																					int vertexFaceCapacity,
-																					int numConcavePairs,
-																					int pairIdx
-																					)
+__kernel void b3FindConcaveSeparatingAxisKernel(__global b3Int4* concavePairs,
+												__global const b3RigidBodyData* rigidBodies,
+												__global const b3Collidable* collidables,
+												__global const b3ConvexPolyhedronData* convexShapes,
+												__global const b3Float4* vertices,
+												__global const b3Float4* uniqueEdges,
+												__global const b3GpuFace* faces,
+												__global const int* indices,
+												__global const b3GpuChildShape* gpuChildShapes,
+												__global b3Aabb* aabbs,
+												__global b3Float4* concaveSeparatingNormalsOut,
+												__global b3Int4* clippingFacesOut,
+												__global b3Vector3* worldVertsA1Out,
+												__global b3Vector3* worldNormalsA1Out,
+												__global b3Vector3* worldVertsB1Out,
+												__global int* hasSeparatingNormals,
+												int vertexFaceCapacity,
+												int numConcavePairs,
+												int pairIdx)
 {
 	int i = pairIdx;
-/*	int i = get_global_id(0);
+	/*	int i = get_global_id(0);
 	if (i>=numConcavePairs)
 		return;
 	int pairIdx = i;
@@ -568,8 +539,8 @@ __kernel void   b3FindConcaveSeparatingAxisKernel( __global b3Int4* concavePairs
 	int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;
 	int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;
 
-	if (collidables[collidableIndexB].m_shapeType!=SHAPE_CONVEX_HULL&&
-		collidables[collidableIndexB].m_shapeType!=SHAPE_COMPOUND_OF_CONVEX_HULLS)
+	if (collidables[collidableIndexB].m_shapeType != SHAPE_CONVEX_HULL &&
+		collidables[collidableIndexB].m_shapeType != SHAPE_COMPOUND_OF_CONVEX_HULLS)
 	{
 		concavePairs[pairIdx].w = -1;
 		return;
@@ -577,121 +548,119 @@ __kernel void   b3FindConcaveSeparatingAxisKernel( __global b3Int4* concavePairs
 
 	hasSeparatingNormals[i] = 0;
 
-//	int numFacesA = convexShapes[shapeIndexA].m_numFaces;
+	//	int numFacesA = convexShapes[shapeIndexA].m_numFaces;
 	int numActualConcaveConvexTests = 0;
-	
+
 	int f = concavePairs[i].z;
-	
+
 	bool overlap = false;
-	
+
 	b3ConvexPolyhedronData convexPolyhedronA;
 
 	//add 3 vertices of the triangle
 	convexPolyhedronA.m_numVertices = 3;
 	convexPolyhedronA.m_vertexOffset = 0;
-	b3Float4	localCenter = b3MakeFloat4(0.f,0.f,0.f,0.f);
+	b3Float4 localCenter = b3MakeFloat4(0.f, 0.f, 0.f, 0.f);
 
-	b3GpuFace face = faces[convexShapes[shapeIndexA].m_faceOffset+f];
+	b3GpuFace face = faces[convexShapes[shapeIndexA].m_faceOffset + f];
 	b3Aabb triAabb;
-	triAabb.m_minVec = b3MakeFloat4(1e30f,1e30f,1e30f,0.f);
-	triAabb.m_maxVec = b3MakeFloat4(-1e30f,-1e30f,-1e30f,0.f);
-	
+	triAabb.m_minVec = b3MakeFloat4(1e30f, 1e30f, 1e30f, 0.f);
+	triAabb.m_maxVec = b3MakeFloat4(-1e30f, -1e30f, -1e30f, 0.f);
+
 	b3Float4 verticesA[3];
-	for (int i=0;i<3;i++)
+	for (int i = 0; i < 3; i++)
 	{
-		int index = indices[face.m_indexOffset+i];
-		b3Float4 vert = vertices[convexShapes[shapeIndexA].m_vertexOffset+index];
+		int index = indices[face.m_indexOffset + i];
+		b3Float4 vert = vertices[convexShapes[shapeIndexA].m_vertexOffset + index];
 		verticesA[i] = vert;
 		localCenter += vert;
-			
-		triAabb.m_minVec = b3MinFloat4(triAabb.m_minVec,vert);		
-		triAabb.m_maxVec = b3MaxFloat4(triAabb.m_maxVec,vert);		
 
+		triAabb.m_minVec = b3MinFloat4(triAabb.m_minVec, vert);
+		triAabb.m_maxVec = b3MaxFloat4(triAabb.m_maxVec, vert);
 	}
 
 	overlap = true;
 	overlap = (triAabb.m_minVec.x > aabbs[bodyIndexB].m_maxVec.x || triAabb.m_maxVec.x < aabbs[bodyIndexB].m_minVec.x) ? false : overlap;
 	overlap = (triAabb.m_minVec.z > aabbs[bodyIndexB].m_maxVec.z || triAabb.m_maxVec.z < aabbs[bodyIndexB].m_minVec.z) ? false : overlap;
 	overlap = (triAabb.m_minVec.y > aabbs[bodyIndexB].m_maxVec.y || triAabb.m_maxVec.y < aabbs[bodyIndexB].m_minVec.y) ? false : overlap;
-		
+
 	if (overlap)
 	{
 		float dmin = FLT_MAX;
-		int hasSeparatingAxis=5;
-		b3Float4 sepAxis=b3MakeFloat4(1,2,3,4);
+		int hasSeparatingAxis = 5;
+		b3Float4 sepAxis = b3MakeFloat4(1, 2, 3, 4);
 
-	//	int localCC=0;
+		//	int localCC=0;
 		numActualConcaveConvexTests++;
 
 		//a triangle has 3 unique edges
 		convexPolyhedronA.m_numUniqueEdges = 3;
 		convexPolyhedronA.m_uniqueEdgesOffset = 0;
 		b3Float4 uniqueEdgesA[3];
-		
-		uniqueEdgesA[0] = (verticesA[1]-verticesA[0]);
-		uniqueEdgesA[1] = (verticesA[2]-verticesA[1]);
-		uniqueEdgesA[2] = (verticesA[0]-verticesA[2]);
 
+		uniqueEdgesA[0] = (verticesA[1] - verticesA[0]);
+		uniqueEdgesA[1] = (verticesA[2] - verticesA[1]);
+		uniqueEdgesA[2] = (verticesA[0] - verticesA[2]);
 
 		convexPolyhedronA.m_faceOffset = 0;
-                                  
-		b3Float4 normal = b3MakeFloat4(face.m_plane.x,face.m_plane.y,face.m_plane.z,0.f);
-                             
+
+		b3Float4 normal = b3MakeFloat4(face.m_plane.x, face.m_plane.y, face.m_plane.z, 0.f);
+
 		b3GpuFace facesA[B3_TRIANGLE_NUM_CONVEX_FACES];
-		int indicesA[3+3+2+2+2];
-		int curUsedIndices=0;
-		int fidx=0;
+		int indicesA[3 + 3 + 2 + 2 + 2];
+		int curUsedIndices = 0;
+		int fidx = 0;
 
 		//front size of triangle
 		{
-			facesA[fidx].m_indexOffset=curUsedIndices;
+			facesA[fidx].m_indexOffset = curUsedIndices;
 			indicesA[0] = 0;
 			indicesA[1] = 1;
 			indicesA[2] = 2;
-			curUsedIndices+=3;
+			curUsedIndices += 3;
 			float c = face.m_plane.w;
 			facesA[fidx].m_plane.x = normal.x;
 			facesA[fidx].m_plane.y = normal.y;
 			facesA[fidx].m_plane.z = normal.z;
 			facesA[fidx].m_plane.w = c;
-			facesA[fidx].m_numIndices=3;
+			facesA[fidx].m_numIndices = 3;
 		}
 		fidx++;
 		//back size of triangle
 		{
-			facesA[fidx].m_indexOffset=curUsedIndices;
-			indicesA[3]=2;
-			indicesA[4]=1;
-			indicesA[5]=0;
-			curUsedIndices+=3;
-			float c = b3Dot(normal,verticesA[0]);
-		//	float c1 = -face.m_plane.w;
+			facesA[fidx].m_indexOffset = curUsedIndices;
+			indicesA[3] = 2;
+			indicesA[4] = 1;
+			indicesA[5] = 0;
+			curUsedIndices += 3;
+			float c = b3Dot(normal, verticesA[0]);
+			//	float c1 = -face.m_plane.w;
 			facesA[fidx].m_plane.x = -normal.x;
 			facesA[fidx].m_plane.y = -normal.y;
 			facesA[fidx].m_plane.z = -normal.z;
 			facesA[fidx].m_plane.w = c;
-			facesA[fidx].m_numIndices=3;
+			facesA[fidx].m_numIndices = 3;
 		}
 		fidx++;
 
 		bool addEdgePlanes = true;
 		if (addEdgePlanes)
 		{
-			int numVertices=3;
-			int prevVertex = numVertices-1;
-			for (int i=0;i<numVertices;i++)
+			int numVertices = 3;
+			int prevVertex = numVertices - 1;
+			for (int i = 0; i < numVertices; i++)
 			{
 				b3Float4 v0 = verticesA[i];
 				b3Float4 v1 = verticesA[prevVertex];
-                                            
-				b3Float4 edgeNormal = b3Normalized(b3Cross(normal,v1-v0));
-				float c = -b3Dot(edgeNormal,v0);
+
+				b3Float4 edgeNormal = b3Normalized(b3Cross(normal, v1 - v0));
+				float c = -b3Dot(edgeNormal, v0);
 
 				facesA[fidx].m_numIndices = 2;
-				facesA[fidx].m_indexOffset=curUsedIndices;
-				indicesA[curUsedIndices++]=i;
-				indicesA[curUsedIndices++]=prevVertex;
-                                            
+				facesA[fidx].m_indexOffset = curUsedIndices;
+				indicesA[curUsedIndices++] = i;
+				indicesA[curUsedIndices++] = prevVertex;
+
 				facesA[fidx].m_plane.x = edgeNormal.x;
 				facesA[fidx].m_plane.y = edgeNormal.y;
 				facesA[fidx].m_plane.z = edgeNormal.z;
@@ -701,8 +670,7 @@ __kernel void   b3FindConcaveSeparatingAxisKernel( __global b3Int4* concavePairs
 			}
 		}
 		convexPolyhedronA.m_numFaces = B3_TRIANGLE_NUM_CONVEX_FACES;
-		convexPolyhedronA.m_localCenter = localCenter*(1.f/3.f);
-
+		convexPolyhedronA.m_localCenter = localCenter * (1.f / 3.f);
 
 		b3Float4 posA = rigidBodies[bodyIndexA].m_pos;
 		posA.w = 0.f;
@@ -710,23 +678,20 @@ __kernel void   b3FindConcaveSeparatingAxisKernel( __global b3Int4* concavePairs
 		posB.w = 0.f;
 
 		b3Quaternion ornA = rigidBodies[bodyIndexA].m_quat;
-		b3Quaternion ornB =rigidBodies[bodyIndexB].m_quat;
-
-		
-
+		b3Quaternion ornB = rigidBodies[bodyIndexB].m_quat;
 
 		///////////////////
 		///compound shape support
 
-		if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)
+		if (collidables[collidableIndexB].m_shapeType == SHAPE_COMPOUND_OF_CONVEX_HULLS)
 		{
 			int compoundChild = concavePairs[pairIdx].w;
-			int childShapeIndexB = compoundChild;//collidables[collidableIndexB].m_shapeIndex+compoundChild;
+			int childShapeIndexB = compoundChild;  //collidables[collidableIndexB].m_shapeIndex+compoundChild;
 			int childColIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;
 			b3Float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;
 			b3Quaternion childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;
-			b3Float4 newPosB = b3TransformPoint(childPosB,posB,ornB);
-			b3Quaternion newOrnB = b3QuatMul(ornB,childOrnB);
+			b3Float4 newPosB = b3TransformPoint(childPosB, posB, ornB);
+			b3Quaternion newOrnB = b3QuatMul(ornB, childOrnB);
 			posB = newPosB;
 			ornB = newOrnB;
 			shapeIndexB = collidables[childColIndexB].m_shapeIndex;
@@ -736,59 +701,61 @@ __kernel void   b3FindConcaveSeparatingAxisKernel( __global b3Int4* concavePairs
 		b3Float4 c0local = convexPolyhedronA.m_localCenter;
 		b3Float4 c0 = b3TransformPoint(c0local, posA, ornA);
 		b3Float4 c1local = convexShapes[shapeIndexB].m_localCenter;
-		b3Float4 c1 = b3TransformPoint(c1local,posB,ornB);
+		b3Float4 c1 = b3TransformPoint(c1local, posB, ornB);
 		const b3Float4 DeltaC2 = c0 - c1;
 
-
-		bool sepA = b3FindSeparatingAxis(	&convexPolyhedronA, &convexShapes[shapeIndexB],
-												posA,ornA,
-												posB,ornB,
-												DeltaC2,
-												verticesA,uniqueEdgesA,facesA,indicesA,
-												vertices,uniqueEdges,faces,indices,
-												&sepAxis,&dmin);
+		bool sepA = b3FindSeparatingAxis(&convexPolyhedronA, &convexShapes[shapeIndexB],
+										 posA, ornA,
+										 posB, ornB,
+										 DeltaC2,
+										 verticesA, uniqueEdgesA, facesA, indicesA,
+										 vertices, uniqueEdges, faces, indices,
+										 &sepAxis, &dmin);
 		hasSeparatingAxis = 4;
 		if (!sepA)
 		{
 			hasSeparatingAxis = 0;
-		} else
+		}
+		else
 		{
-			bool sepB = b3FindSeparatingAxis(	&convexShapes[shapeIndexB],&convexPolyhedronA,
-												posB,ornB,
-												posA,ornA,
-												DeltaC2,
-												vertices,uniqueEdges,faces,indices,
-												verticesA,uniqueEdgesA,facesA,indicesA,
-												&sepAxis,&dmin);
+			bool sepB = b3FindSeparatingAxis(&convexShapes[shapeIndexB], &convexPolyhedronA,
+											 posB, ornB,
+											 posA, ornA,
+											 DeltaC2,
+											 vertices, uniqueEdges, faces, indices,
+											 verticesA, uniqueEdgesA, facesA, indicesA,
+											 &sepAxis, &dmin);
 
 			if (!sepB)
 			{
 				hasSeparatingAxis = 0;
-			} else
+			}
+			else
 			{
-				bool sepEE = b3FindSeparatingAxisEdgeEdge(	&convexPolyhedronA, &convexShapes[shapeIndexB],
-															posA,ornA,
-															posB,ornB,
-															DeltaC2,
-															verticesA,uniqueEdgesA,facesA,indicesA,
-															vertices,uniqueEdges,faces,indices,
-															&sepAxis,&dmin,true);
-	
+				bool sepEE = b3FindSeparatingAxisEdgeEdge(&convexPolyhedronA, &convexShapes[shapeIndexB],
+														  posA, ornA,
+														  posB, ornB,
+														  DeltaC2,
+														  verticesA, uniqueEdgesA, facesA, indicesA,
+														  vertices, uniqueEdges, faces, indices,
+														  &sepAxis, &dmin, true);
+
 				if (!sepEE)
 				{
 					hasSeparatingAxis = 0;
-				} else
+				}
+				else
 				{
 					hasSeparatingAxis = 1;
 				}
 			}
-		}	
-		
+		}
+
 		if (hasSeparatingAxis)
 		{
-			hasSeparatingNormals[i]=1;
+			hasSeparatingNormals[i] = 1;
 			sepAxis.w = dmin;
-			concaveSeparatingNormalsOut[pairIdx]=sepAxis;
+			concaveSeparatingNormalsOut[pairIdx] = sepAxis;
 
 			//now compute clipping faces A and B, and world-space clipping vertices A and B...
 
@@ -796,37 +763,35 @@ __kernel void   b3FindConcaveSeparatingAxisKernel( __global b3Int4* concavePairs
 			float maxDist = 0.02f;
 
 			b3FindClippingFaces(sepAxis,
-                     &convexPolyhedronA,
-					 &convexShapes[shapeIndexB],
-					 posA,ornA,
-					 posB,ornB,
-                       worldVertsA1Out,
-                      worldNormalsA1Out,
-                      worldVertsB1Out,
-					  vertexFaceCapacity,
-                      minDist, maxDist,
-                      verticesA,
-                      facesA,
-                      indicesA,
- 
-					  vertices,
-                      faces,
-                      indices,
-                      clippingFacesOut, pairIdx);
+								&convexPolyhedronA,
+								&convexShapes[shapeIndexB],
+								posA, ornA,
+								posB, ornB,
+								worldVertsA1Out,
+								worldNormalsA1Out,
+								worldVertsB1Out,
+								vertexFaceCapacity,
+								minDist, maxDist,
+								verticesA,
+								facesA,
+								indicesA,
 
-		} else
-		{	
+								vertices,
+								faces,
+								indices,
+								clippingFacesOut, pairIdx);
+		}
+		else
+		{
 			//mark this pair as in-active
 			concavePairs[pairIdx].w = -1;
 		}
 	}
 	else
-	{	
+	{
 		//mark this pair as in-active
 		concavePairs[pairIdx].w = -1;
 	}
 }
 
-
-#endif //B3_FIND_CONCAVE_SEPARATING_AXIS_H
-
+#endif  //B3_FIND_CONCAVE_SEPARATING_AXIS_H
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3FindSeparatingAxis.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3FindSeparatingAxis.h
index 332dbc278c4..b4981ae6540 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3FindSeparatingAxis.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3FindSeparatingAxis.h
@@ -1,28 +1,27 @@
 #ifndef B3_FIND_SEPARATING_AXIS_H
 #define B3_FIND_SEPARATING_AXIS_H
 
-
-inline void b3ProjectAxis(const b3ConvexPolyhedronData& hull,  const b3Float4& pos, const b3Quaternion& orn, const b3Float4& dir, const b3AlignedObjectArray<b3Vector3>& vertices, b3Scalar& min, b3Scalar& max)
+inline void b3ProjectAxis(const b3ConvexPolyhedronData& hull, const b3Float4& pos, const b3Quaternion& orn, const b3Float4& dir, const b3AlignedObjectArray<b3Vector3>& vertices, b3Scalar& min, b3Scalar& max)
 {
 	min = FLT_MAX;
 	max = -FLT_MAX;
 	int numVerts = hull.m_numVertices;
 
-	const b3Float4 localDir = b3QuatRotate(orn.inverse(),dir);
+	const b3Float4 localDir = b3QuatRotate(orn.inverse(), dir);
 
-	b3Scalar offset = b3Dot3F4(pos,dir);
+	b3Scalar offset = b3Dot3F4(pos, dir);
 
-	for(int i=0;i<numVerts;i++)
+	for (int i = 0; i < numVerts; i++)
 	{
 		//b3Vector3 pt = trans * vertices[m_vertexOffset+i];
 		//b3Scalar dp = pt.dot(dir);
 		//b3Vector3 vertex = vertices[hull.m_vertexOffset+i];
-		b3Scalar dp = b3Dot3F4((b3Float4&)vertices[hull.m_vertexOffset+i],localDir);
+		b3Scalar dp = b3Dot3F4((b3Float4&)vertices[hull.m_vertexOffset + i], localDir);
 		//b3Assert(dp==dpL);
-		if(dp < min)	min = dp;
-		if(dp > max)	max = dp;
+		if (dp < min) min = dp;
+		if (dp > max) max = dp;
 	}
-	if(min>max)
+	if (min > max)
 	{
 		b3Scalar tmp = min;
 		min = max;
@@ -32,44 +31,42 @@ inline void b3ProjectAxis(const b3ConvexPolyhedronData& hull,  const b3Float4& p
 	max += offset;
 }
 
-
-inline bool b3TestSepAxis(const b3ConvexPolyhedronData& hullA, const b3ConvexPolyhedronData& hullB, 
-	const b3Float4& posA,const b3Quaternion& ornA,
-	const b3Float4& posB,const b3Quaternion& ornB,
-	const b3Float4& sep_axis, const b3AlignedObjectArray<b3Vector3>& verticesA,const b3AlignedObjectArray<b3Vector3>& verticesB,b3Scalar& depth)
+inline bool b3TestSepAxis(const b3ConvexPolyhedronData& hullA, const b3ConvexPolyhedronData& hullB,
+						  const b3Float4& posA, const b3Quaternion& ornA,
+						  const b3Float4& posB, const b3Quaternion& ornB,
+						  const b3Float4& sep_axis, const b3AlignedObjectArray<b3Vector3>& verticesA, const b3AlignedObjectArray<b3Vector3>& verticesB, b3Scalar& depth)
 {
-	b3Scalar Min0,Max0;
-	b3Scalar Min1,Max1;
-	b3ProjectAxis(hullA,posA,ornA,sep_axis,verticesA, Min0, Max0);
-	b3ProjectAxis(hullB,posB,ornB, sep_axis,verticesB, Min1, Max1);
+	b3Scalar Min0, Max0;
+	b3Scalar Min1, Max1;
+	b3ProjectAxis(hullA, posA, ornA, sep_axis, verticesA, Min0, Max0);
+	b3ProjectAxis(hullB, posB, ornB, sep_axis, verticesB, Min1, Max1);
 
-	if(Max0<Min1 || Max1<Min0)
+	if (Max0 < Min1 || Max1 < Min0)
 		return false;
 
 	b3Scalar d0 = Max0 - Min1;
-	b3Assert(d0>=0.0f);
+	b3Assert(d0 >= 0.0f);
 	b3Scalar d1 = Max1 - Min0;
-	b3Assert(d1>=0.0f);
-	depth = d0<d1 ? d0:d1;
+	b3Assert(d1 >= 0.0f);
+	depth = d0 < d1 ? d0 : d1;
 	return true;
 }
 
+inline bool b3FindSeparatingAxis(const b3ConvexPolyhedronData& hullA, const b3ConvexPolyhedronData& hullB,
+								 const b3Float4& posA1,
+								 const b3Quaternion& ornA,
+								 const b3Float4& posB1,
+								 const b3Quaternion& ornB,
+								 const b3AlignedObjectArray<b3Vector3>& verticesA,
+								 const b3AlignedObjectArray<b3Vector3>& uniqueEdgesA,
+								 const b3AlignedObjectArray<b3GpuFace>& facesA,
+								 const b3AlignedObjectArray<int>& indicesA,
+								 const b3AlignedObjectArray<b3Vector3>& verticesB,
+								 const b3AlignedObjectArray<b3Vector3>& uniqueEdgesB,
+								 const b3AlignedObjectArray<b3GpuFace>& facesB,
+								 const b3AlignedObjectArray<int>& indicesB,
 
-inline bool b3FindSeparatingAxis(	const b3ConvexPolyhedronData& hullA, const b3ConvexPolyhedronData& hullB, 
-	const b3Float4& posA1,
-	const b3Quaternion& ornA,
-	const b3Float4& posB1,
-	const b3Quaternion& ornB,
-	const b3AlignedObjectArray<b3Vector3>& verticesA,
-	const b3AlignedObjectArray<b3Vector3>& uniqueEdgesA, 
-	const b3AlignedObjectArray<b3GpuFace>& facesA,
-	const b3AlignedObjectArray<int>& indicesA,
-	const b3AlignedObjectArray<b3Vector3>& verticesB, 
-	const b3AlignedObjectArray<b3Vector3>& uniqueEdgesB, 
-	const b3AlignedObjectArray<b3GpuFace>& facesB,
-	const b3AlignedObjectArray<int>& indicesB,
-
-	b3Vector3& sep)
+								 b3Vector3& sep)
 {
 	B3_PROFILE("findSeparatingAxis");
 
@@ -77,42 +74,41 @@ inline bool b3FindSeparatingAxis(	const b3ConvexPolyhedronData& hullA, const b3C
 	posA.w = 0.f;
 	b3Float4 posB = posB1;
 	posB.w = 0.f;
-//#ifdef TEST_INTERNAL_OBJECTS
+	//#ifdef TEST_INTERNAL_OBJECTS
 	b3Float4 c0local = (b3Float4&)hullA.m_localCenter;
 
 	b3Float4 c0 = b3TransformPoint(c0local, posA, ornA);
 	b3Float4 c1local = (b3Float4&)hullB.m_localCenter;
-	b3Float4 c1 = b3TransformPoint(c1local,posB,ornB);
+	b3Float4 c1 = b3TransformPoint(c1local, posB, ornB);
 	const b3Float4 deltaC2 = c0 - c1;
-//#endif
+	//#endif
 
 	b3Scalar dmin = FLT_MAX;
-	int curPlaneTests=0;
+	int curPlaneTests = 0;
 
 	int numFacesA = hullA.m_numFaces;
 	// Test normals from hullA
-	for(int i=0;i<numFacesA;i++)
+	for (int i = 0; i < numFacesA; i++)
 	{
-		const b3Float4& normal = (b3Float4&)facesA[hullA.m_faceOffset+i].m_plane;
-		b3Float4 faceANormalWS = b3QuatRotate(ornA,normal);
+		const b3Float4& normal = (b3Float4&)facesA[hullA.m_faceOffset + i].m_plane;
+		b3Float4 faceANormalWS = b3QuatRotate(ornA, normal);
 
-		if (b3Dot3F4(deltaC2,faceANormalWS)<0)
-			faceANormalWS*=-1.f;
+		if (b3Dot3F4(deltaC2, faceANormalWS) < 0)
+			faceANormalWS *= -1.f;
 
 		curPlaneTests++;
 #ifdef TEST_INTERNAL_OBJECTS
 		gExpectedNbTests++;
-		if(gUseInternalObject && !TestInternalObjects(transA,transB, DeltaC2, faceANormalWS, hullA, hullB, dmin))
+		if (gUseInternalObject && !TestInternalObjects(transA, transB, DeltaC2, faceANormalWS, hullA, hullB, dmin))
 			continue;
 		gActualNbTests++;
 #endif
 
-		
 		b3Scalar d;
-		if(!b3TestSepAxis( hullA, hullB, posA,ornA,posB,ornB,faceANormalWS, verticesA, verticesB,d))
+		if (!b3TestSepAxis(hullA, hullB, posA, ornA, posB, ornB, faceANormalWS, verticesA, verticesB, d))
 			return false;
 
-		if(d<dmin)
+		if (d < dmin)
 		{
 			dmin = d;
 			sep = (b3Vector3&)faceANormalWS;
@@ -121,86 +117,81 @@ inline bool b3FindSeparatingAxis(	const b3ConvexPolyhedronData& hullA, const b3C
 
 	int numFacesB = hullB.m_numFaces;
 	// Test normals from hullB
-	for(int i=0;i<numFacesB;i++)
+	for (int i = 0; i < numFacesB; i++)
 	{
-		b3Float4 normal = (b3Float4&)facesB[hullB.m_faceOffset+i].m_plane;
+		b3Float4 normal = (b3Float4&)facesB[hullB.m_faceOffset + i].m_plane;
 		b3Float4 WorldNormal = b3QuatRotate(ornB, normal);
 
-		if (b3Dot3F4(deltaC2,WorldNormal)<0)
+		if (b3Dot3F4(deltaC2, WorldNormal) < 0)
 		{
-			WorldNormal*=-1.f;
+			WorldNormal *= -1.f;
 		}
 		curPlaneTests++;
 #ifdef TEST_INTERNAL_OBJECTS
 		gExpectedNbTests++;
-		if(gUseInternalObject && !TestInternalObjects(transA,transB,DeltaC2, WorldNormal, hullA, hullB, dmin))
+		if (gUseInternalObject && !TestInternalObjects(transA, transB, DeltaC2, WorldNormal, hullA, hullB, dmin))
 			continue;
 		gActualNbTests++;
 #endif
 
 		b3Scalar d;
-		if(!b3TestSepAxis(hullA, hullB,posA,ornA,posB,ornB,WorldNormal,verticesA,verticesB,d))
+		if (!b3TestSepAxis(hullA, hullB, posA, ornA, posB, ornB, WorldNormal, verticesA, verticesB, d))
 			return false;
 
-		if(d<dmin)
+		if (d < dmin)
 		{
 			dmin = d;
 			sep = (b3Vector3&)WorldNormal;
 		}
 	}
 
-//	b3Vector3 edgeAstart,edgeAend,edgeBstart,edgeBend;
+	//	b3Vector3 edgeAstart,edgeAend,edgeBstart,edgeBend;
 
 	int curEdgeEdge = 0;
 	// Test edges
-	for(int e0=0;e0<hullA.m_numUniqueEdges;e0++)
+	for (int e0 = 0; e0 < hullA.m_numUniqueEdges; e0++)
 	{
-		const b3Float4& edge0 = (b3Float4&) uniqueEdgesA[hullA.m_uniqueEdgesOffset+e0];
-		b3Float4 edge0World = b3QuatRotate(ornA,(b3Float4&)edge0);
+		const b3Float4& edge0 = (b3Float4&)uniqueEdgesA[hullA.m_uniqueEdgesOffset + e0];
+		b3Float4 edge0World = b3QuatRotate(ornA, (b3Float4&)edge0);
 
-		for(int e1=0;e1<hullB.m_numUniqueEdges;e1++)
+		for (int e1 = 0; e1 < hullB.m_numUniqueEdges; e1++)
 		{
-			const b3Vector3 edge1 = uniqueEdgesB[hullB.m_uniqueEdgesOffset+e1];
-			b3Float4 edge1World = b3QuatRotate(ornB,(b3Float4&)edge1);
+			const b3Vector3 edge1 = uniqueEdgesB[hullB.m_uniqueEdgesOffset + e1];
+			b3Float4 edge1World = b3QuatRotate(ornB, (b3Float4&)edge1);
 
-
-			b3Float4 crossje = b3Cross3(edge0World,edge1World);
+			b3Float4 crossje = b3Cross3(edge0World, edge1World);
 
 			curEdgeEdge++;
-			if(!b3IsAlmostZero((b3Vector3&)crossje))
+			if (!b3IsAlmostZero((b3Vector3&)crossje))
 			{
 				crossje = b3FastNormalized3(crossje);
-				if (b3Dot3F4(deltaC2,crossje)<0)
-					crossje*=-1.f;
-
+				if (b3Dot3F4(deltaC2, crossje) < 0)
+					crossje *= -1.f;
 
 #ifdef TEST_INTERNAL_OBJECTS
 				gExpectedNbTests++;
-				if(gUseInternalObject && !TestInternalObjects(transA,transB,DeltaC2, Cross, hullA, hullB, dmin))
+				if (gUseInternalObject && !TestInternalObjects(transA, transB, DeltaC2, Cross, hullA, hullB, dmin))
 					continue;
 				gActualNbTests++;
 #endif
 
 				b3Scalar dist;
-				if(!b3TestSepAxis( hullA, hullB, posA,ornA,posB,ornB,crossje, verticesA,verticesB,dist))
+				if (!b3TestSepAxis(hullA, hullB, posA, ornA, posB, ornB, crossje, verticesA, verticesB, dist))
 					return false;
 
-				if(dist<dmin)
+				if (dist < dmin)
 				{
 					dmin = dist;
 					sep = (b3Vector3&)crossje;
 				}
 			}
 		}
-
 	}
 
-	
-	if((b3Dot3F4(-deltaC2,(b3Float4&)sep))>0.0f)
+	if ((b3Dot3F4(-deltaC2, (b3Float4&)sep)) > 0.0f)
 		sep = -sep;
 
 	return true;
 }
 
-#endif //B3_FIND_SEPARATING_AXIS_H
-
+#endif  //B3_FIND_SEPARATING_AXIS_H
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3MprPenetration.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3MprPenetration.h
index 6c3ad7c9dd9..a3bfbf2995a 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3MprPenetration.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3MprPenetration.h
@@ -15,9 +15,6 @@
  *  See the License for more information.
  */
 
-
-
-
 #ifndef B3_MPR_PENETRATION_H
 #define B3_MPR_PENETRATION_H
 
@@ -27,9 +24,6 @@
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3ConvexPolyhedronData.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3Collidable.h"
 
-
-
-
 #ifdef __cplusplus
 #define B3_MPR_SQRT sqrtf
 #else
@@ -41,196 +35,171 @@
 #define B3_MPR_TOLERANCE 1E-6f
 #define B3_MPR_MAX_ITERATIONS 1000
 
-struct _b3MprSupport_t 
+struct _b3MprSupport_t
 {
-    b3Float4 v;  //!< Support point in minkowski sum
-    b3Float4 v1; //!< Support point in obj1
-    b3Float4 v2; //!< Support point in obj2
+	b3Float4 v;   //!< Support point in minkowski sum
+	b3Float4 v1;  //!< Support point in obj1
+	b3Float4 v2;  //!< Support point in obj2
 };
 typedef struct _b3MprSupport_t b3MprSupport_t;
 
-struct _b3MprSimplex_t 
+struct _b3MprSimplex_t
 {
-    b3MprSupport_t ps[4];
-    int last; //!< index of last added point
+	b3MprSupport_t ps[4];
+	int last;  //!< index of last added point
 };
 typedef struct _b3MprSimplex_t b3MprSimplex_t;
 
-inline b3MprSupport_t* b3MprSimplexPointW(b3MprSimplex_t *s, int idx)
+inline b3MprSupport_t *b3MprSimplexPointW(b3MprSimplex_t *s, int idx)
 {
-    return &s->ps[idx];
+	return &s->ps[idx];
 }
 
 inline void b3MprSimplexSetSize(b3MprSimplex_t *s, int size)
 {
-    s->last = size - 1;
+	s->last = size - 1;
 }
 
-
 inline int b3MprSimplexSize(const b3MprSimplex_t *s)
 {
-    return s->last + 1;
+	return s->last + 1;
 }
 
-
-inline const b3MprSupport_t* b3MprSimplexPoint(const b3MprSimplex_t* s, int idx)
+inline const b3MprSupport_t *b3MprSimplexPoint(const b3MprSimplex_t *s, int idx)
 {
-    // here is no check on boundaries
-    return &s->ps[idx];
+	// here is no check on boundaries
+	return &s->ps[idx];
 }
 
 inline void b3MprSupportCopy(b3MprSupport_t *d, const b3MprSupport_t *s)
 {
-    *d = *s;
+	*d = *s;
 }
 
 inline void b3MprSimplexSet(b3MprSimplex_t *s, size_t pos, const b3MprSupport_t *a)
 {
-    b3MprSupportCopy(s->ps + pos, a);
+	b3MprSupportCopy(s->ps + pos, a);
 }
 
-
 inline void b3MprSimplexSwap(b3MprSimplex_t *s, size_t pos1, size_t pos2)
 {
-    b3MprSupport_t supp;
+	b3MprSupport_t supp;
 
-    b3MprSupportCopy(&supp, &s->ps[pos1]);
-    b3MprSupportCopy(&s->ps[pos1], &s->ps[pos2]);
-    b3MprSupportCopy(&s->ps[pos2], &supp);
+	b3MprSupportCopy(&supp, &s->ps[pos1]);
+	b3MprSupportCopy(&s->ps[pos1], &s->ps[pos2]);
+	b3MprSupportCopy(&s->ps[pos2], &supp);
 }
 
-
 inline int b3MprIsZero(float val)
 {
-    return B3_MPR_FABS(val) < FLT_EPSILON;
+	return B3_MPR_FABS(val) < FLT_EPSILON;
 }
 
-
-
 inline int b3MprEq(float _a, float _b)
 {
-    float ab;
-    float a, b;
+	float ab;
+	float a, b;
 
-    ab = B3_MPR_FABS(_a - _b);
-    if (B3_MPR_FABS(ab) < FLT_EPSILON)
-        return 1;
+	ab = B3_MPR_FABS(_a - _b);
+	if (B3_MPR_FABS(ab) < FLT_EPSILON)
+		return 1;
 
-    a = B3_MPR_FABS(_a);
-    b = B3_MPR_FABS(_b);
-    if (b > a){
-        return ab < FLT_EPSILON * b;
-    }else{
-        return ab < FLT_EPSILON * a;
-    }
+	a = B3_MPR_FABS(_a);
+	b = B3_MPR_FABS(_b);
+	if (b > a)
+	{
+		return ab < FLT_EPSILON * b;
+	}
+	else
+	{
+		return ab < FLT_EPSILON * a;
+	}
 }
 
-
-inline int b3MprVec3Eq(const b3Float4* a, const b3Float4 *b)
+inline int b3MprVec3Eq(const b3Float4 *a, const b3Float4 *b)
 {
-    return b3MprEq((*a).x, (*b).x)
-            && b3MprEq((*a).y, (*b).y)
-            && b3MprEq((*a).z, (*b).z);
+	return b3MprEq((*a).x, (*b).x) && b3MprEq((*a).y, (*b).y) && b3MprEq((*a).z, (*b).z);
 }
 
-
-
-inline b3Float4 b3LocalGetSupportVertex(b3Float4ConstArg supportVec,__global const b3ConvexPolyhedronData_t* hull, 	b3ConstArray(b3Float4) verticesA)
+inline b3Float4 b3LocalGetSupportVertex(b3Float4ConstArg supportVec, __global const b3ConvexPolyhedronData_t *hull, b3ConstArray(b3Float4) verticesA)
 {
-	b3Float4 supVec = b3MakeFloat4(0,0,0,0);
+	b3Float4 supVec = b3MakeFloat4(0, 0, 0, 0);
 	float maxDot = -B3_LARGE_FLOAT;
 
-    if( 0 < hull->m_numVertices )
-    {
-        const b3Float4 scaled = supportVec;
+	if (0 < hull->m_numVertices)
+	{
+		const b3Float4 scaled = supportVec;
 		int index = b3MaxDot(scaled, &verticesA[hull->m_vertexOffset], hull->m_numVertices, &maxDot);
-        return verticesA[hull->m_vertexOffset+index];
-    }
-
-    return supVec;
+		return verticesA[hull->m_vertexOffset + index];
+	}
 
+	return supVec;
 }
 
-
-B3_STATIC void b3MprConvexSupport(int pairIndex,int bodyIndex,  b3ConstArray(b3RigidBodyData_t) cpuBodyBuf, 
-													b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData, 
-													b3ConstArray(b3Collidable_t)				cpuCollidables,
-													b3ConstArray(b3Float4)					cpuVertices,
-													__global b3Float4* sepAxis,
-														const b3Float4* _dir, b3Float4* outp, int logme)
+B3_STATIC void b3MprConvexSupport(int pairIndex, int bodyIndex, b3ConstArray(b3RigidBodyData_t) cpuBodyBuf,
+								  b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData,
+								  b3ConstArray(b3Collidable_t) cpuCollidables,
+								  b3ConstArray(b3Float4) cpuVertices,
+								  __global b3Float4 *sepAxis,
+								  const b3Float4 *_dir, b3Float4 *outp, int logme)
 {
 	//dir is in worldspace, move to local space
-	
+
 	b3Float4 pos = cpuBodyBuf[bodyIndex].m_pos;
 	b3Quat orn = cpuBodyBuf[bodyIndex].m_quat;
-	
-	b3Float4 dir = b3MakeFloat4((*_dir).x,(*_dir).y,(*_dir).z,0.f);
-	
-	const b3Float4 localDir = b3QuatRotate(b3QuatInverse(orn),dir);
-	
+
+	b3Float4 dir = b3MakeFloat4((*_dir).x, (*_dir).y, (*_dir).z, 0.f);
+
+	const b3Float4 localDir = b3QuatRotate(b3QuatInverse(orn), dir);
 
 	//find local support vertex
 	int colIndex = cpuBodyBuf[bodyIndex].m_collidableIdx;
-	
-	b3Assert(cpuCollidables[colIndex].m_shapeType==SHAPE_CONVEX_HULL);
-	__global const b3ConvexPolyhedronData_t* hull = &cpuConvexData[cpuCollidables[colIndex].m_shapeIndex];
-	
+
+	b3Assert(cpuCollidables[colIndex].m_shapeType == SHAPE_CONVEX_HULL);
+	__global const b3ConvexPolyhedronData_t *hull = &cpuConvexData[cpuCollidables[colIndex].m_shapeIndex];
+
 	b3Float4 pInA;
 	if (logme)
 	{
-
-
-	//	b3Float4 supVec = b3MakeFloat4(0,0,0,0);
+		//	b3Float4 supVec = b3MakeFloat4(0,0,0,0);
 		float maxDot = -B3_LARGE_FLOAT;
 
-		if( 0 < hull->m_numVertices )
+		if (0 < hull->m_numVertices)
 		{
 			const b3Float4 scaled = localDir;
 			int index = b3MaxDot(scaled, &cpuVertices[hull->m_vertexOffset], hull->m_numVertices, &maxDot);
-			pInA = cpuVertices[hull->m_vertexOffset+index];
-			
+			pInA = cpuVertices[hull->m_vertexOffset + index];
 		}
-
-
-	} else
+	}
+	else
 	{
-		pInA = b3LocalGetSupportVertex(localDir,hull,cpuVertices);
+		pInA = b3LocalGetSupportVertex(localDir, hull, cpuVertices);
 	}
 
 	//move vertex to world space
-	*outp = b3TransformPoint(pInA,pos,orn);
-	
+	*outp = b3TransformPoint(pInA, pos, orn);
 }
 
-inline void b3MprSupport(int pairIndex,int bodyIndexA, int bodyIndexB,   b3ConstArray(b3RigidBodyData_t) cpuBodyBuf, 
-													b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData, 
-													b3ConstArray(b3Collidable_t)				cpuCollidables,
-													b3ConstArray(b3Float4)					cpuVertices,
-													__global b3Float4* sepAxis,
-													const b3Float4* _dir, b3MprSupport_t *supp)
+inline void b3MprSupport(int pairIndex, int bodyIndexA, int bodyIndexB, b3ConstArray(b3RigidBodyData_t) cpuBodyBuf,
+						 b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData,
+						 b3ConstArray(b3Collidable_t) cpuCollidables,
+						 b3ConstArray(b3Float4) cpuVertices,
+						 __global b3Float4 *sepAxis,
+						 const b3Float4 *_dir, b3MprSupport_t *supp)
 {
-    b3Float4 dir;
+	b3Float4 dir;
 	dir = *_dir;
-	b3MprConvexSupport(pairIndex,bodyIndexA,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices,sepAxis,&dir, &supp->v1,0);
-    dir = *_dir*-1.f;
-	b3MprConvexSupport(pairIndex,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices,sepAxis,&dir, &supp->v2,0);
-    supp->v = supp->v1 - supp->v2;
+	b3MprConvexSupport(pairIndex, bodyIndexA, cpuBodyBuf, cpuConvexData, cpuCollidables, cpuVertices, sepAxis, &dir, &supp->v1, 0);
+	dir = *_dir * -1.f;
+	b3MprConvexSupport(pairIndex, bodyIndexB, cpuBodyBuf, cpuConvexData, cpuCollidables, cpuVertices, sepAxis, &dir, &supp->v2, 0);
+	supp->v = supp->v1 - supp->v2;
 }
 
-
-
-
-
-
-
-
-
 inline void b3FindOrigin(int bodyIndexA, int bodyIndexB, b3ConstArray(b3RigidBodyData_t) cpuBodyBuf, b3MprSupport_t *center)
 {
-
-    center->v1 = cpuBodyBuf[bodyIndexA].m_pos;
+	center->v1 = cpuBodyBuf[bodyIndexA].m_pos;
 	center->v2 = cpuBodyBuf[bodyIndexB].m_pos;
-    center->v = center->v1 - center->v2;
+	center->v = center->v1 - center->v2;
 }
 
 inline void b3MprVec3Set(b3Float4 *v, float x, float y, float z)
@@ -243,48 +212,45 @@ inline void b3MprVec3Set(b3Float4 *v, float x, float y, float z)
 
 inline void b3MprVec3Add(b3Float4 *v, const b3Float4 *w)
 {
-    (*v).x += (*w).x;
-    (*v).y += (*w).y;
-    (*v).z += (*w).z;
+	(*v).x += (*w).x;
+	(*v).y += (*w).y;
+	(*v).z += (*w).z;
 }
 
 inline void b3MprVec3Copy(b3Float4 *v, const b3Float4 *w)
 {
-    *v = *w;
+	*v = *w;
 }
 
 inline void b3MprVec3Scale(b3Float4 *d, float k)
 {
-    *d *= k;
+	*d *= k;
 }
 
 inline float b3MprVec3Dot(const b3Float4 *a, const b3Float4 *b)
 {
-    float dot;
+	float dot;
 
-	dot = b3Dot3F4(*a,*b);
-    return dot;
+	dot = b3Dot3F4(*a, *b);
+	return dot;
 }
 
-
 inline float b3MprVec3Len2(const b3Float4 *v)
 {
-    return b3MprVec3Dot(v, v);
+	return b3MprVec3Dot(v, v);
 }
 
 inline void b3MprVec3Normalize(b3Float4 *d)
 {
-    float k = 1.f / B3_MPR_SQRT(b3MprVec3Len2(d));
-    b3MprVec3Scale(d, k);
+	float k = 1.f / B3_MPR_SQRT(b3MprVec3Len2(d));
+	b3MprVec3Scale(d, k);
 }
 
 inline void b3MprVec3Cross(b3Float4 *d, const b3Float4 *a, const b3Float4 *b)
 {
-	*d = b3Cross3(*a,*b);
-	
+	*d = b3Cross3(*a, *b);
 }
 
-
 inline void b3MprVec3Sub2(b3Float4 *d, const b3Float4 *v, const b3Float4 *w)
 {
 	*d = *v - *w;
@@ -292,629 +258,631 @@ inline void b3MprVec3Sub2(b3Float4 *d, const b3Float4 *v, const b3Float4 *w)
 
 inline void b3PortalDir(const b3MprSimplex_t *portal, b3Float4 *dir)
 {
-    b3Float4 v2v1, v3v1;
+	b3Float4 v2v1, v3v1;
 
-    b3MprVec3Sub2(&v2v1, &b3MprSimplexPoint(portal, 2)->v,
-                       &b3MprSimplexPoint(portal, 1)->v);
-    b3MprVec3Sub2(&v3v1, &b3MprSimplexPoint(portal, 3)->v,
-                       &b3MprSimplexPoint(portal, 1)->v);
-    b3MprVec3Cross(dir, &v2v1, &v3v1);
-    b3MprVec3Normalize(dir);
+	b3MprVec3Sub2(&v2v1, &b3MprSimplexPoint(portal, 2)->v,
+				  &b3MprSimplexPoint(portal, 1)->v);
+	b3MprVec3Sub2(&v3v1, &b3MprSimplexPoint(portal, 3)->v,
+				  &b3MprSimplexPoint(portal, 1)->v);
+	b3MprVec3Cross(dir, &v2v1, &v3v1);
+	b3MprVec3Normalize(dir);
 }
 
-
 inline int portalEncapsulesOrigin(const b3MprSimplex_t *portal,
-                                       const b3Float4 *dir)
+								  const b3Float4 *dir)
 {
-    float dot;
-    dot = b3MprVec3Dot(dir, &b3MprSimplexPoint(portal, 1)->v);
-    return b3MprIsZero(dot) || dot > 0.f;
+	float dot;
+	dot = b3MprVec3Dot(dir, &b3MprSimplexPoint(portal, 1)->v);
+	return b3MprIsZero(dot) || dot > 0.f;
 }
 
 inline int portalReachTolerance(const b3MprSimplex_t *portal,
-                                     const b3MprSupport_t *v4,
-                                     const b3Float4 *dir)
+								const b3MprSupport_t *v4,
+								const b3Float4 *dir)
 {
-    float dv1, dv2, dv3, dv4;
-    float dot1, dot2, dot3;
+	float dv1, dv2, dv3, dv4;
+	float dot1, dot2, dot3;
 
-    // find the smallest dot product of dir and {v1-v4, v2-v4, v3-v4}
+	// find the smallest dot product of dir and {v1-v4, v2-v4, v3-v4}
 
-    dv1 = b3MprVec3Dot(&b3MprSimplexPoint(portal, 1)->v, dir);
-    dv2 = b3MprVec3Dot(&b3MprSimplexPoint(portal, 2)->v, dir);
-    dv3 = b3MprVec3Dot(&b3MprSimplexPoint(portal, 3)->v, dir);
-    dv4 = b3MprVec3Dot(&v4->v, dir);
+	dv1 = b3MprVec3Dot(&b3MprSimplexPoint(portal, 1)->v, dir);
+	dv2 = b3MprVec3Dot(&b3MprSimplexPoint(portal, 2)->v, dir);
+	dv3 = b3MprVec3Dot(&b3MprSimplexPoint(portal, 3)->v, dir);
+	dv4 = b3MprVec3Dot(&v4->v, dir);
 
-    dot1 = dv4 - dv1;
-    dot2 = dv4 - dv2;
-    dot3 = dv4 - dv3;
+	dot1 = dv4 - dv1;
+	dot2 = dv4 - dv2;
+	dot3 = dv4 - dv3;
 
-    dot1 = B3_MPR_FMIN(dot1, dot2);
-    dot1 = B3_MPR_FMIN(dot1, dot3);
+	dot1 = B3_MPR_FMIN(dot1, dot2);
+	dot1 = B3_MPR_FMIN(dot1, dot3);
 
-    return b3MprEq(dot1, B3_MPR_TOLERANCE) || dot1 < B3_MPR_TOLERANCE;
+	return b3MprEq(dot1, B3_MPR_TOLERANCE) || dot1 < B3_MPR_TOLERANCE;
 }
 
-inline int portalCanEncapsuleOrigin(const b3MprSimplex_t *portal,   
-                                         const b3MprSupport_t *v4,
-                                         const b3Float4 *dir)
+inline int portalCanEncapsuleOrigin(const b3MprSimplex_t *portal,
+									const b3MprSupport_t *v4,
+									const b3Float4 *dir)
 {
-    float dot;
-    dot = b3MprVec3Dot(&v4->v, dir);
-    return b3MprIsZero(dot) || dot > 0.f;
+	float dot;
+	dot = b3MprVec3Dot(&v4->v, dir);
+	return b3MprIsZero(dot) || dot > 0.f;
 }
 
 inline void b3ExpandPortal(b3MprSimplex_t *portal,
-                              const b3MprSupport_t *v4)
+						   const b3MprSupport_t *v4)
 {
-    float dot;
-    b3Float4 v4v0;
+	float dot;
+	b3Float4 v4v0;
 
-    b3MprVec3Cross(&v4v0, &v4->v, &b3MprSimplexPoint(portal, 0)->v);
-    dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 1)->v, &v4v0);
-    if (dot > 0.f){
-        dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 2)->v, &v4v0);
-        if (dot > 0.f){
-            b3MprSimplexSet(portal, 1, v4);
-        }else{
-            b3MprSimplexSet(portal, 3, v4);
-        }
-    }else{
-        dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 3)->v, &v4v0);
-        if (dot > 0.f){
-            b3MprSimplexSet(portal, 2, v4);
-        }else{
-            b3MprSimplexSet(portal, 1, v4);
-        }
-    }
-}
-
-
-
-B3_STATIC int b3DiscoverPortal(int pairIndex, int bodyIndexA, int bodyIndexB,  b3ConstArray(b3RigidBodyData_t) cpuBodyBuf, 
-													b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData, 
-													b3ConstArray(b3Collidable_t)				cpuCollidables,
-													b3ConstArray(b3Float4)					cpuVertices,
-													__global b3Float4* sepAxis,
-													__global int*	hasSepAxis,
-													b3MprSimplex_t *portal)
-{
-    b3Float4 dir, va, vb;
-    float dot;
-    int cont;
-	
-	
-
-    // vertex 0 is center of portal
-    b3FindOrigin(bodyIndexA,bodyIndexB,cpuBodyBuf, b3MprSimplexPointW(portal, 0));
-    // vertex 0 is center of portal
-    b3MprSimplexSetSize(portal, 1);
-	
-
-
-	b3Float4 zero = b3MakeFloat4(0,0,0,0);
-	b3Float4* b3mpr_vec3_origin = &zero;
-
-    if (b3MprVec3Eq(&b3MprSimplexPoint(portal, 0)->v, b3mpr_vec3_origin)){
-        // Portal's center lies on origin (0,0,0) => we know that objects
-        // intersect but we would need to know penetration info.
-        // So move center little bit...
-        b3MprVec3Set(&va, FLT_EPSILON * 10.f, 0.f, 0.f);
-        b3MprVec3Add(&b3MprSimplexPointW(portal, 0)->v, &va);
-    }
-
-
-    // vertex 1 = support in direction of origin
-    b3MprVec3Copy(&dir, &b3MprSimplexPoint(portal, 0)->v);
-    b3MprVec3Scale(&dir, -1.f);
-    b3MprVec3Normalize(&dir);
-
-
-    b3MprSupport(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&dir, b3MprSimplexPointW(portal, 1));
-
-    b3MprSimplexSetSize(portal, 2);
-
-    // test if origin isn't outside of v1
-    dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 1)->v, &dir);
-	
-
-    if (b3MprIsZero(dot) || dot < 0.f)
-        return -1;
-
-
-    // vertex 2
-    b3MprVec3Cross(&dir, &b3MprSimplexPoint(portal, 0)->v,
-                       &b3MprSimplexPoint(portal, 1)->v);
-    if (b3MprIsZero(b3MprVec3Len2(&dir))){
-        if (b3MprVec3Eq(&b3MprSimplexPoint(portal, 1)->v, b3mpr_vec3_origin)){
-            // origin lies on v1
-            return 1;
-        }else{
-            // origin lies on v0-v1 segment
-            return 2;
-        }
-    }
-
-    b3MprVec3Normalize(&dir);
-	 b3MprSupport(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&dir, b3MprSimplexPointW(portal, 2));
-    
-    dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 2)->v, &dir);
-    if (b3MprIsZero(dot) || dot < 0.f)
-        return -1;
-
-    b3MprSimplexSetSize(portal, 3);
-
-    // vertex 3 direction
-    b3MprVec3Sub2(&va, &b3MprSimplexPoint(portal, 1)->v,
-                     &b3MprSimplexPoint(portal, 0)->v);
-    b3MprVec3Sub2(&vb, &b3MprSimplexPoint(portal, 2)->v,
-                     &b3MprSimplexPoint(portal, 0)->v);
-    b3MprVec3Cross(&dir, &va, &vb);
-    b3MprVec3Normalize(&dir);
-
-    // it is better to form portal faces to be oriented "outside" origin
-    dot = b3MprVec3Dot(&dir, &b3MprSimplexPoint(portal, 0)->v);
-    if (dot > 0.f){
-        b3MprSimplexSwap(portal, 1, 2);
-        b3MprVec3Scale(&dir, -1.f);
-    }
-
-    while (b3MprSimplexSize(portal) < 4){
-		 b3MprSupport(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&dir, b3MprSimplexPointW(portal, 3));
-        
-        dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 3)->v, &dir);
-        if (b3MprIsZero(dot) || dot < 0.f)
-            return -1;
-
-        cont = 0;
-
-        // test if origin is outside (v1, v0, v3) - set v2 as v3 and
-        // continue
-        b3MprVec3Cross(&va, &b3MprSimplexPoint(portal, 1)->v,
-                          &b3MprSimplexPoint(portal, 3)->v);
-        dot = b3MprVec3Dot(&va, &b3MprSimplexPoint(portal, 0)->v);
-        if (dot < 0.f && !b3MprIsZero(dot)){
-            b3MprSimplexSet(portal, 2, b3MprSimplexPoint(portal, 3));
-            cont = 1;
-        }
-
-        if (!cont){
-            // test if origin is outside (v3, v0, v2) - set v1 as v3 and
-            // continue
-            b3MprVec3Cross(&va, &b3MprSimplexPoint(portal, 3)->v,
-                              &b3MprSimplexPoint(portal, 2)->v);
-            dot = b3MprVec3Dot(&va, &b3MprSimplexPoint(portal, 0)->v);
-            if (dot < 0.f && !b3MprIsZero(dot)){
-                b3MprSimplexSet(portal, 1, b3MprSimplexPoint(portal, 3));
-                cont = 1;
-            }
-        }
-
-        if (cont){
-            b3MprVec3Sub2(&va, &b3MprSimplexPoint(portal, 1)->v,
-                             &b3MprSimplexPoint(portal, 0)->v);
-            b3MprVec3Sub2(&vb, &b3MprSimplexPoint(portal, 2)->v,
-                             &b3MprSimplexPoint(portal, 0)->v);
-            b3MprVec3Cross(&dir, &va, &vb);
-            b3MprVec3Normalize(&dir);
-        }else{
-            b3MprSimplexSetSize(portal, 4);
-        }
-    }
-
-    return 0;
-}
-
-
-B3_STATIC int b3RefinePortal(int pairIndex,int bodyIndexA, int bodyIndexB,  b3ConstArray(b3RigidBodyData_t) cpuBodyBuf, 
-													b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData, 
-													b3ConstArray(b3Collidable_t)				cpuCollidables,
-													b3ConstArray(b3Float4)					cpuVertices,
-													__global b3Float4* sepAxis,
-													b3MprSimplex_t *portal)
-{
-    b3Float4 dir;
-    b3MprSupport_t v4;
-
-	for (int i=0;i<B3_MPR_MAX_ITERATIONS;i++)
-    //while (1)
+	b3MprVec3Cross(&v4v0, &v4->v, &b3MprSimplexPoint(portal, 0)->v);
+	dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 1)->v, &v4v0);
+	if (dot > 0.f)
 	{
-        // compute direction outside the portal (from v0 throught v1,v2,v3
-        // face)
-        b3PortalDir(portal, &dir);
-
-        // test if origin is inside the portal
-        if (portalEncapsulesOrigin(portal, &dir))
-            return 0;
-
-        // get next support point
-        
-		 b3MprSupport(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&dir, &v4);
-
-
-        // test if v4 can expand portal to contain origin and if portal
-        // expanding doesn't reach given tolerance
-        if (!portalCanEncapsuleOrigin(portal, &v4, &dir)
-                || portalReachTolerance(portal, &v4, &dir))
+		dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 2)->v, &v4v0);
+		if (dot > 0.f)
 		{
-            return -1;
-        }
+			b3MprSimplexSet(portal, 1, v4);
+		}
+		else
+		{
+			b3MprSimplexSet(portal, 3, v4);
+		}
+	}
+	else
+	{
+		dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 3)->v, &v4v0);
+		if (dot > 0.f)
+		{
+			b3MprSimplexSet(portal, 2, v4);
+		}
+		else
+		{
+			b3MprSimplexSet(portal, 1, v4);
+		}
+	}
+}
 
-        // v1-v2-v3 triangle must be rearranged to face outside Minkowski
-        // difference (direction from v0).
-        b3ExpandPortal(portal, &v4);
-    }
+B3_STATIC int b3DiscoverPortal(int pairIndex, int bodyIndexA, int bodyIndexB, b3ConstArray(b3RigidBodyData_t) cpuBodyBuf,
+							   b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData,
+							   b3ConstArray(b3Collidable_t) cpuCollidables,
+							   b3ConstArray(b3Float4) cpuVertices,
+							   __global b3Float4 *sepAxis,
+							   __global int *hasSepAxis,
+							   b3MprSimplex_t *portal)
+{
+	b3Float4 dir, va, vb;
+	float dot;
+	int cont;
 
-    return -1;
+	// vertex 0 is center of portal
+	b3FindOrigin(bodyIndexA, bodyIndexB, cpuBodyBuf, b3MprSimplexPointW(portal, 0));
+	// vertex 0 is center of portal
+	b3MprSimplexSetSize(portal, 1);
+
+	b3Float4 zero = b3MakeFloat4(0, 0, 0, 0);
+	b3Float4 *b3mpr_vec3_origin = &zero;
+
+	if (b3MprVec3Eq(&b3MprSimplexPoint(portal, 0)->v, b3mpr_vec3_origin))
+	{
+		// Portal's center lies on origin (0,0,0) => we know that objects
+		// intersect but we would need to know penetration info.
+		// So move center little bit...
+		b3MprVec3Set(&va, FLT_EPSILON * 10.f, 0.f, 0.f);
+		b3MprVec3Add(&b3MprSimplexPointW(portal, 0)->v, &va);
+	}
+
+	// vertex 1 = support in direction of origin
+	b3MprVec3Copy(&dir, &b3MprSimplexPoint(portal, 0)->v);
+	b3MprVec3Scale(&dir, -1.f);
+	b3MprVec3Normalize(&dir);
+
+	b3MprSupport(pairIndex, bodyIndexA, bodyIndexB, cpuBodyBuf, cpuConvexData, cpuCollidables, cpuVertices, sepAxis, &dir, b3MprSimplexPointW(portal, 1));
+
+	b3MprSimplexSetSize(portal, 2);
+
+	// test if origin isn't outside of v1
+	dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 1)->v, &dir);
+
+	if (b3MprIsZero(dot) || dot < 0.f)
+		return -1;
+
+	// vertex 2
+	b3MprVec3Cross(&dir, &b3MprSimplexPoint(portal, 0)->v,
+				   &b3MprSimplexPoint(portal, 1)->v);
+	if (b3MprIsZero(b3MprVec3Len2(&dir)))
+	{
+		if (b3MprVec3Eq(&b3MprSimplexPoint(portal, 1)->v, b3mpr_vec3_origin))
+		{
+			// origin lies on v1
+			return 1;
+		}
+		else
+		{
+			// origin lies on v0-v1 segment
+			return 2;
+		}
+	}
+
+	b3MprVec3Normalize(&dir);
+	b3MprSupport(pairIndex, bodyIndexA, bodyIndexB, cpuBodyBuf, cpuConvexData, cpuCollidables, cpuVertices, sepAxis, &dir, b3MprSimplexPointW(portal, 2));
+
+	dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 2)->v, &dir);
+	if (b3MprIsZero(dot) || dot < 0.f)
+		return -1;
+
+	b3MprSimplexSetSize(portal, 3);
+
+	// vertex 3 direction
+	b3MprVec3Sub2(&va, &b3MprSimplexPoint(portal, 1)->v,
+				  &b3MprSimplexPoint(portal, 0)->v);
+	b3MprVec3Sub2(&vb, &b3MprSimplexPoint(portal, 2)->v,
+				  &b3MprSimplexPoint(portal, 0)->v);
+	b3MprVec3Cross(&dir, &va, &vb);
+	b3MprVec3Normalize(&dir);
+
+	// it is better to form portal faces to be oriented "outside" origin
+	dot = b3MprVec3Dot(&dir, &b3MprSimplexPoint(portal, 0)->v);
+	if (dot > 0.f)
+	{
+		b3MprSimplexSwap(portal, 1, 2);
+		b3MprVec3Scale(&dir, -1.f);
+	}
+
+	while (b3MprSimplexSize(portal) < 4)
+	{
+		b3MprSupport(pairIndex, bodyIndexA, bodyIndexB, cpuBodyBuf, cpuConvexData, cpuCollidables, cpuVertices, sepAxis, &dir, b3MprSimplexPointW(portal, 3));
+
+		dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 3)->v, &dir);
+		if (b3MprIsZero(dot) || dot < 0.f)
+			return -1;
+
+		cont = 0;
+
+		// test if origin is outside (v1, v0, v3) - set v2 as v3 and
+		// continue
+		b3MprVec3Cross(&va, &b3MprSimplexPoint(portal, 1)->v,
+					   &b3MprSimplexPoint(portal, 3)->v);
+		dot = b3MprVec3Dot(&va, &b3MprSimplexPoint(portal, 0)->v);
+		if (dot < 0.f && !b3MprIsZero(dot))
+		{
+			b3MprSimplexSet(portal, 2, b3MprSimplexPoint(portal, 3));
+			cont = 1;
+		}
+
+		if (!cont)
+		{
+			// test if origin is outside (v3, v0, v2) - set v1 as v3 and
+			// continue
+			b3MprVec3Cross(&va, &b3MprSimplexPoint(portal, 3)->v,
+						   &b3MprSimplexPoint(portal, 2)->v);
+			dot = b3MprVec3Dot(&va, &b3MprSimplexPoint(portal, 0)->v);
+			if (dot < 0.f && !b3MprIsZero(dot))
+			{
+				b3MprSimplexSet(portal, 1, b3MprSimplexPoint(portal, 3));
+				cont = 1;
+			}
+		}
+
+		if (cont)
+		{
+			b3MprVec3Sub2(&va, &b3MprSimplexPoint(portal, 1)->v,
+						  &b3MprSimplexPoint(portal, 0)->v);
+			b3MprVec3Sub2(&vb, &b3MprSimplexPoint(portal, 2)->v,
+						  &b3MprSimplexPoint(portal, 0)->v);
+			b3MprVec3Cross(&dir, &va, &vb);
+			b3MprVec3Normalize(&dir);
+		}
+		else
+		{
+			b3MprSimplexSetSize(portal, 4);
+		}
+	}
+
+	return 0;
+}
+
+B3_STATIC int b3RefinePortal(int pairIndex, int bodyIndexA, int bodyIndexB, b3ConstArray(b3RigidBodyData_t) cpuBodyBuf,
+							 b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData,
+							 b3ConstArray(b3Collidable_t) cpuCollidables,
+							 b3ConstArray(b3Float4) cpuVertices,
+							 __global b3Float4 *sepAxis,
+							 b3MprSimplex_t *portal)
+{
+	b3Float4 dir;
+	b3MprSupport_t v4;
+
+	for (int i = 0; i < B3_MPR_MAX_ITERATIONS; i++)
+	//while (1)
+	{
+		// compute direction outside the portal (from v0 throught v1,v2,v3
+		// face)
+		b3PortalDir(portal, &dir);
+
+		// test if origin is inside the portal
+		if (portalEncapsulesOrigin(portal, &dir))
+			return 0;
+
+		// get next support point
+
+		b3MprSupport(pairIndex, bodyIndexA, bodyIndexB, cpuBodyBuf, cpuConvexData, cpuCollidables, cpuVertices, sepAxis, &dir, &v4);
+
+		// test if v4 can expand portal to contain origin and if portal
+		// expanding doesn't reach given tolerance
+		if (!portalCanEncapsuleOrigin(portal, &v4, &dir) || portalReachTolerance(portal, &v4, &dir))
+		{
+			return -1;
+		}
+
+		// v1-v2-v3 triangle must be rearranged to face outside Minkowski
+		// difference (direction from v0).
+		b3ExpandPortal(portal, &v4);
+	}
+
+	return -1;
 }
 
 B3_STATIC void b3FindPos(const b3MprSimplex_t *portal, b3Float4 *pos)
 {
+	b3Float4 zero = b3MakeFloat4(0, 0, 0, 0);
+	b3Float4 *b3mpr_vec3_origin = &zero;
 
-	b3Float4 zero = b3MakeFloat4(0,0,0,0);
-	b3Float4* b3mpr_vec3_origin = &zero;
+	b3Float4 dir;
+	size_t i;
+	float b[4], sum, inv;
+	b3Float4 vec, p1, p2;
 
-    b3Float4 dir;
-    size_t i;
-    float b[4], sum, inv;
-    b3Float4 vec, p1, p2;
+	b3PortalDir(portal, &dir);
 
-    b3PortalDir(portal, &dir);
+	// use barycentric coordinates of tetrahedron to find origin
+	b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 1)->v,
+				   &b3MprSimplexPoint(portal, 2)->v);
+	b[0] = b3MprVec3Dot(&vec, &b3MprSimplexPoint(portal, 3)->v);
 
-    // use barycentric coordinates of tetrahedron to find origin
-    b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 1)->v,
-                       &b3MprSimplexPoint(portal, 2)->v);
-    b[0] = b3MprVec3Dot(&vec, &b3MprSimplexPoint(portal, 3)->v);
+	b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 3)->v,
+				   &b3MprSimplexPoint(portal, 2)->v);
+	b[1] = b3MprVec3Dot(&vec, &b3MprSimplexPoint(portal, 0)->v);
 
-    b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 3)->v,
-                       &b3MprSimplexPoint(portal, 2)->v);
-    b[1] = b3MprVec3Dot(&vec, &b3MprSimplexPoint(portal, 0)->v);
+	b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 0)->v,
+				   &b3MprSimplexPoint(portal, 1)->v);
+	b[2] = b3MprVec3Dot(&vec, &b3MprSimplexPoint(portal, 3)->v);
 
-    b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 0)->v,
-                       &b3MprSimplexPoint(portal, 1)->v);
-    b[2] = b3MprVec3Dot(&vec, &b3MprSimplexPoint(portal, 3)->v);
-
-    b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 2)->v,
-                       &b3MprSimplexPoint(portal, 1)->v);
-    b[3] = b3MprVec3Dot(&vec, &b3MprSimplexPoint(portal, 0)->v);
+	b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 2)->v,
+				   &b3MprSimplexPoint(portal, 1)->v);
+	b[3] = b3MprVec3Dot(&vec, &b3MprSimplexPoint(portal, 0)->v);
 
 	sum = b[0] + b[1] + b[2] + b[3];
 
-    if (b3MprIsZero(sum) || sum < 0.f){
+	if (b3MprIsZero(sum) || sum < 0.f)
+	{
 		b[0] = 0.f;
 
-        b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 2)->v,
-                           &b3MprSimplexPoint(portal, 3)->v);
-        b[1] = b3MprVec3Dot(&vec, &dir);
-        b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 3)->v,
-                           &b3MprSimplexPoint(portal, 1)->v);
-        b[2] = b3MprVec3Dot(&vec, &dir);
-        b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 1)->v,
-                           &b3MprSimplexPoint(portal, 2)->v);
-        b[3] = b3MprVec3Dot(&vec, &dir);
+		b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 2)->v,
+					   &b3MprSimplexPoint(portal, 3)->v);
+		b[1] = b3MprVec3Dot(&vec, &dir);
+		b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 3)->v,
+					   &b3MprSimplexPoint(portal, 1)->v);
+		b[2] = b3MprVec3Dot(&vec, &dir);
+		b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 1)->v,
+					   &b3MprSimplexPoint(portal, 2)->v);
+		b[3] = b3MprVec3Dot(&vec, &dir);
 
 		sum = b[1] + b[2] + b[3];
 	}
 
 	inv = 1.f / sum;
 
-    b3MprVec3Copy(&p1, b3mpr_vec3_origin);
-    b3MprVec3Copy(&p2, b3mpr_vec3_origin);
-    for (i = 0; i < 4; i++){
-        b3MprVec3Copy(&vec, &b3MprSimplexPoint(portal, i)->v1);
-        b3MprVec3Scale(&vec, b[i]);
-        b3MprVec3Add(&p1, &vec);
+	b3MprVec3Copy(&p1, b3mpr_vec3_origin);
+	b3MprVec3Copy(&p2, b3mpr_vec3_origin);
+	for (i = 0; i < 4; i++)
+	{
+		b3MprVec3Copy(&vec, &b3MprSimplexPoint(portal, i)->v1);
+		b3MprVec3Scale(&vec, b[i]);
+		b3MprVec3Add(&p1, &vec);
 
-        b3MprVec3Copy(&vec, &b3MprSimplexPoint(portal, i)->v2);
-        b3MprVec3Scale(&vec, b[i]);
-        b3MprVec3Add(&p2, &vec);
-    }
-    b3MprVec3Scale(&p1, inv);
-    b3MprVec3Scale(&p2, inv);
+		b3MprVec3Copy(&vec, &b3MprSimplexPoint(portal, i)->v2);
+		b3MprVec3Scale(&vec, b[i]);
+		b3MprVec3Add(&p2, &vec);
+	}
+	b3MprVec3Scale(&p1, inv);
+	b3MprVec3Scale(&p2, inv);
 
-    b3MprVec3Copy(pos, &p1);
-    b3MprVec3Add(pos, &p2);
-    b3MprVec3Scale(pos, 0.5);
+	b3MprVec3Copy(pos, &p1);
+	b3MprVec3Add(pos, &p2);
+	b3MprVec3Scale(pos, 0.5);
 }
 
 inline float b3MprVec3Dist2(const b3Float4 *a, const b3Float4 *b)
 {
-    b3Float4 ab;
-    b3MprVec3Sub2(&ab, a, b);
-    return b3MprVec3Len2(&ab);
+	b3Float4 ab;
+	b3MprVec3Sub2(&ab, a, b);
+	return b3MprVec3Len2(&ab);
 }
 
 inline float _b3MprVec3PointSegmentDist2(const b3Float4 *P,
-                                                  const b3Float4 *x0,
-                                                  const b3Float4 *b,
-                                                  b3Float4 *witness)
+										 const b3Float4 *x0,
+										 const b3Float4 *b,
+										 b3Float4 *witness)
 {
-    // The computation comes from solving equation of segment:
-    //      S(t) = x0 + t.d
-    //          where - x0 is initial point of segment
-    //                - d is direction of segment from x0 (|d| > 0)
-    //                - t belongs to <0, 1> interval
-    // 
-    // Than, distance from a segment to some point P can be expressed:
-    //      D(t) = |x0 + t.d - P|^2
-    //          which is distance from any point on segment. Minimization
-    //          of this function brings distance from P to segment.
-    // Minimization of D(t) leads to simple quadratic equation that's
-    // solving is straightforward.
-    //
-    // Bonus of this method is witness point for free.
+	// The computation comes from solving equation of segment:
+	//      S(t) = x0 + t.d
+	//          where - x0 is initial point of segment
+	//                - d is direction of segment from x0 (|d| > 0)
+	//                - t belongs to <0, 1> interval
+	//
+	// Than, distance from a segment to some point P can be expressed:
+	//      D(t) = |x0 + t.d - P|^2
+	//          which is distance from any point on segment. Minimization
+	//          of this function brings distance from P to segment.
+	// Minimization of D(t) leads to simple quadratic equation that's
+	// solving is straightforward.
+	//
+	// Bonus of this method is witness point for free.
 
-    float dist, t;
-    b3Float4 d, a;
+	float dist, t;
+	b3Float4 d, a;
 
-    // direction of segment
-    b3MprVec3Sub2(&d, b, x0);
+	// direction of segment
+	b3MprVec3Sub2(&d, b, x0);
 
-    // precompute vector from P to x0
-    b3MprVec3Sub2(&a, x0, P);
+	// precompute vector from P to x0
+	b3MprVec3Sub2(&a, x0, P);
 
-    t  = -1.f * b3MprVec3Dot(&a, &d);
-    t /= b3MprVec3Len2(&d);
+	t = -1.f * b3MprVec3Dot(&a, &d);
+	t /= b3MprVec3Len2(&d);
 
-    if (t < 0.f || b3MprIsZero(t)){
-        dist = b3MprVec3Dist2(x0, P);
-        if (witness)
-            b3MprVec3Copy(witness, x0);
-    }else if (t > 1.f || b3MprEq(t, 1.f)){
-        dist = b3MprVec3Dist2(b, P);
-        if (witness)
-            b3MprVec3Copy(witness, b);
-    }else{
-        if (witness){
-            b3MprVec3Copy(witness, &d);
-            b3MprVec3Scale(witness, t);
-            b3MprVec3Add(witness, x0);
-            dist = b3MprVec3Dist2(witness, P);
-        }else{
-            // recycling variables
-            b3MprVec3Scale(&d, t);
-            b3MprVec3Add(&d, &a);
-            dist = b3MprVec3Len2(&d);
-        }
-    }
+	if (t < 0.f || b3MprIsZero(t))
+	{
+		dist = b3MprVec3Dist2(x0, P);
+		if (witness)
+			b3MprVec3Copy(witness, x0);
+	}
+	else if (t > 1.f || b3MprEq(t, 1.f))
+	{
+		dist = b3MprVec3Dist2(b, P);
+		if (witness)
+			b3MprVec3Copy(witness, b);
+	}
+	else
+	{
+		if (witness)
+		{
+			b3MprVec3Copy(witness, &d);
+			b3MprVec3Scale(witness, t);
+			b3MprVec3Add(witness, x0);
+			dist = b3MprVec3Dist2(witness, P);
+		}
+		else
+		{
+			// recycling variables
+			b3MprVec3Scale(&d, t);
+			b3MprVec3Add(&d, &a);
+			dist = b3MprVec3Len2(&d);
+		}
+	}
 
-    return dist;
+	return dist;
 }
 
-
 inline float b3MprVec3PointTriDist2(const b3Float4 *P,
-                                const b3Float4 *x0, const b3Float4 *B,
-                                const b3Float4 *C,
-                                b3Float4 *witness)
+									const b3Float4 *x0, const b3Float4 *B,
+									const b3Float4 *C,
+									b3Float4 *witness)
 {
-    // Computation comes from analytic expression for triangle (x0, B, C)
-    //      T(s, t) = x0 + s.d1 + t.d2, where d1 = B - x0 and d2 = C - x0 and
-    // Then equation for distance is:
-    //      D(s, t) = | T(s, t) - P |^2
-    // This leads to minimization of quadratic function of two variables.
-    // The solution from is taken only if s is between 0 and 1, t is
-    // between 0 and 1 and t + s < 1, otherwise distance from segment is
-    // computed.
+	// Computation comes from analytic expression for triangle (x0, B, C)
+	//      T(s, t) = x0 + s.d1 + t.d2, where d1 = B - x0 and d2 = C - x0 and
+	// Then equation for distance is:
+	//      D(s, t) = | T(s, t) - P |^2
+	// This leads to minimization of quadratic function of two variables.
+	// The solution from is taken only if s is between 0 and 1, t is
+	// between 0 and 1 and t + s < 1, otherwise distance from segment is
+	// computed.
 
-    b3Float4 d1, d2, a;
-    float u, v, w, p, q, r;
-    float s, t, dist, dist2;
-    b3Float4 witness2;
+	b3Float4 d1, d2, a;
+	float u, v, w, p, q, r;
+	float s, t, dist, dist2;
+	b3Float4 witness2;
 
-    b3MprVec3Sub2(&d1, B, x0);
-    b3MprVec3Sub2(&d2, C, x0);
-    b3MprVec3Sub2(&a, x0, P);
+	b3MprVec3Sub2(&d1, B, x0);
+	b3MprVec3Sub2(&d2, C, x0);
+	b3MprVec3Sub2(&a, x0, P);
 
-    u = b3MprVec3Dot(&a, &a);
-    v = b3MprVec3Dot(&d1, &d1);
-    w = b3MprVec3Dot(&d2, &d2);
-    p = b3MprVec3Dot(&a, &d1);
-    q = b3MprVec3Dot(&a, &d2);
-    r = b3MprVec3Dot(&d1, &d2);
+	u = b3MprVec3Dot(&a, &a);
+	v = b3MprVec3Dot(&d1, &d1);
+	w = b3MprVec3Dot(&d2, &d2);
+	p = b3MprVec3Dot(&a, &d1);
+	q = b3MprVec3Dot(&a, &d2);
+	r = b3MprVec3Dot(&d1, &d2);
 
-    s = (q * r - w * p) / (w * v - r * r);
-    t = (-s * r - q) / w;
+	s = (q * r - w * p) / (w * v - r * r);
+	t = (-s * r - q) / w;
 
-    if ((b3MprIsZero(s) || s > 0.f)
-            && (b3MprEq(s, 1.f) || s < 1.f)
-            && (b3MprIsZero(t) || t > 0.f)
-            && (b3MprEq(t, 1.f) || t < 1.f)
-            && (b3MprEq(t + s, 1.f) || t + s < 1.f)){
+	if ((b3MprIsZero(s) || s > 0.f) && (b3MprEq(s, 1.f) || s < 1.f) && (b3MprIsZero(t) || t > 0.f) && (b3MprEq(t, 1.f) || t < 1.f) && (b3MprEq(t + s, 1.f) || t + s < 1.f))
+	{
+		if (witness)
+		{
+			b3MprVec3Scale(&d1, s);
+			b3MprVec3Scale(&d2, t);
+			b3MprVec3Copy(witness, x0);
+			b3MprVec3Add(witness, &d1);
+			b3MprVec3Add(witness, &d2);
 
-        if (witness){
-            b3MprVec3Scale(&d1, s);
-            b3MprVec3Scale(&d2, t);
-            b3MprVec3Copy(witness, x0);
-            b3MprVec3Add(witness, &d1);
-            b3MprVec3Add(witness, &d2);
+			dist = b3MprVec3Dist2(witness, P);
+		}
+		else
+		{
+			dist = s * s * v;
+			dist += t * t * w;
+			dist += 2.f * s * t * r;
+			dist += 2.f * s * p;
+			dist += 2.f * t * q;
+			dist += u;
+		}
+	}
+	else
+	{
+		dist = _b3MprVec3PointSegmentDist2(P, x0, B, witness);
 
-            dist = b3MprVec3Dist2(witness, P);
-        }else{
-            dist  = s * s * v;
-            dist += t * t * w;
-            dist += 2.f * s * t * r;
-            dist += 2.f * s * p;
-            dist += 2.f * t * q;
-            dist += u;
-        }
-    }else{
-        dist = _b3MprVec3PointSegmentDist2(P, x0, B, witness);
+		dist2 = _b3MprVec3PointSegmentDist2(P, x0, C, &witness2);
+		if (dist2 < dist)
+		{
+			dist = dist2;
+			if (witness)
+				b3MprVec3Copy(witness, &witness2);
+		}
 
-        dist2 = _b3MprVec3PointSegmentDist2(P, x0, C, &witness2);
-        if (dist2 < dist){
-            dist = dist2;
-            if (witness)
-                b3MprVec3Copy(witness, &witness2);
-        }
+		dist2 = _b3MprVec3PointSegmentDist2(P, B, C, &witness2);
+		if (dist2 < dist)
+		{
+			dist = dist2;
+			if (witness)
+				b3MprVec3Copy(witness, &witness2);
+		}
+	}
 
-        dist2 = _b3MprVec3PointSegmentDist2(P, B, C, &witness2);
-        if (dist2 < dist){
-            dist = dist2;
-            if (witness)
-                b3MprVec3Copy(witness, &witness2);
-        }
-    }
-
-    return dist;
+	return dist;
 }
 
-
-B3_STATIC void b3FindPenetr(int pairIndex,int bodyIndexA, int bodyIndexB,  b3ConstArray(b3RigidBodyData_t) cpuBodyBuf, 
-													b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData, 
-													b3ConstArray(b3Collidable_t)				cpuCollidables,
-													b3ConstArray(b3Float4)					cpuVertices,
-													__global b3Float4* sepAxis,
-                       b3MprSimplex_t *portal,
-                       float *depth, b3Float4 *pdir, b3Float4 *pos)
+B3_STATIC void b3FindPenetr(int pairIndex, int bodyIndexA, int bodyIndexB, b3ConstArray(b3RigidBodyData_t) cpuBodyBuf,
+							b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData,
+							b3ConstArray(b3Collidable_t) cpuCollidables,
+							b3ConstArray(b3Float4) cpuVertices,
+							__global b3Float4 *sepAxis,
+							b3MprSimplex_t *portal,
+							float *depth, b3Float4 *pdir, b3Float4 *pos)
 {
-    b3Float4 dir;
-    b3MprSupport_t v4;
-    unsigned long iterations;
+	b3Float4 dir;
+	b3MprSupport_t v4;
+	unsigned long iterations;
 
-	b3Float4 zero = b3MakeFloat4(0,0,0,0);
-	b3Float4* b3mpr_vec3_origin = &zero;
+	b3Float4 zero = b3MakeFloat4(0, 0, 0, 0);
+	b3Float4 *b3mpr_vec3_origin = &zero;
 
-
-    iterations = 1UL;
-	for (int i=0;i<B3_MPR_MAX_ITERATIONS;i++)
-    //while (1)
+	iterations = 1UL;
+	for (int i = 0; i < B3_MPR_MAX_ITERATIONS; i++)
+	//while (1)
 	{
-        // compute portal direction and obtain next support point
-        b3PortalDir(portal, &dir);
-        
-		 b3MprSupport(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&dir, &v4);
+		// compute portal direction and obtain next support point
+		b3PortalDir(portal, &dir);
 
+		b3MprSupport(pairIndex, bodyIndexA, bodyIndexB, cpuBodyBuf, cpuConvexData, cpuCollidables, cpuVertices, sepAxis, &dir, &v4);
 
-        // reached tolerance -> find penetration info
-        if (portalReachTolerance(portal, &v4, &dir)
-                || iterations ==B3_MPR_MAX_ITERATIONS)
+		// reached tolerance -> find penetration info
+		if (portalReachTolerance(portal, &v4, &dir) || iterations == B3_MPR_MAX_ITERATIONS)
 		{
-            *depth = b3MprVec3PointTriDist2(b3mpr_vec3_origin,&b3MprSimplexPoint(portal, 1)->v,&b3MprSimplexPoint(portal, 2)->v,&b3MprSimplexPoint(portal, 3)->v,pdir);
-            *depth = B3_MPR_SQRT(*depth);
-			
+			*depth = b3MprVec3PointTriDist2(b3mpr_vec3_origin, &b3MprSimplexPoint(portal, 1)->v, &b3MprSimplexPoint(portal, 2)->v, &b3MprSimplexPoint(portal, 3)->v, pdir);
+			*depth = B3_MPR_SQRT(*depth);
+
 			if (b3MprIsZero((*pdir).x) && b3MprIsZero((*pdir).y) && b3MprIsZero((*pdir).z))
 			{
-				
 				*pdir = dir;
-			} 
+			}
 			b3MprVec3Normalize(pdir);
-			
-            // barycentric coordinates:
-            b3FindPos(portal, pos);
 
+			// barycentric coordinates:
+			b3FindPos(portal, pos);
 
-            return;
-        }
+			return;
+		}
 
-        b3ExpandPortal(portal, &v4);
+		b3ExpandPortal(portal, &v4);
 
-        iterations++;
-    }
+		iterations++;
+	}
 }
 
-B3_STATIC void b3FindPenetrTouch(b3MprSimplex_t *portal,float *depth, b3Float4 *dir, b3Float4 *pos)
+B3_STATIC void b3FindPenetrTouch(b3MprSimplex_t *portal, float *depth, b3Float4 *dir, b3Float4 *pos)
 {
-    // Touching contact on portal's v1 - so depth is zero and direction
-    // is unimportant and pos can be guessed
-    *depth = 0.f;
-    b3Float4 zero = b3MakeFloat4(0,0,0,0);
-	b3Float4* b3mpr_vec3_origin = &zero;
-
+	// Touching contact on portal's v1 - so depth is zero and direction
+	// is unimportant and pos can be guessed
+	*depth = 0.f;
+	b3Float4 zero = b3MakeFloat4(0, 0, 0, 0);
+	b3Float4 *b3mpr_vec3_origin = &zero;
 
 	b3MprVec3Copy(dir, b3mpr_vec3_origin);
 
-    b3MprVec3Copy(pos, &b3MprSimplexPoint(portal, 1)->v1);
-    b3MprVec3Add(pos, &b3MprSimplexPoint(portal, 1)->v2);
-    b3MprVec3Scale(pos, 0.5);
+	b3MprVec3Copy(pos, &b3MprSimplexPoint(portal, 1)->v1);
+	b3MprVec3Add(pos, &b3MprSimplexPoint(portal, 1)->v2);
+	b3MprVec3Scale(pos, 0.5);
 }
 
 B3_STATIC void b3FindPenetrSegment(b3MprSimplex_t *portal,
-                              float *depth, b3Float4 *dir, b3Float4 *pos)
+								   float *depth, b3Float4 *dir, b3Float4 *pos)
 {
-    
-    // Origin lies on v0-v1 segment.
-    // Depth is distance to v1, direction also and position must be
-    // computed
+	// Origin lies on v0-v1 segment.
+	// Depth is distance to v1, direction also and position must be
+	// computed
 
-    b3MprVec3Copy(pos, &b3MprSimplexPoint(portal, 1)->v1);
-    b3MprVec3Add(pos, &b3MprSimplexPoint(portal, 1)->v2);
-    b3MprVec3Scale(pos, 0.5f);
+	b3MprVec3Copy(pos, &b3MprSimplexPoint(portal, 1)->v1);
+	b3MprVec3Add(pos, &b3MprSimplexPoint(portal, 1)->v2);
+	b3MprVec3Scale(pos, 0.5f);
 
-    
-    b3MprVec3Copy(dir, &b3MprSimplexPoint(portal, 1)->v);
-    *depth = B3_MPR_SQRT(b3MprVec3Len2(dir));
-    b3MprVec3Normalize(dir);
+	b3MprVec3Copy(dir, &b3MprSimplexPoint(portal, 1)->v);
+	*depth = B3_MPR_SQRT(b3MprVec3Len2(dir));
+	b3MprVec3Normalize(dir);
 }
 
-
-
 inline int b3MprPenetration(int pairIndex, int bodyIndexA, int bodyIndexB,
-					b3ConstArray(b3RigidBodyData_t) cpuBodyBuf,
-					b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData, 
-					b3ConstArray(b3Collidable_t)	cpuCollidables,
-					b3ConstArray(b3Float4)	cpuVertices,
-					__global b3Float4* sepAxis,
-					__global int*	hasSepAxis,
-					float *depthOut, b3Float4* dirOut, b3Float4* posOut)
+							b3ConstArray(b3RigidBodyData_t) cpuBodyBuf,
+							b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData,
+							b3ConstArray(b3Collidable_t) cpuCollidables,
+							b3ConstArray(b3Float4) cpuVertices,
+							__global b3Float4 *sepAxis,
+							__global int *hasSepAxis,
+							float *depthOut, b3Float4 *dirOut, b3Float4 *posOut)
 {
-	
-	 b3MprSimplex_t portal;
+	b3MprSimplex_t portal;
 
-	 
-//	if (!hasSepAxis[pairIndex])
+	//	if (!hasSepAxis[pairIndex])
 	//	return -1;
-	
-	hasSepAxis[pairIndex] = 0;
-	 int res;
 
-    // Phase 1: Portal discovery
-    res = b3DiscoverPortal(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices,sepAxis,hasSepAxis, &portal);
-	
-	  
+	hasSepAxis[pairIndex] = 0;
+	int res;
+
+	// Phase 1: Portal discovery
+	res = b3DiscoverPortal(pairIndex, bodyIndexA, bodyIndexB, cpuBodyBuf, cpuConvexData, cpuCollidables, cpuVertices, sepAxis, hasSepAxis, &portal);
+
 	//sepAxis[pairIndex] = *pdir;//or -dir?
 
 	switch (res)
 	{
-	case 0:
+		case 0:
 		{
 			// Phase 2: Portal refinement
-		
-			res = b3RefinePortal(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&portal);
+
+			res = b3RefinePortal(pairIndex, bodyIndexA, bodyIndexB, cpuBodyBuf, cpuConvexData, cpuCollidables, cpuVertices, sepAxis, &portal);
 			if (res < 0)
 				return -1;
 
 			// Phase 3. Penetration info
-			b3FindPenetr(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&portal, depthOut, dirOut, posOut);
+			b3FindPenetr(pairIndex, bodyIndexA, bodyIndexB, cpuBodyBuf, cpuConvexData, cpuCollidables, cpuVertices, sepAxis, &portal, depthOut, dirOut, posOut);
 			hasSepAxis[pairIndex] = 1;
 			sepAxis[pairIndex] = -*dirOut;
 			break;
 		}
-	case 1:
+		case 1:
 		{
-			 // Touching contact on portal's v1.
+			// Touching contact on portal's v1.
 			b3FindPenetrTouch(&portal, depthOut, dirOut, posOut);
 			break;
 		}
-	case 2:
+		case 2:
 		{
-			
-			b3FindPenetrSegment( &portal, depthOut, dirOut, posOut);
+			b3FindPenetrSegment(&portal, depthOut, dirOut, posOut);
 			break;
 		}
-	default:
+		default:
 		{
-			hasSepAxis[pairIndex]=0;
+			hasSepAxis[pairIndex] = 0;
 			//if (res < 0)
 			//{
-				// Origin isn't inside portal - no collision.
-				return -1;
+			// Origin isn't inside portal - no collision.
+			return -1;
 			//}
 		}
 	};
-	
+
 	return 0;
 };
 
-
-
-#endif //B3_MPR_PENETRATION_H
+#endif  //B3_MPR_PENETRATION_H
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3NewContactReduction.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3NewContactReduction.h
index 718222ebcae..6e991e14b05 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3NewContactReduction.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3NewContactReduction.h
@@ -8,189 +8,168 @@
 
 #define GET_NPOINTS(x) (x).m_worldNormalOnB.w
 
-
 int b3ExtractManifoldSequentialGlobal(__global const b3Float4* p, int nPoints, b3Float4ConstArg nearNormal, b3Int4* contactIdx)
 {
-	if( nPoints == 0 )
-        return 0;
-    
-    if (nPoints <=4)
-        return nPoints;
-    
-    
-    if (nPoints >64)
-        nPoints = 64;
-    
-	b3Float4 center = b3MakeFloat4(0,0,0,0);
+	if (nPoints == 0)
+		return 0;
+
+	if (nPoints <= 4)
+		return nPoints;
+
+	if (nPoints > 64)
+		nPoints = 64;
+
+	b3Float4 center = b3MakeFloat4(0, 0, 0, 0);
 	{
-		
-		for (int i=0;i<nPoints;i++)
+		for (int i = 0; i < nPoints; i++)
 			center += p[i];
 		center /= (float)nPoints;
 	}
-    
-	
-    
+
 	//	sample 4 directions
-    
-    b3Float4 aVector = p[0] - center;
-    b3Float4 u = b3Cross( nearNormal, aVector );
-    b3Float4 v = b3Cross( nearNormal, u );
-    u = b3Normalized( u );
-    v = b3Normalized( v );
-    
-    
-    //keep point with deepest penetration
-    float minW= FLT_MAX;
-    
-    int minIndex=-1;
-    
-    b3Float4 maxDots;
-    maxDots.x = FLT_MIN;
-    maxDots.y = FLT_MIN;
-    maxDots.z = FLT_MIN;
-    maxDots.w = FLT_MIN;
-    
-    //	idx, distance
-    for(int ie = 0; ie<nPoints; ie++ )
-    {
-        if (p[ie].w<minW)
-        {
-            minW = p[ie].w;
-            minIndex=ie;
-        }
-        float f;
-        b3Float4 r = p[ie]-center;
-        f = b3Dot( u, r );
-        if (f<maxDots.x)
-        {
-            maxDots.x = f;
-            contactIdx[0].x = ie;
-        }
-        
-        f = b3Dot( -u, r );
-        if (f<maxDots.y)
-        {
-            maxDots.y = f;
-            contactIdx[0].y = ie;
-        }
-        
-        
-        f = b3Dot( v, r );
-        if (f<maxDots.z)
-        {
-            maxDots.z = f;
-            contactIdx[0].z = ie;
-        }
-        
-        f = b3Dot( -v, r );
-        if (f<maxDots.w)
-        {
-            maxDots.w = f;
-            contactIdx[0].w = ie;
-        }
-        
-    }
-    
-    if (contactIdx[0].x != minIndex && contactIdx[0].y != minIndex && contactIdx[0].z != minIndex && contactIdx[0].w != minIndex)
-    {
-        //replace the first contact with minimum (todo: replace contact with least penetration)
-        contactIdx[0].x = minIndex;
-    }
-    
-    return 4;
-    
+
+	b3Float4 aVector = p[0] - center;
+	b3Float4 u = b3Cross(nearNormal, aVector);
+	b3Float4 v = b3Cross(nearNormal, u);
+	u = b3Normalized(u);
+	v = b3Normalized(v);
+
+	//keep point with deepest penetration
+	float minW = FLT_MAX;
+
+	int minIndex = -1;
+
+	b3Float4 maxDots;
+	maxDots.x = FLT_MIN;
+	maxDots.y = FLT_MIN;
+	maxDots.z = FLT_MIN;
+	maxDots.w = FLT_MIN;
+
+	//	idx, distance
+	for (int ie = 0; ie < nPoints; ie++)
+	{
+		if (p[ie].w < minW)
+		{
+			minW = p[ie].w;
+			minIndex = ie;
+		}
+		float f;
+		b3Float4 r = p[ie] - center;
+		f = b3Dot(u, r);
+		if (f < maxDots.x)
+		{
+			maxDots.x = f;
+			contactIdx[0].x = ie;
+		}
+
+		f = b3Dot(-u, r);
+		if (f < maxDots.y)
+		{
+			maxDots.y = f;
+			contactIdx[0].y = ie;
+		}
+
+		f = b3Dot(v, r);
+		if (f < maxDots.z)
+		{
+			maxDots.z = f;
+			contactIdx[0].z = ie;
+		}
+
+		f = b3Dot(-v, r);
+		if (f < maxDots.w)
+		{
+			maxDots.w = f;
+			contactIdx[0].w = ie;
+		}
+	}
+
+	if (contactIdx[0].x != minIndex && contactIdx[0].y != minIndex && contactIdx[0].z != minIndex && contactIdx[0].w != minIndex)
+	{
+		//replace the first contact with minimum (todo: replace contact with least penetration)
+		contactIdx[0].x = minIndex;
+	}
+
+	return 4;
 }
 
-__kernel void   b3NewContactReductionKernel( __global b3Int4* pairs,
-                                                   __global const b3RigidBodyData_t* rigidBodies,
-                                                   __global const b3Float4* separatingNormals,
-                                                   __global const int* hasSeparatingAxis,
-                                                   __global struct b3Contact4Data* globalContactsOut,
-                                                   __global b3Int4* clippingFaces,
-                                                   __global b3Float4* worldVertsB2,
-                                                   volatile __global int* nGlobalContactsOut,
-                                                   int vertexFaceCapacity,
-												   int contactCapacity,
-                                                   int numPairs,
-												   int pairIndex
-                                                   )
+__kernel void b3NewContactReductionKernel(__global b3Int4* pairs,
+										  __global const b3RigidBodyData_t* rigidBodies,
+										  __global const b3Float4* separatingNormals,
+										  __global const int* hasSeparatingAxis,
+										  __global struct b3Contact4Data* globalContactsOut,
+										  __global b3Int4* clippingFaces,
+										  __global b3Float4* worldVertsB2,
+										  volatile __global int* nGlobalContactsOut,
+										  int vertexFaceCapacity,
+										  int contactCapacity,
+										  int numPairs,
+										  int pairIndex)
 {
-//    int i = get_global_id(0);
+	//    int i = get_global_id(0);
 	//int pairIndex = i;
 	int i = pairIndex;
 
-    b3Int4 contactIdx;
-    contactIdx=b3MakeInt4(0,1,2,3);
-    
-	if (i<numPairs)
+	b3Int4 contactIdx;
+	contactIdx = b3MakeInt4(0, 1, 2, 3);
+
+	if (i < numPairs)
 	{
-        
 		if (hasSeparatingAxis[i])
 		{
-            
-			
-            
-            
 			int nPoints = clippingFaces[pairIndex].w;
-           
-            if (nPoints>0)
-            {
 
-                 __global b3Float4* pointsIn = &worldVertsB2[pairIndex*vertexFaceCapacity];
-                b3Float4 normal = -separatingNormals[i];
-                
-                int nReducedContacts = b3ExtractManifoldSequentialGlobal(pointsIn, nPoints, normal, &contactIdx);
-            
-                int dstIdx;
-                dstIdx = b3AtomicInc( nGlobalContactsOut);
-				
-//#if 0
-                b3Assert(dstIdx < contactCapacity);
+			if (nPoints > 0)
+			{
+				__global b3Float4* pointsIn = &worldVertsB2[pairIndex * vertexFaceCapacity];
+				b3Float4 normal = -separatingNormals[i];
+
+				int nReducedContacts = b3ExtractManifoldSequentialGlobal(pointsIn, nPoints, normal, &contactIdx);
+
+				int dstIdx;
+				dstIdx = b3AtomicInc(nGlobalContactsOut);
+
+				//#if 0
+				b3Assert(dstIdx < contactCapacity);
 				if (dstIdx < contactCapacity)
 				{
-
 					__global struct b3Contact4Data* c = &globalContactsOut[dstIdx];
 					c->m_worldNormalOnB = -normal;
-					c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);
+					c->m_restituitionCoeffCmp = (0.f * 0xffff);
+					c->m_frictionCoeffCmp = (0.7f * 0xffff);
 					c->m_batchIdx = pairIndex;
 					int bodyA = pairs[pairIndex].x;
 					int bodyB = pairs[pairIndex].y;
 
 					pairs[pairIndex].w = dstIdx;
 
-					c->m_bodyAPtrAndSignBit = rigidBodies[bodyA].m_invMass==0?-bodyA:bodyA;
-					c->m_bodyBPtrAndSignBit = rigidBodies[bodyB].m_invMass==0?-bodyB:bodyB;
-                    c->m_childIndexA =-1;
-					c->m_childIndexB =-1;
+					c->m_bodyAPtrAndSignBit = rigidBodies[bodyA].m_invMass == 0 ? -bodyA : bodyA;
+					c->m_bodyBPtrAndSignBit = rigidBodies[bodyB].m_invMass == 0 ? -bodyB : bodyB;
+					c->m_childIndexA = -1;
+					c->m_childIndexB = -1;
+
+					switch (nReducedContacts)
+					{
+						case 4:
+							c->m_worldPosB[3] = pointsIn[contactIdx.w];
+						case 3:
+							c->m_worldPosB[2] = pointsIn[contactIdx.z];
+						case 2:
+							c->m_worldPosB[1] = pointsIn[contactIdx.y];
+						case 1:
+							c->m_worldPosB[0] = pointsIn[contactIdx.x];
+						default:
+						{
+						}
+					};
 
-                    switch (nReducedContacts)
-                    {
-                        case 4:
-                            c->m_worldPosB[3] = pointsIn[contactIdx.w];
-                        case 3:
-                            c->m_worldPosB[2] = pointsIn[contactIdx.z];
-                        case 2:
-                            c->m_worldPosB[1] = pointsIn[contactIdx.y];
-                        case 1:
-                            c->m_worldPosB[0] = pointsIn[contactIdx.x];
-                        default:
-                        {
-                        }
-                    };
-                    
 					GET_NPOINTS(*c) = nReducedContacts;
-                    
-                 }
-                 
-                
-//#endif
-				
-			}//		if (numContactsOut>0)
-		}//		if (hasSeparatingAxis[i])
-	}//	if (i<numPairs)
+				}
 
-    
-    
+				//#endif
+
+			}  //		if (numContactsOut>0)
+		}      //		if (hasSeparatingAxis[i])
+	}          //	if (i<numPairs)
 }
 #endif
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3QuantizedBvhNodeData.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3QuantizedBvhNodeData.h
index 3661e43cf1c..ba796eac72f 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3QuantizedBvhNodeData.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3QuantizedBvhNodeData.h
@@ -14,57 +14,56 @@ typedef struct b3QuantizedBvhNodeData b3QuantizedBvhNodeData_t;
 struct b3QuantizedBvhNodeData
 {
 	//12 bytes
-	unsigned short int	m_quantizedAabbMin[3];
-	unsigned short int	m_quantizedAabbMax[3];
+	unsigned short int m_quantizedAabbMin[3];
+	unsigned short int m_quantizedAabbMax[3];
 	//4 bytes
-	int	m_escapeIndexOrTriangleIndex;
+	int m_escapeIndexOrTriangleIndex;
 };
 
-inline int	b3GetTriangleIndex(const b3QuantizedBvhNodeData* rootNode)
+inline int b3GetTriangleIndex(const b3QuantizedBvhNodeData* rootNode)
 {
-	unsigned int x=0;
-	unsigned int y = (~(x&0))<<(31-B3_MAX_NUM_PARTS_IN_BITS);
+	unsigned int x = 0;
+	unsigned int y = (~(x & 0)) << (31 - B3_MAX_NUM_PARTS_IN_BITS);
 	// Get only the lower bits where the triangle index is stored
-	return (rootNode->m_escapeIndexOrTriangleIndex&~(y));
+	return (rootNode->m_escapeIndexOrTriangleIndex & ~(y));
 }
 
 inline int b3IsLeaf(const b3QuantizedBvhNodeData* rootNode)
 {
 	//skipindex is negative (internal node), triangleindex >=0 (leafnode)
-	return (rootNode->m_escapeIndexOrTriangleIndex >= 0)? 1 : 0;
+	return (rootNode->m_escapeIndexOrTriangleIndex >= 0) ? 1 : 0;
 }
-	
+
 inline int b3GetEscapeIndex(const b3QuantizedBvhNodeData* rootNode)
 {
 	return -rootNode->m_escapeIndexOrTriangleIndex;
 }
 
-inline void b3QuantizeWithClamp(unsigned short* out, b3Float4ConstArg point2,int isMax, b3Float4ConstArg bvhAabbMin, b3Float4ConstArg bvhAabbMax, b3Float4ConstArg bvhQuantization)
+inline void b3QuantizeWithClamp(unsigned short* out, b3Float4ConstArg point2, int isMax, b3Float4ConstArg bvhAabbMin, b3Float4ConstArg bvhAabbMax, b3Float4ConstArg bvhQuantization)
 {
-	b3Float4 clampedPoint = b3MaxFloat4(point2,bvhAabbMin);
-	clampedPoint = b3MinFloat4 (clampedPoint, bvhAabbMax);
+	b3Float4 clampedPoint = b3MaxFloat4(point2, bvhAabbMin);
+	clampedPoint = b3MinFloat4(clampedPoint, bvhAabbMax);
 
 	b3Float4 v = (clampedPoint - bvhAabbMin) * bvhQuantization;
 	if (isMax)
 	{
-		out[0] = (unsigned short) (((unsigned short)(v.x+1.f) | 1));
-		out[1] = (unsigned short) (((unsigned short)(v.y+1.f) | 1));
-		out[2] = (unsigned short) (((unsigned short)(v.z+1.f) | 1));
-	} else
-	{
-		out[0] = (unsigned short) (((unsigned short)(v.x) & 0xfffe));
-		out[1] = (unsigned short) (((unsigned short)(v.y) & 0xfffe));
-		out[2] = (unsigned short) (((unsigned short)(v.z) & 0xfffe));
+		out[0] = (unsigned short)(((unsigned short)(v.x + 1.f) | 1));
+		out[1] = (unsigned short)(((unsigned short)(v.y + 1.f) | 1));
+		out[2] = (unsigned short)(((unsigned short)(v.z + 1.f) | 1));
+	}
+	else
+	{
+		out[0] = (unsigned short)(((unsigned short)(v.x) & 0xfffe));
+		out[1] = (unsigned short)(((unsigned short)(v.y) & 0xfffe));
+		out[2] = (unsigned short)(((unsigned short)(v.z) & 0xfffe));
 	}
-
 }
 
-
 inline int b3TestQuantizedAabbAgainstQuantizedAabbSlow(
-								const unsigned short int* aabbMin1,
-								const unsigned short int* aabbMax1,
-								const unsigned short int* aabbMin2,
-								const unsigned short int* aabbMax2)
+	const unsigned short int* aabbMin1,
+	const unsigned short int* aabbMax1,
+	const unsigned short int* aabbMin2,
+	const unsigned short int* aabbMax2)
 {
 	//int overlap = 1;
 	if (aabbMin1[0] > aabbMax2[0])
@@ -86,5 +85,4 @@ inline int b3TestQuantizedAabbAgainstQuantizedAabbSlow(
 	//return overlap;
 }
 
-
-#endif //B3_QUANTIZED_BVH_NODE_H
+#endif  //B3_QUANTIZED_BVH_NODE_H
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3ReduceContacts.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3ReduceContacts.h
index 35b51970063..c108255b9f4 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3ReduceContacts.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3ReduceContacts.h
@@ -3,95 +3,87 @@
 
 inline int b3ReduceContacts(const b3Float4* p, int nPoints, const b3Float4& nearNormal, b3Int4* contactIdx)
 {
-	if( nPoints == 0 )
-        return 0;
-    
-    if (nPoints <=4)
-        return nPoints;
-    
-    
-    if (nPoints >64)
-        nPoints = 64;
-    
-	b3Float4 center = b3MakeFloat4(0,0,0,0);
+	if (nPoints == 0)
+		return 0;
+
+	if (nPoints <= 4)
+		return nPoints;
+
+	if (nPoints > 64)
+		nPoints = 64;
+
+	b3Float4 center = b3MakeFloat4(0, 0, 0, 0);
 	{
-		
-		for (int i=0;i<nPoints;i++)
+		for (int i = 0; i < nPoints; i++)
 			center += p[i];
 		center /= (float)nPoints;
 	}
-    
-	
-    
+
 	//	sample 4 directions
-    
-    b3Float4 aVector = p[0] - center;
-    b3Float4 u = b3Cross3( nearNormal, aVector );
-    b3Float4 v = b3Cross3( nearNormal, u );
-    u = b3FastNormalized3( u );
-    v = b3FastNormalized3( v );
-    
-    
-    //keep point with deepest penetration
-    float minW= FLT_MAX;
-    
-    int minIndex=-1;
-    
-    b3Float4 maxDots;
-    maxDots.x = FLT_MIN;
-    maxDots.y = FLT_MIN;
-    maxDots.z = FLT_MIN;
-    maxDots.w = FLT_MIN;
-    
-    //	idx, distance
-    for(int ie = 0; ie<nPoints; ie++ )
-    {
-        if (p[ie].w<minW)
-        {
-            minW = p[ie].w;
-            minIndex=ie;
-        }
-        float f;
-        b3Float4 r = p[ie]-center;
-        f = b3Dot3F4( u, r );
-        if (f<maxDots.x)
-        {
-            maxDots.x = f;
-            contactIdx[0].x = ie;
-        }
-        
-        f = b3Dot3F4( -u, r );
-        if (f<maxDots.y)
-        {
-            maxDots.y = f;
-            contactIdx[0].y = ie;
-        }
-        
-        
-        f = b3Dot3F4( v, r );
-        if (f<maxDots.z)
-        {
-            maxDots.z = f;
-            contactIdx[0].z = ie;
-        }
-        
-        f = b3Dot3F4( -v, r );
-        if (f<maxDots.w)
-        {
-            maxDots.w = f;
-            contactIdx[0].w = ie;
-        }
-        
-    }
-    
-    if (contactIdx[0].x != minIndex && contactIdx[0].y != minIndex && contactIdx[0].z != minIndex && contactIdx[0].w != minIndex)
-    {
-        //replace the first contact with minimum (todo: replace contact with least penetration)
-        contactIdx[0].x = minIndex;
-    }
-    
-    return 4;
-    
+
+	b3Float4 aVector = p[0] - center;
+	b3Float4 u = b3Cross3(nearNormal, aVector);
+	b3Float4 v = b3Cross3(nearNormal, u);
+	u = b3FastNormalized3(u);
+	v = b3FastNormalized3(v);
+
+	//keep point with deepest penetration
+	float minW = FLT_MAX;
+
+	int minIndex = -1;
+
+	b3Float4 maxDots;
+	maxDots.x = FLT_MIN;
+	maxDots.y = FLT_MIN;
+	maxDots.z = FLT_MIN;
+	maxDots.w = FLT_MIN;
+
+	//	idx, distance
+	for (int ie = 0; ie < nPoints; ie++)
+	{
+		if (p[ie].w < minW)
+		{
+			minW = p[ie].w;
+			minIndex = ie;
+		}
+		float f;
+		b3Float4 r = p[ie] - center;
+		f = b3Dot3F4(u, r);
+		if (f < maxDots.x)
+		{
+			maxDots.x = f;
+			contactIdx[0].x = ie;
+		}
+
+		f = b3Dot3F4(-u, r);
+		if (f < maxDots.y)
+		{
+			maxDots.y = f;
+			contactIdx[0].y = ie;
+		}
+
+		f = b3Dot3F4(v, r);
+		if (f < maxDots.z)
+		{
+			maxDots.z = f;
+			contactIdx[0].z = ie;
+		}
+
+		f = b3Dot3F4(-v, r);
+		if (f < maxDots.w)
+		{
+			maxDots.w = f;
+			contactIdx[0].w = ie;
+		}
+	}
+
+	if (contactIdx[0].x != minIndex && contactIdx[0].y != minIndex && contactIdx[0].z != minIndex && contactIdx[0].w != minIndex)
+	{
+		//replace the first contact with minimum (todo: replace contact with least penetration)
+		contactIdx[0].x = minIndex;
+	}
+
+	return 4;
 }
 
-#endif //B3_REDUCE_CONTACTS_H
+#endif  //B3_REDUCE_CONTACTS_H
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3RigidBodyData.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3RigidBodyData.h
index 50632c871f1..663e946fc1e 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3RigidBodyData.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3RigidBodyData.h
@@ -7,18 +7,17 @@
 
 typedef struct b3RigidBodyData b3RigidBodyData_t;
 
-
 struct b3RigidBodyData
 {
-	b3Float4				m_pos;
-	b3Quat					m_quat;
-	b3Float4				m_linVel;
-	b3Float4				m_angVel;
+	b3Float4 m_pos;
+	b3Quat m_quat;
+	b3Float4 m_linVel;
+	b3Float4 m_angVel;
 
-	int 					m_collidableIdx;
-	float 				m_invMass;
-	float 				m_restituitionCoeff;
-	float 				m_frictionCoeff;
+	int m_collidableIdx;
+	float m_invMass;
+	float m_restituitionCoeff;
+	float m_frictionCoeff;
 };
 
 typedef struct b3InertiaData b3InertiaData_t;
@@ -29,6 +28,4 @@ struct b3InertiaData
 	b3Mat3x3 m_initInvInertia;
 };
 
-
-#endif //B3_RIGIDBODY_DATA_H
-	
\ No newline at end of file
+#endif  //B3_RIGIDBODY_DATA_H
diff --git a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3UpdateAabbs.h b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3UpdateAabbs.h
index 8d40d19a036..e0c3a5cf978 100644
--- a/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3UpdateAabbs.h
+++ b/thirdparty/bullet/Bullet3Collision/NarrowPhaseCollision/shared/b3UpdateAabbs.h
@@ -1,40 +1,35 @@
 #ifndef B3_UPDATE_AABBS_H
 #define B3_UPDATE_AABBS_H
 
-
-
 #include "Bullet3Collision/BroadPhaseCollision/shared/b3Aabb.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3Collidable.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3RigidBodyData.h"
 
-
-
-void b3ComputeWorldAabb(  int bodyId, __global const b3RigidBodyData_t* bodies, __global const  b3Collidable_t* collidables, __global const  b3Aabb_t* localShapeAABB, __global b3Aabb_t* worldAabbs)
+void b3ComputeWorldAabb(int bodyId, __global const b3RigidBodyData_t* bodies, __global const b3Collidable_t* collidables, __global const b3Aabb_t* localShapeAABB, __global b3Aabb_t* worldAabbs)
 {
 	__global const b3RigidBodyData_t* body = &bodies[bodyId];
 
 	b3Float4 position = body->m_pos;
-	b3Quat	orientation = body->m_quat;
-	
+	b3Quat orientation = body->m_quat;
+
 	int collidableIndex = body->m_collidableIdx;
 	int shapeIndex = collidables[collidableIndex].m_shapeIndex;
-		
-	if (shapeIndex>=0)
+
+	if (shapeIndex >= 0)
 	{
-				
 		b3Aabb_t localAabb = localShapeAABB[collidableIndex];
 		b3Aabb_t worldAabb;
-		
-		b3Float4 aabbAMinOut,aabbAMaxOut;	
+
+		b3Float4 aabbAMinOut, aabbAMaxOut;
 		float margin = 0.f;
-		b3TransformAabb2(localAabb.m_minVec,localAabb.m_maxVec,margin,position,orientation,&aabbAMinOut,&aabbAMaxOut);
-		
-		worldAabb.m_minVec =aabbAMinOut;
+		b3TransformAabb2(localAabb.m_minVec, localAabb.m_maxVec, margin, position, orientation, &aabbAMinOut, &aabbAMaxOut);
+
+		worldAabb.m_minVec = aabbAMinOut;
 		worldAabb.m_minIndices[3] = bodyId;
 		worldAabb.m_maxVec = aabbAMaxOut;
-		worldAabb.m_signedMaxIndices[3] = body[bodyId].m_invMass==0.f? 0 : 1;
+		worldAabb.m_signedMaxIndices[3] = body[bodyId].m_invMass == 0.f ? 0 : 1;
 		worldAabbs[bodyId] = worldAabb;
 	}
 }
 
-#endif //B3_UPDATE_AABBS_H
+#endif  //B3_UPDATE_AABBS_H
diff --git a/thirdparty/bullet/Bullet3Common/b3AlignedAllocator.cpp b/thirdparty/bullet/Bullet3Common/b3AlignedAllocator.cpp
index b98e2b4d33e..d546d5e0669 100644
--- a/thirdparty/bullet/Bullet3Common/b3AlignedAllocator.cpp
+++ b/thirdparty/bullet/Bullet3Common/b3AlignedAllocator.cpp
@@ -15,9 +15,11 @@ subject to the following restrictions:
 
 #include "b3AlignedAllocator.h"
 
+#ifdef B3_ALLOCATOR_STATISTICS
 int b3g_numAlignedAllocs = 0;
 int b3g_numAlignedFree = 0;
-int b3g_totalBytesAlignedAllocs = 0;//detect memory leaks
+int b3g_totalBytesAlignedAllocs = 0;  //detect memory leaks
+#endif
 
 static void *b3AllocDefault(size_t size)
 {
@@ -29,12 +31,10 @@ static void b3FreeDefault(void *ptr)
 	free(ptr);
 }
 
-static b3AllocFunc* b3s_allocFunc = b3AllocDefault;
-static b3FreeFunc* b3s_freeFunc = b3FreeDefault;
+static b3AllocFunc *b3s_allocFunc = b3AllocDefault;
+static b3FreeFunc *b3s_freeFunc = b3FreeDefault;
 
-
-
-#if defined (B3_HAS_ALIGNED_ALLOCATOR)
+#if defined(B3_HAS_ALIGNED_ALLOCATOR)
 #include <malloc.h>
 static void *b3AlignedAllocDefault(size_t size, int alignment)
 {
@@ -59,123 +59,128 @@ static inline void b3AlignedFreeDefault(void *ptr)
 }
 #else
 
-
-
-
-
 static inline void *b3AlignedAllocDefault(size_t size, int alignment)
 {
-  void *ret;
-  char *real;
-  real = (char *)b3s_allocFunc(size + sizeof(void *) + (alignment-1));
-  if (real) {
-	ret = b3AlignPointer(real + sizeof(void *),alignment);
-    *((void **)(ret)-1) = (void *)(real);
-  } else {
-    ret = (void *)(real);
-  }
-  return (ret);
+	void *ret;
+	char *real;
+	real = (char *)b3s_allocFunc(size + sizeof(void *) + (alignment - 1));
+	if (real)
+	{
+		ret = b3AlignPointer(real + sizeof(void *), alignment);
+		*((void **)(ret)-1) = (void *)(real);
+	}
+	else
+	{
+		ret = (void *)(real);
+	}
+	return (ret);
 }
 
 static inline void b3AlignedFreeDefault(void *ptr)
 {
-  void* real;
+	void *real;
 
-  if (ptr) {
-    real = *((void **)(ptr)-1);
-    b3s_freeFunc(real);
-  }
+	if (ptr)
+	{
+		real = *((void **)(ptr)-1);
+		b3s_freeFunc(real);
+	}
 }
 #endif
 
-
-static b3AlignedAllocFunc* b3s_alignedAllocFunc = b3AlignedAllocDefault;
-static b3AlignedFreeFunc* b3s_alignedFreeFunc = b3AlignedFreeDefault;
+static b3AlignedAllocFunc *b3s_alignedAllocFunc = b3AlignedAllocDefault;
+static b3AlignedFreeFunc *b3s_alignedFreeFunc = b3AlignedFreeDefault;
 
 void b3AlignedAllocSetCustomAligned(b3AlignedAllocFunc *allocFunc, b3AlignedFreeFunc *freeFunc)
 {
-  b3s_alignedAllocFunc = allocFunc ? allocFunc : b3AlignedAllocDefault;
-  b3s_alignedFreeFunc = freeFunc ? freeFunc : b3AlignedFreeDefault;
+	b3s_alignedAllocFunc = allocFunc ? allocFunc : b3AlignedAllocDefault;
+	b3s_alignedFreeFunc = freeFunc ? freeFunc : b3AlignedFreeDefault;
 }
 
 void b3AlignedAllocSetCustom(b3AllocFunc *allocFunc, b3FreeFunc *freeFunc)
 {
-  b3s_allocFunc = allocFunc ? allocFunc : b3AllocDefault;
-  b3s_freeFunc = freeFunc ? freeFunc : b3FreeDefault;
+	b3s_allocFunc = allocFunc ? allocFunc : b3AllocDefault;
+	b3s_freeFunc = freeFunc ? freeFunc : b3FreeDefault;
 }
 
 #ifdef B3_DEBUG_MEMORY_ALLOCATIONS
 //this generic allocator provides the total allocated number of bytes
 #include <stdio.h>
 
-void*   b3AlignedAllocInternal  (size_t size, int alignment,int line,char* filename)
-{
- void *ret;
- char *real;
-
- b3g_totalBytesAlignedAllocs += size;
- b3g_numAlignedAllocs++;
-
- 
- real = (char *)b3s_allocFunc(size + 2*sizeof(void *) + (alignment-1));
- if (real) {
-   ret = (void*) b3AlignPointer(real + 2*sizeof(void *), alignment);
-   *((void **)(ret)-1) = (void *)(real);
-       *((int*)(ret)-2) = size;
-
- } else {
-   ret = (void *)(real);//??
- }
-
- b3Printf("allocation#%d at address %x, from %s,line %d, size %d\n",b3g_numAlignedAllocs,real, filename,line,size);
-
- int* ptr = (int*)ret;
- *ptr = 12;
- return (ret);
-}
-
-void    b3AlignedFreeInternal   (void* ptr,int line,char* filename)
-{
-
- void* real;
- b3g_numAlignedFree++;
-
- if (ptr) {
-   real = *((void **)(ptr)-1);
-       int size = *((int*)(ptr)-2);
-       b3g_totalBytesAlignedAllocs -= size;
-
-	   b3Printf("free #%d at address %x, from %s,line %d, size %d\n",b3g_numAlignedFree,real, filename,line,size);
-
-   b3s_freeFunc(real);
- } else
- {
-	 b3Printf("NULL ptr\n");
- }
-}
-
-#else //B3_DEBUG_MEMORY_ALLOCATIONS
-
-void*	b3AlignedAllocInternal	(size_t size, int alignment)
+void *b3AlignedAllocInternal(size_t size, int alignment, int line, char *filename)
 {
+	void *ret;
+	char *real;
+#ifdef B3_ALLOCATOR_STATISTICS
+	b3g_totalBytesAlignedAllocs += size;
 	b3g_numAlignedAllocs++;
-	void* ptr;
+#endif
+	real = (char *)b3s_allocFunc(size + 2 * sizeof(void *) + (alignment - 1));
+	if (real)
+	{
+		ret = (void *)b3AlignPointer(real + 2 * sizeof(void *), alignment);
+		*((void **)(ret)-1) = (void *)(real);
+		*((int *)(ret)-2) = size;
+	}
+	else
+	{
+		ret = (void *)(real);  //??
+	}
+
+	b3Printf("allocation#%d at address %x, from %s,line %d, size %d\n", b3g_numAlignedAllocs, real, filename, line, size);
+
+	int *ptr = (int *)ret;
+	*ptr = 12;
+	return (ret);
+}
+
+void b3AlignedFreeInternal(void *ptr, int line, char *filename)
+{
+	void *real;
+#ifdef B3_ALLOCATOR_STATISTICS
+	b3g_numAlignedFree++;
+#endif
+	if (ptr)
+	{
+		real = *((void **)(ptr)-1);
+		int size = *((int *)(ptr)-2);
+#ifdef B3_ALLOCATOR_STATISTICS
+		b3g_totalBytesAlignedAllocs -= size;
+#endif
+		b3Printf("free #%d at address %x, from %s,line %d, size %d\n", b3g_numAlignedFree, real, filename, line, size);
+
+		b3s_freeFunc(real);
+	}
+	else
+	{
+		b3Printf("NULL ptr\n");
+	}
+}
+
+#else  //B3_DEBUG_MEMORY_ALLOCATIONS
+
+void *b3AlignedAllocInternal(size_t size, int alignment)
+{
+#ifdef B3_ALLOCATOR_STATISTICS
+	b3g_numAlignedAllocs++;
+#endif
+	void *ptr;
 	ptr = b3s_alignedAllocFunc(size, alignment);
-//	b3Printf("b3AlignedAllocInternal %d, %x\n",size,ptr);
+	//	b3Printf("b3AlignedAllocInternal %d, %x\n",size,ptr);
 	return ptr;
 }
 
-void	b3AlignedFreeInternal	(void* ptr)
+void b3AlignedFreeInternal(void *ptr)
 {
 	if (!ptr)
 	{
 		return;
 	}
-
+#ifdef B3_ALLOCATOR_STATISTICS
 	b3g_numAlignedFree++;
-//	b3Printf("b3AlignedFreeInternal %x\n",ptr);
+#endif
+	//	b3Printf("b3AlignedFreeInternal %x\n",ptr);
 	b3s_alignedFreeFunc(ptr);
 }
 
-#endif //B3_DEBUG_MEMORY_ALLOCATIONS
-
+#endif  //B3_DEBUG_MEMORY_ALLOCATIONS
diff --git a/thirdparty/bullet/Bullet3Common/b3AlignedAllocator.h b/thirdparty/bullet/Bullet3Common/b3AlignedAllocator.h
index be418bd55f9..bcff9f128e7 100644
--- a/thirdparty/bullet/Bullet3Common/b3AlignedAllocator.h
+++ b/thirdparty/bullet/Bullet3Common/b3AlignedAllocator.h
@@ -24,84 +24,87 @@ subject to the following restrictions:
 //#define B3_DEBUG_MEMORY_ALLOCATIONS 1
 #ifdef B3_DEBUG_MEMORY_ALLOCATIONS
 
-#define b3AlignedAlloc(a,b) \
-		b3AlignedAllocInternal(a,b,__LINE__,__FILE__)
+#define b3AlignedAlloc(a, b) \
+	b3AlignedAllocInternal(a, b, __LINE__, __FILE__)
 
 #define b3AlignedFree(ptr) \
-		b3AlignedFreeInternal(ptr,__LINE__,__FILE__)
+	b3AlignedFreeInternal(ptr, __LINE__, __FILE__)
 
-void*	b3AlignedAllocInternal	(size_t size, int alignment,int line,char* filename);
+void* b3AlignedAllocInternal(size_t size, int alignment, int line, char* filename);
 
-void	b3AlignedFreeInternal	(void* ptr,int line,char* filename);
+void b3AlignedFreeInternal(void* ptr, int line, char* filename);
 
 #else
-	void*	b3AlignedAllocInternal	(size_t size, int alignment);
-	void	b3AlignedFreeInternal	(void* ptr);
+void* b3AlignedAllocInternal(size_t size, int alignment);
+void b3AlignedFreeInternal(void* ptr);
 
-	#define b3AlignedAlloc(size,alignment) b3AlignedAllocInternal(size,alignment)
-	#define b3AlignedFree(ptr) b3AlignedFreeInternal(ptr)
+#define b3AlignedAlloc(size, alignment) b3AlignedAllocInternal(size, alignment)
+#define b3AlignedFree(ptr) b3AlignedFreeInternal(ptr)
 
 #endif
-typedef int	btSizeType;
+typedef int btSizeType;
 
-typedef void *(b3AlignedAllocFunc)(size_t size, int alignment);
-typedef void (b3AlignedFreeFunc)(void *memblock);
-typedef void *(b3AllocFunc)(size_t size);
-typedef void (b3FreeFunc)(void *memblock);
+typedef void*(b3AlignedAllocFunc)(size_t size, int alignment);
+typedef void(b3AlignedFreeFunc)(void* memblock);
+typedef void*(b3AllocFunc)(size_t size);
+typedef void(b3FreeFunc)(void* memblock);
 
 ///The developer can let all Bullet memory allocations go through a custom memory allocator, using b3AlignedAllocSetCustom
-void b3AlignedAllocSetCustom(b3AllocFunc *allocFunc, b3FreeFunc *freeFunc);
+void b3AlignedAllocSetCustom(b3AllocFunc* allocFunc, b3FreeFunc* freeFunc);
 ///If the developer has already an custom aligned allocator, then b3AlignedAllocSetCustomAligned can be used. The default aligned allocator pre-allocates extra memory using the non-aligned allocator, and instruments it.
-void b3AlignedAllocSetCustomAligned(b3AlignedAllocFunc *allocFunc, b3AlignedFreeFunc *freeFunc);
-
+void b3AlignedAllocSetCustomAligned(b3AlignedAllocFunc* allocFunc, b3AlignedFreeFunc* freeFunc);
 
 ///The b3AlignedAllocator is a portable class for aligned memory allocations.
 ///Default implementations for unaligned and aligned allocations can be overridden by a custom allocator using b3AlignedAllocSetCustom and b3AlignedAllocSetCustomAligned.
-template < typename T , unsigned Alignment >
-class b3AlignedAllocator {
-	
-	typedef b3AlignedAllocator< T , Alignment > self_type;
-	
-public:
+template <typename T, unsigned Alignment>
+class b3AlignedAllocator
+{
+	typedef b3AlignedAllocator<T, Alignment> self_type;
 
+public:
 	//just going down a list:
 	b3AlignedAllocator() {}
 	/*
 	b3AlignedAllocator( const self_type & ) {}
 	*/
 
-	template < typename Other >
-	b3AlignedAllocator( const b3AlignedAllocator< Other , Alignment > & ) {}
+	template <typename Other>
+	b3AlignedAllocator(const b3AlignedAllocator<Other, Alignment>&)
+	{
+	}
 
-	typedef const T*         const_pointer;
-	typedef const T&         const_reference;
-	typedef T*               pointer;
-	typedef T&               reference;
-	typedef T                value_type;
+	typedef const T* const_pointer;
+	typedef const T& const_reference;
+	typedef T* pointer;
+	typedef T& reference;
+	typedef T value_type;
 
-	pointer       address   ( reference        ref ) const                           { return &ref; }
-	const_pointer address   ( const_reference  ref ) const                           { return &ref; }
-	pointer       allocate  ( btSizeType        n   , const_pointer *      hint = 0 ) {
+	pointer address(reference ref) const { return &ref; }
+	const_pointer address(const_reference ref) const { return &ref; }
+	pointer allocate(btSizeType n, const_pointer* hint = 0)
+	{
 		(void)hint;
-		return reinterpret_cast< pointer >(b3AlignedAlloc( sizeof(value_type) * n , Alignment ));
+		return reinterpret_cast<pointer>(b3AlignedAlloc(sizeof(value_type) * n, Alignment));
 	}
-	void          construct ( pointer          ptr , const value_type &   value    ) { new (ptr) value_type( value ); }
-	void          deallocate( pointer          ptr ) {
-		b3AlignedFree( reinterpret_cast< void * >( ptr ) );
+	void construct(pointer ptr, const value_type& value) { new (ptr) value_type(value); }
+	void deallocate(pointer ptr)
+	{
+		b3AlignedFree(reinterpret_cast<void*>(ptr));
 	}
-	void          destroy   ( pointer          ptr )                                 { ptr->~value_type(); }
-	
+	void destroy(pointer ptr) { ptr->~value_type(); }
 
-	template < typename O > struct rebind {
-		typedef b3AlignedAllocator< O , Alignment > other;
+	template <typename O>
+	struct rebind
+	{
+		typedef b3AlignedAllocator<O, Alignment> other;
 	};
-	template < typename O >
-	self_type & operator=( const b3AlignedAllocator< O , Alignment > & ) { return *this; }
+	template <typename O>
+	self_type& operator=(const b3AlignedAllocator<O, Alignment>&)
+	{
+		return *this;
+	}
 
-	friend bool operator==( const self_type & , const self_type & ) { return true; }
+	friend bool operator==(const self_type&, const self_type&) { return true; }
 };
 
-
-
-#endif //B3_ALIGNED_ALLOCATOR
-
+#endif  //B3_ALIGNED_ALLOCATOR
diff --git a/thirdparty/bullet/Bullet3Common/b3AlignedObjectArray.h b/thirdparty/bullet/Bullet3Common/b3AlignedObjectArray.h
index ef71016565a..249e381bf1a 100644
--- a/thirdparty/bullet/Bullet3Common/b3AlignedObjectArray.h
+++ b/thirdparty/bullet/Bullet3Common/b3AlignedObjectArray.h
@@ -13,11 +13,10 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef B3_OBJECT_ARRAY__
 #define B3_OBJECT_ARRAY__
 
-#include "b3Scalar.h" // has definitions like B3_FORCE_INLINE
+#include "b3Scalar.h"  // has definitions like B3_FORCE_INLINE
 #include "b3AlignedAllocator.h"
 
 ///If the platform doesn't support placement new, you can disable B3_USE_PLACEMENT_NEW
@@ -28,402 +27,386 @@ subject to the following restrictions:
 
 #define B3_USE_PLACEMENT_NEW 1
 //#define B3_USE_MEMCPY 1 //disable, because it is cumbersome to find out for each platform where memcpy is defined. It can be in <memory.h> or <string.h> or otherwise...
-#define B3_ALLOW_ARRAY_COPY_OPERATOR // enabling this can accidently perform deep copies of data if you are not careful
+#define B3_ALLOW_ARRAY_COPY_OPERATOR  // enabling this can accidently perform deep copies of data if you are not careful
 
 #ifdef B3_USE_MEMCPY
 #include <memory.h>
 #include <string.h>
-#endif //B3_USE_MEMCPY
+#endif  //B3_USE_MEMCPY
 
 #ifdef B3_USE_PLACEMENT_NEW
-#include <new> //for placement new
-#endif //B3_USE_PLACEMENT_NEW
-
+#include <new>  //for placement new
+#endif          //B3_USE_PLACEMENT_NEW
 
 ///The b3AlignedObjectArray template class uses a subset of the stl::vector interface for its methods
 ///It is developed to replace stl::vector to avoid portability issues, including STL alignment issues to add SIMD/SSE data
-template <typename T> 
-//template <class T> 
+template <typename T>
+//template <class T>
 class b3AlignedObjectArray
 {
-	b3AlignedAllocator<T , 16>	m_allocator;
+	b3AlignedAllocator<T, 16> m_allocator;
 
-	int					m_size;
-	int					m_capacity;
-	T*					m_data;
+	int m_size;
+	int m_capacity;
+	T* m_data;
 	//PCK: added this line
-	bool				m_ownsMemory;
+	bool m_ownsMemory;
 
 #ifdef B3_ALLOW_ARRAY_COPY_OPERATOR
 public:
-	B3_FORCE_INLINE b3AlignedObjectArray<T>& operator=(const b3AlignedObjectArray<T> &other)
+	B3_FORCE_INLINE b3AlignedObjectArray<T>& operator=(const b3AlignedObjectArray<T>& other)
 	{
 		copyFromArray(other);
 		return *this;
 	}
-#else//B3_ALLOW_ARRAY_COPY_OPERATOR
+#else   //B3_ALLOW_ARRAY_COPY_OPERATOR
 private:
-		B3_FORCE_INLINE b3AlignedObjectArray<T>& operator=(const b3AlignedObjectArray<T> &other);
-#endif//B3_ALLOW_ARRAY_COPY_OPERATOR
+	B3_FORCE_INLINE b3AlignedObjectArray<T>& operator=(const b3AlignedObjectArray<T>& other);
+#endif  //B3_ALLOW_ARRAY_COPY_OPERATOR
 
 protected:
-		B3_FORCE_INLINE	int	allocSize(int size)
-		{
-			return (size ? size*2 : 1);
-		}
-		B3_FORCE_INLINE	void	copy(int start,int end, T* dest) const
-		{
-			int i;
-			for (i=start;i<end;++i)
+	B3_FORCE_INLINE int allocSize(int size)
+	{
+		return (size ? size * 2 : 1);
+	}
+	B3_FORCE_INLINE void copy(int start, int end, T* dest) const
+	{
+		int i;
+		for (i = start; i < end; ++i)
 #ifdef B3_USE_PLACEMENT_NEW
-				new (&dest[i]) T(m_data[i]);
+			new (&dest[i]) T(m_data[i]);
 #else
-				dest[i] = m_data[i];
-#endif //B3_USE_PLACEMENT_NEW
-		}
+			dest[i] = m_data[i];
+#endif  //B3_USE_PLACEMENT_NEW
+	}
 
-		B3_FORCE_INLINE	void	init()
+	B3_FORCE_INLINE void init()
+	{
+		//PCK: added this line
+		m_ownsMemory = true;
+		m_data = 0;
+		m_size = 0;
+		m_capacity = 0;
+	}
+	B3_FORCE_INLINE void destroy(int first, int last)
+	{
+		int i;
+		for (i = first; i < last; i++)
 		{
-			//PCK: added this line
-			m_ownsMemory = true;
-			m_data = 0;
-			m_size = 0;
-			m_capacity = 0;
+			m_data[i].~T();
 		}
-		B3_FORCE_INLINE	void	destroy(int first,int last)
+	}
+
+	B3_FORCE_INLINE void* allocate(int size)
+	{
+		if (size)
+			return m_allocator.allocate(size);
+		return 0;
+	}
+
+	B3_FORCE_INLINE void deallocate()
+	{
+		if (m_data)
 		{
-			int i;
-			for (i=first; i<last;i++)
+			//PCK: enclosed the deallocation in this block
+			if (m_ownsMemory)
+			{
+				m_allocator.deallocate(m_data);
+			}
+			m_data = 0;
+		}
+	}
+
+public:
+	b3AlignedObjectArray()
+	{
+		init();
+	}
+
+	~b3AlignedObjectArray()
+	{
+		clear();
+	}
+
+	///Generally it is best to avoid using the copy constructor of an b3AlignedObjectArray, and use a (const) reference to the array instead.
+	b3AlignedObjectArray(const b3AlignedObjectArray& otherArray)
+	{
+		init();
+
+		int otherSize = otherArray.size();
+		resize(otherSize);
+		otherArray.copy(0, otherSize, m_data);
+	}
+
+	/// return the number of elements in the array
+	B3_FORCE_INLINE int size() const
+	{
+		return m_size;
+	}
+
+	B3_FORCE_INLINE const T& at(int n) const
+	{
+		b3Assert(n >= 0);
+		b3Assert(n < size());
+		return m_data[n];
+	}
+
+	B3_FORCE_INLINE T& at(int n)
+	{
+		b3Assert(n >= 0);
+		b3Assert(n < size());
+		return m_data[n];
+	}
+
+	B3_FORCE_INLINE const T& operator[](int n) const
+	{
+		b3Assert(n >= 0);
+		b3Assert(n < size());
+		return m_data[n];
+	}
+
+	B3_FORCE_INLINE T& operator[](int n)
+	{
+		b3Assert(n >= 0);
+		b3Assert(n < size());
+		return m_data[n];
+	}
+
+	///clear the array, deallocated memory. Generally it is better to use array.resize(0), to reduce performance overhead of run-time memory (de)allocations.
+	B3_FORCE_INLINE void clear()
+	{
+		destroy(0, size());
+
+		deallocate();
+
+		init();
+	}
+
+	B3_FORCE_INLINE void pop_back()
+	{
+		b3Assert(m_size > 0);
+		m_size--;
+		m_data[m_size].~T();
+	}
+
+	///resize changes the number of elements in the array. If the new size is larger, the new elements will be constructed using the optional second argument.
+	///when the new number of elements is smaller, the destructor will be called, but memory will not be freed, to reduce performance overhead of run-time memory (de)allocations.
+	B3_FORCE_INLINE void resizeNoInitialize(int newsize)
+	{
+		int curSize = size();
+
+		if (newsize < curSize)
+		{
+		}
+		else
+		{
+			if (newsize > size())
+			{
+				reserve(newsize);
+			}
+			//leave this uninitialized
+		}
+		m_size = newsize;
+	}
+
+	B3_FORCE_INLINE void resize(int newsize, const T& fillData = T())
+	{
+		int curSize = size();
+
+		if (newsize < curSize)
+		{
+			for (int i = newsize; i < curSize; i++)
 			{
 				m_data[i].~T();
 			}
 		}
-
-		B3_FORCE_INLINE	void* allocate(int size)
+		else
 		{
-			if (size)
-				return m_allocator.allocate(size);
-			return 0;
-		}
-
-		B3_FORCE_INLINE	void	deallocate()
-		{
-			if(m_data)	{
-				//PCK: enclosed the deallocation in this block
-				if (m_ownsMemory)
-				{
-					m_allocator.deallocate(m_data);
-				}
-				m_data = 0;
+			if (newsize > size())
+			{
+				reserve(newsize);
 			}
-		}
-
-	
-
-
-	public:
-		
-		b3AlignedObjectArray()
-		{
-			init();
-		}
-
-		~b3AlignedObjectArray()
-		{
-			clear();
-		}
-
-		///Generally it is best to avoid using the copy constructor of an b3AlignedObjectArray, and use a (const) reference to the array instead.
-		b3AlignedObjectArray(const b3AlignedObjectArray& otherArray)
-		{
-			init();
-
-			int otherSize = otherArray.size();
-			resize (otherSize);
-			otherArray.copy(0, otherSize, m_data);
-		}
-
-		
-		
-		/// return the number of elements in the array
-		B3_FORCE_INLINE	int size() const
-		{	
-			return m_size;
-		}
-		
-		B3_FORCE_INLINE const T& at(int n) const
-		{
-			b3Assert(n>=0);
-			b3Assert(n<size());
-			return m_data[n];
-		}
-
-		B3_FORCE_INLINE T& at(int n)
-		{
-			b3Assert(n>=0);
-			b3Assert(n<size());
-			return m_data[n];
-		}
-
-		B3_FORCE_INLINE const T& operator[](int n) const
-		{
-			b3Assert(n>=0);
-			b3Assert(n<size());
-			return m_data[n];
-		}
-
-		B3_FORCE_INLINE T& operator[](int n)
-		{
-			b3Assert(n>=0);
-			b3Assert(n<size());
-			return m_data[n];
-		}
-		
-
-		///clear the array, deallocated memory. Generally it is better to use array.resize(0), to reduce performance overhead of run-time memory (de)allocations.
-		B3_FORCE_INLINE	void	clear()
-		{
-			destroy(0,size());
-			
-			deallocate();
-			
-			init();
-		}
-
-		B3_FORCE_INLINE	void	pop_back()
-		{
-			b3Assert(m_size>0);
-			m_size--;
-			m_data[m_size].~T();
-		}
-
-
-		///resize changes the number of elements in the array. If the new size is larger, the new elements will be constructed using the optional second argument.
-		///when the new number of elements is smaller, the destructor will be called, but memory will not be freed, to reduce performance overhead of run-time memory (de)allocations.
-		B3_FORCE_INLINE	void	resizeNoInitialize(int newsize)
-		{
-			int curSize = size();
-
-			if (newsize < curSize)
-			{
-			} else
-			{
-				if (newsize > size())
-				{
-					reserve(newsize);
-				}
-				//leave this uninitialized
-			}
-			m_size = newsize;
-		}
-	
-		B3_FORCE_INLINE	void	resize(int newsize, const T& fillData=T())
-		{
-			int curSize = size();
-
-			if (newsize < curSize)
-			{
-				for(int i = newsize; i < curSize; i++)
-				{
-					m_data[i].~T();
-				}
-			} else
-			{
-				if (newsize > size())
-				{
-					reserve(newsize);
-				}
 #ifdef B3_USE_PLACEMENT_NEW
-				for (int i=curSize;i<newsize;i++)
-				{
-					new ( &m_data[i]) T(fillData);
-				}
-#endif //B3_USE_PLACEMENT_NEW
-
-			}
-
-			m_size = newsize;
-		}
-		B3_FORCE_INLINE	T&  expandNonInitializing( )
-		{	
-			int sz = size();
-			if( sz == capacity() )
+			for (int i = curSize; i < newsize; i++)
 			{
-				reserve( allocSize(size()) );
+				new (&m_data[i]) T(fillData);
 			}
-			m_size++;
-
-			return m_data[sz];		
+#endif  //B3_USE_PLACEMENT_NEW
 		}
 
+		m_size = newsize;
+	}
+	B3_FORCE_INLINE T& expandNonInitializing()
+	{
+		int sz = size();
+		if (sz == capacity())
+		{
+			reserve(allocSize(size()));
+		}
+		m_size++;
 
-		B3_FORCE_INLINE	T&  expand( const T& fillValue=T())
-		{	
-			int sz = size();
-			if( sz == capacity() )
-			{
-				reserve( allocSize(size()) );
-			}
-			m_size++;
+		return m_data[sz];
+	}
+
+	B3_FORCE_INLINE T& expand(const T& fillValue = T())
+	{
+		int sz = size();
+		if (sz == capacity())
+		{
+			reserve(allocSize(size()));
+		}
+		m_size++;
 #ifdef B3_USE_PLACEMENT_NEW
-			new (&m_data[sz]) T(fillValue); //use the in-place new (not really allocating heap memory)
+		new (&m_data[sz]) T(fillValue);  //use the in-place new (not really allocating heap memory)
 #endif
 
-			return m_data[sz];		
+		return m_data[sz];
+	}
+
+	B3_FORCE_INLINE void push_back(const T& _Val)
+	{
+		int sz = size();
+		if (sz == capacity())
+		{
+			reserve(allocSize(size()));
 		}
 
-
-		B3_FORCE_INLINE	void push_back(const T& _Val)
-		{	
-			int sz = size();
-			if( sz == capacity() )
-			{
-				reserve( allocSize(size()) );
-			}
-			
 #ifdef B3_USE_PLACEMENT_NEW
-			new ( &m_data[m_size] ) T(_Val);
+		new (&m_data[m_size]) T(_Val);
 #else
-			m_data[size()] = _Val;			
-#endif //B3_USE_PLACEMENT_NEW
+		m_data[size()] = _Val;
+#endif  //B3_USE_PLACEMENT_NEW
 
-			m_size++;
-		}
+		m_size++;
+	}
 
-	
-		/// return the pre-allocated (reserved) elements, this is at least as large as the total number of elements,see size() and reserve()
-		B3_FORCE_INLINE	int capacity() const
-		{	
-			return m_capacity;
-		}
-		
-		B3_FORCE_INLINE	void reserve(int _Count)
-		{	// determine new minimum length of allocated storage
-			if (capacity() < _Count)
-			{	// not enough room, reallocate
-				T*	s = (T*)allocate(_Count);
-				b3Assert(s);
-				if (s==0)
-				{
-					b3Error("b3AlignedObjectArray reserve out-of-memory\n");
-					_Count=0;
-					m_size=0;
-				}
-				copy(0, size(), s);
-
-				destroy(0,size());
-
-				deallocate();
-				
-				//PCK: added this line
-				m_ownsMemory = true;
-
-				m_data = s;
-				
-				m_capacity = _Count;
+	/// return the pre-allocated (reserved) elements, this is at least as large as the total number of elements,see size() and reserve()
+	B3_FORCE_INLINE int capacity() const
+	{
+		return m_capacity;
+	}
 
+	B3_FORCE_INLINE void reserve(int _Count)
+	{  // determine new minimum length of allocated storage
+		if (capacity() < _Count)
+		{  // not enough room, reallocate
+			T* s = (T*)allocate(_Count);
+			b3Assert(s);
+			if (s == 0)
+			{
+				b3Error("b3AlignedObjectArray reserve out-of-memory\n");
+				_Count = 0;
+				m_size = 0;
 			}
+			copy(0, size(), s);
+
+			destroy(0, size());
+
+			deallocate();
+
+			//PCK: added this line
+			m_ownsMemory = true;
+
+			m_data = s;
+
+			m_capacity = _Count;
 		}
+	}
 
-
-		class less
+	class less
+	{
+	public:
+		bool operator()(const T& a, const T& b)
 		{
-			public:
+			return (a < b);
+		}
+	};
 
-				bool operator() ( const T& a, const T& b )
-				{
-					return ( a < b );
-				}
-		};
-	
-
-		template <typename L>
-		void quickSortInternal(const L& CompareFunc,int lo, int hi)
-		{
+	template <typename L>
+	void quickSortInternal(const L& CompareFunc, int lo, int hi)
+	{
 		//  lo is the lower index, hi is the upper index
 		//  of the region of array a that is to be sorted
-			int i=lo, j=hi;
-			T x=m_data[(lo+hi)/2];
+		int i = lo, j = hi;
+		T x = m_data[(lo + hi) / 2];
 
-			//  partition
-			do
-			{    
-				while (CompareFunc(m_data[i],x)) 
-					i++; 
-				while (CompareFunc(x,m_data[j])) 
-					j--;
-				if (i<=j)
-				{
-					swap(i,j);
-					i++; j--;
-				}
-			} while (i<=j);
-
-			//  recursion
-			if (lo<j) 
-				quickSortInternal( CompareFunc, lo, j);
-			if (i<hi) 
-				quickSortInternal( CompareFunc, i, hi);
-		}
-
-
-		template <typename L>
-		void quickSort(const L& CompareFunc)
+		//  partition
+		do
 		{
-			//don't sort 0 or 1 elements
-			if (size()>1)
+			while (CompareFunc(m_data[i], x))
+				i++;
+			while (CompareFunc(x, m_data[j]))
+				j--;
+			if (i <= j)
 			{
-				quickSortInternal(CompareFunc,0,size()-1);
+				swap(i, j);
+				i++;
+				j--;
+			}
+		} while (i <= j);
+
+		//  recursion
+		if (lo < j)
+			quickSortInternal(CompareFunc, lo, j);
+		if (i < hi)
+			quickSortInternal(CompareFunc, i, hi);
+	}
+
+	template <typename L>
+	void quickSort(const L& CompareFunc)
+	{
+		//don't sort 0 or 1 elements
+		if (size() > 1)
+		{
+			quickSortInternal(CompareFunc, 0, size() - 1);
+		}
+	}
+
+	///heap sort from http://www.csse.monash.edu.au/~lloyd/tildeAlgDS/Sort/Heap/
+	template <typename L>
+	void downHeap(T* pArr, int k, int n, const L& CompareFunc)
+	{
+		/*  PRE: a[k+1..N] is a heap */
+		/* POST:  a[k..N]  is a heap */
+
+		T temp = pArr[k - 1];
+		/* k has child(s) */
+		while (k <= n / 2)
+		{
+			int child = 2 * k;
+
+			if ((child < n) && CompareFunc(pArr[child - 1], pArr[child]))
+			{
+				child++;
+			}
+			/* pick larger child */
+			if (CompareFunc(temp, pArr[child - 1]))
+			{
+				/* move child up */
+				pArr[k - 1] = pArr[child - 1];
+				k = child;
+			}
+			else
+			{
+				break;
 			}
 		}
+		pArr[k - 1] = temp;
+	} /*downHeap*/
 
-
-		///heap sort from http://www.csse.monash.edu.au/~lloyd/tildeAlgDS/Sort/Heap/
-		template <typename L>
-		void downHeap(T *pArr, int k, int n, const L& CompareFunc)
-		{
-			/*  PRE: a[k+1..N] is a heap */
-			/* POST:  a[k..N]  is a heap */
-			
-			T temp = pArr[k - 1];
-			/* k has child(s) */
-			while (k <= n/2) 
-			{
-				int child = 2*k;
-				
-				if ((child < n) && CompareFunc(pArr[child - 1] , pArr[child]))
-				{
-					child++;
-				}
-				/* pick larger child */
-				if (CompareFunc(temp , pArr[child - 1]))
-				{
-					/* move child up */
-					pArr[k - 1] = pArr[child - 1];
-					k = child;
-				}
-				else
-				{
-					break;
-				}
-			}
-			pArr[k - 1] = temp;
-		} /*downHeap*/
-
-		void	swap(int index0,int index1)
-		{
+	void swap(int index0, int index1)
+	{
 #ifdef B3_USE_MEMCPY
-			char	temp[sizeof(T)];
-			memcpy(temp,&m_data[index0],sizeof(T));
-			memcpy(&m_data[index0],&m_data[index1],sizeof(T));
-			memcpy(&m_data[index1],temp,sizeof(T));
+		char temp[sizeof(T)];
+		memcpy(temp, &m_data[index0], sizeof(T));
+		memcpy(&m_data[index0], &m_data[index1], sizeof(T));
+		memcpy(&m_data[index1], temp, sizeof(T));
 #else
-			T temp = m_data[index0];
-			m_data[index0] = m_data[index1];
-			m_data[index1] = temp;
-#endif //B3_USE_PLACEMENT_NEW
-
-		}
+		T temp = m_data[index0];
+		m_data[index0] = m_data[index1];
+		m_data[index1] = temp;
+#endif  //B3_USE_PLACEMENT_NEW
+	}
 
 	template <typename L>
 	void heapSort(const L& CompareFunc)
@@ -431,49 +414,48 @@ protected:
 		/* sort a[0..N-1],  N.B. 0 to N-1 */
 		int k;
 		int n = m_size;
-		for (k = n/2; k > 0; k--) 
+		for (k = n / 2; k > 0; k--)
 		{
 			downHeap(m_data, k, n, CompareFunc);
 		}
 
 		/* a[1..N] is now a heap */
-		while ( n>=1 ) 
+		while (n >= 1)
 		{
-			swap(0,n-1); /* largest of a[0..n-1] */
-
+			swap(0, n - 1); /* largest of a[0..n-1] */
 
 			n = n - 1;
 			/* restore a[1..i-1] heap */
 			downHeap(m_data, 1, n, CompareFunc);
-		} 
+		}
 	}
 
 	///non-recursive binary search, assumes sorted array
-	int	findBinarySearch(const T& key) const
+	int findBinarySearch(const T& key) const
 	{
 		int first = 0;
-		int last = size()-1;
+		int last = size() - 1;
 
 		//assume sorted array
-		while (first <= last) {
+		while (first <= last)
+		{
 			int mid = (first + last) / 2;  // compute mid point.
-			if (key > m_data[mid]) 
+			if (key > m_data[mid])
 				first = mid + 1;  // repeat search in top half.
-			else if (key < m_data[mid]) 
-				last = mid - 1; // repeat search in bottom half.
+			else if (key < m_data[mid])
+				last = mid - 1;  // repeat search in bottom half.
 			else
-				return mid;     // found it. return position /////
+				return mid;  // found it. return position /////
 		}
-		return size();    // failed to find key
+		return size();  // failed to find key
 	}
 
-
-	int	findLinearSearch(const T& key) const
+	int findLinearSearch(const T& key) const
 	{
-		int index=size();
+		int index = size();
 		int i;
 
-		for (i=0;i<size();i++)
+		for (i = 0; i < size(); i++)
 		{
 			if (m_data[i] == key)
 			{
@@ -483,36 +465,35 @@ protected:
 		}
 		return index;
 	}
-    
-    int	findLinearSearch2(const T& key) const
-    {
-        int index=-1;
-        int i;
-        
-        for (i=0;i<size();i++)
-        {
-            if (m_data[i] == key)
-            {
-                index = i;
-                break;
-            }
-        }
-        return index;
-    }
 
-	void	remove(const T& key)
+	int findLinearSearch2(const T& key) const
 	{
+		int index = -1;
+		int i;
 
-		int findIndex = findLinearSearch(key);
-		if (findIndex<size())
+		for (i = 0; i < size(); i++)
 		{
-			swap( findIndex,size()-1);
+			if (m_data[i] == key)
+			{
+				index = i;
+				break;
+			}
+		}
+		return index;
+	}
+
+	void remove(const T& key)
+	{
+		int findIndex = findLinearSearch(key);
+		if (findIndex < size())
+		{
+			swap(findIndex, size() - 1);
 			pop_back();
 		}
 	}
 
 	//PCK: whole function
-	void initializeFromBuffer(void *buffer, int size, int capacity)
+	void initializeFromBuffer(void* buffer, int size, int capacity)
 	{
 		clear();
 		m_ownsMemory = false;
@@ -524,18 +505,18 @@ protected:
 	void copyFromArray(const b3AlignedObjectArray& otherArray)
 	{
 		int otherSize = otherArray.size();
-		resize (otherSize);
+		resize(otherSize);
 		otherArray.copy(0, otherSize, m_data);
 	}
 
 	void removeAtIndex(int index)
-    {
-        if (index<size())
-        {
-            swap( index,size()-1);
-            pop_back();
-        }
-    }
+	{
+		if (index < size())
+		{
+			swap(index, size() - 1);
+			pop_back();
+		}
+	}
 };
 
-#endif //B3_OBJECT_ARRAY__
+#endif  //B3_OBJECT_ARRAY__
diff --git a/thirdparty/bullet/Bullet3Common/b3CommandLineArgs.h b/thirdparty/bullet/Bullet3Common/b3CommandLineArgs.h
index 38df8e26004..5fe4f25f8de 100644
--- a/thirdparty/bullet/Bullet3Common/b3CommandLineArgs.h
+++ b/thirdparty/bullet/Bullet3Common/b3CommandLineArgs.h
@@ -12,51 +12,54 @@
 class b3CommandLineArgs
 {
 protected:
-
 	std::map<std::string, std::string> pairs;
 
 public:
-
 	// Constructor
 	b3CommandLineArgs(int argc, char **argv)
 	{
-		addArgs(argc,argv);
+		addArgs(argc, argv);
 	}
 
-	void addArgs(int argc, char**argv)
+	void addArgs(int argc, char **argv)
 	{
-	    for (int i = 1; i < argc; i++)
-	    {
-	        std::string arg = argv[i];
+		for (int i = 1; i < argc; i++)
+		{
+			std::string arg = argv[i];
 
-			if ((arg.length() < 2) || (arg[0] != '-') || (arg[1] != '-')) {
-	        	continue;
-	        }
-
-        	std::string::size_type pos;
-		    std::string key, val;
-	        if ((pos = arg.find( '=')) == std::string::npos) {
-	        	key = std::string(arg, 2, arg.length() - 2);
-	        	val = "";
-	        } else {
-	        	key = std::string(arg, 2, pos - 2);
-	        	val = std::string(arg, pos + 1, arg.length() - 1);
-	        }
-			
-			//only add new keys, don't replace existing
-			if(pairs.find(key) == pairs.end())
+			if ((arg.length() < 2) || (arg[0] != '-') || (arg[1] != '-'))
 			{
-        		pairs[key] = val;
+				continue;
 			}
-	    }
+
+			std::string::size_type pos;
+			std::string key, val;
+			if ((pos = arg.find('=')) == std::string::npos)
+			{
+				key = std::string(arg, 2, arg.length() - 2);
+				val = "";
+			}
+			else
+			{
+				key = std::string(arg, 2, pos - 2);
+				val = std::string(arg, pos + 1, arg.length() - 1);
+			}
+
+			//only add new keys, don't replace existing
+			if (pairs.find(key) == pairs.end())
+			{
+				pairs[key] = val;
+			}
+		}
 	}
 
-	bool CheckCmdLineFlag(const char* arg_name)
+	bool CheckCmdLineFlag(const char *arg_name)
 	{
 		std::map<std::string, std::string>::iterator itr;
-		if ((itr = pairs.find(arg_name)) != pairs.end()) {
+		if ((itr = pairs.find(arg_name)) != pairs.end())
+		{
 			return true;
-	    }
+		}
 		return false;
 	}
 
@@ -73,29 +76,31 @@ template <typename T>
 inline bool b3CommandLineArgs::GetCmdLineArgument(const char *arg_name, T &val)
 {
 	std::map<std::string, std::string>::iterator itr;
-	if ((itr = pairs.find(arg_name)) != pairs.end()) {
+	if ((itr = pairs.find(arg_name)) != pairs.end())
+	{
 		std::istringstream strstream(itr->second);
 		strstream >> val;
 		return true;
-    }
-	return false;
-}
-
-template <>
-inline bool b3CommandLineArgs::GetCmdLineArgument<char*>(const char* arg_name, char* &val)
-{
-	std::map<std::string, std::string>::iterator itr;
-	if ((itr = pairs.find(arg_name)) != pairs.end()) {
-
-		std::string s = itr->second;
-		val = (char*) malloc(sizeof(char) * (s.length() + 1));
-		std::strcpy(val, s.c_str());
-		return true;
-	} else {
-    	val = NULL;
 	}
 	return false;
 }
 
+template <>
+inline bool b3CommandLineArgs::GetCmdLineArgument<char *>(const char *arg_name, char *&val)
+{
+	std::map<std::string, std::string>::iterator itr;
+	if ((itr = pairs.find(arg_name)) != pairs.end())
+	{
+		std::string s = itr->second;
+		val = (char *)malloc(sizeof(char) * (s.length() + 1));
+		std::strcpy(val, s.c_str());
+		return true;
+	}
+	else
+	{
+		val = NULL;
+	}
+	return false;
+}
 
-#endif //COMMAND_LINE_ARGS_H
+#endif  //COMMAND_LINE_ARGS_H
diff --git a/thirdparty/bullet/Bullet3Common/b3FileUtils.h b/thirdparty/bullet/Bullet3Common/b3FileUtils.h
index b5e8225cf0b..9ded17eaafa 100644
--- a/thirdparty/bullet/Bullet3Common/b3FileUtils.h
+++ b/thirdparty/bullet/Bullet3Common/b3FileUtils.h
@@ -3,7 +3,7 @@
 
 #include <stdio.h>
 #include "b3Scalar.h"
-#include <stddef.h>//ptrdiff_h
+#include <stddef.h>  //ptrdiff_h
 #include <string.h>
 
 struct b3FileUtils
@@ -17,42 +17,42 @@ struct b3FileUtils
 
 	static bool findFile(const char* orgFileName, char* relativeFileName, int maxRelativeFileNameMaxLen)
 	{
-		FILE* f=0;
-		f = fopen(orgFileName,"rb");
-                if (f)
-                {
+		FILE* f = 0;
+		f = fopen(orgFileName, "rb");
+		if (f)
+		{
 			//printf("original file found: [%s]\n", orgFileName);
-			sprintf(relativeFileName,"%s", orgFileName);
+			sprintf(relativeFileName, "%s", orgFileName);
 			fclose(f);
 			return true;
 		}
 
-		//printf("Trying various directories, relative to current working directory\n");	
-			const char* prefix[]={"./","./data/","../data/","../../data/","../../../data/","../../../../data/"};
-			int numPrefixes = sizeof(prefix)/sizeof(const char*);
-	
-			f=0;
-			bool fileFound = false;
+		//printf("Trying various directories, relative to current working directory\n");
+		const char* prefix[] = {"./", "./data/", "../data/", "../../data/", "../../../data/", "../../../../data/"};
+		int numPrefixes = sizeof(prefix) / sizeof(const char*);
 
-			for (int i=0;!f && i<numPrefixes;i++)
-			{
+		f = 0;
+		bool fileFound = false;
+
+		for (int i = 0; !f && i < numPrefixes; i++)
+		{
 #ifdef _MSC_VER
-				sprintf_s(relativeFileName,maxRelativeFileNameMaxLen,"%s%s",prefix[i],orgFileName);
+			sprintf_s(relativeFileName, maxRelativeFileNameMaxLen, "%s%s", prefix[i], orgFileName);
 #else
-				sprintf(relativeFileName,"%s%s",prefix[i],orgFileName);
+			sprintf(relativeFileName, "%s%s", prefix[i], orgFileName);
 #endif
-				f = fopen(relativeFileName,"rb");
-				if (f)
-				{
-					fileFound = true;
-					break;
-				}
-			}
+			f = fopen(relativeFileName, "rb");
 			if (f)
 			{
-				fclose(f);
+				fileFound = true;
+				break;
 			}
-	
+		}
+		if (f)
+		{
+			fclose(f);
+		}
+
 		return fileFound;
 	}
 
@@ -60,8 +60,8 @@ struct b3FileUtils
 	{
 		size_t const patlen = strlen(pattern);
 		size_t patcnt = 0;
-		const char * oriptr;
-		const char * patloc;
+		const char* oriptr;
+		const char* patloc;
 		// find how many times the pattern occurs in the original string
 		for (oriptr = name; (patloc = strstr(oriptr, pattern)); oriptr = patloc + patlen)
 		{
@@ -70,29 +70,27 @@ struct b3FileUtils
 		return oriptr;
 	}
 
-	
-
 	static int extractPath(const char* fileName, char* path, int maxPathLength)
 	{
 		const char* stripped = strip2(fileName, "/");
 		stripped = strip2(stripped, "\\");
 
-		ptrdiff_t len = stripped-fileName;
-		b3Assert((len+1)<maxPathLength);
+		ptrdiff_t len = stripped - fileName;
+		b3Assert((len + 1) < maxPathLength);
 
-		if (len && ((len+1)<maxPathLength))
+		if (len && ((len + 1) < maxPathLength))
 		{
-
-			for (int i=0;i<len;i++)
+			for (int i = 0; i < len; i++)
 			{
 				path[i] = fileName[i];
 			}
-			path[len]=0;
-		} else
+			path[len] = 0;
+		}
+		else
 		{
 			len = 0;
-			b3Assert(maxPathLength>0);
-			if (maxPathLength>0)
+			b3Assert(maxPathLength > 0);
+			if (maxPathLength > 0)
 			{
 				path[len] = 0;
 			}
@@ -102,23 +100,21 @@ struct b3FileUtils
 
 	static char toLowerChar(const char t)
 	{
-		if (t>=(char)'A' && t<=(char)'Z')
+		if (t >= (char)'A' && t <= (char)'Z')
 			return t + ((char)'a' - (char)'A');
 		else
 			return t;
 	}
 
-
 	static void toLower(char* str)
 	{
-		int len=strlen(str);
-		for (int i=0;i<len;i++)
+		int len = strlen(str);
+		for (int i = 0; i < len; i++)
 		{
 			str[i] = toLowerChar(str[i]);
 		}
 	}
 
-
 	/*static const char* strip2(const char* name, const char* pattern)
 	{
 		size_t const patlen = strlen(pattern);
@@ -133,6 +129,5 @@ struct b3FileUtils
 		return oriptr;
 	}
 	*/
-
 };
-#endif //B3_FILE_UTILS_H
+#endif  //B3_FILE_UTILS_H
diff --git a/thirdparty/bullet/Bullet3Common/b3HashMap.h b/thirdparty/bullet/Bullet3Common/b3HashMap.h
index 24a59d9baa3..3009e2cf2fd 100644
--- a/thirdparty/bullet/Bullet3Common/b3HashMap.h
+++ b/thirdparty/bullet/Bullet3Common/b3HashMap.h
@@ -13,86 +13,80 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef B3_HASH_MAP_H
 #define B3_HASH_MAP_H
 
 #include "b3AlignedObjectArray.h"
 
-
 #include <string>
 
 ///very basic hashable string implementation, compatible with b3HashMap
 struct b3HashString
 {
 	std::string m_string;
-	unsigned int	m_hash;
+	unsigned int m_hash;
 
-	B3_FORCE_INLINE	unsigned int getHash()const
+	B3_FORCE_INLINE unsigned int getHash() const
 	{
 		return m_hash;
 	}
 
-
 	b3HashString(const char* name)
-		:m_string(name)
+		: m_string(name)
 	{
-
 		/* magic numbers from http://www.isthe.com/chongo/tech/comp/fnv/ */
-		static const unsigned int  InitialFNV = 2166136261u;
+		static const unsigned int InitialFNV = 2166136261u;
 		static const unsigned int FNVMultiple = 16777619u;
 
 		/* Fowler / Noll / Vo (FNV) Hash */
 		unsigned int hash = InitialFNV;
 		int len = m_string.length();
-		for(int i = 0; i<len; i++)
+		for (int i = 0; i < len; i++)
 		{
-			hash = hash ^ (m_string[i]);       /* xor  the low 8 bits */
-			hash = hash * FNVMultiple;  /* multiply by the magic number */
+			hash = hash ^ (m_string[i]); /* xor  the low 8 bits */
+			hash = hash * FNVMultiple;   /* multiply by the magic number */
 		}
 		m_hash = hash;
 	}
 
-	int portableStringCompare(const char* src,	const char* dst) const
+	int portableStringCompare(const char* src, const char* dst) const
 	{
-			int ret = 0 ;
+		int ret = 0;
 
-			while( ! (ret = *(unsigned char *)src - *(unsigned char *)dst) && *dst)
-					++src, ++dst;
+		while (!(ret = *(unsigned char*)src - *(unsigned char*)dst) && *dst)
+			++src, ++dst;
 
-			if ( ret < 0 )
-					ret = -1 ;
-			else if ( ret > 0 )
-					ret = 1 ;
+		if (ret < 0)
+			ret = -1;
+		else if (ret > 0)
+			ret = 1;
 
-			return( ret );
+		return (ret);
 	}
 
 	bool equals(const b3HashString& other) const
 	{
 		return (m_string == other.m_string);
 	}
-
 };
 
-
-const int B3_HASH_NULL=0xffffffff;
-
+const int B3_HASH_NULL = 0xffffffff;
 
 class b3HashInt
 {
-	int	m_uid;
+	int m_uid;
+
 public:
-	b3HashInt(int uid)	:m_uid(uid)
+	b3HashInt(int uid) : m_uid(uid)
 	{
 	}
 
-	int	getUid1() const
+	int getUid1() const
 	{
 		return m_uid;
 	}
 
-	void	setUid1(int uid)
+	void setUid1(int uid)
 	{
 		m_uid = uid;
 	}
@@ -102,34 +96,34 @@ public:
 		return getUid1() == other.getUid1();
 	}
 	//to our success
-	B3_FORCE_INLINE	unsigned int getHash()const
+	B3_FORCE_INLINE unsigned int getHash() const
 	{
 		int key = m_uid;
 		// Thomas Wang's hash
-		key += ~(key << 15);	key ^=  (key >> 10);	key +=  (key << 3);	key ^=  (key >> 6);	key += ~(key << 11);	key ^=  (key >> 16);
+		key += ~(key << 15);
+		key ^= (key >> 10);
+		key += (key << 3);
+		key ^= (key >> 6);
+		key += ~(key << 11);
+		key ^= (key >> 16);
 		return key;
 	}
 };
 
-
-
 class b3HashPtr
 {
-
-	union
-	{
-		const void*	m_pointer;
-		int	m_hashValues[2];
+	union {
+		const void* m_pointer;
+		int m_hashValues[2];
 	};
 
 public:
-
 	b3HashPtr(const void* ptr)
-		:m_pointer(ptr)
+		: m_pointer(ptr)
 	{
 	}
 
-	const void*	getPointer() const
+	const void* getPointer() const
 	{
 		return m_pointer;
 	}
@@ -140,65 +134,69 @@ public:
 	}
 
 	//to our success
-	B3_FORCE_INLINE	unsigned int getHash()const
+	B3_FORCE_INLINE unsigned int getHash() const
 	{
-		const bool VOID_IS_8 = ((sizeof(void*)==8));
-		
-		int key = VOID_IS_8? m_hashValues[0]+m_hashValues[1] : m_hashValues[0];
-	
+		const bool VOID_IS_8 = ((sizeof(void*) == 8));
+
+		int key = VOID_IS_8 ? m_hashValues[0] + m_hashValues[1] : m_hashValues[0];
+
 		// Thomas Wang's hash
-		key += ~(key << 15);	key ^=  (key >> 10);	key +=  (key << 3);	key ^=  (key >> 6);	key += ~(key << 11);	key ^=  (key >> 16);
+		key += ~(key << 15);
+		key ^= (key >> 10);
+		key += (key << 3);
+		key ^= (key >> 6);
+		key += ~(key << 11);
+		key ^= (key >> 16);
 		return key;
 	}
-
-	
 };
 
-
 template <class Value>
 class b3HashKeyPtr
 {
-        int     m_uid;
+	int m_uid;
+
 public:
+	b3HashKeyPtr(int uid) : m_uid(uid)
+	{
+	}
 
-        b3HashKeyPtr(int uid)    :m_uid(uid)
-        {
-        }
+	int getUid1() const
+	{
+		return m_uid;
+	}
 
-        int     getUid1() const
-        {
-                return m_uid;
-        }
+	bool equals(const b3HashKeyPtr<Value>& other) const
+	{
+		return getUid1() == other.getUid1();
+	}
 
-        bool equals(const b3HashKeyPtr<Value>& other) const
-        {
-                return getUid1() == other.getUid1();
-        }
-
-        //to our success
-        B3_FORCE_INLINE       unsigned int getHash()const
-        {
-                int key = m_uid;
-                // Thomas Wang's hash
-                key += ~(key << 15);	key ^=  (key >> 10);	key +=  (key << 3);	key ^=  (key >> 6);	key += ~(key << 11);	key ^=  (key >> 16);
-                return key;
-        }
-
-        
+	//to our success
+	B3_FORCE_INLINE unsigned int getHash() const
+	{
+		int key = m_uid;
+		// Thomas Wang's hash
+		key += ~(key << 15);
+		key ^= (key >> 10);
+		key += (key << 3);
+		key ^= (key >> 6);
+		key += ~(key << 11);
+		key ^= (key >> 16);
+		return key;
+	}
 };
 
-
 template <class Value>
 class b3HashKey
 {
-	int	m_uid;
-public:
+	int m_uid;
 
-	b3HashKey(int uid)	:m_uid(uid)
+public:
+	b3HashKey(int uid) : m_uid(uid)
 	{
 	}
 
-	int	getUid1() const
+	int getUid1() const
 	{
 		return m_uid;
 	}
@@ -208,30 +206,33 @@ public:
 		return getUid1() == other.getUid1();
 	}
 	//to our success
-	B3_FORCE_INLINE	unsigned int getHash()const
+	B3_FORCE_INLINE unsigned int getHash() const
 	{
 		int key = m_uid;
 		// Thomas Wang's hash
-		key += ~(key << 15);	key ^=  (key >> 10);	key +=  (key << 3);	key ^=  (key >> 6);	key += ~(key << 11);	key ^=  (key >> 16);
+		key += ~(key << 15);
+		key ^= (key >> 10);
+		key += (key << 3);
+		key ^= (key >> 6);
+		key += ~(key << 11);
+		key ^= (key >> 16);
 		return key;
 	}
 };
 
-
 ///The b3HashMap template class implements a generic and lightweight hashmap.
 ///A basic sample of how to use b3HashMap is located in Demos\BasicDemo\main.cpp
 template <class Key, class Value>
 class b3HashMap
 {
-
 protected:
-	b3AlignedObjectArray<int>		m_hashTable;
-	b3AlignedObjectArray<int>		m_next;
-	
-	b3AlignedObjectArray<Value>		m_valueArray;
-	b3AlignedObjectArray<Key>		m_keyArray;
+	b3AlignedObjectArray<int> m_hashTable;
+	b3AlignedObjectArray<int> m_next;
 
-	void	growTables(const Key& /*key*/)
+	b3AlignedObjectArray<Value> m_valueArray;
+	b3AlignedObjectArray<Key> m_keyArray;
+
+	void growTables(const Key& /*key*/)
 	{
 		int newCapacity = m_valueArray.capacity();
 
@@ -245,7 +246,7 @@ protected:
 
 			int i;
 
-			for (i= 0; i < newCapacity; ++i)
+			for (i = 0; i < newCapacity; ++i)
 			{
 				m_hashTable[i] = B3_HASH_NULL;
 			}
@@ -254,30 +255,28 @@ protected:
 				m_next[i] = B3_HASH_NULL;
 			}
 
-			for(i=0;i<curHashtableSize;i++)
+			for (i = 0; i < curHashtableSize; i++)
 			{
 				//const Value& value = m_valueArray[i];
 				//const Key& key = m_keyArray[i];
 
-				int	hashValue = m_keyArray[i].getHash() & (m_valueArray.capacity()-1);	// New hash value with new mask
+				int hashValue = m_keyArray[i].getHash() & (m_valueArray.capacity() - 1);  // New hash value with new mask
 				m_next[i] = m_hashTable[hashValue];
 				m_hashTable[hashValue] = i;
 			}
-
-
 		}
 	}
 
-	public:
-
-	void insert(const Key& key, const Value& value) {
-		int hash = key.getHash() & (m_valueArray.capacity()-1);
+public:
+	void insert(const Key& key, const Value& value)
+	{
+		int hash = key.getHash() & (m_valueArray.capacity() - 1);
 
 		//replace value if the key is already there
 		int index = findIndex(key);
 		if (index != B3_HASH_NULL)
 		{
-			m_valueArray[index]=value;
+			m_valueArray[index] = value;
 			return;
 		}
 
@@ -291,19 +290,19 @@ protected:
 		{
 			growTables(key);
 			//hash with new capacity
-			hash = key.getHash() & (m_valueArray.capacity()-1);
+			hash = key.getHash() & (m_valueArray.capacity() - 1);
 		}
 		m_next[count] = m_hashTable[hash];
 		m_hashTable[hash] = count;
 	}
 
-	void remove(const Key& key) {
-
-		int hash = key.getHash() & (m_valueArray.capacity()-1);
+	void remove(const Key& key)
+	{
+		int hash = key.getHash() & (m_valueArray.capacity() - 1);
 
 		int pairIndex = findIndex(key);
-		
-		if (pairIndex ==B3_HASH_NULL)
+
+		if (pairIndex == B3_HASH_NULL)
 		{
 			return;
 		}
@@ -344,7 +343,7 @@ protected:
 		}
 
 		// Remove the last pair from the hash table.
-		int lastHash = m_keyArray[lastPairIndex].getHash() & (m_valueArray.capacity()-1);
+		int lastHash = m_keyArray[lastPairIndex].getHash() & (m_valueArray.capacity() - 1);
 
 		index = m_hashTable[lastHash];
 		b3Assert(index != B3_HASH_NULL);
@@ -376,10 +375,8 @@ protected:
 
 		m_valueArray.pop_back();
 		m_keyArray.pop_back();
-
 	}
 
-
 	int size() const
 	{
 		return m_valueArray.size();
@@ -399,23 +396,24 @@ protected:
 		return &m_valueArray[index];
 	}
 
-	 Key getKeyAtIndex(int index)
-    {
-        b3Assert(index < m_keyArray.size());
-        return m_keyArray[index];
-    }
-    
-    const Key getKeyAtIndex(int index) const
-    {
-        b3Assert(index < m_keyArray.size());
-        return m_keyArray[index];
-    }
+	Key getKeyAtIndex(int index)
+	{
+		b3Assert(index < m_keyArray.size());
+		return m_keyArray[index];
+	}
 
-	Value* operator[](const Key& key) {
+	const Key getKeyAtIndex(int index) const
+	{
+		b3Assert(index < m_keyArray.size());
+		return m_keyArray[index];
+	}
+
+	Value* operator[](const Key& key)
+	{
 		return find(key);
 	}
 
-	const Value*	find(const Key& key) const
+	const Value* find(const Key& key) const
 	{
 		int index = findIndex(key);
 		if (index == B3_HASH_NULL)
@@ -425,7 +423,7 @@ protected:
 		return &m_valueArray[index];
 	}
 
-	Value*	find(const Key& key)
+	Value* find(const Key& key)
 	{
 		int index = findIndex(key);
 		if (index == B3_HASH_NULL)
@@ -435,10 +433,9 @@ protected:
 		return &m_valueArray[index];
 	}
 
-
-	int	findIndex(const Key& key) const
+	int findIndex(const Key& key) const
 	{
-		unsigned int hash = key.getHash() & (m_valueArray.capacity()-1);
+		unsigned int hash = key.getHash() & (m_valueArray.capacity() - 1);
 
 		if (hash >= (unsigned int)m_hashTable.size())
 		{
@@ -453,14 +450,13 @@ protected:
 		return index;
 	}
 
-	void	clear()
+	void clear()
 	{
 		m_hashTable.clear();
 		m_next.clear();
 		m_valueArray.clear();
 		m_keyArray.clear();
 	}
-
 };
 
-#endif //B3_HASH_MAP_H
+#endif  //B3_HASH_MAP_H
diff --git a/thirdparty/bullet/Bullet3Common/b3Logging.cpp b/thirdparty/bullet/Bullet3Common/b3Logging.cpp
index a8e95071555..9c9f7c09ea9 100644
--- a/thirdparty/bullet/Bullet3Common/b3Logging.cpp
+++ b/thirdparty/bullet/Bullet3Common/b3Logging.cpp
@@ -20,17 +20,16 @@ subject to the following restrictions:
 
 #ifdef _WIN32
 #include <windows.h>
-#endif //_WIN32
-
+#endif  //_WIN32
 
 void b3PrintfFuncDefault(const char* msg)
 {
 #ifdef _WIN32
 	OutputDebugStringA(msg);
 #endif
-	printf("%s",msg);
-    //is this portable?
-    fflush(stdout);
+	printf("%s", msg);
+	//is this portable?
+	fflush(stdout);
 }
 
 void b3WarningMessageFuncDefault(const char* msg)
@@ -38,32 +37,26 @@ void b3WarningMessageFuncDefault(const char* msg)
 #ifdef _WIN32
 	OutputDebugStringA(msg);
 #endif
-	printf("%s",msg);
-    //is this portable?
-    fflush(stdout);
-
+	printf("%s", msg);
+	//is this portable?
+	fflush(stdout);
 }
 
-
 void b3ErrorMessageFuncDefault(const char* msg)
 {
 #ifdef _WIN32
 	OutputDebugStringA(msg);
 #endif
-	printf("%s",msg);
+	printf("%s", msg);
 
-    //is this portable?
-    fflush(stdout);
-    
+	//is this portable?
+	fflush(stdout);
 }
 
-
-
 static b3PrintfFunc* b3s_printfFunc = b3PrintfFuncDefault;
 static b3WarningMessageFunc* b3s_warningMessageFunc = b3WarningMessageFuncDefault;
 static b3ErrorMessageFunc* b3s_errorMessageFunc = b3ErrorMessageFuncDefault;
 
-
 ///The developer can route b3Printf output using their own implementation
 void b3SetCustomPrintfFunc(b3PrintfFunc* printfFunc)
 {
@@ -81,54 +74,50 @@ void b3SetCustomErrorMessageFunc(b3PrintfFunc* errorMessageFunc)
 //#define B3_MAX_DEBUG_STRING_LENGTH 2048
 #define B3_MAX_DEBUG_STRING_LENGTH 32768
 
-
-void b3OutputPrintfVarArgsInternal(const char *str, ...)
+void b3OutputPrintfVarArgsInternal(const char* str, ...)
 {
-    char strDebug[B3_MAX_DEBUG_STRING_LENGTH]={0};
-    va_list argList;
-    va_start(argList, str);
+	char strDebug[B3_MAX_DEBUG_STRING_LENGTH] = {0};
+	va_list argList;
+	va_start(argList, str);
 #ifdef _MSC_VER
-    vsprintf_s(strDebug,B3_MAX_DEBUG_STRING_LENGTH,str,argList);
+	vsprintf_s(strDebug, B3_MAX_DEBUG_STRING_LENGTH, str, argList);
 #else
-    vsnprintf(strDebug,B3_MAX_DEBUG_STRING_LENGTH,str,argList);
+	vsnprintf(strDebug, B3_MAX_DEBUG_STRING_LENGTH, str, argList);
 #endif
-        (b3s_printfFunc)(strDebug);
-    va_end(argList);    
+	(b3s_printfFunc)(strDebug);
+	va_end(argList);
 }
-void b3OutputWarningMessageVarArgsInternal(const char *str, ...)
+void b3OutputWarningMessageVarArgsInternal(const char* str, ...)
 {
-    char strDebug[B3_MAX_DEBUG_STRING_LENGTH]={0};
-    va_list argList;
-    va_start(argList, str);
+	char strDebug[B3_MAX_DEBUG_STRING_LENGTH] = {0};
+	va_list argList;
+	va_start(argList, str);
 #ifdef _MSC_VER
-    vsprintf_s(strDebug,B3_MAX_DEBUG_STRING_LENGTH,str,argList);
+	vsprintf_s(strDebug, B3_MAX_DEBUG_STRING_LENGTH, str, argList);
 #else
-    vsnprintf(strDebug,B3_MAX_DEBUG_STRING_LENGTH,str,argList);
+	vsnprintf(strDebug, B3_MAX_DEBUG_STRING_LENGTH, str, argList);
 #endif
-        (b3s_warningMessageFunc)(strDebug);
-    va_end(argList);    
+	(b3s_warningMessageFunc)(strDebug);
+	va_end(argList);
 }
-void b3OutputErrorMessageVarArgsInternal(const char *str, ...)
+void b3OutputErrorMessageVarArgsInternal(const char* str, ...)
 {
-	
-    char strDebug[B3_MAX_DEBUG_STRING_LENGTH]={0};
-    va_list argList;
-    va_start(argList, str);
+	char strDebug[B3_MAX_DEBUG_STRING_LENGTH] = {0};
+	va_list argList;
+	va_start(argList, str);
 #ifdef _MSC_VER
-    vsprintf_s(strDebug,B3_MAX_DEBUG_STRING_LENGTH,str,argList);
+	vsprintf_s(strDebug, B3_MAX_DEBUG_STRING_LENGTH, str, argList);
 #else
-    vsnprintf(strDebug,B3_MAX_DEBUG_STRING_LENGTH,str,argList);
+	vsnprintf(strDebug, B3_MAX_DEBUG_STRING_LENGTH, str, argList);
 #endif
-        (b3s_errorMessageFunc)(strDebug);
-    va_end(argList);    
-
+	(b3s_errorMessageFunc)(strDebug);
+	va_end(argList);
 }
 
-
-void	b3EnterProfileZoneDefault(const char* name)
+void b3EnterProfileZoneDefault(const char* name)
 {
 }
-void	b3LeaveProfileZoneDefault()
+void b3LeaveProfileZoneDefault()
 {
 }
 static b3EnterProfileZoneFunc* b3s_enterFunc = b3EnterProfileZoneDefault;
@@ -151,10 +140,6 @@ void b3SetCustomLeaveProfileZoneFunc(b3LeaveProfileZoneFunc* leaveFunc)
 	b3s_leaveFunc = leaveFunc;
 }
 
-
-
-
 #ifndef _MSC_VER
 #undef vsprintf_s
 #endif
-
diff --git a/thirdparty/bullet/Bullet3Common/b3Logging.h b/thirdparty/bullet/Bullet3Common/b3Logging.h
index b302effe435..9c92b12ebb9 100644
--- a/thirdparty/bullet/Bullet3Common/b3Logging.h
+++ b/thirdparty/bullet/Bullet3Common/b3Logging.h
@@ -3,75 +3,84 @@
 #define B3_LOGGING_H
 
 #ifdef __cplusplus
-extern "C" {
+extern "C"
+{
 #endif
-    
+
 ///We add the do/while so that the statement "if (condition) b3Printf("test"); else {...}" would fail
 ///You can also customize the message by uncommenting out a different line below
 #define b3Printf(...) b3OutputPrintfVarArgsInternal(__VA_ARGS__)
-//#define b3Printf(...) do {b3OutputPrintfVarArgsInternal("b3Printf[%s,%d]:",__FILE__,__LINE__);b3OutputPrintfVarArgsInternal(__VA_ARGS__); } while(0)
-//#define b3Printf b3OutputPrintfVarArgsInternal
-//#define b3Printf(...) printf(__VA_ARGS__)
-//#define b3Printf(...)
-
-#define b3Warning(...) do {b3OutputWarningMessageVarArgsInternal("b3Warning[%s,%d]:\n",__FILE__,__LINE__);b3OutputWarningMessageVarArgsInternal(__VA_ARGS__); }while(0)
-#define b3Error(...) do {b3OutputErrorMessageVarArgsInternal("b3Error[%s,%d]:\n",__FILE__,__LINE__);b3OutputErrorMessageVarArgsInternal(__VA_ARGS__); } while(0)
+	//#define b3Printf(...) do {b3OutputPrintfVarArgsInternal("b3Printf[%s,%d]:",__FILE__,__LINE__);b3OutputPrintfVarArgsInternal(__VA_ARGS__); } while(0)
+	//#define b3Printf b3OutputPrintfVarArgsInternal
+	//#define b3Printf(...) printf(__VA_ARGS__)
+	//#define b3Printf(...)
 
+#define b3Warning(...)                                                                    \
+	do                                                                                    \
+	{                                                                                     \
+		b3OutputWarningMessageVarArgsInternal("b3Warning[%s,%d]:\n", __FILE__, __LINE__); \
+		b3OutputWarningMessageVarArgsInternal(__VA_ARGS__);                               \
+	} while (0)
+#define b3Error(...)                                                                  \
+	do                                                                                \
+	{                                                                                 \
+		b3OutputErrorMessageVarArgsInternal("b3Error[%s,%d]:\n", __FILE__, __LINE__); \
+		b3OutputErrorMessageVarArgsInternal(__VA_ARGS__);                             \
+	} while (0)
 
 #ifndef B3_NO_PROFILE
 
-void b3EnterProfileZone(const char* name);
-void b3LeaveProfileZone();
+	void b3EnterProfileZone(const char* name);
+	void b3LeaveProfileZone();
 #ifdef __cplusplus
 
-class	b3ProfileZone
-{
-public:
-	b3ProfileZone(const char* name)
-	{ 
-		b3EnterProfileZone( name ); 
-	}
+	class b3ProfileZone
+	{
+	public:
+		b3ProfileZone(const char* name)
+		{
+			b3EnterProfileZone(name);
+		}
 
-	~b3ProfileZone()
-	{ 
-		b3LeaveProfileZone(); 
-	}
-};
+		~b3ProfileZone()
+		{
+			b3LeaveProfileZone();
+		}
+	};
 
-#define	B3_PROFILE( name )			b3ProfileZone __profile( name )
+#define B3_PROFILE(name) b3ProfileZone __profile(name)
 #endif
 
-#else //B3_NO_PROFILE
+#else  //B3_NO_PROFILE
 
-#define	B3_PROFILE( name )
+#define B3_PROFILE(name)
 #define b3StartProfile(a)
 #define b3StopProfile
 
-#endif //#ifndef B3_NO_PROFILE
+#endif  //#ifndef B3_NO_PROFILE
 
+	typedef void(b3PrintfFunc)(const char* msg);
+	typedef void(b3WarningMessageFunc)(const char* msg);
+	typedef void(b3ErrorMessageFunc)(const char* msg);
+	typedef void(b3EnterProfileZoneFunc)(const char* msg);
+	typedef void(b3LeaveProfileZoneFunc)();
 
-typedef void (b3PrintfFunc)(const char* msg);
-typedef void (b3WarningMessageFunc)(const char* msg);
-typedef void (b3ErrorMessageFunc)(const char* msg);
-typedef void (b3EnterProfileZoneFunc)(const char* msg);
-typedef void (b3LeaveProfileZoneFunc)();
+	///The developer can route b3Printf output using their own implementation
+	void b3SetCustomPrintfFunc(b3PrintfFunc* printfFunc);
+	void b3SetCustomWarningMessageFunc(b3WarningMessageFunc* warningMsgFunc);
+	void b3SetCustomErrorMessageFunc(b3ErrorMessageFunc* errorMsgFunc);
 
-///The developer can route b3Printf output using their own implementation
-void b3SetCustomPrintfFunc(b3PrintfFunc* printfFunc);
-void b3SetCustomWarningMessageFunc(b3WarningMessageFunc* warningMsgFunc);
-void b3SetCustomErrorMessageFunc(b3ErrorMessageFunc* errorMsgFunc);
+	///Set custom profile zone functions (zones can be nested)
+	void b3SetCustomEnterProfileZoneFunc(b3EnterProfileZoneFunc* enterFunc);
+	void b3SetCustomLeaveProfileZoneFunc(b3LeaveProfileZoneFunc* leaveFunc);
 
-///Set custom profile zone functions (zones can be nested)
-void b3SetCustomEnterProfileZoneFunc(b3EnterProfileZoneFunc* enterFunc);
-void b3SetCustomLeaveProfileZoneFunc(b3LeaveProfileZoneFunc* leaveFunc);
-
-///Don't use those internal functions directly, use the b3Printf or b3SetCustomPrintfFunc instead (or warning/error version)
-void b3OutputPrintfVarArgsInternal(const char *str, ...);
-void b3OutputWarningMessageVarArgsInternal(const char *str, ...);
-void b3OutputErrorMessageVarArgsInternal(const char *str, ...);
+	///Don't use those internal functions directly, use the b3Printf or b3SetCustomPrintfFunc instead (or warning/error version)
+	void b3OutputPrintfVarArgsInternal(const char* str, ...);
+	void b3OutputWarningMessageVarArgsInternal(const char* str, ...);
+	void b3OutputErrorMessageVarArgsInternal(const char* str, ...);
 
 #ifdef __cplusplus
-    }
+}
 #endif
 
-#endif//B3_LOGGING_H
\ No newline at end of file
+#endif  //B3_LOGGING_H
\ No newline at end of file
diff --git a/thirdparty/bullet/Bullet3Common/b3Matrix3x3.h b/thirdparty/bullet/Bullet3Common/b3Matrix3x3.h
index 89b57cf59a9..6c46536a811 100644
--- a/thirdparty/bullet/Bullet3Common/b3Matrix3x3.h
+++ b/thirdparty/bullet/Bullet3Common/b3Matrix3x3.h
@@ -12,8 +12,7 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-#ifndef	B3_MATRIX3x3_H
+#ifndef B3_MATRIX3x3_H
 #define B3_MATRIX3x3_H
 
 #include "b3Vector3.h"
@@ -32,22 +31,22 @@ const b3SimdFloat4 B3_ATTRIBUTE_ALIGNED16(b3v0010) = {0.0f, 0.0f, 1.0f, 0.0f};
 #endif
 
 #ifdef B3_USE_DOUBLE_PRECISION
-#define b3Matrix3x3Data	b3Matrix3x3DoubleData 
+#define b3Matrix3x3Data b3Matrix3x3DoubleData
 #else
-#define b3Matrix3x3Data	b3Matrix3x3FloatData
-#endif //B3_USE_DOUBLE_PRECISION
-
+#define b3Matrix3x3Data b3Matrix3x3FloatData
+#endif  //B3_USE_DOUBLE_PRECISION
 
 /**@brief The b3Matrix3x3 class implements a 3x3 rotation matrix, to perform linear algebra in combination with b3Quaternion, b3Transform and b3Vector3.
 * Make sure to only include a pure orthogonal matrix without scaling. */
-B3_ATTRIBUTE_ALIGNED16(class) b3Matrix3x3 {
-
+B3_ATTRIBUTE_ALIGNED16(class)
+b3Matrix3x3
+{
 	///Data storage for the matrix, each vector is a row of the matrix
 	b3Vector3 m_el[3];
 
 public:
 	/** @brief No initializaion constructor */
-	b3Matrix3x3 () {}
+	b3Matrix3x3() {}
 
 	//		explicit b3Matrix3x3(const b3Scalar *m) { setFromOpenGLSubMatrix(m); }
 
@@ -62,27 +61,27 @@ public:
 	*/
 	/** @brief Constructor with row major formatting */
 	b3Matrix3x3(const b3Scalar& xx, const b3Scalar& xy, const b3Scalar& xz,
-		const b3Scalar& yx, const b3Scalar& yy, const b3Scalar& yz,
-		const b3Scalar& zx, const b3Scalar& zy, const b3Scalar& zz)
-	{ 
-		setValue(xx, xy, xz, 
-			yx, yy, yz, 
-			zx, zy, zz);
+				const b3Scalar& yx, const b3Scalar& yy, const b3Scalar& yz,
+				const b3Scalar& zx, const b3Scalar& zy, const b3Scalar& zz)
+	{
+		setValue(xx, xy, xz,
+				 yx, yy, yz,
+				 zx, zy, zz);
 	}
 
-#if (defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE))|| defined (B3_USE_NEON)
-	B3_FORCE_INLINE b3Matrix3x3 (const b3SimdFloat4 v0, const b3SimdFloat4 v1, const b3SimdFloat4 v2 ) 
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)) || defined(B3_USE_NEON)
+	B3_FORCE_INLINE b3Matrix3x3(const b3SimdFloat4 v0, const b3SimdFloat4 v1, const b3SimdFloat4 v2)
 	{
-        m_el[0].mVec128 = v0;
-        m_el[1].mVec128 = v1;
-        m_el[2].mVec128 = v2;
+		m_el[0].mVec128 = v0;
+		m_el[1].mVec128 = v1;
+		m_el[2].mVec128 = v2;
 	}
 
-	B3_FORCE_INLINE b3Matrix3x3 (const b3Vector3& v0, const b3Vector3& v1, const b3Vector3& v2 ) 
+	B3_FORCE_INLINE b3Matrix3x3(const b3Vector3& v0, const b3Vector3& v1, const b3Vector3& v2)
 	{
-        m_el[0] = v0;
-        m_el[1] = v1;
-        m_el[2] = v2;
+		m_el[0] = v0;
+		m_el[1] = v1;
+		m_el[2] = v2;
 	}
 
 	// Copy constructor
@@ -94,25 +93,25 @@ public:
 	}
 
 	// Assignment Operator
-	B3_FORCE_INLINE b3Matrix3x3& operator=(const b3Matrix3x3& m) 
+	B3_FORCE_INLINE b3Matrix3x3& operator=(const b3Matrix3x3& m)
 	{
 		m_el[0].mVec128 = m.m_el[0].mVec128;
 		m_el[1].mVec128 = m.m_el[1].mVec128;
 		m_el[2].mVec128 = m.m_el[2].mVec128;
-		
+
 		return *this;
 	}
 
 #else
 
 	/** @brief Copy constructor */
-	B3_FORCE_INLINE b3Matrix3x3 (const b3Matrix3x3& other)
+	B3_FORCE_INLINE b3Matrix3x3(const b3Matrix3x3& other)
 	{
 		m_el[0] = other.m_el[0];
 		m_el[1] = other.m_el[1];
 		m_el[2] = other.m_el[2];
 	}
-    
+
 	/** @brief Assignment Operator */
 	B3_FORCE_INLINE b3Matrix3x3& operator=(const b3Matrix3x3& other)
 	{
@@ -128,10 +127,9 @@ public:
 	*  @param i Column number 0 indexed */
 	B3_FORCE_INLINE b3Vector3 getColumn(int i) const
 	{
-		return b3MakeVector3(m_el[0][i],m_el[1][i],m_el[2][i]);
+		return b3MakeVector3(m_el[0][i], m_el[1][i], m_el[2][i]);
 	}
 
-
 	/** @brief Get a row of the matrix as a vector 
 	*  @param i Row number 0 indexed */
 	B3_FORCE_INLINE const b3Vector3& getRow(int i) const
@@ -142,10 +140,10 @@ public:
 
 	/** @brief Get a mutable reference to a row of the matrix as a vector 
 	*  @param i Row number 0 indexed */
-	B3_FORCE_INLINE b3Vector3&  operator[](int i)
-	{ 
+	B3_FORCE_INLINE b3Vector3& operator[](int i)
+	{
 		b3FullAssert(0 <= i && i < 3);
-		return m_el[i]; 
+		return m_el[i];
 	}
 
 	/** @brief Get a const reference to a row of the matrix as a vector 
@@ -153,32 +151,31 @@ public:
 	B3_FORCE_INLINE const b3Vector3& operator[](int i) const
 	{
 		b3FullAssert(0 <= i && i < 3);
-		return m_el[i]; 
+		return m_el[i];
 	}
 
 	/** @brief Multiply by the target matrix on the right
 	*  @param m Rotation matrix to be applied 
 	* Equivilant to this = this * m */
-	b3Matrix3x3& operator*=(const b3Matrix3x3& m); 
+	b3Matrix3x3& operator*=(const b3Matrix3x3& m);
 
 	/** @brief Adds by the target matrix on the right
 	*  @param m matrix to be applied 
 	* Equivilant to this = this + m */
-	b3Matrix3x3& operator+=(const b3Matrix3x3& m); 
+	b3Matrix3x3& operator+=(const b3Matrix3x3& m);
 
 	/** @brief Substractss by the target matrix on the right
 	*  @param m matrix to be applied 
 	* Equivilant to this = this - m */
-	b3Matrix3x3& operator-=(const b3Matrix3x3& m); 
+	b3Matrix3x3& operator-=(const b3Matrix3x3& m);
 
 	/** @brief Set from the rotational part of a 4x4 OpenGL matrix
 	*  @param m A pointer to the beginning of the array of scalars*/
-	void setFromOpenGLSubMatrix(const b3Scalar *m)
+	void setFromOpenGLSubMatrix(const b3Scalar* m)
 	{
-		m_el[0].setValue(m[0],m[4],m[8]);
-		m_el[1].setValue(m[1],m[5],m[9]);
-		m_el[2].setValue(m[2],m[6],m[10]);
-
+		m_el[0].setValue(m[0], m[4], m[8]);
+		m_el[1].setValue(m[1], m[5], m[9]);
+		m_el[2].setValue(m[2], m[6], m[10]);
 	}
 	/** @brief Set the values of the matrix explicitly (row major)
 	*  @param xx Top left
@@ -190,93 +187,92 @@ public:
 	*  @param zx Bottom Left
 	*  @param zy Bottom Middle
 	*  @param zz Bottom Right*/
-	void setValue(const b3Scalar& xx, const b3Scalar& xy, const b3Scalar& xz, 
-		const b3Scalar& yx, const b3Scalar& yy, const b3Scalar& yz, 
-		const b3Scalar& zx, const b3Scalar& zy, const b3Scalar& zz)
+	void setValue(const b3Scalar& xx, const b3Scalar& xy, const b3Scalar& xz,
+				  const b3Scalar& yx, const b3Scalar& yy, const b3Scalar& yz,
+				  const b3Scalar& zx, const b3Scalar& zy, const b3Scalar& zz)
 	{
-		m_el[0].setValue(xx,xy,xz);
-		m_el[1].setValue(yx,yy,yz);
-		m_el[2].setValue(zx,zy,zz);
+		m_el[0].setValue(xx, xy, xz);
+		m_el[1].setValue(yx, yy, yz);
+		m_el[2].setValue(zx, zy, zz);
 	}
 
 	/** @brief Set the matrix from a quaternion
-	*  @param q The Quaternion to match */  
-	void setRotation(const b3Quaternion& q) 
+	*  @param q The Quaternion to match */
+	void setRotation(const b3Quaternion& q)
 	{
 		b3Scalar d = q.length2();
 		b3FullAssert(d != b3Scalar(0.0));
 		b3Scalar s = b3Scalar(2.0) / d;
-    
-    #if defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
-        __m128	vs, Q = q.get128();
+
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
+		__m128 vs, Q = q.get128();
 		__m128i Qi = b3CastfTo128i(Q);
-        __m128	Y, Z;
-        __m128	V1, V2, V3;
-        __m128	V11, V21, V31;
-        __m128	NQ = _mm_xor_ps(Q, b3vMzeroMask);
+		__m128 Y, Z;
+		__m128 V1, V2, V3;
+		__m128 V11, V21, V31;
+		__m128 NQ = _mm_xor_ps(Q, b3vMzeroMask);
 		__m128i NQi = b3CastfTo128i(NQ);
-        
-        V1 = b3CastiTo128f(_mm_shuffle_epi32 (Qi, B3_SHUFFLE(1,0,2,3)));	// Y X Z W
-		V2 = _mm_shuffle_ps(NQ, Q, B3_SHUFFLE(0,0,1,3));     // -X -X  Y  W
-        V3 = b3CastiTo128f(_mm_shuffle_epi32 (Qi, B3_SHUFFLE(2,1,0,3)));	// Z Y X W
-        V1 = _mm_xor_ps(V1, b3vMPPP);	//	change the sign of the first element
-			
-        V11	= b3CastiTo128f(_mm_shuffle_epi32 (Qi, B3_SHUFFLE(1,1,0,3)));	// Y Y X W
-		V21 = _mm_unpackhi_ps(Q, Q);                    //  Z  Z  W  W
-		V31 = _mm_shuffle_ps(Q, NQ, B3_SHUFFLE(0,2,0,3));	//  X  Z -X -W
 
-		V2 = V2 * V1;	//
-		V1 = V1 * V11;	//
-		V3 = V3 * V31;	//
+		V1 = b3CastiTo128f(_mm_shuffle_epi32(Qi, B3_SHUFFLE(1, 0, 2, 3)));  // Y X Z W
+		V2 = _mm_shuffle_ps(NQ, Q, B3_SHUFFLE(0, 0, 1, 3));                 // -X -X  Y  W
+		V3 = b3CastiTo128f(_mm_shuffle_epi32(Qi, B3_SHUFFLE(2, 1, 0, 3)));  // Z Y X W
+		V1 = _mm_xor_ps(V1, b3vMPPP);                                       //	change the sign of the first element
 
-        V11 = _mm_shuffle_ps(NQ, Q, B3_SHUFFLE(2,3,1,3));	//	-Z -W  Y  W
-		V11 = V11 * V21;	//
-        V21 = _mm_xor_ps(V21, b3vMPPP);	//	change the sign of the first element
-		V31 = _mm_shuffle_ps(Q, NQ, B3_SHUFFLE(3,3,1,3));	//	 W  W -Y -W
-        V31 = _mm_xor_ps(V31, b3vMPPP);	//	change the sign of the first element
-		Y = b3CastiTo128f(_mm_shuffle_epi32 (NQi, B3_SHUFFLE(3,2,0,3)));	// -W -Z -X -W
-		Z = b3CastiTo128f(_mm_shuffle_epi32 (Qi, B3_SHUFFLE(1,0,1,3)));	//  Y  X  Y  W
+		V11 = b3CastiTo128f(_mm_shuffle_epi32(Qi, B3_SHUFFLE(1, 1, 0, 3)));  // Y Y X W
+		V21 = _mm_unpackhi_ps(Q, Q);                                         //  Z  Z  W  W
+		V31 = _mm_shuffle_ps(Q, NQ, B3_SHUFFLE(0, 2, 0, 3));                 //  X  Z -X -W
+
+		V2 = V2 * V1;   //
+		V1 = V1 * V11;  //
+		V3 = V3 * V31;  //
+
+		V11 = _mm_shuffle_ps(NQ, Q, B3_SHUFFLE(2, 3, 1, 3));                //	-Z -W  Y  W
+		V11 = V11 * V21;                                                    //
+		V21 = _mm_xor_ps(V21, b3vMPPP);                                     //	change the sign of the first element
+		V31 = _mm_shuffle_ps(Q, NQ, B3_SHUFFLE(3, 3, 1, 3));                //	 W  W -Y -W
+		V31 = _mm_xor_ps(V31, b3vMPPP);                                     //	change the sign of the first element
+		Y = b3CastiTo128f(_mm_shuffle_epi32(NQi, B3_SHUFFLE(3, 2, 0, 3)));  // -W -Z -X -W
+		Z = b3CastiTo128f(_mm_shuffle_epi32(Qi, B3_SHUFFLE(1, 0, 1, 3)));   //  Y  X  Y  W
 
 		vs = _mm_load_ss(&s);
 		V21 = V21 * Y;
 		V31 = V31 * Z;
 
 		V1 = V1 + V11;
-        V2 = V2 + V21;
-        V3 = V3 + V31;
+		V2 = V2 + V21;
+		V3 = V3 + V31;
 
-        vs = b3_splat3_ps(vs, 0);
-            //	s ready
-        V1 = V1 * vs;
-        V2 = V2 * vs;
-        V3 = V3 * vs;
-        
-        V1 = V1 + b3v1000;
-        V2 = V2 + b3v0100;
-        V3 = V3 + b3v0010;
-        
-        m_el[0] = b3MakeVector3(V1); 
-        m_el[1] = b3MakeVector3(V2);
-        m_el[2] = b3MakeVector3(V3);
-    #else    
-		b3Scalar xs = q.getX() * s,   ys = q.getY() * s,   zs = q.getZ() * s;
-		b3Scalar wx = q.getW() * xs,  wy = q.getW() * ys,  wz = q.getW() * zs;
-		b3Scalar xx = q.getX() * xs,  xy = q.getX() * ys,  xz = q.getX() * zs;
-		b3Scalar yy = q.getY() * ys,  yz = q.getY() * zs,  zz = q.getZ() * zs;
+		vs = b3_splat3_ps(vs, 0);
+		//	s ready
+		V1 = V1 * vs;
+		V2 = V2 * vs;
+		V3 = V3 * vs;
+
+		V1 = V1 + b3v1000;
+		V2 = V2 + b3v0100;
+		V3 = V3 + b3v0010;
+
+		m_el[0] = b3MakeVector3(V1);
+		m_el[1] = b3MakeVector3(V2);
+		m_el[2] = b3MakeVector3(V3);
+#else
+		b3Scalar xs = q.getX() * s, ys = q.getY() * s, zs = q.getZ() * s;
+		b3Scalar wx = q.getW() * xs, wy = q.getW() * ys, wz = q.getW() * zs;
+		b3Scalar xx = q.getX() * xs, xy = q.getX() * ys, xz = q.getX() * zs;
+		b3Scalar yy = q.getY() * ys, yz = q.getY() * zs, zz = q.getZ() * zs;
 		setValue(
-            b3Scalar(1.0) - (yy + zz), xy - wz, xz + wy,
+			b3Scalar(1.0) - (yy + zz), xy - wz, xz + wy,
 			xy + wz, b3Scalar(1.0) - (xx + zz), yz - wx,
 			xz - wy, yz + wx, b3Scalar(1.0) - (xx + yy));
-	#endif
-    }
-
+#endif
+	}
 
 	/** @brief Set the matrix from euler angles using YPR around YXZ respectively
 	*  @param yaw Yaw about Y axis
 	*  @param pitch Pitch about X axis
 	*  @param roll Roll about Z axis 
 	*/
-	void setEulerYPR(const b3Scalar& yaw, const b3Scalar& pitch, const b3Scalar& roll) 
+	void setEulerYPR(const b3Scalar& yaw, const b3Scalar& pitch, const b3Scalar& roll)
 	{
 		setEulerZYX(roll, pitch, yaw);
 	}
@@ -290,182 +286,197 @@ public:
 	* angles are applied in ZYX order. I.e a vector is first rotated 
 	* about X then Y and then Z
 	**/
-	void setEulerZYX(b3Scalar eulerX,b3Scalar eulerY,b3Scalar eulerZ) { 
+	void setEulerZYX(b3Scalar eulerX, b3Scalar eulerY, b3Scalar eulerZ)
+	{
 		///@todo proposed to reverse this since it's labeled zyx but takes arguments xyz and it will match all other parts of the code
-		b3Scalar ci ( b3Cos(eulerX)); 
-		b3Scalar cj ( b3Cos(eulerY)); 
-		b3Scalar ch ( b3Cos(eulerZ)); 
-		b3Scalar si ( b3Sin(eulerX)); 
-		b3Scalar sj ( b3Sin(eulerY)); 
-		b3Scalar sh ( b3Sin(eulerZ)); 
-		b3Scalar cc = ci * ch; 
-		b3Scalar cs = ci * sh; 
-		b3Scalar sc = si * ch; 
+		b3Scalar ci(b3Cos(eulerX));
+		b3Scalar cj(b3Cos(eulerY));
+		b3Scalar ch(b3Cos(eulerZ));
+		b3Scalar si(b3Sin(eulerX));
+		b3Scalar sj(b3Sin(eulerY));
+		b3Scalar sh(b3Sin(eulerZ));
+		b3Scalar cc = ci * ch;
+		b3Scalar cs = ci * sh;
+		b3Scalar sc = si * ch;
 		b3Scalar ss = si * sh;
 
 		setValue(cj * ch, sj * sc - cs, sj * cc + ss,
-			cj * sh, sj * ss + cc, sj * cs - sc, 
-			-sj,      cj * si,      cj * ci);
+				 cj * sh, sj * ss + cc, sj * cs - sc,
+				 -sj, cj * si, cj * ci);
 	}
 
 	/**@brief Set the matrix to the identity */
 	void setIdentity()
-	{ 
-#if (defined(B3_USE_SSE_IN_API)&& defined (B3_USE_SSE)) || defined(B3_USE_NEON)
-			m_el[0] = b3MakeVector3(b3v1000); 
-			m_el[1] = b3MakeVector3(b3v0100);
-			m_el[2] = b3MakeVector3(b3v0010);
+	{
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)) || defined(B3_USE_NEON)
+		m_el[0] = b3MakeVector3(b3v1000);
+		m_el[1] = b3MakeVector3(b3v0100);
+		m_el[2] = b3MakeVector3(b3v0010);
 #else
-		setValue(b3Scalar(1.0), b3Scalar(0.0), b3Scalar(0.0), 
-			b3Scalar(0.0), b3Scalar(1.0), b3Scalar(0.0), 
-			b3Scalar(0.0), b3Scalar(0.0), b3Scalar(1.0)); 
+		setValue(b3Scalar(1.0), b3Scalar(0.0), b3Scalar(0.0),
+				 b3Scalar(0.0), b3Scalar(1.0), b3Scalar(0.0),
+				 b3Scalar(0.0), b3Scalar(0.0), b3Scalar(1.0));
 #endif
 	}
 
-	static const b3Matrix3x3&	getIdentity()
+	static const b3Matrix3x3& getIdentity()
 	{
-#if (defined(B3_USE_SSE_IN_API)&& defined (B3_USE_SSE)) || defined(B3_USE_NEON)
-        static const b3Matrix3x3 
-        identityMatrix(b3v1000, b3v0100, b3v0010);
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)) || defined(B3_USE_NEON)
+		static const b3Matrix3x3
+			identityMatrix(b3v1000, b3v0100, b3v0010);
 #else
-		static const b3Matrix3x3 
-        identityMatrix(
-            b3Scalar(1.0), b3Scalar(0.0), b3Scalar(0.0), 
-			b3Scalar(0.0), b3Scalar(1.0), b3Scalar(0.0), 
-			b3Scalar(0.0), b3Scalar(0.0), b3Scalar(1.0));
+		static const b3Matrix3x3
+			identityMatrix(
+				b3Scalar(1.0), b3Scalar(0.0), b3Scalar(0.0),
+				b3Scalar(0.0), b3Scalar(1.0), b3Scalar(0.0),
+				b3Scalar(0.0), b3Scalar(0.0), b3Scalar(1.0));
 #endif
 		return identityMatrix;
 	}
 
 	/**@brief Fill the rotational part of an OpenGL matrix and clear the shear/perspective
 	* @param m The array to be filled */
-	void getOpenGLSubMatrix(b3Scalar *m) const 
+	void getOpenGLSubMatrix(b3Scalar * m) const
 	{
-#if defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
-        __m128 v0 = m_el[0].mVec128;
-        __m128 v1 = m_el[1].mVec128;
-        __m128 v2 = m_el[2].mVec128;    //  x2 y2 z2 w2
-        __m128 *vm = (__m128 *)m;
-        __m128 vT;
-        
-        v2 = _mm_and_ps(v2, b3vFFF0fMask);  //  x2 y2 z2 0
-        
-        vT = _mm_unpackhi_ps(v0, v1);	//	z0 z1 * *
-        v0 = _mm_unpacklo_ps(v0, v1);	//	x0 x1 y0 y1
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
+		__m128 v0 = m_el[0].mVec128;
+		__m128 v1 = m_el[1].mVec128;
+		__m128 v2 = m_el[2].mVec128;  //  x2 y2 z2 w2
+		__m128* vm = (__m128*)m;
+		__m128 vT;
 
-        v1 = _mm_shuffle_ps(v0, v2, B3_SHUFFLE(2, 3, 1, 3) );	// y0 y1 y2 0
-        v0 = _mm_shuffle_ps(v0, v2, B3_SHUFFLE(0, 1, 0, 3) );	// x0 x1 x2 0
-        v2 = b3CastdTo128f(_mm_move_sd(b3CastfTo128d(v2), b3CastfTo128d(vT)));	// z0 z1 z2 0
+		v2 = _mm_and_ps(v2, b3vFFF0fMask);  //  x2 y2 z2 0
 
-        vm[0] = v0;
-        vm[1] = v1;
-        vm[2] = v2;
+		vT = _mm_unpackhi_ps(v0, v1);  //	z0 z1 * *
+		v0 = _mm_unpacklo_ps(v0, v1);  //	x0 x1 y0 y1
+
+		v1 = _mm_shuffle_ps(v0, v2, B3_SHUFFLE(2, 3, 1, 3));                    // y0 y1 y2 0
+		v0 = _mm_shuffle_ps(v0, v2, B3_SHUFFLE(0, 1, 0, 3));                    // x0 x1 x2 0
+		v2 = b3CastdTo128f(_mm_move_sd(b3CastfTo128d(v2), b3CastfTo128d(vT)));  // z0 z1 z2 0
+
+		vm[0] = v0;
+		vm[1] = v1;
+		vm[2] = v2;
 #elif defined(B3_USE_NEON)
-        // note: zeros the w channel. We can preserve it at the cost of two more vtrn instructions.
-        static const uint32x2_t zMask = (const uint32x2_t) {-1, 0 };
-        float32x4_t *vm = (float32x4_t *)m;
-        float32x4x2_t top = vtrnq_f32( m_el[0].mVec128, m_el[1].mVec128 );  // {x0 x1 z0 z1}, {y0 y1 w0 w1}
-        float32x2x2_t bl = vtrn_f32( vget_low_f32(m_el[2].mVec128), vdup_n_f32(0.0f) );       // {x2  0 }, {y2 0}
-        float32x4_t v0 = vcombine_f32( vget_low_f32(top.val[0]), bl.val[0] );
-        float32x4_t v1 = vcombine_f32( vget_low_f32(top.val[1]), bl.val[1] );
-        float32x2_t q = (float32x2_t) vand_u32( (uint32x2_t) vget_high_f32( m_el[2].mVec128), zMask );
-        float32x4_t v2 = vcombine_f32( vget_high_f32(top.val[0]), q );       // z0 z1 z2  0
+		// note: zeros the w channel. We can preserve it at the cost of two more vtrn instructions.
+		static const uint32x2_t zMask = (const uint32x2_t){-1, 0};
+		float32x4_t* vm = (float32x4_t*)m;
+		float32x4x2_t top = vtrnq_f32(m_el[0].mVec128, m_el[1].mVec128);               // {x0 x1 z0 z1}, {y0 y1 w0 w1}
+		float32x2x2_t bl = vtrn_f32(vget_low_f32(m_el[2].mVec128), vdup_n_f32(0.0f));  // {x2  0 }, {y2 0}
+		float32x4_t v0 = vcombine_f32(vget_low_f32(top.val[0]), bl.val[0]);
+		float32x4_t v1 = vcombine_f32(vget_low_f32(top.val[1]), bl.val[1]);
+		float32x2_t q = (float32x2_t)vand_u32((uint32x2_t)vget_high_f32(m_el[2].mVec128), zMask);
+		float32x4_t v2 = vcombine_f32(vget_high_f32(top.val[0]), q);  // z0 z1 z2  0
 
-        vm[0] = v0;
-        vm[1] = v1;
-        vm[2] = v2;
+		vm[0] = v0;
+		vm[1] = v1;
+		vm[2] = v2;
 #else
-		m[0]  = b3Scalar(m_el[0].getX()); 
-		m[1]  = b3Scalar(m_el[1].getX());
-		m[2]  = b3Scalar(m_el[2].getX());
-		m[3]  = b3Scalar(0.0); 
-		m[4]  = b3Scalar(m_el[0].getY());
-		m[5]  = b3Scalar(m_el[1].getY());
-		m[6]  = b3Scalar(m_el[2].getY());
-		m[7]  = b3Scalar(0.0); 
-		m[8]  = b3Scalar(m_el[0].getZ()); 
-		m[9]  = b3Scalar(m_el[1].getZ());
+		m[0] = b3Scalar(m_el[0].getX());
+		m[1] = b3Scalar(m_el[1].getX());
+		m[2] = b3Scalar(m_el[2].getX());
+		m[3] = b3Scalar(0.0);
+		m[4] = b3Scalar(m_el[0].getY());
+		m[5] = b3Scalar(m_el[1].getY());
+		m[6] = b3Scalar(m_el[2].getY());
+		m[7] = b3Scalar(0.0);
+		m[8] = b3Scalar(m_el[0].getZ());
+		m[9] = b3Scalar(m_el[1].getZ());
 		m[10] = b3Scalar(m_el[2].getZ());
-		m[11] = b3Scalar(0.0); 
+		m[11] = b3Scalar(0.0);
 #endif
 	}
 
 	/**@brief Get the matrix represented as a quaternion 
 	* @param q The quaternion which will be set */
-	void getRotation(b3Quaternion& q) const
+	void getRotation(b3Quaternion & q) const
 	{
-#if (defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE))|| defined (B3_USE_NEON)
-        b3Scalar trace = m_el[0].getX() + m_el[1].getY() + m_el[2].getZ();
-        b3Scalar s, x;
-        
-        union {
-            b3SimdFloat4 vec;
-            b3Scalar f[4];
-        } temp;
-        
-        if (trace > b3Scalar(0.0)) 
-        {
-            x = trace + b3Scalar(1.0);
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)) || defined(B3_USE_NEON)
+		b3Scalar trace = m_el[0].getX() + m_el[1].getY() + m_el[2].getZ();
+		b3Scalar s, x;
 
-            temp.f[0]=m_el[2].getY() - m_el[1].getZ();
-            temp.f[1]=m_el[0].getZ() - m_el[2].getX();
-            temp.f[2]=m_el[1].getX() - m_el[0].getY();
-            temp.f[3]=x;
-            //temp.f[3]= s * b3Scalar(0.5);
-        } 
-        else 
-        {
-            int i, j, k;
-            if(m_el[0].getX() < m_el[1].getY()) 
-            { 
-                if( m_el[1].getY() < m_el[2].getZ() )
-                    { i = 2; j = 0; k = 1; }
-                else
-                    { i = 1; j = 2; k = 0; }
-            }
-            else
-            {
-                if( m_el[0].getX() < m_el[2].getZ())
-                    { i = 2; j = 0; k = 1; }
-                else
-                    { i = 0; j = 1; k = 2; }
-            }
+		union {
+			b3SimdFloat4 vec;
+			b3Scalar f[4];
+		} temp;
 
-            x = m_el[i][i] - m_el[j][j] - m_el[k][k] + b3Scalar(1.0);
+		if (trace > b3Scalar(0.0))
+		{
+			x = trace + b3Scalar(1.0);
 
-            temp.f[3] = (m_el[k][j] - m_el[j][k]);
-            temp.f[j] = (m_el[j][i] + m_el[i][j]);
-            temp.f[k] = (m_el[k][i] + m_el[i][k]);
-            temp.f[i] = x;
-            //temp.f[i] = s * b3Scalar(0.5);
-        }
+			temp.f[0] = m_el[2].getY() - m_el[1].getZ();
+			temp.f[1] = m_el[0].getZ() - m_el[2].getX();
+			temp.f[2] = m_el[1].getX() - m_el[0].getY();
+			temp.f[3] = x;
+			//temp.f[3]= s * b3Scalar(0.5);
+		}
+		else
+		{
+			int i, j, k;
+			if (m_el[0].getX() < m_el[1].getY())
+			{
+				if (m_el[1].getY() < m_el[2].getZ())
+				{
+					i = 2;
+					j = 0;
+					k = 1;
+				}
+				else
+				{
+					i = 1;
+					j = 2;
+					k = 0;
+				}
+			}
+			else
+			{
+				if (m_el[0].getX() < m_el[2].getZ())
+				{
+					i = 2;
+					j = 0;
+					k = 1;
+				}
+				else
+				{
+					i = 0;
+					j = 1;
+					k = 2;
+				}
+			}
 
-        s = b3Sqrt(x);
-        q.set128(temp.vec);
-        s = b3Scalar(0.5) / s;
+			x = m_el[i][i] - m_el[j][j] - m_el[k][k] + b3Scalar(1.0);
 
-        q *= s;
-#else    
+			temp.f[3] = (m_el[k][j] - m_el[j][k]);
+			temp.f[j] = (m_el[j][i] + m_el[i][j]);
+			temp.f[k] = (m_el[k][i] + m_el[i][k]);
+			temp.f[i] = x;
+			//temp.f[i] = s * b3Scalar(0.5);
+		}
+
+		s = b3Sqrt(x);
+		q.set128(temp.vec);
+		s = b3Scalar(0.5) / s;
+
+		q *= s;
+#else
 		b3Scalar trace = m_el[0].getX() + m_el[1].getY() + m_el[2].getZ();
 
 		b3Scalar temp[4];
 
-		if (trace > b3Scalar(0.0)) 
+		if (trace > b3Scalar(0.0))
 		{
 			b3Scalar s = b3Sqrt(trace + b3Scalar(1.0));
-			temp[3]=(s * b3Scalar(0.5));
+			temp[3] = (s * b3Scalar(0.5));
 			s = b3Scalar(0.5) / s;
 
-			temp[0]=((m_el[2].getY() - m_el[1].getZ()) * s);
-			temp[1]=((m_el[0].getZ() - m_el[2].getX()) * s);
-			temp[2]=((m_el[1].getX() - m_el[0].getY()) * s);
-		} 
-		else 
+			temp[0] = ((m_el[2].getY() - m_el[1].getZ()) * s);
+			temp[1] = ((m_el[0].getZ() - m_el[2].getX()) * s);
+			temp[2] = ((m_el[1].getX() - m_el[0].getY()) * s);
+		}
+		else
 		{
-			int i = m_el[0].getX() < m_el[1].getY() ? 
-				(m_el[1].getY() < m_el[2].getZ() ? 2 : 1) :
-				(m_el[0].getX() < m_el[2].getZ() ? 2 : 0); 
-			int j = (i + 1) % 3;  
+			int i = m_el[0].getX() < m_el[1].getY() ? (m_el[1].getY() < m_el[2].getZ() ? 2 : 1) : (m_el[0].getX() < m_el[2].getZ() ? 2 : 0);
+			int j = (i + 1) % 3;
 			int k = (i + 2) % 3;
 
 			b3Scalar s = b3Sqrt(m_el[i][i] - m_el[j][j] - m_el[k][k] + b3Scalar(1.0));
@@ -476,44 +487,42 @@ public:
 			temp[j] = (m_el[j][i] + m_el[i][j]) * s;
 			temp[k] = (m_el[k][i] + m_el[i][k]) * s;
 		}
-		q.setValue(temp[0],temp[1],temp[2],temp[3]);
+		q.setValue(temp[0], temp[1], temp[2], temp[3]);
 #endif
 	}
 
 	/**@brief Get the matrix represented as euler angles around YXZ, roundtrip with setEulerYPR
 	* @param yaw Yaw around Y axis
 	* @param pitch Pitch around X axis
-	* @param roll around Z axis */	
-	void getEulerYPR(b3Scalar& yaw, b3Scalar& pitch, b3Scalar& roll) const
+	* @param roll around Z axis */
+	void getEulerYPR(b3Scalar & yaw, b3Scalar & pitch, b3Scalar & roll) const
 	{
-
 		// first use the normal calculus
 		yaw = b3Scalar(b3Atan2(m_el[1].getX(), m_el[0].getX()));
 		pitch = b3Scalar(b3Asin(-m_el[2].getX()));
 		roll = b3Scalar(b3Atan2(m_el[2].getY(), m_el[2].getZ()));
 
 		// on pitch = +/-HalfPI
-		if (b3Fabs(pitch)==B3_HALF_PI)
+		if (b3Fabs(pitch) == B3_HALF_PI)
 		{
-			if (yaw>0)
-				yaw-=B3_PI;
+			if (yaw > 0)
+				yaw -= B3_PI;
 			else
-				yaw+=B3_PI;
+				yaw += B3_PI;
 
-			if (roll>0)
-				roll-=B3_PI;
+			if (roll > 0)
+				roll -= B3_PI;
 			else
-				roll+=B3_PI;
+				roll += B3_PI;
 		}
 	};
 
-
 	/**@brief Get the matrix represented as euler angles around ZYX
 	* @param yaw Yaw around X axis
 	* @param pitch Pitch around Y axis
 	* @param roll around X axis 
-	* @param solution_number Which solution of two possible solutions ( 1 or 2) are possible values*/	
-	void getEulerZYX(b3Scalar& yaw, b3Scalar& pitch, b3Scalar& roll, unsigned int solution_number = 1) const
+	* @param solution_number Which solution of two possible solutions ( 1 or 2) are possible values*/
+	void getEulerZYX(b3Scalar & yaw, b3Scalar & pitch, b3Scalar & roll, unsigned int solution_number = 1) const
 	{
 		struct Euler
 		{
@@ -523,7 +532,7 @@ public:
 		};
 
 		Euler euler_out;
-		Euler euler_out2; //second solution
+		Euler euler_out2;  //second solution
 		//get the pointer to the raw data
 
 		// Check that pitch is not at a singularity
@@ -533,7 +542,7 @@ public:
 			euler_out2.yaw = 0;
 
 			// From difference of angles formula
-			b3Scalar delta = b3Atan2(m_el[0].getX(),m_el[0].getZ());
+			b3Scalar delta = b3Atan2(m_el[0].getX(), m_el[0].getZ());
 			if (m_el[2].getX() > 0)  //gimbal locked up
 			{
 				euler_out.pitch = B3_PI / b3Scalar(2.0);
@@ -541,7 +550,7 @@ public:
 				euler_out.roll = euler_out.pitch + delta;
 				euler_out2.roll = euler_out.pitch + delta;
 			}
-			else // gimbal locked down
+			else  // gimbal locked down
 			{
 				euler_out.pitch = -B3_PI / b3Scalar(2.0);
 				euler_out2.pitch = -B3_PI / b3Scalar(2.0);
@@ -551,29 +560,29 @@ public:
 		}
 		else
 		{
-			euler_out.pitch = - b3Asin(m_el[2].getX());
+			euler_out.pitch = -b3Asin(m_el[2].getX());
 			euler_out2.pitch = B3_PI - euler_out.pitch;
 
-			euler_out.roll = b3Atan2(m_el[2].getY()/b3Cos(euler_out.pitch), 
-				m_el[2].getZ()/b3Cos(euler_out.pitch));
-			euler_out2.roll = b3Atan2(m_el[2].getY()/b3Cos(euler_out2.pitch), 
-				m_el[2].getZ()/b3Cos(euler_out2.pitch));
+			euler_out.roll = b3Atan2(m_el[2].getY() / b3Cos(euler_out.pitch),
+									 m_el[2].getZ() / b3Cos(euler_out.pitch));
+			euler_out2.roll = b3Atan2(m_el[2].getY() / b3Cos(euler_out2.pitch),
+									  m_el[2].getZ() / b3Cos(euler_out2.pitch));
 
-			euler_out.yaw = b3Atan2(m_el[1].getX()/b3Cos(euler_out.pitch), 
-				m_el[0].getX()/b3Cos(euler_out.pitch));
-			euler_out2.yaw = b3Atan2(m_el[1].getX()/b3Cos(euler_out2.pitch), 
-				m_el[0].getX()/b3Cos(euler_out2.pitch));
+			euler_out.yaw = b3Atan2(m_el[1].getX() / b3Cos(euler_out.pitch),
+									m_el[0].getX() / b3Cos(euler_out.pitch));
+			euler_out2.yaw = b3Atan2(m_el[1].getX() / b3Cos(euler_out2.pitch),
+									 m_el[0].getX() / b3Cos(euler_out2.pitch));
 		}
 
 		if (solution_number == 1)
-		{ 
-			yaw = euler_out.yaw; 
+		{
+			yaw = euler_out.yaw;
 			pitch = euler_out.pitch;
 			roll = euler_out.roll;
 		}
 		else
-		{ 
-			yaw = euler_out2.yaw; 
+		{
+			yaw = euler_out2.yaw;
 			pitch = euler_out2.pitch;
 			roll = euler_out2.roll;
 		}
@@ -584,18 +593,18 @@ public:
 
 	b3Matrix3x3 scaled(const b3Vector3& s) const
 	{
-#if (defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE))|| defined (B3_USE_NEON)
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)) || defined(B3_USE_NEON)
 		return b3Matrix3x3(m_el[0] * s, m_el[1] * s, m_el[2] * s);
-#else		
+#else
 		return b3Matrix3x3(
-            m_el[0].getX() * s.getX(), m_el[0].getY() * s.getY(), m_el[0].getZ() * s.getZ(),
+			m_el[0].getX() * s.getX(), m_el[0].getY() * s.getY(), m_el[0].getZ() * s.getZ(),
 			m_el[1].getX() * s.getX(), m_el[1].getY() * s.getY(), m_el[1].getZ() * s.getZ(),
 			m_el[2].getX() * s.getX(), m_el[2].getY() * s.getY(), m_el[2].getZ() * s.getZ());
 #endif
 	}
 
 	/**@brief Return the determinant of the matrix */
-	b3Scalar            determinant() const;
+	b3Scalar determinant() const;
 	/**@brief Return the adjoint of the matrix */
 	b3Matrix3x3 adjoint() const;
 	/**@brief Return the matrix with all values non negative */
@@ -603,25 +612,24 @@ public:
 	/**@brief Return the transpose of the matrix */
 	b3Matrix3x3 transpose() const;
 	/**@brief Return the inverse of the matrix */
-	b3Matrix3x3 inverse() const; 
+	b3Matrix3x3 inverse() const;
 
 	b3Matrix3x3 transposeTimes(const b3Matrix3x3& m) const;
 	b3Matrix3x3 timesTranspose(const b3Matrix3x3& m) const;
 
-	B3_FORCE_INLINE b3Scalar tdotx(const b3Vector3& v) const 
+	B3_FORCE_INLINE b3Scalar tdotx(const b3Vector3& v) const
 	{
 		return m_el[0].getX() * v.getX() + m_el[1].getX() * v.getY() + m_el[2].getX() * v.getZ();
 	}
-	B3_FORCE_INLINE b3Scalar tdoty(const b3Vector3& v) const 
+	B3_FORCE_INLINE b3Scalar tdoty(const b3Vector3& v) const
 	{
 		return m_el[0].getY() * v.getX() + m_el[1].getY() * v.getY() + m_el[2].getY() * v.getZ();
 	}
-	B3_FORCE_INLINE b3Scalar tdotz(const b3Vector3& v) const 
+	B3_FORCE_INLINE b3Scalar tdotz(const b3Vector3& v) const
 	{
 		return m_el[0].getZ() * v.getX() + m_el[1].getZ() * v.getY() + m_el[2].getZ() * v.getZ();
 	}
 
-
 	/**@brief diagonalizes this matrix by the Jacobi method.
 	* @param rot stores the rotation from the coordinate system in which the matrix is diagonal to the original
 	* coordinate system, i.e., old_this = rot * new_this * rot^T. 
@@ -631,7 +639,7 @@ public:
 	* 
 	* Note that this matrix is assumed to be symmetric. 
 	*/
-	void diagonalize(b3Matrix3x3& rot, b3Scalar threshold, int maxSteps)
+	void diagonalize(b3Matrix3x3 & rot, b3Scalar threshold, int maxSteps)
 	{
 		rot.setIdentity();
 		for (int step = maxSteps; step > 0; step--)
@@ -667,7 +675,7 @@ public:
 				step = 1;
 			}
 
-			// compute Jacobi rotation J which leads to a zero for element [p][q] 
+			// compute Jacobi rotation J which leads to a zero for element [p][q]
 			b3Scalar mpq = m_el[p][q];
 			b3Scalar theta = (m_el[q][q] - m_el[p][p]) / (2 * mpq);
 			b3Scalar theta2 = theta * theta;
@@ -676,7 +684,7 @@ public:
 			if (theta2 * theta2 < b3Scalar(10 / B3_EPSILON))
 			{
 				t = (theta >= 0) ? 1 / (theta + b3Sqrt(1 + theta2))
-					: 1 / (theta - b3Sqrt(1 + theta2));
+								 : 1 / (theta - b3Sqrt(1 + theta2));
 				cos = 1 / b3Sqrt(1 + t * t);
 				sin = cos * t;
 			}
@@ -709,9 +717,6 @@ public:
 		}
 	}
 
-
-
-
 	/**@brief Calculate the matrix cofactor 
 	* @param r1 The first row to use for calculating the cofactor
 	* @param c1 The first column to use for calculating the cofactor
@@ -719,304 +724,298 @@ public:
 	* @param c1 The second column to use for calculating the cofactor
 	* See http://en.wikipedia.org/wiki/Cofactor_(linear_algebra) for more details
 	*/
-	b3Scalar cofac(int r1, int c1, int r2, int c2) const 
+	b3Scalar cofac(int r1, int c1, int r2, int c2) const
 	{
 		return m_el[r1][c1] * m_el[r2][c2] - m_el[r1][c2] * m_el[r2][c1];
 	}
 
-	void	serialize(struct	b3Matrix3x3Data& dataOut) const;
+	void serialize(struct b3Matrix3x3Data & dataOut) const;
 
-	void	serializeFloat(struct	b3Matrix3x3FloatData& dataOut) const;
+	void serializeFloat(struct b3Matrix3x3FloatData & dataOut) const;
 
-	void	deSerialize(const struct	b3Matrix3x3Data& dataIn);
+	void deSerialize(const struct b3Matrix3x3Data& dataIn);
 
-	void	deSerializeFloat(const struct	b3Matrix3x3FloatData& dataIn);
-
-	void	deSerializeDouble(const struct	b3Matrix3x3DoubleData& dataIn);
+	void deSerializeFloat(const struct b3Matrix3x3FloatData& dataIn);
 
+	void deSerializeDouble(const struct b3Matrix3x3DoubleData& dataIn);
 };
 
-
-B3_FORCE_INLINE b3Matrix3x3& 
+B3_FORCE_INLINE b3Matrix3x3&
 b3Matrix3x3::operator*=(const b3Matrix3x3& m)
 {
-#if defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
-    __m128 rv00, rv01, rv02;
-    __m128 rv10, rv11, rv12;
-    __m128 rv20, rv21, rv22;
-    __m128 mv0, mv1, mv2;
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
+	__m128 rv00, rv01, rv02;
+	__m128 rv10, rv11, rv12;
+	__m128 rv20, rv21, rv22;
+	__m128 mv0, mv1, mv2;
 
-    rv02 = m_el[0].mVec128;
-    rv12 = m_el[1].mVec128;
-    rv22 = m_el[2].mVec128;
+	rv02 = m_el[0].mVec128;
+	rv12 = m_el[1].mVec128;
+	rv22 = m_el[2].mVec128;
 
-    mv0 = _mm_and_ps(m[0].mVec128, b3vFFF0fMask); 
-    mv1 = _mm_and_ps(m[1].mVec128, b3vFFF0fMask); 
-    mv2 = _mm_and_ps(m[2].mVec128, b3vFFF0fMask); 
-    
-    // rv0
-    rv00 = b3_splat_ps(rv02, 0);
-    rv01 = b3_splat_ps(rv02, 1);
-    rv02 = b3_splat_ps(rv02, 2);
-    
-    rv00 = _mm_mul_ps(rv00, mv0);
-    rv01 = _mm_mul_ps(rv01, mv1);
-    rv02 = _mm_mul_ps(rv02, mv2);
-    
-    // rv1
-    rv10 = b3_splat_ps(rv12, 0);
-    rv11 = b3_splat_ps(rv12, 1);
-    rv12 = b3_splat_ps(rv12, 2);
-    
-    rv10 = _mm_mul_ps(rv10, mv0);
-    rv11 = _mm_mul_ps(rv11, mv1);
-    rv12 = _mm_mul_ps(rv12, mv2);
-    
-    // rv2
-    rv20 = b3_splat_ps(rv22, 0);
-    rv21 = b3_splat_ps(rv22, 1);
-    rv22 = b3_splat_ps(rv22, 2);
-    
-    rv20 = _mm_mul_ps(rv20, mv0);
-    rv21 = _mm_mul_ps(rv21, mv1);
-    rv22 = _mm_mul_ps(rv22, mv2);
+	mv0 = _mm_and_ps(m[0].mVec128, b3vFFF0fMask);
+	mv1 = _mm_and_ps(m[1].mVec128, b3vFFF0fMask);
+	mv2 = _mm_and_ps(m[2].mVec128, b3vFFF0fMask);
 
-    rv00 = _mm_add_ps(rv00, rv01);
-    rv10 = _mm_add_ps(rv10, rv11);
-    rv20 = _mm_add_ps(rv20, rv21);
+	// rv0
+	rv00 = b3_splat_ps(rv02, 0);
+	rv01 = b3_splat_ps(rv02, 1);
+	rv02 = b3_splat_ps(rv02, 2);
 
-    m_el[0].mVec128 = _mm_add_ps(rv00, rv02);
-    m_el[1].mVec128 = _mm_add_ps(rv10, rv12);
-    m_el[2].mVec128 = _mm_add_ps(rv20, rv22);
+	rv00 = _mm_mul_ps(rv00, mv0);
+	rv01 = _mm_mul_ps(rv01, mv1);
+	rv02 = _mm_mul_ps(rv02, mv2);
+
+	// rv1
+	rv10 = b3_splat_ps(rv12, 0);
+	rv11 = b3_splat_ps(rv12, 1);
+	rv12 = b3_splat_ps(rv12, 2);
+
+	rv10 = _mm_mul_ps(rv10, mv0);
+	rv11 = _mm_mul_ps(rv11, mv1);
+	rv12 = _mm_mul_ps(rv12, mv2);
+
+	// rv2
+	rv20 = b3_splat_ps(rv22, 0);
+	rv21 = b3_splat_ps(rv22, 1);
+	rv22 = b3_splat_ps(rv22, 2);
+
+	rv20 = _mm_mul_ps(rv20, mv0);
+	rv21 = _mm_mul_ps(rv21, mv1);
+	rv22 = _mm_mul_ps(rv22, mv2);
+
+	rv00 = _mm_add_ps(rv00, rv01);
+	rv10 = _mm_add_ps(rv10, rv11);
+	rv20 = _mm_add_ps(rv20, rv21);
+
+	m_el[0].mVec128 = _mm_add_ps(rv00, rv02);
+	m_el[1].mVec128 = _mm_add_ps(rv10, rv12);
+	m_el[2].mVec128 = _mm_add_ps(rv20, rv22);
 
 #elif defined(B3_USE_NEON)
 
-    float32x4_t rv0, rv1, rv2;
-    float32x4_t v0, v1, v2;
-    float32x4_t mv0, mv1, mv2;
+	float32x4_t rv0, rv1, rv2;
+	float32x4_t v0, v1, v2;
+	float32x4_t mv0, mv1, mv2;
 
-    v0 = m_el[0].mVec128;
-    v1 = m_el[1].mVec128;
-    v2 = m_el[2].mVec128;
+	v0 = m_el[0].mVec128;
+	v1 = m_el[1].mVec128;
+	v2 = m_el[2].mVec128;
 
-    mv0 = (float32x4_t) vandq_s32((int32x4_t)m[0].mVec128, b3vFFF0Mask); 
-    mv1 = (float32x4_t) vandq_s32((int32x4_t)m[1].mVec128, b3vFFF0Mask); 
-    mv2 = (float32x4_t) vandq_s32((int32x4_t)m[2].mVec128, b3vFFF0Mask); 
-    
-    rv0 = vmulq_lane_f32(mv0, vget_low_f32(v0), 0);
-    rv1 = vmulq_lane_f32(mv0, vget_low_f32(v1), 0);
-    rv2 = vmulq_lane_f32(mv0, vget_low_f32(v2), 0);
-    
-    rv0 = vmlaq_lane_f32(rv0, mv1, vget_low_f32(v0), 1);
-    rv1 = vmlaq_lane_f32(rv1, mv1, vget_low_f32(v1), 1);
-    rv2 = vmlaq_lane_f32(rv2, mv1, vget_low_f32(v2), 1);
-    
-    rv0 = vmlaq_lane_f32(rv0, mv2, vget_high_f32(v0), 0);
-    rv1 = vmlaq_lane_f32(rv1, mv2, vget_high_f32(v1), 0);
-    rv2 = vmlaq_lane_f32(rv2, mv2, vget_high_f32(v2), 0);
+	mv0 = (float32x4_t)vandq_s32((int32x4_t)m[0].mVec128, b3vFFF0Mask);
+	mv1 = (float32x4_t)vandq_s32((int32x4_t)m[1].mVec128, b3vFFF0Mask);
+	mv2 = (float32x4_t)vandq_s32((int32x4_t)m[2].mVec128, b3vFFF0Mask);
 
-    m_el[0].mVec128 = rv0;
-    m_el[1].mVec128 = rv1;
-    m_el[2].mVec128 = rv2;
-#else    
+	rv0 = vmulq_lane_f32(mv0, vget_low_f32(v0), 0);
+	rv1 = vmulq_lane_f32(mv0, vget_low_f32(v1), 0);
+	rv2 = vmulq_lane_f32(mv0, vget_low_f32(v2), 0);
+
+	rv0 = vmlaq_lane_f32(rv0, mv1, vget_low_f32(v0), 1);
+	rv1 = vmlaq_lane_f32(rv1, mv1, vget_low_f32(v1), 1);
+	rv2 = vmlaq_lane_f32(rv2, mv1, vget_low_f32(v2), 1);
+
+	rv0 = vmlaq_lane_f32(rv0, mv2, vget_high_f32(v0), 0);
+	rv1 = vmlaq_lane_f32(rv1, mv2, vget_high_f32(v1), 0);
+	rv2 = vmlaq_lane_f32(rv2, mv2, vget_high_f32(v2), 0);
+
+	m_el[0].mVec128 = rv0;
+	m_el[1].mVec128 = rv1;
+	m_el[2].mVec128 = rv2;
+#else
 	setValue(
-        m.tdotx(m_el[0]), m.tdoty(m_el[0]), m.tdotz(m_el[0]),
+		m.tdotx(m_el[0]), m.tdoty(m_el[0]), m.tdotz(m_el[0]),
 		m.tdotx(m_el[1]), m.tdoty(m_el[1]), m.tdotz(m_el[1]),
 		m.tdotx(m_el[2]), m.tdoty(m_el[2]), m.tdotz(m_el[2]));
 #endif
 	return *this;
 }
 
-B3_FORCE_INLINE b3Matrix3x3& 
+B3_FORCE_INLINE b3Matrix3x3&
 b3Matrix3x3::operator+=(const b3Matrix3x3& m)
 {
-#if (defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE))|| defined (B3_USE_NEON)
-    m_el[0].mVec128 = m_el[0].mVec128 + m.m_el[0].mVec128;
-    m_el[1].mVec128 = m_el[1].mVec128 + m.m_el[1].mVec128;
-    m_el[2].mVec128 = m_el[2].mVec128 + m.m_el[2].mVec128;
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)) || defined(B3_USE_NEON)
+	m_el[0].mVec128 = m_el[0].mVec128 + m.m_el[0].mVec128;
+	m_el[1].mVec128 = m_el[1].mVec128 + m.m_el[1].mVec128;
+	m_el[2].mVec128 = m_el[2].mVec128 + m.m_el[2].mVec128;
 #else
 	setValue(
-		m_el[0][0]+m.m_el[0][0], 
-		m_el[0][1]+m.m_el[0][1],
-		m_el[0][2]+m.m_el[0][2],
-		m_el[1][0]+m.m_el[1][0], 
-		m_el[1][1]+m.m_el[1][1],
-		m_el[1][2]+m.m_el[1][2],
-		m_el[2][0]+m.m_el[2][0], 
-		m_el[2][1]+m.m_el[2][1],
-		m_el[2][2]+m.m_el[2][2]);
+		m_el[0][0] + m.m_el[0][0],
+		m_el[0][1] + m.m_el[0][1],
+		m_el[0][2] + m.m_el[0][2],
+		m_el[1][0] + m.m_el[1][0],
+		m_el[1][1] + m.m_el[1][1],
+		m_el[1][2] + m.m_el[1][2],
+		m_el[2][0] + m.m_el[2][0],
+		m_el[2][1] + m.m_el[2][1],
+		m_el[2][2] + m.m_el[2][2]);
 #endif
 	return *this;
 }
 
 B3_FORCE_INLINE b3Matrix3x3
-operator*(const b3Matrix3x3& m, const b3Scalar & k)
+operator*(const b3Matrix3x3& m, const b3Scalar& k)
 {
-#if (defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE))
-    __m128 vk = b3_splat_ps(_mm_load_ss((float *)&k), 0x80);
-    return b3Matrix3x3(
-                _mm_mul_ps(m[0].mVec128, vk), 
-                _mm_mul_ps(m[1].mVec128, vk), 
-                _mm_mul_ps(m[2].mVec128, vk)); 
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE))
+	__m128 vk = b3_splat_ps(_mm_load_ss((float*)&k), 0x80);
+	return b3Matrix3x3(
+		_mm_mul_ps(m[0].mVec128, vk),
+		_mm_mul_ps(m[1].mVec128, vk),
+		_mm_mul_ps(m[2].mVec128, vk));
 #elif defined(B3_USE_NEON)
-    return b3Matrix3x3(
-                vmulq_n_f32(m[0].mVec128, k),
-                vmulq_n_f32(m[1].mVec128, k),
-                vmulq_n_f32(m[2].mVec128, k)); 
+	return b3Matrix3x3(
+		vmulq_n_f32(m[0].mVec128, k),
+		vmulq_n_f32(m[1].mVec128, k),
+		vmulq_n_f32(m[2].mVec128, k));
 #else
 	return b3Matrix3x3(
-		m[0].getX()*k,m[0].getY()*k,m[0].getZ()*k,
-		m[1].getX()*k,m[1].getY()*k,m[1].getZ()*k,
-		m[2].getX()*k,m[2].getY()*k,m[2].getZ()*k);
+		m[0].getX() * k, m[0].getY() * k, m[0].getZ() * k,
+		m[1].getX() * k, m[1].getY() * k, m[1].getZ() * k,
+		m[2].getX() * k, m[2].getY() * k, m[2].getZ() * k);
 #endif
 }
 
-B3_FORCE_INLINE b3Matrix3x3 
+B3_FORCE_INLINE b3Matrix3x3
 operator+(const b3Matrix3x3& m1, const b3Matrix3x3& m2)
 {
-#if (defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE))|| defined (B3_USE_NEON)
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)) || defined(B3_USE_NEON)
 	return b3Matrix3x3(
-        m1[0].mVec128 + m2[0].mVec128,
-        m1[1].mVec128 + m2[1].mVec128,
-        m1[2].mVec128 + m2[2].mVec128);
+		m1[0].mVec128 + m2[0].mVec128,
+		m1[1].mVec128 + m2[1].mVec128,
+		m1[2].mVec128 + m2[2].mVec128);
 #else
 	return b3Matrix3x3(
-        m1[0][0]+m2[0][0], 
-        m1[0][1]+m2[0][1],
-        m1[0][2]+m2[0][2],
-        
-        m1[1][0]+m2[1][0], 
-        m1[1][1]+m2[1][1],
-        m1[1][2]+m2[1][2],
-        
-        m1[2][0]+m2[2][0], 
-        m1[2][1]+m2[2][1],
-        m1[2][2]+m2[2][2]);
-#endif    
-}
+		m1[0][0] + m2[0][0],
+		m1[0][1] + m2[0][1],
+		m1[0][2] + m2[0][2],
 
-B3_FORCE_INLINE b3Matrix3x3 
-operator-(const b3Matrix3x3& m1, const b3Matrix3x3& m2)
-{
-#if (defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE))|| defined (B3_USE_NEON)
-	return b3Matrix3x3(
-        m1[0].mVec128 - m2[0].mVec128,
-        m1[1].mVec128 - m2[1].mVec128,
-        m1[2].mVec128 - m2[2].mVec128);
-#else
-	return b3Matrix3x3(
-        m1[0][0]-m2[0][0], 
-        m1[0][1]-m2[0][1],
-        m1[0][2]-m2[0][2],
-        
-        m1[1][0]-m2[1][0], 
-        m1[1][1]-m2[1][1],
-        m1[1][2]-m2[1][2],
-        
-        m1[2][0]-m2[2][0], 
-        m1[2][1]-m2[2][1],
-        m1[2][2]-m2[2][2]);
+		m1[1][0] + m2[1][0],
+		m1[1][1] + m2[1][1],
+		m1[1][2] + m2[1][2],
+
+		m1[2][0] + m2[2][0],
+		m1[2][1] + m2[2][1],
+		m1[2][2] + m2[2][2]);
 #endif
 }
 
+B3_FORCE_INLINE b3Matrix3x3
+operator-(const b3Matrix3x3& m1, const b3Matrix3x3& m2)
+{
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)) || defined(B3_USE_NEON)
+	return b3Matrix3x3(
+		m1[0].mVec128 - m2[0].mVec128,
+		m1[1].mVec128 - m2[1].mVec128,
+		m1[2].mVec128 - m2[2].mVec128);
+#else
+	return b3Matrix3x3(
+		m1[0][0] - m2[0][0],
+		m1[0][1] - m2[0][1],
+		m1[0][2] - m2[0][2],
 
-B3_FORCE_INLINE b3Matrix3x3& 
+		m1[1][0] - m2[1][0],
+		m1[1][1] - m2[1][1],
+		m1[1][2] - m2[1][2],
+
+		m1[2][0] - m2[2][0],
+		m1[2][1] - m2[2][1],
+		m1[2][2] - m2[2][2]);
+#endif
+}
+
+B3_FORCE_INLINE b3Matrix3x3&
 b3Matrix3x3::operator-=(const b3Matrix3x3& m)
 {
-#if (defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE))|| defined (B3_USE_NEON)
-    m_el[0].mVec128 = m_el[0].mVec128 - m.m_el[0].mVec128;
-    m_el[1].mVec128 = m_el[1].mVec128 - m.m_el[1].mVec128;
-    m_el[2].mVec128 = m_el[2].mVec128 - m.m_el[2].mVec128;
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)) || defined(B3_USE_NEON)
+	m_el[0].mVec128 = m_el[0].mVec128 - m.m_el[0].mVec128;
+	m_el[1].mVec128 = m_el[1].mVec128 - m.m_el[1].mVec128;
+	m_el[2].mVec128 = m_el[2].mVec128 - m.m_el[2].mVec128;
 #else
 	setValue(
-	m_el[0][0]-m.m_el[0][0], 
-	m_el[0][1]-m.m_el[0][1],
-	m_el[0][2]-m.m_el[0][2],
-	m_el[1][0]-m.m_el[1][0], 
-	m_el[1][1]-m.m_el[1][1],
-	m_el[1][2]-m.m_el[1][2],
-	m_el[2][0]-m.m_el[2][0], 
-	m_el[2][1]-m.m_el[2][1],
-	m_el[2][2]-m.m_el[2][2]);
+		m_el[0][0] - m.m_el[0][0],
+		m_el[0][1] - m.m_el[0][1],
+		m_el[0][2] - m.m_el[0][2],
+		m_el[1][0] - m.m_el[1][0],
+		m_el[1][1] - m.m_el[1][1],
+		m_el[1][2] - m.m_el[1][2],
+		m_el[2][0] - m.m_el[2][0],
+		m_el[2][1] - m.m_el[2][1],
+		m_el[2][2] - m.m_el[2][2]);
 #endif
 	return *this;
 }
 
-
-B3_FORCE_INLINE b3Scalar 
+B3_FORCE_INLINE b3Scalar
 b3Matrix3x3::determinant() const
-{ 
+{
 	return b3Triple((*this)[0], (*this)[1], (*this)[2]);
 }
 
-
-B3_FORCE_INLINE b3Matrix3x3 
+B3_FORCE_INLINE b3Matrix3x3
 b3Matrix3x3::absolute() const
 {
-#if (defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE))
-    return b3Matrix3x3(
-            _mm_and_ps(m_el[0].mVec128, b3vAbsfMask),
-            _mm_and_ps(m_el[1].mVec128, b3vAbsfMask),
-            _mm_and_ps(m_el[2].mVec128, b3vAbsfMask));
-#elif defined(B3_USE_NEON)
-    return b3Matrix3x3(
-            (float32x4_t)vandq_s32((int32x4_t)m_el[0].mVec128, b3v3AbsMask),
-            (float32x4_t)vandq_s32((int32x4_t)m_el[1].mVec128, b3v3AbsMask),
-            (float32x4_t)vandq_s32((int32x4_t)m_el[2].mVec128, b3v3AbsMask));
-#else	
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE))
 	return b3Matrix3x3(
-            b3Fabs(m_el[0].getX()), b3Fabs(m_el[0].getY()), b3Fabs(m_el[0].getZ()),
-            b3Fabs(m_el[1].getX()), b3Fabs(m_el[1].getY()), b3Fabs(m_el[1].getZ()),
-            b3Fabs(m_el[2].getX()), b3Fabs(m_el[2].getY()), b3Fabs(m_el[2].getZ()));
-#endif
-}
-
-B3_FORCE_INLINE b3Matrix3x3 
-b3Matrix3x3::transpose() const 
-{
-#if (defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE))
-    __m128 v0 = m_el[0].mVec128;
-    __m128 v1 = m_el[1].mVec128;
-    __m128 v2 = m_el[2].mVec128;    //  x2 y2 z2 w2
-    __m128 vT;
-    
-    v2 = _mm_and_ps(v2, b3vFFF0fMask);  //  x2 y2 z2 0
-    
-    vT = _mm_unpackhi_ps(v0, v1);	//	z0 z1 * *
-    v0 = _mm_unpacklo_ps(v0, v1);	//	x0 x1 y0 y1
-
-    v1 = _mm_shuffle_ps(v0, v2, B3_SHUFFLE(2, 3, 1, 3) );	// y0 y1 y2 0
-    v0 = _mm_shuffle_ps(v0, v2, B3_SHUFFLE(0, 1, 0, 3) );	// x0 x1 x2 0
-    v2 = b3CastdTo128f(_mm_move_sd(b3CastfTo128d(v2), b3CastfTo128d(vT)));	// z0 z1 z2 0
-
-
-    return b3Matrix3x3( v0, v1, v2 );
+		_mm_and_ps(m_el[0].mVec128, b3vAbsfMask),
+		_mm_and_ps(m_el[1].mVec128, b3vAbsfMask),
+		_mm_and_ps(m_el[2].mVec128, b3vAbsfMask));
 #elif defined(B3_USE_NEON)
-    // note: zeros the w channel. We can preserve it at the cost of two more vtrn instructions.
-    static const uint32x2_t zMask = (const uint32x2_t) {-1, 0 };
-    float32x4x2_t top = vtrnq_f32( m_el[0].mVec128, m_el[1].mVec128 );  // {x0 x1 z0 z1}, {y0 y1 w0 w1}
-    float32x2x2_t bl = vtrn_f32( vget_low_f32(m_el[2].mVec128), vdup_n_f32(0.0f) );       // {x2  0 }, {y2 0}
-    float32x4_t v0 = vcombine_f32( vget_low_f32(top.val[0]), bl.val[0] );
-    float32x4_t v1 = vcombine_f32( vget_low_f32(top.val[1]), bl.val[1] );
-    float32x2_t q = (float32x2_t) vand_u32( (uint32x2_t) vget_high_f32( m_el[2].mVec128), zMask );
-    float32x4_t v2 = vcombine_f32( vget_high_f32(top.val[0]), q );       // z0 z1 z2  0
-    return b3Matrix3x3( v0, v1, v2 ); 
+	return b3Matrix3x3(
+		(float32x4_t)vandq_s32((int32x4_t)m_el[0].mVec128, b3v3AbsMask),
+		(float32x4_t)vandq_s32((int32x4_t)m_el[1].mVec128, b3v3AbsMask),
+		(float32x4_t)vandq_s32((int32x4_t)m_el[2].mVec128, b3v3AbsMask));
 #else
-	return b3Matrix3x3( m_el[0].getX(), m_el[1].getX(), m_el[2].getX(),
-                        m_el[0].getY(), m_el[1].getY(), m_el[2].getY(),
-                        m_el[0].getZ(), m_el[1].getZ(), m_el[2].getZ());
+	return b3Matrix3x3(
+		b3Fabs(m_el[0].getX()), b3Fabs(m_el[0].getY()), b3Fabs(m_el[0].getZ()),
+		b3Fabs(m_el[1].getX()), b3Fabs(m_el[1].getY()), b3Fabs(m_el[1].getZ()),
+		b3Fabs(m_el[2].getX()), b3Fabs(m_el[2].getY()), b3Fabs(m_el[2].getZ()));
 #endif
 }
 
-B3_FORCE_INLINE b3Matrix3x3 
-b3Matrix3x3::adjoint() const 
+B3_FORCE_INLINE b3Matrix3x3
+b3Matrix3x3::transpose() const
+{
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE))
+	__m128 v0 = m_el[0].mVec128;
+	__m128 v1 = m_el[1].mVec128;
+	__m128 v2 = m_el[2].mVec128;  //  x2 y2 z2 w2
+	__m128 vT;
+
+	v2 = _mm_and_ps(v2, b3vFFF0fMask);  //  x2 y2 z2 0
+
+	vT = _mm_unpackhi_ps(v0, v1);  //	z0 z1 * *
+	v0 = _mm_unpacklo_ps(v0, v1);  //	x0 x1 y0 y1
+
+	v1 = _mm_shuffle_ps(v0, v2, B3_SHUFFLE(2, 3, 1, 3));                    // y0 y1 y2 0
+	v0 = _mm_shuffle_ps(v0, v2, B3_SHUFFLE(0, 1, 0, 3));                    // x0 x1 x2 0
+	v2 = b3CastdTo128f(_mm_move_sd(b3CastfTo128d(v2), b3CastfTo128d(vT)));  // z0 z1 z2 0
+
+	return b3Matrix3x3(v0, v1, v2);
+#elif defined(B3_USE_NEON)
+	// note: zeros the w channel. We can preserve it at the cost of two more vtrn instructions.
+	static const uint32x2_t zMask = (const uint32x2_t){-1, 0};
+	float32x4x2_t top = vtrnq_f32(m_el[0].mVec128, m_el[1].mVec128);               // {x0 x1 z0 z1}, {y0 y1 w0 w1}
+	float32x2x2_t bl = vtrn_f32(vget_low_f32(m_el[2].mVec128), vdup_n_f32(0.0f));  // {x2  0 }, {y2 0}
+	float32x4_t v0 = vcombine_f32(vget_low_f32(top.val[0]), bl.val[0]);
+	float32x4_t v1 = vcombine_f32(vget_low_f32(top.val[1]), bl.val[1]);
+	float32x2_t q = (float32x2_t)vand_u32((uint32x2_t)vget_high_f32(m_el[2].mVec128), zMask);
+	float32x4_t v2 = vcombine_f32(vget_high_f32(top.val[0]), q);  // z0 z1 z2  0
+	return b3Matrix3x3(v0, v1, v2);
+#else
+	return b3Matrix3x3(m_el[0].getX(), m_el[1].getX(), m_el[2].getX(),
+					   m_el[0].getY(), m_el[1].getY(), m_el[2].getY(),
+					   m_el[0].getZ(), m_el[1].getZ(), m_el[2].getZ());
+#endif
+}
+
+B3_FORCE_INLINE b3Matrix3x3
+b3Matrix3x3::adjoint() const
 {
 	return b3Matrix3x3(cofac(1, 1, 2, 2), cofac(0, 2, 2, 1), cofac(0, 1, 1, 2),
-		cofac(1, 2, 2, 0), cofac(0, 0, 2, 2), cofac(0, 2, 1, 0),
-		cofac(1, 0, 2, 1), cofac(0, 1, 2, 0), cofac(0, 0, 1, 1));
+					   cofac(1, 2, 2, 0), cofac(0, 0, 2, 2), cofac(0, 2, 1, 0),
+					   cofac(1, 0, 2, 1), cofac(0, 1, 2, 0), cofac(0, 0, 1, 1));
 }
 
-B3_FORCE_INLINE b3Matrix3x3 
+B3_FORCE_INLINE b3Matrix3x3
 b3Matrix3x3::inverse() const
 {
 	b3Vector3 co = b3MakeVector3(cofac(1, 1, 2, 2), cofac(1, 2, 2, 0), cofac(1, 0, 2, 1));
@@ -1024,54 +1023,54 @@ b3Matrix3x3::inverse() const
 	b3FullAssert(det != b3Scalar(0.0));
 	b3Scalar s = b3Scalar(1.0) / det;
 	return b3Matrix3x3(co.getX() * s, cofac(0, 2, 2, 1) * s, cofac(0, 1, 1, 2) * s,
-		co.getY() * s, cofac(0, 0, 2, 2) * s, cofac(0, 2, 1, 0) * s,
-		co.getZ() * s, cofac(0, 1, 2, 0) * s, cofac(0, 0, 1, 1) * s);
+					   co.getY() * s, cofac(0, 0, 2, 2) * s, cofac(0, 2, 1, 0) * s,
+					   co.getZ() * s, cofac(0, 1, 2, 0) * s, cofac(0, 0, 1, 1) * s);
 }
 
-B3_FORCE_INLINE b3Matrix3x3 
+B3_FORCE_INLINE b3Matrix3x3
 b3Matrix3x3::transposeTimes(const b3Matrix3x3& m) const
 {
-#if (defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE))
-    // zeros w
-//    static const __m128i xyzMask = (const __m128i){ -1ULL, 0xffffffffULL };
-    __m128 row = m_el[0].mVec128;
-    __m128 m0 = _mm_and_ps( m.getRow(0).mVec128, b3vFFF0fMask );
-    __m128 m1 = _mm_and_ps( m.getRow(1).mVec128, b3vFFF0fMask);
-    __m128 m2 = _mm_and_ps( m.getRow(2).mVec128, b3vFFF0fMask );
-    __m128 r0 = _mm_mul_ps(m0, _mm_shuffle_ps(row, row, 0));
-    __m128 r1 = _mm_mul_ps(m0, _mm_shuffle_ps(row, row, 0x55));
-    __m128 r2 = _mm_mul_ps(m0, _mm_shuffle_ps(row, row, 0xaa));
-    row = m_el[1].mVec128;
-    r0 = _mm_add_ps( r0, _mm_mul_ps(m1, _mm_shuffle_ps(row, row, 0)));
-    r1 = _mm_add_ps( r1, _mm_mul_ps(m1, _mm_shuffle_ps(row, row, 0x55)));
-    r2 = _mm_add_ps( r2, _mm_mul_ps(m1, _mm_shuffle_ps(row, row, 0xaa)));
-    row = m_el[2].mVec128;
-    r0 = _mm_add_ps( r0, _mm_mul_ps(m2, _mm_shuffle_ps(row, row, 0)));
-    r1 = _mm_add_ps( r1, _mm_mul_ps(m2, _mm_shuffle_ps(row, row, 0x55)));
-    r2 = _mm_add_ps( r2, _mm_mul_ps(m2, _mm_shuffle_ps(row, row, 0xaa)));
-    return b3Matrix3x3( r0, r1, r2 );
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE))
+	// zeros w
+	//    static const __m128i xyzMask = (const __m128i){ -1ULL, 0xffffffffULL };
+	__m128 row = m_el[0].mVec128;
+	__m128 m0 = _mm_and_ps(m.getRow(0).mVec128, b3vFFF0fMask);
+	__m128 m1 = _mm_and_ps(m.getRow(1).mVec128, b3vFFF0fMask);
+	__m128 m2 = _mm_and_ps(m.getRow(2).mVec128, b3vFFF0fMask);
+	__m128 r0 = _mm_mul_ps(m0, _mm_shuffle_ps(row, row, 0));
+	__m128 r1 = _mm_mul_ps(m0, _mm_shuffle_ps(row, row, 0x55));
+	__m128 r2 = _mm_mul_ps(m0, _mm_shuffle_ps(row, row, 0xaa));
+	row = m_el[1].mVec128;
+	r0 = _mm_add_ps(r0, _mm_mul_ps(m1, _mm_shuffle_ps(row, row, 0)));
+	r1 = _mm_add_ps(r1, _mm_mul_ps(m1, _mm_shuffle_ps(row, row, 0x55)));
+	r2 = _mm_add_ps(r2, _mm_mul_ps(m1, _mm_shuffle_ps(row, row, 0xaa)));
+	row = m_el[2].mVec128;
+	r0 = _mm_add_ps(r0, _mm_mul_ps(m2, _mm_shuffle_ps(row, row, 0)));
+	r1 = _mm_add_ps(r1, _mm_mul_ps(m2, _mm_shuffle_ps(row, row, 0x55)));
+	r2 = _mm_add_ps(r2, _mm_mul_ps(m2, _mm_shuffle_ps(row, row, 0xaa)));
+	return b3Matrix3x3(r0, r1, r2);
 
 #elif defined B3_USE_NEON
-    // zeros w
-    static const uint32x4_t xyzMask = (const uint32x4_t){ -1, -1, -1, 0 };
-    float32x4_t m0 = (float32x4_t) vandq_u32( (uint32x4_t) m.getRow(0).mVec128, xyzMask );
-    float32x4_t m1 = (float32x4_t) vandq_u32( (uint32x4_t) m.getRow(1).mVec128, xyzMask );
-    float32x4_t m2 = (float32x4_t) vandq_u32( (uint32x4_t) m.getRow(2).mVec128, xyzMask );
-    float32x4_t row = m_el[0].mVec128;
-    float32x4_t r0 = vmulq_lane_f32( m0, vget_low_f32(row), 0);
-    float32x4_t r1 = vmulq_lane_f32( m0, vget_low_f32(row), 1);
-    float32x4_t r2 = vmulq_lane_f32( m0, vget_high_f32(row), 0);
-    row = m_el[1].mVec128;
-    r0 = vmlaq_lane_f32( r0, m1, vget_low_f32(row), 0);
-    r1 = vmlaq_lane_f32( r1, m1, vget_low_f32(row), 1);
-    r2 = vmlaq_lane_f32( r2, m1, vget_high_f32(row), 0);
-    row = m_el[2].mVec128;
-    r0 = vmlaq_lane_f32( r0, m2, vget_low_f32(row), 0);
-    r1 = vmlaq_lane_f32( r1, m2, vget_low_f32(row), 1);
-    r2 = vmlaq_lane_f32( r2, m2, vget_high_f32(row), 0);
-    return b3Matrix3x3( r0, r1, r2 );
+	// zeros w
+	static const uint32x4_t xyzMask = (const uint32x4_t){-1, -1, -1, 0};
+	float32x4_t m0 = (float32x4_t)vandq_u32((uint32x4_t)m.getRow(0).mVec128, xyzMask);
+	float32x4_t m1 = (float32x4_t)vandq_u32((uint32x4_t)m.getRow(1).mVec128, xyzMask);
+	float32x4_t m2 = (float32x4_t)vandq_u32((uint32x4_t)m.getRow(2).mVec128, xyzMask);
+	float32x4_t row = m_el[0].mVec128;
+	float32x4_t r0 = vmulq_lane_f32(m0, vget_low_f32(row), 0);
+	float32x4_t r1 = vmulq_lane_f32(m0, vget_low_f32(row), 1);
+	float32x4_t r2 = vmulq_lane_f32(m0, vget_high_f32(row), 0);
+	row = m_el[1].mVec128;
+	r0 = vmlaq_lane_f32(r0, m1, vget_low_f32(row), 0);
+	r1 = vmlaq_lane_f32(r1, m1, vget_low_f32(row), 1);
+	r2 = vmlaq_lane_f32(r2, m1, vget_high_f32(row), 0);
+	row = m_el[2].mVec128;
+	r0 = vmlaq_lane_f32(r0, m2, vget_low_f32(row), 0);
+	r1 = vmlaq_lane_f32(r1, m2, vget_low_f32(row), 1);
+	r2 = vmlaq_lane_f32(r2, m2, vget_high_f32(row), 0);
+	return b3Matrix3x3(r0, r1, r2);
 #else
-    return b3Matrix3x3(
+	return b3Matrix3x3(
 		m_el[0].getX() * m[0].getX() + m_el[1].getX() * m[1].getX() + m_el[2].getX() * m[2].getX(),
 		m_el[0].getX() * m[0].getY() + m_el[1].getX() * m[1].getY() + m_el[2].getX() * m[2].getY(),
 		m_el[0].getX() * m[0].getZ() + m_el[1].getX() * m[1].getZ() + m_el[2].getX() * m[2].getZ(),
@@ -1084,51 +1083,51 @@ b3Matrix3x3::transposeTimes(const b3Matrix3x3& m) const
 #endif
 }
 
-B3_FORCE_INLINE b3Matrix3x3 
+B3_FORCE_INLINE b3Matrix3x3
 b3Matrix3x3::timesTranspose(const b3Matrix3x3& m) const
 {
-#if (defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE))
-    __m128 a0 = m_el[0].mVec128;
-    __m128 a1 = m_el[1].mVec128;
-    __m128 a2 = m_el[2].mVec128;
-    
-    b3Matrix3x3 mT = m.transpose(); // we rely on transpose() zeroing w channel so that we don't have to do it here
-    __m128 mx = mT[0].mVec128;
-    __m128 my = mT[1].mVec128;
-    __m128 mz = mT[2].mVec128;
-    
-    __m128 r0 = _mm_mul_ps(mx, _mm_shuffle_ps(a0, a0, 0x00));
-    __m128 r1 = _mm_mul_ps(mx, _mm_shuffle_ps(a1, a1, 0x00));
-    __m128 r2 = _mm_mul_ps(mx, _mm_shuffle_ps(a2, a2, 0x00));
-    r0 = _mm_add_ps(r0, _mm_mul_ps(my, _mm_shuffle_ps(a0, a0, 0x55)));
-    r1 = _mm_add_ps(r1, _mm_mul_ps(my, _mm_shuffle_ps(a1, a1, 0x55)));
-    r2 = _mm_add_ps(r2, _mm_mul_ps(my, _mm_shuffle_ps(a2, a2, 0x55)));
-    r0 = _mm_add_ps(r0, _mm_mul_ps(mz, _mm_shuffle_ps(a0, a0, 0xaa)));
-    r1 = _mm_add_ps(r1, _mm_mul_ps(mz, _mm_shuffle_ps(a1, a1, 0xaa)));
-    r2 = _mm_add_ps(r2, _mm_mul_ps(mz, _mm_shuffle_ps(a2, a2, 0xaa)));
-    return b3Matrix3x3( r0, r1, r2);
-            
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE))
+	__m128 a0 = m_el[0].mVec128;
+	__m128 a1 = m_el[1].mVec128;
+	__m128 a2 = m_el[2].mVec128;
+
+	b3Matrix3x3 mT = m.transpose();  // we rely on transpose() zeroing w channel so that we don't have to do it here
+	__m128 mx = mT[0].mVec128;
+	__m128 my = mT[1].mVec128;
+	__m128 mz = mT[2].mVec128;
+
+	__m128 r0 = _mm_mul_ps(mx, _mm_shuffle_ps(a0, a0, 0x00));
+	__m128 r1 = _mm_mul_ps(mx, _mm_shuffle_ps(a1, a1, 0x00));
+	__m128 r2 = _mm_mul_ps(mx, _mm_shuffle_ps(a2, a2, 0x00));
+	r0 = _mm_add_ps(r0, _mm_mul_ps(my, _mm_shuffle_ps(a0, a0, 0x55)));
+	r1 = _mm_add_ps(r1, _mm_mul_ps(my, _mm_shuffle_ps(a1, a1, 0x55)));
+	r2 = _mm_add_ps(r2, _mm_mul_ps(my, _mm_shuffle_ps(a2, a2, 0x55)));
+	r0 = _mm_add_ps(r0, _mm_mul_ps(mz, _mm_shuffle_ps(a0, a0, 0xaa)));
+	r1 = _mm_add_ps(r1, _mm_mul_ps(mz, _mm_shuffle_ps(a1, a1, 0xaa)));
+	r2 = _mm_add_ps(r2, _mm_mul_ps(mz, _mm_shuffle_ps(a2, a2, 0xaa)));
+	return b3Matrix3x3(r0, r1, r2);
+
 #elif defined B3_USE_NEON
-    float32x4_t a0 = m_el[0].mVec128;
-    float32x4_t a1 = m_el[1].mVec128;
-    float32x4_t a2 = m_el[2].mVec128;
-    
-    b3Matrix3x3 mT = m.transpose(); // we rely on transpose() zeroing w channel so that we don't have to do it here
-    float32x4_t mx = mT[0].mVec128;
-    float32x4_t my = mT[1].mVec128;
-    float32x4_t mz = mT[2].mVec128;
-    
-    float32x4_t r0 = vmulq_lane_f32( mx, vget_low_f32(a0), 0);
-    float32x4_t r1 = vmulq_lane_f32( mx, vget_low_f32(a1), 0);
-    float32x4_t r2 = vmulq_lane_f32( mx, vget_low_f32(a2), 0);
-    r0 = vmlaq_lane_f32( r0, my, vget_low_f32(a0), 1);
-    r1 = vmlaq_lane_f32( r1, my, vget_low_f32(a1), 1);
-    r2 = vmlaq_lane_f32( r2, my, vget_low_f32(a2), 1);
-    r0 = vmlaq_lane_f32( r0, mz, vget_high_f32(a0), 0);
-    r1 = vmlaq_lane_f32( r1, mz, vget_high_f32(a1), 0);
-    r2 = vmlaq_lane_f32( r2, mz, vget_high_f32(a2), 0);
-    return b3Matrix3x3( r0, r1, r2 );
-    
+	float32x4_t a0 = m_el[0].mVec128;
+	float32x4_t a1 = m_el[1].mVec128;
+	float32x4_t a2 = m_el[2].mVec128;
+
+	b3Matrix3x3 mT = m.transpose();  // we rely on transpose() zeroing w channel so that we don't have to do it here
+	float32x4_t mx = mT[0].mVec128;
+	float32x4_t my = mT[1].mVec128;
+	float32x4_t mz = mT[2].mVec128;
+
+	float32x4_t r0 = vmulq_lane_f32(mx, vget_low_f32(a0), 0);
+	float32x4_t r1 = vmulq_lane_f32(mx, vget_low_f32(a1), 0);
+	float32x4_t r2 = vmulq_lane_f32(mx, vget_low_f32(a2), 0);
+	r0 = vmlaq_lane_f32(r0, my, vget_low_f32(a0), 1);
+	r1 = vmlaq_lane_f32(r1, my, vget_low_f32(a1), 1);
+	r2 = vmlaq_lane_f32(r2, my, vget_low_f32(a2), 1);
+	r0 = vmlaq_lane_f32(r0, mz, vget_high_f32(a0), 0);
+	r1 = vmlaq_lane_f32(r1, mz, vget_high_f32(a1), 0);
+	r2 = vmlaq_lane_f32(r2, mz, vget_high_f32(a2), 0);
+	return b3Matrix3x3(r0, r1, r2);
+
 #else
 	return b3Matrix3x3(
 		m_el[0].dot(m[0]), m_el[0].dot(m[1]), m_el[0].dot(m[2]),
@@ -1137,139 +1136,138 @@ b3Matrix3x3::timesTranspose(const b3Matrix3x3& m) const
 #endif
 }
 
-B3_FORCE_INLINE b3Vector3 
-operator*(const b3Matrix3x3& m, const b3Vector3& v) 
+B3_FORCE_INLINE b3Vector3
+operator*(const b3Matrix3x3& m, const b3Vector3& v)
 {
-#if (defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE))|| defined (B3_USE_NEON)
-    return v.dot3(m[0], m[1], m[2]);
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)) || defined(B3_USE_NEON)
+	return v.dot3(m[0], m[1], m[2]);
 #else
 	return b3MakeVector3(m[0].dot(v), m[1].dot(v), m[2].dot(v));
 #endif
 }
 
-
 B3_FORCE_INLINE b3Vector3
 operator*(const b3Vector3& v, const b3Matrix3x3& m)
 {
-#if (defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE))
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE))
 
-    const __m128 vv = v.mVec128;
+	const __m128 vv = v.mVec128;
 
-    __m128 c0 = b3_splat_ps( vv, 0);
-    __m128 c1 = b3_splat_ps( vv, 1);
-    __m128 c2 = b3_splat_ps( vv, 2);
+	__m128 c0 = b3_splat_ps(vv, 0);
+	__m128 c1 = b3_splat_ps(vv, 1);
+	__m128 c2 = b3_splat_ps(vv, 2);
 
-    c0 = _mm_mul_ps(c0, _mm_and_ps(m[0].mVec128, b3vFFF0fMask) );
-    c1 = _mm_mul_ps(c1, _mm_and_ps(m[1].mVec128, b3vFFF0fMask) );
-    c0 = _mm_add_ps(c0, c1);
-    c2 = _mm_mul_ps(c2, _mm_and_ps(m[2].mVec128, b3vFFF0fMask) );
-    
-    return b3MakeVector3(_mm_add_ps(c0, c2));
+	c0 = _mm_mul_ps(c0, _mm_and_ps(m[0].mVec128, b3vFFF0fMask));
+	c1 = _mm_mul_ps(c1, _mm_and_ps(m[1].mVec128, b3vFFF0fMask));
+	c0 = _mm_add_ps(c0, c1);
+	c2 = _mm_mul_ps(c2, _mm_and_ps(m[2].mVec128, b3vFFF0fMask));
+
+	return b3MakeVector3(_mm_add_ps(c0, c2));
 #elif defined(B3_USE_NEON)
-    const float32x4_t vv = v.mVec128;
-    const float32x2_t vlo = vget_low_f32(vv);
-    const float32x2_t vhi = vget_high_f32(vv);
+	const float32x4_t vv = v.mVec128;
+	const float32x2_t vlo = vget_low_f32(vv);
+	const float32x2_t vhi = vget_high_f32(vv);
 
-    float32x4_t c0, c1, c2;
+	float32x4_t c0, c1, c2;
 
-    c0 = (float32x4_t) vandq_s32((int32x4_t)m[0].mVec128, b3vFFF0Mask);
-    c1 = (float32x4_t) vandq_s32((int32x4_t)m[1].mVec128, b3vFFF0Mask);
-    c2 = (float32x4_t) vandq_s32((int32x4_t)m[2].mVec128, b3vFFF0Mask);
+	c0 = (float32x4_t)vandq_s32((int32x4_t)m[0].mVec128, b3vFFF0Mask);
+	c1 = (float32x4_t)vandq_s32((int32x4_t)m[1].mVec128, b3vFFF0Mask);
+	c2 = (float32x4_t)vandq_s32((int32x4_t)m[2].mVec128, b3vFFF0Mask);
 
-    c0 = vmulq_lane_f32(c0, vlo, 0);
-    c1 = vmulq_lane_f32(c1, vlo, 1);
-    c2 = vmulq_lane_f32(c2, vhi, 0);
-    c0 = vaddq_f32(c0, c1);
-    c0 = vaddq_f32(c0, c2);
-    
-    return b3MakeVector3(c0);
+	c0 = vmulq_lane_f32(c0, vlo, 0);
+	c1 = vmulq_lane_f32(c1, vlo, 1);
+	c2 = vmulq_lane_f32(c2, vhi, 0);
+	c0 = vaddq_f32(c0, c1);
+	c0 = vaddq_f32(c0, c2);
+
+	return b3MakeVector3(c0);
 #else
 	return b3MakeVector3(m.tdotx(v), m.tdoty(v), m.tdotz(v));
 #endif
 }
 
-B3_FORCE_INLINE b3Matrix3x3 
+B3_FORCE_INLINE b3Matrix3x3
 operator*(const b3Matrix3x3& m1, const b3Matrix3x3& m2)
 {
-#if (defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE))
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE))
 
-    __m128 m10 = m1[0].mVec128;  
-    __m128 m11 = m1[1].mVec128;
-    __m128 m12 = m1[2].mVec128;
-    
-    __m128 m2v = _mm_and_ps(m2[0].mVec128, b3vFFF0fMask);
-    
-    __m128 c0 = b3_splat_ps( m10, 0);
-    __m128 c1 = b3_splat_ps( m11, 0);
-    __m128 c2 = b3_splat_ps( m12, 0);
-    
-    c0 = _mm_mul_ps(c0, m2v);
-    c1 = _mm_mul_ps(c1, m2v);
-    c2 = _mm_mul_ps(c2, m2v);
-    
-    m2v = _mm_and_ps(m2[1].mVec128, b3vFFF0fMask);
-    
-    __m128 c0_1 = b3_splat_ps( m10, 1);
-    __m128 c1_1 = b3_splat_ps( m11, 1);
-    __m128 c2_1 = b3_splat_ps( m12, 1);
-    
-    c0_1 = _mm_mul_ps(c0_1, m2v);
-    c1_1 = _mm_mul_ps(c1_1, m2v);
-    c2_1 = _mm_mul_ps(c2_1, m2v);
-    
-    m2v = _mm_and_ps(m2[2].mVec128, b3vFFF0fMask);
-    
-    c0 = _mm_add_ps(c0, c0_1);
-    c1 = _mm_add_ps(c1, c1_1);
-    c2 = _mm_add_ps(c2, c2_1);
-    
-    m10 = b3_splat_ps( m10, 2);
-    m11 = b3_splat_ps( m11, 2);
-    m12 = b3_splat_ps( m12, 2);
-    
-    m10 = _mm_mul_ps(m10, m2v);
-    m11 = _mm_mul_ps(m11, m2v);
-    m12 = _mm_mul_ps(m12, m2v);
-    
-    c0 = _mm_add_ps(c0, m10);
-    c1 = _mm_add_ps(c1, m11);
-    c2 = _mm_add_ps(c2, m12);
-    
-    return b3Matrix3x3(c0, c1, c2);
+	__m128 m10 = m1[0].mVec128;
+	__m128 m11 = m1[1].mVec128;
+	__m128 m12 = m1[2].mVec128;
+
+	__m128 m2v = _mm_and_ps(m2[0].mVec128, b3vFFF0fMask);
+
+	__m128 c0 = b3_splat_ps(m10, 0);
+	__m128 c1 = b3_splat_ps(m11, 0);
+	__m128 c2 = b3_splat_ps(m12, 0);
+
+	c0 = _mm_mul_ps(c0, m2v);
+	c1 = _mm_mul_ps(c1, m2v);
+	c2 = _mm_mul_ps(c2, m2v);
+
+	m2v = _mm_and_ps(m2[1].mVec128, b3vFFF0fMask);
+
+	__m128 c0_1 = b3_splat_ps(m10, 1);
+	__m128 c1_1 = b3_splat_ps(m11, 1);
+	__m128 c2_1 = b3_splat_ps(m12, 1);
+
+	c0_1 = _mm_mul_ps(c0_1, m2v);
+	c1_1 = _mm_mul_ps(c1_1, m2v);
+	c2_1 = _mm_mul_ps(c2_1, m2v);
+
+	m2v = _mm_and_ps(m2[2].mVec128, b3vFFF0fMask);
+
+	c0 = _mm_add_ps(c0, c0_1);
+	c1 = _mm_add_ps(c1, c1_1);
+	c2 = _mm_add_ps(c2, c2_1);
+
+	m10 = b3_splat_ps(m10, 2);
+	m11 = b3_splat_ps(m11, 2);
+	m12 = b3_splat_ps(m12, 2);
+
+	m10 = _mm_mul_ps(m10, m2v);
+	m11 = _mm_mul_ps(m11, m2v);
+	m12 = _mm_mul_ps(m12, m2v);
+
+	c0 = _mm_add_ps(c0, m10);
+	c1 = _mm_add_ps(c1, m11);
+	c2 = _mm_add_ps(c2, m12);
+
+	return b3Matrix3x3(c0, c1, c2);
 
 #elif defined(B3_USE_NEON)
 
-    float32x4_t rv0, rv1, rv2;
-    float32x4_t v0, v1, v2;
-    float32x4_t mv0, mv1, mv2;
+	float32x4_t rv0, rv1, rv2;
+	float32x4_t v0, v1, v2;
+	float32x4_t mv0, mv1, mv2;
 
-    v0 = m1[0].mVec128;
-    v1 = m1[1].mVec128;
-    v2 = m1[2].mVec128;
+	v0 = m1[0].mVec128;
+	v1 = m1[1].mVec128;
+	v2 = m1[2].mVec128;
 
-    mv0 = (float32x4_t) vandq_s32((int32x4_t)m2[0].mVec128, b3vFFF0Mask); 
-    mv1 = (float32x4_t) vandq_s32((int32x4_t)m2[1].mVec128, b3vFFF0Mask); 
-    mv2 = (float32x4_t) vandq_s32((int32x4_t)m2[2].mVec128, b3vFFF0Mask); 
-    
-    rv0 = vmulq_lane_f32(mv0, vget_low_f32(v0), 0);
-    rv1 = vmulq_lane_f32(mv0, vget_low_f32(v1), 0);
-    rv2 = vmulq_lane_f32(mv0, vget_low_f32(v2), 0);
-    
-    rv0 = vmlaq_lane_f32(rv0, mv1, vget_low_f32(v0), 1);
-    rv1 = vmlaq_lane_f32(rv1, mv1, vget_low_f32(v1), 1);
-    rv2 = vmlaq_lane_f32(rv2, mv1, vget_low_f32(v2), 1);
-    
-    rv0 = vmlaq_lane_f32(rv0, mv2, vget_high_f32(v0), 0);
-    rv1 = vmlaq_lane_f32(rv1, mv2, vget_high_f32(v1), 0);
-    rv2 = vmlaq_lane_f32(rv2, mv2, vget_high_f32(v2), 0);
+	mv0 = (float32x4_t)vandq_s32((int32x4_t)m2[0].mVec128, b3vFFF0Mask);
+	mv1 = (float32x4_t)vandq_s32((int32x4_t)m2[1].mVec128, b3vFFF0Mask);
+	mv2 = (float32x4_t)vandq_s32((int32x4_t)m2[2].mVec128, b3vFFF0Mask);
+
+	rv0 = vmulq_lane_f32(mv0, vget_low_f32(v0), 0);
+	rv1 = vmulq_lane_f32(mv0, vget_low_f32(v1), 0);
+	rv2 = vmulq_lane_f32(mv0, vget_low_f32(v2), 0);
+
+	rv0 = vmlaq_lane_f32(rv0, mv1, vget_low_f32(v0), 1);
+	rv1 = vmlaq_lane_f32(rv1, mv1, vget_low_f32(v1), 1);
+	rv2 = vmlaq_lane_f32(rv2, mv1, vget_low_f32(v2), 1);
+
+	rv0 = vmlaq_lane_f32(rv0, mv2, vget_high_f32(v0), 0);
+	rv1 = vmlaq_lane_f32(rv1, mv2, vget_high_f32(v1), 0);
+	rv2 = vmlaq_lane_f32(rv2, mv2, vget_high_f32(v2), 0);
 
 	return b3Matrix3x3(rv0, rv1, rv2);
-        
-#else	
+
+#else
 	return b3Matrix3x3(
-		m2.tdotx( m1[0]), m2.tdoty( m1[0]), m2.tdotz( m1[0]),
-		m2.tdotx( m1[1]), m2.tdoty( m1[1]), m2.tdotz( m1[1]),
-		m2.tdotx( m1[2]), m2.tdoty( m1[2]), m2.tdotz( m1[2]));
+		m2.tdotx(m1[0]), m2.tdoty(m1[0]), m2.tdotz(m1[0]),
+		m2.tdotx(m1[1]), m2.tdoty(m1[1]), m2.tdotz(m1[1]),
+		m2.tdotx(m1[2]), m2.tdoty(m1[2]), m2.tdotz(m1[2]));
 #endif
 }
 
@@ -1292,71 +1290,65 @@ m1[0][2] * m2[0][2] + m1[1][2] * m2[1][2] + m1[2][2] * m2[2][2]);
 * It will test all elements are equal.  */
 B3_FORCE_INLINE bool operator==(const b3Matrix3x3& m1, const b3Matrix3x3& m2)
 {
-#if (defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE))
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE))
 
-    __m128 c0, c1, c2;
+	__m128 c0, c1, c2;
 
-    c0 = _mm_cmpeq_ps(m1[0].mVec128, m2[0].mVec128);
-    c1 = _mm_cmpeq_ps(m1[1].mVec128, m2[1].mVec128);
-    c2 = _mm_cmpeq_ps(m1[2].mVec128, m2[2].mVec128);
-    
-    c0 = _mm_and_ps(c0, c1);
-    c0 = _mm_and_ps(c0, c2);
+	c0 = _mm_cmpeq_ps(m1[0].mVec128, m2[0].mVec128);
+	c1 = _mm_cmpeq_ps(m1[1].mVec128, m2[1].mVec128);
+	c2 = _mm_cmpeq_ps(m1[2].mVec128, m2[2].mVec128);
 
-    return (0x7 == _mm_movemask_ps((__m128)c0));
-#else 
-	return 
-    (   m1[0][0] == m2[0][0] && m1[1][0] == m2[1][0] && m1[2][0] == m2[2][0] &&
-		m1[0][1] == m2[0][1] && m1[1][1] == m2[1][1] && m1[2][1] == m2[2][1] &&
-		m1[0][2] == m2[0][2] && m1[1][2] == m2[1][2] && m1[2][2] == m2[2][2] );
+	c0 = _mm_and_ps(c0, c1);
+	c0 = _mm_and_ps(c0, c2);
+
+	return (0x7 == _mm_movemask_ps((__m128)c0));
+#else
+	return (m1[0][0] == m2[0][0] && m1[1][0] == m2[1][0] && m1[2][0] == m2[2][0] &&
+			m1[0][1] == m2[0][1] && m1[1][1] == m2[1][1] && m1[2][1] == m2[2][1] &&
+			m1[0][2] == m2[0][2] && m1[1][2] == m2[1][2] && m1[2][2] == m2[2][2]);
 #endif
 }
 
 ///for serialization
-struct	b3Matrix3x3FloatData
+struct b3Matrix3x3FloatData
 {
 	b3Vector3FloatData m_el[3];
 };
 
 ///for serialization
-struct	b3Matrix3x3DoubleData
+struct b3Matrix3x3DoubleData
 {
 	b3Vector3DoubleData m_el[3];
 };
 
-
-	
-
-B3_FORCE_INLINE	void	b3Matrix3x3::serialize(struct	b3Matrix3x3Data& dataOut) const
+B3_FORCE_INLINE void b3Matrix3x3::serialize(struct b3Matrix3x3Data& dataOut) const
 {
-	for (int i=0;i<3;i++)
+	for (int i = 0; i < 3; i++)
 		m_el[i].serialize(dataOut.m_el[i]);
 }
 
-B3_FORCE_INLINE	void	b3Matrix3x3::serializeFloat(struct	b3Matrix3x3FloatData& dataOut) const
+B3_FORCE_INLINE void b3Matrix3x3::serializeFloat(struct b3Matrix3x3FloatData& dataOut) const
 {
-	for (int i=0;i<3;i++)
+	for (int i = 0; i < 3; i++)
 		m_el[i].serializeFloat(dataOut.m_el[i]);
 }
 
-
-B3_FORCE_INLINE	void	b3Matrix3x3::deSerialize(const struct	b3Matrix3x3Data& dataIn)
+B3_FORCE_INLINE void b3Matrix3x3::deSerialize(const struct b3Matrix3x3Data& dataIn)
 {
-	for (int i=0;i<3;i++)
+	for (int i = 0; i < 3; i++)
 		m_el[i].deSerialize(dataIn.m_el[i]);
 }
 
-B3_FORCE_INLINE	void	b3Matrix3x3::deSerializeFloat(const struct	b3Matrix3x3FloatData& dataIn)
+B3_FORCE_INLINE void b3Matrix3x3::deSerializeFloat(const struct b3Matrix3x3FloatData& dataIn)
 {
-	for (int i=0;i<3;i++)
+	for (int i = 0; i < 3; i++)
 		m_el[i].deSerializeFloat(dataIn.m_el[i]);
 }
 
-B3_FORCE_INLINE	void	b3Matrix3x3::deSerializeDouble(const struct	b3Matrix3x3DoubleData& dataIn)
+B3_FORCE_INLINE void b3Matrix3x3::deSerializeDouble(const struct b3Matrix3x3DoubleData& dataIn)
 {
-	for (int i=0;i<3;i++)
+	for (int i = 0; i < 3; i++)
 		m_el[i].deSerializeDouble(dataIn.m_el[i]);
 }
 
-#endif //B3_MATRIX3x3_H
-
+#endif  //B3_MATRIX3x3_H
diff --git a/thirdparty/bullet/Bullet3Common/b3MinMax.h b/thirdparty/bullet/Bullet3Common/b3MinMax.h
index 73af23a4f97..c09c3db3f5a 100644
--- a/thirdparty/bullet/Bullet3Common/b3MinMax.h
+++ b/thirdparty/bullet/Bullet3Common/b3MinMax.h
@@ -12,60 +12,58 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #ifndef B3_GEN_MINMAX_H
 #define B3_GEN_MINMAX_H
 
 #include "b3Scalar.h"
 
 template <class T>
-B3_FORCE_INLINE const T& b3Min(const T& a, const T& b) 
+B3_FORCE_INLINE const T& b3Min(const T& a, const T& b)
 {
-  return a < b ? a : b ;
+	return a < b ? a : b;
 }
 
 template <class T>
-B3_FORCE_INLINE const T& b3Max(const T& a, const T& b) 
+B3_FORCE_INLINE const T& b3Max(const T& a, const T& b)
 {
-  return  a > b ? a : b;
+	return a > b ? a : b;
 }
 
 template <class T>
-B3_FORCE_INLINE const T& b3Clamped(const T& a, const T& lb, const T& ub) 
+B3_FORCE_INLINE const T& b3Clamped(const T& a, const T& lb, const T& ub)
 {
-	return a < lb ? lb : (ub < a ? ub : a); 
+	return a < lb ? lb : (ub < a ? ub : a);
 }
 
 template <class T>
-B3_FORCE_INLINE void b3SetMin(T& a, const T& b) 
+B3_FORCE_INLINE void b3SetMin(T& a, const T& b)
 {
-    if (b < a) 
+	if (b < a)
 	{
 		a = b;
 	}
 }
 
 template <class T>
-B3_FORCE_INLINE void b3SetMax(T& a, const T& b) 
+B3_FORCE_INLINE void b3SetMax(T& a, const T& b)
 {
-    if (a < b) 
+	if (a < b)
 	{
 		a = b;
 	}
 }
 
 template <class T>
-B3_FORCE_INLINE void b3Clamp(T& a, const T& lb, const T& ub) 
+B3_FORCE_INLINE void b3Clamp(T& a, const T& lb, const T& ub)
 {
-	if (a < lb) 
+	if (a < lb)
 	{
-		a = lb; 
+		a = lb;
 	}
-	else if (ub < a) 
+	else if (ub < a)
 	{
 		a = ub;
 	}
 }
 
-#endif //B3_GEN_MINMAX_H
+#endif  //B3_GEN_MINMAX_H
diff --git a/thirdparty/bullet/Bullet3Common/b3PoolAllocator.h b/thirdparty/bullet/Bullet3Common/b3PoolAllocator.h
index 2fcdcf5b249..ed56bc627dc 100644
--- a/thirdparty/bullet/Bullet3Common/b3PoolAllocator.h
+++ b/thirdparty/bullet/Bullet3Common/b3PoolAllocator.h
@@ -12,7 +12,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef _BT_POOL_ALLOCATOR_H
 #define _BT_POOL_ALLOCATOR_H
 
@@ -22,37 +21,37 @@ subject to the following restrictions:
 ///The b3PoolAllocator class allows to efficiently allocate a large pool of objects, instead of dynamically allocating them separately.
 class b3PoolAllocator
 {
-	int				m_elemSize;
-	int				m_maxElements;
-	int				m_freeCount;
-	void*			m_firstFree;
-	unsigned char*	m_pool;
+	int m_elemSize;
+	int m_maxElements;
+	int m_freeCount;
+	void* m_firstFree;
+	unsigned char* m_pool;
 
 public:
-
 	b3PoolAllocator(int elemSize, int maxElements)
-		:m_elemSize(elemSize),
-		m_maxElements(maxElements)
+		: m_elemSize(elemSize),
+		  m_maxElements(maxElements)
 	{
-		m_pool = (unsigned char*) b3AlignedAlloc( static_cast<unsigned int>(m_elemSize*m_maxElements),16);
+		m_pool = (unsigned char*)b3AlignedAlloc(static_cast<unsigned int>(m_elemSize * m_maxElements), 16);
 
 		unsigned char* p = m_pool;
-        m_firstFree = p;
-        m_freeCount = m_maxElements;
-        int count = m_maxElements;
-        while (--count) {
-            *(void**)p = (p + m_elemSize);
-            p += m_elemSize;
-        }
-        *(void**)p = 0;
-    }
+		m_firstFree = p;
+		m_freeCount = m_maxElements;
+		int count = m_maxElements;
+		while (--count)
+		{
+			*(void**)p = (p + m_elemSize);
+			p += m_elemSize;
+		}
+		*(void**)p = 0;
+	}
 
 	~b3PoolAllocator()
 	{
-		b3AlignedFree( m_pool);
+		b3AlignedFree(m_pool);
 	}
 
-	int	getFreeCount() const
+	int getFreeCount() const
 	{
 		return m_freeCount;
 	}
@@ -67,21 +66,22 @@ public:
 		return m_maxElements;
 	}
 
-	void*	allocate(int size)
+	void* allocate(int size)
 	{
 		// release mode fix
 		(void)size;
-		b3Assert(!size || size<=m_elemSize);
-		b3Assert(m_freeCount>0);
-        void* result = m_firstFree;
-        m_firstFree = *(void**)m_firstFree;
-        --m_freeCount;
-        return result;
+		b3Assert(!size || size <= m_elemSize);
+		b3Assert(m_freeCount > 0);
+		void* result = m_firstFree;
+		m_firstFree = *(void**)m_firstFree;
+		--m_freeCount;
+		return result;
 	}
 
 	bool validPtr(void* ptr)
 	{
-		if (ptr) {
+		if (ptr)
+		{
 			if (((unsigned char*)ptr >= m_pool && (unsigned char*)ptr < m_pool + m_maxElements * m_elemSize))
 			{
 				return true;
@@ -90,32 +90,32 @@ public:
 		return false;
 	}
 
-	void	freeMemory(void* ptr)
+	void freeMemory(void* ptr)
 	{
-		 if (ptr) {
-            b3Assert((unsigned char*)ptr >= m_pool && (unsigned char*)ptr < m_pool + m_maxElements * m_elemSize);
+		if (ptr)
+		{
+			b3Assert((unsigned char*)ptr >= m_pool && (unsigned char*)ptr < m_pool + m_maxElements * m_elemSize);
 
-            *(void**)ptr = m_firstFree;
-            m_firstFree = ptr;
-            ++m_freeCount;
-        }
+			*(void**)ptr = m_firstFree;
+			m_firstFree = ptr;
+			++m_freeCount;
+		}
 	}
 
-	int	getElementSize() const
+	int getElementSize() const
 	{
 		return m_elemSize;
 	}
 
-	unsigned char*	getPoolAddress()
+	unsigned char* getPoolAddress()
 	{
 		return m_pool;
 	}
 
-	const unsigned char*	getPoolAddress() const
+	const unsigned char* getPoolAddress() const
 	{
 		return m_pool;
 	}
-
 };
 
-#endif //_BT_POOL_ALLOCATOR_H
+#endif  //_BT_POOL_ALLOCATOR_H
diff --git a/thirdparty/bullet/Bullet3Common/b3QuadWord.h b/thirdparty/bullet/Bullet3Common/b3QuadWord.h
index 65c95819775..0def305fac8 100644
--- a/thirdparty/bullet/Bullet3Common/b3QuadWord.h
+++ b/thirdparty/bullet/Bullet3Common/b3QuadWord.h
@@ -12,18 +12,13 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef B3_SIMD_QUADWORD_H
 #define B3_SIMD_QUADWORD_H
 
 #include "b3Scalar.h"
 #include "b3MinMax.h"
 
-
-
-
-
-#if defined (__CELLOS_LV2) && defined (__SPU__)
+#if defined(__CELLOS_LV2) && defined(__SPU__)
 #include <altivec.h>
 #endif
 
@@ -31,58 +26,64 @@ subject to the following restrictions:
  * Some issues under PS3 Linux with IBM 2.1 SDK, gcc compiler prevent from using aligned quadword.
  */
 #ifndef USE_LIBSPE2
-B3_ATTRIBUTE_ALIGNED16(class) b3QuadWord
+B3_ATTRIBUTE_ALIGNED16(class)
+b3QuadWord
 #else
 class b3QuadWord
 #endif
 {
 protected:
-
-#if defined (__SPU__) && defined (__CELLOS_LV2__)
+#if defined(__SPU__) && defined(__CELLOS_LV2__)
 	union {
 		vec_float4 mVec128;
-		b3Scalar	m_floats[4];
+		b3Scalar m_floats[4];
 	};
+
 public:
-	vec_float4	get128() const
+	vec_float4 get128() const
 	{
 		return mVec128;
 	}
 
-#else //__CELLOS_LV2__ __SPU__
+#else  //__CELLOS_LV2__ __SPU__
 
-#if defined(B3_USE_SSE) || defined(B3_USE_NEON) 
+#if defined(B3_USE_SSE) || defined(B3_USE_NEON)
 public:
 	union {
 		b3SimdFloat4 mVec128;
-		b3Scalar	m_floats[4];
-		struct {b3Scalar x,y,z,w;};
+		b3Scalar m_floats[4];
+		struct
+		{
+			b3Scalar x, y, z, w;
+		};
 	};
+
 public:
-	B3_FORCE_INLINE	b3SimdFloat4	get128() const
+	B3_FORCE_INLINE b3SimdFloat4 get128() const
 	{
 		return mVec128;
 	}
-	B3_FORCE_INLINE	void	set128(b3SimdFloat4 v128)
+	B3_FORCE_INLINE void set128(b3SimdFloat4 v128)
 	{
 		mVec128 = v128;
 	}
 #else
 public:
-	union
-	{
-		b3Scalar	m_floats[4];
-		struct {b3Scalar x,y,z,w;};
+	union {
+		b3Scalar m_floats[4];
+		struct
+		{
+			b3Scalar x, y, z, w;
+		};
 	};
-#endif // B3_USE_SSE
+#endif  // B3_USE_SSE
 
-#endif //__CELLOS_LV2__ __SPU__
+#endif  //__CELLOS_LV2__ __SPU__
 
-	public:
-  
+public:
 #if defined(B3_USE_SSE) || defined(B3_USE_NEON)
 
-	// Set Vector 
+	// Set Vector
 	B3_FORCE_INLINE b3QuadWord(const b3SimdFloat4 vec)
 	{
 		mVec128 = vec;
@@ -95,151 +96,147 @@ public:
 	}
 
 	// Assignment Operator
-	B3_FORCE_INLINE b3QuadWord& 
-	operator=(const b3QuadWord& v) 
+	B3_FORCE_INLINE b3QuadWord&
+	operator=(const b3QuadWord& v)
 	{
 		mVec128 = v.mVec128;
-		
+
 		return *this;
 	}
-	
+
 #endif
 
-  /**@brief Return the x value */
-		B3_FORCE_INLINE const b3Scalar& getX() const { return m_floats[0]; }
-  /**@brief Return the y value */
-		B3_FORCE_INLINE const b3Scalar& getY() const { return m_floats[1]; }
-  /**@brief Return the z value */
-		B3_FORCE_INLINE const b3Scalar& getZ() const { return m_floats[2]; }
-  /**@brief Set the x value */
-		B3_FORCE_INLINE void	setX(b3Scalar _x) { m_floats[0] = _x;};
-  /**@brief Set the y value */
-		B3_FORCE_INLINE void	setY(b3Scalar _y) { m_floats[1] = _y;};
-  /**@brief Set the z value */
-		B3_FORCE_INLINE void	setZ(b3Scalar _z) { m_floats[2] = _z;};
-  /**@brief Set the w value */
-		B3_FORCE_INLINE void	setW(b3Scalar _w) { m_floats[3] = _w;};
-  /**@brief Return the x value */
+	/**@brief Return the x value */
+	B3_FORCE_INLINE const b3Scalar& getX() const { return m_floats[0]; }
+	/**@brief Return the y value */
+	B3_FORCE_INLINE const b3Scalar& getY() const { return m_floats[1]; }
+	/**@brief Return the z value */
+	B3_FORCE_INLINE const b3Scalar& getZ() const { return m_floats[2]; }
+	/**@brief Set the x value */
+	B3_FORCE_INLINE void setX(b3Scalar _x) { m_floats[0] = _x; };
+	/**@brief Set the y value */
+	B3_FORCE_INLINE void setY(b3Scalar _y) { m_floats[1] = _y; };
+	/**@brief Set the z value */
+	B3_FORCE_INLINE void setZ(b3Scalar _z) { m_floats[2] = _z; };
+	/**@brief Set the w value */
+	B3_FORCE_INLINE void setW(b3Scalar _w) { m_floats[3] = _w; };
+	/**@brief Return the x value */
 
-
-	//B3_FORCE_INLINE b3Scalar&       operator[](int i)       { return (&m_floats[0])[i];	}      
+	//B3_FORCE_INLINE b3Scalar&       operator[](int i)       { return (&m_floats[0])[i];	}
 	//B3_FORCE_INLINE const b3Scalar& operator[](int i) const { return (&m_floats[0])[i]; }
 	///operator b3Scalar*() replaces operator[], using implicit conversion. We added operator != and operator == to avoid pointer comparisons.
-	B3_FORCE_INLINE	operator       b3Scalar *()       { return &m_floats[0]; }
-	B3_FORCE_INLINE	operator const b3Scalar *() const { return &m_floats[0]; }
+	B3_FORCE_INLINE operator b3Scalar*() { return &m_floats[0]; }
+	B3_FORCE_INLINE operator const b3Scalar*() const { return &m_floats[0]; }
 
-	B3_FORCE_INLINE	bool	operator==(const b3QuadWord& other) const
+	B3_FORCE_INLINE bool operator==(const b3QuadWord& other) const
 	{
 #ifdef B3_USE_SSE
-        return (0xf == _mm_movemask_ps((__m128)_mm_cmpeq_ps(mVec128, other.mVec128)));
-#else 
-		return ((m_floats[3]==other.m_floats[3]) && 
-                (m_floats[2]==other.m_floats[2]) && 
-                (m_floats[1]==other.m_floats[1]) && 
-                (m_floats[0]==other.m_floats[0]));
+		return (0xf == _mm_movemask_ps((__m128)_mm_cmpeq_ps(mVec128, other.mVec128)));
+#else
+		return ((m_floats[3] == other.m_floats[3]) &&
+				(m_floats[2] == other.m_floats[2]) &&
+				(m_floats[1] == other.m_floats[1]) &&
+				(m_floats[0] == other.m_floats[0]));
 #endif
 	}
 
-	B3_FORCE_INLINE	bool	operator!=(const b3QuadWord& other) const
+	B3_FORCE_INLINE bool operator!=(const b3QuadWord& other) const
 	{
 		return !(*this == other);
 	}
 
-  /**@brief Set x,y,z and zero w 
+	/**@brief Set x,y,z and zero w 
    * @param x Value of x
    * @param y Value of y
    * @param z Value of z
    */
-		B3_FORCE_INLINE void 	setValue(const b3Scalar& _x, const b3Scalar& _y, const b3Scalar& _z)
-		{
-			m_floats[0]=_x;
-			m_floats[1]=_y;
-			m_floats[2]=_z;
-			m_floats[3] = 0.f;
-		}
+	B3_FORCE_INLINE void setValue(const b3Scalar& _x, const b3Scalar& _y, const b3Scalar& _z)
+	{
+		m_floats[0] = _x;
+		m_floats[1] = _y;
+		m_floats[2] = _z;
+		m_floats[3] = 0.f;
+	}
 
-/*		void getValue(b3Scalar *m) const 
+	/*		void getValue(b3Scalar *m) const 
 		{
 			m[0] = m_floats[0];
 			m[1] = m_floats[1];
 			m[2] = m_floats[2];
 		}
 */
-/**@brief Set the values 
+	/**@brief Set the values 
    * @param x Value of x
    * @param y Value of y
    * @param z Value of z
    * @param w Value of w
    */
-		B3_FORCE_INLINE void	setValue(const b3Scalar& _x, const b3Scalar& _y, const b3Scalar& _z,const b3Scalar& _w)
-		{
-			m_floats[0]=_x;
-			m_floats[1]=_y;
-			m_floats[2]=_z;
-			m_floats[3]=_w;
-		}
-  /**@brief No initialization constructor */
-		B3_FORCE_INLINE b3QuadWord()
-		//	:m_floats[0](b3Scalar(0.)),m_floats[1](b3Scalar(0.)),m_floats[2](b3Scalar(0.)),m_floats[3](b3Scalar(0.))
-		{
-		}
- 
-  /**@brief Three argument constructor (zeros w)
+	B3_FORCE_INLINE void setValue(const b3Scalar& _x, const b3Scalar& _y, const b3Scalar& _z, const b3Scalar& _w)
+	{
+		m_floats[0] = _x;
+		m_floats[1] = _y;
+		m_floats[2] = _z;
+		m_floats[3] = _w;
+	}
+	/**@brief No initialization constructor */
+	B3_FORCE_INLINE b3QuadWord()
+	//	:m_floats[0](b3Scalar(0.)),m_floats[1](b3Scalar(0.)),m_floats[2](b3Scalar(0.)),m_floats[3](b3Scalar(0.))
+	{
+	}
+
+	/**@brief Three argument constructor (zeros w)
    * @param x Value of x
    * @param y Value of y
    * @param z Value of z
    */
-		B3_FORCE_INLINE b3QuadWord(const b3Scalar& _x, const b3Scalar& _y, const b3Scalar& _z)		
-		{
-			m_floats[0] = _x, m_floats[1] = _y, m_floats[2] = _z, m_floats[3] = 0.0f;
-		}
+	B3_FORCE_INLINE b3QuadWord(const b3Scalar& _x, const b3Scalar& _y, const b3Scalar& _z)
+	{
+		m_floats[0] = _x, m_floats[1] = _y, m_floats[2] = _z, m_floats[3] = 0.0f;
+	}
 
-/**@brief Initializing constructor
+	/**@brief Initializing constructor
    * @param x Value of x
    * @param y Value of y
    * @param z Value of z
    * @param w Value of w
    */
-		B3_FORCE_INLINE b3QuadWord(const b3Scalar& _x, const b3Scalar& _y, const b3Scalar& _z,const b3Scalar& _w) 
-		{
-			m_floats[0] = _x, m_floats[1] = _y, m_floats[2] = _z, m_floats[3] = _w;
-		}
+	B3_FORCE_INLINE b3QuadWord(const b3Scalar& _x, const b3Scalar& _y, const b3Scalar& _z, const b3Scalar& _w)
+	{
+		m_floats[0] = _x, m_floats[1] = _y, m_floats[2] = _z, m_floats[3] = _w;
+	}
 
-  /**@brief Set each element to the max of the current values and the values of another b3QuadWord
+	/**@brief Set each element to the max of the current values and the values of another b3QuadWord
    * @param other The other b3QuadWord to compare with 
    */
-		B3_FORCE_INLINE void	setMax(const b3QuadWord& other)
-		{
-        #ifdef B3_USE_SSE
-            mVec128 = _mm_max_ps(mVec128, other.mVec128);
-        #elif defined(B3_USE_NEON)
-            mVec128 = vmaxq_f32(mVec128, other.mVec128);
-        #else
-        	b3SetMax(m_floats[0], other.m_floats[0]);
-			b3SetMax(m_floats[1], other.m_floats[1]);
-			b3SetMax(m_floats[2], other.m_floats[2]);
-			b3SetMax(m_floats[3], other.m_floats[3]);
-		#endif
-        }
-  /**@brief Set each element to the min of the current values and the values of another b3QuadWord
+	B3_FORCE_INLINE void setMax(const b3QuadWord& other)
+	{
+#ifdef B3_USE_SSE
+		mVec128 = _mm_max_ps(mVec128, other.mVec128);
+#elif defined(B3_USE_NEON)
+		mVec128 = vmaxq_f32(mVec128, other.mVec128);
+#else
+		b3SetMax(m_floats[0], other.m_floats[0]);
+		b3SetMax(m_floats[1], other.m_floats[1]);
+		b3SetMax(m_floats[2], other.m_floats[2]);
+		b3SetMax(m_floats[3], other.m_floats[3]);
+#endif
+	}
+	/**@brief Set each element to the min of the current values and the values of another b3QuadWord
    * @param other The other b3QuadWord to compare with 
    */
-		B3_FORCE_INLINE void	setMin(const b3QuadWord& other)
-		{
-        #ifdef B3_USE_SSE
-            mVec128 = _mm_min_ps(mVec128, other.mVec128);
-        #elif defined(B3_USE_NEON)
-            mVec128 = vminq_f32(mVec128, other.mVec128);
-        #else
-        	b3SetMin(m_floats[0], other.m_floats[0]);
-			b3SetMin(m_floats[1], other.m_floats[1]);
-			b3SetMin(m_floats[2], other.m_floats[2]);
-			b3SetMin(m_floats[3], other.m_floats[3]);
-		#endif
-        }
-
-
-
+	B3_FORCE_INLINE void setMin(const b3QuadWord& other)
+	{
+#ifdef B3_USE_SSE
+		mVec128 = _mm_min_ps(mVec128, other.mVec128);
+#elif defined(B3_USE_NEON)
+		mVec128 = vminq_f32(mVec128, other.mVec128);
+#else
+		b3SetMin(m_floats[0], other.m_floats[0]);
+		b3SetMin(m_floats[1], other.m_floats[1]);
+		b3SetMin(m_floats[2], other.m_floats[2]);
+		b3SetMin(m_floats[3], other.m_floats[3]);
+#endif
+	}
 };
 
-#endif //B3_SIMD_QUADWORD_H
+#endif  //B3_SIMD_QUADWORD_H
diff --git a/thirdparty/bullet/Bullet3Common/b3Quaternion.h b/thirdparty/bullet/Bullet3Common/b3Quaternion.h
index ad205433487..9bd5ff7d90e 100644
--- a/thirdparty/bullet/Bullet3Common/b3Quaternion.h
+++ b/thirdparty/bullet/Bullet3Common/b3Quaternion.h
@@ -12,19 +12,12 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #ifndef B3_SIMD__QUATERNION_H_
 #define B3_SIMD__QUATERNION_H_
 
-
 #include "b3Vector3.h"
 #include "b3QuadWord.h"
 
-
-
-
-
 #ifdef B3_USE_SSE
 
 const __m128 B3_ATTRIBUTE_ALIGNED16(b3vOnes) = {1.0f, 1.0f, 1.0f, 1.0f};
@@ -39,13 +32,14 @@ const b3SimdFloat4 B3_ATTRIBUTE_ALIGNED16(b3vPPPM) = {+0.0f, +0.0f, +0.0f, -0.0f
 #endif
 
 /**@brief The b3Quaternion implements quaternion to perform linear algebra rotations in combination with b3Matrix3x3, b3Vector3 and b3Transform. */
-class b3Quaternion : public b3QuadWord {
+class b3Quaternion : public b3QuadWord
+{
 public:
-  /**@brief No initialization constructor */
+	/**@brief No initialization constructor */
 	b3Quaternion() {}
 
-#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE))|| defined(B3_USE_NEON) 
-	// Set Vector 
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)) || defined(B3_USE_NEON)
+	// Set Vector
 	B3_FORCE_INLINE b3Quaternion(const b3SimdFloat4 vec)
 	{
 		mVec128 = vec;
@@ -58,63 +52,70 @@ public:
 	}
 
 	// Assignment Operator
-	B3_FORCE_INLINE b3Quaternion& 
-	operator=(const b3Quaternion& v) 
+	B3_FORCE_INLINE b3Quaternion&
+	operator=(const b3Quaternion& v)
 	{
 		mVec128 = v.mVec128;
-		
+
 		return *this;
 	}
-	
+
 #endif
 
 	//		template <typename b3Scalar>
 	//		explicit Quaternion(const b3Scalar *v) : Tuple4<b3Scalar>(v) {}
-  /**@brief Constructor from scalars */
-	b3Quaternion(const b3Scalar& _x, const b3Scalar& _y, const b3Scalar& _z, const b3Scalar& _w) 
-		: b3QuadWord(_x, _y, _z, _w) 
+	/**@brief Constructor from scalars */
+	b3Quaternion(const b3Scalar& _x, const b3Scalar& _y, const b3Scalar& _z, const b3Scalar& _w)
+		: b3QuadWord(_x, _y, _z, _w)
 	{
 		//b3Assert(!((_x==1.f) && (_y==0.f) && (_z==0.f) && (_w==0.f)));
 	}
-  /**@brief Axis angle Constructor
+	/**@brief Axis angle Constructor
    * @param axis The axis which the rotation is around
    * @param angle The magnitude of the rotation around the angle (Radians) */
-	b3Quaternion(const b3Vector3& _axis, const b3Scalar& _angle) 
-	{ 
-		setRotation(_axis, _angle); 
+	b3Quaternion(const b3Vector3& _axis, const b3Scalar& _angle)
+	{
+		setRotation(_axis, _angle);
 	}
-  /**@brief Constructor from Euler angles
+	/**@brief Constructor from Euler angles
    * @param yaw Angle around Y unless B3_EULER_DEFAULT_ZYX defined then Z
    * @param pitch Angle around X unless B3_EULER_DEFAULT_ZYX defined then Y
    * @param roll Angle around Z unless B3_EULER_DEFAULT_ZYX defined then X */
 	b3Quaternion(const b3Scalar& yaw, const b3Scalar& pitch, const b3Scalar& roll)
-	{ 
+	{
 #ifndef B3_EULER_DEFAULT_ZYX
-		setEuler(yaw, pitch, roll); 
+		setEuler(yaw, pitch, roll);
 #else
-		setEulerZYX(yaw, pitch, roll); 
-#endif 
+		setEulerZYX(yaw, pitch, roll);
+#endif
 	}
-  /**@brief Set the rotation using axis angle notation 
+	/**@brief Set the rotation using axis angle notation 
    * @param axis The axis around which to rotate
    * @param angle The magnitude of the rotation in Radians */
 	void setRotation(const b3Vector3& axis, const b3Scalar& _angle)
 	{
 		b3Scalar d = axis.length();
 		b3Assert(d != b3Scalar(0.0));
-		b3Scalar s = b3Sin(_angle * b3Scalar(0.5)) / d;
-		setValue(axis.getX() * s, axis.getY() * s, axis.getZ() * s, 
-			b3Cos(_angle * b3Scalar(0.5)));
+		if (d < B3_EPSILON)
+		{
+			setValue(0, 0, 0, 1);
+		}
+		else
+		{
+			b3Scalar s = b3Sin(_angle * b3Scalar(0.5)) / d;
+			setValue(axis.getX() * s, axis.getY() * s, axis.getZ() * s,
+				b3Cos(_angle * b3Scalar(0.5)));
+		}
 	}
-  /**@brief Set the quaternion using Euler angles
+	/**@brief Set the quaternion using Euler angles
    * @param yaw Angle around Y
    * @param pitch Angle around X
    * @param roll Angle around Z */
 	void setEuler(const b3Scalar& yaw, const b3Scalar& pitch, const b3Scalar& roll)
 	{
-		b3Scalar halfYaw = b3Scalar(yaw) * b3Scalar(0.5);  
-		b3Scalar halfPitch = b3Scalar(pitch) * b3Scalar(0.5);  
-		b3Scalar halfRoll = b3Scalar(roll) * b3Scalar(0.5);  
+		b3Scalar halfYaw = b3Scalar(yaw) * b3Scalar(0.5);
+		b3Scalar halfPitch = b3Scalar(pitch) * b3Scalar(0.5);
+		b3Scalar halfRoll = b3Scalar(roll) * b3Scalar(0.5);
 		b3Scalar cosYaw = b3Cos(halfYaw);
 		b3Scalar sinYaw = b3Sin(halfYaw);
 		b3Scalar cosPitch = b3Cos(halfPitch);
@@ -122,34 +123,34 @@ public:
 		b3Scalar cosRoll = b3Cos(halfRoll);
 		b3Scalar sinRoll = b3Sin(halfRoll);
 		setValue(cosRoll * sinPitch * cosYaw + sinRoll * cosPitch * sinYaw,
-			cosRoll * cosPitch * sinYaw - sinRoll * sinPitch * cosYaw,
-			sinRoll * cosPitch * cosYaw - cosRoll * sinPitch * sinYaw,
-			cosRoll * cosPitch * cosYaw + sinRoll * sinPitch * sinYaw);
+				 cosRoll * cosPitch * sinYaw - sinRoll * sinPitch * cosYaw,
+				 sinRoll * cosPitch * cosYaw - cosRoll * sinPitch * sinYaw,
+				 cosRoll * cosPitch * cosYaw + sinRoll * sinPitch * sinYaw);
 	}
- 
+
 	/**@brief Set the quaternion using euler angles 
    * @param yaw Angle around Z
    * @param pitch Angle around Y
    * @param roll Angle around X */
 	void setEulerZYX(const b3Scalar& yawZ, const b3Scalar& pitchY, const b3Scalar& rollX)
 	{
-		b3Scalar halfYaw = b3Scalar(yawZ) * b3Scalar(0.5);  
-		b3Scalar halfPitch = b3Scalar(pitchY) * b3Scalar(0.5);  
-		b3Scalar halfRoll = b3Scalar(rollX) * b3Scalar(0.5);  
+		b3Scalar halfYaw = b3Scalar(yawZ) * b3Scalar(0.5);
+		b3Scalar halfPitch = b3Scalar(pitchY) * b3Scalar(0.5);
+		b3Scalar halfRoll = b3Scalar(rollX) * b3Scalar(0.5);
 		b3Scalar cosYaw = b3Cos(halfYaw);
 		b3Scalar sinYaw = b3Sin(halfYaw);
 		b3Scalar cosPitch = b3Cos(halfPitch);
 		b3Scalar sinPitch = b3Sin(halfPitch);
 		b3Scalar cosRoll = b3Cos(halfRoll);
 		b3Scalar sinRoll = b3Sin(halfRoll);
-		setValue(sinRoll * cosPitch * cosYaw - cosRoll * sinPitch * sinYaw, //x
-                         cosRoll * sinPitch * cosYaw + sinRoll * cosPitch * sinYaw, //y
-                         cosRoll * cosPitch * sinYaw - sinRoll * sinPitch * cosYaw, //z
-                         cosRoll * cosPitch * cosYaw + sinRoll * sinPitch * sinYaw); //formerly yzx
+		setValue(sinRoll * cosPitch * cosYaw - cosRoll * sinPitch * sinYaw,   //x
+				 cosRoll * sinPitch * cosYaw + sinRoll * cosPitch * sinYaw,   //y
+				 cosRoll * cosPitch * sinYaw - sinRoll * sinPitch * cosYaw,   //z
+				 cosRoll * cosPitch * cosYaw + sinRoll * sinPitch * sinYaw);  //formerly yzx
 		normalize();
 	}
 
-	  /**@brief Get the euler angles from this quaternion
+	/**@brief Get the euler angles from this quaternion
 	   * @param yaw Angle around Z
 	   * @param pitch Angle around Y
 	   * @param roll Angle around X */
@@ -166,221 +167,221 @@ public:
 		squ = m_floats[3] * m_floats[3];
 		rollX = b3Atan2(2 * (m_floats[1] * m_floats[2] + m_floats[3] * m_floats[0]), squ - sqx - sqy + sqz);
 		sarg = b3Scalar(-2.) * (m_floats[0] * m_floats[2] - m_floats[3] * m_floats[1]);
-		pitchY = sarg <= b3Scalar(-1.0) ? b3Scalar(-0.5) * B3_PI: (sarg >= b3Scalar(1.0) ? b3Scalar(0.5) * B3_PI : b3Asin(sarg));
+		pitchY = sarg <= b3Scalar(-1.0) ? b3Scalar(-0.5) * B3_PI : (sarg >= b3Scalar(1.0) ? b3Scalar(0.5) * B3_PI : b3Asin(sarg));
 		yawZ = b3Atan2(2 * (m_floats[0] * m_floats[1] + m_floats[3] * m_floats[2]), squ + sqx - sqy - sqz);
 	}
 
-  /**@brief Add two quaternions
+	/**@brief Add two quaternions
    * @param q The quaternion to add to this one */
-	B3_FORCE_INLINE	b3Quaternion& operator+=(const b3Quaternion& q)
+	B3_FORCE_INLINE b3Quaternion& operator+=(const b3Quaternion& q)
 	{
-#if defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 		mVec128 = _mm_add_ps(mVec128, q.mVec128);
 #elif defined(B3_USE_NEON)
 		mVec128 = vaddq_f32(mVec128, q.mVec128);
-#else	
-		m_floats[0] += q.getX(); 
-        m_floats[1] += q.getY(); 
-        m_floats[2] += q.getZ(); 
-        m_floats[3] += q.m_floats[3];
+#else
+		m_floats[0] += q.getX();
+		m_floats[1] += q.getY();
+		m_floats[2] += q.getZ();
+		m_floats[3] += q.m_floats[3];
 #endif
 		return *this;
 	}
 
-  /**@brief Subtract out a quaternion
+	/**@brief Subtract out a quaternion
    * @param q The quaternion to subtract from this one */
-	b3Quaternion& operator-=(const b3Quaternion& q) 
+	b3Quaternion& operator-=(const b3Quaternion& q)
 	{
-#if defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 		mVec128 = _mm_sub_ps(mVec128, q.mVec128);
 #elif defined(B3_USE_NEON)
 		mVec128 = vsubq_f32(mVec128, q.mVec128);
-#else	
-		m_floats[0] -= q.getX(); 
-        m_floats[1] -= q.getY(); 
-        m_floats[2] -= q.getZ(); 
-        m_floats[3] -= q.m_floats[3];
+#else
+		m_floats[0] -= q.getX();
+		m_floats[1] -= q.getY();
+		m_floats[2] -= q.getZ();
+		m_floats[3] -= q.m_floats[3];
 #endif
-        return *this;
+		return *this;
 	}
 
-  /**@brief Scale this quaternion
+	/**@brief Scale this quaternion
    * @param s The scalar to scale by */
 	b3Quaternion& operator*=(const b3Scalar& s)
 	{
-#if defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
-		__m128	vs = _mm_load_ss(&s);	//	(S 0 0 0)
-		vs = b3_pshufd_ps(vs, 0);	//	(S S S S)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
+		__m128 vs = _mm_load_ss(&s);  //	(S 0 0 0)
+		vs = b3_pshufd_ps(vs, 0);     //	(S S S S)
 		mVec128 = _mm_mul_ps(mVec128, vs);
 #elif defined(B3_USE_NEON)
 		mVec128 = vmulq_n_f32(mVec128, s);
 #else
-		m_floats[0] *= s; 
-        m_floats[1] *= s; 
-        m_floats[2] *= s; 
-        m_floats[3] *= s;
+		m_floats[0] *= s;
+		m_floats[1] *= s;
+		m_floats[2] *= s;
+		m_floats[3] *= s;
 #endif
 		return *this;
 	}
 
-  /**@brief Multiply this quaternion by q on the right
+	/**@brief Multiply this quaternion by q on the right
    * @param q The other quaternion 
    * Equivilant to this = this * q */
 	b3Quaternion& operator*=(const b3Quaternion& q)
 	{
-#if defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 		__m128 vQ2 = q.get128();
-		
-		__m128 A1 = b3_pshufd_ps(mVec128, B3_SHUFFLE(0,1,2,0));
-		__m128 B1 = b3_pshufd_ps(vQ2, B3_SHUFFLE(3,3,3,0));
-		
+
+		__m128 A1 = b3_pshufd_ps(mVec128, B3_SHUFFLE(0, 1, 2, 0));
+		__m128 B1 = b3_pshufd_ps(vQ2, B3_SHUFFLE(3, 3, 3, 0));
+
 		A1 = A1 * B1;
-		
-		__m128 A2 = b3_pshufd_ps(mVec128, B3_SHUFFLE(1,2,0,1));
-		__m128 B2 = b3_pshufd_ps(vQ2, B3_SHUFFLE(2,0,1,1));
-		
+
+		__m128 A2 = b3_pshufd_ps(mVec128, B3_SHUFFLE(1, 2, 0, 1));
+		__m128 B2 = b3_pshufd_ps(vQ2, B3_SHUFFLE(2, 0, 1, 1));
+
 		A2 = A2 * B2;
-		
-		B1 = b3_pshufd_ps(mVec128, B3_SHUFFLE(2,0,1,2));
-		B2 = b3_pshufd_ps(vQ2, B3_SHUFFLE(1,2,0,2));
-		
-		B1 = B1 * B2;	//	A3 *= B3
-		
-		mVec128 = b3_splat_ps(mVec128, 3);	//	A0
-		mVec128 = mVec128 * vQ2;	//	A0 * B0
-		
-		A1 = A1 + A2;	//	AB12
-		mVec128 = mVec128 - B1;	//	AB03 = AB0 - AB3 
-		A1 = _mm_xor_ps(A1, b3vPPPM);	//	change sign of the last element
-		mVec128 = mVec128+ A1;	//	AB03 + AB12
 
-#elif defined(B3_USE_NEON)     
+		B1 = b3_pshufd_ps(mVec128, B3_SHUFFLE(2, 0, 1, 2));
+		B2 = b3_pshufd_ps(vQ2, B3_SHUFFLE(1, 2, 0, 2));
 
-        float32x4_t vQ1 = mVec128;
-        float32x4_t vQ2 = q.get128();
-        float32x4_t A0, A1, B1, A2, B2, A3, B3;
-        float32x2_t vQ1zx, vQ2wx, vQ1yz, vQ2zx, vQ2yz, vQ2xz;
-        
-        {
-        float32x2x2_t tmp;
-        tmp = vtrn_f32( vget_high_f32(vQ1), vget_low_f32(vQ1) );       // {z x}, {w y}
-        vQ1zx = tmp.val[0];
+		B1 = B1 * B2;  //	A3 *= B3
 
-        tmp = vtrn_f32( vget_high_f32(vQ2), vget_low_f32(vQ2) );       // {z x}, {w y}
-        vQ2zx = tmp.val[0];
-        }
-        vQ2wx = vext_f32(vget_high_f32(vQ2), vget_low_f32(vQ2), 1); 
+		mVec128 = b3_splat_ps(mVec128, 3);  //	A0
+		mVec128 = mVec128 * vQ2;            //	A0 * B0
 
-        vQ1yz = vext_f32(vget_low_f32(vQ1), vget_high_f32(vQ1), 1);
+		A1 = A1 + A2;                  //	AB12
+		mVec128 = mVec128 - B1;        //	AB03 = AB0 - AB3
+		A1 = _mm_xor_ps(A1, b3vPPPM);  //	change sign of the last element
+		mVec128 = mVec128 + A1;        //	AB03 + AB12
 
-        vQ2yz = vext_f32(vget_low_f32(vQ2), vget_high_f32(vQ2), 1);
-        vQ2xz = vext_f32(vQ2zx, vQ2zx, 1);
+#elif defined(B3_USE_NEON)
 
-        A1 = vcombine_f32(vget_low_f32(vQ1), vQ1zx);                    // X Y  z x 
-        B1 = vcombine_f32(vdup_lane_f32(vget_high_f32(vQ2), 1), vQ2wx); // W W  W X 
+		float32x4_t vQ1 = mVec128;
+		float32x4_t vQ2 = q.get128();
+		float32x4_t A0, A1, B1, A2, B2, A3, B3;
+		float32x2_t vQ1zx, vQ2wx, vQ1yz, vQ2zx, vQ2yz, vQ2xz;
 
-        A2 = vcombine_f32(vQ1yz, vget_low_f32(vQ1));
-        B2 = vcombine_f32(vQ2zx, vdup_lane_f32(vget_low_f32(vQ2), 1));
+		{
+			float32x2x2_t tmp;
+			tmp = vtrn_f32(vget_high_f32(vQ1), vget_low_f32(vQ1));  // {z x}, {w y}
+			vQ1zx = tmp.val[0];
 
-        A3 = vcombine_f32(vQ1zx, vQ1yz);        // Z X Y Z
-        B3 = vcombine_f32(vQ2yz, vQ2xz);        // Y Z x z
+			tmp = vtrn_f32(vget_high_f32(vQ2), vget_low_f32(vQ2));  // {z x}, {w y}
+			vQ2zx = tmp.val[0];
+		}
+		vQ2wx = vext_f32(vget_high_f32(vQ2), vget_low_f32(vQ2), 1);
 
-        A1 = vmulq_f32(A1, B1);
-        A2 = vmulq_f32(A2, B2);
-        A3 = vmulq_f32(A3, B3);	//	A3 *= B3
-        A0 = vmulq_lane_f32(vQ2, vget_high_f32(vQ1), 1); //	A0 * B0
+		vQ1yz = vext_f32(vget_low_f32(vQ1), vget_high_f32(vQ1), 1);
 
-        A1 = vaddq_f32(A1, A2);	//	AB12 = AB1 + AB2
-        A0 = vsubq_f32(A0, A3);	//	AB03 = AB0 - AB3 
-        
-        //	change the sign of the last element
-        A1 = (b3SimdFloat4)veorq_s32((int32x4_t)A1, (int32x4_t)b3vPPPM);	
-        A0 = vaddq_f32(A0, A1);	//	AB03 + AB12
-        
-        mVec128 = A0;
+		vQ2yz = vext_f32(vget_low_f32(vQ2), vget_high_f32(vQ2), 1);
+		vQ2xz = vext_f32(vQ2zx, vQ2zx, 1);
+
+		A1 = vcombine_f32(vget_low_f32(vQ1), vQ1zx);                     // X Y  z x
+		B1 = vcombine_f32(vdup_lane_f32(vget_high_f32(vQ2), 1), vQ2wx);  // W W  W X
+
+		A2 = vcombine_f32(vQ1yz, vget_low_f32(vQ1));
+		B2 = vcombine_f32(vQ2zx, vdup_lane_f32(vget_low_f32(vQ2), 1));
+
+		A3 = vcombine_f32(vQ1zx, vQ1yz);  // Z X Y Z
+		B3 = vcombine_f32(vQ2yz, vQ2xz);  // Y Z x z
+
+		A1 = vmulq_f32(A1, B1);
+		A2 = vmulq_f32(A2, B2);
+		A3 = vmulq_f32(A3, B3);                           //	A3 *= B3
+		A0 = vmulq_lane_f32(vQ2, vget_high_f32(vQ1), 1);  //	A0 * B0
+
+		A1 = vaddq_f32(A1, A2);  //	AB12 = AB1 + AB2
+		A0 = vsubq_f32(A0, A3);  //	AB03 = AB0 - AB3
+
+		//	change the sign of the last element
+		A1 = (b3SimdFloat4)veorq_s32((int32x4_t)A1, (int32x4_t)b3vPPPM);
+		A0 = vaddq_f32(A0, A1);  //	AB03 + AB12
+
+		mVec128 = A0;
 #else
 		setValue(
-            m_floats[3] * q.getX() + m_floats[0] * q.m_floats[3] + m_floats[1] * q.getZ() - m_floats[2] * q.getY(),
+			m_floats[3] * q.getX() + m_floats[0] * q.m_floats[3] + m_floats[1] * q.getZ() - m_floats[2] * q.getY(),
 			m_floats[3] * q.getY() + m_floats[1] * q.m_floats[3] + m_floats[2] * q.getX() - m_floats[0] * q.getZ(),
 			m_floats[3] * q.getZ() + m_floats[2] * q.m_floats[3] + m_floats[0] * q.getY() - m_floats[1] * q.getX(),
 			m_floats[3] * q.m_floats[3] - m_floats[0] * q.getX() - m_floats[1] * q.getY() - m_floats[2] * q.getZ());
 #endif
 		return *this;
 	}
-  /**@brief Return the dot product between this quaternion and another
+	/**@brief Return the dot product between this quaternion and another
    * @param q The other quaternion */
 	b3Scalar dot(const b3Quaternion& q) const
 	{
-#if defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
-		__m128	vd;
-		
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
+		__m128 vd;
+
 		vd = _mm_mul_ps(mVec128, q.mVec128);
-		
-        __m128 t = _mm_movehl_ps(vd, vd);
+
+		__m128 t = _mm_movehl_ps(vd, vd);
 		vd = _mm_add_ps(vd, t);
 		t = _mm_shuffle_ps(vd, vd, 0x55);
 		vd = _mm_add_ss(vd, t);
-		
-        return _mm_cvtss_f32(vd);
+
+		return _mm_cvtss_f32(vd);
 #elif defined(B3_USE_NEON)
 		float32x4_t vd = vmulq_f32(mVec128, q.mVec128);
-		float32x2_t x = vpadd_f32(vget_low_f32(vd), vget_high_f32(vd));  
+		float32x2_t x = vpadd_f32(vget_low_f32(vd), vget_high_f32(vd));
 		x = vpadd_f32(x, x);
 		return vget_lane_f32(x, 0);
-#else    
-		return  m_floats[0] * q.getX() + 
-                m_floats[1] * q.getY() + 
-                m_floats[2] * q.getZ() + 
-                m_floats[3] * q.m_floats[3];
+#else
+		return m_floats[0] * q.getX() +
+			   m_floats[1] * q.getY() +
+			   m_floats[2] * q.getZ() +
+			   m_floats[3] * q.m_floats[3];
 #endif
 	}
 
-  /**@brief Return the length squared of the quaternion */
+	/**@brief Return the length squared of the quaternion */
 	b3Scalar length2() const
 	{
 		return dot(*this);
 	}
 
-  /**@brief Return the length of the quaternion */
+	/**@brief Return the length of the quaternion */
 	b3Scalar length() const
 	{
 		return b3Sqrt(length2());
 	}
 
-  /**@brief Normalize the quaternion 
+	/**@brief Normalize the quaternion 
    * Such that x^2 + y^2 + z^2 +w^2 = 1 */
-	b3Quaternion& normalize() 
+	b3Quaternion& normalize()
 	{
-#if defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
-		__m128	vd;
-		
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
+		__m128 vd;
+
 		vd = _mm_mul_ps(mVec128, mVec128);
-		
-        __m128 t = _mm_movehl_ps(vd, vd);
+
+		__m128 t = _mm_movehl_ps(vd, vd);
 		vd = _mm_add_ps(vd, t);
 		t = _mm_shuffle_ps(vd, vd, 0x55);
 		vd = _mm_add_ss(vd, t);
 
 		vd = _mm_sqrt_ss(vd);
 		vd = _mm_div_ss(b3vOnes, vd);
-        vd = b3_pshufd_ps(vd, 0); // splat
+		vd = b3_pshufd_ps(vd, 0);  // splat
 		mVec128 = _mm_mul_ps(mVec128, vd);
-    
+
 		return *this;
-#else    
+#else
 		return *this /= length();
 #endif
 	}
 
-  /**@brief Return a scaled version of this quaternion
+	/**@brief Return a scaled version of this quaternion
    * @param s The scale factor */
 	B3_FORCE_INLINE b3Quaternion
 	operator*(const b3Scalar& s) const
 	{
-#if defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
-		__m128	vs = _mm_load_ss(&s);	//	(S 0 0 0)
-		vs = b3_pshufd_ps(vs, 0x00);	//	(S S S S)
-		
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
+		__m128 vs = _mm_load_ss(&s);  //	(S 0 0 0)
+		vs = b3_pshufd_ps(vs, 0x00);  //	(S S S S)
+
 		return b3Quaternion(_mm_mul_ps(mVec128, vs));
 #elif defined(B3_USE_NEON)
 		return b3Quaternion(vmulq_n_f32(mVec128, s));
@@ -389,7 +390,7 @@ public:
 #endif
 	}
 
-  /**@brief Return an inversely scaled versionof this quaternion
+	/**@brief Return an inversely scaled versionof this quaternion
    * @param s The inverse scale factor */
 	b3Quaternion operator/(const b3Scalar& s) const
 	{
@@ -397,29 +398,29 @@ public:
 		return *this * (b3Scalar(1.0) / s);
 	}
 
-  /**@brief Inversely scale this quaternion
+	/**@brief Inversely scale this quaternion
    * @param s The scale factor */
-	b3Quaternion& operator/=(const b3Scalar& s) 
+	b3Quaternion& operator/=(const b3Scalar& s)
 	{
 		b3Assert(s != b3Scalar(0.0));
 		return *this *= b3Scalar(1.0) / s;
 	}
 
-  /**@brief Return a normalized version of this quaternion */
-	b3Quaternion normalized() const 
+	/**@brief Return a normalized version of this quaternion */
+	b3Quaternion normalized() const
 	{
 		return *this / length();
-	} 
-  /**@brief Return the angle between this quaternion and the other 
+	}
+	/**@brief Return the angle between this quaternion and the other 
    * @param q The other quaternion */
-	b3Scalar angle(const b3Quaternion& q) const 
+	b3Scalar angle(const b3Quaternion& q) const
 	{
 		b3Scalar s = b3Sqrt(length2() * q.length2());
 		b3Assert(s != b3Scalar(0.0));
 		return b3Acos(dot(q) / s);
 	}
-  /**@brief Return the angle of rotation represented by this quaternion */
-	b3Scalar getAngle() const 
+	/**@brief Return the angle of rotation represented by this quaternion */
+	b3Scalar getAngle() const
 	{
 		b3Scalar s = b3Scalar(2.) * b3Acos(m_floats[3]);
 		return s;
@@ -428,117 +429,116 @@ public:
 	/**@brief Return the axis of the rotation represented by this quaternion */
 	b3Vector3 getAxis() const
 	{
-		b3Scalar s_squared = 1.f-m_floats[3]*m_floats[3];
-		
-		if (s_squared < b3Scalar(10.) * B3_EPSILON) //Check for divide by zero
-			return b3MakeVector3(1.0, 0.0, 0.0);  // Arbitrary
-		b3Scalar s = 1.f/b3Sqrt(s_squared);
+		b3Scalar s_squared = 1.f - m_floats[3] * m_floats[3];
+
+		if (s_squared < b3Scalar(10.) * B3_EPSILON)  //Check for divide by zero
+			return b3MakeVector3(1.0, 0.0, 0.0);     // Arbitrary
+		b3Scalar s = 1.f / b3Sqrt(s_squared);
 		return b3MakeVector3(m_floats[0] * s, m_floats[1] * s, m_floats[2] * s);
 	}
 
 	/**@brief Return the inverse of this quaternion */
 	b3Quaternion inverse() const
 	{
-#if defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 		return b3Quaternion(_mm_xor_ps(mVec128, b3vQInv));
 #elif defined(B3_USE_NEON)
-        return b3Quaternion((b3SimdFloat4)veorq_s32((int32x4_t)mVec128, (int32x4_t)b3vQInv));
-#else	
+		return b3Quaternion((b3SimdFloat4)veorq_s32((int32x4_t)mVec128, (int32x4_t)b3vQInv));
+#else
 		return b3Quaternion(-m_floats[0], -m_floats[1], -m_floats[2], m_floats[3]);
 #endif
 	}
 
-  /**@brief Return the sum of this quaternion and the other 
+	/**@brief Return the sum of this quaternion and the other 
    * @param q2 The other quaternion */
 	B3_FORCE_INLINE b3Quaternion
 	operator+(const b3Quaternion& q2) const
 	{
-#if defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 		return b3Quaternion(_mm_add_ps(mVec128, q2.mVec128));
 #elif defined(B3_USE_NEON)
-        return b3Quaternion(vaddq_f32(mVec128, q2.mVec128));
-#else	
+		return b3Quaternion(vaddq_f32(mVec128, q2.mVec128));
+#else
 		const b3Quaternion& q1 = *this;
 		return b3Quaternion(q1.getX() + q2.getX(), q1.getY() + q2.getY(), q1.getZ() + q2.getZ(), q1.m_floats[3] + q2.m_floats[3]);
 #endif
 	}
 
-  /**@brief Return the difference between this quaternion and the other 
+	/**@brief Return the difference between this quaternion and the other 
    * @param q2 The other quaternion */
 	B3_FORCE_INLINE b3Quaternion
 	operator-(const b3Quaternion& q2) const
 	{
-#if defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 		return b3Quaternion(_mm_sub_ps(mVec128, q2.mVec128));
 #elif defined(B3_USE_NEON)
-        return b3Quaternion(vsubq_f32(mVec128, q2.mVec128));
-#else	
+		return b3Quaternion(vsubq_f32(mVec128, q2.mVec128));
+#else
 		const b3Quaternion& q1 = *this;
 		return b3Quaternion(q1.getX() - q2.getX(), q1.getY() - q2.getY(), q1.getZ() - q2.getZ(), q1.m_floats[3] - q2.m_floats[3]);
 #endif
 	}
 
-  /**@brief Return the negative of this quaternion 
+	/**@brief Return the negative of this quaternion 
    * This simply negates each element */
 	B3_FORCE_INLINE b3Quaternion operator-() const
 	{
-#if defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 		return b3Quaternion(_mm_xor_ps(mVec128, b3vMzeroMask));
 #elif defined(B3_USE_NEON)
-		return b3Quaternion((b3SimdFloat4)veorq_s32((int32x4_t)mVec128, (int32x4_t)b3vMzeroMask) );
-#else	
+		return b3Quaternion((b3SimdFloat4)veorq_s32((int32x4_t)mVec128, (int32x4_t)b3vMzeroMask));
+#else
 		const b3Quaternion& q2 = *this;
-		return b3Quaternion( - q2.getX(), - q2.getY(),  - q2.getZ(),  - q2.m_floats[3]);
+		return b3Quaternion(-q2.getX(), -q2.getY(), -q2.getZ(), -q2.m_floats[3]);
 #endif
 	}
-  /**@todo document this and it's use */
-	B3_FORCE_INLINE b3Quaternion farthest( const b3Quaternion& qd) const 
+	/**@todo document this and it's use */
+	B3_FORCE_INLINE b3Quaternion farthest(const b3Quaternion& qd) const
 	{
-		b3Quaternion diff,sum;
+		b3Quaternion diff, sum;
 		diff = *this - qd;
 		sum = *this + qd;
-		if( diff.dot(diff) > sum.dot(sum) )
+		if (diff.dot(diff) > sum.dot(sum))
 			return qd;
 		return (-qd);
 	}
 
 	/**@todo document this and it's use */
-	B3_FORCE_INLINE b3Quaternion nearest( const b3Quaternion& qd) const 
+	B3_FORCE_INLINE b3Quaternion nearest(const b3Quaternion& qd) const
 	{
-		b3Quaternion diff,sum;
+		b3Quaternion diff, sum;
 		diff = *this - qd;
 		sum = *this + qd;
-		if( diff.dot(diff) < sum.dot(sum) )
+		if (diff.dot(diff) < sum.dot(sum))
 			return qd;
 		return (-qd);
 	}
 
-
-  /**@brief Return the quaternion which is the result of Spherical Linear Interpolation between this and the other quaternion
+	/**@brief Return the quaternion which is the result of Spherical Linear Interpolation between this and the other quaternion
    * @param q The other quaternion to interpolate with 
    * @param t The ratio between this and q to interpolate.  If t = 0 the result is this, if t=1 the result is q.
    * Slerp interpolates assuming constant velocity.  */
 	b3Quaternion slerp(const b3Quaternion& q, const b3Scalar& t) const
 	{
-	  b3Scalar magnitude = b3Sqrt(length2() * q.length2()); 
-	  b3Assert(magnitude > b3Scalar(0));
+		b3Scalar magnitude = b3Sqrt(length2() * q.length2());
+		b3Assert(magnitude > b3Scalar(0));
 
-    b3Scalar product = dot(q) / magnitude;
-    if (b3Fabs(product) < b3Scalar(1))
+		b3Scalar product = dot(q) / magnitude;
+		if (b3Fabs(product) < b3Scalar(1))
 		{
-      // Take care of long angle case see http://en.wikipedia.org/wiki/Slerp
-      const b3Scalar sign = (product < 0) ? b3Scalar(-1) : b3Scalar(1);
+			// Take care of long angle case see http://en.wikipedia.org/wiki/Slerp
+			const b3Scalar sign = (product < 0) ? b3Scalar(-1) : b3Scalar(1);
 
-      const b3Scalar theta = b3Acos(sign * product);
-      const b3Scalar s1 = b3Sin(sign * t * theta);   
-      const b3Scalar d = b3Scalar(1.0) / b3Sin(theta);
-      const b3Scalar s0 = b3Sin((b3Scalar(1.0) - t) * theta);
+			const b3Scalar theta = b3Acos(sign * product);
+			const b3Scalar s1 = b3Sin(sign * t * theta);
+			const b3Scalar d = b3Scalar(1.0) / b3Sin(theta);
+			const b3Scalar s0 = b3Sin((b3Scalar(1.0) - t) * theta);
 
-      return b3Quaternion(
-          (m_floats[0] * s0 + q.getX() * s1) * d,
-          (m_floats[1] * s0 + q.getY() * s1) * d,
-          (m_floats[2] * s0 + q.getZ() * s1) * d,
-          (m_floats[3] * s0 + q.m_floats[3] * s1) * d);
+			return b3Quaternion(
+				(m_floats[0] * s0 + q.getX() * s1) * d,
+				(m_floats[1] * s0 + q.getY() * s1) * d,
+				(m_floats[2] * s0 + q.getZ() * s1) * d,
+				(m_floats[3] * s0 + q.m_floats[3] * s1) * d);
 		}
 		else
 		{
@@ -546,301 +546,294 @@ public:
 		}
 	}
 
-	static const b3Quaternion&	getIdentity()
+	static const b3Quaternion& getIdentity()
 	{
-		static const b3Quaternion identityQuat(b3Scalar(0.),b3Scalar(0.),b3Scalar(0.),b3Scalar(1.));
+		static const b3Quaternion identityQuat(b3Scalar(0.), b3Scalar(0.), b3Scalar(0.), b3Scalar(1.));
 		return identityQuat;
 	}
 
 	B3_FORCE_INLINE const b3Scalar& getW() const { return m_floats[3]; }
-
-	
 };
 
-
-
-
-
 /**@brief Return the product of two quaternions */
 B3_FORCE_INLINE b3Quaternion
-operator*(const b3Quaternion& q1, const b3Quaternion& q2) 
+operator*(const b3Quaternion& q1, const b3Quaternion& q2)
 {
-#if defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 	__m128 vQ1 = q1.get128();
 	__m128 vQ2 = q2.get128();
 	__m128 A0, A1, B1, A2, B2;
-    
-	A1 = b3_pshufd_ps(vQ1, B3_SHUFFLE(0,1,2,0)); // X Y  z x     //      vtrn
-	B1 = b3_pshufd_ps(vQ2, B3_SHUFFLE(3,3,3,0)); // W W  W X     // vdup vext
+
+	A1 = b3_pshufd_ps(vQ1, B3_SHUFFLE(0, 1, 2, 0));  // X Y  z x     //      vtrn
+	B1 = b3_pshufd_ps(vQ2, B3_SHUFFLE(3, 3, 3, 0));  // W W  W X     // vdup vext
 
 	A1 = A1 * B1;
-	
-	A2 = b3_pshufd_ps(vQ1, B3_SHUFFLE(1,2,0,1)); // Y Z  X Y     // vext 
-	B2 = b3_pshufd_ps(vQ2, B3_SHUFFLE(2,0,1,1)); // z x  Y Y     // vtrn vdup
+
+	A2 = b3_pshufd_ps(vQ1, B3_SHUFFLE(1, 2, 0, 1));  // Y Z  X Y     // vext
+	B2 = b3_pshufd_ps(vQ2, B3_SHUFFLE(2, 0, 1, 1));  // z x  Y Y     // vtrn vdup
 
 	A2 = A2 * B2;
 
-	B1 = b3_pshufd_ps(vQ1, B3_SHUFFLE(2,0,1,2)); // z x Y Z      // vtrn vext
-	B2 = b3_pshufd_ps(vQ2, B3_SHUFFLE(1,2,0,2)); // Y Z x z      // vext vtrn
-	
-	B1 = B1 * B2;	//	A3 *= B3
+	B1 = b3_pshufd_ps(vQ1, B3_SHUFFLE(2, 0, 1, 2));  // z x Y Z      // vtrn vext
+	B2 = b3_pshufd_ps(vQ2, B3_SHUFFLE(1, 2, 0, 2));  // Y Z x z      // vext vtrn
 
-	A0 = b3_splat_ps(vQ1, 3);	//	A0
-	A0 = A0 * vQ2;	//	A0 * B0
+	B1 = B1 * B2;  //	A3 *= B3
+
+	A0 = b3_splat_ps(vQ1, 3);  //	A0
+	A0 = A0 * vQ2;             //	A0 * B0
+
+	A1 = A1 + A2;  //	AB12
+	A0 = A0 - B1;  //	AB03 = AB0 - AB3
+
+	A1 = _mm_xor_ps(A1, b3vPPPM);  //	change sign of the last element
+	A0 = A0 + A1;                  //	AB03 + AB12
 
-	A1 = A1 + A2;	//	AB12
-	A0 =  A0 - B1;	//	AB03 = AB0 - AB3 
-	
-    A1 = _mm_xor_ps(A1, b3vPPPM);	//	change sign of the last element
-	A0 = A0 + A1;	//	AB03 + AB12
-	
 	return b3Quaternion(A0);
 
-#elif defined(B3_USE_NEON)     
+#elif defined(B3_USE_NEON)
 
 	float32x4_t vQ1 = q1.get128();
 	float32x4_t vQ2 = q2.get128();
 	float32x4_t A0, A1, B1, A2, B2, A3, B3;
-    float32x2_t vQ1zx, vQ2wx, vQ1yz, vQ2zx, vQ2yz, vQ2xz;
-    
-    {
-    float32x2x2_t tmp;
-    tmp = vtrn_f32( vget_high_f32(vQ1), vget_low_f32(vQ1) );       // {z x}, {w y}
-    vQ1zx = tmp.val[0];
+	float32x2_t vQ1zx, vQ2wx, vQ1yz, vQ2zx, vQ2yz, vQ2xz;
 
-    tmp = vtrn_f32( vget_high_f32(vQ2), vget_low_f32(vQ2) );       // {z x}, {w y}
-    vQ2zx = tmp.val[0];
-    }
-    vQ2wx = vext_f32(vget_high_f32(vQ2), vget_low_f32(vQ2), 1); 
+	{
+		float32x2x2_t tmp;
+		tmp = vtrn_f32(vget_high_f32(vQ1), vget_low_f32(vQ1));  // {z x}, {w y}
+		vQ1zx = tmp.val[0];
 
-    vQ1yz = vext_f32(vget_low_f32(vQ1), vget_high_f32(vQ1), 1);
+		tmp = vtrn_f32(vget_high_f32(vQ2), vget_low_f32(vQ2));  // {z x}, {w y}
+		vQ2zx = tmp.val[0];
+	}
+	vQ2wx = vext_f32(vget_high_f32(vQ2), vget_low_f32(vQ2), 1);
 
-    vQ2yz = vext_f32(vget_low_f32(vQ2), vget_high_f32(vQ2), 1);
-    vQ2xz = vext_f32(vQ2zx, vQ2zx, 1);
+	vQ1yz = vext_f32(vget_low_f32(vQ1), vget_high_f32(vQ1), 1);
 
-    A1 = vcombine_f32(vget_low_f32(vQ1), vQ1zx);                    // X Y  z x 
-    B1 = vcombine_f32(vdup_lane_f32(vget_high_f32(vQ2), 1), vQ2wx); // W W  W X 
+	vQ2yz = vext_f32(vget_low_f32(vQ2), vget_high_f32(vQ2), 1);
+	vQ2xz = vext_f32(vQ2zx, vQ2zx, 1);
+
+	A1 = vcombine_f32(vget_low_f32(vQ1), vQ1zx);                     // X Y  z x
+	B1 = vcombine_f32(vdup_lane_f32(vget_high_f32(vQ2), 1), vQ2wx);  // W W  W X
 
 	A2 = vcombine_f32(vQ1yz, vget_low_f32(vQ1));
-    B2 = vcombine_f32(vQ2zx, vdup_lane_f32(vget_low_f32(vQ2), 1));
+	B2 = vcombine_f32(vQ2zx, vdup_lane_f32(vget_low_f32(vQ2), 1));
 
-    A3 = vcombine_f32(vQ1zx, vQ1yz);        // Z X Y Z
-    B3 = vcombine_f32(vQ2yz, vQ2xz);        // Y Z x z
+	A3 = vcombine_f32(vQ1zx, vQ1yz);  // Z X Y Z
+	B3 = vcombine_f32(vQ2yz, vQ2xz);  // Y Z x z
 
 	A1 = vmulq_f32(A1, B1);
 	A2 = vmulq_f32(A2, B2);
-	A3 = vmulq_f32(A3, B3);	//	A3 *= B3
-	A0 = vmulq_lane_f32(vQ2, vget_high_f32(vQ1), 1); //	A0 * B0
+	A3 = vmulq_f32(A3, B3);                           //	A3 *= B3
+	A0 = vmulq_lane_f32(vQ2, vget_high_f32(vQ1), 1);  //	A0 * B0
+
+	A1 = vaddq_f32(A1, A2);  //	AB12 = AB1 + AB2
+	A0 = vsubq_f32(A0, A3);  //	AB03 = AB0 - AB3
+
+	//	change the sign of the last element
+	A1 = (b3SimdFloat4)veorq_s32((int32x4_t)A1, (int32x4_t)b3vPPPM);
+	A0 = vaddq_f32(A0, A1);  //	AB03 + AB12
 
-	A1 = vaddq_f32(A1, A2);	//	AB12 = AB1 + AB2
-	A0 = vsubq_f32(A0, A3);	//	AB03 = AB0 - AB3 
-	
-    //	change the sign of the last element
-    A1 = (b3SimdFloat4)veorq_s32((int32x4_t)A1, (int32x4_t)b3vPPPM);	
-	A0 = vaddq_f32(A0, A1);	//	AB03 + AB12
-	
 	return b3Quaternion(A0);
 
 #else
 	return b3Quaternion(
-        q1.getW() * q2.getX() + q1.getX() * q2.getW() + q1.getY() * q2.getZ() - q1.getZ() * q2.getY(),
+		q1.getW() * q2.getX() + q1.getX() * q2.getW() + q1.getY() * q2.getZ() - q1.getZ() * q2.getY(),
 		q1.getW() * q2.getY() + q1.getY() * q2.getW() + q1.getZ() * q2.getX() - q1.getX() * q2.getZ(),
 		q1.getW() * q2.getZ() + q1.getZ() * q2.getW() + q1.getX() * q2.getY() - q1.getY() * q2.getX(),
-		q1.getW() * q2.getW() - q1.getX() * q2.getX() - q1.getY() * q2.getY() - q1.getZ() * q2.getZ()); 
+		q1.getW() * q2.getW() - q1.getX() * q2.getX() - q1.getY() * q2.getY() - q1.getZ() * q2.getZ());
 #endif
 }
 
 B3_FORCE_INLINE b3Quaternion
 operator*(const b3Quaternion& q, const b3Vector3& w)
 {
-#if defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 	__m128 vQ1 = q.get128();
 	__m128 vQ2 = w.get128();
 	__m128 A1, B1, A2, B2, A3, B3;
-	
-	A1 = b3_pshufd_ps(vQ1, B3_SHUFFLE(3,3,3,0));
-	B1 = b3_pshufd_ps(vQ2, B3_SHUFFLE(0,1,2,0));
+
+	A1 = b3_pshufd_ps(vQ1, B3_SHUFFLE(3, 3, 3, 0));
+	B1 = b3_pshufd_ps(vQ2, B3_SHUFFLE(0, 1, 2, 0));
 
 	A1 = A1 * B1;
-	
-	A2 = b3_pshufd_ps(vQ1, B3_SHUFFLE(1,2,0,1));
-	B2 = b3_pshufd_ps(vQ2, B3_SHUFFLE(2,0,1,1));
+
+	A2 = b3_pshufd_ps(vQ1, B3_SHUFFLE(1, 2, 0, 1));
+	B2 = b3_pshufd_ps(vQ2, B3_SHUFFLE(2, 0, 1, 1));
 
 	A2 = A2 * B2;
 
-	A3 = b3_pshufd_ps(vQ1, B3_SHUFFLE(2,0,1,2));
-	B3 = b3_pshufd_ps(vQ2, B3_SHUFFLE(1,2,0,2));
-	
-	A3 = A3 * B3;	//	A3 *= B3
+	A3 = b3_pshufd_ps(vQ1, B3_SHUFFLE(2, 0, 1, 2));
+	B3 = b3_pshufd_ps(vQ2, B3_SHUFFLE(1, 2, 0, 2));
+
+	A3 = A3 * B3;  //	A3 *= B3
+
+	A1 = A1 + A2;                  //	AB12
+	A1 = _mm_xor_ps(A1, b3vPPPM);  //	change sign of the last element
+	A1 = A1 - A3;                  //	AB123 = AB12 - AB3
 
-	A1 = A1 + A2;	//	AB12
-	A1 = _mm_xor_ps(A1, b3vPPPM);	//	change sign of the last element
-    A1 = A1 - A3;	//	AB123 = AB12 - AB3 
-	
 	return b3Quaternion(A1);
-    
-#elif defined(B3_USE_NEON)     
+
+#elif defined(B3_USE_NEON)
 
 	float32x4_t vQ1 = q.get128();
 	float32x4_t vQ2 = w.get128();
 	float32x4_t A1, B1, A2, B2, A3, B3;
-    float32x2_t vQ1wx, vQ2zx, vQ1yz, vQ2yz, vQ1zx, vQ2xz;
-    
-    vQ1wx = vext_f32(vget_high_f32(vQ1), vget_low_f32(vQ1), 1); 
-    {
-    float32x2x2_t tmp;
+	float32x2_t vQ1wx, vQ2zx, vQ1yz, vQ2yz, vQ1zx, vQ2xz;
 
-    tmp = vtrn_f32( vget_high_f32(vQ2), vget_low_f32(vQ2) );       // {z x}, {w y}
-    vQ2zx = tmp.val[0];
+	vQ1wx = vext_f32(vget_high_f32(vQ1), vget_low_f32(vQ1), 1);
+	{
+		float32x2x2_t tmp;
 
-    tmp = vtrn_f32( vget_high_f32(vQ1), vget_low_f32(vQ1) );       // {z x}, {w y}
-    vQ1zx = tmp.val[0];
-    }
+		tmp = vtrn_f32(vget_high_f32(vQ2), vget_low_f32(vQ2));  // {z x}, {w y}
+		vQ2zx = tmp.val[0];
 
-    vQ1yz = vext_f32(vget_low_f32(vQ1), vget_high_f32(vQ1), 1);
+		tmp = vtrn_f32(vget_high_f32(vQ1), vget_low_f32(vQ1));  // {z x}, {w y}
+		vQ1zx = tmp.val[0];
+	}
 
-    vQ2yz = vext_f32(vget_low_f32(vQ2), vget_high_f32(vQ2), 1);
-    vQ2xz = vext_f32(vQ2zx, vQ2zx, 1);
+	vQ1yz = vext_f32(vget_low_f32(vQ1), vget_high_f32(vQ1), 1);
 
-    A1 = vcombine_f32(vdup_lane_f32(vget_high_f32(vQ1), 1), vQ1wx); // W W  W X 
-    B1 = vcombine_f32(vget_low_f32(vQ2), vQ2zx);                    // X Y  z x 
+	vQ2yz = vext_f32(vget_low_f32(vQ2), vget_high_f32(vQ2), 1);
+	vQ2xz = vext_f32(vQ2zx, vQ2zx, 1);
+
+	A1 = vcombine_f32(vdup_lane_f32(vget_high_f32(vQ1), 1), vQ1wx);  // W W  W X
+	B1 = vcombine_f32(vget_low_f32(vQ2), vQ2zx);                     // X Y  z x
 
 	A2 = vcombine_f32(vQ1yz, vget_low_f32(vQ1));
-    B2 = vcombine_f32(vQ2zx, vdup_lane_f32(vget_low_f32(vQ2), 1));
+	B2 = vcombine_f32(vQ2zx, vdup_lane_f32(vget_low_f32(vQ2), 1));
 
-    A3 = vcombine_f32(vQ1zx, vQ1yz);        // Z X Y Z
-    B3 = vcombine_f32(vQ2yz, vQ2xz);        // Y Z x z
+	A3 = vcombine_f32(vQ1zx, vQ1yz);  // Z X Y Z
+	B3 = vcombine_f32(vQ2yz, vQ2xz);  // Y Z x z
 
 	A1 = vmulq_f32(A1, B1);
 	A2 = vmulq_f32(A2, B2);
-	A3 = vmulq_f32(A3, B3);	//	A3 *= B3
+	A3 = vmulq_f32(A3, B3);  //	A3 *= B3
+
+	A1 = vaddq_f32(A1, A2);  //	AB12 = AB1 + AB2
+
+	//	change the sign of the last element
+	A1 = (b3SimdFloat4)veorq_s32((int32x4_t)A1, (int32x4_t)b3vPPPM);
+
+	A1 = vsubq_f32(A1, A3);  //	AB123 = AB12 - AB3
 
-	A1 = vaddq_f32(A1, A2);	//	AB12 = AB1 + AB2
-	
-    //	change the sign of the last element
-    A1 = (b3SimdFloat4)veorq_s32((int32x4_t)A1, (int32x4_t)b3vPPPM);	
-	
-    A1 = vsubq_f32(A1, A3);	//	AB123 = AB12 - AB3
-	
 	return b3Quaternion(A1);
-    
+
 #else
-	return b3Quaternion( 
-         q.getW() * w.getX() + q.getY() * w.getZ() - q.getZ() * w.getY(),
-		 q.getW() * w.getY() + q.getZ() * w.getX() - q.getX() * w.getZ(),
-		 q.getW() * w.getZ() + q.getX() * w.getY() - q.getY() * w.getX(),
-		-q.getX() * w.getX() - q.getY() * w.getY() - q.getZ() * w.getZ()); 
+	return b3Quaternion(
+		q.getW() * w.getX() + q.getY() * w.getZ() - q.getZ() * w.getY(),
+		q.getW() * w.getY() + q.getZ() * w.getX() - q.getX() * w.getZ(),
+		q.getW() * w.getZ() + q.getX() * w.getY() - q.getY() * w.getX(),
+		-q.getX() * w.getX() - q.getY() * w.getY() - q.getZ() * w.getZ());
 #endif
 }
 
 B3_FORCE_INLINE b3Quaternion
 operator*(const b3Vector3& w, const b3Quaternion& q)
 {
-#if defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 	__m128 vQ1 = w.get128();
 	__m128 vQ2 = q.get128();
 	__m128 A1, B1, A2, B2, A3, B3;
-	
-	A1 = b3_pshufd_ps(vQ1, B3_SHUFFLE(0,1,2,0));  // X Y  z x
-	B1 = b3_pshufd_ps(vQ2, B3_SHUFFLE(3,3,3,0));  // W W  W X 
+
+	A1 = b3_pshufd_ps(vQ1, B3_SHUFFLE(0, 1, 2, 0));  // X Y  z x
+	B1 = b3_pshufd_ps(vQ2, B3_SHUFFLE(3, 3, 3, 0));  // W W  W X
 
 	A1 = A1 * B1;
-	
-	A2 = b3_pshufd_ps(vQ1, B3_SHUFFLE(1,2,0,1));
-	B2 = b3_pshufd_ps(vQ2, B3_SHUFFLE(2,0,1,1));
 
-	A2 = A2 *B2;
+	A2 = b3_pshufd_ps(vQ1, B3_SHUFFLE(1, 2, 0, 1));
+	B2 = b3_pshufd_ps(vQ2, B3_SHUFFLE(2, 0, 1, 1));
 
-	A3 = b3_pshufd_ps(vQ1, B3_SHUFFLE(2,0,1,2));
-	B3 = b3_pshufd_ps(vQ2, B3_SHUFFLE(1,2,0,2));
-	
-	A3 = A3 * B3;	//	A3 *= B3
+	A2 = A2 * B2;
+
+	A3 = b3_pshufd_ps(vQ1, B3_SHUFFLE(2, 0, 1, 2));
+	B3 = b3_pshufd_ps(vQ2, B3_SHUFFLE(1, 2, 0, 2));
+
+	A3 = A3 * B3;  //	A3 *= B3
+
+	A1 = A1 + A2;                  //	AB12
+	A1 = _mm_xor_ps(A1, b3vPPPM);  //	change sign of the last element
+	A1 = A1 - A3;                  //	AB123 = AB12 - AB3
 
-	A1 = A1 + A2;	//	AB12
-	A1 = _mm_xor_ps(A1, b3vPPPM);	//	change sign of the last element
-	A1 = A1 - A3;	//	AB123 = AB12 - AB3 
-	
 	return b3Quaternion(A1);
 
-#elif defined(B3_USE_NEON)     
+#elif defined(B3_USE_NEON)
 
 	float32x4_t vQ1 = w.get128();
 	float32x4_t vQ2 = q.get128();
-	float32x4_t  A1, B1, A2, B2, A3, B3;
-    float32x2_t vQ1zx, vQ2wx, vQ1yz, vQ2zx, vQ2yz, vQ2xz;
-    
-    {
-    float32x2x2_t tmp;
-   
-    tmp = vtrn_f32( vget_high_f32(vQ1), vget_low_f32(vQ1) );       // {z x}, {w y}
-    vQ1zx = tmp.val[0];
+	float32x4_t A1, B1, A2, B2, A3, B3;
+	float32x2_t vQ1zx, vQ2wx, vQ1yz, vQ2zx, vQ2yz, vQ2xz;
 
-    tmp = vtrn_f32( vget_high_f32(vQ2), vget_low_f32(vQ2) );       // {z x}, {w y}
-    vQ2zx = tmp.val[0];
-    }
-    vQ2wx = vext_f32(vget_high_f32(vQ2), vget_low_f32(vQ2), 1); 
+	{
+		float32x2x2_t tmp;
 
-    vQ1yz = vext_f32(vget_low_f32(vQ1), vget_high_f32(vQ1), 1);
+		tmp = vtrn_f32(vget_high_f32(vQ1), vget_low_f32(vQ1));  // {z x}, {w y}
+		vQ1zx = tmp.val[0];
 
-    vQ2yz = vext_f32(vget_low_f32(vQ2), vget_high_f32(vQ2), 1);
-    vQ2xz = vext_f32(vQ2zx, vQ2zx, 1);
+		tmp = vtrn_f32(vget_high_f32(vQ2), vget_low_f32(vQ2));  // {z x}, {w y}
+		vQ2zx = tmp.val[0];
+	}
+	vQ2wx = vext_f32(vget_high_f32(vQ2), vget_low_f32(vQ2), 1);
 
-    A1 = vcombine_f32(vget_low_f32(vQ1), vQ1zx);                    // X Y  z x 
-    B1 = vcombine_f32(vdup_lane_f32(vget_high_f32(vQ2), 1), vQ2wx); // W W  W X 
+	vQ1yz = vext_f32(vget_low_f32(vQ1), vget_high_f32(vQ1), 1);
+
+	vQ2yz = vext_f32(vget_low_f32(vQ2), vget_high_f32(vQ2), 1);
+	vQ2xz = vext_f32(vQ2zx, vQ2zx, 1);
+
+	A1 = vcombine_f32(vget_low_f32(vQ1), vQ1zx);                     // X Y  z x
+	B1 = vcombine_f32(vdup_lane_f32(vget_high_f32(vQ2), 1), vQ2wx);  // W W  W X
 
 	A2 = vcombine_f32(vQ1yz, vget_low_f32(vQ1));
-    B2 = vcombine_f32(vQ2zx, vdup_lane_f32(vget_low_f32(vQ2), 1));
+	B2 = vcombine_f32(vQ2zx, vdup_lane_f32(vget_low_f32(vQ2), 1));
 
-    A3 = vcombine_f32(vQ1zx, vQ1yz);        // Z X Y Z
-    B3 = vcombine_f32(vQ2yz, vQ2xz);        // Y Z x z
+	A3 = vcombine_f32(vQ1zx, vQ1yz);  // Z X Y Z
+	B3 = vcombine_f32(vQ2yz, vQ2xz);  // Y Z x z
 
 	A1 = vmulq_f32(A1, B1);
 	A2 = vmulq_f32(A2, B2);
-	A3 = vmulq_f32(A3, B3);	//	A3 *= B3
+	A3 = vmulq_f32(A3, B3);  //	A3 *= B3
+
+	A1 = vaddq_f32(A1, A2);  //	AB12 = AB1 + AB2
+
+	//	change the sign of the last element
+	A1 = (b3SimdFloat4)veorq_s32((int32x4_t)A1, (int32x4_t)b3vPPPM);
+
+	A1 = vsubq_f32(A1, A3);  //	AB123 = AB12 - AB3
 
-	A1 = vaddq_f32(A1, A2);	//	AB12 = AB1 + AB2
-	
-    //	change the sign of the last element
-    A1 = (b3SimdFloat4)veorq_s32((int32x4_t)A1, (int32x4_t)b3vPPPM);	
-	
-    A1 = vsubq_f32(A1, A3);	//	AB123 = AB12 - AB3
-	
 	return b3Quaternion(A1);
-    
+
 #else
-	return b3Quaternion( 
-        +w.getX() * q.getW() + w.getY() * q.getZ() - w.getZ() * q.getY(),
+	return b3Quaternion(
+		+w.getX() * q.getW() + w.getY() * q.getZ() - w.getZ() * q.getY(),
 		+w.getY() * q.getW() + w.getZ() * q.getX() - w.getX() * q.getZ(),
 		+w.getZ() * q.getW() + w.getX() * q.getY() - w.getY() * q.getX(),
-		-w.getX() * q.getX() - w.getY() * q.getY() - w.getZ() * q.getZ()); 
+		-w.getX() * q.getX() - w.getY() * q.getY() - w.getZ() * q.getZ());
 #endif
 }
 
 /**@brief Calculate the dot product between two quaternions */
-B3_FORCE_INLINE b3Scalar 
-b3Dot(const b3Quaternion& q1, const b3Quaternion& q2) 
-{ 
-	return q1.dot(q2); 
+B3_FORCE_INLINE b3Scalar
+b3Dot(const b3Quaternion& q1, const b3Quaternion& q2)
+{
+	return q1.dot(q2);
 }
 
-
 /**@brief Return the length of a quaternion */
 B3_FORCE_INLINE b3Scalar
-b3Length(const b3Quaternion& q) 
-{ 
-	return q.length(); 
+b3Length(const b3Quaternion& q)
+{
+	return q.length();
 }
 
 /**@brief Return the angle between two quaternions*/
 B3_FORCE_INLINE b3Scalar
-b3Angle(const b3Quaternion& q1, const b3Quaternion& q2) 
-{ 
-	return q1.angle(q2); 
+b3Angle(const b3Quaternion& q1, const b3Quaternion& q2)
+{
+	return q1.angle(q2);
 }
 
 /**@brief Return the inverse of a quaternion*/
 B3_FORCE_INLINE b3Quaternion
-b3Inverse(const b3Quaternion& q) 
+b3Inverse(const b3Quaternion& q)
 {
 	return q.inverse();
 }
@@ -851,7 +844,7 @@ b3Inverse(const b3Quaternion& q)
  * @param t The ration between q1 and q2.  t = 0 return q1, t=1 returns q2 
  * Slerp assumes constant velocity between positions. */
 B3_FORCE_INLINE b3Quaternion
-b3Slerp(const b3Quaternion& q1, const b3Quaternion& q2, const b3Scalar& t) 
+b3Slerp(const b3Quaternion& q1, const b3Quaternion& q2, const b3Scalar& t)
 {
 	return q1.slerp(q2, t);
 }
@@ -859,7 +852,7 @@ b3Slerp(const b3Quaternion& q1, const b3Quaternion& q2, const b3Scalar& t)
 B3_FORCE_INLINE b3Quaternion
 b3QuatMul(const b3Quaternion& rot0, const b3Quaternion& rot1)
 {
-	return rot0*rot1;
+	return rot0 * rot1;
 }
 
 B3_FORCE_INLINE b3Quaternion
@@ -868,51 +861,45 @@ b3QuatNormalized(const b3Quaternion& orn)
 	return orn.normalized();
 }
 
-
-
-B3_FORCE_INLINE b3Vector3 
-b3QuatRotate(const b3Quaternion& rotation, const b3Vector3& v) 
+B3_FORCE_INLINE b3Vector3
+b3QuatRotate(const b3Quaternion& rotation, const b3Vector3& v)
 {
 	b3Quaternion q = rotation * v;
 	q *= rotation.inverse();
-#if defined (B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 	return b3MakeVector3(_mm_and_ps(q.get128(), b3vFFF0fMask));
 #elif defined(B3_USE_NEON)
-    return b3MakeVector3((float32x4_t)vandq_s32((int32x4_t)q.get128(), b3vFFF0Mask));
-#else	
-	return b3MakeVector3(q.getX(),q.getY(),q.getZ());
+	return b3MakeVector3((float32x4_t)vandq_s32((int32x4_t)q.get128(), b3vFFF0Mask));
+#else
+	return b3MakeVector3(q.getX(), q.getY(), q.getZ());
 #endif
 }
 
-B3_FORCE_INLINE b3Quaternion 
-b3ShortestArcQuat(const b3Vector3& v0, const b3Vector3& v1) // Game Programming Gems 2.10. make sure v0,v1 are normalized
+B3_FORCE_INLINE b3Quaternion
+b3ShortestArcQuat(const b3Vector3& v0, const b3Vector3& v1)  // Game Programming Gems 2.10. make sure v0,v1 are normalized
 {
 	b3Vector3 c = v0.cross(v1);
-	b3Scalar  d = v0.dot(v1);
+	b3Scalar d = v0.dot(v1);
 
 	if (d < -1.0 + B3_EPSILON)
 	{
-		b3Vector3 n,unused;
-		b3PlaneSpace1(v0,n,unused);
-		return b3Quaternion(n.getX(),n.getY(),n.getZ(),0.0f); // just pick any vector that is orthogonal to v0
+		b3Vector3 n, unused;
+		b3PlaneSpace1(v0, n, unused);
+		return b3Quaternion(n.getX(), n.getY(), n.getZ(), 0.0f);  // just pick any vector that is orthogonal to v0
 	}
 
-	b3Scalar  s = b3Sqrt((1.0f + d) * 2.0f);
+	b3Scalar s = b3Sqrt((1.0f + d) * 2.0f);
 	b3Scalar rs = 1.0f / s;
 
-	return b3Quaternion(c.getX()*rs,c.getY()*rs,c.getZ()*rs,s * 0.5f);
-	
+	return b3Quaternion(c.getX() * rs, c.getY() * rs, c.getZ() * rs, s * 0.5f);
 }
 
-B3_FORCE_INLINE b3Quaternion 
-b3ShortestArcQuatNormalize2(b3Vector3& v0,b3Vector3& v1)
+B3_FORCE_INLINE b3Quaternion
+b3ShortestArcQuatNormalize2(b3Vector3& v0, b3Vector3& v1)
 {
 	v0.normalize();
 	v1.normalize();
-	return b3ShortestArcQuat(v0,v1);
+	return b3ShortestArcQuat(v0, v1);
 }
 
-#endif //B3_SIMD__QUATERNION_H_
-
-
-
+#endif  //B3_SIMD__QUATERNION_H_
diff --git a/thirdparty/bullet/Bullet3Common/b3Random.h b/thirdparty/bullet/Bullet3Common/b3Random.h
index dc040f15626..c2e21496c76 100644
--- a/thirdparty/bullet/Bullet3Common/b3Random.h
+++ b/thirdparty/bullet/Bullet3Common/b3Random.h
@@ -12,8 +12,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #ifndef B3_GEN_RANDOM_H
 #define B3_GEN_RANDOM_H
 
@@ -26,8 +24,8 @@ subject to the following restrictions:
 
 #define B3_RAND_MAX UINT_MAX
 
-B3_FORCE_INLINE void         b3Srand(unsigned int seed) { init_genrand(seed); }
-B3_FORCE_INLINE unsigned int b3rand()                   { return genrand_int32(); }
+B3_FORCE_INLINE void b3Srand(unsigned int seed) { init_genrand(seed); }
+B3_FORCE_INLINE unsigned int b3rand() { return genrand_int32(); }
 
 #else
 
@@ -35,8 +33,8 @@ B3_FORCE_INLINE unsigned int b3rand()                   { return genrand_int32()
 
 #define B3_RAND_MAX RAND_MAX
 
-B3_FORCE_INLINE void         b3Srand(unsigned int seed) { srand(seed); } 
-B3_FORCE_INLINE unsigned int b3rand()                   { return rand(); }
+B3_FORCE_INLINE void b3Srand(unsigned int seed) { srand(seed); }
+B3_FORCE_INLINE unsigned int b3rand() { return rand(); }
 
 #endif
 
@@ -45,6 +43,4 @@ inline b3Scalar b3RandRange(b3Scalar minRange, b3Scalar maxRange)
 	return (b3rand() / (b3Scalar(B3_RAND_MAX) + b3Scalar(1.0))) * (maxRange - minRange) + minRange;
 }
 
-
-#endif //B3_GEN_RANDOM_H
-
+#endif  //B3_GEN_RANDOM_H
diff --git a/thirdparty/bullet/Bullet3Common/b3ResizablePool.h b/thirdparty/bullet/Bullet3Common/b3ResizablePool.h
index 06ad8a778d3..cafe3ff3964 100644
--- a/thirdparty/bullet/Bullet3Common/b3ResizablePool.h
+++ b/thirdparty/bullet/Bullet3Common/b3ResizablePool.h
@@ -4,10 +4,10 @@
 
 #include "Bullet3Common/b3AlignedObjectArray.h"
 
-enum 
+enum
 {
-	B3_POOL_HANDLE_TERMINAL_FREE=-1,
-	B3_POOL_HANDLE_TERMINAL_USED =-2
+	B3_POOL_HANDLE_TERMINAL_FREE = -1,
+	B3_POOL_HANDLE_TERMINAL_USED = -2
 };
 
 template <typename U>
@@ -20,25 +20,23 @@ struct b3PoolBodyHandle : public U
 	{
 		m_nextFreeHandle = next;
 	}
-	int	getNextFree() const
+	int getNextFree() const
 	{
 		return m_nextFreeHandle;
 	}
 };
 
-template <typename T> 
+template <typename T>
 class b3ResizablePool
 {
-
 protected:
-	b3AlignedObjectArray<T>	m_bodyHandles;
-	int m_numUsedHandles;						// number of active handles
-	int	m_firstFreeHandle;		// free handles list
+	b3AlignedObjectArray<T> m_bodyHandles;
+	int m_numUsedHandles;   // number of active handles
+	int m_firstFreeHandle;  // free handles list
 
 	T* getHandleInternal(int handle)
 	{
 		return &m_bodyHandles[handle];
-
 	}
 	const T* getHandleInternal(int handle) const
 	{
@@ -46,17 +44,16 @@ protected:
 	}
 
 public:
-	
 	b3ResizablePool()
 	{
 		initHandles();
 	}
-	
+
 	virtual ~b3ResizablePool()
 	{
 		exitHandles();
 	}
-///handle management
+	///handle management
 
 	int getNumHandles() const
 	{
@@ -65,44 +62,40 @@ public:
 
 	void getUsedHandles(b3AlignedObjectArray<int>& usedHandles) const
 	{
-
-		for (int i=0;i<m_bodyHandles.size();i++)
+		for (int i = 0; i < m_bodyHandles.size(); i++)
 		{
-			if (m_bodyHandles[i].getNextFree()==B3_POOL_HANDLE_TERMINAL_USED)
+			if (m_bodyHandles[i].getNextFree() == B3_POOL_HANDLE_TERMINAL_USED)
 			{
 				usedHandles.push_back(i);
 			}
 		}
 	}
 
-	
-
 	T* getHandle(int handle)
 	{
-		b3Assert(handle>=0);
-		b3Assert(handle<m_bodyHandles.size());
-		if ((handle<0) || (handle>=m_bodyHandles.size()))
+		b3Assert(handle >= 0);
+		b3Assert(handle < m_bodyHandles.size());
+		if ((handle < 0) || (handle >= m_bodyHandles.size()))
 		{
 			return 0;
 		}
 
-		if (m_bodyHandles[handle].getNextFree()==B3_POOL_HANDLE_TERMINAL_USED)
+		if (m_bodyHandles[handle].getNextFree() == B3_POOL_HANDLE_TERMINAL_USED)
 		{
 			return &m_bodyHandles[handle];
 		}
 		return 0;
-
 	}
 	const T* getHandle(int handle) const
 	{
-		b3Assert(handle>=0);
-		b3Assert(handle<m_bodyHandles.size());
-		if ((handle<0) || (handle>=m_bodyHandles.size()))
+		b3Assert(handle >= 0);
+		b3Assert(handle < m_bodyHandles.size());
+		if ((handle < 0) || (handle >= m_bodyHandles.size()))
 		{
 			return 0;
 		}
 
-		if (m_bodyHandles[handle].getNextFree()==B3_POOL_HANDLE_TERMINAL_USED)
+		if (m_bodyHandles[handle].getNextFree() == B3_POOL_HANDLE_TERMINAL_USED)
 		{
 			return &m_bodyHandles[handle];
 		}
@@ -120,7 +113,6 @@ public:
 			for (int i = curCapacity; i < newCapacity; i++)
 				m_bodyHandles[i].setNextFree(i + 1);
 
-
 			m_bodyHandles[newCapacity - 1].setNextFree(-1);
 		}
 		m_firstFreeHandle = curCapacity;
@@ -142,19 +134,18 @@ public:
 
 	int allocHandle()
 	{
-		b3Assert(m_firstFreeHandle>=0);
+		b3Assert(m_firstFreeHandle >= 0);
 
 		int handle = m_firstFreeHandle;
 		m_firstFreeHandle = getHandleInternal(handle)->getNextFree();
 		m_numUsedHandles++;
 
-		if (m_firstFreeHandle<0)
+		if (m_firstFreeHandle < 0)
 		{
 			//int curCapacity = m_bodyHandles.size();
-			int additionalCapacity= m_bodyHandles.size();
+			int additionalCapacity = m_bodyHandles.size();
 			increaseHandleCapacity(additionalCapacity);
 
-
 			getHandleInternal(handle)->setNextFree(m_firstFreeHandle);
 		}
 		getHandleInternal(handle)->setNextFree(B3_POOL_HANDLE_TERMINAL_USED);
@@ -162,12 +153,11 @@ public:
 		return handle;
 	}
 
-
 	void freeHandle(int handle)
 	{
 		b3Assert(handle >= 0);
 
-		if (m_bodyHandles[handle].getNextFree()==B3_POOL_HANDLE_TERMINAL_USED)
+		if (m_bodyHandles[handle].getNextFree() == B3_POOL_HANDLE_TERMINAL_USED)
 		{
 			getHandleInternal(handle)->clear();
 			getHandleInternal(handle)->setNextFree(m_firstFreeHandle);
@@ -176,7 +166,6 @@ public:
 		}
 	}
 };
-	///end handle management
-	
-	#endif //B3_RESIZABLE_POOL_H
-	
\ No newline at end of file
+///end handle management
+
+#endif  //B3_RESIZABLE_POOL_H
diff --git a/thirdparty/bullet/Bullet3Common/b3Scalar.h b/thirdparty/bullet/Bullet3Common/b3Scalar.h
index dbc7fea397d..0db5eb6f4fd 100644
--- a/thirdparty/bullet/Bullet3Common/b3Scalar.h
+++ b/thirdparty/bullet/Bullet3Common/b3Scalar.h
@@ -12,8 +12,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #ifndef B3_SCALAR_H
 #define B3_SCALAR_H
 
@@ -22,238 +20,252 @@ subject to the following restrictions:
 #pragma unmanaged
 #endif
 
-
-
 #include <math.h>
-#include <stdlib.h>//size_t for MSVC 6.0
+#include <stdlib.h>  //size_t for MSVC 6.0
 #include <float.h>
 
 //Original repository is at http://github.com/erwincoumans/bullet3
 #define B3_BULLET_VERSION 300
 
-inline int	b3GetVersion()
+inline int b3GetVersion()
 {
 	return B3_BULLET_VERSION;
 }
 
-#if defined(DEBUG) || defined (_DEBUG)
+#if defined(DEBUG) || defined(_DEBUG)
 #define B3_DEBUG
 #endif
 
-#include "b3Logging.h"//for b3Error
-
+#include "b3Logging.h"  //for b3Error
 
 #ifdef _WIN32
 
-		#if defined(__MINGW32__) || defined(__CYGWIN__) || (defined (_MSC_VER) && _MSC_VER < 1300)
+#if defined(__MINGW32__) || defined(__CYGWIN__) || (defined(_MSC_VER) && _MSC_VER < 1300)
 
-			#define B3_FORCE_INLINE inline
-			#define B3_ATTRIBUTE_ALIGNED16(a) a
-			#define B3_ATTRIBUTE_ALIGNED64(a) a
-			#define B3_ATTRIBUTE_ALIGNED128(a) a
-		#else
-			//#define B3_HAS_ALIGNED_ALLOCATOR
-			#pragma warning(disable : 4324) // disable padding warning
+#define B3_FORCE_INLINE inline
+#define B3_ATTRIBUTE_ALIGNED16(a) a
+#define B3_ATTRIBUTE_ALIGNED64(a) a
+#define B3_ATTRIBUTE_ALIGNED128(a) a
+#else
+//#define B3_HAS_ALIGNED_ALLOCATOR
+#pragma warning(disable : 4324)  // disable padding warning
 //			#pragma warning(disable:4530) // Disable the exception disable but used in MSCV Stl warning.
-			#pragma warning(disable:4996) //Turn off warnings about deprecated C routines
+#pragma warning(disable : 4996)  //Turn off warnings about deprecated C routines
 //			#pragma warning(disable:4786) // Disable the "debug name too long" warning
 
-			#define B3_FORCE_INLINE __forceinline
-			#define B3_ATTRIBUTE_ALIGNED16(a) __declspec(align(16)) a
-			#define B3_ATTRIBUTE_ALIGNED64(a) __declspec(align(64)) a
-			#define B3_ATTRIBUTE_ALIGNED128(a) __declspec (align(128)) a
-		#ifdef _XBOX
-			#define B3_USE_VMX128
+#define B3_FORCE_INLINE __forceinline
+#define B3_ATTRIBUTE_ALIGNED16(a) __declspec(align(16)) a
+#define B3_ATTRIBUTE_ALIGNED64(a) __declspec(align(64)) a
+#define B3_ATTRIBUTE_ALIGNED128(a) __declspec(align(128)) a
+#ifdef _XBOX
+#define B3_USE_VMX128
 
-			#include <ppcintrinsics.h>
- 			#define B3_HAVE_NATIVE_FSEL
- 			#define b3Fsel(a,b,c) __fsel((a),(b),(c))
-		#else
+#include <ppcintrinsics.h>
+#define B3_HAVE_NATIVE_FSEL
+#define b3Fsel(a, b, c) __fsel((a), (b), (c))
+#else
 
-#if (defined (_WIN32) && (_MSC_VER) && _MSC_VER >= 1400) && (!defined (B3_USE_DOUBLE_PRECISION))
-	#if (defined (_M_IX86) || defined (_M_X64))
-			#define B3_USE_SSE
-			#ifdef B3_USE_SSE
-			//B3_USE_SSE_IN_API is disabled under Windows by default, because 
-			//it makes it harder to integrate Bullet into your application under Windows 
-			//(structured embedding Bullet structs/classes need to be 16-byte aligned)
-			//with relatively little performance gain
-			//If you are not embedded Bullet data in your classes, or make sure that you align those classes on 16-byte boundaries
-			//you can manually enable this line or set it in the build system for a bit of performance gain (a few percent, dependent on usage)
-			//#define B3_USE_SSE_IN_API
-			#endif //B3_USE_SSE
-			#include <emmintrin.h>
-	#endif
+#if (defined(_WIN32) && (_MSC_VER) && _MSC_VER >= 1400) && (!defined(B3_USE_DOUBLE_PRECISION))
+#if (defined(_M_IX86) || defined(_M_X64))
+#define B3_USE_SSE
+#ifdef B3_USE_SSE
+//B3_USE_SSE_IN_API is disabled under Windows by default, because
+//it makes it harder to integrate Bullet into your application under Windows
+//(structured embedding Bullet structs/classes need to be 16-byte aligned)
+//with relatively little performance gain
+//If you are not embedded Bullet data in your classes, or make sure that you align those classes on 16-byte boundaries
+//you can manually enable this line or set it in the build system for a bit of performance gain (a few percent, dependent on usage)
+//#define B3_USE_SSE_IN_API
+#endif  //B3_USE_SSE
+#include <emmintrin.h>
+#endif
 #endif
 
-		#endif//_XBOX
+#endif  //_XBOX
 
-		#endif //__MINGW32__
+#endif  //__MINGW32__
 
 #ifdef B3_DEBUG
-	#ifdef _MSC_VER
-		#include <stdio.h>
-		#define b3Assert(x) { if(!(x)){b3Error("Assert "__FILE__ ":%u ("#x")\n", __LINE__);__debugbreak();	}}
-	#else//_MSC_VER
-		#include <assert.h>
-		#define b3Assert assert
-	#endif//_MSC_VER
+#ifdef _MSC_VER
+#include <stdio.h>
+#define b3Assert(x)               \
+	{                             \
+		if (!(x))                 \
+		{                         \
+			b3Error(              \
+				"Assert "__FILE__ \
+				":%u (" #x ")\n", \
+				__LINE__);        \
+			__debugbreak();       \
+		}                         \
+	}
+#else  //_MSC_VER
+#include <assert.h>
+#define b3Assert assert
+#endif  //_MSC_VER
 #else
-		#define b3Assert(x)
+#define b3Assert(x)
 #endif
-		//b3FullAssert is optional, slows down a lot
-		#define b3FullAssert(x)
+//b3FullAssert is optional, slows down a lot
+#define b3FullAssert(x)
 
-		#define b3Likely(_c)  _c
-		#define b3Unlikely(_c) _c
+#define b3Likely(_c) _c
+#define b3Unlikely(_c) _c
 
 #else
-	
-#if defined	(__CELLOS_LV2__)
-		#define B3_FORCE_INLINE inline __attribute__((always_inline))
-		#define B3_ATTRIBUTE_ALIGNED16(a) a __attribute__ ((aligned (16)))
-		#define B3_ATTRIBUTE_ALIGNED64(a) a __attribute__ ((aligned (64)))
-		#define B3_ATTRIBUTE_ALIGNED128(a) a __attribute__ ((aligned (128)))
-		#ifndef assert
-		#include <assert.h>
-		#endif
+
+#if defined(__CELLOS_LV2__)
+#define B3_FORCE_INLINE inline __attribute__((always_inline))
+#define B3_ATTRIBUTE_ALIGNED16(a) a __attribute__((aligned(16)))
+#define B3_ATTRIBUTE_ALIGNED64(a) a __attribute__((aligned(64)))
+#define B3_ATTRIBUTE_ALIGNED128(a) a __attribute__((aligned(128)))
+#ifndef assert
+#include <assert.h>
+#endif
 #ifdef B3_DEBUG
 #ifdef __SPU__
 #include <spu_printf.h>
 #define printf spu_printf
-	#define b3Assert(x) {if(!(x)){b3Error("Assert "__FILE__ ":%u ("#x")\n", __LINE__);spu_hcmpeq(0,0);}}
+#define b3Assert(x)               \
+	{                             \
+		if (!(x))                 \
+		{                         \
+			b3Error(              \
+				"Assert "__FILE__ \
+				":%u (" #x ")\n", \
+				__LINE__);        \
+			spu_hcmpeq(0, 0);     \
+		}                         \
+	}
 #else
-	#define b3Assert assert
+#define b3Assert assert
 #endif
-	
-#else
-		#define b3Assert(x)
-#endif
-		//b3FullAssert is optional, slows down a lot
-		#define b3FullAssert(x)
 
-		#define b3Likely(_c)  _c
-		#define b3Unlikely(_c) _c
+#else
+#define b3Assert(x)
+#endif
+//b3FullAssert is optional, slows down a lot
+#define b3FullAssert(x)
+
+#define b3Likely(_c) _c
+#define b3Unlikely(_c) _c
 
 #else
 
 #ifdef USE_LIBSPE2
 
-		#define B3_FORCE_INLINE __inline
-		#define B3_ATTRIBUTE_ALIGNED16(a) a __attribute__ ((aligned (16)))
-		#define B3_ATTRIBUTE_ALIGNED64(a) a __attribute__ ((aligned (64)))
-		#define B3_ATTRIBUTE_ALIGNED128(a) a __attribute__ ((aligned (128)))
-		#ifndef assert
-		#include <assert.h>
-		#endif
+#define B3_FORCE_INLINE __inline
+#define B3_ATTRIBUTE_ALIGNED16(a) a __attribute__((aligned(16)))
+#define B3_ATTRIBUTE_ALIGNED64(a) a __attribute__((aligned(64)))
+#define B3_ATTRIBUTE_ALIGNED128(a) a __attribute__((aligned(128)))
+#ifndef assert
+#include <assert.h>
+#endif
 #ifdef B3_DEBUG
-		#define b3Assert assert
+#define b3Assert assert
 #else
-		#define b3Assert(x)
+#define b3Assert(x)
 #endif
-		//b3FullAssert is optional, slows down a lot
-		#define b3FullAssert(x)
+//b3FullAssert is optional, slows down a lot
+#define b3FullAssert(x)
 
-
-		#define b3Likely(_c)   __builtin_expect((_c), 1)
-		#define b3Unlikely(_c) __builtin_expect((_c), 0)
-		
+#define b3Likely(_c) __builtin_expect((_c), 1)
+#define b3Unlikely(_c) __builtin_expect((_c), 0)
 
 #else
-	//non-windows systems
+//non-windows systems
 
-#if (defined (__APPLE__) && (!defined (B3_USE_DOUBLE_PRECISION)))
-    #if defined (__i386__) || defined (__x86_64__)
-        #define B3_USE_SSE
-		//B3_USE_SSE_IN_API is enabled on Mac OSX by default, because memory is automatically aligned on 16-byte boundaries
-		//if apps run into issues, we will disable the next line
-		#define B3_USE_SSE_IN_API
-        #ifdef B3_USE_SSE
-            // include appropriate SSE level
-            #if defined (__SSE4_1__)
-                #include <smmintrin.h>
-            #elif defined (__SSSE3__)
-                #include <tmmintrin.h>
-            #elif defined (__SSE3__)
-                #include <pmmintrin.h>
-            #else
-                #include <emmintrin.h>
-            #endif
-        #endif //B3_USE_SSE
-    #elif defined( __armv7__ )
-        #ifdef __clang__
-            #define B3_USE_NEON 1
+#if (defined(__APPLE__) && (!defined(B3_USE_DOUBLE_PRECISION)))
+#if defined(__i386__) || defined(__x86_64__)
+#define B3_USE_SSE
+//B3_USE_SSE_IN_API is enabled on Mac OSX by default, because memory is automatically aligned on 16-byte boundaries
+//if apps run into issues, we will disable the next line
+#define B3_USE_SSE_IN_API
+#ifdef B3_USE_SSE
+// include appropriate SSE level
+#if defined(__SSE4_1__)
+#include <smmintrin.h>
+#elif defined(__SSSE3__)
+#include <tmmintrin.h>
+#elif defined(__SSE3__)
+#include <pmmintrin.h>
+#else
+#include <emmintrin.h>
+#endif
+#endif  //B3_USE_SSE
+#elif defined(__armv7__)
+#ifdef __clang__
+#define B3_USE_NEON 1
 
-            #if defined B3_USE_NEON && defined (__clang__)
-                #include <arm_neon.h>
-            #endif//B3_USE_NEON
-       #endif //__clang__
-    #endif//__arm__
+#if defined B3_USE_NEON && defined(__clang__)
+#include <arm_neon.h>
+#endif  //B3_USE_NEON
+#endif  //__clang__
+#endif  //__arm__
 
-	#define B3_FORCE_INLINE inline __attribute__ ((always_inline))
+#define B3_FORCE_INLINE inline __attribute__((always_inline))
 ///@todo: check out alignment methods for other platforms/compilers
-	#define B3_ATTRIBUTE_ALIGNED16(a) a __attribute__ ((aligned (16)))
-	#define B3_ATTRIBUTE_ALIGNED64(a) a __attribute__ ((aligned (64)))
-	#define B3_ATTRIBUTE_ALIGNED128(a) a __attribute__ ((aligned (128)))
-	#ifndef assert
-	#include <assert.h>
-	#endif
+#define B3_ATTRIBUTE_ALIGNED16(a) a __attribute__((aligned(16)))
+#define B3_ATTRIBUTE_ALIGNED64(a) a __attribute__((aligned(64)))
+#define B3_ATTRIBUTE_ALIGNED128(a) a __attribute__((aligned(128)))
+#ifndef assert
+#include <assert.h>
+#endif
 
-	#if defined(DEBUG) || defined (_DEBUG)
-	 #if defined (__i386__) || defined (__x86_64__)
-	#include <stdio.h>
-	 #define b3Assert(x)\
-	{\
-	if(!(x))\
-	{\
-		b3Error("Assert %s in line %d, file %s\n",#x, __LINE__, __FILE__);\
-		asm volatile ("int3");\
-	}\
+#if defined(DEBUG) || defined(_DEBUG)
+#if defined(__i386__) || defined(__x86_64__)
+#include <stdio.h>
+#define b3Assert(x)                                                             \
+	{                                                                           \
+		if (!(x))                                                               \
+		{                                                                       \
+			b3Error("Assert %s in line %d, file %s\n", #x, __LINE__, __FILE__); \
+			asm volatile("int3");                                               \
+		}                                                                       \
 	}
-	#else//defined (__i386__) || defined (__x86_64__)
-		#define b3Assert assert
-	#endif//defined (__i386__) || defined (__x86_64__)
-	#else//defined(DEBUG) || defined (_DEBUG)
-		#define b3Assert(x)
-	#endif//defined(DEBUG) || defined (_DEBUG)
+#else  //defined (__i386__) || defined (__x86_64__)
+#define b3Assert assert
+#endif  //defined (__i386__) || defined (__x86_64__)
+#else   //defined(DEBUG) || defined (_DEBUG)
+#define b3Assert(x)
+#endif  //defined(DEBUG) || defined (_DEBUG)
 
-	//b3FullAssert is optional, slows down a lot
-	#define b3FullAssert(x)
-	#define b3Likely(_c)  _c
-	#define b3Unlikely(_c) _c
+//b3FullAssert is optional, slows down a lot
+#define b3FullAssert(x)
+#define b3Likely(_c) _c
+#define b3Unlikely(_c) _c
 
 #else
 
-		#define B3_FORCE_INLINE inline
-		///@todo: check out alignment methods for other platforms/compilers
-		#define B3_ATTRIBUTE_ALIGNED16(a) a __attribute__ ((aligned (16)))
-		#define B3_ATTRIBUTE_ALIGNED64(a) a __attribute__ ((aligned (64)))
-		#define B3_ATTRIBUTE_ALIGNED128(a) a __attribute__ ((aligned (128)))
-		///#define B3_ATTRIBUTE_ALIGNED16(a) a
-		///#define B3_ATTRIBUTE_ALIGNED64(a) a
-		///#define B3_ATTRIBUTE_ALIGNED128(a) a
-		#ifndef assert
-		#include <assert.h>
-		#endif
+#define B3_FORCE_INLINE inline
+///@todo: check out alignment methods for other platforms/compilers
+#define B3_ATTRIBUTE_ALIGNED16(a) a __attribute__((aligned(16)))
+#define B3_ATTRIBUTE_ALIGNED64(a) a __attribute__((aligned(64)))
+#define B3_ATTRIBUTE_ALIGNED128(a) a __attribute__((aligned(128)))
+///#define B3_ATTRIBUTE_ALIGNED16(a) a
+///#define B3_ATTRIBUTE_ALIGNED64(a) a
+///#define B3_ATTRIBUTE_ALIGNED128(a) a
+#ifndef assert
+#include <assert.h>
+#endif
 
-#if defined(DEBUG) || defined (_DEBUG)
-		#define b3Assert assert
+#if defined(DEBUG) || defined(_DEBUG)
+#define b3Assert assert
 #else
-		#define b3Assert(x)
+#define b3Assert(x)
 #endif
 
-		//b3FullAssert is optional, slows down a lot
-		#define b3FullAssert(x)
-		#define b3Likely(_c)  _c
-		#define b3Unlikely(_c) _c
-#endif //__APPLE__ 
+//b3FullAssert is optional, slows down a lot
+#define b3FullAssert(x)
+#define b3Likely(_c) _c
+#define b3Unlikely(_c) _c
+#endif  //__APPLE__
 
-#endif // LIBSPE2
+#endif  // LIBSPE2
 
-#endif	//__CELLOS_LV2__
+#endif  //__CELLOS_LV2__
 #endif
 
-
 ///The b3Scalar type abstracts floating point numbers, to easily switch between double and single floating point precision.
 #if defined(B3_USE_DOUBLE_PRECISION)
 typedef double b3Scalar;
@@ -267,34 +279,34 @@ typedef float b3Scalar;
 
 #ifdef B3_USE_SSE
 typedef __m128 b3SimdFloat4;
-#endif//B3_USE_SSE
+#endif  //B3_USE_SSE
 
-#if defined B3_USE_SSE_IN_API && defined (B3_USE_SSE)
+#if defined B3_USE_SSE_IN_API && defined(B3_USE_SSE)
 #ifdef _WIN32
 
 #ifndef B3_NAN
 static int b3NanMask = 0x7F800001;
-#define B3_NAN (*(float*)&b3NanMask)
+#define B3_NAN (*(float *)&b3NanMask)
 #endif
 
 #ifndef B3_INFINITY_MASK
-static  int b3InfinityMask = 0x7F800000;
-#define B3_INFINITY_MASK (*(float*)&b3InfinityMask)
+static int b3InfinityMask = 0x7F800000;
+#define B3_INFINITY_MASK (*(float *)&b3InfinityMask)
 #endif
 
-inline __m128 operator + (const __m128 A, const __m128 B)
+inline __m128 operator+(const __m128 A, const __m128 B)
 {
-    return _mm_add_ps(A, B);
+	return _mm_add_ps(A, B);
 }
 
-inline __m128 operator - (const __m128 A, const __m128 B)
+inline __m128 operator-(const __m128 A, const __m128 B)
 {
-    return _mm_sub_ps(A, B);
+	return _mm_sub_ps(A, B);
 }
 
-inline __m128 operator * (const __m128 A, const __m128 B)
+inline __m128 operator*(const __m128 A, const __m128 B)
 {
-    return _mm_mul_ps(A, B);
+	return _mm_mul_ps(A, B);
 }
 
 #define b3CastfTo128i(a) (_mm_castps_si128(a))
@@ -302,18 +314,19 @@ inline __m128 operator * (const __m128 A, const __m128 B)
 #define b3CastiTo128f(a) (_mm_castsi128_ps(a))
 #define b3CastdTo128f(a) (_mm_castpd_ps(a))
 #define b3CastdTo128i(a) (_mm_castpd_si128(a))
-#define b3Assign128(r0,r1,r2,r3) _mm_setr_ps(r0,r1,r2,r3)
+#define b3Assign128(r0, r1, r2, r3) _mm_setr_ps(r0, r1, r2, r3)
 
-#else//_WIN32
+#else  //_WIN32
 
 #define b3CastfTo128i(a) ((__m128i)(a))
 #define b3CastfTo128d(a) ((__m128d)(a))
-#define b3CastiTo128f(a)  ((__m128) (a))
-#define b3CastdTo128f(a) ((__m128) (a))
+#define b3CastiTo128f(a) ((__m128)(a))
+#define b3CastdTo128f(a) ((__m128)(a))
 #define b3CastdTo128i(a) ((__m128i)(a))
-#define b3Assign128(r0,r1,r2,r3) (__m128){r0,r1,r2,r3}
-#endif//_WIN32
-#endif //B3_USE_SSE_IN_API
+#define b3Assign128(r0, r1, r2, r3) \
+	(__m128) { r0, r1, r2, r3 }
+#endif  //_WIN32
+#endif  //B3_USE_SSE_IN_API
 
 #ifdef B3_USE_NEON
 #include <arm_neon.h>
@@ -321,142 +334,160 @@ inline __m128 operator * (const __m128 A, const __m128 B)
 typedef float32x4_t b3SimdFloat4;
 #define B3_INFINITY INFINITY
 #define B3_NAN NAN
-#define b3Assign128(r0,r1,r2,r3) (float32x4_t){r0,r1,r2,r3}
+#define b3Assign128(r0, r1, r2, r3) \
+	(float32x4_t) { r0, r1, r2, r3 }
 #endif
 
-
-
-
-
-#define B3_DECLARE_ALIGNED_ALLOCATOR() \
-   B3_FORCE_INLINE void* operator new(size_t sizeInBytes)   { return b3AlignedAlloc(sizeInBytes,16); }   \
-   B3_FORCE_INLINE void  operator delete(void* ptr)         { b3AlignedFree(ptr); }   \
-   B3_FORCE_INLINE void* operator new(size_t, void* ptr)   { return ptr; }   \
-   B3_FORCE_INLINE void  operator delete(void*, void*)      { }   \
-   B3_FORCE_INLINE void* operator new[](size_t sizeInBytes)   { return b3AlignedAlloc(sizeInBytes,16); }   \
-   B3_FORCE_INLINE void  operator delete[](void* ptr)         { b3AlignedFree(ptr); }   \
-   B3_FORCE_INLINE void* operator new[](size_t, void* ptr)   { return ptr; }   \
-   B3_FORCE_INLINE void  operator delete[](void*, void*)      { }   \
-
-
+#define B3_DECLARE_ALIGNED_ALLOCATOR()                                                                   \
+	B3_FORCE_INLINE void *operator new(size_t sizeInBytes) { return b3AlignedAlloc(sizeInBytes, 16); }   \
+	B3_FORCE_INLINE void operator delete(void *ptr) { b3AlignedFree(ptr); }                              \
+	B3_FORCE_INLINE void *operator new(size_t, void *ptr) { return ptr; }                                \
+	B3_FORCE_INLINE void operator delete(void *, void *) {}                                              \
+	B3_FORCE_INLINE void *operator new[](size_t sizeInBytes) { return b3AlignedAlloc(sizeInBytes, 16); } \
+	B3_FORCE_INLINE void operator delete[](void *ptr) { b3AlignedFree(ptr); }                            \
+	B3_FORCE_INLINE void *operator new[](size_t, void *ptr) { return ptr; }                              \
+	B3_FORCE_INLINE void operator delete[](void *, void *) {}
 
 #if defined(B3_USE_DOUBLE_PRECISION) || defined(B3_FORCE_DOUBLE_FUNCTIONS)
-		
-B3_FORCE_INLINE b3Scalar b3Sqrt(b3Scalar x) { return sqrt(x); }
+
+B3_FORCE_INLINE b3Scalar b3Sqrt(b3Scalar x)
+{
+	return sqrt(x);
+}
 B3_FORCE_INLINE b3Scalar b3Fabs(b3Scalar x) { return fabs(x); }
 B3_FORCE_INLINE b3Scalar b3Cos(b3Scalar x) { return cos(x); }
 B3_FORCE_INLINE b3Scalar b3Sin(b3Scalar x) { return sin(x); }
 B3_FORCE_INLINE b3Scalar b3Tan(b3Scalar x) { return tan(x); }
-B3_FORCE_INLINE b3Scalar b3Acos(b3Scalar x) { if (x<b3Scalar(-1))	x=b3Scalar(-1); if (x>b3Scalar(1))	x=b3Scalar(1); return acos(x); }
-B3_FORCE_INLINE b3Scalar b3Asin(b3Scalar x) { if (x<b3Scalar(-1))	x=b3Scalar(-1); if (x>b3Scalar(1))	x=b3Scalar(1); return asin(x); }
+B3_FORCE_INLINE b3Scalar b3Acos(b3Scalar x)
+{
+	if (x < b3Scalar(-1)) x = b3Scalar(-1);
+	if (x > b3Scalar(1)) x = b3Scalar(1);
+	return acos(x);
+}
+B3_FORCE_INLINE b3Scalar b3Asin(b3Scalar x)
+{
+	if (x < b3Scalar(-1)) x = b3Scalar(-1);
+	if (x > b3Scalar(1)) x = b3Scalar(1);
+	return asin(x);
+}
 B3_FORCE_INLINE b3Scalar b3Atan(b3Scalar x) { return atan(x); }
 B3_FORCE_INLINE b3Scalar b3Atan2(b3Scalar x, b3Scalar y) { return atan2(x, y); }
 B3_FORCE_INLINE b3Scalar b3Exp(b3Scalar x) { return exp(x); }
 B3_FORCE_INLINE b3Scalar b3Log(b3Scalar x) { return log(x); }
-B3_FORCE_INLINE b3Scalar b3Pow(b3Scalar x,b3Scalar y) { return pow(x,y); }
-B3_FORCE_INLINE b3Scalar b3Fmod(b3Scalar x,b3Scalar y) { return fmod(x,y); }
+B3_FORCE_INLINE b3Scalar b3Pow(b3Scalar x, b3Scalar y) { return pow(x, y); }
+B3_FORCE_INLINE b3Scalar b3Fmod(b3Scalar x, b3Scalar y) { return fmod(x, y); }
 
 #else
-		
-B3_FORCE_INLINE b3Scalar b3Sqrt(b3Scalar y) 
-{ 
+
+B3_FORCE_INLINE b3Scalar b3Sqrt(b3Scalar y)
+{
 #ifdef USE_APPROXIMATION
-    double x, z, tempf;
-    unsigned long *tfptr = ((unsigned long *)&tempf) + 1;
+	double x, z, tempf;
+	unsigned long *tfptr = ((unsigned long *)&tempf) + 1;
 
 	tempf = y;
-	*tfptr = (0xbfcdd90a - *tfptr)>>1; /* estimate of 1/sqrt(y) */
-	x =  tempf;
-	z =  y*b3Scalar(0.5);
-	x = (b3Scalar(1.5)*x)-(x*x)*(x*z);         /* iteration formula     */
-	x = (b3Scalar(1.5)*x)-(x*x)*(x*z);
-	x = (b3Scalar(1.5)*x)-(x*x)*(x*z);
-	x = (b3Scalar(1.5)*x)-(x*x)*(x*z);
-	x = (b3Scalar(1.5)*x)-(x*x)*(x*z);
-	return x*y;
+	*tfptr = (0xbfcdd90a - *tfptr) >> 1; /* estimate of 1/sqrt(y) */
+	x = tempf;
+	z = y * b3Scalar(0.5);
+	x = (b3Scalar(1.5) * x) - (x * x) * (x * z); /* iteration formula     */
+	x = (b3Scalar(1.5) * x) - (x * x) * (x * z);
+	x = (b3Scalar(1.5) * x) - (x * x) * (x * z);
+	x = (b3Scalar(1.5) * x) - (x * x) * (x * z);
+	x = (b3Scalar(1.5) * x) - (x * x) * (x * z);
+	return x * y;
 #else
-	return sqrtf(y); 
+	return sqrtf(y);
 #endif
 }
 B3_FORCE_INLINE b3Scalar b3Fabs(b3Scalar x) { return fabsf(x); }
 B3_FORCE_INLINE b3Scalar b3Cos(b3Scalar x) { return cosf(x); }
 B3_FORCE_INLINE b3Scalar b3Sin(b3Scalar x) { return sinf(x); }
 B3_FORCE_INLINE b3Scalar b3Tan(b3Scalar x) { return tanf(x); }
-B3_FORCE_INLINE b3Scalar b3Acos(b3Scalar x) { 
-	if (x<b3Scalar(-1))	
-		x=b3Scalar(-1); 
-	if (x>b3Scalar(1))	
-		x=b3Scalar(1);
-	return acosf(x); 
+B3_FORCE_INLINE b3Scalar b3Acos(b3Scalar x)
+{
+	if (x < b3Scalar(-1))
+		x = b3Scalar(-1);
+	if (x > b3Scalar(1))
+		x = b3Scalar(1);
+	return acosf(x);
 }
-B3_FORCE_INLINE b3Scalar b3Asin(b3Scalar x) { 
-	if (x<b3Scalar(-1))	
-		x=b3Scalar(-1); 
-	if (x>b3Scalar(1))	
-		x=b3Scalar(1);
-	return asinf(x); 
+B3_FORCE_INLINE b3Scalar b3Asin(b3Scalar x)
+{
+	if (x < b3Scalar(-1))
+		x = b3Scalar(-1);
+	if (x > b3Scalar(1))
+		x = b3Scalar(1);
+	return asinf(x);
 }
 B3_FORCE_INLINE b3Scalar b3Atan(b3Scalar x) { return atanf(x); }
 B3_FORCE_INLINE b3Scalar b3Atan2(b3Scalar x, b3Scalar y) { return atan2f(x, y); }
 B3_FORCE_INLINE b3Scalar b3Exp(b3Scalar x) { return expf(x); }
 B3_FORCE_INLINE b3Scalar b3Log(b3Scalar x) { return logf(x); }
-B3_FORCE_INLINE b3Scalar b3Pow(b3Scalar x,b3Scalar y) { return powf(x,y); }
-B3_FORCE_INLINE b3Scalar b3Fmod(b3Scalar x,b3Scalar y) { return fmodf(x,y); }
-	
+B3_FORCE_INLINE b3Scalar b3Pow(b3Scalar x, b3Scalar y) { return powf(x, y); }
+B3_FORCE_INLINE b3Scalar b3Fmod(b3Scalar x, b3Scalar y) { return fmodf(x, y); }
+
 #endif
 
-#define B3_2_PI         b3Scalar(6.283185307179586232)
-#define B3_PI           (B3_2_PI * b3Scalar(0.5))
-#define B3_HALF_PI      (B3_2_PI * b3Scalar(0.25))
+#define B3_2_PI b3Scalar(6.283185307179586232)
+#define B3_PI (B3_2_PI * b3Scalar(0.5))
+#define B3_HALF_PI (B3_2_PI * b3Scalar(0.25))
 #define B3_RADS_PER_DEG (B3_2_PI / b3Scalar(360.0))
-#define B3_DEGS_PER_RAD  (b3Scalar(360.0) / B3_2_PI)
+#define B3_DEGS_PER_RAD (b3Scalar(360.0) / B3_2_PI)
 #define B3_SQRT12 b3Scalar(0.7071067811865475244008443621048490)
 
-#define b3RecipSqrt(x) ((b3Scalar)(b3Scalar(1.0)/b3Sqrt(b3Scalar(x))))		/* reciprocal square root */
-
+#define b3RecipSqrt(x) ((b3Scalar)(b3Scalar(1.0) / b3Sqrt(b3Scalar(x)))) /* reciprocal square root */
 
 #ifdef B3_USE_DOUBLE_PRECISION
-#define B3_EPSILON      DBL_EPSILON
-#define B3_INFINITY     DBL_MAX
+#define B3_EPSILON DBL_EPSILON
+#define B3_INFINITY DBL_MAX
 #else
-#define B3_EPSILON      FLT_EPSILON
-#define B3_INFINITY     FLT_MAX
+#define B3_EPSILON FLT_EPSILON
+#define B3_INFINITY FLT_MAX
 #endif
 
-B3_FORCE_INLINE b3Scalar b3Atan2Fast(b3Scalar y, b3Scalar x) 
+B3_FORCE_INLINE b3Scalar b3Atan2Fast(b3Scalar y, b3Scalar x)
 {
 	b3Scalar coeff_1 = B3_PI / 4.0f;
 	b3Scalar coeff_2 = 3.0f * coeff_1;
 	b3Scalar abs_y = b3Fabs(y);
 	b3Scalar angle;
-	if (x >= 0.0f) {
+	if (x >= 0.0f)
+	{
 		b3Scalar r = (x - abs_y) / (x + abs_y);
 		angle = coeff_1 - coeff_1 * r;
-	} else {
+	}
+	else
+	{
 		b3Scalar r = (x + abs_y) / (abs_y - x);
 		angle = coeff_2 - coeff_1 * r;
 	}
 	return (y < 0.0f) ? -angle : angle;
 }
 
-B3_FORCE_INLINE bool      b3FuzzyZero(b3Scalar x) { return b3Fabs(x) < B3_EPSILON; }
+B3_FORCE_INLINE bool b3FuzzyZero(b3Scalar x) { return b3Fabs(x) < B3_EPSILON; }
 
-B3_FORCE_INLINE bool	b3Equal(b3Scalar a, b3Scalar eps) {
+B3_FORCE_INLINE bool b3Equal(b3Scalar a, b3Scalar eps)
+{
 	return (((a) <= eps) && !((a) < -eps));
 }
-B3_FORCE_INLINE bool	b3GreaterEqual (b3Scalar a, b3Scalar eps) {
+B3_FORCE_INLINE bool b3GreaterEqual(b3Scalar a, b3Scalar eps)
+{
 	return (!((a) <= eps));
 }
 
-
-B3_FORCE_INLINE int       b3IsNegative(b3Scalar x) {
-    return x < b3Scalar(0.0) ? 1 : 0;
+B3_FORCE_INLINE int b3IsNegative(b3Scalar x)
+{
+	return x < b3Scalar(0.0) ? 1 : 0;
 }
 
 B3_FORCE_INLINE b3Scalar b3Radians(b3Scalar x) { return x * B3_RADS_PER_DEG; }
 B3_FORCE_INLINE b3Scalar b3Degrees(b3Scalar x) { return x * B3_DEGS_PER_RAD; }
 
-#define B3_DECLARE_HANDLE(name) typedef struct name##__ { int unused; } *name
+#define B3_DECLARE_HANDLE(name) \
+	typedef struct name##__     \
+	{                           \
+		int unused;             \
+	} * name
 
 #ifndef b3Fsel
 B3_FORCE_INLINE b3Scalar b3Fsel(b3Scalar a, b3Scalar b, b3Scalar c)
@@ -464,60 +495,57 @@ B3_FORCE_INLINE b3Scalar b3Fsel(b3Scalar a, b3Scalar b, b3Scalar c)
 	return a >= 0 ? b : c;
 }
 #endif
-#define b3Fsels(a,b,c) (b3Scalar)b3Fsel(a,b,c)
-
+#define b3Fsels(a, b, c) (b3Scalar) b3Fsel(a, b, c)
 
 B3_FORCE_INLINE bool b3MachineIsLittleEndian()
 {
-   long int i = 1;
-   const char *p = (const char *) &i;
-   if (p[0] == 1)  // Lowest address contains the least significant byte
-	   return true;
-   else
-	   return false;
+	long int i = 1;
+	const char *p = (const char *)&i;
+	if (p[0] == 1)  // Lowest address contains the least significant byte
+		return true;
+	else
+		return false;
 }
 
-
-
 ///b3Select avoids branches, which makes performance much better for consoles like Playstation 3 and XBox 360
 ///Thanks Phil Knight. See also http://www.cellperformance.com/articles/2006/04/more_techniques_for_eliminatin_1.html
-B3_FORCE_INLINE unsigned b3Select(unsigned condition, unsigned valueIfConditionNonZero, unsigned valueIfConditionZero) 
+B3_FORCE_INLINE unsigned b3Select(unsigned condition, unsigned valueIfConditionNonZero, unsigned valueIfConditionZero)
 {
-    // Set testNz to 0xFFFFFFFF if condition is nonzero, 0x00000000 if condition is zero
-    // Rely on positive value or'ed with its negative having sign bit on
-    // and zero value or'ed with its negative (which is still zero) having sign bit off 
-    // Use arithmetic shift right, shifting the sign bit through all 32 bits
-    unsigned testNz = (unsigned)(((int)condition | -(int)condition) >> 31);
-    unsigned testEqz = ~testNz;
-    return ((valueIfConditionNonZero & testNz) | (valueIfConditionZero & testEqz)); 
+	// Set testNz to 0xFFFFFFFF if condition is nonzero, 0x00000000 if condition is zero
+	// Rely on positive value or'ed with its negative having sign bit on
+	// and zero value or'ed with its negative (which is still zero) having sign bit off
+	// Use arithmetic shift right, shifting the sign bit through all 32 bits
+	unsigned testNz = (unsigned)(((int)condition | -(int)condition) >> 31);
+	unsigned testEqz = ~testNz;
+	return ((valueIfConditionNonZero & testNz) | (valueIfConditionZero & testEqz));
 }
 B3_FORCE_INLINE int b3Select(unsigned condition, int valueIfConditionNonZero, int valueIfConditionZero)
 {
-    unsigned testNz = (unsigned)(((int)condition | -(int)condition) >> 31);
-    unsigned testEqz = ~testNz; 
-    return static_cast<int>((valueIfConditionNonZero & testNz) | (valueIfConditionZero & testEqz));
+	unsigned testNz = (unsigned)(((int)condition | -(int)condition) >> 31);
+	unsigned testEqz = ~testNz;
+	return static_cast<int>((valueIfConditionNonZero & testNz) | (valueIfConditionZero & testEqz));
 }
 B3_FORCE_INLINE float b3Select(unsigned condition, float valueIfConditionNonZero, float valueIfConditionZero)
 {
 #ifdef B3_HAVE_NATIVE_FSEL
-    return (float)b3Fsel((b3Scalar)condition - b3Scalar(1.0f), valueIfConditionNonZero, valueIfConditionZero);
+	return (float)b3Fsel((b3Scalar)condition - b3Scalar(1.0f), valueIfConditionNonZero, valueIfConditionZero);
 #else
-    return (condition != 0) ? valueIfConditionNonZero : valueIfConditionZero; 
+	return (condition != 0) ? valueIfConditionNonZero : valueIfConditionZero;
 #endif
 }
 
-template<typename T> B3_FORCE_INLINE void b3Swap(T& a, T& b)
+template <typename T>
+B3_FORCE_INLINE void b3Swap(T &a, T &b)
 {
 	T tmp = a;
 	a = b;
 	b = tmp;
 }
 
-
 //PCK: endian swapping functions
 B3_FORCE_INLINE unsigned b3SwapEndian(unsigned val)
 {
-	return (((val & 0xff000000) >> 24) | ((val & 0x00ff0000) >> 8) | ((val & 0x0000ff00) << 8)  | ((val & 0x000000ff) << 24));
+	return (((val & 0xff000000) >> 24) | ((val & 0x00ff0000) >> 8) | ((val & 0x0000ff00) << 8) | ((val & 0x000000ff) << 24));
 }
 
 B3_FORCE_INLINE unsigned short b3SwapEndian(unsigned short val)
@@ -532,87 +560,85 @@ B3_FORCE_INLINE unsigned b3SwapEndian(int val)
 
 B3_FORCE_INLINE unsigned short b3SwapEndian(short val)
 {
-	return b3SwapEndian((unsigned short) val);
+	return b3SwapEndian((unsigned short)val);
 }
 
 ///b3SwapFloat uses using char pointers to swap the endianness
 ////b3SwapFloat/b3SwapDouble will NOT return a float, because the machine might 'correct' invalid floating point values
-///Not all values of sign/exponent/mantissa are valid floating point numbers according to IEEE 754. 
-///When a floating point unit is faced with an invalid value, it may actually change the value, or worse, throw an exception. 
-///In most systems, running user mode code, you wouldn't get an exception, but instead the hardware/os/runtime will 'fix' the number for you. 
+///Not all values of sign/exponent/mantissa are valid floating point numbers according to IEEE 754.
+///When a floating point unit is faced with an invalid value, it may actually change the value, or worse, throw an exception.
+///In most systems, running user mode code, you wouldn't get an exception, but instead the hardware/os/runtime will 'fix' the number for you.
 ///so instead of returning a float/double, we return integer/long long integer
-B3_FORCE_INLINE unsigned int  b3SwapEndianFloat(float d)
+B3_FORCE_INLINE unsigned int b3SwapEndianFloat(float d)
 {
-    unsigned int a = 0;
-    unsigned char *dst = (unsigned char *)&a;
-    unsigned char *src = (unsigned char *)&d;
+	unsigned int a = 0;
+	unsigned char *dst = (unsigned char *)&a;
+	unsigned char *src = (unsigned char *)&d;
 
-    dst[0] = src[3];
-    dst[1] = src[2];
-    dst[2] = src[1];
-    dst[3] = src[0];
-    return a;
+	dst[0] = src[3];
+	dst[1] = src[2];
+	dst[2] = src[1];
+	dst[3] = src[0];
+	return a;
 }
 
 // unswap using char pointers
-B3_FORCE_INLINE float b3UnswapEndianFloat(unsigned int a) 
+B3_FORCE_INLINE float b3UnswapEndianFloat(unsigned int a)
 {
-    float d = 0.0f;
-    unsigned char *src = (unsigned char *)&a;
-    unsigned char *dst = (unsigned char *)&d;
+	float d = 0.0f;
+	unsigned char *src = (unsigned char *)&a;
+	unsigned char *dst = (unsigned char *)&d;
 
-    dst[0] = src[3];
-    dst[1] = src[2];
-    dst[2] = src[1];
-    dst[3] = src[0];
+	dst[0] = src[3];
+	dst[1] = src[2];
+	dst[2] = src[1];
+	dst[3] = src[0];
 
-    return d;
+	return d;
 }
 
-
 // swap using char pointers
-B3_FORCE_INLINE void  b3SwapEndianDouble(double d, unsigned char* dst)
+B3_FORCE_INLINE void b3SwapEndianDouble(double d, unsigned char *dst)
 {
-    unsigned char *src = (unsigned char *)&d;
-
-    dst[0] = src[7];
-    dst[1] = src[6];
-    dst[2] = src[5];
-    dst[3] = src[4];
-    dst[4] = src[3];
-    dst[5] = src[2];
-    dst[6] = src[1];
-    dst[7] = src[0];
+	unsigned char *src = (unsigned char *)&d;
 
+	dst[0] = src[7];
+	dst[1] = src[6];
+	dst[2] = src[5];
+	dst[3] = src[4];
+	dst[4] = src[3];
+	dst[5] = src[2];
+	dst[6] = src[1];
+	dst[7] = src[0];
 }
 
 // unswap using char pointers
-B3_FORCE_INLINE double b3UnswapEndianDouble(const unsigned char *src) 
+B3_FORCE_INLINE double b3UnswapEndianDouble(const unsigned char *src)
 {
-    double d = 0.0;
-    unsigned char *dst = (unsigned char *)&d;
+	double d = 0.0;
+	unsigned char *dst = (unsigned char *)&d;
 
-    dst[0] = src[7];
-    dst[1] = src[6];
-    dst[2] = src[5];
-    dst[3] = src[4];
-    dst[4] = src[3];
-    dst[5] = src[2];
-    dst[6] = src[1];
-    dst[7] = src[0];
+	dst[0] = src[7];
+	dst[1] = src[6];
+	dst[2] = src[5];
+	dst[3] = src[4];
+	dst[4] = src[3];
+	dst[5] = src[2];
+	dst[6] = src[1];
+	dst[7] = src[0];
 
 	return d;
 }
 
 // returns normalized value in range [-B3_PI, B3_PI]
-B3_FORCE_INLINE b3Scalar b3NormalizeAngle(b3Scalar angleInRadians) 
+B3_FORCE_INLINE b3Scalar b3NormalizeAngle(b3Scalar angleInRadians)
 {
 	angleInRadians = b3Fmod(angleInRadians, B3_2_PI);
-	if(angleInRadians < -B3_PI)
+	if (angleInRadians < -B3_PI)
 	{
 		return angleInRadians + B3_2_PI;
 	}
-	else if(angleInRadians > B3_PI)
+	else if (angleInRadians > B3_PI)
 	{
 		return angleInRadians - B3_2_PI;
 	}
@@ -626,38 +652,34 @@ B3_FORCE_INLINE b3Scalar b3NormalizeAngle(b3Scalar angleInRadians)
 struct b3TypedObject
 {
 	b3TypedObject(int objectType)
-		:m_objectType(objectType)
+		: m_objectType(objectType)
 	{
 	}
-	int	m_objectType;
+	int m_objectType;
 	inline int getObjectType() const
 	{
 		return m_objectType;
 	}
 };
 
-
-  
 ///align a pointer to the provided alignment, upwards
-template <typename T>T* b3AlignPointer(T* unalignedPtr, size_t alignment)
+template <typename T>
+T *b3AlignPointer(T *unalignedPtr, size_t alignment)
 {
-		
 	struct b3ConvertPointerSizeT
 	{
-		union 
-		{
-				T* ptr;
-				size_t integer;
+		union {
+			T *ptr;
+			size_t integer;
 		};
 	};
-    b3ConvertPointerSizeT converter;
-    
-    
+	b3ConvertPointerSizeT converter;
+
 	const size_t bit_mask = ~(alignment - 1);
-    converter.ptr = unalignedPtr;
-	converter.integer += alignment-1;
+	converter.ptr = unalignedPtr;
+	converter.integer += alignment - 1;
 	converter.integer &= bit_mask;
 	return converter.ptr;
 }
 
-#endif //B3_SCALAR_H
+#endif  //B3_SCALAR_H
diff --git a/thirdparty/bullet/Bullet3Common/b3StackAlloc.h b/thirdparty/bullet/Bullet3Common/b3StackAlloc.h
index de7de056b50..4972236ac7c 100644
--- a/thirdparty/bullet/Bullet3Common/b3StackAlloc.h
+++ b/thirdparty/bullet/Bullet3Common/b3StackAlloc.h
@@ -20,97 +20,99 @@ Nov.2006
 #ifndef B3_STACK_ALLOC
 #define B3_STACK_ALLOC
 
-#include "b3Scalar.h" //for b3Assert
+#include "b3Scalar.h"  //for b3Assert
 #include "b3AlignedAllocator.h"
 
 ///The b3Block class is an internal structure for the b3StackAlloc memory allocator.
 struct b3Block
 {
-	b3Block*			previous;
-	unsigned char*		address;
+	b3Block* previous;
+	unsigned char* address;
 };
 
 ///The StackAlloc class provides some fast stack-based memory allocator (LIFO last-in first-out)
 class b3StackAlloc
 {
 public:
+	b3StackAlloc(unsigned int size)
+	{
+		ctor();
+		create(size);
+	}
+	~b3StackAlloc() { destroy(); }
 
-	b3StackAlloc(unsigned int size)	{ ctor();create(size); }
-	~b3StackAlloc()		{ destroy(); }
-	
-	inline void		create(unsigned int size)
+	inline void create(unsigned int size)
 	{
 		destroy();
-		data		=  (unsigned char*) b3AlignedAlloc(size,16);
-		totalsize	=	size;
+		data = (unsigned char*)b3AlignedAlloc(size, 16);
+		totalsize = size;
 	}
-	inline void		destroy()
+	inline void destroy()
 	{
-		b3Assert(usedsize==0);
+		b3Assert(usedsize == 0);
 		//Raise(L"StackAlloc is still in use");
 
-		if(usedsize==0)
+		if (usedsize == 0)
 		{
-			if(!ischild && data)		
+			if (!ischild && data)
 				b3AlignedFree(data);
 
-			data				=	0;
-			usedsize			=	0;
+			data = 0;
+			usedsize = 0;
 		}
-		
 	}
 
-	int	getAvailableMemory() const
+	int getAvailableMemory() const
 	{
 		return static_cast<int>(totalsize - usedsize);
 	}
 
-	unsigned char*			allocate(unsigned int size)
+	unsigned char* allocate(unsigned int size)
 	{
-		const unsigned int	nus(usedsize+size);
-		if(nus<totalsize)
+		const unsigned int nus(usedsize + size);
+		if (nus < totalsize)
 		{
-			usedsize=nus;
-			return(data+(usedsize-size));
+			usedsize = nus;
+			return (data + (usedsize - size));
 		}
 		b3Assert(0);
 		//&& (L"Not enough memory"));
-		
-		return(0);
+
+		return (0);
 	}
-	B3_FORCE_INLINE b3Block*		beginBlock()
+	B3_FORCE_INLINE b3Block* beginBlock()
 	{
-		b3Block*	pb = (b3Block*)allocate(sizeof(b3Block));
-		pb->previous	=	current;
-		pb->address		=	data+usedsize;
-		current			=	pb;
-		return(pb);
+		b3Block* pb = (b3Block*)allocate(sizeof(b3Block));
+		pb->previous = current;
+		pb->address = data + usedsize;
+		current = pb;
+		return (pb);
 	}
-	B3_FORCE_INLINE void		endBlock(b3Block* block)
+	B3_FORCE_INLINE void endBlock(b3Block* block)
 	{
-		b3Assert(block==current);
+		b3Assert(block == current);
 		//Raise(L"Unmatched blocks");
-		if(block==current)
+		if (block == current)
 		{
-			current		=	block->previous;
-			usedsize	=	(unsigned int)((block->address-data)-sizeof(b3Block));
+			current = block->previous;
+			usedsize = (unsigned int)((block->address - data) - sizeof(b3Block));
 		}
 	}
 
 private:
-	void		ctor()
+	void ctor()
 	{
-		data		=	0;
-		totalsize	=	0;
-		usedsize	=	0;
-		current		=	0;
-		ischild		=	false;
+		data = 0;
+		totalsize = 0;
+		usedsize = 0;
+		current = 0;
+		ischild = false;
 	}
-	unsigned char*		data;
-	unsigned int		totalsize;
-	unsigned int		usedsize;
-	b3Block*	current;
-	bool		ischild;
+	unsigned char* data;
+	unsigned int totalsize;
+	unsigned int usedsize;
+	b3Block* current;
+	bool ischild;
 };
 
-#endif //B3_STACK_ALLOC
+#endif  //B3_STACK_ALLOC
diff --git a/thirdparty/bullet/Bullet3Common/b3Transform.h b/thirdparty/bullet/Bullet3Common/b3Transform.h
index fa480759beb..149da9d1484 100644
--- a/thirdparty/bullet/Bullet3Common/b3Transform.h
+++ b/thirdparty/bullet/Bullet3Common/b3Transform.h
@@ -12,11 +12,9 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef B3_TRANSFORM_H
 #define B3_TRANSFORM_H
 
-
 #include "b3Matrix3x3.h"
 
 #ifdef B3_USE_DOUBLE_PRECISION
@@ -25,46 +23,45 @@ subject to the following restrictions:
 #define b3TransformData b3TransformFloatData
 #endif
 
-
-
-
 /**@brief The b3Transform class supports rigid transforms with only translation and rotation and no scaling/shear.
  *It can be used in combination with b3Vector3, b3Quaternion and b3Matrix3x3 linear algebra classes. */
-B3_ATTRIBUTE_ALIGNED16(class) b3Transform {
-	
-  ///Storage for the rotation
+B3_ATTRIBUTE_ALIGNED16(class)
+b3Transform
+{
+	///Storage for the rotation
 	b3Matrix3x3 m_basis;
-  ///Storage for the translation
-	b3Vector3   m_origin;
+	///Storage for the translation
+	b3Vector3 m_origin;
 
 public:
-	
-  /**@brief No initialization constructor */
+	/**@brief No initialization constructor */
 	b3Transform() {}
-  /**@brief Constructor from b3Quaternion (optional b3Vector3 )
+	/**@brief Constructor from b3Quaternion (optional b3Vector3 )
    * @param q Rotation from quaternion 
    * @param c Translation from Vector (default 0,0,0) */
-	explicit B3_FORCE_INLINE b3Transform(const b3Quaternion& q, 
-		const b3Vector3& c = b3MakeVector3(b3Scalar(0), b3Scalar(0), b3Scalar(0))) 
+	explicit B3_FORCE_INLINE b3Transform(const b3Quaternion& q,
+										 const b3Vector3& c = b3MakeVector3(b3Scalar(0), b3Scalar(0), b3Scalar(0)))
 		: m_basis(q),
-		m_origin(c)
-	{}
-
-  /**@brief Constructor from b3Matrix3x3 (optional b3Vector3)
-   * @param b Rotation from Matrix 
-   * @param c Translation from Vector default (0,0,0)*/
-	explicit B3_FORCE_INLINE b3Transform(const b3Matrix3x3& b, 
-		const b3Vector3& c = b3MakeVector3(b3Scalar(0), b3Scalar(0), b3Scalar(0)))
-		: m_basis(b),
-		m_origin(c)
-	{}
-  /**@brief Copy constructor */
-	B3_FORCE_INLINE b3Transform (const b3Transform& other)
-		: m_basis(other.m_basis),
-		m_origin(other.m_origin)
+		  m_origin(c)
 	{
 	}
-  /**@brief Assignment Operator */
+
+	/**@brief Constructor from b3Matrix3x3 (optional b3Vector3)
+   * @param b Rotation from Matrix 
+   * @param c Translation from Vector default (0,0,0)*/
+	explicit B3_FORCE_INLINE b3Transform(const b3Matrix3x3& b,
+										 const b3Vector3& c = b3MakeVector3(b3Scalar(0), b3Scalar(0), b3Scalar(0)))
+		: m_basis(b),
+		  m_origin(c)
+	{
+	}
+	/**@brief Copy constructor */
+	B3_FORCE_INLINE b3Transform(const b3Transform& other)
+		: m_basis(other.m_basis),
+		  m_origin(other.m_origin)
+	{
+	}
+	/**@brief Assignment Operator */
 	B3_FORCE_INLINE b3Transform& operator=(const b3Transform& other)
 	{
 		m_basis = other.m_basis;
@@ -72,70 +69,70 @@ public:
 		return *this;
 	}
 
-
-  /**@brief Set the current transform as the value of the product of two transforms
+	/**@brief Set the current transform as the value of the product of two transforms
    * @param t1 Transform 1
    * @param t2 Transform 2
    * This = Transform1 * Transform2 */
-		B3_FORCE_INLINE void mult(const b3Transform& t1, const b3Transform& t2) {
-			m_basis = t1.m_basis * t2.m_basis;
-			m_origin = t1(t2.m_origin);
-		}
+	B3_FORCE_INLINE void mult(const b3Transform& t1, const b3Transform& t2)
+	{
+		m_basis = t1.m_basis * t2.m_basis;
+		m_origin = t1(t2.m_origin);
+	}
 
-/*		void multInverseLeft(const b3Transform& t1, const b3Transform& t2) {
+	/*		void multInverseLeft(const b3Transform& t1, const b3Transform& t2) {
 			b3Vector3 v = t2.m_origin - t1.m_origin;
 			m_basis = b3MultTransposeLeft(t1.m_basis, t2.m_basis);
 			m_origin = v * t1.m_basis;
 		}
 		*/
 
-/**@brief Return the transform of the vector */
+	/**@brief Return the transform of the vector */
 	B3_FORCE_INLINE b3Vector3 operator()(const b3Vector3& x) const
 	{
-        return x.dot3(m_basis[0], m_basis[1], m_basis[2]) + m_origin;
+		return x.dot3(m_basis[0], m_basis[1], m_basis[2]) + m_origin;
 	}
 
-  /**@brief Return the transform of the vector */
+	/**@brief Return the transform of the vector */
 	B3_FORCE_INLINE b3Vector3 operator*(const b3Vector3& x) const
 	{
 		return (*this)(x);
 	}
 
-  /**@brief Return the transform of the b3Quaternion */
+	/**@brief Return the transform of the b3Quaternion */
 	B3_FORCE_INLINE b3Quaternion operator*(const b3Quaternion& q) const
 	{
 		return getRotation() * q;
 	}
 
-  /**@brief Return the basis matrix for the rotation */
-	B3_FORCE_INLINE b3Matrix3x3&       getBasis()          { return m_basis; }
-  /**@brief Return the basis matrix for the rotation */
-	B3_FORCE_INLINE const b3Matrix3x3& getBasis()    const { return m_basis; }
+	/**@brief Return the basis matrix for the rotation */
+	B3_FORCE_INLINE b3Matrix3x3& getBasis() { return m_basis; }
+	/**@brief Return the basis matrix for the rotation */
+	B3_FORCE_INLINE const b3Matrix3x3& getBasis() const { return m_basis; }
 
-  /**@brief Return the origin vector translation */
-	B3_FORCE_INLINE b3Vector3&         getOrigin()         { return m_origin; }
-  /**@brief Return the origin vector translation */
-	B3_FORCE_INLINE const b3Vector3&   getOrigin()   const { return m_origin; }
+	/**@brief Return the origin vector translation */
+	B3_FORCE_INLINE b3Vector3& getOrigin() { return m_origin; }
+	/**@brief Return the origin vector translation */
+	B3_FORCE_INLINE const b3Vector3& getOrigin() const { return m_origin; }
 
-  /**@brief Return a quaternion representing the rotation */
-	b3Quaternion getRotation() const { 
+	/**@brief Return a quaternion representing the rotation */
+	b3Quaternion getRotation() const
+	{
 		b3Quaternion q;
 		m_basis.getRotation(q);
 		return q;
 	}
-	
-	
-  /**@brief Set from an array 
+
+	/**@brief Set from an array 
    * @param m A pointer to a 15 element array (12 rotation(row major padded on the right by 1), and 3 translation */
-	void setFromOpenGLMatrix(const b3Scalar *m)
+	void setFromOpenGLMatrix(const b3Scalar* m)
 	{
 		m_basis.setFromOpenGLSubMatrix(m);
-		m_origin.setValue(m[12],m[13],m[14]);
+		m_origin.setValue(m[12], m[13], m[14]);
 	}
 
-  /**@brief Fill an array representation
+	/**@brief Fill an array representation
    * @param m A pointer to a 15 element array (12 rotation(row major padded on the right by 1), and 3 translation */
-	void getOpenGLMatrix(b3Scalar *m) const 
+	void getOpenGLMatrix(b3Scalar * m) const
 	{
 		m_basis.getOpenGLSubMatrix(m);
 		m[12] = m_origin.getX();
@@ -144,80 +141,76 @@ public:
 		m[15] = b3Scalar(1.0);
 	}
 
-  /**@brief Set the translational element
+	/**@brief Set the translational element
    * @param origin The vector to set the translation to */
-	B3_FORCE_INLINE void setOrigin(const b3Vector3& origin) 
-	{ 
+	B3_FORCE_INLINE void setOrigin(const b3Vector3& origin)
+	{
 		m_origin = origin;
 	}
 
 	B3_FORCE_INLINE b3Vector3 invXform(const b3Vector3& inVec) const;
 
-
-  /**@brief Set the rotational element by b3Matrix3x3 */
+	/**@brief Set the rotational element by b3Matrix3x3 */
 	B3_FORCE_INLINE void setBasis(const b3Matrix3x3& basis)
-	{ 
+	{
 		m_basis = basis;
 	}
 
-  /**@brief Set the rotational element by b3Quaternion */
+	/**@brief Set the rotational element by b3Quaternion */
 	B3_FORCE_INLINE void setRotation(const b3Quaternion& q)
 	{
 		m_basis.setRotation(q);
 	}
 
-
-  /**@brief Set this transformation to the identity */
+	/**@brief Set this transformation to the identity */
 	void setIdentity()
 	{
 		m_basis.setIdentity();
 		m_origin.setValue(b3Scalar(0.0), b3Scalar(0.0), b3Scalar(0.0));
 	}
 
-  /**@brief Multiply this Transform by another(this = this * another) 
+	/**@brief Multiply this Transform by another(this = this * another) 
    * @param t The other transform */
-	b3Transform& operator*=(const b3Transform& t) 
+	b3Transform& operator*=(const b3Transform& t)
 	{
 		m_origin += m_basis * t.m_origin;
 		m_basis *= t.m_basis;
 		return *this;
 	}
 
-  /**@brief Return the inverse of this transform */
+	/**@brief Return the inverse of this transform */
 	b3Transform inverse() const
-	{ 
+	{
 		b3Matrix3x3 inv = m_basis.transpose();
 		return b3Transform(inv, inv * -m_origin);
 	}
 
-  /**@brief Return the inverse of this transform times the other transform
+	/**@brief Return the inverse of this transform times the other transform
    * @param t The other transform 
    * return this.inverse() * the other */
-	b3Transform inverseTimes(const b3Transform& t) const;  
+	b3Transform inverseTimes(const b3Transform& t) const;
 
-  /**@brief Return the product of this transform and the other */
+	/**@brief Return the product of this transform and the other */
 	b3Transform operator*(const b3Transform& t) const;
 
-  /**@brief Return an identity transform */
-	static const b3Transform&	getIdentity()
+	/**@brief Return an identity transform */
+	static const b3Transform& getIdentity()
 	{
 		static const b3Transform identityTransform(b3Matrix3x3::getIdentity());
 		return identityTransform;
 	}
 
-	void	serialize(struct	b3TransformData& dataOut) const;
+	void serialize(struct b3TransformData & dataOut) const;
 
-	void	serializeFloat(struct	b3TransformFloatData& dataOut) const;
+	void serializeFloat(struct b3TransformFloatData & dataOut) const;
 
-	void	deSerialize(const struct	b3TransformData& dataIn);
+	void deSerialize(const struct b3TransformData& dataIn);
 
-	void	deSerializeDouble(const struct	b3TransformDoubleData& dataIn);
-
-	void	deSerializeFloat(const struct	b3TransformFloatData& dataIn);
+	void deSerializeDouble(const struct b3TransformDoubleData& dataIn);
 
+	void deSerializeFloat(const struct b3TransformFloatData& dataIn);
 };
 
-
 B3_FORCE_INLINE b3Vector3
 b3Transform::invXform(const b3Vector3& inVec) const
 {
@@ -225,80 +218,69 @@ b3Transform::invXform(const b3Vector3& inVec) const
 	return (m_basis.transpose() * v);
 }
 
-B3_FORCE_INLINE b3Transform 
-b3Transform::inverseTimes(const b3Transform& t) const  
+B3_FORCE_INLINE b3Transform
+b3Transform::inverseTimes(const b3Transform& t) const
 {
 	b3Vector3 v = t.getOrigin() - m_origin;
-		return b3Transform(m_basis.transposeTimes(t.m_basis),
-			v * m_basis);
+	return b3Transform(m_basis.transposeTimes(t.m_basis),
+					   v * m_basis);
 }
 
-B3_FORCE_INLINE b3Transform 
-b3Transform::operator*(const b3Transform& t) const
+B3_FORCE_INLINE b3Transform
+	b3Transform::operator*(const b3Transform& t) const
 {
-	return b3Transform(m_basis * t.m_basis, 
-		(*this)(t.m_origin));
+	return b3Transform(m_basis * t.m_basis,
+					   (*this)(t.m_origin));
 }
 
 /**@brief Test if two transforms have all elements equal */
 B3_FORCE_INLINE bool operator==(const b3Transform& t1, const b3Transform& t2)
 {
-   return ( t1.getBasis()  == t2.getBasis() &&
-            t1.getOrigin() == t2.getOrigin() );
+	return (t1.getBasis() == t2.getBasis() &&
+			t1.getOrigin() == t2.getOrigin());
 }
 
-
 ///for serialization
-struct	b3TransformFloatData
+struct b3TransformFloatData
 {
-	b3Matrix3x3FloatData	m_basis;
-	b3Vector3FloatData	m_origin;
+	b3Matrix3x3FloatData m_basis;
+	b3Vector3FloatData m_origin;
 };
 
-struct	b3TransformDoubleData
+struct b3TransformDoubleData
 {
-	b3Matrix3x3DoubleData	m_basis;
-	b3Vector3DoubleData	m_origin;
+	b3Matrix3x3DoubleData m_basis;
+	b3Vector3DoubleData m_origin;
 };
 
-
-
-B3_FORCE_INLINE	void	b3Transform::serialize(b3TransformData& dataOut) const
+B3_FORCE_INLINE void b3Transform::serialize(b3TransformData& dataOut) const
 {
 	m_basis.serialize(dataOut.m_basis);
 	m_origin.serialize(dataOut.m_origin);
 }
 
-B3_FORCE_INLINE	void	b3Transform::serializeFloat(b3TransformFloatData& dataOut) const
+B3_FORCE_INLINE void b3Transform::serializeFloat(b3TransformFloatData& dataOut) const
 {
 	m_basis.serializeFloat(dataOut.m_basis);
 	m_origin.serializeFloat(dataOut.m_origin);
 }
 
-
-B3_FORCE_INLINE	void	b3Transform::deSerialize(const b3TransformData& dataIn)
+B3_FORCE_INLINE void b3Transform::deSerialize(const b3TransformData& dataIn)
 {
 	m_basis.deSerialize(dataIn.m_basis);
 	m_origin.deSerialize(dataIn.m_origin);
 }
 
-B3_FORCE_INLINE	void	b3Transform::deSerializeFloat(const b3TransformFloatData& dataIn)
+B3_FORCE_INLINE void b3Transform::deSerializeFloat(const b3TransformFloatData& dataIn)
 {
 	m_basis.deSerializeFloat(dataIn.m_basis);
 	m_origin.deSerializeFloat(dataIn.m_origin);
 }
 
-B3_FORCE_INLINE	void	b3Transform::deSerializeDouble(const b3TransformDoubleData& dataIn)
+B3_FORCE_INLINE void b3Transform::deSerializeDouble(const b3TransformDoubleData& dataIn)
 {
 	m_basis.deSerializeDouble(dataIn.m_basis);
 	m_origin.deSerializeDouble(dataIn.m_origin);
 }
 
-
-#endif //B3_TRANSFORM_H
-
-
-
-
-
-
+#endif  //B3_TRANSFORM_H
diff --git a/thirdparty/bullet/Bullet3Common/b3TransformUtil.h b/thirdparty/bullet/Bullet3Common/b3TransformUtil.h
index 6ce580c1328..1850a9be5f2 100644
--- a/thirdparty/bullet/Bullet3Common/b3TransformUtil.h
+++ b/thirdparty/bullet/Bullet3Common/b3TransformUtil.h
@@ -12,204 +12,189 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef B3_TRANSFORM_UTIL_H
 #define B3_TRANSFORM_UTIL_H
 
 #include "b3Transform.h"
-#define B3_ANGULAR_MOTION_THRESHOLD b3Scalar(0.5)*B3_HALF_PI
+#define B3_ANGULAR_MOTION_THRESHOLD b3Scalar(0.5) * B3_HALF_PI
 
-
-
-
-B3_FORCE_INLINE b3Vector3 b3AabbSupport(const b3Vector3& halfExtents,const b3Vector3& supportDir)
+B3_FORCE_INLINE b3Vector3 b3AabbSupport(const b3Vector3& halfExtents, const b3Vector3& supportDir)
 {
 	return b3MakeVector3(supportDir.getX() < b3Scalar(0.0) ? -halfExtents.getX() : halfExtents.getX(),
-      supportDir.getY() < b3Scalar(0.0) ? -halfExtents.getY() : halfExtents.getY(),
-      supportDir.getZ() < b3Scalar(0.0) ? -halfExtents.getZ() : halfExtents.getZ()); 
+						 supportDir.getY() < b3Scalar(0.0) ? -halfExtents.getY() : halfExtents.getY(),
+						 supportDir.getZ() < b3Scalar(0.0) ? -halfExtents.getZ() : halfExtents.getZ());
 }
 
-
-
-
-
-
 /// Utils related to temporal transforms
 class b3TransformUtil
 {
-
 public:
-
-	static void integrateTransform(const b3Transform& curTrans,const b3Vector3& linvel,const b3Vector3& angvel,b3Scalar timeStep,b3Transform& predictedTransform)
+	static void integrateTransform(const b3Transform& curTrans, const b3Vector3& linvel, const b3Vector3& angvel, b3Scalar timeStep, b3Transform& predictedTransform)
 	{
 		predictedTransform.setOrigin(curTrans.getOrigin() + linvel * timeStep);
-//	#define QUATERNION_DERIVATIVE
-	#ifdef QUATERNION_DERIVATIVE
+		//	#define QUATERNION_DERIVATIVE
+#ifdef QUATERNION_DERIVATIVE
 		b3Quaternion predictedOrn = curTrans.getRotation();
 		predictedOrn += (angvel * predictedOrn) * (timeStep * b3Scalar(0.5));
 		predictedOrn.normalize();
-	#else
+#else
 		//Exponential map
 		//google for "Practical Parameterization of Rotations Using the Exponential Map", F. Sebastian Grassia
 
 		b3Vector3 axis;
-		b3Scalar	fAngle = angvel.length(); 
+		b3Scalar fAngle = angvel.length();
 		//limit the angular motion
-		if (fAngle*timeStep > B3_ANGULAR_MOTION_THRESHOLD)
+		if (fAngle * timeStep > B3_ANGULAR_MOTION_THRESHOLD)
 		{
 			fAngle = B3_ANGULAR_MOTION_THRESHOLD / timeStep;
 		}
 
-		if ( fAngle < b3Scalar(0.001) )
+		if (fAngle < b3Scalar(0.001))
 		{
 			// use Taylor's expansions of sync function
-			axis   = angvel*( b3Scalar(0.5)*timeStep-(timeStep*timeStep*timeStep)*(b3Scalar(0.020833333333))*fAngle*fAngle );
+			axis = angvel * (b3Scalar(0.5) * timeStep - (timeStep * timeStep * timeStep) * (b3Scalar(0.020833333333)) * fAngle * fAngle);
 		}
 		else
 		{
 			// sync(fAngle) = sin(c*fAngle)/t
-			axis   = angvel*( b3Sin(b3Scalar(0.5)*fAngle*timeStep)/fAngle );
+			axis = angvel * (b3Sin(b3Scalar(0.5) * fAngle * timeStep) / fAngle);
 		}
-		b3Quaternion dorn (axis.getX(),axis.getY(),axis.getZ(),b3Cos( fAngle*timeStep*b3Scalar(0.5) ));
+		b3Quaternion dorn(axis.getX(), axis.getY(), axis.getZ(), b3Cos(fAngle * timeStep * b3Scalar(0.5)));
 		b3Quaternion orn0 = curTrans.getRotation();
 
 		b3Quaternion predictedOrn = dorn * orn0;
 		predictedOrn.normalize();
-	#endif
+#endif
 		predictedTransform.setRotation(predictedOrn);
 	}
 
-	static void	calculateVelocityQuaternion(const b3Vector3& pos0,const b3Vector3& pos1,const b3Quaternion& orn0,const b3Quaternion& orn1,b3Scalar timeStep,b3Vector3& linVel,b3Vector3& angVel)
+	static void calculateVelocityQuaternion(const b3Vector3& pos0, const b3Vector3& pos1, const b3Quaternion& orn0, const b3Quaternion& orn1, b3Scalar timeStep, b3Vector3& linVel, b3Vector3& angVel)
 	{
 		linVel = (pos1 - pos0) / timeStep;
 		b3Vector3 axis;
-		b3Scalar  angle;
+		b3Scalar angle;
 		if (orn0 != orn1)
 		{
-			calculateDiffAxisAngleQuaternion(orn0,orn1,axis,angle);
+			calculateDiffAxisAngleQuaternion(orn0, orn1, axis, angle);
 			angVel = axis * angle / timeStep;
-		} else
+		}
+		else
 		{
-			angVel.setValue(0,0,0);
+			angVel.setValue(0, 0, 0);
 		}
 	}
 
-	static void calculateDiffAxisAngleQuaternion(const b3Quaternion& orn0,const b3Quaternion& orn1a,b3Vector3& axis,b3Scalar& angle)
+	static void calculateDiffAxisAngleQuaternion(const b3Quaternion& orn0, const b3Quaternion& orn1a, b3Vector3& axis, b3Scalar& angle)
 	{
 		b3Quaternion orn1 = orn0.nearest(orn1a);
 		b3Quaternion dorn = orn1 * orn0.inverse();
 		angle = dorn.getAngle();
-		axis = b3MakeVector3(dorn.getX(),dorn.getY(),dorn.getZ());
+		axis = b3MakeVector3(dorn.getX(), dorn.getY(), dorn.getZ());
 		axis[3] = b3Scalar(0.);
 		//check for axis length
 		b3Scalar len = axis.length2();
-		if (len < B3_EPSILON*B3_EPSILON)
-			axis = b3MakeVector3(b3Scalar(1.),b3Scalar(0.),b3Scalar(0.));
+		if (len < B3_EPSILON * B3_EPSILON)
+			axis = b3MakeVector3(b3Scalar(1.), b3Scalar(0.), b3Scalar(0.));
 		else
 			axis /= b3Sqrt(len);
 	}
 
-	static void	calculateVelocity(const b3Transform& transform0,const b3Transform& transform1,b3Scalar timeStep,b3Vector3& linVel,b3Vector3& angVel)
+	static void calculateVelocity(const b3Transform& transform0, const b3Transform& transform1, b3Scalar timeStep, b3Vector3& linVel, b3Vector3& angVel)
 	{
 		linVel = (transform1.getOrigin() - transform0.getOrigin()) / timeStep;
 		b3Vector3 axis;
-		b3Scalar  angle;
-		calculateDiffAxisAngle(transform0,transform1,axis,angle);
+		b3Scalar angle;
+		calculateDiffAxisAngle(transform0, transform1, axis, angle);
 		angVel = axis * angle / timeStep;
 	}
 
-	static void calculateDiffAxisAngle(const b3Transform& transform0,const b3Transform& transform1,b3Vector3& axis,b3Scalar& angle)
+	static void calculateDiffAxisAngle(const b3Transform& transform0, const b3Transform& transform1, b3Vector3& axis, b3Scalar& angle)
 	{
 		b3Matrix3x3 dmat = transform1.getBasis() * transform0.getBasis().inverse();
 		b3Quaternion dorn;
 		dmat.getRotation(dorn);
 
-		///floating point inaccuracy can lead to w component > 1..., which breaks 
+		///floating point inaccuracy can lead to w component > 1..., which breaks
 		dorn.normalize();
-		
+
 		angle = dorn.getAngle();
-		axis = b3MakeVector3(dorn.getX(),dorn.getY(),dorn.getZ());
+		axis = b3MakeVector3(dorn.getX(), dorn.getY(), dorn.getZ());
 		axis[3] = b3Scalar(0.);
 		//check for axis length
 		b3Scalar len = axis.length2();
-		if (len < B3_EPSILON*B3_EPSILON)
-			axis = b3MakeVector3(b3Scalar(1.),b3Scalar(0.),b3Scalar(0.));
+		if (len < B3_EPSILON * B3_EPSILON)
+			axis = b3MakeVector3(b3Scalar(1.), b3Scalar(0.), b3Scalar(0.));
 		else
 			axis /= b3Sqrt(len);
 	}
-
 };
 
-
-///The b3ConvexSeparatingDistanceUtil can help speed up convex collision detection 
+///The b3ConvexSeparatingDistanceUtil can help speed up convex collision detection
 ///by conservatively updating a cached separating distance/vector instead of re-calculating the closest distance
-class	b3ConvexSeparatingDistanceUtil
+class b3ConvexSeparatingDistanceUtil
 {
-	b3Quaternion	m_ornA;
-	b3Quaternion	m_ornB;
-	b3Vector3	m_posA;
-	b3Vector3	m_posB;
-	
-	b3Vector3	m_separatingNormal;
+	b3Quaternion m_ornA;
+	b3Quaternion m_ornB;
+	b3Vector3 m_posA;
+	b3Vector3 m_posB;
 
-	b3Scalar	m_boundingRadiusA;
-	b3Scalar	m_boundingRadiusB;
-	b3Scalar	m_separatingDistance;
+	b3Vector3 m_separatingNormal;
+
+	b3Scalar m_boundingRadiusA;
+	b3Scalar m_boundingRadiusB;
+	b3Scalar m_separatingDistance;
 
 public:
-
-	b3ConvexSeparatingDistanceUtil(b3Scalar	boundingRadiusA,b3Scalar	boundingRadiusB)
-		:m_boundingRadiusA(boundingRadiusA),
-		m_boundingRadiusB(boundingRadiusB),
-		m_separatingDistance(0.f)
+	b3ConvexSeparatingDistanceUtil(b3Scalar boundingRadiusA, b3Scalar boundingRadiusB)
+		: m_boundingRadiusA(boundingRadiusA),
+		  m_boundingRadiusB(boundingRadiusB),
+		  m_separatingDistance(0.f)
 	{
 	}
 
-	b3Scalar	getConservativeSeparatingDistance()
+	b3Scalar getConservativeSeparatingDistance()
 	{
 		return m_separatingDistance;
 	}
 
-	void	updateSeparatingDistance(const b3Transform& transA,const b3Transform& transB)
+	void updateSeparatingDistance(const b3Transform& transA, const b3Transform& transB)
 	{
 		const b3Vector3& toPosA = transA.getOrigin();
 		const b3Vector3& toPosB = transB.getOrigin();
 		b3Quaternion toOrnA = transA.getRotation();
 		b3Quaternion toOrnB = transB.getRotation();
 
-		if (m_separatingDistance>0.f)
+		if (m_separatingDistance > 0.f)
 		{
-			
-
-			b3Vector3 linVelA,angVelA,linVelB,angVelB;
-			b3TransformUtil::calculateVelocityQuaternion(m_posA,toPosA,m_ornA,toOrnA,b3Scalar(1.),linVelA,angVelA);
-			b3TransformUtil::calculateVelocityQuaternion(m_posB,toPosB,m_ornB,toOrnB,b3Scalar(1.),linVelB,angVelB);
+			b3Vector3 linVelA, angVelA, linVelB, angVelB;
+			b3TransformUtil::calculateVelocityQuaternion(m_posA, toPosA, m_ornA, toOrnA, b3Scalar(1.), linVelA, angVelA);
+			b3TransformUtil::calculateVelocityQuaternion(m_posB, toPosB, m_ornB, toOrnB, b3Scalar(1.), linVelB, angVelB);
 			b3Scalar maxAngularProjectedVelocity = angVelA.length() * m_boundingRadiusA + angVelB.length() * m_boundingRadiusB;
-			b3Vector3 relLinVel = (linVelB-linVelA);
+			b3Vector3 relLinVel = (linVelB - linVelA);
 			b3Scalar relLinVelocLength = relLinVel.dot(m_separatingNormal);
-			if (relLinVelocLength<0.f)
+			if (relLinVelocLength < 0.f)
 			{
 				relLinVelocLength = 0.f;
 			}
-	
-			b3Scalar	projectedMotion = maxAngularProjectedVelocity +relLinVelocLength;
+
+			b3Scalar projectedMotion = maxAngularProjectedVelocity + relLinVelocLength;
 			m_separatingDistance -= projectedMotion;
 		}
-	
+
 		m_posA = toPosA;
 		m_posB = toPosB;
 		m_ornA = toOrnA;
 		m_ornB = toOrnB;
 	}
 
-	void	initSeparatingDistance(const b3Vector3& separatingVector,b3Scalar separatingDistance,const b3Transform& transA,const b3Transform& transB)
+	void initSeparatingDistance(const b3Vector3& separatingVector, b3Scalar separatingDistance, const b3Transform& transA, const b3Transform& transB)
 	{
 		m_separatingDistance = separatingDistance;
 
-		if (m_separatingDistance>0.f)
+		if (m_separatingDistance > 0.f)
 		{
 			m_separatingNormal = separatingVector;
-			
+
 			const b3Vector3& toPosA = transA.getOrigin();
 			const b3Vector3& toPosB = transB.getOrigin();
 			b3Quaternion toOrnA = transA.getRotation();
@@ -220,9 +205,6 @@ public:
 			m_ornB = toOrnB;
 		}
 	}
-
 };
 
-
-#endif //B3_TRANSFORM_UTIL_H
-
+#endif  //B3_TRANSFORM_UTIL_H
diff --git a/thirdparty/bullet/Bullet3Common/b3Vector3.cpp b/thirdparty/bullet/Bullet3Common/b3Vector3.cpp
index 5f5ac4ac043..100fb774c1b 100644
--- a/thirdparty/bullet/Bullet3Common/b3Vector3.cpp
+++ b/thirdparty/bullet/Bullet3Common/b3Vector3.cpp
@@ -14,274 +14,281 @@
  This source version has been altered.
  */
 
-#if defined (_WIN32) || defined (__i386__)
+#if defined(_WIN32) || defined(__i386__)
 #define B3_USE_SSE_IN_API
 #endif
 
 #include "b3Vector3.h"
 
-#if defined (B3_USE_SSE) || defined (B3_USE_NEON)
+#if defined(B3_USE_SSE) || defined(B3_USE_NEON)
 
 #ifdef __APPLE__
 #include <stdint.h>
-typedef  float float4 __attribute__ ((vector_size(16)));
+typedef float float4 __attribute__((vector_size(16)));
 #else
 #define float4 __m128
 #endif
 //typedef  uint32_t uint4 __attribute__ ((vector_size(16)));
 
-
 #if defined B3_USE_SSE || defined _WIN32
 
-#define LOG2_ARRAY_SIZE     6
-#define STACK_ARRAY_COUNT   (1UL << LOG2_ARRAY_SIZE)
+#define LOG2_ARRAY_SIZE 6
+#define STACK_ARRAY_COUNT (1UL << LOG2_ARRAY_SIZE)
 
 #include <emmintrin.h>
 
-long b3_maxdot_large( const float *vv, const float *vec, unsigned long count, float *dotResult );
-long b3_maxdot_large( const float *vv, const float *vec, unsigned long count, float *dotResult )
+long b3_maxdot_large(const float *vv, const float *vec, unsigned long count, float *dotResult);
+long b3_maxdot_large(const float *vv, const float *vec, unsigned long count, float *dotResult)
 {
-    const float4 *vertices = (const float4*) vv;
-    static const unsigned char indexTable[16] = {(unsigned char)-1, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0 };
-    float4 dotMax = b3Assign128( -B3_INFINITY,  -B3_INFINITY,  -B3_INFINITY,  -B3_INFINITY );
-    float4 vvec = _mm_loadu_ps( vec );
-    float4 vHi = b3CastiTo128f(_mm_shuffle_epi32( b3CastfTo128i( vvec), 0xaa ));          /// zzzz
-    float4 vLo = _mm_movelh_ps( vvec, vvec );                               /// xyxy
+	const float4 *vertices = (const float4 *)vv;
+	static const unsigned char indexTable[16] = {(unsigned char)-1, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0};
+	float4 dotMax = b3Assign128(-B3_INFINITY, -B3_INFINITY, -B3_INFINITY, -B3_INFINITY);
+	float4 vvec = _mm_loadu_ps(vec);
+	float4 vHi = b3CastiTo128f(_mm_shuffle_epi32(b3CastfTo128i(vvec), 0xaa));  /// zzzz
+	float4 vLo = _mm_movelh_ps(vvec, vvec);                                    /// xyxy
 
-    long maxIndex = -1L;
+	long maxIndex = -1L;
 
-    size_t segment = 0;
-    float4 stack_array[ STACK_ARRAY_COUNT ];
+	size_t segment = 0;
+	float4 stack_array[STACK_ARRAY_COUNT];
 
 #if DEBUG
-    // memset( stack_array, -1, STACK_ARRAY_COUNT * sizeof(stack_array[0]) );
+	// memset( stack_array, -1, STACK_ARRAY_COUNT * sizeof(stack_array[0]) );
 #endif
 
-    size_t index;
-    float4 max;
-    // Faster loop without cleanup code for full tiles
-    for ( segment = 0; segment + STACK_ARRAY_COUNT*4 <= count; segment += STACK_ARRAY_COUNT*4 )
-    {
-        max = dotMax;
+	size_t index;
+	float4 max;
+	// Faster loop without cleanup code for full tiles
+	for (segment = 0; segment + STACK_ARRAY_COUNT * 4 <= count; segment += STACK_ARRAY_COUNT * 4)
+	{
+		max = dotMax;
 
-        for( index = 0; index < STACK_ARRAY_COUNT; index+= 4 )
-        { // do four dot products at a time. Carefully avoid touching the w element.
-            float4 v0 = vertices[0];
-            float4 v1 = vertices[1];
-            float4 v2 = vertices[2];
-            float4 v3 = vertices[3];            vertices += 4;
+		for (index = 0; index < STACK_ARRAY_COUNT; index += 4)
+		{  // do four dot products at a time. Carefully avoid touching the w element.
+			float4 v0 = vertices[0];
+			float4 v1 = vertices[1];
+			float4 v2 = vertices[2];
+			float4 v3 = vertices[3];
+			vertices += 4;
 
-            float4 lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            float4 hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            float4 lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            float4 hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
+			float4 lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			float4 hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			float4 lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			float4 hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
 
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index] = x;
-            max = _mm_max_ps( x, max );         // control the order here so that max is never NaN even if x is nan
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index] = x;
+			max = _mm_max_ps(x, max);  // control the order here so that max is never NaN even if x is nan
 
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
 
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
 
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+1] = x;
-            max = _mm_max_ps( x, max );         // control the order here so that max is never NaN even if x is nan
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 1] = x;
+			max = _mm_max_ps(x, max);  // control the order here so that max is never NaN even if x is nan
 
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
 
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
 
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+2] = x;
-            max = _mm_max_ps( x, max );         // control the order here so that max is never NaN even if x is nan
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 2] = x;
+			max = _mm_max_ps(x, max);  // control the order here so that max is never NaN even if x is nan
 
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
 
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
 
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+3] = x;
-            max = _mm_max_ps( x, max );         // control the order here so that max is never NaN even if x is nan
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 3] = x;
+			max = _mm_max_ps(x, max);  // control the order here so that max is never NaN even if x is nan
 
-            // It is too costly to keep the index of the max here. We will look for it again later.  We save a lot of work this way.
-        }
+			// It is too costly to keep the index of the max here. We will look for it again later.  We save a lot of work this way.
+		}
 
-        // If we found a new max
-        if( 0xf != _mm_movemask_ps( (float4) _mm_cmpeq_ps(max, dotMax)))
-        {
-            // copy the new max across all lanes of our max accumulator
-            max = _mm_max_ps(max, (float4) _mm_shuffle_ps( max, max, 0x4e));
-            max = _mm_max_ps(max, (float4) _mm_shuffle_ps( max, max, 0xb1));
+		// If we found a new max
+		if (0xf != _mm_movemask_ps((float4)_mm_cmpeq_ps(max, dotMax)))
+		{
+			// copy the new max across all lanes of our max accumulator
+			max = _mm_max_ps(max, (float4)_mm_shuffle_ps(max, max, 0x4e));
+			max = _mm_max_ps(max, (float4)_mm_shuffle_ps(max, max, 0xb1));
 
-            dotMax = max;
+			dotMax = max;
 
-            // find first occurrence of that max
-            size_t test;
-            for( index = 0; 0 == (test=_mm_movemask_ps( _mm_cmpeq_ps( stack_array[index], max))); index++ )   // local_count must be a multiple of 4
-            {}
-            // record where it is.
-            maxIndex = 4*index + segment + indexTable[test];
-        }
-    }
+			// find first occurrence of that max
+			size_t test;
+			for (index = 0; 0 == (test = _mm_movemask_ps(_mm_cmpeq_ps(stack_array[index], max))); index++)  // local_count must be a multiple of 4
+			{
+			}
+			// record where it is.
+			maxIndex = 4 * index + segment + indexTable[test];
+		}
+	}
 
-    // account for work we've already done
-    count -= segment;
+	// account for work we've already done
+	count -= segment;
 
-    // Deal with the last < STACK_ARRAY_COUNT vectors
-    max = dotMax;
-    index = 0;
+	// Deal with the last < STACK_ARRAY_COUNT vectors
+	max = dotMax;
+	index = 0;
 
+	if (b3Unlikely(count > 16))
+	{
+		for (; index + 4 <= count / 4; index += 4)
+		{  // do four dot products at a time. Carefully avoid touching the w element.
+			float4 v0 = vertices[0];
+			float4 v1 = vertices[1];
+			float4 v2 = vertices[2];
+			float4 v3 = vertices[3];
+			vertices += 4;
 
-    if( b3Unlikely( count > 16) )
-    {
-        for( ; index + 4 <= count / 4; index+=4 )
-        { // do four dot products at a time. Carefully avoid touching the w element.
-            float4 v0 = vertices[0];
-            float4 v1 = vertices[1];
-            float4 v2 = vertices[2];
-            float4 v3 = vertices[3];            vertices += 4;
+			float4 lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			float4 hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			float4 lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			float4 hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
 
-            float4 lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            float4 hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            float4 lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            float4 hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index] = x;
+			max = _mm_max_ps(x, max);  // control the order here so that max is never NaN even if x is nan
 
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index] = x;
-            max = _mm_max_ps( x, max );         // control the order here so that max is never NaN even if x is nan
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
 
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
 
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 1] = x;
+			max = _mm_max_ps(x, max);  // control the order here so that max is never NaN even if x is nan
 
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+1] = x;
-            max = _mm_max_ps( x, max );         // control the order here so that max is never NaN even if x is nan
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
 
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
 
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 2] = x;
+			max = _mm_max_ps(x, max);  // control the order here so that max is never NaN even if x is nan
 
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+2] = x;
-            max = _mm_max_ps( x, max );         // control the order here so that max is never NaN even if x is nan
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
 
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
 
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 3] = x;
+			max = _mm_max_ps(x, max);  // control the order here so that max is never NaN even if x is nan
 
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+3] = x;
-            max = _mm_max_ps( x, max );         // control the order here so that max is never NaN even if x is nan
+			// It is too costly to keep the index of the max here. We will look for it again later.  We save a lot of work this way.
+		}
+	}
 
-            // It is too costly to keep the index of the max here. We will look for it again later.  We save a lot of work this way.
-        }
-    }
-
-    size_t localCount = (count & -4L) - 4*index;
-    if( localCount )
-    {
+	size_t localCount = (count & -4L) - 4 * index;
+	if (localCount)
+	{
 #ifdef __APPLE__
-        float4 t0, t1, t2, t3, t4;
-        float4 * sap = &stack_array[index + localCount / 4];
-          vertices += localCount;      // counter the offset
-         size_t byteIndex = -(localCount) * sizeof(float);
-        //AT&T Code style assembly
-        asm volatile
-        (   ".align 4                                                                   \n\
+		float4 t0, t1, t2, t3, t4;
+		float4 *sap = &stack_array[index + localCount / 4];
+		vertices += localCount;  // counter the offset
+		size_t byteIndex = -(localCount) * sizeof(float);
+		//AT&T Code style assembly
+		asm volatile(
+			".align 4                                                                   \n\
              0: movaps  %[max], %[t2]                            // move max out of the way to avoid propagating NaNs in max \n\
           movaps  (%[vertices], %[byteIndex], 4),    %[t0]    // vertices[0]      \n\
           movaps  16(%[vertices], %[byteIndex], 4),  %[t1]    // vertices[1]      \n\
@@ -307,369 +314,375 @@ long b3_maxdot_large( const float *vv, const float *vec, unsigned long count, fl
          add     $16, %[byteIndex]                           // advance loop counter\n\
          jnz     0b                                          \n\
      "
-         : [max] "+x" (max), [t0] "=&x" (t0), [t1] "=&x" (t1), [t2] "=&x" (t2), [t3] "=&x" (t3), [t4] "=&x" (t4), [byteIndex] "+r" (byteIndex)
-         : [vLo] "x" (vLo), [vHi] "x" (vHi), [vertices] "r" (vertices), [sap] "r" (sap)
-         : "memory", "cc"
-         );
-        index += localCount/4;
+			: [max] "+x"(max), [t0] "=&x"(t0), [t1] "=&x"(t1), [t2] "=&x"(t2), [t3] "=&x"(t3), [t4] "=&x"(t4), [byteIndex] "+r"(byteIndex)
+			: [vLo] "x"(vLo), [vHi] "x"(vHi), [vertices] "r"(vertices), [sap] "r"(sap)
+			: "memory", "cc");
+		index += localCount / 4;
 #else
-        {
-            for( unsigned int i=0; i<localCount/4; i++,index++)
-            { // do four dot products at a time. Carefully avoid touching the w element.
-                float4 v0 = vertices[0];
-                float4 v1 = vertices[1];
-                float4 v2 = vertices[2];
-                float4 v3 = vertices[3];
-                vertices += 4;
+		{
+			for (unsigned int i = 0; i < localCount / 4; i++, index++)
+			{  // do four dot products at a time. Carefully avoid touching the w element.
+				float4 v0 = vertices[0];
+				float4 v1 = vertices[1];
+				float4 v2 = vertices[2];
+				float4 v3 = vertices[3];
+				vertices += 4;
 
-                float4 lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-                float4 hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-                float4 lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-                float4 hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
+				float4 lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+				float4 hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+				float4 lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+				float4 hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
 
-                lo0 = lo0*vLo;
-                lo1 = lo1*vLo;
-                float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
-                float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
-                float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-                z = z*vHi;
-                x = x+y;
-                x = x+z;
-                stack_array[index] = x;
-                max = _mm_max_ps( x, max );         // control the order here so that max is never NaN even if x is nan
-            }
-        }
-#endif //__APPLE__
-    }
+				lo0 = lo0 * vLo;
+				lo1 = lo1 * vLo;
+				float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
+				float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
+				float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+				z = z * vHi;
+				x = x + y;
+				x = x + z;
+				stack_array[index] = x;
+				max = _mm_max_ps(x, max);  // control the order here so that max is never NaN even if x is nan
+			}
+		}
+#endif  //__APPLE__
+	}
 
-    // process the last few points
-    if( count & 3 )
-    {
-        float4 v0, v1, v2, x, y, z;
-        switch( count & 3 )
-        {
-            case 3:
-            {
-                v0 = vertices[0];
-                v1 = vertices[1];
-                v2 = vertices[2];
+	// process the last few points
+	if (count & 3)
+	{
+		float4 v0, v1, v2, x, y, z;
+		switch (count & 3)
+		{
+			case 3:
+			{
+				v0 = vertices[0];
+				v1 = vertices[1];
+				v2 = vertices[2];
 
-                // Calculate 3 dot products, transpose, duplicate v2
-                float4 lo0 = _mm_movelh_ps( v0, v1);        // xyxy.lo
-                float4 hi0 = _mm_movehl_ps( v1, v0);        // z?z?.lo
-                lo0 = lo0*vLo;
-                z = _mm_shuffle_ps(hi0, v2,  0xa8 );           // z0z1z2z2
-                z = z*vHi;
-                float4 lo1 = _mm_movelh_ps(v2, v2);          // xyxy
-                lo1 = lo1*vLo;
-                x = _mm_shuffle_ps(lo0, lo1, 0x88);
-                y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            }
-                break;
-            case 2:
-            {
-                v0 = vertices[0];
-                v1 = vertices[1];
-                float4 xy = _mm_movelh_ps(v0, v1);
-                z = _mm_movehl_ps(v1, v0);
-                xy = xy*vLo;
-                z = _mm_shuffle_ps( z, z,  0xa8);
-                x = _mm_shuffle_ps( xy, xy, 0xa8);
-                y = _mm_shuffle_ps( xy, xy, 0xfd);
-                z = z*vHi;
-            }
-                break;
-            case 1:
-            {
-                float4 xy = vertices[0];
-                z =  _mm_shuffle_ps( xy, xy, 0xaa);
-                xy = xy*vLo;
-                z = z*vHi;
-                x = _mm_shuffle_ps(xy, xy, 0);
-                y = _mm_shuffle_ps(xy, xy, 0x55);
-            }
-                break;
-        }
-        x = x+y;
-        x = x+z;
-        stack_array[index] = x;
-        max = _mm_max_ps( x, max );         // control the order here so that max is never NaN even if x is nan
-        index++;
-    }
+				// Calculate 3 dot products, transpose, duplicate v2
+				float4 lo0 = _mm_movelh_ps(v0, v1);  // xyxy.lo
+				float4 hi0 = _mm_movehl_ps(v1, v0);  // z?z?.lo
+				lo0 = lo0 * vLo;
+				z = _mm_shuffle_ps(hi0, v2, 0xa8);  // z0z1z2z2
+				z = z * vHi;
+				float4 lo1 = _mm_movelh_ps(v2, v2);  // xyxy
+				lo1 = lo1 * vLo;
+				x = _mm_shuffle_ps(lo0, lo1, 0x88);
+				y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			}
+			break;
+			case 2:
+			{
+				v0 = vertices[0];
+				v1 = vertices[1];
+				float4 xy = _mm_movelh_ps(v0, v1);
+				z = _mm_movehl_ps(v1, v0);
+				xy = xy * vLo;
+				z = _mm_shuffle_ps(z, z, 0xa8);
+				x = _mm_shuffle_ps(xy, xy, 0xa8);
+				y = _mm_shuffle_ps(xy, xy, 0xfd);
+				z = z * vHi;
+			}
+			break;
+			case 1:
+			{
+				float4 xy = vertices[0];
+				z = _mm_shuffle_ps(xy, xy, 0xaa);
+				xy = xy * vLo;
+				z = z * vHi;
+				x = _mm_shuffle_ps(xy, xy, 0);
+				y = _mm_shuffle_ps(xy, xy, 0x55);
+			}
+			break;
+		}
+		x = x + y;
+		x = x + z;
+		stack_array[index] = x;
+		max = _mm_max_ps(x, max);  // control the order here so that max is never NaN even if x is nan
+		index++;
+	}
 
-    // if we found a new max.
-    if( 0 == segment || 0xf != _mm_movemask_ps( (float4) _mm_cmpeq_ps(max, dotMax)))
-    { // we found a new max. Search for it
-      // find max across the max vector, place in all elements of max -- big latency hit here
-        max = _mm_max_ps(max, (float4) _mm_shuffle_ps( max, max, 0x4e));
-        max = _mm_max_ps(max, (float4) _mm_shuffle_ps( max, max, 0xb1));
+	// if we found a new max.
+	if (0 == segment || 0xf != _mm_movemask_ps((float4)_mm_cmpeq_ps(max, dotMax)))
+	{  // we found a new max. Search for it
+		// find max across the max vector, place in all elements of max -- big latency hit here
+		max = _mm_max_ps(max, (float4)_mm_shuffle_ps(max, max, 0x4e));
+		max = _mm_max_ps(max, (float4)_mm_shuffle_ps(max, max, 0xb1));
 
-        // It is slightly faster to do this part in scalar code when count < 8. However, the common case for
-        // this where it actually makes a difference is handled in the early out at the top of the function,
-        // so it is less than a 1% difference here. I opted for improved code size, fewer branches and reduced
-        // complexity, and removed it.
+		// It is slightly faster to do this part in scalar code when count < 8. However, the common case for
+		// this where it actually makes a difference is handled in the early out at the top of the function,
+		// so it is less than a 1% difference here. I opted for improved code size, fewer branches and reduced
+		// complexity, and removed it.
 
-        dotMax = max;
+		dotMax = max;
 
-        // scan for the first occurence of max in the array
-        size_t test;
-        for( index = 0; 0 == (test=_mm_movemask_ps( _mm_cmpeq_ps( stack_array[index], max))); index++ )   // local_count must be a multiple of 4
-        {}
-        maxIndex = 4*index + segment + indexTable[test];
-    }
+		// scan for the first occurence of max in the array
+		size_t test;
+		for (index = 0; 0 == (test = _mm_movemask_ps(_mm_cmpeq_ps(stack_array[index], max))); index++)  // local_count must be a multiple of 4
+		{
+		}
+		maxIndex = 4 * index + segment + indexTable[test];
+	}
 
-    _mm_store_ss( dotResult, dotMax);
-    return maxIndex;
+	_mm_store_ss(dotResult, dotMax);
+	return maxIndex;
 }
 
-long b3_mindot_large( const float *vv, const float *vec, unsigned long count, float *dotResult );
+long b3_mindot_large(const float *vv, const float *vec, unsigned long count, float *dotResult);
 
-long b3_mindot_large( const float *vv, const float *vec, unsigned long count, float *dotResult )
+long b3_mindot_large(const float *vv, const float *vec, unsigned long count, float *dotResult)
 {
-    const float4 *vertices = (const float4*) vv;
-    static const unsigned char indexTable[16] = {(unsigned char)-1, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0 };
+	const float4 *vertices = (const float4 *)vv;
+	static const unsigned char indexTable[16] = {(unsigned char)-1, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0};
 
-    float4 dotmin = b3Assign128( B3_INFINITY,  B3_INFINITY,  B3_INFINITY,  B3_INFINITY );
-    float4 vvec = _mm_loadu_ps( vec );
-    float4 vHi = b3CastiTo128f(_mm_shuffle_epi32( b3CastfTo128i( vvec), 0xaa ));          /// zzzz
-    float4 vLo = _mm_movelh_ps( vvec, vvec );                               /// xyxy
+	float4 dotmin = b3Assign128(B3_INFINITY, B3_INFINITY, B3_INFINITY, B3_INFINITY);
+	float4 vvec = _mm_loadu_ps(vec);
+	float4 vHi = b3CastiTo128f(_mm_shuffle_epi32(b3CastfTo128i(vvec), 0xaa));  /// zzzz
+	float4 vLo = _mm_movelh_ps(vvec, vvec);                                    /// xyxy
 
-    long minIndex = -1L;
+	long minIndex = -1L;
 
-    size_t segment = 0;
-    float4 stack_array[ STACK_ARRAY_COUNT ];
+	size_t segment = 0;
+	float4 stack_array[STACK_ARRAY_COUNT];
 
 #if DEBUG
-    // memset( stack_array, -1, STACK_ARRAY_COUNT * sizeof(stack_array[0]) );
+	// memset( stack_array, -1, STACK_ARRAY_COUNT * sizeof(stack_array[0]) );
 #endif
 
-    size_t index;
-    float4 min;
-    // Faster loop without cleanup code for full tiles
-    for ( segment = 0; segment + STACK_ARRAY_COUNT*4 <= count; segment += STACK_ARRAY_COUNT*4 )
-    {
-        min = dotmin;
+	size_t index;
+	float4 min;
+	// Faster loop without cleanup code for full tiles
+	for (segment = 0; segment + STACK_ARRAY_COUNT * 4 <= count; segment += STACK_ARRAY_COUNT * 4)
+	{
+		min = dotmin;
 
-        for( index = 0; index < STACK_ARRAY_COUNT; index+= 4 )
-        { // do four dot products at a time. Carefully avoid touching the w element.
-            float4 v0 = vertices[0];
-            float4 v1 = vertices[1];
-            float4 v2 = vertices[2];
-            float4 v3 = vertices[3];            vertices += 4;
+		for (index = 0; index < STACK_ARRAY_COUNT; index += 4)
+		{  // do four dot products at a time. Carefully avoid touching the w element.
+			float4 v0 = vertices[0];
+			float4 v1 = vertices[1];
+			float4 v2 = vertices[2];
+			float4 v3 = vertices[3];
+			vertices += 4;
 
-            float4 lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            float4 hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            float4 lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            float4 hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
+			float4 lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			float4 hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			float4 lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			float4 hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
 
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index] = x;
-            min = _mm_min_ps( x, min );         // control the order here so that min is never NaN even if x is nan
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index] = x;
+			min = _mm_min_ps(x, min);  // control the order here so that min is never NaN even if x is nan
 
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
 
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
 
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+1] = x;
-            min = _mm_min_ps( x, min );         // control the order here so that min is never NaN even if x is nan
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 1] = x;
+			min = _mm_min_ps(x, min);  // control the order here so that min is never NaN even if x is nan
 
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
 
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
 
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+2] = x;
-            min = _mm_min_ps( x, min );         // control the order here so that min is never NaN even if x is nan
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 2] = x;
+			min = _mm_min_ps(x, min);  // control the order here so that min is never NaN even if x is nan
 
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
 
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
 
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+3] = x;
-            min = _mm_min_ps( x, min );         // control the order here so that min is never NaN even if x is nan
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 3] = x;
+			min = _mm_min_ps(x, min);  // control the order here so that min is never NaN even if x is nan
 
-            // It is too costly to keep the index of the min here. We will look for it again later.  We save a lot of work this way.
-        }
+			// It is too costly to keep the index of the min here. We will look for it again later.  We save a lot of work this way.
+		}
 
-        // If we found a new min
-        if( 0xf != _mm_movemask_ps( (float4) _mm_cmpeq_ps(min, dotmin)))
-        {
-            // copy the new min across all lanes of our min accumulator
-            min = _mm_min_ps(min, (float4) _mm_shuffle_ps( min, min, 0x4e));
-            min = _mm_min_ps(min, (float4) _mm_shuffle_ps( min, min, 0xb1));
+		// If we found a new min
+		if (0xf != _mm_movemask_ps((float4)_mm_cmpeq_ps(min, dotmin)))
+		{
+			// copy the new min across all lanes of our min accumulator
+			min = _mm_min_ps(min, (float4)_mm_shuffle_ps(min, min, 0x4e));
+			min = _mm_min_ps(min, (float4)_mm_shuffle_ps(min, min, 0xb1));
 
-            dotmin = min;
+			dotmin = min;
 
-            // find first occurrence of that min
-            size_t test;
-            for( index = 0; 0 == (test=_mm_movemask_ps( _mm_cmpeq_ps( stack_array[index], min))); index++ )   // local_count must be a multiple of 4
-            {}
-            // record where it is.
-            minIndex = 4*index + segment + indexTable[test];
-        }
-    }
+			// find first occurrence of that min
+			size_t test;
+			for (index = 0; 0 == (test = _mm_movemask_ps(_mm_cmpeq_ps(stack_array[index], min))); index++)  // local_count must be a multiple of 4
+			{
+			}
+			// record where it is.
+			minIndex = 4 * index + segment + indexTable[test];
+		}
+	}
 
-    // account for work we've already done
-    count -= segment;
+	// account for work we've already done
+	count -= segment;
 
-    // Deal with the last < STACK_ARRAY_COUNT vectors
-    min = dotmin;
-    index = 0;
+	// Deal with the last < STACK_ARRAY_COUNT vectors
+	min = dotmin;
+	index = 0;
 
+	if (b3Unlikely(count > 16))
+	{
+		for (; index + 4 <= count / 4; index += 4)
+		{  // do four dot products at a time. Carefully avoid touching the w element.
+			float4 v0 = vertices[0];
+			float4 v1 = vertices[1];
+			float4 v2 = vertices[2];
+			float4 v3 = vertices[3];
+			vertices += 4;
 
-    if(b3Unlikely( count > 16) )
-    {
-        for( ; index + 4 <= count / 4; index+=4 )
-        { // do four dot products at a time. Carefully avoid touching the w element.
-            float4 v0 = vertices[0];
-            float4 v1 = vertices[1];
-            float4 v2 = vertices[2];
-            float4 v3 = vertices[3];            vertices += 4;
+			float4 lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			float4 hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			float4 lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			float4 hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
 
-            float4 lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            float4 hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            float4 lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            float4 hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index] = x;
+			min = _mm_min_ps(x, min);  // control the order here so that min is never NaN even if x is nan
 
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index] = x;
-            min = _mm_min_ps( x, min );         // control the order here so that min is never NaN even if x is nan
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
 
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
 
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 1] = x;
+			min = _mm_min_ps(x, min);  // control the order here so that min is never NaN even if x is nan
 
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+1] = x;
-            min = _mm_min_ps( x, min );         // control the order here so that min is never NaN even if x is nan
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
 
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
 
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 2] = x;
+			min = _mm_min_ps(x, min);  // control the order here so that min is never NaN even if x is nan
 
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+2] = x;
-            min = _mm_min_ps( x, min );         // control the order here so that min is never NaN even if x is nan
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
 
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
 
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
-
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+3] = x;
-            min = _mm_min_ps( x, min );         // control the order here so that min is never NaN even if x is nan
-
-            // It is too costly to keep the index of the min here. We will look for it again later.  We save a lot of work this way.
-        }
-    }
-
-    size_t localCount = (count & -4L) - 4*index;
-    if( localCount )
-    {
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 3] = x;
+			min = _mm_min_ps(x, min);  // control the order here so that min is never NaN even if x is nan
 
+			// It is too costly to keep the index of the min here. We will look for it again later.  We save a lot of work this way.
+		}
+	}
 
+	size_t localCount = (count & -4L) - 4 * index;
+	if (localCount)
+	{
 #ifdef __APPLE__
-        vertices += localCount;      // counter the offset
-        float4 t0, t1, t2, t3, t4;
-        size_t byteIndex = -(localCount) * sizeof(float);
-        float4 * sap = &stack_array[index + localCount / 4];
+		vertices += localCount;  // counter the offset
+		float4 t0, t1, t2, t3, t4;
+		size_t byteIndex = -(localCount) * sizeof(float);
+		float4 *sap = &stack_array[index + localCount / 4];
 
-        asm volatile
-        (   ".align 4                                                                   \n\
+		asm volatile(
+			".align 4                                                                   \n\
              0: movaps  %[min], %[t2]                            // move min out of the way to avoid propagating NaNs in min \n\
              movaps  (%[vertices], %[byteIndex], 4),    %[t0]    // vertices[0]      \n\
              movaps  16(%[vertices], %[byteIndex], 4),  %[t1]    // vertices[1]      \n\
@@ -695,937 +708,930 @@ long b3_mindot_large( const float *vv, const float *vec, unsigned long count, fl
              add     $16, %[byteIndex]                           // advance loop counter\n\
              jnz     0b                                          \n\
              "
-         : [min] "+x" (min), [t0] "=&x" (t0), [t1] "=&x" (t1), [t2] "=&x" (t2), [t3] "=&x" (t3), [t4] "=&x" (t4), [byteIndex] "+r" (byteIndex)
-         : [vLo] "x" (vLo), [vHi] "x" (vHi), [vertices] "r" (vertices), [sap] "r" (sap)
-         : "memory", "cc"
-         );
-        index += localCount/4;
+			: [min] "+x"(min), [t0] "=&x"(t0), [t1] "=&x"(t1), [t2] "=&x"(t2), [t3] "=&x"(t3), [t4] "=&x"(t4), [byteIndex] "+r"(byteIndex)
+			: [vLo] "x"(vLo), [vHi] "x"(vHi), [vertices] "r"(vertices), [sap] "r"(sap)
+			: "memory", "cc");
+		index += localCount / 4;
 #else
-        {
-            for( unsigned int i=0; i<localCount/4; i++,index++)
-            { // do four dot products at a time. Carefully avoid touching the w element.
-                float4 v0 = vertices[0];
-                float4 v1 = vertices[1];
-                float4 v2 = vertices[2];
-                float4 v3 = vertices[3];
-                vertices += 4;
+		{
+			for (unsigned int i = 0; i < localCount / 4; i++, index++)
+			{  // do four dot products at a time. Carefully avoid touching the w element.
+				float4 v0 = vertices[0];
+				float4 v1 = vertices[1];
+				float4 v2 = vertices[2];
+				float4 v3 = vertices[3];
+				vertices += 4;
 
-                float4 lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-                float4 hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-                float4 lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-                float4 hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
+				float4 lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+				float4 hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+				float4 lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+				float4 hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
 
-                lo0 = lo0*vLo;
-                lo1 = lo1*vLo;
-                float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
-                float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
-                float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-                z = z*vHi;
-                x = x+y;
-                x = x+z;
-                stack_array[index] = x;
-                min = _mm_min_ps( x, min );         // control the order here so that max is never NaN even if x is nan
-            }
-        }
+				lo0 = lo0 * vLo;
+				lo1 = lo1 * vLo;
+				float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
+				float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
+				float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+				z = z * vHi;
+				x = x + y;
+				x = x + z;
+				stack_array[index] = x;
+				min = _mm_min_ps(x, min);  // control the order here so that max is never NaN even if x is nan
+			}
+		}
 
 #endif
-    }
+	}
 
-    // process the last few points
-    if( count & 3 )
-    {
-        float4 v0, v1, v2, x, y, z;
-        switch( count & 3 )
-        {
-            case 3:
-            {
-                v0 = vertices[0];
-                v1 = vertices[1];
-                v2 = vertices[2];
+	// process the last few points
+	if (count & 3)
+	{
+		float4 v0, v1, v2, x, y, z;
+		switch (count & 3)
+		{
+			case 3:
+			{
+				v0 = vertices[0];
+				v1 = vertices[1];
+				v2 = vertices[2];
 
-                // Calculate 3 dot products, transpose, duplicate v2
-                float4 lo0 = _mm_movelh_ps( v0, v1);        // xyxy.lo
-                float4 hi0 = _mm_movehl_ps( v1, v0);        // z?z?.lo
-                lo0 = lo0*vLo;
-                z = _mm_shuffle_ps(hi0, v2,  0xa8 );           // z0z1z2z2
-                z = z*vHi;
-                float4 lo1 = _mm_movelh_ps(v2, v2);          // xyxy
-                lo1 = lo1*vLo;
-                x = _mm_shuffle_ps(lo0, lo1, 0x88);
-                y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            }
-                break;
-            case 2:
-            {
-                v0 = vertices[0];
-                v1 = vertices[1];
-                float4 xy = _mm_movelh_ps(v0, v1);
-                z = _mm_movehl_ps(v1, v0);
-                xy = xy*vLo;
-                z = _mm_shuffle_ps( z, z,  0xa8);
-                x = _mm_shuffle_ps( xy, xy, 0xa8);
-                y = _mm_shuffle_ps( xy, xy, 0xfd);
-                z = z*vHi;
-            }
-                break;
-            case 1:
-            {
-                float4 xy = vertices[0];
-                z =  _mm_shuffle_ps( xy, xy, 0xaa);
-                xy = xy*vLo;
-                z = z*vHi;
-                x = _mm_shuffle_ps(xy, xy, 0);
-                y = _mm_shuffle_ps(xy, xy, 0x55);
-            }
-                break;
-        }
-        x = x+y;
-        x = x+z;
-        stack_array[index] = x;
-        min = _mm_min_ps( x, min );         // control the order here so that min is never NaN even if x is nan
-        index++;
-    }
+				// Calculate 3 dot products, transpose, duplicate v2
+				float4 lo0 = _mm_movelh_ps(v0, v1);  // xyxy.lo
+				float4 hi0 = _mm_movehl_ps(v1, v0);  // z?z?.lo
+				lo0 = lo0 * vLo;
+				z = _mm_shuffle_ps(hi0, v2, 0xa8);  // z0z1z2z2
+				z = z * vHi;
+				float4 lo1 = _mm_movelh_ps(v2, v2);  // xyxy
+				lo1 = lo1 * vLo;
+				x = _mm_shuffle_ps(lo0, lo1, 0x88);
+				y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			}
+			break;
+			case 2:
+			{
+				v0 = vertices[0];
+				v1 = vertices[1];
+				float4 xy = _mm_movelh_ps(v0, v1);
+				z = _mm_movehl_ps(v1, v0);
+				xy = xy * vLo;
+				z = _mm_shuffle_ps(z, z, 0xa8);
+				x = _mm_shuffle_ps(xy, xy, 0xa8);
+				y = _mm_shuffle_ps(xy, xy, 0xfd);
+				z = z * vHi;
+			}
+			break;
+			case 1:
+			{
+				float4 xy = vertices[0];
+				z = _mm_shuffle_ps(xy, xy, 0xaa);
+				xy = xy * vLo;
+				z = z * vHi;
+				x = _mm_shuffle_ps(xy, xy, 0);
+				y = _mm_shuffle_ps(xy, xy, 0x55);
+			}
+			break;
+		}
+		x = x + y;
+		x = x + z;
+		stack_array[index] = x;
+		min = _mm_min_ps(x, min);  // control the order here so that min is never NaN even if x is nan
+		index++;
+	}
 
-    // if we found a new min.
-    if( 0 == segment || 0xf != _mm_movemask_ps( (float4) _mm_cmpeq_ps(min, dotmin)))
-    { // we found a new min. Search for it
-      // find min across the min vector, place in all elements of min -- big latency hit here
-        min = _mm_min_ps(min, (float4) _mm_shuffle_ps( min, min, 0x4e));
-        min = _mm_min_ps(min, (float4) _mm_shuffle_ps( min, min, 0xb1));
+	// if we found a new min.
+	if (0 == segment || 0xf != _mm_movemask_ps((float4)_mm_cmpeq_ps(min, dotmin)))
+	{  // we found a new min. Search for it
+		// find min across the min vector, place in all elements of min -- big latency hit here
+		min = _mm_min_ps(min, (float4)_mm_shuffle_ps(min, min, 0x4e));
+		min = _mm_min_ps(min, (float4)_mm_shuffle_ps(min, min, 0xb1));
 
-        // It is slightly faster to do this part in scalar code when count < 8. However, the common case for
-        // this where it actually makes a difference is handled in the early out at the top of the function,
-        // so it is less than a 1% difference here. I opted for improved code size, fewer branches and reduced
-        // complexity, and removed it.
+		// It is slightly faster to do this part in scalar code when count < 8. However, the common case for
+		// this where it actually makes a difference is handled in the early out at the top of the function,
+		// so it is less than a 1% difference here. I opted for improved code size, fewer branches and reduced
+		// complexity, and removed it.
 
-        dotmin = min;
+		dotmin = min;
 
-        // scan for the first occurence of min in the array
-        size_t test;
-        for( index = 0; 0 == (test=_mm_movemask_ps( _mm_cmpeq_ps( stack_array[index], min))); index++ )   // local_count must be a multiple of 4
-        {}
-        minIndex = 4*index + segment + indexTable[test];
-    }
+		// scan for the first occurence of min in the array
+		size_t test;
+		for (index = 0; 0 == (test = _mm_movemask_ps(_mm_cmpeq_ps(stack_array[index], min))); index++)  // local_count must be a multiple of 4
+		{
+		}
+		minIndex = 4 * index + segment + indexTable[test];
+	}
 
-    _mm_store_ss( dotResult, dotmin);
-    return minIndex;
+	_mm_store_ss(dotResult, dotmin);
+	return minIndex;
 }
 
-
 #elif defined B3_USE_NEON
-#define ARM_NEON_GCC_COMPATIBILITY  1
+#define ARM_NEON_GCC_COMPATIBILITY 1
 #include <arm_neon.h>
 
+static long b3_maxdot_large_v0(const float *vv, const float *vec, unsigned long count, float *dotResult);
+static long b3_maxdot_large_v1(const float *vv, const float *vec, unsigned long count, float *dotResult);
+static long b3_maxdot_large_sel(const float *vv, const float *vec, unsigned long count, float *dotResult);
+static long b3_mindot_large_v0(const float *vv, const float *vec, unsigned long count, float *dotResult);
+static long b3_mindot_large_v1(const float *vv, const float *vec, unsigned long count, float *dotResult);
+static long b3_mindot_large_sel(const float *vv, const float *vec, unsigned long count, float *dotResult);
 
-static long b3_maxdot_large_v0( const float *vv, const float *vec, unsigned long count, float *dotResult );
-static long b3_maxdot_large_v1( const float *vv, const float *vec, unsigned long count, float *dotResult );
-static long b3_maxdot_large_sel( const float *vv, const float *vec, unsigned long count, float *dotResult );
-static long b3_mindot_large_v0( const float *vv, const float *vec, unsigned long count, float *dotResult );
-static long b3_mindot_large_v1( const float *vv, const float *vec, unsigned long count, float *dotResult );
-static long b3_mindot_large_sel( const float *vv, const float *vec, unsigned long count, float *dotResult );
+long (*b3_maxdot_large)(const float *vv, const float *vec, unsigned long count, float *dotResult) = b3_maxdot_large_sel;
+long (*b3_mindot_large)(const float *vv, const float *vec, unsigned long count, float *dotResult) = b3_mindot_large_sel;
 
-long (*b3_maxdot_large)( const float *vv, const float *vec, unsigned long count, float *dotResult ) = b3_maxdot_large_sel;
-long (*b3_mindot_large)( const float *vv, const float *vec, unsigned long count, float *dotResult ) = b3_mindot_large_sel;
-
-extern "C" {int  _get_cpu_capabilities( void );}
-
-static long b3_maxdot_large_sel( const float *vv, const float *vec, unsigned long count, float *dotResult )
+extern "C"
 {
-    if( _get_cpu_capabilities() & 0x2000 )
-        b3_maxdot_large = _maxdot_large_v1;
-    else
-        b3_maxdot_large = _maxdot_large_v0;
-
-    return b3_maxdot_large(vv, vec, count, dotResult);
+	int _get_cpu_capabilities(void);
 }
 
-static long b3_mindot_large_sel( const float *vv, const float *vec, unsigned long count, float *dotResult )
+static long b3_maxdot_large_sel(const float *vv, const float *vec, unsigned long count, float *dotResult)
 {
-    if( _get_cpu_capabilities() & 0x2000 )
-        b3_mindot_large = _mindot_large_v1;
-    else
-        b3_mindot_large = _mindot_large_v0;
+	if (_get_cpu_capabilities() & 0x2000)
+		b3_maxdot_large = _maxdot_large_v1;
+	else
+		b3_maxdot_large = _maxdot_large_v0;
 
-    return b3_mindot_large(vv, vec, count, dotResult);
+	return b3_maxdot_large(vv, vec, count, dotResult);
 }
 
-
-
-#define vld1q_f32_aligned_postincrement( _ptr ) ({ float32x4_t _r; asm( "vld1.f32  {%0}, [%1, :128]!\n" : "=w" (_r), "+r" (_ptr) ); /*return*/ _r; })
-
-
-long b3_maxdot_large_v0( const float *vv, const float *vec, unsigned long count, float *dotResult )
+static long b3_mindot_large_sel(const float *vv, const float *vec, unsigned long count, float *dotResult)
 {
-    unsigned long i = 0;
-    float32x4_t vvec = vld1q_f32_aligned_postincrement( vec );
-    float32x2_t vLo = vget_low_f32(vvec);
-    float32x2_t vHi = vdup_lane_f32(vget_high_f32(vvec), 0);
-    float32x2_t dotMaxLo = (float32x2_t) { -B3_INFINITY, -B3_INFINITY };
-    float32x2_t dotMaxHi = (float32x2_t) { -B3_INFINITY, -B3_INFINITY };
-    uint32x2_t indexLo = (uint32x2_t) {0, 1};
-    uint32x2_t indexHi = (uint32x2_t) {2, 3};
-    uint32x2_t iLo = (uint32x2_t) {-1, -1};
-    uint32x2_t iHi = (uint32x2_t) {-1, -1};
-    const uint32x2_t four = (uint32x2_t) {4,4};
+	if (_get_cpu_capabilities() & 0x2000)
+		b3_mindot_large = _mindot_large_v1;
+	else
+		b3_mindot_large = _mindot_large_v0;
 
-    for( ; i+8 <= count; i+= 8 )
-    {
-        float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v3 = vld1q_f32_aligned_postincrement( vv );
-
-        float32x2_t xy0 = vmul_f32( vget_low_f32(v0), vLo);
-        float32x2_t xy1 = vmul_f32( vget_low_f32(v1), vLo);
-        float32x2_t xy2 = vmul_f32( vget_low_f32(v2), vLo);
-        float32x2_t xy3 = vmul_f32( vget_low_f32(v3), vLo);
-
-        float32x2x2_t z0 = vtrn_f32( vget_high_f32(v0), vget_high_f32(v1));
-        float32x2x2_t z1 = vtrn_f32( vget_high_f32(v2), vget_high_f32(v3));
-        float32x2_t zLo = vmul_f32( z0.val[0], vHi);
-        float32x2_t zHi = vmul_f32( z1.val[0], vHi);
-
-        float32x2_t rLo = vpadd_f32( xy0, xy1);
-        float32x2_t rHi = vpadd_f32( xy2, xy3);
-        rLo = vadd_f32(rLo, zLo);
-        rHi = vadd_f32(rHi, zHi);
-
-        uint32x2_t maskLo = vcgt_f32( rLo, dotMaxLo );
-        uint32x2_t maskHi = vcgt_f32( rHi, dotMaxHi );
-        dotMaxLo = vbsl_f32( maskLo, rLo, dotMaxLo);
-        dotMaxHi = vbsl_f32( maskHi, rHi, dotMaxHi);
-        iLo = vbsl_u32(maskLo, indexLo, iLo);
-        iHi = vbsl_u32(maskHi, indexHi, iHi);
-        indexLo = vadd_u32(indexLo, four);
-        indexHi = vadd_u32(indexHi, four);
-
-        v0 = vld1q_f32_aligned_postincrement( vv );
-        v1 = vld1q_f32_aligned_postincrement( vv );
-        v2 = vld1q_f32_aligned_postincrement( vv );
-        v3 = vld1q_f32_aligned_postincrement( vv );
-
-        xy0 = vmul_f32( vget_low_f32(v0), vLo);
-        xy1 = vmul_f32( vget_low_f32(v1), vLo);
-        xy2 = vmul_f32( vget_low_f32(v2), vLo);
-        xy3 = vmul_f32( vget_low_f32(v3), vLo);
-
-        z0 = vtrn_f32( vget_high_f32(v0), vget_high_f32(v1));
-        z1 = vtrn_f32( vget_high_f32(v2), vget_high_f32(v3));
-        zLo = vmul_f32( z0.val[0], vHi);
-        zHi = vmul_f32( z1.val[0], vHi);
-
-        rLo = vpadd_f32( xy0, xy1);
-        rHi = vpadd_f32( xy2, xy3);
-        rLo = vadd_f32(rLo, zLo);
-        rHi = vadd_f32(rHi, zHi);
-
-        maskLo = vcgt_f32( rLo, dotMaxLo );
-        maskHi = vcgt_f32( rHi, dotMaxHi );
-        dotMaxLo = vbsl_f32( maskLo, rLo, dotMaxLo);
-        dotMaxHi = vbsl_f32( maskHi, rHi, dotMaxHi);
-        iLo = vbsl_u32(maskLo, indexLo, iLo);
-        iHi = vbsl_u32(maskHi, indexHi, iHi);
-        indexLo = vadd_u32(indexLo, four);
-        indexHi = vadd_u32(indexHi, four);
-    }
-
-    for( ; i+4 <= count; i+= 4 )
-    {
-        float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v3 = vld1q_f32_aligned_postincrement( vv );
-
-        float32x2_t xy0 = vmul_f32( vget_low_f32(v0), vLo);
-        float32x2_t xy1 = vmul_f32( vget_low_f32(v1), vLo);
-        float32x2_t xy2 = vmul_f32( vget_low_f32(v2), vLo);
-        float32x2_t xy3 = vmul_f32( vget_low_f32(v3), vLo);
-
-        float32x2x2_t z0 = vtrn_f32( vget_high_f32(v0), vget_high_f32(v1));
-        float32x2x2_t z1 = vtrn_f32( vget_high_f32(v2), vget_high_f32(v3));
-        float32x2_t zLo = vmul_f32( z0.val[0], vHi);
-        float32x2_t zHi = vmul_f32( z1.val[0], vHi);
-
-        float32x2_t rLo = vpadd_f32( xy0, xy1);
-        float32x2_t rHi = vpadd_f32( xy2, xy3);
-        rLo = vadd_f32(rLo, zLo);
-        rHi = vadd_f32(rHi, zHi);
-
-        uint32x2_t maskLo = vcgt_f32( rLo, dotMaxLo );
-        uint32x2_t maskHi = vcgt_f32( rHi, dotMaxHi );
-        dotMaxLo = vbsl_f32( maskLo, rLo, dotMaxLo);
-        dotMaxHi = vbsl_f32( maskHi, rHi, dotMaxHi);
-        iLo = vbsl_u32(maskLo, indexLo, iLo);
-        iHi = vbsl_u32(maskHi, indexHi, iHi);
-        indexLo = vadd_u32(indexLo, four);
-        indexHi = vadd_u32(indexHi, four);
-    }
-
-    switch( count & 3 )
-    {
-        case 3:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-
-            float32x2_t xy0 = vmul_f32( vget_low_f32(v0), vLo);
-            float32x2_t xy1 = vmul_f32( vget_low_f32(v1), vLo);
-            float32x2_t xy2 = vmul_f32( vget_low_f32(v2), vLo);
-
-            float32x2x2_t z0 = vtrn_f32( vget_high_f32(v0), vget_high_f32(v1));
-            float32x2_t zLo = vmul_f32( z0.val[0], vHi);
-            float32x2_t zHi = vmul_f32( vdup_lane_f32(vget_high_f32(v2), 0), vHi);
-
-            float32x2_t rLo = vpadd_f32( xy0, xy1);
-            float32x2_t rHi = vpadd_f32( xy2, xy2);
-            rLo = vadd_f32(rLo, zLo);
-            rHi = vadd_f32(rHi, zHi);
-
-            uint32x2_t maskLo = vcgt_f32( rLo, dotMaxLo );
-            uint32x2_t maskHi = vcgt_f32( rHi, dotMaxHi );
-            dotMaxLo = vbsl_f32( maskLo, rLo, dotMaxLo);
-            dotMaxHi = vbsl_f32( maskHi, rHi, dotMaxHi);
-            iLo = vbsl_u32(maskLo, indexLo, iLo);
-            iHi = vbsl_u32(maskHi, indexHi, iHi);
-        }
-            break;
-        case 2:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-
-            float32x2_t xy0 = vmul_f32( vget_low_f32(v0), vLo);
-            float32x2_t xy1 = vmul_f32( vget_low_f32(v1), vLo);
-
-            float32x2x2_t z0 = vtrn_f32( vget_high_f32(v0), vget_high_f32(v1));
-            float32x2_t zLo = vmul_f32( z0.val[0], vHi);
-
-            float32x2_t rLo = vpadd_f32( xy0, xy1);
-            rLo = vadd_f32(rLo, zLo);
-
-            uint32x2_t maskLo = vcgt_f32( rLo, dotMaxLo );
-            dotMaxLo = vbsl_f32( maskLo, rLo, dotMaxLo);
-            iLo = vbsl_u32(maskLo, indexLo, iLo);
-        }
-            break;
-        case 1:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            float32x2_t xy0 = vmul_f32( vget_low_f32(v0), vLo);
-            float32x2_t z0 = vdup_lane_f32(vget_high_f32(v0), 0);
-            float32x2_t zLo = vmul_f32( z0, vHi);
-            float32x2_t rLo = vpadd_f32( xy0, xy0);
-            rLo = vadd_f32(rLo, zLo);
-            uint32x2_t maskLo = vcgt_f32( rLo, dotMaxLo );
-            dotMaxLo = vbsl_f32( maskLo, rLo, dotMaxLo);
-            iLo = vbsl_u32(maskLo, indexLo, iLo);
-        }
-            break;
-
-        default:
-            break;
-    }
-
-    // select best answer between hi and lo results
-    uint32x2_t mask = vcgt_f32( dotMaxHi, dotMaxLo );
-    dotMaxLo = vbsl_f32(mask, dotMaxHi, dotMaxLo);
-    iLo = vbsl_u32(mask, iHi, iLo);
-
-    // select best answer between even and odd results
-    dotMaxHi = vdup_lane_f32(dotMaxLo, 1);
-    iHi = vdup_lane_u32(iLo, 1);
-    mask = vcgt_f32( dotMaxHi, dotMaxLo );
-    dotMaxLo = vbsl_f32(mask, dotMaxHi, dotMaxLo);
-    iLo = vbsl_u32(mask, iHi, iLo);
-
-    *dotResult = vget_lane_f32( dotMaxLo, 0);
-    return vget_lane_u32(iLo, 0);
+	return b3_mindot_large(vv, vec, count, dotResult);
 }
 
+#define vld1q_f32_aligned_postincrement(_ptr) ({ float32x4_t _r; asm( "vld1.f32  {%0}, [%1, :128]!\n" : "=w" (_r), "+r" (_ptr) ); /*return*/ _r; })
 
-long b3_maxdot_large_v1( const float *vv, const float *vec, unsigned long count, float *dotResult )
+long b3_maxdot_large_v0(const float *vv, const float *vec, unsigned long count, float *dotResult)
 {
-    float32x4_t vvec = vld1q_f32_aligned_postincrement( vec );
-    float32x4_t vLo = vcombine_f32(vget_low_f32(vvec), vget_low_f32(vvec));
-    float32x4_t vHi = vdupq_lane_f32(vget_high_f32(vvec), 0);
-    const uint32x4_t four = (uint32x4_t){ 4, 4, 4, 4 };
-    uint32x4_t local_index = (uint32x4_t) {0, 1, 2, 3};
-    uint32x4_t index = (uint32x4_t) { -1, -1, -1, -1 };
-    float32x4_t maxDot = (float32x4_t) { -B3_INFINITY, -B3_INFINITY, -B3_INFINITY, -B3_INFINITY };
+	unsigned long i = 0;
+	float32x4_t vvec = vld1q_f32_aligned_postincrement(vec);
+	float32x2_t vLo = vget_low_f32(vvec);
+	float32x2_t vHi = vdup_lane_f32(vget_high_f32(vvec), 0);
+	float32x2_t dotMaxLo = (float32x2_t){-B3_INFINITY, -B3_INFINITY};
+	float32x2_t dotMaxHi = (float32x2_t){-B3_INFINITY, -B3_INFINITY};
+	uint32x2_t indexLo = (uint32x2_t){0, 1};
+	uint32x2_t indexHi = (uint32x2_t){2, 3};
+	uint32x2_t iLo = (uint32x2_t){-1, -1};
+	uint32x2_t iHi = (uint32x2_t){-1, -1};
+	const uint32x2_t four = (uint32x2_t){4, 4};
 
-    unsigned long i = 0;
-    for( ; i + 8 <= count; i += 8 )
-    {
-        float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v3 = vld1q_f32_aligned_postincrement( vv );
+	for (; i + 8 <= count; i += 8)
+	{
+		float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v3 = vld1q_f32_aligned_postincrement(vv);
 
-        // the next two lines should resolve to a single vswp d, d
-        float32x4_t xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v1));
-        float32x4_t xy1 = vcombine_f32( vget_low_f32(v2), vget_low_f32(v3));
-        // the next two lines should resolve to a single vswp d, d
-        float32x4_t z0 = vcombine_f32( vget_high_f32(v0), vget_high_f32(v1));
-        float32x4_t z1 = vcombine_f32( vget_high_f32(v2), vget_high_f32(v3));
+		float32x2_t xy0 = vmul_f32(vget_low_f32(v0), vLo);
+		float32x2_t xy1 = vmul_f32(vget_low_f32(v1), vLo);
+		float32x2_t xy2 = vmul_f32(vget_low_f32(v2), vLo);
+		float32x2_t xy3 = vmul_f32(vget_low_f32(v3), vLo);
 
-        xy0 = vmulq_f32(xy0, vLo);
-        xy1 = vmulq_f32(xy1, vLo);
+		float32x2x2_t z0 = vtrn_f32(vget_high_f32(v0), vget_high_f32(v1));
+		float32x2x2_t z1 = vtrn_f32(vget_high_f32(v2), vget_high_f32(v3));
+		float32x2_t zLo = vmul_f32(z0.val[0], vHi);
+		float32x2_t zHi = vmul_f32(z1.val[0], vHi);
 
-        float32x4x2_t zb = vuzpq_f32( z0, z1);
-        float32x4_t z = vmulq_f32( zb.val[0], vHi);
-        float32x4x2_t xy = vuzpq_f32( xy0, xy1);
-        float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
-        x = vaddq_f32(x, z);
+		float32x2_t rLo = vpadd_f32(xy0, xy1);
+		float32x2_t rHi = vpadd_f32(xy2, xy3);
+		rLo = vadd_f32(rLo, zLo);
+		rHi = vadd_f32(rHi, zHi);
 
-        uint32x4_t mask = vcgtq_f32(x, maxDot);
-        maxDot = vbslq_f32( mask, x, maxDot);
-        index = vbslq_u32(mask, local_index, index);
-        local_index = vaddq_u32(local_index, four);
+		uint32x2_t maskLo = vcgt_f32(rLo, dotMaxLo);
+		uint32x2_t maskHi = vcgt_f32(rHi, dotMaxHi);
+		dotMaxLo = vbsl_f32(maskLo, rLo, dotMaxLo);
+		dotMaxHi = vbsl_f32(maskHi, rHi, dotMaxHi);
+		iLo = vbsl_u32(maskLo, indexLo, iLo);
+		iHi = vbsl_u32(maskHi, indexHi, iHi);
+		indexLo = vadd_u32(indexLo, four);
+		indexHi = vadd_u32(indexHi, four);
 
-        v0 = vld1q_f32_aligned_postincrement( vv );
-        v1 = vld1q_f32_aligned_postincrement( vv );
-        v2 = vld1q_f32_aligned_postincrement( vv );
-        v3 = vld1q_f32_aligned_postincrement( vv );
+		v0 = vld1q_f32_aligned_postincrement(vv);
+		v1 = vld1q_f32_aligned_postincrement(vv);
+		v2 = vld1q_f32_aligned_postincrement(vv);
+		v3 = vld1q_f32_aligned_postincrement(vv);
 
-        // the next two lines should resolve to a single vswp d, d
-        xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v1));
-        xy1 = vcombine_f32( vget_low_f32(v2), vget_low_f32(v3));
-        // the next two lines should resolve to a single vswp d, d
-        z0 = vcombine_f32( vget_high_f32(v0), vget_high_f32(v1));
-        z1 = vcombine_f32( vget_high_f32(v2), vget_high_f32(v3));
+		xy0 = vmul_f32(vget_low_f32(v0), vLo);
+		xy1 = vmul_f32(vget_low_f32(v1), vLo);
+		xy2 = vmul_f32(vget_low_f32(v2), vLo);
+		xy3 = vmul_f32(vget_low_f32(v3), vLo);
 
-        xy0 = vmulq_f32(xy0, vLo);
-        xy1 = vmulq_f32(xy1, vLo);
+		z0 = vtrn_f32(vget_high_f32(v0), vget_high_f32(v1));
+		z1 = vtrn_f32(vget_high_f32(v2), vget_high_f32(v3));
+		zLo = vmul_f32(z0.val[0], vHi);
+		zHi = vmul_f32(z1.val[0], vHi);
 
-        zb = vuzpq_f32( z0, z1);
-        z = vmulq_f32( zb.val[0], vHi);
-        xy = vuzpq_f32( xy0, xy1);
-        x = vaddq_f32(xy.val[0], xy.val[1]);
-        x = vaddq_f32(x, z);
+		rLo = vpadd_f32(xy0, xy1);
+		rHi = vpadd_f32(xy2, xy3);
+		rLo = vadd_f32(rLo, zLo);
+		rHi = vadd_f32(rHi, zHi);
 
-        mask = vcgtq_f32(x, maxDot);
-        maxDot = vbslq_f32( mask, x, maxDot);
-        index = vbslq_u32(mask, local_index, index);
-        local_index = vaddq_u32(local_index, four);
-    }
+		maskLo = vcgt_f32(rLo, dotMaxLo);
+		maskHi = vcgt_f32(rHi, dotMaxHi);
+		dotMaxLo = vbsl_f32(maskLo, rLo, dotMaxLo);
+		dotMaxHi = vbsl_f32(maskHi, rHi, dotMaxHi);
+		iLo = vbsl_u32(maskLo, indexLo, iLo);
+		iHi = vbsl_u32(maskHi, indexHi, iHi);
+		indexLo = vadd_u32(indexLo, four);
+		indexHi = vadd_u32(indexHi, four);
+	}
 
-    for( ; i + 4 <= count; i += 4 )
-    {
-        float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v3 = vld1q_f32_aligned_postincrement( vv );
+	for (; i + 4 <= count; i += 4)
+	{
+		float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v3 = vld1q_f32_aligned_postincrement(vv);
 
-        // the next two lines should resolve to a single vswp d, d
-        float32x4_t xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v1));
-        float32x4_t xy1 = vcombine_f32( vget_low_f32(v2), vget_low_f32(v3));
-        // the next two lines should resolve to a single vswp d, d
-        float32x4_t z0 = vcombine_f32( vget_high_f32(v0), vget_high_f32(v1));
-        float32x4_t z1 = vcombine_f32( vget_high_f32(v2), vget_high_f32(v3));
+		float32x2_t xy0 = vmul_f32(vget_low_f32(v0), vLo);
+		float32x2_t xy1 = vmul_f32(vget_low_f32(v1), vLo);
+		float32x2_t xy2 = vmul_f32(vget_low_f32(v2), vLo);
+		float32x2_t xy3 = vmul_f32(vget_low_f32(v3), vLo);
 
-        xy0 = vmulq_f32(xy0, vLo);
-        xy1 = vmulq_f32(xy1, vLo);
+		float32x2x2_t z0 = vtrn_f32(vget_high_f32(v0), vget_high_f32(v1));
+		float32x2x2_t z1 = vtrn_f32(vget_high_f32(v2), vget_high_f32(v3));
+		float32x2_t zLo = vmul_f32(z0.val[0], vHi);
+		float32x2_t zHi = vmul_f32(z1.val[0], vHi);
 
-        float32x4x2_t zb = vuzpq_f32( z0, z1);
-        float32x4_t z = vmulq_f32( zb.val[0], vHi);
-        float32x4x2_t xy = vuzpq_f32( xy0, xy1);
-        float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
-        x = vaddq_f32(x, z);
+		float32x2_t rLo = vpadd_f32(xy0, xy1);
+		float32x2_t rHi = vpadd_f32(xy2, xy3);
+		rLo = vadd_f32(rLo, zLo);
+		rHi = vadd_f32(rHi, zHi);
 
-        uint32x4_t mask = vcgtq_f32(x, maxDot);
-        maxDot = vbslq_f32( mask, x, maxDot);
-        index = vbslq_u32(mask, local_index, index);
-        local_index = vaddq_u32(local_index, four);
-    }
+		uint32x2_t maskLo = vcgt_f32(rLo, dotMaxLo);
+		uint32x2_t maskHi = vcgt_f32(rHi, dotMaxHi);
+		dotMaxLo = vbsl_f32(maskLo, rLo, dotMaxLo);
+		dotMaxHi = vbsl_f32(maskHi, rHi, dotMaxHi);
+		iLo = vbsl_u32(maskLo, indexLo, iLo);
+		iHi = vbsl_u32(maskHi, indexHi, iHi);
+		indexLo = vadd_u32(indexLo, four);
+		indexHi = vadd_u32(indexHi, four);
+	}
 
-    switch (count & 3) {
-        case 3:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
+	switch (count & 3)
+	{
+		case 3:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
 
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v1));
-            float32x4_t xy1 = vcombine_f32( vget_low_f32(v2), vget_low_f32(v2));
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t z0 = vcombine_f32( vget_high_f32(v0), vget_high_f32(v1));
-            float32x4_t z1 = vcombine_f32( vget_high_f32(v2), vget_high_f32(v2));
+			float32x2_t xy0 = vmul_f32(vget_low_f32(v0), vLo);
+			float32x2_t xy1 = vmul_f32(vget_low_f32(v1), vLo);
+			float32x2_t xy2 = vmul_f32(vget_low_f32(v2), vLo);
 
-            xy0 = vmulq_f32(xy0, vLo);
-            xy1 = vmulq_f32(xy1, vLo);
+			float32x2x2_t z0 = vtrn_f32(vget_high_f32(v0), vget_high_f32(v1));
+			float32x2_t zLo = vmul_f32(z0.val[0], vHi);
+			float32x2_t zHi = vmul_f32(vdup_lane_f32(vget_high_f32(v2), 0), vHi);
 
-            float32x4x2_t zb = vuzpq_f32( z0, z1);
-            float32x4_t z = vmulq_f32( zb.val[0], vHi);
-            float32x4x2_t xy = vuzpq_f32( xy0, xy1);
-            float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
-            x = vaddq_f32(x, z);
+			float32x2_t rLo = vpadd_f32(xy0, xy1);
+			float32x2_t rHi = vpadd_f32(xy2, xy2);
+			rLo = vadd_f32(rLo, zLo);
+			rHi = vadd_f32(rHi, zHi);
 
-            uint32x4_t mask = vcgtq_f32(x, maxDot);
-            maxDot = vbslq_f32( mask, x, maxDot);
-            index = vbslq_u32(mask, local_index, index);
-            local_index = vaddq_u32(local_index, four);
-        }
-            break;
+			uint32x2_t maskLo = vcgt_f32(rLo, dotMaxLo);
+			uint32x2_t maskHi = vcgt_f32(rHi, dotMaxHi);
+			dotMaxLo = vbsl_f32(maskLo, rLo, dotMaxLo);
+			dotMaxHi = vbsl_f32(maskHi, rHi, dotMaxHi);
+			iLo = vbsl_u32(maskLo, indexLo, iLo);
+			iHi = vbsl_u32(maskHi, indexHi, iHi);
+		}
+		break;
+		case 2:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
 
-        case 2:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
+			float32x2_t xy0 = vmul_f32(vget_low_f32(v0), vLo);
+			float32x2_t xy1 = vmul_f32(vget_low_f32(v1), vLo);
 
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v1));
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t z0 = vcombine_f32( vget_high_f32(v0), vget_high_f32(v1));
+			float32x2x2_t z0 = vtrn_f32(vget_high_f32(v0), vget_high_f32(v1));
+			float32x2_t zLo = vmul_f32(z0.val[0], vHi);
 
-            xy0 = vmulq_f32(xy0, vLo);
+			float32x2_t rLo = vpadd_f32(xy0, xy1);
+			rLo = vadd_f32(rLo, zLo);
 
-            float32x4x2_t zb = vuzpq_f32( z0, z0);
-            float32x4_t z = vmulq_f32( zb.val[0], vHi);
-            float32x4x2_t xy = vuzpq_f32( xy0, xy0);
-            float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
-            x = vaddq_f32(x, z);
+			uint32x2_t maskLo = vcgt_f32(rLo, dotMaxLo);
+			dotMaxLo = vbsl_f32(maskLo, rLo, dotMaxLo);
+			iLo = vbsl_u32(maskLo, indexLo, iLo);
+		}
+		break;
+		case 1:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+			float32x2_t xy0 = vmul_f32(vget_low_f32(v0), vLo);
+			float32x2_t z0 = vdup_lane_f32(vget_high_f32(v0), 0);
+			float32x2_t zLo = vmul_f32(z0, vHi);
+			float32x2_t rLo = vpadd_f32(xy0, xy0);
+			rLo = vadd_f32(rLo, zLo);
+			uint32x2_t maskLo = vcgt_f32(rLo, dotMaxLo);
+			dotMaxLo = vbsl_f32(maskLo, rLo, dotMaxLo);
+			iLo = vbsl_u32(maskLo, indexLo, iLo);
+		}
+		break;
 
-            uint32x4_t mask = vcgtq_f32(x, maxDot);
-            maxDot = vbslq_f32( mask, x, maxDot);
-            index = vbslq_u32(mask, local_index, index);
-            local_index = vaddq_u32(local_index, four);
-        }
-            break;
+		default:
+			break;
+	}
 
-        case 1:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
+	// select best answer between hi and lo results
+	uint32x2_t mask = vcgt_f32(dotMaxHi, dotMaxLo);
+	dotMaxLo = vbsl_f32(mask, dotMaxHi, dotMaxLo);
+	iLo = vbsl_u32(mask, iHi, iLo);
 
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v0));
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t z = vdupq_lane_f32(vget_high_f32(v0), 0);
-
-            xy0 = vmulq_f32(xy0, vLo);
-
-            z = vmulq_f32( z, vHi);
-            float32x4x2_t xy = vuzpq_f32( xy0, xy0);
-            float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
-            x = vaddq_f32(x, z);
-
-            uint32x4_t mask = vcgtq_f32(x, maxDot);
-            maxDot = vbslq_f32( mask, x, maxDot);
-            index = vbslq_u32(mask, local_index, index);
-            local_index = vaddq_u32(local_index, four);
-        }
-            break;
-
-        default:
-            break;
-    }
-
-
-    // select best answer between hi and lo results
-    uint32x2_t mask = vcgt_f32( vget_high_f32(maxDot), vget_low_f32(maxDot));
-    float32x2_t maxDot2 = vbsl_f32(mask, vget_high_f32(maxDot), vget_low_f32(maxDot));
-    uint32x2_t index2 = vbsl_u32(mask, vget_high_u32(index), vget_low_u32(index));
-
-    // select best answer between even and odd results
-    float32x2_t maxDotO = vdup_lane_f32(maxDot2, 1);
-    uint32x2_t indexHi = vdup_lane_u32(index2, 1);
-    mask = vcgt_f32( maxDotO, maxDot2 );
-    maxDot2 = vbsl_f32(mask, maxDotO, maxDot2);
-    index2 = vbsl_u32(mask, indexHi, index2);
-
-    *dotResult = vget_lane_f32( maxDot2, 0);
-    return vget_lane_u32(index2, 0);
+	// select best answer between even and odd results
+	dotMaxHi = vdup_lane_f32(dotMaxLo, 1);
+	iHi = vdup_lane_u32(iLo, 1);
+	mask = vcgt_f32(dotMaxHi, dotMaxLo);
+	dotMaxLo = vbsl_f32(mask, dotMaxHi, dotMaxLo);
+	iLo = vbsl_u32(mask, iHi, iLo);
 
+	*dotResult = vget_lane_f32(dotMaxLo, 0);
+	return vget_lane_u32(iLo, 0);
 }
 
-long b3_mindot_large_v0( const float *vv, const float *vec, unsigned long count, float *dotResult )
+long b3_maxdot_large_v1(const float *vv, const float *vec, unsigned long count, float *dotResult)
 {
-    unsigned long i = 0;
-    float32x4_t vvec = vld1q_f32_aligned_postincrement( vec );
-    float32x2_t vLo = vget_low_f32(vvec);
-    float32x2_t vHi = vdup_lane_f32(vget_high_f32(vvec), 0);
-    float32x2_t dotMinLo = (float32x2_t) { B3_INFINITY, B3_INFINITY };
-    float32x2_t dotMinHi = (float32x2_t) { B3_INFINITY, B3_INFINITY };
-    uint32x2_t indexLo = (uint32x2_t) {0, 1};
-    uint32x2_t indexHi = (uint32x2_t) {2, 3};
-    uint32x2_t iLo = (uint32x2_t) {-1, -1};
-    uint32x2_t iHi = (uint32x2_t) {-1, -1};
-    const uint32x2_t four = (uint32x2_t) {4,4};
+	float32x4_t vvec = vld1q_f32_aligned_postincrement(vec);
+	float32x4_t vLo = vcombine_f32(vget_low_f32(vvec), vget_low_f32(vvec));
+	float32x4_t vHi = vdupq_lane_f32(vget_high_f32(vvec), 0);
+	const uint32x4_t four = (uint32x4_t){4, 4, 4, 4};
+	uint32x4_t local_index = (uint32x4_t){0, 1, 2, 3};
+	uint32x4_t index = (uint32x4_t){-1, -1, -1, -1};
+	float32x4_t maxDot = (float32x4_t){-B3_INFINITY, -B3_INFINITY, -B3_INFINITY, -B3_INFINITY};
 
-    for( ; i+8 <= count; i+= 8 )
-    {
-        float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v3 = vld1q_f32_aligned_postincrement( vv );
+	unsigned long i = 0;
+	for (; i + 8 <= count; i += 8)
+	{
+		float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v3 = vld1q_f32_aligned_postincrement(vv);
 
-        float32x2_t xy0 = vmul_f32( vget_low_f32(v0), vLo);
-        float32x2_t xy1 = vmul_f32( vget_low_f32(v1), vLo);
-        float32x2_t xy2 = vmul_f32( vget_low_f32(v2), vLo);
-        float32x2_t xy3 = vmul_f32( vget_low_f32(v3), vLo);
+		// the next two lines should resolve to a single vswp d, d
+		float32x4_t xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v1));
+		float32x4_t xy1 = vcombine_f32(vget_low_f32(v2), vget_low_f32(v3));
+		// the next two lines should resolve to a single vswp d, d
+		float32x4_t z0 = vcombine_f32(vget_high_f32(v0), vget_high_f32(v1));
+		float32x4_t z1 = vcombine_f32(vget_high_f32(v2), vget_high_f32(v3));
 
-        float32x2x2_t z0 = vtrn_f32( vget_high_f32(v0), vget_high_f32(v1));
-        float32x2x2_t z1 = vtrn_f32( vget_high_f32(v2), vget_high_f32(v3));
-        float32x2_t zLo = vmul_f32( z0.val[0], vHi);
-        float32x2_t zHi = vmul_f32( z1.val[0], vHi);
+		xy0 = vmulq_f32(xy0, vLo);
+		xy1 = vmulq_f32(xy1, vLo);
 
-        float32x2_t rLo = vpadd_f32( xy0, xy1);
-        float32x2_t rHi = vpadd_f32( xy2, xy3);
-        rLo = vadd_f32(rLo, zLo);
-        rHi = vadd_f32(rHi, zHi);
+		float32x4x2_t zb = vuzpq_f32(z0, z1);
+		float32x4_t z = vmulq_f32(zb.val[0], vHi);
+		float32x4x2_t xy = vuzpq_f32(xy0, xy1);
+		float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
+		x = vaddq_f32(x, z);
 
-        uint32x2_t maskLo = vclt_f32( rLo, dotMinLo );
-        uint32x2_t maskHi = vclt_f32( rHi, dotMinHi );
-        dotMinLo = vbsl_f32( maskLo, rLo, dotMinLo);
-        dotMinHi = vbsl_f32( maskHi, rHi, dotMinHi);
-        iLo = vbsl_u32(maskLo, indexLo, iLo);
-        iHi = vbsl_u32(maskHi, indexHi, iHi);
-        indexLo = vadd_u32(indexLo, four);
-        indexHi = vadd_u32(indexHi, four);
+		uint32x4_t mask = vcgtq_f32(x, maxDot);
+		maxDot = vbslq_f32(mask, x, maxDot);
+		index = vbslq_u32(mask, local_index, index);
+		local_index = vaddq_u32(local_index, four);
 
-        v0 = vld1q_f32_aligned_postincrement( vv );
-        v1 = vld1q_f32_aligned_postincrement( vv );
-        v2 = vld1q_f32_aligned_postincrement( vv );
-        v3 = vld1q_f32_aligned_postincrement( vv );
+		v0 = vld1q_f32_aligned_postincrement(vv);
+		v1 = vld1q_f32_aligned_postincrement(vv);
+		v2 = vld1q_f32_aligned_postincrement(vv);
+		v3 = vld1q_f32_aligned_postincrement(vv);
 
-        xy0 = vmul_f32( vget_low_f32(v0), vLo);
-        xy1 = vmul_f32( vget_low_f32(v1), vLo);
-        xy2 = vmul_f32( vget_low_f32(v2), vLo);
-        xy3 = vmul_f32( vget_low_f32(v3), vLo);
+		// the next two lines should resolve to a single vswp d, d
+		xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v1));
+		xy1 = vcombine_f32(vget_low_f32(v2), vget_low_f32(v3));
+		// the next two lines should resolve to a single vswp d, d
+		z0 = vcombine_f32(vget_high_f32(v0), vget_high_f32(v1));
+		z1 = vcombine_f32(vget_high_f32(v2), vget_high_f32(v3));
 
-        z0 = vtrn_f32( vget_high_f32(v0), vget_high_f32(v1));
-        z1 = vtrn_f32( vget_high_f32(v2), vget_high_f32(v3));
-        zLo = vmul_f32( z0.val[0], vHi);
-        zHi = vmul_f32( z1.val[0], vHi);
+		xy0 = vmulq_f32(xy0, vLo);
+		xy1 = vmulq_f32(xy1, vLo);
 
-        rLo = vpadd_f32( xy0, xy1);
-        rHi = vpadd_f32( xy2, xy3);
-        rLo = vadd_f32(rLo, zLo);
-        rHi = vadd_f32(rHi, zHi);
+		zb = vuzpq_f32(z0, z1);
+		z = vmulq_f32(zb.val[0], vHi);
+		xy = vuzpq_f32(xy0, xy1);
+		x = vaddq_f32(xy.val[0], xy.val[1]);
+		x = vaddq_f32(x, z);
 
-        maskLo = vclt_f32( rLo, dotMinLo );
-        maskHi = vclt_f32( rHi, dotMinHi );
-        dotMinLo = vbsl_f32( maskLo, rLo, dotMinLo);
-        dotMinHi = vbsl_f32( maskHi, rHi, dotMinHi);
-        iLo = vbsl_u32(maskLo, indexLo, iLo);
-        iHi = vbsl_u32(maskHi, indexHi, iHi);
-        indexLo = vadd_u32(indexLo, four);
-        indexHi = vadd_u32(indexHi, four);
-    }
+		mask = vcgtq_f32(x, maxDot);
+		maxDot = vbslq_f32(mask, x, maxDot);
+		index = vbslq_u32(mask, local_index, index);
+		local_index = vaddq_u32(local_index, four);
+	}
 
-    for( ; i+4 <= count; i+= 4 )
-    {
-        float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v3 = vld1q_f32_aligned_postincrement( vv );
+	for (; i + 4 <= count; i += 4)
+	{
+		float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v3 = vld1q_f32_aligned_postincrement(vv);
 
-        float32x2_t xy0 = vmul_f32( vget_low_f32(v0), vLo);
-        float32x2_t xy1 = vmul_f32( vget_low_f32(v1), vLo);
-        float32x2_t xy2 = vmul_f32( vget_low_f32(v2), vLo);
-        float32x2_t xy3 = vmul_f32( vget_low_f32(v3), vLo);
+		// the next two lines should resolve to a single vswp d, d
+		float32x4_t xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v1));
+		float32x4_t xy1 = vcombine_f32(vget_low_f32(v2), vget_low_f32(v3));
+		// the next two lines should resolve to a single vswp d, d
+		float32x4_t z0 = vcombine_f32(vget_high_f32(v0), vget_high_f32(v1));
+		float32x4_t z1 = vcombine_f32(vget_high_f32(v2), vget_high_f32(v3));
 
-        float32x2x2_t z0 = vtrn_f32( vget_high_f32(v0), vget_high_f32(v1));
-        float32x2x2_t z1 = vtrn_f32( vget_high_f32(v2), vget_high_f32(v3));
-        float32x2_t zLo = vmul_f32( z0.val[0], vHi);
-        float32x2_t zHi = vmul_f32( z1.val[0], vHi);
+		xy0 = vmulq_f32(xy0, vLo);
+		xy1 = vmulq_f32(xy1, vLo);
 
-        float32x2_t rLo = vpadd_f32( xy0, xy1);
-        float32x2_t rHi = vpadd_f32( xy2, xy3);
-        rLo = vadd_f32(rLo, zLo);
-        rHi = vadd_f32(rHi, zHi);
+		float32x4x2_t zb = vuzpq_f32(z0, z1);
+		float32x4_t z = vmulq_f32(zb.val[0], vHi);
+		float32x4x2_t xy = vuzpq_f32(xy0, xy1);
+		float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
+		x = vaddq_f32(x, z);
 
-        uint32x2_t maskLo = vclt_f32( rLo, dotMinLo );
-        uint32x2_t maskHi = vclt_f32( rHi, dotMinHi );
-        dotMinLo = vbsl_f32( maskLo, rLo, dotMinLo);
-        dotMinHi = vbsl_f32( maskHi, rHi, dotMinHi);
-        iLo = vbsl_u32(maskLo, indexLo, iLo);
-        iHi = vbsl_u32(maskHi, indexHi, iHi);
-        indexLo = vadd_u32(indexLo, four);
-        indexHi = vadd_u32(indexHi, four);
-    }
-    switch( count & 3 )
-    {
-        case 3:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
+		uint32x4_t mask = vcgtq_f32(x, maxDot);
+		maxDot = vbslq_f32(mask, x, maxDot);
+		index = vbslq_u32(mask, local_index, index);
+		local_index = vaddq_u32(local_index, four);
+	}
 
-            float32x2_t xy0 = vmul_f32( vget_low_f32(v0), vLo);
-            float32x2_t xy1 = vmul_f32( vget_low_f32(v1), vLo);
-            float32x2_t xy2 = vmul_f32( vget_low_f32(v2), vLo);
+	switch (count & 3)
+	{
+		case 3:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
 
-            float32x2x2_t z0 = vtrn_f32( vget_high_f32(v0), vget_high_f32(v1));
-            float32x2_t zLo = vmul_f32( z0.val[0], vHi);
-            float32x2_t zHi = vmul_f32( vdup_lane_f32(vget_high_f32(v2), 0), vHi);
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v1));
+			float32x4_t xy1 = vcombine_f32(vget_low_f32(v2), vget_low_f32(v2));
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t z0 = vcombine_f32(vget_high_f32(v0), vget_high_f32(v1));
+			float32x4_t z1 = vcombine_f32(vget_high_f32(v2), vget_high_f32(v2));
 
-            float32x2_t rLo = vpadd_f32( xy0, xy1);
-            float32x2_t rHi = vpadd_f32( xy2, xy2);
-            rLo = vadd_f32(rLo, zLo);
-            rHi = vadd_f32(rHi, zHi);
+			xy0 = vmulq_f32(xy0, vLo);
+			xy1 = vmulq_f32(xy1, vLo);
 
-            uint32x2_t maskLo = vclt_f32( rLo, dotMinLo );
-            uint32x2_t maskHi = vclt_f32( rHi, dotMinHi );
-            dotMinLo = vbsl_f32( maskLo, rLo, dotMinLo);
-            dotMinHi = vbsl_f32( maskHi, rHi, dotMinHi);
-            iLo = vbsl_u32(maskLo, indexLo, iLo);
-            iHi = vbsl_u32(maskHi, indexHi, iHi);
-        }
-            break;
-        case 2:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
+			float32x4x2_t zb = vuzpq_f32(z0, z1);
+			float32x4_t z = vmulq_f32(zb.val[0], vHi);
+			float32x4x2_t xy = vuzpq_f32(xy0, xy1);
+			float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
+			x = vaddq_f32(x, z);
 
-            float32x2_t xy0 = vmul_f32( vget_low_f32(v0), vLo);
-            float32x2_t xy1 = vmul_f32( vget_low_f32(v1), vLo);
+			uint32x4_t mask = vcgtq_f32(x, maxDot);
+			maxDot = vbslq_f32(mask, x, maxDot);
+			index = vbslq_u32(mask, local_index, index);
+			local_index = vaddq_u32(local_index, four);
+		}
+		break;
 
-            float32x2x2_t z0 = vtrn_f32( vget_high_f32(v0), vget_high_f32(v1));
-            float32x2_t zLo = vmul_f32( z0.val[0], vHi);
+		case 2:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
 
-            float32x2_t rLo = vpadd_f32( xy0, xy1);
-            rLo = vadd_f32(rLo, zLo);
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v1));
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t z0 = vcombine_f32(vget_high_f32(v0), vget_high_f32(v1));
 
-            uint32x2_t maskLo = vclt_f32( rLo, dotMinLo );
-            dotMinLo = vbsl_f32( maskLo, rLo, dotMinLo);
-            iLo = vbsl_u32(maskLo, indexLo, iLo);
-        }
-            break;
-        case 1:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            float32x2_t xy0 = vmul_f32( vget_low_f32(v0), vLo);
-            float32x2_t z0 = vdup_lane_f32(vget_high_f32(v0), 0);
-            float32x2_t zLo = vmul_f32( z0, vHi);
-            float32x2_t rLo = vpadd_f32( xy0, xy0);
-            rLo = vadd_f32(rLo, zLo);
-            uint32x2_t maskLo = vclt_f32( rLo, dotMinLo );
-            dotMinLo = vbsl_f32( maskLo, rLo, dotMinLo);
-            iLo = vbsl_u32(maskLo, indexLo, iLo);
-        }
-            break;
+			xy0 = vmulq_f32(xy0, vLo);
 
-        default:
-            break;
-    }
+			float32x4x2_t zb = vuzpq_f32(z0, z0);
+			float32x4_t z = vmulq_f32(zb.val[0], vHi);
+			float32x4x2_t xy = vuzpq_f32(xy0, xy0);
+			float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
+			x = vaddq_f32(x, z);
 
-    // select best answer between hi and lo results
-    uint32x2_t mask = vclt_f32( dotMinHi, dotMinLo );
-    dotMinLo = vbsl_f32(mask, dotMinHi, dotMinLo);
-    iLo = vbsl_u32(mask, iHi, iLo);
+			uint32x4_t mask = vcgtq_f32(x, maxDot);
+			maxDot = vbslq_f32(mask, x, maxDot);
+			index = vbslq_u32(mask, local_index, index);
+			local_index = vaddq_u32(local_index, four);
+		}
+		break;
 
-    // select best answer between even and odd results
-    dotMinHi = vdup_lane_f32(dotMinLo, 1);
-    iHi = vdup_lane_u32(iLo, 1);
-    mask = vclt_f32( dotMinHi, dotMinLo );
-    dotMinLo = vbsl_f32(mask, dotMinHi, dotMinLo);
-    iLo = vbsl_u32(mask, iHi, iLo);
+		case 1:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
 
-    *dotResult = vget_lane_f32( dotMinLo, 0);
-    return vget_lane_u32(iLo, 0);
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v0));
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t z = vdupq_lane_f32(vget_high_f32(v0), 0);
+
+			xy0 = vmulq_f32(xy0, vLo);
+
+			z = vmulq_f32(z, vHi);
+			float32x4x2_t xy = vuzpq_f32(xy0, xy0);
+			float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
+			x = vaddq_f32(x, z);
+
+			uint32x4_t mask = vcgtq_f32(x, maxDot);
+			maxDot = vbslq_f32(mask, x, maxDot);
+			index = vbslq_u32(mask, local_index, index);
+			local_index = vaddq_u32(local_index, four);
+		}
+		break;
+
+		default:
+			break;
+	}
+
+	// select best answer between hi and lo results
+	uint32x2_t mask = vcgt_f32(vget_high_f32(maxDot), vget_low_f32(maxDot));
+	float32x2_t maxDot2 = vbsl_f32(mask, vget_high_f32(maxDot), vget_low_f32(maxDot));
+	uint32x2_t index2 = vbsl_u32(mask, vget_high_u32(index), vget_low_u32(index));
+
+	// select best answer between even and odd results
+	float32x2_t maxDotO = vdup_lane_f32(maxDot2, 1);
+	uint32x2_t indexHi = vdup_lane_u32(index2, 1);
+	mask = vcgt_f32(maxDotO, maxDot2);
+	maxDot2 = vbsl_f32(mask, maxDotO, maxDot2);
+	index2 = vbsl_u32(mask, indexHi, index2);
+
+	*dotResult = vget_lane_f32(maxDot2, 0);
+	return vget_lane_u32(index2, 0);
 }
 
-long b3_mindot_large_v1( const float *vv, const float *vec, unsigned long count, float *dotResult )
+long b3_mindot_large_v0(const float *vv, const float *vec, unsigned long count, float *dotResult)
 {
-    float32x4_t vvec = vld1q_f32_aligned_postincrement( vec );
-    float32x4_t vLo = vcombine_f32(vget_low_f32(vvec), vget_low_f32(vvec));
-    float32x4_t vHi = vdupq_lane_f32(vget_high_f32(vvec), 0);
-    const uint32x4_t four = (uint32x4_t){ 4, 4, 4, 4 };
-    uint32x4_t local_index = (uint32x4_t) {0, 1, 2, 3};
-    uint32x4_t index = (uint32x4_t) { -1, -1, -1, -1 };
-    float32x4_t minDot = (float32x4_t) { B3_INFINITY, B3_INFINITY, B3_INFINITY, B3_INFINITY };
+	unsigned long i = 0;
+	float32x4_t vvec = vld1q_f32_aligned_postincrement(vec);
+	float32x2_t vLo = vget_low_f32(vvec);
+	float32x2_t vHi = vdup_lane_f32(vget_high_f32(vvec), 0);
+	float32x2_t dotMinLo = (float32x2_t){B3_INFINITY, B3_INFINITY};
+	float32x2_t dotMinHi = (float32x2_t){B3_INFINITY, B3_INFINITY};
+	uint32x2_t indexLo = (uint32x2_t){0, 1};
+	uint32x2_t indexHi = (uint32x2_t){2, 3};
+	uint32x2_t iLo = (uint32x2_t){-1, -1};
+	uint32x2_t iHi = (uint32x2_t){-1, -1};
+	const uint32x2_t four = (uint32x2_t){4, 4};
 
-    unsigned long i = 0;
-    for( ; i + 8 <= count; i += 8 )
-    {
-        float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v3 = vld1q_f32_aligned_postincrement( vv );
+	for (; i + 8 <= count; i += 8)
+	{
+		float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v3 = vld1q_f32_aligned_postincrement(vv);
 
-        // the next two lines should resolve to a single vswp d, d
-        float32x4_t xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v1));
-        float32x4_t xy1 = vcombine_f32( vget_low_f32(v2), vget_low_f32(v3));
-        // the next two lines should resolve to a single vswp d, d
-        float32x4_t z0 = vcombine_f32( vget_high_f32(v0), vget_high_f32(v1));
-        float32x4_t z1 = vcombine_f32( vget_high_f32(v2), vget_high_f32(v3));
+		float32x2_t xy0 = vmul_f32(vget_low_f32(v0), vLo);
+		float32x2_t xy1 = vmul_f32(vget_low_f32(v1), vLo);
+		float32x2_t xy2 = vmul_f32(vget_low_f32(v2), vLo);
+		float32x2_t xy3 = vmul_f32(vget_low_f32(v3), vLo);
 
-        xy0 = vmulq_f32(xy0, vLo);
-        xy1 = vmulq_f32(xy1, vLo);
+		float32x2x2_t z0 = vtrn_f32(vget_high_f32(v0), vget_high_f32(v1));
+		float32x2x2_t z1 = vtrn_f32(vget_high_f32(v2), vget_high_f32(v3));
+		float32x2_t zLo = vmul_f32(z0.val[0], vHi);
+		float32x2_t zHi = vmul_f32(z1.val[0], vHi);
 
-        float32x4x2_t zb = vuzpq_f32( z0, z1);
-        float32x4_t z = vmulq_f32( zb.val[0], vHi);
-        float32x4x2_t xy = vuzpq_f32( xy0, xy1);
-        float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
-        x = vaddq_f32(x, z);
+		float32x2_t rLo = vpadd_f32(xy0, xy1);
+		float32x2_t rHi = vpadd_f32(xy2, xy3);
+		rLo = vadd_f32(rLo, zLo);
+		rHi = vadd_f32(rHi, zHi);
 
-        uint32x4_t mask = vcltq_f32(x, minDot);
-        minDot = vbslq_f32( mask, x, minDot);
-        index = vbslq_u32(mask, local_index, index);
-        local_index = vaddq_u32(local_index, four);
+		uint32x2_t maskLo = vclt_f32(rLo, dotMinLo);
+		uint32x2_t maskHi = vclt_f32(rHi, dotMinHi);
+		dotMinLo = vbsl_f32(maskLo, rLo, dotMinLo);
+		dotMinHi = vbsl_f32(maskHi, rHi, dotMinHi);
+		iLo = vbsl_u32(maskLo, indexLo, iLo);
+		iHi = vbsl_u32(maskHi, indexHi, iHi);
+		indexLo = vadd_u32(indexLo, four);
+		indexHi = vadd_u32(indexHi, four);
 
-        v0 = vld1q_f32_aligned_postincrement( vv );
-        v1 = vld1q_f32_aligned_postincrement( vv );
-        v2 = vld1q_f32_aligned_postincrement( vv );
-        v3 = vld1q_f32_aligned_postincrement( vv );
+		v0 = vld1q_f32_aligned_postincrement(vv);
+		v1 = vld1q_f32_aligned_postincrement(vv);
+		v2 = vld1q_f32_aligned_postincrement(vv);
+		v3 = vld1q_f32_aligned_postincrement(vv);
 
-        // the next two lines should resolve to a single vswp d, d
-        xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v1));
-        xy1 = vcombine_f32( vget_low_f32(v2), vget_low_f32(v3));
-        // the next two lines should resolve to a single vswp d, d
-        z0 = vcombine_f32( vget_high_f32(v0), vget_high_f32(v1));
-        z1 = vcombine_f32( vget_high_f32(v2), vget_high_f32(v3));
+		xy0 = vmul_f32(vget_low_f32(v0), vLo);
+		xy1 = vmul_f32(vget_low_f32(v1), vLo);
+		xy2 = vmul_f32(vget_low_f32(v2), vLo);
+		xy3 = vmul_f32(vget_low_f32(v3), vLo);
 
-        xy0 = vmulq_f32(xy0, vLo);
-        xy1 = vmulq_f32(xy1, vLo);
+		z0 = vtrn_f32(vget_high_f32(v0), vget_high_f32(v1));
+		z1 = vtrn_f32(vget_high_f32(v2), vget_high_f32(v3));
+		zLo = vmul_f32(z0.val[0], vHi);
+		zHi = vmul_f32(z1.val[0], vHi);
 
-        zb = vuzpq_f32( z0, z1);
-        z = vmulq_f32( zb.val[0], vHi);
-        xy = vuzpq_f32( xy0, xy1);
-        x = vaddq_f32(xy.val[0], xy.val[1]);
-        x = vaddq_f32(x, z);
+		rLo = vpadd_f32(xy0, xy1);
+		rHi = vpadd_f32(xy2, xy3);
+		rLo = vadd_f32(rLo, zLo);
+		rHi = vadd_f32(rHi, zHi);
 
-        mask = vcltq_f32(x, minDot);
-        minDot = vbslq_f32( mask, x, minDot);
-        index = vbslq_u32(mask, local_index, index);
-        local_index = vaddq_u32(local_index, four);
-    }
+		maskLo = vclt_f32(rLo, dotMinLo);
+		maskHi = vclt_f32(rHi, dotMinHi);
+		dotMinLo = vbsl_f32(maskLo, rLo, dotMinLo);
+		dotMinHi = vbsl_f32(maskHi, rHi, dotMinHi);
+		iLo = vbsl_u32(maskLo, indexLo, iLo);
+		iHi = vbsl_u32(maskHi, indexHi, iHi);
+		indexLo = vadd_u32(indexLo, four);
+		indexHi = vadd_u32(indexHi, four);
+	}
 
-    for( ; i + 4 <= count; i += 4 )
-    {
-        float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v3 = vld1q_f32_aligned_postincrement( vv );
+	for (; i + 4 <= count; i += 4)
+	{
+		float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v3 = vld1q_f32_aligned_postincrement(vv);
 
-        // the next two lines should resolve to a single vswp d, d
-        float32x4_t xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v1));
-        float32x4_t xy1 = vcombine_f32( vget_low_f32(v2), vget_low_f32(v3));
-        // the next two lines should resolve to a single vswp d, d
-        float32x4_t z0 = vcombine_f32( vget_high_f32(v0), vget_high_f32(v1));
-        float32x4_t z1 = vcombine_f32( vget_high_f32(v2), vget_high_f32(v3));
+		float32x2_t xy0 = vmul_f32(vget_low_f32(v0), vLo);
+		float32x2_t xy1 = vmul_f32(vget_low_f32(v1), vLo);
+		float32x2_t xy2 = vmul_f32(vget_low_f32(v2), vLo);
+		float32x2_t xy3 = vmul_f32(vget_low_f32(v3), vLo);
 
-        xy0 = vmulq_f32(xy0, vLo);
-        xy1 = vmulq_f32(xy1, vLo);
+		float32x2x2_t z0 = vtrn_f32(vget_high_f32(v0), vget_high_f32(v1));
+		float32x2x2_t z1 = vtrn_f32(vget_high_f32(v2), vget_high_f32(v3));
+		float32x2_t zLo = vmul_f32(z0.val[0], vHi);
+		float32x2_t zHi = vmul_f32(z1.val[0], vHi);
 
-        float32x4x2_t zb = vuzpq_f32( z0, z1);
-        float32x4_t z = vmulq_f32( zb.val[0], vHi);
-        float32x4x2_t xy = vuzpq_f32( xy0, xy1);
-        float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
-        x = vaddq_f32(x, z);
+		float32x2_t rLo = vpadd_f32(xy0, xy1);
+		float32x2_t rHi = vpadd_f32(xy2, xy3);
+		rLo = vadd_f32(rLo, zLo);
+		rHi = vadd_f32(rHi, zHi);
 
-        uint32x4_t mask = vcltq_f32(x, minDot);
-        minDot = vbslq_f32( mask, x, minDot);
-        index = vbslq_u32(mask, local_index, index);
-        local_index = vaddq_u32(local_index, four);
-    }
+		uint32x2_t maskLo = vclt_f32(rLo, dotMinLo);
+		uint32x2_t maskHi = vclt_f32(rHi, dotMinHi);
+		dotMinLo = vbsl_f32(maskLo, rLo, dotMinLo);
+		dotMinHi = vbsl_f32(maskHi, rHi, dotMinHi);
+		iLo = vbsl_u32(maskLo, indexLo, iLo);
+		iHi = vbsl_u32(maskHi, indexHi, iHi);
+		indexLo = vadd_u32(indexLo, four);
+		indexHi = vadd_u32(indexHi, four);
+	}
+	switch (count & 3)
+	{
+		case 3:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
 
-    switch (count & 3) {
-        case 3:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
+			float32x2_t xy0 = vmul_f32(vget_low_f32(v0), vLo);
+			float32x2_t xy1 = vmul_f32(vget_low_f32(v1), vLo);
+			float32x2_t xy2 = vmul_f32(vget_low_f32(v2), vLo);
 
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v1));
-            float32x4_t xy1 = vcombine_f32( vget_low_f32(v2), vget_low_f32(v2));
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t z0 = vcombine_f32( vget_high_f32(v0), vget_high_f32(v1));
-            float32x4_t z1 = vcombine_f32( vget_high_f32(v2), vget_high_f32(v2));
+			float32x2x2_t z0 = vtrn_f32(vget_high_f32(v0), vget_high_f32(v1));
+			float32x2_t zLo = vmul_f32(z0.val[0], vHi);
+			float32x2_t zHi = vmul_f32(vdup_lane_f32(vget_high_f32(v2), 0), vHi);
 
-            xy0 = vmulq_f32(xy0, vLo);
-            xy1 = vmulq_f32(xy1, vLo);
+			float32x2_t rLo = vpadd_f32(xy0, xy1);
+			float32x2_t rHi = vpadd_f32(xy2, xy2);
+			rLo = vadd_f32(rLo, zLo);
+			rHi = vadd_f32(rHi, zHi);
 
-            float32x4x2_t zb = vuzpq_f32( z0, z1);
-            float32x4_t z = vmulq_f32( zb.val[0], vHi);
-            float32x4x2_t xy = vuzpq_f32( xy0, xy1);
-            float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
-            x = vaddq_f32(x, z);
+			uint32x2_t maskLo = vclt_f32(rLo, dotMinLo);
+			uint32x2_t maskHi = vclt_f32(rHi, dotMinHi);
+			dotMinLo = vbsl_f32(maskLo, rLo, dotMinLo);
+			dotMinHi = vbsl_f32(maskHi, rHi, dotMinHi);
+			iLo = vbsl_u32(maskLo, indexLo, iLo);
+			iHi = vbsl_u32(maskHi, indexHi, iHi);
+		}
+		break;
+		case 2:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
 
-            uint32x4_t mask = vcltq_f32(x, minDot);
-            minDot = vbslq_f32( mask, x, minDot);
-            index = vbslq_u32(mask, local_index, index);
-            local_index = vaddq_u32(local_index, four);
-        }
-            break;
+			float32x2_t xy0 = vmul_f32(vget_low_f32(v0), vLo);
+			float32x2_t xy1 = vmul_f32(vget_low_f32(v1), vLo);
 
-        case 2:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
+			float32x2x2_t z0 = vtrn_f32(vget_high_f32(v0), vget_high_f32(v1));
+			float32x2_t zLo = vmul_f32(z0.val[0], vHi);
 
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v1));
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t z0 = vcombine_f32( vget_high_f32(v0), vget_high_f32(v1));
+			float32x2_t rLo = vpadd_f32(xy0, xy1);
+			rLo = vadd_f32(rLo, zLo);
 
-            xy0 = vmulq_f32(xy0, vLo);
+			uint32x2_t maskLo = vclt_f32(rLo, dotMinLo);
+			dotMinLo = vbsl_f32(maskLo, rLo, dotMinLo);
+			iLo = vbsl_u32(maskLo, indexLo, iLo);
+		}
+		break;
+		case 1:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+			float32x2_t xy0 = vmul_f32(vget_low_f32(v0), vLo);
+			float32x2_t z0 = vdup_lane_f32(vget_high_f32(v0), 0);
+			float32x2_t zLo = vmul_f32(z0, vHi);
+			float32x2_t rLo = vpadd_f32(xy0, xy0);
+			rLo = vadd_f32(rLo, zLo);
+			uint32x2_t maskLo = vclt_f32(rLo, dotMinLo);
+			dotMinLo = vbsl_f32(maskLo, rLo, dotMinLo);
+			iLo = vbsl_u32(maskLo, indexLo, iLo);
+		}
+		break;
 
-            float32x4x2_t zb = vuzpq_f32( z0, z0);
-            float32x4_t z = vmulq_f32( zb.val[0], vHi);
-            float32x4x2_t xy = vuzpq_f32( xy0, xy0);
-            float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
-            x = vaddq_f32(x, z);
+		default:
+			break;
+	}
 
-            uint32x4_t mask = vcltq_f32(x, minDot);
-            minDot = vbslq_f32( mask, x, minDot);
-            index = vbslq_u32(mask, local_index, index);
-            local_index = vaddq_u32(local_index, four);
-        }
-            break;
+	// select best answer between hi and lo results
+	uint32x2_t mask = vclt_f32(dotMinHi, dotMinLo);
+	dotMinLo = vbsl_f32(mask, dotMinHi, dotMinLo);
+	iLo = vbsl_u32(mask, iHi, iLo);
 
-        case 1:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
+	// select best answer between even and odd results
+	dotMinHi = vdup_lane_f32(dotMinLo, 1);
+	iHi = vdup_lane_u32(iLo, 1);
+	mask = vclt_f32(dotMinHi, dotMinLo);
+	dotMinLo = vbsl_f32(mask, dotMinHi, dotMinLo);
+	iLo = vbsl_u32(mask, iHi, iLo);
 
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v0));
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t z = vdupq_lane_f32(vget_high_f32(v0), 0);
+	*dotResult = vget_lane_f32(dotMinLo, 0);
+	return vget_lane_u32(iLo, 0);
+}
 
-            xy0 = vmulq_f32(xy0, vLo);
+long b3_mindot_large_v1(const float *vv, const float *vec, unsigned long count, float *dotResult)
+{
+	float32x4_t vvec = vld1q_f32_aligned_postincrement(vec);
+	float32x4_t vLo = vcombine_f32(vget_low_f32(vvec), vget_low_f32(vvec));
+	float32x4_t vHi = vdupq_lane_f32(vget_high_f32(vvec), 0);
+	const uint32x4_t four = (uint32x4_t){4, 4, 4, 4};
+	uint32x4_t local_index = (uint32x4_t){0, 1, 2, 3};
+	uint32x4_t index = (uint32x4_t){-1, -1, -1, -1};
+	float32x4_t minDot = (float32x4_t){B3_INFINITY, B3_INFINITY, B3_INFINITY, B3_INFINITY};
 
-            z = vmulq_f32( z, vHi);
-            float32x4x2_t xy = vuzpq_f32( xy0, xy0);
-            float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
-            x = vaddq_f32(x, z);
+	unsigned long i = 0;
+	for (; i + 8 <= count; i += 8)
+	{
+		float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v3 = vld1q_f32_aligned_postincrement(vv);
 
-            uint32x4_t mask = vcltq_f32(x, minDot);
-            minDot = vbslq_f32( mask, x, minDot);
-            index = vbslq_u32(mask, local_index, index);
-            local_index = vaddq_u32(local_index, four);
-        }
-            break;
+		// the next two lines should resolve to a single vswp d, d
+		float32x4_t xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v1));
+		float32x4_t xy1 = vcombine_f32(vget_low_f32(v2), vget_low_f32(v3));
+		// the next two lines should resolve to a single vswp d, d
+		float32x4_t z0 = vcombine_f32(vget_high_f32(v0), vget_high_f32(v1));
+		float32x4_t z1 = vcombine_f32(vget_high_f32(v2), vget_high_f32(v3));
 
-        default:
-            break;
-    }
+		xy0 = vmulq_f32(xy0, vLo);
+		xy1 = vmulq_f32(xy1, vLo);
 
+		float32x4x2_t zb = vuzpq_f32(z0, z1);
+		float32x4_t z = vmulq_f32(zb.val[0], vHi);
+		float32x4x2_t xy = vuzpq_f32(xy0, xy1);
+		float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
+		x = vaddq_f32(x, z);
 
-    // select best answer between hi and lo results
-    uint32x2_t mask = vclt_f32( vget_high_f32(minDot), vget_low_f32(minDot));
-    float32x2_t minDot2 = vbsl_f32(mask, vget_high_f32(minDot), vget_low_f32(minDot));
-    uint32x2_t index2 = vbsl_u32(mask, vget_high_u32(index), vget_low_u32(index));
+		uint32x4_t mask = vcltq_f32(x, minDot);
+		minDot = vbslq_f32(mask, x, minDot);
+		index = vbslq_u32(mask, local_index, index);
+		local_index = vaddq_u32(local_index, four);
 
-    // select best answer between even and odd results
-    float32x2_t minDotO = vdup_lane_f32(minDot2, 1);
-    uint32x2_t indexHi = vdup_lane_u32(index2, 1);
-    mask = vclt_f32( minDotO, minDot2 );
-    minDot2 = vbsl_f32(mask, minDotO, minDot2);
-    index2 = vbsl_u32(mask, indexHi, index2);
+		v0 = vld1q_f32_aligned_postincrement(vv);
+		v1 = vld1q_f32_aligned_postincrement(vv);
+		v2 = vld1q_f32_aligned_postincrement(vv);
+		v3 = vld1q_f32_aligned_postincrement(vv);
 
-    *dotResult = vget_lane_f32( minDot2, 0);
-    return vget_lane_u32(index2, 0);
+		// the next two lines should resolve to a single vswp d, d
+		xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v1));
+		xy1 = vcombine_f32(vget_low_f32(v2), vget_low_f32(v3));
+		// the next two lines should resolve to a single vswp d, d
+		z0 = vcombine_f32(vget_high_f32(v0), vget_high_f32(v1));
+		z1 = vcombine_f32(vget_high_f32(v2), vget_high_f32(v3));
 
+		xy0 = vmulq_f32(xy0, vLo);
+		xy1 = vmulq_f32(xy1, vLo);
+
+		zb = vuzpq_f32(z0, z1);
+		z = vmulq_f32(zb.val[0], vHi);
+		xy = vuzpq_f32(xy0, xy1);
+		x = vaddq_f32(xy.val[0], xy.val[1]);
+		x = vaddq_f32(x, z);
+
+		mask = vcltq_f32(x, minDot);
+		minDot = vbslq_f32(mask, x, minDot);
+		index = vbslq_u32(mask, local_index, index);
+		local_index = vaddq_u32(local_index, four);
+	}
+
+	for (; i + 4 <= count; i += 4)
+	{
+		float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v3 = vld1q_f32_aligned_postincrement(vv);
+
+		// the next two lines should resolve to a single vswp d, d
+		float32x4_t xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v1));
+		float32x4_t xy1 = vcombine_f32(vget_low_f32(v2), vget_low_f32(v3));
+		// the next two lines should resolve to a single vswp d, d
+		float32x4_t z0 = vcombine_f32(vget_high_f32(v0), vget_high_f32(v1));
+		float32x4_t z1 = vcombine_f32(vget_high_f32(v2), vget_high_f32(v3));
+
+		xy0 = vmulq_f32(xy0, vLo);
+		xy1 = vmulq_f32(xy1, vLo);
+
+		float32x4x2_t zb = vuzpq_f32(z0, z1);
+		float32x4_t z = vmulq_f32(zb.val[0], vHi);
+		float32x4x2_t xy = vuzpq_f32(xy0, xy1);
+		float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
+		x = vaddq_f32(x, z);
+
+		uint32x4_t mask = vcltq_f32(x, minDot);
+		minDot = vbslq_f32(mask, x, minDot);
+		index = vbslq_u32(mask, local_index, index);
+		local_index = vaddq_u32(local_index, four);
+	}
+
+	switch (count & 3)
+	{
+		case 3:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v1));
+			float32x4_t xy1 = vcombine_f32(vget_low_f32(v2), vget_low_f32(v2));
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t z0 = vcombine_f32(vget_high_f32(v0), vget_high_f32(v1));
+			float32x4_t z1 = vcombine_f32(vget_high_f32(v2), vget_high_f32(v2));
+
+			xy0 = vmulq_f32(xy0, vLo);
+			xy1 = vmulq_f32(xy1, vLo);
+
+			float32x4x2_t zb = vuzpq_f32(z0, z1);
+			float32x4_t z = vmulq_f32(zb.val[0], vHi);
+			float32x4x2_t xy = vuzpq_f32(xy0, xy1);
+			float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
+			x = vaddq_f32(x, z);
+
+			uint32x4_t mask = vcltq_f32(x, minDot);
+			minDot = vbslq_f32(mask, x, minDot);
+			index = vbslq_u32(mask, local_index, index);
+			local_index = vaddq_u32(local_index, four);
+		}
+		break;
+
+		case 2:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v1));
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t z0 = vcombine_f32(vget_high_f32(v0), vget_high_f32(v1));
+
+			xy0 = vmulq_f32(xy0, vLo);
+
+			float32x4x2_t zb = vuzpq_f32(z0, z0);
+			float32x4_t z = vmulq_f32(zb.val[0], vHi);
+			float32x4x2_t xy = vuzpq_f32(xy0, xy0);
+			float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
+			x = vaddq_f32(x, z);
+
+			uint32x4_t mask = vcltq_f32(x, minDot);
+			minDot = vbslq_f32(mask, x, minDot);
+			index = vbslq_u32(mask, local_index, index);
+			local_index = vaddq_u32(local_index, four);
+		}
+		break;
+
+		case 1:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v0));
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t z = vdupq_lane_f32(vget_high_f32(v0), 0);
+
+			xy0 = vmulq_f32(xy0, vLo);
+
+			z = vmulq_f32(z, vHi);
+			float32x4x2_t xy = vuzpq_f32(xy0, xy0);
+			float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
+			x = vaddq_f32(x, z);
+
+			uint32x4_t mask = vcltq_f32(x, minDot);
+			minDot = vbslq_f32(mask, x, minDot);
+			index = vbslq_u32(mask, local_index, index);
+			local_index = vaddq_u32(local_index, four);
+		}
+		break;
+
+		default:
+			break;
+	}
+
+	// select best answer between hi and lo results
+	uint32x2_t mask = vclt_f32(vget_high_f32(minDot), vget_low_f32(minDot));
+	float32x2_t minDot2 = vbsl_f32(mask, vget_high_f32(minDot), vget_low_f32(minDot));
+	uint32x2_t index2 = vbsl_u32(mask, vget_high_u32(index), vget_low_u32(index));
+
+	// select best answer between even and odd results
+	float32x2_t minDotO = vdup_lane_f32(minDot2, 1);
+	uint32x2_t indexHi = vdup_lane_u32(index2, 1);
+	mask = vclt_f32(minDotO, minDot2);
+	minDot2 = vbsl_f32(mask, minDotO, minDot2);
+	index2 = vbsl_u32(mask, indexHi, index2);
+
+	*dotResult = vget_lane_f32(minDot2, 0);
+	return vget_lane_u32(index2, 0);
 }
 
 #else
-    #error Unhandled __APPLE__ arch
+#error Unhandled __APPLE__ arch
 #endif
 
-#endif  /* __APPLE__ */
-
-
+#endif /* __APPLE__ */
diff --git a/thirdparty/bullet/Bullet3Common/b3Vector3.h b/thirdparty/bullet/Bullet3Common/b3Vector3.h
index 16ec02b0eda..56e6c13311d 100644
--- a/thirdparty/bullet/Bullet3Common/b3Vector3.h
+++ b/thirdparty/bullet/Bullet3Common/b3Vector3.h
@@ -12,8 +12,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #ifndef B3_VECTOR3_H
 #define B3_VECTOR3_H
 
@@ -28,37 +26,34 @@ subject to the following restrictions:
 #else
 #define b3Vector3Data b3Vector3FloatData
 #define b3Vector3DataName "b3Vector3FloatData"
-#endif //B3_USE_DOUBLE_PRECISION
+#endif  //B3_USE_DOUBLE_PRECISION
 
 #if defined B3_USE_SSE
 
 //typedef  uint32_t __m128i __attribute__ ((vector_size(16)));
 
 #ifdef _MSC_VER
-#pragma warning(disable: 4556) // value of intrinsic immediate argument '4294967239' is out of range '0 - 255'
+#pragma warning(disable : 4556)  // value of intrinsic immediate argument '4294967239' is out of range '0 - 255'
 #endif
 
-
-#define B3_SHUFFLE(x,y,z,w) ((w)<<6 | (z)<<4 | (y)<<2 | (x))
+#define B3_SHUFFLE(x, y, z, w) ((w) << 6 | (z) << 4 | (y) << 2 | (x))
 //#define b3_pshufd_ps( _a, _mask ) (__m128) _mm_shuffle_epi32((__m128i)(_a), (_mask) )
-#define b3_pshufd_ps( _a, _mask ) _mm_shuffle_ps((_a), (_a), (_mask) )
-#define b3_splat3_ps( _a, _i ) b3_pshufd_ps((_a), B3_SHUFFLE(_i,_i,_i, 3) )
-#define b3_splat_ps( _a, _i )  b3_pshufd_ps((_a), B3_SHUFFLE(_i,_i,_i,_i) )
+#define b3_pshufd_ps(_a, _mask) _mm_shuffle_ps((_a), (_a), (_mask))
+#define b3_splat3_ps(_a, _i) b3_pshufd_ps((_a), B3_SHUFFLE(_i, _i, _i, 3))
+#define b3_splat_ps(_a, _i) b3_pshufd_ps((_a), B3_SHUFFLE(_i, _i, _i, _i))
 
 #define b3v3AbsiMask (_mm_set_epi32(0x00000000, 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF))
-#define b3vAbsMask (_mm_set_epi32( 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF))
+#define b3vAbsMask (_mm_set_epi32(0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF))
 #define b3vFFF0Mask (_mm_set_epi32(0x00000000, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF))
 #define b3v3AbsfMask b3CastiTo128f(b3v3AbsiMask)
 #define b3vFFF0fMask b3CastiTo128f(b3vFFF0Mask)
 #define b3vxyzMaskf b3vFFF0fMask
 #define b3vAbsfMask b3CastiTo128f(b3vAbsMask)
 
-
-
 const __m128 B3_ATTRIBUTE_ALIGNED16(b3vMzeroMask) = {-0.0f, -0.0f, -0.0f, -0.0f};
 const __m128 B3_ATTRIBUTE_ALIGNED16(b3v1110) = {1.0f, 1.0f, 1.0f, 0.0f};
 const __m128 B3_ATTRIBUTE_ALIGNED16(b3vHalf) = {0.5f, 0.5f, 0.5f, 0.5f};
-const __m128 B3_ATTRIBUTE_ALIGNED16(b3v1_5)  = {1.5f, 1.5f, 1.5f, 1.5f};
+const __m128 B3_ATTRIBUTE_ALIGNED16(b3v1_5) = {1.5f, 1.5f, 1.5f, 1.5f};
 
 #endif
 
@@ -74,70 +69,69 @@ const int32x4_t B3_ATTRIBUTE_ALIGNED16(b3v3AbsMask) = (int32x4_t){0x7FFFFFFF, 0x
 class b3Vector3;
 class b3Vector4;
 
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 //#if defined (B3_USE_SSE) || defined (B3_USE_NEON)
-inline b3Vector3 b3MakeVector3( b3SimdFloat4 v);
-inline b3Vector4 b3MakeVector4( b3SimdFloat4 vec);
+inline b3Vector3 b3MakeVector3(b3SimdFloat4 v);
+inline b3Vector4 b3MakeVector4(b3SimdFloat4 vec);
 #endif
 
-inline b3Vector3 b3MakeVector3(b3Scalar x,b3Scalar y,b3Scalar z);
-inline b3Vector3 b3MakeVector3(b3Scalar x,b3Scalar y,b3Scalar z, b3Scalar w);
-inline b3Vector4 b3MakeVector4(b3Scalar x,b3Scalar y,b3Scalar z,b3Scalar w);
-
+inline b3Vector3 b3MakeVector3(b3Scalar x, b3Scalar y, b3Scalar z);
+inline b3Vector3 b3MakeVector3(b3Scalar x, b3Scalar y, b3Scalar z, b3Scalar w);
+inline b3Vector4 b3MakeVector4(b3Scalar x, b3Scalar y, b3Scalar z, b3Scalar w);
 
 /**@brief b3Vector3 can be used to represent 3D points and vectors.
  * It has an un-used w component to suit 16-byte alignment when b3Vector3 is stored in containers. This extra component can be used by derived classes (Quaternion?) or by user
  * Ideally, this class should be replaced by a platform optimized SIMD version that keeps the data in registers
  */
-B3_ATTRIBUTE_ALIGNED16(class) b3Vector3
+B3_ATTRIBUTE_ALIGNED16(class)
+b3Vector3
 {
 public:
-#if defined (B3_USE_SSE) || defined(B3_USE_NEON) // _WIN32 || ARM
-        union {
-            b3SimdFloat4      mVec128;
-            float	m_floats[4];
-			struct {float x,y,z,w;};
-
-        };
+#if defined(B3_USE_SSE) || defined(B3_USE_NEON)  // _WIN32 || ARM
+	union {
+		b3SimdFloat4 mVec128;
+		float m_floats[4];
+		struct
+		{
+			float x, y, z, w;
+		};
+	};
 #else
-	union
-	{
-        	float	m_floats[4];
-			struct {float	x,y,z,w;};
+	union {
+		float m_floats[4];
+		struct
+		{
+			float x, y, z, w;
+		};
 	};
 #endif
 
-
 public:
-
 	B3_DECLARE_ALIGNED_ALLOCATOR();
 
-#if defined (B3_USE_SSE) || defined(B3_USE_NEON) // _WIN32 || ARM
+#if defined(B3_USE_SSE) || defined(B3_USE_NEON)  // _WIN32 || ARM
 
 	/*B3_FORCE_INLINE		b3Vector3()
 	{
 	}
 	*/
 
-    B3_FORCE_INLINE	b3SimdFloat4	get128() const
-    {
-        return mVec128;
-    }
-    B3_FORCE_INLINE	void	set128(b3SimdFloat4 v128)
-    {
-        mVec128 = v128;
-    }
+	B3_FORCE_INLINE b3SimdFloat4 get128() const
+	{
+		return mVec128;
+	}
+	B3_FORCE_INLINE void set128(b3SimdFloat4 v128)
+	{
+		mVec128 = v128;
+	}
 #endif
 
-	public:
-
-
-
-/**@brief Add a vector to this one
+public:
+	/**@brief Add a vector to this one
  * @param The vector to add to this one */
 	B3_FORCE_INLINE b3Vector3& operator+=(const b3Vector3& v)
 	{
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 		mVec128 = _mm_add_ps(mVec128, v.mVec128);
 #elif defined(B3_USE_NEON)
 		mVec128 = vaddq_f32(mVec128, v.mVec128);
@@ -149,12 +143,11 @@ public:
 		return *this;
 	}
 
-
-  /**@brief Subtract a vector from this one
+	/**@brief Subtract a vector from this one
    * @param The vector to subtract */
 	B3_FORCE_INLINE b3Vector3& operator-=(const b3Vector3& v)
 	{
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 		mVec128 = _mm_sub_ps(mVec128, v.mVec128);
 #elif defined(B3_USE_NEON)
 		mVec128 = vsubq_f32(mVec128, v.mVec128);
@@ -166,13 +159,13 @@ public:
 		return *this;
 	}
 
-  /**@brief Scale the vector
+	/**@brief Scale the vector
    * @param s Scale factor */
 	B3_FORCE_INLINE b3Vector3& operator*=(const b3Scalar& s)
 	{
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
-		__m128	vs = _mm_load_ss(&s);	//	(S 0 0 0)
-		vs = b3_pshufd_ps(vs, 0x80);	//	(S S S 0.0)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
+		__m128 vs = _mm_load_ss(&s);  //	(S 0 0 0)
+		vs = b3_pshufd_ps(vs, 0x80);  //	(S S S 0.0)
 		mVec128 = _mm_mul_ps(mVec128, vs);
 #elif defined(B3_USE_NEON)
 		mVec128 = vmulq_n_f32(mVec128, s);
@@ -184,13 +177,13 @@ public:
 		return *this;
 	}
 
-  /**@brief Inversely scale the vector
+	/**@brief Inversely scale the vector
    * @param s Scale factor to divide by */
 	B3_FORCE_INLINE b3Vector3& operator/=(const b3Scalar& s)
 	{
 		b3FullAssert(s != b3Scalar(0.0));
 
-#if 0 //defined(B3_USE_SSE_IN_API)
+#if 0  //defined(B3_USE_SSE_IN_API)
 // this code is not faster !
 		__m128 vs = _mm_load_ss(&s);
 		vs = _mm_div_ss(b3v1110, vs);
@@ -204,11 +197,11 @@ public:
 #endif
 	}
 
-  /**@brief Return the dot product
+	/**@brief Return the dot product
    * @param v The other vector in the dot product */
 	B3_FORCE_INLINE b3Scalar dot(const b3Vector3& v) const
 	{
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 		__m128 vd = _mm_mul_ps(mVec128, v.mVec128);
 		__m128 z = _mm_movehl_ps(vd, vd);
 		__m128 y = _mm_shuffle_ps(vd, vd, 0x55);
@@ -221,29 +214,29 @@ public:
 		x = vadd_f32(x, vget_high_f32(vd));
 		return vget_lane_f32(x, 0);
 #else
-		return	m_floats[0] * v.m_floats[0] +
-				m_floats[1] * v.m_floats[1] +
-				m_floats[2] * v.m_floats[2];
+		return m_floats[0] * v.m_floats[0] +
+			   m_floats[1] * v.m_floats[1] +
+			   m_floats[2] * v.m_floats[2];
 #endif
 	}
 
-  /**@brief Return the length of the vector squared */
+	/**@brief Return the length of the vector squared */
 	B3_FORCE_INLINE b3Scalar length2() const
 	{
 		return dot(*this);
 	}
 
-  /**@brief Return the length of the vector */
+	/**@brief Return the length of the vector */
 	B3_FORCE_INLINE b3Scalar length() const
 	{
 		return b3Sqrt(length2());
 	}
 
-  /**@brief Return the distance squared between the ends of this and another vector
+	/**@brief Return the distance squared between the ends of this and another vector
    * This is symantically treating the vector like a point */
 	B3_FORCE_INLINE b3Scalar distance2(const b3Vector3& v) const;
 
-  /**@brief Return the distance between the ends of this and another vector
+	/**@brief Return the distance between the ends of this and another vector
    * This is symantically treating the vector like a point */
 	B3_FORCE_INLINE b3Scalar distance(const b3Vector3& v) const;
 
@@ -251,7 +244,7 @@ public:
 	{
 		b3Scalar l2 = length2();
 		//triNormal.normalize();
-		if (l2 >= B3_EPSILON*B3_EPSILON)
+		if (l2 >= B3_EPSILON * B3_EPSILON)
 		{
 			(*this) /= b3Sqrt(l2);
 		}
@@ -262,43 +255,42 @@ public:
 		return *this;
 	}
 
-  /**@brief Normalize this vector
+	/**@brief Normalize this vector
    * x^2 + y^2 + z^2 = 1 */
 	B3_FORCE_INLINE b3Vector3& normalize()
 	{
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
-        // dot product first
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
+		// dot product first
 		__m128 vd = _mm_mul_ps(mVec128, mVec128);
 		__m128 z = _mm_movehl_ps(vd, vd);
 		__m128 y = _mm_shuffle_ps(vd, vd, 0x55);
 		vd = _mm_add_ss(vd, y);
 		vd = _mm_add_ss(vd, z);
 
-        #if 0
+#if 0
         vd = _mm_sqrt_ss(vd);
 		vd = _mm_div_ss(b3v1110, vd);
 		vd = b3_splat_ps(vd, 0x80);
 		mVec128 = _mm_mul_ps(mVec128, vd);
-        #else
+#else
 
-        // NR step 1/sqrt(x) - vd is x, y is output
-        y = _mm_rsqrt_ss(vd); // estimate
+		// NR step 1/sqrt(x) - vd is x, y is output
+		y = _mm_rsqrt_ss(vd);  // estimate
 
-        //  one step NR
-        z = b3v1_5;
-        vd = _mm_mul_ss(vd, b3vHalf); // vd * 0.5
-        //x2 = vd;
-        vd = _mm_mul_ss(vd, y); // vd * 0.5 * y0
-        vd = _mm_mul_ss(vd, y); // vd * 0.5 * y0 * y0
-        z = _mm_sub_ss(z, vd);  // 1.5 - vd * 0.5 * y0 * y0
+		//  one step NR
+		z = b3v1_5;
+		vd = _mm_mul_ss(vd, b3vHalf);  // vd * 0.5
+		//x2 = vd;
+		vd = _mm_mul_ss(vd, y);  // vd * 0.5 * y0
+		vd = _mm_mul_ss(vd, y);  // vd * 0.5 * y0 * y0
+		z = _mm_sub_ss(z, vd);   // 1.5 - vd * 0.5 * y0 * y0
 
-        y = _mm_mul_ss(y, z);   // y0 * (1.5 - vd * 0.5 * y0 * y0)
+		y = _mm_mul_ss(y, z);  // y0 * (1.5 - vd * 0.5 * y0 * y0)
 
 		y = b3_splat_ps(y, 0x80);
 		mVec128 = _mm_mul_ps(mVec128, y);
 
-        #endif
-
+#endif
 
 		return *this;
 #else
@@ -306,15 +298,15 @@ public:
 #endif
 	}
 
-  /**@brief Return a normalized version of this vector */
+	/**@brief Return a normalized version of this vector */
 	B3_FORCE_INLINE b3Vector3 normalized() const;
 
-  /**@brief Return a rotated version of this vector
+	/**@brief Return a rotated version of this vector
    * @param wAxis The axis to rotate about
    * @param angle The angle to rotate by */
-	B3_FORCE_INLINE b3Vector3 rotate( const b3Vector3& wAxis, const b3Scalar angle ) const;
+	B3_FORCE_INLINE b3Vector3 rotate(const b3Vector3& wAxis, const b3Scalar angle) const;
 
-  /**@brief Return the angle between this and another vector
+	/**@brief Return the angle between this and another vector
    * @param v The other vector */
 	B3_FORCE_INLINE b3Scalar angle(const b3Vector3& v) const
 	{
@@ -323,10 +315,10 @@ public:
 		return b3Acos(dot(v) / s);
 	}
 
-  /**@brief Return a vector will the absolute values of each element */
+	/**@brief Return a vector will the absolute values of each element */
 	B3_FORCE_INLINE b3Vector3 absolute() const
 	{
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 		return b3MakeVector3(_mm_and_ps(mVec128, b3v3AbsfMask));
 #elif defined(B3_USE_NEON)
 		return b3Vector3(vabsq_f32(mVec128));
@@ -338,15 +330,15 @@ public:
 #endif
 	}
 
-  /**@brief Return the cross product between this and another vector
+	/**@brief Return the cross product between this and another vector
    * @param v The other vector */
 	B3_FORCE_INLINE b3Vector3 cross(const b3Vector3& v) const
 	{
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
-		__m128	T, V;
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
+		__m128 T, V;
 
-		T = b3_pshufd_ps(mVec128, B3_SHUFFLE(1, 2, 0, 3));	//	(Y Z X 0)
-		V = b3_pshufd_ps(v.mVec128, B3_SHUFFLE(1, 2, 0, 3));	//	(Y Z X 0)
+		T = b3_pshufd_ps(mVec128, B3_SHUFFLE(1, 2, 0, 3));    //	(Y Z X 0)
+		V = b3_pshufd_ps(v.mVec128, B3_SHUFFLE(1, 2, 0, 3));  //	(Y Z X 0)
 
 		V = _mm_mul_ps(V, mVec128);
 		T = _mm_mul_ps(T, v.mVec128);
@@ -381,10 +373,10 @@ public:
 
 	B3_FORCE_INLINE b3Scalar triple(const b3Vector3& v1, const b3Vector3& v2) const
 	{
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 		// cross:
-		__m128 T = _mm_shuffle_ps(v1.mVec128, v1.mVec128, B3_SHUFFLE(1, 2, 0, 3));	//	(Y Z X 0)
-		__m128 V = _mm_shuffle_ps(v2.mVec128, v2.mVec128, B3_SHUFFLE(1, 2, 0, 3));	//	(Y Z X 0)
+		__m128 T = _mm_shuffle_ps(v1.mVec128, v1.mVec128, B3_SHUFFLE(1, 2, 0, 3));  //	(Y Z X 0)
+		__m128 V = _mm_shuffle_ps(v2.mVec128, v2.mVec128, B3_SHUFFLE(1, 2, 0, 3));  //	(Y Z X 0)
 
 		V = _mm_mul_ps(V, v1.mVec128);
 		T = _mm_mul_ps(T, v2.mVec128);
@@ -422,25 +414,24 @@ public:
 		x = vadd_f32(x, vget_high_f32(V));
 		return vget_lane_f32(x, 0);
 #else
-		return
-			m_floats[0] * (v1.m_floats[1] * v2.m_floats[2] - v1.m_floats[2] * v2.m_floats[1]) +
-			m_floats[1] * (v1.m_floats[2] * v2.m_floats[0] - v1.m_floats[0] * v2.m_floats[2]) +
-			m_floats[2] * (v1.m_floats[0] * v2.m_floats[1] - v1.m_floats[1] * v2.m_floats[0]);
+		return m_floats[0] * (v1.m_floats[1] * v2.m_floats[2] - v1.m_floats[2] * v2.m_floats[1]) +
+			   m_floats[1] * (v1.m_floats[2] * v2.m_floats[0] - v1.m_floats[0] * v2.m_floats[2]) +
+			   m_floats[2] * (v1.m_floats[0] * v2.m_floats[1] - v1.m_floats[1] * v2.m_floats[0]);
 #endif
 	}
 
-  /**@brief Return the axis with the smallest value
+	/**@brief Return the axis with the smallest value
    * Note return values are 0,1,2 for x, y, or z */
 	B3_FORCE_INLINE int minAxis() const
 	{
-		return m_floats[0] < m_floats[1] ? (m_floats[0] <m_floats[2] ? 0 : 2) : (m_floats[1] <m_floats[2] ? 1 : 2);
+		return m_floats[0] < m_floats[1] ? (m_floats[0] < m_floats[2] ? 0 : 2) : (m_floats[1] < m_floats[2] ? 1 : 2);
 	}
 
-  /**@brief Return the axis with the largest value
+	/**@brief Return the axis with the largest value
    * Note return values are 0,1,2 for x, y, or z */
 	B3_FORCE_INLINE int maxAxis() const
 	{
-		return m_floats[0] < m_floats[1] ? (m_floats[1] <m_floats[2] ? 2 : 1) : (m_floats[0] <m_floats[2] ? 2 : 0);
+		return m_floats[0] < m_floats[1] ? (m_floats[1] < m_floats[2] ? 2 : 1) : (m_floats[0] < m_floats[2] ? 2 : 0);
 	}
 
 	B3_FORCE_INLINE int furthestAxis() const
@@ -453,18 +444,17 @@ public:
 		return absolute().maxAxis();
 	}
 
-
 	B3_FORCE_INLINE void setInterpolate3(const b3Vector3& v0, const b3Vector3& v1, b3Scalar rt)
 	{
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
-		__m128	vrt = _mm_load_ss(&rt);	//	(rt 0 0 0)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
+		__m128 vrt = _mm_load_ss(&rt);  //	(rt 0 0 0)
 		b3Scalar s = b3Scalar(1.0) - rt;
-		__m128	vs = _mm_load_ss(&s);	//	(S 0 0 0)
-		vs = b3_pshufd_ps(vs, 0x80);	//	(S S S 0.0)
+		__m128 vs = _mm_load_ss(&s);  //	(S 0 0 0)
+		vs = b3_pshufd_ps(vs, 0x80);  //	(S S S 0.0)
 		__m128 r0 = _mm_mul_ps(v0.mVec128, vs);
-		vrt = b3_pshufd_ps(vrt, 0x80);	//	(rt rt rt 0.0)
+		vrt = b3_pshufd_ps(vrt, 0x80);  //	(rt rt rt 0.0)
 		__m128 r1 = _mm_mul_ps(v1.mVec128, vrt);
-		__m128 tmp3 = _mm_add_ps(r0,r1);
+		__m128 tmp3 = _mm_add_ps(r0, r1);
 		mVec128 = tmp3;
 #elif defined(B3_USE_NEON)
 		float32x4_t vl = vsubq_f32(v1.mVec128, v0.mVec128);
@@ -480,14 +470,14 @@ public:
 #endif
 	}
 
-  /**@brief Return the linear interpolation between this and another vector
+	/**@brief Return the linear interpolation between this and another vector
    * @param v The other vector
    * @param t The ration of this to v (t = 0 => return this, t=1 => return other) */
 	B3_FORCE_INLINE b3Vector3 lerp(const b3Vector3& v, const b3Scalar& t) const
 	{
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
-		__m128	vt = _mm_load_ss(&t);	//	(t 0 0 0)
-		vt = b3_pshufd_ps(vt, 0x80);	//	(rt rt rt 0.0)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
+		__m128 vt = _mm_load_ss(&t);  //	(t 0 0 0)
+		vt = b3_pshufd_ps(vt, 0x80);  //	(rt rt rt 0.0)
 		__m128 vl = _mm_sub_ps(v.mVec128, mVec128);
 		vl = _mm_mul_ps(vl, vt);
 		vl = _mm_add_ps(vl, mVec128);
@@ -500,18 +490,17 @@ public:
 
 		return b3Vector3(vl);
 #else
-		return
-			b3MakeVector3(	m_floats[0] + (v.m_floats[0] - m_floats[0]) * t,
-						m_floats[1] + (v.m_floats[1] - m_floats[1]) * t,
-						m_floats[2] + (v.m_floats[2] - m_floats[2]) * t);
+		return b3MakeVector3(m_floats[0] + (v.m_floats[0] - m_floats[0]) * t,
+							 m_floats[1] + (v.m_floats[1] - m_floats[1]) * t,
+							 m_floats[2] + (v.m_floats[2] - m_floats[2]) * t);
 #endif
 	}
 
-  /**@brief Elementwise multiply this vector by the other
+	/**@brief Elementwise multiply this vector by the other
    * @param v The other vector */
 	B3_FORCE_INLINE b3Vector3& operator*=(const b3Vector3& v)
 	{
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 		mVec128 = _mm_mul_ps(mVec128, v.mVec128);
 #elif defined(B3_USE_NEON)
 		mVec128 = vmulq_f32(mVec128, v.mVec128);
@@ -523,53 +512,53 @@ public:
 		return *this;
 	}
 
-	 /**@brief Return the x value */
-		B3_FORCE_INLINE const b3Scalar& getX() const { return m_floats[0]; }
-  /**@brief Return the y value */
-		B3_FORCE_INLINE const b3Scalar& getY() const { return m_floats[1]; }
-  /**@brief Return the z value */
-		B3_FORCE_INLINE const b3Scalar& getZ() const { return m_floats[2]; }
-/**@brief Return the w value */
-		B3_FORCE_INLINE const b3Scalar& getW() const { return m_floats[3]; }
+	/**@brief Return the x value */
+	B3_FORCE_INLINE const b3Scalar& getX() const { return m_floats[0]; }
+	/**@brief Return the y value */
+	B3_FORCE_INLINE const b3Scalar& getY() const { return m_floats[1]; }
+	/**@brief Return the z value */
+	B3_FORCE_INLINE const b3Scalar& getZ() const { return m_floats[2]; }
+	/**@brief Return the w value */
+	B3_FORCE_INLINE const b3Scalar& getW() const { return m_floats[3]; }
 
-  /**@brief Set the x value */
-		B3_FORCE_INLINE void	setX(b3Scalar _x) { m_floats[0] = _x;};
-  /**@brief Set the y value */
-		B3_FORCE_INLINE void	setY(b3Scalar _y) { m_floats[1] = _y;};
-  /**@brief Set the z value */
-		B3_FORCE_INLINE void	setZ(b3Scalar _z) { m_floats[2] = _z;};
-  /**@brief Set the w value */
-		B3_FORCE_INLINE void	setW(b3Scalar _w) { m_floats[3] = _w;};
+	/**@brief Set the x value */
+	B3_FORCE_INLINE void setX(b3Scalar _x) { m_floats[0] = _x; };
+	/**@brief Set the y value */
+	B3_FORCE_INLINE void setY(b3Scalar _y) { m_floats[1] = _y; };
+	/**@brief Set the z value */
+	B3_FORCE_INLINE void setZ(b3Scalar _z) { m_floats[2] = _z; };
+	/**@brief Set the w value */
+	B3_FORCE_INLINE void setW(b3Scalar _w) { m_floats[3] = _w; };
 
 	//B3_FORCE_INLINE b3Scalar&       operator[](int i)       { return (&m_floats[0])[i];	}
 	//B3_FORCE_INLINE const b3Scalar& operator[](int i) const { return (&m_floats[0])[i]; }
 	///operator b3Scalar*() replaces operator[], using implicit conversion. We added operator != and operator == to avoid pointer comparisons.
-	B3_FORCE_INLINE	operator       b3Scalar *()       { return &m_floats[0]; }
-	B3_FORCE_INLINE	operator const b3Scalar *() const { return &m_floats[0]; }
+	B3_FORCE_INLINE operator b3Scalar*() { return &m_floats[0]; }
+	B3_FORCE_INLINE operator const b3Scalar*() const { return &m_floats[0]; }
 
-	B3_FORCE_INLINE	bool	operator==(const b3Vector3& other) const
+	B3_FORCE_INLINE bool operator==(const b3Vector3& other) const
 	{
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
-        return (0xf == _mm_movemask_ps((__m128)_mm_cmpeq_ps(mVec128, other.mVec128)));
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
+		return (0xf == _mm_movemask_ps((__m128)_mm_cmpeq_ps(mVec128, other.mVec128)));
 #else
-		return ((m_floats[3]==other.m_floats[3]) &&
-                (m_floats[2]==other.m_floats[2]) &&
-                (m_floats[1]==other.m_floats[1]) &&
-                (m_floats[0]==other.m_floats[0]));
+		return ((m_floats[3] == other.m_floats[3]) &&
+				(m_floats[2] == other.m_floats[2]) &&
+				(m_floats[1] == other.m_floats[1]) &&
+				(m_floats[0] == other.m_floats[0]));
 #endif
 	}
 
-	B3_FORCE_INLINE	bool	operator!=(const b3Vector3& other) const
+	B3_FORCE_INLINE bool operator!=(const b3Vector3& other) const
 	{
 		return !(*this == other);
 	}
 
-  /**@brief Set each element to the max of the current values and the values of another b3Vector3
+	/**@brief Set each element to the max of the current values and the values of another b3Vector3
    * @param other The other b3Vector3 to compare with
    */
-	B3_FORCE_INLINE void	setMax(const b3Vector3& other)
+	B3_FORCE_INLINE void setMax(const b3Vector3& other)
 	{
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 		mVec128 = _mm_max_ps(mVec128, other.mVec128);
 #elif defined(B3_USE_NEON)
 		mVec128 = vmaxq_f32(mVec128, other.mVec128);
@@ -581,12 +570,12 @@ public:
 #endif
 	}
 
-  /**@brief Set each element to the min of the current values and the values of another b3Vector3
+	/**@brief Set each element to the min of the current values and the values of another b3Vector3
    * @param other The other b3Vector3 to compare with
    */
-	B3_FORCE_INLINE void	setMin(const b3Vector3& other)
+	B3_FORCE_INLINE void setMin(const b3Vector3& other)
 	{
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 		mVec128 = _mm_min_ps(mVec128, other.mVec128);
 #elif defined(B3_USE_NEON)
 		mVec128 = vminq_f32(mVec128, other.mVec128);
@@ -598,46 +587,46 @@ public:
 #endif
 	}
 
-	B3_FORCE_INLINE void 	setValue(const b3Scalar& _x, const b3Scalar& _y, const b3Scalar& _z)
+	B3_FORCE_INLINE void setValue(const b3Scalar& _x, const b3Scalar& _y, const b3Scalar& _z)
 	{
-		m_floats[0]=_x;
-		m_floats[1]=_y;
-		m_floats[2]=_z;
+		m_floats[0] = _x;
+		m_floats[1] = _y;
+		m_floats[2] = _z;
 		m_floats[3] = b3Scalar(0.f);
 	}
 
-	void	getSkewSymmetricMatrix(b3Vector3* v0,b3Vector3* v1,b3Vector3* v2) const
+	void getSkewSymmetricMatrix(b3Vector3 * v0, b3Vector3 * v1, b3Vector3 * v2) const
 	{
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 
-		__m128 V  = _mm_and_ps(mVec128, b3vFFF0fMask);
+		__m128 V = _mm_and_ps(mVec128, b3vFFF0fMask);
 		__m128 V0 = _mm_xor_ps(b3vMzeroMask, V);
 		__m128 V2 = _mm_movelh_ps(V0, V);
 
 		__m128 V1 = _mm_shuffle_ps(V, V0, 0xCE);
 
-        V0 = _mm_shuffle_ps(V0, V, 0xDB);
+		V0 = _mm_shuffle_ps(V0, V, 0xDB);
 		V2 = _mm_shuffle_ps(V2, V, 0xF9);
 
 		v0->mVec128 = V0;
 		v1->mVec128 = V1;
 		v2->mVec128 = V2;
 #else
-		v0->setValue(0.		,-getZ()		,getY());
-		v1->setValue(getZ()	,0.			,-getX());
-		v2->setValue(-getY()	,getX()	,0.);
+		v0->setValue(0., -getZ(), getY());
+		v1->setValue(getZ(), 0., -getX());
+		v2->setValue(-getY(), getX(), 0.);
 #endif
 	}
 
 	void setZero()
 	{
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 		mVec128 = (__m128)_mm_xor_ps(mVec128, mVec128);
 #elif defined(B3_USE_NEON)
 		int32x4_t vi = vdupq_n_s32(0);
 		mVec128 = vreinterpretq_f32_s32(vi);
 #else
-		setValue(b3Scalar(0.),b3Scalar(0.),b3Scalar(0.));
+		setValue(b3Scalar(0.), b3Scalar(0.), b3Scalar(0.));
 #endif
 	}
 
@@ -651,76 +640,76 @@ public:
 		return length2() < B3_EPSILON;
 	}
 
-	B3_FORCE_INLINE	void	serialize(struct	b3Vector3Data& dataOut) const;
+	B3_FORCE_INLINE void serialize(struct b3Vector3Data & dataOut) const;
 
-	B3_FORCE_INLINE	void	deSerialize(const struct	b3Vector3Data& dataIn);
+	B3_FORCE_INLINE void deSerialize(const struct b3Vector3Data& dataIn);
 
-	B3_FORCE_INLINE	void	serializeFloat(struct	b3Vector3FloatData& dataOut) const;
+	B3_FORCE_INLINE void serializeFloat(struct b3Vector3FloatData & dataOut) const;
 
-	B3_FORCE_INLINE	void	deSerializeFloat(const struct	b3Vector3FloatData& dataIn);
+	B3_FORCE_INLINE void deSerializeFloat(const struct b3Vector3FloatData& dataIn);
 
-	B3_FORCE_INLINE	void	serializeDouble(struct	b3Vector3DoubleData& dataOut) const;
+	B3_FORCE_INLINE void serializeDouble(struct b3Vector3DoubleData & dataOut) const;
 
-	B3_FORCE_INLINE	void	deSerializeDouble(const struct	b3Vector3DoubleData& dataIn);
+	B3_FORCE_INLINE void deSerializeDouble(const struct b3Vector3DoubleData& dataIn);
 
-        /**@brief returns index of maximum dot product between this and vectors in array[]
+	/**@brief returns index of maximum dot product between this and vectors in array[]
          * @param array The other vectors
          * @param array_count The number of other vectors
          * @param dotOut The maximum dot product */
-        B3_FORCE_INLINE   long    maxDot( const b3Vector3 *array, long array_count, b3Scalar &dotOut ) const;
+	B3_FORCE_INLINE long maxDot(const b3Vector3* array, long array_count, b3Scalar& dotOut) const;
 
-        /**@brief returns index of minimum dot product between this and vectors in array[]
+	/**@brief returns index of minimum dot product between this and vectors in array[]
          * @param array The other vectors
          * @param array_count The number of other vectors
          * @param dotOut The minimum dot product */
-        B3_FORCE_INLINE   long    minDot( const b3Vector3 *array, long array_count, b3Scalar &dotOut ) const;
+	B3_FORCE_INLINE long minDot(const b3Vector3* array, long array_count, b3Scalar& dotOut) const;
 
-    /* create a vector as  b3Vector3( this->dot( b3Vector3 v0 ), this->dot( b3Vector3 v1), this->dot( b3Vector3 v2 ))  */
-    B3_FORCE_INLINE b3Vector3  dot3( const b3Vector3 &v0, const b3Vector3 &v1, const b3Vector3 &v2 ) const
-    {
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+	/* create a vector as  b3Vector3( this->dot( b3Vector3 v0 ), this->dot( b3Vector3 v1), this->dot( b3Vector3 v2 ))  */
+	B3_FORCE_INLINE b3Vector3 dot3(const b3Vector3& v0, const b3Vector3& v1, const b3Vector3& v2) const
+	{
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 
-        __m128 a0 = _mm_mul_ps( v0.mVec128, this->mVec128 );
-        __m128 a1 = _mm_mul_ps( v1.mVec128, this->mVec128 );
-        __m128 a2 = _mm_mul_ps( v2.mVec128, this->mVec128 );
-        __m128 b0 = _mm_unpacklo_ps( a0, a1 );
-        __m128 b1 = _mm_unpackhi_ps( a0, a1 );
-        __m128 b2 = _mm_unpacklo_ps( a2, _mm_setzero_ps() );
-        __m128 r = _mm_movelh_ps( b0, b2 );
-        r = _mm_add_ps( r, _mm_movehl_ps( b2, b0 ));
-        a2 = _mm_and_ps( a2, b3vxyzMaskf);
-        r = _mm_add_ps( r, b3CastdTo128f (_mm_move_sd( b3CastfTo128d(a2), b3CastfTo128d(b1) )));
-        return b3MakeVector3(r);
+		__m128 a0 = _mm_mul_ps(v0.mVec128, this->mVec128);
+		__m128 a1 = _mm_mul_ps(v1.mVec128, this->mVec128);
+		__m128 a2 = _mm_mul_ps(v2.mVec128, this->mVec128);
+		__m128 b0 = _mm_unpacklo_ps(a0, a1);
+		__m128 b1 = _mm_unpackhi_ps(a0, a1);
+		__m128 b2 = _mm_unpacklo_ps(a2, _mm_setzero_ps());
+		__m128 r = _mm_movelh_ps(b0, b2);
+		r = _mm_add_ps(r, _mm_movehl_ps(b2, b0));
+		a2 = _mm_and_ps(a2, b3vxyzMaskf);
+		r = _mm_add_ps(r, b3CastdTo128f(_mm_move_sd(b3CastfTo128d(a2), b3CastfTo128d(b1))));
+		return b3MakeVector3(r);
 
 #elif defined(B3_USE_NEON)
-        static const uint32x4_t xyzMask = (const uint32x4_t){ -1, -1, -1, 0 };
-        float32x4_t a0 = vmulq_f32( v0.mVec128, this->mVec128);
-        float32x4_t a1 = vmulq_f32( v1.mVec128, this->mVec128);
-        float32x4_t a2 = vmulq_f32( v2.mVec128, this->mVec128);
-        float32x2x2_t zLo = vtrn_f32( vget_high_f32(a0), vget_high_f32(a1));
-        a2 = (float32x4_t) vandq_u32((uint32x4_t) a2, xyzMask );
-        float32x2_t b0 = vadd_f32( vpadd_f32( vget_low_f32(a0), vget_low_f32(a1)), zLo.val[0] );
-        float32x2_t b1 = vpadd_f32( vpadd_f32( vget_low_f32(a2), vget_high_f32(a2)), vdup_n_f32(0.0f));
-        return b3Vector3( vcombine_f32(b0, b1) );
+		static const uint32x4_t xyzMask = (const uint32x4_t){-1, -1, -1, 0};
+		float32x4_t a0 = vmulq_f32(v0.mVec128, this->mVec128);
+		float32x4_t a1 = vmulq_f32(v1.mVec128, this->mVec128);
+		float32x4_t a2 = vmulq_f32(v2.mVec128, this->mVec128);
+		float32x2x2_t zLo = vtrn_f32(vget_high_f32(a0), vget_high_f32(a1));
+		a2 = (float32x4_t)vandq_u32((uint32x4_t)a2, xyzMask);
+		float32x2_t b0 = vadd_f32(vpadd_f32(vget_low_f32(a0), vget_low_f32(a1)), zLo.val[0]);
+		float32x2_t b1 = vpadd_f32(vpadd_f32(vget_low_f32(a2), vget_high_f32(a2)), vdup_n_f32(0.0f));
+		return b3Vector3(vcombine_f32(b0, b1));
 #else
-		return b3MakeVector3( dot(v0), dot(v1), dot(v2));
+		return b3MakeVector3(dot(v0), dot(v1), dot(v2));
 #endif
-    }
+	}
 };
 
 /**@brief Return the sum of two vectors (Point symantics)*/
 B3_FORCE_INLINE b3Vector3
 operator+(const b3Vector3& v1, const b3Vector3& v2)
 {
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 	return b3MakeVector3(_mm_add_ps(v1.mVec128, v2.mVec128));
 #elif defined(B3_USE_NEON)
 	return b3MakeVector3(vaddq_f32(v1.mVec128, v2.mVec128));
 #else
 	return b3MakeVector3(
-			v1.m_floats[0] + v2.m_floats[0],
-			v1.m_floats[1] + v2.m_floats[1],
-			v1.m_floats[2] + v2.m_floats[2]);
+		v1.m_floats[0] + v2.m_floats[0],
+		v1.m_floats[1] + v2.m_floats[1],
+		v1.m_floats[2] + v2.m_floats[2]);
 #endif
 }
 
@@ -728,15 +717,15 @@ operator+(const b3Vector3& v1, const b3Vector3& v2)
 B3_FORCE_INLINE b3Vector3
 operator*(const b3Vector3& v1, const b3Vector3& v2)
 {
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 	return b3MakeVector3(_mm_mul_ps(v1.mVec128, v2.mVec128));
 #elif defined(B3_USE_NEON)
 	return b3MakeVector3(vmulq_f32(v1.mVec128, v2.mVec128));
 #else
 	return b3MakeVector3(
-			v1.m_floats[0] * v2.m_floats[0],
-			v1.m_floats[1] * v2.m_floats[1],
-			v1.m_floats[2] * v2.m_floats[2]);
+		v1.m_floats[0] * v2.m_floats[0],
+		v1.m_floats[1] * v2.m_floats[1],
+		v1.m_floats[2] * v2.m_floats[2]);
 #endif
 }
 
@@ -744,7 +733,7 @@ operator*(const b3Vector3& v1, const b3Vector3& v2)
 B3_FORCE_INLINE b3Vector3
 operator-(const b3Vector3& v1, const b3Vector3& v2)
 {
-#if (defined(B3_USE_SSE_IN_API)  && defined(B3_USE_SSE))
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE))
 
 	//	without _mm_and_ps this code causes slowdown in Concave moving
 	__m128 r = _mm_sub_ps(v1.mVec128, v2.mVec128);
@@ -754,9 +743,9 @@ operator-(const b3Vector3& v1, const b3Vector3& v2)
 	return b3MakeVector3((float32x4_t)vandq_s32((int32x4_t)r, b3vFFF0Mask));
 #else
 	return b3MakeVector3(
-			v1.m_floats[0] - v2.m_floats[0],
-			v1.m_floats[1] - v2.m_floats[1],
-			v1.m_floats[2] - v2.m_floats[2]);
+		v1.m_floats[0] - v2.m_floats[0],
+		v1.m_floats[1] - v2.m_floats[1],
+		v1.m_floats[2] - v2.m_floats[2]);
 #endif
 }
 
@@ -764,7 +753,7 @@ operator-(const b3Vector3& v1, const b3Vector3& v2)
 B3_FORCE_INLINE b3Vector3
 operator-(const b3Vector3& v)
 {
-#if (defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE))
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE))
 	__m128 r = _mm_xor_ps(v.mVec128, b3vMzeroMask);
 	return b3MakeVector3(_mm_and_ps(r, b3vFFF0fMask));
 #elif defined(B3_USE_NEON)
@@ -778,9 +767,9 @@ operator-(const b3Vector3& v)
 B3_FORCE_INLINE b3Vector3
 operator*(const b3Vector3& v, const b3Scalar& s)
 {
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
-	__m128	vs = _mm_load_ss(&s);	//	(S 0 0 0)
-	vs = b3_pshufd_ps(vs, 0x80);	//	(S S S 0.0)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
+	__m128 vs = _mm_load_ss(&s);  //	(S 0 0 0)
+	vs = b3_pshufd_ps(vs, 0x80);  //	(S S S 0.0)
 	return b3MakeVector3(_mm_mul_ps(v.mVec128, vs));
 #elif defined(B3_USE_NEON)
 	float32x4_t r = vmulq_n_f32(v.mVec128, s);
@@ -802,7 +791,7 @@ B3_FORCE_INLINE b3Vector3
 operator/(const b3Vector3& v, const b3Scalar& s)
 {
 	b3FullAssert(s != b3Scalar(0.0));
-#if 0 //defined(B3_USE_SSE_IN_API)
+#if 0  //defined(B3_USE_SSE_IN_API)
 // this code is not faster !
 	__m128 vs = _mm_load_ss(&s);
     vs = _mm_div_ss(b3v1110, vs);
@@ -818,7 +807,7 @@ operator/(const b3Vector3& v, const b3Scalar& s)
 B3_FORCE_INLINE b3Vector3
 operator/(const b3Vector3& v1, const b3Vector3& v2)
 {
-#if (defined(B3_USE_SSE_IN_API)&& defined (B3_USE_SSE))
+#if (defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE))
 	__m128 vec = _mm_div_ps(v1.mVec128, v2.mVec128);
 	vec = _mm_and_ps(vec, b3vFFF0fMask);
 	return b3MakeVector3(vec);
@@ -828,19 +817,19 @@ operator/(const b3Vector3& v1, const b3Vector3& v2)
 	x = v1.mVec128;
 	y = v2.mVec128;
 
-	v = vrecpeq_f32(y);			// v ~ 1/y
-	m = vrecpsq_f32(y, v);		// m = (2-v*y)
-	v = vmulq_f32(v, m);		// vv = v*m ~~ 1/y
-	m = vrecpsq_f32(y, v);		// mm = (2-vv*y)
-	v = vmulq_f32(v, x);		// x*vv
-	v = vmulq_f32(v, m);		// (x*vv)*(2-vv*y) = x*(vv(2-vv*y)) ~~~ x/y
+	v = vrecpeq_f32(y);     // v ~ 1/y
+	m = vrecpsq_f32(y, v);  // m = (2-v*y)
+	v = vmulq_f32(v, m);    // vv = v*m ~~ 1/y
+	m = vrecpsq_f32(y, v);  // mm = (2-vv*y)
+	v = vmulq_f32(v, x);    // x*vv
+	v = vmulq_f32(v, m);    // (x*vv)*(2-vv*y) = x*(vv(2-vv*y)) ~~~ x/y
 
 	return b3Vector3(v);
 #else
 	return b3MakeVector3(
-			v1.m_floats[0] / v2.m_floats[0],
-			v1.m_floats[1] / v2.m_floats[1],
-			v1.m_floats[2] / v2.m_floats[2]);
+		v1.m_floats[0] / v2.m_floats[0],
+		v1.m_floats[1] / v2.m_floats[1],
+		v1.m_floats[2] / v2.m_floats[2]);
 #endif
 }
 
@@ -851,7 +840,6 @@ b3Dot(const b3Vector3& v1, const b3Vector3& v2)
 	return v1.dot(v2);
 }
 
-
 /**@brief Return the distance squared between two vectors */
 B3_FORCE_INLINE b3Scalar
 b3Distance2(const b3Vector3& v1, const b3Vector3& v2)
@@ -859,7 +847,6 @@ b3Distance2(const b3Vector3& v1, const b3Vector3& v2)
 	return v1.distance2(v2);
 }
 
-
 /**@brief Return the distance between two vectors */
 B3_FORCE_INLINE b3Scalar
 b3Distance(const b3Vector3& v1, const b3Vector3& v2)
@@ -897,8 +884,6 @@ b3Lerp(const b3Vector3& v1, const b3Vector3& v2, const b3Scalar& t)
 	return v1.lerp(v2, t);
 }
 
-
-
 B3_FORCE_INLINE b3Scalar b3Vector3::distance2(const b3Vector3& v) const
 {
 	return (v - *this).length2();
@@ -911,7 +896,7 @@ B3_FORCE_INLINE b3Scalar b3Vector3::distance(const b3Vector3& v) const
 
 B3_FORCE_INLINE b3Vector3 b3Vector3::normalized() const
 {
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 	b3Vector3 norm = *this;
 
 	return norm.normalize();
@@ -920,143 +905,136 @@ B3_FORCE_INLINE b3Vector3 b3Vector3::normalized() const
 #endif
 }
 
-B3_FORCE_INLINE b3Vector3 b3Vector3::rotate( const b3Vector3& wAxis, const b3Scalar _angle ) const
+B3_FORCE_INLINE b3Vector3 b3Vector3::rotate(const b3Vector3& wAxis, const b3Scalar _angle) const
 {
 	// wAxis must be a unit lenght vector
 
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 
-    __m128 O = _mm_mul_ps(wAxis.mVec128, mVec128);
-	b3Scalar ssin = b3Sin( _angle );
-    __m128 C = wAxis.cross( b3MakeVector3(mVec128) ).mVec128;
+	__m128 O = _mm_mul_ps(wAxis.mVec128, mVec128);
+	b3Scalar ssin = b3Sin(_angle);
+	__m128 C = wAxis.cross(b3MakeVector3(mVec128)).mVec128;
 	O = _mm_and_ps(O, b3vFFF0fMask);
-    b3Scalar scos = b3Cos( _angle );
+	b3Scalar scos = b3Cos(_angle);
 
-	__m128 vsin = _mm_load_ss(&ssin);	//	(S 0 0 0)
-    __m128 vcos = _mm_load_ss(&scos);	//	(S 0 0 0)
+	__m128 vsin = _mm_load_ss(&ssin);  //	(S 0 0 0)
+	__m128 vcos = _mm_load_ss(&scos);  //	(S 0 0 0)
 
-	__m128 Y = b3_pshufd_ps(O, 0xC9);	//	(Y Z X 0)
-	__m128 Z = b3_pshufd_ps(O, 0xD2);	//	(Z X Y 0)
+	__m128 Y = b3_pshufd_ps(O, 0xC9);  //	(Y Z X 0)
+	__m128 Z = b3_pshufd_ps(O, 0xD2);  //	(Z X Y 0)
 	O = _mm_add_ps(O, Y);
-	vsin = b3_pshufd_ps(vsin, 0x80);	//	(S S S 0)
+	vsin = b3_pshufd_ps(vsin, 0x80);  //	(S S S 0)
 	O = _mm_add_ps(O, Z);
-    vcos = b3_pshufd_ps(vcos, 0x80);	//	(S S S 0)
+	vcos = b3_pshufd_ps(vcos, 0x80);  //	(S S S 0)
 
-    vsin = vsin * C;
+	vsin = vsin * C;
 	O = O * wAxis.mVec128;
 	__m128 X = mVec128 - O;
 
-    O = O + vsin;
+	O = O + vsin;
 	vcos = vcos * X;
 	O = O + vcos;
 
 	return b3MakeVector3(O);
 #else
-	b3Vector3 o = wAxis * wAxis.dot( *this );
+	b3Vector3 o = wAxis * wAxis.dot(*this);
 	b3Vector3 _x = *this - o;
 	b3Vector3 _y;
 
-	_y = wAxis.cross( *this );
+	_y = wAxis.cross(*this);
 
-	return ( o + _x * b3Cos( _angle ) + _y * b3Sin( _angle ) );
+	return (o + _x * b3Cos(_angle) + _y * b3Sin(_angle));
 #endif
 }
 
-B3_FORCE_INLINE   long    b3Vector3::maxDot( const b3Vector3 *array, long array_count, b3Scalar &dotOut ) const
+B3_FORCE_INLINE long b3Vector3::maxDot(const b3Vector3* array, long array_count, b3Scalar& dotOut) const
 {
-#if defined (B3_USE_SSE) || defined (B3_USE_NEON)
-    #if defined _WIN32 || defined (B3_USE_SSE)
-        const long scalar_cutoff = 10;
-        long b3_maxdot_large( const float *array, const float *vec, unsigned long array_count, float *dotOut );
-    #elif defined B3_USE_NEON
-        const long scalar_cutoff = 4;
-        extern long (*_maxdot_large)( const float *array, const float *vec, unsigned long array_count, float *dotOut );
-    #endif
-    if( array_count < scalar_cutoff )
+#if defined(B3_USE_SSE) || defined(B3_USE_NEON)
+#if defined _WIN32 || defined(B3_USE_SSE)
+	const long scalar_cutoff = 10;
+	long b3_maxdot_large(const float* array, const float* vec, unsigned long array_count, float* dotOut);
+#elif defined B3_USE_NEON
+	const long scalar_cutoff = 4;
+	extern long (*_maxdot_large)(const float* array, const float* vec, unsigned long array_count, float* dotOut);
+#endif
+	if (array_count < scalar_cutoff)
 #else
 
-#endif//B3_USE_SSE || B3_USE_NEON
-    {
-        b3Scalar maxDot = -B3_INFINITY;
-        int i = 0;
-        int ptIndex = -1;
-        for( i = 0; i < array_count; i++ )
-        {
-            b3Scalar dot = array[i].dot(*this);
+#endif  //B3_USE_SSE || B3_USE_NEON
+	{
+		b3Scalar maxDot = -B3_INFINITY;
+		int i = 0;
+		int ptIndex = -1;
+		for (i = 0; i < array_count; i++)
+		{
+			b3Scalar dot = array[i].dot(*this);
 
-            if( dot > maxDot )
-            {
-                maxDot = dot;
-                ptIndex = i;
-            }
-        }
+			if (dot > maxDot)
+			{
+				maxDot = dot;
+				ptIndex = i;
+			}
+		}
 
-		b3Assert(ptIndex>=0);
-        if (ptIndex<0)
+		b3Assert(ptIndex >= 0);
+		if (ptIndex < 0)
 		{
 			ptIndex = 0;
 		}
-        dotOut = maxDot;
-        return ptIndex;
-    }
-#if defined (B3_USE_SSE) || defined (B3_USE_NEON)
-    return b3_maxdot_large( (float*) array, (float*) &m_floats[0], array_count, &dotOut );
+		dotOut = maxDot;
+		return ptIndex;
+	}
+#if defined(B3_USE_SSE) || defined(B3_USE_NEON)
+	return b3_maxdot_large((float*)array, (float*)&m_floats[0], array_count, &dotOut);
 #endif
 }
 
-B3_FORCE_INLINE   long    b3Vector3::minDot( const b3Vector3 *array, long array_count, b3Scalar &dotOut ) const
+B3_FORCE_INLINE long b3Vector3::minDot(const b3Vector3* array, long array_count, b3Scalar& dotOut) const
 {
-#if defined (B3_USE_SSE) || defined (B3_USE_NEON)
-    #if defined B3_USE_SSE
-        const long scalar_cutoff = 10;
-        long b3_mindot_large( const float *array, const float *vec, unsigned long array_count, float *dotOut );
-    #elif defined B3_USE_NEON
-        const long scalar_cutoff = 4;
-        extern long (*b3_mindot_large)( const float *array, const float *vec, unsigned long array_count, float *dotOut );
-    #else
-        #error unhandled arch!
-    #endif
+#if defined(B3_USE_SSE) || defined(B3_USE_NEON)
+#if defined B3_USE_SSE
+	const long scalar_cutoff = 10;
+	long b3_mindot_large(const float* array, const float* vec, unsigned long array_count, float* dotOut);
+#elif defined B3_USE_NEON
+	const long scalar_cutoff = 4;
+	extern long (*b3_mindot_large)(const float* array, const float* vec, unsigned long array_count, float* dotOut);
+#else
+#error unhandled arch!
+#endif
 
-    if( array_count < scalar_cutoff )
-#endif//B3_USE_SSE || B3_USE_NEON
-    {
-        b3Scalar  minDot = B3_INFINITY;
-        int i = 0;
-        int ptIndex = -1;
+	if (array_count < scalar_cutoff)
+#endif  //B3_USE_SSE || B3_USE_NEON
+	{
+		b3Scalar minDot = B3_INFINITY;
+		int i = 0;
+		int ptIndex = -1;
 
-        for( i = 0; i < array_count; i++ )
-        {
-            b3Scalar dot = array[i].dot(*this);
+		for (i = 0; i < array_count; i++)
+		{
+			b3Scalar dot = array[i].dot(*this);
 
-            if( dot < minDot )
-            {
-                minDot = dot;
-                ptIndex = i;
-            }
-        }
+			if (dot < minDot)
+			{
+				minDot = dot;
+				ptIndex = i;
+			}
+		}
 
-        dotOut = minDot;
+		dotOut = minDot;
 
-        return ptIndex;
-    }
-#if defined (B3_USE_SSE) || defined (B3_USE_NEON)
-    return b3_mindot_large( (float*) array, (float*) &m_floats[0], array_count, &dotOut );
+		return ptIndex;
+	}
+#if defined(B3_USE_SSE) || defined(B3_USE_NEON)
+	return b3_mindot_large((float*)array, (float*)&m_floats[0], array_count, &dotOut);
 #endif
 }
 
-
 class b3Vector4 : public b3Vector3
 {
 public:
-
-
-
-
-
-
 	B3_FORCE_INLINE b3Vector4 absolute4() const
 	{
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 		return b3MakeVector4(_mm_and_ps(mVec128, b3vAbsfMask));
 #elif defined(B3_USE_NEON)
 		return b3Vector4(vabsq_f32(mVec128));
@@ -1069,11 +1047,9 @@ public:
 #endif
 	}
 
+	b3Scalar getW() const { return m_floats[3]; }
 
-	b3Scalar	getW() const { return m_floats[3];}
-
-
-		B3_FORCE_INLINE int maxAxis4() const
+	B3_FORCE_INLINE int maxAxis4() const
 	{
 		int maxIndex = -1;
 		b3Scalar maxVal = b3Scalar(-B3_LARGE_FLOAT);
@@ -1090,7 +1066,7 @@ public:
 		if (m_floats[2] > maxVal)
 		{
 			maxIndex = 2;
-			maxVal =m_floats[2];
+			maxVal = m_floats[2];
 		}
 		if (m_floats[3] > maxVal)
 		{
@@ -1100,7 +1076,6 @@ public:
 		return maxIndex;
 	}
 
-
 	B3_FORCE_INLINE int minAxis4() const
 	{
 		int minIndex = -1;
@@ -1118,7 +1093,7 @@ public:
 		if (m_floats[2] < minVal)
 		{
 			minIndex = 2;
-			minVal =m_floats[2];
+			minVal = m_floats[2];
 		}
 		if (m_floats[3] < minVal)
 		{
@@ -1129,216 +1104,200 @@ public:
 		return minIndex;
 	}
 
-
 	B3_FORCE_INLINE int closestAxis4() const
 	{
 		return absolute4().maxAxis4();
 	}
 
-
-
-
-  /**@brief Set x,y,z and zero w
+	/**@brief Set x,y,z and zero w
    * @param x Value of x
    * @param y Value of y
    * @param z Value of z
    */
 
-
-/*		void getValue(b3Scalar *m) const
+	/*		void getValue(b3Scalar *m) const
 		{
 			m[0] = m_floats[0];
 			m[1] = m_floats[1];
 			m[2] =m_floats[2];
 		}
 */
-/**@brief Set the values
+	/**@brief Set the values
    * @param x Value of x
    * @param y Value of y
    * @param z Value of z
    * @param w Value of w
    */
-		B3_FORCE_INLINE void	setValue(const b3Scalar& _x, const b3Scalar& _y, const b3Scalar& _z,const b3Scalar& _w)
-		{
-			m_floats[0]=_x;
-			m_floats[1]=_y;
-			m_floats[2]=_z;
-			m_floats[3]=_w;
-		}
-
-
+	B3_FORCE_INLINE void setValue(const b3Scalar& _x, const b3Scalar& _y, const b3Scalar& _z, const b3Scalar& _w)
+	{
+		m_floats[0] = _x;
+		m_floats[1] = _y;
+		m_floats[2] = _z;
+		m_floats[3] = _w;
+	}
 };
 
-
 ///b3SwapVector3Endian swaps vector endianness, useful for network and cross-platform serialization
-B3_FORCE_INLINE void	b3SwapScalarEndian(const b3Scalar& sourceVal, b3Scalar& destVal)
+B3_FORCE_INLINE void b3SwapScalarEndian(const b3Scalar& sourceVal, b3Scalar& destVal)
 {
-	#ifdef B3_USE_DOUBLE_PRECISION
-	unsigned char* dest = (unsigned char*) &destVal;
-	unsigned char* src  = (unsigned char*) &sourceVal;
+#ifdef B3_USE_DOUBLE_PRECISION
+	unsigned char* dest = (unsigned char*)&destVal;
+	unsigned char* src = (unsigned char*)&sourceVal;
 	dest[0] = src[7];
-    dest[1] = src[6];
-    dest[2] = src[5];
-    dest[3] = src[4];
-    dest[4] = src[3];
-    dest[5] = src[2];
-    dest[6] = src[1];
-    dest[7] = src[0];
+	dest[1] = src[6];
+	dest[2] = src[5];
+	dest[3] = src[4];
+	dest[4] = src[3];
+	dest[5] = src[2];
+	dest[6] = src[1];
+	dest[7] = src[0];
 #else
-	unsigned char* dest = (unsigned char*) &destVal;
-	unsigned char* src  = (unsigned char*) &sourceVal;
+	unsigned char* dest = (unsigned char*)&destVal;
+	unsigned char* src = (unsigned char*)&sourceVal;
 	dest[0] = src[3];
-    dest[1] = src[2];
-    dest[2] = src[1];
-    dest[3] = src[0];
-#endif //B3_USE_DOUBLE_PRECISION
+	dest[1] = src[2];
+	dest[2] = src[1];
+	dest[3] = src[0];
+#endif  //B3_USE_DOUBLE_PRECISION
 }
 ///b3SwapVector3Endian swaps vector endianness, useful for network and cross-platform serialization
-B3_FORCE_INLINE void	b3SwapVector3Endian(const b3Vector3& sourceVec, b3Vector3& destVec)
+B3_FORCE_INLINE void b3SwapVector3Endian(const b3Vector3& sourceVec, b3Vector3& destVec)
 {
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 	{
-		b3SwapScalarEndian(sourceVec[i],destVec[i]);
+		b3SwapScalarEndian(sourceVec[i], destVec[i]);
 	}
-
 }
 
 ///b3UnSwapVector3Endian swaps vector endianness, useful for network and cross-platform serialization
-B3_FORCE_INLINE void	b3UnSwapVector3Endian(b3Vector3& vector)
+B3_FORCE_INLINE void b3UnSwapVector3Endian(b3Vector3& vector)
 {
-
-	b3Vector3	swappedVec;
-	for (int i=0;i<4;i++)
+	b3Vector3 swappedVec;
+	for (int i = 0; i < 4; i++)
 	{
-		b3SwapScalarEndian(vector[i],swappedVec[i]);
+		b3SwapScalarEndian(vector[i], swappedVec[i]);
 	}
 	vector = swappedVec;
 }
 
 template <class T>
-B3_FORCE_INLINE void b3PlaneSpace1 (const T& n, T& p, T& q)
+B3_FORCE_INLINE void b3PlaneSpace1(const T& n, T& p, T& q)
 {
-  if (b3Fabs(n[2]) > B3_SQRT12) {
-    // choose p in y-z plane
-    b3Scalar a = n[1]*n[1] + n[2]*n[2];
-    b3Scalar k = b3RecipSqrt (a);
-    p[0] = 0;
-	p[1] = -n[2]*k;
-	p[2] = n[1]*k;
-    // set q = n x p
-    q[0] = a*k;
-	q[1] = -n[0]*p[2];
-	q[2] = n[0]*p[1];
-  }
-  else {
-    // choose p in x-y plane
-    b3Scalar a = n[0]*n[0] + n[1]*n[1];
-    b3Scalar k = b3RecipSqrt (a);
-    p[0] = -n[1]*k;
-	p[1] = n[0]*k;
-	p[2] = 0;
-    // set q = n x p
-    q[0] = -n[2]*p[1];
-	q[1] = n[2]*p[0];
-	q[2] = a*k;
-  }
+	if (b3Fabs(n[2]) > B3_SQRT12)
+	{
+		// choose p in y-z plane
+		b3Scalar a = n[1] * n[1] + n[2] * n[2];
+		b3Scalar k = b3RecipSqrt(a);
+		p[0] = 0;
+		p[1] = -n[2] * k;
+		p[2] = n[1] * k;
+		// set q = n x p
+		q[0] = a * k;
+		q[1] = -n[0] * p[2];
+		q[2] = n[0] * p[1];
+	}
+	else
+	{
+		// choose p in x-y plane
+		b3Scalar a = n[0] * n[0] + n[1] * n[1];
+		b3Scalar k = b3RecipSqrt(a);
+		p[0] = -n[1] * k;
+		p[1] = n[0] * k;
+		p[2] = 0;
+		// set q = n x p
+		q[0] = -n[2] * p[1];
+		q[1] = n[2] * p[0];
+		q[2] = a * k;
+	}
 }
 
-
-struct	b3Vector3FloatData
+struct b3Vector3FloatData
 {
-	float	m_floats[4];
+	float m_floats[4];
 };
 
-struct	b3Vector3DoubleData
+struct b3Vector3DoubleData
 {
-	double	m_floats[4];
-
+	double m_floats[4];
 };
 
-B3_FORCE_INLINE	void	b3Vector3::serializeFloat(struct	b3Vector3FloatData& dataOut) const
+B3_FORCE_INLINE void b3Vector3::serializeFloat(struct b3Vector3FloatData& dataOut) const
 {
 	///could also do a memcpy, check if it is worth it
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 		dataOut.m_floats[i] = float(m_floats[i]);
 }
 
-B3_FORCE_INLINE void	b3Vector3::deSerializeFloat(const struct	b3Vector3FloatData& dataIn)
+B3_FORCE_INLINE void b3Vector3::deSerializeFloat(const struct b3Vector3FloatData& dataIn)
 {
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 		m_floats[i] = b3Scalar(dataIn.m_floats[i]);
 }
 
-
-B3_FORCE_INLINE	void	b3Vector3::serializeDouble(struct	b3Vector3DoubleData& dataOut) const
+B3_FORCE_INLINE void b3Vector3::serializeDouble(struct b3Vector3DoubleData& dataOut) const
 {
 	///could also do a memcpy, check if it is worth it
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 		dataOut.m_floats[i] = double(m_floats[i]);
 }
 
-B3_FORCE_INLINE void	b3Vector3::deSerializeDouble(const struct	b3Vector3DoubleData& dataIn)
+B3_FORCE_INLINE void b3Vector3::deSerializeDouble(const struct b3Vector3DoubleData& dataIn)
 {
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 		m_floats[i] = b3Scalar(dataIn.m_floats[i]);
 }
 
-
-B3_FORCE_INLINE	void	b3Vector3::serialize(struct	b3Vector3Data& dataOut) const
+B3_FORCE_INLINE void b3Vector3::serialize(struct b3Vector3Data& dataOut) const
 {
 	///could also do a memcpy, check if it is worth it
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 		dataOut.m_floats[i] = m_floats[i];
 }
 
-B3_FORCE_INLINE void	b3Vector3::deSerialize(const struct	b3Vector3Data& dataIn)
+B3_FORCE_INLINE void b3Vector3::deSerialize(const struct b3Vector3Data& dataIn)
 {
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 		m_floats[i] = dataIn.m_floats[i];
 }
 
-
-
-
-inline b3Vector3 b3MakeVector3(b3Scalar x,b3Scalar y,b3Scalar z)
+inline b3Vector3 b3MakeVector3(b3Scalar x, b3Scalar y, b3Scalar z)
 {
-	b3Vector3	tmp;
-	tmp.setValue(x,y,z);
+	b3Vector3 tmp;
+	tmp.setValue(x, y, z);
 	return tmp;
 }
 
-inline b3Vector3 b3MakeVector3(b3Scalar x,b3Scalar y,b3Scalar z, b3Scalar w)
+inline b3Vector3 b3MakeVector3(b3Scalar x, b3Scalar y, b3Scalar z, b3Scalar w)
 {
-	b3Vector3	tmp;
-	tmp.setValue(x,y,z);
+	b3Vector3 tmp;
+	tmp.setValue(x, y, z);
 	tmp.w = w;
 	return tmp;
 }
 
-inline b3Vector4 b3MakeVector4(b3Scalar x,b3Scalar y,b3Scalar z,b3Scalar w)
+inline b3Vector4 b3MakeVector4(b3Scalar x, b3Scalar y, b3Scalar z, b3Scalar w)
 {
-	b3Vector4	tmp;
-	tmp.setValue(x,y,z,w);
+	b3Vector4 tmp;
+	tmp.setValue(x, y, z, w);
 	return tmp;
 }
 
-#if defined(B3_USE_SSE_IN_API) && defined (B3_USE_SSE)
+#if defined(B3_USE_SSE_IN_API) && defined(B3_USE_SSE)
 
-inline b3Vector3 b3MakeVector3( b3SimdFloat4 v)
+inline b3Vector3 b3MakeVector3(b3SimdFloat4 v)
 {
-        b3Vector3 tmp;
-        tmp.set128(v);
-        return tmp;
+	b3Vector3 tmp;
+	tmp.set128(v);
+	return tmp;
 }
 
 inline b3Vector4 b3MakeVector4(b3SimdFloat4 vec)
 {
-	b3Vector4	tmp;
+	b3Vector4 tmp;
 	tmp.set128(vec);
 	return tmp;
 }
 
 #endif
 
-
-#endif //B3_VECTOR3_H
+#endif  //B3_VECTOR3_H
diff --git a/thirdparty/bullet/Bullet3Common/shared/b3Float4.h b/thirdparty/bullet/Bullet3Common/shared/b3Float4.h
index 5e4b95bcee1..d8a9f474113 100644
--- a/thirdparty/bullet/Bullet3Common/shared/b3Float4.h
+++ b/thirdparty/bullet/Bullet3Common/shared/b3Float4.h
@@ -4,94 +4,87 @@
 #include "Bullet3Common/shared/b3PlatformDefinitions.h"
 
 #ifdef __cplusplus
-	#include "Bullet3Common/b3Vector3.h"
-	#define b3Float4 b3Vector3
-	#define b3Float4ConstArg const b3Vector3&
-	#define b3Dot3F4 b3Dot
-	#define b3Cross3 b3Cross
-	#define	b3MakeFloat4  b3MakeVector3
-	inline b3Vector3 b3Normalized(const b3Vector3& vec)
-	{
-		return vec.normalized();
-	}
-
-	inline b3Float4 b3FastNormalized3(b3Float4ConstArg v)
-	{
-		return v.normalized();
-	}
-
-	inline b3Float4 b3MaxFloat4 (const b3Float4& a, const b3Float4& b)
-	{
-		b3Float4 tmp = a;
-		tmp.setMax(b);
-		return tmp;
-	}
-	inline b3Float4 b3MinFloat4 (const b3Float4& a, const b3Float4& b)
-	{
-		b3Float4 tmp = a;
-		tmp.setMin(b);
-		return tmp;
-	}
+#include "Bullet3Common/b3Vector3.h"
+#define b3Float4 b3Vector3
+#define b3Float4ConstArg const b3Vector3&
+#define b3Dot3F4 b3Dot
+#define b3Cross3 b3Cross
+#define b3MakeFloat4 b3MakeVector3
+inline b3Vector3 b3Normalized(const b3Vector3& vec)
+{
+	return vec.normalized();
+}
 
+inline b3Float4 b3FastNormalized3(b3Float4ConstArg v)
+{
+	return v.normalized();
+}
 
+inline b3Float4 b3MaxFloat4(const b3Float4& a, const b3Float4& b)
+{
+	b3Float4 tmp = a;
+	tmp.setMax(b);
+	return tmp;
+}
+inline b3Float4 b3MinFloat4(const b3Float4& a, const b3Float4& b)
+{
+	b3Float4 tmp = a;
+	tmp.setMin(b);
+	return tmp;
+}
 
 #else
-	typedef float4	b3Float4;
-	#define b3Float4ConstArg const b3Float4
-	#define b3MakeFloat4 (float4)
-	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)
-	{
-		float4 a1 = b3MakeFloat4(v0.xyz,0.f);
-		float4 b1 = b3MakeFloat4(v1.xyz,0.f);
-		return dot(a1, b1);
-	}
-	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)
-	{
-		float4 a1 = b3MakeFloat4(v0.xyz,0.f);
-		float4 b1 = b3MakeFloat4(v1.xyz,0.f);
-		return cross(a1, b1);
-	}
-	#define b3MinFloat4 min
-	#define b3MaxFloat4 max
+typedef float4 b3Float4;
+#define b3Float4ConstArg const b3Float4
+#define b3MakeFloat4 (float4)
+float b3Dot3F4(b3Float4ConstArg v0, b3Float4ConstArg v1)
+{
+	float4 a1 = b3MakeFloat4(v0.xyz, 0.f);
+	float4 b1 = b3MakeFloat4(v1.xyz, 0.f);
+	return dot(a1, b1);
+}
+b3Float4 b3Cross3(b3Float4ConstArg v0, b3Float4ConstArg v1)
+{
+	float4 a1 = b3MakeFloat4(v0.xyz, 0.f);
+	float4 b1 = b3MakeFloat4(v1.xyz, 0.f);
+	return cross(a1, b1);
+}
+#define b3MinFloat4 min
+#define b3MaxFloat4 max
 
-	#define b3Normalized(a) normalize(a)
+#define b3Normalized(a) normalize(a)
 
-#endif 
+#endif
 
-
-		
 inline bool b3IsAlmostZero(b3Float4ConstArg v)
 {
-	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	
+	if (b3Fabs(v.x) > 1e-6 || b3Fabs(v.y) > 1e-6 || b3Fabs(v.z) > 1e-6)
 		return false;
 	return true;
 }
 
-
-inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )
+inline int b3MaxDot(b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut)
 {
-    float maxDot = -B3_INFINITY;
-    int i = 0;
-    int ptIndex = -1;
-    for( i = 0; i < vecLen; i++ )
-    {
-        float dot = b3Dot3F4(vecArray[i],vec);
-            
-        if( dot > maxDot )
-        {
-            maxDot = dot;
-            ptIndex = i;
-        }
-    }
-	b3Assert(ptIndex>=0);
-    if (ptIndex<0)
+	float maxDot = -B3_INFINITY;
+	int i = 0;
+	int ptIndex = -1;
+	for (i = 0; i < vecLen; i++)
+	{
+		float dot = b3Dot3F4(vecArray[i], vec);
+
+		if (dot > maxDot)
+		{
+			maxDot = dot;
+			ptIndex = i;
+		}
+	}
+	b3Assert(ptIndex >= 0);
+	if (ptIndex < 0)
 	{
 		ptIndex = 0;
 	}
-    *dotOut = maxDot;
-    return ptIndex;
+	*dotOut = maxDot;
+	return ptIndex;
 }
 
-
-
-#endif //B3_FLOAT4_H
+#endif  //B3_FLOAT4_H
diff --git a/thirdparty/bullet/Bullet3Common/shared/b3Int2.h b/thirdparty/bullet/Bullet3Common/shared/b3Int2.h
index f1d01f81a57..7b84de4436d 100644
--- a/thirdparty/bullet/Bullet3Common/shared/b3Int2.h
+++ b/thirdparty/bullet/Bullet3Common/shared/b3Int2.h
@@ -20,11 +20,10 @@ subject to the following restrictions:
 
 struct b3UnsignedInt2
 {
-	union
-	{
+	union {
 		struct
 		{
-			unsigned int x,y;
+			unsigned int x, y;
 		};
 		struct
 		{
@@ -35,11 +34,10 @@ struct b3UnsignedInt2
 
 struct b3Int2
 {
-	union
-	{
+	union {
 		struct
 		{
-			int x,y;
+			int x, y;
 		};
 		struct
 		{
@@ -51,7 +49,8 @@ struct b3Int2
 inline b3Int2 b3MakeInt2(int x, int y)
 {
 	b3Int2 v;
-	v.s[0] = x; v.s[1] = y;
+	v.s[0] = x;
+	v.s[1] = y;
 	return v;
 }
 #else
@@ -60,5 +59,5 @@ inline b3Int2 b3MakeInt2(int x, int y)
 #define b3Int2 int2
 #define b3MakeInt2 (int2)
 
-#endif //__cplusplus
+#endif  //__cplusplus
 #endif
\ No newline at end of file
diff --git a/thirdparty/bullet/Bullet3Common/shared/b3Int4.h b/thirdparty/bullet/Bullet3Common/shared/b3Int4.h
index aa02d6beef9..f6a1754245c 100644
--- a/thirdparty/bullet/Bullet3Common/shared/b3Int4.h
+++ b/thirdparty/bullet/Bullet3Common/shared/b3Int4.h
@@ -5,16 +5,15 @@
 
 #include "Bullet3Common/b3Scalar.h"
 
-
-B3_ATTRIBUTE_ALIGNED16(struct) b3UnsignedInt4
+B3_ATTRIBUTE_ALIGNED16(struct)
+b3UnsignedInt4
 {
 	B3_DECLARE_ALIGNED_ALLOCATOR();
 
-	union
-	{
+	union {
 		struct
 		{
-			unsigned int x,y,z,w;
+			unsigned int x, y, z, w;
 		};
 		struct
 		{
@@ -23,15 +22,15 @@ B3_ATTRIBUTE_ALIGNED16(struct) b3UnsignedInt4
 	};
 };
 
-B3_ATTRIBUTE_ALIGNED16(struct) b3Int4
+B3_ATTRIBUTE_ALIGNED16(struct)
+b3Int4
 {
 	B3_DECLARE_ALIGNED_ALLOCATOR();
 
-	union
-	{
+	union {
 		struct
 		{
-			int x,y,z,w;
+			int x, y, z, w;
 		};
 		struct
 		{
@@ -43,26 +42,30 @@ B3_ATTRIBUTE_ALIGNED16(struct) b3Int4
 B3_FORCE_INLINE b3Int4 b3MakeInt4(int x, int y, int z, int w = 0)
 {
 	b3Int4 v;
-	v.s[0] = x; v.s[1] = y; v.s[2] = z; v.s[3] = w;
+	v.s[0] = x;
+	v.s[1] = y;
+	v.s[2] = z;
+	v.s[3] = w;
 	return v;
 }
 
 B3_FORCE_INLINE b3UnsignedInt4 b3MakeUnsignedInt4(unsigned int x, unsigned int y, unsigned int z, unsigned int w = 0)
 {
 	b3UnsignedInt4 v;
-	v.s[0] = x; v.s[1] = y; v.s[2] = z; v.s[3] = w;
+	v.s[0] = x;
+	v.s[1] = y;
+	v.s[2] = z;
+	v.s[3] = w;
 	return v;
 }
 
 #else
 
-
 #define b3UnsignedInt4 uint4
 #define b3Int4 int4
 #define b3MakeInt4 (int4)
 #define b3MakeUnsignedInt4 (uint4)
 
+#endif  //__cplusplus
 
-#endif //__cplusplus
-
-#endif //B3_INT4_H
+#endif  //B3_INT4_H
diff --git a/thirdparty/bullet/Bullet3Common/shared/b3Mat3x3.h b/thirdparty/bullet/Bullet3Common/shared/b3Mat3x3.h
index 7b1fef32f84..ce6482b5a6c 100644
--- a/thirdparty/bullet/Bullet3Common/shared/b3Mat3x3.h
+++ b/thirdparty/bullet/Bullet3Common/shared/b3Mat3x3.h
@@ -4,7 +4,6 @@
 
 #include "Bullet3Common/shared/b3Quat.h"
 
-
 #ifdef __cplusplus
 
 #include "Bullet3Common/b3Matrix3x3.h"
@@ -22,43 +21,41 @@ inline b3Mat3x3 b3AbsoluteMat3x3(b3Mat3x3ConstArg mat)
 	return mat.absolute();
 }
 
-#define b3GetRow(m,row) m.getRow(row)
+#define b3GetRow(m, row) m.getRow(row)
 
-__inline
-b3Float4 mtMul3(b3Float4ConstArg a, b3Mat3x3ConstArg b)
+__inline b3Float4 mtMul3(b3Float4ConstArg a, b3Mat3x3ConstArg b)
 {
-	return b*a;
+	return b * a;
 }
 
-
 #else
 
 typedef struct
 {
 	b3Float4 m_row[3];
-}b3Mat3x3;
+} b3Mat3x3;
 
 #define b3Mat3x3ConstArg const b3Mat3x3
-#define b3GetRow(m,row) (m.m_row[row])
+#define b3GetRow(m, row) (m.m_row[row])
 
 inline b3Mat3x3 b3QuatGetRotationMatrix(b3Quat quat)
 {
-	b3Float4 quat2 = (b3Float4)(quat.x*quat.x, quat.y*quat.y, quat.z*quat.z, 0.f);
+	b3Float4 quat2 = (b3Float4)(quat.x * quat.x, quat.y * quat.y, quat.z * quat.z, 0.f);
 	b3Mat3x3 out;
 
-	out.m_row[0].x=1-2*quat2.y-2*quat2.z;
-	out.m_row[0].y=2*quat.x*quat.y-2*quat.w*quat.z;
-	out.m_row[0].z=2*quat.x*quat.z+2*quat.w*quat.y;
+	out.m_row[0].x = 1 - 2 * quat2.y - 2 * quat2.z;
+	out.m_row[0].y = 2 * quat.x * quat.y - 2 * quat.w * quat.z;
+	out.m_row[0].z = 2 * quat.x * quat.z + 2 * quat.w * quat.y;
 	out.m_row[0].w = 0.f;
 
-	out.m_row[1].x=2*quat.x*quat.y+2*quat.w*quat.z;
-	out.m_row[1].y=1-2*quat2.x-2*quat2.z;
-	out.m_row[1].z=2*quat.y*quat.z-2*quat.w*quat.x;
+	out.m_row[1].x = 2 * quat.x * quat.y + 2 * quat.w * quat.z;
+	out.m_row[1].y = 1 - 2 * quat2.x - 2 * quat2.z;
+	out.m_row[1].z = 2 * quat.y * quat.z - 2 * quat.w * quat.x;
 	out.m_row[1].w = 0.f;
 
-	out.m_row[2].x=2*quat.x*quat.z-2*quat.w*quat.y;
-	out.m_row[2].y=2*quat.y*quat.z+2*quat.w*quat.x;
-	out.m_row[2].z=1-2*quat2.x-2*quat2.y;
+	out.m_row[2].x = 2 * quat.x * quat.z - 2 * quat.w * quat.y;
+	out.m_row[2].y = 2 * quat.y * quat.z + 2 * quat.w * quat.x;
+	out.m_row[2].z = 1 - 2 * quat2.x - 2 * quat2.y;
 	out.m_row[2].w = 0.f;
 
 	return out;
@@ -73,27 +70,19 @@ inline b3Mat3x3 b3AbsoluteMat3x3(b3Mat3x3ConstArg matIn)
 	return out;
 }
 
+__inline b3Mat3x3 mtZero();
 
-__inline
-b3Mat3x3 mtZero();
+__inline b3Mat3x3 mtIdentity();
 
-__inline
-b3Mat3x3 mtIdentity();
+__inline b3Mat3x3 mtTranspose(b3Mat3x3 m);
 
-__inline
-b3Mat3x3 mtTranspose(b3Mat3x3 m);
+__inline b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b);
 
-__inline
-b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b);
+__inline b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b);
 
-__inline
-b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b);
+__inline b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b);
 
-__inline
-b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b);
-
-__inline
-b3Mat3x3 mtZero()
+__inline b3Mat3x3 mtZero()
 {
 	b3Mat3x3 m;
 	m.m_row[0] = (b3Float4)(0.f);
@@ -102,18 +91,16 @@ b3Mat3x3 mtZero()
 	return m;
 }
 
-__inline
-b3Mat3x3 mtIdentity()
+__inline b3Mat3x3 mtIdentity()
 {
 	b3Mat3x3 m;
-	m.m_row[0] = (b3Float4)(1,0,0,0);
-	m.m_row[1] = (b3Float4)(0,1,0,0);
-	m.m_row[2] = (b3Float4)(0,0,1,0);
+	m.m_row[0] = (b3Float4)(1, 0, 0, 0);
+	m.m_row[1] = (b3Float4)(0, 1, 0, 0);
+	m.m_row[2] = (b3Float4)(0, 0, 1, 0);
 	return m;
 }
 
-__inline
-b3Mat3x3 mtTranspose(b3Mat3x3 m)
+__inline b3Mat3x3 mtTranspose(b3Mat3x3 m)
 {
 	b3Mat3x3 out;
 	out.m_row[0] = (b3Float4)(m.m_row[0].x, m.m_row[1].x, m.m_row[2].x, 0.f);
@@ -122,58 +109,49 @@ b3Mat3x3 mtTranspose(b3Mat3x3 m)
 	return out;
 }
 
-__inline
-b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b)
+__inline b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b)
 {
 	b3Mat3x3 transB;
-	transB = mtTranspose( b );
+	transB = mtTranspose(b);
 	b3Mat3x3 ans;
 	//	why this doesn't run when 0ing in the for{}
 	a.m_row[0].w = 0.f;
 	a.m_row[1].w = 0.f;
 	a.m_row[2].w = 0.f;
-	for(int i=0; i<3; i++)
+	for (int i = 0; i < 3; i++)
 	{
-//	a.m_row[i].w = 0.f;
-		ans.m_row[i].x = b3Dot3F4(a.m_row[i],transB.m_row[0]);
-		ans.m_row[i].y = b3Dot3F4(a.m_row[i],transB.m_row[1]);
-		ans.m_row[i].z = b3Dot3F4(a.m_row[i],transB.m_row[2]);
+		//	a.m_row[i].w = 0.f;
+		ans.m_row[i].x = b3Dot3F4(a.m_row[i], transB.m_row[0]);
+		ans.m_row[i].y = b3Dot3F4(a.m_row[i], transB.m_row[1]);
+		ans.m_row[i].z = b3Dot3F4(a.m_row[i], transB.m_row[2]);
 		ans.m_row[i].w = 0.f;
 	}
 	return ans;
 }
 
-__inline
-b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b)
+__inline b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b)
 {
 	b3Float4 ans;
-	ans.x = b3Dot3F4( a.m_row[0], b );
-	ans.y = b3Dot3F4( a.m_row[1], b );
-	ans.z = b3Dot3F4( a.m_row[2], b );
+	ans.x = b3Dot3F4(a.m_row[0], b);
+	ans.y = b3Dot3F4(a.m_row[1], b);
+	ans.z = b3Dot3F4(a.m_row[2], b);
 	ans.w = 0.f;
 	return ans;
 }
 
-__inline
-b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b)
+__inline b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b)
 {
 	b3Float4 colx = b3MakeFloat4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);
 	b3Float4 coly = b3MakeFloat4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);
 	b3Float4 colz = b3MakeFloat4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);
 
 	b3Float4 ans;
-	ans.x = b3Dot3F4( a, colx );
-	ans.y = b3Dot3F4( a, coly );
-	ans.z = b3Dot3F4( a, colz );
+	ans.x = b3Dot3F4(a, colx);
+	ans.y = b3Dot3F4(a, coly);
+	ans.z = b3Dot3F4(a, colz);
 	return ans;
 }
 
-
 #endif
 
-
-
-
-
-
-#endif //B3_MAT3x3_H
+#endif  //B3_MAT3x3_H
diff --git a/thirdparty/bullet/Bullet3Common/shared/b3PlatformDefinitions.h b/thirdparty/bullet/Bullet3Common/shared/b3PlatformDefinitions.h
index 1c133fb0889..b72bee93106 100644
--- a/thirdparty/bullet/Bullet3Common/shared/b3PlatformDefinitions.h
+++ b/thirdparty/bullet/Bullet3Common/shared/b3PlatformDefinitions.h
@@ -8,18 +8,18 @@ struct MyTest
 
 #ifdef __cplusplus
 //#define b3ConstArray(a) const b3AlignedObjectArray<a>&
-#define b3ConstArray(a) const a*
+#define b3ConstArray(a) const a *
 #define b3AtomicInc(a) ((*a)++)
 
-inline int b3AtomicAdd (volatile int *p, int val)
+inline int b3AtomicAdd(volatile int *p, int val)
 {
 	int oldValue = *p;
-	int newValue = oldValue+val;
+	int newValue = oldValue + val;
 	*p = newValue;
 	return oldValue;
 }
 
-#define __global 
+#define __global
 
 #define B3_STATIC static
 #else
@@ -27,7 +27,7 @@ inline int b3AtomicAdd (volatile int *p, int val)
 #define B3_LARGE_FLOAT 1e18f
 #define B3_INFINITY 1e18f
 #define b3Assert(a)
-#define b3ConstArray(a) __global const a*
+#define b3ConstArray(a) __global const a *
 #define b3AtomicInc atomic_inc
 #define b3AtomicAdd atomic_add
 #define b3Fabs fabs
diff --git a/thirdparty/bullet/Bullet3Common/shared/b3Quat.h b/thirdparty/bullet/Bullet3Common/shared/b3Quat.h
index f262d5e08f4..940610c77b9 100644
--- a/thirdparty/bullet/Bullet3Common/shared/b3Quat.h
+++ b/thirdparty/bullet/Bullet3Common/shared/b3Quat.h
@@ -5,35 +5,34 @@
 #include "Bullet3Common/shared/b3Float4.h"
 
 #ifdef __cplusplus
-	#include "Bullet3Common/b3Quaternion.h"
-	#include "Bullet3Common/b3Transform.h"
+#include "Bullet3Common/b3Quaternion.h"
+#include "Bullet3Common/b3Transform.h"
 
-	#define b3Quat b3Quaternion
-	#define b3QuatConstArg const b3Quaternion&
-	inline b3Quat b3QuatInverse(b3QuatConstArg orn)
-	{
-		return orn.inverse();
-	}
+#define b3Quat b3Quaternion
+#define b3QuatConstArg const b3Quaternion&
+inline b3Quat b3QuatInverse(b3QuatConstArg orn)
+{
+	return orn.inverse();
+}
 
-	inline b3Float4 b3TransformPoint(b3Float4ConstArg point, b3Float4ConstArg translation, b3QuatConstArg  orientation)
-	{
-		b3Transform tr;
-		tr.setOrigin(translation);
-		tr.setRotation(orientation);
-		return tr(point);
-	}
+inline b3Float4 b3TransformPoint(b3Float4ConstArg point, b3Float4ConstArg translation, b3QuatConstArg orientation)
+{
+	b3Transform tr;
+	tr.setOrigin(translation);
+	tr.setRotation(orientation);
+	return tr(point);
+}
 
 #else
-	typedef float4	b3Quat;
-	#define b3QuatConstArg const b3Quat
-	
-	
+typedef float4 b3Quat;
+#define b3QuatConstArg const b3Quat
+
 inline float4 b3FastNormalize4(float4 v)
 {
-	v = (float4)(v.xyz,0.f);
+	v = (float4)(v.xyz, 0.f);
 	return fast_normalize(v);
 }
-	
+
 inline b3Quat b3QuatMul(b3Quat a, b3Quat b);
 inline b3Quat b3QuatNormalized(b3QuatConstArg in);
 inline b3Quat b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec);
@@ -43,20 +42,20 @@ inline b3Quat b3QuatInverse(b3QuatConstArg q);
 inline b3Quat b3QuatMul(b3QuatConstArg a, b3QuatConstArg b)
 {
 	b3Quat ans;
-	ans = b3Cross3( a, b );
-	ans += a.w*b+b.w*a;
-//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);
-	ans.w = a.w*b.w - b3Dot3F4(a, b);
+	ans = b3Cross3(a, b);
+	ans += a.w * b + b.w * a;
+	//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);
+	ans.w = a.w * b.w - b3Dot3F4(a, b);
 	return ans;
 }
 
 inline b3Quat b3QuatNormalized(b3QuatConstArg in)
 {
 	b3Quat q;
-	q=in;
+	q = in;
 	//return b3FastNormalize4(in);
 	float len = native_sqrt(dot(q, q));
-	if(len > 0.f)
+	if (len > 0.f)
 	{
 		q *= 1.f / len;
 	}
@@ -69,15 +68,13 @@ inline b3Quat b3QuatNormalized(b3QuatConstArg in)
 }
 inline float4 b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec)
 {
-	b3Quat qInv = b3QuatInvert( q );
+	b3Quat qInv = b3QuatInvert(q);
 	float4 vcpy = vec;
 	vcpy.w = 0.f;
-	float4 out = b3QuatMul(b3QuatMul(q,vcpy),qInv);
+	float4 out = b3QuatMul(b3QuatMul(q, vcpy), qInv);
 	return out;
 }
 
-
-
 inline b3Quat b3QuatInverse(b3QuatConstArg q)
 {
 	return (b3Quat)(-q.xyz, q.w);
@@ -90,14 +87,14 @@ inline b3Quat b3QuatInvert(b3QuatConstArg q)
 
 inline float4 b3QuatInvRotate(b3QuatConstArg q, b3QuatConstArg vec)
 {
-	return b3QuatRotate( b3QuatInvert( q ), vec );
+	return b3QuatRotate(b3QuatInvert(q), vec);
 }
 
-inline b3Float4 b3TransformPoint(b3Float4ConstArg point, b3Float4ConstArg translation, b3QuatConstArg  orientation)
+inline b3Float4 b3TransformPoint(b3Float4ConstArg point, b3Float4ConstArg translation, b3QuatConstArg orientation)
 {
-	return b3QuatRotate( orientation, point ) + (translation);
+	return b3QuatRotate(orientation, point) + (translation);
 }
-	
-#endif 
 
-#endif //B3_QUAT_H
+#endif
+
+#endif  //B3_QUAT_H
diff --git a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3ContactSolverInfo.h b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3ContactSolverInfo.h
index 7a12257b33c..049c9116fd9 100644
--- a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3ContactSolverInfo.h
+++ b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3ContactSolverInfo.h
@@ -18,7 +18,7 @@ subject to the following restrictions:
 
 #include "Bullet3Common/b3Scalar.h"
 
-enum	b3SolverMode
+enum b3SolverMode
 {
 	B3_SOLVER_RANDMIZE_ORDER = 1,
 	B3_SOLVER_FRICTION_SEPARATE = 2,
@@ -34,45 +34,38 @@ enum	b3SolverMode
 
 struct b3ContactSolverInfoData
 {
-	
-
-	b3Scalar	m_tau;
-	b3Scalar	m_damping;//global non-contact constraint damping, can be locally overridden by constraints during 'getInfo2'.
-	b3Scalar	m_friction;
-	b3Scalar	m_timeStep;
-	b3Scalar	m_restitution;
-	int		m_numIterations;
-	b3Scalar	m_maxErrorReduction;
-	b3Scalar	m_sor;
-	b3Scalar	m_erp;//used as Baumgarte factor
-	b3Scalar	m_erp2;//used in Split Impulse
-	b3Scalar	m_globalCfm;//constraint force mixing
-	int			m_splitImpulse;
-	b3Scalar	m_splitImpulsePenetrationThreshold;
-	b3Scalar	m_splitImpulseTurnErp;
-	b3Scalar	m_linearSlop;
-	b3Scalar	m_warmstartingFactor;
-
-	int			m_solverMode;
-	int	m_restingContactRestitutionThreshold;
-	int			m_minimumSolverBatchSize;
-	b3Scalar	m_maxGyroscopicForce;
-	b3Scalar	m_singleAxisRollingFrictionThreshold;
-
+	b3Scalar m_tau;
+	b3Scalar m_damping;  //global non-contact constraint damping, can be locally overridden by constraints during 'getInfo2'.
+	b3Scalar m_friction;
+	b3Scalar m_timeStep;
+	b3Scalar m_restitution;
+	int m_numIterations;
+	b3Scalar m_maxErrorReduction;
+	b3Scalar m_sor;
+	b3Scalar m_erp;        //used as Baumgarte factor
+	b3Scalar m_erp2;       //used in Split Impulse
+	b3Scalar m_globalCfm;  //constraint force mixing
+	int m_splitImpulse;
+	b3Scalar m_splitImpulsePenetrationThreshold;
+	b3Scalar m_splitImpulseTurnErp;
+	b3Scalar m_linearSlop;
+	b3Scalar m_warmstartingFactor;
 
+	int m_solverMode;
+	int m_restingContactRestitutionThreshold;
+	int m_minimumSolverBatchSize;
+	b3Scalar m_maxGyroscopicForce;
+	b3Scalar m_singleAxisRollingFrictionThreshold;
 };
 
 struct b3ContactSolverInfo : public b3ContactSolverInfoData
 {
-
-	
-
 	inline b3ContactSolverInfo()
 	{
 		m_tau = b3Scalar(0.6);
 		m_damping = b3Scalar(1.0);
 		m_friction = b3Scalar(0.3);
-		m_timeStep = b3Scalar(1.f/60.f);
+		m_timeStep = b3Scalar(1.f / 60.f);
 		m_restitution = b3Scalar(0.);
 		m_maxErrorReduction = b3Scalar(20.);
 		m_numIterations = 10;
@@ -84,76 +77,73 @@ struct b3ContactSolverInfo : public b3ContactSolverInfoData
 		m_splitImpulsePenetrationThreshold = -.04f;
 		m_splitImpulseTurnErp = 0.1f;
 		m_linearSlop = b3Scalar(0.0);
-		m_warmstartingFactor=b3Scalar(0.85);
+		m_warmstartingFactor = b3Scalar(0.85);
 		//m_solverMode =  B3_SOLVER_USE_WARMSTARTING |  B3_SOLVER_SIMD | B3_SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION|B3_SOLVER_USE_2_FRICTION_DIRECTIONS|B3_SOLVER_ENABLE_FRICTION_DIRECTION_CACHING;// | B3_SOLVER_RANDMIZE_ORDER;
-		m_solverMode = B3_SOLVER_USE_WARMSTARTING | B3_SOLVER_SIMD;// | B3_SOLVER_RANDMIZE_ORDER;
-		m_restingContactRestitutionThreshold = 2;//unused as of 2.81
-		m_minimumSolverBatchSize = 128; //try to combine islands until the amount of constraints reaches this limit
-		m_maxGyroscopicForce = 100.f; ///only used to clamp forces for bodies that have their B3_ENABLE_GYROPSCOPIC_FORCE flag set (using b3RigidBody::setFlag)
-		m_singleAxisRollingFrictionThreshold = 1e30f;///if the velocity is above this threshold, it will use a single constraint row (axis), otherwise 3 rows.
+		m_solverMode = B3_SOLVER_USE_WARMSTARTING | B3_SOLVER_SIMD;  // | B3_SOLVER_RANDMIZE_ORDER;
+		m_restingContactRestitutionThreshold = 2;                    //unused as of 2.81
+		m_minimumSolverBatchSize = 128;                              //try to combine islands until the amount of constraints reaches this limit
+		m_maxGyroscopicForce = 100.f;                                ///only used to clamp forces for bodies that have their B3_ENABLE_GYROPSCOPIC_FORCE flag set (using b3RigidBody::setFlag)
+		m_singleAxisRollingFrictionThreshold = 1e30f;                ///if the velocity is above this threshold, it will use a single constraint row (axis), otherwise 3 rows.
 	}
 };
 
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
 struct b3ContactSolverInfoDoubleData
 {
-	double		m_tau;
-	double		m_damping;//global non-contact constraint damping, can be locally overridden by constraints during 'getInfo2'.
-	double		m_friction;
-	double		m_timeStep;
-	double		m_restitution;
-	double		m_maxErrorReduction;
-	double		m_sor;
-	double		m_erp;//used as Baumgarte factor
-	double		m_erp2;//used in Split Impulse
-	double		m_globalCfm;//constraint force mixing
-	double		m_splitImpulsePenetrationThreshold;
-	double		m_splitImpulseTurnErp;
-	double		m_linearSlop;
-	double		m_warmstartingFactor;
-	double		m_maxGyroscopicForce;
-	double		m_singleAxisRollingFrictionThreshold;
-
-	int			m_numIterations;
-	int			m_solverMode;
-	int			m_restingContactRestitutionThreshold;
-	int			m_minimumSolverBatchSize;
-	int			m_splitImpulse;
-	char		m_padding[4];
+	double m_tau;
+	double m_damping;  //global non-contact constraint damping, can be locally overridden by constraints during 'getInfo2'.
+	double m_friction;
+	double m_timeStep;
+	double m_restitution;
+	double m_maxErrorReduction;
+	double m_sor;
+	double m_erp;        //used as Baumgarte factor
+	double m_erp2;       //used in Split Impulse
+	double m_globalCfm;  //constraint force mixing
+	double m_splitImpulsePenetrationThreshold;
+	double m_splitImpulseTurnErp;
+	double m_linearSlop;
+	double m_warmstartingFactor;
+	double m_maxGyroscopicForce;
+	double m_singleAxisRollingFrictionThreshold;
 
+	int m_numIterations;
+	int m_solverMode;
+	int m_restingContactRestitutionThreshold;
+	int m_minimumSolverBatchSize;
+	int m_splitImpulse;
+	char m_padding[4];
 };
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
 struct b3ContactSolverInfoFloatData
 {
-	float		m_tau;
-	float		m_damping;//global non-contact constraint damping, can be locally overridden by constraints during 'getInfo2'.
-	float		m_friction;
-	float		m_timeStep;
+	float m_tau;
+	float m_damping;  //global non-contact constraint damping, can be locally overridden by constraints during 'getInfo2'.
+	float m_friction;
+	float m_timeStep;
 
-	float		m_restitution;
-	float		m_maxErrorReduction;
-	float		m_sor;
-	float		m_erp;//used as Baumgarte factor
+	float m_restitution;
+	float m_maxErrorReduction;
+	float m_sor;
+	float m_erp;  //used as Baumgarte factor
 
-	float		m_erp2;//used in Split Impulse
-	float		m_globalCfm;//constraint force mixing
-	float		m_splitImpulsePenetrationThreshold;
-	float		m_splitImpulseTurnErp;
+	float m_erp2;       //used in Split Impulse
+	float m_globalCfm;  //constraint force mixing
+	float m_splitImpulsePenetrationThreshold;
+	float m_splitImpulseTurnErp;
 
-	float		m_linearSlop;
-	float		m_warmstartingFactor;
-	float		m_maxGyroscopicForce;
-	float		m_singleAxisRollingFrictionThreshold;
+	float m_linearSlop;
+	float m_warmstartingFactor;
+	float m_maxGyroscopicForce;
+	float m_singleAxisRollingFrictionThreshold;
 
-	int			m_numIterations;
-	int			m_solverMode;
-	int			m_restingContactRestitutionThreshold;
-	int			m_minimumSolverBatchSize;
+	int m_numIterations;
+	int m_solverMode;
+	int m_restingContactRestitutionThreshold;
+	int m_minimumSolverBatchSize;
 
-	int			m_splitImpulse;
-	char		m_padding[4];
+	int m_splitImpulse;
+	char m_padding[4];
 };
 
-
-
-#endif //B3_CONTACT_SOLVER_INFO
+#endif  //B3_CONTACT_SOLVER_INFO
diff --git a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3FixedConstraint.cpp b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3FixedConstraint.cpp
index 5e11e749357..ace4b183884 100644
--- a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3FixedConstraint.cpp
+++ b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3FixedConstraint.cpp
@@ -4,105 +4,100 @@
 #include "Bullet3Common/b3TransformUtil.h"
 #include <new>
 
-
-b3FixedConstraint::b3FixedConstraint(int rbA,int rbB, const b3Transform& frameInA,const b3Transform& frameInB)
-:b3TypedConstraint(B3_FIXED_CONSTRAINT_TYPE,rbA,rbB)
+b3FixedConstraint::b3FixedConstraint(int rbA, int rbB, const b3Transform& frameInA, const b3Transform& frameInB)
+	: b3TypedConstraint(B3_FIXED_CONSTRAINT_TYPE, rbA, rbB)
 {
 	m_pivotInA = frameInA.getOrigin();
 	m_pivotInB = frameInB.getOrigin();
-	m_relTargetAB = frameInA.getRotation()*frameInB.getRotation().inverse();
-
+	m_relTargetAB = frameInA.getRotation() * frameInB.getRotation().inverse();
 }
 
-b3FixedConstraint::~b3FixedConstraint ()
+b3FixedConstraint::~b3FixedConstraint()
 {
 }
 
-	
-void b3FixedConstraint::getInfo1 (b3ConstraintInfo1* info,const b3RigidBodyData* bodies)
+void b3FixedConstraint::getInfo1(b3ConstraintInfo1* info, const b3RigidBodyData* bodies)
 {
 	info->m_numConstraintRows = 6;
 	info->nub = 6;
 }
 
-void b3FixedConstraint::getInfo2 (b3ConstraintInfo2* info, const b3RigidBodyData* bodies)
+void b3FixedConstraint::getInfo2(b3ConstraintInfo2* info, const b3RigidBodyData* bodies)
 {
 	//fix the 3 linear degrees of freedom
 
 	const b3Vector3& worldPosA = bodies[m_rbA].m_pos;
 	const b3Quaternion& worldOrnA = bodies[m_rbA].m_quat;
-	const b3Vector3& worldPosB= bodies[m_rbB].m_pos;
+	const b3Vector3& worldPosB = bodies[m_rbB].m_pos;
 	const b3Quaternion& worldOrnB = bodies[m_rbB].m_quat;
 
 	info->m_J1linearAxis[0] = 1;
-	info->m_J1linearAxis[info->rowskip+1] = 1;
-	info->m_J1linearAxis[2*info->rowskip+2] = 1;
+	info->m_J1linearAxis[info->rowskip + 1] = 1;
+	info->m_J1linearAxis[2 * info->rowskip + 2] = 1;
 
-	b3Vector3 a1 = b3QuatRotate(worldOrnA,m_pivotInA);
+	b3Vector3 a1 = b3QuatRotate(worldOrnA, m_pivotInA);
 	{
 		b3Vector3* angular0 = (b3Vector3*)(info->m_J1angularAxis);
-		b3Vector3* angular1 = (b3Vector3*)(info->m_J1angularAxis+info->rowskip);
-		b3Vector3* angular2 = (b3Vector3*)(info->m_J1angularAxis+2*info->rowskip);
+		b3Vector3* angular1 = (b3Vector3*)(info->m_J1angularAxis + info->rowskip);
+		b3Vector3* angular2 = (b3Vector3*)(info->m_J1angularAxis + 2 * info->rowskip);
 		b3Vector3 a1neg = -a1;
-		a1neg.getSkewSymmetricMatrix(angular0,angular1,angular2);
+		a1neg.getSkewSymmetricMatrix(angular0, angular1, angular2);
 	}
-    
+
 	if (info->m_J2linearAxis)
 	{
 		info->m_J2linearAxis[0] = -1;
-		info->m_J2linearAxis[info->rowskip+1] = -1;
-		info->m_J2linearAxis[2*info->rowskip+2] = -1;
+		info->m_J2linearAxis[info->rowskip + 1] = -1;
+		info->m_J2linearAxis[2 * info->rowskip + 2] = -1;
 	}
-	
-	b3Vector3 a2 = b3QuatRotate(worldOrnB,m_pivotInB);
-   
+
+	b3Vector3 a2 = b3QuatRotate(worldOrnB, m_pivotInB);
+
 	{
-	//	b3Vector3 a2n = -a2;
+		//	b3Vector3 a2n = -a2;
 		b3Vector3* angular0 = (b3Vector3*)(info->m_J2angularAxis);
-		b3Vector3* angular1 = (b3Vector3*)(info->m_J2angularAxis+info->rowskip);
-		b3Vector3* angular2 = (b3Vector3*)(info->m_J2angularAxis+2*info->rowskip);
-		a2.getSkewSymmetricMatrix(angular0,angular1,angular2);
+		b3Vector3* angular1 = (b3Vector3*)(info->m_J2angularAxis + info->rowskip);
+		b3Vector3* angular2 = (b3Vector3*)(info->m_J2angularAxis + 2 * info->rowskip);
+		a2.getSkewSymmetricMatrix(angular0, angular1, angular2);
 	}
 
-    // set right hand side for the linear dofs
+	// set right hand side for the linear dofs
 	b3Scalar k = info->fps * info->erp;
-	b3Vector3 linearError = k*(a2+worldPosB-a1-worldPosA);
-    int j;
-	for (j=0; j<3; j++)
-    {
-        info->m_constraintError[j*info->rowskip] = linearError[j];
+	b3Vector3 linearError = k * (a2 + worldPosB - a1 - worldPosA);
+	int j;
+	for (j = 0; j < 3; j++)
+	{
+		info->m_constraintError[j * info->rowskip] = linearError[j];
 		//printf("info->m_constraintError[%d]=%f\n",j,info->m_constraintError[j]);
-    }
+	}
 
-		//fix the 3 angular degrees of freedom
+	//fix the 3 angular degrees of freedom
 
 	int start_row = 3;
 	int s = info->rowskip;
-    int start_index = start_row * s;
+	int start_index = start_row * s;
 
-    // 3 rows to make body rotations equal
+	// 3 rows to make body rotations equal
 	info->m_J1angularAxis[start_index] = 1;
-    info->m_J1angularAxis[start_index + s + 1] = 1;
-    info->m_J1angularAxis[start_index + s*2+2] = 1;
-    if ( info->m_J2angularAxis)
-    {
-        info->m_J2angularAxis[start_index] = -1;
-        info->m_J2angularAxis[start_index + s+1] = -1;
-        info->m_J2angularAxis[start_index + s*2+2] = -1;
-    }
+	info->m_J1angularAxis[start_index + s + 1] = 1;
+	info->m_J1angularAxis[start_index + s * 2 + 2] = 1;
+	if (info->m_J2angularAxis)
+	{
+		info->m_J2angularAxis[start_index] = -1;
+		info->m_J2angularAxis[start_index + s + 1] = -1;
+		info->m_J2angularAxis[start_index + s * 2 + 2] = -1;
+	}
 
-
-    // set right hand side for the angular dofs
+	// set right hand side for the angular dofs
 
 	b3Vector3 diff;
 	b3Scalar angle;
-	b3Quaternion qrelCur = worldOrnA *worldOrnB.inverse();
-
-	b3TransformUtil::calculateDiffAxisAngleQuaternion(m_relTargetAB,qrelCur,diff,angle);
-	diff*=-angle;
-	for (j=0; j<3; j++)
-    {
-        info->m_constraintError[(3+j)*info->rowskip] = k * diff[j];
-    }
+	b3Quaternion qrelCur = worldOrnA * worldOrnB.inverse();
 
+	b3TransformUtil::calculateDiffAxisAngleQuaternion(m_relTargetAB, qrelCur, diff, angle);
+	diff *= -angle;
+	for (j = 0; j < 3; j++)
+	{
+		info->m_constraintError[(3 + j) * info->rowskip] = k * diff[j];
+	}
 }
\ No newline at end of file
diff --git a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3FixedConstraint.h b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3FixedConstraint.h
index e884a829129..64809666e48 100644
--- a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3FixedConstraint.h
+++ b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3FixedConstraint.h
@@ -4,32 +4,31 @@
 
 #include "b3TypedConstraint.h"
 
-B3_ATTRIBUTE_ALIGNED16(class) b3FixedConstraint : public b3TypedConstraint
+B3_ATTRIBUTE_ALIGNED16(class)
+b3FixedConstraint : public b3TypedConstraint
 {
 	b3Vector3 m_pivotInA;
 	b3Vector3 m_pivotInB;
 	b3Quaternion m_relTargetAB;
 
 public:
-	b3FixedConstraint(int  rbA,int rbB, const b3Transform& frameInA,const b3Transform& frameInB);
-	
+	b3FixedConstraint(int rbA, int rbB, const b3Transform& frameInA, const b3Transform& frameInB);
+
 	virtual ~b3FixedConstraint();
 
-	
-	virtual void getInfo1 (b3ConstraintInfo1* info,const b3RigidBodyData* bodies);
+	virtual void getInfo1(b3ConstraintInfo1 * info, const b3RigidBodyData* bodies);
 
-	virtual void getInfo2 (b3ConstraintInfo2* info, const b3RigidBodyData* bodies);
+	virtual void getInfo2(b3ConstraintInfo2 * info, const b3RigidBodyData* bodies);
 
-	virtual	void	setParam(int num, b3Scalar value, int axis = -1)
+	virtual void setParam(int num, b3Scalar value, int axis = -1)
 	{
 		b3Assert(0);
 	}
-	virtual	b3Scalar getParam(int num, int axis = -1) const
+	virtual b3Scalar getParam(int num, int axis = -1) const
 	{
 		b3Assert(0);
 		return 0.f;
 	}
-
 };
 
-#endif //B3_FIXED_CONSTRAINT_H
+#endif  //B3_FIXED_CONSTRAINT_H
diff --git a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3Generic6DofConstraint.cpp b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3Generic6DofConstraint.cpp
index 3ae2922e586..0d5bb2014bf 100644
--- a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3Generic6DofConstraint.cpp
+++ b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3Generic6DofConstraint.cpp
@@ -26,69 +26,48 @@ http://gimpact.sf.net
 #include "Bullet3Common/b3TransformUtil.h"
 #include <new>
 
-
-
 #define D6_USE_OBSOLETE_METHOD false
 #define D6_USE_FRAME_OFFSET true
 
-
-
-
-
-
-b3Generic6DofConstraint::b3Generic6DofConstraint(int rbA,int  rbB, const b3Transform& frameInA, const b3Transform& frameInB, bool useLinearReferenceFrameA, const b3RigidBodyData* bodies)
-: b3TypedConstraint(B3_D6_CONSTRAINT_TYPE, rbA, rbB)
-, m_frameInA(frameInA)
-, m_frameInB(frameInB),
-m_useLinearReferenceFrameA(useLinearReferenceFrameA),
-m_useOffsetForConstraintFrame(D6_USE_FRAME_OFFSET),
-m_flags(0)
+b3Generic6DofConstraint::b3Generic6DofConstraint(int rbA, int rbB, const b3Transform& frameInA, const b3Transform& frameInB, bool useLinearReferenceFrameA, const b3RigidBodyData* bodies)
+	: b3TypedConstraint(B3_D6_CONSTRAINT_TYPE, rbA, rbB), m_frameInA(frameInA), m_frameInB(frameInB), m_useLinearReferenceFrameA(useLinearReferenceFrameA), m_useOffsetForConstraintFrame(D6_USE_FRAME_OFFSET), m_flags(0)
 {
 	calculateTransforms(bodies);
 }
 
-
-
-
-
-
 #define GENERIC_D6_DISABLE_WARMSTARTING 1
 
-
-
 b3Scalar btGetMatrixElem(const b3Matrix3x3& mat, int index);
 b3Scalar btGetMatrixElem(const b3Matrix3x3& mat, int index)
 {
-	int i = index%3;
-	int j = index/3;
+	int i = index % 3;
+	int j = index / 3;
 	return mat[i][j];
 }
 
-
-
 ///MatrixToEulerXYZ from http://www.geometrictools.com/LibFoundation/Mathematics/Wm4Matrix3.inl.html
-bool	matrixToEulerXYZ(const b3Matrix3x3& mat,b3Vector3& xyz);
-bool	matrixToEulerXYZ(const b3Matrix3x3& mat,b3Vector3& xyz)
+bool matrixToEulerXYZ(const b3Matrix3x3& mat, b3Vector3& xyz);
+bool matrixToEulerXYZ(const b3Matrix3x3& mat, b3Vector3& xyz)
 {
 	//	// rot =  cy*cz          -cy*sz           sy
 	//	//        cz*sx*sy+cx*sz  cx*cz-sx*sy*sz -cy*sx
 	//	//       -cx*cz*sy+sx*sz  cz*sx+cx*sy*sz  cx*cy
 	//
 
-	b3Scalar fi = btGetMatrixElem(mat,2);
+	b3Scalar fi = btGetMatrixElem(mat, 2);
 	if (fi < b3Scalar(1.0f))
 	{
 		if (fi > b3Scalar(-1.0f))
 		{
-			xyz[0] = b3Atan2(-btGetMatrixElem(mat,5),btGetMatrixElem(mat,8));
-			xyz[1] = b3Asin(btGetMatrixElem(mat,2));
-			xyz[2] = b3Atan2(-btGetMatrixElem(mat,1),btGetMatrixElem(mat,0));
+			xyz[0] = b3Atan2(-btGetMatrixElem(mat, 5), btGetMatrixElem(mat, 8));
+			xyz[1] = b3Asin(btGetMatrixElem(mat, 2));
+			xyz[2] = b3Atan2(-btGetMatrixElem(mat, 1), btGetMatrixElem(mat, 0));
 			return true;
 		}
 		else
 		{
 			// WARNING.  Not unique.  XA - ZA = -atan2(r10,r11)
-			xyz[0] = -b3Atan2(btGetMatrixElem(mat,3),btGetMatrixElem(mat,4));
+			xyz[0] = -b3Atan2(btGetMatrixElem(mat, 3), btGetMatrixElem(mat, 4));
 			xyz[1] = -B3_HALF_PI;
 			xyz[2] = b3Scalar(0.0);
 			return false;
@@ -97,7 +76,7 @@ bool	matrixToEulerXYZ(const b3Matrix3x3& mat,b3Vector3& xyz)
 	else
 	{
 		// WARNING.  Not unique.  XAngle + ZAngle = atan2(r10,r11)
-		xyz[0] = b3Atan2(btGetMatrixElem(mat,3),btGetMatrixElem(mat,4));
+		xyz[0] = b3Atan2(btGetMatrixElem(mat, 3), btGetMatrixElem(mat, 4));
 		xyz[1] = B3_HALF_PI;
 		xyz[2] = 0.0;
 	}
@@ -108,85 +87,75 @@ bool	matrixToEulerXYZ(const b3Matrix3x3& mat,b3Vector3& xyz)
 
 int b3RotationalLimitMotor::testLimitValue(b3Scalar test_value)
 {
-	if(m_loLimit>m_hiLimit)
+	if (m_loLimit > m_hiLimit)
 	{
-		m_currentLimit = 0;//Free from violation
+		m_currentLimit = 0;  //Free from violation
 		return 0;
 	}
 	if (test_value < m_loLimit)
 	{
-		m_currentLimit = 1;//low limit violation
-		m_currentLimitError =  test_value - m_loLimit;
-		if(m_currentLimitError>B3_PI) 
-			m_currentLimitError-=B3_2_PI;
-		else if(m_currentLimitError<-B3_PI) 
-			m_currentLimitError+=B3_2_PI;
+		m_currentLimit = 1;  //low limit violation
+		m_currentLimitError = test_value - m_loLimit;
+		if (m_currentLimitError > B3_PI)
+			m_currentLimitError -= B3_2_PI;
+		else if (m_currentLimitError < -B3_PI)
+			m_currentLimitError += B3_2_PI;
 		return 1;
 	}
-	else if (test_value> m_hiLimit)
+	else if (test_value > m_hiLimit)
 	{
-		m_currentLimit = 2;//High limit violation
+		m_currentLimit = 2;  //High limit violation
 		m_currentLimitError = test_value - m_hiLimit;
-		if(m_currentLimitError>B3_PI) 
-			m_currentLimitError-=B3_2_PI;
-		else if(m_currentLimitError<-B3_PI) 
-			m_currentLimitError+=B3_2_PI;
+		if (m_currentLimitError > B3_PI)
+			m_currentLimitError -= B3_2_PI;
+		else if (m_currentLimitError < -B3_PI)
+			m_currentLimitError += B3_2_PI;
 		return 2;
 	};
 
-	m_currentLimit = 0;//Free from violation
+	m_currentLimit = 0;  //Free from violation
 	return 0;
-
 }
 
-
-
-
 //////////////////////////// End b3RotationalLimitMotor ////////////////////////////////////
 
-
-
-
 //////////////////////////// b3TranslationalLimitMotor ////////////////////////////////////
 
-
 int b3TranslationalLimitMotor::testLimitValue(int limitIndex, b3Scalar test_value)
 {
 	b3Scalar loLimit = m_lowerLimit[limitIndex];
 	b3Scalar hiLimit = m_upperLimit[limitIndex];
-	if(loLimit > hiLimit)
+	if (loLimit > hiLimit)
 	{
-		m_currentLimit[limitIndex] = 0;//Free from violation
+		m_currentLimit[limitIndex] = 0;  //Free from violation
 		m_currentLimitError[limitIndex] = b3Scalar(0.f);
 		return 0;
 	}
 
 	if (test_value < loLimit)
 	{
-		m_currentLimit[limitIndex] = 2;//low limit violation
-		m_currentLimitError[limitIndex] =  test_value - loLimit;
+		m_currentLimit[limitIndex] = 2;  //low limit violation
+		m_currentLimitError[limitIndex] = test_value - loLimit;
 		return 2;
 	}
-	else if (test_value> hiLimit)
+	else if (test_value > hiLimit)
 	{
-		m_currentLimit[limitIndex] = 1;//High limit violation
+		m_currentLimit[limitIndex] = 1;  //High limit violation
 		m_currentLimitError[limitIndex] = test_value - hiLimit;
 		return 1;
 	};
 
-	m_currentLimit[limitIndex] = 0;//Free from violation
+	m_currentLimit[limitIndex] = 0;  //Free from violation
 	m_currentLimitError[limitIndex] = b3Scalar(0.f);
 	return 0;
 }
 
-
-
 //////////////////////////// b3TranslationalLimitMotor ////////////////////////////////////
 
 void b3Generic6DofConstraint::calculateAngleInfo()
 {
-	b3Matrix3x3 relative_frame = m_calculatedTransformA.getBasis().inverse()*m_calculatedTransformB.getBasis();
-	matrixToEulerXYZ(relative_frame,m_calculatedAxisAngleDiff);
+	b3Matrix3x3 relative_frame = m_calculatedTransformA.getBasis().inverse() * m_calculatedTransformB.getBasis();
+	matrixToEulerXYZ(relative_frame, m_calculatedAxisAngleDiff);
 	// in euler angle mode we do not actually constrain the angular velocity
 	// along the axes axis[0] and axis[2] (although we do use axis[1]) :
 	//
@@ -211,12 +180,11 @@ void b3Generic6DofConstraint::calculateAngleInfo()
 	m_calculatedAxis[0].normalize();
 	m_calculatedAxis[1].normalize();
 	m_calculatedAxis[2].normalize();
-
 }
 
 static b3Transform getCenterOfMassTransform(const b3RigidBodyData& body)
 {
-	b3Transform tr(body.m_quat,body.m_pos);
+	b3Transform tr(body.m_quat, body.m_pos);
 	return tr;
 }
 
@@ -226,26 +194,26 @@ void b3Generic6DofConstraint::calculateTransforms(const b3RigidBodyData* bodies)
 	b3Transform transB;
 	transA = getCenterOfMassTransform(bodies[m_rbA]);
 	transB = getCenterOfMassTransform(bodies[m_rbB]);
-	calculateTransforms(transA,transB,bodies);
+	calculateTransforms(transA, transB, bodies);
 }
 
-void b3Generic6DofConstraint::calculateTransforms(const b3Transform& transA,const b3Transform& transB,const b3RigidBodyData* bodies)
+void b3Generic6DofConstraint::calculateTransforms(const b3Transform& transA, const b3Transform& transB, const b3RigidBodyData* bodies)
 {
 	m_calculatedTransformA = transA * m_frameInA;
 	m_calculatedTransformB = transB * m_frameInB;
 	calculateLinearInfo();
 	calculateAngleInfo();
-	if(m_useOffsetForConstraintFrame)
-	{	//  get weight factors depending on masses
+	if (m_useOffsetForConstraintFrame)
+	{  //  get weight factors depending on masses
 		b3Scalar miA = bodies[m_rbA].m_invMass;
 		b3Scalar miB = bodies[m_rbB].m_invMass;
 		m_hasStaticBody = (miA < B3_EPSILON) || (miB < B3_EPSILON);
 		b3Scalar miS = miA + miB;
-		if(miS > b3Scalar(0.f))
+		if (miS > b3Scalar(0.f))
 		{
 			m_factA = miB / miS;
 		}
-		else 
+		else
 		{
 			m_factA = b3Scalar(0.5f);
 		}
@@ -253,12 +221,6 @@ void b3Generic6DofConstraint::calculateTransforms(const b3Transform& transA,cons
 	}
 }
 
-
-
-
-
-
-
 bool b3Generic6DofConstraint::testAngularLimitMotor(int axis_index)
 {
 	b3Scalar angle = m_calculatedAxisAngleDiff[axis_index];
@@ -269,48 +231,43 @@ bool b3Generic6DofConstraint::testAngularLimitMotor(int axis_index)
 	return m_angularLimits[axis_index].needApplyTorques();
 }
 
-
-
-
-void b3Generic6DofConstraint::getInfo1 (b3ConstraintInfo1* info,const b3RigidBodyData* bodies)
+void b3Generic6DofConstraint::getInfo1(b3ConstraintInfo1* info, const b3RigidBodyData* bodies)
 {
 	//prepare constraint
-	calculateTransforms(getCenterOfMassTransform(bodies[m_rbA]),getCenterOfMassTransform(bodies[m_rbB]),bodies);
+	calculateTransforms(getCenterOfMassTransform(bodies[m_rbA]), getCenterOfMassTransform(bodies[m_rbB]), bodies);
 	info->m_numConstraintRows = 0;
 	info->nub = 6;
 	int i;
 	//test linear limits
-	for(i = 0; i < 3; i++)
+	for (i = 0; i < 3; i++)
 	{
-		if(m_linearLimits.needApplyForce(i))
+		if (m_linearLimits.needApplyForce(i))
 		{
 			info->m_numConstraintRows++;
 			info->nub--;
 		}
 	}
 	//test angular limits
-	for (i=0;i<3 ;i++ )
+	for (i = 0; i < 3; i++)
 	{
-		if(testAngularLimitMotor(i))
+		if (testAngularLimitMotor(i))
 		{
 			info->m_numConstraintRows++;
 			info->nub--;
 		}
 	}
-//	printf("info->m_numConstraintRows=%d\n",info->m_numConstraintRows);
+	//	printf("info->m_numConstraintRows=%d\n",info->m_numConstraintRows);
 }
 
-void b3Generic6DofConstraint::getInfo1NonVirtual (b3ConstraintInfo1* info,const b3RigidBodyData* bodies)
+void b3Generic6DofConstraint::getInfo1NonVirtual(b3ConstraintInfo1* info, const b3RigidBodyData* bodies)
 {
 	//pre-allocate all 6
 	info->m_numConstraintRows = 6;
 	info->nub = 0;
 }
 
-
-void b3Generic6DofConstraint::getInfo2 (b3ConstraintInfo2* info,const b3RigidBodyData* bodies)
+void b3Generic6DofConstraint::getInfo2(b3ConstraintInfo2* info, const b3RigidBodyData* bodies)
 {
-
 	b3Transform transA = getCenterOfMassTransform(bodies[m_rbA]);
 	b3Transform transB = getCenterOfMassTransform(bodies[m_rbB]);
 	const b3Vector3& linVelA = bodies[m_rbA].m_linVel;
@@ -318,136 +275,124 @@ void b3Generic6DofConstraint::getInfo2 (b3ConstraintInfo2* info,const b3RigidBod
 	const b3Vector3& angVelA = bodies[m_rbA].m_angVel;
 	const b3Vector3& angVelB = bodies[m_rbB].m_angVel;
 
-	if(m_useOffsetForConstraintFrame)
-	{ // for stability better to solve angular limits first
-		int row = setAngularLimits(info, 0,transA,transB,linVelA,linVelB,angVelA,angVelB);
-		setLinearLimits(info, row, transA,transB,linVelA,linVelB,angVelA,angVelB);
+	if (m_useOffsetForConstraintFrame)
+	{  // for stability better to solve angular limits first
+		int row = setAngularLimits(info, 0, transA, transB, linVelA, linVelB, angVelA, angVelB);
+		setLinearLimits(info, row, transA, transB, linVelA, linVelB, angVelA, angVelB);
 	}
 	else
-	{ // leave old version for compatibility
-		int row = setLinearLimits(info, 0, transA,transB,linVelA,linVelB,angVelA,angVelB);
-		setAngularLimits(info, row,transA,transB,linVelA,linVelB,angVelA,angVelB);
+	{  // leave old version for compatibility
+		int row = setLinearLimits(info, 0, transA, transB, linVelA, linVelB, angVelA, angVelB);
+		setAngularLimits(info, row, transA, transB, linVelA, linVelB, angVelA, angVelB);
 	}
-
 }
 
-
-void b3Generic6DofConstraint::getInfo2NonVirtual (b3ConstraintInfo2* info, const b3Transform& transA,const b3Transform& transB,const b3Vector3& linVelA,const b3Vector3& linVelB,const b3Vector3& angVelA,const b3Vector3& angVelB,const b3RigidBodyData* bodies)
+void b3Generic6DofConstraint::getInfo2NonVirtual(b3ConstraintInfo2* info, const b3Transform& transA, const b3Transform& transB, const b3Vector3& linVelA, const b3Vector3& linVelB, const b3Vector3& angVelA, const b3Vector3& angVelB, const b3RigidBodyData* bodies)
 {
-	
 	//prepare constraint
-	calculateTransforms(transA,transB,bodies);
+	calculateTransforms(transA, transB, bodies);
 
 	int i;
-	for (i=0;i<3 ;i++ )
+	for (i = 0; i < 3; i++)
 	{
 		testAngularLimitMotor(i);
 	}
 
-	if(m_useOffsetForConstraintFrame)
-	{ // for stability better to solve angular limits first
-		int row = setAngularLimits(info, 0,transA,transB,linVelA,linVelB,angVelA,angVelB);
-		setLinearLimits(info, row, transA,transB,linVelA,linVelB,angVelA,angVelB);
+	if (m_useOffsetForConstraintFrame)
+	{  // for stability better to solve angular limits first
+		int row = setAngularLimits(info, 0, transA, transB, linVelA, linVelB, angVelA, angVelB);
+		setLinearLimits(info, row, transA, transB, linVelA, linVelB, angVelA, angVelB);
 	}
 	else
-	{ // leave old version for compatibility
-		int row = setLinearLimits(info, 0, transA,transB,linVelA,linVelB,angVelA,angVelB);
-		setAngularLimits(info, row,transA,transB,linVelA,linVelB,angVelA,angVelB);
+	{  // leave old version for compatibility
+		int row = setLinearLimits(info, 0, transA, transB, linVelA, linVelB, angVelA, angVelB);
+		setAngularLimits(info, row, transA, transB, linVelA, linVelB, angVelA, angVelB);
 	}
 }
 
-
-
-int b3Generic6DofConstraint::setLinearLimits(b3ConstraintInfo2* info, int row, const b3Transform& transA,const b3Transform& transB,const b3Vector3& linVelA,const b3Vector3& linVelB,const b3Vector3& angVelA,const b3Vector3& angVelB)
+int b3Generic6DofConstraint::setLinearLimits(b3ConstraintInfo2* info, int row, const b3Transform& transA, const b3Transform& transB, const b3Vector3& linVelA, const b3Vector3& linVelB, const b3Vector3& angVelA, const b3Vector3& angVelB)
 {
-//	int row = 0;
+	//	int row = 0;
 	//solve linear limits
 	b3RotationalLimitMotor limot;
-	for (int i=0;i<3 ;i++ )
+	for (int i = 0; i < 3; i++)
 	{
-		if(m_linearLimits.needApplyForce(i))
-		{ // re-use rotational motor code
+		if (m_linearLimits.needApplyForce(i))
+		{  // re-use rotational motor code
 			limot.m_bounce = b3Scalar(0.f);
 			limot.m_currentLimit = m_linearLimits.m_currentLimit[i];
 			limot.m_currentPosition = m_linearLimits.m_currentLinearDiff[i];
-			limot.m_currentLimitError  = m_linearLimits.m_currentLimitError[i];
-			limot.m_damping  = m_linearLimits.m_damping;
-			limot.m_enableMotor  = m_linearLimits.m_enableMotor[i];
-			limot.m_hiLimit  = m_linearLimits.m_upperLimit[i];
-			limot.m_limitSoftness  = m_linearLimits.m_limitSoftness;
-			limot.m_loLimit  = m_linearLimits.m_lowerLimit[i];
-			limot.m_maxLimitForce  = b3Scalar(0.f);
-			limot.m_maxMotorForce  = m_linearLimits.m_maxMotorForce[i];
-			limot.m_targetVelocity  = m_linearLimits.m_targetVelocity[i];
+			limot.m_currentLimitError = m_linearLimits.m_currentLimitError[i];
+			limot.m_damping = m_linearLimits.m_damping;
+			limot.m_enableMotor = m_linearLimits.m_enableMotor[i];
+			limot.m_hiLimit = m_linearLimits.m_upperLimit[i];
+			limot.m_limitSoftness = m_linearLimits.m_limitSoftness;
+			limot.m_loLimit = m_linearLimits.m_lowerLimit[i];
+			limot.m_maxLimitForce = b3Scalar(0.f);
+			limot.m_maxMotorForce = m_linearLimits.m_maxMotorForce[i];
+			limot.m_targetVelocity = m_linearLimits.m_targetVelocity[i];
 			b3Vector3 axis = m_calculatedTransformA.getBasis().getColumn(i);
 			int flags = m_flags >> (i * B3_6DOF_FLAGS_AXIS_SHIFT);
-			limot.m_normalCFM	= (flags & B3_6DOF_FLAGS_CFM_NORM) ? m_linearLimits.m_normalCFM[i] : info->cfm[0];
-			limot.m_stopCFM		= (flags & B3_6DOF_FLAGS_CFM_STOP) ? m_linearLimits.m_stopCFM[i] : info->cfm[0];
-			limot.m_stopERP		= (flags & B3_6DOF_FLAGS_ERP_STOP) ? m_linearLimits.m_stopERP[i] : info->erp;
-			if(m_useOffsetForConstraintFrame)
+			limot.m_normalCFM = (flags & B3_6DOF_FLAGS_CFM_NORM) ? m_linearLimits.m_normalCFM[i] : info->cfm[0];
+			limot.m_stopCFM = (flags & B3_6DOF_FLAGS_CFM_STOP) ? m_linearLimits.m_stopCFM[i] : info->cfm[0];
+			limot.m_stopERP = (flags & B3_6DOF_FLAGS_ERP_STOP) ? m_linearLimits.m_stopERP[i] : info->erp;
+			if (m_useOffsetForConstraintFrame)
 			{
 				int indx1 = (i + 1) % 3;
 				int indx2 = (i + 2) % 3;
-				int rotAllowed = 1; // rotations around orthos to current axis
-				if(m_angularLimits[indx1].m_currentLimit && m_angularLimits[indx2].m_currentLimit)
+				int rotAllowed = 1;  // rotations around orthos to current axis
+				if (m_angularLimits[indx1].m_currentLimit && m_angularLimits[indx2].m_currentLimit)
 				{
 					rotAllowed = 0;
 				}
-				row += get_limit_motor_info2(&limot, transA,transB,linVelA,linVelB,angVelA,angVelB, info, row, axis, 0, rotAllowed);
+				row += get_limit_motor_info2(&limot, transA, transB, linVelA, linVelB, angVelA, angVelB, info, row, axis, 0, rotAllowed);
 			}
 			else
 			{
-				row += get_limit_motor_info2(&limot, transA,transB,linVelA,linVelB,angVelA,angVelB, info, row, axis, 0);
+				row += get_limit_motor_info2(&limot, transA, transB, linVelA, linVelB, angVelA, angVelB, info, row, axis, 0);
 			}
 		}
 	}
 	return row;
 }
 
-
-
-int b3Generic6DofConstraint::setAngularLimits(b3ConstraintInfo2 *info, int row_offset, const b3Transform& transA,const b3Transform& transB,const b3Vector3& linVelA,const b3Vector3& linVelB,const b3Vector3& angVelA,const b3Vector3& angVelB)
+int b3Generic6DofConstraint::setAngularLimits(b3ConstraintInfo2* info, int row_offset, const b3Transform& transA, const b3Transform& transB, const b3Vector3& linVelA, const b3Vector3& linVelB, const b3Vector3& angVelA, const b3Vector3& angVelB)
 {
-	b3Generic6DofConstraint * d6constraint = this;
+	b3Generic6DofConstraint* d6constraint = this;
 	int row = row_offset;
 	//solve angular limits
-	for (int i=0;i<3 ;i++ )
+	for (int i = 0; i < 3; i++)
 	{
-		if(d6constraint->getRotationalLimitMotor(i)->needApplyTorques())
+		if (d6constraint->getRotationalLimitMotor(i)->needApplyTorques())
 		{
 			b3Vector3 axis = d6constraint->getAxis(i);
 			int flags = m_flags >> ((i + 3) * B3_6DOF_FLAGS_AXIS_SHIFT);
-			if(!(flags & B3_6DOF_FLAGS_CFM_NORM))
+			if (!(flags & B3_6DOF_FLAGS_CFM_NORM))
 			{
 				m_angularLimits[i].m_normalCFM = info->cfm[0];
 			}
-			if(!(flags & B3_6DOF_FLAGS_CFM_STOP))
+			if (!(flags & B3_6DOF_FLAGS_CFM_STOP))
 			{
 				m_angularLimits[i].m_stopCFM = info->cfm[0];
 			}
-			if(!(flags & B3_6DOF_FLAGS_ERP_STOP))
+			if (!(flags & B3_6DOF_FLAGS_ERP_STOP))
 			{
 				m_angularLimits[i].m_stopERP = info->erp;
 			}
 			row += get_limit_motor_info2(d6constraint->getRotationalLimitMotor(i),
-												transA,transB,linVelA,linVelB,angVelA,angVelB, info,row,axis,1);
+										 transA, transB, linVelA, linVelB, angVelA, angVelB, info, row, axis, 1);
 		}
 	}
 
 	return row;
 }
 
-
-
-
-void	b3Generic6DofConstraint::updateRHS(b3Scalar	timeStep)
+void b3Generic6DofConstraint::updateRHS(b3Scalar timeStep)
 {
 	(void)timeStep;
-
 }
 
-
-void b3Generic6DofConstraint::setFrames(const b3Transform& frameA, const b3Transform& frameB,const b3RigidBodyData* bodies)
+void b3Generic6DofConstraint::setFrames(const b3Transform& frameA, const b3Transform& frameB, const b3RigidBodyData* bodies)
 {
 	m_frameInA = frameA;
 	m_frameInB = frameB;
@@ -455,33 +400,27 @@ void b3Generic6DofConstraint::setFrames(const b3Transform& frameA, const b3Trans
 	calculateTransforms(bodies);
 }
 
-
-
 b3Vector3 b3Generic6DofConstraint::getAxis(int axis_index) const
 {
 	return m_calculatedAxis[axis_index];
 }
 
-
-b3Scalar	b3Generic6DofConstraint::getRelativePivotPosition(int axisIndex) const
+b3Scalar b3Generic6DofConstraint::getRelativePivotPosition(int axisIndex) const
 {
 	return m_calculatedLinearDiff[axisIndex];
 }
 
-
 b3Scalar b3Generic6DofConstraint::getAngle(int axisIndex) const
 {
 	return m_calculatedAxisAngleDiff[axisIndex];
 }
 
-
-
 void b3Generic6DofConstraint::calcAnchorPos(const b3RigidBodyData* bodies)
 {
 	b3Scalar imA = bodies[m_rbA].m_invMass;
 	b3Scalar imB = bodies[m_rbB].m_invMass;
 	b3Scalar weight;
-	if(imB == b3Scalar(0.0))
+	if (imB == b3Scalar(0.0))
 	{
 		weight = b3Scalar(1.0);
 	}
@@ -495,47 +434,43 @@ void b3Generic6DofConstraint::calcAnchorPos(const b3RigidBodyData* bodies)
 	return;
 }
 
-
-
 void b3Generic6DofConstraint::calculateLinearInfo()
 {
 	m_calculatedLinearDiff = m_calculatedTransformB.getOrigin() - m_calculatedTransformA.getOrigin();
 	m_calculatedLinearDiff = m_calculatedTransformA.getBasis().inverse() * m_calculatedLinearDiff;
-	for(int i = 0; i < 3; i++)
+	for (int i = 0; i < 3; i++)
 	{
 		m_linearLimits.m_currentLinearDiff[i] = m_calculatedLinearDiff[i];
 		m_linearLimits.testLimitValue(i, m_calculatedLinearDiff[i]);
 	}
 }
 
-
-
 int b3Generic6DofConstraint::get_limit_motor_info2(
-	b3RotationalLimitMotor * limot,
-	const b3Transform& transA,const b3Transform& transB,const b3Vector3& linVelA,const b3Vector3& linVelB,const b3Vector3& angVelA,const b3Vector3& angVelB,
-	b3ConstraintInfo2 *info, int row, b3Vector3& ax1, int rotational,int rotAllowed)
+	b3RotationalLimitMotor* limot,
+	const b3Transform& transA, const b3Transform& transB, const b3Vector3& linVelA, const b3Vector3& linVelB, const b3Vector3& angVelA, const b3Vector3& angVelB,
+	b3ConstraintInfo2* info, int row, b3Vector3& ax1, int rotational, int rotAllowed)
 {
-    int srow = row * info->rowskip;
-    bool powered = limot->m_enableMotor;
-    int limit = limot->m_currentLimit;
-    if (powered || limit)
-    {   // if the joint is powered, or has joint limits, add in the extra row
-        b3Scalar *J1 = rotational ? info->m_J1angularAxis : info->m_J1linearAxis;
-        b3Scalar *J2 = rotational ? info->m_J2angularAxis : info->m_J2linearAxis;
+	int srow = row * info->rowskip;
+	bool powered = limot->m_enableMotor;
+	int limit = limot->m_currentLimit;
+	if (powered || limit)
+	{  // if the joint is powered, or has joint limits, add in the extra row
+		b3Scalar* J1 = rotational ? info->m_J1angularAxis : info->m_J1linearAxis;
+		b3Scalar* J2 = rotational ? info->m_J2angularAxis : info->m_J2linearAxis;
 		if (J1)
 		{
-			J1[srow+0] = ax1[0];
-			J1[srow+1] = ax1[1];
-			J1[srow+2] = ax1[2];
+			J1[srow + 0] = ax1[0];
+			J1[srow + 1] = ax1[1];
+			J1[srow + 2] = ax1[2];
 		}
 		if (J2)
 		{
-			J2[srow+0] = -ax1[0];
-			J2[srow+1] = -ax1[1];
-			J2[srow+2] = -ax1[2];
+			J2[srow + 0] = -ax1[0];
+			J2[srow + 1] = -ax1[1];
+			J2[srow + 2] = -ax1[2];
 		}
-		if((!rotational))
-        {
+		if ((!rotational))
+		{
 			if (m_useOffsetForConstraintFrame)
 			{
 				b3Vector3 tmpA, tmpB, relA, relB;
@@ -558,55 +493,56 @@ int b3Generic6DofConstraint::get_limit_motor_info2(
 				relB = orthoB - totalDist * m_factB;
 				tmpA = relA.cross(ax1);
 				tmpB = relB.cross(ax1);
-				if(m_hasStaticBody && (!rotAllowed))
+				if (m_hasStaticBody && (!rotAllowed))
 				{
 					tmpA *= m_factA;
 					tmpB *= m_factB;
 				}
 				int i;
-				for (i=0; i<3; i++) info->m_J1angularAxis[srow+i] = tmpA[i];
-				for (i=0; i<3; i++) info->m_J2angularAxis[srow+i] = -tmpB[i];
-			} else
+				for (i = 0; i < 3; i++) info->m_J1angularAxis[srow + i] = tmpA[i];
+				for (i = 0; i < 3; i++) info->m_J2angularAxis[srow + i] = -tmpB[i];
+			}
+			else
 			{
-				b3Vector3 ltd;	// Linear Torque Decoupling vector
+				b3Vector3 ltd;  // Linear Torque Decoupling vector
 				b3Vector3 c = m_calculatedTransformB.getOrigin() - transA.getOrigin();
 				ltd = c.cross(ax1);
-				info->m_J1angularAxis[srow+0] = ltd[0];
-				info->m_J1angularAxis[srow+1] = ltd[1];
-				info->m_J1angularAxis[srow+2] = ltd[2];
+				info->m_J1angularAxis[srow + 0] = ltd[0];
+				info->m_J1angularAxis[srow + 1] = ltd[1];
+				info->m_J1angularAxis[srow + 2] = ltd[2];
 
 				c = m_calculatedTransformB.getOrigin() - transB.getOrigin();
 				ltd = -c.cross(ax1);
-				info->m_J2angularAxis[srow+0] = ltd[0];
-				info->m_J2angularAxis[srow+1] = ltd[1];
-				info->m_J2angularAxis[srow+2] = ltd[2];
+				info->m_J2angularAxis[srow + 0] = ltd[0];
+				info->m_J2angularAxis[srow + 1] = ltd[1];
+				info->m_J2angularAxis[srow + 2] = ltd[2];
 			}
-        }
-        // if we're limited low and high simultaneously, the joint motor is
-        // ineffective
-        if (limit && (limot->m_loLimit == limot->m_hiLimit)) powered = false;
-        info->m_constraintError[srow] = b3Scalar(0.f);
-        if (powered)
-        {
+		}
+		// if we're limited low and high simultaneously, the joint motor is
+		// ineffective
+		if (limit && (limot->m_loLimit == limot->m_hiLimit)) powered = false;
+		info->m_constraintError[srow] = b3Scalar(0.f);
+		if (powered)
+		{
 			info->cfm[srow] = limot->m_normalCFM;
-            if(!limit)
-            {
+			if (!limit)
+			{
 				b3Scalar tag_vel = rotational ? limot->m_targetVelocity : -limot->m_targetVelocity;
 
-				b3Scalar mot_fact = getMotorFactor(	limot->m_currentPosition, 
-													limot->m_loLimit,
-													limot->m_hiLimit, 
-													tag_vel, 
-													info->fps * limot->m_stopERP);
+				b3Scalar mot_fact = getMotorFactor(limot->m_currentPosition,
+												   limot->m_loLimit,
+												   limot->m_hiLimit,
+												   tag_vel,
+												   info->fps * limot->m_stopERP);
 				info->m_constraintError[srow] += mot_fact * limot->m_targetVelocity;
-                info->m_lowerLimit[srow] = -limot->m_maxMotorForce / info->fps;
-                info->m_upperLimit[srow] = limot->m_maxMotorForce / info->fps;
-            }
-        }
-        if(limit)
-        {
-            b3Scalar k = info->fps * limot->m_stopERP;
-			if(!rotational)
+				info->m_lowerLimit[srow] = -limot->m_maxMotorForce / info->fps;
+				info->m_upperLimit[srow] = limot->m_maxMotorForce / info->fps;
+			}
+		}
+		if (limit)
+		{
+			b3Scalar k = info->fps * limot->m_stopERP;
+			if (!rotational)
 			{
 				info->m_constraintError[srow] += k * limot->m_currentLimitError;
 			}
@@ -615,116 +551,112 @@ int b3Generic6DofConstraint::get_limit_motor_info2(
 				info->m_constraintError[srow] += -k * limot->m_currentLimitError;
 			}
 			info->cfm[srow] = limot->m_stopCFM;
-            if (limot->m_loLimit == limot->m_hiLimit)
-            {   // limited low and high simultaneously
-                info->m_lowerLimit[srow] = -B3_INFINITY;
-                info->m_upperLimit[srow] = B3_INFINITY;
-            }
-            else
-            {
-                if (limit == 1)
-                {
-                    info->m_lowerLimit[srow] = 0;
-                    info->m_upperLimit[srow] = B3_INFINITY;
-                }
-                else
-                {
-                    info->m_lowerLimit[srow] = -B3_INFINITY;
-                    info->m_upperLimit[srow] = 0;
-                }
-                // deal with bounce
-                if (limot->m_bounce > 0)
-                {
-                    // calculate joint velocity
-                    b3Scalar vel;
-                    if (rotational)
-                    {
-                        vel = angVelA.dot(ax1);
-//make sure that if no body -> angVelB == zero vec
-//                        if (body1)
-                            vel -= angVelB.dot(ax1);
-                    }
-                    else
-                    {
-                        vel = linVelA.dot(ax1);
-//make sure that if no body -> angVelB == zero vec
-//                        if (body1)
-                            vel -= linVelB.dot(ax1);
-                    }
-                    // only apply bounce if the velocity is incoming, and if the
-                    // resulting c[] exceeds what we already have.
-                    if (limit == 1)
-                    {
-                        if (vel < 0)
-                        {
-                            b3Scalar newc = -limot->m_bounce* vel;
-                            if (newc > info->m_constraintError[srow]) 
+			if (limot->m_loLimit == limot->m_hiLimit)
+			{  // limited low and high simultaneously
+				info->m_lowerLimit[srow] = -B3_INFINITY;
+				info->m_upperLimit[srow] = B3_INFINITY;
+			}
+			else
+			{
+				if (limit == 1)
+				{
+					info->m_lowerLimit[srow] = 0;
+					info->m_upperLimit[srow] = B3_INFINITY;
+				}
+				else
+				{
+					info->m_lowerLimit[srow] = -B3_INFINITY;
+					info->m_upperLimit[srow] = 0;
+				}
+				// deal with bounce
+				if (limot->m_bounce > 0)
+				{
+					// calculate joint velocity
+					b3Scalar vel;
+					if (rotational)
+					{
+						vel = angVelA.dot(ax1);
+						//make sure that if no body -> angVelB == zero vec
+						//                        if (body1)
+						vel -= angVelB.dot(ax1);
+					}
+					else
+					{
+						vel = linVelA.dot(ax1);
+						//make sure that if no body -> angVelB == zero vec
+						//                        if (body1)
+						vel -= linVelB.dot(ax1);
+					}
+					// only apply bounce if the velocity is incoming, and if the
+					// resulting c[] exceeds what we already have.
+					if (limit == 1)
+					{
+						if (vel < 0)
+						{
+							b3Scalar newc = -limot->m_bounce * vel;
+							if (newc > info->m_constraintError[srow])
 								info->m_constraintError[srow] = newc;
-                        }
-                    }
-                    else
-                    {
-                        if (vel > 0)
-                        {
-                            b3Scalar newc = -limot->m_bounce * vel;
-                            if (newc < info->m_constraintError[srow]) 
+						}
+					}
+					else
+					{
+						if (vel > 0)
+						{
+							b3Scalar newc = -limot->m_bounce * vel;
+							if (newc < info->m_constraintError[srow])
 								info->m_constraintError[srow] = newc;
-                        }
-                    }
-                }
-            }
-        }
-        return 1;
-    }
-    else return 0;
+						}
+					}
+				}
+			}
+		}
+		return 1;
+	}
+	else
+		return 0;
 }
 
-
-
-
-
-
-	///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5). 
-	///If no axis is provided, it uses the default axis for this constraint.
+///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5).
+///If no axis is provided, it uses the default axis for this constraint.
 void b3Generic6DofConstraint::setParam(int num, b3Scalar value, int axis)
 {
-	if((axis >= 0) && (axis < 3))
+	if ((axis >= 0) && (axis < 3))
 	{
-		switch(num)
+		switch (num)
 		{
-			case B3_CONSTRAINT_STOP_ERP : 
+			case B3_CONSTRAINT_STOP_ERP:
 				m_linearLimits.m_stopERP[axis] = value;
 				m_flags |= B3_6DOF_FLAGS_ERP_STOP << (axis * B3_6DOF_FLAGS_AXIS_SHIFT);
 				break;
-			case B3_CONSTRAINT_STOP_CFM : 
+			case B3_CONSTRAINT_STOP_CFM:
 				m_linearLimits.m_stopCFM[axis] = value;
 				m_flags |= B3_6DOF_FLAGS_CFM_STOP << (axis * B3_6DOF_FLAGS_AXIS_SHIFT);
 				break;
-			case B3_CONSTRAINT_CFM : 
+			case B3_CONSTRAINT_CFM:
 				m_linearLimits.m_normalCFM[axis] = value;
 				m_flags |= B3_6DOF_FLAGS_CFM_NORM << (axis * B3_6DOF_FLAGS_AXIS_SHIFT);
 				break;
-			default : 
+			default:
 				b3AssertConstrParams(0);
 		}
 	}
-	else if((axis >=3) && (axis < 6))
+	else if ((axis >= 3) && (axis < 6))
 	{
-		switch(num)
+		switch (num)
 		{
-			case B3_CONSTRAINT_STOP_ERP : 
+			case B3_CONSTRAINT_STOP_ERP:
 				m_angularLimits[axis - 3].m_stopERP = value;
 				m_flags |= B3_6DOF_FLAGS_ERP_STOP << (axis * B3_6DOF_FLAGS_AXIS_SHIFT);
 				break;
-			case B3_CONSTRAINT_STOP_CFM : 
+			case B3_CONSTRAINT_STOP_CFM:
 				m_angularLimits[axis - 3].m_stopCFM = value;
 				m_flags |= B3_6DOF_FLAGS_CFM_STOP << (axis * B3_6DOF_FLAGS_AXIS_SHIFT);
 				break;
-			case B3_CONSTRAINT_CFM : 
+			case B3_CONSTRAINT_CFM:
 				m_angularLimits[axis - 3].m_normalCFM = value;
 				m_flags |= B3_6DOF_FLAGS_CFM_NORM << (axis * B3_6DOF_FLAGS_AXIS_SHIFT);
 				break;
-			default : 
+			default:
 				b3AssertConstrParams(0);
 		}
 	}
@@ -734,47 +666,47 @@ void b3Generic6DofConstraint::setParam(int num, b3Scalar value, int axis)
 	}
 }
 
-	///return the local value of parameter
-b3Scalar b3Generic6DofConstraint::getParam(int num, int axis) const 
+///return the local value of parameter
+b3Scalar b3Generic6DofConstraint::getParam(int num, int axis) const
 {
 	b3Scalar retVal = 0;
-	if((axis >= 0) && (axis < 3))
+	if ((axis >= 0) && (axis < 3))
 	{
-		switch(num)
+		switch (num)
 		{
-			case B3_CONSTRAINT_STOP_ERP : 
+			case B3_CONSTRAINT_STOP_ERP:
 				b3AssertConstrParams(m_flags & (B3_6DOF_FLAGS_ERP_STOP << (axis * B3_6DOF_FLAGS_AXIS_SHIFT)));
 				retVal = m_linearLimits.m_stopERP[axis];
 				break;
-			case B3_CONSTRAINT_STOP_CFM : 
+			case B3_CONSTRAINT_STOP_CFM:
 				b3AssertConstrParams(m_flags & (B3_6DOF_FLAGS_CFM_STOP << (axis * B3_6DOF_FLAGS_AXIS_SHIFT)));
 				retVal = m_linearLimits.m_stopCFM[axis];
 				break;
-			case B3_CONSTRAINT_CFM : 
+			case B3_CONSTRAINT_CFM:
 				b3AssertConstrParams(m_flags & (B3_6DOF_FLAGS_CFM_NORM << (axis * B3_6DOF_FLAGS_AXIS_SHIFT)));
 				retVal = m_linearLimits.m_normalCFM[axis];
 				break;
-			default : 
+			default:
 				b3AssertConstrParams(0);
 		}
 	}
-	else if((axis >=3) && (axis < 6))
+	else if ((axis >= 3) && (axis < 6))
 	{
-		switch(num)
+		switch (num)
 		{
-			case B3_CONSTRAINT_STOP_ERP : 
+			case B3_CONSTRAINT_STOP_ERP:
 				b3AssertConstrParams(m_flags & (B3_6DOF_FLAGS_ERP_STOP << (axis * B3_6DOF_FLAGS_AXIS_SHIFT)));
 				retVal = m_angularLimits[axis - 3].m_stopERP;
 				break;
-			case B3_CONSTRAINT_STOP_CFM : 
+			case B3_CONSTRAINT_STOP_CFM:
 				b3AssertConstrParams(m_flags & (B3_6DOF_FLAGS_CFM_STOP << (axis * B3_6DOF_FLAGS_AXIS_SHIFT)));
 				retVal = m_angularLimits[axis - 3].m_stopCFM;
 				break;
-			case B3_CONSTRAINT_CFM : 
+			case B3_CONSTRAINT_CFM:
 				b3AssertConstrParams(m_flags & (B3_6DOF_FLAGS_CFM_NORM << (axis * B3_6DOF_FLAGS_AXIS_SHIFT)));
 				retVal = m_angularLimits[axis - 3].m_normalCFM;
 				break;
-			default : 
+			default:
 				b3AssertConstrParams(0);
 		}
 	}
@@ -785,23 +717,21 @@ b3Scalar b3Generic6DofConstraint::getParam(int num, int axis) const
 	return retVal;
 }
 
- 
-
-void b3Generic6DofConstraint::setAxis(const b3Vector3& axis1,const b3Vector3& axis2, const b3RigidBodyData* bodies)
+void b3Generic6DofConstraint::setAxis(const b3Vector3& axis1, const b3Vector3& axis2, const b3RigidBodyData* bodies)
 {
 	b3Vector3 zAxis = axis1.normalized();
 	b3Vector3 yAxis = axis2.normalized();
-	b3Vector3 xAxis = yAxis.cross(zAxis); // we want right coordinate system
-	
+	b3Vector3 xAxis = yAxis.cross(zAxis);  // we want right coordinate system
+
 	b3Transform frameInW;
 	frameInW.setIdentity();
-	frameInW.getBasis().setValue(	xAxis[0], yAxis[0], zAxis[0],	
-	                                xAxis[1], yAxis[1], zAxis[1],
-	                               xAxis[2], yAxis[2], zAxis[2]);
-	
+	frameInW.getBasis().setValue(xAxis[0], yAxis[0], zAxis[0],
+								 xAxis[1], yAxis[1], zAxis[1],
+								 xAxis[2], yAxis[2], zAxis[2]);
+
 	// now get constraint frame in local coordinate systems
 	m_frameInA = getCenterOfMassTransform(bodies[m_rbA]).inverse() * frameInW;
 	m_frameInB = getCenterOfMassTransform(bodies[m_rbB]).inverse() * frameInW;
-	
+
 	calculateTransforms(bodies);
 }
diff --git a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3Generic6DofConstraint.h b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3Generic6DofConstraint.h
index 169b1b94ade..1597809db3b 100644
--- a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3Generic6DofConstraint.h
+++ b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3Generic6DofConstraint.h
@@ -23,7 +23,6 @@ email: projectileman@yahoo.com
 http://gimpact.sf.net
 */
 
-
 #ifndef B3_GENERIC_6DOF_CONSTRAINT_H
 #define B3_GENERIC_6DOF_CONSTRAINT_H
 
@@ -33,88 +32,83 @@ http://gimpact.sf.net
 
 struct b3RigidBodyData;
 
-
-
-
 //! Rotation Limit structure for generic joints
 class b3RotationalLimitMotor
 {
 public:
-    //! limit_parameters
-    //!@{
-    b3Scalar m_loLimit;//!< joint limit
-    b3Scalar m_hiLimit;//!< joint limit
-    b3Scalar m_targetVelocity;//!< target motor velocity
-    b3Scalar m_maxMotorForce;//!< max force on motor
-    b3Scalar m_maxLimitForce;//!< max force on limit
-    b3Scalar m_damping;//!< Damping.
-    b3Scalar m_limitSoftness;//! Relaxation factor
-    b3Scalar m_normalCFM;//!< Constraint force mixing factor
-    b3Scalar m_stopERP;//!< Error tolerance factor when joint is at limit
-    b3Scalar m_stopCFM;//!< Constraint force mixing factor when joint is at limit
-    b3Scalar m_bounce;//!< restitution factor
-    bool m_enableMotor;
+	//! limit_parameters
+	//!@{
+	b3Scalar m_loLimit;         //!< joint limit
+	b3Scalar m_hiLimit;         //!< joint limit
+	b3Scalar m_targetVelocity;  //!< target motor velocity
+	b3Scalar m_maxMotorForce;   //!< max force on motor
+	b3Scalar m_maxLimitForce;   //!< max force on limit
+	b3Scalar m_damping;         //!< Damping.
+	b3Scalar m_limitSoftness;   //! Relaxation factor
+	b3Scalar m_normalCFM;       //!< Constraint force mixing factor
+	b3Scalar m_stopERP;         //!< Error tolerance factor when joint is at limit
+	b3Scalar m_stopCFM;         //!< Constraint force mixing factor when joint is at limit
+	b3Scalar m_bounce;          //!< restitution factor
+	bool m_enableMotor;
 
-    //!@}
+	//!@}
 
-    //! temp_variables
-    //!@{
-    b3Scalar m_currentLimitError;//!  How much is violated this limit
-    b3Scalar m_currentPosition;     //!  current value of angle 
-    int m_currentLimit;//!< 0=free, 1=at lo limit, 2=at hi limit
-    b3Scalar m_accumulatedImpulse;
-    //!@}
+	//! temp_variables
+	//!@{
+	b3Scalar m_currentLimitError;  //!  How much is violated this limit
+	b3Scalar m_currentPosition;    //!  current value of angle
+	int m_currentLimit;            //!< 0=free, 1=at lo limit, 2=at hi limit
+	b3Scalar m_accumulatedImpulse;
+	//!@}
 
-    b3RotationalLimitMotor()
-    {
-    	m_accumulatedImpulse = 0.f;
-        m_targetVelocity = 0;
-        m_maxMotorForce = 6.0f;
-        m_maxLimitForce = 300.0f;
-        m_loLimit = 1.0f;
-        m_hiLimit = -1.0f;
+	b3RotationalLimitMotor()
+	{
+		m_accumulatedImpulse = 0.f;
+		m_targetVelocity = 0;
+		m_maxMotorForce = 6.0f;
+		m_maxLimitForce = 300.0f;
+		m_loLimit = 1.0f;
+		m_hiLimit = -1.0f;
 		m_normalCFM = 0.f;
 		m_stopERP = 0.2f;
 		m_stopCFM = 0.f;
-        m_bounce = 0.0f;
-        m_damping = 1.0f;
-        m_limitSoftness = 0.5f;
-        m_currentLimit = 0;
-        m_currentLimitError = 0;
-        m_enableMotor = false;
-    }
+		m_bounce = 0.0f;
+		m_damping = 1.0f;
+		m_limitSoftness = 0.5f;
+		m_currentLimit = 0;
+		m_currentLimitError = 0;
+		m_enableMotor = false;
+	}
 
-    b3RotationalLimitMotor(const b3RotationalLimitMotor & limot)
-    {
-        m_targetVelocity = limot.m_targetVelocity;
-        m_maxMotorForce = limot.m_maxMotorForce;
-        m_limitSoftness = limot.m_limitSoftness;
-        m_loLimit = limot.m_loLimit;
-        m_hiLimit = limot.m_hiLimit;
+	b3RotationalLimitMotor(const b3RotationalLimitMotor& limot)
+	{
+		m_targetVelocity = limot.m_targetVelocity;
+		m_maxMotorForce = limot.m_maxMotorForce;
+		m_limitSoftness = limot.m_limitSoftness;
+		m_loLimit = limot.m_loLimit;
+		m_hiLimit = limot.m_hiLimit;
 		m_normalCFM = limot.m_normalCFM;
 		m_stopERP = limot.m_stopERP;
-		m_stopCFM =	limot.m_stopCFM;
-        m_bounce = limot.m_bounce;
-        m_currentLimit = limot.m_currentLimit;
-        m_currentLimitError = limot.m_currentLimitError;
-        m_enableMotor = limot.m_enableMotor;
-    }
-
-
+		m_stopCFM = limot.m_stopCFM;
+		m_bounce = limot.m_bounce;
+		m_currentLimit = limot.m_currentLimit;
+		m_currentLimitError = limot.m_currentLimitError;
+		m_enableMotor = limot.m_enableMotor;
+	}
 
 	//! Is limited
-    bool isLimited()
-    {
-    	if(m_loLimit > m_hiLimit) return false;
-    	return true;
-    }
+	bool isLimited()
+	{
+		if (m_loLimit > m_hiLimit) return false;
+		return true;
+	}
 
 	//! Need apply correction
-    bool needApplyTorques()
-    {
-    	if(m_currentLimit == 0 && m_enableMotor == false) return false;
-    	return true;
-    }
+	bool needApplyTorques()
+	{
+		if (m_currentLimit == 0 && m_enableMotor == false) return false;
+		return true;
+	}
 
 	//! calculates  error
 	/*!
@@ -123,104 +117,98 @@ public:
 	int testLimitValue(b3Scalar test_value);
 
 	//! apply the correction impulses for two bodies
-    b3Scalar solveAngularLimits(b3Scalar timeStep,b3Vector3& axis, b3Scalar jacDiagABInv,b3RigidBodyData * body0, b3RigidBodyData * body1);
-
+	b3Scalar solveAngularLimits(b3Scalar timeStep, b3Vector3& axis, b3Scalar jacDiagABInv, b3RigidBodyData* body0, b3RigidBodyData* body1);
 };
 
-
-
 class b3TranslationalLimitMotor
 {
 public:
-	b3Vector3 m_lowerLimit;//!< the constraint lower limits
-    b3Vector3 m_upperLimit;//!< the constraint upper limits
-    b3Vector3 m_accumulatedImpulse;
-    //! Linear_Limit_parameters
-    //!@{
-	b3Vector3	m_normalCFM;//!< Constraint force mixing factor
-    b3Vector3	m_stopERP;//!< Error tolerance factor when joint is at limit
-	b3Vector3	m_stopCFM;//!< Constraint force mixing factor when joint is at limit
-    b3Vector3	m_targetVelocity;//!< target motor velocity
-    b3Vector3	m_maxMotorForce;//!< max force on motor
-    b3Vector3	m_currentLimitError;//!  How much is violated this limit
-    b3Vector3	m_currentLinearDiff;//!  Current relative offset of constraint frames
-	b3Scalar	m_limitSoftness;//!< Softness for linear limit
-    b3Scalar	m_damping;//!< Damping for linear limit
-    b3Scalar	m_restitution;//! Bounce parameter for linear limit
+	b3Vector3 m_lowerLimit;  //!< the constraint lower limits
+	b3Vector3 m_upperLimit;  //!< the constraint upper limits
+	b3Vector3 m_accumulatedImpulse;
+	//! Linear_Limit_parameters
+	//!@{
+	b3Vector3 m_normalCFM;          //!< Constraint force mixing factor
+	b3Vector3 m_stopERP;            //!< Error tolerance factor when joint is at limit
+	b3Vector3 m_stopCFM;            //!< Constraint force mixing factor when joint is at limit
+	b3Vector3 m_targetVelocity;     //!< target motor velocity
+	b3Vector3 m_maxMotorForce;      //!< max force on motor
+	b3Vector3 m_currentLimitError;  //!  How much is violated this limit
+	b3Vector3 m_currentLinearDiff;  //!  Current relative offset of constraint frames
+	b3Scalar m_limitSoftness;       //!< Softness for linear limit
+	b3Scalar m_damping;             //!< Damping for linear limit
+	b3Scalar m_restitution;         //! Bounce parameter for linear limit
 	//!@}
-	bool		m_enableMotor[3];
-	int			m_currentLimit[3];//!< 0=free, 1=at lower limit, 2=at upper limit
+	bool m_enableMotor[3];
+	int m_currentLimit[3];  //!< 0=free, 1=at lower limit, 2=at upper limit
 
-    b3TranslationalLimitMotor()
-    {
-    	m_lowerLimit.setValue(0.f,0.f,0.f);
-    	m_upperLimit.setValue(0.f,0.f,0.f);
-    	m_accumulatedImpulse.setValue(0.f,0.f,0.f);
+	b3TranslationalLimitMotor()
+	{
+		m_lowerLimit.setValue(0.f, 0.f, 0.f);
+		m_upperLimit.setValue(0.f, 0.f, 0.f);
+		m_accumulatedImpulse.setValue(0.f, 0.f, 0.f);
 		m_normalCFM.setValue(0.f, 0.f, 0.f);
 		m_stopERP.setValue(0.2f, 0.2f, 0.2f);
 		m_stopCFM.setValue(0.f, 0.f, 0.f);
 
-    	m_limitSoftness = 0.7f;
-    	m_damping = b3Scalar(1.0f);
-    	m_restitution = b3Scalar(0.5f);
-		for(int i=0; i < 3; i++) 
+		m_limitSoftness = 0.7f;
+		m_damping = b3Scalar(1.0f);
+		m_restitution = b3Scalar(0.5f);
+		for (int i = 0; i < 3; i++)
 		{
 			m_enableMotor[i] = false;
 			m_targetVelocity[i] = b3Scalar(0.f);
 			m_maxMotorForce[i] = b3Scalar(0.f);
 		}
-    }
+	}
 
-    b3TranslationalLimitMotor(const b3TranslationalLimitMotor & other )
-    {
-    	m_lowerLimit = other.m_lowerLimit;
-    	m_upperLimit = other.m_upperLimit;
-    	m_accumulatedImpulse = other.m_accumulatedImpulse;
+	b3TranslationalLimitMotor(const b3TranslationalLimitMotor& other)
+	{
+		m_lowerLimit = other.m_lowerLimit;
+		m_upperLimit = other.m_upperLimit;
+		m_accumulatedImpulse = other.m_accumulatedImpulse;
 
-    	m_limitSoftness = other.m_limitSoftness ;
-    	m_damping = other.m_damping;
-    	m_restitution = other.m_restitution;
+		m_limitSoftness = other.m_limitSoftness;
+		m_damping = other.m_damping;
+		m_restitution = other.m_restitution;
 		m_normalCFM = other.m_normalCFM;
 		m_stopERP = other.m_stopERP;
 		m_stopCFM = other.m_stopCFM;
 
-		for(int i=0; i < 3; i++) 
+		for (int i = 0; i < 3; i++)
 		{
 			m_enableMotor[i] = other.m_enableMotor[i];
 			m_targetVelocity[i] = other.m_targetVelocity[i];
 			m_maxMotorForce[i] = other.m_maxMotorForce[i];
 		}
-    }
+	}
 
-    //! Test limit
+	//! Test limit
 	/*!
     - free means upper < lower,
     - locked means upper == lower
     - limited means upper > lower
     - limitIndex: first 3 are linear, next 3 are angular
     */
-    inline bool	isLimited(int limitIndex)
-    {
-       return (m_upperLimit[limitIndex] >= m_lowerLimit[limitIndex]);
-    }
-    inline bool needApplyForce(int limitIndex)
-    {
-    	if(m_currentLimit[limitIndex] == 0 && m_enableMotor[limitIndex] == false) return false;
-    	return true;
-    }
+	inline bool isLimited(int limitIndex)
+	{
+		return (m_upperLimit[limitIndex] >= m_lowerLimit[limitIndex]);
+	}
+	inline bool needApplyForce(int limitIndex)
+	{
+		if (m_currentLimit[limitIndex] == 0 && m_enableMotor[limitIndex] == false) return false;
+		return true;
+	}
 	int testLimitValue(int limitIndex, b3Scalar test_value);
 
-
-    b3Scalar solveLinearAxis(
-    	b3Scalar timeStep,
-        b3Scalar jacDiagABInv,
-        b3RigidBodyData& body1,const b3Vector3 &pointInA,
-        b3RigidBodyData& body2,const b3Vector3 &pointInB,
-        int limit_index,
-        const b3Vector3 & axis_normal_on_a,
-		const b3Vector3 & anchorPos);
-
-
+	b3Scalar solveLinearAxis(
+		b3Scalar timeStep,
+		b3Scalar jacDiagABInv,
+		b3RigidBodyData& body1, const b3Vector3& pointInA,
+		b3RigidBodyData& body2, const b3Vector3& pointInB,
+		int limit_index,
+		const b3Vector3& axis_normal_on_a,
+		const b3Vector3& anchorPos);
 };
 
 enum b36DofFlags
@@ -229,8 +217,7 @@ enum b36DofFlags
 	B3_6DOF_FLAGS_CFM_STOP = 2,
 	B3_6DOF_FLAGS_ERP_STOP = 4
 };
-#define B3_6DOF_FLAGS_AXIS_SHIFT 3 // bits per axis
-
+#define B3_6DOF_FLAGS_AXIS_SHIFT 3  // bits per axis
 
 /// b3Generic6DofConstraint between two rigidbodies each with a pivotpoint that descibes the axis location in local space
 /*!
@@ -268,240 +255,229 @@ This brings support for limit parameters and motors. </li>
 </ul>
 
 */
-B3_ATTRIBUTE_ALIGNED16(class) b3Generic6DofConstraint : public b3TypedConstraint
+B3_ATTRIBUTE_ALIGNED16(class)
+b3Generic6DofConstraint : public b3TypedConstraint
 {
 protected:
-
 	//! relative_frames
-    //!@{
-	b3Transform	m_frameInA;//!< the constraint space w.r.t body A
-    b3Transform	m_frameInB;//!< the constraint space w.r.t body B
-    //!@}
-
-    //! Jacobians
-    //!@{
-//    b3JacobianEntry	m_jacLinear[3];//!< 3 orthogonal linear constraints
-//    b3JacobianEntry	m_jacAng[3];//!< 3 orthogonal angular constraints
-    //!@}
-
-	//! Linear_Limit_parameters
-    //!@{
-    b3TranslationalLimitMotor m_linearLimits;
-    //!@}
-
-
-    //! hinge_parameters
-    //!@{
-    b3RotationalLimitMotor m_angularLimits[3];
+	//!@{
+	b3Transform m_frameInA;  //!< the constraint space w.r.t body A
+	b3Transform m_frameInB;  //!< the constraint space w.r.t body B
 	//!@}
 
+	//! Jacobians
+	//!@{
+	//    b3JacobianEntry	m_jacLinear[3];//!< 3 orthogonal linear constraints
+	//    b3JacobianEntry	m_jacAng[3];//!< 3 orthogonal angular constraints
+	//!@}
+
+	//! Linear_Limit_parameters
+	//!@{
+	b3TranslationalLimitMotor m_linearLimits;
+	//!@}
+
+	//! hinge_parameters
+	//!@{
+	b3RotationalLimitMotor m_angularLimits[3];
+	//!@}
 
 protected:
-    //! temporal variables
-    //!@{
-    b3Transform m_calculatedTransformA;
-    b3Transform m_calculatedTransformB;
-    b3Vector3 m_calculatedAxisAngleDiff;
-    b3Vector3 m_calculatedAxis[3];
-    b3Vector3 m_calculatedLinearDiff;
-    b3Scalar m_timeStep;
-	b3Scalar	m_factA;
-	b3Scalar	m_factB;
-	bool		m_hasStaticBody;
-    
-	b3Vector3 m_AnchorPos; // point betwen pivots of bodies A and B to solve linear axes
+	//! temporal variables
+	//!@{
+	b3Transform m_calculatedTransformA;
+	b3Transform m_calculatedTransformB;
+	b3Vector3 m_calculatedAxisAngleDiff;
+	b3Vector3 m_calculatedAxis[3];
+	b3Vector3 m_calculatedLinearDiff;
+	b3Scalar m_timeStep;
+	b3Scalar m_factA;
+	b3Scalar m_factB;
+	bool m_hasStaticBody;
 
-    bool	m_useLinearReferenceFrameA;
-	bool	m_useOffsetForConstraintFrame;
-    
-	int		m_flags;
+	b3Vector3 m_AnchorPos;  // point betwen pivots of bodies A and B to solve linear axes
 
-    //!@}
+	bool m_useLinearReferenceFrameA;
+	bool m_useOffsetForConstraintFrame;
 
-    b3Generic6DofConstraint&	operator=(b3Generic6DofConstraint&	other)
-    {
-        b3Assert(0);
-        (void) other;
-        return *this;
-    }
+	int m_flags;
 
+	//!@}
 
-	int setAngularLimits(b3ConstraintInfo2 *info, int row_offset,const b3Transform& transA,const b3Transform& transB,const b3Vector3& linVelA,const b3Vector3& linVelB,const b3Vector3& angVelA,const b3Vector3& angVelB);
+	b3Generic6DofConstraint& operator=(b3Generic6DofConstraint& other)
+	{
+		b3Assert(0);
+		(void)other;
+		return *this;
+	}
 
-	int setLinearLimits(b3ConstraintInfo2 *info, int row, const b3Transform& transA,const b3Transform& transB,const b3Vector3& linVelA,const b3Vector3& linVelB,const b3Vector3& angVelA,const b3Vector3& angVelB);
+	int setAngularLimits(b3ConstraintInfo2 * info, int row_offset, const b3Transform& transA, const b3Transform& transB, const b3Vector3& linVelA, const b3Vector3& linVelB, const b3Vector3& angVelA, const b3Vector3& angVelB);
 
+	int setLinearLimits(b3ConstraintInfo2 * info, int row, const b3Transform& transA, const b3Transform& transB, const b3Vector3& linVelA, const b3Vector3& linVelB, const b3Vector3& angVelA, const b3Vector3& angVelB);
 
 	// tests linear limits
 	void calculateLinearInfo();
 
 	//! calcs the euler angles between the two bodies.
-    void calculateAngleInfo();
-
-
+	void calculateAngleInfo();
 
 public:
-
 	B3_DECLARE_ALIGNED_ALLOCATOR();
-	
-    b3Generic6DofConstraint(int rbA, int rbB, const b3Transform& frameInA, const b3Transform& frameInB ,bool useLinearReferenceFrameA,const b3RigidBodyData* bodies);
-    
+
+	b3Generic6DofConstraint(int rbA, int rbB, const b3Transform& frameInA, const b3Transform& frameInB, bool useLinearReferenceFrameA, const b3RigidBodyData* bodies);
+
 	//! Calcs global transform of the offsets
 	/*!
 	Calcs the global transform for the joint offset for body A an B, and also calcs the agle differences between the bodies.
 	\sa b3Generic6DofConstraint.getCalculatedTransformA , b3Generic6DofConstraint.getCalculatedTransformB, b3Generic6DofConstraint.calculateAngleInfo
 	*/
-    void calculateTransforms(const b3Transform& transA,const b3Transform& transB,const b3RigidBodyData* bodies);
+	void calculateTransforms(const b3Transform& transA, const b3Transform& transB, const b3RigidBodyData* bodies);
 
 	void calculateTransforms(const b3RigidBodyData* bodies);
 
 	//! Gets the global transform of the offset for body A
-    /*!
+	/*!
     \sa b3Generic6DofConstraint.getFrameOffsetA, b3Generic6DofConstraint.getFrameOffsetB, b3Generic6DofConstraint.calculateAngleInfo.
     */
-    const b3Transform & getCalculatedTransformA() const
-    {
-    	return m_calculatedTransformA;
-    }
+	const b3Transform& getCalculatedTransformA() const
+	{
+		return m_calculatedTransformA;
+	}
 
-    //! Gets the global transform of the offset for body B
-    /*!
+	//! Gets the global transform of the offset for body B
+	/*!
     \sa b3Generic6DofConstraint.getFrameOffsetA, b3Generic6DofConstraint.getFrameOffsetB, b3Generic6DofConstraint.calculateAngleInfo.
     */
-    const b3Transform & getCalculatedTransformB() const
-    {
-    	return m_calculatedTransformB;
-    }
+	const b3Transform& getCalculatedTransformB() const
+	{
+		return m_calculatedTransformB;
+	}
 
-    const b3Transform & getFrameOffsetA() const
-    {
-    	return m_frameInA;
-    }
+	const b3Transform& getFrameOffsetA() const
+	{
+		return m_frameInA;
+	}
 
-    const b3Transform & getFrameOffsetB() const
-    {
-    	return m_frameInB;
-    }
+	const b3Transform& getFrameOffsetB() const
+	{
+		return m_frameInB;
+	}
 
+	b3Transform& getFrameOffsetA()
+	{
+		return m_frameInA;
+	}
 
-    b3Transform & getFrameOffsetA()
-    {
-    	return m_frameInA;
-    }
+	b3Transform& getFrameOffsetB()
+	{
+		return m_frameInB;
+	}
 
-    b3Transform & getFrameOffsetB()
-    {
-    	return m_frameInB;
-    }
+	virtual void getInfo1(b3ConstraintInfo1 * info, const b3RigidBodyData* bodies);
 
+	void getInfo1NonVirtual(b3ConstraintInfo1 * info, const b3RigidBodyData* bodies);
 
+	virtual void getInfo2(b3ConstraintInfo2 * info, const b3RigidBodyData* bodies);
 
-	virtual void getInfo1 (b3ConstraintInfo1* info,const b3RigidBodyData* bodies);
+	void getInfo2NonVirtual(b3ConstraintInfo2 * info, const b3Transform& transA, const b3Transform& transB, const b3Vector3& linVelA, const b3Vector3& linVelB, const b3Vector3& angVelA, const b3Vector3& angVelB, const b3RigidBodyData* bodies);
 
-	void getInfo1NonVirtual (b3ConstraintInfo1* info,const b3RigidBodyData* bodies);
-
-	virtual void getInfo2 (b3ConstraintInfo2* info,const b3RigidBodyData* bodies);
-
-	void getInfo2NonVirtual (b3ConstraintInfo2* info,const b3Transform& transA,const b3Transform& transB,const b3Vector3& linVelA,const b3Vector3& linVelB,const b3Vector3& angVelA,const b3Vector3& angVelB,const b3RigidBodyData* bodies);
-
-
-    void	updateRHS(b3Scalar	timeStep);
+	void updateRHS(b3Scalar timeStep);
 
 	//! Get the rotation axis in global coordinates
-    b3Vector3 getAxis(int axis_index) const;
+	b3Vector3 getAxis(int axis_index) const;
 
-    //! Get the relative Euler angle
-    /*!
+	//! Get the relative Euler angle
+	/*!
 	\pre b3Generic6DofConstraint::calculateTransforms() must be called previously.
 	*/
-    b3Scalar getAngle(int axis_index) const;
+	b3Scalar getAngle(int axis_index) const;
 
 	//! Get the relative position of the constraint pivot
-    /*!
+	/*!
 	\pre b3Generic6DofConstraint::calculateTransforms() must be called previously.
 	*/
 	b3Scalar getRelativePivotPosition(int axis_index) const;
 
-	void setFrames(const b3Transform & frameA, const b3Transform & frameB, const b3RigidBodyData* bodies);
+	void setFrames(const b3Transform& frameA, const b3Transform& frameB, const b3RigidBodyData* bodies);
 
 	//! Test angular limit.
 	/*!
 	Calculates angular correction and returns true if limit needs to be corrected.
 	\pre b3Generic6DofConstraint::calculateTransforms() must be called previously.
 	*/
-    bool testAngularLimitMotor(int axis_index);
+	bool testAngularLimitMotor(int axis_index);
 
-    void	setLinearLowerLimit(const b3Vector3& linearLower)
-    {
-    	m_linearLimits.m_lowerLimit = linearLower;
-    }
+	void setLinearLowerLimit(const b3Vector3& linearLower)
+	{
+		m_linearLimits.m_lowerLimit = linearLower;
+	}
 
-	void	getLinearLowerLimit(b3Vector3& linearLower)
+	void getLinearLowerLimit(b3Vector3 & linearLower)
 	{
 		linearLower = m_linearLimits.m_lowerLimit;
 	}
 
-	void	setLinearUpperLimit(const b3Vector3& linearUpper)
+	void setLinearUpperLimit(const b3Vector3& linearUpper)
 	{
 		m_linearLimits.m_upperLimit = linearUpper;
 	}
 
-	void	getLinearUpperLimit(b3Vector3& linearUpper)
+	void getLinearUpperLimit(b3Vector3 & linearUpper)
 	{
 		linearUpper = m_linearLimits.m_upperLimit;
 	}
 
-    void	setAngularLowerLimit(const b3Vector3& angularLower)
-    {
-		for(int i = 0; i < 3; i++) 
-			m_angularLimits[i].m_loLimit = b3NormalizeAngle(angularLower[i]);
-    }
-
-	void	getAngularLowerLimit(b3Vector3& angularLower)
+	void setAngularLowerLimit(const b3Vector3& angularLower)
 	{
-		for(int i = 0; i < 3; i++) 
+		for (int i = 0; i < 3; i++)
+			m_angularLimits[i].m_loLimit = b3NormalizeAngle(angularLower[i]);
+	}
+
+	void getAngularLowerLimit(b3Vector3 & angularLower)
+	{
+		for (int i = 0; i < 3; i++)
 			angularLower[i] = m_angularLimits[i].m_loLimit;
 	}
 
-    void	setAngularUpperLimit(const b3Vector3& angularUpper)
-    {
-		for(int i = 0; i < 3; i++)
-			m_angularLimits[i].m_hiLimit = b3NormalizeAngle(angularUpper[i]);
-    }
-
-	void	getAngularUpperLimit(b3Vector3& angularUpper)
+	void setAngularUpperLimit(const b3Vector3& angularUpper)
 	{
-		for(int i = 0; i < 3; i++)
+		for (int i = 0; i < 3; i++)
+			m_angularLimits[i].m_hiLimit = b3NormalizeAngle(angularUpper[i]);
+	}
+
+	void getAngularUpperLimit(b3Vector3 & angularUpper)
+	{
+		for (int i = 0; i < 3; i++)
 			angularUpper[i] = m_angularLimits[i].m_hiLimit;
 	}
 
 	//! Retrieves the angular limit informacion
-    b3RotationalLimitMotor * getRotationalLimitMotor(int index)
-    {
-    	return &m_angularLimits[index];
-    }
+	b3RotationalLimitMotor* getRotationalLimitMotor(int index)
+	{
+		return &m_angularLimits[index];
+	}
 
-    //! Retrieves the  limit informacion
-    b3TranslationalLimitMotor * getTranslationalLimitMotor()
-    {
-    	return &m_linearLimits;
-    }
+	//! Retrieves the  limit informacion
+	b3TranslationalLimitMotor* getTranslationalLimitMotor()
+	{
+		return &m_linearLimits;
+	}
 
-    //first 3 are linear, next 3 are angular
-    void setLimit(int axis, b3Scalar lo, b3Scalar hi)
-    {
-    	if(axis<3)
-    	{
-    		m_linearLimits.m_lowerLimit[axis] = lo;
-    		m_linearLimits.m_upperLimit[axis] = hi;
-    	}
-    	else
-    	{
+	//first 3 are linear, next 3 are angular
+	void setLimit(int axis, b3Scalar lo, b3Scalar hi)
+	{
+		if (axis < 3)
+		{
+			m_linearLimits.m_lowerLimit[axis] = lo;
+			m_linearLimits.m_upperLimit[axis] = hi;
+		}
+		else
+		{
 			lo = b3NormalizeAngle(lo);
 			hi = b3NormalizeAngle(hi);
-    		m_angularLimits[axis-3].m_loLimit = lo;
-    		m_angularLimits[axis-3].m_hiLimit = hi;
-    	}
-    }
+			m_angularLimits[axis - 3].m_loLimit = lo;
+			m_angularLimits[axis - 3].m_hiLimit = hi;
+		}
+	}
 
 	//! Test limit
 	/*!
@@ -510,41 +486,32 @@ public:
     - limited means upper > lower
     - limitIndex: first 3 are linear, next 3 are angular
     */
-    bool	isLimited(int limitIndex)
-    {
-    	if(limitIndex<3)
-    	{
+	bool isLimited(int limitIndex)
+	{
+		if (limitIndex < 3)
+		{
 			return m_linearLimits.isLimited(limitIndex);
+		}
+		return m_angularLimits[limitIndex - 3].isLimited();
+	}
 
-    	}
-        return m_angularLimits[limitIndex-3].isLimited();
-    }
+	virtual void calcAnchorPos(const b3RigidBodyData* bodies);  // overridable
 
-	virtual void calcAnchorPos(const b3RigidBodyData* bodies); // overridable
-
-	int get_limit_motor_info2(	b3RotationalLimitMotor * limot,
-								const b3Transform& transA,const b3Transform& transB,const b3Vector3& linVelA,const b3Vector3& linVelB,const b3Vector3& angVelA,const b3Vector3& angVelB,
-								b3ConstraintInfo2 *info, int row, b3Vector3& ax1, int rotational, int rotAllowed = false);
+	int get_limit_motor_info2(b3RotationalLimitMotor * limot,
+							  const b3Transform& transA, const b3Transform& transB, const b3Vector3& linVelA, const b3Vector3& linVelB, const b3Vector3& angVelA, const b3Vector3& angVelB,
+							  b3ConstraintInfo2* info, int row, b3Vector3& ax1, int rotational, int rotAllowed = false);
 
 	// access for UseFrameOffset
 	bool getUseFrameOffset() { return m_useOffsetForConstraintFrame; }
 	void setUseFrameOffset(bool frameOffsetOnOff) { m_useOffsetForConstraintFrame = frameOffsetOnOff; }
 
-	///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5). 
+	///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5).
 	///If no axis is provided, it uses the default axis for this constraint.
-	virtual	void setParam(int num, b3Scalar value, int axis = -1);
+	virtual void setParam(int num, b3Scalar value, int axis = -1);
 	///return the local value of parameter
-	virtual	b3Scalar getParam(int num, int axis = -1) const;
+	virtual b3Scalar getParam(int num, int axis = -1) const;
 
-	void setAxis( const b3Vector3& axis1, const b3Vector3& axis2,const b3RigidBodyData* bodies);
-
-
-
-	
+	void setAxis(const b3Vector3& axis1, const b3Vector3& axis2, const b3RigidBodyData* bodies);
 };
 
-
-
-
-
-#endif //B3_GENERIC_6DOF_CONSTRAINT_H
+#endif  //B3_GENERIC_6DOF_CONSTRAINT_H
diff --git a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3JacobianEntry.h b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3JacobianEntry.h
index a55168eb38f..13269debf65 100644
--- a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3JacobianEntry.h
+++ b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3JacobianEntry.h
@@ -18,7 +18,6 @@ subject to the following restrictions:
 
 #include "Bullet3Common/b3Matrix3x3.h"
 
-
 //notes:
 // Another memory optimization would be to store m_1MinvJt in the remaining 3 w components
 // which makes the b3JacobianEntry memory layout 16 bytes
@@ -27,25 +26,26 @@ subject to the following restrictions:
 /// Jacobian entry is an abstraction that allows to describe constraints
 /// it can be used in combination with a constraint solver
 /// Can be used to relate the effect of an impulse to the constraint error
-B3_ATTRIBUTE_ALIGNED16(class) b3JacobianEntry
+B3_ATTRIBUTE_ALIGNED16(class)
+b3JacobianEntry
 {
 public:
-	b3JacobianEntry() {};
+	b3JacobianEntry(){};
 	//constraint between two different rigidbodies
 	b3JacobianEntry(
 		const b3Matrix3x3& world2A,
 		const b3Matrix3x3& world2B,
-		const b3Vector3& rel_pos1,const b3Vector3& rel_pos2,
+		const b3Vector3& rel_pos1, const b3Vector3& rel_pos2,
 		const b3Vector3& jointAxis,
-		const b3Vector3& inertiaInvA, 
+		const b3Vector3& inertiaInvA,
 		const b3Scalar massInvA,
 		const b3Vector3& inertiaInvB,
 		const b3Scalar massInvB)
-		:m_linearJointAxis(jointAxis)
+		: m_linearJointAxis(jointAxis)
 	{
-		m_aJ = world2A*(rel_pos1.cross(m_linearJointAxis));
-		m_bJ = world2B*(rel_pos2.cross(-m_linearJointAxis));
-		m_0MinvJt	= inertiaInvA * m_aJ;
+		m_aJ = world2A * (rel_pos1.cross(m_linearJointAxis));
+		m_bJ = world2B * (rel_pos2.cross(-m_linearJointAxis));
+		m_0MinvJt = inertiaInvA * m_aJ;
 		m_1MinvJt = inertiaInvB * m_bJ;
 		m_Adiag = massInvA + m_0MinvJt.dot(m_aJ) + massInvB + m_1MinvJt.dot(m_bJ);
 
@@ -54,33 +54,31 @@ public:
 
 	//angular constraint between two different rigidbodies
 	b3JacobianEntry(const b3Vector3& jointAxis,
-		const b3Matrix3x3& world2A,
-		const b3Matrix3x3& world2B,
-		const b3Vector3& inertiaInvA,
-		const b3Vector3& inertiaInvB)
-		:m_linearJointAxis(b3MakeVector3(b3Scalar(0.),b3Scalar(0.),b3Scalar(0.)))
+					const b3Matrix3x3& world2A,
+					const b3Matrix3x3& world2B,
+					const b3Vector3& inertiaInvA,
+					const b3Vector3& inertiaInvB)
+		: m_linearJointAxis(b3MakeVector3(b3Scalar(0.), b3Scalar(0.), b3Scalar(0.)))
 	{
-		m_aJ= world2A*jointAxis;
-		m_bJ = world2B*-jointAxis;
-		m_0MinvJt	= inertiaInvA * m_aJ;
+		m_aJ = world2A * jointAxis;
+		m_bJ = world2B * -jointAxis;
+		m_0MinvJt = inertiaInvA * m_aJ;
 		m_1MinvJt = inertiaInvB * m_bJ;
-		m_Adiag =  m_0MinvJt.dot(m_aJ) + m_1MinvJt.dot(m_bJ);
+		m_Adiag = m_0MinvJt.dot(m_aJ) + m_1MinvJt.dot(m_bJ);
 
 		b3Assert(m_Adiag > b3Scalar(0.0));
 	}
 
 	//angular constraint between two different rigidbodies
 	b3JacobianEntry(const b3Vector3& axisInA,
-		const b3Vector3& axisInB,
-		const b3Vector3& inertiaInvA,
-		const b3Vector3& inertiaInvB)
-		: m_linearJointAxis(b3MakeVector3(b3Scalar(0.),b3Scalar(0.),b3Scalar(0.)))
-		, m_aJ(axisInA)
-		, m_bJ(-axisInB)
+					const b3Vector3& axisInB,
+					const b3Vector3& inertiaInvA,
+					const b3Vector3& inertiaInvB)
+		: m_linearJointAxis(b3MakeVector3(b3Scalar(0.), b3Scalar(0.), b3Scalar(0.))), m_aJ(axisInA), m_bJ(-axisInB)
 	{
-		m_0MinvJt	= inertiaInvA * m_aJ;
+		m_0MinvJt = inertiaInvA * m_aJ;
 		m_1MinvJt = inertiaInvB * m_bJ;
-		m_Adiag =  m_0MinvJt.dot(m_aJ) + m_1MinvJt.dot(m_bJ);
+		m_Adiag = m_0MinvJt.dot(m_aJ) + m_1MinvJt.dot(m_bJ);
 
 		b3Assert(m_Adiag > b3Scalar(0.0));
 	}
@@ -88,25 +86,25 @@ public:
 	//constraint on one rigidbody
 	b3JacobianEntry(
 		const b3Matrix3x3& world2A,
-		const b3Vector3& rel_pos1,const b3Vector3& rel_pos2,
+		const b3Vector3& rel_pos1, const b3Vector3& rel_pos2,
 		const b3Vector3& jointAxis,
-		const b3Vector3& inertiaInvA, 
+		const b3Vector3& inertiaInvA,
 		const b3Scalar massInvA)
-		:m_linearJointAxis(jointAxis)
+		: m_linearJointAxis(jointAxis)
 	{
-		m_aJ= world2A*(rel_pos1.cross(jointAxis));
-		m_bJ = world2A*(rel_pos2.cross(-jointAxis));
-		m_0MinvJt	= inertiaInvA * m_aJ;
-		m_1MinvJt = b3MakeVector3(b3Scalar(0.),b3Scalar(0.),b3Scalar(0.));
+		m_aJ = world2A * (rel_pos1.cross(jointAxis));
+		m_bJ = world2A * (rel_pos2.cross(-jointAxis));
+		m_0MinvJt = inertiaInvA * m_aJ;
+		m_1MinvJt = b3MakeVector3(b3Scalar(0.), b3Scalar(0.), b3Scalar(0.));
 		m_Adiag = massInvA + m_0MinvJt.dot(m_aJ);
 
 		b3Assert(m_Adiag > b3Scalar(0.0));
 	}
 
-	b3Scalar	getDiagonal() const { return m_Adiag; }
+	b3Scalar getDiagonal() const { return m_Adiag; }
 
 	// for two constraints on the same rigidbody (for example vehicle friction)
-	b3Scalar	getNonDiagonal(const b3JacobianEntry& jacB, const b3Scalar massInvA) const
+	b3Scalar getNonDiagonal(const b3JacobianEntry& jacB, const b3Scalar massInvA) const
 	{
 		const b3JacobianEntry& jacA = *this;
 		b3Scalar lin = massInvA * jacA.m_linearJointAxis.dot(jacB.m_linearJointAxis);
@@ -114,42 +112,39 @@ public:
 		return lin + ang;
 	}
 
-	
-
 	// for two constraints on sharing two same rigidbodies (for example two contact points between two rigidbodies)
-	b3Scalar	getNonDiagonal(const b3JacobianEntry& jacB,const b3Scalar massInvA,const b3Scalar massInvB) const
+	b3Scalar getNonDiagonal(const b3JacobianEntry& jacB, const b3Scalar massInvA, const b3Scalar massInvB) const
 	{
 		const b3JacobianEntry& jacA = *this;
 		b3Vector3 lin = jacA.m_linearJointAxis * jacB.m_linearJointAxis;
 		b3Vector3 ang0 = jacA.m_0MinvJt * jacB.m_aJ;
 		b3Vector3 ang1 = jacA.m_1MinvJt * jacB.m_bJ;
-		b3Vector3 lin0 = massInvA * lin ;
+		b3Vector3 lin0 = massInvA * lin;
 		b3Vector3 lin1 = massInvB * lin;
-		b3Vector3 sum = ang0+ang1+lin0+lin1;
-		return sum[0]+sum[1]+sum[2];
+		b3Vector3 sum = ang0 + ang1 + lin0 + lin1;
+		return sum[0] + sum[1] + sum[2];
 	}
 
-	b3Scalar getRelativeVelocity(const b3Vector3& linvelA,const b3Vector3& angvelA,const b3Vector3& linvelB,const b3Vector3& angvelB)
+	b3Scalar getRelativeVelocity(const b3Vector3& linvelA, const b3Vector3& angvelA, const b3Vector3& linvelB, const b3Vector3& angvelB)
 	{
 		b3Vector3 linrel = linvelA - linvelB;
-		b3Vector3 angvela  = angvelA * m_aJ;
-		b3Vector3 angvelb  = angvelB * m_bJ;
+		b3Vector3 angvela = angvelA * m_aJ;
+		b3Vector3 angvelb = angvelB * m_bJ;
 		linrel *= m_linearJointAxis;
 		angvela += angvelb;
 		angvela += linrel;
-		b3Scalar rel_vel2 = angvela[0]+angvela[1]+angvela[2];
+		b3Scalar rel_vel2 = angvela[0] + angvela[1] + angvela[2];
 		return rel_vel2 + B3_EPSILON;
 	}
-//private:
+	//private:
 
-	b3Vector3	m_linearJointAxis;
-	b3Vector3	m_aJ;
-	b3Vector3	m_bJ;
-	b3Vector3	m_0MinvJt;
-	b3Vector3	m_1MinvJt;
+	b3Vector3 m_linearJointAxis;
+	b3Vector3 m_aJ;
+	b3Vector3 m_bJ;
+	b3Vector3 m_0MinvJt;
+	b3Vector3 m_1MinvJt;
 	//Optimization: can be stored in the w/last component of one of the vectors
-	b3Scalar	m_Adiag;
-
+	b3Scalar m_Adiag;
 };
 
-#endif //B3_JACOBIAN_ENTRY_H
+#endif  //B3_JACOBIAN_ENTRY_H
diff --git a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3PgsJacobiSolver.cpp b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3PgsJacobiSolver.cpp
index de729d4556a..b7050b10705 100644
--- a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3PgsJacobiSolver.cpp
+++ b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3PgsJacobiSolver.cpp
@@ -29,14 +29,13 @@ subject to the following restrictions:
 //#include "b3SolverBody.h"
 //#include "b3SolverConstraint.h"
 #include "Bullet3Common/b3AlignedObjectArray.h"
-#include <string.h> //for memset
+#include <string.h>  //for memset
 //#include "../../dynamics/basic_demo/Stubs/AdlContact4.h"
 #include "Bullet3Collision/NarrowPhaseCollision/b3Contact4.h"
 
-
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3RigidBodyData.h"
 
-static b3Transform	getWorldTransform(b3RigidBodyData* rb)
+static b3Transform getWorldTransform(b3RigidBodyData* rb)
 {
 	b3Transform newTrans;
 	newTrans.setOrigin(rb->m_pos);
@@ -44,19 +43,17 @@ static b3Transform	getWorldTransform(b3RigidBodyData* rb)
 	return newTrans;
 }
 
-static const b3Matrix3x3&	getInvInertiaTensorWorld(b3InertiaData* inertia)
+static const b3Matrix3x3& getInvInertiaTensorWorld(b3InertiaData* inertia)
 {
 	return inertia->m_invInertiaWorld;
 }
 
-
-
-static const b3Vector3&	getLinearVelocity(b3RigidBodyData* rb)
+static const b3Vector3& getLinearVelocity(b3RigidBodyData* rb)
 {
 	return rb->m_linVel;
 }
 
-static const b3Vector3&	getAngularVelocity(b3RigidBodyData* rb)
+static const b3Vector3& getAngularVelocity(b3RigidBodyData* rb)
 {
 	return rb->m_angVel;
 }
@@ -65,47 +62,46 @@ static b3Vector3 getVelocityInLocalPoint(b3RigidBodyData* rb, const b3Vector3& r
 {
 	//we also calculate lin/ang velocity for kinematic objects
 	return getLinearVelocity(rb) + getAngularVelocity(rb).cross(rel_pos);
-	
 }
 
-struct	b3ContactPoint
+struct b3ContactPoint
 {
-	b3Vector3	m_positionWorldOnA;
-	b3Vector3	m_positionWorldOnB;
-	b3Vector3	m_normalWorldOnB;
-	b3Scalar	m_appliedImpulse;
-	b3Scalar	m_distance;
-	b3Scalar	m_combinedRestitution;
+	b3Vector3 m_positionWorldOnA;
+	b3Vector3 m_positionWorldOnB;
+	b3Vector3 m_normalWorldOnB;
+	b3Scalar m_appliedImpulse;
+	b3Scalar m_distance;
+	b3Scalar m_combinedRestitution;
 
 	///information related to friction
-	b3Scalar	m_combinedFriction;
-	b3Vector3	m_lateralFrictionDir1;
-	b3Vector3	m_lateralFrictionDir2;
-	b3Scalar	m_appliedImpulseLateral1;
-	b3Scalar	m_appliedImpulseLateral2;	
-	b3Scalar	m_combinedRollingFriction;
-	b3Scalar	m_contactMotion1;
-	b3Scalar	m_contactMotion2;
-	b3Scalar	m_contactCFM1;
-	b3Scalar	m_contactCFM2;
-	
-	bool		m_lateralFrictionInitialized;
+	b3Scalar m_combinedFriction;
+	b3Vector3 m_lateralFrictionDir1;
+	b3Vector3 m_lateralFrictionDir2;
+	b3Scalar m_appliedImpulseLateral1;
+	b3Scalar m_appliedImpulseLateral2;
+	b3Scalar m_combinedRollingFriction;
+	b3Scalar m_contactMotion1;
+	b3Scalar m_contactMotion2;
+	b3Scalar m_contactCFM1;
+	b3Scalar m_contactCFM2;
 
-	b3Vector3	getPositionWorldOnA()
+	bool m_lateralFrictionInitialized;
+
+	b3Vector3 getPositionWorldOnA()
 	{
 		return m_positionWorldOnA;
 	}
-	b3Vector3	getPositionWorldOnB()
+	b3Vector3 getPositionWorldOnB()
 	{
 		return m_positionWorldOnB;
 	}
-	b3Scalar	getDistance()
+	b3Scalar getDistance()
 	{
 		return m_distance;
 	}
 };
 
-void	getContactPoint(b3Contact4* contact, int contactIndex, b3ContactPoint& pointOut)
+void getContactPoint(b3Contact4* contact, int contactIndex, b3ContactPoint& pointOut)
 {
 	pointOut.m_appliedImpulse = 0.f;
 	pointOut.m_appliedImpulseLateral1 = 0.f;
@@ -117,160 +113,145 @@ void	getContactPoint(b3Contact4* contact, int contactIndex, b3ContactPoint& poin
 	pointOut.m_contactCFM2 = 0.f;
 	pointOut.m_contactMotion1 = 0.f;
 	pointOut.m_contactMotion2 = 0.f;
-	pointOut.m_distance = contact->getPenetration(contactIndex);//??0.01f
+	pointOut.m_distance = contact->getPenetration(contactIndex);  //??0.01f
 	b3Vector3 normalOnB = contact->m_worldNormalOnB;
-	normalOnB.normalize();//is this needed?
+	normalOnB.normalize();  //is this needed?
 
-	b3Vector3 l1,l2;
-	b3PlaneSpace1(normalOnB,l1,l2);
+	b3Vector3 l1, l2;
+	b3PlaneSpace1(normalOnB, l1, l2);
 
 	pointOut.m_normalWorldOnB = normalOnB;
 	//printf("normalOnB = %f,%f,%f\n",normalOnB.getX(),normalOnB.getY(),normalOnB.getZ());
 	pointOut.m_lateralFrictionDir1 = l1;
 	pointOut.m_lateralFrictionDir2 = l2;
 	pointOut.m_lateralFrictionInitialized = true;
-	
-	
+
 	b3Vector3 worldPosB = contact->m_worldPosB[contactIndex];
 	pointOut.m_positionWorldOnB = worldPosB;
-	pointOut.m_positionWorldOnA = worldPosB+normalOnB*pointOut.m_distance;
+	pointOut.m_positionWorldOnA = worldPosB + normalOnB * pointOut.m_distance;
 }
 
-int	getNumContacts(b3Contact4* contact)
+int getNumContacts(b3Contact4* contact)
 {
 	return contact->getNPoints();
 }
 
 b3PgsJacobiSolver::b3PgsJacobiSolver(bool usePgs)
-:m_usePgs(usePgs),
-m_numSplitImpulseRecoveries(0),
-m_btSeed2(0)
+	: m_usePgs(usePgs),
+	  m_numSplitImpulseRecoveries(0),
+	  m_btSeed2(0)
 {
-
 }
 
 b3PgsJacobiSolver::~b3PgsJacobiSolver()
 {
 }
 
-void	b3PgsJacobiSolver::solveContacts(int numBodies, b3RigidBodyData* bodies, b3InertiaData* inertias, int numContacts, b3Contact4* contacts, int numConstraints, b3TypedConstraint** constraints)
+void b3PgsJacobiSolver::solveContacts(int numBodies, b3RigidBodyData* bodies, b3InertiaData* inertias, int numContacts, b3Contact4* contacts, int numConstraints, b3TypedConstraint** constraints)
 {
 	b3ContactSolverInfo infoGlobal;
 	infoGlobal.m_splitImpulse = false;
-	infoGlobal.m_timeStep = 1.f/60.f;
-	infoGlobal.m_numIterations = 4;//4;
-//	infoGlobal.m_solverMode|=B3_SOLVER_USE_2_FRICTION_DIRECTIONS|B3_SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS|B3_SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION;
+	infoGlobal.m_timeStep = 1.f / 60.f;
+	infoGlobal.m_numIterations = 4;  //4;
+									 //	infoGlobal.m_solverMode|=B3_SOLVER_USE_2_FRICTION_DIRECTIONS|B3_SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS|B3_SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION;
 	//infoGlobal.m_solverMode|=B3_SOLVER_USE_2_FRICTION_DIRECTIONS|B3_SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS;
-	infoGlobal.m_solverMode|=B3_SOLVER_USE_2_FRICTION_DIRECTIONS;
+	infoGlobal.m_solverMode |= B3_SOLVER_USE_2_FRICTION_DIRECTIONS;
 
 	//if (infoGlobal.m_solverMode & B3_SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS)
 	//if ((infoGlobal.m_solverMode & B3_SOLVER_USE_2_FRICTION_DIRECTIONS) && (infoGlobal.m_solverMode & B3_SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION))
-				
 
-	solveGroup(bodies,inertias,numBodies,contacts,numContacts,constraints,numConstraints,infoGlobal);
+	solveGroup(bodies, inertias, numBodies, contacts, numContacts, constraints, numConstraints, infoGlobal);
 
 	if (!numContacts)
 		return;
 }
 
-
-
-
 /// b3PgsJacobiSolver Sequentially applies impulses
 b3Scalar b3PgsJacobiSolver::solveGroup(b3RigidBodyData* bodies,
-										b3InertiaData* inertias, 
-										int numBodies,
-										b3Contact4* manifoldPtr, 
-										int numManifolds,
-										b3TypedConstraint** constraints,
-										int numConstraints,
-										const b3ContactSolverInfo& infoGlobal)
+									   b3InertiaData* inertias,
+									   int numBodies,
+									   b3Contact4* manifoldPtr,
+									   int numManifolds,
+									   b3TypedConstraint** constraints,
+									   int numConstraints,
+									   const b3ContactSolverInfo& infoGlobal)
 {
-
 	B3_PROFILE("solveGroup");
 	//you need to provide at least some bodies
-	
-	solveGroupCacheFriendlySetup( bodies, inertias,numBodies, manifoldPtr,  numManifolds,constraints, numConstraints,infoGlobal);
 
-	solveGroupCacheFriendlyIterations(constraints, numConstraints,infoGlobal);
+	solveGroupCacheFriendlySetup(bodies, inertias, numBodies, manifoldPtr, numManifolds, constraints, numConstraints, infoGlobal);
+
+	solveGroupCacheFriendlyIterations(constraints, numConstraints, infoGlobal);
+
+	solveGroupCacheFriendlyFinish(bodies, inertias, numBodies, infoGlobal);
 
-	solveGroupCacheFriendlyFinish(bodies, inertias,numBodies, infoGlobal);
-	
 	return 0.f;
 }
 
-
-
-
-
-
-
-
-
 #ifdef USE_SIMD
 #include <emmintrin.h>
-#define b3VecSplat(x, e) _mm_shuffle_ps(x, x, _MM_SHUFFLE(e,e,e,e))
-static inline __m128 b3SimdDot3( __m128 vec0, __m128 vec1 )
+#define b3VecSplat(x, e) _mm_shuffle_ps(x, x, _MM_SHUFFLE(e, e, e, e))
+static inline __m128 b3SimdDot3(__m128 vec0, __m128 vec1)
 {
-	__m128 result = _mm_mul_ps( vec0, vec1);
-	return _mm_add_ps( b3VecSplat( result, 0 ), _mm_add_ps( b3VecSplat( result, 1 ), b3VecSplat( result, 2 ) ) );
+	__m128 result = _mm_mul_ps(vec0, vec1);
+	return _mm_add_ps(b3VecSplat(result, 0), _mm_add_ps(b3VecSplat(result, 1), b3VecSplat(result, 2)));
 }
-#endif//USE_SIMD
+#endif  //USE_SIMD
 
 // Project Gauss Seidel or the equivalent Sequential Impulse
-void b3PgsJacobiSolver::resolveSingleConstraintRowGenericSIMD(b3SolverBody& body1,b3SolverBody& body2,const b3SolverConstraint& c)
+void b3PgsJacobiSolver::resolveSingleConstraintRowGenericSIMD(b3SolverBody& body1, b3SolverBody& body2, const b3SolverConstraint& c)
 {
 #ifdef USE_SIMD
 	__m128 cpAppliedImp = _mm_set1_ps(c.m_appliedImpulse);
-	__m128	lowerLimit1 = _mm_set1_ps(c.m_lowerLimit);
-	__m128	upperLimit1 = _mm_set1_ps(c.m_upperLimit);
-	__m128 deltaImpulse = _mm_sub_ps(_mm_set1_ps(c.m_rhs), _mm_mul_ps(_mm_set1_ps(c.m_appliedImpulse),_mm_set1_ps(c.m_cfm)));
-	__m128 deltaVel1Dotn	=	_mm_add_ps(b3SimdDot3(c.m_contactNormal.mVec128,body1.internalGetDeltaLinearVelocity().mVec128), b3SimdDot3(c.m_relpos1CrossNormal.mVec128,body1.internalGetDeltaAngularVelocity().mVec128));
-	__m128 deltaVel2Dotn	=	_mm_sub_ps(b3SimdDot3(c.m_relpos2CrossNormal.mVec128,body2.internalGetDeltaAngularVelocity().mVec128),b3SimdDot3((c.m_contactNormal).mVec128,body2.internalGetDeltaLinearVelocity().mVec128));
-	deltaImpulse	=	_mm_sub_ps(deltaImpulse,_mm_mul_ps(deltaVel1Dotn,_mm_set1_ps(c.m_jacDiagABInv)));
-	deltaImpulse	=	_mm_sub_ps(deltaImpulse,_mm_mul_ps(deltaVel2Dotn,_mm_set1_ps(c.m_jacDiagABInv)));
-	b3SimdScalar sum = _mm_add_ps(cpAppliedImp,deltaImpulse);
-	b3SimdScalar resultLowerLess,resultUpperLess;
-	resultLowerLess = _mm_cmplt_ps(sum,lowerLimit1);
-	resultUpperLess = _mm_cmplt_ps(sum,upperLimit1);
-	__m128 lowMinApplied = _mm_sub_ps(lowerLimit1,cpAppliedImp);
-	deltaImpulse = _mm_or_ps( _mm_and_ps(resultLowerLess, lowMinApplied), _mm_andnot_ps(resultLowerLess, deltaImpulse) );
-	c.m_appliedImpulse = _mm_or_ps( _mm_and_ps(resultLowerLess, lowerLimit1), _mm_andnot_ps(resultLowerLess, sum) );
-	__m128 upperMinApplied = _mm_sub_ps(upperLimit1,cpAppliedImp);
-	deltaImpulse = _mm_or_ps( _mm_and_ps(resultUpperLess, deltaImpulse), _mm_andnot_ps(resultUpperLess, upperMinApplied) );
-	c.m_appliedImpulse = _mm_or_ps( _mm_and_ps(resultUpperLess, c.m_appliedImpulse), _mm_andnot_ps(resultUpperLess, upperLimit1) );
-	__m128	linearComponentA = _mm_mul_ps(c.m_contactNormal.mVec128,body1.internalGetInvMass().mVec128);
-	__m128	linearComponentB = _mm_mul_ps((c.m_contactNormal).mVec128,body2.internalGetInvMass().mVec128);
+	__m128 lowerLimit1 = _mm_set1_ps(c.m_lowerLimit);
+	__m128 upperLimit1 = _mm_set1_ps(c.m_upperLimit);
+	__m128 deltaImpulse = _mm_sub_ps(_mm_set1_ps(c.m_rhs), _mm_mul_ps(_mm_set1_ps(c.m_appliedImpulse), _mm_set1_ps(c.m_cfm)));
+	__m128 deltaVel1Dotn = _mm_add_ps(b3SimdDot3(c.m_contactNormal.mVec128, body1.internalGetDeltaLinearVelocity().mVec128), b3SimdDot3(c.m_relpos1CrossNormal.mVec128, body1.internalGetDeltaAngularVelocity().mVec128));
+	__m128 deltaVel2Dotn = _mm_sub_ps(b3SimdDot3(c.m_relpos2CrossNormal.mVec128, body2.internalGetDeltaAngularVelocity().mVec128), b3SimdDot3((c.m_contactNormal).mVec128, body2.internalGetDeltaLinearVelocity().mVec128));
+	deltaImpulse = _mm_sub_ps(deltaImpulse, _mm_mul_ps(deltaVel1Dotn, _mm_set1_ps(c.m_jacDiagABInv)));
+	deltaImpulse = _mm_sub_ps(deltaImpulse, _mm_mul_ps(deltaVel2Dotn, _mm_set1_ps(c.m_jacDiagABInv)));
+	b3SimdScalar sum = _mm_add_ps(cpAppliedImp, deltaImpulse);
+	b3SimdScalar resultLowerLess, resultUpperLess;
+	resultLowerLess = _mm_cmplt_ps(sum, lowerLimit1);
+	resultUpperLess = _mm_cmplt_ps(sum, upperLimit1);
+	__m128 lowMinApplied = _mm_sub_ps(lowerLimit1, cpAppliedImp);
+	deltaImpulse = _mm_or_ps(_mm_and_ps(resultLowerLess, lowMinApplied), _mm_andnot_ps(resultLowerLess, deltaImpulse));
+	c.m_appliedImpulse = _mm_or_ps(_mm_and_ps(resultLowerLess, lowerLimit1), _mm_andnot_ps(resultLowerLess, sum));
+	__m128 upperMinApplied = _mm_sub_ps(upperLimit1, cpAppliedImp);
+	deltaImpulse = _mm_or_ps(_mm_and_ps(resultUpperLess, deltaImpulse), _mm_andnot_ps(resultUpperLess, upperMinApplied));
+	c.m_appliedImpulse = _mm_or_ps(_mm_and_ps(resultUpperLess, c.m_appliedImpulse), _mm_andnot_ps(resultUpperLess, upperLimit1));
+	__m128 linearComponentA = _mm_mul_ps(c.m_contactNormal.mVec128, body1.internalGetInvMass().mVec128);
+	__m128 linearComponentB = _mm_mul_ps((c.m_contactNormal).mVec128, body2.internalGetInvMass().mVec128);
 	__m128 impulseMagnitude = deltaImpulse;
-	body1.internalGetDeltaLinearVelocity().mVec128 = _mm_add_ps(body1.internalGetDeltaLinearVelocity().mVec128,_mm_mul_ps(linearComponentA,impulseMagnitude));
-	body1.internalGetDeltaAngularVelocity().mVec128 = _mm_add_ps(body1.internalGetDeltaAngularVelocity().mVec128 ,_mm_mul_ps(c.m_angularComponentA.mVec128,impulseMagnitude));
-	body2.internalGetDeltaLinearVelocity().mVec128 = _mm_sub_ps(body2.internalGetDeltaLinearVelocity().mVec128,_mm_mul_ps(linearComponentB,impulseMagnitude));
-	body2.internalGetDeltaAngularVelocity().mVec128 = _mm_add_ps(body2.internalGetDeltaAngularVelocity().mVec128 ,_mm_mul_ps(c.m_angularComponentB.mVec128,impulseMagnitude));
+	body1.internalGetDeltaLinearVelocity().mVec128 = _mm_add_ps(body1.internalGetDeltaLinearVelocity().mVec128, _mm_mul_ps(linearComponentA, impulseMagnitude));
+	body1.internalGetDeltaAngularVelocity().mVec128 = _mm_add_ps(body1.internalGetDeltaAngularVelocity().mVec128, _mm_mul_ps(c.m_angularComponentA.mVec128, impulseMagnitude));
+	body2.internalGetDeltaLinearVelocity().mVec128 = _mm_sub_ps(body2.internalGetDeltaLinearVelocity().mVec128, _mm_mul_ps(linearComponentB, impulseMagnitude));
+	body2.internalGetDeltaAngularVelocity().mVec128 = _mm_add_ps(body2.internalGetDeltaAngularVelocity().mVec128, _mm_mul_ps(c.m_angularComponentB.mVec128, impulseMagnitude));
 #else
-	resolveSingleConstraintRowGeneric(body1,body2,c);
+	resolveSingleConstraintRowGeneric(body1, body2, c);
 #endif
 }
 
 // Project Gauss Seidel or the equivalent Sequential Impulse
- void b3PgsJacobiSolver::resolveSingleConstraintRowGeneric(b3SolverBody& body1,b3SolverBody& body2,const b3SolverConstraint& c)
+void b3PgsJacobiSolver::resolveSingleConstraintRowGeneric(b3SolverBody& body1, b3SolverBody& body2, const b3SolverConstraint& c)
 {
-	b3Scalar deltaImpulse = c.m_rhs-b3Scalar(c.m_appliedImpulse)*c.m_cfm;
-	const b3Scalar deltaVel1Dotn	=	c.m_contactNormal.dot(body1.internalGetDeltaLinearVelocity()) 	+ c.m_relpos1CrossNormal.dot(body1.internalGetDeltaAngularVelocity());
-	const b3Scalar deltaVel2Dotn	=	-c.m_contactNormal.dot(body2.internalGetDeltaLinearVelocity()) + c.m_relpos2CrossNormal.dot(body2.internalGetDeltaAngularVelocity());
+	b3Scalar deltaImpulse = c.m_rhs - b3Scalar(c.m_appliedImpulse) * c.m_cfm;
+	const b3Scalar deltaVel1Dotn = c.m_contactNormal.dot(body1.internalGetDeltaLinearVelocity()) + c.m_relpos1CrossNormal.dot(body1.internalGetDeltaAngularVelocity());
+	const b3Scalar deltaVel2Dotn = -c.m_contactNormal.dot(body2.internalGetDeltaLinearVelocity()) + c.m_relpos2CrossNormal.dot(body2.internalGetDeltaAngularVelocity());
 
-//	const b3Scalar delta_rel_vel	=	deltaVel1Dotn-deltaVel2Dotn;
-	deltaImpulse	-=	deltaVel1Dotn*c.m_jacDiagABInv;
-	deltaImpulse	-=	deltaVel2Dotn*c.m_jacDiagABInv;
+	//	const b3Scalar delta_rel_vel	=	deltaVel1Dotn-deltaVel2Dotn;
+	deltaImpulse -= deltaVel1Dotn * c.m_jacDiagABInv;
+	deltaImpulse -= deltaVel2Dotn * c.m_jacDiagABInv;
 
 	const b3Scalar sum = b3Scalar(c.m_appliedImpulse) + deltaImpulse;
 	if (sum < c.m_lowerLimit)
 	{
-		deltaImpulse = c.m_lowerLimit-c.m_appliedImpulse;
+		deltaImpulse = c.m_lowerLimit - c.m_appliedImpulse;
 		c.m_appliedImpulse = c.m_lowerLimit;
 	}
-	else if (sum > c.m_upperLimit) 
+	else if (sum > c.m_upperLimit)
 	{
-		deltaImpulse = c.m_upperLimit-c.m_appliedImpulse;
+		deltaImpulse = c.m_upperLimit - c.m_appliedImpulse;
 		c.m_appliedImpulse = c.m_upperLimit;
 	}
 	else
@@ -278,94 +259,93 @@ void b3PgsJacobiSolver::resolveSingleConstraintRowGenericSIMD(b3SolverBody& body
 		c.m_appliedImpulse = sum;
 	}
 
-	body1.internalApplyImpulse(c.m_contactNormal*body1.internalGetInvMass(),c.m_angularComponentA,deltaImpulse);
-	body2.internalApplyImpulse(-c.m_contactNormal*body2.internalGetInvMass(),c.m_angularComponentB,deltaImpulse);
+	body1.internalApplyImpulse(c.m_contactNormal * body1.internalGetInvMass(), c.m_angularComponentA, deltaImpulse);
+	body2.internalApplyImpulse(-c.m_contactNormal * body2.internalGetInvMass(), c.m_angularComponentB, deltaImpulse);
 }
 
- void b3PgsJacobiSolver::resolveSingleConstraintRowLowerLimitSIMD(b3SolverBody& body1,b3SolverBody& body2,const b3SolverConstraint& c)
+void b3PgsJacobiSolver::resolveSingleConstraintRowLowerLimitSIMD(b3SolverBody& body1, b3SolverBody& body2, const b3SolverConstraint& c)
 {
 #ifdef USE_SIMD
 	__m128 cpAppliedImp = _mm_set1_ps(c.m_appliedImpulse);
-	__m128	lowerLimit1 = _mm_set1_ps(c.m_lowerLimit);
-	__m128	upperLimit1 = _mm_set1_ps(c.m_upperLimit);
-	__m128 deltaImpulse = _mm_sub_ps(_mm_set1_ps(c.m_rhs), _mm_mul_ps(_mm_set1_ps(c.m_appliedImpulse),_mm_set1_ps(c.m_cfm)));
-	__m128 deltaVel1Dotn	=	_mm_add_ps(b3SimdDot3(c.m_contactNormal.mVec128,body1.internalGetDeltaLinearVelocity().mVec128), b3SimdDot3(c.m_relpos1CrossNormal.mVec128,body1.internalGetDeltaAngularVelocity().mVec128));
-	__m128 deltaVel2Dotn	=	_mm_sub_ps(b3SimdDot3(c.m_relpos2CrossNormal.mVec128,body2.internalGetDeltaAngularVelocity().mVec128),b3SimdDot3((c.m_contactNormal).mVec128,body2.internalGetDeltaLinearVelocity().mVec128));
-	deltaImpulse	=	_mm_sub_ps(deltaImpulse,_mm_mul_ps(deltaVel1Dotn,_mm_set1_ps(c.m_jacDiagABInv)));
-	deltaImpulse	=	_mm_sub_ps(deltaImpulse,_mm_mul_ps(deltaVel2Dotn,_mm_set1_ps(c.m_jacDiagABInv)));
-	b3SimdScalar sum = _mm_add_ps(cpAppliedImp,deltaImpulse);
-	b3SimdScalar resultLowerLess,resultUpperLess;
-	resultLowerLess = _mm_cmplt_ps(sum,lowerLimit1);
-	resultUpperLess = _mm_cmplt_ps(sum,upperLimit1);
-	__m128 lowMinApplied = _mm_sub_ps(lowerLimit1,cpAppliedImp);
-	deltaImpulse = _mm_or_ps( _mm_and_ps(resultLowerLess, lowMinApplied), _mm_andnot_ps(resultLowerLess, deltaImpulse) );
-	c.m_appliedImpulse = _mm_or_ps( _mm_and_ps(resultLowerLess, lowerLimit1), _mm_andnot_ps(resultLowerLess, sum) );
-	__m128	linearComponentA = _mm_mul_ps(c.m_contactNormal.mVec128,body1.internalGetInvMass().mVec128);
-	__m128	linearComponentB = _mm_mul_ps((c.m_contactNormal).mVec128,body2.internalGetInvMass().mVec128);
+	__m128 lowerLimit1 = _mm_set1_ps(c.m_lowerLimit);
+	__m128 upperLimit1 = _mm_set1_ps(c.m_upperLimit);
+	__m128 deltaImpulse = _mm_sub_ps(_mm_set1_ps(c.m_rhs), _mm_mul_ps(_mm_set1_ps(c.m_appliedImpulse), _mm_set1_ps(c.m_cfm)));
+	__m128 deltaVel1Dotn = _mm_add_ps(b3SimdDot3(c.m_contactNormal.mVec128, body1.internalGetDeltaLinearVelocity().mVec128), b3SimdDot3(c.m_relpos1CrossNormal.mVec128, body1.internalGetDeltaAngularVelocity().mVec128));
+	__m128 deltaVel2Dotn = _mm_sub_ps(b3SimdDot3(c.m_relpos2CrossNormal.mVec128, body2.internalGetDeltaAngularVelocity().mVec128), b3SimdDot3((c.m_contactNormal).mVec128, body2.internalGetDeltaLinearVelocity().mVec128));
+	deltaImpulse = _mm_sub_ps(deltaImpulse, _mm_mul_ps(deltaVel1Dotn, _mm_set1_ps(c.m_jacDiagABInv)));
+	deltaImpulse = _mm_sub_ps(deltaImpulse, _mm_mul_ps(deltaVel2Dotn, _mm_set1_ps(c.m_jacDiagABInv)));
+	b3SimdScalar sum = _mm_add_ps(cpAppliedImp, deltaImpulse);
+	b3SimdScalar resultLowerLess, resultUpperLess;
+	resultLowerLess = _mm_cmplt_ps(sum, lowerLimit1);
+	resultUpperLess = _mm_cmplt_ps(sum, upperLimit1);
+	__m128 lowMinApplied = _mm_sub_ps(lowerLimit1, cpAppliedImp);
+	deltaImpulse = _mm_or_ps(_mm_and_ps(resultLowerLess, lowMinApplied), _mm_andnot_ps(resultLowerLess, deltaImpulse));
+	c.m_appliedImpulse = _mm_or_ps(_mm_and_ps(resultLowerLess, lowerLimit1), _mm_andnot_ps(resultLowerLess, sum));
+	__m128 linearComponentA = _mm_mul_ps(c.m_contactNormal.mVec128, body1.internalGetInvMass().mVec128);
+	__m128 linearComponentB = _mm_mul_ps((c.m_contactNormal).mVec128, body2.internalGetInvMass().mVec128);
 	__m128 impulseMagnitude = deltaImpulse;
-	body1.internalGetDeltaLinearVelocity().mVec128 = _mm_add_ps(body1.internalGetDeltaLinearVelocity().mVec128,_mm_mul_ps(linearComponentA,impulseMagnitude));
-	body1.internalGetDeltaAngularVelocity().mVec128 = _mm_add_ps(body1.internalGetDeltaAngularVelocity().mVec128 ,_mm_mul_ps(c.m_angularComponentA.mVec128,impulseMagnitude));
-	body2.internalGetDeltaLinearVelocity().mVec128 = _mm_sub_ps(body2.internalGetDeltaLinearVelocity().mVec128,_mm_mul_ps(linearComponentB,impulseMagnitude));
-	body2.internalGetDeltaAngularVelocity().mVec128 = _mm_add_ps(body2.internalGetDeltaAngularVelocity().mVec128 ,_mm_mul_ps(c.m_angularComponentB.mVec128,impulseMagnitude));
+	body1.internalGetDeltaLinearVelocity().mVec128 = _mm_add_ps(body1.internalGetDeltaLinearVelocity().mVec128, _mm_mul_ps(linearComponentA, impulseMagnitude));
+	body1.internalGetDeltaAngularVelocity().mVec128 = _mm_add_ps(body1.internalGetDeltaAngularVelocity().mVec128, _mm_mul_ps(c.m_angularComponentA.mVec128, impulseMagnitude));
+	body2.internalGetDeltaLinearVelocity().mVec128 = _mm_sub_ps(body2.internalGetDeltaLinearVelocity().mVec128, _mm_mul_ps(linearComponentB, impulseMagnitude));
+	body2.internalGetDeltaAngularVelocity().mVec128 = _mm_add_ps(body2.internalGetDeltaAngularVelocity().mVec128, _mm_mul_ps(c.m_angularComponentB.mVec128, impulseMagnitude));
 #else
-	resolveSingleConstraintRowLowerLimit(body1,body2,c);
+	resolveSingleConstraintRowLowerLimit(body1, body2, c);
 #endif
 }
 
 // Project Gauss Seidel or the equivalent Sequential Impulse
- void b3PgsJacobiSolver::resolveSingleConstraintRowLowerLimit(b3SolverBody& body1,b3SolverBody& body2,const b3SolverConstraint& c)
+void b3PgsJacobiSolver::resolveSingleConstraintRowLowerLimit(b3SolverBody& body1, b3SolverBody& body2, const b3SolverConstraint& c)
 {
-	b3Scalar deltaImpulse = c.m_rhs-b3Scalar(c.m_appliedImpulse)*c.m_cfm;
-	const b3Scalar deltaVel1Dotn	=	c.m_contactNormal.dot(body1.internalGetDeltaLinearVelocity()) 	+ c.m_relpos1CrossNormal.dot(body1.internalGetDeltaAngularVelocity());
-	const b3Scalar deltaVel2Dotn	=	-c.m_contactNormal.dot(body2.internalGetDeltaLinearVelocity()) + c.m_relpos2CrossNormal.dot(body2.internalGetDeltaAngularVelocity());
+	b3Scalar deltaImpulse = c.m_rhs - b3Scalar(c.m_appliedImpulse) * c.m_cfm;
+	const b3Scalar deltaVel1Dotn = c.m_contactNormal.dot(body1.internalGetDeltaLinearVelocity()) + c.m_relpos1CrossNormal.dot(body1.internalGetDeltaAngularVelocity());
+	const b3Scalar deltaVel2Dotn = -c.m_contactNormal.dot(body2.internalGetDeltaLinearVelocity()) + c.m_relpos2CrossNormal.dot(body2.internalGetDeltaAngularVelocity());
 
-	deltaImpulse	-=	deltaVel1Dotn*c.m_jacDiagABInv;
-	deltaImpulse	-=	deltaVel2Dotn*c.m_jacDiagABInv;
+	deltaImpulse -= deltaVel1Dotn * c.m_jacDiagABInv;
+	deltaImpulse -= deltaVel2Dotn * c.m_jacDiagABInv;
 	const b3Scalar sum = b3Scalar(c.m_appliedImpulse) + deltaImpulse;
 	if (sum < c.m_lowerLimit)
 	{
-		deltaImpulse = c.m_lowerLimit-c.m_appliedImpulse;
+		deltaImpulse = c.m_lowerLimit - c.m_appliedImpulse;
 		c.m_appliedImpulse = c.m_lowerLimit;
 	}
 	else
 	{
 		c.m_appliedImpulse = sum;
 	}
-	body1.internalApplyImpulse(c.m_contactNormal*body1.internalGetInvMass(),c.m_angularComponentA,deltaImpulse);
-	body2.internalApplyImpulse(-c.m_contactNormal*body2.internalGetInvMass(),c.m_angularComponentB,deltaImpulse);
+	body1.internalApplyImpulse(c.m_contactNormal * body1.internalGetInvMass(), c.m_angularComponentA, deltaImpulse);
+	body2.internalApplyImpulse(-c.m_contactNormal * body2.internalGetInvMass(), c.m_angularComponentB, deltaImpulse);
 }
 
-
-void	b3PgsJacobiSolver::resolveSplitPenetrationImpulseCacheFriendly(
-        b3SolverBody& body1,
-        b3SolverBody& body2,
-        const b3SolverConstraint& c)
+void b3PgsJacobiSolver::resolveSplitPenetrationImpulseCacheFriendly(
+	b3SolverBody& body1,
+	b3SolverBody& body2,
+	const b3SolverConstraint& c)
 {
-		if (c.m_rhsPenetration)
-        {
-			m_numSplitImpulseRecoveries++;
-			b3Scalar deltaImpulse = c.m_rhsPenetration-b3Scalar(c.m_appliedPushImpulse)*c.m_cfm;
-			const b3Scalar deltaVel1Dotn	=	c.m_contactNormal.dot(body1.internalGetPushVelocity()) 	+ c.m_relpos1CrossNormal.dot(body1.internalGetTurnVelocity());
-			const b3Scalar deltaVel2Dotn	=	-c.m_contactNormal.dot(body2.internalGetPushVelocity()) + c.m_relpos2CrossNormal.dot(body2.internalGetTurnVelocity());
+	if (c.m_rhsPenetration)
+	{
+		m_numSplitImpulseRecoveries++;
+		b3Scalar deltaImpulse = c.m_rhsPenetration - b3Scalar(c.m_appliedPushImpulse) * c.m_cfm;
+		const b3Scalar deltaVel1Dotn = c.m_contactNormal.dot(body1.internalGetPushVelocity()) + c.m_relpos1CrossNormal.dot(body1.internalGetTurnVelocity());
+		const b3Scalar deltaVel2Dotn = -c.m_contactNormal.dot(body2.internalGetPushVelocity()) + c.m_relpos2CrossNormal.dot(body2.internalGetTurnVelocity());
 
-			deltaImpulse	-=	deltaVel1Dotn*c.m_jacDiagABInv;
-			deltaImpulse	-=	deltaVel2Dotn*c.m_jacDiagABInv;
-			const b3Scalar sum = b3Scalar(c.m_appliedPushImpulse) + deltaImpulse;
-			if (sum < c.m_lowerLimit)
-			{
-				deltaImpulse = c.m_lowerLimit-c.m_appliedPushImpulse;
-				c.m_appliedPushImpulse = c.m_lowerLimit;
-			}
-			else
-			{
-				c.m_appliedPushImpulse = sum;
-			}
-			body1.internalApplyPushImpulse(c.m_contactNormal*body1.internalGetInvMass(),c.m_angularComponentA,deltaImpulse);
-			body2.internalApplyPushImpulse(-c.m_contactNormal*body2.internalGetInvMass(),c.m_angularComponentB,deltaImpulse);
-        }
+		deltaImpulse -= deltaVel1Dotn * c.m_jacDiagABInv;
+		deltaImpulse -= deltaVel2Dotn * c.m_jacDiagABInv;
+		const b3Scalar sum = b3Scalar(c.m_appliedPushImpulse) + deltaImpulse;
+		if (sum < c.m_lowerLimit)
+		{
+			deltaImpulse = c.m_lowerLimit - c.m_appliedPushImpulse;
+			c.m_appliedPushImpulse = c.m_lowerLimit;
+		}
+		else
+		{
+			c.m_appliedPushImpulse = sum;
+		}
+		body1.internalApplyPushImpulse(c.m_contactNormal * body1.internalGetInvMass(), c.m_angularComponentA, deltaImpulse);
+		body2.internalApplyPushImpulse(-c.m_contactNormal * body2.internalGetInvMass(), c.m_angularComponentB, deltaImpulse);
+	}
 }
 
- void b3PgsJacobiSolver::resolveSplitPenetrationSIMD(b3SolverBody& body1,b3SolverBody& body2,const b3SolverConstraint& c)
+void b3PgsJacobiSolver::resolveSplitPenetrationSIMD(b3SolverBody& body1, b3SolverBody& body2, const b3SolverConstraint& c)
 {
 #ifdef USE_SIMD
 	if (!c.m_rhsPenetration)
@@ -374,44 +354,40 @@ void	b3PgsJacobiSolver::resolveSplitPenetrationImpulseCacheFriendly(
 	m_numSplitImpulseRecoveries++;
 
 	__m128 cpAppliedImp = _mm_set1_ps(c.m_appliedPushImpulse);
-	__m128	lowerLimit1 = _mm_set1_ps(c.m_lowerLimit);
-	__m128	upperLimit1 = _mm_set1_ps(c.m_upperLimit);
-	__m128 deltaImpulse = _mm_sub_ps(_mm_set1_ps(c.m_rhsPenetration), _mm_mul_ps(_mm_set1_ps(c.m_appliedPushImpulse),_mm_set1_ps(c.m_cfm)));
-	__m128 deltaVel1Dotn	=	_mm_add_ps(b3SimdDot3(c.m_contactNormal.mVec128,body1.internalGetPushVelocity().mVec128), b3SimdDot3(c.m_relpos1CrossNormal.mVec128,body1.internalGetTurnVelocity().mVec128));
-	__m128 deltaVel2Dotn	=	_mm_sub_ps(b3SimdDot3(c.m_relpos2CrossNormal.mVec128,body2.internalGetTurnVelocity().mVec128),b3SimdDot3((c.m_contactNormal).mVec128,body2.internalGetPushVelocity().mVec128));
-	deltaImpulse	=	_mm_sub_ps(deltaImpulse,_mm_mul_ps(deltaVel1Dotn,_mm_set1_ps(c.m_jacDiagABInv)));
-	deltaImpulse	=	_mm_sub_ps(deltaImpulse,_mm_mul_ps(deltaVel2Dotn,_mm_set1_ps(c.m_jacDiagABInv)));
-	b3SimdScalar sum = _mm_add_ps(cpAppliedImp,deltaImpulse);
-	b3SimdScalar resultLowerLess,resultUpperLess;
-	resultLowerLess = _mm_cmplt_ps(sum,lowerLimit1);
-	resultUpperLess = _mm_cmplt_ps(sum,upperLimit1);
-	__m128 lowMinApplied = _mm_sub_ps(lowerLimit1,cpAppliedImp);
-	deltaImpulse = _mm_or_ps( _mm_and_ps(resultLowerLess, lowMinApplied), _mm_andnot_ps(resultLowerLess, deltaImpulse) );
-	c.m_appliedPushImpulse = _mm_or_ps( _mm_and_ps(resultLowerLess, lowerLimit1), _mm_andnot_ps(resultLowerLess, sum) );
-	__m128	linearComponentA = _mm_mul_ps(c.m_contactNormal.mVec128,body1.internalGetInvMass().mVec128);
-	__m128	linearComponentB = _mm_mul_ps((c.m_contactNormal).mVec128,body2.internalGetInvMass().mVec128);
+	__m128 lowerLimit1 = _mm_set1_ps(c.m_lowerLimit);
+	__m128 upperLimit1 = _mm_set1_ps(c.m_upperLimit);
+	__m128 deltaImpulse = _mm_sub_ps(_mm_set1_ps(c.m_rhsPenetration), _mm_mul_ps(_mm_set1_ps(c.m_appliedPushImpulse), _mm_set1_ps(c.m_cfm)));
+	__m128 deltaVel1Dotn = _mm_add_ps(b3SimdDot3(c.m_contactNormal.mVec128, body1.internalGetPushVelocity().mVec128), b3SimdDot3(c.m_relpos1CrossNormal.mVec128, body1.internalGetTurnVelocity().mVec128));
+	__m128 deltaVel2Dotn = _mm_sub_ps(b3SimdDot3(c.m_relpos2CrossNormal.mVec128, body2.internalGetTurnVelocity().mVec128), b3SimdDot3((c.m_contactNormal).mVec128, body2.internalGetPushVelocity().mVec128));
+	deltaImpulse = _mm_sub_ps(deltaImpulse, _mm_mul_ps(deltaVel1Dotn, _mm_set1_ps(c.m_jacDiagABInv)));
+	deltaImpulse = _mm_sub_ps(deltaImpulse, _mm_mul_ps(deltaVel2Dotn, _mm_set1_ps(c.m_jacDiagABInv)));
+	b3SimdScalar sum = _mm_add_ps(cpAppliedImp, deltaImpulse);
+	b3SimdScalar resultLowerLess, resultUpperLess;
+	resultLowerLess = _mm_cmplt_ps(sum, lowerLimit1);
+	resultUpperLess = _mm_cmplt_ps(sum, upperLimit1);
+	__m128 lowMinApplied = _mm_sub_ps(lowerLimit1, cpAppliedImp);
+	deltaImpulse = _mm_or_ps(_mm_and_ps(resultLowerLess, lowMinApplied), _mm_andnot_ps(resultLowerLess, deltaImpulse));
+	c.m_appliedPushImpulse = _mm_or_ps(_mm_and_ps(resultLowerLess, lowerLimit1), _mm_andnot_ps(resultLowerLess, sum));
+	__m128 linearComponentA = _mm_mul_ps(c.m_contactNormal.mVec128, body1.internalGetInvMass().mVec128);
+	__m128 linearComponentB = _mm_mul_ps((c.m_contactNormal).mVec128, body2.internalGetInvMass().mVec128);
 	__m128 impulseMagnitude = deltaImpulse;
-	body1.internalGetPushVelocity().mVec128 = _mm_add_ps(body1.internalGetPushVelocity().mVec128,_mm_mul_ps(linearComponentA,impulseMagnitude));
-	body1.internalGetTurnVelocity().mVec128 = _mm_add_ps(body1.internalGetTurnVelocity().mVec128 ,_mm_mul_ps(c.m_angularComponentA.mVec128,impulseMagnitude));
-	body2.internalGetPushVelocity().mVec128 = _mm_sub_ps(body2.internalGetPushVelocity().mVec128,_mm_mul_ps(linearComponentB,impulseMagnitude));
-	body2.internalGetTurnVelocity().mVec128 = _mm_add_ps(body2.internalGetTurnVelocity().mVec128 ,_mm_mul_ps(c.m_angularComponentB.mVec128,impulseMagnitude));
+	body1.internalGetPushVelocity().mVec128 = _mm_add_ps(body1.internalGetPushVelocity().mVec128, _mm_mul_ps(linearComponentA, impulseMagnitude));
+	body1.internalGetTurnVelocity().mVec128 = _mm_add_ps(body1.internalGetTurnVelocity().mVec128, _mm_mul_ps(c.m_angularComponentA.mVec128, impulseMagnitude));
+	body2.internalGetPushVelocity().mVec128 = _mm_sub_ps(body2.internalGetPushVelocity().mVec128, _mm_mul_ps(linearComponentB, impulseMagnitude));
+	body2.internalGetTurnVelocity().mVec128 = _mm_add_ps(body2.internalGetTurnVelocity().mVec128, _mm_mul_ps(c.m_angularComponentB.mVec128, impulseMagnitude));
 #else
-	resolveSplitPenetrationImpulseCacheFriendly(body1,body2,c);
+	resolveSplitPenetrationImpulseCacheFriendly(body1, body2, c);
 #endif
 }
 
-
-
 unsigned long b3PgsJacobiSolver::b3Rand2()
 {
-	m_btSeed2 = (1664525L*m_btSeed2 + 1013904223L) & 0xffffffff;
+	m_btSeed2 = (1664525L * m_btSeed2 + 1013904223L) & 0xffffffff;
 	return m_btSeed2;
 }
 
-
-
 //See ODE: adam's all-int straightforward(?) dRandInt (0..n-1)
-int b3PgsJacobiSolver::b3RandInt2 (int n)
+int b3PgsJacobiSolver::b3RandInt2(int n)
 {
 	// seems good; xor-fold and modulus
 	const unsigned long un = static_cast<unsigned long>(n);
@@ -419,15 +395,20 @@ int b3PgsJacobiSolver::b3RandInt2 (int n)
 
 	// note: probably more aggressive than it needs to be -- might be
 	//       able to get away without one or two of the innermost branches.
-	if (un <= 0x00010000UL) {
+	if (un <= 0x00010000UL)
+	{
 		r ^= (r >> 16);
-		if (un <= 0x00000100UL) {
+		if (un <= 0x00000100UL)
+		{
 			r ^= (r >> 8);
-			if (un <= 0x00000010UL) {
+			if (un <= 0x00000010UL)
+			{
 				r ^= (r >> 4);
-				if (un <= 0x00000004UL) {
+				if (un <= 0x00000004UL)
+				{
 					r ^= (r >> 2);
-					if (un <= 0x00000002UL) {
+					if (un <= 0x00000002UL)
+					{
 						r ^= (r >> 1);
 					}
 				}
@@ -435,62 +416,46 @@ int b3PgsJacobiSolver::b3RandInt2 (int n)
 		}
 	}
 
-	return (int) (r % un);
+	return (int)(r % un);
 }
 
-
-
-void	b3PgsJacobiSolver::initSolverBody(int bodyIndex, b3SolverBody* solverBody, b3RigidBodyData* rb)
+void b3PgsJacobiSolver::initSolverBody(int bodyIndex, b3SolverBody* solverBody, b3RigidBodyData* rb)
 {
-
-	solverBody->m_deltaLinearVelocity.setValue(0.f,0.f,0.f);
-	solverBody->m_deltaAngularVelocity.setValue(0.f,0.f,0.f);
-	solverBody->internalGetPushVelocity().setValue(0.f,0.f,0.f);
-	solverBody->internalGetTurnVelocity().setValue(0.f,0.f,0.f);
+	solverBody->m_deltaLinearVelocity.setValue(0.f, 0.f, 0.f);
+	solverBody->m_deltaAngularVelocity.setValue(0.f, 0.f, 0.f);
+	solverBody->internalGetPushVelocity().setValue(0.f, 0.f, 0.f);
+	solverBody->internalGetTurnVelocity().setValue(0.f, 0.f, 0.f);
 
 	if (rb)
 	{
 		solverBody->m_worldTransform = getWorldTransform(rb);
-		solverBody->internalSetInvMass(b3MakeVector3(rb->m_invMass,rb->m_invMass,rb->m_invMass));
+		solverBody->internalSetInvMass(b3MakeVector3(rb->m_invMass, rb->m_invMass, rb->m_invMass));
 		solverBody->m_originalBodyIndex = bodyIndex;
-		solverBody->m_angularFactor = b3MakeVector3(1,1,1);
-		solverBody->m_linearFactor = b3MakeVector3(1,1,1);
+		solverBody->m_angularFactor = b3MakeVector3(1, 1, 1);
+		solverBody->m_linearFactor = b3MakeVector3(1, 1, 1);
 		solverBody->m_linearVelocity = getLinearVelocity(rb);
 		solverBody->m_angularVelocity = getAngularVelocity(rb);
-	} else
+	}
+	else
 	{
 		solverBody->m_worldTransform.setIdentity();
-		solverBody->internalSetInvMass(b3MakeVector3(0,0,0));
+		solverBody->internalSetInvMass(b3MakeVector3(0, 0, 0));
 		solverBody->m_originalBodyIndex = bodyIndex;
-		solverBody->m_angularFactor.setValue(1,1,1);
-		solverBody->m_linearFactor.setValue(1,1,1);
-		solverBody->m_linearVelocity.setValue(0,0,0);
-		solverBody->m_angularVelocity.setValue(0,0,0);
+		solverBody->m_angularFactor.setValue(1, 1, 1);
+		solverBody->m_linearFactor.setValue(1, 1, 1);
+		solverBody->m_linearVelocity.setValue(0, 0, 0);
+		solverBody->m_angularVelocity.setValue(0, 0, 0);
 	}
-
-
 }
 
-
-
-
-
-
 b3Scalar b3PgsJacobiSolver::restitutionCurve(b3Scalar rel_vel, b3Scalar restitution)
 {
 	b3Scalar rest = restitution * -rel_vel;
 	return rest;
 }
 
-
-
-
-
-
-void b3PgsJacobiSolver::setupFrictionConstraint(b3RigidBodyData* bodies,b3InertiaData* inertias, b3SolverConstraint& solverConstraint, const b3Vector3& normalAxis,int  solverBodyIdA,int solverBodyIdB,b3ContactPoint& cp,const b3Vector3& rel_pos1,const b3Vector3& rel_pos2,b3RigidBodyData* colObj0,b3RigidBodyData* colObj1, b3Scalar relaxation, b3Scalar desiredVelocity, b3Scalar cfmSlip)
+void b3PgsJacobiSolver::setupFrictionConstraint(b3RigidBodyData* bodies, b3InertiaData* inertias, b3SolverConstraint& solverConstraint, const b3Vector3& normalAxis, int solverBodyIdA, int solverBodyIdB, b3ContactPoint& cp, const b3Vector3& rel_pos1, const b3Vector3& rel_pos2, b3RigidBodyData* colObj0, b3RigidBodyData* colObj1, b3Scalar relaxation, b3Scalar desiredVelocity, b3Scalar cfmSlip)
 {
-
-	
 	solverConstraint.m_contactNormal = normalAxis;
 	b3SolverBody& solverBodyA = m_tmpSolverBodyPool[solverBodyIdA];
 	b3SolverBody& solverBodyB = m_tmpSolverBodyPool[solverBodyIdB];
@@ -498,7 +463,6 @@ void b3PgsJacobiSolver::setupFrictionConstraint(b3RigidBodyData* bodies,b3Inerti
 	b3RigidBodyData* body0 = &bodies[solverBodyA.m_originalBodyIndex];
 	b3RigidBodyData* body1 = &bodies[solverBodyB.m_originalBodyIndex];
 
-
 	solverConstraint.m_solverBodyIdA = solverBodyIdA;
 	solverConstraint.m_solverBodyIdB = solverBodyIdB;
 
@@ -511,12 +475,12 @@ void b3PgsJacobiSolver::setupFrictionConstraint(b3RigidBodyData* bodies,b3Inerti
 	{
 		b3Vector3 ftorqueAxis1 = rel_pos1.cross(solverConstraint.m_contactNormal);
 		solverConstraint.m_relpos1CrossNormal = ftorqueAxis1;
-		solverConstraint.m_angularComponentA = body0 ? getInvInertiaTensorWorld(&inertias[solverBodyA.m_originalBodyIndex])*ftorqueAxis1 : b3MakeVector3(0,0,0);
+		solverConstraint.m_angularComponentA = body0 ? getInvInertiaTensorWorld(&inertias[solverBodyA.m_originalBodyIndex]) * ftorqueAxis1 : b3MakeVector3(0, 0, 0);
 	}
 	{
 		b3Vector3 ftorqueAxis1 = rel_pos2.cross(-solverConstraint.m_contactNormal);
 		solverConstraint.m_relpos2CrossNormal = ftorqueAxis1;
-		solverConstraint.m_angularComponentB = body1 ? getInvInertiaTensorWorld(&inertias[solverBodyB.m_originalBodyIndex])*ftorqueAxis1 : b3MakeVector3(0,0,0);
+		solverConstraint.m_angularComponentB = body1 ? getInvInertiaTensorWorld(&inertias[solverBodyB.m_originalBodyIndex]) * ftorqueAxis1 : b3MakeVector3(0, 0, 0);
 	}
 
 	b3Scalar scaledDenom;
@@ -527,72 +491,66 @@ void b3PgsJacobiSolver::setupFrictionConstraint(b3RigidBodyData* bodies,b3Inerti
 		b3Scalar denom1 = 0.f;
 		if (body0)
 		{
-			vec = ( solverConstraint.m_angularComponentA).cross(rel_pos1);
+			vec = (solverConstraint.m_angularComponentA).cross(rel_pos1);
 			denom0 = body0->m_invMass + normalAxis.dot(vec);
 		}
 		if (body1)
 		{
-			vec = ( -solverConstraint.m_angularComponentB).cross(rel_pos2);
+			vec = (-solverConstraint.m_angularComponentB).cross(rel_pos2);
 			denom1 = body1->m_invMass + normalAxis.dot(vec);
 		}
 
 		b3Scalar denom;
 		if (m_usePgs)
 		{
-			scaledDenom = denom = relaxation/(denom0+denom1);
-		} else
+			scaledDenom = denom = relaxation / (denom0 + denom1);
+		}
+		else
 		{
-			denom = relaxation/(denom0+denom1);
-			b3Scalar countA = body0->m_invMass ? b3Scalar(m_bodyCount[solverBodyA.m_originalBodyIndex]): 1.f;
-			b3Scalar countB = body1->m_invMass ? b3Scalar(m_bodyCount[solverBodyB.m_originalBodyIndex]): 1.f;
+			denom = relaxation / (denom0 + denom1);
+			b3Scalar countA = body0->m_invMass ? b3Scalar(m_bodyCount[solverBodyA.m_originalBodyIndex]) : 1.f;
+			b3Scalar countB = body1->m_invMass ? b3Scalar(m_bodyCount[solverBodyB.m_originalBodyIndex]) : 1.f;
 
-			scaledDenom = relaxation/(denom0*countA+denom1*countB);
+			scaledDenom = relaxation / (denom0 * countA + denom1 * countB);
 		}
 
 		solverConstraint.m_jacDiagABInv = denom;
 	}
 
 	{
-		
-
 		b3Scalar rel_vel;
-		b3Scalar vel1Dotn = solverConstraint.m_contactNormal.dot(body0?solverBodyA.m_linearVelocity:b3MakeVector3(0,0,0)) 
-			+ solverConstraint.m_relpos1CrossNormal.dot(body0?solverBodyA.m_angularVelocity:b3MakeVector3(0,0,0));
-		b3Scalar vel2Dotn = -solverConstraint.m_contactNormal.dot(body1?solverBodyB.m_linearVelocity:b3MakeVector3(0,0,0)) 
-			+ solverConstraint.m_relpos2CrossNormal.dot(body1?solverBodyB.m_angularVelocity:b3MakeVector3(0,0,0));
+		b3Scalar vel1Dotn = solverConstraint.m_contactNormal.dot(body0 ? solverBodyA.m_linearVelocity : b3MakeVector3(0, 0, 0)) + solverConstraint.m_relpos1CrossNormal.dot(body0 ? solverBodyA.m_angularVelocity : b3MakeVector3(0, 0, 0));
+		b3Scalar vel2Dotn = -solverConstraint.m_contactNormal.dot(body1 ? solverBodyB.m_linearVelocity : b3MakeVector3(0, 0, 0)) + solverConstraint.m_relpos2CrossNormal.dot(body1 ? solverBodyB.m_angularVelocity : b3MakeVector3(0, 0, 0));
 
-		rel_vel = vel1Dotn+vel2Dotn;
+		rel_vel = vel1Dotn + vel2Dotn;
 
-//		b3Scalar positionalError = 0.f;
+		//		b3Scalar positionalError = 0.f;
 
-		b3SimdScalar velocityError =  desiredVelocity - rel_vel;
-		b3SimdScalar	velocityImpulse = velocityError * b3SimdScalar(scaledDenom);//solverConstraint.m_jacDiagABInv);
+		b3SimdScalar velocityError = desiredVelocity - rel_vel;
+		b3SimdScalar velocityImpulse = velocityError * b3SimdScalar(scaledDenom);  //solverConstraint.m_jacDiagABInv);
 		solverConstraint.m_rhs = velocityImpulse;
 		solverConstraint.m_cfm = cfmSlip;
 		solverConstraint.m_lowerLimit = 0;
 		solverConstraint.m_upperLimit = 1e10f;
-		
 	}
 }
 
-b3SolverConstraint&	b3PgsJacobiSolver::addFrictionConstraint(b3RigidBodyData* bodies,b3InertiaData* inertias, const b3Vector3& normalAxis,int solverBodyIdA,int solverBodyIdB,int frictionIndex,b3ContactPoint& cp,const b3Vector3& rel_pos1,const b3Vector3& rel_pos2,b3RigidBodyData* colObj0,b3RigidBodyData* colObj1, b3Scalar relaxation, b3Scalar desiredVelocity, b3Scalar cfmSlip)
+b3SolverConstraint& b3PgsJacobiSolver::addFrictionConstraint(b3RigidBodyData* bodies, b3InertiaData* inertias, const b3Vector3& normalAxis, int solverBodyIdA, int solverBodyIdB, int frictionIndex, b3ContactPoint& cp, const b3Vector3& rel_pos1, const b3Vector3& rel_pos2, b3RigidBodyData* colObj0, b3RigidBodyData* colObj1, b3Scalar relaxation, b3Scalar desiredVelocity, b3Scalar cfmSlip)
 {
 	b3SolverConstraint& solverConstraint = m_tmpSolverContactFrictionConstraintPool.expandNonInitializing();
 	solverConstraint.m_frictionIndex = frictionIndex;
-	setupFrictionConstraint(bodies,inertias,solverConstraint, normalAxis, solverBodyIdA, solverBodyIdB, cp, rel_pos1, rel_pos2, 
+	setupFrictionConstraint(bodies, inertias, solverConstraint, normalAxis, solverBodyIdA, solverBodyIdB, cp, rel_pos1, rel_pos2,
 							colObj0, colObj1, relaxation, desiredVelocity, cfmSlip);
 	return solverConstraint;
 }
 
-
-void b3PgsJacobiSolver::setupRollingFrictionConstraint(b3RigidBodyData* bodies,b3InertiaData* inertias,	b3SolverConstraint& solverConstraint, const b3Vector3& normalAxis1,int solverBodyIdA,int  solverBodyIdB,
-									b3ContactPoint& cp,const b3Vector3& rel_pos1,const b3Vector3& rel_pos2,
-									b3RigidBodyData* colObj0,b3RigidBodyData* colObj1, b3Scalar relaxation, 
-									b3Scalar desiredVelocity, b3Scalar cfmSlip)
+void b3PgsJacobiSolver::setupRollingFrictionConstraint(b3RigidBodyData* bodies, b3InertiaData* inertias, b3SolverConstraint& solverConstraint, const b3Vector3& normalAxis1, int solverBodyIdA, int solverBodyIdB,
+													   b3ContactPoint& cp, const b3Vector3& rel_pos1, const b3Vector3& rel_pos2,
+													   b3RigidBodyData* colObj0, b3RigidBodyData* colObj1, b3Scalar relaxation,
+													   b3Scalar desiredVelocity, b3Scalar cfmSlip)
 
 {
-	b3Vector3 normalAxis=b3MakeVector3(0,0,0);
-
+	b3Vector3 normalAxis = b3MakeVector3(0, 0, 0);
 
 	solverConstraint.m_contactNormal = normalAxis;
 	b3SolverBody& solverBodyA = m_tmpSolverBodyPool[solverBodyIdA];
@@ -613,272 +571,244 @@ void b3PgsJacobiSolver::setupRollingFrictionConstraint(b3RigidBodyData* bodies,b
 	{
 		b3Vector3 ftorqueAxis1 = -normalAxis1;
 		solverConstraint.m_relpos1CrossNormal = ftorqueAxis1;
-		solverConstraint.m_angularComponentA = body0 ? getInvInertiaTensorWorld(&inertias[solverBodyA.m_originalBodyIndex])*ftorqueAxis1 : b3MakeVector3(0,0,0);
+		solverConstraint.m_angularComponentA = body0 ? getInvInertiaTensorWorld(&inertias[solverBodyA.m_originalBodyIndex]) * ftorqueAxis1 : b3MakeVector3(0, 0, 0);
 	}
 	{
 		b3Vector3 ftorqueAxis1 = normalAxis1;
 		solverConstraint.m_relpos2CrossNormal = ftorqueAxis1;
-		solverConstraint.m_angularComponentB = body1 ? getInvInertiaTensorWorld(&inertias[solverBodyB.m_originalBodyIndex])*ftorqueAxis1 : b3MakeVector3(0,0,0);
+		solverConstraint.m_angularComponentB = body1 ? getInvInertiaTensorWorld(&inertias[solverBodyB.m_originalBodyIndex]) * ftorqueAxis1 : b3MakeVector3(0, 0, 0);
 	}
 
-
 	{
-		b3Vector3 iMJaA = body0?getInvInertiaTensorWorld(&inertias[solverBodyA.m_originalBodyIndex])*solverConstraint.m_relpos1CrossNormal:b3MakeVector3(0,0,0);
-		b3Vector3 iMJaB = body1?getInvInertiaTensorWorld(&inertias[solverBodyB.m_originalBodyIndex])*solverConstraint.m_relpos2CrossNormal:b3MakeVector3(0,0,0);
+		b3Vector3 iMJaA = body0 ? getInvInertiaTensorWorld(&inertias[solverBodyA.m_originalBodyIndex]) * solverConstraint.m_relpos1CrossNormal : b3MakeVector3(0, 0, 0);
+		b3Vector3 iMJaB = body1 ? getInvInertiaTensorWorld(&inertias[solverBodyB.m_originalBodyIndex]) * solverConstraint.m_relpos2CrossNormal : b3MakeVector3(0, 0, 0);
 		b3Scalar sum = 0;
 		sum += iMJaA.dot(solverConstraint.m_relpos1CrossNormal);
 		sum += iMJaB.dot(solverConstraint.m_relpos2CrossNormal);
-		solverConstraint.m_jacDiagABInv = b3Scalar(1.)/sum;
+		solverConstraint.m_jacDiagABInv = b3Scalar(1.) / sum;
 	}
 
 	{
-		
-
 		b3Scalar rel_vel;
-		b3Scalar vel1Dotn = solverConstraint.m_contactNormal.dot(body0?solverBodyA.m_linearVelocity:b3MakeVector3(0,0,0)) 
-			+ solverConstraint.m_relpos1CrossNormal.dot(body0?solverBodyA.m_angularVelocity:b3MakeVector3(0,0,0));
-		b3Scalar vel2Dotn = -solverConstraint.m_contactNormal.dot(body1?solverBodyB.m_linearVelocity:b3MakeVector3(0,0,0)) 
-			+ solverConstraint.m_relpos2CrossNormal.dot(body1?solverBodyB.m_angularVelocity:b3MakeVector3(0,0,0));
+		b3Scalar vel1Dotn = solverConstraint.m_contactNormal.dot(body0 ? solverBodyA.m_linearVelocity : b3MakeVector3(0, 0, 0)) + solverConstraint.m_relpos1CrossNormal.dot(body0 ? solverBodyA.m_angularVelocity : b3MakeVector3(0, 0, 0));
+		b3Scalar vel2Dotn = -solverConstraint.m_contactNormal.dot(body1 ? solverBodyB.m_linearVelocity : b3MakeVector3(0, 0, 0)) + solverConstraint.m_relpos2CrossNormal.dot(body1 ? solverBodyB.m_angularVelocity : b3MakeVector3(0, 0, 0));
 
-		rel_vel = vel1Dotn+vel2Dotn;
+		rel_vel = vel1Dotn + vel2Dotn;
 
-//		b3Scalar positionalError = 0.f;
+		//		b3Scalar positionalError = 0.f;
 
-		b3SimdScalar velocityError =  desiredVelocity - rel_vel;
-		b3SimdScalar	velocityImpulse = velocityError * b3SimdScalar(solverConstraint.m_jacDiagABInv);
+		b3SimdScalar velocityError = desiredVelocity - rel_vel;
+		b3SimdScalar velocityImpulse = velocityError * b3SimdScalar(solverConstraint.m_jacDiagABInv);
 		solverConstraint.m_rhs = velocityImpulse;
 		solverConstraint.m_cfm = cfmSlip;
 		solverConstraint.m_lowerLimit = 0;
 		solverConstraint.m_upperLimit = 1e10f;
-		
 	}
 }
 
-
-
-
-
-
-
-
-b3SolverConstraint&	b3PgsJacobiSolver::addRollingFrictionConstraint(b3RigidBodyData* bodies,b3InertiaData* inertias,const b3Vector3& normalAxis,int solverBodyIdA,int solverBodyIdB,int frictionIndex,b3ContactPoint& cp,const b3Vector3& rel_pos1,const b3Vector3& rel_pos2,b3RigidBodyData* colObj0,b3RigidBodyData* colObj1, b3Scalar relaxation, b3Scalar desiredVelocity, b3Scalar cfmSlip)
+b3SolverConstraint& b3PgsJacobiSolver::addRollingFrictionConstraint(b3RigidBodyData* bodies, b3InertiaData* inertias, const b3Vector3& normalAxis, int solverBodyIdA, int solverBodyIdB, int frictionIndex, b3ContactPoint& cp, const b3Vector3& rel_pos1, const b3Vector3& rel_pos2, b3RigidBodyData* colObj0, b3RigidBodyData* colObj1, b3Scalar relaxation, b3Scalar desiredVelocity, b3Scalar cfmSlip)
 {
 	b3SolverConstraint& solverConstraint = m_tmpSolverContactRollingFrictionConstraintPool.expandNonInitializing();
 	solverConstraint.m_frictionIndex = frictionIndex;
-	setupRollingFrictionConstraint(bodies,inertias,solverConstraint, normalAxis, solverBodyIdA, solverBodyIdB, cp, rel_pos1, rel_pos2, 
-							colObj0, colObj1, relaxation, desiredVelocity, cfmSlip);
+	setupRollingFrictionConstraint(bodies, inertias, solverConstraint, normalAxis, solverBodyIdA, solverBodyIdB, cp, rel_pos1, rel_pos2,
+								   colObj0, colObj1, relaxation, desiredVelocity, cfmSlip);
 	return solverConstraint;
 }
 
-
-int	b3PgsJacobiSolver::getOrInitSolverBody(int bodyIndex, b3RigidBodyData* bodies,b3InertiaData* inertias)
+int b3PgsJacobiSolver::getOrInitSolverBody(int bodyIndex, b3RigidBodyData* bodies, b3InertiaData* inertias)
 {
 	//b3Assert(bodyIndex< m_tmpSolverBodyPool.size());
 
 	b3RigidBodyData& body = bodies[bodyIndex];
 	int curIndex = -1;
-	if (m_usePgs || body.m_invMass==0.f)
+	if (m_usePgs || body.m_invMass == 0.f)
 	{
-		if (m_bodyCount[bodyIndex]<0)
+		if (m_bodyCount[bodyIndex] < 0)
 		{
 			curIndex = m_tmpSolverBodyPool.size();
 			b3SolverBody& solverBody = m_tmpSolverBodyPool.expand();
-			initSolverBody(bodyIndex,&solverBody,&body);
+			initSolverBody(bodyIndex, &solverBody, &body);
 			solverBody.m_originalBodyIndex = bodyIndex;
 			m_bodyCount[bodyIndex] = curIndex;
-		} else
+		}
+		else
 		{
 			curIndex = m_bodyCount[bodyIndex];
 		}
-	} else
+	}
+	else
 	{
-		b3Assert(m_bodyCount[bodyIndex]>0);
+		b3Assert(m_bodyCount[bodyIndex] > 0);
 		m_bodyCountCheck[bodyIndex]++;
 		curIndex = m_tmpSolverBodyPool.size();
 		b3SolverBody& solverBody = m_tmpSolverBodyPool.expand();
-		initSolverBody(bodyIndex,&solverBody,&body);
+		initSolverBody(bodyIndex, &solverBody, &body);
 		solverBody.m_originalBodyIndex = bodyIndex;
 	}
 
-	b3Assert(curIndex>=0);
+	b3Assert(curIndex >= 0);
 	return curIndex;
-
 }
 #include <stdio.h>
 
-
-void b3PgsJacobiSolver::setupContactConstraint(b3RigidBodyData* bodies, b3InertiaData* inertias,b3SolverConstraint& solverConstraint, 
-																 int solverBodyIdA, int solverBodyIdB,
-																 b3ContactPoint& cp, const b3ContactSolverInfo& infoGlobal,
-																 b3Vector3& vel, b3Scalar& rel_vel, b3Scalar& relaxation,
-																 b3Vector3& rel_pos1, b3Vector3& rel_pos2)
-{
-			
-			const b3Vector3& pos1 = cp.getPositionWorldOnA();
-			const b3Vector3& pos2 = cp.getPositionWorldOnB();
-
-			b3SolverBody* bodyA = &m_tmpSolverBodyPool[solverBodyIdA];
-			b3SolverBody* bodyB = &m_tmpSolverBodyPool[solverBodyIdB];
-
-			b3RigidBodyData* rb0 = &bodies[bodyA->m_originalBodyIndex];
-			b3RigidBodyData* rb1 = &bodies[bodyB->m_originalBodyIndex];
-
-//			b3Vector3 rel_pos1 = pos1 - colObj0->getWorldTransform().getOrigin(); 
-//			b3Vector3 rel_pos2 = pos2 - colObj1->getWorldTransform().getOrigin();
-			rel_pos1 = pos1 - bodyA->getWorldTransform().getOrigin(); 
-			rel_pos2 = pos2 - bodyB->getWorldTransform().getOrigin();
-
-			relaxation = 1.f;
-
-			b3Vector3 torqueAxis0 = rel_pos1.cross(cp.m_normalWorldOnB);
-			solverConstraint.m_angularComponentA = rb0 ? getInvInertiaTensorWorld(&inertias[bodyA->m_originalBodyIndex])*torqueAxis0 : b3MakeVector3(0,0,0);
-			b3Vector3 torqueAxis1 = rel_pos2.cross(cp.m_normalWorldOnB);		
-			solverConstraint.m_angularComponentB = rb1 ? getInvInertiaTensorWorld(&inertias[bodyB->m_originalBodyIndex])*-torqueAxis1 : b3MakeVector3(0,0,0);
-
-			b3Scalar scaledDenom;
-				{
-#ifdef COMPUTE_IMPULSE_DENOM
-					b3Scalar denom0 = rb0->computeImpulseDenominator(pos1,cp.m_normalWorldOnB);
-					b3Scalar denom1 = rb1->computeImpulseDenominator(pos2,cp.m_normalWorldOnB);
-#else							
-					b3Vector3 vec;
-					b3Scalar denom0 = 0.f;
-					b3Scalar denom1 = 0.f;
-					if (rb0)
-					{
-						vec = ( solverConstraint.m_angularComponentA).cross(rel_pos1);
-						denom0 = rb0->m_invMass + cp.m_normalWorldOnB.dot(vec);
-					}
-					if (rb1)
-					{
-						vec = ( -solverConstraint.m_angularComponentB).cross(rel_pos2);
-						denom1 = rb1->m_invMass + cp.m_normalWorldOnB.dot(vec);
-					}
-#endif //COMPUTE_IMPULSE_DENOM		
-
-					
-					b3Scalar denom;
-					if (m_usePgs)
-					{
-						scaledDenom = denom = relaxation/(denom0+denom1);
-					} else
-					{
-						denom = relaxation/(denom0+denom1);
-
-						b3Scalar countA = rb0->m_invMass? b3Scalar(m_bodyCount[bodyA->m_originalBodyIndex]) : 1.f;
-						b3Scalar countB = rb1->m_invMass? b3Scalar(m_bodyCount[bodyB->m_originalBodyIndex]) : 1.f;
-						scaledDenom = relaxation/(denom0*countA+denom1*countB);
-					}
-					solverConstraint.m_jacDiagABInv = denom;
-				}
-
-				solverConstraint.m_contactNormal = cp.m_normalWorldOnB;
-				solverConstraint.m_relpos1CrossNormal = torqueAxis0;
-				solverConstraint.m_relpos2CrossNormal = -torqueAxis1;
-
-				b3Scalar restitution = 0.f;
-				b3Scalar penetration = cp.getDistance()+infoGlobal.m_linearSlop;
-
-				{
-					b3Vector3 vel1,vel2;
-
-					vel1 = rb0? getVelocityInLocalPoint(rb0,rel_pos1) : b3MakeVector3(0,0,0);
-					vel2 = rb1? getVelocityInLocalPoint(rb1, rel_pos2) : b3MakeVector3(0,0,0);
-
-	//			b3Vector3 vel2 = rb1 ? rb1->getVelocityInLocalPoint(rel_pos2) : b3Vector3(0,0,0);
-					vel  = vel1 - vel2;
-					rel_vel = cp.m_normalWorldOnB.dot(vel);
-
-					
-
-					solverConstraint.m_friction = cp.m_combinedFriction;
-
-				
-					restitution =  restitutionCurve(rel_vel, cp.m_combinedRestitution);
-					if (restitution <= b3Scalar(0.))
-					{
-						restitution = 0.f;
-					};
-				}
-
-
-				///warm starting (or zero if disabled)
-				if (infoGlobal.m_solverMode & B3_SOLVER_USE_WARMSTARTING)
-				{
-					solverConstraint.m_appliedImpulse = cp.m_appliedImpulse * infoGlobal.m_warmstartingFactor;
-					if (rb0)
-						bodyA->internalApplyImpulse(solverConstraint.m_contactNormal*bodyA->internalGetInvMass(),solverConstraint.m_angularComponentA,solverConstraint.m_appliedImpulse);
-					if (rb1)
-						bodyB->internalApplyImpulse(solverConstraint.m_contactNormal*bodyB->internalGetInvMass(),-solverConstraint.m_angularComponentB,-(b3Scalar)solverConstraint.m_appliedImpulse);
-				} else
-				{
-					solverConstraint.m_appliedImpulse = 0.f;
-				}
-
-				solverConstraint.m_appliedPushImpulse = 0.f;
-
-				{
-					b3Scalar vel1Dotn = solverConstraint.m_contactNormal.dot(rb0?bodyA->m_linearVelocity:b3MakeVector3(0,0,0)) 
-						+ solverConstraint.m_relpos1CrossNormal.dot(rb0?bodyA->m_angularVelocity:b3MakeVector3(0,0,0));
-					b3Scalar vel2Dotn = -solverConstraint.m_contactNormal.dot(rb1?bodyB->m_linearVelocity:b3MakeVector3(0,0,0)) 
-						+ solverConstraint.m_relpos2CrossNormal.dot(rb1?bodyB->m_angularVelocity:b3MakeVector3(0,0,0));
-					b3Scalar rel_vel = vel1Dotn+vel2Dotn;
-
-					b3Scalar positionalError = 0.f;
-					b3Scalar	velocityError = restitution - rel_vel;// * damping;
-					
-
-					b3Scalar erp = infoGlobal.m_erp2;
-					if (!infoGlobal.m_splitImpulse || (penetration > infoGlobal.m_splitImpulsePenetrationThreshold))
-					{
-						erp = infoGlobal.m_erp;
-					}
-
-					if (penetration>0)
-					{
-						positionalError = 0;
-
-						velocityError -= penetration / infoGlobal.m_timeStep;
-					} else
-					{
-						positionalError = -penetration * erp/infoGlobal.m_timeStep;
-					}
-
-					b3Scalar  penetrationImpulse = positionalError*scaledDenom;//solverConstraint.m_jacDiagABInv;
-					b3Scalar velocityImpulse = velocityError *scaledDenom;//solverConstraint.m_jacDiagABInv;
-
-					if (!infoGlobal.m_splitImpulse || (penetration > infoGlobal.m_splitImpulsePenetrationThreshold))
-					{
-						//combine position and velocity into rhs
-						solverConstraint.m_rhs = penetrationImpulse+velocityImpulse;
-						solverConstraint.m_rhsPenetration = 0.f;
-
-					} else
-					{
-						//split position and velocity into rhs and m_rhsPenetration
-						solverConstraint.m_rhs = velocityImpulse;
-						solverConstraint.m_rhsPenetration = penetrationImpulse;
-					}
-					solverConstraint.m_cfm = 0.f;
-					solverConstraint.m_lowerLimit = 0;
-					solverConstraint.m_upperLimit = 1e10f;
-				}
-
-
-
-
-}
-
-
-
-void b3PgsJacobiSolver::setFrictionConstraintImpulse( b3RigidBodyData* bodies, b3InertiaData* inertias,b3SolverConstraint& solverConstraint, 
-																		int solverBodyIdA, int solverBodyIdB,
-																 b3ContactPoint& cp, const b3ContactSolverInfo& infoGlobal)
+void b3PgsJacobiSolver::setupContactConstraint(b3RigidBodyData* bodies, b3InertiaData* inertias, b3SolverConstraint& solverConstraint,
+											   int solverBodyIdA, int solverBodyIdB,
+											   b3ContactPoint& cp, const b3ContactSolverInfo& infoGlobal,
+											   b3Vector3& vel, b3Scalar& rel_vel, b3Scalar& relaxation,
+											   b3Vector3& rel_pos1, b3Vector3& rel_pos2)
 {
+	const b3Vector3& pos1 = cp.getPositionWorldOnA();
+	const b3Vector3& pos2 = cp.getPositionWorldOnB();
 
 	b3SolverBody* bodyA = &m_tmpSolverBodyPool[solverBodyIdA];
 	b3SolverBody* bodyB = &m_tmpSolverBodyPool[solverBodyIdB];
 
+	b3RigidBodyData* rb0 = &bodies[bodyA->m_originalBodyIndex];
+	b3RigidBodyData* rb1 = &bodies[bodyB->m_originalBodyIndex];
+
+	//			b3Vector3 rel_pos1 = pos1 - colObj0->getWorldTransform().getOrigin();
+	//			b3Vector3 rel_pos2 = pos2 - colObj1->getWorldTransform().getOrigin();
+	rel_pos1 = pos1 - bodyA->getWorldTransform().getOrigin();
+	rel_pos2 = pos2 - bodyB->getWorldTransform().getOrigin();
+
+	relaxation = 1.f;
+
+	b3Vector3 torqueAxis0 = rel_pos1.cross(cp.m_normalWorldOnB);
+	solverConstraint.m_angularComponentA = rb0 ? getInvInertiaTensorWorld(&inertias[bodyA->m_originalBodyIndex]) * torqueAxis0 : b3MakeVector3(0, 0, 0);
+	b3Vector3 torqueAxis1 = rel_pos2.cross(cp.m_normalWorldOnB);
+	solverConstraint.m_angularComponentB = rb1 ? getInvInertiaTensorWorld(&inertias[bodyB->m_originalBodyIndex]) * -torqueAxis1 : b3MakeVector3(0, 0, 0);
+
+	b3Scalar scaledDenom;
+	{
+#ifdef COMPUTE_IMPULSE_DENOM
+		b3Scalar denom0 = rb0->computeImpulseDenominator(pos1, cp.m_normalWorldOnB);
+		b3Scalar denom1 = rb1->computeImpulseDenominator(pos2, cp.m_normalWorldOnB);
+#else
+		b3Vector3 vec;
+		b3Scalar denom0 = 0.f;
+		b3Scalar denom1 = 0.f;
+		if (rb0)
+		{
+			vec = (solverConstraint.m_angularComponentA).cross(rel_pos1);
+			denom0 = rb0->m_invMass + cp.m_normalWorldOnB.dot(vec);
+		}
+		if (rb1)
+		{
+			vec = (-solverConstraint.m_angularComponentB).cross(rel_pos2);
+			denom1 = rb1->m_invMass + cp.m_normalWorldOnB.dot(vec);
+		}
+#endif  //COMPUTE_IMPULSE_DENOM
+
+		b3Scalar denom;
+		if (m_usePgs)
+		{
+			scaledDenom = denom = relaxation / (denom0 + denom1);
+		}
+		else
+		{
+			denom = relaxation / (denom0 + denom1);
+
+			b3Scalar countA = rb0->m_invMass ? b3Scalar(m_bodyCount[bodyA->m_originalBodyIndex]) : 1.f;
+			b3Scalar countB = rb1->m_invMass ? b3Scalar(m_bodyCount[bodyB->m_originalBodyIndex]) : 1.f;
+			scaledDenom = relaxation / (denom0 * countA + denom1 * countB);
+		}
+		solverConstraint.m_jacDiagABInv = denom;
+	}
+
+	solverConstraint.m_contactNormal = cp.m_normalWorldOnB;
+	solverConstraint.m_relpos1CrossNormal = torqueAxis0;
+	solverConstraint.m_relpos2CrossNormal = -torqueAxis1;
+
+	b3Scalar restitution = 0.f;
+	b3Scalar penetration = cp.getDistance() + infoGlobal.m_linearSlop;
+
+	{
+		b3Vector3 vel1, vel2;
+
+		vel1 = rb0 ? getVelocityInLocalPoint(rb0, rel_pos1) : b3MakeVector3(0, 0, 0);
+		vel2 = rb1 ? getVelocityInLocalPoint(rb1, rel_pos2) : b3MakeVector3(0, 0, 0);
+
+		//			b3Vector3 vel2 = rb1 ? rb1->getVelocityInLocalPoint(rel_pos2) : b3Vector3(0,0,0);
+		vel = vel1 - vel2;
+		rel_vel = cp.m_normalWorldOnB.dot(vel);
+
+		solverConstraint.m_friction = cp.m_combinedFriction;
+
+		restitution = restitutionCurve(rel_vel, cp.m_combinedRestitution);
+		if (restitution <= b3Scalar(0.))
+		{
+			restitution = 0.f;
+		};
+	}
+
+	///warm starting (or zero if disabled)
+	if (infoGlobal.m_solverMode & B3_SOLVER_USE_WARMSTARTING)
+	{
+		solverConstraint.m_appliedImpulse = cp.m_appliedImpulse * infoGlobal.m_warmstartingFactor;
+		if (rb0)
+			bodyA->internalApplyImpulse(solverConstraint.m_contactNormal * bodyA->internalGetInvMass(), solverConstraint.m_angularComponentA, solverConstraint.m_appliedImpulse);
+		if (rb1)
+			bodyB->internalApplyImpulse(solverConstraint.m_contactNormal * bodyB->internalGetInvMass(), -solverConstraint.m_angularComponentB, -(b3Scalar)solverConstraint.m_appliedImpulse);
+	}
+	else
+	{
+		solverConstraint.m_appliedImpulse = 0.f;
+	}
+
+	solverConstraint.m_appliedPushImpulse = 0.f;
+
+	{
+		b3Scalar vel1Dotn = solverConstraint.m_contactNormal.dot(rb0 ? bodyA->m_linearVelocity : b3MakeVector3(0, 0, 0)) + solverConstraint.m_relpos1CrossNormal.dot(rb0 ? bodyA->m_angularVelocity : b3MakeVector3(0, 0, 0));
+		b3Scalar vel2Dotn = -solverConstraint.m_contactNormal.dot(rb1 ? bodyB->m_linearVelocity : b3MakeVector3(0, 0, 0)) + solverConstraint.m_relpos2CrossNormal.dot(rb1 ? bodyB->m_angularVelocity : b3MakeVector3(0, 0, 0));
+		b3Scalar rel_vel = vel1Dotn + vel2Dotn;
+
+		b3Scalar positionalError = 0.f;
+		b3Scalar velocityError = restitution - rel_vel;  // * damping;
+
+		b3Scalar erp = infoGlobal.m_erp2;
+		if (!infoGlobal.m_splitImpulse || (penetration > infoGlobal.m_splitImpulsePenetrationThreshold))
+		{
+			erp = infoGlobal.m_erp;
+		}
+
+		if (penetration > 0)
+		{
+			positionalError = 0;
+
+			velocityError -= penetration / infoGlobal.m_timeStep;
+		}
+		else
+		{
+			positionalError = -penetration * erp / infoGlobal.m_timeStep;
+		}
+
+		b3Scalar penetrationImpulse = positionalError * scaledDenom;  //solverConstraint.m_jacDiagABInv;
+		b3Scalar velocityImpulse = velocityError * scaledDenom;       //solverConstraint.m_jacDiagABInv;
+
+		if (!infoGlobal.m_splitImpulse || (penetration > infoGlobal.m_splitImpulsePenetrationThreshold))
+		{
+			//combine position and velocity into rhs
+			solverConstraint.m_rhs = penetrationImpulse + velocityImpulse;
+			solverConstraint.m_rhsPenetration = 0.f;
+		}
+		else
+		{
+			//split position and velocity into rhs and m_rhsPenetration
+			solverConstraint.m_rhs = velocityImpulse;
+			solverConstraint.m_rhsPenetration = penetrationImpulse;
+		}
+		solverConstraint.m_cfm = 0.f;
+		solverConstraint.m_lowerLimit = 0;
+		solverConstraint.m_upperLimit = 1e10f;
+	}
+}
+
+void b3PgsJacobiSolver::setFrictionConstraintImpulse(b3RigidBodyData* bodies, b3InertiaData* inertias, b3SolverConstraint& solverConstraint,
+													 int solverBodyIdA, int solverBodyIdB,
+													 b3ContactPoint& cp, const b3ContactSolverInfo& infoGlobal)
+{
+	b3SolverBody* bodyA = &m_tmpSolverBodyPool[solverBodyIdA];
+	b3SolverBody* bodyB = &m_tmpSolverBodyPool[solverBodyIdB];
 
 	{
 		b3SolverConstraint& frictionConstraint1 = m_tmpSolverContactFrictionConstraintPool[solverConstraint.m_frictionIndex];
@@ -886,10 +816,11 @@ void b3PgsJacobiSolver::setFrictionConstraintImpulse( b3RigidBodyData* bodies, b
 		{
 			frictionConstraint1.m_appliedImpulse = cp.m_appliedImpulseLateral1 * infoGlobal.m_warmstartingFactor;
 			if (bodies[bodyA->m_originalBodyIndex].m_invMass)
-				bodyA->internalApplyImpulse(frictionConstraint1.m_contactNormal*bodies[bodyA->m_originalBodyIndex].m_invMass,frictionConstraint1.m_angularComponentA,frictionConstraint1.m_appliedImpulse);
+				bodyA->internalApplyImpulse(frictionConstraint1.m_contactNormal * bodies[bodyA->m_originalBodyIndex].m_invMass, frictionConstraint1.m_angularComponentA, frictionConstraint1.m_appliedImpulse);
 			if (bodies[bodyB->m_originalBodyIndex].m_invMass)
-				bodyB->internalApplyImpulse(frictionConstraint1.m_contactNormal*bodies[bodyB->m_originalBodyIndex].m_invMass,-frictionConstraint1.m_angularComponentB,-(b3Scalar)frictionConstraint1.m_appliedImpulse);
-		} else
+				bodyB->internalApplyImpulse(frictionConstraint1.m_contactNormal * bodies[bodyB->m_originalBodyIndex].m_invMass, -frictionConstraint1.m_angularComponentB, -(b3Scalar)frictionConstraint1.m_appliedImpulse);
+		}
+		else
 		{
 			frictionConstraint1.m_appliedImpulse = 0.f;
 		}
@@ -897,51 +828,45 @@ void b3PgsJacobiSolver::setFrictionConstraintImpulse( b3RigidBodyData* bodies, b
 
 	if ((infoGlobal.m_solverMode & B3_SOLVER_USE_2_FRICTION_DIRECTIONS))
 	{
-		b3SolverConstraint& frictionConstraint2 = m_tmpSolverContactFrictionConstraintPool[solverConstraint.m_frictionIndex+1];
+		b3SolverConstraint& frictionConstraint2 = m_tmpSolverContactFrictionConstraintPool[solverConstraint.m_frictionIndex + 1];
 		if (infoGlobal.m_solverMode & B3_SOLVER_USE_WARMSTARTING)
 		{
-			frictionConstraint2.m_appliedImpulse = cp.m_appliedImpulseLateral2  * infoGlobal.m_warmstartingFactor;
+			frictionConstraint2.m_appliedImpulse = cp.m_appliedImpulseLateral2 * infoGlobal.m_warmstartingFactor;
 			if (bodies[bodyA->m_originalBodyIndex].m_invMass)
-				bodyA->internalApplyImpulse(frictionConstraint2.m_contactNormal*bodies[bodyA->m_originalBodyIndex].m_invMass,frictionConstraint2.m_angularComponentA,frictionConstraint2.m_appliedImpulse);
+				bodyA->internalApplyImpulse(frictionConstraint2.m_contactNormal * bodies[bodyA->m_originalBodyIndex].m_invMass, frictionConstraint2.m_angularComponentA, frictionConstraint2.m_appliedImpulse);
 			if (bodies[bodyB->m_originalBodyIndex].m_invMass)
-				bodyB->internalApplyImpulse(frictionConstraint2.m_contactNormal*bodies[bodyB->m_originalBodyIndex].m_invMass,-frictionConstraint2.m_angularComponentB,-(b3Scalar)frictionConstraint2.m_appliedImpulse);
-		} else
+				bodyB->internalApplyImpulse(frictionConstraint2.m_contactNormal * bodies[bodyB->m_originalBodyIndex].m_invMass, -frictionConstraint2.m_angularComponentB, -(b3Scalar)frictionConstraint2.m_appliedImpulse);
+		}
+		else
 		{
 			frictionConstraint2.m_appliedImpulse = 0.f;
 		}
 	}
 }
 
-
-
-
-void	b3PgsJacobiSolver::convertContact(b3RigidBodyData* bodies, b3InertiaData* inertias,b3Contact4* manifold,const b3ContactSolverInfo& infoGlobal)
+void b3PgsJacobiSolver::convertContact(b3RigidBodyData* bodies, b3InertiaData* inertias, b3Contact4* manifold, const b3ContactSolverInfo& infoGlobal)
 {
-	b3RigidBodyData* colObj0=0,*colObj1=0;
+	b3RigidBodyData *colObj0 = 0, *colObj1 = 0;
 
-	
-	int solverBodyIdA = getOrInitSolverBody(manifold->getBodyA(),bodies,inertias);
-	int solverBodyIdB = getOrInitSolverBody(manifold->getBodyB(),bodies,inertias);
+	int solverBodyIdA = getOrInitSolverBody(manifold->getBodyA(), bodies, inertias);
+	int solverBodyIdB = getOrInitSolverBody(manifold->getBodyB(), bodies, inertias);
 
-//	b3RigidBody* bodyA = b3RigidBody::upcast(colObj0);
-//	b3RigidBody* bodyB = b3RigidBody::upcast(colObj1);
+	//	b3RigidBody* bodyA = b3RigidBody::upcast(colObj0);
+	//	b3RigidBody* bodyB = b3RigidBody::upcast(colObj1);
 
 	b3SolverBody* solverBodyA = &m_tmpSolverBodyPool[solverBodyIdA];
 	b3SolverBody* solverBodyB = &m_tmpSolverBodyPool[solverBodyIdB];
 
-
-
 	///avoid collision response between two static objects
 	if (solverBodyA->m_invMass.isZero() && solverBodyB->m_invMass.isZero())
 		return;
 
-	int rollingFriction=1;
+	int rollingFriction = 1;
 	int numContacts = getNumContacts(manifold);
-	for (int j=0;j<numContacts;j++)
+	for (int j = 0; j < numContacts; j++)
 	{
-
 		b3ContactPoint cp;
-		getContactPoint(manifold,j,cp);
+		getContactPoint(manifold, j, cp);
 
 		if (cp.getDistance() <= getContactProcessingThreshold(manifold))
 		{
@@ -953,61 +878,60 @@ void	b3PgsJacobiSolver::convertContact(b3RigidBodyData* bodies, b3InertiaData* i
 
 			int frictionIndex = m_tmpSolverContactConstraintPool.size();
 			b3SolverConstraint& solverConstraint = m_tmpSolverContactConstraintPool.expandNonInitializing();
-//			b3RigidBody* rb0 = b3RigidBody::upcast(colObj0);
-//			b3RigidBody* rb1 = b3RigidBody::upcast(colObj1);
+			//			b3RigidBody* rb0 = b3RigidBody::upcast(colObj0);
+			//			b3RigidBody* rb1 = b3RigidBody::upcast(colObj1);
 			solverConstraint.m_solverBodyIdA = solverBodyIdA;
 			solverConstraint.m_solverBodyIdB = solverBodyIdB;
 
 			solverConstraint.m_originalContactPoint = &cp;
 
-			setupContactConstraint(bodies,inertias,solverConstraint, solverBodyIdA, solverBodyIdB, cp, infoGlobal, vel, rel_vel, relaxation, rel_pos1, rel_pos2);
+			setupContactConstraint(bodies, inertias, solverConstraint, solverBodyIdA, solverBodyIdB, cp, infoGlobal, vel, rel_vel, relaxation, rel_pos1, rel_pos2);
 
-//			const b3Vector3& pos1 = cp.getPositionWorldOnA();
-//			const b3Vector3& pos2 = cp.getPositionWorldOnB();
+			//			const b3Vector3& pos1 = cp.getPositionWorldOnA();
+			//			const b3Vector3& pos2 = cp.getPositionWorldOnB();
 
 			/////setup the friction constraints
 
 			solverConstraint.m_frictionIndex = m_tmpSolverContactFrictionConstraintPool.size();
 
-			b3Vector3 angVelA,angVelB;
+			b3Vector3 angVelA, angVelB;
 			solverBodyA->getAngularVelocity(angVelA);
-			solverBodyB->getAngularVelocity(angVelB);			
-			b3Vector3 relAngVel = angVelB-angVelA;
+			solverBodyB->getAngularVelocity(angVelB);
+			b3Vector3 relAngVel = angVelB - angVelA;
 
-			if ((cp.m_combinedRollingFriction>0.f) && (rollingFriction>0))
+			if ((cp.m_combinedRollingFriction > 0.f) && (rollingFriction > 0))
 			{
 				//only a single rollingFriction per manifold
 				rollingFriction--;
-				if (relAngVel.length()>infoGlobal.m_singleAxisRollingFrictionThreshold)
+				if (relAngVel.length() > infoGlobal.m_singleAxisRollingFrictionThreshold)
 				{
 					relAngVel.normalize();
-					if (relAngVel.length()>0.001)
-						addRollingFrictionConstraint(bodies,inertias,relAngVel,solverBodyIdA,solverBodyIdB,frictionIndex,cp,rel_pos1,rel_pos2,colObj0,colObj1, relaxation);
-
-				} else
+					if (relAngVel.length() > 0.001)
+						addRollingFrictionConstraint(bodies, inertias, relAngVel, solverBodyIdA, solverBodyIdB, frictionIndex, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation);
+				}
+				else
 				{
-					addRollingFrictionConstraint(bodies,inertias,cp.m_normalWorldOnB,solverBodyIdA,solverBodyIdB,frictionIndex,cp,rel_pos1,rel_pos2,colObj0,colObj1, relaxation);
-					b3Vector3 axis0,axis1;
-					b3PlaneSpace1(cp.m_normalWorldOnB,axis0,axis1);
-					if (axis0.length()>0.001)
-						addRollingFrictionConstraint(bodies,inertias,axis0,solverBodyIdA,solverBodyIdB,frictionIndex,cp,rel_pos1,rel_pos2,colObj0,colObj1, relaxation);
-					if (axis1.length()>0.001)
-						addRollingFrictionConstraint(bodies,inertias,axis1,solverBodyIdA,solverBodyIdB,frictionIndex,cp,rel_pos1,rel_pos2,colObj0,colObj1, relaxation);
-		
+					addRollingFrictionConstraint(bodies, inertias, cp.m_normalWorldOnB, solverBodyIdA, solverBodyIdB, frictionIndex, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation);
+					b3Vector3 axis0, axis1;
+					b3PlaneSpace1(cp.m_normalWorldOnB, axis0, axis1);
+					if (axis0.length() > 0.001)
+						addRollingFrictionConstraint(bodies, inertias, axis0, solverBodyIdA, solverBodyIdB, frictionIndex, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation);
+					if (axis1.length() > 0.001)
+						addRollingFrictionConstraint(bodies, inertias, axis1, solverBodyIdA, solverBodyIdB, frictionIndex, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation);
 				}
 			}
 
 			///Bullet has several options to set the friction directions
-			///By default, each contact has only a single friction direction that is recomputed automatically very frame 
+			///By default, each contact has only a single friction direction that is recomputed automatically very frame
 			///based on the relative linear velocity.
 			///If the relative velocity it zero, it will automatically compute a friction direction.
-			
+
 			///You can also enable two friction directions, using the B3_SOLVER_USE_2_FRICTION_DIRECTIONS.
 			///In that case, the second friction direction will be orthogonal to both contact normal and first friction direction.
 			///
 			///If you choose B3_SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION, then the friction will be independent from the relative projected velocity.
 			///
-			///The user can manually override the friction directions for certain contacts using a contact callback, 
+			///The user can manually override the friction directions for certain contacts using a contact callback,
 			///and set the cp.m_lateralFrictionInitialized to true
 			///In that case, you can set the target relative motion in each friction direction (cp.m_contactMotion1 and cp.m_contactMotion2)
 			///this will give a conveyor belt effect
@@ -1018,99 +942,91 @@ void	b3PgsJacobiSolver::convertContact(b3RigidBodyData* bodies, b3InertiaData* i
 				b3Scalar lat_rel_vel = cp.m_lateralFrictionDir1.length2();
 				if (!(infoGlobal.m_solverMode & B3_SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION) && lat_rel_vel > B3_EPSILON)
 				{
-					cp.m_lateralFrictionDir1 *= 1.f/b3Sqrt(lat_rel_vel);
-					if((infoGlobal.m_solverMode & B3_SOLVER_USE_2_FRICTION_DIRECTIONS))
+					cp.m_lateralFrictionDir1 *= 1.f / b3Sqrt(lat_rel_vel);
+					if ((infoGlobal.m_solverMode & B3_SOLVER_USE_2_FRICTION_DIRECTIONS))
 					{
 						cp.m_lateralFrictionDir2 = cp.m_lateralFrictionDir1.cross(cp.m_normalWorldOnB);
-						cp.m_lateralFrictionDir2.normalize();//??
-						addFrictionConstraint(bodies,inertias,cp.m_lateralFrictionDir2,solverBodyIdA,solverBodyIdB,frictionIndex,cp,rel_pos1,rel_pos2,colObj0,colObj1, relaxation);
-
+						cp.m_lateralFrictionDir2.normalize();  //??
+						addFrictionConstraint(bodies, inertias, cp.m_lateralFrictionDir2, solverBodyIdA, solverBodyIdB, frictionIndex, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation);
 					}
 
-					addFrictionConstraint(bodies,inertias,cp.m_lateralFrictionDir1,solverBodyIdA,solverBodyIdB,frictionIndex,cp,rel_pos1,rel_pos2,colObj0,colObj1, relaxation);
-
-				} else
+					addFrictionConstraint(bodies, inertias, cp.m_lateralFrictionDir1, solverBodyIdA, solverBodyIdB, frictionIndex, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation);
+				}
+				else
 				{
-					b3PlaneSpace1(cp.m_normalWorldOnB,cp.m_lateralFrictionDir1,cp.m_lateralFrictionDir2);
+					b3PlaneSpace1(cp.m_normalWorldOnB, cp.m_lateralFrictionDir1, cp.m_lateralFrictionDir2);
 
 					if ((infoGlobal.m_solverMode & B3_SOLVER_USE_2_FRICTION_DIRECTIONS))
 					{
-						addFrictionConstraint(bodies,inertias,cp.m_lateralFrictionDir2,solverBodyIdA,solverBodyIdB,frictionIndex,cp,rel_pos1,rel_pos2,colObj0,colObj1, relaxation);
+						addFrictionConstraint(bodies, inertias, cp.m_lateralFrictionDir2, solverBodyIdA, solverBodyIdB, frictionIndex, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation);
 					}
 
-					addFrictionConstraint(bodies,inertias,cp.m_lateralFrictionDir1,solverBodyIdA,solverBodyIdB,frictionIndex,cp,rel_pos1,rel_pos2,colObj0,colObj1, relaxation);
+					addFrictionConstraint(bodies, inertias, cp.m_lateralFrictionDir1, solverBodyIdA, solverBodyIdB, frictionIndex, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation);
 
 					if ((infoGlobal.m_solverMode & B3_SOLVER_USE_2_FRICTION_DIRECTIONS) && (infoGlobal.m_solverMode & B3_SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION))
 					{
 						cp.m_lateralFrictionInitialized = true;
 					}
 				}
-
-			} else
+			}
+			else
 			{
-				addFrictionConstraint(bodies,inertias,cp.m_lateralFrictionDir1,solverBodyIdA,solverBodyIdB,frictionIndex,cp,rel_pos1,rel_pos2,colObj0,colObj1, relaxation,cp.m_contactMotion1, cp.m_contactCFM1);
+				addFrictionConstraint(bodies, inertias, cp.m_lateralFrictionDir1, solverBodyIdA, solverBodyIdB, frictionIndex, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, cp.m_contactMotion1, cp.m_contactCFM1);
 
 				if ((infoGlobal.m_solverMode & B3_SOLVER_USE_2_FRICTION_DIRECTIONS))
-					addFrictionConstraint(bodies,inertias,cp.m_lateralFrictionDir2,solverBodyIdA,solverBodyIdB,frictionIndex,cp,rel_pos1,rel_pos2,colObj0,colObj1, relaxation, cp.m_contactMotion2, cp.m_contactCFM2);
+					addFrictionConstraint(bodies, inertias, cp.m_lateralFrictionDir2, solverBodyIdA, solverBodyIdB, frictionIndex, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, cp.m_contactMotion2, cp.m_contactCFM2);
 
-				setFrictionConstraintImpulse( bodies,inertias,solverConstraint, solverBodyIdA, solverBodyIdB, cp, infoGlobal);
+				setFrictionConstraintImpulse(bodies, inertias, solverConstraint, solverBodyIdA, solverBodyIdB, cp, infoGlobal);
 			}
-		
-
-			
-
 		}
 	}
 }
 
-b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlySetup(b3RigidBodyData* bodies, b3InertiaData* inertias, int numBodies, b3Contact4* manifoldPtr, int numManifolds,b3TypedConstraint** constraints,int numConstraints,const b3ContactSolverInfo& infoGlobal)
+b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlySetup(b3RigidBodyData* bodies, b3InertiaData* inertias, int numBodies, b3Contact4* manifoldPtr, int numManifolds, b3TypedConstraint** constraints, int numConstraints, const b3ContactSolverInfo& infoGlobal)
 {
 	B3_PROFILE("solveGroupCacheFriendlySetup");
 
-
 	m_maxOverrideNumSolverIterations = 0;
 
-
-
 	m_tmpSolverBodyPool.resize(0);
-	
-	
+
 	m_bodyCount.resize(0);
-	m_bodyCount.resize(numBodies,0);
+	m_bodyCount.resize(numBodies, 0);
 	m_bodyCountCheck.resize(0);
-	m_bodyCountCheck.resize(numBodies,0);
-	
+	m_bodyCountCheck.resize(numBodies, 0);
+
 	m_deltaLinearVelocities.resize(0);
-	m_deltaLinearVelocities.resize(numBodies,b3MakeVector3(0,0,0));
+	m_deltaLinearVelocities.resize(numBodies, b3MakeVector3(0, 0, 0));
 	m_deltaAngularVelocities.resize(0);
-	m_deltaAngularVelocities.resize(numBodies,b3MakeVector3(0,0,0));
-	
+	m_deltaAngularVelocities.resize(numBodies, b3MakeVector3(0, 0, 0));
+
 	//int totalBodies = 0;
 
-	for (int i=0;i<numConstraints;i++)
+	for (int i = 0; i < numConstraints; i++)
 	{
 		int bodyIndexA = constraints[i]->getRigidBodyA();
 		int bodyIndexB = constraints[i]->getRigidBodyB();
 		if (m_usePgs)
 		{
-			m_bodyCount[bodyIndexA]=-1;
-			m_bodyCount[bodyIndexB]=-1;
-		} else
+			m_bodyCount[bodyIndexA] = -1;
+			m_bodyCount[bodyIndexB] = -1;
+		}
+		else
 		{
 			//didn't implement joints with Jacobi version yet
 			b3Assert(0);
 		}
-
 	}
-	for (int i=0;i<numManifolds;i++)
+	for (int i = 0; i < numManifolds; i++)
 	{
 		int bodyIndexA = manifoldPtr[i].getBodyA();
 		int bodyIndexB = manifoldPtr[i].getBodyB();
 		if (m_usePgs)
 		{
-			m_bodyCount[bodyIndexA]=-1;
-			m_bodyCount[bodyIndexB]=-1;
-		} else
+			m_bodyCount[bodyIndexA] = -1;
+			m_bodyCount[bodyIndexB] = -1;
+		}
+		else
 		{
 			if (bodies[bodyIndexA].m_invMass)
 			{
@@ -1118,26 +1034,23 @@ b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlySetup(b3RigidBodyData* bodies
 				m_bodyCount[bodyIndexA]++;
 			}
 			else
-				m_bodyCount[bodyIndexA]=-1;
+				m_bodyCount[bodyIndexA] = -1;
 
 			if (bodies[bodyIndexB].m_invMass)
-			//	m_bodyCount[bodyIndexB]+=manifoldPtr[i].getNPoints();
+				//	m_bodyCount[bodyIndexB]+=manifoldPtr[i].getNPoints();
 				m_bodyCount[bodyIndexB]++;
 			else
-				m_bodyCount[bodyIndexB]=-1;
+				m_bodyCount[bodyIndexB] = -1;
 		}
-
 	}
 
-
-	
 	if (1)
 	{
 		int j;
-		for (j=0;j<numConstraints;j++)
+		for (j = 0; j < numConstraints; j++)
 		{
 			b3TypedConstraint* constraint = constraints[j];
-			
+
 			constraint->internalSetAppliedImpulse(0.0f);
 		}
 	}
@@ -1146,13 +1059,12 @@ b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlySetup(b3RigidBodyData* bodies
 	//if (1)
 	{
 		{
-
 			int totalNumRows = 0;
 			int i;
-			
+
 			m_tmpConstraintSizesPool.resizeNoInitialize(numConstraints);
 			//calculate the total number of contraint rows
-			for (i=0;i<numConstraints;i++)
+			for (i = 0; i < numConstraints; i++)
 			{
 				b3TypedConstraint::b3ConstraintInfo1& info1 = m_tmpConstraintSizesPool[i];
 				b3JointFeedback* fb = constraints[i]->getJointFeedback();
@@ -1169,8 +1081,9 @@ b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlySetup(b3RigidBodyData* bodies
 				}
 				if (constraints[i]->isEnabled())
 				{
-					constraints[i]->getInfo1(&info1,bodies);
-				} else
+					constraints[i]->getInfo1(&info1, bodies);
+				}
+				else
 				{
 					info1.m_numConstraintRows = 0;
 					info1.nub = 0;
@@ -1179,45 +1092,40 @@ b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlySetup(b3RigidBodyData* bodies
 			}
 			m_tmpSolverNonContactConstraintPool.resizeNoInitialize(totalNumRows);
 
-			
 #ifndef DISABLE_JOINTS
 			///setup the b3SolverConstraints
 			int currentRow = 0;
 
-			for (i=0;i<numConstraints;i++)
+			for (i = 0; i < numConstraints; i++)
 			{
 				const b3TypedConstraint::b3ConstraintInfo1& info1 = m_tmpConstraintSizesPool[i];
-				
+
 				if (info1.m_numConstraintRows)
 				{
-					b3Assert(currentRow<totalNumRows);
+					b3Assert(currentRow < totalNumRows);
 
 					b3SolverConstraint* currentConstraintRow = &m_tmpSolverNonContactConstraintPool[currentRow];
 					b3TypedConstraint* constraint = constraints[i];
 
-					b3RigidBodyData& rbA = bodies[ constraint->getRigidBodyA()];
+					b3RigidBodyData& rbA = bodies[constraint->getRigidBodyA()];
 					//b3RigidBody& rbA = constraint->getRigidBodyA();
-	//				b3RigidBody& rbB = constraint->getRigidBodyB();
-					b3RigidBodyData& rbB = bodies[ constraint->getRigidBodyB()];
-
-                    int solverBodyIdA = getOrInitSolverBody(constraint->getRigidBodyA(),bodies,inertias);
-                    int solverBodyIdB = getOrInitSolverBody(constraint->getRigidBodyB(),bodies,inertias);
-
-                    b3SolverBody* bodyAPtr = &m_tmpSolverBodyPool[solverBodyIdA];
-                    b3SolverBody* bodyBPtr = &m_tmpSolverBodyPool[solverBodyIdB];
-
+					//				b3RigidBody& rbB = constraint->getRigidBodyB();
+					b3RigidBodyData& rbB = bodies[constraint->getRigidBodyB()];
 
+					int solverBodyIdA = getOrInitSolverBody(constraint->getRigidBodyA(), bodies, inertias);
+					int solverBodyIdB = getOrInitSolverBody(constraint->getRigidBodyB(), bodies, inertias);
 
+					b3SolverBody* bodyAPtr = &m_tmpSolverBodyPool[solverBodyIdA];
+					b3SolverBody* bodyBPtr = &m_tmpSolverBodyPool[solverBodyIdB];
 
 					int overrideNumSolverIterations = constraint->getOverrideNumSolverIterations() > 0 ? constraint->getOverrideNumSolverIterations() : infoGlobal.m_numIterations;
-					if (overrideNumSolverIterations>m_maxOverrideNumSolverIterations)
+					if (overrideNumSolverIterations > m_maxOverrideNumSolverIterations)
 						m_maxOverrideNumSolverIterations = overrideNumSolverIterations;
 
-
 					int j;
-					for ( j=0;j<info1.m_numConstraintRows;j++)
+					for (j = 0; j < info1.m_numConstraintRows; j++)
 					{
-						memset(&currentConstraintRow[j],0,sizeof(b3SolverConstraint));
+						memset(&currentConstraintRow[j], 0, sizeof(b3SolverConstraint));
 						currentConstraintRow[j].m_lowerLimit = -B3_INFINITY;
 						currentConstraintRow[j].m_upperLimit = B3_INFINITY;
 						currentConstraintRow[j].m_appliedImpulse = 0.f;
@@ -1227,26 +1135,25 @@ b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlySetup(b3RigidBodyData* bodies
 						currentConstraintRow[j].m_overrideNumSolverIterations = overrideNumSolverIterations;
 					}
 
-					bodyAPtr->internalGetDeltaLinearVelocity().setValue(0.f,0.f,0.f);
-					bodyAPtr->internalGetDeltaAngularVelocity().setValue(0.f,0.f,0.f);
-					bodyAPtr->internalGetPushVelocity().setValue(0.f,0.f,0.f);
-					bodyAPtr->internalGetTurnVelocity().setValue(0.f,0.f,0.f);
-					bodyBPtr->internalGetDeltaLinearVelocity().setValue(0.f,0.f,0.f);
-					bodyBPtr->internalGetDeltaAngularVelocity().setValue(0.f,0.f,0.f);
-					bodyBPtr->internalGetPushVelocity().setValue(0.f,0.f,0.f);
-					bodyBPtr->internalGetTurnVelocity().setValue(0.f,0.f,0.f);
-
+					bodyAPtr->internalGetDeltaLinearVelocity().setValue(0.f, 0.f, 0.f);
+					bodyAPtr->internalGetDeltaAngularVelocity().setValue(0.f, 0.f, 0.f);
+					bodyAPtr->internalGetPushVelocity().setValue(0.f, 0.f, 0.f);
+					bodyAPtr->internalGetTurnVelocity().setValue(0.f, 0.f, 0.f);
+					bodyBPtr->internalGetDeltaLinearVelocity().setValue(0.f, 0.f, 0.f);
+					bodyBPtr->internalGetDeltaAngularVelocity().setValue(0.f, 0.f, 0.f);
+					bodyBPtr->internalGetPushVelocity().setValue(0.f, 0.f, 0.f);
+					bodyBPtr->internalGetTurnVelocity().setValue(0.f, 0.f, 0.f);
 
 					b3TypedConstraint::b3ConstraintInfo2 info2;
-					info2.fps = 1.f/infoGlobal.m_timeStep;
+					info2.fps = 1.f / infoGlobal.m_timeStep;
 					info2.erp = infoGlobal.m_erp;
 					info2.m_J1linearAxis = currentConstraintRow->m_contactNormal;
 					info2.m_J1angularAxis = currentConstraintRow->m_relpos1CrossNormal;
 					info2.m_J2linearAxis = 0;
 					info2.m_J2angularAxis = currentConstraintRow->m_relpos2CrossNormal;
-					info2.rowskip = sizeof(b3SolverConstraint)/sizeof(b3Scalar);//check this
-					///the size of b3SolverConstraint needs be a multiple of b3Scalar
-		            b3Assert(info2.rowskip*sizeof(b3Scalar)== sizeof(b3SolverConstraint));
+					info2.rowskip = sizeof(b3SolverConstraint) / sizeof(b3Scalar);  //check this
+																					///the size of b3SolverConstraint needs be a multiple of b3Scalar
+					b3Assert(info2.rowskip * sizeof(b3Scalar) == sizeof(b3SolverConstraint));
 					info2.m_constraintError = &currentConstraintRow->m_rhs;
 					currentConstraintRow->m_cfm = infoGlobal.m_globalCfm;
 					info2.m_damping = infoGlobal.m_damping;
@@ -1254,47 +1161,45 @@ b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlySetup(b3RigidBodyData* bodies
 					info2.m_lowerLimit = &currentConstraintRow->m_lowerLimit;
 					info2.m_upperLimit = &currentConstraintRow->m_upperLimit;
 					info2.m_numIterations = infoGlobal.m_numIterations;
-					constraints[i]->getInfo2(&info2,bodies);
+					constraints[i]->getInfo2(&info2, bodies);
 
 					///finalize the constraint setup
-					for ( j=0;j<info1.m_numConstraintRows;j++)
+					for (j = 0; j < info1.m_numConstraintRows; j++)
 					{
 						b3SolverConstraint& solverConstraint = currentConstraintRow[j];
 
-						if (solverConstraint.m_upperLimit>=constraints[i]->getBreakingImpulseThreshold())
+						if (solverConstraint.m_upperLimit >= constraints[i]->getBreakingImpulseThreshold())
 						{
 							solverConstraint.m_upperLimit = constraints[i]->getBreakingImpulseThreshold();
 						}
 
-						if (solverConstraint.m_lowerLimit<=-constraints[i]->getBreakingImpulseThreshold())
+						if (solverConstraint.m_lowerLimit <= -constraints[i]->getBreakingImpulseThreshold())
 						{
 							solverConstraint.m_lowerLimit = -constraints[i]->getBreakingImpulseThreshold();
 						}
 
 						solverConstraint.m_originalContactPoint = constraint;
-							
-						b3Matrix3x3& invInertiaWorldA= inertias[constraint->getRigidBodyA()].m_invInertiaWorld;
-						{
 
+						b3Matrix3x3& invInertiaWorldA = inertias[constraint->getRigidBodyA()].m_invInertiaWorld;
+						{
 							//b3Vector3 angularFactorA(1,1,1);
 							const b3Vector3& ftorqueAxis1 = solverConstraint.m_relpos1CrossNormal;
-							solverConstraint.m_angularComponentA = invInertiaWorldA*ftorqueAxis1;//*angularFactorA;
+							solverConstraint.m_angularComponentA = invInertiaWorldA * ftorqueAxis1;  //*angularFactorA;
 						}
-						
-						b3Matrix3x3& invInertiaWorldB= inertias[constraint->getRigidBodyB()].m_invInertiaWorld;
-						{
 
+						b3Matrix3x3& invInertiaWorldB = inertias[constraint->getRigidBodyB()].m_invInertiaWorld;
+						{
 							const b3Vector3& ftorqueAxis2 = solverConstraint.m_relpos2CrossNormal;
-							solverConstraint.m_angularComponentB = invInertiaWorldB*ftorqueAxis2;//*constraint->getRigidBodyB().getAngularFactor();
+							solverConstraint.m_angularComponentB = invInertiaWorldB * ftorqueAxis2;  //*constraint->getRigidBodyB().getAngularFactor();
 						}
 
 						{
 							//it is ok to use solverConstraint.m_contactNormal instead of -solverConstraint.m_contactNormal
 							//because it gets multiplied iMJlB
-							b3Vector3 iMJlA = solverConstraint.m_contactNormal*rbA.m_invMass;
-							b3Vector3 iMJaA = invInertiaWorldA*solverConstraint.m_relpos1CrossNormal;
-							b3Vector3 iMJlB = solverConstraint.m_contactNormal*rbB.m_invMass;//sign of normal?
-							b3Vector3 iMJaB = invInertiaWorldB*solverConstraint.m_relpos2CrossNormal;
+							b3Vector3 iMJlA = solverConstraint.m_contactNormal * rbA.m_invMass;
+							b3Vector3 iMJaA = invInertiaWorldA * solverConstraint.m_relpos1CrossNormal;
+							b3Vector3 iMJlB = solverConstraint.m_contactNormal * rbB.m_invMass;  //sign of normal?
+							b3Vector3 iMJaB = invInertiaWorldB * solverConstraint.m_relpos2CrossNormal;
 
 							b3Scalar sum = iMJlA.dot(solverConstraint.m_contactNormal);
 							sum += iMJaA.dot(solverConstraint.m_relpos1CrossNormal);
@@ -1302,10 +1207,9 @@ b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlySetup(b3RigidBodyData* bodies
 							sum += iMJaB.dot(solverConstraint.m_relpos2CrossNormal);
 							b3Scalar fsum = b3Fabs(sum);
 							b3Assert(fsum > B3_EPSILON);
-							solverConstraint.m_jacDiagABInv = fsum>B3_EPSILON?b3Scalar(1.)/sum : 0.f;
+							solverConstraint.m_jacDiagABInv = fsum > B3_EPSILON ? b3Scalar(1.) / sum : 0.f;
 						}
 
-
 						///fix rhs
 						///todo: add force/torque accelerators
 						{
@@ -1313,38 +1217,35 @@ b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlySetup(b3RigidBodyData* bodies
 							b3Scalar vel1Dotn = solverConstraint.m_contactNormal.dot(rbA.m_linVel) + solverConstraint.m_relpos1CrossNormal.dot(rbA.m_angVel);
 							b3Scalar vel2Dotn = -solverConstraint.m_contactNormal.dot(rbB.m_linVel) + solverConstraint.m_relpos2CrossNormal.dot(rbB.m_angVel);
 
-							rel_vel = vel1Dotn+vel2Dotn;
+							rel_vel = vel1Dotn + vel2Dotn;
 
 							b3Scalar restitution = 0.f;
-							b3Scalar positionalError = solverConstraint.m_rhs;//already filled in by getConstraintInfo2
-							b3Scalar	velocityError = restitution - rel_vel * info2.m_damping;
-							b3Scalar	penetrationImpulse = positionalError*solverConstraint.m_jacDiagABInv;
-							b3Scalar	velocityImpulse = velocityError *solverConstraint.m_jacDiagABInv;
-							solverConstraint.m_rhs = penetrationImpulse+velocityImpulse;
+							b3Scalar positionalError = solverConstraint.m_rhs;  //already filled in by getConstraintInfo2
+							b3Scalar velocityError = restitution - rel_vel * info2.m_damping;
+							b3Scalar penetrationImpulse = positionalError * solverConstraint.m_jacDiagABInv;
+							b3Scalar velocityImpulse = velocityError * solverConstraint.m_jacDiagABInv;
+							solverConstraint.m_rhs = penetrationImpulse + velocityImpulse;
 							solverConstraint.m_appliedImpulse = 0.f;
-
 						}
 					}
 				}
-				currentRow+=m_tmpConstraintSizesPool[i].m_numConstraintRows;
+				currentRow += m_tmpConstraintSizesPool[i].m_numConstraintRows;
 			}
-#endif //DISABLE_JOINTS
+#endif  //DISABLE_JOINTS
 		}
 
-
 		{
 			int i;
 
-			for (i=0;i<numManifolds;i++)
+			for (i = 0; i < numManifolds; i++)
 			{
 				b3Contact4& manifold = manifoldPtr[i];
-				convertContact(bodies,inertias,&manifold,infoGlobal);
+				convertContact(bodies, inertias, &manifold, infoGlobal);
 			}
 		}
 	}
 
-//	b3ContactSolverInfo info = infoGlobal;
-
+	//	b3ContactSolverInfo info = infoGlobal;
 
 	int numNonContactPool = m_tmpSolverNonContactConstraintPool.size();
 	int numConstraintPool = m_tmpSolverContactConstraintPool.size();
@@ -1353,64 +1254,63 @@ b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlySetup(b3RigidBodyData* bodies
 	///@todo: use stack allocator for such temporarily memory, same for solver bodies/constraints
 	m_orderNonContactConstraintPool.resizeNoInitialize(numNonContactPool);
 	if ((infoGlobal.m_solverMode & B3_SOLVER_USE_2_FRICTION_DIRECTIONS))
-		m_orderTmpConstraintPool.resizeNoInitialize(numConstraintPool*2);
+		m_orderTmpConstraintPool.resizeNoInitialize(numConstraintPool * 2);
 	else
 		m_orderTmpConstraintPool.resizeNoInitialize(numConstraintPool);
 
 	m_orderFrictionConstraintPool.resizeNoInitialize(numFrictionPool);
 	{
 		int i;
-		for (i=0;i<numNonContactPool;i++)
+		for (i = 0; i < numNonContactPool; i++)
 		{
 			m_orderNonContactConstraintPool[i] = i;
 		}
-		for (i=0;i<numConstraintPool;i++)
+		for (i = 0; i < numConstraintPool; i++)
 		{
 			m_orderTmpConstraintPool[i] = i;
 		}
-		for (i=0;i<numFrictionPool;i++)
+		for (i = 0; i < numFrictionPool; i++)
 		{
 			m_orderFrictionConstraintPool[i] = i;
 		}
 	}
 
 	return 0.f;
-
 }
 
-
-b3Scalar b3PgsJacobiSolver::solveSingleIteration(int iteration,b3TypedConstraint** constraints,int numConstraints,const b3ContactSolverInfo& infoGlobal)
+b3Scalar b3PgsJacobiSolver::solveSingleIteration(int iteration, b3TypedConstraint** constraints, int numConstraints, const b3ContactSolverInfo& infoGlobal)
 {
-
 	int numNonContactPool = m_tmpSolverNonContactConstraintPool.size();
 	int numConstraintPool = m_tmpSolverContactConstraintPool.size();
 	int numFrictionPool = m_tmpSolverContactFrictionConstraintPool.size();
-	
+
 	if (infoGlobal.m_solverMode & B3_SOLVER_RANDMIZE_ORDER)
 	{
-		if (1)			// uncomment this for a bit less random ((iteration & 7) == 0)
+		if (1)  // uncomment this for a bit less random ((iteration & 7) == 0)
 		{
-
-			for (int j=0; j<numNonContactPool; ++j) {
+			for (int j = 0; j < numNonContactPool; ++j)
+			{
 				int tmp = m_orderNonContactConstraintPool[j];
-				int swapi = b3RandInt2(j+1);
+				int swapi = b3RandInt2(j + 1);
 				m_orderNonContactConstraintPool[j] = m_orderNonContactConstraintPool[swapi];
 				m_orderNonContactConstraintPool[swapi] = tmp;
 			}
 
-			//contact/friction constraints are not solved more than 
-			if (iteration< infoGlobal.m_numIterations)
+			//contact/friction constraints are not solved more than
+			if (iteration < infoGlobal.m_numIterations)
 			{
-				for (int j=0; j<numConstraintPool; ++j) {
+				for (int j = 0; j < numConstraintPool; ++j)
+				{
 					int tmp = m_orderTmpConstraintPool[j];
-					int swapi = b3RandInt2(j+1);
+					int swapi = b3RandInt2(j + 1);
 					m_orderTmpConstraintPool[j] = m_orderTmpConstraintPool[swapi];
 					m_orderTmpConstraintPool[swapi] = tmp;
 				}
 
-				for (int j=0; j<numFrictionPool; ++j) {
+				for (int j = 0; j < numFrictionPool; ++j)
+				{
 					int tmp = m_orderFrictionConstraintPool[j];
-					int swapi = b3RandInt2(j+1);
+					int swapi = b3RandInt2(j + 1);
 					m_orderFrictionConstraintPool[j] = m_orderFrictionConstraintPool[swapi];
 					m_orderFrictionConstraintPool[swapi] = tmp;
 				}
@@ -1421,173 +1321,163 @@ b3Scalar b3PgsJacobiSolver::solveSingleIteration(int iteration,b3TypedConstraint
 	if (infoGlobal.m_solverMode & B3_SOLVER_SIMD)
 	{
 		///solve all joint constraints, using SIMD, if available
-		for (int j=0;j<m_tmpSolverNonContactConstraintPool.size();j++)
+		for (int j = 0; j < m_tmpSolverNonContactConstraintPool.size(); j++)
 		{
 			b3SolverConstraint& constraint = m_tmpSolverNonContactConstraintPool[m_orderNonContactConstraintPool[j]];
 			if (iteration < constraint.m_overrideNumSolverIterations)
-				resolveSingleConstraintRowGenericSIMD(m_tmpSolverBodyPool[constraint.m_solverBodyIdA],m_tmpSolverBodyPool[constraint.m_solverBodyIdB],constraint);
+				resolveSingleConstraintRowGenericSIMD(m_tmpSolverBodyPool[constraint.m_solverBodyIdA], m_tmpSolverBodyPool[constraint.m_solverBodyIdB], constraint);
 		}
 
-		if (iteration< infoGlobal.m_numIterations)
+		if (iteration < infoGlobal.m_numIterations)
 		{
-
 			///solve all contact constraints using SIMD, if available
 			if (infoGlobal.m_solverMode & B3_SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS)
 			{
 				int numPoolConstraints = m_tmpSolverContactConstraintPool.size();
-				int multiplier = (infoGlobal.m_solverMode & B3_SOLVER_USE_2_FRICTION_DIRECTIONS)? 2 : 1;
+				int multiplier = (infoGlobal.m_solverMode & B3_SOLVER_USE_2_FRICTION_DIRECTIONS) ? 2 : 1;
 
-				for (int c=0;c<numPoolConstraints;c++)
+				for (int c = 0; c < numPoolConstraints; c++)
 				{
-					b3Scalar totalImpulse =0;
+					b3Scalar totalImpulse = 0;
 
 					{
 						const b3SolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[m_orderTmpConstraintPool[c]];
-						resolveSingleConstraintRowLowerLimitSIMD(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA],m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB],solveManifold);
+						resolveSingleConstraintRowLowerLimitSIMD(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
 						totalImpulse = solveManifold.m_appliedImpulse;
 					}
 					bool applyFriction = true;
 					if (applyFriction)
 					{
 						{
+							b3SolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[m_orderFrictionConstraintPool[c * multiplier]];
 
-							b3SolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[m_orderFrictionConstraintPool[c*multiplier]];
-
-							if (totalImpulse>b3Scalar(0))
+							if (totalImpulse > b3Scalar(0))
 							{
-								solveManifold.m_lowerLimit = -(solveManifold.m_friction*totalImpulse);
-								solveManifold.m_upperLimit = solveManifold.m_friction*totalImpulse;
+								solveManifold.m_lowerLimit = -(solveManifold.m_friction * totalImpulse);
+								solveManifold.m_upperLimit = solveManifold.m_friction * totalImpulse;
 
-								resolveSingleConstraintRowGenericSIMD(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA],m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB],solveManifold);
+								resolveSingleConstraintRowGenericSIMD(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
 							}
 						}
 
 						if (infoGlobal.m_solverMode & B3_SOLVER_USE_2_FRICTION_DIRECTIONS)
 						{
+							b3SolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[m_orderFrictionConstraintPool[c * multiplier + 1]];
 
-							b3SolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[m_orderFrictionConstraintPool[c*multiplier+1]];
-				
-							if (totalImpulse>b3Scalar(0))
+							if (totalImpulse > b3Scalar(0))
 							{
-								solveManifold.m_lowerLimit = -(solveManifold.m_friction*totalImpulse);
-								solveManifold.m_upperLimit = solveManifold.m_friction*totalImpulse;
+								solveManifold.m_lowerLimit = -(solveManifold.m_friction * totalImpulse);
+								solveManifold.m_upperLimit = solveManifold.m_friction * totalImpulse;
 
-								resolveSingleConstraintRowGenericSIMD(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA],m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB],solveManifold);
+								resolveSingleConstraintRowGenericSIMD(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
 							}
 						}
 					}
 				}
-
 			}
-			else//B3_SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS
+			else  //B3_SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS
 			{
 				//solve the friction constraints after all contact constraints, don't interleave them
 				int numPoolConstraints = m_tmpSolverContactConstraintPool.size();
 				int j;
 
-				for (j=0;j<numPoolConstraints;j++)
+				for (j = 0; j < numPoolConstraints; j++)
 				{
 					const b3SolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[m_orderTmpConstraintPool[j]];
-					resolveSingleConstraintRowLowerLimitSIMD(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA],m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB],solveManifold);
-
+					resolveSingleConstraintRowLowerLimitSIMD(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
 				}
 
 				if (!m_usePgs)
 					averageVelocities();
-				
 
 				///solve all friction constraints, using SIMD, if available
 
 				int numFrictionPoolConstraints = m_tmpSolverContactFrictionConstraintPool.size();
-				for (j=0;j<numFrictionPoolConstraints;j++)
+				for (j = 0; j < numFrictionPoolConstraints; j++)
 				{
 					b3SolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[m_orderFrictionConstraintPool[j]];
 					b3Scalar totalImpulse = m_tmpSolverContactConstraintPool[solveManifold.m_frictionIndex].m_appliedImpulse;
 
-					if (totalImpulse>b3Scalar(0))
+					if (totalImpulse > b3Scalar(0))
 					{
-						solveManifold.m_lowerLimit = -(solveManifold.m_friction*totalImpulse);
-						solveManifold.m_upperLimit = solveManifold.m_friction*totalImpulse;
+						solveManifold.m_lowerLimit = -(solveManifold.m_friction * totalImpulse);
+						solveManifold.m_upperLimit = solveManifold.m_friction * totalImpulse;
 
-						resolveSingleConstraintRowGenericSIMD(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA],m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB],solveManifold);
+						resolveSingleConstraintRowGenericSIMD(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
 					}
 				}
 
-				
 				int numRollingFrictionPoolConstraints = m_tmpSolverContactRollingFrictionConstraintPool.size();
-				for (j=0;j<numRollingFrictionPoolConstraints;j++)
+				for (j = 0; j < numRollingFrictionPoolConstraints; j++)
 				{
-
 					b3SolverConstraint& rollingFrictionConstraint = m_tmpSolverContactRollingFrictionConstraintPool[j];
 					b3Scalar totalImpulse = m_tmpSolverContactConstraintPool[rollingFrictionConstraint.m_frictionIndex].m_appliedImpulse;
-					if (totalImpulse>b3Scalar(0))
+					if (totalImpulse > b3Scalar(0))
 					{
-						b3Scalar rollingFrictionMagnitude = rollingFrictionConstraint.m_friction*totalImpulse;
-						if (rollingFrictionMagnitude>rollingFrictionConstraint.m_friction)
+						b3Scalar rollingFrictionMagnitude = rollingFrictionConstraint.m_friction * totalImpulse;
+						if (rollingFrictionMagnitude > rollingFrictionConstraint.m_friction)
 							rollingFrictionMagnitude = rollingFrictionConstraint.m_friction;
 
 						rollingFrictionConstraint.m_lowerLimit = -rollingFrictionMagnitude;
 						rollingFrictionConstraint.m_upperLimit = rollingFrictionMagnitude;
 
-						resolveSingleConstraintRowGenericSIMD(m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdA],m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdB],rollingFrictionConstraint);
+						resolveSingleConstraintRowGenericSIMD(m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdA], m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdB], rollingFrictionConstraint);
 					}
 				}
-				
-
-			}			
+			}
 		}
-	} else
+	}
+	else
 	{
 		//non-SIMD version
 		///solve all joint constraints
-		for (int j=0;j<m_tmpSolverNonContactConstraintPool.size();j++)
+		for (int j = 0; j < m_tmpSolverNonContactConstraintPool.size(); j++)
 		{
 			b3SolverConstraint& constraint = m_tmpSolverNonContactConstraintPool[m_orderNonContactConstraintPool[j]];
 			if (iteration < constraint.m_overrideNumSolverIterations)
-				resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[constraint.m_solverBodyIdA],m_tmpSolverBodyPool[constraint.m_solverBodyIdB],constraint);
+				resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[constraint.m_solverBodyIdA], m_tmpSolverBodyPool[constraint.m_solverBodyIdB], constraint);
 		}
 
-		if (iteration< infoGlobal.m_numIterations)
+		if (iteration < infoGlobal.m_numIterations)
 		{
-
 			///solve all contact constraints
 			int numPoolConstraints = m_tmpSolverContactConstraintPool.size();
-			for (int j=0;j<numPoolConstraints;j++)
+			for (int j = 0; j < numPoolConstraints; j++)
 			{
 				const b3SolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[m_orderTmpConstraintPool[j]];
-				resolveSingleConstraintRowLowerLimit(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA],m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB],solveManifold);
+				resolveSingleConstraintRowLowerLimit(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
 			}
 			///solve all friction constraints
 			int numFrictionPoolConstraints = m_tmpSolverContactFrictionConstraintPool.size();
-			for (int j=0;j<numFrictionPoolConstraints;j++)
+			for (int j = 0; j < numFrictionPoolConstraints; j++)
 			{
 				b3SolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[m_orderFrictionConstraintPool[j]];
 				b3Scalar totalImpulse = m_tmpSolverContactConstraintPool[solveManifold.m_frictionIndex].m_appliedImpulse;
 
-				if (totalImpulse>b3Scalar(0))
+				if (totalImpulse > b3Scalar(0))
 				{
-					solveManifold.m_lowerLimit = -(solveManifold.m_friction*totalImpulse);
-					solveManifold.m_upperLimit = solveManifold.m_friction*totalImpulse;
+					solveManifold.m_lowerLimit = -(solveManifold.m_friction * totalImpulse);
+					solveManifold.m_upperLimit = solveManifold.m_friction * totalImpulse;
 
-					resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA],m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB],solveManifold);
+					resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
 				}
 			}
 
 			int numRollingFrictionPoolConstraints = m_tmpSolverContactRollingFrictionConstraintPool.size();
-			for (int j=0;j<numRollingFrictionPoolConstraints;j++)
+			for (int j = 0; j < numRollingFrictionPoolConstraints; j++)
 			{
 				b3SolverConstraint& rollingFrictionConstraint = m_tmpSolverContactRollingFrictionConstraintPool[j];
 				b3Scalar totalImpulse = m_tmpSolverContactConstraintPool[rollingFrictionConstraint.m_frictionIndex].m_appliedImpulse;
-				if (totalImpulse>b3Scalar(0))
+				if (totalImpulse > b3Scalar(0))
 				{
-					b3Scalar rollingFrictionMagnitude = rollingFrictionConstraint.m_friction*totalImpulse;
-					if (rollingFrictionMagnitude>rollingFrictionConstraint.m_friction)
+					b3Scalar rollingFrictionMagnitude = rollingFrictionConstraint.m_friction * totalImpulse;
+					if (rollingFrictionMagnitude > rollingFrictionConstraint.m_friction)
 						rollingFrictionMagnitude = rollingFrictionConstraint.m_friction;
 
 					rollingFrictionConstraint.m_lowerLimit = -rollingFrictionMagnitude;
 					rollingFrictionConstraint.m_upperLimit = rollingFrictionMagnitude;
 
-					resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdA],m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdB],rollingFrictionConstraint);
+					resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdA], m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdB], rollingFrictionConstraint);
 				}
 			}
 		}
@@ -1595,40 +1485,39 @@ b3Scalar b3PgsJacobiSolver::solveSingleIteration(int iteration,b3TypedConstraint
 	return 0.f;
 }
 
-
-void b3PgsJacobiSolver::solveGroupCacheFriendlySplitImpulseIterations(b3TypedConstraint** constraints,int numConstraints,const b3ContactSolverInfo& infoGlobal)
+void b3PgsJacobiSolver::solveGroupCacheFriendlySplitImpulseIterations(b3TypedConstraint** constraints, int numConstraints, const b3ContactSolverInfo& infoGlobal)
 {
 	int iteration;
 	if (infoGlobal.m_splitImpulse)
 	{
 		if (infoGlobal.m_solverMode & B3_SOLVER_SIMD)
 		{
-			for ( iteration = 0;iteration<infoGlobal.m_numIterations;iteration++)
+			for (iteration = 0; iteration < infoGlobal.m_numIterations; iteration++)
 			{
 				{
 					int numPoolConstraints = m_tmpSolverContactConstraintPool.size();
 					int j;
-					for (j=0;j<numPoolConstraints;j++)
+					for (j = 0; j < numPoolConstraints; j++)
 					{
 						const b3SolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[m_orderTmpConstraintPool[j]];
 
-						resolveSplitPenetrationSIMD(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA],m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB],solveManifold);
+						resolveSplitPenetrationSIMD(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
 					}
 				}
 			}
 		}
 		else
 		{
-			for ( iteration = 0;iteration<infoGlobal.m_numIterations;iteration++)
+			for (iteration = 0; iteration < infoGlobal.m_numIterations; iteration++)
 			{
 				{
 					int numPoolConstraints = m_tmpSolverContactConstraintPool.size();
 					int j;
-					for (j=0;j<numPoolConstraints;j++)
+					for (j = 0; j < numPoolConstraints; j++)
 					{
 						const b3SolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[m_orderTmpConstraintPool[j]];
 
-						resolveSplitPenetrationImpulseCacheFriendly(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA],m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB],solveManifold);
+						resolveSplitPenetrationImpulseCacheFriendly(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
 					}
 				}
 			}
@@ -1636,100 +1525,95 @@ void b3PgsJacobiSolver::solveGroupCacheFriendlySplitImpulseIterations(b3TypedCon
 	}
 }
 
-b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlyIterations(b3TypedConstraint** constraints,int numConstraints,const b3ContactSolverInfo& infoGlobal)
+b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlyIterations(b3TypedConstraint** constraints, int numConstraints, const b3ContactSolverInfo& infoGlobal)
 {
 	B3_PROFILE("solveGroupCacheFriendlyIterations");
 
 	{
 		///this is a special step to resolve penetrations (just for contacts)
-		solveGroupCacheFriendlySplitImpulseIterations(constraints,numConstraints,infoGlobal);
+		solveGroupCacheFriendlySplitImpulseIterations(constraints, numConstraints, infoGlobal);
 
-		int maxIterations = m_maxOverrideNumSolverIterations > infoGlobal.m_numIterations? m_maxOverrideNumSolverIterations : infoGlobal.m_numIterations;
+		int maxIterations = m_maxOverrideNumSolverIterations > infoGlobal.m_numIterations ? m_maxOverrideNumSolverIterations : infoGlobal.m_numIterations;
 
-		for ( int iteration = 0 ; iteration< maxIterations ; iteration++)
+		for (int iteration = 0; iteration < maxIterations; iteration++)
 		//for ( int iteration = maxIterations-1  ; iteration >= 0;iteration--)
-		{			
-			
-			solveSingleIteration(iteration, constraints,numConstraints,infoGlobal);
-
+		{
+			solveSingleIteration(iteration, constraints, numConstraints, infoGlobal);
 
 			if (!m_usePgs)
 			{
 				averageVelocities();
 			}
 		}
-		
 	}
 	return 0.f;
 }
 
-void	b3PgsJacobiSolver::averageVelocities()
+void b3PgsJacobiSolver::averageVelocities()
 {
 	B3_PROFILE("averaging");
 	//average the velocities
 	int numBodies = m_bodyCount.size();
 
 	m_deltaLinearVelocities.resize(0);
-	m_deltaLinearVelocities.resize(numBodies,b3MakeVector3(0,0,0));
+	m_deltaLinearVelocities.resize(numBodies, b3MakeVector3(0, 0, 0));
 	m_deltaAngularVelocities.resize(0);
-	m_deltaAngularVelocities.resize(numBodies,b3MakeVector3(0,0,0));
+	m_deltaAngularVelocities.resize(numBodies, b3MakeVector3(0, 0, 0));
 
-	for (int i=0;i<m_tmpSolverBodyPool.size();i++)
+	for (int i = 0; i < m_tmpSolverBodyPool.size(); i++)
 	{
 		if (!m_tmpSolverBodyPool[i].m_invMass.isZero())
 		{
 			int orgBodyIndex = m_tmpSolverBodyPool[i].m_originalBodyIndex;
-			m_deltaLinearVelocities[orgBodyIndex]+=m_tmpSolverBodyPool[i].getDeltaLinearVelocity();
-			m_deltaAngularVelocities[orgBodyIndex]+=m_tmpSolverBodyPool[i].getDeltaAngularVelocity();
+			m_deltaLinearVelocities[orgBodyIndex] += m_tmpSolverBodyPool[i].getDeltaLinearVelocity();
+			m_deltaAngularVelocities[orgBodyIndex] += m_tmpSolverBodyPool[i].getDeltaAngularVelocity();
 		}
 	}
-				
-	for (int i=0;i<m_tmpSolverBodyPool.size();i++)
+
+	for (int i = 0; i < m_tmpSolverBodyPool.size(); i++)
 	{
 		int orgBodyIndex = m_tmpSolverBodyPool[i].m_originalBodyIndex;
 
 		if (!m_tmpSolverBodyPool[i].m_invMass.isZero())
 		{
-						
 			b3Assert(m_bodyCount[orgBodyIndex] == m_bodyCountCheck[orgBodyIndex]);
-						
-			b3Scalar factor = 1.f/b3Scalar(m_bodyCount[orgBodyIndex]);
-						
 
-			m_tmpSolverBodyPool[i].m_deltaLinearVelocity = m_deltaLinearVelocities[orgBodyIndex]*factor;
-			m_tmpSolverBodyPool[i].m_deltaAngularVelocity = m_deltaAngularVelocities[orgBodyIndex]*factor;
+			b3Scalar factor = 1.f / b3Scalar(m_bodyCount[orgBodyIndex]);
+
+			m_tmpSolverBodyPool[i].m_deltaLinearVelocity = m_deltaLinearVelocities[orgBodyIndex] * factor;
+			m_tmpSolverBodyPool[i].m_deltaAngularVelocity = m_deltaAngularVelocities[orgBodyIndex] * factor;
 		}
 	}
 }
 
-b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlyFinish(b3RigidBodyData* bodies,b3InertiaData* inertias,int numBodies,const b3ContactSolverInfo& infoGlobal)
+b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlyFinish(b3RigidBodyData* bodies, b3InertiaData* inertias, int numBodies, const b3ContactSolverInfo& infoGlobal)
 {
 	B3_PROFILE("solveGroupCacheFriendlyFinish");
 	int numPoolConstraints = m_tmpSolverContactConstraintPool.size();
-	int i,j;
+	int i, j;
 
 	if (infoGlobal.m_solverMode & B3_SOLVER_USE_WARMSTARTING)
 	{
-		for (j=0;j<numPoolConstraints;j++)
+		for (j = 0; j < numPoolConstraints; j++)
 		{
 			const b3SolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[j];
-			b3ContactPoint* pt = (b3ContactPoint*) solveManifold.m_originalContactPoint;
+			b3ContactPoint* pt = (b3ContactPoint*)solveManifold.m_originalContactPoint;
 			b3Assert(pt);
 			pt->m_appliedImpulse = solveManifold.m_appliedImpulse;
-		//	float f = m_tmpSolverContactFrictionConstraintPool[solveManifold.m_frictionIndex].m_appliedImpulse;
+			//	float f = m_tmpSolverContactFrictionConstraintPool[solveManifold.m_frictionIndex].m_appliedImpulse;
 			//	printf("pt->m_appliedImpulseLateral1 = %f\n", f);
 			pt->m_appliedImpulseLateral1 = m_tmpSolverContactFrictionConstraintPool[solveManifold.m_frictionIndex].m_appliedImpulse;
 			//printf("pt->m_appliedImpulseLateral1 = %f\n", pt->m_appliedImpulseLateral1);
 			if ((infoGlobal.m_solverMode & B3_SOLVER_USE_2_FRICTION_DIRECTIONS))
 			{
-				pt->m_appliedImpulseLateral2 = m_tmpSolverContactFrictionConstraintPool[solveManifold.m_frictionIndex+1].m_appliedImpulse;
+				pt->m_appliedImpulseLateral2 = m_tmpSolverContactFrictionConstraintPool[solveManifold.m_frictionIndex + 1].m_appliedImpulse;
 			}
 			//do a callback here?
 		}
 	}
 
 	numPoolConstraints = m_tmpSolverNonContactConstraintPool.size();
-	for (j=0;j<numPoolConstraints;j++)
+	for (j = 0; j < numPoolConstraints; j++)
 	{
 		const b3SolverConstraint& solverConstr = m_tmpSolverNonContactConstraintPool[j];
 		b3TypedConstraint* constr = (b3TypedConstraint*)solverConstr.m_originalContactPoint;
@@ -1739,15 +1623,14 @@ b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlyFinish(b3RigidBodyData* bodie
 			b3SolverBody* bodyA = &m_tmpSolverBodyPool[solverConstr.m_solverBodyIdA];
 			b3SolverBody* bodyB = &m_tmpSolverBodyPool[solverConstr.m_solverBodyIdB];
 
-			fb->m_appliedForceBodyA += solverConstr.m_contactNormal*solverConstr.m_appliedImpulse*bodyA->m_linearFactor/infoGlobal.m_timeStep;
-			fb->m_appliedForceBodyB += -solverConstr.m_contactNormal*solverConstr.m_appliedImpulse*bodyB->m_linearFactor/infoGlobal.m_timeStep;
-			fb->m_appliedTorqueBodyA += solverConstr.m_relpos1CrossNormal* bodyA->m_angularFactor*solverConstr.m_appliedImpulse/infoGlobal.m_timeStep;
-			fb->m_appliedTorqueBodyB += -solverConstr.m_relpos1CrossNormal* bodyB->m_angularFactor*solverConstr.m_appliedImpulse/infoGlobal.m_timeStep;
-			
+			fb->m_appliedForceBodyA += solverConstr.m_contactNormal * solverConstr.m_appliedImpulse * bodyA->m_linearFactor / infoGlobal.m_timeStep;
+			fb->m_appliedForceBodyB += -solverConstr.m_contactNormal * solverConstr.m_appliedImpulse * bodyB->m_linearFactor / infoGlobal.m_timeStep;
+			fb->m_appliedTorqueBodyA += solverConstr.m_relpos1CrossNormal * bodyA->m_angularFactor * solverConstr.m_appliedImpulse / infoGlobal.m_timeStep;
+			fb->m_appliedTorqueBodyB += -solverConstr.m_relpos1CrossNormal * bodyB->m_angularFactor * solverConstr.m_appliedImpulse / infoGlobal.m_timeStep;
 		}
 
 		constr->internalSetAppliedImpulse(solverConstr.m_appliedImpulse);
-		if (b3Fabs(solverConstr.m_appliedImpulse)>=constr->getBreakingImpulseThreshold())
+		if (b3Fabs(solverConstr.m_appliedImpulse) >= constr->getBreakingImpulseThreshold())
 		{
 			constr->setEnabled(false);
 		}
@@ -1755,7 +1638,7 @@ b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlyFinish(b3RigidBodyData* bodie
 
 	{
 		B3_PROFILE("write back velocities and transforms");
-		for ( i=0;i<m_tmpSolverBodyPool.size();i++)
+		for (i = 0; i < m_tmpSolverBodyPool.size(); i++)
 		{
 			int bodyIndex = m_tmpSolverBodyPool[i].m_originalBodyIndex;
 			//b3Assert(i==bodyIndex);
@@ -1772,12 +1655,13 @@ b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlyFinish(b3RigidBodyData* bodie
 				{
 					body->m_linVel = m_tmpSolverBodyPool[i].m_linearVelocity;
 					body->m_angVel = m_tmpSolverBodyPool[i].m_angularVelocity;
-				} else
+				}
+				else
 				{
-					b3Scalar factor = 1.f/b3Scalar(m_bodyCount[bodyIndex]);
+					b3Scalar factor = 1.f / b3Scalar(m_bodyCount[bodyIndex]);
 
-					b3Vector3 deltaLinVel = m_deltaLinearVelocities[bodyIndex]*factor;
-					b3Vector3 deltaAngVel = m_deltaAngularVelocities[bodyIndex]*factor;
+					b3Vector3 deltaLinVel = m_deltaLinearVelocities[bodyIndex] * factor;
+					b3Vector3 deltaAngVel = m_deltaAngularVelocities[bodyIndex] * factor;
 					//printf("body %d\n",bodyIndex);
 					//printf("deltaLinVel = %f,%f,%f\n",deltaLinVel.getX(),deltaLinVel.getY(),deltaLinVel.getZ());
 					//printf("deltaAngVel = %f,%f,%f\n",deltaAngVel.getX(),deltaAngVel.getY(),deltaAngVel.getZ());
@@ -1785,7 +1669,7 @@ b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlyFinish(b3RigidBodyData* bodie
 					body->m_linVel += deltaLinVel;
 					body->m_angVel += deltaAngVel;
 				}
-			
+
 				if (infoGlobal.m_splitImpulse)
 				{
 					body->m_pos = m_tmpSolverBodyPool[i].m_worldTransform.getOrigin();
@@ -1797,7 +1681,6 @@ b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlyFinish(b3RigidBodyData* bodie
 		}
 	}
 
-
 	m_tmpSolverContactConstraintPool.resizeNoInitialize(0);
 	m_tmpSolverNonContactConstraintPool.resizeNoInitialize(0);
 	m_tmpSolverContactFrictionConstraintPool.resizeNoInitialize(0);
@@ -1807,9 +1690,7 @@ b3Scalar b3PgsJacobiSolver::solveGroupCacheFriendlyFinish(b3RigidBodyData* bodie
 	return 0.f;
 }
 
-
-
-void	b3PgsJacobiSolver::reset()
+void b3PgsJacobiSolver::reset()
 {
 	m_btSeed2 = 0;
 }
\ No newline at end of file
diff --git a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3PgsJacobiSolver.h b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3PgsJacobiSolver.h
index d2ca307fab3..5b616541d95 100644
--- a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3PgsJacobiSolver.h
+++ b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3PgsJacobiSolver.h
@@ -1,11 +1,9 @@
 #ifndef B3_PGS_JACOBI_SOLVER
 #define B3_PGS_JACOBI_SOLVER
 
-
 struct b3Contact4;
 struct b3ContactPoint;
 
-
 class b3Dispatcher;
 
 #include "b3TypedConstraint.h"
@@ -18,132 +16,118 @@ struct b3InertiaData;
 
 class b3PgsJacobiSolver
 {
-
 protected:
-	b3AlignedObjectArray<b3SolverBody>      m_tmpSolverBodyPool;
-	b3ConstraintArray			m_tmpSolverContactConstraintPool;
-	b3ConstraintArray			m_tmpSolverNonContactConstraintPool;
-	b3ConstraintArray			m_tmpSolverContactFrictionConstraintPool;
-	b3ConstraintArray			m_tmpSolverContactRollingFrictionConstraintPool;
+	b3AlignedObjectArray<b3SolverBody> m_tmpSolverBodyPool;
+	b3ConstraintArray m_tmpSolverContactConstraintPool;
+	b3ConstraintArray m_tmpSolverNonContactConstraintPool;
+	b3ConstraintArray m_tmpSolverContactFrictionConstraintPool;
+	b3ConstraintArray m_tmpSolverContactRollingFrictionConstraintPool;
 
-	b3AlignedObjectArray<int>	m_orderTmpConstraintPool;
-	b3AlignedObjectArray<int>	m_orderNonContactConstraintPool;
-	b3AlignedObjectArray<int>	m_orderFrictionConstraintPool;
+	b3AlignedObjectArray<int> m_orderTmpConstraintPool;
+	b3AlignedObjectArray<int> m_orderNonContactConstraintPool;
+	b3AlignedObjectArray<int> m_orderFrictionConstraintPool;
 	b3AlignedObjectArray<b3TypedConstraint::b3ConstraintInfo1> m_tmpConstraintSizesPool;
-	
-	b3AlignedObjectArray<int>		m_bodyCount;
-	b3AlignedObjectArray<int>		m_bodyCountCheck;
-	
-	b3AlignedObjectArray<b3Vector3>	m_deltaLinearVelocities;
-	b3AlignedObjectArray<b3Vector3>	m_deltaAngularVelocities;
 
-	bool						m_usePgs;
-	void						averageVelocities();
+	b3AlignedObjectArray<int> m_bodyCount;
+	b3AlignedObjectArray<int> m_bodyCountCheck;
 
-	int							m_maxOverrideNumSolverIterations;
+	b3AlignedObjectArray<b3Vector3> m_deltaLinearVelocities;
+	b3AlignedObjectArray<b3Vector3> m_deltaAngularVelocities;
 
-	int							m_numSplitImpulseRecoveries;
+	bool m_usePgs;
+	void averageVelocities();
 
-	b3Scalar	getContactProcessingThreshold(b3Contact4* contact)
+	int m_maxOverrideNumSolverIterations;
+
+	int m_numSplitImpulseRecoveries;
+
+	b3Scalar getContactProcessingThreshold(b3Contact4* contact)
 	{
 		return 0.02f;
 	}
-	void setupFrictionConstraint(	b3RigidBodyData* bodies,b3InertiaData* inertias, b3SolverConstraint& solverConstraint, const b3Vector3& normalAxis,int solverBodyIdA,int  solverBodyIdB,
-									b3ContactPoint& cp,const b3Vector3& rel_pos1,const b3Vector3& rel_pos2,
-									b3RigidBodyData* colObj0,b3RigidBodyData* colObj1, b3Scalar relaxation, 
-									b3Scalar desiredVelocity=0., b3Scalar cfmSlip=0.);
+	void setupFrictionConstraint(b3RigidBodyData* bodies, b3InertiaData* inertias, b3SolverConstraint& solverConstraint, const b3Vector3& normalAxis, int solverBodyIdA, int solverBodyIdB,
+								 b3ContactPoint& cp, const b3Vector3& rel_pos1, const b3Vector3& rel_pos2,
+								 b3RigidBodyData* colObj0, b3RigidBodyData* colObj1, b3Scalar relaxation,
+								 b3Scalar desiredVelocity = 0., b3Scalar cfmSlip = 0.);
 
-	void setupRollingFrictionConstraint(b3RigidBodyData* bodies,b3InertiaData* inertias,	b3SolverConstraint& solverConstraint, const b3Vector3& normalAxis,int solverBodyIdA,int  solverBodyIdB,
-									b3ContactPoint& cp,const b3Vector3& rel_pos1,const b3Vector3& rel_pos2,
-									b3RigidBodyData* colObj0,b3RigidBodyData* colObj1, b3Scalar relaxation, 
-									b3Scalar desiredVelocity=0., b3Scalar cfmSlip=0.);
-
-	b3SolverConstraint&	addFrictionConstraint(b3RigidBodyData* bodies,b3InertiaData* inertias,const b3Vector3& normalAxis,int solverBodyIdA,int solverBodyIdB,int frictionIndex,b3ContactPoint& cp,const b3Vector3& rel_pos1,const b3Vector3& rel_pos2,b3RigidBodyData* colObj0,b3RigidBodyData* colObj1, b3Scalar relaxation, b3Scalar desiredVelocity=0., b3Scalar cfmSlip=0.);
-	b3SolverConstraint&	addRollingFrictionConstraint(b3RigidBodyData* bodies,b3InertiaData* inertias,const b3Vector3& normalAxis,int solverBodyIdA,int solverBodyIdB,int frictionIndex,b3ContactPoint& cp,const b3Vector3& rel_pos1,const b3Vector3& rel_pos2,b3RigidBodyData* colObj0,b3RigidBodyData* colObj1, b3Scalar relaxation, b3Scalar desiredVelocity=0, b3Scalar cfmSlip=0.f);
+	void setupRollingFrictionConstraint(b3RigidBodyData* bodies, b3InertiaData* inertias, b3SolverConstraint& solverConstraint, const b3Vector3& normalAxis, int solverBodyIdA, int solverBodyIdB,
+										b3ContactPoint& cp, const b3Vector3& rel_pos1, const b3Vector3& rel_pos2,
+										b3RigidBodyData* colObj0, b3RigidBodyData* colObj1, b3Scalar relaxation,
+										b3Scalar desiredVelocity = 0., b3Scalar cfmSlip = 0.);
 
+	b3SolverConstraint& addFrictionConstraint(b3RigidBodyData* bodies, b3InertiaData* inertias, const b3Vector3& normalAxis, int solverBodyIdA, int solverBodyIdB, int frictionIndex, b3ContactPoint& cp, const b3Vector3& rel_pos1, const b3Vector3& rel_pos2, b3RigidBodyData* colObj0, b3RigidBodyData* colObj1, b3Scalar relaxation, b3Scalar desiredVelocity = 0., b3Scalar cfmSlip = 0.);
+	b3SolverConstraint& addRollingFrictionConstraint(b3RigidBodyData* bodies, b3InertiaData* inertias, const b3Vector3& normalAxis, int solverBodyIdA, int solverBodyIdB, int frictionIndex, b3ContactPoint& cp, const b3Vector3& rel_pos1, const b3Vector3& rel_pos2, b3RigidBodyData* colObj0, b3RigidBodyData* colObj1, b3Scalar relaxation, b3Scalar desiredVelocity = 0, b3Scalar cfmSlip = 0.f);
 
 	void setupContactConstraint(b3RigidBodyData* bodies, b3InertiaData* inertias,
-								b3SolverConstraint& solverConstraint, int solverBodyIdA, int solverBodyIdB, b3ContactPoint& cp, 
-								const b3ContactSolverInfo& infoGlobal, b3Vector3& vel, b3Scalar& rel_vel, b3Scalar& relaxation, 
+								b3SolverConstraint& solverConstraint, int solverBodyIdA, int solverBodyIdB, b3ContactPoint& cp,
+								const b3ContactSolverInfo& infoGlobal, b3Vector3& vel, b3Scalar& rel_vel, b3Scalar& relaxation,
 								b3Vector3& rel_pos1, b3Vector3& rel_pos2);
 
-	void setFrictionConstraintImpulse( b3RigidBodyData* bodies, b3InertiaData* inertias,b3SolverConstraint& solverConstraint, int solverBodyIdA,int solverBodyIdB, 
-										 b3ContactPoint& cp, const b3ContactSolverInfo& infoGlobal);
+	void setFrictionConstraintImpulse(b3RigidBodyData* bodies, b3InertiaData* inertias, b3SolverConstraint& solverConstraint, int solverBodyIdA, int solverBodyIdB,
+									  b3ContactPoint& cp, const b3ContactSolverInfo& infoGlobal);
 
 	///m_btSeed2 is used for re-arranging the constraint rows. improves convergence/quality of friction
-	unsigned long	m_btSeed2;
+	unsigned long m_btSeed2;
 
-	
 	b3Scalar restitutionCurve(b3Scalar rel_vel, b3Scalar restitution);
 
-	void	convertContact(b3RigidBodyData* bodies, b3InertiaData* inertias,b3Contact4* manifold,const b3ContactSolverInfo& infoGlobal);
+	void convertContact(b3RigidBodyData* bodies, b3InertiaData* inertias, b3Contact4* manifold, const b3ContactSolverInfo& infoGlobal);
 
+	void resolveSplitPenetrationSIMD(
+		b3SolverBody& bodyA, b3SolverBody& bodyB,
+		const b3SolverConstraint& contactConstraint);
 
-	void	resolveSplitPenetrationSIMD(
-     b3SolverBody& bodyA,b3SolverBody& bodyB,
-        const b3SolverConstraint& contactConstraint);
-
-	void	resolveSplitPenetrationImpulseCacheFriendly(
-       b3SolverBody& bodyA,b3SolverBody& bodyB,
-        const b3SolverConstraint& contactConstraint);
+	void resolveSplitPenetrationImpulseCacheFriendly(
+		b3SolverBody& bodyA, b3SolverBody& bodyB,
+		const b3SolverConstraint& contactConstraint);
 
 	//internal method
-	int		getOrInitSolverBody(int bodyIndex, b3RigidBodyData* bodies,b3InertiaData* inertias);
-	void	initSolverBody(int bodyIndex, b3SolverBody* solverBody, b3RigidBodyData* collisionObject);
+	int getOrInitSolverBody(int bodyIndex, b3RigidBodyData* bodies, b3InertiaData* inertias);
+	void initSolverBody(int bodyIndex, b3SolverBody* solverBody, b3RigidBodyData* collisionObject);
 
-	void	resolveSingleConstraintRowGeneric(b3SolverBody& bodyA,b3SolverBody& bodyB,const b3SolverConstraint& contactConstraint);
+	void resolveSingleConstraintRowGeneric(b3SolverBody& bodyA, b3SolverBody& bodyB, const b3SolverConstraint& contactConstraint);
+
+	void resolveSingleConstraintRowGenericSIMD(b3SolverBody& bodyA, b3SolverBody& bodyB, const b3SolverConstraint& contactConstraint);
+
+	void resolveSingleConstraintRowLowerLimit(b3SolverBody& bodyA, b3SolverBody& bodyB, const b3SolverConstraint& contactConstraint);
+
+	void resolveSingleConstraintRowLowerLimitSIMD(b3SolverBody& bodyA, b3SolverBody& bodyB, const b3SolverConstraint& contactConstraint);
 
-	void	resolveSingleConstraintRowGenericSIMD(b3SolverBody& bodyA,b3SolverBody& bodyB,const b3SolverConstraint& contactConstraint);
-	
-	void	resolveSingleConstraintRowLowerLimit(b3SolverBody& bodyA,b3SolverBody& bodyB,const b3SolverConstraint& contactConstraint);
-	
-	void	resolveSingleConstraintRowLowerLimitSIMD(b3SolverBody& bodyA,b3SolverBody& bodyB,const b3SolverConstraint& contactConstraint);
-		
 protected:
+	virtual b3Scalar solveGroupCacheFriendlySetup(b3RigidBodyData* bodies, b3InertiaData* inertias, int numBodies, b3Contact4* manifoldPtr, int numManifolds, b3TypedConstraint** constraints, int numConstraints, const b3ContactSolverInfo& infoGlobal);
 
-	virtual b3Scalar solveGroupCacheFriendlySetup(b3RigidBodyData* bodies, b3InertiaData* inertias,int numBodies,b3Contact4* manifoldPtr, int numManifolds,b3TypedConstraint** constraints,int numConstraints,const b3ContactSolverInfo& infoGlobal);
-
-
-	virtual b3Scalar solveGroupCacheFriendlyIterations(b3TypedConstraint** constraints,int numConstraints,const b3ContactSolverInfo& infoGlobal);
-	virtual void solveGroupCacheFriendlySplitImpulseIterations(b3TypedConstraint** constraints,int numConstraints,const b3ContactSolverInfo& infoGlobal);
-	b3Scalar solveSingleIteration(int iteration, b3TypedConstraint** constraints,int numConstraints,const b3ContactSolverInfo& infoGlobal);
-
-
-	virtual b3Scalar solveGroupCacheFriendlyFinish(b3RigidBodyData* bodies, b3InertiaData* inertias,int numBodies,const b3ContactSolverInfo& infoGlobal);
+	virtual b3Scalar solveGroupCacheFriendlyIterations(b3TypedConstraint** constraints, int numConstraints, const b3ContactSolverInfo& infoGlobal);
+	virtual void solveGroupCacheFriendlySplitImpulseIterations(b3TypedConstraint** constraints, int numConstraints, const b3ContactSolverInfo& infoGlobal);
+	b3Scalar solveSingleIteration(int iteration, b3TypedConstraint** constraints, int numConstraints, const b3ContactSolverInfo& infoGlobal);
 
+	virtual b3Scalar solveGroupCacheFriendlyFinish(b3RigidBodyData* bodies, b3InertiaData* inertias, int numBodies, const b3ContactSolverInfo& infoGlobal);
 
 public:
-
 	B3_DECLARE_ALIGNED_ALLOCATOR();
-	
+
 	b3PgsJacobiSolver(bool usePgs);
 	virtual ~b3PgsJacobiSolver();
 
-//	void	solveContacts(int numBodies, b3RigidBodyData* bodies, b3InertiaData* inertias, int numContacts, b3Contact4* contacts);
-	void	solveContacts(int numBodies, b3RigidBodyData* bodies, b3InertiaData* inertias, int numContacts, b3Contact4* contacts, int numConstraints, b3TypedConstraint** constraints);
+	//	void	solveContacts(int numBodies, b3RigidBodyData* bodies, b3InertiaData* inertias, int numContacts, b3Contact4* contacts);
+	void solveContacts(int numBodies, b3RigidBodyData* bodies, b3InertiaData* inertias, int numContacts, b3Contact4* contacts, int numConstraints, b3TypedConstraint** constraints);
 
-	b3Scalar solveGroup(b3RigidBodyData* bodies,b3InertiaData* inertias,int numBodies,b3Contact4* manifoldPtr, int numManifolds,b3TypedConstraint** constraints,int numConstraints,const b3ContactSolverInfo& infoGlobal);
+	b3Scalar solveGroup(b3RigidBodyData* bodies, b3InertiaData* inertias, int numBodies, b3Contact4* manifoldPtr, int numManifolds, b3TypedConstraint** constraints, int numConstraints, const b3ContactSolverInfo& infoGlobal);
 
 	///clear internal cached data and reset random seed
-	virtual	void	reset();
-	
+	virtual void reset();
+
 	unsigned long b3Rand2();
 
-	int b3RandInt2 (int n);
+	int b3RandInt2(int n);
 
-	void	setRandSeed(unsigned long seed)
+	void setRandSeed(unsigned long seed)
 	{
 		m_btSeed2 = seed;
 	}
-	unsigned long	getRandSeed() const
+	unsigned long getRandSeed() const
 	{
 		return m_btSeed2;
 	}
-
-
-
-
 };
 
-#endif //B3_PGS_JACOBI_SOLVER
-
+#endif  //B3_PGS_JACOBI_SOLVER
diff --git a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3Point2PointConstraint.cpp b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3Point2PointConstraint.cpp
index 02c11db3209..cfa7c7dd114 100644
--- a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3Point2PointConstraint.cpp
+++ b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3Point2PointConstraint.cpp
@@ -13,21 +13,14 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "b3Point2PointConstraint.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3RigidBodyData.h"
 
 #include <new>
 
-
-
-
-
-b3Point2PointConstraint::b3Point2PointConstraint(int rbA,int rbB, const b3Vector3& pivotInA,const b3Vector3& pivotInB)
-:b3TypedConstraint(B3_POINT2POINT_CONSTRAINT_TYPE,rbA,rbB),m_pivotInA(pivotInA),m_pivotInB(pivotInB),
-m_flags(0)
+b3Point2PointConstraint::b3Point2PointConstraint(int rbA, int rbB, const b3Vector3& pivotInA, const b3Vector3& pivotInB)
+	: b3TypedConstraint(B3_POINT2POINT_CONSTRAINT_TYPE, rbA, rbB), m_pivotInA(pivotInA), m_pivotInB(pivotInB), m_flags(0)
 {
-
 }
 
 /*
@@ -40,22 +33,18 @@ m_useSolveConstraintObsolete(false)
 }
 */
 
-
-void b3Point2PointConstraint::getInfo1 (b3ConstraintInfo1* info,const b3RigidBodyData* bodies)
+void b3Point2PointConstraint::getInfo1(b3ConstraintInfo1* info, const b3RigidBodyData* bodies)
 {
-	getInfo1NonVirtual(info,bodies);
+	getInfo1NonVirtual(info, bodies);
 }
 
-void b3Point2PointConstraint::getInfo1NonVirtual (b3ConstraintInfo1* info,const b3RigidBodyData* bodies)
+void b3Point2PointConstraint::getInfo1NonVirtual(b3ConstraintInfo1* info, const b3RigidBodyData* bodies)
 {
-		info->m_numConstraintRows = 3;
-		info->nub = 3;
+	info->m_numConstraintRows = 3;
+	info->nub = 3;
 }
 
-
-
-
-void b3Point2PointConstraint::getInfo2 (b3ConstraintInfo2* info, const b3RigidBodyData* bodies)
+void b3Point2PointConstraint::getInfo2(b3ConstraintInfo2* info, const b3RigidBodyData* bodies)
 {
 	b3Transform trA;
 	trA.setIdentity();
@@ -67,143 +56,135 @@ void b3Point2PointConstraint::getInfo2 (b3ConstraintInfo2* info, const b3RigidBo
 	trB.setOrigin(bodies[m_rbB].m_pos);
 	trB.setRotation(bodies[m_rbB].m_quat);
 
-	getInfo2NonVirtual(info, trA,trB);
+	getInfo2NonVirtual(info, trA, trB);
 }
 
-void b3Point2PointConstraint::getInfo2NonVirtual (b3ConstraintInfo2* info, const b3Transform& body0_trans, const b3Transform& body1_trans)
+void b3Point2PointConstraint::getInfo2NonVirtual(b3ConstraintInfo2* info, const b3Transform& body0_trans, const b3Transform& body1_trans)
 {
-
-	 //retrieve matrices
+	//retrieve matrices
 
 	// anchor points in global coordinates with respect to body PORs.
-   
-    // set jacobian
-    info->m_J1linearAxis[0] = 1;
-	info->m_J1linearAxis[info->rowskip+1] = 1;
-	info->m_J1linearAxis[2*info->rowskip+2] = 1;
 
-	b3Vector3 a1 = body0_trans.getBasis()*getPivotInA();
+	// set jacobian
+	info->m_J1linearAxis[0] = 1;
+	info->m_J1linearAxis[info->rowskip + 1] = 1;
+	info->m_J1linearAxis[2 * info->rowskip + 2] = 1;
+
+	b3Vector3 a1 = body0_trans.getBasis() * getPivotInA();
 	//b3Vector3 a1a = b3QuatRotate(body0_trans.getRotation(),getPivotInA());
 
 	{
 		b3Vector3* angular0 = (b3Vector3*)(info->m_J1angularAxis);
-		b3Vector3* angular1 = (b3Vector3*)(info->m_J1angularAxis+info->rowskip);
-		b3Vector3* angular2 = (b3Vector3*)(info->m_J1angularAxis+2*info->rowskip);
+		b3Vector3* angular1 = (b3Vector3*)(info->m_J1angularAxis + info->rowskip);
+		b3Vector3* angular2 = (b3Vector3*)(info->m_J1angularAxis + 2 * info->rowskip);
 		b3Vector3 a1neg = -a1;
-		a1neg.getSkewSymmetricMatrix(angular0,angular1,angular2);
+		a1neg.getSkewSymmetricMatrix(angular0, angular1, angular2);
 	}
-    
+
 	if (info->m_J2linearAxis)
 	{
 		info->m_J2linearAxis[0] = -1;
-		info->m_J2linearAxis[info->rowskip+1] = -1;
-		info->m_J2linearAxis[2*info->rowskip+2] = -1;
+		info->m_J2linearAxis[info->rowskip + 1] = -1;
+		info->m_J2linearAxis[2 * info->rowskip + 2] = -1;
 	}
-	
-	b3Vector3 a2 = body1_trans.getBasis()*getPivotInB();
-   
+
+	b3Vector3 a2 = body1_trans.getBasis() * getPivotInB();
+
 	{
-	//	b3Vector3 a2n = -a2;
+		//	b3Vector3 a2n = -a2;
 		b3Vector3* angular0 = (b3Vector3*)(info->m_J2angularAxis);
-		b3Vector3* angular1 = (b3Vector3*)(info->m_J2angularAxis+info->rowskip);
-		b3Vector3* angular2 = (b3Vector3*)(info->m_J2angularAxis+2*info->rowskip);
-		a2.getSkewSymmetricMatrix(angular0,angular1,angular2);
+		b3Vector3* angular1 = (b3Vector3*)(info->m_J2angularAxis + info->rowskip);
+		b3Vector3* angular2 = (b3Vector3*)(info->m_J2angularAxis + 2 * info->rowskip);
+		a2.getSkewSymmetricMatrix(angular0, angular1, angular2);
 	}
-    
 
-
-    // set right hand side
+	// set right hand side
 	b3Scalar currERP = (m_flags & B3_P2P_FLAGS_ERP) ? m_erp : info->erp;
-    b3Scalar k = info->fps * currERP;
-    int j;
-	for (j=0; j<3; j++)
-    {
-        info->m_constraintError[j*info->rowskip] = k * (a2[j] + body1_trans.getOrigin()[j] - a1[j] - body0_trans.getOrigin()[j]);
-		//printf("info->m_constraintError[%d]=%f\n",j,info->m_constraintError[j]);
-    }
-	if(m_flags & B3_P2P_FLAGS_CFM)
+	b3Scalar k = info->fps * currERP;
+	int j;
+	for (j = 0; j < 3; j++)
 	{
-		for (j=0; j<3; j++)
+		info->m_constraintError[j * info->rowskip] = k * (a2[j] + body1_trans.getOrigin()[j] - a1[j] - body0_trans.getOrigin()[j]);
+		//printf("info->m_constraintError[%d]=%f\n",j,info->m_constraintError[j]);
+	}
+	if (m_flags & B3_P2P_FLAGS_CFM)
+	{
+		for (j = 0; j < 3; j++)
 		{
-			info->cfm[j*info->rowskip] = m_cfm;
+			info->cfm[j * info->rowskip] = m_cfm;
 		}
 	}
 
-	b3Scalar impulseClamp = m_setting.m_impulseClamp;//
-	for (j=0; j<3; j++)
-    {
+	b3Scalar impulseClamp = m_setting.m_impulseClamp;  //
+	for (j = 0; j < 3; j++)
+	{
 		if (m_setting.m_impulseClamp > 0)
 		{
-			info->m_lowerLimit[j*info->rowskip] = -impulseClamp;
-			info->m_upperLimit[j*info->rowskip] = impulseClamp;
+			info->m_lowerLimit[j * info->rowskip] = -impulseClamp;
+			info->m_upperLimit[j * info->rowskip] = impulseClamp;
 		}
 	}
 	info->m_damping = m_setting.m_damping;
-	
 }
 
-
-
-void	b3Point2PointConstraint::updateRHS(b3Scalar	timeStep)
+void b3Point2PointConstraint::updateRHS(b3Scalar timeStep)
 {
 	(void)timeStep;
-
 }
 
-///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5). 
+///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5).
 ///If no axis is provided, it uses the default axis for this constraint.
 void b3Point2PointConstraint::setParam(int num, b3Scalar value, int axis)
 {
-	if(axis != -1)
+	if (axis != -1)
 	{
 		b3AssertConstrParams(0);
 	}
 	else
 	{
-		switch(num)
+		switch (num)
 		{
-			case B3_CONSTRAINT_ERP :
-			case B3_CONSTRAINT_STOP_ERP :
-				m_erp = value; 
+			case B3_CONSTRAINT_ERP:
+			case B3_CONSTRAINT_STOP_ERP:
+				m_erp = value;
 				m_flags |= B3_P2P_FLAGS_ERP;
 				break;
-			case B3_CONSTRAINT_CFM :
-			case B3_CONSTRAINT_STOP_CFM :
-				m_cfm = value; 
+			case B3_CONSTRAINT_CFM:
+			case B3_CONSTRAINT_STOP_CFM:
+				m_cfm = value;
 				m_flags |= B3_P2P_FLAGS_CFM;
 				break;
-			default: 
+			default:
 				b3AssertConstrParams(0);
 		}
 	}
 }
 
 ///return the local value of parameter
-b3Scalar b3Point2PointConstraint::getParam(int num, int axis) const 
+b3Scalar b3Point2PointConstraint::getParam(int num, int axis) const
 {
 	b3Scalar retVal(B3_INFINITY);
-	if(axis != -1)
+	if (axis != -1)
 	{
 		b3AssertConstrParams(0);
 	}
 	else
 	{
-		switch(num)
+		switch (num)
 		{
-			case B3_CONSTRAINT_ERP :
-			case B3_CONSTRAINT_STOP_ERP :
+			case B3_CONSTRAINT_ERP:
+			case B3_CONSTRAINT_STOP_ERP:
 				b3AssertConstrParams(m_flags & B3_P2P_FLAGS_ERP);
-				retVal = m_erp; 
+				retVal = m_erp;
 				break;
-			case B3_CONSTRAINT_CFM :
-			case B3_CONSTRAINT_STOP_CFM :
+			case B3_CONSTRAINT_CFM:
+			case B3_CONSTRAINT_STOP_CFM:
 				b3AssertConstrParams(m_flags & B3_P2P_FLAGS_CFM);
-				retVal = m_cfm; 
+				retVal = m_cfm;
 				break;
-			default: 
+			default:
 				b3AssertConstrParams(0);
 		}
 	}
 	return retVal;
 }
-	
diff --git a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3Point2PointConstraint.h b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3Point2PointConstraint.h
index 681b487334c..14762a3e358 100644
--- a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3Point2PointConstraint.h
+++ b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3Point2PointConstraint.h
@@ -22,26 +22,24 @@ subject to the following restrictions:
 
 class b3RigidBody;
 
-
 #ifdef B3_USE_DOUBLE_PRECISION
-#define b3Point2PointConstraintData	b3Point2PointConstraintDoubleData
-#define b3Point2PointConstraintDataName	"b3Point2PointConstraintDoubleData"
+#define b3Point2PointConstraintData b3Point2PointConstraintDoubleData
+#define b3Point2PointConstraintDataName "b3Point2PointConstraintDoubleData"
 #else
-#define b3Point2PointConstraintData	b3Point2PointConstraintFloatData
-#define b3Point2PointConstraintDataName	"b3Point2PointConstraintFloatData"
-#endif //B3_USE_DOUBLE_PRECISION
+#define b3Point2PointConstraintData b3Point2PointConstraintFloatData
+#define b3Point2PointConstraintDataName "b3Point2PointConstraintFloatData"
+#endif  //B3_USE_DOUBLE_PRECISION
 
-struct	b3ConstraintSetting
+struct b3ConstraintSetting
 {
-	b3ConstraintSetting()	:
-		m_tau(b3Scalar(0.3)),
-		m_damping(b3Scalar(1.)),
-		m_impulseClamp(b3Scalar(0.))
+	b3ConstraintSetting() : m_tau(b3Scalar(0.3)),
+							m_damping(b3Scalar(1.)),
+							m_impulseClamp(b3Scalar(0.))
 	{
 	}
-	b3Scalar		m_tau;
-	b3Scalar		m_damping;
-	b3Scalar		m_impulseClamp;
+	b3Scalar m_tau;
+	b3Scalar m_damping;
+	b3Scalar m_impulseClamp;
 };
 
 enum b3Point2PointFlags
@@ -51,47 +49,45 @@ enum b3Point2PointFlags
 };
 
 /// point to point constraint between two rigidbodies each with a pivotpoint that descibes the 'ballsocket' location in local space
-B3_ATTRIBUTE_ALIGNED16(class) b3Point2PointConstraint : public b3TypedConstraint
+B3_ATTRIBUTE_ALIGNED16(class)
+b3Point2PointConstraint : public b3TypedConstraint
 {
 #ifdef IN_PARALLELL_SOLVER
 public:
 #endif
-	
-	b3Vector3	m_pivotInA;
-	b3Vector3	m_pivotInB;
-	
-	int			m_flags;
-	b3Scalar	m_erp;
-	b3Scalar	m_cfm;
-	
-public:
 
+	b3Vector3 m_pivotInA;
+	b3Vector3 m_pivotInB;
+
+	int m_flags;
+	b3Scalar m_erp;
+	b3Scalar m_cfm;
+
+public:
 	B3_DECLARE_ALIGNED_ALLOCATOR();
 
-	b3ConstraintSetting	m_setting;
+	b3ConstraintSetting m_setting;
 
-	b3Point2PointConstraint(int  rbA,int rbB, const b3Vector3& pivotInA,const b3Vector3& pivotInB);
+	b3Point2PointConstraint(int rbA, int rbB, const b3Vector3& pivotInA, const b3Vector3& pivotInB);
 
 	//b3Point2PointConstraint(int  rbA,const b3Vector3& pivotInA);
 
+	virtual void getInfo1(b3ConstraintInfo1 * info, const b3RigidBodyData* bodies);
 
+	void getInfo1NonVirtual(b3ConstraintInfo1 * info, const b3RigidBodyData* bodies);
 
-	virtual void getInfo1 (b3ConstraintInfo1* info,const b3RigidBodyData* bodies);
+	virtual void getInfo2(b3ConstraintInfo2 * info, const b3RigidBodyData* bodies);
 
-	void getInfo1NonVirtual (b3ConstraintInfo1* info,const b3RigidBodyData* bodies);
+	void getInfo2NonVirtual(b3ConstraintInfo2 * info, const b3Transform& body0_trans, const b3Transform& body1_trans);
 
-	virtual void getInfo2 (b3ConstraintInfo2* info, const b3RigidBodyData* bodies);
+	void updateRHS(b3Scalar timeStep);
 
-	void getInfo2NonVirtual (b3ConstraintInfo2* info, const b3Transform& body0_trans, const b3Transform& body1_trans);
-
-	void	updateRHS(b3Scalar	timeStep);
-
-	void	setPivotA(const b3Vector3& pivotA)
+	void setPivotA(const b3Vector3& pivotA)
 	{
 		m_pivotInA = pivotA;
 	}
 
-	void	setPivotB(const b3Vector3& pivotB)
+	void setPivotB(const b3Vector3& pivotB)
 	{
 		m_pivotInB = pivotB;
 	}
@@ -106,34 +102,32 @@ public:
 		return m_pivotInB;
 	}
 
-	///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5). 
+	///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5).
 	///If no axis is provided, it uses the default axis for this constraint.
-	virtual	void	setParam(int num, b3Scalar value, int axis = -1);
+	virtual void setParam(int num, b3Scalar value, int axis = -1);
 	///return the local value of parameter
-	virtual	b3Scalar getParam(int num, int axis = -1) const;
+	virtual b3Scalar getParam(int num, int axis = -1) const;
 
-//	virtual	int	calculateSerializeBufferSize() const;
+	//	virtual	int	calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-//	virtual	const char*	serialize(void* dataBuffer, b3Serializer* serializer) const;
-
-
+	//	virtual	const char*	serialize(void* dataBuffer, b3Serializer* serializer) const;
 };
 
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
-struct	b3Point2PointConstraintFloatData
+struct b3Point2PointConstraintFloatData
 {
-	b3TypedConstraintData	m_typeConstraintData;
-	b3Vector3FloatData	m_pivotInA;
-	b3Vector3FloatData	m_pivotInB;
+	b3TypedConstraintData m_typeConstraintData;
+	b3Vector3FloatData m_pivotInA;
+	b3Vector3FloatData m_pivotInB;
 };
 
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
-struct	b3Point2PointConstraintDoubleData
+struct b3Point2PointConstraintDoubleData
 {
-	b3TypedConstraintData	m_typeConstraintData;
-	b3Vector3DoubleData	m_pivotInA;
-	b3Vector3DoubleData	m_pivotInB;
+	b3TypedConstraintData m_typeConstraintData;
+	b3Vector3DoubleData m_pivotInA;
+	b3Vector3DoubleData m_pivotInB;
 };
 
 /*
@@ -156,4 +150,4 @@ B3_FORCE_INLINE	const char*	b3Point2PointConstraint::serialize(void* dataBuffer,
 }
 */
 
-#endif //B3_POINT2POINTCONSTRAINT_H
+#endif  //B3_POINT2POINTCONSTRAINT_H
diff --git a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3SolverBody.h b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3SolverBody.h
index 0049317d981..196d0e5793b 100644
--- a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3SolverBody.h
+++ b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3SolverBody.h
@@ -16,7 +16,6 @@ subject to the following restrictions:
 #ifndef B3_SOLVER_BODY_H
 #define B3_SOLVER_BODY_H
 
-
 #include "Bullet3Common/b3Vector3.h"
 #include "Bullet3Common/b3Matrix3x3.h"
 
@@ -26,110 +25,104 @@ subject to the following restrictions:
 ///Until we get other contributions, only use SIMD on Windows, when using Visual Studio 2008 or later, and not double precision
 #ifdef B3_USE_SSE
 #define USE_SIMD 1
-#endif //
-
+#endif  //
 
 #ifdef USE_SIMD
 
-struct	b3SimdScalar
+struct b3SimdScalar
 {
-	B3_FORCE_INLINE	b3SimdScalar()
-	{
-
-	}
-
-	B3_FORCE_INLINE	b3SimdScalar(float	fl)
-	:m_vec128 (_mm_set1_ps(fl))
+	B3_FORCE_INLINE b3SimdScalar()
 	{
 	}
 
-	B3_FORCE_INLINE	b3SimdScalar(__m128 v128)
-		:m_vec128(v128)
+	B3_FORCE_INLINE b3SimdScalar(float fl)
+		: m_vec128(_mm_set1_ps(fl))
 	{
 	}
-	union
+
+	B3_FORCE_INLINE b3SimdScalar(__m128 v128)
+		: m_vec128(v128)
 	{
-		__m128		m_vec128;
-		float		m_floats[4];
-		float		x,y,z,w;
-		int			m_ints[4];
-		b3Scalar	m_unusedPadding;
+	}
+	union {
+		__m128 m_vec128;
+		float m_floats[4];
+		float x, y, z, w;
+		int m_ints[4];
+		b3Scalar m_unusedPadding;
 	};
-	B3_FORCE_INLINE	__m128	get128()
+	B3_FORCE_INLINE __m128 get128()
 	{
 		return m_vec128;
 	}
 
-	B3_FORCE_INLINE	const __m128	get128() const
+	B3_FORCE_INLINE const __m128 get128() const
 	{
 		return m_vec128;
 	}
 
-	B3_FORCE_INLINE	void	set128(__m128 v128)
+	B3_FORCE_INLINE void set128(__m128 v128)
 	{
 		m_vec128 = v128;
 	}
 
-	B3_FORCE_INLINE	operator       __m128()       
-	{ 
-		return m_vec128; 
+	B3_FORCE_INLINE operator __m128()
+	{
+		return m_vec128;
 	}
-	B3_FORCE_INLINE	operator const __m128() const 
-	{ 
-		return m_vec128; 
-	}
-	
-	B3_FORCE_INLINE	operator float() const 
-	{ 
-		return m_floats[0]; 
+	B3_FORCE_INLINE operator const __m128() const
+	{
+		return m_vec128;
 	}
 
+	B3_FORCE_INLINE operator float() const
+	{
+		return m_floats[0];
+	}
 };
 
 ///@brief Return the elementwise product of two b3SimdScalar
-B3_FORCE_INLINE b3SimdScalar 
-operator*(const b3SimdScalar& v1, const b3SimdScalar& v2) 
+B3_FORCE_INLINE b3SimdScalar
+operator*(const b3SimdScalar& v1, const b3SimdScalar& v2)
 {
-	return b3SimdScalar(_mm_mul_ps(v1.get128(),v2.get128()));
+	return b3SimdScalar(_mm_mul_ps(v1.get128(), v2.get128()));
 }
 
 ///@brief Return the elementwise product of two b3SimdScalar
-B3_FORCE_INLINE b3SimdScalar 
-operator+(const b3SimdScalar& v1, const b3SimdScalar& v2) 
+B3_FORCE_INLINE b3SimdScalar
+operator+(const b3SimdScalar& v1, const b3SimdScalar& v2)
 {
-	return b3SimdScalar(_mm_add_ps(v1.get128(),v2.get128()));
+	return b3SimdScalar(_mm_add_ps(v1.get128(), v2.get128()));
 }
 
-
 #else
 #define b3SimdScalar b3Scalar
 #endif
 
 ///The b3SolverBody is an internal datastructure for the constraint solver. Only necessary data is packed to increase cache coherence/performance.
-B3_ATTRIBUTE_ALIGNED16 (struct)	b3SolverBody
+B3_ATTRIBUTE_ALIGNED16(struct)
+b3SolverBody
 {
 	B3_DECLARE_ALIGNED_ALLOCATOR();
-	b3Transform		m_worldTransform;
-	b3Vector3		m_deltaLinearVelocity;
-	b3Vector3		m_deltaAngularVelocity;
-	b3Vector3		m_angularFactor;
-	b3Vector3		m_linearFactor;
-	b3Vector3		m_invMass;
-	b3Vector3		m_pushVelocity;
-	b3Vector3		m_turnVelocity;
-	b3Vector3		m_linearVelocity;
-	b3Vector3		m_angularVelocity;
+	b3Transform m_worldTransform;
+	b3Vector3 m_deltaLinearVelocity;
+	b3Vector3 m_deltaAngularVelocity;
+	b3Vector3 m_angularFactor;
+	b3Vector3 m_linearFactor;
+	b3Vector3 m_invMass;
+	b3Vector3 m_pushVelocity;
+	b3Vector3 m_turnVelocity;
+	b3Vector3 m_linearVelocity;
+	b3Vector3 m_angularVelocity;
 
-	union 
-	{
-		void*	m_originalBody;
-		int		m_originalBodyIndex;
+	union {
+		void* m_originalBody;
+		int m_originalBodyIndex;
 	};
 
 	int padding[3];
 
-
-	void	setWorldTransform(const b3Transform& worldTransform)
+	void setWorldTransform(const b3Transform& worldTransform)
 	{
 		m_worldTransform = worldTransform;
 	}
@@ -138,45 +131,42 @@ B3_ATTRIBUTE_ALIGNED16 (struct)	b3SolverBody
 	{
 		return m_worldTransform;
 	}
-	
-	B3_FORCE_INLINE void	getVelocityInLocalPointObsolete(const b3Vector3& rel_pos, b3Vector3& velocity ) const
+
+	B3_FORCE_INLINE void getVelocityInLocalPointObsolete(const b3Vector3& rel_pos, b3Vector3& velocity) const
 	{
 		if (m_originalBody)
-			velocity = m_linearVelocity+m_deltaLinearVelocity + (m_angularVelocity+m_deltaAngularVelocity).cross(rel_pos);
+			velocity = m_linearVelocity + m_deltaLinearVelocity + (m_angularVelocity + m_deltaAngularVelocity).cross(rel_pos);
 		else
-			velocity.setValue(0,0,0);
+			velocity.setValue(0, 0, 0);
 	}
 
-	B3_FORCE_INLINE void	getAngularVelocity(b3Vector3& angVel) const
+	B3_FORCE_INLINE void getAngularVelocity(b3Vector3 & angVel) const
 	{
 		if (m_originalBody)
-			angVel =m_angularVelocity+m_deltaAngularVelocity;
+			angVel = m_angularVelocity + m_deltaAngularVelocity;
 		else
-			angVel.setValue(0,0,0);
+			angVel.setValue(0, 0, 0);
 	}
 
-
 	//Optimization for the iterative solver: avoid calculating constant terms involving inertia, normal, relative position
-	B3_FORCE_INLINE void applyImpulse(const b3Vector3& linearComponent, const b3Vector3& angularComponent,const b3Scalar impulseMagnitude)
+	B3_FORCE_INLINE void applyImpulse(const b3Vector3& linearComponent, const b3Vector3& angularComponent, const b3Scalar impulseMagnitude)
 	{
 		if (m_originalBody)
 		{
-			m_deltaLinearVelocity += linearComponent*impulseMagnitude*m_linearFactor;
-			m_deltaAngularVelocity += angularComponent*(impulseMagnitude*m_angularFactor);
+			m_deltaLinearVelocity += linearComponent * impulseMagnitude * m_linearFactor;
+			m_deltaAngularVelocity += angularComponent * (impulseMagnitude * m_angularFactor);
 		}
 	}
 
-	B3_FORCE_INLINE void internalApplyPushImpulse(const b3Vector3& linearComponent, const b3Vector3& angularComponent,b3Scalar impulseMagnitude)
+	B3_FORCE_INLINE void internalApplyPushImpulse(const b3Vector3& linearComponent, const b3Vector3& angularComponent, b3Scalar impulseMagnitude)
 	{
 		if (m_originalBody)
 		{
-			m_pushVelocity += linearComponent*impulseMagnitude*m_linearFactor;
-			m_turnVelocity += angularComponent*(impulseMagnitude*m_angularFactor);
+			m_pushVelocity += linearComponent * impulseMagnitude * m_linearFactor;
+			m_turnVelocity += angularComponent * (impulseMagnitude * m_angularFactor);
 		}
 	}
 
-
-
 	const b3Vector3& getDeltaLinearVelocity() const
 	{
 		return m_deltaLinearVelocity;
@@ -187,20 +177,19 @@ B3_ATTRIBUTE_ALIGNED16 (struct)	b3SolverBody
 		return m_deltaAngularVelocity;
 	}
 
-	const b3Vector3& getPushVelocity() const 
+	const b3Vector3& getPushVelocity() const
 	{
 		return m_pushVelocity;
 	}
 
-	const b3Vector3& getTurnVelocity() const 
+	const b3Vector3& getTurnVelocity() const
 	{
 		return m_turnVelocity;
 	}
 
-
 	////////////////////////////////////////////////
 	///some internal methods, don't use them
-		
+
 	b3Vector3& internalGetDeltaLinearVelocity()
 	{
 		return m_deltaLinearVelocity;
@@ -225,7 +214,7 @@ B3_ATTRIBUTE_ALIGNED16 (struct)	b3SolverBody
 	{
 		m_invMass = invMass;
 	}
-	
+
 	b3Vector3& internalGetPushVelocity()
 	{
 		return m_pushVelocity;
@@ -236,67 +225,57 @@ B3_ATTRIBUTE_ALIGNED16 (struct)	b3SolverBody
 		return m_turnVelocity;
 	}
 
-	B3_FORCE_INLINE void	internalGetVelocityInLocalPointObsolete(const b3Vector3& rel_pos, b3Vector3& velocity ) const
+	B3_FORCE_INLINE void internalGetVelocityInLocalPointObsolete(const b3Vector3& rel_pos, b3Vector3& velocity) const
 	{
-		velocity = m_linearVelocity+m_deltaLinearVelocity + (m_angularVelocity+m_deltaAngularVelocity).cross(rel_pos);
+		velocity = m_linearVelocity + m_deltaLinearVelocity + (m_angularVelocity + m_deltaAngularVelocity).cross(rel_pos);
 	}
 
-	B3_FORCE_INLINE void	internalGetAngularVelocity(b3Vector3& angVel) const
+	B3_FORCE_INLINE void internalGetAngularVelocity(b3Vector3 & angVel) const
 	{
-		angVel = m_angularVelocity+m_deltaAngularVelocity;
+		angVel = m_angularVelocity + m_deltaAngularVelocity;
 	}
 
-
 	//Optimization for the iterative solver: avoid calculating constant terms involving inertia, normal, relative position
-	B3_FORCE_INLINE void internalApplyImpulse(const b3Vector3& linearComponent, const b3Vector3& angularComponent,const b3Scalar impulseMagnitude)
+	B3_FORCE_INLINE void internalApplyImpulse(const b3Vector3& linearComponent, const b3Vector3& angularComponent, const b3Scalar impulseMagnitude)
 	{
 		//if (m_originalBody)
 		{
-			m_deltaLinearVelocity += linearComponent*impulseMagnitude*m_linearFactor;
-			m_deltaAngularVelocity += angularComponent*(impulseMagnitude*m_angularFactor);
+			m_deltaLinearVelocity += linearComponent * impulseMagnitude * m_linearFactor;
+			m_deltaAngularVelocity += angularComponent * (impulseMagnitude * m_angularFactor);
 		}
 	}
-		
-	
-	
 
-	void	writebackVelocity()
+	void writebackVelocity()
 	{
 		//if (m_originalBody>=0)
 		{
-			m_linearVelocity +=m_deltaLinearVelocity;
+			m_linearVelocity += m_deltaLinearVelocity;
 			m_angularVelocity += m_deltaAngularVelocity;
-			
+
 			//m_originalBody->setCompanionId(-1);
 		}
 	}
 
-
-	void	writebackVelocityAndTransform(b3Scalar timeStep, b3Scalar splitImpulseTurnErp)
+	void writebackVelocityAndTransform(b3Scalar timeStep, b3Scalar splitImpulseTurnErp)
 	{
-        (void) timeStep;
+		(void)timeStep;
 		if (m_originalBody)
 		{
 			m_linearVelocity += m_deltaLinearVelocity;
 			m_angularVelocity += m_deltaAngularVelocity;
-			
+
 			//correct the position/orientation based on push/turn recovery
 			b3Transform newTransform;
-			if (m_pushVelocity[0]!=0.f || m_pushVelocity[1]!=0 || m_pushVelocity[2]!=0 || m_turnVelocity[0]!=0.f || m_turnVelocity[1]!=0 || m_turnVelocity[2]!=0)
+			if (m_pushVelocity[0] != 0.f || m_pushVelocity[1] != 0 || m_pushVelocity[2] != 0 || m_turnVelocity[0] != 0.f || m_turnVelocity[1] != 0 || m_turnVelocity[2] != 0)
 			{
-			//	b3Quaternion orn = m_worldTransform.getRotation();
-				b3TransformUtil::integrateTransform(m_worldTransform,m_pushVelocity,m_turnVelocity*splitImpulseTurnErp,timeStep,newTransform);
+				//	b3Quaternion orn = m_worldTransform.getRotation();
+				b3TransformUtil::integrateTransform(m_worldTransform, m_pushVelocity, m_turnVelocity * splitImpulseTurnErp, timeStep, newTransform);
 				m_worldTransform = newTransform;
 			}
 			//m_worldTransform.setRotation(orn);
 			//m_originalBody->setCompanionId(-1);
 		}
 	}
-	
-
-
 };
 
-#endif //B3_SOLVER_BODY_H
-
-
+#endif  //B3_SOLVER_BODY_H
diff --git a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3SolverConstraint.h b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3SolverConstraint.h
index bce83d46089..4927ae42884 100644
--- a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3SolverConstraint.h
+++ b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3SolverConstraint.h
@@ -16,7 +16,6 @@ subject to the following restrictions:
 #ifndef B3_SOLVER_CONSTRAINT_H
 #define B3_SOLVER_CONSTRAINT_H
 
-
 #include "Bullet3Common/b3Vector3.h"
 #include "Bullet3Common/b3Matrix3x3.h"
 //#include "b3JacobianEntry.h"
@@ -25,56 +24,50 @@ subject to the following restrictions:
 //#define NO_FRICTION_TANGENTIALS 1
 #include "b3SolverBody.h"
 
-
 ///1D constraint along a normal axis between bodyA and bodyB. It can be combined to solve contact and friction constraints.
-B3_ATTRIBUTE_ALIGNED16 (struct)	b3SolverConstraint
+B3_ATTRIBUTE_ALIGNED16(struct)
+b3SolverConstraint
 {
 	B3_DECLARE_ALIGNED_ALLOCATOR();
 
-	b3Vector3		m_relpos1CrossNormal;
-	b3Vector3		m_contactNormal;
+	b3Vector3 m_relpos1CrossNormal;
+	b3Vector3 m_contactNormal;
 
-	b3Vector3		m_relpos2CrossNormal;
+	b3Vector3 m_relpos2CrossNormal;
 	//b3Vector3		m_contactNormal2;//usually m_contactNormal2 == -m_contactNormal
 
-	b3Vector3		m_angularComponentA;
-	b3Vector3		m_angularComponentB;
-	
-	mutable b3SimdScalar	m_appliedPushImpulse;
-	mutable b3SimdScalar	m_appliedImpulse;
+	b3Vector3 m_angularComponentA;
+	b3Vector3 m_angularComponentB;
+
+	mutable b3SimdScalar m_appliedPushImpulse;
+	mutable b3SimdScalar m_appliedImpulse;
 	int m_padding1;
 	int m_padding2;
-	b3Scalar	m_friction;
-	b3Scalar	m_jacDiagABInv;
-	b3Scalar		m_rhs;
-	b3Scalar		m_cfm;
-	
-    b3Scalar		m_lowerLimit;
-	b3Scalar		m_upperLimit;
-	b3Scalar		m_rhsPenetration;
-    union
-	{
-		void*		m_originalContactPoint;
-		b3Scalar	m_unusedPadding4;
+	b3Scalar m_friction;
+	b3Scalar m_jacDiagABInv;
+	b3Scalar m_rhs;
+	b3Scalar m_cfm;
+
+	b3Scalar m_lowerLimit;
+	b3Scalar m_upperLimit;
+	b3Scalar m_rhsPenetration;
+	union {
+		void* m_originalContactPoint;
+		b3Scalar m_unusedPadding4;
 	};
 
-	int	m_overrideNumSolverIterations;
-    int			m_frictionIndex;
+	int m_overrideNumSolverIterations;
+	int m_frictionIndex;
 	int m_solverBodyIdA;
 	int m_solverBodyIdB;
 
-    
-	enum		b3SolverConstraintType
+	enum b3SolverConstraintType
 	{
 		B3_SOLVER_CONTACT_1D = 0,
 		B3_SOLVER_FRICTION_1D
 	};
 };
 
-typedef b3AlignedObjectArray<b3SolverConstraint>	b3ConstraintArray;
-
-
-#endif //B3_SOLVER_CONSTRAINT_H
-
-
+typedef b3AlignedObjectArray<b3SolverConstraint> b3ConstraintArray;
 
+#endif  //B3_SOLVER_CONSTRAINT_H
diff --git a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3TypedConstraint.cpp b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3TypedConstraint.cpp
index 699c481d64a..885e277d8c2 100644
--- a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3TypedConstraint.cpp
+++ b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3TypedConstraint.cpp
@@ -13,53 +13,46 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "b3TypedConstraint.h"
 //#include "Bullet3Common/b3Serializer.h"
 
-
 #define B3_DEFAULT_DEBUGDRAW_SIZE b3Scalar(0.3f)
 
-
-
-b3TypedConstraint::b3TypedConstraint(b3TypedConstraintType type, int rbA,int rbB)
-:b3TypedObject(type),
-m_userConstraintType(-1),
-m_userConstraintPtr((void*)-1),
-m_breakingImpulseThreshold(B3_INFINITY),
-m_isEnabled(true),
-m_needsFeedback(false),
-m_overrideNumSolverIterations(-1),
-m_rbA(rbA),
-m_rbB(rbB),
-m_appliedImpulse(b3Scalar(0.)),
-m_dbgDrawSize(B3_DEFAULT_DEBUGDRAW_SIZE),
-m_jointFeedback(0)
+b3TypedConstraint::b3TypedConstraint(b3TypedConstraintType type, int rbA, int rbB)
+	: b3TypedObject(type),
+	  m_userConstraintType(-1),
+	  m_userConstraintPtr((void*)-1),
+	  m_breakingImpulseThreshold(B3_INFINITY),
+	  m_isEnabled(true),
+	  m_needsFeedback(false),
+	  m_overrideNumSolverIterations(-1),
+	  m_rbA(rbA),
+	  m_rbB(rbB),
+	  m_appliedImpulse(b3Scalar(0.)),
+	  m_dbgDrawSize(B3_DEFAULT_DEBUGDRAW_SIZE),
+	  m_jointFeedback(0)
 {
 }
 
-
-
-
 b3Scalar b3TypedConstraint::getMotorFactor(b3Scalar pos, b3Scalar lowLim, b3Scalar uppLim, b3Scalar vel, b3Scalar timeFact)
 {
-	if(lowLim > uppLim)
+	if (lowLim > uppLim)
 	{
 		return b3Scalar(1.0f);
 	}
-	else if(lowLim == uppLim)
+	else if (lowLim == uppLim)
 	{
 		return b3Scalar(0.0f);
 	}
 	b3Scalar lim_fact = b3Scalar(1.0f);
 	b3Scalar delta_max = vel / timeFact;
-	if(delta_max < b3Scalar(0.0f))
+	if (delta_max < b3Scalar(0.0f))
 	{
-		if((pos >= lowLim) && (pos < (lowLim - delta_max)))
+		if ((pos >= lowLim) && (pos < (lowLim - delta_max)))
 		{
 			lim_fact = (lowLim - pos) / delta_max;
 		}
-		else if(pos  < lowLim)
+		else if (pos < lowLim)
 		{
 			lim_fact = b3Scalar(0.0f);
 		}
@@ -68,13 +61,13 @@ b3Scalar b3TypedConstraint::getMotorFactor(b3Scalar pos, b3Scalar lowLim, b3Scal
 			lim_fact = b3Scalar(1.0f);
 		}
 	}
-	else if(delta_max > b3Scalar(0.0f))
+	else if (delta_max > b3Scalar(0.0f))
 	{
-		if((pos <= uppLim) && (pos > (uppLim - delta_max)))
+		if ((pos <= uppLim) && (pos > (uppLim - delta_max)))
 		{
 			lim_fact = (uppLim - pos) / delta_max;
 		}
-		else if(pos  > uppLim)
+		else if (pos > uppLim)
 		{
 			lim_fact = b3Scalar(0.0f);
 		}
@@ -85,18 +78,16 @@ b3Scalar b3TypedConstraint::getMotorFactor(b3Scalar pos, b3Scalar lowLim, b3Scal
 	}
 	else
 	{
-			lim_fact = b3Scalar(0.0f);
+		lim_fact = b3Scalar(0.0f);
 	}
 	return lim_fact;
 }
 
-
-
 void b3AngularLimit::set(b3Scalar low, b3Scalar high, b3Scalar _softness, b3Scalar _biasFactor, b3Scalar _relaxationFactor)
 {
 	m_halfRange = (high - low) / 2.0f;
 	m_center = b3NormalizeAngle(low + m_halfRange);
-	m_softness =  _softness;
+	m_softness = _softness;
 	m_biasFactor = _biasFactor;
 	m_relaxationFactor = _relaxationFactor;
 }
@@ -113,7 +104,7 @@ void b3AngularLimit::test(const b3Scalar angle)
 		if (deviation < -m_halfRange)
 		{
 			m_solveLimit = true;
-			m_correction = - (deviation + m_halfRange);
+			m_correction = -(deviation + m_halfRange);
 			m_sign = +1.0f;
 		}
 		else if (deviation > m_halfRange)
@@ -125,7 +116,6 @@ void b3AngularLimit::test(const b3Scalar angle)
 	}
 }
 
-
 b3Scalar b3AngularLimit::getError() const
 {
 	return m_correction * m_sign;
diff --git a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3TypedConstraint.h b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3TypedConstraint.h
index cf9cec0d5ed..f74aec4d3c8 100644
--- a/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3TypedConstraint.h
+++ b/thirdparty/bullet/Bullet3Dynamics/ConstraintSolver/b3TypedConstraint.h
@@ -16,7 +16,6 @@ subject to the following restrictions:
 #ifndef B3_TYPED_CONSTRAINT_H
 #define B3_TYPED_CONSTRAINT_H
 
-
 #include "Bullet3Common/b3Scalar.h"
 #include "b3SolverConstraint.h"
 
@@ -25,7 +24,7 @@ class b3Serializer;
 //Don't change any of the existing enum values, so add enum types at the end for serialization compatibility
 enum b3TypedConstraintType
 {
-	B3_POINT2POINT_CONSTRAINT_TYPE=3,
+	B3_POINT2POINT_CONSTRAINT_TYPE = 3,
 	B3_HINGE_CONSTRAINT_TYPE,
 	B3_CONETWIST_CONSTRAINT_TYPE,
 	B3_D6_CONSTRAINT_TYPE,
@@ -37,92 +36,86 @@ enum b3TypedConstraintType
 	B3_MAX_CONSTRAINT_TYPE
 };
 
-
 enum b3ConstraintParams
 {
-	B3_CONSTRAINT_ERP=1,
+	B3_CONSTRAINT_ERP = 1,
 	B3_CONSTRAINT_STOP_ERP,
 	B3_CONSTRAINT_CFM,
 	B3_CONSTRAINT_STOP_CFM
 };
 
 #if 1
-	#define b3AssertConstrParams(_par) b3Assert(_par) 
+#define b3AssertConstrParams(_par) b3Assert(_par)
 #else
-	#define b3AssertConstrParams(_par)
+#define b3AssertConstrParams(_par)
 #endif
 
-
-B3_ATTRIBUTE_ALIGNED16(struct)	b3JointFeedback
+B3_ATTRIBUTE_ALIGNED16(struct)
+b3JointFeedback
 {
-	b3Vector3	m_appliedForceBodyA;
-	b3Vector3	m_appliedTorqueBodyA;
-	b3Vector3	m_appliedForceBodyB;
-	b3Vector3	m_appliedTorqueBodyB;
+	b3Vector3 m_appliedForceBodyA;
+	b3Vector3 m_appliedTorqueBodyA;
+	b3Vector3 m_appliedForceBodyB;
+	b3Vector3 m_appliedTorqueBodyB;
 };
 
-
 struct b3RigidBodyData;
 
-
 ///TypedConstraint is the baseclass for Bullet constraints and vehicles
-B3_ATTRIBUTE_ALIGNED16(class) b3TypedConstraint : public b3TypedObject
+B3_ATTRIBUTE_ALIGNED16(class)
+b3TypedConstraint : public b3TypedObject
 {
-	int	m_userConstraintType;
+	int m_userConstraintType;
 
-	union
-	{
-		int	m_userConstraintId;
+	union {
+		int m_userConstraintId;
 		void* m_userConstraintPtr;
 	};
 
-	b3Scalar	m_breakingImpulseThreshold;
-	bool		m_isEnabled;
-	bool		m_needsFeedback;
-	int			m_overrideNumSolverIterations;
+	b3Scalar m_breakingImpulseThreshold;
+	bool m_isEnabled;
+	bool m_needsFeedback;
+	int m_overrideNumSolverIterations;
 
-
-	b3TypedConstraint&	operator=(b3TypedConstraint&	other)
+	b3TypedConstraint& operator=(b3TypedConstraint& other)
 	{
 		b3Assert(0);
-		(void) other;
+		(void)other;
 		return *this;
 	}
 
 protected:
-	int				m_rbA;
-	int				m_rbB;
-	b3Scalar	m_appliedImpulse;
-	b3Scalar	m_dbgDrawSize;
-	b3JointFeedback*	m_jointFeedback;
+	int m_rbA;
+	int m_rbB;
+	b3Scalar m_appliedImpulse;
+	b3Scalar m_dbgDrawSize;
+	b3JointFeedback* m_jointFeedback;
 
 	///internal method used by the constraint solver, don't use them directly
 	b3Scalar getMotorFactor(b3Scalar pos, b3Scalar lowLim, b3Scalar uppLim, b3Scalar vel, b3Scalar timeFact);
-	
 
 public:
-
 	B3_DECLARE_ALIGNED_ALLOCATOR();
 
-	virtual ~b3TypedConstraint() {};
-	b3TypedConstraint(b3TypedConstraintType type, int bodyA,int bodyB);
+	virtual ~b3TypedConstraint(){};
+	b3TypedConstraint(b3TypedConstraintType type, int bodyA, int bodyB);
 
-	struct b3ConstraintInfo1 {
-		int m_numConstraintRows,nub;
+	struct b3ConstraintInfo1
+	{
+		int m_numConstraintRows, nub;
 	};
 
-	
-
-	struct b3ConstraintInfo2 {
+	struct b3ConstraintInfo2
+	{
 		// integrator parameters: frames per second (1/stepsize), default error
 		// reduction parameter (0..1).
-		b3Scalar fps,erp;
+		b3Scalar fps, erp;
 
 		// for the first and second body, pointers to two (linear and angular)
 		// n*3 jacobian sub matrices, stored by rows. these matrices will have
 		// been initialized to 0 on entry. if the second body is zero then the
 		// J2xx pointers may be 0.
-		b3Scalar *m_J1linearAxis,*m_J1angularAxis,*m_J2linearAxis,*m_J2angularAxis;
+		b3Scalar *m_J1linearAxis, *m_J1angularAxis, *m_J2linearAxis, *m_J2angularAxis;
 
 		// elements to jump from one row to the next in J's
 		int rowskip;
@@ -130,24 +123,24 @@ public:
 		// right hand sides of the equation J*v = c + cfm * lambda. cfm is the
 		// "constraint force mixing" vector. c is set to zero on entry, cfm is
 		// set to a constant value (typically very small or zero) value on entry.
-		b3Scalar *m_constraintError,*cfm;
+		b3Scalar *m_constraintError, *cfm;
 
 		// lo and hi limits for variables (set to -/+ infinity on entry).
-		b3Scalar *m_lowerLimit,*m_upperLimit;
+		b3Scalar *m_lowerLimit, *m_upperLimit;
 
 		// findex vector for variables. see the LCP solver interface for a
 		// description of what this does. this is set to -1 on entry.
 		// note that the returned indexes are relative to the first index of
 		// the constraint.
-		int *findex;
+		int* findex;
 		// number of solver iterations
 		int m_numIterations;
 
 		//damping of the velocity
-		b3Scalar	m_damping;
+		b3Scalar m_damping;
 	};
 
-	int	getOverrideNumSolverIterations() const
+	int getOverrideNumSolverIterations() const
 	{
 		return m_overrideNumSolverIterations;
 	}
@@ -159,59 +152,55 @@ public:
 		m_overrideNumSolverIterations = overideNumIterations;
 	}
 
-
 	///internal method used by the constraint solver, don't use them directly
-	virtual	void	setupSolverConstraint(b3ConstraintArray& ca, int solverBodyA,int solverBodyB, b3Scalar timeStep)
+	virtual void setupSolverConstraint(b3ConstraintArray & ca, int solverBodyA, int solverBodyB, b3Scalar timeStep)
 	{
-        (void)ca;
-        (void)solverBodyA;
-        (void)solverBodyB;
-        (void)timeStep;
+		(void)ca;
+		(void)solverBodyA;
+		(void)solverBodyB;
+		(void)timeStep;
 	}
-	
-	///internal method used by the constraint solver, don't use them directly
-	virtual void getInfo1 (b3ConstraintInfo1* info,const b3RigidBodyData* bodies)=0;
 
 	///internal method used by the constraint solver, don't use them directly
-	virtual void getInfo2 (b3ConstraintInfo2* info,  const b3RigidBodyData* bodies)=0;
+	virtual void getInfo1(b3ConstraintInfo1 * info, const b3RigidBodyData* bodies) = 0;
 
 	///internal method used by the constraint solver, don't use them directly
-	void	internalSetAppliedImpulse(b3Scalar appliedImpulse)
+	virtual void getInfo2(b3ConstraintInfo2 * info, const b3RigidBodyData* bodies) = 0;
+
+	///internal method used by the constraint solver, don't use them directly
+	void internalSetAppliedImpulse(b3Scalar appliedImpulse)
 	{
 		m_appliedImpulse = appliedImpulse;
 	}
 	///internal method used by the constraint solver, don't use them directly
-	b3Scalar	internalGetAppliedImpulse()
+	b3Scalar internalGetAppliedImpulse()
 	{
 		return m_appliedImpulse;
 	}
 
-
-	b3Scalar	getBreakingImpulseThreshold() const
+	b3Scalar getBreakingImpulseThreshold() const
 	{
-		return 	m_breakingImpulseThreshold;
+		return m_breakingImpulseThreshold;
 	}
 
-	void	setBreakingImpulseThreshold(b3Scalar threshold)
+	void setBreakingImpulseThreshold(b3Scalar threshold)
 	{
 		m_breakingImpulseThreshold = threshold;
 	}
 
-	bool	isEnabled() const
+	bool isEnabled() const
 	{
 		return m_isEnabled;
 	}
 
-	void	setEnabled(bool enabled)
+	void setEnabled(bool enabled)
 	{
-		m_isEnabled=enabled;
+		m_isEnabled = enabled;
 	}
 
-
 	///internal method used by the constraint solver, don't use them directly
-	virtual	void	solveConstraintObsolete(b3SolverBody& /*bodyA*/,b3SolverBody& /*bodyB*/,b3Scalar	/*timeStep*/) {};
+	virtual void solveConstraintObsolete(b3SolverBody& /*bodyA*/, b3SolverBody& /*bodyB*/, b3Scalar /*timeStep*/){};
 
-	
 	int getRigidBodyA() const
 	{
 		return m_rbA;
@@ -221,8 +210,7 @@ public:
 		return m_rbB;
 	}
 
-
-	int getRigidBodyA() 
+	int getRigidBodyA()
 	{
 		return m_rbA;
 	}
@@ -233,15 +221,15 @@ public:
 
 	int getUserConstraintType() const
 	{
-		return m_userConstraintType ;
+		return m_userConstraintType;
 	}
 
-	void	setUserConstraintType(int userConstraintType)
+	void setUserConstraintType(int userConstraintType)
 	{
 		m_userConstraintType = userConstraintType;
 	};
 
-	void	setUserConstraintId(int uid)
+	void setUserConstraintId(int uid)
 	{
 		m_userConstraintId = uid;
 	}
@@ -251,17 +239,17 @@ public:
 		return m_userConstraintId;
 	}
 
-	void	setUserConstraintPtr(void* ptr)
+	void setUserConstraintPtr(void* ptr)
 	{
 		m_userConstraintPtr = ptr;
 	}
 
-	void*	getUserConstraintPtr()
+	void* getUserConstraintPtr()
 	{
 		return m_userConstraintPtr;
 	}
 
-	void	setJointFeedback(b3JointFeedback* jointFeedback)
+	void setJointFeedback(b3JointFeedback * jointFeedback)
 	{
 		m_jointFeedback = jointFeedback;
 	}
@@ -276,37 +264,36 @@ public:
 		return m_jointFeedback;
 	}
 
-
 	int getUid() const
 	{
-		return m_userConstraintId;   
-	} 
+		return m_userConstraintId;
+	}
 
-	bool	needsFeedback() const
+	bool needsFeedback() const
 	{
 		return m_needsFeedback;
 	}
 
 	///enableFeedback will allow to read the applied linear and angular impulse
 	///use getAppliedImpulse, getAppliedLinearImpulse and getAppliedAngularImpulse to read feedback information
-	void	enableFeedback(bool needsFeedback)
+	void enableFeedback(bool needsFeedback)
 	{
 		m_needsFeedback = needsFeedback;
 	}
 
-	///getAppliedImpulse is an estimated total applied impulse. 
+	///getAppliedImpulse is an estimated total applied impulse.
 	///This feedback could be used to determine breaking constraints or playing sounds.
-	b3Scalar	getAppliedImpulse() const
+	b3Scalar getAppliedImpulse() const
 	{
 		b3Assert(m_needsFeedback);
 		return m_appliedImpulse;
 	}
 
-	b3TypedConstraintType getConstraintType () const
+	b3TypedConstraintType getConstraintType() const
 	{
 		return b3TypedConstraintType(m_objectType);
 	}
-	
+
 	void setDbgDrawSize(b3Scalar dbgDrawSize)
 	{
 		m_dbgDrawSize = dbgDrawSize;
@@ -316,35 +303,34 @@ public:
 		return m_dbgDrawSize;
 	}
 
-	///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5). 
+	///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5).
 	///If no axis is provided, it uses the default axis for this constraint.
-	virtual	void	setParam(int num, b3Scalar value, int axis = -1) = 0;
+	virtual void setParam(int num, b3Scalar value, int axis = -1) = 0;
 
 	///return the local value of parameter
-	virtual	b3Scalar getParam(int num, int axis = -1) const = 0;
-	
-//	virtual	int	calculateSerializeBufferSize() const;
+	virtual b3Scalar getParam(int num, int axis = -1) const = 0;
+
+	//	virtual	int	calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
 	//virtual	const char*	serialize(void* dataBuffer, b3Serializer* serializer) const;
-
 };
 
-// returns angle in range [-B3_2_PI, B3_2_PI], closest to one of the limits 
+// returns angle in range [-B3_2_PI, B3_2_PI], closest to one of the limits
 // all arguments should be normalized angles (i.e. in range [-B3_PI, B3_PI])
 B3_FORCE_INLINE b3Scalar b3AdjustAngleToLimits(b3Scalar angleInRadians, b3Scalar angleLowerLimitInRadians, b3Scalar angleUpperLimitInRadians)
 {
-	if(angleLowerLimitInRadians >= angleUpperLimitInRadians)
+	if (angleLowerLimitInRadians >= angleUpperLimitInRadians)
 	{
 		return angleInRadians;
 	}
-	else if(angleInRadians < angleLowerLimitInRadians)
+	else if (angleInRadians < angleLowerLimitInRadians)
 	{
 		b3Scalar diffLo = b3Fabs(b3NormalizeAngle(angleLowerLimitInRadians - angleInRadians));
 		b3Scalar diffHi = b3Fabs(b3NormalizeAngle(angleUpperLimitInRadians - angleInRadians));
 		return (diffLo < diffHi) ? angleInRadians : (angleInRadians + B3_2_PI);
 	}
-	else if(angleInRadians > angleUpperLimitInRadians)
+	else if (angleInRadians > angleUpperLimitInRadians)
 	{
 		b3Scalar diffHi = b3Fabs(b3NormalizeAngle(angleInRadians - angleUpperLimitInRadians));
 		b3Scalar diffLo = b3Fabs(b3NormalizeAngle(angleInRadians - angleLowerLimitInRadians));
@@ -356,6 +342,7 @@ B3_FORCE_INLINE b3Scalar b3AdjustAngleToLimits(b3Scalar angleInRadians, b3Scalar
 	}
 }
 
+// clang-format off
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
 struct	b3TypedConstraintData
 {
@@ -379,17 +366,18 @@ struct	b3TypedConstraintData
 	
 };
 
+// clang-format on
+
 /*B3_FORCE_INLINE	int	b3TypedConstraint::calculateSerializeBufferSize() const
 {
 	return sizeof(b3TypedConstraintData);
 }
 */
 
-
 class b3AngularLimit
 {
 private:
-	b3Scalar 
+	b3Scalar
 		m_center,
 		m_halfRange,
 		m_softness,
@@ -404,15 +392,16 @@ private:
 public:
 	/// Default constructor initializes limit as inactive, allowing free constraint movement
 	b3AngularLimit()
-		:m_center(0.0f),
-		m_halfRange(-1.0f),
-		m_softness(0.9f),
-		m_biasFactor(0.3f),
-		m_relaxationFactor(1.0f),
-		m_correction(0.0f),
-		m_sign(0.0f),
-		m_solveLimit(false)
-	{}
+		: m_center(0.0f),
+		  m_halfRange(-1.0f),
+		  m_softness(0.9f),
+		  m_biasFactor(0.3f),
+		  m_relaxationFactor(1.0f),
+		  m_correction(0.0f),
+		  m_sign(0.0f),
+		  m_solveLimit(false)
+	{
+	}
 
 	/// Sets all limit's parameters.
 	/// When low > high limit becomes inactive.
@@ -441,13 +430,13 @@ public:
 		return m_relaxationFactor;
 	}
 
-	/// Returns correction value evaluated when test() was invoked 
+	/// Returns correction value evaluated when test() was invoked
 	inline b3Scalar getCorrection() const
 	{
 		return m_correction;
 	}
 
-	/// Returns sign value evaluated when test() was invoked 
+	/// Returns sign value evaluated when test() was invoked
 	inline b3Scalar getSign() const
 	{
 		return m_sign;
@@ -475,9 +464,6 @@ public:
 	b3Scalar getLow() const;
 
 	b3Scalar getHigh() const;
-
 };
 
-
-
-#endif //B3_TYPED_CONSTRAINT_H
+#endif  //B3_TYPED_CONSTRAINT_H
diff --git a/thirdparty/bullet/Bullet3Dynamics/b3CpuRigidBodyPipeline.cpp b/thirdparty/bullet/Bullet3Dynamics/b3CpuRigidBodyPipeline.cpp
index fbc84cc28d6..f1080d9d5e1 100644
--- a/thirdparty/bullet/Bullet3Dynamics/b3CpuRigidBodyPipeline.cpp
+++ b/thirdparty/bullet/Bullet3Dynamics/b3CpuRigidBodyPipeline.cpp
@@ -11,7 +11,6 @@
 #include "Bullet3Dynamics/shared/b3ContactConstraint4.h"
 #include "Bullet3Dynamics/shared/b3Inertia.h"
 
-
 struct b3CpuRigidBodyPipelineInternalData
 {
 	b3AlignedObjectArray<b3RigidBodyData> m_rigidBodies;
@@ -22,7 +21,6 @@ struct b3CpuRigidBodyPipelineInternalData
 	b3CpuNarrowPhase* m_np;
 	b3Config m_config;
 };
-	
 
 b3CpuRigidBodyPipeline::b3CpuRigidBodyPipeline(class b3CpuNarrowPhase* narrowphase, struct b3DynamicBvhBroadphase* broadphaseDbvt, const b3Config& config)
 {
@@ -39,49 +37,43 @@ b3CpuRigidBodyPipeline::~b3CpuRigidBodyPipeline()
 
 void b3CpuRigidBodyPipeline::updateAabbWorldSpace()
 {
-
-	for (int i=0;i<this->getNumBodies();i++)
+	for (int i = 0; i < this->getNumBodies(); i++)
 	{
 		b3RigidBodyData* body = &m_data->m_rigidBodies[i];
 		b3Float4 position = body->m_pos;
-		b3Quat	orientation = body->m_quat;
+		b3Quat orientation = body->m_quat;
 
 		int collidableIndex = body->m_collidableIdx;
 		b3Collidable& collidable = m_data->m_np->getCollidableCpu(collidableIndex);
 		int shapeIndex = collidable.m_shapeIndex;
-		
-		if (shapeIndex>=0)
-		{
-			
 
+		if (shapeIndex >= 0)
+		{
 			b3Aabb localAabb = m_data->m_np->getLocalSpaceAabb(shapeIndex);
 			b3Aabb& worldAabb = m_data->m_aabbWorldSpace[i];
-			float margin=0.f;
-			b3TransformAabb2(localAabb.m_minVec,localAabb.m_maxVec,margin,position,orientation,&worldAabb.m_minVec,&worldAabb.m_maxVec);
-			m_data->m_bp->setAabb(i,worldAabb.m_minVec,worldAabb.m_maxVec,0);
+			float margin = 0.f;
+			b3TransformAabb2(localAabb.m_minVec, localAabb.m_maxVec, margin, position, orientation, &worldAabb.m_minVec, &worldAabb.m_maxVec);
+			m_data->m_bp->setAabb(i, worldAabb.m_minVec, worldAabb.m_maxVec, 0);
 		}
 	}
 }
 
-void	b3CpuRigidBodyPipeline::computeOverlappingPairs()
+void b3CpuRigidBodyPipeline::computeOverlappingPairs()
 {
 	int numPairs = m_data->m_bp->getOverlappingPairCache()->getNumOverlappingPairs();
 	m_data->m_bp->calculateOverlappingPairs();
 	numPairs = m_data->m_bp->getOverlappingPairCache()->getNumOverlappingPairs();
-	printf("numPairs=%d\n",numPairs);
+	printf("numPairs=%d\n", numPairs);
 }
 
 void b3CpuRigidBodyPipeline::computeContactPoints()
 {
-	
 	b3AlignedObjectArray<b3Int4>& pairs = m_data->m_bp->getOverlappingPairCache()->getOverlappingPairArray();
-	
-	m_data->m_np->computeContacts(pairs,m_data->m_aabbWorldSpace, m_data->m_rigidBodies);
 
+	m_data->m_np->computeContacts(pairs, m_data->m_aabbWorldSpace, m_data->m_rigidBodies);
 }
-void	b3CpuRigidBodyPipeline::stepSimulation(float deltaTime)
+void b3CpuRigidBodyPipeline::stepSimulation(float deltaTime)
 {
-	
 	//update world space aabb's
 	updateAabbWorldSpace();
 
@@ -92,73 +84,71 @@ void	b3CpuRigidBodyPipeline::stepSimulation(float deltaTime)
 	computeContactPoints();
 
 	//solve contacts
-	
+
 	//update transforms
 	integrate(deltaTime);
-	
-	
 }
 
-
-static	inline	float b3CalcRelVel(const b3Vector3& l0, const b3Vector3& l1, const b3Vector3& a0, const b3Vector3& a1, 
-					 const b3Vector3& linVel0, const b3Vector3& angVel0, const b3Vector3& linVel1, const b3Vector3& angVel1)
+static inline float b3CalcRelVel(const b3Vector3& l0, const b3Vector3& l1, const b3Vector3& a0, const b3Vector3& a1,
+								 const b3Vector3& linVel0, const b3Vector3& angVel0, const b3Vector3& linVel1, const b3Vector3& angVel1)
 {
 	return b3Dot(l0, linVel0) + b3Dot(a0, angVel0) + b3Dot(l1, linVel1) + b3Dot(a1, angVel1);
 }
 
-
-static	inline	void b3SetLinearAndAngular(const b3Vector3& n, const b3Vector3& r0, const b3Vector3& r1,
-							 b3Vector3& linear, b3Vector3& angular0, b3Vector3& angular1)
+static inline void b3SetLinearAndAngular(const b3Vector3& n, const b3Vector3& r0, const b3Vector3& r1,
+										 b3Vector3& linear, b3Vector3& angular0, b3Vector3& angular1)
 {
 	linear = -n;
 	angular0 = -b3Cross(r0, n);
 	angular1 = b3Cross(r1, n);
 }
 
-
-
-static inline void b3SolveContact(b3ContactConstraint4& cs, 
-	const b3Vector3& posA, b3Vector3& linVelA, b3Vector3& angVelA, float invMassA, const b3Matrix3x3& invInertiaA,
-	const b3Vector3& posB, b3Vector3& linVelB, b3Vector3& angVelB, float invMassB, const b3Matrix3x3& invInertiaB, 
-	float maxRambdaDt[4], float minRambdaDt[4])
+static inline void b3SolveContact(b3ContactConstraint4& cs,
+								  const b3Vector3& posA, b3Vector3& linVelA, b3Vector3& angVelA, float invMassA, const b3Matrix3x3& invInertiaA,
+								  const b3Vector3& posB, b3Vector3& linVelB, b3Vector3& angVelB, float invMassB, const b3Matrix3x3& invInertiaB,
+								  float maxRambdaDt[4], float minRambdaDt[4])
 {
+	b3Vector3 dLinVelA;
+	dLinVelA.setZero();
+	b3Vector3 dAngVelA;
+	dAngVelA.setZero();
+	b3Vector3 dLinVelB;
+	dLinVelB.setZero();
+	b3Vector3 dAngVelB;
+	dAngVelB.setZero();
 
-	b3Vector3 dLinVelA; dLinVelA.setZero();
-	b3Vector3 dAngVelA; dAngVelA.setZero();
-	b3Vector3 dLinVelB; dLinVelB.setZero();
-	b3Vector3 dAngVelB; dAngVelB.setZero();
-
-	for(int ic=0; ic<4; ic++)
+	for (int ic = 0; ic < 4; ic++)
 	{
 		//	dont necessary because this makes change to 0
-		if( cs.m_jacCoeffInv[ic] == 0.f ) continue;
+		if (cs.m_jacCoeffInv[ic] == 0.f) continue;
 
 		{
 			b3Vector3 angular0, angular1, linear;
 			b3Vector3 r0 = cs.m_worldPos[ic] - (b3Vector3&)posA;
 			b3Vector3 r1 = cs.m_worldPos[ic] - (b3Vector3&)posB;
-			b3SetLinearAndAngular( (const b3Vector3 &)-cs.m_linear, (const b3Vector3 &)r0, (const b3Vector3 &)r1, linear, angular0, angular1 );
+			b3SetLinearAndAngular((const b3Vector3&)-cs.m_linear, (const b3Vector3&)r0, (const b3Vector3&)r1, linear, angular0, angular1);
 
-			float rambdaDt = b3CalcRelVel((const b3Vector3 &)cs.m_linear,(const b3Vector3 &) -cs.m_linear, angular0, angular1,
-				linVelA, angVelA, linVelB, angVelB ) + cs.m_b[ic];
+			float rambdaDt = b3CalcRelVel((const b3Vector3&)cs.m_linear, (const b3Vector3&)-cs.m_linear, angular0, angular1,
+										  linVelA, angVelA, linVelB, angVelB) +
+							 cs.m_b[ic];
 			rambdaDt *= cs.m_jacCoeffInv[ic];
 
 			{
 				float prevSum = cs.m_appliedRambdaDt[ic];
 				float updated = prevSum;
 				updated += rambdaDt;
-				updated = b3Max( updated, minRambdaDt[ic] );
-				updated = b3Min( updated, maxRambdaDt[ic] );
+				updated = b3Max(updated, minRambdaDt[ic]);
+				updated = b3Min(updated, maxRambdaDt[ic]);
 				rambdaDt = updated - prevSum;
 				cs.m_appliedRambdaDt[ic] = updated;
 			}
 
-			b3Vector3 linImp0 = invMassA*linear*rambdaDt;
-			b3Vector3 linImp1 = invMassB*(-linear)*rambdaDt;
-			b3Vector3 angImp0 = (invInertiaA* angular0)*rambdaDt;
-			b3Vector3 angImp1 = (invInertiaB* angular1)*rambdaDt;
+			b3Vector3 linImp0 = invMassA * linear * rambdaDt;
+			b3Vector3 linImp1 = invMassB * (-linear) * rambdaDt;
+			b3Vector3 angImp0 = (invInertiaA * angular0) * rambdaDt;
+			b3Vector3 angImp1 = (invInertiaB * angular1) * rambdaDt;
 #ifdef _WIN32
-            b3Assert(_finite(linImp0.getX()));
+			b3Assert(_finite(linImp0.getX()));
 			b3Assert(_finite(linImp1.getX()));
 #endif
 			{
@@ -169,53 +159,46 @@ static inline void b3SolveContact(b3ContactConstraint4& cs,
 			}
 		}
 	}
-
-
 }
 
-
-
-
-
-static inline void b3SolveFriction(b3ContactConstraint4& cs, 
-		const b3Vector3& posA, b3Vector3& linVelA, b3Vector3& angVelA, float invMassA, const b3Matrix3x3& invInertiaA,
-		const b3Vector3& posB, b3Vector3& linVelB, b3Vector3& angVelB, float invMassB, const b3Matrix3x3& invInertiaB, 
-		float maxRambdaDt[4], float minRambdaDt[4])
+static inline void b3SolveFriction(b3ContactConstraint4& cs,
+								   const b3Vector3& posA, b3Vector3& linVelA, b3Vector3& angVelA, float invMassA, const b3Matrix3x3& invInertiaA,
+								   const b3Vector3& posB, b3Vector3& linVelB, b3Vector3& angVelB, float invMassB, const b3Matrix3x3& invInertiaB,
+								   float maxRambdaDt[4], float minRambdaDt[4])
 {
-
-	if( cs.m_fJacCoeffInv[0] == 0 && cs.m_fJacCoeffInv[0] == 0 ) return;
+	if (cs.m_fJacCoeffInv[0] == 0 && cs.m_fJacCoeffInv[0] == 0) return;
 	const b3Vector3& center = (const b3Vector3&)cs.m_center;
 
 	b3Vector3 n = -(const b3Vector3&)cs.m_linear;
 
 	b3Vector3 tangent[2];
 
-	b3PlaneSpace1 (n, tangent[0],tangent[1]);
+	b3PlaneSpace1(n, tangent[0], tangent[1]);
 
 	b3Vector3 angular0, angular1, linear;
 	b3Vector3 r0 = center - posA;
 	b3Vector3 r1 = center - posB;
-	for(int i=0; i<2; i++)
+	for (int i = 0; i < 2; i++)
 	{
-		b3SetLinearAndAngular( tangent[i], r0, r1, linear, angular0, angular1 );
+		b3SetLinearAndAngular(tangent[i], r0, r1, linear, angular0, angular1);
 		float rambdaDt = b3CalcRelVel(linear, -linear, angular0, angular1,
-			linVelA, angVelA, linVelB, angVelB );
+									  linVelA, angVelA, linVelB, angVelB);
 		rambdaDt *= cs.m_fJacCoeffInv[i];
 
-			{
-				float prevSum = cs.m_fAppliedRambdaDt[i];
-				float updated = prevSum;
-				updated += rambdaDt;
-				updated = b3Max( updated, minRambdaDt[i] );
-				updated = b3Min( updated, maxRambdaDt[i] );
-				rambdaDt = updated - prevSum;
-				cs.m_fAppliedRambdaDt[i] = updated;
-			}
+		{
+			float prevSum = cs.m_fAppliedRambdaDt[i];
+			float updated = prevSum;
+			updated += rambdaDt;
+			updated = b3Max(updated, minRambdaDt[i]);
+			updated = b3Min(updated, maxRambdaDt[i]);
+			rambdaDt = updated - prevSum;
+			cs.m_fAppliedRambdaDt[i] = updated;
+		}
 
-		b3Vector3 linImp0 = invMassA*linear*rambdaDt;
-		b3Vector3 linImp1 = invMassB*(-linear)*rambdaDt;
-		b3Vector3 angImp0 = (invInertiaA* angular0)*rambdaDt;
-		b3Vector3 angImp1 = (invInertiaB* angular1)*rambdaDt;
+		b3Vector3 linImp0 = invMassA * linear * rambdaDt;
+		b3Vector3 linImp1 = invMassB * (-linear) * rambdaDt;
+		b3Vector3 angImp0 = (invInertiaA * angular0) * rambdaDt;
+		b3Vector3 angImp1 = (invInertiaB * angular1) * rambdaDt;
 #ifdef _WIN32
 		b3Assert(_finite(linImp0.getX()));
 		b3Assert(_finite(linImp1.getX()));
@@ -226,57 +209,45 @@ static inline void b3SolveFriction(b3ContactConstraint4& cs,
 		angVelB += angImp1;
 	}
 
-	{	//	angular damping for point constraint
-		b3Vector3 ab = ( posB - posA ).normalized();
-		b3Vector3 ac = ( center - posA ).normalized();
-		if( b3Dot( ab, ac ) > 0.95f || (invMassA == 0.f || invMassB == 0.f))
+	{  //	angular damping for point constraint
+		b3Vector3 ab = (posB - posA).normalized();
+		b3Vector3 ac = (center - posA).normalized();
+		if (b3Dot(ab, ac) > 0.95f || (invMassA == 0.f || invMassB == 0.f))
 		{
-			float angNA = b3Dot( n, angVelA );
-			float angNB = b3Dot( n, angVelB );
+			float angNA = b3Dot(n, angVelA);
+			float angNB = b3Dot(n, angVelB);
 
-			angVelA -= (angNA*0.1f)*n;
-			angVelB -= (angNB*0.1f)*n;
+			angVelA -= (angNA * 0.1f) * n;
+			angVelB -= (angNB * 0.1f) * n;
 		}
 	}
-
 }
 
-
-
-
-
-struct b3SolveTask// : public ThreadPool::Task
+struct b3SolveTask  // : public ThreadPool::Task
 {
-	b3SolveTask(b3AlignedObjectArray<b3RigidBodyData>& bodies,  
-				b3AlignedObjectArray<b3Inertia>& shapes, 
+	b3SolveTask(b3AlignedObjectArray<b3RigidBodyData>& bodies,
+				b3AlignedObjectArray<b3Inertia>& shapes,
 				b3AlignedObjectArray<b3ContactConstraint4>& constraints,
 				int start, int nConstraints,
 				int maxNumBatches,
-				b3AlignedObjectArray<int>* wgUsedBodies, int curWgidx
-				)
-		: m_bodies( bodies ), m_shapes( shapes ), m_constraints( constraints ), 
-		m_wgUsedBodies(wgUsedBodies),m_curWgidx(curWgidx),
-m_start( start ), 
-	m_nConstraints( nConstraints ),
-	m_solveFriction( true ),
-	m_maxNumBatches(maxNumBatches)
-	{}
+				b3AlignedObjectArray<int>* wgUsedBodies, int curWgidx)
+		: m_bodies(bodies), m_shapes(shapes), m_constraints(constraints), m_wgUsedBodies(wgUsedBodies), m_curWgidx(curWgidx), m_start(start), m_nConstraints(nConstraints), m_solveFriction(true), m_maxNumBatches(maxNumBatches)
+	{
+	}
 
-	unsigned short int getType(){ return 0; }
+	unsigned short int getType() { return 0; }
 
 	void run(int tIdx)
 	{
 		b3AlignedObjectArray<int> usedBodies;
 		//printf("run..............\n");
 
-		
-		for (int bb=0;bb<m_maxNumBatches;bb++)
+		for (int bb = 0; bb < m_maxNumBatches; bb++)
 		{
 			usedBodies.resize(0);
-			for(int ic=m_nConstraints-1; ic>=0; ic--)
+			for (int ic = m_nConstraints - 1; ic >= 0; ic--)
 			//for(int ic=0; ic<m_nConstraints; ic++)
 			{
-				
 				int i = m_start + ic;
 				if (m_constraints[i].m_batchIdx != bb)
 					continue;
@@ -298,87 +269,80 @@ m_start( start ),
 					//printf("ic(b)=%d, localBatch=%d\n",ic,localBatch);
 				}
 #endif
-				if (aIdx==10)
+				if (aIdx == 10)
 				{
 					//printf("ic(a)=%d, localBatch=%d\n",ic,localBatch);
 				}
-				if (usedBodies.size()<(aIdx+1))
+				if (usedBodies.size() < (aIdx + 1))
 				{
-					usedBodies.resize(aIdx+1,0);
+					usedBodies.resize(aIdx + 1, 0);
 				}
-		
-				if (usedBodies.size()<(bIdx+1))
+
+				if (usedBodies.size() < (bIdx + 1))
 				{
-					usedBodies.resize(bIdx+1,0);
+					usedBodies.resize(bIdx + 1, 0);
 				}
 
 				if (bodyA.m_invMass)
 				{
-					b3Assert(usedBodies[aIdx]==0);
+					b3Assert(usedBodies[aIdx] == 0);
 					usedBodies[aIdx]++;
 				}
-				
+
 				if (bodyB.m_invMass)
 				{
-					b3Assert(usedBodies[bIdx]==0);
+					b3Assert(usedBodies[bIdx] == 0);
 					usedBodies[bIdx]++;
 				}
-				
 
-				if( !m_solveFriction )
+				if (!m_solveFriction)
 				{
-					float maxRambdaDt[4] = {FLT_MAX,FLT_MAX,FLT_MAX,FLT_MAX};
-					float minRambdaDt[4] = {0.f,0.f,0.f,0.f};
-
-					b3SolveContact( m_constraints[i], (b3Vector3&)bodyA.m_pos, (b3Vector3&)bodyA.m_linVel, (b3Vector3&)bodyA.m_angVel, bodyA.m_invMass, (const b3Matrix3x3 &)m_shapes[aIdx].m_invInertiaWorld, 
-							(b3Vector3&)bodyB.m_pos, (b3Vector3&)bodyB.m_linVel, (b3Vector3&)bodyB.m_angVel, bodyB.m_invMass, (const b3Matrix3x3 &)m_shapes[bIdx].m_invInertiaWorld,
-						maxRambdaDt, minRambdaDt );
+					float maxRambdaDt[4] = {FLT_MAX, FLT_MAX, FLT_MAX, FLT_MAX};
+					float minRambdaDt[4] = {0.f, 0.f, 0.f, 0.f};
 
+					b3SolveContact(m_constraints[i], (b3Vector3&)bodyA.m_pos, (b3Vector3&)bodyA.m_linVel, (b3Vector3&)bodyA.m_angVel, bodyA.m_invMass, (const b3Matrix3x3&)m_shapes[aIdx].m_invInertiaWorld,
+								   (b3Vector3&)bodyB.m_pos, (b3Vector3&)bodyB.m_linVel, (b3Vector3&)bodyB.m_angVel, bodyB.m_invMass, (const b3Matrix3x3&)m_shapes[bIdx].m_invInertiaWorld,
+								   maxRambdaDt, minRambdaDt);
 				}
 				else
 				{
-					float maxRambdaDt[4] = {FLT_MAX,FLT_MAX,FLT_MAX,FLT_MAX};
-					float minRambdaDt[4] = {0.f,0.f,0.f,0.f};
+					float maxRambdaDt[4] = {FLT_MAX, FLT_MAX, FLT_MAX, FLT_MAX};
+					float minRambdaDt[4] = {0.f, 0.f, 0.f, 0.f};
 
 					float sum = 0;
-					for(int j=0; j<4; j++)
+					for (int j = 0; j < 4; j++)
 					{
-						sum +=m_constraints[i].m_appliedRambdaDt[j];
+						sum += m_constraints[i].m_appliedRambdaDt[j];
 					}
 					frictionCoeff = 0.7f;
-					for(int j=0; j<4; j++)
+					for (int j = 0; j < 4; j++)
 					{
-						maxRambdaDt[j] = frictionCoeff*sum;
+						maxRambdaDt[j] = frictionCoeff * sum;
 						minRambdaDt[j] = -maxRambdaDt[j];
 					}
 
-				b3SolveFriction( m_constraints[i], (b3Vector3&)bodyA.m_pos, (b3Vector3&)bodyA.m_linVel, (b3Vector3&)bodyA.m_angVel, bodyA.m_invMass,(const b3Matrix3x3 &) m_shapes[aIdx].m_invInertiaWorld, 
-						(b3Vector3&)bodyB.m_pos, (b3Vector3&)bodyB.m_linVel, (b3Vector3&)bodyB.m_angVel, bodyB.m_invMass,(const b3Matrix3x3 &) m_shapes[bIdx].m_invInertiaWorld,
-						maxRambdaDt, minRambdaDt );
-			
+					b3SolveFriction(m_constraints[i], (b3Vector3&)bodyA.m_pos, (b3Vector3&)bodyA.m_linVel, (b3Vector3&)bodyA.m_angVel, bodyA.m_invMass, (const b3Matrix3x3&)m_shapes[aIdx].m_invInertiaWorld,
+									(b3Vector3&)bodyB.m_pos, (b3Vector3&)bodyB.m_linVel, (b3Vector3&)bodyB.m_angVel, bodyB.m_invMass, (const b3Matrix3x3&)m_shapes[bIdx].m_invInertiaWorld,
+									maxRambdaDt, minRambdaDt);
 				}
 			}
 
 			if (m_wgUsedBodies)
 			{
-				if (m_wgUsedBodies[m_curWgidx].size()<usedBodies.size())
+				if (m_wgUsedBodies[m_curWgidx].size() < usedBodies.size())
 				{
 					m_wgUsedBodies[m_curWgidx].resize(usedBodies.size());
 				}
-				for (int i=0;i<usedBodies.size();i++)
+				for (int i = 0; i < usedBodies.size(); i++)
 				{
 					if (usedBodies[i])
 					{
 						//printf("cell %d uses body %d\n", m_curWgidx,i);
-						m_wgUsedBodies[m_curWgidx][i]=1;
+						m_wgUsedBodies[m_curWgidx][i] = 1;
 					}
 				}
 			}
-
 		}
-
-
-		
 	}
 
 	b3AlignedObjectArray<b3RigidBodyData>& m_bodies;
@@ -397,24 +361,22 @@ void b3CpuRigidBodyPipeline::solveContactConstraints()
 	int m_nIterations = 4;
 
 	b3AlignedObjectArray<b3ContactConstraint4> contactConstraints;
-//	const b3AlignedObjectArray<b3Contact4Data>& contacts = m_data->m_np->getContacts();
+	//	const b3AlignedObjectArray<b3Contact4Data>& contacts = m_data->m_np->getContacts();
 	int n = contactConstraints.size();
 	//convert contacts...
 
-
-	
 	int maxNumBatches = 250;
 
-	for(int iter=0; iter<m_nIterations; iter++)
+	for (int iter = 0; iter < m_nIterations; iter++)
 	{
-		b3SolveTask task( m_data->m_rigidBodies, m_data->m_inertias, contactConstraints, 0, n ,maxNumBatches,0,0);
+		b3SolveTask task(m_data->m_rigidBodies, m_data->m_inertias, contactConstraints, 0, n, maxNumBatches, 0, 0);
 		task.m_solveFriction = false;
 		task.run(0);
 	}
 
-	for(int iter=0; iter<m_nIterations; iter++)
+	for (int iter = 0; iter < m_nIterations; iter++)
 	{
-		b3SolveTask task( m_data->m_rigidBodies, m_data->m_inertias, contactConstraints, 0, n ,maxNumBatches,0,0);
+		b3SolveTask task(m_data->m_rigidBodies, m_data->m_inertias, contactConstraints, 0, n, maxNumBatches, 0, 0);
 		task.m_solveFriction = true;
 		task.run(0);
 	}
@@ -422,53 +384,51 @@ void b3CpuRigidBodyPipeline::solveContactConstraints()
 
 void b3CpuRigidBodyPipeline::integrate(float deltaTime)
 {
-	float angDamping=0.f;
-	b3Vector3 gravityAcceleration=b3MakeVector3(0,-9,0);
+	float angDamping = 0.f;
+	b3Vector3 gravityAcceleration = b3MakeVector3(0, -9, 0);
 
 	//integrate transforms (external forces/gravity should be moved into constraint solver)
-	for (int i=0;i<m_data->m_rigidBodies.size();i++)
+	for (int i = 0; i < m_data->m_rigidBodies.size(); i++)
 	{
-		b3IntegrateTransform(&m_data->m_rigidBodies[i],deltaTime,angDamping,gravityAcceleration);
+		b3IntegrateTransform(&m_data->m_rigidBodies[i], deltaTime, angDamping, gravityAcceleration);
 	}
-
 }
 
-int		b3CpuRigidBodyPipeline::registerPhysicsInstance(float mass, const float* position, const float* orientation, int collidableIndex, int userData)
+int b3CpuRigidBodyPipeline::registerPhysicsInstance(float mass, const float* position, const float* orientation, int collidableIndex, int userData)
 {
 	b3RigidBodyData body;
 	int bodyIndex = m_data->m_rigidBodies.size();
-	body.m_invMass = mass ? 1.f/mass : 0.f;
-	body.m_angVel.setValue(0,0,0);
+	body.m_invMass = mass ? 1.f / mass : 0.f;
+	body.m_angVel.setValue(0, 0, 0);
 	body.m_collidableIdx = collidableIndex;
 	body.m_frictionCoeff = 0.3f;
-	body.m_linVel.setValue(0,0,0);
-	body.m_pos.setValue(position[0],position[1],position[2]);
-	body.m_quat.setValue(orientation[0],orientation[1],orientation[2],orientation[3]);
+	body.m_linVel.setValue(0, 0, 0);
+	body.m_pos.setValue(position[0], position[1], position[2]);
+	body.m_quat.setValue(orientation[0], orientation[1], orientation[2], orientation[3]);
 	body.m_restituitionCoeff = 0.f;
 
 	m_data->m_rigidBodies.push_back(body);
 
-	
-	if (collidableIndex>=0)
+	if (collidableIndex >= 0)
 	{
 		b3Aabb& worldAabb = m_data->m_aabbWorldSpace.expand();
 
 		b3Aabb localAabb = m_data->m_np->getLocalSpaceAabb(collidableIndex);
-		b3Vector3 localAabbMin=b3MakeVector3(localAabb.m_min[0],localAabb.m_min[1],localAabb.m_min[2]);
-		b3Vector3 localAabbMax=b3MakeVector3(localAabb.m_max[0],localAabb.m_max[1],localAabb.m_max[2]);
-		
+		b3Vector3 localAabbMin = b3MakeVector3(localAabb.m_min[0], localAabb.m_min[1], localAabb.m_min[2]);
+		b3Vector3 localAabbMax = b3MakeVector3(localAabb.m_max[0], localAabb.m_max[1], localAabb.m_max[2]);
+
 		b3Scalar margin = 0.01f;
 		b3Transform t;
 		t.setIdentity();
-		t.setOrigin(b3MakeVector3(position[0],position[1],position[2]));
-		t.setRotation(b3Quaternion(orientation[0],orientation[1],orientation[2],orientation[3]));
-		b3TransformAabb(localAabbMin,localAabbMax, margin,t,worldAabb.m_minVec,worldAabb.m_maxVec);
+		t.setOrigin(b3MakeVector3(position[0], position[1], position[2]));
+		t.setRotation(b3Quaternion(orientation[0], orientation[1], orientation[2], orientation[3]));
+		b3TransformAabb(localAabbMin, localAabbMax, margin, t, worldAabb.m_minVec, worldAabb.m_maxVec);
 
-		m_data->m_bp->createProxy(worldAabb.m_minVec,worldAabb.m_maxVec,bodyIndex,0,1,1);
-//		b3Vector3 aabbMin,aabbMax;
-	//	m_data->m_bp->getAabb(bodyIndex,aabbMin,aabbMax);
-
-	} else
+		m_data->m_bp->createProxy(worldAabb.m_minVec, worldAabb.m_maxVec, bodyIndex, 0, 1, 1);
+		//		b3Vector3 aabbMin,aabbMax;
+		//	m_data->m_bp->getAabb(bodyIndex,aabbMin,aabbMax);
+	}
+	else
 	{
 		b3Error("registerPhysicsInstance using invalid collidableIndex\n");
 	}
@@ -476,13 +436,12 @@ int		b3CpuRigidBodyPipeline::registerPhysicsInstance(float mass, const float* po
 	return bodyIndex;
 }
 
-
 const struct b3RigidBodyData* b3CpuRigidBodyPipeline::getBodyBuffer() const
 {
 	return m_data->m_rigidBodies.size() ? &m_data->m_rigidBodies[0] : 0;
 }
 
-int	b3CpuRigidBodyPipeline::getNumBodies() const
+int b3CpuRigidBodyPipeline::getNumBodies() const
 {
 	return m_data->m_rigidBodies.size();
 }
diff --git a/thirdparty/bullet/Bullet3Dynamics/b3CpuRigidBodyPipeline.h b/thirdparty/bullet/Bullet3Dynamics/b3CpuRigidBodyPipeline.h
index 2f3c2ae77e0..9c65419f26b 100644
--- a/thirdparty/bullet/Bullet3Dynamics/b3CpuRigidBodyPipeline.h
+++ b/thirdparty/bullet/Bullet3Dynamics/b3CpuRigidBodyPipeline.h
@@ -16,52 +16,47 @@ subject to the following restrictions:
 #ifndef B3_CPU_RIGIDBODY_PIPELINE_H
 #define B3_CPU_RIGIDBODY_PIPELINE_H
 
-
-
 #include "Bullet3Common/b3AlignedObjectArray.h"
 #include "Bullet3Collision/NarrowPhaseCollision/b3RaycastInfo.h"
 
 class b3CpuRigidBodyPipeline
 {
 protected:
-	struct b3CpuRigidBodyPipelineInternalData*	m_data;
+	struct b3CpuRigidBodyPipelineInternalData* m_data;
 
 	int allocateCollidable();
 
 public:
-
-
 	b3CpuRigidBodyPipeline(class b3CpuNarrowPhase* narrowphase, struct b3DynamicBvhBroadphase* broadphaseDbvt, const struct b3Config& config);
 	virtual ~b3CpuRigidBodyPipeline();
 
-	virtual void	stepSimulation(float deltaTime);
-	virtual void	integrate(float timeStep);
-	virtual void	updateAabbWorldSpace();
-	virtual void	computeOverlappingPairs();
-	virtual void	computeContactPoints();
-	virtual void	solveContactConstraints();
+	virtual void stepSimulation(float deltaTime);
+	virtual void integrate(float timeStep);
+	virtual void updateAabbWorldSpace();
+	virtual void computeOverlappingPairs();
+	virtual void computeContactPoints();
+	virtual void solveContactConstraints();
 
-	int		registerConvexPolyhedron(class b3ConvexUtility* convex);
+	int registerConvexPolyhedron(class b3ConvexUtility* convex);
 
-	int		registerPhysicsInstance(float mass, const float* position, const float* orientation, int collisionShapeIndex, int userData);
-	void	writeAllInstancesToGpu();
-	void	copyConstraintsToHost();
-	void	setGravity(const float* grav);
-	void	reset();
-	
-	int createPoint2PointConstraint(int bodyA, int bodyB, const float* pivotInA, const float* pivotInB,float breakingThreshold);
+	int registerPhysicsInstance(float mass, const float* position, const float* orientation, int collisionShapeIndex, int userData);
+	void writeAllInstancesToGpu();
+	void copyConstraintsToHost();
+	void setGravity(const float* grav);
+	void reset();
+
+	int createPoint2PointConstraint(int bodyA, int bodyB, const float* pivotInA, const float* pivotInB, float breakingThreshold);
 	int createFixedConstraint(int bodyA, int bodyB, const float* pivotInA, const float* pivotInB, const float* relTargetAB, float breakingThreshold);
 	void removeConstraintByUid(int uid);
 
-	void	addConstraint(class b3TypedConstraint* constraint);
-	void	removeConstraint(b3TypedConstraint* constraint);
+	void addConstraint(class b3TypedConstraint* constraint);
+	void removeConstraint(b3TypedConstraint* constraint);
 
-	void	castRays(const b3AlignedObjectArray<b3RayInfo>& rays,	b3AlignedObjectArray<b3RayHit>& hitResults);
+	void castRays(const b3AlignedObjectArray<b3RayInfo>& rays, b3AlignedObjectArray<b3RayHit>& hitResults);
 
 	const struct b3RigidBodyData* getBodyBuffer() const;
 
-	int	getNumBodies() const;
-
+	int getNumBodies() const;
 };
 
-#endif //B3_CPU_RIGIDBODY_PIPELINE_H
\ No newline at end of file
+#endif  //B3_CPU_RIGIDBODY_PIPELINE_H
\ No newline at end of file
diff --git a/thirdparty/bullet/Bullet3Dynamics/shared/b3ContactConstraint4.h b/thirdparty/bullet/Bullet3Dynamics/shared/b3ContactConstraint4.h
index 68cf65e3129..cf2eed0e7c4 100644
--- a/thirdparty/bullet/Bullet3Dynamics/shared/b3ContactConstraint4.h
+++ b/thirdparty/bullet/Bullet3Dynamics/shared/b3ContactConstraint4.h
@@ -5,30 +5,27 @@
 
 typedef struct b3ContactConstraint4 b3ContactConstraint4_t;
 
-
 struct b3ContactConstraint4
 {
-
-	b3Float4 m_linear;//normal?
+	b3Float4 m_linear;  //normal?
 	b3Float4 m_worldPos[4];
-	b3Float4 m_center;	//	friction
+	b3Float4 m_center;  //	friction
 	float m_jacCoeffInv[4];
 	float m_b[4];
 	float m_appliedRambdaDt[4];
-	float m_fJacCoeffInv[2];	//	friction
-	float m_fAppliedRambdaDt[2];	//	friction
+	float m_fJacCoeffInv[2];      //	friction
+	float m_fAppliedRambdaDt[2];  //	friction
 
 	unsigned int m_bodyA;
 	unsigned int m_bodyB;
-	int			m_batchIdx;
+	int m_batchIdx;
 	unsigned int m_paddings;
-
 };
 
 //inline	void setFrictionCoeff(float value) { m_linear[3] = value; }
-inline	float b3GetFrictionCoeff(b3ContactConstraint4_t* constraint) 
+inline float b3GetFrictionCoeff(b3ContactConstraint4_t* constraint)
 {
-	return constraint->m_linear.w; 
+	return constraint->m_linear.w;
 }
 
-#endif //B3_CONTACT_CONSTRAINT5_H
+#endif  //B3_CONTACT_CONSTRAINT5_H
diff --git a/thirdparty/bullet/Bullet3Dynamics/shared/b3ConvertConstraint4.h b/thirdparty/bullet/Bullet3Dynamics/shared/b3ConvertConstraint4.h
index 805a2bd3ea0..3e72f1c3f21 100644
--- a/thirdparty/bullet/Bullet3Dynamics/shared/b3ConvertConstraint4.h
+++ b/thirdparty/bullet/Bullet3Dynamics/shared/b3ConvertConstraint4.h
@@ -4,89 +4,84 @@
 #include "Bullet3Dynamics/shared/b3ContactConstraint4.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3RigidBodyData.h"
 
-
-void b3PlaneSpace1 (b3Float4ConstArg n, b3Float4* p, b3Float4* q);
- void b3PlaneSpace1 (b3Float4ConstArg n, b3Float4* p, b3Float4* q)
+void b3PlaneSpace1(b3Float4ConstArg n, b3Float4* p, b3Float4* q);
+void b3PlaneSpace1(b3Float4ConstArg n, b3Float4* p, b3Float4* q)
 {
-  if (b3Fabs(n.z) > 0.70710678f) {
-    // choose p in y-z plane
-    float a = n.y*n.y + n.z*n.z;
-    float k = 1.f/sqrt(a);
-    p[0].x = 0;
-	p[0].y = -n.z*k;
-	p[0].z = n.y*k;
-    // set q = n x p
-    q[0].x = a*k;
-	q[0].y = -n.x*p[0].z;
-	q[0].z = n.x*p[0].y;
-  }
-  else {
-    // choose p in x-y plane
-    float a = n.x*n.x + n.y*n.y;
-    float k = 1.f/sqrt(a);
-    p[0].x = -n.y*k;
-	p[0].y = n.x*k;
-	p[0].z = 0;
-    // set q = n x p
-    q[0].x = -n.z*p[0].y;
-	q[0].y = n.z*p[0].x;
-	q[0].z = a*k;
-  }
+	if (b3Fabs(n.z) > 0.70710678f)
+	{
+		// choose p in y-z plane
+		float a = n.y * n.y + n.z * n.z;
+		float k = 1.f / sqrt(a);
+		p[0].x = 0;
+		p[0].y = -n.z * k;
+		p[0].z = n.y * k;
+		// set q = n x p
+		q[0].x = a * k;
+		q[0].y = -n.x * p[0].z;
+		q[0].z = n.x * p[0].y;
+	}
+	else
+	{
+		// choose p in x-y plane
+		float a = n.x * n.x + n.y * n.y;
+		float k = 1.f / sqrt(a);
+		p[0].x = -n.y * k;
+		p[0].y = n.x * k;
+		p[0].z = 0;
+		// set q = n x p
+		q[0].x = -n.z * p[0].y;
+		q[0].y = n.z * p[0].x;
+		q[0].z = a * k;
+	}
 }
 
-
- 
-void setLinearAndAngular( b3Float4ConstArg n, b3Float4ConstArg r0, b3Float4ConstArg r1, b3Float4* linear, b3Float4* angular0, b3Float4* angular1)
+void setLinearAndAngular(b3Float4ConstArg n, b3Float4ConstArg r0, b3Float4ConstArg r1, b3Float4* linear, b3Float4* angular0, b3Float4* angular1)
 {
-	*linear = b3MakeFloat4(n.x,n.y,n.z,0.f);
+	*linear = b3MakeFloat4(n.x, n.y, n.z, 0.f);
 	*angular0 = b3Cross3(r0, n);
 	*angular1 = -b3Cross3(r1, n);
 }
 
-
-float calcRelVel( b3Float4ConstArg l0, b3Float4ConstArg l1, b3Float4ConstArg a0, b3Float4ConstArg a1, b3Float4ConstArg linVel0,
-	b3Float4ConstArg angVel0, b3Float4ConstArg linVel1, b3Float4ConstArg angVel1 )
+float calcRelVel(b3Float4ConstArg l0, b3Float4ConstArg l1, b3Float4ConstArg a0, b3Float4ConstArg a1, b3Float4ConstArg linVel0,
+				 b3Float4ConstArg angVel0, b3Float4ConstArg linVel1, b3Float4ConstArg angVel1)
 {
 	return b3Dot3F4(l0, linVel0) + b3Dot3F4(a0, angVel0) + b3Dot3F4(l1, linVel1) + b3Dot3F4(a1, angVel1);
 }
 
-
 float calcJacCoeff(b3Float4ConstArg linear0, b3Float4ConstArg linear1, b3Float4ConstArg angular0, b3Float4ConstArg angular1,
-					float invMass0, const b3Mat3x3* invInertia0, float invMass1, const b3Mat3x3* invInertia1)
+				   float invMass0, const b3Mat3x3* invInertia0, float invMass1, const b3Mat3x3* invInertia1)
 {
 	//	linear0,1 are normlized
-	float jmj0 = invMass0;//b3Dot3F4(linear0, linear0)*invMass0;
-	float jmj1 = b3Dot3F4(mtMul3(angular0,*invInertia0), angular0);
-	float jmj2 = invMass1;//b3Dot3F4(linear1, linear1)*invMass1;
-	float jmj3 = b3Dot3F4(mtMul3(angular1,*invInertia1), angular1);
-	return -1.f/(jmj0+jmj1+jmj2+jmj3);
+	float jmj0 = invMass0;  //b3Dot3F4(linear0, linear0)*invMass0;
+	float jmj1 = b3Dot3F4(mtMul3(angular0, *invInertia0), angular0);
+	float jmj2 = invMass1;  //b3Dot3F4(linear1, linear1)*invMass1;
+	float jmj3 = b3Dot3F4(mtMul3(angular1, *invInertia1), angular1);
+	return -1.f / (jmj0 + jmj1 + jmj2 + jmj3);
 }
 
-
-void setConstraint4( b3Float4ConstArg posA, b3Float4ConstArg linVelA, b3Float4ConstArg angVelA, float invMassA, b3Mat3x3ConstArg invInertiaA,
-	b3Float4ConstArg posB, b3Float4ConstArg linVelB, b3Float4ConstArg angVelB, float invMassB, b3Mat3x3ConstArg invInertiaB, 
-	__global struct b3Contact4Data* src, float dt, float positionDrift, float positionConstraintCoeff,
-	b3ContactConstraint4_t* dstC )
+void setConstraint4(b3Float4ConstArg posA, b3Float4ConstArg linVelA, b3Float4ConstArg angVelA, float invMassA, b3Mat3x3ConstArg invInertiaA,
+					b3Float4ConstArg posB, b3Float4ConstArg linVelB, b3Float4ConstArg angVelB, float invMassB, b3Mat3x3ConstArg invInertiaB,
+					__global struct b3Contact4Data* src, float dt, float positionDrift, float positionConstraintCoeff,
+					b3ContactConstraint4_t* dstC)
 {
 	dstC->m_bodyA = abs(src->m_bodyAPtrAndSignBit);
 	dstC->m_bodyB = abs(src->m_bodyBPtrAndSignBit);
 
-	float dtInv = 1.f/dt;
-	for(int ic=0; ic<4; ic++)
+	float dtInv = 1.f / dt;
+	for (int ic = 0; ic < 4; ic++)
 	{
 		dstC->m_appliedRambdaDt[ic] = 0.f;
 	}
 	dstC->m_fJacCoeffInv[0] = dstC->m_fJacCoeffInv[1] = 0.f;
 
-
 	dstC->m_linear = src->m_worldNormalOnB;
-	dstC->m_linear.w = 0.7f ;//src->getFrictionCoeff() );
-	for(int ic=0; ic<4; ic++)
+	dstC->m_linear.w = 0.7f;  //src->getFrictionCoeff() );
+	for (int ic = 0; ic < 4; ic++)
 	{
 		b3Float4 r0 = src->m_worldPosB[ic] - posA;
 		b3Float4 r1 = src->m_worldPosB[ic] - posB;
 
-		if( ic >= src->m_worldNormalOnB.w )//npoints
+		if (ic >= src->m_worldNormalOnB.w)  //npoints
 		{
 			dstC->m_jacCoeffInv[ic] = 0.f;
 			continue;
@@ -98,56 +93,56 @@ void setConstraint4( b3Float4ConstArg posA, b3Float4ConstArg linVelA, b3Float4Co
 			setLinearAndAngular(src->m_worldNormalOnB, r0, r1, &linear, &angular0, &angular1);
 
 			dstC->m_jacCoeffInv[ic] = calcJacCoeff(linear, -linear, angular0, angular1,
-				invMassA, &invInertiaA, invMassB, &invInertiaB );
+												   invMassA, &invInertiaA, invMassB, &invInertiaB);
 
 			relVelN = calcRelVel(linear, -linear, angular0, angular1,
-				linVelA, angVelA, linVelB, angVelB);
+								 linVelA, angVelA, linVelB, angVelB);
 
-			float e = 0.f;//src->getRestituitionCoeff();
-			if( relVelN*relVelN < 0.004f ) e = 0.f;
+			float e = 0.f;  //src->getRestituitionCoeff();
+			if (relVelN * relVelN < 0.004f) e = 0.f;
 
-			dstC->m_b[ic] = e*relVelN;
+			dstC->m_b[ic] = e * relVelN;
 			//float penetration = src->m_worldPosB[ic].w;
-			dstC->m_b[ic] += (src->m_worldPosB[ic].w + positionDrift)*positionConstraintCoeff*dtInv;
+			dstC->m_b[ic] += (src->m_worldPosB[ic].w + positionDrift) * positionConstraintCoeff * dtInv;
 			dstC->m_appliedRambdaDt[ic] = 0.f;
 		}
 	}
 
-	if( src->m_worldNormalOnB.w > 0 )//npoints
-	{	//	prepare friction
-		b3Float4 center = b3MakeFloat4(0.f,0.f,0.f,0.f);
-		for(int i=0; i<src->m_worldNormalOnB.w; i++) 
+	if (src->m_worldNormalOnB.w > 0)  //npoints
+	{                                 //	prepare friction
+		b3Float4 center = b3MakeFloat4(0.f, 0.f, 0.f, 0.f);
+		for (int i = 0; i < src->m_worldNormalOnB.w; i++)
 			center += src->m_worldPosB[i];
 		center /= (float)src->m_worldNormalOnB.w;
 
 		b3Float4 tangent[2];
-		b3PlaneSpace1(src->m_worldNormalOnB,&tangent[0],&tangent[1]);
-		
+		b3PlaneSpace1(src->m_worldNormalOnB, &tangent[0], &tangent[1]);
+
 		b3Float4 r[2];
 		r[0] = center - posA;
 		r[1] = center - posB;
 
-		for(int i=0; i<2; i++)
+		for (int i = 0; i < 2; i++)
 		{
 			b3Float4 linear, angular0, angular1;
 			setLinearAndAngular(tangent[i], r[0], r[1], &linear, &angular0, &angular1);
 
 			dstC->m_fJacCoeffInv[i] = calcJacCoeff(linear, -linear, angular0, angular1,
-				invMassA, &invInertiaA, invMassB, &invInertiaB );
+												   invMassA, &invInertiaA, invMassB, &invInertiaB);
 			dstC->m_fAppliedRambdaDt[i] = 0.f;
 		}
 		dstC->m_center = center;
 	}
 
-	for(int i=0; i<4; i++)
+	for (int i = 0; i < 4; i++)
 	{
-		if( i<src->m_worldNormalOnB.w )
+		if (i < src->m_worldNormalOnB.w)
 		{
 			dstC->m_worldPos[i] = src->m_worldPosB[i];
 		}
 		else
 		{
-			dstC->m_worldPos[i] = b3MakeFloat4(0.f,0.f,0.f,0.f);
+			dstC->m_worldPos[i] = b3MakeFloat4(0.f, 0.f, 0.f, 0.f);
 		}
 	}
 }
diff --git a/thirdparty/bullet/Bullet3Dynamics/shared/b3Inertia.h b/thirdparty/bullet/Bullet3Dynamics/shared/b3Inertia.h
index 96fe9f8b398..602a1335aa8 100644
--- a/thirdparty/bullet/Bullet3Dynamics/shared/b3Inertia.h
+++ b/thirdparty/bullet/Bullet3Dynamics/shared/b3Inertia.h
@@ -11,5 +11,4 @@ struct b3Inertia
 	b3Mat3x3 m_initInvInertia;
 };
 
-
-#endif //B3_INERTIA_H
\ No newline at end of file
+#endif  //B3_INERTIA_H
\ No newline at end of file
diff --git a/thirdparty/bullet/Bullet3Dynamics/shared/b3IntegrateTransforms.h b/thirdparty/bullet/Bullet3Dynamics/shared/b3IntegrateTransforms.h
index e96f90d3f32..56d9118f95e 100644
--- a/thirdparty/bullet/Bullet3Dynamics/shared/b3IntegrateTransforms.h
+++ b/thirdparty/bullet/Bullet3Dynamics/shared/b3IntegrateTransforms.h
@@ -2,11 +2,8 @@
 
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3RigidBodyData.h"
 
-
-
-inline void integrateSingleTransform( __global b3RigidBodyData_t* bodies,int nodeID, float timeStep, float angularDamping, b3Float4ConstArg gravityAcceleration)
+inline void integrateSingleTransform(__global b3RigidBodyData_t* bodies, int nodeID, float timeStep, float angularDamping, b3Float4ConstArg gravityAcceleration)
 {
-	
 	if (bodies[nodeID].m_invMass != 0.f)
 	{
 		float BT_GPU_ANGULAR_MOTION_THRESHOLD = (0.25f * 3.14159254f);
@@ -18,27 +15,27 @@ inline void integrateSingleTransform( __global b3RigidBodyData_t* bodies,int nod
 			bodies[nodeID].m_angVel.x *= angularDamping;
 			bodies[nodeID].m_angVel.y *= angularDamping;
 			bodies[nodeID].m_angVel.z *= angularDamping;
-			
+
 			b3Float4 angvel = bodies[nodeID].m_angVel;
 
 			float fAngle = b3Sqrt(b3Dot3F4(angvel, angvel));
-			
+
 			//limit the angular motion
-			if(fAngle*timeStep > BT_GPU_ANGULAR_MOTION_THRESHOLD)
+			if (fAngle * timeStep > BT_GPU_ANGULAR_MOTION_THRESHOLD)
 			{
 				fAngle = BT_GPU_ANGULAR_MOTION_THRESHOLD / timeStep;
 			}
-			if(fAngle < 0.001f)
+			if (fAngle < 0.001f)
 			{
 				// use Taylor's expansions of sync function
-				axis = angvel * (0.5f*timeStep-(timeStep*timeStep*timeStep)*0.020833333333f * fAngle * fAngle);
+				axis = angvel * (0.5f * timeStep - (timeStep * timeStep * timeStep) * 0.020833333333f * fAngle * fAngle);
 			}
 			else
 			{
 				// sync(fAngle) = sin(c*fAngle)/t
-				axis = angvel * ( b3Sin(0.5f * fAngle * timeStep) / fAngle);
+				axis = angvel * (b3Sin(0.5f * fAngle * timeStep) / fAngle);
 			}
-			
+
 			b3Quat dorn;
 			dorn.x = axis.x;
 			dorn.y = axis.y;
@@ -47,23 +44,21 @@ inline void integrateSingleTransform( __global b3RigidBodyData_t* bodies,int nod
 			b3Quat orn0 = bodies[nodeID].m_quat;
 			b3Quat predictedOrn = b3QuatMul(dorn, orn0);
 			predictedOrn = b3QuatNormalized(predictedOrn);
-			bodies[nodeID].m_quat=predictedOrn;
+			bodies[nodeID].m_quat = predictedOrn;
 		}
-		//linear velocity		
-		bodies[nodeID].m_pos +=  bodies[nodeID].m_linVel * timeStep;
-		
+		//linear velocity
+		bodies[nodeID].m_pos += bodies[nodeID].m_linVel * timeStep;
+
 		//apply gravity
 		bodies[nodeID].m_linVel += gravityAcceleration * timeStep;
-		
 	}
-	
 }
 
-inline void b3IntegrateTransform( __global b3RigidBodyData_t* body, float timeStep, float angularDamping, b3Float4ConstArg gravityAcceleration)
+inline void b3IntegrateTransform(__global b3RigidBodyData_t* body, float timeStep, float angularDamping, b3Float4ConstArg gravityAcceleration)
 {
 	float BT_GPU_ANGULAR_MOTION_THRESHOLD = (0.25f * 3.14159254f);
-	
-	if( (body->m_invMass != 0.f))
+
+	if ((body->m_invMass != 0.f))
 	{
 		//angular velocity
 		{
@@ -72,23 +67,23 @@ inline void b3IntegrateTransform( __global b3RigidBodyData_t* body, float timeSt
 			body->m_angVel.x *= angularDamping;
 			body->m_angVel.y *= angularDamping;
 			body->m_angVel.z *= angularDamping;
-			
+
 			b3Float4 angvel = body->m_angVel;
 			float fAngle = b3Sqrt(b3Dot3F4(angvel, angvel));
 			//limit the angular motion
-			if(fAngle*timeStep > BT_GPU_ANGULAR_MOTION_THRESHOLD)
+			if (fAngle * timeStep > BT_GPU_ANGULAR_MOTION_THRESHOLD)
 			{
 				fAngle = BT_GPU_ANGULAR_MOTION_THRESHOLD / timeStep;
 			}
-			if(fAngle < 0.001f)
+			if (fAngle < 0.001f)
 			{
 				// use Taylor's expansions of sync function
-				axis = angvel * (0.5f*timeStep-(timeStep*timeStep*timeStep)*0.020833333333f * fAngle * fAngle);
+				axis = angvel * (0.5f * timeStep - (timeStep * timeStep * timeStep) * 0.020833333333f * fAngle * fAngle);
 			}
 			else
 			{
 				// sync(fAngle) = sin(c*fAngle)/t
-				axis = angvel * ( b3Sin(0.5f * fAngle * timeStep) / fAngle);
+				axis = angvel * (b3Sin(0.5f * fAngle * timeStep) / fAngle);
 			}
 			b3Quat dorn;
 			dorn.x = axis.x;
@@ -99,15 +94,13 @@ inline void b3IntegrateTransform( __global b3RigidBodyData_t* body, float timeSt
 
 			b3Quat predictedOrn = b3QuatMul(dorn, orn0);
 			predictedOrn = b3QuatNormalized(predictedOrn);
-			body->m_quat=predictedOrn;
+			body->m_quat = predictedOrn;
 		}
 
 		//apply gravity
 		body->m_linVel += gravityAcceleration * timeStep;
 
-		//linear velocity		
-		body->m_pos +=  body->m_linVel * timeStep;
-		
+		//linear velocity
+		body->m_pos += body->m_linVel * timeStep;
 	}
-	
 }
diff --git a/thirdparty/bullet/Bullet3Geometry/b3AabbUtil.h b/thirdparty/bullet/Bullet3Geometry/b3AabbUtil.h
index 4c72d5bbfc0..396a4014501 100644
--- a/thirdparty/bullet/Bullet3Geometry/b3AabbUtil.h
+++ b/thirdparty/bullet/Bullet3Geometry/b3AabbUtil.h
@@ -12,8 +12,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #ifndef B3_AABB_UTIL2
 #define B3_AABB_UTIL2
 
@@ -21,20 +19,18 @@ subject to the following restrictions:
 #include "Bullet3Common/b3Vector3.h"
 #include "Bullet3Common/b3MinMax.h"
 
-
-
-B3_FORCE_INLINE void b3AabbExpand (b3Vector3& aabbMin,
-								   b3Vector3& aabbMax,
-								   const b3Vector3& expansionMin,
-								   const b3Vector3& expansionMax)
+B3_FORCE_INLINE void b3AabbExpand(b3Vector3& aabbMin,
+								  b3Vector3& aabbMax,
+								  const b3Vector3& expansionMin,
+								  const b3Vector3& expansionMax)
 {
 	aabbMin = aabbMin + expansionMin;
 	aabbMax = aabbMax + expansionMax;
 }
 
 /// conservative test for overlap between two aabbs
-B3_FORCE_INLINE bool b3TestPointAgainstAabb2(const b3Vector3 &aabbMin1, const b3Vector3 &aabbMax1,
-								const b3Vector3 &point)
+B3_FORCE_INLINE bool b3TestPointAgainstAabb2(const b3Vector3& aabbMin1, const b3Vector3& aabbMax1,
+											 const b3Vector3& point)
 {
 	bool overlap = true;
 	overlap = (aabbMin1.getX() > point.getX() || aabbMax1.getX() < point.getX()) ? false : overlap;
@@ -43,10 +39,9 @@ B3_FORCE_INLINE bool b3TestPointAgainstAabb2(const b3Vector3 &aabbMin1, const b3
 	return overlap;
 }
 
-
 /// conservative test for overlap between two aabbs
-B3_FORCE_INLINE bool b3TestAabbAgainstAabb2(const b3Vector3 &aabbMin1, const b3Vector3 &aabbMax1,
-								const b3Vector3 &aabbMin2, const b3Vector3 &aabbMax2)
+B3_FORCE_INLINE bool b3TestAabbAgainstAabb2(const b3Vector3& aabbMin1, const b3Vector3& aabbMax1,
+											const b3Vector3& aabbMin2, const b3Vector3& aabbMax2)
 {
 	bool overlap = true;
 	overlap = (aabbMin1.getX() > aabbMax2.getX() || aabbMax1.getX() < aabbMin2.getX()) ? false : overlap;
@@ -56,52 +51,49 @@ B3_FORCE_INLINE bool b3TestAabbAgainstAabb2(const b3Vector3 &aabbMin1, const b3V
 }
 
 /// conservative test for overlap between triangle and aabb
-B3_FORCE_INLINE bool b3TestTriangleAgainstAabb2(const b3Vector3 *vertices,
-									const b3Vector3 &aabbMin, const b3Vector3 &aabbMax)
+B3_FORCE_INLINE bool b3TestTriangleAgainstAabb2(const b3Vector3* vertices,
+												const b3Vector3& aabbMin, const b3Vector3& aabbMax)
 {
-	const b3Vector3 &p1 = vertices[0];
-	const b3Vector3 &p2 = vertices[1];
-	const b3Vector3 &p3 = vertices[2];
+	const b3Vector3& p1 = vertices[0];
+	const b3Vector3& p2 = vertices[1];
+	const b3Vector3& p3 = vertices[2];
 
 	if (b3Min(b3Min(p1[0], p2[0]), p3[0]) > aabbMax[0]) return false;
 	if (b3Max(b3Max(p1[0], p2[0]), p3[0]) < aabbMin[0]) return false;
 
 	if (b3Min(b3Min(p1[2], p2[2]), p3[2]) > aabbMax[2]) return false;
 	if (b3Max(b3Max(p1[2], p2[2]), p3[2]) < aabbMin[2]) return false;
-  
+
 	if (b3Min(b3Min(p1[1], p2[1]), p3[1]) > aabbMax[1]) return false;
 	if (b3Max(b3Max(p1[1], p2[1]), p3[1]) < aabbMin[1]) return false;
 	return true;
 }
 
-
-B3_FORCE_INLINE int	b3Outcode(const b3Vector3& p,const b3Vector3& halfExtent) 
+B3_FORCE_INLINE int b3Outcode(const b3Vector3& p, const b3Vector3& halfExtent)
 {
-	return (p.getX()  < -halfExtent.getX() ? 0x01 : 0x0) |    
-		   (p.getX() >  halfExtent.getX() ? 0x08 : 0x0) |
-		   (p.getY() < -halfExtent.getY() ? 0x02 : 0x0) |    
-		   (p.getY() >  halfExtent.getY() ? 0x10 : 0x0) |
-		   (p.getZ() < -halfExtent.getZ() ? 0x4 : 0x0) |    
-		   (p.getZ() >  halfExtent.getZ() ? 0x20 : 0x0);
+	return (p.getX() < -halfExtent.getX() ? 0x01 : 0x0) |
+		   (p.getX() > halfExtent.getX() ? 0x08 : 0x0) |
+		   (p.getY() < -halfExtent.getY() ? 0x02 : 0x0) |
+		   (p.getY() > halfExtent.getY() ? 0x10 : 0x0) |
+		   (p.getZ() < -halfExtent.getZ() ? 0x4 : 0x0) |
+		   (p.getZ() > halfExtent.getZ() ? 0x20 : 0x0);
 }
 
-
-
 B3_FORCE_INLINE bool b3RayAabb2(const b3Vector3& rayFrom,
-								  const b3Vector3& rayInvDirection,
-								  const unsigned int raySign[3],
-								  const b3Vector3 bounds[2],
-								  b3Scalar& tmin,
-								  b3Scalar lambda_min,
-								  b3Scalar lambda_max)
+								const b3Vector3& rayInvDirection,
+								const unsigned int raySign[3],
+								const b3Vector3 bounds[2],
+								b3Scalar& tmin,
+								b3Scalar lambda_min,
+								b3Scalar lambda_max)
 {
 	b3Scalar tmax, tymin, tymax, tzmin, tzmax;
 	tmin = (bounds[raySign[0]].getX() - rayFrom.getX()) * rayInvDirection.getX();
-	tmax = (bounds[1-raySign[0]].getX() - rayFrom.getX()) * rayInvDirection.getX();
+	tmax = (bounds[1 - raySign[0]].getX() - rayFrom.getX()) * rayInvDirection.getX();
 	tymin = (bounds[raySign[1]].getY() - rayFrom.getY()) * rayInvDirection.getY();
-	tymax = (bounds[1-raySign[1]].getY() - rayFrom.getY()) * rayInvDirection.getY();
+	tymax = (bounds[1 - raySign[1]].getY() - rayFrom.getY()) * rayInvDirection.getY();
 
-	if ( (tmin > tymax) || (tymin > tmax) )
+	if ((tmin > tymax) || (tymin > tmax))
 		return false;
 
 	if (tymin > tmin)
@@ -111,59 +103,59 @@ B3_FORCE_INLINE bool b3RayAabb2(const b3Vector3& rayFrom,
 		tmax = tymax;
 
 	tzmin = (bounds[raySign[2]].getZ() - rayFrom.getZ()) * rayInvDirection.getZ();
-	tzmax = (bounds[1-raySign[2]].getZ() - rayFrom.getZ()) * rayInvDirection.getZ();
+	tzmax = (bounds[1 - raySign[2]].getZ() - rayFrom.getZ()) * rayInvDirection.getZ();
 
-	if ( (tmin > tzmax) || (tzmin > tmax) )
+	if ((tmin > tzmax) || (tzmin > tmax))
 		return false;
 	if (tzmin > tmin)
 		tmin = tzmin;
 	if (tzmax < tmax)
 		tmax = tzmax;
-	return ( (tmin < lambda_max) && (tmax > lambda_min) );
+	return ((tmin < lambda_max) && (tmax > lambda_min));
 }
 
-B3_FORCE_INLINE bool b3RayAabb(const b3Vector3& rayFrom, 
-								 const b3Vector3& rayTo, 
-								 const b3Vector3& aabbMin, 
-								 const b3Vector3& aabbMax,
-					  b3Scalar& param, b3Vector3& normal) 
+B3_FORCE_INLINE bool b3RayAabb(const b3Vector3& rayFrom,
+							   const b3Vector3& rayTo,
+							   const b3Vector3& aabbMin,
+							   const b3Vector3& aabbMax,
+							   b3Scalar& param, b3Vector3& normal)
 {
-	b3Vector3 aabbHalfExtent = (aabbMax-aabbMin)* b3Scalar(0.5);
-	b3Vector3 aabbCenter = (aabbMax+aabbMin)* b3Scalar(0.5);
-	b3Vector3	source = rayFrom - aabbCenter;
-	b3Vector3	target = rayTo - aabbCenter;
-	int	sourceOutcode = b3Outcode(source,aabbHalfExtent);
-	int targetOutcode = b3Outcode(target,aabbHalfExtent);
+	b3Vector3 aabbHalfExtent = (aabbMax - aabbMin) * b3Scalar(0.5);
+	b3Vector3 aabbCenter = (aabbMax + aabbMin) * b3Scalar(0.5);
+	b3Vector3 source = rayFrom - aabbCenter;
+	b3Vector3 target = rayTo - aabbCenter;
+	int sourceOutcode = b3Outcode(source, aabbHalfExtent);
+	int targetOutcode = b3Outcode(target, aabbHalfExtent);
 	if ((sourceOutcode & targetOutcode) == 0x0)
 	{
 		b3Scalar lambda_enter = b3Scalar(0.0);
-		b3Scalar lambda_exit  = param;
+		b3Scalar lambda_exit = param;
 		b3Vector3 r = target - source;
 		int i;
-		b3Scalar	normSign = 1;
-		b3Vector3	hitNormal = b3MakeVector3(0,0,0);
-		int bit=1;
+		b3Scalar normSign = 1;
+		b3Vector3 hitNormal = b3MakeVector3(0, 0, 0);
+		int bit = 1;
 
-		for (int j=0;j<2;j++)
+		for (int j = 0; j < 2; j++)
 		{
 			for (i = 0; i != 3; ++i)
 			{
 				if (sourceOutcode & bit)
 				{
-					b3Scalar lambda = (-source[i] - aabbHalfExtent[i]*normSign) / r[i];
+					b3Scalar lambda = (-source[i] - aabbHalfExtent[i] * normSign) / r[i];
 					if (lambda_enter <= lambda)
 					{
 						lambda_enter = lambda;
-						hitNormal.setValue(0,0,0);
+						hitNormal.setValue(0, 0, 0);
 						hitNormal[i] = normSign;
 					}
 				}
-				else if (targetOutcode & bit) 
+				else if (targetOutcode & bit)
 				{
-					b3Scalar lambda = (-source[i] - aabbHalfExtent[i]*normSign) / r[i];
+					b3Scalar lambda = (-source[i] - aabbHalfExtent[i] * normSign) / r[i];
 					b3SetMin(lambda_exit, lambda);
 				}
-				bit<<=1;
+				bit <<= 1;
 			}
 			normSign = b3Scalar(-1.);
 		}
@@ -177,56 +169,49 @@ B3_FORCE_INLINE bool b3RayAabb(const b3Vector3& rayFrom,
 	return false;
 }
 
-
-
-B3_FORCE_INLINE	void b3TransformAabb(const b3Vector3& halfExtents, b3Scalar margin,const b3Transform& t,b3Vector3& aabbMinOut,b3Vector3& aabbMaxOut)
+B3_FORCE_INLINE void b3TransformAabb(const b3Vector3& halfExtents, b3Scalar margin, const b3Transform& t, b3Vector3& aabbMinOut, b3Vector3& aabbMaxOut)
 {
-	b3Vector3 halfExtentsWithMargin = halfExtents+b3MakeVector3(margin,margin,margin);
-	b3Matrix3x3 abs_b = t.getBasis().absolute();  
+	b3Vector3 halfExtentsWithMargin = halfExtents + b3MakeVector3(margin, margin, margin);
+	b3Matrix3x3 abs_b = t.getBasis().absolute();
 	b3Vector3 center = t.getOrigin();
-    b3Vector3 extent = halfExtentsWithMargin.dot3( abs_b[0], abs_b[1], abs_b[2] );
+	b3Vector3 extent = halfExtentsWithMargin.dot3(abs_b[0], abs_b[1], abs_b[2]);
 	aabbMinOut = center - extent;
 	aabbMaxOut = center + extent;
 }
 
-
-B3_FORCE_INLINE	void b3TransformAabb(const b3Vector3& localAabbMin,const b3Vector3& localAabbMax, b3Scalar margin,const b3Transform& trans,b3Vector3& aabbMinOut,b3Vector3& aabbMaxOut)
+B3_FORCE_INLINE void b3TransformAabb(const b3Vector3& localAabbMin, const b3Vector3& localAabbMax, b3Scalar margin, const b3Transform& trans, b3Vector3& aabbMinOut, b3Vector3& aabbMaxOut)
 {
-		//b3Assert(localAabbMin.getX() <= localAabbMax.getX());
-		//b3Assert(localAabbMin.getY() <= localAabbMax.getY());
-		//b3Assert(localAabbMin.getZ() <= localAabbMax.getZ());
-		b3Vector3 localHalfExtents = b3Scalar(0.5)*(localAabbMax-localAabbMin);
-		localHalfExtents+=b3MakeVector3(margin,margin,margin);
+	//b3Assert(localAabbMin.getX() <= localAabbMax.getX());
+	//b3Assert(localAabbMin.getY() <= localAabbMax.getY());
+	//b3Assert(localAabbMin.getZ() <= localAabbMax.getZ());
+	b3Vector3 localHalfExtents = b3Scalar(0.5) * (localAabbMax - localAabbMin);
+	localHalfExtents += b3MakeVector3(margin, margin, margin);
 
-		b3Vector3 localCenter = b3Scalar(0.5)*(localAabbMax+localAabbMin);
-		b3Matrix3x3 abs_b = trans.getBasis().absolute();  
-		b3Vector3 center = trans(localCenter);
-        b3Vector3 extent = localHalfExtents.dot3( abs_b[0], abs_b[1], abs_b[2] );
-		aabbMinOut = center-extent;
-		aabbMaxOut = center+extent;
+	b3Vector3 localCenter = b3Scalar(0.5) * (localAabbMax + localAabbMin);
+	b3Matrix3x3 abs_b = trans.getBasis().absolute();
+	b3Vector3 center = trans(localCenter);
+	b3Vector3 extent = localHalfExtents.dot3(abs_b[0], abs_b[1], abs_b[2]);
+	aabbMinOut = center - extent;
+	aabbMaxOut = center + extent;
 }
 
 #define B3_USE_BANCHLESS 1
 #ifdef B3_USE_BANCHLESS
-	//This block replaces the block below and uses no branches, and replaces the 8 bit return with a 32 bit return for improved performance (~3x on XBox 360)
-	B3_FORCE_INLINE unsigned b3TestQuantizedAabbAgainstQuantizedAabb(const unsigned short int* aabbMin1,const unsigned short int* aabbMax1,const unsigned short int* aabbMin2,const unsigned short int* aabbMax2)
-	{		
-		return static_cast<unsigned int>(b3Select((unsigned)((aabbMin1[0] <= aabbMax2[0]) & (aabbMax1[0] >= aabbMin2[0])
-			& (aabbMin1[2] <= aabbMax2[2]) & (aabbMax1[2] >= aabbMin2[2])
-			& (aabbMin1[1] <= aabbMax2[1]) & (aabbMax1[1] >= aabbMin2[1])),
-			1, 0));
-	}
+//This block replaces the block below and uses no branches, and replaces the 8 bit return with a 32 bit return for improved performance (~3x on XBox 360)
+B3_FORCE_INLINE unsigned b3TestQuantizedAabbAgainstQuantizedAabb(const unsigned short int* aabbMin1, const unsigned short int* aabbMax1, const unsigned short int* aabbMin2, const unsigned short int* aabbMax2)
+{
+	return static_cast<unsigned int>(b3Select((unsigned)((aabbMin1[0] <= aabbMax2[0]) & (aabbMax1[0] >= aabbMin2[0]) & (aabbMin1[2] <= aabbMax2[2]) & (aabbMax1[2] >= aabbMin2[2]) & (aabbMin1[1] <= aabbMax2[1]) & (aabbMax1[1] >= aabbMin2[1])),
+											  1, 0));
+}
 #else
-	B3_FORCE_INLINE bool b3TestQuantizedAabbAgainstQuantizedAabb(const unsigned short int* aabbMin1,const unsigned short int* aabbMax1,const unsigned short int* aabbMin2,const unsigned short int* aabbMax2)
-	{
-		bool overlap = true;
-		overlap = (aabbMin1[0] > aabbMax2[0] || aabbMax1[0] < aabbMin2[0]) ? false : overlap;
-		overlap = (aabbMin1[2] > aabbMax2[2] || aabbMax1[2] < aabbMin2[2]) ? false : overlap;
-		overlap = (aabbMin1[1] > aabbMax2[1] || aabbMax1[1] < aabbMin2[1]) ? false : overlap;
-		return overlap;
-	}
-#endif //B3_USE_BANCHLESS
-
-#endif //B3_AABB_UTIL2
-
+B3_FORCE_INLINE bool b3TestQuantizedAabbAgainstQuantizedAabb(const unsigned short int* aabbMin1, const unsigned short int* aabbMax1, const unsigned short int* aabbMin2, const unsigned short int* aabbMax2)
+{
+	bool overlap = true;
+	overlap = (aabbMin1[0] > aabbMax2[0] || aabbMax1[0] < aabbMin2[0]) ? false : overlap;
+	overlap = (aabbMin1[2] > aabbMax2[2] || aabbMax1[2] < aabbMin2[2]) ? false : overlap;
+	overlap = (aabbMin1[1] > aabbMax2[1] || aabbMax1[1] < aabbMin2[1]) ? false : overlap;
+	return overlap;
+}
+#endif  //B3_USE_BANCHLESS
 
+#endif  //B3_AABB_UTIL2
diff --git a/thirdparty/bullet/Bullet3Geometry/b3ConvexHullComputer.cpp b/thirdparty/bullet/Bullet3Geometry/b3ConvexHullComputer.cpp
index 18835c38d5a..b37652456e8 100644
--- a/thirdparty/bullet/Bullet3Geometry/b3ConvexHullComputer.cpp
+++ b/thirdparty/bullet/Bullet3Geometry/b3ConvexHullComputer.cpp
@@ -20,850 +20,851 @@ subject to the following restrictions:
 #include "Bullet3Common/b3Vector3.h"
 
 #ifdef __GNUC__
-	#include <stdint.h>
-	typedef int32_t btInt32_t;
-	typedef int64_t btInt64_t;
-	typedef uint32_t btUint32_t;
-	typedef uint64_t btUint64_t;
+#include <stdint.h>
+typedef int32_t btInt32_t;
+typedef int64_t btInt64_t;
+typedef uint32_t btUint32_t;
+typedef uint64_t btUint64_t;
 #elif defined(_MSC_VER)
-	typedef __int32 btInt32_t;
-	typedef __int64 btInt64_t;
-	typedef unsigned __int32 btUint32_t;
-	typedef unsigned __int64 btUint64_t;
+typedef __int32 btInt32_t;
+typedef __int64 btInt64_t;
+typedef unsigned __int32 btUint32_t;
+typedef unsigned __int64 btUint64_t;
 #else
-	typedef int btInt32_t;
-	typedef long long int btInt64_t;
-	typedef unsigned int btUint32_t;
-	typedef unsigned long long int btUint64_t;
+typedef int btInt32_t;
+typedef long long int btInt64_t;
+typedef unsigned int btUint32_t;
+typedef unsigned long long int btUint64_t;
 #endif
 
-
 //The definition of USE_X86_64_ASM is moved into the build system. You can enable it manually by commenting out the following lines
 //#if (defined(__GNUC__) && defined(__x86_64__) && !defined(__ICL))  // || (defined(__ICL) && defined(_M_X64))   bug in Intel compiler, disable inline assembly
 //	#define USE_X86_64_ASM
 //#endif
 
-
 //#define DEBUG_CONVEX_HULL
 //#define SHOW_ITERATIONS
 
 #if defined(DEBUG_CONVEX_HULL) || defined(SHOW_ITERATIONS)
-	#include <stdio.h>
+#include <stdio.h>
 #endif
 
 // Convex hull implementation based on Preparata and Hong
 // Ole Kniemeyer, MAXON Computer GmbH
 class b3ConvexHullInternal
 {
+public:
+	class Point64
+	{
 	public:
-		
-		class Point64
+		btInt64_t x;
+		btInt64_t y;
+		btInt64_t z;
+
+		Point64(btInt64_t x, btInt64_t y, btInt64_t z) : x(x), y(y), z(z)
 		{
-			public:
-				btInt64_t x;
-				btInt64_t y;
-				btInt64_t z;
-				
-				Point64(btInt64_t x, btInt64_t y, btInt64_t z): x(x), y(y), z(z)
-				{
-				}
-
-				bool isZero()
-				{
-					return (x == 0) && (y == 0) && (z == 0);
-				}
-
-				btInt64_t dot(const Point64& b) const
-				{
-					return x * b.x + y * b.y + z * b.z;
-				}
-		};
-		
-		class Point32
-		{
-			public:
-				btInt32_t x;
-				btInt32_t y;
-				btInt32_t z;
-				int index;
-				
-				Point32()
-				{
-				}
-				
-				Point32(btInt32_t x, btInt32_t y, btInt32_t z): x(x), y(y), z(z), index(-1)
-				{
-				}
-				
-				bool operator==(const Point32& b) const
-				{
-					return (x == b.x) && (y == b.y) && (z == b.z);
-				}
-
-				bool operator!=(const Point32& b) const
-				{
-					return (x != b.x) || (y != b.y) || (z != b.z);
-				}
-
-				bool isZero()
-				{
-					return (x == 0) && (y == 0) && (z == 0);
-				}
-
-				Point64 cross(const Point32& b) const
-				{
-					return Point64(y * b.z - z * b.y, z * b.x - x * b.z, x * b.y - y * b.x);
-				}
-
-				Point64 cross(const Point64& b) const
-				{
-					return Point64(y * b.z - z * b.y, z * b.x - x * b.z, x * b.y - y * b.x);
-				}
-
-				btInt64_t dot(const Point32& b) const
-				{
-					return x * b.x + y * b.y + z * b.z;
-				}
-
-				btInt64_t dot(const Point64& b) const
-				{
-					return x * b.x + y * b.y + z * b.z;
-				}
-
-				Point32 operator+(const Point32& b) const
-				{
-					return Point32(x + b.x, y + b.y, z + b.z);
-				}
-
-				Point32 operator-(const Point32& b) const
-				{
-					return Point32(x - b.x, y - b.y, z - b.z);
-				}
-		};
-
-		class Int128
-		{
-			public:
-				btUint64_t low;
-				btUint64_t high;
-
-				Int128()
-				{
-				}
-
-				Int128(btUint64_t low, btUint64_t high): low(low), high(high)
-				{
-				}
-
-				Int128(btUint64_t low): low(low), high(0)
-				{
-				}
-
-				Int128(btInt64_t value): low(value), high((value >= 0) ? 0 : (btUint64_t) -1LL)
-				{
-				}
-
-				static Int128 mul(btInt64_t a, btInt64_t b);
-
-				static Int128 mul(btUint64_t a, btUint64_t b);
-
-				Int128 operator-() const
-				{
-					return Int128((btUint64_t) -(btInt64_t)low, ~high + (low == 0));
-				}
-
-				Int128 operator+(const Int128& b) const
-				{
-#ifdef USE_X86_64_ASM
-					Int128 result;
-					__asm__ ("addq %[bl], %[rl]\n\t"
-									 "adcq %[bh], %[rh]\n\t"
-									 : [rl] "=r" (result.low), [rh] "=r" (result.high)
-									 : "0"(low), "1"(high), [bl] "g"(b.low), [bh] "g"(b.high)
-									 : "cc" );
-					return result;
-#else
-					btUint64_t lo = low + b.low;
-					return Int128(lo, high + b.high + (lo < low));
-#endif
-				}
-
-				Int128 operator-(const Int128& b) const
-				{
-#ifdef USE_X86_64_ASM
-					Int128 result;
-					__asm__ ("subq %[bl], %[rl]\n\t"
-									 "sbbq %[bh], %[rh]\n\t"
-									 : [rl] "=r" (result.low), [rh] "=r" (result.high)
-									 : "0"(low), "1"(high), [bl] "g"(b.low), [bh] "g"(b.high)
-									 : "cc" );
-					return result;
-#else
-					return *this + -b;
-#endif
-				}
-
-				Int128& operator+=(const Int128& b)
-				{
-#ifdef USE_X86_64_ASM
-					__asm__ ("addq %[bl], %[rl]\n\t"
-									 "adcq %[bh], %[rh]\n\t"
-									 : [rl] "=r" (low), [rh] "=r" (high)
-									 : "0"(low), "1"(high), [bl] "g"(b.low), [bh] "g"(b.high)
-									 : "cc" );
-#else
-					btUint64_t lo = low + b.low;
-					if (lo < low)
-					{
-						++high;
-					}
-					low = lo;
-					high += b.high;
-#endif
-					return *this;
-				}
-
-				Int128& operator++()
-				{
-					if (++low == 0)
-					{
-						++high;
-					}
-					return *this;
-				}
-
-				Int128 operator*(btInt64_t b) const;
-
-				b3Scalar toScalar() const
-				{
-					return ((btInt64_t) high >= 0) ? b3Scalar(high) * (b3Scalar(0x100000000LL) * b3Scalar(0x100000000LL)) + b3Scalar(low)
-						: -(-*this).toScalar();
-				}
-
-				int getSign() const
-				{
-					return ((btInt64_t) high < 0) ? -1 : (high || low) ? 1 : 0;
-				}
-
-				bool operator<(const Int128& b) const
-				{
-					return (high < b.high) || ((high == b.high) && (low < b.low));
-				}
-
-				int ucmp(const Int128&b) const
-				{
-					if (high < b.high)
-					{
-						return -1;
-					}
-					if (high > b.high)
-					{
-						return 1;
-					}
-					if (low < b.low)
-					{
-						return -1;
-					}
-					if (low > b.low)
-					{
-						return 1;
-					}
-					return 0;
-				}
-		};
-
-
-		class Rational64
-		{
-			private:
-				btUint64_t m_numerator;
-				btUint64_t m_denominator;
-				int sign;
-				
-			public:
-				Rational64(btInt64_t numerator, btInt64_t denominator)
-				{
-					if (numerator > 0)
-					{
-						sign = 1;
-						m_numerator = (btUint64_t) numerator;
-					}
-					else if (numerator < 0)
-					{
-						sign = -1;
-						m_numerator = (btUint64_t) -numerator;
-					}
-					else
-					{
-						sign = 0;
-						m_numerator = 0;
-					}
-					if (denominator > 0)
-					{
-						m_denominator = (btUint64_t) denominator;
-					}
-					else if (denominator < 0)
-					{
-						sign = -sign;
-						m_denominator = (btUint64_t) -denominator;
-					}
-					else
-					{
-						m_denominator = 0;
-					}
-				}
-				
-				bool isNegativeInfinity() const
-				{
-					return (sign < 0) && (m_denominator == 0);
-				}
-				
-				bool isNaN() const
-				{
-					return (sign == 0) && (m_denominator == 0);
-				}
-				
-				int compare(const Rational64& b) const;
-				
-				b3Scalar toScalar() const
-				{
-					return sign * ((m_denominator == 0) ? B3_INFINITY : (b3Scalar) m_numerator / m_denominator);
-				}
-		};
-
-
-		class Rational128
-		{
-			private:
-				Int128 numerator;
-				Int128 denominator;
-				int sign;
-				bool isInt64;
-
-			public:
-				Rational128(btInt64_t value)
-				{
-					if (value > 0)
-					{
-						sign = 1;
-						this->numerator = value;
-					}
-					else if (value < 0)
-					{
-						sign = -1;
-						this->numerator = -value;
-					}
-					else
-					{
-						sign = 0;
-						this->numerator = (btUint64_t) 0;
-					}
-					this->denominator = (btUint64_t) 1;
-					isInt64 = true;
-				}
-
-				Rational128(const Int128& numerator, const Int128& denominator)
-				{
-					sign = numerator.getSign();
-					if (sign >= 0)
-					{
-						this->numerator = numerator;
-					}
-					else
-					{
-						this->numerator = -numerator;
-					}
-					int dsign = denominator.getSign();
-					if (dsign >= 0)
-					{
-						this->denominator = denominator;
-					}
-					else
-					{
-						sign = -sign;
-						this->denominator = -denominator;
-					}
-					isInt64 = false;
-				}
-
-				int compare(const Rational128& b) const;
-
-				int compare(btInt64_t b) const;
-
-				b3Scalar toScalar() const
-				{
-					return sign * ((denominator.getSign() == 0) ? B3_INFINITY : numerator.toScalar() / denominator.toScalar());
-				}
-		};
-
-		class PointR128
-		{
-			public:
-				Int128 x;
-				Int128 y;
-				Int128 z;
-				Int128 denominator;
-
-				PointR128()
-				{
-				}
-
-				PointR128(Int128 x, Int128 y, Int128 z, Int128 denominator): x(x), y(y), z(z), denominator(denominator)
-				{
-				}
-
-				b3Scalar xvalue() const
-				{
-					return x.toScalar() / denominator.toScalar();
-				}
-
-				b3Scalar yvalue() const
-				{
-					return y.toScalar() / denominator.toScalar();
-				}
-
-				b3Scalar zvalue() const
-				{
-					return z.toScalar() / denominator.toScalar();
-				}
-		};
-
-
-		class Edge;
-		class Face;
-
-		class Vertex
-		{
-			public:
-				Vertex* next;
-				Vertex* prev;
-				Edge* edges;
-				Face* firstNearbyFace;
-				Face* lastNearbyFace;
-				PointR128 point128;
-				Point32 point;
-				int copy;
-				
-				Vertex(): next(NULL), prev(NULL), edges(NULL), firstNearbyFace(NULL), lastNearbyFace(NULL), copy(-1)
-				{
-				}
-
-#ifdef DEBUG_CONVEX_HULL
-				void print()
-				{
-					b3Printf("V%d (%d, %d, %d)", point.index, point.x, point.y, point.z);
-				}
-
-				void printGraph();
-#endif
-
-				Point32 operator-(const Vertex& b) const
-				{
-					return point - b.point;
-				}
-
-				Rational128 dot(const Point64& b) const
-				{
-					return (point.index >= 0) ? Rational128(point.dot(b))
-						: Rational128(point128.x * b.x + point128.y * b.y + point128.z * b.z, point128.denominator);
-				}
-
-				b3Scalar xvalue() const
-				{
-					return (point.index >= 0) ? b3Scalar(point.x) : point128.xvalue();
-				}
-
-				b3Scalar yvalue() const
-				{
-					return (point.index >= 0) ? b3Scalar(point.y) : point128.yvalue();
-				}
-
-				b3Scalar zvalue() const
-				{
-					return (point.index >= 0) ? b3Scalar(point.z) : point128.zvalue();
-				}
-
-				void receiveNearbyFaces(Vertex* src)
-				{
-					if (lastNearbyFace)
-					{
-						lastNearbyFace->nextWithSameNearbyVertex = src->firstNearbyFace;
-					}
-					else
-					{
-						firstNearbyFace = src->firstNearbyFace;
-					}
-					if (src->lastNearbyFace)
-					{
-						lastNearbyFace = src->lastNearbyFace;
-					}
-					for (Face* f = src->firstNearbyFace; f; f = f->nextWithSameNearbyVertex)
-					{
-						b3Assert(f->nearbyVertex == src);
-						f->nearbyVertex = this;
-					}
-					src->firstNearbyFace = NULL;
-					src->lastNearbyFace = NULL;
-				}
-		};
-
-
-		class Edge
-		{
-			public:
-				Edge* next;
-				Edge* prev;
-				Edge* reverse;
-				Vertex* target;
-				Face* face;
-				int copy;
-
-				~Edge()
-				{
-					next = NULL;
-					prev = NULL;
-					reverse = NULL;
-					target = NULL;
-					face = NULL;
-				}
-
-				void link(Edge* n)
-				{
-					b3Assert(reverse->target == n->reverse->target);
-					next = n;
-					n->prev = this;
-				}
-
-#ifdef DEBUG_CONVEX_HULL
-				void print()
-				{
-					b3Printf("E%p : %d -> %d,  n=%p p=%p   (0 %d\t%d\t%d) -> (%d %d %d)", this, reverse->target->point.index, target->point.index, next, prev,
-								 reverse->target->point.x, reverse->target->point.y, reverse->target->point.z, target->point.x, target->point.y, target->point.z);
-				}
-#endif
-		};
-
-		class Face
-		{
-			public:
-				Face* next;
-				Vertex* nearbyVertex;
-				Face* nextWithSameNearbyVertex;
-				Point32 origin;
-				Point32 dir0;
-				Point32 dir1;
-
-				Face(): next(NULL), nearbyVertex(NULL), nextWithSameNearbyVertex(NULL)
-				{
-				}
-
-				void init(Vertex* a, Vertex* b, Vertex* c)
-				{
-					nearbyVertex = a;
-					origin = a->point;
-					dir0 = *b - *a;
-					dir1 = *c - *a;
-					if (a->lastNearbyFace)
-					{
-						a->lastNearbyFace->nextWithSameNearbyVertex = this;
-					}
-					else
-					{
-						a->firstNearbyFace = this;
-					}
-					a->lastNearbyFace = this;
-				}
-
-				Point64 getNormal()
-				{
-					return dir0.cross(dir1);
-				}
-		};
-
-		template<typename UWord, typename UHWord> class DMul
-		{
-			private:
-				static btUint32_t high(btUint64_t value)
-				{
-					return (btUint32_t) (value >> 32);
-				}
-				
-				static btUint32_t low(btUint64_t value)
-				{
-					return (btUint32_t) value;
-				}
-				
-				static btUint64_t mul(btUint32_t a, btUint32_t b)
-				{
-					return (btUint64_t) a * (btUint64_t) b;
-				}
-				
-				static void shlHalf(btUint64_t& value)
-				{
-					value <<= 32;
-				}
-				
-				static btUint64_t high(Int128 value)
-				{
-					return value.high;
-				}
-				
-				static btUint64_t low(Int128 value)
-				{
-					return value.low;
-				}
-				
-				static Int128 mul(btUint64_t a, btUint64_t b)
-				{
-					return Int128::mul(a, b);
-				}
-				
-				static void shlHalf(Int128& value)
-				{
-					value.high = value.low;
-					value.low = 0;
-				}
-				
-			public:
-				
-				static void mul(UWord a, UWord b, UWord& resLow, UWord& resHigh)
-				{
-					UWord p00 = mul(low(a), low(b));
-					UWord p01 = mul(low(a), high(b));
-					UWord p10 = mul(high(a), low(b));
-					UWord p11 = mul(high(a), high(b));
-					UWord p0110 = UWord(low(p01)) + UWord(low(p10));
-					p11 += high(p01);
-					p11 += high(p10);
-					p11 += high(p0110);
-					shlHalf(p0110);
-					p00 += p0110;
-					if (p00 < p0110)
-					{
-						++p11;
-					}
-					resLow = p00;
-					resHigh = p11;
-				}
-		};
-	
-	private:
-
-		class IntermediateHull
-		{
-			public:
-				Vertex* minXy;
-				Vertex* maxXy;
-				Vertex* minYx;
-				Vertex* maxYx;
-				
-				IntermediateHull(): minXy(NULL), maxXy(NULL), minYx(NULL), maxYx(NULL)
-				{
-				}
-				
-				void print();
-		};
-	
-		enum Orientation {NONE, CLOCKWISE, COUNTER_CLOCKWISE};
-
-		template <typename T> class PoolArray
-		{
-			private:
-				T* array;
-				int size;
-
-			public:
-				PoolArray<T>* next;
-
-				PoolArray(int size): size(size), next(NULL)
-				{
-					array = (T*) b3AlignedAlloc(sizeof(T) * size, 16);
-				}
-
-				~PoolArray()
-				{
-					b3AlignedFree(array);
-				}
-
-				T* init()
-				{
-					T* o = array;
-					for (int i = 0; i < size; i++, o++)
-					{
-						o->next = (i+1 < size) ? o + 1 : NULL;
-					}
-					return array;
-				}
-		};
-
-		template <typename T> class Pool
-		{
-			private:
-				PoolArray<T>* arrays;
-				PoolArray<T>* nextArray;
-				T* freeObjects;
-				int arraySize;
-
-			public:
-				Pool(): arrays(NULL), nextArray(NULL), freeObjects(NULL), arraySize(256)
-				{
-				}
-
-				~Pool()
-				{
-					while (arrays)
-					{
-						PoolArray<T>* p = arrays;
-						arrays = p->next;
-						p->~PoolArray<T>();
-						b3AlignedFree(p);
-					}
-				}
-
-				void reset()
-				{
-					nextArray = arrays;
-					freeObjects = NULL;
-				}
-
-				void setArraySize(int arraySize)
-				{
-					this->arraySize = arraySize;
-				}
-
-				T* newObject()
-				{
-					T* o = freeObjects;
-					if (!o)
-					{
-						PoolArray<T>* p = nextArray;
-						if (p)
-						{
-							nextArray = p->next;
-						}
-						else
-						{
-							p = new(b3AlignedAlloc(sizeof(PoolArray<T>), 16)) PoolArray<T>(arraySize);
-							p->next = arrays;
-							arrays = p;
-						}
-						o = p->init();
-					}
-					freeObjects = o->next;
-					return new(o) T();
-				};
-
-				void freeObject(T* object)
-				{
-					object->~T();
-					object->next = freeObjects;
-					freeObjects = object;
-				}
-		};
-
-		b3Vector3 scaling;
-		b3Vector3 center;
-		Pool<Vertex> vertexPool;
-		Pool<Edge> edgePool;
-		Pool<Face> facePool;
-		b3AlignedObjectArray<Vertex*> originalVertices;
-		int mergeStamp;
-		int minAxis;
-		int medAxis;
-		int maxAxis;
-		int usedEdgePairs;
-		int maxUsedEdgePairs;
-
-		static Orientation getOrientation(const Edge* prev, const Edge* next, const Point32& s, const Point32& t);
-		Edge* findMaxAngle(bool ccw, const Vertex* start, const Point32& s, const Point64& rxs, const Point64& sxrxs, Rational64& minCot);
-		void findEdgeForCoplanarFaces(Vertex* c0, Vertex* c1, Edge*& e0, Edge*& e1, Vertex* stop0, Vertex* stop1);
-
-		Edge* newEdgePair(Vertex* from, Vertex* to);
-
-		void removeEdgePair(Edge* edge)
-		{
-			Edge* n = edge->next;
-			Edge* r = edge->reverse;
-
-			b3Assert(edge->target && r->target);
-
-			if (n != edge)
-			{
-				n->prev = edge->prev;
-				edge->prev->next = n;
-				r->target->edges = n;
-			}
-			else
-			{
-				r->target->edges = NULL;
-			}
-			
-			n = r->next;
-			
-			if (n != r)
-			{
-				n->prev = r->prev;
-				r->prev->next = n;
-				edge->target->edges = n;
-			}
-			else
-			{
-				edge->target->edges = NULL;
-			}
-
-			edgePool.freeObject(edge);
-			edgePool.freeObject(r);
-			usedEdgePairs--;
 		}
-		
-		void computeInternal(int start, int end, IntermediateHull& result);
-		
-		bool mergeProjection(IntermediateHull& h0, IntermediateHull& h1, Vertex*& c0, Vertex*& c1);
-		
-		void merge(IntermediateHull& h0, IntermediateHull& h1);
 
-		b3Vector3 toBtVector(const Point32& v);
+		bool isZero()
+		{
+			return (x == 0) && (y == 0) && (z == 0);
+		}
 
-		b3Vector3 getBtNormal(Face* face);
+		btInt64_t dot(const Point64& b) const
+		{
+			return x * b.x + y * b.y + z * b.z;
+		}
+	};
 
-		bool shiftFace(Face* face, b3Scalar amount, b3AlignedObjectArray<Vertex*> stack);
+	class Point32
+	{
+	public:
+		btInt32_t x;
+		btInt32_t y;
+		btInt32_t z;
+		int index;
+
+		Point32()
+		{
+		}
+
+		Point32(btInt32_t x, btInt32_t y, btInt32_t z) : x(x), y(y), z(z), index(-1)
+		{
+		}
+
+		bool operator==(const Point32& b) const
+		{
+			return (x == b.x) && (y == b.y) && (z == b.z);
+		}
+
+		bool operator!=(const Point32& b) const
+		{
+			return (x != b.x) || (y != b.y) || (z != b.z);
+		}
+
+		bool isZero()
+		{
+			return (x == 0) && (y == 0) && (z == 0);
+		}
+
+		Point64 cross(const Point32& b) const
+		{
+			return Point64(y * b.z - z * b.y, z * b.x - x * b.z, x * b.y - y * b.x);
+		}
+
+		Point64 cross(const Point64& b) const
+		{
+			return Point64(y * b.z - z * b.y, z * b.x - x * b.z, x * b.y - y * b.x);
+		}
+
+		btInt64_t dot(const Point32& b) const
+		{
+			return x * b.x + y * b.y + z * b.z;
+		}
+
+		btInt64_t dot(const Point64& b) const
+		{
+			return x * b.x + y * b.y + z * b.z;
+		}
+
+		Point32 operator+(const Point32& b) const
+		{
+			return Point32(x + b.x, y + b.y, z + b.z);
+		}
+
+		Point32 operator-(const Point32& b) const
+		{
+			return Point32(x - b.x, y - b.y, z - b.z);
+		}
+	};
+
+	class Int128
+	{
+	public:
+		btUint64_t low;
+		btUint64_t high;
+
+		Int128()
+		{
+		}
+
+		Int128(btUint64_t low, btUint64_t high) : low(low), high(high)
+		{
+		}
+
+		Int128(btUint64_t low) : low(low), high(0)
+		{
+		}
+
+		Int128(btInt64_t value) : low(value), high((value >= 0) ? 0 : (btUint64_t)-1LL)
+		{
+		}
+
+		static Int128 mul(btInt64_t a, btInt64_t b);
+
+		static Int128 mul(btUint64_t a, btUint64_t b);
+
+		Int128 operator-() const
+		{
+			return Int128((btUint64_t) - (btInt64_t)low, ~high + (low == 0));
+		}
+
+		Int128 operator+(const Int128& b) const
+		{
+#ifdef USE_X86_64_ASM
+			Int128 result;
+			__asm__(
+				"addq %[bl], %[rl]\n\t"
+				"adcq %[bh], %[rh]\n\t"
+				: [rl] "=r"(result.low), [rh] "=r"(result.high)
+				: "0"(low), "1"(high), [bl] "g"(b.low), [bh] "g"(b.high)
+				: "cc");
+			return result;
+#else
+			btUint64_t lo = low + b.low;
+			return Int128(lo, high + b.high + (lo < low));
+#endif
+		}
+
+		Int128 operator-(const Int128& b) const
+		{
+#ifdef USE_X86_64_ASM
+			Int128 result;
+			__asm__(
+				"subq %[bl], %[rl]\n\t"
+				"sbbq %[bh], %[rh]\n\t"
+				: [rl] "=r"(result.low), [rh] "=r"(result.high)
+				: "0"(low), "1"(high), [bl] "g"(b.low), [bh] "g"(b.high)
+				: "cc");
+			return result;
+#else
+			return *this + -b;
+#endif
+		}
+
+		Int128& operator+=(const Int128& b)
+		{
+#ifdef USE_X86_64_ASM
+			__asm__(
+				"addq %[bl], %[rl]\n\t"
+				"adcq %[bh], %[rh]\n\t"
+				: [rl] "=r"(low), [rh] "=r"(high)
+				: "0"(low), "1"(high), [bl] "g"(b.low), [bh] "g"(b.high)
+				: "cc");
+#else
+			btUint64_t lo = low + b.low;
+			if (lo < low)
+			{
+				++high;
+			}
+			low = lo;
+			high += b.high;
+#endif
+			return *this;
+		}
+
+		Int128& operator++()
+		{
+			if (++low == 0)
+			{
+				++high;
+			}
+			return *this;
+		}
+
+		Int128 operator*(btInt64_t b) const;
+
+		b3Scalar toScalar() const
+		{
+			return ((btInt64_t)high >= 0) ? b3Scalar(high) * (b3Scalar(0x100000000LL) * b3Scalar(0x100000000LL)) + b3Scalar(low)
+										  : -(-*this).toScalar();
+		}
+
+		int getSign() const
+		{
+			return ((btInt64_t)high < 0) ? -1 : (high || low) ? 1 : 0;
+		}
+
+		bool operator<(const Int128& b) const
+		{
+			return (high < b.high) || ((high == b.high) && (low < b.low));
+		}
+
+		int ucmp(const Int128& b) const
+		{
+			if (high < b.high)
+			{
+				return -1;
+			}
+			if (high > b.high)
+			{
+				return 1;
+			}
+			if (low < b.low)
+			{
+				return -1;
+			}
+			if (low > b.low)
+			{
+				return 1;
+			}
+			return 0;
+		}
+	};
+
+	class Rational64
+	{
+	private:
+		btUint64_t m_numerator;
+		btUint64_t m_denominator;
+		int sign;
 
 	public:
-		Vertex* vertexList;
+		Rational64(btInt64_t numerator, btInt64_t denominator)
+		{
+			if (numerator > 0)
+			{
+				sign = 1;
+				m_numerator = (btUint64_t)numerator;
+			}
+			else if (numerator < 0)
+			{
+				sign = -1;
+				m_numerator = (btUint64_t)-numerator;
+			}
+			else
+			{
+				sign = 0;
+				m_numerator = 0;
+			}
+			if (denominator > 0)
+			{
+				m_denominator = (btUint64_t)denominator;
+			}
+			else if (denominator < 0)
+			{
+				sign = -sign;
+				m_denominator = (btUint64_t)-denominator;
+			}
+			else
+			{
+				m_denominator = 0;
+			}
+		}
 
-		void compute(const void* coords, bool doubleCoords, int stride, int count);
+		bool isNegativeInfinity() const
+		{
+			return (sign < 0) && (m_denominator == 0);
+		}
 
-		b3Vector3 getCoordinates(const Vertex* v);
+		bool isNaN() const
+		{
+			return (sign == 0) && (m_denominator == 0);
+		}
 
-		b3Scalar shrink(b3Scalar amount, b3Scalar clampAmount);
+		int compare(const Rational64& b) const;
+
+		b3Scalar toScalar() const
+		{
+			return sign * ((m_denominator == 0) ? B3_INFINITY : (b3Scalar)m_numerator / m_denominator);
+		}
+	};
+
+	class Rational128
+	{
+	private:
+		Int128 numerator;
+		Int128 denominator;
+		int sign;
+		bool isInt64;
+
+	public:
+		Rational128(btInt64_t value)
+		{
+			if (value > 0)
+			{
+				sign = 1;
+				this->numerator = value;
+			}
+			else if (value < 0)
+			{
+				sign = -1;
+				this->numerator = -value;
+			}
+			else
+			{
+				sign = 0;
+				this->numerator = (btUint64_t)0;
+			}
+			this->denominator = (btUint64_t)1;
+			isInt64 = true;
+		}
+
+		Rational128(const Int128& numerator, const Int128& denominator)
+		{
+			sign = numerator.getSign();
+			if (sign >= 0)
+			{
+				this->numerator = numerator;
+			}
+			else
+			{
+				this->numerator = -numerator;
+			}
+			int dsign = denominator.getSign();
+			if (dsign >= 0)
+			{
+				this->denominator = denominator;
+			}
+			else
+			{
+				sign = -sign;
+				this->denominator = -denominator;
+			}
+			isInt64 = false;
+		}
+
+		int compare(const Rational128& b) const;
+
+		int compare(btInt64_t b) const;
+
+		b3Scalar toScalar() const
+		{
+			return sign * ((denominator.getSign() == 0) ? B3_INFINITY : numerator.toScalar() / denominator.toScalar());
+		}
+	};
+
+	class PointR128
+	{
+	public:
+		Int128 x;
+		Int128 y;
+		Int128 z;
+		Int128 denominator;
+
+		PointR128()
+		{
+		}
+
+		PointR128(Int128 x, Int128 y, Int128 z, Int128 denominator) : x(x), y(y), z(z), denominator(denominator)
+		{
+		}
+
+		b3Scalar xvalue() const
+		{
+			return x.toScalar() / denominator.toScalar();
+		}
+
+		b3Scalar yvalue() const
+		{
+			return y.toScalar() / denominator.toScalar();
+		}
+
+		b3Scalar zvalue() const
+		{
+			return z.toScalar() / denominator.toScalar();
+		}
+	};
+
+	class Edge;
+	class Face;
+
+	class Vertex
+	{
+	public:
+		Vertex* next;
+		Vertex* prev;
+		Edge* edges;
+		Face* firstNearbyFace;
+		Face* lastNearbyFace;
+		PointR128 point128;
+		Point32 point;
+		int copy;
+
+		Vertex() : next(NULL), prev(NULL), edges(NULL), firstNearbyFace(NULL), lastNearbyFace(NULL), copy(-1)
+		{
+		}
+
+#ifdef DEBUG_CONVEX_HULL
+		void print()
+		{
+			b3Printf("V%d (%d, %d, %d)", point.index, point.x, point.y, point.z);
+		}
+
+		void printGraph();
+#endif
+
+		Point32 operator-(const Vertex& b) const
+		{
+			return point - b.point;
+		}
+
+		Rational128 dot(const Point64& b) const
+		{
+			return (point.index >= 0) ? Rational128(point.dot(b))
+									  : Rational128(point128.x * b.x + point128.y * b.y + point128.z * b.z, point128.denominator);
+		}
+
+		b3Scalar xvalue() const
+		{
+			return (point.index >= 0) ? b3Scalar(point.x) : point128.xvalue();
+		}
+
+		b3Scalar yvalue() const
+		{
+			return (point.index >= 0) ? b3Scalar(point.y) : point128.yvalue();
+		}
+
+		b3Scalar zvalue() const
+		{
+			return (point.index >= 0) ? b3Scalar(point.z) : point128.zvalue();
+		}
+
+		void receiveNearbyFaces(Vertex* src)
+		{
+			if (lastNearbyFace)
+			{
+				lastNearbyFace->nextWithSameNearbyVertex = src->firstNearbyFace;
+			}
+			else
+			{
+				firstNearbyFace = src->firstNearbyFace;
+			}
+			if (src->lastNearbyFace)
+			{
+				lastNearbyFace = src->lastNearbyFace;
+			}
+			for (Face* f = src->firstNearbyFace; f; f = f->nextWithSameNearbyVertex)
+			{
+				b3Assert(f->nearbyVertex == src);
+				f->nearbyVertex = this;
+			}
+			src->firstNearbyFace = NULL;
+			src->lastNearbyFace = NULL;
+		}
+	};
+
+	class Edge
+	{
+	public:
+		Edge* next;
+		Edge* prev;
+		Edge* reverse;
+		Vertex* target;
+		Face* face;
+		int copy;
+
+		~Edge()
+		{
+			next = NULL;
+			prev = NULL;
+			reverse = NULL;
+			target = NULL;
+			face = NULL;
+		}
+
+		void link(Edge* n)
+		{
+			b3Assert(reverse->target == n->reverse->target);
+			next = n;
+			n->prev = this;
+		}
+
+#ifdef DEBUG_CONVEX_HULL
+		void print()
+		{
+			b3Printf("E%p : %d -> %d,  n=%p p=%p   (0 %d\t%d\t%d) -> (%d %d %d)", this, reverse->target->point.index, target->point.index, next, prev,
+					 reverse->target->point.x, reverse->target->point.y, reverse->target->point.z, target->point.x, target->point.y, target->point.z);
+		}
+#endif
+	};
+
+	class Face
+	{
+	public:
+		Face* next;
+		Vertex* nearbyVertex;
+		Face* nextWithSameNearbyVertex;
+		Point32 origin;
+		Point32 dir0;
+		Point32 dir1;
+
+		Face() : next(NULL), nearbyVertex(NULL), nextWithSameNearbyVertex(NULL)
+		{
+		}
+
+		void init(Vertex* a, Vertex* b, Vertex* c)
+		{
+			nearbyVertex = a;
+			origin = a->point;
+			dir0 = *b - *a;
+			dir1 = *c - *a;
+			if (a->lastNearbyFace)
+			{
+				a->lastNearbyFace->nextWithSameNearbyVertex = this;
+			}
+			else
+			{
+				a->firstNearbyFace = this;
+			}
+			a->lastNearbyFace = this;
+		}
+
+		Point64 getNormal()
+		{
+			return dir0.cross(dir1);
+		}
+	};
+
+	template <typename UWord, typename UHWord>
+	class DMul
+	{
+	private:
+		static btUint32_t high(btUint64_t value)
+		{
+			return (btUint32_t)(value >> 32);
+		}
+
+		static btUint32_t low(btUint64_t value)
+		{
+			return (btUint32_t)value;
+		}
+
+		static btUint64_t mul(btUint32_t a, btUint32_t b)
+		{
+			return (btUint64_t)a * (btUint64_t)b;
+		}
+
+		static void shlHalf(btUint64_t& value)
+		{
+			value <<= 32;
+		}
+
+		static btUint64_t high(Int128 value)
+		{
+			return value.high;
+		}
+
+		static btUint64_t low(Int128 value)
+		{
+			return value.low;
+		}
+
+		static Int128 mul(btUint64_t a, btUint64_t b)
+		{
+			return Int128::mul(a, b);
+		}
+
+		static void shlHalf(Int128& value)
+		{
+			value.high = value.low;
+			value.low = 0;
+		}
+
+	public:
+		static void mul(UWord a, UWord b, UWord& resLow, UWord& resHigh)
+		{
+			UWord p00 = mul(low(a), low(b));
+			UWord p01 = mul(low(a), high(b));
+			UWord p10 = mul(high(a), low(b));
+			UWord p11 = mul(high(a), high(b));
+			UWord p0110 = UWord(low(p01)) + UWord(low(p10));
+			p11 += high(p01);
+			p11 += high(p10);
+			p11 += high(p0110);
+			shlHalf(p0110);
+			p00 += p0110;
+			if (p00 < p0110)
+			{
+				++p11;
+			}
+			resLow = p00;
+			resHigh = p11;
+		}
+	};
+
+private:
+	class IntermediateHull
+	{
+	public:
+		Vertex* minXy;
+		Vertex* maxXy;
+		Vertex* minYx;
+		Vertex* maxYx;
+
+		IntermediateHull() : minXy(NULL), maxXy(NULL), minYx(NULL), maxYx(NULL)
+		{
+		}
+
+		void print();
+	};
+
+	enum Orientation
+	{
+		NONE,
+		CLOCKWISE,
+		COUNTER_CLOCKWISE
+	};
+
+	template <typename T>
+	class PoolArray
+	{
+	private:
+		T* array;
+		int size;
+
+	public:
+		PoolArray<T>* next;
+
+		PoolArray(int size) : size(size), next(NULL)
+		{
+			array = (T*)b3AlignedAlloc(sizeof(T) * size, 16);
+		}
+
+		~PoolArray()
+		{
+			b3AlignedFree(array);
+		}
+
+		T* init()
+		{
+			T* o = array;
+			for (int i = 0; i < size; i++, o++)
+			{
+				o->next = (i + 1 < size) ? o + 1 : NULL;
+			}
+			return array;
+		}
+	};
+
+	template <typename T>
+	class Pool
+	{
+	private:
+		PoolArray<T>* arrays;
+		PoolArray<T>* nextArray;
+		T* freeObjects;
+		int arraySize;
+
+	public:
+		Pool() : arrays(NULL), nextArray(NULL), freeObjects(NULL), arraySize(256)
+		{
+		}
+
+		~Pool()
+		{
+			while (arrays)
+			{
+				PoolArray<T>* p = arrays;
+				arrays = p->next;
+				p->~PoolArray<T>();
+				b3AlignedFree(p);
+			}
+		}
+
+		void reset()
+		{
+			nextArray = arrays;
+			freeObjects = NULL;
+		}
+
+		void setArraySize(int arraySize)
+		{
+			this->arraySize = arraySize;
+		}
+
+		T* newObject()
+		{
+			T* o = freeObjects;
+			if (!o)
+			{
+				PoolArray<T>* p = nextArray;
+				if (p)
+				{
+					nextArray = p->next;
+				}
+				else
+				{
+					p = new (b3AlignedAlloc(sizeof(PoolArray<T>), 16)) PoolArray<T>(arraySize);
+					p->next = arrays;
+					arrays = p;
+				}
+				o = p->init();
+			}
+			freeObjects = o->next;
+			return new (o) T();
+		};
+
+		void freeObject(T* object)
+		{
+			object->~T();
+			object->next = freeObjects;
+			freeObjects = object;
+		}
+	};
+
+	b3Vector3 scaling;
+	b3Vector3 center;
+	Pool<Vertex> vertexPool;
+	Pool<Edge> edgePool;
+	Pool<Face> facePool;
+	b3AlignedObjectArray<Vertex*> originalVertices;
+	int mergeStamp;
+	int minAxis;
+	int medAxis;
+	int maxAxis;
+	int usedEdgePairs;
+	int maxUsedEdgePairs;
+
+	static Orientation getOrientation(const Edge* prev, const Edge* next, const Point32& s, const Point32& t);
+	Edge* findMaxAngle(bool ccw, const Vertex* start, const Point32& s, const Point64& rxs, const Point64& sxrxs, Rational64& minCot);
+	void findEdgeForCoplanarFaces(Vertex* c0, Vertex* c1, Edge*& e0, Edge*& e1, Vertex* stop0, Vertex* stop1);
+
+	Edge* newEdgePair(Vertex* from, Vertex* to);
+
+	void removeEdgePair(Edge* edge)
+	{
+		Edge* n = edge->next;
+		Edge* r = edge->reverse;
+
+		b3Assert(edge->target && r->target);
+
+		if (n != edge)
+		{
+			n->prev = edge->prev;
+			edge->prev->next = n;
+			r->target->edges = n;
+		}
+		else
+		{
+			r->target->edges = NULL;
+		}
+
+		n = r->next;
+
+		if (n != r)
+		{
+			n->prev = r->prev;
+			r->prev->next = n;
+			edge->target->edges = n;
+		}
+		else
+		{
+			edge->target->edges = NULL;
+		}
+
+		edgePool.freeObject(edge);
+		edgePool.freeObject(r);
+		usedEdgePairs--;
+	}
+
+	void computeInternal(int start, int end, IntermediateHull& result);
+
+	bool mergeProjection(IntermediateHull& h0, IntermediateHull& h1, Vertex*& c0, Vertex*& c1);
+
+	void merge(IntermediateHull& h0, IntermediateHull& h1);
+
+	b3Vector3 toBtVector(const Point32& v);
+
+	b3Vector3 getBtNormal(Face* face);
+
+	bool shiftFace(Face* face, b3Scalar amount, b3AlignedObjectArray<Vertex*> stack);
+
+public:
+	Vertex* vertexList;
+
+	void compute(const void* coords, bool doubleCoords, int stride, int count);
+
+	b3Vector3 getCoordinates(const Vertex* v);
+
+	b3Scalar shrink(b3Scalar amount, b3Scalar clampAmount);
 };
 
-
 b3ConvexHullInternal::Int128 b3ConvexHullInternal::Int128::operator*(btInt64_t b) const
 {
-	bool negative = (btInt64_t) high < 0;
+	bool negative = (btInt64_t)high < 0;
 	Int128 a = negative ? -*this : *this;
 	if (b < 0)
 	{
 		negative = !negative;
 		b = -b;
 	}
-	Int128 result = mul(a.low, (btUint64_t) b);
-	result.high += a.high * (btUint64_t) b;
+	Int128 result = mul(a.low, (btUint64_t)b);
+	result.high += a.high * (btUint64_t)b;
 	return negative ? -result : result;
 }
 
 b3ConvexHullInternal::Int128 b3ConvexHullInternal::Int128::mul(btInt64_t a, btInt64_t b)
 {
 	Int128 result;
-	
+
 #ifdef USE_X86_64_ASM
-	__asm__ ("imulq %[b]"
-					 : "=a" (result.low), "=d" (result.high)
-					 : "0"(a), [b] "r"(b)
-					 : "cc" );
+	__asm__("imulq %[b]"
+			: "=a"(result.low), "=d"(result.high)
+			: "0"(a), [b] "r"(b)
+			: "cc");
 	return result;
-	
+
 #else
 	bool negative = a < 0;
 	if (negative)
@@ -875,7 +876,7 @@ b3ConvexHullInternal::Int128 b3ConvexHullInternal::Int128::mul(btInt64_t a, btIn
 		negative = !negative;
 		b = -b;
 	}
-	DMul<btUint64_t, btUint32_t>::mul((btUint64_t) a, (btUint64_t) b, result.low, result.high);
+	DMul<btUint64_t, btUint32_t>::mul((btUint64_t)a, (btUint64_t)b, result.low, result.high);
 	return negative ? -result : result;
 #endif
 }
@@ -885,10 +886,10 @@ b3ConvexHullInternal::Int128 b3ConvexHullInternal::Int128::mul(btUint64_t a, btU
 	Int128 result;
 
 #ifdef USE_X86_64_ASM
-	__asm__ ("mulq %[b]"
-					 : "=a" (result.low), "=d" (result.high)
-					 : "0"(a), [b] "r"(b)
-					 : "cc" );
+	__asm__("mulq %[b]"
+			: "=a"(result.low), "=d"(result.high)
+			: "0"(a), [b] "r"(b)
+			: "cc");
 
 #else
 	DMul<btUint64_t, btUint32_t>::mul(a, b, result.low, result.high);
@@ -915,24 +916,25 @@ int b3ConvexHullInternal::Rational64::compare(const Rational64& b) const
 	int result;
 	btInt64_t tmp;
 	btInt64_t dummy;
-	__asm__ ("mulq %[bn]\n\t"
-					 "movq %%rax, %[tmp]\n\t"
-					 "movq %%rdx, %%rbx\n\t"
-					 "movq %[tn], %%rax\n\t"
-					 "mulq %[bd]\n\t"
-					 "subq %[tmp], %%rax\n\t"
-					 "sbbq %%rbx, %%rdx\n\t" // rdx:rax contains 128-bit-difference "numerator*b.denominator - b.numerator*denominator"
-					 "setnsb %%bh\n\t" // bh=1 if difference is non-negative, bh=0 otherwise
-					 "orq %%rdx, %%rax\n\t"
-					 "setnzb %%bl\n\t" // bl=1 if difference if non-zero, bl=0 if it is zero
-					 "decb %%bh\n\t" // now bx=0x0000 if difference is zero, 0xff01 if it is negative, 0x0001 if it is positive (i.e., same sign as difference)
-					 "shll $16, %%ebx\n\t" // ebx has same sign as difference
-					 : "=&b"(result), [tmp] "=&r"(tmp), "=a"(dummy)
-					 : "a"(denominator), [bn] "g"(b.numerator), [tn] "g"(numerator), [bd] "g"(b.denominator)
-					 : "%rdx", "cc" );
-	return result ? result ^ sign // if sign is +1, only bit 0 of result is inverted, which does not change the sign of result (and cannot result in zero)
-																// if sign is -1, all bits of result are inverted, which changes the sign of result (and again cannot result in zero)
-								: 0;
+	__asm__(
+		"mulq %[bn]\n\t"
+		"movq %%rax, %[tmp]\n\t"
+		"movq %%rdx, %%rbx\n\t"
+		"movq %[tn], %%rax\n\t"
+		"mulq %[bd]\n\t"
+		"subq %[tmp], %%rax\n\t"
+		"sbbq %%rbx, %%rdx\n\t"  // rdx:rax contains 128-bit-difference "numerator*b.denominator - b.numerator*denominator"
+		"setnsb %%bh\n\t"        // bh=1 if difference is non-negative, bh=0 otherwise
+		"orq %%rdx, %%rax\n\t"
+		"setnzb %%bl\n\t"      // bl=1 if difference if non-zero, bl=0 if it is zero
+		"decb %%bh\n\t"        // now bx=0x0000 if difference is zero, 0xff01 if it is negative, 0x0001 if it is positive (i.e., same sign as difference)
+		"shll $16, %%ebx\n\t"  // ebx has same sign as difference
+		: "=&b"(result), [tmp] "=&r"(tmp), "=a"(dummy)
+		: "a"(denominator), [bn] "g"(b.numerator), [tn] "g"(numerator), [bd] "g"(b.denominator)
+		: "%rdx", "cc");
+	return result ? result ^ sign  // if sign is +1, only bit 0 of result is inverted, which does not change the sign of result (and cannot result in zero)
+								   // if sign is -1, all bits of result are inverted, which changes the sign of result (and again cannot result in zero)
+				  : 0;
 
 #else
 
@@ -953,7 +955,7 @@ int b3ConvexHullInternal::Rational128::compare(const Rational128& b) const
 	}
 	if (isInt64)
 	{
-		return -b.compare(sign * (btInt64_t) numerator.low);
+		return -b.compare(sign * (btInt64_t)numerator.low);
 	}
 
 	Int128 nbdLow, nbdHigh, dbnLow, dbnHigh;
@@ -972,7 +974,7 @@ int b3ConvexHullInternal::Rational128::compare(btInt64_t b) const
 {
 	if (isInt64)
 	{
-		btInt64_t a = sign * (btInt64_t) numerator.low;
+		btInt64_t a = sign * (btInt64_t)numerator.low;
 		return (a > b) ? 1 : (a < b) ? -1 : 0;
 	}
 	if (b > 0)
@@ -998,7 +1000,6 @@ int b3ConvexHullInternal::Rational128::compare(btInt64_t b) const
 	return numerator.ucmp(denominator * b) * sign;
 }
 
-
 b3ConvexHullInternal::Edge* b3ConvexHullInternal::newEdgePair(Vertex* from, Vertex* to)
 {
 	b3Assert(from && to);
@@ -1066,7 +1067,7 @@ bool b3ConvexHullInternal::mergeProjection(IntermediateHull& h0, IntermediateHul
 			}
 		}
 	}
-	
+
 	v0 = h0.maxXy;
 	v1 = h1.maxXy;
 	Vertex* v00 = NULL;
@@ -1074,7 +1075,7 @@ bool b3ConvexHullInternal::mergeProjection(IntermediateHull& h0, IntermediateHul
 	btInt32_t sign = 1;
 
 	for (int side = 0; side <= 1; side++)
-	{		
+	{
 		btInt32_t dx = (v1->point.x - v0->point.x) * sign;
 		if (dx > 0)
 		{
@@ -1117,7 +1118,7 @@ bool b3ConvexHullInternal::mergeProjection(IntermediateHull& h0, IntermediateHul
 			while (true)
 			{
 				btInt32_t dy = v1->point.y - v0->point.y;
-				
+
 				Vertex* w1 = side ? v1->prev : v1->next;
 				if (w1 != v1)
 				{
@@ -1130,7 +1131,7 @@ bool b3ConvexHullInternal::mergeProjection(IntermediateHull& h0, IntermediateHul
 						continue;
 					}
 				}
-				
+
 				Vertex* w0 = side ? v0->prev : v0->next;
 				if (w0 != v0)
 				{
@@ -1144,7 +1145,7 @@ bool b3ConvexHullInternal::mergeProjection(IntermediateHull& h0, IntermediateHul
 						continue;
 					}
 				}
-				
+
 				break;
 			}
 		}
@@ -1170,7 +1171,7 @@ bool b3ConvexHullInternal::mergeProjection(IntermediateHull& h0, IntermediateHul
 			}
 			v1 = w1;
 		}
-		
+
 		if (side == 0)
 		{
 			v00 = v0;
@@ -1196,7 +1197,7 @@ bool b3ConvexHullInternal::mergeProjection(IntermediateHull& h0, IntermediateHul
 	{
 		h0.maxXy = h1.maxXy;
 	}
-	
+
 	h0.maxYx = h1.maxYx;
 
 	c0 = v00;
@@ -1300,7 +1301,7 @@ void b3ConvexHullInternal::computeInternal(int start, int end, IntermediateHull&
 	}
 
 	int split0 = start + n / 2;
-	Point32 p = originalVertices[split0-1]->point;
+	Point32 p = originalVertices[split0 - 1]->point;
 	int split1 = split0;
 	while ((split1 < end) && (originalVertices[split1]->point == p))
 	{
@@ -1325,7 +1326,7 @@ void b3ConvexHullInternal::computeInternal(int start, int end, IntermediateHull&
 void b3ConvexHullInternal::IntermediateHull::print()
 {
 	b3Printf("    Hull\n");
-	for (Vertex* v = minXy; v; )
+	for (Vertex* v = minXy; v;)
 	{
 		b3Printf("      ");
 		v->print();
@@ -1353,7 +1354,7 @@ void b3ConvexHullInternal::IntermediateHull::print()
 		}
 	}
 	if (minXy)
-	{		
+	{
 		minXy->copy = (minXy->copy == -1) ? -2 : -1;
 		minXy->printGraph();
 	}
@@ -1429,7 +1430,7 @@ b3ConvexHullInternal::Edge* b3ConvexHullInternal::findMaxAngle(bool ccw, const V
 				Point32 t = *e->target - *start;
 				Rational64 cot(t.dot(sxrxs), t.dot(rxs));
 #ifdef DEBUG_CONVEX_HULL
-				b3Printf("      Angle is %f (%d) for ", (float) b3Atan(cot.toScalar()), (int) cot.isNaN());
+				b3Printf("      Angle is %f (%d) for ", (float)b3Atan(cot.toScalar()), (int)cot.isNaN());
 				e->print();
 #endif
 				if (cot.isNaN())
@@ -1476,7 +1477,7 @@ void b3ConvexHullInternal::findEdgeForCoplanarFaces(Vertex* c0, Vertex* c1, Edge
 	b3Assert(!start1 || (start1->target->point.dot(normal) == dist));
 	Point64 perp = s.cross(normal);
 	b3Assert(!perp.isZero());
-	
+
 #ifdef DEBUG_CONVEX_HULL
 	b3Printf("   Advancing %d %d  (%p %p, %d %d)\n", c0->point.index, c1->point.index, start0, start1, start0 ? start0->target->point.index : -1, start1 ? start1->target->point.index : -1);
 #endif
@@ -1506,7 +1507,7 @@ void b3ConvexHullInternal::findEdgeForCoplanarFaces(Vertex* c0, Vertex* c1, Edge
 			et0 = e->target->point;
 		}
 	}
-	
+
 	btInt64_t maxDot1 = et1.dot(perp);
 	if (e1)
 	{
@@ -1543,7 +1544,7 @@ void b3ConvexHullInternal::findEdgeForCoplanarFaces(Vertex* c0, Vertex* c1, Edge
 		while (true)
 		{
 			btInt64_t dy = (et1 - et0).dot(s);
-			
+
 			if (e0 && (e0->target != stop0))
 			{
 				Edge* f0 = e0->next->reverse;
@@ -1560,7 +1561,7 @@ void b3ConvexHullInternal::findEdgeForCoplanarFaces(Vertex* c0, Vertex* c1, Edge
 					}
 				}
 			}
-			
+
 			if (e1 && (e1->target != stop1))
 			{
 				Edge* f1 = e1->reverse->next;
@@ -1595,7 +1596,7 @@ void b3ConvexHullInternal::findEdgeForCoplanarFaces(Vertex* c0, Vertex* c1, Edge
 		while (true)
 		{
 			btInt64_t dy = (et1 - et0).dot(s);
-			
+
 			if (e1 && (e1->target != stop1))
 			{
 				Edge* f1 = e1->prev->reverse;
@@ -1612,7 +1613,7 @@ void b3ConvexHullInternal::findEdgeForCoplanarFaces(Vertex* c0, Vertex* c1, Edge
 					}
 				}
 			}
-			
+
 			if (e0 && (e0->target != stop0))
 			{
 				Edge* f0 = e0->reverse->prev;
@@ -1647,7 +1648,6 @@ void b3ConvexHullInternal::findEdgeForCoplanarFaces(Vertex* c0, Vertex* c1, Edge
 #endif
 }
 
-
 void b3ConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 {
 	if (!h1.maxXy)
@@ -1659,7 +1659,7 @@ void b3ConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 		h0 = h1;
 		return;
 	}
-	
+
 	mergeStamp--;
 
 	Vertex* c0 = NULL;
@@ -1699,7 +1699,7 @@ void b3ConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 				e = e->next;
 			} while (e != c0->edges);
 		}
-		
+
 		e = c1->edges;
 		Edge* start1 = NULL;
 		if (e)
@@ -1751,7 +1751,7 @@ void b3ConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 		Point32 r = prevPoint - c0->point;
 		Point64 rxs = r.cross(s);
 		Point64 sxrxs = s.cross(rxs);
-		
+
 #ifdef DEBUG_CONVEX_HULL
 		b3Printf("\n  Checking %d %d\n", c0->point.index, c1->point.index);
 #endif
@@ -1802,7 +1802,7 @@ void b3ConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 				e->prev = pendingTail1;
 				pendingTail1 = e;
 			}
-			
+
 			Edge* e0 = min0;
 			Edge* e1 = min1;
 
@@ -1819,7 +1819,7 @@ void b3ConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 			{
 				if (toPrev1)
 				{
-					for (Edge* e = toPrev1->next, *n = NULL; e != min1; e = n)
+					for (Edge *e = toPrev1->next, *n = NULL; e != min1; e = n)
 					{
 						n = e->next;
 						removeEdgePair(e);
@@ -1855,7 +1855,7 @@ void b3ConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 			{
 				if (toPrev0)
 				{
-					for (Edge* e = toPrev0->prev, *n = NULL; e != min0; e = n)
+					for (Edge *e = toPrev0->prev, *n = NULL; e != min0; e = n)
 					{
 						n = e->prev;
 						removeEdgePair(e);
@@ -1897,7 +1897,7 @@ void b3ConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 			}
 			else
 			{
-				for (Edge* e = toPrev0->prev, *n = NULL; e != firstNew0; e = n)
+				for (Edge *e = toPrev0->prev, *n = NULL; e != firstNew0; e = n)
 				{
 					n = e->prev;
 					removeEdgePair(e);
@@ -1916,7 +1916,7 @@ void b3ConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 			}
 			else
 			{
-				for (Edge* e = toPrev1->next, *n = NULL; e != firstNew1; e = n)
+				for (Edge *e = toPrev1->next, *n = NULL; e != firstNew1; e = n)
 				{
 					n = e->next;
 					removeEdgePair(e);
@@ -1927,7 +1927,7 @@ void b3ConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 					pendingTail1->link(firstNew1);
 				}
 			}
-			
+
 			return;
 		}
 
@@ -1935,7 +1935,6 @@ void b3ConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 	}
 }
 
-
 static bool b3PointCmp(const b3ConvexHullInternal::Point32& p, const b3ConvexHullInternal::Point32& q)
 {
 	return (p.y < q.y) || ((p.y == q.y) && ((p.x < q.x) || ((p.x == q.x) && (p.z < q.z))));
@@ -1943,14 +1942,14 @@ static bool b3PointCmp(const b3ConvexHullInternal::Point32& p, const b3ConvexHul
 
 void b3ConvexHullInternal::compute(const void* coords, bool doubleCoords, int stride, int count)
 {
-	b3Vector3 min = b3MakeVector3(b3Scalar(1e30), b3Scalar(1e30), b3Scalar(1e30)),  max = b3MakeVector3(b3Scalar(-1e30), b3Scalar(-1e30), b3Scalar(-1e30));
-	const char* ptr = (const char*) coords;
+	b3Vector3 min = b3MakeVector3(b3Scalar(1e30), b3Scalar(1e30), b3Scalar(1e30)), max = b3MakeVector3(b3Scalar(-1e30), b3Scalar(-1e30), b3Scalar(-1e30));
+	const char* ptr = (const char*)coords;
 	if (doubleCoords)
 	{
 		for (int i = 0; i < count; i++)
 		{
-			const double* v = (const double*) ptr;
-			b3Vector3 p = b3MakeVector3((b3Scalar) v[0], (b3Scalar) v[1], (b3Scalar) v[2]);
+			const double* v = (const double*)ptr;
+			b3Vector3 p = b3MakeVector3((b3Scalar)v[0], (b3Scalar)v[1], (b3Scalar)v[2]);
 			ptr += stride;
 			min.setMin(p);
 			max.setMax(p);
@@ -1960,7 +1959,7 @@ void b3ConvexHullInternal::compute(const void* coords, bool doubleCoords, int st
 	{
 		for (int i = 0; i < count; i++)
 		{
-			const float* v = (const float*) ptr;
+			const float* v = (const float*)ptr;
 			b3Vector3 p = b3MakeVector3(v[0], v[1], v[2]);
 			ptr += stride;
 			min.setMin(p);
@@ -2001,18 +2000,18 @@ void b3ConvexHullInternal::compute(const void* coords, bool doubleCoords, int st
 
 	b3AlignedObjectArray<Point32> points;
 	points.resize(count);
-	ptr = (const char*) coords;
+	ptr = (const char*)coords;
 	if (doubleCoords)
 	{
 		for (int i = 0; i < count; i++)
 		{
-			const double* v = (const double*) ptr;
-			b3Vector3 p = b3MakeVector3((b3Scalar) v[0], (b3Scalar) v[1], (b3Scalar) v[2]);
+			const double* v = (const double*)ptr;
+			b3Vector3 p = b3MakeVector3((b3Scalar)v[0], (b3Scalar)v[1], (b3Scalar)v[2]);
 			ptr += stride;
 			p = (p - center) * s;
-			points[i].x = (btInt32_t) p[medAxis];
-			points[i].y = (btInt32_t) p[maxAxis];
-			points[i].z = (btInt32_t) p[minAxis];
+			points[i].x = (btInt32_t)p[medAxis];
+			points[i].y = (btInt32_t)p[maxAxis];
+			points[i].z = (btInt32_t)p[minAxis];
 			points[i].index = i;
 		}
 	}
@@ -2020,13 +2019,13 @@ void b3ConvexHullInternal::compute(const void* coords, bool doubleCoords, int st
 	{
 		for (int i = 0; i < count; i++)
 		{
-			const float* v = (const float*) ptr;
+			const float* v = (const float*)ptr;
 			b3Vector3 p = b3MakeVector3(v[0], v[1], v[2]);
 			ptr += stride;
 			p = (p - center) * s;
-			points[i].x = (btInt32_t) p[medAxis];
-			points[i].y = (btInt32_t) p[maxAxis];
-			points[i].z = (btInt32_t) p[minAxis];
+			points[i].x = (btInt32_t)p[medAxis];
+			points[i].y = (btInt32_t)p[maxAxis];
+			points[i].z = (btInt32_t)p[minAxis];
 			points[i].index = i;
 		}
 	}
@@ -2180,7 +2179,7 @@ b3Scalar b3ConvexHullInternal::shrink(b3Scalar amount, b3Scalar clampAmount)
 				minDist = dist;
 			}
 		}
-		
+
 		if (minDist <= 0)
 		{
 			return 0;
@@ -2221,7 +2220,7 @@ bool b3ConvexHullInternal::shiftFace(Face* face, b3Scalar amount, b3AlignedObjec
 	{
 		origShift[2] /= scaling[2];
 	}
-	Point32 shift((btInt32_t) origShift[medAxis], (btInt32_t) origShift[maxAxis], (btInt32_t) origShift[minAxis]);
+	Point32 shift((btInt32_t)origShift[medAxis], (btInt32_t)origShift[maxAxis], (btInt32_t)origShift[minAxis]);
 	if (shift.isZero())
 	{
 		return true;
@@ -2229,7 +2228,7 @@ bool b3ConvexHullInternal::shiftFace(Face* face, b3Scalar amount, b3AlignedObjec
 	Point64 normal = face->getNormal();
 #ifdef DEBUG_CONVEX_HULL
 	b3Printf("\nShrinking face (%d %d %d) (%d %d %d) (%d %d %d) by (%d %d %d)\n",
-				 face->origin.x, face->origin.y, face->origin.z, face->dir0.x, face->dir0.y, face->dir0.z, face->dir1.x, face->dir1.y, face->dir1.z, shift.x, shift.y, shift.z);
+			 face->origin.x, face->origin.y, face->origin.z, face->dir0.x, face->dir0.y, face->dir0.z, face->dir1.x, face->dir1.y, face->dir1.z, shift.x, shift.y, shift.z);
 #endif
 	btInt64_t origDot = face->origin.dot(normal);
 	Point32 shiftedOrigin = face->origin + shift;
@@ -2266,7 +2265,7 @@ bool b3ConvexHullInternal::shiftFace(Face* face, b3Scalar amount, b3AlignedObjec
 #ifdef DEBUG_CONVEX_HULL
 			b3Printf("Moving downwards, edge is ");
 			e->print();
-			b3Printf(", dot is %f (%f %lld)\n", (float) dot.toScalar(), (float) optDot.toScalar(), shiftedDot);
+			b3Printf(", dot is %f (%f %lld)\n", (float)dot.toScalar(), (float)optDot.toScalar(), shiftedDot);
 #endif
 			if (dot.compare(optDot) < 0)
 			{
@@ -2302,7 +2301,7 @@ bool b3ConvexHullInternal::shiftFace(Face* face, b3Scalar amount, b3AlignedObjec
 #ifdef DEBUG_CONVEX_HULL
 			b3Printf("Moving upwards, edge is ");
 			e->print();
-			b3Printf(", dot is %f (%f %lld)\n", (float) dot.toScalar(), (float) optDot.toScalar(), shiftedDot);
+			b3Printf(", dot is %f (%f %lld)\n", (float)dot.toScalar(), (float)optDot.toScalar(), shiftedDot);
 #endif
 			if (dot.compare(optDot) > 0)
 			{
@@ -2318,7 +2317,7 @@ bool b3ConvexHullInternal::shiftFace(Face* face, b3Scalar amount, b3AlignedObjec
 			}
 			e = e->prev;
 		} while (e != startEdge);
-		
+
 		if (!intersection)
 		{
 			return true;
@@ -2355,7 +2354,7 @@ bool b3ConvexHullInternal::shiftFace(Face* face, b3Scalar amount, b3AlignedObjec
 		b3Printf("Needed %d iterations to check for complete containment\n", n);
 #endif
 	}
-	
+
 	Edge* firstIntersection = NULL;
 	Edge* faceEdge = NULL;
 	Edge* firstFaceEdge = NULL;
@@ -2464,7 +2463,7 @@ bool b3ConvexHullInternal::shiftFace(Face* face, b3Scalar amount, b3AlignedObjec
 #ifdef DEBUG_CONVEX_HULL
 			b3Printf("1: Removed part contains (%d %d %d)\n", removed->point.x, removed->point.y, removed->point.z);
 #endif
-			
+
 			Point64 n0 = intersection->face->getNormal();
 			Point64 n1 = intersection->reverse->face->getNormal();
 			btInt64_t m00 = face->dir0.dot(n0);
@@ -2478,16 +2477,13 @@ bool b3ConvexHullInternal::shiftFace(Face* face, b3Scalar amount, b3AlignedObjec
 			Vertex* v = vertexPool.newObject();
 			v->point.index = -1;
 			v->copy = -1;
-			v->point128 = PointR128(Int128::mul(face->dir0.x * r0, m11) - Int128::mul(face->dir0.x * r1, m01)
-															+ Int128::mul(face->dir1.x * r1, m00) - Int128::mul(face->dir1.x * r0, m10) + det * shiftedOrigin.x,
-															Int128::mul(face->dir0.y * r0, m11) - Int128::mul(face->dir0.y * r1, m01)
-															+ Int128::mul(face->dir1.y * r1, m00) - Int128::mul(face->dir1.y * r0, m10) + det * shiftedOrigin.y,
-															Int128::mul(face->dir0.z * r0, m11) - Int128::mul(face->dir0.z * r1, m01)
-															+ Int128::mul(face->dir1.z * r1, m00) - Int128::mul(face->dir1.z * r0, m10) + det * shiftedOrigin.z,
-															det);
-			v->point.x = (btInt32_t) v->point128.xvalue();
-			v->point.y = (btInt32_t) v->point128.yvalue();
-			v->point.z = (btInt32_t) v->point128.zvalue();
+			v->point128 = PointR128(Int128::mul(face->dir0.x * r0, m11) - Int128::mul(face->dir0.x * r1, m01) + Int128::mul(face->dir1.x * r1, m00) - Int128::mul(face->dir1.x * r0, m10) + det * shiftedOrigin.x,
+									Int128::mul(face->dir0.y * r0, m11) - Int128::mul(face->dir0.y * r1, m01) + Int128::mul(face->dir1.y * r1, m00) - Int128::mul(face->dir1.y * r0, m10) + det * shiftedOrigin.y,
+									Int128::mul(face->dir0.z * r0, m11) - Int128::mul(face->dir0.z * r1, m01) + Int128::mul(face->dir1.z * r1, m00) - Int128::mul(face->dir1.z * r0, m10) + det * shiftedOrigin.z,
+									det);
+			v->point.x = (btInt32_t)v->point128.xvalue();
+			v->point.y = (btInt32_t)v->point128.yvalue();
+			v->point.z = (btInt32_t)v->point128.zvalue();
 			intersection->target = v;
 			v->edges = e;
 
@@ -2626,7 +2622,6 @@ bool b3ConvexHullInternal::shiftFace(Face* face, b3Scalar amount, b3AlignedObjec
 	return true;
 }
 
-
 static int getVertexCopy(b3ConvexHullInternal::Vertex* vertex, b3AlignedObjectArray<b3ConvexHullInternal::Vertex*>& vertices)
 {
 	int index = vertex->copy;
@@ -2748,8 +2743,3 @@ b3Scalar b3ConvexHullComputer::compute(const void* coords, bool doubleCoords, in
 
 	return shift;
 }
-
-
-
-
-
diff --git a/thirdparty/bullet/Bullet3Geometry/b3ConvexHullComputer.h b/thirdparty/bullet/Bullet3Geometry/b3ConvexHullComputer.h
index 6dcc931a78c..8852c5a524f 100644
--- a/thirdparty/bullet/Bullet3Geometry/b3ConvexHullComputer.h
+++ b/thirdparty/bullet/Bullet3Geometry/b3ConvexHullComputer.h
@@ -23,58 +23,56 @@ subject to the following restrictions:
 /// Ole Kniemeyer, MAXON Computer GmbH
 class b3ConvexHullComputer
 {
+private:
+	b3Scalar compute(const void* coords, bool doubleCoords, int stride, int count, b3Scalar shrink, b3Scalar shrinkClamp);
+
+public:
+	class Edge
+	{
 	private:
-		b3Scalar compute(const void* coords, bool doubleCoords, int stride, int count, b3Scalar shrink, b3Scalar shrinkClamp);
+		int next;
+		int reverse;
+		int targetVertex;
+
+		friend class b3ConvexHullComputer;
 
 	public:
-
-		class Edge
+		int getSourceVertex() const
 		{
-			private:
-				int next;
-				int reverse;
-				int targetVertex;
+			return (this + reverse)->targetVertex;
+		}
 
-				friend class b3ConvexHullComputer;
+		int getTargetVertex() const
+		{
+			return targetVertex;
+		}
 
-			public:
-				int getSourceVertex() const
-				{
-					return (this + reverse)->targetVertex;
-				}
+		const Edge* getNextEdgeOfVertex() const  // clockwise list of all edges of a vertex
+		{
+			return this + next;
+		}
 
-				int getTargetVertex() const
-				{
-					return targetVertex;
-				}
+		const Edge* getNextEdgeOfFace() const  // counter-clockwise list of all edges of a face
+		{
+			return (this + reverse)->getNextEdgeOfVertex();
+		}
 
-				const Edge* getNextEdgeOfVertex() const // clockwise list of all edges of a vertex
-				{
-					return this + next;
-				}
+		const Edge* getReverseEdge() const
+		{
+			return this + reverse;
+		}
+	};
 
-				const Edge* getNextEdgeOfFace() const // counter-clockwise list of all edges of a face
-				{
-					return (this + reverse)->getNextEdgeOfVertex();
-				}
+	// Vertices of the output hull
+	b3AlignedObjectArray<b3Vector3> vertices;
 
-				const Edge* getReverseEdge() const
-				{
-					return this + reverse;
-				}
-		};
+	// Edges of the output hull
+	b3AlignedObjectArray<Edge> edges;
 
+	// Faces of the convex hull. Each entry is an index into the "edges" array pointing to an edge of the face. Faces are planar n-gons
+	b3AlignedObjectArray<int> faces;
 
-		// Vertices of the output hull
-		b3AlignedObjectArray<b3Vector3> vertices;
-
-		// Edges of the output hull
-		b3AlignedObjectArray<Edge> edges;
-
-		// Faces of the convex hull. Each entry is an index into the "edges" array pointing to an edge of the face. Faces are planar n-gons
-		b3AlignedObjectArray<int> faces;
-
-		/*
+	/*
 		Compute convex hull of "count" vertices stored in "coords". "stride" is the difference in bytes
 		between the addresses of consecutive vertices. If "shrink" is positive, the convex hull is shrunken
 		by that amount (each face is moved by "shrink" length units towards the center along its normal).
@@ -86,18 +84,16 @@ class b3ConvexHullComputer
 
 		The output convex hull can be found in the member variables "vertices", "edges", "faces".
 		*/
-		b3Scalar compute(const float* coords, int stride, int count, b3Scalar shrink, b3Scalar shrinkClamp)
-		{
-			return compute(coords, false, stride, count, shrink, shrinkClamp);
-		}
+	b3Scalar compute(const float* coords, int stride, int count, b3Scalar shrink, b3Scalar shrinkClamp)
+	{
+		return compute(coords, false, stride, count, shrink, shrinkClamp);
+	}
 
-		// same as above, but double precision
-		b3Scalar compute(const double* coords, int stride, int count, b3Scalar shrink, b3Scalar shrinkClamp)
-		{
-			return compute(coords, true, stride, count, shrink, shrinkClamp);
-		}
+	// same as above, but double precision
+	b3Scalar compute(const double* coords, int stride, int count, b3Scalar shrink, b3Scalar shrinkClamp)
+	{
+		return compute(coords, true, stride, count, shrink, shrinkClamp);
+	}
 };
 
-
-#endif //B3_CONVEX_HULL_COMPUTER_H
-
+#endif  //B3_CONVEX_HULL_COMPUTER_H
diff --git a/thirdparty/bullet/Bullet3Geometry/b3GeometryUtil.cpp b/thirdparty/bullet/Bullet3Geometry/b3GeometryUtil.cpp
index dd80fed6bd9..c4041003caf 100644
--- a/thirdparty/bullet/Bullet3Geometry/b3GeometryUtil.cpp
+++ b/thirdparty/bullet/Bullet3Geometry/b3GeometryUtil.cpp
@@ -12,49 +12,43 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #include "b3GeometryUtil.h"
 
-
 /*
   Make sure this dummy function never changes so that it
   can be used by probes that are checking whether the
   library is actually installed.
 */
 extern "C"
-{	
-	void b3BulletMathProbe ();
+{
+	void b3BulletMathProbe();
 
-	void b3BulletMathProbe () {}
+	void b3BulletMathProbe() {}
 }
 
-
-bool	b3GeometryUtil::isPointInsidePlanes(const b3AlignedObjectArray<b3Vector3>& planeEquations, const b3Vector3& point, b3Scalar	margin)
+bool b3GeometryUtil::isPointInsidePlanes(const b3AlignedObjectArray<b3Vector3>& planeEquations, const b3Vector3& point, b3Scalar margin)
 {
 	int numbrushes = planeEquations.size();
-	for (int i=0;i<numbrushes;i++)
+	for (int i = 0; i < numbrushes; i++)
 	{
 		const b3Vector3& N1 = planeEquations[i];
-		b3Scalar dist = b3Scalar(N1.dot(point))+b3Scalar(N1[3])-margin;
-		if (dist>b3Scalar(0.))
+		b3Scalar dist = b3Scalar(N1.dot(point)) + b3Scalar(N1[3]) - margin;
+		if (dist > b3Scalar(0.))
 		{
 			return false;
 		}
 	}
 	return true;
-		
 }
 
-
-bool	b3GeometryUtil::areVerticesBehindPlane(const b3Vector3& planeNormal, const b3AlignedObjectArray<b3Vector3>& vertices, b3Scalar	margin)
+bool b3GeometryUtil::areVerticesBehindPlane(const b3Vector3& planeNormal, const b3AlignedObjectArray<b3Vector3>& vertices, b3Scalar margin)
 {
 	int numvertices = vertices.size();
-	for (int i=0;i<numvertices;i++)
+	for (int i = 0; i < numvertices; i++)
 	{
 		const b3Vector3& N1 = vertices[i];
-		b3Scalar dist = b3Scalar(planeNormal.dot(N1))+b3Scalar(planeNormal[3])-margin;
-		if (dist>b3Scalar(0.))
+		b3Scalar dist = b3Scalar(planeNormal.dot(N1)) + b3Scalar(planeNormal[3]) - margin;
+		if (dist > b3Scalar(0.))
 		{
 			return false;
 		}
@@ -62,102 +56,98 @@ bool	b3GeometryUtil::areVerticesBehindPlane(const b3Vector3& planeNormal, const
 	return true;
 }
 
-bool notExist(const b3Vector3& planeEquation,const b3AlignedObjectArray<b3Vector3>& planeEquations);
+bool notExist(const b3Vector3& planeEquation, const b3AlignedObjectArray<b3Vector3>& planeEquations);
 
-bool notExist(const b3Vector3& planeEquation,const b3AlignedObjectArray<b3Vector3>& planeEquations)
+bool notExist(const b3Vector3& planeEquation, const b3AlignedObjectArray<b3Vector3>& planeEquations)
 {
 	int numbrushes = planeEquations.size();
-	for (int i=0;i<numbrushes;i++)
+	for (int i = 0; i < numbrushes; i++)
 	{
 		const b3Vector3& N1 = planeEquations[i];
 		if (planeEquation.dot(N1) > b3Scalar(0.999))
 		{
 			return false;
-		} 
+		}
 	}
 	return true;
 }
 
-void	b3GeometryUtil::getPlaneEquationsFromVertices(b3AlignedObjectArray<b3Vector3>& vertices, b3AlignedObjectArray<b3Vector3>& planeEquationsOut )
+void b3GeometryUtil::getPlaneEquationsFromVertices(b3AlignedObjectArray<b3Vector3>& vertices, b3AlignedObjectArray<b3Vector3>& planeEquationsOut)
 {
-		const int numvertices = vertices.size();
+	const int numvertices = vertices.size();
 	// brute force:
-	for (int i=0;i<numvertices;i++)
+	for (int i = 0; i < numvertices; i++)
 	{
 		const b3Vector3& N1 = vertices[i];
-		
 
-		for (int j=i+1;j<numvertices;j++)
+		for (int j = i + 1; j < numvertices; j++)
 		{
 			const b3Vector3& N2 = vertices[j];
-				
-			for (int k=j+1;k<numvertices;k++)
-			{
 
+			for (int k = j + 1; k < numvertices; k++)
+			{
 				const b3Vector3& N3 = vertices[k];
 
-				b3Vector3 planeEquation,edge0,edge1;
-				edge0 = N2-N1;
-				edge1 = N3-N1;
+				b3Vector3 planeEquation, edge0, edge1;
+				edge0 = N2 - N1;
+				edge1 = N3 - N1;
 				b3Scalar normalSign = b3Scalar(1.);
-				for (int ww=0;ww<2;ww++)
+				for (int ww = 0; ww < 2; ww++)
 				{
 					planeEquation = normalSign * edge0.cross(edge1);
 					if (planeEquation.length2() > b3Scalar(0.0001))
 					{
 						planeEquation.normalize();
-						if (notExist(planeEquation,planeEquationsOut))
+						if (notExist(planeEquation, planeEquationsOut))
 						{
 							planeEquation[3] = -planeEquation.dot(N1);
-							
-								//check if inside, and replace supportingVertexOut if needed
-								if (areVerticesBehindPlane(planeEquation,vertices,b3Scalar(0.01)))
-								{
-									planeEquationsOut.push_back(planeEquation);
-								}
+
+							//check if inside, and replace supportingVertexOut if needed
+							if (areVerticesBehindPlane(planeEquation, vertices, b3Scalar(0.01)))
+							{
+								planeEquationsOut.push_back(planeEquation);
+							}
 						}
 					}
 					normalSign = b3Scalar(-1.);
 				}
-			
 			}
 		}
 	}
-
 }
 
-void	b3GeometryUtil::getVerticesFromPlaneEquations(const b3AlignedObjectArray<b3Vector3>& planeEquations , b3AlignedObjectArray<b3Vector3>& verticesOut )
+void b3GeometryUtil::getVerticesFromPlaneEquations(const b3AlignedObjectArray<b3Vector3>& planeEquations, b3AlignedObjectArray<b3Vector3>& verticesOut)
 {
 	const int numbrushes = planeEquations.size();
 	// brute force:
-	for (int i=0;i<numbrushes;i++)
+	for (int i = 0; i < numbrushes; i++)
 	{
 		const b3Vector3& N1 = planeEquations[i];
-		
 
-		for (int j=i+1;j<numbrushes;j++)
+		for (int j = i + 1; j < numbrushes; j++)
 		{
 			const b3Vector3& N2 = planeEquations[j];
-				
-			for (int k=j+1;k<numbrushes;k++)
-			{
 
+			for (int k = j + 1; k < numbrushes; k++)
+			{
 				const b3Vector3& N3 = planeEquations[k];
 
-				b3Vector3 n2n3; n2n3 = N2.cross(N3);
-				b3Vector3 n3n1; n3n1 = N3.cross(N1);
-				b3Vector3 n1n2; n1n2 = N1.cross(N2);
-				
-				if ( ( n2n3.length2() > b3Scalar(0.0001) ) &&
-					 ( n3n1.length2() > b3Scalar(0.0001) ) &&
-					 ( n1n2.length2() > b3Scalar(0.0001) ) )
+				b3Vector3 n2n3;
+				n2n3 = N2.cross(N3);
+				b3Vector3 n3n1;
+				n3n1 = N3.cross(N1);
+				b3Vector3 n1n2;
+				n1n2 = N1.cross(N2);
+
+				if ((n2n3.length2() > b3Scalar(0.0001)) &&
+					(n3n1.length2() > b3Scalar(0.0001)) &&
+					(n1n2.length2() > b3Scalar(0.0001)))
 				{
 					//point P out of 3 plane equations:
 
-					//	d1 ( N2 * N3 ) + d2 ( N3 * N1 ) + d3 ( N1 * N2 )  
-					//P =  -------------------------------------------------------------------------  
-					//   N1 . ( N2 * N3 )  
-
+					//	d1 ( N2 * N3 ) + d2 ( N3 * N1 ) + d3 ( N1 * N2 )
+					//P =  -------------------------------------------------------------------------
+					//   N1 . ( N2 * N3 )
 
 					b3Scalar quotient = (N1.dot(n2n3));
 					if (b3Fabs(quotient) > b3Scalar(0.000001))
@@ -172,7 +162,7 @@ void	b3GeometryUtil::getVerticesFromPlaneEquations(const b3AlignedObjectArray<b3
 						potentialVertex *= quotient;
 
 						//check if inside, and replace supportingVertexOut if needed
-						if (isPointInsidePlanes(planeEquations,potentialVertex,b3Scalar(0.01)))
+						if (isPointInsidePlanes(planeEquations, potentialVertex, b3Scalar(0.01)))
 						{
 							verticesOut.push_back(potentialVertex);
 						}
@@ -182,4 +172,3 @@ void	b3GeometryUtil::getVerticesFromPlaneEquations(const b3AlignedObjectArray<b3
 		}
 	}
 }
-
diff --git a/thirdparty/bullet/Bullet3Geometry/b3GeometryUtil.h b/thirdparty/bullet/Bullet3Geometry/b3GeometryUtil.h
index 8b5fd7ad62e..967c8d67e9a 100644
--- a/thirdparty/bullet/Bullet3Geometry/b3GeometryUtil.h
+++ b/thirdparty/bullet/Bullet3Geometry/b3GeometryUtil.h
@@ -12,7 +12,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef B3_GEOMETRY_UTIL_H
 #define B3_GEOMETRY_UTIL_H
 
@@ -22,21 +21,16 @@ subject to the following restrictions:
 ///The b3GeometryUtil helper class provides a few methods to convert between plane equations and vertices.
 class b3GeometryUtil
 {
-	public:
-	
-	
-		static void	getPlaneEquationsFromVertices(b3AlignedObjectArray<b3Vector3>& vertices, b3AlignedObjectArray<b3Vector3>& planeEquationsOut );
+public:
+	static void getPlaneEquationsFromVertices(b3AlignedObjectArray<b3Vector3>& vertices, b3AlignedObjectArray<b3Vector3>& planeEquationsOut);
 
-		static void	getVerticesFromPlaneEquations(const b3AlignedObjectArray<b3Vector3>& planeEquations , b3AlignedObjectArray<b3Vector3>& verticesOut );
-	
-		static bool	isInside(const b3AlignedObjectArray<b3Vector3>& vertices, const b3Vector3& planeNormal, b3Scalar	margin);
-		
-		static bool	isPointInsidePlanes(const b3AlignedObjectArray<b3Vector3>& planeEquations, const b3Vector3& point, b3Scalar	margin);
+	static void getVerticesFromPlaneEquations(const b3AlignedObjectArray<b3Vector3>& planeEquations, b3AlignedObjectArray<b3Vector3>& verticesOut);
 
-		static bool	areVerticesBehindPlane(const b3Vector3& planeNormal, const b3AlignedObjectArray<b3Vector3>& vertices, b3Scalar	margin);
+	static bool isInside(const b3AlignedObjectArray<b3Vector3>& vertices, const b3Vector3& planeNormal, b3Scalar margin);
 
+	static bool isPointInsidePlanes(const b3AlignedObjectArray<b3Vector3>& planeEquations, const b3Vector3& point, b3Scalar margin);
+
+	static bool areVerticesBehindPlane(const b3Vector3& planeNormal, const b3AlignedObjectArray<b3Vector3>& vertices, b3Scalar margin);
 };
 
-
-#endif //B3_GEOMETRY_UTIL_H
-
+#endif  //B3_GEOMETRY_UTIL_H
diff --git a/thirdparty/bullet/Bullet3Geometry/b3GrahamScan2dConvexHull.h b/thirdparty/bullet/Bullet3Geometry/b3GrahamScan2dConvexHull.h
index 1b933c52645..8881c9a6384 100644
--- a/thirdparty/bullet/Bullet3Geometry/b3GrahamScan2dConvexHull.h
+++ b/thirdparty/bullet/Bullet3Geometry/b3GrahamScan2dConvexHull.h
@@ -13,41 +13,40 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef B3_GRAHAM_SCAN_2D_CONVEX_HULL_H
 #define B3_GRAHAM_SCAN_2D_CONVEX_HULL_H
 
-
 #include "Bullet3Common/b3Vector3.h"
 #include "Bullet3Common/b3AlignedObjectArray.h"
 
 struct b3GrahamVector3 : public b3Vector3
 {
 	b3GrahamVector3(const b3Vector3& org, int orgIndex)
-		:b3Vector3(org),
-			m_orgIndex(orgIndex)
+		: b3Vector3(org),
+		  m_orgIndex(orgIndex)
 	{
 	}
-	b3Scalar	m_angle;
+	b3Scalar m_angle;
 	int m_orgIndex;
 };
 
-
-struct b3AngleCompareFunc {
+struct b3AngleCompareFunc
+{
 	b3Vector3 m_anchor;
 	b3AngleCompareFunc(const b3Vector3& anchor)
-	: m_anchor(anchor) 
+		: m_anchor(anchor)
 	{
 	}
-	bool operator()(const b3GrahamVector3& a, const b3GrahamVector3& b) const {
+	bool operator()(const b3GrahamVector3& a, const b3GrahamVector3& b) const
+	{
 		if (a.m_angle != b.m_angle)
 			return a.m_angle < b.m_angle;
 		else
 		{
-			b3Scalar al = (a-m_anchor).length2();
-			b3Scalar bl = (b-m_anchor).length2();
+			b3Scalar al = (a - m_anchor).length2();
+			b3Scalar bl = (b - m_anchor).length2();
 			if (al != bl)
-				return  al < bl;
+				return al < bl;
 			else
 			{
 				return a.m_orgIndex < b.m_orgIndex;
@@ -58,60 +57,60 @@ struct b3AngleCompareFunc {
 
 inline void b3GrahamScanConvexHull2D(b3AlignedObjectArray<b3GrahamVector3>& originalPoints, b3AlignedObjectArray<b3GrahamVector3>& hull, const b3Vector3& normalAxis)
 {
-	b3Vector3 axis0,axis1;
-	b3PlaneSpace1(normalAxis,axis0,axis1);
-	
+	b3Vector3 axis0, axis1;
+	b3PlaneSpace1(normalAxis, axis0, axis1);
 
-	if (originalPoints.size()<=1)
+	if (originalPoints.size() <= 1)
 	{
-		for (int i=0;i<originalPoints.size();i++)
+		for (int i = 0; i < originalPoints.size(); i++)
 			hull.push_back(originalPoints[0]);
 		return;
 	}
 	//step1 : find anchor point with smallest projection on axis0 and move it to first location
-	for (int i=0;i<originalPoints.size();i++)
+	for (int i = 0; i < originalPoints.size(); i++)
 	{
-//		const b3Vector3& left = originalPoints[i];
-//		const b3Vector3& right = originalPoints[0];
+		//		const b3Vector3& left = originalPoints[i];
+		//		const b3Vector3& right = originalPoints[0];
 		b3Scalar projL = originalPoints[i].dot(axis0);
 		b3Scalar projR = originalPoints[0].dot(axis0);
 		if (projL < projR)
 		{
-			originalPoints.swap(0,i);
+			originalPoints.swap(0, i);
 		}
 	}
 
 	//also precompute angles
 	originalPoints[0].m_angle = -1e30f;
-	for (int i=1;i<originalPoints.size();i++)
+	for (int i = 1; i < originalPoints.size(); i++)
 	{
 		b3Vector3 xvec = axis0;
-		b3Vector3 ar = originalPoints[i]-originalPoints[0];
+		b3Vector3 ar = originalPoints[i] - originalPoints[0];
 		originalPoints[i].m_angle = b3Cross(xvec, ar).dot(normalAxis) / ar.length();
 	}
 
 	//step 2: sort all points, based on 'angle' with this anchor
 	b3AngleCompareFunc comp(originalPoints[0]);
-	originalPoints.quickSortInternal(comp,1,originalPoints.size()-1);
+	originalPoints.quickSortInternal(comp, 1, originalPoints.size() - 1);
 
 	int i;
-	for (i = 0; i<2; i++) 
+	for (i = 0; i < 2; i++)
 		hull.push_back(originalPoints[i]);
 
 	//step 3: keep all 'convex' points and discard concave points (using back tracking)
-	for (; i != originalPoints.size(); i++) 
+	for (; i != originalPoints.size(); i++)
 	{
 		bool isConvex = false;
-		while (!isConvex&& hull.size()>1) {
-			b3Vector3& a = hull[hull.size()-2];
-			b3Vector3& b = hull[hull.size()-1];
-			isConvex = b3Cross(a-b,a-originalPoints[i]).dot(normalAxis)> 0;
+		while (!isConvex && hull.size() > 1)
+		{
+			b3Vector3& a = hull[hull.size() - 2];
+			b3Vector3& b = hull[hull.size() - 1];
+			isConvex = b3Cross(a - b, a - originalPoints[i]).dot(normalAxis) > 0;
 			if (!isConvex)
 				hull.pop_back();
-			else 
+			else
 				hull.push_back(originalPoints[i]);
 		}
 	}
 }
 
-#endif //B3_GRAHAM_SCAN_2D_CONVEX_HULL_H
+#endif  //B3_GRAHAM_SCAN_2D_CONVEX_HULL_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuBroadphaseInterface.h b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuBroadphaseInterface.h
index 0ed8aa82326..b296992525f 100644
--- a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuBroadphaseInterface.h
+++ b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuBroadphaseInterface.h
@@ -12,33 +12,31 @@
 class b3GpuBroadphaseInterface
 {
 public:
-
-	typedef class b3GpuBroadphaseInterface* (CreateFunc)(cl_context ctx,cl_device_id device, cl_command_queue  q);
+	typedef class b3GpuBroadphaseInterface*(CreateFunc)(cl_context ctx, cl_device_id device, cl_command_queue q);
 
 	virtual ~b3GpuBroadphaseInterface()
 	{
 	}
 
-	virtual void createProxy(const b3Vector3& aabbMin,  const b3Vector3& aabbMax, int userPtr , int collisionFilterGroup, int collisionFilterMask)=0;
-	virtual void createLargeProxy(const b3Vector3& aabbMin,  const b3Vector3& aabbMax, int userPtr , int collisionFilterGroup, int collisionFilterMask)=0;
+	virtual void createProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, int userPtr, int collisionFilterGroup, int collisionFilterMask) = 0;
+	virtual void createLargeProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, int userPtr, int collisionFilterGroup, int collisionFilterMask) = 0;
 
-	virtual void  calculateOverlappingPairs(int maxPairs)=0;
-	virtual void  calculateOverlappingPairsHost(int maxPairs)=0;
+	virtual void calculateOverlappingPairs(int maxPairs) = 0;
+	virtual void calculateOverlappingPairsHost(int maxPairs) = 0;
 
 	//call writeAabbsToGpu after done making all changes (createProxy etc)
-	virtual void writeAabbsToGpu()=0;
+	virtual void writeAabbsToGpu() = 0;
 
-	virtual cl_mem	getAabbBufferWS()=0;
-	virtual int	getNumOverlap()=0;
-	virtual cl_mem	getOverlappingPairBuffer()=0;
+	virtual cl_mem getAabbBufferWS() = 0;
+	virtual int getNumOverlap() = 0;
+	virtual cl_mem getOverlappingPairBuffer() = 0;
+
+	virtual b3OpenCLArray<b3SapAabb>& getAllAabbsGPU() = 0;
+	virtual b3AlignedObjectArray<b3SapAabb>& getAllAabbsCPU() = 0;
 
-	virtual b3OpenCLArray<b3SapAabb>&	getAllAabbsGPU()=0;
-	virtual b3AlignedObjectArray<b3SapAabb>&	getAllAabbsCPU()=0;
-	
 	virtual b3OpenCLArray<b3Int4>& getOverlappingPairsGPU() = 0;
 	virtual b3OpenCLArray<int>& getSmallAabbIndicesGPU() = 0;
 	virtual b3OpenCLArray<int>& getLargeAabbIndicesGPU() = 0;
-
 };
 
-#endif //B3_GPU_BROADPHASE_INTERFACE_H
+#endif  //B3_GPU_BROADPHASE_INTERFACE_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuGridBroadphase.cpp b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuGridBroadphase.cpp
index 74d0c8056cc..e714fadac30 100644
--- a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuGridBroadphase.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuGridBroadphase.cpp
@@ -5,12 +5,9 @@
 #include "kernels/sapKernels.h"
 //#include "kernels/gridBroadphase.cl"
 
-
 #include "Bullet3OpenCL/Initialize/b3OpenCLUtils.h"
 #include "Bullet3OpenCL/ParallelPrimitives/b3LauncherCL.h"
 
-
-
 #define B3_BROADPHASE_SAP_PATH "src/Bullet3OpenCL/BroadphaseCollision/kernels/sap.cl"
 #define B3_GRID_BROADPHASE_PATH "src/Bullet3OpenCL/BroadphaseCollision/kernels/gridBroadphase.cl"
 
@@ -21,31 +18,25 @@ cl_kernel kFindOverlappingPairs;
 cl_kernel m_copyAabbsKernel;
 cl_kernel m_sap2Kernel;
 
-
-
-
-
 //int maxPairsPerBody = 64;
-int maxBodiesPerCell = 256;//??
+int maxBodiesPerCell = 256;  //??
 
-b3GpuGridBroadphase::b3GpuGridBroadphase(cl_context ctx,cl_device_id device, cl_command_queue  q )
-:m_context(ctx),
-m_device(device),
-m_queue(q),
-m_allAabbsGPU1(ctx,q),
-m_smallAabbsMappingGPU(ctx,q),
-m_largeAabbsMappingGPU(ctx,q),
-m_gpuPairs(ctx,q),
+b3GpuGridBroadphase::b3GpuGridBroadphase(cl_context ctx, cl_device_id device, cl_command_queue q)
+	: m_context(ctx),
+	  m_device(device),
+	  m_queue(q),
+	  m_allAabbsGPU1(ctx, q),
+	  m_smallAabbsMappingGPU(ctx, q),
+	  m_largeAabbsMappingGPU(ctx, q),
+	  m_gpuPairs(ctx, q),
 
-m_hashGpu(ctx,q),
+	  m_hashGpu(ctx, q),
 
-m_cellStartGpu(ctx,q),
-m_paramsGPU(ctx,q)
+	  m_cellStartGpu(ctx, q),
+	  m_paramsGPU(ctx, q)
 {
-
-	
-	b3Vector3 gridSize = b3MakeVector3(3,3,3);
-	b3Vector3 invGridSize = b3MakeVector3(1.f/gridSize[0],1.f/gridSize[1],1.f/gridSize[2]);
+	b3Vector3 gridSize = b3MakeVector3(3, 3, 3);
+	b3Vector3 invGridSize = b3MakeVector3(1.f / gridSize[0], 1.f / gridSize[1], 1.f / gridSize[2]);
 
 	m_paramsCPU.m_gridSize[0] = 128;
 	m_paramsCPU.m_gridSize[1] = 128;
@@ -58,92 +49,79 @@ m_paramsGPU(ctx,q)
 	m_paramsCPU.m_invCellSize[3] = 0.f;
 	m_paramsGPU.push_back(m_paramsCPU);
 
-	cl_int errNum=0;
+	cl_int errNum = 0;
 
 	{
 		const char* sapSrc = sapCL;
-		cl_program sapProg = b3OpenCLUtils::compileCLProgramFromString(m_context,m_device,sapSrc,&errNum,"",B3_BROADPHASE_SAP_PATH);
-		b3Assert(errNum==CL_SUCCESS);
-		m_copyAabbsKernel= b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,sapSrc, "copyAabbsKernel",&errNum,sapProg );
-		m_sap2Kernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,sapSrc, "computePairsKernelTwoArrays",&errNum,sapProg );
-		b3Assert(errNum==CL_SUCCESS);
+		cl_program sapProg = b3OpenCLUtils::compileCLProgramFromString(m_context, m_device, sapSrc, &errNum, "", B3_BROADPHASE_SAP_PATH);
+		b3Assert(errNum == CL_SUCCESS);
+		m_copyAabbsKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, sapSrc, "copyAabbsKernel", &errNum, sapProg);
+		m_sap2Kernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, sapSrc, "computePairsKernelTwoArrays", &errNum, sapProg);
+		b3Assert(errNum == CL_SUCCESS);
 	}
 
 	{
-		
-		cl_program gridProg = b3OpenCLUtils::compileCLProgramFromString(m_context,m_device,gridBroadphaseCL,&errNum,"",B3_GRID_BROADPHASE_PATH);
-		b3Assert(errNum==CL_SUCCESS);
+		cl_program gridProg = b3OpenCLUtils::compileCLProgramFromString(m_context, m_device, gridBroadphaseCL, &errNum, "", B3_GRID_BROADPHASE_PATH);
+		b3Assert(errNum == CL_SUCCESS);
 
-		kCalcHashAABB = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,gridBroadphaseCL, "kCalcHashAABB",&errNum,gridProg);
-		b3Assert(errNum==CL_SUCCESS);
-	
-		kClearCellStart = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,gridBroadphaseCL, "kClearCellStart",&errNum,gridProg);
-		b3Assert(errNum==CL_SUCCESS);
+		kCalcHashAABB = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, gridBroadphaseCL, "kCalcHashAABB", &errNum, gridProg);
+		b3Assert(errNum == CL_SUCCESS);
 
-		kFindCellStart = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,gridBroadphaseCL, "kFindCellStart",&errNum,gridProg);
-		b3Assert(errNum==CL_SUCCESS);
+		kClearCellStart = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, gridBroadphaseCL, "kClearCellStart", &errNum, gridProg);
+		b3Assert(errNum == CL_SUCCESS);
 
-	
-		kFindOverlappingPairs = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,gridBroadphaseCL, "kFindOverlappingPairs",&errNum,gridProg);
-		b3Assert(errNum==CL_SUCCESS);
+		kFindCellStart = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, gridBroadphaseCL, "kFindCellStart", &errNum, gridProg);
+		b3Assert(errNum == CL_SUCCESS);
 
-		
-		
-		
+		kFindOverlappingPairs = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, gridBroadphaseCL, "kFindOverlappingPairs", &errNum, gridProg);
+		b3Assert(errNum == CL_SUCCESS);
 	}
 
-	m_sorter = new b3RadixSort32CL(m_context,m_device,m_queue);
-
+	m_sorter = new b3RadixSort32CL(m_context, m_device, m_queue);
 }
 b3GpuGridBroadphase::~b3GpuGridBroadphase()
 {
-	clReleaseKernel( kCalcHashAABB);
-	clReleaseKernel( kClearCellStart);
-	clReleaseKernel( kFindCellStart);
-	clReleaseKernel( kFindOverlappingPairs);
-	clReleaseKernel( m_sap2Kernel);
-	clReleaseKernel( m_copyAabbsKernel);
-	
-	
-	
+	clReleaseKernel(kCalcHashAABB);
+	clReleaseKernel(kClearCellStart);
+	clReleaseKernel(kFindCellStart);
+	clReleaseKernel(kFindOverlappingPairs);
+	clReleaseKernel(m_sap2Kernel);
+	clReleaseKernel(m_copyAabbsKernel);
+
 	delete m_sorter;
 }
 
-
-
-void b3GpuGridBroadphase::createProxy(const b3Vector3& aabbMin,  const b3Vector3& aabbMax, int userPtr , int collisionFilterGroup, int collisionFilterMask)
+void b3GpuGridBroadphase::createProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, int userPtr, int collisionFilterGroup, int collisionFilterMask)
 {
 	b3SapAabb aabb;
 	aabb.m_minVec = aabbMin;
 	aabb.m_maxVec = aabbMax;
 	aabb.m_minIndices[3] = userPtr;
-	aabb.m_signedMaxIndices[3] = m_allAabbsCPU1.size();//NOT userPtr;
+	aabb.m_signedMaxIndices[3] = m_allAabbsCPU1.size();  //NOT userPtr;
 	m_smallAabbsMappingCPU.push_back(m_allAabbsCPU1.size());
 
 	m_allAabbsCPU1.push_back(aabb);
-
 }
-void b3GpuGridBroadphase::createLargeProxy(const b3Vector3& aabbMin,  const b3Vector3& aabbMax, int userPtr , int collisionFilterGroup, int collisionFilterMask)
+void b3GpuGridBroadphase::createLargeProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, int userPtr, int collisionFilterGroup, int collisionFilterMask)
 {
 	b3SapAabb aabb;
 	aabb.m_minVec = aabbMin;
 	aabb.m_maxVec = aabbMax;
 	aabb.m_minIndices[3] = userPtr;
-	aabb.m_signedMaxIndices[3] = m_allAabbsCPU1.size();//NOT userPtr;
+	aabb.m_signedMaxIndices[3] = m_allAabbsCPU1.size();  //NOT userPtr;
 	m_largeAabbsMappingCPU.push_back(m_allAabbsCPU1.size());
 
 	m_allAabbsCPU1.push_back(aabb);
 }
 
-void  b3GpuGridBroadphase::calculateOverlappingPairs(int maxPairs)
+void b3GpuGridBroadphase::calculateOverlappingPairs(int maxPairs)
 {
 	B3_PROFILE("b3GpuGridBroadphase::calculateOverlappingPairs");
-	
 
 	if (0)
 	{
 		calculateOverlappingPairsHost(maxPairs);
-	/*
+		/*
 		b3AlignedObjectArray<b3Int4> cpuPairs;
 		m_gpuPairs.copyToHost(cpuPairs);
 		printf("host m_gpuPairs.size()=%d\n",m_gpuPairs.size());
@@ -154,57 +132,50 @@ void  b3GpuGridBroadphase::calculateOverlappingPairs(int maxPairs)
 		*/
 		return;
 	}
-	
-	
 
-
-	
 	int numSmallAabbs = m_smallAabbsMappingGPU.size();
 
-	b3OpenCLArray<int> pairCount(m_context,m_queue);
+	b3OpenCLArray<int> pairCount(m_context, m_queue);
 	pairCount.push_back(0);
-	m_gpuPairs.resize(maxPairs);//numSmallAabbs*maxPairsPerBody);
+	m_gpuPairs.resize(maxPairs);  //numSmallAabbs*maxPairsPerBody);
 
 	{
 		int numLargeAabbs = m_largeAabbsMappingGPU.size();
 		if (numLargeAabbs && numSmallAabbs)
 		{
 			B3_PROFILE("sap2Kernel");
-			b3BufferInfoCL bInfo[] = { 
-				b3BufferInfoCL( m_allAabbsGPU1.getBufferCL() ),
-				b3BufferInfoCL( m_largeAabbsMappingGPU.getBufferCL() ),
-				b3BufferInfoCL( m_smallAabbsMappingGPU.getBufferCL() ), 
-				b3BufferInfoCL( m_gpuPairs.getBufferCL() ), 
+			b3BufferInfoCL bInfo[] = {
+				b3BufferInfoCL(m_allAabbsGPU1.getBufferCL()),
+				b3BufferInfoCL(m_largeAabbsMappingGPU.getBufferCL()),
+				b3BufferInfoCL(m_smallAabbsMappingGPU.getBufferCL()),
+				b3BufferInfoCL(m_gpuPairs.getBufferCL()),
 				b3BufferInfoCL(pairCount.getBufferCL())};
-			b3LauncherCL launcher(m_queue, m_sap2Kernel,"m_sap2Kernel");
-			launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-			launcher.setConst(   numLargeAabbs  );
-			launcher.setConst( numSmallAabbs);
-			launcher.setConst( 0  );//axis is not used
-			launcher.setConst( maxPairs  );
-	//@todo: use actual maximum work item sizes of the device instead of hardcoded values
-			launcher.launch2D( numLargeAabbs, numSmallAabbs,4,64);
-                
+			b3LauncherCL launcher(m_queue, m_sap2Kernel, "m_sap2Kernel");
+			launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+			launcher.setConst(numLargeAabbs);
+			launcher.setConst(numSmallAabbs);
+			launcher.setConst(0);  //axis is not used
+			launcher.setConst(maxPairs);
+			//@todo: use actual maximum work item sizes of the device instead of hardcoded values
+			launcher.launch2D(numLargeAabbs, numSmallAabbs, 4, 64);
+
 			int numPairs = pairCount.at(0);
-			
-			if (numPairs >maxPairs)
+
+			if (numPairs > maxPairs)
 			{
 				b3Error("Error running out of pairs: numPairs = %d, maxPairs = %d.\n", numPairs, maxPairs);
-				numPairs =maxPairs;
+				numPairs = maxPairs;
 			}
 		}
 	}
 
-
-
-
 	if (numSmallAabbs)
 	{
 		B3_PROFILE("gridKernel");
 		m_hashGpu.resize(numSmallAabbs);
 		{
 			B3_PROFILE("kCalcHashAABB");
-			b3LauncherCL launch(m_queue,kCalcHashAABB,"kCalcHashAABB");
+			b3LauncherCL launch(m_queue, kCalcHashAABB, "kCalcHashAABB");
 			launch.setConst(numSmallAabbs);
 			launch.setBuffer(m_allAabbsGPU1.getBufferCL());
 			launch.setBuffer(m_smallAabbsMappingGPU.getBufferCL());
@@ -214,117 +185,104 @@ void  b3GpuGridBroadphase::calculateOverlappingPairs(int maxPairs)
 		}
 
 		m_sorter->execute(m_hashGpu);
-		
-		int numCells = this->m_paramsCPU.m_gridSize[0]*this->m_paramsCPU.m_gridSize[1]*this->m_paramsCPU.m_gridSize[2];
+
+		int numCells = this->m_paramsCPU.m_gridSize[0] * this->m_paramsCPU.m_gridSize[1] * this->m_paramsCPU.m_gridSize[2];
 		m_cellStartGpu.resize(numCells);
 		//b3AlignedObjectArray<int >			cellStartCpu;
-				
-		
+
 		{
 			B3_PROFILE("kClearCellStart");
-			b3LauncherCL launch(m_queue,kClearCellStart,"kClearCellStart");
+			b3LauncherCL launch(m_queue, kClearCellStart, "kClearCellStart");
 			launch.setConst(numCells);
 			launch.setBuffer(m_cellStartGpu.getBufferCL());
 			launch.launch1D(numCells);
 			//m_cellStartGpu.copyToHost(cellStartCpu);
 			//printf("??\n");
-
 		}
 
-
 		{
 			B3_PROFILE("kFindCellStart");
-			b3LauncherCL launch(m_queue,kFindCellStart,"kFindCellStart");
+			b3LauncherCL launch(m_queue, kFindCellStart, "kFindCellStart");
 			launch.setConst(numSmallAabbs);
 			launch.setBuffer(m_hashGpu.getBufferCL());
 			launch.setBuffer(m_cellStartGpu.getBufferCL());
 			launch.launch1D(numSmallAabbs);
 			//m_cellStartGpu.copyToHost(cellStartCpu);
 			//printf("??\n");
-
 		}
-		
+
 		{
 			B3_PROFILE("kFindOverlappingPairs");
-			
-			
-			b3LauncherCL launch(m_queue,kFindOverlappingPairs,"kFindOverlappingPairs");
+
+			b3LauncherCL launch(m_queue, kFindOverlappingPairs, "kFindOverlappingPairs");
 			launch.setConst(numSmallAabbs);
 			launch.setBuffer(m_allAabbsGPU1.getBufferCL());
 			launch.setBuffer(m_smallAabbsMappingGPU.getBufferCL());
 			launch.setBuffer(m_hashGpu.getBufferCL());
 			launch.setBuffer(m_cellStartGpu.getBufferCL());
-			
+
 			launch.setBuffer(m_paramsGPU.getBufferCL());
 			//launch.setBuffer(0);
 			launch.setBuffer(pairCount.getBufferCL());
 			launch.setBuffer(m_gpuPairs.getBufferCL());
-			
+
 			launch.setConst(maxPairs);
 			launch.launch1D(numSmallAabbs);
-			
 
 			int numPairs = pairCount.at(0);
-			if (numPairs >maxPairs)
+			if (numPairs > maxPairs)
 			{
 				b3Error("Error running out of pairs: numPairs = %d, maxPairs = %d.\n", numPairs, maxPairs);
-				numPairs =maxPairs;
+				numPairs = maxPairs;
 			}
-			
+
 			m_gpuPairs.resize(numPairs);
-	
+
 			if (0)
 			{
 				b3AlignedObjectArray<b3Int4> pairsCpu;
 				m_gpuPairs.copyToHost(pairsCpu);
 
 				int sz = m_gpuPairs.size();
-				printf("m_gpuPairs.size()=%d\n",sz);
-				for (int i=0;i<m_gpuPairs.size();i++)
+				printf("m_gpuPairs.size()=%d\n", sz);
+				for (int i = 0; i < m_gpuPairs.size(); i++)
 				{
-					printf("pair %d = %d,%d\n",i,pairsCpu[i].x,pairsCpu[i].y);
+					printf("pair %d = %d,%d\n", i, pairsCpu[i].x, pairsCpu[i].y);
 				}
 
 				printf("?!?\n");
 			}
-			
 		}
-	
-
 	}
 
-	
-
-
-
 	//calculateOverlappingPairsHost(maxPairs);
 }
-void  b3GpuGridBroadphase::calculateOverlappingPairsHost(int maxPairs)
+void b3GpuGridBroadphase::calculateOverlappingPairsHost(int maxPairs)
 {
-
 	m_hostPairs.resize(0);
 	m_allAabbsGPU1.copyToHost(m_allAabbsCPU1);
-	for (int i=0;i<m_allAabbsCPU1.size();i++)
+	for (int i = 0; i < m_allAabbsCPU1.size(); i++)
 	{
-		for (int j=i+1;j<m_allAabbsCPU1.size();j++)
+		for (int j = i + 1; j < m_allAabbsCPU1.size(); j++)
 		{
 			if (b3TestAabbAgainstAabb2(m_allAabbsCPU1[i].m_minVec, m_allAabbsCPU1[i].m_maxVec,
-				m_allAabbsCPU1[j].m_minVec,m_allAabbsCPU1[j].m_maxVec))
+									   m_allAabbsCPU1[j].m_minVec, m_allAabbsCPU1[j].m_maxVec))
 			{
 				b3Int4 pair;
 				int a = m_allAabbsCPU1[j].m_minIndices[3];
 				int b = m_allAabbsCPU1[i].m_minIndices[3];
-				if (a<=b)
+				if (a <= b)
 				{
-					pair.x = a; 
-					pair.y = b;//store the original index in the unsorted aabb array
-				} else
+					pair.x = a;
+					pair.y = b;  //store the original index in the unsorted aabb array
+				}
+				else
 				{
 					pair.x = b;
-					pair.y = a;//store the original index in the unsorted aabb array
+					pair.y = a;  //store the original index in the unsorted aabb array
 				}
-					
-				if (m_hostPairs.size()<maxPairs)
+
+				if (m_hostPairs.size() < maxPairs)
 				{
 					m_hostPairs.push_back(pair);
 				}
@@ -332,40 +290,36 @@ void  b3GpuGridBroadphase::calculateOverlappingPairsHost(int maxPairs)
 		}
 	}
 
-
 	m_gpuPairs.copyFromHost(m_hostPairs);
-
-
 }
 
-	//call writeAabbsToGpu after done making all changes (createProxy etc)
+//call writeAabbsToGpu after done making all changes (createProxy etc)
 void b3GpuGridBroadphase::writeAabbsToGpu()
 {
 	m_allAabbsGPU1.copyFromHost(m_allAabbsCPU1);
 	m_smallAabbsMappingGPU.copyFromHost(m_smallAabbsMappingCPU);
 	m_largeAabbsMappingGPU.copyFromHost(m_largeAabbsMappingCPU);
-
 }
 
-cl_mem	b3GpuGridBroadphase::getAabbBufferWS()
+cl_mem b3GpuGridBroadphase::getAabbBufferWS()
 {
 	return this->m_allAabbsGPU1.getBufferCL();
 }
-int	b3GpuGridBroadphase::getNumOverlap()
+int b3GpuGridBroadphase::getNumOverlap()
 {
 	return m_gpuPairs.size();
 }
-cl_mem	b3GpuGridBroadphase::getOverlappingPairBuffer()
+cl_mem b3GpuGridBroadphase::getOverlappingPairBuffer()
 {
 	return m_gpuPairs.getBufferCL();
 }
 
-b3OpenCLArray<b3SapAabb>&	b3GpuGridBroadphase::getAllAabbsGPU()
+b3OpenCLArray<b3SapAabb>& b3GpuGridBroadphase::getAllAabbsGPU()
 {
 	return m_allAabbsGPU1;
 }
 
-b3AlignedObjectArray<b3SapAabb>&	b3GpuGridBroadphase::getAllAabbsCPU()
+b3AlignedObjectArray<b3SapAabb>& b3GpuGridBroadphase::getAllAabbsCPU()
 {
 	return m_allAabbsCPU1;
 }
@@ -382,4 +336,3 @@ b3OpenCLArray<int>& b3GpuGridBroadphase::getLargeAabbIndicesGPU()
 {
 	return m_largeAabbsMappingGPU;
 }
-
diff --git a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuGridBroadphase.h b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuGridBroadphase.h
index ec18c9f7165..b76cb43b68d 100644
--- a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuGridBroadphase.h
+++ b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuGridBroadphase.h
@@ -6,83 +6,75 @@
 
 struct b3ParamsGridBroadphaseCL
 {
-
 	float m_invCellSize[4];
-	int   m_gridSize[4];
+	int m_gridSize[4];
 
-	int	getMaxBodiesPerCell() const
+	int getMaxBodiesPerCell() const
 	{
 		return m_gridSize[3];
 	}
 
-	void setMaxBodiesPerCell(int maxOverlap) 
+	void setMaxBodiesPerCell(int maxOverlap)
 	{
 		m_gridSize[3] = maxOverlap;
 	}
 };
 
-
 class b3GpuGridBroadphase : public b3GpuBroadphaseInterface
 {
 protected:
-	cl_context				m_context;
-	cl_device_id			m_device;
-	cl_command_queue		m_queue;
+	cl_context m_context;
+	cl_device_id m_device;
+	cl_command_queue m_queue;
 
-	b3OpenCLArray<b3SapAabb>	m_allAabbsGPU1;
-	b3AlignedObjectArray<b3SapAabb>	m_allAabbsCPU1;
+	b3OpenCLArray<b3SapAabb> m_allAabbsGPU1;
+	b3AlignedObjectArray<b3SapAabb> m_allAabbsCPU1;
 
-	b3OpenCLArray<int>	m_smallAabbsMappingGPU;
+	b3OpenCLArray<int> m_smallAabbsMappingGPU;
 	b3AlignedObjectArray<int> m_smallAabbsMappingCPU;
 
-	b3OpenCLArray<int>	m_largeAabbsMappingGPU;
+	b3OpenCLArray<int> m_largeAabbsMappingGPU;
 	b3AlignedObjectArray<int> m_largeAabbsMappingCPU;
 
 	b3AlignedObjectArray<b3Int4> m_hostPairs;
-	b3OpenCLArray<b3Int4>			m_gpuPairs;
+	b3OpenCLArray<b3Int4> m_gpuPairs;
 
-	b3OpenCLArray<b3SortData>			m_hashGpu;
-	b3OpenCLArray<int>			m_cellStartGpu;
-	
+	b3OpenCLArray<b3SortData> m_hashGpu;
+	b3OpenCLArray<int> m_cellStartGpu;
 
-	b3ParamsGridBroadphaseCL		m_paramsCPU;
-	b3OpenCLArray<b3ParamsGridBroadphaseCL>		m_paramsGPU;
+	b3ParamsGridBroadphaseCL m_paramsCPU;
+	b3OpenCLArray<b3ParamsGridBroadphaseCL> m_paramsGPU;
 
-	class b3RadixSort32CL*			m_sorter;
+	class b3RadixSort32CL* m_sorter;
 
 public:
-
-	b3GpuGridBroadphase(cl_context ctx,cl_device_id device, cl_command_queue  q );
+	b3GpuGridBroadphase(cl_context ctx, cl_device_id device, cl_command_queue q);
 	virtual ~b3GpuGridBroadphase();
 
-	static b3GpuBroadphaseInterface* CreateFunc(cl_context ctx,cl_device_id device, cl_command_queue  q)
+	static b3GpuBroadphaseInterface* CreateFunc(cl_context ctx, cl_device_id device, cl_command_queue q)
 	{
-		return new b3GpuGridBroadphase(ctx,device,q);
+		return new b3GpuGridBroadphase(ctx, device, q);
 	}
 
-	
+	virtual void createProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, int userPtr, int collisionFilterGroup, int collisionFilterMask);
+	virtual void createLargeProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, int userPtr, int collisionFilterGroup, int collisionFilterMask);
 
-
-	virtual void createProxy(const b3Vector3& aabbMin,  const b3Vector3& aabbMax, int userPtr , int collisionFilterGroup, int collisionFilterMask);
-	virtual void createLargeProxy(const b3Vector3& aabbMin,  const b3Vector3& aabbMax, int userPtr , int collisionFilterGroup, int collisionFilterMask);
-
-	virtual void  calculateOverlappingPairs(int maxPairs);
-	virtual void  calculateOverlappingPairsHost(int maxPairs);
+	virtual void calculateOverlappingPairs(int maxPairs);
+	virtual void calculateOverlappingPairsHost(int maxPairs);
 
 	//call writeAabbsToGpu after done making all changes (createProxy etc)
 	virtual void writeAabbsToGpu();
 
-	virtual cl_mem	getAabbBufferWS();
-	virtual int	getNumOverlap();
-	virtual cl_mem	getOverlappingPairBuffer();
+	virtual cl_mem getAabbBufferWS();
+	virtual int getNumOverlap();
+	virtual cl_mem getOverlappingPairBuffer();
+
+	virtual b3OpenCLArray<b3SapAabb>& getAllAabbsGPU();
+	virtual b3AlignedObjectArray<b3SapAabb>& getAllAabbsCPU();
 
-	virtual b3OpenCLArray<b3SapAabb>&	getAllAabbsGPU();
-	virtual b3AlignedObjectArray<b3SapAabb>&	getAllAabbsCPU();
-	
 	virtual b3OpenCLArray<b3Int4>& getOverlappingPairsGPU();
 	virtual b3OpenCLArray<int>& getSmallAabbIndicesGPU();
 	virtual b3OpenCLArray<int>& getLargeAabbIndicesGPU();
-
 };
 
-#endif //B3_GPU_GRID_BROADPHASE_H
\ No newline at end of file
+#endif  //B3_GPU_GRID_BROADPHASE_H
\ No newline at end of file
diff --git a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuParallelLinearBvh.cpp b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuParallelLinearBvh.cpp
index 641df9eb128..0721928684f 100644
--- a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuParallelLinearBvh.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuParallelLinearBvh.cpp
@@ -16,177 +16,174 @@ subject to the following restrictions:
 
 #include "b3GpuParallelLinearBvh.h"
 
-b3GpuParallelLinearBvh::b3GpuParallelLinearBvh(cl_context context, cl_device_id device, cl_command_queue queue) :
-	m_queue(queue),
-	m_radixSorter(context, device, queue),
-	
-	m_rootNodeIndex(context, queue),
-	m_maxDistanceFromRoot(context, queue),
-	m_temp(context, queue),
-	
-	m_internalNodeAabbs(context, queue),
-	m_internalNodeLeafIndexRanges(context, queue),
-	m_internalNodeChildNodes(context, queue),
-	m_internalNodeParentNodes(context, queue),
-	
-	m_commonPrefixes(context, queue),
-	m_commonPrefixLengths(context, queue),
-	m_distanceFromRoot(context, queue),
-	
-	m_leafNodeParentNodes(context, queue),
-	m_mortonCodesAndAabbIndicies(context, queue),
-	m_mergedAabb(context, queue),
-	m_leafNodeAabbs(context, queue),
-	
-	m_largeAabbs(context, queue)
+b3GpuParallelLinearBvh::b3GpuParallelLinearBvh(cl_context context, cl_device_id device, cl_command_queue queue) : m_queue(queue),
+																												  m_radixSorter(context, device, queue),
+
+																												  m_rootNodeIndex(context, queue),
+																												  m_maxDistanceFromRoot(context, queue),
+																												  m_temp(context, queue),
+
+																												  m_internalNodeAabbs(context, queue),
+																												  m_internalNodeLeafIndexRanges(context, queue),
+																												  m_internalNodeChildNodes(context, queue),
+																												  m_internalNodeParentNodes(context, queue),
+
+																												  m_commonPrefixes(context, queue),
+																												  m_commonPrefixLengths(context, queue),
+																												  m_distanceFromRoot(context, queue),
+
+																												  m_leafNodeParentNodes(context, queue),
+																												  m_mortonCodesAndAabbIndicies(context, queue),
+																												  m_mergedAabb(context, queue),
+																												  m_leafNodeAabbs(context, queue),
+
+																												  m_largeAabbs(context, queue)
 {
 	m_rootNodeIndex.resize(1);
 	m_maxDistanceFromRoot.resize(1);
 	m_temp.resize(1);
-	
+
 	//
 	const char CL_PROGRAM_PATH[] = "src/Bullet3OpenCL/BroadphaseCollision/kernels/parallelLinearBvh.cl";
-	
-	const char* kernelSource = parallelLinearBvhCL;	//parallelLinearBvhCL.h
+
+	const char* kernelSource = parallelLinearBvhCL;  //parallelLinearBvhCL.h
 	cl_int error;
 	char* additionalMacros = 0;
 	m_parallelLinearBvhProgram = b3OpenCLUtils::compileCLProgramFromString(context, device, kernelSource, &error, additionalMacros, CL_PROGRAM_PATH);
 	b3Assert(m_parallelLinearBvhProgram);
-	
-	m_separateAabbsKernel = b3OpenCLUtils::compileCLKernelFromString( context, device, kernelSource, "separateAabbs", &error, m_parallelLinearBvhProgram, additionalMacros );
+
+	m_separateAabbsKernel = b3OpenCLUtils::compileCLKernelFromString(context, device, kernelSource, "separateAabbs", &error, m_parallelLinearBvhProgram, additionalMacros);
 	b3Assert(m_separateAabbsKernel);
-	m_findAllNodesMergedAabbKernel = b3OpenCLUtils::compileCLKernelFromString( context, device, kernelSource, "findAllNodesMergedAabb", &error, m_parallelLinearBvhProgram, additionalMacros );
+	m_findAllNodesMergedAabbKernel = b3OpenCLUtils::compileCLKernelFromString(context, device, kernelSource, "findAllNodesMergedAabb", &error, m_parallelLinearBvhProgram, additionalMacros);
 	b3Assert(m_findAllNodesMergedAabbKernel);
-	m_assignMortonCodesAndAabbIndiciesKernel = b3OpenCLUtils::compileCLKernelFromString( context, device, kernelSource, "assignMortonCodesAndAabbIndicies", &error, m_parallelLinearBvhProgram, additionalMacros );
+	m_assignMortonCodesAndAabbIndiciesKernel = b3OpenCLUtils::compileCLKernelFromString(context, device, kernelSource, "assignMortonCodesAndAabbIndicies", &error, m_parallelLinearBvhProgram, additionalMacros);
 	b3Assert(m_assignMortonCodesAndAabbIndiciesKernel);
-	
-	m_computeAdjacentPairCommonPrefixKernel = b3OpenCLUtils::compileCLKernelFromString( context, device, kernelSource, "computeAdjacentPairCommonPrefix", &error, m_parallelLinearBvhProgram, additionalMacros );
+
+	m_computeAdjacentPairCommonPrefixKernel = b3OpenCLUtils::compileCLKernelFromString(context, device, kernelSource, "computeAdjacentPairCommonPrefix", &error, m_parallelLinearBvhProgram, additionalMacros);
 	b3Assert(m_computeAdjacentPairCommonPrefixKernel);
-	m_buildBinaryRadixTreeLeafNodesKernel = b3OpenCLUtils::compileCLKernelFromString( context, device, kernelSource, "buildBinaryRadixTreeLeafNodes", &error, m_parallelLinearBvhProgram, additionalMacros );
+	m_buildBinaryRadixTreeLeafNodesKernel = b3OpenCLUtils::compileCLKernelFromString(context, device, kernelSource, "buildBinaryRadixTreeLeafNodes", &error, m_parallelLinearBvhProgram, additionalMacros);
 	b3Assert(m_buildBinaryRadixTreeLeafNodesKernel);
-	m_buildBinaryRadixTreeInternalNodesKernel = b3OpenCLUtils::compileCLKernelFromString( context, device, kernelSource, "buildBinaryRadixTreeInternalNodes", &error, m_parallelLinearBvhProgram, additionalMacros );
+	m_buildBinaryRadixTreeInternalNodesKernel = b3OpenCLUtils::compileCLKernelFromString(context, device, kernelSource, "buildBinaryRadixTreeInternalNodes", &error, m_parallelLinearBvhProgram, additionalMacros);
 	b3Assert(m_buildBinaryRadixTreeInternalNodesKernel);
-	m_findDistanceFromRootKernel = b3OpenCLUtils::compileCLKernelFromString( context, device, kernelSource, "findDistanceFromRoot", &error, m_parallelLinearBvhProgram, additionalMacros );
+	m_findDistanceFromRootKernel = b3OpenCLUtils::compileCLKernelFromString(context, device, kernelSource, "findDistanceFromRoot", &error, m_parallelLinearBvhProgram, additionalMacros);
 	b3Assert(m_findDistanceFromRootKernel);
-	m_buildBinaryRadixTreeAabbsRecursiveKernel = b3OpenCLUtils::compileCLKernelFromString( context, device, kernelSource, "buildBinaryRadixTreeAabbsRecursive", &error, m_parallelLinearBvhProgram, additionalMacros );
+	m_buildBinaryRadixTreeAabbsRecursiveKernel = b3OpenCLUtils::compileCLKernelFromString(context, device, kernelSource, "buildBinaryRadixTreeAabbsRecursive", &error, m_parallelLinearBvhProgram, additionalMacros);
 	b3Assert(m_buildBinaryRadixTreeAabbsRecursiveKernel);
-	
-	m_findLeafIndexRangesKernel = b3OpenCLUtils::compileCLKernelFromString( context, device, kernelSource, "findLeafIndexRanges", &error, m_parallelLinearBvhProgram, additionalMacros );
+
+	m_findLeafIndexRangesKernel = b3OpenCLUtils::compileCLKernelFromString(context, device, kernelSource, "findLeafIndexRanges", &error, m_parallelLinearBvhProgram, additionalMacros);
 	b3Assert(m_findLeafIndexRangesKernel);
-	
-	m_plbvhCalculateOverlappingPairsKernel = b3OpenCLUtils::compileCLKernelFromString( context, device, kernelSource, "plbvhCalculateOverlappingPairs", &error, m_parallelLinearBvhProgram, additionalMacros );
+
+	m_plbvhCalculateOverlappingPairsKernel = b3OpenCLUtils::compileCLKernelFromString(context, device, kernelSource, "plbvhCalculateOverlappingPairs", &error, m_parallelLinearBvhProgram, additionalMacros);
 	b3Assert(m_plbvhCalculateOverlappingPairsKernel);
-	m_plbvhRayTraverseKernel = b3OpenCLUtils::compileCLKernelFromString( context, device, kernelSource, "plbvhRayTraverse", &error, m_parallelLinearBvhProgram, additionalMacros );
+	m_plbvhRayTraverseKernel = b3OpenCLUtils::compileCLKernelFromString(context, device, kernelSource, "plbvhRayTraverse", &error, m_parallelLinearBvhProgram, additionalMacros);
 	b3Assert(m_plbvhRayTraverseKernel);
-	m_plbvhLargeAabbAabbTestKernel = b3OpenCLUtils::compileCLKernelFromString( context, device, kernelSource, "plbvhLargeAabbAabbTest", &error, m_parallelLinearBvhProgram, additionalMacros );
+	m_plbvhLargeAabbAabbTestKernel = b3OpenCLUtils::compileCLKernelFromString(context, device, kernelSource, "plbvhLargeAabbAabbTest", &error, m_parallelLinearBvhProgram, additionalMacros);
 	b3Assert(m_plbvhLargeAabbAabbTestKernel);
-	m_plbvhLargeAabbRayTestKernel = b3OpenCLUtils::compileCLKernelFromString( context, device, kernelSource, "plbvhLargeAabbRayTest", &error, m_parallelLinearBvhProgram, additionalMacros );
+	m_plbvhLargeAabbRayTestKernel = b3OpenCLUtils::compileCLKernelFromString(context, device, kernelSource, "plbvhLargeAabbRayTest", &error, m_parallelLinearBvhProgram, additionalMacros);
 	b3Assert(m_plbvhLargeAabbRayTestKernel);
 }
 
-b3GpuParallelLinearBvh::~b3GpuParallelLinearBvh() 
+b3GpuParallelLinearBvh::~b3GpuParallelLinearBvh()
 {
 	clReleaseKernel(m_separateAabbsKernel);
 	clReleaseKernel(m_findAllNodesMergedAabbKernel);
 	clReleaseKernel(m_assignMortonCodesAndAabbIndiciesKernel);
-	
+
 	clReleaseKernel(m_computeAdjacentPairCommonPrefixKernel);
 	clReleaseKernel(m_buildBinaryRadixTreeLeafNodesKernel);
 	clReleaseKernel(m_buildBinaryRadixTreeInternalNodesKernel);
 	clReleaseKernel(m_findDistanceFromRootKernel);
 	clReleaseKernel(m_buildBinaryRadixTreeAabbsRecursiveKernel);
-	
+
 	clReleaseKernel(m_findLeafIndexRangesKernel);
-	
+
 	clReleaseKernel(m_plbvhCalculateOverlappingPairsKernel);
 	clReleaseKernel(m_plbvhRayTraverseKernel);
 	clReleaseKernel(m_plbvhLargeAabbAabbTestKernel);
 	clReleaseKernel(m_plbvhLargeAabbRayTestKernel);
-	
+
 	clReleaseProgram(m_parallelLinearBvhProgram);
 }
 
-void b3GpuParallelLinearBvh::build(const b3OpenCLArray<b3SapAabb>& worldSpaceAabbs, const b3OpenCLArray<int>& smallAabbIndices, 
-									const b3OpenCLArray<int>& largeAabbIndices)
+void b3GpuParallelLinearBvh::build(const b3OpenCLArray<b3SapAabb>& worldSpaceAabbs, const b3OpenCLArray<int>& smallAabbIndices,
+								   const b3OpenCLArray<int>& largeAabbIndices)
 {
 	B3_PROFILE("b3ParallelLinearBvh::build()");
-	
+
 	int numLargeAabbs = largeAabbIndices.size();
 	int numSmallAabbs = smallAabbIndices.size();
-	
-	//Since all AABBs(both large and small) are input as a contiguous array, 
+
+	//Since all AABBs(both large and small) are input as a contiguous array,
 	//with 2 additional arrays used to indicate the indices of large and small AABBs,
 	//it is necessary to separate the AABBs so that the large AABBs will not degrade the quality of the BVH.
 	{
 		B3_PROFILE("Separate large and small AABBs");
-		
+
 		m_largeAabbs.resize(numLargeAabbs);
 		m_leafNodeAabbs.resize(numSmallAabbs);
-		
+
 		//Write large AABBs into m_largeAabbs
 		{
-			b3BufferInfoCL bufferInfo[] = 
-			{
-				b3BufferInfoCL( worldSpaceAabbs.getBufferCL() ),
-				b3BufferInfoCL( largeAabbIndices.getBufferCL() ),
-				
-				b3BufferInfoCL( m_largeAabbs.getBufferCL() )
-			};
-			
+			b3BufferInfoCL bufferInfo[] =
+				{
+					b3BufferInfoCL(worldSpaceAabbs.getBufferCL()),
+					b3BufferInfoCL(largeAabbIndices.getBufferCL()),
+
+					b3BufferInfoCL(m_largeAabbs.getBufferCL())};
+
 			b3LauncherCL launcher(m_queue, m_separateAabbsKernel, "m_separateAabbsKernel");
-			launcher.setBuffers( bufferInfo, sizeof(bufferInfo)/sizeof(b3BufferInfoCL) );
+			launcher.setBuffers(bufferInfo, sizeof(bufferInfo) / sizeof(b3BufferInfoCL));
 			launcher.setConst(numLargeAabbs);
-			
+
 			launcher.launch1D(numLargeAabbs);
 		}
-		
+
 		//Write small AABBs into m_leafNodeAabbs
 		{
-			b3BufferInfoCL bufferInfo[] = 
-			{
-				b3BufferInfoCL( worldSpaceAabbs.getBufferCL() ),
-				b3BufferInfoCL( smallAabbIndices.getBufferCL() ),
-				
-				b3BufferInfoCL( m_leafNodeAabbs.getBufferCL() )
-			};
-			
+			b3BufferInfoCL bufferInfo[] =
+				{
+					b3BufferInfoCL(worldSpaceAabbs.getBufferCL()),
+					b3BufferInfoCL(smallAabbIndices.getBufferCL()),
+
+					b3BufferInfoCL(m_leafNodeAabbs.getBufferCL())};
+
 			b3LauncherCL launcher(m_queue, m_separateAabbsKernel, "m_separateAabbsKernel");
-			launcher.setBuffers( bufferInfo, sizeof(bufferInfo)/sizeof(b3BufferInfoCL) );
+			launcher.setBuffers(bufferInfo, sizeof(bufferInfo) / sizeof(b3BufferInfoCL));
 			launcher.setConst(numSmallAabbs);
-			
+
 			launcher.launch1D(numSmallAabbs);
 		}
-		
+
 		clFinish(m_queue);
 	}
-	
+
 	//
-	int numLeaves = numSmallAabbs;	//Number of leaves in the BVH == Number of rigid bodies with small AABBs
+	int numLeaves = numSmallAabbs;  //Number of leaves in the BVH == Number of rigid bodies with small AABBs
 	int numInternalNodes = numLeaves - 1;
-	
-	if(numLeaves < 2)
+
+	if (numLeaves < 2)
 	{
 		//Number of leaf nodes is checked in calculateOverlappingPairs() and testRaysAgainstBvhAabbs(),
 		//so it does not matter if numLeaves == 0 and rootNodeIndex == -1
 		int rootNodeIndex = numLeaves - 1;
 		m_rootNodeIndex.copyFromHostPointer(&rootNodeIndex, 1);
-		
+
 		//Since the AABBs need to be rearranged(sorted) for the BVH construction algorithm,
 		//m_mortonCodesAndAabbIndicies.m_value is used to map a sorted AABB index to the unsorted AABB index
 		//instead of directly moving the AABBs. It needs to be set for the ray cast traversal kernel to work.
 		//( m_mortonCodesAndAabbIndicies[].m_value == unsorted index == index of m_leafNodeAabbs )
-		if(numLeaves == 1)
+		if (numLeaves == 1)
 		{
 			b3SortData leaf;
-			leaf.m_value = 0;		//1 leaf so index is always 0; leaf.m_key does not need to be set
-			
+			leaf.m_value = 0;  //1 leaf so index is always 0; leaf.m_key does not need to be set
+
 			m_mortonCodesAndAabbIndicies.resize(1);
 			m_mortonCodesAndAabbIndicies.copyFromHostPointer(&leaf, 1);
 		}
-		
+
 		return;
 	}
-	
+
 	//
 	{
 		m_internalNodeAabbs.resize(numInternalNodes);
@@ -197,37 +194,37 @@ void b3GpuParallelLinearBvh::build(const b3OpenCLArray<b3SapAabb>& worldSpaceAab
 		m_commonPrefixes.resize(numInternalNodes);
 		m_commonPrefixLengths.resize(numInternalNodes);
 		m_distanceFromRoot.resize(numInternalNodes);
-	
+
 		m_leafNodeParentNodes.resize(numLeaves);
 		m_mortonCodesAndAabbIndicies.resize(numLeaves);
 		m_mergedAabb.resize(numLeaves);
 	}
-	
-	//Find the merged AABB of all small AABBs; this is used to define the size of 
+
+	//Find the merged AABB of all small AABBs; this is used to define the size of
 	//each cell in the virtual grid for the next kernel(2^10 cells in each dimension).
 	{
 		B3_PROFILE("Find AABB of merged nodes");
-	
-		m_mergedAabb.copyFromOpenCLArray(m_leafNodeAabbs);	//Need to make a copy since the kernel modifies the array
-			
-		for(int numAabbsNeedingMerge = numLeaves; numAabbsNeedingMerge >= 2; 
-				numAabbsNeedingMerge = numAabbsNeedingMerge / 2 + numAabbsNeedingMerge % 2)
+
+		m_mergedAabb.copyFromOpenCLArray(m_leafNodeAabbs);  //Need to make a copy since the kernel modifies the array
+
+		for (int numAabbsNeedingMerge = numLeaves; numAabbsNeedingMerge >= 2;
+			 numAabbsNeedingMerge = numAabbsNeedingMerge / 2 + numAabbsNeedingMerge % 2)
 		{
-			b3BufferInfoCL bufferInfo[] = 
-			{
-				b3BufferInfoCL( m_mergedAabb.getBufferCL() )		//Resulting AABB is stored in m_mergedAabb[0]
-			};
-			
+			b3BufferInfoCL bufferInfo[] =
+				{
+					b3BufferInfoCL(m_mergedAabb.getBufferCL())  //Resulting AABB is stored in m_mergedAabb[0]
+				};
+
 			b3LauncherCL launcher(m_queue, m_findAllNodesMergedAabbKernel, "m_findAllNodesMergedAabbKernel");
-			launcher.setBuffers( bufferInfo, sizeof(bufferInfo)/sizeof(b3BufferInfoCL) );
+			launcher.setBuffers(bufferInfo, sizeof(bufferInfo) / sizeof(b3BufferInfoCL));
 			launcher.setConst(numAabbsNeedingMerge);
-			
+
 			launcher.launch1D(numAabbsNeedingMerge);
 		}
-		
+
 		clFinish(m_queue);
 	}
-	
+
 	//Insert the center of the AABBs into a virtual grid,
 	//then convert the discrete grid coordinates into a morton code
 	//For each element in m_mortonCodesAndAabbIndicies, set
@@ -235,34 +232,32 @@ void b3GpuParallelLinearBvh::build(const b3OpenCLArray<b3SapAabb>& worldSpaceAab
 	//	m_value == small AABB index
 	{
 		B3_PROFILE("Assign morton codes");
-	
-		b3BufferInfoCL bufferInfo[] = 
-		{
-			b3BufferInfoCL( m_leafNodeAabbs.getBufferCL() ),
-			b3BufferInfoCL( m_mergedAabb.getBufferCL() ),
-			b3BufferInfoCL( m_mortonCodesAndAabbIndicies.getBufferCL() )
-		};
-		
+
+		b3BufferInfoCL bufferInfo[] =
+			{
+				b3BufferInfoCL(m_leafNodeAabbs.getBufferCL()),
+				b3BufferInfoCL(m_mergedAabb.getBufferCL()),
+				b3BufferInfoCL(m_mortonCodesAndAabbIndicies.getBufferCL())};
+
 		b3LauncherCL launcher(m_queue, m_assignMortonCodesAndAabbIndiciesKernel, "m_assignMortonCodesAndAabbIndiciesKernel");
-		launcher.setBuffers( bufferInfo, sizeof(bufferInfo)/sizeof(b3BufferInfoCL) );
+		launcher.setBuffers(bufferInfo, sizeof(bufferInfo) / sizeof(b3BufferInfoCL));
 		launcher.setConst(numLeaves);
-		
+
 		launcher.launch1D(numLeaves);
 		clFinish(m_queue);
 	}
-	
+
 	//
 	{
 		B3_PROFILE("Sort leaves by morton codes");
-	
+
 		m_radixSorter.execute(m_mortonCodesAndAabbIndicies);
 		clFinish(m_queue);
 	}
-	
+
 	//
 	constructBinaryRadixTree();
-	
-	
+
 	//Since it is a sorted binary radix tree, each internal node contains a contiguous subset of leaf node indices.
 	//The root node contains leaf node indices in the range [0, numLeafNodes - 1].
 	//The child nodes of each node split their parent's index range into 2 contiguous halves.
@@ -273,17 +268,16 @@ void b3GpuParallelLinearBvh::build(const b3OpenCLArray<b3SapAabb>& worldSpaceAab
 	//This property can be used for optimizing calculateOverlappingPairs(), to avoid testing each AABB pair twice
 	{
 		B3_PROFILE("m_findLeafIndexRangesKernel");
-	
-		b3BufferInfoCL bufferInfo[] = 
-		{
-			b3BufferInfoCL( m_internalNodeChildNodes.getBufferCL() ),
-			b3BufferInfoCL( m_internalNodeLeafIndexRanges.getBufferCL() )
-		};
-		
+
+		b3BufferInfoCL bufferInfo[] =
+			{
+				b3BufferInfoCL(m_internalNodeChildNodes.getBufferCL()),
+				b3BufferInfoCL(m_internalNodeLeafIndexRanges.getBufferCL())};
+
 		b3LauncherCL launcher(m_queue, m_findLeafIndexRangesKernel, "m_findLeafIndexRangesKernel");
-		launcher.setBuffers( bufferInfo, sizeof(bufferInfo)/sizeof(b3BufferInfoCL) );
+		launcher.setBuffers(bufferInfo, sizeof(bufferInfo) / sizeof(b3BufferInfoCL));
 		launcher.setConst(numInternalNodes);
-		
+
 		launcher.launch1D(numInternalNodes);
 		clFinish(m_queue);
 	}
@@ -293,285 +287,271 @@ void b3GpuParallelLinearBvh::calculateOverlappingPairs(b3OpenCLArray<b3Int4>& ou
 {
 	int maxPairs = out_overlappingPairs.size();
 	b3OpenCLArray<int>& numPairsGpu = m_temp;
-	
+
 	int reset = 0;
 	numPairsGpu.copyFromHostPointer(&reset, 1);
-	
+
 	//
-	if( m_leafNodeAabbs.size() > 1 )
+	if (m_leafNodeAabbs.size() > 1)
 	{
 		B3_PROFILE("PLBVH small-small AABB test");
-	
+
 		int numQueryAabbs = m_leafNodeAabbs.size();
-		
-		b3BufferInfoCL bufferInfo[] = 
-		{
-			b3BufferInfoCL( m_leafNodeAabbs.getBufferCL() ),
-			
-			b3BufferInfoCL( m_rootNodeIndex.getBufferCL() ),
-			b3BufferInfoCL( m_internalNodeChildNodes.getBufferCL() ),
-			b3BufferInfoCL( m_internalNodeAabbs.getBufferCL() ),
-			b3BufferInfoCL( m_internalNodeLeafIndexRanges.getBufferCL() ),
-			b3BufferInfoCL( m_mortonCodesAndAabbIndicies.getBufferCL() ),
-			
-			b3BufferInfoCL( numPairsGpu.getBufferCL() ),
-			b3BufferInfoCL( out_overlappingPairs.getBufferCL() )
-		};
-		
+
+		b3BufferInfoCL bufferInfo[] =
+			{
+				b3BufferInfoCL(m_leafNodeAabbs.getBufferCL()),
+
+				b3BufferInfoCL(m_rootNodeIndex.getBufferCL()),
+				b3BufferInfoCL(m_internalNodeChildNodes.getBufferCL()),
+				b3BufferInfoCL(m_internalNodeAabbs.getBufferCL()),
+				b3BufferInfoCL(m_internalNodeLeafIndexRanges.getBufferCL()),
+				b3BufferInfoCL(m_mortonCodesAndAabbIndicies.getBufferCL()),
+
+				b3BufferInfoCL(numPairsGpu.getBufferCL()),
+				b3BufferInfoCL(out_overlappingPairs.getBufferCL())};
+
 		b3LauncherCL launcher(m_queue, m_plbvhCalculateOverlappingPairsKernel, "m_plbvhCalculateOverlappingPairsKernel");
-		launcher.setBuffers( bufferInfo, sizeof(bufferInfo)/sizeof(b3BufferInfoCL) );
+		launcher.setBuffers(bufferInfo, sizeof(bufferInfo) / sizeof(b3BufferInfoCL));
 		launcher.setConst(maxPairs);
 		launcher.setConst(numQueryAabbs);
-		
+
 		launcher.launch1D(numQueryAabbs);
 		clFinish(m_queue);
 	}
-	
+
 	int numLargeAabbRigids = m_largeAabbs.size();
-	if( numLargeAabbRigids > 0 && m_leafNodeAabbs.size() > 0 )
+	if (numLargeAabbRigids > 0 && m_leafNodeAabbs.size() > 0)
 	{
 		B3_PROFILE("PLBVH large-small AABB test");
-	
+
 		int numQueryAabbs = m_leafNodeAabbs.size();
-		
-		b3BufferInfoCL bufferInfo[] = 
-		{
-			b3BufferInfoCL( m_leafNodeAabbs.getBufferCL() ),
-			b3BufferInfoCL( m_largeAabbs.getBufferCL() ),
-			
-			b3BufferInfoCL( numPairsGpu.getBufferCL() ),
-			b3BufferInfoCL( out_overlappingPairs.getBufferCL() )
-		};
-		
+
+		b3BufferInfoCL bufferInfo[] =
+			{
+				b3BufferInfoCL(m_leafNodeAabbs.getBufferCL()),
+				b3BufferInfoCL(m_largeAabbs.getBufferCL()),
+
+				b3BufferInfoCL(numPairsGpu.getBufferCL()),
+				b3BufferInfoCL(out_overlappingPairs.getBufferCL())};
+
 		b3LauncherCL launcher(m_queue, m_plbvhLargeAabbAabbTestKernel, "m_plbvhLargeAabbAabbTestKernel");
-		launcher.setBuffers( bufferInfo, sizeof(bufferInfo)/sizeof(b3BufferInfoCL) );
+		launcher.setBuffers(bufferInfo, sizeof(bufferInfo) / sizeof(b3BufferInfoCL));
 		launcher.setConst(maxPairs);
 		launcher.setConst(numLargeAabbRigids);
 		launcher.setConst(numQueryAabbs);
-		
+
 		launcher.launch1D(numQueryAabbs);
 		clFinish(m_queue);
 	}
-	
-	
+
 	//
 	int numPairs = -1;
 	numPairsGpu.copyToHostPointer(&numPairs, 1);
-	if(numPairs > maxPairs)
+	if (numPairs > maxPairs)
 	{
 		b3Error("Error running out of pairs: numPairs = %d, maxPairs = %d.\n", numPairs, maxPairs);
 		numPairs = maxPairs;
 		numPairsGpu.copyFromHostPointer(&maxPairs, 1);
 	}
-	
+
 	out_overlappingPairs.resize(numPairs);
 }
 
-
-void b3GpuParallelLinearBvh::testRaysAgainstBvhAabbs(const b3OpenCLArray<b3RayInfo>& rays, 
-							b3OpenCLArray<int>& out_numRayRigidPairs, b3OpenCLArray<b3Int2>& out_rayRigidPairs)
+void b3GpuParallelLinearBvh::testRaysAgainstBvhAabbs(const b3OpenCLArray<b3RayInfo>& rays,
+													 b3OpenCLArray<int>& out_numRayRigidPairs, b3OpenCLArray<b3Int2>& out_rayRigidPairs)
 {
 	B3_PROFILE("PLBVH testRaysAgainstBvhAabbs()");
-	
+
 	int numRays = rays.size();
 	int maxRayRigidPairs = out_rayRigidPairs.size();
-	
+
 	int reset = 0;
 	out_numRayRigidPairs.copyFromHostPointer(&reset, 1);
-	
+
 	//
-	if( m_leafNodeAabbs.size() > 0 )
+	if (m_leafNodeAabbs.size() > 0)
 	{
 		B3_PROFILE("PLBVH ray test small AABB");
-	
-		b3BufferInfoCL bufferInfo[] = 
-		{
-			b3BufferInfoCL( m_leafNodeAabbs.getBufferCL() ),
-			
-			b3BufferInfoCL( m_rootNodeIndex.getBufferCL() ),
-			b3BufferInfoCL( m_internalNodeChildNodes.getBufferCL() ),
-			b3BufferInfoCL( m_internalNodeAabbs.getBufferCL() ),
-			b3BufferInfoCL( m_internalNodeLeafIndexRanges.getBufferCL() ),
-			b3BufferInfoCL( m_mortonCodesAndAabbIndicies.getBufferCL() ),
-			
-			b3BufferInfoCL( rays.getBufferCL() ),
-			
-			b3BufferInfoCL( out_numRayRigidPairs.getBufferCL() ),
-			b3BufferInfoCL( out_rayRigidPairs.getBufferCL() )
-		};
-		
+
+		b3BufferInfoCL bufferInfo[] =
+			{
+				b3BufferInfoCL(m_leafNodeAabbs.getBufferCL()),
+
+				b3BufferInfoCL(m_rootNodeIndex.getBufferCL()),
+				b3BufferInfoCL(m_internalNodeChildNodes.getBufferCL()),
+				b3BufferInfoCL(m_internalNodeAabbs.getBufferCL()),
+				b3BufferInfoCL(m_internalNodeLeafIndexRanges.getBufferCL()),
+				b3BufferInfoCL(m_mortonCodesAndAabbIndicies.getBufferCL()),
+
+				b3BufferInfoCL(rays.getBufferCL()),
+
+				b3BufferInfoCL(out_numRayRigidPairs.getBufferCL()),
+				b3BufferInfoCL(out_rayRigidPairs.getBufferCL())};
+
 		b3LauncherCL launcher(m_queue, m_plbvhRayTraverseKernel, "m_plbvhRayTraverseKernel");
-		launcher.setBuffers( bufferInfo, sizeof(bufferInfo)/sizeof(b3BufferInfoCL) );
+		launcher.setBuffers(bufferInfo, sizeof(bufferInfo) / sizeof(b3BufferInfoCL));
 		launcher.setConst(maxRayRigidPairs);
 		launcher.setConst(numRays);
-		
+
 		launcher.launch1D(numRays);
 		clFinish(m_queue);
 	}
-	
+
 	int numLargeAabbRigids = m_largeAabbs.size();
-	if(numLargeAabbRigids > 0)
+	if (numLargeAabbRigids > 0)
 	{
 		B3_PROFILE("PLBVH ray test large AABB");
-		
-		b3BufferInfoCL bufferInfo[] = 
-		{
-			b3BufferInfoCL( m_largeAabbs.getBufferCL() ),
-			b3BufferInfoCL( rays.getBufferCL() ),
-			
-			b3BufferInfoCL( out_numRayRigidPairs.getBufferCL() ),
-			b3BufferInfoCL( out_rayRigidPairs.getBufferCL() )
-		};
-		
+
+		b3BufferInfoCL bufferInfo[] =
+			{
+				b3BufferInfoCL(m_largeAabbs.getBufferCL()),
+				b3BufferInfoCL(rays.getBufferCL()),
+
+				b3BufferInfoCL(out_numRayRigidPairs.getBufferCL()),
+				b3BufferInfoCL(out_rayRigidPairs.getBufferCL())};
+
 		b3LauncherCL launcher(m_queue, m_plbvhLargeAabbRayTestKernel, "m_plbvhLargeAabbRayTestKernel");
-		launcher.setBuffers( bufferInfo, sizeof(bufferInfo)/sizeof(b3BufferInfoCL) );
+		launcher.setBuffers(bufferInfo, sizeof(bufferInfo) / sizeof(b3BufferInfoCL));
 		launcher.setConst(numLargeAabbRigids);
 		launcher.setConst(maxRayRigidPairs);
 		launcher.setConst(numRays);
-		
+
 		launcher.launch1D(numRays);
 		clFinish(m_queue);
 	}
-	
+
 	//
 	int numRayRigidPairs = -1;
 	out_numRayRigidPairs.copyToHostPointer(&numRayRigidPairs, 1);
-	
-	if(numRayRigidPairs > maxRayRigidPairs)
+
+	if (numRayRigidPairs > maxRayRigidPairs)
 		b3Error("Error running out of rayRigid pairs: numRayRigidPairs = %d, maxRayRigidPairs = %d.\n", numRayRigidPairs, maxRayRigidPairs);
-	
 }
 
 void b3GpuParallelLinearBvh::constructBinaryRadixTree()
 {
 	B3_PROFILE("b3GpuParallelLinearBvh::constructBinaryRadixTree()");
-	
+
 	int numLeaves = m_leafNodeAabbs.size();
 	int numInternalNodes = numLeaves - 1;
-	
+
 	//Each internal node is placed in between 2 leaf nodes.
 	//By using this arrangement and computing the common prefix between
 	//these 2 adjacent leaf nodes, it is possible to quickly construct a binary radix tree.
 	{
 		B3_PROFILE("m_computeAdjacentPairCommonPrefixKernel");
-		
-		b3BufferInfoCL bufferInfo[] = 
-		{
-			b3BufferInfoCL( m_mortonCodesAndAabbIndicies.getBufferCL() ),
-			b3BufferInfoCL( m_commonPrefixes.getBufferCL() ),
-			b3BufferInfoCL( m_commonPrefixLengths.getBufferCL() )
-		};
-		
+
+		b3BufferInfoCL bufferInfo[] =
+			{
+				b3BufferInfoCL(m_mortonCodesAndAabbIndicies.getBufferCL()),
+				b3BufferInfoCL(m_commonPrefixes.getBufferCL()),
+				b3BufferInfoCL(m_commonPrefixLengths.getBufferCL())};
+
 		b3LauncherCL launcher(m_queue, m_computeAdjacentPairCommonPrefixKernel, "m_computeAdjacentPairCommonPrefixKernel");
-		launcher.setBuffers( bufferInfo, sizeof(bufferInfo)/sizeof(b3BufferInfoCL) );
+		launcher.setBuffers(bufferInfo, sizeof(bufferInfo) / sizeof(b3BufferInfoCL));
 		launcher.setConst(numInternalNodes);
-		
+
 		launcher.launch1D(numInternalNodes);
 		clFinish(m_queue);
 	}
-	
-	//For each leaf node, select its parent node by 
+
+	//For each leaf node, select its parent node by
 	//comparing the 2 nearest internal nodes and assign child node indices
 	{
 		B3_PROFILE("m_buildBinaryRadixTreeLeafNodesKernel");
-		
-		b3BufferInfoCL bufferInfo[] = 
-		{
-			b3BufferInfoCL( m_commonPrefixLengths.getBufferCL() ),
-			b3BufferInfoCL( m_leafNodeParentNodes.getBufferCL() ),
-			b3BufferInfoCL( m_internalNodeChildNodes.getBufferCL() )
-		};
-		
+
+		b3BufferInfoCL bufferInfo[] =
+			{
+				b3BufferInfoCL(m_commonPrefixLengths.getBufferCL()),
+				b3BufferInfoCL(m_leafNodeParentNodes.getBufferCL()),
+				b3BufferInfoCL(m_internalNodeChildNodes.getBufferCL())};
+
 		b3LauncherCL launcher(m_queue, m_buildBinaryRadixTreeLeafNodesKernel, "m_buildBinaryRadixTreeLeafNodesKernel");
-		launcher.setBuffers( bufferInfo, sizeof(bufferInfo)/sizeof(b3BufferInfoCL) );
+		launcher.setBuffers(bufferInfo, sizeof(bufferInfo) / sizeof(b3BufferInfoCL));
 		launcher.setConst(numLeaves);
-		
+
 		launcher.launch1D(numLeaves);
 		clFinish(m_queue);
 	}
-	
+
 	//For each internal node, perform 2 binary searches among the other internal nodes
 	//to its left and right to find its potential parent nodes and assign child node indices
 	{
 		B3_PROFILE("m_buildBinaryRadixTreeInternalNodesKernel");
-		
-		b3BufferInfoCL bufferInfo[] = 
-		{
-			b3BufferInfoCL( m_commonPrefixes.getBufferCL() ),
-			b3BufferInfoCL( m_commonPrefixLengths.getBufferCL() ),
-			b3BufferInfoCL( m_internalNodeChildNodes.getBufferCL() ),
-			b3BufferInfoCL( m_internalNodeParentNodes.getBufferCL() ),
-			b3BufferInfoCL( m_rootNodeIndex.getBufferCL() )
-		};
-		
+
+		b3BufferInfoCL bufferInfo[] =
+			{
+				b3BufferInfoCL(m_commonPrefixes.getBufferCL()),
+				b3BufferInfoCL(m_commonPrefixLengths.getBufferCL()),
+				b3BufferInfoCL(m_internalNodeChildNodes.getBufferCL()),
+				b3BufferInfoCL(m_internalNodeParentNodes.getBufferCL()),
+				b3BufferInfoCL(m_rootNodeIndex.getBufferCL())};
+
 		b3LauncherCL launcher(m_queue, m_buildBinaryRadixTreeInternalNodesKernel, "m_buildBinaryRadixTreeInternalNodesKernel");
-		launcher.setBuffers( bufferInfo, sizeof(bufferInfo)/sizeof(b3BufferInfoCL) );
+		launcher.setBuffers(bufferInfo, sizeof(bufferInfo) / sizeof(b3BufferInfoCL));
 		launcher.setConst(numInternalNodes);
-		
+
 		launcher.launch1D(numInternalNodes);
 		clFinish(m_queue);
 	}
-	
+
 	//Find the number of nodes seperating each internal node and the root node
 	//so that the AABBs can be set using the next kernel.
 	//Also determine the maximum number of nodes separating an internal node and the root node.
 	{
 		B3_PROFILE("m_findDistanceFromRootKernel");
-	
-		b3BufferInfoCL bufferInfo[] = 
-		{
-			b3BufferInfoCL( m_rootNodeIndex.getBufferCL() ),
-			b3BufferInfoCL( m_internalNodeParentNodes.getBufferCL() ),
-			b3BufferInfoCL( m_maxDistanceFromRoot.getBufferCL() ),
-			b3BufferInfoCL( m_distanceFromRoot.getBufferCL() )
-		};
-		
+
+		b3BufferInfoCL bufferInfo[] =
+			{
+				b3BufferInfoCL(m_rootNodeIndex.getBufferCL()),
+				b3BufferInfoCL(m_internalNodeParentNodes.getBufferCL()),
+				b3BufferInfoCL(m_maxDistanceFromRoot.getBufferCL()),
+				b3BufferInfoCL(m_distanceFromRoot.getBufferCL())};
+
 		b3LauncherCL launcher(m_queue, m_findDistanceFromRootKernel, "m_findDistanceFromRootKernel");
-		launcher.setBuffers( bufferInfo, sizeof(bufferInfo)/sizeof(b3BufferInfoCL) );
+		launcher.setBuffers(bufferInfo, sizeof(bufferInfo) / sizeof(b3BufferInfoCL));
 		launcher.setConst(numInternalNodes);
-		
+
 		launcher.launch1D(numInternalNodes);
 		clFinish(m_queue);
 	}
-	
+
 	//Starting from the internal nodes nearest to the leaf nodes, recursively move up
 	//the tree towards the root to set the AABBs of each internal node; each internal node
 	//checks its children and merges their AABBs
 	{
 		B3_PROFILE("m_buildBinaryRadixTreeAabbsRecursiveKernel");
-		
+
 		int maxDistanceFromRoot = -1;
 		{
 			B3_PROFILE("copy maxDistanceFromRoot to CPU");
 			m_maxDistanceFromRoot.copyToHostPointer(&maxDistanceFromRoot, 1);
 			clFinish(m_queue);
 		}
-		
-		for(int distanceFromRoot = maxDistanceFromRoot; distanceFromRoot >= 0; --distanceFromRoot)
+
+		for (int distanceFromRoot = maxDistanceFromRoot; distanceFromRoot >= 0; --distanceFromRoot)
 		{
-			b3BufferInfoCL bufferInfo[] = 
-			{
-				b3BufferInfoCL( m_distanceFromRoot.getBufferCL() ),
-				b3BufferInfoCL( m_mortonCodesAndAabbIndicies.getBufferCL() ),
-				b3BufferInfoCL( m_internalNodeChildNodes.getBufferCL() ),
-				b3BufferInfoCL( m_leafNodeAabbs.getBufferCL() ),
-				b3BufferInfoCL( m_internalNodeAabbs.getBufferCL() )
-			};
-			
+			b3BufferInfoCL bufferInfo[] =
+				{
+					b3BufferInfoCL(m_distanceFromRoot.getBufferCL()),
+					b3BufferInfoCL(m_mortonCodesAndAabbIndicies.getBufferCL()),
+					b3BufferInfoCL(m_internalNodeChildNodes.getBufferCL()),
+					b3BufferInfoCL(m_leafNodeAabbs.getBufferCL()),
+					b3BufferInfoCL(m_internalNodeAabbs.getBufferCL())};
+
 			b3LauncherCL launcher(m_queue, m_buildBinaryRadixTreeAabbsRecursiveKernel, "m_buildBinaryRadixTreeAabbsRecursiveKernel");
-			launcher.setBuffers( bufferInfo, sizeof(bufferInfo)/sizeof(b3BufferInfoCL) );
+			launcher.setBuffers(bufferInfo, sizeof(bufferInfo) / sizeof(b3BufferInfoCL));
 			launcher.setConst(maxDistanceFromRoot);
 			launcher.setConst(distanceFromRoot);
 			launcher.setConst(numInternalNodes);
-			
+
 			//It may seem inefficent to launch a thread for each internal node when a
 			//much smaller number of nodes is actually processed, but this is actually
-			//faster than determining the exact nodes that are ready to merge their child AABBs. 
+			//faster than determining the exact nodes that are ready to merge their child AABBs.
 			launcher.launch1D(numInternalNodes);
 		}
-		
+
 		clFinish(m_queue);
 	}
 }
-
-	
\ No newline at end of file
diff --git a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuParallelLinearBvh.h b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuParallelLinearBvh.h
index effe617b7b8..b3907751298 100644
--- a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuParallelLinearBvh.h
+++ b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuParallelLinearBvh.h
@@ -37,10 +37,10 @@ subject to the following restrictions:
 ///"Maximizing Parallelism in the Construction of BVHs, Octrees, and k-d trees" [Karras 2012] \n
 ///@par
 ///The basic algorithm for building the BVH as presented in [Lauterbach et al. 2009] consists of 4 stages:
-/// - [fully parallel] Assign morton codes for each AABB using its center (after quantizing the AABB centers into a virtual grid) 
+/// - [fully parallel] Assign morton codes for each AABB using its center (after quantizing the AABB centers into a virtual grid)
 /// - [fully parallel] Sort morton codes
-/// - [somewhat parallel] Build binary radix tree (assign parent/child pointers for internal nodes of the BVH) 
-/// - [somewhat parallel] Set internal node AABBs 
+/// - [somewhat parallel] Build binary radix tree (assign parent/child pointers for internal nodes of the BVH)
+/// - [somewhat parallel] Set internal node AABBs
 ///@par
 ///[Karras 2012] improves on the algorithm by introducing fully parallel methods for the last 2 stages.
 ///The BVH implementation here shares many concepts with [Karras 2012], but a different method is used for constructing the tree.
@@ -49,75 +49,75 @@ subject to the following restrictions:
 class b3GpuParallelLinearBvh
 {
 	cl_command_queue m_queue;
-	
+
 	cl_program m_parallelLinearBvhProgram;
-	
+
 	cl_kernel m_separateAabbsKernel;
 	cl_kernel m_findAllNodesMergedAabbKernel;
 	cl_kernel m_assignMortonCodesAndAabbIndiciesKernel;
-	
+
 	//Binary radix tree construction kernels
 	cl_kernel m_computeAdjacentPairCommonPrefixKernel;
 	cl_kernel m_buildBinaryRadixTreeLeafNodesKernel;
 	cl_kernel m_buildBinaryRadixTreeInternalNodesKernel;
 	cl_kernel m_findDistanceFromRootKernel;
 	cl_kernel m_buildBinaryRadixTreeAabbsRecursiveKernel;
-	
+
 	cl_kernel m_findLeafIndexRangesKernel;
-	
+
 	//Traversal kernels
 	cl_kernel m_plbvhCalculateOverlappingPairsKernel;
 	cl_kernel m_plbvhRayTraverseKernel;
 	cl_kernel m_plbvhLargeAabbAabbTestKernel;
 	cl_kernel m_plbvhLargeAabbRayTestKernel;
-	
+
 	b3RadixSort32CL m_radixSorter;
-	
+
 	//1 element
-	b3OpenCLArray<int> m_rootNodeIndex;							//Most significant bit(0x80000000) is set to indicate internal node
-	b3OpenCLArray<int> m_maxDistanceFromRoot;					//Max number of internal nodes between an internal node and the root node
-	b3OpenCLArray<int> m_temp;									//Used to hold the number of pairs in calculateOverlappingPairs()
-	
+	b3OpenCLArray<int> m_rootNodeIndex;        //Most significant bit(0x80000000) is set to indicate internal node
+	b3OpenCLArray<int> m_maxDistanceFromRoot;  //Max number of internal nodes between an internal node and the root node
+	b3OpenCLArray<int> m_temp;                 //Used to hold the number of pairs in calculateOverlappingPairs()
+
 	//1 element per internal node (number_of_internal_nodes == number_of_leaves - 1)
 	b3OpenCLArray<b3SapAabb> m_internalNodeAabbs;
-	b3OpenCLArray<b3Int2> m_internalNodeLeafIndexRanges;		//x == min leaf index, y == max leaf index
-	b3OpenCLArray<b3Int2> m_internalNodeChildNodes;				//x == left child, y == right child; msb(0x80000000) is set to indicate internal node
-	b3OpenCLArray<int> m_internalNodeParentNodes;				//For parent node index, msb(0x80000000) is not set since it is always internal
-	
+	b3OpenCLArray<b3Int2> m_internalNodeLeafIndexRanges;  //x == min leaf index, y == max leaf index
+	b3OpenCLArray<b3Int2> m_internalNodeChildNodes;       //x == left child, y == right child; msb(0x80000000) is set to indicate internal node
+	b3OpenCLArray<int> m_internalNodeParentNodes;         //For parent node index, msb(0x80000000) is not set since it is always internal
+
 	//1 element per internal node; for binary radix tree construction
 	b3OpenCLArray<b3Int64> m_commonPrefixes;
 	b3OpenCLArray<int> m_commonPrefixLengths;
-	b3OpenCLArray<int> m_distanceFromRoot;						//Number of internal nodes between this node and the root
-	
+	b3OpenCLArray<int> m_distanceFromRoot;  //Number of internal nodes between this node and the root
+
 	//1 element per leaf node (leaf nodes only include small AABBs)
-	b3OpenCLArray<int> m_leafNodeParentNodes;					//For parent node index, msb(0x80000000) is not set since it is always internal
-	b3OpenCLArray<b3SortData> m_mortonCodesAndAabbIndicies;		//m_key == morton code, m_value == aabb index in m_leafNodeAabbs
-	b3OpenCLArray<b3SapAabb> m_mergedAabb;						//m_mergedAabb[0] contains the merged AABB of all leaf nodes
-	b3OpenCLArray<b3SapAabb> m_leafNodeAabbs;					//Contains only small AABBs
-	
+	b3OpenCLArray<int> m_leafNodeParentNodes;                //For parent node index, msb(0x80000000) is not set since it is always internal
+	b3OpenCLArray<b3SortData> m_mortonCodesAndAabbIndicies;  //m_key == morton code, m_value == aabb index in m_leafNodeAabbs
+	b3OpenCLArray<b3SapAabb> m_mergedAabb;                   //m_mergedAabb[0] contains the merged AABB of all leaf nodes
+	b3OpenCLArray<b3SapAabb> m_leafNodeAabbs;                //Contains only small AABBs
+
 	//1 element per large AABB, which is not stored in the BVH
 	b3OpenCLArray<b3SapAabb> m_largeAabbs;
-	
+
 public:
 	b3GpuParallelLinearBvh(cl_context context, cl_device_id device, cl_command_queue queue);
 	virtual ~b3GpuParallelLinearBvh();
-	
+
 	///Must be called before any other function
-	void build(const b3OpenCLArray<b3SapAabb>& worldSpaceAabbs, const b3OpenCLArray<int>& smallAabbIndices, 
-				const b3OpenCLArray<int>& largeAabbIndices);
-	
+	void build(const b3OpenCLArray<b3SapAabb>& worldSpaceAabbs, const b3OpenCLArray<int>& smallAabbIndices,
+			   const b3OpenCLArray<int>& largeAabbIndices);
+
 	///calculateOverlappingPairs() uses the worldSpaceAabbs parameter of b3GpuParallelLinearBvh::build() as the query AABBs.
 	///@param out_overlappingPairs The size() of this array is used to determine the max number of pairs.
 	///If the number of overlapping pairs is < out_overlappingPairs.size(), out_overlappingPairs is resized.
 	void calculateOverlappingPairs(b3OpenCLArray<b3Int4>& out_overlappingPairs);
-	
+
 	///@param out_numRigidRayPairs Array of length 1; contains the number of detected ray-rigid AABB intersections;
 	///this value may be greater than out_rayRigidPairs.size() if out_rayRigidPairs is not large enough.
 	///@param out_rayRigidPairs Contains an array of rays intersecting rigid AABBs; x == ray index, y == rigid body index.
 	///If the size of this array is insufficient to hold all ray-rigid AABB intersections, additional intersections are discarded.
-	void testRaysAgainstBvhAabbs(const b3OpenCLArray<b3RayInfo>& rays, 
-								b3OpenCLArray<int>& out_numRayRigidPairs, b3OpenCLArray<b3Int2>& out_rayRigidPairs);
-								
+	void testRaysAgainstBvhAabbs(const b3OpenCLArray<b3RayInfo>& rays,
+								 b3OpenCLArray<int>& out_numRayRigidPairs, b3OpenCLArray<b3Int2>& out_rayRigidPairs);
+
 private:
 	void constructBinaryRadixTree();
 };
diff --git a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuParallelLinearBvhBroadphase.cpp b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuParallelLinearBvhBroadphase.cpp
index d2618024ac4..62ea7a32df4 100644
--- a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuParallelLinearBvhBroadphase.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuParallelLinearBvhBroadphase.cpp
@@ -13,45 +13,44 @@ subject to the following restrictions:
 
 #include "b3GpuParallelLinearBvhBroadphase.h"
 
-b3GpuParallelLinearBvhBroadphase::b3GpuParallelLinearBvhBroadphase(cl_context context, cl_device_id device, cl_command_queue queue) : 
-	m_plbvh(context, device, queue),
-	
-	m_overlappingPairsGpu(context, queue),
-	
-	m_aabbsGpu(context, queue),
-	m_smallAabbsMappingGpu(context, queue),
-	m_largeAabbsMappingGpu(context, queue)
+b3GpuParallelLinearBvhBroadphase::b3GpuParallelLinearBvhBroadphase(cl_context context, cl_device_id device, cl_command_queue queue) : m_plbvh(context, device, queue),
+
+																																	  m_overlappingPairsGpu(context, queue),
+
+																																	  m_aabbsGpu(context, queue),
+																																	  m_smallAabbsMappingGpu(context, queue),
+																																	  m_largeAabbsMappingGpu(context, queue)
 {
 }
 
-void b3GpuParallelLinearBvhBroadphase::createProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, int userPtr,  int collisionFilterGroup,  int collisionFilterMask)
+void b3GpuParallelLinearBvhBroadphase::createProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, int userPtr, int collisionFilterGroup, int collisionFilterMask)
 {
 	int newAabbIndex = m_aabbsCpu.size();
 
 	b3SapAabb aabb;
 	aabb.m_minVec = aabbMin;
 	aabb.m_maxVec = aabbMax;
-	
+
 	aabb.m_minIndices[3] = userPtr;
 	aabb.m_signedMaxIndices[3] = newAabbIndex;
-	
+
 	m_smallAabbsMappingCpu.push_back(newAabbIndex);
-	
+
 	m_aabbsCpu.push_back(aabb);
 }
-void b3GpuParallelLinearBvhBroadphase::createLargeProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, int userPtr,  int collisionFilterGroup,  int collisionFilterMask)
+void b3GpuParallelLinearBvhBroadphase::createLargeProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, int userPtr, int collisionFilterGroup, int collisionFilterMask)
 {
 	int newAabbIndex = m_aabbsCpu.size();
 
 	b3SapAabb aabb;
 	aabb.m_minVec = aabbMin;
 	aabb.m_maxVec = aabbMax;
-	
+
 	aabb.m_minIndices[3] = userPtr;
 	aabb.m_signedMaxIndices[3] = newAabbIndex;
-	
+
 	m_largeAabbsMappingCpu.push_back(newAabbIndex);
-	
+
 	m_aabbsCpu.push_back(aabb);
 }
 
@@ -59,22 +58,19 @@ void b3GpuParallelLinearBvhBroadphase::calculateOverlappingPairs(int maxPairs)
 {
 	//Reconstruct BVH
 	m_plbvh.build(m_aabbsGpu, m_smallAabbsMappingGpu, m_largeAabbsMappingGpu);
-	
+
 	//
 	m_overlappingPairsGpu.resize(maxPairs);
 	m_plbvh.calculateOverlappingPairs(m_overlappingPairsGpu);
 }
 void b3GpuParallelLinearBvhBroadphase::calculateOverlappingPairsHost(int maxPairs)
 {
-	b3Assert(0);	//CPU version not implemented
+	b3Assert(0);  //CPU version not implemented
 }
 
-void b3GpuParallelLinearBvhBroadphase::writeAabbsToGpu() 
-{ 
-	m_aabbsGpu.copyFromHost(m_aabbsCpu); 
+void b3GpuParallelLinearBvhBroadphase::writeAabbsToGpu()
+{
+	m_aabbsGpu.copyFromHost(m_aabbsCpu);
 	m_smallAabbsMappingGpu.copyFromHost(m_smallAabbsMappingCpu);
 	m_largeAabbsMappingGpu.copyFromHost(m_largeAabbsMappingCpu);
 }
-	
-	
-	
diff --git a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuParallelLinearBvhBroadphase.h b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuParallelLinearBvhBroadphase.h
index e5185006371..dda0eea7bea 100644
--- a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuParallelLinearBvhBroadphase.h
+++ b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuParallelLinearBvhBroadphase.h
@@ -21,42 +21,42 @@ subject to the following restrictions:
 class b3GpuParallelLinearBvhBroadphase : public b3GpuBroadphaseInterface
 {
 	b3GpuParallelLinearBvh m_plbvh;
-	
+
 	b3OpenCLArray<b3Int4> m_overlappingPairsGpu;
-	
+
 	b3OpenCLArray<b3SapAabb> m_aabbsGpu;
 	b3OpenCLArray<int> m_smallAabbsMappingGpu;
 	b3OpenCLArray<int> m_largeAabbsMappingGpu;
-	
+
 	b3AlignedObjectArray<b3SapAabb> m_aabbsCpu;
 	b3AlignedObjectArray<int> m_smallAabbsMappingCpu;
 	b3AlignedObjectArray<int> m_largeAabbsMappingCpu;
-	
+
 public:
 	b3GpuParallelLinearBvhBroadphase(cl_context context, cl_device_id device, cl_command_queue queue);
 	virtual ~b3GpuParallelLinearBvhBroadphase() {}
 
-	virtual void createProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, int userPtr,  int collisionFilterGroup,  int collisionFilterMask);
-	virtual void createLargeProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, int userPtr,  int collisionFilterGroup,  int collisionFilterMask);
-	
+	virtual void createProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, int userPtr, int collisionFilterGroup, int collisionFilterMask);
+	virtual void createLargeProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, int userPtr, int collisionFilterGroup, int collisionFilterMask);
+
 	virtual void calculateOverlappingPairs(int maxPairs);
 	virtual void calculateOverlappingPairsHost(int maxPairs);
 
 	//call writeAabbsToGpu after done making all changes (createProxy etc)
 	virtual void writeAabbsToGpu();
-	
-	virtual int	getNumOverlap() { return m_overlappingPairsGpu.size(); }
+
+	virtual int getNumOverlap() { return m_overlappingPairsGpu.size(); }
 	virtual cl_mem getOverlappingPairBuffer() { return m_overlappingPairsGpu.getBufferCL(); }
 
 	virtual cl_mem getAabbBufferWS() { return m_aabbsGpu.getBufferCL(); }
 	virtual b3OpenCLArray<b3SapAabb>& getAllAabbsGPU() { return m_aabbsGpu; }
-	
+
 	virtual b3OpenCLArray<b3Int4>& getOverlappingPairsGPU() { return m_overlappingPairsGpu; }
 	virtual b3OpenCLArray<int>& getSmallAabbIndicesGPU() { return m_smallAabbsMappingGpu; }
 	virtual b3OpenCLArray<int>& getLargeAabbIndicesGPU() { return m_largeAabbsMappingGpu; }
-	
+
 	virtual b3AlignedObjectArray<b3SapAabb>& getAllAabbsCPU() { return m_aabbsCpu; }
-	
+
 	static b3GpuBroadphaseInterface* CreateFunc(cl_context context, cl_device_id device, cl_command_queue queue)
 	{
 		return new b3GpuParallelLinearBvhBroadphase(context, device, queue);
diff --git a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuSapBroadphase.cpp b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuSapBroadphase.cpp
index c45fbbdcaa3..4126d03ed0a 100644
--- a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuSapBroadphase.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuSapBroadphase.cpp
@@ -6,7 +6,6 @@ bool searchIncremental3dSapOnGpu = true;
 #include "Bullet3OpenCL/ParallelPrimitives/b3LauncherCL.h"
 #include "Bullet3OpenCL/ParallelPrimitives/b3PrefixScanFloat4CL.h"
 
-
 #include "Bullet3OpenCL/Initialize/b3OpenCLUtils.h"
 #include "kernels/sapKernels.h"
 
@@ -56,110 +55,105 @@ bool searchIncremental3dSapOnGpu = true;
 	class b3PrefixScanFloat4CL*		m_prefixScanFloat4;
  */
 
-b3GpuSapBroadphase::b3GpuSapBroadphase(cl_context ctx,cl_device_id device, cl_command_queue  q , b3GpuSapKernelType kernelType)
-:m_context(ctx),
-m_device(device),
-m_queue(q),
+b3GpuSapBroadphase::b3GpuSapBroadphase(cl_context ctx, cl_device_id device, cl_command_queue q, b3GpuSapKernelType kernelType)
+	: m_context(ctx),
+	  m_device(device),
+	  m_queue(q),
 
-m_objectMinMaxIndexGPUaxis0(ctx,q),
-m_objectMinMaxIndexGPUaxis1(ctx,q),
-m_objectMinMaxIndexGPUaxis2(ctx,q),
-m_objectMinMaxIndexGPUaxis0prev(ctx,q),
-m_objectMinMaxIndexGPUaxis1prev(ctx,q),
-m_objectMinMaxIndexGPUaxis2prev(ctx,q),
-m_sortedAxisGPU0(ctx,q),
-m_sortedAxisGPU1(ctx,q),
-m_sortedAxisGPU2(ctx,q),
-m_sortedAxisGPU0prev(ctx,q),
-m_sortedAxisGPU1prev(ctx,q),
-m_sortedAxisGPU2prev(ctx,q),
-m_addedHostPairsGPU(ctx,q),
-m_removedHostPairsGPU(ctx,q),
-m_addedCountGPU(ctx,q),
-m_removedCountGPU(ctx,q),
-m_currentBuffer(-1),
-m_pairCount(ctx,q),
-m_allAabbsGPU(ctx,q),
-m_sum(ctx,q),
-m_sum2(ctx,q),
-m_dst(ctx,q),
-m_smallAabbsMappingGPU(ctx,q),
-m_largeAabbsMappingGPU(ctx,q),
-m_overlappingPairs(ctx,q),
-m_gpuSmallSortData(ctx,q),
-m_gpuSmallSortedAabbs(ctx,q)
+	  m_objectMinMaxIndexGPUaxis0(ctx, q),
+	  m_objectMinMaxIndexGPUaxis1(ctx, q),
+	  m_objectMinMaxIndexGPUaxis2(ctx, q),
+	  m_objectMinMaxIndexGPUaxis0prev(ctx, q),
+	  m_objectMinMaxIndexGPUaxis1prev(ctx, q),
+	  m_objectMinMaxIndexGPUaxis2prev(ctx, q),
+	  m_sortedAxisGPU0(ctx, q),
+	  m_sortedAxisGPU1(ctx, q),
+	  m_sortedAxisGPU2(ctx, q),
+	  m_sortedAxisGPU0prev(ctx, q),
+	  m_sortedAxisGPU1prev(ctx, q),
+	  m_sortedAxisGPU2prev(ctx, q),
+	  m_addedHostPairsGPU(ctx, q),
+	  m_removedHostPairsGPU(ctx, q),
+	  m_addedCountGPU(ctx, q),
+	  m_removedCountGPU(ctx, q),
+	  m_currentBuffer(-1),
+	  m_pairCount(ctx, q),
+	  m_allAabbsGPU(ctx, q),
+	  m_sum(ctx, q),
+	  m_sum2(ctx, q),
+	  m_dst(ctx, q),
+	  m_smallAabbsMappingGPU(ctx, q),
+	  m_largeAabbsMappingGPU(ctx, q),
+	  m_overlappingPairs(ctx, q),
+	  m_gpuSmallSortData(ctx, q),
+	  m_gpuSmallSortedAabbs(ctx, q)
 {
 	const char* sapSrc = sapCL;
-    
-    
-	cl_int errNum=0;
+
+	cl_int errNum = 0;
 
 	b3Assert(m_context);
 	b3Assert(m_device);
-	cl_program sapProg = b3OpenCLUtils::compileCLProgramFromString(m_context,m_device,sapSrc,&errNum,"",B3_BROADPHASE_SAP_PATH);
-	b3Assert(errNum==CL_SUCCESS);
+	cl_program sapProg = b3OpenCLUtils::compileCLProgramFromString(m_context, m_device, sapSrc, &errNum, "", B3_BROADPHASE_SAP_PATH);
+	b3Assert(errNum == CL_SUCCESS);
 
-
-	b3Assert(errNum==CL_SUCCESS);
+	b3Assert(errNum == CL_SUCCESS);
 #ifndef __APPLE__
-	m_prefixScanFloat4 = new b3PrefixScanFloat4CL(m_context,m_device,m_queue);
+	m_prefixScanFloat4 = new b3PrefixScanFloat4CL(m_context, m_device, m_queue);
 #else
 	m_prefixScanFloat4 = 0;
 #endif
 	m_sapKernel = 0;
-	
+
 	switch (kernelType)
 	{
 		case B3_GPU_SAP_KERNEL_BRUTE_FORCE_CPU:
 		{
-			m_sapKernel=0;
+			m_sapKernel = 0;
 			break;
 		}
-		case 		B3_GPU_SAP_KERNEL_BRUTE_FORCE_GPU:
+		case B3_GPU_SAP_KERNEL_BRUTE_FORCE_GPU:
 		{
-			m_sapKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,sapSrc, "computePairsKernelBruteForce",&errNum,sapProg );
+			m_sapKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, sapSrc, "computePairsKernelBruteForce", &errNum, sapProg);
 			break;
 		}
 
 		case B3_GPU_SAP_KERNEL_ORIGINAL:
 		{
-			m_sapKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,sapSrc, "computePairsKernelOriginal",&errNum,sapProg );
+			m_sapKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, sapSrc, "computePairsKernelOriginal", &errNum, sapProg);
 			break;
 		}
 		case B3_GPU_SAP_KERNEL_BARRIER:
 		{
-			m_sapKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,sapSrc, "computePairsKernelBarrier",&errNum,sapProg );
+			m_sapKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, sapSrc, "computePairsKernelBarrier", &errNum, sapProg);
 			break;
 		}
 		case B3_GPU_SAP_KERNEL_LOCAL_SHARED_MEMORY:
 		{
-			m_sapKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,sapSrc, "computePairsKernelLocalSharedMemory",&errNum,sapProg );
+			m_sapKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, sapSrc, "computePairsKernelLocalSharedMemory", &errNum, sapProg);
 			break;
 		}
 
 		default:
 		{
-			m_sapKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,sapSrc, "computePairsKernelLocalSharedMemory",&errNum,sapProg );
+			m_sapKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, sapSrc, "computePairsKernelLocalSharedMemory", &errNum, sapProg);
 			b3Error("Unknown 3D GPU SAP provided, fallback to computePairsKernelLocalSharedMemory");
 		}
 	};
-	
-		
-	
-	m_sap2Kernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,sapSrc, "computePairsKernelTwoArrays",&errNum,sapProg );
-	b3Assert(errNum==CL_SUCCESS);
 
-	m_prepareSumVarianceKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,sapSrc, "prepareSumVarianceKernel",&errNum,sapProg );
-	b3Assert(errNum==CL_SUCCESS);
+	m_sap2Kernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, sapSrc, "computePairsKernelTwoArrays", &errNum, sapProg);
+	b3Assert(errNum == CL_SUCCESS);
 
-		
-	m_flipFloatKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,sapSrc, "flipFloatKernel",&errNum,sapProg );
+	m_prepareSumVarianceKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, sapSrc, "prepareSumVarianceKernel", &errNum, sapProg);
+	b3Assert(errNum == CL_SUCCESS);
 
-	m_copyAabbsKernel= b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,sapSrc, "copyAabbsKernel",&errNum,sapProg );
+	m_flipFloatKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, sapSrc, "flipFloatKernel", &errNum, sapProg);
 
-	m_scatterKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,sapSrc, "scatterKernel",&errNum,sapProg );
+	m_copyAabbsKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, sapSrc, "copyAabbsKernel", &errNum, sapProg);
 
-	m_sorter = new b3RadixSort32CL(m_context,m_device,m_queue);
+	m_scatterKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, sapSrc, "scatterKernel", &errNum, sapProg);
+
+	m_sorter = new b3RadixSort32CL(m_context, m_device, m_queue);
 }
 
 b3GpuSapBroadphase::~b3GpuSapBroadphase()
@@ -173,13 +167,11 @@ b3GpuSapBroadphase::~b3GpuSapBroadphase()
 	clReleaseKernel(m_sapKernel);
 	clReleaseKernel(m_sap2Kernel);
 	clReleaseKernel(m_prepareSumVarianceKernel);
-	
-
 }
 
 /// conservative test for overlap between two aabbs
-static bool TestAabbAgainstAabb2(const b3Vector3 &aabbMin1, const b3Vector3 &aabbMax1,
-								const b3Vector3 &aabbMin2, const b3Vector3 &aabbMax2)
+static bool TestAabbAgainstAabb2(const b3Vector3& aabbMin1, const b3Vector3& aabbMax1,
+								 const b3Vector3& aabbMin2, const b3Vector3& aabbMax2)
 {
 	bool overlap = true;
 	overlap = (aabbMin1.getX() > aabbMax2.getX() || aabbMax1.getX() < aabbMin2.getX()) ? false : overlap;
@@ -188,8 +180,6 @@ static bool TestAabbAgainstAabb2(const b3Vector3 &aabbMin1, const b3Vector3 &aab
 	return overlap;
 }
 
-
-
 //http://stereopsis.com/radix.html
 static unsigned int FloatFlip(float fl)
 {
@@ -198,79 +188,77 @@ static unsigned int FloatFlip(float fl)
 	return f ^ mask;
 };
 
-void  b3GpuSapBroadphase::init3dSap()
+void b3GpuSapBroadphase::init3dSap()
 {
-	if (m_currentBuffer<0)
+	if (m_currentBuffer < 0)
 	{
 		m_allAabbsGPU.copyToHost(m_allAabbsCPU);
 
 		m_currentBuffer = 0;
-		for (int axis=0;axis<3;axis++)
+		for (int axis = 0; axis < 3; axis++)
 		{
-			for (int buf=0;buf<2;buf++)
+			for (int buf = 0; buf < 2; buf++)
 			{
 				int totalNumAabbs = m_allAabbsCPU.size();
-				int numEndPoints = 2*totalNumAabbs;
+				int numEndPoints = 2 * totalNumAabbs;
 				m_sortedAxisCPU[axis][buf].resize(numEndPoints);
 
-				if (buf==m_currentBuffer)
+				if (buf == m_currentBuffer)
 				{
-					for (int i=0;i<totalNumAabbs;i++)
+					for (int i = 0; i < totalNumAabbs; i++)
 					{
-						m_sortedAxisCPU[axis][buf][i*2].m_key = FloatFlip(m_allAabbsCPU[i].m_min[axis])-1;
-						m_sortedAxisCPU[axis][buf][i*2].m_value = i*2;
-						m_sortedAxisCPU[axis][buf][i*2+1].m_key = FloatFlip(m_allAabbsCPU[i].m_max[axis])+1;
-						m_sortedAxisCPU[axis][buf][i*2+1].m_value = i*2+1;
+						m_sortedAxisCPU[axis][buf][i * 2].m_key = FloatFlip(m_allAabbsCPU[i].m_min[axis]) - 1;
+						m_sortedAxisCPU[axis][buf][i * 2].m_value = i * 2;
+						m_sortedAxisCPU[axis][buf][i * 2 + 1].m_key = FloatFlip(m_allAabbsCPU[i].m_max[axis]) + 1;
+						m_sortedAxisCPU[axis][buf][i * 2 + 1].m_value = i * 2 + 1;
 					}
 				}
 			}
 		}
 
-		for (int axis=0;axis<3;axis++)
+		for (int axis = 0; axis < 3; axis++)
 		{
 			m_sorter->executeHost(m_sortedAxisCPU[axis][m_currentBuffer]);
 		}
 
-		for (int axis=0;axis<3;axis++)
+		for (int axis = 0; axis < 3; axis++)
 		{
 			//int totalNumAabbs = m_allAabbsCPU.size();
 			int numEndPoints = m_sortedAxisCPU[axis][m_currentBuffer].size();
 			m_objectMinMaxIndexCPU[axis][m_currentBuffer].resize(numEndPoints);
-			for (int i=0;i<numEndPoints;i++)
+			for (int i = 0; i < numEndPoints; i++)
 			{
 				int destIndex = m_sortedAxisCPU[axis][m_currentBuffer][i].m_value;
-				int newDest = destIndex/2;
-				if (destIndex&1)
+				int newDest = destIndex / 2;
+				if (destIndex & 1)
 				{
-					m_objectMinMaxIndexCPU[axis][m_currentBuffer][newDest].y=i;
-				} else
+					m_objectMinMaxIndexCPU[axis][m_currentBuffer][newDest].y = i;
+				}
+				else
 				{
-					m_objectMinMaxIndexCPU[axis][m_currentBuffer][newDest].x=i;
+					m_objectMinMaxIndexCPU[axis][m_currentBuffer][newDest].x = i;
 				}
 			}
 		}
-
 	}
 }
 
-
 static bool b3PairCmp(const b3Int4& p, const b3Int4& q)
 {
-	return ((p.x<q.x) || ((p.x==q.x) && (p.y<q.y)));
+	return ((p.x < q.x) || ((p.x == q.x) && (p.y < q.y)));
 }
 
-
-static bool operator==(const b3Int4& a,const b3Int4& b)
+static bool operator==(const b3Int4& a, const b3Int4& b)
 {
 	return a.x == b.x && a.y == b.y;
 };
 
-static bool operator<(const b3Int4& a,const b3Int4& b)
+static bool operator<(const b3Int4& a, const b3Int4& b)
 {
 	return a.x < b.x || (a.x == b.x && a.y < b.y);
 };
 
-static bool operator>(const b3Int4& a,const b3Int4& b)
+static bool operator>(const b3Int4& a, const b3Int4& b)
 {
 	return a.x > b.x || (a.x == b.x && a.y > b.y);
 };
@@ -278,31 +266,29 @@ static bool operator>(const b3Int4& a,const b3Int4& b)
 b3AlignedObjectArray<b3Int4> addedHostPairs;
 b3AlignedObjectArray<b3Int4> removedHostPairs;
 
-b3AlignedObjectArray<b3SapAabb>	preAabbs;
+b3AlignedObjectArray<b3SapAabb> preAabbs;
 
-void  b3GpuSapBroadphase::calculateOverlappingPairsHostIncremental3Sap()
+void b3GpuSapBroadphase::calculateOverlappingPairsHostIncremental3Sap()
 {
 	//static int framepje = 0;
 	//printf("framepje=%d\n",framepje++);
-	
 
 	B3_PROFILE("calculateOverlappingPairsHostIncremental3Sap");
 
 	addedHostPairs.resize(0);
 	removedHostPairs.resize(0);
 
-	b3Assert(m_currentBuffer>=0);
-	
+	b3Assert(m_currentBuffer >= 0);
+
 	{
 		preAabbs.resize(m_allAabbsCPU.size());
-		for (int i=0;i<preAabbs.size();i++)
+		for (int i = 0; i < preAabbs.size(); i++)
 		{
-			preAabbs[i]=m_allAabbsCPU[i];
+			preAabbs[i] = m_allAabbsCPU[i];
 		}
 	}
-	
 
-	if (m_currentBuffer<0)
+	if (m_currentBuffer < 0)
 		return;
 	{
 		B3_PROFILE("m_allAabbsGPU.copyToHost");
@@ -316,100 +302,87 @@ void  b3GpuSapBroadphase::calculateOverlappingPairsHostIncremental3Sap()
 	}
 	if (0)
 	{
-	{
-		printf("ab[40].min=%f,%f,%f,ab[40].max=%f,%f,%f\n",
-		m_allAabbsCPU[40].m_min[0],	m_allAabbsCPU[40].m_min[1],m_allAabbsCPU[40].m_min[2],
-		m_allAabbsCPU[40].m_max[0],	m_allAabbsCPU[40].m_max[1],m_allAabbsCPU[40].m_max[2]);
-	}
-
-	{
-		printf("ab[53].min=%f,%f,%f,ab[53].max=%f,%f,%f\n",
-		m_allAabbsCPU[53].m_min[0],	m_allAabbsCPU[53].m_min[1],m_allAabbsCPU[53].m_min[2],
-		m_allAabbsCPU[53].m_max[0],	m_allAabbsCPU[53].m_max[1],m_allAabbsCPU[53].m_max[2]);
-	}
-	
-
-	{
-	b3Int4 newPair;
-	newPair.x = 40;
-	newPair.y = 53;
-		int index = allPairs.findBinarySearch(newPair);
-		printf("hasPair(40,53)=%d out of %d\n",index, allPairs.size());
-
 		{
-			int overlap = TestAabbAgainstAabb2((const b3Vector3&)m_allAabbsCPU[40].m_min, (const b3Vector3&)m_allAabbsCPU[40].m_max,(const b3Vector3&)m_allAabbsCPU[53].m_min,(const b3Vector3&)m_allAabbsCPU[53].m_max);
-			printf("overlap=%d\n",overlap);
+			printf("ab[40].min=%f,%f,%f,ab[40].max=%f,%f,%f\n",
+				   m_allAabbsCPU[40].m_min[0], m_allAabbsCPU[40].m_min[1], m_allAabbsCPU[40].m_min[2],
+				   m_allAabbsCPU[40].m_max[0], m_allAabbsCPU[40].m_max[1], m_allAabbsCPU[40].m_max[2]);
 		}
 
-		if (preAabbs.size())
 		{
-			int prevOverlap = TestAabbAgainstAabb2((const b3Vector3&)preAabbs[40].m_min, (const b3Vector3&)preAabbs[40].m_max,(const b3Vector3&)preAabbs[53].m_min,(const b3Vector3&)preAabbs[53].m_max);
-			printf("prevoverlap=%d\n",prevOverlap);
-		} else
-		{
-			printf("unknown prevoverlap\n");
+			printf("ab[53].min=%f,%f,%f,ab[53].max=%f,%f,%f\n",
+				   m_allAabbsCPU[53].m_min[0], m_allAabbsCPU[53].m_min[1], m_allAabbsCPU[53].m_min[2],
+				   m_allAabbsCPU[53].m_max[0], m_allAabbsCPU[53].m_max[1], m_allAabbsCPU[53].m_max[2]);
 		}
 
-	}
-	}
-
-
-	if (0)
-	{
-		for (int i=0;i<m_allAabbsCPU.size();i++)
 		{
-			//printf("aabb[%d] min=%f,%f,%f max=%f,%f,%f\n",i,m_allAabbsCPU[i].m_min[0],m_allAabbsCPU[i].m_min[1],m_allAabbsCPU[i].m_min[2],			m_allAabbsCPU[i].m_max[0],m_allAabbsCPU[i].m_max[1],m_allAabbsCPU[i].m_max[2]);
+			b3Int4 newPair;
+			newPair.x = 40;
+			newPair.y = 53;
+			int index = allPairs.findBinarySearch(newPair);
+			printf("hasPair(40,53)=%d out of %d\n", index, allPairs.size());
 
-
-		}
-
-		for (int axis=0;axis<3;axis++)
-		{
-			for (int buf=0;buf<2;buf++)
 			{
-				b3Assert(m_sortedAxisCPU[axis][buf].size() == m_allAabbsCPU.size()*2);
+				int overlap = TestAabbAgainstAabb2((const b3Vector3&)m_allAabbsCPU[40].m_min, (const b3Vector3&)m_allAabbsCPU[40].m_max, (const b3Vector3&)m_allAabbsCPU[53].m_min, (const b3Vector3&)m_allAabbsCPU[53].m_max);
+				printf("overlap=%d\n", overlap);
+			}
+
+			if (preAabbs.size())
+			{
+				int prevOverlap = TestAabbAgainstAabb2((const b3Vector3&)preAabbs[40].m_min, (const b3Vector3&)preAabbs[40].m_max, (const b3Vector3&)preAabbs[53].m_min, (const b3Vector3&)preAabbs[53].m_max);
+				printf("prevoverlap=%d\n", prevOverlap);
+			}
+			else
+			{
+				printf("unknown prevoverlap\n");
 			}
 		}
 	}
 
+	if (0)
+	{
+		for (int i = 0; i < m_allAabbsCPU.size(); i++)
+		{
+			//printf("aabb[%d] min=%f,%f,%f max=%f,%f,%f\n",i,m_allAabbsCPU[i].m_min[0],m_allAabbsCPU[i].m_min[1],m_allAabbsCPU[i].m_min[2],			m_allAabbsCPU[i].m_max[0],m_allAabbsCPU[i].m_max[1],m_allAabbsCPU[i].m_max[2]);
+		}
 
+		for (int axis = 0; axis < 3; axis++)
+		{
+			for (int buf = 0; buf < 2; buf++)
+			{
+				b3Assert(m_sortedAxisCPU[axis][buf].size() == m_allAabbsCPU.size() * 2);
+			}
+		}
+	}
 
-	m_currentBuffer = 1-m_currentBuffer;
-	
-
+	m_currentBuffer = 1 - m_currentBuffer;
 
 	int totalNumAabbs = m_allAabbsCPU.size();
 
 	{
 		B3_PROFILE("assign m_sortedAxisCPU(FloatFlip)");
-		for (int i=0;i<totalNumAabbs;i++)
+		for (int i = 0; i < totalNumAabbs; i++)
 		{
-		
-
 			unsigned int keyMin[3];
 			unsigned int keyMax[3];
-			for (int axis=0;axis<3;axis++)
+			for (int axis = 0; axis < 3; axis++)
 			{
-				float vmin=m_allAabbsCPU[i].m_min[axis];
+				float vmin = m_allAabbsCPU[i].m_min[axis];
 				float vmax = m_allAabbsCPU[i].m_max[axis];
 				keyMin[axis] = FloatFlip(vmin);
 				keyMax[axis] = FloatFlip(vmax);
-			
-				m_sortedAxisCPU[axis][m_currentBuffer][i*2].m_key = keyMin[axis]-1;
-				m_sortedAxisCPU[axis][m_currentBuffer][i*2].m_value = i*2;
-				m_sortedAxisCPU[axis][m_currentBuffer][i*2+1].m_key = keyMax[axis]+1;
-				m_sortedAxisCPU[axis][m_currentBuffer][i*2+1].m_value = i*2+1;
+
+				m_sortedAxisCPU[axis][m_currentBuffer][i * 2].m_key = keyMin[axis] - 1;
+				m_sortedAxisCPU[axis][m_currentBuffer][i * 2].m_value = i * 2;
+				m_sortedAxisCPU[axis][m_currentBuffer][i * 2 + 1].m_key = keyMax[axis] + 1;
+				m_sortedAxisCPU[axis][m_currentBuffer][i * 2 + 1].m_value = i * 2 + 1;
 			}
 			//printf("aabb[%d] min=%u,%u,%u max %u,%u,%u\n", i,keyMin[0],keyMin[1],keyMin[2],keyMax[0],keyMax[1],keyMax[2]);
-
 		}
 	}
-	
-	
 
 	{
 		B3_PROFILE("sort m_sortedAxisCPU");
-		for (int axis=0;axis<3;axis++)
+		for (int axis = 0; axis < 3; axis++)
 			m_sorter->executeHost(m_sortedAxisCPU[axis][m_currentBuffer]);
 	}
 
@@ -432,21 +405,22 @@ void  b3GpuSapBroadphase::calculateOverlappingPairsHostIncremental3Sap()
 
 	{
 		B3_PROFILE("assign m_objectMinMaxIndexCPU");
-		for (int axis=0;axis<3;axis++)
+		for (int axis = 0; axis < 3; axis++)
 		{
 			int totalNumAabbs = m_allAabbsCPU.size();
 			int numEndPoints = m_sortedAxisCPU[axis][m_currentBuffer].size();
 			m_objectMinMaxIndexCPU[axis][m_currentBuffer].resize(totalNumAabbs);
-			for (int i=0;i<numEndPoints;i++)
+			for (int i = 0; i < numEndPoints; i++)
 			{
 				int destIndex = m_sortedAxisCPU[axis][m_currentBuffer][i].m_value;
-				int newDest = destIndex/2;
-				if (destIndex&1)
+				int newDest = destIndex / 2;
+				if (destIndex & 1)
 				{
-					m_objectMinMaxIndexCPU[axis][m_currentBuffer][newDest].y=i;
-				} else
+					m_objectMinMaxIndexCPU[axis][m_currentBuffer][newDest].y = i;
+				}
+				else
 				{
-					m_objectMinMaxIndexCPU[axis][m_currentBuffer][newDest].x=i;
+					m_objectMinMaxIndexCPU[axis][m_currentBuffer][newDest].x = i;
 				}
 			}
 		}
@@ -485,12 +459,11 @@ void  b3GpuSapBroadphase::calculateOverlappingPairsHostIncremental3Sap()
 	}
 #endif
 
-
 	int a = m_objectMinMaxIndexCPU[0][m_currentBuffer].size();
 	int b = m_objectMinMaxIndexCPU[1][m_currentBuffer].size();
 	int c = m_objectMinMaxIndexCPU[2][m_currentBuffer].size();
-	b3Assert(a==b);
-	b3Assert(b==c);
+	b3Assert(a == b);
+	b3Assert(b == c);
 	/*
 	if (searchIncremental3dSapOnGpu)
 	{
@@ -574,175 +547,170 @@ void  b3GpuSapBroadphase::calculateOverlappingPairsHostIncremental3Sap()
 		int numObjects = m_objectMinMaxIndexCPU[0][m_currentBuffer].size();
 
 		B3_PROFILE("actual search");
-		for (int i=0;i<numObjects;i++)
+		for (int i = 0; i < numObjects; i++)
 		{
 			//int numObjects = m_objectMinMaxIndexCPU[axis][m_currentBuffer].size();
 			//int checkObjects[]={40,53};
 			//int numCheckObjects = sizeof(checkObjects)/sizeof(int);
-			
+
 			//for (int a=0;a<numCheckObjects ;a++)
-			
-			for (int axis=0;axis<3;axis++)
+
+			for (int axis = 0; axis < 3; axis++)
 			{
 				//int i = checkObjects[a];
 
 				unsigned int curMinIndex = m_objectMinMaxIndexCPU[axis][m_currentBuffer][i].x;
 				unsigned int curMaxIndex = m_objectMinMaxIndexCPU[axis][m_currentBuffer][i].y;
-				unsigned int prevMinIndex = m_objectMinMaxIndexCPU[axis][1-m_currentBuffer][i].x;
+				unsigned int prevMinIndex = m_objectMinMaxIndexCPU[axis][1 - m_currentBuffer][i].x;
 				int dmin = curMinIndex - prevMinIndex;
-				
-				unsigned int prevMaxIndex = m_objectMinMaxIndexCPU[axis][1-m_currentBuffer][i].y;
 
-				
+				unsigned int prevMaxIndex = m_objectMinMaxIndexCPU[axis][1 - m_currentBuffer][i].y;
 
 				int dmax = curMaxIndex - prevMaxIndex;
-				if (dmin!=0)
+				if (dmin != 0)
 				{
 					//printf("for object %d, dmin=%d\n",i,dmin);
 				}
-				if (dmax!=0)
+				if (dmax != 0)
 				{
 					//printf("for object %d, dmax=%d\n",i,dmax);
 				}
-				for (int otherbuffer = 0;otherbuffer<2;otherbuffer++)
+				for (int otherbuffer = 0; otherbuffer < 2; otherbuffer++)
 				{
-					if (dmin!=0)
+					if (dmin != 0)
 					{
-						int stepMin = dmin<0 ? -1 : 1;
-						for (int j=prevMinIndex;j!=curMinIndex;j+=stepMin)
+						int stepMin = dmin < 0 ? -1 : 1;
+						for (int j = prevMinIndex; j != curMinIndex; j += stepMin)
 						{
 							int otherIndex2 = m_sortedAxisCPU[axis][otherbuffer][j].y;
-							int otherIndex = otherIndex2/2;
-							if (otherIndex!=i)
+							int otherIndex = otherIndex2 / 2;
+							if (otherIndex != i)
 							{
-								bool otherIsMax = ((otherIndex2&1)!=0);
+								bool otherIsMax = ((otherIndex2 & 1) != 0);
 
 								if (otherIsMax)
 								{
 									//bool overlap = TestAabbAgainstAabb2((const b3Vector3&)m_allAabbsCPU[i].m_min, (const b3Vector3&)m_allAabbsCPU[i].m_max,(const b3Vector3&)m_allAabbsCPU[otherIndex].m_min,(const b3Vector3&)m_allAabbsCPU[otherIndex].m_max);
 									//bool prevOverlap = TestAabbAgainstAabb2((const b3Vector3&)preAabbs[i].m_min, (const b3Vector3&)preAabbs[i].m_max,(const b3Vector3&)preAabbs[otherIndex].m_min,(const b3Vector3&)preAabbs[otherIndex].m_max);
-									
+
 									bool overlap = true;
 
-									for (int ax=0;ax<3;ax++)
+									for (int ax = 0; ax < 3; ax++)
 									{
 										if ((m_objectMinMaxIndexCPU[ax][m_currentBuffer][i].x > m_objectMinMaxIndexCPU[ax][m_currentBuffer][otherIndex].y) ||
 											(m_objectMinMaxIndexCPU[ax][m_currentBuffer][i].y < m_objectMinMaxIndexCPU[ax][m_currentBuffer][otherIndex].x))
-											overlap=false;
+											overlap = false;
 									}
 
-								//	b3Assert(overlap2==overlap);
+									//	b3Assert(overlap2==overlap);
 
 									bool prevOverlap = true;
 
-									for (int ax=0;ax<3;ax++)
+									for (int ax = 0; ax < 3; ax++)
 									{
-										if ((m_objectMinMaxIndexCPU[ax][1-m_currentBuffer][i].x > m_objectMinMaxIndexCPU[ax][1-m_currentBuffer][otherIndex].y) ||
-											(m_objectMinMaxIndexCPU[ax][1-m_currentBuffer][i].y < m_objectMinMaxIndexCPU[ax][1-m_currentBuffer][otherIndex].x))
-											prevOverlap=false;
+										if ((m_objectMinMaxIndexCPU[ax][1 - m_currentBuffer][i].x > m_objectMinMaxIndexCPU[ax][1 - m_currentBuffer][otherIndex].y) ||
+											(m_objectMinMaxIndexCPU[ax][1 - m_currentBuffer][i].y < m_objectMinMaxIndexCPU[ax][1 - m_currentBuffer][otherIndex].x))
+											prevOverlap = false;
 									}
-									
 
 									//b3Assert(overlap==overlap2);
-								
 
-
-									if (dmin<0)
+									if (dmin < 0)
 									{
 										if (overlap && !prevOverlap)
 										{
 											//add a pair
 											b3Int4 newPair;
-											if (i<=otherIndex)
+											if (i <= otherIndex)
 											{
 												newPair.x = i;
 												newPair.y = otherIndex;
-											} else
+											}
+											else
 											{
 												newPair.x = otherIndex;
 												newPair.y = i;
 											}
 											addedHostPairs.push_back(newPair);
 										}
-									} 
+									}
 									else
 									{
 										if (!overlap && prevOverlap)
 										{
-									
 											//remove a pair
 											b3Int4 removedPair;
-											if (i<=otherIndex)
+											if (i <= otherIndex)
 											{
 												removedPair.x = i;
 												removedPair.y = otherIndex;
-											} else
+											}
+											else
 											{
 												removedPair.x = otherIndex;
 												removedPair.y = i;
 											}
 											removedHostPairs.push_back(removedPair);
 										}
-									}//otherisMax
-								}//if (dmin<0)
-							}//if (otherIndex!=i)
-						}//for (int j=
+									}  //otherisMax
+								}      //if (dmin<0)
+							}          //if (otherIndex!=i)
+						}              //for (int j=
 					}
-				
-					if (dmax!=0)
+
+					if (dmax != 0)
 					{
-						int stepMax = dmax<0 ? -1 : 1;
-						for (int j=prevMaxIndex;j!=curMaxIndex;j+=stepMax)
+						int stepMax = dmax < 0 ? -1 : 1;
+						for (int j = prevMaxIndex; j != curMaxIndex; j += stepMax)
 						{
 							int otherIndex2 = m_sortedAxisCPU[axis][otherbuffer][j].y;
-							int otherIndex = otherIndex2/2;
-							if (otherIndex!=i)
+							int otherIndex = otherIndex2 / 2;
+							if (otherIndex != i)
 							{
 								//bool otherIsMin = ((otherIndex2&1)==0);
 								//if (otherIsMin)
 								{
 									//bool overlap = TestAabbAgainstAabb2((const b3Vector3&)m_allAabbsCPU[i].m_min, (const b3Vector3&)m_allAabbsCPU[i].m_max,(const b3Vector3&)m_allAabbsCPU[otherIndex].m_min,(const b3Vector3&)m_allAabbsCPU[otherIndex].m_max);
 									//bool prevOverlap = TestAabbAgainstAabb2((const b3Vector3&)preAabbs[i].m_min, (const b3Vector3&)preAabbs[i].m_max,(const b3Vector3&)preAabbs[otherIndex].m_min,(const b3Vector3&)preAabbs[otherIndex].m_max);
-									
+
 									bool overlap = true;
 
-									for (int ax=0;ax<3;ax++)
+									for (int ax = 0; ax < 3; ax++)
 									{
 										if ((m_objectMinMaxIndexCPU[ax][m_currentBuffer][i].x > m_objectMinMaxIndexCPU[ax][m_currentBuffer][otherIndex].y) ||
 											(m_objectMinMaxIndexCPU[ax][m_currentBuffer][i].y < m_objectMinMaxIndexCPU[ax][m_currentBuffer][otherIndex].x))
-											overlap=false;
+											overlap = false;
 									}
 									//b3Assert(overlap2==overlap);
 
 									bool prevOverlap = true;
 
-									for (int ax=0;ax<3;ax++)
+									for (int ax = 0; ax < 3; ax++)
 									{
-										if ((m_objectMinMaxIndexCPU[ax][1-m_currentBuffer][i].x > m_objectMinMaxIndexCPU[ax][1-m_currentBuffer][otherIndex].y) ||
-											(m_objectMinMaxIndexCPU[ax][1-m_currentBuffer][i].y < m_objectMinMaxIndexCPU[ax][1-m_currentBuffer][otherIndex].x))
-											prevOverlap=false;
+										if ((m_objectMinMaxIndexCPU[ax][1 - m_currentBuffer][i].x > m_objectMinMaxIndexCPU[ax][1 - m_currentBuffer][otherIndex].y) ||
+											(m_objectMinMaxIndexCPU[ax][1 - m_currentBuffer][i].y < m_objectMinMaxIndexCPU[ax][1 - m_currentBuffer][otherIndex].x))
+											prevOverlap = false;
 									}
-									
 
-									if (dmax>0)
+									if (dmax > 0)
 									{
 										if (overlap && !prevOverlap)
 										{
 											//add a pair
 											b3Int4 newPair;
-											if (i<=otherIndex)
+											if (i <= otherIndex)
 											{
 												newPair.x = i;
 												newPair.y = otherIndex;
-											} else
+											}
+											else
 											{
 												newPair.x = otherIndex;
 												newPair.y = i;
 											}
 											addedHostPairs.push_back(newPair);
-							
 										}
-									} 
+									}
 									else
 									{
 										if (!overlap && prevOverlap)
@@ -750,33 +718,31 @@ void  b3GpuSapBroadphase::calculateOverlappingPairsHostIncremental3Sap()
 											//if (otherIndex2&1==0) -> min?
 											//remove a pair
 											b3Int4 removedPair;
-											if (i<=otherIndex)
+											if (i <= otherIndex)
 											{
 												removedPair.x = i;
 												removedPair.y = otherIndex;
-											} else
+											}
+											else
 											{
 												removedPair.x = otherIndex;
 												removedPair.y = i;
 											}
 											removedHostPairs.push_back(removedPair);
-								
 										}
 									}
-							
-								}//if (dmin<0)
-							}//if (otherIndex!=i)
-						}//for (int j=
+
+								}  //if (dmin<0)
+							}      //if (otherIndex!=i)
+						}          //for (int j=
 					}
-				}//for (int otherbuffer
-			}//for (int axis=0;
-		}//for (int i=0;i<numObjects
+				}  //for (int otherbuffer
+			}      //for (int axis=0;
+		}          //for (int i=0;i<numObjects
 	}
 
 	//remove duplicates and add/remove then to existing m_overlappingPairs
-	
-	
-	
+
 	{
 		{
 			B3_PROFILE("sort allPairs");
@@ -795,31 +761,28 @@ void  b3GpuSapBroadphase::calculateOverlappingPairsHostIncremental3Sap()
 	b3Int4 prevPair;
 	prevPair.x = -1;
 	prevPair.y = -1;
-	
+
 	int uniqueRemovedPairs = 0;
 
 	b3AlignedObjectArray<int> removedPositions;
 
 	{
 		B3_PROFILE("actual removing");
-		for (int i=0;i<removedHostPairs.size();i++)
+		for (int i = 0; i < removedHostPairs.size(); i++)
 		{
 			b3Int4 removedPair = removedHostPairs[i];
 			if ((removedPair.x != prevPair.x) || (removedPair.y != prevPair.y))
 			{
+				int index1 = allPairs.findBinarySearch(removedPair);
 
-			int index1 = allPairs.findBinarySearch(removedPair);
+				//#ifdef _DEBUG
 
-	//#ifdef _DEBUG
-				
-				
-	
 				int index2 = allPairs.findLinearSearch(removedPair);
-				b3Assert(index1==index2);
-	
+				b3Assert(index1 == index2);
+
 				//b3Assert(index1!=allPairs.size());
-				if (index1<allPairs.size())
-	//#endif//_DEBUG
+				if (index1 < allPairs.size())
+				//#endif//_DEBUG
 				{
 					uniqueRemovedPairs++;
 					removedPositions.push_back(index1);
@@ -833,13 +796,13 @@ void  b3GpuSapBroadphase::calculateOverlappingPairsHostIncremental3Sap()
 
 		if (uniqueRemovedPairs)
 		{
-			for (int i=0;i<removedPositions.size();i++)
+			for (int i = 0; i < removedPositions.size(); i++)
 			{
-				allPairs[removedPositions[i]].x = INT_MAX ;
-				allPairs[removedPositions[i]].y = INT_MAX ;
+				allPairs[removedPositions[i]].x = INT_MAX;
+				allPairs[removedPositions[i]].y = INT_MAX;
 			}
 			allPairs.quickSort(b3PairCmp);
-			allPairs.resize(allPairs.size()-uniqueRemovedPairs);
+			allPairs.resize(allPairs.size() - uniqueRemovedPairs);
 		}
 	}
 	//if (uniqueRemovedPairs)
@@ -848,33 +811,31 @@ void  b3GpuSapBroadphase::calculateOverlappingPairsHostIncremental3Sap()
 
 	prevPair.x = -1;
 	prevPair.y = -1;
-	
-	int uniqueAddedPairs=0;
+
+	int uniqueAddedPairs = 0;
 	b3AlignedObjectArray<b3Int4> actualAddedPairs;
 
 	{
 		B3_PROFILE("actual adding");
-		for (int i=0;i<addedHostPairs.size();i++)
+		for (int i = 0; i < addedHostPairs.size(); i++)
 		{
 			b3Int4 newPair = addedHostPairs[i];
 			if ((newPair.x != prevPair.x) || (newPair.y != prevPair.y))
 			{
-//#ifdef _DEBUG		
+				//#ifdef _DEBUG
 				int index1 = allPairs.findBinarySearch(newPair);
-				
-	
-				int index2 = allPairs.findLinearSearch(newPair);
-				b3Assert(index1==index2);
-	
 
-				b3Assert(index1==allPairs.size());
-				if (index1!=allPairs.size())
+				int index2 = allPairs.findLinearSearch(newPair);
+				b3Assert(index1 == index2);
+
+				b3Assert(index1 == allPairs.size());
+				if (index1 != allPairs.size())
 				{
 					printf("??\n");
 				}
 
-				if (index1==allPairs.size())
-//#endif //_DEBUG
+				if (index1 == allPairs.size())
+				//#endif //_DEBUG
 				{
 					uniqueAddedPairs++;
 					actualAddedPairs.push_back(newPair);
@@ -882,94 +843,83 @@ void  b3GpuSapBroadphase::calculateOverlappingPairsHostIncremental3Sap()
 			}
 			prevPair = newPair;
 		}
-		for (int i=0;i<actualAddedPairs.size();i++)
+		for (int i = 0; i < actualAddedPairs.size(); i++)
 		{
 			//printf("framepje (%d), new pair(%d):%d,%d\n",framepje,i,actualAddedPairs[i].x,actualAddedPairs[i].y);
 			allPairs.push_back(actualAddedPairs[i]);
 		}
 	}
-	
+
 	//if (uniqueAddedPairs)
 	//	printf("uniqueAddedPairs=%d\n", uniqueAddedPairs);
 
-
 	{
 		B3_PROFILE("m_overlappingPairs.copyFromHost");
 		m_overlappingPairs.copyFromHost(allPairs);
 	}
-
-	
 }
 
-
-
-
-void  b3GpuSapBroadphase::calculateOverlappingPairsHost(int maxPairs)
+void b3GpuSapBroadphase::calculateOverlappingPairsHost(int maxPairs)
 {
 	//test
-//	if (m_currentBuffer>=0)
+	//	if (m_currentBuffer>=0)
 	//	return calculateOverlappingPairsHostIncremental3Sap();
 
 	b3Assert(m_allAabbsCPU.size() == m_allAabbsGPU.size());
 	m_allAabbsGPU.copyToHost(m_allAabbsCPU);
 
-
-	
-	int axis=0;
+	int axis = 0;
 	{
 		B3_PROFILE("CPU compute best variance axis");
-		b3Vector3 s=b3MakeVector3(0,0,0),s2=b3MakeVector3(0,0,0);
+		b3Vector3 s = b3MakeVector3(0, 0, 0), s2 = b3MakeVector3(0, 0, 0);
 		int numRigidBodies = m_smallAabbsMappingCPU.size();
 
-		for(int i=0;i<numRigidBodies;i++) 
+		for (int i = 0; i < numRigidBodies; i++)
 		{
 			b3SapAabb aabb = this->m_allAabbsCPU[m_smallAabbsMappingCPU[i]];
 
-			b3Vector3 maxAabb=b3MakeVector3(aabb.m_max[0],aabb.m_max[1],aabb.m_max[2]);
-			b3Vector3 minAabb=b3MakeVector3(aabb.m_min[0],aabb.m_min[1],aabb.m_min[2]);
-			b3Vector3 centerAabb=(maxAabb+minAabb)*0.5f;
-						
+			b3Vector3 maxAabb = b3MakeVector3(aabb.m_max[0], aabb.m_max[1], aabb.m_max[2]);
+			b3Vector3 minAabb = b3MakeVector3(aabb.m_min[0], aabb.m_min[1], aabb.m_min[2]);
+			b3Vector3 centerAabb = (maxAabb + minAabb) * 0.5f;
+
 			s += centerAabb;
-			s2 += centerAabb*centerAabb;
+			s2 += centerAabb * centerAabb;
 		}
-		b3Vector3 v = s2 - (s*s) / (float)numRigidBodies;
-		
-		if(v[1] > v[0]) 
+		b3Vector3 v = s2 - (s * s) / (float)numRigidBodies;
+
+		if (v[1] > v[0])
 			axis = 1;
-		if(v[2] > v[axis]) 
+		if (v[2] > v[axis])
 			axis = 2;
 	}
 
-
-	
-
 	b3AlignedObjectArray<b3Int4> hostPairs;
 
 	{
 		int numSmallAabbs = m_smallAabbsMappingCPU.size();
-		for (int i=0;i<numSmallAabbs;i++)
+		for (int i = 0; i < numSmallAabbs; i++)
 		{
 			b3SapAabb smallAabbi = m_allAabbsCPU[m_smallAabbsMappingCPU[i]];
 			//float reference = smallAabbi.m_max[axis];
 
-			for (int j=i+1;j<numSmallAabbs;j++)
+			for (int j = i + 1; j < numSmallAabbs; j++)
 			{
-
 				b3SapAabb smallAabbj = m_allAabbsCPU[m_smallAabbsMappingCPU[j]];
 
 				if (TestAabbAgainstAabb2((b3Vector3&)smallAabbi.m_min, (b3Vector3&)smallAabbi.m_max,
-					(b3Vector3&)smallAabbj.m_min,(b3Vector3&)smallAabbj.m_max))
+										 (b3Vector3&)smallAabbj.m_min, (b3Vector3&)smallAabbj.m_max))
 				{
 					b3Int4 pair;
 					int a = smallAabbi.m_minIndices[3];
 					int b = smallAabbj.m_minIndices[3];
-					if (a<=b)
+					if (a <= b)
 					{
-						pair.x = a;//store the original index in the unsorted aabb array
+						pair.x = a;  //store the original index in the unsorted aabb array
 						pair.y = b;
-					} else
+					}
+					else
 					{
-						pair.x = b;//store the original index in the unsorted aabb array
+						pair.x = b;  //store the original index in the unsorted aabb array
 						pair.y = a;
 					}
 					hostPairs.push_back(pair);
@@ -978,35 +928,35 @@ void  b3GpuSapBroadphase::calculateOverlappingPairsHost(int maxPairs)
 		}
 	}
 
-	
 	{
 		int numSmallAabbs = m_smallAabbsMappingCPU.size();
-		for (int i=0;i<numSmallAabbs;i++)
+		for (int i = 0; i < numSmallAabbs; i++)
 		{
 			b3SapAabb smallAabbi = m_allAabbsCPU[m_smallAabbsMappingCPU[i]];
 
 			//float reference = smallAabbi.m_max[axis];
 			int numLargeAabbs = m_largeAabbsMappingCPU.size();
 
-			for (int j=0;j<numLargeAabbs;j++)
+			for (int j = 0; j < numLargeAabbs; j++)
 			{
 				b3SapAabb largeAabbj = m_allAabbsCPU[m_largeAabbsMappingCPU[j]];
 				if (TestAabbAgainstAabb2((b3Vector3&)smallAabbi.m_min, (b3Vector3&)smallAabbi.m_max,
-					(b3Vector3&)largeAabbj.m_min,(b3Vector3&)largeAabbj.m_max))
+										 (b3Vector3&)largeAabbj.m_min, (b3Vector3&)largeAabbj.m_max))
 				{
 					b3Int4 pair;
 					int a = largeAabbj.m_minIndices[3];
 					int b = smallAabbi.m_minIndices[3];
-					if (a<=b)
+					if (a <= b)
 					{
-						pair.x = a; 
-						pair.y = b;//store the original index in the unsorted aabb array
-					} else
+						pair.x = a;
+						pair.y = b;  //store the original index in the unsorted aabb array
+					}
+					else
 					{
 						pair.x = b;
-						pair.y = a;//store the original index in the unsorted aabb array
+						pair.y = a;  //store the original index in the unsorted aabb array
 					}
-					
+
 					hostPairs.push_back(pair);
 				}
 			}
@@ -1021,21 +971,20 @@ void  b3GpuSapBroadphase::calculateOverlappingPairsHost(int maxPairs)
 	if (hostPairs.size())
 	{
 		m_overlappingPairs.copyFromHost(hostPairs);
-	} else
+	}
+	else
 	{
 		m_overlappingPairs.resize(0);
 	}
 
 	//init3dSap();
-
 }
 
-void  b3GpuSapBroadphase::reset()
+void b3GpuSapBroadphase::reset()
 {
 	m_allAabbsGPU.resize(0);
 	m_allAabbsCPU.resize(0);
 
-
 	m_smallAabbsMappingGPU.resize(0);
 	m_smallAabbsMappingCPU.resize(0);
 
@@ -1043,13 +992,11 @@ void  b3GpuSapBroadphase::reset()
 
 	m_largeAabbsMappingGPU.resize(0);
 	m_largeAabbsMappingCPU.resize(0);
-
 }
 
-
-void  b3GpuSapBroadphase::calculateOverlappingPairs(int maxPairs)
+void b3GpuSapBroadphase::calculateOverlappingPairs(int maxPairs)
 {
-	if (m_sapKernel==0)
+	if (m_sapKernel == 0)
 	{
 		calculateOverlappingPairsHost(maxPairs);
 		return;
@@ -1065,68 +1012,62 @@ void  b3GpuSapBroadphase::calculateOverlappingPairs(int maxPairs)
 	int axis = 0;
 
 	{
+		//bool syncOnHost = false;
 
-	//bool syncOnHost = false;
-
-	int numSmallAabbs = m_smallAabbsMappingCPU.size();
-	if (m_prefixScanFloat4 && numSmallAabbs)
-	{
-		B3_PROFILE("GPU compute best variance axis");
-		
-		if (m_dst.size()!=(numSmallAabbs+1))
+		int numSmallAabbs = m_smallAabbsMappingCPU.size();
+		if (m_prefixScanFloat4 && numSmallAabbs)
 		{
-			m_dst.resize(numSmallAabbs+128);
-			m_sum.resize(numSmallAabbs+128);
-			m_sum2.resize(numSmallAabbs+128);
-			m_sum.at(numSmallAabbs)=b3MakeVector3(0,0,0); //slow?
-			m_sum2.at(numSmallAabbs)=b3MakeVector3(0,0,0); //slow?
+			B3_PROFILE("GPU compute best variance axis");
+
+			if (m_dst.size() != (numSmallAabbs + 1))
+			{
+				m_dst.resize(numSmallAabbs + 128);
+				m_sum.resize(numSmallAabbs + 128);
+				m_sum2.resize(numSmallAabbs + 128);
+				m_sum.at(numSmallAabbs) = b3MakeVector3(0, 0, 0);   //slow?
+				m_sum2.at(numSmallAabbs) = b3MakeVector3(0, 0, 0);  //slow?
+			}
+
+			b3LauncherCL launcher(m_queue, m_prepareSumVarianceKernel, "m_prepareSumVarianceKernel");
+			launcher.setBuffer(m_allAabbsGPU.getBufferCL());
+
+			launcher.setBuffer(m_smallAabbsMappingGPU.getBufferCL());
+			launcher.setBuffer(m_sum.getBufferCL());
+			launcher.setBuffer(m_sum2.getBufferCL());
+			launcher.setConst(numSmallAabbs);
+			int num = numSmallAabbs;
+			launcher.launch1D(num);
+
+			b3Vector3 s;
+			b3Vector3 s2;
+			m_prefixScanFloat4->execute(m_sum, m_dst, numSmallAabbs + 1, &s);
+			m_prefixScanFloat4->execute(m_sum2, m_dst, numSmallAabbs + 1, &s2);
+
+			b3Vector3 v = s2 - (s * s) / (float)numSmallAabbs;
+
+			if (v[1] > v[0])
+				axis = 1;
+			if (v[2] > v[axis])
+				axis = 2;
 		}
 
-		b3LauncherCL launcher(m_queue, m_prepareSumVarianceKernel ,"m_prepareSumVarianceKernel");
-		launcher.setBuffer(m_allAabbsGPU.getBufferCL());
-		
-		launcher.setBuffer(m_smallAabbsMappingGPU.getBufferCL());
-		launcher.setBuffer(m_sum.getBufferCL());
-		launcher.setBuffer(m_sum2.getBufferCL());
-		launcher.setConst( numSmallAabbs  );
-		int num = numSmallAabbs;
-		launcher.launch1D( num);
-		
-
-		b3Vector3 s;
-		b3Vector3 s2;
-		m_prefixScanFloat4->execute(m_sum,m_dst,numSmallAabbs+1,&s);
-		m_prefixScanFloat4->execute(m_sum2,m_dst,numSmallAabbs+1,&s2);
-
-		b3Vector3 v = s2 - (s*s) / (float)numSmallAabbs;
-		
-		if(v[1] > v[0]) 
-			axis = 1;
-		if(v[2] > v[axis]) 
-			axis = 2;
-	}
-
-
-		
 		m_gpuSmallSortData.resize(numSmallAabbs);
-		
 
 #if 1
 		if (m_smallAabbsMappingGPU.size())
 		{
-			
 			B3_PROFILE("flipFloatKernel");
-			b3BufferInfoCL bInfo[] = { 
-				b3BufferInfoCL( m_allAabbsGPU.getBufferCL(), true ), 
-				b3BufferInfoCL( m_smallAabbsMappingGPU.getBufferCL(), true),
-				b3BufferInfoCL( m_gpuSmallSortData.getBufferCL())};
-			b3LauncherCL launcher(m_queue, m_flipFloatKernel ,"m_flipFloatKernel");
-			launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-			launcher.setConst( numSmallAabbs  );
-			launcher.setConst( axis  );
-			
+			b3BufferInfoCL bInfo[] = {
+				b3BufferInfoCL(m_allAabbsGPU.getBufferCL(), true),
+				b3BufferInfoCL(m_smallAabbsMappingGPU.getBufferCL(), true),
+				b3BufferInfoCL(m_gpuSmallSortData.getBufferCL())};
+			b3LauncherCL launcher(m_queue, m_flipFloatKernel, "m_flipFloatKernel");
+			launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+			launcher.setConst(numSmallAabbs);
+			launcher.setConst(axis);
+
 			int num = numSmallAabbs;
-			launcher.launch1D( num);
+			launcher.launch1D(num);
 			clFinish(m_queue);
 		}
 
@@ -1141,69 +1082,66 @@ void  b3GpuSapBroadphase::calculateOverlappingPairs(int maxPairs)
 		if (numSmallAabbs)
 		{
 			B3_PROFILE("scatterKernel");
-			
-			b3BufferInfoCL bInfo[] = { 
-				b3BufferInfoCL( m_allAabbsGPU.getBufferCL(), true ),
-				b3BufferInfoCL( m_smallAabbsMappingGPU.getBufferCL(), true),
-				b3BufferInfoCL( m_gpuSmallSortData.getBufferCL(),true),
+
+			b3BufferInfoCL bInfo[] = {
+				b3BufferInfoCL(m_allAabbsGPU.getBufferCL(), true),
+				b3BufferInfoCL(m_smallAabbsMappingGPU.getBufferCL(), true),
+				b3BufferInfoCL(m_gpuSmallSortData.getBufferCL(), true),
 				b3BufferInfoCL(m_gpuSmallSortedAabbs.getBufferCL())};
-			b3LauncherCL launcher(m_queue, m_scatterKernel ,"m_scatterKernel ");
-			launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-			launcher.setConst( numSmallAabbs);
+			b3LauncherCL launcher(m_queue, m_scatterKernel, "m_scatterKernel ");
+			launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+			launcher.setConst(numSmallAabbs);
 			int num = numSmallAabbs;
-			launcher.launch1D( num);
+			launcher.launch1D(num);
 			clFinish(m_queue);
-			
 		}
-        
 
-			m_overlappingPairs.resize(maxPairs);
+		m_overlappingPairs.resize(maxPairs);
 
-			m_pairCount.resize(0);
-			m_pairCount.push_back(0);
-            int numPairs=0;
+		m_pairCount.resize(0);
+		m_pairCount.push_back(0);
+		int numPairs = 0;
 
+		{
+			int numLargeAabbs = m_largeAabbsMappingGPU.size();
+			if (numLargeAabbs && numSmallAabbs)
 			{
-				int numLargeAabbs = m_largeAabbsMappingGPU.size();
-				if (numLargeAabbs && numSmallAabbs)
+				//@todo
+				B3_PROFILE("sap2Kernel");
+				b3BufferInfoCL bInfo[] = {
+					b3BufferInfoCL(m_allAabbsGPU.getBufferCL()),
+					b3BufferInfoCL(m_largeAabbsMappingGPU.getBufferCL()),
+					b3BufferInfoCL(m_smallAabbsMappingGPU.getBufferCL()),
+					b3BufferInfoCL(m_overlappingPairs.getBufferCL()),
+					b3BufferInfoCL(m_pairCount.getBufferCL())};
+				b3LauncherCL launcher(m_queue, m_sap2Kernel, "m_sap2Kernel");
+				launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+				launcher.setConst(numLargeAabbs);
+				launcher.setConst(numSmallAabbs);
+				launcher.setConst(axis);
+				launcher.setConst(maxPairs);
+				//@todo: use actual maximum work item sizes of the device instead of hardcoded values
+				launcher.launch2D(numLargeAabbs, numSmallAabbs, 4, 64);
+
+				numPairs = m_pairCount.at(0);
+				if (numPairs > maxPairs)
 				{
-					//@todo
-					B3_PROFILE("sap2Kernel");
-					b3BufferInfoCL bInfo[] = { 
-						b3BufferInfoCL( m_allAabbsGPU.getBufferCL() ),
-						b3BufferInfoCL( m_largeAabbsMappingGPU.getBufferCL() ),
-						b3BufferInfoCL( m_smallAabbsMappingGPU.getBufferCL() ), 
-						b3BufferInfoCL( m_overlappingPairs.getBufferCL() ), 
-						b3BufferInfoCL(m_pairCount.getBufferCL())};
-					b3LauncherCL launcher(m_queue, m_sap2Kernel,"m_sap2Kernel");
-					launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-					launcher.setConst(   numLargeAabbs  );
-					launcher.setConst( numSmallAabbs);
-					launcher.setConst( axis  );
-					launcher.setConst( maxPairs  );
-//@todo: use actual maximum work item sizes of the device instead of hardcoded values
-					launcher.launch2D( numLargeAabbs, numSmallAabbs,4,64);
-                
-					numPairs = m_pairCount.at(0);
-					if (numPairs >maxPairs)
-					{
-						b3Error("Error running out of pairs: numPairs = %d, maxPairs = %d.\n", numPairs, maxPairs);
-						numPairs =maxPairs;
-					}
+					b3Error("Error running out of pairs: numPairs = %d, maxPairs = %d.\n", numPairs, maxPairs);
+					numPairs = maxPairs;
 				}
 			}
-			if (m_gpuSmallSortedAabbs.size())
-			{
-				B3_PROFILE("sapKernel");
-				b3BufferInfoCL bInfo[] = { b3BufferInfoCL( m_gpuSmallSortedAabbs.getBufferCL() ), b3BufferInfoCL( m_overlappingPairs.getBufferCL() ), b3BufferInfoCL(m_pairCount.getBufferCL())};
-				b3LauncherCL launcher(m_queue, m_sapKernel,"m_sapKernel");
-				launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-				launcher.setConst( numSmallAabbs  );
-				launcher.setConst( axis  );
-				launcher.setConst( maxPairs  );
+		}
+		if (m_gpuSmallSortedAabbs.size())
+		{
+			B3_PROFILE("sapKernel");
+			b3BufferInfoCL bInfo[] = {b3BufferInfoCL(m_gpuSmallSortedAabbs.getBufferCL()), b3BufferInfoCL(m_overlappingPairs.getBufferCL()), b3BufferInfoCL(m_pairCount.getBufferCL())};
+			b3LauncherCL launcher(m_queue, m_sapKernel, "m_sapKernel");
+			launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+			launcher.setConst(numSmallAabbs);
+			launcher.setConst(axis);
+			launcher.setConst(maxPairs);
 
-			
-				int num = numSmallAabbs;
+			int num = numSmallAabbs;
 #if 0                
                 int buffSize = launcher.getSerializationBufferSize();
                 unsigned char* buf = new unsigned char[buffSize+sizeof(int)];
@@ -1225,73 +1163,71 @@ void  b3GpuSapBroadphase::calculateOverlappingPairs(int maxPairs)
                 FILE* f = fopen("m_sapKernelArgs.bin","wb");
                 fwrite(buf,buffSize+sizeof(int),1,f);
                 fclose(f);
-#endif//
+#endif  //
 
-                launcher.launch1D( num);
-				clFinish(m_queue);
-                
-                numPairs = m_pairCount.at(0);
-                if (numPairs>maxPairs)
-				{
-					b3Error("Error running out of pairs: numPairs = %d, maxPairs = %d.\n", numPairs, maxPairs);
-					numPairs = maxPairs;
-					m_pairCount.resize(0);
-					m_pairCount.push_back(maxPairs);
-				}
+			launcher.launch1D(num);
+			clFinish(m_queue);
+
+			numPairs = m_pairCount.at(0);
+			if (numPairs > maxPairs)
+			{
+				b3Error("Error running out of pairs: numPairs = %d, maxPairs = %d.\n", numPairs, maxPairs);
+				numPairs = maxPairs;
+				m_pairCount.resize(0);
+				m_pairCount.push_back(maxPairs);
 			}
-			
+		}
+
 #else
-        int numPairs = 0;
-        
-        
-        b3LauncherCL launcher(m_queue, m_sapKernel);
+		int numPairs = 0;
 
-        const char* fileName = "m_sapKernelArgs.bin";
-        FILE* f = fopen(fileName,"rb");
-        if (f)
-        {
-            int sizeInBytes=0;
-            if (fseek(f, 0, SEEK_END) || (sizeInBytes = ftell(f)) == EOF || fseek(f, 0, SEEK_SET)) 
-            {
-                printf("error, cannot get file size\n");
-                exit(0);
-            }
-            
-            unsigned char* buf = (unsigned char*) malloc(sizeInBytes);
-            fread(buf,sizeInBytes,1,f);
-            int serializedBytes = launcher.deserializeArgs(buf, sizeInBytes,m_context);
-            int num = *(int*)&buf[serializedBytes];
-            launcher.launch1D( num);
-            
-            b3OpenCLArray<int> pairCount(m_context, m_queue);
-            int numElements = launcher.m_arrays[2]->size()/sizeof(int);
-            pairCount.setFromOpenCLBuffer(launcher.m_arrays[2]->getBufferCL(),numElements);
-            numPairs = pairCount.at(0);
-            //printf("overlapping pairs = %d\n",numPairs);
-            b3AlignedObjectArray<b3Int4>		hostOoverlappingPairs;
-            b3OpenCLArray<b3Int4> tmpGpuPairs(m_context,m_queue);
-            tmpGpuPairs.setFromOpenCLBuffer(launcher.m_arrays[1]->getBufferCL(),numPairs );
-   
-            tmpGpuPairs.copyToHost(hostOoverlappingPairs);
-            m_overlappingPairs.copyFromHost(hostOoverlappingPairs);
-            //printf("hello %d\n", m_overlappingPairs.size());
-            free(buf);
-            fclose(f);
-            
-        } else {
-            printf("error: cannot find file %s\n",fileName);
-        }
-        
-        clFinish(m_queue);
+		b3LauncherCL launcher(m_queue, m_sapKernel);
+
+		const char* fileName = "m_sapKernelArgs.bin";
+		FILE* f = fopen(fileName, "rb");
+		if (f)
+		{
+			int sizeInBytes = 0;
+			if (fseek(f, 0, SEEK_END) || (sizeInBytes = ftell(f)) == EOF || fseek(f, 0, SEEK_SET))
+			{
+				printf("error, cannot get file size\n");
+				exit(0);
+			}
+
+			unsigned char* buf = (unsigned char*)malloc(sizeInBytes);
+			fread(buf, sizeInBytes, 1, f);
+			int serializedBytes = launcher.deserializeArgs(buf, sizeInBytes, m_context);
+			int num = *(int*)&buf[serializedBytes];
+			launcher.launch1D(num);
+
+			b3OpenCLArray<int> pairCount(m_context, m_queue);
+			int numElements = launcher.m_arrays[2]->size() / sizeof(int);
+			pairCount.setFromOpenCLBuffer(launcher.m_arrays[2]->getBufferCL(), numElements);
+			numPairs = pairCount.at(0);
+			//printf("overlapping pairs = %d\n",numPairs);
+			b3AlignedObjectArray<b3Int4> hostOoverlappingPairs;
+			b3OpenCLArray<b3Int4> tmpGpuPairs(m_context, m_queue);
+			tmpGpuPairs.setFromOpenCLBuffer(launcher.m_arrays[1]->getBufferCL(), numPairs);
+
+			tmpGpuPairs.copyToHost(hostOoverlappingPairs);
+			m_overlappingPairs.copyFromHost(hostOoverlappingPairs);
+			//printf("hello %d\n", m_overlappingPairs.size());
+			free(buf);
+			fclose(f);
+		}
+		else
+		{
+			printf("error: cannot find file %s\n", fileName);
+		}
+
+		clFinish(m_queue);
 
-        
 #endif
 
-			
-        m_overlappingPairs.resize(numPairs);
-		
-	}//B3_PROFILE("GPU_RADIX SORT");
-	//init3dSap();
+		m_overlappingPairs.resize(numPairs);
+
+	}  //B3_PROFILE("GPU_RADIX SORT");
+	   //init3dSap();
 }
 
 void b3GpuSapBroadphase::writeAabbsToGpu()
@@ -1299,17 +1235,14 @@ void b3GpuSapBroadphase::writeAabbsToGpu()
 	m_smallAabbsMappingGPU.copyFromHost(m_smallAabbsMappingCPU);
 	m_largeAabbsMappingGPU.copyFromHost(m_largeAabbsMappingCPU);
 
-	m_allAabbsGPU.copyFromHost(m_allAabbsCPU);//might not be necessary, the 'setupGpuAabbsFull' already takes care of this
-	
-	
-
+	m_allAabbsGPU.copyFromHost(m_allAabbsCPU);  //might not be necessary, the 'setupGpuAabbsFull' already takes care of this
 }
 
-void b3GpuSapBroadphase::createLargeProxy(const b3Vector3& aabbMin,  const b3Vector3& aabbMax, int userPtr , int collisionFilterGroup, int collisionFilterMask)
+void b3GpuSapBroadphase::createLargeProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, int userPtr, int collisionFilterGroup, int collisionFilterMask)
 {
 	int index = userPtr;
 	b3SapAabb aabb;
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 	{
 		aabb.m_min[i] = aabbMin[i];
 		aabb.m_max[i] = aabbMax[i];
@@ -1317,15 +1250,15 @@ void b3GpuSapBroadphase::createLargeProxy(const b3Vector3& aabbMin,  const b3Vec
 	aabb.m_minIndices[3] = index;
 	aabb.m_signedMaxIndices[3] = m_allAabbsCPU.size();
 	m_largeAabbsMappingCPU.push_back(m_allAabbsCPU.size());
-	
+
 	m_allAabbsCPU.push_back(aabb);
 }
 
-void b3GpuSapBroadphase::createProxy(const b3Vector3& aabbMin,  const b3Vector3& aabbMax, int userPtr , int collisionFilterGroup, int collisionFilterMask)
+void b3GpuSapBroadphase::createProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, int userPtr, int collisionFilterGroup, int collisionFilterMask)
 {
 	int index = userPtr;
 	b3SapAabb aabb;
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 	{
 		aabb.m_min[i] = aabbMin[i];
 		aabb.m_max[i] = aabbMax[i];
@@ -1334,20 +1267,19 @@ void b3GpuSapBroadphase::createProxy(const b3Vector3& aabbMin,  const b3Vector3&
 	aabb.m_signedMaxIndices[3] = m_allAabbsCPU.size();
 	m_smallAabbsMappingCPU.push_back(m_allAabbsCPU.size());
 
-
 	m_allAabbsCPU.push_back(aabb);
 }
 
-cl_mem	b3GpuSapBroadphase::getAabbBufferWS()
+cl_mem b3GpuSapBroadphase::getAabbBufferWS()
 {
 	return m_allAabbsGPU.getBufferCL();
 }
 
-int	b3GpuSapBroadphase::getNumOverlap()
+int b3GpuSapBroadphase::getNumOverlap()
 {
 	return m_overlappingPairs.size();
 }
-cl_mem	b3GpuSapBroadphase::getOverlappingPairBuffer()
+cl_mem b3GpuSapBroadphase::getOverlappingPairBuffer()
 {
 	return m_overlappingPairs.getBufferCL();
 }
diff --git a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuSapBroadphase.h b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuSapBroadphase.h
index 8d36ac78f28..d17590b14a0 100644
--- a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuSapBroadphase.h
+++ b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3GpuSapBroadphase.h
@@ -2,7 +2,7 @@
 #define B3_GPU_SAP_BROADPHASE_H
 
 #include "Bullet3OpenCL/ParallelPrimitives/b3OpenCLArray.h"
-#include "Bullet3OpenCL/ParallelPrimitives/b3FillCL.h" //b3Int2
+#include "Bullet3OpenCL/ParallelPrimitives/b3FillCL.h"  //b3Int2
 class b3Vector3;
 #include "Bullet3OpenCL/ParallelPrimitives/b3RadixSort32CL.h"
 
@@ -11,141 +11,133 @@ class b3Vector3;
 
 #include "b3GpuBroadphaseInterface.h"
 
-
 class b3GpuSapBroadphase : public b3GpuBroadphaseInterface
 {
-	
-	cl_context				m_context;
-	cl_device_id			m_device;
-	cl_command_queue		m_queue;
-	cl_kernel				m_flipFloatKernel;
-	cl_kernel				m_scatterKernel ;
-	cl_kernel				m_copyAabbsKernel;
-	cl_kernel				m_sapKernel;
-	cl_kernel				m_sap2Kernel;
-	cl_kernel				m_prepareSumVarianceKernel;
-	
+	cl_context m_context;
+	cl_device_id m_device;
+	cl_command_queue m_queue;
+	cl_kernel m_flipFloatKernel;
+	cl_kernel m_scatterKernel;
+	cl_kernel m_copyAabbsKernel;
+	cl_kernel m_sapKernel;
+	cl_kernel m_sap2Kernel;
+	cl_kernel m_prepareSumVarianceKernel;
 
 	class b3RadixSort32CL* m_sorter;
 
 	///test for 3d SAP
-	b3AlignedObjectArray<b3SortData>		m_sortedAxisCPU[3][2];
-	b3AlignedObjectArray<b3UnsignedInt2>	m_objectMinMaxIndexCPU[3][2];
-	b3OpenCLArray<b3UnsignedInt2>			m_objectMinMaxIndexGPUaxis0;
-	b3OpenCLArray<b3UnsignedInt2>			m_objectMinMaxIndexGPUaxis1;
-	b3OpenCLArray<b3UnsignedInt2>			m_objectMinMaxIndexGPUaxis2;
-	b3OpenCLArray<b3UnsignedInt2>			m_objectMinMaxIndexGPUaxis0prev;
-	b3OpenCLArray<b3UnsignedInt2>			m_objectMinMaxIndexGPUaxis1prev;
-	b3OpenCLArray<b3UnsignedInt2>			m_objectMinMaxIndexGPUaxis2prev;
+	b3AlignedObjectArray<b3SortData> m_sortedAxisCPU[3][2];
+	b3AlignedObjectArray<b3UnsignedInt2> m_objectMinMaxIndexCPU[3][2];
+	b3OpenCLArray<b3UnsignedInt2> m_objectMinMaxIndexGPUaxis0;
+	b3OpenCLArray<b3UnsignedInt2> m_objectMinMaxIndexGPUaxis1;
+	b3OpenCLArray<b3UnsignedInt2> m_objectMinMaxIndexGPUaxis2;
+	b3OpenCLArray<b3UnsignedInt2> m_objectMinMaxIndexGPUaxis0prev;
+	b3OpenCLArray<b3UnsignedInt2> m_objectMinMaxIndexGPUaxis1prev;
+	b3OpenCLArray<b3UnsignedInt2> m_objectMinMaxIndexGPUaxis2prev;
 
-	b3OpenCLArray<b3SortData>				m_sortedAxisGPU0;
-	b3OpenCLArray<b3SortData>				m_sortedAxisGPU1;
-	b3OpenCLArray<b3SortData>				m_sortedAxisGPU2;
-	b3OpenCLArray<b3SortData>				m_sortedAxisGPU0prev;
-	b3OpenCLArray<b3SortData>				m_sortedAxisGPU1prev;
-	b3OpenCLArray<b3SortData>				m_sortedAxisGPU2prev;
+	b3OpenCLArray<b3SortData> m_sortedAxisGPU0;
+	b3OpenCLArray<b3SortData> m_sortedAxisGPU1;
+	b3OpenCLArray<b3SortData> m_sortedAxisGPU2;
+	b3OpenCLArray<b3SortData> m_sortedAxisGPU0prev;
+	b3OpenCLArray<b3SortData> m_sortedAxisGPU1prev;
+	b3OpenCLArray<b3SortData> m_sortedAxisGPU2prev;
 
+	b3OpenCLArray<b3Int4> m_addedHostPairsGPU;
+	b3OpenCLArray<b3Int4> m_removedHostPairsGPU;
+	b3OpenCLArray<int> m_addedCountGPU;
+	b3OpenCLArray<int> m_removedCountGPU;
 
-	b3OpenCLArray<b3Int4>					m_addedHostPairsGPU;
-	b3OpenCLArray<b3Int4>					m_removedHostPairsGPU;
-	b3OpenCLArray<int>						m_addedCountGPU;
-	b3OpenCLArray<int>						m_removedCountGPU;
-	
-	int	m_currentBuffer;
+	int m_currentBuffer;
 
 public:
-
 	b3OpenCLArray<int> m_pairCount;
 
+	b3OpenCLArray<b3SapAabb> m_allAabbsGPU;
+	b3AlignedObjectArray<b3SapAabb> m_allAabbsCPU;
 
-	b3OpenCLArray<b3SapAabb>	m_allAabbsGPU;
-	b3AlignedObjectArray<b3SapAabb>	m_allAabbsCPU;
-
-	virtual b3OpenCLArray<b3SapAabb>&	getAllAabbsGPU()
+	virtual b3OpenCLArray<b3SapAabb>& getAllAabbsGPU()
 	{
 		return m_allAabbsGPU;
 	}
-	virtual b3AlignedObjectArray<b3SapAabb>&	getAllAabbsCPU()
+	virtual b3AlignedObjectArray<b3SapAabb>& getAllAabbsCPU()
 	{
 		return m_allAabbsCPU;
 	}
 
-	b3OpenCLArray<b3Vector3>	m_sum;
-	b3OpenCLArray<b3Vector3>	m_sum2;
-	b3OpenCLArray<b3Vector3>	m_dst;
+	b3OpenCLArray<b3Vector3> m_sum;
+	b3OpenCLArray<b3Vector3> m_sum2;
+	b3OpenCLArray<b3Vector3> m_dst;
 
-	b3OpenCLArray<int>	m_smallAabbsMappingGPU;
+	b3OpenCLArray<int> m_smallAabbsMappingGPU;
 	b3AlignedObjectArray<int> m_smallAabbsMappingCPU;
 
-	b3OpenCLArray<int>	m_largeAabbsMappingGPU;
+	b3OpenCLArray<int> m_largeAabbsMappingGPU;
 	b3AlignedObjectArray<int> m_largeAabbsMappingCPU;
 
-	
-	b3OpenCLArray<b3Int4>		m_overlappingPairs;
+	b3OpenCLArray<b3Int4> m_overlappingPairs;
 
 	//temporary gpu work memory
-	b3OpenCLArray<b3SortData>	m_gpuSmallSortData;
-	b3OpenCLArray<b3SapAabb>	m_gpuSmallSortedAabbs;
+	b3OpenCLArray<b3SortData> m_gpuSmallSortData;
+	b3OpenCLArray<b3SapAabb> m_gpuSmallSortedAabbs;
 
-	class b3PrefixScanFloat4CL*		m_prefixScanFloat4;
+	class b3PrefixScanFloat4CL* m_prefixScanFloat4;
 
 	enum b3GpuSapKernelType
 	{
-		B3_GPU_SAP_KERNEL_BRUTE_FORCE_CPU=1,
+		B3_GPU_SAP_KERNEL_BRUTE_FORCE_CPU = 1,
 		B3_GPU_SAP_KERNEL_BRUTE_FORCE_GPU,
 		B3_GPU_SAP_KERNEL_ORIGINAL,
 		B3_GPU_SAP_KERNEL_BARRIER,
 		B3_GPU_SAP_KERNEL_LOCAL_SHARED_MEMORY
 	};
 
-	b3GpuSapBroadphase(cl_context ctx,cl_device_id device, cl_command_queue  q , b3GpuSapKernelType kernelType=B3_GPU_SAP_KERNEL_LOCAL_SHARED_MEMORY);
+	b3GpuSapBroadphase(cl_context ctx, cl_device_id device, cl_command_queue q, b3GpuSapKernelType kernelType = B3_GPU_SAP_KERNEL_LOCAL_SHARED_MEMORY);
 	virtual ~b3GpuSapBroadphase();
-	
-	static b3GpuBroadphaseInterface* CreateFuncBruteForceCpu(cl_context ctx,cl_device_id device, cl_command_queue  q)
+
+	static b3GpuBroadphaseInterface* CreateFuncBruteForceCpu(cl_context ctx, cl_device_id device, cl_command_queue q)
 	{
-		return new b3GpuSapBroadphase(ctx,device,q,B3_GPU_SAP_KERNEL_BRUTE_FORCE_CPU);
+		return new b3GpuSapBroadphase(ctx, device, q, B3_GPU_SAP_KERNEL_BRUTE_FORCE_CPU);
 	}
 
-	static b3GpuBroadphaseInterface* CreateFuncBruteForceGpu(cl_context ctx,cl_device_id device, cl_command_queue  q)
+	static b3GpuBroadphaseInterface* CreateFuncBruteForceGpu(cl_context ctx, cl_device_id device, cl_command_queue q)
 	{
-		return new b3GpuSapBroadphase(ctx,device,q,B3_GPU_SAP_KERNEL_BRUTE_FORCE_GPU);
+		return new b3GpuSapBroadphase(ctx, device, q, B3_GPU_SAP_KERNEL_BRUTE_FORCE_GPU);
 	}
 
-	static b3GpuBroadphaseInterface* CreateFuncOriginal(cl_context ctx,cl_device_id device, cl_command_queue  q)
+	static b3GpuBroadphaseInterface* CreateFuncOriginal(cl_context ctx, cl_device_id device, cl_command_queue q)
 	{
-		return new b3GpuSapBroadphase(ctx,device,q,B3_GPU_SAP_KERNEL_ORIGINAL);
+		return new b3GpuSapBroadphase(ctx, device, q, B3_GPU_SAP_KERNEL_ORIGINAL);
 	}
-	static b3GpuBroadphaseInterface* CreateFuncBarrier(cl_context ctx,cl_device_id device, cl_command_queue  q)
+	static b3GpuBroadphaseInterface* CreateFuncBarrier(cl_context ctx, cl_device_id device, cl_command_queue q)
 	{
-		return new b3GpuSapBroadphase(ctx,device,q,B3_GPU_SAP_KERNEL_BARRIER);
+		return new b3GpuSapBroadphase(ctx, device, q, B3_GPU_SAP_KERNEL_BARRIER);
 	}
-	static b3GpuBroadphaseInterface* CreateFuncLocalMemory(cl_context ctx,cl_device_id device, cl_command_queue  q)
+	static b3GpuBroadphaseInterface* CreateFuncLocalMemory(cl_context ctx, cl_device_id device, cl_command_queue q)
 	{
-		return new b3GpuSapBroadphase(ctx,device,q,B3_GPU_SAP_KERNEL_LOCAL_SHARED_MEMORY);
+		return new b3GpuSapBroadphase(ctx, device, q, B3_GPU_SAP_KERNEL_LOCAL_SHARED_MEMORY);
 	}
-	
 
-	virtual void  calculateOverlappingPairs(int maxPairs);
-	virtual void  calculateOverlappingPairsHost(int maxPairs);
-	
-	void  reset();
+	virtual void calculateOverlappingPairs(int maxPairs);
+	virtual void calculateOverlappingPairsHost(int maxPairs);
+
+	void reset();
 
 	void init3dSap();
 	virtual void calculateOverlappingPairsHostIncremental3Sap();
 
-	virtual void createProxy(const b3Vector3& aabbMin,  const b3Vector3& aabbMax, int userPtr , int collisionFilterGroup, int collisionFilterMask);
-	virtual void createLargeProxy(const b3Vector3& aabbMin,  const b3Vector3& aabbMax, int userPtr , int collisionFilterGroup, int collisionFilterMask);
+	virtual void createProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, int userPtr, int collisionFilterGroup, int collisionFilterMask);
+	virtual void createLargeProxy(const b3Vector3& aabbMin, const b3Vector3& aabbMax, int userPtr, int collisionFilterGroup, int collisionFilterMask);
 
 	//call writeAabbsToGpu after done making all changes (createProxy etc)
 	virtual void writeAabbsToGpu();
 
-	virtual cl_mem	getAabbBufferWS();
-	virtual int	getNumOverlap();
-	virtual cl_mem	getOverlappingPairBuffer();
-	
+	virtual cl_mem getAabbBufferWS();
+	virtual int getNumOverlap();
+	virtual cl_mem getOverlappingPairBuffer();
+
 	virtual b3OpenCLArray<b3Int4>& getOverlappingPairsGPU();
 	virtual b3OpenCLArray<int>& getSmallAabbIndicesGPU();
 	virtual b3OpenCLArray<int>& getLargeAabbIndicesGPU();
 };
 
-#endif //B3_GPU_SAP_BROADPHASE_H
\ No newline at end of file
+#endif  //B3_GPU_SAP_BROADPHASE_H
\ No newline at end of file
diff --git a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3SapAabb.h b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3SapAabb.h
index ea6550fedea..60570f2605c 100644
--- a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3SapAabb.h
+++ b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/b3SapAabb.h
@@ -5,10 +5,9 @@
 #include "Bullet3Collision/BroadPhaseCollision/shared/b3Aabb.h"
 
 ///just make sure that the b3Aabb is 16-byte aligned
-B3_ATTRIBUTE_ALIGNED16(struct) b3SapAabb : public b3Aabb
-{
+B3_ATTRIBUTE_ALIGNED16(struct)
+b3SapAabb : public b3Aabb{
 
-};
+			};
 
-
-#endif //B3_SAP_AABB_H
+#endif  //B3_SAP_AABB_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/kernels/gridBroadphaseKernels.h b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/kernels/gridBroadphaseKernels.h
index dad42477c37..01854177869 100644
--- a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/kernels/gridBroadphaseKernels.h
+++ b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/kernels/gridBroadphaseKernels.h
@@ -1,199 +1,198 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* gridBroadphaseCL= \
-"int getPosHash(int4 gridPos, __global float4* pParams)\n"
-"{\n"
-"	int4 gridDim = *((__global int4*)(pParams + 1));\n"
-"	gridPos.x &= gridDim.x - 1;\n"
-"	gridPos.y &= gridDim.y - 1;\n"
-"	gridPos.z &= gridDim.z - 1;\n"
-"	int hash = gridPos.z * gridDim.y * gridDim.x + gridPos.y * gridDim.x + gridPos.x;\n"
-"	return hash;\n"
-"} \n"
-"int4 getGridPos(float4 worldPos, __global float4* pParams)\n"
-"{\n"
-"    int4 gridPos;\n"
-"	int4 gridDim = *((__global int4*)(pParams + 1));\n"
-"    gridPos.x = (int)floor(worldPos.x * pParams[0].x) & (gridDim.x - 1);\n"
-"    gridPos.y = (int)floor(worldPos.y * pParams[0].y) & (gridDim.y - 1);\n"
-"    gridPos.z = (int)floor(worldPos.z * pParams[0].z) & (gridDim.z - 1);\n"
-"    return gridPos;\n"
-"}\n"
-"// calculate grid hash value for each body using its AABB\n"
-"__kernel void kCalcHashAABB(int numObjects, __global float4* allpAABB, __global const int* smallAabbMapping, __global int2* pHash, __global float4* pParams )\n"
-"{\n"
-"    int index = get_global_id(0);\n"
-"    if(index >= numObjects)\n"
-"	{\n"
-"		return;\n"
-"	}\n"
-"	float4 bbMin = allpAABB[smallAabbMapping[index]*2];\n"
-"	float4 bbMax = allpAABB[smallAabbMapping[index]*2 + 1];\n"
-"	float4 pos;\n"
-"	pos.x = (bbMin.x + bbMax.x) * 0.5f;\n"
-"	pos.y = (bbMin.y + bbMax.y) * 0.5f;\n"
-"	pos.z = (bbMin.z + bbMax.z) * 0.5f;\n"
-"	pos.w = 0.f;\n"
-"    // get address in grid\n"
-"    int4 gridPos = getGridPos(pos, pParams);\n"
-"    int gridHash = getPosHash(gridPos, pParams);\n"
-"    // store grid hash and body index\n"
-"    int2 hashVal;\n"
-"    hashVal.x = gridHash;\n"
-"    hashVal.y = index;\n"
-"    pHash[index] = hashVal;\n"
-"}\n"
-"__kernel void kClearCellStart(	int numCells, \n"
-"								__global int* pCellStart )\n"
-"{\n"
-"    int index = get_global_id(0);\n"
-"    if(index >= numCells)\n"
-"	{\n"
-"		return;\n"
-"	}\n"
-"	pCellStart[index] = -1;\n"
-"}\n"
-"__kernel void kFindCellStart(int numObjects, __global int2* pHash, __global int* cellStart )\n"
-"{\n"
-"	__local int sharedHash[513];\n"
-"    int index = get_global_id(0);\n"
-"	int2 sortedData;\n"
-"    if(index < numObjects)\n"
-"	{\n"
-"		sortedData = pHash[index];\n"
-"		// Load hash data into shared memory so that we can look \n"
-"		// at neighboring body's hash value without loading\n"
-"		// two hash values per thread\n"
-"		sharedHash[get_local_id(0) + 1] = sortedData.x;\n"
-"		if((index > 0) && (get_local_id(0) == 0))\n"
-"		{\n"
-"			// first thread in block must load neighbor body hash\n"
-"			sharedHash[0] = pHash[index-1].x;\n"
-"		}\n"
-"	}\n"
-"    barrier(CLK_LOCAL_MEM_FENCE);\n"
-"    if(index < numObjects)\n"
-"	{\n"
-"		if((index == 0) || (sortedData.x != sharedHash[get_local_id(0)]))\n"
-"		{\n"
-"			cellStart[sortedData.x] = index;\n"
-"		}\n"
-"	}\n"
-"}\n"
-"int testAABBOverlap(float4 min0, float4 max0, float4 min1, float4 max1)\n"
-"{\n"
-"	return	(min0.x <= max1.x)&& (min1.x <= max0.x) && \n"
-"			(min0.y <= max1.y)&& (min1.y <= max0.y) && \n"
-"			(min0.z <= max1.z)&& (min1.z <= max0.z); \n"
-"}\n"
-"//search for AABB 'index' against other AABBs' in this cell\n"
-"void findPairsInCell(	int numObjects,\n"
-"						int4	gridPos,\n"
-"						int    index,\n"
-"						__global int2*  pHash,\n"
-"						__global int*   pCellStart,\n"
-"						__global float4* allpAABB, \n"
-"						__global const int* smallAabbMapping,\n"
-"						__global float4* pParams,\n"
-"							volatile  __global int* pairCount,\n"
-"						__global int4*   pPairBuff2,\n"
-"						int maxPairs\n"
-"						)\n"
-"{\n"
-"	int4 pGridDim = *((__global int4*)(pParams + 1));\n"
-"	int maxBodiesPerCell = pGridDim.w;\n"
-"    int gridHash = getPosHash(gridPos, pParams);\n"
-"    // get start of bucket for this cell\n"
-"    int bucketStart = pCellStart[gridHash];\n"
-"    if (bucketStart == -1)\n"
-"	{\n"
-"        return;   // cell empty\n"
-"	}\n"
-"	// iterate over bodies in this cell\n"
-"    int2 sortedData = pHash[index];\n"
-"	int unsorted_indx = sortedData.y;\n"
-"    float4 min0 = allpAABB[smallAabbMapping[unsorted_indx]*2 + 0]; \n"
-"	float4 max0 = allpAABB[smallAabbMapping[unsorted_indx]*2 + 1];\n"
-"	int handleIndex =  as_int(min0.w);\n"
-"	\n"
-"	int bucketEnd = bucketStart + maxBodiesPerCell;\n"
-"	bucketEnd = (bucketEnd > numObjects) ? numObjects : bucketEnd;\n"
-"	for(int index2 = bucketStart; index2 < bucketEnd; index2++) \n"
-"	{\n"
-"        int2 cellData = pHash[index2];\n"
-"        if (cellData.x != gridHash)\n"
-"        {\n"
-"			break;   // no longer in same bucket\n"
-"		}\n"
-"		int unsorted_indx2 = cellData.y;\n"
-"        //if (unsorted_indx2 < unsorted_indx) // check not colliding with self\n"
-"		if (unsorted_indx2 != unsorted_indx) // check not colliding with self\n"
-"        {   \n"
-"			float4 min1 = allpAABB[smallAabbMapping[unsorted_indx2]*2 + 0];\n"
-"			float4 max1 = allpAABB[smallAabbMapping[unsorted_indx2]*2 + 1];\n"
-"			if(testAABBOverlap(min0, max0, min1, max1))\n"
-"			{\n"
-"				if (pairCount)\n"
-"				{\n"
-"					int handleIndex2 = as_int(min1.w);\n"
-"					if (handleIndex<handleIndex2)\n"
-"					{\n"
-"						int curPair = atomic_add(pairCount,1);\n"
-"						if (curPair<maxPairs)\n"
-"						{\n"
-"							int4 newpair;\n"
-"							newpair.x = handleIndex;\n"
-"							newpair.y = handleIndex2;\n"
-"							newpair.z = -1;\n"
-"							newpair.w = -1;\n"
-"							pPairBuff2[curPair] = newpair;\n"
-"						}\n"
-"					}\n"
-"				\n"
-"				}\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"}\n"
-"__kernel void kFindOverlappingPairs(	int numObjects,\n"
-"										__global float4* allpAABB, \n"
-"										__global const int* smallAabbMapping,\n"
-"										__global int2* pHash, \n"
-"										__global int* pCellStart, \n"
-"										__global float4* pParams ,\n"
-"										volatile  __global int* pairCount,\n"
-"										__global int4*   pPairBuff2,\n"
-"										int maxPairs\n"
-"										)\n"
-"{\n"
-"    int index = get_global_id(0);\n"
-"    if(index >= numObjects)\n"
-"	{\n"
-"		return;\n"
-"	}\n"
-"    int2 sortedData = pHash[index];\n"
-"	int unsorted_indx = sortedData.y;\n"
-"	float4 bbMin = allpAABB[smallAabbMapping[unsorted_indx]*2 + 0];\n"
-"	float4 bbMax = allpAABB[smallAabbMapping[unsorted_indx]*2 + 1];\n"
-"	float4 pos;\n"
-"	pos.x = (bbMin.x + bbMax.x) * 0.5f;\n"
-"	pos.y = (bbMin.y + bbMax.y) * 0.5f;\n"
-"	pos.z = (bbMin.z + bbMax.z) * 0.5f;\n"
-"    // get address in grid\n"
-"    int4 gridPosA = getGridPos(pos, pParams);\n"
-"    int4 gridPosB; \n"
-"    // examine only neighbouring cells\n"
-"    for(int z=-1; z<=1; z++) \n"
-"    {\n"
-"		gridPosB.z = gridPosA.z + z;\n"
-"        for(int y=-1; y<=1; y++) \n"
-"        {\n"
-"			gridPosB.y = gridPosA.y + y;\n"
-"            for(int x=-1; x<=1; x++) \n"
-"            {\n"
-"				gridPosB.x = gridPosA.x + x;\n"
-"                findPairsInCell(numObjects, gridPosB, index, pHash, pCellStart, allpAABB,smallAabbMapping, pParams, pairCount,pPairBuff2, maxPairs);\n"
-"            }\n"
-"        }\n"
-"    }\n"
-"}\n"
-;
+static const char* gridBroadphaseCL =
+	"int getPosHash(int4 gridPos, __global float4* pParams)\n"
+	"{\n"
+	"	int4 gridDim = *((__global int4*)(pParams + 1));\n"
+	"	gridPos.x &= gridDim.x - 1;\n"
+	"	gridPos.y &= gridDim.y - 1;\n"
+	"	gridPos.z &= gridDim.z - 1;\n"
+	"	int hash = gridPos.z * gridDim.y * gridDim.x + gridPos.y * gridDim.x + gridPos.x;\n"
+	"	return hash;\n"
+	"} \n"
+	"int4 getGridPos(float4 worldPos, __global float4* pParams)\n"
+	"{\n"
+	"    int4 gridPos;\n"
+	"	int4 gridDim = *((__global int4*)(pParams + 1));\n"
+	"    gridPos.x = (int)floor(worldPos.x * pParams[0].x) & (gridDim.x - 1);\n"
+	"    gridPos.y = (int)floor(worldPos.y * pParams[0].y) & (gridDim.y - 1);\n"
+	"    gridPos.z = (int)floor(worldPos.z * pParams[0].z) & (gridDim.z - 1);\n"
+	"    return gridPos;\n"
+	"}\n"
+	"// calculate grid hash value for each body using its AABB\n"
+	"__kernel void kCalcHashAABB(int numObjects, __global float4* allpAABB, __global const int* smallAabbMapping, __global int2* pHash, __global float4* pParams )\n"
+	"{\n"
+	"    int index = get_global_id(0);\n"
+	"    if(index >= numObjects)\n"
+	"	{\n"
+	"		return;\n"
+	"	}\n"
+	"	float4 bbMin = allpAABB[smallAabbMapping[index]*2];\n"
+	"	float4 bbMax = allpAABB[smallAabbMapping[index]*2 + 1];\n"
+	"	float4 pos;\n"
+	"	pos.x = (bbMin.x + bbMax.x) * 0.5f;\n"
+	"	pos.y = (bbMin.y + bbMax.y) * 0.5f;\n"
+	"	pos.z = (bbMin.z + bbMax.z) * 0.5f;\n"
+	"	pos.w = 0.f;\n"
+	"    // get address in grid\n"
+	"    int4 gridPos = getGridPos(pos, pParams);\n"
+	"    int gridHash = getPosHash(gridPos, pParams);\n"
+	"    // store grid hash and body index\n"
+	"    int2 hashVal;\n"
+	"    hashVal.x = gridHash;\n"
+	"    hashVal.y = index;\n"
+	"    pHash[index] = hashVal;\n"
+	"}\n"
+	"__kernel void kClearCellStart(	int numCells, \n"
+	"								__global int* pCellStart )\n"
+	"{\n"
+	"    int index = get_global_id(0);\n"
+	"    if(index >= numCells)\n"
+	"	{\n"
+	"		return;\n"
+	"	}\n"
+	"	pCellStart[index] = -1;\n"
+	"}\n"
+	"__kernel void kFindCellStart(int numObjects, __global int2* pHash, __global int* cellStart )\n"
+	"{\n"
+	"	__local int sharedHash[513];\n"
+	"    int index = get_global_id(0);\n"
+	"	int2 sortedData;\n"
+	"    if(index < numObjects)\n"
+	"	{\n"
+	"		sortedData = pHash[index];\n"
+	"		// Load hash data into shared memory so that we can look \n"
+	"		// at neighboring body's hash value without loading\n"
+	"		// two hash values per thread\n"
+	"		sharedHash[get_local_id(0) + 1] = sortedData.x;\n"
+	"		if((index > 0) && (get_local_id(0) == 0))\n"
+	"		{\n"
+	"			// first thread in block must load neighbor body hash\n"
+	"			sharedHash[0] = pHash[index-1].x;\n"
+	"		}\n"
+	"	}\n"
+	"    barrier(CLK_LOCAL_MEM_FENCE);\n"
+	"    if(index < numObjects)\n"
+	"	{\n"
+	"		if((index == 0) || (sortedData.x != sharedHash[get_local_id(0)]))\n"
+	"		{\n"
+	"			cellStart[sortedData.x] = index;\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"int testAABBOverlap(float4 min0, float4 max0, float4 min1, float4 max1)\n"
+	"{\n"
+	"	return	(min0.x <= max1.x)&& (min1.x <= max0.x) && \n"
+	"			(min0.y <= max1.y)&& (min1.y <= max0.y) && \n"
+	"			(min0.z <= max1.z)&& (min1.z <= max0.z); \n"
+	"}\n"
+	"//search for AABB 'index' against other AABBs' in this cell\n"
+	"void findPairsInCell(	int numObjects,\n"
+	"						int4	gridPos,\n"
+	"						int    index,\n"
+	"						__global int2*  pHash,\n"
+	"						__global int*   pCellStart,\n"
+	"						__global float4* allpAABB, \n"
+	"						__global const int* smallAabbMapping,\n"
+	"						__global float4* pParams,\n"
+	"							volatile  __global int* pairCount,\n"
+	"						__global int4*   pPairBuff2,\n"
+	"						int maxPairs\n"
+	"						)\n"
+	"{\n"
+	"	int4 pGridDim = *((__global int4*)(pParams + 1));\n"
+	"	int maxBodiesPerCell = pGridDim.w;\n"
+	"    int gridHash = getPosHash(gridPos, pParams);\n"
+	"    // get start of bucket for this cell\n"
+	"    int bucketStart = pCellStart[gridHash];\n"
+	"    if (bucketStart == -1)\n"
+	"	{\n"
+	"        return;   // cell empty\n"
+	"	}\n"
+	"	// iterate over bodies in this cell\n"
+	"    int2 sortedData = pHash[index];\n"
+	"	int unsorted_indx = sortedData.y;\n"
+	"    float4 min0 = allpAABB[smallAabbMapping[unsorted_indx]*2 + 0]; \n"
+	"	float4 max0 = allpAABB[smallAabbMapping[unsorted_indx]*2 + 1];\n"
+	"	int handleIndex =  as_int(min0.w);\n"
+	"	\n"
+	"	int bucketEnd = bucketStart + maxBodiesPerCell;\n"
+	"	bucketEnd = (bucketEnd > numObjects) ? numObjects : bucketEnd;\n"
+	"	for(int index2 = bucketStart; index2 < bucketEnd; index2++) \n"
+	"	{\n"
+	"        int2 cellData = pHash[index2];\n"
+	"        if (cellData.x != gridHash)\n"
+	"        {\n"
+	"			break;   // no longer in same bucket\n"
+	"		}\n"
+	"		int unsorted_indx2 = cellData.y;\n"
+	"        //if (unsorted_indx2 < unsorted_indx) // check not colliding with self\n"
+	"		if (unsorted_indx2 != unsorted_indx) // check not colliding with self\n"
+	"        {   \n"
+	"			float4 min1 = allpAABB[smallAabbMapping[unsorted_indx2]*2 + 0];\n"
+	"			float4 max1 = allpAABB[smallAabbMapping[unsorted_indx2]*2 + 1];\n"
+	"			if(testAABBOverlap(min0, max0, min1, max1))\n"
+	"			{\n"
+	"				if (pairCount)\n"
+	"				{\n"
+	"					int handleIndex2 = as_int(min1.w);\n"
+	"					if (handleIndex<handleIndex2)\n"
+	"					{\n"
+	"						int curPair = atomic_add(pairCount,1);\n"
+	"						if (curPair<maxPairs)\n"
+	"						{\n"
+	"							int4 newpair;\n"
+	"							newpair.x = handleIndex;\n"
+	"							newpair.y = handleIndex2;\n"
+	"							newpair.z = -1;\n"
+	"							newpair.w = -1;\n"
+	"							pPairBuff2[curPair] = newpair;\n"
+	"						}\n"
+	"					}\n"
+	"				\n"
+	"				}\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"__kernel void kFindOverlappingPairs(	int numObjects,\n"
+	"										__global float4* allpAABB, \n"
+	"										__global const int* smallAabbMapping,\n"
+	"										__global int2* pHash, \n"
+	"										__global int* pCellStart, \n"
+	"										__global float4* pParams ,\n"
+	"										volatile  __global int* pairCount,\n"
+	"										__global int4*   pPairBuff2,\n"
+	"										int maxPairs\n"
+	"										)\n"
+	"{\n"
+	"    int index = get_global_id(0);\n"
+	"    if(index >= numObjects)\n"
+	"	{\n"
+	"		return;\n"
+	"	}\n"
+	"    int2 sortedData = pHash[index];\n"
+	"	int unsorted_indx = sortedData.y;\n"
+	"	float4 bbMin = allpAABB[smallAabbMapping[unsorted_indx]*2 + 0];\n"
+	"	float4 bbMax = allpAABB[smallAabbMapping[unsorted_indx]*2 + 1];\n"
+	"	float4 pos;\n"
+	"	pos.x = (bbMin.x + bbMax.x) * 0.5f;\n"
+	"	pos.y = (bbMin.y + bbMax.y) * 0.5f;\n"
+	"	pos.z = (bbMin.z + bbMax.z) * 0.5f;\n"
+	"    // get address in grid\n"
+	"    int4 gridPosA = getGridPos(pos, pParams);\n"
+	"    int4 gridPosB; \n"
+	"    // examine only neighbouring cells\n"
+	"    for(int z=-1; z<=1; z++) \n"
+	"    {\n"
+	"		gridPosB.z = gridPosA.z + z;\n"
+	"        for(int y=-1; y<=1; y++) \n"
+	"        {\n"
+	"			gridPosB.y = gridPosA.y + y;\n"
+	"            for(int x=-1; x<=1; x++) \n"
+	"            {\n"
+	"				gridPosB.x = gridPosA.x + x;\n"
+	"                findPairsInCell(numObjects, gridPosB, index, pHash, pCellStart, allpAABB,smallAabbMapping, pParams, pairCount,pPairBuff2, maxPairs);\n"
+	"            }\n"
+	"        }\n"
+	"    }\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/kernels/parallelLinearBvhKernels.h b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/kernels/parallelLinearBvhKernels.h
index 5eb8f45b164..c02877dde9c 100644
--- a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/kernels/parallelLinearBvhKernels.h
+++ b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/kernels/parallelLinearBvhKernels.h
@@ -1,729 +1,728 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* parallelLinearBvhCL= \
-"/*\n"
-"This software is provided 'as-is', without any express or implied warranty.\n"
-"In no event will the authors be held liable for any damages arising from the use of this software.\n"
-"Permission is granted to anyone to use this software for any purpose,\n"
-"including commercial applications, and to alter it and redistribute it freely,\n"
-"subject to the following restrictions:\n"
-"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
-"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
-"3. This notice may not be removed or altered from any source distribution.\n"
-"*/\n"
-"//Initial Author Jackson Lee, 2014\n"
-"typedef float b3Scalar;\n"
-"typedef float4 b3Vector3;\n"
-"#define b3Max max\n"
-"#define b3Min min\n"
-"#define b3Sqrt sqrt\n"
-"typedef struct\n"
-"{\n"
-"	unsigned int m_key;\n"
-"	unsigned int m_value;\n"
-"} SortDataCL;\n"
-"typedef struct \n"
-"{\n"
-"	union\n"
-"	{\n"
-"		float4	m_min;\n"
-"		float   m_minElems[4];\n"
-"		int			m_minIndices[4];\n"
-"	};\n"
-"	union\n"
-"	{\n"
-"		float4	m_max;\n"
-"		float   m_maxElems[4];\n"
-"		int			m_maxIndices[4];\n"
-"	};\n"
-"} b3AabbCL;\n"
-"unsigned int interleaveBits(unsigned int x)\n"
-"{\n"
-"	//........ ........ ......12 3456789A	//x\n"
-"	//....1..2 ..3..4.. 5..6..7. .8..9..A	//x after interleaving bits\n"
-"	\n"
-"	//......12 3456789A ......12 3456789A	//x ^ (x << 16)\n"
-"	//11111111 ........ ........ 11111111	//0x FF 00 00 FF\n"
-"	//......12 ........ ........ 3456789A	//x = (x ^ (x << 16)) & 0xFF0000FF;\n"
-"	\n"
-"	//......12 ........ 3456789A 3456789A	//x ^ (x <<  8)\n"
-"	//......11 ........ 1111.... ....1111	//0x 03 00 F0 0F\n"
-"	//......12 ........ 3456.... ....789A	//x = (x ^ (x <<  8)) & 0x0300F00F;\n"
-"	\n"
-"	//..12..12 ....3456 3456.... 789A789A	//x ^ (x <<  4)\n"
-"	//......11 ....11.. ..11.... 11....11	//0x 03 0C 30 C3\n"
-"	//......12 ....34.. ..56.... 78....9A	//x = (x ^ (x <<  4)) & 0x030C30C3;\n"
-"	\n"
-"	//....1212 ..3434.. 5656..78 78..9A9A	//x ^ (x <<  2)\n"
-"	//....1..1 ..1..1.. 1..1..1. .1..1..1	//0x 09 24 92 49\n"
-"	//....1..2 ..3..4.. 5..6..7. .8..9..A	//x = (x ^ (x <<  2)) & 0x09249249;\n"
-"	\n"
-"	//........ ........ ......11 11111111	//0x000003FF\n"
-"	x &= 0x000003FF;		//Clear all bits above bit 10\n"
-"	\n"
-"	x = (x ^ (x << 16)) & 0xFF0000FF;\n"
-"	x = (x ^ (x <<  8)) & 0x0300F00F;\n"
-"	x = (x ^ (x <<  4)) & 0x030C30C3;\n"
-"	x = (x ^ (x <<  2)) & 0x09249249;\n"
-"	\n"
-"	return x;\n"
-"}\n"
-"unsigned int getMortonCode(unsigned int x, unsigned int y, unsigned int z)\n"
-"{\n"
-"	return interleaveBits(x) << 0 | interleaveBits(y) << 1 | interleaveBits(z) << 2;\n"
-"}\n"
-"__kernel void separateAabbs(__global b3AabbCL* unseparatedAabbs, __global int* aabbIndices, __global b3AabbCL* out_aabbs, int numAabbsToSeparate)\n"
-"{\n"
-"	int separatedAabbIndex = get_global_id(0);\n"
-"	if(separatedAabbIndex >= numAabbsToSeparate) return;\n"
-"	int unseparatedAabbIndex = aabbIndices[separatedAabbIndex];\n"
-"	out_aabbs[separatedAabbIndex] = unseparatedAabbs[unseparatedAabbIndex];\n"
-"}\n"
-"//Should replace with an optimized parallel reduction\n"
-"__kernel void findAllNodesMergedAabb(__global b3AabbCL* out_mergedAabb, int numAabbsNeedingMerge)\n"
-"{\n"
-"	//Each time this kernel is added to the command queue, \n"
-"	//the number of AABBs needing to be merged is halved\n"
-"	//\n"
-"	//Example with 159 AABBs:\n"
-"	//	numRemainingAabbs == 159 / 2 + 159 % 2 == 80\n"
-"	//	numMergedAabbs == 159 - 80 == 79\n"
-"	//So, indices [0, 78] are merged with [0 + 80, 78 + 80]\n"
-"	\n"
-"	int numRemainingAabbs = numAabbsNeedingMerge / 2 + numAabbsNeedingMerge % 2;\n"
-"	int numMergedAabbs = numAabbsNeedingMerge - numRemainingAabbs;\n"
-"	\n"
-"	int aabbIndex = get_global_id(0);\n"
-"	if(aabbIndex >= numMergedAabbs) return;\n"
-"	\n"
-"	int otherAabbIndex = aabbIndex + numRemainingAabbs;\n"
-"	\n"
-"	b3AabbCL aabb = out_mergedAabb[aabbIndex];\n"
-"	b3AabbCL otherAabb = out_mergedAabb[otherAabbIndex];\n"
-"		\n"
-"	b3AabbCL mergedAabb;\n"
-"	mergedAabb.m_min = b3Min(aabb.m_min, otherAabb.m_min);\n"
-"	mergedAabb.m_max = b3Max(aabb.m_max, otherAabb.m_max);\n"
-"	out_mergedAabb[aabbIndex] = mergedAabb;\n"
-"}\n"
-"__kernel void assignMortonCodesAndAabbIndicies(__global b3AabbCL* worldSpaceAabbs, __global b3AabbCL* mergedAabbOfAllNodes, \n"
-"												__global SortDataCL* out_mortonCodesAndAabbIndices, int numAabbs)\n"
-"{\n"
-"	int leafNodeIndex = get_global_id(0);	//Leaf node index == AABB index\n"
-"	if(leafNodeIndex >= numAabbs) return;\n"
-"	\n"
-"	b3AabbCL mergedAabb = mergedAabbOfAllNodes[0];\n"
-"	b3Vector3 gridCenter = (mergedAabb.m_min + mergedAabb.m_max) * 0.5f;\n"
-"	b3Vector3 gridCellSize = (mergedAabb.m_max - mergedAabb.m_min) / (float)1024;\n"
-"	\n"
-"	b3AabbCL aabb = worldSpaceAabbs[leafNodeIndex];\n"
-"	b3Vector3 aabbCenter = (aabb.m_min + aabb.m_max) * 0.5f;\n"
-"	b3Vector3 aabbCenterRelativeToGrid = aabbCenter - gridCenter;\n"
-"	\n"
-"	//Quantize into integer coordinates\n"
-"	//floor() is needed to prevent the center cell, at (0,0,0) from being twice the size\n"
-"	b3Vector3 gridPosition = aabbCenterRelativeToGrid / gridCellSize;\n"
-"	\n"
-"	int4 discretePosition;\n"
-"	discretePosition.x = (int)( (gridPosition.x >= 0.0f) ? gridPosition.x : floor(gridPosition.x) );\n"
-"	discretePosition.y = (int)( (gridPosition.y >= 0.0f) ? gridPosition.y : floor(gridPosition.y) );\n"
-"	discretePosition.z = (int)( (gridPosition.z >= 0.0f) ? gridPosition.z : floor(gridPosition.z) );\n"
-"	\n"
-"	//Clamp coordinates into [-512, 511], then convert range from [-512, 511] to [0, 1023]\n"
-"	discretePosition = b3Max( -512, b3Min(discretePosition, 511) );\n"
-"	discretePosition += 512;\n"
-"	\n"
-"	//Interleave bits(assign a morton code, also known as a z-curve)\n"
-"	unsigned int mortonCode = getMortonCode(discretePosition.x, discretePosition.y, discretePosition.z);\n"
-"	\n"
-"	//\n"
-"	SortDataCL mortonCodeIndexPair;\n"
-"	mortonCodeIndexPair.m_key = mortonCode;\n"
-"	mortonCodeIndexPair.m_value = leafNodeIndex;\n"
-"	\n"
-"	out_mortonCodesAndAabbIndices[leafNodeIndex] = mortonCodeIndexPair;\n"
-"}\n"
-"#define B3_PLVBH_TRAVERSE_MAX_STACK_SIZE 128\n"
-"//The most significant bit(0x80000000) of a int32 is used to distinguish between leaf and internal nodes.\n"
-"//If it is set, then the index is for an internal node; otherwise, it is a leaf node. \n"
-"//In both cases, the bit should be cleared to access the actual node index.\n"
-"int isLeafNode(int index) { return (index >> 31 == 0); }\n"
-"int getIndexWithInternalNodeMarkerRemoved(int index) { return index & (~0x80000000); }\n"
-"int getIndexWithInternalNodeMarkerSet(int isLeaf, int index) { return (isLeaf) ? index : (index | 0x80000000); }\n"
-"//From sap.cl\n"
-"#define NEW_PAIR_MARKER -1\n"
-"bool TestAabbAgainstAabb2(const b3AabbCL* aabb1, const b3AabbCL* aabb2)\n"
-"{\n"
-"	bool overlap = true;\n"
-"	overlap = (aabb1->m_min.x > aabb2->m_max.x || aabb1->m_max.x < aabb2->m_min.x) ? false : overlap;\n"
-"	overlap = (aabb1->m_min.z > aabb2->m_max.z || aabb1->m_max.z < aabb2->m_min.z) ? false : overlap;\n"
-"	overlap = (aabb1->m_min.y > aabb2->m_max.y || aabb1->m_max.y < aabb2->m_min.y) ? false : overlap;\n"
-"	return overlap;\n"
-"}\n"
-"//From sap.cl\n"
-"__kernel void plbvhCalculateOverlappingPairs(__global b3AabbCL* rigidAabbs, \n"
-"											__global int* rootNodeIndex, \n"
-"											__global int2* internalNodeChildIndices, \n"
-"											__global b3AabbCL* internalNodeAabbs,\n"
-"											__global int2* internalNodeLeafIndexRanges,\n"
-"											\n"
-"											__global SortDataCL* mortonCodesAndAabbIndices,\n"
-"											__global int* out_numPairs, __global int4* out_overlappingPairs, \n"
-"											int maxPairs, int numQueryAabbs)\n"
-"{\n"
-"	//Using get_group_id()/get_local_id() is Faster than get_global_id(0) since\n"
-"	//mortonCodesAndAabbIndices[] contains rigid body indices sorted along the z-curve (more spatially coherent)\n"
-"	int queryBvhNodeIndex = get_group_id(0) * get_local_size(0) + get_local_id(0);\n"
-"	if(queryBvhNodeIndex >= numQueryAabbs) return;\n"
-"	\n"
-"	int queryRigidIndex = mortonCodesAndAabbIndices[queryBvhNodeIndex].m_value;\n"
-"	b3AabbCL queryAabb = rigidAabbs[queryRigidIndex];\n"
-"	\n"
-"	int stack[B3_PLVBH_TRAVERSE_MAX_STACK_SIZE];\n"
-"	\n"
-"	int stackSize = 1;\n"
-"	stack[0] = *rootNodeIndex;\n"
-"	\n"
-"	while(stackSize)\n"
-"	{\n"
-"		int internalOrLeafNodeIndex = stack[ stackSize - 1 ];\n"
-"		--stackSize;\n"
-"		\n"
-"		int isLeaf = isLeafNode(internalOrLeafNodeIndex);	//Internal node if false\n"
-"		int bvhNodeIndex = getIndexWithInternalNodeMarkerRemoved(internalOrLeafNodeIndex);\n"
-"		\n"
-"		//Optimization - if the BVH is structured as a binary radix tree, then\n"
-"		//each internal node corresponds to a contiguous range of leaf nodes(internalNodeLeafIndexRanges[]).\n"
-"		//This can be used to avoid testing each AABB-AABB pair twice, including preventing each node from colliding with itself.\n"
-"		{\n"
-"			int highestLeafIndex = (isLeaf) ? bvhNodeIndex : internalNodeLeafIndexRanges[bvhNodeIndex].y;\n"
-"			if(highestLeafIndex <= queryBvhNodeIndex) continue;\n"
-"		}\n"
-"		\n"
-"		//bvhRigidIndex is not used if internal node\n"
-"		int bvhRigidIndex = (isLeaf) ? mortonCodesAndAabbIndices[bvhNodeIndex].m_value : -1;\n"
-"	\n"
-"		b3AabbCL bvhNodeAabb = (isLeaf) ? rigidAabbs[bvhRigidIndex] : internalNodeAabbs[bvhNodeIndex];\n"
-"		if( TestAabbAgainstAabb2(&queryAabb, &bvhNodeAabb) )\n"
-"		{\n"
-"			if(isLeaf)\n"
-"			{\n"
-"				int4 pair;\n"
-"				pair.x = rigidAabbs[queryRigidIndex].m_minIndices[3];\n"
-"				pair.y = rigidAabbs[bvhRigidIndex].m_minIndices[3];\n"
-"				pair.z = NEW_PAIR_MARKER;\n"
-"				pair.w = NEW_PAIR_MARKER;\n"
-"				\n"
-"				int pairIndex = atomic_inc(out_numPairs);\n"
-"				if(pairIndex < maxPairs) out_overlappingPairs[pairIndex] = pair;\n"
-"			}\n"
-"			\n"
-"			if(!isLeaf)	//Internal node\n"
-"			{\n"
-"				if(stackSize + 2 > B3_PLVBH_TRAVERSE_MAX_STACK_SIZE)\n"
-"				{\n"
-"					//Error\n"
-"				}\n"
-"				else\n"
-"				{\n"
-"					stack[ stackSize++ ] = internalNodeChildIndices[bvhNodeIndex].x;\n"
-"					stack[ stackSize++ ] = internalNodeChildIndices[bvhNodeIndex].y;\n"
-"				}\n"
-"			}\n"
-"		}\n"
-"		\n"
-"	}\n"
-"}\n"
-"//From rayCastKernels.cl\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_from;\n"
-"	float4 m_to;\n"
-"} b3RayInfo;\n"
-"//From rayCastKernels.cl\n"
-"b3Vector3 b3Vector3_normalize(b3Vector3 v)\n"
-"{\n"
-"	b3Vector3 normal = (b3Vector3){v.x, v.y, v.z, 0.f};\n"
-"	return normalize(normal);	//OpenCL normalize == vector4 normalize\n"
-"}\n"
-"b3Scalar b3Vector3_length2(b3Vector3 v) { return v.x*v.x + v.y*v.y + v.z*v.z; }\n"
-"b3Scalar b3Vector3_dot(b3Vector3 a, b3Vector3 b) { return a.x*b.x + a.y*b.y + a.z*b.z; }\n"
-"int rayIntersectsAabb(b3Vector3 rayOrigin, b3Scalar rayLength, b3Vector3 rayNormalizedDirection, b3AabbCL aabb)\n"
-"{\n"
-"	//AABB is considered as 3 pairs of 2 planes( {x_min, x_max}, {y_min, y_max}, {z_min, z_max} ).\n"
-"	//t_min is the point of intersection with the closer plane, t_max is the point of intersection with the farther plane.\n"
-"	//\n"
-"	//if (rayNormalizedDirection.x < 0.0f), then max.x will be the near plane \n"
-"	//and min.x will be the far plane; otherwise, it is reversed.\n"
-"	//\n"
-"	//In order for there to be a collision, the t_min and t_max of each pair must overlap.\n"
-"	//This can be tested for by selecting the highest t_min and lowest t_max and comparing them.\n"
-"	\n"
-"	int4 isNegative = isless( rayNormalizedDirection, ((b3Vector3){0.0f, 0.0f, 0.0f, 0.0f}) );	//isless(x,y) returns (x < y)\n"
-"	\n"
-"	//When using vector types, the select() function checks the most signficant bit, \n"
-"	//but isless() sets the least significant bit.\n"
-"	isNegative <<= 31;\n"
-"	//select(b, a, condition) == condition ? a : b\n"
-"	//When using select() with vector types, (condition[i]) is true if its most significant bit is 1\n"
-"	b3Vector3 t_min = ( select(aabb.m_min, aabb.m_max, isNegative) - rayOrigin ) / rayNormalizedDirection;\n"
-"	b3Vector3 t_max = ( select(aabb.m_max, aabb.m_min, isNegative) - rayOrigin ) / rayNormalizedDirection;\n"
-"	\n"
-"	b3Scalar t_min_final = 0.0f;\n"
-"	b3Scalar t_max_final = rayLength;\n"
-"	\n"
-"	//Must use fmin()/fmax(); if one of the parameters is NaN, then the parameter that is not NaN is returned. \n"
-"	//Behavior of min()/max() with NaNs is undefined. (See OpenCL Specification 1.2 [6.12.2] and [6.12.4])\n"
-"	//Since the innermost fmin()/fmax() is always not NaN, this should never return NaN.\n"
-"	t_min_final = fmax( t_min.z, fmax(t_min.y, fmax(t_min.x, t_min_final)) );\n"
-"	t_max_final = fmin( t_max.z, fmin(t_max.y, fmin(t_max.x, t_max_final)) );\n"
-"	\n"
-"	return (t_min_final <= t_max_final);\n"
-"}\n"
-"__kernel void plbvhRayTraverse(__global b3AabbCL* rigidAabbs,\n"
-"								__global int* rootNodeIndex, \n"
-"								__global int2* internalNodeChildIndices, \n"
-"								__global b3AabbCL* internalNodeAabbs,\n"
-"								__global int2* internalNodeLeafIndexRanges,\n"
-"								__global SortDataCL* mortonCodesAndAabbIndices,\n"
-"								\n"
-"								__global b3RayInfo* rays,\n"
-"								\n"
-"								__global int* out_numRayRigidPairs, \n"
-"								__global int2* out_rayRigidPairs,\n"
-"								int maxRayRigidPairs, int numRays)\n"
-"{\n"
-"	int rayIndex = get_global_id(0);\n"
-"	if(rayIndex >= numRays) return;\n"
-"	\n"
-"	//\n"
-"	b3Vector3 rayFrom = rays[rayIndex].m_from;\n"
-"	b3Vector3 rayTo = rays[rayIndex].m_to;\n"
-"	b3Vector3 rayNormalizedDirection = b3Vector3_normalize(rayTo - rayFrom);\n"
-"	b3Scalar rayLength = b3Sqrt( b3Vector3_length2(rayTo - rayFrom) );\n"
-"	\n"
-"	//\n"
-"	int stack[B3_PLVBH_TRAVERSE_MAX_STACK_SIZE];\n"
-"	\n"
-"	int stackSize = 1;\n"
-"	stack[0] = *rootNodeIndex;\n"
-"	\n"
-"	while(stackSize)\n"
-"	{\n"
-"		int internalOrLeafNodeIndex = stack[ stackSize - 1 ];\n"
-"		--stackSize;\n"
-"		\n"
-"		int isLeaf = isLeafNode(internalOrLeafNodeIndex);	//Internal node if false\n"
-"		int bvhNodeIndex = getIndexWithInternalNodeMarkerRemoved(internalOrLeafNodeIndex);\n"
-"		\n"
-"		//bvhRigidIndex is not used if internal node\n"
-"		int bvhRigidIndex = (isLeaf) ? mortonCodesAndAabbIndices[bvhNodeIndex].m_value : -1;\n"
-"	\n"
-"		b3AabbCL bvhNodeAabb = (isLeaf) ? rigidAabbs[bvhRigidIndex] : internalNodeAabbs[bvhNodeIndex];\n"
-"		if( rayIntersectsAabb(rayFrom, rayLength, rayNormalizedDirection, bvhNodeAabb)  )\n"
-"		{\n"
-"			if(isLeaf)\n"
-"			{\n"
-"				int2 rayRigidPair;\n"
-"				rayRigidPair.x = rayIndex;\n"
-"				rayRigidPair.y = rigidAabbs[bvhRigidIndex].m_minIndices[3];\n"
-"				\n"
-"				int pairIndex = atomic_inc(out_numRayRigidPairs);\n"
-"				if(pairIndex < maxRayRigidPairs) out_rayRigidPairs[pairIndex] = rayRigidPair;\n"
-"			}\n"
-"			\n"
-"			if(!isLeaf)	//Internal node\n"
-"			{\n"
-"				if(stackSize + 2 > B3_PLVBH_TRAVERSE_MAX_STACK_SIZE)\n"
-"				{\n"
-"					//Error\n"
-"				}\n"
-"				else\n"
-"				{\n"
-"					stack[ stackSize++ ] = internalNodeChildIndices[bvhNodeIndex].x;\n"
-"					stack[ stackSize++ ] = internalNodeChildIndices[bvhNodeIndex].y;\n"
-"				}\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"}\n"
-"__kernel void plbvhLargeAabbAabbTest(__global b3AabbCL* smallAabbs, __global b3AabbCL* largeAabbs, \n"
-"									__global int* out_numPairs, __global int4* out_overlappingPairs, \n"
-"									int maxPairs, int numLargeAabbRigids, int numSmallAabbRigids)\n"
-"{\n"
-"	int smallAabbIndex = get_global_id(0);\n"
-"	if(smallAabbIndex >= numSmallAabbRigids) return;\n"
-"	\n"
-"	b3AabbCL smallAabb = smallAabbs[smallAabbIndex];\n"
-"	for(int i = 0; i < numLargeAabbRigids; ++i)\n"
-"	{\n"
-"		b3AabbCL largeAabb = largeAabbs[i];\n"
-"		if( TestAabbAgainstAabb2(&smallAabb, &largeAabb) )\n"
-"		{\n"
-"			int4 pair;\n"
-"			pair.x = largeAabb.m_minIndices[3];\n"
-"			pair.y = smallAabb.m_minIndices[3];\n"
-"			pair.z = NEW_PAIR_MARKER;\n"
-"			pair.w = NEW_PAIR_MARKER;\n"
-"			\n"
-"			int pairIndex = atomic_inc(out_numPairs);\n"
-"			if(pairIndex < maxPairs) out_overlappingPairs[pairIndex] = pair;\n"
-"		}\n"
-"	}\n"
-"}\n"
-"__kernel void plbvhLargeAabbRayTest(__global b3AabbCL* largeRigidAabbs, __global b3RayInfo* rays,\n"
-"									__global int* out_numRayRigidPairs,  __global int2* out_rayRigidPairs,\n"
-"									int numLargeAabbRigids, int maxRayRigidPairs, int numRays)\n"
-"{\n"
-"	int rayIndex = get_global_id(0);\n"
-"	if(rayIndex >= numRays) return;\n"
-"	\n"
-"	b3Vector3 rayFrom = rays[rayIndex].m_from;\n"
-"	b3Vector3 rayTo = rays[rayIndex].m_to;\n"
-"	b3Vector3 rayNormalizedDirection = b3Vector3_normalize(rayTo - rayFrom);\n"
-"	b3Scalar rayLength = b3Sqrt( b3Vector3_length2(rayTo - rayFrom) );\n"
-"	\n"
-"	for(int i = 0; i < numLargeAabbRigids; ++i)\n"
-"	{\n"
-"		b3AabbCL rigidAabb = largeRigidAabbs[i];\n"
-"		if( rayIntersectsAabb(rayFrom, rayLength, rayNormalizedDirection, rigidAabb) )\n"
-"		{\n"
-"			int2 rayRigidPair;\n"
-"			rayRigidPair.x = rayIndex;\n"
-"			rayRigidPair.y = rigidAabb.m_minIndices[3];\n"
-"			\n"
-"			int pairIndex = atomic_inc(out_numRayRigidPairs);\n"
-"			if(pairIndex < maxRayRigidPairs) out_rayRigidPairs[pairIndex] = rayRigidPair;\n"
-"		}\n"
-"	}\n"
-"}\n"
-"//Set so that it is always greater than the actual common prefixes, and never selected as a parent node.\n"
-"//If there are no duplicates, then the highest common prefix is 32 or 64, depending on the number of bits used for the z-curve.\n"
-"//Duplicate common prefixes increase the highest common prefix at most by the number of bits used to index the leaf node.\n"
-"//Since 32 bit ints are used to index leaf nodes, the max prefix is 64(32 + 32 bit z-curve) or 96(32 + 64 bit z-curve).\n"
-"#define B3_PLBVH_INVALID_COMMON_PREFIX 128\n"
-"#define B3_PLBVH_ROOT_NODE_MARKER -1\n"
-"#define b3Int64 long\n"
-"int computeCommonPrefixLength(b3Int64 i, b3Int64 j) { return (int)clz(i ^ j); }\n"
-"b3Int64 computeCommonPrefix(b3Int64 i, b3Int64 j) \n"
-"{\n"
-"	//This function only needs to return (i & j) in order for the algorithm to work,\n"
-"	//but it may help with debugging to mask out the lower bits.\n"
-"	b3Int64 commonPrefixLength = (b3Int64)computeCommonPrefixLength(i, j);\n"
-"	b3Int64 sharedBits = i & j;\n"
-"	b3Int64 bitmask = ((b3Int64)(~0)) << (64 - commonPrefixLength);	//Set all bits after the common prefix to 0\n"
-"	\n"
-"	return sharedBits & bitmask;\n"
-"}\n"
-"//Same as computeCommonPrefixLength(), but allows for prefixes with different lengths\n"
-"int getSharedPrefixLength(b3Int64 prefixA, int prefixLengthA, b3Int64 prefixB, int prefixLengthB)\n"
-"{\n"
-"	return b3Min( computeCommonPrefixLength(prefixA, prefixB), b3Min(prefixLengthA, prefixLengthB) );\n"
-"}\n"
-"__kernel void computeAdjacentPairCommonPrefix(__global SortDataCL* mortonCodesAndAabbIndices,\n"
-"											__global b3Int64* out_commonPrefixes,\n"
-"											__global int* out_commonPrefixLengths,\n"
-"											int numInternalNodes)\n"
-"{\n"
-"	int internalNodeIndex = get_global_id(0);\n"
-"	if (internalNodeIndex >= numInternalNodes) return;\n"
-"	\n"
-"	//Here, (internalNodeIndex + 1) is never out of bounds since it is a leaf node index,\n"
-"	//and the number of internal nodes is always numLeafNodes - 1\n"
-"	int leftLeafIndex = internalNodeIndex;\n"
-"	int rightLeafIndex = internalNodeIndex + 1;\n"
-"	\n"
-"	int leftLeafMortonCode = mortonCodesAndAabbIndices[leftLeafIndex].m_key;\n"
-"	int rightLeafMortonCode = mortonCodesAndAabbIndices[rightLeafIndex].m_key;\n"
-"	\n"
-"	//Binary radix tree construction algorithm does not work if there are duplicate morton codes.\n"
-"	//Append the index of each leaf node to each morton code so that there are no duplicates.\n"
-"	//The algorithm also requires that the morton codes are sorted in ascending order; this requirement\n"
-"	//is also satisfied with this method, as (leftLeafIndex < rightLeafIndex) is always true.\n"
-"	//\n"
-"	//upsample(a, b) == ( ((b3Int64)a) << 32) | b\n"
-"	b3Int64 nonduplicateLeftMortonCode = upsample(leftLeafMortonCode, leftLeafIndex);\n"
-"	b3Int64 nonduplicateRightMortonCode = upsample(rightLeafMortonCode, rightLeafIndex);\n"
-"	\n"
-"	out_commonPrefixes[internalNodeIndex] = computeCommonPrefix(nonduplicateLeftMortonCode, nonduplicateRightMortonCode);\n"
-"	out_commonPrefixLengths[internalNodeIndex] = computeCommonPrefixLength(nonduplicateLeftMortonCode, nonduplicateRightMortonCode);\n"
-"}\n"
-"__kernel void buildBinaryRadixTreeLeafNodes(__global int* commonPrefixLengths, __global int* out_leafNodeParentNodes,\n"
-"											__global int2* out_childNodes, int numLeafNodes)\n"
-"{\n"
-"	int leafNodeIndex = get_global_id(0);\n"
-"	if (leafNodeIndex >= numLeafNodes) return;\n"
-"	\n"
-"	int numInternalNodes = numLeafNodes - 1;\n"
-"	\n"
-"	int leftSplitIndex = leafNodeIndex - 1;\n"
-"	int rightSplitIndex = leafNodeIndex;\n"
-"	\n"
-"	int leftCommonPrefix = (leftSplitIndex >= 0) ? commonPrefixLengths[leftSplitIndex] : B3_PLBVH_INVALID_COMMON_PREFIX;\n"
-"	int rightCommonPrefix = (rightSplitIndex < numInternalNodes) ? commonPrefixLengths[rightSplitIndex] : B3_PLBVH_INVALID_COMMON_PREFIX;\n"
-"	\n"
-"	//Parent node is the highest adjacent common prefix that is lower than the node's common prefix\n"
-"	//Leaf nodes are considered as having the highest common prefix\n"
-"	int isLeftHigherCommonPrefix = (leftCommonPrefix > rightCommonPrefix);\n"
-"	\n"
-"	//Handle cases for the edge nodes; the first and last node\n"
-"	//For leaf nodes, leftCommonPrefix and rightCommonPrefix should never both be B3_PLBVH_INVALID_COMMON_PREFIX\n"
-"	if(leftCommonPrefix == B3_PLBVH_INVALID_COMMON_PREFIX) isLeftHigherCommonPrefix = false;\n"
-"	if(rightCommonPrefix == B3_PLBVH_INVALID_COMMON_PREFIX) isLeftHigherCommonPrefix = true;\n"
-"	\n"
-"	int parentNodeIndex = (isLeftHigherCommonPrefix) ? leftSplitIndex : rightSplitIndex;\n"
-"	out_leafNodeParentNodes[leafNodeIndex] = parentNodeIndex;\n"
-"	\n"
-"	int isRightChild = (isLeftHigherCommonPrefix);	//If the left node is the parent, then this node is its right child and vice versa\n"
-"	\n"
-"	//out_childNodesAsInt[0] == int2.x == left child\n"
-"	//out_childNodesAsInt[1] == int2.y == right child\n"
-"	int isLeaf = 1;\n"
-"	__global int* out_childNodesAsInt = (__global int*)(&out_childNodes[parentNodeIndex]);\n"
-"	out_childNodesAsInt[isRightChild] = getIndexWithInternalNodeMarkerSet(isLeaf, leafNodeIndex);\n"
-"}\n"
-"__kernel void buildBinaryRadixTreeInternalNodes(__global b3Int64* commonPrefixes, __global int* commonPrefixLengths,\n"
-"												__global int2* out_childNodes,\n"
-"												__global int* out_internalNodeParentNodes, __global int* out_rootNodeIndex,\n"
-"												int numInternalNodes)\n"
-"{\n"
-"	int internalNodeIndex = get_group_id(0) * get_local_size(0) + get_local_id(0);\n"
-"	if(internalNodeIndex >= numInternalNodes) return;\n"
-"	\n"
-"	b3Int64 nodePrefix = commonPrefixes[internalNodeIndex];\n"
-"	int nodePrefixLength = commonPrefixLengths[internalNodeIndex];\n"
-"	\n"
-"//#define USE_LINEAR_SEARCH\n"
-"#ifdef USE_LINEAR_SEARCH\n"
-"	int leftIndex = -1;\n"
-"	int rightIndex = -1;\n"
-"	\n"
-"	//Find nearest element to left with a lower common prefix\n"
-"	for(int i = internalNodeIndex - 1; i >= 0; --i)\n"
-"	{\n"
-"		int nodeLeftSharedPrefixLength = getSharedPrefixLength(nodePrefix, nodePrefixLength, commonPrefixes[i], commonPrefixLengths[i]);\n"
-"		if(nodeLeftSharedPrefixLength < nodePrefixLength)\n"
-"		{\n"
-"			leftIndex = i;\n"
-"			break;\n"
-"		}\n"
-"	}\n"
-"	\n"
-"	//Find nearest element to right with a lower common prefix\n"
-"	for(int i = internalNodeIndex + 1; i < numInternalNodes; ++i)\n"
-"	{\n"
-"		int nodeRightSharedPrefixLength = getSharedPrefixLength(nodePrefix, nodePrefixLength, commonPrefixes[i], commonPrefixLengths[i]);\n"
-"		if(nodeRightSharedPrefixLength < nodePrefixLength)\n"
-"		{\n"
-"			rightIndex = i;\n"
-"			break;\n"
-"		}\n"
-"	}\n"
-"	\n"
-"#else //Use binary search\n"
-"	//Find nearest element to left with a lower common prefix\n"
-"	int leftIndex = -1;\n"
-"	{\n"
-"		int lower = 0;\n"
-"		int upper = internalNodeIndex - 1;\n"
-"		\n"
-"		while(lower <= upper)\n"
-"		{\n"
-"			int mid = (lower + upper) / 2;\n"
-"			b3Int64 midPrefix = commonPrefixes[mid];\n"
-"			int midPrefixLength = commonPrefixLengths[mid];\n"
-"			\n"
-"			int nodeMidSharedPrefixLength = getSharedPrefixLength(nodePrefix, nodePrefixLength, midPrefix, midPrefixLength);\n"
-"			if(nodeMidSharedPrefixLength < nodePrefixLength) \n"
-"			{\n"
-"				int right = mid + 1;\n"
-"				if(right < internalNodeIndex)\n"
-"				{\n"
-"					b3Int64 rightPrefix = commonPrefixes[right];\n"
-"					int rightPrefixLength = commonPrefixLengths[right];\n"
-"					\n"
-"					int nodeRightSharedPrefixLength = getSharedPrefixLength(nodePrefix, nodePrefixLength, rightPrefix, rightPrefixLength);\n"
-"					if(nodeRightSharedPrefixLength < nodePrefixLength) \n"
-"					{\n"
-"						lower = right;\n"
-"						leftIndex = right;\n"
-"					}\n"
-"					else \n"
-"					{\n"
-"						leftIndex = mid;\n"
-"						break;\n"
-"					}\n"
-"				}\n"
-"				else \n"
-"				{\n"
-"					leftIndex = mid;\n"
-"					break;\n"
-"				}\n"
-"			}\n"
-"			else upper = mid - 1;\n"
-"		}\n"
-"	}\n"
-"	\n"
-"	//Find nearest element to right with a lower common prefix\n"
-"	int rightIndex = -1;\n"
-"	{\n"
-"		int lower = internalNodeIndex + 1;\n"
-"		int upper = numInternalNodes - 1;\n"
-"		\n"
-"		while(lower <= upper)\n"
-"		{\n"
-"			int mid = (lower + upper) / 2;\n"
-"			b3Int64 midPrefix = commonPrefixes[mid];\n"
-"			int midPrefixLength = commonPrefixLengths[mid];\n"
-"			\n"
-"			int nodeMidSharedPrefixLength = getSharedPrefixLength(nodePrefix, nodePrefixLength, midPrefix, midPrefixLength);\n"
-"			if(nodeMidSharedPrefixLength < nodePrefixLength) \n"
-"			{\n"
-"				int left = mid - 1;\n"
-"				if(left > internalNodeIndex)\n"
-"				{\n"
-"					b3Int64 leftPrefix = commonPrefixes[left];\n"
-"					int leftPrefixLength = commonPrefixLengths[left];\n"
-"				\n"
-"					int nodeLeftSharedPrefixLength = getSharedPrefixLength(nodePrefix, nodePrefixLength, leftPrefix, leftPrefixLength);\n"
-"					if(nodeLeftSharedPrefixLength < nodePrefixLength) \n"
-"					{\n"
-"						upper = left;\n"
-"						rightIndex = left;\n"
-"					}\n"
-"					else \n"
-"					{\n"
-"						rightIndex = mid;\n"
-"						break;\n"
-"					}\n"
-"				}\n"
-"				else \n"
-"				{\n"
-"					rightIndex = mid;\n"
-"					break;\n"
-"				}\n"
-"			}\n"
-"			else lower = mid + 1;\n"
-"		}\n"
-"	}\n"
-"#endif\n"
-"	\n"
-"	//Select parent\n"
-"	{\n"
-"		int leftPrefixLength = (leftIndex != -1) ? commonPrefixLengths[leftIndex] : B3_PLBVH_INVALID_COMMON_PREFIX;\n"
-"		int rightPrefixLength =  (rightIndex != -1) ? commonPrefixLengths[rightIndex] : B3_PLBVH_INVALID_COMMON_PREFIX;\n"
-"		\n"
-"		int isLeftHigherPrefixLength = (leftPrefixLength > rightPrefixLength);\n"
-"		\n"
-"		if(leftPrefixLength == B3_PLBVH_INVALID_COMMON_PREFIX) isLeftHigherPrefixLength = false;\n"
-"		else if(rightPrefixLength == B3_PLBVH_INVALID_COMMON_PREFIX) isLeftHigherPrefixLength = true;\n"
-"		\n"
-"		int parentNodeIndex = (isLeftHigherPrefixLength) ? leftIndex : rightIndex;\n"
-"		\n"
-"		int isRootNode = (leftIndex == -1 && rightIndex == -1);\n"
-"		out_internalNodeParentNodes[internalNodeIndex] = (!isRootNode) ? parentNodeIndex : B3_PLBVH_ROOT_NODE_MARKER;\n"
-"		\n"
-"		int isLeaf = 0;\n"
-"		if(!isRootNode)\n"
-"		{\n"
-"			int isRightChild = (isLeftHigherPrefixLength);	//If the left node is the parent, then this node is its right child and vice versa\n"
-"			\n"
-"			//out_childNodesAsInt[0] == int2.x == left child\n"
-"			//out_childNodesAsInt[1] == int2.y == right child\n"
-"			__global int* out_childNodesAsInt = (__global int*)(&out_childNodes[parentNodeIndex]);\n"
-"			out_childNodesAsInt[isRightChild] = getIndexWithInternalNodeMarkerSet(isLeaf, internalNodeIndex);\n"
-"		}\n"
-"		else *out_rootNodeIndex = getIndexWithInternalNodeMarkerSet(isLeaf, internalNodeIndex);\n"
-"	}\n"
-"}\n"
-"__kernel void findDistanceFromRoot(__global int* rootNodeIndex, __global int* internalNodeParentNodes,\n"
-"									__global int* out_maxDistanceFromRoot, __global int* out_distanceFromRoot, int numInternalNodes)\n"
-"{\n"
-"	if( get_global_id(0) == 0 ) atomic_xchg(out_maxDistanceFromRoot, 0);\n"
-"	int internalNodeIndex = get_global_id(0);\n"
-"	if(internalNodeIndex >= numInternalNodes) return;\n"
-"	\n"
-"	//\n"
-"	int distanceFromRoot = 0;\n"
-"	{\n"
-"		int parentIndex = internalNodeParentNodes[internalNodeIndex];\n"
-"		while(parentIndex != B3_PLBVH_ROOT_NODE_MARKER)\n"
-"		{\n"
-"			parentIndex = internalNodeParentNodes[parentIndex];\n"
-"			++distanceFromRoot;\n"
-"		}\n"
-"	}\n"
-"	out_distanceFromRoot[internalNodeIndex] = distanceFromRoot;\n"
-"	\n"
-"	//\n"
-"	__local int localMaxDistanceFromRoot;\n"
-"	if( get_local_id(0) == 0 ) localMaxDistanceFromRoot = 0;\n"
-"	barrier(CLK_LOCAL_MEM_FENCE);\n"
-"	\n"
-"	atomic_max(&localMaxDistanceFromRoot, distanceFromRoot);\n"
-"	barrier(CLK_LOCAL_MEM_FENCE);\n"
-"	\n"
-"	if( get_local_id(0) == 0 ) atomic_max(out_maxDistanceFromRoot, localMaxDistanceFromRoot);\n"
-"}\n"
-"__kernel void buildBinaryRadixTreeAabbsRecursive(__global int* distanceFromRoot, __global SortDataCL* mortonCodesAndAabbIndices,\n"
-"												__global int2* childNodes,\n"
-"												__global b3AabbCL* leafNodeAabbs, __global b3AabbCL* internalNodeAabbs,\n"
-"												int maxDistanceFromRoot, int processedDistance, int numInternalNodes)\n"
-"{\n"
-"	int internalNodeIndex = get_global_id(0);\n"
-"	if(internalNodeIndex >= numInternalNodes) return;\n"
-"	\n"
-"	int distance = distanceFromRoot[internalNodeIndex];\n"
-"	\n"
-"	if(distance == processedDistance)\n"
-"	{\n"
-"		int leftChildIndex = childNodes[internalNodeIndex].x;\n"
-"		int rightChildIndex = childNodes[internalNodeIndex].y;\n"
-"		\n"
-"		int isLeftChildLeaf = isLeafNode(leftChildIndex);\n"
-"		int isRightChildLeaf = isLeafNode(rightChildIndex);\n"
-"		\n"
-"		leftChildIndex = getIndexWithInternalNodeMarkerRemoved(leftChildIndex);\n"
-"		rightChildIndex = getIndexWithInternalNodeMarkerRemoved(rightChildIndex);\n"
-"		\n"
-"		//leftRigidIndex/rightRigidIndex is not used if internal node\n"
-"		int leftRigidIndex = (isLeftChildLeaf) ? mortonCodesAndAabbIndices[leftChildIndex].m_value : -1;\n"
-"		int rightRigidIndex = (isRightChildLeaf) ? mortonCodesAndAabbIndices[rightChildIndex].m_value : -1;\n"
-"		\n"
-"		b3AabbCL leftChildAabb = (isLeftChildLeaf) ? leafNodeAabbs[leftRigidIndex] : internalNodeAabbs[leftChildIndex];\n"
-"		b3AabbCL rightChildAabb = (isRightChildLeaf) ? leafNodeAabbs[rightRigidIndex] : internalNodeAabbs[rightChildIndex];\n"
-"		\n"
-"		b3AabbCL mergedAabb;\n"
-"		mergedAabb.m_min = b3Min(leftChildAabb.m_min, rightChildAabb.m_min);\n"
-"		mergedAabb.m_max = b3Max(leftChildAabb.m_max, rightChildAabb.m_max);\n"
-"		internalNodeAabbs[internalNodeIndex] = mergedAabb;\n"
-"	}\n"
-"}\n"
-"__kernel void findLeafIndexRanges(__global int2* internalNodeChildNodes, __global int2* out_leafIndexRanges, int numInternalNodes)\n"
-"{\n"
-"	int internalNodeIndex = get_global_id(0);\n"
-"	if(internalNodeIndex >= numInternalNodes) return;\n"
-"	\n"
-"	int numLeafNodes = numInternalNodes + 1;\n"
-"	\n"
-"	int2 childNodes = internalNodeChildNodes[internalNodeIndex];\n"
-"	\n"
-"	int2 leafIndexRange;	//x == min leaf index, y == max leaf index\n"
-"	\n"
-"	//Find lowest leaf index covered by this internal node\n"
-"	{\n"
-"		int lowestIndex = childNodes.x;		//childNodes.x == Left child\n"
-"		while( !isLeafNode(lowestIndex) ) lowestIndex = internalNodeChildNodes[ getIndexWithInternalNodeMarkerRemoved(lowestIndex) ].x;\n"
-"		leafIndexRange.x = lowestIndex;\n"
-"	}\n"
-"	\n"
-"	//Find highest leaf index covered by this internal node\n"
-"	{\n"
-"		int highestIndex = childNodes.y;	//childNodes.y == Right child\n"
-"		while( !isLeafNode(highestIndex) ) highestIndex = internalNodeChildNodes[ getIndexWithInternalNodeMarkerRemoved(highestIndex) ].y;\n"
-"		leafIndexRange.y = highestIndex;\n"
-"	}\n"
-"	\n"
-"	//\n"
-"	out_leafIndexRanges[internalNodeIndex] = leafIndexRange;\n"
-"}\n"
-;
+static const char* parallelLinearBvhCL =
+	"/*\n"
+	"This software is provided 'as-is', without any express or implied warranty.\n"
+	"In no event will the authors be held liable for any damages arising from the use of this software.\n"
+	"Permission is granted to anyone to use this software for any purpose,\n"
+	"including commercial applications, and to alter it and redistribute it freely,\n"
+	"subject to the following restrictions:\n"
+	"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
+	"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
+	"3. This notice may not be removed or altered from any source distribution.\n"
+	"*/\n"
+	"//Initial Author Jackson Lee, 2014\n"
+	"typedef float b3Scalar;\n"
+	"typedef float4 b3Vector3;\n"
+	"#define b3Max max\n"
+	"#define b3Min min\n"
+	"#define b3Sqrt sqrt\n"
+	"typedef struct\n"
+	"{\n"
+	"	unsigned int m_key;\n"
+	"	unsigned int m_value;\n"
+	"} SortDataCL;\n"
+	"typedef struct \n"
+	"{\n"
+	"	union\n"
+	"	{\n"
+	"		float4	m_min;\n"
+	"		float   m_minElems[4];\n"
+	"		int			m_minIndices[4];\n"
+	"	};\n"
+	"	union\n"
+	"	{\n"
+	"		float4	m_max;\n"
+	"		float   m_maxElems[4];\n"
+	"		int			m_maxIndices[4];\n"
+	"	};\n"
+	"} b3AabbCL;\n"
+	"unsigned int interleaveBits(unsigned int x)\n"
+	"{\n"
+	"	//........ ........ ......12 3456789A	//x\n"
+	"	//....1..2 ..3..4.. 5..6..7. .8..9..A	//x after interleaving bits\n"
+	"	\n"
+	"	//......12 3456789A ......12 3456789A	//x ^ (x << 16)\n"
+	"	//11111111 ........ ........ 11111111	//0x FF 00 00 FF\n"
+	"	//......12 ........ ........ 3456789A	//x = (x ^ (x << 16)) & 0xFF0000FF;\n"
+	"	\n"
+	"	//......12 ........ 3456789A 3456789A	//x ^ (x <<  8)\n"
+	"	//......11 ........ 1111.... ....1111	//0x 03 00 F0 0F\n"
+	"	//......12 ........ 3456.... ....789A	//x = (x ^ (x <<  8)) & 0x0300F00F;\n"
+	"	\n"
+	"	//..12..12 ....3456 3456.... 789A789A	//x ^ (x <<  4)\n"
+	"	//......11 ....11.. ..11.... 11....11	//0x 03 0C 30 C3\n"
+	"	//......12 ....34.. ..56.... 78....9A	//x = (x ^ (x <<  4)) & 0x030C30C3;\n"
+	"	\n"
+	"	//....1212 ..3434.. 5656..78 78..9A9A	//x ^ (x <<  2)\n"
+	"	//....1..1 ..1..1.. 1..1..1. .1..1..1	//0x 09 24 92 49\n"
+	"	//....1..2 ..3..4.. 5..6..7. .8..9..A	//x = (x ^ (x <<  2)) & 0x09249249;\n"
+	"	\n"
+	"	//........ ........ ......11 11111111	//0x000003FF\n"
+	"	x &= 0x000003FF;		//Clear all bits above bit 10\n"
+	"	\n"
+	"	x = (x ^ (x << 16)) & 0xFF0000FF;\n"
+	"	x = (x ^ (x <<  8)) & 0x0300F00F;\n"
+	"	x = (x ^ (x <<  4)) & 0x030C30C3;\n"
+	"	x = (x ^ (x <<  2)) & 0x09249249;\n"
+	"	\n"
+	"	return x;\n"
+	"}\n"
+	"unsigned int getMortonCode(unsigned int x, unsigned int y, unsigned int z)\n"
+	"{\n"
+	"	return interleaveBits(x) << 0 | interleaveBits(y) << 1 | interleaveBits(z) << 2;\n"
+	"}\n"
+	"__kernel void separateAabbs(__global b3AabbCL* unseparatedAabbs, __global int* aabbIndices, __global b3AabbCL* out_aabbs, int numAabbsToSeparate)\n"
+	"{\n"
+	"	int separatedAabbIndex = get_global_id(0);\n"
+	"	if(separatedAabbIndex >= numAabbsToSeparate) return;\n"
+	"	int unseparatedAabbIndex = aabbIndices[separatedAabbIndex];\n"
+	"	out_aabbs[separatedAabbIndex] = unseparatedAabbs[unseparatedAabbIndex];\n"
+	"}\n"
+	"//Should replace with an optimized parallel reduction\n"
+	"__kernel void findAllNodesMergedAabb(__global b3AabbCL* out_mergedAabb, int numAabbsNeedingMerge)\n"
+	"{\n"
+	"	//Each time this kernel is added to the command queue, \n"
+	"	//the number of AABBs needing to be merged is halved\n"
+	"	//\n"
+	"	//Example with 159 AABBs:\n"
+	"	//	numRemainingAabbs == 159 / 2 + 159 % 2 == 80\n"
+	"	//	numMergedAabbs == 159 - 80 == 79\n"
+	"	//So, indices [0, 78] are merged with [0 + 80, 78 + 80]\n"
+	"	\n"
+	"	int numRemainingAabbs = numAabbsNeedingMerge / 2 + numAabbsNeedingMerge % 2;\n"
+	"	int numMergedAabbs = numAabbsNeedingMerge - numRemainingAabbs;\n"
+	"	\n"
+	"	int aabbIndex = get_global_id(0);\n"
+	"	if(aabbIndex >= numMergedAabbs) return;\n"
+	"	\n"
+	"	int otherAabbIndex = aabbIndex + numRemainingAabbs;\n"
+	"	\n"
+	"	b3AabbCL aabb = out_mergedAabb[aabbIndex];\n"
+	"	b3AabbCL otherAabb = out_mergedAabb[otherAabbIndex];\n"
+	"		\n"
+	"	b3AabbCL mergedAabb;\n"
+	"	mergedAabb.m_min = b3Min(aabb.m_min, otherAabb.m_min);\n"
+	"	mergedAabb.m_max = b3Max(aabb.m_max, otherAabb.m_max);\n"
+	"	out_mergedAabb[aabbIndex] = mergedAabb;\n"
+	"}\n"
+	"__kernel void assignMortonCodesAndAabbIndicies(__global b3AabbCL* worldSpaceAabbs, __global b3AabbCL* mergedAabbOfAllNodes, \n"
+	"												__global SortDataCL* out_mortonCodesAndAabbIndices, int numAabbs)\n"
+	"{\n"
+	"	int leafNodeIndex = get_global_id(0);	//Leaf node index == AABB index\n"
+	"	if(leafNodeIndex >= numAabbs) return;\n"
+	"	\n"
+	"	b3AabbCL mergedAabb = mergedAabbOfAllNodes[0];\n"
+	"	b3Vector3 gridCenter = (mergedAabb.m_min + mergedAabb.m_max) * 0.5f;\n"
+	"	b3Vector3 gridCellSize = (mergedAabb.m_max - mergedAabb.m_min) / (float)1024;\n"
+	"	\n"
+	"	b3AabbCL aabb = worldSpaceAabbs[leafNodeIndex];\n"
+	"	b3Vector3 aabbCenter = (aabb.m_min + aabb.m_max) * 0.5f;\n"
+	"	b3Vector3 aabbCenterRelativeToGrid = aabbCenter - gridCenter;\n"
+	"	\n"
+	"	//Quantize into integer coordinates\n"
+	"	//floor() is needed to prevent the center cell, at (0,0,0) from being twice the size\n"
+	"	b3Vector3 gridPosition = aabbCenterRelativeToGrid / gridCellSize;\n"
+	"	\n"
+	"	int4 discretePosition;\n"
+	"	discretePosition.x = (int)( (gridPosition.x >= 0.0f) ? gridPosition.x : floor(gridPosition.x) );\n"
+	"	discretePosition.y = (int)( (gridPosition.y >= 0.0f) ? gridPosition.y : floor(gridPosition.y) );\n"
+	"	discretePosition.z = (int)( (gridPosition.z >= 0.0f) ? gridPosition.z : floor(gridPosition.z) );\n"
+	"	\n"
+	"	//Clamp coordinates into [-512, 511], then convert range from [-512, 511] to [0, 1023]\n"
+	"	discretePosition = b3Max( -512, b3Min(discretePosition, 511) );\n"
+	"	discretePosition += 512;\n"
+	"	\n"
+	"	//Interleave bits(assign a morton code, also known as a z-curve)\n"
+	"	unsigned int mortonCode = getMortonCode(discretePosition.x, discretePosition.y, discretePosition.z);\n"
+	"	\n"
+	"	//\n"
+	"	SortDataCL mortonCodeIndexPair;\n"
+	"	mortonCodeIndexPair.m_key = mortonCode;\n"
+	"	mortonCodeIndexPair.m_value = leafNodeIndex;\n"
+	"	\n"
+	"	out_mortonCodesAndAabbIndices[leafNodeIndex] = mortonCodeIndexPair;\n"
+	"}\n"
+	"#define B3_PLVBH_TRAVERSE_MAX_STACK_SIZE 128\n"
+	"//The most significant bit(0x80000000) of a int32 is used to distinguish between leaf and internal nodes.\n"
+	"//If it is set, then the index is for an internal node; otherwise, it is a leaf node. \n"
+	"//In both cases, the bit should be cleared to access the actual node index.\n"
+	"int isLeafNode(int index) { return (index >> 31 == 0); }\n"
+	"int getIndexWithInternalNodeMarkerRemoved(int index) { return index & (~0x80000000); }\n"
+	"int getIndexWithInternalNodeMarkerSet(int isLeaf, int index) { return (isLeaf) ? index : (index | 0x80000000); }\n"
+	"//From sap.cl\n"
+	"#define NEW_PAIR_MARKER -1\n"
+	"bool TestAabbAgainstAabb2(const b3AabbCL* aabb1, const b3AabbCL* aabb2)\n"
+	"{\n"
+	"	bool overlap = true;\n"
+	"	overlap = (aabb1->m_min.x > aabb2->m_max.x || aabb1->m_max.x < aabb2->m_min.x) ? false : overlap;\n"
+	"	overlap = (aabb1->m_min.z > aabb2->m_max.z || aabb1->m_max.z < aabb2->m_min.z) ? false : overlap;\n"
+	"	overlap = (aabb1->m_min.y > aabb2->m_max.y || aabb1->m_max.y < aabb2->m_min.y) ? false : overlap;\n"
+	"	return overlap;\n"
+	"}\n"
+	"//From sap.cl\n"
+	"__kernel void plbvhCalculateOverlappingPairs(__global b3AabbCL* rigidAabbs, \n"
+	"											__global int* rootNodeIndex, \n"
+	"											__global int2* internalNodeChildIndices, \n"
+	"											__global b3AabbCL* internalNodeAabbs,\n"
+	"											__global int2* internalNodeLeafIndexRanges,\n"
+	"											\n"
+	"											__global SortDataCL* mortonCodesAndAabbIndices,\n"
+	"											__global int* out_numPairs, __global int4* out_overlappingPairs, \n"
+	"											int maxPairs, int numQueryAabbs)\n"
+	"{\n"
+	"	//Using get_group_id()/get_local_id() is Faster than get_global_id(0) since\n"
+	"	//mortonCodesAndAabbIndices[] contains rigid body indices sorted along the z-curve (more spatially coherent)\n"
+	"	int queryBvhNodeIndex = get_group_id(0) * get_local_size(0) + get_local_id(0);\n"
+	"	if(queryBvhNodeIndex >= numQueryAabbs) return;\n"
+	"	\n"
+	"	int queryRigidIndex = mortonCodesAndAabbIndices[queryBvhNodeIndex].m_value;\n"
+	"	b3AabbCL queryAabb = rigidAabbs[queryRigidIndex];\n"
+	"	\n"
+	"	int stack[B3_PLVBH_TRAVERSE_MAX_STACK_SIZE];\n"
+	"	\n"
+	"	int stackSize = 1;\n"
+	"	stack[0] = *rootNodeIndex;\n"
+	"	\n"
+	"	while(stackSize)\n"
+	"	{\n"
+	"		int internalOrLeafNodeIndex = stack[ stackSize - 1 ];\n"
+	"		--stackSize;\n"
+	"		\n"
+	"		int isLeaf = isLeafNode(internalOrLeafNodeIndex);	//Internal node if false\n"
+	"		int bvhNodeIndex = getIndexWithInternalNodeMarkerRemoved(internalOrLeafNodeIndex);\n"
+	"		\n"
+	"		//Optimization - if the BVH is structured as a binary radix tree, then\n"
+	"		//each internal node corresponds to a contiguous range of leaf nodes(internalNodeLeafIndexRanges[]).\n"
+	"		//This can be used to avoid testing each AABB-AABB pair twice, including preventing each node from colliding with itself.\n"
+	"		{\n"
+	"			int highestLeafIndex = (isLeaf) ? bvhNodeIndex : internalNodeLeafIndexRanges[bvhNodeIndex].y;\n"
+	"			if(highestLeafIndex <= queryBvhNodeIndex) continue;\n"
+	"		}\n"
+	"		\n"
+	"		//bvhRigidIndex is not used if internal node\n"
+	"		int bvhRigidIndex = (isLeaf) ? mortonCodesAndAabbIndices[bvhNodeIndex].m_value : -1;\n"
+	"	\n"
+	"		b3AabbCL bvhNodeAabb = (isLeaf) ? rigidAabbs[bvhRigidIndex] : internalNodeAabbs[bvhNodeIndex];\n"
+	"		if( TestAabbAgainstAabb2(&queryAabb, &bvhNodeAabb) )\n"
+	"		{\n"
+	"			if(isLeaf)\n"
+	"			{\n"
+	"				int4 pair;\n"
+	"				pair.x = rigidAabbs[queryRigidIndex].m_minIndices[3];\n"
+	"				pair.y = rigidAabbs[bvhRigidIndex].m_minIndices[3];\n"
+	"				pair.z = NEW_PAIR_MARKER;\n"
+	"				pair.w = NEW_PAIR_MARKER;\n"
+	"				\n"
+	"				int pairIndex = atomic_inc(out_numPairs);\n"
+	"				if(pairIndex < maxPairs) out_overlappingPairs[pairIndex] = pair;\n"
+	"			}\n"
+	"			\n"
+	"			if(!isLeaf)	//Internal node\n"
+	"			{\n"
+	"				if(stackSize + 2 > B3_PLVBH_TRAVERSE_MAX_STACK_SIZE)\n"
+	"				{\n"
+	"					//Error\n"
+	"				}\n"
+	"				else\n"
+	"				{\n"
+	"					stack[ stackSize++ ] = internalNodeChildIndices[bvhNodeIndex].x;\n"
+	"					stack[ stackSize++ ] = internalNodeChildIndices[bvhNodeIndex].y;\n"
+	"				}\n"
+	"			}\n"
+	"		}\n"
+	"		\n"
+	"	}\n"
+	"}\n"
+	"//From rayCastKernels.cl\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_from;\n"
+	"	float4 m_to;\n"
+	"} b3RayInfo;\n"
+	"//From rayCastKernels.cl\n"
+	"b3Vector3 b3Vector3_normalize(b3Vector3 v)\n"
+	"{\n"
+	"	b3Vector3 normal = (b3Vector3){v.x, v.y, v.z, 0.f};\n"
+	"	return normalize(normal);	//OpenCL normalize == vector4 normalize\n"
+	"}\n"
+	"b3Scalar b3Vector3_length2(b3Vector3 v) { return v.x*v.x + v.y*v.y + v.z*v.z; }\n"
+	"b3Scalar b3Vector3_dot(b3Vector3 a, b3Vector3 b) { return a.x*b.x + a.y*b.y + a.z*b.z; }\n"
+	"int rayIntersectsAabb(b3Vector3 rayOrigin, b3Scalar rayLength, b3Vector3 rayNormalizedDirection, b3AabbCL aabb)\n"
+	"{\n"
+	"	//AABB is considered as 3 pairs of 2 planes( {x_min, x_max}, {y_min, y_max}, {z_min, z_max} ).\n"
+	"	//t_min is the point of intersection with the closer plane, t_max is the point of intersection with the farther plane.\n"
+	"	//\n"
+	"	//if (rayNormalizedDirection.x < 0.0f), then max.x will be the near plane \n"
+	"	//and min.x will be the far plane; otherwise, it is reversed.\n"
+	"	//\n"
+	"	//In order for there to be a collision, the t_min and t_max of each pair must overlap.\n"
+	"	//This can be tested for by selecting the highest t_min and lowest t_max and comparing them.\n"
+	"	\n"
+	"	int4 isNegative = isless( rayNormalizedDirection, ((b3Vector3){0.0f, 0.0f, 0.0f, 0.0f}) );	//isless(x,y) returns (x < y)\n"
+	"	\n"
+	"	//When using vector types, the select() function checks the most signficant bit, \n"
+	"	//but isless() sets the least significant bit.\n"
+	"	isNegative <<= 31;\n"
+	"	//select(b, a, condition) == condition ? a : b\n"
+	"	//When using select() with vector types, (condition[i]) is true if its most significant bit is 1\n"
+	"	b3Vector3 t_min = ( select(aabb.m_min, aabb.m_max, isNegative) - rayOrigin ) / rayNormalizedDirection;\n"
+	"	b3Vector3 t_max = ( select(aabb.m_max, aabb.m_min, isNegative) - rayOrigin ) / rayNormalizedDirection;\n"
+	"	\n"
+	"	b3Scalar t_min_final = 0.0f;\n"
+	"	b3Scalar t_max_final = rayLength;\n"
+	"	\n"
+	"	//Must use fmin()/fmax(); if one of the parameters is NaN, then the parameter that is not NaN is returned. \n"
+	"	//Behavior of min()/max() with NaNs is undefined. (See OpenCL Specification 1.2 [6.12.2] and [6.12.4])\n"
+	"	//Since the innermost fmin()/fmax() is always not NaN, this should never return NaN.\n"
+	"	t_min_final = fmax( t_min.z, fmax(t_min.y, fmax(t_min.x, t_min_final)) );\n"
+	"	t_max_final = fmin( t_max.z, fmin(t_max.y, fmin(t_max.x, t_max_final)) );\n"
+	"	\n"
+	"	return (t_min_final <= t_max_final);\n"
+	"}\n"
+	"__kernel void plbvhRayTraverse(__global b3AabbCL* rigidAabbs,\n"
+	"								__global int* rootNodeIndex, \n"
+	"								__global int2* internalNodeChildIndices, \n"
+	"								__global b3AabbCL* internalNodeAabbs,\n"
+	"								__global int2* internalNodeLeafIndexRanges,\n"
+	"								__global SortDataCL* mortonCodesAndAabbIndices,\n"
+	"								\n"
+	"								__global b3RayInfo* rays,\n"
+	"								\n"
+	"								__global int* out_numRayRigidPairs, \n"
+	"								__global int2* out_rayRigidPairs,\n"
+	"								int maxRayRigidPairs, int numRays)\n"
+	"{\n"
+	"	int rayIndex = get_global_id(0);\n"
+	"	if(rayIndex >= numRays) return;\n"
+	"	\n"
+	"	//\n"
+	"	b3Vector3 rayFrom = rays[rayIndex].m_from;\n"
+	"	b3Vector3 rayTo = rays[rayIndex].m_to;\n"
+	"	b3Vector3 rayNormalizedDirection = b3Vector3_normalize(rayTo - rayFrom);\n"
+	"	b3Scalar rayLength = b3Sqrt( b3Vector3_length2(rayTo - rayFrom) );\n"
+	"	\n"
+	"	//\n"
+	"	int stack[B3_PLVBH_TRAVERSE_MAX_STACK_SIZE];\n"
+	"	\n"
+	"	int stackSize = 1;\n"
+	"	stack[0] = *rootNodeIndex;\n"
+	"	\n"
+	"	while(stackSize)\n"
+	"	{\n"
+	"		int internalOrLeafNodeIndex = stack[ stackSize - 1 ];\n"
+	"		--stackSize;\n"
+	"		\n"
+	"		int isLeaf = isLeafNode(internalOrLeafNodeIndex);	//Internal node if false\n"
+	"		int bvhNodeIndex = getIndexWithInternalNodeMarkerRemoved(internalOrLeafNodeIndex);\n"
+	"		\n"
+	"		//bvhRigidIndex is not used if internal node\n"
+	"		int bvhRigidIndex = (isLeaf) ? mortonCodesAndAabbIndices[bvhNodeIndex].m_value : -1;\n"
+	"	\n"
+	"		b3AabbCL bvhNodeAabb = (isLeaf) ? rigidAabbs[bvhRigidIndex] : internalNodeAabbs[bvhNodeIndex];\n"
+	"		if( rayIntersectsAabb(rayFrom, rayLength, rayNormalizedDirection, bvhNodeAabb)  )\n"
+	"		{\n"
+	"			if(isLeaf)\n"
+	"			{\n"
+	"				int2 rayRigidPair;\n"
+	"				rayRigidPair.x = rayIndex;\n"
+	"				rayRigidPair.y = rigidAabbs[bvhRigidIndex].m_minIndices[3];\n"
+	"				\n"
+	"				int pairIndex = atomic_inc(out_numRayRigidPairs);\n"
+	"				if(pairIndex < maxRayRigidPairs) out_rayRigidPairs[pairIndex] = rayRigidPair;\n"
+	"			}\n"
+	"			\n"
+	"			if(!isLeaf)	//Internal node\n"
+	"			{\n"
+	"				if(stackSize + 2 > B3_PLVBH_TRAVERSE_MAX_STACK_SIZE)\n"
+	"				{\n"
+	"					//Error\n"
+	"				}\n"
+	"				else\n"
+	"				{\n"
+	"					stack[ stackSize++ ] = internalNodeChildIndices[bvhNodeIndex].x;\n"
+	"					stack[ stackSize++ ] = internalNodeChildIndices[bvhNodeIndex].y;\n"
+	"				}\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"__kernel void plbvhLargeAabbAabbTest(__global b3AabbCL* smallAabbs, __global b3AabbCL* largeAabbs, \n"
+	"									__global int* out_numPairs, __global int4* out_overlappingPairs, \n"
+	"									int maxPairs, int numLargeAabbRigids, int numSmallAabbRigids)\n"
+	"{\n"
+	"	int smallAabbIndex = get_global_id(0);\n"
+	"	if(smallAabbIndex >= numSmallAabbRigids) return;\n"
+	"	\n"
+	"	b3AabbCL smallAabb = smallAabbs[smallAabbIndex];\n"
+	"	for(int i = 0; i < numLargeAabbRigids; ++i)\n"
+	"	{\n"
+	"		b3AabbCL largeAabb = largeAabbs[i];\n"
+	"		if( TestAabbAgainstAabb2(&smallAabb, &largeAabb) )\n"
+	"		{\n"
+	"			int4 pair;\n"
+	"			pair.x = largeAabb.m_minIndices[3];\n"
+	"			pair.y = smallAabb.m_minIndices[3];\n"
+	"			pair.z = NEW_PAIR_MARKER;\n"
+	"			pair.w = NEW_PAIR_MARKER;\n"
+	"			\n"
+	"			int pairIndex = atomic_inc(out_numPairs);\n"
+	"			if(pairIndex < maxPairs) out_overlappingPairs[pairIndex] = pair;\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"__kernel void plbvhLargeAabbRayTest(__global b3AabbCL* largeRigidAabbs, __global b3RayInfo* rays,\n"
+	"									__global int* out_numRayRigidPairs,  __global int2* out_rayRigidPairs,\n"
+	"									int numLargeAabbRigids, int maxRayRigidPairs, int numRays)\n"
+	"{\n"
+	"	int rayIndex = get_global_id(0);\n"
+	"	if(rayIndex >= numRays) return;\n"
+	"	\n"
+	"	b3Vector3 rayFrom = rays[rayIndex].m_from;\n"
+	"	b3Vector3 rayTo = rays[rayIndex].m_to;\n"
+	"	b3Vector3 rayNormalizedDirection = b3Vector3_normalize(rayTo - rayFrom);\n"
+	"	b3Scalar rayLength = b3Sqrt( b3Vector3_length2(rayTo - rayFrom) );\n"
+	"	\n"
+	"	for(int i = 0; i < numLargeAabbRigids; ++i)\n"
+	"	{\n"
+	"		b3AabbCL rigidAabb = largeRigidAabbs[i];\n"
+	"		if( rayIntersectsAabb(rayFrom, rayLength, rayNormalizedDirection, rigidAabb) )\n"
+	"		{\n"
+	"			int2 rayRigidPair;\n"
+	"			rayRigidPair.x = rayIndex;\n"
+	"			rayRigidPair.y = rigidAabb.m_minIndices[3];\n"
+	"			\n"
+	"			int pairIndex = atomic_inc(out_numRayRigidPairs);\n"
+	"			if(pairIndex < maxRayRigidPairs) out_rayRigidPairs[pairIndex] = rayRigidPair;\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"//Set so that it is always greater than the actual common prefixes, and never selected as a parent node.\n"
+	"//If there are no duplicates, then the highest common prefix is 32 or 64, depending on the number of bits used for the z-curve.\n"
+	"//Duplicate common prefixes increase the highest common prefix at most by the number of bits used to index the leaf node.\n"
+	"//Since 32 bit ints are used to index leaf nodes, the max prefix is 64(32 + 32 bit z-curve) or 96(32 + 64 bit z-curve).\n"
+	"#define B3_PLBVH_INVALID_COMMON_PREFIX 128\n"
+	"#define B3_PLBVH_ROOT_NODE_MARKER -1\n"
+	"#define b3Int64 long\n"
+	"int computeCommonPrefixLength(b3Int64 i, b3Int64 j) { return (int)clz(i ^ j); }\n"
+	"b3Int64 computeCommonPrefix(b3Int64 i, b3Int64 j) \n"
+	"{\n"
+	"	//This function only needs to return (i & j) in order for the algorithm to work,\n"
+	"	//but it may help with debugging to mask out the lower bits.\n"
+	"	b3Int64 commonPrefixLength = (b3Int64)computeCommonPrefixLength(i, j);\n"
+	"	b3Int64 sharedBits = i & j;\n"
+	"	b3Int64 bitmask = ((b3Int64)(~0)) << (64 - commonPrefixLength);	//Set all bits after the common prefix to 0\n"
+	"	\n"
+	"	return sharedBits & bitmask;\n"
+	"}\n"
+	"//Same as computeCommonPrefixLength(), but allows for prefixes with different lengths\n"
+	"int getSharedPrefixLength(b3Int64 prefixA, int prefixLengthA, b3Int64 prefixB, int prefixLengthB)\n"
+	"{\n"
+	"	return b3Min( computeCommonPrefixLength(prefixA, prefixB), b3Min(prefixLengthA, prefixLengthB) );\n"
+	"}\n"
+	"__kernel void computeAdjacentPairCommonPrefix(__global SortDataCL* mortonCodesAndAabbIndices,\n"
+	"											__global b3Int64* out_commonPrefixes,\n"
+	"											__global int* out_commonPrefixLengths,\n"
+	"											int numInternalNodes)\n"
+	"{\n"
+	"	int internalNodeIndex = get_global_id(0);\n"
+	"	if (internalNodeIndex >= numInternalNodes) return;\n"
+	"	\n"
+	"	//Here, (internalNodeIndex + 1) is never out of bounds since it is a leaf node index,\n"
+	"	//and the number of internal nodes is always numLeafNodes - 1\n"
+	"	int leftLeafIndex = internalNodeIndex;\n"
+	"	int rightLeafIndex = internalNodeIndex + 1;\n"
+	"	\n"
+	"	int leftLeafMortonCode = mortonCodesAndAabbIndices[leftLeafIndex].m_key;\n"
+	"	int rightLeafMortonCode = mortonCodesAndAabbIndices[rightLeafIndex].m_key;\n"
+	"	\n"
+	"	//Binary radix tree construction algorithm does not work if there are duplicate morton codes.\n"
+	"	//Append the index of each leaf node to each morton code so that there are no duplicates.\n"
+	"	//The algorithm also requires that the morton codes are sorted in ascending order; this requirement\n"
+	"	//is also satisfied with this method, as (leftLeafIndex < rightLeafIndex) is always true.\n"
+	"	//\n"
+	"	//upsample(a, b) == ( ((b3Int64)a) << 32) | b\n"
+	"	b3Int64 nonduplicateLeftMortonCode = upsample(leftLeafMortonCode, leftLeafIndex);\n"
+	"	b3Int64 nonduplicateRightMortonCode = upsample(rightLeafMortonCode, rightLeafIndex);\n"
+	"	\n"
+	"	out_commonPrefixes[internalNodeIndex] = computeCommonPrefix(nonduplicateLeftMortonCode, nonduplicateRightMortonCode);\n"
+	"	out_commonPrefixLengths[internalNodeIndex] = computeCommonPrefixLength(nonduplicateLeftMortonCode, nonduplicateRightMortonCode);\n"
+	"}\n"
+	"__kernel void buildBinaryRadixTreeLeafNodes(__global int* commonPrefixLengths, __global int* out_leafNodeParentNodes,\n"
+	"											__global int2* out_childNodes, int numLeafNodes)\n"
+	"{\n"
+	"	int leafNodeIndex = get_global_id(0);\n"
+	"	if (leafNodeIndex >= numLeafNodes) return;\n"
+	"	\n"
+	"	int numInternalNodes = numLeafNodes - 1;\n"
+	"	\n"
+	"	int leftSplitIndex = leafNodeIndex - 1;\n"
+	"	int rightSplitIndex = leafNodeIndex;\n"
+	"	\n"
+	"	int leftCommonPrefix = (leftSplitIndex >= 0) ? commonPrefixLengths[leftSplitIndex] : B3_PLBVH_INVALID_COMMON_PREFIX;\n"
+	"	int rightCommonPrefix = (rightSplitIndex < numInternalNodes) ? commonPrefixLengths[rightSplitIndex] : B3_PLBVH_INVALID_COMMON_PREFIX;\n"
+	"	\n"
+	"	//Parent node is the highest adjacent common prefix that is lower than the node's common prefix\n"
+	"	//Leaf nodes are considered as having the highest common prefix\n"
+	"	int isLeftHigherCommonPrefix = (leftCommonPrefix > rightCommonPrefix);\n"
+	"	\n"
+	"	//Handle cases for the edge nodes; the first and last node\n"
+	"	//For leaf nodes, leftCommonPrefix and rightCommonPrefix should never both be B3_PLBVH_INVALID_COMMON_PREFIX\n"
+	"	if(leftCommonPrefix == B3_PLBVH_INVALID_COMMON_PREFIX) isLeftHigherCommonPrefix = false;\n"
+	"	if(rightCommonPrefix == B3_PLBVH_INVALID_COMMON_PREFIX) isLeftHigherCommonPrefix = true;\n"
+	"	\n"
+	"	int parentNodeIndex = (isLeftHigherCommonPrefix) ? leftSplitIndex : rightSplitIndex;\n"
+	"	out_leafNodeParentNodes[leafNodeIndex] = parentNodeIndex;\n"
+	"	\n"
+	"	int isRightChild = (isLeftHigherCommonPrefix);	//If the left node is the parent, then this node is its right child and vice versa\n"
+	"	\n"
+	"	//out_childNodesAsInt[0] == int2.x == left child\n"
+	"	//out_childNodesAsInt[1] == int2.y == right child\n"
+	"	int isLeaf = 1;\n"
+	"	__global int* out_childNodesAsInt = (__global int*)(&out_childNodes[parentNodeIndex]);\n"
+	"	out_childNodesAsInt[isRightChild] = getIndexWithInternalNodeMarkerSet(isLeaf, leafNodeIndex);\n"
+	"}\n"
+	"__kernel void buildBinaryRadixTreeInternalNodes(__global b3Int64* commonPrefixes, __global int* commonPrefixLengths,\n"
+	"												__global int2* out_childNodes,\n"
+	"												__global int* out_internalNodeParentNodes, __global int* out_rootNodeIndex,\n"
+	"												int numInternalNodes)\n"
+	"{\n"
+	"	int internalNodeIndex = get_group_id(0) * get_local_size(0) + get_local_id(0);\n"
+	"	if(internalNodeIndex >= numInternalNodes) return;\n"
+	"	\n"
+	"	b3Int64 nodePrefix = commonPrefixes[internalNodeIndex];\n"
+	"	int nodePrefixLength = commonPrefixLengths[internalNodeIndex];\n"
+	"	\n"
+	"//#define USE_LINEAR_SEARCH\n"
+	"#ifdef USE_LINEAR_SEARCH\n"
+	"	int leftIndex = -1;\n"
+	"	int rightIndex = -1;\n"
+	"	\n"
+	"	//Find nearest element to left with a lower common prefix\n"
+	"	for(int i = internalNodeIndex - 1; i >= 0; --i)\n"
+	"	{\n"
+	"		int nodeLeftSharedPrefixLength = getSharedPrefixLength(nodePrefix, nodePrefixLength, commonPrefixes[i], commonPrefixLengths[i]);\n"
+	"		if(nodeLeftSharedPrefixLength < nodePrefixLength)\n"
+	"		{\n"
+	"			leftIndex = i;\n"
+	"			break;\n"
+	"		}\n"
+	"	}\n"
+	"	\n"
+	"	//Find nearest element to right with a lower common prefix\n"
+	"	for(int i = internalNodeIndex + 1; i < numInternalNodes; ++i)\n"
+	"	{\n"
+	"		int nodeRightSharedPrefixLength = getSharedPrefixLength(nodePrefix, nodePrefixLength, commonPrefixes[i], commonPrefixLengths[i]);\n"
+	"		if(nodeRightSharedPrefixLength < nodePrefixLength)\n"
+	"		{\n"
+	"			rightIndex = i;\n"
+	"			break;\n"
+	"		}\n"
+	"	}\n"
+	"	\n"
+	"#else //Use binary search\n"
+	"	//Find nearest element to left with a lower common prefix\n"
+	"	int leftIndex = -1;\n"
+	"	{\n"
+	"		int lower = 0;\n"
+	"		int upper = internalNodeIndex - 1;\n"
+	"		\n"
+	"		while(lower <= upper)\n"
+	"		{\n"
+	"			int mid = (lower + upper) / 2;\n"
+	"			b3Int64 midPrefix = commonPrefixes[mid];\n"
+	"			int midPrefixLength = commonPrefixLengths[mid];\n"
+	"			\n"
+	"			int nodeMidSharedPrefixLength = getSharedPrefixLength(nodePrefix, nodePrefixLength, midPrefix, midPrefixLength);\n"
+	"			if(nodeMidSharedPrefixLength < nodePrefixLength) \n"
+	"			{\n"
+	"				int right = mid + 1;\n"
+	"				if(right < internalNodeIndex)\n"
+	"				{\n"
+	"					b3Int64 rightPrefix = commonPrefixes[right];\n"
+	"					int rightPrefixLength = commonPrefixLengths[right];\n"
+	"					\n"
+	"					int nodeRightSharedPrefixLength = getSharedPrefixLength(nodePrefix, nodePrefixLength, rightPrefix, rightPrefixLength);\n"
+	"					if(nodeRightSharedPrefixLength < nodePrefixLength) \n"
+	"					{\n"
+	"						lower = right;\n"
+	"						leftIndex = right;\n"
+	"					}\n"
+	"					else \n"
+	"					{\n"
+	"						leftIndex = mid;\n"
+	"						break;\n"
+	"					}\n"
+	"				}\n"
+	"				else \n"
+	"				{\n"
+	"					leftIndex = mid;\n"
+	"					break;\n"
+	"				}\n"
+	"			}\n"
+	"			else upper = mid - 1;\n"
+	"		}\n"
+	"	}\n"
+	"	\n"
+	"	//Find nearest element to right with a lower common prefix\n"
+	"	int rightIndex = -1;\n"
+	"	{\n"
+	"		int lower = internalNodeIndex + 1;\n"
+	"		int upper = numInternalNodes - 1;\n"
+	"		\n"
+	"		while(lower <= upper)\n"
+	"		{\n"
+	"			int mid = (lower + upper) / 2;\n"
+	"			b3Int64 midPrefix = commonPrefixes[mid];\n"
+	"			int midPrefixLength = commonPrefixLengths[mid];\n"
+	"			\n"
+	"			int nodeMidSharedPrefixLength = getSharedPrefixLength(nodePrefix, nodePrefixLength, midPrefix, midPrefixLength);\n"
+	"			if(nodeMidSharedPrefixLength < nodePrefixLength) \n"
+	"			{\n"
+	"				int left = mid - 1;\n"
+	"				if(left > internalNodeIndex)\n"
+	"				{\n"
+	"					b3Int64 leftPrefix = commonPrefixes[left];\n"
+	"					int leftPrefixLength = commonPrefixLengths[left];\n"
+	"				\n"
+	"					int nodeLeftSharedPrefixLength = getSharedPrefixLength(nodePrefix, nodePrefixLength, leftPrefix, leftPrefixLength);\n"
+	"					if(nodeLeftSharedPrefixLength < nodePrefixLength) \n"
+	"					{\n"
+	"						upper = left;\n"
+	"						rightIndex = left;\n"
+	"					}\n"
+	"					else \n"
+	"					{\n"
+	"						rightIndex = mid;\n"
+	"						break;\n"
+	"					}\n"
+	"				}\n"
+	"				else \n"
+	"				{\n"
+	"					rightIndex = mid;\n"
+	"					break;\n"
+	"				}\n"
+	"			}\n"
+	"			else lower = mid + 1;\n"
+	"		}\n"
+	"	}\n"
+	"#endif\n"
+	"	\n"
+	"	//Select parent\n"
+	"	{\n"
+	"		int leftPrefixLength = (leftIndex != -1) ? commonPrefixLengths[leftIndex] : B3_PLBVH_INVALID_COMMON_PREFIX;\n"
+	"		int rightPrefixLength =  (rightIndex != -1) ? commonPrefixLengths[rightIndex] : B3_PLBVH_INVALID_COMMON_PREFIX;\n"
+	"		\n"
+	"		int isLeftHigherPrefixLength = (leftPrefixLength > rightPrefixLength);\n"
+	"		\n"
+	"		if(leftPrefixLength == B3_PLBVH_INVALID_COMMON_PREFIX) isLeftHigherPrefixLength = false;\n"
+	"		else if(rightPrefixLength == B3_PLBVH_INVALID_COMMON_PREFIX) isLeftHigherPrefixLength = true;\n"
+	"		\n"
+	"		int parentNodeIndex = (isLeftHigherPrefixLength) ? leftIndex : rightIndex;\n"
+	"		\n"
+	"		int isRootNode = (leftIndex == -1 && rightIndex == -1);\n"
+	"		out_internalNodeParentNodes[internalNodeIndex] = (!isRootNode) ? parentNodeIndex : B3_PLBVH_ROOT_NODE_MARKER;\n"
+	"		\n"
+	"		int isLeaf = 0;\n"
+	"		if(!isRootNode)\n"
+	"		{\n"
+	"			int isRightChild = (isLeftHigherPrefixLength);	//If the left node is the parent, then this node is its right child and vice versa\n"
+	"			\n"
+	"			//out_childNodesAsInt[0] == int2.x == left child\n"
+	"			//out_childNodesAsInt[1] == int2.y == right child\n"
+	"			__global int* out_childNodesAsInt = (__global int*)(&out_childNodes[parentNodeIndex]);\n"
+	"			out_childNodesAsInt[isRightChild] = getIndexWithInternalNodeMarkerSet(isLeaf, internalNodeIndex);\n"
+	"		}\n"
+	"		else *out_rootNodeIndex = getIndexWithInternalNodeMarkerSet(isLeaf, internalNodeIndex);\n"
+	"	}\n"
+	"}\n"
+	"__kernel void findDistanceFromRoot(__global int* rootNodeIndex, __global int* internalNodeParentNodes,\n"
+	"									__global int* out_maxDistanceFromRoot, __global int* out_distanceFromRoot, int numInternalNodes)\n"
+	"{\n"
+	"	if( get_global_id(0) == 0 ) atomic_xchg(out_maxDistanceFromRoot, 0);\n"
+	"	int internalNodeIndex = get_global_id(0);\n"
+	"	if(internalNodeIndex >= numInternalNodes) return;\n"
+	"	\n"
+	"	//\n"
+	"	int distanceFromRoot = 0;\n"
+	"	{\n"
+	"		int parentIndex = internalNodeParentNodes[internalNodeIndex];\n"
+	"		while(parentIndex != B3_PLBVH_ROOT_NODE_MARKER)\n"
+	"		{\n"
+	"			parentIndex = internalNodeParentNodes[parentIndex];\n"
+	"			++distanceFromRoot;\n"
+	"		}\n"
+	"	}\n"
+	"	out_distanceFromRoot[internalNodeIndex] = distanceFromRoot;\n"
+	"	\n"
+	"	//\n"
+	"	__local int localMaxDistanceFromRoot;\n"
+	"	if( get_local_id(0) == 0 ) localMaxDistanceFromRoot = 0;\n"
+	"	barrier(CLK_LOCAL_MEM_FENCE);\n"
+	"	\n"
+	"	atomic_max(&localMaxDistanceFromRoot, distanceFromRoot);\n"
+	"	barrier(CLK_LOCAL_MEM_FENCE);\n"
+	"	\n"
+	"	if( get_local_id(0) == 0 ) atomic_max(out_maxDistanceFromRoot, localMaxDistanceFromRoot);\n"
+	"}\n"
+	"__kernel void buildBinaryRadixTreeAabbsRecursive(__global int* distanceFromRoot, __global SortDataCL* mortonCodesAndAabbIndices,\n"
+	"												__global int2* childNodes,\n"
+	"												__global b3AabbCL* leafNodeAabbs, __global b3AabbCL* internalNodeAabbs,\n"
+	"												int maxDistanceFromRoot, int processedDistance, int numInternalNodes)\n"
+	"{\n"
+	"	int internalNodeIndex = get_global_id(0);\n"
+	"	if(internalNodeIndex >= numInternalNodes) return;\n"
+	"	\n"
+	"	int distance = distanceFromRoot[internalNodeIndex];\n"
+	"	\n"
+	"	if(distance == processedDistance)\n"
+	"	{\n"
+	"		int leftChildIndex = childNodes[internalNodeIndex].x;\n"
+	"		int rightChildIndex = childNodes[internalNodeIndex].y;\n"
+	"		\n"
+	"		int isLeftChildLeaf = isLeafNode(leftChildIndex);\n"
+	"		int isRightChildLeaf = isLeafNode(rightChildIndex);\n"
+	"		\n"
+	"		leftChildIndex = getIndexWithInternalNodeMarkerRemoved(leftChildIndex);\n"
+	"		rightChildIndex = getIndexWithInternalNodeMarkerRemoved(rightChildIndex);\n"
+	"		\n"
+	"		//leftRigidIndex/rightRigidIndex is not used if internal node\n"
+	"		int leftRigidIndex = (isLeftChildLeaf) ? mortonCodesAndAabbIndices[leftChildIndex].m_value : -1;\n"
+	"		int rightRigidIndex = (isRightChildLeaf) ? mortonCodesAndAabbIndices[rightChildIndex].m_value : -1;\n"
+	"		\n"
+	"		b3AabbCL leftChildAabb = (isLeftChildLeaf) ? leafNodeAabbs[leftRigidIndex] : internalNodeAabbs[leftChildIndex];\n"
+	"		b3AabbCL rightChildAabb = (isRightChildLeaf) ? leafNodeAabbs[rightRigidIndex] : internalNodeAabbs[rightChildIndex];\n"
+	"		\n"
+	"		b3AabbCL mergedAabb;\n"
+	"		mergedAabb.m_min = b3Min(leftChildAabb.m_min, rightChildAabb.m_min);\n"
+	"		mergedAabb.m_max = b3Max(leftChildAabb.m_max, rightChildAabb.m_max);\n"
+	"		internalNodeAabbs[internalNodeIndex] = mergedAabb;\n"
+	"	}\n"
+	"}\n"
+	"__kernel void findLeafIndexRanges(__global int2* internalNodeChildNodes, __global int2* out_leafIndexRanges, int numInternalNodes)\n"
+	"{\n"
+	"	int internalNodeIndex = get_global_id(0);\n"
+	"	if(internalNodeIndex >= numInternalNodes) return;\n"
+	"	\n"
+	"	int numLeafNodes = numInternalNodes + 1;\n"
+	"	\n"
+	"	int2 childNodes = internalNodeChildNodes[internalNodeIndex];\n"
+	"	\n"
+	"	int2 leafIndexRange;	//x == min leaf index, y == max leaf index\n"
+	"	\n"
+	"	//Find lowest leaf index covered by this internal node\n"
+	"	{\n"
+	"		int lowestIndex = childNodes.x;		//childNodes.x == Left child\n"
+	"		while( !isLeafNode(lowestIndex) ) lowestIndex = internalNodeChildNodes[ getIndexWithInternalNodeMarkerRemoved(lowestIndex) ].x;\n"
+	"		leafIndexRange.x = lowestIndex;\n"
+	"	}\n"
+	"	\n"
+	"	//Find highest leaf index covered by this internal node\n"
+	"	{\n"
+	"		int highestIndex = childNodes.y;	//childNodes.y == Right child\n"
+	"		while( !isLeafNode(highestIndex) ) highestIndex = internalNodeChildNodes[ getIndexWithInternalNodeMarkerRemoved(highestIndex) ].y;\n"
+	"		leafIndexRange.y = highestIndex;\n"
+	"	}\n"
+	"	\n"
+	"	//\n"
+	"	out_leafIndexRanges[internalNodeIndex] = leafIndexRange;\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/kernels/sapKernels.h b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/kernels/sapKernels.h
index 04d40fcf26e..d6999b94cb3 100644
--- a/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/kernels/sapKernels.h
+++ b/thirdparty/bullet/Bullet3OpenCL/BroadphaseCollision/kernels/sapKernels.h
@@ -1,342 +1,341 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* sapCL= \
-"/*\n"
-"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
-"This software is provided 'as-is', without any express or implied warranty.\n"
-"In no event will the authors be held liable for any damages arising from the use of this software.\n"
-"Permission is granted to anyone to use this software for any purpose, \n"
-"including commercial applications, and to alter it and redistribute it freely, \n"
-"subject to the following restrictions:\n"
-"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
-"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
-"3. This notice may not be removed or altered from any source distribution.\n"
-"*/\n"
-"//Originally written by Erwin Coumans\n"
-"#define NEW_PAIR_MARKER -1\n"
-"typedef struct \n"
-"{\n"
-"	union\n"
-"	{\n"
-"		float4	m_min;\n"
-"		float   m_minElems[4];\n"
-"		int			m_minIndices[4];\n"
-"	};\n"
-"	union\n"
-"	{\n"
-"		float4	m_max;\n"
-"		float   m_maxElems[4];\n"
-"		int			m_maxIndices[4];\n"
-"	};\n"
-"} btAabbCL;\n"
-"/// conservative test for overlap between two aabbs\n"
-"bool TestAabbAgainstAabb2(const btAabbCL* aabb1, __local const btAabbCL* aabb2);\n"
-"bool TestAabbAgainstAabb2(const btAabbCL* aabb1, __local const btAabbCL* aabb2)\n"
-"{\n"
-"	bool overlap = true;\n"
-"	overlap = (aabb1->m_min.x > aabb2->m_max.x || aabb1->m_max.x < aabb2->m_min.x) ? false : overlap;\n"
-"	overlap = (aabb1->m_min.z > aabb2->m_max.z || aabb1->m_max.z < aabb2->m_min.z) ? false : overlap;\n"
-"	overlap = (aabb1->m_min.y > aabb2->m_max.y || aabb1->m_max.y < aabb2->m_min.y) ? false : overlap;\n"
-"	return overlap;\n"
-"}\n"
-"bool TestAabbAgainstAabb2GlobalGlobal(__global const btAabbCL* aabb1, __global const btAabbCL* aabb2);\n"
-"bool TestAabbAgainstAabb2GlobalGlobal(__global const btAabbCL* aabb1, __global const btAabbCL* aabb2)\n"
-"{\n"
-"	bool overlap = true;\n"
-"	overlap = (aabb1->m_min.x > aabb2->m_max.x || aabb1->m_max.x < aabb2->m_min.x) ? false : overlap;\n"
-"	overlap = (aabb1->m_min.z > aabb2->m_max.z || aabb1->m_max.z < aabb2->m_min.z) ? false : overlap;\n"
-"	overlap = (aabb1->m_min.y > aabb2->m_max.y || aabb1->m_max.y < aabb2->m_min.y) ? false : overlap;\n"
-"	return overlap;\n"
-"}\n"
-"bool TestAabbAgainstAabb2Global(const btAabbCL* aabb1, __global const btAabbCL* aabb2);\n"
-"bool TestAabbAgainstAabb2Global(const btAabbCL* aabb1, __global const btAabbCL* aabb2)\n"
-"{\n"
-"	bool overlap = true;\n"
-"	overlap = (aabb1->m_min.x > aabb2->m_max.x || aabb1->m_max.x < aabb2->m_min.x) ? false : overlap;\n"
-"	overlap = (aabb1->m_min.z > aabb2->m_max.z || aabb1->m_max.z < aabb2->m_min.z) ? false : overlap;\n"
-"	overlap = (aabb1->m_min.y > aabb2->m_max.y || aabb1->m_max.y < aabb2->m_min.y) ? false : overlap;\n"
-"	return overlap;\n"
-"}\n"
-"__kernel void   computePairsKernelTwoArrays( __global const btAabbCL* unsortedAabbs, __global const int* unsortedAabbMapping,  __global const int* unsortedAabbMapping2, volatile __global int4* pairsOut,volatile  __global int* pairCount, int numUnsortedAabbs, int numUnSortedAabbs2, int axis, int maxPairs)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	if (i>=numUnsortedAabbs)\n"
-"		return;\n"
-"	int j = get_global_id(1);\n"
-"	if (j>=numUnSortedAabbs2)\n"
-"		return;\n"
-"	__global const btAabbCL* unsortedAabbPtr = &unsortedAabbs[unsortedAabbMapping[i]];\n"
-"	__global const btAabbCL* unsortedAabbPtr2 = &unsortedAabbs[unsortedAabbMapping2[j]];\n"
-"	if (TestAabbAgainstAabb2GlobalGlobal(unsortedAabbPtr,unsortedAabbPtr2))\n"
-"	{\n"
-"		int4 myPair;\n"
-"		\n"
-"		int xIndex = unsortedAabbPtr[0].m_minIndices[3];\n"
-"		int yIndex = unsortedAabbPtr2[0].m_minIndices[3];\n"
-"		if (xIndex>yIndex)\n"
-"		{\n"
-"			int tmp = xIndex;\n"
-"			xIndex=yIndex;\n"
-"			yIndex=tmp;\n"
-"		}\n"
-"		\n"
-"		myPair.x = xIndex;\n"
-"		myPair.y = yIndex;\n"
-"		myPair.z = NEW_PAIR_MARKER;\n"
-"		myPair.w = NEW_PAIR_MARKER;\n"
-"		int curPair = atomic_inc (pairCount);\n"
-"		if (curPair<maxPairs)\n"
-"		{\n"
-"				pairsOut[curPair] = myPair; //flush to main memory\n"
-"		}\n"
-"	}\n"
-"}\n"
-"__kernel void   computePairsKernelBruteForce( __global const btAabbCL* aabbs, volatile __global int4* pairsOut,volatile  __global int* pairCount, int numObjects, int axis, int maxPairs)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	if (i>=numObjects)\n"
-"		return;\n"
-"	for (int j=i+1;j<numObjects;j++)\n"
-"	{\n"
-"		if (TestAabbAgainstAabb2GlobalGlobal(&aabbs[i],&aabbs[j]))\n"
-"		{\n"
-"			int4 myPair;\n"
-"			myPair.x = aabbs[i].m_minIndices[3];\n"
-"			myPair.y = aabbs[j].m_minIndices[3];\n"
-"			myPair.z = NEW_PAIR_MARKER;\n"
-"			myPair.w = NEW_PAIR_MARKER;\n"
-"			int curPair = atomic_inc (pairCount);\n"
-"			if (curPair<maxPairs)\n"
-"			{\n"
-"					pairsOut[curPair] = myPair; //flush to main memory\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"}\n"
-"__kernel void   computePairsKernelOriginal( __global const btAabbCL* aabbs, volatile __global int4* pairsOut,volatile  __global int* pairCount, int numObjects, int axis, int maxPairs)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	if (i>=numObjects)\n"
-"		return;\n"
-"	for (int j=i+1;j<numObjects;j++)\n"
-"	{\n"
-"  	if(aabbs[i].m_maxElems[axis] < (aabbs[j].m_minElems[axis])) \n"
-"		{\n"
-"			break;\n"
-"		}\n"
-"		if (TestAabbAgainstAabb2GlobalGlobal(&aabbs[i],&aabbs[j]))\n"
-"		{\n"
-"			int4 myPair;\n"
-"			myPair.x = aabbs[i].m_minIndices[3];\n"
-"			myPair.y = aabbs[j].m_minIndices[3];\n"
-"			myPair.z = NEW_PAIR_MARKER;\n"
-"			myPair.w = NEW_PAIR_MARKER;\n"
-"			int curPair = atomic_inc (pairCount);\n"
-"			if (curPair<maxPairs)\n"
-"			{\n"
-"					pairsOut[curPair] = myPair; //flush to main memory\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"}\n"
-"__kernel void   computePairsKernelBarrier( __global const btAabbCL* aabbs, volatile __global int4* pairsOut,volatile  __global int* pairCount, int numObjects, int axis, int maxPairs)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	int localId = get_local_id(0);\n"
-"	__local int numActiveWgItems[1];\n"
-"	__local int breakRequest[1];\n"
-"	if (localId==0)\n"
-"	{\n"
-"		numActiveWgItems[0] = 0;\n"
-"		breakRequest[0] = 0;\n"
-"	}\n"
-"	barrier(CLK_LOCAL_MEM_FENCE);\n"
-"	atomic_inc(numActiveWgItems);\n"
-"	barrier(CLK_LOCAL_MEM_FENCE);\n"
-"	int localBreak = 0;\n"
-"	int j=i+1;\n"
-"	do\n"
-"	{\n"
-"		barrier(CLK_LOCAL_MEM_FENCE);\n"
-"	\n"
-"		if (j<numObjects)\n"
-"		{\n"
-"	  	if(aabbs[i].m_maxElems[axis] < (aabbs[j].m_minElems[axis])) \n"
-"			{\n"
-"				if (!localBreak)\n"
-"				{\n"
-"					atomic_inc(breakRequest);\n"
-"					localBreak = 1;\n"
-"				}\n"
-"			}\n"
-"		}\n"
-"		\n"
-"		barrier(CLK_LOCAL_MEM_FENCE);\n"
-"		\n"
-"		if (j>=numObjects && !localBreak)\n"
-"		{\n"
-"			atomic_inc(breakRequest);\n"
-"			localBreak = 1;\n"
-"		}\n"
-"		barrier(CLK_LOCAL_MEM_FENCE);\n"
-"		\n"
-"		if (!localBreak)\n"
-"		{\n"
-"			if (TestAabbAgainstAabb2GlobalGlobal(&aabbs[i],&aabbs[j]))\n"
-"			{\n"
-"				int4 myPair;\n"
-"				myPair.x = aabbs[i].m_minIndices[3];\n"
-"				myPair.y = aabbs[j].m_minIndices[3];\n"
-"				myPair.z = NEW_PAIR_MARKER;\n"
-"				myPair.w = NEW_PAIR_MARKER;\n"
-"				int curPair = atomic_inc (pairCount);\n"
-"				if (curPair<maxPairs)\n"
-"				{\n"
-"						pairsOut[curPair] = myPair; //flush to main memory\n"
-"				}\n"
-"			}\n"
-"		}\n"
-"		j++;\n"
-"	} while (breakRequest[0]<numActiveWgItems[0]);\n"
-"}\n"
-"__kernel void   computePairsKernelLocalSharedMemory( __global const btAabbCL* aabbs, volatile __global int4* pairsOut,volatile  __global int* pairCount, int numObjects, int axis, int maxPairs)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	int localId = get_local_id(0);\n"
-"	__local int numActiveWgItems[1];\n"
-"	__local int breakRequest[1];\n"
-"	__local btAabbCL localAabbs[128];// = aabbs[i];\n"
-"	\n"
-"	btAabbCL myAabb;\n"
-"	\n"
-"	myAabb = (i<numObjects)? aabbs[i]:aabbs[0];\n"
-"	float testValue = 	myAabb.m_maxElems[axis];\n"
-"	\n"
-"	if (localId==0)\n"
-"	{\n"
-"		numActiveWgItems[0] = 0;\n"
-"		breakRequest[0] = 0;\n"
-"	}\n"
-"	int localCount=0;\n"
-"	int block=0;\n"
-"	localAabbs[localId] = (i+block)<numObjects? aabbs[i+block] : aabbs[0];\n"
-"	localAabbs[localId+64] = (i+block+64)<numObjects? aabbs[i+block+64]: aabbs[0];\n"
-"	\n"
-"	barrier(CLK_LOCAL_MEM_FENCE);\n"
-"	atomic_inc(numActiveWgItems);\n"
-"	barrier(CLK_LOCAL_MEM_FENCE);\n"
-"	int localBreak = 0;\n"
-"	\n"
-"	int j=i+1;\n"
-"	do\n"
-"	{\n"
-"		barrier(CLK_LOCAL_MEM_FENCE);\n"
-"	\n"
-"		if (j<numObjects)\n"
-"		{\n"
-"	  	if(testValue < (localAabbs[localCount+localId+1].m_minElems[axis])) \n"
-"			{\n"
-"				if (!localBreak)\n"
-"				{\n"
-"					atomic_inc(breakRequest);\n"
-"					localBreak = 1;\n"
-"				}\n"
-"			}\n"
-"		}\n"
-"		\n"
-"		barrier(CLK_LOCAL_MEM_FENCE);\n"
-"		\n"
-"		if (j>=numObjects && !localBreak)\n"
-"		{\n"
-"			atomic_inc(breakRequest);\n"
-"			localBreak = 1;\n"
-"		}\n"
-"		barrier(CLK_LOCAL_MEM_FENCE);\n"
-"		\n"
-"		if (!localBreak)\n"
-"		{\n"
-"			if (TestAabbAgainstAabb2(&myAabb,&localAabbs[localCount+localId+1]))\n"
-"			{\n"
-"				int4 myPair;\n"
-"				myPair.x = myAabb.m_minIndices[3];\n"
-"				myPair.y = localAabbs[localCount+localId+1].m_minIndices[3];\n"
-"				myPair.z = NEW_PAIR_MARKER;\n"
-"				myPair.w = NEW_PAIR_MARKER;\n"
-"				int curPair = atomic_inc (pairCount);\n"
-"				if (curPair<maxPairs)\n"
-"				{\n"
-"						pairsOut[curPair] = myPair; //flush to main memory\n"
-"				}\n"
-"			}\n"
-"		}\n"
-"		\n"
-"		barrier(CLK_LOCAL_MEM_FENCE);\n"
-"		localCount++;\n"
-"		if (localCount==64)\n"
-"		{\n"
-"			localCount = 0;\n"
-"			block+=64;			\n"
-"			localAabbs[localId] = ((i+block)<numObjects) ? aabbs[i+block] : aabbs[0];\n"
-"			localAabbs[localId+64] = ((i+64+block)<numObjects) ? aabbs[i+block+64] : aabbs[0];\n"
-"		}\n"
-"		j++;\n"
-"		\n"
-"	} while (breakRequest[0]<numActiveWgItems[0]);\n"
-"	\n"
-"}\n"
-"//http://stereopsis.com/radix.html\n"
-"unsigned int FloatFlip(float fl);\n"
-"unsigned int FloatFlip(float fl)\n"
-"{\n"
-"	unsigned int f = *(unsigned int*)&fl;\n"
-"	unsigned int mask = -(int)(f >> 31) | 0x80000000;\n"
-"	return f ^ mask;\n"
-"}\n"
-"float IFloatFlip(unsigned int f);\n"
-"float IFloatFlip(unsigned int f)\n"
-"{\n"
-"	unsigned int mask = ((f >> 31) - 1) | 0x80000000;\n"
-"	unsigned int fl = f ^ mask;\n"
-"	return *(float*)&fl;\n"
-"}\n"
-"__kernel void   copyAabbsKernel( __global const btAabbCL* allAabbs, __global btAabbCL* destAabbs, int numObjects)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	if (i>=numObjects)\n"
-"		return;\n"
-"	int src = destAabbs[i].m_maxIndices[3];\n"
-"	destAabbs[i] = allAabbs[src];\n"
-"	destAabbs[i].m_maxIndices[3] = src;\n"
-"}\n"
-"__kernel void   flipFloatKernel( __global const btAabbCL* allAabbs, __global const int* smallAabbMapping, __global int2* sortData, int numObjects, int axis)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	if (i>=numObjects)\n"
-"		return;\n"
-"	\n"
-"	\n"
-"	sortData[i].x = FloatFlip(allAabbs[smallAabbMapping[i]].m_minElems[axis]);\n"
-"	sortData[i].y = i;\n"
-"		\n"
-"}\n"
-"__kernel void   scatterKernel( __global const btAabbCL* allAabbs, __global const int* smallAabbMapping, volatile __global const int2* sortData, __global btAabbCL* sortedAabbs, int numObjects)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	if (i>=numObjects)\n"
-"		return;\n"
-"	\n"
-"	sortedAabbs[i] = allAabbs[smallAabbMapping[sortData[i].y]];\n"
-"}\n"
-"__kernel void   prepareSumVarianceKernel( __global const btAabbCL* allAabbs, __global const int* smallAabbMapping, __global float4* sum, __global float4* sum2,int numAabbs)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	if (i>=numAabbs)\n"
-"		return;\n"
-"	\n"
-"	btAabbCL smallAabb = allAabbs[smallAabbMapping[i]];\n"
-"	\n"
-"	float4 s;\n"
-"	s = (smallAabb.m_max+smallAabb.m_min)*0.5f;\n"
-"	sum[i]=s;\n"
-"	sum2[i]=s*s;	\n"
-"}\n"
-;
+static const char* sapCL =
+	"/*\n"
+	"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
+	"This software is provided 'as-is', without any express or implied warranty.\n"
+	"In no event will the authors be held liable for any damages arising from the use of this software.\n"
+	"Permission is granted to anyone to use this software for any purpose, \n"
+	"including commercial applications, and to alter it and redistribute it freely, \n"
+	"subject to the following restrictions:\n"
+	"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
+	"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
+	"3. This notice may not be removed or altered from any source distribution.\n"
+	"*/\n"
+	"//Originally written by Erwin Coumans\n"
+	"#define NEW_PAIR_MARKER -1\n"
+	"typedef struct \n"
+	"{\n"
+	"	union\n"
+	"	{\n"
+	"		float4	m_min;\n"
+	"		float   m_minElems[4];\n"
+	"		int			m_minIndices[4];\n"
+	"	};\n"
+	"	union\n"
+	"	{\n"
+	"		float4	m_max;\n"
+	"		float   m_maxElems[4];\n"
+	"		int			m_maxIndices[4];\n"
+	"	};\n"
+	"} btAabbCL;\n"
+	"/// conservative test for overlap between two aabbs\n"
+	"bool TestAabbAgainstAabb2(const btAabbCL* aabb1, __local const btAabbCL* aabb2);\n"
+	"bool TestAabbAgainstAabb2(const btAabbCL* aabb1, __local const btAabbCL* aabb2)\n"
+	"{\n"
+	"	bool overlap = true;\n"
+	"	overlap = (aabb1->m_min.x > aabb2->m_max.x || aabb1->m_max.x < aabb2->m_min.x) ? false : overlap;\n"
+	"	overlap = (aabb1->m_min.z > aabb2->m_max.z || aabb1->m_max.z < aabb2->m_min.z) ? false : overlap;\n"
+	"	overlap = (aabb1->m_min.y > aabb2->m_max.y || aabb1->m_max.y < aabb2->m_min.y) ? false : overlap;\n"
+	"	return overlap;\n"
+	"}\n"
+	"bool TestAabbAgainstAabb2GlobalGlobal(__global const btAabbCL* aabb1, __global const btAabbCL* aabb2);\n"
+	"bool TestAabbAgainstAabb2GlobalGlobal(__global const btAabbCL* aabb1, __global const btAabbCL* aabb2)\n"
+	"{\n"
+	"	bool overlap = true;\n"
+	"	overlap = (aabb1->m_min.x > aabb2->m_max.x || aabb1->m_max.x < aabb2->m_min.x) ? false : overlap;\n"
+	"	overlap = (aabb1->m_min.z > aabb2->m_max.z || aabb1->m_max.z < aabb2->m_min.z) ? false : overlap;\n"
+	"	overlap = (aabb1->m_min.y > aabb2->m_max.y || aabb1->m_max.y < aabb2->m_min.y) ? false : overlap;\n"
+	"	return overlap;\n"
+	"}\n"
+	"bool TestAabbAgainstAabb2Global(const btAabbCL* aabb1, __global const btAabbCL* aabb2);\n"
+	"bool TestAabbAgainstAabb2Global(const btAabbCL* aabb1, __global const btAabbCL* aabb2)\n"
+	"{\n"
+	"	bool overlap = true;\n"
+	"	overlap = (aabb1->m_min.x > aabb2->m_max.x || aabb1->m_max.x < aabb2->m_min.x) ? false : overlap;\n"
+	"	overlap = (aabb1->m_min.z > aabb2->m_max.z || aabb1->m_max.z < aabb2->m_min.z) ? false : overlap;\n"
+	"	overlap = (aabb1->m_min.y > aabb2->m_max.y || aabb1->m_max.y < aabb2->m_min.y) ? false : overlap;\n"
+	"	return overlap;\n"
+	"}\n"
+	"__kernel void   computePairsKernelTwoArrays( __global const btAabbCL* unsortedAabbs, __global const int* unsortedAabbMapping,  __global const int* unsortedAabbMapping2, volatile __global int4* pairsOut,volatile  __global int* pairCount, int numUnsortedAabbs, int numUnSortedAabbs2, int axis, int maxPairs)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	if (i>=numUnsortedAabbs)\n"
+	"		return;\n"
+	"	int j = get_global_id(1);\n"
+	"	if (j>=numUnSortedAabbs2)\n"
+	"		return;\n"
+	"	__global const btAabbCL* unsortedAabbPtr = &unsortedAabbs[unsortedAabbMapping[i]];\n"
+	"	__global const btAabbCL* unsortedAabbPtr2 = &unsortedAabbs[unsortedAabbMapping2[j]];\n"
+	"	if (TestAabbAgainstAabb2GlobalGlobal(unsortedAabbPtr,unsortedAabbPtr2))\n"
+	"	{\n"
+	"		int4 myPair;\n"
+	"		\n"
+	"		int xIndex = unsortedAabbPtr[0].m_minIndices[3];\n"
+	"		int yIndex = unsortedAabbPtr2[0].m_minIndices[3];\n"
+	"		if (xIndex>yIndex)\n"
+	"		{\n"
+	"			int tmp = xIndex;\n"
+	"			xIndex=yIndex;\n"
+	"			yIndex=tmp;\n"
+	"		}\n"
+	"		\n"
+	"		myPair.x = xIndex;\n"
+	"		myPair.y = yIndex;\n"
+	"		myPair.z = NEW_PAIR_MARKER;\n"
+	"		myPair.w = NEW_PAIR_MARKER;\n"
+	"		int curPair = atomic_inc (pairCount);\n"
+	"		if (curPair<maxPairs)\n"
+	"		{\n"
+	"				pairsOut[curPair] = myPair; //flush to main memory\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"__kernel void   computePairsKernelBruteForce( __global const btAabbCL* aabbs, volatile __global int4* pairsOut,volatile  __global int* pairCount, int numObjects, int axis, int maxPairs)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	if (i>=numObjects)\n"
+	"		return;\n"
+	"	for (int j=i+1;j<numObjects;j++)\n"
+	"	{\n"
+	"		if (TestAabbAgainstAabb2GlobalGlobal(&aabbs[i],&aabbs[j]))\n"
+	"		{\n"
+	"			int4 myPair;\n"
+	"			myPair.x = aabbs[i].m_minIndices[3];\n"
+	"			myPair.y = aabbs[j].m_minIndices[3];\n"
+	"			myPair.z = NEW_PAIR_MARKER;\n"
+	"			myPair.w = NEW_PAIR_MARKER;\n"
+	"			int curPair = atomic_inc (pairCount);\n"
+	"			if (curPair<maxPairs)\n"
+	"			{\n"
+	"					pairsOut[curPair] = myPair; //flush to main memory\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"__kernel void   computePairsKernelOriginal( __global const btAabbCL* aabbs, volatile __global int4* pairsOut,volatile  __global int* pairCount, int numObjects, int axis, int maxPairs)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	if (i>=numObjects)\n"
+	"		return;\n"
+	"	for (int j=i+1;j<numObjects;j++)\n"
+	"	{\n"
+	"  	if(aabbs[i].m_maxElems[axis] < (aabbs[j].m_minElems[axis])) \n"
+	"		{\n"
+	"			break;\n"
+	"		}\n"
+	"		if (TestAabbAgainstAabb2GlobalGlobal(&aabbs[i],&aabbs[j]))\n"
+	"		{\n"
+	"			int4 myPair;\n"
+	"			myPair.x = aabbs[i].m_minIndices[3];\n"
+	"			myPair.y = aabbs[j].m_minIndices[3];\n"
+	"			myPair.z = NEW_PAIR_MARKER;\n"
+	"			myPair.w = NEW_PAIR_MARKER;\n"
+	"			int curPair = atomic_inc (pairCount);\n"
+	"			if (curPair<maxPairs)\n"
+	"			{\n"
+	"					pairsOut[curPair] = myPair; //flush to main memory\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"__kernel void   computePairsKernelBarrier( __global const btAabbCL* aabbs, volatile __global int4* pairsOut,volatile  __global int* pairCount, int numObjects, int axis, int maxPairs)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	int localId = get_local_id(0);\n"
+	"	__local int numActiveWgItems[1];\n"
+	"	__local int breakRequest[1];\n"
+	"	if (localId==0)\n"
+	"	{\n"
+	"		numActiveWgItems[0] = 0;\n"
+	"		breakRequest[0] = 0;\n"
+	"	}\n"
+	"	barrier(CLK_LOCAL_MEM_FENCE);\n"
+	"	atomic_inc(numActiveWgItems);\n"
+	"	barrier(CLK_LOCAL_MEM_FENCE);\n"
+	"	int localBreak = 0;\n"
+	"	int j=i+1;\n"
+	"	do\n"
+	"	{\n"
+	"		barrier(CLK_LOCAL_MEM_FENCE);\n"
+	"	\n"
+	"		if (j<numObjects)\n"
+	"		{\n"
+	"	  	if(aabbs[i].m_maxElems[axis] < (aabbs[j].m_minElems[axis])) \n"
+	"			{\n"
+	"				if (!localBreak)\n"
+	"				{\n"
+	"					atomic_inc(breakRequest);\n"
+	"					localBreak = 1;\n"
+	"				}\n"
+	"			}\n"
+	"		}\n"
+	"		\n"
+	"		barrier(CLK_LOCAL_MEM_FENCE);\n"
+	"		\n"
+	"		if (j>=numObjects && !localBreak)\n"
+	"		{\n"
+	"			atomic_inc(breakRequest);\n"
+	"			localBreak = 1;\n"
+	"		}\n"
+	"		barrier(CLK_LOCAL_MEM_FENCE);\n"
+	"		\n"
+	"		if (!localBreak)\n"
+	"		{\n"
+	"			if (TestAabbAgainstAabb2GlobalGlobal(&aabbs[i],&aabbs[j]))\n"
+	"			{\n"
+	"				int4 myPair;\n"
+	"				myPair.x = aabbs[i].m_minIndices[3];\n"
+	"				myPair.y = aabbs[j].m_minIndices[3];\n"
+	"				myPair.z = NEW_PAIR_MARKER;\n"
+	"				myPair.w = NEW_PAIR_MARKER;\n"
+	"				int curPair = atomic_inc (pairCount);\n"
+	"				if (curPair<maxPairs)\n"
+	"				{\n"
+	"						pairsOut[curPair] = myPair; //flush to main memory\n"
+	"				}\n"
+	"			}\n"
+	"		}\n"
+	"		j++;\n"
+	"	} while (breakRequest[0]<numActiveWgItems[0]);\n"
+	"}\n"
+	"__kernel void   computePairsKernelLocalSharedMemory( __global const btAabbCL* aabbs, volatile __global int4* pairsOut,volatile  __global int* pairCount, int numObjects, int axis, int maxPairs)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	int localId = get_local_id(0);\n"
+	"	__local int numActiveWgItems[1];\n"
+	"	__local int breakRequest[1];\n"
+	"	__local btAabbCL localAabbs[128];// = aabbs[i];\n"
+	"	\n"
+	"	btAabbCL myAabb;\n"
+	"	\n"
+	"	myAabb = (i<numObjects)? aabbs[i]:aabbs[0];\n"
+	"	float testValue = 	myAabb.m_maxElems[axis];\n"
+	"	\n"
+	"	if (localId==0)\n"
+	"	{\n"
+	"		numActiveWgItems[0] = 0;\n"
+	"		breakRequest[0] = 0;\n"
+	"	}\n"
+	"	int localCount=0;\n"
+	"	int block=0;\n"
+	"	localAabbs[localId] = (i+block)<numObjects? aabbs[i+block] : aabbs[0];\n"
+	"	localAabbs[localId+64] = (i+block+64)<numObjects? aabbs[i+block+64]: aabbs[0];\n"
+	"	\n"
+	"	barrier(CLK_LOCAL_MEM_FENCE);\n"
+	"	atomic_inc(numActiveWgItems);\n"
+	"	barrier(CLK_LOCAL_MEM_FENCE);\n"
+	"	int localBreak = 0;\n"
+	"	\n"
+	"	int j=i+1;\n"
+	"	do\n"
+	"	{\n"
+	"		barrier(CLK_LOCAL_MEM_FENCE);\n"
+	"	\n"
+	"		if (j<numObjects)\n"
+	"		{\n"
+	"	  	if(testValue < (localAabbs[localCount+localId+1].m_minElems[axis])) \n"
+	"			{\n"
+	"				if (!localBreak)\n"
+	"				{\n"
+	"					atomic_inc(breakRequest);\n"
+	"					localBreak = 1;\n"
+	"				}\n"
+	"			}\n"
+	"		}\n"
+	"		\n"
+	"		barrier(CLK_LOCAL_MEM_FENCE);\n"
+	"		\n"
+	"		if (j>=numObjects && !localBreak)\n"
+	"		{\n"
+	"			atomic_inc(breakRequest);\n"
+	"			localBreak = 1;\n"
+	"		}\n"
+	"		barrier(CLK_LOCAL_MEM_FENCE);\n"
+	"		\n"
+	"		if (!localBreak)\n"
+	"		{\n"
+	"			if (TestAabbAgainstAabb2(&myAabb,&localAabbs[localCount+localId+1]))\n"
+	"			{\n"
+	"				int4 myPair;\n"
+	"				myPair.x = myAabb.m_minIndices[3];\n"
+	"				myPair.y = localAabbs[localCount+localId+1].m_minIndices[3];\n"
+	"				myPair.z = NEW_PAIR_MARKER;\n"
+	"				myPair.w = NEW_PAIR_MARKER;\n"
+	"				int curPair = atomic_inc (pairCount);\n"
+	"				if (curPair<maxPairs)\n"
+	"				{\n"
+	"						pairsOut[curPair] = myPair; //flush to main memory\n"
+	"				}\n"
+	"			}\n"
+	"		}\n"
+	"		\n"
+	"		barrier(CLK_LOCAL_MEM_FENCE);\n"
+	"		localCount++;\n"
+	"		if (localCount==64)\n"
+	"		{\n"
+	"			localCount = 0;\n"
+	"			block+=64;			\n"
+	"			localAabbs[localId] = ((i+block)<numObjects) ? aabbs[i+block] : aabbs[0];\n"
+	"			localAabbs[localId+64] = ((i+64+block)<numObjects) ? aabbs[i+block+64] : aabbs[0];\n"
+	"		}\n"
+	"		j++;\n"
+	"		\n"
+	"	} while (breakRequest[0]<numActiveWgItems[0]);\n"
+	"	\n"
+	"}\n"
+	"//http://stereopsis.com/radix.html\n"
+	"unsigned int FloatFlip(float fl);\n"
+	"unsigned int FloatFlip(float fl)\n"
+	"{\n"
+	"	unsigned int f = *(unsigned int*)&fl;\n"
+	"	unsigned int mask = -(int)(f >> 31) | 0x80000000;\n"
+	"	return f ^ mask;\n"
+	"}\n"
+	"float IFloatFlip(unsigned int f);\n"
+	"float IFloatFlip(unsigned int f)\n"
+	"{\n"
+	"	unsigned int mask = ((f >> 31) - 1) | 0x80000000;\n"
+	"	unsigned int fl = f ^ mask;\n"
+	"	return *(float*)&fl;\n"
+	"}\n"
+	"__kernel void   copyAabbsKernel( __global const btAabbCL* allAabbs, __global btAabbCL* destAabbs, int numObjects)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	if (i>=numObjects)\n"
+	"		return;\n"
+	"	int src = destAabbs[i].m_maxIndices[3];\n"
+	"	destAabbs[i] = allAabbs[src];\n"
+	"	destAabbs[i].m_maxIndices[3] = src;\n"
+	"}\n"
+	"__kernel void   flipFloatKernel( __global const btAabbCL* allAabbs, __global const int* smallAabbMapping, __global int2* sortData, int numObjects, int axis)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	if (i>=numObjects)\n"
+	"		return;\n"
+	"	\n"
+	"	\n"
+	"	sortData[i].x = FloatFlip(allAabbs[smallAabbMapping[i]].m_minElems[axis]);\n"
+	"	sortData[i].y = i;\n"
+	"		\n"
+	"}\n"
+	"__kernel void   scatterKernel( __global const btAabbCL* allAabbs, __global const int* smallAabbMapping, volatile __global const int2* sortData, __global btAabbCL* sortedAabbs, int numObjects)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	if (i>=numObjects)\n"
+	"		return;\n"
+	"	\n"
+	"	sortedAabbs[i] = allAabbs[smallAabbMapping[sortData[i].y]];\n"
+	"}\n"
+	"__kernel void   prepareSumVarianceKernel( __global const btAabbCL* allAabbs, __global const int* smallAabbMapping, __global float4* sum, __global float4* sum2,int numAabbs)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	if (i>=numAabbs)\n"
+	"		return;\n"
+	"	\n"
+	"	btAabbCL smallAabb = allAabbs[smallAabbMapping[i]];\n"
+	"	\n"
+	"	float4 s;\n"
+	"	s = (smallAabb.m_max+smallAabb.m_min)*0.5f;\n"
+	"	sum[i]=s;\n"
+	"	sum2[i]=s*s;	\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/Initialize/b3OpenCLInclude.h b/thirdparty/bullet/Bullet3OpenCL/Initialize/b3OpenCLInclude.h
index e79182d7cb8..61465382630 100644
--- a/thirdparty/bullet/Bullet3OpenCL/Initialize/b3OpenCLInclude.h
+++ b/thirdparty/bullet/Bullet3OpenCL/Initialize/b3OpenCLInclude.h
@@ -17,7 +17,7 @@ subject to the following restrictions:
 #define B3_OPENCL_INCLUDE_H
 
 #ifdef B3_USE_CLEW
-	#include "clew/clew.h"
+#include "clew/clew.h"
 #else
 
 #ifdef __APPLE__
@@ -25,7 +25,7 @@ subject to the following restrictions:
 #include <MiniCL/cl.h>
 #else
 #include <OpenCL/cl.h>
-#include <OpenCL/cl_ext.h> //clLogMessagesToStderrAPPLE
+#include <OpenCL/cl_ext.h>  //clLogMessagesToStderrAPPLE
 #endif
 #else
 #ifdef USE_MINICL
@@ -34,15 +34,18 @@ subject to the following restrictions:
 #include <CL/cl.h>
 #ifdef _WIN32
 #include "CL/cl_gl.h"
-#endif //_WIN32
+#endif  //_WIN32
 #endif
-#endif //__APPLE__
-#endif //B3_USE_CLEW
+#endif  //__APPLE__
+#endif  //B3_USE_CLEW
 
 #include <assert.h>
 #include <stdio.h>
-#define oclCHECKERROR(a, b) if((a)!=(b)) { printf("OCL Error : %d\n", (a)); assert((a) == (b)); }
-
-
-#endif //B3_OPENCL_INCLUDE_H
+#define oclCHECKERROR(a, b)              \
+	if ((a) != (b))                      \
+	{                                    \
+		printf("OCL Error : %d\n", (a)); \
+		assert((a) == (b));              \
+	}
 
+#endif  //B3_OPENCL_INCLUDE_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/Initialize/b3OpenCLUtils.cpp b/thirdparty/bullet/Bullet3OpenCL/Initialize/b3OpenCLUtils.cpp
index 896191c89c9..fe54ea5ec90 100644
--- a/thirdparty/bullet/Bullet3OpenCL/Initialize/b3OpenCLUtils.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/Initialize/b3OpenCLUtils.cpp
@@ -16,7 +16,6 @@ subject to the following restrictions:
 //Original author: Roman Ponomarev
 //Mostly Reimplemented by Erwin Coumans
 
-
 bool gDebugForceLoadingFromSource = false;
 bool gDebugSkipLoadingBinary = false;
 
@@ -25,7 +24,7 @@ bool gDebugSkipLoadingBinary = false;
 #include <string.h>
 
 #ifdef _WIN32
-#pragma warning (disable:4996)
+#pragma warning(disable : 4996)
 #endif
 #include "b3OpenCLUtils.h"
 //#include "b3OpenCLInclude.h"
@@ -33,7 +32,7 @@ bool gDebugSkipLoadingBinary = false;
 #include <stdio.h>
 #include <stdlib.h>
 
-#define B3_MAX_CL_DEVICES 16 //who needs 16 devices?
+#define B3_MAX_CL_DEVICES 16  //who needs 16 devices?
 
 #ifdef _WIN32
 #include <windows.h>
@@ -46,53 +45,49 @@ bool gDebugSkipLoadingBinary = false;
 
 #endif
 
-static const char* sCachedBinaryPath="cache";
-
+static const char* sCachedBinaryPath = "cache";
 
 //Set the preferred platform vendor using the OpenCL SDK
 static const char* spPlatformVendor =
 #if defined(CL_PLATFORM_MINI_CL)
-"MiniCL, SCEA";
+	"MiniCL, SCEA";
 #elif defined(CL_PLATFORM_AMD)
-"Advanced Micro Devices, Inc.";
+	"Advanced Micro Devices, Inc.";
 #elif defined(CL_PLATFORM_NVIDIA)
-"NVIDIA Corporation";
+	"NVIDIA Corporation";
 #elif defined(CL_PLATFORM_INTEL)
-"Intel(R) Corporation";
+	"Intel(R) Corporation";
 #elif defined(B3_USE_CLEW)
-"clew (OpenCL Extension Wrangler library)";
+	"clew (OpenCL Extension Wrangler library)";
 #else
-"Unknown Vendor";
+	"Unknown Vendor";
 #endif
 
 #ifndef CL_PLATFORM_MINI_CL
 #ifdef _WIN32
 #ifndef B3_USE_CLEW
 #include "CL/cl_gl.h"
-#endif //B3_USE_CLEW
-#endif //_WIN32
+#endif  //B3_USE_CLEW
+#endif  //_WIN32
 #endif
 
-
-void MyFatalBreakAPPLE(   const char *  errstr ,
-                       const void *  private_info ,
-                       size_t        cb ,
-                       void *        user_data  )
+void MyFatalBreakAPPLE(const char* errstr,
+					   const void* private_info,
+					   size_t cb,
+					   void* user_data)
 {
-   
+	const char* patloc = strstr(errstr, "Warning");
+	//find out if it is a warning or error, exit if error
 
-    const char* patloc = strstr(errstr, "Warning");
-    //find out if it is a warning or error, exit if error
-
-    if (patloc)
-    {
+	if (patloc)
+	{
 		b3Warning("Warning: %s\n", errstr);
-    } else
-    {
+	}
+	else
+	{
 		b3Error("Error: %s\n", errstr);
-        b3Assert(0);
-    }
-
+		b3Assert(0);
+	}
 }
 
 #ifdef B3_USE_CLEW
@@ -102,30 +97,31 @@ int b3OpenCLUtils_clewInit()
 	int result = -1;
 
 #ifdef _WIN32
-        const char* cl = "OpenCL.dll";
+	const char* cl = "OpenCL.dll";
 #elif defined __APPLE__
-        const char* cl = "/System/Library/Frameworks/OpenCL.framework/Versions/Current/OpenCL";
-#else//presumable Linux?
-        //linux (tested on Ubuntu 12.10 with Catalyst 13.4 beta drivers, not that there is no symbolic link from libOpenCL.so
-        const char* cl = "libOpenCL.so.1";
-        result = clewInit(cl);
-        if (result != CLEW_SUCCESS)
-        {
-                cl = "libOpenCL.so";
-        } else
-        {
-                clewExit();
-        }
+	const char* cl = "/System/Library/Frameworks/OpenCL.framework/Versions/Current/OpenCL";
+#else  //presumable Linux? \
+	   //linux (tested on Ubuntu 12.10 with Catalyst 13.4 beta drivers, not that there is no symbolic link from libOpenCL.so
+	const char* cl = "libOpenCL.so.1";
+	result = clewInit(cl);
+	if (result != CLEW_SUCCESS)
+	{
+		cl = "libOpenCL.so";
+	}
+	else
+	{
+		clewExit();
+	}
 #endif
-        result = clewInit(cl);
-        if (result!=CLEW_SUCCESS)
-		{
-                b3Error("clewInit failed with error code %d\n",result);
-		}
-        else
-        {
-                b3Printf("clewInit succesfull using %s\n",cl);
-        }
+	result = clewInit(cl);
+	if (result != CLEW_SUCCESS)
+	{
+		b3Error("clewInit failed with error code %d\n", result);
+	}
+	else
+	{
+		b3Printf("clewInit succesfull using %s\n", cl);
+	}
 	return result;
 }
 #endif
@@ -136,19 +132,18 @@ int b3OpenCLUtils_getNumPlatforms(cl_int* pErrNum)
 	b3OpenCLUtils_clewInit();
 #endif
 
-	cl_platform_id pPlatforms[10] = { 0 };
+	cl_platform_id pPlatforms[10] = {0};
 
-    cl_uint numPlatforms = 0;
-    cl_int ciErrNum = clGetPlatformIDs(10, pPlatforms, &numPlatforms);
+	cl_uint numPlatforms = 0;
+	cl_int ciErrNum = clGetPlatformIDs(10, pPlatforms, &numPlatforms);
 	//cl_int ciErrNum = clGetPlatformIDs(0, NULL, &numPlatforms);
 
-	if(ciErrNum != CL_SUCCESS)
+	if (ciErrNum != CL_SUCCESS)
 	{
-		if(pErrNum != NULL)
+		if (pErrNum != NULL)
 			*pErrNum = ciErrNum;
 	}
 	return numPlatforms;
-
 }
 
 const char* b3OpenCLUtils_getSdkVendorName()
@@ -164,28 +159,28 @@ void b3OpenCLUtils_setCachePath(const char* path)
 cl_platform_id b3OpenCLUtils_getPlatform(int platformIndex0, cl_int* pErrNum)
 {
 #ifdef B3_USE_CLEW
-        b3OpenCLUtils_clewInit();
+	b3OpenCLUtils_clewInit();
 #endif
 
 	cl_platform_id platform = 0;
-	unsigned int platformIndex = (unsigned int )platformIndex0;
+	unsigned int platformIndex = (unsigned int)platformIndex0;
 	cl_uint numPlatforms;
 	cl_int ciErrNum = clGetPlatformIDs(0, NULL, &numPlatforms);
 
-	if (platformIndex<numPlatforms)
+	if (platformIndex < numPlatforms)
 	{
-		cl_platform_id* platforms = (cl_platform_id*) malloc (sizeof(cl_platform_id)*numPlatforms);
+		cl_platform_id* platforms = (cl_platform_id*)malloc(sizeof(cl_platform_id) * numPlatforms);
 		ciErrNum = clGetPlatformIDs(numPlatforms, platforms, NULL);
-		if(ciErrNum != CL_SUCCESS)
+		if (ciErrNum != CL_SUCCESS)
 		{
-			if(pErrNum != NULL)
+			if (pErrNum != NULL)
 				*pErrNum = ciErrNum;
 			return platform;
 		}
 
 		platform = platforms[platformIndex];
 
-		free (platforms);
+		free(platforms);
 	}
 
 	return platform;
@@ -195,30 +190,28 @@ void b3OpenCLUtils::getPlatformInfo(cl_platform_id platform, b3OpenCLPlatformInf
 {
 	b3Assert(platform);
 	cl_int ciErrNum;
-	ciErrNum = clGetPlatformInfo(	platform,CL_PLATFORM_VENDOR,B3_MAX_STRING_LENGTH,platformInfo->m_platformVendor,NULL);
-	oclCHECKERROR(ciErrNum,CL_SUCCESS);
-	ciErrNum = clGetPlatformInfo(	platform,CL_PLATFORM_NAME,B3_MAX_STRING_LENGTH,platformInfo->m_platformName,NULL);
-	oclCHECKERROR(ciErrNum,CL_SUCCESS);
-	ciErrNum = clGetPlatformInfo(	platform,CL_PLATFORM_VERSION,B3_MAX_STRING_LENGTH,platformInfo->m_platformVersion,NULL);
-	oclCHECKERROR(ciErrNum,CL_SUCCESS);
+	ciErrNum = clGetPlatformInfo(platform, CL_PLATFORM_VENDOR, B3_MAX_STRING_LENGTH, platformInfo->m_platformVendor, NULL);
+	oclCHECKERROR(ciErrNum, CL_SUCCESS);
+	ciErrNum = clGetPlatformInfo(platform, CL_PLATFORM_NAME, B3_MAX_STRING_LENGTH, platformInfo->m_platformName, NULL);
+	oclCHECKERROR(ciErrNum, CL_SUCCESS);
+	ciErrNum = clGetPlatformInfo(platform, CL_PLATFORM_VERSION, B3_MAX_STRING_LENGTH, platformInfo->m_platformVersion, NULL);
+	oclCHECKERROR(ciErrNum, CL_SUCCESS);
 }
 
-void b3OpenCLUtils_printPlatformInfo( cl_platform_id platform)
+void b3OpenCLUtils_printPlatformInfo(cl_platform_id platform)
 {
 	b3OpenCLPlatformInfo platformInfo;
-	b3OpenCLUtils::getPlatformInfo (platform, &platformInfo);
+	b3OpenCLUtils::getPlatformInfo(platform, &platformInfo);
 	b3Printf("Platform info:\n");
-	b3Printf("  CL_PLATFORM_VENDOR: \t\t\t%s\n",platformInfo.m_platformVendor);
-	b3Printf("  CL_PLATFORM_NAME: \t\t\t%s\n",platformInfo.m_platformName);
-	b3Printf("  CL_PLATFORM_VERSION: \t\t\t%s\n",platformInfo.m_platformVersion);
+	b3Printf("  CL_PLATFORM_VENDOR: \t\t\t%s\n", platformInfo.m_platformVendor);
+	b3Printf("  CL_PLATFORM_NAME: \t\t\t%s\n", platformInfo.m_platformName);
+	b3Printf("  CL_PLATFORM_VERSION: \t\t\t%s\n", platformInfo.m_platformVersion);
 }
 
-
-
 cl_context b3OpenCLUtils_createContextFromPlatform(cl_platform_id platform, cl_device_type deviceType, cl_int* pErrNum, void* pGLContext, void* pGLDC, int preferredDeviceIndex, int preferredPlatformIndex)
 {
 	cl_context retContext = 0;
-	cl_int ciErrNum=0;
+	cl_int ciErrNum = 0;
 	cl_uint num_entries;
 	cl_device_id devices[B3_MAX_CL_DEVICES];
 	cl_uint num_devices;
@@ -228,7 +221,7 @@ cl_context b3OpenCLUtils_createContextFromPlatform(cl_platform_id platform, cl_d
 	* If we could find our platform, use it. Otherwise pass a NULL and get whatever the
 	* implementation thinks we should be using.
 	*/
-	cl_context_properties cps[7] = {0,0,0,0,0,0,0};
+	cl_context_properties cps[7] = {0, 0, 0, 0, 0, 0, 0};
 	cps[0] = CL_CONTEXT_PLATFORM;
 	cps[1] = (cl_context_properties)platform;
 #ifdef _WIN32
@@ -240,25 +233,24 @@ cl_context b3OpenCLUtils_createContextFromPlatform(cl_platform_id platform, cl_d
 		cps[4] = CL_WGL_HDC_KHR;
 		cps[5] = (cl_context_properties)pGLDC;
 	}
-#endif //B3_USE_CLEW
-#endif //_WIN32
+#endif  //B3_USE_CLEW
+#endif  //_WIN32
 	num_entries = B3_MAX_CL_DEVICES;
 
-
-	num_devices=-1;
+	num_devices = -1;
 
 	ciErrNum = clGetDeviceIDs(
 		platform,
 		deviceType,
- 		num_entries,
- 		devices,
- 		&num_devices);
+		num_entries,
+		devices,
+		&num_devices);
 
-    if (ciErrNum<0)
-    {
-        b3Printf("clGetDeviceIDs returned %d\n",ciErrNum);
-        return 0;
-    }
+	if (ciErrNum < 0)
+	{
+		b3Printf("clGetDeviceIDs returned %d\n", ciErrNum);
+		return 0;
+	}
 	cprops = (NULL == platform) ? NULL : cps;
 
 	if (!num_devices)
@@ -268,32 +260,33 @@ cl_context b3OpenCLUtils_createContextFromPlatform(cl_platform_id platform, cl_d
 	{
 		//search for the GPU that relates to the OpenCL context
 		unsigned int i;
-		for (i=0;i<num_devices;i++)
+		for (i = 0; i < num_devices; i++)
 		{
-			retContext = clCreateContext(cprops,1,&devices[i],NULL,NULL,&ciErrNum);
-			if (ciErrNum==CL_SUCCESS)
+			retContext = clCreateContext(cprops, 1, &devices[i], NULL, NULL, &ciErrNum);
+			if (ciErrNum == CL_SUCCESS)
 				break;
 		}
 	}
 	else
 	{
-		if (preferredDeviceIndex>=0 && (unsigned int)preferredDeviceIndex<num_devices)
+		if (preferredDeviceIndex >= 0 && (unsigned int)preferredDeviceIndex < num_devices)
 		{
 			//create a context of the preferred device index
-			retContext = clCreateContext(cprops,1,&devices[preferredDeviceIndex],NULL,NULL,&ciErrNum);
-		} else
+			retContext = clCreateContext(cprops, 1, &devices[preferredDeviceIndex], NULL, NULL, &ciErrNum);
+		}
+		else
 		{
 			//create a context of all devices
-#if defined (__APPLE__)
-			retContext = clCreateContext(cprops,num_devices,devices,MyFatalBreakAPPLE,NULL,&ciErrNum);
+#if defined(__APPLE__)
+			retContext = clCreateContext(cprops, num_devices, devices, MyFatalBreakAPPLE, NULL, &ciErrNum);
 #else
-        b3Printf("numDevices=%d\n",num_devices);
+			b3Printf("numDevices=%d\n", num_devices);
 
-			retContext = clCreateContext(cprops,num_devices,devices,NULL,NULL,&ciErrNum);
+			retContext = clCreateContext(cprops, num_devices, devices, NULL, NULL, &ciErrNum);
 #endif
 		}
 	}
-	if(pErrNum != NULL)
+	if (pErrNum != NULL)
 	{
 		*pErrNum = ciErrNum;
 	};
@@ -301,60 +294,58 @@ cl_context b3OpenCLUtils_createContextFromPlatform(cl_platform_id platform, cl_d
 	return retContext;
 }
 
-cl_context b3OpenCLUtils_createContextFromType(cl_device_type deviceType, cl_int* pErrNum, void* pGLContext, void* pGLDC , int preferredDeviceIndex, int preferredPlatformIndex, cl_platform_id* retPlatformId)
+cl_context b3OpenCLUtils_createContextFromType(cl_device_type deviceType, cl_int* pErrNum, void* pGLContext, void* pGLDC, int preferredDeviceIndex, int preferredPlatformIndex, cl_platform_id* retPlatformId)
 {
 #ifdef B3_USE_CLEW
-        b3OpenCLUtils_clewInit();
+	b3OpenCLUtils_clewInit();
 #endif
 
-
 	cl_uint numPlatforms;
 	cl_context retContext = 0;
 	unsigned int i;
 
 	cl_int ciErrNum = clGetPlatformIDs(0, NULL, &numPlatforms);
-	if(ciErrNum != CL_SUCCESS)
+	if (ciErrNum != CL_SUCCESS)
 	{
-		if(pErrNum != NULL) *pErrNum = ciErrNum;
+		if (pErrNum != NULL) *pErrNum = ciErrNum;
 		return NULL;
 	}
-	if(numPlatforms > 0)
+	if (numPlatforms > 0)
 	{
-		cl_platform_id* platforms = (cl_platform_id*) malloc (sizeof(cl_platform_id)*numPlatforms);
+		cl_platform_id* platforms = (cl_platform_id*)malloc(sizeof(cl_platform_id) * numPlatforms);
 		ciErrNum = clGetPlatformIDs(numPlatforms, platforms, NULL);
-		if(ciErrNum != CL_SUCCESS)
+		if (ciErrNum != CL_SUCCESS)
 		{
-			if(pErrNum != NULL)
+			if (pErrNum != NULL)
 				*pErrNum = ciErrNum;
 			free(platforms);
 			return NULL;
 		}
 
-
-
-		for ( i = 0; i < numPlatforms; ++i)
+		for (i = 0; i < numPlatforms; ++i)
 		{
 			char pbuf[128];
-			ciErrNum = clGetPlatformInfo(	platforms[i],
-				CL_PLATFORM_VENDOR,
-				sizeof(pbuf),
-				pbuf,
-				NULL);
-			if(ciErrNum != CL_SUCCESS)
+			ciErrNum = clGetPlatformInfo(platforms[i],
+										 CL_PLATFORM_VENDOR,
+										 sizeof(pbuf),
+										 pbuf,
+										 NULL);
+			if (ciErrNum != CL_SUCCESS)
 			{
-				if(pErrNum != NULL) *pErrNum = ciErrNum;
+				if (pErrNum != NULL) *pErrNum = ciErrNum;
 				return NULL;
 			}
 
-			if (preferredPlatformIndex>=0 && i==preferredPlatformIndex)
+			if (preferredPlatformIndex >= 0 && i == preferredPlatformIndex)
 			{
 				cl_platform_id tmpPlatform = platforms[0];
 				platforms[0] = platforms[i];
 				platforms[i] = tmpPlatform;
 				break;
-			} else
+			}
+			else
 			{
-				if(!strcmp(pbuf, spPlatformVendor))
+				if (!strcmp(pbuf, spPlatformVendor))
 				{
 					cl_platform_id tmpPlatform = platforms[0];
 					platforms[0] = platforms[i];
@@ -368,11 +359,11 @@ cl_context b3OpenCLUtils_createContextFromType(cl_device_type deviceType, cl_int
 			cl_platform_id platform = platforms[i];
 			assert(platform);
 
-			retContext = b3OpenCLUtils_createContextFromPlatform(platform,deviceType,pErrNum,pGLContext,pGLDC,preferredDeviceIndex,preferredPlatformIndex);
+			retContext = b3OpenCLUtils_createContextFromPlatform(platform, deviceType, pErrNum, pGLContext, pGLDC, preferredDeviceIndex, preferredPlatformIndex);
 
 			if (retContext)
 			{
-//				printf("OpenCL platform details:\n");
+				//				printf("OpenCL platform details:\n");
 				b3OpenCLPlatformInfo platformInfo;
 
 				b3OpenCLUtils::getPlatformInfo(platform, &platformInfo);
@@ -384,12 +375,11 @@ cl_context b3OpenCLUtils_createContextFromType(cl_device_type deviceType, cl_int
 			}
 		}
 
-		free (platforms);
+		free(platforms);
 	}
 	return retContext;
 }
 
-
 //////////////////////////////////////////////////////////////////////////////
 //! Gets the id of the nth device from the context
 //!
@@ -403,16 +393,17 @@ cl_device_id b3OpenCLUtils_getDevice(cl_context cxMainContext, int deviceIndex)
 
 	size_t szParmDataBytes;
 	cl_device_id* cdDevices;
-	cl_device_id device ;
+	cl_device_id device;
 
 	// get the list of devices associated with context
 	clGetContextInfo(cxMainContext, CL_CONTEXT_DEVICES, 0, NULL, &szParmDataBytes);
 
-	if( szParmDataBytes / sizeof(cl_device_id) < (unsigned int)deviceIndex ) {
+	if (szParmDataBytes / sizeof(cl_device_id) < (unsigned int)deviceIndex)
+	{
 		return (cl_device_id)-1;
 	}
 
-	cdDevices = (cl_device_id*) malloc(szParmDataBytes);
+	cdDevices = (cl_device_id*)malloc(szParmDataBytes);
 
 	clGetContextInfo(cxMainContext, CL_CONTEXT_DEVICES, szParmDataBytes, cdDevices, NULL);
 
@@ -427,12 +418,10 @@ int b3OpenCLUtils_getNumDevices(cl_context cxMainContext)
 	size_t szParamDataBytes;
 	int device_count;
 	clGetContextInfo(cxMainContext, CL_CONTEXT_DEVICES, 0, NULL, &szParamDataBytes);
-	device_count = (int) szParamDataBytes/ sizeof(cl_device_id);
+	device_count = (int)szParamDataBytes / sizeof(cl_device_id);
 	return device_count;
 }
 
-
-
 void b3OpenCLUtils::getDeviceInfo(cl_device_id device, b3OpenCLDeviceInfo* info)
 {
 	// CL_DEVICE_NAME
@@ -514,23 +503,22 @@ void b3OpenCLUtils::getDeviceInfo(cl_device_id device, b3OpenCLDeviceInfo* info)
 	clGetDeviceInfo(device, CL_DEVICE_PREFERRED_VECTOR_WIDTH_DOUBLE, sizeof(cl_uint), &info->m_vecWidthDouble, NULL);
 }
 
-
 void b3OpenCLUtils_printDeviceInfo(cl_device_id device)
 {
 	b3OpenCLDeviceInfo info;
-	b3OpenCLUtils::getDeviceInfo(device,&info);
+	b3OpenCLUtils::getDeviceInfo(device, &info);
 	b3Printf("Device Info:\n");
 	b3Printf("  CL_DEVICE_NAME: \t\t\t%s\n", info.m_deviceName);
 	b3Printf("  CL_DEVICE_VENDOR: \t\t\t%s\n", info.m_deviceVendor);
 	b3Printf("  CL_DRIVER_VERSION: \t\t\t%s\n", info.m_driverVersion);
 
-	if( info.m_deviceType & CL_DEVICE_TYPE_CPU )
+	if (info.m_deviceType & CL_DEVICE_TYPE_CPU)
 		b3Printf("  CL_DEVICE_TYPE:\t\t\t%s\n", "CL_DEVICE_TYPE_CPU");
-	if( info.m_deviceType & CL_DEVICE_TYPE_GPU )
+	if (info.m_deviceType & CL_DEVICE_TYPE_GPU)
 		b3Printf("  CL_DEVICE_TYPE:\t\t\t%s\n", "CL_DEVICE_TYPE_GPU");
-	if( info.m_deviceType & CL_DEVICE_TYPE_ACCELERATOR )
+	if (info.m_deviceType & CL_DEVICE_TYPE_ACCELERATOR)
 		b3Printf("  CL_DEVICE_TYPE:\t\t\t%s\n", "CL_DEVICE_TYPE_ACCELERATOR");
-	if( info.m_deviceType & CL_DEVICE_TYPE_DEFAULT )
+	if (info.m_deviceType & CL_DEVICE_TYPE_DEFAULT)
 		b3Printf("  CL_DEVICE_TYPE:\t\t\t%s\n", "CL_DEVICE_TYPE_DEFAULT");
 
 	b3Printf("  CL_DEVICE_MAX_COMPUTE_UNITS:\t\t%u\n", info.m_computeUnits);
@@ -539,15 +527,15 @@ void b3OpenCLUtils_printDeviceInfo(cl_device_id device)
 	b3Printf("  CL_DEVICE_MAX_WORK_GROUP_SIZE:\t%u\n", info.m_workgroupSize);
 	b3Printf("  CL_DEVICE_MAX_CLOCK_FREQUENCY:\t%u MHz\n", info.m_clockFrequency);
 	b3Printf("  CL_DEVICE_ADDRESS_BITS:\t\t%u\n", info.m_addressBits);
-	b3Printf("  CL_DEVICE_MAX_MEM_ALLOC_SIZE:\t\t%u MByte\n", (unsigned int)(info.m_maxMemAllocSize/ (1024 * 1024)));
-	b3Printf("  CL_DEVICE_GLOBAL_MEM_SIZE:\t\t%u MByte\n", (unsigned int)(info.m_globalMemSize/ (1024 * 1024)));
-	b3Printf("  CL_DEVICE_ERROR_CORRECTION_SUPPORT:\t%s\n", info.m_errorCorrectionSupport== CL_TRUE ? "yes" : "no");
+	b3Printf("  CL_DEVICE_MAX_MEM_ALLOC_SIZE:\t\t%u MByte\n", (unsigned int)(info.m_maxMemAllocSize / (1024 * 1024)));
+	b3Printf("  CL_DEVICE_GLOBAL_MEM_SIZE:\t\t%u MByte\n", (unsigned int)(info.m_globalMemSize / (1024 * 1024)));
+	b3Printf("  CL_DEVICE_ERROR_CORRECTION_SUPPORT:\t%s\n", info.m_errorCorrectionSupport == CL_TRUE ? "yes" : "no");
 	b3Printf("  CL_DEVICE_LOCAL_MEM_TYPE:\t\t%s\n", info.m_localMemType == 1 ? "local" : "global");
 	b3Printf("  CL_DEVICE_LOCAL_MEM_SIZE:\t\t%u KByte\n", (unsigned int)(info.m_localMemSize / 1024));
 	b3Printf("  CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE:\t%u KByte\n", (unsigned int)(info.m_constantBufferSize / 1024));
-	if( info.m_queueProperties  & CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE )
+	if (info.m_queueProperties & CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE)
 		b3Printf("  CL_DEVICE_QUEUE_PROPERTIES:\t\t%s\n", "CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE");
-	if( info.m_queueProperties & CL_QUEUE_PROFILING_ENABLE )
+	if (info.m_queueProperties & CL_QUEUE_PROFILING_ENABLE)
 		b3Printf("  CL_DEVICE_QUEUE_PROPERTIES:\t\t%s\n", "CL_QUEUE_PROFILING_ENABLE");
 
 	b3Printf("  CL_DEVICE_IMAGE_SUPPORT:\t\t%u\n", info.m_imageSupport);
@@ -562,7 +550,7 @@ void b3OpenCLUtils_printDeviceInfo(cl_device_id device)
 	b3Printf("\t\t\t\t\t3D_MAX_DEPTH\t %u\n", info.m_image3dMaxDepth);
 	if (*info.m_deviceExtensions != 0)
 	{
-		b3Printf("\n  CL_DEVICE_EXTENSIONS:%s\n",info.m_deviceExtensions);
+		b3Printf("\n  CL_DEVICE_EXTENSIONS:%s\n", info.m_deviceExtensions);
 	}
 	else
 	{
@@ -570,36 +558,33 @@ void b3OpenCLUtils_printDeviceInfo(cl_device_id device)
 	}
 	b3Printf("  CL_DEVICE_PREFERRED_VECTOR_WIDTH_<t>\t");
 	b3Printf("CHAR %u, SHORT %u, INT %u,LONG %u, FLOAT %u, DOUBLE %u\n\n\n",
-		info.m_vecWidthChar, info.m_vecWidthShort, info.m_vecWidthInt, info.m_vecWidthLong,info.m_vecWidthFloat, info.m_vecWidthDouble);
-
-
+			 info.m_vecWidthChar, info.m_vecWidthShort, info.m_vecWidthInt, info.m_vecWidthLong, info.m_vecWidthFloat, info.m_vecWidthDouble);
 }
 
-
 static const char* strip2(const char* name, const char* pattern)
 {
-	  size_t const patlen = strlen(pattern);
-  	size_t patcnt = 0;
-	  const char * oriptr;
-	  const char * patloc;
-		// find how many times the pattern occurs in the original string
-	  for (oriptr = name; (patloc = strstr(oriptr, pattern)); oriptr = patloc + patlen)
-	  {
+	size_t const patlen = strlen(pattern);
+	size_t patcnt = 0;
+	const char* oriptr;
+	const char* patloc;
+	// find how many times the pattern occurs in the original string
+	for (oriptr = name; (patloc = strstr(oriptr, pattern)); oriptr = patloc + patlen)
+	{
 		patcnt++;
-	  }
-	  return oriptr;
+	}
+	return oriptr;
 }
 
-cl_program b3OpenCLUtils_compileCLProgramFromString(cl_context clContext, cl_device_id device, const char* kernelSourceOrg, cl_int* pErrNum, const char* additionalMacrosArg , const char* clFileNameForCaching, bool disableBinaryCaching)
+cl_program b3OpenCLUtils_compileCLProgramFromString(cl_context clContext, cl_device_id device, const char* kernelSourceOrg, cl_int* pErrNum, const char* additionalMacrosArg, const char* clFileNameForCaching, bool disableBinaryCaching)
 {
-	const char* additionalMacros = additionalMacrosArg?additionalMacrosArg:"";
+	const char* additionalMacros = additionalMacrosArg ? additionalMacrosArg : "";
 
 	if (disableBinaryCaching)
 	{
 		//kernelSourceOrg = 0;
 	}
 
-	cl_program m_cpProgram=0;
+	cl_program m_cpProgram = 0;
 	cl_int status;
 
 	char binaryFileName[B3_MAX_STRING_LENGTH];
@@ -609,67 +594,64 @@ cl_program b3OpenCLUtils_compileCLProgramFromString(cl_context clContext, cl_dev
 	const char* strippedName;
 	int fileUpToDate = 0;
 #ifdef _WIN32
-	int binaryFileValid=0;
-#endif	
+	int binaryFileValid = 0;
+#endif
 	if (!disableBinaryCaching && clFileNameForCaching)
 	{
 		clGetDeviceInfo(device, CL_DEVICE_NAME, 256, &deviceName, NULL);
 		clGetDeviceInfo(device, CL_DRIVER_VERSION, 256, &driverVersion, NULL);
-	
-		strippedName = strip2(clFileNameForCaching,"\\");
-		strippedName = strip2(strippedName,"/");
-	
+
+		strippedName = strip2(clFileNameForCaching, "\\");
+		strippedName = strip2(strippedName, "/");
+
 #ifdef _MSC_VER
-		sprintf_s(binaryFileName,B3_MAX_STRING_LENGTH,"%s/%s.%s.%s.bin",sCachedBinaryPath,strippedName, deviceName,driverVersion );
+		sprintf_s(binaryFileName, B3_MAX_STRING_LENGTH, "%s/%s.%s.%s.bin", sCachedBinaryPath, strippedName, deviceName, driverVersion);
 #else
-		sprintf(binaryFileName,"%s/%s.%s.%s.bin",sCachedBinaryPath,strippedName, deviceName,driverVersion );
+		sprintf(binaryFileName, "%s/%s.%s.%s.bin", sCachedBinaryPath, strippedName, deviceName, driverVersion);
 #endif
 	}
-	if (clFileNameForCaching && !(disableBinaryCaching || gDebugSkipLoadingBinary||gDebugForceLoadingFromSource) )
+	if (clFileNameForCaching && !(disableBinaryCaching || gDebugSkipLoadingBinary || gDebugForceLoadingFromSource))
 	{
-		
 #ifdef _WIN32
-	char* bla=0;
-
-	
+		char* bla = 0;
 
 		//printf("searching for %s\n", binaryFileName);
 
-
 		FILETIME modtimeBinary;
-		CreateDirectoryA(sCachedBinaryPath,0);
+		CreateDirectoryA(sCachedBinaryPath, 0);
 		{
-
-			HANDLE binaryFileHandle = CreateFileA(binaryFileName,GENERIC_READ,0,0,OPEN_EXISTING,FILE_ATTRIBUTE_NORMAL,0);
-			if (binaryFileHandle ==INVALID_HANDLE_VALUE)
+			HANDLE binaryFileHandle = CreateFileA(binaryFileName, GENERIC_READ, 0, 0, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, 0);
+			if (binaryFileHandle == INVALID_HANDLE_VALUE)
 			{
 				DWORD errorCode;
 				errorCode = GetLastError();
 				switch (errorCode)
 				{
-				case ERROR_FILE_NOT_FOUND:
+					case ERROR_FILE_NOT_FOUND:
 					{
 						b3Warning("\nCached file not found %s\n", binaryFileName);
 						break;
 					}
-				case ERROR_PATH_NOT_FOUND:
+					case ERROR_PATH_NOT_FOUND:
 					{
 						b3Warning("\nCached file path not found %s\n", binaryFileName);
 						break;
 					}
-				default:
+					default:
 					{
 						b3Warning("\nFailed reading cached file with errorCode = %d\n", errorCode);
 					}
 				}
-			} else
+			}
+			else
 			{
-				if (GetFileTime(binaryFileHandle, NULL, NULL, &modtimeBinary)==0)
+				if (GetFileTime(binaryFileHandle, NULL, NULL, &modtimeBinary) == 0)
 				{
 					DWORD errorCode;
 					errorCode = GetLastError();
 					b3Warning("\nGetFileTime errorCode = %d\n", errorCode);
-				} else
+				}
+				else
 				{
 					binaryFileValid = 1;
 				}
@@ -678,37 +660,35 @@ cl_program b3OpenCLUtils_compileCLProgramFromString(cl_context clContext, cl_dev
 
 			if (binaryFileValid)
 			{
-				HANDLE srcFileHandle = CreateFileA(clFileNameForCaching,GENERIC_READ,0,0,OPEN_EXISTING,FILE_ATTRIBUTE_NORMAL,0);
+				HANDLE srcFileHandle = CreateFileA(clFileNameForCaching, GENERIC_READ, 0, 0, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, 0);
 
-				if (srcFileHandle==INVALID_HANDLE_VALUE)
+				if (srcFileHandle == INVALID_HANDLE_VALUE)
 				{
-					const char* prefix[]={"./","../","../../","../../../","../../../../"};
-					for (int i=0;(srcFileHandle==INVALID_HANDLE_VALUE) && i<5;i++)
+					const char* prefix[] = {"./", "../", "../../", "../../../", "../../../../"};
+					for (int i = 0; (srcFileHandle == INVALID_HANDLE_VALUE) && i < 5; i++)
 					{
 						char relativeFileName[1024];
-						sprintf(relativeFileName,"%s%s",prefix[i],clFileNameForCaching);
-						srcFileHandle = CreateFileA(relativeFileName,GENERIC_READ,0,0,OPEN_EXISTING,FILE_ATTRIBUTE_NORMAL,0);
+						sprintf(relativeFileName, "%s%s", prefix[i], clFileNameForCaching);
+						srcFileHandle = CreateFileA(relativeFileName, GENERIC_READ, 0, 0, OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, 0);
 					}
-
 				}
 
-
-				if (srcFileHandle!=INVALID_HANDLE_VALUE)
+				if (srcFileHandle != INVALID_HANDLE_VALUE)
 				{
 					FILETIME modtimeSrc;
-					if (GetFileTime(srcFileHandle, NULL, NULL, &modtimeSrc)==0)
+					if (GetFileTime(srcFileHandle, NULL, NULL, &modtimeSrc) == 0)
 					{
 						DWORD errorCode;
 						errorCode = GetLastError();
 						b3Warning("\nGetFileTime errorCode = %d\n", errorCode);
 					}
-					if (  ( modtimeSrc.dwHighDateTime < modtimeBinary.dwHighDateTime)
-						||(( modtimeSrc.dwHighDateTime == modtimeBinary.dwHighDateTime)&&(modtimeSrc.dwLowDateTime <= modtimeBinary.dwLowDateTime)))
+					if ((modtimeSrc.dwHighDateTime < modtimeBinary.dwHighDateTime) || ((modtimeSrc.dwHighDateTime == modtimeBinary.dwHighDateTime) && (modtimeSrc.dwLowDateTime <= modtimeBinary.dwLowDateTime)))
 					{
-						fileUpToDate=1;
-					} else
+						fileUpToDate = 1;
+					}
+					else
 					{
-						b3Warning("\nCached binary file out-of-date (%s)\n",binaryFileName);
+						b3Warning("\nCached binary file out-of-date (%s)\n", binaryFileName);
 					}
 					CloseHandle(srcFileHandle);
 				}
@@ -719,25 +699,25 @@ cl_program b3OpenCLUtils_compileCLProgramFromString(cl_context clContext, cl_dev
 					errorCode = GetLastError();
 					switch (errorCode)
 					{
-					case ERROR_FILE_NOT_FOUND:
+						case ERROR_FILE_NOT_FOUND:
 						{
 							b3Warning("\nSrc file not found %s\n", clFileNameForCaching);
 							break;
 						}
-					case ERROR_PATH_NOT_FOUND:
+						case ERROR_PATH_NOT_FOUND:
 						{
 							b3Warning("\nSrc path not found %s\n", clFileNameForCaching);
 							break;
 						}
-					default:
+						default:
 						{
 							b3Warning("\nnSrc file reading errorCode = %d\n", errorCode);
 						}
 					}
 
 					//we should make sure the src file exists so we can verify the timestamp with binary
-//					assert(0);
-					b3Warning("Warning: cannot find OpenCL kernel %s to verify timestamp of binary cached kernel %s\n",clFileNameForCaching, binaryFileName);
+					//					assert(0);
+					b3Warning("Warning: cannot find OpenCL kernel %s to verify timestamp of binary cached kernel %s\n", clFileNameForCaching, binaryFileName);
 					fileUpToDate = true;
 #else
 					//if we cannot find the source, assume it is OK in release builds
@@ -745,126 +725,109 @@ cl_program b3OpenCLUtils_compileCLProgramFromString(cl_context clContext, cl_dev
 #endif
 				}
 			}
-
-
 		}
 
-	
-
 #else
-	fileUpToDate = true;
-	if (mkdir(sCachedBinaryPath,0777) == -1)
-	{
+		fileUpToDate = true;
+		if (mkdir(sCachedBinaryPath, 0777) == -1)
+		{
+		}
+		else
+		{
+			b3Printf("Succesfully created cache directory: %s\n", sCachedBinaryPath);
+		}
+#endif  //_WIN32
 	}
-	else
-	{
-		b3Printf("Succesfully created cache directory: %s\n", sCachedBinaryPath);
-	}
-#endif //_WIN32
-	}
-	
-	
-	if( fileUpToDate)
+
+	if (fileUpToDate)
 	{
 #ifdef _MSC_VER
 		FILE* file;
-		if (fopen_s(&file,binaryFileName, "rb")!=0)
-			file=0;
+		if (fopen_s(&file, binaryFileName, "rb") != 0)
+			file = 0;
 #else
 		FILE* file = fopen(binaryFileName, "rb");
 #endif
-		
+
 		if (file)
 		{
-			size_t binarySize=0;
-			char* binary =0;
-			
-			fseek( file, 0L, SEEK_END );
-			binarySize = ftell( file );
-			rewind( file );
-			binary = (char*)malloc(sizeof(char)*binarySize);
+			size_t binarySize = 0;
+			char* binary = 0;
+
+			fseek(file, 0L, SEEK_END);
+			binarySize = ftell(file);
+			rewind(file);
+			binary = (char*)malloc(sizeof(char) * binarySize);
 			int bytesRead;
-			bytesRead = fread( binary, sizeof(char), binarySize, file );
-			fclose( file );
-			
-			m_cpProgram = clCreateProgramWithBinary( clContext, 1,&device, &binarySize, (const unsigned char**)&binary, 0, &status );
-			b3Assert( status == CL_SUCCESS );
-			status = clBuildProgram( m_cpProgram, 1, &device, additionalMacros, 0, 0 );
-			b3Assert( status == CL_SUCCESS );
-			
-			if( status != CL_SUCCESS )
+			bytesRead = fread(binary, sizeof(char), binarySize, file);
+			fclose(file);
+
+			m_cpProgram = clCreateProgramWithBinary(clContext, 1, &device, &binarySize, (const unsigned char**)&binary, 0, &status);
+			b3Assert(status == CL_SUCCESS);
+			status = clBuildProgram(m_cpProgram, 1, &device, additionalMacros, 0, 0);
+			b3Assert(status == CL_SUCCESS);
+
+			if (status != CL_SUCCESS)
 			{
-				char *build_log;
+				char* build_log;
 				size_t ret_val_size;
 				clGetProgramBuildInfo(m_cpProgram, device, CL_PROGRAM_BUILD_LOG, 0, NULL, &ret_val_size);
-				build_log = (char*)malloc(sizeof(char)*(ret_val_size+1));
+				build_log = (char*)malloc(sizeof(char) * (ret_val_size + 1));
 				clGetProgramBuildInfo(m_cpProgram, device, CL_PROGRAM_BUILD_LOG, ret_val_size, build_log, NULL);
 				build_log[ret_val_size] = '\0';
 				b3Error("%s\n", build_log);
-				free (build_log);
+				free(build_log);
 				b3Assert(0);
 				m_cpProgram = 0;
 
-				b3Warning("clBuildProgram reported failure on cached binary: %s\n",binaryFileName);
-
-			} else
-			{
-				b3Printf("clBuildProgram successfully compiled cached binary: %s\n",binaryFileName);	
+				b3Warning("clBuildProgram reported failure on cached binary: %s\n", binaryFileName);
 			}
-			free (binary);
-			
-		} else
+			else
+			{
+				b3Printf("clBuildProgram successfully compiled cached binary: %s\n", binaryFileName);
+			}
+			free(binary);
+		}
+		else
 		{
-			b3Warning("Cannot open cached binary: %s\n",binaryFileName);
+			b3Warning("Cannot open cached binary: %s\n", binaryFileName);
 		}
 	}
-	
-	
-	
-	
-	
-	
-	
-	
-	
+
 	if (!m_cpProgram)
 	{
-
 		cl_int localErrNum;
 		char* compileFlags;
 		int flagsize;
 
-
-
 		const char* kernelSource = kernelSourceOrg;
 
 		if (!kernelSourceOrg || gDebugForceLoadingFromSource)
 		{
 			if (clFileNameForCaching)
 			{
-
 				FILE* file = fopen(clFileNameForCaching, "rb");
 				//in many cases the relative path is a few levels up the directory hierarchy, so try it
 				if (!file)
 				{
-					const char* prefix[]={"../","../../","../../../","../../../../"};
-					for (int i=0;!file && i<3;i++)
+					const char* prefix[] = {"../", "../../", "../../../", "../../../../"};
+					for (int i = 0; !file && i < 3; i++)
 					{
 						char relativeFileName[1024];
-						sprintf(relativeFileName,"%s%s",prefix[i],clFileNameForCaching);
+						sprintf(relativeFileName, "%s%s", prefix[i], clFileNameForCaching);
 						file = fopen(relativeFileName, "rb");
 					}
 				}
 
 				if (file)
 				{
-					char* kernelSrc=0;
-					fseek( file, 0L, SEEK_END );
-					int kernelSize = ftell( file );
-					rewind( file );
-					kernelSrc = (char*)malloc(kernelSize+1);
+					char* kernelSrc = 0;
+					fseek(file, 0L, SEEK_END);
+					int kernelSize = ftell(file);
+					rewind(file);
+					kernelSrc = (char*)malloc(kernelSize + 1);
 					int readBytes;
-					readBytes = fread((void*)kernelSrc,1,kernelSize, file);
+					readBytes = fread((void*)kernelSrc, 1, kernelSize, file);
 					kernelSrc[kernelSize] = 0;
 					fclose(file);
 					kernelSource = kernelSrc;
@@ -873,15 +836,14 @@ cl_program b3OpenCLUtils_compileCLProgramFromString(cl_context clContext, cl_dev
 		}
 
 		size_t program_length = kernelSource ? strlen(kernelSource) : 0;
-#ifdef MAC //or __APPLE__?
+#ifdef MAC  //or __APPLE__?
 		char* flags = "-cl-mad-enable -DMAC ";
 #else
 		const char* flags = "";
 #endif
 
-
 		m_cpProgram = clCreateProgramWithSource(clContext, 1, (const char**)&kernelSource, &program_length, &localErrNum);
-		if (localErrNum!= CL_SUCCESS)
+		if (localErrNum != CL_SUCCESS)
 		{
 			if (pErrNum)
 				*pErrNum = localErrNum;
@@ -890,108 +852,100 @@ cl_program b3OpenCLUtils_compileCLProgramFromString(cl_context clContext, cl_dev
 
 		// Build the program with 'mad' Optimization option
 
-
-
-        		flagsize = sizeof(char)*(strlen(additionalMacros) + strlen(flags) + 5);
-		compileFlags = (char*) malloc(flagsize);
+		flagsize = sizeof(char) * (strlen(additionalMacros) + strlen(flags) + 5);
+		compileFlags = (char*)malloc(flagsize);
 #ifdef _MSC_VER
-		sprintf_s(compileFlags,flagsize, "%s %s", flags, additionalMacros);
+		sprintf_s(compileFlags, flagsize, "%s %s", flags, additionalMacros);
 #else
 		sprintf(compileFlags, "%s %s", flags, additionalMacros);
 #endif
 		localErrNum = clBuildProgram(m_cpProgram, 1, &device, compileFlags, NULL, NULL);
-		if (localErrNum!= CL_SUCCESS)
+		if (localErrNum != CL_SUCCESS)
 		{
-			char *build_log;
+			char* build_log;
 			size_t ret_val_size;
 			clGetProgramBuildInfo(m_cpProgram, device, CL_PROGRAM_BUILD_LOG, 0, NULL, &ret_val_size);
-			build_log = (char*) malloc(sizeof(char)*(ret_val_size+1));
+			build_log = (char*)malloc(sizeof(char) * (ret_val_size + 1));
 			clGetProgramBuildInfo(m_cpProgram, device, CL_PROGRAM_BUILD_LOG, ret_val_size, build_log, NULL);
 
 			// to be carefully, terminate with \0
 			// there's no information in the reference whether the string is 0 terminated or not
 			build_log[ret_val_size] = '\0';
 
-
 			b3Error("Error in clBuildProgram, Line %u in file %s, Log: \n%s\n !!!\n\n", __LINE__, __FILE__, build_log);
-			free (build_log);
+			free(build_log);
 			if (pErrNum)
 				*pErrNum = localErrNum;
 			return 0;
 		}
 
-
-		if( !disableBinaryCaching && clFileNameForCaching )
-		{	//	write to binary
+		if (!disableBinaryCaching && clFileNameForCaching)
+		{  //	write to binary
 
 			cl_uint numAssociatedDevices;
-			status = clGetProgramInfo( m_cpProgram, CL_PROGRAM_NUM_DEVICES, sizeof(cl_uint), &numAssociatedDevices, 0 );
-			b3Assert( status == CL_SUCCESS );
-			if (numAssociatedDevices==1)
+			status = clGetProgramInfo(m_cpProgram, CL_PROGRAM_NUM_DEVICES, sizeof(cl_uint), &numAssociatedDevices, 0);
+			b3Assert(status == CL_SUCCESS);
+			if (numAssociatedDevices == 1)
 			{
-
 				size_t binarySize;
-				char* binary ;
+				char* binary;
 
-				status = clGetProgramInfo( m_cpProgram, CL_PROGRAM_BINARY_SIZES, sizeof(size_t), &binarySize, 0 );
-				b3Assert( status == CL_SUCCESS );
+				status = clGetProgramInfo(m_cpProgram, CL_PROGRAM_BINARY_SIZES, sizeof(size_t), &binarySize, 0);
+				b3Assert(status == CL_SUCCESS);
 
-				binary = (char*)malloc(sizeof(char)*binarySize);
+				binary = (char*)malloc(sizeof(char) * binarySize);
 
-				status = clGetProgramInfo( m_cpProgram, CL_PROGRAM_BINARIES, sizeof(char*), &binary, 0 );
-				b3Assert( status == CL_SUCCESS );
+				status = clGetProgramInfo(m_cpProgram, CL_PROGRAM_BINARIES, sizeof(char*), &binary, 0);
+				b3Assert(status == CL_SUCCESS);
 
 				{
-					FILE* file=0;
+					FILE* file = 0;
 #ifdef _MSC_VER
-					if (fopen_s(&file,binaryFileName, "wb")!=0)
-						file=0;
+					if (fopen_s(&file, binaryFileName, "wb") != 0)
+						file = 0;
 #else
 					file = fopen(binaryFileName, "wb");
 #endif
 					if (file)
 					{
-						fwrite( binary, sizeof(char), binarySize, file );
-						fclose( file );
-					} else
+						fwrite(binary, sizeof(char), binarySize, file);
+						fclose(file);
+					}
+					else
 					{
 						b3Warning("cannot write file %s\n", binaryFileName);
 					}
 				}
 
-				free (binary);
+				free(binary);
 			}
 		}
 
 		free(compileFlags);
-
 	}
 	return m_cpProgram;
 }
 
-
-cl_kernel b3OpenCLUtils_compileCLKernelFromString(cl_context clContext, cl_device_id device, const char* kernelSource, const char* kernelName, cl_int* pErrNum, cl_program prog, const char* additionalMacros )
+cl_kernel b3OpenCLUtils_compileCLKernelFromString(cl_context clContext, cl_device_id device, const char* kernelSource, const char* kernelName, cl_int* pErrNum, cl_program prog, const char* additionalMacros)
 {
-
 	cl_kernel kernel;
 	cl_int localErrNum;
 
 	cl_program m_cpProgram = prog;
 
-	b3Printf("compiling kernel %s ",kernelName);
+	b3Printf("compiling kernel %s ", kernelName);
 
 	if (!m_cpProgram)
 	{
-		m_cpProgram = b3OpenCLUtils_compileCLProgramFromString(clContext,device,kernelSource,pErrNum, additionalMacros,0, false);
+		m_cpProgram = b3OpenCLUtils_compileCLProgramFromString(clContext, device, kernelSource, pErrNum, additionalMacros, 0, false);
 	}
 
-
 	// Create the kernel
 	kernel = clCreateKernel(m_cpProgram, kernelName, &localErrNum);
 	if (localErrNum != CL_SUCCESS)
 	{
 		b3Error("Error in clCreateKernel, Line %u in file %s, cannot find kernel function %s !!!\n\n", __LINE__, __FILE__, kernelName);
-        assert(0);
+		assert(0);
 		if (pErrNum)
 			*pErrNum = localErrNum;
 		return 0;
@@ -1003,9 +957,7 @@ cl_kernel b3OpenCLUtils_compileCLKernelFromString(cl_context clContext, cl_devic
 	}
 	b3Printf("ready. \n");
 
-
 	if (pErrNum)
-			*pErrNum = CL_SUCCESS;
+		*pErrNum = CL_SUCCESS;
 	return kernel;
-
 }
diff --git a/thirdparty/bullet/Bullet3OpenCL/Initialize/b3OpenCLUtils.h b/thirdparty/bullet/Bullet3OpenCL/Initialize/b3OpenCLUtils.h
index db6466e76b2..6c82eed2a6b 100644
--- a/thirdparty/bullet/Bullet3OpenCL/Initialize/b3OpenCLUtils.h
+++ b/thirdparty/bullet/Bullet3OpenCL/Initialize/b3OpenCLUtils.h
@@ -22,42 +22,41 @@ subject to the following restrictions:
 #include "b3OpenCLInclude.h"
 
 #ifdef __cplusplus
-extern "C" {
+extern "C"
+{
 #endif
 
+	///C API for OpenCL utilities: convenience functions, see below for C++ API
 
-///C API for OpenCL utilities: convenience functions, see below for C++ API
+	/// CL Context optionally takes a GL context. This is a generic type because we don't really want this code
+	/// to have to understand GL types. It is a HGLRC in _WIN32 or a GLXContext otherwise.
+	cl_context b3OpenCLUtils_createContextFromType(cl_device_type deviceType, cl_int* pErrNum, void* pGLCtx, void* pGLDC, int preferredDeviceIndex, int preferredPlatformIndex, cl_platform_id* platformId);
 
-/// CL Context optionally takes a GL context. This is a generic type because we don't really want this code
-/// to have to understand GL types. It is a HGLRC in _WIN32 or a GLXContext otherwise.
-cl_context 	b3OpenCLUtils_createContextFromType(cl_device_type deviceType, cl_int* pErrNum, void* pGLCtx , void* pGLDC , int preferredDeviceIndex , int preferredPlatformIndex, cl_platform_id* platformId);
-	
-int b3OpenCLUtils_getNumDevices(cl_context cxMainContext);
+	int b3OpenCLUtils_getNumDevices(cl_context cxMainContext);
 
-cl_device_id b3OpenCLUtils_getDevice(cl_context cxMainContext, int nr);
+	cl_device_id b3OpenCLUtils_getDevice(cl_context cxMainContext, int nr);
 
-void b3OpenCLUtils_printDeviceInfo(cl_device_id device);
+	void b3OpenCLUtils_printDeviceInfo(cl_device_id device);
 
-cl_kernel b3OpenCLUtils_compileCLKernelFromString( cl_context clContext,cl_device_id device, const char* kernelSource, const char* kernelName, cl_int* pErrNum, cl_program prog,const char* additionalMacros);
+	cl_kernel b3OpenCLUtils_compileCLKernelFromString(cl_context clContext, cl_device_id device, const char* kernelSource, const char* kernelName, cl_int* pErrNum, cl_program prog, const char* additionalMacros);
 
-//optional
-cl_program b3OpenCLUtils_compileCLProgramFromString( cl_context clContext,cl_device_id device, const char* kernelSource, cl_int* pErrNum,const char* additionalMacros  , const char* srcFileNameForCaching, bool disableBinaryCaching);
+	//optional
+	cl_program b3OpenCLUtils_compileCLProgramFromString(cl_context clContext, cl_device_id device, const char* kernelSource, cl_int* pErrNum, const char* additionalMacros, const char* srcFileNameForCaching, bool disableBinaryCaching);
 
-//the following optional APIs provide access using specific platform information
-int b3OpenCLUtils_getNumPlatforms(cl_int* pErrNum);
+	//the following optional APIs provide access using specific platform information
+	int b3OpenCLUtils_getNumPlatforms(cl_int* pErrNum);
 
-///get the nr'th platform, where nr is in the range [0..getNumPlatforms)
-cl_platform_id b3OpenCLUtils_getPlatform(int nr, cl_int* pErrNum);
+	///get the nr'th platform, where nr is in the range [0..getNumPlatforms)
+	cl_platform_id b3OpenCLUtils_getPlatform(int nr, cl_int* pErrNum);
 
+	void b3OpenCLUtils_printPlatformInfo(cl_platform_id platform);
 
-void b3OpenCLUtils_printPlatformInfo(cl_platform_id platform);
+	const char* b3OpenCLUtils_getSdkVendorName();
 
-const char* b3OpenCLUtils_getSdkVendorName();
-	
-///set the path (directory/folder) where the compiled OpenCL kernel are stored
-void b3OpenCLUtils_setCachePath(const char* path);
-	
-cl_context 	b3OpenCLUtils_createContextFromPlatform(cl_platform_id platform, cl_device_type deviceType, cl_int* pErrNum, void* pGLCtx , void* pGLDC ,int preferredDeviceIndex , int preferredPlatformIndex);
+	///set the path (directory/folder) where the compiled OpenCL kernel are stored
+	void b3OpenCLUtils_setCachePath(const char* path);
+
+	cl_context b3OpenCLUtils_createContextFromPlatform(cl_platform_id platform, cl_device_type deviceType, cl_int* pErrNum, void* pGLCtx, void* pGLDC, int preferredDeviceIndex, int preferredPlatformIndex);
 
 #ifdef __cplusplus
 }
@@ -71,37 +70,35 @@ typedef struct
 	char m_driverVersion[B3_MAX_STRING_LENGTH];
 	char m_deviceExtensions[B3_MAX_STRING_LENGTH];
 
-	cl_device_type		m_deviceType;
-	cl_uint 				m_computeUnits;
-	size_t 					m_workitemDims;
-	size_t 					m_workItemSize[3];
-	size_t 					m_image2dMaxWidth;
-	size_t 					m_image2dMaxHeight;
-	size_t 					m_image3dMaxWidth;
-	size_t 					m_image3dMaxHeight;
-	size_t 					m_image3dMaxDepth;
-	size_t 					m_workgroupSize;
-	cl_uint 				m_clockFrequency;
-	cl_ulong				m_constantBufferSize;
-	cl_ulong				m_localMemSize;
-	cl_ulong				m_globalMemSize;
-    cl_bool					m_errorCorrectionSupport;
+	cl_device_type m_deviceType;
+	cl_uint m_computeUnits;
+	size_t m_workitemDims;
+	size_t m_workItemSize[3];
+	size_t m_image2dMaxWidth;
+	size_t m_image2dMaxHeight;
+	size_t m_image3dMaxWidth;
+	size_t m_image3dMaxHeight;
+	size_t m_image3dMaxDepth;
+	size_t m_workgroupSize;
+	cl_uint m_clockFrequency;
+	cl_ulong m_constantBufferSize;
+	cl_ulong m_localMemSize;
+	cl_ulong m_globalMemSize;
+	cl_bool m_errorCorrectionSupport;
 	cl_device_local_mem_type m_localMemType;
-	cl_uint					m_maxReadImageArgs;
-	cl_uint					m_maxWriteImageArgs;
+	cl_uint m_maxReadImageArgs;
+	cl_uint m_maxWriteImageArgs;
 
-
-
-	cl_uint 				m_addressBits;
-	cl_ulong				m_maxMemAllocSize;
+	cl_uint m_addressBits;
+	cl_ulong m_maxMemAllocSize;
 	cl_command_queue_properties m_queueProperties;
-	cl_bool					m_imageSupport;
-	cl_uint					m_vecWidthChar;
-	cl_uint					m_vecWidthShort;
-	cl_uint					m_vecWidthInt;
-	cl_uint					m_vecWidthLong;
-	cl_uint					m_vecWidthFloat;
-	cl_uint					m_vecWidthDouble;
+	cl_bool m_imageSupport;
+	cl_uint m_vecWidthChar;
+	cl_uint m_vecWidthShort;
+	cl_uint m_vecWidthInt;
+	cl_uint m_vecWidthLong;
+	cl_uint m_vecWidthFloat;
+	cl_uint m_vecWidthDouble;
 
 } b3OpenCLDeviceInfo;
 
@@ -110,33 +107,32 @@ struct b3OpenCLPlatformInfo
 	char m_platformVendor[B3_MAX_STRING_LENGTH];
 	char m_platformName[B3_MAX_STRING_LENGTH];
 	char m_platformVersion[B3_MAX_STRING_LENGTH];
-	
+
 	b3OpenCLPlatformInfo()
 	{
-		m_platformVendor[0]=0;
-		m_platformName[0]=0;
-		m_platformVersion[0]=0;
+		m_platformVendor[0] = 0;
+		m_platformName[0] = 0;
+		m_platformVersion[0] = 0;
 	}
 };
 
-
 ///C++ API for OpenCL utilities: convenience functions
 struct b3OpenCLUtils
 {
 	/// CL Context optionally takes a GL context. This is a generic type because we don't really want this code
 	/// to have to understand GL types. It is a HGLRC in _WIN32 or a GLXContext otherwise.
-	static inline cl_context 	createContextFromType(cl_device_type deviceType, cl_int* pErrNum, void* pGLCtx = 0, void* pGLDC = 0, int preferredDeviceIndex = -1, int preferredPlatformIndex= - 1, cl_platform_id* platformId=0)
+	static inline cl_context createContextFromType(cl_device_type deviceType, cl_int* pErrNum, void* pGLCtx = 0, void* pGLDC = 0, int preferredDeviceIndex = -1, int preferredPlatformIndex = -1, cl_platform_id* platformId = 0)
 	{
-		return b3OpenCLUtils_createContextFromType(deviceType, pErrNum, pGLCtx , pGLDC , preferredDeviceIndex, preferredPlatformIndex, platformId);
+		return b3OpenCLUtils_createContextFromType(deviceType, pErrNum, pGLCtx, pGLDC, preferredDeviceIndex, preferredPlatformIndex, platformId);
 	}
-	
+
 	static inline int getNumDevices(cl_context cxMainContext)
 	{
 		return b3OpenCLUtils_getNumDevices(cxMainContext);
 	}
 	static inline cl_device_id getDevice(cl_context cxMainContext, int nr)
 	{
-		return b3OpenCLUtils_getDevice(cxMainContext,nr);
+		return b3OpenCLUtils_getDevice(cxMainContext, nr);
 	}
 
 	static void getDeviceInfo(cl_device_id device, b3OpenCLDeviceInfo* info);
@@ -146,28 +142,28 @@ struct b3OpenCLUtils
 		b3OpenCLUtils_printDeviceInfo(device);
 	}
 
-	static inline cl_kernel compileCLKernelFromString( cl_context clContext,cl_device_id device, const char* kernelSource, const char* kernelName, cl_int* pErrNum=0, cl_program prog=0,const char* additionalMacros = "" )
+	static inline cl_kernel compileCLKernelFromString(cl_context clContext, cl_device_id device, const char* kernelSource, const char* kernelName, cl_int* pErrNum = 0, cl_program prog = 0, const char* additionalMacros = "")
 	{
-		return b3OpenCLUtils_compileCLKernelFromString(clContext,device, kernelSource,  kernelName, pErrNum, prog,additionalMacros);
+		return b3OpenCLUtils_compileCLKernelFromString(clContext, device, kernelSource, kernelName, pErrNum, prog, additionalMacros);
 	}
 
 	//optional
-	static inline cl_program compileCLProgramFromString( cl_context clContext,cl_device_id device, const char* kernelSource, cl_int* pErrNum=0,const char* additionalMacros = "" , const char* srcFileNameForCaching=0, bool disableBinaryCaching=false)
+	static inline cl_program compileCLProgramFromString(cl_context clContext, cl_device_id device, const char* kernelSource, cl_int* pErrNum = 0, const char* additionalMacros = "", const char* srcFileNameForCaching = 0, bool disableBinaryCaching = false)
 	{
-		return b3OpenCLUtils_compileCLProgramFromString(clContext,device, kernelSource, pErrNum,additionalMacros, srcFileNameForCaching, disableBinaryCaching);
+		return b3OpenCLUtils_compileCLProgramFromString(clContext, device, kernelSource, pErrNum, additionalMacros, srcFileNameForCaching, disableBinaryCaching);
 	}
 
 	//the following optional APIs provide access using specific platform information
-	static inline int getNumPlatforms(cl_int* pErrNum=0)
+	static inline int getNumPlatforms(cl_int* pErrNum = 0)
 	{
 		return b3OpenCLUtils_getNumPlatforms(pErrNum);
 	}
 	///get the nr'th platform, where nr is in the range [0..getNumPlatforms)
-	static inline cl_platform_id getPlatform(int nr, cl_int* pErrNum=0)
+	static inline cl_platform_id getPlatform(int nr, cl_int* pErrNum = 0)
 	{
-		return b3OpenCLUtils_getPlatform(nr,pErrNum);
+		return b3OpenCLUtils_getPlatform(nr, pErrNum);
 	}
-	
+
 	static void getPlatformInfo(cl_platform_id platform, b3OpenCLPlatformInfo* platformInfo);
 
 	static inline void printPlatformInfo(cl_platform_id platform)
@@ -179,9 +175,9 @@ struct b3OpenCLUtils
 	{
 		return b3OpenCLUtils_getSdkVendorName();
 	}
-	static inline cl_context 	createContextFromPlatform(cl_platform_id platform, cl_device_type deviceType, cl_int* pErrNum, void* pGLCtx = 0, void* pGLDC = 0,int preferredDeviceIndex = -1, int preferredPlatformIndex= -1)
+	static inline cl_context createContextFromPlatform(cl_platform_id platform, cl_device_type deviceType, cl_int* pErrNum, void* pGLCtx = 0, void* pGLDC = 0, int preferredDeviceIndex = -1, int preferredPlatformIndex = -1)
 	{
-		return b3OpenCLUtils_createContextFromPlatform(platform, deviceType, pErrNum, pGLCtx,pGLDC,preferredDeviceIndex, preferredPlatformIndex);
+		return b3OpenCLUtils_createContextFromPlatform(platform, deviceType, pErrNum, pGLCtx, pGLDC, preferredDeviceIndex, preferredPlatformIndex);
 	}
 	static void setCachePath(const char* path)
 	{
@@ -189,6 +185,6 @@ struct b3OpenCLUtils
 	}
 };
 
-#endif //__cplusplus
+#endif  //__cplusplus
 
-#endif // B3_OPENCL_UTILS_H
+#endif  // B3_OPENCL_UTILS_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3BvhInfo.h b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3BvhInfo.h
index 872f0395064..27835bb7474 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3BvhInfo.h
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3BvhInfo.h
@@ -5,14 +5,13 @@
 
 struct b3BvhInfo
 {
-	b3Vector3	m_aabbMin;
-	b3Vector3	m_aabbMax;
-	b3Vector3	m_quantization;
-	int			m_numNodes;
-	int			m_numSubTrees;
-	int			m_nodeOffset;
-	int			m_subTreeOffset;
-
+	b3Vector3 m_aabbMin;
+	b3Vector3 m_aabbMax;
+	b3Vector3 m_quantization;
+	int m_numNodes;
+	int m_numSubTrees;
+	int m_nodeOffset;
+	int m_subTreeOffset;
 };
 
-#endif //B3_BVH_INFO_H
\ No newline at end of file
+#endif  //B3_BVH_INFO_H
\ No newline at end of file
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3ContactCache.cpp b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3ContactCache.cpp
index cb30ee939b8..4db717f8c36 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3ContactCache.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3ContactCache.cpp
@@ -15,7 +15,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "b3ContactCache.h"
 #include "Bullet3Common/b3Transform.h"
 
@@ -69,7 +68,7 @@ int b3ContactCache::sortCachedPoints(const b3Vector3& pt)
 				maxPenetration = m_pointCache[i].getDistance();
 			}
 		}
-#endif //KEEP_DEEPEST_POINT
+#endif  //KEEP_DEEPEST_POINT
 		
 		b3Scalar res0(b3Scalar(0.)),res1(b3Scalar(0.)),res2(b3Scalar(0.)),res3(b3Scalar(0.));
 
@@ -251,8 +250,4 @@ void b3ContactCache::refreshContactPoints(const b3Transform& trA,const b3Transfo
 
 }
 
-
-
-
-
 #endif
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3ContactCache.h b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3ContactCache.h
index d6c9b0a07e4..a15fd0b2a9e 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3ContactCache.h
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3ContactCache.h
@@ -17,17 +17,13 @@ subject to the following restrictions:
 #ifndef B3_CONTACT_CACHE_H
 #define B3_CONTACT_CACHE_H
 
-
 #include "Bullet3Common/b3Vector3.h"
 #include "Bullet3Common/b3Transform.h"
 #include "Bullet3Common/b3AlignedAllocator.h"
 
-
 ///maximum contact breaking and merging threshold
 extern b3Scalar gContactBreakingThreshold;
 
-
-
 #define MANIFOLD_CACHE_SIZE 4
 
 ///b3ContactCache is a contact point cache, it stays persistent as long as objects are overlapping in the broadphase.
@@ -37,24 +33,16 @@ extern b3Scalar gContactBreakingThreshold;
 ///reduces the cache to 4 points, when more then 4 points are added, using following rules:
 ///the contact point with deepest penetration is always kept, and it tries to maximuze the area covered by the points
 ///note that some pairs of objects might have more then one contact manifold.
-B3_ATTRIBUTE_ALIGNED16( class) b3ContactCache
+B3_ATTRIBUTE_ALIGNED16(class)
+b3ContactCache
 {
-
-	
-
-	
 	/// sort cached points so most isolated points come first
-	int	sortCachedPoints(const b3Vector3& pt);
-
-	
+	int sortCachedPoints(const b3Vector3& pt);
 
 public:
-
 	B3_DECLARE_ALIGNED_ALLOCATOR();
 
-	
-	
-	int addManifoldPoint( const b3Vector3& newPoint);
+	int addManifoldPoint(const b3Vector3& newPoint);
 
 	/*void replaceContactPoint(const b3Vector3& newPoint,int insertIndex)
 	{
@@ -63,18 +51,12 @@ public:
 	}
 	*/
 
-
-	
 	static bool validContactDistance(const b3Vector3& pt);
-	
+
 	/// calculated new worldspace coordinates and depth, and reject points that exceed the collision margin
-	static void	refreshContactPoints(  const b3Transform& trA,const b3Transform& trB, struct b3Contact4Data& newContactCache);
-
-	static void removeContactPoint(struct b3Contact4Data& newContactCache,int i);
-	
+	static void refreshContactPoints(const b3Transform& trA, const b3Transform& trB, struct b3Contact4Data& newContactCache);
 
+	static void removeContactPoint(struct b3Contact4Data & newContactCache, int i);
 };
 
-
-
-#endif //B3_CONTACT_CACHE_H
+#endif  //B3_CONTACT_CACHE_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3ConvexHullContact.cpp b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3ConvexHullContact.cpp
index fb435aa7fdd..54a104c5c86 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3ConvexHullContact.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3ConvexHullContact.cpp
@@ -16,19 +16,18 @@ subject to the following restrictions:
 bool findSeparatingAxisOnGpu = true;
 bool splitSearchSepAxisConcave = false;
 bool splitSearchSepAxisConvex = true;
-bool useMprGpu = true;//use mpr for edge-edge  (+contact point) or sat. Needs testing on main OpenCL platforms, before enabling...
+bool useMprGpu = true;  //use mpr for edge-edge  (+contact point) or sat. Needs testing on main OpenCL platforms, before enabling...
 bool bvhTraversalKernelGPU = true;
 bool findConcaveSeparatingAxisKernelGPU = true;
-bool clipConcaveFacesAndFindContactsCPU = false;//false;//true;
-bool clipConvexFacesAndFindContactsCPU = false;//false;//true;
-bool reduceConcaveContactsOnGPU = true;//false;
-bool reduceConvexContactsOnGPU = true;//false;
+bool clipConcaveFacesAndFindContactsCPU = false;  //false;//true;
+bool clipConvexFacesAndFindContactsCPU = false;   //false;//true;
+bool reduceConcaveContactsOnGPU = true;           //false;
+bool reduceConvexContactsOnGPU = true;            //false;
 bool findConvexClippingFacesGPU = true;
-bool useGjk = false;///option for CPU/host testing, when findSeparatingAxisOnGpu = false
-bool useGjkContacts = false;//////option for CPU/host testing when findSeparatingAxisOnGpu = false
+bool useGjk = false;          ///option for CPU/host testing, when findSeparatingAxisOnGpu = false
+bool useGjkContacts = false;  //////option for CPU/host testing when findSeparatingAxisOnGpu = false
 
-
-static int myframecount=0;///for testing
+static int myframecount = 0;  ///for testing
 
 ///This file was written by Erwin Coumans
 ///Separating axis rest based on work from Pierre Terdiman, see
@@ -42,10 +41,10 @@ static int myframecount=0;///for testing
 //#define PERSISTENT_CONTACTS_HOST
 #endif
 
-int b3g_actualSATPairTests=0;
+int b3g_actualSATPairTests = 0;
 
 #include "b3ConvexHullContact.h"
-#include <string.h>//memcpy
+#include <string.h>  //memcpy
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3ConvexPolyhedronData.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3MprPenetration.h"
 
@@ -54,8 +53,7 @@ int b3g_actualSATPairTests=0;
 
 typedef b3AlignedObjectArray<b3Vector3> b3VertexArray;
 
-
-#include <float.h> //for FLT_MAX
+#include <float.h>  //for FLT_MAX
 #include "Bullet3OpenCL/Initialize/b3OpenCLUtils.h"
 #include "Bullet3OpenCL/ParallelPrimitives/b3LauncherCL.h"
 //#include "AdlQuaternion.h"
@@ -69,7 +67,6 @@ typedef b3AlignedObjectArray<b3Vector3> b3VertexArray;
 #include "kernels/bvhTraversal.h"
 #include "kernels/primitiveContacts.h"
 
-
 #include "Bullet3Geometry/b3AabbUtil.h"
 
 #define BT_NARROWPHASE_SAT_PATH "src/Bullet3OpenCL/NarrowphaseCollision/kernels/sat.cl"
@@ -77,12 +74,10 @@ typedef b3AlignedObjectArray<b3Vector3> b3VertexArray;
 
 #define BT_NARROWPHASE_MPR_PATH "src/Bullet3OpenCL/NarrowphaseCollision/kernels/mpr.cl"
 
-
 #define BT_NARROWPHASE_CLIPHULL_PATH "src/Bullet3OpenCL/NarrowphaseCollision/kernels/satClipHullContacts.cl"
 #define BT_NARROWPHASE_BVH_TRAVERSAL_PATH "src/Bullet3OpenCL/NarrowphaseCollision/kernels/bvhTraversal.cl"
 #define BT_NARROWPHASE_PRIMITIVE_CONTACT_PATH "src/Bullet3OpenCL/NarrowphaseCollision/kernels/primitiveContacts.cl"
 
-
 #ifndef __global
 #define __global
 #endif
@@ -91,204 +86,184 @@ typedef b3AlignedObjectArray<b3Vector3> b3VertexArray;
 #define __kernel
 #endif
 
-
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3BvhTraversal.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3FindConcaveSatAxis.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3ClipFaces.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3NewContactReduction.h"
 
-
-
 #define dot3F4 b3Dot
 
-GpuSatCollision::GpuSatCollision(cl_context ctx,cl_device_id device, cl_command_queue  q )
-:m_context(ctx),
-m_device(device),
-m_queue(q),
+GpuSatCollision::GpuSatCollision(cl_context ctx, cl_device_id device, cl_command_queue q)
+	: m_context(ctx),
+	  m_device(device),
+	  m_queue(q),
 
-m_findSeparatingAxisKernel(0),
-m_findSeparatingAxisVertexFaceKernel(0),
-m_findSeparatingAxisEdgeEdgeKernel(0),
-m_unitSphereDirections(m_context,m_queue),
+	  m_findSeparatingAxisKernel(0),
+	  m_findSeparatingAxisVertexFaceKernel(0),
+	  m_findSeparatingAxisEdgeEdgeKernel(0),
+	  m_unitSphereDirections(m_context, m_queue),
 
-m_totalContactsOut(m_context, m_queue),
-m_sepNormals(m_context, m_queue),
-m_dmins(m_context,m_queue),
+	  m_totalContactsOut(m_context, m_queue),
+	  m_sepNormals(m_context, m_queue),
+	  m_dmins(m_context, m_queue),
 
-m_hasSeparatingNormals(m_context, m_queue),
-m_concaveSepNormals(m_context, m_queue),
-m_concaveHasSeparatingNormals(m_context,m_queue),
-m_numConcavePairsOut(m_context, m_queue),
+	  m_hasSeparatingNormals(m_context, m_queue),
+	  m_concaveSepNormals(m_context, m_queue),
+	  m_concaveHasSeparatingNormals(m_context, m_queue),
+	  m_numConcavePairsOut(m_context, m_queue),
 
+	  m_gpuCompoundPairs(m_context, m_queue),
 
-m_gpuCompoundPairs(m_context, m_queue),
+	  m_gpuCompoundSepNormals(m_context, m_queue),
+	  m_gpuHasCompoundSepNormals(m_context, m_queue),
 
-
-m_gpuCompoundSepNormals(m_context, m_queue),
-m_gpuHasCompoundSepNormals(m_context, m_queue),
-
-m_numCompoundPairsOut(m_context, m_queue)
+	  m_numCompoundPairsOut(m_context, m_queue)
 {
 	m_totalContactsOut.push_back(0);
-	
-	cl_int errNum=0;
+
+	cl_int errNum = 0;
 
 	if (1)
 	{
 		const char* mprSrc = mprKernelsCL;
-		
+
 		const char* srcConcave = satConcaveKernelsCL;
-		char flags[1024]={0};
-//#ifdef CL_PLATFORM_INTEL
-//		sprintf(flags,"-g -s \"%s\"","C:/develop/bullet3_experiments2/opencl/gpu_narrowphase/kernels/sat.cl");
-//#endif
-		m_mprPenetrationKernel  = 0;
+		char flags[1024] = {0};
+		//#ifdef CL_PLATFORM_INTEL
+		//		sprintf(flags,"-g -s \"%s\"","C:/develop/bullet3_experiments2/opencl/gpu_narrowphase/kernels/sat.cl");
+		//#endif
+		m_mprPenetrationKernel = 0;
 		m_findSeparatingAxisUnitSphereKernel = 0;
 
 		if (useMprGpu)
 		{
-			cl_program mprProg = b3OpenCLUtils::compileCLProgramFromString(m_context,m_device,mprSrc,&errNum,flags,BT_NARROWPHASE_MPR_PATH);
-			b3Assert(errNum==CL_SUCCESS);
-		
-			m_mprPenetrationKernel  = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,mprSrc, "mprPenetrationKernel",&errNum,mprProg );
+			cl_program mprProg = b3OpenCLUtils::compileCLProgramFromString(m_context, m_device, mprSrc, &errNum, flags, BT_NARROWPHASE_MPR_PATH);
+			b3Assert(errNum == CL_SUCCESS);
+
+			m_mprPenetrationKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, mprSrc, "mprPenetrationKernel", &errNum, mprProg);
 			b3Assert(m_mprPenetrationKernel);
-			b3Assert(errNum==CL_SUCCESS);
+			b3Assert(errNum == CL_SUCCESS);
 
-			m_findSeparatingAxisUnitSphereKernel =  b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,mprSrc, "findSeparatingAxisUnitSphereKernel",&errNum,mprProg );
+			m_findSeparatingAxisUnitSphereKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, mprSrc, "findSeparatingAxisUnitSphereKernel", &errNum, mprProg);
 			b3Assert(m_findSeparatingAxisUnitSphereKernel);
-            b3Assert(errNum==CL_SUCCESS);
+			b3Assert(errNum == CL_SUCCESS);
 
-
-			int numDirections = sizeof(unitSphere162)/sizeof(b3Vector3);
+			int numDirections = sizeof(unitSphere162) / sizeof(b3Vector3);
 			m_unitSphereDirections.resize(numDirections);
-			m_unitSphereDirections.copyFromHostPointer(unitSphere162,numDirections,0,true);
-
-
+			m_unitSphereDirections.copyFromHostPointer(unitSphere162, numDirections, 0, true);
 		}
 
+		cl_program satProg = b3OpenCLUtils::compileCLProgramFromString(m_context, m_device, satKernelsCL, &errNum, flags, BT_NARROWPHASE_SAT_PATH);
+		b3Assert(errNum == CL_SUCCESS);
 
-		cl_program satProg = b3OpenCLUtils::compileCLProgramFromString(m_context,m_device,satKernelsCL,&errNum,flags,BT_NARROWPHASE_SAT_PATH);
-		b3Assert(errNum==CL_SUCCESS);
+		cl_program satConcaveProg = b3OpenCLUtils::compileCLProgramFromString(m_context, m_device, srcConcave, &errNum, flags, BT_NARROWPHASE_SAT_CONCAVE_PATH);
+		b3Assert(errNum == CL_SUCCESS);
 
-		cl_program satConcaveProg = b3OpenCLUtils::compileCLProgramFromString(m_context,m_device,srcConcave,&errNum,flags,BT_NARROWPHASE_SAT_CONCAVE_PATH);
-		b3Assert(errNum==CL_SUCCESS);
-
-		m_findSeparatingAxisKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,satKernelsCL, "findSeparatingAxisKernel",&errNum,satProg );
+		m_findSeparatingAxisKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, satKernelsCL, "findSeparatingAxisKernel", &errNum, satProg);
 		b3Assert(m_findSeparatingAxisKernel);
-		b3Assert(errNum==CL_SUCCESS);
+		b3Assert(errNum == CL_SUCCESS);
 
-
-		m_findSeparatingAxisVertexFaceKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,satKernelsCL, "findSeparatingAxisVertexFaceKernel",&errNum,satProg );
+		m_findSeparatingAxisVertexFaceKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, satKernelsCL, "findSeparatingAxisVertexFaceKernel", &errNum, satProg);
 		b3Assert(m_findSeparatingAxisVertexFaceKernel);
 
-		m_findSeparatingAxisEdgeEdgeKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,satKernelsCL, "findSeparatingAxisEdgeEdgeKernel",&errNum,satProg );
+		m_findSeparatingAxisEdgeEdgeKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, satKernelsCL, "findSeparatingAxisEdgeEdgeKernel", &errNum, satProg);
 		b3Assert(m_findSeparatingAxisVertexFaceKernel);
 
-
-		m_findConcaveSeparatingAxisKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,satKernelsCL, "findConcaveSeparatingAxisKernel",&errNum,satProg );
+		m_findConcaveSeparatingAxisKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, satKernelsCL, "findConcaveSeparatingAxisKernel", &errNum, satProg);
 		b3Assert(m_findConcaveSeparatingAxisKernel);
-		b3Assert(errNum==CL_SUCCESS);
-        
-        m_findConcaveSeparatingAxisVertexFaceKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,srcConcave, "findConcaveSeparatingAxisVertexFaceKernel",&errNum,satConcaveProg );
+		b3Assert(errNum == CL_SUCCESS);
+
+		m_findConcaveSeparatingAxisVertexFaceKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, srcConcave, "findConcaveSeparatingAxisVertexFaceKernel", &errNum, satConcaveProg);
 		b3Assert(m_findConcaveSeparatingAxisVertexFaceKernel);
-		b3Assert(errNum==CL_SUCCESS);
-        
-        m_findConcaveSeparatingAxisEdgeEdgeKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,srcConcave, "findConcaveSeparatingAxisEdgeEdgeKernel",&errNum,satConcaveProg );
+		b3Assert(errNum == CL_SUCCESS);
+
+		m_findConcaveSeparatingAxisEdgeEdgeKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, srcConcave, "findConcaveSeparatingAxisEdgeEdgeKernel", &errNum, satConcaveProg);
 		b3Assert(m_findConcaveSeparatingAxisEdgeEdgeKernel);
-		b3Assert(errNum==CL_SUCCESS);
-        
-     
-        
-		
-		m_findCompoundPairsKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,satKernelsCL, "findCompoundPairsKernel",&errNum,satProg );
+		b3Assert(errNum == CL_SUCCESS);
+
+		m_findCompoundPairsKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, satKernelsCL, "findCompoundPairsKernel", &errNum, satProg);
 		b3Assert(m_findCompoundPairsKernel);
-		b3Assert(errNum==CL_SUCCESS);
-		m_processCompoundPairsKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,satKernelsCL, "processCompoundPairsKernel",&errNum,satProg );
+		b3Assert(errNum == CL_SUCCESS);
+		m_processCompoundPairsKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, satKernelsCL, "processCompoundPairsKernel", &errNum, satProg);
 		b3Assert(m_processCompoundPairsKernel);
-		b3Assert(errNum==CL_SUCCESS);
+		b3Assert(errNum == CL_SUCCESS);
 	}
 
 	if (1)
 	{
 		const char* srcClip = satClipKernelsCL;
 
-		char flags[1024]={0};
-//#ifdef CL_PLATFORM_INTEL
-//		sprintf(flags,"-g -s \"%s\"","C:/develop/bullet3_experiments2/opencl/gpu_narrowphase/kernels/satClipHullContacts.cl");
-//#endif
+		char flags[1024] = {0};
+		//#ifdef CL_PLATFORM_INTEL
+		//		sprintf(flags,"-g -s \"%s\"","C:/develop/bullet3_experiments2/opencl/gpu_narrowphase/kernels/satClipHullContacts.cl");
+		//#endif
 
-		cl_program satClipContactsProg = b3OpenCLUtils::compileCLProgramFromString(m_context,m_device,srcClip,&errNum,flags,BT_NARROWPHASE_CLIPHULL_PATH);
-		b3Assert(errNum==CL_SUCCESS);
+		cl_program satClipContactsProg = b3OpenCLUtils::compileCLProgramFromString(m_context, m_device, srcClip, &errNum, flags, BT_NARROWPHASE_CLIPHULL_PATH);
+		b3Assert(errNum == CL_SUCCESS);
 
-		m_clipHullHullKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,srcClip, "clipHullHullKernel",&errNum,satClipContactsProg);
-		b3Assert(errNum==CL_SUCCESS);
+		m_clipHullHullKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, srcClip, "clipHullHullKernel", &errNum, satClipContactsProg);
+		b3Assert(errNum == CL_SUCCESS);
 
-		m_clipCompoundsHullHullKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,srcClip, "clipCompoundsHullHullKernel",&errNum,satClipContactsProg);
-		b3Assert(errNum==CL_SUCCESS);
-		
+		m_clipCompoundsHullHullKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, srcClip, "clipCompoundsHullHullKernel", &errNum, satClipContactsProg);
+		b3Assert(errNum == CL_SUCCESS);
 
-        m_findClippingFacesKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,srcClip, "findClippingFacesKernel",&errNum,satClipContactsProg);
-		b3Assert(errNum==CL_SUCCESS);
+		m_findClippingFacesKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, srcClip, "findClippingFacesKernel", &errNum, satClipContactsProg);
+		b3Assert(errNum == CL_SUCCESS);
 
-        m_clipFacesAndFindContacts = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,srcClip, "clipFacesAndFindContactsKernel",&errNum,satClipContactsProg);
-		b3Assert(errNum==CL_SUCCESS);        
+		m_clipFacesAndFindContacts = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, srcClip, "clipFacesAndFindContactsKernel", &errNum, satClipContactsProg);
+		b3Assert(errNum == CL_SUCCESS);
 
-		m_clipHullHullConcaveConvexKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,srcClip, "clipHullHullConcaveConvexKernel",&errNum,satClipContactsProg);
-		b3Assert(errNum==CL_SUCCESS);
+		m_clipHullHullConcaveConvexKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, srcClip, "clipHullHullConcaveConvexKernel", &errNum, satClipContactsProg);
+		b3Assert(errNum == CL_SUCCESS);
 
-//		m_extractManifoldAndAddContactKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,srcClip, "extractManifoldAndAddContactKernel",&errNum,satClipContactsProg);
-	//	b3Assert(errNum==CL_SUCCESS);
+		//		m_extractManifoldAndAddContactKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,srcClip, "extractManifoldAndAddContactKernel",&errNum,satClipContactsProg);
+		//	b3Assert(errNum==CL_SUCCESS);
 
-        m_newContactReductionKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,srcClip,
-                            "newContactReductionKernel",&errNum,satClipContactsProg);
-		b3Assert(errNum==CL_SUCCESS);
+		m_newContactReductionKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, srcClip,
+																			   "newContactReductionKernel", &errNum, satClipContactsProg);
+		b3Assert(errNum == CL_SUCCESS);
 	}
-   else
+	else
 	{
-		m_clipHullHullKernel=0;
+		m_clipHullHullKernel = 0;
 		m_clipCompoundsHullHullKernel = 0;
-        m_findClippingFacesKernel = 0;
-        m_newContactReductionKernel=0;
-        m_clipFacesAndFindContacts = 0;
+		m_findClippingFacesKernel = 0;
+		m_newContactReductionKernel = 0;
+		m_clipFacesAndFindContacts = 0;
 		m_clipHullHullConcaveConvexKernel = 0;
-//		m_extractManifoldAndAddContactKernel = 0;
+		//		m_extractManifoldAndAddContactKernel = 0;
 	}
 
-	 if (1)
+	if (1)
 	{
 		const char* srcBvh = bvhTraversalKernelCL;
-		cl_program bvhTraversalProg = b3OpenCLUtils::compileCLProgramFromString(m_context,m_device,srcBvh,&errNum,"",BT_NARROWPHASE_BVH_TRAVERSAL_PATH);
-		b3Assert(errNum==CL_SUCCESS);
-
-		m_bvhTraversalKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,srcBvh, "bvhTraversalKernel",&errNum,bvhTraversalProg,"");
-		b3Assert(errNum==CL_SUCCESS);
+		cl_program bvhTraversalProg = b3OpenCLUtils::compileCLProgramFromString(m_context, m_device, srcBvh, &errNum, "", BT_NARROWPHASE_BVH_TRAVERSAL_PATH);
+		b3Assert(errNum == CL_SUCCESS);
 
+		m_bvhTraversalKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, srcBvh, "bvhTraversalKernel", &errNum, bvhTraversalProg, "");
+		b3Assert(errNum == CL_SUCCESS);
 	}
-        
-	 {
-		 const char* primitiveContactsSrc = primitiveContactsKernelsCL;
-		cl_program primitiveContactsProg = b3OpenCLUtils::compileCLProgramFromString(m_context,m_device,primitiveContactsSrc,&errNum,"",BT_NARROWPHASE_PRIMITIVE_CONTACT_PATH);
-		b3Assert(errNum==CL_SUCCESS);
 
-		m_primitiveContactsKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,primitiveContactsSrc, "primitiveContactsKernel",&errNum,primitiveContactsProg,"");
-		b3Assert(errNum==CL_SUCCESS);
+	{
+		const char* primitiveContactsSrc = primitiveContactsKernelsCL;
+		cl_program primitiveContactsProg = b3OpenCLUtils::compileCLProgramFromString(m_context, m_device, primitiveContactsSrc, &errNum, "", BT_NARROWPHASE_PRIMITIVE_CONTACT_PATH);
+		b3Assert(errNum == CL_SUCCESS);
 
-		m_findConcaveSphereContactsKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,primitiveContactsSrc, "findConcaveSphereContactsKernel",&errNum,primitiveContactsProg );
-		b3Assert(errNum==CL_SUCCESS);
+		m_primitiveContactsKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, primitiveContactsSrc, "primitiveContactsKernel", &errNum, primitiveContactsProg, "");
+		b3Assert(errNum == CL_SUCCESS);
+
+		m_findConcaveSphereContactsKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, primitiveContactsSrc, "findConcaveSphereContactsKernel", &errNum, primitiveContactsProg);
+		b3Assert(errNum == CL_SUCCESS);
 		b3Assert(m_findConcaveSphereContactsKernel);
 
-		m_processCompoundPairsPrimitivesKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device,primitiveContactsSrc, "processCompoundPairsPrimitivesKernel",&errNum,primitiveContactsProg,"");
-		b3Assert(errNum==CL_SUCCESS);
+		m_processCompoundPairsPrimitivesKernel = b3OpenCLUtils::compileCLKernelFromString(m_context, m_device, primitiveContactsSrc, "processCompoundPairsPrimitivesKernel", &errNum, primitiveContactsProg, "");
+		b3Assert(errNum == CL_SUCCESS);
 		b3Assert(m_processCompoundPairsPrimitivesKernel);
-		 
-	 }
-	
-
+	}
 }
 
 GpuSatCollision::~GpuSatCollision()
 {
-	
 	if (m_findSeparatingAxisVertexFaceKernel)
 		clReleaseKernel(m_findSeparatingAxisVertexFaceKernel);
 
@@ -301,17 +276,15 @@ GpuSatCollision::~GpuSatCollision()
 	if (m_mprPenetrationKernel)
 		clReleaseKernel(m_mprPenetrationKernel);
 
-
 	if (m_findSeparatingAxisKernel)
 		clReleaseKernel(m_findSeparatingAxisKernel);
 
-    if (m_findConcaveSeparatingAxisVertexFaceKernel)
-        clReleaseKernel(m_findConcaveSeparatingAxisVertexFaceKernel);
+	if (m_findConcaveSeparatingAxisVertexFaceKernel)
+		clReleaseKernel(m_findConcaveSeparatingAxisVertexFaceKernel);
+
+	if (m_findConcaveSeparatingAxisEdgeEdgeKernel)
+		clReleaseKernel(m_findConcaveSeparatingAxisEdgeEdgeKernel);
 
-    
-    if (m_findConcaveSeparatingAxisEdgeEdgeKernel)
-        clReleaseKernel(m_findConcaveSeparatingAxisEdgeEdgeKernel);
-    
 	if (m_findConcaveSeparatingAxisKernel)
 		clReleaseKernel(m_findConcaveSeparatingAxisKernel);
 
@@ -320,17 +293,17 @@ GpuSatCollision::~GpuSatCollision()
 
 	if (m_processCompoundPairsKernel)
 		clReleaseKernel(m_processCompoundPairsKernel);
-    
-    if (m_findClippingFacesKernel)
-        clReleaseKernel(m_findClippingFacesKernel);
-   
-    if (m_clipFacesAndFindContacts)
-        clReleaseKernel(m_clipFacesAndFindContacts);
-    if (m_newContactReductionKernel)
-        clReleaseKernel(m_newContactReductionKernel);
+
+	if (m_findClippingFacesKernel)
+		clReleaseKernel(m_findClippingFacesKernel);
+
+	if (m_clipFacesAndFindContacts)
+		clReleaseKernel(m_clipFacesAndFindContacts);
+	if (m_newContactReductionKernel)
+		clReleaseKernel(m_newContactReductionKernel);
 	if (m_primitiveContactsKernel)
 		clReleaseKernel(m_primitiveContactsKernel);
-    
+
 	if (m_findConcaveSphereContactsKernel)
 		clReleaseKernel(m_findConcaveSphereContactsKernel);
 
@@ -344,12 +317,11 @@ GpuSatCollision::~GpuSatCollision()
 
 	if (m_clipHullHullConcaveConvexKernel)
 		clReleaseKernel(m_clipHullHullConcaveConvexKernel);
-//	if (m_extractManifoldAndAddContactKernel)
+	//	if (m_extractManifoldAndAddContactKernel)
 	//	clReleaseKernel(m_extractManifoldAndAddContactKernel);
 
 	if (m_bvhTraversalKernel)
 		clReleaseKernel(m_bvhTraversalKernel);
-
 }
 
 struct MyTriangleCallback : public b3NodeOverlapCallback
@@ -359,14 +331,13 @@ struct MyTriangleCallback : public b3NodeOverlapCallback
 
 	virtual void processNode(int subPart, int triangleIndex)
 	{
-		printf("bodyIndexA %d, bodyIndexB %d\n",m_bodyIndexA,m_bodyIndexB);
+		printf("bodyIndexA %d, bodyIndexB %d\n", m_bodyIndexA, m_bodyIndexB);
 		printf("triangleIndex %d\n", triangleIndex);
 	}
 };
 
-
 #define float4 b3Vector3
-#define make_float4(x,y,z,w) b3MakeVector3(x,y,z,w)
+#define make_float4(x, y, z, w) b3MakeVector3(x, y, z, w)
 
 float signedDistanceFromPointToPlane(const float4& point, const float4& planeEqn, float4* closestPointOnFace)
 {
@@ -377,9 +348,7 @@ float signedDistanceFromPointToPlane(const float4& point, const float4& planeEqn
 	return dist;
 }
 
-
-
-#define cross3(a,b) (a.cross(b))
+#define cross3(a, b) (a.cross(b))
 b3Vector3 transform(const b3Vector3* v, const b3Vector3* pos, const b3Quaternion* orn)
 {
 	b3Transform tr;
@@ -390,184 +359,170 @@ b3Vector3 transform(const b3Vector3* v, const b3Vector3* pos, const b3Quaternion
 	return res;
 }
 
-
-inline bool IsPointInPolygon(const float4& p, 
-							const b3GpuFace* face,
+inline bool IsPointInPolygon(const float4& p,
+							 const b3GpuFace* face,
 							 const float4* baseVertex,
-							const  int* convexIndices,
-							float4* out)
+							 const int* convexIndices,
+							 float4* out)
 {
-    float4 a;
-    float4 b;
-    float4 ab;
-    float4 ap;
-    float4 v;
+	float4 a;
+	float4 b;
+	float4 ab;
+	float4 ap;
+	float4 v;
 
-	float4 plane = b3MakeVector3(face->m_plane.x,face->m_plane.y,face->m_plane.z,0.f);
-	
-	if (face->m_numIndices<2)
+	float4 plane = b3MakeVector3(face->m_plane.x, face->m_plane.y, face->m_plane.z, 0.f);
+
+	if (face->m_numIndices < 2)
 		return false;
 
-	
-	float4 v0 = baseVertex[convexIndices[face->m_indexOffset + face->m_numIndices-1]];
+	float4 v0 = baseVertex[convexIndices[face->m_indexOffset + face->m_numIndices - 1]];
 	b = v0;
 
-    for(unsigned i=0; i != face->m_numIndices; ++i)
-    {
+	for (unsigned i = 0; i != face->m_numIndices; ++i)
+	{
 		a = b;
 		float4 vi = baseVertex[convexIndices[face->m_indexOffset + i]];
 		b = vi;
-        ab = b-a;
-        ap = p-a;
-        v = cross3(ab,plane);
+		ab = b - a;
+		ap = p - a;
+		v = cross3(ab, plane);
 
-        if (b3Dot(ap, v) > 0.f)
-        {
-            float ab_m2 = b3Dot(ab, ab);
-            float rt = ab_m2 != 0.f ? b3Dot(ab, ap) / ab_m2 : 0.f;
-            if (rt <= 0.f)
-            {
-                *out = a;
-            }
-            else if (rt >= 1.f) 
-            {
-                *out = b;
-            }
-            else
-            {
-            	float s = 1.f - rt;
+		if (b3Dot(ap, v) > 0.f)
+		{
+			float ab_m2 = b3Dot(ab, ab);
+			float rt = ab_m2 != 0.f ? b3Dot(ab, ap) / ab_m2 : 0.f;
+			if (rt <= 0.f)
+			{
+				*out = a;
+			}
+			else if (rt >= 1.f)
+			{
+				*out = b;
+			}
+			else
+			{
+				float s = 1.f - rt;
 				out[0].x = s * a.x + rt * b.x;
 				out[0].y = s * a.y + rt * b.y;
 				out[0].z = s * a.z + rt * b.z;
-            }
-            return false;
-        }
-    }
-    return true;
+			}
+			return false;
+		}
+	}
+	return true;
 }
 
 #define normalize3(a) (a.normalize())
 
-
-int extractManifoldSequentialGlobal( const float4* p, int nPoints, const float4& nearNormal, b3Int4* contactIdx)
+int extractManifoldSequentialGlobal(const float4* p, int nPoints, const float4& nearNormal, b3Int4* contactIdx)
 {
-	if( nPoints == 0 )
-        return 0;
-    
-    if (nPoints <=4)
-        return nPoints;
-    
-    
-    if (nPoints >64)
-        nPoints = 64;
-    
-	float4 center = b3MakeVector3(0,0,0,0);
+	if (nPoints == 0)
+		return 0;
+
+	if (nPoints <= 4)
+		return nPoints;
+
+	if (nPoints > 64)
+		nPoints = 64;
+
+	float4 center = b3MakeVector3(0, 0, 0, 0);
 	{
-		
-		for (int i=0;i<nPoints;i++)
+		for (int i = 0; i < nPoints; i++)
 			center += p[i];
 		center /= (float)nPoints;
 	}
-    
-	
-    
+
 	//	sample 4 directions
-    
-    float4 aVector = p[0] - center;
-    float4 u = cross3( nearNormal, aVector );
-    float4 v = cross3( nearNormal, u );
-    u = normalize3( u );
-    v = normalize3( v );
-    
-    
-    //keep point with deepest penetration
-    float minW= FLT_MAX;
-    
-    int minIndex=-1;
-    
-    float4 maxDots;
-    maxDots.x = FLT_MIN;
-    maxDots.y = FLT_MIN;
-    maxDots.z = FLT_MIN;
-    maxDots.w = FLT_MIN;
-    
-    //	idx, distance
-    for(int ie = 0; ie<nPoints; ie++ )
-    {
-        if (p[ie].w<minW)
-        {
-            minW = p[ie].w;
-            minIndex=ie;
-        }
-        float f;
-        float4 r = p[ie]-center;
-        f = dot3F4( u, r );
-        if (f<maxDots.x)
-        {
-            maxDots.x = f;
-            contactIdx[0].x = ie;
-        }
-        
-        f = dot3F4( -u, r );
-        if (f<maxDots.y)
-        {
-            maxDots.y = f;
-            contactIdx[0].y = ie;
-        }
-        
-        
-        f = dot3F4( v, r );
-        if (f<maxDots.z)
-        {
-            maxDots.z = f;
-            contactIdx[0].z = ie;
-        }
-        
-        f = dot3F4( -v, r );
-        if (f<maxDots.w)
-        {
-            maxDots.w = f;
-            contactIdx[0].w = ie;
-        }
-        
-    }
-    
-    if (contactIdx[0].x != minIndex && contactIdx[0].y != minIndex && contactIdx[0].z != minIndex && contactIdx[0].w != minIndex)
-    {
-        //replace the first contact with minimum (todo: replace contact with least penetration)
-        contactIdx[0].x = minIndex;
-    }
-    
-    return 4;
-    
+
+	float4 aVector = p[0] - center;
+	float4 u = cross3(nearNormal, aVector);
+	float4 v = cross3(nearNormal, u);
+	u = normalize3(u);
+	v = normalize3(v);
+
+	//keep point with deepest penetration
+	float minW = FLT_MAX;
+
+	int minIndex = -1;
+
+	float4 maxDots;
+	maxDots.x = FLT_MIN;
+	maxDots.y = FLT_MIN;
+	maxDots.z = FLT_MIN;
+	maxDots.w = FLT_MIN;
+
+	//	idx, distance
+	for (int ie = 0; ie < nPoints; ie++)
+	{
+		if (p[ie].w < minW)
+		{
+			minW = p[ie].w;
+			minIndex = ie;
+		}
+		float f;
+		float4 r = p[ie] - center;
+		f = dot3F4(u, r);
+		if (f < maxDots.x)
+		{
+			maxDots.x = f;
+			contactIdx[0].x = ie;
+		}
+
+		f = dot3F4(-u, r);
+		if (f < maxDots.y)
+		{
+			maxDots.y = f;
+			contactIdx[0].y = ie;
+		}
+
+		f = dot3F4(v, r);
+		if (f < maxDots.z)
+		{
+			maxDots.z = f;
+			contactIdx[0].z = ie;
+		}
+
+		f = dot3F4(-v, r);
+		if (f < maxDots.w)
+		{
+			maxDots.w = f;
+			contactIdx[0].w = ie;
+		}
+	}
+
+	if (contactIdx[0].x != minIndex && contactIdx[0].y != minIndex && contactIdx[0].z != minIndex && contactIdx[0].w != minIndex)
+	{
+		//replace the first contact with minimum (todo: replace contact with least penetration)
+		contactIdx[0].x = minIndex;
+	}
+
+	return 4;
 }
 
-
-
 #define MAX_VERTS 1024
 
-
-inline void project(const b3ConvexPolyhedronData& hull,  const float4& pos, const b3Quaternion& orn, const float4& dir, const b3AlignedObjectArray<b3Vector3>& vertices, b3Scalar& min, b3Scalar& max)
+inline void project(const b3ConvexPolyhedronData& hull, const float4& pos, const b3Quaternion& orn, const float4& dir, const b3AlignedObjectArray<b3Vector3>& vertices, b3Scalar& min, b3Scalar& max)
 {
 	min = FLT_MAX;
 	max = -FLT_MAX;
 	int numVerts = hull.m_numVertices;
 
-	const float4 localDir = b3QuatRotate(orn.inverse(),dir);
+	const float4 localDir = b3QuatRotate(orn.inverse(), dir);
 
-	b3Scalar offset = dot3F4(pos,dir);
+	b3Scalar offset = dot3F4(pos, dir);
 
-	for(int i=0;i<numVerts;i++)
+	for (int i = 0; i < numVerts; i++)
 	{
 		//b3Vector3 pt = trans * vertices[m_vertexOffset+i];
 		//b3Scalar dp = pt.dot(dir);
 		//b3Vector3 vertex = vertices[hull.m_vertexOffset+i];
-		b3Scalar dp = dot3F4((float4&)vertices[hull.m_vertexOffset+i],localDir);
+		b3Scalar dp = dot3F4((float4&)vertices[hull.m_vertexOffset + i], localDir);
 		//b3Assert(dp==dpL);
-		if(dp < min)	min = dp;
-		if(dp > max)	max = dp;
+		if (dp < min) min = dp;
+		if (dp > max) max = dp;
 	}
-	if(min>max)
+	if (min > max)
 	{
 		b3Scalar tmp = min;
 		min = max;
@@ -577,50 +532,48 @@ inline void project(const b3ConvexPolyhedronData& hull,  const float4& pos, cons
 	max += offset;
 }
 
-
-static bool TestSepAxis(const b3ConvexPolyhedronData& hullA, const b3ConvexPolyhedronData& hullB, 
-	const float4& posA,const b3Quaternion& ornA,
-	const float4& posB,const b3Quaternion& ornB,
-	const float4& sep_axis, const b3AlignedObjectArray<b3Vector3>& verticesA,const b3AlignedObjectArray<b3Vector3>& verticesB,b3Scalar& depth)
+static bool TestSepAxis(const b3ConvexPolyhedronData& hullA, const b3ConvexPolyhedronData& hullB,
+						const float4& posA, const b3Quaternion& ornA,
+						const float4& posB, const b3Quaternion& ornB,
+						const float4& sep_axis, const b3AlignedObjectArray<b3Vector3>& verticesA, const b3AlignedObjectArray<b3Vector3>& verticesB, b3Scalar& depth)
 {
-	b3Scalar Min0,Max0;
-	b3Scalar Min1,Max1;
-	project(hullA,posA,ornA,sep_axis,verticesA, Min0, Max0);
-	project(hullB,posB,ornB, sep_axis,verticesB, Min1, Max1);
+	b3Scalar Min0, Max0;
+	b3Scalar Min1, Max1;
+	project(hullA, posA, ornA, sep_axis, verticesA, Min0, Max0);
+	project(hullB, posB, ornB, sep_axis, verticesB, Min1, Max1);
 
-	if(Max0<Min1 || Max1<Min0)
+	if (Max0 < Min1 || Max1 < Min0)
 		return false;
 
 	b3Scalar d0 = Max0 - Min1;
-	assert(d0>=0.0f);
+	assert(d0 >= 0.0f);
 	b3Scalar d1 = Max1 - Min0;
-	assert(d1>=0.0f);
-	depth = d0<d1 ? d0:d1;
+	assert(d1 >= 0.0f);
+	depth = d0 < d1 ? d0 : d1;
 	return true;
 }
 
 inline bool IsAlmostZero(const b3Vector3& v)
 {
-	if(fabsf(v.x)>1e-6 || fabsf(v.y)>1e-6 || fabsf(v.z)>1e-6)	return false;
+	if (fabsf(v.x) > 1e-6 || fabsf(v.y) > 1e-6 || fabsf(v.z) > 1e-6) return false;
 	return true;
 }
 
+static bool findSeparatingAxis(const b3ConvexPolyhedronData& hullA, const b3ConvexPolyhedronData& hullB,
+							   const float4& posA1,
+							   const b3Quaternion& ornA,
+							   const float4& posB1,
+							   const b3Quaternion& ornB,
+							   const b3AlignedObjectArray<b3Vector3>& verticesA,
+							   const b3AlignedObjectArray<b3Vector3>& uniqueEdgesA,
+							   const b3AlignedObjectArray<b3GpuFace>& facesA,
+							   const b3AlignedObjectArray<int>& indicesA,
+							   const b3AlignedObjectArray<b3Vector3>& verticesB,
+							   const b3AlignedObjectArray<b3Vector3>& uniqueEdgesB,
+							   const b3AlignedObjectArray<b3GpuFace>& facesB,
+							   const b3AlignedObjectArray<int>& indicesB,
 
-static bool findSeparatingAxis(	const b3ConvexPolyhedronData& hullA, const b3ConvexPolyhedronData& hullB, 
-	const float4& posA1,
-	const b3Quaternion& ornA,
-	const float4& posB1,
-	const b3Quaternion& ornB,
-	const b3AlignedObjectArray<b3Vector3>& verticesA,
-	const b3AlignedObjectArray<b3Vector3>& uniqueEdgesA, 
-	const b3AlignedObjectArray<b3GpuFace>& facesA,
-	const b3AlignedObjectArray<int>& indicesA,
-	const b3AlignedObjectArray<b3Vector3>& verticesB, 
-	const b3AlignedObjectArray<b3Vector3>& uniqueEdgesB, 
-	const b3AlignedObjectArray<b3GpuFace>& facesB,
-	const b3AlignedObjectArray<int>& indicesB,
-
-	b3Vector3& sep)
+							   b3Vector3& sep)
 {
 	B3_PROFILE("findSeparatingAxis");
 
@@ -629,41 +582,40 @@ static bool findSeparatingAxis(	const b3ConvexPolyhedronData& hullA, const b3Con
 	posA.w = 0.f;
 	float4 posB = posB1;
 	posB.w = 0.f;
-//#ifdef TEST_INTERNAL_OBJECTS
+	//#ifdef TEST_INTERNAL_OBJECTS
 	float4 c0local = (float4&)hullA.m_localCenter;
 	float4 c0 = transform(&c0local, &posA, &ornA);
 	float4 c1local = (float4&)hullB.m_localCenter;
-	float4 c1 = transform(&c1local,&posB,&ornB);
+	float4 c1 = transform(&c1local, &posB, &ornB);
 	const float4 deltaC2 = c0 - c1;
-//#endif
+	//#endif
 
 	b3Scalar dmin = FLT_MAX;
-	int curPlaneTests=0;
+	int curPlaneTests = 0;
 
 	int numFacesA = hullA.m_numFaces;
 	// Test normals from hullA
-	for(int i=0;i<numFacesA;i++)
+	for (int i = 0; i < numFacesA; i++)
 	{
-		const float4& normal = (float4&)facesA[hullA.m_faceOffset+i].m_plane;
-		float4 faceANormalWS = b3QuatRotate(ornA,normal);
+		const float4& normal = (float4&)facesA[hullA.m_faceOffset + i].m_plane;
+		float4 faceANormalWS = b3QuatRotate(ornA, normal);
 
-		if (dot3F4(deltaC2,faceANormalWS)<0)
-			faceANormalWS*=-1.f;
+		if (dot3F4(deltaC2, faceANormalWS) < 0)
+			faceANormalWS *= -1.f;
 
 		curPlaneTests++;
 #ifdef TEST_INTERNAL_OBJECTS
 		gExpectedNbTests++;
-		if(gUseInternalObject && !TestInternalObjects(transA,transB, DeltaC2, faceANormalWS, hullA, hullB, dmin))
+		if (gUseInternalObject && !TestInternalObjects(transA, transB, DeltaC2, faceANormalWS, hullA, hullB, dmin))
 			continue;
 		gActualNbTests++;
 #endif
 
-		
 		b3Scalar d;
-		if(!TestSepAxis( hullA, hullB, posA,ornA,posB,ornB,faceANormalWS, verticesA, verticesB,d))
+		if (!TestSepAxis(hullA, hullB, posA, ornA, posB, ornB, faceANormalWS, verticesA, verticesB, d))
 			return false;
 
-		if(d<dmin)
+		if (d < dmin)
 		{
 			dmin = d;
 			sep = (b3Vector3&)faceANormalWS;
@@ -672,28 +624,28 @@ static bool findSeparatingAxis(	const b3ConvexPolyhedronData& hullA, const b3Con
 
 	int numFacesB = hullB.m_numFaces;
 	// Test normals from hullB
-	for(int i=0;i<numFacesB;i++)
+	for (int i = 0; i < numFacesB; i++)
 	{
-		float4 normal = (float4&)facesB[hullB.m_faceOffset+i].m_plane;
+		float4 normal = (float4&)facesB[hullB.m_faceOffset + i].m_plane;
 		float4 WorldNormal = b3QuatRotate(ornB, normal);
 
-		if (dot3F4(deltaC2,WorldNormal)<0)
+		if (dot3F4(deltaC2, WorldNormal) < 0)
 		{
-			WorldNormal*=-1.f;
+			WorldNormal *= -1.f;
 		}
 		curPlaneTests++;
 #ifdef TEST_INTERNAL_OBJECTS
 		gExpectedNbTests++;
-		if(gUseInternalObject && !TestInternalObjects(transA,transB,DeltaC2, WorldNormal, hullA, hullB, dmin))
+		if (gUseInternalObject && !TestInternalObjects(transA, transB, DeltaC2, WorldNormal, hullA, hullB, dmin))
 			continue;
 		gActualNbTests++;
 #endif
 
 		b3Scalar d;
-		if(!TestSepAxis(hullA, hullB,posA,ornA,posB,ornB,WorldNormal,verticesA,verticesB,d))
+		if (!TestSepAxis(hullA, hullB, posA, ornA, posB, ornB, WorldNormal, verticesA, verticesB, d))
 			return false;
 
-		if(d<dmin)
+		if (d < dmin)
 		{
 			dmin = d;
 			sep = (b3Vector3&)WorldNormal;
@@ -702,70 +654,65 @@ static bool findSeparatingAxis(	const b3ConvexPolyhedronData& hullA, const b3Con
 
 	int curEdgeEdge = 0;
 	// Test edges
-	for(int e0=0;e0<hullA.m_numUniqueEdges;e0++)
+	for (int e0 = 0; e0 < hullA.m_numUniqueEdges; e0++)
 	{
-		const float4& edge0 = (float4&) uniqueEdgesA[hullA.m_uniqueEdgesOffset+e0];
-		float4 edge0World = b3QuatRotate(ornA,(float4&)edge0);
+		const float4& edge0 = (float4&)uniqueEdgesA[hullA.m_uniqueEdgesOffset + e0];
+		float4 edge0World = b3QuatRotate(ornA, (float4&)edge0);
 
-		for(int e1=0;e1<hullB.m_numUniqueEdges;e1++)
+		for (int e1 = 0; e1 < hullB.m_numUniqueEdges; e1++)
 		{
-			const b3Vector3 edge1 = uniqueEdgesB[hullB.m_uniqueEdgesOffset+e1];
-			float4 edge1World = b3QuatRotate(ornB,(float4&)edge1);
+			const b3Vector3 edge1 = uniqueEdgesB[hullB.m_uniqueEdgesOffset + e1];
+			float4 edge1World = b3QuatRotate(ornB, (float4&)edge1);
 
-
-			float4 crossje = cross3(edge0World,edge1World);
+			float4 crossje = cross3(edge0World, edge1World);
 
 			curEdgeEdge++;
-			if(!IsAlmostZero((b3Vector3&)crossje))
+			if (!IsAlmostZero((b3Vector3&)crossje))
 			{
 				crossje = normalize3(crossje);
-				if (dot3F4(deltaC2,crossje)<0)
-					crossje*=-1.f;
-
+				if (dot3F4(deltaC2, crossje) < 0)
+					crossje *= -1.f;
 
 #ifdef TEST_INTERNAL_OBJECTS
 				gExpectedNbTests++;
-				if(gUseInternalObject && !TestInternalObjects(transA,transB,DeltaC2, Cross, hullA, hullB, dmin))
+				if (gUseInternalObject && !TestInternalObjects(transA, transB, DeltaC2, Cross, hullA, hullB, dmin))
 					continue;
 				gActualNbTests++;
 #endif
 
 				b3Scalar dist;
-				if(!TestSepAxis( hullA, hullB, posA,ornA,posB,ornB,crossje, verticesA,verticesB,dist))
+				if (!TestSepAxis(hullA, hullB, posA, ornA, posB, ornB, crossje, verticesA, verticesB, dist))
 					return false;
 
-				if(dist<dmin)
+				if (dist < dmin)
 				{
 					dmin = dist;
 					sep = (b3Vector3&)crossje;
 				}
 			}
 		}
-
 	}
 
-	
-	if((dot3F4(-deltaC2,(float4&)sep))>0.0f)
+	if ((dot3F4(-deltaC2, (float4&)sep)) > 0.0f)
 		sep = -sep;
 
 	return true;
 }
 
-
-bool findSeparatingAxisEdgeEdge(	__global const b3ConvexPolyhedronData* hullA, __global const b3ConvexPolyhedronData* hullB, 
-	const b3Float4& posA1,
-	const b3Quat& ornA,
-	const b3Float4& posB1,
-	const b3Quat& ornB,
-	const b3Float4& DeltaC2,
-	__global const b3AlignedObjectArray<float4>& vertices, 
-	__global const b3AlignedObjectArray<float4>& uniqueEdges, 
-	__global const b3AlignedObjectArray<b3GpuFace>& faces,
-	__global const b3AlignedObjectArray<int>&  indices,
-	float4* sep,
-	float* dmin)
+bool findSeparatingAxisEdgeEdge(__global const b3ConvexPolyhedronData* hullA, __global const b3ConvexPolyhedronData* hullB,
+								const b3Float4& posA1,
+								const b3Quat& ornA,
+								const b3Float4& posB1,
+								const b3Quat& ornB,
+								const b3Float4& DeltaC2,
+								__global const b3AlignedObjectArray<float4>& vertices,
+								__global const b3AlignedObjectArray<float4>& uniqueEdges,
+								__global const b3AlignedObjectArray<b3GpuFace>& faces,
+								__global const b3AlignedObjectArray<int>& indices,
+								float4* sep,
+								float* dmin)
 {
-//	int i = get_global_id(0);
+	//	int i = get_global_id(0);
 
 	float4 posA = posA1;
 	posA.w = 0.f;
@@ -776,97 +723,89 @@ bool findSeparatingAxisEdgeEdge(	__global const b3ConvexPolyhedronData* hullA, _
 
 	int curEdgeEdge = 0;
 	// Test edges
-	for(int e0=0;e0<hullA->m_numUniqueEdges;e0++)
+	for (int e0 = 0; e0 < hullA->m_numUniqueEdges; e0++)
 	{
-		const float4 edge0 = uniqueEdges[hullA->m_uniqueEdgesOffset+e0];
-		float4 edge0World = b3QuatRotate(ornA,edge0);
+		const float4 edge0 = uniqueEdges[hullA->m_uniqueEdgesOffset + e0];
+		float4 edge0World = b3QuatRotate(ornA, edge0);
 
-		for(int e1=0;e1<hullB->m_numUniqueEdges;e1++)
+		for (int e1 = 0; e1 < hullB->m_numUniqueEdges; e1++)
 		{
-			const float4 edge1 = uniqueEdges[hullB->m_uniqueEdgesOffset+e1];
-			float4 edge1World = b3QuatRotate(ornB,edge1);
+			const float4 edge1 = uniqueEdges[hullB->m_uniqueEdgesOffset + e1];
+			float4 edge1World = b3QuatRotate(ornB, edge1);
 
-
-			float4 crossje = cross3(edge0World,edge1World);
+			float4 crossje = cross3(edge0World, edge1World);
 
 			curEdgeEdge++;
-			if(!IsAlmostZero(crossje))
+			if (!IsAlmostZero(crossje))
 			{
 				crossje = normalize3(crossje);
-				if (dot3F4(DeltaC2,crossje)<0)
-					crossje*=-1.f;
-					
+				if (dot3F4(DeltaC2, crossje) < 0)
+					crossje *= -1.f;
+
 				float dist;
 				bool result = true;
 				{
-					float Min0,Max0;
-					float Min1,Max1;
-					project(*hullA,posA,ornA,crossje,vertices, Min0, Max0);
-					project(*hullB,posB,ornB,crossje,vertices, Min1, Max1);
-				
-					if(Max0<Min1 || Max1<Min0)
+					float Min0, Max0;
+					float Min1, Max1;
+					project(*hullA, posA, ornA, crossje, vertices, Min0, Max0);
+					project(*hullB, posB, ornB, crossje, vertices, Min1, Max1);
+
+					if (Max0 < Min1 || Max1 < Min0)
 						result = false;
-				
+
 					float d0 = Max0 - Min1;
 					float d1 = Max1 - Min0;
-					dist = d0<d1 ? d0:d1;
+					dist = d0 < d1 ? d0 : d1;
 					result = true;
-
 				}
-				
 
-				if(dist<*dmin)
+				if (dist < *dmin)
 				{
 					*dmin = dist;
 					*sep = crossje;
 				}
 			}
 		}
-
 	}
 
-	
-	if((dot3F4(-DeltaC2,*sep))>0.0f)
+	if ((dot3F4(-DeltaC2, *sep)) > 0.0f)
 	{
 		*sep = -(*sep);
 	}
 	return true;
 }
 
-
-__inline float4 lerp3(const float4& a,const float4& b, float  t)
+__inline float4 lerp3(const float4& a, const float4& b, float t)
 {
-	return b3MakeVector3(	a.x + (b.x - a.x) * t,
-						a.y + (b.y - a.y) * t,
-						a.z + (b.z - a.z) * t,
-						0.f);
+	return b3MakeVector3(a.x + (b.x - a.x) * t,
+						 a.y + (b.y - a.y) * t,
+						 a.z + (b.z - a.z) * t,
+						 0.f);
 }
 
-
 // Clips a face to the back of a plane, return the number of vertices out, stored in ppVtxOut
-int clipFace(const float4* pVtxIn, int numVertsIn, float4& planeNormalWS,float planeEqWS, float4* ppVtxOut)
+int clipFace(const float4* pVtxIn, int numVertsIn, float4& planeNormalWS, float planeEqWS, float4* ppVtxOut)
 {
-	
 	int ve;
 	float ds, de;
 	int numVertsOut = 0;
 	if (numVertsIn < 2)
 		return 0;
 
-	float4 firstVertex=pVtxIn[numVertsIn-1];
+	float4 firstVertex = pVtxIn[numVertsIn - 1];
 	float4 endVertex = pVtxIn[0];
-	
-	ds = dot3F4(planeNormalWS,firstVertex)+planeEqWS;
+
+	ds = dot3F4(planeNormalWS, firstVertex) + planeEqWS;
 
 	for (ve = 0; ve < numVertsIn; ve++)
 	{
-		endVertex=pVtxIn[ve];
+		endVertex = pVtxIn[ve];
 
-		de = dot3F4(planeNormalWS,endVertex)+planeEqWS;
+		de = dot3F4(planeNormalWS, endVertex) + planeEqWS;
 
-		if (ds<0)
+		if (ds < 0)
 		{
-			if (de<0)
+			if (de < 0)
 			{
 				// Start < 0, end < 0, so output endVertex
 				ppVtxOut[numVertsOut++] = endVertex;
@@ -874,15 +813,15 @@ int clipFace(const float4* pVtxIn, int numVertsIn, float4& planeNormalWS,float p
 			else
 			{
 				// Start < 0, end >= 0, so output intersection
-				ppVtxOut[numVertsOut++] = lerp3(firstVertex, endVertex,(ds * 1.f/(ds - de)) );
+				ppVtxOut[numVertsOut++] = lerp3(firstVertex, endVertex, (ds * 1.f / (ds - de)));
 			}
 		}
 		else
 		{
-			if (de<0)
+			if (de < 0)
 			{
 				// Start >= 0, end < 0 so output intersection and end
-				ppVtxOut[numVertsOut++] = lerp3(firstVertex, endVertex,(ds * 1.f/(ds - de)) );
+				ppVtxOut[numVertsOut++] = lerp3(firstVertex, endVertex, (ds * 1.f / (ds - de)));
 				ppVtxOut[numVertsOut++] = endVertex;
 			}
 		}
@@ -892,36 +831,35 @@ int clipFace(const float4* pVtxIn, int numVertsIn, float4& planeNormalWS,float p
 	return numVertsOut;
 }
 
-
-int clipFaceAgainstHull(const float4& separatingNormal, const b3ConvexPolyhedronData* hullA,  
-	const float4& posA, const b3Quaternion& ornA, float4* worldVertsB1, int numWorldVertsB1,
-	float4* worldVertsB2, int capacityWorldVertsB2,
-	const float minDist, float maxDist,
-	const b3AlignedObjectArray<float4>& verticesA,	const b3AlignedObjectArray<b3GpuFace>& facesA,	const b3AlignedObjectArray<int>& indicesA,
-	//const float4* verticesB,	const b3GpuFace* facesB,	const int* indicesB,
-	float4* contactsOut,
-	int contactCapacity)
+int clipFaceAgainstHull(const float4& separatingNormal, const b3ConvexPolyhedronData* hullA,
+						const float4& posA, const b3Quaternion& ornA, float4* worldVertsB1, int numWorldVertsB1,
+						float4* worldVertsB2, int capacityWorldVertsB2,
+						const float minDist, float maxDist,
+						const b3AlignedObjectArray<float4>& verticesA, const b3AlignedObjectArray<b3GpuFace>& facesA, const b3AlignedObjectArray<int>& indicesA,
+						//const float4* verticesB,	const b3GpuFace* facesB,	const int* indicesB,
+						float4* contactsOut,
+						int contactCapacity)
 {
 	int numContactsOut = 0;
 
 	float4* pVtxIn = worldVertsB1;
 	float4* pVtxOut = worldVertsB2;
-	
+
 	int numVertsIn = numWorldVertsB1;
 	int numVertsOut = 0;
 
-	int closestFaceA=-1;
+	int closestFaceA = -1;
 	{
 		float dmin = FLT_MAX;
-		for(int face=0;face<hullA->m_numFaces;face++)
+		for (int face = 0; face < hullA->m_numFaces; face++)
 		{
 			const float4 Normal = b3MakeVector3(
-				facesA[hullA->m_faceOffset+face].m_plane.x, 
-				facesA[hullA->m_faceOffset+face].m_plane.y, 
-				facesA[hullA->m_faceOffset+face].m_plane.z,0.f);
-			const float4 faceANormalWS = b3QuatRotate(ornA,Normal);
-		
-			float d = dot3F4(faceANormalWS,separatingNormal);
+				facesA[hullA->m_faceOffset + face].m_plane.x,
+				facesA[hullA->m_faceOffset + face].m_plane.y,
+				facesA[hullA->m_faceOffset + face].m_plane.z, 0.f);
+			const float4 faceANormalWS = b3QuatRotate(ornA, Normal);
+
+			float d = dot3F4(faceANormalWS, separatingNormal);
 			if (d < dmin)
 			{
 				dmin = d;
@@ -929,33 +867,33 @@ int clipFaceAgainstHull(const float4& separatingNormal, const b3ConvexPolyhedron
 			}
 		}
 	}
-	if (closestFaceA<0)
+	if (closestFaceA < 0)
 		return numContactsOut;
 
-	b3GpuFace polyA = facesA[hullA->m_faceOffset+closestFaceA];
+	b3GpuFace polyA = facesA[hullA->m_faceOffset + closestFaceA];
 
 	// clip polygon to back of planes of all faces of hull A that are adjacent to witness face
-//	int numContacts = numWorldVertsB1;
+	//	int numContacts = numWorldVertsB1;
 	int numVerticesA = polyA.m_numIndices;
-	for(int e0=0;e0<numVerticesA;e0++)
+	for (int e0 = 0; e0 < numVerticesA; e0++)
 	{
-		const float4 a = verticesA[hullA->m_vertexOffset+indicesA[polyA.m_indexOffset+e0]];
-		const float4 b = verticesA[hullA->m_vertexOffset+indicesA[polyA.m_indexOffset+((e0+1)%numVerticesA)]];
+		const float4 a = verticesA[hullA->m_vertexOffset + indicesA[polyA.m_indexOffset + e0]];
+		const float4 b = verticesA[hullA->m_vertexOffset + indicesA[polyA.m_indexOffset + ((e0 + 1) % numVerticesA)]];
 		const float4 edge0 = a - b;
-		const float4 WorldEdge0 = b3QuatRotate(ornA,edge0);
-		float4 planeNormalA = make_float4(polyA.m_plane.x,polyA.m_plane.y,polyA.m_plane.z,0.f);
-		float4 worldPlaneAnormal1 = b3QuatRotate(ornA,planeNormalA);
+		const float4 WorldEdge0 = b3QuatRotate(ornA, edge0);
+		float4 planeNormalA = make_float4(polyA.m_plane.x, polyA.m_plane.y, polyA.m_plane.z, 0.f);
+		float4 worldPlaneAnormal1 = b3QuatRotate(ornA, planeNormalA);
+
+		float4 planeNormalWS1 = -cross3(WorldEdge0, worldPlaneAnormal1);
+		float4 worldA1 = transform(&a, &posA, &ornA);
+		float planeEqWS1 = -dot3F4(worldA1, planeNormalWS1);
 
-		float4 planeNormalWS1 = -cross3(WorldEdge0,worldPlaneAnormal1);
-		float4 worldA1 = transform(&a,&posA,&ornA);
-		float planeEqWS1 = -dot3F4(worldA1,planeNormalWS1);
-		
 		float4 planeNormalWS = planeNormalWS1;
-		float planeEqWS=planeEqWS1;
-		
+		float planeEqWS = planeEqWS1;
+
 		//clip face
 		//clipFace(*pVtxIn, *pVtxOut,planeNormalWS,planeEqWS);
-		numVertsOut = clipFace(pVtxIn, numVertsIn, planeNormalWS,planeEqWS, pVtxOut);
+		numVertsOut = clipFace(pVtxIn, numVertsIn, planeNormalWS, planeEqWS, pVtxOut);
 
 		//btSwap(pVtxIn,pVtxOut);
 		float4* tmp = pVtxOut;
@@ -965,32 +903,32 @@ int clipFaceAgainstHull(const float4& separatingNormal, const b3ConvexPolyhedron
 		numVertsOut = 0;
 	}
 
-	
 	// only keep points that are behind the witness face
 	{
-		float4 localPlaneNormal  = make_float4(polyA.m_plane.x,polyA.m_plane.y,polyA.m_plane.z,0.f);
+		float4 localPlaneNormal = make_float4(polyA.m_plane.x, polyA.m_plane.y, polyA.m_plane.z, 0.f);
 		float localPlaneEq = polyA.m_plane.w;
-		float4 planeNormalWS = b3QuatRotate(ornA,localPlaneNormal);
-		float planeEqWS=localPlaneEq-dot3F4(planeNormalWS,posA);
-		for (int i=0;i<numVertsIn;i++)
+		float4 planeNormalWS = b3QuatRotate(ornA, localPlaneNormal);
+		float planeEqWS = localPlaneEq - dot3F4(planeNormalWS, posA);
+		for (int i = 0; i < numVertsIn; i++)
 		{
-			float depth = dot3F4(planeNormalWS,pVtxIn[i])+planeEqWS;
-			if (depth <=minDist)
+			float depth = dot3F4(planeNormalWS, pVtxIn[i]) + planeEqWS;
+			if (depth <= minDist)
 			{
 				depth = minDist;
 			}
-			if (numContactsOut<contactCapacity)
+			if (numContactsOut < contactCapacity)
 			{
-				if (depth <=maxDist)
+				if (depth <= maxDist)
 				{
 					float4 pointInWorld = pVtxIn[i];
 					//resultOut.addContactPoint(separatingNormal,point,depth);
-					contactsOut[numContactsOut++] = b3MakeVector3(pointInWorld.x,pointInWorld.y,pointInWorld.z,depth);
+					contactsOut[numContactsOut++] = b3MakeVector3(pointInWorld.x, pointInWorld.y, pointInWorld.z, depth);
 					//printf("depth=%f\n",depth);
 				}
-			} else
+			}
+			else
 			{
-				b3Error("exceeding contact capacity (%d,%df)\n", numContactsOut,contactCapacity);
+				b3Error("exceeding contact capacity (%d,%df)\n", numContactsOut, contactCapacity);
 			}
 		}
 	}
@@ -998,62 +936,60 @@ int clipFaceAgainstHull(const float4& separatingNormal, const b3ConvexPolyhedron
 	return numContactsOut;
 }
 
+static int clipHullAgainstHull(const float4& separatingNormal,
+							   const b3ConvexPolyhedronData& hullA, const b3ConvexPolyhedronData& hullB,
+							   const float4& posA, const b3Quaternion& ornA, const float4& posB, const b3Quaternion& ornB,
+							   float4* worldVertsB1, float4* worldVertsB2, int capacityWorldVerts,
+							   const float minDist, float maxDist,
+							   const b3AlignedObjectArray<float4>& verticesA, const b3AlignedObjectArray<b3GpuFace>& facesA, const b3AlignedObjectArray<int>& indicesA,
+							   const b3AlignedObjectArray<float4>& verticesB, const b3AlignedObjectArray<b3GpuFace>& facesB, const b3AlignedObjectArray<int>& indicesB,
 
-
-static int	clipHullAgainstHull(const float4& separatingNormal, 
-	const b3ConvexPolyhedronData& hullA, const b3ConvexPolyhedronData& hullB, 
-	const float4& posA, const b3Quaternion& ornA,const float4& posB, const b3Quaternion& ornB, 
-	float4* worldVertsB1, float4* worldVertsB2, int capacityWorldVerts,
-	const float minDist, float maxDist,
-	const b3AlignedObjectArray<float4>& verticesA,	const b3AlignedObjectArray<b3GpuFace>& facesA,	const b3AlignedObjectArray<int>& indicesA,
-	const b3AlignedObjectArray<float4>& verticesB,	const b3AlignedObjectArray<b3GpuFace>& facesB,	const b3AlignedObjectArray<int>& indicesB,
-
-	float4*	contactsOut,
-	int contactCapacity)
+							   float4* contactsOut,
+							   int contactCapacity)
 {
 	int numContactsOut = 0;
-	int numWorldVertsB1= 0;
-	
+	int numWorldVertsB1 = 0;
+
 	B3_PROFILE("clipHullAgainstHull");
 
-//	float curMaxDist=maxDist;
-	int closestFaceB=-1;
+	//	float curMaxDist=maxDist;
+	int closestFaceB = -1;
 	float dmax = -FLT_MAX;
 
 	{
 		//B3_PROFILE("closestFaceB");
-		if (hullB.m_numFaces!=1)
+		if (hullB.m_numFaces != 1)
 		{
 			//printf("wtf\n");
 		}
 		static bool once = true;
 		//printf("separatingNormal=%f,%f,%f\n",separatingNormal.x,separatingNormal.y,separatingNormal.z);
-		
-		for(int face=0;face<hullB.m_numFaces;face++)
+
+		for (int face = 0; face < hullB.m_numFaces; face++)
 		{
 #ifdef BT_DEBUG_SAT_FACE
 			if (once)
-				printf("face %d\n",face);
-			const b3GpuFace* faceB = &facesB[hullB.m_faceOffset+face];
+				printf("face %d\n", face);
+			const b3GpuFace* faceB = &facesB[hullB.m_faceOffset + face];
 			if (once)
 			{
-				for (int i=0;i<faceB->m_numIndices;i++)
+				for (int i = 0; i < faceB->m_numIndices; i++)
 				{
-					float4 vert = verticesB[hullB.m_vertexOffset+indicesB[faceB->m_indexOffset+i]];
-					printf("vert[%d] = %f,%f,%f\n",i,vert.x,vert.y,vert.z);
+					float4 vert = verticesB[hullB.m_vertexOffset + indicesB[faceB->m_indexOffset + i]];
+					printf("vert[%d] = %f,%f,%f\n", i, vert.x, vert.y, vert.z);
 				}
 			}
-#endif //BT_DEBUG_SAT_FACE
-			//if (facesB[hullB.m_faceOffset+face].m_numIndices>2)
+#endif  //BT_DEBUG_SAT_FACE \
+	//if (facesB[hullB.m_faceOffset+face].m_numIndices>2)
 			{
-				const float4 Normal = b3MakeVector3(facesB[hullB.m_faceOffset+face].m_plane.x, 
-					facesB[hullB.m_faceOffset+face].m_plane.y, facesB[hullB.m_faceOffset+face].m_plane.z,0.f);
+				const float4 Normal = b3MakeVector3(facesB[hullB.m_faceOffset + face].m_plane.x,
+													facesB[hullB.m_faceOffset + face].m_plane.y, facesB[hullB.m_faceOffset + face].m_plane.z, 0.f);
 				const float4 WorldNormal = b3QuatRotate(ornB, Normal);
 #ifdef BT_DEBUG_SAT_FACE
 				if (once)
-					printf("faceNormal = %f,%f,%f\n",Normal.x,Normal.y,Normal.z);
+					printf("faceNormal = %f,%f,%f\n", Normal.x, Normal.y, Normal.z);
 #endif
-				float d = dot3F4(WorldNormal,separatingNormal);
+				float d = dot3F4(WorldNormal, separatingNormal);
 				if (d > dmax)
 				{
 					dmax = d;
@@ -1064,184 +1000,176 @@ static int	clipHullAgainstHull(const float4& separatingNormal,
 		once = false;
 	}
 
-	
-	b3Assert(closestFaceB>=0);
+	b3Assert(closestFaceB >= 0);
 	{
 		//B3_PROFILE("worldVertsB1");
-		const b3GpuFace& polyB = facesB[hullB.m_faceOffset+closestFaceB];
+		const b3GpuFace& polyB = facesB[hullB.m_faceOffset + closestFaceB];
 		const int numVertices = polyB.m_numIndices;
-		for(int e0=0;e0<numVertices;e0++)
+		for (int e0 = 0; e0 < numVertices; e0++)
 		{
-			const float4& b = verticesB[hullB.m_vertexOffset+indicesB[polyB.m_indexOffset+e0]];
-			worldVertsB1[numWorldVertsB1++] = transform(&b,&posB,&ornB);
+			const float4& b = verticesB[hullB.m_vertexOffset + indicesB[polyB.m_indexOffset + e0]];
+			worldVertsB1[numWorldVertsB1++] = transform(&b, &posB, &ornB);
 		}
 	}
 
-	if (closestFaceB>=0)
+	if (closestFaceB >= 0)
 	{
 		//B3_PROFILE("clipFaceAgainstHull");
-		numContactsOut = clipFaceAgainstHull((float4&)separatingNormal, &hullA, 
-				posA,ornA,
-				worldVertsB1,numWorldVertsB1,worldVertsB2,capacityWorldVerts, minDist, maxDist,
-				verticesA,				facesA,				indicesA,
-				contactsOut,contactCapacity);
+		numContactsOut = clipFaceAgainstHull((float4&)separatingNormal, &hullA,
+											 posA, ornA,
+											 worldVertsB1, numWorldVertsB1, worldVertsB2, capacityWorldVerts, minDist, maxDist,
+											 verticesA, facesA, indicesA,
+											 contactsOut, contactCapacity);
 	}
 
 	return numContactsOut;
 }
 
-
-
-
-
-
-#define PARALLEL_SUM(v, n) for(int j=1; j<n; j++) v[0] += v[j];
-#define PARALLEL_DO(execution, n) for(int ie=0; ie<n; ie++){execution;}
-#define REDUCE_MAX(v, n) {int i=0;\
-for(int offset=0; offset<n; offset++) v[i] = (v[i].y > v[i+offset].y)? v[i]: v[i+offset]; }
-#define REDUCE_MIN(v, n) {int i=0;\
-for(int offset=0; offset<n; offset++) v[i] = (v[i].y < v[i+offset].y)? v[i]: v[i+offset]; }
+#define PARALLEL_SUM(v, n) \
+	for (int j = 1; j < n; j++) v[0] += v[j];
+#define PARALLEL_DO(execution, n)  \
+	for (int ie = 0; ie < n; ie++) \
+	{                              \
+		execution;                 \
+	}
+#define REDUCE_MAX(v, n)                                                                                     \
+	{                                                                                                        \
+		int i = 0;                                                                                           \
+		for (int offset = 0; offset < n; offset++) v[i] = (v[i].y > v[i + offset].y) ? v[i] : v[i + offset]; \
+	}
+#define REDUCE_MIN(v, n)                                                                                     \
+	{                                                                                                        \
+		int i = 0;                                                                                           \
+		for (int offset = 0; offset < n; offset++) v[i] = (v[i].y < v[i + offset].y) ? v[i] : v[i + offset]; \
+	}
 
 int extractManifold(const float4* p, int nPoints, const float4& nearNormal, b3Int4* contactIdx)
 {
-	if( nPoints == 0 )
-        return 0;
-    
-    if (nPoints <=4)
-        return nPoints;
-    
-    
-    if (nPoints >64)
-        nPoints = 64;
-    
-	float4 center = make_float4(0,0,0,0);
+	if (nPoints == 0)
+		return 0;
+
+	if (nPoints <= 4)
+		return nPoints;
+
+	if (nPoints > 64)
+		nPoints = 64;
+
+	float4 center = make_float4(0, 0, 0, 0);
 	{
-		
-		for (int i=0;i<nPoints;i++)
+		for (int i = 0; i < nPoints; i++)
 			center += p[i];
 		center /= (float)nPoints;
 	}
-    
-	
-    
+
 	//	sample 4 directions
-    
-    float4 aVector = p[0] - center;
-    float4 u = cross3( nearNormal, aVector );
-    float4 v = cross3( nearNormal, u );
-    u = normalize3( u );
-    v = normalize3( v );
-    
-    
-    //keep point with deepest penetration
-    float minW= FLT_MAX;
-    
-    int minIndex=-1;
-    
-    float4 maxDots;
-    maxDots.x = FLT_MIN;
-    maxDots.y = FLT_MIN;
-    maxDots.z = FLT_MIN;
-    maxDots.w = FLT_MIN;
-    
-    //	idx, distance
-    for(int ie = 0; ie<nPoints; ie++ )
-    {
-        if (p[ie].w<minW)
-        {
-            minW = p[ie].w;
-            minIndex=ie;
-        }
-        float f;
-        float4 r = p[ie]-center;
-        f = dot3F4( u, r );
-        if (f<maxDots.x)
-        {
-            maxDots.x = f;
-            contactIdx[0].x = ie;
-        }
-        
-        f = dot3F4( -u, r );
-        if (f<maxDots.y)
-        {
-            maxDots.y = f;
-            contactIdx[0].y = ie;
-        }
-        
-        
-        f = dot3F4( v, r );
-        if (f<maxDots.z)
-        {
-            maxDots.z = f;
-            contactIdx[0].z = ie;
-        }
-        
-        f = dot3F4( -v, r );
-        if (f<maxDots.w)
-        {
-            maxDots.w = f;
-            contactIdx[0].w = ie;
-        }
-        
-    }
-    
-    if (contactIdx[0].x != minIndex && contactIdx[0].y != minIndex && contactIdx[0].z != minIndex && contactIdx[0].w != minIndex)
-    {
-        //replace the first contact with minimum (todo: replace contact with least penetration)
-        contactIdx[0].x = minIndex;
-    }
-    
-    return 4;
-    
+
+	float4 aVector = p[0] - center;
+	float4 u = cross3(nearNormal, aVector);
+	float4 v = cross3(nearNormal, u);
+	u = normalize3(u);
+	v = normalize3(v);
+
+	//keep point with deepest penetration
+	float minW = FLT_MAX;
+
+	int minIndex = -1;
+
+	float4 maxDots;
+	maxDots.x = FLT_MIN;
+	maxDots.y = FLT_MIN;
+	maxDots.z = FLT_MIN;
+	maxDots.w = FLT_MIN;
+
+	//	idx, distance
+	for (int ie = 0; ie < nPoints; ie++)
+	{
+		if (p[ie].w < minW)
+		{
+			minW = p[ie].w;
+			minIndex = ie;
+		}
+		float f;
+		float4 r = p[ie] - center;
+		f = dot3F4(u, r);
+		if (f < maxDots.x)
+		{
+			maxDots.x = f;
+			contactIdx[0].x = ie;
+		}
+
+		f = dot3F4(-u, r);
+		if (f < maxDots.y)
+		{
+			maxDots.y = f;
+			contactIdx[0].y = ie;
+		}
+
+		f = dot3F4(v, r);
+		if (f < maxDots.z)
+		{
+			maxDots.z = f;
+			contactIdx[0].z = ie;
+		}
+
+		f = dot3F4(-v, r);
+		if (f < maxDots.w)
+		{
+			maxDots.w = f;
+			contactIdx[0].w = ie;
+		}
+	}
+
+	if (contactIdx[0].x != minIndex && contactIdx[0].y != minIndex && contactIdx[0].z != minIndex && contactIdx[0].w != minIndex)
+	{
+		//replace the first contact with minimum (todo: replace contact with least penetration)
+		contactIdx[0].x = minIndex;
+	}
+
+	return 4;
 }
 
-
-
-
 int clipHullHullSingle(
-			int bodyIndexA, int bodyIndexB,
-										 const float4& posA,
-										 const b3Quaternion& ornA,
-										 const float4& posB,
-										 const b3Quaternion& ornB,
+	int bodyIndexA, int bodyIndexB,
+	const float4& posA,
+	const b3Quaternion& ornA,
+	const float4& posB,
+	const b3Quaternion& ornB,
 
-			int collidableIndexA, int collidableIndexB,
+	int collidableIndexA, int collidableIndexB,
 
-			const b3AlignedObjectArray<b3RigidBodyData>* bodyBuf, 
-			b3AlignedObjectArray<b3Contact4>* globalContactOut, 
-			int& nContacts,
-			
-			const b3AlignedObjectArray<b3ConvexPolyhedronData>& hostConvexDataA,
-			const b3AlignedObjectArray<b3ConvexPolyhedronData>& hostConvexDataB,
-	
-			const b3AlignedObjectArray<b3Vector3>& verticesA, 
-			const b3AlignedObjectArray<b3Vector3>& uniqueEdgesA, 
-			const b3AlignedObjectArray<b3GpuFace>& facesA,
-			const b3AlignedObjectArray<int>& indicesA,
-	
-			const b3AlignedObjectArray<b3Vector3>& verticesB,
-			const b3AlignedObjectArray<b3Vector3>& uniqueEdgesB,
-			const b3AlignedObjectArray<b3GpuFace>& facesB,
-			const b3AlignedObjectArray<int>& indicesB,
+	const b3AlignedObjectArray<b3RigidBodyData>* bodyBuf,
+	b3AlignedObjectArray<b3Contact4>* globalContactOut,
+	int& nContacts,
 
-			const b3AlignedObjectArray<b3Collidable>& hostCollidablesA,
-			const b3AlignedObjectArray<b3Collidable>& hostCollidablesB,
-			const b3Vector3& sepNormalWorldSpace,
-			int maxContactCapacity			)
+	const b3AlignedObjectArray<b3ConvexPolyhedronData>& hostConvexDataA,
+	const b3AlignedObjectArray<b3ConvexPolyhedronData>& hostConvexDataB,
+
+	const b3AlignedObjectArray<b3Vector3>& verticesA,
+	const b3AlignedObjectArray<b3Vector3>& uniqueEdgesA,
+	const b3AlignedObjectArray<b3GpuFace>& facesA,
+	const b3AlignedObjectArray<int>& indicesA,
+
+	const b3AlignedObjectArray<b3Vector3>& verticesB,
+	const b3AlignedObjectArray<b3Vector3>& uniqueEdgesB,
+	const b3AlignedObjectArray<b3GpuFace>& facesB,
+	const b3AlignedObjectArray<int>& indicesB,
+
+	const b3AlignedObjectArray<b3Collidable>& hostCollidablesA,
+	const b3AlignedObjectArray<b3Collidable>& hostCollidablesB,
+	const b3Vector3& sepNormalWorldSpace,
+	int maxContactCapacity)
 {
 	int contactIndex = -1;
 	b3ConvexPolyhedronData hullA, hullB;
-    
-    b3Collidable colA = hostCollidablesA[collidableIndexA];
-    hullA = hostConvexDataA[colA.m_shapeIndex];
-    //printf("numvertsA = %d\n",hullA.m_numVertices);
-    
-    
-    b3Collidable colB = hostCollidablesB[collidableIndexB];
-    hullB = hostConvexDataB[colB.m_shapeIndex];
-    //printf("numvertsB = %d\n",hullB.m_numVertices);
-    
-	
+
+	b3Collidable colA = hostCollidablesA[collidableIndexA];
+	hullA = hostConvexDataA[colA.m_shapeIndex];
+	//printf("numvertsA = %d\n",hullA.m_numVertices);
+
+	b3Collidable colB = hostCollidablesB[collidableIndexB];
+	hullB = hostConvexDataB[colB.m_shapeIndex];
+	//printf("numvertsB = %d\n",hullB.m_numVertices);
+
 	float4 contactsOut[MAX_VERTS];
 	int localContactCapacity = MAX_VERTS;
 
@@ -1249,133 +1177,125 @@ int clipHullHullSingle(
 	b3Assert(_finite(bodyBuf->at(bodyIndexA).m_pos.x));
 	b3Assert(_finite(bodyBuf->at(bodyIndexB).m_pos.x));
 #endif
-	
-	
+
 	{
-		
 		float4 worldVertsB1[MAX_VERTS];
 		float4 worldVertsB2[MAX_VERTS];
 		int capacityWorldVerts = MAX_VERTS;
 
-		float4 hostNormal = make_float4(sepNormalWorldSpace.x,sepNormalWorldSpace.y,sepNormalWorldSpace.z,0.f);
+		float4 hostNormal = make_float4(sepNormalWorldSpace.x, sepNormalWorldSpace.y, sepNormalWorldSpace.z, 0.f);
 		int shapeA = hostCollidablesA[collidableIndexA].m_shapeIndex;
 		int shapeB = hostCollidablesB[collidableIndexB].m_shapeIndex;
 
 		b3Scalar minDist = -1;
 		b3Scalar maxDist = 0.;
 
-		        
-
-		b3Transform trA,trB;
+		b3Transform trA, trB;
 		{
-		//B3_PROFILE("transform computation");
-		//trA.setIdentity();
-		trA.setOrigin(b3MakeVector3(posA.x,posA.y,posA.z));
-		trA.setRotation(b3Quaternion(ornA.x,ornA.y,ornA.z,ornA.w));
-				
-		//trB.setIdentity();
-		trB.setOrigin(b3MakeVector3(posB.x,posB.y,posB.z));
-		trB.setRotation(b3Quaternion(ornB.x,ornB.y,ornB.z,ornB.w));
+			//B3_PROFILE("transform computation");
+			//trA.setIdentity();
+			trA.setOrigin(b3MakeVector3(posA.x, posA.y, posA.z));
+			trA.setRotation(b3Quaternion(ornA.x, ornA.y, ornA.z, ornA.w));
+
+			//trB.setIdentity();
+			trB.setOrigin(b3MakeVector3(posB.x, posB.y, posB.z));
+			trB.setRotation(b3Quaternion(ornB.x, ornB.y, ornB.z, ornB.w));
 		}
 
 		b3Quaternion trAorn = trA.getRotation();
-        b3Quaternion trBorn = trB.getRotation();
-        
-		int numContactsOut = clipHullAgainstHull(hostNormal, 
-						hostConvexDataA.at(shapeA), 
-						hostConvexDataB.at(shapeB),
-								(float4&)trA.getOrigin(), (b3Quaternion&)trAorn,
-								(float4&)trB.getOrigin(), (b3Quaternion&)trBorn,
-								worldVertsB1,worldVertsB2,capacityWorldVerts,
-								minDist, maxDist,
-								verticesA,	facesA,indicesA,
-								verticesB,	facesB,indicesB,
-								
-								contactsOut,localContactCapacity);
+		b3Quaternion trBorn = trB.getRotation();
 
-		if (numContactsOut>0)
+		int numContactsOut = clipHullAgainstHull(hostNormal,
+												 hostConvexDataA.at(shapeA),
+												 hostConvexDataB.at(shapeB),
+												 (float4&)trA.getOrigin(), (b3Quaternion&)trAorn,
+												 (float4&)trB.getOrigin(), (b3Quaternion&)trBorn,
+												 worldVertsB1, worldVertsB2, capacityWorldVerts,
+												 minDist, maxDist,
+												 verticesA, facesA, indicesA,
+												 verticesB, facesB, indicesB,
+
+												 contactsOut, localContactCapacity);
+
+		if (numContactsOut > 0)
 		{
 			B3_PROFILE("overlap");
 
 			float4 normalOnSurfaceB = (float4&)hostNormal;
-			
+
 			b3Int4 contactIdx;
 			contactIdx.x = 0;
 			contactIdx.y = 1;
 			contactIdx.z = 2;
 			contactIdx.w = 3;
-			
+
 			int numPoints = 0;
-					
+
 			{
-			//	B3_PROFILE("extractManifold");
-				numPoints = extractManifold(contactsOut, numContactsOut, normalOnSurfaceB,  &contactIdx);
+				//	B3_PROFILE("extractManifold");
+				numPoints = extractManifold(contactsOut, numContactsOut, normalOnSurfaceB, &contactIdx);
 			}
-					
+
 			b3Assert(numPoints);
-					
-			if (nContacts<maxContactCapacity)
+
+			if (nContacts < maxContactCapacity)
 			{
 				contactIndex = nContacts;
 				globalContactOut->expand();
 				b3Contact4& contact = globalContactOut->at(nContacts);
-				contact.m_batchIdx = 0;//i;
-				contact.m_bodyAPtrAndSignBit = (bodyBuf->at(bodyIndexA).m_invMass==0)? -bodyIndexA:bodyIndexA;
-				contact.m_bodyBPtrAndSignBit = (bodyBuf->at(bodyIndexB).m_invMass==0)? -bodyIndexB:bodyIndexB;
+				contact.m_batchIdx = 0;  //i;
+				contact.m_bodyAPtrAndSignBit = (bodyBuf->at(bodyIndexA).m_invMass == 0) ? -bodyIndexA : bodyIndexA;
+				contact.m_bodyBPtrAndSignBit = (bodyBuf->at(bodyIndexB).m_invMass == 0) ? -bodyIndexB : bodyIndexB;
 
 				contact.m_frictionCoeffCmp = 45874;
 				contact.m_restituitionCoeffCmp = 0;
-					
-	//			float distance = 0.f;
-				for (int p=0;p<numPoints;p++)
+
+				//			float distance = 0.f;
+				for (int p = 0; p < numPoints; p++)
 				{
-					contact.m_worldPosB[p] = contactsOut[contactIdx.s[p]];//check if it is actually on B
-					contact.m_worldNormalOnB = normalOnSurfaceB; 
+					contact.m_worldPosB[p] = contactsOut[contactIdx.s[p]];  //check if it is actually on B
+					contact.m_worldNormalOnB = normalOnSurfaceB;
 				}
 				//printf("bodyIndexA %d,bodyIndexB %d,normal=%f,%f,%f numPoints %d\n",bodyIndexA,bodyIndexB,normalOnSurfaceB.x,normalOnSurfaceB.y,normalOnSurfaceB.z,numPoints);
 				contact.m_worldNormalOnB.w = (b3Scalar)numPoints;
 				nContacts++;
-			} else
+			}
+			else
 			{
-				b3Error("Error: exceeding contact capacity (%d/%d)\n", nContacts,maxContactCapacity);
+				b3Error("Error: exceeding contact capacity (%d/%d)\n", nContacts, maxContactCapacity);
 			}
 		}
 	}
 	return contactIndex;
 }
 
-	
-
-
-
 void computeContactPlaneConvex(int pairIndex,
-																int bodyIndexA, int bodyIndexB, 
-																int collidableIndexA, int collidableIndexB, 
-																const b3RigidBodyData* rigidBodies, 
-																const b3Collidable* collidables,
-																const b3ConvexPolyhedronData* convexShapes,
-																const b3Vector3* convexVertices,
-																const int* convexIndices,
-																const b3GpuFace* faces,
-																b3Contact4* globalContactsOut,
-																int& nGlobalContactsOut,
-																int maxContactCapacity)
+							   int bodyIndexA, int bodyIndexB,
+							   int collidableIndexA, int collidableIndexB,
+							   const b3RigidBodyData* rigidBodies,
+							   const b3Collidable* collidables,
+							   const b3ConvexPolyhedronData* convexShapes,
+							   const b3Vector3* convexVertices,
+							   const int* convexIndices,
+							   const b3GpuFace* faces,
+							   b3Contact4* globalContactsOut,
+							   int& nGlobalContactsOut,
+							   int maxContactCapacity)
 {
-
-		int shapeIndex = collidables[collidableIndexB].m_shapeIndex;
+	int shapeIndex = collidables[collidableIndexB].m_shapeIndex;
 	const b3ConvexPolyhedronData* hullB = &convexShapes[shapeIndex];
-	
+
 	b3Vector3 posB = rigidBodies[bodyIndexB].m_pos;
 	b3Quaternion ornB = rigidBodies[bodyIndexB].m_quat;
 	b3Vector3 posA = rigidBodies[bodyIndexA].m_pos;
 	b3Quaternion ornA = rigidBodies[bodyIndexA].m_quat;
 
-//	int numContactsOut = 0;
-//	int numWorldVertsB1= 0;
+	//	int numContactsOut = 0;
+	//	int numWorldVertsB1= 0;
 
 	b3Vector3 planeEq = faces[collidables[collidableIndexA].m_shapeIndex].m_plane;
-	b3Vector3 planeNormal=b3MakeVector3(planeEq.x,planeEq.y,planeEq.z);
-	b3Vector3 planeNormalWorld = b3QuatRotate(ornA,planeNormal);
+	b3Vector3 planeNormal = b3MakeVector3(planeEq.x, planeEq.y, planeEq.z);
+	b3Vector3 planeNormalWorld = b3QuatRotate(ornA, planeNormal);
 	float planeConstant = planeEq.w;
 	b3Transform convexWorldTransform;
 	convexWorldTransform.setIdentity();
@@ -1387,13 +1307,13 @@ void computeContactPlaneConvex(int pairIndex,
 	planeTransform.setRotation(ornA);
 
 	b3Transform planeInConvex;
-	planeInConvex= convexWorldTransform.inverse() * planeTransform;
+	planeInConvex = convexWorldTransform.inverse() * planeTransform;
 	b3Transform convexInPlane;
 	convexInPlane = planeTransform.inverse() * convexWorldTransform;
-	
-	b3Vector3 planeNormalInConvex = planeInConvex.getBasis()*-planeNormal;
+
+	b3Vector3 planeNormalInConvex = planeInConvex.getBasis() * -planeNormal;
 	float maxDot = -1e30;
-	int hitVertex=-1;
+	int hitVertex = -1;
 	b3Vector3 hitVtx;
 
 #define MAX_PLANE_CONVEX_POINTS 64
@@ -1406,54 +1326,52 @@ void computeContactPlaneConvex(int pairIndex,
 	contactIdx.s[1] = 1;
 	contactIdx.s[2] = 2;
 	contactIdx.s[3] = 3;
-	
-	for (int i=0;i<hullB->m_numVertices;i++)
+
+	for (int i = 0; i < hullB->m_numVertices; i++)
 	{
-		b3Vector3 vtx = convexVertices[hullB->m_vertexOffset+i];
+		b3Vector3 vtx = convexVertices[hullB->m_vertexOffset + i];
 		float curDot = vtx.dot(planeNormalInConvex);
 
-
-		if (curDot>maxDot)
+		if (curDot > maxDot)
 		{
-			hitVertex=i;
-			maxDot=curDot;
+			hitVertex = i;
+			maxDot = curDot;
 			hitVtx = vtx;
 			//make sure the deepest points is always included
-			if (numPoints==MAX_PLANE_CONVEX_POINTS)
+			if (numPoints == MAX_PLANE_CONVEX_POINTS)
 				numPoints--;
 		}
 
-		if (numPoints<MAX_PLANE_CONVEX_POINTS)
+		if (numPoints < MAX_PLANE_CONVEX_POINTS)
 		{
-			b3Vector3 vtxWorld = convexWorldTransform*vtx;
-			b3Vector3 vtxInPlane = planeTransform.inverse()*vtxWorld;
-			float dist = planeNormal.dot(vtxInPlane)-planeConstant;
-			if (dist<0.f)
+			b3Vector3 vtxWorld = convexWorldTransform * vtx;
+			b3Vector3 vtxInPlane = planeTransform.inverse() * vtxWorld;
+			float dist = planeNormal.dot(vtxInPlane) - planeConstant;
+			if (dist < 0.f)
 			{
 				vtxWorld.w = dist;
 				contactPoints[numPoints] = vtxWorld;
 				numPoints++;
 			}
 		}
-
 	}
 
-	int numReducedPoints  = 0;
+	int numReducedPoints = 0;
 
 	numReducedPoints = numPoints;
-	
-	if (numPoints>4)
+
+	if (numPoints > 4)
 	{
-		numReducedPoints = extractManifoldSequentialGlobal( contactPoints, numPoints, planeNormalInConvex, &contactIdx);
+		numReducedPoints = extractManifoldSequentialGlobal(contactPoints, numPoints, planeNormalInConvex, &contactIdx);
 	}
 	int dstIdx;
-//    dstIdx = nGlobalContactsOut++;//AppendInc( nGlobalContactsOut, dstIdx );
-		
-	if (numReducedPoints>0)
+	//    dstIdx = nGlobalContactsOut++;//AppendInc( nGlobalContactsOut, dstIdx );
+
+	if (numReducedPoints > 0)
 	{
 		if (nGlobalContactsOut < maxContactCapacity)
 		{
-			dstIdx=nGlobalContactsOut;
+			dstIdx = nGlobalContactsOut;
 			nGlobalContactsOut++;
 
 			b3Contact4* c = &globalContactsOut[dstIdx];
@@ -1462,38 +1380,33 @@ void computeContactPlaneConvex(int pairIndex,
 			c->setRestituitionCoeff(0.f);
 
 			c->m_batchIdx = pairIndex;
-			c->m_bodyAPtrAndSignBit = rigidBodies[bodyIndexA].m_invMass==0?-bodyIndexA:bodyIndexA;
-			c->m_bodyBPtrAndSignBit = rigidBodies[bodyIndexB].m_invMass==0?-bodyIndexB:bodyIndexB;
-			for (int i=0;i<numReducedPoints;i++)
+			c->m_bodyAPtrAndSignBit = rigidBodies[bodyIndexA].m_invMass == 0 ? -bodyIndexA : bodyIndexA;
+			c->m_bodyBPtrAndSignBit = rigidBodies[bodyIndexB].m_invMass == 0 ? -bodyIndexB : bodyIndexB;
+			for (int i = 0; i < numReducedPoints; i++)
 			{
 				b3Vector3 pOnB1 = contactPoints[contactIdx.s[i]];
 				c->m_worldPosB[i] = pOnB1;
 			}
 			c->m_worldNormalOnB.w = (b3Scalar)numReducedPoints;
-		}//if (dstIdx < numPairs)
-	}	
-		
+		}  //if (dstIdx < numPairs)
+	}
 
-
-//	printf("computeContactPlaneConvex\n");
+	//	printf("computeContactPlaneConvex\n");
 }
 
-
-
-B3_FORCE_INLINE b3Vector3	MyUnQuantize(const unsigned short* vecIn, const b3Vector3& quantization, const b3Vector3& bvhAabbMin)
-	{
-			b3Vector3	vecOut;
-			vecOut.setValue(
-			(b3Scalar)(vecIn[0]) / (quantization.x),
-			(b3Scalar)(vecIn[1]) / (quantization.y),
-			(b3Scalar)(vecIn[2]) / (quantization.z));
-			vecOut += bvhAabbMin;
-			return vecOut;
-	}
+B3_FORCE_INLINE b3Vector3 MyUnQuantize(const unsigned short* vecIn, const b3Vector3& quantization, const b3Vector3& bvhAabbMin)
+{
+	b3Vector3 vecOut;
+	vecOut.setValue(
+		(b3Scalar)(vecIn[0]) / (quantization.x),
+		(b3Scalar)(vecIn[1]) / (quantization.y),
+		(b3Scalar)(vecIn[2]) / (quantization.z));
+	vecOut += bvhAabbMin;
+	return vecOut;
+}
 
 void traverseTreeTree()
 {
-
 }
 
 #include "Bullet3Common/shared/b3Mat3x3.h"
@@ -1503,44 +1416,40 @@ int maxNumAabbChecks = 0;
 int maxDepth = 0;
 
 // work-in-progress
-__kernel void   findCompoundPairsKernel( 
+__kernel void findCompoundPairsKernel(
 	int pairIndex,
 	int bodyIndexA,
 	int bodyIndexB,
 	int collidableIndexA,
 	int collidableIndexB,
-	__global const b3RigidBodyData* rigidBodies, 
+	__global const b3RigidBodyData* rigidBodies,
 	__global const b3Collidable* collidables,
-	__global const b3ConvexPolyhedronData* convexShapes, 
+	__global const b3ConvexPolyhedronData* convexShapes,
 	__global const b3AlignedObjectArray<b3Float4>& vertices,
 	__global const b3AlignedObjectArray<b3Aabb>& aabbsWorldSpace,
 	__global const b3AlignedObjectArray<b3Aabb>& aabbsLocalSpace,
 	__global const b3GpuChildShape* gpuChildShapes,
 	__global b3Int4* gpuCompoundPairsOut,
-	__global  int* numCompoundPairsOut,
+	__global int* numCompoundPairsOut,
 	int maxNumCompoundPairsCapacity,
-	b3AlignedObjectArray<b3QuantizedBvhNode>&	treeNodesCPU,
-	b3AlignedObjectArray<b3BvhSubtreeInfo>&	subTreesCPU,
-	b3AlignedObjectArray<b3BvhInfo>&	bvhInfoCPU
-	)
+	b3AlignedObjectArray<b3QuantizedBvhNode>& treeNodesCPU,
+	b3AlignedObjectArray<b3BvhSubtreeInfo>& subTreesCPU,
+	b3AlignedObjectArray<b3BvhInfo>& bvhInfoCPU)
 {
-	numAabbChecks=0;
-	maxNumAabbChecks=0;
-//	int i = pairIndex;
+	numAabbChecks = 0;
+	maxNumAabbChecks = 0;
+	//	int i = pairIndex;
 	{
-		
-
 		int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;
 		int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;
 
-
 		//once the broadphase avoids static-static pairs, we can remove this test
-		if ((rigidBodies[bodyIndexA].m_invMass==0) &&(rigidBodies[bodyIndexB].m_invMass==0))
+		if ((rigidBodies[bodyIndexA].m_invMass == 0) && (rigidBodies[bodyIndexB].m_invMass == 0))
 		{
 			return;
 		}
 
-		if ((collidables[collidableIndexA].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS) &&(collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS))
+		if ((collidables[collidableIndexA].m_shapeType == SHAPE_COMPOUND_OF_CONVEX_HULLS) && (collidables[collidableIndexB].m_shapeType == SHAPE_COMPOUND_OF_CONVEX_HULLS))
 		{
 			int bvhA = collidables[collidableIndexA].m_compoundBvhIndex;
 			int bvhB = collidables[collidableIndexB].m_compoundBvhIndex;
@@ -1548,9 +1457,8 @@ __kernel void   findCompoundPairsKernel(
 			int subTreesOffsetA = bvhInfoCPU[bvhA].m_subTreeOffset;
 			int subTreesOffsetB = bvhInfoCPU[bvhB].m_subTreeOffset;
 
-
 			int numSubTreesB = bvhInfoCPU[bvhB].m_numSubTrees;
-			
+
 			float4 posA = rigidBodies[bodyIndexA].m_pos;
 			b3Quat ornA = rigidBodies[bodyIndexA].m_quat;
 
@@ -1567,41 +1475,37 @@ __kernel void   findCompoundPairsKernel(
 			transB.setOrigin(posB);
 			transB.setRotation(ornB);
 
-
-
-			for (int p=0;p<numSubTreesA;p++)
+			for (int p = 0; p < numSubTreesA; p++)
 			{
-				b3BvhSubtreeInfo subtreeA = subTreesCPU[subTreesOffsetA+p];
+				b3BvhSubtreeInfo subtreeA = subTreesCPU[subTreesOffsetA + p];
 				//bvhInfoCPU[bvhA].m_quantization
-				b3Vector3 treeAminLocal = MyUnQuantize(subtreeA.m_quantizedAabbMin,bvhInfoCPU[bvhA].m_quantization,bvhInfoCPU[bvhA].m_aabbMin);
-				b3Vector3 treeAmaxLocal = MyUnQuantize(subtreeA.m_quantizedAabbMax,bvhInfoCPU[bvhA].m_quantization,bvhInfoCPU[bvhA].m_aabbMin);
+				b3Vector3 treeAminLocal = MyUnQuantize(subtreeA.m_quantizedAabbMin, bvhInfoCPU[bvhA].m_quantization, bvhInfoCPU[bvhA].m_aabbMin);
+				b3Vector3 treeAmaxLocal = MyUnQuantize(subtreeA.m_quantizedAabbMax, bvhInfoCPU[bvhA].m_quantization, bvhInfoCPU[bvhA].m_aabbMin);
 
-				b3Vector3 aabbAMinOut,aabbAMaxOut;
-				float margin=0.f;
-				b3TransformAabb2(treeAminLocal,treeAmaxLocal, margin,transA.getOrigin(),transA.getRotation(),&aabbAMinOut,&aabbAMaxOut);
+				b3Vector3 aabbAMinOut, aabbAMaxOut;
+				float margin = 0.f;
+				b3TransformAabb2(treeAminLocal, treeAmaxLocal, margin, transA.getOrigin(), transA.getRotation(), &aabbAMinOut, &aabbAMaxOut);
 
-				for (int q=0;q<numSubTreesB;q++)
+				for (int q = 0; q < numSubTreesB; q++)
 				{
-					b3BvhSubtreeInfo subtreeB = subTreesCPU[subTreesOffsetB+q];
+					b3BvhSubtreeInfo subtreeB = subTreesCPU[subTreesOffsetB + q];
 
-					b3Vector3 treeBminLocal = MyUnQuantize(subtreeB.m_quantizedAabbMin,bvhInfoCPU[bvhB].m_quantization,bvhInfoCPU[bvhB].m_aabbMin);
-					b3Vector3 treeBmaxLocal = MyUnQuantize(subtreeB.m_quantizedAabbMax,bvhInfoCPU[bvhB].m_quantization,bvhInfoCPU[bvhB].m_aabbMin);
+					b3Vector3 treeBminLocal = MyUnQuantize(subtreeB.m_quantizedAabbMin, bvhInfoCPU[bvhB].m_quantization, bvhInfoCPU[bvhB].m_aabbMin);
+					b3Vector3 treeBmaxLocal = MyUnQuantize(subtreeB.m_quantizedAabbMax, bvhInfoCPU[bvhB].m_quantization, bvhInfoCPU[bvhB].m_aabbMin);
 
-					b3Vector3 aabbBMinOut,aabbBMaxOut;
-					float margin=0.f;
-					b3TransformAabb2(treeBminLocal,treeBmaxLocal, margin,transB.getOrigin(),transB.getRotation(),&aabbBMinOut,&aabbBMaxOut);
+					b3Vector3 aabbBMinOut, aabbBMaxOut;
+					float margin = 0.f;
+					b3TransformAabb2(treeBminLocal, treeBmaxLocal, margin, transB.getOrigin(), transB.getRotation(), &aabbBMinOut, &aabbBMaxOut);
 
-					
-					numAabbChecks=0;
-					bool aabbOverlap = b3TestAabbAgainstAabb(aabbAMinOut,aabbAMaxOut,aabbBMinOut,aabbBMaxOut);
+					numAabbChecks = 0;
+					bool aabbOverlap = b3TestAabbAgainstAabb(aabbAMinOut, aabbAMaxOut, aabbBMinOut, aabbBMaxOut);
 					if (aabbOverlap)
 					{
-						
-						int startNodeIndexA = subtreeA.m_rootNodeIndex+bvhInfoCPU[bvhA].m_nodeOffset;
-		//				int endNodeIndexA = startNodeIndexA+subtreeA.m_subtreeSize;
+						int startNodeIndexA = subtreeA.m_rootNodeIndex + bvhInfoCPU[bvhA].m_nodeOffset;
+						//				int endNodeIndexA = startNodeIndexA+subtreeA.m_subtreeSize;
 
-						int startNodeIndexB = subtreeB.m_rootNodeIndex+bvhInfoCPU[bvhB].m_nodeOffset;
-		//				int endNodeIndexB = startNodeIndexB+subtreeB.m_subtreeSize;
+						int startNodeIndexB = subtreeB.m_rootNodeIndex + bvhInfoCPU[bvhB].m_nodeOffset;
+						//				int endNodeIndexB = startNodeIndexB+subtreeB.m_subtreeSize;
 
 						b3AlignedObjectArray<b3Int2> nodeStack;
 						b3Int2 node0;
@@ -1610,33 +1514,33 @@ __kernel void   findCompoundPairsKernel(
 
 						int maxStackDepth = 1024;
 						nodeStack.resize(maxStackDepth);
-						int depth=0;
-						nodeStack[depth++]=node0;
+						int depth = 0;
+						nodeStack[depth++] = node0;
 
 						do
 						{
 							if (depth > maxDepth)
 							{
-								maxDepth=depth;
-								printf("maxDepth=%d\n",maxDepth);
+								maxDepth = depth;
+								printf("maxDepth=%d\n", maxDepth);
 							}
 							b3Int2 node = nodeStack[--depth];
-							
-							b3Vector3 aMinLocal = MyUnQuantize(treeNodesCPU[node.x].m_quantizedAabbMin,bvhInfoCPU[bvhA].m_quantization,bvhInfoCPU[bvhA].m_aabbMin);
-							b3Vector3 aMaxLocal = MyUnQuantize(treeNodesCPU[node.x].m_quantizedAabbMax,bvhInfoCPU[bvhA].m_quantization,bvhInfoCPU[bvhA].m_aabbMin);
 
-							b3Vector3 bMinLocal = MyUnQuantize(treeNodesCPU[node.y].m_quantizedAabbMin,bvhInfoCPU[bvhB].m_quantization,bvhInfoCPU[bvhB].m_aabbMin);
-							b3Vector3 bMaxLocal = MyUnQuantize(treeNodesCPU[node.y].m_quantizedAabbMax,bvhInfoCPU[bvhB].m_quantization,bvhInfoCPU[bvhB].m_aabbMin);
+							b3Vector3 aMinLocal = MyUnQuantize(treeNodesCPU[node.x].m_quantizedAabbMin, bvhInfoCPU[bvhA].m_quantization, bvhInfoCPU[bvhA].m_aabbMin);
+							b3Vector3 aMaxLocal = MyUnQuantize(treeNodesCPU[node.x].m_quantizedAabbMax, bvhInfoCPU[bvhA].m_quantization, bvhInfoCPU[bvhA].m_aabbMin);
 
-							float margin=0.f;
-							b3Vector3 aabbAMinOut,aabbAMaxOut;
-							b3TransformAabb2(aMinLocal,aMaxLocal, margin,transA.getOrigin(),transA.getRotation(),&aabbAMinOut,&aabbAMaxOut);
+							b3Vector3 bMinLocal = MyUnQuantize(treeNodesCPU[node.y].m_quantizedAabbMin, bvhInfoCPU[bvhB].m_quantization, bvhInfoCPU[bvhB].m_aabbMin);
+							b3Vector3 bMaxLocal = MyUnQuantize(treeNodesCPU[node.y].m_quantizedAabbMax, bvhInfoCPU[bvhB].m_quantization, bvhInfoCPU[bvhB].m_aabbMin);
 
-							b3Vector3 aabbBMinOut,aabbBMaxOut;
-							b3TransformAabb2(bMinLocal,bMaxLocal, margin,transB.getOrigin(),transB.getRotation(),&aabbBMinOut,&aabbBMaxOut);
+							float margin = 0.f;
+							b3Vector3 aabbAMinOut, aabbAMaxOut;
+							b3TransformAabb2(aMinLocal, aMaxLocal, margin, transA.getOrigin(), transA.getRotation(), &aabbAMinOut, &aabbAMaxOut);
+
+							b3Vector3 aabbBMinOut, aabbBMaxOut;
+							b3TransformAabb2(bMinLocal, bMaxLocal, margin, transB.getOrigin(), transB.getRotation(), &aabbBMinOut, &aabbBMaxOut);
 
 							numAabbChecks++;
-							bool nodeOverlap = b3TestAabbAgainstAabb(aabbAMinOut,aabbAMaxOut,aabbBMinOut,aabbBMaxOut);
+							bool nodeOverlap = b3TestAabbAgainstAabb(aabbAMinOut, aabbAMaxOut, aabbBMinOut, aabbBMaxOut);
 							if (nodeOverlap)
 							{
 								bool isLeafA = treeNodesCPU[node.x].isLeafNode();
@@ -1645,23 +1549,23 @@ __kernel void   findCompoundPairsKernel(
 								bool isInternalB = !isLeafB;
 
 								//fail, even though it might hit two leaf nodes
-								if (depth+4>maxStackDepth && !(isLeafA && isLeafB))
+								if (depth + 4 > maxStackDepth && !(isLeafA && isLeafB))
 								{
 									b3Error("Error: traversal exceeded maxStackDepth\n");
 									continue;
 								}
 
-								if(isInternalA)
+								if (isInternalA)
 								{
-									int nodeAleftChild = node.x+1;
-									bool isNodeALeftChildLeaf = treeNodesCPU[node.x+1].isLeafNode();
-									int nodeArightChild = isNodeALeftChildLeaf? node.x+2 : node.x+1 + treeNodesCPU[node.x+1].getEscapeIndex();
+									int nodeAleftChild = node.x + 1;
+									bool isNodeALeftChildLeaf = treeNodesCPU[node.x + 1].isLeafNode();
+									int nodeArightChild = isNodeALeftChildLeaf ? node.x + 2 : node.x + 1 + treeNodesCPU[node.x + 1].getEscapeIndex();
 
-									if(isInternalB)
-									{					
-										int nodeBleftChild = node.y+1;
-										bool isNodeBLeftChildLeaf = treeNodesCPU[node.y+1].isLeafNode();
-										int nodeBrightChild = isNodeBLeftChildLeaf? node.y+2 : node.y+1 + treeNodesCPU[node.y+1].getEscapeIndex();
+									if (isInternalB)
+									{
+										int nodeBleftChild = node.y + 1;
+										bool isNodeBLeftChildLeaf = treeNodesCPU[node.y + 1].isLeafNode();
+										int nodeBrightChild = isNodeBLeftChildLeaf ? node.y + 2 : node.y + 1 + treeNodesCPU[node.y + 1].getEscapeIndex();
 
 										nodeStack[depth++] = b3MakeInt2(nodeAleftChild, nodeBleftChild);
 										nodeStack[depth++] = b3MakeInt2(nodeArightChild, nodeBleftChild);
@@ -1670,90 +1574,83 @@ __kernel void   findCompoundPairsKernel(
 									}
 									else
 									{
-										nodeStack[depth++] = b3MakeInt2(nodeAleftChild,node.y);
-										nodeStack[depth++] = b3MakeInt2(nodeArightChild,node.y);
+										nodeStack[depth++] = b3MakeInt2(nodeAleftChild, node.y);
+										nodeStack[depth++] = b3MakeInt2(nodeArightChild, node.y);
 									}
 								}
 								else
 								{
-									if(isInternalB)
+									if (isInternalB)
 									{
-										int nodeBleftChild = node.y+1;
-										bool isNodeBLeftChildLeaf = treeNodesCPU[node.y+1].isLeafNode();
-										int nodeBrightChild = isNodeBLeftChildLeaf? node.y+2 : node.y+1 + treeNodesCPU[node.y+1].getEscapeIndex();
-										nodeStack[depth++] = b3MakeInt2(node.x,nodeBleftChild);
-										nodeStack[depth++] = b3MakeInt2(node.x,nodeBrightChild);
+										int nodeBleftChild = node.y + 1;
+										bool isNodeBLeftChildLeaf = treeNodesCPU[node.y + 1].isLeafNode();
+										int nodeBrightChild = isNodeBLeftChildLeaf ? node.y + 2 : node.y + 1 + treeNodesCPU[node.y + 1].getEscapeIndex();
+										nodeStack[depth++] = b3MakeInt2(node.x, nodeBleftChild);
+										nodeStack[depth++] = b3MakeInt2(node.x, nodeBrightChild);
 									}
 									else
 									{
 										int compoundPairIdx = b3AtomicInc(numCompoundPairsOut);
-										if (compoundPairIdx<maxNumCompoundPairsCapacity)
+										if (compoundPairIdx < maxNumCompoundPairsCapacity)
 										{
 											int childShapeIndexA = treeNodesCPU[node.x].getTriangleIndex();
 											int childShapeIndexB = treeNodesCPU[node.y].getTriangleIndex();
-											gpuCompoundPairsOut[compoundPairIdx]  = b3MakeInt4(bodyIndexA,bodyIndexB,childShapeIndexA,childShapeIndexB);
+											gpuCompoundPairsOut[compoundPairIdx] = b3MakeInt4(bodyIndexA, bodyIndexB, childShapeIndexA, childShapeIndexB);
 										}
 									}
 								}
 							}
 						} while (depth);
-						maxNumAabbChecks = b3Max(numAabbChecks,maxNumAabbChecks);
+						maxNumAabbChecks = b3Max(numAabbChecks, maxNumAabbChecks);
 					}
 				}
 			}
-			
+
 			return;
 		}
 
-		if ((collidables[collidableIndexA].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS) ||(collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS))
+		if ((collidables[collidableIndexA].m_shapeType == SHAPE_COMPOUND_OF_CONVEX_HULLS) || (collidables[collidableIndexB].m_shapeType == SHAPE_COMPOUND_OF_CONVEX_HULLS))
 		{
-
-			if (collidables[collidableIndexA].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS) 
+			if (collidables[collidableIndexA].m_shapeType == SHAPE_COMPOUND_OF_CONVEX_HULLS)
 			{
-
 				int numChildrenA = collidables[collidableIndexA].m_numChildShapes;
-				for (int c=0;c<numChildrenA;c++)
+				for (int c = 0; c < numChildrenA; c++)
 				{
-					int childShapeIndexA = collidables[collidableIndexA].m_shapeIndex+c;
+					int childShapeIndexA = collidables[collidableIndexA].m_shapeIndex + c;
 					int childColIndexA = gpuChildShapes[childShapeIndexA].m_shapeIndex;
 
 					float4 posA = rigidBodies[bodyIndexA].m_pos;
 					b3Quat ornA = rigidBodies[bodyIndexA].m_quat;
 					float4 childPosA = gpuChildShapes[childShapeIndexA].m_childPosition;
 					b3Quat childOrnA = gpuChildShapes[childShapeIndexA].m_childOrientation;
-					float4 newPosA = b3QuatRotate(ornA,childPosA)+posA;
-					b3Quat newOrnA = b3QuatMul(ornA,childOrnA);
-					
+					float4 newPosA = b3QuatRotate(ornA, childPosA) + posA;
+					b3Quat newOrnA = b3QuatMul(ornA, childOrnA);
 
-					
 					b3Aabb aabbA = aabbsLocalSpace[childColIndexA];
 
-					
 					b3Transform transA;
 					transA.setIdentity();
 					transA.setOrigin(newPosA);
 					transA.setRotation(newOrnA);
-					b3Scalar margin=0.0f;
+					b3Scalar margin = 0.0f;
 
-					b3Vector3 aabbAMinOut,aabbAMaxOut;
+					b3Vector3 aabbAMinOut, aabbAMaxOut;
 
-					b3TransformAabb2((const b3Float4&)aabbA.m_min,(const b3Float4&)aabbA.m_max, margin,transA.getOrigin(),transA.getRotation(),&aabbAMinOut,&aabbAMaxOut);
+					b3TransformAabb2((const b3Float4&)aabbA.m_min, (const b3Float4&)aabbA.m_max, margin, transA.getOrigin(), transA.getRotation(), &aabbAMinOut, &aabbAMaxOut);
 
-					if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)
+					if (collidables[collidableIndexB].m_shapeType == SHAPE_COMPOUND_OF_CONVEX_HULLS)
 					{
 						int numChildrenB = collidables[collidableIndexB].m_numChildShapes;
-						for (int b=0;b<numChildrenB;b++)
+						for (int b = 0; b < numChildrenB; b++)
 						{
-							int childShapeIndexB = collidables[collidableIndexB].m_shapeIndex+b;
+							int childShapeIndexB = collidables[collidableIndexB].m_shapeIndex + b;
 							int childColIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;
 							b3Quat ornB = rigidBodies[bodyIndexB].m_quat;
 							float4 posB = rigidBodies[bodyIndexB].m_pos;
 							float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;
 							b3Quat childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;
-							float4 newPosB = transform(&childPosB,&posB,&ornB);
-							b3Quat newOrnB = b3QuatMul(ornB,childOrnB);
-
-							
+							float4 newPosB = transform(&childPosB, &posB, &ornB);
+							b3Quat newOrnB = b3QuatMul(ornB, childOrnB);
 
 							b3Aabb aabbB = aabbsLocalSpace[childColIndexB];
 
@@ -1762,11 +1659,11 @@ __kernel void   findCompoundPairsKernel(
 							transB.setOrigin(newPosB);
 							transB.setRotation(newOrnB);
 
-							b3Vector3 aabbBMinOut,aabbBMaxOut;
-							b3TransformAabb2((const b3Float4&)aabbB.m_min,(const b3Float4&)aabbB.m_max, margin,transB.getOrigin(),transB.getRotation(),&aabbBMinOut,&aabbBMaxOut);
+							b3Vector3 aabbBMinOut, aabbBMaxOut;
+							b3TransformAabb2((const b3Float4&)aabbB.m_min, (const b3Float4&)aabbB.m_max, margin, transB.getOrigin(), transB.getRotation(), &aabbBMinOut, &aabbBMaxOut);
 
 							numAabbChecks++;
-							bool aabbOverlap = b3TestAabbAgainstAabb(aabbAMinOut,aabbAMaxOut,aabbBMinOut,aabbBMaxOut);
+							bool aabbOverlap = b3TestAabbAgainstAabb(aabbAMinOut, aabbAMaxOut, aabbBMinOut, aabbBMaxOut);
 							if (aabbOverlap)
 							{
 								/*
@@ -1784,22 +1681,22 @@ __kernel void   findCompoundPairsKernel(
 								float4 c1 = transform(&c1local,&posB,&ornB);
 								const float4 DeltaC2 = c0 - c1;
 								*/
-								{//
+								{  //
 									int compoundPairIdx = b3AtomicInc(numCompoundPairsOut);
-									if (compoundPairIdx<maxNumCompoundPairsCapacity)
+									if (compoundPairIdx < maxNumCompoundPairsCapacity)
 									{
-										gpuCompoundPairsOut[compoundPairIdx]  = b3MakeInt4(bodyIndexA,bodyIndexB,childShapeIndexA,childShapeIndexB);
+										gpuCompoundPairsOut[compoundPairIdx] = b3MakeInt4(bodyIndexA, bodyIndexB, childShapeIndexA, childShapeIndexB);
 									}
-								}//
-							}//fi(1)
-						} //for (int b=0
-					}//if (collidables[collidableIndexB].
-					else//if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)
+								}  //
+							}      //fi(1)
+						}          //for (int b=0
+					}              //if (collidables[collidableIndexB].
+					else           //if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)
 					{
 						if (1)
 						{
-						//	int numFacesA = convexShapes[shapeIndexA].m_numFaces;
-						//	float dmin = FLT_MAX;
+							//	int numFacesA = convexShapes[shapeIndexA].m_numFaces;
+							//	float dmin = FLT_MAX;
 							float4 posA = newPosA;
 							posA.w = 0.f;
 							float4 posB = rigidBodies[bodyIndexB].m_pos;
@@ -1811,45 +1708,43 @@ __kernel void   findCompoundPairsKernel(
 							float4 c1local = convexShapes[shapeIndexB].m_localCenter;
 							b3Quat ornB = rigidBodies[bodyIndexB].m_quat;
 							float4 c1;
-							c1 = transform(&c1local,&posB,&ornB);
-						//	const float4 DeltaC2 = c0 - c1;
+							c1 = transform(&c1local, &posB, &ornB);
+							//	const float4 DeltaC2 = c0 - c1;
 
 							{
 								int compoundPairIdx = b3AtomicInc(numCompoundPairsOut);
-								if (compoundPairIdx<maxNumCompoundPairsCapacity)
+								if (compoundPairIdx < maxNumCompoundPairsCapacity)
 								{
-									gpuCompoundPairsOut[compoundPairIdx] = b3MakeInt4(bodyIndexA,bodyIndexB,childShapeIndexA,-1);
-								}//if (compoundPairIdx<maxNumCompoundPairsCapacity)
-							}//
-						}//fi (1)
-					}//if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)
-				}//for (int b=0;b<numChildrenB;b++)	
+									gpuCompoundPairsOut[compoundPairIdx] = b3MakeInt4(bodyIndexA, bodyIndexB, childShapeIndexA, -1);
+								}  //if (compoundPairIdx<maxNumCompoundPairsCapacity)
+							}      //
+						}          //fi (1)
+					}              //if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)
+				}                  //for (int b=0;b<numChildrenB;b++)
 				return;
-			}//if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)
-			if ((collidables[collidableIndexA].m_shapeType!=SHAPE_CONCAVE_TRIMESH) 
-				&& (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS))
+			}  //if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)
+			if ((collidables[collidableIndexA].m_shapeType != SHAPE_CONCAVE_TRIMESH) && (collidables[collidableIndexB].m_shapeType == SHAPE_COMPOUND_OF_CONVEX_HULLS))
 			{
 				int numChildrenB = collidables[collidableIndexB].m_numChildShapes;
-				for (int b=0;b<numChildrenB;b++)
+				for (int b = 0; b < numChildrenB; b++)
 				{
-					int childShapeIndexB = collidables[collidableIndexB].m_shapeIndex+b;
+					int childShapeIndexB = collidables[collidableIndexB].m_shapeIndex + b;
 					int childColIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;
 					b3Quat ornB = rigidBodies[bodyIndexB].m_quat;
 					float4 posB = rigidBodies[bodyIndexB].m_pos;
 					float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;
 					b3Quat childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;
-					float4 newPosB = b3QuatRotate(ornB,childPosB)+posB;
-					b3Quat newOrnB = b3QuatMul(ornB,childOrnB);
+					float4 newPosB = b3QuatRotate(ornB, childPosB) + posB;
+					b3Quat newOrnB = b3QuatMul(ornB, childOrnB);
 
 					int shapeIndexB = collidables[childColIndexB].m_shapeIndex;
 
-
 					//////////////////////////////////////
 
 					if (1)
 					{
-					//	int numFacesA = convexShapes[shapeIndexA].m_numFaces;
-					//	float dmin = FLT_MAX;
+						//	int numFacesA = convexShapes[shapeIndexA].m_numFaces;
+						//	float dmin = FLT_MAX;
 						float4 posA = rigidBodies[bodyIndexA].m_pos;
 						posA.w = 0.f;
 						float4 posB = newPosB;
@@ -1859,99 +1754,96 @@ __kernel void   findCompoundPairsKernel(
 						float4 c0;
 						c0 = transform(&c0local, &posA, &ornA);
 						float4 c1local = convexShapes[shapeIndexB].m_localCenter;
-						b3Quat ornB =newOrnB;
+						b3Quat ornB = newOrnB;
 						float4 c1;
-						c1 = transform(&c1local,&posB,&ornB);
-					//	const float4 DeltaC2 = c0 - c1;
-						{//
+						c1 = transform(&c1local, &posB, &ornB);
+						//	const float4 DeltaC2 = c0 - c1;
+						{  //
 							int compoundPairIdx = b3AtomicInc(numCompoundPairsOut);
-							if (compoundPairIdx<maxNumCompoundPairsCapacity)
+							if (compoundPairIdx < maxNumCompoundPairsCapacity)
 							{
-								gpuCompoundPairsOut[compoundPairIdx] = b3MakeInt4(bodyIndexA,bodyIndexB,-1,childShapeIndexB);
-							}//fi (compoundPairIdx<maxNumCompoundPairsCapacity)
-						}//
-					}//fi (1)	
-				}//for (int b=0;b<numChildrenB;b++)
+								gpuCompoundPairsOut[compoundPairIdx] = b3MakeInt4(bodyIndexA, bodyIndexB, -1, childShapeIndexB);
+							}  //fi (compoundPairIdx<maxNumCompoundPairsCapacity)
+						}      //
+					}          //fi (1)
+				}              //for (int b=0;b<numChildrenB;b++)
 				return;
-			}//if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)
+			}  //if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)
 			return;
-		}//fi ((collidables[collidableIndexA].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS) ||(collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS))
-	}//i<numPairs
+		}  //fi ((collidables[collidableIndexA].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS) ||(collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS))
+	}      //i<numPairs
 }
 
-
-
-__kernel void   processCompoundPairsKernel( __global const b3Int4* gpuCompoundPairs,
-										__global const b3RigidBodyData* rigidBodies, 
-										__global const b3Collidable* collidables,
-										__global const b3ConvexPolyhedronData* convexShapes, 
-										__global const b3AlignedObjectArray<b3Float4>& vertices,
-										__global const b3AlignedObjectArray<b3Float4>& uniqueEdges,
-										__global const b3AlignedObjectArray<b3GpuFace>& faces,
-										__global const b3AlignedObjectArray<int>& indices,
-										__global b3Aabb* aabbs,
-										__global const b3GpuChildShape* gpuChildShapes,
-										__global b3AlignedObjectArray<b3Float4>& gpuCompoundSepNormalsOut,
-										__global b3AlignedObjectArray<int>& gpuHasCompoundSepNormalsOut,
-										int numCompoundPairs,
-										int i
-										)
+__kernel void processCompoundPairsKernel(__global const b3Int4* gpuCompoundPairs,
+										 __global const b3RigidBodyData* rigidBodies,
+										 __global const b3Collidable* collidables,
+										 __global const b3ConvexPolyhedronData* convexShapes,
+										 __global const b3AlignedObjectArray<b3Float4>& vertices,
+										 __global const b3AlignedObjectArray<b3Float4>& uniqueEdges,
+										 __global const b3AlignedObjectArray<b3GpuFace>& faces,
+										 __global const b3AlignedObjectArray<int>& indices,
+										 __global b3Aabb* aabbs,
+										 __global const b3GpuChildShape* gpuChildShapes,
+										 __global b3AlignedObjectArray<b3Float4>& gpuCompoundSepNormalsOut,
+										 __global b3AlignedObjectArray<int>& gpuHasCompoundSepNormalsOut,
+										 int numCompoundPairs,
+										 int i)
 {
-
-//	int i = get_global_id(0);
-	if (i<numCompoundPairs)
+	//	int i = get_global_id(0);
+	if (i < numCompoundPairs)
 	{
 		int bodyIndexA = gpuCompoundPairs[i].x;
 		int bodyIndexB = gpuCompoundPairs[i].y;
 
 		int childShapeIndexA = gpuCompoundPairs[i].z;
 		int childShapeIndexB = gpuCompoundPairs[i].w;
-		
+
 		int collidableIndexA = -1;
 		int collidableIndexB = -1;
-		
+
 		b3Quat ornA = rigidBodies[bodyIndexA].m_quat;
 		float4 posA = rigidBodies[bodyIndexA].m_pos;
-		
+
 		b3Quat ornB = rigidBodies[bodyIndexB].m_quat;
 		float4 posB = rigidBodies[bodyIndexB].m_pos;
-							
+
 		if (childShapeIndexA >= 0)
 		{
 			collidableIndexA = gpuChildShapes[childShapeIndexA].m_shapeIndex;
 			float4 childPosA = gpuChildShapes[childShapeIndexA].m_childPosition;
-			b3Quat	childOrnA = gpuChildShapes[childShapeIndexA].m_childOrientation;
-			float4 newPosA = b3QuatRotate(ornA,childPosA)+posA;
-			b3Quat newOrnA = b3QuatMul(ornA,childOrnA);
+			b3Quat childOrnA = gpuChildShapes[childShapeIndexA].m_childOrientation;
+			float4 newPosA = b3QuatRotate(ornA, childPosA) + posA;
+			b3Quat newOrnA = b3QuatMul(ornA, childOrnA);
 			posA = newPosA;
 			ornA = newOrnA;
-		} else
+		}
+		else
 		{
 			collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;
 		}
-		
-		if (childShapeIndexB>=0)
+
+		if (childShapeIndexB >= 0)
 		{
 			collidableIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;
 			float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;
 			b3Quat childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;
-			float4 newPosB = b3QuatRotate(ornB,childPosB)+posB;
-			b3Quat newOrnB = b3QuatMul(ornB,childOrnB);
+			float4 newPosB = b3QuatRotate(ornB, childPosB) + posB;
+			b3Quat newOrnB = b3QuatMul(ornB, childOrnB);
 			posB = newPosB;
 			ornB = newOrnB;
-		} else
-		{
-			collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;	
 		}
-	
+		else
+		{
+			collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;
+		}
+
 		gpuHasCompoundSepNormalsOut[i] = 0;
-	
+
 		int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;
 		int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;
-	
+
 		int shapeTypeA = collidables[collidableIndexA].m_shapeType;
 		int shapeTypeB = collidables[collidableIndexB].m_shapeType;
-	
 
 		if ((shapeTypeA != SHAPE_CONVEX_HULL) || (shapeTypeB != SHAPE_CONVEX_HULL))
 		{
@@ -1959,145 +1851,142 @@ __kernel void   processCompoundPairsKernel( __global const b3Int4* gpuCompoundPa
 		}
 
 		int hasSeparatingAxis = 5;
-							
-	//	int numFacesA = convexShapes[shapeIndexA].m_numFaces;
+
+		//	int numFacesA = convexShapes[shapeIndexA].m_numFaces;
 		float dmin = FLT_MAX;
 		posA.w = 0.f;
 		posB.w = 0.f;
 		float4 c0local = convexShapes[shapeIndexA].m_localCenter;
 		float4 c0 = transform(&c0local, &posA, &ornA);
 		float4 c1local = convexShapes[shapeIndexB].m_localCenter;
-		float4 c1 = transform(&c1local,&posB,&ornB);
+		float4 c1 = transform(&c1local, &posB, &ornB);
 		const float4 DeltaC2 = c0 - c1;
-		float4 sepNormal = make_float4(1,0,0,0);
-//		bool sepA = findSeparatingAxis(	convexShapes[shapeIndexA], convexShapes[shapeIndexB],posA,ornA,posB,ornB,DeltaC2,vertices,uniqueEdges,faces,indices,&sepNormal,&dmin);
-		bool sepA = findSeparatingAxis(	convexShapes[shapeIndexA], convexShapes[shapeIndexB],posA,ornA,posB,ornB,vertices,uniqueEdges,faces,indices,vertices,uniqueEdges,faces,indices,sepNormal);//,&dmin);
-	
+		float4 sepNormal = make_float4(1, 0, 0, 0);
+		//		bool sepA = findSeparatingAxis(	convexShapes[shapeIndexA], convexShapes[shapeIndexB],posA,ornA,posB,ornB,DeltaC2,vertices,uniqueEdges,faces,indices,&sepNormal,&dmin);
+		bool sepA = findSeparatingAxis(convexShapes[shapeIndexA], convexShapes[shapeIndexB], posA, ornA, posB, ornB, vertices, uniqueEdges, faces, indices, vertices, uniqueEdges, faces, indices, sepNormal);  //,&dmin);
+
 		hasSeparatingAxis = 4;
 		if (!sepA)
 		{
 			hasSeparatingAxis = 0;
-		} else
+		}
+		else
 		{
-			bool sepB = findSeparatingAxis(	convexShapes[shapeIndexB],convexShapes[shapeIndexA],posB,ornB,posA,ornA,vertices,uniqueEdges,faces,indices,vertices,uniqueEdges,faces,indices,sepNormal);//,&dmin);
+			bool sepB = findSeparatingAxis(convexShapes[shapeIndexB], convexShapes[shapeIndexA], posB, ornB, posA, ornA, vertices, uniqueEdges, faces, indices, vertices, uniqueEdges, faces, indices, sepNormal);  //,&dmin);
 
 			if (!sepB)
 			{
 				hasSeparatingAxis = 0;
-			} else//(!sepB)
+			}
+			else  //(!sepB)
 			{
-				bool sepEE = findSeparatingAxisEdgeEdge(	&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],posA,ornA,posB,ornB,DeltaC2,vertices,uniqueEdges,faces,indices,&sepNormal,&dmin);
+				bool sepEE = findSeparatingAxisEdgeEdge(&convexShapes[shapeIndexA], &convexShapes[shapeIndexB], posA, ornA, posB, ornB, DeltaC2, vertices, uniqueEdges, faces, indices, &sepNormal, &dmin);
 				if (sepEE)
 				{
-						gpuCompoundSepNormalsOut[i] = sepNormal;//fastNormalize4(sepNormal);
-						gpuHasCompoundSepNormalsOut[i] = 1;
-				}//sepEE
-			}//(!sepB)
-		}//(!sepA)
-		
-		
+					gpuCompoundSepNormalsOut[i] = sepNormal;  //fastNormalize4(sepNormal);
+					gpuHasCompoundSepNormalsOut[i] = 1;
+				}  //sepEE
+			}      //(!sepB)
+		}          //(!sepA)
 	}
-		
 }
 
-
-__kernel void   clipCompoundsHullHullKernel( __global const b3Int4* gpuCompoundPairs, 
-																					__global const b3RigidBodyData* rigidBodies, 
-																					__global const b3Collidable* collidables,
-																					__global const b3ConvexPolyhedronData* convexShapes, 
-																					__global const b3AlignedObjectArray<b3Float4>& vertices,
-																					__global const b3AlignedObjectArray<b3Float4>& uniqueEdges,
-																					__global const b3AlignedObjectArray<b3GpuFace>& faces,
-																					__global const b3AlignedObjectArray<int>& indices,
-																					__global const b3GpuChildShape* gpuChildShapes,
-																					__global const b3AlignedObjectArray<b3Float4>& gpuCompoundSepNormalsOut,
-																					__global const b3AlignedObjectArray<int>& gpuHasCompoundSepNormalsOut,
-																					__global struct b3Contact4Data* globalContactsOut,
-																					int* nGlobalContactsOut,
-																					int numCompoundPairs, int maxContactCapacity, int i)
+__kernel void clipCompoundsHullHullKernel(__global const b3Int4* gpuCompoundPairs,
+										  __global const b3RigidBodyData* rigidBodies,
+										  __global const b3Collidable* collidables,
+										  __global const b3ConvexPolyhedronData* convexShapes,
+										  __global const b3AlignedObjectArray<b3Float4>& vertices,
+										  __global const b3AlignedObjectArray<b3Float4>& uniqueEdges,
+										  __global const b3AlignedObjectArray<b3GpuFace>& faces,
+										  __global const b3AlignedObjectArray<int>& indices,
+										  __global const b3GpuChildShape* gpuChildShapes,
+										  __global const b3AlignedObjectArray<b3Float4>& gpuCompoundSepNormalsOut,
+										  __global const b3AlignedObjectArray<int>& gpuHasCompoundSepNormalsOut,
+										  __global struct b3Contact4Data* globalContactsOut,
+										  int* nGlobalContactsOut,
+										  int numCompoundPairs, int maxContactCapacity, int i)
 {
-
-//	int i = get_global_id(0);
+	//	int i = get_global_id(0);
 	int pairIndex = i;
-	
+
 	float4 worldVertsB1[64];
 	float4 worldVertsB2[64];
-	int capacityWorldVerts = 64;	
+	int capacityWorldVerts = 64;
 
 	float4 localContactsOut[64];
-	int localContactCapacity=64;
-	
+	int localContactCapacity = 64;
+
 	float minDist = -1e30f;
 	float maxDist = 0.0f;
 
-	if (i<numCompoundPairs)
+	if (i < numCompoundPairs)
 	{
-
 		if (gpuHasCompoundSepNormalsOut[i])
 		{
-
 			int bodyIndexA = gpuCompoundPairs[i].x;
 			int bodyIndexB = gpuCompoundPairs[i].y;
-			
+
 			int childShapeIndexA = gpuCompoundPairs[i].z;
 			int childShapeIndexB = gpuCompoundPairs[i].w;
-			
+
 			int collidableIndexA = -1;
 			int collidableIndexB = -1;
-			
+
 			b3Quat ornA = rigidBodies[bodyIndexA].m_quat;
 			float4 posA = rigidBodies[bodyIndexA].m_pos;
-			
+
 			b3Quat ornB = rigidBodies[bodyIndexB].m_quat;
 			float4 posB = rigidBodies[bodyIndexB].m_pos;
-								
+
 			if (childShapeIndexA >= 0)
 			{
 				collidableIndexA = gpuChildShapes[childShapeIndexA].m_shapeIndex;
 				float4 childPosA = gpuChildShapes[childShapeIndexA].m_childPosition;
 				b3Quat childOrnA = gpuChildShapes[childShapeIndexA].m_childOrientation;
-				float4 newPosA = b3QuatRotate(ornA,childPosA)+posA;
-				b3Quat newOrnA = b3QuatMul(ornA,childOrnA);
+				float4 newPosA = b3QuatRotate(ornA, childPosA) + posA;
+				b3Quat newOrnA = b3QuatMul(ornA, childOrnA);
 				posA = newPosA;
 				ornA = newOrnA;
-			} else
+			}
+			else
 			{
 				collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;
 			}
-			
-			if (childShapeIndexB>=0)
+
+			if (childShapeIndexB >= 0)
 			{
 				collidableIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;
 				float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;
-				b3Quat  childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;
-				float4 newPosB = b3QuatRotate(ornB,childPosB)+posB;
-				b3Quat  newOrnB = b3QuatMul(ornB,childOrnB);
+				b3Quat childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;
+				float4 newPosB = b3QuatRotate(ornB, childPosB) + posB;
+				b3Quat newOrnB = b3QuatMul(ornB, childOrnB);
 				posB = newPosB;
 				ornB = newOrnB;
-			} else
-			{
-				collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;	
 			}
-			
+			else
+			{
+				collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;
+			}
+
 			int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;
 			int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;
-		
+
 			int numLocalContactsOut = clipHullAgainstHull(gpuCompoundSepNormalsOut[i],
-														convexShapes[shapeIndexA], convexShapes[shapeIndexB],
-														posA,ornA,
-													  posB,ornB,
-													  worldVertsB1,worldVertsB2,capacityWorldVerts,
-														minDist, maxDist,
-														vertices,faces,indices,
-														vertices,faces,indices,
-														localContactsOut,localContactCapacity);
-												
-		if (numLocalContactsOut>0)
-		{
+														  convexShapes[shapeIndexA], convexShapes[shapeIndexB],
+														  posA, ornA,
+														  posB, ornB,
+														  worldVertsB1, worldVertsB2, capacityWorldVerts,
+														  minDist, maxDist,
+														  vertices, faces, indices,
+														  vertices, faces, indices,
+														  localContactsOut, localContactCapacity);
+
+			if (numLocalContactsOut > 0)
+			{
 				float4 normal = -gpuCompoundSepNormalsOut[i];
 				int nPoints = numLocalContactsOut;
 				float4* pointsIn = localContactsOut;
-				b3Int4 contactIdx;// = {-1,-1,-1,-1};
+				b3Int4 contactIdx;  // = {-1,-1,-1,-1};
 
 				contactIdx.s[0] = 0;
 				contactIdx.s[1] = 1;
@@ -2105,111 +1994,106 @@ __kernel void   clipCompoundsHullHullKernel( __global const b3Int4* gpuCompoundP
 				contactIdx.s[3] = 3;
 
 				int nReducedContacts = extractManifoldSequentialGlobal(pointsIn, nPoints, normal, &contactIdx);
-		
+
 				int dstIdx;
-				dstIdx = b3AtomicInc( nGlobalContactsOut);
-				if ((dstIdx+nReducedContacts) < maxContactCapacity)
+				dstIdx = b3AtomicInc(nGlobalContactsOut);
+				if ((dstIdx + nReducedContacts) < maxContactCapacity)
 				{
-					__global struct b3Contact4Data* c = globalContactsOut+ dstIdx;
+					__global struct b3Contact4Data* c = globalContactsOut + dstIdx;
 					c->m_worldNormalOnB = -normal;
-					c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);
+					c->m_restituitionCoeffCmp = (0.f * 0xffff);
+					c->m_frictionCoeffCmp = (0.7f * 0xffff);
 					c->m_batchIdx = pairIndex;
 					int bodyA = gpuCompoundPairs[pairIndex].x;
 					int bodyB = gpuCompoundPairs[pairIndex].y;
-					c->m_bodyAPtrAndSignBit = rigidBodies[bodyA].m_invMass==0?-bodyA:bodyA;
-					c->m_bodyBPtrAndSignBit = rigidBodies[bodyB].m_invMass==0?-bodyB:bodyB;
+					c->m_bodyAPtrAndSignBit = rigidBodies[bodyA].m_invMass == 0 ? -bodyA : bodyA;
+					c->m_bodyBPtrAndSignBit = rigidBodies[bodyB].m_invMass == 0 ? -bodyB : bodyB;
 					c->m_childIndexA = childShapeIndexA;
 					c->m_childIndexB = childShapeIndexB;
-					for (int i=0;i<nReducedContacts;i++)
+					for (int i = 0; i < nReducedContacts; i++)
 					{
 						c->m_worldPosB[i] = pointsIn[contactIdx.s[i]];
 					}
-					b3Contact4Data_setNumPoints(c,nReducedContacts);
+					b3Contact4Data_setNumPoints(c, nReducedContacts);
 				}
-				
-			}//		if (numContactsOut>0)
-		}//		if (gpuHasCompoundSepNormalsOut[i])
-	}//	if (i<numCompoundPairs)
 
+			}  //		if (numContactsOut>0)
+		}      //		if (gpuHasCompoundSepNormalsOut[i])
+	}          //	if (i<numCompoundPairs)
 }
 
-
 void computeContactCompoundCompound(int pairIndex,
-																int bodyIndexA, int bodyIndexB, 
-																int collidableIndexA, int collidableIndexB, 
-																const b3RigidBodyData* rigidBodies, 
-																const b3Collidable* collidables,
-																const b3ConvexPolyhedronData* convexShapes,
-																const b3GpuChildShape* cpuChildShapes,
-																const b3AlignedObjectArray<b3Aabb>& hostAabbsWorldSpace,
-																const b3AlignedObjectArray<b3Aabb>& hostAabbsLocalSpace,
+									int bodyIndexA, int bodyIndexB,
+									int collidableIndexA, int collidableIndexB,
+									const b3RigidBodyData* rigidBodies,
+									const b3Collidable* collidables,
+									const b3ConvexPolyhedronData* convexShapes,
+									const b3GpuChildShape* cpuChildShapes,
+									const b3AlignedObjectArray<b3Aabb>& hostAabbsWorldSpace,
+									const b3AlignedObjectArray<b3Aabb>& hostAabbsLocalSpace,
 
-																const b3AlignedObjectArray<b3Vector3>& convexVertices,
-																const b3AlignedObjectArray<b3Vector3>& hostUniqueEdges,
-																const b3AlignedObjectArray<int>& convexIndices,
-																const b3AlignedObjectArray<b3GpuFace>& faces,
-																
-																b3Contact4* globalContactsOut,
-																int& nGlobalContactsOut,
-																int maxContactCapacity,
-																b3AlignedObjectArray<b3QuantizedBvhNode>&	treeNodesCPU,
-																b3AlignedObjectArray<b3BvhSubtreeInfo>&	subTreesCPU,
-																b3AlignedObjectArray<b3BvhInfo>&	bvhInfoCPU
-																)
+									const b3AlignedObjectArray<b3Vector3>& convexVertices,
+									const b3AlignedObjectArray<b3Vector3>& hostUniqueEdges,
+									const b3AlignedObjectArray<int>& convexIndices,
+									const b3AlignedObjectArray<b3GpuFace>& faces,
+
+									b3Contact4* globalContactsOut,
+									int& nGlobalContactsOut,
+									int maxContactCapacity,
+									b3AlignedObjectArray<b3QuantizedBvhNode>& treeNodesCPU,
+									b3AlignedObjectArray<b3BvhSubtreeInfo>& subTreesCPU,
+									b3AlignedObjectArray<b3BvhInfo>& bvhInfoCPU)
 {
-
 	int shapeTypeB = collidables[collidableIndexB].m_shapeType;
 	b3Assert(shapeTypeB == SHAPE_COMPOUND_OF_CONVEX_HULLS);
 
 	b3AlignedObjectArray<b3Int4> cpuCompoundPairsOut;
-	int numCompoundPairsOut=0;
-	int maxNumCompoundPairsCapacity = 8192;//1024;
+	int numCompoundPairsOut = 0;
+	int maxNumCompoundPairsCapacity = 8192;  //1024;
 	cpuCompoundPairsOut.resize(maxNumCompoundPairsCapacity);
 
 	// work-in-progress
-	findCompoundPairsKernel( 
-							pairIndex,
-							bodyIndexA,bodyIndexB,
-							collidableIndexA,collidableIndexB,
-							rigidBodies, 
-							collidables,
-							convexShapes, 
-							convexVertices,
-							hostAabbsWorldSpace,
-							hostAabbsLocalSpace,
-							cpuChildShapes,
-							&cpuCompoundPairsOut[0],
-							&numCompoundPairsOut,
-							maxNumCompoundPairsCapacity	,
-							treeNodesCPU,
-							subTreesCPU,
-							bvhInfoCPU
-							);
+	findCompoundPairsKernel(
+		pairIndex,
+		bodyIndexA, bodyIndexB,
+		collidableIndexA, collidableIndexB,
+		rigidBodies,
+		collidables,
+		convexShapes,
+		convexVertices,
+		hostAabbsWorldSpace,
+		hostAabbsLocalSpace,
+		cpuChildShapes,
+		&cpuCompoundPairsOut[0],
+		&numCompoundPairsOut,
+		maxNumCompoundPairsCapacity,
+		treeNodesCPU,
+		subTreesCPU,
+		bvhInfoCPU);
 
-	printf("maxNumAabbChecks=%d\n",maxNumAabbChecks);
-	if (numCompoundPairsOut>maxNumCompoundPairsCapacity)
+	printf("maxNumAabbChecks=%d\n", maxNumAabbChecks);
+	if (numCompoundPairsOut > maxNumCompoundPairsCapacity)
 	{
-		b3Error("numCompoundPairsOut exceeded maxNumCompoundPairsCapacity (%d)\n",maxNumCompoundPairsCapacity);
-		numCompoundPairsOut=maxNumCompoundPairsCapacity;
+		b3Error("numCompoundPairsOut exceeded maxNumCompoundPairsCapacity (%d)\n", maxNumCompoundPairsCapacity);
+		numCompoundPairsOut = maxNumCompoundPairsCapacity;
 	}
 	b3AlignedObjectArray<b3Float4> cpuCompoundSepNormalsOut;
 	b3AlignedObjectArray<int> cpuHasCompoundSepNormalsOut;
 	cpuCompoundSepNormalsOut.resize(numCompoundPairsOut);
 	cpuHasCompoundSepNormalsOut.resize(numCompoundPairsOut);
 
-	for (int i=0;i<numCompoundPairsOut;i++)
+	for (int i = 0; i < numCompoundPairsOut; i++)
 	{
-
-		processCompoundPairsKernel(&cpuCompoundPairsOut[0],rigidBodies,collidables,convexShapes,convexVertices,hostUniqueEdges,faces,convexIndices,0,cpuChildShapes,
-			cpuCompoundSepNormalsOut,cpuHasCompoundSepNormalsOut,numCompoundPairsOut,i);
+		processCompoundPairsKernel(&cpuCompoundPairsOut[0], rigidBodies, collidables, convexShapes, convexVertices, hostUniqueEdges, faces, convexIndices, 0, cpuChildShapes,
+								   cpuCompoundSepNormalsOut, cpuHasCompoundSepNormalsOut, numCompoundPairsOut, i);
 	}
 
-	for (int i=0;i<numCompoundPairsOut;i++)
+	for (int i = 0; i < numCompoundPairsOut; i++)
 	{
-		clipCompoundsHullHullKernel(&cpuCompoundPairsOut[0],rigidBodies,collidables,convexShapes,convexVertices,hostUniqueEdges,faces,convexIndices,cpuChildShapes,
-			cpuCompoundSepNormalsOut,cpuHasCompoundSepNormalsOut,globalContactsOut,&nGlobalContactsOut,numCompoundPairsOut,maxContactCapacity,i);
+		clipCompoundsHullHullKernel(&cpuCompoundPairsOut[0], rigidBodies, collidables, convexShapes, convexVertices, hostUniqueEdges, faces, convexIndices, cpuChildShapes,
+									cpuCompoundSepNormalsOut, cpuHasCompoundSepNormalsOut, globalContactsOut, &nGlobalContactsOut, numCompoundPairsOut, maxContactCapacity, i);
 	}
-		/*
+	/*
 		int childColIndexA = gpuChildShapes[childShapeIndexA].m_shapeIndex;
 
 					float4 posA = rigidBodies[bodyIndexA].m_pos;
@@ -2235,7 +2119,6 @@ void computeContactCompoundCompound(int pairIndex,
 							);
 							*/
 
-	
 	/*
 	if (foundSepAxis)
 	{
@@ -2271,8 +2154,8 @@ void computeContactCompoundCompound(int pairIndex,
 	}
 	*/
 
-//	return contactIndex;
-	
+	//	return contactIndex;
+
 	/*
 
 	int numChildrenB = collidables[collidableIndexB].m_numChildShapes;
@@ -2294,56 +2177,52 @@ void computeContactCompoundCompound(int pairIndex,
 
 	}
 	*/
-
 }
 
 void computeContactPlaneCompound(int pairIndex,
-																int bodyIndexA, int bodyIndexB, 
-																int collidableIndexA, int collidableIndexB, 
-																const b3RigidBodyData* rigidBodies, 
-																const b3Collidable* collidables,
-																const b3ConvexPolyhedronData* convexShapes,
-																const b3GpuChildShape* cpuChildShapes,
-																const b3Vector3* convexVertices,
-																const int* convexIndices,
-																const b3GpuFace* faces,
-																
-																b3Contact4* globalContactsOut,
-																int& nGlobalContactsOut,
-																int maxContactCapacity)
-{
+								 int bodyIndexA, int bodyIndexB,
+								 int collidableIndexA, int collidableIndexB,
+								 const b3RigidBodyData* rigidBodies,
+								 const b3Collidable* collidables,
+								 const b3ConvexPolyhedronData* convexShapes,
+								 const b3GpuChildShape* cpuChildShapes,
+								 const b3Vector3* convexVertices,
+								 const int* convexIndices,
+								 const b3GpuFace* faces,
 
+								 b3Contact4* globalContactsOut,
+								 int& nGlobalContactsOut,
+								 int maxContactCapacity)
+{
 	int shapeTypeB = collidables[collidableIndexB].m_shapeType;
 	b3Assert(shapeTypeB == SHAPE_COMPOUND_OF_CONVEX_HULLS);
 
-
 	int numChildrenB = collidables[collidableIndexB].m_numChildShapes;
-	for (int c=0;c<numChildrenB;c++)
+	for (int c = 0; c < numChildrenB; c++)
 	{
-		int childShapeIndexB = collidables[collidableIndexB].m_shapeIndex+c;
+		int childShapeIndexB = collidables[collidableIndexB].m_shapeIndex + c;
 		int childColIndexB = cpuChildShapes[childShapeIndexB].m_shapeIndex;
 
 		float4 rootPosB = rigidBodies[bodyIndexB].m_pos;
 		b3Quaternion rootOrnB = rigidBodies[bodyIndexB].m_quat;
 		b3Vector3 childPosB = cpuChildShapes[childShapeIndexB].m_childPosition;
 		b3Quaternion childOrnB = cpuChildShapes[childShapeIndexB].m_childOrientation;
-		float4  posB = b3QuatRotate(rootOrnB,childPosB)+rootPosB;
-		b3Quaternion ornB = rootOrnB*childOrnB;//b3QuatMul(ornB,childOrnB);
+		float4 posB = b3QuatRotate(rootOrnB, childPosB) + rootPosB;
+		b3Quaternion ornB = rootOrnB * childOrnB;  //b3QuatMul(ornB,childOrnB);
 
 		int shapeIndexB = collidables[childColIndexB].m_shapeIndex;
 
 		const b3ConvexPolyhedronData* hullB = &convexShapes[shapeIndexB];
-	
-		
+
 		b3Vector3 posA = rigidBodies[bodyIndexA].m_pos;
 		b3Quaternion ornA = rigidBodies[bodyIndexA].m_quat;
 
-	//	int numContactsOut = 0;
-	//	int numWorldVertsB1= 0;
+		//	int numContactsOut = 0;
+		//	int numWorldVertsB1= 0;
 
 		b3Vector3 planeEq = faces[collidables[collidableIndexA].m_shapeIndex].m_plane;
-		b3Vector3 planeNormal=b3MakeVector3(planeEq.x,planeEq.y,planeEq.z);
-		b3Vector3 planeNormalWorld = b3QuatRotate(ornA,planeNormal);
+		b3Vector3 planeNormal = b3MakeVector3(planeEq.x, planeEq.y, planeEq.z);
+		b3Vector3 planeNormalWorld = b3QuatRotate(ornA, planeNormal);
 		float planeConstant = planeEq.w;
 		b3Transform convexWorldTransform;
 		convexWorldTransform.setIdentity();
@@ -2355,16 +2234,16 @@ void computeContactPlaneCompound(int pairIndex,
 		planeTransform.setRotation(ornA);
 
 		b3Transform planeInConvex;
-		planeInConvex= convexWorldTransform.inverse() * planeTransform;
+		planeInConvex = convexWorldTransform.inverse() * planeTransform;
 		b3Transform convexInPlane;
 		convexInPlane = planeTransform.inverse() * convexWorldTransform;
-	
-		b3Vector3 planeNormalInConvex = planeInConvex.getBasis()*-planeNormal;
+
+		b3Vector3 planeNormalInConvex = planeInConvex.getBasis() * -planeNormal;
 		float maxDot = -1e30;
-		int hitVertex=-1;
+		int hitVertex = -1;
 		b3Vector3 hitVtx;
 
-	#define MAX_PLANE_CONVEX_POINTS 64
+#define MAX_PLANE_CONVEX_POINTS 64
 
 		b3Vector3 contactPoints[MAX_PLANE_CONVEX_POINTS];
 		int numPoints = 0;
@@ -2374,54 +2253,52 @@ void computeContactPlaneCompound(int pairIndex,
 		contactIdx.s[1] = 1;
 		contactIdx.s[2] = 2;
 		contactIdx.s[3] = 3;
-	
-		for (int i=0;i<hullB->m_numVertices;i++)
+
+		for (int i = 0; i < hullB->m_numVertices; i++)
 		{
-			b3Vector3 vtx = convexVertices[hullB->m_vertexOffset+i];
+			b3Vector3 vtx = convexVertices[hullB->m_vertexOffset + i];
 			float curDot = vtx.dot(planeNormalInConvex);
 
-
-			if (curDot>maxDot)
+			if (curDot > maxDot)
 			{
-				hitVertex=i;
-				maxDot=curDot;
+				hitVertex = i;
+				maxDot = curDot;
 				hitVtx = vtx;
 				//make sure the deepest points is always included
-				if (numPoints==MAX_PLANE_CONVEX_POINTS)
+				if (numPoints == MAX_PLANE_CONVEX_POINTS)
 					numPoints--;
 			}
 
-			if (numPoints<MAX_PLANE_CONVEX_POINTS)
+			if (numPoints < MAX_PLANE_CONVEX_POINTS)
 			{
-				b3Vector3 vtxWorld = convexWorldTransform*vtx;
-				b3Vector3 vtxInPlane = planeTransform.inverse()*vtxWorld;
-				float dist = planeNormal.dot(vtxInPlane)-planeConstant;
-				if (dist<0.f)
+				b3Vector3 vtxWorld = convexWorldTransform * vtx;
+				b3Vector3 vtxInPlane = planeTransform.inverse() * vtxWorld;
+				float dist = planeNormal.dot(vtxInPlane) - planeConstant;
+				if (dist < 0.f)
 				{
 					vtxWorld.w = dist;
 					contactPoints[numPoints] = vtxWorld;
 					numPoints++;
 				}
 			}
-
 		}
 
-		int numReducedPoints  = 0;
+		int numReducedPoints = 0;
 
 		numReducedPoints = numPoints;
-	
-		if (numPoints>4)
+
+		if (numPoints > 4)
 		{
-			numReducedPoints = extractManifoldSequentialGlobal( contactPoints, numPoints, planeNormalInConvex, &contactIdx);
+			numReducedPoints = extractManifoldSequentialGlobal(contactPoints, numPoints, planeNormalInConvex, &contactIdx);
 		}
 		int dstIdx;
-	//    dstIdx = nGlobalContactsOut++;//AppendInc( nGlobalContactsOut, dstIdx );
-		
-		if (numReducedPoints>0)
+		//    dstIdx = nGlobalContactsOut++;//AppendInc( nGlobalContactsOut, dstIdx );
+
+		if (numReducedPoints > 0)
 		{
 			if (nGlobalContactsOut < maxContactCapacity)
 			{
-				dstIdx=nGlobalContactsOut;
+				dstIdx = nGlobalContactsOut;
 				nGlobalContactsOut++;
 
 				b3Contact4* c = &globalContactsOut[dstIdx];
@@ -2430,48 +2307,37 @@ void computeContactPlaneCompound(int pairIndex,
 				c->setRestituitionCoeff(0.f);
 
 				c->m_batchIdx = pairIndex;
-				c->m_bodyAPtrAndSignBit = rigidBodies[bodyIndexA].m_invMass==0?-bodyIndexA:bodyIndexA;
-				c->m_bodyBPtrAndSignBit = rigidBodies[bodyIndexB].m_invMass==0?-bodyIndexB:bodyIndexB;
-				for (int i=0;i<numReducedPoints;i++)
+				c->m_bodyAPtrAndSignBit = rigidBodies[bodyIndexA].m_invMass == 0 ? -bodyIndexA : bodyIndexA;
+				c->m_bodyBPtrAndSignBit = rigidBodies[bodyIndexB].m_invMass == 0 ? -bodyIndexB : bodyIndexB;
+				for (int i = 0; i < numReducedPoints; i++)
 				{
 					b3Vector3 pOnB1 = contactPoints[contactIdx.s[i]];
 					c->m_worldPosB[i] = pOnB1;
 				}
 				c->m_worldNormalOnB.w = (b3Scalar)numReducedPoints;
-			}//if (dstIdx < numPairs)
-		}	
-		
+			}  //if (dstIdx < numPairs)
+		}
 	}
-
-	
 }
 
-
-
-
-
-void	computeContactSphereConvex(int pairIndex,
-																int bodyIndexA, int bodyIndexB, 
-																int collidableIndexA, int collidableIndexB, 
-																const b3RigidBodyData* rigidBodies, 
-																const b3Collidable* collidables,
-																const b3ConvexPolyhedronData* convexShapes,
-																const b3Vector3* convexVertices,
-																const int* convexIndices,
-																const b3GpuFace* faces,
-																b3Contact4* globalContactsOut,
-																int& nGlobalContactsOut,
-																int maxContactCapacity)
+void computeContactSphereConvex(int pairIndex,
+								int bodyIndexA, int bodyIndexB,
+								int collidableIndexA, int collidableIndexB,
+								const b3RigidBodyData* rigidBodies,
+								const b3Collidable* collidables,
+								const b3ConvexPolyhedronData* convexShapes,
+								const b3Vector3* convexVertices,
+								const int* convexIndices,
+								const b3GpuFace* faces,
+								b3Contact4* globalContactsOut,
+								int& nGlobalContactsOut,
+								int maxContactCapacity)
 {
-
 	float radius = collidables[collidableIndexA].m_radius;
 	float4 spherePos1 = rigidBodies[bodyIndexA].m_pos;
 	b3Quaternion sphereOrn = rigidBodies[bodyIndexA].m_quat;
 
-
-
 	float4 pos = rigidBodies[bodyIndexB].m_pos;
-	
 
 	b3Quaternion quat = rigidBodies[bodyIndexB].m_quat;
 
@@ -2487,64 +2353,65 @@ void	computeContactSphereConvex(int pairIndex,
 	int shapeIndex = collidables[collidableIndex].m_shapeIndex;
 	int numFaces = convexShapes[shapeIndex].m_numFaces;
 	float4 closestPnt = b3MakeVector3(0, 0, 0, 0);
-//	float4 hitNormalWorld = b3MakeVector3(0, 0, 0, 0);
-	float minDist = -1000000.f; // TODO: What is the largest/smallest float?
+	//	float4 hitNormalWorld = b3MakeVector3(0, 0, 0, 0);
+	float minDist = -1000000.f;  // TODO: What is the largest/smallest float?
 	bool bCollide = true;
 	int region = -1;
 	float4 localHitNormal;
-	for ( int f = 0; f < numFaces; f++ )
+	for (int f = 0; f < numFaces; f++)
 	{
-		b3GpuFace face = faces[convexShapes[shapeIndex].m_faceOffset+f];
+		b3GpuFace face = faces[convexShapes[shapeIndex].m_faceOffset + f];
 		float4 planeEqn;
-		float4 localPlaneNormal = b3MakeVector3(face.m_plane.x,face.m_plane.y,face.m_plane.z,0.f);
-		float4 n1 = localPlaneNormal;//quatRotate(quat,localPlaneNormal);
+		float4 localPlaneNormal = b3MakeVector3(face.m_plane.x, face.m_plane.y, face.m_plane.z, 0.f);
+		float4 n1 = localPlaneNormal;  //quatRotate(quat,localPlaneNormal);
 		planeEqn = n1;
 		planeEqn[3] = face.m_plane.w;
 
 		float4 pntReturn;
 		float dist = signedDistanceFromPointToPlane(spherePos, planeEqn, &pntReturn);
 
-		if ( dist > radius)
+		if (dist > radius)
 		{
 			bCollide = false;
 			break;
 		}
 
-		if ( dist > 0 )
+		if (dist > 0)
 		{
 			//might hit an edge or vertex
 			b3Vector3 out;
 
 			bool isInPoly = IsPointInPolygon(spherePos,
-					&face,
-					&convexVertices[convexShapes[shapeIndex].m_vertexOffset],
-					convexIndices,
-                    &out);
+											 &face,
+											 &convexVertices[convexShapes[shapeIndex].m_vertexOffset],
+											 convexIndices,
+											 &out);
 			if (isInPoly)
 			{
-				if (dist>minDist)
+				if (dist > minDist)
 				{
 					minDist = dist;
 					closestPnt = pntReturn;
 					localHitNormal = planeEqn;
-					region=1;
+					region = 1;
 				}
-			} else
+			}
+			else
 			{
-				b3Vector3 tmp = spherePos-out;
+				b3Vector3 tmp = spherePos - out;
 				b3Scalar l2 = tmp.length2();
-				if (l2<radius*radius)
+				if (l2 < radius * radius)
 				{
-					dist  = b3Sqrt(l2);
-					if (dist>minDist)
+					dist = b3Sqrt(l2);
+					if (dist > minDist)
 					{
 						minDist = dist;
 						closestPnt = out;
-						localHitNormal = tmp/dist;
-						region=2;
+						localHitNormal = tmp / dist;
+						region = 2;
 					}
-					
-				} else
+				}
+				else
 				{
 					bCollide = false;
 					break;
@@ -2553,12 +2420,12 @@ void	computeContactSphereConvex(int pairIndex,
 		}
 		else
 		{
-			if ( dist > minDist )
+			if (dist > minDist)
 			{
 				minDist = dist;
 				closestPnt = pntReturn;
 				localHitNormal = planeEqn;
-				region=3;
+				region = 3;
 			}
 		}
 	}
@@ -2567,128 +2434,113 @@ void	computeContactSphereConvex(int pairIndex,
 
 	if (bCollide && minDist > -10000)
 	{
-		
-		float4 normalOnSurfaceB1 = tr.getBasis()*localHitNormal;//-hitNormalWorld;
+		float4 normalOnSurfaceB1 = tr.getBasis() * localHitNormal;  //-hitNormalWorld;
 		float4 pOnB1 = tr(closestPnt);
 		//printf("dist ,%f,",minDist);
-		float actualDepth = minDist-radius;
-		if (actualDepth<0)
+		float actualDepth = minDist - radius;
+		if (actualDepth < 0)
 		{
-		//printf("actualDepth = ,%f,", actualDepth);
-		//printf("normalOnSurfaceB1 = ,%f,%f,%f,", normalOnSurfaceB1.x,normalOnSurfaceB1.y,normalOnSurfaceB1.z);
-		//printf("region=,%d,\n", region);
-		pOnB1[3] = actualDepth;
+			//printf("actualDepth = ,%f,", actualDepth);
+			//printf("normalOnSurfaceB1 = ,%f,%f,%f,", normalOnSurfaceB1.x,normalOnSurfaceB1.y,normalOnSurfaceB1.z);
+			//printf("region=,%d,\n", region);
+			pOnB1[3] = actualDepth;
 
-		int dstIdx;
-//    dstIdx = nGlobalContactsOut++;//AppendInc( nGlobalContactsOut, dstIdx );
-		
-		if (nGlobalContactsOut < maxContactCapacity)
-		{
-			dstIdx=nGlobalContactsOut;
-			nGlobalContactsOut++;
+			int dstIdx;
+			//    dstIdx = nGlobalContactsOut++;//AppendInc( nGlobalContactsOut, dstIdx );
 
-			b3Contact4* c = &globalContactsOut[dstIdx];
-			c->m_worldNormalOnB = normalOnSurfaceB1;
-			c->setFrictionCoeff(0.7);
-			c->setRestituitionCoeff(0.f);
+			if (nGlobalContactsOut < maxContactCapacity)
+			{
+				dstIdx = nGlobalContactsOut;
+				nGlobalContactsOut++;
 
-			c->m_batchIdx = pairIndex;
-			c->m_bodyAPtrAndSignBit = rigidBodies[bodyIndexA].m_invMass==0?-bodyIndexA:bodyIndexA;
-			c->m_bodyBPtrAndSignBit = rigidBodies[bodyIndexB].m_invMass==0?-bodyIndexB:bodyIndexB;
-			c->m_worldPosB[0] = pOnB1;
-			int numPoints = 1;
-			c->m_worldNormalOnB.w = (b3Scalar)numPoints;
-		}//if (dstIdx < numPairs)
+				b3Contact4* c = &globalContactsOut[dstIdx];
+				c->m_worldNormalOnB = normalOnSurfaceB1;
+				c->setFrictionCoeff(0.7);
+				c->setRestituitionCoeff(0.f);
+
+				c->m_batchIdx = pairIndex;
+				c->m_bodyAPtrAndSignBit = rigidBodies[bodyIndexA].m_invMass == 0 ? -bodyIndexA : bodyIndexA;
+				c->m_bodyBPtrAndSignBit = rigidBodies[bodyIndexB].m_invMass == 0 ? -bodyIndexB : bodyIndexB;
+				c->m_worldPosB[0] = pOnB1;
+				int numPoints = 1;
+				c->m_worldNormalOnB.w = (b3Scalar)numPoints;
+			}  //if (dstIdx < numPairs)
 		}
-	}//if (hasCollision)
-	
+	}  //if (hasCollision)
 }
 
-
-
-
 int computeContactConvexConvex2(
-																int pairIndex,
-																int bodyIndexA, int bodyIndexB, 
-																int collidableIndexA, int collidableIndexB, 
-																const b3AlignedObjectArray<b3RigidBodyData>& rigidBodies, 
-																const b3AlignedObjectArray<b3Collidable>& collidables,
-																const b3AlignedObjectArray<b3ConvexPolyhedronData>& convexShapes,
-																const b3AlignedObjectArray<b3Vector3>& convexVertices,
-																const b3AlignedObjectArray<b3Vector3>& uniqueEdges,
-																const b3AlignedObjectArray<int>& convexIndices,
-																const b3AlignedObjectArray<b3GpuFace>& faces,
-																b3AlignedObjectArray<b3Contact4>& globalContactsOut,
-																int& nGlobalContactsOut,
-																int maxContactCapacity,
-																const b3AlignedObjectArray<b3Contact4>& oldContacts
-																)
+	int pairIndex,
+	int bodyIndexA, int bodyIndexB,
+	int collidableIndexA, int collidableIndexB,
+	const b3AlignedObjectArray<b3RigidBodyData>& rigidBodies,
+	const b3AlignedObjectArray<b3Collidable>& collidables,
+	const b3AlignedObjectArray<b3ConvexPolyhedronData>& convexShapes,
+	const b3AlignedObjectArray<b3Vector3>& convexVertices,
+	const b3AlignedObjectArray<b3Vector3>& uniqueEdges,
+	const b3AlignedObjectArray<int>& convexIndices,
+	const b3AlignedObjectArray<b3GpuFace>& faces,
+	b3AlignedObjectArray<b3Contact4>& globalContactsOut,
+	int& nGlobalContactsOut,
+	int maxContactCapacity,
+	const b3AlignedObjectArray<b3Contact4>& oldContacts)
 {
 	int contactIndex = -1;
 	b3Vector3 posA = rigidBodies[bodyIndexA].m_pos;
 	b3Quaternion ornA = rigidBodies[bodyIndexA].m_quat;
 	b3Vector3 posB = rigidBodies[bodyIndexB].m_pos;
 	b3Quaternion ornB = rigidBodies[bodyIndexB].m_quat;
-	
 
 	b3ConvexPolyhedronData hullA, hullB;
-    
+
 	b3Vector3 sepNormalWorldSpace;
 
-	
+	b3Collidable colA = collidables[collidableIndexA];
+	hullA = convexShapes[colA.m_shapeIndex];
+	//printf("numvertsA = %d\n",hullA.m_numVertices);
 
-    b3Collidable colA = collidables[collidableIndexA];
-    hullA = convexShapes[colA.m_shapeIndex];
-    //printf("numvertsA = %d\n",hullA.m_numVertices);
-    
-    
-    b3Collidable colB = collidables[collidableIndexB];
-    hullB = convexShapes[colB.m_shapeIndex];
-    //printf("numvertsB = %d\n",hullB.m_numVertices);
+	b3Collidable colB = collidables[collidableIndexB];
+	hullB = convexShapes[colB.m_shapeIndex];
+	//printf("numvertsB = %d\n",hullB.m_numVertices);
 
-//	int contactCapacity = MAX_VERTS;
+	//	int contactCapacity = MAX_VERTS;
 	//int numContactsOut=0;
 
-
 #ifdef _WIN32
 	b3Assert(_finite(rigidBodies[bodyIndexA].m_pos.x));
 	b3Assert(_finite(rigidBodies[bodyIndexB].m_pos.x));
 #endif
-	
-		bool foundSepAxis = findSeparatingAxis(hullA,hullB,
-							posA,
-							ornA,
-							posB,
-							ornB,
 
-							convexVertices,uniqueEdges,faces,convexIndices,
-							convexVertices,uniqueEdges,faces,convexIndices,
-							
-							sepNormalWorldSpace
-							);
+	bool foundSepAxis = findSeparatingAxis(hullA, hullB,
+										   posA,
+										   ornA,
+										   posB,
+										   ornB,
+
+										   convexVertices, uniqueEdges, faces, convexIndices,
+										   convexVertices, uniqueEdges, faces, convexIndices,
+
+										   sepNormalWorldSpace);
 
-	
 	if (foundSepAxis)
 	{
-		
-		
 		contactIndex = clipHullHullSingle(
 			bodyIndexA, bodyIndexB,
-						   posA,ornA,
-						   posB,ornB,
+			posA, ornA,
+			posB, ornB,
 			collidableIndexA, collidableIndexB,
-			&rigidBodies, 
+			&rigidBodies,
 			&globalContactsOut,
 			nGlobalContactsOut,
-			
+
 			convexShapes,
 			convexShapes,
-	
-			convexVertices, 
-			uniqueEdges, 
+
+			convexVertices,
+			uniqueEdges,
 			faces,
 			convexIndices,
-	
+
 			convexVertices,
 			uniqueEdges,
 			faces,
@@ -2698,50 +2550,42 @@ int computeContactConvexConvex2(
 			collidables,
 			sepNormalWorldSpace,
 			maxContactCapacity);
-			
 	}
 
 	return contactIndex;
 }
 
+void GpuSatCollision::computeConvexConvexContactsGPUSAT(b3OpenCLArray<b3Int4>* pairs, int nPairs,
+														const b3OpenCLArray<b3RigidBodyData>* bodyBuf,
+														b3OpenCLArray<b3Contact4>* contactOut, int& nContacts,
+														const b3OpenCLArray<b3Contact4>* oldContacts,
+														int maxContactCapacity,
+														int compoundPairCapacity,
+														const b3OpenCLArray<b3ConvexPolyhedronData>& convexData,
+														const b3OpenCLArray<b3Vector3>& gpuVertices,
+														const b3OpenCLArray<b3Vector3>& gpuUniqueEdges,
+														const b3OpenCLArray<b3GpuFace>& gpuFaces,
+														const b3OpenCLArray<int>& gpuIndices,
+														const b3OpenCLArray<b3Collidable>& gpuCollidables,
+														const b3OpenCLArray<b3GpuChildShape>& gpuChildShapes,
 
+														const b3OpenCLArray<b3Aabb>& clAabbsWorldSpace,
+														const b3OpenCLArray<b3Aabb>& clAabbsLocalSpace,
 
+														b3OpenCLArray<b3Vector3>& worldVertsB1GPU,
+														b3OpenCLArray<b3Int4>& clippingFacesOutGPU,
+														b3OpenCLArray<b3Vector3>& worldNormalsAGPU,
+														b3OpenCLArray<b3Vector3>& worldVertsA1GPU,
+														b3OpenCLArray<b3Vector3>& worldVertsB2GPU,
+														b3AlignedObjectArray<class b3OptimizedBvh*>& bvhDataUnused,
+														b3OpenCLArray<b3QuantizedBvhNode>* treeNodesGPU,
+														b3OpenCLArray<b3BvhSubtreeInfo>* subTreesGPU,
+														b3OpenCLArray<b3BvhInfo>* bvhInfo,
 
-
-																
-																
-void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>* pairs, int nPairs,
-			const b3OpenCLArray<b3RigidBodyData>* bodyBuf,
-			b3OpenCLArray<b3Contact4>* contactOut, int& nContacts,
-			const b3OpenCLArray<b3Contact4>* oldContacts,
-			int maxContactCapacity,
-			int compoundPairCapacity,
-			const b3OpenCLArray<b3ConvexPolyhedronData>& convexData,
-			const b3OpenCLArray<b3Vector3>& gpuVertices,
-			const b3OpenCLArray<b3Vector3>& gpuUniqueEdges,
-			const b3OpenCLArray<b3GpuFace>& gpuFaces,
-			const b3OpenCLArray<int>& gpuIndices,
-			const b3OpenCLArray<b3Collidable>& gpuCollidables,
-			const b3OpenCLArray<b3GpuChildShape>& gpuChildShapes,
-
-			const b3OpenCLArray<b3Aabb>& clAabbsWorldSpace,
-			const b3OpenCLArray<b3Aabb>& clAabbsLocalSpace,
-
-            b3OpenCLArray<b3Vector3>& worldVertsB1GPU,
-            b3OpenCLArray<b3Int4>& clippingFacesOutGPU,
-            b3OpenCLArray<b3Vector3>& worldNormalsAGPU,
-            b3OpenCLArray<b3Vector3>& worldVertsA1GPU,
-            b3OpenCLArray<b3Vector3>& worldVertsB2GPU,    
-			b3AlignedObjectArray<class b3OptimizedBvh*>& bvhDataUnused,
-			b3OpenCLArray<b3QuantizedBvhNode>*	treeNodesGPU,
-			b3OpenCLArray<b3BvhSubtreeInfo>*	subTreesGPU,
-			b3OpenCLArray<b3BvhInfo>*	bvhInfo,
-
-			int numObjects,
-			int maxTriConvexPairCapacity,
-			b3OpenCLArray<b3Int4>& triangleConvexPairsOut,
-			int& numTriConvexPairsOut
-			)
+														int numObjects,
+														int maxTriConvexPairCapacity,
+														b3OpenCLArray<b3Int4>& triangleConvexPairsOut,
+														int& numTriConvexPairsOut)
 {
 	myframecount++;
 
@@ -2750,14 +2594,13 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 
 #ifdef CHECK_ON_HOST
 
-
-	b3AlignedObjectArray<b3QuantizedBvhNode>	treeNodesCPU;
+	b3AlignedObjectArray<b3QuantizedBvhNode> treeNodesCPU;
 	treeNodesGPU->copyToHost(treeNodesCPU);
 
-	b3AlignedObjectArray<b3BvhSubtreeInfo>	subTreesCPU;
+	b3AlignedObjectArray<b3BvhSubtreeInfo> subTreesCPU;
 	subTreesGPU->copyToHost(subTreesCPU);
 
-	b3AlignedObjectArray<b3BvhInfo>	bvhInfoCPU;
+	b3AlignedObjectArray<b3BvhInfo> bvhInfoCPU;
 	bvhInfo->copyToHost(bvhInfoCPU);
 
 	b3AlignedObjectArray<b3Aabb> hostAabbsWorldSpace;
@@ -2772,8 +2615,6 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 	b3AlignedObjectArray<b3RigidBodyData> hostBodyBuf;
 	bodyBuf->copyToHost(hostBodyBuf);
 
-	
-
 	b3AlignedObjectArray<b3ConvexPolyhedronData> hostConvexData;
 	convexData.copyToHost(hostConvexData);
 
@@ -2788,10 +2629,9 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 	gpuIndices.copyToHost(hostIndices);
 	b3AlignedObjectArray<b3Collidable> hostCollidables;
 	gpuCollidables.copyToHost(hostCollidables);
-	
+
 	b3AlignedObjectArray<b3GpuChildShape> cpuChildShapes;
 	gpuChildShapes.copyToHost(cpuChildShapes);
-	
 
 	b3AlignedObjectArray<b3Int4> hostTriangleConvexPairs;
 
@@ -2802,16 +2642,15 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 	}
 
 	b3AlignedObjectArray<b3Contact4> oldHostContacts;
-	
+
 	if (oldContacts->size())
 	{
 		oldContacts->copyToHost(oldHostContacts);
 	}
 
-
 	hostContacts.resize(maxContactCapacity);
 
-	for (int i=0;i<nPairs;i++)
+	for (int i = 0; i < nPairs; i++)
 	{
 		int bodyIndexA = hostPairs[i].x;
 		int bodyIndexB = hostPairs[i].y;
@@ -2821,84 +2660,73 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 		if (hostCollidables[collidableIndexA].m_shapeType == SHAPE_SPHERE &&
 			hostCollidables[collidableIndexB].m_shapeType == SHAPE_CONVEX_HULL)
 		{
-			computeContactSphereConvex(i,bodyIndexA,bodyIndexB,collidableIndexA,collidableIndexB,&hostBodyBuf[0],
-				&hostCollidables[0],&hostConvexData[0],&hostVertices[0],&hostIndices[0],&hostFaces[0],&hostContacts[0],nContacts,maxContactCapacity);
+			computeContactSphereConvex(i, bodyIndexA, bodyIndexB, collidableIndexA, collidableIndexB, &hostBodyBuf[0],
+									   &hostCollidables[0], &hostConvexData[0], &hostVertices[0], &hostIndices[0], &hostFaces[0], &hostContacts[0], nContacts, maxContactCapacity);
 		}
 
 		if (hostCollidables[collidableIndexA].m_shapeType == SHAPE_CONVEX_HULL &&
 			hostCollidables[collidableIndexB].m_shapeType == SHAPE_SPHERE)
 		{
-			computeContactSphereConvex(i,bodyIndexB,bodyIndexA,collidableIndexB,collidableIndexA,&hostBodyBuf[0],
-				&hostCollidables[0],&hostConvexData[0],&hostVertices[0],&hostIndices[0],&hostFaces[0],&hostContacts[0],nContacts,maxContactCapacity);
+			computeContactSphereConvex(i, bodyIndexB, bodyIndexA, collidableIndexB, collidableIndexA, &hostBodyBuf[0],
+									   &hostCollidables[0], &hostConvexData[0], &hostVertices[0], &hostIndices[0], &hostFaces[0], &hostContacts[0], nContacts, maxContactCapacity);
 			//printf("convex-sphere\n");
-			
 		}
 
 		if (hostCollidables[collidableIndexA].m_shapeType == SHAPE_CONVEX_HULL &&
 			hostCollidables[collidableIndexB].m_shapeType == SHAPE_PLANE)
 		{
-			computeContactPlaneConvex(i,bodyIndexB,bodyIndexA,collidableIndexB,collidableIndexA,&hostBodyBuf[0],
-			&hostCollidables[0],&hostConvexData[0],&hostVertices[0],&hostIndices[0],&hostFaces[0],&hostContacts[0],nContacts,maxContactCapacity);
-//			printf("convex-plane\n");
-			
+			computeContactPlaneConvex(i, bodyIndexB, bodyIndexA, collidableIndexB, collidableIndexA, &hostBodyBuf[0],
+									  &hostCollidables[0], &hostConvexData[0], &hostVertices[0], &hostIndices[0], &hostFaces[0], &hostContacts[0], nContacts, maxContactCapacity);
+			//			printf("convex-plane\n");
 		}
 
 		if (hostCollidables[collidableIndexA].m_shapeType == SHAPE_PLANE &&
 			hostCollidables[collidableIndexB].m_shapeType == SHAPE_CONVEX_HULL)
 		{
-			computeContactPlaneConvex(i,bodyIndexA,bodyIndexB,collidableIndexA,collidableIndexB,&hostBodyBuf[0],
-			&hostCollidables[0],&hostConvexData[0],&hostVertices[0],&hostIndices[0],&hostFaces[0],&hostContacts[0],nContacts,maxContactCapacity);
-//			printf("plane-convex\n");
-			
+			computeContactPlaneConvex(i, bodyIndexA, bodyIndexB, collidableIndexA, collidableIndexB, &hostBodyBuf[0],
+									  &hostCollidables[0], &hostConvexData[0], &hostVertices[0], &hostIndices[0], &hostFaces[0], &hostContacts[0], nContacts, maxContactCapacity);
+			//			printf("plane-convex\n");
 		}
 
-			if (hostCollidables[collidableIndexA].m_shapeType == SHAPE_COMPOUND_OF_CONVEX_HULLS &&
+		if (hostCollidables[collidableIndexA].m_shapeType == SHAPE_COMPOUND_OF_CONVEX_HULLS &&
 			hostCollidables[collidableIndexB].m_shapeType == SHAPE_COMPOUND_OF_CONVEX_HULLS)
 		{
-			computeContactCompoundCompound(i,bodyIndexB,bodyIndexA,collidableIndexB,collidableIndexA,&hostBodyBuf[0],
-			&hostCollidables[0],&hostConvexData[0],&cpuChildShapes[0], hostAabbsWorldSpace,hostAabbsLocalSpace,hostVertices,hostUniqueEdges,hostIndices,hostFaces,&hostContacts[0],
-			nContacts,maxContactCapacity,treeNodesCPU,subTreesCPU,bvhInfoCPU);	
-//			printf("convex-plane\n");
-			
+			computeContactCompoundCompound(i, bodyIndexB, bodyIndexA, collidableIndexB, collidableIndexA, &hostBodyBuf[0],
+										   &hostCollidables[0], &hostConvexData[0], &cpuChildShapes[0], hostAabbsWorldSpace, hostAabbsLocalSpace, hostVertices, hostUniqueEdges, hostIndices, hostFaces, &hostContacts[0],
+										   nContacts, maxContactCapacity, treeNodesCPU, subTreesCPU, bvhInfoCPU);
+			//			printf("convex-plane\n");
 		}
 
-
-				if (hostCollidables[collidableIndexA].m_shapeType == SHAPE_COMPOUND_OF_CONVEX_HULLS &&
+		if (hostCollidables[collidableIndexA].m_shapeType == SHAPE_COMPOUND_OF_CONVEX_HULLS &&
 			hostCollidables[collidableIndexB].m_shapeType == SHAPE_PLANE)
 		{
-			computeContactPlaneCompound(i,bodyIndexB,bodyIndexA,collidableIndexB,collidableIndexA,&hostBodyBuf[0],
-			&hostCollidables[0],&hostConvexData[0],&cpuChildShapes[0], &hostVertices[0],&hostIndices[0],&hostFaces[0],&hostContacts[0],nContacts,maxContactCapacity);
-//			printf("convex-plane\n");
-			
+			computeContactPlaneCompound(i, bodyIndexB, bodyIndexA, collidableIndexB, collidableIndexA, &hostBodyBuf[0],
+										&hostCollidables[0], &hostConvexData[0], &cpuChildShapes[0], &hostVertices[0], &hostIndices[0], &hostFaces[0], &hostContacts[0], nContacts, maxContactCapacity);
+			//			printf("convex-plane\n");
 		}
 
 		if (hostCollidables[collidableIndexA].m_shapeType == SHAPE_PLANE &&
 			hostCollidables[collidableIndexB].m_shapeType == SHAPE_COMPOUND_OF_CONVEX_HULLS)
 		{
-			computeContactPlaneCompound(i,bodyIndexA,bodyIndexB,collidableIndexA,collidableIndexB,&hostBodyBuf[0],
-			&hostCollidables[0],&hostConvexData[0],&cpuChildShapes[0],&hostVertices[0],&hostIndices[0],&hostFaces[0],&hostContacts[0],nContacts,maxContactCapacity);
-//			printf("plane-convex\n");
-			
+			computeContactPlaneCompound(i, bodyIndexA, bodyIndexB, collidableIndexA, collidableIndexB, &hostBodyBuf[0],
+										&hostCollidables[0], &hostConvexData[0], &cpuChildShapes[0], &hostVertices[0], &hostIndices[0], &hostFaces[0], &hostContacts[0], nContacts, maxContactCapacity);
+			//			printf("plane-convex\n");
 		}
 
 		if (hostCollidables[collidableIndexA].m_shapeType == SHAPE_CONVEX_HULL &&
 			hostCollidables[collidableIndexB].m_shapeType == SHAPE_CONVEX_HULL)
 		{
 			//printf("hostPairs[i].z=%d\n",hostPairs[i].z);
-			int contactIndex = computeContactConvexConvex2(           i,bodyIndexA,bodyIndexB,collidableIndexA,collidableIndexB,hostBodyBuf, hostCollidables,hostConvexData,hostVertices,hostUniqueEdges,hostIndices,hostFaces,hostContacts,nContacts,maxContactCapacity,oldHostContacts);
+			int contactIndex = computeContactConvexConvex2(i, bodyIndexA, bodyIndexB, collidableIndexA, collidableIndexB, hostBodyBuf, hostCollidables, hostConvexData, hostVertices, hostUniqueEdges, hostIndices, hostFaces, hostContacts, nContacts, maxContactCapacity, oldHostContacts);
 			//int contactIndex = computeContactConvexConvex(hostPairs,i,bodyIndexA,bodyIndexB,collidableIndexA,collidableIndexB,hostBodyBuf,hostCollidables,hostConvexData,hostVertices,hostUniqueEdges,hostIndices,hostFaces,hostContacts,nContacts,maxContactCapacity,oldHostContacts);
 
-
-			if (contactIndex>=0)
+			if (contactIndex >= 0)
 			{
-//				printf("convex convex contactIndex = %d\n",contactIndex);
+				//				printf("convex convex contactIndex = %d\n",contactIndex);
 				hostPairs[i].z = contactIndex;
 			}
-//			printf("plane-convex\n");
-			
+			//			printf("plane-convex\n");
 		}
-
-
 	}
 
 	if (hostPairs.size())
@@ -2908,81 +2736,76 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 
 	hostContacts.resize(nContacts);
 	if (nContacts)
-		{
-			
-			contactOut->copyFromHost(hostContacts);
-		} else
+	{
+		contactOut->copyFromHost(hostContacts);
+	}
+	else
 	{
 		contactOut->resize(0);
-		}
+	}
 
-		m_totalContactsOut.copyFromHostPointer(&nContacts,1,0,true);
-		//printf("(HOST) nContacts = %d\n",nContacts);
+	m_totalContactsOut.copyFromHostPointer(&nContacts, 1, 0, true);
+	//printf("(HOST) nContacts = %d\n",nContacts);
 
 #else
 
 	{
 		if (nPairs)
 		{
-			m_totalContactsOut.copyFromHostPointer(&nContacts,1,0,true);
+			m_totalContactsOut.copyFromHostPointer(&nContacts, 1, 0, true);
 
 			B3_PROFILE("primitiveContactsKernel");
 			b3BufferInfoCL bInfo[] = {
-				b3BufferInfoCL( pairs->getBufferCL(), true ), 
-				b3BufferInfoCL( bodyBuf->getBufferCL(),true), 
-				b3BufferInfoCL( gpuCollidables.getBufferCL(),true), 
-				b3BufferInfoCL( convexData.getBufferCL(),true),
-				b3BufferInfoCL( gpuVertices.getBufferCL(),true),
-				b3BufferInfoCL( gpuUniqueEdges.getBufferCL(),true),
-				b3BufferInfoCL( gpuFaces.getBufferCL(),true),
-				b3BufferInfoCL( gpuIndices.getBufferCL(),true),
-				b3BufferInfoCL( contactOut->getBufferCL()),
-				b3BufferInfoCL( m_totalContactsOut.getBufferCL())	
-			};
-			
-			b3LauncherCL launcher(m_queue, m_primitiveContactsKernel,"m_primitiveContactsKernel");
-			launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-			launcher.setConst( nPairs  );
+				b3BufferInfoCL(pairs->getBufferCL(), true),
+				b3BufferInfoCL(bodyBuf->getBufferCL(), true),
+				b3BufferInfoCL(gpuCollidables.getBufferCL(), true),
+				b3BufferInfoCL(convexData.getBufferCL(), true),
+				b3BufferInfoCL(gpuVertices.getBufferCL(), true),
+				b3BufferInfoCL(gpuUniqueEdges.getBufferCL(), true),
+				b3BufferInfoCL(gpuFaces.getBufferCL(), true),
+				b3BufferInfoCL(gpuIndices.getBufferCL(), true),
+				b3BufferInfoCL(contactOut->getBufferCL()),
+				b3BufferInfoCL(m_totalContactsOut.getBufferCL())};
+
+			b3LauncherCL launcher(m_queue, m_primitiveContactsKernel, "m_primitiveContactsKernel");
+			launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+			launcher.setConst(nPairs);
 			launcher.setConst(maxContactCapacity);
 			int num = nPairs;
-			launcher.launch1D( num);
+			launcher.launch1D(num);
 			clFinish(m_queue);
-		
+
 			nContacts = m_totalContactsOut.at(0);
 			contactOut->resize(nContacts);
 		}
 	}
 
-	
-#endif//CHECK_ON_HOST
-	
+#endif  //CHECK_ON_HOST
+
 	B3_PROFILE("computeConvexConvexContactsGPUSAT");
-   // printf("nContacts = %d\n",nContacts);
-    
-	
+	// printf("nContacts = %d\n",nContacts);
+
 	m_sepNormals.resize(nPairs);
 	m_hasSeparatingNormals.resize(nPairs);
-	
-	int concaveCapacity=maxTriConvexPairCapacity;
+
+	int concaveCapacity = maxTriConvexPairCapacity;
 	m_concaveSepNormals.resize(concaveCapacity);
 	m_concaveHasSeparatingNormals.resize(concaveCapacity);
 	m_numConcavePairsOut.resize(0);
 	m_numConcavePairsOut.push_back(0);
 
-	
 	m_gpuCompoundPairs.resize(compoundPairCapacity);
 
 	m_gpuCompoundSepNormals.resize(compoundPairCapacity);
-	
-	
+
 	m_gpuHasCompoundSepNormals.resize(compoundPairCapacity);
-	
+
 	m_numCompoundPairsOut.resize(0);
 	m_numCompoundPairsOut.push_back(0);
 
 	int numCompoundPairs = 0;
 
-	int numConcavePairs =0;
+	int numConcavePairs = 0;
 
 	{
 		clFinish(m_queue);
@@ -2991,33 +2814,30 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 			m_dmins.resize(nPairs);
 			if (splitSearchSepAxisConvex)
 			{
-					
-
 				if (useMprGpu)
 				{
 					nContacts = m_totalContactsOut.at(0);
 					{
 						B3_PROFILE("mprPenetrationKernel");
-						b3BufferInfoCL bInfo[] = { 
-							b3BufferInfoCL( pairs->getBufferCL(), true ), 
-							b3BufferInfoCL( bodyBuf->getBufferCL(),true), 
-							b3BufferInfoCL( gpuCollidables.getBufferCL(),true), 
-							b3BufferInfoCL( convexData.getBufferCL(),true),
-							b3BufferInfoCL( gpuVertices.getBufferCL(),true),
-							b3BufferInfoCL( m_sepNormals.getBufferCL()),
-							b3BufferInfoCL( m_hasSeparatingNormals.getBufferCL()),
-							b3BufferInfoCL( contactOut->getBufferCL()),
-							b3BufferInfoCL( m_totalContactsOut.getBufferCL())
-						};
+						b3BufferInfoCL bInfo[] = {
+							b3BufferInfoCL(pairs->getBufferCL(), true),
+							b3BufferInfoCL(bodyBuf->getBufferCL(), true),
+							b3BufferInfoCL(gpuCollidables.getBufferCL(), true),
+							b3BufferInfoCL(convexData.getBufferCL(), true),
+							b3BufferInfoCL(gpuVertices.getBufferCL(), true),
+							b3BufferInfoCL(m_sepNormals.getBufferCL()),
+							b3BufferInfoCL(m_hasSeparatingNormals.getBufferCL()),
+							b3BufferInfoCL(contactOut->getBufferCL()),
+							b3BufferInfoCL(m_totalContactsOut.getBufferCL())};
 
-						b3LauncherCL launcher(m_queue, m_mprPenetrationKernel,"mprPenetrationKernel");
-						launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
+						b3LauncherCL launcher(m_queue, m_mprPenetrationKernel, "mprPenetrationKernel");
+						launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
 
 						launcher.setConst(maxContactCapacity);
-						launcher.setConst( nPairs  );
+						launcher.setConst(nPairs);
 
 						int num = nPairs;
-						launcher.launch1D( num);
+						launcher.launch1D(num);
 						clFinish(m_queue);
 						/*
 						b3AlignedObjectArray<int>hostHasSepAxis;
@@ -3027,173 +2847,160 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 						*/
 						nContacts = m_totalContactsOut.at(0);
 						contactOut->resize(nContacts);
-					//	printf("nContacts (after mprPenetrationKernel) = %d\n",nContacts);
-						if (nContacts>maxContactCapacity)
+						//	printf("nContacts (after mprPenetrationKernel) = %d\n",nContacts);
+						if (nContacts > maxContactCapacity)
 						{
-                
 							b3Error("Error: contacts exceeds capacity (%d/%d)\n", nContacts, maxContactCapacity);
 							nContacts = maxContactCapacity;
 						}
-
 					}
 				}
-				
+
 				if (1)
 				{
-
 					if (1)
 					{
-					{
-						B3_PROFILE("findSeparatingAxisVertexFaceKernel");
-						b3BufferInfoCL bInfo[] = { 
-							b3BufferInfoCL( pairs->getBufferCL(), true ), 
-							b3BufferInfoCL( bodyBuf->getBufferCL(),true), 
-							b3BufferInfoCL( gpuCollidables.getBufferCL(),true), 
-							b3BufferInfoCL( convexData.getBufferCL(),true),
-							b3BufferInfoCL( gpuVertices.getBufferCL(),true),
-							b3BufferInfoCL( gpuUniqueEdges.getBufferCL(),true),
-							b3BufferInfoCL( gpuFaces.getBufferCL(),true),
-							b3BufferInfoCL( gpuIndices.getBufferCL(),true),
-							b3BufferInfoCL( clAabbsWorldSpace.getBufferCL(),true),
-							b3BufferInfoCL( m_sepNormals.getBufferCL()),
-							b3BufferInfoCL( m_hasSeparatingNormals.getBufferCL()),
-							b3BufferInfoCL( m_dmins.getBufferCL())
-						};
+						{
+							B3_PROFILE("findSeparatingAxisVertexFaceKernel");
+							b3BufferInfoCL bInfo[] = {
+								b3BufferInfoCL(pairs->getBufferCL(), true),
+								b3BufferInfoCL(bodyBuf->getBufferCL(), true),
+								b3BufferInfoCL(gpuCollidables.getBufferCL(), true),
+								b3BufferInfoCL(convexData.getBufferCL(), true),
+								b3BufferInfoCL(gpuVertices.getBufferCL(), true),
+								b3BufferInfoCL(gpuUniqueEdges.getBufferCL(), true),
+								b3BufferInfoCL(gpuFaces.getBufferCL(), true),
+								b3BufferInfoCL(gpuIndices.getBufferCL(), true),
+								b3BufferInfoCL(clAabbsWorldSpace.getBufferCL(), true),
+								b3BufferInfoCL(m_sepNormals.getBufferCL()),
+								b3BufferInfoCL(m_hasSeparatingNormals.getBufferCL()),
+								b3BufferInfoCL(m_dmins.getBufferCL())};
 
-						b3LauncherCL launcher(m_queue, m_findSeparatingAxisVertexFaceKernel,"findSeparatingAxisVertexFaceKernel");
-						launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-						launcher.setConst( nPairs  );
+							b3LauncherCL launcher(m_queue, m_findSeparatingAxisVertexFaceKernel, "findSeparatingAxisVertexFaceKernel");
+							launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+							launcher.setConst(nPairs);
 
-						int num = nPairs;
-						launcher.launch1D( num);
-						clFinish(m_queue);
-					}
+							int num = nPairs;
+							launcher.launch1D(num);
+							clFinish(m_queue);
+						}
 
+						int numDirections = sizeof(unitSphere162) / sizeof(b3Vector3);
 
-					int numDirections = sizeof(unitSphere162)/sizeof(b3Vector3);
-					
-					{
-						B3_PROFILE("findSeparatingAxisEdgeEdgeKernel");
-						b3BufferInfoCL bInfo[] = { 
-							b3BufferInfoCL( pairs->getBufferCL(), true ), 
-							b3BufferInfoCL( bodyBuf->getBufferCL(),true), 
-							b3BufferInfoCL( gpuCollidables.getBufferCL(),true), 
-							b3BufferInfoCL( convexData.getBufferCL(),true),
-							b3BufferInfoCL( gpuVertices.getBufferCL(),true),
-							b3BufferInfoCL( gpuUniqueEdges.getBufferCL(),true),
-							b3BufferInfoCL( gpuFaces.getBufferCL(),true),
-							b3BufferInfoCL( gpuIndices.getBufferCL(),true),
-							b3BufferInfoCL( clAabbsWorldSpace.getBufferCL(),true),
-							b3BufferInfoCL( m_sepNormals.getBufferCL()),
-							b3BufferInfoCL( m_hasSeparatingNormals.getBufferCL()),
-							b3BufferInfoCL( m_dmins.getBufferCL()),
-							b3BufferInfoCL( m_unitSphereDirections.getBufferCL(),true)
+						{
+							B3_PROFILE("findSeparatingAxisEdgeEdgeKernel");
+							b3BufferInfoCL bInfo[] = {
+								b3BufferInfoCL(pairs->getBufferCL(), true),
+								b3BufferInfoCL(bodyBuf->getBufferCL(), true),
+								b3BufferInfoCL(gpuCollidables.getBufferCL(), true),
+								b3BufferInfoCL(convexData.getBufferCL(), true),
+								b3BufferInfoCL(gpuVertices.getBufferCL(), true),
+								b3BufferInfoCL(gpuUniqueEdges.getBufferCL(), true),
+								b3BufferInfoCL(gpuFaces.getBufferCL(), true),
+								b3BufferInfoCL(gpuIndices.getBufferCL(), true),
+								b3BufferInfoCL(clAabbsWorldSpace.getBufferCL(), true),
+								b3BufferInfoCL(m_sepNormals.getBufferCL()),
+								b3BufferInfoCL(m_hasSeparatingNormals.getBufferCL()),
+								b3BufferInfoCL(m_dmins.getBufferCL()),
+								b3BufferInfoCL(m_unitSphereDirections.getBufferCL(), true)
 
-						};
+							};
 
-						b3LauncherCL launcher(m_queue, m_findSeparatingAxisEdgeEdgeKernel,"findSeparatingAxisEdgeEdgeKernel");
-						launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-						launcher.setConst( numDirections);
-						launcher.setConst( nPairs  );
-						int num = nPairs;
-						launcher.launch1D( num);
-						clFinish(m_queue);
-
-					}
+							b3LauncherCL launcher(m_queue, m_findSeparatingAxisEdgeEdgeKernel, "findSeparatingAxisEdgeEdgeKernel");
+							launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+							launcher.setConst(numDirections);
+							launcher.setConst(nPairs);
+							int num = nPairs;
+							launcher.launch1D(num);
+							clFinish(m_queue);
+						}
 					}
 					if (useMprGpu)
 					{
 						B3_PROFILE("findSeparatingAxisUnitSphereKernel");
-						b3BufferInfoCL bInfo[] = { 
-								b3BufferInfoCL( pairs->getBufferCL(), true ), 
-								b3BufferInfoCL( bodyBuf->getBufferCL(),true), 
-								b3BufferInfoCL( gpuCollidables.getBufferCL(),true), 
-								b3BufferInfoCL( convexData.getBufferCL(),true),
-								b3BufferInfoCL( gpuVertices.getBufferCL(),true),
-								b3BufferInfoCL( m_unitSphereDirections.getBufferCL(),true),
-								b3BufferInfoCL( m_sepNormals.getBufferCL()),
-								b3BufferInfoCL( m_hasSeparatingNormals.getBufferCL()),
-								b3BufferInfoCL( m_dmins.getBufferCL())
-						};
+						b3BufferInfoCL bInfo[] = {
+							b3BufferInfoCL(pairs->getBufferCL(), true),
+							b3BufferInfoCL(bodyBuf->getBufferCL(), true),
+							b3BufferInfoCL(gpuCollidables.getBufferCL(), true),
+							b3BufferInfoCL(convexData.getBufferCL(), true),
+							b3BufferInfoCL(gpuVertices.getBufferCL(), true),
+							b3BufferInfoCL(m_unitSphereDirections.getBufferCL(), true),
+							b3BufferInfoCL(m_sepNormals.getBufferCL()),
+							b3BufferInfoCL(m_hasSeparatingNormals.getBufferCL()),
+							b3BufferInfoCL(m_dmins.getBufferCL())};
 
-						b3LauncherCL launcher(m_queue, m_findSeparatingAxisUnitSphereKernel,"findSeparatingAxisUnitSphereKernel");
-						launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-						int numDirections = sizeof(unitSphere162)/sizeof(b3Vector3);
-						launcher.setConst( numDirections);
+						b3LauncherCL launcher(m_queue, m_findSeparatingAxisUnitSphereKernel, "findSeparatingAxisUnitSphereKernel");
+						launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+						int numDirections = sizeof(unitSphere162) / sizeof(b3Vector3);
+						launcher.setConst(numDirections);
+
+						launcher.setConst(nPairs);
 
-						launcher.setConst( nPairs  );
-                                                
 						int num = nPairs;
-						launcher.launch1D( num);
+						launcher.launch1D(num);
 						clFinish(m_queue);
 					}
+				}
 			}
-				
-
-			} else
+			else
 			{
 				B3_PROFILE("findSeparatingAxisKernel");
-				b3BufferInfoCL bInfo[] = { 
-					b3BufferInfoCL( pairs->getBufferCL(), true ), 
-					b3BufferInfoCL( bodyBuf->getBufferCL(),true), 
-					b3BufferInfoCL( gpuCollidables.getBufferCL(),true), 
-					b3BufferInfoCL( convexData.getBufferCL(),true),
-					b3BufferInfoCL( gpuVertices.getBufferCL(),true),
-					b3BufferInfoCL( gpuUniqueEdges.getBufferCL(),true),
-					b3BufferInfoCL( gpuFaces.getBufferCL(),true),
-					b3BufferInfoCL( gpuIndices.getBufferCL(),true),
-					b3BufferInfoCL( clAabbsWorldSpace.getBufferCL(),true),
-					b3BufferInfoCL( m_sepNormals.getBufferCL()),
-					b3BufferInfoCL( m_hasSeparatingNormals.getBufferCL())
-				};
+				b3BufferInfoCL bInfo[] = {
+					b3BufferInfoCL(pairs->getBufferCL(), true),
+					b3BufferInfoCL(bodyBuf->getBufferCL(), true),
+					b3BufferInfoCL(gpuCollidables.getBufferCL(), true),
+					b3BufferInfoCL(convexData.getBufferCL(), true),
+					b3BufferInfoCL(gpuVertices.getBufferCL(), true),
+					b3BufferInfoCL(gpuUniqueEdges.getBufferCL(), true),
+					b3BufferInfoCL(gpuFaces.getBufferCL(), true),
+					b3BufferInfoCL(gpuIndices.getBufferCL(), true),
+					b3BufferInfoCL(clAabbsWorldSpace.getBufferCL(), true),
+					b3BufferInfoCL(m_sepNormals.getBufferCL()),
+					b3BufferInfoCL(m_hasSeparatingNormals.getBufferCL())};
 
-				b3LauncherCL launcher(m_queue, m_findSeparatingAxisKernel,"m_findSeparatingAxisKernel");
-				launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-				launcher.setConst( nPairs  );
+				b3LauncherCL launcher(m_queue, m_findSeparatingAxisKernel, "m_findSeparatingAxisKernel");
+				launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+				launcher.setConst(nPairs);
 
 				int num = nPairs;
-				launcher.launch1D( num);
+				launcher.launch1D(num);
 				clFinish(m_queue);
 			}
-			
-			
 		}
-        else
-        {
-            
+		else
+		{
 			B3_PROFILE("findSeparatingAxisKernel CPU");
-           
-            
-            b3AlignedObjectArray<b3Int4> hostPairs;
-            pairs->copyToHost(hostPairs);
-            b3AlignedObjectArray<b3RigidBodyData> hostBodyBuf;
-            bodyBuf->copyToHost(hostBodyBuf);
 
-            b3AlignedObjectArray<b3Collidable> hostCollidables;
-            gpuCollidables.copyToHost(hostCollidables);
-            
-            b3AlignedObjectArray<b3GpuChildShape> cpuChildShapes;
-            gpuChildShapes.copyToHost(cpuChildShapes);
-            
-            b3AlignedObjectArray<b3ConvexPolyhedronData> hostConvexShapeData;
-            convexData.copyToHost(hostConvexShapeData);
-            
-            b3AlignedObjectArray<b3Vector3> hostVertices;
-            gpuVertices.copyToHost(hostVertices);
-            
-            b3AlignedObjectArray<int> hostHasSepAxis;
-            hostHasSepAxis.resize(nPairs);
-            b3AlignedObjectArray<b3Vector3> hostSepAxis;
-            hostSepAxis.resize(nPairs);
-            
-            b3AlignedObjectArray<b3Vector3> hostUniqueEdges;
-            gpuUniqueEdges.copyToHost(hostUniqueEdges);
-            b3AlignedObjectArray<b3GpuFace> hostFaces;
-            gpuFaces.copyToHost(hostFaces);
-            
-            b3AlignedObjectArray<int> hostIndices;
-            gpuIndices.copyToHost(hostIndices);
-			
+			b3AlignedObjectArray<b3Int4> hostPairs;
+			pairs->copyToHost(hostPairs);
+			b3AlignedObjectArray<b3RigidBodyData> hostBodyBuf;
+			bodyBuf->copyToHost(hostBodyBuf);
+
+			b3AlignedObjectArray<b3Collidable> hostCollidables;
+			gpuCollidables.copyToHost(hostCollidables);
+
+			b3AlignedObjectArray<b3GpuChildShape> cpuChildShapes;
+			gpuChildShapes.copyToHost(cpuChildShapes);
+
+			b3AlignedObjectArray<b3ConvexPolyhedronData> hostConvexShapeData;
+			convexData.copyToHost(hostConvexShapeData);
+
+			b3AlignedObjectArray<b3Vector3> hostVertices;
+			gpuVertices.copyToHost(hostVertices);
+
+			b3AlignedObjectArray<int> hostHasSepAxis;
+			hostHasSepAxis.resize(nPairs);
+			b3AlignedObjectArray<b3Vector3> hostSepAxis;
+			hostSepAxis.resize(nPairs);
+
+			b3AlignedObjectArray<b3Vector3> hostUniqueEdges;
+			gpuUniqueEdges.copyToHost(hostUniqueEdges);
+			b3AlignedObjectArray<b3GpuFace> hostFaces;
+			gpuFaces.copyToHost(hostFaces);
+
+			b3AlignedObjectArray<int> hostIndices;
+			gpuIndices.copyToHost(hostIndices);
+
 			b3AlignedObjectArray<b3Contact4> hostContacts;
 			if (nContacts)
 			{
@@ -3201,61 +3008,56 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 			}
 			hostContacts.resize(maxContactCapacity);
 			int nGlobalContactsOut = nContacts;
-			
-            
-            for (int i=0;i<nPairs;i++)
-            {
-                
-                int bodyIndexA = hostPairs[i].x;
-                int bodyIndexB = hostPairs[i].y;
-                int collidableIndexA = hostBodyBuf[bodyIndexA].m_collidableIdx;
-                int collidableIndexB = hostBodyBuf[bodyIndexB].m_collidableIdx;
-                
-                int shapeIndexA = hostCollidables[collidableIndexA].m_shapeIndex;
-                int shapeIndexB = hostCollidables[collidableIndexB].m_shapeIndex;
-                
-                hostHasSepAxis[i] = 0;
-                
-                //once the broadphase avoids static-static pairs, we can remove this test
-                if ((hostBodyBuf[bodyIndexA].m_invMass==0) &&(hostBodyBuf[bodyIndexB].m_invMass==0))
-                {
-                    continue;
-                }
-                
-                
-                if ((hostCollidables[collidableIndexA].m_shapeType!=SHAPE_CONVEX_HULL) ||(hostCollidables[collidableIndexB].m_shapeType!=SHAPE_CONVEX_HULL))
-                {
-                    continue;
-                }
-                
-                float dmin = FLT_MAX;
-                
-                b3ConvexPolyhedronData* convexShapeA = &hostConvexShapeData[shapeIndexA];
-                b3ConvexPolyhedronData* convexShapeB = &hostConvexShapeData[shapeIndexB];
-                b3Vector3 posA = hostBodyBuf[bodyIndexA].m_pos;
-                b3Vector3 posB = hostBodyBuf[bodyIndexB].m_pos;
-                b3Quaternion ornA =hostBodyBuf[bodyIndexA].m_quat;
-                b3Quaternion ornB =hostBodyBuf[bodyIndexB].m_quat;
-				
-				
+
+			for (int i = 0; i < nPairs; i++)
+			{
+				int bodyIndexA = hostPairs[i].x;
+				int bodyIndexB = hostPairs[i].y;
+				int collidableIndexA = hostBodyBuf[bodyIndexA].m_collidableIdx;
+				int collidableIndexB = hostBodyBuf[bodyIndexB].m_collidableIdx;
+
+				int shapeIndexA = hostCollidables[collidableIndexA].m_shapeIndex;
+				int shapeIndexB = hostCollidables[collidableIndexB].m_shapeIndex;
+
+				hostHasSepAxis[i] = 0;
+
+				//once the broadphase avoids static-static pairs, we can remove this test
+				if ((hostBodyBuf[bodyIndexA].m_invMass == 0) && (hostBodyBuf[bodyIndexB].m_invMass == 0))
+				{
+					continue;
+				}
+
+				if ((hostCollidables[collidableIndexA].m_shapeType != SHAPE_CONVEX_HULL) || (hostCollidables[collidableIndexB].m_shapeType != SHAPE_CONVEX_HULL))
+				{
+					continue;
+				}
+
+				float dmin = FLT_MAX;
+
+				b3ConvexPolyhedronData* convexShapeA = &hostConvexShapeData[shapeIndexA];
+				b3ConvexPolyhedronData* convexShapeB = &hostConvexShapeData[shapeIndexB];
+				b3Vector3 posA = hostBodyBuf[bodyIndexA].m_pos;
+				b3Vector3 posB = hostBodyBuf[bodyIndexB].m_pos;
+				b3Quaternion ornA = hostBodyBuf[bodyIndexA].m_quat;
+				b3Quaternion ornB = hostBodyBuf[bodyIndexB].m_quat;
+
 				if (useGjk)
 				{
-
 					//first approximate the separating axis, to 'fail-proof' GJK+EPA or MPR
 					{
 						b3Vector3 c0local = hostConvexShapeData[shapeIndexA].m_localCenter;
 						b3Vector3 c0 = b3TransformPoint(c0local, posA, ornA);
 						b3Vector3 c1local = hostConvexShapeData[shapeIndexB].m_localCenter;
-						b3Vector3 c1 = b3TransformPoint(c1local,posB,ornB);
+						b3Vector3 c1 = b3TransformPoint(c1local, posB, ornB);
 						b3Vector3 DeltaC2 = c0 - c1;
-                
+
 						b3Vector3 sepAxis;
-                
+
 						bool hasSepAxisA = b3FindSeparatingAxis(convexShapeA, convexShapeB, posA, ornA, posB, ornB, DeltaC2,
-							&hostVertices.at(0), &hostUniqueEdges.at(0), &hostFaces.at(0), &hostIndices.at(0),
-							&hostVertices.at(0), &hostUniqueEdges.at(0), &hostFaces.at(0), &hostIndices.at(0),
-											 &sepAxis, &dmin);
-                
+																&hostVertices.at(0), &hostUniqueEdges.at(0), &hostFaces.at(0), &hostIndices.at(0),
+																&hostVertices.at(0), &hostUniqueEdges.at(0), &hostFaces.at(0), &hostIndices.at(0),
+																&sepAxis, &dmin);
+
 						if (hasSepAxisA)
 						{
 							bool hasSepAxisB = b3FindSeparatingAxis(convexShapeB, convexShapeA, posB, ornB, posA, ornA, DeltaC2,
@@ -3264,11 +3066,11 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 																	&sepAxis, &dmin);
 							if (hasSepAxisB)
 							{
-								bool hasEdgeEdge =b3FindSeparatingAxisEdgeEdge(convexShapeA, convexShapeB, posA, ornA, posB, ornB, DeltaC2,
-															 &hostVertices.at(0), &hostUniqueEdges.at(0), &hostFaces.at(0), &hostIndices.at(0),
-															 &hostVertices.at(0), &hostUniqueEdges.at(0), &hostFaces.at(0), &hostIndices.at(0),
-															 &sepAxis, &dmin,false);
-													 
+								bool hasEdgeEdge = b3FindSeparatingAxisEdgeEdge(convexShapeA, convexShapeB, posA, ornA, posB, ornB, DeltaC2,
+																				&hostVertices.at(0), &hostUniqueEdges.at(0), &hostFaces.at(0), &hostIndices.at(0),
+																				&hostVertices.at(0), &hostUniqueEdges.at(0), &hostFaces.at(0), &hostIndices.at(0),
+																				&sepAxis, &dmin, false);
+
 								if (hasEdgeEdge)
 								{
 									hostHasSepAxis[i] = 1;
@@ -3282,163 +3084,150 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 					if (hostHasSepAxis[i])
 					{
 						int pairIndex = i;
-				
+
 						bool useMpr = true;
 						if (useMpr)
 						{
-							int res=0;
+							int res = 0;
 							float depth = 0.f;
-							b3Vector3 sepAxis2 = b3MakeVector3(1,0,0);
-							b3Vector3 resultPointOnBWorld = b3MakeVector3(0,0,0);
+							b3Vector3 sepAxis2 = b3MakeVector3(1, 0, 0);
+							b3Vector3 resultPointOnBWorld = b3MakeVector3(0, 0, 0);
 
-						float depthOut;
-						b3Vector3 dirOut;
-						b3Vector3 posOut;
-						
+							float depthOut;
+							b3Vector3 dirOut;
+							b3Vector3 posOut;
 
-						//res = b3MprPenetration(bodyIndexA,bodyIndexB,hostBodyBuf,hostConvexShapeData,hostCollidables,hostVertices,&mprConfig,&depthOut,&dirOut,&posOut);
-						res = b3MprPenetration(pairIndex,bodyIndexA,bodyIndexB,&hostBodyBuf[0],&hostConvexShapeData[0],&hostCollidables[0],&hostVertices[0],&hostSepAxis[0],&hostHasSepAxis[0],&depthOut,&dirOut,&posOut);
-						depth = depthOut;
-						sepAxis2 =  b3MakeVector3(-dirOut.x,-dirOut.y,-dirOut.z);
-						resultPointOnBWorld = posOut;
-						//hostHasSepAxis[i] = 0;
+							//res = b3MprPenetration(bodyIndexA,bodyIndexB,hostBodyBuf,hostConvexShapeData,hostCollidables,hostVertices,&mprConfig,&depthOut,&dirOut,&posOut);
+							res = b3MprPenetration(pairIndex, bodyIndexA, bodyIndexB, &hostBodyBuf[0], &hostConvexShapeData[0], &hostCollidables[0], &hostVertices[0], &hostSepAxis[0], &hostHasSepAxis[0], &depthOut, &dirOut, &posOut);
+							depth = depthOut;
+							sepAxis2 = b3MakeVector3(-dirOut.x, -dirOut.y, -dirOut.z);
+							resultPointOnBWorld = posOut;
+							//hostHasSepAxis[i] = 0;
 
-
-						if (res==0)
-						{
-							//add point?
-							//printf("depth = %f\n",depth);
-							//printf("normal = %f,%f,%f\n",dir.v[0],dir.v[1],dir.v[2]);
-							//qprintf("pos = %f,%f,%f\n",pos.v[0],pos.v[1],pos.v[2]);
-						
-							
-
-							float dist=0.f;
-
-							const b3ConvexPolyhedronData& hullA = hostConvexShapeData[hostCollidables[hostBodyBuf[bodyIndexA].m_collidableIdx].m_shapeIndex];
-							const b3ConvexPolyhedronData& hullB = hostConvexShapeData[hostCollidables[hostBodyBuf[bodyIndexB].m_collidableIdx].m_shapeIndex];
-
-							if(b3TestSepAxis( &hullA, &hullB, posA,ornA,posB,ornB,&sepAxis2, &hostVertices[0], &hostVertices[0],&dist))
+							if (res == 0)
 							{
-								if (depth > dist)
+								//add point?
+								//printf("depth = %f\n",depth);
+								//printf("normal = %f,%f,%f\n",dir.v[0],dir.v[1],dir.v[2]);
+								//qprintf("pos = %f,%f,%f\n",pos.v[0],pos.v[1],pos.v[2]);
+
+								float dist = 0.f;
+
+								const b3ConvexPolyhedronData& hullA = hostConvexShapeData[hostCollidables[hostBodyBuf[bodyIndexA].m_collidableIdx].m_shapeIndex];
+								const b3ConvexPolyhedronData& hullB = hostConvexShapeData[hostCollidables[hostBodyBuf[bodyIndexB].m_collidableIdx].m_shapeIndex];
+
+								if (b3TestSepAxis(&hullA, &hullB, posA, ornA, posB, ornB, &sepAxis2, &hostVertices[0], &hostVertices[0], &dist))
 								{
-									float diff = depth - dist;
-									
-									static float maxdiff = 0.f;
-									if (maxdiff < diff)
+									if (depth > dist)
 									{
-										maxdiff = diff;
-										printf("maxdiff = %20.10f\n",maxdiff);
+										float diff = depth - dist;
+
+										static float maxdiff = 0.f;
+										if (maxdiff < diff)
+										{
+											maxdiff = diff;
+											printf("maxdiff = %20.10f\n", maxdiff);
+										}
 									}
 								}
-							}
-							if (depth > dmin)
-							{
-								b3Vector3 oldAxis = hostSepAxis[i];
-								depth = dmin;
-								sepAxis2 = oldAxis;
-							}
-
-							
-
-							if(b3TestSepAxis( &hullA, &hullB, posA,ornA,posB,ornB,&sepAxis2, &hostVertices[0], &hostVertices[0],&dist))
-							{
-								if (depth > dist)
+								if (depth > dmin)
 								{
-									float diff = depth - dist;
-									//printf("?diff  = %f\n",diff );
-									static float maxdiff = 0.f;
-									if (maxdiff < diff)
+									b3Vector3 oldAxis = hostSepAxis[i];
+									depth = dmin;
+									sepAxis2 = oldAxis;
+								}
+
+								if (b3TestSepAxis(&hullA, &hullB, posA, ornA, posB, ornB, &sepAxis2, &hostVertices[0], &hostVertices[0], &dist))
+								{
+									if (depth > dist)
 									{
-										maxdiff = diff;
-										printf("maxdiff = %20.10f\n",maxdiff);
+										float diff = depth - dist;
+										//printf("?diff  = %f\n",diff );
+										static float maxdiff = 0.f;
+										if (maxdiff < diff)
+										{
+											maxdiff = diff;
+											printf("maxdiff = %20.10f\n", maxdiff);
+										}
 									}
+									//this is used for SAT
+									//hostHasSepAxis[i] = 1;
+									//hostSepAxis[i] = sepAxis2;
+
+									//add contact point
+
+									//int contactIndex = nGlobalContactsOut;
+									b3Contact4& newContact = hostContacts.at(nGlobalContactsOut);
+									nGlobalContactsOut++;
+									newContact.m_batchIdx = 0;  //i;
+									newContact.m_bodyAPtrAndSignBit = (hostBodyBuf.at(bodyIndexA).m_invMass == 0) ? -bodyIndexA : bodyIndexA;
+									newContact.m_bodyBPtrAndSignBit = (hostBodyBuf.at(bodyIndexB).m_invMass == 0) ? -bodyIndexB : bodyIndexB;
+
+									newContact.m_frictionCoeffCmp = 45874;
+									newContact.m_restituitionCoeffCmp = 0;
+
+									static float maxDepth = 0.f;
+
+									if (depth > maxDepth)
+									{
+										maxDepth = depth;
+										printf("MPR maxdepth = %f\n", maxDepth);
+									}
+
+									resultPointOnBWorld.w = -depth;
+									newContact.m_worldPosB[0] = resultPointOnBWorld;
+									//b3Vector3 resultPointOnAWorld = resultPointOnBWorld+depth*sepAxis2;
+									newContact.m_worldNormalOnB = sepAxis2;
+									newContact.m_worldNormalOnB.w = (b3Scalar)1;
 								}
-								//this is used for SAT
-								//hostHasSepAxis[i] = 1;
-								//hostSepAxis[i] = sepAxis2;
-
-								//add contact point
-
-								//int contactIndex = nGlobalContactsOut;
-								b3Contact4& newContact = hostContacts.at(nGlobalContactsOut);
-								nGlobalContactsOut++;
-								newContact.m_batchIdx = 0;//i;
-								newContact.m_bodyAPtrAndSignBit = (hostBodyBuf.at(bodyIndexA).m_invMass==0)? -bodyIndexA:bodyIndexA;
-								newContact.m_bodyBPtrAndSignBit = (hostBodyBuf.at(bodyIndexB).m_invMass==0)? -bodyIndexB:bodyIndexB;
-
-								newContact.m_frictionCoeffCmp = 45874;
-								newContact.m_restituitionCoeffCmp = 0;
-						
-								
-								static float maxDepth = 0.f;
-							
-								if (depth > maxDepth)
+								else
 								{
-									maxDepth  = depth;
-									printf("MPR maxdepth = %f\n",maxDepth );
-							
+									printf("rejected\n");
 								}
-							
-
-								resultPointOnBWorld.w = -depth;
-								newContact.m_worldPosB[0] = resultPointOnBWorld;
-								//b3Vector3 resultPointOnAWorld = resultPointOnBWorld+depth*sepAxis2;
-								newContact.m_worldNormalOnB = sepAxis2;
-								newContact.m_worldNormalOnB.w = (b3Scalar)1;
-							} else
-							{
-								printf("rejected\n");
 							}
-
-			
 						}
-						} else
+						else
 						{
-
-			
-				
-			//int contactIndex = computeContactConvexConvex2(           i,bodyIndexA,bodyIndexB,collidableIndexA,collidableIndexB,hostBodyBuf, hostCollidables,hostConvexData,hostVertices,hostUniqueEdges,hostIndices,hostFaces,hostContacts,nContacts,maxContactCapacity,oldHostContacts);
-							b3AlignedObjectArray<b3Contact4> oldHostContacts;	
+							//int contactIndex = computeContactConvexConvex2(           i,bodyIndexA,bodyIndexB,collidableIndexA,collidableIndexB,hostBodyBuf, hostCollidables,hostConvexData,hostVertices,hostUniqueEdges,hostIndices,hostFaces,hostContacts,nContacts,maxContactCapacity,oldHostContacts);
+							b3AlignedObjectArray<b3Contact4> oldHostContacts;
 							int result;
-							result = computeContactConvexConvex2( //hostPairs,
-													   pairIndex,
-													bodyIndexA, bodyIndexB,
-													   collidableIndexA, collidableIndexB,
-													   hostBodyBuf,
-													   hostCollidables,
-													   hostConvexShapeData,
-													   hostVertices,
-													   hostUniqueEdges,
-													   hostIndices,
-													   hostFaces,
-													   hostContacts,
-													   nGlobalContactsOut,
-														maxContactCapacity,
-														oldHostContacts
-														//hostHasSepAxis,
-														//hostSepAxis
-														
-																);
-						}//mpr
-					}//hostHasSepAxis[i] = 1;
-					
-				} else
+							result = computeContactConvexConvex2(  //hostPairs,
+								pairIndex,
+								bodyIndexA, bodyIndexB,
+								collidableIndexA, collidableIndexB,
+								hostBodyBuf,
+								hostCollidables,
+								hostConvexShapeData,
+								hostVertices,
+								hostUniqueEdges,
+								hostIndices,
+								hostFaces,
+								hostContacts,
+								nGlobalContactsOut,
+								maxContactCapacity,
+								oldHostContacts
+								//hostHasSepAxis,
+								//hostSepAxis
+
+							);
+						}  //mpr
+					}      //hostHasSepAxis[i] = 1;
+				}
+				else
 				{
-				
 					b3Vector3 c0local = hostConvexShapeData[shapeIndexA].m_localCenter;
 					b3Vector3 c0 = b3TransformPoint(c0local, posA, ornA);
 					b3Vector3 c1local = hostConvexShapeData[shapeIndexB].m_localCenter;
-					b3Vector3 c1 = b3TransformPoint(c1local,posB,ornB);
+					b3Vector3 c1 = b3TransformPoint(c1local, posB, ornB);
 					b3Vector3 DeltaC2 = c0 - c1;
-                
+
 					b3Vector3 sepAxis;
-                
+
 					bool hasSepAxisA = b3FindSeparatingAxis(convexShapeA, convexShapeB, posA, ornA, posB, ornB, DeltaC2,
-						&hostVertices.at(0), &hostUniqueEdges.at(0), &hostFaces.at(0), &hostIndices.at(0),
-						&hostVertices.at(0), &hostUniqueEdges.at(0), &hostFaces.at(0), &hostIndices.at(0),
-										 &sepAxis, &dmin);
-                
+															&hostVertices.at(0), &hostUniqueEdges.at(0), &hostFaces.at(0), &hostIndices.at(0),
+															&hostVertices.at(0), &hostUniqueEdges.at(0), &hostFaces.at(0), &hostIndices.at(0),
+															&sepAxis, &dmin);
+
 					if (hasSepAxisA)
 					{
 						bool hasSepAxisB = b3FindSeparatingAxis(convexShapeB, convexShapeA, posB, ornB, posA, ornA, DeltaC2,
@@ -3447,11 +3236,11 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 																&sepAxis, &dmin);
 						if (hasSepAxisB)
 						{
-							bool hasEdgeEdge =b3FindSeparatingAxisEdgeEdge(convexShapeA, convexShapeB, posA, ornA, posB, ornB, DeltaC2,
-														 &hostVertices.at(0), &hostUniqueEdges.at(0), &hostFaces.at(0), &hostIndices.at(0),
-														 &hostVertices.at(0), &hostUniqueEdges.at(0), &hostFaces.at(0), &hostIndices.at(0),
-														 &sepAxis, &dmin,true);
-													 
+							bool hasEdgeEdge = b3FindSeparatingAxisEdgeEdge(convexShapeA, convexShapeB, posA, ornA, posB, ornB, DeltaC2,
+																			&hostVertices.at(0), &hostUniqueEdges.at(0), &hostFaces.at(0), &hostIndices.at(0),
+																			&hostVertices.at(0), &hostUniqueEdges.at(0), &hostFaces.at(0), &hostIndices.at(0),
+																			&sepAxis, &dmin, true);
+
 							if (hasEdgeEdge)
 							{
 								hostHasSepAxis[i] = 1;
@@ -3460,21 +3249,21 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 						}
 					}
 				}
-            }
-            
-			if (useGjkContacts)//nGlobalContactsOut>0)
+			}
+
+			if (useGjkContacts)  //nGlobalContactsOut>0)
 			{
 				//printf("nGlobalContactsOut=%d\n",nGlobalContactsOut);
 				nContacts = nGlobalContactsOut;
 				contactOut->copyFromHost(hostContacts);
-	
-				m_totalContactsOut.copyFromHostPointer(&nContacts,1,0,true);
+
+				m_totalContactsOut.copyFromHostPointer(&nContacts, 1, 0, true);
 			}
-            
-            m_hasSeparatingNormals.copyFromHost(hostHasSepAxis);
-            m_sepNormals.copyFromHost(hostSepAxis);
-        
-            /*
+
+			m_hasSeparatingNormals.copyFromHost(hostHasSepAxis);
+			m_sepNormals.copyFromHost(hostSepAxis);
+
+			/*
              //double-check results from GPU (comment-out the 'else' so both paths are executed
             b3AlignedObjectArray<int> checkHasSepAxis;
             m_hasSeparatingNormals.copyToHost(checkHasSepAxis);
@@ -3491,352 +3280,314 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
             //m_hasSeparatingNormals.copyFromHost(hostHasSepAxis);
             //    m_sepNormals.copyFromHost(hostSepAxis);
             */
-        }
-        
-        
-        numCompoundPairs = m_numCompoundPairsOut.at(0);
-        bool useGpuFindCompoundPairs=true;
-        if (useGpuFindCompoundPairs)
-        {
-            B3_PROFILE("findCompoundPairsKernel");
-            b3BufferInfoCL bInfo[] = 
-            { 
-                b3BufferInfoCL( pairs->getBufferCL(), true ), 
-                b3BufferInfoCL( bodyBuf->getBufferCL(),true), 
-                b3BufferInfoCL( gpuCollidables.getBufferCL(),true), 
-                b3BufferInfoCL( convexData.getBufferCL(),true),
-                b3BufferInfoCL( gpuVertices.getBufferCL(),true),
-                b3BufferInfoCL( gpuUniqueEdges.getBufferCL(),true),
-                b3BufferInfoCL( gpuFaces.getBufferCL(),true),
-                b3BufferInfoCL( gpuIndices.getBufferCL(),true),
-                b3BufferInfoCL( clAabbsLocalSpace.getBufferCL(),true),
-                b3BufferInfoCL( gpuChildShapes.getBufferCL(),true),
-                b3BufferInfoCL( m_gpuCompoundPairs.getBufferCL()),
-                b3BufferInfoCL( m_numCompoundPairsOut.getBufferCL()),
-                b3BufferInfoCL(subTreesGPU->getBufferCL()),
-                b3BufferInfoCL(treeNodesGPU->getBufferCL()),
-                b3BufferInfoCL(bvhInfo->getBufferCL())
-            };
+		}
 
-            b3LauncherCL launcher(m_queue, m_findCompoundPairsKernel,"m_findCompoundPairsKernel");
-            launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-            launcher.setConst( nPairs  );
-            launcher.setConst( compoundPairCapacity);
+		numCompoundPairs = m_numCompoundPairsOut.at(0);
+		bool useGpuFindCompoundPairs = true;
+		if (useGpuFindCompoundPairs)
+		{
+			B3_PROFILE("findCompoundPairsKernel");
+			b3BufferInfoCL bInfo[] =
+				{
+					b3BufferInfoCL(pairs->getBufferCL(), true),
+					b3BufferInfoCL(bodyBuf->getBufferCL(), true),
+					b3BufferInfoCL(gpuCollidables.getBufferCL(), true),
+					b3BufferInfoCL(convexData.getBufferCL(), true),
+					b3BufferInfoCL(gpuVertices.getBufferCL(), true),
+					b3BufferInfoCL(gpuUniqueEdges.getBufferCL(), true),
+					b3BufferInfoCL(gpuFaces.getBufferCL(), true),
+					b3BufferInfoCL(gpuIndices.getBufferCL(), true),
+					b3BufferInfoCL(clAabbsLocalSpace.getBufferCL(), true),
+					b3BufferInfoCL(gpuChildShapes.getBufferCL(), true),
+					b3BufferInfoCL(m_gpuCompoundPairs.getBufferCL()),
+					b3BufferInfoCL(m_numCompoundPairsOut.getBufferCL()),
+					b3BufferInfoCL(subTreesGPU->getBufferCL()),
+					b3BufferInfoCL(treeNodesGPU->getBufferCL()),
+					b3BufferInfoCL(bvhInfo->getBufferCL())};
 
-            int num = nPairs;
-            launcher.launch1D( num);
-            clFinish(m_queue);
+			b3LauncherCL launcher(m_queue, m_findCompoundPairsKernel, "m_findCompoundPairsKernel");
+			launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+			launcher.setConst(nPairs);
+			launcher.setConst(compoundPairCapacity);
 
-            numCompoundPairs = m_numCompoundPairsOut.at(0);
-            //printf("numCompoundPairs =%d\n",numCompoundPairs );
-            if (numCompoundPairs)
-            {
-                //printf("numCompoundPairs=%d\n",numCompoundPairs);
-            }
-            
+			int num = nPairs;
+			launcher.launch1D(num);
+			clFinish(m_queue);
 
-        } else
-        {
+			numCompoundPairs = m_numCompoundPairsOut.at(0);
+			//printf("numCompoundPairs =%d\n",numCompoundPairs );
+			if (numCompoundPairs)
+			{
+				//printf("numCompoundPairs=%d\n",numCompoundPairs);
+			}
+		}
+		else
+		{
+			b3AlignedObjectArray<b3QuantizedBvhNode> treeNodesCPU;
+			treeNodesGPU->copyToHost(treeNodesCPU);
 
+			b3AlignedObjectArray<b3BvhSubtreeInfo> subTreesCPU;
+			subTreesGPU->copyToHost(subTreesCPU);
 
-            b3AlignedObjectArray<b3QuantizedBvhNode>	treeNodesCPU;
-            treeNodesGPU->copyToHost(treeNodesCPU);
+			b3AlignedObjectArray<b3BvhInfo> bvhInfoCPU;
+			bvhInfo->copyToHost(bvhInfoCPU);
 
-            b3AlignedObjectArray<b3BvhSubtreeInfo>	subTreesCPU;
-            subTreesGPU->copyToHost(subTreesCPU);
+			b3AlignedObjectArray<b3Aabb> hostAabbsWorldSpace;
+			clAabbsWorldSpace.copyToHost(hostAabbsWorldSpace);
 
-            b3AlignedObjectArray<b3BvhInfo>	bvhInfoCPU;
-            bvhInfo->copyToHost(bvhInfoCPU);
+			b3AlignedObjectArray<b3Aabb> hostAabbsLocalSpace;
+			clAabbsLocalSpace.copyToHost(hostAabbsLocalSpace);
 
-            b3AlignedObjectArray<b3Aabb> hostAabbsWorldSpace;
-            clAabbsWorldSpace.copyToHost(hostAabbsWorldSpace);
+			b3AlignedObjectArray<b3Int4> hostPairs;
+			pairs->copyToHost(hostPairs);
 
-            b3AlignedObjectArray<b3Aabb> hostAabbsLocalSpace;
-            clAabbsLocalSpace.copyToHost(hostAabbsLocalSpace);
+			b3AlignedObjectArray<b3RigidBodyData> hostBodyBuf;
+			bodyBuf->copyToHost(hostBodyBuf);
 
-            b3AlignedObjectArray<b3Int4> hostPairs;
-            pairs->copyToHost(hostPairs);
+			b3AlignedObjectArray<b3Int4> cpuCompoundPairsOut;
+			cpuCompoundPairsOut.resize(compoundPairCapacity);
 
-            b3AlignedObjectArray<b3RigidBodyData> hostBodyBuf;
-            bodyBuf->copyToHost(hostBodyBuf);
+			b3AlignedObjectArray<b3Collidable> hostCollidables;
+			gpuCollidables.copyToHost(hostCollidables);
 
+			b3AlignedObjectArray<b3GpuChildShape> cpuChildShapes;
+			gpuChildShapes.copyToHost(cpuChildShapes);
 
-            b3AlignedObjectArray<b3Int4> cpuCompoundPairsOut;
-            cpuCompoundPairsOut.resize(compoundPairCapacity);
+			b3AlignedObjectArray<b3ConvexPolyhedronData> hostConvexData;
+			convexData.copyToHost(hostConvexData);
 
-            b3AlignedObjectArray<b3Collidable> hostCollidables;
-            gpuCollidables.copyToHost(hostCollidables);
+			b3AlignedObjectArray<b3Vector3> hostVertices;
+			gpuVertices.copyToHost(hostVertices);
 
-            b3AlignedObjectArray<b3GpuChildShape> cpuChildShapes;
-            gpuChildShapes.copyToHost(cpuChildShapes);
-
-            b3AlignedObjectArray<b3ConvexPolyhedronData> hostConvexData;
-            convexData.copyToHost(hostConvexData);
-
-            b3AlignedObjectArray<b3Vector3> hostVertices;
-            gpuVertices.copyToHost(hostVertices);
-
-
-
-
-            for (int pairIndex=0;pairIndex<nPairs;pairIndex++)
-            {
-                int bodyIndexA = hostPairs[pairIndex].x;
-                int bodyIndexB = hostPairs[pairIndex].y;
-                int collidableIndexA = hostBodyBuf[bodyIndexA].m_collidableIdx;
-                int collidableIndexB = hostBodyBuf[bodyIndexB].m_collidableIdx;
+			for (int pairIndex = 0; pairIndex < nPairs; pairIndex++)
+			{
+				int bodyIndexA = hostPairs[pairIndex].x;
+				int bodyIndexB = hostPairs[pairIndex].y;
+				int collidableIndexA = hostBodyBuf[bodyIndexA].m_collidableIdx;
+				int collidableIndexB = hostBodyBuf[bodyIndexB].m_collidableIdx;
 				if (cpuChildShapes.size())
 				{
-                findCompoundPairsKernel( 
-                            pairIndex,
-                            bodyIndexA,
-                            bodyIndexB,
-                            collidableIndexA,
-                            collidableIndexB,
-                            &hostBodyBuf[0],
-                            &hostCollidables[0],
-                            &hostConvexData[0],
-                            hostVertices,
-                            hostAabbsWorldSpace,
-                            hostAabbsLocalSpace,
-                            &cpuChildShapes[0],
-                            &cpuCompoundPairsOut[0],
-                            &numCompoundPairs,
-                            compoundPairCapacity,
-                            treeNodesCPU,
-                            subTreesCPU,
-                            bvhInfoCPU
-                            );
+					findCompoundPairsKernel(
+						pairIndex,
+						bodyIndexA,
+						bodyIndexB,
+						collidableIndexA,
+						collidableIndexB,
+						&hostBodyBuf[0],
+						&hostCollidables[0],
+						&hostConvexData[0],
+						hostVertices,
+						hostAabbsWorldSpace,
+						hostAabbsLocalSpace,
+						&cpuChildShapes[0],
+						&cpuCompoundPairsOut[0],
+						&numCompoundPairs,
+						compoundPairCapacity,
+						treeNodesCPU,
+						subTreesCPU,
+						bvhInfoCPU);
 				}
-            }
-            
+			}
 
-			m_numCompoundPairsOut.copyFromHostPointer(&numCompoundPairs,1,0,true);
+			m_numCompoundPairsOut.copyFromHostPointer(&numCompoundPairs, 1, 0, true);
 			if (numCompoundPairs)
 			{
 				b3CompoundOverlappingPair* ptr = (b3CompoundOverlappingPair*)&cpuCompoundPairsOut[0];
-				m_gpuCompoundPairs.copyFromHostPointer(ptr,numCompoundPairs,0,true);
+				m_gpuCompoundPairs.copyFromHostPointer(ptr, numCompoundPairs, 0, true);
 			}
 			//cpuCompoundPairsOut
-            
-        }
+		}
 		if (numCompoundPairs)
 		{
-			printf("numCompoundPairs=%d\n",numCompoundPairs);
+			printf("numCompoundPairs=%d\n", numCompoundPairs);
 		}
 
-        if (numCompoundPairs > compoundPairCapacity)
-        {
-            b3Error("Exceeded compound pair capacity (%d/%d)\n", numCompoundPairs,  compoundPairCapacity);
-            numCompoundPairs = compoundPairCapacity;
-        }
+		if (numCompoundPairs > compoundPairCapacity)
+		{
+			b3Error("Exceeded compound pair capacity (%d/%d)\n", numCompoundPairs, compoundPairCapacity);
+			numCompoundPairs = compoundPairCapacity;
+		}
 
-        
+		m_gpuCompoundPairs.resize(numCompoundPairs);
+		m_gpuHasCompoundSepNormals.resize(numCompoundPairs);
+		m_gpuCompoundSepNormals.resize(numCompoundPairs);
 
-        m_gpuCompoundPairs.resize(numCompoundPairs);
-        m_gpuHasCompoundSepNormals.resize(numCompoundPairs);
-        m_gpuCompoundSepNormals.resize(numCompoundPairs);
-        
+		if (numCompoundPairs)
+		{
+			B3_PROFILE("processCompoundPairsPrimitivesKernel");
+			b3BufferInfoCL bInfo[] =
+				{
+					b3BufferInfoCL(m_gpuCompoundPairs.getBufferCL(), true),
+					b3BufferInfoCL(bodyBuf->getBufferCL(), true),
+					b3BufferInfoCL(gpuCollidables.getBufferCL(), true),
+					b3BufferInfoCL(convexData.getBufferCL(), true),
+					b3BufferInfoCL(gpuVertices.getBufferCL(), true),
+					b3BufferInfoCL(gpuUniqueEdges.getBufferCL(), true),
+					b3BufferInfoCL(gpuFaces.getBufferCL(), true),
+					b3BufferInfoCL(gpuIndices.getBufferCL(), true),
+					b3BufferInfoCL(clAabbsWorldSpace.getBufferCL(), true),
+					b3BufferInfoCL(gpuChildShapes.getBufferCL(), true),
+					b3BufferInfoCL(contactOut->getBufferCL()),
+					b3BufferInfoCL(m_totalContactsOut.getBufferCL())};
 
-        if (numCompoundPairs)
-        {
-            B3_PROFILE("processCompoundPairsPrimitivesKernel");
-            b3BufferInfoCL bInfo[] = 
-            { 
-                b3BufferInfoCL( m_gpuCompoundPairs.getBufferCL(), true ), 
-                b3BufferInfoCL( bodyBuf->getBufferCL(),true), 
-                b3BufferInfoCL( gpuCollidables.getBufferCL(),true), 
-                b3BufferInfoCL( convexData.getBufferCL(),true),
-                b3BufferInfoCL( gpuVertices.getBufferCL(),true),
-                b3BufferInfoCL( gpuUniqueEdges.getBufferCL(),true),
-                b3BufferInfoCL( gpuFaces.getBufferCL(),true),
-                b3BufferInfoCL( gpuIndices.getBufferCL(),true),
-                b3BufferInfoCL( clAabbsWorldSpace.getBufferCL(),true),
-                b3BufferInfoCL( gpuChildShapes.getBufferCL(),true),
-                b3BufferInfoCL( contactOut->getBufferCL()),
-                b3BufferInfoCL( m_totalContactsOut.getBufferCL())	
-            };
+			b3LauncherCL launcher(m_queue, m_processCompoundPairsPrimitivesKernel, "m_processCompoundPairsPrimitivesKernel");
+			launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+			launcher.setConst(numCompoundPairs);
+			launcher.setConst(maxContactCapacity);
 
-            b3LauncherCL launcher(m_queue, m_processCompoundPairsPrimitivesKernel,"m_processCompoundPairsPrimitivesKernel");
-            launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-            launcher.setConst( numCompoundPairs  );
-            launcher.setConst(maxContactCapacity);
+			int num = numCompoundPairs;
+			launcher.launch1D(num);
+			clFinish(m_queue);
+			nContacts = m_totalContactsOut.at(0);
+			//printf("nContacts (after processCompoundPairsPrimitivesKernel) = %d\n",nContacts);
+			if (nContacts > maxContactCapacity)
+			{
+				b3Error("Error: contacts exceeds capacity (%d/%d)\n", nContacts, maxContactCapacity);
+				nContacts = maxContactCapacity;
+			}
+		}
 
-            int num = numCompoundPairs;
-            launcher.launch1D( num);
-            clFinish(m_queue);
-            nContacts = m_totalContactsOut.at(0);
-            //printf("nContacts (after processCompoundPairsPrimitivesKernel) = %d\n",nContacts);
-            if (nContacts>maxContactCapacity)
-            {
-                
-                b3Error("Error: contacts exceeds capacity (%d/%d)\n", nContacts, maxContactCapacity);
-                nContacts = maxContactCapacity;
-            }
-        }
-        
+		if (numCompoundPairs)
+		{
+			B3_PROFILE("processCompoundPairsKernel");
+			b3BufferInfoCL bInfo[] =
+				{
+					b3BufferInfoCL(m_gpuCompoundPairs.getBufferCL(), true),
+					b3BufferInfoCL(bodyBuf->getBufferCL(), true),
+					b3BufferInfoCL(gpuCollidables.getBufferCL(), true),
+					b3BufferInfoCL(convexData.getBufferCL(), true),
+					b3BufferInfoCL(gpuVertices.getBufferCL(), true),
+					b3BufferInfoCL(gpuUniqueEdges.getBufferCL(), true),
+					b3BufferInfoCL(gpuFaces.getBufferCL(), true),
+					b3BufferInfoCL(gpuIndices.getBufferCL(), true),
+					b3BufferInfoCL(clAabbsWorldSpace.getBufferCL(), true),
+					b3BufferInfoCL(gpuChildShapes.getBufferCL(), true),
+					b3BufferInfoCL(m_gpuCompoundSepNormals.getBufferCL()),
+					b3BufferInfoCL(m_gpuHasCompoundSepNormals.getBufferCL())};
 
-        if (numCompoundPairs)
-        {
-            B3_PROFILE("processCompoundPairsKernel");
-            b3BufferInfoCL bInfo[] = 
-            { 
-                b3BufferInfoCL( m_gpuCompoundPairs.getBufferCL(), true ), 
-                b3BufferInfoCL( bodyBuf->getBufferCL(),true), 
-                b3BufferInfoCL( gpuCollidables.getBufferCL(),true), 
-                b3BufferInfoCL( convexData.getBufferCL(),true),
-                b3BufferInfoCL( gpuVertices.getBufferCL(),true),
-                b3BufferInfoCL( gpuUniqueEdges.getBufferCL(),true),
-                b3BufferInfoCL( gpuFaces.getBufferCL(),true),
-                b3BufferInfoCL( gpuIndices.getBufferCL(),true),
-                b3BufferInfoCL( clAabbsWorldSpace.getBufferCL(),true),
-                b3BufferInfoCL( gpuChildShapes.getBufferCL(),true),
-                b3BufferInfoCL( m_gpuCompoundSepNormals.getBufferCL()),
-                b3BufferInfoCL( m_gpuHasCompoundSepNormals.getBufferCL())
-            };
+			b3LauncherCL launcher(m_queue, m_processCompoundPairsKernel, "m_processCompoundPairsKernel");
+			launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+			launcher.setConst(numCompoundPairs);
 
-            b3LauncherCL launcher(m_queue, m_processCompoundPairsKernel,"m_processCompoundPairsKernel");
-            launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-            launcher.setConst( numCompoundPairs  );
+			int num = numCompoundPairs;
+			launcher.launch1D(num);
+			clFinish(m_queue);
+		}
 
-            int num = numCompoundPairs;
-            launcher.launch1D( num);
-            clFinish(m_queue);
-        
-        }
+		//printf("numConcave  = %d\n",numConcave);
 
-
-        //printf("numConcave  = %d\n",numConcave);
-
-    
-
-//		printf("hostNormals.size()=%d\n",hostNormals.size());
+		//		printf("hostNormals.size()=%d\n",hostNormals.size());
 		//int numPairs = pairCount.at(0);
-		
-		
-		
 	}
 	int vertexFaceCapacity = 64;
 
-
-		
 	{
 		//now perform the tree query on GPU
-			
-					
-				
-				
+
 		if (treeNodesGPU->size() && treeNodesGPU->size())
 		{
 			if (bvhTraversalKernelGPU)
 			{
-						
 				B3_PROFILE("m_bvhTraversalKernel");
-						
-						
+
 				numConcavePairs = m_numConcavePairsOut.at(0);
-						
-				b3LauncherCL launcher(m_queue, m_bvhTraversalKernel,"m_bvhTraversalKernel");
-				launcher.setBuffer( pairs->getBufferCL());
-				launcher.setBuffer(  bodyBuf->getBufferCL());
-				launcher.setBuffer( gpuCollidables.getBufferCL());
-				launcher.setBuffer( clAabbsWorldSpace.getBufferCL());
-				launcher.setBuffer( triangleConvexPairsOut.getBufferCL());
-				launcher.setBuffer( m_numConcavePairsOut.getBufferCL());
-				launcher.setBuffer( subTreesGPU->getBufferCL());
-				launcher.setBuffer( treeNodesGPU->getBufferCL());
-				launcher.setBuffer( bvhInfo->getBufferCL());
-						
-				launcher.setConst( nPairs  );
-				launcher.setConst( maxTriConvexPairCapacity);
+
+				b3LauncherCL launcher(m_queue, m_bvhTraversalKernel, "m_bvhTraversalKernel");
+				launcher.setBuffer(pairs->getBufferCL());
+				launcher.setBuffer(bodyBuf->getBufferCL());
+				launcher.setBuffer(gpuCollidables.getBufferCL());
+				launcher.setBuffer(clAabbsWorldSpace.getBufferCL());
+				launcher.setBuffer(triangleConvexPairsOut.getBufferCL());
+				launcher.setBuffer(m_numConcavePairsOut.getBufferCL());
+				launcher.setBuffer(subTreesGPU->getBufferCL());
+				launcher.setBuffer(treeNodesGPU->getBufferCL());
+				launcher.setBuffer(bvhInfo->getBufferCL());
+
+				launcher.setConst(nPairs);
+				launcher.setConst(maxTriConvexPairCapacity);
 				int num = nPairs;
-				launcher.launch1D( num);
+				launcher.launch1D(num);
 				clFinish(m_queue);
 				numConcavePairs = m_numConcavePairsOut.at(0);
-			} else
+			}
+			else
 			{
-					b3AlignedObjectArray<b3Int4> hostPairs;
-					pairs->copyToHost(hostPairs);
-					b3AlignedObjectArray<b3RigidBodyData> hostBodyBuf;
-					bodyBuf->copyToHost(hostBodyBuf);
-					b3AlignedObjectArray<b3Collidable> hostCollidables;
-					gpuCollidables.copyToHost(hostCollidables);
-					b3AlignedObjectArray<b3Aabb> hostAabbsWorldSpace;
-					clAabbsWorldSpace.copyToHost(hostAabbsWorldSpace);
+				b3AlignedObjectArray<b3Int4> hostPairs;
+				pairs->copyToHost(hostPairs);
+				b3AlignedObjectArray<b3RigidBodyData> hostBodyBuf;
+				bodyBuf->copyToHost(hostBodyBuf);
+				b3AlignedObjectArray<b3Collidable> hostCollidables;
+				gpuCollidables.copyToHost(hostCollidables);
+				b3AlignedObjectArray<b3Aabb> hostAabbsWorldSpace;
+				clAabbsWorldSpace.copyToHost(hostAabbsWorldSpace);
 
-					//int maxTriConvexPairCapacity,
-					b3AlignedObjectArray<b3Int4> triangleConvexPairsOutHost;
-					triangleConvexPairsOutHost.resize(maxTriConvexPairCapacity);
+				//int maxTriConvexPairCapacity,
+				b3AlignedObjectArray<b3Int4> triangleConvexPairsOutHost;
+				triangleConvexPairsOutHost.resize(maxTriConvexPairCapacity);
 
-					//int numTriConvexPairsOutHost=0;
-					numConcavePairs = 0;
-					//m_numConcavePairsOut
+				//int numTriConvexPairsOutHost=0;
+				numConcavePairs = 0;
+				//m_numConcavePairsOut
 
-					b3AlignedObjectArray<b3QuantizedBvhNode>	treeNodesCPU;
-					treeNodesGPU->copyToHost(treeNodesCPU);
-					b3AlignedObjectArray<b3BvhSubtreeInfo>	subTreesCPU;
-					subTreesGPU->copyToHost(subTreesCPU);
-					b3AlignedObjectArray<b3BvhInfo>	bvhInfoCPU;
-					bvhInfo->copyToHost(bvhInfoCPU);
-					//compute it...
+				b3AlignedObjectArray<b3QuantizedBvhNode> treeNodesCPU;
+				treeNodesGPU->copyToHost(treeNodesCPU);
+				b3AlignedObjectArray<b3BvhSubtreeInfo> subTreesCPU;
+				subTreesGPU->copyToHost(subTreesCPU);
+				b3AlignedObjectArray<b3BvhInfo> bvhInfoCPU;
+				bvhInfo->copyToHost(bvhInfoCPU);
+				//compute it...
 
-					volatile int hostNumConcavePairsOut=0;
+				volatile int hostNumConcavePairsOut = 0;
 
-					//
-					for (int i=0;i<nPairs;i++)
-					{
-						b3BvhTraversal( &hostPairs.at(0), 
-						&hostBodyBuf.at(0),
-						&hostCollidables.at(0),
-						&hostAabbsWorldSpace.at(0),
-						&triangleConvexPairsOutHost.at(0),
-						&hostNumConcavePairsOut,
-						&subTreesCPU.at(0),
-						&treeNodesCPU.at(0),
-						&bvhInfoCPU.at(0),
-						nPairs,
-						maxTriConvexPairCapacity,
-						i);
-					}
-					numConcavePairs = hostNumConcavePairsOut;
+				//
+				for (int i = 0; i < nPairs; i++)
+				{
+					b3BvhTraversal(&hostPairs.at(0),
+								   &hostBodyBuf.at(0),
+								   &hostCollidables.at(0),
+								   &hostAabbsWorldSpace.at(0),
+								   &triangleConvexPairsOutHost.at(0),
+								   &hostNumConcavePairsOut,
+								   &subTreesCPU.at(0),
+								   &treeNodesCPU.at(0),
+								   &bvhInfoCPU.at(0),
+								   nPairs,
+								   maxTriConvexPairCapacity,
+								   i);
+				}
+				numConcavePairs = hostNumConcavePairsOut;
 
-					if (hostNumConcavePairsOut)
-					{
-						triangleConvexPairsOutHost.resize(hostNumConcavePairsOut);
-						triangleConvexPairsOut.copyFromHost(triangleConvexPairsOutHost);
-					}
-					//
+				if (hostNumConcavePairsOut)
+				{
+					triangleConvexPairsOutHost.resize(hostNumConcavePairsOut);
+					triangleConvexPairsOut.copyFromHost(triangleConvexPairsOutHost);
+				}
+				//
 
-					m_numConcavePairsOut.resize(0);
-					m_numConcavePairsOut.push_back(numConcavePairs);
+				m_numConcavePairsOut.resize(0);
+				m_numConcavePairsOut.push_back(numConcavePairs);
 			}
 
-				//printf("numConcavePairs=%d (max = %d\n",numConcavePairs,maxTriConvexPairCapacity);
-						
+			//printf("numConcavePairs=%d (max = %d\n",numConcavePairs,maxTriConvexPairCapacity);
+
 			if (numConcavePairs > maxTriConvexPairCapacity)
 			{
 				static int exceeded_maxTriConvexPairCapacity_count = 0;
 				b3Error("Exceeded the maxTriConvexPairCapacity (found %d but max is %d, it happened %d times)\n",
-					numConcavePairs,maxTriConvexPairCapacity,exceeded_maxTriConvexPairCapacity_count++);
+						numConcavePairs, maxTriConvexPairCapacity, exceeded_maxTriConvexPairCapacity_count++);
 				numConcavePairs = maxTriConvexPairCapacity;
 			}
 			triangleConvexPairsOut.resize(numConcavePairs);
-	
+
 			if (numConcavePairs)
 			{
-
-				
-						
-	
 				clippingFacesOutGPU.resize(numConcavePairs);
 				worldNormalsAGPU.resize(numConcavePairs);
-				worldVertsA1GPU.resize(vertexFaceCapacity*(numConcavePairs));
-				worldVertsB1GPU.resize(vertexFaceCapacity*(numConcavePairs));
-
+				worldVertsA1GPU.resize(vertexFaceCapacity * (numConcavePairs));
+				worldVertsB1GPU.resize(vertexFaceCapacity * (numConcavePairs));
 
 				if (findConcaveSeparatingAxisKernelGPU)
 				{
-
 					/*
 					m_concaveHasSeparatingNormals.copyFromHost(concaveHasSeparatingNormalsCPU);
 						clippingFacesOutGPU.copyFromHost(clippingFacesOutCPU);
@@ -3846,236 +3597,213 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 					*/
 
 					//now perform a SAT test for each triangle-convex element (stored in triangleConvexPairsOut)
-                    if (splitSearchSepAxisConcave)
-                    {
-                        //printf("numConcavePairs = %d\n",numConcavePairs);
-                        m_dmins.resize(numConcavePairs);
-                        {
-                            B3_PROFILE("findConcaveSeparatingAxisVertexFaceKernel");
-                            b3BufferInfoCL bInfo[] = {
-                                b3BufferInfoCL( triangleConvexPairsOut.getBufferCL() ),
-                                b3BufferInfoCL( bodyBuf->getBufferCL(),true),
-                                b3BufferInfoCL( gpuCollidables.getBufferCL(),true),
-                                b3BufferInfoCL( convexData.getBufferCL(),true),
-                                b3BufferInfoCL( gpuVertices.getBufferCL(),true),
-                                b3BufferInfoCL( gpuUniqueEdges.getBufferCL(),true),
-                                b3BufferInfoCL( gpuFaces.getBufferCL(),true),
-                                b3BufferInfoCL( gpuIndices.getBufferCL(),true),
-                                b3BufferInfoCL( gpuChildShapes.getBufferCL(),true),
-                                b3BufferInfoCL( clAabbsWorldSpace.getBufferCL(),true),
-                                b3BufferInfoCL( m_concaveSepNormals.getBufferCL()),
-                                b3BufferInfoCL( m_concaveHasSeparatingNormals.getBufferCL()),
-                                b3BufferInfoCL( clippingFacesOutGPU.getBufferCL()),
-                                b3BufferInfoCL( worldVertsA1GPU.getBufferCL()),
-                                b3BufferInfoCL(worldNormalsAGPU.getBufferCL()),
-                                b3BufferInfoCL(worldVertsB1GPU.getBufferCL()),
-                                b3BufferInfoCL(m_dmins.getBufferCL())
-                            };
-                            
-                            b3LauncherCL launcher(m_queue, m_findConcaveSeparatingAxisVertexFaceKernel,"m_findConcaveSeparatingAxisVertexFaceKernel");
-                            launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-                            launcher.setConst(vertexFaceCapacity);
-                            launcher.setConst( numConcavePairs  );
-                            
-                            int num = numConcavePairs;
-                            launcher.launch1D( num);
-                            clFinish(m_queue);
-
-                            
-                        }
-//                        numConcavePairs = 0;
-                        if (1)
-                        {
-                            B3_PROFILE("findConcaveSeparatingAxisEdgeEdgeKernel");
-                            b3BufferInfoCL bInfo[] = {
-                                b3BufferInfoCL( triangleConvexPairsOut.getBufferCL() ),
-                                b3BufferInfoCL( bodyBuf->getBufferCL(),true),
-                                b3BufferInfoCL( gpuCollidables.getBufferCL(),true),
-                                b3BufferInfoCL( convexData.getBufferCL(),true),
-                                b3BufferInfoCL( gpuVertices.getBufferCL(),true),
-                                b3BufferInfoCL( gpuUniqueEdges.getBufferCL(),true),
-                                b3BufferInfoCL( gpuFaces.getBufferCL(),true),
-                                b3BufferInfoCL( gpuIndices.getBufferCL(),true),
-                                b3BufferInfoCL( gpuChildShapes.getBufferCL(),true),
-                                b3BufferInfoCL( clAabbsWorldSpace.getBufferCL(),true),
-                                b3BufferInfoCL( m_concaveSepNormals.getBufferCL()),
-                                b3BufferInfoCL( m_concaveHasSeparatingNormals.getBufferCL()),
-                                b3BufferInfoCL( clippingFacesOutGPU.getBufferCL()),
-                                b3BufferInfoCL( worldVertsA1GPU.getBufferCL()),
-                                b3BufferInfoCL(worldNormalsAGPU.getBufferCL()),
-                                b3BufferInfoCL(worldVertsB1GPU.getBufferCL()),
-                                b3BufferInfoCL(m_dmins.getBufferCL())
-                            };
-                            
-                            b3LauncherCL launcher(m_queue, m_findConcaveSeparatingAxisEdgeEdgeKernel,"m_findConcaveSeparatingAxisEdgeEdgeKernel");
-                            launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-                            launcher.setConst(vertexFaceCapacity);
-                            launcher.setConst( numConcavePairs  );
-                            
-                            int num = numConcavePairs;
-                            launcher.launch1D( num);
-                            clFinish(m_queue);
-                        }
-                      
-                        
-                        // numConcavePairs = 0;
-                        
-                        
-                        
-                        
-                        
-                        
-                    } else
-                    {
-                        B3_PROFILE("findConcaveSeparatingAxisKernel");
-                        b3BufferInfoCL bInfo[] = { 
-                            b3BufferInfoCL( triangleConvexPairsOut.getBufferCL() ), 
-                            b3BufferInfoCL( bodyBuf->getBufferCL(),true), 
-                            b3BufferInfoCL( gpuCollidables.getBufferCL(),true), 
-                            b3BufferInfoCL( convexData.getBufferCL(),true),
-                            b3BufferInfoCL( gpuVertices.getBufferCL(),true),
-                            b3BufferInfoCL( gpuUniqueEdges.getBufferCL(),true),
-                            b3BufferInfoCL( gpuFaces.getBufferCL(),true),
-                            b3BufferInfoCL( gpuIndices.getBufferCL(),true),
-                            b3BufferInfoCL( gpuChildShapes.getBufferCL(),true),
-                            b3BufferInfoCL( clAabbsWorldSpace.getBufferCL(),true),
-                            b3BufferInfoCL( m_concaveSepNormals.getBufferCL()),
-                            b3BufferInfoCL( m_concaveHasSeparatingNormals.getBufferCL()),
-                            b3BufferInfoCL( clippingFacesOutGPU.getBufferCL()),
-                            b3BufferInfoCL( worldVertsA1GPU.getBufferCL()),
-                            b3BufferInfoCL(worldNormalsAGPU.getBufferCL()),
-                            b3BufferInfoCL(worldVertsB1GPU.getBufferCL())
-                        };
-
-                        b3LauncherCL launcher(m_queue, m_findConcaveSeparatingAxisKernel,"m_findConcaveSeparatingAxisKernel");
-                        launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-                        launcher.setConst(vertexFaceCapacity);
-                        launcher.setConst( numConcavePairs  );
-
-                        int num = numConcavePairs;
-                        launcher.launch1D( num);
-                        clFinish(m_queue);
-                    }
-                    
-                    
-				} else
-				{
-
-						b3AlignedObjectArray<b3Int4> clippingFacesOutCPU;
-						b3AlignedObjectArray<b3Vector3> worldVertsA1CPU;
-						b3AlignedObjectArray<b3Vector3> worldNormalsACPU;
-						b3AlignedObjectArray<b3Vector3> worldVertsB1CPU;
-						b3AlignedObjectArray<int>concaveHasSeparatingNormalsCPU;
-
-						b3AlignedObjectArray<b3Int4> triangleConvexPairsOutHost;
-						triangleConvexPairsOut.copyToHost(triangleConvexPairsOutHost);
-						//triangleConvexPairsOutHost.resize(maxTriConvexPairCapacity);
-						b3AlignedObjectArray<b3RigidBodyData> hostBodyBuf;
-						bodyBuf->copyToHost(hostBodyBuf);
-						b3AlignedObjectArray<b3Collidable> hostCollidables;
-						gpuCollidables.copyToHost(hostCollidables);
-						b3AlignedObjectArray<b3Aabb> hostAabbsWorldSpace;
-						clAabbsWorldSpace.copyToHost(hostAabbsWorldSpace);
-
-						b3AlignedObjectArray<b3ConvexPolyhedronData> hostConvexData;
-						convexData.copyToHost(hostConvexData);
-
-						b3AlignedObjectArray<b3Vector3> hostVertices;
-						gpuVertices.copyToHost(hostVertices);
-
-						b3AlignedObjectArray<b3Vector3> hostUniqueEdges;
-						gpuUniqueEdges.copyToHost(hostUniqueEdges);
-						b3AlignedObjectArray<b3GpuFace> hostFaces;
-						gpuFaces.copyToHost(hostFaces);
-						b3AlignedObjectArray<int> hostIndices;
-						gpuIndices.copyToHost(hostIndices);
-						b3AlignedObjectArray<b3GpuChildShape> cpuChildShapes;
-						gpuChildShapes.copyToHost(cpuChildShapes);
-
-
-								
-						b3AlignedObjectArray<b3Vector3> concaveSepNormalsHost;
-						m_concaveSepNormals.copyToHost(concaveSepNormalsHost);
-						concaveHasSeparatingNormalsCPU.resize(concaveSepNormalsHost.size());
-
-						b3GpuChildShape* childShapePointerCPU = 0;
-						if (cpuChildShapes.size())
-							childShapePointerCPU  = &cpuChildShapes.at(0);
-
-						clippingFacesOutCPU.resize(clippingFacesOutGPU.size());
-						worldVertsA1CPU.resize(worldVertsA1GPU.size());
-    					worldNormalsACPU.resize(worldNormalsAGPU.size());
-						worldVertsB1CPU.resize(worldVertsB1GPU.size());
-
-						for (int i=0;i<numConcavePairs;i++)
+					if (splitSearchSepAxisConcave)
+					{
+						//printf("numConcavePairs = %d\n",numConcavePairs);
+						m_dmins.resize(numConcavePairs);
 						{
-							b3FindConcaveSeparatingAxisKernel(&triangleConvexPairsOutHost.at(0),
-								&hostBodyBuf.at(0),
-								&hostCollidables.at(0),
-								&hostConvexData.at(0), &hostVertices.at(0),&hostUniqueEdges.at(0),
-								&hostFaces.at(0),&hostIndices.at(0),childShapePointerCPU,
-								&hostAabbsWorldSpace.at(0),
-								&concaveSepNormalsHost.at(0),
-								&clippingFacesOutCPU.at(0),
-								&worldVertsA1CPU.at(0),
-								&worldNormalsACPU.at(0),
-								&worldVertsB1CPU.at(0),
-								&concaveHasSeparatingNormalsCPU.at(0),
-								vertexFaceCapacity,
-								numConcavePairs,i);
-						};
+							B3_PROFILE("findConcaveSeparatingAxisVertexFaceKernel");
+							b3BufferInfoCL bInfo[] = {
+								b3BufferInfoCL(triangleConvexPairsOut.getBufferCL()),
+								b3BufferInfoCL(bodyBuf->getBufferCL(), true),
+								b3BufferInfoCL(gpuCollidables.getBufferCL(), true),
+								b3BufferInfoCL(convexData.getBufferCL(), true),
+								b3BufferInfoCL(gpuVertices.getBufferCL(), true),
+								b3BufferInfoCL(gpuUniqueEdges.getBufferCL(), true),
+								b3BufferInfoCL(gpuFaces.getBufferCL(), true),
+								b3BufferInfoCL(gpuIndices.getBufferCL(), true),
+								b3BufferInfoCL(gpuChildShapes.getBufferCL(), true),
+								b3BufferInfoCL(clAabbsWorldSpace.getBufferCL(), true),
+								b3BufferInfoCL(m_concaveSepNormals.getBufferCL()),
+								b3BufferInfoCL(m_concaveHasSeparatingNormals.getBufferCL()),
+								b3BufferInfoCL(clippingFacesOutGPU.getBufferCL()),
+								b3BufferInfoCL(worldVertsA1GPU.getBufferCL()),
+								b3BufferInfoCL(worldNormalsAGPU.getBufferCL()),
+								b3BufferInfoCL(worldVertsB1GPU.getBufferCL()),
+								b3BufferInfoCL(m_dmins.getBufferCL())};
 
-						m_concaveSepNormals.copyFromHost(concaveSepNormalsHost);
-						m_concaveHasSeparatingNormals.copyFromHost(concaveHasSeparatingNormalsCPU);
-						clippingFacesOutGPU.copyFromHost(clippingFacesOutCPU);
-						worldVertsA1GPU.copyFromHost(worldVertsA1CPU);
-						worldNormalsAGPU.copyFromHost(worldNormalsACPU);
-						worldVertsB1GPU.copyFromHost(worldVertsB1CPU);
+							b3LauncherCL launcher(m_queue, m_findConcaveSeparatingAxisVertexFaceKernel, "m_findConcaveSeparatingAxisVertexFaceKernel");
+							launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+							launcher.setConst(vertexFaceCapacity);
+							launcher.setConst(numConcavePairs);
 
+							int num = numConcavePairs;
+							launcher.launch1D(num);
+							clFinish(m_queue);
+						}
+						//                        numConcavePairs = 0;
+						if (1)
+						{
+							B3_PROFILE("findConcaveSeparatingAxisEdgeEdgeKernel");
+							b3BufferInfoCL bInfo[] = {
+								b3BufferInfoCL(triangleConvexPairsOut.getBufferCL()),
+								b3BufferInfoCL(bodyBuf->getBufferCL(), true),
+								b3BufferInfoCL(gpuCollidables.getBufferCL(), true),
+								b3BufferInfoCL(convexData.getBufferCL(), true),
+								b3BufferInfoCL(gpuVertices.getBufferCL(), true),
+								b3BufferInfoCL(gpuUniqueEdges.getBufferCL(), true),
+								b3BufferInfoCL(gpuFaces.getBufferCL(), true),
+								b3BufferInfoCL(gpuIndices.getBufferCL(), true),
+								b3BufferInfoCL(gpuChildShapes.getBufferCL(), true),
+								b3BufferInfoCL(clAabbsWorldSpace.getBufferCL(), true),
+								b3BufferInfoCL(m_concaveSepNormals.getBufferCL()),
+								b3BufferInfoCL(m_concaveHasSeparatingNormals.getBufferCL()),
+								b3BufferInfoCL(clippingFacesOutGPU.getBufferCL()),
+								b3BufferInfoCL(worldVertsA1GPU.getBufferCL()),
+								b3BufferInfoCL(worldNormalsAGPU.getBufferCL()),
+								b3BufferInfoCL(worldVertsB1GPU.getBufferCL()),
+								b3BufferInfoCL(m_dmins.getBufferCL())};
 
+							b3LauncherCL launcher(m_queue, m_findConcaveSeparatingAxisEdgeEdgeKernel, "m_findConcaveSeparatingAxisEdgeEdgeKernel");
+							launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+							launcher.setConst(vertexFaceCapacity);
+							launcher.setConst(numConcavePairs);
 
+							int num = numConcavePairs;
+							launcher.launch1D(num);
+							clFinish(m_queue);
+						}
+
+						// numConcavePairs = 0;
+					}
+					else
+					{
+						B3_PROFILE("findConcaveSeparatingAxisKernel");
+						b3BufferInfoCL bInfo[] = {
+							b3BufferInfoCL(triangleConvexPairsOut.getBufferCL()),
+							b3BufferInfoCL(bodyBuf->getBufferCL(), true),
+							b3BufferInfoCL(gpuCollidables.getBufferCL(), true),
+							b3BufferInfoCL(convexData.getBufferCL(), true),
+							b3BufferInfoCL(gpuVertices.getBufferCL(), true),
+							b3BufferInfoCL(gpuUniqueEdges.getBufferCL(), true),
+							b3BufferInfoCL(gpuFaces.getBufferCL(), true),
+							b3BufferInfoCL(gpuIndices.getBufferCL(), true),
+							b3BufferInfoCL(gpuChildShapes.getBufferCL(), true),
+							b3BufferInfoCL(clAabbsWorldSpace.getBufferCL(), true),
+							b3BufferInfoCL(m_concaveSepNormals.getBufferCL()),
+							b3BufferInfoCL(m_concaveHasSeparatingNormals.getBufferCL()),
+							b3BufferInfoCL(clippingFacesOutGPU.getBufferCL()),
+							b3BufferInfoCL(worldVertsA1GPU.getBufferCL()),
+							b3BufferInfoCL(worldNormalsAGPU.getBufferCL()),
+							b3BufferInfoCL(worldVertsB1GPU.getBufferCL())};
+
+						b3LauncherCL launcher(m_queue, m_findConcaveSeparatingAxisKernel, "m_findConcaveSeparatingAxisKernel");
+						launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+						launcher.setConst(vertexFaceCapacity);
+						launcher.setConst(numConcavePairs);
+
+						int num = numConcavePairs;
+						launcher.launch1D(num);
+						clFinish(m_queue);
+					}
 				}
-//							b3AlignedObjectArray<b3Vector3> cpuCompoundSepNormals;
-//						m_concaveSepNormals.copyToHost(cpuCompoundSepNormals);
-//					b3AlignedObjectArray<b3Int4> cpuConcavePairs;
-//				triangleConvexPairsOut.copyToHost(cpuConcavePairs);
+				else
+				{
+					b3AlignedObjectArray<b3Int4> clippingFacesOutCPU;
+					b3AlignedObjectArray<b3Vector3> worldVertsA1CPU;
+					b3AlignedObjectArray<b3Vector3> worldNormalsACPU;
+					b3AlignedObjectArray<b3Vector3> worldVertsB1CPU;
+					b3AlignedObjectArray<int> concaveHasSeparatingNormalsCPU;
 
+					b3AlignedObjectArray<b3Int4> triangleConvexPairsOutHost;
+					triangleConvexPairsOut.copyToHost(triangleConvexPairsOutHost);
+					//triangleConvexPairsOutHost.resize(maxTriConvexPairCapacity);
+					b3AlignedObjectArray<b3RigidBodyData> hostBodyBuf;
+					bodyBuf->copyToHost(hostBodyBuf);
+					b3AlignedObjectArray<b3Collidable> hostCollidables;
+					gpuCollidables.copyToHost(hostCollidables);
+					b3AlignedObjectArray<b3Aabb> hostAabbsWorldSpace;
+					clAabbsWorldSpace.copyToHost(hostAabbsWorldSpace);
 
+					b3AlignedObjectArray<b3ConvexPolyhedronData> hostConvexData;
+					convexData.copyToHost(hostConvexData);
+
+					b3AlignedObjectArray<b3Vector3> hostVertices;
+					gpuVertices.copyToHost(hostVertices);
+
+					b3AlignedObjectArray<b3Vector3> hostUniqueEdges;
+					gpuUniqueEdges.copyToHost(hostUniqueEdges);
+					b3AlignedObjectArray<b3GpuFace> hostFaces;
+					gpuFaces.copyToHost(hostFaces);
+					b3AlignedObjectArray<int> hostIndices;
+					gpuIndices.copyToHost(hostIndices);
+					b3AlignedObjectArray<b3GpuChildShape> cpuChildShapes;
+					gpuChildShapes.copyToHost(cpuChildShapes);
+
+					b3AlignedObjectArray<b3Vector3> concaveSepNormalsHost;
+					m_concaveSepNormals.copyToHost(concaveSepNormalsHost);
+					concaveHasSeparatingNormalsCPU.resize(concaveSepNormalsHost.size());
+
+					b3GpuChildShape* childShapePointerCPU = 0;
+					if (cpuChildShapes.size())
+						childShapePointerCPU = &cpuChildShapes.at(0);
+
+					clippingFacesOutCPU.resize(clippingFacesOutGPU.size());
+					worldVertsA1CPU.resize(worldVertsA1GPU.size());
+					worldNormalsACPU.resize(worldNormalsAGPU.size());
+					worldVertsB1CPU.resize(worldVertsB1GPU.size());
+
+					for (int i = 0; i < numConcavePairs; i++)
+					{
+						b3FindConcaveSeparatingAxisKernel(&triangleConvexPairsOutHost.at(0),
+														  &hostBodyBuf.at(0),
+														  &hostCollidables.at(0),
+														  &hostConvexData.at(0), &hostVertices.at(0), &hostUniqueEdges.at(0),
+														  &hostFaces.at(0), &hostIndices.at(0), childShapePointerCPU,
+														  &hostAabbsWorldSpace.at(0),
+														  &concaveSepNormalsHost.at(0),
+														  &clippingFacesOutCPU.at(0),
+														  &worldVertsA1CPU.at(0),
+														  &worldNormalsACPU.at(0),
+														  &worldVertsB1CPU.at(0),
+														  &concaveHasSeparatingNormalsCPU.at(0),
+														  vertexFaceCapacity,
+														  numConcavePairs, i);
+					};
+
+					m_concaveSepNormals.copyFromHost(concaveSepNormalsHost);
+					m_concaveHasSeparatingNormals.copyFromHost(concaveHasSeparatingNormalsCPU);
+					clippingFacesOutGPU.copyFromHost(clippingFacesOutCPU);
+					worldVertsA1GPU.copyFromHost(worldVertsA1CPU);
+					worldNormalsAGPU.copyFromHost(worldNormalsACPU);
+					worldVertsB1GPU.copyFromHost(worldVertsB1CPU);
+				}
+				//							b3AlignedObjectArray<b3Vector3> cpuCompoundSepNormals;
+				//						m_concaveSepNormals.copyToHost(cpuCompoundSepNormals);
+				//					b3AlignedObjectArray<b3Int4> cpuConcavePairs;
+				//				triangleConvexPairsOut.copyToHost(cpuConcavePairs);
 			}
 		}
-		
-		
 	}
 
 	if (numConcavePairs)
 	{
-			if (numConcavePairs)
+		if (numConcavePairs)
 		{
 			B3_PROFILE("findConcaveSphereContactsKernel");
-				nContacts = m_totalContactsOut.at(0);
-//				printf("nContacts1 = %d\n",nContacts);
-			b3BufferInfoCL bInfo[] = { 
-				b3BufferInfoCL( triangleConvexPairsOut.getBufferCL() ), 
-				b3BufferInfoCL( bodyBuf->getBufferCL(),true), 
-				b3BufferInfoCL( gpuCollidables.getBufferCL(),true), 
-				b3BufferInfoCL( convexData.getBufferCL(),true),
-				b3BufferInfoCL( gpuVertices.getBufferCL(),true),
-				b3BufferInfoCL( gpuUniqueEdges.getBufferCL(),true),
-				b3BufferInfoCL( gpuFaces.getBufferCL(),true),
-				b3BufferInfoCL( gpuIndices.getBufferCL(),true),
-				b3BufferInfoCL( clAabbsWorldSpace.getBufferCL(),true),
-				b3BufferInfoCL( contactOut->getBufferCL()),
-				b3BufferInfoCL( m_totalContactsOut.getBufferCL())
-			};
+			nContacts = m_totalContactsOut.at(0);
+			//				printf("nContacts1 = %d\n",nContacts);
+			b3BufferInfoCL bInfo[] = {
+				b3BufferInfoCL(triangleConvexPairsOut.getBufferCL()),
+				b3BufferInfoCL(bodyBuf->getBufferCL(), true),
+				b3BufferInfoCL(gpuCollidables.getBufferCL(), true),
+				b3BufferInfoCL(convexData.getBufferCL(), true),
+				b3BufferInfoCL(gpuVertices.getBufferCL(), true),
+				b3BufferInfoCL(gpuUniqueEdges.getBufferCL(), true),
+				b3BufferInfoCL(gpuFaces.getBufferCL(), true),
+				b3BufferInfoCL(gpuIndices.getBufferCL(), true),
+				b3BufferInfoCL(clAabbsWorldSpace.getBufferCL(), true),
+				b3BufferInfoCL(contactOut->getBufferCL()),
+				b3BufferInfoCL(m_totalContactsOut.getBufferCL())};
 
-			b3LauncherCL launcher(m_queue, m_findConcaveSphereContactsKernel,"m_findConcaveSphereContactsKernel");
-			launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
+			b3LauncherCL launcher(m_queue, m_findConcaveSphereContactsKernel, "m_findConcaveSphereContactsKernel");
+			launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
 
-			launcher.setConst( numConcavePairs  );
+			launcher.setConst(numConcavePairs);
 			launcher.setConst(maxContactCapacity);
 
 			int num = numConcavePairs;
-			launcher.launch1D( num);
+			launcher.launch1D(num);
 			clFinish(m_queue);
 			nContacts = m_totalContactsOut.at(0);
 			//printf("nContacts (after findConcaveSphereContactsKernel) = %d\n",nContacts);
@@ -4088,11 +3816,8 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 				nContacts = maxContactCapacity;
 			}
 		}
-		
 	}
 
-
-
 #ifdef __APPLE__
 	bool contactClippingOnGpu = true;
 #else
@@ -4101,9 +3826,8 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 
 	if (contactClippingOnGpu)
 	{
-		m_totalContactsOut.copyFromHostPointer(&nContacts,1,0,true);
-//		printf("nContacts3 = %d\n",nContacts);
-
+		m_totalContactsOut.copyFromHostPointer(&nContacts, 1, 0, true);
+		//		printf("nContacts3 = %d\n",nContacts);
 
 		//B3_PROFILE("clipHullHullKernel");
 
@@ -4122,15 +3846,12 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 
 			if (breakupConcaveConvexKernel)
 			{
-
-				worldVertsB2GPU.resize(vertexFaceCapacity*numConcavePairs);
-
+				worldVertsB2GPU.resize(vertexFaceCapacity * numConcavePairs);
 
 				//clipFacesAndFindContacts
 
 				if (clipConcaveFacesAndFindContactsCPU)
 				{
-
 					b3AlignedObjectArray<b3Int4> clippingFacesOutCPU;
 					b3AlignedObjectArray<b3Vector3> worldVertsA1CPU;
 					b3AlignedObjectArray<b3Vector3> worldNormalsACPU;
@@ -4141,120 +3862,108 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 					worldNormalsAGPU.copyToHost(worldNormalsACPU);
 					worldVertsB1GPU.copyToHost(worldVertsB1CPU);
 
-
-
-					b3AlignedObjectArray<int>concaveHasSeparatingNormalsCPU;
+					b3AlignedObjectArray<int> concaveHasSeparatingNormalsCPU;
 					m_concaveHasSeparatingNormals.copyToHost(concaveHasSeparatingNormalsCPU);
 
 					b3AlignedObjectArray<b3Vector3> concaveSepNormalsHost;
 					m_concaveSepNormals.copyToHost(concaveSepNormalsHost);
 
-					b3AlignedObjectArray<b3Vector3> worldVertsB2CPU;  
+					b3AlignedObjectArray<b3Vector3> worldVertsB2CPU;
 					worldVertsB2CPU.resize(worldVertsB2GPU.size());
 
-
-					for (int i=0;i<numConcavePairs;i++)
+					for (int i = 0; i < numConcavePairs; i++)
 					{
-
-						clipFacesAndFindContactsKernel(   &concaveSepNormalsHost.at(0),
-							&concaveHasSeparatingNormalsCPU.at(0),
-							&clippingFacesOutCPU.at(0),
-							&worldVertsA1CPU.at(0),
-							&worldNormalsACPU.at(0),
-							&worldVertsB1CPU.at(0),
-							&worldVertsB2CPU.at(0),
-							vertexFaceCapacity,
-							i);
+						clipFacesAndFindContactsKernel(&concaveSepNormalsHost.at(0),
+													   &concaveHasSeparatingNormalsCPU.at(0),
+													   &clippingFacesOutCPU.at(0),
+													   &worldVertsA1CPU.at(0),
+													   &worldNormalsACPU.at(0),
+													   &worldVertsB1CPU.at(0),
+													   &worldVertsB2CPU.at(0),
+													   vertexFaceCapacity,
+													   i);
 					}
 
 					clippingFacesOutGPU.copyFromHost(clippingFacesOutCPU);
 					worldVertsB2GPU.copyFromHost(worldVertsB2CPU);
-
-
-				} else
+				}
+				else
 				{
-
 					if (1)
 					{
-
-
-
 						B3_PROFILE("clipFacesAndFindContacts");
 						//nContacts = m_totalContactsOut.at(0);
 						//int h = m_hasSeparatingNormals.at(0);
 						//int4 p = clippingFacesOutGPU.at(0);
 						b3BufferInfoCL bInfo[] = {
-							b3BufferInfoCL( m_concaveSepNormals.getBufferCL()),
-							b3BufferInfoCL( m_concaveHasSeparatingNormals.getBufferCL()),
-							b3BufferInfoCL( clippingFacesOutGPU.getBufferCL()),
-							b3BufferInfoCL( worldVertsA1GPU.getBufferCL()),
-							b3BufferInfoCL( worldNormalsAGPU.getBufferCL()),
-							b3BufferInfoCL( worldVertsB1GPU.getBufferCL()),
-							b3BufferInfoCL( worldVertsB2GPU.getBufferCL())
-						};
-						b3LauncherCL launcher(m_queue, m_clipFacesAndFindContacts,"m_clipFacesAndFindContacts");
-						launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
+							b3BufferInfoCL(m_concaveSepNormals.getBufferCL()),
+							b3BufferInfoCL(m_concaveHasSeparatingNormals.getBufferCL()),
+							b3BufferInfoCL(clippingFacesOutGPU.getBufferCL()),
+							b3BufferInfoCL(worldVertsA1GPU.getBufferCL()),
+							b3BufferInfoCL(worldNormalsAGPU.getBufferCL()),
+							b3BufferInfoCL(worldVertsB1GPU.getBufferCL()),
+							b3BufferInfoCL(worldVertsB2GPU.getBufferCL())};
+						b3LauncherCL launcher(m_queue, m_clipFacesAndFindContacts, "m_clipFacesAndFindContacts");
+						launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
 						launcher.setConst(vertexFaceCapacity);
 
-						launcher.setConst( numConcavePairs  );
+						launcher.setConst(numConcavePairs);
 						int debugMode = 0;
-						launcher.setConst( debugMode);
+						launcher.setConst(debugMode);
 						int num = numConcavePairs;
-						launcher.launch1D( num);
+						launcher.launch1D(num);
 						clFinish(m_queue);
 						//int bla = m_totalContactsOut.at(0);
 					}
 				}
 				//contactReduction
 				{
-					int newContactCapacity=nContacts+numConcavePairs; 
+					int newContactCapacity = nContacts + numConcavePairs;
 					contactOut->reserve(newContactCapacity);
 					if (reduceConcaveContactsOnGPU)
 					{
-//						printf("newReservation = %d\n",newReservation);
+						//						printf("newReservation = %d\n",newReservation);
 						{
 							B3_PROFILE("newContactReductionKernel");
 							b3BufferInfoCL bInfo[] =
-							{
-								b3BufferInfoCL( triangleConvexPairsOut.getBufferCL(), true ),
-								b3BufferInfoCL( bodyBuf->getBufferCL(),true),
-								b3BufferInfoCL( m_concaveSepNormals.getBufferCL()),
-								b3BufferInfoCL( m_concaveHasSeparatingNormals.getBufferCL()),
-								b3BufferInfoCL( contactOut->getBufferCL()),
-								b3BufferInfoCL( clippingFacesOutGPU.getBufferCL()),
-								b3BufferInfoCL( worldVertsB2GPU.getBufferCL()),
-								b3BufferInfoCL( m_totalContactsOut.getBufferCL())
-							};
+								{
+									b3BufferInfoCL(triangleConvexPairsOut.getBufferCL(), true),
+									b3BufferInfoCL(bodyBuf->getBufferCL(), true),
+									b3BufferInfoCL(m_concaveSepNormals.getBufferCL()),
+									b3BufferInfoCL(m_concaveHasSeparatingNormals.getBufferCL()),
+									b3BufferInfoCL(contactOut->getBufferCL()),
+									b3BufferInfoCL(clippingFacesOutGPU.getBufferCL()),
+									b3BufferInfoCL(worldVertsB2GPU.getBufferCL()),
+									b3BufferInfoCL(m_totalContactsOut.getBufferCL())};
 
-							b3LauncherCL launcher(m_queue, m_newContactReductionKernel,"m_newContactReductionKernel");
-							launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
+							b3LauncherCL launcher(m_queue, m_newContactReductionKernel, "m_newContactReductionKernel");
+							launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
 							launcher.setConst(vertexFaceCapacity);
 							launcher.setConst(newContactCapacity);
-							launcher.setConst( numConcavePairs  );
+							launcher.setConst(numConcavePairs);
 							int num = numConcavePairs;
 
-							launcher.launch1D( num);
+							launcher.launch1D(num);
 						}
 						nContacts = m_totalContactsOut.at(0);
 						contactOut->resize(nContacts);
 
 						//printf("contactOut4 (after newContactReductionKernel) = %d\n",nContacts);
-					}else
+					}
+					else
 					{
-						
 						volatile int nGlobalContactsOut = nContacts;
 						b3AlignedObjectArray<b3Int4> triangleConvexPairsOutHost;
 						triangleConvexPairsOut.copyToHost(triangleConvexPairsOutHost);
 						b3AlignedObjectArray<b3RigidBodyData> hostBodyBuf;
 						bodyBuf->copyToHost(hostBodyBuf);
 
-						b3AlignedObjectArray<int>concaveHasSeparatingNormalsCPU;
+						b3AlignedObjectArray<int> concaveHasSeparatingNormalsCPU;
 						m_concaveHasSeparatingNormals.copyToHost(concaveHasSeparatingNormalsCPU);
 
 						b3AlignedObjectArray<b3Vector3> concaveSepNormalsHost;
 						m_concaveSepNormals.copyToHost(concaveSepNormalsHost);
 
-
 						b3AlignedObjectArray<b3Contact4> hostContacts;
 						if (nContacts)
 						{
@@ -4268,67 +3977,59 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 						clippingFacesOutGPU.copyToHost(clippingFacesOutCPU);
 						worldVertsB2GPU.copyToHost(worldVertsB2CPU);
 
-
-
-						for (int i=0;i<numConcavePairs;i++)
+						for (int i = 0; i < numConcavePairs; i++)
 						{
-							b3NewContactReductionKernel( &triangleConvexPairsOutHost.at(0),
-                                                   &hostBodyBuf.at(0),
-												   &concaveSepNormalsHost.at(0),
-												   &concaveHasSeparatingNormalsCPU.at(0),
-												   &hostContacts.at(0),
-                                                   &clippingFacesOutCPU.at(0),
-                                                   &worldVertsB2CPU.at(0),
-                                                   &nGlobalContactsOut,
-                                                   vertexFaceCapacity,
-												   newContactCapacity,
-                                                   numConcavePairs,
-												   i
-                                                   );
-
+							b3NewContactReductionKernel(&triangleConvexPairsOutHost.at(0),
+														&hostBodyBuf.at(0),
+														&concaveSepNormalsHost.at(0),
+														&concaveHasSeparatingNormalsCPU.at(0),
+														&hostContacts.at(0),
+														&clippingFacesOutCPU.at(0),
+														&worldVertsB2CPU.at(0),
+														&nGlobalContactsOut,
+														vertexFaceCapacity,
+														newContactCapacity,
+														numConcavePairs,
+														i);
 						}
 
-
 						nContacts = nGlobalContactsOut;
-						m_totalContactsOut.copyFromHostPointer(&nContacts,1,0,true);
-//						nContacts = m_totalContactsOut.at(0);
+						m_totalContactsOut.copyFromHostPointer(&nContacts, 1, 0, true);
+						//						nContacts = m_totalContactsOut.at(0);
 						//contactOut->resize(nContacts);
 						hostContacts.resize(nContacts);
 						//printf("contactOut4 (after newContactReductionKernel) = %d\n",nContacts);
 						contactOut->copyFromHost(hostContacts);
 					}
-
 				}
 				//re-use?
-
-
-			} else
+			}
+			else
 			{
 				B3_PROFILE("clipHullHullConcaveConvexKernel");
 				nContacts = m_totalContactsOut.at(0);
 				int newContactCapacity = contactOut->capacity();
 
 				//printf("contactOut5 = %d\n",nContacts);
-				b3BufferInfoCL bInfo[] = { 
-					b3BufferInfoCL( triangleConvexPairsOut.getBufferCL(), true ), 
-					b3BufferInfoCL( bodyBuf->getBufferCL(),true), 
-					b3BufferInfoCL( gpuCollidables.getBufferCL(),true), 
-					b3BufferInfoCL( convexData.getBufferCL(),true),
-					b3BufferInfoCL( gpuVertices.getBufferCL(),true),
-					b3BufferInfoCL( gpuUniqueEdges.getBufferCL(),true),
-					b3BufferInfoCL( gpuFaces.getBufferCL(),true),
-					b3BufferInfoCL( gpuIndices.getBufferCL(),true),
-					b3BufferInfoCL( gpuChildShapes.getBufferCL(),true),
-					b3BufferInfoCL( m_concaveSepNormals.getBufferCL()),
-					b3BufferInfoCL( contactOut->getBufferCL()),
-					b3BufferInfoCL( m_totalContactsOut.getBufferCL())	
-				};
-				b3LauncherCL launcher(m_queue, m_clipHullHullConcaveConvexKernel,"m_clipHullHullConcaveConvexKernel");
-				launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
+				b3BufferInfoCL bInfo[] = {
+					b3BufferInfoCL(triangleConvexPairsOut.getBufferCL(), true),
+					b3BufferInfoCL(bodyBuf->getBufferCL(), true),
+					b3BufferInfoCL(gpuCollidables.getBufferCL(), true),
+					b3BufferInfoCL(convexData.getBufferCL(), true),
+					b3BufferInfoCL(gpuVertices.getBufferCL(), true),
+					b3BufferInfoCL(gpuUniqueEdges.getBufferCL(), true),
+					b3BufferInfoCL(gpuFaces.getBufferCL(), true),
+					b3BufferInfoCL(gpuIndices.getBufferCL(), true),
+					b3BufferInfoCL(gpuChildShapes.getBufferCL(), true),
+					b3BufferInfoCL(m_concaveSepNormals.getBufferCL()),
+					b3BufferInfoCL(contactOut->getBufferCL()),
+					b3BufferInfoCL(m_totalContactsOut.getBufferCL())};
+				b3LauncherCL launcher(m_queue, m_clipHullHullConcaveConvexKernel, "m_clipHullHullConcaveConvexKernel");
+				launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
 				launcher.setConst(newContactCapacity);
-				launcher.setConst( numConcavePairs  );
+				launcher.setConst(numConcavePairs);
 				int num = numConcavePairs;
-				launcher.launch1D( num);
+				launcher.launch1D(num);
 				clFinish(m_queue);
 				nContacts = m_totalContactsOut.at(0);
 				contactOut->resize(nContacts);
@@ -4337,12 +4038,10 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 				contactOut->copyToHost(cpuContacts);
 			}
 			//			printf("nContacts after = %d\n", nContacts);
-		}//numConcavePairs
-
-
+		}  //numConcavePairs
 
 		//convex-convex contact clipping
-		
+
 		bool breakupKernel = false;
 
 #ifdef __APPLE__
@@ -4350,166 +4049,149 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
 #endif
 
 #ifdef CHECK_ON_HOST
-	bool computeConvexConvex = false;
+		bool computeConvexConvex = false;
 #else
-	bool computeConvexConvex = true;
-#endif//CHECK_ON_HOST
+		bool computeConvexConvex = true;
+#endif  //CHECK_ON_HOST
 		if (computeConvexConvex)
 		{
 			B3_PROFILE("clipHullHullKernel");
-		if (breakupKernel)
-		{
-
-
-
-
-			worldVertsB1GPU.resize(vertexFaceCapacity*nPairs);
-			clippingFacesOutGPU.resize(nPairs);
-			worldNormalsAGPU.resize(nPairs);
-			worldVertsA1GPU.resize(vertexFaceCapacity*nPairs);
-			worldVertsB2GPU.resize(vertexFaceCapacity*nPairs);
-
-			if (findConvexClippingFacesGPU)
+			if (breakupKernel)
 			{
-				B3_PROFILE("findClippingFacesKernel");
-				b3BufferInfoCL bInfo[] = {
-					b3BufferInfoCL( pairs->getBufferCL(), true ),
-					b3BufferInfoCL( bodyBuf->getBufferCL(),true),
-					b3BufferInfoCL( gpuCollidables.getBufferCL(),true),
-					b3BufferInfoCL( convexData.getBufferCL(),true),
-					b3BufferInfoCL( gpuVertices.getBufferCL(),true),
-					b3BufferInfoCL( gpuUniqueEdges.getBufferCL(),true),
-					b3BufferInfoCL( gpuFaces.getBufferCL(),true), 
-					b3BufferInfoCL( gpuIndices.getBufferCL(),true),
-					b3BufferInfoCL( m_sepNormals.getBufferCL()),
-					b3BufferInfoCL( m_hasSeparatingNormals.getBufferCL()),
-					b3BufferInfoCL( clippingFacesOutGPU.getBufferCL()),
-					b3BufferInfoCL( worldVertsA1GPU.getBufferCL()),
-					b3BufferInfoCL( worldNormalsAGPU.getBufferCL()),
-					b3BufferInfoCL( worldVertsB1GPU.getBufferCL())
-				};
+				worldVertsB1GPU.resize(vertexFaceCapacity * nPairs);
+				clippingFacesOutGPU.resize(nPairs);
+				worldNormalsAGPU.resize(nPairs);
+				worldVertsA1GPU.resize(vertexFaceCapacity * nPairs);
+				worldVertsB2GPU.resize(vertexFaceCapacity * nPairs);
 
-				b3LauncherCL launcher(m_queue, m_findClippingFacesKernel,"m_findClippingFacesKernel");
-				launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-				launcher.setConst( vertexFaceCapacity);
-				launcher.setConst( nPairs  );
-				int num = nPairs;
-				launcher.launch1D( num);
-				clFinish(m_queue);
-
-			} else
-			{
-				
-				float minDist = -1e30f;
-				float maxDist = 0.02f;
-
-				b3AlignedObjectArray<b3ConvexPolyhedronData> hostConvexData;
-				convexData.copyToHost(hostConvexData);
-				b3AlignedObjectArray<b3Collidable> hostCollidables;
-				gpuCollidables.copyToHost(hostCollidables);
-
-				b3AlignedObjectArray<int> hostHasSepNormals;
-				m_hasSeparatingNormals.copyToHost(hostHasSepNormals);
-				b3AlignedObjectArray<b3Vector3> cpuSepNormals;
-				m_sepNormals.copyToHost(cpuSepNormals);
-
-				b3AlignedObjectArray<b3Int4> hostPairs;
-				pairs->copyToHost(hostPairs);
-				b3AlignedObjectArray<b3RigidBodyData> hostBodyBuf;
-				bodyBuf->copyToHost(hostBodyBuf);
-
-
-				//worldVertsB1GPU.resize(vertexFaceCapacity*nPairs);
-				b3AlignedObjectArray<b3Vector3> worldVertsB1CPU;
-				worldVertsB1GPU.copyToHost(worldVertsB1CPU);
-
-				b3AlignedObjectArray<b3Int4> clippingFacesOutCPU;
-				clippingFacesOutGPU.copyToHost(clippingFacesOutCPU);
-
-				b3AlignedObjectArray<b3Vector3> worldNormalsACPU;
-				worldNormalsACPU.resize(nPairs);
-
-				b3AlignedObjectArray<b3Vector3> worldVertsA1CPU;
-				worldVertsA1CPU.resize(worldVertsA1GPU.size());
-			
-			
-				b3AlignedObjectArray<b3Vector3> hostVertices;
-				gpuVertices.copyToHost(hostVertices);
-				b3AlignedObjectArray<b3GpuFace> hostFaces;
-				gpuFaces.copyToHost(hostFaces);
-				b3AlignedObjectArray<int> hostIndices;
-				gpuIndices.copyToHost(hostIndices);
-				
-
-				for (int i=0;i<nPairs;i++)
+				if (findConvexClippingFacesGPU)
 				{
+					B3_PROFILE("findClippingFacesKernel");
+					b3BufferInfoCL bInfo[] = {
+						b3BufferInfoCL(pairs->getBufferCL(), true),
+						b3BufferInfoCL(bodyBuf->getBufferCL(), true),
+						b3BufferInfoCL(gpuCollidables.getBufferCL(), true),
+						b3BufferInfoCL(convexData.getBufferCL(), true),
+						b3BufferInfoCL(gpuVertices.getBufferCL(), true),
+						b3BufferInfoCL(gpuUniqueEdges.getBufferCL(), true),
+						b3BufferInfoCL(gpuFaces.getBufferCL(), true),
+						b3BufferInfoCL(gpuIndices.getBufferCL(), true),
+						b3BufferInfoCL(m_sepNormals.getBufferCL()),
+						b3BufferInfoCL(m_hasSeparatingNormals.getBufferCL()),
+						b3BufferInfoCL(clippingFacesOutGPU.getBufferCL()),
+						b3BufferInfoCL(worldVertsA1GPU.getBufferCL()),
+						b3BufferInfoCL(worldNormalsAGPU.getBufferCL()),
+						b3BufferInfoCL(worldVertsB1GPU.getBufferCL())};
 
-					int bodyIndexA = hostPairs[i].x;
-					int bodyIndexB = hostPairs[i].y;
-			
-					int collidableIndexA = hostBodyBuf[bodyIndexA].m_collidableIdx;
-					int collidableIndexB = hostBodyBuf[bodyIndexB].m_collidableIdx;
-			
-					int shapeIndexA = hostCollidables[collidableIndexA].m_shapeIndex;
-					int shapeIndexB = hostCollidables[collidableIndexB].m_shapeIndex;
-			
-
-					if (hostHasSepNormals[i])
-					{
-						b3FindClippingFaces(cpuSepNormals[i],
-							&hostConvexData[shapeIndexA],
-							&hostConvexData[shapeIndexB],
-							hostBodyBuf[bodyIndexA].m_pos,hostBodyBuf[bodyIndexA].m_quat,
-							hostBodyBuf[bodyIndexB].m_pos,hostBodyBuf[bodyIndexB].m_quat,
-							&worldVertsA1CPU.at(0),&worldNormalsACPU.at(0),
-							&worldVertsB1CPU.at(0),
-							vertexFaceCapacity,minDist,maxDist,
-							&hostVertices.at(0),&hostFaces.at(0),
-							&hostIndices.at(0),
-							&hostVertices.at(0),&hostFaces.at(0),
-							&hostIndices.at(0),&clippingFacesOutCPU.at(0),i);
-					}
+					b3LauncherCL launcher(m_queue, m_findClippingFacesKernel, "m_findClippingFacesKernel");
+					launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+					launcher.setConst(vertexFaceCapacity);
+					launcher.setConst(nPairs);
+					int num = nPairs;
+					launcher.launch1D(num);
+					clFinish(m_queue);
 				}
-
-				clippingFacesOutGPU.copyFromHost(clippingFacesOutCPU);
-				worldVertsA1GPU.copyFromHost(worldVertsA1CPU);
-				worldNormalsAGPU.copyFromHost(worldNormalsACPU);
-				worldVertsB1GPU.copyFromHost(worldVertsB1CPU);
-
-			}
-
-
-
-
-
-			///clip face B against face A, reduce contacts and append them to a global contact array
-			if (1)
-			{
-				if (clipConvexFacesAndFindContactsCPU)
+				else
 				{
+					float minDist = -1e30f;
+					float maxDist = 0.02f;
 
-					//b3AlignedObjectArray<b3Int4> hostPairs;
-					//pairs->copyToHost(hostPairs);
+					b3AlignedObjectArray<b3ConvexPolyhedronData> hostConvexData;
+					convexData.copyToHost(hostConvexData);
+					b3AlignedObjectArray<b3Collidable> hostCollidables;
+					gpuCollidables.copyToHost(hostCollidables);
 
-					b3AlignedObjectArray<b3Vector3> hostSepNormals;
-					m_sepNormals.copyToHost(hostSepNormals);
-					b3AlignedObjectArray<int> hostHasSepAxis;
-					m_hasSeparatingNormals.copyToHost(hostHasSepAxis);
+					b3AlignedObjectArray<int> hostHasSepNormals;
+					m_hasSeparatingNormals.copyToHost(hostHasSepNormals);
+					b3AlignedObjectArray<b3Vector3> cpuSepNormals;
+					m_sepNormals.copyToHost(cpuSepNormals);
 
-					b3AlignedObjectArray<b3Int4> hostClippingFaces;
-					clippingFacesOutGPU.copyToHost(hostClippingFaces);
-					b3AlignedObjectArray<b3Vector3> worldVertsB2CPU;
-					worldVertsB2CPU.resize(vertexFaceCapacity*nPairs);
-					
-					b3AlignedObjectArray<b3Vector3>worldVertsA1CPU;
-					worldVertsA1GPU.copyToHost(worldVertsA1CPU);
-					b3AlignedObjectArray<b3Vector3> worldNormalsACPU;
-					worldNormalsAGPU.copyToHost(worldNormalsACPU);
+					b3AlignedObjectArray<b3Int4> hostPairs;
+					pairs->copyToHost(hostPairs);
+					b3AlignedObjectArray<b3RigidBodyData> hostBodyBuf;
+					bodyBuf->copyToHost(hostBodyBuf);
 
-					b3AlignedObjectArray<b3Vector3>  worldVertsB1CPU;
+					//worldVertsB1GPU.resize(vertexFaceCapacity*nPairs);
+					b3AlignedObjectArray<b3Vector3> worldVertsB1CPU;
 					worldVertsB1GPU.copyToHost(worldVertsB1CPU);
 
-					/*
+					b3AlignedObjectArray<b3Int4> clippingFacesOutCPU;
+					clippingFacesOutGPU.copyToHost(clippingFacesOutCPU);
+
+					b3AlignedObjectArray<b3Vector3> worldNormalsACPU;
+					worldNormalsACPU.resize(nPairs);
+
+					b3AlignedObjectArray<b3Vector3> worldVertsA1CPU;
+					worldVertsA1CPU.resize(worldVertsA1GPU.size());
+
+					b3AlignedObjectArray<b3Vector3> hostVertices;
+					gpuVertices.copyToHost(hostVertices);
+					b3AlignedObjectArray<b3GpuFace> hostFaces;
+					gpuFaces.copyToHost(hostFaces);
+					b3AlignedObjectArray<int> hostIndices;
+					gpuIndices.copyToHost(hostIndices);
+
+					for (int i = 0; i < nPairs; i++)
+					{
+						int bodyIndexA = hostPairs[i].x;
+						int bodyIndexB = hostPairs[i].y;
+
+						int collidableIndexA = hostBodyBuf[bodyIndexA].m_collidableIdx;
+						int collidableIndexB = hostBodyBuf[bodyIndexB].m_collidableIdx;
+
+						int shapeIndexA = hostCollidables[collidableIndexA].m_shapeIndex;
+						int shapeIndexB = hostCollidables[collidableIndexB].m_shapeIndex;
+
+						if (hostHasSepNormals[i])
+						{
+							b3FindClippingFaces(cpuSepNormals[i],
+												&hostConvexData[shapeIndexA],
+												&hostConvexData[shapeIndexB],
+												hostBodyBuf[bodyIndexA].m_pos, hostBodyBuf[bodyIndexA].m_quat,
+												hostBodyBuf[bodyIndexB].m_pos, hostBodyBuf[bodyIndexB].m_quat,
+												&worldVertsA1CPU.at(0), &worldNormalsACPU.at(0),
+												&worldVertsB1CPU.at(0),
+												vertexFaceCapacity, minDist, maxDist,
+												&hostVertices.at(0), &hostFaces.at(0),
+												&hostIndices.at(0),
+												&hostVertices.at(0), &hostFaces.at(0),
+												&hostIndices.at(0), &clippingFacesOutCPU.at(0), i);
+						}
+					}
+
+					clippingFacesOutGPU.copyFromHost(clippingFacesOutCPU);
+					worldVertsA1GPU.copyFromHost(worldVertsA1CPU);
+					worldNormalsAGPU.copyFromHost(worldNormalsACPU);
+					worldVertsB1GPU.copyFromHost(worldVertsB1CPU);
+				}
+
+				///clip face B against face A, reduce contacts and append them to a global contact array
+				if (1)
+				{
+					if (clipConvexFacesAndFindContactsCPU)
+					{
+						//b3AlignedObjectArray<b3Int4> hostPairs;
+						//pairs->copyToHost(hostPairs);
+
+						b3AlignedObjectArray<b3Vector3> hostSepNormals;
+						m_sepNormals.copyToHost(hostSepNormals);
+						b3AlignedObjectArray<int> hostHasSepAxis;
+						m_hasSeparatingNormals.copyToHost(hostHasSepAxis);
+
+						b3AlignedObjectArray<b3Int4> hostClippingFaces;
+						clippingFacesOutGPU.copyToHost(hostClippingFaces);
+						b3AlignedObjectArray<b3Vector3> worldVertsB2CPU;
+						worldVertsB2CPU.resize(vertexFaceCapacity * nPairs);
+
+						b3AlignedObjectArray<b3Vector3> worldVertsA1CPU;
+						worldVertsA1GPU.copyToHost(worldVertsA1CPU);
+						b3AlignedObjectArray<b3Vector3> worldNormalsACPU;
+						worldNormalsAGPU.copyToHost(worldNormalsACPU);
+
+						b3AlignedObjectArray<b3Vector3> worldVertsB1CPU;
+						worldVertsB1GPU.copyToHost(worldVertsB1CPU);
+
+						/*
 					  __global const b3Float4* separatingNormals,
                                                    __global const int* hasSeparatingAxis,
                                                    __global b3Int4* clippingFacesOut,
@@ -4520,214 +4202,207 @@ void GpuSatCollision::computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>*
                                                     int vertexFaceCapacity,
 															int pairIndex
 					*/
-					for (int i=0;i<nPairs;i++)
-					{
-						clipFacesAndFindContactsKernel(
-							&hostSepNormals.at(0),
-							&hostHasSepAxis.at(0),
-							&hostClippingFaces.at(0),
-							&worldVertsA1CPU.at(0),
-							&worldNormalsACPU.at(0),
-							&worldVertsB1CPU.at(0),
-							&worldVertsB2CPU.at(0),
-
-						vertexFaceCapacity,
-							i);
-					}
-					
-					clippingFacesOutGPU.copyFromHost(hostClippingFaces);
-					worldVertsB2GPU.copyFromHost(worldVertsB2CPU);
-
-				} else
-				{
-					B3_PROFILE("clipFacesAndFindContacts");
-					//nContacts = m_totalContactsOut.at(0);
-					//int h = m_hasSeparatingNormals.at(0);
-					//int4 p = clippingFacesOutGPU.at(0);
-					b3BufferInfoCL bInfo[] = {
-						b3BufferInfoCL( m_sepNormals.getBufferCL()),
-						b3BufferInfoCL( m_hasSeparatingNormals.getBufferCL()),
-						b3BufferInfoCL( clippingFacesOutGPU.getBufferCL()),
-						b3BufferInfoCL( worldVertsA1GPU.getBufferCL()),
-						b3BufferInfoCL( worldNormalsAGPU.getBufferCL()),
-						b3BufferInfoCL( worldVertsB1GPU.getBufferCL()),
-						b3BufferInfoCL( worldVertsB2GPU.getBufferCL())
-					};
-
-					b3LauncherCL launcher(m_queue, m_clipFacesAndFindContacts,"m_clipFacesAndFindContacts");
-					launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-					launcher.setConst(vertexFaceCapacity);
-
-					launcher.setConst( nPairs  );
-					int debugMode = 0;
-					launcher.setConst( debugMode);
-					int num = nPairs;
-					launcher.launch1D( num);
-					clFinish(m_queue);
-				} 
-
-				{
-					nContacts = m_totalContactsOut.at(0);
-					//printf("nContacts = %d\n",nContacts);
-
-					int newContactCapacity = nContacts+nPairs;
-					contactOut->reserve(newContactCapacity);
-
-					if (reduceConvexContactsOnGPU)
-					{
+						for (int i = 0; i < nPairs; i++)
 						{
-							B3_PROFILE("newContactReductionKernel");
-							b3BufferInfoCL bInfo[] =
-							{
-								b3BufferInfoCL( pairs->getBufferCL(), true ),
-								b3BufferInfoCL( bodyBuf->getBufferCL(),true),
-								b3BufferInfoCL( m_sepNormals.getBufferCL()),
-								b3BufferInfoCL( m_hasSeparatingNormals.getBufferCL()),
-								b3BufferInfoCL( contactOut->getBufferCL()),
-								b3BufferInfoCL( clippingFacesOutGPU.getBufferCL()),
-								b3BufferInfoCL( worldVertsB2GPU.getBufferCL()),
-								b3BufferInfoCL( m_totalContactsOut.getBufferCL())
-							};
-
-							b3LauncherCL launcher(m_queue, m_newContactReductionKernel,"m_newContactReductionKernel");
-							launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-							launcher.setConst(vertexFaceCapacity);
-							launcher.setConst(newContactCapacity);
-							launcher.setConst( nPairs  );
-							int num = nPairs;
-
-							launcher.launch1D( num);
-						}
-						nContacts = m_totalContactsOut.at(0);
-						contactOut->resize(nContacts);
-					} else
-					{
-
-						volatile int nGlobalContactsOut = nContacts;
-						b3AlignedObjectArray<b3Int4> hostPairs;
-						pairs->copyToHost(hostPairs);
-						b3AlignedObjectArray<b3RigidBodyData> hostBodyBuf;
-						bodyBuf->copyToHost(hostBodyBuf);
-						b3AlignedObjectArray<b3Vector3> hostSepNormals;
-						m_sepNormals.copyToHost(hostSepNormals);
-						b3AlignedObjectArray<int> hostHasSepAxis;
-						m_hasSeparatingNormals.copyToHost(hostHasSepAxis);
-						b3AlignedObjectArray<b3Contact4> hostContactsOut;
-						contactOut->copyToHost(hostContactsOut);
-						hostContactsOut.resize(newContactCapacity);
-
-						b3AlignedObjectArray<b3Int4> hostClippingFaces;
-						clippingFacesOutGPU.copyToHost(hostClippingFaces);
-						b3AlignedObjectArray<b3Vector3> worldVertsB2CPU;
-						worldVertsB2GPU.copyToHost(worldVertsB2CPU);
-
-						for (int i=0;i<nPairs;i++)
-						{
-							b3NewContactReductionKernel(&hostPairs.at(0),
-								&hostBodyBuf.at(0),
+							clipFacesAndFindContactsKernel(
 								&hostSepNormals.at(0),
 								&hostHasSepAxis.at(0),
-								&hostContactsOut.at(0),
 								&hostClippingFaces.at(0),
+								&worldVertsA1CPU.at(0),
+								&worldNormalsACPU.at(0),
+								&worldVertsB1CPU.at(0),
 								&worldVertsB2CPU.at(0),
-								&nGlobalContactsOut,
+
 								vertexFaceCapacity,
-								newContactCapacity,
-								nPairs,
 								i);
 						}
 
-						nContacts = nGlobalContactsOut;
-						m_totalContactsOut.copyFromHostPointer(&nContacts,1,0,true);
-						hostContactsOut.resize(nContacts);
-						//printf("contactOut4 (after newContactReductionKernel) = %d\n",nContacts);
-						contactOut->copyFromHost(hostContactsOut);
+						clippingFacesOutGPU.copyFromHost(hostClippingFaces);
+						worldVertsB2GPU.copyFromHost(worldVertsB2CPU);
+					}
+					else
+					{
+						B3_PROFILE("clipFacesAndFindContacts");
+						//nContacts = m_totalContactsOut.at(0);
+						//int h = m_hasSeparatingNormals.at(0);
+						//int4 p = clippingFacesOutGPU.at(0);
+						b3BufferInfoCL bInfo[] = {
+							b3BufferInfoCL(m_sepNormals.getBufferCL()),
+							b3BufferInfoCL(m_hasSeparatingNormals.getBufferCL()),
+							b3BufferInfoCL(clippingFacesOutGPU.getBufferCL()),
+							b3BufferInfoCL(worldVertsA1GPU.getBufferCL()),
+							b3BufferInfoCL(worldNormalsAGPU.getBufferCL()),
+							b3BufferInfoCL(worldVertsB1GPU.getBufferCL()),
+							b3BufferInfoCL(worldVertsB2GPU.getBufferCL())};
+
+						b3LauncherCL launcher(m_queue, m_clipFacesAndFindContacts, "m_clipFacesAndFindContacts");
+						launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+						launcher.setConst(vertexFaceCapacity);
+
+						launcher.setConst(nPairs);
+						int debugMode = 0;
+						launcher.setConst(debugMode);
+						int num = nPairs;
+						launcher.launch1D(num);
+						clFinish(m_queue);
+					}
+
+					{
+						nContacts = m_totalContactsOut.at(0);
+						//printf("nContacts = %d\n",nContacts);
+
+						int newContactCapacity = nContacts + nPairs;
+						contactOut->reserve(newContactCapacity);
+
+						if (reduceConvexContactsOnGPU)
+						{
+							{
+								B3_PROFILE("newContactReductionKernel");
+								b3BufferInfoCL bInfo[] =
+									{
+										b3BufferInfoCL(pairs->getBufferCL(), true),
+										b3BufferInfoCL(bodyBuf->getBufferCL(), true),
+										b3BufferInfoCL(m_sepNormals.getBufferCL()),
+										b3BufferInfoCL(m_hasSeparatingNormals.getBufferCL()),
+										b3BufferInfoCL(contactOut->getBufferCL()),
+										b3BufferInfoCL(clippingFacesOutGPU.getBufferCL()),
+										b3BufferInfoCL(worldVertsB2GPU.getBufferCL()),
+										b3BufferInfoCL(m_totalContactsOut.getBufferCL())};
+
+								b3LauncherCL launcher(m_queue, m_newContactReductionKernel, "m_newContactReductionKernel");
+								launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+								launcher.setConst(vertexFaceCapacity);
+								launcher.setConst(newContactCapacity);
+								launcher.setConst(nPairs);
+								int num = nPairs;
+
+								launcher.launch1D(num);
+							}
+							nContacts = m_totalContactsOut.at(0);
+							contactOut->resize(nContacts);
+						}
+						else
+						{
+							volatile int nGlobalContactsOut = nContacts;
+							b3AlignedObjectArray<b3Int4> hostPairs;
+							pairs->copyToHost(hostPairs);
+							b3AlignedObjectArray<b3RigidBodyData> hostBodyBuf;
+							bodyBuf->copyToHost(hostBodyBuf);
+							b3AlignedObjectArray<b3Vector3> hostSepNormals;
+							m_sepNormals.copyToHost(hostSepNormals);
+							b3AlignedObjectArray<int> hostHasSepAxis;
+							m_hasSeparatingNormals.copyToHost(hostHasSepAxis);
+							b3AlignedObjectArray<b3Contact4> hostContactsOut;
+							contactOut->copyToHost(hostContactsOut);
+							hostContactsOut.resize(newContactCapacity);
+
+							b3AlignedObjectArray<b3Int4> hostClippingFaces;
+							clippingFacesOutGPU.copyToHost(hostClippingFaces);
+							b3AlignedObjectArray<b3Vector3> worldVertsB2CPU;
+							worldVertsB2GPU.copyToHost(worldVertsB2CPU);
+
+							for (int i = 0; i < nPairs; i++)
+							{
+								b3NewContactReductionKernel(&hostPairs.at(0),
+															&hostBodyBuf.at(0),
+															&hostSepNormals.at(0),
+															&hostHasSepAxis.at(0),
+															&hostContactsOut.at(0),
+															&hostClippingFaces.at(0),
+															&worldVertsB2CPU.at(0),
+															&nGlobalContactsOut,
+															vertexFaceCapacity,
+															newContactCapacity,
+															nPairs,
+															i);
+							}
+
+							nContacts = nGlobalContactsOut;
+							m_totalContactsOut.copyFromHostPointer(&nContacts, 1, 0, true);
+							hostContactsOut.resize(nContacts);
+							//printf("contactOut4 (after newContactReductionKernel) = %d\n",nContacts);
+							contactOut->copyFromHost(hostContactsOut);
+						}
+						//                    b3Contact4 pt = contactOut->at(0);
+						//                  printf("nContacts = %d\n",nContacts);
 					}
-					//                    b3Contact4 pt = contactOut->at(0);
-					//                  printf("nContacts = %d\n",nContacts);
 				}
 			}
-		}            
-		else//breakupKernel
-		{
+			else  //breakupKernel
+			{
+				if (nPairs)
+				{
+					b3BufferInfoCL bInfo[] = {
+						b3BufferInfoCL(pairs->getBufferCL(), true),
+						b3BufferInfoCL(bodyBuf->getBufferCL(), true),
+						b3BufferInfoCL(gpuCollidables.getBufferCL(), true),
+						b3BufferInfoCL(convexData.getBufferCL(), true),
+						b3BufferInfoCL(gpuVertices.getBufferCL(), true),
+						b3BufferInfoCL(gpuUniqueEdges.getBufferCL(), true),
+						b3BufferInfoCL(gpuFaces.getBufferCL(), true),
+						b3BufferInfoCL(gpuIndices.getBufferCL(), true),
+						b3BufferInfoCL(m_sepNormals.getBufferCL()),
+						b3BufferInfoCL(m_hasSeparatingNormals.getBufferCL()),
+						b3BufferInfoCL(contactOut->getBufferCL()),
+						b3BufferInfoCL(m_totalContactsOut.getBufferCL())};
+					b3LauncherCL launcher(m_queue, m_clipHullHullKernel, "m_clipHullHullKernel");
+					launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+					launcher.setConst(nPairs);
+					launcher.setConst(maxContactCapacity);
 
-			if (nPairs)
+					int num = nPairs;
+					launcher.launch1D(num);
+					clFinish(m_queue);
+
+					nContacts = m_totalContactsOut.at(0);
+					if (nContacts >= maxContactCapacity)
+					{
+						b3Error("Exceeded contact capacity (%d/%d)\n", nContacts, maxContactCapacity);
+						nContacts = maxContactCapacity;
+					}
+					contactOut->resize(nContacts);
+				}
+			}
+
+			int nCompoundsPairs = m_gpuCompoundPairs.size();
+
+			if (nCompoundsPairs)
 			{
 				b3BufferInfoCL bInfo[] = {
-					b3BufferInfoCL( pairs->getBufferCL(), true ), 
-					b3BufferInfoCL( bodyBuf->getBufferCL(),true), 
-					b3BufferInfoCL( gpuCollidables.getBufferCL(),true), 
-					b3BufferInfoCL( convexData.getBufferCL(),true),
-					b3BufferInfoCL( gpuVertices.getBufferCL(),true),
-					b3BufferInfoCL( gpuUniqueEdges.getBufferCL(),true),
-					b3BufferInfoCL( gpuFaces.getBufferCL(),true),
-					b3BufferInfoCL( gpuIndices.getBufferCL(),true),
-					b3BufferInfoCL( m_sepNormals.getBufferCL()),
-					b3BufferInfoCL( m_hasSeparatingNormals.getBufferCL()),
-					b3BufferInfoCL( contactOut->getBufferCL()),
-					b3BufferInfoCL( m_totalContactsOut.getBufferCL())	
-				};
-				b3LauncherCL launcher(m_queue, m_clipHullHullKernel,"m_clipHullHullKernel");
-				launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-				launcher.setConst( nPairs  );
+					b3BufferInfoCL(m_gpuCompoundPairs.getBufferCL(), true),
+					b3BufferInfoCL(bodyBuf->getBufferCL(), true),
+					b3BufferInfoCL(gpuCollidables.getBufferCL(), true),
+					b3BufferInfoCL(convexData.getBufferCL(), true),
+					b3BufferInfoCL(gpuVertices.getBufferCL(), true),
+					b3BufferInfoCL(gpuUniqueEdges.getBufferCL(), true),
+					b3BufferInfoCL(gpuFaces.getBufferCL(), true),
+					b3BufferInfoCL(gpuIndices.getBufferCL(), true),
+					b3BufferInfoCL(gpuChildShapes.getBufferCL(), true),
+					b3BufferInfoCL(m_gpuCompoundSepNormals.getBufferCL(), true),
+					b3BufferInfoCL(m_gpuHasCompoundSepNormals.getBufferCL(), true),
+					b3BufferInfoCL(contactOut->getBufferCL()),
+					b3BufferInfoCL(m_totalContactsOut.getBufferCL())};
+				b3LauncherCL launcher(m_queue, m_clipCompoundsHullHullKernel, "m_clipCompoundsHullHullKernel");
+				launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+				launcher.setConst(nCompoundsPairs);
 				launcher.setConst(maxContactCapacity);
 
-				int num = nPairs;
-				launcher.launch1D( num);
+				int num = nCompoundsPairs;
+				launcher.launch1D(num);
 				clFinish(m_queue);
 
 				nContacts = m_totalContactsOut.at(0);
-				if (nContacts >= maxContactCapacity)
+				if (nContacts > maxContactCapacity)
 				{
-					b3Error("Exceeded contact capacity (%d/%d)\n",nContacts,maxContactCapacity);
+					b3Error("Error: contacts exceeds capacity (%d/%d)\n", nContacts, maxContactCapacity);
 					nContacts = maxContactCapacity;
 				}
 				contactOut->resize(nContacts);
-			}
+			}  //if nCompoundsPairs
 		}
-
-
-		int nCompoundsPairs = m_gpuCompoundPairs.size();
-
-		if (nCompoundsPairs)
-		{
-			b3BufferInfoCL bInfo[] = {
-				b3BufferInfoCL( m_gpuCompoundPairs.getBufferCL(), true ), 
-				b3BufferInfoCL( bodyBuf->getBufferCL(),true), 
-				b3BufferInfoCL( gpuCollidables.getBufferCL(),true), 
-				b3BufferInfoCL( convexData.getBufferCL(),true),
-				b3BufferInfoCL( gpuVertices.getBufferCL(),true),
-				b3BufferInfoCL( gpuUniqueEdges.getBufferCL(),true),
-				b3BufferInfoCL( gpuFaces.getBufferCL(),true),
-				b3BufferInfoCL( gpuIndices.getBufferCL(),true),
-				b3BufferInfoCL( gpuChildShapes.getBufferCL(),true),
-				b3BufferInfoCL( m_gpuCompoundSepNormals.getBufferCL(),true),
-				b3BufferInfoCL( m_gpuHasCompoundSepNormals.getBufferCL(),true),
-				b3BufferInfoCL( contactOut->getBufferCL()),
-				b3BufferInfoCL( m_totalContactsOut.getBufferCL())	
-			};
-			b3LauncherCL launcher(m_queue, m_clipCompoundsHullHullKernel,"m_clipCompoundsHullHullKernel");
-			launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-			launcher.setConst( nCompoundsPairs  );
-			launcher.setConst(maxContactCapacity);
-
-			int num = nCompoundsPairs;
-			launcher.launch1D( num);
-			clFinish(m_queue);
-
-			nContacts = m_totalContactsOut.at(0);
-			if (nContacts>maxContactCapacity)
-			{
-
-				b3Error("Error: contacts exceeds capacity (%d/%d)\n", nContacts, maxContactCapacity);
-				nContacts = maxContactCapacity;
-			}
-			contactOut->resize(nContacts);
-		}//if nCompoundsPairs
-		}
-	}//contactClippingOnGpu
+	}  //contactClippingOnGpu
 
 	//printf("nContacts end = %d\n",nContacts);
-	
+
 	//printf("frameCount = %d\n",frameCount++);
 }
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3ConvexHullContact.h b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3ConvexHullContact.h
index e24c1579c6d..53e8c4ed4d6 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3ConvexHullContact.h
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3ConvexHullContact.h
@@ -17,102 +17,90 @@
 
 //#include "../../dynamics/basic_demo/Stubs/ChNarrowPhase.h"
 
-
-
-
 struct GpuSatCollision
 {
-	cl_context				m_context;
-	cl_device_id			m_device;
-	cl_command_queue		m_queue;
-	cl_kernel				m_findSeparatingAxisKernel;
-	cl_kernel				m_mprPenetrationKernel;
-	cl_kernel				m_findSeparatingAxisUnitSphereKernel;
-
+	cl_context m_context;
+	cl_device_id m_device;
+	cl_command_queue m_queue;
+	cl_kernel m_findSeparatingAxisKernel;
+	cl_kernel m_mprPenetrationKernel;
+	cl_kernel m_findSeparatingAxisUnitSphereKernel;
 
 	cl_kernel m_findSeparatingAxisVertexFaceKernel;
 	cl_kernel m_findSeparatingAxisEdgeEdgeKernel;
-	
-	cl_kernel				m_findConcaveSeparatingAxisKernel;
-    cl_kernel				m_findConcaveSeparatingAxisVertexFaceKernel;
-    cl_kernel				m_findConcaveSeparatingAxisEdgeEdgeKernel;
- 
-    
-    
-    
-	cl_kernel				m_findCompoundPairsKernel;
-	cl_kernel				m_processCompoundPairsKernel;
 
-	cl_kernel				m_clipHullHullKernel;
-	cl_kernel				m_clipCompoundsHullHullKernel;
-    
-    cl_kernel               m_clipFacesAndFindContacts;
-    cl_kernel               m_findClippingFacesKernel;
-    
-	cl_kernel				m_clipHullHullConcaveConvexKernel;
-//	cl_kernel				m_extractManifoldAndAddContactKernel;
-    cl_kernel               m_newContactReductionKernel;
+	cl_kernel m_findConcaveSeparatingAxisKernel;
+	cl_kernel m_findConcaveSeparatingAxisVertexFaceKernel;
+	cl_kernel m_findConcaveSeparatingAxisEdgeEdgeKernel;
 
-	cl_kernel				m_bvhTraversalKernel;
-	cl_kernel				m_primitiveContactsKernel;
-	cl_kernel				m_findConcaveSphereContactsKernel;
+	cl_kernel m_findCompoundPairsKernel;
+	cl_kernel m_processCompoundPairsKernel;
+
+	cl_kernel m_clipHullHullKernel;
+	cl_kernel m_clipCompoundsHullHullKernel;
+
+	cl_kernel m_clipFacesAndFindContacts;
+	cl_kernel m_findClippingFacesKernel;
+
+	cl_kernel m_clipHullHullConcaveConvexKernel;
+	//	cl_kernel				m_extractManifoldAndAddContactKernel;
+	cl_kernel m_newContactReductionKernel;
+
+	cl_kernel m_bvhTraversalKernel;
+	cl_kernel m_primitiveContactsKernel;
+	cl_kernel m_findConcaveSphereContactsKernel;
+
+	cl_kernel m_processCompoundPairsPrimitivesKernel;
 
-	cl_kernel				m_processCompoundPairsPrimitivesKernel;
-    
 	b3OpenCLArray<b3Vector3> m_unitSphereDirections;
 
-	b3OpenCLArray<int>		m_totalContactsOut;
+	b3OpenCLArray<int> m_totalContactsOut;
 
 	b3OpenCLArray<b3Vector3> m_sepNormals;
 	b3OpenCLArray<float> m_dmins;
 
-	b3OpenCLArray<int>		m_hasSeparatingNormals;
+	b3OpenCLArray<int> m_hasSeparatingNormals;
 	b3OpenCLArray<b3Vector3> m_concaveSepNormals;
-	b3OpenCLArray<int>		m_concaveHasSeparatingNormals;
-	b3OpenCLArray<int>		m_numConcavePairsOut;
+	b3OpenCLArray<int> m_concaveHasSeparatingNormals;
+	b3OpenCLArray<int> m_numConcavePairsOut;
 	b3OpenCLArray<b3CompoundOverlappingPair> m_gpuCompoundPairs;
 	b3OpenCLArray<b3Vector3> m_gpuCompoundSepNormals;
-	b3OpenCLArray<int>		m_gpuHasCompoundSepNormals;
-	b3OpenCLArray<int>		m_numCompoundPairsOut;
-	
+	b3OpenCLArray<int> m_gpuHasCompoundSepNormals;
+	b3OpenCLArray<int> m_numCompoundPairsOut;
 
-	GpuSatCollision(cl_context ctx,cl_device_id device, cl_command_queue  q );
+	GpuSatCollision(cl_context ctx, cl_device_id device, cl_command_queue q);
 	virtual ~GpuSatCollision();
-	
 
-	void computeConvexConvexContactsGPUSAT( b3OpenCLArray<b3Int4>* pairs, int nPairs, 
-			const b3OpenCLArray<b3RigidBodyData>* bodyBuf,
-			b3OpenCLArray<b3Contact4>* contactOut, int& nContacts,
-			const b3OpenCLArray<b3Contact4>* oldContacts,
-			int maxContactCapacity,
-			int compoundPairCapacity,
-			const b3OpenCLArray<b3ConvexPolyhedronData>& hostConvexData,
-			const b3OpenCLArray<b3Vector3>& vertices,
-			const b3OpenCLArray<b3Vector3>& uniqueEdges,
-			const b3OpenCLArray<b3GpuFace>& faces,
-			const b3OpenCLArray<int>& indices,
-			const b3OpenCLArray<b3Collidable>& gpuCollidables,
-			const b3OpenCLArray<b3GpuChildShape>& gpuChildShapes,
-
-			const b3OpenCLArray<b3Aabb>& clAabbsWorldSpace,
-			const b3OpenCLArray<b3Aabb>& clAabbsLocalSpace,
-
-           b3OpenCLArray<b3Vector3>& worldVertsB1GPU,
-           b3OpenCLArray<b3Int4>& clippingFacesOutGPU,
-           b3OpenCLArray<b3Vector3>& worldNormalsAGPU,
-           b3OpenCLArray<b3Vector3>& worldVertsA1GPU,
-           b3OpenCLArray<b3Vector3>& worldVertsB2GPU,
-		   b3AlignedObjectArray<class b3OptimizedBvh*>& bvhData,
-		   b3OpenCLArray<b3QuantizedBvhNode>*	treeNodesGPU,
-			b3OpenCLArray<b3BvhSubtreeInfo>*	subTreesGPU,
-			b3OpenCLArray<b3BvhInfo>*	bvhInfo,
-			int numObjects,
-			int maxTriConvexPairCapacity,
-			b3OpenCLArray<b3Int4>& triangleConvexPairs,
-			int& numTriConvexPairsOut
-			);
+	void computeConvexConvexContactsGPUSAT(b3OpenCLArray<b3Int4>* pairs, int nPairs,
+										   const b3OpenCLArray<b3RigidBodyData>* bodyBuf,
+										   b3OpenCLArray<b3Contact4>* contactOut, int& nContacts,
+										   const b3OpenCLArray<b3Contact4>* oldContacts,
+										   int maxContactCapacity,
+										   int compoundPairCapacity,
+										   const b3OpenCLArray<b3ConvexPolyhedronData>& hostConvexData,
+										   const b3OpenCLArray<b3Vector3>& vertices,
+										   const b3OpenCLArray<b3Vector3>& uniqueEdges,
+										   const b3OpenCLArray<b3GpuFace>& faces,
+										   const b3OpenCLArray<int>& indices,
+										   const b3OpenCLArray<b3Collidable>& gpuCollidables,
+										   const b3OpenCLArray<b3GpuChildShape>& gpuChildShapes,
 
+										   const b3OpenCLArray<b3Aabb>& clAabbsWorldSpace,
+										   const b3OpenCLArray<b3Aabb>& clAabbsLocalSpace,
 
+										   b3OpenCLArray<b3Vector3>& worldVertsB1GPU,
+										   b3OpenCLArray<b3Int4>& clippingFacesOutGPU,
+										   b3OpenCLArray<b3Vector3>& worldNormalsAGPU,
+										   b3OpenCLArray<b3Vector3>& worldVertsA1GPU,
+										   b3OpenCLArray<b3Vector3>& worldVertsB2GPU,
+										   b3AlignedObjectArray<class b3OptimizedBvh*>& bvhData,
+										   b3OpenCLArray<b3QuantizedBvhNode>* treeNodesGPU,
+										   b3OpenCLArray<b3BvhSubtreeInfo>* subTreesGPU,
+										   b3OpenCLArray<b3BvhInfo>* bvhInfo,
+										   int numObjects,
+										   int maxTriConvexPairCapacity,
+										   b3OpenCLArray<b3Int4>& triangleConvexPairs,
+										   int& numTriConvexPairsOut);
 };
 
-#endif //_CONVEX_HULL_CONTACT_H
+#endif  //_CONVEX_HULL_CONTACT_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3ConvexPolyhedronCL.h b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3ConvexPolyhedronCL.h
index 337100fb1a6..c4cf7000765 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3ConvexPolyhedronCL.h
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3ConvexPolyhedronCL.h
@@ -4,6 +4,4 @@
 #include "Bullet3Common/b3Transform.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3ConvexPolyhedronData.h"
 
-
-
-#endif //CONVEX_POLYHEDRON_CL
+#endif  //CONVEX_POLYHEDRON_CL
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3GjkEpa.cpp b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3GjkEpa.cpp
index d636f983c69..974b246f030 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3GjkEpa.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3GjkEpa.cpp
@@ -29,902 +29,951 @@ GJK-EPA collision solver by Nathanael Presson, 2008
 
 namespace gjkepa2_impl2
 {
+// Config
 
-	// Config
+/* GJK	*/
+#define GJK_MAX_ITERATIONS 128
+#define GJK_ACCURACY ((b3Scalar)0.0001)
+#define GJK_MIN_DISTANCE ((b3Scalar)0.0001)
+#define GJK_DUPLICATED_EPS ((b3Scalar)0.0001)
+#define GJK_SIMPLEX2_EPS ((b3Scalar)0.0)
+#define GJK_SIMPLEX3_EPS ((b3Scalar)0.0)
+#define GJK_SIMPLEX4_EPS ((b3Scalar)0.0)
 
-	/* GJK	*/ 
-#define GJK_MAX_ITERATIONS	128
-#define GJK_ACCURACY		((b3Scalar)0.0001)
-#define GJK_MIN_DISTANCE	((b3Scalar)0.0001)
-#define GJK_DUPLICATED_EPS	((b3Scalar)0.0001)
-#define GJK_SIMPLEX2_EPS	((b3Scalar)0.0)
-#define GJK_SIMPLEX3_EPS	((b3Scalar)0.0)
-#define GJK_SIMPLEX4_EPS	((b3Scalar)0.0)
+/* EPA	*/
+#define EPA_MAX_VERTICES 64
+#define EPA_MAX_FACES (EPA_MAX_VERTICES * 2)
+#define EPA_MAX_ITERATIONS 255
+#define EPA_ACCURACY ((b3Scalar)0.0001)
+#define EPA_FALLBACK (10 * EPA_ACCURACY)
+#define EPA_PLANE_EPS ((b3Scalar)0.00001)
+#define EPA_INSIDE_EPS ((b3Scalar)0.01)
 
-	/* EPA	*/ 
-#define EPA_MAX_VERTICES	64
-#define EPA_MAX_FACES		(EPA_MAX_VERTICES*2)
-#define EPA_MAX_ITERATIONS	255
-#define EPA_ACCURACY		((b3Scalar)0.0001)
-#define EPA_FALLBACK		(10*EPA_ACCURACY)
-#define EPA_PLANE_EPS		((b3Scalar)0.00001)
-#define EPA_INSIDE_EPS		((b3Scalar)0.01)
+// Shorthands
 
+// MinkowskiDiff
+struct b3MinkowskiDiff
+{
+	const b3ConvexPolyhedronData* m_shapes[2];
 
-	// Shorthands
-	
+	b3Matrix3x3 m_toshape1;
+	b3Transform m_toshape0;
 
-	// MinkowskiDiff
-	struct	b3MinkowskiDiff
+	bool m_enableMargin;
+
+	void EnableMargin(bool enable)
 	{
-		
-
-		const b3ConvexPolyhedronData*	m_shapes[2];
-		
-
-		b3Matrix3x3				m_toshape1;
-		b3Transform				m_toshape0;
-
-		bool					m_enableMargin;
-		
-
-			void					EnableMargin(bool enable)
+		m_enableMargin = enable;
+	}
+	inline b3Vector3 Support0(const b3Vector3& d, const b3AlignedObjectArray<b3Vector3>& verticesA) const
+	{
+		if (m_enableMargin)
 		{
-			m_enableMargin = enable;
-		}	
-		inline b3Vector3		Support0(const b3Vector3& d, const b3AlignedObjectArray<b3Vector3>& verticesA) const
-		{
-			if (m_enableMargin)
-			{
-				return localGetSupportVertexWithMargin(d,m_shapes[0],verticesA,0.f);
-			} else
-			{
-				return localGetSupportVertexWithoutMargin(d,m_shapes[0],verticesA);
-			}
+			return localGetSupportVertexWithMargin(d, m_shapes[0], verticesA, 0.f);
 		}
-		inline b3Vector3		Support1(const b3Vector3& d, const b3AlignedObjectArray<b3Vector3>& verticesB) const
+		else
 		{
-			if (m_enableMargin)
-			{
-				return m_toshape0*(localGetSupportVertexWithMargin(m_toshape1*d,m_shapes[1],verticesB,0.f));
-			} else
-			{
-				return m_toshape0*(localGetSupportVertexWithoutMargin(m_toshape1*d,m_shapes[1],verticesB));
-			}
+			return localGetSupportVertexWithoutMargin(d, m_shapes[0], verticesA);
 		}
+	}
+	inline b3Vector3 Support1(const b3Vector3& d, const b3AlignedObjectArray<b3Vector3>& verticesB) const
+	{
+		if (m_enableMargin)
+		{
+			return m_toshape0 * (localGetSupportVertexWithMargin(m_toshape1 * d, m_shapes[1], verticesB, 0.f));
+		}
+		else
+		{
+			return m_toshape0 * (localGetSupportVertexWithoutMargin(m_toshape1 * d, m_shapes[1], verticesB));
+		}
+	}
 
-		inline b3Vector3		Support(const b3Vector3& d,  const b3AlignedObjectArray<b3Vector3>& verticesA,  const b3AlignedObjectArray<b3Vector3>& verticesB) const
-		{
-			return(Support0(d,verticesA)-Support1(-d,verticesB));
-		}
-		b3Vector3				Support(const b3Vector3& d,unsigned int index,const b3AlignedObjectArray<b3Vector3>& verticesA,  const b3AlignedObjectArray<b3Vector3>& verticesB) const
-		{
-			if(index)
-				return(Support1(d,verticesA));
-			else
-				return(Support0(d,verticesB));
-		}
+	inline b3Vector3 Support(const b3Vector3& d, const b3AlignedObjectArray<b3Vector3>& verticesA, const b3AlignedObjectArray<b3Vector3>& verticesB) const
+	{
+		return (Support0(d, verticesA) - Support1(-d, verticesB));
+	}
+	b3Vector3 Support(const b3Vector3& d, unsigned int index, const b3AlignedObjectArray<b3Vector3>& verticesA, const b3AlignedObjectArray<b3Vector3>& verticesB) const
+	{
+		if (index)
+			return (Support1(d, verticesA));
+		else
+			return (Support0(d, verticesB));
+	}
+};
+
+typedef b3MinkowskiDiff tShape;
+
+// GJK
+struct b3GJK
+{
+	/* Types		*/
+	struct sSV
+	{
+		b3Vector3 d, w;
 	};
-
-	typedef	b3MinkowskiDiff	tShape;
-
-
-	// GJK
-	struct	b3GJK
+	struct sSimplex
 	{
-		/* Types		*/ 
-		struct	sSV
+		sSV* c[4];
+		b3Scalar p[4];
+		unsigned int rank;
+	};
+	struct eStatus
+	{
+		enum _
 		{
-			b3Vector3	d,w;
-		};
-		struct	sSimplex
-		{
-			sSV*		c[4];
-			b3Scalar	p[4];
-			unsigned int			rank;
-		};
-		struct	eStatus	{ enum _ {
 			Valid,
 			Inside,
-			Failed		};};
-			/* Fields		*/ 
-			tShape			m_shape;
-			const b3AlignedObjectArray<b3Vector3>& m_verticesA;
-			const b3AlignedObjectArray<b3Vector3>& m_verticesB;
-			b3Vector3		m_ray;
-			b3Scalar		m_distance;
-			sSimplex		m_simplices[2];
-			sSV				m_store[4];
-			sSV*			m_free[4];
-			unsigned int				m_nfree;
-			unsigned int				m_current;
-			sSimplex*		m_simplex;
-			eStatus::_		m_status;
-			/* Methods		*/ 
-			b3GJK(const b3AlignedObjectArray<b3Vector3>& verticesA,const b3AlignedObjectArray<b3Vector3>& verticesB)
-				:m_verticesA(verticesA),m_verticesB(verticesB)
-			{
-				Initialize();
-			}
-			void				Initialize()
-			{
-				m_ray		=	b3MakeVector3(0,0,0);
-				m_nfree		=	0;
-				m_status	=	eStatus::Failed;
-				m_current	=	0;
-				m_distance	=	0;
-			}
-			eStatus::_			Evaluate(const tShape& shapearg,const b3Vector3& guess)
-			{
-				unsigned int			iterations=0;
-				b3Scalar	sqdist=0;
-				b3Scalar	alpha=0;
-				b3Vector3	lastw[4];
-				unsigned int			clastw=0;
-				/* Initialize solver		*/ 
-				m_free[0]			=	&m_store[0];
-				m_free[1]			=	&m_store[1];
-				m_free[2]			=	&m_store[2];
-				m_free[3]			=	&m_store[3];
-				m_nfree				=	4;
-				m_current			=	0;
-				m_status			=	eStatus::Valid;
-				m_shape				=	shapearg;
-				m_distance			=	0;
-				/* Initialize simplex		*/ 
-				m_simplices[0].rank	=	0;
-				m_ray				=	guess;
-				const b3Scalar	sqrl=	m_ray.length2();
-				appendvertice(m_simplices[0],sqrl>0?-m_ray:b3MakeVector3(1,0,0));
-				m_simplices[0].p[0]	=	1;
-				m_ray				=	m_simplices[0].c[0]->w;	
-				sqdist				=	sqrl;
-				lastw[0]			=
-					lastw[1]			=
-					lastw[2]			=
-					lastw[3]			=	m_ray;
-				/* Loop						*/ 
-				do	{
-					const unsigned int		next=1-m_current;
-					sSimplex&	cs=m_simplices[m_current];
-					sSimplex&	ns=m_simplices[next];
-					/* Check zero							*/ 
-					const b3Scalar	rl=m_ray.length();
-					if(rl<GJK_MIN_DISTANCE)
-					{/* Touching or inside				*/ 
-						m_status=eStatus::Inside;
-						break;
-					}
-					/* Append new vertice in -'v' direction	*/ 
-					appendvertice(cs,-m_ray);
-					const b3Vector3&	w=cs.c[cs.rank-1]->w;
-					bool				found=false;
-					for(unsigned int i=0;i<4;++i)
-					{
-						if((w-lastw[i]).length2()<GJK_DUPLICATED_EPS)
-						{ found=true;break; }
-					}
-					if(found)
-					{/* Return old simplex				*/ 
-						removevertice(m_simplices[m_current]);
-						break;
-					}
-					else
-					{/* Update lastw					*/ 
-						lastw[clastw=(clastw+1)&3]=w;
-					}
-					/* Check for termination				*/ 
-					const b3Scalar	omega=b3Dot(m_ray,w)/rl;
-					alpha=b3Max(omega,alpha);
-					if(((rl-alpha)-(GJK_ACCURACY*rl))<=0)
-					{/* Return old simplex				*/ 
-						removevertice(m_simplices[m_current]);
-						break;
-					}		
-					/* Reduce simplex						*/ 
-					b3Scalar	weights[4];
-					unsigned int			mask=0;
-					switch(cs.rank)
-					{
-					case	2:	sqdist=projectorigin(	cs.c[0]->w,
-									cs.c[1]->w,
-									weights,mask);break;
-					case	3:	sqdist=projectorigin(	cs.c[0]->w,
-									cs.c[1]->w,
-									cs.c[2]->w,
-									weights,mask);break;
-					case	4:	sqdist=projectorigin(	cs.c[0]->w,
-									cs.c[1]->w,
-									cs.c[2]->w,
-									cs.c[3]->w,
-									weights,mask);break;
-					}
-					if(sqdist>=0)
-					{/* Valid	*/ 
-						ns.rank		=	0;
-						m_ray		=	b3MakeVector3(0,0,0);
-						m_current	=	next;
-						for(unsigned int i=0,ni=cs.rank;i<ni;++i)
-						{
-							if(mask&(1<<i))
-							{
-								ns.c[ns.rank]		=	cs.c[i];
-								ns.p[ns.rank++]		=	weights[i];
-								m_ray				+=	cs.c[i]->w*weights[i];
-							}
-							else
-							{
-								m_free[m_nfree++]	=	cs.c[i];
-							}
-						}
-						if(mask==15) m_status=eStatus::Inside;
-					}
-					else
-					{/* Return old simplex				*/ 
-						removevertice(m_simplices[m_current]);
-						break;
-					}
-					m_status=((++iterations)<GJK_MAX_ITERATIONS)?m_status:eStatus::Failed;
-				} while(m_status==eStatus::Valid);
-				m_simplex=&m_simplices[m_current];
-				switch(m_status)
-				{
-				case	eStatus::Valid:		m_distance=m_ray.length();break;
-				case	eStatus::Inside:	m_distance=0;break;
-				default:
-					{
-					}
-				}	
-				return(m_status);
-			}
-			bool					EncloseOrigin()
-			{
-				switch(m_simplex->rank)
-				{
-				case	1:
-					{
-						for(unsigned int i=0;i<3;++i)
-						{
-							b3Vector3		axis=b3MakeVector3(0,0,0);
-							axis[i]=1;
-							appendvertice(*m_simplex, axis);
-							if(EncloseOrigin())	return(true);
-							removevertice(*m_simplex);
-							appendvertice(*m_simplex,-axis);
-							if(EncloseOrigin())	return(true);
-							removevertice(*m_simplex);
-						}
-					}
-					break;
-				case	2:
-					{
-						const b3Vector3	d=m_simplex->c[1]->w-m_simplex->c[0]->w;
-						for(unsigned int i=0;i<3;++i)
-						{
-							b3Vector3		axis=b3MakeVector3(0,0,0);
-							axis[i]=1;
-							const b3Vector3	p=b3Cross(d,axis);
-							if(p.length2()>0)
-							{
-								appendvertice(*m_simplex, p);
-								if(EncloseOrigin())	return(true);
-								removevertice(*m_simplex);
-								appendvertice(*m_simplex,-p);
-								if(EncloseOrigin())	return(true);
-								removevertice(*m_simplex);
-							}
-						}
-					}
-					break;
-				case	3:
-					{
-						const b3Vector3	n=b3Cross(m_simplex->c[1]->w-m_simplex->c[0]->w,
-							m_simplex->c[2]->w-m_simplex->c[0]->w);
-						if(n.length2()>0)
-						{
-							appendvertice(*m_simplex,n);
-							if(EncloseOrigin())	return(true);
-							removevertice(*m_simplex);
-							appendvertice(*m_simplex,-n);
-							if(EncloseOrigin())	return(true);
-							removevertice(*m_simplex);
-						}
-					}
-					break;
-				case	4:
-					{
-						if(b3Fabs(det(	m_simplex->c[0]->w-m_simplex->c[3]->w,
-							m_simplex->c[1]->w-m_simplex->c[3]->w,
-							m_simplex->c[2]->w-m_simplex->c[3]->w))>0)
-							return(true);
-					}
-					break;
-				}
-				return(false);
-			}
-			/* Internals	*/ 
-			void				getsupport(const b3Vector3& d,sSV& sv) const
-			{
-				sv.d	=	d/d.length();
-				sv.w	=	m_shape.Support(sv.d,m_verticesA,m_verticesB);
-			}
-			void				removevertice(sSimplex& simplex)
-			{
-				m_free[m_nfree++]=simplex.c[--simplex.rank];
-			}
-			void				appendvertice(sSimplex& simplex,const b3Vector3& v)
-			{
-				simplex.p[simplex.rank]=0;
-				simplex.c[simplex.rank]=m_free[--m_nfree];
-				getsupport(v,*simplex.c[simplex.rank++]);
-			}
-			static b3Scalar		det(const b3Vector3& a,const b3Vector3& b,const b3Vector3& c)
-			{
-				return(	a.y*b.z*c.x+a.z*b.x*c.y-
-					a.x*b.z*c.y-a.y*b.x*c.z+
-					a.x*b.y*c.z-a.z*b.y*c.x);
-			}
-			static b3Scalar		projectorigin(	const b3Vector3& a,
-				const b3Vector3& b,
-				b3Scalar* w,unsigned int& m)
-			{
-				const b3Vector3	d=b-a;
-				const b3Scalar	l=d.length2();
-				if(l>GJK_SIMPLEX2_EPS)
-				{
-					const b3Scalar	t(l>0?-b3Dot(a,d)/l:0);
-					if(t>=1)		{ w[0]=0;w[1]=1;m=2;return(b.length2()); }
-					else if(t<=0)	{ w[0]=1;w[1]=0;m=1;return(a.length2()); }
-					else			{ w[0]=1-(w[1]=t);m=3;return((a+d*t).length2()); }
-				}
-				return(-1);
-			}
-			static b3Scalar		projectorigin(	const b3Vector3& a,
-				const b3Vector3& b,
-				const b3Vector3& c,
-				b3Scalar* w,unsigned int& m)
-			{
-				static const unsigned int		imd3[]={1,2,0};
-				const b3Vector3*	vt[]={&a,&b,&c};
-				const b3Vector3		dl[]={a-b,b-c,c-a};
-				const b3Vector3		n=b3Cross(dl[0],dl[1]);
-				const b3Scalar		l=n.length2();
-				if(l>GJK_SIMPLEX3_EPS)
-				{
-					b3Scalar	mindist=-1;
-					b3Scalar	subw[2]={0.f,0.f};
-					unsigned int			subm(0);
-					for(unsigned int i=0;i<3;++i)
-					{
-						if(b3Dot(*vt[i],b3Cross(dl[i],n))>0)
-						{
-							const unsigned int			j=imd3[i];
-							const b3Scalar	subd(projectorigin(*vt[i],*vt[j],subw,subm));
-							if((mindist<0)||(subd<mindist))
-							{
-								mindist		=	subd;
-								m			=	static_cast<unsigned int>(((subm&1)?1<<i:0)+((subm&2)?1<<j:0));
-								w[i]		=	subw[0];
-								w[j]		=	subw[1];
-								w[imd3[j]]	=	0;				
-							}
-						}
-					}
-					if(mindist<0)
-					{
-						const b3Scalar	d=b3Dot(a,n);	
-						const b3Scalar	s=b3Sqrt(l);
-						const b3Vector3	p=n*(d/l);
-						mindist	=	p.length2();
-						m		=	7;
-						w[0]	=	(b3Cross(dl[1],b-p)).length()/s;
-						w[1]	=	(b3Cross(dl[2],c-p)).length()/s;
-						w[2]	=	1-(w[0]+w[1]);
-					}
-					return(mindist);
-				}
-				return(-1);
-			}
-			static b3Scalar		projectorigin(	const b3Vector3& a,
-				const b3Vector3& b,
-				const b3Vector3& c,
-				const b3Vector3& d,
-				b3Scalar* w,unsigned int& m)
-			{
-				static const unsigned int		imd3[]={1,2,0};
-				const b3Vector3*	vt[]={&a,&b,&c,&d};
-				const b3Vector3		dl[]={a-d,b-d,c-d};
-				const b3Scalar		vl=det(dl[0],dl[1],dl[2]);
-				const bool			ng=(vl*b3Dot(a,b3Cross(b-c,a-b)))<=0;
-				if(ng&&(b3Fabs(vl)>GJK_SIMPLEX4_EPS))
-				{
-					b3Scalar	mindist=-1;
-					b3Scalar	subw[3]={0.f,0.f,0.f};
-					unsigned int			subm(0);
-					for(unsigned int i=0;i<3;++i)
-					{
-						const unsigned int			j=imd3[i];
-						const b3Scalar	s=vl*b3Dot(d,b3Cross(dl[i],dl[j]));
-						if(s>0)
-						{
-							const b3Scalar	subd=projectorigin(*vt[i],*vt[j],d,subw,subm);
-							if((mindist<0)||(subd<mindist))
-							{
-								mindist		=	subd;
-								m			=	static_cast<unsigned int>((subm&1?1<<i:0)+
-									(subm&2?1<<j:0)+
-									(subm&4?8:0));
-								w[i]		=	subw[0];
-								w[j]		=	subw[1];
-								w[imd3[j]]	=	0;
-								w[3]		=	subw[2];
-							}
-						}
-					}
-					if(mindist<0)
-					{
-						mindist	=	0;
-						m		=	15;
-						w[0]	=	det(c,b,d)/vl;
-						w[1]	=	det(a,c,d)/vl;
-						w[2]	=	det(b,a,d)/vl;
-						w[3]	=	1-(w[0]+w[1]+w[2]);
-					}
-					return(mindist);
-				}
-				return(-1);
-			}
+			Failed
+		};
 	};
-
-	// EPA
-	struct	b3EPA
+	/* Fields		*/
+	tShape m_shape;
+	const b3AlignedObjectArray<b3Vector3>& m_verticesA;
+	const b3AlignedObjectArray<b3Vector3>& m_verticesB;
+	b3Vector3 m_ray;
+	b3Scalar m_distance;
+	sSimplex m_simplices[2];
+	sSV m_store[4];
+	sSV* m_free[4];
+	unsigned int m_nfree;
+	unsigned int m_current;
+	sSimplex* m_simplex;
+	eStatus::_ m_status;
+	/* Methods		*/
+	b3GJK(const b3AlignedObjectArray<b3Vector3>& verticesA, const b3AlignedObjectArray<b3Vector3>& verticesB)
+		: m_verticesA(verticesA), m_verticesB(verticesB)
 	{
-		/* Types		*/ 
-		typedef	b3GJK::sSV	sSV;
-		struct	sFace
+		Initialize();
+	}
+	void Initialize()
+	{
+		m_ray = b3MakeVector3(0, 0, 0);
+		m_nfree = 0;
+		m_status = eStatus::Failed;
+		m_current = 0;
+		m_distance = 0;
+	}
+	eStatus::_ Evaluate(const tShape& shapearg, const b3Vector3& guess)
+	{
+		unsigned int iterations = 0;
+		b3Scalar sqdist = 0;
+		b3Scalar alpha = 0;
+		b3Vector3 lastw[4];
+		unsigned int clastw = 0;
+		/* Initialize solver		*/
+		m_free[0] = &m_store[0];
+		m_free[1] = &m_store[1];
+		m_free[2] = &m_store[2];
+		m_free[3] = &m_store[3];
+		m_nfree = 4;
+		m_current = 0;
+		m_status = eStatus::Valid;
+		m_shape = shapearg;
+		m_distance = 0;
+		/* Initialize simplex		*/
+		m_simplices[0].rank = 0;
+		m_ray = guess;
+		const b3Scalar sqrl = m_ray.length2();
+		appendvertice(m_simplices[0], sqrl > 0 ? -m_ray : b3MakeVector3(1, 0, 0));
+		m_simplices[0].p[0] = 1;
+		m_ray = m_simplices[0].c[0]->w;
+		sqdist = sqrl;
+		lastw[0] =
+			lastw[1] =
+				lastw[2] =
+					lastw[3] = m_ray;
+		/* Loop						*/
+		do
 		{
-			b3Vector3	n;
-			b3Scalar	d;
-			sSV*		c[3];
-			sFace*		f[3];
-			sFace*		l[2];
-			unsigned char			e[3];
-			unsigned char			pass;
-		};
-		struct	sList
+			const unsigned int next = 1 - m_current;
+			sSimplex& cs = m_simplices[m_current];
+			sSimplex& ns = m_simplices[next];
+			/* Check zero							*/
+			const b3Scalar rl = m_ray.length();
+			if (rl < GJK_MIN_DISTANCE)
+			{ /* Touching or inside				*/
+				m_status = eStatus::Inside;
+				break;
+			}
+			/* Append new vertice in -'v' direction	*/
+			appendvertice(cs, -m_ray);
+			const b3Vector3& w = cs.c[cs.rank - 1]->w;
+			bool found = false;
+			for (unsigned int i = 0; i < 4; ++i)
+			{
+				if ((w - lastw[i]).length2() < GJK_DUPLICATED_EPS)
+				{
+					found = true;
+					break;
+				}
+			}
+			if (found)
+			{ /* Return old simplex				*/
+				removevertice(m_simplices[m_current]);
+				break;
+			}
+			else
+			{ /* Update lastw					*/
+				lastw[clastw = (clastw + 1) & 3] = w;
+			}
+			/* Check for termination				*/
+			const b3Scalar omega = b3Dot(m_ray, w) / rl;
+			alpha = b3Max(omega, alpha);
+			if (((rl - alpha) - (GJK_ACCURACY * rl)) <= 0)
+			{ /* Return old simplex				*/
+				removevertice(m_simplices[m_current]);
+				break;
+			}
+			/* Reduce simplex						*/
+			b3Scalar weights[4];
+			unsigned int mask = 0;
+			switch (cs.rank)
+			{
+				case 2:
+					sqdist = projectorigin(cs.c[0]->w,
+										   cs.c[1]->w,
+										   weights, mask);
+					break;
+				case 3:
+					sqdist = projectorigin(cs.c[0]->w,
+										   cs.c[1]->w,
+										   cs.c[2]->w,
+										   weights, mask);
+					break;
+				case 4:
+					sqdist = projectorigin(cs.c[0]->w,
+										   cs.c[1]->w,
+										   cs.c[2]->w,
+										   cs.c[3]->w,
+										   weights, mask);
+					break;
+			}
+			if (sqdist >= 0)
+			{ /* Valid	*/
+				ns.rank = 0;
+				m_ray = b3MakeVector3(0, 0, 0);
+				m_current = next;
+				for (unsigned int i = 0, ni = cs.rank; i < ni; ++i)
+				{
+					if (mask & (1 << i))
+					{
+						ns.c[ns.rank] = cs.c[i];
+						ns.p[ns.rank++] = weights[i];
+						m_ray += cs.c[i]->w * weights[i];
+					}
+					else
+					{
+						m_free[m_nfree++] = cs.c[i];
+					}
+				}
+				if (mask == 15) m_status = eStatus::Inside;
+			}
+			else
+			{ /* Return old simplex				*/
+				removevertice(m_simplices[m_current]);
+				break;
+			}
+			m_status = ((++iterations) < GJK_MAX_ITERATIONS) ? m_status : eStatus::Failed;
+		} while (m_status == eStatus::Valid);
+		m_simplex = &m_simplices[m_current];
+		switch (m_status)
 		{
-			sFace*		root;
-			unsigned int			count;
-			sList() : root(0),count(0)	{}
-		};
-		struct	sHorizon
+			case eStatus::Valid:
+				m_distance = m_ray.length();
+				break;
+			case eStatus::Inside:
+				m_distance = 0;
+				break;
+			default:
+			{
+			}
+		}
+		return (m_status);
+	}
+	bool EncloseOrigin()
+	{
+		switch (m_simplex->rank)
+		{
+			case 1:
+			{
+				for (unsigned int i = 0; i < 3; ++i)
+				{
+					b3Vector3 axis = b3MakeVector3(0, 0, 0);
+					axis[i] = 1;
+					appendvertice(*m_simplex, axis);
+					if (EncloseOrigin()) return (true);
+					removevertice(*m_simplex);
+					appendvertice(*m_simplex, -axis);
+					if (EncloseOrigin()) return (true);
+					removevertice(*m_simplex);
+				}
+			}
+			break;
+			case 2:
+			{
+				const b3Vector3 d = m_simplex->c[1]->w - m_simplex->c[0]->w;
+				for (unsigned int i = 0; i < 3; ++i)
+				{
+					b3Vector3 axis = b3MakeVector3(0, 0, 0);
+					axis[i] = 1;
+					const b3Vector3 p = b3Cross(d, axis);
+					if (p.length2() > 0)
+					{
+						appendvertice(*m_simplex, p);
+						if (EncloseOrigin()) return (true);
+						removevertice(*m_simplex);
+						appendvertice(*m_simplex, -p);
+						if (EncloseOrigin()) return (true);
+						removevertice(*m_simplex);
+					}
+				}
+			}
+			break;
+			case 3:
+			{
+				const b3Vector3 n = b3Cross(m_simplex->c[1]->w - m_simplex->c[0]->w,
+											m_simplex->c[2]->w - m_simplex->c[0]->w);
+				if (n.length2() > 0)
+				{
+					appendvertice(*m_simplex, n);
+					if (EncloseOrigin()) return (true);
+					removevertice(*m_simplex);
+					appendvertice(*m_simplex, -n);
+					if (EncloseOrigin()) return (true);
+					removevertice(*m_simplex);
+				}
+			}
+			break;
+			case 4:
+			{
+				if (b3Fabs(det(m_simplex->c[0]->w - m_simplex->c[3]->w,
+							   m_simplex->c[1]->w - m_simplex->c[3]->w,
+							   m_simplex->c[2]->w - m_simplex->c[3]->w)) > 0)
+					return (true);
+			}
+			break;
+		}
+		return (false);
+	}
+	/* Internals	*/
+	void getsupport(const b3Vector3& d, sSV& sv) const
+	{
+		sv.d = d / d.length();
+		sv.w = m_shape.Support(sv.d, m_verticesA, m_verticesB);
+	}
+	void removevertice(sSimplex& simplex)
+	{
+		m_free[m_nfree++] = simplex.c[--simplex.rank];
+	}
+	void appendvertice(sSimplex& simplex, const b3Vector3& v)
+	{
+		simplex.p[simplex.rank] = 0;
+		simplex.c[simplex.rank] = m_free[--m_nfree];
+		getsupport(v, *simplex.c[simplex.rank++]);
+	}
+	static b3Scalar det(const b3Vector3& a, const b3Vector3& b, const b3Vector3& c)
+	{
+		return (a.y * b.z * c.x + a.z * b.x * c.y -
+				a.x * b.z * c.y - a.y * b.x * c.z +
+				a.x * b.y * c.z - a.z * b.y * c.x);
+	}
+	static b3Scalar projectorigin(const b3Vector3& a,
+								  const b3Vector3& b,
+								  b3Scalar* w, unsigned int& m)
+	{
+		const b3Vector3 d = b - a;
+		const b3Scalar l = d.length2();
+		if (l > GJK_SIMPLEX2_EPS)
+		{
+			const b3Scalar t(l > 0 ? -b3Dot(a, d) / l : 0);
+			if (t >= 1)
+			{
+				w[0] = 0;
+				w[1] = 1;
+				m = 2;
+				return (b.length2());
+			}
+			else if (t <= 0)
+			{
+				w[0] = 1;
+				w[1] = 0;
+				m = 1;
+				return (a.length2());
+			}
+			else
+			{
+				w[0] = 1 - (w[1] = t);
+				m = 3;
+				return ((a + d * t).length2());
+			}
+		}
+		return (-1);
+	}
+	static b3Scalar projectorigin(const b3Vector3& a,
+								  const b3Vector3& b,
+								  const b3Vector3& c,
+								  b3Scalar* w, unsigned int& m)
+	{
+		static const unsigned int imd3[] = {1, 2, 0};
+		const b3Vector3* vt[] = {&a, &b, &c};
+		const b3Vector3 dl[] = {a - b, b - c, c - a};
+		const b3Vector3 n = b3Cross(dl[0], dl[1]);
+		const b3Scalar l = n.length2();
+		if (l > GJK_SIMPLEX3_EPS)
+		{
+			b3Scalar mindist = -1;
+			b3Scalar subw[2] = {0.f, 0.f};
+			unsigned int subm(0);
+			for (unsigned int i = 0; i < 3; ++i)
+			{
+				if (b3Dot(*vt[i], b3Cross(dl[i], n)) > 0)
+				{
+					const unsigned int j = imd3[i];
+					const b3Scalar subd(projectorigin(*vt[i], *vt[j], subw, subm));
+					if ((mindist < 0) || (subd < mindist))
+					{
+						mindist = subd;
+						m = static_cast<unsigned int>(((subm & 1) ? 1 << i : 0) + ((subm & 2) ? 1 << j : 0));
+						w[i] = subw[0];
+						w[j] = subw[1];
+						w[imd3[j]] = 0;
+					}
+				}
+			}
+			if (mindist < 0)
+			{
+				const b3Scalar d = b3Dot(a, n);
+				const b3Scalar s = b3Sqrt(l);
+				const b3Vector3 p = n * (d / l);
+				mindist = p.length2();
+				m = 7;
+				w[0] = (b3Cross(dl[1], b - p)).length() / s;
+				w[1] = (b3Cross(dl[2], c - p)).length() / s;
+				w[2] = 1 - (w[0] + w[1]);
+			}
+			return (mindist);
+		}
+		return (-1);
+	}
+	static b3Scalar projectorigin(const b3Vector3& a,
+								  const b3Vector3& b,
+								  const b3Vector3& c,
+								  const b3Vector3& d,
+								  b3Scalar* w, unsigned int& m)
+	{
+		static const unsigned int imd3[] = {1, 2, 0};
+		const b3Vector3* vt[] = {&a, &b, &c, &d};
+		const b3Vector3 dl[] = {a - d, b - d, c - d};
+		const b3Scalar vl = det(dl[0], dl[1], dl[2]);
+		const bool ng = (vl * b3Dot(a, b3Cross(b - c, a - b))) <= 0;
+		if (ng && (b3Fabs(vl) > GJK_SIMPLEX4_EPS))
+		{
+			b3Scalar mindist = -1;
+			b3Scalar subw[3] = {0.f, 0.f, 0.f};
+			unsigned int subm(0);
+			for (unsigned int i = 0; i < 3; ++i)
+			{
+				const unsigned int j = imd3[i];
+				const b3Scalar s = vl * b3Dot(d, b3Cross(dl[i], dl[j]));
+				if (s > 0)
+				{
+					const b3Scalar subd = projectorigin(*vt[i], *vt[j], d, subw, subm);
+					if ((mindist < 0) || (subd < mindist))
+					{
+						mindist = subd;
+						m = static_cast<unsigned int>((subm & 1 ? 1 << i : 0) +
+													  (subm & 2 ? 1 << j : 0) +
+													  (subm & 4 ? 8 : 0));
+						w[i] = subw[0];
+						w[j] = subw[1];
+						w[imd3[j]] = 0;
+						w[3] = subw[2];
+					}
+				}
+			}
+			if (mindist < 0)
+			{
+				mindist = 0;
+				m = 15;
+				w[0] = det(c, b, d) / vl;
+				w[1] = det(a, c, d) / vl;
+				w[2] = det(b, a, d) / vl;
+				w[3] = 1 - (w[0] + w[1] + w[2]);
+			}
+			return (mindist);
+		}
+		return (-1);
+	}
+};
+
+// EPA
+struct b3EPA
+{
+	/* Types		*/
+	typedef b3GJK::sSV sSV;
+	struct sFace
+	{
+		b3Vector3 n;
+		b3Scalar d;
+		sSV* c[3];
+		sFace* f[3];
+		sFace* l[2];
+		unsigned char e[3];
+		unsigned char pass;
+	};
+	struct sList
+	{
+		sFace* root;
+		unsigned int count;
+		sList() : root(0), count(0) {}
+	};
+	struct sHorizon
+	{
+		sFace* cf;
+		sFace* ff;
+		unsigned int nf;
+		sHorizon() : cf(0), ff(0), nf(0) {}
+	};
+	struct eStatus
+	{
+		enum _
 		{
-			sFace*		cf;
-			sFace*		ff;
-			unsigned int			nf;
-			sHorizon() : cf(0),ff(0),nf(0)	{}
-		};
-		struct	eStatus { enum _ {
 			Valid,
 			Touching,
 			Degenerated,
 			NonConvex,
-			InvalidHull,		
+			InvalidHull,
 			OutOfFaces,
 			OutOfVertices,
 			AccuraryReached,
 			FallBack,
-			Failed		};};
-			/* Fields		*/ 
-			eStatus::_		m_status;
-			b3GJK::sSimplex	m_result;
-			b3Vector3		m_normal;
-			b3Scalar		m_depth;
-			sSV				m_sv_store[EPA_MAX_VERTICES];
-			sFace			m_fc_store[EPA_MAX_FACES];
-			unsigned int				m_nextsv;
-			sList			m_hull;
-			sList			m_stock;
-			/* Methods		*/ 
-			b3EPA()
-			{
-				Initialize();	
-			}
-
-
-			static inline void		bind(sFace* fa,unsigned int ea,sFace* fb,unsigned int eb)
-			{
-				fa->e[ea]=(unsigned char)eb;fa->f[ea]=fb;
-				fb->e[eb]=(unsigned char)ea;fb->f[eb]=fa;
-			}
-			static inline void		append(sList& list,sFace* face)
-			{
-				face->l[0]	=	0;
-				face->l[1]	=	list.root;
-				if(list.root) list.root->l[0]=face;
-				list.root	=	face;
-				++list.count;
-			}
-			static inline void		remove(sList& list,sFace* face)
-			{
-				if(face->l[1]) face->l[1]->l[0]=face->l[0];
-				if(face->l[0]) face->l[0]->l[1]=face->l[1];
-				if(face==list.root) list.root=face->l[1];
-				--list.count;
-			}
-
-
-			void				Initialize()
-			{
-				m_status	=	eStatus::Failed;
-				m_normal	=	b3MakeVector3(0,0,0);
-				m_depth		=	0;
-				m_nextsv	=	0;
-				for(unsigned int i=0;i<EPA_MAX_FACES;++i)
-				{
-					append(m_stock,&m_fc_store[EPA_MAX_FACES-i-1]);
-				}
-			}
-			eStatus::_			Evaluate(b3GJK& gjk,const b3Vector3& guess)
-			{
-				b3GJK::sSimplex&	simplex=*gjk.m_simplex;
-				if((simplex.rank>1)&&gjk.EncloseOrigin())
-				{
-
-					/* Clean up				*/ 
-					while(m_hull.root)
-					{
-						sFace*	f = m_hull.root;
-						remove(m_hull,f);
-						append(m_stock,f);
-					}
-					m_status	=	eStatus::Valid;
-					m_nextsv	=	0;
-					/* Orient simplex		*/ 
-					if(gjk.det(	simplex.c[0]->w-simplex.c[3]->w,
-						simplex.c[1]->w-simplex.c[3]->w,
-						simplex.c[2]->w-simplex.c[3]->w)<0)
-					{
-						b3Swap(simplex.c[0],simplex.c[1]);
-						b3Swap(simplex.p[0],simplex.p[1]);
-					}
-					/* Build initial hull	*/ 
-					sFace*	tetra[]={newface(simplex.c[0],simplex.c[1],simplex.c[2],true),
-						newface(simplex.c[1],simplex.c[0],simplex.c[3],true),
-						newface(simplex.c[2],simplex.c[1],simplex.c[3],true),
-						newface(simplex.c[0],simplex.c[2],simplex.c[3],true)};
-					if(m_hull.count==4)
-					{
-						sFace*		best=findbest();
-						sFace		outer=*best;
-						unsigned int			pass=0;
-						unsigned int			iterations=0;
-						bind(tetra[0],0,tetra[1],0);
-						bind(tetra[0],1,tetra[2],0);
-						bind(tetra[0],2,tetra[3],0);
-						bind(tetra[1],1,tetra[3],2);
-						bind(tetra[1],2,tetra[2],1);
-						bind(tetra[2],2,tetra[3],1);
-						m_status=eStatus::Valid;
-						for(;iterations<EPA_MAX_ITERATIONS;++iterations)
-						{
-							if(m_nextsv<EPA_MAX_VERTICES)
-							{	
-								sHorizon		horizon;
-								sSV*			w=&m_sv_store[m_nextsv++];
-								bool			valid=true;					
-								best->pass	=	(unsigned char)(++pass);
-								gjk.getsupport(best->n,*w);
-								const b3Scalar	wdist=b3Dot(best->n,w->w)-best->d;
-								if(wdist>EPA_ACCURACY)
-								{
-									for(unsigned int j=0;(j<3)&&valid;++j)
-									{
-										valid&=expand(	pass,w,
-											best->f[j],best->e[j],
-											horizon);
-									}
-									if(valid&&(horizon.nf>=3))
-									{
-										bind(horizon.cf,1,horizon.ff,2);
-										remove(m_hull,best);
-										append(m_stock,best);
-										best=findbest();
-										outer=*best;
-									} else { 
-										m_status=eStatus::Failed;
-										//m_status=eStatus::InvalidHull;
-									break; }
-								} else { m_status=eStatus::AccuraryReached;break; }
-							} else { m_status=eStatus::OutOfVertices;break; }
-						}
-						const b3Vector3	projection=outer.n*outer.d;
-						m_normal	=	outer.n;
-						m_depth		=	outer.d;
-						m_result.rank	=	3;
-						m_result.c[0]	=	outer.c[0];
-						m_result.c[1]	=	outer.c[1];
-						m_result.c[2]	=	outer.c[2];
-						m_result.p[0]	=	b3Cross(	outer.c[1]->w-projection,
-							outer.c[2]->w-projection).length();
-						m_result.p[1]	=	b3Cross(	outer.c[2]->w-projection,
-							outer.c[0]->w-projection).length();
-						m_result.p[2]	=	b3Cross(	outer.c[0]->w-projection,
-							outer.c[1]->w-projection).length();
-						const b3Scalar	sum=m_result.p[0]+m_result.p[1]+m_result.p[2];
-						m_result.p[0]	/=	sum;
-						m_result.p[1]	/=	sum;
-						m_result.p[2]	/=	sum;
-						return(m_status);
-					}
-				}
-				/* Fallback		*/ 
-				m_status	=	eStatus::FallBack;
-				m_normal	=	-guess;
-				const b3Scalar	nl=m_normal.length();
-				if(nl>0)
-					m_normal	=	m_normal/nl;
-				else
-					m_normal	=	b3MakeVector3(1,0,0);
-				m_depth	=	0;
-				m_result.rank=1;
-				m_result.c[0]=simplex.c[0];
-				m_result.p[0]=1;	
-				return(m_status);
-			}
-			bool getedgedist(sFace* face, sSV* a, sSV* b, b3Scalar& dist)
-			{
-				const b3Vector3 ba = b->w - a->w;
-				const b3Vector3 n_ab = b3Cross(ba, face->n); // Outward facing edge normal direction, on triangle plane
-				const b3Scalar a_dot_nab = b3Dot(a->w, n_ab); // Only care about the sign to determine inside/outside, so not normalization required
-
-				if(a_dot_nab < 0)
-				{
-					// Outside of edge a->b
-
-					const b3Scalar ba_l2 = ba.length2();
-					const b3Scalar a_dot_ba = b3Dot(a->w, ba);
-					const b3Scalar b_dot_ba = b3Dot(b->w, ba);
-
-					if(a_dot_ba > 0)
-					{
-						// Pick distance vertex a
-						dist = a->w.length();
-					}
-					else if(b_dot_ba < 0)
-					{
-						// Pick distance vertex b
-						dist = b->w.length();
-					}
-					else
-					{
-						// Pick distance to edge a->b
-						const b3Scalar a_dot_b = b3Dot(a->w, b->w);
-						dist = b3Sqrt(b3Max((a->w.length2() * b->w.length2() - a_dot_b * a_dot_b) / ba_l2, (b3Scalar)0));
-					}
-
-					return true;
-				}
-
-				return false;
-			}
-			sFace*				newface(sSV* a,sSV* b,sSV* c,bool forced)
-			{
-				if(m_stock.root)
-				{
-					sFace*	face=m_stock.root;
-					remove(m_stock,face);
-					append(m_hull,face);
-					face->pass	=	0;
-					face->c[0]	=	a;
-					face->c[1]	=	b;
-					face->c[2]	=	c;
-					face->n		=	b3Cross(b->w-a->w,c->w-a->w);
-					const b3Scalar	l=face->n.length();
-					const bool		v=l>EPA_ACCURACY;
-
-					if(v)
-					{
-						if(!(getedgedist(face, a, b, face->d) ||
-							 getedgedist(face, b, c, face->d) ||
-							 getedgedist(face, c, a, face->d)))
-						{
-							// Origin projects to the interior of the triangle
-							// Use distance to triangle plane
-							face->d = b3Dot(a->w, face->n) / l;
-						}
-
-						face->n /= l;
-						if(forced || (face->d >= -EPA_PLANE_EPS))
-						{
-							return face;
-						}
-						else
-							m_status=eStatus::NonConvex;
-					}
-					else
-						m_status=eStatus::Degenerated;
-
-					remove(m_hull, face);
-					append(m_stock, face);
-					return 0;
-
-				}
-				m_status = m_stock.root ? eStatus::OutOfVertices : eStatus::OutOfFaces;
-				return 0;
-			}
-			sFace*				findbest()
-			{
-				sFace*		minf=m_hull.root;
-				b3Scalar	mind=minf->d*minf->d;
-				for(sFace* f=minf->l[1];f;f=f->l[1])
-				{
-					const b3Scalar	sqd=f->d*f->d;
-					if(sqd<mind)
-					{
-						minf=f;
-						mind=sqd;
-					}
-				}
-				return(minf);
-			}
-			bool				expand(unsigned int pass,sSV* w,sFace* f,unsigned int e,sHorizon& horizon)
-			{
-				static const unsigned int	i1m3[]={1,2,0};
-				static const unsigned int	i2m3[]={2,0,1};
-				if(f->pass!=pass)
-				{
-					const unsigned int	e1=i1m3[e];
-					if((b3Dot(f->n,w->w)-f->d)<-EPA_PLANE_EPS)
-					{
-						sFace*	nf=newface(f->c[e1],f->c[e],w,false);
-						if(nf)
-						{
-							bind(nf,0,f,e);
-							if(horizon.cf) bind(horizon.cf,1,nf,2); else horizon.ff=nf;
-							horizon.cf=nf;
-							++horizon.nf;
-							return(true);
-						}
-					}
-					else
-					{
-						const unsigned int	e2=i2m3[e];
-						f->pass		=	(unsigned char)pass;
-						if(	expand(pass,w,f->f[e1],f->e[e1],horizon)&&
-							expand(pass,w,f->f[e2],f->e[e2],horizon))
-						{
-							remove(m_hull,f);
-							append(m_stock,f);
-							return(true);
-						}
-					}
-				}
-				return(false);
-			}
-
+			Failed
+		};
 	};
-
-	//
-	static void	Initialize(const b3Transform&	transA, const b3Transform&	transB,
-								const b3ConvexPolyhedronData* hullA, const b3ConvexPolyhedronData* hullB, 
-								const b3AlignedObjectArray<b3Vector3>& verticesA,
-								const b3AlignedObjectArray<b3Vector3>& verticesB,
-		b3GjkEpaSolver2::sResults& results,
-		tShape& shape,
-		bool withmargins)
+	/* Fields		*/
+	eStatus::_ m_status;
+	b3GJK::sSimplex m_result;
+	b3Vector3 m_normal;
+	b3Scalar m_depth;
+	sSV m_sv_store[EPA_MAX_VERTICES];
+	sFace m_fc_store[EPA_MAX_FACES];
+	unsigned int m_nextsv;
+	sList m_hull;
+	sList m_stock;
+	/* Methods		*/
+	b3EPA()
 	{
-		/* Results		*/ 
-		results.witnesses[0]	=
-			results.witnesses[1]	=	b3MakeVector3(0,0,0);
-		results.status			=	b3GjkEpaSolver2::sResults::Separated;
-		/* Shape		*/ 
-		shape.m_shapes[0]		=	hullA;
-		shape.m_shapes[1]		=	hullB;
-		shape.m_toshape1		=	transB.getBasis().transposeTimes(transA.getBasis());
-		shape.m_toshape0		=	transA.inverseTimes(transB);
-		shape.EnableMargin(withmargins);
+		Initialize();
 	}
 
+	static inline void bind(sFace* fa, unsigned int ea, sFace* fb, unsigned int eb)
+	{
+		fa->e[ea] = (unsigned char)eb;
+		fa->f[ea] = fb;
+		fb->e[eb] = (unsigned char)ea;
+		fb->f[eb] = fa;
+	}
+	static inline void append(sList& list, sFace* face)
+	{
+		face->l[0] = 0;
+		face->l[1] = list.root;
+		if (list.root) list.root->l[0] = face;
+		list.root = face;
+		++list.count;
+	}
+	static inline void remove(sList& list, sFace* face)
+	{
+		if (face->l[1]) face->l[1]->l[0] = face->l[0];
+		if (face->l[0]) face->l[0]->l[1] = face->l[1];
+		if (face == list.root) list.root = face->l[1];
+		--list.count;
+	}
+
+	void Initialize()
+	{
+		m_status = eStatus::Failed;
+		m_normal = b3MakeVector3(0, 0, 0);
+		m_depth = 0;
+		m_nextsv = 0;
+		for (unsigned int i = 0; i < EPA_MAX_FACES; ++i)
+		{
+			append(m_stock, &m_fc_store[EPA_MAX_FACES - i - 1]);
+		}
+	}
+	eStatus::_ Evaluate(b3GJK& gjk, const b3Vector3& guess)
+	{
+		b3GJK::sSimplex& simplex = *gjk.m_simplex;
+		if ((simplex.rank > 1) && gjk.EncloseOrigin())
+		{
+			/* Clean up				*/
+			while (m_hull.root)
+			{
+				sFace* f = m_hull.root;
+				remove(m_hull, f);
+				append(m_stock, f);
+			}
+			m_status = eStatus::Valid;
+			m_nextsv = 0;
+			/* Orient simplex		*/
+			if (gjk.det(simplex.c[0]->w - simplex.c[3]->w,
+						simplex.c[1]->w - simplex.c[3]->w,
+						simplex.c[2]->w - simplex.c[3]->w) < 0)
+			{
+				b3Swap(simplex.c[0], simplex.c[1]);
+				b3Swap(simplex.p[0], simplex.p[1]);
+			}
+			/* Build initial hull	*/
+			sFace* tetra[] = {newface(simplex.c[0], simplex.c[1], simplex.c[2], true),
+							  newface(simplex.c[1], simplex.c[0], simplex.c[3], true),
+							  newface(simplex.c[2], simplex.c[1], simplex.c[3], true),
+							  newface(simplex.c[0], simplex.c[2], simplex.c[3], true)};
+			if (m_hull.count == 4)
+			{
+				sFace* best = findbest();
+				sFace outer = *best;
+				unsigned int pass = 0;
+				unsigned int iterations = 0;
+				bind(tetra[0], 0, tetra[1], 0);
+				bind(tetra[0], 1, tetra[2], 0);
+				bind(tetra[0], 2, tetra[3], 0);
+				bind(tetra[1], 1, tetra[3], 2);
+				bind(tetra[1], 2, tetra[2], 1);
+				bind(tetra[2], 2, tetra[3], 1);
+				m_status = eStatus::Valid;
+				for (; iterations < EPA_MAX_ITERATIONS; ++iterations)
+				{
+					if (m_nextsv < EPA_MAX_VERTICES)
+					{
+						sHorizon horizon;
+						sSV* w = &m_sv_store[m_nextsv++];
+						bool valid = true;
+						best->pass = (unsigned char)(++pass);
+						gjk.getsupport(best->n, *w);
+						const b3Scalar wdist = b3Dot(best->n, w->w) - best->d;
+						if (wdist > EPA_ACCURACY)
+						{
+							for (unsigned int j = 0; (j < 3) && valid; ++j)
+							{
+								valid &= expand(pass, w,
+												best->f[j], best->e[j],
+												horizon);
+							}
+							if (valid && (horizon.nf >= 3))
+							{
+								bind(horizon.cf, 1, horizon.ff, 2);
+								remove(m_hull, best);
+								append(m_stock, best);
+								best = findbest();
+								outer = *best;
+							}
+							else
+							{
+								m_status = eStatus::Failed;
+								//m_status=eStatus::InvalidHull;
+								break;
+							}
+						}
+						else
+						{
+							m_status = eStatus::AccuraryReached;
+							break;
+						}
+					}
+					else
+					{
+						m_status = eStatus::OutOfVertices;
+						break;
+					}
+				}
+				const b3Vector3 projection = outer.n * outer.d;
+				m_normal = outer.n;
+				m_depth = outer.d;
+				m_result.rank = 3;
+				m_result.c[0] = outer.c[0];
+				m_result.c[1] = outer.c[1];
+				m_result.c[2] = outer.c[2];
+				m_result.p[0] = b3Cross(outer.c[1]->w - projection,
+										outer.c[2]->w - projection)
+									.length();
+				m_result.p[1] = b3Cross(outer.c[2]->w - projection,
+										outer.c[0]->w - projection)
+									.length();
+				m_result.p[2] = b3Cross(outer.c[0]->w - projection,
+										outer.c[1]->w - projection)
+									.length();
+				const b3Scalar sum = m_result.p[0] + m_result.p[1] + m_result.p[2];
+				m_result.p[0] /= sum;
+				m_result.p[1] /= sum;
+				m_result.p[2] /= sum;
+				return (m_status);
+			}
+		}
+		/* Fallback		*/
+		m_status = eStatus::FallBack;
+		m_normal = -guess;
+		const b3Scalar nl = m_normal.length();
+		if (nl > 0)
+			m_normal = m_normal / nl;
+		else
+			m_normal = b3MakeVector3(1, 0, 0);
+		m_depth = 0;
+		m_result.rank = 1;
+		m_result.c[0] = simplex.c[0];
+		m_result.p[0] = 1;
+		return (m_status);
+	}
+	bool getedgedist(sFace* face, sSV* a, sSV* b, b3Scalar& dist)
+	{
+		const b3Vector3 ba = b->w - a->w;
+		const b3Vector3 n_ab = b3Cross(ba, face->n);   // Outward facing edge normal direction, on triangle plane
+		const b3Scalar a_dot_nab = b3Dot(a->w, n_ab);  // Only care about the sign to determine inside/outside, so not normalization required
+
+		if (a_dot_nab < 0)
+		{
+			// Outside of edge a->b
+
+			const b3Scalar ba_l2 = ba.length2();
+			const b3Scalar a_dot_ba = b3Dot(a->w, ba);
+			const b3Scalar b_dot_ba = b3Dot(b->w, ba);
+
+			if (a_dot_ba > 0)
+			{
+				// Pick distance vertex a
+				dist = a->w.length();
+			}
+			else if (b_dot_ba < 0)
+			{
+				// Pick distance vertex b
+				dist = b->w.length();
+			}
+			else
+			{
+				// Pick distance to edge a->b
+				const b3Scalar a_dot_b = b3Dot(a->w, b->w);
+				dist = b3Sqrt(b3Max((a->w.length2() * b->w.length2() - a_dot_b * a_dot_b) / ba_l2, (b3Scalar)0));
+			}
+
+			return true;
+		}
+
+		return false;
+	}
+	sFace* newface(sSV* a, sSV* b, sSV* c, bool forced)
+	{
+		if (m_stock.root)
+		{
+			sFace* face = m_stock.root;
+			remove(m_stock, face);
+			append(m_hull, face);
+			face->pass = 0;
+			face->c[0] = a;
+			face->c[1] = b;
+			face->c[2] = c;
+			face->n = b3Cross(b->w - a->w, c->w - a->w);
+			const b3Scalar l = face->n.length();
+			const bool v = l > EPA_ACCURACY;
+
+			if (v)
+			{
+				if (!(getedgedist(face, a, b, face->d) ||
+					  getedgedist(face, b, c, face->d) ||
+					  getedgedist(face, c, a, face->d)))
+				{
+					// Origin projects to the interior of the triangle
+					// Use distance to triangle plane
+					face->d = b3Dot(a->w, face->n) / l;
+				}
+
+				face->n /= l;
+				if (forced || (face->d >= -EPA_PLANE_EPS))
+				{
+					return face;
+				}
+				else
+					m_status = eStatus::NonConvex;
+			}
+			else
+				m_status = eStatus::Degenerated;
+
+			remove(m_hull, face);
+			append(m_stock, face);
+			return 0;
+		}
+		m_status = m_stock.root ? eStatus::OutOfVertices : eStatus::OutOfFaces;
+		return 0;
+	}
+	sFace* findbest()
+	{
+		sFace* minf = m_hull.root;
+		b3Scalar mind = minf->d * minf->d;
+		for (sFace* f = minf->l[1]; f; f = f->l[1])
+		{
+			const b3Scalar sqd = f->d * f->d;
+			if (sqd < mind)
+			{
+				minf = f;
+				mind = sqd;
+			}
+		}
+		return (minf);
+	}
+	bool expand(unsigned int pass, sSV* w, sFace* f, unsigned int e, sHorizon& horizon)
+	{
+		static const unsigned int i1m3[] = {1, 2, 0};
+		static const unsigned int i2m3[] = {2, 0, 1};
+		if (f->pass != pass)
+		{
+			const unsigned int e1 = i1m3[e];
+			if ((b3Dot(f->n, w->w) - f->d) < -EPA_PLANE_EPS)
+			{
+				sFace* nf = newface(f->c[e1], f->c[e], w, false);
+				if (nf)
+				{
+					bind(nf, 0, f, e);
+					if (horizon.cf)
+						bind(horizon.cf, 1, nf, 2);
+					else
+						horizon.ff = nf;
+					horizon.cf = nf;
+					++horizon.nf;
+					return (true);
+				}
+			}
+			else
+			{
+				const unsigned int e2 = i2m3[e];
+				f->pass = (unsigned char)pass;
+				if (expand(pass, w, f->f[e1], f->e[e1], horizon) &&
+					expand(pass, w, f->f[e2], f->e[e2], horizon))
+				{
+					remove(m_hull, f);
+					append(m_stock, f);
+					return (true);
+				}
+			}
+		}
+		return (false);
+	}
+};
+
+//
+static void Initialize(const b3Transform& transA, const b3Transform& transB,
+					   const b3ConvexPolyhedronData* hullA, const b3ConvexPolyhedronData* hullB,
+					   const b3AlignedObjectArray<b3Vector3>& verticesA,
+					   const b3AlignedObjectArray<b3Vector3>& verticesB,
+					   b3GjkEpaSolver2::sResults& results,
+					   tShape& shape,
+					   bool withmargins)
+{
+	/* Results		*/
+	results.witnesses[0] =
+		results.witnesses[1] = b3MakeVector3(0, 0, 0);
+	results.status = b3GjkEpaSolver2::sResults::Separated;
+	/* Shape		*/
+	shape.m_shapes[0] = hullA;
+	shape.m_shapes[1] = hullB;
+	shape.m_toshape1 = transB.getBasis().transposeTimes(transA.getBasis());
+	shape.m_toshape0 = transA.inverseTimes(transB);
+	shape.EnableMargin(withmargins);
 }
 
+}  // namespace gjkepa2_impl2
+
 //
 // Api
 //
 
-using namespace	gjkepa2_impl2;
+using namespace gjkepa2_impl2;
 
 //
-int			b3GjkEpaSolver2::StackSizeRequirement()
+int b3GjkEpaSolver2::StackSizeRequirement()
 {
-	return(sizeof(b3GJK)+sizeof(b3EPA));
+	return (sizeof(b3GJK) + sizeof(b3EPA));
 }
 
 //
-bool		b3GjkEpaSolver2::Distance(	const b3Transform&	transA, const b3Transform&	transB,
-										const b3ConvexPolyhedronData* hullA, const b3ConvexPolyhedronData* hullB, 
-										const b3AlignedObjectArray<b3Vector3>& verticesA,
-										const b3AlignedObjectArray<b3Vector3>& verticesB,
-									  const b3Vector3&		guess,
-									  sResults&				results)
+bool b3GjkEpaSolver2::Distance(const b3Transform& transA, const b3Transform& transB,
+							   const b3ConvexPolyhedronData* hullA, const b3ConvexPolyhedronData* hullB,
+							   const b3AlignedObjectArray<b3Vector3>& verticesA,
+							   const b3AlignedObjectArray<b3Vector3>& verticesB,
+							   const b3Vector3& guess,
+							   sResults& results)
 {
-	tShape			shape;
-	Initialize(transA,transB,hullA,hullB,verticesA,verticesB,results,shape,false);
-	b3GJK				gjk(verticesA,verticesB);
-	b3GJK::eStatus::_	gjk_status=gjk.Evaluate(shape,guess);
-	if(gjk_status==b3GJK::eStatus::Valid)
+	tShape shape;
+	Initialize(transA, transB, hullA, hullB, verticesA, verticesB, results, shape, false);
+	b3GJK gjk(verticesA, verticesB);
+	b3GJK::eStatus::_ gjk_status = gjk.Evaluate(shape, guess);
+	if (gjk_status == b3GJK::eStatus::Valid)
 	{
-		b3Vector3	w0=b3MakeVector3(0,0,0);
-		b3Vector3	w1=b3MakeVector3(0,0,0);
-		for(unsigned int i=0;i<gjk.m_simplex->rank;++i)
+		b3Vector3 w0 = b3MakeVector3(0, 0, 0);
+		b3Vector3 w1 = b3MakeVector3(0, 0, 0);
+		for (unsigned int i = 0; i < gjk.m_simplex->rank; ++i)
 		{
-			const b3Scalar	p=gjk.m_simplex->p[i];
-			w0+=shape.Support( gjk.m_simplex->c[i]->d,0,verticesA,verticesB)*p;
-			w1+=shape.Support(-gjk.m_simplex->c[i]->d,1,verticesA,verticesB)*p;
+			const b3Scalar p = gjk.m_simplex->p[i];
+			w0 += shape.Support(gjk.m_simplex->c[i]->d, 0, verticesA, verticesB) * p;
+			w1 += shape.Support(-gjk.m_simplex->c[i]->d, 1, verticesA, verticesB) * p;
 		}
-		results.witnesses[0]	=	transA*w0;
-		results.witnesses[1]	=	transA*w1;
-		results.normal			=	w0-w1;
-		results.distance		=	results.normal.length();
-		results.normal			/=	results.distance>GJK_MIN_DISTANCE?results.distance:1;
-		return(true);
+		results.witnesses[0] = transA * w0;
+		results.witnesses[1] = transA * w1;
+		results.normal = w0 - w1;
+		results.distance = results.normal.length();
+		results.normal /= results.distance > GJK_MIN_DISTANCE ? results.distance : 1;
+		return (true);
 	}
 	else
 	{
-		results.status	=	gjk_status==b3GJK::eStatus::Inside?
-			sResults::Penetrating	:
-		sResults::GJK_Failed	;
-		return(false);
+		results.status = gjk_status == b3GJK::eStatus::Inside ? sResults::Penetrating : sResults::GJK_Failed;
+		return (false);
 	}
 }
 
 //
-bool	b3GjkEpaSolver2::Penetration(	const b3Transform&	transA, const b3Transform&	transB,
-										const b3ConvexPolyhedronData* hullA, const b3ConvexPolyhedronData* hullB, 
-										const b3AlignedObjectArray<b3Vector3>& verticesA,
-										const b3AlignedObjectArray<b3Vector3>& verticesB,
-									 const b3Vector3&		guess,
-									 sResults&				results,
-									 bool					usemargins)
+bool b3GjkEpaSolver2::Penetration(const b3Transform& transA, const b3Transform& transB,
+								  const b3ConvexPolyhedronData* hullA, const b3ConvexPolyhedronData* hullB,
+								  const b3AlignedObjectArray<b3Vector3>& verticesA,
+								  const b3AlignedObjectArray<b3Vector3>& verticesB,
+								  const b3Vector3& guess,
+								  sResults& results,
+								  bool usemargins)
 {
-
-	tShape			shape;
-	Initialize(transA,transB,hullA,hullB,verticesA,verticesB,results,shape,usemargins);
-	b3GJK				gjk(verticesA,verticesB);
-	b3GJK::eStatus::_	gjk_status=gjk.Evaluate(shape,guess);
-	switch(gjk_status)
+	tShape shape;
+	Initialize(transA, transB, hullA, hullB, verticesA, verticesB, results, shape, usemargins);
+	b3GJK gjk(verticesA, verticesB);
+	b3GJK::eStatus::_ gjk_status = gjk.Evaluate(shape, guess);
+	switch (gjk_status)
 	{
-	case	b3GJK::eStatus::Inside:
+		case b3GJK::eStatus::Inside:
 		{
-			b3EPA				epa;
-			b3EPA::eStatus::_	epa_status=epa.Evaluate(gjk,-guess);
-			if(epa_status!=b3EPA::eStatus::Failed)
+			b3EPA epa;
+			b3EPA::eStatus::_ epa_status = epa.Evaluate(gjk, -guess);
+			if (epa_status != b3EPA::eStatus::Failed)
 			{
-				b3Vector3	w0=b3MakeVector3(0,0,0);
-				for(unsigned int i=0;i<epa.m_result.rank;++i)
+				b3Vector3 w0 = b3MakeVector3(0, 0, 0);
+				for (unsigned int i = 0; i < epa.m_result.rank; ++i)
 				{
-					w0+=shape.Support(epa.m_result.c[i]->d,0,verticesA,verticesB)*epa.m_result.p[i];
+					w0 += shape.Support(epa.m_result.c[i]->d, 0, verticesA, verticesB) * epa.m_result.p[i];
 				}
-				results.status			=	sResults::Penetrating;
-				results.witnesses[0]	=	transA*w0;
-				results.witnesses[1]	=	transA*(w0-epa.m_normal*epa.m_depth);
-				results.normal			=	-epa.m_normal;
-				results.distance		=	-epa.m_depth;
-				return(true);
-			} else results.status=sResults::EPA_Failed;
+				results.status = sResults::Penetrating;
+				results.witnesses[0] = transA * w0;
+				results.witnesses[1] = transA * (w0 - epa.m_normal * epa.m_depth);
+				results.normal = -epa.m_normal;
+				results.distance = -epa.m_depth;
+				return (true);
+			}
+			else
+				results.status = sResults::EPA_Failed;
 		}
 		break;
-	case	b3GJK::eStatus::Failed:
-		results.status=sResults::GJK_Failed;
-		break;
+		case b3GJK::eStatus::Failed:
+			results.status = sResults::GJK_Failed;
+			break;
 		default:
-					{
-					}
+		{
+		}
 	}
-	return(false);
+	return (false);
 }
 
-
 #if 0
 //
 b3Scalar	b3GjkEpaSolver2::SignedDistance(const b3Vector3& position,
@@ -994,8 +1043,7 @@ bool	b3GjkEpaSolver2::SignedDistance(const btConvexShape*	shape0,
 }
 #endif
 
-
-/* Symbols cleanup		*/ 
+/* Symbols cleanup		*/
 
 #undef GJK_MAX_ITERATIONS
 #undef GJK_ACCURACY
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3GjkEpa.h b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3GjkEpa.h
index 976238a04c8..7db32c63098 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3GjkEpa.h
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3GjkEpa.h
@@ -29,40 +29,39 @@ GJK-EPA collision solver by Nathanael Presson, 2008
 #include "Bullet3Common/b3Transform.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3ConvexPolyhedronData.h"
 
-
 ///btGjkEpaSolver contributed under zlib by Nathanael Presson
-struct	b3GjkEpaSolver2
+struct b3GjkEpaSolver2
 {
-struct	sResults
+	struct sResults
 	{
-	enum eStatus
+		enum eStatus
 		{
-		Separated,		/* Shapes doesnt penetrate												*/ 
-		Penetrating,	/* Shapes are penetrating												*/ 
-		GJK_Failed,		/* GJK phase fail, no big issue, shapes are probably just 'touching'	*/ 
-		EPA_Failed		/* EPA phase fail, bigger problem, need to save parameters, and debug	*/ 
-		}		status;
-	b3Vector3	witnesses[2];
-	b3Vector3	normal;
-	b3Scalar	distance;
+			Separated,   /* Shapes doesnt penetrate												*/
+			Penetrating, /* Shapes are penetrating												*/
+			GJK_Failed,  /* GJK phase fail, no big issue, shapes are probably just 'touching'	*/
+			EPA_Failed   /* EPA phase fail, bigger problem, need to save parameters, and debug	*/
+		} status;
+		b3Vector3 witnesses[2];
+		b3Vector3 normal;
+		b3Scalar distance;
 	};
 
-static int		StackSizeRequirement();
+	static int StackSizeRequirement();
 
-static bool		Distance(	 const b3Transform&	transA, const b3Transform&	transB,
-							const b3ConvexPolyhedronData* hullA, const b3ConvexPolyhedronData* hullB, 
-							const b3AlignedObjectArray<b3Vector3>& verticesA,
-							const b3AlignedObjectArray<b3Vector3>& verticesB,
-							const b3Vector3& guess,
-							sResults& results);
+	static bool Distance(const b3Transform& transA, const b3Transform& transB,
+						 const b3ConvexPolyhedronData* hullA, const b3ConvexPolyhedronData* hullB,
+						 const b3AlignedObjectArray<b3Vector3>& verticesA,
+						 const b3AlignedObjectArray<b3Vector3>& verticesB,
+						 const b3Vector3& guess,
+						 sResults& results);
 
-static bool		Penetration( const b3Transform&	transA, const b3Transform&	transB,
-							const b3ConvexPolyhedronData* hullA, const b3ConvexPolyhedronData* hullB, 
+	static bool Penetration(const b3Transform& transA, const b3Transform& transB,
+							const b3ConvexPolyhedronData* hullA, const b3ConvexPolyhedronData* hullB,
 							const b3AlignedObjectArray<b3Vector3>& verticesA,
 							const b3AlignedObjectArray<b3Vector3>& verticesB,
 							const b3Vector3& guess,
 							sResults& results,
-							bool usemargins=true);
+							bool usemargins = true);
 #if 0
 static b3Scalar	SignedDistance(	const b3Vector3& position,
 								b3Scalar margin,
@@ -74,9 +73,7 @@ static bool		SignedDistance(	const btConvexShape* shape0,const btTransform& wtrs
 								const btConvexShape* shape1,const btTransform& wtrs1,
 								const b3Vector3& guess,
 								sResults& results);
-#endif 
-
+#endif
 };
 
-#endif //B3_GJK_EPA2_H
-
+#endif  //B3_GJK_EPA2_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3OptimizedBvh.cpp b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3OptimizedBvh.cpp
index e9e51d5a36c..6f2c5251a0f 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3OptimizedBvh.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3OptimizedBvh.cpp
@@ -13,50 +13,45 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "b3OptimizedBvh.h"
 #include "b3StridingMeshInterface.h"
 #include "Bullet3Geometry/b3AabbUtil.h"
 
-
 b3OptimizedBvh::b3OptimizedBvh()
-{ 
+{
 }
 
 b3OptimizedBvh::~b3OptimizedBvh()
 {
 }
 
-
 void b3OptimizedBvh::build(b3StridingMeshInterface* triangles, bool useQuantizedAabbCompression, const b3Vector3& bvhAabbMin, const b3Vector3& bvhAabbMax)
 {
 	m_useQuantization = useQuantizedAabbCompression;
 
-
 	// NodeArray	triangleNodes;
 
-	struct	NodeTriangleCallback : public b3InternalTriangleIndexCallback
+	struct NodeTriangleCallback : public b3InternalTriangleIndexCallback
 	{
-
-		NodeArray&	m_triangleNodes;
+		NodeArray& m_triangleNodes;
 
 		NodeTriangleCallback& operator=(NodeTriangleCallback& other)
 		{
 			m_triangleNodes.copyFromArray(other.m_triangleNodes);
 			return *this;
 		}
-		
-		NodeTriangleCallback(NodeArray&	triangleNodes)
-			:m_triangleNodes(triangleNodes)
+
+		NodeTriangleCallback(NodeArray& triangleNodes)
+			: m_triangleNodes(triangleNodes)
 		{
 		}
 
-		virtual void internalProcessTriangleIndex(b3Vector3* triangle,int partId,int  triangleIndex)
+		virtual void internalProcessTriangleIndex(b3Vector3* triangle, int partId, int triangleIndex)
 		{
 			b3OptimizedBvhNode node;
-			b3Vector3	aabbMin,aabbMax;
-			aabbMin.setValue(b3Scalar(B3_LARGE_FLOAT),b3Scalar(B3_LARGE_FLOAT),b3Scalar(B3_LARGE_FLOAT));
-			aabbMax.setValue(b3Scalar(-B3_LARGE_FLOAT),b3Scalar(-B3_LARGE_FLOAT),b3Scalar(-B3_LARGE_FLOAT)); 
+			b3Vector3 aabbMin, aabbMax;
+			aabbMin.setValue(b3Scalar(B3_LARGE_FLOAT), b3Scalar(B3_LARGE_FLOAT), b3Scalar(B3_LARGE_FLOAT));
+			aabbMax.setValue(b3Scalar(-B3_LARGE_FLOAT), b3Scalar(-B3_LARGE_FLOAT), b3Scalar(-B3_LARGE_FLOAT));
 			aabbMin.setMin(triangle[0]);
 			aabbMax.setMax(triangle[0]);
 			aabbMin.setMin(triangle[1]);
@@ -69,17 +64,17 @@ void b3OptimizedBvh::build(b3StridingMeshInterface* triangles, bool useQuantized
 			node.m_aabbMaxOrg = aabbMax;
 
 			node.m_escapeIndex = -1;
-	
+
 			//for child nodes
 			node.m_subPart = partId;
 			node.m_triangleIndex = triangleIndex;
 			m_triangleNodes.push_back(node);
 		}
 	};
-	struct	QuantizedNodeTriangleCallback : public b3InternalTriangleIndexCallback
+	struct QuantizedNodeTriangleCallback : public b3InternalTriangleIndexCallback
 	{
-		QuantizedNodeArray&	m_triangleNodes;
-		const b3QuantizedBvh* m_optimizedTree; // for quantization
+		QuantizedNodeArray& m_triangleNodes;
+		const b3QuantizedBvh* m_optimizedTree;  // for quantization
 
 		QuantizedNodeTriangleCallback& operator=(QuantizedNodeTriangleCallback& other)
 		{
@@ -88,23 +83,23 @@ void b3OptimizedBvh::build(b3StridingMeshInterface* triangles, bool useQuantized
 			return *this;
 		}
 
-		QuantizedNodeTriangleCallback(QuantizedNodeArray&	triangleNodes,const b3QuantizedBvh* tree)
-			:m_triangleNodes(triangleNodes),m_optimizedTree(tree)
+		QuantizedNodeTriangleCallback(QuantizedNodeArray& triangleNodes, const b3QuantizedBvh* tree)
+			: m_triangleNodes(triangleNodes), m_optimizedTree(tree)
 		{
 		}
 
-		virtual void internalProcessTriangleIndex(b3Vector3* triangle,int partId,int  triangleIndex)
+		virtual void internalProcessTriangleIndex(b3Vector3* triangle, int partId, int triangleIndex)
 		{
 			// The partId and triangle index must fit in the same (positive) integer
-			b3Assert(partId < (1<<MAX_NUM_PARTS_IN_BITS));
-			b3Assert(triangleIndex < (1<<(31-MAX_NUM_PARTS_IN_BITS)));
+			b3Assert(partId < (1 << MAX_NUM_PARTS_IN_BITS));
+			b3Assert(triangleIndex < (1 << (31 - MAX_NUM_PARTS_IN_BITS)));
 			//negative indices are reserved for escapeIndex
-			b3Assert(triangleIndex>=0);
+			b3Assert(triangleIndex >= 0);
 
 			b3QuantizedBvhNode node;
-			b3Vector3	aabbMin,aabbMax;
-			aabbMin.setValue(b3Scalar(B3_LARGE_FLOAT),b3Scalar(B3_LARGE_FLOAT),b3Scalar(B3_LARGE_FLOAT));
-			aabbMax.setValue(b3Scalar(-B3_LARGE_FLOAT),b3Scalar(-B3_LARGE_FLOAT),b3Scalar(-B3_LARGE_FLOAT)); 
+			b3Vector3 aabbMin, aabbMax;
+			aabbMin.setValue(b3Scalar(B3_LARGE_FLOAT), b3Scalar(B3_LARGE_FLOAT), b3Scalar(B3_LARGE_FLOAT));
+			aabbMax.setValue(b3Scalar(-B3_LARGE_FLOAT), b3Scalar(-B3_LARGE_FLOAT), b3Scalar(-B3_LARGE_FLOAT));
 			aabbMin.setMin(triangle[0]);
 			aabbMax.setMax(triangle[0]);
 			aabbMin.setMin(triangle[1]);
@@ -131,59 +126,52 @@ void b3OptimizedBvh::build(b3StridingMeshInterface* triangles, bool useQuantized
 				aabbMin.setZ(aabbMin.getZ() - MIN_AABB_HALF_DIMENSION);
 			}
 
-			m_optimizedTree->quantize(&node.m_quantizedAabbMin[0],aabbMin,0);
-			m_optimizedTree->quantize(&node.m_quantizedAabbMax[0],aabbMax,1);
+			m_optimizedTree->quantize(&node.m_quantizedAabbMin[0], aabbMin, 0);
+			m_optimizedTree->quantize(&node.m_quantizedAabbMax[0], aabbMax, 1);
 
-			node.m_escapeIndexOrTriangleIndex = (partId<<(31-MAX_NUM_PARTS_IN_BITS)) | triangleIndex;
+			node.m_escapeIndexOrTriangleIndex = (partId << (31 - MAX_NUM_PARTS_IN_BITS)) | triangleIndex;
 
 			m_triangleNodes.push_back(node);
 		}
 	};
-	
-
 
 	int numLeafNodes = 0;
 
-	
 	if (m_useQuantization)
 	{
-
 		//initialize quantization values
-		setQuantizationValues(bvhAabbMin,bvhAabbMax);
+		setQuantizationValues(bvhAabbMin, bvhAabbMax);
 
-		QuantizedNodeTriangleCallback	callback(m_quantizedLeafNodes,this);
+		QuantizedNodeTriangleCallback callback(m_quantizedLeafNodes, this);
 
-	
-		triangles->InternalProcessAllTriangles(&callback,m_bvhAabbMin,m_bvhAabbMax);
+		triangles->InternalProcessAllTriangles(&callback, m_bvhAabbMin, m_bvhAabbMax);
 
 		//now we have an array of leafnodes in m_leafNodes
 		numLeafNodes = m_quantizedLeafNodes.size();
 
-
-		m_quantizedContiguousNodes.resize(2*numLeafNodes);
-
-
-	} else
+		m_quantizedContiguousNodes.resize(2 * numLeafNodes);
+	}
+	else
 	{
-		NodeTriangleCallback	callback(m_leafNodes);
+		NodeTriangleCallback callback(m_leafNodes);
 
-		b3Vector3 aabbMin=b3MakeVector3(b3Scalar(-B3_LARGE_FLOAT),b3Scalar(-B3_LARGE_FLOAT),b3Scalar(-B3_LARGE_FLOAT));
-		b3Vector3 aabbMax=b3MakeVector3(b3Scalar(B3_LARGE_FLOAT),b3Scalar(B3_LARGE_FLOAT),b3Scalar(B3_LARGE_FLOAT));
+		b3Vector3 aabbMin = b3MakeVector3(b3Scalar(-B3_LARGE_FLOAT), b3Scalar(-B3_LARGE_FLOAT), b3Scalar(-B3_LARGE_FLOAT));
+		b3Vector3 aabbMax = b3MakeVector3(b3Scalar(B3_LARGE_FLOAT), b3Scalar(B3_LARGE_FLOAT), b3Scalar(B3_LARGE_FLOAT));
 
-		triangles->InternalProcessAllTriangles(&callback,aabbMin,aabbMax);
+		triangles->InternalProcessAllTriangles(&callback, aabbMin, aabbMax);
 
 		//now we have an array of leafnodes in m_leafNodes
 		numLeafNodes = m_leafNodes.size();
 
-		m_contiguousNodes.resize(2*numLeafNodes);
+		m_contiguousNodes.resize(2 * numLeafNodes);
 	}
 
 	m_curNodeIndex = 0;
 
-	buildTree(0,numLeafNodes);
+	buildTree(0, numLeafNodes);
 
 	///if the entire tree is small then subtree size, we need to create a header info for the tree
-	if(m_useQuantization && !m_SubtreeHeaders.size())
+	if (m_useQuantization && !m_SubtreeHeaders.size())
 	{
 		b3BvhSubtreeInfo& subtree = m_SubtreeHeaders.expand();
 		subtree.setAabbFromQuantizeNode(m_quantizedContiguousNodes[0]);
@@ -199,37 +187,29 @@ void b3OptimizedBvh::build(b3StridingMeshInterface* triangles, bool useQuantized
 	m_leafNodes.clear();
 }
 
-
-
-
-void	b3OptimizedBvh::refit(b3StridingMeshInterface* meshInterface,const b3Vector3& aabbMin,const b3Vector3& aabbMax)
+void b3OptimizedBvh::refit(b3StridingMeshInterface* meshInterface, const b3Vector3& aabbMin, const b3Vector3& aabbMax)
 {
 	if (m_useQuantization)
 	{
+		setQuantizationValues(aabbMin, aabbMax);
 
-		setQuantizationValues(aabbMin,aabbMax);
-
-		updateBvhNodes(meshInterface,0,m_curNodeIndex,0);
+		updateBvhNodes(meshInterface, 0, m_curNodeIndex, 0);
 
 		///now update all subtree headers
 
 		int i;
-		for (i=0;i<m_SubtreeHeaders.size();i++)
+		for (i = 0; i < m_SubtreeHeaders.size(); i++)
 		{
 			b3BvhSubtreeInfo& subtree = m_SubtreeHeaders[i];
 			subtree.setAabbFromQuantizeNode(m_quantizedContiguousNodes[subtree.m_rootNodeIndex]);
 		}
-
-	} else
+	}
+	else
 	{
-
 	}
 }
 
-
-
-
-void	b3OptimizedBvh::refitPartial(b3StridingMeshInterface* meshInterface,const b3Vector3& aabbMin,const b3Vector3& aabbMax)
+void b3OptimizedBvh::refitPartial(b3StridingMeshInterface* meshInterface, const b3Vector3& aabbMin, const b3Vector3& aabbMax)
 {
 	//incrementally initialize quantization values
 	b3Assert(m_useQuantization);
@@ -244,147 +224,135 @@ void	b3OptimizedBvh::refitPartial(b3StridingMeshInterface* meshInterface,const b
 
 	///we should update all quantization values, using updateBvhNodes(meshInterface);
 	///but we only update chunks that overlap the given aabb
-	
-	unsigned short	quantizedQueryAabbMin[3];
-	unsigned short	quantizedQueryAabbMax[3];
 
-	quantize(&quantizedQueryAabbMin[0],aabbMin,0);
-	quantize(&quantizedQueryAabbMax[0],aabbMax,1);
+	unsigned short quantizedQueryAabbMin[3];
+	unsigned short quantizedQueryAabbMax[3];
+
+	quantize(&quantizedQueryAabbMin[0], aabbMin, 0);
+	quantize(&quantizedQueryAabbMax[0], aabbMax, 1);
 
 	int i;
-	for (i=0;i<this->m_SubtreeHeaders.size();i++)
+	for (i = 0; i < this->m_SubtreeHeaders.size(); i++)
 	{
 		b3BvhSubtreeInfo& subtree = m_SubtreeHeaders[i];
 
 		//PCK: unsigned instead of bool
-		unsigned overlap = b3TestQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin,quantizedQueryAabbMax,subtree.m_quantizedAabbMin,subtree.m_quantizedAabbMax);
+		unsigned overlap = b3TestQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin, quantizedQueryAabbMax, subtree.m_quantizedAabbMin, subtree.m_quantizedAabbMax);
 		if (overlap != 0)
 		{
-			updateBvhNodes(meshInterface,subtree.m_rootNodeIndex,subtree.m_rootNodeIndex+subtree.m_subtreeSize,i);
+			updateBvhNodes(meshInterface, subtree.m_rootNodeIndex, subtree.m_rootNodeIndex + subtree.m_subtreeSize, i);
 
 			subtree.setAabbFromQuantizeNode(m_quantizedContiguousNodes[subtree.m_rootNodeIndex]);
 		}
 	}
-	
 }
 
-void	b3OptimizedBvh::updateBvhNodes(b3StridingMeshInterface* meshInterface,int firstNode,int endNode,int index)
+void b3OptimizedBvh::updateBvhNodes(b3StridingMeshInterface* meshInterface, int firstNode, int endNode, int index)
 {
 	(void)index;
 
 	b3Assert(m_useQuantization);
 
-	int curNodeSubPart=-1;
+	int curNodeSubPart = -1;
 
 	//get access info to trianglemesh data
-		const unsigned char *vertexbase = 0;
-		int numverts = 0;
-		PHY_ScalarType type = PHY_INTEGER;
-		int stride = 0;
-		const unsigned char *indexbase = 0;
-		int indexstride = 0;
-		int numfaces = 0;
-		PHY_ScalarType indicestype = PHY_INTEGER;
+	const unsigned char* vertexbase = 0;
+	int numverts = 0;
+	PHY_ScalarType type = PHY_INTEGER;
+	int stride = 0;
+	const unsigned char* indexbase = 0;
+	int indexstride = 0;
+	int numfaces = 0;
+	PHY_ScalarType indicestype = PHY_INTEGER;
 
-		b3Vector3	triangleVerts[3];
-		b3Vector3	aabbMin,aabbMax;
-		const b3Vector3& meshScaling = meshInterface->getScaling();
-		
-		int i;
-		for (i=endNode-1;i>=firstNode;i--)
+	b3Vector3 triangleVerts[3];
+	b3Vector3 aabbMin, aabbMax;
+	const b3Vector3& meshScaling = meshInterface->getScaling();
+
+	int i;
+	for (i = endNode - 1; i >= firstNode; i--)
+	{
+		b3QuantizedBvhNode& curNode = m_quantizedContiguousNodes[i];
+		if (curNode.isLeafNode())
 		{
-
-
-			b3QuantizedBvhNode& curNode = m_quantizedContiguousNodes[i];
-			if (curNode.isLeafNode())
+			//recalc aabb from triangle data
+			int nodeSubPart = curNode.getPartId();
+			int nodeTriangleIndex = curNode.getTriangleIndex();
+			if (nodeSubPart != curNodeSubPart)
 			{
-				//recalc aabb from triangle data
-				int nodeSubPart = curNode.getPartId();
-				int nodeTriangleIndex = curNode.getTriangleIndex();
-				if (nodeSubPart != curNodeSubPart)
-				{
-					if (curNodeSubPart >= 0)
-						meshInterface->unLockReadOnlyVertexBase(curNodeSubPart);
-					meshInterface->getLockedReadOnlyVertexIndexBase(&vertexbase,numverts,	type,stride,&indexbase,indexstride,numfaces,indicestype,nodeSubPart);
+				if (curNodeSubPart >= 0)
+					meshInterface->unLockReadOnlyVertexBase(curNodeSubPart);
+				meshInterface->getLockedReadOnlyVertexIndexBase(&vertexbase, numverts, type, stride, &indexbase, indexstride, numfaces, indicestype, nodeSubPart);
 
-					curNodeSubPart = nodeSubPart;
-					b3Assert(indicestype==PHY_INTEGER||indicestype==PHY_SHORT);
-				}
-				//triangles->getLockedReadOnlyVertexIndexBase(vertexBase,numVerts,
+				curNodeSubPart = nodeSubPart;
+				b3Assert(indicestype == PHY_INTEGER || indicestype == PHY_SHORT);
+			}
+			//triangles->getLockedReadOnlyVertexIndexBase(vertexBase,numVerts,
 
-				unsigned int* gfxbase = (unsigned int*)(indexbase+nodeTriangleIndex*indexstride);
-				
-				
-				for (int j=2;j>=0;j--)
-				{
-					
-					int graphicsindex = indicestype==PHY_SHORT?((unsigned short*)gfxbase)[j]:gfxbase[j];
-					if (type == PHY_FLOAT)
-					{
-						float* graphicsbase = (float*)(vertexbase+graphicsindex*stride);
-						triangleVerts[j] = b3MakeVector3(
-							graphicsbase[0]*meshScaling.getX(),
-							graphicsbase[1]*meshScaling.getY(),
-							graphicsbase[2]*meshScaling.getZ());
-					}
-					else
-					{
-						double* graphicsbase = (double*)(vertexbase+graphicsindex*stride);
-						triangleVerts[j] = b3MakeVector3( b3Scalar(graphicsbase[0]*meshScaling.getX()), b3Scalar(graphicsbase[1]*meshScaling.getY()), b3Scalar(graphicsbase[2]*meshScaling.getZ()));
-					}
-				}
+			unsigned int* gfxbase = (unsigned int*)(indexbase + nodeTriangleIndex * indexstride);
 
-
-				
-				aabbMin.setValue(b3Scalar(B3_LARGE_FLOAT),b3Scalar(B3_LARGE_FLOAT),b3Scalar(B3_LARGE_FLOAT));
-				aabbMax.setValue(b3Scalar(-B3_LARGE_FLOAT),b3Scalar(-B3_LARGE_FLOAT),b3Scalar(-B3_LARGE_FLOAT)); 
-				aabbMin.setMin(triangleVerts[0]);
-				aabbMax.setMax(triangleVerts[0]);
-				aabbMin.setMin(triangleVerts[1]);
-				aabbMax.setMax(triangleVerts[1]);
-				aabbMin.setMin(triangleVerts[2]);
-				aabbMax.setMax(triangleVerts[2]);
-
-				quantize(&curNode.m_quantizedAabbMin[0],aabbMin,0);
-				quantize(&curNode.m_quantizedAabbMax[0],aabbMax,1);
-				
-			} else
+			for (int j = 2; j >= 0; j--)
 			{
-				//combine aabb from both children
-
-				b3QuantizedBvhNode* leftChildNode = &m_quantizedContiguousNodes[i+1];
-				
-				b3QuantizedBvhNode* rightChildNode = leftChildNode->isLeafNode() ? &m_quantizedContiguousNodes[i+2] :
-					&m_quantizedContiguousNodes[i+1+leftChildNode->getEscapeIndex()];
-				
-
+				int graphicsindex = indicestype == PHY_SHORT ? ((unsigned short*)gfxbase)[j] : gfxbase[j];
+				if (type == PHY_FLOAT)
 				{
-					for (int i=0;i<3;i++)
-					{
-						curNode.m_quantizedAabbMin[i] = leftChildNode->m_quantizedAabbMin[i];
-						if (curNode.m_quantizedAabbMin[i]>rightChildNode->m_quantizedAabbMin[i])
-							curNode.m_quantizedAabbMin[i]=rightChildNode->m_quantizedAabbMin[i];
-
-						curNode.m_quantizedAabbMax[i] = leftChildNode->m_quantizedAabbMax[i];
-						if (curNode.m_quantizedAabbMax[i] < rightChildNode->m_quantizedAabbMax[i])
-							curNode.m_quantizedAabbMax[i] = rightChildNode->m_quantizedAabbMax[i];
-					}
+					float* graphicsbase = (float*)(vertexbase + graphicsindex * stride);
+					triangleVerts[j] = b3MakeVector3(
+						graphicsbase[0] * meshScaling.getX(),
+						graphicsbase[1] * meshScaling.getY(),
+						graphicsbase[2] * meshScaling.getZ());
+				}
+				else
+				{
+					double* graphicsbase = (double*)(vertexbase + graphicsindex * stride);
+					triangleVerts[j] = b3MakeVector3(b3Scalar(graphicsbase[0] * meshScaling.getX()), b3Scalar(graphicsbase[1] * meshScaling.getY()), b3Scalar(graphicsbase[2] * meshScaling.getZ()));
 				}
 			}
 
+			aabbMin.setValue(b3Scalar(B3_LARGE_FLOAT), b3Scalar(B3_LARGE_FLOAT), b3Scalar(B3_LARGE_FLOAT));
+			aabbMax.setValue(b3Scalar(-B3_LARGE_FLOAT), b3Scalar(-B3_LARGE_FLOAT), b3Scalar(-B3_LARGE_FLOAT));
+			aabbMin.setMin(triangleVerts[0]);
+			aabbMax.setMax(triangleVerts[0]);
+			aabbMin.setMin(triangleVerts[1]);
+			aabbMax.setMax(triangleVerts[1]);
+			aabbMin.setMin(triangleVerts[2]);
+			aabbMax.setMax(triangleVerts[2]);
+
+			quantize(&curNode.m_quantizedAabbMin[0], aabbMin, 0);
+			quantize(&curNode.m_quantizedAabbMax[0], aabbMax, 1);
 		}
+		else
+		{
+			//combine aabb from both children
 
-		if (curNodeSubPart >= 0)
-			meshInterface->unLockReadOnlyVertexBase(curNodeSubPart);
+			b3QuantizedBvhNode* leftChildNode = &m_quantizedContiguousNodes[i + 1];
 
-		
+			b3QuantizedBvhNode* rightChildNode = leftChildNode->isLeafNode() ? &m_quantizedContiguousNodes[i + 2] : &m_quantizedContiguousNodes[i + 1 + leftChildNode->getEscapeIndex()];
+
+			{
+				for (int i = 0; i < 3; i++)
+				{
+					curNode.m_quantizedAabbMin[i] = leftChildNode->m_quantizedAabbMin[i];
+					if (curNode.m_quantizedAabbMin[i] > rightChildNode->m_quantizedAabbMin[i])
+						curNode.m_quantizedAabbMin[i] = rightChildNode->m_quantizedAabbMin[i];
+
+					curNode.m_quantizedAabbMax[i] = leftChildNode->m_quantizedAabbMax[i];
+					if (curNode.m_quantizedAabbMax[i] < rightChildNode->m_quantizedAabbMax[i])
+						curNode.m_quantizedAabbMax[i] = rightChildNode->m_quantizedAabbMax[i];
+				}
+			}
+		}
+	}
+
+	if (curNodeSubPart >= 0)
+		meshInterface->unLockReadOnlyVertexBase(curNodeSubPart);
 }
 
 ///deSerializeInPlace loads and initializes a BVH from a buffer in memory 'in place'
-b3OptimizedBvh* b3OptimizedBvh::deSerializeInPlace(void *i_alignedDataBuffer, unsigned int i_dataBufferSize, bool i_swapEndian)
+b3OptimizedBvh* b3OptimizedBvh::deSerializeInPlace(void* i_alignedDataBuffer, unsigned int i_dataBufferSize, bool i_swapEndian)
 {
-	b3QuantizedBvh* bvh = b3QuantizedBvh::deSerializeInPlace(i_alignedDataBuffer,i_dataBufferSize,i_swapEndian);
-	
+	b3QuantizedBvh* bvh = b3QuantizedBvh::deSerializeInPlace(i_alignedDataBuffer, i_dataBufferSize, i_swapEndian);
+
 	//we don't add additional data so just do a static upcast
 	return static_cast<b3OptimizedBvh*>(bvh);
 }
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3OptimizedBvh.h b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3OptimizedBvh.h
index 0272ef83bfc..12865529399 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3OptimizedBvh.h
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3OptimizedBvh.h
@@ -22,44 +22,35 @@ subject to the following restrictions:
 
 class b3StridingMeshInterface;
 
-
 ///The b3OptimizedBvh extends the b3QuantizedBvh to create AABB tree for triangle meshes, through the b3StridingMeshInterface.
-B3_ATTRIBUTE_ALIGNED16(class) b3OptimizedBvh : public b3QuantizedBvh
+B3_ATTRIBUTE_ALIGNED16(class)
+b3OptimizedBvh : public b3QuantizedBvh
 {
-	
 public:
 	B3_DECLARE_ALIGNED_ALLOCATOR();
 
 protected:
-
 public:
-
 	b3OptimizedBvh();
 
 	virtual ~b3OptimizedBvh();
 
-	void	build(b3StridingMeshInterface* triangles,bool useQuantizedAabbCompression, const b3Vector3& bvhAabbMin, const b3Vector3& bvhAabbMax);
+	void build(b3StridingMeshInterface * triangles, bool useQuantizedAabbCompression, const b3Vector3& bvhAabbMin, const b3Vector3& bvhAabbMax);
 
-	void	refit(b3StridingMeshInterface* triangles,const b3Vector3& aabbMin,const b3Vector3& aabbMax);
+	void refit(b3StridingMeshInterface * triangles, const b3Vector3& aabbMin, const b3Vector3& aabbMax);
 
-	void	refitPartial(b3StridingMeshInterface* triangles,const b3Vector3& aabbMin, const b3Vector3& aabbMax);
+	void refitPartial(b3StridingMeshInterface * triangles, const b3Vector3& aabbMin, const b3Vector3& aabbMax);
 
-	void	updateBvhNodes(b3StridingMeshInterface* meshInterface,int firstNode,int endNode,int index);
+	void updateBvhNodes(b3StridingMeshInterface * meshInterface, int firstNode, int endNode, int index);
 
 	/// Data buffer MUST be 16 byte aligned
-	virtual bool serializeInPlace(void *o_alignedDataBuffer, unsigned i_dataBufferSize, bool i_swapEndian) const
+	virtual bool serializeInPlace(void* o_alignedDataBuffer, unsigned i_dataBufferSize, bool i_swapEndian) const
 	{
-		return b3QuantizedBvh::serialize(o_alignedDataBuffer,i_dataBufferSize,i_swapEndian);
-
+		return b3QuantizedBvh::serialize(o_alignedDataBuffer, i_dataBufferSize, i_swapEndian);
 	}
 
 	///deSerializeInPlace loads and initializes a BVH from a buffer in memory 'in place'
-	static b3OptimizedBvh *deSerializeInPlace(void *i_alignedDataBuffer, unsigned int i_dataBufferSize, bool i_swapEndian);
-
-
+	static b3OptimizedBvh* deSerializeInPlace(void* i_alignedDataBuffer, unsigned int i_dataBufferSize, bool i_swapEndian);
 };
 
-
-#endif //B3_OPTIMIZED_BVH_H
-
-
+#endif  //B3_OPTIMIZED_BVH_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3QuantizedBvh.cpp b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3QuantizedBvh.cpp
index 52027e11184..9a448495f31 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3QuantizedBvh.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3QuantizedBvh.cpp
@@ -17,46 +17,40 @@ subject to the following restrictions:
 
 #include "Bullet3Geometry/b3AabbUtil.h"
 
-
 #define RAYAABB2
 
-b3QuantizedBvh::b3QuantizedBvh() : 
-					m_bulletVersion(B3_BULLET_VERSION),
-					m_useQuantization(false), 
-					m_traversalMode(TRAVERSAL_STACKLESS_CACHE_FRIENDLY)
-					//m_traversalMode(TRAVERSAL_STACKLESS)
-					//m_traversalMode(TRAVERSAL_RECURSIVE)
-					,m_subtreeHeaderCount(0) //PCK: add this line
+b3QuantizedBvh::b3QuantizedBvh() : m_bulletVersion(B3_BULLET_VERSION),
+								   m_useQuantization(false),
+								   m_traversalMode(TRAVERSAL_STACKLESS_CACHE_FRIENDLY)
+								   //m_traversalMode(TRAVERSAL_STACKLESS)
+								   //m_traversalMode(TRAVERSAL_RECURSIVE)
+								   ,
+								   m_subtreeHeaderCount(0)  //PCK: add this line
 {
-	m_bvhAabbMin.setValue(-B3_INFINITY,-B3_INFINITY,-B3_INFINITY);
-	m_bvhAabbMax.setValue(B3_INFINITY,B3_INFINITY,B3_INFINITY);
+	m_bvhAabbMin.setValue(-B3_INFINITY, -B3_INFINITY, -B3_INFINITY);
+	m_bvhAabbMax.setValue(B3_INFINITY, B3_INFINITY, B3_INFINITY);
 }
 
-
-
-
-
 void b3QuantizedBvh::buildInternal()
 {
 	///assumes that caller filled in the m_quantizedLeafNodes
 	m_useQuantization = true;
 	int numLeafNodes = 0;
-	
+
 	if (m_useQuantization)
 	{
 		//now we have an array of leafnodes in m_leafNodes
 		numLeafNodes = m_quantizedLeafNodes.size();
 
-		m_quantizedContiguousNodes.resize(2*numLeafNodes);
-
+		m_quantizedContiguousNodes.resize(2 * numLeafNodes);
 	}
 
 	m_curNodeIndex = 0;
 
-	buildTree(0,numLeafNodes);
+	buildTree(0, numLeafNodes);
 
 	///if the entire tree is small then subtree size, we need to create a header info for the tree
-	if(m_useQuantization && !m_SubtreeHeaders.size())
+	if (m_useQuantization && !m_SubtreeHeaders.size())
 	{
 		b3BvhSubtreeInfo& subtree = m_SubtreeHeaders.expand();
 		subtree.setAabbFromQuantizeNode(m_quantizedContiguousNodes[0]);
@@ -72,35 +66,27 @@ void b3QuantizedBvh::buildInternal()
 	m_leafNodes.clear();
 }
 
-
-
 ///just for debugging, to visualize the individual patches/subtrees
 #ifdef DEBUG_PATCH_COLORS
-b3Vector3 color[4]=
-{
-	b3Vector3(1,0,0),
-	b3Vector3(0,1,0),
-	b3Vector3(0,0,1),
-	b3Vector3(0,1,1)
-};
-#endif //DEBUG_PATCH_COLORS
+b3Vector3 color[4] =
+	{
+		b3Vector3(1, 0, 0),
+		b3Vector3(0, 1, 0),
+		b3Vector3(0, 0, 1),
+		b3Vector3(0, 1, 1)};
+#endif  //DEBUG_PATCH_COLORS
 
-
-
-void	b3QuantizedBvh::setQuantizationValues(const b3Vector3& bvhAabbMin,const b3Vector3& bvhAabbMax,b3Scalar quantizationMargin)
+void b3QuantizedBvh::setQuantizationValues(const b3Vector3& bvhAabbMin, const b3Vector3& bvhAabbMax, b3Scalar quantizationMargin)
 {
 	//enlarge the AABB to avoid division by zero when initializing the quantization values
-	b3Vector3 clampValue =b3MakeVector3(quantizationMargin,quantizationMargin,quantizationMargin);
+	b3Vector3 clampValue = b3MakeVector3(quantizationMargin, quantizationMargin, quantizationMargin);
 	m_bvhAabbMin = bvhAabbMin - clampValue;
 	m_bvhAabbMax = bvhAabbMax + clampValue;
 	b3Vector3 aabbSize = m_bvhAabbMax - m_bvhAabbMin;
-	m_bvhQuantization = b3MakeVector3(b3Scalar(65533.0),b3Scalar(65533.0),b3Scalar(65533.0)) / aabbSize;
+	m_bvhQuantization = b3MakeVector3(b3Scalar(65533.0), b3Scalar(65533.0), b3Scalar(65533.0)) / aabbSize;
 	m_useQuantization = true;
 }
 
-
-
-
 b3QuantizedBvh::~b3QuantizedBvh()
 {
 }
@@ -108,104 +94,100 @@ b3QuantizedBvh::~b3QuantizedBvh()
 #ifdef DEBUG_TREE_BUILDING
 int gStackDepth = 0;
 int gMaxStackDepth = 0;
-#endif //DEBUG_TREE_BUILDING
+#endif  //DEBUG_TREE_BUILDING
 
-void	b3QuantizedBvh::buildTree	(int startIndex,int endIndex)
+void b3QuantizedBvh::buildTree(int startIndex, int endIndex)
 {
 #ifdef DEBUG_TREE_BUILDING
 	gStackDepth++;
 	if (gStackDepth > gMaxStackDepth)
 		gMaxStackDepth = gStackDepth;
-#endif //DEBUG_TREE_BUILDING
-
+#endif  //DEBUG_TREE_BUILDING
 
 	int splitAxis, splitIndex, i;
-	int numIndices =endIndex-startIndex;
+	int numIndices = endIndex - startIndex;
 	int curIndex = m_curNodeIndex;
 
-	b3Assert(numIndices>0);
+	b3Assert(numIndices > 0);
 
-	if (numIndices==1)
+	if (numIndices == 1)
 	{
 #ifdef DEBUG_TREE_BUILDING
 		gStackDepth--;
-#endif //DEBUG_TREE_BUILDING
-		
-		assignInternalNodeFromLeafNode(m_curNodeIndex,startIndex);
+#endif  //DEBUG_TREE_BUILDING
+
+		assignInternalNodeFromLeafNode(m_curNodeIndex, startIndex);
 
 		m_curNodeIndex++;
-		return;	
+		return;
 	}
 	//calculate Best Splitting Axis and where to split it. Sort the incoming 'leafNodes' array within range 'startIndex/endIndex'.
-	
-	splitAxis = calcSplittingAxis(startIndex,endIndex);
 
-	splitIndex = sortAndCalcSplittingIndex(startIndex,endIndex,splitAxis);
+	splitAxis = calcSplittingAxis(startIndex, endIndex);
+
+	splitIndex = sortAndCalcSplittingIndex(startIndex, endIndex, splitAxis);
 
 	int internalNodeIndex = m_curNodeIndex;
-	
+
 	//set the min aabb to 'inf' or a max value, and set the max aabb to a -inf/minimum value.
 	//the aabb will be expanded during buildTree/mergeInternalNodeAabb with actual node values
-	setInternalNodeAabbMin(m_curNodeIndex,m_bvhAabbMax);//can't use b3Vector3(B3_INFINITY,B3_INFINITY,B3_INFINITY)) because of quantization
-	setInternalNodeAabbMax(m_curNodeIndex,m_bvhAabbMin);//can't use b3Vector3(-B3_INFINITY,-B3_INFINITY,-B3_INFINITY)) because of quantization
-	
-	
-	for (i=startIndex;i<endIndex;i++)
+	setInternalNodeAabbMin(m_curNodeIndex, m_bvhAabbMax);  //can't use b3Vector3(B3_INFINITY,B3_INFINITY,B3_INFINITY)) because of quantization
+	setInternalNodeAabbMax(m_curNodeIndex, m_bvhAabbMin);  //can't use b3Vector3(-B3_INFINITY,-B3_INFINITY,-B3_INFINITY)) because of quantization
+
+	for (i = startIndex; i < endIndex; i++)
 	{
-		mergeInternalNodeAabb(m_curNodeIndex,getAabbMin(i),getAabbMax(i));
+		mergeInternalNodeAabb(m_curNodeIndex, getAabbMin(i), getAabbMax(i));
 	}
 
 	m_curNodeIndex++;
-	
 
 	//internalNode->m_escapeIndex;
-	
+
 	int leftChildNodexIndex = m_curNodeIndex;
 
 	//build left child tree
-	buildTree(startIndex,splitIndex);
+	buildTree(startIndex, splitIndex);
 
 	int rightChildNodexIndex = m_curNodeIndex;
 	//build right child tree
-	buildTree(splitIndex,endIndex);
+	buildTree(splitIndex, endIndex);
 
 #ifdef DEBUG_TREE_BUILDING
 	gStackDepth--;
-#endif //DEBUG_TREE_BUILDING
+#endif  //DEBUG_TREE_BUILDING
 
 	int escapeIndex = m_curNodeIndex - curIndex;
 
 	if (m_useQuantization)
 	{
 		//escapeIndex is the number of nodes of this subtree
-		const int sizeQuantizedNode =sizeof(b3QuantizedBvhNode);
+		const int sizeQuantizedNode = sizeof(b3QuantizedBvhNode);
 		const int treeSizeInBytes = escapeIndex * sizeQuantizedNode;
 		if (treeSizeInBytes > MAX_SUBTREE_SIZE_IN_BYTES)
 		{
-			updateSubtreeHeaders(leftChildNodexIndex,rightChildNodexIndex);
+			updateSubtreeHeaders(leftChildNodexIndex, rightChildNodexIndex);
 		}
-	} else
+	}
+	else
 	{
-
 	}
 
-	setInternalNodeEscapeIndex(internalNodeIndex,escapeIndex);
-
+	setInternalNodeEscapeIndex(internalNodeIndex, escapeIndex);
 }
 
-void	b3QuantizedBvh::updateSubtreeHeaders(int leftChildNodexIndex,int rightChildNodexIndex)
+void b3QuantizedBvh::updateSubtreeHeaders(int leftChildNodexIndex, int rightChildNodexIndex)
 {
 	b3Assert(m_useQuantization);
 
 	b3QuantizedBvhNode& leftChildNode = m_quantizedContiguousNodes[leftChildNodexIndex];
 	int leftSubTreeSize = leftChildNode.isLeafNode() ? 1 : leftChildNode.getEscapeIndex();
-	int leftSubTreeSizeInBytes =  leftSubTreeSize * static_cast<int>(sizeof(b3QuantizedBvhNode));
-	
+	int leftSubTreeSizeInBytes = leftSubTreeSize * static_cast<int>(sizeof(b3QuantizedBvhNode));
+
 	b3QuantizedBvhNode& rightChildNode = m_quantizedContiguousNodes[rightChildNodexIndex];
 	int rightSubTreeSize = rightChildNode.isLeafNode() ? 1 : rightChildNode.getEscapeIndex();
-	int rightSubTreeSizeInBytes =  rightSubTreeSize *  static_cast<int>(sizeof(b3QuantizedBvhNode));
+	int rightSubTreeSizeInBytes = rightSubTreeSize * static_cast<int>(sizeof(b3QuantizedBvhNode));
 
-	if(leftSubTreeSizeInBytes <= MAX_SUBTREE_SIZE_IN_BYTES)
+	if (leftSubTreeSizeInBytes <= MAX_SUBTREE_SIZE_IN_BYTES)
 	{
 		b3BvhSubtreeInfo& subtree = m_SubtreeHeaders.expand();
 		subtree.setAabbFromQuantizeNode(leftChildNode);
@@ -213,7 +195,7 @@ void	b3QuantizedBvh::updateSubtreeHeaders(int leftChildNodexIndex,int rightChild
 		subtree.m_subtreeSize = leftSubTreeSize;
 	}
 
-	if(rightSubTreeSizeInBytes <= MAX_SUBTREE_SIZE_IN_BYTES)
+	if (rightSubTreeSizeInBytes <= MAX_SUBTREE_SIZE_IN_BYTES)
 	{
 		b3BvhSubtreeInfo& subtree = m_SubtreeHeaders.expand();
 		subtree.setAabbFromQuantizeNode(rightChildNode);
@@ -225,32 +207,31 @@ void	b3QuantizedBvh::updateSubtreeHeaders(int leftChildNodexIndex,int rightChild
 	m_subtreeHeaderCount = m_SubtreeHeaders.size();
 }
 
-
-int	b3QuantizedBvh::sortAndCalcSplittingIndex(int startIndex,int endIndex,int splitAxis)
+int b3QuantizedBvh::sortAndCalcSplittingIndex(int startIndex, int endIndex, int splitAxis)
 {
 	int i;
-	int splitIndex =startIndex;
+	int splitIndex = startIndex;
 	int numIndices = endIndex - startIndex;
 	b3Scalar splitValue;
 
-	b3Vector3 means=b3MakeVector3(b3Scalar(0.),b3Scalar(0.),b3Scalar(0.));
-	for (i=startIndex;i<endIndex;i++)
+	b3Vector3 means = b3MakeVector3(b3Scalar(0.), b3Scalar(0.), b3Scalar(0.));
+	for (i = startIndex; i < endIndex; i++)
 	{
-		b3Vector3 center = b3Scalar(0.5)*(getAabbMax(i)+getAabbMin(i));
-		means+=center;
+		b3Vector3 center = b3Scalar(0.5) * (getAabbMax(i) + getAabbMin(i));
+		means += center;
 	}
-	means *= (b3Scalar(1.)/(b3Scalar)numIndices);
-	
+	means *= (b3Scalar(1.) / (b3Scalar)numIndices);
+
 	splitValue = means[splitAxis];
-	
+
 	//sort leafNodes so all values larger then splitValue comes first, and smaller values start from 'splitIndex'.
-	for (i=startIndex;i<endIndex;i++)
+	for (i = startIndex; i < endIndex; i++)
 	{
-		b3Vector3 center = b3Scalar(0.5)*(getAabbMax(i)+getAabbMin(i));
+		b3Vector3 center = b3Scalar(0.5) * (getAabbMax(i) + getAabbMin(i));
 		if (center[splitAxis] > splitValue)
 		{
 			//swap
-			swapLeafNodes(i,splitIndex);
+			swapLeafNodes(i, splitIndex);
 			splitIndex++;
 		}
 	}
@@ -260,56 +241,53 @@ int	b3QuantizedBvh::sortAndCalcSplittingIndex(int startIndex,int endIndex,int sp
 	//unbalanced1 is unsafe: it can cause stack overflows
 	//bool unbalanced1 = ((splitIndex==startIndex) || (splitIndex == (endIndex-1)));
 
-	//unbalanced2 should work too: always use center (perfect balanced trees)	
+	//unbalanced2 should work too: always use center (perfect balanced trees)
 	//bool unbalanced2 = true;
 
 	//this should be safe too:
-	int rangeBalancedIndices = numIndices/3;
-	bool unbalanced = ((splitIndex<=(startIndex+rangeBalancedIndices)) || (splitIndex >=(endIndex-1-rangeBalancedIndices)));
-	
+	int rangeBalancedIndices = numIndices / 3;
+	bool unbalanced = ((splitIndex <= (startIndex + rangeBalancedIndices)) || (splitIndex >= (endIndex - 1 - rangeBalancedIndices)));
+
 	if (unbalanced)
 	{
-		splitIndex = startIndex+ (numIndices>>1);
+		splitIndex = startIndex + (numIndices >> 1);
 	}
 
-	bool unbal = (splitIndex==startIndex) || (splitIndex == (endIndex));
+	bool unbal = (splitIndex == startIndex) || (splitIndex == (endIndex));
 	(void)unbal;
 	b3Assert(!unbal);
 
 	return splitIndex;
 }
 
-
-int	b3QuantizedBvh::calcSplittingAxis(int startIndex,int endIndex)
+int b3QuantizedBvh::calcSplittingAxis(int startIndex, int endIndex)
 {
 	int i;
 
-	b3Vector3 means=b3MakeVector3(b3Scalar(0.),b3Scalar(0.),b3Scalar(0.));
-	b3Vector3 variance=b3MakeVector3(b3Scalar(0.),b3Scalar(0.),b3Scalar(0.));
-	int numIndices = endIndex-startIndex;
+	b3Vector3 means = b3MakeVector3(b3Scalar(0.), b3Scalar(0.), b3Scalar(0.));
+	b3Vector3 variance = b3MakeVector3(b3Scalar(0.), b3Scalar(0.), b3Scalar(0.));
+	int numIndices = endIndex - startIndex;
 
-	for (i=startIndex;i<endIndex;i++)
+	for (i = startIndex; i < endIndex; i++)
 	{
-		b3Vector3 center = b3Scalar(0.5)*(getAabbMax(i)+getAabbMin(i));
-		means+=center;
+		b3Vector3 center = b3Scalar(0.5) * (getAabbMax(i) + getAabbMin(i));
+		means += center;
 	}
-	means *= (b3Scalar(1.)/(b3Scalar)numIndices);
-		
-	for (i=startIndex;i<endIndex;i++)
+	means *= (b3Scalar(1.) / (b3Scalar)numIndices);
+
+	for (i = startIndex; i < endIndex; i++)
 	{
-		b3Vector3 center = b3Scalar(0.5)*(getAabbMax(i)+getAabbMin(i));
-		b3Vector3 diff2 = center-means;
+		b3Vector3 center = b3Scalar(0.5) * (getAabbMax(i) + getAabbMin(i));
+		b3Vector3 diff2 = center - means;
 		diff2 = diff2 * diff2;
 		variance += diff2;
 	}
-	variance *= (b3Scalar(1.)/	((b3Scalar)numIndices-1)	);
-	
+	variance *= (b3Scalar(1.) / ((b3Scalar)numIndices - 1));
+
 	return variance.maxAxis();
 }
 
-
-
-void	b3QuantizedBvh::reportAabbOverlappingNodex(b3NodeOverlapCallback* nodeCallback,const b3Vector3& aabbMin,const b3Vector3& aabbMax) const
+void b3QuantizedBvh::reportAabbOverlappingNodex(b3NodeOverlapCallback* nodeCallback, const b3Vector3& aabbMin, const b3Vector3& aabbMax) const
 {
 	//either choose recursive traversal (walkTree) or stackless (walkStacklessTree)
 
@@ -318,38 +296,37 @@ void	b3QuantizedBvh::reportAabbOverlappingNodex(b3NodeOverlapCallback* nodeCallb
 		///quantize query AABB
 		unsigned short int quantizedQueryAabbMin[3];
 		unsigned short int quantizedQueryAabbMax[3];
-		quantizeWithClamp(quantizedQueryAabbMin,aabbMin,0);
-		quantizeWithClamp(quantizedQueryAabbMax,aabbMax,1);
+		quantizeWithClamp(quantizedQueryAabbMin, aabbMin, 0);
+		quantizeWithClamp(quantizedQueryAabbMax, aabbMax, 1);
 
 		switch (m_traversalMode)
 		{
-		case TRAVERSAL_STACKLESS:
-				walkStacklessQuantizedTree(nodeCallback,quantizedQueryAabbMin,quantizedQueryAabbMax,0,m_curNodeIndex);
-			break;
-		case TRAVERSAL_STACKLESS_CACHE_FRIENDLY:
-				walkStacklessQuantizedTreeCacheFriendly(nodeCallback,quantizedQueryAabbMin,quantizedQueryAabbMax);
-			break;
-		case TRAVERSAL_RECURSIVE:
+			case TRAVERSAL_STACKLESS:
+				walkStacklessQuantizedTree(nodeCallback, quantizedQueryAabbMin, quantizedQueryAabbMax, 0, m_curNodeIndex);
+				break;
+			case TRAVERSAL_STACKLESS_CACHE_FRIENDLY:
+				walkStacklessQuantizedTreeCacheFriendly(nodeCallback, quantizedQueryAabbMin, quantizedQueryAabbMax);
+				break;
+			case TRAVERSAL_RECURSIVE:
 			{
 				const b3QuantizedBvhNode* rootNode = &m_quantizedContiguousNodes[0];
-				walkRecursiveQuantizedTreeAgainstQueryAabb(rootNode,nodeCallback,quantizedQueryAabbMin,quantizedQueryAabbMax);
+				walkRecursiveQuantizedTreeAgainstQueryAabb(rootNode, nodeCallback, quantizedQueryAabbMin, quantizedQueryAabbMax);
 			}
 			break;
-		default:
-			//unsupported
-			b3Assert(0);
+			default:
+				//unsupported
+				b3Assert(0);
 		}
-	} else
+	}
+	else
 	{
-		walkStacklessTree(nodeCallback,aabbMin,aabbMax);
+		walkStacklessTree(nodeCallback, aabbMin, aabbMax);
 	}
 }
 
-
 static int b3s_maxIterations = 0;
 
-
-void	b3QuantizedBvh::walkStacklessTree(b3NodeOverlapCallback* nodeCallback,const b3Vector3& aabbMin,const b3Vector3& aabbMax) const
+void b3QuantizedBvh::walkStacklessTree(b3NodeOverlapCallback* nodeCallback, const b3Vector3& aabbMin, const b3Vector3& aabbMax) const
 {
 	b3Assert(!m_useQuantization);
 
@@ -363,24 +340,25 @@ void	b3QuantizedBvh::walkStacklessTree(b3NodeOverlapCallback* nodeCallback,const
 	while (curIndex < m_curNodeIndex)
 	{
 		//catch bugs in tree data
-		b3Assert (walkIterations < m_curNodeIndex);
+		b3Assert(walkIterations < m_curNodeIndex);
 
 		walkIterations++;
-		aabbOverlap = b3TestAabbAgainstAabb2(aabbMin,aabbMax,rootNode->m_aabbMinOrg,rootNode->m_aabbMaxOrg);
+		aabbOverlap = b3TestAabbAgainstAabb2(aabbMin, aabbMax, rootNode->m_aabbMinOrg, rootNode->m_aabbMaxOrg);
 		isLeafNode = rootNode->m_escapeIndex == -1;
-		
+
 		//PCK: unsigned instead of bool
 		if (isLeafNode && (aabbOverlap != 0))
 		{
-			nodeCallback->processNode(rootNode->m_subPart,rootNode->m_triangleIndex);
-		} 
-		
+			nodeCallback->processNode(rootNode->m_subPart, rootNode->m_triangleIndex);
+		}
+
 		//PCK: unsigned instead of bool
 		if ((aabbOverlap != 0) || isLeafNode)
 		{
 			rootNode++;
 			curIndex++;
-		} else
+		}
+		else
 		{
 			escapeIndex = rootNode->m_escapeIndex;
 			rootNode += escapeIndex;
@@ -389,7 +367,6 @@ void	b3QuantizedBvh::walkStacklessTree(b3NodeOverlapCallback* nodeCallback,const
 	}
 	if (b3s_maxIterations < walkIterations)
 		b3s_maxIterations = walkIterations;
-
 }
 
 /*
@@ -413,39 +390,38 @@ void	b3QuantizedBvh::walkTree(b3OptimizedBvhNode* rootNode,b3NodeOverlapCallback
 }
 */
 
-void b3QuantizedBvh::walkRecursiveQuantizedTreeAgainstQueryAabb(const b3QuantizedBvhNode* currentNode,b3NodeOverlapCallback* nodeCallback,unsigned short int* quantizedQueryAabbMin,unsigned short int* quantizedQueryAabbMax) const
+void b3QuantizedBvh::walkRecursiveQuantizedTreeAgainstQueryAabb(const b3QuantizedBvhNode* currentNode, b3NodeOverlapCallback* nodeCallback, unsigned short int* quantizedQueryAabbMin, unsigned short int* quantizedQueryAabbMax) const
 {
 	b3Assert(m_useQuantization);
-	
+
 	bool isLeafNode;
 	//PCK: unsigned instead of bool
 	unsigned aabbOverlap;
 
 	//PCK: unsigned instead of bool
-	aabbOverlap = b3TestQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin,quantizedQueryAabbMax,currentNode->m_quantizedAabbMin,currentNode->m_quantizedAabbMax);
+	aabbOverlap = b3TestQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin, quantizedQueryAabbMax, currentNode->m_quantizedAabbMin, currentNode->m_quantizedAabbMax);
 	isLeafNode = currentNode->isLeafNode();
-		
+
 	//PCK: unsigned instead of bool
 	if (aabbOverlap != 0)
 	{
 		if (isLeafNode)
 		{
-			nodeCallback->processNode(currentNode->getPartId(),currentNode->getTriangleIndex());
-		} else
+			nodeCallback->processNode(currentNode->getPartId(), currentNode->getTriangleIndex());
+		}
+		else
 		{
 			//process left and right children
-			const b3QuantizedBvhNode* leftChildNode = currentNode+1;
-			walkRecursiveQuantizedTreeAgainstQueryAabb(leftChildNode,nodeCallback,quantizedQueryAabbMin,quantizedQueryAabbMax);
+			const b3QuantizedBvhNode* leftChildNode = currentNode + 1;
+			walkRecursiveQuantizedTreeAgainstQueryAabb(leftChildNode, nodeCallback, quantizedQueryAabbMin, quantizedQueryAabbMax);
 
-			const b3QuantizedBvhNode* rightChildNode = leftChildNode->isLeafNode() ? leftChildNode+1:leftChildNode+leftChildNode->getEscapeIndex();
-			walkRecursiveQuantizedTreeAgainstQueryAabb(rightChildNode,nodeCallback,quantizedQueryAabbMin,quantizedQueryAabbMax);
+			const b3QuantizedBvhNode* rightChildNode = leftChildNode->isLeafNode() ? leftChildNode + 1 : leftChildNode + leftChildNode->getEscapeIndex();
+			walkRecursiveQuantizedTreeAgainstQueryAabb(rightChildNode, nodeCallback, quantizedQueryAabbMin, quantizedQueryAabbMax);
 		}
-	}		
+	}
 }
 
-
-
-void	b3QuantizedBvh::walkStacklessTreeAgainstRay(b3NodeOverlapCallback* nodeCallback, const b3Vector3& raySource, const b3Vector3& rayTarget, const b3Vector3& aabbMin, const b3Vector3& aabbMax, int startNodeIndex,int endNodeIndex) const
+void b3QuantizedBvh::walkStacklessTreeAgainstRay(b3NodeOverlapCallback* nodeCallback, const b3Vector3& raySource, const b3Vector3& rayTarget, const b3Vector3& aabbMin, const b3Vector3& aabbMax, int startNodeIndex, int endNodeIndex) const
 {
 	b3Assert(!m_useQuantization);
 
@@ -454,11 +430,11 @@ void	b3QuantizedBvh::walkStacklessTreeAgainstRay(b3NodeOverlapCallback* nodeCall
 	int walkIterations = 0;
 	bool isLeafNode;
 	//PCK: unsigned instead of bool
-	unsigned aabbOverlap=0;
-	unsigned rayBoxOverlap=0;
+	unsigned aabbOverlap = 0;
+	unsigned rayBoxOverlap = 0;
 	b3Scalar lambda_max = 1.0;
-	
-		/* Quick pruning by quantized box */
+
+	/* Quick pruning by quantized box */
 	b3Vector3 rayAabbMin = raySource;
 	b3Vector3 rayAabbMax = raySource;
 	rayAabbMin.setMin(rayTarget);
@@ -469,15 +445,15 @@ void	b3QuantizedBvh::walkStacklessTreeAgainstRay(b3NodeOverlapCallback* nodeCall
 	rayAabbMax += aabbMax;
 
 #ifdef RAYAABB2
-	b3Vector3 rayDir = (rayTarget-raySource);
-	rayDir.normalize ();
-	lambda_max = rayDir.dot(rayTarget-raySource);
+	b3Vector3 rayDir = (rayTarget - raySource);
+	rayDir.normalize();
+	lambda_max = rayDir.dot(rayTarget - raySource);
 	///what about division by zero? --> just set rayDirection[i] to 1.0
 	b3Vector3 rayDirectionInverse;
 	rayDirectionInverse[0] = rayDir[0] == b3Scalar(0.0) ? b3Scalar(B3_LARGE_FLOAT) : b3Scalar(1.0) / rayDir[0];
 	rayDirectionInverse[1] = rayDir[1] == b3Scalar(0.0) ? b3Scalar(B3_LARGE_FLOAT) : b3Scalar(1.0) / rayDir[1];
 	rayDirectionInverse[2] = rayDir[2] == b3Scalar(0.0) ? b3Scalar(B3_LARGE_FLOAT) : b3Scalar(1.0) / rayDir[2];
-	unsigned int sign[3] = { rayDirectionInverse[0] < 0.0, rayDirectionInverse[1] < 0.0, rayDirectionInverse[2] < 0.0};
+	unsigned int sign[3] = {rayDirectionInverse[0] < 0.0, rayDirectionInverse[1] < 0.0, rayDirectionInverse[2] < 0.0};
 #endif
 
 	b3Vector3 bounds[2];
@@ -486,7 +462,7 @@ void	b3QuantizedBvh::walkStacklessTreeAgainstRay(b3NodeOverlapCallback* nodeCall
 	{
 		b3Scalar param = 1.0;
 		//catch bugs in tree data
-		b3Assert (walkIterations < m_curNodeIndex);
+		b3Assert(walkIterations < m_curNodeIndex);
 
 		walkIterations++;
 
@@ -496,34 +472,35 @@ void	b3QuantizedBvh::walkStacklessTreeAgainstRay(b3NodeOverlapCallback* nodeCall
 		bounds[0] -= aabbMax;
 		bounds[1] -= aabbMin;
 
-		aabbOverlap = b3TestAabbAgainstAabb2(rayAabbMin,rayAabbMax,rootNode->m_aabbMinOrg,rootNode->m_aabbMaxOrg);
+		aabbOverlap = b3TestAabbAgainstAabb2(rayAabbMin, rayAabbMax, rootNode->m_aabbMinOrg, rootNode->m_aabbMaxOrg);
 		//perhaps profile if it is worth doing the aabbOverlap test first
 
 #ifdef RAYAABB2
-			///careful with this check: need to check division by zero (above) and fix the unQuantize method
-			///thanks Joerg/hiker for the reproduction case!
-			///http://www.bulletphysics.com/Bullet/phpBB3/viewtopic.php?f=9&t=1858
-		rayBoxOverlap = aabbOverlap ? b3RayAabb2 (raySource, rayDirectionInverse, sign, bounds, param, 0.0f, lambda_max) : false;
+		///careful with this check: need to check division by zero (above) and fix the unQuantize method
+		///thanks Joerg/hiker for the reproduction case!
+		///http://www.bulletphysics.com/Bullet/phpBB3/viewtopic.php?f=9&t=1858
+		rayBoxOverlap = aabbOverlap ? b3RayAabb2(raySource, rayDirectionInverse, sign, bounds, param, 0.0f, lambda_max) : false;
 
 #else
 		b3Vector3 normal;
-		rayBoxOverlap = b3RayAabb(raySource, rayTarget,bounds[0],bounds[1],param, normal);
+		rayBoxOverlap = b3RayAabb(raySource, rayTarget, bounds[0], bounds[1], param, normal);
 #endif
 
 		isLeafNode = rootNode->m_escapeIndex == -1;
-		
+
 		//PCK: unsigned instead of bool
 		if (isLeafNode && (rayBoxOverlap != 0))
 		{
-			nodeCallback->processNode(rootNode->m_subPart,rootNode->m_triangleIndex);
-		} 
-		
+			nodeCallback->processNode(rootNode->m_subPart, rootNode->m_triangleIndex);
+		}
+
 		//PCK: unsigned instead of bool
 		if ((rayBoxOverlap != 0) || isLeafNode)
 		{
 			rootNode++;
 			curIndex++;
-		} else
+		}
+		else
 		{
 			escapeIndex = rootNode->m_escapeIndex;
 			rootNode += escapeIndex;
@@ -532,15 +509,12 @@ void	b3QuantizedBvh::walkStacklessTreeAgainstRay(b3NodeOverlapCallback* nodeCall
 	}
 	if (b3s_maxIterations < walkIterations)
 		b3s_maxIterations = walkIterations;
-
 }
 
-
-
-void	b3QuantizedBvh::walkStacklessQuantizedTreeAgainstRay(b3NodeOverlapCallback* nodeCallback, const b3Vector3& raySource, const b3Vector3& rayTarget, const b3Vector3& aabbMin, const b3Vector3& aabbMax, int startNodeIndex,int endNodeIndex) const
+void b3QuantizedBvh::walkStacklessQuantizedTreeAgainstRay(b3NodeOverlapCallback* nodeCallback, const b3Vector3& raySource, const b3Vector3& rayTarget, const b3Vector3& aabbMin, const b3Vector3& aabbMax, int startNodeIndex, int endNodeIndex) const
 {
 	b3Assert(m_useQuantization);
-	
+
 	int curIndex = startNodeIndex;
 	int walkIterations = 0;
 	int subTreeSize = endNodeIndex - startNodeIndex;
@@ -548,7 +522,7 @@ void	b3QuantizedBvh::walkStacklessQuantizedTreeAgainstRay(b3NodeOverlapCallback*
 
 	const b3QuantizedBvhNode* rootNode = &m_quantizedContiguousNodes[startNodeIndex];
 	int escapeIndex;
-	
+
 	bool isLeafNode;
 	//PCK: unsigned instead of bool
 	unsigned boxBoxOverlap = 0;
@@ -557,14 +531,14 @@ void	b3QuantizedBvh::walkStacklessQuantizedTreeAgainstRay(b3NodeOverlapCallback*
 	b3Scalar lambda_max = 1.0;
 
 #ifdef RAYAABB2
-	b3Vector3 rayDirection = (rayTarget-raySource);
-	rayDirection.normalize ();
-	lambda_max = rayDirection.dot(rayTarget-raySource);
+	b3Vector3 rayDirection = (rayTarget - raySource);
+	rayDirection.normalize();
+	lambda_max = rayDirection.dot(rayTarget - raySource);
 	///what about division by zero? --> just set rayDirection[i] to 1.0
 	rayDirection[0] = rayDirection[0] == b3Scalar(0.0) ? b3Scalar(B3_LARGE_FLOAT) : b3Scalar(1.0) / rayDirection[0];
 	rayDirection[1] = rayDirection[1] == b3Scalar(0.0) ? b3Scalar(B3_LARGE_FLOAT) : b3Scalar(1.0) / rayDirection[1];
 	rayDirection[2] = rayDirection[2] == b3Scalar(0.0) ? b3Scalar(B3_LARGE_FLOAT) : b3Scalar(1.0) / rayDirection[2];
-	unsigned int sign[3] = { rayDirection[0] < 0.0, rayDirection[1] < 0.0, rayDirection[2] < 0.0};
+	unsigned int sign[3] = {rayDirection[0] < 0.0, rayDirection[1] < 0.0, rayDirection[2] < 0.0};
 #endif
 
 	/* Quick pruning by quantized box */
@@ -579,37 +553,36 @@ void	b3QuantizedBvh::walkStacklessQuantizedTreeAgainstRay(b3NodeOverlapCallback*
 
 	unsigned short int quantizedQueryAabbMin[3];
 	unsigned short int quantizedQueryAabbMax[3];
-	quantizeWithClamp(quantizedQueryAabbMin,rayAabbMin,0);
-	quantizeWithClamp(quantizedQueryAabbMax,rayAabbMax,1);
+	quantizeWithClamp(quantizedQueryAabbMin, rayAabbMin, 0);
+	quantizeWithClamp(quantizedQueryAabbMax, rayAabbMax, 1);
 
 	while (curIndex < endNodeIndex)
 	{
-
 //#define VISUALLY_ANALYZE_BVH 1
 #ifdef VISUALLY_ANALYZE_BVH
 		//some code snippet to debugDraw aabb, to visually analyze bvh structure
 		static int drawPatch = 0;
 		//need some global access to a debugDrawer
 		extern b3IDebugDraw* debugDrawerPtr;
-		if (curIndex==drawPatch)
+		if (curIndex == drawPatch)
 		{
-			b3Vector3 aabbMin,aabbMax;
+			b3Vector3 aabbMin, aabbMax;
 			aabbMin = unQuantize(rootNode->m_quantizedAabbMin);
 			aabbMax = unQuantize(rootNode->m_quantizedAabbMax);
-			b3Vector3	color(1,0,0);
-			debugDrawerPtr->drawAabb(aabbMin,aabbMax,color);
+			b3Vector3 color(1, 0, 0);
+			debugDrawerPtr->drawAabb(aabbMin, aabbMax, color);
 		}
-#endif//VISUALLY_ANALYZE_BVH
+#endif  //VISUALLY_ANALYZE_BVH
 
 		//catch bugs in tree data
-		b3Assert (walkIterations < subTreeSize);
+		b3Assert(walkIterations < subTreeSize);
 
 		walkIterations++;
 		//PCK: unsigned instead of bool
 		// only interested if this is closer than any previous hit
 		b3Scalar param = 1.0;
 		rayBoxOverlap = 0;
-		boxBoxOverlap = b3TestQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin,quantizedQueryAabbMax,rootNode->m_quantizedAabbMin,rootNode->m_quantizedAabbMax);
+		boxBoxOverlap = b3TestQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin, quantizedQueryAabbMax, rootNode->m_quantizedAabbMin, rootNode->m_quantizedAabbMax);
 		isLeafNode = rootNode->isLeafNode();
 		if (boxBoxOverlap)
 		{
@@ -634,24 +607,25 @@ void	b3QuantizedBvh::walkStacklessQuantizedTreeAgainstRay(b3NodeOverlapCallback*
 			///http://www.bulletphysics.com/Bullet/phpBB3/viewtopic.php?f=9&t=1858
 
 			//B3_PROFILE("b3RayAabb2");
-			rayBoxOverlap = b3RayAabb2 (raySource, rayDirection, sign, bounds, param, 0.0f, lambda_max);
-			
+			rayBoxOverlap = b3RayAabb2(raySource, rayDirection, sign, bounds, param, 0.0f, lambda_max);
+
 #else
-			rayBoxOverlap = true;//b3RayAabb(raySource, rayTarget, bounds[0], bounds[1], param, normal);
+			rayBoxOverlap = true;  //b3RayAabb(raySource, rayTarget, bounds[0], bounds[1], param, normal);
 #endif
 		}
-		
+
 		if (isLeafNode && rayBoxOverlap)
 		{
-			nodeCallback->processNode(rootNode->getPartId(),rootNode->getTriangleIndex());
+			nodeCallback->processNode(rootNode->getPartId(), rootNode->getTriangleIndex());
 		}
-		
+
 		//PCK: unsigned instead of bool
 		if ((rayBoxOverlap != 0) || isLeafNode)
 		{
 			rootNode++;
 			curIndex++;
-		} else
+		}
+		else
 		{
 			escapeIndex = rootNode->getEscapeIndex();
 			rootNode += escapeIndex;
@@ -660,13 +634,12 @@ void	b3QuantizedBvh::walkStacklessQuantizedTreeAgainstRay(b3NodeOverlapCallback*
 	}
 	if (b3s_maxIterations < walkIterations)
 		b3s_maxIterations = walkIterations;
-
 }
 
-void	b3QuantizedBvh::walkStacklessQuantizedTree(b3NodeOverlapCallback* nodeCallback,unsigned short int* quantizedQueryAabbMin,unsigned short int* quantizedQueryAabbMax,int startNodeIndex,int endNodeIndex) const
+void b3QuantizedBvh::walkStacklessQuantizedTree(b3NodeOverlapCallback* nodeCallback, unsigned short int* quantizedQueryAabbMin, unsigned short int* quantizedQueryAabbMax, int startNodeIndex, int endNodeIndex) const
 {
 	b3Assert(m_useQuantization);
-	
+
 	int curIndex = startNodeIndex;
 	int walkIterations = 0;
 	int subTreeSize = endNodeIndex - startNodeIndex;
@@ -674,49 +647,49 @@ void	b3QuantizedBvh::walkStacklessQuantizedTree(b3NodeOverlapCallback* nodeCallb
 
 	const b3QuantizedBvhNode* rootNode = &m_quantizedContiguousNodes[startNodeIndex];
 	int escapeIndex;
-	
+
 	bool isLeafNode;
 	//PCK: unsigned instead of bool
 	unsigned aabbOverlap;
 
 	while (curIndex < endNodeIndex)
 	{
-
 //#define VISUALLY_ANALYZE_BVH 1
 #ifdef VISUALLY_ANALYZE_BVH
 		//some code snippet to debugDraw aabb, to visually analyze bvh structure
 		static int drawPatch = 0;
 		//need some global access to a debugDrawer
 		extern b3IDebugDraw* debugDrawerPtr;
-		if (curIndex==drawPatch)
+		if (curIndex == drawPatch)
 		{
-			b3Vector3 aabbMin,aabbMax;
+			b3Vector3 aabbMin, aabbMax;
 			aabbMin = unQuantize(rootNode->m_quantizedAabbMin);
 			aabbMax = unQuantize(rootNode->m_quantizedAabbMax);
-			b3Vector3	color(1,0,0);
-			debugDrawerPtr->drawAabb(aabbMin,aabbMax,color);
+			b3Vector3 color(1, 0, 0);
+			debugDrawerPtr->drawAabb(aabbMin, aabbMax, color);
 		}
-#endif//VISUALLY_ANALYZE_BVH
+#endif  //VISUALLY_ANALYZE_BVH
 
 		//catch bugs in tree data
-		b3Assert (walkIterations < subTreeSize);
+		b3Assert(walkIterations < subTreeSize);
 
 		walkIterations++;
 		//PCK: unsigned instead of bool
-		aabbOverlap = b3TestQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin,quantizedQueryAabbMax,rootNode->m_quantizedAabbMin,rootNode->m_quantizedAabbMax);
+		aabbOverlap = b3TestQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin, quantizedQueryAabbMax, rootNode->m_quantizedAabbMin, rootNode->m_quantizedAabbMax);
 		isLeafNode = rootNode->isLeafNode();
-		
+
 		if (isLeafNode && aabbOverlap)
 		{
-			nodeCallback->processNode(rootNode->getPartId(),rootNode->getTriangleIndex());
-		} 
-		
+			nodeCallback->processNode(rootNode->getPartId(), rootNode->getTriangleIndex());
+		}
+
 		//PCK: unsigned instead of bool
 		if ((aabbOverlap != 0) || isLeafNode)
 		{
 			rootNode++;
 			curIndex++;
-		} else
+		}
+		else
 		{
 			escapeIndex = rootNode->getEscapeIndex();
 			rootNode += escapeIndex;
@@ -725,40 +698,36 @@ void	b3QuantizedBvh::walkStacklessQuantizedTree(b3NodeOverlapCallback* nodeCallb
 	}
 	if (b3s_maxIterations < walkIterations)
 		b3s_maxIterations = walkIterations;
-
 }
 
 //This traversal can be called from Playstation 3 SPU
-void	b3QuantizedBvh::walkStacklessQuantizedTreeCacheFriendly(b3NodeOverlapCallback* nodeCallback,unsigned short int* quantizedQueryAabbMin,unsigned short int* quantizedQueryAabbMax) const
+void b3QuantizedBvh::walkStacklessQuantizedTreeCacheFriendly(b3NodeOverlapCallback* nodeCallback, unsigned short int* quantizedQueryAabbMin, unsigned short int* quantizedQueryAabbMax) const
 {
 	b3Assert(m_useQuantization);
 
 	int i;
 
-
-	for (i=0;i<this->m_SubtreeHeaders.size();i++)
+	for (i = 0; i < this->m_SubtreeHeaders.size(); i++)
 	{
 		const b3BvhSubtreeInfo& subtree = m_SubtreeHeaders[i];
 
 		//PCK: unsigned instead of bool
-		unsigned overlap = b3TestQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin,quantizedQueryAabbMax,subtree.m_quantizedAabbMin,subtree.m_quantizedAabbMax);
+		unsigned overlap = b3TestQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin, quantizedQueryAabbMax, subtree.m_quantizedAabbMin, subtree.m_quantizedAabbMax);
 		if (overlap != 0)
 		{
-			walkStacklessQuantizedTree(nodeCallback,quantizedQueryAabbMin,quantizedQueryAabbMax,
-				subtree.m_rootNodeIndex,
-				subtree.m_rootNodeIndex+subtree.m_subtreeSize);
+			walkStacklessQuantizedTree(nodeCallback, quantizedQueryAabbMin, quantizedQueryAabbMax,
+									   subtree.m_rootNodeIndex,
+									   subtree.m_rootNodeIndex + subtree.m_subtreeSize);
 		}
 	}
 }
 
-
-void	b3QuantizedBvh::reportRayOverlappingNodex (b3NodeOverlapCallback* nodeCallback, const b3Vector3& raySource, const b3Vector3& rayTarget) const
+void b3QuantizedBvh::reportRayOverlappingNodex(b3NodeOverlapCallback* nodeCallback, const b3Vector3& raySource, const b3Vector3& rayTarget) const
 {
-	reportBoxCastOverlappingNodex(nodeCallback,raySource,rayTarget,b3MakeVector3(0,0,0),b3MakeVector3(0,0,0));
+	reportBoxCastOverlappingNodex(nodeCallback, raySource, rayTarget, b3MakeVector3(0, 0, 0), b3MakeVector3(0, 0, 0));
 }
 
-
-void	b3QuantizedBvh::reportBoxCastOverlappingNodex(b3NodeOverlapCallback* nodeCallback, const b3Vector3& raySource, const b3Vector3& rayTarget, const b3Vector3& aabbMin,const b3Vector3& aabbMax) const
+void b3QuantizedBvh::reportBoxCastOverlappingNodex(b3NodeOverlapCallback* nodeCallback, const b3Vector3& raySource, const b3Vector3& rayTarget, const b3Vector3& aabbMin, const b3Vector3& aabbMax) const
 {
 	//always use stackless
 
@@ -782,31 +751,31 @@ void	b3QuantizedBvh::reportBoxCastOverlappingNodex(b3NodeOverlapCallback* nodeCa
 		reportAabbOverlappingNodex(nodeCallback,qaabbMin,qaabbMax);
 	}
 	*/
-
 }
 
-
-void	b3QuantizedBvh::swapLeafNodes(int i,int splitIndex)
+void b3QuantizedBvh::swapLeafNodes(int i, int splitIndex)
 {
 	if (m_useQuantization)
 	{
-			b3QuantizedBvhNode tmp = m_quantizedLeafNodes[i];
-			m_quantizedLeafNodes[i] = m_quantizedLeafNodes[splitIndex];
-			m_quantizedLeafNodes[splitIndex] = tmp;
-	} else
+		b3QuantizedBvhNode tmp = m_quantizedLeafNodes[i];
+		m_quantizedLeafNodes[i] = m_quantizedLeafNodes[splitIndex];
+		m_quantizedLeafNodes[splitIndex] = tmp;
+	}
+	else
 	{
-			b3OptimizedBvhNode tmp = m_leafNodes[i];
-			m_leafNodes[i] = m_leafNodes[splitIndex];
-			m_leafNodes[splitIndex] = tmp;
+		b3OptimizedBvhNode tmp = m_leafNodes[i];
+		m_leafNodes[i] = m_leafNodes[splitIndex];
+		m_leafNodes[splitIndex] = tmp;
 	}
 }
 
-void	b3QuantizedBvh::assignInternalNodeFromLeafNode(int internalNode,int leafNodeIndex)
+void b3QuantizedBvh::assignInternalNodeFromLeafNode(int internalNode, int leafNodeIndex)
 {
 	if (m_useQuantization)
 	{
 		m_quantizedContiguousNodes[internalNode] = m_quantizedLeafNodes[leafNodeIndex];
-	} else
+	}
+	else
 	{
 		m_contiguousNodes[internalNode] = m_leafNodes[leafNodeIndex];
 	}
@@ -823,11 +792,10 @@ static const unsigned BVH_ALIGNMENT_MASK = BVH_ALIGNMENT-1;
 static const unsigned BVH_ALIGNMENT_BLOCKS = 2;
 #endif
 
-
 unsigned int b3QuantizedBvh::getAlignmentSerializationPadding()
 {
 	// I changed this to 0 since the extra padding is not needed or used.
-	return 0;//BVH_ALIGNMENT_BLOCKS * BVH_ALIGNMENT;
+	return 0;  //BVH_ALIGNMENT_BLOCKS * BVH_ALIGNMENT;
 }
 
 unsigned b3QuantizedBvh::calculateSerializeBufferSize() const
@@ -841,12 +809,12 @@ unsigned b3QuantizedBvh::calculateSerializeBufferSize() const
 	return baseSize + m_curNodeIndex * sizeof(b3OptimizedBvhNode);
 }
 
-bool b3QuantizedBvh::serialize(void *o_alignedDataBuffer, unsigned /*i_dataBufferSize */, bool i_swapEndian) const
+bool b3QuantizedBvh::serialize(void* o_alignedDataBuffer, unsigned /*i_dataBufferSize */, bool i_swapEndian) const
 {
 	b3Assert(m_subtreeHeaderCount == m_SubtreeHeaders.size());
 	m_subtreeHeaderCount = m_SubtreeHeaders.size();
 
-/*	if (i_dataBufferSize < calculateSerializeBufferSize() || o_alignedDataBuffer == NULL || (((unsigned)o_alignedDataBuffer & BVH_ALIGNMENT_MASK) != 0))
+	/*	if (i_dataBufferSize < calculateSerializeBufferSize() || o_alignedDataBuffer == NULL || (((unsigned)o_alignedDataBuffer & BVH_ALIGNMENT_MASK) != 0))
 	{
 		///check alignedment for buffer?
 		b3Assert(0);
@@ -854,7 +822,7 @@ bool b3QuantizedBvh::serialize(void *o_alignedDataBuffer, unsigned /*i_dataBuffe
 	}
 */
 
-	b3QuantizedBvh *targetBvh = (b3QuantizedBvh *)o_alignedDataBuffer;
+	b3QuantizedBvh* targetBvh = (b3QuantizedBvh*)o_alignedDataBuffer;
 
 	// construct the class so the virtual function table, etc will be set up
 	// Also, m_leafNodes and m_quantizedLeafNodes will be initialized to default values by the constructor
@@ -864,10 +832,9 @@ bool b3QuantizedBvh::serialize(void *o_alignedDataBuffer, unsigned /*i_dataBuffe
 	{
 		targetBvh->m_curNodeIndex = static_cast<int>(b3SwapEndian(m_curNodeIndex));
 
-
-		b3SwapVector3Endian(m_bvhAabbMin,targetBvh->m_bvhAabbMin);
-		b3SwapVector3Endian(m_bvhAabbMax,targetBvh->m_bvhAabbMax);
-		b3SwapVector3Endian(m_bvhQuantization,targetBvh->m_bvhQuantization);
+		b3SwapVector3Endian(m_bvhAabbMin, targetBvh->m_bvhAabbMin);
+		b3SwapVector3Endian(m_bvhAabbMax, targetBvh->m_bvhAabbMax);
+		b3SwapVector3Endian(m_bvhQuantization, targetBvh->m_bvhQuantization);
 
 		targetBvh->m_traversalMode = (b3TraversalMode)b3SwapEndian(m_traversalMode);
 		targetBvh->m_subtreeHeaderCount = static_cast<int>(b3SwapEndian(m_subtreeHeaderCount));
@@ -884,12 +851,12 @@ bool b3QuantizedBvh::serialize(void *o_alignedDataBuffer, unsigned /*i_dataBuffe
 
 	targetBvh->m_useQuantization = m_useQuantization;
 
-	unsigned char *nodeData = (unsigned char *)targetBvh;
+	unsigned char* nodeData = (unsigned char*)targetBvh;
 	nodeData += sizeof(b3QuantizedBvh);
-	
-	unsigned sizeToAdd = 0;//(BVH_ALIGNMENT-((unsigned)nodeData & BVH_ALIGNMENT_MASK))&BVH_ALIGNMENT_MASK;
+
+	unsigned sizeToAdd = 0;  //(BVH_ALIGNMENT-((unsigned)nodeData & BVH_ALIGNMENT_MASK))&BVH_ALIGNMENT_MASK;
 	nodeData += sizeToAdd;
-	
+
 	int nodeCount = m_curNodeIndex;
 
 	if (m_useQuantization)
@@ -915,7 +882,6 @@ bool b3QuantizedBvh::serialize(void *o_alignedDataBuffer, unsigned /*i_dataBuffe
 		{
 			for (int nodeIndex = 0; nodeIndex < nodeCount; nodeIndex++)
 			{
-	
 				targetBvh->m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMin[0] = m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMin[0];
 				targetBvh->m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMin[1] = m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMin[1];
 				targetBvh->m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMin[2] = m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMin[2];
@@ -925,8 +891,6 @@ bool b3QuantizedBvh::serialize(void *o_alignedDataBuffer, unsigned /*i_dataBuffe
 				targetBvh->m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMax[2] = m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMax[2];
 
 				targetBvh->m_quantizedContiguousNodes[nodeIndex].m_escapeIndexOrTriangleIndex = m_quantizedContiguousNodes[nodeIndex].m_escapeIndexOrTriangleIndex;
-
-
 			}
 		}
 		nodeData += sizeof(b3QuantizedBvhNode) * nodeCount;
@@ -972,7 +936,7 @@ bool b3QuantizedBvh::serialize(void *o_alignedDataBuffer, unsigned /*i_dataBuffe
 		targetBvh->m_contiguousNodes.initializeFromBuffer(NULL, 0, 0);
 	}
 
-	sizeToAdd = 0;//(BVH_ALIGNMENT-((unsigned)nodeData & BVH_ALIGNMENT_MASK))&BVH_ALIGNMENT_MASK;
+	sizeToAdd = 0;  //(BVH_ALIGNMENT-((unsigned)nodeData & BVH_ALIGNMENT_MASK))&BVH_ALIGNMENT_MASK;
 	nodeData += sizeToAdd;
 
 	// Now serialize the subtree headers
@@ -1027,14 +991,13 @@ bool b3QuantizedBvh::serialize(void *o_alignedDataBuffer, unsigned /*i_dataBuffe
 	return true;
 }
 
-b3QuantizedBvh *b3QuantizedBvh::deSerializeInPlace(void *i_alignedDataBuffer, unsigned int i_dataBufferSize, bool i_swapEndian)
+b3QuantizedBvh* b3QuantizedBvh::deSerializeInPlace(void* i_alignedDataBuffer, unsigned int i_dataBufferSize, bool i_swapEndian)
 {
-
-	if (i_alignedDataBuffer == NULL)// || (((unsigned)i_alignedDataBuffer & BVH_ALIGNMENT_MASK) != 0))
+	if (i_alignedDataBuffer == NULL)  // || (((unsigned)i_alignedDataBuffer & BVH_ALIGNMENT_MASK) != 0))
 	{
 		return NULL;
 	}
-	b3QuantizedBvh *bvh = (b3QuantizedBvh *)i_alignedDataBuffer;
+	b3QuantizedBvh* bvh = (b3QuantizedBvh*)i_alignedDataBuffer;
 
 	if (i_swapEndian)
 	{
@@ -1056,12 +1019,12 @@ b3QuantizedBvh *b3QuantizedBvh::deSerializeInPlace(void *i_alignedDataBuffer, un
 		return NULL;
 	}
 
-	unsigned char *nodeData = (unsigned char *)bvh;
+	unsigned char* nodeData = (unsigned char*)bvh;
 	nodeData += sizeof(b3QuantizedBvh);
-	
-	unsigned sizeToAdd = 0;//(BVH_ALIGNMENT-((unsigned)nodeData & BVH_ALIGNMENT_MASK))&BVH_ALIGNMENT_MASK;
+
+	unsigned sizeToAdd = 0;  //(BVH_ALIGNMENT-((unsigned)nodeData & BVH_ALIGNMENT_MASK))&BVH_ALIGNMENT_MASK;
 	nodeData += sizeToAdd;
-	
+
 	int nodeCount = bvh->m_curNodeIndex;
 
 	// Must call placement new to fill in virtual function table, etc, but we don't want to overwrite most data, so call a special version of the constructor
@@ -1099,7 +1062,7 @@ b3QuantizedBvh *b3QuantizedBvh::deSerializeInPlace(void *i_alignedDataBuffer, un
 			{
 				b3UnSwapVector3Endian(bvh->m_contiguousNodes[nodeIndex].m_aabbMinOrg);
 				b3UnSwapVector3Endian(bvh->m_contiguousNodes[nodeIndex].m_aabbMaxOrg);
-				
+
 				bvh->m_contiguousNodes[nodeIndex].m_escapeIndex = static_cast<int>(b3SwapEndian(bvh->m_contiguousNodes[nodeIndex].m_escapeIndex));
 				bvh->m_contiguousNodes[nodeIndex].m_subPart = static_cast<int>(b3SwapEndian(bvh->m_contiguousNodes[nodeIndex].m_subPart));
 				bvh->m_contiguousNodes[nodeIndex].m_triangleIndex = static_cast<int>(b3SwapEndian(bvh->m_contiguousNodes[nodeIndex].m_triangleIndex));
@@ -1108,7 +1071,7 @@ b3QuantizedBvh *b3QuantizedBvh::deSerializeInPlace(void *i_alignedDataBuffer, un
 		nodeData += sizeof(b3OptimizedBvhNode) * nodeCount;
 	}
 
-	sizeToAdd = 0;//(BVH_ALIGNMENT-((unsigned)nodeData & BVH_ALIGNMENT_MASK))&BVH_ALIGNMENT_MASK;
+	sizeToAdd = 0;  //(BVH_ALIGNMENT-((unsigned)nodeData & BVH_ALIGNMENT_MASK))&BVH_ALIGNMENT_MASK;
 	nodeData += sizeToAdd;
 
 	// Now serialize the subtree headers
@@ -1134,13 +1097,11 @@ b3QuantizedBvh *b3QuantizedBvh::deSerializeInPlace(void *i_alignedDataBuffer, un
 }
 
 // Constructor that prevents b3Vector3's default constructor from being called
-b3QuantizedBvh::b3QuantizedBvh(b3QuantizedBvh &self, bool /* ownsMemory */) :
-m_bvhAabbMin(self.m_bvhAabbMin),
-m_bvhAabbMax(self.m_bvhAabbMax),
-m_bvhQuantization(self.m_bvhQuantization),
-m_bulletVersion(B3_BULLET_VERSION)
+b3QuantizedBvh::b3QuantizedBvh(b3QuantizedBvh& self, bool /* ownsMemory */) : m_bvhAabbMin(self.m_bvhAabbMin),
+																			  m_bvhAabbMax(self.m_bvhAabbMax),
+																			  m_bvhQuantization(self.m_bvhQuantization),
+																			  m_bulletVersion(B3_BULLET_VERSION)
 {
-
 }
 
 void b3QuantizedBvh::deSerializeFloat(struct b3QuantizedBvhFloatData& quantizedBvhFloatData)
@@ -1150,8 +1111,8 @@ void b3QuantizedBvh::deSerializeFloat(struct b3QuantizedBvhFloatData& quantizedB
 	m_bvhQuantization.deSerializeFloat(quantizedBvhFloatData.m_bvhQuantization);
 
 	m_curNodeIndex = quantizedBvhFloatData.m_curNodeIndex;
-	m_useQuantization = quantizedBvhFloatData.m_useQuantization!=0;
-	
+	m_useQuantization = quantizedBvhFloatData.m_useQuantization != 0;
+
 	{
 		int numElem = quantizedBvhFloatData.m_numContiguousLeafNodes;
 		m_contiguousNodes.resize(numElem);
@@ -1160,7 +1121,7 @@ void b3QuantizedBvh::deSerializeFloat(struct b3QuantizedBvhFloatData& quantizedB
 		{
 			b3OptimizedBvhNodeFloatData* memPtr = quantizedBvhFloatData.m_contiguousNodesPtr;
 
-			for (int i=0;i<numElem;i++,memPtr++)
+			for (int i = 0; i < numElem; i++, memPtr++)
 			{
 				m_contiguousNodes[i].m_aabbMaxOrg.deSerializeFloat(memPtr->m_aabbMaxOrg);
 				m_contiguousNodes[i].m_aabbMinOrg.deSerializeFloat(memPtr->m_aabbMinOrg);
@@ -1174,11 +1135,11 @@ void b3QuantizedBvh::deSerializeFloat(struct b3QuantizedBvhFloatData& quantizedB
 	{
 		int numElem = quantizedBvhFloatData.m_numQuantizedContiguousNodes;
 		m_quantizedContiguousNodes.resize(numElem);
-		
+
 		if (numElem)
 		{
 			b3QuantizedBvhNodeData* memPtr = quantizedBvhFloatData.m_quantizedContiguousNodesPtr;
-			for (int i=0;i<numElem;i++,memPtr++)
+			for (int i = 0; i < numElem; i++, memPtr++)
 			{
 				m_quantizedContiguousNodes[i].m_escapeIndexOrTriangleIndex = memPtr->m_escapeIndexOrTriangleIndex;
 				m_quantizedContiguousNodes[i].m_quantizedAabbMax[0] = memPtr->m_quantizedAabbMax[0];
@@ -1192,16 +1153,16 @@ void b3QuantizedBvh::deSerializeFloat(struct b3QuantizedBvhFloatData& quantizedB
 	}
 
 	m_traversalMode = b3TraversalMode(quantizedBvhFloatData.m_traversalMode);
-	
+
 	{
 		int numElem = quantizedBvhFloatData.m_numSubtreeHeaders;
 		m_SubtreeHeaders.resize(numElem);
 		if (numElem)
 		{
 			b3BvhSubtreeInfoData* memPtr = quantizedBvhFloatData.m_subTreeInfoPtr;
-			for (int i=0;i<numElem;i++,memPtr++)
+			for (int i = 0; i < numElem; i++, memPtr++)
 			{
-				m_SubtreeHeaders[i].m_quantizedAabbMax[0] = memPtr->m_quantizedAabbMax[0] ;
+				m_SubtreeHeaders[i].m_quantizedAabbMax[0] = memPtr->m_quantizedAabbMax[0];
 				m_SubtreeHeaders[i].m_quantizedAabbMax[1] = memPtr->m_quantizedAabbMax[1];
 				m_SubtreeHeaders[i].m_quantizedAabbMax[2] = memPtr->m_quantizedAabbMax[2];
 				m_SubtreeHeaders[i].m_quantizedAabbMin[0] = memPtr->m_quantizedAabbMin[0];
@@ -1221,8 +1182,8 @@ void b3QuantizedBvh::deSerializeDouble(struct b3QuantizedBvhDoubleData& quantize
 	m_bvhQuantization.deSerializeDouble(quantizedBvhDoubleData.m_bvhQuantization);
 
 	m_curNodeIndex = quantizedBvhDoubleData.m_curNodeIndex;
-	m_useQuantization = quantizedBvhDoubleData.m_useQuantization!=0;
-	
+	m_useQuantization = quantizedBvhDoubleData.m_useQuantization != 0;
+
 	{
 		int numElem = quantizedBvhDoubleData.m_numContiguousLeafNodes;
 		m_contiguousNodes.resize(numElem);
@@ -1231,7 +1192,7 @@ void b3QuantizedBvh::deSerializeDouble(struct b3QuantizedBvhDoubleData& quantize
 		{
 			b3OptimizedBvhNodeDoubleData* memPtr = quantizedBvhDoubleData.m_contiguousNodesPtr;
 
-			for (int i=0;i<numElem;i++,memPtr++)
+			for (int i = 0; i < numElem; i++, memPtr++)
 			{
 				m_contiguousNodes[i].m_aabbMaxOrg.deSerializeDouble(memPtr->m_aabbMaxOrg);
 				m_contiguousNodes[i].m_aabbMinOrg.deSerializeDouble(memPtr->m_aabbMinOrg);
@@ -1245,11 +1206,11 @@ void b3QuantizedBvh::deSerializeDouble(struct b3QuantizedBvhDoubleData& quantize
 	{
 		int numElem = quantizedBvhDoubleData.m_numQuantizedContiguousNodes;
 		m_quantizedContiguousNodes.resize(numElem);
-		
+
 		if (numElem)
 		{
 			b3QuantizedBvhNodeData* memPtr = quantizedBvhDoubleData.m_quantizedContiguousNodesPtr;
-			for (int i=0;i<numElem;i++,memPtr++)
+			for (int i = 0; i < numElem; i++, memPtr++)
 			{
 				m_quantizedContiguousNodes[i].m_escapeIndexOrTriangleIndex = memPtr->m_escapeIndexOrTriangleIndex;
 				m_quantizedContiguousNodes[i].m_quantizedAabbMax[0] = memPtr->m_quantizedAabbMax[0];
@@ -1263,16 +1224,16 @@ void b3QuantizedBvh::deSerializeDouble(struct b3QuantizedBvhDoubleData& quantize
 	}
 
 	m_traversalMode = b3TraversalMode(quantizedBvhDoubleData.m_traversalMode);
-	
+
 	{
 		int numElem = quantizedBvhDoubleData.m_numSubtreeHeaders;
 		m_SubtreeHeaders.resize(numElem);
 		if (numElem)
 		{
 			b3BvhSubtreeInfoData* memPtr = quantizedBvhDoubleData.m_subTreeInfoPtr;
-			for (int i=0;i<numElem;i++,memPtr++)
+			for (int i = 0; i < numElem; i++, memPtr++)
 			{
-				m_SubtreeHeaders[i].m_quantizedAabbMax[0] = memPtr->m_quantizedAabbMax[0] ;
+				m_SubtreeHeaders[i].m_quantizedAabbMax[0] = memPtr->m_quantizedAabbMax[0];
 				m_SubtreeHeaders[i].m_quantizedAabbMax[1] = memPtr->m_quantizedAabbMax[1];
 				m_SubtreeHeaders[i].m_quantizedAabbMax[2] = memPtr->m_quantizedAabbMax[2];
 				m_SubtreeHeaders[i].m_quantizedAabbMin[0] = memPtr->m_quantizedAabbMin[0];
@@ -1283,19 +1244,11 @@ void b3QuantizedBvh::deSerializeDouble(struct b3QuantizedBvhDoubleData& quantize
 			}
 		}
 	}
-
 }
 
-
-
 ///fills the dataBuffer and returns the struct name (and 0 on failure)
-const char*	b3QuantizedBvh::serialize(void* dataBuffer, b3Serializer* serializer) const
+const char* b3QuantizedBvh::serialize(void* dataBuffer, b3Serializer* serializer) const
 {
 	b3Assert(0);
 	return 0;
 }
-
-
-
-
-
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3QuantizedBvh.h b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3QuantizedBvh.h
index 63c523c7584..48b41abcad3 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3QuantizedBvh.h
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3QuantizedBvh.h
@@ -22,11 +22,11 @@ class b3Serializer;
 #ifdef DEBUG_CHECK_DEQUANTIZATION
 #ifdef __SPU__
 #define printf spu_printf
-#endif //__SPU__
+#endif  //__SPU__
 
 #include <stdio.h>
 #include <stdlib.h>
-#endif //DEBUG_CHECK_DEQUANTIZATION
+#endif  //DEBUG_CHECK_DEQUANTIZATION
 
 #include "Bullet3Common/b3Vector3.h"
 #include "Bullet3Common/b3AlignedAllocator.h"
@@ -44,13 +44,10 @@ class b3Serializer;
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3QuantizedBvhNodeData.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3BvhSubtreeInfoData.h"
 
-
-
 //http://msdn.microsoft.com/library/default.asp?url=/library/en-us/vclang/html/vclrf__m128.asp
 
-
 //Note: currently we have 16 bytes per quantized node
-#define MAX_SUBTREE_SIZE_IN_BYTES  2048
+#define MAX_SUBTREE_SIZE_IN_BYTES 2048
 
 // 10 gives the potential for 1024 parts, with at most 2^21 (2097152) (minus one
 // actually) triangles each (since the sign bit is reserved
@@ -58,7 +55,8 @@ class b3Serializer;
 
 ///b3QuantizedBvhNode is a compressed aabb node, 16 bytes.
 ///Node can be used for leafnode or internal node. Leafnodes can point to 32-bit triangle index (non-negative range).
-B3_ATTRIBUTE_ALIGNED16	(struct) b3QuantizedBvhNode : public b3QuantizedBvhNodeData
+B3_ATTRIBUTE_ALIGNED16(struct)
+b3QuantizedBvhNode : public b3QuantizedBvhNodeData
 {
 	B3_DECLARE_ALIGNED_ALLOCATOR();
 
@@ -72,48 +70,48 @@ B3_ATTRIBUTE_ALIGNED16	(struct) b3QuantizedBvhNode : public b3QuantizedBvhNodeDa
 		b3Assert(!isLeafNode());
 		return -m_escapeIndexOrTriangleIndex;
 	}
-	int	getTriangleIndex() const
+	int getTriangleIndex() const
 	{
 		b3Assert(isLeafNode());
-		unsigned int x=0;
-		unsigned int y = (~(x&0))<<(31-MAX_NUM_PARTS_IN_BITS);
+		unsigned int x = 0;
+		unsigned int y = (~(x & 0)) << (31 - MAX_NUM_PARTS_IN_BITS);
 		// Get only the lower bits where the triangle index is stored
-		return (m_escapeIndexOrTriangleIndex&~(y));
+		return (m_escapeIndexOrTriangleIndex & ~(y));
 	}
-	int	getPartId() const
+	int getPartId() const
 	{
 		b3Assert(isLeafNode());
 		// Get only the highest bits where the part index is stored
-		return (m_escapeIndexOrTriangleIndex>>(31-MAX_NUM_PARTS_IN_BITS));
+		return (m_escapeIndexOrTriangleIndex >> (31 - MAX_NUM_PARTS_IN_BITS));
 	}
-}
-;
+};
 
 /// b3OptimizedBvhNode contains both internal and leaf node information.
 /// Total node size is 44 bytes / node. You can use the compressed version of 16 bytes.
-B3_ATTRIBUTE_ALIGNED16 (struct) b3OptimizedBvhNode
+B3_ATTRIBUTE_ALIGNED16(struct)
+b3OptimizedBvhNode
 {
 	B3_DECLARE_ALIGNED_ALLOCATOR();
 
 	//32 bytes
-	b3Vector3	m_aabbMinOrg;
-	b3Vector3	m_aabbMaxOrg;
+	b3Vector3 m_aabbMinOrg;
+	b3Vector3 m_aabbMaxOrg;
 
 	//4
-	int	m_escapeIndex;
+	int m_escapeIndex;
 
 	//8
 	//for child nodes
-	int	m_subPart;
-	int	m_triangleIndex;
+	int m_subPart;
+	int m_triangleIndex;
 
-//pad the size to 64 bytes
-	char	m_padding[20];
+	//pad the size to 64 bytes
+	char m_padding[20];
 };
 
-
 ///b3BvhSubtreeInfo provides info to gather a subtree of limited size
-B3_ATTRIBUTE_ALIGNED16(class) b3BvhSubtreeInfo : public b3BvhSubtreeInfoData
+B3_ATTRIBUTE_ALIGNED16(class)
+b3BvhSubtreeInfo : public b3BvhSubtreeInfoData
 {
 public:
 	B3_DECLARE_ALIGNED_ALLOCATOR();
@@ -123,8 +121,7 @@ public:
 		//memset(&m_padding[0], 0, sizeof(m_padding));
 	}
 
-
-	void	setAabbFromQuantizeNode(const b3QuantizedBvhNode& quantizedNode)
+	void setAabbFromQuantizeNode(const b3QuantizedBvhNode& quantizedNode)
 	{
 		m_quantizedAabbMin[0] = quantizedNode.m_quantizedAabbMin[0];
 		m_quantizedAabbMin[1] = quantizedNode.m_quantizedAabbMin[1];
@@ -133,14 +130,12 @@ public:
 		m_quantizedAabbMax[1] = quantizedNode.m_quantizedAabbMax[1];
 		m_quantizedAabbMax[2] = quantizedNode.m_quantizedAabbMax[2];
 	}
-}
-;
-
+};
 
 class b3NodeOverlapCallback
 {
 public:
-	virtual ~b3NodeOverlapCallback() {};
+	virtual ~b3NodeOverlapCallback(){};
 
 	virtual void processNode(int subPart, int triangleIndex) = 0;
 };
@@ -148,18 +143,16 @@ public:
 #include "Bullet3Common/b3AlignedAllocator.h"
 #include "Bullet3Common/b3AlignedObjectArray.h"
 
-
-
 ///for code readability:
-typedef b3AlignedObjectArray<b3OptimizedBvhNode>	NodeArray;
-typedef b3AlignedObjectArray<b3QuantizedBvhNode>	QuantizedNodeArray;
-typedef b3AlignedObjectArray<b3BvhSubtreeInfo>		BvhSubtreeInfoArray;
-
+typedef b3AlignedObjectArray<b3OptimizedBvhNode> NodeArray;
+typedef b3AlignedObjectArray<b3QuantizedBvhNode> QuantizedNodeArray;
+typedef b3AlignedObjectArray<b3BvhSubtreeInfo> BvhSubtreeInfoArray;
 
 ///The b3QuantizedBvh class stores an AABB tree that can be quickly traversed on CPU and Cell SPU.
 ///It is used by the b3BvhTriangleMeshShape as midphase
 ///It is recommended to use quantization for better performance and lower memory requirements.
-B3_ATTRIBUTE_ALIGNED16(class) b3QuantizedBvh
+B3_ATTRIBUTE_ALIGNED16(class)
+b3QuantizedBvh
 {
 public:
 	enum b3TraversalMode
@@ -169,56 +162,48 @@ public:
 		TRAVERSAL_RECURSIVE
 	};
 
-
-
-
-	b3Vector3			m_bvhAabbMin;
-	b3Vector3			m_bvhAabbMax;
-	b3Vector3			m_bvhQuantization;
+	b3Vector3 m_bvhAabbMin;
+	b3Vector3 m_bvhAabbMax;
+	b3Vector3 m_bvhQuantization;
 
 protected:
-	int					m_bulletVersion;	//for serialization versioning. It could also be used to detect endianess.
+	int m_bulletVersion;  //for serialization versioning. It could also be used to detect endianess.
 
-	int					m_curNodeIndex;
+	int m_curNodeIndex;
 	//quantization data
-	bool				m_useQuantization;
+	bool m_useQuantization;
 
+	NodeArray m_leafNodes;
+	NodeArray m_contiguousNodes;
+	QuantizedNodeArray m_quantizedLeafNodes;
+	QuantizedNodeArray m_quantizedContiguousNodes;
 
-
-	NodeArray			m_leafNodes;
-	NodeArray			m_contiguousNodes;
-	QuantizedNodeArray	m_quantizedLeafNodes;
-	QuantizedNodeArray	m_quantizedContiguousNodes;
-	
-	b3TraversalMode	m_traversalMode;
-	BvhSubtreeInfoArray		m_SubtreeHeaders;
+	b3TraversalMode m_traversalMode;
+	BvhSubtreeInfoArray m_SubtreeHeaders;
 
 	//This is only used for serialization so we don't have to add serialization directly to b3AlignedObjectArray
 	mutable int m_subtreeHeaderCount;
 
-	
-
-
-
 	///two versions, one for quantized and normal nodes. This allows code-reuse while maintaining readability (no template/macro!)
 	///this might be refactored into a virtual, it is usually not calculated at run-time
-	void	setInternalNodeAabbMin(int nodeIndex, const b3Vector3& aabbMin)
+	void setInternalNodeAabbMin(int nodeIndex, const b3Vector3& aabbMin)
 	{
 		if (m_useQuantization)
 		{
-			quantize(&m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMin[0] ,aabbMin,0);
-		} else
+			quantize(&m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMin[0], aabbMin, 0);
+		}
+		else
 		{
 			m_contiguousNodes[nodeIndex].m_aabbMinOrg = aabbMin;
-
 		}
 	}
-	void	setInternalNodeAabbMax(int nodeIndex,const b3Vector3& aabbMax)
+	void setInternalNodeAabbMax(int nodeIndex, const b3Vector3& aabbMax)
 	{
 		if (m_useQuantization)
 		{
-			quantize(&m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMax[0],aabbMax,1);
-		} else
+			quantize(&m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMax[0], aabbMax, 1);
+		}
+		else
 		{
 			m_contiguousNodes[nodeIndex].m_aabbMaxOrg = aabbMax;
 		}
@@ -232,115 +217,102 @@ protected:
 		}
 		//non-quantized
 		return m_leafNodes[nodeIndex].m_aabbMinOrg;
-
 	}
 	b3Vector3 getAabbMax(int nodeIndex) const
 	{
 		if (m_useQuantization)
 		{
 			return unQuantize(&m_quantizedLeafNodes[nodeIndex].m_quantizedAabbMax[0]);
-		} 
+		}
 		//non-quantized
 		return m_leafNodes[nodeIndex].m_aabbMaxOrg;
-		
 	}
 
-	
-	void	setInternalNodeEscapeIndex(int nodeIndex, int escapeIndex)
+	void setInternalNodeEscapeIndex(int nodeIndex, int escapeIndex)
 	{
 		if (m_useQuantization)
 		{
 			m_quantizedContiguousNodes[nodeIndex].m_escapeIndexOrTriangleIndex = -escapeIndex;
-		} 
+		}
 		else
 		{
 			m_contiguousNodes[nodeIndex].m_escapeIndex = escapeIndex;
 		}
-
 	}
 
-	void mergeInternalNodeAabb(int nodeIndex,const b3Vector3& newAabbMin,const b3Vector3& newAabbMax) 
+	void mergeInternalNodeAabb(int nodeIndex, const b3Vector3& newAabbMin, const b3Vector3& newAabbMax)
 	{
 		if (m_useQuantization)
 		{
 			unsigned short int quantizedAabbMin[3];
 			unsigned short int quantizedAabbMax[3];
-			quantize(quantizedAabbMin,newAabbMin,0);
-			quantize(quantizedAabbMax,newAabbMax,1);
-			for (int i=0;i<3;i++)
+			quantize(quantizedAabbMin, newAabbMin, 0);
+			quantize(quantizedAabbMax, newAabbMax, 1);
+			for (int i = 0; i < 3; i++)
 			{
 				if (m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMin[i] > quantizedAabbMin[i])
 					m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMin[i] = quantizedAabbMin[i];
 
 				if (m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMax[i] < quantizedAabbMax[i])
 					m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMax[i] = quantizedAabbMax[i];
-
 			}
-		} else
+		}
+		else
 		{
 			//non-quantized
 			m_contiguousNodes[nodeIndex].m_aabbMinOrg.setMin(newAabbMin);
-			m_contiguousNodes[nodeIndex].m_aabbMaxOrg.setMax(newAabbMax);		
+			m_contiguousNodes[nodeIndex].m_aabbMaxOrg.setMax(newAabbMax);
 		}
 	}
 
-	void	swapLeafNodes(int firstIndex,int secondIndex);
+	void swapLeafNodes(int firstIndex, int secondIndex);
 
-	void	assignInternalNodeFromLeafNode(int internalNode,int leafNodeIndex);
+	void assignInternalNodeFromLeafNode(int internalNode, int leafNodeIndex);
 
 protected:
+	void buildTree(int startIndex, int endIndex);
 
-	
+	int calcSplittingAxis(int startIndex, int endIndex);
 
-	void	buildTree	(int startIndex,int endIndex);
+	int sortAndCalcSplittingIndex(int startIndex, int endIndex, int splitAxis);
 
-	int	calcSplittingAxis(int startIndex,int endIndex);
+	void walkStacklessTree(b3NodeOverlapCallback * nodeCallback, const b3Vector3& aabbMin, const b3Vector3& aabbMax) const;
 
-	int	sortAndCalcSplittingIndex(int startIndex,int endIndex,int splitAxis);
-	
-	void	walkStacklessTree(b3NodeOverlapCallback* nodeCallback,const b3Vector3& aabbMin,const b3Vector3& aabbMax) const;
-
-	void	walkStacklessQuantizedTreeAgainstRay(b3NodeOverlapCallback* nodeCallback, const b3Vector3& raySource, const b3Vector3& rayTarget, const b3Vector3& aabbMin, const b3Vector3& aabbMax, int startNodeIndex,int endNodeIndex) const;
-	void	walkStacklessQuantizedTree(b3NodeOverlapCallback* nodeCallback,unsigned short int* quantizedQueryAabbMin,unsigned short int* quantizedQueryAabbMax,int startNodeIndex,int endNodeIndex) const;
-	void	walkStacklessTreeAgainstRay(b3NodeOverlapCallback* nodeCallback, const b3Vector3& raySource, const b3Vector3& rayTarget, const b3Vector3& aabbMin, const b3Vector3& aabbMax, int startNodeIndex,int endNodeIndex) const;
+	void walkStacklessQuantizedTreeAgainstRay(b3NodeOverlapCallback * nodeCallback, const b3Vector3& raySource, const b3Vector3& rayTarget, const b3Vector3& aabbMin, const b3Vector3& aabbMax, int startNodeIndex, int endNodeIndex) const;
+	void walkStacklessQuantizedTree(b3NodeOverlapCallback * nodeCallback, unsigned short int* quantizedQueryAabbMin, unsigned short int* quantizedQueryAabbMax, int startNodeIndex, int endNodeIndex) const;
+	void walkStacklessTreeAgainstRay(b3NodeOverlapCallback * nodeCallback, const b3Vector3& raySource, const b3Vector3& rayTarget, const b3Vector3& aabbMin, const b3Vector3& aabbMax, int startNodeIndex, int endNodeIndex) const;
 
 	///tree traversal designed for small-memory processors like PS3 SPU
-	void	walkStacklessQuantizedTreeCacheFriendly(b3NodeOverlapCallback* nodeCallback,unsigned short int* quantizedQueryAabbMin,unsigned short int* quantizedQueryAabbMax) const;
+	void walkStacklessQuantizedTreeCacheFriendly(b3NodeOverlapCallback * nodeCallback, unsigned short int* quantizedQueryAabbMin, unsigned short int* quantizedQueryAabbMax) const;
 
 	///use the 16-byte stackless 'skipindex' node tree to do a recursive traversal
-	void	walkRecursiveQuantizedTreeAgainstQueryAabb(const b3QuantizedBvhNode* currentNode,b3NodeOverlapCallback* nodeCallback,unsigned short int* quantizedQueryAabbMin,unsigned short int* quantizedQueryAabbMax) const;
+	void walkRecursiveQuantizedTreeAgainstQueryAabb(const b3QuantizedBvhNode* currentNode, b3NodeOverlapCallback* nodeCallback, unsigned short int* quantizedQueryAabbMin, unsigned short int* quantizedQueryAabbMax) const;
 
 	///use the 16-byte stackless 'skipindex' node tree to do a recursive traversal
-	void	walkRecursiveQuantizedTreeAgainstQuantizedTree(const b3QuantizedBvhNode* treeNodeA,const b3QuantizedBvhNode* treeNodeB,b3NodeOverlapCallback* nodeCallback) const;
-	
+	void walkRecursiveQuantizedTreeAgainstQuantizedTree(const b3QuantizedBvhNode* treeNodeA, const b3QuantizedBvhNode* treeNodeB, b3NodeOverlapCallback* nodeCallback) const;
 
-
-
-	void	updateSubtreeHeaders(int leftChildNodexIndex,int rightChildNodexIndex);
+	void updateSubtreeHeaders(int leftChildNodexIndex, int rightChildNodexIndex);
 
 public:
-	
 	B3_DECLARE_ALIGNED_ALLOCATOR();
 
 	b3QuantizedBvh();
 
 	virtual ~b3QuantizedBvh();
 
-	
 	///***************************************** expert/internal use only *************************
-	void	setQuantizationValues(const b3Vector3& bvhAabbMin,const b3Vector3& bvhAabbMax,b3Scalar quantizationMargin=b3Scalar(1.0));
-	QuantizedNodeArray&	getLeafNodeArray() {			return	m_quantizedLeafNodes;	}
+	void setQuantizationValues(const b3Vector3& bvhAabbMin, const b3Vector3& bvhAabbMax, b3Scalar quantizationMargin = b3Scalar(1.0));
+	QuantizedNodeArray& getLeafNodeArray() { return m_quantizedLeafNodes; }
 	///buildInternal is expert use only: assumes that setQuantizationValues and LeafNodeArray are initialized
-	void	buildInternal();
+	void buildInternal();
 	///***************************************** expert/internal use only *************************
 
-	void	reportAabbOverlappingNodex(b3NodeOverlapCallback* nodeCallback,const b3Vector3& aabbMin,const b3Vector3& aabbMax) const;
-	void	reportRayOverlappingNodex (b3NodeOverlapCallback* nodeCallback, const b3Vector3& raySource, const b3Vector3& rayTarget) const;
-	void	reportBoxCastOverlappingNodex(b3NodeOverlapCallback* nodeCallback, const b3Vector3& raySource, const b3Vector3& rayTarget, const b3Vector3& aabbMin,const b3Vector3& aabbMax) const;
+	void reportAabbOverlappingNodex(b3NodeOverlapCallback * nodeCallback, const b3Vector3& aabbMin, const b3Vector3& aabbMax) const;
+	void reportRayOverlappingNodex(b3NodeOverlapCallback * nodeCallback, const b3Vector3& raySource, const b3Vector3& rayTarget) const;
+	void reportBoxCastOverlappingNodex(b3NodeOverlapCallback * nodeCallback, const b3Vector3& raySource, const b3Vector3& rayTarget, const b3Vector3& aabbMin, const b3Vector3& aabbMax) const;
 
-		B3_FORCE_INLINE void quantize(unsigned short* out, const b3Vector3& point,int isMax) const
+	B3_FORCE_INLINE void quantize(unsigned short* out, const b3Vector3& point, int isMax) const
 	{
-
 		b3Assert(m_useQuantization);
 
 		b3Assert(point.getX() <= m_bvhAabbMax.getX());
@@ -357,16 +329,16 @@ public:
 		///@todo: double-check this
 		if (isMax)
 		{
-			out[0] = (unsigned short) (((unsigned short)(v.getX()+b3Scalar(1.)) | 1));
-			out[1] = (unsigned short) (((unsigned short)(v.getY()+b3Scalar(1.)) | 1));
-			out[2] = (unsigned short) (((unsigned short)(v.getZ()+b3Scalar(1.)) | 1));
-		} else
-		{
-			out[0] = (unsigned short) (((unsigned short)(v.getX()) & 0xfffe));
-			out[1] = (unsigned short) (((unsigned short)(v.getY()) & 0xfffe));
-			out[2] = (unsigned short) (((unsigned short)(v.getZ()) & 0xfffe));
+			out[0] = (unsigned short)(((unsigned short)(v.getX() + b3Scalar(1.)) | 1));
+			out[1] = (unsigned short)(((unsigned short)(v.getY() + b3Scalar(1.)) | 1));
+			out[2] = (unsigned short)(((unsigned short)(v.getZ() + b3Scalar(1.)) | 1));
+		}
+		else
+		{
+			out[0] = (unsigned short)(((unsigned short)(v.getX()) & 0xfffe));
+			out[1] = (unsigned short)(((unsigned short)(v.getY()) & 0xfffe));
+			out[2] = (unsigned short)(((unsigned short)(v.getZ()) & 0xfffe));
 		}
-
 
 #ifdef DEBUG_CHECK_DEQUANTIZATION
 		b3Vector3 newPoint = unQuantize(out);
@@ -374,105 +346,97 @@ public:
 		{
 			if (newPoint.getX() < point.getX())
 			{
-				printf("unconservative X, diffX = %f, oldX=%f,newX=%f\n",newPoint.getX()-point.getX(), newPoint.getX(),point.getX());
+				printf("unconservative X, diffX = %f, oldX=%f,newX=%f\n", newPoint.getX() - point.getX(), newPoint.getX(), point.getX());
 			}
 			if (newPoint.getY() < point.getY())
 			{
-				printf("unconservative Y, diffY = %f, oldY=%f,newY=%f\n",newPoint.getY()-point.getY(), newPoint.getY(),point.getY());
+				printf("unconservative Y, diffY = %f, oldY=%f,newY=%f\n", newPoint.getY() - point.getY(), newPoint.getY(), point.getY());
 			}
 			if (newPoint.getZ() < point.getZ())
 			{
-
-				printf("unconservative Z, diffZ = %f, oldZ=%f,newZ=%f\n",newPoint.getZ()-point.getZ(), newPoint.getZ(),point.getZ());
+				printf("unconservative Z, diffZ = %f, oldZ=%f,newZ=%f\n", newPoint.getZ() - point.getZ(), newPoint.getZ(), point.getZ());
 			}
-		} else
+		}
+		else
 		{
 			if (newPoint.getX() > point.getX())
 			{
-				printf("unconservative X, diffX = %f, oldX=%f,newX=%f\n",newPoint.getX()-point.getX(), newPoint.getX(),point.getX());
+				printf("unconservative X, diffX = %f, oldX=%f,newX=%f\n", newPoint.getX() - point.getX(), newPoint.getX(), point.getX());
 			}
 			if (newPoint.getY() > point.getY())
 			{
-				printf("unconservative Y, diffY = %f, oldY=%f,newY=%f\n",newPoint.getY()-point.getY(), newPoint.getY(),point.getY());
+				printf("unconservative Y, diffY = %f, oldY=%f,newY=%f\n", newPoint.getY() - point.getY(), newPoint.getY(), point.getY());
 			}
 			if (newPoint.getZ() > point.getZ())
 			{
-				printf("unconservative Z, diffZ = %f, oldZ=%f,newZ=%f\n",newPoint.getZ()-point.getZ(), newPoint.getZ(),point.getZ());
+				printf("unconservative Z, diffZ = %f, oldZ=%f,newZ=%f\n", newPoint.getZ() - point.getZ(), newPoint.getZ(), point.getZ());
 			}
 		}
-#endif //DEBUG_CHECK_DEQUANTIZATION
-
+#endif  //DEBUG_CHECK_DEQUANTIZATION
 	}
 
-
-	B3_FORCE_INLINE void quantizeWithClamp(unsigned short* out, const b3Vector3& point2,int isMax) const
+	B3_FORCE_INLINE void quantizeWithClamp(unsigned short* out, const b3Vector3& point2, int isMax) const
 	{
-
 		b3Assert(m_useQuantization);
 
 		b3Vector3 clampedPoint(point2);
 		clampedPoint.setMax(m_bvhAabbMin);
 		clampedPoint.setMin(m_bvhAabbMax);
 
-		quantize(out,clampedPoint,isMax);
-
+		quantize(out, clampedPoint, isMax);
 	}
-	
-	B3_FORCE_INLINE b3Vector3	unQuantize(const unsigned short* vecIn) const
+
+	B3_FORCE_INLINE b3Vector3 unQuantize(const unsigned short* vecIn) const
 	{
-			b3Vector3	vecOut;
-			vecOut.setValue(
+		b3Vector3 vecOut;
+		vecOut.setValue(
 			(b3Scalar)(vecIn[0]) / (m_bvhQuantization.getX()),
 			(b3Scalar)(vecIn[1]) / (m_bvhQuantization.getY()),
 			(b3Scalar)(vecIn[2]) / (m_bvhQuantization.getZ()));
-			vecOut += m_bvhAabbMin;
-			return vecOut;
+		vecOut += m_bvhAabbMin;
+		return vecOut;
 	}
 
 	///setTraversalMode let's you choose between stackless, recursive or stackless cache friendly tree traversal. Note this is only implemented for quantized trees.
-	void	setTraversalMode(b3TraversalMode	traversalMode)
+	void setTraversalMode(b3TraversalMode traversalMode)
 	{
 		m_traversalMode = traversalMode;
 	}
 
-
-	B3_FORCE_INLINE QuantizedNodeArray&	getQuantizedNodeArray()
-	{	
-		return	m_quantizedContiguousNodes;
+	B3_FORCE_INLINE QuantizedNodeArray& getQuantizedNodeArray()
+	{
+		return m_quantizedContiguousNodes;
 	}
 
-
-	B3_FORCE_INLINE BvhSubtreeInfoArray&	getSubtreeInfoArray()
+	B3_FORCE_INLINE BvhSubtreeInfoArray& getSubtreeInfoArray()
 	{
 		return m_SubtreeHeaders;
 	}
 
-////////////////////////////////////////////////////////////////////
+	////////////////////////////////////////////////////////////////////
 
 	/////Calculate space needed to store BVH for serialization
 	unsigned calculateSerializeBufferSize() const;
 
 	/// Data buffer MUST be 16 byte aligned
-	virtual bool serialize(void *o_alignedDataBuffer, unsigned i_dataBufferSize, bool i_swapEndian) const;
+	virtual bool serialize(void* o_alignedDataBuffer, unsigned i_dataBufferSize, bool i_swapEndian) const;
 
 	///deSerializeInPlace loads and initializes a BVH from a buffer in memory 'in place'
-	static b3QuantizedBvh *deSerializeInPlace(void *i_alignedDataBuffer, unsigned int i_dataBufferSize, bool i_swapEndian);
+	static b3QuantizedBvh* deSerializeInPlace(void* i_alignedDataBuffer, unsigned int i_dataBufferSize, bool i_swapEndian);
 
 	static unsigned int getAlignmentSerializationPadding();
-//////////////////////////////////////////////////////////////////////
+	//////////////////////////////////////////////////////////////////////
 
-	
-	virtual	int	calculateSerializeBufferSizeNew() const;
+	virtual int calculateSerializeBufferSizeNew() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, b3Serializer* serializer) const;
+	virtual const char* serialize(void* dataBuffer, b3Serializer* serializer) const;
 
-	virtual	void deSerializeFloat(struct b3QuantizedBvhFloatData& quantizedBvhFloatData);
+	virtual void deSerializeFloat(struct b3QuantizedBvhFloatData & quantizedBvhFloatData);
 
-	virtual	void deSerializeDouble(struct b3QuantizedBvhDoubleData& quantizedBvhDoubleData);
+	virtual void deSerializeDouble(struct b3QuantizedBvhDoubleData & quantizedBvhDoubleData);
 
-
-////////////////////////////////////////////////////////////////////
+	////////////////////////////////////////////////////////////////////
 
 	B3_FORCE_INLINE bool isQuantized()
 	{
@@ -483,74 +447,65 @@ private:
 	// Special "copy" constructor that allows for in-place deserialization
 	// Prevents b3Vector3's default constructor from being called, but doesn't inialize much else
 	// ownsMemory should most likely be false if deserializing, and if you are not, don't call this (it also changes the function signature, which we need)
-	b3QuantizedBvh(b3QuantizedBvh &other, bool ownsMemory);
-
-}
-;
-
+	b3QuantizedBvh(b3QuantizedBvh & other, bool ownsMemory);
+};
 
 struct b3OptimizedBvhNodeFloatData
 {
-	b3Vector3FloatData	m_aabbMinOrg;
-	b3Vector3FloatData	m_aabbMaxOrg;
-	int	m_escapeIndex;
-	int	m_subPart;
-	int	m_triangleIndex;
+	b3Vector3FloatData m_aabbMinOrg;
+	b3Vector3FloatData m_aabbMaxOrg;
+	int m_escapeIndex;
+	int m_subPart;
+	int m_triangleIndex;
 	char m_pad[4];
 };
 
 struct b3OptimizedBvhNodeDoubleData
 {
-	b3Vector3DoubleData	m_aabbMinOrg;
-	b3Vector3DoubleData	m_aabbMaxOrg;
-	int	m_escapeIndex;
-	int	m_subPart;
-	int	m_triangleIndex;
-	char	m_pad[4];
+	b3Vector3DoubleData m_aabbMinOrg;
+	b3Vector3DoubleData m_aabbMaxOrg;
+	int m_escapeIndex;
+	int m_subPart;
+	int m_triangleIndex;
+	char m_pad[4];
 };
 
-
-
-struct	b3QuantizedBvhFloatData
+struct b3QuantizedBvhFloatData
 {
-	b3Vector3FloatData			m_bvhAabbMin;
-	b3Vector3FloatData			m_bvhAabbMax;
-	b3Vector3FloatData			m_bvhQuantization;
-	int					m_curNodeIndex;
-	int					m_useQuantization;
-	int					m_numContiguousLeafNodes;
-	int					m_numQuantizedContiguousNodes;
-	b3OptimizedBvhNodeFloatData	*m_contiguousNodesPtr;
-	b3QuantizedBvhNodeData		*m_quantizedContiguousNodesPtr;
-	b3BvhSubtreeInfoData	*m_subTreeInfoPtr;
-	int					m_traversalMode;
-	int					m_numSubtreeHeaders;
-	
+	b3Vector3FloatData m_bvhAabbMin;
+	b3Vector3FloatData m_bvhAabbMax;
+	b3Vector3FloatData m_bvhQuantization;
+	int m_curNodeIndex;
+	int m_useQuantization;
+	int m_numContiguousLeafNodes;
+	int m_numQuantizedContiguousNodes;
+	b3OptimizedBvhNodeFloatData* m_contiguousNodesPtr;
+	b3QuantizedBvhNodeData* m_quantizedContiguousNodesPtr;
+	b3BvhSubtreeInfoData* m_subTreeInfoPtr;
+	int m_traversalMode;
+	int m_numSubtreeHeaders;
 };
 
-struct	b3QuantizedBvhDoubleData
+struct b3QuantizedBvhDoubleData
 {
-	b3Vector3DoubleData			m_bvhAabbMin;
-	b3Vector3DoubleData			m_bvhAabbMax;
-	b3Vector3DoubleData			m_bvhQuantization;
-	int							m_curNodeIndex;
-	int							m_useQuantization;
-	int							m_numContiguousLeafNodes;
-	int							m_numQuantizedContiguousNodes;
-	b3OptimizedBvhNodeDoubleData	*m_contiguousNodesPtr;
-	b3QuantizedBvhNodeData			*m_quantizedContiguousNodesPtr;
+	b3Vector3DoubleData m_bvhAabbMin;
+	b3Vector3DoubleData m_bvhAabbMax;
+	b3Vector3DoubleData m_bvhQuantization;
+	int m_curNodeIndex;
+	int m_useQuantization;
+	int m_numContiguousLeafNodes;
+	int m_numQuantizedContiguousNodes;
+	b3OptimizedBvhNodeDoubleData* m_contiguousNodesPtr;
+	b3QuantizedBvhNodeData* m_quantizedContiguousNodesPtr;
 
-	int							m_traversalMode;
-	int							m_numSubtreeHeaders;
-	b3BvhSubtreeInfoData		*m_subTreeInfoPtr;
+	int m_traversalMode;
+	int m_numSubtreeHeaders;
+	b3BvhSubtreeInfoData* m_subTreeInfoPtr;
 };
 
-
-B3_FORCE_INLINE	int	b3QuantizedBvh::calculateSerializeBufferSizeNew() const
+B3_FORCE_INLINE int b3QuantizedBvh::calculateSerializeBufferSizeNew() const
 {
 	return sizeof(b3QuantizedBvhData);
 }
 
-
-
-#endif //B3_QUANTIZED_BVH_H
+#endif  //B3_QUANTIZED_BVH_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3StridingMeshInterface.cpp b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3StridingMeshInterface.cpp
index 4d97f7f62b9..6b0c941f23e 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3StridingMeshInterface.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3StridingMeshInterface.cpp
@@ -15,35 +15,32 @@ subject to the following restrictions:
 
 #include "b3StridingMeshInterface.h"
 
-
 b3StridingMeshInterface::~b3StridingMeshInterface()
 {
-
 }
 
-
-void	b3StridingMeshInterface::InternalProcessAllTriangles(b3InternalTriangleIndexCallback* callback,const b3Vector3& aabbMin,const b3Vector3& aabbMax) const
+void b3StridingMeshInterface::InternalProcessAllTriangles(b3InternalTriangleIndexCallback* callback, const b3Vector3& aabbMin, const b3Vector3& aabbMax) const
 {
 	(void)aabbMin;
 	(void)aabbMax;
 	int numtotalphysicsverts = 0;
-	int part,graphicssubparts = getNumSubParts();
-	const unsigned char * vertexbase;
-	const unsigned char * indexbase;
+	int part, graphicssubparts = getNumSubParts();
+	const unsigned char* vertexbase;
+	const unsigned char* indexbase;
 	int indexstride;
 	PHY_ScalarType type;
 	PHY_ScalarType gfxindextype;
-	int stride,numverts,numtriangles;
+	int stride, numverts, numtriangles;
 	int gfxindex;
 	b3Vector3 triangle[3];
 
 	b3Vector3 meshScaling = getScaling();
 
 	///if the number of parts is big, the performance might drop due to the innerloop switch on indextype
-	for (part=0;part<graphicssubparts ;part++)
+	for (part = 0; part < graphicssubparts; part++)
 	{
-		getLockedReadOnlyVertexIndexBase(&vertexbase,numverts,type,stride,&indexbase,indexstride,numtriangles,gfxindextype,part);
-		numtotalphysicsverts+=numtriangles*3; //upper bound
+		getLockedReadOnlyVertexIndexBase(&vertexbase, numverts, type, stride, &indexbase, indexstride, numtriangles, gfxindextype, part);
+		numtotalphysicsverts += numtriangles * 3;  //upper bound
 
 		///unlike that developers want to pass in double-precision meshes in single-precision Bullet build
 		///so disable this feature by default
@@ -51,143 +48,141 @@ void	b3StridingMeshInterface::InternalProcessAllTriangles(b3InternalTriangleInde
 
 		switch (type)
 		{
-		case PHY_FLOAT:
-		 {
+			case PHY_FLOAT:
+			{
+				float* graphicsbase;
 
-			 float* graphicsbase;
+				switch (gfxindextype)
+				{
+					case PHY_INTEGER:
+					{
+						for (gfxindex = 0; gfxindex < numtriangles; gfxindex++)
+						{
+							unsigned int* tri_indices = (unsigned int*)(indexbase + gfxindex * indexstride);
+							graphicsbase = (float*)(vertexbase + tri_indices[0] * stride);
+							triangle[0].setValue(graphicsbase[0] * meshScaling.getX(), graphicsbase[1] * meshScaling.getY(), graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (float*)(vertexbase + tri_indices[1] * stride);
+							triangle[1].setValue(graphicsbase[0] * meshScaling.getX(), graphicsbase[1] * meshScaling.getY(), graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (float*)(vertexbase + tri_indices[2] * stride);
+							triangle[2].setValue(graphicsbase[0] * meshScaling.getX(), graphicsbase[1] * meshScaling.getY(), graphicsbase[2] * meshScaling.getZ());
+							callback->internalProcessTriangleIndex(triangle, part, gfxindex);
+						}
+						break;
+					}
+					case PHY_SHORT:
+					{
+						for (gfxindex = 0; gfxindex < numtriangles; gfxindex++)
+						{
+							unsigned short int* tri_indices = (unsigned short int*)(indexbase + gfxindex * indexstride);
+							graphicsbase = (float*)(vertexbase + tri_indices[0] * stride);
+							triangle[0].setValue(graphicsbase[0] * meshScaling.getX(), graphicsbase[1] * meshScaling.getY(), graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (float*)(vertexbase + tri_indices[1] * stride);
+							triangle[1].setValue(graphicsbase[0] * meshScaling.getX(), graphicsbase[1] * meshScaling.getY(), graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (float*)(vertexbase + tri_indices[2] * stride);
+							triangle[2].setValue(graphicsbase[0] * meshScaling.getX(), graphicsbase[1] * meshScaling.getY(), graphicsbase[2] * meshScaling.getZ());
+							callback->internalProcessTriangleIndex(triangle, part, gfxindex);
+						}
+						break;
+					}
+					case PHY_UCHAR:
+					{
+						for (gfxindex = 0; gfxindex < numtriangles; gfxindex++)
+						{
+							unsigned char* tri_indices = (unsigned char*)(indexbase + gfxindex * indexstride);
+							graphicsbase = (float*)(vertexbase + tri_indices[0] * stride);
+							triangle[0].setValue(graphicsbase[0] * meshScaling.getX(), graphicsbase[1] * meshScaling.getY(), graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (float*)(vertexbase + tri_indices[1] * stride);
+							triangle[1].setValue(graphicsbase[0] * meshScaling.getX(), graphicsbase[1] * meshScaling.getY(), graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (float*)(vertexbase + tri_indices[2] * stride);
+							triangle[2].setValue(graphicsbase[0] * meshScaling.getX(), graphicsbase[1] * meshScaling.getY(), graphicsbase[2] * meshScaling.getZ());
+							callback->internalProcessTriangleIndex(triangle, part, gfxindex);
+						}
+						break;
+					}
+					default:
+						b3Assert((gfxindextype == PHY_INTEGER) || (gfxindextype == PHY_SHORT));
+				}
+				break;
+			}
 
-			 switch (gfxindextype)
-			 {
-			 case PHY_INTEGER:
-				 {
-					 for (gfxindex=0;gfxindex<numtriangles;gfxindex++)
-					 {
-						 unsigned int* tri_indices= (unsigned int*)(indexbase+gfxindex*indexstride);
-						 graphicsbase = (float*)(vertexbase+tri_indices[0]*stride);
-						 triangle[0].setValue(graphicsbase[0]*meshScaling.getX(),graphicsbase[1]*meshScaling.getY(),graphicsbase[2]*meshScaling.getZ());
-						 graphicsbase = (float*)(vertexbase+tri_indices[1]*stride);
-						 triangle[1].setValue(graphicsbase[0]*meshScaling.getX(),graphicsbase[1]*meshScaling.getY(),	graphicsbase[2]*meshScaling.getZ());
-						 graphicsbase = (float*)(vertexbase+tri_indices[2]*stride);
-						 triangle[2].setValue(graphicsbase[0]*meshScaling.getX(),graphicsbase[1]*meshScaling.getY(),	graphicsbase[2]*meshScaling.getZ());
-						 callback->internalProcessTriangleIndex(triangle,part,gfxindex);
-					 }
-					 break;
-				 }
-			 case PHY_SHORT:
-				 {
-					 for (gfxindex=0;gfxindex<numtriangles;gfxindex++)
-					 {
-						 unsigned short int* tri_indices= (unsigned short int*)(indexbase+gfxindex*indexstride);
-						 graphicsbase = (float*)(vertexbase+tri_indices[0]*stride);
-						 triangle[0].setValue(graphicsbase[0]*meshScaling.getX(),graphicsbase[1]*meshScaling.getY(),graphicsbase[2]*meshScaling.getZ());
-						 graphicsbase = (float*)(vertexbase+tri_indices[1]*stride);
-						 triangle[1].setValue(graphicsbase[0]*meshScaling.getX(),graphicsbase[1]*meshScaling.getY(),	graphicsbase[2]*meshScaling.getZ());
-						 graphicsbase = (float*)(vertexbase+tri_indices[2]*stride);
-						 triangle[2].setValue(graphicsbase[0]*meshScaling.getX(),graphicsbase[1]*meshScaling.getY(),	graphicsbase[2]*meshScaling.getZ());
-						 callback->internalProcessTriangleIndex(triangle,part,gfxindex);
-					 }
-					 break;
-				 }
-			case PHY_UCHAR:
-				 {
-					 for (gfxindex=0;gfxindex<numtriangles;gfxindex++)
-					 {
-						 unsigned char* tri_indices= (unsigned char*)(indexbase+gfxindex*indexstride);
-						 graphicsbase = (float*)(vertexbase+tri_indices[0]*stride);
-						 triangle[0].setValue(graphicsbase[0]*meshScaling.getX(),graphicsbase[1]*meshScaling.getY(),graphicsbase[2]*meshScaling.getZ());
-						 graphicsbase = (float*)(vertexbase+tri_indices[1]*stride);
-						 triangle[1].setValue(graphicsbase[0]*meshScaling.getX(),graphicsbase[1]*meshScaling.getY(),	graphicsbase[2]*meshScaling.getZ());
-						 graphicsbase = (float*)(vertexbase+tri_indices[2]*stride);
-						 triangle[2].setValue(graphicsbase[0]*meshScaling.getX(),graphicsbase[1]*meshScaling.getY(),	graphicsbase[2]*meshScaling.getZ());
-						 callback->internalProcessTriangleIndex(triangle,part,gfxindex);
-					 }
-					 break;
-				 }
-			 default:
-				 b3Assert((gfxindextype == PHY_INTEGER) || (gfxindextype == PHY_SHORT));
-			 }
-			 break;
-		 }
-
-		case PHY_DOUBLE:
+			case PHY_DOUBLE:
 			{
 				double* graphicsbase;
 
 				switch (gfxindextype)
 				{
-				case PHY_INTEGER:
+					case PHY_INTEGER:
 					{
-						for (gfxindex=0;gfxindex<numtriangles;gfxindex++)
+						for (gfxindex = 0; gfxindex < numtriangles; gfxindex++)
 						{
-							unsigned int* tri_indices= (unsigned int*)(indexbase+gfxindex*indexstride);
-							graphicsbase = (double*)(vertexbase+tri_indices[0]*stride);
-							triangle[0].setValue((b3Scalar)graphicsbase[0]*meshScaling.getX(),(b3Scalar)graphicsbase[1]*meshScaling.getY(),(b3Scalar)graphicsbase[2]*meshScaling.getZ());
-							graphicsbase = (double*)(vertexbase+tri_indices[1]*stride);
-							triangle[1].setValue((b3Scalar)graphicsbase[0]*meshScaling.getX(),(b3Scalar)graphicsbase[1]*meshScaling.getY(),  (b3Scalar)graphicsbase[2]*meshScaling.getZ());
-							graphicsbase = (double*)(vertexbase+tri_indices[2]*stride);
-							triangle[2].setValue((b3Scalar)graphicsbase[0]*meshScaling.getX(),(b3Scalar)graphicsbase[1]*meshScaling.getY(),  (b3Scalar)graphicsbase[2]*meshScaling.getZ());
-							callback->internalProcessTriangleIndex(triangle,part,gfxindex);
+							unsigned int* tri_indices = (unsigned int*)(indexbase + gfxindex * indexstride);
+							graphicsbase = (double*)(vertexbase + tri_indices[0] * stride);
+							triangle[0].setValue((b3Scalar)graphicsbase[0] * meshScaling.getX(), (b3Scalar)graphicsbase[1] * meshScaling.getY(), (b3Scalar)graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (double*)(vertexbase + tri_indices[1] * stride);
+							triangle[1].setValue((b3Scalar)graphicsbase[0] * meshScaling.getX(), (b3Scalar)graphicsbase[1] * meshScaling.getY(), (b3Scalar)graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (double*)(vertexbase + tri_indices[2] * stride);
+							triangle[2].setValue((b3Scalar)graphicsbase[0] * meshScaling.getX(), (b3Scalar)graphicsbase[1] * meshScaling.getY(), (b3Scalar)graphicsbase[2] * meshScaling.getZ());
+							callback->internalProcessTriangleIndex(triangle, part, gfxindex);
 						}
 						break;
 					}
-				case PHY_SHORT:
+					case PHY_SHORT:
 					{
-						for (gfxindex=0;gfxindex<numtriangles;gfxindex++)
+						for (gfxindex = 0; gfxindex < numtriangles; gfxindex++)
 						{
-							unsigned short int* tri_indices= (unsigned short int*)(indexbase+gfxindex*indexstride);
-							graphicsbase = (double*)(vertexbase+tri_indices[0]*stride);
-							triangle[0].setValue((b3Scalar)graphicsbase[0]*meshScaling.getX(),(b3Scalar)graphicsbase[1]*meshScaling.getY(),(b3Scalar)graphicsbase[2]*meshScaling.getZ());
-							graphicsbase = (double*)(vertexbase+tri_indices[1]*stride);
-							triangle[1].setValue((b3Scalar)graphicsbase[0]*meshScaling.getX(),(b3Scalar)graphicsbase[1]*meshScaling.getY(),  (b3Scalar)graphicsbase[2]*meshScaling.getZ());
-							graphicsbase = (double*)(vertexbase+tri_indices[2]*stride);
-							triangle[2].setValue((b3Scalar)graphicsbase[0]*meshScaling.getX(),(b3Scalar)graphicsbase[1]*meshScaling.getY(),  (b3Scalar)graphicsbase[2]*meshScaling.getZ());
-							callback->internalProcessTriangleIndex(triangle,part,gfxindex);
+							unsigned short int* tri_indices = (unsigned short int*)(indexbase + gfxindex * indexstride);
+							graphicsbase = (double*)(vertexbase + tri_indices[0] * stride);
+							triangle[0].setValue((b3Scalar)graphicsbase[0] * meshScaling.getX(), (b3Scalar)graphicsbase[1] * meshScaling.getY(), (b3Scalar)graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (double*)(vertexbase + tri_indices[1] * stride);
+							triangle[1].setValue((b3Scalar)graphicsbase[0] * meshScaling.getX(), (b3Scalar)graphicsbase[1] * meshScaling.getY(), (b3Scalar)graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (double*)(vertexbase + tri_indices[2] * stride);
+							triangle[2].setValue((b3Scalar)graphicsbase[0] * meshScaling.getX(), (b3Scalar)graphicsbase[1] * meshScaling.getY(), (b3Scalar)graphicsbase[2] * meshScaling.getZ());
+							callback->internalProcessTriangleIndex(triangle, part, gfxindex);
 						}
 						break;
 					}
-				case PHY_UCHAR:
+					case PHY_UCHAR:
 					{
-						for (gfxindex=0;gfxindex<numtriangles;gfxindex++)
+						for (gfxindex = 0; gfxindex < numtriangles; gfxindex++)
 						{
-							unsigned char* tri_indices= (unsigned char*)(indexbase+gfxindex*indexstride);
-							graphicsbase = (double*)(vertexbase+tri_indices[0]*stride);
-							triangle[0].setValue((b3Scalar)graphicsbase[0]*meshScaling.getX(),(b3Scalar)graphicsbase[1]*meshScaling.getY(),(b3Scalar)graphicsbase[2]*meshScaling.getZ());
-							graphicsbase = (double*)(vertexbase+tri_indices[1]*stride);
-							triangle[1].setValue((b3Scalar)graphicsbase[0]*meshScaling.getX(),(b3Scalar)graphicsbase[1]*meshScaling.getY(),  (b3Scalar)graphicsbase[2]*meshScaling.getZ());
-							graphicsbase = (double*)(vertexbase+tri_indices[2]*stride);
-							triangle[2].setValue((b3Scalar)graphicsbase[0]*meshScaling.getX(),(b3Scalar)graphicsbase[1]*meshScaling.getY(),  (b3Scalar)graphicsbase[2]*meshScaling.getZ());
-							callback->internalProcessTriangleIndex(triangle,part,gfxindex);
+							unsigned char* tri_indices = (unsigned char*)(indexbase + gfxindex * indexstride);
+							graphicsbase = (double*)(vertexbase + tri_indices[0] * stride);
+							triangle[0].setValue((b3Scalar)graphicsbase[0] * meshScaling.getX(), (b3Scalar)graphicsbase[1] * meshScaling.getY(), (b3Scalar)graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (double*)(vertexbase + tri_indices[1] * stride);
+							triangle[1].setValue((b3Scalar)graphicsbase[0] * meshScaling.getX(), (b3Scalar)graphicsbase[1] * meshScaling.getY(), (b3Scalar)graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (double*)(vertexbase + tri_indices[2] * stride);
+							triangle[2].setValue((b3Scalar)graphicsbase[0] * meshScaling.getX(), (b3Scalar)graphicsbase[1] * meshScaling.getY(), (b3Scalar)graphicsbase[2] * meshScaling.getZ());
+							callback->internalProcessTriangleIndex(triangle, part, gfxindex);
 						}
 						break;
 					}
-				default:
-					b3Assert((gfxindextype == PHY_INTEGER) || (gfxindextype == PHY_SHORT));
+					default:
+						b3Assert((gfxindextype == PHY_INTEGER) || (gfxindextype == PHY_SHORT));
 				}
 				break;
 			}
-		default:
-			b3Assert((type == PHY_FLOAT) || (type == PHY_DOUBLE));
+			default:
+				b3Assert((type == PHY_FLOAT) || (type == PHY_DOUBLE));
 		}
 
 		unLockReadOnlyVertexBase(part);
 	}
 }
 
-void	b3StridingMeshInterface::calculateAabbBruteForce(b3Vector3& aabbMin,b3Vector3& aabbMax)
+void b3StridingMeshInterface::calculateAabbBruteForce(b3Vector3& aabbMin, b3Vector3& aabbMax)
 {
-
-	struct	AabbCalculationCallback : public b3InternalTriangleIndexCallback
+	struct AabbCalculationCallback : public b3InternalTriangleIndexCallback
 	{
-		b3Vector3	m_aabbMin;
-		b3Vector3	m_aabbMax;
+		b3Vector3 m_aabbMin;
+		b3Vector3 m_aabbMax;
 
 		AabbCalculationCallback()
 		{
-			m_aabbMin.setValue(b3Scalar(B3_LARGE_FLOAT),b3Scalar(B3_LARGE_FLOAT),b3Scalar(B3_LARGE_FLOAT));
-			m_aabbMax.setValue(b3Scalar(-B3_LARGE_FLOAT),b3Scalar(-B3_LARGE_FLOAT),b3Scalar(-B3_LARGE_FLOAT));
+			m_aabbMin.setValue(b3Scalar(B3_LARGE_FLOAT), b3Scalar(B3_LARGE_FLOAT), b3Scalar(B3_LARGE_FLOAT));
+			m_aabbMax.setValue(b3Scalar(-B3_LARGE_FLOAT), b3Scalar(-B3_LARGE_FLOAT), b3Scalar(-B3_LARGE_FLOAT));
 		}
 
-		virtual void internalProcessTriangleIndex(b3Vector3* triangle,int partId,int  triangleIndex)
+		virtual void internalProcessTriangleIndex(b3Vector3* triangle, int partId, int triangleIndex)
 		{
 			(void)partId;
 			(void)triangleIndex;
@@ -202,13 +197,11 @@ void	b3StridingMeshInterface::calculateAabbBruteForce(b3Vector3& aabbMin,b3Vecto
 	};
 
 	//first calculate the total aabb for all triangles
-	AabbCalculationCallback	aabbCallback;
-	aabbMin.setValue(b3Scalar(-B3_LARGE_FLOAT),b3Scalar(-B3_LARGE_FLOAT),b3Scalar(-B3_LARGE_FLOAT));
-	aabbMax.setValue(b3Scalar(B3_LARGE_FLOAT),b3Scalar(B3_LARGE_FLOAT),b3Scalar(B3_LARGE_FLOAT));
-	InternalProcessAllTriangles(&aabbCallback,aabbMin,aabbMax);
+	AabbCalculationCallback aabbCallback;
+	aabbMin.setValue(b3Scalar(-B3_LARGE_FLOAT), b3Scalar(-B3_LARGE_FLOAT), b3Scalar(-B3_LARGE_FLOAT));
+	aabbMax.setValue(b3Scalar(B3_LARGE_FLOAT), b3Scalar(B3_LARGE_FLOAT), b3Scalar(B3_LARGE_FLOAT));
+	InternalProcessAllTriangles(&aabbCallback, aabbMin, aabbMax);
 
 	aabbMin = aabbCallback.m_aabbMin;
 	aabbMax = aabbCallback.m_aabbMax;
 }
-
-
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3StridingMeshInterface.h b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3StridingMeshInterface.h
index 9513f68f77e..087b30f3e6a 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3StridingMeshInterface.h
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3StridingMeshInterface.h
@@ -20,148 +20,139 @@ subject to the following restrictions:
 #include "b3TriangleCallback.h"
 //#include "b3ConcaveShape.h"
 
-
-enum  	PHY_ScalarType { 
-  PHY_FLOAT, PHY_DOUBLE, PHY_INTEGER, PHY_SHORT, 
-  PHY_FIXEDPOINT88, PHY_UCHAR 
+enum PHY_ScalarType
+{
+	PHY_FLOAT,
+	PHY_DOUBLE,
+	PHY_INTEGER,
+	PHY_SHORT,
+	PHY_FIXEDPOINT88,
+	PHY_UCHAR
 };
 
-
 ///	The b3StridingMeshInterface is the interface class for high performance generic access to triangle meshes, used in combination with b3BvhTriangleMeshShape and some other collision shapes.
 /// Using index striding of 3*sizeof(integer) it can use triangle arrays, using index striding of 1*sizeof(integer) it can handle triangle strips.
 /// It allows for sharing graphics and collision meshes. Also it provides locking/unlocking of graphics meshes that are in gpu memory.
-B3_ATTRIBUTE_ALIGNED16(class ) b3StridingMeshInterface
+B3_ATTRIBUTE_ALIGNED16(class)
+b3StridingMeshInterface
 {
-	protected:
-	
-		b3Vector3 m_scaling;
+protected:
+	b3Vector3 m_scaling;
 
-	public:
-		B3_DECLARE_ALIGNED_ALLOCATOR();
-		
-		b3StridingMeshInterface() :m_scaling(b3MakeVector3(b3Scalar(1.),b3Scalar(1.),b3Scalar(1.)))
-		{
+public:
+	B3_DECLARE_ALIGNED_ALLOCATOR();
 
-		}
+	b3StridingMeshInterface() : m_scaling(b3MakeVector3(b3Scalar(1.), b3Scalar(1.), b3Scalar(1.)))
+	{
+	}
 
-		virtual ~b3StridingMeshInterface();
+	virtual ~b3StridingMeshInterface();
 
+	virtual void InternalProcessAllTriangles(b3InternalTriangleIndexCallback * callback, const b3Vector3& aabbMin, const b3Vector3& aabbMax) const;
 
+	///brute force method to calculate aabb
+	void calculateAabbBruteForce(b3Vector3 & aabbMin, b3Vector3 & aabbMax);
 
-		virtual void	InternalProcessAllTriangles(b3InternalTriangleIndexCallback* callback,const b3Vector3& aabbMin,const b3Vector3& aabbMax) const;
+	/// get read and write access to a subpart of a triangle mesh
+	/// this subpart has a continuous array of vertices and indices
+	/// in this way the mesh can be handled as chunks of memory with striding
+	/// very similar to OpenGL vertexarray support
+	/// make a call to unLockVertexBase when the read and write access is finished
+	virtual void getLockedVertexIndexBase(unsigned char** vertexbase, int& numverts, PHY_ScalarType& type, int& stride, unsigned char** indexbase, int& indexstride, int& numfaces, PHY_ScalarType& indicestype, int subpart = 0) = 0;
 
-		///brute force method to calculate aabb
-		void	calculateAabbBruteForce(b3Vector3& aabbMin,b3Vector3& aabbMax);
+	virtual void getLockedReadOnlyVertexIndexBase(const unsigned char** vertexbase, int& numverts, PHY_ScalarType& type, int& stride, const unsigned char** indexbase, int& indexstride, int& numfaces, PHY_ScalarType& indicestype, int subpart = 0) const = 0;
 
-		/// get read and write access to a subpart of a triangle mesh
-		/// this subpart has a continuous array of vertices and indices
-		/// in this way the mesh can be handled as chunks of memory with striding
-		/// very similar to OpenGL vertexarray support
-		/// make a call to unLockVertexBase when the read and write access is finished	
-		virtual void	getLockedVertexIndexBase(unsigned char **vertexbase, int& numverts,PHY_ScalarType& type, int& stride,unsigned char **indexbase,int & indexstride,int& numfaces,PHY_ScalarType& indicestype,int subpart=0)=0;
-		
-		virtual void	getLockedReadOnlyVertexIndexBase(const unsigned char **vertexbase, int& numverts,PHY_ScalarType& type, int& stride,const unsigned char **indexbase,int & indexstride,int& numfaces,PHY_ScalarType& indicestype,int subpart=0) const=0;
-	
-		/// unLockVertexBase finishes the access to a subpart of the triangle mesh
-		/// make a call to unLockVertexBase when the read and write access (using getLockedVertexIndexBase) is finished
-		virtual void	unLockVertexBase(int subpart)=0;
+	/// unLockVertexBase finishes the access to a subpart of the triangle mesh
+	/// make a call to unLockVertexBase when the read and write access (using getLockedVertexIndexBase) is finished
+	virtual void unLockVertexBase(int subpart) = 0;
 
-		virtual void	unLockReadOnlyVertexBase(int subpart) const=0;
+	virtual void unLockReadOnlyVertexBase(int subpart) const = 0;
 
+	/// getNumSubParts returns the number of seperate subparts
+	/// each subpart has a continuous array of vertices and indices
+	virtual int getNumSubParts() const = 0;
 
-		/// getNumSubParts returns the number of seperate subparts
-		/// each subpart has a continuous array of vertices and indices
-		virtual int		getNumSubParts() const=0;
+	virtual void preallocateVertices(int numverts) = 0;
+	virtual void preallocateIndices(int numindices) = 0;
 
-		virtual void	preallocateVertices(int numverts)=0;
-		virtual void	preallocateIndices(int numindices)=0;
+	virtual bool hasPremadeAabb() const { return false; }
+	virtual void setPremadeAabb(const b3Vector3& aabbMin, const b3Vector3& aabbMax) const
+	{
+		(void)aabbMin;
+		(void)aabbMax;
+	}
+	virtual void getPremadeAabb(b3Vector3 * aabbMin, b3Vector3 * aabbMax) const
+	{
+		(void)aabbMin;
+		(void)aabbMax;
+	}
 
-		virtual bool	hasPremadeAabb() const { return false; }
-		virtual void	setPremadeAabb(const b3Vector3& aabbMin, const b3Vector3& aabbMax ) const
-                {
-                        (void) aabbMin;
-                        (void) aabbMax;
-                }
-		virtual void	getPremadeAabb(b3Vector3* aabbMin, b3Vector3* aabbMax ) const
-        {
-            (void) aabbMin;
-            (void) aabbMax;
-        }
-
-		const b3Vector3&	getScaling() const {
-			return m_scaling;
-		}
-		void	setScaling(const b3Vector3& scaling)
-		{
-			m_scaling = scaling;
-		}
-
-		virtual	int	calculateSerializeBufferSize() const;
-
-		///fills the dataBuffer and returns the struct name (and 0 on failure)
-		//virtual	const char*	serialize(void* dataBuffer, b3Serializer* serializer) const;
+	const b3Vector3& getScaling() const
+	{
+		return m_scaling;
+	}
+	void setScaling(const b3Vector3& scaling)
+	{
+		m_scaling = scaling;
+	}
 
+	virtual int calculateSerializeBufferSize() const;
 
+	///fills the dataBuffer and returns the struct name (and 0 on failure)
+	//virtual	const char*	serialize(void* dataBuffer, b3Serializer* serializer) const;
 };
 
-struct	b3IntIndexData
+struct b3IntIndexData
 {
-	int	m_value;
+	int m_value;
 };
 
-struct	b3ShortIntIndexData
+struct b3ShortIntIndexData
 {
 	short m_value;
 	char m_pad[2];
 };
 
-struct	b3ShortIntIndexTripletData
+struct b3ShortIntIndexTripletData
 {
-	short	m_values[3];
-	char	m_pad[2];
+	short m_values[3];
+	char m_pad[2];
 };
 
-struct	b3CharIndexTripletData
+struct b3CharIndexTripletData
 {
 	unsigned char m_values[3];
-	char	m_pad;
+	char m_pad;
 };
 
-
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
-struct	b3MeshPartData
+struct b3MeshPartData
 {
-	b3Vector3FloatData			*m_vertices3f;
-	b3Vector3DoubleData			*m_vertices3d;
+	b3Vector3FloatData* m_vertices3f;
+	b3Vector3DoubleData* m_vertices3d;
 
-	b3IntIndexData				*m_indices32;
-	b3ShortIntIndexTripletData	*m_3indices16;
-	b3CharIndexTripletData		*m_3indices8;
+	b3IntIndexData* m_indices32;
+	b3ShortIntIndexTripletData* m_3indices16;
+	b3CharIndexTripletData* m_3indices8;
 
-	b3ShortIntIndexData			*m_indices16;//backwards compatibility
+	b3ShortIntIndexData* m_indices16;  //backwards compatibility
 
-	int                     m_numTriangles;//length of m_indices = m_numTriangles
-	int                     m_numVertices;
+	int m_numTriangles;  //length of m_indices = m_numTriangles
+	int m_numVertices;
 };
 
-
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
-struct	b3StridingMeshInterfaceData
+struct b3StridingMeshInterfaceData
 {
-	b3MeshPartData	*m_meshPartsPtr;
-	b3Vector3FloatData	m_scaling;
-	int	m_numMeshParts;
+	b3MeshPartData* m_meshPartsPtr;
+	b3Vector3FloatData m_scaling;
+	int m_numMeshParts;
 	char m_padding[4];
 };
 
-
-
-
-B3_FORCE_INLINE	int	b3StridingMeshInterface::calculateSerializeBufferSize() const
+B3_FORCE_INLINE int b3StridingMeshInterface::calculateSerializeBufferSize() const
 {
 	return sizeof(b3StridingMeshInterfaceData);
 }
 
-
-
-#endif //B3_STRIDING_MESHINTERFACE_H
+#endif  //B3_STRIDING_MESHINTERFACE_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3SupportMappings.h b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3SupportMappings.h
index d073ee57c37..9ca1e229496 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3SupportMappings.h
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3SupportMappings.h
@@ -6,33 +6,29 @@
 #include "Bullet3Common/b3AlignedObjectArray.h"
 #include "b3VectorFloat4.h"
 
-
 struct b3GjkPairDetector;
 
-
-
-inline b3Vector3 localGetSupportVertexWithMargin(const float4& supportVec,const struct b3ConvexPolyhedronData* hull, 
-	const b3AlignedObjectArray<b3Vector3>& verticesA, b3Scalar margin)
+inline b3Vector3 localGetSupportVertexWithMargin(const float4& supportVec, const struct b3ConvexPolyhedronData* hull,
+												 const b3AlignedObjectArray<b3Vector3>& verticesA, b3Scalar margin)
 {
-	b3Vector3 supVec = b3MakeVector3(b3Scalar(0.),b3Scalar(0.),b3Scalar(0.));
+	b3Vector3 supVec = b3MakeVector3(b3Scalar(0.), b3Scalar(0.), b3Scalar(0.));
 	b3Scalar maxDot = b3Scalar(-B3_LARGE_FLOAT);
 
-    // Here we take advantage of dot(a, b*c) = dot(a*b, c).  Note: This is true mathematically, but not numerically. 
-    if( 0 < hull->m_numVertices )
-    {
-        const b3Vector3 scaled = supportVec;
-		int index = (int) scaled.maxDot( &verticesA[hull->m_vertexOffset], hull->m_numVertices, maxDot); 
-        return verticesA[hull->m_vertexOffset+index];
-    }
-
-    return supVec;
+	// Here we take advantage of dot(a, b*c) = dot(a*b, c).  Note: This is true mathematically, but not numerically.
+	if (0 < hull->m_numVertices)
+	{
+		const b3Vector3 scaled = supportVec;
+		int index = (int)scaled.maxDot(&verticesA[hull->m_vertexOffset], hull->m_numVertices, maxDot);
+		return verticesA[hull->m_vertexOffset + index];
+	}
 
+	return supVec;
 }
 
-inline b3Vector3 localGetSupportVertexWithoutMargin(const float4& supportVec,const struct b3ConvexPolyhedronData* hull, 
-	const b3AlignedObjectArray<b3Vector3>& verticesA)
+inline b3Vector3 localGetSupportVertexWithoutMargin(const float4& supportVec, const struct b3ConvexPolyhedronData* hull,
+													const b3AlignedObjectArray<b3Vector3>& verticesA)
 {
-	return localGetSupportVertexWithMargin(supportVec,hull,verticesA,0.f);
+	return localGetSupportVertexWithMargin(supportVec, hull, verticesA, 0.f);
 }
 
-#endif //B3_SUPPORT_MAPPINGS_H
+#endif  //B3_SUPPORT_MAPPINGS_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3TriangleCallback.cpp b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3TriangleCallback.cpp
index 9066451884e..3908c6de896 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3TriangleCallback.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3TriangleCallback.cpp
@@ -17,12 +17,8 @@ subject to the following restrictions:
 
 b3TriangleCallback::~b3TriangleCallback()
 {
-	
 }
 
-
 b3InternalTriangleIndexCallback::~b3InternalTriangleIndexCallback()
 {
-
 }
-
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3TriangleCallback.h b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3TriangleCallback.h
index 3059fa4f212..a0fd3e7ac71 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3TriangleCallback.h
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3TriangleCallback.h
@@ -18,13 +18,11 @@ subject to the following restrictions:
 
 #include "Bullet3Common/b3Vector3.h"
 
-
 ///The b3TriangleCallback provides a callback for each overlapping triangle when calling processAllTriangles.
 ///This callback is called by processAllTriangles for all b3ConcaveShape derived class, such as  b3BvhTriangleMeshShape, b3StaticPlaneShape and b3HeightfieldTerrainShape.
 class b3TriangleCallback
 {
 public:
-
 	virtual ~b3TriangleCallback();
 	virtual void processTriangle(b3Vector3* triangle, int partId, int triangleIndex) = 0;
 };
@@ -32,11 +30,8 @@ public:
 class b3InternalTriangleIndexCallback
 {
 public:
-
 	virtual ~b3InternalTriangleIndexCallback();
-	virtual void internalProcessTriangleIndex(b3Vector3* triangle,int partId,int  triangleIndex) = 0;
+	virtual void internalProcessTriangleIndex(b3Vector3* triangle, int partId, int triangleIndex) = 0;
 };
 
-
-
-#endif //B3_TRIANGLE_CALLBACK_H
+#endif  //B3_TRIANGLE_CALLBACK_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3TriangleIndexVertexArray.cpp b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3TriangleIndexVertexArray.cpp
index a0f59babbee..73faadbdd0f 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3TriangleIndexVertexArray.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3TriangleIndexVertexArray.cpp
@@ -15,81 +15,76 @@ subject to the following restrictions:
 
 #include "b3TriangleIndexVertexArray.h"
 
-b3TriangleIndexVertexArray::b3TriangleIndexVertexArray(int numTriangles,int* triangleIndexBase,int triangleIndexStride,int numVertices,b3Scalar* vertexBase,int vertexStride)
-: m_hasAabb(0)
+b3TriangleIndexVertexArray::b3TriangleIndexVertexArray(int numTriangles, int* triangleIndexBase, int triangleIndexStride, int numVertices, b3Scalar* vertexBase, int vertexStride)
+	: m_hasAabb(0)
 {
 	b3IndexedMesh mesh;
 
 	mesh.m_numTriangles = numTriangles;
-	mesh.m_triangleIndexBase = (const unsigned char *)triangleIndexBase;
+	mesh.m_triangleIndexBase = (const unsigned char*)triangleIndexBase;
 	mesh.m_triangleIndexStride = triangleIndexStride;
 	mesh.m_numVertices = numVertices;
-	mesh.m_vertexBase = (const unsigned char *)vertexBase;
+	mesh.m_vertexBase = (const unsigned char*)vertexBase;
 	mesh.m_vertexStride = vertexStride;
 
 	addIndexedMesh(mesh);
-
 }
 
 b3TriangleIndexVertexArray::~b3TriangleIndexVertexArray()
 {
-
 }
 
-void	b3TriangleIndexVertexArray::getLockedVertexIndexBase(unsigned char **vertexbase, int& numverts,PHY_ScalarType& type, int& vertexStride,unsigned char **indexbase,int & indexstride,int& numfaces,PHY_ScalarType& indicestype,int subpart)
+void b3TriangleIndexVertexArray::getLockedVertexIndexBase(unsigned char** vertexbase, int& numverts, PHY_ScalarType& type, int& vertexStride, unsigned char** indexbase, int& indexstride, int& numfaces, PHY_ScalarType& indicestype, int subpart)
 {
-	b3Assert(subpart< getNumSubParts() );
+	b3Assert(subpart < getNumSubParts());
 
 	b3IndexedMesh& mesh = m_indexedMeshes[subpart];
 
 	numverts = mesh.m_numVertices;
-	(*vertexbase) = (unsigned char *) mesh.m_vertexBase;
+	(*vertexbase) = (unsigned char*)mesh.m_vertexBase;
 
-   type = mesh.m_vertexType;
+	type = mesh.m_vertexType;
 
 	vertexStride = mesh.m_vertexStride;
 
 	numfaces = mesh.m_numTriangles;
 
-	(*indexbase) = (unsigned char *)mesh.m_triangleIndexBase;
+	(*indexbase) = (unsigned char*)mesh.m_triangleIndexBase;
 	indexstride = mesh.m_triangleIndexStride;
 	indicestype = mesh.m_indexType;
 }
 
-void	b3TriangleIndexVertexArray::getLockedReadOnlyVertexIndexBase(const unsigned char **vertexbase, int& numverts,PHY_ScalarType& type, int& vertexStride,const unsigned char **indexbase,int & indexstride,int& numfaces,PHY_ScalarType& indicestype,int subpart) const
+void b3TriangleIndexVertexArray::getLockedReadOnlyVertexIndexBase(const unsigned char** vertexbase, int& numverts, PHY_ScalarType& type, int& vertexStride, const unsigned char** indexbase, int& indexstride, int& numfaces, PHY_ScalarType& indicestype, int subpart) const
 {
 	const b3IndexedMesh& mesh = m_indexedMeshes[subpart];
 
 	numverts = mesh.m_numVertices;
-	(*vertexbase) = (const unsigned char *)mesh.m_vertexBase;
+	(*vertexbase) = (const unsigned char*)mesh.m_vertexBase;
+
+	type = mesh.m_vertexType;
 
-   type = mesh.m_vertexType;
-   
 	vertexStride = mesh.m_vertexStride;
 
 	numfaces = mesh.m_numTriangles;
-	(*indexbase) = (const unsigned char *)mesh.m_triangleIndexBase;
+	(*indexbase) = (const unsigned char*)mesh.m_triangleIndexBase;
 	indexstride = mesh.m_triangleIndexStride;
 	indicestype = mesh.m_indexType;
 }
 
-bool	b3TriangleIndexVertexArray::hasPremadeAabb() const
+bool b3TriangleIndexVertexArray::hasPremadeAabb() const
 {
 	return (m_hasAabb == 1);
 }
 
-
-void	b3TriangleIndexVertexArray::setPremadeAabb(const b3Vector3& aabbMin, const b3Vector3& aabbMax ) const
+void b3TriangleIndexVertexArray::setPremadeAabb(const b3Vector3& aabbMin, const b3Vector3& aabbMax) const
 {
 	m_aabbMin = aabbMin;
 	m_aabbMax = aabbMax;
-	m_hasAabb = 1; // this is intentionally an int see notes in header
+	m_hasAabb = 1;  // this is intentionally an int see notes in header
 }
 
-void	b3TriangleIndexVertexArray::getPremadeAabb(b3Vector3* aabbMin, b3Vector3* aabbMax ) const
+void b3TriangleIndexVertexArray::getPremadeAabb(b3Vector3* aabbMin, b3Vector3* aabbMax) const
 {
 	*aabbMin = m_aabbMin;
 	*aabbMax = m_aabbMax;
 }
-
-
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3TriangleIndexVertexArray.h b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3TriangleIndexVertexArray.h
index d26b2893bc5..b6ceb8df10f 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3TriangleIndexVertexArray.h
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3TriangleIndexVertexArray.h
@@ -20,62 +20,59 @@ subject to the following restrictions:
 #include "Bullet3Common/b3AlignedObjectArray.h"
 #include "Bullet3Common/b3Scalar.h"
 
-
 ///The b3IndexedMesh indexes a single vertex and index array. Multiple b3IndexedMesh objects can be passed into a b3TriangleIndexVertexArray using addIndexedMesh.
 ///Instead of the number of indices, we pass the number of triangles.
-B3_ATTRIBUTE_ALIGNED16( struct)	b3IndexedMesh
+B3_ATTRIBUTE_ALIGNED16(struct)
+b3IndexedMesh
 {
 	B3_DECLARE_ALIGNED_ALLOCATOR();
 
-   int                     m_numTriangles;
-   const unsigned char *   m_triangleIndexBase;
-   // Size in byte of the indices for one triangle (3*sizeof(index_type) if the indices are tightly packed)
-   int                     m_triangleIndexStride;
-   int                     m_numVertices;
-   const unsigned char *   m_vertexBase;
-   // Size of a vertex, in bytes
-   int                     m_vertexStride;
+	int m_numTriangles;
+	const unsigned char* m_triangleIndexBase;
+	// Size in byte of the indices for one triangle (3*sizeof(index_type) if the indices are tightly packed)
+	int m_triangleIndexStride;
+	int m_numVertices;
+	const unsigned char* m_vertexBase;
+	// Size of a vertex, in bytes
+	int m_vertexStride;
 
-   // The index type is set when adding an indexed mesh to the
-   // b3TriangleIndexVertexArray, do not set it manually
-   PHY_ScalarType m_indexType;
+	// The index type is set when adding an indexed mesh to the
+	// b3TriangleIndexVertexArray, do not set it manually
+	PHY_ScalarType m_indexType;
 
-   // The vertex type has a default type similar to Bullet's precision mode (float or double)
-   // but can be set manually if you for example run Bullet with double precision but have
-   // mesh data in single precision..
-   PHY_ScalarType m_vertexType;
+	// The vertex type has a default type similar to Bullet's precision mode (float or double)
+	// but can be set manually if you for example run Bullet with double precision but have
+	// mesh data in single precision..
+	PHY_ScalarType m_vertexType;
 
-
-   b3IndexedMesh()
-	   :m_indexType(PHY_INTEGER),
+	b3IndexedMesh()
+		: m_indexType(PHY_INTEGER),
 #ifdef B3_USE_DOUBLE_PRECISION
-      m_vertexType(PHY_DOUBLE)
-#else // B3_USE_DOUBLE_PRECISION
-      m_vertexType(PHY_FLOAT)
-#endif // B3_USE_DOUBLE_PRECISION
-      {
-      }
-}
-;
+		  m_vertexType(PHY_DOUBLE)
+#else   // B3_USE_DOUBLE_PRECISION
+		  m_vertexType(PHY_FLOAT)
+#endif  // B3_USE_DOUBLE_PRECISION
+	{
+	}
+};
 
-
-typedef b3AlignedObjectArray<b3IndexedMesh>	IndexedMeshArray;
+typedef b3AlignedObjectArray<b3IndexedMesh> IndexedMeshArray;
 
 ///The b3TriangleIndexVertexArray allows to access multiple triangle meshes, by indexing into existing triangle/index arrays.
 ///Additional meshes can be added using addIndexedMesh
 ///No duplcate is made of the vertex/index data, it only indexes into external vertex/index arrays.
 ///So keep those arrays around during the lifetime of this b3TriangleIndexVertexArray.
-B3_ATTRIBUTE_ALIGNED16( class) b3TriangleIndexVertexArray : public b3StridingMeshInterface
+B3_ATTRIBUTE_ALIGNED16(class)
+b3TriangleIndexVertexArray : public b3StridingMeshInterface
 {
 protected:
-	IndexedMeshArray	m_indexedMeshes;
+	IndexedMeshArray m_indexedMeshes;
 	int m_pad[2];
-	mutable int m_hasAabb; // using int instead of bool to maintain alignment
+	mutable int m_hasAabb;  // using int instead of bool to maintain alignment
 	mutable b3Vector3 m_aabbMin;
 	mutable b3Vector3 m_aabbMax;
 
 public:
-
 	B3_DECLARE_ALIGNED_ALLOCATOR();
 
 	b3TriangleIndexVertexArray() : m_hasAabb(0)
@@ -85,49 +82,47 @@ public:
 	virtual ~b3TriangleIndexVertexArray();
 
 	//just to be backwards compatible
-	b3TriangleIndexVertexArray(int numTriangles,int* triangleIndexBase,int triangleIndexStride,int numVertices,b3Scalar* vertexBase,int vertexStride);
-	
-	void	addIndexedMesh(const b3IndexedMesh& mesh, PHY_ScalarType indexType = PHY_INTEGER)
+	b3TriangleIndexVertexArray(int numTriangles, int* triangleIndexBase, int triangleIndexStride, int numVertices, b3Scalar* vertexBase, int vertexStride);
+
+	void addIndexedMesh(const b3IndexedMesh& mesh, PHY_ScalarType indexType = PHY_INTEGER)
 	{
 		m_indexedMeshes.push_back(mesh);
-		m_indexedMeshes[m_indexedMeshes.size()-1].m_indexType = indexType;
+		m_indexedMeshes[m_indexedMeshes.size() - 1].m_indexType = indexType;
 	}
-	
-	
-	virtual void	getLockedVertexIndexBase(unsigned char **vertexbase, int& numverts,PHY_ScalarType& type, int& vertexStride,unsigned char **indexbase,int & indexstride,int& numfaces,PHY_ScalarType& indicestype,int subpart=0);
 
-	virtual void	getLockedReadOnlyVertexIndexBase(const unsigned char **vertexbase, int& numverts,PHY_ScalarType& type, int& vertexStride,const unsigned char **indexbase,int & indexstride,int& numfaces,PHY_ScalarType& indicestype,int subpart=0) const;
+	virtual void getLockedVertexIndexBase(unsigned char** vertexbase, int& numverts, PHY_ScalarType& type, int& vertexStride, unsigned char** indexbase, int& indexstride, int& numfaces, PHY_ScalarType& indicestype, int subpart = 0);
+
+	virtual void getLockedReadOnlyVertexIndexBase(const unsigned char** vertexbase, int& numverts, PHY_ScalarType& type, int& vertexStride, const unsigned char** indexbase, int& indexstride, int& numfaces, PHY_ScalarType& indicestype, int subpart = 0) const;
 
 	/// unLockVertexBase finishes the access to a subpart of the triangle mesh
 	/// make a call to unLockVertexBase when the read and write access (using getLockedVertexIndexBase) is finished
-	virtual void	unLockVertexBase(int subpart) {(void)subpart;}
+	virtual void unLockVertexBase(int subpart) { (void)subpart; }
 
-	virtual void	unLockReadOnlyVertexBase(int subpart) const {(void)subpart;}
+	virtual void unLockReadOnlyVertexBase(int subpart) const { (void)subpart; }
 
 	/// getNumSubParts returns the number of seperate subparts
 	/// each subpart has a continuous array of vertices and indices
-	virtual int		getNumSubParts() const { 
+	virtual int getNumSubParts() const
+	{
 		return (int)m_indexedMeshes.size();
 	}
 
-	IndexedMeshArray&	getIndexedMeshArray()
+	IndexedMeshArray& getIndexedMeshArray()
 	{
 		return m_indexedMeshes;
 	}
 
-	const IndexedMeshArray&	getIndexedMeshArray() const
+	const IndexedMeshArray& getIndexedMeshArray() const
 	{
 		return m_indexedMeshes;
 	}
 
-	virtual void	preallocateVertices(int numverts){(void) numverts;}
-	virtual void	preallocateIndices(int numindices){(void) numindices;}
+	virtual void preallocateVertices(int numverts) { (void)numverts; }
+	virtual void preallocateIndices(int numindices) { (void)numindices; }
 
-	virtual bool	hasPremadeAabb() const;
-	virtual void	setPremadeAabb(const b3Vector3& aabbMin, const b3Vector3& aabbMax ) const;
-	virtual void	getPremadeAabb(b3Vector3* aabbMin, b3Vector3* aabbMax ) const;
+	virtual bool hasPremadeAabb() const;
+	virtual void setPremadeAabb(const b3Vector3& aabbMin, const b3Vector3& aabbMax) const;
+	virtual void getPremadeAabb(b3Vector3 * aabbMin, b3Vector3 * aabbMax) const;
+};
 
-}
-;
-
-#endif //B3_TRIANGLE_INDEX_VERTEX_ARRAY_H
+#endif  //B3_TRIANGLE_INDEX_VERTEX_ARRAY_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3VectorFloat4.h b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3VectorFloat4.h
index f6f65f7719d..5cc4b5a6262 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3VectorFloat4.h
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3VectorFloat4.h
@@ -7,5 +7,4 @@
 #define float4 b3Vector3
 //#define make_float4(x,y,z,w) b3Vector4(x,y,z,w)
 
-
-#endif //B3_VECTOR_FLOAT4_H
+#endif  //B3_VECTOR_FLOAT4_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3VoronoiSimplexSolver.cpp b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3VoronoiSimplexSolver.cpp
index cf3d5ef49d4..dae61d45810 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3VoronoiSimplexSolver.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3VoronoiSimplexSolver.cpp
@@ -23,26 +23,24 @@ subject to the following restrictions:
 		
 */
 
-
 #include "b3VoronoiSimplexSolver.h"
 
-#define VERTA  0
-#define VERTB  1
-#define VERTC  2
-#define VERTD  3
+#define VERTA 0
+#define VERTB 1
+#define VERTC 2
+#define VERTD 3
 
 #define B3_CATCH_DEGENERATE_TETRAHEDRON 1
-void	b3VoronoiSimplexSolver::removeVertex(int index)
+void b3VoronoiSimplexSolver::removeVertex(int index)
 {
-	
-	b3Assert(m_numVertices>0);
+	b3Assert(m_numVertices > 0);
 	m_numVertices--;
 	m_simplexVectorW[index] = m_simplexVectorW[m_numVertices];
 	m_simplexPointsP[index] = m_simplexPointsP[m_numVertices];
 	m_simplexPointsQ[index] = m_simplexPointsQ[m_numVertices];
 }
 
-void	b3VoronoiSimplexSolver::reduceVertices (const b3UsageBitfield& usedVerts)
+void b3VoronoiSimplexSolver::reduceVertices(const b3UsageBitfield& usedVerts)
 {
 	if ((numVertices() >= 4) && (!usedVerts.usedVertexD))
 		removeVertex(3);
@@ -52,29 +50,22 @@ void	b3VoronoiSimplexSolver::reduceVertices (const b3UsageBitfield& usedVerts)
 
 	if ((numVertices() >= 2) && (!usedVerts.usedVertexB))
 		removeVertex(1);
-	
+
 	if ((numVertices() >= 1) && (!usedVerts.usedVertexA))
 		removeVertex(0);
-
 }
 
-
-
-
-
 //clear the simplex, remove all the vertices
 void b3VoronoiSimplexSolver::reset()
 {
 	m_cachedValidClosest = false;
 	m_numVertices = 0;
 	m_needsUpdate = true;
-	m_lastW = b3MakeVector3(b3Scalar(B3_LARGE_FLOAT),b3Scalar(B3_LARGE_FLOAT),b3Scalar(B3_LARGE_FLOAT));
+	m_lastW = b3MakeVector3(b3Scalar(B3_LARGE_FLOAT), b3Scalar(B3_LARGE_FLOAT), b3Scalar(B3_LARGE_FLOAT));
 	m_cachedBC.reset();
 }
 
-
-
-	//add a vertex
+//add a vertex
 void b3VoronoiSimplexSolver::addVertex(const b3Vector3& w, const b3Vector3& p, const b3Vector3& q)
 {
 	m_lastW = w;
@@ -87,9 +78,8 @@ void b3VoronoiSimplexSolver::addVertex(const b3Vector3& w, const b3Vector3& p, c
 	m_numVertices++;
 }
 
-bool	b3VoronoiSimplexSolver::updateClosestVectorAndPoints()
+bool b3VoronoiSimplexSolver::updateClosestVectorAndPoints()
 {
-	
 	if (m_needsUpdate)
 	{
 		m_cachedBC.reset();
@@ -98,127 +88,131 @@ bool	b3VoronoiSimplexSolver::updateClosestVectorAndPoints()
 
 		switch (numVertices())
 		{
-		case 0:
+			case 0:
 				m_cachedValidClosest = false;
 				break;
-		case 1:
+			case 1:
 			{
 				m_cachedP1 = m_simplexPointsP[0];
 				m_cachedP2 = m_simplexPointsQ[0];
-				m_cachedV = m_cachedP1-m_cachedP2; //== m_simplexVectorW[0]
+				m_cachedV = m_cachedP1 - m_cachedP2;  //== m_simplexVectorW[0]
 				m_cachedBC.reset();
-				m_cachedBC.setBarycentricCoordinates(b3Scalar(1.),b3Scalar(0.),b3Scalar(0.),b3Scalar(0.));
+				m_cachedBC.setBarycentricCoordinates(b3Scalar(1.), b3Scalar(0.), b3Scalar(0.), b3Scalar(0.));
 				m_cachedValidClosest = m_cachedBC.isValid();
 				break;
 			};
-		case 2:
+			case 2:
 			{
-			//closest point origin from line segment
-					const b3Vector3& from = m_simplexVectorW[0];
-					const b3Vector3& to = m_simplexVectorW[1];
-					b3Vector3 nearest;
+				//closest point origin from line segment
+				const b3Vector3& from = m_simplexVectorW[0];
+				const b3Vector3& to = m_simplexVectorW[1];
+				b3Vector3 nearest;
 
-					b3Vector3 p =b3MakeVector3(b3Scalar(0.),b3Scalar(0.),b3Scalar(0.));
-					b3Vector3 diff = p - from;
-					b3Vector3 v = to - from;
-					b3Scalar t = v.dot(diff);
-					
-					if (t > 0) {
-						b3Scalar dotVV = v.dot(v);
-						if (t < dotVV) {
-							t /= dotVV;
-							diff -= t*v;
-							m_cachedBC.m_usedVertices.usedVertexA = true;
-							m_cachedBC.m_usedVertices.usedVertexB = true;
-						} else {
-							t = 1;
-							diff -= v;
-							//reduce to 1 point
-							m_cachedBC.m_usedVertices.usedVertexB = true;
-						}
-					} else
+				b3Vector3 p = b3MakeVector3(b3Scalar(0.), b3Scalar(0.), b3Scalar(0.));
+				b3Vector3 diff = p - from;
+				b3Vector3 v = to - from;
+				b3Scalar t = v.dot(diff);
+
+				if (t > 0)
+				{
+					b3Scalar dotVV = v.dot(v);
+					if (t < dotVV)
 					{
-						t = 0;
-						//reduce to 1 point
+						t /= dotVV;
+						diff -= t * v;
 						m_cachedBC.m_usedVertices.usedVertexA = true;
+						m_cachedBC.m_usedVertices.usedVertexB = true;
 					}
-					m_cachedBC.setBarycentricCoordinates(1-t,t);
-					nearest = from + t*v;
+					else
+					{
+						t = 1;
+						diff -= v;
+						//reduce to 1 point
+						m_cachedBC.m_usedVertices.usedVertexB = true;
+					}
+				}
+				else
+				{
+					t = 0;
+					//reduce to 1 point
+					m_cachedBC.m_usedVertices.usedVertexA = true;
+				}
+				m_cachedBC.setBarycentricCoordinates(1 - t, t);
+				nearest = from + t * v;
 
-					m_cachedP1 = m_simplexPointsP[0] + t * (m_simplexPointsP[1] - m_simplexPointsP[0]);
-					m_cachedP2 = m_simplexPointsQ[0] + t * (m_simplexPointsQ[1] - m_simplexPointsQ[0]);
-					m_cachedV = m_cachedP1 - m_cachedP2;
-					
-					reduceVertices(m_cachedBC.m_usedVertices);
+				m_cachedP1 = m_simplexPointsP[0] + t * (m_simplexPointsP[1] - m_simplexPointsP[0]);
+				m_cachedP2 = m_simplexPointsQ[0] + t * (m_simplexPointsQ[1] - m_simplexPointsQ[0]);
+				m_cachedV = m_cachedP1 - m_cachedP2;
 
-					m_cachedValidClosest = m_cachedBC.isValid();
-					break;
+				reduceVertices(m_cachedBC.m_usedVertices);
+
+				m_cachedValidClosest = m_cachedBC.isValid();
+				break;
 			}
-		case 3: 
-			{ 
-				//closest point origin from triangle 
-				b3Vector3 p =b3MakeVector3(b3Scalar(0.),b3Scalar(0.),b3Scalar(0.)); 
-
-				const b3Vector3& a = m_simplexVectorW[0]; 
-				const b3Vector3& b = m_simplexVectorW[1]; 
-				const b3Vector3& c = m_simplexVectorW[2]; 
-
-				closestPtPointTriangle(p,a,b,c,m_cachedBC); 
-				m_cachedP1 = m_simplexPointsP[0] * m_cachedBC.m_barycentricCoords[0] + 
-				m_simplexPointsP[1] * m_cachedBC.m_barycentricCoords[1] + 
-				m_simplexPointsP[2] * m_cachedBC.m_barycentricCoords[2]; 
-
-				m_cachedP2 = m_simplexPointsQ[0] * m_cachedBC.m_barycentricCoords[0] + 
-				m_simplexPointsQ[1] * m_cachedBC.m_barycentricCoords[1] + 
-				m_simplexPointsQ[2] * m_cachedBC.m_barycentricCoords[2]; 
-
-				m_cachedV = m_cachedP1-m_cachedP2; 
-
-				reduceVertices (m_cachedBC.m_usedVertices); 
-				m_cachedValidClosest = m_cachedBC.isValid(); 
-
-				break; 
-			}
-		case 4:
+			case 3:
 			{
+				//closest point origin from triangle
+				b3Vector3 p = b3MakeVector3(b3Scalar(0.), b3Scalar(0.), b3Scalar(0.));
+
+				const b3Vector3& a = m_simplexVectorW[0];
+				const b3Vector3& b = m_simplexVectorW[1];
+				const b3Vector3& c = m_simplexVectorW[2];
+
+				closestPtPointTriangle(p, a, b, c, m_cachedBC);
+				m_cachedP1 = m_simplexPointsP[0] * m_cachedBC.m_barycentricCoords[0] +
+							 m_simplexPointsP[1] * m_cachedBC.m_barycentricCoords[1] +
+							 m_simplexPointsP[2] * m_cachedBC.m_barycentricCoords[2];
+
+				m_cachedP2 = m_simplexPointsQ[0] * m_cachedBC.m_barycentricCoords[0] +
+							 m_simplexPointsQ[1] * m_cachedBC.m_barycentricCoords[1] +
+							 m_simplexPointsQ[2] * m_cachedBC.m_barycentricCoords[2];
+
+				m_cachedV = m_cachedP1 - m_cachedP2;
+
+				reduceVertices(m_cachedBC.m_usedVertices);
+				m_cachedValidClosest = m_cachedBC.isValid();
+
+				break;
+			}
+			case 4:
+			{
+				b3Vector3 p = b3MakeVector3(b3Scalar(0.), b3Scalar(0.), b3Scalar(0.));
 
-				
-				b3Vector3 p =b3MakeVector3(b3Scalar(0.),b3Scalar(0.),b3Scalar(0.));
-				
 				const b3Vector3& a = m_simplexVectorW[0];
 				const b3Vector3& b = m_simplexVectorW[1];
 				const b3Vector3& c = m_simplexVectorW[2];
 				const b3Vector3& d = m_simplexVectorW[3];
 
-				bool hasSeperation = closestPtPointTetrahedron(p,a,b,c,d,m_cachedBC);
+				bool hasSeperation = closestPtPointTetrahedron(p, a, b, c, d, m_cachedBC);
 
 				if (hasSeperation)
 				{
-
 					m_cachedP1 = m_simplexPointsP[0] * m_cachedBC.m_barycentricCoords[0] +
-						m_simplexPointsP[1] * m_cachedBC.m_barycentricCoords[1] +
-						m_simplexPointsP[2] * m_cachedBC.m_barycentricCoords[2] +
-						m_simplexPointsP[3] * m_cachedBC.m_barycentricCoords[3];
+								 m_simplexPointsP[1] * m_cachedBC.m_barycentricCoords[1] +
+								 m_simplexPointsP[2] * m_cachedBC.m_barycentricCoords[2] +
+								 m_simplexPointsP[3] * m_cachedBC.m_barycentricCoords[3];
 
 					m_cachedP2 = m_simplexPointsQ[0] * m_cachedBC.m_barycentricCoords[0] +
-						m_simplexPointsQ[1] * m_cachedBC.m_barycentricCoords[1] +
-						m_simplexPointsQ[2] * m_cachedBC.m_barycentricCoords[2] +
-						m_simplexPointsQ[3] * m_cachedBC.m_barycentricCoords[3];
+								 m_simplexPointsQ[1] * m_cachedBC.m_barycentricCoords[1] +
+								 m_simplexPointsQ[2] * m_cachedBC.m_barycentricCoords[2] +
+								 m_simplexPointsQ[3] * m_cachedBC.m_barycentricCoords[3];
 
-					m_cachedV = m_cachedP1-m_cachedP2;
-					reduceVertices (m_cachedBC.m_usedVertices);
-				} else
+					m_cachedV = m_cachedP1 - m_cachedP2;
+					reduceVertices(m_cachedBC.m_usedVertices);
+				}
+				else
 				{
-//					printf("sub distance got penetration\n");
+					//					printf("sub distance got penetration\n");
 
 					if (m_cachedBC.m_degenerate)
 					{
 						m_cachedValidClosest = false;
-					} else
+					}
+					else
 					{
 						m_cachedValidClosest = true;
 						//degenerate case == false, penetration = true + zero
-						m_cachedV.setValue(b3Scalar(0.),b3Scalar(0.),b3Scalar(0.));
+						m_cachedV.setValue(b3Scalar(0.), b3Scalar(0.), b3Scalar(0.));
 					}
 					break;
 				}
@@ -228,7 +222,7 @@ bool	b3VoronoiSimplexSolver::updateClosestVectorAndPoints()
 				//closest point origin from tetrahedron
 				break;
 			}
-		default:
+			default:
 			{
 				m_cachedValidClosest = false;
 			}
@@ -236,7 +230,6 @@ bool	b3VoronoiSimplexSolver::updateClosestVectorAndPoints()
 	}
 
 	return m_cachedValidClosest;
-
 }
 
 //return/calculate the closest vertex
@@ -247,13 +240,11 @@ bool b3VoronoiSimplexSolver::closest(b3Vector3& v)
 	return succes;
 }
 
-
-
 b3Scalar b3VoronoiSimplexSolver::maxVertex()
 {
 	int i, numverts = numVertices();
 	b3Scalar maxV = b3Scalar(0.);
-	for (i=0;i<numverts;i++)
+	for (i = 0; i < numverts; i++)
 	{
 		b3Scalar curLen2 = m_simplexVectorW[i].length2();
 		if (maxV < curLen2)
@@ -262,13 +253,11 @@ b3Scalar b3VoronoiSimplexSolver::maxVertex()
 	return maxV;
 }
 
-
-
-	//return the current simplex
-int b3VoronoiSimplexSolver::getSimplex(b3Vector3 *pBuf, b3Vector3 *qBuf, b3Vector3 *yBuf) const
+//return the current simplex
+int b3VoronoiSimplexSolver::getSimplex(b3Vector3* pBuf, b3Vector3* qBuf, b3Vector3* yBuf) const
 {
 	int i;
-	for (i=0;i<numVertices();i++)
+	for (i = 0; i < numVertices(); i++)
 	{
 		yBuf[i] = m_simplexVectorW[i];
 		pBuf[i] = m_simplexPointsP[i];
@@ -277,20 +266,17 @@ int b3VoronoiSimplexSolver::getSimplex(b3Vector3 *pBuf, b3Vector3 *qBuf, b3Vecto
 	return numVertices();
 }
 
-
-
-
 bool b3VoronoiSimplexSolver::inSimplex(const b3Vector3& w)
 {
 	bool found = false;
 	int i, numverts = numVertices();
 	//b3Scalar maxV = b3Scalar(0.);
-	
+
 	//w is in the current (reduced) simplex
-	for (i=0;i<numverts;i++)
+	for (i = 0; i < numverts; i++)
 	{
 #ifdef BT_USE_EQUAL_VERTEX_THRESHOLD
-		if ( m_simplexVectorW[i].distance2(w) <= m_equalVertexThreshold)
+		if (m_simplexVectorW[i].distance2(w) <= m_equalVertexThreshold)
 #else
 		if (m_simplexVectorW[i] == w)
 #endif
@@ -300,199 +286,190 @@ bool b3VoronoiSimplexSolver::inSimplex(const b3Vector3& w)
 	//check in case lastW is already removed
 	if (w == m_lastW)
 		return true;
-    	
+
 	return found;
 }
 
-void b3VoronoiSimplexSolver::backup_closest(b3Vector3& v) 
+void b3VoronoiSimplexSolver::backup_closest(b3Vector3& v)
 {
 	v = m_cachedV;
 }
 
-
-bool b3VoronoiSimplexSolver::emptySimplex() const 
+bool b3VoronoiSimplexSolver::emptySimplex() const
 {
 	return (numVertices() == 0);
-
 }
 
-void b3VoronoiSimplexSolver::compute_points(b3Vector3& p1, b3Vector3& p2) 
+void b3VoronoiSimplexSolver::compute_points(b3Vector3& p1, b3Vector3& p2)
 {
 	updateClosestVectorAndPoints();
 	p1 = m_cachedP1;
 	p2 = m_cachedP2;
-
 }
 
-
-
-
-bool	b3VoronoiSimplexSolver::closestPtPointTriangle(const b3Vector3& p, const b3Vector3& a, const b3Vector3& b, const b3Vector3& c,b3SubSimplexClosestResult& result)
+bool b3VoronoiSimplexSolver::closestPtPointTriangle(const b3Vector3& p, const b3Vector3& a, const b3Vector3& b, const b3Vector3& c, b3SubSimplexClosestResult& result)
 {
 	result.m_usedVertices.reset();
 
-    // Check if P in vertex region outside A
-    b3Vector3 ab = b - a;
-    b3Vector3 ac = c - a;
-    b3Vector3 ap = p - a;
-    b3Scalar d1 = ab.dot(ap);
-    b3Scalar d2 = ac.dot(ap);
-    if (d1 <= b3Scalar(0.0) && d2 <= b3Scalar(0.0)) 
+	// Check if P in vertex region outside A
+	b3Vector3 ab = b - a;
+	b3Vector3 ac = c - a;
+	b3Vector3 ap = p - a;
+	b3Scalar d1 = ab.dot(ap);
+	b3Scalar d2 = ac.dot(ap);
+	if (d1 <= b3Scalar(0.0) && d2 <= b3Scalar(0.0))
 	{
 		result.m_closestPointOnSimplex = a;
 		result.m_usedVertices.usedVertexA = true;
-		result.setBarycentricCoordinates(1,0,0);
-		return true;// a; // barycentric coordinates (1,0,0)
+		result.setBarycentricCoordinates(1, 0, 0);
+		return true;  // a; // barycentric coordinates (1,0,0)
 	}
 
-    // Check if P in vertex region outside B
-    b3Vector3 bp = p - b;
-    b3Scalar d3 = ab.dot(bp);
-    b3Scalar d4 = ac.dot(bp);
-    if (d3 >= b3Scalar(0.0) && d4 <= d3) 
+	// Check if P in vertex region outside B
+	b3Vector3 bp = p - b;
+	b3Scalar d3 = ab.dot(bp);
+	b3Scalar d4 = ac.dot(bp);
+	if (d3 >= b3Scalar(0.0) && d4 <= d3)
 	{
 		result.m_closestPointOnSimplex = b;
 		result.m_usedVertices.usedVertexB = true;
-		result.setBarycentricCoordinates(0,1,0);
+		result.setBarycentricCoordinates(0, 1, 0);
 
-		return true; // b; // barycentric coordinates (0,1,0)
+		return true;  // b; // barycentric coordinates (0,1,0)
 	}
-    // Check if P in edge region of AB, if so return projection of P onto AB
-    b3Scalar vc = d1*d4 - d3*d2;
-    if (vc <= b3Scalar(0.0) && d1 >= b3Scalar(0.0) && d3 <= b3Scalar(0.0)) {
-        b3Scalar v = d1 / (d1 - d3);
+	// Check if P in edge region of AB, if so return projection of P onto AB
+	b3Scalar vc = d1 * d4 - d3 * d2;
+	if (vc <= b3Scalar(0.0) && d1 >= b3Scalar(0.0) && d3 <= b3Scalar(0.0))
+	{
+		b3Scalar v = d1 / (d1 - d3);
 		result.m_closestPointOnSimplex = a + v * ab;
 		result.m_usedVertices.usedVertexA = true;
 		result.m_usedVertices.usedVertexB = true;
-		result.setBarycentricCoordinates(1-v,v,0);
+		result.setBarycentricCoordinates(1 - v, v, 0);
 		return true;
-        //return a + v * ab; // barycentric coordinates (1-v,v,0)
-    }
+		//return a + v * ab; // barycentric coordinates (1-v,v,0)
+	}
 
-    // Check if P in vertex region outside C
-    b3Vector3 cp = p - c;
-    b3Scalar d5 = ab.dot(cp);
-    b3Scalar d6 = ac.dot(cp);
-    if (d6 >= b3Scalar(0.0) && d5 <= d6) 
+	// Check if P in vertex region outside C
+	b3Vector3 cp = p - c;
+	b3Scalar d5 = ab.dot(cp);
+	b3Scalar d6 = ac.dot(cp);
+	if (d6 >= b3Scalar(0.0) && d5 <= d6)
 	{
 		result.m_closestPointOnSimplex = c;
 		result.m_usedVertices.usedVertexC = true;
-		result.setBarycentricCoordinates(0,0,1);
-		return true;//c; // barycentric coordinates (0,0,1)
+		result.setBarycentricCoordinates(0, 0, 1);
+		return true;  //c; // barycentric coordinates (0,0,1)
 	}
 
-    // Check if P in edge region of AC, if so return projection of P onto AC
-    b3Scalar vb = d5*d2 - d1*d6;
-    if (vb <= b3Scalar(0.0) && d2 >= b3Scalar(0.0) && d6 <= b3Scalar(0.0)) {
-        b3Scalar w = d2 / (d2 - d6);
+	// Check if P in edge region of AC, if so return projection of P onto AC
+	b3Scalar vb = d5 * d2 - d1 * d6;
+	if (vb <= b3Scalar(0.0) && d2 >= b3Scalar(0.0) && d6 <= b3Scalar(0.0))
+	{
+		b3Scalar w = d2 / (d2 - d6);
 		result.m_closestPointOnSimplex = a + w * ac;
 		result.m_usedVertices.usedVertexA = true;
 		result.m_usedVertices.usedVertexC = true;
-		result.setBarycentricCoordinates(1-w,0,w);
+		result.setBarycentricCoordinates(1 - w, 0, w);
 		return true;
-        //return a + w * ac; // barycentric coordinates (1-w,0,w)
-    }
+		//return a + w * ac; // barycentric coordinates (1-w,0,w)
+	}
+
+	// Check if P in edge region of BC, if so return projection of P onto BC
+	b3Scalar va = d3 * d6 - d5 * d4;
+	if (va <= b3Scalar(0.0) && (d4 - d3) >= b3Scalar(0.0) && (d5 - d6) >= b3Scalar(0.0))
+	{
+		b3Scalar w = (d4 - d3) / ((d4 - d3) + (d5 - d6));
 
-    // Check if P in edge region of BC, if so return projection of P onto BC
-    b3Scalar va = d3*d6 - d5*d4;
-    if (va <= b3Scalar(0.0) && (d4 - d3) >= b3Scalar(0.0) && (d5 - d6) >= b3Scalar(0.0)) {
-        b3Scalar w = (d4 - d3) / ((d4 - d3) + (d5 - d6));
-		
 		result.m_closestPointOnSimplex = b + w * (c - b);
 		result.m_usedVertices.usedVertexB = true;
 		result.m_usedVertices.usedVertexC = true;
-		result.setBarycentricCoordinates(0,1-w,w);
-		return true;		
-       // return b + w * (c - b); // barycentric coordinates (0,1-w,w)
-    }
+		result.setBarycentricCoordinates(0, 1 - w, w);
+		return true;
+		// return b + w * (c - b); // barycentric coordinates (0,1-w,w)
+	}
+
+	// P inside face region. Compute Q through its barycentric coordinates (u,v,w)
+	b3Scalar denom = b3Scalar(1.0) / (va + vb + vc);
+	b3Scalar v = vb * denom;
+	b3Scalar w = vc * denom;
 
-    // P inside face region. Compute Q through its barycentric coordinates (u,v,w)
-    b3Scalar denom = b3Scalar(1.0) / (va + vb + vc);
-    b3Scalar v = vb * denom;
-    b3Scalar w = vc * denom;
-    
 	result.m_closestPointOnSimplex = a + ab * v + ac * w;
 	result.m_usedVertices.usedVertexA = true;
 	result.m_usedVertices.usedVertexB = true;
 	result.m_usedVertices.usedVertexC = true;
-	result.setBarycentricCoordinates(1-v-w,v,w);
-	
+	result.setBarycentricCoordinates(1 - v - w, v, w);
+
 	return true;
-//	return a + ab * v + ac * w; // = u*a + v*b + w*c, u = va * denom = b3Scalar(1.0) - v - w
-
+	//	return a + ab * v + ac * w; // = u*a + v*b + w*c, u = va * denom = b3Scalar(1.0) - v - w
 }
 
-
-
-
-
 /// Test if point p and d lie on opposite sides of plane through abc
 int b3VoronoiSimplexSolver::pointOutsideOfPlane(const b3Vector3& p, const b3Vector3& a, const b3Vector3& b, const b3Vector3& c, const b3Vector3& d)
 {
-	b3Vector3 normal = (b-a).cross(c-a);
+	b3Vector3 normal = (b - a).cross(c - a);
 
-    b3Scalar signp = (p - a).dot(normal); // [AP AB AC]
-    b3Scalar signd = (d - a).dot( normal); // [AD AB AC]
+	b3Scalar signp = (p - a).dot(normal);  // [AP AB AC]
+	b3Scalar signd = (d - a).dot(normal);  // [AD AB AC]
 
 #ifdef B3_CATCH_DEGENERATE_TETRAHEDRON
 #ifdef BT_USE_DOUBLE_PRECISION
-if (signd * signd < (b3Scalar(1e-8) * b3Scalar(1e-8)))
+	if (signd * signd < (b3Scalar(1e-8) * b3Scalar(1e-8)))
 	{
 		return -1;
 	}
 #else
 	if (signd * signd < (b3Scalar(1e-4) * b3Scalar(1e-4)))
 	{
-//		printf("affine dependent/degenerate\n");//
+		//		printf("affine dependent/degenerate\n");//
 		return -1;
 	}
 #endif
 
 #endif
 	// Points on opposite sides if expression signs are opposite
-    return signp * signd < b3Scalar(0.);
+	return signp * signd < b3Scalar(0.);
 }
 
-
-bool	b3VoronoiSimplexSolver::closestPtPointTetrahedron(const b3Vector3& p, const b3Vector3& a, const b3Vector3& b, const b3Vector3& c, const b3Vector3& d, b3SubSimplexClosestResult& finalResult)
+bool b3VoronoiSimplexSolver::closestPtPointTetrahedron(const b3Vector3& p, const b3Vector3& a, const b3Vector3& b, const b3Vector3& c, const b3Vector3& d, b3SubSimplexClosestResult& finalResult)
 {
 	b3SubSimplexClosestResult tempResult;
 
-    // Start out assuming point inside all halfspaces, so closest to itself
+	// Start out assuming point inside all halfspaces, so closest to itself
 	finalResult.m_closestPointOnSimplex = p;
 	finalResult.m_usedVertices.reset();
-    finalResult.m_usedVertices.usedVertexA = true;
+	finalResult.m_usedVertices.usedVertexA = true;
 	finalResult.m_usedVertices.usedVertexB = true;
 	finalResult.m_usedVertices.usedVertexC = true;
 	finalResult.m_usedVertices.usedVertexD = true;
 
-    int pointOutsideABC = pointOutsideOfPlane(p, a, b, c, d);
+	int pointOutsideABC = pointOutsideOfPlane(p, a, b, c, d);
 	int pointOutsideACD = pointOutsideOfPlane(p, a, c, d, b);
-  	int	pointOutsideADB = pointOutsideOfPlane(p, a, d, b, c);
-	int	pointOutsideBDC = pointOutsideOfPlane(p, b, d, c, a);
+	int pointOutsideADB = pointOutsideOfPlane(p, a, d, b, c);
+	int pointOutsideBDC = pointOutsideOfPlane(p, b, d, c, a);
 
-   if (pointOutsideABC < 0 || pointOutsideACD < 0 || pointOutsideADB < 0 || pointOutsideBDC < 0)
-   {
-	   finalResult.m_degenerate = true;
-	   return false;
-   }
-
-   if (!pointOutsideABC  && !pointOutsideACD && !pointOutsideADB && !pointOutsideBDC)
-	 {
-		 return false;
-	 }
-
-
-    b3Scalar bestSqDist = FLT_MAX;
-    // If point outside face abc then compute closest point on abc
-	if (pointOutsideABC) 
+	if (pointOutsideABC < 0 || pointOutsideACD < 0 || pointOutsideADB < 0 || pointOutsideBDC < 0)
 	{
-        closestPtPointTriangle(p, a, b, c,tempResult);
+		finalResult.m_degenerate = true;
+		return false;
+	}
+
+	if (!pointOutsideABC && !pointOutsideACD && !pointOutsideADB && !pointOutsideBDC)
+	{
+		return false;
+	}
+
+	b3Scalar bestSqDist = FLT_MAX;
+	// If point outside face abc then compute closest point on abc
+	if (pointOutsideABC)
+	{
+		closestPtPointTriangle(p, a, b, c, tempResult);
 		b3Vector3 q = tempResult.m_closestPointOnSimplex;
-		
-        b3Scalar sqDist = (q - p).dot( q - p);
-        // Update best closest point if (squared) distance is less than current best
-        if (sqDist < bestSqDist) {
+
+		b3Scalar sqDist = (q - p).dot(q - p);
+		// Update best closest point if (squared) distance is less than current best
+		if (sqDist < bestSqDist)
+		{
 			bestSqDist = sqDist;
 			finalResult.m_closestPointOnSimplex = q;
 			//convert result bitmask!
@@ -501,25 +478,22 @@ bool	b3VoronoiSimplexSolver::closestPtPointTetrahedron(const b3Vector3& p, const
 			finalResult.m_usedVertices.usedVertexB = tempResult.m_usedVertices.usedVertexB;
 			finalResult.m_usedVertices.usedVertexC = tempResult.m_usedVertices.usedVertexC;
 			finalResult.setBarycentricCoordinates(
-					tempResult.m_barycentricCoords[VERTA],
-					tempResult.m_barycentricCoords[VERTB],
-					tempResult.m_barycentricCoords[VERTC],
-					0
-			);
-
+				tempResult.m_barycentricCoords[VERTA],
+				tempResult.m_barycentricCoords[VERTB],
+				tempResult.m_barycentricCoords[VERTC],
+				0);
 		}
-    }
-  
+	}
 
 	// Repeat test for face acd
-	if (pointOutsideACD) 
+	if (pointOutsideACD)
 	{
-        closestPtPointTriangle(p, a, c, d,tempResult);
+		closestPtPointTriangle(p, a, c, d, tempResult);
 		b3Vector3 q = tempResult.m_closestPointOnSimplex;
 		//convert result bitmask!
 
-        b3Scalar sqDist = (q - p).dot( q - p);
-        if (sqDist < bestSqDist) 
+		b3Scalar sqDist = (q - p).dot(q - p);
+		if (sqDist < bestSqDist)
 		{
 			bestSqDist = sqDist;
 			finalResult.m_closestPointOnSimplex = q;
@@ -529,52 +503,46 @@ bool	b3VoronoiSimplexSolver::closestPtPointTetrahedron(const b3Vector3& p, const
 			finalResult.m_usedVertices.usedVertexC = tempResult.m_usedVertices.usedVertexB;
 			finalResult.m_usedVertices.usedVertexD = tempResult.m_usedVertices.usedVertexC;
 			finalResult.setBarycentricCoordinates(
-					tempResult.m_barycentricCoords[VERTA],
-					0,
-					tempResult.m_barycentricCoords[VERTB],
-					tempResult.m_barycentricCoords[VERTC]
-			);
-
+				tempResult.m_barycentricCoords[VERTA],
+				0,
+				tempResult.m_barycentricCoords[VERTB],
+				tempResult.m_barycentricCoords[VERTC]);
 		}
-    }
-    // Repeat test for face adb
+	}
+	// Repeat test for face adb
 
-	
 	if (pointOutsideADB)
 	{
-		closestPtPointTriangle(p, a, d, b,tempResult);
+		closestPtPointTriangle(p, a, d, b, tempResult);
 		b3Vector3 q = tempResult.m_closestPointOnSimplex;
 		//convert result bitmask!
 
-        b3Scalar sqDist = (q - p).dot( q - p);
-        if (sqDist < bestSqDist) 
+		b3Scalar sqDist = (q - p).dot(q - p);
+		if (sqDist < bestSqDist)
 		{
 			bestSqDist = sqDist;
 			finalResult.m_closestPointOnSimplex = q;
 			finalResult.m_usedVertices.reset();
 			finalResult.m_usedVertices.usedVertexA = tempResult.m_usedVertices.usedVertexA;
 			finalResult.m_usedVertices.usedVertexB = tempResult.m_usedVertices.usedVertexC;
-			
+
 			finalResult.m_usedVertices.usedVertexD = tempResult.m_usedVertices.usedVertexB;
 			finalResult.setBarycentricCoordinates(
-					tempResult.m_barycentricCoords[VERTA],
-					tempResult.m_barycentricCoords[VERTC],
-					0,
-					tempResult.m_barycentricCoords[VERTB]
-			);
-
+				tempResult.m_barycentricCoords[VERTA],
+				tempResult.m_barycentricCoords[VERTC],
+				0,
+				tempResult.m_barycentricCoords[VERTB]);
 		}
-    }
-    // Repeat test for face bdc
-    
+	}
+	// Repeat test for face bdc
 
 	if (pointOutsideBDC)
 	{
-        closestPtPointTriangle(p, b, d, c,tempResult);
+		closestPtPointTriangle(p, b, d, c, tempResult);
 		b3Vector3 q = tempResult.m_closestPointOnSimplex;
 		//convert result bitmask!
-        b3Scalar sqDist = (q - p).dot( q - p);
-        if (sqDist < bestSqDist) 
+		b3Scalar sqDist = (q - p).dot(q - p);
+		if (sqDist < bestSqDist)
 		{
 			bestSqDist = sqDist;
 			finalResult.m_closestPointOnSimplex = q;
@@ -585,25 +553,22 @@ bool	b3VoronoiSimplexSolver::closestPtPointTetrahedron(const b3Vector3& p, const
 			finalResult.m_usedVertices.usedVertexD = tempResult.m_usedVertices.usedVertexB;
 
 			finalResult.setBarycentricCoordinates(
-					0,
-					tempResult.m_barycentricCoords[VERTA],
-					tempResult.m_barycentricCoords[VERTC],
-					tempResult.m_barycentricCoords[VERTB]
-			);
-
+				0,
+				tempResult.m_barycentricCoords[VERTA],
+				tempResult.m_barycentricCoords[VERTC],
+				tempResult.m_barycentricCoords[VERTB]);
 		}
-    }
+	}
 
 	//help! we ended up full !
-	
+
 	if (finalResult.m_usedVertices.usedVertexA &&
 		finalResult.m_usedVertices.usedVertexB &&
 		finalResult.m_usedVertices.usedVertexC &&
-		finalResult.m_usedVertices.usedVertexD) 
+		finalResult.m_usedVertices.usedVertexD)
 	{
 		return true;
 	}
 
-    return true;
+	return true;
 }
-
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3VoronoiSimplexSolver.h b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3VoronoiSimplexSolver.h
index a6e27667d89..b40b169978e 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3VoronoiSimplexSolver.h
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/b3VoronoiSimplexSolver.h
@@ -13,22 +13,19 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #ifndef B3_VORONOI_SIMPLEX_SOLVER_H
 #define B3_VORONOI_SIMPLEX_SOLVER_H
 
 #include "Bullet3Common/b3Vector3.h"
 
-
 #define VORONOI_SIMPLEX_MAX_VERTS 5
 
 ///disable next define, or use defaultCollisionConfiguration->getSimplexSolver()->setEqualVertexThreshold(0.f) to disable/configure
 //#define BT_USE_EQUAL_VERTEX_THRESHOLD
 #define VORONOI_DEFAULT_EQUAL_VERTEX_THRESHOLD 0.0001f
 
-
-struct b3UsageBitfield{
+struct b3UsageBitfield
+{
 	b3UsageBitfield()
 	{
 		reset();
@@ -41,137 +38,127 @@ struct b3UsageBitfield{
 		usedVertexC = false;
 		usedVertexD = false;
 	}
-	unsigned short usedVertexA	: 1;
-	unsigned short usedVertexB	: 1;
-	unsigned short usedVertexC	: 1;
-	unsigned short usedVertexD	: 1;
-	unsigned short unused1		: 1;
-	unsigned short unused2		: 1;
-	unsigned short unused3		: 1;
-	unsigned short unused4		: 1;
+	unsigned short usedVertexA : 1;
+	unsigned short usedVertexB : 1;
+	unsigned short usedVertexC : 1;
+	unsigned short usedVertexD : 1;
+	unsigned short unused1 : 1;
+	unsigned short unused2 : 1;
+	unsigned short unused3 : 1;
+	unsigned short unused4 : 1;
 };
 
-
-struct	b3SubSimplexClosestResult
+struct b3SubSimplexClosestResult
 {
-	b3Vector3	m_closestPointOnSimplex;
+	b3Vector3 m_closestPointOnSimplex;
 	//MASK for m_usedVertices
-	//stores the simplex vertex-usage, using the MASK, 
+	//stores the simplex vertex-usage, using the MASK,
 	// if m_usedVertices & MASK then the related vertex is used
-	b3UsageBitfield	m_usedVertices;
-	b3Scalar	m_barycentricCoords[4];
+	b3UsageBitfield m_usedVertices;
+	b3Scalar m_barycentricCoords[4];
 	bool m_degenerate;
 
-	void	reset()
+	void reset()
 	{
 		m_degenerate = false;
 		setBarycentricCoordinates();
 		m_usedVertices.reset();
 	}
-	bool	isValid()
+	bool isValid()
 	{
 		bool valid = (m_barycentricCoords[0] >= b3Scalar(0.)) &&
-			(m_barycentricCoords[1] >= b3Scalar(0.)) &&
-			(m_barycentricCoords[2] >= b3Scalar(0.)) &&
-			(m_barycentricCoords[3] >= b3Scalar(0.));
-
+					 (m_barycentricCoords[1] >= b3Scalar(0.)) &&
+					 (m_barycentricCoords[2] >= b3Scalar(0.)) &&
+					 (m_barycentricCoords[3] >= b3Scalar(0.));
 
 		return valid;
 	}
-	void	setBarycentricCoordinates(b3Scalar a=b3Scalar(0.),b3Scalar b=b3Scalar(0.),b3Scalar c=b3Scalar(0.),b3Scalar d=b3Scalar(0.))
+	void setBarycentricCoordinates(b3Scalar a = b3Scalar(0.), b3Scalar b = b3Scalar(0.), b3Scalar c = b3Scalar(0.), b3Scalar d = b3Scalar(0.))
 	{
 		m_barycentricCoords[0] = a;
 		m_barycentricCoords[1] = b;
 		m_barycentricCoords[2] = c;
 		m_barycentricCoords[3] = d;
 	}
-
 };
 
 /// b3VoronoiSimplexSolver is an implementation of the closest point distance algorithm from a 1-4 points simplex to the origin.
 /// Can be used with GJK, as an alternative to Johnson distance algorithm.
 
-B3_ATTRIBUTE_ALIGNED16(class) b3VoronoiSimplexSolver 
+B3_ATTRIBUTE_ALIGNED16(class)
+b3VoronoiSimplexSolver
 {
 public:
-
 	B3_DECLARE_ALIGNED_ALLOCATOR();
 
-	int	m_numVertices;
+	int m_numVertices;
 
-	b3Vector3	m_simplexVectorW[VORONOI_SIMPLEX_MAX_VERTS];
-	b3Vector3	m_simplexPointsP[VORONOI_SIMPLEX_MAX_VERTS];
-	b3Vector3	m_simplexPointsQ[VORONOI_SIMPLEX_MAX_VERTS];
+	b3Vector3 m_simplexVectorW[VORONOI_SIMPLEX_MAX_VERTS];
+	b3Vector3 m_simplexPointsP[VORONOI_SIMPLEX_MAX_VERTS];
+	b3Vector3 m_simplexPointsQ[VORONOI_SIMPLEX_MAX_VERTS];
 
-	
-
-	b3Vector3	m_cachedP1;
-	b3Vector3	m_cachedP2;
-	b3Vector3	m_cachedV;
-	b3Vector3	m_lastW;
-	
-	b3Scalar	m_equalVertexThreshold;
-	bool		m_cachedValidClosest;
+	b3Vector3 m_cachedP1;
+	b3Vector3 m_cachedP2;
+	b3Vector3 m_cachedV;
+	b3Vector3 m_lastW;
 
+	b3Scalar m_equalVertexThreshold;
+	bool m_cachedValidClosest;
 
 	b3SubSimplexClosestResult m_cachedBC;
 
-	bool	m_needsUpdate;
-	
-	void	removeVertex(int index);
-	void	reduceVertices (const b3UsageBitfield& usedVerts);
-	bool	updateClosestVectorAndPoints();
+	bool m_needsUpdate;
 
-	bool	closestPtPointTetrahedron(const b3Vector3& p, const b3Vector3& a, const b3Vector3& b, const b3Vector3& c, const b3Vector3& d, b3SubSimplexClosestResult& finalResult);
-	int		pointOutsideOfPlane(const b3Vector3& p, const b3Vector3& a, const b3Vector3& b, const b3Vector3& c, const b3Vector3& d);
-	bool	closestPtPointTriangle(const b3Vector3& p, const b3Vector3& a, const b3Vector3& b, const b3Vector3& c,b3SubSimplexClosestResult& result);
+	void removeVertex(int index);
+	void reduceVertices(const b3UsageBitfield& usedVerts);
+	bool updateClosestVectorAndPoints();
+
+	bool closestPtPointTetrahedron(const b3Vector3& p, const b3Vector3& a, const b3Vector3& b, const b3Vector3& c, const b3Vector3& d, b3SubSimplexClosestResult& finalResult);
+	int pointOutsideOfPlane(const b3Vector3& p, const b3Vector3& a, const b3Vector3& b, const b3Vector3& c, const b3Vector3& d);
+	bool closestPtPointTriangle(const b3Vector3& p, const b3Vector3& a, const b3Vector3& b, const b3Vector3& c, b3SubSimplexClosestResult& result);
 
 public:
-
 	b3VoronoiSimplexSolver()
-		:  m_equalVertexThreshold(VORONOI_DEFAULT_EQUAL_VERTEX_THRESHOLD)
+		: m_equalVertexThreshold(VORONOI_DEFAULT_EQUAL_VERTEX_THRESHOLD)
 	{
 	}
-	 void reset();
+	void reset();
 
-	 void addVertex(const b3Vector3& w, const b3Vector3& p, const b3Vector3& q);
+	void addVertex(const b3Vector3& w, const b3Vector3& p, const b3Vector3& q);
 
-	 void	setEqualVertexThreshold(b3Scalar threshold)
-	 {
-		 m_equalVertexThreshold = threshold;
-	 }
+	void setEqualVertexThreshold(b3Scalar threshold)
+	{
+		m_equalVertexThreshold = threshold;
+	}
 
-	 b3Scalar	getEqualVertexThreshold() const
-	 {
-		 return m_equalVertexThreshold;
-	 }
+	b3Scalar getEqualVertexThreshold() const
+	{
+		return m_equalVertexThreshold;
+	}
 
-	 bool closest(b3Vector3& v);
+	bool closest(b3Vector3 & v);
 
-	 b3Scalar maxVertex();
+	b3Scalar maxVertex();
 
-	 bool fullSimplex() const
-	 {
-		 return (m_numVertices == 4);
-	 }
+	bool fullSimplex() const
+	{
+		return (m_numVertices == 4);
+	}
 
-	 int getSimplex(b3Vector3 *pBuf, b3Vector3 *qBuf, b3Vector3 *yBuf) const;
+	int getSimplex(b3Vector3 * pBuf, b3Vector3 * qBuf, b3Vector3 * yBuf) const;
 
-	 bool inSimplex(const b3Vector3& w);
-	
-	 void backup_closest(b3Vector3& v) ;
+	bool inSimplex(const b3Vector3& w);
 
-	 bool emptySimplex() const ;
+	void backup_closest(b3Vector3 & v);
 
-	 void compute_points(b3Vector3& p1, b3Vector3& p2) ;
-
-	 int numVertices() const 
-	 {
-		 return m_numVertices;
-	 }
+	bool emptySimplex() const;
 
+	void compute_points(b3Vector3 & p1, b3Vector3 & p2);
 
+	int numVertices() const
+	{
+		return m_numVertices;
+	}
 };
 
-#endif //B3_VORONOI_SIMPLEX_SOLVER_H
-
+#endif  //B3_VORONOI_SIMPLEX_SOLVER_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/bvhTraversal.h b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/bvhTraversal.h
index 4b3b49eae8d..f1df8a69706 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/bvhTraversal.h
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/bvhTraversal.h
@@ -1,258 +1,257 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* bvhTraversalKernelCL= \
-"//keep this enum in sync with the CPU version (in btCollidable.h)\n"
-"//written by Erwin Coumans\n"
-"#define SHAPE_CONVEX_HULL 3\n"
-"#define SHAPE_CONCAVE_TRIMESH 5\n"
-"#define TRIANGLE_NUM_CONVEX_FACES 5\n"
-"#define SHAPE_COMPOUND_OF_CONVEX_HULLS 6\n"
-"#define SHAPE_SPHERE 7\n"
-"typedef unsigned int u32;\n"
-"#define MAX_NUM_PARTS_IN_BITS 10\n"
-"///btQuantizedBvhNode is a compressed aabb node, 16 bytes.\n"
-"///Node can be used for leafnode or internal node. Leafnodes can point to 32-bit triangle index (non-negative range).\n"
-"typedef struct\n"
-"{\n"
-"	//12 bytes\n"
-"	unsigned short int	m_quantizedAabbMin[3];\n"
-"	unsigned short int	m_quantizedAabbMax[3];\n"
-"	//4 bytes\n"
-"	int	m_escapeIndexOrTriangleIndex;\n"
-"} btQuantizedBvhNode;\n"
-"typedef struct\n"
-"{\n"
-"	float4		m_aabbMin;\n"
-"	float4		m_aabbMax;\n"
-"	float4		m_quantization;\n"
-"	int			m_numNodes;\n"
-"	int			m_numSubTrees;\n"
-"	int			m_nodeOffset;\n"
-"	int			m_subTreeOffset;\n"
-"} b3BvhInfo;\n"
-"int	getTriangleIndex(const btQuantizedBvhNode* rootNode)\n"
-"{\n"
-"	unsigned int x=0;\n"
-"	unsigned int y = (~(x&0))<<(31-MAX_NUM_PARTS_IN_BITS);\n"
-"	// Get only the lower bits where the triangle index is stored\n"
-"	return (rootNode->m_escapeIndexOrTriangleIndex&~(y));\n"
-"}\n"
-"int isLeaf(const btQuantizedBvhNode* rootNode)\n"
-"{\n"
-"	//skipindex is negative (internal node), triangleindex >=0 (leafnode)\n"
-"	return (rootNode->m_escapeIndexOrTriangleIndex >= 0)? 1 : 0;\n"
-"}\n"
-"	\n"
-"int getEscapeIndex(const btQuantizedBvhNode* rootNode)\n"
-"{\n"
-"	return -rootNode->m_escapeIndexOrTriangleIndex;\n"
-"}\n"
-"typedef struct\n"
-"{\n"
-"	//12 bytes\n"
-"	unsigned short int	m_quantizedAabbMin[3];\n"
-"	unsigned short int	m_quantizedAabbMax[3];\n"
-"	//4 bytes, points to the root of the subtree\n"
-"	int			m_rootNodeIndex;\n"
-"	//4 bytes\n"
-"	int			m_subtreeSize;\n"
-"	int			m_padding[3];\n"
-"} btBvhSubtreeInfo;\n"
-"///keep this in sync with btCollidable.h\n"
-"typedef struct\n"
-"{\n"
-"	int m_numChildShapes;\n"
-"	int blaat2;\n"
-"	int m_shapeType;\n"
-"	int m_shapeIndex;\n"
-"	\n"
-"} btCollidableGpu;\n"
-"typedef struct\n"
-"{\n"
-"	float4	m_childPosition;\n"
-"	float4	m_childOrientation;\n"
-"	int m_shapeIndex;\n"
-"	int m_unused0;\n"
-"	int m_unused1;\n"
-"	int m_unused2;\n"
-"} btGpuChildShape;\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_pos;\n"
-"	float4 m_quat;\n"
-"	float4 m_linVel;\n"
-"	float4 m_angVel;\n"
-"	u32 m_collidableIdx;\n"
-"	float m_invMass;\n"
-"	float m_restituitionCoeff;\n"
-"	float m_frictionCoeff;\n"
-"} BodyData;\n"
-"typedef struct \n"
-"{\n"
-"	union\n"
-"	{\n"
-"		float4	m_min;\n"
-"		float   m_minElems[4];\n"
-"		int			m_minIndices[4];\n"
-"	};\n"
-"	union\n"
-"	{\n"
-"		float4	m_max;\n"
-"		float   m_maxElems[4];\n"
-"		int			m_maxIndices[4];\n"
-"	};\n"
-"} btAabbCL;\n"
-"int testQuantizedAabbAgainstQuantizedAabb(\n"
-"								const unsigned short int* aabbMin1,\n"
-"								const unsigned short int* aabbMax1,\n"
-"								const unsigned short int* aabbMin2,\n"
-"								const unsigned short int* aabbMax2)\n"
-"{\n"
-"	//int overlap = 1;\n"
-"	if (aabbMin1[0] > aabbMax2[0])\n"
-"		return 0;\n"
-"	if (aabbMax1[0] < aabbMin2[0])\n"
-"		return 0;\n"
-"	if (aabbMin1[1] > aabbMax2[1])\n"
-"		return 0;\n"
-"	if (aabbMax1[1] < aabbMin2[1])\n"
-"		return 0;\n"
-"	if (aabbMin1[2] > aabbMax2[2])\n"
-"		return 0;\n"
-"	if (aabbMax1[2] < aabbMin2[2])\n"
-"		return 0;\n"
-"	return 1;\n"
-"	//overlap = ((aabbMin1[0] > aabbMax2[0]) || (aabbMax1[0] < aabbMin2[0])) ? 0 : overlap;\n"
-"	//overlap = ((aabbMin1[2] > aabbMax2[2]) || (aabbMax1[2] < aabbMin2[2])) ? 0 : overlap;\n"
-"	//overlap = ((aabbMin1[1] > aabbMax2[1]) || (aabbMax1[1] < aabbMin2[1])) ? 0 : overlap;\n"
-"	//return overlap;\n"
-"}\n"
-"void quantizeWithClamp(unsigned short* out, float4 point2,int isMax, float4 bvhAabbMin, float4 bvhAabbMax, float4 bvhQuantization)\n"
-"{\n"
-"	float4 clampedPoint = max(point2,bvhAabbMin);\n"
-"	clampedPoint = min (clampedPoint, bvhAabbMax);\n"
-"	float4 v = (clampedPoint - bvhAabbMin) * bvhQuantization;\n"
-"	if (isMax)\n"
-"	{\n"
-"		out[0] = (unsigned short) (((unsigned short)(v.x+1.f) | 1));\n"
-"		out[1] = (unsigned short) (((unsigned short)(v.y+1.f) | 1));\n"
-"		out[2] = (unsigned short) (((unsigned short)(v.z+1.f) | 1));\n"
-"	} else\n"
-"	{\n"
-"		out[0] = (unsigned short) (((unsigned short)(v.x) & 0xfffe));\n"
-"		out[1] = (unsigned short) (((unsigned short)(v.y) & 0xfffe));\n"
-"		out[2] = (unsigned short) (((unsigned short)(v.z) & 0xfffe));\n"
-"	}\n"
-"}\n"
-"// work-in-progress\n"
-"__kernel void   bvhTraversalKernel( __global const int4* pairs, \n"
-"									__global const BodyData* rigidBodies, \n"
-"									__global const btCollidableGpu* collidables,\n"
-"									__global btAabbCL* aabbs,\n"
-"									__global int4* concavePairsOut,\n"
-"									__global volatile int* numConcavePairsOut,\n"
-"									__global const btBvhSubtreeInfo* subtreeHeadersRoot,\n"
-"									__global const btQuantizedBvhNode* quantizedNodesRoot,\n"
-"									__global const b3BvhInfo* bvhInfos,\n"
-"									int numPairs,\n"
-"									int maxNumConcavePairsCapacity)\n"
-"{\n"
-"	int id = get_global_id(0);\n"
-"	if (id>=numPairs)\n"
-"		return;\n"
-"	\n"
-"	int bodyIndexA = pairs[id].x;\n"
-"	int bodyIndexB = pairs[id].y;\n"
-"	int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
-"	int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
-"	\n"
-"	//once the broadphase avoids static-static pairs, we can remove this test\n"
-"	if ((rigidBodies[bodyIndexA].m_invMass==0) &&(rigidBodies[bodyIndexB].m_invMass==0))\n"
-"	{\n"
-"		return;\n"
-"	}\n"
-"		\n"
-"	if (collidables[collidableIndexA].m_shapeType!=SHAPE_CONCAVE_TRIMESH)\n"
-"		return;\n"
-"	int shapeTypeB = collidables[collidableIndexB].m_shapeType;\n"
-"		\n"
-"	if (shapeTypeB!=SHAPE_CONVEX_HULL &&\n"
-"		shapeTypeB!=SHAPE_SPHERE	&&\n"
-"		shapeTypeB!=SHAPE_COMPOUND_OF_CONVEX_HULLS\n"
-"		)\n"
-"		return;\n"
-"	b3BvhInfo bvhInfo = bvhInfos[collidables[collidableIndexA].m_numChildShapes];\n"
-"	float4 bvhAabbMin = bvhInfo.m_aabbMin;\n"
-"	float4 bvhAabbMax = bvhInfo.m_aabbMax;\n"
-"	float4 bvhQuantization = bvhInfo.m_quantization;\n"
-"	int numSubtreeHeaders = bvhInfo.m_numSubTrees;\n"
-"	__global const btBvhSubtreeInfo* subtreeHeaders = &subtreeHeadersRoot[bvhInfo.m_subTreeOffset];\n"
-"	__global const btQuantizedBvhNode* quantizedNodes = &quantizedNodesRoot[bvhInfo.m_nodeOffset];\n"
-"	\n"
-"	unsigned short int quantizedQueryAabbMin[3];\n"
-"	unsigned short int quantizedQueryAabbMax[3];\n"
-"	quantizeWithClamp(quantizedQueryAabbMin,aabbs[bodyIndexB].m_min,false,bvhAabbMin, bvhAabbMax,bvhQuantization);\n"
-"	quantizeWithClamp(quantizedQueryAabbMax,aabbs[bodyIndexB].m_max,true ,bvhAabbMin, bvhAabbMax,bvhQuantization);\n"
-"	\n"
-"	for (int i=0;i<numSubtreeHeaders;i++)\n"
-"	{\n"
-"		btBvhSubtreeInfo subtree = subtreeHeaders[i];\n"
-"				\n"
-"		int overlap = testQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin,quantizedQueryAabbMax,subtree.m_quantizedAabbMin,subtree.m_quantizedAabbMax);\n"
-"		if (overlap != 0)\n"
-"		{\n"
-"			int startNodeIndex = subtree.m_rootNodeIndex;\n"
-"			int endNodeIndex = subtree.m_rootNodeIndex+subtree.m_subtreeSize;\n"
-"			int curIndex = startNodeIndex;\n"
-"			int escapeIndex;\n"
-"			int isLeafNode;\n"
-"			int aabbOverlap;\n"
-"			while (curIndex < endNodeIndex)\n"
-"			{\n"
-"				btQuantizedBvhNode rootNode = quantizedNodes[curIndex];\n"
-"				aabbOverlap = testQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin,quantizedQueryAabbMax,rootNode.m_quantizedAabbMin,rootNode.m_quantizedAabbMax);\n"
-"				isLeafNode = isLeaf(&rootNode);\n"
-"				if (aabbOverlap)\n"
-"				{\n"
-"					if (isLeafNode)\n"
-"					{\n"
-"						int triangleIndex = getTriangleIndex(&rootNode);\n"
-"						if (shapeTypeB==SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
-"						{\n"
-"								int numChildrenB = collidables[collidableIndexB].m_numChildShapes;\n"
-"								int pairIdx = atomic_add(numConcavePairsOut,numChildrenB);\n"
-"								for (int b=0;b<numChildrenB;b++)\n"
-"								{\n"
-"									if ((pairIdx+b)<maxNumConcavePairsCapacity)\n"
-"									{\n"
-"										int childShapeIndexB = collidables[collidableIndexB].m_shapeIndex+b;\n"
-"										int4 newPair = (int4)(bodyIndexA,bodyIndexB,triangleIndex,childShapeIndexB);\n"
-"										concavePairsOut[pairIdx+b] = newPair;\n"
-"									}\n"
-"								}\n"
-"						} else\n"
-"						{\n"
-"							int pairIdx = atomic_inc(numConcavePairsOut);\n"
-"							if (pairIdx<maxNumConcavePairsCapacity)\n"
-"							{\n"
-"								int4 newPair = (int4)(bodyIndexA,bodyIndexB,triangleIndex,0);\n"
-"								concavePairsOut[pairIdx] = newPair;\n"
-"							}\n"
-"						}\n"
-"					} \n"
-"					curIndex++;\n"
-"				} else\n"
-"				{\n"
-"					if (isLeafNode)\n"
-"					{\n"
-"						curIndex++;\n"
-"					} else\n"
-"					{\n"
-"						escapeIndex = getEscapeIndex(&rootNode);\n"
-"						curIndex += escapeIndex;\n"
-"					}\n"
-"				}\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"}\n"
-;
+static const char* bvhTraversalKernelCL =
+	"//keep this enum in sync with the CPU version (in btCollidable.h)\n"
+	"//written by Erwin Coumans\n"
+	"#define SHAPE_CONVEX_HULL 3\n"
+	"#define SHAPE_CONCAVE_TRIMESH 5\n"
+	"#define TRIANGLE_NUM_CONVEX_FACES 5\n"
+	"#define SHAPE_COMPOUND_OF_CONVEX_HULLS 6\n"
+	"#define SHAPE_SPHERE 7\n"
+	"typedef unsigned int u32;\n"
+	"#define MAX_NUM_PARTS_IN_BITS 10\n"
+	"///btQuantizedBvhNode is a compressed aabb node, 16 bytes.\n"
+	"///Node can be used for leafnode or internal node. Leafnodes can point to 32-bit triangle index (non-negative range).\n"
+	"typedef struct\n"
+	"{\n"
+	"	//12 bytes\n"
+	"	unsigned short int	m_quantizedAabbMin[3];\n"
+	"	unsigned short int	m_quantizedAabbMax[3];\n"
+	"	//4 bytes\n"
+	"	int	m_escapeIndexOrTriangleIndex;\n"
+	"} btQuantizedBvhNode;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4		m_aabbMin;\n"
+	"	float4		m_aabbMax;\n"
+	"	float4		m_quantization;\n"
+	"	int			m_numNodes;\n"
+	"	int			m_numSubTrees;\n"
+	"	int			m_nodeOffset;\n"
+	"	int			m_subTreeOffset;\n"
+	"} b3BvhInfo;\n"
+	"int	getTriangleIndex(const btQuantizedBvhNode* rootNode)\n"
+	"{\n"
+	"	unsigned int x=0;\n"
+	"	unsigned int y = (~(x&0))<<(31-MAX_NUM_PARTS_IN_BITS);\n"
+	"	// Get only the lower bits where the triangle index is stored\n"
+	"	return (rootNode->m_escapeIndexOrTriangleIndex&~(y));\n"
+	"}\n"
+	"int isLeaf(const btQuantizedBvhNode* rootNode)\n"
+	"{\n"
+	"	//skipindex is negative (internal node), triangleindex >=0 (leafnode)\n"
+	"	return (rootNode->m_escapeIndexOrTriangleIndex >= 0)? 1 : 0;\n"
+	"}\n"
+	"	\n"
+	"int getEscapeIndex(const btQuantizedBvhNode* rootNode)\n"
+	"{\n"
+	"	return -rootNode->m_escapeIndexOrTriangleIndex;\n"
+	"}\n"
+	"typedef struct\n"
+	"{\n"
+	"	//12 bytes\n"
+	"	unsigned short int	m_quantizedAabbMin[3];\n"
+	"	unsigned short int	m_quantizedAabbMax[3];\n"
+	"	//4 bytes, points to the root of the subtree\n"
+	"	int			m_rootNodeIndex;\n"
+	"	//4 bytes\n"
+	"	int			m_subtreeSize;\n"
+	"	int			m_padding[3];\n"
+	"} btBvhSubtreeInfo;\n"
+	"///keep this in sync with btCollidable.h\n"
+	"typedef struct\n"
+	"{\n"
+	"	int m_numChildShapes;\n"
+	"	int blaat2;\n"
+	"	int m_shapeType;\n"
+	"	int m_shapeIndex;\n"
+	"	\n"
+	"} btCollidableGpu;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4	m_childPosition;\n"
+	"	float4	m_childOrientation;\n"
+	"	int m_shapeIndex;\n"
+	"	int m_unused0;\n"
+	"	int m_unused1;\n"
+	"	int m_unused2;\n"
+	"} btGpuChildShape;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_pos;\n"
+	"	float4 m_quat;\n"
+	"	float4 m_linVel;\n"
+	"	float4 m_angVel;\n"
+	"	u32 m_collidableIdx;\n"
+	"	float m_invMass;\n"
+	"	float m_restituitionCoeff;\n"
+	"	float m_frictionCoeff;\n"
+	"} BodyData;\n"
+	"typedef struct \n"
+	"{\n"
+	"	union\n"
+	"	{\n"
+	"		float4	m_min;\n"
+	"		float   m_minElems[4];\n"
+	"		int			m_minIndices[4];\n"
+	"	};\n"
+	"	union\n"
+	"	{\n"
+	"		float4	m_max;\n"
+	"		float   m_maxElems[4];\n"
+	"		int			m_maxIndices[4];\n"
+	"	};\n"
+	"} btAabbCL;\n"
+	"int testQuantizedAabbAgainstQuantizedAabb(\n"
+	"								const unsigned short int* aabbMin1,\n"
+	"								const unsigned short int* aabbMax1,\n"
+	"								const unsigned short int* aabbMin2,\n"
+	"								const unsigned short int* aabbMax2)\n"
+	"{\n"
+	"	//int overlap = 1;\n"
+	"	if (aabbMin1[0] > aabbMax2[0])\n"
+	"		return 0;\n"
+	"	if (aabbMax1[0] < aabbMin2[0])\n"
+	"		return 0;\n"
+	"	if (aabbMin1[1] > aabbMax2[1])\n"
+	"		return 0;\n"
+	"	if (aabbMax1[1] < aabbMin2[1])\n"
+	"		return 0;\n"
+	"	if (aabbMin1[2] > aabbMax2[2])\n"
+	"		return 0;\n"
+	"	if (aabbMax1[2] < aabbMin2[2])\n"
+	"		return 0;\n"
+	"	return 1;\n"
+	"	//overlap = ((aabbMin1[0] > aabbMax2[0]) || (aabbMax1[0] < aabbMin2[0])) ? 0 : overlap;\n"
+	"	//overlap = ((aabbMin1[2] > aabbMax2[2]) || (aabbMax1[2] < aabbMin2[2])) ? 0 : overlap;\n"
+	"	//overlap = ((aabbMin1[1] > aabbMax2[1]) || (aabbMax1[1] < aabbMin2[1])) ? 0 : overlap;\n"
+	"	//return overlap;\n"
+	"}\n"
+	"void quantizeWithClamp(unsigned short* out, float4 point2,int isMax, float4 bvhAabbMin, float4 bvhAabbMax, float4 bvhQuantization)\n"
+	"{\n"
+	"	float4 clampedPoint = max(point2,bvhAabbMin);\n"
+	"	clampedPoint = min (clampedPoint, bvhAabbMax);\n"
+	"	float4 v = (clampedPoint - bvhAabbMin) * bvhQuantization;\n"
+	"	if (isMax)\n"
+	"	{\n"
+	"		out[0] = (unsigned short) (((unsigned short)(v.x+1.f) | 1));\n"
+	"		out[1] = (unsigned short) (((unsigned short)(v.y+1.f) | 1));\n"
+	"		out[2] = (unsigned short) (((unsigned short)(v.z+1.f) | 1));\n"
+	"	} else\n"
+	"	{\n"
+	"		out[0] = (unsigned short) (((unsigned short)(v.x) & 0xfffe));\n"
+	"		out[1] = (unsigned short) (((unsigned short)(v.y) & 0xfffe));\n"
+	"		out[2] = (unsigned short) (((unsigned short)(v.z) & 0xfffe));\n"
+	"	}\n"
+	"}\n"
+	"// work-in-progress\n"
+	"__kernel void   bvhTraversalKernel( __global const int4* pairs, \n"
+	"									__global const BodyData* rigidBodies, \n"
+	"									__global const btCollidableGpu* collidables,\n"
+	"									__global btAabbCL* aabbs,\n"
+	"									__global int4* concavePairsOut,\n"
+	"									__global volatile int* numConcavePairsOut,\n"
+	"									__global const btBvhSubtreeInfo* subtreeHeadersRoot,\n"
+	"									__global const btQuantizedBvhNode* quantizedNodesRoot,\n"
+	"									__global const b3BvhInfo* bvhInfos,\n"
+	"									int numPairs,\n"
+	"									int maxNumConcavePairsCapacity)\n"
+	"{\n"
+	"	int id = get_global_id(0);\n"
+	"	if (id>=numPairs)\n"
+	"		return;\n"
+	"	\n"
+	"	int bodyIndexA = pairs[id].x;\n"
+	"	int bodyIndexB = pairs[id].y;\n"
+	"	int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
+	"	int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
+	"	\n"
+	"	//once the broadphase avoids static-static pairs, we can remove this test\n"
+	"	if ((rigidBodies[bodyIndexA].m_invMass==0) &&(rigidBodies[bodyIndexB].m_invMass==0))\n"
+	"	{\n"
+	"		return;\n"
+	"	}\n"
+	"		\n"
+	"	if (collidables[collidableIndexA].m_shapeType!=SHAPE_CONCAVE_TRIMESH)\n"
+	"		return;\n"
+	"	int shapeTypeB = collidables[collidableIndexB].m_shapeType;\n"
+	"		\n"
+	"	if (shapeTypeB!=SHAPE_CONVEX_HULL &&\n"
+	"		shapeTypeB!=SHAPE_SPHERE	&&\n"
+	"		shapeTypeB!=SHAPE_COMPOUND_OF_CONVEX_HULLS\n"
+	"		)\n"
+	"		return;\n"
+	"	b3BvhInfo bvhInfo = bvhInfos[collidables[collidableIndexA].m_numChildShapes];\n"
+	"	float4 bvhAabbMin = bvhInfo.m_aabbMin;\n"
+	"	float4 bvhAabbMax = bvhInfo.m_aabbMax;\n"
+	"	float4 bvhQuantization = bvhInfo.m_quantization;\n"
+	"	int numSubtreeHeaders = bvhInfo.m_numSubTrees;\n"
+	"	__global const btBvhSubtreeInfo* subtreeHeaders = &subtreeHeadersRoot[bvhInfo.m_subTreeOffset];\n"
+	"	__global const btQuantizedBvhNode* quantizedNodes = &quantizedNodesRoot[bvhInfo.m_nodeOffset];\n"
+	"	\n"
+	"	unsigned short int quantizedQueryAabbMin[3];\n"
+	"	unsigned short int quantizedQueryAabbMax[3];\n"
+	"	quantizeWithClamp(quantizedQueryAabbMin,aabbs[bodyIndexB].m_min,false,bvhAabbMin, bvhAabbMax,bvhQuantization);\n"
+	"	quantizeWithClamp(quantizedQueryAabbMax,aabbs[bodyIndexB].m_max,true ,bvhAabbMin, bvhAabbMax,bvhQuantization);\n"
+	"	\n"
+	"	for (int i=0;i<numSubtreeHeaders;i++)\n"
+	"	{\n"
+	"		btBvhSubtreeInfo subtree = subtreeHeaders[i];\n"
+	"				\n"
+	"		int overlap = testQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin,quantizedQueryAabbMax,subtree.m_quantizedAabbMin,subtree.m_quantizedAabbMax);\n"
+	"		if (overlap != 0)\n"
+	"		{\n"
+	"			int startNodeIndex = subtree.m_rootNodeIndex;\n"
+	"			int endNodeIndex = subtree.m_rootNodeIndex+subtree.m_subtreeSize;\n"
+	"			int curIndex = startNodeIndex;\n"
+	"			int escapeIndex;\n"
+	"			int isLeafNode;\n"
+	"			int aabbOverlap;\n"
+	"			while (curIndex < endNodeIndex)\n"
+	"			{\n"
+	"				btQuantizedBvhNode rootNode = quantizedNodes[curIndex];\n"
+	"				aabbOverlap = testQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin,quantizedQueryAabbMax,rootNode.m_quantizedAabbMin,rootNode.m_quantizedAabbMax);\n"
+	"				isLeafNode = isLeaf(&rootNode);\n"
+	"				if (aabbOverlap)\n"
+	"				{\n"
+	"					if (isLeafNode)\n"
+	"					{\n"
+	"						int triangleIndex = getTriangleIndex(&rootNode);\n"
+	"						if (shapeTypeB==SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
+	"						{\n"
+	"								int numChildrenB = collidables[collidableIndexB].m_numChildShapes;\n"
+	"								int pairIdx = atomic_add(numConcavePairsOut,numChildrenB);\n"
+	"								for (int b=0;b<numChildrenB;b++)\n"
+	"								{\n"
+	"									if ((pairIdx+b)<maxNumConcavePairsCapacity)\n"
+	"									{\n"
+	"										int childShapeIndexB = collidables[collidableIndexB].m_shapeIndex+b;\n"
+	"										int4 newPair = (int4)(bodyIndexA,bodyIndexB,triangleIndex,childShapeIndexB);\n"
+	"										concavePairsOut[pairIdx+b] = newPair;\n"
+	"									}\n"
+	"								}\n"
+	"						} else\n"
+	"						{\n"
+	"							int pairIdx = atomic_inc(numConcavePairsOut);\n"
+	"							if (pairIdx<maxNumConcavePairsCapacity)\n"
+	"							{\n"
+	"								int4 newPair = (int4)(bodyIndexA,bodyIndexB,triangleIndex,0);\n"
+	"								concavePairsOut[pairIdx] = newPair;\n"
+	"							}\n"
+	"						}\n"
+	"					} \n"
+	"					curIndex++;\n"
+	"				} else\n"
+	"				{\n"
+	"					if (isLeafNode)\n"
+	"					{\n"
+	"						curIndex++;\n"
+	"					} else\n"
+	"					{\n"
+	"						escapeIndex = getEscapeIndex(&rootNode);\n"
+	"						curIndex += escapeIndex;\n"
+	"					}\n"
+	"				}\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/mprKernels.h b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/mprKernels.h
index 7ed4b382c3d..74959a931cd 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/mprKernels.h
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/mprKernels.h
@@ -1,1446 +1,1445 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* mprKernelsCL= \
-"/***\n"
-" * ---------------------------------\n"
-" * Copyright (c)2012 Daniel Fiser <danfis@danfis.cz>\n"
-" *\n"
-" *  This file was ported from mpr.c file, part of libccd.\n"
-" *  The Minkoski Portal Refinement implementation was ported \n"
-" *  to OpenCL by Erwin Coumans for the Bullet 3 Physics library.\n"
-" *  at http://github.com/erwincoumans/bullet3\n"
-" *\n"
-" *  Distributed under the OSI-approved BSD License (the \"License\");\n"
-" *  see <http://www.opensource.org/licenses/bsd-license.php>.\n"
-" *  This software is distributed WITHOUT ANY WARRANTY; without even the\n"
-" *  implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.\n"
-" *  See the License for more information.\n"
-" */\n"
-"#ifndef B3_MPR_PENETRATION_H\n"
-"#define B3_MPR_PENETRATION_H\n"
-"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
-"#define B3_PLATFORM_DEFINITIONS_H\n"
-"struct MyTest\n"
-"{\n"
-"	int bla;\n"
-"};\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
-"#define B3_LARGE_FLOAT 1e18f\n"
-"#define B3_INFINITY 1e18f\n"
-"#define b3Assert(a)\n"
-"#define b3ConstArray(a) __global const a*\n"
-"#define b3AtomicInc atomic_inc\n"
-"#define b3AtomicAdd atomic_add\n"
-"#define b3Fabs fabs\n"
-"#define b3Sqrt native_sqrt\n"
-"#define b3Sin native_sin\n"
-"#define b3Cos native_cos\n"
-"#define B3_STATIC\n"
-"#endif\n"
-"#endif\n"
-"#ifndef B3_FLOAT4_H\n"
-"#define B3_FLOAT4_H\n"
-"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif\n"
-"#endif\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"	typedef float4	b3Float4;\n"
-"	#define b3Float4ConstArg const b3Float4\n"
-"	#define b3MakeFloat4 (float4)\n"
-"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return dot(a1, b1);\n"
-"	}\n"
-"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return cross(a1, b1);\n"
-"	}\n"
-"	#define b3MinFloat4 min\n"
-"	#define b3MaxFloat4 max\n"
-"	#define b3Normalized(a) normalize(a)\n"
-"#endif \n"
-"		\n"
-"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
-"{\n"
-"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
-"		return false;\n"
-"	return true;\n"
-"}\n"
-"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
-"{\n"
-"    float maxDot = -B3_INFINITY;\n"
-"    int i = 0;\n"
-"    int ptIndex = -1;\n"
-"    for( i = 0; i < vecLen; i++ )\n"
-"    {\n"
-"        float dot = b3Dot3F4(vecArray[i],vec);\n"
-"            \n"
-"        if( dot > maxDot )\n"
-"        {\n"
-"            maxDot = dot;\n"
-"            ptIndex = i;\n"
-"        }\n"
-"    }\n"
-"	b3Assert(ptIndex>=0);\n"
-"    if (ptIndex<0)\n"
-"	{\n"
-"		ptIndex = 0;\n"
-"	}\n"
-"    *dotOut = maxDot;\n"
-"    return ptIndex;\n"
-"}\n"
-"#endif //B3_FLOAT4_H\n"
-"#ifndef B3_RIGIDBODY_DATA_H\n"
-"#define B3_RIGIDBODY_DATA_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_FLOAT4_H\n"
-"#ifndef B3_QUAT_H\n"
-"#define B3_QUAT_H\n"
-"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif\n"
-"#endif\n"
-"#ifndef B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"	typedef float4	b3Quat;\n"
-"	#define b3QuatConstArg const b3Quat\n"
-"	\n"
-"	\n"
-"inline float4 b3FastNormalize4(float4 v)\n"
-"{\n"
-"	v = (float4)(v.xyz,0.f);\n"
-"	return fast_normalize(v);\n"
-"}\n"
-"	\n"
-"inline b3Quat b3QuatMul(b3Quat a, b3Quat b);\n"
-"inline b3Quat b3QuatNormalized(b3QuatConstArg in);\n"
-"inline b3Quat b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec);\n"
-"inline b3Quat b3QuatInvert(b3QuatConstArg q);\n"
-"inline b3Quat b3QuatInverse(b3QuatConstArg q);\n"
-"inline b3Quat b3QuatMul(b3QuatConstArg a, b3QuatConstArg b)\n"
-"{\n"
-"	b3Quat ans;\n"
-"	ans = b3Cross3( a, b );\n"
-"	ans += a.w*b+b.w*a;\n"
-"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
-"	ans.w = a.w*b.w - b3Dot3F4(a, b);\n"
-"	return ans;\n"
-"}\n"
-"inline b3Quat b3QuatNormalized(b3QuatConstArg in)\n"
-"{\n"
-"	b3Quat q;\n"
-"	q=in;\n"
-"	//return b3FastNormalize4(in);\n"
-"	float len = native_sqrt(dot(q, q));\n"
-"	if(len > 0.f)\n"
-"	{\n"
-"		q *= 1.f / len;\n"
-"	}\n"
-"	else\n"
-"	{\n"
-"		q.x = q.y = q.z = 0.f;\n"
-"		q.w = 1.f;\n"
-"	}\n"
-"	return q;\n"
-"}\n"
-"inline float4 b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
-"{\n"
-"	b3Quat qInv = b3QuatInvert( q );\n"
-"	float4 vcpy = vec;\n"
-"	vcpy.w = 0.f;\n"
-"	float4 out = b3QuatMul(b3QuatMul(q,vcpy),qInv);\n"
-"	return out;\n"
-"}\n"
-"inline b3Quat b3QuatInverse(b3QuatConstArg q)\n"
-"{\n"
-"	return (b3Quat)(-q.xyz, q.w);\n"
-"}\n"
-"inline b3Quat b3QuatInvert(b3QuatConstArg q)\n"
-"{\n"
-"	return (b3Quat)(-q.xyz, q.w);\n"
-"}\n"
-"inline float4 b3QuatInvRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
-"{\n"
-"	return b3QuatRotate( b3QuatInvert( q ), vec );\n"
-"}\n"
-"inline b3Float4 b3TransformPoint(b3Float4ConstArg point, b3Float4ConstArg translation, b3QuatConstArg  orientation)\n"
-"{\n"
-"	return b3QuatRotate( orientation, point ) + (translation);\n"
-"}\n"
-"	\n"
-"#endif \n"
-"#endif //B3_QUAT_H\n"
-"#ifndef B3_MAT3x3_H\n"
-"#define B3_MAT3x3_H\n"
-"#ifndef B3_QUAT_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_QUAT_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"typedef struct\n"
-"{\n"
-"	b3Float4 m_row[3];\n"
-"}b3Mat3x3;\n"
-"#define b3Mat3x3ConstArg const b3Mat3x3\n"
-"#define b3GetRow(m,row) (m.m_row[row])\n"
-"inline b3Mat3x3 b3QuatGetRotationMatrix(b3Quat quat)\n"
-"{\n"
-"	b3Float4 quat2 = (b3Float4)(quat.x*quat.x, quat.y*quat.y, quat.z*quat.z, 0.f);\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0].x=1-2*quat2.y-2*quat2.z;\n"
-"	out.m_row[0].y=2*quat.x*quat.y-2*quat.w*quat.z;\n"
-"	out.m_row[0].z=2*quat.x*quat.z+2*quat.w*quat.y;\n"
-"	out.m_row[0].w = 0.f;\n"
-"	out.m_row[1].x=2*quat.x*quat.y+2*quat.w*quat.z;\n"
-"	out.m_row[1].y=1-2*quat2.x-2*quat2.z;\n"
-"	out.m_row[1].z=2*quat.y*quat.z-2*quat.w*quat.x;\n"
-"	out.m_row[1].w = 0.f;\n"
-"	out.m_row[2].x=2*quat.x*quat.z-2*quat.w*quat.y;\n"
-"	out.m_row[2].y=2*quat.y*quat.z+2*quat.w*quat.x;\n"
-"	out.m_row[2].z=1-2*quat2.x-2*quat2.y;\n"
-"	out.m_row[2].w = 0.f;\n"
-"	return out;\n"
-"}\n"
-"inline b3Mat3x3 b3AbsoluteMat3x3(b3Mat3x3ConstArg matIn)\n"
-"{\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0] = fabs(matIn.m_row[0]);\n"
-"	out.m_row[1] = fabs(matIn.m_row[1]);\n"
-"	out.m_row[2] = fabs(matIn.m_row[2]);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtZero();\n"
-"__inline\n"
-"b3Mat3x3 mtIdentity();\n"
-"__inline\n"
-"b3Mat3x3 mtTranspose(b3Mat3x3 m);\n"
-"__inline\n"
-"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b);\n"
-"__inline\n"
-"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b);\n"
-"__inline\n"
-"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b);\n"
-"__inline\n"
-"b3Mat3x3 mtZero()\n"
-"{\n"
-"	b3Mat3x3 m;\n"
-"	m.m_row[0] = (b3Float4)(0.f);\n"
-"	m.m_row[1] = (b3Float4)(0.f);\n"
-"	m.m_row[2] = (b3Float4)(0.f);\n"
-"	return m;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtIdentity()\n"
-"{\n"
-"	b3Mat3x3 m;\n"
-"	m.m_row[0] = (b3Float4)(1,0,0,0);\n"
-"	m.m_row[1] = (b3Float4)(0,1,0,0);\n"
-"	m.m_row[2] = (b3Float4)(0,0,1,0);\n"
-"	return m;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtTranspose(b3Mat3x3 m)\n"
-"{\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0] = (b3Float4)(m.m_row[0].x, m.m_row[1].x, m.m_row[2].x, 0.f);\n"
-"	out.m_row[1] = (b3Float4)(m.m_row[0].y, m.m_row[1].y, m.m_row[2].y, 0.f);\n"
-"	out.m_row[2] = (b3Float4)(m.m_row[0].z, m.m_row[1].z, m.m_row[2].z, 0.f);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b)\n"
-"{\n"
-"	b3Mat3x3 transB;\n"
-"	transB = mtTranspose( b );\n"
-"	b3Mat3x3 ans;\n"
-"	//	why this doesn't run when 0ing in the for{}\n"
-"	a.m_row[0].w = 0.f;\n"
-"	a.m_row[1].w = 0.f;\n"
-"	a.m_row[2].w = 0.f;\n"
-"	for(int i=0; i<3; i++)\n"
-"	{\n"
-"//	a.m_row[i].w = 0.f;\n"
-"		ans.m_row[i].x = b3Dot3F4(a.m_row[i],transB.m_row[0]);\n"
-"		ans.m_row[i].y = b3Dot3F4(a.m_row[i],transB.m_row[1]);\n"
-"		ans.m_row[i].z = b3Dot3F4(a.m_row[i],transB.m_row[2]);\n"
-"		ans.m_row[i].w = 0.f;\n"
-"	}\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b)\n"
-"{\n"
-"	b3Float4 ans;\n"
-"	ans.x = b3Dot3F4( a.m_row[0], b );\n"
-"	ans.y = b3Dot3F4( a.m_row[1], b );\n"
-"	ans.z = b3Dot3F4( a.m_row[2], b );\n"
-"	ans.w = 0.f;\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b)\n"
-"{\n"
-"	b3Float4 colx = b3MakeFloat4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
-"	b3Float4 coly = b3MakeFloat4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
-"	b3Float4 colz = b3MakeFloat4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
-"	b3Float4 ans;\n"
-"	ans.x = b3Dot3F4( a, colx );\n"
-"	ans.y = b3Dot3F4( a, coly );\n"
-"	ans.z = b3Dot3F4( a, colz );\n"
-"	return ans;\n"
-"}\n"
-"#endif\n"
-"#endif //B3_MAT3x3_H\n"
-"typedef struct b3RigidBodyData b3RigidBodyData_t;\n"
-"struct b3RigidBodyData\n"
-"{\n"
-"	b3Float4				m_pos;\n"
-"	b3Quat					m_quat;\n"
-"	b3Float4				m_linVel;\n"
-"	b3Float4				m_angVel;\n"
-"	int 					m_collidableIdx;\n"
-"	float 				m_invMass;\n"
-"	float 				m_restituitionCoeff;\n"
-"	float 				m_frictionCoeff;\n"
-"};\n"
-"typedef struct b3InertiaData b3InertiaData_t;\n"
-"struct b3InertiaData\n"
-"{\n"
-"	b3Mat3x3 m_invInertiaWorld;\n"
-"	b3Mat3x3 m_initInvInertia;\n"
-"};\n"
-"#endif //B3_RIGIDBODY_DATA_H\n"
-"	\n"
-"#ifndef B3_CONVEX_POLYHEDRON_DATA_H\n"
-"#define B3_CONVEX_POLYHEDRON_DATA_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_FLOAT4_H\n"
-"#ifndef B3_QUAT_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_QUAT_H\n"
-"typedef struct b3GpuFace b3GpuFace_t;\n"
-"struct b3GpuFace\n"
-"{\n"
-"	b3Float4 m_plane;\n"
-"	int m_indexOffset;\n"
-"	int m_numIndices;\n"
-"	int m_unusedPadding1;\n"
-"	int m_unusedPadding2;\n"
-"};\n"
-"typedef struct b3ConvexPolyhedronData b3ConvexPolyhedronData_t;\n"
-"struct b3ConvexPolyhedronData\n"
-"{\n"
-"	b3Float4		m_localCenter;\n"
-"	b3Float4		m_extents;\n"
-"	b3Float4		mC;\n"
-"	b3Float4		mE;\n"
-"	float			m_radius;\n"
-"	int	m_faceOffset;\n"
-"	int m_numFaces;\n"
-"	int	m_numVertices;\n"
-"	int m_vertexOffset;\n"
-"	int	m_uniqueEdgesOffset;\n"
-"	int	m_numUniqueEdges;\n"
-"	int m_unused;\n"
-"};\n"
-"#endif //B3_CONVEX_POLYHEDRON_DATA_H\n"
-"#ifndef B3_COLLIDABLE_H\n"
-"#define B3_COLLIDABLE_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_FLOAT4_H\n"
-"#ifndef B3_QUAT_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_QUAT_H\n"
-"enum b3ShapeTypes\n"
-"{\n"
-"	SHAPE_HEIGHT_FIELD=1,\n"
-"	SHAPE_CONVEX_HULL=3,\n"
-"	SHAPE_PLANE=4,\n"
-"	SHAPE_CONCAVE_TRIMESH=5,\n"
-"	SHAPE_COMPOUND_OF_CONVEX_HULLS=6,\n"
-"	SHAPE_SPHERE=7,\n"
-"	MAX_NUM_SHAPE_TYPES,\n"
-"};\n"
-"typedef struct b3Collidable b3Collidable_t;\n"
-"struct b3Collidable\n"
-"{\n"
-"	union {\n"
-"		int m_numChildShapes;\n"
-"		int m_bvhIndex;\n"
-"	};\n"
-"	union\n"
-"	{\n"
-"		float m_radius;\n"
-"		int	m_compoundBvhIndex;\n"
-"	};\n"
-"	int m_shapeType;\n"
-"	int m_shapeIndex;\n"
-"};\n"
-"typedef struct b3GpuChildShape b3GpuChildShape_t;\n"
-"struct b3GpuChildShape\n"
-"{\n"
-"	b3Float4	m_childPosition;\n"
-"	b3Quat		m_childOrientation;\n"
-"	int m_shapeIndex;\n"
-"	int m_unused0;\n"
-"	int m_unused1;\n"
-"	int m_unused2;\n"
-"};\n"
-"struct b3CompoundOverlappingPair\n"
-"{\n"
-"	int m_bodyIndexA;\n"
-"	int m_bodyIndexB;\n"
-"//	int	m_pairType;\n"
-"	int m_childShapeIndexA;\n"
-"	int m_childShapeIndexB;\n"
-"};\n"
-"#endif //B3_COLLIDABLE_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#define B3_MPR_SQRT sqrt\n"
-"#endif\n"
-"#define B3_MPR_FMIN(x, y) ((x) < (y) ? (x) : (y))\n"
-"#define B3_MPR_FABS fabs\n"
-"#define B3_MPR_TOLERANCE 1E-6f\n"
-"#define B3_MPR_MAX_ITERATIONS 1000\n"
-"struct _b3MprSupport_t \n"
-"{\n"
-"    b3Float4 v;  //!< Support point in minkowski sum\n"
-"    b3Float4 v1; //!< Support point in obj1\n"
-"    b3Float4 v2; //!< Support point in obj2\n"
-"};\n"
-"typedef struct _b3MprSupport_t b3MprSupport_t;\n"
-"struct _b3MprSimplex_t \n"
-"{\n"
-"    b3MprSupport_t ps[4];\n"
-"    int last; //!< index of last added point\n"
-"};\n"
-"typedef struct _b3MprSimplex_t b3MprSimplex_t;\n"
-"inline b3MprSupport_t* b3MprSimplexPointW(b3MprSimplex_t *s, int idx)\n"
-"{\n"
-"    return &s->ps[idx];\n"
-"}\n"
-"inline void b3MprSimplexSetSize(b3MprSimplex_t *s, int size)\n"
-"{\n"
-"    s->last = size - 1;\n"
-"}\n"
-"inline int b3MprSimplexSize(const b3MprSimplex_t *s)\n"
-"{\n"
-"    return s->last + 1;\n"
-"}\n"
-"inline const b3MprSupport_t* b3MprSimplexPoint(const b3MprSimplex_t* s, int idx)\n"
-"{\n"
-"    // here is no check on boundaries\n"
-"    return &s->ps[idx];\n"
-"}\n"
-"inline void b3MprSupportCopy(b3MprSupport_t *d, const b3MprSupport_t *s)\n"
-"{\n"
-"    *d = *s;\n"
-"}\n"
-"inline void b3MprSimplexSet(b3MprSimplex_t *s, size_t pos, const b3MprSupport_t *a)\n"
-"{\n"
-"    b3MprSupportCopy(s->ps + pos, a);\n"
-"}\n"
-"inline void b3MprSimplexSwap(b3MprSimplex_t *s, size_t pos1, size_t pos2)\n"
-"{\n"
-"    b3MprSupport_t supp;\n"
-"    b3MprSupportCopy(&supp, &s->ps[pos1]);\n"
-"    b3MprSupportCopy(&s->ps[pos1], &s->ps[pos2]);\n"
-"    b3MprSupportCopy(&s->ps[pos2], &supp);\n"
-"}\n"
-"inline int b3MprIsZero(float val)\n"
-"{\n"
-"    return B3_MPR_FABS(val) < FLT_EPSILON;\n"
-"}\n"
-"inline int b3MprEq(float _a, float _b)\n"
-"{\n"
-"    float ab;\n"
-"    float a, b;\n"
-"    ab = B3_MPR_FABS(_a - _b);\n"
-"    if (B3_MPR_FABS(ab) < FLT_EPSILON)\n"
-"        return 1;\n"
-"    a = B3_MPR_FABS(_a);\n"
-"    b = B3_MPR_FABS(_b);\n"
-"    if (b > a){\n"
-"        return ab < FLT_EPSILON * b;\n"
-"    }else{\n"
-"        return ab < FLT_EPSILON * a;\n"
-"    }\n"
-"}\n"
-"inline int b3MprVec3Eq(const b3Float4* a, const b3Float4 *b)\n"
-"{\n"
-"    return b3MprEq((*a).x, (*b).x)\n"
-"            && b3MprEq((*a).y, (*b).y)\n"
-"            && b3MprEq((*a).z, (*b).z);\n"
-"}\n"
-"inline b3Float4 b3LocalGetSupportVertex(b3Float4ConstArg supportVec,__global const b3ConvexPolyhedronData_t* hull, 	b3ConstArray(b3Float4) verticesA)\n"
-"{\n"
-"	b3Float4 supVec = b3MakeFloat4(0,0,0,0);\n"
-"	float maxDot = -B3_LARGE_FLOAT;\n"
-"    if( 0 < hull->m_numVertices )\n"
-"    {\n"
-"        const b3Float4 scaled = supportVec;\n"
-"		int index = b3MaxDot(scaled, &verticesA[hull->m_vertexOffset], hull->m_numVertices, &maxDot);\n"
-"        return verticesA[hull->m_vertexOffset+index];\n"
-"    }\n"
-"    return supVec;\n"
-"}\n"
-"B3_STATIC void b3MprConvexSupport(int pairIndex,int bodyIndex,  b3ConstArray(b3RigidBodyData_t) cpuBodyBuf, \n"
-"													b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData, \n"
-"													b3ConstArray(b3Collidable_t)				cpuCollidables,\n"
-"													b3ConstArray(b3Float4)					cpuVertices,\n"
-"													__global b3Float4* sepAxis,\n"
-"														const b3Float4* _dir, b3Float4* outp, int logme)\n"
-"{\n"
-"	//dir is in worldspace, move to local space\n"
-"	\n"
-"	b3Float4 pos = cpuBodyBuf[bodyIndex].m_pos;\n"
-"	b3Quat orn = cpuBodyBuf[bodyIndex].m_quat;\n"
-"	\n"
-"	b3Float4 dir = b3MakeFloat4((*_dir).x,(*_dir).y,(*_dir).z,0.f);\n"
-"	\n"
-"	const b3Float4 localDir = b3QuatRotate(b3QuatInverse(orn),dir);\n"
-"	\n"
-"	//find local support vertex\n"
-"	int colIndex = cpuBodyBuf[bodyIndex].m_collidableIdx;\n"
-"	\n"
-"	b3Assert(cpuCollidables[colIndex].m_shapeType==SHAPE_CONVEX_HULL);\n"
-"	__global const b3ConvexPolyhedronData_t* hull = &cpuConvexData[cpuCollidables[colIndex].m_shapeIndex];\n"
-"	\n"
-"	b3Float4 pInA;\n"
-"	if (logme)\n"
-"	{\n"
-"		b3Float4 supVec = b3MakeFloat4(0,0,0,0);\n"
-"		float maxDot = -B3_LARGE_FLOAT;\n"
-"		if( 0 < hull->m_numVertices )\n"
-"		{\n"
-"			const b3Float4 scaled = localDir;\n"
-"			int index = b3MaxDot(scaled, &cpuVertices[hull->m_vertexOffset], hull->m_numVertices, &maxDot);\n"
-"			pInA = cpuVertices[hull->m_vertexOffset+index];\n"
-"			\n"
-"		}\n"
-"	} else\n"
-"	{\n"
-"		pInA = b3LocalGetSupportVertex(localDir,hull,cpuVertices);\n"
-"	}\n"
-"	//move vertex to world space\n"
-"	*outp = b3TransformPoint(pInA,pos,orn);\n"
-"	\n"
-"}\n"
-"inline void b3MprSupport(int pairIndex,int bodyIndexA, int bodyIndexB,   b3ConstArray(b3RigidBodyData_t) cpuBodyBuf, \n"
-"													b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData, \n"
-"													b3ConstArray(b3Collidable_t)				cpuCollidables,\n"
-"													b3ConstArray(b3Float4)					cpuVertices,\n"
-"													__global b3Float4* sepAxis,\n"
-"													const b3Float4* _dir, b3MprSupport_t *supp)\n"
-"{\n"
-"    b3Float4 dir;\n"
-"	dir = *_dir;\n"
-"	b3MprConvexSupport(pairIndex,bodyIndexA,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices,sepAxis,&dir, &supp->v1,0);\n"
-"    dir = *_dir*-1.f;\n"
-"	b3MprConvexSupport(pairIndex,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices,sepAxis,&dir, &supp->v2,0);\n"
-"    supp->v = supp->v1 - supp->v2;\n"
-"}\n"
-"inline void b3FindOrigin(int bodyIndexA, int bodyIndexB, b3ConstArray(b3RigidBodyData_t) cpuBodyBuf, b3MprSupport_t *center)\n"
-"{\n"
-"    center->v1 = cpuBodyBuf[bodyIndexA].m_pos;\n"
-"	center->v2 = cpuBodyBuf[bodyIndexB].m_pos;\n"
-"    center->v = center->v1 - center->v2;\n"
-"}\n"
-"inline void b3MprVec3Set(b3Float4 *v, float x, float y, float z)\n"
-"{\n"
-"	(*v).x = x;\n"
-"	(*v).y = y;\n"
-"	(*v).z = z;\n"
-"	(*v).w = 0.f;\n"
-"}\n"
-"inline void b3MprVec3Add(b3Float4 *v, const b3Float4 *w)\n"
-"{\n"
-"    (*v).x += (*w).x;\n"
-"    (*v).y += (*w).y;\n"
-"    (*v).z += (*w).z;\n"
-"}\n"
-"inline void b3MprVec3Copy(b3Float4 *v, const b3Float4 *w)\n"
-"{\n"
-"    *v = *w;\n"
-"}\n"
-"inline void b3MprVec3Scale(b3Float4 *d, float k)\n"
-"{\n"
-"    *d *= k;\n"
-"}\n"
-"inline float b3MprVec3Dot(const b3Float4 *a, const b3Float4 *b)\n"
-"{\n"
-"    float dot;\n"
-"	dot = b3Dot3F4(*a,*b);\n"
-"    return dot;\n"
-"}\n"
-"inline float b3MprVec3Len2(const b3Float4 *v)\n"
-"{\n"
-"    return b3MprVec3Dot(v, v);\n"
-"}\n"
-"inline void b3MprVec3Normalize(b3Float4 *d)\n"
-"{\n"
-"    float k = 1.f / B3_MPR_SQRT(b3MprVec3Len2(d));\n"
-"    b3MprVec3Scale(d, k);\n"
-"}\n"
-"inline void b3MprVec3Cross(b3Float4 *d, const b3Float4 *a, const b3Float4 *b)\n"
-"{\n"
-"	*d = b3Cross3(*a,*b);\n"
-"	\n"
-"}\n"
-"inline void b3MprVec3Sub2(b3Float4 *d, const b3Float4 *v, const b3Float4 *w)\n"
-"{\n"
-"	*d = *v - *w;\n"
-"}\n"
-"inline void b3PortalDir(const b3MprSimplex_t *portal, b3Float4 *dir)\n"
-"{\n"
-"    b3Float4 v2v1, v3v1;\n"
-"    b3MprVec3Sub2(&v2v1, &b3MprSimplexPoint(portal, 2)->v,\n"
-"                       &b3MprSimplexPoint(portal, 1)->v);\n"
-"    b3MprVec3Sub2(&v3v1, &b3MprSimplexPoint(portal, 3)->v,\n"
-"                       &b3MprSimplexPoint(portal, 1)->v);\n"
-"    b3MprVec3Cross(dir, &v2v1, &v3v1);\n"
-"    b3MprVec3Normalize(dir);\n"
-"}\n"
-"inline int portalEncapsulesOrigin(const b3MprSimplex_t *portal,\n"
-"                                       const b3Float4 *dir)\n"
-"{\n"
-"    float dot;\n"
-"    dot = b3MprVec3Dot(dir, &b3MprSimplexPoint(portal, 1)->v);\n"
-"    return b3MprIsZero(dot) || dot > 0.f;\n"
-"}\n"
-"inline int portalReachTolerance(const b3MprSimplex_t *portal,\n"
-"                                     const b3MprSupport_t *v4,\n"
-"                                     const b3Float4 *dir)\n"
-"{\n"
-"    float dv1, dv2, dv3, dv4;\n"
-"    float dot1, dot2, dot3;\n"
-"    // find the smallest dot product of dir and {v1-v4, v2-v4, v3-v4}\n"
-"    dv1 = b3MprVec3Dot(&b3MprSimplexPoint(portal, 1)->v, dir);\n"
-"    dv2 = b3MprVec3Dot(&b3MprSimplexPoint(portal, 2)->v, dir);\n"
-"    dv3 = b3MprVec3Dot(&b3MprSimplexPoint(portal, 3)->v, dir);\n"
-"    dv4 = b3MprVec3Dot(&v4->v, dir);\n"
-"    dot1 = dv4 - dv1;\n"
-"    dot2 = dv4 - dv2;\n"
-"    dot3 = dv4 - dv3;\n"
-"    dot1 = B3_MPR_FMIN(dot1, dot2);\n"
-"    dot1 = B3_MPR_FMIN(dot1, dot3);\n"
-"    return b3MprEq(dot1, B3_MPR_TOLERANCE) || dot1 < B3_MPR_TOLERANCE;\n"
-"}\n"
-"inline int portalCanEncapsuleOrigin(const b3MprSimplex_t *portal,   \n"
-"                                         const b3MprSupport_t *v4,\n"
-"                                         const b3Float4 *dir)\n"
-"{\n"
-"    float dot;\n"
-"    dot = b3MprVec3Dot(&v4->v, dir);\n"
-"    return b3MprIsZero(dot) || dot > 0.f;\n"
-"}\n"
-"inline void b3ExpandPortal(b3MprSimplex_t *portal,\n"
-"                              const b3MprSupport_t *v4)\n"
-"{\n"
-"    float dot;\n"
-"    b3Float4 v4v0;\n"
-"    b3MprVec3Cross(&v4v0, &v4->v, &b3MprSimplexPoint(portal, 0)->v);\n"
-"    dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 1)->v, &v4v0);\n"
-"    if (dot > 0.f){\n"
-"        dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 2)->v, &v4v0);\n"
-"        if (dot > 0.f){\n"
-"            b3MprSimplexSet(portal, 1, v4);\n"
-"        }else{\n"
-"            b3MprSimplexSet(portal, 3, v4);\n"
-"        }\n"
-"    }else{\n"
-"        dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 3)->v, &v4v0);\n"
-"        if (dot > 0.f){\n"
-"            b3MprSimplexSet(portal, 2, v4);\n"
-"        }else{\n"
-"            b3MprSimplexSet(portal, 1, v4);\n"
-"        }\n"
-"    }\n"
-"}\n"
-"B3_STATIC int b3DiscoverPortal(int pairIndex, int bodyIndexA, int bodyIndexB,  b3ConstArray(b3RigidBodyData_t) cpuBodyBuf, \n"
-"													b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData, \n"
-"													b3ConstArray(b3Collidable_t)				cpuCollidables,\n"
-"													b3ConstArray(b3Float4)					cpuVertices,\n"
-"													__global b3Float4* sepAxis,\n"
-"													__global int*	hasSepAxis,\n"
-"													b3MprSimplex_t *portal)\n"
-"{\n"
-"    b3Float4 dir, va, vb;\n"
-"    float dot;\n"
-"    int cont;\n"
-"	\n"
-"	\n"
-"    // vertex 0 is center of portal\n"
-"    b3FindOrigin(bodyIndexA,bodyIndexB,cpuBodyBuf, b3MprSimplexPointW(portal, 0));\n"
-"    // vertex 0 is center of portal\n"
-"    b3MprSimplexSetSize(portal, 1);\n"
-"	\n"
-"	b3Float4 zero = b3MakeFloat4(0,0,0,0);\n"
-"	b3Float4* b3mpr_vec3_origin = &zero;\n"
-"    if (b3MprVec3Eq(&b3MprSimplexPoint(portal, 0)->v, b3mpr_vec3_origin)){\n"
-"        // Portal's center lies on origin (0,0,0) => we know that objects\n"
-"        // intersect but we would need to know penetration info.\n"
-"        // So move center little bit...\n"
-"        b3MprVec3Set(&va, FLT_EPSILON * 10.f, 0.f, 0.f);\n"
-"        b3MprVec3Add(&b3MprSimplexPointW(portal, 0)->v, &va);\n"
-"    }\n"
-"    // vertex 1 = support in direction of origin\n"
-"    b3MprVec3Copy(&dir, &b3MprSimplexPoint(portal, 0)->v);\n"
-"    b3MprVec3Scale(&dir, -1.f);\n"
-"    b3MprVec3Normalize(&dir);\n"
-"    b3MprSupport(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&dir, b3MprSimplexPointW(portal, 1));\n"
-"    b3MprSimplexSetSize(portal, 2);\n"
-"    // test if origin isn't outside of v1\n"
-"    dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 1)->v, &dir);\n"
-"	\n"
-"    if (b3MprIsZero(dot) || dot < 0.f)\n"
-"        return -1;\n"
-"    // vertex 2\n"
-"    b3MprVec3Cross(&dir, &b3MprSimplexPoint(portal, 0)->v,\n"
-"                       &b3MprSimplexPoint(portal, 1)->v);\n"
-"    if (b3MprIsZero(b3MprVec3Len2(&dir))){\n"
-"        if (b3MprVec3Eq(&b3MprSimplexPoint(portal, 1)->v, b3mpr_vec3_origin)){\n"
-"            // origin lies on v1\n"
-"            return 1;\n"
-"        }else{\n"
-"            // origin lies on v0-v1 segment\n"
-"            return 2;\n"
-"        }\n"
-"    }\n"
-"    b3MprVec3Normalize(&dir);\n"
-"	 b3MprSupport(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&dir, b3MprSimplexPointW(portal, 2));\n"
-"    \n"
-"    dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 2)->v, &dir);\n"
-"    if (b3MprIsZero(dot) || dot < 0.f)\n"
-"        return -1;\n"
-"    b3MprSimplexSetSize(portal, 3);\n"
-"    // vertex 3 direction\n"
-"    b3MprVec3Sub2(&va, &b3MprSimplexPoint(portal, 1)->v,\n"
-"                     &b3MprSimplexPoint(portal, 0)->v);\n"
-"    b3MprVec3Sub2(&vb, &b3MprSimplexPoint(portal, 2)->v,\n"
-"                     &b3MprSimplexPoint(portal, 0)->v);\n"
-"    b3MprVec3Cross(&dir, &va, &vb);\n"
-"    b3MprVec3Normalize(&dir);\n"
-"    // it is better to form portal faces to be oriented \"outside\" origin\n"
-"    dot = b3MprVec3Dot(&dir, &b3MprSimplexPoint(portal, 0)->v);\n"
-"    if (dot > 0.f){\n"
-"        b3MprSimplexSwap(portal, 1, 2);\n"
-"        b3MprVec3Scale(&dir, -1.f);\n"
-"    }\n"
-"    while (b3MprSimplexSize(portal) < 4){\n"
-"		 b3MprSupport(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&dir, b3MprSimplexPointW(portal, 3));\n"
-"        \n"
-"        dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 3)->v, &dir);\n"
-"        if (b3MprIsZero(dot) || dot < 0.f)\n"
-"            return -1;\n"
-"        cont = 0;\n"
-"        // test if origin is outside (v1, v0, v3) - set v2 as v3 and\n"
-"        // continue\n"
-"        b3MprVec3Cross(&va, &b3MprSimplexPoint(portal, 1)->v,\n"
-"                          &b3MprSimplexPoint(portal, 3)->v);\n"
-"        dot = b3MprVec3Dot(&va, &b3MprSimplexPoint(portal, 0)->v);\n"
-"        if (dot < 0.f && !b3MprIsZero(dot)){\n"
-"            b3MprSimplexSet(portal, 2, b3MprSimplexPoint(portal, 3));\n"
-"            cont = 1;\n"
-"        }\n"
-"        if (!cont){\n"
-"            // test if origin is outside (v3, v0, v2) - set v1 as v3 and\n"
-"            // continue\n"
-"            b3MprVec3Cross(&va, &b3MprSimplexPoint(portal, 3)->v,\n"
-"                              &b3MprSimplexPoint(portal, 2)->v);\n"
-"            dot = b3MprVec3Dot(&va, &b3MprSimplexPoint(portal, 0)->v);\n"
-"            if (dot < 0.f && !b3MprIsZero(dot)){\n"
-"                b3MprSimplexSet(portal, 1, b3MprSimplexPoint(portal, 3));\n"
-"                cont = 1;\n"
-"            }\n"
-"        }\n"
-"        if (cont){\n"
-"            b3MprVec3Sub2(&va, &b3MprSimplexPoint(portal, 1)->v,\n"
-"                             &b3MprSimplexPoint(portal, 0)->v);\n"
-"            b3MprVec3Sub2(&vb, &b3MprSimplexPoint(portal, 2)->v,\n"
-"                             &b3MprSimplexPoint(portal, 0)->v);\n"
-"            b3MprVec3Cross(&dir, &va, &vb);\n"
-"            b3MprVec3Normalize(&dir);\n"
-"        }else{\n"
-"            b3MprSimplexSetSize(portal, 4);\n"
-"        }\n"
-"    }\n"
-"    return 0;\n"
-"}\n"
-"B3_STATIC int b3RefinePortal(int pairIndex,int bodyIndexA, int bodyIndexB,  b3ConstArray(b3RigidBodyData_t) cpuBodyBuf, \n"
-"													b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData, \n"
-"													b3ConstArray(b3Collidable_t)				cpuCollidables,\n"
-"													b3ConstArray(b3Float4)					cpuVertices,\n"
-"													__global b3Float4* sepAxis,\n"
-"													b3MprSimplex_t *portal)\n"
-"{\n"
-"    b3Float4 dir;\n"
-"    b3MprSupport_t v4;\n"
-"	for (int i=0;i<B3_MPR_MAX_ITERATIONS;i++)\n"
-"    //while (1)\n"
-"	{\n"
-"        // compute direction outside the portal (from v0 throught v1,v2,v3\n"
-"        // face)\n"
-"        b3PortalDir(portal, &dir);\n"
-"        // test if origin is inside the portal\n"
-"        if (portalEncapsulesOrigin(portal, &dir))\n"
-"            return 0;\n"
-"        // get next support point\n"
-"        \n"
-"		 b3MprSupport(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&dir, &v4);\n"
-"        // test if v4 can expand portal to contain origin and if portal\n"
-"        // expanding doesn't reach given tolerance\n"
-"        if (!portalCanEncapsuleOrigin(portal, &v4, &dir)\n"
-"                || portalReachTolerance(portal, &v4, &dir))\n"
-"		{\n"
-"            return -1;\n"
-"        }\n"
-"        // v1-v2-v3 triangle must be rearranged to face outside Minkowski\n"
-"        // difference (direction from v0).\n"
-"        b3ExpandPortal(portal, &v4);\n"
-"    }\n"
-"    return -1;\n"
-"}\n"
-"B3_STATIC void b3FindPos(const b3MprSimplex_t *portal, b3Float4 *pos)\n"
-"{\n"
-"	b3Float4 zero = b3MakeFloat4(0,0,0,0);\n"
-"	b3Float4* b3mpr_vec3_origin = &zero;\n"
-"    b3Float4 dir;\n"
-"    size_t i;\n"
-"    float b[4], sum, inv;\n"
-"    b3Float4 vec, p1, p2;\n"
-"    b3PortalDir(portal, &dir);\n"
-"    // use barycentric coordinates of tetrahedron to find origin\n"
-"    b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 1)->v,\n"
-"                       &b3MprSimplexPoint(portal, 2)->v);\n"
-"    b[0] = b3MprVec3Dot(&vec, &b3MprSimplexPoint(portal, 3)->v);\n"
-"    b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 3)->v,\n"
-"                       &b3MprSimplexPoint(portal, 2)->v);\n"
-"    b[1] = b3MprVec3Dot(&vec, &b3MprSimplexPoint(portal, 0)->v);\n"
-"    b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 0)->v,\n"
-"                       &b3MprSimplexPoint(portal, 1)->v);\n"
-"    b[2] = b3MprVec3Dot(&vec, &b3MprSimplexPoint(portal, 3)->v);\n"
-"    b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 2)->v,\n"
-"                       &b3MprSimplexPoint(portal, 1)->v);\n"
-"    b[3] = b3MprVec3Dot(&vec, &b3MprSimplexPoint(portal, 0)->v);\n"
-"	sum = b[0] + b[1] + b[2] + b[3];\n"
-"    if (b3MprIsZero(sum) || sum < 0.f){\n"
-"		b[0] = 0.f;\n"
-"        b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 2)->v,\n"
-"                           &b3MprSimplexPoint(portal, 3)->v);\n"
-"        b[1] = b3MprVec3Dot(&vec, &dir);\n"
-"        b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 3)->v,\n"
-"                           &b3MprSimplexPoint(portal, 1)->v);\n"
-"        b[2] = b3MprVec3Dot(&vec, &dir);\n"
-"        b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 1)->v,\n"
-"                           &b3MprSimplexPoint(portal, 2)->v);\n"
-"        b[3] = b3MprVec3Dot(&vec, &dir);\n"
-"		sum = b[1] + b[2] + b[3];\n"
-"	}\n"
-"	inv = 1.f / sum;\n"
-"    b3MprVec3Copy(&p1, b3mpr_vec3_origin);\n"
-"    b3MprVec3Copy(&p2, b3mpr_vec3_origin);\n"
-"    for (i = 0; i < 4; i++){\n"
-"        b3MprVec3Copy(&vec, &b3MprSimplexPoint(portal, i)->v1);\n"
-"        b3MprVec3Scale(&vec, b[i]);\n"
-"        b3MprVec3Add(&p1, &vec);\n"
-"        b3MprVec3Copy(&vec, &b3MprSimplexPoint(portal, i)->v2);\n"
-"        b3MprVec3Scale(&vec, b[i]);\n"
-"        b3MprVec3Add(&p2, &vec);\n"
-"    }\n"
-"    b3MprVec3Scale(&p1, inv);\n"
-"    b3MprVec3Scale(&p2, inv);\n"
-"    b3MprVec3Copy(pos, &p1);\n"
-"    b3MprVec3Add(pos, &p2);\n"
-"    b3MprVec3Scale(pos, 0.5);\n"
-"}\n"
-"inline float b3MprVec3Dist2(const b3Float4 *a, const b3Float4 *b)\n"
-"{\n"
-"    b3Float4 ab;\n"
-"    b3MprVec3Sub2(&ab, a, b);\n"
-"    return b3MprVec3Len2(&ab);\n"
-"}\n"
-"inline float _b3MprVec3PointSegmentDist2(const b3Float4 *P,\n"
-"                                                  const b3Float4 *x0,\n"
-"                                                  const b3Float4 *b,\n"
-"                                                  b3Float4 *witness)\n"
-"{\n"
-"    // The computation comes from solving equation of segment:\n"
-"    //      S(t) = x0 + t.d\n"
-"    //          where - x0 is initial point of segment\n"
-"    //                - d is direction of segment from x0 (|d| > 0)\n"
-"    //                - t belongs to <0, 1> interval\n"
-"    // \n"
-"    // Than, distance from a segment to some point P can be expressed:\n"
-"    //      D(t) = |x0 + t.d - P|^2\n"
-"    //          which is distance from any point on segment. Minimization\n"
-"    //          of this function brings distance from P to segment.\n"
-"    // Minimization of D(t) leads to simple quadratic equation that's\n"
-"    // solving is straightforward.\n"
-"    //\n"
-"    // Bonus of this method is witness point for free.\n"
-"    float dist, t;\n"
-"    b3Float4 d, a;\n"
-"    // direction of segment\n"
-"    b3MprVec3Sub2(&d, b, x0);\n"
-"    // precompute vector from P to x0\n"
-"    b3MprVec3Sub2(&a, x0, P);\n"
-"    t  = -1.f * b3MprVec3Dot(&a, &d);\n"
-"    t /= b3MprVec3Len2(&d);\n"
-"    if (t < 0.f || b3MprIsZero(t)){\n"
-"        dist = b3MprVec3Dist2(x0, P);\n"
-"        if (witness)\n"
-"            b3MprVec3Copy(witness, x0);\n"
-"    }else if (t > 1.f || b3MprEq(t, 1.f)){\n"
-"        dist = b3MprVec3Dist2(b, P);\n"
-"        if (witness)\n"
-"            b3MprVec3Copy(witness, b);\n"
-"    }else{\n"
-"        if (witness){\n"
-"            b3MprVec3Copy(witness, &d);\n"
-"            b3MprVec3Scale(witness, t);\n"
-"            b3MprVec3Add(witness, x0);\n"
-"            dist = b3MprVec3Dist2(witness, P);\n"
-"        }else{\n"
-"            // recycling variables\n"
-"            b3MprVec3Scale(&d, t);\n"
-"            b3MprVec3Add(&d, &a);\n"
-"            dist = b3MprVec3Len2(&d);\n"
-"        }\n"
-"    }\n"
-"    return dist;\n"
-"}\n"
-"inline float b3MprVec3PointTriDist2(const b3Float4 *P,\n"
-"                                const b3Float4 *x0, const b3Float4 *B,\n"
-"                                const b3Float4 *C,\n"
-"                                b3Float4 *witness)\n"
-"{\n"
-"    // Computation comes from analytic expression for triangle (x0, B, C)\n"
-"    //      T(s, t) = x0 + s.d1 + t.d2, where d1 = B - x0 and d2 = C - x0 and\n"
-"    // Then equation for distance is:\n"
-"    //      D(s, t) = | T(s, t) - P |^2\n"
-"    // This leads to minimization of quadratic function of two variables.\n"
-"    // The solution from is taken only if s is between 0 and 1, t is\n"
-"    // between 0 and 1 and t + s < 1, otherwise distance from segment is\n"
-"    // computed.\n"
-"    b3Float4 d1, d2, a;\n"
-"    float u, v, w, p, q, r;\n"
-"    float s, t, dist, dist2;\n"
-"    b3Float4 witness2;\n"
-"    b3MprVec3Sub2(&d1, B, x0);\n"
-"    b3MprVec3Sub2(&d2, C, x0);\n"
-"    b3MprVec3Sub2(&a, x0, P);\n"
-"    u = b3MprVec3Dot(&a, &a);\n"
-"    v = b3MprVec3Dot(&d1, &d1);\n"
-"    w = b3MprVec3Dot(&d2, &d2);\n"
-"    p = b3MprVec3Dot(&a, &d1);\n"
-"    q = b3MprVec3Dot(&a, &d2);\n"
-"    r = b3MprVec3Dot(&d1, &d2);\n"
-"    s = (q * r - w * p) / (w * v - r * r);\n"
-"    t = (-s * r - q) / w;\n"
-"    if ((b3MprIsZero(s) || s > 0.f)\n"
-"            && (b3MprEq(s, 1.f) || s < 1.f)\n"
-"            && (b3MprIsZero(t) || t > 0.f)\n"
-"            && (b3MprEq(t, 1.f) || t < 1.f)\n"
-"            && (b3MprEq(t + s, 1.f) || t + s < 1.f)){\n"
-"        if (witness){\n"
-"            b3MprVec3Scale(&d1, s);\n"
-"            b3MprVec3Scale(&d2, t);\n"
-"            b3MprVec3Copy(witness, x0);\n"
-"            b3MprVec3Add(witness, &d1);\n"
-"            b3MprVec3Add(witness, &d2);\n"
-"            dist = b3MprVec3Dist2(witness, P);\n"
-"        }else{\n"
-"            dist  = s * s * v;\n"
-"            dist += t * t * w;\n"
-"            dist += 2.f * s * t * r;\n"
-"            dist += 2.f * s * p;\n"
-"            dist += 2.f * t * q;\n"
-"            dist += u;\n"
-"        }\n"
-"    }else{\n"
-"        dist = _b3MprVec3PointSegmentDist2(P, x0, B, witness);\n"
-"        dist2 = _b3MprVec3PointSegmentDist2(P, x0, C, &witness2);\n"
-"        if (dist2 < dist){\n"
-"            dist = dist2;\n"
-"            if (witness)\n"
-"                b3MprVec3Copy(witness, &witness2);\n"
-"        }\n"
-"        dist2 = _b3MprVec3PointSegmentDist2(P, B, C, &witness2);\n"
-"        if (dist2 < dist){\n"
-"            dist = dist2;\n"
-"            if (witness)\n"
-"                b3MprVec3Copy(witness, &witness2);\n"
-"        }\n"
-"    }\n"
-"    return dist;\n"
-"}\n"
-"B3_STATIC void b3FindPenetr(int pairIndex,int bodyIndexA, int bodyIndexB,  b3ConstArray(b3RigidBodyData_t) cpuBodyBuf, \n"
-"													b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData, \n"
-"													b3ConstArray(b3Collidable_t)				cpuCollidables,\n"
-"													b3ConstArray(b3Float4)					cpuVertices,\n"
-"													__global b3Float4* sepAxis,\n"
-"                       b3MprSimplex_t *portal,\n"
-"                       float *depth, b3Float4 *pdir, b3Float4 *pos)\n"
-"{\n"
-"    b3Float4 dir;\n"
-"    b3MprSupport_t v4;\n"
-"    unsigned long iterations;\n"
-"	b3Float4 zero = b3MakeFloat4(0,0,0,0);\n"
-"	b3Float4* b3mpr_vec3_origin = &zero;\n"
-"    iterations = 1UL;\n"
-"	for (int i=0;i<B3_MPR_MAX_ITERATIONS;i++)\n"
-"    //while (1)\n"
-"	{\n"
-"        // compute portal direction and obtain next support point\n"
-"        b3PortalDir(portal, &dir);\n"
-"        \n"
-"		 b3MprSupport(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&dir, &v4);\n"
-"        // reached tolerance -> find penetration info\n"
-"        if (portalReachTolerance(portal, &v4, &dir)\n"
-"                || iterations ==B3_MPR_MAX_ITERATIONS)\n"
-"		{\n"
-"            *depth = b3MprVec3PointTriDist2(b3mpr_vec3_origin,&b3MprSimplexPoint(portal, 1)->v,&b3MprSimplexPoint(portal, 2)->v,&b3MprSimplexPoint(portal, 3)->v,pdir);\n"
-"            *depth = B3_MPR_SQRT(*depth);\n"
-"			\n"
-"			if (b3MprIsZero((*pdir).x) && b3MprIsZero((*pdir).y) && b3MprIsZero((*pdir).z))\n"
-"			{\n"
-"				\n"
-"				*pdir = dir;\n"
-"			} \n"
-"			b3MprVec3Normalize(pdir);\n"
-"			\n"
-"            // barycentric coordinates:\n"
-"            b3FindPos(portal, pos);\n"
-"            return;\n"
-"        }\n"
-"        b3ExpandPortal(portal, &v4);\n"
-"        iterations++;\n"
-"    }\n"
-"}\n"
-"B3_STATIC void b3FindPenetrTouch(b3MprSimplex_t *portal,float *depth, b3Float4 *dir, b3Float4 *pos)\n"
-"{\n"
-"    // Touching contact on portal's v1 - so depth is zero and direction\n"
-"    // is unimportant and pos can be guessed\n"
-"    *depth = 0.f;\n"
-"    b3Float4 zero = b3MakeFloat4(0,0,0,0);\n"
-"	b3Float4* b3mpr_vec3_origin = &zero;\n"
-"	b3MprVec3Copy(dir, b3mpr_vec3_origin);\n"
-"    b3MprVec3Copy(pos, &b3MprSimplexPoint(portal, 1)->v1);\n"
-"    b3MprVec3Add(pos, &b3MprSimplexPoint(portal, 1)->v2);\n"
-"    b3MprVec3Scale(pos, 0.5);\n"
-"}\n"
-"B3_STATIC void b3FindPenetrSegment(b3MprSimplex_t *portal,\n"
-"                              float *depth, b3Float4 *dir, b3Float4 *pos)\n"
-"{\n"
-"    \n"
-"    // Origin lies on v0-v1 segment.\n"
-"    // Depth is distance to v1, direction also and position must be\n"
-"    // computed\n"
-"    b3MprVec3Copy(pos, &b3MprSimplexPoint(portal, 1)->v1);\n"
-"    b3MprVec3Add(pos, &b3MprSimplexPoint(portal, 1)->v2);\n"
-"    b3MprVec3Scale(pos, 0.5f);\n"
-"    \n"
-"    b3MprVec3Copy(dir, &b3MprSimplexPoint(portal, 1)->v);\n"
-"    *depth = B3_MPR_SQRT(b3MprVec3Len2(dir));\n"
-"    b3MprVec3Normalize(dir);\n"
-"}\n"
-"inline int b3MprPenetration(int pairIndex, int bodyIndexA, int bodyIndexB,\n"
-"					b3ConstArray(b3RigidBodyData_t) cpuBodyBuf,\n"
-"					b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData, \n"
-"					b3ConstArray(b3Collidable_t)	cpuCollidables,\n"
-"					b3ConstArray(b3Float4)	cpuVertices,\n"
-"					__global b3Float4* sepAxis,\n"
-"					__global int*	hasSepAxis,\n"
-"					float *depthOut, b3Float4* dirOut, b3Float4* posOut)\n"
-"{\n"
-"	\n"
-"	 b3MprSimplex_t portal;\n"
-"	 \n"
-"//	if (!hasSepAxis[pairIndex])\n"
-"	//	return -1;\n"
-"	\n"
-"	hasSepAxis[pairIndex] = 0;\n"
-"	 int res;\n"
-"    // Phase 1: Portal discovery\n"
-"    res = b3DiscoverPortal(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices,sepAxis,hasSepAxis, &portal);\n"
-"	\n"
-"	  \n"
-"	//sepAxis[pairIndex] = *pdir;//or -dir?\n"
-"	switch (res)\n"
-"	{\n"
-"	case 0:\n"
-"		{\n"
-"			// Phase 2: Portal refinement\n"
-"		\n"
-"			res = b3RefinePortal(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&portal);\n"
-"			if (res < 0)\n"
-"				return -1;\n"
-"			// Phase 3. Penetration info\n"
-"			b3FindPenetr(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&portal, depthOut, dirOut, posOut);\n"
-"			hasSepAxis[pairIndex] = 1;\n"
-"			sepAxis[pairIndex] = -*dirOut;\n"
-"			break;\n"
-"		}\n"
-"	case 1:\n"
-"		{\n"
-"			 // Touching contact on portal's v1.\n"
-"			b3FindPenetrTouch(&portal, depthOut, dirOut, posOut);\n"
-"			break;\n"
-"		}\n"
-"	case 2:\n"
-"		{\n"
-"			\n"
-"			b3FindPenetrSegment( &portal, depthOut, dirOut, posOut);\n"
-"			break;\n"
-"		}\n"
-"	default:\n"
-"		{\n"
-"			hasSepAxis[pairIndex]=0;\n"
-"			//if (res < 0)\n"
-"			//{\n"
-"				// Origin isn't inside portal - no collision.\n"
-"				return -1;\n"
-"			//}\n"
-"		}\n"
-"	};\n"
-"	\n"
-"	return 0;\n"
-"};\n"
-"#endif //B3_MPR_PENETRATION_H\n"
-"#ifndef B3_CONTACT4DATA_H\n"
-"#define B3_CONTACT4DATA_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_FLOAT4_H\n"
-"typedef  struct b3Contact4Data b3Contact4Data_t;\n"
-"struct b3Contact4Data\n"
-"{\n"
-"	b3Float4	m_worldPosB[4];\n"
-"//	b3Float4	m_localPosA[4];\n"
-"//	b3Float4	m_localPosB[4];\n"
-"	b3Float4	m_worldNormalOnB;	//	w: m_nPoints\n"
-"	unsigned short  m_restituitionCoeffCmp;\n"
-"	unsigned short  m_frictionCoeffCmp;\n"
-"	int m_batchIdx;\n"
-"	int m_bodyAPtrAndSignBit;//x:m_bodyAPtr, y:m_bodyBPtr\n"
-"	int m_bodyBPtrAndSignBit;\n"
-"	int	m_childIndexA;\n"
-"	int	m_childIndexB;\n"
-"	int m_unused1;\n"
-"	int m_unused2;\n"
-"};\n"
-"inline int b3Contact4Data_getNumPoints(const struct b3Contact4Data* contact)\n"
-"{\n"
-"	return (int)contact->m_worldNormalOnB.w;\n"
-"};\n"
-"inline void b3Contact4Data_setNumPoints(struct b3Contact4Data* contact, int numPoints)\n"
-"{\n"
-"	contact->m_worldNormalOnB.w = (float)numPoints;\n"
-"};\n"
-"#endif //B3_CONTACT4DATA_H\n"
-"#define AppendInc(x, out) out = atomic_inc(x)\n"
-"#define GET_NPOINTS(x) (x).m_worldNormalOnB.w\n"
-"#ifdef cl_ext_atomic_counters_32\n"
-"	#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
-"#else\n"
-"	#define counter32_t volatile __global int*\n"
-"#endif\n"
-"__kernel void   mprPenetrationKernel( __global int4* pairs,\n"
-"																					__global const b3RigidBodyData_t* rigidBodies, \n"
-"																					__global const b3Collidable_t* collidables,\n"
-"																					__global const b3ConvexPolyhedronData_t* convexShapes, \n"
-"																					__global const float4* vertices,\n"
-"																					__global float4* separatingNormals,\n"
-"																					__global int* hasSeparatingAxis,\n"
-"																					__global struct b3Contact4Data* restrict globalContactsOut,\n"
-"																					counter32_t nGlobalContactsOut,\n"
-"																					int contactCapacity,\n"
-"																					int numPairs)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	int pairIndex = i;\n"
-"	if (i<numPairs)\n"
-"	{\n"
-"		int bodyIndexA = pairs[i].x;\n"
-"		int bodyIndexB = pairs[i].y;\n"
-"		int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
-"		int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
-"	\n"
-"		int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
-"		int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
-"		\n"
-"		\n"
-"		//once the broadphase avoids static-static pairs, we can remove this test\n"
-"		if ((rigidBodies[bodyIndexA].m_invMass==0) &&(rigidBodies[bodyIndexB].m_invMass==0))\n"
-"		{\n"
-"			return;\n"
-"		}\n"
-"		\n"
-"		if ((collidables[collidableIndexA].m_shapeType!=SHAPE_CONVEX_HULL) ||(collidables[collidableIndexB].m_shapeType!=SHAPE_CONVEX_HULL))\n"
-"		{\n"
-"			return;\n"
-"		}\n"
-"		float depthOut;\n"
-"		b3Float4 dirOut;\n"
-"		b3Float4 posOut;\n"
-"		int res = b3MprPenetration(pairIndex, bodyIndexA, bodyIndexB,rigidBodies,convexShapes,collidables,vertices,separatingNormals,hasSeparatingAxis,&depthOut, &dirOut, &posOut);\n"
-"		\n"
-"		\n"
-"		\n"
-"		\n"
-"		if (res==0)\n"
-"		{\n"
-"			//add a contact\n"
-"			int dstIdx;\n"
-"			AppendInc( nGlobalContactsOut, dstIdx );\n"
-"			if (dstIdx<contactCapacity)\n"
-"			{\n"
-"				pairs[pairIndex].z = dstIdx;\n"
-"				__global struct b3Contact4Data* c = globalContactsOut + dstIdx;\n"
-"				c->m_worldNormalOnB = -dirOut;//normal;\n"
-"				c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
-"				c->m_batchIdx = pairIndex;\n"
-"				int bodyA = pairs[pairIndex].x;\n"
-"				int bodyB = pairs[pairIndex].y;\n"
-"				c->m_bodyAPtrAndSignBit = rigidBodies[bodyA].m_invMass==0 ? -bodyA:bodyA;\n"
-"				c->m_bodyBPtrAndSignBit = rigidBodies[bodyB].m_invMass==0 ? -bodyB:bodyB;\n"
-"				c->m_childIndexA = -1;\n"
-"				c->m_childIndexB = -1;\n"
-"				//for (int i=0;i<nContacts;i++)\n"
-"				posOut.w = -depthOut;\n"
-"				c->m_worldPosB[0] = posOut;//localPoints[contactIdx[i]];\n"
-"				GET_NPOINTS(*c) = 1;//nContacts;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"}\n"
-"typedef float4 Quaternion;\n"
-"#define make_float4 (float4)\n"
-"__inline\n"
-"float dot3F4(float4 a, float4 b)\n"
-"{\n"
-"	float4 a1 = make_float4(a.xyz,0.f);\n"
-"	float4 b1 = make_float4(b.xyz,0.f);\n"
-"	return dot(a1, b1);\n"
-"}\n"
-"__inline\n"
-"float4 cross3(float4 a, float4 b)\n"
-"{\n"
-"	return cross(a,b);\n"
-"}\n"
-"__inline\n"
-"Quaternion qtMul(Quaternion a, Quaternion b)\n"
-"{\n"
-"	Quaternion ans;\n"
-"	ans = cross3( a, b );\n"
-"	ans += a.w*b+b.w*a;\n"
-"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
-"	ans.w = a.w*b.w - dot3F4(a, b);\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"Quaternion qtInvert(Quaternion q)\n"
-"{\n"
-"	return (Quaternion)(-q.xyz, q.w);\n"
-"}\n"
-"__inline\n"
-"float4 qtRotate(Quaternion q, float4 vec)\n"
-"{\n"
-"	Quaternion qInv = qtInvert( q );\n"
-"	float4 vcpy = vec;\n"
-"	vcpy.w = 0.f;\n"
-"	float4 out = qtMul(qtMul(q,vcpy),qInv);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"float4 transform(const float4* p, const float4* translation, const Quaternion* orientation)\n"
-"{\n"
-"	return qtRotate( *orientation, *p ) + (*translation);\n"
-"}\n"
-"__inline\n"
-"float4 qtInvRotate(const Quaternion q, float4 vec)\n"
-"{\n"
-"	return qtRotate( qtInvert( q ), vec );\n"
-"}\n"
-"inline void project(__global const b3ConvexPolyhedronData_t* hull,  const float4 pos, const float4 orn, \n"
-"const float4* dir, __global const float4* vertices, float* min, float* max)\n"
-"{\n"
-"	min[0] = FLT_MAX;\n"
-"	max[0] = -FLT_MAX;\n"
-"	int numVerts = hull->m_numVertices;\n"
-"	const float4 localDir = qtInvRotate(orn,*dir);\n"
-"	float offset = dot(pos,*dir);\n"
-"	for(int i=0;i<numVerts;i++)\n"
-"	{\n"
-"		float dp = dot(vertices[hull->m_vertexOffset+i],localDir);\n"
-"		if(dp < min[0])	\n"
-"			min[0] = dp;\n"
-"		if(dp > max[0])	\n"
-"			max[0] = dp;\n"
-"	}\n"
-"	if(min[0]>max[0])\n"
-"	{\n"
-"		float tmp = min[0];\n"
-"		min[0] = max[0];\n"
-"		max[0] = tmp;\n"
-"	}\n"
-"	min[0] += offset;\n"
-"	max[0] += offset;\n"
-"}\n"
-"bool findSeparatingAxisUnitSphere(	__global const b3ConvexPolyhedronData_t* hullA, __global const b3ConvexPolyhedronData_t* hullB, \n"
-"	const float4 posA1,\n"
-"	const float4 ornA,\n"
-"	const float4 posB1,\n"
-"	const float4 ornB,\n"
-"	const float4 DeltaC2,\n"
-"	__global const float4* vertices,\n"
-"	__global const float4* unitSphereDirections,\n"
-"	int numUnitSphereDirections,\n"
-"	float4* sep,\n"
-"	float* dmin)\n"
-"{\n"
-"	\n"
-"	float4 posA = posA1;\n"
-"	posA.w = 0.f;\n"
-"	float4 posB = posB1;\n"
-"	posB.w = 0.f;\n"
-"	int curPlaneTests=0;\n"
-"	int curEdgeEdge = 0;\n"
-"	// Test unit sphere directions\n"
-"	for (int i=0;i<numUnitSphereDirections;i++)\n"
-"	{\n"
-"		float4 crossje;\n"
-"		crossje = unitSphereDirections[i];	\n"
-"		if (dot3F4(DeltaC2,crossje)>0)\n"
-"			crossje *= -1.f;\n"
-"		{\n"
-"			float dist;\n"
-"			bool result = true;\n"
-"			float Min0,Max0;\n"
-"			float Min1,Max1;\n"
-"			project(hullA,posA,ornA,&crossje,vertices, &Min0, &Max0);\n"
-"			project(hullB,posB,ornB,&crossje,vertices, &Min1, &Max1);\n"
-"		\n"
-"			if(Max0<Min1 || Max1<Min0)\n"
-"				return false;\n"
-"		\n"
-"			float d0 = Max0 - Min1;\n"
-"			float d1 = Max1 - Min0;\n"
-"			dist = d0<d1 ? d0:d1;\n"
-"			result = true;\n"
-"	\n"
-"			if(dist<*dmin)\n"
-"			{\n"
-"				*dmin = dist;\n"
-"				*sep = crossje;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	\n"
-"	if((dot3F4(-DeltaC2,*sep))>0.0f)\n"
-"	{\n"
-"		*sep = -(*sep);\n"
-"	}\n"
-"	return true;\n"
-"}\n"
-"__kernel void   findSeparatingAxisUnitSphereKernel( __global const int4* pairs, \n"
-"																					__global const b3RigidBodyData_t* rigidBodies, \n"
-"																					__global const b3Collidable_t* collidables,\n"
-"																					__global const b3ConvexPolyhedronData_t* convexShapes, \n"
-"																					__global const float4* vertices,\n"
-"																					__global const float4* unitSphereDirections,\n"
-"																					__global  float4* separatingNormals,\n"
-"																					__global  int* hasSeparatingAxis,\n"
-"																					__global  float* dmins,\n"
-"																					int numUnitSphereDirections,\n"
-"																					int numPairs\n"
-"																					)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	\n"
-"	if (i<numPairs)\n"
-"	{\n"
-"		if (hasSeparatingAxis[i])\n"
-"		{\n"
-"	\n"
-"			int bodyIndexA = pairs[i].x;\n"
-"			int bodyIndexB = pairs[i].y;\n"
-"	\n"
-"			int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
-"			int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
-"		\n"
-"			int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
-"			int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
-"			\n"
-"			\n"
-"			int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
-"	\n"
-"			float dmin = dmins[i];\n"
-"	\n"
-"			float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
-"			posA.w = 0.f;\n"
-"			float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
-"			posB.w = 0.f;\n"
-"			float4 c0local = convexShapes[shapeIndexA].m_localCenter;\n"
-"			float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
-"			float4 c0 = transform(&c0local, &posA, &ornA);\n"
-"			float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
-"			float4 ornB =rigidBodies[bodyIndexB].m_quat;\n"
-"			float4 c1 = transform(&c1local,&posB,&ornB);\n"
-"			const float4 DeltaC2 = c0 - c1;\n"
-"			float4 sepNormal = separatingNormals[i];\n"
-"			\n"
-"			int numEdgeEdgeDirections = convexShapes[shapeIndexA].m_numUniqueEdges*convexShapes[shapeIndexB].m_numUniqueEdges;\n"
-"			if (numEdgeEdgeDirections>numUnitSphereDirections)\n"
-"			{\n"
-"				bool sepEE = findSeparatingAxisUnitSphere(	&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],posA,ornA,\n"
-"																										posB,ornB,\n"
-"																										DeltaC2,\n"
-"																										vertices,unitSphereDirections,numUnitSphereDirections,&sepNormal,&dmin);\n"
-"				if (!sepEE)\n"
-"				{\n"
-"					hasSeparatingAxis[i] = 0;\n"
-"				} else\n"
-"				{\n"
-"					hasSeparatingAxis[i] = 1;\n"
-"					separatingNormals[i] = sepNormal;\n"
-"				}\n"
-"			}\n"
-"		}		//if (hasSeparatingAxis[i])\n"
-"	}//(i<numPairs)\n"
-"}\n"
-;
+static const char* mprKernelsCL =
+	"/***\n"
+	" * ---------------------------------\n"
+	" * Copyright (c)2012 Daniel Fiser <danfis@danfis.cz>\n"
+	" *\n"
+	" *  This file was ported from mpr.c file, part of libccd.\n"
+	" *  The Minkoski Portal Refinement implementation was ported \n"
+	" *  to OpenCL by Erwin Coumans for the Bullet 3 Physics library.\n"
+	" *  at http://github.com/erwincoumans/bullet3\n"
+	" *\n"
+	" *  Distributed under the OSI-approved BSD License (the \"License\");\n"
+	" *  see <http://www.opensource.org/licenses/bsd-license.php>.\n"
+	" *  This software is distributed WITHOUT ANY WARRANTY; without even the\n"
+	" *  implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.\n"
+	" *  See the License for more information.\n"
+	" */\n"
+	"#ifndef B3_MPR_PENETRATION_H\n"
+	"#define B3_MPR_PENETRATION_H\n"
+	"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
+	"#define B3_PLATFORM_DEFINITIONS_H\n"
+	"struct MyTest\n"
+	"{\n"
+	"	int bla;\n"
+	"};\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
+	"#define B3_LARGE_FLOAT 1e18f\n"
+	"#define B3_INFINITY 1e18f\n"
+	"#define b3Assert(a)\n"
+	"#define b3ConstArray(a) __global const a*\n"
+	"#define b3AtomicInc atomic_inc\n"
+	"#define b3AtomicAdd atomic_add\n"
+	"#define b3Fabs fabs\n"
+	"#define b3Sqrt native_sqrt\n"
+	"#define b3Sin native_sin\n"
+	"#define b3Cos native_cos\n"
+	"#define B3_STATIC\n"
+	"#endif\n"
+	"#endif\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#define B3_FLOAT4_H\n"
+	"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif\n"
+	"#endif\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"	typedef float4	b3Float4;\n"
+	"	#define b3Float4ConstArg const b3Float4\n"
+	"	#define b3MakeFloat4 (float4)\n"
+	"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return dot(a1, b1);\n"
+	"	}\n"
+	"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return cross(a1, b1);\n"
+	"	}\n"
+	"	#define b3MinFloat4 min\n"
+	"	#define b3MaxFloat4 max\n"
+	"	#define b3Normalized(a) normalize(a)\n"
+	"#endif \n"
+	"		\n"
+	"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
+	"{\n"
+	"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
+	"		return false;\n"
+	"	return true;\n"
+	"}\n"
+	"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
+	"{\n"
+	"    float maxDot = -B3_INFINITY;\n"
+	"    int i = 0;\n"
+	"    int ptIndex = -1;\n"
+	"    for( i = 0; i < vecLen; i++ )\n"
+	"    {\n"
+	"        float dot = b3Dot3F4(vecArray[i],vec);\n"
+	"            \n"
+	"        if( dot > maxDot )\n"
+	"        {\n"
+	"            maxDot = dot;\n"
+	"            ptIndex = i;\n"
+	"        }\n"
+	"    }\n"
+	"	b3Assert(ptIndex>=0);\n"
+	"    if (ptIndex<0)\n"
+	"	{\n"
+	"		ptIndex = 0;\n"
+	"	}\n"
+	"    *dotOut = maxDot;\n"
+	"    return ptIndex;\n"
+	"}\n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifndef B3_RIGIDBODY_DATA_H\n"
+	"#define B3_RIGIDBODY_DATA_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifndef B3_QUAT_H\n"
+	"#define B3_QUAT_H\n"
+	"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif\n"
+	"#endif\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"	typedef float4	b3Quat;\n"
+	"	#define b3QuatConstArg const b3Quat\n"
+	"	\n"
+	"	\n"
+	"inline float4 b3FastNormalize4(float4 v)\n"
+	"{\n"
+	"	v = (float4)(v.xyz,0.f);\n"
+	"	return fast_normalize(v);\n"
+	"}\n"
+	"	\n"
+	"inline b3Quat b3QuatMul(b3Quat a, b3Quat b);\n"
+	"inline b3Quat b3QuatNormalized(b3QuatConstArg in);\n"
+	"inline b3Quat b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec);\n"
+	"inline b3Quat b3QuatInvert(b3QuatConstArg q);\n"
+	"inline b3Quat b3QuatInverse(b3QuatConstArg q);\n"
+	"inline b3Quat b3QuatMul(b3QuatConstArg a, b3QuatConstArg b)\n"
+	"{\n"
+	"	b3Quat ans;\n"
+	"	ans = b3Cross3( a, b );\n"
+	"	ans += a.w*b+b.w*a;\n"
+	"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
+	"	ans.w = a.w*b.w - b3Dot3F4(a, b);\n"
+	"	return ans;\n"
+	"}\n"
+	"inline b3Quat b3QuatNormalized(b3QuatConstArg in)\n"
+	"{\n"
+	"	b3Quat q;\n"
+	"	q=in;\n"
+	"	//return b3FastNormalize4(in);\n"
+	"	float len = native_sqrt(dot(q, q));\n"
+	"	if(len > 0.f)\n"
+	"	{\n"
+	"		q *= 1.f / len;\n"
+	"	}\n"
+	"	else\n"
+	"	{\n"
+	"		q.x = q.y = q.z = 0.f;\n"
+	"		q.w = 1.f;\n"
+	"	}\n"
+	"	return q;\n"
+	"}\n"
+	"inline float4 b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
+	"{\n"
+	"	b3Quat qInv = b3QuatInvert( q );\n"
+	"	float4 vcpy = vec;\n"
+	"	vcpy.w = 0.f;\n"
+	"	float4 out = b3QuatMul(b3QuatMul(q,vcpy),qInv);\n"
+	"	return out;\n"
+	"}\n"
+	"inline b3Quat b3QuatInverse(b3QuatConstArg q)\n"
+	"{\n"
+	"	return (b3Quat)(-q.xyz, q.w);\n"
+	"}\n"
+	"inline b3Quat b3QuatInvert(b3QuatConstArg q)\n"
+	"{\n"
+	"	return (b3Quat)(-q.xyz, q.w);\n"
+	"}\n"
+	"inline float4 b3QuatInvRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
+	"{\n"
+	"	return b3QuatRotate( b3QuatInvert( q ), vec );\n"
+	"}\n"
+	"inline b3Float4 b3TransformPoint(b3Float4ConstArg point, b3Float4ConstArg translation, b3QuatConstArg  orientation)\n"
+	"{\n"
+	"	return b3QuatRotate( orientation, point ) + (translation);\n"
+	"}\n"
+	"	\n"
+	"#endif \n"
+	"#endif //B3_QUAT_H\n"
+	"#ifndef B3_MAT3x3_H\n"
+	"#define B3_MAT3x3_H\n"
+	"#ifndef B3_QUAT_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_QUAT_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"typedef struct\n"
+	"{\n"
+	"	b3Float4 m_row[3];\n"
+	"}b3Mat3x3;\n"
+	"#define b3Mat3x3ConstArg const b3Mat3x3\n"
+	"#define b3GetRow(m,row) (m.m_row[row])\n"
+	"inline b3Mat3x3 b3QuatGetRotationMatrix(b3Quat quat)\n"
+	"{\n"
+	"	b3Float4 quat2 = (b3Float4)(quat.x*quat.x, quat.y*quat.y, quat.z*quat.z, 0.f);\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0].x=1-2*quat2.y-2*quat2.z;\n"
+	"	out.m_row[0].y=2*quat.x*quat.y-2*quat.w*quat.z;\n"
+	"	out.m_row[0].z=2*quat.x*quat.z+2*quat.w*quat.y;\n"
+	"	out.m_row[0].w = 0.f;\n"
+	"	out.m_row[1].x=2*quat.x*quat.y+2*quat.w*quat.z;\n"
+	"	out.m_row[1].y=1-2*quat2.x-2*quat2.z;\n"
+	"	out.m_row[1].z=2*quat.y*quat.z-2*quat.w*quat.x;\n"
+	"	out.m_row[1].w = 0.f;\n"
+	"	out.m_row[2].x=2*quat.x*quat.z-2*quat.w*quat.y;\n"
+	"	out.m_row[2].y=2*quat.y*quat.z+2*quat.w*quat.x;\n"
+	"	out.m_row[2].z=1-2*quat2.x-2*quat2.y;\n"
+	"	out.m_row[2].w = 0.f;\n"
+	"	return out;\n"
+	"}\n"
+	"inline b3Mat3x3 b3AbsoluteMat3x3(b3Mat3x3ConstArg matIn)\n"
+	"{\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0] = fabs(matIn.m_row[0]);\n"
+	"	out.m_row[1] = fabs(matIn.m_row[1]);\n"
+	"	out.m_row[2] = fabs(matIn.m_row[2]);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtZero();\n"
+	"__inline\n"
+	"b3Mat3x3 mtIdentity();\n"
+	"__inline\n"
+	"b3Mat3x3 mtTranspose(b3Mat3x3 m);\n"
+	"__inline\n"
+	"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b);\n"
+	"__inline\n"
+	"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b);\n"
+	"__inline\n"
+	"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b);\n"
+	"__inline\n"
+	"b3Mat3x3 mtZero()\n"
+	"{\n"
+	"	b3Mat3x3 m;\n"
+	"	m.m_row[0] = (b3Float4)(0.f);\n"
+	"	m.m_row[1] = (b3Float4)(0.f);\n"
+	"	m.m_row[2] = (b3Float4)(0.f);\n"
+	"	return m;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtIdentity()\n"
+	"{\n"
+	"	b3Mat3x3 m;\n"
+	"	m.m_row[0] = (b3Float4)(1,0,0,0);\n"
+	"	m.m_row[1] = (b3Float4)(0,1,0,0);\n"
+	"	m.m_row[2] = (b3Float4)(0,0,1,0);\n"
+	"	return m;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtTranspose(b3Mat3x3 m)\n"
+	"{\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0] = (b3Float4)(m.m_row[0].x, m.m_row[1].x, m.m_row[2].x, 0.f);\n"
+	"	out.m_row[1] = (b3Float4)(m.m_row[0].y, m.m_row[1].y, m.m_row[2].y, 0.f);\n"
+	"	out.m_row[2] = (b3Float4)(m.m_row[0].z, m.m_row[1].z, m.m_row[2].z, 0.f);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b)\n"
+	"{\n"
+	"	b3Mat3x3 transB;\n"
+	"	transB = mtTranspose( b );\n"
+	"	b3Mat3x3 ans;\n"
+	"	//	why this doesn't run when 0ing in the for{}\n"
+	"	a.m_row[0].w = 0.f;\n"
+	"	a.m_row[1].w = 0.f;\n"
+	"	a.m_row[2].w = 0.f;\n"
+	"	for(int i=0; i<3; i++)\n"
+	"	{\n"
+	"//	a.m_row[i].w = 0.f;\n"
+	"		ans.m_row[i].x = b3Dot3F4(a.m_row[i],transB.m_row[0]);\n"
+	"		ans.m_row[i].y = b3Dot3F4(a.m_row[i],transB.m_row[1]);\n"
+	"		ans.m_row[i].z = b3Dot3F4(a.m_row[i],transB.m_row[2]);\n"
+	"		ans.m_row[i].w = 0.f;\n"
+	"	}\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b)\n"
+	"{\n"
+	"	b3Float4 ans;\n"
+	"	ans.x = b3Dot3F4( a.m_row[0], b );\n"
+	"	ans.y = b3Dot3F4( a.m_row[1], b );\n"
+	"	ans.z = b3Dot3F4( a.m_row[2], b );\n"
+	"	ans.w = 0.f;\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b)\n"
+	"{\n"
+	"	b3Float4 colx = b3MakeFloat4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
+	"	b3Float4 coly = b3MakeFloat4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
+	"	b3Float4 colz = b3MakeFloat4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
+	"	b3Float4 ans;\n"
+	"	ans.x = b3Dot3F4( a, colx );\n"
+	"	ans.y = b3Dot3F4( a, coly );\n"
+	"	ans.z = b3Dot3F4( a, colz );\n"
+	"	return ans;\n"
+	"}\n"
+	"#endif\n"
+	"#endif //B3_MAT3x3_H\n"
+	"typedef struct b3RigidBodyData b3RigidBodyData_t;\n"
+	"struct b3RigidBodyData\n"
+	"{\n"
+	"	b3Float4				m_pos;\n"
+	"	b3Quat					m_quat;\n"
+	"	b3Float4				m_linVel;\n"
+	"	b3Float4				m_angVel;\n"
+	"	int 					m_collidableIdx;\n"
+	"	float 				m_invMass;\n"
+	"	float 				m_restituitionCoeff;\n"
+	"	float 				m_frictionCoeff;\n"
+	"};\n"
+	"typedef struct b3InertiaData b3InertiaData_t;\n"
+	"struct b3InertiaData\n"
+	"{\n"
+	"	b3Mat3x3 m_invInertiaWorld;\n"
+	"	b3Mat3x3 m_initInvInertia;\n"
+	"};\n"
+	"#endif //B3_RIGIDBODY_DATA_H\n"
+	"	\n"
+	"#ifndef B3_CONVEX_POLYHEDRON_DATA_H\n"
+	"#define B3_CONVEX_POLYHEDRON_DATA_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifndef B3_QUAT_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_QUAT_H\n"
+	"typedef struct b3GpuFace b3GpuFace_t;\n"
+	"struct b3GpuFace\n"
+	"{\n"
+	"	b3Float4 m_plane;\n"
+	"	int m_indexOffset;\n"
+	"	int m_numIndices;\n"
+	"	int m_unusedPadding1;\n"
+	"	int m_unusedPadding2;\n"
+	"};\n"
+	"typedef struct b3ConvexPolyhedronData b3ConvexPolyhedronData_t;\n"
+	"struct b3ConvexPolyhedronData\n"
+	"{\n"
+	"	b3Float4		m_localCenter;\n"
+	"	b3Float4		m_extents;\n"
+	"	b3Float4		mC;\n"
+	"	b3Float4		mE;\n"
+	"	float			m_radius;\n"
+	"	int	m_faceOffset;\n"
+	"	int m_numFaces;\n"
+	"	int	m_numVertices;\n"
+	"	int m_vertexOffset;\n"
+	"	int	m_uniqueEdgesOffset;\n"
+	"	int	m_numUniqueEdges;\n"
+	"	int m_unused;\n"
+	"};\n"
+	"#endif //B3_CONVEX_POLYHEDRON_DATA_H\n"
+	"#ifndef B3_COLLIDABLE_H\n"
+	"#define B3_COLLIDABLE_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifndef B3_QUAT_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_QUAT_H\n"
+	"enum b3ShapeTypes\n"
+	"{\n"
+	"	SHAPE_HEIGHT_FIELD=1,\n"
+	"	SHAPE_CONVEX_HULL=3,\n"
+	"	SHAPE_PLANE=4,\n"
+	"	SHAPE_CONCAVE_TRIMESH=5,\n"
+	"	SHAPE_COMPOUND_OF_CONVEX_HULLS=6,\n"
+	"	SHAPE_SPHERE=7,\n"
+	"	MAX_NUM_SHAPE_TYPES,\n"
+	"};\n"
+	"typedef struct b3Collidable b3Collidable_t;\n"
+	"struct b3Collidable\n"
+	"{\n"
+	"	union {\n"
+	"		int m_numChildShapes;\n"
+	"		int m_bvhIndex;\n"
+	"	};\n"
+	"	union\n"
+	"	{\n"
+	"		float m_radius;\n"
+	"		int	m_compoundBvhIndex;\n"
+	"	};\n"
+	"	int m_shapeType;\n"
+	"	int m_shapeIndex;\n"
+	"};\n"
+	"typedef struct b3GpuChildShape b3GpuChildShape_t;\n"
+	"struct b3GpuChildShape\n"
+	"{\n"
+	"	b3Float4	m_childPosition;\n"
+	"	b3Quat		m_childOrientation;\n"
+	"	int m_shapeIndex;\n"
+	"	int m_unused0;\n"
+	"	int m_unused1;\n"
+	"	int m_unused2;\n"
+	"};\n"
+	"struct b3CompoundOverlappingPair\n"
+	"{\n"
+	"	int m_bodyIndexA;\n"
+	"	int m_bodyIndexB;\n"
+	"//	int	m_pairType;\n"
+	"	int m_childShapeIndexA;\n"
+	"	int m_childShapeIndexB;\n"
+	"};\n"
+	"#endif //B3_COLLIDABLE_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#define B3_MPR_SQRT sqrt\n"
+	"#endif\n"
+	"#define B3_MPR_FMIN(x, y) ((x) < (y) ? (x) : (y))\n"
+	"#define B3_MPR_FABS fabs\n"
+	"#define B3_MPR_TOLERANCE 1E-6f\n"
+	"#define B3_MPR_MAX_ITERATIONS 1000\n"
+	"struct _b3MprSupport_t \n"
+	"{\n"
+	"    b3Float4 v;  //!< Support point in minkowski sum\n"
+	"    b3Float4 v1; //!< Support point in obj1\n"
+	"    b3Float4 v2; //!< Support point in obj2\n"
+	"};\n"
+	"typedef struct _b3MprSupport_t b3MprSupport_t;\n"
+	"struct _b3MprSimplex_t \n"
+	"{\n"
+	"    b3MprSupport_t ps[4];\n"
+	"    int last; //!< index of last added point\n"
+	"};\n"
+	"typedef struct _b3MprSimplex_t b3MprSimplex_t;\n"
+	"inline b3MprSupport_t* b3MprSimplexPointW(b3MprSimplex_t *s, int idx)\n"
+	"{\n"
+	"    return &s->ps[idx];\n"
+	"}\n"
+	"inline void b3MprSimplexSetSize(b3MprSimplex_t *s, int size)\n"
+	"{\n"
+	"    s->last = size - 1;\n"
+	"}\n"
+	"inline int b3MprSimplexSize(const b3MprSimplex_t *s)\n"
+	"{\n"
+	"    return s->last + 1;\n"
+	"}\n"
+	"inline const b3MprSupport_t* b3MprSimplexPoint(const b3MprSimplex_t* s, int idx)\n"
+	"{\n"
+	"    // here is no check on boundaries\n"
+	"    return &s->ps[idx];\n"
+	"}\n"
+	"inline void b3MprSupportCopy(b3MprSupport_t *d, const b3MprSupport_t *s)\n"
+	"{\n"
+	"    *d = *s;\n"
+	"}\n"
+	"inline void b3MprSimplexSet(b3MprSimplex_t *s, size_t pos, const b3MprSupport_t *a)\n"
+	"{\n"
+	"    b3MprSupportCopy(s->ps + pos, a);\n"
+	"}\n"
+	"inline void b3MprSimplexSwap(b3MprSimplex_t *s, size_t pos1, size_t pos2)\n"
+	"{\n"
+	"    b3MprSupport_t supp;\n"
+	"    b3MprSupportCopy(&supp, &s->ps[pos1]);\n"
+	"    b3MprSupportCopy(&s->ps[pos1], &s->ps[pos2]);\n"
+	"    b3MprSupportCopy(&s->ps[pos2], &supp);\n"
+	"}\n"
+	"inline int b3MprIsZero(float val)\n"
+	"{\n"
+	"    return B3_MPR_FABS(val) < FLT_EPSILON;\n"
+	"}\n"
+	"inline int b3MprEq(float _a, float _b)\n"
+	"{\n"
+	"    float ab;\n"
+	"    float a, b;\n"
+	"    ab = B3_MPR_FABS(_a - _b);\n"
+	"    if (B3_MPR_FABS(ab) < FLT_EPSILON)\n"
+	"        return 1;\n"
+	"    a = B3_MPR_FABS(_a);\n"
+	"    b = B3_MPR_FABS(_b);\n"
+	"    if (b > a){\n"
+	"        return ab < FLT_EPSILON * b;\n"
+	"    }else{\n"
+	"        return ab < FLT_EPSILON * a;\n"
+	"    }\n"
+	"}\n"
+	"inline int b3MprVec3Eq(const b3Float4* a, const b3Float4 *b)\n"
+	"{\n"
+	"    return b3MprEq((*a).x, (*b).x)\n"
+	"            && b3MprEq((*a).y, (*b).y)\n"
+	"            && b3MprEq((*a).z, (*b).z);\n"
+	"}\n"
+	"inline b3Float4 b3LocalGetSupportVertex(b3Float4ConstArg supportVec,__global const b3ConvexPolyhedronData_t* hull, 	b3ConstArray(b3Float4) verticesA)\n"
+	"{\n"
+	"	b3Float4 supVec = b3MakeFloat4(0,0,0,0);\n"
+	"	float maxDot = -B3_LARGE_FLOAT;\n"
+	"    if( 0 < hull->m_numVertices )\n"
+	"    {\n"
+	"        const b3Float4 scaled = supportVec;\n"
+	"		int index = b3MaxDot(scaled, &verticesA[hull->m_vertexOffset], hull->m_numVertices, &maxDot);\n"
+	"        return verticesA[hull->m_vertexOffset+index];\n"
+	"    }\n"
+	"    return supVec;\n"
+	"}\n"
+	"B3_STATIC void b3MprConvexSupport(int pairIndex,int bodyIndex,  b3ConstArray(b3RigidBodyData_t) cpuBodyBuf, \n"
+	"													b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData, \n"
+	"													b3ConstArray(b3Collidable_t)				cpuCollidables,\n"
+	"													b3ConstArray(b3Float4)					cpuVertices,\n"
+	"													__global b3Float4* sepAxis,\n"
+	"														const b3Float4* _dir, b3Float4* outp, int logme)\n"
+	"{\n"
+	"	//dir is in worldspace, move to local space\n"
+	"	\n"
+	"	b3Float4 pos = cpuBodyBuf[bodyIndex].m_pos;\n"
+	"	b3Quat orn = cpuBodyBuf[bodyIndex].m_quat;\n"
+	"	\n"
+	"	b3Float4 dir = b3MakeFloat4((*_dir).x,(*_dir).y,(*_dir).z,0.f);\n"
+	"	\n"
+	"	const b3Float4 localDir = b3QuatRotate(b3QuatInverse(orn),dir);\n"
+	"	\n"
+	"	//find local support vertex\n"
+	"	int colIndex = cpuBodyBuf[bodyIndex].m_collidableIdx;\n"
+	"	\n"
+	"	b3Assert(cpuCollidables[colIndex].m_shapeType==SHAPE_CONVEX_HULL);\n"
+	"	__global const b3ConvexPolyhedronData_t* hull = &cpuConvexData[cpuCollidables[colIndex].m_shapeIndex];\n"
+	"	\n"
+	"	b3Float4 pInA;\n"
+	"	if (logme)\n"
+	"	{\n"
+	"		b3Float4 supVec = b3MakeFloat4(0,0,0,0);\n"
+	"		float maxDot = -B3_LARGE_FLOAT;\n"
+	"		if( 0 < hull->m_numVertices )\n"
+	"		{\n"
+	"			const b3Float4 scaled = localDir;\n"
+	"			int index = b3MaxDot(scaled, &cpuVertices[hull->m_vertexOffset], hull->m_numVertices, &maxDot);\n"
+	"			pInA = cpuVertices[hull->m_vertexOffset+index];\n"
+	"			\n"
+	"		}\n"
+	"	} else\n"
+	"	{\n"
+	"		pInA = b3LocalGetSupportVertex(localDir,hull,cpuVertices);\n"
+	"	}\n"
+	"	//move vertex to world space\n"
+	"	*outp = b3TransformPoint(pInA,pos,orn);\n"
+	"	\n"
+	"}\n"
+	"inline void b3MprSupport(int pairIndex,int bodyIndexA, int bodyIndexB,   b3ConstArray(b3RigidBodyData_t) cpuBodyBuf, \n"
+	"													b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData, \n"
+	"													b3ConstArray(b3Collidable_t)				cpuCollidables,\n"
+	"													b3ConstArray(b3Float4)					cpuVertices,\n"
+	"													__global b3Float4* sepAxis,\n"
+	"													const b3Float4* _dir, b3MprSupport_t *supp)\n"
+	"{\n"
+	"    b3Float4 dir;\n"
+	"	dir = *_dir;\n"
+	"	b3MprConvexSupport(pairIndex,bodyIndexA,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices,sepAxis,&dir, &supp->v1,0);\n"
+	"    dir = *_dir*-1.f;\n"
+	"	b3MprConvexSupport(pairIndex,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices,sepAxis,&dir, &supp->v2,0);\n"
+	"    supp->v = supp->v1 - supp->v2;\n"
+	"}\n"
+	"inline void b3FindOrigin(int bodyIndexA, int bodyIndexB, b3ConstArray(b3RigidBodyData_t) cpuBodyBuf, b3MprSupport_t *center)\n"
+	"{\n"
+	"    center->v1 = cpuBodyBuf[bodyIndexA].m_pos;\n"
+	"	center->v2 = cpuBodyBuf[bodyIndexB].m_pos;\n"
+	"    center->v = center->v1 - center->v2;\n"
+	"}\n"
+	"inline void b3MprVec3Set(b3Float4 *v, float x, float y, float z)\n"
+	"{\n"
+	"	(*v).x = x;\n"
+	"	(*v).y = y;\n"
+	"	(*v).z = z;\n"
+	"	(*v).w = 0.f;\n"
+	"}\n"
+	"inline void b3MprVec3Add(b3Float4 *v, const b3Float4 *w)\n"
+	"{\n"
+	"    (*v).x += (*w).x;\n"
+	"    (*v).y += (*w).y;\n"
+	"    (*v).z += (*w).z;\n"
+	"}\n"
+	"inline void b3MprVec3Copy(b3Float4 *v, const b3Float4 *w)\n"
+	"{\n"
+	"    *v = *w;\n"
+	"}\n"
+	"inline void b3MprVec3Scale(b3Float4 *d, float k)\n"
+	"{\n"
+	"    *d *= k;\n"
+	"}\n"
+	"inline float b3MprVec3Dot(const b3Float4 *a, const b3Float4 *b)\n"
+	"{\n"
+	"    float dot;\n"
+	"	dot = b3Dot3F4(*a,*b);\n"
+	"    return dot;\n"
+	"}\n"
+	"inline float b3MprVec3Len2(const b3Float4 *v)\n"
+	"{\n"
+	"    return b3MprVec3Dot(v, v);\n"
+	"}\n"
+	"inline void b3MprVec3Normalize(b3Float4 *d)\n"
+	"{\n"
+	"    float k = 1.f / B3_MPR_SQRT(b3MprVec3Len2(d));\n"
+	"    b3MprVec3Scale(d, k);\n"
+	"}\n"
+	"inline void b3MprVec3Cross(b3Float4 *d, const b3Float4 *a, const b3Float4 *b)\n"
+	"{\n"
+	"	*d = b3Cross3(*a,*b);\n"
+	"	\n"
+	"}\n"
+	"inline void b3MprVec3Sub2(b3Float4 *d, const b3Float4 *v, const b3Float4 *w)\n"
+	"{\n"
+	"	*d = *v - *w;\n"
+	"}\n"
+	"inline void b3PortalDir(const b3MprSimplex_t *portal, b3Float4 *dir)\n"
+	"{\n"
+	"    b3Float4 v2v1, v3v1;\n"
+	"    b3MprVec3Sub2(&v2v1, &b3MprSimplexPoint(portal, 2)->v,\n"
+	"                       &b3MprSimplexPoint(portal, 1)->v);\n"
+	"    b3MprVec3Sub2(&v3v1, &b3MprSimplexPoint(portal, 3)->v,\n"
+	"                       &b3MprSimplexPoint(portal, 1)->v);\n"
+	"    b3MprVec3Cross(dir, &v2v1, &v3v1);\n"
+	"    b3MprVec3Normalize(dir);\n"
+	"}\n"
+	"inline int portalEncapsulesOrigin(const b3MprSimplex_t *portal,\n"
+	"                                       const b3Float4 *dir)\n"
+	"{\n"
+	"    float dot;\n"
+	"    dot = b3MprVec3Dot(dir, &b3MprSimplexPoint(portal, 1)->v);\n"
+	"    return b3MprIsZero(dot) || dot > 0.f;\n"
+	"}\n"
+	"inline int portalReachTolerance(const b3MprSimplex_t *portal,\n"
+	"                                     const b3MprSupport_t *v4,\n"
+	"                                     const b3Float4 *dir)\n"
+	"{\n"
+	"    float dv1, dv2, dv3, dv4;\n"
+	"    float dot1, dot2, dot3;\n"
+	"    // find the smallest dot product of dir and {v1-v4, v2-v4, v3-v4}\n"
+	"    dv1 = b3MprVec3Dot(&b3MprSimplexPoint(portal, 1)->v, dir);\n"
+	"    dv2 = b3MprVec3Dot(&b3MprSimplexPoint(portal, 2)->v, dir);\n"
+	"    dv3 = b3MprVec3Dot(&b3MprSimplexPoint(portal, 3)->v, dir);\n"
+	"    dv4 = b3MprVec3Dot(&v4->v, dir);\n"
+	"    dot1 = dv4 - dv1;\n"
+	"    dot2 = dv4 - dv2;\n"
+	"    dot3 = dv4 - dv3;\n"
+	"    dot1 = B3_MPR_FMIN(dot1, dot2);\n"
+	"    dot1 = B3_MPR_FMIN(dot1, dot3);\n"
+	"    return b3MprEq(dot1, B3_MPR_TOLERANCE) || dot1 < B3_MPR_TOLERANCE;\n"
+	"}\n"
+	"inline int portalCanEncapsuleOrigin(const b3MprSimplex_t *portal,   \n"
+	"                                         const b3MprSupport_t *v4,\n"
+	"                                         const b3Float4 *dir)\n"
+	"{\n"
+	"    float dot;\n"
+	"    dot = b3MprVec3Dot(&v4->v, dir);\n"
+	"    return b3MprIsZero(dot) || dot > 0.f;\n"
+	"}\n"
+	"inline void b3ExpandPortal(b3MprSimplex_t *portal,\n"
+	"                              const b3MprSupport_t *v4)\n"
+	"{\n"
+	"    float dot;\n"
+	"    b3Float4 v4v0;\n"
+	"    b3MprVec3Cross(&v4v0, &v4->v, &b3MprSimplexPoint(portal, 0)->v);\n"
+	"    dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 1)->v, &v4v0);\n"
+	"    if (dot > 0.f){\n"
+	"        dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 2)->v, &v4v0);\n"
+	"        if (dot > 0.f){\n"
+	"            b3MprSimplexSet(portal, 1, v4);\n"
+	"        }else{\n"
+	"            b3MprSimplexSet(portal, 3, v4);\n"
+	"        }\n"
+	"    }else{\n"
+	"        dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 3)->v, &v4v0);\n"
+	"        if (dot > 0.f){\n"
+	"            b3MprSimplexSet(portal, 2, v4);\n"
+	"        }else{\n"
+	"            b3MprSimplexSet(portal, 1, v4);\n"
+	"        }\n"
+	"    }\n"
+	"}\n"
+	"B3_STATIC int b3DiscoverPortal(int pairIndex, int bodyIndexA, int bodyIndexB,  b3ConstArray(b3RigidBodyData_t) cpuBodyBuf, \n"
+	"													b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData, \n"
+	"													b3ConstArray(b3Collidable_t)				cpuCollidables,\n"
+	"													b3ConstArray(b3Float4)					cpuVertices,\n"
+	"													__global b3Float4* sepAxis,\n"
+	"													__global int*	hasSepAxis,\n"
+	"													b3MprSimplex_t *portal)\n"
+	"{\n"
+	"    b3Float4 dir, va, vb;\n"
+	"    float dot;\n"
+	"    int cont;\n"
+	"	\n"
+	"	\n"
+	"    // vertex 0 is center of portal\n"
+	"    b3FindOrigin(bodyIndexA,bodyIndexB,cpuBodyBuf, b3MprSimplexPointW(portal, 0));\n"
+	"    // vertex 0 is center of portal\n"
+	"    b3MprSimplexSetSize(portal, 1);\n"
+	"	\n"
+	"	b3Float4 zero = b3MakeFloat4(0,0,0,0);\n"
+	"	b3Float4* b3mpr_vec3_origin = &zero;\n"
+	"    if (b3MprVec3Eq(&b3MprSimplexPoint(portal, 0)->v, b3mpr_vec3_origin)){\n"
+	"        // Portal's center lies on origin (0,0,0) => we know that objects\n"
+	"        // intersect but we would need to know penetration info.\n"
+	"        // So move center little bit...\n"
+	"        b3MprVec3Set(&va, FLT_EPSILON * 10.f, 0.f, 0.f);\n"
+	"        b3MprVec3Add(&b3MprSimplexPointW(portal, 0)->v, &va);\n"
+	"    }\n"
+	"    // vertex 1 = support in direction of origin\n"
+	"    b3MprVec3Copy(&dir, &b3MprSimplexPoint(portal, 0)->v);\n"
+	"    b3MprVec3Scale(&dir, -1.f);\n"
+	"    b3MprVec3Normalize(&dir);\n"
+	"    b3MprSupport(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&dir, b3MprSimplexPointW(portal, 1));\n"
+	"    b3MprSimplexSetSize(portal, 2);\n"
+	"    // test if origin isn't outside of v1\n"
+	"    dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 1)->v, &dir);\n"
+	"	\n"
+	"    if (b3MprIsZero(dot) || dot < 0.f)\n"
+	"        return -1;\n"
+	"    // vertex 2\n"
+	"    b3MprVec3Cross(&dir, &b3MprSimplexPoint(portal, 0)->v,\n"
+	"                       &b3MprSimplexPoint(portal, 1)->v);\n"
+	"    if (b3MprIsZero(b3MprVec3Len2(&dir))){\n"
+	"        if (b3MprVec3Eq(&b3MprSimplexPoint(portal, 1)->v, b3mpr_vec3_origin)){\n"
+	"            // origin lies on v1\n"
+	"            return 1;\n"
+	"        }else{\n"
+	"            // origin lies on v0-v1 segment\n"
+	"            return 2;\n"
+	"        }\n"
+	"    }\n"
+	"    b3MprVec3Normalize(&dir);\n"
+	"	 b3MprSupport(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&dir, b3MprSimplexPointW(portal, 2));\n"
+	"    \n"
+	"    dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 2)->v, &dir);\n"
+	"    if (b3MprIsZero(dot) || dot < 0.f)\n"
+	"        return -1;\n"
+	"    b3MprSimplexSetSize(portal, 3);\n"
+	"    // vertex 3 direction\n"
+	"    b3MprVec3Sub2(&va, &b3MprSimplexPoint(portal, 1)->v,\n"
+	"                     &b3MprSimplexPoint(portal, 0)->v);\n"
+	"    b3MprVec3Sub2(&vb, &b3MprSimplexPoint(portal, 2)->v,\n"
+	"                     &b3MprSimplexPoint(portal, 0)->v);\n"
+	"    b3MprVec3Cross(&dir, &va, &vb);\n"
+	"    b3MprVec3Normalize(&dir);\n"
+	"    // it is better to form portal faces to be oriented \"outside\" origin\n"
+	"    dot = b3MprVec3Dot(&dir, &b3MprSimplexPoint(portal, 0)->v);\n"
+	"    if (dot > 0.f){\n"
+	"        b3MprSimplexSwap(portal, 1, 2);\n"
+	"        b3MprVec3Scale(&dir, -1.f);\n"
+	"    }\n"
+	"    while (b3MprSimplexSize(portal) < 4){\n"
+	"		 b3MprSupport(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&dir, b3MprSimplexPointW(portal, 3));\n"
+	"        \n"
+	"        dot = b3MprVec3Dot(&b3MprSimplexPoint(portal, 3)->v, &dir);\n"
+	"        if (b3MprIsZero(dot) || dot < 0.f)\n"
+	"            return -1;\n"
+	"        cont = 0;\n"
+	"        // test if origin is outside (v1, v0, v3) - set v2 as v3 and\n"
+	"        // continue\n"
+	"        b3MprVec3Cross(&va, &b3MprSimplexPoint(portal, 1)->v,\n"
+	"                          &b3MprSimplexPoint(portal, 3)->v);\n"
+	"        dot = b3MprVec3Dot(&va, &b3MprSimplexPoint(portal, 0)->v);\n"
+	"        if (dot < 0.f && !b3MprIsZero(dot)){\n"
+	"            b3MprSimplexSet(portal, 2, b3MprSimplexPoint(portal, 3));\n"
+	"            cont = 1;\n"
+	"        }\n"
+	"        if (!cont){\n"
+	"            // test if origin is outside (v3, v0, v2) - set v1 as v3 and\n"
+	"            // continue\n"
+	"            b3MprVec3Cross(&va, &b3MprSimplexPoint(portal, 3)->v,\n"
+	"                              &b3MprSimplexPoint(portal, 2)->v);\n"
+	"            dot = b3MprVec3Dot(&va, &b3MprSimplexPoint(portal, 0)->v);\n"
+	"            if (dot < 0.f && !b3MprIsZero(dot)){\n"
+	"                b3MprSimplexSet(portal, 1, b3MprSimplexPoint(portal, 3));\n"
+	"                cont = 1;\n"
+	"            }\n"
+	"        }\n"
+	"        if (cont){\n"
+	"            b3MprVec3Sub2(&va, &b3MprSimplexPoint(portal, 1)->v,\n"
+	"                             &b3MprSimplexPoint(portal, 0)->v);\n"
+	"            b3MprVec3Sub2(&vb, &b3MprSimplexPoint(portal, 2)->v,\n"
+	"                             &b3MprSimplexPoint(portal, 0)->v);\n"
+	"            b3MprVec3Cross(&dir, &va, &vb);\n"
+	"            b3MprVec3Normalize(&dir);\n"
+	"        }else{\n"
+	"            b3MprSimplexSetSize(portal, 4);\n"
+	"        }\n"
+	"    }\n"
+	"    return 0;\n"
+	"}\n"
+	"B3_STATIC int b3RefinePortal(int pairIndex,int bodyIndexA, int bodyIndexB,  b3ConstArray(b3RigidBodyData_t) cpuBodyBuf, \n"
+	"													b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData, \n"
+	"													b3ConstArray(b3Collidable_t)				cpuCollidables,\n"
+	"													b3ConstArray(b3Float4)					cpuVertices,\n"
+	"													__global b3Float4* sepAxis,\n"
+	"													b3MprSimplex_t *portal)\n"
+	"{\n"
+	"    b3Float4 dir;\n"
+	"    b3MprSupport_t v4;\n"
+	"	for (int i=0;i<B3_MPR_MAX_ITERATIONS;i++)\n"
+	"    //while (1)\n"
+	"	{\n"
+	"        // compute direction outside the portal (from v0 throught v1,v2,v3\n"
+	"        // face)\n"
+	"        b3PortalDir(portal, &dir);\n"
+	"        // test if origin is inside the portal\n"
+	"        if (portalEncapsulesOrigin(portal, &dir))\n"
+	"            return 0;\n"
+	"        // get next support point\n"
+	"        \n"
+	"		 b3MprSupport(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&dir, &v4);\n"
+	"        // test if v4 can expand portal to contain origin and if portal\n"
+	"        // expanding doesn't reach given tolerance\n"
+	"        if (!portalCanEncapsuleOrigin(portal, &v4, &dir)\n"
+	"                || portalReachTolerance(portal, &v4, &dir))\n"
+	"		{\n"
+	"            return -1;\n"
+	"        }\n"
+	"        // v1-v2-v3 triangle must be rearranged to face outside Minkowski\n"
+	"        // difference (direction from v0).\n"
+	"        b3ExpandPortal(portal, &v4);\n"
+	"    }\n"
+	"    return -1;\n"
+	"}\n"
+	"B3_STATIC void b3FindPos(const b3MprSimplex_t *portal, b3Float4 *pos)\n"
+	"{\n"
+	"	b3Float4 zero = b3MakeFloat4(0,0,0,0);\n"
+	"	b3Float4* b3mpr_vec3_origin = &zero;\n"
+	"    b3Float4 dir;\n"
+	"    size_t i;\n"
+	"    float b[4], sum, inv;\n"
+	"    b3Float4 vec, p1, p2;\n"
+	"    b3PortalDir(portal, &dir);\n"
+	"    // use barycentric coordinates of tetrahedron to find origin\n"
+	"    b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 1)->v,\n"
+	"                       &b3MprSimplexPoint(portal, 2)->v);\n"
+	"    b[0] = b3MprVec3Dot(&vec, &b3MprSimplexPoint(portal, 3)->v);\n"
+	"    b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 3)->v,\n"
+	"                       &b3MprSimplexPoint(portal, 2)->v);\n"
+	"    b[1] = b3MprVec3Dot(&vec, &b3MprSimplexPoint(portal, 0)->v);\n"
+	"    b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 0)->v,\n"
+	"                       &b3MprSimplexPoint(portal, 1)->v);\n"
+	"    b[2] = b3MprVec3Dot(&vec, &b3MprSimplexPoint(portal, 3)->v);\n"
+	"    b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 2)->v,\n"
+	"                       &b3MprSimplexPoint(portal, 1)->v);\n"
+	"    b[3] = b3MprVec3Dot(&vec, &b3MprSimplexPoint(portal, 0)->v);\n"
+	"	sum = b[0] + b[1] + b[2] + b[3];\n"
+	"    if (b3MprIsZero(sum) || sum < 0.f){\n"
+	"		b[0] = 0.f;\n"
+	"        b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 2)->v,\n"
+	"                           &b3MprSimplexPoint(portal, 3)->v);\n"
+	"        b[1] = b3MprVec3Dot(&vec, &dir);\n"
+	"        b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 3)->v,\n"
+	"                           &b3MprSimplexPoint(portal, 1)->v);\n"
+	"        b[2] = b3MprVec3Dot(&vec, &dir);\n"
+	"        b3MprVec3Cross(&vec, &b3MprSimplexPoint(portal, 1)->v,\n"
+	"                           &b3MprSimplexPoint(portal, 2)->v);\n"
+	"        b[3] = b3MprVec3Dot(&vec, &dir);\n"
+	"		sum = b[1] + b[2] + b[3];\n"
+	"	}\n"
+	"	inv = 1.f / sum;\n"
+	"    b3MprVec3Copy(&p1, b3mpr_vec3_origin);\n"
+	"    b3MprVec3Copy(&p2, b3mpr_vec3_origin);\n"
+	"    for (i = 0; i < 4; i++){\n"
+	"        b3MprVec3Copy(&vec, &b3MprSimplexPoint(portal, i)->v1);\n"
+	"        b3MprVec3Scale(&vec, b[i]);\n"
+	"        b3MprVec3Add(&p1, &vec);\n"
+	"        b3MprVec3Copy(&vec, &b3MprSimplexPoint(portal, i)->v2);\n"
+	"        b3MprVec3Scale(&vec, b[i]);\n"
+	"        b3MprVec3Add(&p2, &vec);\n"
+	"    }\n"
+	"    b3MprVec3Scale(&p1, inv);\n"
+	"    b3MprVec3Scale(&p2, inv);\n"
+	"    b3MprVec3Copy(pos, &p1);\n"
+	"    b3MprVec3Add(pos, &p2);\n"
+	"    b3MprVec3Scale(pos, 0.5);\n"
+	"}\n"
+	"inline float b3MprVec3Dist2(const b3Float4 *a, const b3Float4 *b)\n"
+	"{\n"
+	"    b3Float4 ab;\n"
+	"    b3MprVec3Sub2(&ab, a, b);\n"
+	"    return b3MprVec3Len2(&ab);\n"
+	"}\n"
+	"inline float _b3MprVec3PointSegmentDist2(const b3Float4 *P,\n"
+	"                                                  const b3Float4 *x0,\n"
+	"                                                  const b3Float4 *b,\n"
+	"                                                  b3Float4 *witness)\n"
+	"{\n"
+	"    // The computation comes from solving equation of segment:\n"
+	"    //      S(t) = x0 + t.d\n"
+	"    //          where - x0 is initial point of segment\n"
+	"    //                - d is direction of segment from x0 (|d| > 0)\n"
+	"    //                - t belongs to <0, 1> interval\n"
+	"    // \n"
+	"    // Than, distance from a segment to some point P can be expressed:\n"
+	"    //      D(t) = |x0 + t.d - P|^2\n"
+	"    //          which is distance from any point on segment. Minimization\n"
+	"    //          of this function brings distance from P to segment.\n"
+	"    // Minimization of D(t) leads to simple quadratic equation that's\n"
+	"    // solving is straightforward.\n"
+	"    //\n"
+	"    // Bonus of this method is witness point for free.\n"
+	"    float dist, t;\n"
+	"    b3Float4 d, a;\n"
+	"    // direction of segment\n"
+	"    b3MprVec3Sub2(&d, b, x0);\n"
+	"    // precompute vector from P to x0\n"
+	"    b3MprVec3Sub2(&a, x0, P);\n"
+	"    t  = -1.f * b3MprVec3Dot(&a, &d);\n"
+	"    t /= b3MprVec3Len2(&d);\n"
+	"    if (t < 0.f || b3MprIsZero(t)){\n"
+	"        dist = b3MprVec3Dist2(x0, P);\n"
+	"        if (witness)\n"
+	"            b3MprVec3Copy(witness, x0);\n"
+	"    }else if (t > 1.f || b3MprEq(t, 1.f)){\n"
+	"        dist = b3MprVec3Dist2(b, P);\n"
+	"        if (witness)\n"
+	"            b3MprVec3Copy(witness, b);\n"
+	"    }else{\n"
+	"        if (witness){\n"
+	"            b3MprVec3Copy(witness, &d);\n"
+	"            b3MprVec3Scale(witness, t);\n"
+	"            b3MprVec3Add(witness, x0);\n"
+	"            dist = b3MprVec3Dist2(witness, P);\n"
+	"        }else{\n"
+	"            // recycling variables\n"
+	"            b3MprVec3Scale(&d, t);\n"
+	"            b3MprVec3Add(&d, &a);\n"
+	"            dist = b3MprVec3Len2(&d);\n"
+	"        }\n"
+	"    }\n"
+	"    return dist;\n"
+	"}\n"
+	"inline float b3MprVec3PointTriDist2(const b3Float4 *P,\n"
+	"                                const b3Float4 *x0, const b3Float4 *B,\n"
+	"                                const b3Float4 *C,\n"
+	"                                b3Float4 *witness)\n"
+	"{\n"
+	"    // Computation comes from analytic expression for triangle (x0, B, C)\n"
+	"    //      T(s, t) = x0 + s.d1 + t.d2, where d1 = B - x0 and d2 = C - x0 and\n"
+	"    // Then equation for distance is:\n"
+	"    //      D(s, t) = | T(s, t) - P |^2\n"
+	"    // This leads to minimization of quadratic function of two variables.\n"
+	"    // The solution from is taken only if s is between 0 and 1, t is\n"
+	"    // between 0 and 1 and t + s < 1, otherwise distance from segment is\n"
+	"    // computed.\n"
+	"    b3Float4 d1, d2, a;\n"
+	"    float u, v, w, p, q, r;\n"
+	"    float s, t, dist, dist2;\n"
+	"    b3Float4 witness2;\n"
+	"    b3MprVec3Sub2(&d1, B, x0);\n"
+	"    b3MprVec3Sub2(&d2, C, x0);\n"
+	"    b3MprVec3Sub2(&a, x0, P);\n"
+	"    u = b3MprVec3Dot(&a, &a);\n"
+	"    v = b3MprVec3Dot(&d1, &d1);\n"
+	"    w = b3MprVec3Dot(&d2, &d2);\n"
+	"    p = b3MprVec3Dot(&a, &d1);\n"
+	"    q = b3MprVec3Dot(&a, &d2);\n"
+	"    r = b3MprVec3Dot(&d1, &d2);\n"
+	"    s = (q * r - w * p) / (w * v - r * r);\n"
+	"    t = (-s * r - q) / w;\n"
+	"    if ((b3MprIsZero(s) || s > 0.f)\n"
+	"            && (b3MprEq(s, 1.f) || s < 1.f)\n"
+	"            && (b3MprIsZero(t) || t > 0.f)\n"
+	"            && (b3MprEq(t, 1.f) || t < 1.f)\n"
+	"            && (b3MprEq(t + s, 1.f) || t + s < 1.f)){\n"
+	"        if (witness){\n"
+	"            b3MprVec3Scale(&d1, s);\n"
+	"            b3MprVec3Scale(&d2, t);\n"
+	"            b3MprVec3Copy(witness, x0);\n"
+	"            b3MprVec3Add(witness, &d1);\n"
+	"            b3MprVec3Add(witness, &d2);\n"
+	"            dist = b3MprVec3Dist2(witness, P);\n"
+	"        }else{\n"
+	"            dist  = s * s * v;\n"
+	"            dist += t * t * w;\n"
+	"            dist += 2.f * s * t * r;\n"
+	"            dist += 2.f * s * p;\n"
+	"            dist += 2.f * t * q;\n"
+	"            dist += u;\n"
+	"        }\n"
+	"    }else{\n"
+	"        dist = _b3MprVec3PointSegmentDist2(P, x0, B, witness);\n"
+	"        dist2 = _b3MprVec3PointSegmentDist2(P, x0, C, &witness2);\n"
+	"        if (dist2 < dist){\n"
+	"            dist = dist2;\n"
+	"            if (witness)\n"
+	"                b3MprVec3Copy(witness, &witness2);\n"
+	"        }\n"
+	"        dist2 = _b3MprVec3PointSegmentDist2(P, B, C, &witness2);\n"
+	"        if (dist2 < dist){\n"
+	"            dist = dist2;\n"
+	"            if (witness)\n"
+	"                b3MprVec3Copy(witness, &witness2);\n"
+	"        }\n"
+	"    }\n"
+	"    return dist;\n"
+	"}\n"
+	"B3_STATIC void b3FindPenetr(int pairIndex,int bodyIndexA, int bodyIndexB,  b3ConstArray(b3RigidBodyData_t) cpuBodyBuf, \n"
+	"													b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData, \n"
+	"													b3ConstArray(b3Collidable_t)				cpuCollidables,\n"
+	"													b3ConstArray(b3Float4)					cpuVertices,\n"
+	"													__global b3Float4* sepAxis,\n"
+	"                       b3MprSimplex_t *portal,\n"
+	"                       float *depth, b3Float4 *pdir, b3Float4 *pos)\n"
+	"{\n"
+	"    b3Float4 dir;\n"
+	"    b3MprSupport_t v4;\n"
+	"    unsigned long iterations;\n"
+	"	b3Float4 zero = b3MakeFloat4(0,0,0,0);\n"
+	"	b3Float4* b3mpr_vec3_origin = &zero;\n"
+	"    iterations = 1UL;\n"
+	"	for (int i=0;i<B3_MPR_MAX_ITERATIONS;i++)\n"
+	"    //while (1)\n"
+	"	{\n"
+	"        // compute portal direction and obtain next support point\n"
+	"        b3PortalDir(portal, &dir);\n"
+	"        \n"
+	"		 b3MprSupport(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&dir, &v4);\n"
+	"        // reached tolerance -> find penetration info\n"
+	"        if (portalReachTolerance(portal, &v4, &dir)\n"
+	"                || iterations ==B3_MPR_MAX_ITERATIONS)\n"
+	"		{\n"
+	"            *depth = b3MprVec3PointTriDist2(b3mpr_vec3_origin,&b3MprSimplexPoint(portal, 1)->v,&b3MprSimplexPoint(portal, 2)->v,&b3MprSimplexPoint(portal, 3)->v,pdir);\n"
+	"            *depth = B3_MPR_SQRT(*depth);\n"
+	"			\n"
+	"			if (b3MprIsZero((*pdir).x) && b3MprIsZero((*pdir).y) && b3MprIsZero((*pdir).z))\n"
+	"			{\n"
+	"				\n"
+	"				*pdir = dir;\n"
+	"			} \n"
+	"			b3MprVec3Normalize(pdir);\n"
+	"			\n"
+	"            // barycentric coordinates:\n"
+	"            b3FindPos(portal, pos);\n"
+	"            return;\n"
+	"        }\n"
+	"        b3ExpandPortal(portal, &v4);\n"
+	"        iterations++;\n"
+	"    }\n"
+	"}\n"
+	"B3_STATIC void b3FindPenetrTouch(b3MprSimplex_t *portal,float *depth, b3Float4 *dir, b3Float4 *pos)\n"
+	"{\n"
+	"    // Touching contact on portal's v1 - so depth is zero and direction\n"
+	"    // is unimportant and pos can be guessed\n"
+	"    *depth = 0.f;\n"
+	"    b3Float4 zero = b3MakeFloat4(0,0,0,0);\n"
+	"	b3Float4* b3mpr_vec3_origin = &zero;\n"
+	"	b3MprVec3Copy(dir, b3mpr_vec3_origin);\n"
+	"    b3MprVec3Copy(pos, &b3MprSimplexPoint(portal, 1)->v1);\n"
+	"    b3MprVec3Add(pos, &b3MprSimplexPoint(portal, 1)->v2);\n"
+	"    b3MprVec3Scale(pos, 0.5);\n"
+	"}\n"
+	"B3_STATIC void b3FindPenetrSegment(b3MprSimplex_t *portal,\n"
+	"                              float *depth, b3Float4 *dir, b3Float4 *pos)\n"
+	"{\n"
+	"    \n"
+	"    // Origin lies on v0-v1 segment.\n"
+	"    // Depth is distance to v1, direction also and position must be\n"
+	"    // computed\n"
+	"    b3MprVec3Copy(pos, &b3MprSimplexPoint(portal, 1)->v1);\n"
+	"    b3MprVec3Add(pos, &b3MprSimplexPoint(portal, 1)->v2);\n"
+	"    b3MprVec3Scale(pos, 0.5f);\n"
+	"    \n"
+	"    b3MprVec3Copy(dir, &b3MprSimplexPoint(portal, 1)->v);\n"
+	"    *depth = B3_MPR_SQRT(b3MprVec3Len2(dir));\n"
+	"    b3MprVec3Normalize(dir);\n"
+	"}\n"
+	"inline int b3MprPenetration(int pairIndex, int bodyIndexA, int bodyIndexB,\n"
+	"					b3ConstArray(b3RigidBodyData_t) cpuBodyBuf,\n"
+	"					b3ConstArray(b3ConvexPolyhedronData_t) cpuConvexData, \n"
+	"					b3ConstArray(b3Collidable_t)	cpuCollidables,\n"
+	"					b3ConstArray(b3Float4)	cpuVertices,\n"
+	"					__global b3Float4* sepAxis,\n"
+	"					__global int*	hasSepAxis,\n"
+	"					float *depthOut, b3Float4* dirOut, b3Float4* posOut)\n"
+	"{\n"
+	"	\n"
+	"	 b3MprSimplex_t portal;\n"
+	"	 \n"
+	"//	if (!hasSepAxis[pairIndex])\n"
+	"	//	return -1;\n"
+	"	\n"
+	"	hasSepAxis[pairIndex] = 0;\n"
+	"	 int res;\n"
+	"    // Phase 1: Portal discovery\n"
+	"    res = b3DiscoverPortal(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices,sepAxis,hasSepAxis, &portal);\n"
+	"	\n"
+	"	  \n"
+	"	//sepAxis[pairIndex] = *pdir;//or -dir?\n"
+	"	switch (res)\n"
+	"	{\n"
+	"	case 0:\n"
+	"		{\n"
+	"			// Phase 2: Portal refinement\n"
+	"		\n"
+	"			res = b3RefinePortal(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&portal);\n"
+	"			if (res < 0)\n"
+	"				return -1;\n"
+	"			// Phase 3. Penetration info\n"
+	"			b3FindPenetr(pairIndex,bodyIndexA,bodyIndexB,cpuBodyBuf,cpuConvexData,cpuCollidables,cpuVertices, sepAxis,&portal, depthOut, dirOut, posOut);\n"
+	"			hasSepAxis[pairIndex] = 1;\n"
+	"			sepAxis[pairIndex] = -*dirOut;\n"
+	"			break;\n"
+	"		}\n"
+	"	case 1:\n"
+	"		{\n"
+	"			 // Touching contact on portal's v1.\n"
+	"			b3FindPenetrTouch(&portal, depthOut, dirOut, posOut);\n"
+	"			break;\n"
+	"		}\n"
+	"	case 2:\n"
+	"		{\n"
+	"			\n"
+	"			b3FindPenetrSegment( &portal, depthOut, dirOut, posOut);\n"
+	"			break;\n"
+	"		}\n"
+	"	default:\n"
+	"		{\n"
+	"			hasSepAxis[pairIndex]=0;\n"
+	"			//if (res < 0)\n"
+	"			//{\n"
+	"				// Origin isn't inside portal - no collision.\n"
+	"				return -1;\n"
+	"			//}\n"
+	"		}\n"
+	"	};\n"
+	"	\n"
+	"	return 0;\n"
+	"};\n"
+	"#endif //B3_MPR_PENETRATION_H\n"
+	"#ifndef B3_CONTACT4DATA_H\n"
+	"#define B3_CONTACT4DATA_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_FLOAT4_H\n"
+	"typedef  struct b3Contact4Data b3Contact4Data_t;\n"
+	"struct b3Contact4Data\n"
+	"{\n"
+	"	b3Float4	m_worldPosB[4];\n"
+	"//	b3Float4	m_localPosA[4];\n"
+	"//	b3Float4	m_localPosB[4];\n"
+	"	b3Float4	m_worldNormalOnB;	//	w: m_nPoints\n"
+	"	unsigned short  m_restituitionCoeffCmp;\n"
+	"	unsigned short  m_frictionCoeffCmp;\n"
+	"	int m_batchIdx;\n"
+	"	int m_bodyAPtrAndSignBit;//x:m_bodyAPtr, y:m_bodyBPtr\n"
+	"	int m_bodyBPtrAndSignBit;\n"
+	"	int	m_childIndexA;\n"
+	"	int	m_childIndexB;\n"
+	"	int m_unused1;\n"
+	"	int m_unused2;\n"
+	"};\n"
+	"inline int b3Contact4Data_getNumPoints(const struct b3Contact4Data* contact)\n"
+	"{\n"
+	"	return (int)contact->m_worldNormalOnB.w;\n"
+	"};\n"
+	"inline void b3Contact4Data_setNumPoints(struct b3Contact4Data* contact, int numPoints)\n"
+	"{\n"
+	"	contact->m_worldNormalOnB.w = (float)numPoints;\n"
+	"};\n"
+	"#endif //B3_CONTACT4DATA_H\n"
+	"#define AppendInc(x, out) out = atomic_inc(x)\n"
+	"#define GET_NPOINTS(x) (x).m_worldNormalOnB.w\n"
+	"#ifdef cl_ext_atomic_counters_32\n"
+	"	#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
+	"#else\n"
+	"	#define counter32_t volatile __global int*\n"
+	"#endif\n"
+	"__kernel void   mprPenetrationKernel( __global int4* pairs,\n"
+	"																					__global const b3RigidBodyData_t* rigidBodies, \n"
+	"																					__global const b3Collidable_t* collidables,\n"
+	"																					__global const b3ConvexPolyhedronData_t* convexShapes, \n"
+	"																					__global const float4* vertices,\n"
+	"																					__global float4* separatingNormals,\n"
+	"																					__global int* hasSeparatingAxis,\n"
+	"																					__global struct b3Contact4Data* restrict globalContactsOut,\n"
+	"																					counter32_t nGlobalContactsOut,\n"
+	"																					int contactCapacity,\n"
+	"																					int numPairs)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	int pairIndex = i;\n"
+	"	if (i<numPairs)\n"
+	"	{\n"
+	"		int bodyIndexA = pairs[i].x;\n"
+	"		int bodyIndexB = pairs[i].y;\n"
+	"		int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
+	"		int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
+	"	\n"
+	"		int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
+	"		int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
+	"		\n"
+	"		\n"
+	"		//once the broadphase avoids static-static pairs, we can remove this test\n"
+	"		if ((rigidBodies[bodyIndexA].m_invMass==0) &&(rigidBodies[bodyIndexB].m_invMass==0))\n"
+	"		{\n"
+	"			return;\n"
+	"		}\n"
+	"		\n"
+	"		if ((collidables[collidableIndexA].m_shapeType!=SHAPE_CONVEX_HULL) ||(collidables[collidableIndexB].m_shapeType!=SHAPE_CONVEX_HULL))\n"
+	"		{\n"
+	"			return;\n"
+	"		}\n"
+	"		float depthOut;\n"
+	"		b3Float4 dirOut;\n"
+	"		b3Float4 posOut;\n"
+	"		int res = b3MprPenetration(pairIndex, bodyIndexA, bodyIndexB,rigidBodies,convexShapes,collidables,vertices,separatingNormals,hasSeparatingAxis,&depthOut, &dirOut, &posOut);\n"
+	"		\n"
+	"		\n"
+	"		\n"
+	"		\n"
+	"		if (res==0)\n"
+	"		{\n"
+	"			//add a contact\n"
+	"			int dstIdx;\n"
+	"			AppendInc( nGlobalContactsOut, dstIdx );\n"
+	"			if (dstIdx<contactCapacity)\n"
+	"			{\n"
+	"				pairs[pairIndex].z = dstIdx;\n"
+	"				__global struct b3Contact4Data* c = globalContactsOut + dstIdx;\n"
+	"				c->m_worldNormalOnB = -dirOut;//normal;\n"
+	"				c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
+	"				c->m_batchIdx = pairIndex;\n"
+	"				int bodyA = pairs[pairIndex].x;\n"
+	"				int bodyB = pairs[pairIndex].y;\n"
+	"				c->m_bodyAPtrAndSignBit = rigidBodies[bodyA].m_invMass==0 ? -bodyA:bodyA;\n"
+	"				c->m_bodyBPtrAndSignBit = rigidBodies[bodyB].m_invMass==0 ? -bodyB:bodyB;\n"
+	"				c->m_childIndexA = -1;\n"
+	"				c->m_childIndexB = -1;\n"
+	"				//for (int i=0;i<nContacts;i++)\n"
+	"				posOut.w = -depthOut;\n"
+	"				c->m_worldPosB[0] = posOut;//localPoints[contactIdx[i]];\n"
+	"				GET_NPOINTS(*c) = 1;//nContacts;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"typedef float4 Quaternion;\n"
+	"#define make_float4 (float4)\n"
+	"__inline\n"
+	"float dot3F4(float4 a, float4 b)\n"
+	"{\n"
+	"	float4 a1 = make_float4(a.xyz,0.f);\n"
+	"	float4 b1 = make_float4(b.xyz,0.f);\n"
+	"	return dot(a1, b1);\n"
+	"}\n"
+	"__inline\n"
+	"float4 cross3(float4 a, float4 b)\n"
+	"{\n"
+	"	return cross(a,b);\n"
+	"}\n"
+	"__inline\n"
+	"Quaternion qtMul(Quaternion a, Quaternion b)\n"
+	"{\n"
+	"	Quaternion ans;\n"
+	"	ans = cross3( a, b );\n"
+	"	ans += a.w*b+b.w*a;\n"
+	"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
+	"	ans.w = a.w*b.w - dot3F4(a, b);\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"Quaternion qtInvert(Quaternion q)\n"
+	"{\n"
+	"	return (Quaternion)(-q.xyz, q.w);\n"
+	"}\n"
+	"__inline\n"
+	"float4 qtRotate(Quaternion q, float4 vec)\n"
+	"{\n"
+	"	Quaternion qInv = qtInvert( q );\n"
+	"	float4 vcpy = vec;\n"
+	"	vcpy.w = 0.f;\n"
+	"	float4 out = qtMul(qtMul(q,vcpy),qInv);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"float4 transform(const float4* p, const float4* translation, const Quaternion* orientation)\n"
+	"{\n"
+	"	return qtRotate( *orientation, *p ) + (*translation);\n"
+	"}\n"
+	"__inline\n"
+	"float4 qtInvRotate(const Quaternion q, float4 vec)\n"
+	"{\n"
+	"	return qtRotate( qtInvert( q ), vec );\n"
+	"}\n"
+	"inline void project(__global const b3ConvexPolyhedronData_t* hull,  const float4 pos, const float4 orn, \n"
+	"const float4* dir, __global const float4* vertices, float* min, float* max)\n"
+	"{\n"
+	"	min[0] = FLT_MAX;\n"
+	"	max[0] = -FLT_MAX;\n"
+	"	int numVerts = hull->m_numVertices;\n"
+	"	const float4 localDir = qtInvRotate(orn,*dir);\n"
+	"	float offset = dot(pos,*dir);\n"
+	"	for(int i=0;i<numVerts;i++)\n"
+	"	{\n"
+	"		float dp = dot(vertices[hull->m_vertexOffset+i],localDir);\n"
+	"		if(dp < min[0])	\n"
+	"			min[0] = dp;\n"
+	"		if(dp > max[0])	\n"
+	"			max[0] = dp;\n"
+	"	}\n"
+	"	if(min[0]>max[0])\n"
+	"	{\n"
+	"		float tmp = min[0];\n"
+	"		min[0] = max[0];\n"
+	"		max[0] = tmp;\n"
+	"	}\n"
+	"	min[0] += offset;\n"
+	"	max[0] += offset;\n"
+	"}\n"
+	"bool findSeparatingAxisUnitSphere(	__global const b3ConvexPolyhedronData_t* hullA, __global const b3ConvexPolyhedronData_t* hullB, \n"
+	"	const float4 posA1,\n"
+	"	const float4 ornA,\n"
+	"	const float4 posB1,\n"
+	"	const float4 ornB,\n"
+	"	const float4 DeltaC2,\n"
+	"	__global const float4* vertices,\n"
+	"	__global const float4* unitSphereDirections,\n"
+	"	int numUnitSphereDirections,\n"
+	"	float4* sep,\n"
+	"	float* dmin)\n"
+	"{\n"
+	"	\n"
+	"	float4 posA = posA1;\n"
+	"	posA.w = 0.f;\n"
+	"	float4 posB = posB1;\n"
+	"	posB.w = 0.f;\n"
+	"	int curPlaneTests=0;\n"
+	"	int curEdgeEdge = 0;\n"
+	"	// Test unit sphere directions\n"
+	"	for (int i=0;i<numUnitSphereDirections;i++)\n"
+	"	{\n"
+	"		float4 crossje;\n"
+	"		crossje = unitSphereDirections[i];	\n"
+	"		if (dot3F4(DeltaC2,crossje)>0)\n"
+	"			crossje *= -1.f;\n"
+	"		{\n"
+	"			float dist;\n"
+	"			bool result = true;\n"
+	"			float Min0,Max0;\n"
+	"			float Min1,Max1;\n"
+	"			project(hullA,posA,ornA,&crossje,vertices, &Min0, &Max0);\n"
+	"			project(hullB,posB,ornB,&crossje,vertices, &Min1, &Max1);\n"
+	"		\n"
+	"			if(Max0<Min1 || Max1<Min0)\n"
+	"				return false;\n"
+	"		\n"
+	"			float d0 = Max0 - Min1;\n"
+	"			float d1 = Max1 - Min0;\n"
+	"			dist = d0<d1 ? d0:d1;\n"
+	"			result = true;\n"
+	"	\n"
+	"			if(dist<*dmin)\n"
+	"			{\n"
+	"				*dmin = dist;\n"
+	"				*sep = crossje;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	\n"
+	"	if((dot3F4(-DeltaC2,*sep))>0.0f)\n"
+	"	{\n"
+	"		*sep = -(*sep);\n"
+	"	}\n"
+	"	return true;\n"
+	"}\n"
+	"__kernel void   findSeparatingAxisUnitSphereKernel( __global const int4* pairs, \n"
+	"																					__global const b3RigidBodyData_t* rigidBodies, \n"
+	"																					__global const b3Collidable_t* collidables,\n"
+	"																					__global const b3ConvexPolyhedronData_t* convexShapes, \n"
+	"																					__global const float4* vertices,\n"
+	"																					__global const float4* unitSphereDirections,\n"
+	"																					__global  float4* separatingNormals,\n"
+	"																					__global  int* hasSeparatingAxis,\n"
+	"																					__global  float* dmins,\n"
+	"																					int numUnitSphereDirections,\n"
+	"																					int numPairs\n"
+	"																					)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	\n"
+	"	if (i<numPairs)\n"
+	"	{\n"
+	"		if (hasSeparatingAxis[i])\n"
+	"		{\n"
+	"	\n"
+	"			int bodyIndexA = pairs[i].x;\n"
+	"			int bodyIndexB = pairs[i].y;\n"
+	"	\n"
+	"			int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
+	"			int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
+	"		\n"
+	"			int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
+	"			int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
+	"			\n"
+	"			\n"
+	"			int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
+	"	\n"
+	"			float dmin = dmins[i];\n"
+	"	\n"
+	"			float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
+	"			posA.w = 0.f;\n"
+	"			float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
+	"			posB.w = 0.f;\n"
+	"			float4 c0local = convexShapes[shapeIndexA].m_localCenter;\n"
+	"			float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
+	"			float4 c0 = transform(&c0local, &posA, &ornA);\n"
+	"			float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
+	"			float4 ornB =rigidBodies[bodyIndexB].m_quat;\n"
+	"			float4 c1 = transform(&c1local,&posB,&ornB);\n"
+	"			const float4 DeltaC2 = c0 - c1;\n"
+	"			float4 sepNormal = separatingNormals[i];\n"
+	"			\n"
+	"			int numEdgeEdgeDirections = convexShapes[shapeIndexA].m_numUniqueEdges*convexShapes[shapeIndexB].m_numUniqueEdges;\n"
+	"			if (numEdgeEdgeDirections>numUnitSphereDirections)\n"
+	"			{\n"
+	"				bool sepEE = findSeparatingAxisUnitSphere(	&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],posA,ornA,\n"
+	"																										posB,ornB,\n"
+	"																										DeltaC2,\n"
+	"																										vertices,unitSphereDirections,numUnitSphereDirections,&sepNormal,&dmin);\n"
+	"				if (!sepEE)\n"
+	"				{\n"
+	"					hasSeparatingAxis[i] = 0;\n"
+	"				} else\n"
+	"				{\n"
+	"					hasSeparatingAxis[i] = 1;\n"
+	"					separatingNormals[i] = sepNormal;\n"
+	"				}\n"
+	"			}\n"
+	"		}		//if (hasSeparatingAxis[i])\n"
+	"	}//(i<numPairs)\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/primitiveContacts.h b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/primitiveContacts.h
index b0103fe6748..b2e0a2dd472 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/primitiveContacts.h
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/primitiveContacts.h
@@ -1,1289 +1,1288 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* primitiveContactsKernelsCL= \
-"#ifndef B3_CONTACT4DATA_H\n"
-"#define B3_CONTACT4DATA_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#define B3_FLOAT4_H\n"
-"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
-"#define B3_PLATFORM_DEFINITIONS_H\n"
-"struct MyTest\n"
-"{\n"
-"	int bla;\n"
-"};\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
-"#define B3_LARGE_FLOAT 1e18f\n"
-"#define B3_INFINITY 1e18f\n"
-"#define b3Assert(a)\n"
-"#define b3ConstArray(a) __global const a*\n"
-"#define b3AtomicInc atomic_inc\n"
-"#define b3AtomicAdd atomic_add\n"
-"#define b3Fabs fabs\n"
-"#define b3Sqrt native_sqrt\n"
-"#define b3Sin native_sin\n"
-"#define b3Cos native_cos\n"
-"#define B3_STATIC\n"
-"#endif\n"
-"#endif\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"	typedef float4	b3Float4;\n"
-"	#define b3Float4ConstArg const b3Float4\n"
-"	#define b3MakeFloat4 (float4)\n"
-"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return dot(a1, b1);\n"
-"	}\n"
-"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return cross(a1, b1);\n"
-"	}\n"
-"	#define b3MinFloat4 min\n"
-"	#define b3MaxFloat4 max\n"
-"	#define b3Normalized(a) normalize(a)\n"
-"#endif \n"
-"		\n"
-"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
-"{\n"
-"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
-"		return false;\n"
-"	return true;\n"
-"}\n"
-"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
-"{\n"
-"    float maxDot = -B3_INFINITY;\n"
-"    int i = 0;\n"
-"    int ptIndex = -1;\n"
-"    for( i = 0; i < vecLen; i++ )\n"
-"    {\n"
-"        float dot = b3Dot3F4(vecArray[i],vec);\n"
-"            \n"
-"        if( dot > maxDot )\n"
-"        {\n"
-"            maxDot = dot;\n"
-"            ptIndex = i;\n"
-"        }\n"
-"    }\n"
-"	b3Assert(ptIndex>=0);\n"
-"    if (ptIndex<0)\n"
-"	{\n"
-"		ptIndex = 0;\n"
-"	}\n"
-"    *dotOut = maxDot;\n"
-"    return ptIndex;\n"
-"}\n"
-"#endif //B3_FLOAT4_H\n"
-"typedef  struct b3Contact4Data b3Contact4Data_t;\n"
-"struct b3Contact4Data\n"
-"{\n"
-"	b3Float4	m_worldPosB[4];\n"
-"//	b3Float4	m_localPosA[4];\n"
-"//	b3Float4	m_localPosB[4];\n"
-"	b3Float4	m_worldNormalOnB;	//	w: m_nPoints\n"
-"	unsigned short  m_restituitionCoeffCmp;\n"
-"	unsigned short  m_frictionCoeffCmp;\n"
-"	int m_batchIdx;\n"
-"	int m_bodyAPtrAndSignBit;//x:m_bodyAPtr, y:m_bodyBPtr\n"
-"	int m_bodyBPtrAndSignBit;\n"
-"	int	m_childIndexA;\n"
-"	int	m_childIndexB;\n"
-"	int m_unused1;\n"
-"	int m_unused2;\n"
-"};\n"
-"inline int b3Contact4Data_getNumPoints(const struct b3Contact4Data* contact)\n"
-"{\n"
-"	return (int)contact->m_worldNormalOnB.w;\n"
-"};\n"
-"inline void b3Contact4Data_setNumPoints(struct b3Contact4Data* contact, int numPoints)\n"
-"{\n"
-"	contact->m_worldNormalOnB.w = (float)numPoints;\n"
-"};\n"
-"#endif //B3_CONTACT4DATA_H\n"
-"#define SHAPE_CONVEX_HULL 3\n"
-"#define SHAPE_PLANE 4\n"
-"#define SHAPE_CONCAVE_TRIMESH 5\n"
-"#define SHAPE_COMPOUND_OF_CONVEX_HULLS 6\n"
-"#define SHAPE_SPHERE 7\n"
-"#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_extended_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics : enable\n"
-"#ifdef cl_ext_atomic_counters_32\n"
-"#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
-"#else\n"
-"#define counter32_t volatile __global int*\n"
-"#endif\n"
-"#define GET_GROUP_IDX get_group_id(0)\n"
-"#define GET_LOCAL_IDX get_local_id(0)\n"
-"#define GET_GLOBAL_IDX get_global_id(0)\n"
-"#define GET_GROUP_SIZE get_local_size(0)\n"
-"#define GET_NUM_GROUPS get_num_groups(0)\n"
-"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
-"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
-"#define AtomInc(x) atom_inc(&(x))\n"
-"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
-"#define AppendInc(x, out) out = atomic_inc(x)\n"
-"#define AtomAdd(x, value) atom_add(&(x), value)\n"
-"#define AtomCmpxhg(x, cmp, value) atom_cmpxchg( &(x), cmp, value )\n"
-"#define AtomXhg(x, value) atom_xchg ( &(x), value )\n"
-"#define max2 max\n"
-"#define min2 min\n"
-"typedef unsigned int u32;\n"
-"typedef struct \n"
-"{\n"
-"	union\n"
-"	{\n"
-"		float4	m_min;\n"
-"		float   m_minElems[4];\n"
-"		int			m_minIndices[4];\n"
-"	};\n"
-"	union\n"
-"	{\n"
-"		float4	m_max;\n"
-"		float   m_maxElems[4];\n"
-"		int			m_maxIndices[4];\n"
-"	};\n"
-"} btAabbCL;\n"
-"///keep this in sync with btCollidable.h\n"
-"typedef struct\n"
-"{\n"
-"	int m_numChildShapes;\n"
-"	float m_radius;\n"
-"	int m_shapeType;\n"
-"	int m_shapeIndex;\n"
-"	\n"
-"} btCollidableGpu;\n"
-"typedef struct\n"
-"{\n"
-"	float4	m_childPosition;\n"
-"	float4	m_childOrientation;\n"
-"	int m_shapeIndex;\n"
-"	int m_unused0;\n"
-"	int m_unused1;\n"
-"	int m_unused2;\n"
-"} btGpuChildShape;\n"
-"#define GET_NPOINTS(x) (x).m_worldNormalOnB.w\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_pos;\n"
-"	float4 m_quat;\n"
-"	float4 m_linVel;\n"
-"	float4 m_angVel;\n"
-"	u32 m_collidableIdx;	\n"
-"	float m_invMass;\n"
-"	float m_restituitionCoeff;\n"
-"	float m_frictionCoeff;\n"
-"} BodyData;\n"
-"typedef struct  \n"
-"{\n"
-"	float4		m_localCenter;\n"
-"	float4		m_extents;\n"
-"	float4		mC;\n"
-"	float4		mE;\n"
-"	\n"
-"	float			m_radius;\n"
-"	int	m_faceOffset;\n"
-"	int m_numFaces;\n"
-"	int	m_numVertices;\n"
-"	\n"
-"	int m_vertexOffset;\n"
-"	int	m_uniqueEdgesOffset;\n"
-"	int	m_numUniqueEdges;\n"
-"	int m_unused;\n"
-"} ConvexPolyhedronCL;\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_plane;\n"
-"	int m_indexOffset;\n"
-"	int m_numIndices;\n"
-"} btGpuFace;\n"
-"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
-"#define make_float4 (float4)\n"
-"#define make_float2 (float2)\n"
-"#define make_uint4 (uint4)\n"
-"#define make_int4 (int4)\n"
-"#define make_uint2 (uint2)\n"
-"#define make_int2 (int2)\n"
-"__inline\n"
-"float fastDiv(float numerator, float denominator)\n"
-"{\n"
-"	return native_divide(numerator, denominator);	\n"
-"//	return numerator/denominator;	\n"
-"}\n"
-"__inline\n"
-"float4 fastDiv4(float4 numerator, float4 denominator)\n"
-"{\n"
-"	return native_divide(numerator, denominator);	\n"
-"}\n"
-"__inline\n"
-"float4 cross3(float4 a, float4 b)\n"
-"{\n"
-"	return cross(a,b);\n"
-"}\n"
-"//#define dot3F4 dot\n"
-"__inline\n"
-"float dot3F4(float4 a, float4 b)\n"
-"{\n"
-"	float4 a1 = make_float4(a.xyz,0.f);\n"
-"	float4 b1 = make_float4(b.xyz,0.f);\n"
-"	return dot(a1, b1);\n"
-"}\n"
-"__inline\n"
-"float4 fastNormalize4(float4 v)\n"
-"{\n"
-"	return fast_normalize(v);\n"
-"}\n"
-"///////////////////////////////////////\n"
-"//	Quaternion\n"
-"///////////////////////////////////////\n"
-"typedef float4 Quaternion;\n"
-"__inline\n"
-"Quaternion qtMul(Quaternion a, Quaternion b);\n"
-"__inline\n"
-"Quaternion qtNormalize(Quaternion in);\n"
-"__inline\n"
-"float4 qtRotate(Quaternion q, float4 vec);\n"
-"__inline\n"
-"Quaternion qtInvert(Quaternion q);\n"
-"__inline\n"
-"Quaternion qtMul(Quaternion a, Quaternion b)\n"
-"{\n"
-"	Quaternion ans;\n"
-"	ans = cross3( a, b );\n"
-"	ans += a.w*b+b.w*a;\n"
-"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
-"	ans.w = a.w*b.w - dot3F4(a, b);\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"Quaternion qtNormalize(Quaternion in)\n"
-"{\n"
-"	return fastNormalize4(in);\n"
-"//	in /= length( in );\n"
-"//	return in;\n"
-"}\n"
-"__inline\n"
-"float4 qtRotate(Quaternion q, float4 vec)\n"
-"{\n"
-"	Quaternion qInv = qtInvert( q );\n"
-"	float4 vcpy = vec;\n"
-"	vcpy.w = 0.f;\n"
-"	float4 out = qtMul(qtMul(q,vcpy),qInv);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"Quaternion qtInvert(Quaternion q)\n"
-"{\n"
-"	return (Quaternion)(-q.xyz, q.w);\n"
-"}\n"
-"__inline\n"
-"float4 qtInvRotate(const Quaternion q, float4 vec)\n"
-"{\n"
-"	return qtRotate( qtInvert( q ), vec );\n"
-"}\n"
-"__inline\n"
-"float4 transform(const float4* p, const float4* translation, const Quaternion* orientation)\n"
-"{\n"
-"	return qtRotate( *orientation, *p ) + (*translation);\n"
-"}\n"
-"void	trInverse(float4 translationIn, Quaternion orientationIn,\n"
-"		float4* translationOut, Quaternion* orientationOut)\n"
-"{\n"
-"	*orientationOut = qtInvert(orientationIn);\n"
-"	*translationOut = qtRotate(*orientationOut, -translationIn);\n"
-"}\n"
-"void	trMul(float4 translationA, Quaternion orientationA,\n"
-"						float4 translationB, Quaternion orientationB,\n"
-"		float4* translationOut, Quaternion* orientationOut)\n"
-"{\n"
-"	*orientationOut = qtMul(orientationA,orientationB);\n"
-"	*translationOut = transform(&translationB,&translationA,&orientationA);\n"
-"}\n"
-"__inline\n"
-"float4 normalize3(const float4 a)\n"
-"{\n"
-"	float4 n = make_float4(a.x, a.y, a.z, 0.f);\n"
-"	return fastNormalize4( n );\n"
-"}\n"
-"__inline float4 lerp3(const float4 a,const float4 b, float  t)\n"
-"{\n"
-"	return make_float4(	a.x + (b.x - a.x) * t,\n"
-"						a.y + (b.y - a.y) * t,\n"
-"						a.z + (b.z - a.z) * t,\n"
-"						0.f);\n"
-"}\n"
-"float signedDistanceFromPointToPlane(float4 point, float4 planeEqn, float4* closestPointOnFace)\n"
-"{\n"
-"	float4 n = (float4)(planeEqn.x, planeEqn.y, planeEqn.z, 0);\n"
-"	float dist = dot3F4(n, point) + planeEqn.w;\n"
-"	*closestPointOnFace = point - dist * n;\n"
-"	return dist;\n"
-"}\n"
-"inline bool IsPointInPolygon(float4 p, \n"
-"							const btGpuFace* face,\n"
-"							__global const float4* baseVertex,\n"
-"							__global const  int* convexIndices,\n"
-"							float4* out)\n"
-"{\n"
-"    float4 a;\n"
-"    float4 b;\n"
-"    float4 ab;\n"
-"    float4 ap;\n"
-"    float4 v;\n"
-"	float4 plane = make_float4(face->m_plane.x,face->m_plane.y,face->m_plane.z,0.f);\n"
-"	\n"
-"	if (face->m_numIndices<2)\n"
-"		return false;\n"
-"	\n"
-"	float4 v0 = baseVertex[convexIndices[face->m_indexOffset + face->m_numIndices-1]];\n"
-"	\n"
-"	b = v0;\n"
-"    for(unsigned i=0; i != face->m_numIndices; ++i)\n"
-"    {\n"
-"		a = b;\n"
-"		float4 vi = baseVertex[convexIndices[face->m_indexOffset + i]];\n"
-"		b = vi;\n"
-"        ab = b-a;\n"
-"        ap = p-a;\n"
-"        v = cross3(ab,plane);\n"
-"        if (dot(ap, v) > 0.f)\n"
-"        {\n"
-"            float ab_m2 = dot(ab, ab);\n"
-"            float rt = ab_m2 != 0.f ? dot(ab, ap) / ab_m2 : 0.f;\n"
-"            if (rt <= 0.f)\n"
-"            {\n"
-"                *out = a;\n"
-"            }\n"
-"            else if (rt >= 1.f) \n"
-"            {\n"
-"                *out = b;\n"
-"            }\n"
-"            else\n"
-"            {\n"
-"            	float s = 1.f - rt;\n"
-"				out[0].x = s * a.x + rt * b.x;\n"
-"				out[0].y = s * a.y + rt * b.y;\n"
-"				out[0].z = s * a.z + rt * b.z;\n"
-"            }\n"
-"            return false;\n"
-"        }\n"
-"    }\n"
-"    return true;\n"
-"}\n"
-"void	computeContactSphereConvex(int pairIndex,\n"
-"																int bodyIndexA, int bodyIndexB, \n"
-"																int collidableIndexA, int collidableIndexB, \n"
-"																__global const BodyData* rigidBodies, \n"
-"																__global const btCollidableGpu* collidables,\n"
-"																__global const ConvexPolyhedronCL* convexShapes,\n"
-"																__global const float4* convexVertices,\n"
-"																__global const int* convexIndices,\n"
-"																__global const btGpuFace* faces,\n"
-"																__global struct b3Contact4Data* restrict globalContactsOut,\n"
-"																counter32_t nGlobalContactsOut,\n"
-"																int maxContactCapacity,\n"
-"																float4 spherePos2,\n"
-"																float radius,\n"
-"																float4 pos,\n"
-"																float4 quat\n"
-"																)\n"
-"{\n"
-"	float4 invPos;\n"
-"	float4 invOrn;\n"
-"	trInverse(pos,quat, &invPos,&invOrn);\n"
-"	float4 spherePos = transform(&spherePos2,&invPos,&invOrn);\n"
-"	int shapeIndex = collidables[collidableIndexB].m_shapeIndex;\n"
-"	int numFaces = convexShapes[shapeIndex].m_numFaces;\n"
-"	float4 closestPnt = (float4)(0, 0, 0, 0);\n"
-"	float4 hitNormalWorld = (float4)(0, 0, 0, 0);\n"
-"	float minDist = -1000000.f;\n"
-"	bool bCollide = true;\n"
-"	for ( int f = 0; f < numFaces; f++ )\n"
-"	{\n"
-"		btGpuFace face = faces[convexShapes[shapeIndex].m_faceOffset+f];\n"
-"		// set up a plane equation \n"
-"		float4 planeEqn;\n"
-"		float4 n1 = face.m_plane;\n"
-"		n1.w = 0.f;\n"
-"		planeEqn = n1;\n"
-"		planeEqn.w = face.m_plane.w;\n"
-"		\n"
-"	\n"
-"		// compute a signed distance from the vertex in cloth to the face of rigidbody.\n"
-"		float4 pntReturn;\n"
-"		float dist = signedDistanceFromPointToPlane(spherePos, planeEqn, &pntReturn);\n"
-"		// If the distance is positive, the plane is a separating plane. \n"
-"		if ( dist > radius )\n"
-"		{\n"
-"			bCollide = false;\n"
-"			break;\n"
-"		}\n"
-"		if (dist>0)\n"
-"		{\n"
-"			//might hit an edge or vertex\n"
-"			float4 out;\n"
-"			float4 zeroPos = make_float4(0,0,0,0);\n"
-"			bool isInPoly = IsPointInPolygon(spherePos,\n"
-"					&face,\n"
-"					&convexVertices[convexShapes[shapeIndex].m_vertexOffset],\n"
-"					convexIndices,\n"
-"           &out);\n"
-"			if (isInPoly)\n"
-"			{\n"
-"				if (dist>minDist)\n"
-"				{\n"
-"					minDist = dist;\n"
-"					closestPnt = pntReturn;\n"
-"					hitNormalWorld = planeEqn;\n"
-"					\n"
-"				}\n"
-"			} else\n"
-"			{\n"
-"				float4 tmp = spherePos-out;\n"
-"				float l2 = dot(tmp,tmp);\n"
-"				if (l2<radius*radius)\n"
-"				{\n"
-"					dist  = sqrt(l2);\n"
-"					if (dist>minDist)\n"
-"					{\n"
-"						minDist = dist;\n"
-"						closestPnt = out;\n"
-"						hitNormalWorld = tmp/dist;\n"
-"						\n"
-"					}\n"
-"					\n"
-"				} else\n"
-"				{\n"
-"					bCollide = false;\n"
-"					break;\n"
-"				}\n"
-"			}\n"
-"		} else\n"
-"		{\n"
-"			if ( dist > minDist )\n"
-"			{\n"
-"				minDist = dist;\n"
-"				closestPnt = pntReturn;\n"
-"				hitNormalWorld.xyz = planeEqn.xyz;\n"
-"			}\n"
-"		}\n"
-"		\n"
-"	}\n"
-"	\n"
-"	if (bCollide && minDist > -10000)\n"
-"	{\n"
-"		float4 normalOnSurfaceB1 = qtRotate(quat,-hitNormalWorld);\n"
-"		float4 pOnB1 = transform(&closestPnt,&pos,&quat);\n"
-"		\n"
-"		float actualDepth = minDist-radius;\n"
-"		if (actualDepth<=0.f)\n"
-"		{\n"
-"			\n"
-"			pOnB1.w = actualDepth;\n"
-"			int dstIdx;\n"
-"			AppendInc( nGlobalContactsOut, dstIdx );\n"
-"		\n"
-"			\n"
-"			if (1)//dstIdx < maxContactCapacity)\n"
-"			{\n"
-"				__global struct b3Contact4Data* c = &globalContactsOut[dstIdx];\n"
-"				c->m_worldNormalOnB = -normalOnSurfaceB1;\n"
-"				c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
-"				c->m_batchIdx = pairIndex;\n"
-"				c->m_bodyAPtrAndSignBit = rigidBodies[bodyIndexA].m_invMass==0?-bodyIndexA:bodyIndexA;\n"
-"				c->m_bodyBPtrAndSignBit = rigidBodies[bodyIndexB].m_invMass==0?-bodyIndexB:bodyIndexB;\n"
-"				c->m_worldPosB[0] = pOnB1;\n"
-"				c->m_childIndexA = -1;\n"
-"				c->m_childIndexB = -1;\n"
-"				GET_NPOINTS(*c) = 1;\n"
-"			} \n"
-"		}\n"
-"	}//if (hasCollision)\n"
-"}\n"
-"							\n"
-"int extractManifoldSequential(const float4* p, int nPoints, float4 nearNormal, int4* contactIdx)\n"
-"{\n"
-"	if( nPoints == 0 )\n"
-"        return 0;\n"
-"    \n"
-"    if (nPoints <=4)\n"
-"        return nPoints;\n"
-"    \n"
-"    \n"
-"    if (nPoints >64)\n"
-"        nPoints = 64;\n"
-"    \n"
-"	float4 center = make_float4(0.f);\n"
-"	{\n"
-"		\n"
-"		for (int i=0;i<nPoints;i++)\n"
-"			center += p[i];\n"
-"		center /= (float)nPoints;\n"
-"	}\n"
-"    \n"
-"	\n"
-"    \n"
-"	//	sample 4 directions\n"
-"    \n"
-"    float4 aVector = p[0] - center;\n"
-"    float4 u = cross3( nearNormal, aVector );\n"
-"    float4 v = cross3( nearNormal, u );\n"
-"    u = normalize3( u );\n"
-"    v = normalize3( v );\n"
-"    \n"
-"    \n"
-"    //keep point with deepest penetration\n"
-"    float minW= FLT_MAX;\n"
-"    \n"
-"    int minIndex=-1;\n"
-"    \n"
-"    float4 maxDots;\n"
-"    maxDots.x = FLT_MIN;\n"
-"    maxDots.y = FLT_MIN;\n"
-"    maxDots.z = FLT_MIN;\n"
-"    maxDots.w = FLT_MIN;\n"
-"    \n"
-"    //	idx, distance\n"
-"    for(int ie = 0; ie<nPoints; ie++ )\n"
-"    {\n"
-"        if (p[ie].w<minW)\n"
-"        {\n"
-"            minW = p[ie].w;\n"
-"            minIndex=ie;\n"
-"        }\n"
-"        float f;\n"
-"        float4 r = p[ie]-center;\n"
-"        f = dot3F4( u, r );\n"
-"        if (f<maxDots.x)\n"
-"        {\n"
-"            maxDots.x = f;\n"
-"            contactIdx[0].x = ie;\n"
-"        }\n"
-"        \n"
-"        f = dot3F4( -u, r );\n"
-"        if (f<maxDots.y)\n"
-"        {\n"
-"            maxDots.y = f;\n"
-"            contactIdx[0].y = ie;\n"
-"        }\n"
-"        \n"
-"        \n"
-"        f = dot3F4( v, r );\n"
-"        if (f<maxDots.z)\n"
-"        {\n"
-"            maxDots.z = f;\n"
-"            contactIdx[0].z = ie;\n"
-"        }\n"
-"        \n"
-"        f = dot3F4( -v, r );\n"
-"        if (f<maxDots.w)\n"
-"        {\n"
-"            maxDots.w = f;\n"
-"            contactIdx[0].w = ie;\n"
-"        }\n"
-"        \n"
-"    }\n"
-"    \n"
-"    if (contactIdx[0].x != minIndex && contactIdx[0].y != minIndex && contactIdx[0].z != minIndex && contactIdx[0].w != minIndex)\n"
-"    {\n"
-"        //replace the first contact with minimum (todo: replace contact with least penetration)\n"
-"        contactIdx[0].x = minIndex;\n"
-"    }\n"
-"    \n"
-"    return 4;\n"
-"    \n"
-"}\n"
-"#define MAX_PLANE_CONVEX_POINTS 64\n"
-"int computeContactPlaneConvex(int pairIndex,\n"
-"								int bodyIndexA, int bodyIndexB, \n"
-"								int collidableIndexA, int collidableIndexB, \n"
-"								__global const BodyData* rigidBodies, \n"
-"								__global const btCollidableGpu*collidables,\n"
-"								__global const ConvexPolyhedronCL* convexShapes,\n"
-"								__global const float4* convexVertices,\n"
-"								__global const int* convexIndices,\n"
-"								__global const btGpuFace* faces,\n"
-"								__global struct b3Contact4Data* restrict globalContactsOut,\n"
-"								counter32_t nGlobalContactsOut,\n"
-"								int maxContactCapacity,\n"
-"								float4 posB,\n"
-"								Quaternion ornB\n"
-"								)\n"
-"{\n"
-"	int resultIndex=-1;\n"
-"		int shapeIndex = collidables[collidableIndexB].m_shapeIndex;\n"
-"	__global const ConvexPolyhedronCL* hullB = &convexShapes[shapeIndex];\n"
-"	\n"
-"	float4 posA;\n"
-"	posA = rigidBodies[bodyIndexA].m_pos;\n"
-"	Quaternion ornA;\n"
-"	ornA = rigidBodies[bodyIndexA].m_quat;\n"
-"	int numContactsOut = 0;\n"
-"	int numWorldVertsB1= 0;\n"
-"	float4 planeEq;\n"
-"	 planeEq = faces[collidables[collidableIndexA].m_shapeIndex].m_plane;\n"
-"	float4 planeNormal = make_float4(planeEq.x,planeEq.y,planeEq.z,0.f);\n"
-"	float4 planeNormalWorld;\n"
-"	planeNormalWorld = qtRotate(ornA,planeNormal);\n"
-"	float planeConstant = planeEq.w;\n"
-"	\n"
-"	float4 invPosA;Quaternion invOrnA;\n"
-"	float4 convexInPlaneTransPos1; Quaternion convexInPlaneTransOrn1;\n"
-"	{\n"
-"		\n"
-"		trInverse(posA,ornA,&invPosA,&invOrnA);\n"
-"		trMul(invPosA,invOrnA,posB,ornB,&convexInPlaneTransPos1,&convexInPlaneTransOrn1);\n"
-"	}\n"
-"	float4 invPosB;Quaternion invOrnB;\n"
-"	float4 planeInConvexPos1;	Quaternion planeInConvexOrn1;\n"
-"	{\n"
-"		\n"
-"		trInverse(posB,ornB,&invPosB,&invOrnB);\n"
-"		trMul(invPosB,invOrnB,posA,ornA,&planeInConvexPos1,&planeInConvexOrn1);	\n"
-"	}\n"
-"	\n"
-"	float4 planeNormalInConvex = qtRotate(planeInConvexOrn1,-planeNormal);\n"
-"	float maxDot = -1e30;\n"
-"	int hitVertex=-1;\n"
-"	float4 hitVtx;\n"
-"	float4 contactPoints[MAX_PLANE_CONVEX_POINTS];\n"
-"	int numPoints = 0;\n"
-"	int4 contactIdx;\n"
-"	contactIdx=make_int4(0,1,2,3);\n"
-"    \n"
-"	\n"
-"	for (int i=0;i<hullB->m_numVertices;i++)\n"
-"	{\n"
-"		float4 vtx = convexVertices[hullB->m_vertexOffset+i];\n"
-"		float curDot = dot(vtx,planeNormalInConvex);\n"
-"		if (curDot>maxDot)\n"
-"		{\n"
-"			hitVertex=i;\n"
-"			maxDot=curDot;\n"
-"			hitVtx = vtx;\n"
-"			//make sure the deepest points is always included\n"
-"			if (numPoints==MAX_PLANE_CONVEX_POINTS)\n"
-"				numPoints--;\n"
-"		}\n"
-"		if (numPoints<MAX_PLANE_CONVEX_POINTS)\n"
-"		{\n"
-"			float4 vtxWorld = transform(&vtx, &posB, &ornB);\n"
-"			float4 vtxInPlane = transform(&vtxWorld, &invPosA, &invOrnA);//oplaneTransform.inverse()*vtxWorld;\n"
-"			float dist = dot(planeNormal,vtxInPlane)-planeConstant;\n"
-"			if (dist<0.f)\n"
-"			{\n"
-"				vtxWorld.w = dist;\n"
-"				contactPoints[numPoints] = vtxWorld;\n"
-"				numPoints++;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	int numReducedPoints  = numPoints;\n"
-"	if (numPoints>4)\n"
-"	{\n"
-"		numReducedPoints = extractManifoldSequential( contactPoints, numPoints, planeNormalInConvex, &contactIdx);\n"
-"	}\n"
-"	if (numReducedPoints>0)\n"
-"	{\n"
-"		int dstIdx;\n"
-"	    AppendInc( nGlobalContactsOut, dstIdx );\n"
-"		if (dstIdx < maxContactCapacity)\n"
-"		{\n"
-"			resultIndex = dstIdx;\n"
-"			__global struct b3Contact4Data* c = &globalContactsOut[dstIdx];\n"
-"			c->m_worldNormalOnB = -planeNormalWorld;\n"
-"			//c->setFrictionCoeff(0.7);\n"
-"			//c->setRestituitionCoeff(0.f);\n"
-"			c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
-"			c->m_batchIdx = pairIndex;\n"
-"			c->m_bodyAPtrAndSignBit = rigidBodies[bodyIndexA].m_invMass==0?-bodyIndexA:bodyIndexA;\n"
-"			c->m_bodyBPtrAndSignBit = rigidBodies[bodyIndexB].m_invMass==0?-bodyIndexB:bodyIndexB;\n"
-"			c->m_childIndexA = -1;\n"
-"			c->m_childIndexB = -1;\n"
-"			switch (numReducedPoints)\n"
-"            {\n"
-"                case 4:\n"
-"                    c->m_worldPosB[3] = contactPoints[contactIdx.w];\n"
-"                case 3:\n"
-"                    c->m_worldPosB[2] = contactPoints[contactIdx.z];\n"
-"                case 2:\n"
-"                    c->m_worldPosB[1] = contactPoints[contactIdx.y];\n"
-"                case 1:\n"
-"                    c->m_worldPosB[0] = contactPoints[contactIdx.x];\n"
-"                default:\n"
-"                {\n"
-"                }\n"
-"            };\n"
-"			\n"
-"			GET_NPOINTS(*c) = numReducedPoints;\n"
-"		}//if (dstIdx < numPairs)\n"
-"	}	\n"
-"	return resultIndex;\n"
-"}\n"
-"void	computeContactPlaneSphere(int pairIndex,\n"
-"																int bodyIndexA, int bodyIndexB, \n"
-"																int collidableIndexA, int collidableIndexB, \n"
-"																__global const BodyData* rigidBodies, \n"
-"																__global const btCollidableGpu* collidables,\n"
-"																__global const btGpuFace* faces,\n"
-"																__global struct b3Contact4Data* restrict globalContactsOut,\n"
-"																counter32_t nGlobalContactsOut,\n"
-"																int maxContactCapacity)\n"
-"{\n"
-"	float4 planeEq = faces[collidables[collidableIndexA].m_shapeIndex].m_plane;\n"
-"	float radius = collidables[collidableIndexB].m_radius;\n"
-"	float4 posA1 = rigidBodies[bodyIndexA].m_pos;\n"
-"	float4 ornA1 = rigidBodies[bodyIndexA].m_quat;\n"
-"	float4 posB1 = rigidBodies[bodyIndexB].m_pos;\n"
-"	float4 ornB1 = rigidBodies[bodyIndexB].m_quat;\n"
-"	\n"
-"	bool hasCollision = false;\n"
-"	float4 planeNormal1 = make_float4(planeEq.x,planeEq.y,planeEq.z,0.f);\n"
-"	float planeConstant = planeEq.w;\n"
-"	float4 convexInPlaneTransPos1; Quaternion convexInPlaneTransOrn1;\n"
-"	{\n"
-"		float4 invPosA;Quaternion invOrnA;\n"
-"		trInverse(posA1,ornA1,&invPosA,&invOrnA);\n"
-"		trMul(invPosA,invOrnA,posB1,ornB1,&convexInPlaneTransPos1,&convexInPlaneTransOrn1);\n"
-"	}\n"
-"	float4 planeInConvexPos1;	Quaternion planeInConvexOrn1;\n"
-"	{\n"
-"		float4 invPosB;Quaternion invOrnB;\n"
-"		trInverse(posB1,ornB1,&invPosB,&invOrnB);\n"
-"		trMul(invPosB,invOrnB,posA1,ornA1,&planeInConvexPos1,&planeInConvexOrn1);	\n"
-"	}\n"
-"	float4 vtx1 = qtRotate(planeInConvexOrn1,-planeNormal1)*radius;\n"
-"	float4 vtxInPlane1 = transform(&vtx1,&convexInPlaneTransPos1,&convexInPlaneTransOrn1);\n"
-"	float distance = dot3F4(planeNormal1,vtxInPlane1) - planeConstant;\n"
-"	hasCollision = distance < 0.f;//m_manifoldPtr->getContactBreakingThreshold();\n"
-"	if (hasCollision)\n"
-"	{\n"
-"		float4 vtxInPlaneProjected1 = vtxInPlane1 -   distance*planeNormal1;\n"
-"		float4 vtxInPlaneWorld1 = transform(&vtxInPlaneProjected1,&posA1,&ornA1);\n"
-"		float4 normalOnSurfaceB1 = qtRotate(ornA1,planeNormal1);\n"
-"		float4 pOnB1 = vtxInPlaneWorld1+normalOnSurfaceB1*distance;\n"
-"		pOnB1.w = distance;\n"
-"		int dstIdx;\n"
-"    AppendInc( nGlobalContactsOut, dstIdx );\n"
-"		\n"
-"		if (dstIdx < maxContactCapacity)\n"
-"		{\n"
-"			__global struct b3Contact4Data* c = &globalContactsOut[dstIdx];\n"
-"			c->m_worldNormalOnB = -normalOnSurfaceB1;\n"
-"			c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
-"			c->m_batchIdx = pairIndex;\n"
-"			c->m_bodyAPtrAndSignBit = rigidBodies[bodyIndexA].m_invMass==0?-bodyIndexA:bodyIndexA;\n"
-"			c->m_bodyBPtrAndSignBit = rigidBodies[bodyIndexB].m_invMass==0?-bodyIndexB:bodyIndexB;\n"
-"			c->m_worldPosB[0] = pOnB1;\n"
-"			c->m_childIndexA = -1;\n"
-"			c->m_childIndexB = -1;\n"
-"			GET_NPOINTS(*c) = 1;\n"
-"		}//if (dstIdx < numPairs)\n"
-"	}//if (hasCollision)\n"
-"}\n"
-"__kernel void   primitiveContactsKernel( __global int4* pairs, \n"
-"																					__global const BodyData* rigidBodies, \n"
-"																					__global const btCollidableGpu* collidables,\n"
-"																					__global const ConvexPolyhedronCL* convexShapes, \n"
-"																					__global const float4* vertices,\n"
-"																					__global const float4* uniqueEdges,\n"
-"																					__global const btGpuFace* faces,\n"
-"																					__global const int* indices,\n"
-"																					__global struct b3Contact4Data* restrict globalContactsOut,\n"
-"																					counter32_t nGlobalContactsOut,\n"
-"																					int numPairs, int maxContactCapacity)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	int pairIndex = i;\n"
-"	\n"
-"	float4 worldVertsB1[64];\n"
-"	float4 worldVertsB2[64];\n"
-"	int capacityWorldVerts = 64;	\n"
-"	float4 localContactsOut[64];\n"
-"	int localContactCapacity=64;\n"
-"	\n"
-"	float minDist = -1e30f;\n"
-"	float maxDist = 0.02f;\n"
-"	if (i<numPairs)\n"
-"	{\n"
-"		int bodyIndexA = pairs[i].x;\n"
-"		int bodyIndexB = pairs[i].y;\n"
-"			\n"
-"		int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
-"		int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
-"	\n"
-"		if (collidables[collidableIndexA].m_shapeType == SHAPE_PLANE &&\n"
-"			collidables[collidableIndexB].m_shapeType == SHAPE_CONVEX_HULL)\n"
-"		{\n"
-"			float4 posB;\n"
-"			posB = rigidBodies[bodyIndexB].m_pos;\n"
-"			Quaternion ornB;\n"
-"			ornB = rigidBodies[bodyIndexB].m_quat;\n"
-"			int contactIndex = computeContactPlaneConvex(pairIndex, bodyIndexA, bodyIndexB, collidableIndexA, collidableIndexB, \n"
-"																rigidBodies,collidables,convexShapes,vertices,indices,\n"
-"																faces,	globalContactsOut, nGlobalContactsOut,maxContactCapacity, posB,ornB);\n"
-"			if (contactIndex>=0)\n"
-"				pairs[pairIndex].z = contactIndex;\n"
-"			return;\n"
-"		}\n"
-"		if (collidables[collidableIndexA].m_shapeType == SHAPE_CONVEX_HULL &&\n"
-"			collidables[collidableIndexB].m_shapeType == SHAPE_PLANE)\n"
-"		{\n"
-"			float4 posA;\n"
-"			posA = rigidBodies[bodyIndexA].m_pos;\n"
-"			Quaternion ornA;\n"
-"			ornA = rigidBodies[bodyIndexA].m_quat;\n"
-"			int contactIndex = computeContactPlaneConvex( pairIndex, bodyIndexB,bodyIndexA,  collidableIndexB,collidableIndexA, \n"
-"																rigidBodies,collidables,convexShapes,vertices,indices,\n"
-"																faces,	globalContactsOut, nGlobalContactsOut,maxContactCapacity,posA,ornA);\n"
-"			if (contactIndex>=0)\n"
-"				pairs[pairIndex].z = contactIndex;\n"
-"			return;\n"
-"		}\n"
-"		if (collidables[collidableIndexA].m_shapeType == SHAPE_PLANE &&\n"
-"			collidables[collidableIndexB].m_shapeType == SHAPE_SPHERE)\n"
-"		{\n"
-"			computeContactPlaneSphere(pairIndex, bodyIndexA, bodyIndexB, collidableIndexA, collidableIndexB, \n"
-"																rigidBodies,collidables,faces,	globalContactsOut, nGlobalContactsOut,maxContactCapacity);\n"
-"			return;\n"
-"		}\n"
-"		if (collidables[collidableIndexA].m_shapeType == SHAPE_SPHERE &&\n"
-"			collidables[collidableIndexB].m_shapeType == SHAPE_PLANE)\n"
-"		{\n"
-"			computeContactPlaneSphere( pairIndex, bodyIndexB,bodyIndexA,  collidableIndexB,collidableIndexA, \n"
-"																rigidBodies,collidables,\n"
-"																faces,	globalContactsOut, nGlobalContactsOut,maxContactCapacity);\n"
-"			return;\n"
-"		}\n"
-"		\n"
-"	\n"
-"		if (collidables[collidableIndexA].m_shapeType == SHAPE_SPHERE &&\n"
-"			collidables[collidableIndexB].m_shapeType == SHAPE_CONVEX_HULL)\n"
-"		{\n"
-"		\n"
-"			float4 spherePos = rigidBodies[bodyIndexA].m_pos;\n"
-"			float sphereRadius = collidables[collidableIndexA].m_radius;\n"
-"			float4 convexPos = rigidBodies[bodyIndexB].m_pos;\n"
-"			float4 convexOrn = rigidBodies[bodyIndexB].m_quat;\n"
-"			computeContactSphereConvex(pairIndex, bodyIndexA, bodyIndexB, collidableIndexA, collidableIndexB, \n"
-"																rigidBodies,collidables,convexShapes,vertices,indices,faces, globalContactsOut, nGlobalContactsOut,maxContactCapacity,\n"
-"																spherePos,sphereRadius,convexPos,convexOrn);\n"
-"			return;\n"
-"		}\n"
-"		if (collidables[collidableIndexA].m_shapeType == SHAPE_CONVEX_HULL &&\n"
-"			collidables[collidableIndexB].m_shapeType == SHAPE_SPHERE)\n"
-"		{\n"
-"		\n"
-"			float4 spherePos = rigidBodies[bodyIndexB].m_pos;\n"
-"			float sphereRadius = collidables[collidableIndexB].m_radius;\n"
-"			float4 convexPos = rigidBodies[bodyIndexA].m_pos;\n"
-"			float4 convexOrn = rigidBodies[bodyIndexA].m_quat;\n"
-"			computeContactSphereConvex(pairIndex, bodyIndexB, bodyIndexA, collidableIndexB, collidableIndexA, \n"
-"																rigidBodies,collidables,convexShapes,vertices,indices,faces, globalContactsOut, nGlobalContactsOut,maxContactCapacity,\n"
-"																spherePos,sphereRadius,convexPos,convexOrn);\n"
-"			return;\n"
-"		}\n"
-"	\n"
-"	\n"
-"	\n"
-"		\n"
-"	\n"
-"	\n"
-"		if (collidables[collidableIndexA].m_shapeType == SHAPE_SPHERE &&\n"
-"			collidables[collidableIndexB].m_shapeType == SHAPE_SPHERE)\n"
-"		{\n"
-"			//sphere-sphere\n"
-"			float radiusA = collidables[collidableIndexA].m_radius;\n"
-"			float radiusB = collidables[collidableIndexB].m_radius;\n"
-"			float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
-"			float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
-"			float4 diff = posA-posB;\n"
-"			float len = length(diff);\n"
-"			\n"
-"			///iff distance positive, don't generate a new contact\n"
-"			if ( len <= (radiusA+radiusB))\n"
-"			{\n"
-"				///distance (negative means penetration)\n"
-"				float dist = len - (radiusA+radiusB);\n"
-"				float4 normalOnSurfaceB = make_float4(1.f,0.f,0.f,0.f);\n"
-"				if (len > 0.00001)\n"
-"				{\n"
-"					normalOnSurfaceB = diff / len;\n"
-"				}\n"
-"				float4 contactPosB = posB + normalOnSurfaceB*radiusB;\n"
-"				contactPosB.w = dist;\n"
-"								\n"
-"				int dstIdx;\n"
-"				 AppendInc( nGlobalContactsOut, dstIdx );\n"
-"				\n"
-"				if (dstIdx < maxContactCapacity)\n"
-"				{\n"
-"					__global struct b3Contact4Data* c = &globalContactsOut[dstIdx];\n"
-"					c->m_worldNormalOnB = normalOnSurfaceB;\n"
-"					c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
-"					c->m_batchIdx = pairIndex;\n"
-"					int bodyA = pairs[pairIndex].x;\n"
-"					int bodyB = pairs[pairIndex].y;\n"
-"					c->m_bodyAPtrAndSignBit = rigidBodies[bodyA].m_invMass==0?-bodyA:bodyA;\n"
-"					c->m_bodyBPtrAndSignBit = rigidBodies[bodyB].m_invMass==0?-bodyB:bodyB;\n"
-"					c->m_worldPosB[0] = contactPosB;\n"
-"					c->m_childIndexA = -1;\n"
-"					c->m_childIndexB = -1;\n"
-"					GET_NPOINTS(*c) = 1;\n"
-"				}//if (dstIdx < numPairs)\n"
-"			}//if ( len <= (radiusA+radiusB))\n"
-"			return;\n"
-"		}//SHAPE_SPHERE SHAPE_SPHERE\n"
-"	}//	if (i<numPairs)\n"
-"}\n"
-"// work-in-progress\n"
-"__kernel void   processCompoundPairsPrimitivesKernel( __global const int4* gpuCompoundPairs,\n"
-"													__global const BodyData* rigidBodies, \n"
-"													__global const btCollidableGpu* collidables,\n"
-"													__global const ConvexPolyhedronCL* convexShapes, \n"
-"													__global const float4* vertices,\n"
-"													__global const float4* uniqueEdges,\n"
-"													__global const btGpuFace* faces,\n"
-"													__global const int* indices,\n"
-"													__global btAabbCL* aabbs,\n"
-"													__global const btGpuChildShape* gpuChildShapes,\n"
-"													__global struct b3Contact4Data* restrict globalContactsOut,\n"
-"													counter32_t nGlobalContactsOut,\n"
-"													int numCompoundPairs, int maxContactCapacity\n"
-"													)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	if (i<numCompoundPairs)\n"
-"	{\n"
-"		int bodyIndexA = gpuCompoundPairs[i].x;\n"
-"		int bodyIndexB = gpuCompoundPairs[i].y;\n"
-"		int childShapeIndexA = gpuCompoundPairs[i].z;\n"
-"		int childShapeIndexB = gpuCompoundPairs[i].w;\n"
-"		\n"
-"		int collidableIndexA = -1;\n"
-"		int collidableIndexB = -1;\n"
-"		\n"
-"		float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
-"		float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
-"		\n"
-"		float4 ornB = rigidBodies[bodyIndexB].m_quat;\n"
-"		float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
-"							\n"
-"		if (childShapeIndexA >= 0)\n"
-"		{\n"
-"			collidableIndexA = gpuChildShapes[childShapeIndexA].m_shapeIndex;\n"
-"			float4 childPosA = gpuChildShapes[childShapeIndexA].m_childPosition;\n"
-"			float4 childOrnA = gpuChildShapes[childShapeIndexA].m_childOrientation;\n"
-"			float4 newPosA = qtRotate(ornA,childPosA)+posA;\n"
-"			float4 newOrnA = qtMul(ornA,childOrnA);\n"
-"			posA = newPosA;\n"
-"			ornA = newOrnA;\n"
-"		} else\n"
-"		{\n"
-"			collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
-"		}\n"
-"		\n"
-"		if (childShapeIndexB>=0)\n"
-"		{\n"
-"			collidableIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;\n"
-"			float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;\n"
-"			float4 childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;\n"
-"			float4 newPosB = transform(&childPosB,&posB,&ornB);\n"
-"			float4 newOrnB = qtMul(ornB,childOrnB);\n"
-"			posB = newPosB;\n"
-"			ornB = newOrnB;\n"
-"		} else\n"
-"		{\n"
-"			collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;	\n"
-"		}\n"
-"	\n"
-"		int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
-"		int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
-"	\n"
-"		int shapeTypeA = collidables[collidableIndexA].m_shapeType;\n"
-"		int shapeTypeB = collidables[collidableIndexB].m_shapeType;\n"
-"		int pairIndex = i;\n"
-"		if ((shapeTypeA == SHAPE_PLANE) && (shapeTypeB==SHAPE_CONVEX_HULL))\n"
-"		{\n"
-"			computeContactPlaneConvex( pairIndex, bodyIndexA,bodyIndexB,  collidableIndexA,collidableIndexB, \n"
-"																rigidBodies,collidables,convexShapes,vertices,indices,\n"
-"																faces,	globalContactsOut, nGlobalContactsOut,maxContactCapacity,posB,ornB);\n"
-"			return;\n"
-"		}\n"
-"		if ((shapeTypeA == SHAPE_CONVEX_HULL) && (shapeTypeB==SHAPE_PLANE))\n"
-"		{\n"
-"			computeContactPlaneConvex( pairIndex, bodyIndexB,bodyIndexA,  collidableIndexB,collidableIndexA, \n"
-"																rigidBodies,collidables,convexShapes,vertices,indices,\n"
-"																faces,	globalContactsOut, nGlobalContactsOut,maxContactCapacity,posA,ornA);\n"
-"			return;\n"
-"		}\n"
-"		if ((shapeTypeA == SHAPE_CONVEX_HULL) && (shapeTypeB == SHAPE_SPHERE))\n"
-"		{\n"
-"			float4 spherePos = rigidBodies[bodyIndexB].m_pos;\n"
-"			float sphereRadius = collidables[collidableIndexB].m_radius;\n"
-"			float4 convexPos = posA;\n"
-"			float4 convexOrn = ornA;\n"
-"			\n"
-"			computeContactSphereConvex(pairIndex, bodyIndexB, bodyIndexA , collidableIndexB,collidableIndexA, \n"
-"										rigidBodies,collidables,convexShapes,vertices,indices,faces, globalContactsOut, nGlobalContactsOut,maxContactCapacity,\n"
-"										spherePos,sphereRadius,convexPos,convexOrn);\n"
-"	\n"
-"			return;\n"
-"		}\n"
-"		if ((shapeTypeA == SHAPE_SPHERE) && (shapeTypeB == SHAPE_CONVEX_HULL))\n"
-"		{\n"
-"			float4 spherePos = rigidBodies[bodyIndexA].m_pos;\n"
-"			float sphereRadius = collidables[collidableIndexA].m_radius;\n"
-"			float4 convexPos = posB;\n"
-"			float4 convexOrn = ornB;\n"
-"			\n"
-"			computeContactSphereConvex(pairIndex, bodyIndexA, bodyIndexB, collidableIndexA, collidableIndexB, \n"
-"										rigidBodies,collidables,convexShapes,vertices,indices,faces, globalContactsOut, nGlobalContactsOut,maxContactCapacity,\n"
-"										spherePos,sphereRadius,convexPos,convexOrn);\n"
-"	\n"
-"			return;\n"
-"		}\n"
-"	}//	if (i<numCompoundPairs)\n"
-"}\n"
-"bool pointInTriangle(const float4* vertices, const float4* normal, float4 *p )\n"
-"{\n"
-"	const float4* p1 = &vertices[0];\n"
-"	const float4* p2 = &vertices[1];\n"
-"	const float4* p3 = &vertices[2];\n"
-"	float4 edge1;	edge1 = (*p2 - *p1);\n"
-"	float4 edge2;	edge2 = ( *p3 - *p2 );\n"
-"	float4 edge3;	edge3 = ( *p1 - *p3 );\n"
-"	\n"
-"	float4 p1_to_p; p1_to_p = ( *p - *p1 );\n"
-"	float4 p2_to_p; p2_to_p = ( *p - *p2 );\n"
-"	float4 p3_to_p; p3_to_p = ( *p - *p3 );\n"
-"	float4 edge1_normal; edge1_normal = ( cross(edge1,*normal));\n"
-"	float4 edge2_normal; edge2_normal = ( cross(edge2,*normal));\n"
-"	float4 edge3_normal; edge3_normal = ( cross(edge3,*normal));\n"
-"	\n"
-"	\n"
-"	float r1, r2, r3;\n"
-"	r1 = dot(edge1_normal,p1_to_p );\n"
-"	r2 = dot(edge2_normal,p2_to_p );\n"
-"	r3 = dot(edge3_normal,p3_to_p );\n"
-"	\n"
-"	if ( r1 > 0 && r2 > 0 && r3 > 0 )\n"
-"		return true;\n"
-"    if ( r1 <= 0 && r2 <= 0 && r3 <= 0 ) \n"
-"		return true;\n"
-"	return false;\n"
-"}\n"
-"float segmentSqrDistance(float4 from, float4 to,float4 p, float4* nearest) \n"
-"{\n"
-"	float4 diff = p - from;\n"
-"	float4 v = to - from;\n"
-"	float t = dot(v,diff);\n"
-"	\n"
-"	if (t > 0) \n"
-"	{\n"
-"		float dotVV = dot(v,v);\n"
-"		if (t < dotVV) \n"
-"		{\n"
-"			t /= dotVV;\n"
-"			diff -= t*v;\n"
-"		} else \n"
-"		{\n"
-"			t = 1;\n"
-"			diff -= v;\n"
-"		}\n"
-"	} else\n"
-"	{\n"
-"		t = 0;\n"
-"	}\n"
-"	*nearest = from + t*v;\n"
-"	return dot(diff,diff);	\n"
-"}\n"
-"void	computeContactSphereTriangle(int pairIndex,\n"
-"									int bodyIndexA, int bodyIndexB,\n"
-"									int collidableIndexA, int collidableIndexB, \n"
-"									__global const BodyData* rigidBodies, \n"
-"									__global const btCollidableGpu* collidables,\n"
-"									const float4* triangleVertices,\n"
-"									__global struct b3Contact4Data* restrict globalContactsOut,\n"
-"									counter32_t nGlobalContactsOut,\n"
-"									int maxContactCapacity,\n"
-"									float4 spherePos2,\n"
-"									float radius,\n"
-"									float4 pos,\n"
-"									float4 quat,\n"
-"									int faceIndex\n"
-"									)\n"
-"{\n"
-"	float4 invPos;\n"
-"	float4 invOrn;\n"
-"	trInverse(pos,quat, &invPos,&invOrn);\n"
-"	float4 spherePos = transform(&spherePos2,&invPos,&invOrn);\n"
-"	int numFaces = 3;\n"
-"	float4 closestPnt = (float4)(0, 0, 0, 0);\n"
-"	float4 hitNormalWorld = (float4)(0, 0, 0, 0);\n"
-"	float minDist = -1000000.f;\n"
-"	bool bCollide = false;\n"
-"	\n"
-"	//////////////////////////////////////\n"
-"	float4 sphereCenter;\n"
-"	sphereCenter = spherePos;\n"
-"	const float4* vertices = triangleVertices;\n"
-"	float contactBreakingThreshold = 0.f;//todo?\n"
-"	float radiusWithThreshold = radius + contactBreakingThreshold;\n"
-"	float4 edge10;\n"
-"	edge10 = vertices[1]-vertices[0];\n"
-"	edge10.w = 0.f;//is this needed?\n"
-"	float4 edge20;\n"
-"	edge20 = vertices[2]-vertices[0];\n"
-"	edge20.w = 0.f;//is this needed?\n"
-"	float4 normal = cross3(edge10,edge20);\n"
-"	normal = normalize(normal);\n"
-"	float4 p1ToCenter;\n"
-"	p1ToCenter = sphereCenter - vertices[0];\n"
-"	\n"
-"	float distanceFromPlane = dot(p1ToCenter,normal);\n"
-"	if (distanceFromPlane < 0.f)\n"
-"	{\n"
-"		//triangle facing the other way\n"
-"		distanceFromPlane *= -1.f;\n"
-"		normal *= -1.f;\n"
-"	}\n"
-"	hitNormalWorld = normal;\n"
-"	bool isInsideContactPlane = distanceFromPlane < radiusWithThreshold;\n"
-"	\n"
-"	// Check for contact / intersection\n"
-"	bool hasContact = false;\n"
-"	float4 contactPoint;\n"
-"	if (isInsideContactPlane) \n"
-"	{\n"
-"	\n"
-"		if (pointInTriangle(vertices,&normal, &sphereCenter)) \n"
-"		{\n"
-"			// Inside the contact wedge - touches a point on the shell plane\n"
-"			hasContact = true;\n"
-"			contactPoint = sphereCenter - normal*distanceFromPlane;\n"
-"			\n"
-"		} else {\n"
-"			// Could be inside one of the contact capsules\n"
-"			float contactCapsuleRadiusSqr = radiusWithThreshold*radiusWithThreshold;\n"
-"			float4 nearestOnEdge;\n"
-"			int numEdges = 3;\n"
-"			for (int i = 0; i < numEdges; i++) \n"
-"			{\n"
-"				float4 pa =vertices[i];\n"
-"				float4 pb = vertices[(i+1)%3];\n"
-"				float distanceSqr = segmentSqrDistance(pa,pb,sphereCenter, &nearestOnEdge);\n"
-"				if (distanceSqr < contactCapsuleRadiusSqr) \n"
-"				{\n"
-"					// Yep, we're inside a capsule\n"
-"					hasContact = true;\n"
-"					contactPoint = nearestOnEdge;\n"
-"					\n"
-"				}\n"
-"				\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	if (hasContact) \n"
-"	{\n"
-"		closestPnt = contactPoint;\n"
-"		float4 contactToCenter = sphereCenter - contactPoint;\n"
-"		minDist = length(contactToCenter);\n"
-"		if (minDist>FLT_EPSILON)\n"
-"		{\n"
-"			hitNormalWorld = normalize(contactToCenter);//*(1./minDist);\n"
-"			bCollide  = true;\n"
-"		}\n"
-"		\n"
-"	}\n"
-"	/////////////////////////////////////\n"
-"	if (bCollide && minDist > -10000)\n"
-"	{\n"
-"		\n"
-"		float4 normalOnSurfaceB1 = qtRotate(quat,-hitNormalWorld);\n"
-"		float4 pOnB1 = transform(&closestPnt,&pos,&quat);\n"
-"		float actualDepth = minDist-radius;\n"
-"		\n"
-"		if (actualDepth<=0.f)\n"
-"		{\n"
-"			pOnB1.w = actualDepth;\n"
-"			int dstIdx;\n"
-"			\n"
-"			float lenSqr = dot3F4(normalOnSurfaceB1,normalOnSurfaceB1);\n"
-"			if (lenSqr>FLT_EPSILON)\n"
-"			{\n"
-"				AppendInc( nGlobalContactsOut, dstIdx );\n"
-"			\n"
-"				if (dstIdx < maxContactCapacity)\n"
-"				{\n"
-"					__global struct b3Contact4Data* c = &globalContactsOut[dstIdx];\n"
-"					c->m_worldNormalOnB = -normalOnSurfaceB1;\n"
-"					c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
-"					c->m_batchIdx = pairIndex;\n"
-"					c->m_bodyAPtrAndSignBit = rigidBodies[bodyIndexA].m_invMass==0?-bodyIndexA:bodyIndexA;\n"
-"					c->m_bodyBPtrAndSignBit = rigidBodies[bodyIndexB].m_invMass==0?-bodyIndexB:bodyIndexB;\n"
-"					c->m_worldPosB[0] = pOnB1;\n"
-"					c->m_childIndexA = -1;\n"
-"					c->m_childIndexB = faceIndex;\n"
-"					GET_NPOINTS(*c) = 1;\n"
-"				} \n"
-"			}\n"
-"		}\n"
-"	}//if (hasCollision)\n"
-"}\n"
-"// work-in-progress\n"
-"__kernel void   findConcaveSphereContactsKernel( __global int4* concavePairs,\n"
-"												__global const BodyData* rigidBodies,\n"
-"												__global const btCollidableGpu* collidables,\n"
-"												__global const ConvexPolyhedronCL* convexShapes, \n"
-"												__global const float4* vertices,\n"
-"												__global const float4* uniqueEdges,\n"
-"												__global const btGpuFace* faces,\n"
-"												__global const int* indices,\n"
-"												__global btAabbCL* aabbs,\n"
-"												__global struct b3Contact4Data* restrict globalContactsOut,\n"
-"												counter32_t nGlobalContactsOut,\n"
-"													int numConcavePairs, int maxContactCapacity\n"
-"												)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	if (i>=numConcavePairs)\n"
-"		return;\n"
-"	int pairIdx = i;\n"
-"	int bodyIndexA = concavePairs[i].x;\n"
-"	int bodyIndexB = concavePairs[i].y;\n"
-"	int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
-"	int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
-"	int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
-"	int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
-"	if (collidables[collidableIndexB].m_shapeType==SHAPE_SPHERE)\n"
-"	{\n"
-"		int f = concavePairs[i].z;\n"
-"		btGpuFace face = faces[convexShapes[shapeIndexA].m_faceOffset+f];\n"
-"		\n"
-"		float4 verticesA[3];\n"
-"		for (int i=0;i<3;i++)\n"
-"		{\n"
-"			int index = indices[face.m_indexOffset+i];\n"
-"			float4 vert = vertices[convexShapes[shapeIndexA].m_vertexOffset+index];\n"
-"			verticesA[i] = vert;\n"
-"		}\n"
-"		float4 spherePos = rigidBodies[bodyIndexB].m_pos;\n"
-"		float sphereRadius = collidables[collidableIndexB].m_radius;\n"
-"		float4 convexPos = rigidBodies[bodyIndexA].m_pos;\n"
-"		float4 convexOrn = rigidBodies[bodyIndexA].m_quat;\n"
-"		computeContactSphereTriangle(i, bodyIndexB, bodyIndexA, collidableIndexB, collidableIndexA, \n"
-"																rigidBodies,collidables,\n"
-"																verticesA,\n"
-"																globalContactsOut, nGlobalContactsOut,maxContactCapacity,\n"
-"																spherePos,sphereRadius,convexPos,convexOrn, f);\n"
-"		return;\n"
-"	}\n"
-"}\n"
-;
+static const char* primitiveContactsKernelsCL =
+	"#ifndef B3_CONTACT4DATA_H\n"
+	"#define B3_CONTACT4DATA_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#define B3_FLOAT4_H\n"
+	"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
+	"#define B3_PLATFORM_DEFINITIONS_H\n"
+	"struct MyTest\n"
+	"{\n"
+	"	int bla;\n"
+	"};\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
+	"#define B3_LARGE_FLOAT 1e18f\n"
+	"#define B3_INFINITY 1e18f\n"
+	"#define b3Assert(a)\n"
+	"#define b3ConstArray(a) __global const a*\n"
+	"#define b3AtomicInc atomic_inc\n"
+	"#define b3AtomicAdd atomic_add\n"
+	"#define b3Fabs fabs\n"
+	"#define b3Sqrt native_sqrt\n"
+	"#define b3Sin native_sin\n"
+	"#define b3Cos native_cos\n"
+	"#define B3_STATIC\n"
+	"#endif\n"
+	"#endif\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"	typedef float4	b3Float4;\n"
+	"	#define b3Float4ConstArg const b3Float4\n"
+	"	#define b3MakeFloat4 (float4)\n"
+	"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return dot(a1, b1);\n"
+	"	}\n"
+	"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return cross(a1, b1);\n"
+	"	}\n"
+	"	#define b3MinFloat4 min\n"
+	"	#define b3MaxFloat4 max\n"
+	"	#define b3Normalized(a) normalize(a)\n"
+	"#endif \n"
+	"		\n"
+	"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
+	"{\n"
+	"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
+	"		return false;\n"
+	"	return true;\n"
+	"}\n"
+	"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
+	"{\n"
+	"    float maxDot = -B3_INFINITY;\n"
+	"    int i = 0;\n"
+	"    int ptIndex = -1;\n"
+	"    for( i = 0; i < vecLen; i++ )\n"
+	"    {\n"
+	"        float dot = b3Dot3F4(vecArray[i],vec);\n"
+	"            \n"
+	"        if( dot > maxDot )\n"
+	"        {\n"
+	"            maxDot = dot;\n"
+	"            ptIndex = i;\n"
+	"        }\n"
+	"    }\n"
+	"	b3Assert(ptIndex>=0);\n"
+	"    if (ptIndex<0)\n"
+	"	{\n"
+	"		ptIndex = 0;\n"
+	"	}\n"
+	"    *dotOut = maxDot;\n"
+	"    return ptIndex;\n"
+	"}\n"
+	"#endif //B3_FLOAT4_H\n"
+	"typedef  struct b3Contact4Data b3Contact4Data_t;\n"
+	"struct b3Contact4Data\n"
+	"{\n"
+	"	b3Float4	m_worldPosB[4];\n"
+	"//	b3Float4	m_localPosA[4];\n"
+	"//	b3Float4	m_localPosB[4];\n"
+	"	b3Float4	m_worldNormalOnB;	//	w: m_nPoints\n"
+	"	unsigned short  m_restituitionCoeffCmp;\n"
+	"	unsigned short  m_frictionCoeffCmp;\n"
+	"	int m_batchIdx;\n"
+	"	int m_bodyAPtrAndSignBit;//x:m_bodyAPtr, y:m_bodyBPtr\n"
+	"	int m_bodyBPtrAndSignBit;\n"
+	"	int	m_childIndexA;\n"
+	"	int	m_childIndexB;\n"
+	"	int m_unused1;\n"
+	"	int m_unused2;\n"
+	"};\n"
+	"inline int b3Contact4Data_getNumPoints(const struct b3Contact4Data* contact)\n"
+	"{\n"
+	"	return (int)contact->m_worldNormalOnB.w;\n"
+	"};\n"
+	"inline void b3Contact4Data_setNumPoints(struct b3Contact4Data* contact, int numPoints)\n"
+	"{\n"
+	"	contact->m_worldNormalOnB.w = (float)numPoints;\n"
+	"};\n"
+	"#endif //B3_CONTACT4DATA_H\n"
+	"#define SHAPE_CONVEX_HULL 3\n"
+	"#define SHAPE_PLANE 4\n"
+	"#define SHAPE_CONCAVE_TRIMESH 5\n"
+	"#define SHAPE_COMPOUND_OF_CONVEX_HULLS 6\n"
+	"#define SHAPE_SPHERE 7\n"
+	"#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_extended_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics : enable\n"
+	"#ifdef cl_ext_atomic_counters_32\n"
+	"#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
+	"#else\n"
+	"#define counter32_t volatile __global int*\n"
+	"#endif\n"
+	"#define GET_GROUP_IDX get_group_id(0)\n"
+	"#define GET_LOCAL_IDX get_local_id(0)\n"
+	"#define GET_GLOBAL_IDX get_global_id(0)\n"
+	"#define GET_GROUP_SIZE get_local_size(0)\n"
+	"#define GET_NUM_GROUPS get_num_groups(0)\n"
+	"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
+	"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
+	"#define AtomInc(x) atom_inc(&(x))\n"
+	"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
+	"#define AppendInc(x, out) out = atomic_inc(x)\n"
+	"#define AtomAdd(x, value) atom_add(&(x), value)\n"
+	"#define AtomCmpxhg(x, cmp, value) atom_cmpxchg( &(x), cmp, value )\n"
+	"#define AtomXhg(x, value) atom_xchg ( &(x), value )\n"
+	"#define max2 max\n"
+	"#define min2 min\n"
+	"typedef unsigned int u32;\n"
+	"typedef struct \n"
+	"{\n"
+	"	union\n"
+	"	{\n"
+	"		float4	m_min;\n"
+	"		float   m_minElems[4];\n"
+	"		int			m_minIndices[4];\n"
+	"	};\n"
+	"	union\n"
+	"	{\n"
+	"		float4	m_max;\n"
+	"		float   m_maxElems[4];\n"
+	"		int			m_maxIndices[4];\n"
+	"	};\n"
+	"} btAabbCL;\n"
+	"///keep this in sync with btCollidable.h\n"
+	"typedef struct\n"
+	"{\n"
+	"	int m_numChildShapes;\n"
+	"	float m_radius;\n"
+	"	int m_shapeType;\n"
+	"	int m_shapeIndex;\n"
+	"	\n"
+	"} btCollidableGpu;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4	m_childPosition;\n"
+	"	float4	m_childOrientation;\n"
+	"	int m_shapeIndex;\n"
+	"	int m_unused0;\n"
+	"	int m_unused1;\n"
+	"	int m_unused2;\n"
+	"} btGpuChildShape;\n"
+	"#define GET_NPOINTS(x) (x).m_worldNormalOnB.w\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_pos;\n"
+	"	float4 m_quat;\n"
+	"	float4 m_linVel;\n"
+	"	float4 m_angVel;\n"
+	"	u32 m_collidableIdx;	\n"
+	"	float m_invMass;\n"
+	"	float m_restituitionCoeff;\n"
+	"	float m_frictionCoeff;\n"
+	"} BodyData;\n"
+	"typedef struct  \n"
+	"{\n"
+	"	float4		m_localCenter;\n"
+	"	float4		m_extents;\n"
+	"	float4		mC;\n"
+	"	float4		mE;\n"
+	"	\n"
+	"	float			m_radius;\n"
+	"	int	m_faceOffset;\n"
+	"	int m_numFaces;\n"
+	"	int	m_numVertices;\n"
+	"	\n"
+	"	int m_vertexOffset;\n"
+	"	int	m_uniqueEdgesOffset;\n"
+	"	int	m_numUniqueEdges;\n"
+	"	int m_unused;\n"
+	"} ConvexPolyhedronCL;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_plane;\n"
+	"	int m_indexOffset;\n"
+	"	int m_numIndices;\n"
+	"} btGpuFace;\n"
+	"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
+	"#define make_float4 (float4)\n"
+	"#define make_float2 (float2)\n"
+	"#define make_uint4 (uint4)\n"
+	"#define make_int4 (int4)\n"
+	"#define make_uint2 (uint2)\n"
+	"#define make_int2 (int2)\n"
+	"__inline\n"
+	"float fastDiv(float numerator, float denominator)\n"
+	"{\n"
+	"	return native_divide(numerator, denominator);	\n"
+	"//	return numerator/denominator;	\n"
+	"}\n"
+	"__inline\n"
+	"float4 fastDiv4(float4 numerator, float4 denominator)\n"
+	"{\n"
+	"	return native_divide(numerator, denominator);	\n"
+	"}\n"
+	"__inline\n"
+	"float4 cross3(float4 a, float4 b)\n"
+	"{\n"
+	"	return cross(a,b);\n"
+	"}\n"
+	"//#define dot3F4 dot\n"
+	"__inline\n"
+	"float dot3F4(float4 a, float4 b)\n"
+	"{\n"
+	"	float4 a1 = make_float4(a.xyz,0.f);\n"
+	"	float4 b1 = make_float4(b.xyz,0.f);\n"
+	"	return dot(a1, b1);\n"
+	"}\n"
+	"__inline\n"
+	"float4 fastNormalize4(float4 v)\n"
+	"{\n"
+	"	return fast_normalize(v);\n"
+	"}\n"
+	"///////////////////////////////////////\n"
+	"//	Quaternion\n"
+	"///////////////////////////////////////\n"
+	"typedef float4 Quaternion;\n"
+	"__inline\n"
+	"Quaternion qtMul(Quaternion a, Quaternion b);\n"
+	"__inline\n"
+	"Quaternion qtNormalize(Quaternion in);\n"
+	"__inline\n"
+	"float4 qtRotate(Quaternion q, float4 vec);\n"
+	"__inline\n"
+	"Quaternion qtInvert(Quaternion q);\n"
+	"__inline\n"
+	"Quaternion qtMul(Quaternion a, Quaternion b)\n"
+	"{\n"
+	"	Quaternion ans;\n"
+	"	ans = cross3( a, b );\n"
+	"	ans += a.w*b+b.w*a;\n"
+	"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
+	"	ans.w = a.w*b.w - dot3F4(a, b);\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"Quaternion qtNormalize(Quaternion in)\n"
+	"{\n"
+	"	return fastNormalize4(in);\n"
+	"//	in /= length( in );\n"
+	"//	return in;\n"
+	"}\n"
+	"__inline\n"
+	"float4 qtRotate(Quaternion q, float4 vec)\n"
+	"{\n"
+	"	Quaternion qInv = qtInvert( q );\n"
+	"	float4 vcpy = vec;\n"
+	"	vcpy.w = 0.f;\n"
+	"	float4 out = qtMul(qtMul(q,vcpy),qInv);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"Quaternion qtInvert(Quaternion q)\n"
+	"{\n"
+	"	return (Quaternion)(-q.xyz, q.w);\n"
+	"}\n"
+	"__inline\n"
+	"float4 qtInvRotate(const Quaternion q, float4 vec)\n"
+	"{\n"
+	"	return qtRotate( qtInvert( q ), vec );\n"
+	"}\n"
+	"__inline\n"
+	"float4 transform(const float4* p, const float4* translation, const Quaternion* orientation)\n"
+	"{\n"
+	"	return qtRotate( *orientation, *p ) + (*translation);\n"
+	"}\n"
+	"void	trInverse(float4 translationIn, Quaternion orientationIn,\n"
+	"		float4* translationOut, Quaternion* orientationOut)\n"
+	"{\n"
+	"	*orientationOut = qtInvert(orientationIn);\n"
+	"	*translationOut = qtRotate(*orientationOut, -translationIn);\n"
+	"}\n"
+	"void	trMul(float4 translationA, Quaternion orientationA,\n"
+	"						float4 translationB, Quaternion orientationB,\n"
+	"		float4* translationOut, Quaternion* orientationOut)\n"
+	"{\n"
+	"	*orientationOut = qtMul(orientationA,orientationB);\n"
+	"	*translationOut = transform(&translationB,&translationA,&orientationA);\n"
+	"}\n"
+	"__inline\n"
+	"float4 normalize3(const float4 a)\n"
+	"{\n"
+	"	float4 n = make_float4(a.x, a.y, a.z, 0.f);\n"
+	"	return fastNormalize4( n );\n"
+	"}\n"
+	"__inline float4 lerp3(const float4 a,const float4 b, float  t)\n"
+	"{\n"
+	"	return make_float4(	a.x + (b.x - a.x) * t,\n"
+	"						a.y + (b.y - a.y) * t,\n"
+	"						a.z + (b.z - a.z) * t,\n"
+	"						0.f);\n"
+	"}\n"
+	"float signedDistanceFromPointToPlane(float4 point, float4 planeEqn, float4* closestPointOnFace)\n"
+	"{\n"
+	"	float4 n = (float4)(planeEqn.x, planeEqn.y, planeEqn.z, 0);\n"
+	"	float dist = dot3F4(n, point) + planeEqn.w;\n"
+	"	*closestPointOnFace = point - dist * n;\n"
+	"	return dist;\n"
+	"}\n"
+	"inline bool IsPointInPolygon(float4 p, \n"
+	"							const btGpuFace* face,\n"
+	"							__global const float4* baseVertex,\n"
+	"							__global const  int* convexIndices,\n"
+	"							float4* out)\n"
+	"{\n"
+	"    float4 a;\n"
+	"    float4 b;\n"
+	"    float4 ab;\n"
+	"    float4 ap;\n"
+	"    float4 v;\n"
+	"	float4 plane = make_float4(face->m_plane.x,face->m_plane.y,face->m_plane.z,0.f);\n"
+	"	\n"
+	"	if (face->m_numIndices<2)\n"
+	"		return false;\n"
+	"	\n"
+	"	float4 v0 = baseVertex[convexIndices[face->m_indexOffset + face->m_numIndices-1]];\n"
+	"	\n"
+	"	b = v0;\n"
+	"    for(unsigned i=0; i != face->m_numIndices; ++i)\n"
+	"    {\n"
+	"		a = b;\n"
+	"		float4 vi = baseVertex[convexIndices[face->m_indexOffset + i]];\n"
+	"		b = vi;\n"
+	"        ab = b-a;\n"
+	"        ap = p-a;\n"
+	"        v = cross3(ab,plane);\n"
+	"        if (dot(ap, v) > 0.f)\n"
+	"        {\n"
+	"            float ab_m2 = dot(ab, ab);\n"
+	"            float rt = ab_m2 != 0.f ? dot(ab, ap) / ab_m2 : 0.f;\n"
+	"            if (rt <= 0.f)\n"
+	"            {\n"
+	"                *out = a;\n"
+	"            }\n"
+	"            else if (rt >= 1.f) \n"
+	"            {\n"
+	"                *out = b;\n"
+	"            }\n"
+	"            else\n"
+	"            {\n"
+	"            	float s = 1.f - rt;\n"
+	"				out[0].x = s * a.x + rt * b.x;\n"
+	"				out[0].y = s * a.y + rt * b.y;\n"
+	"				out[0].z = s * a.z + rt * b.z;\n"
+	"            }\n"
+	"            return false;\n"
+	"        }\n"
+	"    }\n"
+	"    return true;\n"
+	"}\n"
+	"void	computeContactSphereConvex(int pairIndex,\n"
+	"																int bodyIndexA, int bodyIndexB, \n"
+	"																int collidableIndexA, int collidableIndexB, \n"
+	"																__global const BodyData* rigidBodies, \n"
+	"																__global const btCollidableGpu* collidables,\n"
+	"																__global const ConvexPolyhedronCL* convexShapes,\n"
+	"																__global const float4* convexVertices,\n"
+	"																__global const int* convexIndices,\n"
+	"																__global const btGpuFace* faces,\n"
+	"																__global struct b3Contact4Data* restrict globalContactsOut,\n"
+	"																counter32_t nGlobalContactsOut,\n"
+	"																int maxContactCapacity,\n"
+	"																float4 spherePos2,\n"
+	"																float radius,\n"
+	"																float4 pos,\n"
+	"																float4 quat\n"
+	"																)\n"
+	"{\n"
+	"	float4 invPos;\n"
+	"	float4 invOrn;\n"
+	"	trInverse(pos,quat, &invPos,&invOrn);\n"
+	"	float4 spherePos = transform(&spherePos2,&invPos,&invOrn);\n"
+	"	int shapeIndex = collidables[collidableIndexB].m_shapeIndex;\n"
+	"	int numFaces = convexShapes[shapeIndex].m_numFaces;\n"
+	"	float4 closestPnt = (float4)(0, 0, 0, 0);\n"
+	"	float4 hitNormalWorld = (float4)(0, 0, 0, 0);\n"
+	"	float minDist = -1000000.f;\n"
+	"	bool bCollide = true;\n"
+	"	for ( int f = 0; f < numFaces; f++ )\n"
+	"	{\n"
+	"		btGpuFace face = faces[convexShapes[shapeIndex].m_faceOffset+f];\n"
+	"		// set up a plane equation \n"
+	"		float4 planeEqn;\n"
+	"		float4 n1 = face.m_plane;\n"
+	"		n1.w = 0.f;\n"
+	"		planeEqn = n1;\n"
+	"		planeEqn.w = face.m_plane.w;\n"
+	"		\n"
+	"	\n"
+	"		// compute a signed distance from the vertex in cloth to the face of rigidbody.\n"
+	"		float4 pntReturn;\n"
+	"		float dist = signedDistanceFromPointToPlane(spherePos, planeEqn, &pntReturn);\n"
+	"		// If the distance is positive, the plane is a separating plane. \n"
+	"		if ( dist > radius )\n"
+	"		{\n"
+	"			bCollide = false;\n"
+	"			break;\n"
+	"		}\n"
+	"		if (dist>0)\n"
+	"		{\n"
+	"			//might hit an edge or vertex\n"
+	"			float4 out;\n"
+	"			float4 zeroPos = make_float4(0,0,0,0);\n"
+	"			bool isInPoly = IsPointInPolygon(spherePos,\n"
+	"					&face,\n"
+	"					&convexVertices[convexShapes[shapeIndex].m_vertexOffset],\n"
+	"					convexIndices,\n"
+	"           &out);\n"
+	"			if (isInPoly)\n"
+	"			{\n"
+	"				if (dist>minDist)\n"
+	"				{\n"
+	"					minDist = dist;\n"
+	"					closestPnt = pntReturn;\n"
+	"					hitNormalWorld = planeEqn;\n"
+	"					\n"
+	"				}\n"
+	"			} else\n"
+	"			{\n"
+	"				float4 tmp = spherePos-out;\n"
+	"				float l2 = dot(tmp,tmp);\n"
+	"				if (l2<radius*radius)\n"
+	"				{\n"
+	"					dist  = sqrt(l2);\n"
+	"					if (dist>minDist)\n"
+	"					{\n"
+	"						minDist = dist;\n"
+	"						closestPnt = out;\n"
+	"						hitNormalWorld = tmp/dist;\n"
+	"						\n"
+	"					}\n"
+	"					\n"
+	"				} else\n"
+	"				{\n"
+	"					bCollide = false;\n"
+	"					break;\n"
+	"				}\n"
+	"			}\n"
+	"		} else\n"
+	"		{\n"
+	"			if ( dist > minDist )\n"
+	"			{\n"
+	"				minDist = dist;\n"
+	"				closestPnt = pntReturn;\n"
+	"				hitNormalWorld.xyz = planeEqn.xyz;\n"
+	"			}\n"
+	"		}\n"
+	"		\n"
+	"	}\n"
+	"	\n"
+	"	if (bCollide && minDist > -10000)\n"
+	"	{\n"
+	"		float4 normalOnSurfaceB1 = qtRotate(quat,-hitNormalWorld);\n"
+	"		float4 pOnB1 = transform(&closestPnt,&pos,&quat);\n"
+	"		\n"
+	"		float actualDepth = minDist-radius;\n"
+	"		if (actualDepth<=0.f)\n"
+	"		{\n"
+	"			\n"
+	"			pOnB1.w = actualDepth;\n"
+	"			int dstIdx;\n"
+	"			AppendInc( nGlobalContactsOut, dstIdx );\n"
+	"		\n"
+	"			\n"
+	"			if (1)//dstIdx < maxContactCapacity)\n"
+	"			{\n"
+	"				__global struct b3Contact4Data* c = &globalContactsOut[dstIdx];\n"
+	"				c->m_worldNormalOnB = -normalOnSurfaceB1;\n"
+	"				c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
+	"				c->m_batchIdx = pairIndex;\n"
+	"				c->m_bodyAPtrAndSignBit = rigidBodies[bodyIndexA].m_invMass==0?-bodyIndexA:bodyIndexA;\n"
+	"				c->m_bodyBPtrAndSignBit = rigidBodies[bodyIndexB].m_invMass==0?-bodyIndexB:bodyIndexB;\n"
+	"				c->m_worldPosB[0] = pOnB1;\n"
+	"				c->m_childIndexA = -1;\n"
+	"				c->m_childIndexB = -1;\n"
+	"				GET_NPOINTS(*c) = 1;\n"
+	"			} \n"
+	"		}\n"
+	"	}//if (hasCollision)\n"
+	"}\n"
+	"							\n"
+	"int extractManifoldSequential(const float4* p, int nPoints, float4 nearNormal, int4* contactIdx)\n"
+	"{\n"
+	"	if( nPoints == 0 )\n"
+	"        return 0;\n"
+	"    \n"
+	"    if (nPoints <=4)\n"
+	"        return nPoints;\n"
+	"    \n"
+	"    \n"
+	"    if (nPoints >64)\n"
+	"        nPoints = 64;\n"
+	"    \n"
+	"	float4 center = make_float4(0.f);\n"
+	"	{\n"
+	"		\n"
+	"		for (int i=0;i<nPoints;i++)\n"
+	"			center += p[i];\n"
+	"		center /= (float)nPoints;\n"
+	"	}\n"
+	"    \n"
+	"	\n"
+	"    \n"
+	"	//	sample 4 directions\n"
+	"    \n"
+	"    float4 aVector = p[0] - center;\n"
+	"    float4 u = cross3( nearNormal, aVector );\n"
+	"    float4 v = cross3( nearNormal, u );\n"
+	"    u = normalize3( u );\n"
+	"    v = normalize3( v );\n"
+	"    \n"
+	"    \n"
+	"    //keep point with deepest penetration\n"
+	"    float minW= FLT_MAX;\n"
+	"    \n"
+	"    int minIndex=-1;\n"
+	"    \n"
+	"    float4 maxDots;\n"
+	"    maxDots.x = FLT_MIN;\n"
+	"    maxDots.y = FLT_MIN;\n"
+	"    maxDots.z = FLT_MIN;\n"
+	"    maxDots.w = FLT_MIN;\n"
+	"    \n"
+	"    //	idx, distance\n"
+	"    for(int ie = 0; ie<nPoints; ie++ )\n"
+	"    {\n"
+	"        if (p[ie].w<minW)\n"
+	"        {\n"
+	"            minW = p[ie].w;\n"
+	"            minIndex=ie;\n"
+	"        }\n"
+	"        float f;\n"
+	"        float4 r = p[ie]-center;\n"
+	"        f = dot3F4( u, r );\n"
+	"        if (f<maxDots.x)\n"
+	"        {\n"
+	"            maxDots.x = f;\n"
+	"            contactIdx[0].x = ie;\n"
+	"        }\n"
+	"        \n"
+	"        f = dot3F4( -u, r );\n"
+	"        if (f<maxDots.y)\n"
+	"        {\n"
+	"            maxDots.y = f;\n"
+	"            contactIdx[0].y = ie;\n"
+	"        }\n"
+	"        \n"
+	"        \n"
+	"        f = dot3F4( v, r );\n"
+	"        if (f<maxDots.z)\n"
+	"        {\n"
+	"            maxDots.z = f;\n"
+	"            contactIdx[0].z = ie;\n"
+	"        }\n"
+	"        \n"
+	"        f = dot3F4( -v, r );\n"
+	"        if (f<maxDots.w)\n"
+	"        {\n"
+	"            maxDots.w = f;\n"
+	"            contactIdx[0].w = ie;\n"
+	"        }\n"
+	"        \n"
+	"    }\n"
+	"    \n"
+	"    if (contactIdx[0].x != minIndex && contactIdx[0].y != minIndex && contactIdx[0].z != minIndex && contactIdx[0].w != minIndex)\n"
+	"    {\n"
+	"        //replace the first contact with minimum (todo: replace contact with least penetration)\n"
+	"        contactIdx[0].x = minIndex;\n"
+	"    }\n"
+	"    \n"
+	"    return 4;\n"
+	"    \n"
+	"}\n"
+	"#define MAX_PLANE_CONVEX_POINTS 64\n"
+	"int computeContactPlaneConvex(int pairIndex,\n"
+	"								int bodyIndexA, int bodyIndexB, \n"
+	"								int collidableIndexA, int collidableIndexB, \n"
+	"								__global const BodyData* rigidBodies, \n"
+	"								__global const btCollidableGpu*collidables,\n"
+	"								__global const ConvexPolyhedronCL* convexShapes,\n"
+	"								__global const float4* convexVertices,\n"
+	"								__global const int* convexIndices,\n"
+	"								__global const btGpuFace* faces,\n"
+	"								__global struct b3Contact4Data* restrict globalContactsOut,\n"
+	"								counter32_t nGlobalContactsOut,\n"
+	"								int maxContactCapacity,\n"
+	"								float4 posB,\n"
+	"								Quaternion ornB\n"
+	"								)\n"
+	"{\n"
+	"	int resultIndex=-1;\n"
+	"		int shapeIndex = collidables[collidableIndexB].m_shapeIndex;\n"
+	"	__global const ConvexPolyhedronCL* hullB = &convexShapes[shapeIndex];\n"
+	"	\n"
+	"	float4 posA;\n"
+	"	posA = rigidBodies[bodyIndexA].m_pos;\n"
+	"	Quaternion ornA;\n"
+	"	ornA = rigidBodies[bodyIndexA].m_quat;\n"
+	"	int numContactsOut = 0;\n"
+	"	int numWorldVertsB1= 0;\n"
+	"	float4 planeEq;\n"
+	"	 planeEq = faces[collidables[collidableIndexA].m_shapeIndex].m_plane;\n"
+	"	float4 planeNormal = make_float4(planeEq.x,planeEq.y,planeEq.z,0.f);\n"
+	"	float4 planeNormalWorld;\n"
+	"	planeNormalWorld = qtRotate(ornA,planeNormal);\n"
+	"	float planeConstant = planeEq.w;\n"
+	"	\n"
+	"	float4 invPosA;Quaternion invOrnA;\n"
+	"	float4 convexInPlaneTransPos1; Quaternion convexInPlaneTransOrn1;\n"
+	"	{\n"
+	"		\n"
+	"		trInverse(posA,ornA,&invPosA,&invOrnA);\n"
+	"		trMul(invPosA,invOrnA,posB,ornB,&convexInPlaneTransPos1,&convexInPlaneTransOrn1);\n"
+	"	}\n"
+	"	float4 invPosB;Quaternion invOrnB;\n"
+	"	float4 planeInConvexPos1;	Quaternion planeInConvexOrn1;\n"
+	"	{\n"
+	"		\n"
+	"		trInverse(posB,ornB,&invPosB,&invOrnB);\n"
+	"		trMul(invPosB,invOrnB,posA,ornA,&planeInConvexPos1,&planeInConvexOrn1);	\n"
+	"	}\n"
+	"	\n"
+	"	float4 planeNormalInConvex = qtRotate(planeInConvexOrn1,-planeNormal);\n"
+	"	float maxDot = -1e30;\n"
+	"	int hitVertex=-1;\n"
+	"	float4 hitVtx;\n"
+	"	float4 contactPoints[MAX_PLANE_CONVEX_POINTS];\n"
+	"	int numPoints = 0;\n"
+	"	int4 contactIdx;\n"
+	"	contactIdx=make_int4(0,1,2,3);\n"
+	"    \n"
+	"	\n"
+	"	for (int i=0;i<hullB->m_numVertices;i++)\n"
+	"	{\n"
+	"		float4 vtx = convexVertices[hullB->m_vertexOffset+i];\n"
+	"		float curDot = dot(vtx,planeNormalInConvex);\n"
+	"		if (curDot>maxDot)\n"
+	"		{\n"
+	"			hitVertex=i;\n"
+	"			maxDot=curDot;\n"
+	"			hitVtx = vtx;\n"
+	"			//make sure the deepest points is always included\n"
+	"			if (numPoints==MAX_PLANE_CONVEX_POINTS)\n"
+	"				numPoints--;\n"
+	"		}\n"
+	"		if (numPoints<MAX_PLANE_CONVEX_POINTS)\n"
+	"		{\n"
+	"			float4 vtxWorld = transform(&vtx, &posB, &ornB);\n"
+	"			float4 vtxInPlane = transform(&vtxWorld, &invPosA, &invOrnA);//oplaneTransform.inverse()*vtxWorld;\n"
+	"			float dist = dot(planeNormal,vtxInPlane)-planeConstant;\n"
+	"			if (dist<0.f)\n"
+	"			{\n"
+	"				vtxWorld.w = dist;\n"
+	"				contactPoints[numPoints] = vtxWorld;\n"
+	"				numPoints++;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	int numReducedPoints  = numPoints;\n"
+	"	if (numPoints>4)\n"
+	"	{\n"
+	"		numReducedPoints = extractManifoldSequential( contactPoints, numPoints, planeNormalInConvex, &contactIdx);\n"
+	"	}\n"
+	"	if (numReducedPoints>0)\n"
+	"	{\n"
+	"		int dstIdx;\n"
+	"	    AppendInc( nGlobalContactsOut, dstIdx );\n"
+	"		if (dstIdx < maxContactCapacity)\n"
+	"		{\n"
+	"			resultIndex = dstIdx;\n"
+	"			__global struct b3Contact4Data* c = &globalContactsOut[dstIdx];\n"
+	"			c->m_worldNormalOnB = -planeNormalWorld;\n"
+	"			//c->setFrictionCoeff(0.7);\n"
+	"			//c->setRestituitionCoeff(0.f);\n"
+	"			c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
+	"			c->m_batchIdx = pairIndex;\n"
+	"			c->m_bodyAPtrAndSignBit = rigidBodies[bodyIndexA].m_invMass==0?-bodyIndexA:bodyIndexA;\n"
+	"			c->m_bodyBPtrAndSignBit = rigidBodies[bodyIndexB].m_invMass==0?-bodyIndexB:bodyIndexB;\n"
+	"			c->m_childIndexA = -1;\n"
+	"			c->m_childIndexB = -1;\n"
+	"			switch (numReducedPoints)\n"
+	"            {\n"
+	"                case 4:\n"
+	"                    c->m_worldPosB[3] = contactPoints[contactIdx.w];\n"
+	"                case 3:\n"
+	"                    c->m_worldPosB[2] = contactPoints[contactIdx.z];\n"
+	"                case 2:\n"
+	"                    c->m_worldPosB[1] = contactPoints[contactIdx.y];\n"
+	"                case 1:\n"
+	"                    c->m_worldPosB[0] = contactPoints[contactIdx.x];\n"
+	"                default:\n"
+	"                {\n"
+	"                }\n"
+	"            };\n"
+	"			\n"
+	"			GET_NPOINTS(*c) = numReducedPoints;\n"
+	"		}//if (dstIdx < numPairs)\n"
+	"	}	\n"
+	"	return resultIndex;\n"
+	"}\n"
+	"void	computeContactPlaneSphere(int pairIndex,\n"
+	"																int bodyIndexA, int bodyIndexB, \n"
+	"																int collidableIndexA, int collidableIndexB, \n"
+	"																__global const BodyData* rigidBodies, \n"
+	"																__global const btCollidableGpu* collidables,\n"
+	"																__global const btGpuFace* faces,\n"
+	"																__global struct b3Contact4Data* restrict globalContactsOut,\n"
+	"																counter32_t nGlobalContactsOut,\n"
+	"																int maxContactCapacity)\n"
+	"{\n"
+	"	float4 planeEq = faces[collidables[collidableIndexA].m_shapeIndex].m_plane;\n"
+	"	float radius = collidables[collidableIndexB].m_radius;\n"
+	"	float4 posA1 = rigidBodies[bodyIndexA].m_pos;\n"
+	"	float4 ornA1 = rigidBodies[bodyIndexA].m_quat;\n"
+	"	float4 posB1 = rigidBodies[bodyIndexB].m_pos;\n"
+	"	float4 ornB1 = rigidBodies[bodyIndexB].m_quat;\n"
+	"	\n"
+	"	bool hasCollision = false;\n"
+	"	float4 planeNormal1 = make_float4(planeEq.x,planeEq.y,planeEq.z,0.f);\n"
+	"	float planeConstant = planeEq.w;\n"
+	"	float4 convexInPlaneTransPos1; Quaternion convexInPlaneTransOrn1;\n"
+	"	{\n"
+	"		float4 invPosA;Quaternion invOrnA;\n"
+	"		trInverse(posA1,ornA1,&invPosA,&invOrnA);\n"
+	"		trMul(invPosA,invOrnA,posB1,ornB1,&convexInPlaneTransPos1,&convexInPlaneTransOrn1);\n"
+	"	}\n"
+	"	float4 planeInConvexPos1;	Quaternion planeInConvexOrn1;\n"
+	"	{\n"
+	"		float4 invPosB;Quaternion invOrnB;\n"
+	"		trInverse(posB1,ornB1,&invPosB,&invOrnB);\n"
+	"		trMul(invPosB,invOrnB,posA1,ornA1,&planeInConvexPos1,&planeInConvexOrn1);	\n"
+	"	}\n"
+	"	float4 vtx1 = qtRotate(planeInConvexOrn1,-planeNormal1)*radius;\n"
+	"	float4 vtxInPlane1 = transform(&vtx1,&convexInPlaneTransPos1,&convexInPlaneTransOrn1);\n"
+	"	float distance = dot3F4(planeNormal1,vtxInPlane1) - planeConstant;\n"
+	"	hasCollision = distance < 0.f;//m_manifoldPtr->getContactBreakingThreshold();\n"
+	"	if (hasCollision)\n"
+	"	{\n"
+	"		float4 vtxInPlaneProjected1 = vtxInPlane1 -   distance*planeNormal1;\n"
+	"		float4 vtxInPlaneWorld1 = transform(&vtxInPlaneProjected1,&posA1,&ornA1);\n"
+	"		float4 normalOnSurfaceB1 = qtRotate(ornA1,planeNormal1);\n"
+	"		float4 pOnB1 = vtxInPlaneWorld1+normalOnSurfaceB1*distance;\n"
+	"		pOnB1.w = distance;\n"
+	"		int dstIdx;\n"
+	"    AppendInc( nGlobalContactsOut, dstIdx );\n"
+	"		\n"
+	"		if (dstIdx < maxContactCapacity)\n"
+	"		{\n"
+	"			__global struct b3Contact4Data* c = &globalContactsOut[dstIdx];\n"
+	"			c->m_worldNormalOnB = -normalOnSurfaceB1;\n"
+	"			c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
+	"			c->m_batchIdx = pairIndex;\n"
+	"			c->m_bodyAPtrAndSignBit = rigidBodies[bodyIndexA].m_invMass==0?-bodyIndexA:bodyIndexA;\n"
+	"			c->m_bodyBPtrAndSignBit = rigidBodies[bodyIndexB].m_invMass==0?-bodyIndexB:bodyIndexB;\n"
+	"			c->m_worldPosB[0] = pOnB1;\n"
+	"			c->m_childIndexA = -1;\n"
+	"			c->m_childIndexB = -1;\n"
+	"			GET_NPOINTS(*c) = 1;\n"
+	"		}//if (dstIdx < numPairs)\n"
+	"	}//if (hasCollision)\n"
+	"}\n"
+	"__kernel void   primitiveContactsKernel( __global int4* pairs, \n"
+	"																					__global const BodyData* rigidBodies, \n"
+	"																					__global const btCollidableGpu* collidables,\n"
+	"																					__global const ConvexPolyhedronCL* convexShapes, \n"
+	"																					__global const float4* vertices,\n"
+	"																					__global const float4* uniqueEdges,\n"
+	"																					__global const btGpuFace* faces,\n"
+	"																					__global const int* indices,\n"
+	"																					__global struct b3Contact4Data* restrict globalContactsOut,\n"
+	"																					counter32_t nGlobalContactsOut,\n"
+	"																					int numPairs, int maxContactCapacity)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	int pairIndex = i;\n"
+	"	\n"
+	"	float4 worldVertsB1[64];\n"
+	"	float4 worldVertsB2[64];\n"
+	"	int capacityWorldVerts = 64;	\n"
+	"	float4 localContactsOut[64];\n"
+	"	int localContactCapacity=64;\n"
+	"	\n"
+	"	float minDist = -1e30f;\n"
+	"	float maxDist = 0.02f;\n"
+	"	if (i<numPairs)\n"
+	"	{\n"
+	"		int bodyIndexA = pairs[i].x;\n"
+	"		int bodyIndexB = pairs[i].y;\n"
+	"			\n"
+	"		int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
+	"		int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
+	"	\n"
+	"		if (collidables[collidableIndexA].m_shapeType == SHAPE_PLANE &&\n"
+	"			collidables[collidableIndexB].m_shapeType == SHAPE_CONVEX_HULL)\n"
+	"		{\n"
+	"			float4 posB;\n"
+	"			posB = rigidBodies[bodyIndexB].m_pos;\n"
+	"			Quaternion ornB;\n"
+	"			ornB = rigidBodies[bodyIndexB].m_quat;\n"
+	"			int contactIndex = computeContactPlaneConvex(pairIndex, bodyIndexA, bodyIndexB, collidableIndexA, collidableIndexB, \n"
+	"																rigidBodies,collidables,convexShapes,vertices,indices,\n"
+	"																faces,	globalContactsOut, nGlobalContactsOut,maxContactCapacity, posB,ornB);\n"
+	"			if (contactIndex>=0)\n"
+	"				pairs[pairIndex].z = contactIndex;\n"
+	"			return;\n"
+	"		}\n"
+	"		if (collidables[collidableIndexA].m_shapeType == SHAPE_CONVEX_HULL &&\n"
+	"			collidables[collidableIndexB].m_shapeType == SHAPE_PLANE)\n"
+	"		{\n"
+	"			float4 posA;\n"
+	"			posA = rigidBodies[bodyIndexA].m_pos;\n"
+	"			Quaternion ornA;\n"
+	"			ornA = rigidBodies[bodyIndexA].m_quat;\n"
+	"			int contactIndex = computeContactPlaneConvex( pairIndex, bodyIndexB,bodyIndexA,  collidableIndexB,collidableIndexA, \n"
+	"																rigidBodies,collidables,convexShapes,vertices,indices,\n"
+	"																faces,	globalContactsOut, nGlobalContactsOut,maxContactCapacity,posA,ornA);\n"
+	"			if (contactIndex>=0)\n"
+	"				pairs[pairIndex].z = contactIndex;\n"
+	"			return;\n"
+	"		}\n"
+	"		if (collidables[collidableIndexA].m_shapeType == SHAPE_PLANE &&\n"
+	"			collidables[collidableIndexB].m_shapeType == SHAPE_SPHERE)\n"
+	"		{\n"
+	"			computeContactPlaneSphere(pairIndex, bodyIndexA, bodyIndexB, collidableIndexA, collidableIndexB, \n"
+	"																rigidBodies,collidables,faces,	globalContactsOut, nGlobalContactsOut,maxContactCapacity);\n"
+	"			return;\n"
+	"		}\n"
+	"		if (collidables[collidableIndexA].m_shapeType == SHAPE_SPHERE &&\n"
+	"			collidables[collidableIndexB].m_shapeType == SHAPE_PLANE)\n"
+	"		{\n"
+	"			computeContactPlaneSphere( pairIndex, bodyIndexB,bodyIndexA,  collidableIndexB,collidableIndexA, \n"
+	"																rigidBodies,collidables,\n"
+	"																faces,	globalContactsOut, nGlobalContactsOut,maxContactCapacity);\n"
+	"			return;\n"
+	"		}\n"
+	"		\n"
+	"	\n"
+	"		if (collidables[collidableIndexA].m_shapeType == SHAPE_SPHERE &&\n"
+	"			collidables[collidableIndexB].m_shapeType == SHAPE_CONVEX_HULL)\n"
+	"		{\n"
+	"		\n"
+	"			float4 spherePos = rigidBodies[bodyIndexA].m_pos;\n"
+	"			float sphereRadius = collidables[collidableIndexA].m_radius;\n"
+	"			float4 convexPos = rigidBodies[bodyIndexB].m_pos;\n"
+	"			float4 convexOrn = rigidBodies[bodyIndexB].m_quat;\n"
+	"			computeContactSphereConvex(pairIndex, bodyIndexA, bodyIndexB, collidableIndexA, collidableIndexB, \n"
+	"																rigidBodies,collidables,convexShapes,vertices,indices,faces, globalContactsOut, nGlobalContactsOut,maxContactCapacity,\n"
+	"																spherePos,sphereRadius,convexPos,convexOrn);\n"
+	"			return;\n"
+	"		}\n"
+	"		if (collidables[collidableIndexA].m_shapeType == SHAPE_CONVEX_HULL &&\n"
+	"			collidables[collidableIndexB].m_shapeType == SHAPE_SPHERE)\n"
+	"		{\n"
+	"		\n"
+	"			float4 spherePos = rigidBodies[bodyIndexB].m_pos;\n"
+	"			float sphereRadius = collidables[collidableIndexB].m_radius;\n"
+	"			float4 convexPos = rigidBodies[bodyIndexA].m_pos;\n"
+	"			float4 convexOrn = rigidBodies[bodyIndexA].m_quat;\n"
+	"			computeContactSphereConvex(pairIndex, bodyIndexB, bodyIndexA, collidableIndexB, collidableIndexA, \n"
+	"																rigidBodies,collidables,convexShapes,vertices,indices,faces, globalContactsOut, nGlobalContactsOut,maxContactCapacity,\n"
+	"																spherePos,sphereRadius,convexPos,convexOrn);\n"
+	"			return;\n"
+	"		}\n"
+	"	\n"
+	"	\n"
+	"	\n"
+	"		\n"
+	"	\n"
+	"	\n"
+	"		if (collidables[collidableIndexA].m_shapeType == SHAPE_SPHERE &&\n"
+	"			collidables[collidableIndexB].m_shapeType == SHAPE_SPHERE)\n"
+	"		{\n"
+	"			//sphere-sphere\n"
+	"			float radiusA = collidables[collidableIndexA].m_radius;\n"
+	"			float radiusB = collidables[collidableIndexB].m_radius;\n"
+	"			float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
+	"			float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
+	"			float4 diff = posA-posB;\n"
+	"			float len = length(diff);\n"
+	"			\n"
+	"			///iff distance positive, don't generate a new contact\n"
+	"			if ( len <= (radiusA+radiusB))\n"
+	"			{\n"
+	"				///distance (negative means penetration)\n"
+	"				float dist = len - (radiusA+radiusB);\n"
+	"				float4 normalOnSurfaceB = make_float4(1.f,0.f,0.f,0.f);\n"
+	"				if (len > 0.00001)\n"
+	"				{\n"
+	"					normalOnSurfaceB = diff / len;\n"
+	"				}\n"
+	"				float4 contactPosB = posB + normalOnSurfaceB*radiusB;\n"
+	"				contactPosB.w = dist;\n"
+	"								\n"
+	"				int dstIdx;\n"
+	"				 AppendInc( nGlobalContactsOut, dstIdx );\n"
+	"				\n"
+	"				if (dstIdx < maxContactCapacity)\n"
+	"				{\n"
+	"					__global struct b3Contact4Data* c = &globalContactsOut[dstIdx];\n"
+	"					c->m_worldNormalOnB = normalOnSurfaceB;\n"
+	"					c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
+	"					c->m_batchIdx = pairIndex;\n"
+	"					int bodyA = pairs[pairIndex].x;\n"
+	"					int bodyB = pairs[pairIndex].y;\n"
+	"					c->m_bodyAPtrAndSignBit = rigidBodies[bodyA].m_invMass==0?-bodyA:bodyA;\n"
+	"					c->m_bodyBPtrAndSignBit = rigidBodies[bodyB].m_invMass==0?-bodyB:bodyB;\n"
+	"					c->m_worldPosB[0] = contactPosB;\n"
+	"					c->m_childIndexA = -1;\n"
+	"					c->m_childIndexB = -1;\n"
+	"					GET_NPOINTS(*c) = 1;\n"
+	"				}//if (dstIdx < numPairs)\n"
+	"			}//if ( len <= (radiusA+radiusB))\n"
+	"			return;\n"
+	"		}//SHAPE_SPHERE SHAPE_SPHERE\n"
+	"	}//	if (i<numPairs)\n"
+	"}\n"
+	"// work-in-progress\n"
+	"__kernel void   processCompoundPairsPrimitivesKernel( __global const int4* gpuCompoundPairs,\n"
+	"													__global const BodyData* rigidBodies, \n"
+	"													__global const btCollidableGpu* collidables,\n"
+	"													__global const ConvexPolyhedronCL* convexShapes, \n"
+	"													__global const float4* vertices,\n"
+	"													__global const float4* uniqueEdges,\n"
+	"													__global const btGpuFace* faces,\n"
+	"													__global const int* indices,\n"
+	"													__global btAabbCL* aabbs,\n"
+	"													__global const btGpuChildShape* gpuChildShapes,\n"
+	"													__global struct b3Contact4Data* restrict globalContactsOut,\n"
+	"													counter32_t nGlobalContactsOut,\n"
+	"													int numCompoundPairs, int maxContactCapacity\n"
+	"													)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	if (i<numCompoundPairs)\n"
+	"	{\n"
+	"		int bodyIndexA = gpuCompoundPairs[i].x;\n"
+	"		int bodyIndexB = gpuCompoundPairs[i].y;\n"
+	"		int childShapeIndexA = gpuCompoundPairs[i].z;\n"
+	"		int childShapeIndexB = gpuCompoundPairs[i].w;\n"
+	"		\n"
+	"		int collidableIndexA = -1;\n"
+	"		int collidableIndexB = -1;\n"
+	"		\n"
+	"		float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
+	"		float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
+	"		\n"
+	"		float4 ornB = rigidBodies[bodyIndexB].m_quat;\n"
+	"		float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
+	"							\n"
+	"		if (childShapeIndexA >= 0)\n"
+	"		{\n"
+	"			collidableIndexA = gpuChildShapes[childShapeIndexA].m_shapeIndex;\n"
+	"			float4 childPosA = gpuChildShapes[childShapeIndexA].m_childPosition;\n"
+	"			float4 childOrnA = gpuChildShapes[childShapeIndexA].m_childOrientation;\n"
+	"			float4 newPosA = qtRotate(ornA,childPosA)+posA;\n"
+	"			float4 newOrnA = qtMul(ornA,childOrnA);\n"
+	"			posA = newPosA;\n"
+	"			ornA = newOrnA;\n"
+	"		} else\n"
+	"		{\n"
+	"			collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
+	"		}\n"
+	"		\n"
+	"		if (childShapeIndexB>=0)\n"
+	"		{\n"
+	"			collidableIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;\n"
+	"			float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;\n"
+	"			float4 childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;\n"
+	"			float4 newPosB = transform(&childPosB,&posB,&ornB);\n"
+	"			float4 newOrnB = qtMul(ornB,childOrnB);\n"
+	"			posB = newPosB;\n"
+	"			ornB = newOrnB;\n"
+	"		} else\n"
+	"		{\n"
+	"			collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;	\n"
+	"		}\n"
+	"	\n"
+	"		int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
+	"		int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
+	"	\n"
+	"		int shapeTypeA = collidables[collidableIndexA].m_shapeType;\n"
+	"		int shapeTypeB = collidables[collidableIndexB].m_shapeType;\n"
+	"		int pairIndex = i;\n"
+	"		if ((shapeTypeA == SHAPE_PLANE) && (shapeTypeB==SHAPE_CONVEX_HULL))\n"
+	"		{\n"
+	"			computeContactPlaneConvex( pairIndex, bodyIndexA,bodyIndexB,  collidableIndexA,collidableIndexB, \n"
+	"																rigidBodies,collidables,convexShapes,vertices,indices,\n"
+	"																faces,	globalContactsOut, nGlobalContactsOut,maxContactCapacity,posB,ornB);\n"
+	"			return;\n"
+	"		}\n"
+	"		if ((shapeTypeA == SHAPE_CONVEX_HULL) && (shapeTypeB==SHAPE_PLANE))\n"
+	"		{\n"
+	"			computeContactPlaneConvex( pairIndex, bodyIndexB,bodyIndexA,  collidableIndexB,collidableIndexA, \n"
+	"																rigidBodies,collidables,convexShapes,vertices,indices,\n"
+	"																faces,	globalContactsOut, nGlobalContactsOut,maxContactCapacity,posA,ornA);\n"
+	"			return;\n"
+	"		}\n"
+	"		if ((shapeTypeA == SHAPE_CONVEX_HULL) && (shapeTypeB == SHAPE_SPHERE))\n"
+	"		{\n"
+	"			float4 spherePos = rigidBodies[bodyIndexB].m_pos;\n"
+	"			float sphereRadius = collidables[collidableIndexB].m_radius;\n"
+	"			float4 convexPos = posA;\n"
+	"			float4 convexOrn = ornA;\n"
+	"			\n"
+	"			computeContactSphereConvex(pairIndex, bodyIndexB, bodyIndexA , collidableIndexB,collidableIndexA, \n"
+	"										rigidBodies,collidables,convexShapes,vertices,indices,faces, globalContactsOut, nGlobalContactsOut,maxContactCapacity,\n"
+	"										spherePos,sphereRadius,convexPos,convexOrn);\n"
+	"	\n"
+	"			return;\n"
+	"		}\n"
+	"		if ((shapeTypeA == SHAPE_SPHERE) && (shapeTypeB == SHAPE_CONVEX_HULL))\n"
+	"		{\n"
+	"			float4 spherePos = rigidBodies[bodyIndexA].m_pos;\n"
+	"			float sphereRadius = collidables[collidableIndexA].m_radius;\n"
+	"			float4 convexPos = posB;\n"
+	"			float4 convexOrn = ornB;\n"
+	"			\n"
+	"			computeContactSphereConvex(pairIndex, bodyIndexA, bodyIndexB, collidableIndexA, collidableIndexB, \n"
+	"										rigidBodies,collidables,convexShapes,vertices,indices,faces, globalContactsOut, nGlobalContactsOut,maxContactCapacity,\n"
+	"										spherePos,sphereRadius,convexPos,convexOrn);\n"
+	"	\n"
+	"			return;\n"
+	"		}\n"
+	"	}//	if (i<numCompoundPairs)\n"
+	"}\n"
+	"bool pointInTriangle(const float4* vertices, const float4* normal, float4 *p )\n"
+	"{\n"
+	"	const float4* p1 = &vertices[0];\n"
+	"	const float4* p2 = &vertices[1];\n"
+	"	const float4* p3 = &vertices[2];\n"
+	"	float4 edge1;	edge1 = (*p2 - *p1);\n"
+	"	float4 edge2;	edge2 = ( *p3 - *p2 );\n"
+	"	float4 edge3;	edge3 = ( *p1 - *p3 );\n"
+	"	\n"
+	"	float4 p1_to_p; p1_to_p = ( *p - *p1 );\n"
+	"	float4 p2_to_p; p2_to_p = ( *p - *p2 );\n"
+	"	float4 p3_to_p; p3_to_p = ( *p - *p3 );\n"
+	"	float4 edge1_normal; edge1_normal = ( cross(edge1,*normal));\n"
+	"	float4 edge2_normal; edge2_normal = ( cross(edge2,*normal));\n"
+	"	float4 edge3_normal; edge3_normal = ( cross(edge3,*normal));\n"
+	"	\n"
+	"	\n"
+	"	float r1, r2, r3;\n"
+	"	r1 = dot(edge1_normal,p1_to_p );\n"
+	"	r2 = dot(edge2_normal,p2_to_p );\n"
+	"	r3 = dot(edge3_normal,p3_to_p );\n"
+	"	\n"
+	"	if ( r1 > 0 && r2 > 0 && r3 > 0 )\n"
+	"		return true;\n"
+	"    if ( r1 <= 0 && r2 <= 0 && r3 <= 0 ) \n"
+	"		return true;\n"
+	"	return false;\n"
+	"}\n"
+	"float segmentSqrDistance(float4 from, float4 to,float4 p, float4* nearest) \n"
+	"{\n"
+	"	float4 diff = p - from;\n"
+	"	float4 v = to - from;\n"
+	"	float t = dot(v,diff);\n"
+	"	\n"
+	"	if (t > 0) \n"
+	"	{\n"
+	"		float dotVV = dot(v,v);\n"
+	"		if (t < dotVV) \n"
+	"		{\n"
+	"			t /= dotVV;\n"
+	"			diff -= t*v;\n"
+	"		} else \n"
+	"		{\n"
+	"			t = 1;\n"
+	"			diff -= v;\n"
+	"		}\n"
+	"	} else\n"
+	"	{\n"
+	"		t = 0;\n"
+	"	}\n"
+	"	*nearest = from + t*v;\n"
+	"	return dot(diff,diff);	\n"
+	"}\n"
+	"void	computeContactSphereTriangle(int pairIndex,\n"
+	"									int bodyIndexA, int bodyIndexB,\n"
+	"									int collidableIndexA, int collidableIndexB, \n"
+	"									__global const BodyData* rigidBodies, \n"
+	"									__global const btCollidableGpu* collidables,\n"
+	"									const float4* triangleVertices,\n"
+	"									__global struct b3Contact4Data* restrict globalContactsOut,\n"
+	"									counter32_t nGlobalContactsOut,\n"
+	"									int maxContactCapacity,\n"
+	"									float4 spherePos2,\n"
+	"									float radius,\n"
+	"									float4 pos,\n"
+	"									float4 quat,\n"
+	"									int faceIndex\n"
+	"									)\n"
+	"{\n"
+	"	float4 invPos;\n"
+	"	float4 invOrn;\n"
+	"	trInverse(pos,quat, &invPos,&invOrn);\n"
+	"	float4 spherePos = transform(&spherePos2,&invPos,&invOrn);\n"
+	"	int numFaces = 3;\n"
+	"	float4 closestPnt = (float4)(0, 0, 0, 0);\n"
+	"	float4 hitNormalWorld = (float4)(0, 0, 0, 0);\n"
+	"	float minDist = -1000000.f;\n"
+	"	bool bCollide = false;\n"
+	"	\n"
+	"	//////////////////////////////////////\n"
+	"	float4 sphereCenter;\n"
+	"	sphereCenter = spherePos;\n"
+	"	const float4* vertices = triangleVertices;\n"
+	"	float contactBreakingThreshold = 0.f;//todo?\n"
+	"	float radiusWithThreshold = radius + contactBreakingThreshold;\n"
+	"	float4 edge10;\n"
+	"	edge10 = vertices[1]-vertices[0];\n"
+	"	edge10.w = 0.f;//is this needed?\n"
+	"	float4 edge20;\n"
+	"	edge20 = vertices[2]-vertices[0];\n"
+	"	edge20.w = 0.f;//is this needed?\n"
+	"	float4 normal = cross3(edge10,edge20);\n"
+	"	normal = normalize(normal);\n"
+	"	float4 p1ToCenter;\n"
+	"	p1ToCenter = sphereCenter - vertices[0];\n"
+	"	\n"
+	"	float distanceFromPlane = dot(p1ToCenter,normal);\n"
+	"	if (distanceFromPlane < 0.f)\n"
+	"	{\n"
+	"		//triangle facing the other way\n"
+	"		distanceFromPlane *= -1.f;\n"
+	"		normal *= -1.f;\n"
+	"	}\n"
+	"	hitNormalWorld = normal;\n"
+	"	bool isInsideContactPlane = distanceFromPlane < radiusWithThreshold;\n"
+	"	\n"
+	"	// Check for contact / intersection\n"
+	"	bool hasContact = false;\n"
+	"	float4 contactPoint;\n"
+	"	if (isInsideContactPlane) \n"
+	"	{\n"
+	"	\n"
+	"		if (pointInTriangle(vertices,&normal, &sphereCenter)) \n"
+	"		{\n"
+	"			// Inside the contact wedge - touches a point on the shell plane\n"
+	"			hasContact = true;\n"
+	"			contactPoint = sphereCenter - normal*distanceFromPlane;\n"
+	"			\n"
+	"		} else {\n"
+	"			// Could be inside one of the contact capsules\n"
+	"			float contactCapsuleRadiusSqr = radiusWithThreshold*radiusWithThreshold;\n"
+	"			float4 nearestOnEdge;\n"
+	"			int numEdges = 3;\n"
+	"			for (int i = 0; i < numEdges; i++) \n"
+	"			{\n"
+	"				float4 pa =vertices[i];\n"
+	"				float4 pb = vertices[(i+1)%3];\n"
+	"				float distanceSqr = segmentSqrDistance(pa,pb,sphereCenter, &nearestOnEdge);\n"
+	"				if (distanceSqr < contactCapsuleRadiusSqr) \n"
+	"				{\n"
+	"					// Yep, we're inside a capsule\n"
+	"					hasContact = true;\n"
+	"					contactPoint = nearestOnEdge;\n"
+	"					\n"
+	"				}\n"
+	"				\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	if (hasContact) \n"
+	"	{\n"
+	"		closestPnt = contactPoint;\n"
+	"		float4 contactToCenter = sphereCenter - contactPoint;\n"
+	"		minDist = length(contactToCenter);\n"
+	"		if (minDist>FLT_EPSILON)\n"
+	"		{\n"
+	"			hitNormalWorld = normalize(contactToCenter);//*(1./minDist);\n"
+	"			bCollide  = true;\n"
+	"		}\n"
+	"		\n"
+	"	}\n"
+	"	/////////////////////////////////////\n"
+	"	if (bCollide && minDist > -10000)\n"
+	"	{\n"
+	"		\n"
+	"		float4 normalOnSurfaceB1 = qtRotate(quat,-hitNormalWorld);\n"
+	"		float4 pOnB1 = transform(&closestPnt,&pos,&quat);\n"
+	"		float actualDepth = minDist-radius;\n"
+	"		\n"
+	"		if (actualDepth<=0.f)\n"
+	"		{\n"
+	"			pOnB1.w = actualDepth;\n"
+	"			int dstIdx;\n"
+	"			\n"
+	"			float lenSqr = dot3F4(normalOnSurfaceB1,normalOnSurfaceB1);\n"
+	"			if (lenSqr>FLT_EPSILON)\n"
+	"			{\n"
+	"				AppendInc( nGlobalContactsOut, dstIdx );\n"
+	"			\n"
+	"				if (dstIdx < maxContactCapacity)\n"
+	"				{\n"
+	"					__global struct b3Contact4Data* c = &globalContactsOut[dstIdx];\n"
+	"					c->m_worldNormalOnB = -normalOnSurfaceB1;\n"
+	"					c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
+	"					c->m_batchIdx = pairIndex;\n"
+	"					c->m_bodyAPtrAndSignBit = rigidBodies[bodyIndexA].m_invMass==0?-bodyIndexA:bodyIndexA;\n"
+	"					c->m_bodyBPtrAndSignBit = rigidBodies[bodyIndexB].m_invMass==0?-bodyIndexB:bodyIndexB;\n"
+	"					c->m_worldPosB[0] = pOnB1;\n"
+	"					c->m_childIndexA = -1;\n"
+	"					c->m_childIndexB = faceIndex;\n"
+	"					GET_NPOINTS(*c) = 1;\n"
+	"				} \n"
+	"			}\n"
+	"		}\n"
+	"	}//if (hasCollision)\n"
+	"}\n"
+	"// work-in-progress\n"
+	"__kernel void   findConcaveSphereContactsKernel( __global int4* concavePairs,\n"
+	"												__global const BodyData* rigidBodies,\n"
+	"												__global const btCollidableGpu* collidables,\n"
+	"												__global const ConvexPolyhedronCL* convexShapes, \n"
+	"												__global const float4* vertices,\n"
+	"												__global const float4* uniqueEdges,\n"
+	"												__global const btGpuFace* faces,\n"
+	"												__global const int* indices,\n"
+	"												__global btAabbCL* aabbs,\n"
+	"												__global struct b3Contact4Data* restrict globalContactsOut,\n"
+	"												counter32_t nGlobalContactsOut,\n"
+	"													int numConcavePairs, int maxContactCapacity\n"
+	"												)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	if (i>=numConcavePairs)\n"
+	"		return;\n"
+	"	int pairIdx = i;\n"
+	"	int bodyIndexA = concavePairs[i].x;\n"
+	"	int bodyIndexB = concavePairs[i].y;\n"
+	"	int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
+	"	int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
+	"	int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
+	"	int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
+	"	if (collidables[collidableIndexB].m_shapeType==SHAPE_SPHERE)\n"
+	"	{\n"
+	"		int f = concavePairs[i].z;\n"
+	"		btGpuFace face = faces[convexShapes[shapeIndexA].m_faceOffset+f];\n"
+	"		\n"
+	"		float4 verticesA[3];\n"
+	"		for (int i=0;i<3;i++)\n"
+	"		{\n"
+	"			int index = indices[face.m_indexOffset+i];\n"
+	"			float4 vert = vertices[convexShapes[shapeIndexA].m_vertexOffset+index];\n"
+	"			verticesA[i] = vert;\n"
+	"		}\n"
+	"		float4 spherePos = rigidBodies[bodyIndexB].m_pos;\n"
+	"		float sphereRadius = collidables[collidableIndexB].m_radius;\n"
+	"		float4 convexPos = rigidBodies[bodyIndexA].m_pos;\n"
+	"		float4 convexOrn = rigidBodies[bodyIndexA].m_quat;\n"
+	"		computeContactSphereTriangle(i, bodyIndexB, bodyIndexA, collidableIndexB, collidableIndexA, \n"
+	"																rigidBodies,collidables,\n"
+	"																verticesA,\n"
+	"																globalContactsOut, nGlobalContactsOut,maxContactCapacity,\n"
+	"																spherePos,sphereRadius,convexPos,convexOrn, f);\n"
+	"		return;\n"
+	"	}\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/satClipHullContacts.h b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/satClipHullContacts.h
index f0ecfc78511..907809d8bd3 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/satClipHullContacts.h
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/satClipHullContacts.h
@@ -1,2099 +1,2098 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* satClipKernelsCL= \
-"#define TRIANGLE_NUM_CONVEX_FACES 5\n"
-"#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_extended_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics : enable\n"
-"#ifdef cl_ext_atomic_counters_32\n"
-"#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
-"#else\n"
-"#define counter32_t volatile __global int*\n"
-"#endif\n"
-"#define GET_GROUP_IDX get_group_id(0)\n"
-"#define GET_LOCAL_IDX get_local_id(0)\n"
-"#define GET_GLOBAL_IDX get_global_id(0)\n"
-"#define GET_GROUP_SIZE get_local_size(0)\n"
-"#define GET_NUM_GROUPS get_num_groups(0)\n"
-"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
-"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
-"#define AtomInc(x) atom_inc(&(x))\n"
-"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
-"#define AppendInc(x, out) out = atomic_inc(x)\n"
-"#define AtomAdd(x, value) atom_add(&(x), value)\n"
-"#define AtomCmpxhg(x, cmp, value) atom_cmpxchg( &(x), cmp, value )\n"
-"#define AtomXhg(x, value) atom_xchg ( &(x), value )\n"
-"#define max2 max\n"
-"#define min2 min\n"
-"typedef unsigned int u32;\n"
-"#ifndef B3_CONTACT4DATA_H\n"
-"#define B3_CONTACT4DATA_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#define B3_FLOAT4_H\n"
-"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
-"#define B3_PLATFORM_DEFINITIONS_H\n"
-"struct MyTest\n"
-"{\n"
-"	int bla;\n"
-"};\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
-"#define B3_LARGE_FLOAT 1e18f\n"
-"#define B3_INFINITY 1e18f\n"
-"#define b3Assert(a)\n"
-"#define b3ConstArray(a) __global const a*\n"
-"#define b3AtomicInc atomic_inc\n"
-"#define b3AtomicAdd atomic_add\n"
-"#define b3Fabs fabs\n"
-"#define b3Sqrt native_sqrt\n"
-"#define b3Sin native_sin\n"
-"#define b3Cos native_cos\n"
-"#define B3_STATIC\n"
-"#endif\n"
-"#endif\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"	typedef float4	b3Float4;\n"
-"	#define b3Float4ConstArg const b3Float4\n"
-"	#define b3MakeFloat4 (float4)\n"
-"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return dot(a1, b1);\n"
-"	}\n"
-"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return cross(a1, b1);\n"
-"	}\n"
-"	#define b3MinFloat4 min\n"
-"	#define b3MaxFloat4 max\n"
-"	#define b3Normalized(a) normalize(a)\n"
-"#endif \n"
-"		\n"
-"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
-"{\n"
-"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
-"		return false;\n"
-"	return true;\n"
-"}\n"
-"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
-"{\n"
-"    float maxDot = -B3_INFINITY;\n"
-"    int i = 0;\n"
-"    int ptIndex = -1;\n"
-"    for( i = 0; i < vecLen; i++ )\n"
-"    {\n"
-"        float dot = b3Dot3F4(vecArray[i],vec);\n"
-"            \n"
-"        if( dot > maxDot )\n"
-"        {\n"
-"            maxDot = dot;\n"
-"            ptIndex = i;\n"
-"        }\n"
-"    }\n"
-"	b3Assert(ptIndex>=0);\n"
-"    if (ptIndex<0)\n"
-"	{\n"
-"		ptIndex = 0;\n"
-"	}\n"
-"    *dotOut = maxDot;\n"
-"    return ptIndex;\n"
-"}\n"
-"#endif //B3_FLOAT4_H\n"
-"typedef  struct b3Contact4Data b3Contact4Data_t;\n"
-"struct b3Contact4Data\n"
-"{\n"
-"	b3Float4	m_worldPosB[4];\n"
-"//	b3Float4	m_localPosA[4];\n"
-"//	b3Float4	m_localPosB[4];\n"
-"	b3Float4	m_worldNormalOnB;	//	w: m_nPoints\n"
-"	unsigned short  m_restituitionCoeffCmp;\n"
-"	unsigned short  m_frictionCoeffCmp;\n"
-"	int m_batchIdx;\n"
-"	int m_bodyAPtrAndSignBit;//x:m_bodyAPtr, y:m_bodyBPtr\n"
-"	int m_bodyBPtrAndSignBit;\n"
-"	int	m_childIndexA;\n"
-"	int	m_childIndexB;\n"
-"	int m_unused1;\n"
-"	int m_unused2;\n"
-"};\n"
-"inline int b3Contact4Data_getNumPoints(const struct b3Contact4Data* contact)\n"
-"{\n"
-"	return (int)contact->m_worldNormalOnB.w;\n"
-"};\n"
-"inline void b3Contact4Data_setNumPoints(struct b3Contact4Data* contact, int numPoints)\n"
-"{\n"
-"	contact->m_worldNormalOnB.w = (float)numPoints;\n"
-"};\n"
-"#endif //B3_CONTACT4DATA_H\n"
-"#ifndef B3_CONVEX_POLYHEDRON_DATA_H\n"
-"#define B3_CONVEX_POLYHEDRON_DATA_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_FLOAT4_H\n"
-"#ifndef B3_QUAT_H\n"
-"#define B3_QUAT_H\n"
-"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif\n"
-"#endif\n"
-"#ifndef B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"	typedef float4	b3Quat;\n"
-"	#define b3QuatConstArg const b3Quat\n"
-"	\n"
-"	\n"
-"inline float4 b3FastNormalize4(float4 v)\n"
-"{\n"
-"	v = (float4)(v.xyz,0.f);\n"
-"	return fast_normalize(v);\n"
-"}\n"
-"	\n"
-"inline b3Quat b3QuatMul(b3Quat a, b3Quat b);\n"
-"inline b3Quat b3QuatNormalized(b3QuatConstArg in);\n"
-"inline b3Quat b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec);\n"
-"inline b3Quat b3QuatInvert(b3QuatConstArg q);\n"
-"inline b3Quat b3QuatInverse(b3QuatConstArg q);\n"
-"inline b3Quat b3QuatMul(b3QuatConstArg a, b3QuatConstArg b)\n"
-"{\n"
-"	b3Quat ans;\n"
-"	ans = b3Cross3( a, b );\n"
-"	ans += a.w*b+b.w*a;\n"
-"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
-"	ans.w = a.w*b.w - b3Dot3F4(a, b);\n"
-"	return ans;\n"
-"}\n"
-"inline b3Quat b3QuatNormalized(b3QuatConstArg in)\n"
-"{\n"
-"	b3Quat q;\n"
-"	q=in;\n"
-"	//return b3FastNormalize4(in);\n"
-"	float len = native_sqrt(dot(q, q));\n"
-"	if(len > 0.f)\n"
-"	{\n"
-"		q *= 1.f / len;\n"
-"	}\n"
-"	else\n"
-"	{\n"
-"		q.x = q.y = q.z = 0.f;\n"
-"		q.w = 1.f;\n"
-"	}\n"
-"	return q;\n"
-"}\n"
-"inline float4 b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
-"{\n"
-"	b3Quat qInv = b3QuatInvert( q );\n"
-"	float4 vcpy = vec;\n"
-"	vcpy.w = 0.f;\n"
-"	float4 out = b3QuatMul(b3QuatMul(q,vcpy),qInv);\n"
-"	return out;\n"
-"}\n"
-"inline b3Quat b3QuatInverse(b3QuatConstArg q)\n"
-"{\n"
-"	return (b3Quat)(-q.xyz, q.w);\n"
-"}\n"
-"inline b3Quat b3QuatInvert(b3QuatConstArg q)\n"
-"{\n"
-"	return (b3Quat)(-q.xyz, q.w);\n"
-"}\n"
-"inline float4 b3QuatInvRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
-"{\n"
-"	return b3QuatRotate( b3QuatInvert( q ), vec );\n"
-"}\n"
-"inline b3Float4 b3TransformPoint(b3Float4ConstArg point, b3Float4ConstArg translation, b3QuatConstArg  orientation)\n"
-"{\n"
-"	return b3QuatRotate( orientation, point ) + (translation);\n"
-"}\n"
-"	\n"
-"#endif \n"
-"#endif //B3_QUAT_H\n"
-"typedef struct b3GpuFace b3GpuFace_t;\n"
-"struct b3GpuFace\n"
-"{\n"
-"	b3Float4 m_plane;\n"
-"	int m_indexOffset;\n"
-"	int m_numIndices;\n"
-"	int m_unusedPadding1;\n"
-"	int m_unusedPadding2;\n"
-"};\n"
-"typedef struct b3ConvexPolyhedronData b3ConvexPolyhedronData_t;\n"
-"struct b3ConvexPolyhedronData\n"
-"{\n"
-"	b3Float4		m_localCenter;\n"
-"	b3Float4		m_extents;\n"
-"	b3Float4		mC;\n"
-"	b3Float4		mE;\n"
-"	float			m_radius;\n"
-"	int	m_faceOffset;\n"
-"	int m_numFaces;\n"
-"	int	m_numVertices;\n"
-"	int m_vertexOffset;\n"
-"	int	m_uniqueEdgesOffset;\n"
-"	int	m_numUniqueEdges;\n"
-"	int m_unused;\n"
-"};\n"
-"#endif //B3_CONVEX_POLYHEDRON_DATA_H\n"
-"#ifndef B3_COLLIDABLE_H\n"
-"#define B3_COLLIDABLE_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_FLOAT4_H\n"
-"#ifndef B3_QUAT_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_QUAT_H\n"
-"enum b3ShapeTypes\n"
-"{\n"
-"	SHAPE_HEIGHT_FIELD=1,\n"
-"	SHAPE_CONVEX_HULL=3,\n"
-"	SHAPE_PLANE=4,\n"
-"	SHAPE_CONCAVE_TRIMESH=5,\n"
-"	SHAPE_COMPOUND_OF_CONVEX_HULLS=6,\n"
-"	SHAPE_SPHERE=7,\n"
-"	MAX_NUM_SHAPE_TYPES,\n"
-"};\n"
-"typedef struct b3Collidable b3Collidable_t;\n"
-"struct b3Collidable\n"
-"{\n"
-"	union {\n"
-"		int m_numChildShapes;\n"
-"		int m_bvhIndex;\n"
-"	};\n"
-"	union\n"
-"	{\n"
-"		float m_radius;\n"
-"		int	m_compoundBvhIndex;\n"
-"	};\n"
-"	int m_shapeType;\n"
-"	int m_shapeIndex;\n"
-"};\n"
-"typedef struct b3GpuChildShape b3GpuChildShape_t;\n"
-"struct b3GpuChildShape\n"
-"{\n"
-"	b3Float4	m_childPosition;\n"
-"	b3Quat		m_childOrientation;\n"
-"	int m_shapeIndex;\n"
-"	int m_unused0;\n"
-"	int m_unused1;\n"
-"	int m_unused2;\n"
-"};\n"
-"struct b3CompoundOverlappingPair\n"
-"{\n"
-"	int m_bodyIndexA;\n"
-"	int m_bodyIndexB;\n"
-"//	int	m_pairType;\n"
-"	int m_childShapeIndexA;\n"
-"	int m_childShapeIndexB;\n"
-"};\n"
-"#endif //B3_COLLIDABLE_H\n"
-"#ifndef B3_RIGIDBODY_DATA_H\n"
-"#define B3_RIGIDBODY_DATA_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_FLOAT4_H\n"
-"#ifndef B3_QUAT_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_QUAT_H\n"
-"#ifndef B3_MAT3x3_H\n"
-"#define B3_MAT3x3_H\n"
-"#ifndef B3_QUAT_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_QUAT_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"typedef struct\n"
-"{\n"
-"	b3Float4 m_row[3];\n"
-"}b3Mat3x3;\n"
-"#define b3Mat3x3ConstArg const b3Mat3x3\n"
-"#define b3GetRow(m,row) (m.m_row[row])\n"
-"inline b3Mat3x3 b3QuatGetRotationMatrix(b3Quat quat)\n"
-"{\n"
-"	b3Float4 quat2 = (b3Float4)(quat.x*quat.x, quat.y*quat.y, quat.z*quat.z, 0.f);\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0].x=1-2*quat2.y-2*quat2.z;\n"
-"	out.m_row[0].y=2*quat.x*quat.y-2*quat.w*quat.z;\n"
-"	out.m_row[0].z=2*quat.x*quat.z+2*quat.w*quat.y;\n"
-"	out.m_row[0].w = 0.f;\n"
-"	out.m_row[1].x=2*quat.x*quat.y+2*quat.w*quat.z;\n"
-"	out.m_row[1].y=1-2*quat2.x-2*quat2.z;\n"
-"	out.m_row[1].z=2*quat.y*quat.z-2*quat.w*quat.x;\n"
-"	out.m_row[1].w = 0.f;\n"
-"	out.m_row[2].x=2*quat.x*quat.z-2*quat.w*quat.y;\n"
-"	out.m_row[2].y=2*quat.y*quat.z+2*quat.w*quat.x;\n"
-"	out.m_row[2].z=1-2*quat2.x-2*quat2.y;\n"
-"	out.m_row[2].w = 0.f;\n"
-"	return out;\n"
-"}\n"
-"inline b3Mat3x3 b3AbsoluteMat3x3(b3Mat3x3ConstArg matIn)\n"
-"{\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0] = fabs(matIn.m_row[0]);\n"
-"	out.m_row[1] = fabs(matIn.m_row[1]);\n"
-"	out.m_row[2] = fabs(matIn.m_row[2]);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtZero();\n"
-"__inline\n"
-"b3Mat3x3 mtIdentity();\n"
-"__inline\n"
-"b3Mat3x3 mtTranspose(b3Mat3x3 m);\n"
-"__inline\n"
-"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b);\n"
-"__inline\n"
-"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b);\n"
-"__inline\n"
-"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b);\n"
-"__inline\n"
-"b3Mat3x3 mtZero()\n"
-"{\n"
-"	b3Mat3x3 m;\n"
-"	m.m_row[0] = (b3Float4)(0.f);\n"
-"	m.m_row[1] = (b3Float4)(0.f);\n"
-"	m.m_row[2] = (b3Float4)(0.f);\n"
-"	return m;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtIdentity()\n"
-"{\n"
-"	b3Mat3x3 m;\n"
-"	m.m_row[0] = (b3Float4)(1,0,0,0);\n"
-"	m.m_row[1] = (b3Float4)(0,1,0,0);\n"
-"	m.m_row[2] = (b3Float4)(0,0,1,0);\n"
-"	return m;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtTranspose(b3Mat3x3 m)\n"
-"{\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0] = (b3Float4)(m.m_row[0].x, m.m_row[1].x, m.m_row[2].x, 0.f);\n"
-"	out.m_row[1] = (b3Float4)(m.m_row[0].y, m.m_row[1].y, m.m_row[2].y, 0.f);\n"
-"	out.m_row[2] = (b3Float4)(m.m_row[0].z, m.m_row[1].z, m.m_row[2].z, 0.f);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b)\n"
-"{\n"
-"	b3Mat3x3 transB;\n"
-"	transB = mtTranspose( b );\n"
-"	b3Mat3x3 ans;\n"
-"	//	why this doesn't run when 0ing in the for{}\n"
-"	a.m_row[0].w = 0.f;\n"
-"	a.m_row[1].w = 0.f;\n"
-"	a.m_row[2].w = 0.f;\n"
-"	for(int i=0; i<3; i++)\n"
-"	{\n"
-"//	a.m_row[i].w = 0.f;\n"
-"		ans.m_row[i].x = b3Dot3F4(a.m_row[i],transB.m_row[0]);\n"
-"		ans.m_row[i].y = b3Dot3F4(a.m_row[i],transB.m_row[1]);\n"
-"		ans.m_row[i].z = b3Dot3F4(a.m_row[i],transB.m_row[2]);\n"
-"		ans.m_row[i].w = 0.f;\n"
-"	}\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b)\n"
-"{\n"
-"	b3Float4 ans;\n"
-"	ans.x = b3Dot3F4( a.m_row[0], b );\n"
-"	ans.y = b3Dot3F4( a.m_row[1], b );\n"
-"	ans.z = b3Dot3F4( a.m_row[2], b );\n"
-"	ans.w = 0.f;\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b)\n"
-"{\n"
-"	b3Float4 colx = b3MakeFloat4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
-"	b3Float4 coly = b3MakeFloat4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
-"	b3Float4 colz = b3MakeFloat4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
-"	b3Float4 ans;\n"
-"	ans.x = b3Dot3F4( a, colx );\n"
-"	ans.y = b3Dot3F4( a, coly );\n"
-"	ans.z = b3Dot3F4( a, colz );\n"
-"	return ans;\n"
-"}\n"
-"#endif\n"
-"#endif //B3_MAT3x3_H\n"
-"typedef struct b3RigidBodyData b3RigidBodyData_t;\n"
-"struct b3RigidBodyData\n"
-"{\n"
-"	b3Float4				m_pos;\n"
-"	b3Quat					m_quat;\n"
-"	b3Float4				m_linVel;\n"
-"	b3Float4				m_angVel;\n"
-"	int 					m_collidableIdx;\n"
-"	float 				m_invMass;\n"
-"	float 				m_restituitionCoeff;\n"
-"	float 				m_frictionCoeff;\n"
-"};\n"
-"typedef struct b3InertiaData b3InertiaData_t;\n"
-"struct b3InertiaData\n"
-"{\n"
-"	b3Mat3x3 m_invInertiaWorld;\n"
-"	b3Mat3x3 m_initInvInertia;\n"
-"};\n"
-"#endif //B3_RIGIDBODY_DATA_H\n"
-"	\n"
-"#define GET_NPOINTS(x) (x).m_worldNormalOnB.w\n"
-"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
-"#define make_float4 (float4)\n"
-"#define make_float2 (float2)\n"
-"#define make_uint4 (uint4)\n"
-"#define make_int4 (int4)\n"
-"#define make_uint2 (uint2)\n"
-"#define make_int2 (int2)\n"
-"__inline\n"
-"float fastDiv(float numerator, float denominator)\n"
-"{\n"
-"	return native_divide(numerator, denominator);	\n"
-"//	return numerator/denominator;	\n"
-"}\n"
-"__inline\n"
-"float4 fastDiv4(float4 numerator, float4 denominator)\n"
-"{\n"
-"	return native_divide(numerator, denominator);	\n"
-"}\n"
-"__inline\n"
-"float4 cross3(float4 a, float4 b)\n"
-"{\n"
-"	return cross(a,b);\n"
-"}\n"
-"//#define dot3F4 dot\n"
-"__inline\n"
-"float dot3F4(float4 a, float4 b)\n"
-"{\n"
-"	float4 a1 = make_float4(a.xyz,0.f);\n"
-"	float4 b1 = make_float4(b.xyz,0.f);\n"
-"	return dot(a1, b1);\n"
-"}\n"
-"__inline\n"
-"float4 fastNormalize4(float4 v)\n"
-"{\n"
-"	return fast_normalize(v);\n"
-"}\n"
-"///////////////////////////////////////\n"
-"//	Quaternion\n"
-"///////////////////////////////////////\n"
-"typedef float4 Quaternion;\n"
-"__inline\n"
-"Quaternion qtMul(Quaternion a, Quaternion b);\n"
-"__inline\n"
-"Quaternion qtNormalize(Quaternion in);\n"
-"__inline\n"
-"float4 qtRotate(Quaternion q, float4 vec);\n"
-"__inline\n"
-"Quaternion qtInvert(Quaternion q);\n"
-"__inline\n"
-"Quaternion qtMul(Quaternion a, Quaternion b)\n"
-"{\n"
-"	Quaternion ans;\n"
-"	ans = cross3( a, b );\n"
-"	ans += a.w*b+b.w*a;\n"
-"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
-"	ans.w = a.w*b.w - dot3F4(a, b);\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"Quaternion qtNormalize(Quaternion in)\n"
-"{\n"
-"	return fastNormalize4(in);\n"
-"//	in /= length( in );\n"
-"//	return in;\n"
-"}\n"
-"__inline\n"
-"float4 qtRotate(Quaternion q, float4 vec)\n"
-"{\n"
-"	Quaternion qInv = qtInvert( q );\n"
-"	float4 vcpy = vec;\n"
-"	vcpy.w = 0.f;\n"
-"	float4 out = qtMul(qtMul(q,vcpy),qInv);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"Quaternion qtInvert(Quaternion q)\n"
-"{\n"
-"	return (Quaternion)(-q.xyz, q.w);\n"
-"}\n"
-"__inline\n"
-"float4 qtInvRotate(const Quaternion q, float4 vec)\n"
-"{\n"
-"	return qtRotate( qtInvert( q ), vec );\n"
-"}\n"
-"__inline\n"
-"float4 transform(const float4* p, const float4* translation, const Quaternion* orientation)\n"
-"{\n"
-"	return qtRotate( *orientation, *p ) + (*translation);\n"
-"}\n"
-"__inline\n"
-"float4 normalize3(const float4 a)\n"
-"{\n"
-"	float4 n = make_float4(a.x, a.y, a.z, 0.f);\n"
-"	return fastNormalize4( n );\n"
-"}\n"
-"__inline float4 lerp3(const float4 a,const float4 b, float  t)\n"
-"{\n"
-"	return make_float4(	a.x + (b.x - a.x) * t,\n"
-"						a.y + (b.y - a.y) * t,\n"
-"						a.z + (b.z - a.z) * t,\n"
-"						0.f);\n"
-"}\n"
-"// Clips a face to the back of a plane, return the number of vertices out, stored in ppVtxOut\n"
-"int clipFaceGlobal(__global const float4* pVtxIn, int numVertsIn, float4 planeNormalWS,float planeEqWS, __global float4* ppVtxOut)\n"
-"{\n"
-"	\n"
-"	int ve;\n"
-"	float ds, de;\n"
-"	int numVertsOut = 0;\n"
-"    //double-check next test\n"
-"    	if (numVertsIn < 2)\n"
-"    		return 0;\n"
-"    \n"
-"	float4 firstVertex=pVtxIn[numVertsIn-1];\n"
-"	float4 endVertex = pVtxIn[0];\n"
-"	\n"
-"	ds = dot3F4(planeNormalWS,firstVertex)+planeEqWS;\n"
-"    \n"
-"	for (ve = 0; ve < numVertsIn; ve++)\n"
-"	{\n"
-"		endVertex=pVtxIn[ve];\n"
-"		de = dot3F4(planeNormalWS,endVertex)+planeEqWS;\n"
-"		if (ds<0)\n"
-"		{\n"
-"			if (de<0)\n"
-"			{\n"
-"				// Start < 0, end < 0, so output endVertex\n"
-"				ppVtxOut[numVertsOut++] = endVertex;\n"
-"			}\n"
-"			else\n"
-"			{\n"
-"				// Start < 0, end >= 0, so output intersection\n"
-"				ppVtxOut[numVertsOut++] = lerp3(firstVertex, endVertex,(ds * 1.f/(ds - de)) );\n"
-"			}\n"
-"		}\n"
-"		else\n"
-"		{\n"
-"			if (de<0)\n"
-"			{\n"
-"				// Start >= 0, end < 0 so output intersection and end\n"
-"				ppVtxOut[numVertsOut++] = lerp3(firstVertex, endVertex,(ds * 1.f/(ds - de)) );\n"
-"				ppVtxOut[numVertsOut++] = endVertex;\n"
-"			}\n"
-"		}\n"
-"		firstVertex = endVertex;\n"
-"		ds = de;\n"
-"	}\n"
-"	return numVertsOut;\n"
-"}\n"
-"// Clips a face to the back of a plane, return the number of vertices out, stored in ppVtxOut\n"
-"int clipFace(const float4* pVtxIn, int numVertsIn, float4 planeNormalWS,float planeEqWS, float4* ppVtxOut)\n"
-"{\n"
-"	\n"
-"	int ve;\n"
-"	float ds, de;\n"
-"	int numVertsOut = 0;\n"
-"//double-check next test\n"
-"	if (numVertsIn < 2)\n"
-"		return 0;\n"
-"	float4 firstVertex=pVtxIn[numVertsIn-1];\n"
-"	float4 endVertex = pVtxIn[0];\n"
-"	\n"
-"	ds = dot3F4(planeNormalWS,firstVertex)+planeEqWS;\n"
-"	for (ve = 0; ve < numVertsIn; ve++)\n"
-"	{\n"
-"		endVertex=pVtxIn[ve];\n"
-"		de = dot3F4(planeNormalWS,endVertex)+planeEqWS;\n"
-"		if (ds<0)\n"
-"		{\n"
-"			if (de<0)\n"
-"			{\n"
-"				// Start < 0, end < 0, so output endVertex\n"
-"				ppVtxOut[numVertsOut++] = endVertex;\n"
-"			}\n"
-"			else\n"
-"			{\n"
-"				// Start < 0, end >= 0, so output intersection\n"
-"				ppVtxOut[numVertsOut++] = lerp3(firstVertex, endVertex,(ds * 1.f/(ds - de)) );\n"
-"			}\n"
-"		}\n"
-"		else\n"
-"		{\n"
-"			if (de<0)\n"
-"			{\n"
-"				// Start >= 0, end < 0 so output intersection and end\n"
-"				ppVtxOut[numVertsOut++] = lerp3(firstVertex, endVertex,(ds * 1.f/(ds - de)) );\n"
-"				ppVtxOut[numVertsOut++] = endVertex;\n"
-"			}\n"
-"		}\n"
-"		firstVertex = endVertex;\n"
-"		ds = de;\n"
-"	}\n"
-"	return numVertsOut;\n"
-"}\n"
-"int clipFaceAgainstHull(const float4 separatingNormal, __global const b3ConvexPolyhedronData_t* hullA,  \n"
-"	const float4 posA, const Quaternion ornA, float4* worldVertsB1, int numWorldVertsB1,\n"
-"	float4* worldVertsB2, int capacityWorldVertsB2,\n"
-"	const float minDist, float maxDist,\n"
-"	__global const float4* vertices,\n"
-"	__global const b3GpuFace_t* faces,\n"
-"	__global const int* indices,\n"
-"	float4* contactsOut,\n"
-"	int contactCapacity)\n"
-"{\n"
-"	int numContactsOut = 0;\n"
-"	float4* pVtxIn = worldVertsB1;\n"
-"	float4* pVtxOut = worldVertsB2;\n"
-"	\n"
-"	int numVertsIn = numWorldVertsB1;\n"
-"	int numVertsOut = 0;\n"
-"	int closestFaceA=-1;\n"
-"	{\n"
-"		float dmin = FLT_MAX;\n"
-"		for(int face=0;face<hullA->m_numFaces;face++)\n"
-"		{\n"
-"			const float4 Normal = make_float4(\n"
-"				faces[hullA->m_faceOffset+face].m_plane.x, \n"
-"				faces[hullA->m_faceOffset+face].m_plane.y, \n"
-"				faces[hullA->m_faceOffset+face].m_plane.z,0.f);\n"
-"			const float4 faceANormalWS = qtRotate(ornA,Normal);\n"
-"		\n"
-"			float d = dot3F4(faceANormalWS,separatingNormal);\n"
-"			if (d < dmin)\n"
-"			{\n"
-"				dmin = d;\n"
-"				closestFaceA = face;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	if (closestFaceA<0)\n"
-"		return numContactsOut;\n"
-"	b3GpuFace_t polyA = faces[hullA->m_faceOffset+closestFaceA];\n"
-"	// clip polygon to back of planes of all faces of hull A that are adjacent to witness face\n"
-"	int numVerticesA = polyA.m_numIndices;\n"
-"	for(int e0=0;e0<numVerticesA;e0++)\n"
-"	{\n"
-"		const float4 a = vertices[hullA->m_vertexOffset+indices[polyA.m_indexOffset+e0]];\n"
-"		const float4 b = vertices[hullA->m_vertexOffset+indices[polyA.m_indexOffset+((e0+1)%numVerticesA)]];\n"
-"		const float4 edge0 = a - b;\n"
-"		const float4 WorldEdge0 = qtRotate(ornA,edge0);\n"
-"		float4 planeNormalA = make_float4(polyA.m_plane.x,polyA.m_plane.y,polyA.m_plane.z,0.f);\n"
-"		float4 worldPlaneAnormal1 = qtRotate(ornA,planeNormalA);\n"
-"		float4 planeNormalWS1 = -cross3(WorldEdge0,worldPlaneAnormal1);\n"
-"		float4 worldA1 = transform(&a,&posA,&ornA);\n"
-"		float planeEqWS1 = -dot3F4(worldA1,planeNormalWS1);\n"
-"		\n"
-"		float4 planeNormalWS = planeNormalWS1;\n"
-"		float planeEqWS=planeEqWS1;\n"
-"		\n"
-"		//clip face\n"
-"		//clipFace(*pVtxIn, *pVtxOut,planeNormalWS,planeEqWS);\n"
-"		numVertsOut = clipFace(pVtxIn, numVertsIn, planeNormalWS,planeEqWS, pVtxOut);\n"
-"		//btSwap(pVtxIn,pVtxOut);\n"
-"		float4* tmp = pVtxOut;\n"
-"		pVtxOut = pVtxIn;\n"
-"		pVtxIn = tmp;\n"
-"		numVertsIn = numVertsOut;\n"
-"		numVertsOut = 0;\n"
-"	}\n"
-"	\n"
-"	// only keep points that are behind the witness face\n"
-"	{\n"
-"		float4 localPlaneNormal  = make_float4(polyA.m_plane.x,polyA.m_plane.y,polyA.m_plane.z,0.f);\n"
-"		float localPlaneEq = polyA.m_plane.w;\n"
-"		float4 planeNormalWS = qtRotate(ornA,localPlaneNormal);\n"
-"		float planeEqWS=localPlaneEq-dot3F4(planeNormalWS,posA);\n"
-"		for (int i=0;i<numVertsIn;i++)\n"
-"		{\n"
-"			float depth = dot3F4(planeNormalWS,pVtxIn[i])+planeEqWS;\n"
-"			if (depth <=minDist)\n"
-"			{\n"
-"				depth = minDist;\n"
-"			}\n"
-"			if (depth <=maxDist)\n"
-"			{\n"
-"				float4 pointInWorld = pVtxIn[i];\n"
-"				//resultOut.addContactPoint(separatingNormal,point,depth);\n"
-"				contactsOut[numContactsOut++] = make_float4(pointInWorld.x,pointInWorld.y,pointInWorld.z,depth);\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	return numContactsOut;\n"
-"}\n"
-"int clipFaceAgainstHullLocalA(const float4 separatingNormal, const b3ConvexPolyhedronData_t* hullA,  \n"
-"	const float4 posA, const Quaternion ornA, float4* worldVertsB1, int numWorldVertsB1,\n"
-"	float4* worldVertsB2, int capacityWorldVertsB2,\n"
-"	const float minDist, float maxDist,\n"
-"	const float4* verticesA,\n"
-"	const b3GpuFace_t* facesA,\n"
-"	const int* indicesA,\n"
-"	__global const float4* verticesB,\n"
-"	__global const b3GpuFace_t* facesB,\n"
-"	__global const int* indicesB,\n"
-"	float4* contactsOut,\n"
-"	int contactCapacity)\n"
-"{\n"
-"	int numContactsOut = 0;\n"
-"	float4* pVtxIn = worldVertsB1;\n"
-"	float4* pVtxOut = worldVertsB2;\n"
-"	\n"
-"	int numVertsIn = numWorldVertsB1;\n"
-"	int numVertsOut = 0;\n"
-"	int closestFaceA=-1;\n"
-"	{\n"
-"		float dmin = FLT_MAX;\n"
-"		for(int face=0;face<hullA->m_numFaces;face++)\n"
-"		{\n"
-"			const float4 Normal = make_float4(\n"
-"				facesA[hullA->m_faceOffset+face].m_plane.x, \n"
-"				facesA[hullA->m_faceOffset+face].m_plane.y, \n"
-"				facesA[hullA->m_faceOffset+face].m_plane.z,0.f);\n"
-"			const float4 faceANormalWS = qtRotate(ornA,Normal);\n"
-"		\n"
-"			float d = dot3F4(faceANormalWS,separatingNormal);\n"
-"			if (d < dmin)\n"
-"			{\n"
-"				dmin = d;\n"
-"				closestFaceA = face;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	if (closestFaceA<0)\n"
-"		return numContactsOut;\n"
-"	b3GpuFace_t polyA = facesA[hullA->m_faceOffset+closestFaceA];\n"
-"	// clip polygon to back of planes of all faces of hull A that are adjacent to witness face\n"
-"	int numVerticesA = polyA.m_numIndices;\n"
-"	for(int e0=0;e0<numVerticesA;e0++)\n"
-"	{\n"
-"		const float4 a = verticesA[hullA->m_vertexOffset+indicesA[polyA.m_indexOffset+e0]];\n"
-"		const float4 b = verticesA[hullA->m_vertexOffset+indicesA[polyA.m_indexOffset+((e0+1)%numVerticesA)]];\n"
-"		const float4 edge0 = a - b;\n"
-"		const float4 WorldEdge0 = qtRotate(ornA,edge0);\n"
-"		float4 planeNormalA = make_float4(polyA.m_plane.x,polyA.m_plane.y,polyA.m_plane.z,0.f);\n"
-"		float4 worldPlaneAnormal1 = qtRotate(ornA,planeNormalA);\n"
-"		float4 planeNormalWS1 = -cross3(WorldEdge0,worldPlaneAnormal1);\n"
-"		float4 worldA1 = transform(&a,&posA,&ornA);\n"
-"		float planeEqWS1 = -dot3F4(worldA1,planeNormalWS1);\n"
-"		\n"
-"		float4 planeNormalWS = planeNormalWS1;\n"
-"		float planeEqWS=planeEqWS1;\n"
-"		\n"
-"		//clip face\n"
-"		//clipFace(*pVtxIn, *pVtxOut,planeNormalWS,planeEqWS);\n"
-"		numVertsOut = clipFace(pVtxIn, numVertsIn, planeNormalWS,planeEqWS, pVtxOut);\n"
-"		//btSwap(pVtxIn,pVtxOut);\n"
-"		float4* tmp = pVtxOut;\n"
-"		pVtxOut = pVtxIn;\n"
-"		pVtxIn = tmp;\n"
-"		numVertsIn = numVertsOut;\n"
-"		numVertsOut = 0;\n"
-"	}\n"
-"	\n"
-"	// only keep points that are behind the witness face\n"
-"	{\n"
-"		float4 localPlaneNormal  = make_float4(polyA.m_plane.x,polyA.m_plane.y,polyA.m_plane.z,0.f);\n"
-"		float localPlaneEq = polyA.m_plane.w;\n"
-"		float4 planeNormalWS = qtRotate(ornA,localPlaneNormal);\n"
-"		float planeEqWS=localPlaneEq-dot3F4(planeNormalWS,posA);\n"
-"		for (int i=0;i<numVertsIn;i++)\n"
-"		{\n"
-"			float depth = dot3F4(planeNormalWS,pVtxIn[i])+planeEqWS;\n"
-"			if (depth <=minDist)\n"
-"			{\n"
-"				depth = minDist;\n"
-"			}\n"
-"			if (depth <=maxDist)\n"
-"			{\n"
-"				float4 pointInWorld = pVtxIn[i];\n"
-"				//resultOut.addContactPoint(separatingNormal,point,depth);\n"
-"				contactsOut[numContactsOut++] = make_float4(pointInWorld.x,pointInWorld.y,pointInWorld.z,depth);\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	return numContactsOut;\n"
-"}\n"
-"int	clipHullAgainstHull(const float4 separatingNormal,\n"
-"	__global const b3ConvexPolyhedronData_t* hullA, __global const b3ConvexPolyhedronData_t* hullB, \n"
-"	const float4 posA, const Quaternion ornA,const float4 posB, const Quaternion ornB, \n"
-"	float4* worldVertsB1, float4* worldVertsB2, int capacityWorldVerts,\n"
-"	const float minDist, float maxDist,\n"
-"	__global const float4* vertices,\n"
-"	__global const b3GpuFace_t* faces,\n"
-"	__global const int* indices,\n"
-"	float4*	localContactsOut,\n"
-"	int localContactCapacity)\n"
-"{\n"
-"	int numContactsOut = 0;\n"
-"	int numWorldVertsB1= 0;\n"
-"	int closestFaceB=-1;\n"
-"	float dmax = -FLT_MAX;\n"
-"	{\n"
-"		for(int face=0;face<hullB->m_numFaces;face++)\n"
-"		{\n"
-"			const float4 Normal = make_float4(faces[hullB->m_faceOffset+face].m_plane.x, \n"
-"				faces[hullB->m_faceOffset+face].m_plane.y, faces[hullB->m_faceOffset+face].m_plane.z,0.f);\n"
-"			const float4 WorldNormal = qtRotate(ornB, Normal);\n"
-"			float d = dot3F4(WorldNormal,separatingNormal);\n"
-"			if (d > dmax)\n"
-"			{\n"
-"				dmax = d;\n"
-"				closestFaceB = face;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	{\n"
-"		const b3GpuFace_t polyB = faces[hullB->m_faceOffset+closestFaceB];\n"
-"		const int numVertices = polyB.m_numIndices;\n"
-"		for(int e0=0;e0<numVertices;e0++)\n"
-"		{\n"
-"			const float4 b = vertices[hullB->m_vertexOffset+indices[polyB.m_indexOffset+e0]];\n"
-"			worldVertsB1[numWorldVertsB1++] = transform(&b,&posB,&ornB);\n"
-"		}\n"
-"	}\n"
-"	if (closestFaceB>=0)\n"
-"	{\n"
-"		numContactsOut = clipFaceAgainstHull(separatingNormal, hullA, \n"
-"				posA,ornA,\n"
-"				worldVertsB1,numWorldVertsB1,worldVertsB2,capacityWorldVerts, minDist, maxDist,vertices,\n"
-"				faces,\n"
-"				indices,localContactsOut,localContactCapacity);\n"
-"	}\n"
-"	return numContactsOut;\n"
-"}\n"
-"int	clipHullAgainstHullLocalA(const float4 separatingNormal,\n"
-"	const b3ConvexPolyhedronData_t* hullA, __global const b3ConvexPolyhedronData_t* hullB, \n"
-"	const float4 posA, const Quaternion ornA,const float4 posB, const Quaternion ornB, \n"
-"	float4* worldVertsB1, float4* worldVertsB2, int capacityWorldVerts,\n"
-"	const float minDist, float maxDist,\n"
-"	const float4* verticesA,\n"
-"	const b3GpuFace_t* facesA,\n"
-"	const int* indicesA,\n"
-"	__global const float4* verticesB,\n"
-"	__global const b3GpuFace_t* facesB,\n"
-"	__global const int* indicesB,\n"
-"	float4*	localContactsOut,\n"
-"	int localContactCapacity)\n"
-"{\n"
-"	int numContactsOut = 0;\n"
-"	int numWorldVertsB1= 0;\n"
-"	int closestFaceB=-1;\n"
-"	float dmax = -FLT_MAX;\n"
-"	{\n"
-"		for(int face=0;face<hullB->m_numFaces;face++)\n"
-"		{\n"
-"			const float4 Normal = make_float4(facesB[hullB->m_faceOffset+face].m_plane.x, \n"
-"				facesB[hullB->m_faceOffset+face].m_plane.y, facesB[hullB->m_faceOffset+face].m_plane.z,0.f);\n"
-"			const float4 WorldNormal = qtRotate(ornB, Normal);\n"
-"			float d = dot3F4(WorldNormal,separatingNormal);\n"
-"			if (d > dmax)\n"
-"			{\n"
-"				dmax = d;\n"
-"				closestFaceB = face;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	{\n"
-"		const b3GpuFace_t polyB = facesB[hullB->m_faceOffset+closestFaceB];\n"
-"		const int numVertices = polyB.m_numIndices;\n"
-"		for(int e0=0;e0<numVertices;e0++)\n"
-"		{\n"
-"			const float4 b = verticesB[hullB->m_vertexOffset+indicesB[polyB.m_indexOffset+e0]];\n"
-"			worldVertsB1[numWorldVertsB1++] = transform(&b,&posB,&ornB);\n"
-"		}\n"
-"	}\n"
-"	if (closestFaceB>=0)\n"
-"	{\n"
-"		numContactsOut = clipFaceAgainstHullLocalA(separatingNormal, hullA, \n"
-"				posA,ornA,\n"
-"				worldVertsB1,numWorldVertsB1,worldVertsB2,capacityWorldVerts, minDist, maxDist,\n"
-"				verticesA,facesA,indicesA,\n"
-"				verticesB,facesB,indicesB,\n"
-"				localContactsOut,localContactCapacity);\n"
-"	}\n"
-"	return numContactsOut;\n"
-"}\n"
-"#define PARALLEL_SUM(v, n) for(int j=1; j<n; j++) v[0] += v[j];\n"
-"#define PARALLEL_DO(execution, n) for(int ie=0; ie<n; ie++){execution;}\n"
-"#define REDUCE_MAX(v, n) {int i=0; for(int offset=0; offset<n; offset++) v[i] = (v[i].y > v[i+offset].y)? v[i]: v[i+offset]; }\n"
-"#define REDUCE_MIN(v, n) {int i=0; for(int offset=0; offset<n; offset++) v[i] = (v[i].y < v[i+offset].y)? v[i]: v[i+offset]; }\n"
-"int extractManifoldSequentialGlobal(__global const float4* p, int nPoints, float4 nearNormal, int4* contactIdx)\n"
-"{\n"
-"	if( nPoints == 0 )\n"
-"        return 0;\n"
-"    \n"
-"    if (nPoints <=4)\n"
-"        return nPoints;\n"
-"    \n"
-"    \n"
-"    if (nPoints >64)\n"
-"        nPoints = 64;\n"
-"    \n"
-"	float4 center = make_float4(0.f);\n"
-"	{\n"
-"		\n"
-"		for (int i=0;i<nPoints;i++)\n"
-"			center += p[i];\n"
-"		center /= (float)nPoints;\n"
-"	}\n"
-"    \n"
-"	\n"
-"    \n"
-"	//	sample 4 directions\n"
-"    \n"
-"    float4 aVector = p[0] - center;\n"
-"    float4 u = cross3( nearNormal, aVector );\n"
-"    float4 v = cross3( nearNormal, u );\n"
-"    u = normalize3( u );\n"
-"    v = normalize3( v );\n"
-"    \n"
-"    \n"
-"    //keep point with deepest penetration\n"
-"    float minW= FLT_MAX;\n"
-"    \n"
-"    int minIndex=-1;\n"
-"    \n"
-"    float4 maxDots;\n"
-"    maxDots.x = FLT_MIN;\n"
-"    maxDots.y = FLT_MIN;\n"
-"    maxDots.z = FLT_MIN;\n"
-"    maxDots.w = FLT_MIN;\n"
-"    \n"
-"    //	idx, distance\n"
-"    for(int ie = 0; ie<nPoints; ie++ )\n"
-"    {\n"
-"        if (p[ie].w<minW)\n"
-"        {\n"
-"            minW = p[ie].w;\n"
-"            minIndex=ie;\n"
-"        }\n"
-"        float f;\n"
-"        float4 r = p[ie]-center;\n"
-"        f = dot3F4( u, r );\n"
-"        if (f<maxDots.x)\n"
-"        {\n"
-"            maxDots.x = f;\n"
-"            contactIdx[0].x = ie;\n"
-"        }\n"
-"        \n"
-"        f = dot3F4( -u, r );\n"
-"        if (f<maxDots.y)\n"
-"        {\n"
-"            maxDots.y = f;\n"
-"            contactIdx[0].y = ie;\n"
-"        }\n"
-"        \n"
-"        \n"
-"        f = dot3F4( v, r );\n"
-"        if (f<maxDots.z)\n"
-"        {\n"
-"            maxDots.z = f;\n"
-"            contactIdx[0].z = ie;\n"
-"        }\n"
-"        \n"
-"        f = dot3F4( -v, r );\n"
-"        if (f<maxDots.w)\n"
-"        {\n"
-"            maxDots.w = f;\n"
-"            contactIdx[0].w = ie;\n"
-"        }\n"
-"        \n"
-"    }\n"
-"    \n"
-"    if (contactIdx[0].x != minIndex && contactIdx[0].y != minIndex && contactIdx[0].z != minIndex && contactIdx[0].w != minIndex)\n"
-"    {\n"
-"        //replace the first contact with minimum (todo: replace contact with least penetration)\n"
-"        contactIdx[0].x = minIndex;\n"
-"    }\n"
-"    \n"
-"    return 4;\n"
-"    \n"
-"}\n"
-"int extractManifoldSequentialGlobalFake(__global const float4* p, int nPoints, float4 nearNormal, int* contactIdx)\n"
-"{\n"
-"    contactIdx[0] = 0;\n"
-"    contactIdx[1] = 1;\n"
-"    contactIdx[2] = 2;\n"
-"    contactIdx[3] = 3;\n"
-"    \n"
-"	if( nPoints == 0 ) return 0;\n"
-"    \n"
-"	nPoints = min2( nPoints, 4 );\n"
-"    return nPoints;\n"
-"    \n"
-"}\n"
-"int extractManifoldSequential(const float4* p, int nPoints, float4 nearNormal, int* contactIdx)\n"
-"{\n"
-"	if( nPoints == 0 ) return 0;\n"
-"	nPoints = min2( nPoints, 64 );\n"
-"	float4 center = make_float4(0.f);\n"
-"	{\n"
-"		float4 v[64];\n"
-"		for (int i=0;i<nPoints;i++)\n"
-"			v[i] = p[i];\n"
-"		//memcpy( v, p, nPoints*sizeof(float4) );\n"
-"		PARALLEL_SUM( v, nPoints );\n"
-"		center = v[0]/(float)nPoints;\n"
-"	}\n"
-"	\n"
-"	{	//	sample 4 directions\n"
-"		if( nPoints < 4 )\n"
-"		{\n"
-"			for(int i=0; i<nPoints; i++) \n"
-"				contactIdx[i] = i;\n"
-"			return nPoints;\n"
-"		}\n"
-"		float4 aVector = p[0] - center;\n"
-"		float4 u = cross3( nearNormal, aVector );\n"
-"		float4 v = cross3( nearNormal, u );\n"
-"		u = normalize3( u );\n"
-"		v = normalize3( v );\n"
-"		int idx[4];\n"
-"		float2 max00 = make_float2(0,FLT_MAX);\n"
-"		{\n"
-"			//	idx, distance\n"
-"			{\n"
-"				{\n"
-"					int4 a[64];\n"
-"					for(int ie = 0; ie<nPoints; ie++ )\n"
-"					{\n"
-"						\n"
-"						\n"
-"						float f;\n"
-"						float4 r = p[ie]-center;\n"
-"						f = dot3F4( u, r );\n"
-"						a[ie].x = ((*(u32*)&f) & 0xffffff00) | (0xff & ie);\n"
-"						f = dot3F4( -u, r );\n"
-"						a[ie].y = ((*(u32*)&f) & 0xffffff00) | (0xff & ie);\n"
-"						f = dot3F4( v, r );\n"
-"						a[ie].z = ((*(u32*)&f) & 0xffffff00) | (0xff & ie);\n"
-"						f = dot3F4( -v, r );\n"
-"						a[ie].w = ((*(u32*)&f) & 0xffffff00) | (0xff & ie);\n"
-"					}\n"
-"					for(int ie=0; ie<nPoints; ie++)\n"
-"					{\n"
-"						a[0].x = (a[0].x > a[ie].x )? a[0].x: a[ie].x;\n"
-"						a[0].y = (a[0].y > a[ie].y )? a[0].y: a[ie].y;\n"
-"						a[0].z = (a[0].z > a[ie].z )? a[0].z: a[ie].z;\n"
-"						a[0].w = (a[0].w > a[ie].w )? a[0].w: a[ie].w;\n"
-"					}\n"
-"					idx[0] = (int)a[0].x & 0xff;\n"
-"					idx[1] = (int)a[0].y & 0xff;\n"
-"					idx[2] = (int)a[0].z & 0xff;\n"
-"					idx[3] = (int)a[0].w & 0xff;\n"
-"				}\n"
-"			}\n"
-"			{\n"
-"				float2 h[64];\n"
-"				PARALLEL_DO( h[ie] = make_float2((float)ie, p[ie].w), nPoints );\n"
-"				REDUCE_MIN( h, nPoints );\n"
-"				max00 = h[0];\n"
-"			}\n"
-"		}\n"
-"		contactIdx[0] = idx[0];\n"
-"		contactIdx[1] = idx[1];\n"
-"		contactIdx[2] = idx[2];\n"
-"		contactIdx[3] = idx[3];\n"
-"		return 4;\n"
-"	}\n"
-"}\n"
-"__kernel void   extractManifoldAndAddContactKernel(__global const int4* pairs, \n"
-"																	__global const b3RigidBodyData_t* rigidBodies, \n"
-"																	__global const float4* closestPointsWorld,\n"
-"																	__global const float4* separatingNormalsWorld,\n"
-"																	__global const int* contactCounts,\n"
-"																	__global const int* contactOffsets,\n"
-"																	__global struct b3Contact4Data* restrict contactsOut,\n"
-"																	counter32_t nContactsOut,\n"
-"																	int contactCapacity,\n"
-"																	int numPairs,\n"
-"																	int pairIndex\n"
-"																	)\n"
-"{\n"
-"	int idx = get_global_id(0);\n"
-"	\n"
-"	if (idx<numPairs)\n"
-"	{\n"
-"		float4 normal = separatingNormalsWorld[idx];\n"
-"		int nPoints = contactCounts[idx];\n"
-"		__global const float4* pointsIn = &closestPointsWorld[contactOffsets[idx]];\n"
-"		float4 localPoints[64];\n"
-"		for (int i=0;i<nPoints;i++)\n"
-"		{\n"
-"			localPoints[i] = pointsIn[i];\n"
-"		}\n"
-"		int contactIdx[4];// = {-1,-1,-1,-1};\n"
-"		contactIdx[0] = -1;\n"
-"		contactIdx[1] = -1;\n"
-"		contactIdx[2] = -1;\n"
-"		contactIdx[3] = -1;\n"
-"		int nContacts = extractManifoldSequential(localPoints, nPoints, normal, contactIdx);\n"
-"		int dstIdx;\n"
-"		AppendInc( nContactsOut, dstIdx );\n"
-"		if (dstIdx<contactCapacity)\n"
-"		{\n"
-"			__global struct b3Contact4Data* c = contactsOut + dstIdx;\n"
-"			c->m_worldNormalOnB = -normal;\n"
-"			c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
-"			c->m_batchIdx = idx;\n"
-"			int bodyA = pairs[pairIndex].x;\n"
-"			int bodyB = pairs[pairIndex].y;\n"
-"			c->m_bodyAPtrAndSignBit = rigidBodies[bodyA].m_invMass==0 ? -bodyA:bodyA;\n"
-"			c->m_bodyBPtrAndSignBit = rigidBodies[bodyB].m_invMass==0 ? -bodyB:bodyB;\n"
-"			c->m_childIndexA = -1;\n"
-"			c->m_childIndexB = -1;\n"
-"			for (int i=0;i<nContacts;i++)\n"
-"			{\n"
-"				c->m_worldPosB[i] = localPoints[contactIdx[i]];\n"
-"			}\n"
-"			GET_NPOINTS(*c) = nContacts;\n"
-"		}\n"
-"	}\n"
-"}\n"
-"void	trInverse(float4 translationIn, Quaternion orientationIn,\n"
-"		float4* translationOut, Quaternion* orientationOut)\n"
-"{\n"
-"	*orientationOut = qtInvert(orientationIn);\n"
-"	*translationOut = qtRotate(*orientationOut, -translationIn);\n"
-"}\n"
-"void	trMul(float4 translationA, Quaternion orientationA,\n"
-"						float4 translationB, Quaternion orientationB,\n"
-"		float4* translationOut, Quaternion* orientationOut)\n"
-"{\n"
-"	*orientationOut = qtMul(orientationA,orientationB);\n"
-"	*translationOut = transform(&translationB,&translationA,&orientationA);\n"
-"}\n"
-"__kernel void   clipHullHullKernel( __global int4* pairs, \n"
-"																					__global const b3RigidBodyData_t* rigidBodies, \n"
-"																					__global const b3Collidable_t* collidables,\n"
-"																					__global const b3ConvexPolyhedronData_t* convexShapes, \n"
-"																					__global const float4* vertices,\n"
-"																					__global const float4* uniqueEdges,\n"
-"																					__global const b3GpuFace_t* faces,\n"
-"																					__global const int* indices,\n"
-"																					__global const float4* separatingNormals,\n"
-"																					__global const int* hasSeparatingAxis,\n"
-"																					__global struct b3Contact4Data* restrict globalContactsOut,\n"
-"																					counter32_t nGlobalContactsOut,\n"
-"																					int numPairs,\n"
-"																					int contactCapacity)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	int pairIndex = i;\n"
-"	\n"
-"	float4 worldVertsB1[64];\n"
-"	float4 worldVertsB2[64];\n"
-"	int capacityWorldVerts = 64;	\n"
-"	float4 localContactsOut[64];\n"
-"	int localContactCapacity=64;\n"
-"	\n"
-"	float minDist = -1e30f;\n"
-"	float maxDist = 0.02f;\n"
-"	if (i<numPairs)\n"
-"	{\n"
-"		int bodyIndexA = pairs[i].x;\n"
-"		int bodyIndexB = pairs[i].y;\n"
-"			\n"
-"		int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
-"		int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
-"		if (hasSeparatingAxis[i])\n"
-"		{\n"
-"			\n"
-"			int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
-"			int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
-"			\n"
-"		\n"
-"			int numLocalContactsOut = clipHullAgainstHull(separatingNormals[i],\n"
-"														&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],\n"
-"														rigidBodies[bodyIndexA].m_pos,rigidBodies[bodyIndexA].m_quat,\n"
-"													  rigidBodies[bodyIndexB].m_pos,rigidBodies[bodyIndexB].m_quat,\n"
-"													  worldVertsB1,worldVertsB2,capacityWorldVerts,\n"
-"														minDist, maxDist,\n"
-"														vertices,faces,indices,\n"
-"														localContactsOut,localContactCapacity);\n"
-"												\n"
-"		if (numLocalContactsOut>0)\n"
-"		{\n"
-"				float4 normal = -separatingNormals[i];\n"
-"				int nPoints = numLocalContactsOut;\n"
-"				float4* pointsIn = localContactsOut;\n"
-"				int contactIdx[4];// = {-1,-1,-1,-1};\n"
-"				contactIdx[0] = -1;\n"
-"				contactIdx[1] = -1;\n"
-"				contactIdx[2] = -1;\n"
-"				contactIdx[3] = -1;\n"
-"		\n"
-"				int nReducedContacts = extractManifoldSequential(pointsIn, nPoints, normal, contactIdx);\n"
-"		\n"
-"				\n"
-"				int mprContactIndex = pairs[pairIndex].z;\n"
-"				int dstIdx = mprContactIndex;\n"
-"				if (dstIdx<0)\n"
-"				{\n"
-"					AppendInc( nGlobalContactsOut, dstIdx );\n"
-"				}\n"
-"				if (dstIdx<contactCapacity)\n"
-"				{\n"
-"					pairs[pairIndex].z = dstIdx;\n"
-"					__global struct b3Contact4Data* c = globalContactsOut+ dstIdx;\n"
-"					c->m_worldNormalOnB = -normal;\n"
-"					c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
-"					c->m_batchIdx = pairIndex;\n"
-"					int bodyA = pairs[pairIndex].x;\n"
-"					int bodyB = pairs[pairIndex].y;\n"
-"					c->m_bodyAPtrAndSignBit = rigidBodies[bodyA].m_invMass==0?-bodyA:bodyA;\n"
-"					c->m_bodyBPtrAndSignBit = rigidBodies[bodyB].m_invMass==0?-bodyB:bodyB;\n"
-"					c->m_childIndexA = -1;\n"
-"					c->m_childIndexB = -1;\n"
-"					for (int i=0;i<nReducedContacts;i++)\n"
-"					{\n"
-"					//this condition means: overwrite contact point, unless at index i==0 we have a valid 'mpr' contact\n"
-"						if (i>0||(mprContactIndex<0))\n"
-"						{\n"
-"							c->m_worldPosB[i] = pointsIn[contactIdx[i]];\n"
-"						}\n"
-"					}\n"
-"					GET_NPOINTS(*c) = nReducedContacts;\n"
-"				}\n"
-"				\n"
-"			}//		if (numContactsOut>0)\n"
-"		}//		if (hasSeparatingAxis[i])\n"
-"	}//	if (i<numPairs)\n"
-"}\n"
-"__kernel void   clipCompoundsHullHullKernel( __global const int4* gpuCompoundPairs, \n"
-"																					__global const b3RigidBodyData_t* rigidBodies, \n"
-"																					__global const b3Collidable_t* collidables,\n"
-"																					__global const b3ConvexPolyhedronData_t* convexShapes, \n"
-"																					__global const float4* vertices,\n"
-"																					__global const float4* uniqueEdges,\n"
-"																					__global const b3GpuFace_t* faces,\n"
-"																					__global const int* indices,\n"
-"																					__global const b3GpuChildShape_t* gpuChildShapes,\n"
-"																					__global const float4* gpuCompoundSepNormalsOut,\n"
-"																					__global const int* gpuHasCompoundSepNormalsOut,\n"
-"																					__global struct b3Contact4Data* restrict globalContactsOut,\n"
-"																					counter32_t nGlobalContactsOut,\n"
-"																					int numCompoundPairs, int maxContactCapacity)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	int pairIndex = i;\n"
-"	\n"
-"	float4 worldVertsB1[64];\n"
-"	float4 worldVertsB2[64];\n"
-"	int capacityWorldVerts = 64;	\n"
-"	float4 localContactsOut[64];\n"
-"	int localContactCapacity=64;\n"
-"	\n"
-"	float minDist = -1e30f;\n"
-"	float maxDist = 0.02f;\n"
-"	if (i<numCompoundPairs)\n"
-"	{\n"
-"		if (gpuHasCompoundSepNormalsOut[i])\n"
-"		{\n"
-"			int bodyIndexA = gpuCompoundPairs[i].x;\n"
-"			int bodyIndexB = gpuCompoundPairs[i].y;\n"
-"			\n"
-"			int childShapeIndexA = gpuCompoundPairs[i].z;\n"
-"			int childShapeIndexB = gpuCompoundPairs[i].w;\n"
-"			\n"
-"			int collidableIndexA = -1;\n"
-"			int collidableIndexB = -1;\n"
-"			\n"
-"			float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
-"			float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
-"			\n"
-"			float4 ornB = rigidBodies[bodyIndexB].m_quat;\n"
-"			float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
-"								\n"
-"			if (childShapeIndexA >= 0)\n"
-"			{\n"
-"				collidableIndexA = gpuChildShapes[childShapeIndexA].m_shapeIndex;\n"
-"				float4 childPosA = gpuChildShapes[childShapeIndexA].m_childPosition;\n"
-"				float4 childOrnA = gpuChildShapes[childShapeIndexA].m_childOrientation;\n"
-"				float4 newPosA = qtRotate(ornA,childPosA)+posA;\n"
-"				float4 newOrnA = qtMul(ornA,childOrnA);\n"
-"				posA = newPosA;\n"
-"				ornA = newOrnA;\n"
-"			} else\n"
-"			{\n"
-"				collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
-"			}\n"
-"			\n"
-"			if (childShapeIndexB>=0)\n"
-"			{\n"
-"				collidableIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;\n"
-"				float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;\n"
-"				float4 childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;\n"
-"				float4 newPosB = transform(&childPosB,&posB,&ornB);\n"
-"				float4 newOrnB = qtMul(ornB,childOrnB);\n"
-"				posB = newPosB;\n"
-"				ornB = newOrnB;\n"
-"			} else\n"
-"			{\n"
-"				collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;	\n"
-"			}\n"
-"			\n"
-"			int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
-"			int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
-"		\n"
-"			int numLocalContactsOut = clipHullAgainstHull(gpuCompoundSepNormalsOut[i],\n"
-"														&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],\n"
-"														posA,ornA,\n"
-"													  posB,ornB,\n"
-"													  worldVertsB1,worldVertsB2,capacityWorldVerts,\n"
-"														minDist, maxDist,\n"
-"														vertices,faces,indices,\n"
-"														localContactsOut,localContactCapacity);\n"
-"												\n"
-"		if (numLocalContactsOut>0)\n"
-"		{\n"
-"				float4 normal = -gpuCompoundSepNormalsOut[i];\n"
-"				int nPoints = numLocalContactsOut;\n"
-"				float4* pointsIn = localContactsOut;\n"
-"				int contactIdx[4];// = {-1,-1,-1,-1};\n"
-"				contactIdx[0] = -1;\n"
-"				contactIdx[1] = -1;\n"
-"				contactIdx[2] = -1;\n"
-"				contactIdx[3] = -1;\n"
-"		\n"
-"				int nReducedContacts = extractManifoldSequential(pointsIn, nPoints, normal, contactIdx);\n"
-"		\n"
-"				int dstIdx;\n"
-"				AppendInc( nGlobalContactsOut, dstIdx );\n"
-"				if ((dstIdx+nReducedContacts) < maxContactCapacity)\n"
-"				{\n"
-"					__global struct b3Contact4Data* c = globalContactsOut+ dstIdx;\n"
-"					c->m_worldNormalOnB = -normal;\n"
-"					c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
-"					c->m_batchIdx = pairIndex;\n"
-"					int bodyA = gpuCompoundPairs[pairIndex].x;\n"
-"					int bodyB = gpuCompoundPairs[pairIndex].y;\n"
-"					c->m_bodyAPtrAndSignBit = rigidBodies[bodyA].m_invMass==0?-bodyA:bodyA;\n"
-"					c->m_bodyBPtrAndSignBit = rigidBodies[bodyB].m_invMass==0?-bodyB:bodyB;\n"
-"					c->m_childIndexA = childShapeIndexA;\n"
-"					c->m_childIndexB = childShapeIndexB;\n"
-"					for (int i=0;i<nReducedContacts;i++)\n"
-"					{\n"
-"						c->m_worldPosB[i] = pointsIn[contactIdx[i]];\n"
-"					}\n"
-"					GET_NPOINTS(*c) = nReducedContacts;\n"
-"				}\n"
-"				\n"
-"			}//		if (numContactsOut>0)\n"
-"		}//		if (gpuHasCompoundSepNormalsOut[i])\n"
-"	}//	if (i<numCompoundPairs)\n"
-"}\n"
-"__kernel void   sphereSphereCollisionKernel( __global const int4* pairs, \n"
-"																					__global const b3RigidBodyData_t* rigidBodies, \n"
-"																					__global const b3Collidable_t* collidables,\n"
-"																					__global const float4* separatingNormals,\n"
-"																					__global const int* hasSeparatingAxis,\n"
-"																					__global struct b3Contact4Data* restrict globalContactsOut,\n"
-"																					counter32_t nGlobalContactsOut,\n"
-"																					int contactCapacity,\n"
-"																					int numPairs)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	int pairIndex = i;\n"
-"	\n"
-"	if (i<numPairs)\n"
-"	{\n"
-"		int bodyIndexA = pairs[i].x;\n"
-"		int bodyIndexB = pairs[i].y;\n"
-"			\n"
-"		int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
-"		int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
-"		if (collidables[collidableIndexA].m_shapeType == SHAPE_SPHERE &&\n"
-"			collidables[collidableIndexB].m_shapeType == SHAPE_SPHERE)\n"
-"		{\n"
-"			//sphere-sphere\n"
-"			float radiusA = collidables[collidableIndexA].m_radius;\n"
-"			float radiusB = collidables[collidableIndexB].m_radius;\n"
-"			float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
-"			float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
-"			float4 diff = posA-posB;\n"
-"			float len = length(diff);\n"
-"			\n"
-"			///iff distance positive, don't generate a new contact\n"
-"			if ( len <= (radiusA+radiusB))\n"
-"			{\n"
-"				///distance (negative means penetration)\n"
-"				float dist = len - (radiusA+radiusB);\n"
-"				float4 normalOnSurfaceB = make_float4(1.f,0.f,0.f,0.f);\n"
-"				if (len > 0.00001)\n"
-"				{\n"
-"					normalOnSurfaceB = diff / len;\n"
-"				}\n"
-"				float4 contactPosB = posB + normalOnSurfaceB*radiusB;\n"
-"				contactPosB.w = dist;\n"
-"								\n"
-"				int dstIdx;\n"
-"				AppendInc( nGlobalContactsOut, dstIdx );\n"
-"				if (dstIdx < contactCapacity)\n"
-"				{\n"
-"					__global struct b3Contact4Data* c = &globalContactsOut[dstIdx];\n"
-"					c->m_worldNormalOnB = -normalOnSurfaceB;\n"
-"					c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
-"					c->m_batchIdx = pairIndex;\n"
-"					int bodyA = pairs[pairIndex].x;\n"
-"					int bodyB = pairs[pairIndex].y;\n"
-"					c->m_bodyAPtrAndSignBit = rigidBodies[bodyA].m_invMass==0?-bodyA:bodyA;\n"
-"					c->m_bodyBPtrAndSignBit = rigidBodies[bodyB].m_invMass==0?-bodyB:bodyB;\n"
-"					c->m_worldPosB[0] = contactPosB;\n"
-"					c->m_childIndexA = -1;\n"
-"					c->m_childIndexB = -1;\n"
-"					GET_NPOINTS(*c) = 1;\n"
-"				}//if (dstIdx < numPairs)\n"
-"			}//if ( len <= (radiusA+radiusB))\n"
-"		}//SHAPE_SPHERE SHAPE_SPHERE\n"
-"	}//if (i<numPairs)\n"
-"}				\n"
-"__kernel void   clipHullHullConcaveConvexKernel( __global int4* concavePairsIn,\n"
-"																					__global const b3RigidBodyData_t* rigidBodies, \n"
-"																					__global const b3Collidable_t* collidables,\n"
-"																					__global const b3ConvexPolyhedronData_t* convexShapes, \n"
-"																					__global const float4* vertices,\n"
-"																					__global const float4* uniqueEdges,\n"
-"																					__global const b3GpuFace_t* faces,\n"
-"																					__global const int* indices,\n"
-"																					__global const b3GpuChildShape_t* gpuChildShapes,\n"
-"																					__global const float4* separatingNormals,\n"
-"																					__global struct b3Contact4Data* restrict globalContactsOut,\n"
-"																					counter32_t nGlobalContactsOut,\n"
-"																					int contactCapacity,\n"
-"																					int numConcavePairs)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	int pairIndex = i;\n"
-"	\n"
-"	float4 worldVertsB1[64];\n"
-"	float4 worldVertsB2[64];\n"
-"	int capacityWorldVerts = 64;	\n"
-"	float4 localContactsOut[64];\n"
-"	int localContactCapacity=64;\n"
-"	\n"
-"	float minDist = -1e30f;\n"
-"	float maxDist = 0.02f;\n"
-"	if (i<numConcavePairs)\n"
-"	{\n"
-"		//negative value means that the pair is invalid\n"
-"		if (concavePairsIn[i].w<0)\n"
-"			return;\n"
-"		int bodyIndexA = concavePairsIn[i].x;\n"
-"		int bodyIndexB = concavePairsIn[i].y;\n"
-"		int f = concavePairsIn[i].z;\n"
-"		int childShapeIndexA = f;\n"
-"		\n"
-"		int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
-"		int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
-"		\n"
-"		int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
-"		int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
-"		\n"
-"		///////////////////////////////////////////////////////////////\n"
-"		\n"
-"	\n"
-"		bool overlap = false;\n"
-"		\n"
-"		b3ConvexPolyhedronData_t convexPolyhedronA;\n"
-"	//add 3 vertices of the triangle\n"
-"		convexPolyhedronA.m_numVertices = 3;\n"
-"		convexPolyhedronA.m_vertexOffset = 0;\n"
-"		float4	localCenter = make_float4(0.f,0.f,0.f,0.f);\n"
-"		b3GpuFace_t face = faces[convexShapes[shapeIndexA].m_faceOffset+f];\n"
-"		\n"
-"		float4 verticesA[3];\n"
-"		for (int i=0;i<3;i++)\n"
-"		{\n"
-"			int index = indices[face.m_indexOffset+i];\n"
-"			float4 vert = vertices[convexShapes[shapeIndexA].m_vertexOffset+index];\n"
-"			verticesA[i] = vert;\n"
-"			localCenter += vert;\n"
-"		}\n"
-"		float dmin = FLT_MAX;\n"
-"		int localCC=0;\n"
-"		//a triangle has 3 unique edges\n"
-"		convexPolyhedronA.m_numUniqueEdges = 3;\n"
-"		convexPolyhedronA.m_uniqueEdgesOffset = 0;\n"
-"		float4 uniqueEdgesA[3];\n"
-"		\n"
-"		uniqueEdgesA[0] = (verticesA[1]-verticesA[0]);\n"
-"		uniqueEdgesA[1] = (verticesA[2]-verticesA[1]);\n"
-"		uniqueEdgesA[2] = (verticesA[0]-verticesA[2]);\n"
-"		convexPolyhedronA.m_faceOffset = 0;\n"
-"                                  \n"
-"		float4 normal = make_float4(face.m_plane.x,face.m_plane.y,face.m_plane.z,0.f);\n"
-"                             \n"
-"		b3GpuFace_t facesA[TRIANGLE_NUM_CONVEX_FACES];\n"
-"		int indicesA[3+3+2+2+2];\n"
-"		int curUsedIndices=0;\n"
-"		int fidx=0;\n"
-"		//front size of triangle\n"
-"		{\n"
-"			facesA[fidx].m_indexOffset=curUsedIndices;\n"
-"			indicesA[0] = 0;\n"
-"			indicesA[1] = 1;\n"
-"			indicesA[2] = 2;\n"
-"			curUsedIndices+=3;\n"
-"			float c = face.m_plane.w;\n"
-"			facesA[fidx].m_plane.x = normal.x;\n"
-"			facesA[fidx].m_plane.y = normal.y;\n"
-"			facesA[fidx].m_plane.z = normal.z;\n"
-"			facesA[fidx].m_plane.w = c;\n"
-"			facesA[fidx].m_numIndices=3;\n"
-"		}\n"
-"		fidx++;\n"
-"		//back size of triangle\n"
-"		{\n"
-"			facesA[fidx].m_indexOffset=curUsedIndices;\n"
-"			indicesA[3]=2;\n"
-"			indicesA[4]=1;\n"
-"			indicesA[5]=0;\n"
-"			curUsedIndices+=3;\n"
-"			float c = dot3F4(normal,verticesA[0]);\n"
-"			float c1 = -face.m_plane.w;\n"
-"			facesA[fidx].m_plane.x = -normal.x;\n"
-"			facesA[fidx].m_plane.y = -normal.y;\n"
-"			facesA[fidx].m_plane.z = -normal.z;\n"
-"			facesA[fidx].m_plane.w = c;\n"
-"			facesA[fidx].m_numIndices=3;\n"
-"		}\n"
-"		fidx++;\n"
-"		bool addEdgePlanes = true;\n"
-"		if (addEdgePlanes)\n"
-"		{\n"
-"			int numVertices=3;\n"
-"			int prevVertex = numVertices-1;\n"
-"			for (int i=0;i<numVertices;i++)\n"
-"			{\n"
-"				float4 v0 = verticesA[i];\n"
-"				float4 v1 = verticesA[prevVertex];\n"
-"                                            \n"
-"				float4 edgeNormal = normalize(cross(normal,v1-v0));\n"
-"				float c = -dot3F4(edgeNormal,v0);\n"
-"				facesA[fidx].m_numIndices = 2;\n"
-"				facesA[fidx].m_indexOffset=curUsedIndices;\n"
-"				indicesA[curUsedIndices++]=i;\n"
-"				indicesA[curUsedIndices++]=prevVertex;\n"
-"                                            \n"
-"				facesA[fidx].m_plane.x = edgeNormal.x;\n"
-"				facesA[fidx].m_plane.y = edgeNormal.y;\n"
-"				facesA[fidx].m_plane.z = edgeNormal.z;\n"
-"				facesA[fidx].m_plane.w = c;\n"
-"				fidx++;\n"
-"				prevVertex = i;\n"
-"			}\n"
-"		}\n"
-"		convexPolyhedronA.m_numFaces = TRIANGLE_NUM_CONVEX_FACES;\n"
-"		convexPolyhedronA.m_localCenter = localCenter*(1.f/3.f);\n"
-"		float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
-"		posA.w = 0.f;\n"
-"		float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
-"		posB.w = 0.f;\n"
-"		float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
-"		float4 ornB =rigidBodies[bodyIndexB].m_quat;\n"
-"		float4 sepAxis = separatingNormals[i];\n"
-"		\n"
-"		int shapeTypeB = collidables[collidableIndexB].m_shapeType;\n"
-"		int childShapeIndexB =-1;\n"
-"		if (shapeTypeB==SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
-"		{\n"
-"			///////////////////\n"
-"			///compound shape support\n"
-"			\n"
-"			childShapeIndexB = concavePairsIn[pairIndex].w;\n"
-"			int childColIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;\n"
-"			shapeIndexB = collidables[childColIndexB].m_shapeIndex;\n"
-"			float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;\n"
-"			float4 childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;\n"
-"			float4 newPosB = transform(&childPosB,&posB,&ornB);\n"
-"			float4 newOrnB = qtMul(ornB,childOrnB);\n"
-"			posB = newPosB;\n"
-"			ornB = newOrnB;\n"
-"			\n"
-"		}\n"
-"		\n"
-"		////////////////////////////////////////\n"
-"		\n"
-"		\n"
-"		\n"
-"		int numLocalContactsOut = clipHullAgainstHullLocalA(sepAxis,\n"
-"														&convexPolyhedronA, &convexShapes[shapeIndexB],\n"
-"														posA,ornA,\n"
-"													  posB,ornB,\n"
-"													  worldVertsB1,worldVertsB2,capacityWorldVerts,\n"
-"														minDist, maxDist,\n"
-"														&verticesA,&facesA,&indicesA,\n"
-"														vertices,faces,indices,\n"
-"														localContactsOut,localContactCapacity);\n"
-"												\n"
-"		if (numLocalContactsOut>0)\n"
-"		{\n"
-"			float4 normal = -separatingNormals[i];\n"
-"			int nPoints = numLocalContactsOut;\n"
-"			float4* pointsIn = localContactsOut;\n"
-"			int contactIdx[4];// = {-1,-1,-1,-1};\n"
-"			contactIdx[0] = -1;\n"
-"			contactIdx[1] = -1;\n"
-"			contactIdx[2] = -1;\n"
-"			contactIdx[3] = -1;\n"
-"	\n"
-"			int nReducedContacts = extractManifoldSequential(pointsIn, nPoints, normal, contactIdx);\n"
-"	\n"
-"			int dstIdx;\n"
-"			AppendInc( nGlobalContactsOut, dstIdx );\n"
-"			if (dstIdx<contactCapacity)\n"
-"			{\n"
-"				__global struct b3Contact4Data* c = globalContactsOut+ dstIdx;\n"
-"				c->m_worldNormalOnB = -normal;\n"
-"				c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
-"				c->m_batchIdx = pairIndex;\n"
-"				int bodyA = concavePairsIn[pairIndex].x;\n"
-"				int bodyB = concavePairsIn[pairIndex].y;\n"
-"				c->m_bodyAPtrAndSignBit = rigidBodies[bodyA].m_invMass==0?-bodyA:bodyA;\n"
-"				c->m_bodyBPtrAndSignBit = rigidBodies[bodyB].m_invMass==0?-bodyB:bodyB;\n"
-"				c->m_childIndexA = childShapeIndexA;\n"
-"				c->m_childIndexB = childShapeIndexB;\n"
-"				for (int i=0;i<nReducedContacts;i++)\n"
-"				{\n"
-"					c->m_worldPosB[i] = pointsIn[contactIdx[i]];\n"
-"				}\n"
-"				GET_NPOINTS(*c) = nReducedContacts;\n"
-"			}\n"
-"				\n"
-"		}//		if (numContactsOut>0)\n"
-"	}//	if (i<numPairs)\n"
-"}\n"
-"int	findClippingFaces(const float4 separatingNormal,\n"
-"                      __global const b3ConvexPolyhedronData_t* hullA, __global const b3ConvexPolyhedronData_t* hullB,\n"
-"                      const float4 posA, const Quaternion ornA,const float4 posB, const Quaternion ornB,\n"
-"                       __global float4* worldVertsA1,\n"
-"                      __global float4* worldNormalsA1,\n"
-"                      __global float4* worldVertsB1,\n"
-"                      int capacityWorldVerts,\n"
-"                      const float minDist, float maxDist,\n"
-"                      __global const float4* vertices,\n"
-"                      __global const b3GpuFace_t* faces,\n"
-"                      __global const int* indices,\n"
-"                      __global int4* clippingFaces, int pairIndex)\n"
-"{\n"
-"	int numContactsOut = 0;\n"
-"	int numWorldVertsB1= 0;\n"
-"    \n"
-"    \n"
-"	int closestFaceB=-1;\n"
-"	float dmax = -FLT_MAX;\n"
-"    \n"
-"	{\n"
-"		for(int face=0;face<hullB->m_numFaces;face++)\n"
-"		{\n"
-"			const float4 Normal = make_float4(faces[hullB->m_faceOffset+face].m_plane.x,\n"
-"                                              faces[hullB->m_faceOffset+face].m_plane.y, faces[hullB->m_faceOffset+face].m_plane.z,0.f);\n"
-"			const float4 WorldNormal = qtRotate(ornB, Normal);\n"
-"			float d = dot3F4(WorldNormal,separatingNormal);\n"
-"			if (d > dmax)\n"
-"			{\n"
-"				dmax = d;\n"
-"				closestFaceB = face;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"    \n"
-"	{\n"
-"		const b3GpuFace_t polyB = faces[hullB->m_faceOffset+closestFaceB];\n"
-"		const int numVertices = polyB.m_numIndices;\n"
-"		for(int e0=0;e0<numVertices;e0++)\n"
-"		{\n"
-"			const float4 b = vertices[hullB->m_vertexOffset+indices[polyB.m_indexOffset+e0]];\n"
-"			worldVertsB1[pairIndex*capacityWorldVerts+numWorldVertsB1++] = transform(&b,&posB,&ornB);\n"
-"		}\n"
-"	}\n"
-"    \n"
-"    int closestFaceA=-1;\n"
-"	{\n"
-"		float dmin = FLT_MAX;\n"
-"		for(int face=0;face<hullA->m_numFaces;face++)\n"
-"		{\n"
-"			const float4 Normal = make_float4(\n"
-"                                              faces[hullA->m_faceOffset+face].m_plane.x,\n"
-"                                              faces[hullA->m_faceOffset+face].m_plane.y,\n"
-"                                              faces[hullA->m_faceOffset+face].m_plane.z,\n"
-"                                              0.f);\n"
-"			const float4 faceANormalWS = qtRotate(ornA,Normal);\n"
-"            \n"
-"			float d = dot3F4(faceANormalWS,separatingNormal);\n"
-"			if (d < dmin)\n"
-"			{\n"
-"				dmin = d;\n"
-"				closestFaceA = face;\n"
-"                worldNormalsA1[pairIndex] = faceANormalWS;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"    \n"
-"    int numVerticesA = faces[hullA->m_faceOffset+closestFaceA].m_numIndices;\n"
-"	for(int e0=0;e0<numVerticesA;e0++)\n"
-"	{\n"
-"        const float4 a = vertices[hullA->m_vertexOffset+indices[faces[hullA->m_faceOffset+closestFaceA].m_indexOffset+e0]];\n"
-"        worldVertsA1[pairIndex*capacityWorldVerts+e0] = transform(&a, &posA,&ornA);\n"
-"    }\n"
-"    \n"
-"    clippingFaces[pairIndex].x = closestFaceA;\n"
-"    clippingFaces[pairIndex].y = closestFaceB;\n"
-"    clippingFaces[pairIndex].z = numVerticesA;\n"
-"    clippingFaces[pairIndex].w = numWorldVertsB1;\n"
-"    \n"
-"    \n"
-"	return numContactsOut;\n"
-"}\n"
-"int clipFaces(__global float4* worldVertsA1,\n"
-"              __global float4* worldNormalsA1,\n"
-"              __global float4* worldVertsB1,\n"
-"              __global float4* worldVertsB2, \n"
-"              int capacityWorldVertsB2,\n"
-"              const float minDist, float maxDist,\n"
-"              __global int4* clippingFaces,\n"
-"              int pairIndex)\n"
-"{\n"
-"	int numContactsOut = 0;\n"
-"    \n"
-"    int closestFaceA = clippingFaces[pairIndex].x;\n"
-"    int closestFaceB = clippingFaces[pairIndex].y;\n"
-"	int numVertsInA = clippingFaces[pairIndex].z;\n"
-"	int numVertsInB = clippingFaces[pairIndex].w;\n"
-"    \n"
-"	int numVertsOut = 0;\n"
-"    \n"
-"	if (closestFaceA<0)\n"
-"		return numContactsOut;\n"
-"    \n"
-"    __global float4* pVtxIn = &worldVertsB1[pairIndex*capacityWorldVertsB2];\n"
-"    __global float4* pVtxOut = &worldVertsB2[pairIndex*capacityWorldVertsB2];\n"
-"    \n"
-"    \n"
-"	\n"
-"	// clip polygon to back of planes of all faces of hull A that are adjacent to witness face\n"
-"    \n"
-"	for(int e0=0;e0<numVertsInA;e0++)\n"
-"	{\n"
-"		const float4 aw = worldVertsA1[pairIndex*capacityWorldVertsB2+e0];\n"
-"		const float4 bw = worldVertsA1[pairIndex*capacityWorldVertsB2+((e0+1)%numVertsInA)];\n"
-"		const float4 WorldEdge0 = aw - bw;\n"
-"		float4 worldPlaneAnormal1 = worldNormalsA1[pairIndex];\n"
-"		float4 planeNormalWS1 = -cross3(WorldEdge0,worldPlaneAnormal1);\n"
-"		float4 worldA1 = aw;\n"
-"		float planeEqWS1 = -dot3F4(worldA1,planeNormalWS1);\n"
-"		float4 planeNormalWS = planeNormalWS1;\n"
-"		float planeEqWS=planeEqWS1;\n"
-"		numVertsOut = clipFaceGlobal(pVtxIn, numVertsInB, planeNormalWS,planeEqWS, pVtxOut);\n"
-"		__global float4* tmp = pVtxOut;\n"
-"		pVtxOut = pVtxIn;\n"
-"		pVtxIn = tmp;\n"
-"		numVertsInB = numVertsOut;\n"
-"		numVertsOut = 0;\n"
-"	}\n"
-"    \n"
-"    //float4 planeNormalWS = worldNormalsA1[pairIndex];\n"
-"    //float planeEqWS=-dot3F4(planeNormalWS,worldVertsA1[pairIndex*capacityWorldVertsB2]);\n"
-"    \n"
-"    /*for (int i=0;i<numVertsInB;i++)\n"
-"    {\n"
-"        pVtxOut[i] = pVtxIn[i];\n"
-"    }*/\n"
-"    \n"
-"    \n"
-"    \n"
-"    \n"
-"    //numVertsInB=0;\n"
-"	\n"
-"    float4 planeNormalWS = worldNormalsA1[pairIndex];\n"
-"    float planeEqWS=-dot3F4(planeNormalWS,worldVertsA1[pairIndex*capacityWorldVertsB2]);\n"
-"    for (int i=0;i<numVertsInB;i++)\n"
-"    {\n"
-"        float depth = dot3F4(planeNormalWS,pVtxIn[i])+planeEqWS;\n"
-"        if (depth <=minDist)\n"
-"        {\n"
-"            depth = minDist;\n"
-"        }\n"
-"        \n"
-"        if (depth <=maxDist)\n"
-"        {\n"
-"            float4 pointInWorld = pVtxIn[i];\n"
-"            pVtxOut[numContactsOut++] = make_float4(pointInWorld.x,pointInWorld.y,pointInWorld.z,depth);\n"
-"        }\n"
-"    }\n"
-"   \n"
-"    clippingFaces[pairIndex].w =numContactsOut;\n"
-"   \n"
-"    \n"
-"	return numContactsOut;\n"
-"}\n"
-"__kernel void   findClippingFacesKernel(  __global const int4* pairs,\n"
-"                                        __global const b3RigidBodyData_t* rigidBodies,\n"
-"                                        __global const b3Collidable_t* collidables,\n"
-"                                        __global const b3ConvexPolyhedronData_t* convexShapes,\n"
-"                                        __global const float4* vertices,\n"
-"                                        __global const float4* uniqueEdges,\n"
-"                                        __global const b3GpuFace_t* faces,\n"
-"                                        __global const int* indices,\n"
-"                                        __global const float4* separatingNormals,\n"
-"                                        __global const int* hasSeparatingAxis,\n"
-"                                        __global int4* clippingFacesOut,\n"
-"                                        __global float4* worldVertsA1,\n"
-"                                        __global float4* worldNormalsA1,\n"
-"                                        __global float4* worldVertsB1,\n"
-"                                        int capacityWorldVerts,\n"
-"                                        int numPairs\n"
-"                                        )\n"
-"{\n"
-"    \n"
-"	int i = get_global_id(0);\n"
-"	int pairIndex = i;\n"
-"    \n"
-"	\n"
-"	float minDist = -1e30f;\n"
-"	float maxDist = 0.02f;\n"
-"    \n"
-"	if (i<numPairs)\n"
-"	{\n"
-"        \n"
-"		if (hasSeparatingAxis[i])\n"
-"		{\n"
-"            \n"
-"			int bodyIndexA = pairs[i].x;\n"
-"			int bodyIndexB = pairs[i].y;\n"
-"			\n"
-"			int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
-"			int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
-"			\n"
-"			int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
-"			int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
-"			\n"
-"            \n"
-"            \n"
-"			int numLocalContactsOut = findClippingFaces(separatingNormals[i],\n"
-"                                                        &convexShapes[shapeIndexA], &convexShapes[shapeIndexB],\n"
-"                                                        rigidBodies[bodyIndexA].m_pos,rigidBodies[bodyIndexA].m_quat,\n"
-"                                                        rigidBodies[bodyIndexB].m_pos,rigidBodies[bodyIndexB].m_quat,\n"
-"                                                        worldVertsA1,\n"
-"                                                        worldNormalsA1,\n"
-"                                                        worldVertsB1,capacityWorldVerts,\n"
-"                                                        minDist, maxDist,\n"
-"                                                        vertices,faces,indices,\n"
-"                                                        clippingFacesOut,i);\n"
-"            \n"
-"            \n"
-"		}//		if (hasSeparatingAxis[i])\n"
-"	}//	if (i<numPairs)\n"
-"    \n"
-"}\n"
-"__kernel void   clipFacesAndFindContactsKernel(    __global const float4* separatingNormals,\n"
-"                                                   __global const int* hasSeparatingAxis,\n"
-"                                                   __global int4* clippingFacesOut,\n"
-"                                                   __global float4* worldVertsA1,\n"
-"                                                   __global float4* worldNormalsA1,\n"
-"                                                   __global float4* worldVertsB1,\n"
-"                                                   __global float4* worldVertsB2,\n"
-"                                                    int vertexFaceCapacity,\n"
-"                                                   int numPairs,\n"
-"					                                        int debugMode\n"
-"                                                   )\n"
-"{\n"
-"    int i = get_global_id(0);\n"
-"	int pairIndex = i;\n"
-"	\n"
-"    \n"
-"	float minDist = -1e30f;\n"
-"	float maxDist = 0.02f;\n"
-"    \n"
-"	if (i<numPairs)\n"
-"	{\n"
-"        \n"
-"		if (hasSeparatingAxis[i])\n"
-"		{\n"
-"            \n"
-"//			int bodyIndexA = pairs[i].x;\n"
-"	//		int bodyIndexB = pairs[i].y;\n"
-"		    \n"
-"            int numLocalContactsOut = 0;\n"
-"            int capacityWorldVertsB2 = vertexFaceCapacity;\n"
-"            \n"
-"            __global float4* pVtxIn = &worldVertsB1[pairIndex*capacityWorldVertsB2];\n"
-"            __global float4* pVtxOut = &worldVertsB2[pairIndex*capacityWorldVertsB2];\n"
-"            \n"
-"            {\n"
-"                __global int4* clippingFaces = clippingFacesOut;\n"
-"            \n"
-"                \n"
-"                int closestFaceA = clippingFaces[pairIndex].x;\n"
-"                int closestFaceB = clippingFaces[pairIndex].y;\n"
-"                int numVertsInA = clippingFaces[pairIndex].z;\n"
-"                int numVertsInB = clippingFaces[pairIndex].w;\n"
-"                \n"
-"                int numVertsOut = 0;\n"
-"                \n"
-"                if (closestFaceA>=0)\n"
-"                {\n"
-"                    \n"
-"                    \n"
-"                    \n"
-"                    // clip polygon to back of planes of all faces of hull A that are adjacent to witness face\n"
-"                    \n"
-"                    for(int e0=0;e0<numVertsInA;e0++)\n"
-"                    {\n"
-"                        const float4 aw = worldVertsA1[pairIndex*capacityWorldVertsB2+e0];\n"
-"                        const float4 bw = worldVertsA1[pairIndex*capacityWorldVertsB2+((e0+1)%numVertsInA)];\n"
-"                        const float4 WorldEdge0 = aw - bw;\n"
-"                        float4 worldPlaneAnormal1 = worldNormalsA1[pairIndex];\n"
-"                        float4 planeNormalWS1 = -cross3(WorldEdge0,worldPlaneAnormal1);\n"
-"                        float4 worldA1 = aw;\n"
-"                        float planeEqWS1 = -dot3F4(worldA1,planeNormalWS1);\n"
-"                        float4 planeNormalWS = planeNormalWS1;\n"
-"                        float planeEqWS=planeEqWS1;\n"
-"                        numVertsOut = clipFaceGlobal(pVtxIn, numVertsInB, planeNormalWS,planeEqWS, pVtxOut);\n"
-"                        __global float4* tmp = pVtxOut;\n"
-"                        pVtxOut = pVtxIn;\n"
-"                        pVtxIn = tmp;\n"
-"                        numVertsInB = numVertsOut;\n"
-"                        numVertsOut = 0;\n"
-"                    }\n"
-"                    \n"
-"                    float4 planeNormalWS = worldNormalsA1[pairIndex];\n"
-"                    float planeEqWS=-dot3F4(planeNormalWS,worldVertsA1[pairIndex*capacityWorldVertsB2]);\n"
-"                    \n"
-"                    for (int i=0;i<numVertsInB;i++)\n"
-"                    {\n"
-"                        float depth = dot3F4(planeNormalWS,pVtxIn[i])+planeEqWS;\n"
-"                        if (depth <=minDist)\n"
-"                        {\n"
-"                            depth = minDist;\n"
-"                        }\n"
-"                        \n"
-"                        if (depth <=maxDist)\n"
-"                        {\n"
-"                            float4 pointInWorld = pVtxIn[i];\n"
-"                            pVtxOut[numLocalContactsOut++] = make_float4(pointInWorld.x,pointInWorld.y,pointInWorld.z,depth);\n"
-"                        }\n"
-"                    }\n"
-"                    \n"
-"                }\n"
-"                clippingFaces[pairIndex].w =numLocalContactsOut;\n"
-"                \n"
-"            }\n"
-"            \n"
-"            for (int i=0;i<numLocalContactsOut;i++)\n"
-"                pVtxIn[i] = pVtxOut[i];\n"
-"                \n"
-"		}//		if (hasSeparatingAxis[i])\n"
-"	}//	if (i<numPairs)\n"
-"    \n"
-"}\n"
-"__kernel void   newContactReductionKernel( __global int4* pairs,\n"
-"                                                   __global const b3RigidBodyData_t* rigidBodies,\n"
-"                                                   __global const float4* separatingNormals,\n"
-"                                                   __global const int* hasSeparatingAxis,\n"
-"                                                   __global struct b3Contact4Data* globalContactsOut,\n"
-"                                                   __global int4* clippingFaces,\n"
-"                                                   __global float4* worldVertsB2,\n"
-"                                                   volatile __global int* nGlobalContactsOut,\n"
-"                                                   int vertexFaceCapacity,\n"
-"												   int contactCapacity,\n"
-"                                                   int numPairs\n"
-"                                                   )\n"
-"{\n"
-"    int i = get_global_id(0);\n"
-"	int pairIndex = i;\n"
-"	\n"
-"    int4 contactIdx;\n"
-"    contactIdx=make_int4(0,1,2,3);\n"
-"    \n"
-"	if (i<numPairs)\n"
-"	{\n"
-"        \n"
-"		if (hasSeparatingAxis[i])\n"
-"		{\n"
-"            \n"
-"			\n"
-"            \n"
-"            \n"
-"			int nPoints = clippingFaces[pairIndex].w;\n"
-"           \n"
-"            if (nPoints>0)\n"
-"            {\n"
-"                 __global float4* pointsIn = &worldVertsB2[pairIndex*vertexFaceCapacity];\n"
-"                float4 normal = -separatingNormals[i];\n"
-"                \n"
-"                int nReducedContacts = extractManifoldSequentialGlobal(pointsIn, nPoints, normal, &contactIdx);\n"
-"            \n"
-"				int mprContactIndex = pairs[pairIndex].z;\n"
-"                int dstIdx = mprContactIndex;\n"
-"				if (dstIdx<0)\n"
-"				{\n"
-"	                AppendInc( nGlobalContactsOut, dstIdx );\n"
-"				}\n"
-"//#if 0\n"
-"                \n"
-"				if (dstIdx < contactCapacity)\n"
-"				{\n"
-"					__global struct b3Contact4Data* c = &globalContactsOut[dstIdx];\n"
-"					c->m_worldNormalOnB = -normal;\n"
-"					c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
-"					c->m_batchIdx = pairIndex;\n"
-"					int bodyA = pairs[pairIndex].x;\n"
-"					int bodyB = pairs[pairIndex].y;\n"
-"					pairs[pairIndex].w = dstIdx;\n"
-"					c->m_bodyAPtrAndSignBit = rigidBodies[bodyA].m_invMass==0?-bodyA:bodyA;\n"
-"					c->m_bodyBPtrAndSignBit = rigidBodies[bodyB].m_invMass==0?-bodyB:bodyB;\n"
-"                    c->m_childIndexA =-1;\n"
-"					c->m_childIndexB =-1;\n"
-"                    switch (nReducedContacts)\n"
-"                    {\n"
-"                        case 4:\n"
-"                            c->m_worldPosB[3] = pointsIn[contactIdx.w];\n"
-"                        case 3:\n"
-"                            c->m_worldPosB[2] = pointsIn[contactIdx.z];\n"
-"                        case 2:\n"
-"                            c->m_worldPosB[1] = pointsIn[contactIdx.y];\n"
-"                        case 1:\n"
-"							if (mprContactIndex<0)//test\n"
-"	                            c->m_worldPosB[0] = pointsIn[contactIdx.x];\n"
-"                        default:\n"
-"                        {\n"
-"                        }\n"
-"                    };\n"
-"                    \n"
-"					GET_NPOINTS(*c) = nReducedContacts;\n"
-"                    \n"
-"                 }\n"
-"                 \n"
-"                \n"
-"//#endif\n"
-"				\n"
-"			}//		if (numContactsOut>0)\n"
-"		}//		if (hasSeparatingAxis[i])\n"
-"	}//	if (i<numPairs)\n"
-"    \n"
-"    \n"
-"}\n"
-;
+static const char* satClipKernelsCL =
+	"#define TRIANGLE_NUM_CONVEX_FACES 5\n"
+	"#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_extended_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics : enable\n"
+	"#ifdef cl_ext_atomic_counters_32\n"
+	"#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
+	"#else\n"
+	"#define counter32_t volatile __global int*\n"
+	"#endif\n"
+	"#define GET_GROUP_IDX get_group_id(0)\n"
+	"#define GET_LOCAL_IDX get_local_id(0)\n"
+	"#define GET_GLOBAL_IDX get_global_id(0)\n"
+	"#define GET_GROUP_SIZE get_local_size(0)\n"
+	"#define GET_NUM_GROUPS get_num_groups(0)\n"
+	"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
+	"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
+	"#define AtomInc(x) atom_inc(&(x))\n"
+	"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
+	"#define AppendInc(x, out) out = atomic_inc(x)\n"
+	"#define AtomAdd(x, value) atom_add(&(x), value)\n"
+	"#define AtomCmpxhg(x, cmp, value) atom_cmpxchg( &(x), cmp, value )\n"
+	"#define AtomXhg(x, value) atom_xchg ( &(x), value )\n"
+	"#define max2 max\n"
+	"#define min2 min\n"
+	"typedef unsigned int u32;\n"
+	"#ifndef B3_CONTACT4DATA_H\n"
+	"#define B3_CONTACT4DATA_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#define B3_FLOAT4_H\n"
+	"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
+	"#define B3_PLATFORM_DEFINITIONS_H\n"
+	"struct MyTest\n"
+	"{\n"
+	"	int bla;\n"
+	"};\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
+	"#define B3_LARGE_FLOAT 1e18f\n"
+	"#define B3_INFINITY 1e18f\n"
+	"#define b3Assert(a)\n"
+	"#define b3ConstArray(a) __global const a*\n"
+	"#define b3AtomicInc atomic_inc\n"
+	"#define b3AtomicAdd atomic_add\n"
+	"#define b3Fabs fabs\n"
+	"#define b3Sqrt native_sqrt\n"
+	"#define b3Sin native_sin\n"
+	"#define b3Cos native_cos\n"
+	"#define B3_STATIC\n"
+	"#endif\n"
+	"#endif\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"	typedef float4	b3Float4;\n"
+	"	#define b3Float4ConstArg const b3Float4\n"
+	"	#define b3MakeFloat4 (float4)\n"
+	"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return dot(a1, b1);\n"
+	"	}\n"
+	"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return cross(a1, b1);\n"
+	"	}\n"
+	"	#define b3MinFloat4 min\n"
+	"	#define b3MaxFloat4 max\n"
+	"	#define b3Normalized(a) normalize(a)\n"
+	"#endif \n"
+	"		\n"
+	"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
+	"{\n"
+	"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
+	"		return false;\n"
+	"	return true;\n"
+	"}\n"
+	"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
+	"{\n"
+	"    float maxDot = -B3_INFINITY;\n"
+	"    int i = 0;\n"
+	"    int ptIndex = -1;\n"
+	"    for( i = 0; i < vecLen; i++ )\n"
+	"    {\n"
+	"        float dot = b3Dot3F4(vecArray[i],vec);\n"
+	"            \n"
+	"        if( dot > maxDot )\n"
+	"        {\n"
+	"            maxDot = dot;\n"
+	"            ptIndex = i;\n"
+	"        }\n"
+	"    }\n"
+	"	b3Assert(ptIndex>=0);\n"
+	"    if (ptIndex<0)\n"
+	"	{\n"
+	"		ptIndex = 0;\n"
+	"	}\n"
+	"    *dotOut = maxDot;\n"
+	"    return ptIndex;\n"
+	"}\n"
+	"#endif //B3_FLOAT4_H\n"
+	"typedef  struct b3Contact4Data b3Contact4Data_t;\n"
+	"struct b3Contact4Data\n"
+	"{\n"
+	"	b3Float4	m_worldPosB[4];\n"
+	"//	b3Float4	m_localPosA[4];\n"
+	"//	b3Float4	m_localPosB[4];\n"
+	"	b3Float4	m_worldNormalOnB;	//	w: m_nPoints\n"
+	"	unsigned short  m_restituitionCoeffCmp;\n"
+	"	unsigned short  m_frictionCoeffCmp;\n"
+	"	int m_batchIdx;\n"
+	"	int m_bodyAPtrAndSignBit;//x:m_bodyAPtr, y:m_bodyBPtr\n"
+	"	int m_bodyBPtrAndSignBit;\n"
+	"	int	m_childIndexA;\n"
+	"	int	m_childIndexB;\n"
+	"	int m_unused1;\n"
+	"	int m_unused2;\n"
+	"};\n"
+	"inline int b3Contact4Data_getNumPoints(const struct b3Contact4Data* contact)\n"
+	"{\n"
+	"	return (int)contact->m_worldNormalOnB.w;\n"
+	"};\n"
+	"inline void b3Contact4Data_setNumPoints(struct b3Contact4Data* contact, int numPoints)\n"
+	"{\n"
+	"	contact->m_worldNormalOnB.w = (float)numPoints;\n"
+	"};\n"
+	"#endif //B3_CONTACT4DATA_H\n"
+	"#ifndef B3_CONVEX_POLYHEDRON_DATA_H\n"
+	"#define B3_CONVEX_POLYHEDRON_DATA_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifndef B3_QUAT_H\n"
+	"#define B3_QUAT_H\n"
+	"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif\n"
+	"#endif\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"	typedef float4	b3Quat;\n"
+	"	#define b3QuatConstArg const b3Quat\n"
+	"	\n"
+	"	\n"
+	"inline float4 b3FastNormalize4(float4 v)\n"
+	"{\n"
+	"	v = (float4)(v.xyz,0.f);\n"
+	"	return fast_normalize(v);\n"
+	"}\n"
+	"	\n"
+	"inline b3Quat b3QuatMul(b3Quat a, b3Quat b);\n"
+	"inline b3Quat b3QuatNormalized(b3QuatConstArg in);\n"
+	"inline b3Quat b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec);\n"
+	"inline b3Quat b3QuatInvert(b3QuatConstArg q);\n"
+	"inline b3Quat b3QuatInverse(b3QuatConstArg q);\n"
+	"inline b3Quat b3QuatMul(b3QuatConstArg a, b3QuatConstArg b)\n"
+	"{\n"
+	"	b3Quat ans;\n"
+	"	ans = b3Cross3( a, b );\n"
+	"	ans += a.w*b+b.w*a;\n"
+	"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
+	"	ans.w = a.w*b.w - b3Dot3F4(a, b);\n"
+	"	return ans;\n"
+	"}\n"
+	"inline b3Quat b3QuatNormalized(b3QuatConstArg in)\n"
+	"{\n"
+	"	b3Quat q;\n"
+	"	q=in;\n"
+	"	//return b3FastNormalize4(in);\n"
+	"	float len = native_sqrt(dot(q, q));\n"
+	"	if(len > 0.f)\n"
+	"	{\n"
+	"		q *= 1.f / len;\n"
+	"	}\n"
+	"	else\n"
+	"	{\n"
+	"		q.x = q.y = q.z = 0.f;\n"
+	"		q.w = 1.f;\n"
+	"	}\n"
+	"	return q;\n"
+	"}\n"
+	"inline float4 b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
+	"{\n"
+	"	b3Quat qInv = b3QuatInvert( q );\n"
+	"	float4 vcpy = vec;\n"
+	"	vcpy.w = 0.f;\n"
+	"	float4 out = b3QuatMul(b3QuatMul(q,vcpy),qInv);\n"
+	"	return out;\n"
+	"}\n"
+	"inline b3Quat b3QuatInverse(b3QuatConstArg q)\n"
+	"{\n"
+	"	return (b3Quat)(-q.xyz, q.w);\n"
+	"}\n"
+	"inline b3Quat b3QuatInvert(b3QuatConstArg q)\n"
+	"{\n"
+	"	return (b3Quat)(-q.xyz, q.w);\n"
+	"}\n"
+	"inline float4 b3QuatInvRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
+	"{\n"
+	"	return b3QuatRotate( b3QuatInvert( q ), vec );\n"
+	"}\n"
+	"inline b3Float4 b3TransformPoint(b3Float4ConstArg point, b3Float4ConstArg translation, b3QuatConstArg  orientation)\n"
+	"{\n"
+	"	return b3QuatRotate( orientation, point ) + (translation);\n"
+	"}\n"
+	"	\n"
+	"#endif \n"
+	"#endif //B3_QUAT_H\n"
+	"typedef struct b3GpuFace b3GpuFace_t;\n"
+	"struct b3GpuFace\n"
+	"{\n"
+	"	b3Float4 m_plane;\n"
+	"	int m_indexOffset;\n"
+	"	int m_numIndices;\n"
+	"	int m_unusedPadding1;\n"
+	"	int m_unusedPadding2;\n"
+	"};\n"
+	"typedef struct b3ConvexPolyhedronData b3ConvexPolyhedronData_t;\n"
+	"struct b3ConvexPolyhedronData\n"
+	"{\n"
+	"	b3Float4		m_localCenter;\n"
+	"	b3Float4		m_extents;\n"
+	"	b3Float4		mC;\n"
+	"	b3Float4		mE;\n"
+	"	float			m_radius;\n"
+	"	int	m_faceOffset;\n"
+	"	int m_numFaces;\n"
+	"	int	m_numVertices;\n"
+	"	int m_vertexOffset;\n"
+	"	int	m_uniqueEdgesOffset;\n"
+	"	int	m_numUniqueEdges;\n"
+	"	int m_unused;\n"
+	"};\n"
+	"#endif //B3_CONVEX_POLYHEDRON_DATA_H\n"
+	"#ifndef B3_COLLIDABLE_H\n"
+	"#define B3_COLLIDABLE_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifndef B3_QUAT_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_QUAT_H\n"
+	"enum b3ShapeTypes\n"
+	"{\n"
+	"	SHAPE_HEIGHT_FIELD=1,\n"
+	"	SHAPE_CONVEX_HULL=3,\n"
+	"	SHAPE_PLANE=4,\n"
+	"	SHAPE_CONCAVE_TRIMESH=5,\n"
+	"	SHAPE_COMPOUND_OF_CONVEX_HULLS=6,\n"
+	"	SHAPE_SPHERE=7,\n"
+	"	MAX_NUM_SHAPE_TYPES,\n"
+	"};\n"
+	"typedef struct b3Collidable b3Collidable_t;\n"
+	"struct b3Collidable\n"
+	"{\n"
+	"	union {\n"
+	"		int m_numChildShapes;\n"
+	"		int m_bvhIndex;\n"
+	"	};\n"
+	"	union\n"
+	"	{\n"
+	"		float m_radius;\n"
+	"		int	m_compoundBvhIndex;\n"
+	"	};\n"
+	"	int m_shapeType;\n"
+	"	int m_shapeIndex;\n"
+	"};\n"
+	"typedef struct b3GpuChildShape b3GpuChildShape_t;\n"
+	"struct b3GpuChildShape\n"
+	"{\n"
+	"	b3Float4	m_childPosition;\n"
+	"	b3Quat		m_childOrientation;\n"
+	"	int m_shapeIndex;\n"
+	"	int m_unused0;\n"
+	"	int m_unused1;\n"
+	"	int m_unused2;\n"
+	"};\n"
+	"struct b3CompoundOverlappingPair\n"
+	"{\n"
+	"	int m_bodyIndexA;\n"
+	"	int m_bodyIndexB;\n"
+	"//	int	m_pairType;\n"
+	"	int m_childShapeIndexA;\n"
+	"	int m_childShapeIndexB;\n"
+	"};\n"
+	"#endif //B3_COLLIDABLE_H\n"
+	"#ifndef B3_RIGIDBODY_DATA_H\n"
+	"#define B3_RIGIDBODY_DATA_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifndef B3_QUAT_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_QUAT_H\n"
+	"#ifndef B3_MAT3x3_H\n"
+	"#define B3_MAT3x3_H\n"
+	"#ifndef B3_QUAT_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_QUAT_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"typedef struct\n"
+	"{\n"
+	"	b3Float4 m_row[3];\n"
+	"}b3Mat3x3;\n"
+	"#define b3Mat3x3ConstArg const b3Mat3x3\n"
+	"#define b3GetRow(m,row) (m.m_row[row])\n"
+	"inline b3Mat3x3 b3QuatGetRotationMatrix(b3Quat quat)\n"
+	"{\n"
+	"	b3Float4 quat2 = (b3Float4)(quat.x*quat.x, quat.y*quat.y, quat.z*quat.z, 0.f);\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0].x=1-2*quat2.y-2*quat2.z;\n"
+	"	out.m_row[0].y=2*quat.x*quat.y-2*quat.w*quat.z;\n"
+	"	out.m_row[0].z=2*quat.x*quat.z+2*quat.w*quat.y;\n"
+	"	out.m_row[0].w = 0.f;\n"
+	"	out.m_row[1].x=2*quat.x*quat.y+2*quat.w*quat.z;\n"
+	"	out.m_row[1].y=1-2*quat2.x-2*quat2.z;\n"
+	"	out.m_row[1].z=2*quat.y*quat.z-2*quat.w*quat.x;\n"
+	"	out.m_row[1].w = 0.f;\n"
+	"	out.m_row[2].x=2*quat.x*quat.z-2*quat.w*quat.y;\n"
+	"	out.m_row[2].y=2*quat.y*quat.z+2*quat.w*quat.x;\n"
+	"	out.m_row[2].z=1-2*quat2.x-2*quat2.y;\n"
+	"	out.m_row[2].w = 0.f;\n"
+	"	return out;\n"
+	"}\n"
+	"inline b3Mat3x3 b3AbsoluteMat3x3(b3Mat3x3ConstArg matIn)\n"
+	"{\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0] = fabs(matIn.m_row[0]);\n"
+	"	out.m_row[1] = fabs(matIn.m_row[1]);\n"
+	"	out.m_row[2] = fabs(matIn.m_row[2]);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtZero();\n"
+	"__inline\n"
+	"b3Mat3x3 mtIdentity();\n"
+	"__inline\n"
+	"b3Mat3x3 mtTranspose(b3Mat3x3 m);\n"
+	"__inline\n"
+	"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b);\n"
+	"__inline\n"
+	"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b);\n"
+	"__inline\n"
+	"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b);\n"
+	"__inline\n"
+	"b3Mat3x3 mtZero()\n"
+	"{\n"
+	"	b3Mat3x3 m;\n"
+	"	m.m_row[0] = (b3Float4)(0.f);\n"
+	"	m.m_row[1] = (b3Float4)(0.f);\n"
+	"	m.m_row[2] = (b3Float4)(0.f);\n"
+	"	return m;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtIdentity()\n"
+	"{\n"
+	"	b3Mat3x3 m;\n"
+	"	m.m_row[0] = (b3Float4)(1,0,0,0);\n"
+	"	m.m_row[1] = (b3Float4)(0,1,0,0);\n"
+	"	m.m_row[2] = (b3Float4)(0,0,1,0);\n"
+	"	return m;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtTranspose(b3Mat3x3 m)\n"
+	"{\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0] = (b3Float4)(m.m_row[0].x, m.m_row[1].x, m.m_row[2].x, 0.f);\n"
+	"	out.m_row[1] = (b3Float4)(m.m_row[0].y, m.m_row[1].y, m.m_row[2].y, 0.f);\n"
+	"	out.m_row[2] = (b3Float4)(m.m_row[0].z, m.m_row[1].z, m.m_row[2].z, 0.f);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b)\n"
+	"{\n"
+	"	b3Mat3x3 transB;\n"
+	"	transB = mtTranspose( b );\n"
+	"	b3Mat3x3 ans;\n"
+	"	//	why this doesn't run when 0ing in the for{}\n"
+	"	a.m_row[0].w = 0.f;\n"
+	"	a.m_row[1].w = 0.f;\n"
+	"	a.m_row[2].w = 0.f;\n"
+	"	for(int i=0; i<3; i++)\n"
+	"	{\n"
+	"//	a.m_row[i].w = 0.f;\n"
+	"		ans.m_row[i].x = b3Dot3F4(a.m_row[i],transB.m_row[0]);\n"
+	"		ans.m_row[i].y = b3Dot3F4(a.m_row[i],transB.m_row[1]);\n"
+	"		ans.m_row[i].z = b3Dot3F4(a.m_row[i],transB.m_row[2]);\n"
+	"		ans.m_row[i].w = 0.f;\n"
+	"	}\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b)\n"
+	"{\n"
+	"	b3Float4 ans;\n"
+	"	ans.x = b3Dot3F4( a.m_row[0], b );\n"
+	"	ans.y = b3Dot3F4( a.m_row[1], b );\n"
+	"	ans.z = b3Dot3F4( a.m_row[2], b );\n"
+	"	ans.w = 0.f;\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b)\n"
+	"{\n"
+	"	b3Float4 colx = b3MakeFloat4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
+	"	b3Float4 coly = b3MakeFloat4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
+	"	b3Float4 colz = b3MakeFloat4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
+	"	b3Float4 ans;\n"
+	"	ans.x = b3Dot3F4( a, colx );\n"
+	"	ans.y = b3Dot3F4( a, coly );\n"
+	"	ans.z = b3Dot3F4( a, colz );\n"
+	"	return ans;\n"
+	"}\n"
+	"#endif\n"
+	"#endif //B3_MAT3x3_H\n"
+	"typedef struct b3RigidBodyData b3RigidBodyData_t;\n"
+	"struct b3RigidBodyData\n"
+	"{\n"
+	"	b3Float4				m_pos;\n"
+	"	b3Quat					m_quat;\n"
+	"	b3Float4				m_linVel;\n"
+	"	b3Float4				m_angVel;\n"
+	"	int 					m_collidableIdx;\n"
+	"	float 				m_invMass;\n"
+	"	float 				m_restituitionCoeff;\n"
+	"	float 				m_frictionCoeff;\n"
+	"};\n"
+	"typedef struct b3InertiaData b3InertiaData_t;\n"
+	"struct b3InertiaData\n"
+	"{\n"
+	"	b3Mat3x3 m_invInertiaWorld;\n"
+	"	b3Mat3x3 m_initInvInertia;\n"
+	"};\n"
+	"#endif //B3_RIGIDBODY_DATA_H\n"
+	"	\n"
+	"#define GET_NPOINTS(x) (x).m_worldNormalOnB.w\n"
+	"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
+	"#define make_float4 (float4)\n"
+	"#define make_float2 (float2)\n"
+	"#define make_uint4 (uint4)\n"
+	"#define make_int4 (int4)\n"
+	"#define make_uint2 (uint2)\n"
+	"#define make_int2 (int2)\n"
+	"__inline\n"
+	"float fastDiv(float numerator, float denominator)\n"
+	"{\n"
+	"	return native_divide(numerator, denominator);	\n"
+	"//	return numerator/denominator;	\n"
+	"}\n"
+	"__inline\n"
+	"float4 fastDiv4(float4 numerator, float4 denominator)\n"
+	"{\n"
+	"	return native_divide(numerator, denominator);	\n"
+	"}\n"
+	"__inline\n"
+	"float4 cross3(float4 a, float4 b)\n"
+	"{\n"
+	"	return cross(a,b);\n"
+	"}\n"
+	"//#define dot3F4 dot\n"
+	"__inline\n"
+	"float dot3F4(float4 a, float4 b)\n"
+	"{\n"
+	"	float4 a1 = make_float4(a.xyz,0.f);\n"
+	"	float4 b1 = make_float4(b.xyz,0.f);\n"
+	"	return dot(a1, b1);\n"
+	"}\n"
+	"__inline\n"
+	"float4 fastNormalize4(float4 v)\n"
+	"{\n"
+	"	return fast_normalize(v);\n"
+	"}\n"
+	"///////////////////////////////////////\n"
+	"//	Quaternion\n"
+	"///////////////////////////////////////\n"
+	"typedef float4 Quaternion;\n"
+	"__inline\n"
+	"Quaternion qtMul(Quaternion a, Quaternion b);\n"
+	"__inline\n"
+	"Quaternion qtNormalize(Quaternion in);\n"
+	"__inline\n"
+	"float4 qtRotate(Quaternion q, float4 vec);\n"
+	"__inline\n"
+	"Quaternion qtInvert(Quaternion q);\n"
+	"__inline\n"
+	"Quaternion qtMul(Quaternion a, Quaternion b)\n"
+	"{\n"
+	"	Quaternion ans;\n"
+	"	ans = cross3( a, b );\n"
+	"	ans += a.w*b+b.w*a;\n"
+	"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
+	"	ans.w = a.w*b.w - dot3F4(a, b);\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"Quaternion qtNormalize(Quaternion in)\n"
+	"{\n"
+	"	return fastNormalize4(in);\n"
+	"//	in /= length( in );\n"
+	"//	return in;\n"
+	"}\n"
+	"__inline\n"
+	"float4 qtRotate(Quaternion q, float4 vec)\n"
+	"{\n"
+	"	Quaternion qInv = qtInvert( q );\n"
+	"	float4 vcpy = vec;\n"
+	"	vcpy.w = 0.f;\n"
+	"	float4 out = qtMul(qtMul(q,vcpy),qInv);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"Quaternion qtInvert(Quaternion q)\n"
+	"{\n"
+	"	return (Quaternion)(-q.xyz, q.w);\n"
+	"}\n"
+	"__inline\n"
+	"float4 qtInvRotate(const Quaternion q, float4 vec)\n"
+	"{\n"
+	"	return qtRotate( qtInvert( q ), vec );\n"
+	"}\n"
+	"__inline\n"
+	"float4 transform(const float4* p, const float4* translation, const Quaternion* orientation)\n"
+	"{\n"
+	"	return qtRotate( *orientation, *p ) + (*translation);\n"
+	"}\n"
+	"__inline\n"
+	"float4 normalize3(const float4 a)\n"
+	"{\n"
+	"	float4 n = make_float4(a.x, a.y, a.z, 0.f);\n"
+	"	return fastNormalize4( n );\n"
+	"}\n"
+	"__inline float4 lerp3(const float4 a,const float4 b, float  t)\n"
+	"{\n"
+	"	return make_float4(	a.x + (b.x - a.x) * t,\n"
+	"						a.y + (b.y - a.y) * t,\n"
+	"						a.z + (b.z - a.z) * t,\n"
+	"						0.f);\n"
+	"}\n"
+	"// Clips a face to the back of a plane, return the number of vertices out, stored in ppVtxOut\n"
+	"int clipFaceGlobal(__global const float4* pVtxIn, int numVertsIn, float4 planeNormalWS,float planeEqWS, __global float4* ppVtxOut)\n"
+	"{\n"
+	"	\n"
+	"	int ve;\n"
+	"	float ds, de;\n"
+	"	int numVertsOut = 0;\n"
+	"    //double-check next test\n"
+	"    	if (numVertsIn < 2)\n"
+	"    		return 0;\n"
+	"    \n"
+	"	float4 firstVertex=pVtxIn[numVertsIn-1];\n"
+	"	float4 endVertex = pVtxIn[0];\n"
+	"	\n"
+	"	ds = dot3F4(planeNormalWS,firstVertex)+planeEqWS;\n"
+	"    \n"
+	"	for (ve = 0; ve < numVertsIn; ve++)\n"
+	"	{\n"
+	"		endVertex=pVtxIn[ve];\n"
+	"		de = dot3F4(planeNormalWS,endVertex)+planeEqWS;\n"
+	"		if (ds<0)\n"
+	"		{\n"
+	"			if (de<0)\n"
+	"			{\n"
+	"				// Start < 0, end < 0, so output endVertex\n"
+	"				ppVtxOut[numVertsOut++] = endVertex;\n"
+	"			}\n"
+	"			else\n"
+	"			{\n"
+	"				// Start < 0, end >= 0, so output intersection\n"
+	"				ppVtxOut[numVertsOut++] = lerp3(firstVertex, endVertex,(ds * 1.f/(ds - de)) );\n"
+	"			}\n"
+	"		}\n"
+	"		else\n"
+	"		{\n"
+	"			if (de<0)\n"
+	"			{\n"
+	"				// Start >= 0, end < 0 so output intersection and end\n"
+	"				ppVtxOut[numVertsOut++] = lerp3(firstVertex, endVertex,(ds * 1.f/(ds - de)) );\n"
+	"				ppVtxOut[numVertsOut++] = endVertex;\n"
+	"			}\n"
+	"		}\n"
+	"		firstVertex = endVertex;\n"
+	"		ds = de;\n"
+	"	}\n"
+	"	return numVertsOut;\n"
+	"}\n"
+	"// Clips a face to the back of a plane, return the number of vertices out, stored in ppVtxOut\n"
+	"int clipFace(const float4* pVtxIn, int numVertsIn, float4 planeNormalWS,float planeEqWS, float4* ppVtxOut)\n"
+	"{\n"
+	"	\n"
+	"	int ve;\n"
+	"	float ds, de;\n"
+	"	int numVertsOut = 0;\n"
+	"//double-check next test\n"
+	"	if (numVertsIn < 2)\n"
+	"		return 0;\n"
+	"	float4 firstVertex=pVtxIn[numVertsIn-1];\n"
+	"	float4 endVertex = pVtxIn[0];\n"
+	"	\n"
+	"	ds = dot3F4(planeNormalWS,firstVertex)+planeEqWS;\n"
+	"	for (ve = 0; ve < numVertsIn; ve++)\n"
+	"	{\n"
+	"		endVertex=pVtxIn[ve];\n"
+	"		de = dot3F4(planeNormalWS,endVertex)+planeEqWS;\n"
+	"		if (ds<0)\n"
+	"		{\n"
+	"			if (de<0)\n"
+	"			{\n"
+	"				// Start < 0, end < 0, so output endVertex\n"
+	"				ppVtxOut[numVertsOut++] = endVertex;\n"
+	"			}\n"
+	"			else\n"
+	"			{\n"
+	"				// Start < 0, end >= 0, so output intersection\n"
+	"				ppVtxOut[numVertsOut++] = lerp3(firstVertex, endVertex,(ds * 1.f/(ds - de)) );\n"
+	"			}\n"
+	"		}\n"
+	"		else\n"
+	"		{\n"
+	"			if (de<0)\n"
+	"			{\n"
+	"				// Start >= 0, end < 0 so output intersection and end\n"
+	"				ppVtxOut[numVertsOut++] = lerp3(firstVertex, endVertex,(ds * 1.f/(ds - de)) );\n"
+	"				ppVtxOut[numVertsOut++] = endVertex;\n"
+	"			}\n"
+	"		}\n"
+	"		firstVertex = endVertex;\n"
+	"		ds = de;\n"
+	"	}\n"
+	"	return numVertsOut;\n"
+	"}\n"
+	"int clipFaceAgainstHull(const float4 separatingNormal, __global const b3ConvexPolyhedronData_t* hullA,  \n"
+	"	const float4 posA, const Quaternion ornA, float4* worldVertsB1, int numWorldVertsB1,\n"
+	"	float4* worldVertsB2, int capacityWorldVertsB2,\n"
+	"	const float minDist, float maxDist,\n"
+	"	__global const float4* vertices,\n"
+	"	__global const b3GpuFace_t* faces,\n"
+	"	__global const int* indices,\n"
+	"	float4* contactsOut,\n"
+	"	int contactCapacity)\n"
+	"{\n"
+	"	int numContactsOut = 0;\n"
+	"	float4* pVtxIn = worldVertsB1;\n"
+	"	float4* pVtxOut = worldVertsB2;\n"
+	"	\n"
+	"	int numVertsIn = numWorldVertsB1;\n"
+	"	int numVertsOut = 0;\n"
+	"	int closestFaceA=-1;\n"
+	"	{\n"
+	"		float dmin = FLT_MAX;\n"
+	"		for(int face=0;face<hullA->m_numFaces;face++)\n"
+	"		{\n"
+	"			const float4 Normal = make_float4(\n"
+	"				faces[hullA->m_faceOffset+face].m_plane.x, \n"
+	"				faces[hullA->m_faceOffset+face].m_plane.y, \n"
+	"				faces[hullA->m_faceOffset+face].m_plane.z,0.f);\n"
+	"			const float4 faceANormalWS = qtRotate(ornA,Normal);\n"
+	"		\n"
+	"			float d = dot3F4(faceANormalWS,separatingNormal);\n"
+	"			if (d < dmin)\n"
+	"			{\n"
+	"				dmin = d;\n"
+	"				closestFaceA = face;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	if (closestFaceA<0)\n"
+	"		return numContactsOut;\n"
+	"	b3GpuFace_t polyA = faces[hullA->m_faceOffset+closestFaceA];\n"
+	"	// clip polygon to back of planes of all faces of hull A that are adjacent to witness face\n"
+	"	int numVerticesA = polyA.m_numIndices;\n"
+	"	for(int e0=0;e0<numVerticesA;e0++)\n"
+	"	{\n"
+	"		const float4 a = vertices[hullA->m_vertexOffset+indices[polyA.m_indexOffset+e0]];\n"
+	"		const float4 b = vertices[hullA->m_vertexOffset+indices[polyA.m_indexOffset+((e0+1)%numVerticesA)]];\n"
+	"		const float4 edge0 = a - b;\n"
+	"		const float4 WorldEdge0 = qtRotate(ornA,edge0);\n"
+	"		float4 planeNormalA = make_float4(polyA.m_plane.x,polyA.m_plane.y,polyA.m_plane.z,0.f);\n"
+	"		float4 worldPlaneAnormal1 = qtRotate(ornA,planeNormalA);\n"
+	"		float4 planeNormalWS1 = -cross3(WorldEdge0,worldPlaneAnormal1);\n"
+	"		float4 worldA1 = transform(&a,&posA,&ornA);\n"
+	"		float planeEqWS1 = -dot3F4(worldA1,planeNormalWS1);\n"
+	"		\n"
+	"		float4 planeNormalWS = planeNormalWS1;\n"
+	"		float planeEqWS=planeEqWS1;\n"
+	"		\n"
+	"		//clip face\n"
+	"		//clipFace(*pVtxIn, *pVtxOut,planeNormalWS,planeEqWS);\n"
+	"		numVertsOut = clipFace(pVtxIn, numVertsIn, planeNormalWS,planeEqWS, pVtxOut);\n"
+	"		//btSwap(pVtxIn,pVtxOut);\n"
+	"		float4* tmp = pVtxOut;\n"
+	"		pVtxOut = pVtxIn;\n"
+	"		pVtxIn = tmp;\n"
+	"		numVertsIn = numVertsOut;\n"
+	"		numVertsOut = 0;\n"
+	"	}\n"
+	"	\n"
+	"	// only keep points that are behind the witness face\n"
+	"	{\n"
+	"		float4 localPlaneNormal  = make_float4(polyA.m_plane.x,polyA.m_plane.y,polyA.m_plane.z,0.f);\n"
+	"		float localPlaneEq = polyA.m_plane.w;\n"
+	"		float4 planeNormalWS = qtRotate(ornA,localPlaneNormal);\n"
+	"		float planeEqWS=localPlaneEq-dot3F4(planeNormalWS,posA);\n"
+	"		for (int i=0;i<numVertsIn;i++)\n"
+	"		{\n"
+	"			float depth = dot3F4(planeNormalWS,pVtxIn[i])+planeEqWS;\n"
+	"			if (depth <=minDist)\n"
+	"			{\n"
+	"				depth = minDist;\n"
+	"			}\n"
+	"			if (depth <=maxDist)\n"
+	"			{\n"
+	"				float4 pointInWorld = pVtxIn[i];\n"
+	"				//resultOut.addContactPoint(separatingNormal,point,depth);\n"
+	"				contactsOut[numContactsOut++] = make_float4(pointInWorld.x,pointInWorld.y,pointInWorld.z,depth);\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	return numContactsOut;\n"
+	"}\n"
+	"int clipFaceAgainstHullLocalA(const float4 separatingNormal, const b3ConvexPolyhedronData_t* hullA,  \n"
+	"	const float4 posA, const Quaternion ornA, float4* worldVertsB1, int numWorldVertsB1,\n"
+	"	float4* worldVertsB2, int capacityWorldVertsB2,\n"
+	"	const float minDist, float maxDist,\n"
+	"	const float4* verticesA,\n"
+	"	const b3GpuFace_t* facesA,\n"
+	"	const int* indicesA,\n"
+	"	__global const float4* verticesB,\n"
+	"	__global const b3GpuFace_t* facesB,\n"
+	"	__global const int* indicesB,\n"
+	"	float4* contactsOut,\n"
+	"	int contactCapacity)\n"
+	"{\n"
+	"	int numContactsOut = 0;\n"
+	"	float4* pVtxIn = worldVertsB1;\n"
+	"	float4* pVtxOut = worldVertsB2;\n"
+	"	\n"
+	"	int numVertsIn = numWorldVertsB1;\n"
+	"	int numVertsOut = 0;\n"
+	"	int closestFaceA=-1;\n"
+	"	{\n"
+	"		float dmin = FLT_MAX;\n"
+	"		for(int face=0;face<hullA->m_numFaces;face++)\n"
+	"		{\n"
+	"			const float4 Normal = make_float4(\n"
+	"				facesA[hullA->m_faceOffset+face].m_plane.x, \n"
+	"				facesA[hullA->m_faceOffset+face].m_plane.y, \n"
+	"				facesA[hullA->m_faceOffset+face].m_plane.z,0.f);\n"
+	"			const float4 faceANormalWS = qtRotate(ornA,Normal);\n"
+	"		\n"
+	"			float d = dot3F4(faceANormalWS,separatingNormal);\n"
+	"			if (d < dmin)\n"
+	"			{\n"
+	"				dmin = d;\n"
+	"				closestFaceA = face;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	if (closestFaceA<0)\n"
+	"		return numContactsOut;\n"
+	"	b3GpuFace_t polyA = facesA[hullA->m_faceOffset+closestFaceA];\n"
+	"	// clip polygon to back of planes of all faces of hull A that are adjacent to witness face\n"
+	"	int numVerticesA = polyA.m_numIndices;\n"
+	"	for(int e0=0;e0<numVerticesA;e0++)\n"
+	"	{\n"
+	"		const float4 a = verticesA[hullA->m_vertexOffset+indicesA[polyA.m_indexOffset+e0]];\n"
+	"		const float4 b = verticesA[hullA->m_vertexOffset+indicesA[polyA.m_indexOffset+((e0+1)%numVerticesA)]];\n"
+	"		const float4 edge0 = a - b;\n"
+	"		const float4 WorldEdge0 = qtRotate(ornA,edge0);\n"
+	"		float4 planeNormalA = make_float4(polyA.m_plane.x,polyA.m_plane.y,polyA.m_plane.z,0.f);\n"
+	"		float4 worldPlaneAnormal1 = qtRotate(ornA,planeNormalA);\n"
+	"		float4 planeNormalWS1 = -cross3(WorldEdge0,worldPlaneAnormal1);\n"
+	"		float4 worldA1 = transform(&a,&posA,&ornA);\n"
+	"		float planeEqWS1 = -dot3F4(worldA1,planeNormalWS1);\n"
+	"		\n"
+	"		float4 planeNormalWS = planeNormalWS1;\n"
+	"		float planeEqWS=planeEqWS1;\n"
+	"		\n"
+	"		//clip face\n"
+	"		//clipFace(*pVtxIn, *pVtxOut,planeNormalWS,planeEqWS);\n"
+	"		numVertsOut = clipFace(pVtxIn, numVertsIn, planeNormalWS,planeEqWS, pVtxOut);\n"
+	"		//btSwap(pVtxIn,pVtxOut);\n"
+	"		float4* tmp = pVtxOut;\n"
+	"		pVtxOut = pVtxIn;\n"
+	"		pVtxIn = tmp;\n"
+	"		numVertsIn = numVertsOut;\n"
+	"		numVertsOut = 0;\n"
+	"	}\n"
+	"	\n"
+	"	// only keep points that are behind the witness face\n"
+	"	{\n"
+	"		float4 localPlaneNormal  = make_float4(polyA.m_plane.x,polyA.m_plane.y,polyA.m_plane.z,0.f);\n"
+	"		float localPlaneEq = polyA.m_plane.w;\n"
+	"		float4 planeNormalWS = qtRotate(ornA,localPlaneNormal);\n"
+	"		float planeEqWS=localPlaneEq-dot3F4(planeNormalWS,posA);\n"
+	"		for (int i=0;i<numVertsIn;i++)\n"
+	"		{\n"
+	"			float depth = dot3F4(planeNormalWS,pVtxIn[i])+planeEqWS;\n"
+	"			if (depth <=minDist)\n"
+	"			{\n"
+	"				depth = minDist;\n"
+	"			}\n"
+	"			if (depth <=maxDist)\n"
+	"			{\n"
+	"				float4 pointInWorld = pVtxIn[i];\n"
+	"				//resultOut.addContactPoint(separatingNormal,point,depth);\n"
+	"				contactsOut[numContactsOut++] = make_float4(pointInWorld.x,pointInWorld.y,pointInWorld.z,depth);\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	return numContactsOut;\n"
+	"}\n"
+	"int	clipHullAgainstHull(const float4 separatingNormal,\n"
+	"	__global const b3ConvexPolyhedronData_t* hullA, __global const b3ConvexPolyhedronData_t* hullB, \n"
+	"	const float4 posA, const Quaternion ornA,const float4 posB, const Quaternion ornB, \n"
+	"	float4* worldVertsB1, float4* worldVertsB2, int capacityWorldVerts,\n"
+	"	const float minDist, float maxDist,\n"
+	"	__global const float4* vertices,\n"
+	"	__global const b3GpuFace_t* faces,\n"
+	"	__global const int* indices,\n"
+	"	float4*	localContactsOut,\n"
+	"	int localContactCapacity)\n"
+	"{\n"
+	"	int numContactsOut = 0;\n"
+	"	int numWorldVertsB1= 0;\n"
+	"	int closestFaceB=-1;\n"
+	"	float dmax = -FLT_MAX;\n"
+	"	{\n"
+	"		for(int face=0;face<hullB->m_numFaces;face++)\n"
+	"		{\n"
+	"			const float4 Normal = make_float4(faces[hullB->m_faceOffset+face].m_plane.x, \n"
+	"				faces[hullB->m_faceOffset+face].m_plane.y, faces[hullB->m_faceOffset+face].m_plane.z,0.f);\n"
+	"			const float4 WorldNormal = qtRotate(ornB, Normal);\n"
+	"			float d = dot3F4(WorldNormal,separatingNormal);\n"
+	"			if (d > dmax)\n"
+	"			{\n"
+	"				dmax = d;\n"
+	"				closestFaceB = face;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	{\n"
+	"		const b3GpuFace_t polyB = faces[hullB->m_faceOffset+closestFaceB];\n"
+	"		const int numVertices = polyB.m_numIndices;\n"
+	"		for(int e0=0;e0<numVertices;e0++)\n"
+	"		{\n"
+	"			const float4 b = vertices[hullB->m_vertexOffset+indices[polyB.m_indexOffset+e0]];\n"
+	"			worldVertsB1[numWorldVertsB1++] = transform(&b,&posB,&ornB);\n"
+	"		}\n"
+	"	}\n"
+	"	if (closestFaceB>=0)\n"
+	"	{\n"
+	"		numContactsOut = clipFaceAgainstHull(separatingNormal, hullA, \n"
+	"				posA,ornA,\n"
+	"				worldVertsB1,numWorldVertsB1,worldVertsB2,capacityWorldVerts, minDist, maxDist,vertices,\n"
+	"				faces,\n"
+	"				indices,localContactsOut,localContactCapacity);\n"
+	"	}\n"
+	"	return numContactsOut;\n"
+	"}\n"
+	"int	clipHullAgainstHullLocalA(const float4 separatingNormal,\n"
+	"	const b3ConvexPolyhedronData_t* hullA, __global const b3ConvexPolyhedronData_t* hullB, \n"
+	"	const float4 posA, const Quaternion ornA,const float4 posB, const Quaternion ornB, \n"
+	"	float4* worldVertsB1, float4* worldVertsB2, int capacityWorldVerts,\n"
+	"	const float minDist, float maxDist,\n"
+	"	const float4* verticesA,\n"
+	"	const b3GpuFace_t* facesA,\n"
+	"	const int* indicesA,\n"
+	"	__global const float4* verticesB,\n"
+	"	__global const b3GpuFace_t* facesB,\n"
+	"	__global const int* indicesB,\n"
+	"	float4*	localContactsOut,\n"
+	"	int localContactCapacity)\n"
+	"{\n"
+	"	int numContactsOut = 0;\n"
+	"	int numWorldVertsB1= 0;\n"
+	"	int closestFaceB=-1;\n"
+	"	float dmax = -FLT_MAX;\n"
+	"	{\n"
+	"		for(int face=0;face<hullB->m_numFaces;face++)\n"
+	"		{\n"
+	"			const float4 Normal = make_float4(facesB[hullB->m_faceOffset+face].m_plane.x, \n"
+	"				facesB[hullB->m_faceOffset+face].m_plane.y, facesB[hullB->m_faceOffset+face].m_plane.z,0.f);\n"
+	"			const float4 WorldNormal = qtRotate(ornB, Normal);\n"
+	"			float d = dot3F4(WorldNormal,separatingNormal);\n"
+	"			if (d > dmax)\n"
+	"			{\n"
+	"				dmax = d;\n"
+	"				closestFaceB = face;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	{\n"
+	"		const b3GpuFace_t polyB = facesB[hullB->m_faceOffset+closestFaceB];\n"
+	"		const int numVertices = polyB.m_numIndices;\n"
+	"		for(int e0=0;e0<numVertices;e0++)\n"
+	"		{\n"
+	"			const float4 b = verticesB[hullB->m_vertexOffset+indicesB[polyB.m_indexOffset+e0]];\n"
+	"			worldVertsB1[numWorldVertsB1++] = transform(&b,&posB,&ornB);\n"
+	"		}\n"
+	"	}\n"
+	"	if (closestFaceB>=0)\n"
+	"	{\n"
+	"		numContactsOut = clipFaceAgainstHullLocalA(separatingNormal, hullA, \n"
+	"				posA,ornA,\n"
+	"				worldVertsB1,numWorldVertsB1,worldVertsB2,capacityWorldVerts, minDist, maxDist,\n"
+	"				verticesA,facesA,indicesA,\n"
+	"				verticesB,facesB,indicesB,\n"
+	"				localContactsOut,localContactCapacity);\n"
+	"	}\n"
+	"	return numContactsOut;\n"
+	"}\n"
+	"#define PARALLEL_SUM(v, n) for(int j=1; j<n; j++) v[0] += v[j];\n"
+	"#define PARALLEL_DO(execution, n) for(int ie=0; ie<n; ie++){execution;}\n"
+	"#define REDUCE_MAX(v, n) {int i=0; for(int offset=0; offset<n; offset++) v[i] = (v[i].y > v[i+offset].y)? v[i]: v[i+offset]; }\n"
+	"#define REDUCE_MIN(v, n) {int i=0; for(int offset=0; offset<n; offset++) v[i] = (v[i].y < v[i+offset].y)? v[i]: v[i+offset]; }\n"
+	"int extractManifoldSequentialGlobal(__global const float4* p, int nPoints, float4 nearNormal, int4* contactIdx)\n"
+	"{\n"
+	"	if( nPoints == 0 )\n"
+	"        return 0;\n"
+	"    \n"
+	"    if (nPoints <=4)\n"
+	"        return nPoints;\n"
+	"    \n"
+	"    \n"
+	"    if (nPoints >64)\n"
+	"        nPoints = 64;\n"
+	"    \n"
+	"	float4 center = make_float4(0.f);\n"
+	"	{\n"
+	"		\n"
+	"		for (int i=0;i<nPoints;i++)\n"
+	"			center += p[i];\n"
+	"		center /= (float)nPoints;\n"
+	"	}\n"
+	"    \n"
+	"	\n"
+	"    \n"
+	"	//	sample 4 directions\n"
+	"    \n"
+	"    float4 aVector = p[0] - center;\n"
+	"    float4 u = cross3( nearNormal, aVector );\n"
+	"    float4 v = cross3( nearNormal, u );\n"
+	"    u = normalize3( u );\n"
+	"    v = normalize3( v );\n"
+	"    \n"
+	"    \n"
+	"    //keep point with deepest penetration\n"
+	"    float minW= FLT_MAX;\n"
+	"    \n"
+	"    int minIndex=-1;\n"
+	"    \n"
+	"    float4 maxDots;\n"
+	"    maxDots.x = FLT_MIN;\n"
+	"    maxDots.y = FLT_MIN;\n"
+	"    maxDots.z = FLT_MIN;\n"
+	"    maxDots.w = FLT_MIN;\n"
+	"    \n"
+	"    //	idx, distance\n"
+	"    for(int ie = 0; ie<nPoints; ie++ )\n"
+	"    {\n"
+	"        if (p[ie].w<minW)\n"
+	"        {\n"
+	"            minW = p[ie].w;\n"
+	"            minIndex=ie;\n"
+	"        }\n"
+	"        float f;\n"
+	"        float4 r = p[ie]-center;\n"
+	"        f = dot3F4( u, r );\n"
+	"        if (f<maxDots.x)\n"
+	"        {\n"
+	"            maxDots.x = f;\n"
+	"            contactIdx[0].x = ie;\n"
+	"        }\n"
+	"        \n"
+	"        f = dot3F4( -u, r );\n"
+	"        if (f<maxDots.y)\n"
+	"        {\n"
+	"            maxDots.y = f;\n"
+	"            contactIdx[0].y = ie;\n"
+	"        }\n"
+	"        \n"
+	"        \n"
+	"        f = dot3F4( v, r );\n"
+	"        if (f<maxDots.z)\n"
+	"        {\n"
+	"            maxDots.z = f;\n"
+	"            contactIdx[0].z = ie;\n"
+	"        }\n"
+	"        \n"
+	"        f = dot3F4( -v, r );\n"
+	"        if (f<maxDots.w)\n"
+	"        {\n"
+	"            maxDots.w = f;\n"
+	"            contactIdx[0].w = ie;\n"
+	"        }\n"
+	"        \n"
+	"    }\n"
+	"    \n"
+	"    if (contactIdx[0].x != minIndex && contactIdx[0].y != minIndex && contactIdx[0].z != minIndex && contactIdx[0].w != minIndex)\n"
+	"    {\n"
+	"        //replace the first contact with minimum (todo: replace contact with least penetration)\n"
+	"        contactIdx[0].x = minIndex;\n"
+	"    }\n"
+	"    \n"
+	"    return 4;\n"
+	"    \n"
+	"}\n"
+	"int extractManifoldSequentialGlobalFake(__global const float4* p, int nPoints, float4 nearNormal, int* contactIdx)\n"
+	"{\n"
+	"    contactIdx[0] = 0;\n"
+	"    contactIdx[1] = 1;\n"
+	"    contactIdx[2] = 2;\n"
+	"    contactIdx[3] = 3;\n"
+	"    \n"
+	"	if( nPoints == 0 ) return 0;\n"
+	"    \n"
+	"	nPoints = min2( nPoints, 4 );\n"
+	"    return nPoints;\n"
+	"    \n"
+	"}\n"
+	"int extractManifoldSequential(const float4* p, int nPoints, float4 nearNormal, int* contactIdx)\n"
+	"{\n"
+	"	if( nPoints == 0 ) return 0;\n"
+	"	nPoints = min2( nPoints, 64 );\n"
+	"	float4 center = make_float4(0.f);\n"
+	"	{\n"
+	"		float4 v[64];\n"
+	"		for (int i=0;i<nPoints;i++)\n"
+	"			v[i] = p[i];\n"
+	"		//memcpy( v, p, nPoints*sizeof(float4) );\n"
+	"		PARALLEL_SUM( v, nPoints );\n"
+	"		center = v[0]/(float)nPoints;\n"
+	"	}\n"
+	"	\n"
+	"	{	//	sample 4 directions\n"
+	"		if( nPoints < 4 )\n"
+	"		{\n"
+	"			for(int i=0; i<nPoints; i++) \n"
+	"				contactIdx[i] = i;\n"
+	"			return nPoints;\n"
+	"		}\n"
+	"		float4 aVector = p[0] - center;\n"
+	"		float4 u = cross3( nearNormal, aVector );\n"
+	"		float4 v = cross3( nearNormal, u );\n"
+	"		u = normalize3( u );\n"
+	"		v = normalize3( v );\n"
+	"		int idx[4];\n"
+	"		float2 max00 = make_float2(0,FLT_MAX);\n"
+	"		{\n"
+	"			//	idx, distance\n"
+	"			{\n"
+	"				{\n"
+	"					int4 a[64];\n"
+	"					for(int ie = 0; ie<nPoints; ie++ )\n"
+	"					{\n"
+	"						\n"
+	"						\n"
+	"						float f;\n"
+	"						float4 r = p[ie]-center;\n"
+	"						f = dot3F4( u, r );\n"
+	"						a[ie].x = ((*(u32*)&f) & 0xffffff00) | (0xff & ie);\n"
+	"						f = dot3F4( -u, r );\n"
+	"						a[ie].y = ((*(u32*)&f) & 0xffffff00) | (0xff & ie);\n"
+	"						f = dot3F4( v, r );\n"
+	"						a[ie].z = ((*(u32*)&f) & 0xffffff00) | (0xff & ie);\n"
+	"						f = dot3F4( -v, r );\n"
+	"						a[ie].w = ((*(u32*)&f) & 0xffffff00) | (0xff & ie);\n"
+	"					}\n"
+	"					for(int ie=0; ie<nPoints; ie++)\n"
+	"					{\n"
+	"						a[0].x = (a[0].x > a[ie].x )? a[0].x: a[ie].x;\n"
+	"						a[0].y = (a[0].y > a[ie].y )? a[0].y: a[ie].y;\n"
+	"						a[0].z = (a[0].z > a[ie].z )? a[0].z: a[ie].z;\n"
+	"						a[0].w = (a[0].w > a[ie].w )? a[0].w: a[ie].w;\n"
+	"					}\n"
+	"					idx[0] = (int)a[0].x & 0xff;\n"
+	"					idx[1] = (int)a[0].y & 0xff;\n"
+	"					idx[2] = (int)a[0].z & 0xff;\n"
+	"					idx[3] = (int)a[0].w & 0xff;\n"
+	"				}\n"
+	"			}\n"
+	"			{\n"
+	"				float2 h[64];\n"
+	"				PARALLEL_DO( h[ie] = make_float2((float)ie, p[ie].w), nPoints );\n"
+	"				REDUCE_MIN( h, nPoints );\n"
+	"				max00 = h[0];\n"
+	"			}\n"
+	"		}\n"
+	"		contactIdx[0] = idx[0];\n"
+	"		contactIdx[1] = idx[1];\n"
+	"		contactIdx[2] = idx[2];\n"
+	"		contactIdx[3] = idx[3];\n"
+	"		return 4;\n"
+	"	}\n"
+	"}\n"
+	"__kernel void   extractManifoldAndAddContactKernel(__global const int4* pairs, \n"
+	"																	__global const b3RigidBodyData_t* rigidBodies, \n"
+	"																	__global const float4* closestPointsWorld,\n"
+	"																	__global const float4* separatingNormalsWorld,\n"
+	"																	__global const int* contactCounts,\n"
+	"																	__global const int* contactOffsets,\n"
+	"																	__global struct b3Contact4Data* restrict contactsOut,\n"
+	"																	counter32_t nContactsOut,\n"
+	"																	int contactCapacity,\n"
+	"																	int numPairs,\n"
+	"																	int pairIndex\n"
+	"																	)\n"
+	"{\n"
+	"	int idx = get_global_id(0);\n"
+	"	\n"
+	"	if (idx<numPairs)\n"
+	"	{\n"
+	"		float4 normal = separatingNormalsWorld[idx];\n"
+	"		int nPoints = contactCounts[idx];\n"
+	"		__global const float4* pointsIn = &closestPointsWorld[contactOffsets[idx]];\n"
+	"		float4 localPoints[64];\n"
+	"		for (int i=0;i<nPoints;i++)\n"
+	"		{\n"
+	"			localPoints[i] = pointsIn[i];\n"
+	"		}\n"
+	"		int contactIdx[4];// = {-1,-1,-1,-1};\n"
+	"		contactIdx[0] = -1;\n"
+	"		contactIdx[1] = -1;\n"
+	"		contactIdx[2] = -1;\n"
+	"		contactIdx[3] = -1;\n"
+	"		int nContacts = extractManifoldSequential(localPoints, nPoints, normal, contactIdx);\n"
+	"		int dstIdx;\n"
+	"		AppendInc( nContactsOut, dstIdx );\n"
+	"		if (dstIdx<contactCapacity)\n"
+	"		{\n"
+	"			__global struct b3Contact4Data* c = contactsOut + dstIdx;\n"
+	"			c->m_worldNormalOnB = -normal;\n"
+	"			c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
+	"			c->m_batchIdx = idx;\n"
+	"			int bodyA = pairs[pairIndex].x;\n"
+	"			int bodyB = pairs[pairIndex].y;\n"
+	"			c->m_bodyAPtrAndSignBit = rigidBodies[bodyA].m_invMass==0 ? -bodyA:bodyA;\n"
+	"			c->m_bodyBPtrAndSignBit = rigidBodies[bodyB].m_invMass==0 ? -bodyB:bodyB;\n"
+	"			c->m_childIndexA = -1;\n"
+	"			c->m_childIndexB = -1;\n"
+	"			for (int i=0;i<nContacts;i++)\n"
+	"			{\n"
+	"				c->m_worldPosB[i] = localPoints[contactIdx[i]];\n"
+	"			}\n"
+	"			GET_NPOINTS(*c) = nContacts;\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"void	trInverse(float4 translationIn, Quaternion orientationIn,\n"
+	"		float4* translationOut, Quaternion* orientationOut)\n"
+	"{\n"
+	"	*orientationOut = qtInvert(orientationIn);\n"
+	"	*translationOut = qtRotate(*orientationOut, -translationIn);\n"
+	"}\n"
+	"void	trMul(float4 translationA, Quaternion orientationA,\n"
+	"						float4 translationB, Quaternion orientationB,\n"
+	"		float4* translationOut, Quaternion* orientationOut)\n"
+	"{\n"
+	"	*orientationOut = qtMul(orientationA,orientationB);\n"
+	"	*translationOut = transform(&translationB,&translationA,&orientationA);\n"
+	"}\n"
+	"__kernel void   clipHullHullKernel( __global int4* pairs, \n"
+	"																					__global const b3RigidBodyData_t* rigidBodies, \n"
+	"																					__global const b3Collidable_t* collidables,\n"
+	"																					__global const b3ConvexPolyhedronData_t* convexShapes, \n"
+	"																					__global const float4* vertices,\n"
+	"																					__global const float4* uniqueEdges,\n"
+	"																					__global const b3GpuFace_t* faces,\n"
+	"																					__global const int* indices,\n"
+	"																					__global const float4* separatingNormals,\n"
+	"																					__global const int* hasSeparatingAxis,\n"
+	"																					__global struct b3Contact4Data* restrict globalContactsOut,\n"
+	"																					counter32_t nGlobalContactsOut,\n"
+	"																					int numPairs,\n"
+	"																					int contactCapacity)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	int pairIndex = i;\n"
+	"	\n"
+	"	float4 worldVertsB1[64];\n"
+	"	float4 worldVertsB2[64];\n"
+	"	int capacityWorldVerts = 64;	\n"
+	"	float4 localContactsOut[64];\n"
+	"	int localContactCapacity=64;\n"
+	"	\n"
+	"	float minDist = -1e30f;\n"
+	"	float maxDist = 0.02f;\n"
+	"	if (i<numPairs)\n"
+	"	{\n"
+	"		int bodyIndexA = pairs[i].x;\n"
+	"		int bodyIndexB = pairs[i].y;\n"
+	"			\n"
+	"		int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
+	"		int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
+	"		if (hasSeparatingAxis[i])\n"
+	"		{\n"
+	"			\n"
+	"			int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
+	"			int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
+	"			\n"
+	"		\n"
+	"			int numLocalContactsOut = clipHullAgainstHull(separatingNormals[i],\n"
+	"														&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],\n"
+	"														rigidBodies[bodyIndexA].m_pos,rigidBodies[bodyIndexA].m_quat,\n"
+	"													  rigidBodies[bodyIndexB].m_pos,rigidBodies[bodyIndexB].m_quat,\n"
+	"													  worldVertsB1,worldVertsB2,capacityWorldVerts,\n"
+	"														minDist, maxDist,\n"
+	"														vertices,faces,indices,\n"
+	"														localContactsOut,localContactCapacity);\n"
+	"												\n"
+	"		if (numLocalContactsOut>0)\n"
+	"		{\n"
+	"				float4 normal = -separatingNormals[i];\n"
+	"				int nPoints = numLocalContactsOut;\n"
+	"				float4* pointsIn = localContactsOut;\n"
+	"				int contactIdx[4];// = {-1,-1,-1,-1};\n"
+	"				contactIdx[0] = -1;\n"
+	"				contactIdx[1] = -1;\n"
+	"				contactIdx[2] = -1;\n"
+	"				contactIdx[3] = -1;\n"
+	"		\n"
+	"				int nReducedContacts = extractManifoldSequential(pointsIn, nPoints, normal, contactIdx);\n"
+	"		\n"
+	"				\n"
+	"				int mprContactIndex = pairs[pairIndex].z;\n"
+	"				int dstIdx = mprContactIndex;\n"
+	"				if (dstIdx<0)\n"
+	"				{\n"
+	"					AppendInc( nGlobalContactsOut, dstIdx );\n"
+	"				}\n"
+	"				if (dstIdx<contactCapacity)\n"
+	"				{\n"
+	"					pairs[pairIndex].z = dstIdx;\n"
+	"					__global struct b3Contact4Data* c = globalContactsOut+ dstIdx;\n"
+	"					c->m_worldNormalOnB = -normal;\n"
+	"					c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
+	"					c->m_batchIdx = pairIndex;\n"
+	"					int bodyA = pairs[pairIndex].x;\n"
+	"					int bodyB = pairs[pairIndex].y;\n"
+	"					c->m_bodyAPtrAndSignBit = rigidBodies[bodyA].m_invMass==0?-bodyA:bodyA;\n"
+	"					c->m_bodyBPtrAndSignBit = rigidBodies[bodyB].m_invMass==0?-bodyB:bodyB;\n"
+	"					c->m_childIndexA = -1;\n"
+	"					c->m_childIndexB = -1;\n"
+	"					for (int i=0;i<nReducedContacts;i++)\n"
+	"					{\n"
+	"					//this condition means: overwrite contact point, unless at index i==0 we have a valid 'mpr' contact\n"
+	"						if (i>0||(mprContactIndex<0))\n"
+	"						{\n"
+	"							c->m_worldPosB[i] = pointsIn[contactIdx[i]];\n"
+	"						}\n"
+	"					}\n"
+	"					GET_NPOINTS(*c) = nReducedContacts;\n"
+	"				}\n"
+	"				\n"
+	"			}//		if (numContactsOut>0)\n"
+	"		}//		if (hasSeparatingAxis[i])\n"
+	"	}//	if (i<numPairs)\n"
+	"}\n"
+	"__kernel void   clipCompoundsHullHullKernel( __global const int4* gpuCompoundPairs, \n"
+	"																					__global const b3RigidBodyData_t* rigidBodies, \n"
+	"																					__global const b3Collidable_t* collidables,\n"
+	"																					__global const b3ConvexPolyhedronData_t* convexShapes, \n"
+	"																					__global const float4* vertices,\n"
+	"																					__global const float4* uniqueEdges,\n"
+	"																					__global const b3GpuFace_t* faces,\n"
+	"																					__global const int* indices,\n"
+	"																					__global const b3GpuChildShape_t* gpuChildShapes,\n"
+	"																					__global const float4* gpuCompoundSepNormalsOut,\n"
+	"																					__global const int* gpuHasCompoundSepNormalsOut,\n"
+	"																					__global struct b3Contact4Data* restrict globalContactsOut,\n"
+	"																					counter32_t nGlobalContactsOut,\n"
+	"																					int numCompoundPairs, int maxContactCapacity)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	int pairIndex = i;\n"
+	"	\n"
+	"	float4 worldVertsB1[64];\n"
+	"	float4 worldVertsB2[64];\n"
+	"	int capacityWorldVerts = 64;	\n"
+	"	float4 localContactsOut[64];\n"
+	"	int localContactCapacity=64;\n"
+	"	\n"
+	"	float minDist = -1e30f;\n"
+	"	float maxDist = 0.02f;\n"
+	"	if (i<numCompoundPairs)\n"
+	"	{\n"
+	"		if (gpuHasCompoundSepNormalsOut[i])\n"
+	"		{\n"
+	"			int bodyIndexA = gpuCompoundPairs[i].x;\n"
+	"			int bodyIndexB = gpuCompoundPairs[i].y;\n"
+	"			\n"
+	"			int childShapeIndexA = gpuCompoundPairs[i].z;\n"
+	"			int childShapeIndexB = gpuCompoundPairs[i].w;\n"
+	"			\n"
+	"			int collidableIndexA = -1;\n"
+	"			int collidableIndexB = -1;\n"
+	"			\n"
+	"			float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
+	"			float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
+	"			\n"
+	"			float4 ornB = rigidBodies[bodyIndexB].m_quat;\n"
+	"			float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
+	"								\n"
+	"			if (childShapeIndexA >= 0)\n"
+	"			{\n"
+	"				collidableIndexA = gpuChildShapes[childShapeIndexA].m_shapeIndex;\n"
+	"				float4 childPosA = gpuChildShapes[childShapeIndexA].m_childPosition;\n"
+	"				float4 childOrnA = gpuChildShapes[childShapeIndexA].m_childOrientation;\n"
+	"				float4 newPosA = qtRotate(ornA,childPosA)+posA;\n"
+	"				float4 newOrnA = qtMul(ornA,childOrnA);\n"
+	"				posA = newPosA;\n"
+	"				ornA = newOrnA;\n"
+	"			} else\n"
+	"			{\n"
+	"				collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
+	"			}\n"
+	"			\n"
+	"			if (childShapeIndexB>=0)\n"
+	"			{\n"
+	"				collidableIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;\n"
+	"				float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;\n"
+	"				float4 childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;\n"
+	"				float4 newPosB = transform(&childPosB,&posB,&ornB);\n"
+	"				float4 newOrnB = qtMul(ornB,childOrnB);\n"
+	"				posB = newPosB;\n"
+	"				ornB = newOrnB;\n"
+	"			} else\n"
+	"			{\n"
+	"				collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;	\n"
+	"			}\n"
+	"			\n"
+	"			int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
+	"			int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
+	"		\n"
+	"			int numLocalContactsOut = clipHullAgainstHull(gpuCompoundSepNormalsOut[i],\n"
+	"														&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],\n"
+	"														posA,ornA,\n"
+	"													  posB,ornB,\n"
+	"													  worldVertsB1,worldVertsB2,capacityWorldVerts,\n"
+	"														minDist, maxDist,\n"
+	"														vertices,faces,indices,\n"
+	"														localContactsOut,localContactCapacity);\n"
+	"												\n"
+	"		if (numLocalContactsOut>0)\n"
+	"		{\n"
+	"				float4 normal = -gpuCompoundSepNormalsOut[i];\n"
+	"				int nPoints = numLocalContactsOut;\n"
+	"				float4* pointsIn = localContactsOut;\n"
+	"				int contactIdx[4];// = {-1,-1,-1,-1};\n"
+	"				contactIdx[0] = -1;\n"
+	"				contactIdx[1] = -1;\n"
+	"				contactIdx[2] = -1;\n"
+	"				contactIdx[3] = -1;\n"
+	"		\n"
+	"				int nReducedContacts = extractManifoldSequential(pointsIn, nPoints, normal, contactIdx);\n"
+	"		\n"
+	"				int dstIdx;\n"
+	"				AppendInc( nGlobalContactsOut, dstIdx );\n"
+	"				if ((dstIdx+nReducedContacts) < maxContactCapacity)\n"
+	"				{\n"
+	"					__global struct b3Contact4Data* c = globalContactsOut+ dstIdx;\n"
+	"					c->m_worldNormalOnB = -normal;\n"
+	"					c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
+	"					c->m_batchIdx = pairIndex;\n"
+	"					int bodyA = gpuCompoundPairs[pairIndex].x;\n"
+	"					int bodyB = gpuCompoundPairs[pairIndex].y;\n"
+	"					c->m_bodyAPtrAndSignBit = rigidBodies[bodyA].m_invMass==0?-bodyA:bodyA;\n"
+	"					c->m_bodyBPtrAndSignBit = rigidBodies[bodyB].m_invMass==0?-bodyB:bodyB;\n"
+	"					c->m_childIndexA = childShapeIndexA;\n"
+	"					c->m_childIndexB = childShapeIndexB;\n"
+	"					for (int i=0;i<nReducedContacts;i++)\n"
+	"					{\n"
+	"						c->m_worldPosB[i] = pointsIn[contactIdx[i]];\n"
+	"					}\n"
+	"					GET_NPOINTS(*c) = nReducedContacts;\n"
+	"				}\n"
+	"				\n"
+	"			}//		if (numContactsOut>0)\n"
+	"		}//		if (gpuHasCompoundSepNormalsOut[i])\n"
+	"	}//	if (i<numCompoundPairs)\n"
+	"}\n"
+	"__kernel void   sphereSphereCollisionKernel( __global const int4* pairs, \n"
+	"																					__global const b3RigidBodyData_t* rigidBodies, \n"
+	"																					__global const b3Collidable_t* collidables,\n"
+	"																					__global const float4* separatingNormals,\n"
+	"																					__global const int* hasSeparatingAxis,\n"
+	"																					__global struct b3Contact4Data* restrict globalContactsOut,\n"
+	"																					counter32_t nGlobalContactsOut,\n"
+	"																					int contactCapacity,\n"
+	"																					int numPairs)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	int pairIndex = i;\n"
+	"	\n"
+	"	if (i<numPairs)\n"
+	"	{\n"
+	"		int bodyIndexA = pairs[i].x;\n"
+	"		int bodyIndexB = pairs[i].y;\n"
+	"			\n"
+	"		int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
+	"		int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
+	"		if (collidables[collidableIndexA].m_shapeType == SHAPE_SPHERE &&\n"
+	"			collidables[collidableIndexB].m_shapeType == SHAPE_SPHERE)\n"
+	"		{\n"
+	"			//sphere-sphere\n"
+	"			float radiusA = collidables[collidableIndexA].m_radius;\n"
+	"			float radiusB = collidables[collidableIndexB].m_radius;\n"
+	"			float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
+	"			float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
+	"			float4 diff = posA-posB;\n"
+	"			float len = length(diff);\n"
+	"			\n"
+	"			///iff distance positive, don't generate a new contact\n"
+	"			if ( len <= (radiusA+radiusB))\n"
+	"			{\n"
+	"				///distance (negative means penetration)\n"
+	"				float dist = len - (radiusA+radiusB);\n"
+	"				float4 normalOnSurfaceB = make_float4(1.f,0.f,0.f,0.f);\n"
+	"				if (len > 0.00001)\n"
+	"				{\n"
+	"					normalOnSurfaceB = diff / len;\n"
+	"				}\n"
+	"				float4 contactPosB = posB + normalOnSurfaceB*radiusB;\n"
+	"				contactPosB.w = dist;\n"
+	"								\n"
+	"				int dstIdx;\n"
+	"				AppendInc( nGlobalContactsOut, dstIdx );\n"
+	"				if (dstIdx < contactCapacity)\n"
+	"				{\n"
+	"					__global struct b3Contact4Data* c = &globalContactsOut[dstIdx];\n"
+	"					c->m_worldNormalOnB = -normalOnSurfaceB;\n"
+	"					c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
+	"					c->m_batchIdx = pairIndex;\n"
+	"					int bodyA = pairs[pairIndex].x;\n"
+	"					int bodyB = pairs[pairIndex].y;\n"
+	"					c->m_bodyAPtrAndSignBit = rigidBodies[bodyA].m_invMass==0?-bodyA:bodyA;\n"
+	"					c->m_bodyBPtrAndSignBit = rigidBodies[bodyB].m_invMass==0?-bodyB:bodyB;\n"
+	"					c->m_worldPosB[0] = contactPosB;\n"
+	"					c->m_childIndexA = -1;\n"
+	"					c->m_childIndexB = -1;\n"
+	"					GET_NPOINTS(*c) = 1;\n"
+	"				}//if (dstIdx < numPairs)\n"
+	"			}//if ( len <= (radiusA+radiusB))\n"
+	"		}//SHAPE_SPHERE SHAPE_SPHERE\n"
+	"	}//if (i<numPairs)\n"
+	"}				\n"
+	"__kernel void   clipHullHullConcaveConvexKernel( __global int4* concavePairsIn,\n"
+	"																					__global const b3RigidBodyData_t* rigidBodies, \n"
+	"																					__global const b3Collidable_t* collidables,\n"
+	"																					__global const b3ConvexPolyhedronData_t* convexShapes, \n"
+	"																					__global const float4* vertices,\n"
+	"																					__global const float4* uniqueEdges,\n"
+	"																					__global const b3GpuFace_t* faces,\n"
+	"																					__global const int* indices,\n"
+	"																					__global const b3GpuChildShape_t* gpuChildShapes,\n"
+	"																					__global const float4* separatingNormals,\n"
+	"																					__global struct b3Contact4Data* restrict globalContactsOut,\n"
+	"																					counter32_t nGlobalContactsOut,\n"
+	"																					int contactCapacity,\n"
+	"																					int numConcavePairs)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	int pairIndex = i;\n"
+	"	\n"
+	"	float4 worldVertsB1[64];\n"
+	"	float4 worldVertsB2[64];\n"
+	"	int capacityWorldVerts = 64;	\n"
+	"	float4 localContactsOut[64];\n"
+	"	int localContactCapacity=64;\n"
+	"	\n"
+	"	float minDist = -1e30f;\n"
+	"	float maxDist = 0.02f;\n"
+	"	if (i<numConcavePairs)\n"
+	"	{\n"
+	"		//negative value means that the pair is invalid\n"
+	"		if (concavePairsIn[i].w<0)\n"
+	"			return;\n"
+	"		int bodyIndexA = concavePairsIn[i].x;\n"
+	"		int bodyIndexB = concavePairsIn[i].y;\n"
+	"		int f = concavePairsIn[i].z;\n"
+	"		int childShapeIndexA = f;\n"
+	"		\n"
+	"		int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
+	"		int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
+	"		\n"
+	"		int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
+	"		int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
+	"		\n"
+	"		///////////////////////////////////////////////////////////////\n"
+	"		\n"
+	"	\n"
+	"		bool overlap = false;\n"
+	"		\n"
+	"		b3ConvexPolyhedronData_t convexPolyhedronA;\n"
+	"	//add 3 vertices of the triangle\n"
+	"		convexPolyhedronA.m_numVertices = 3;\n"
+	"		convexPolyhedronA.m_vertexOffset = 0;\n"
+	"		float4	localCenter = make_float4(0.f,0.f,0.f,0.f);\n"
+	"		b3GpuFace_t face = faces[convexShapes[shapeIndexA].m_faceOffset+f];\n"
+	"		\n"
+	"		float4 verticesA[3];\n"
+	"		for (int i=0;i<3;i++)\n"
+	"		{\n"
+	"			int index = indices[face.m_indexOffset+i];\n"
+	"			float4 vert = vertices[convexShapes[shapeIndexA].m_vertexOffset+index];\n"
+	"			verticesA[i] = vert;\n"
+	"			localCenter += vert;\n"
+	"		}\n"
+	"		float dmin = FLT_MAX;\n"
+	"		int localCC=0;\n"
+	"		//a triangle has 3 unique edges\n"
+	"		convexPolyhedronA.m_numUniqueEdges = 3;\n"
+	"		convexPolyhedronA.m_uniqueEdgesOffset = 0;\n"
+	"		float4 uniqueEdgesA[3];\n"
+	"		\n"
+	"		uniqueEdgesA[0] = (verticesA[1]-verticesA[0]);\n"
+	"		uniqueEdgesA[1] = (verticesA[2]-verticesA[1]);\n"
+	"		uniqueEdgesA[2] = (verticesA[0]-verticesA[2]);\n"
+	"		convexPolyhedronA.m_faceOffset = 0;\n"
+	"                                  \n"
+	"		float4 normal = make_float4(face.m_plane.x,face.m_plane.y,face.m_plane.z,0.f);\n"
+	"                             \n"
+	"		b3GpuFace_t facesA[TRIANGLE_NUM_CONVEX_FACES];\n"
+	"		int indicesA[3+3+2+2+2];\n"
+	"		int curUsedIndices=0;\n"
+	"		int fidx=0;\n"
+	"		//front size of triangle\n"
+	"		{\n"
+	"			facesA[fidx].m_indexOffset=curUsedIndices;\n"
+	"			indicesA[0] = 0;\n"
+	"			indicesA[1] = 1;\n"
+	"			indicesA[2] = 2;\n"
+	"			curUsedIndices+=3;\n"
+	"			float c = face.m_plane.w;\n"
+	"			facesA[fidx].m_plane.x = normal.x;\n"
+	"			facesA[fidx].m_plane.y = normal.y;\n"
+	"			facesA[fidx].m_plane.z = normal.z;\n"
+	"			facesA[fidx].m_plane.w = c;\n"
+	"			facesA[fidx].m_numIndices=3;\n"
+	"		}\n"
+	"		fidx++;\n"
+	"		//back size of triangle\n"
+	"		{\n"
+	"			facesA[fidx].m_indexOffset=curUsedIndices;\n"
+	"			indicesA[3]=2;\n"
+	"			indicesA[4]=1;\n"
+	"			indicesA[5]=0;\n"
+	"			curUsedIndices+=3;\n"
+	"			float c = dot3F4(normal,verticesA[0]);\n"
+	"			float c1 = -face.m_plane.w;\n"
+	"			facesA[fidx].m_plane.x = -normal.x;\n"
+	"			facesA[fidx].m_plane.y = -normal.y;\n"
+	"			facesA[fidx].m_plane.z = -normal.z;\n"
+	"			facesA[fidx].m_plane.w = c;\n"
+	"			facesA[fidx].m_numIndices=3;\n"
+	"		}\n"
+	"		fidx++;\n"
+	"		bool addEdgePlanes = true;\n"
+	"		if (addEdgePlanes)\n"
+	"		{\n"
+	"			int numVertices=3;\n"
+	"			int prevVertex = numVertices-1;\n"
+	"			for (int i=0;i<numVertices;i++)\n"
+	"			{\n"
+	"				float4 v0 = verticesA[i];\n"
+	"				float4 v1 = verticesA[prevVertex];\n"
+	"                                            \n"
+	"				float4 edgeNormal = normalize(cross(normal,v1-v0));\n"
+	"				float c = -dot3F4(edgeNormal,v0);\n"
+	"				facesA[fidx].m_numIndices = 2;\n"
+	"				facesA[fidx].m_indexOffset=curUsedIndices;\n"
+	"				indicesA[curUsedIndices++]=i;\n"
+	"				indicesA[curUsedIndices++]=prevVertex;\n"
+	"                                            \n"
+	"				facesA[fidx].m_plane.x = edgeNormal.x;\n"
+	"				facesA[fidx].m_plane.y = edgeNormal.y;\n"
+	"				facesA[fidx].m_plane.z = edgeNormal.z;\n"
+	"				facesA[fidx].m_plane.w = c;\n"
+	"				fidx++;\n"
+	"				prevVertex = i;\n"
+	"			}\n"
+	"		}\n"
+	"		convexPolyhedronA.m_numFaces = TRIANGLE_NUM_CONVEX_FACES;\n"
+	"		convexPolyhedronA.m_localCenter = localCenter*(1.f/3.f);\n"
+	"		float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
+	"		posA.w = 0.f;\n"
+	"		float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
+	"		posB.w = 0.f;\n"
+	"		float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
+	"		float4 ornB =rigidBodies[bodyIndexB].m_quat;\n"
+	"		float4 sepAxis = separatingNormals[i];\n"
+	"		\n"
+	"		int shapeTypeB = collidables[collidableIndexB].m_shapeType;\n"
+	"		int childShapeIndexB =-1;\n"
+	"		if (shapeTypeB==SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
+	"		{\n"
+	"			///////////////////\n"
+	"			///compound shape support\n"
+	"			\n"
+	"			childShapeIndexB = concavePairsIn[pairIndex].w;\n"
+	"			int childColIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;\n"
+	"			shapeIndexB = collidables[childColIndexB].m_shapeIndex;\n"
+	"			float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;\n"
+	"			float4 childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;\n"
+	"			float4 newPosB = transform(&childPosB,&posB,&ornB);\n"
+	"			float4 newOrnB = qtMul(ornB,childOrnB);\n"
+	"			posB = newPosB;\n"
+	"			ornB = newOrnB;\n"
+	"			\n"
+	"		}\n"
+	"		\n"
+	"		////////////////////////////////////////\n"
+	"		\n"
+	"		\n"
+	"		\n"
+	"		int numLocalContactsOut = clipHullAgainstHullLocalA(sepAxis,\n"
+	"														&convexPolyhedronA, &convexShapes[shapeIndexB],\n"
+	"														posA,ornA,\n"
+	"													  posB,ornB,\n"
+	"													  worldVertsB1,worldVertsB2,capacityWorldVerts,\n"
+	"														minDist, maxDist,\n"
+	"														&verticesA,&facesA,&indicesA,\n"
+	"														vertices,faces,indices,\n"
+	"														localContactsOut,localContactCapacity);\n"
+	"												\n"
+	"		if (numLocalContactsOut>0)\n"
+	"		{\n"
+	"			float4 normal = -separatingNormals[i];\n"
+	"			int nPoints = numLocalContactsOut;\n"
+	"			float4* pointsIn = localContactsOut;\n"
+	"			int contactIdx[4];// = {-1,-1,-1,-1};\n"
+	"			contactIdx[0] = -1;\n"
+	"			contactIdx[1] = -1;\n"
+	"			contactIdx[2] = -1;\n"
+	"			contactIdx[3] = -1;\n"
+	"	\n"
+	"			int nReducedContacts = extractManifoldSequential(pointsIn, nPoints, normal, contactIdx);\n"
+	"	\n"
+	"			int dstIdx;\n"
+	"			AppendInc( nGlobalContactsOut, dstIdx );\n"
+	"			if (dstIdx<contactCapacity)\n"
+	"			{\n"
+	"				__global struct b3Contact4Data* c = globalContactsOut+ dstIdx;\n"
+	"				c->m_worldNormalOnB = -normal;\n"
+	"				c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
+	"				c->m_batchIdx = pairIndex;\n"
+	"				int bodyA = concavePairsIn[pairIndex].x;\n"
+	"				int bodyB = concavePairsIn[pairIndex].y;\n"
+	"				c->m_bodyAPtrAndSignBit = rigidBodies[bodyA].m_invMass==0?-bodyA:bodyA;\n"
+	"				c->m_bodyBPtrAndSignBit = rigidBodies[bodyB].m_invMass==0?-bodyB:bodyB;\n"
+	"				c->m_childIndexA = childShapeIndexA;\n"
+	"				c->m_childIndexB = childShapeIndexB;\n"
+	"				for (int i=0;i<nReducedContacts;i++)\n"
+	"				{\n"
+	"					c->m_worldPosB[i] = pointsIn[contactIdx[i]];\n"
+	"				}\n"
+	"				GET_NPOINTS(*c) = nReducedContacts;\n"
+	"			}\n"
+	"				\n"
+	"		}//		if (numContactsOut>0)\n"
+	"	}//	if (i<numPairs)\n"
+	"}\n"
+	"int	findClippingFaces(const float4 separatingNormal,\n"
+	"                      __global const b3ConvexPolyhedronData_t* hullA, __global const b3ConvexPolyhedronData_t* hullB,\n"
+	"                      const float4 posA, const Quaternion ornA,const float4 posB, const Quaternion ornB,\n"
+	"                       __global float4* worldVertsA1,\n"
+	"                      __global float4* worldNormalsA1,\n"
+	"                      __global float4* worldVertsB1,\n"
+	"                      int capacityWorldVerts,\n"
+	"                      const float minDist, float maxDist,\n"
+	"                      __global const float4* vertices,\n"
+	"                      __global const b3GpuFace_t* faces,\n"
+	"                      __global const int* indices,\n"
+	"                      __global int4* clippingFaces, int pairIndex)\n"
+	"{\n"
+	"	int numContactsOut = 0;\n"
+	"	int numWorldVertsB1= 0;\n"
+	"    \n"
+	"    \n"
+	"	int closestFaceB=-1;\n"
+	"	float dmax = -FLT_MAX;\n"
+	"    \n"
+	"	{\n"
+	"		for(int face=0;face<hullB->m_numFaces;face++)\n"
+	"		{\n"
+	"			const float4 Normal = make_float4(faces[hullB->m_faceOffset+face].m_plane.x,\n"
+	"                                              faces[hullB->m_faceOffset+face].m_plane.y, faces[hullB->m_faceOffset+face].m_plane.z,0.f);\n"
+	"			const float4 WorldNormal = qtRotate(ornB, Normal);\n"
+	"			float d = dot3F4(WorldNormal,separatingNormal);\n"
+	"			if (d > dmax)\n"
+	"			{\n"
+	"				dmax = d;\n"
+	"				closestFaceB = face;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"    \n"
+	"	{\n"
+	"		const b3GpuFace_t polyB = faces[hullB->m_faceOffset+closestFaceB];\n"
+	"		const int numVertices = polyB.m_numIndices;\n"
+	"		for(int e0=0;e0<numVertices;e0++)\n"
+	"		{\n"
+	"			const float4 b = vertices[hullB->m_vertexOffset+indices[polyB.m_indexOffset+e0]];\n"
+	"			worldVertsB1[pairIndex*capacityWorldVerts+numWorldVertsB1++] = transform(&b,&posB,&ornB);\n"
+	"		}\n"
+	"	}\n"
+	"    \n"
+	"    int closestFaceA=-1;\n"
+	"	{\n"
+	"		float dmin = FLT_MAX;\n"
+	"		for(int face=0;face<hullA->m_numFaces;face++)\n"
+	"		{\n"
+	"			const float4 Normal = make_float4(\n"
+	"                                              faces[hullA->m_faceOffset+face].m_plane.x,\n"
+	"                                              faces[hullA->m_faceOffset+face].m_plane.y,\n"
+	"                                              faces[hullA->m_faceOffset+face].m_plane.z,\n"
+	"                                              0.f);\n"
+	"			const float4 faceANormalWS = qtRotate(ornA,Normal);\n"
+	"            \n"
+	"			float d = dot3F4(faceANormalWS,separatingNormal);\n"
+	"			if (d < dmin)\n"
+	"			{\n"
+	"				dmin = d;\n"
+	"				closestFaceA = face;\n"
+	"                worldNormalsA1[pairIndex] = faceANormalWS;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"    \n"
+	"    int numVerticesA = faces[hullA->m_faceOffset+closestFaceA].m_numIndices;\n"
+	"	for(int e0=0;e0<numVerticesA;e0++)\n"
+	"	{\n"
+	"        const float4 a = vertices[hullA->m_vertexOffset+indices[faces[hullA->m_faceOffset+closestFaceA].m_indexOffset+e0]];\n"
+	"        worldVertsA1[pairIndex*capacityWorldVerts+e0] = transform(&a, &posA,&ornA);\n"
+	"    }\n"
+	"    \n"
+	"    clippingFaces[pairIndex].x = closestFaceA;\n"
+	"    clippingFaces[pairIndex].y = closestFaceB;\n"
+	"    clippingFaces[pairIndex].z = numVerticesA;\n"
+	"    clippingFaces[pairIndex].w = numWorldVertsB1;\n"
+	"    \n"
+	"    \n"
+	"	return numContactsOut;\n"
+	"}\n"
+	"int clipFaces(__global float4* worldVertsA1,\n"
+	"              __global float4* worldNormalsA1,\n"
+	"              __global float4* worldVertsB1,\n"
+	"              __global float4* worldVertsB2, \n"
+	"              int capacityWorldVertsB2,\n"
+	"              const float minDist, float maxDist,\n"
+	"              __global int4* clippingFaces,\n"
+	"              int pairIndex)\n"
+	"{\n"
+	"	int numContactsOut = 0;\n"
+	"    \n"
+	"    int closestFaceA = clippingFaces[pairIndex].x;\n"
+	"    int closestFaceB = clippingFaces[pairIndex].y;\n"
+	"	int numVertsInA = clippingFaces[pairIndex].z;\n"
+	"	int numVertsInB = clippingFaces[pairIndex].w;\n"
+	"    \n"
+	"	int numVertsOut = 0;\n"
+	"    \n"
+	"	if (closestFaceA<0)\n"
+	"		return numContactsOut;\n"
+	"    \n"
+	"    __global float4* pVtxIn = &worldVertsB1[pairIndex*capacityWorldVertsB2];\n"
+	"    __global float4* pVtxOut = &worldVertsB2[pairIndex*capacityWorldVertsB2];\n"
+	"    \n"
+	"    \n"
+	"	\n"
+	"	// clip polygon to back of planes of all faces of hull A that are adjacent to witness face\n"
+	"    \n"
+	"	for(int e0=0;e0<numVertsInA;e0++)\n"
+	"	{\n"
+	"		const float4 aw = worldVertsA1[pairIndex*capacityWorldVertsB2+e0];\n"
+	"		const float4 bw = worldVertsA1[pairIndex*capacityWorldVertsB2+((e0+1)%numVertsInA)];\n"
+	"		const float4 WorldEdge0 = aw - bw;\n"
+	"		float4 worldPlaneAnormal1 = worldNormalsA1[pairIndex];\n"
+	"		float4 planeNormalWS1 = -cross3(WorldEdge0,worldPlaneAnormal1);\n"
+	"		float4 worldA1 = aw;\n"
+	"		float planeEqWS1 = -dot3F4(worldA1,planeNormalWS1);\n"
+	"		float4 planeNormalWS = planeNormalWS1;\n"
+	"		float planeEqWS=planeEqWS1;\n"
+	"		numVertsOut = clipFaceGlobal(pVtxIn, numVertsInB, planeNormalWS,planeEqWS, pVtxOut);\n"
+	"		__global float4* tmp = pVtxOut;\n"
+	"		pVtxOut = pVtxIn;\n"
+	"		pVtxIn = tmp;\n"
+	"		numVertsInB = numVertsOut;\n"
+	"		numVertsOut = 0;\n"
+	"	}\n"
+	"    \n"
+	"    //float4 planeNormalWS = worldNormalsA1[pairIndex];\n"
+	"    //float planeEqWS=-dot3F4(planeNormalWS,worldVertsA1[pairIndex*capacityWorldVertsB2]);\n"
+	"    \n"
+	"    /*for (int i=0;i<numVertsInB;i++)\n"
+	"    {\n"
+	"        pVtxOut[i] = pVtxIn[i];\n"
+	"    }*/\n"
+	"    \n"
+	"    \n"
+	"    \n"
+	"    \n"
+	"    //numVertsInB=0;\n"
+	"	\n"
+	"    float4 planeNormalWS = worldNormalsA1[pairIndex];\n"
+	"    float planeEqWS=-dot3F4(planeNormalWS,worldVertsA1[pairIndex*capacityWorldVertsB2]);\n"
+	"    for (int i=0;i<numVertsInB;i++)\n"
+	"    {\n"
+	"        float depth = dot3F4(planeNormalWS,pVtxIn[i])+planeEqWS;\n"
+	"        if (depth <=minDist)\n"
+	"        {\n"
+	"            depth = minDist;\n"
+	"        }\n"
+	"        \n"
+	"        if (depth <=maxDist)\n"
+	"        {\n"
+	"            float4 pointInWorld = pVtxIn[i];\n"
+	"            pVtxOut[numContactsOut++] = make_float4(pointInWorld.x,pointInWorld.y,pointInWorld.z,depth);\n"
+	"        }\n"
+	"    }\n"
+	"   \n"
+	"    clippingFaces[pairIndex].w =numContactsOut;\n"
+	"   \n"
+	"    \n"
+	"	return numContactsOut;\n"
+	"}\n"
+	"__kernel void   findClippingFacesKernel(  __global const int4* pairs,\n"
+	"                                        __global const b3RigidBodyData_t* rigidBodies,\n"
+	"                                        __global const b3Collidable_t* collidables,\n"
+	"                                        __global const b3ConvexPolyhedronData_t* convexShapes,\n"
+	"                                        __global const float4* vertices,\n"
+	"                                        __global const float4* uniqueEdges,\n"
+	"                                        __global const b3GpuFace_t* faces,\n"
+	"                                        __global const int* indices,\n"
+	"                                        __global const float4* separatingNormals,\n"
+	"                                        __global const int* hasSeparatingAxis,\n"
+	"                                        __global int4* clippingFacesOut,\n"
+	"                                        __global float4* worldVertsA1,\n"
+	"                                        __global float4* worldNormalsA1,\n"
+	"                                        __global float4* worldVertsB1,\n"
+	"                                        int capacityWorldVerts,\n"
+	"                                        int numPairs\n"
+	"                                        )\n"
+	"{\n"
+	"    \n"
+	"	int i = get_global_id(0);\n"
+	"	int pairIndex = i;\n"
+	"    \n"
+	"	\n"
+	"	float minDist = -1e30f;\n"
+	"	float maxDist = 0.02f;\n"
+	"    \n"
+	"	if (i<numPairs)\n"
+	"	{\n"
+	"        \n"
+	"		if (hasSeparatingAxis[i])\n"
+	"		{\n"
+	"            \n"
+	"			int bodyIndexA = pairs[i].x;\n"
+	"			int bodyIndexB = pairs[i].y;\n"
+	"			\n"
+	"			int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
+	"			int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
+	"			\n"
+	"			int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
+	"			int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
+	"			\n"
+	"            \n"
+	"            \n"
+	"			int numLocalContactsOut = findClippingFaces(separatingNormals[i],\n"
+	"                                                        &convexShapes[shapeIndexA], &convexShapes[shapeIndexB],\n"
+	"                                                        rigidBodies[bodyIndexA].m_pos,rigidBodies[bodyIndexA].m_quat,\n"
+	"                                                        rigidBodies[bodyIndexB].m_pos,rigidBodies[bodyIndexB].m_quat,\n"
+	"                                                        worldVertsA1,\n"
+	"                                                        worldNormalsA1,\n"
+	"                                                        worldVertsB1,capacityWorldVerts,\n"
+	"                                                        minDist, maxDist,\n"
+	"                                                        vertices,faces,indices,\n"
+	"                                                        clippingFacesOut,i);\n"
+	"            \n"
+	"            \n"
+	"		}//		if (hasSeparatingAxis[i])\n"
+	"	}//	if (i<numPairs)\n"
+	"    \n"
+	"}\n"
+	"__kernel void   clipFacesAndFindContactsKernel(    __global const float4* separatingNormals,\n"
+	"                                                   __global const int* hasSeparatingAxis,\n"
+	"                                                   __global int4* clippingFacesOut,\n"
+	"                                                   __global float4* worldVertsA1,\n"
+	"                                                   __global float4* worldNormalsA1,\n"
+	"                                                   __global float4* worldVertsB1,\n"
+	"                                                   __global float4* worldVertsB2,\n"
+	"                                                    int vertexFaceCapacity,\n"
+	"                                                   int numPairs,\n"
+	"					                                        int debugMode\n"
+	"                                                   )\n"
+	"{\n"
+	"    int i = get_global_id(0);\n"
+	"	int pairIndex = i;\n"
+	"	\n"
+	"    \n"
+	"	float minDist = -1e30f;\n"
+	"	float maxDist = 0.02f;\n"
+	"    \n"
+	"	if (i<numPairs)\n"
+	"	{\n"
+	"        \n"
+	"		if (hasSeparatingAxis[i])\n"
+	"		{\n"
+	"            \n"
+	"//			int bodyIndexA = pairs[i].x;\n"
+	"	//		int bodyIndexB = pairs[i].y;\n"
+	"		    \n"
+	"            int numLocalContactsOut = 0;\n"
+	"            int capacityWorldVertsB2 = vertexFaceCapacity;\n"
+	"            \n"
+	"            __global float4* pVtxIn = &worldVertsB1[pairIndex*capacityWorldVertsB2];\n"
+	"            __global float4* pVtxOut = &worldVertsB2[pairIndex*capacityWorldVertsB2];\n"
+	"            \n"
+	"            {\n"
+	"                __global int4* clippingFaces = clippingFacesOut;\n"
+	"            \n"
+	"                \n"
+	"                int closestFaceA = clippingFaces[pairIndex].x;\n"
+	"                int closestFaceB = clippingFaces[pairIndex].y;\n"
+	"                int numVertsInA = clippingFaces[pairIndex].z;\n"
+	"                int numVertsInB = clippingFaces[pairIndex].w;\n"
+	"                \n"
+	"                int numVertsOut = 0;\n"
+	"                \n"
+	"                if (closestFaceA>=0)\n"
+	"                {\n"
+	"                    \n"
+	"                    \n"
+	"                    \n"
+	"                    // clip polygon to back of planes of all faces of hull A that are adjacent to witness face\n"
+	"                    \n"
+	"                    for(int e0=0;e0<numVertsInA;e0++)\n"
+	"                    {\n"
+	"                        const float4 aw = worldVertsA1[pairIndex*capacityWorldVertsB2+e0];\n"
+	"                        const float4 bw = worldVertsA1[pairIndex*capacityWorldVertsB2+((e0+1)%numVertsInA)];\n"
+	"                        const float4 WorldEdge0 = aw - bw;\n"
+	"                        float4 worldPlaneAnormal1 = worldNormalsA1[pairIndex];\n"
+	"                        float4 planeNormalWS1 = -cross3(WorldEdge0,worldPlaneAnormal1);\n"
+	"                        float4 worldA1 = aw;\n"
+	"                        float planeEqWS1 = -dot3F4(worldA1,planeNormalWS1);\n"
+	"                        float4 planeNormalWS = planeNormalWS1;\n"
+	"                        float planeEqWS=planeEqWS1;\n"
+	"                        numVertsOut = clipFaceGlobal(pVtxIn, numVertsInB, planeNormalWS,planeEqWS, pVtxOut);\n"
+	"                        __global float4* tmp = pVtxOut;\n"
+	"                        pVtxOut = pVtxIn;\n"
+	"                        pVtxIn = tmp;\n"
+	"                        numVertsInB = numVertsOut;\n"
+	"                        numVertsOut = 0;\n"
+	"                    }\n"
+	"                    \n"
+	"                    float4 planeNormalWS = worldNormalsA1[pairIndex];\n"
+	"                    float planeEqWS=-dot3F4(planeNormalWS,worldVertsA1[pairIndex*capacityWorldVertsB2]);\n"
+	"                    \n"
+	"                    for (int i=0;i<numVertsInB;i++)\n"
+	"                    {\n"
+	"                        float depth = dot3F4(planeNormalWS,pVtxIn[i])+planeEqWS;\n"
+	"                        if (depth <=minDist)\n"
+	"                        {\n"
+	"                            depth = minDist;\n"
+	"                        }\n"
+	"                        \n"
+	"                        if (depth <=maxDist)\n"
+	"                        {\n"
+	"                            float4 pointInWorld = pVtxIn[i];\n"
+	"                            pVtxOut[numLocalContactsOut++] = make_float4(pointInWorld.x,pointInWorld.y,pointInWorld.z,depth);\n"
+	"                        }\n"
+	"                    }\n"
+	"                    \n"
+	"                }\n"
+	"                clippingFaces[pairIndex].w =numLocalContactsOut;\n"
+	"                \n"
+	"            }\n"
+	"            \n"
+	"            for (int i=0;i<numLocalContactsOut;i++)\n"
+	"                pVtxIn[i] = pVtxOut[i];\n"
+	"                \n"
+	"		}//		if (hasSeparatingAxis[i])\n"
+	"	}//	if (i<numPairs)\n"
+	"    \n"
+	"}\n"
+	"__kernel void   newContactReductionKernel( __global int4* pairs,\n"
+	"                                                   __global const b3RigidBodyData_t* rigidBodies,\n"
+	"                                                   __global const float4* separatingNormals,\n"
+	"                                                   __global const int* hasSeparatingAxis,\n"
+	"                                                   __global struct b3Contact4Data* globalContactsOut,\n"
+	"                                                   __global int4* clippingFaces,\n"
+	"                                                   __global float4* worldVertsB2,\n"
+	"                                                   volatile __global int* nGlobalContactsOut,\n"
+	"                                                   int vertexFaceCapacity,\n"
+	"												   int contactCapacity,\n"
+	"                                                   int numPairs\n"
+	"                                                   )\n"
+	"{\n"
+	"    int i = get_global_id(0);\n"
+	"	int pairIndex = i;\n"
+	"	\n"
+	"    int4 contactIdx;\n"
+	"    contactIdx=make_int4(0,1,2,3);\n"
+	"    \n"
+	"	if (i<numPairs)\n"
+	"	{\n"
+	"        \n"
+	"		if (hasSeparatingAxis[i])\n"
+	"		{\n"
+	"            \n"
+	"			\n"
+	"            \n"
+	"            \n"
+	"			int nPoints = clippingFaces[pairIndex].w;\n"
+	"           \n"
+	"            if (nPoints>0)\n"
+	"            {\n"
+	"                 __global float4* pointsIn = &worldVertsB2[pairIndex*vertexFaceCapacity];\n"
+	"                float4 normal = -separatingNormals[i];\n"
+	"                \n"
+	"                int nReducedContacts = extractManifoldSequentialGlobal(pointsIn, nPoints, normal, &contactIdx);\n"
+	"            \n"
+	"				int mprContactIndex = pairs[pairIndex].z;\n"
+	"                int dstIdx = mprContactIndex;\n"
+	"				if (dstIdx<0)\n"
+	"				{\n"
+	"	                AppendInc( nGlobalContactsOut, dstIdx );\n"
+	"				}\n"
+	"//#if 0\n"
+	"                \n"
+	"				if (dstIdx < contactCapacity)\n"
+	"				{\n"
+	"					__global struct b3Contact4Data* c = &globalContactsOut[dstIdx];\n"
+	"					c->m_worldNormalOnB = -normal;\n"
+	"					c->m_restituitionCoeffCmp = (0.f*0xffff);c->m_frictionCoeffCmp = (0.7f*0xffff);\n"
+	"					c->m_batchIdx = pairIndex;\n"
+	"					int bodyA = pairs[pairIndex].x;\n"
+	"					int bodyB = pairs[pairIndex].y;\n"
+	"					pairs[pairIndex].w = dstIdx;\n"
+	"					c->m_bodyAPtrAndSignBit = rigidBodies[bodyA].m_invMass==0?-bodyA:bodyA;\n"
+	"					c->m_bodyBPtrAndSignBit = rigidBodies[bodyB].m_invMass==0?-bodyB:bodyB;\n"
+	"                    c->m_childIndexA =-1;\n"
+	"					c->m_childIndexB =-1;\n"
+	"                    switch (nReducedContacts)\n"
+	"                    {\n"
+	"                        case 4:\n"
+	"                            c->m_worldPosB[3] = pointsIn[contactIdx.w];\n"
+	"                        case 3:\n"
+	"                            c->m_worldPosB[2] = pointsIn[contactIdx.z];\n"
+	"                        case 2:\n"
+	"                            c->m_worldPosB[1] = pointsIn[contactIdx.y];\n"
+	"                        case 1:\n"
+	"							if (mprContactIndex<0)//test\n"
+	"	                            c->m_worldPosB[0] = pointsIn[contactIdx.x];\n"
+	"                        default:\n"
+	"                        {\n"
+	"                        }\n"
+	"                    };\n"
+	"                    \n"
+	"					GET_NPOINTS(*c) = nReducedContacts;\n"
+	"                    \n"
+	"                 }\n"
+	"                 \n"
+	"                \n"
+	"//#endif\n"
+	"				\n"
+	"			}//		if (numContactsOut>0)\n"
+	"		}//		if (hasSeparatingAxis[i])\n"
+	"	}//	if (i<numPairs)\n"
+	"    \n"
+	"    \n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/satConcaveKernels.h b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/satConcaveKernels.h
index 611569cacfb..a60702ca62e 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/satConcaveKernels.h
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/satConcaveKernels.h
@@ -1,1457 +1,1456 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* satConcaveKernelsCL= \
-"//keep this enum in sync with the CPU version (in btCollidable.h)\n"
-"//written by Erwin Coumans\n"
-"#define SHAPE_CONVEX_HULL 3\n"
-"#define SHAPE_CONCAVE_TRIMESH 5\n"
-"#define TRIANGLE_NUM_CONVEX_FACES 5\n"
-"#define SHAPE_COMPOUND_OF_CONVEX_HULLS 6\n"
-"#define B3_MAX_STACK_DEPTH 256\n"
-"typedef unsigned int u32;\n"
-"///keep this in sync with btCollidable.h\n"
-"typedef struct\n"
-"{\n"
-"	union {\n"
-"		int m_numChildShapes;\n"
-"		int m_bvhIndex;\n"
-"	};\n"
-"	union\n"
-"	{\n"
-"		float m_radius;\n"
-"		int	m_compoundBvhIndex;\n"
-"	};\n"
-"	\n"
-"	int m_shapeType;\n"
-"	int m_shapeIndex;\n"
-"	\n"
-"} btCollidableGpu;\n"
-"#define MAX_NUM_PARTS_IN_BITS 10\n"
-"///b3QuantizedBvhNode is a compressed aabb node, 16 bytes.\n"
-"///Node can be used for leafnode or internal node. Leafnodes can point to 32-bit triangle index (non-negative range).\n"
-"typedef struct\n"
-"{\n"
-"	//12 bytes\n"
-"	unsigned short int	m_quantizedAabbMin[3];\n"
-"	unsigned short int	m_quantizedAabbMax[3];\n"
-"	//4 bytes\n"
-"	int	m_escapeIndexOrTriangleIndex;\n"
-"} b3QuantizedBvhNode;\n"
-"typedef struct\n"
-"{\n"
-"	float4		m_aabbMin;\n"
-"	float4		m_aabbMax;\n"
-"	float4		m_quantization;\n"
-"	int			m_numNodes;\n"
-"	int			m_numSubTrees;\n"
-"	int			m_nodeOffset;\n"
-"	int			m_subTreeOffset;\n"
-"} b3BvhInfo;\n"
-"int	getTriangleIndex(const b3QuantizedBvhNode* rootNode)\n"
-"{\n"
-"	unsigned int x=0;\n"
-"	unsigned int y = (~(x&0))<<(31-MAX_NUM_PARTS_IN_BITS);\n"
-"	// Get only the lower bits where the triangle index is stored\n"
-"	return (rootNode->m_escapeIndexOrTriangleIndex&~(y));\n"
-"}\n"
-"int	getTriangleIndexGlobal(__global const b3QuantizedBvhNode* rootNode)\n"
-"{\n"
-"	unsigned int x=0;\n"
-"	unsigned int y = (~(x&0))<<(31-MAX_NUM_PARTS_IN_BITS);\n"
-"	// Get only the lower bits where the triangle index is stored\n"
-"	return (rootNode->m_escapeIndexOrTriangleIndex&~(y));\n"
-"}\n"
-"int isLeafNode(const b3QuantizedBvhNode* rootNode)\n"
-"{\n"
-"	//skipindex is negative (internal node), triangleindex >=0 (leafnode)\n"
-"	return (rootNode->m_escapeIndexOrTriangleIndex >= 0)? 1 : 0;\n"
-"}\n"
-"int isLeafNodeGlobal(__global const b3QuantizedBvhNode* rootNode)\n"
-"{\n"
-"	//skipindex is negative (internal node), triangleindex >=0 (leafnode)\n"
-"	return (rootNode->m_escapeIndexOrTriangleIndex >= 0)? 1 : 0;\n"
-"}\n"
-"	\n"
-"int getEscapeIndex(const b3QuantizedBvhNode* rootNode)\n"
-"{\n"
-"	return -rootNode->m_escapeIndexOrTriangleIndex;\n"
-"}\n"
-"int getEscapeIndexGlobal(__global const b3QuantizedBvhNode* rootNode)\n"
-"{\n"
-"	return -rootNode->m_escapeIndexOrTriangleIndex;\n"
-"}\n"
-"typedef struct\n"
-"{\n"
-"	//12 bytes\n"
-"	unsigned short int	m_quantizedAabbMin[3];\n"
-"	unsigned short int	m_quantizedAabbMax[3];\n"
-"	//4 bytes, points to the root of the subtree\n"
-"	int			m_rootNodeIndex;\n"
-"	//4 bytes\n"
-"	int			m_subtreeSize;\n"
-"	int			m_padding[3];\n"
-"} b3BvhSubtreeInfo;\n"
-"typedef struct\n"
-"{\n"
-"	float4	m_childPosition;\n"
-"	float4	m_childOrientation;\n"
-"	int m_shapeIndex;\n"
-"	int m_unused0;\n"
-"	int m_unused1;\n"
-"	int m_unused2;\n"
-"} btGpuChildShape;\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_pos;\n"
-"	float4 m_quat;\n"
-"	float4 m_linVel;\n"
-"	float4 m_angVel;\n"
-"	u32 m_collidableIdx;\n"
-"	float m_invMass;\n"
-"	float m_restituitionCoeff;\n"
-"	float m_frictionCoeff;\n"
-"} BodyData;\n"
-"typedef struct  \n"
-"{\n"
-"	float4		m_localCenter;\n"
-"	float4		m_extents;\n"
-"	float4		mC;\n"
-"	float4		mE;\n"
-"	\n"
-"	float			m_radius;\n"
-"	int	m_faceOffset;\n"
-"	int m_numFaces;\n"
-"	int	m_numVertices;\n"
-"	int m_vertexOffset;\n"
-"	int	m_uniqueEdgesOffset;\n"
-"	int	m_numUniqueEdges;\n"
-"	int m_unused;\n"
-"} ConvexPolyhedronCL;\n"
-"typedef struct \n"
-"{\n"
-"	union\n"
-"	{\n"
-"		float4	m_min;\n"
-"		float   m_minElems[4];\n"
-"		int			m_minIndices[4];\n"
-"	};\n"
-"	union\n"
-"	{\n"
-"		float4	m_max;\n"
-"		float   m_maxElems[4];\n"
-"		int			m_maxIndices[4];\n"
-"	};\n"
-"} btAabbCL;\n"
-"#ifndef B3_AABB_H\n"
-"#define B3_AABB_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#define B3_FLOAT4_H\n"
-"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
-"#define B3_PLATFORM_DEFINITIONS_H\n"
-"struct MyTest\n"
-"{\n"
-"	int bla;\n"
-"};\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
-"#define B3_LARGE_FLOAT 1e18f\n"
-"#define B3_INFINITY 1e18f\n"
-"#define b3Assert(a)\n"
-"#define b3ConstArray(a) __global const a*\n"
-"#define b3AtomicInc atomic_inc\n"
-"#define b3AtomicAdd atomic_add\n"
-"#define b3Fabs fabs\n"
-"#define b3Sqrt native_sqrt\n"
-"#define b3Sin native_sin\n"
-"#define b3Cos native_cos\n"
-"#define B3_STATIC\n"
-"#endif\n"
-"#endif\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"	typedef float4	b3Float4;\n"
-"	#define b3Float4ConstArg const b3Float4\n"
-"	#define b3MakeFloat4 (float4)\n"
-"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return dot(a1, b1);\n"
-"	}\n"
-"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return cross(a1, b1);\n"
-"	}\n"
-"	#define b3MinFloat4 min\n"
-"	#define b3MaxFloat4 max\n"
-"	#define b3Normalized(a) normalize(a)\n"
-"#endif \n"
-"		\n"
-"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
-"{\n"
-"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
-"		return false;\n"
-"	return true;\n"
-"}\n"
-"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
-"{\n"
-"    float maxDot = -B3_INFINITY;\n"
-"    int i = 0;\n"
-"    int ptIndex = -1;\n"
-"    for( i = 0; i < vecLen; i++ )\n"
-"    {\n"
-"        float dot = b3Dot3F4(vecArray[i],vec);\n"
-"            \n"
-"        if( dot > maxDot )\n"
-"        {\n"
-"            maxDot = dot;\n"
-"            ptIndex = i;\n"
-"        }\n"
-"    }\n"
-"	b3Assert(ptIndex>=0);\n"
-"    if (ptIndex<0)\n"
-"	{\n"
-"		ptIndex = 0;\n"
-"	}\n"
-"    *dotOut = maxDot;\n"
-"    return ptIndex;\n"
-"}\n"
-"#endif //B3_FLOAT4_H\n"
-"#ifndef B3_MAT3x3_H\n"
-"#define B3_MAT3x3_H\n"
-"#ifndef B3_QUAT_H\n"
-"#define B3_QUAT_H\n"
-"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif\n"
-"#endif\n"
-"#ifndef B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"	typedef float4	b3Quat;\n"
-"	#define b3QuatConstArg const b3Quat\n"
-"	\n"
-"	\n"
-"inline float4 b3FastNormalize4(float4 v)\n"
-"{\n"
-"	v = (float4)(v.xyz,0.f);\n"
-"	return fast_normalize(v);\n"
-"}\n"
-"	\n"
-"inline b3Quat b3QuatMul(b3Quat a, b3Quat b);\n"
-"inline b3Quat b3QuatNormalized(b3QuatConstArg in);\n"
-"inline b3Quat b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec);\n"
-"inline b3Quat b3QuatInvert(b3QuatConstArg q);\n"
-"inline b3Quat b3QuatInverse(b3QuatConstArg q);\n"
-"inline b3Quat b3QuatMul(b3QuatConstArg a, b3QuatConstArg b)\n"
-"{\n"
-"	b3Quat ans;\n"
-"	ans = b3Cross3( a, b );\n"
-"	ans += a.w*b+b.w*a;\n"
-"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
-"	ans.w = a.w*b.w - b3Dot3F4(a, b);\n"
-"	return ans;\n"
-"}\n"
-"inline b3Quat b3QuatNormalized(b3QuatConstArg in)\n"
-"{\n"
-"	b3Quat q;\n"
-"	q=in;\n"
-"	//return b3FastNormalize4(in);\n"
-"	float len = native_sqrt(dot(q, q));\n"
-"	if(len > 0.f)\n"
-"	{\n"
-"		q *= 1.f / len;\n"
-"	}\n"
-"	else\n"
-"	{\n"
-"		q.x = q.y = q.z = 0.f;\n"
-"		q.w = 1.f;\n"
-"	}\n"
-"	return q;\n"
-"}\n"
-"inline float4 b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
-"{\n"
-"	b3Quat qInv = b3QuatInvert( q );\n"
-"	float4 vcpy = vec;\n"
-"	vcpy.w = 0.f;\n"
-"	float4 out = b3QuatMul(b3QuatMul(q,vcpy),qInv);\n"
-"	return out;\n"
-"}\n"
-"inline b3Quat b3QuatInverse(b3QuatConstArg q)\n"
-"{\n"
-"	return (b3Quat)(-q.xyz, q.w);\n"
-"}\n"
-"inline b3Quat b3QuatInvert(b3QuatConstArg q)\n"
-"{\n"
-"	return (b3Quat)(-q.xyz, q.w);\n"
-"}\n"
-"inline float4 b3QuatInvRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
-"{\n"
-"	return b3QuatRotate( b3QuatInvert( q ), vec );\n"
-"}\n"
-"inline b3Float4 b3TransformPoint(b3Float4ConstArg point, b3Float4ConstArg translation, b3QuatConstArg  orientation)\n"
-"{\n"
-"	return b3QuatRotate( orientation, point ) + (translation);\n"
-"}\n"
-"	\n"
-"#endif \n"
-"#endif //B3_QUAT_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"typedef struct\n"
-"{\n"
-"	b3Float4 m_row[3];\n"
-"}b3Mat3x3;\n"
-"#define b3Mat3x3ConstArg const b3Mat3x3\n"
-"#define b3GetRow(m,row) (m.m_row[row])\n"
-"inline b3Mat3x3 b3QuatGetRotationMatrix(b3Quat quat)\n"
-"{\n"
-"	b3Float4 quat2 = (b3Float4)(quat.x*quat.x, quat.y*quat.y, quat.z*quat.z, 0.f);\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0].x=1-2*quat2.y-2*quat2.z;\n"
-"	out.m_row[0].y=2*quat.x*quat.y-2*quat.w*quat.z;\n"
-"	out.m_row[0].z=2*quat.x*quat.z+2*quat.w*quat.y;\n"
-"	out.m_row[0].w = 0.f;\n"
-"	out.m_row[1].x=2*quat.x*quat.y+2*quat.w*quat.z;\n"
-"	out.m_row[1].y=1-2*quat2.x-2*quat2.z;\n"
-"	out.m_row[1].z=2*quat.y*quat.z-2*quat.w*quat.x;\n"
-"	out.m_row[1].w = 0.f;\n"
-"	out.m_row[2].x=2*quat.x*quat.z-2*quat.w*quat.y;\n"
-"	out.m_row[2].y=2*quat.y*quat.z+2*quat.w*quat.x;\n"
-"	out.m_row[2].z=1-2*quat2.x-2*quat2.y;\n"
-"	out.m_row[2].w = 0.f;\n"
-"	return out;\n"
-"}\n"
-"inline b3Mat3x3 b3AbsoluteMat3x3(b3Mat3x3ConstArg matIn)\n"
-"{\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0] = fabs(matIn.m_row[0]);\n"
-"	out.m_row[1] = fabs(matIn.m_row[1]);\n"
-"	out.m_row[2] = fabs(matIn.m_row[2]);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtZero();\n"
-"__inline\n"
-"b3Mat3x3 mtIdentity();\n"
-"__inline\n"
-"b3Mat3x3 mtTranspose(b3Mat3x3 m);\n"
-"__inline\n"
-"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b);\n"
-"__inline\n"
-"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b);\n"
-"__inline\n"
-"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b);\n"
-"__inline\n"
-"b3Mat3x3 mtZero()\n"
-"{\n"
-"	b3Mat3x3 m;\n"
-"	m.m_row[0] = (b3Float4)(0.f);\n"
-"	m.m_row[1] = (b3Float4)(0.f);\n"
-"	m.m_row[2] = (b3Float4)(0.f);\n"
-"	return m;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtIdentity()\n"
-"{\n"
-"	b3Mat3x3 m;\n"
-"	m.m_row[0] = (b3Float4)(1,0,0,0);\n"
-"	m.m_row[1] = (b3Float4)(0,1,0,0);\n"
-"	m.m_row[2] = (b3Float4)(0,0,1,0);\n"
-"	return m;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtTranspose(b3Mat3x3 m)\n"
-"{\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0] = (b3Float4)(m.m_row[0].x, m.m_row[1].x, m.m_row[2].x, 0.f);\n"
-"	out.m_row[1] = (b3Float4)(m.m_row[0].y, m.m_row[1].y, m.m_row[2].y, 0.f);\n"
-"	out.m_row[2] = (b3Float4)(m.m_row[0].z, m.m_row[1].z, m.m_row[2].z, 0.f);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b)\n"
-"{\n"
-"	b3Mat3x3 transB;\n"
-"	transB = mtTranspose( b );\n"
-"	b3Mat3x3 ans;\n"
-"	//	why this doesn't run when 0ing in the for{}\n"
-"	a.m_row[0].w = 0.f;\n"
-"	a.m_row[1].w = 0.f;\n"
-"	a.m_row[2].w = 0.f;\n"
-"	for(int i=0; i<3; i++)\n"
-"	{\n"
-"//	a.m_row[i].w = 0.f;\n"
-"		ans.m_row[i].x = b3Dot3F4(a.m_row[i],transB.m_row[0]);\n"
-"		ans.m_row[i].y = b3Dot3F4(a.m_row[i],transB.m_row[1]);\n"
-"		ans.m_row[i].z = b3Dot3F4(a.m_row[i],transB.m_row[2]);\n"
-"		ans.m_row[i].w = 0.f;\n"
-"	}\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b)\n"
-"{\n"
-"	b3Float4 ans;\n"
-"	ans.x = b3Dot3F4( a.m_row[0], b );\n"
-"	ans.y = b3Dot3F4( a.m_row[1], b );\n"
-"	ans.z = b3Dot3F4( a.m_row[2], b );\n"
-"	ans.w = 0.f;\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b)\n"
-"{\n"
-"	b3Float4 colx = b3MakeFloat4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
-"	b3Float4 coly = b3MakeFloat4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
-"	b3Float4 colz = b3MakeFloat4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
-"	b3Float4 ans;\n"
-"	ans.x = b3Dot3F4( a, colx );\n"
-"	ans.y = b3Dot3F4( a, coly );\n"
-"	ans.z = b3Dot3F4( a, colz );\n"
-"	return ans;\n"
-"}\n"
-"#endif\n"
-"#endif //B3_MAT3x3_H\n"
-"typedef struct b3Aabb b3Aabb_t;\n"
-"struct b3Aabb\n"
-"{\n"
-"	union\n"
-"	{\n"
-"		float m_min[4];\n"
-"		b3Float4 m_minVec;\n"
-"		int m_minIndices[4];\n"
-"	};\n"
-"	union\n"
-"	{\n"
-"		float	m_max[4];\n"
-"		b3Float4 m_maxVec;\n"
-"		int m_signedMaxIndices[4];\n"
-"	};\n"
-"};\n"
-"inline void b3TransformAabb2(b3Float4ConstArg localAabbMin,b3Float4ConstArg localAabbMax, float margin,\n"
-"						b3Float4ConstArg pos,\n"
-"						b3QuatConstArg orn,\n"
-"						b3Float4* aabbMinOut,b3Float4* aabbMaxOut)\n"
-"{\n"
-"		b3Float4 localHalfExtents = 0.5f*(localAabbMax-localAabbMin);\n"
-"		localHalfExtents+=b3MakeFloat4(margin,margin,margin,0.f);\n"
-"		b3Float4 localCenter = 0.5f*(localAabbMax+localAabbMin);\n"
-"		b3Mat3x3 m;\n"
-"		m = b3QuatGetRotationMatrix(orn);\n"
-"		b3Mat3x3 abs_b = b3AbsoluteMat3x3(m);\n"
-"		b3Float4 center = b3TransformPoint(localCenter,pos,orn);\n"
-"		\n"
-"		b3Float4 extent = b3MakeFloat4(b3Dot3F4(localHalfExtents,b3GetRow(abs_b,0)),\n"
-"										 b3Dot3F4(localHalfExtents,b3GetRow(abs_b,1)),\n"
-"										 b3Dot3F4(localHalfExtents,b3GetRow(abs_b,2)),\n"
-"										 0.f);\n"
-"		*aabbMinOut = center-extent;\n"
-"		*aabbMaxOut = center+extent;\n"
-"}\n"
-"/// conservative test for overlap between two aabbs\n"
-"inline bool b3TestAabbAgainstAabb(b3Float4ConstArg aabbMin1,b3Float4ConstArg aabbMax1,\n"
-"								b3Float4ConstArg aabbMin2, b3Float4ConstArg aabbMax2)\n"
-"{\n"
-"	bool overlap = true;\n"
-"	overlap = (aabbMin1.x > aabbMax2.x || aabbMax1.x < aabbMin2.x) ? false : overlap;\n"
-"	overlap = (aabbMin1.z > aabbMax2.z || aabbMax1.z < aabbMin2.z) ? false : overlap;\n"
-"	overlap = (aabbMin1.y > aabbMax2.y || aabbMax1.y < aabbMin2.y) ? false : overlap;\n"
-"	return overlap;\n"
-"}\n"
-"#endif //B3_AABB_H\n"
-"/*\n"
-"Bullet Continuous Collision Detection and Physics Library\n"
-"Copyright (c) 2003-2013 Erwin Coumans  http://bulletphysics.org\n"
-"This software is provided 'as-is', without any express or implied warranty.\n"
-"In no event will the authors be held liable for any damages arising from the use of this software.\n"
-"Permission is granted to anyone to use this software for any purpose,\n"
-"including commercial applications, and to alter it and redistribute it freely,\n"
-"subject to the following restrictions:\n"
-"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
-"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
-"3. This notice may not be removed or altered from any source distribution.\n"
-"*/\n"
-"#ifndef B3_INT2_H\n"
-"#define B3_INT2_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#define b3UnsignedInt2 uint2\n"
-"#define b3Int2 int2\n"
-"#define b3MakeInt2 (int2)\n"
-"#endif //__cplusplus\n"
-"#endif\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_plane;\n"
-"	int m_indexOffset;\n"
-"	int m_numIndices;\n"
-"} btGpuFace;\n"
-"#define make_float4 (float4)\n"
-"__inline\n"
-"float4 cross3(float4 a, float4 b)\n"
-"{\n"
-"	return cross(a,b);\n"
-"	\n"
-"//	float4 a1 = make_float4(a.xyz,0.f);\n"
-"//	float4 b1 = make_float4(b.xyz,0.f);\n"
-"//	return cross(a1,b1);\n"
-"//float4 c = make_float4(a.y*b.z - a.z*b.y,a.z*b.x - a.x*b.z,a.x*b.y - a.y*b.x,0.f);\n"
-"	\n"
-"	//	float4 c = make_float4(a.y*b.z - a.z*b.y,1.f,a.x*b.y - a.y*b.x,0.f);\n"
-"	\n"
-"	//return c;\n"
-"}\n"
-"__inline\n"
-"float dot3F4(float4 a, float4 b)\n"
-"{\n"
-"	float4 a1 = make_float4(a.xyz,0.f);\n"
-"	float4 b1 = make_float4(b.xyz,0.f);\n"
-"	return dot(a1, b1);\n"
-"}\n"
-"__inline\n"
-"float4 fastNormalize4(float4 v)\n"
-"{\n"
-"	v = make_float4(v.xyz,0.f);\n"
-"	return fast_normalize(v);\n"
-"}\n"
-"///////////////////////////////////////\n"
-"//	Quaternion\n"
-"///////////////////////////////////////\n"
-"typedef float4 Quaternion;\n"
-"__inline\n"
-"Quaternion qtMul(Quaternion a, Quaternion b);\n"
-"__inline\n"
-"Quaternion qtNormalize(Quaternion in);\n"
-"__inline\n"
-"float4 qtRotate(Quaternion q, float4 vec);\n"
-"__inline\n"
-"Quaternion qtInvert(Quaternion q);\n"
-"__inline\n"
-"Quaternion qtMul(Quaternion a, Quaternion b)\n"
-"{\n"
-"	Quaternion ans;\n"
-"	ans = cross3( a, b );\n"
-"	ans += a.w*b+b.w*a;\n"
-"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
-"	ans.w = a.w*b.w - dot3F4(a, b);\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"Quaternion qtNormalize(Quaternion in)\n"
-"{\n"
-"	return fastNormalize4(in);\n"
-"//	in /= length( in );\n"
-"//	return in;\n"
-"}\n"
-"__inline\n"
-"float4 qtRotate(Quaternion q, float4 vec)\n"
-"{\n"
-"	Quaternion qInv = qtInvert( q );\n"
-"	float4 vcpy = vec;\n"
-"	vcpy.w = 0.f;\n"
-"	float4 out = qtMul(qtMul(q,vcpy),qInv);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"Quaternion qtInvert(Quaternion q)\n"
-"{\n"
-"	return (Quaternion)(-q.xyz, q.w);\n"
-"}\n"
-"__inline\n"
-"float4 qtInvRotate(const Quaternion q, float4 vec)\n"
-"{\n"
-"	return qtRotate( qtInvert( q ), vec );\n"
-"}\n"
-"__inline\n"
-"float4 transform(const float4* p, const float4* translation, const Quaternion* orientation)\n"
-"{\n"
-"	return qtRotate( *orientation, *p ) + (*translation);\n"
-"}\n"
-"__inline\n"
-"float4 normalize3(const float4 a)\n"
-"{\n"
-"	float4 n = make_float4(a.x, a.y, a.z, 0.f);\n"
-"	return fastNormalize4( n );\n"
-"}\n"
-"inline void projectLocal(const ConvexPolyhedronCL* hull,  const float4 pos, const float4 orn, \n"
-"const float4* dir, const float4* vertices, float* min, float* max)\n"
-"{\n"
-"	min[0] = FLT_MAX;\n"
-"	max[0] = -FLT_MAX;\n"
-"	int numVerts = hull->m_numVertices;\n"
-"	const float4 localDir = qtInvRotate(orn,*dir);\n"
-"	float offset = dot(pos,*dir);\n"
-"	for(int i=0;i<numVerts;i++)\n"
-"	{\n"
-"		float dp = dot(vertices[hull->m_vertexOffset+i],localDir);\n"
-"		if(dp < min[0])	\n"
-"			min[0] = dp;\n"
-"		if(dp > max[0])	\n"
-"			max[0] = dp;\n"
-"	}\n"
-"	if(min[0]>max[0])\n"
-"	{\n"
-"		float tmp = min[0];\n"
-"		min[0] = max[0];\n"
-"		max[0] = tmp;\n"
-"	}\n"
-"	min[0] += offset;\n"
-"	max[0] += offset;\n"
-"}\n"
-"inline void project(__global const ConvexPolyhedronCL* hull,  const float4 pos, const float4 orn, \n"
-"const float4* dir, __global const float4* vertices, float* min, float* max)\n"
-"{\n"
-"	min[0] = FLT_MAX;\n"
-"	max[0] = -FLT_MAX;\n"
-"	int numVerts = hull->m_numVertices;\n"
-"	const float4 localDir = qtInvRotate(orn,*dir);\n"
-"	float offset = dot(pos,*dir);\n"
-"	for(int i=0;i<numVerts;i++)\n"
-"	{\n"
-"		float dp = dot(vertices[hull->m_vertexOffset+i],localDir);\n"
-"		if(dp < min[0])	\n"
-"			min[0] = dp;\n"
-"		if(dp > max[0])	\n"
-"			max[0] = dp;\n"
-"	}\n"
-"	if(min[0]>max[0])\n"
-"	{\n"
-"		float tmp = min[0];\n"
-"		min[0] = max[0];\n"
-"		max[0] = tmp;\n"
-"	}\n"
-"	min[0] += offset;\n"
-"	max[0] += offset;\n"
-"}\n"
-"inline bool TestSepAxisLocalA(const ConvexPolyhedronCL* hullA, __global const ConvexPolyhedronCL* hullB, \n"
-"	const float4 posA,const float4 ornA,\n"
-"	const float4 posB,const float4 ornB,\n"
-"	float4* sep_axis, const float4* verticesA, __global const float4* verticesB,float* depth)\n"
-"{\n"
-"	float Min0,Max0;\n"
-"	float Min1,Max1;\n"
-"	projectLocal(hullA,posA,ornA,sep_axis,verticesA, &Min0, &Max0);\n"
-"	project(hullB,posB,ornB, sep_axis,verticesB, &Min1, &Max1);\n"
-"	if(Max0<Min1 || Max1<Min0)\n"
-"		return false;\n"
-"	float d0 = Max0 - Min1;\n"
-"	float d1 = Max1 - Min0;\n"
-"	*depth = d0<d1 ? d0:d1;\n"
-"	return true;\n"
-"}\n"
-"inline bool IsAlmostZero(const float4 v)\n"
-"{\n"
-"	if(fabs(v.x)>1e-6f || fabs(v.y)>1e-6f || fabs(v.z)>1e-6f)\n"
-"		return false;\n"
-"	return true;\n"
-"}\n"
-"bool findSeparatingAxisLocalA(	const ConvexPolyhedronCL* hullA, __global const ConvexPolyhedronCL* hullB, \n"
-"	const float4 posA1,\n"
-"	const float4 ornA,\n"
-"	const float4 posB1,\n"
-"	const float4 ornB,\n"
-"	const float4 DeltaC2,\n"
-"	\n"
-"	const float4* verticesA, \n"
-"	const float4* uniqueEdgesA, \n"
-"	const btGpuFace* facesA,\n"
-"	const int*  indicesA,\n"
-"	__global const float4* verticesB, \n"
-"	__global const float4* uniqueEdgesB, \n"
-"	__global const btGpuFace* facesB,\n"
-"	__global const int*  indicesB,\n"
-"	float4* sep,\n"
-"	float* dmin)\n"
-"{\n"
-"	\n"
-"	float4 posA = posA1;\n"
-"	posA.w = 0.f;\n"
-"	float4 posB = posB1;\n"
-"	posB.w = 0.f;\n"
-"	int curPlaneTests=0;\n"
-"	{\n"
-"		int numFacesA = hullA->m_numFaces;\n"
-"		// Test normals from hullA\n"
-"		for(int i=0;i<numFacesA;i++)\n"
-"		{\n"
-"			const float4 normal = facesA[hullA->m_faceOffset+i].m_plane;\n"
-"			float4 faceANormalWS = qtRotate(ornA,normal);\n"
-"			if (dot3F4(DeltaC2,faceANormalWS)<0)\n"
-"				faceANormalWS*=-1.f;\n"
-"			curPlaneTests++;\n"
-"			float d;\n"
-"			if(!TestSepAxisLocalA( hullA, hullB, posA,ornA,posB,ornB,&faceANormalWS, verticesA, verticesB,&d))\n"
-"				return false;\n"
-"			if(d<*dmin)\n"
-"			{\n"
-"				*dmin = d;\n"
-"				*sep = faceANormalWS;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	if((dot3F4(-DeltaC2,*sep))>0.0f)\n"
-"	{\n"
-"		*sep = -(*sep);\n"
-"	}\n"
-"	return true;\n"
-"}\n"
-"bool findSeparatingAxisLocalB(	__global const ConvexPolyhedronCL* hullA,  const ConvexPolyhedronCL* hullB, \n"
-"	const float4 posA1,\n"
-"	const float4 ornA,\n"
-"	const float4 posB1,\n"
-"	const float4 ornB,\n"
-"	const float4 DeltaC2,\n"
-"	__global const float4* verticesA, \n"
-"	__global const float4* uniqueEdgesA, \n"
-"	__global const btGpuFace* facesA,\n"
-"	__global const int*  indicesA,\n"
-"	const float4* verticesB,\n"
-"	const float4* uniqueEdgesB, \n"
-"	const btGpuFace* facesB,\n"
-"	const int*  indicesB,\n"
-"	float4* sep,\n"
-"	float* dmin)\n"
-"{\n"
-"	float4 posA = posA1;\n"
-"	posA.w = 0.f;\n"
-"	float4 posB = posB1;\n"
-"	posB.w = 0.f;\n"
-"	int curPlaneTests=0;\n"
-"	{\n"
-"		int numFacesA = hullA->m_numFaces;\n"
-"		// Test normals from hullA\n"
-"		for(int i=0;i<numFacesA;i++)\n"
-"		{\n"
-"			const float4 normal = facesA[hullA->m_faceOffset+i].m_plane;\n"
-"			float4 faceANormalWS = qtRotate(ornA,normal);\n"
-"			if (dot3F4(DeltaC2,faceANormalWS)<0)\n"
-"				faceANormalWS *= -1.f;\n"
-"			curPlaneTests++;\n"
-"			float d;\n"
-"			if(!TestSepAxisLocalA( hullB, hullA, posB,ornB,posA,ornA, &faceANormalWS, verticesB,verticesA, &d))\n"
-"				return false;\n"
-"			if(d<*dmin)\n"
-"			{\n"
-"				*dmin = d;\n"
-"				*sep = faceANormalWS;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	if((dot3F4(-DeltaC2,*sep))>0.0f)\n"
-"	{\n"
-"		*sep = -(*sep);\n"
-"	}\n"
-"	return true;\n"
-"}\n"
-"bool findSeparatingAxisEdgeEdgeLocalA(	const ConvexPolyhedronCL* hullA, __global const ConvexPolyhedronCL* hullB, \n"
-"	const float4 posA1,\n"
-"	const float4 ornA,\n"
-"	const float4 posB1,\n"
-"	const float4 ornB,\n"
-"	const float4 DeltaC2,\n"
-"	const float4* verticesA, \n"
-"	const float4* uniqueEdgesA, \n"
-"	const btGpuFace* facesA,\n"
-"	const int*  indicesA,\n"
-"	__global const float4* verticesB, \n"
-"	__global const float4* uniqueEdgesB, \n"
-"	__global const btGpuFace* facesB,\n"
-"	__global const int*  indicesB,\n"
-"		float4* sep,\n"
-"	float* dmin)\n"
-"{\n"
-"	float4 posA = posA1;\n"
-"	posA.w = 0.f;\n"
-"	float4 posB = posB1;\n"
-"	posB.w = 0.f;\n"
-"	int curPlaneTests=0;\n"
-"	int curEdgeEdge = 0;\n"
-"	// Test edges\n"
-"	for(int e0=0;e0<hullA->m_numUniqueEdges;e0++)\n"
-"	{\n"
-"		const float4 edge0 = uniqueEdgesA[hullA->m_uniqueEdgesOffset+e0];\n"
-"		float4 edge0World = qtRotate(ornA,edge0);\n"
-"		for(int e1=0;e1<hullB->m_numUniqueEdges;e1++)\n"
-"		{\n"
-"			const float4 edge1 = uniqueEdgesB[hullB->m_uniqueEdgesOffset+e1];\n"
-"			float4 edge1World = qtRotate(ornB,edge1);\n"
-"			float4 crossje = cross3(edge0World,edge1World);\n"
-"			curEdgeEdge++;\n"
-"			if(!IsAlmostZero(crossje))\n"
-"			{\n"
-"				crossje = normalize3(crossje);\n"
-"				if (dot3F4(DeltaC2,crossje)<0)\n"
-"					crossje *= -1.f;\n"
-"				float dist;\n"
-"				bool result = true;\n"
-"				{\n"
-"					float Min0,Max0;\n"
-"					float Min1,Max1;\n"
-"					projectLocal(hullA,posA,ornA,&crossje,verticesA, &Min0, &Max0);\n"
-"					project(hullB,posB,ornB,&crossje,verticesB, &Min1, &Max1);\n"
-"				\n"
-"					if(Max0<Min1 || Max1<Min0)\n"
-"						result = false;\n"
-"				\n"
-"					float d0 = Max0 - Min1;\n"
-"					float d1 = Max1 - Min0;\n"
-"					dist = d0<d1 ? d0:d1;\n"
-"					result = true;\n"
-"				}\n"
-"				\n"
-"				if(dist<*dmin)\n"
-"				{\n"
-"					*dmin = dist;\n"
-"					*sep = crossje;\n"
-"				}\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	\n"
-"	if((dot3F4(-DeltaC2,*sep))>0.0f)\n"
-"	{\n"
-"		*sep = -(*sep);\n"
-"	}\n"
-"	return true;\n"
-"}\n"
-"inline int	findClippingFaces(const float4 separatingNormal,\n"
-"                      const ConvexPolyhedronCL* hullA, \n"
-"					  __global const ConvexPolyhedronCL* hullB,\n"
-"                      const float4 posA, const Quaternion ornA,const float4 posB, const Quaternion ornB,\n"
-"                       __global float4* worldVertsA1,\n"
-"                      __global float4* worldNormalsA1,\n"
-"                      __global float4* worldVertsB1,\n"
-"                      int capacityWorldVerts,\n"
-"                      const float minDist, float maxDist,\n"
-"					  const float4* verticesA,\n"
-"                      const btGpuFace* facesA,\n"
-"                      const int* indicesA,\n"
-"					  __global const float4* verticesB,\n"
-"                      __global const btGpuFace* facesB,\n"
-"                      __global const int* indicesB,\n"
-"                      __global int4* clippingFaces, int pairIndex)\n"
-"{\n"
-"	int numContactsOut = 0;\n"
-"	int numWorldVertsB1= 0;\n"
-"    \n"
-"    \n"
-"	int closestFaceB=0;\n"
-"	float dmax = -FLT_MAX;\n"
-"    \n"
-"	{\n"
-"		for(int face=0;face<hullB->m_numFaces;face++)\n"
-"		{\n"
-"			const float4 Normal = make_float4(facesB[hullB->m_faceOffset+face].m_plane.x,\n"
-"                                              facesB[hullB->m_faceOffset+face].m_plane.y, facesB[hullB->m_faceOffset+face].m_plane.z,0.f);\n"
-"			const float4 WorldNormal = qtRotate(ornB, Normal);\n"
-"			float d = dot3F4(WorldNormal,separatingNormal);\n"
-"			if (d > dmax)\n"
-"			{\n"
-"				dmax = d;\n"
-"				closestFaceB = face;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"    \n"
-"	{\n"
-"		const btGpuFace polyB = facesB[hullB->m_faceOffset+closestFaceB];\n"
-"		int numVertices = polyB.m_numIndices;\n"
-"        if (numVertices>capacityWorldVerts)\n"
-"            numVertices = capacityWorldVerts;\n"
-"        if (numVertices<0)\n"
-"            numVertices = 0;\n"
-"        \n"
-"		for(int e0=0;e0<numVertices;e0++)\n"
-"		{\n"
-"            if (e0<capacityWorldVerts)\n"
-"            {\n"
-"                const float4 b = verticesB[hullB->m_vertexOffset+indicesB[polyB.m_indexOffset+e0]];\n"
-"                worldVertsB1[pairIndex*capacityWorldVerts+numWorldVertsB1++] = transform(&b,&posB,&ornB);\n"
-"            }\n"
-"		}\n"
-"	}\n"
-"    \n"
-"    int closestFaceA=0;\n"
-"	{\n"
-"		float dmin = FLT_MAX;\n"
-"		for(int face=0;face<hullA->m_numFaces;face++)\n"
-"		{\n"
-"			const float4 Normal = make_float4(\n"
-"                                              facesA[hullA->m_faceOffset+face].m_plane.x,\n"
-"                                              facesA[hullA->m_faceOffset+face].m_plane.y,\n"
-"                                              facesA[hullA->m_faceOffset+face].m_plane.z,\n"
-"                                              0.f);\n"
-"			const float4 faceANormalWS = qtRotate(ornA,Normal);\n"
-"            \n"
-"			float d = dot3F4(faceANormalWS,separatingNormal);\n"
-"			if (d < dmin)\n"
-"			{\n"
-"				dmin = d;\n"
-"				closestFaceA = face;\n"
-"                worldNormalsA1[pairIndex] = faceANormalWS;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"    \n"
-"    int numVerticesA = facesA[hullA->m_faceOffset+closestFaceA].m_numIndices;\n"
-"    if (numVerticesA>capacityWorldVerts)\n"
-"       numVerticesA = capacityWorldVerts;\n"
-"    if (numVerticesA<0)\n"
-"        numVerticesA=0;\n"
-"    \n"
-"	for(int e0=0;e0<numVerticesA;e0++)\n"
-"	{\n"
-"        if (e0<capacityWorldVerts)\n"
-"        {\n"
-"            const float4 a = verticesA[hullA->m_vertexOffset+indicesA[facesA[hullA->m_faceOffset+closestFaceA].m_indexOffset+e0]];\n"
-"            worldVertsA1[pairIndex*capacityWorldVerts+e0] = transform(&a, &posA,&ornA);\n"
-"        }\n"
-"    }\n"
-"    \n"
-"    clippingFaces[pairIndex].x = closestFaceA;\n"
-"    clippingFaces[pairIndex].y = closestFaceB;\n"
-"    clippingFaces[pairIndex].z = numVerticesA;\n"
-"    clippingFaces[pairIndex].w = numWorldVertsB1;\n"
-"    \n"
-"    \n"
-"	return numContactsOut;\n"
-"}\n"
-"// work-in-progress\n"
-"__kernel void   findConcaveSeparatingAxisVertexFaceKernel( __global int4* concavePairs,\n"
-"                                                __global const BodyData* rigidBodies,\n"
-"                                                __global const btCollidableGpu* collidables,\n"
-"                                                __global const ConvexPolyhedronCL* convexShapes,\n"
-"                                                __global const float4* vertices,\n"
-"                                                __global const float4* uniqueEdges,\n"
-"                                                __global const btGpuFace* faces,\n"
-"                                                __global const int* indices,\n"
-"                                                __global const btGpuChildShape* gpuChildShapes,\n"
-"                                                __global btAabbCL* aabbs,\n"
-"                                                __global float4* concaveSeparatingNormalsOut,\n"
-"                                                __global int* concaveHasSeparatingNormals,\n"
-"                                                __global int4* clippingFacesOut,\n"
-"                                                __global float4* worldVertsA1GPU,\n"
-"                                                __global float4*  worldNormalsAGPU,\n"
-"                                                __global float4* worldVertsB1GPU,\n"
-"                                                __global float* dmins,\n"
-"                                                int vertexFaceCapacity,\n"
-"                                                int numConcavePairs\n"
-"                                                )\n"
-"{\n"
-"    \n"
-"	int i = get_global_id(0);\n"
-"	if (i>=numConcavePairs)\n"
-"		return;\n"
-"    \n"
-"	concaveHasSeparatingNormals[i] = 0;\n"
-"    \n"
-"	int pairIdx = i;\n"
-"    \n"
-"	int bodyIndexA = concavePairs[i].x;\n"
-"	int bodyIndexB = concavePairs[i].y;\n"
-"    \n"
-"	int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
-"	int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
-"    \n"
-"	int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
-"	int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
-"    \n"
-"	if (collidables[collidableIndexB].m_shapeType!=SHAPE_CONVEX_HULL&&\n"
-"		collidables[collidableIndexB].m_shapeType!=SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
-"	{\n"
-"		concavePairs[pairIdx].w = -1;\n"
-"		return;\n"
-"	}\n"
-"    \n"
-"    \n"
-"    \n"
-"	int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
-"	int numActualConcaveConvexTests = 0;\n"
-"	\n"
-"	int f = concavePairs[i].z;\n"
-"	\n"
-"	bool overlap = false;\n"
-"	\n"
-"	ConvexPolyhedronCL convexPolyhedronA;\n"
-"    \n"
-"	//add 3 vertices of the triangle\n"
-"	convexPolyhedronA.m_numVertices = 3;\n"
-"	convexPolyhedronA.m_vertexOffset = 0;\n"
-"	float4	localCenter = make_float4(0.f,0.f,0.f,0.f);\n"
-"    \n"
-"	btGpuFace face = faces[convexShapes[shapeIndexA].m_faceOffset+f];\n"
-"	float4 triMinAabb, triMaxAabb;\n"
-"	btAabbCL triAabb;\n"
-"	triAabb.m_min = make_float4(1e30f,1e30f,1e30f,0.f);\n"
-"	triAabb.m_max = make_float4(-1e30f,-1e30f,-1e30f,0.f);\n"
-"	\n"
-"	float4 verticesA[3];\n"
-"	for (int i=0;i<3;i++)\n"
-"	{\n"
-"		int index = indices[face.m_indexOffset+i];\n"
-"		float4 vert = vertices[convexShapes[shapeIndexA].m_vertexOffset+index];\n"
-"		verticesA[i] = vert;\n"
-"		localCenter += vert;\n"
-"        \n"
-"		triAabb.m_min = min(triAabb.m_min,vert);\n"
-"		triAabb.m_max = max(triAabb.m_max,vert);\n"
-"        \n"
-"	}\n"
-"    \n"
-"	overlap = true;\n"
-"	overlap = (triAabb.m_min.x > aabbs[bodyIndexB].m_max.x || triAabb.m_max.x < aabbs[bodyIndexB].m_min.x) ? false : overlap;\n"
-"	overlap = (triAabb.m_min.z > aabbs[bodyIndexB].m_max.z || triAabb.m_max.z < aabbs[bodyIndexB].m_min.z) ? false : overlap;\n"
-"	overlap = (triAabb.m_min.y > aabbs[bodyIndexB].m_max.y || triAabb.m_max.y < aabbs[bodyIndexB].m_min.y) ? false : overlap;\n"
-"    \n"
-"	if (overlap)\n"
-"	{\n"
-"		float dmin = FLT_MAX;\n"
-"		int hasSeparatingAxis=5;\n"
-"		float4 sepAxis=make_float4(1,2,3,4);\n"
-"        \n"
-"		int localCC=0;\n"
-"		numActualConcaveConvexTests++;\n"
-"        \n"
-"		//a triangle has 3 unique edges\n"
-"		convexPolyhedronA.m_numUniqueEdges = 3;\n"
-"		convexPolyhedronA.m_uniqueEdgesOffset = 0;\n"
-"		float4 uniqueEdgesA[3];\n"
-"		\n"
-"		uniqueEdgesA[0] = (verticesA[1]-verticesA[0]);\n"
-"		uniqueEdgesA[1] = (verticesA[2]-verticesA[1]);\n"
-"		uniqueEdgesA[2] = (verticesA[0]-verticesA[2]);\n"
-"        \n"
-"        \n"
-"		convexPolyhedronA.m_faceOffset = 0;\n"
-"        \n"
-"		float4 normal = make_float4(face.m_plane.x,face.m_plane.y,face.m_plane.z,0.f);\n"
-"        \n"
-"		btGpuFace facesA[TRIANGLE_NUM_CONVEX_FACES];\n"
-"		int indicesA[3+3+2+2+2];\n"
-"		int curUsedIndices=0;\n"
-"		int fidx=0;\n"
-"        \n"
-"		//front size of triangle\n"
-"		{\n"
-"			facesA[fidx].m_indexOffset=curUsedIndices;\n"
-"			indicesA[0] = 0;\n"
-"			indicesA[1] = 1;\n"
-"			indicesA[2] = 2;\n"
-"			curUsedIndices+=3;\n"
-"			float c = face.m_plane.w;\n"
-"			facesA[fidx].m_plane.x = normal.x;\n"
-"			facesA[fidx].m_plane.y = normal.y;\n"
-"			facesA[fidx].m_plane.z = normal.z;\n"
-"			facesA[fidx].m_plane.w = c;\n"
-"			facesA[fidx].m_numIndices=3;\n"
-"		}\n"
-"		fidx++;\n"
-"		//back size of triangle\n"
-"		{\n"
-"			facesA[fidx].m_indexOffset=curUsedIndices;\n"
-"			indicesA[3]=2;\n"
-"			indicesA[4]=1;\n"
-"			indicesA[5]=0;\n"
-"			curUsedIndices+=3;\n"
-"			float c = dot(normal,verticesA[0]);\n"
-"			float c1 = -face.m_plane.w;\n"
-"			facesA[fidx].m_plane.x = -normal.x;\n"
-"			facesA[fidx].m_plane.y = -normal.y;\n"
-"			facesA[fidx].m_plane.z = -normal.z;\n"
-"			facesA[fidx].m_plane.w = c;\n"
-"			facesA[fidx].m_numIndices=3;\n"
-"		}\n"
-"		fidx++;\n"
-"        \n"
-"		bool addEdgePlanes = true;\n"
-"		if (addEdgePlanes)\n"
-"		{\n"
-"			int numVertices=3;\n"
-"			int prevVertex = numVertices-1;\n"
-"			for (int i=0;i<numVertices;i++)\n"
-"			{\n"
-"				float4 v0 = verticesA[i];\n"
-"				float4 v1 = verticesA[prevVertex];\n"
-"                \n"
-"				float4 edgeNormal = normalize(cross(normal,v1-v0));\n"
-"				float c = -dot(edgeNormal,v0);\n"
-"                \n"
-"				facesA[fidx].m_numIndices = 2;\n"
-"				facesA[fidx].m_indexOffset=curUsedIndices;\n"
-"				indicesA[curUsedIndices++]=i;\n"
-"				indicesA[curUsedIndices++]=prevVertex;\n"
-"                \n"
-"				facesA[fidx].m_plane.x = edgeNormal.x;\n"
-"				facesA[fidx].m_plane.y = edgeNormal.y;\n"
-"				facesA[fidx].m_plane.z = edgeNormal.z;\n"
-"				facesA[fidx].m_plane.w = c;\n"
-"				fidx++;\n"
-"				prevVertex = i;\n"
-"			}\n"
-"		}\n"
-"		convexPolyhedronA.m_numFaces = TRIANGLE_NUM_CONVEX_FACES;\n"
-"		convexPolyhedronA.m_localCenter = localCenter*(1.f/3.f);\n"
-"        \n"
-"        \n"
-"		float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
-"		posA.w = 0.f;\n"
-"		float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
-"		posB.w = 0.f;\n"
-"        \n"
-"		float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
-"		float4 ornB =rigidBodies[bodyIndexB].m_quat;\n"
-"        \n"
-"		\n"
-"        \n"
-"        \n"
-"		///////////////////\n"
-"		///compound shape support\n"
-"        \n"
-"		if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
-"		{\n"
-"			int compoundChild = concavePairs[pairIdx].w;\n"
-"			int childShapeIndexB = compoundChild;//collidables[collidableIndexB].m_shapeIndex+compoundChild;\n"
-"			int childColIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;\n"
-"			float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;\n"
-"			float4 childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;\n"
-"			float4 newPosB = transform(&childPosB,&posB,&ornB);\n"
-"			float4 newOrnB = qtMul(ornB,childOrnB);\n"
-"			posB = newPosB;\n"
-"			ornB = newOrnB;\n"
-"			shapeIndexB = collidables[childColIndexB].m_shapeIndex;\n"
-"		}\n"
-"		//////////////////\n"
-"        \n"
-"		float4 c0local = convexPolyhedronA.m_localCenter;\n"
-"		float4 c0 = transform(&c0local, &posA, &ornA);\n"
-"		float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
-"		float4 c1 = transform(&c1local,&posB,&ornB);\n"
-"		const float4 DeltaC2 = c0 - c1;\n"
-"        \n"
-"        \n"
-"		bool sepA = findSeparatingAxisLocalA(	&convexPolyhedronA, &convexShapes[shapeIndexB],\n"
-"                                             posA,ornA,\n"
-"                                             posB,ornB,\n"
-"                                             DeltaC2,\n"
-"                                             verticesA,uniqueEdgesA,facesA,indicesA,\n"
-"                                             vertices,uniqueEdges,faces,indices,\n"
-"                                             &sepAxis,&dmin);\n"
-"		hasSeparatingAxis = 4;\n"
-"		if (!sepA)\n"
-"		{\n"
-"			hasSeparatingAxis = 0;\n"
-"		} else\n"
-"		{\n"
-"			bool sepB = findSeparatingAxisLocalB(	&convexShapes[shapeIndexB],&convexPolyhedronA,\n"
-"                                                 posB,ornB,\n"
-"                                                 posA,ornA,\n"
-"                                                 DeltaC2,\n"
-"                                                 vertices,uniqueEdges,faces,indices,\n"
-"                                                 verticesA,uniqueEdgesA,facesA,indicesA,\n"
-"                                                 &sepAxis,&dmin);\n"
-"            \n"
-"			if (!sepB)\n"
-"			{\n"
-"				hasSeparatingAxis = 0;\n"
-"			} else\n"
-"			{\n"
-"				hasSeparatingAxis = 1;\n"
-"			}\n"
-"		}	\n"
-"		\n"
-"		if (hasSeparatingAxis)\n"
-"		{\n"
-"            dmins[i] = dmin;\n"
-"			concaveSeparatingNormalsOut[pairIdx]=sepAxis;\n"
-"			concaveHasSeparatingNormals[i]=1;\n"
-"            \n"
-"		} else\n"
-"		{	\n"
-"			//mark this pair as in-active\n"
-"			concavePairs[pairIdx].w = -1;\n"
-"		}\n"
-"	}\n"
-"	else\n"
-"	{	\n"
-"		//mark this pair as in-active\n"
-"		concavePairs[pairIdx].w = -1;\n"
-"	}\n"
-"}\n"
-"// work-in-progress\n"
-"__kernel void   findConcaveSeparatingAxisEdgeEdgeKernel( __global int4* concavePairs,\n"
-"                                                          __global const BodyData* rigidBodies,\n"
-"                                                          __global const btCollidableGpu* collidables,\n"
-"                                                          __global const ConvexPolyhedronCL* convexShapes,\n"
-"                                                          __global const float4* vertices,\n"
-"                                                          __global const float4* uniqueEdges,\n"
-"                                                          __global const btGpuFace* faces,\n"
-"                                                          __global const int* indices,\n"
-"                                                          __global const btGpuChildShape* gpuChildShapes,\n"
-"                                                          __global btAabbCL* aabbs,\n"
-"                                                          __global float4* concaveSeparatingNormalsOut,\n"
-"                                                          __global int* concaveHasSeparatingNormals,\n"
-"                                                          __global int4* clippingFacesOut,\n"
-"                                                          __global float4* worldVertsA1GPU,\n"
-"                                                          __global float4*  worldNormalsAGPU,\n"
-"                                                          __global float4* worldVertsB1GPU,\n"
-"                                                          __global float* dmins,\n"
-"                                                          int vertexFaceCapacity,\n"
-"                                                          int numConcavePairs\n"
-"                                                          )\n"
-"{\n"
-"    \n"
-"	int i = get_global_id(0);\n"
-"	if (i>=numConcavePairs)\n"
-"		return;\n"
-"    \n"
-"	if (!concaveHasSeparatingNormals[i])\n"
-"        return;\n"
-"    \n"
-"	int pairIdx = i;\n"
-"    \n"
-"	int bodyIndexA = concavePairs[i].x;\n"
-"	int bodyIndexB = concavePairs[i].y;\n"
-"    \n"
-"	int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
-"	int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
-"    \n"
-"	int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
-"	int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
-"    \n"
-"    \n"
-"	int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
-"	int numActualConcaveConvexTests = 0;\n"
-"	\n"
-"	int f = concavePairs[i].z;\n"
-"	\n"
-"	bool overlap = false;\n"
-"	\n"
-"	ConvexPolyhedronCL convexPolyhedronA;\n"
-"    \n"
-"	//add 3 vertices of the triangle\n"
-"	convexPolyhedronA.m_numVertices = 3;\n"
-"	convexPolyhedronA.m_vertexOffset = 0;\n"
-"	float4	localCenter = make_float4(0.f,0.f,0.f,0.f);\n"
-"    \n"
-"	btGpuFace face = faces[convexShapes[shapeIndexA].m_faceOffset+f];\n"
-"	float4 triMinAabb, triMaxAabb;\n"
-"	btAabbCL triAabb;\n"
-"	triAabb.m_min = make_float4(1e30f,1e30f,1e30f,0.f);\n"
-"	triAabb.m_max = make_float4(-1e30f,-1e30f,-1e30f,0.f);\n"
-"	\n"
-"	float4 verticesA[3];\n"
-"	for (int i=0;i<3;i++)\n"
-"	{\n"
-"		int index = indices[face.m_indexOffset+i];\n"
-"		float4 vert = vertices[convexShapes[shapeIndexA].m_vertexOffset+index];\n"
-"		verticesA[i] = vert;\n"
-"		localCenter += vert;\n"
-"        \n"
-"		triAabb.m_min = min(triAabb.m_min,vert);\n"
-"		triAabb.m_max = max(triAabb.m_max,vert);\n"
-"        \n"
-"	}\n"
-"    \n"
-"	overlap = true;\n"
-"	overlap = (triAabb.m_min.x > aabbs[bodyIndexB].m_max.x || triAabb.m_max.x < aabbs[bodyIndexB].m_min.x) ? false : overlap;\n"
-"	overlap = (triAabb.m_min.z > aabbs[bodyIndexB].m_max.z || triAabb.m_max.z < aabbs[bodyIndexB].m_min.z) ? false : overlap;\n"
-"	overlap = (triAabb.m_min.y > aabbs[bodyIndexB].m_max.y || triAabb.m_max.y < aabbs[bodyIndexB].m_min.y) ? false : overlap;\n"
-"    \n"
-"	if (overlap)\n"
-"	{\n"
-"		float dmin = dmins[i];\n"
-"		int hasSeparatingAxis=5;\n"
-"		float4 sepAxis=make_float4(1,2,3,4);\n"
-"        sepAxis = concaveSeparatingNormalsOut[pairIdx];\n"
-"        \n"
-"		int localCC=0;\n"
-"		numActualConcaveConvexTests++;\n"
-"        \n"
-"		//a triangle has 3 unique edges\n"
-"		convexPolyhedronA.m_numUniqueEdges = 3;\n"
-"		convexPolyhedronA.m_uniqueEdgesOffset = 0;\n"
-"		float4 uniqueEdgesA[3];\n"
-"		\n"
-"		uniqueEdgesA[0] = (verticesA[1]-verticesA[0]);\n"
-"		uniqueEdgesA[1] = (verticesA[2]-verticesA[1]);\n"
-"		uniqueEdgesA[2] = (verticesA[0]-verticesA[2]);\n"
-"        \n"
-"        \n"
-"		convexPolyhedronA.m_faceOffset = 0;\n"
-"        \n"
-"		float4 normal = make_float4(face.m_plane.x,face.m_plane.y,face.m_plane.z,0.f);\n"
-"        \n"
-"		btGpuFace facesA[TRIANGLE_NUM_CONVEX_FACES];\n"
-"		int indicesA[3+3+2+2+2];\n"
-"		int curUsedIndices=0;\n"
-"		int fidx=0;\n"
-"        \n"
-"		//front size of triangle\n"
-"		{\n"
-"			facesA[fidx].m_indexOffset=curUsedIndices;\n"
-"			indicesA[0] = 0;\n"
-"			indicesA[1] = 1;\n"
-"			indicesA[2] = 2;\n"
-"			curUsedIndices+=3;\n"
-"			float c = face.m_plane.w;\n"
-"			facesA[fidx].m_plane.x = normal.x;\n"
-"			facesA[fidx].m_plane.y = normal.y;\n"
-"			facesA[fidx].m_plane.z = normal.z;\n"
-"			facesA[fidx].m_plane.w = c;\n"
-"			facesA[fidx].m_numIndices=3;\n"
-"		}\n"
-"		fidx++;\n"
-"		//back size of triangle\n"
-"		{\n"
-"			facesA[fidx].m_indexOffset=curUsedIndices;\n"
-"			indicesA[3]=2;\n"
-"			indicesA[4]=1;\n"
-"			indicesA[5]=0;\n"
-"			curUsedIndices+=3;\n"
-"			float c = dot(normal,verticesA[0]);\n"
-"			float c1 = -face.m_plane.w;\n"
-"			facesA[fidx].m_plane.x = -normal.x;\n"
-"			facesA[fidx].m_plane.y = -normal.y;\n"
-"			facesA[fidx].m_plane.z = -normal.z;\n"
-"			facesA[fidx].m_plane.w = c;\n"
-"			facesA[fidx].m_numIndices=3;\n"
-"		}\n"
-"		fidx++;\n"
-"        \n"
-"		bool addEdgePlanes = true;\n"
-"		if (addEdgePlanes)\n"
-"		{\n"
-"			int numVertices=3;\n"
-"			int prevVertex = numVertices-1;\n"
-"			for (int i=0;i<numVertices;i++)\n"
-"			{\n"
-"				float4 v0 = verticesA[i];\n"
-"				float4 v1 = verticesA[prevVertex];\n"
-"                \n"
-"				float4 edgeNormal = normalize(cross(normal,v1-v0));\n"
-"				float c = -dot(edgeNormal,v0);\n"
-"                \n"
-"				facesA[fidx].m_numIndices = 2;\n"
-"				facesA[fidx].m_indexOffset=curUsedIndices;\n"
-"				indicesA[curUsedIndices++]=i;\n"
-"				indicesA[curUsedIndices++]=prevVertex;\n"
-"                \n"
-"				facesA[fidx].m_plane.x = edgeNormal.x;\n"
-"				facesA[fidx].m_plane.y = edgeNormal.y;\n"
-"				facesA[fidx].m_plane.z = edgeNormal.z;\n"
-"				facesA[fidx].m_plane.w = c;\n"
-"				fidx++;\n"
-"				prevVertex = i;\n"
-"			}\n"
-"		}\n"
-"		convexPolyhedronA.m_numFaces = TRIANGLE_NUM_CONVEX_FACES;\n"
-"		convexPolyhedronA.m_localCenter = localCenter*(1.f/3.f);\n"
-"        \n"
-"        \n"
-"		float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
-"		posA.w = 0.f;\n"
-"		float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
-"		posB.w = 0.f;\n"
-"        \n"
-"		float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
-"		float4 ornB =rigidBodies[bodyIndexB].m_quat;\n"
-"        \n"
-"		\n"
-"        \n"
-"        \n"
-"		///////////////////\n"
-"		///compound shape support\n"
-"        \n"
-"		if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
-"		{\n"
-"			int compoundChild = concavePairs[pairIdx].w;\n"
-"			int childShapeIndexB = compoundChild;//collidables[collidableIndexB].m_shapeIndex+compoundChild;\n"
-"			int childColIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;\n"
-"			float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;\n"
-"			float4 childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;\n"
-"			float4 newPosB = transform(&childPosB,&posB,&ornB);\n"
-"			float4 newOrnB = qtMul(ornB,childOrnB);\n"
-"			posB = newPosB;\n"
-"			ornB = newOrnB;\n"
-"			shapeIndexB = collidables[childColIndexB].m_shapeIndex;\n"
-"		}\n"
-"		//////////////////\n"
-"        \n"
-"		float4 c0local = convexPolyhedronA.m_localCenter;\n"
-"		float4 c0 = transform(&c0local, &posA, &ornA);\n"
-"		float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
-"		float4 c1 = transform(&c1local,&posB,&ornB);\n"
-"		const float4 DeltaC2 = c0 - c1;\n"
-"        \n"
-"        \n"
-"		{\n"
-"			bool sepEE = findSeparatingAxisEdgeEdgeLocalA(	&convexPolyhedronA, &convexShapes[shapeIndexB],\n"
-"                                                              posA,ornA,\n"
-"                                                              posB,ornB,\n"
-"                                                              DeltaC2,\n"
-"                                                              verticesA,uniqueEdgesA,facesA,indicesA,\n"
-"                                                              vertices,uniqueEdges,faces,indices,\n"
-"                                                              &sepAxis,&dmin);\n"
-"                \n"
-"			if (!sepEE)\n"
-"			{\n"
-"				hasSeparatingAxis = 0;\n"
-"			} else\n"
-"			{\n"
-"				hasSeparatingAxis = 1;\n"
-"			}\n"
-"		}\n"
-"		\n"
-"		\n"
-"		if (hasSeparatingAxis)\n"
-"		{\n"
-"			sepAxis.w = dmin;\n"
-"            dmins[i] = dmin;\n"
-"			concaveSeparatingNormalsOut[pairIdx]=sepAxis;\n"
-"			concaveHasSeparatingNormals[i]=1;\n"
-"           \n"
-" 	float minDist = -1e30f;\n"
-"			float maxDist = 0.02f;\n"
-"            \n"
-"            findClippingFaces(sepAxis,\n"
-"                              &convexPolyhedronA,\n"
-"                              &convexShapes[shapeIndexB],\n"
-"                              posA,ornA,\n"
-"                              posB,ornB,\n"
-"                              worldVertsA1GPU,\n"
-"                              worldNormalsAGPU,\n"
-"                              worldVertsB1GPU,\n"
-"                              vertexFaceCapacity,\n"
-"                              minDist, maxDist,\n"
-"                              verticesA,\n"
-"                              facesA,\n"
-"                              indicesA,\n"
-"                              vertices,\n"
-"                              faces,\n"
-"                              indices,\n"
-"                              clippingFacesOut, pairIdx);\n"
-"	           \n"
-"            \n"
-"		} else\n"
-"		{	\n"
-"			//mark this pair as in-active\n"
-"			concavePairs[pairIdx].w = -1;\n"
-"		}\n"
-"	}\n"
-"	else\n"
-"	{	\n"
-"		//mark this pair as in-active\n"
-"		concavePairs[pairIdx].w = -1;\n"
-"	}\n"
-"	\n"
-"	concavePairs[i].z = -1;//for the next stage, z is used to determine existing contact points\n"
-"}\n"
-;
+static const char* satConcaveKernelsCL =
+	"//keep this enum in sync with the CPU version (in btCollidable.h)\n"
+	"//written by Erwin Coumans\n"
+	"#define SHAPE_CONVEX_HULL 3\n"
+	"#define SHAPE_CONCAVE_TRIMESH 5\n"
+	"#define TRIANGLE_NUM_CONVEX_FACES 5\n"
+	"#define SHAPE_COMPOUND_OF_CONVEX_HULLS 6\n"
+	"#define B3_MAX_STACK_DEPTH 256\n"
+	"typedef unsigned int u32;\n"
+	"///keep this in sync with btCollidable.h\n"
+	"typedef struct\n"
+	"{\n"
+	"	union {\n"
+	"		int m_numChildShapes;\n"
+	"		int m_bvhIndex;\n"
+	"	};\n"
+	"	union\n"
+	"	{\n"
+	"		float m_radius;\n"
+	"		int	m_compoundBvhIndex;\n"
+	"	};\n"
+	"	\n"
+	"	int m_shapeType;\n"
+	"	int m_shapeIndex;\n"
+	"	\n"
+	"} btCollidableGpu;\n"
+	"#define MAX_NUM_PARTS_IN_BITS 10\n"
+	"///b3QuantizedBvhNode is a compressed aabb node, 16 bytes.\n"
+	"///Node can be used for leafnode or internal node. Leafnodes can point to 32-bit triangle index (non-negative range).\n"
+	"typedef struct\n"
+	"{\n"
+	"	//12 bytes\n"
+	"	unsigned short int	m_quantizedAabbMin[3];\n"
+	"	unsigned short int	m_quantizedAabbMax[3];\n"
+	"	//4 bytes\n"
+	"	int	m_escapeIndexOrTriangleIndex;\n"
+	"} b3QuantizedBvhNode;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4		m_aabbMin;\n"
+	"	float4		m_aabbMax;\n"
+	"	float4		m_quantization;\n"
+	"	int			m_numNodes;\n"
+	"	int			m_numSubTrees;\n"
+	"	int			m_nodeOffset;\n"
+	"	int			m_subTreeOffset;\n"
+	"} b3BvhInfo;\n"
+	"int	getTriangleIndex(const b3QuantizedBvhNode* rootNode)\n"
+	"{\n"
+	"	unsigned int x=0;\n"
+	"	unsigned int y = (~(x&0))<<(31-MAX_NUM_PARTS_IN_BITS);\n"
+	"	// Get only the lower bits where the triangle index is stored\n"
+	"	return (rootNode->m_escapeIndexOrTriangleIndex&~(y));\n"
+	"}\n"
+	"int	getTriangleIndexGlobal(__global const b3QuantizedBvhNode* rootNode)\n"
+	"{\n"
+	"	unsigned int x=0;\n"
+	"	unsigned int y = (~(x&0))<<(31-MAX_NUM_PARTS_IN_BITS);\n"
+	"	// Get only the lower bits where the triangle index is stored\n"
+	"	return (rootNode->m_escapeIndexOrTriangleIndex&~(y));\n"
+	"}\n"
+	"int isLeafNode(const b3QuantizedBvhNode* rootNode)\n"
+	"{\n"
+	"	//skipindex is negative (internal node), triangleindex >=0 (leafnode)\n"
+	"	return (rootNode->m_escapeIndexOrTriangleIndex >= 0)? 1 : 0;\n"
+	"}\n"
+	"int isLeafNodeGlobal(__global const b3QuantizedBvhNode* rootNode)\n"
+	"{\n"
+	"	//skipindex is negative (internal node), triangleindex >=0 (leafnode)\n"
+	"	return (rootNode->m_escapeIndexOrTriangleIndex >= 0)? 1 : 0;\n"
+	"}\n"
+	"	\n"
+	"int getEscapeIndex(const b3QuantizedBvhNode* rootNode)\n"
+	"{\n"
+	"	return -rootNode->m_escapeIndexOrTriangleIndex;\n"
+	"}\n"
+	"int getEscapeIndexGlobal(__global const b3QuantizedBvhNode* rootNode)\n"
+	"{\n"
+	"	return -rootNode->m_escapeIndexOrTriangleIndex;\n"
+	"}\n"
+	"typedef struct\n"
+	"{\n"
+	"	//12 bytes\n"
+	"	unsigned short int	m_quantizedAabbMin[3];\n"
+	"	unsigned short int	m_quantizedAabbMax[3];\n"
+	"	//4 bytes, points to the root of the subtree\n"
+	"	int			m_rootNodeIndex;\n"
+	"	//4 bytes\n"
+	"	int			m_subtreeSize;\n"
+	"	int			m_padding[3];\n"
+	"} b3BvhSubtreeInfo;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4	m_childPosition;\n"
+	"	float4	m_childOrientation;\n"
+	"	int m_shapeIndex;\n"
+	"	int m_unused0;\n"
+	"	int m_unused1;\n"
+	"	int m_unused2;\n"
+	"} btGpuChildShape;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_pos;\n"
+	"	float4 m_quat;\n"
+	"	float4 m_linVel;\n"
+	"	float4 m_angVel;\n"
+	"	u32 m_collidableIdx;\n"
+	"	float m_invMass;\n"
+	"	float m_restituitionCoeff;\n"
+	"	float m_frictionCoeff;\n"
+	"} BodyData;\n"
+	"typedef struct  \n"
+	"{\n"
+	"	float4		m_localCenter;\n"
+	"	float4		m_extents;\n"
+	"	float4		mC;\n"
+	"	float4		mE;\n"
+	"	\n"
+	"	float			m_radius;\n"
+	"	int	m_faceOffset;\n"
+	"	int m_numFaces;\n"
+	"	int	m_numVertices;\n"
+	"	int m_vertexOffset;\n"
+	"	int	m_uniqueEdgesOffset;\n"
+	"	int	m_numUniqueEdges;\n"
+	"	int m_unused;\n"
+	"} ConvexPolyhedronCL;\n"
+	"typedef struct \n"
+	"{\n"
+	"	union\n"
+	"	{\n"
+	"		float4	m_min;\n"
+	"		float   m_minElems[4];\n"
+	"		int			m_minIndices[4];\n"
+	"	};\n"
+	"	union\n"
+	"	{\n"
+	"		float4	m_max;\n"
+	"		float   m_maxElems[4];\n"
+	"		int			m_maxIndices[4];\n"
+	"	};\n"
+	"} btAabbCL;\n"
+	"#ifndef B3_AABB_H\n"
+	"#define B3_AABB_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#define B3_FLOAT4_H\n"
+	"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
+	"#define B3_PLATFORM_DEFINITIONS_H\n"
+	"struct MyTest\n"
+	"{\n"
+	"	int bla;\n"
+	"};\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
+	"#define B3_LARGE_FLOAT 1e18f\n"
+	"#define B3_INFINITY 1e18f\n"
+	"#define b3Assert(a)\n"
+	"#define b3ConstArray(a) __global const a*\n"
+	"#define b3AtomicInc atomic_inc\n"
+	"#define b3AtomicAdd atomic_add\n"
+	"#define b3Fabs fabs\n"
+	"#define b3Sqrt native_sqrt\n"
+	"#define b3Sin native_sin\n"
+	"#define b3Cos native_cos\n"
+	"#define B3_STATIC\n"
+	"#endif\n"
+	"#endif\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"	typedef float4	b3Float4;\n"
+	"	#define b3Float4ConstArg const b3Float4\n"
+	"	#define b3MakeFloat4 (float4)\n"
+	"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return dot(a1, b1);\n"
+	"	}\n"
+	"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return cross(a1, b1);\n"
+	"	}\n"
+	"	#define b3MinFloat4 min\n"
+	"	#define b3MaxFloat4 max\n"
+	"	#define b3Normalized(a) normalize(a)\n"
+	"#endif \n"
+	"		\n"
+	"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
+	"{\n"
+	"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
+	"		return false;\n"
+	"	return true;\n"
+	"}\n"
+	"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
+	"{\n"
+	"    float maxDot = -B3_INFINITY;\n"
+	"    int i = 0;\n"
+	"    int ptIndex = -1;\n"
+	"    for( i = 0; i < vecLen; i++ )\n"
+	"    {\n"
+	"        float dot = b3Dot3F4(vecArray[i],vec);\n"
+	"            \n"
+	"        if( dot > maxDot )\n"
+	"        {\n"
+	"            maxDot = dot;\n"
+	"            ptIndex = i;\n"
+	"        }\n"
+	"    }\n"
+	"	b3Assert(ptIndex>=0);\n"
+	"    if (ptIndex<0)\n"
+	"	{\n"
+	"		ptIndex = 0;\n"
+	"	}\n"
+	"    *dotOut = maxDot;\n"
+	"    return ptIndex;\n"
+	"}\n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifndef B3_MAT3x3_H\n"
+	"#define B3_MAT3x3_H\n"
+	"#ifndef B3_QUAT_H\n"
+	"#define B3_QUAT_H\n"
+	"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif\n"
+	"#endif\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"	typedef float4	b3Quat;\n"
+	"	#define b3QuatConstArg const b3Quat\n"
+	"	\n"
+	"	\n"
+	"inline float4 b3FastNormalize4(float4 v)\n"
+	"{\n"
+	"	v = (float4)(v.xyz,0.f);\n"
+	"	return fast_normalize(v);\n"
+	"}\n"
+	"	\n"
+	"inline b3Quat b3QuatMul(b3Quat a, b3Quat b);\n"
+	"inline b3Quat b3QuatNormalized(b3QuatConstArg in);\n"
+	"inline b3Quat b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec);\n"
+	"inline b3Quat b3QuatInvert(b3QuatConstArg q);\n"
+	"inline b3Quat b3QuatInverse(b3QuatConstArg q);\n"
+	"inline b3Quat b3QuatMul(b3QuatConstArg a, b3QuatConstArg b)\n"
+	"{\n"
+	"	b3Quat ans;\n"
+	"	ans = b3Cross3( a, b );\n"
+	"	ans += a.w*b+b.w*a;\n"
+	"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
+	"	ans.w = a.w*b.w - b3Dot3F4(a, b);\n"
+	"	return ans;\n"
+	"}\n"
+	"inline b3Quat b3QuatNormalized(b3QuatConstArg in)\n"
+	"{\n"
+	"	b3Quat q;\n"
+	"	q=in;\n"
+	"	//return b3FastNormalize4(in);\n"
+	"	float len = native_sqrt(dot(q, q));\n"
+	"	if(len > 0.f)\n"
+	"	{\n"
+	"		q *= 1.f / len;\n"
+	"	}\n"
+	"	else\n"
+	"	{\n"
+	"		q.x = q.y = q.z = 0.f;\n"
+	"		q.w = 1.f;\n"
+	"	}\n"
+	"	return q;\n"
+	"}\n"
+	"inline float4 b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
+	"{\n"
+	"	b3Quat qInv = b3QuatInvert( q );\n"
+	"	float4 vcpy = vec;\n"
+	"	vcpy.w = 0.f;\n"
+	"	float4 out = b3QuatMul(b3QuatMul(q,vcpy),qInv);\n"
+	"	return out;\n"
+	"}\n"
+	"inline b3Quat b3QuatInverse(b3QuatConstArg q)\n"
+	"{\n"
+	"	return (b3Quat)(-q.xyz, q.w);\n"
+	"}\n"
+	"inline b3Quat b3QuatInvert(b3QuatConstArg q)\n"
+	"{\n"
+	"	return (b3Quat)(-q.xyz, q.w);\n"
+	"}\n"
+	"inline float4 b3QuatInvRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
+	"{\n"
+	"	return b3QuatRotate( b3QuatInvert( q ), vec );\n"
+	"}\n"
+	"inline b3Float4 b3TransformPoint(b3Float4ConstArg point, b3Float4ConstArg translation, b3QuatConstArg  orientation)\n"
+	"{\n"
+	"	return b3QuatRotate( orientation, point ) + (translation);\n"
+	"}\n"
+	"	\n"
+	"#endif \n"
+	"#endif //B3_QUAT_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"typedef struct\n"
+	"{\n"
+	"	b3Float4 m_row[3];\n"
+	"}b3Mat3x3;\n"
+	"#define b3Mat3x3ConstArg const b3Mat3x3\n"
+	"#define b3GetRow(m,row) (m.m_row[row])\n"
+	"inline b3Mat3x3 b3QuatGetRotationMatrix(b3Quat quat)\n"
+	"{\n"
+	"	b3Float4 quat2 = (b3Float4)(quat.x*quat.x, quat.y*quat.y, quat.z*quat.z, 0.f);\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0].x=1-2*quat2.y-2*quat2.z;\n"
+	"	out.m_row[0].y=2*quat.x*quat.y-2*quat.w*quat.z;\n"
+	"	out.m_row[0].z=2*quat.x*quat.z+2*quat.w*quat.y;\n"
+	"	out.m_row[0].w = 0.f;\n"
+	"	out.m_row[1].x=2*quat.x*quat.y+2*quat.w*quat.z;\n"
+	"	out.m_row[1].y=1-2*quat2.x-2*quat2.z;\n"
+	"	out.m_row[1].z=2*quat.y*quat.z-2*quat.w*quat.x;\n"
+	"	out.m_row[1].w = 0.f;\n"
+	"	out.m_row[2].x=2*quat.x*quat.z-2*quat.w*quat.y;\n"
+	"	out.m_row[2].y=2*quat.y*quat.z+2*quat.w*quat.x;\n"
+	"	out.m_row[2].z=1-2*quat2.x-2*quat2.y;\n"
+	"	out.m_row[2].w = 0.f;\n"
+	"	return out;\n"
+	"}\n"
+	"inline b3Mat3x3 b3AbsoluteMat3x3(b3Mat3x3ConstArg matIn)\n"
+	"{\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0] = fabs(matIn.m_row[0]);\n"
+	"	out.m_row[1] = fabs(matIn.m_row[1]);\n"
+	"	out.m_row[2] = fabs(matIn.m_row[2]);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtZero();\n"
+	"__inline\n"
+	"b3Mat3x3 mtIdentity();\n"
+	"__inline\n"
+	"b3Mat3x3 mtTranspose(b3Mat3x3 m);\n"
+	"__inline\n"
+	"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b);\n"
+	"__inline\n"
+	"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b);\n"
+	"__inline\n"
+	"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b);\n"
+	"__inline\n"
+	"b3Mat3x3 mtZero()\n"
+	"{\n"
+	"	b3Mat3x3 m;\n"
+	"	m.m_row[0] = (b3Float4)(0.f);\n"
+	"	m.m_row[1] = (b3Float4)(0.f);\n"
+	"	m.m_row[2] = (b3Float4)(0.f);\n"
+	"	return m;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtIdentity()\n"
+	"{\n"
+	"	b3Mat3x3 m;\n"
+	"	m.m_row[0] = (b3Float4)(1,0,0,0);\n"
+	"	m.m_row[1] = (b3Float4)(0,1,0,0);\n"
+	"	m.m_row[2] = (b3Float4)(0,0,1,0);\n"
+	"	return m;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtTranspose(b3Mat3x3 m)\n"
+	"{\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0] = (b3Float4)(m.m_row[0].x, m.m_row[1].x, m.m_row[2].x, 0.f);\n"
+	"	out.m_row[1] = (b3Float4)(m.m_row[0].y, m.m_row[1].y, m.m_row[2].y, 0.f);\n"
+	"	out.m_row[2] = (b3Float4)(m.m_row[0].z, m.m_row[1].z, m.m_row[2].z, 0.f);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b)\n"
+	"{\n"
+	"	b3Mat3x3 transB;\n"
+	"	transB = mtTranspose( b );\n"
+	"	b3Mat3x3 ans;\n"
+	"	//	why this doesn't run when 0ing in the for{}\n"
+	"	a.m_row[0].w = 0.f;\n"
+	"	a.m_row[1].w = 0.f;\n"
+	"	a.m_row[2].w = 0.f;\n"
+	"	for(int i=0; i<3; i++)\n"
+	"	{\n"
+	"//	a.m_row[i].w = 0.f;\n"
+	"		ans.m_row[i].x = b3Dot3F4(a.m_row[i],transB.m_row[0]);\n"
+	"		ans.m_row[i].y = b3Dot3F4(a.m_row[i],transB.m_row[1]);\n"
+	"		ans.m_row[i].z = b3Dot3F4(a.m_row[i],transB.m_row[2]);\n"
+	"		ans.m_row[i].w = 0.f;\n"
+	"	}\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b)\n"
+	"{\n"
+	"	b3Float4 ans;\n"
+	"	ans.x = b3Dot3F4( a.m_row[0], b );\n"
+	"	ans.y = b3Dot3F4( a.m_row[1], b );\n"
+	"	ans.z = b3Dot3F4( a.m_row[2], b );\n"
+	"	ans.w = 0.f;\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b)\n"
+	"{\n"
+	"	b3Float4 colx = b3MakeFloat4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
+	"	b3Float4 coly = b3MakeFloat4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
+	"	b3Float4 colz = b3MakeFloat4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
+	"	b3Float4 ans;\n"
+	"	ans.x = b3Dot3F4( a, colx );\n"
+	"	ans.y = b3Dot3F4( a, coly );\n"
+	"	ans.z = b3Dot3F4( a, colz );\n"
+	"	return ans;\n"
+	"}\n"
+	"#endif\n"
+	"#endif //B3_MAT3x3_H\n"
+	"typedef struct b3Aabb b3Aabb_t;\n"
+	"struct b3Aabb\n"
+	"{\n"
+	"	union\n"
+	"	{\n"
+	"		float m_min[4];\n"
+	"		b3Float4 m_minVec;\n"
+	"		int m_minIndices[4];\n"
+	"	};\n"
+	"	union\n"
+	"	{\n"
+	"		float	m_max[4];\n"
+	"		b3Float4 m_maxVec;\n"
+	"		int m_signedMaxIndices[4];\n"
+	"	};\n"
+	"};\n"
+	"inline void b3TransformAabb2(b3Float4ConstArg localAabbMin,b3Float4ConstArg localAabbMax, float margin,\n"
+	"						b3Float4ConstArg pos,\n"
+	"						b3QuatConstArg orn,\n"
+	"						b3Float4* aabbMinOut,b3Float4* aabbMaxOut)\n"
+	"{\n"
+	"		b3Float4 localHalfExtents = 0.5f*(localAabbMax-localAabbMin);\n"
+	"		localHalfExtents+=b3MakeFloat4(margin,margin,margin,0.f);\n"
+	"		b3Float4 localCenter = 0.5f*(localAabbMax+localAabbMin);\n"
+	"		b3Mat3x3 m;\n"
+	"		m = b3QuatGetRotationMatrix(orn);\n"
+	"		b3Mat3x3 abs_b = b3AbsoluteMat3x3(m);\n"
+	"		b3Float4 center = b3TransformPoint(localCenter,pos,orn);\n"
+	"		\n"
+	"		b3Float4 extent = b3MakeFloat4(b3Dot3F4(localHalfExtents,b3GetRow(abs_b,0)),\n"
+	"										 b3Dot3F4(localHalfExtents,b3GetRow(abs_b,1)),\n"
+	"										 b3Dot3F4(localHalfExtents,b3GetRow(abs_b,2)),\n"
+	"										 0.f);\n"
+	"		*aabbMinOut = center-extent;\n"
+	"		*aabbMaxOut = center+extent;\n"
+	"}\n"
+	"/// conservative test for overlap between two aabbs\n"
+	"inline bool b3TestAabbAgainstAabb(b3Float4ConstArg aabbMin1,b3Float4ConstArg aabbMax1,\n"
+	"								b3Float4ConstArg aabbMin2, b3Float4ConstArg aabbMax2)\n"
+	"{\n"
+	"	bool overlap = true;\n"
+	"	overlap = (aabbMin1.x > aabbMax2.x || aabbMax1.x < aabbMin2.x) ? false : overlap;\n"
+	"	overlap = (aabbMin1.z > aabbMax2.z || aabbMax1.z < aabbMin2.z) ? false : overlap;\n"
+	"	overlap = (aabbMin1.y > aabbMax2.y || aabbMax1.y < aabbMin2.y) ? false : overlap;\n"
+	"	return overlap;\n"
+	"}\n"
+	"#endif //B3_AABB_H\n"
+	"/*\n"
+	"Bullet Continuous Collision Detection and Physics Library\n"
+	"Copyright (c) 2003-2013 Erwin Coumans  http://bulletphysics.org\n"
+	"This software is provided 'as-is', without any express or implied warranty.\n"
+	"In no event will the authors be held liable for any damages arising from the use of this software.\n"
+	"Permission is granted to anyone to use this software for any purpose,\n"
+	"including commercial applications, and to alter it and redistribute it freely,\n"
+	"subject to the following restrictions:\n"
+	"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
+	"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
+	"3. This notice may not be removed or altered from any source distribution.\n"
+	"*/\n"
+	"#ifndef B3_INT2_H\n"
+	"#define B3_INT2_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#define b3UnsignedInt2 uint2\n"
+	"#define b3Int2 int2\n"
+	"#define b3MakeInt2 (int2)\n"
+	"#endif //__cplusplus\n"
+	"#endif\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_plane;\n"
+	"	int m_indexOffset;\n"
+	"	int m_numIndices;\n"
+	"} btGpuFace;\n"
+	"#define make_float4 (float4)\n"
+	"__inline\n"
+	"float4 cross3(float4 a, float4 b)\n"
+	"{\n"
+	"	return cross(a,b);\n"
+	"	\n"
+	"//	float4 a1 = make_float4(a.xyz,0.f);\n"
+	"//	float4 b1 = make_float4(b.xyz,0.f);\n"
+	"//	return cross(a1,b1);\n"
+	"//float4 c = make_float4(a.y*b.z - a.z*b.y,a.z*b.x - a.x*b.z,a.x*b.y - a.y*b.x,0.f);\n"
+	"	\n"
+	"	//	float4 c = make_float4(a.y*b.z - a.z*b.y,1.f,a.x*b.y - a.y*b.x,0.f);\n"
+	"	\n"
+	"	//return c;\n"
+	"}\n"
+	"__inline\n"
+	"float dot3F4(float4 a, float4 b)\n"
+	"{\n"
+	"	float4 a1 = make_float4(a.xyz,0.f);\n"
+	"	float4 b1 = make_float4(b.xyz,0.f);\n"
+	"	return dot(a1, b1);\n"
+	"}\n"
+	"__inline\n"
+	"float4 fastNormalize4(float4 v)\n"
+	"{\n"
+	"	v = make_float4(v.xyz,0.f);\n"
+	"	return fast_normalize(v);\n"
+	"}\n"
+	"///////////////////////////////////////\n"
+	"//	Quaternion\n"
+	"///////////////////////////////////////\n"
+	"typedef float4 Quaternion;\n"
+	"__inline\n"
+	"Quaternion qtMul(Quaternion a, Quaternion b);\n"
+	"__inline\n"
+	"Quaternion qtNormalize(Quaternion in);\n"
+	"__inline\n"
+	"float4 qtRotate(Quaternion q, float4 vec);\n"
+	"__inline\n"
+	"Quaternion qtInvert(Quaternion q);\n"
+	"__inline\n"
+	"Quaternion qtMul(Quaternion a, Quaternion b)\n"
+	"{\n"
+	"	Quaternion ans;\n"
+	"	ans = cross3( a, b );\n"
+	"	ans += a.w*b+b.w*a;\n"
+	"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
+	"	ans.w = a.w*b.w - dot3F4(a, b);\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"Quaternion qtNormalize(Quaternion in)\n"
+	"{\n"
+	"	return fastNormalize4(in);\n"
+	"//	in /= length( in );\n"
+	"//	return in;\n"
+	"}\n"
+	"__inline\n"
+	"float4 qtRotate(Quaternion q, float4 vec)\n"
+	"{\n"
+	"	Quaternion qInv = qtInvert( q );\n"
+	"	float4 vcpy = vec;\n"
+	"	vcpy.w = 0.f;\n"
+	"	float4 out = qtMul(qtMul(q,vcpy),qInv);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"Quaternion qtInvert(Quaternion q)\n"
+	"{\n"
+	"	return (Quaternion)(-q.xyz, q.w);\n"
+	"}\n"
+	"__inline\n"
+	"float4 qtInvRotate(const Quaternion q, float4 vec)\n"
+	"{\n"
+	"	return qtRotate( qtInvert( q ), vec );\n"
+	"}\n"
+	"__inline\n"
+	"float4 transform(const float4* p, const float4* translation, const Quaternion* orientation)\n"
+	"{\n"
+	"	return qtRotate( *orientation, *p ) + (*translation);\n"
+	"}\n"
+	"__inline\n"
+	"float4 normalize3(const float4 a)\n"
+	"{\n"
+	"	float4 n = make_float4(a.x, a.y, a.z, 0.f);\n"
+	"	return fastNormalize4( n );\n"
+	"}\n"
+	"inline void projectLocal(const ConvexPolyhedronCL* hull,  const float4 pos, const float4 orn, \n"
+	"const float4* dir, const float4* vertices, float* min, float* max)\n"
+	"{\n"
+	"	min[0] = FLT_MAX;\n"
+	"	max[0] = -FLT_MAX;\n"
+	"	int numVerts = hull->m_numVertices;\n"
+	"	const float4 localDir = qtInvRotate(orn,*dir);\n"
+	"	float offset = dot(pos,*dir);\n"
+	"	for(int i=0;i<numVerts;i++)\n"
+	"	{\n"
+	"		float dp = dot(vertices[hull->m_vertexOffset+i],localDir);\n"
+	"		if(dp < min[0])	\n"
+	"			min[0] = dp;\n"
+	"		if(dp > max[0])	\n"
+	"			max[0] = dp;\n"
+	"	}\n"
+	"	if(min[0]>max[0])\n"
+	"	{\n"
+	"		float tmp = min[0];\n"
+	"		min[0] = max[0];\n"
+	"		max[0] = tmp;\n"
+	"	}\n"
+	"	min[0] += offset;\n"
+	"	max[0] += offset;\n"
+	"}\n"
+	"inline void project(__global const ConvexPolyhedronCL* hull,  const float4 pos, const float4 orn, \n"
+	"const float4* dir, __global const float4* vertices, float* min, float* max)\n"
+	"{\n"
+	"	min[0] = FLT_MAX;\n"
+	"	max[0] = -FLT_MAX;\n"
+	"	int numVerts = hull->m_numVertices;\n"
+	"	const float4 localDir = qtInvRotate(orn,*dir);\n"
+	"	float offset = dot(pos,*dir);\n"
+	"	for(int i=0;i<numVerts;i++)\n"
+	"	{\n"
+	"		float dp = dot(vertices[hull->m_vertexOffset+i],localDir);\n"
+	"		if(dp < min[0])	\n"
+	"			min[0] = dp;\n"
+	"		if(dp > max[0])	\n"
+	"			max[0] = dp;\n"
+	"	}\n"
+	"	if(min[0]>max[0])\n"
+	"	{\n"
+	"		float tmp = min[0];\n"
+	"		min[0] = max[0];\n"
+	"		max[0] = tmp;\n"
+	"	}\n"
+	"	min[0] += offset;\n"
+	"	max[0] += offset;\n"
+	"}\n"
+	"inline bool TestSepAxisLocalA(const ConvexPolyhedronCL* hullA, __global const ConvexPolyhedronCL* hullB, \n"
+	"	const float4 posA,const float4 ornA,\n"
+	"	const float4 posB,const float4 ornB,\n"
+	"	float4* sep_axis, const float4* verticesA, __global const float4* verticesB,float* depth)\n"
+	"{\n"
+	"	float Min0,Max0;\n"
+	"	float Min1,Max1;\n"
+	"	projectLocal(hullA,posA,ornA,sep_axis,verticesA, &Min0, &Max0);\n"
+	"	project(hullB,posB,ornB, sep_axis,verticesB, &Min1, &Max1);\n"
+	"	if(Max0<Min1 || Max1<Min0)\n"
+	"		return false;\n"
+	"	float d0 = Max0 - Min1;\n"
+	"	float d1 = Max1 - Min0;\n"
+	"	*depth = d0<d1 ? d0:d1;\n"
+	"	return true;\n"
+	"}\n"
+	"inline bool IsAlmostZero(const float4 v)\n"
+	"{\n"
+	"	if(fabs(v.x)>1e-6f || fabs(v.y)>1e-6f || fabs(v.z)>1e-6f)\n"
+	"		return false;\n"
+	"	return true;\n"
+	"}\n"
+	"bool findSeparatingAxisLocalA(	const ConvexPolyhedronCL* hullA, __global const ConvexPolyhedronCL* hullB, \n"
+	"	const float4 posA1,\n"
+	"	const float4 ornA,\n"
+	"	const float4 posB1,\n"
+	"	const float4 ornB,\n"
+	"	const float4 DeltaC2,\n"
+	"	\n"
+	"	const float4* verticesA, \n"
+	"	const float4* uniqueEdgesA, \n"
+	"	const btGpuFace* facesA,\n"
+	"	const int*  indicesA,\n"
+	"	__global const float4* verticesB, \n"
+	"	__global const float4* uniqueEdgesB, \n"
+	"	__global const btGpuFace* facesB,\n"
+	"	__global const int*  indicesB,\n"
+	"	float4* sep,\n"
+	"	float* dmin)\n"
+	"{\n"
+	"	\n"
+	"	float4 posA = posA1;\n"
+	"	posA.w = 0.f;\n"
+	"	float4 posB = posB1;\n"
+	"	posB.w = 0.f;\n"
+	"	int curPlaneTests=0;\n"
+	"	{\n"
+	"		int numFacesA = hullA->m_numFaces;\n"
+	"		// Test normals from hullA\n"
+	"		for(int i=0;i<numFacesA;i++)\n"
+	"		{\n"
+	"			const float4 normal = facesA[hullA->m_faceOffset+i].m_plane;\n"
+	"			float4 faceANormalWS = qtRotate(ornA,normal);\n"
+	"			if (dot3F4(DeltaC2,faceANormalWS)<0)\n"
+	"				faceANormalWS*=-1.f;\n"
+	"			curPlaneTests++;\n"
+	"			float d;\n"
+	"			if(!TestSepAxisLocalA( hullA, hullB, posA,ornA,posB,ornB,&faceANormalWS, verticesA, verticesB,&d))\n"
+	"				return false;\n"
+	"			if(d<*dmin)\n"
+	"			{\n"
+	"				*dmin = d;\n"
+	"				*sep = faceANormalWS;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	if((dot3F4(-DeltaC2,*sep))>0.0f)\n"
+	"	{\n"
+	"		*sep = -(*sep);\n"
+	"	}\n"
+	"	return true;\n"
+	"}\n"
+	"bool findSeparatingAxisLocalB(	__global const ConvexPolyhedronCL* hullA,  const ConvexPolyhedronCL* hullB, \n"
+	"	const float4 posA1,\n"
+	"	const float4 ornA,\n"
+	"	const float4 posB1,\n"
+	"	const float4 ornB,\n"
+	"	const float4 DeltaC2,\n"
+	"	__global const float4* verticesA, \n"
+	"	__global const float4* uniqueEdgesA, \n"
+	"	__global const btGpuFace* facesA,\n"
+	"	__global const int*  indicesA,\n"
+	"	const float4* verticesB,\n"
+	"	const float4* uniqueEdgesB, \n"
+	"	const btGpuFace* facesB,\n"
+	"	const int*  indicesB,\n"
+	"	float4* sep,\n"
+	"	float* dmin)\n"
+	"{\n"
+	"	float4 posA = posA1;\n"
+	"	posA.w = 0.f;\n"
+	"	float4 posB = posB1;\n"
+	"	posB.w = 0.f;\n"
+	"	int curPlaneTests=0;\n"
+	"	{\n"
+	"		int numFacesA = hullA->m_numFaces;\n"
+	"		// Test normals from hullA\n"
+	"		for(int i=0;i<numFacesA;i++)\n"
+	"		{\n"
+	"			const float4 normal = facesA[hullA->m_faceOffset+i].m_plane;\n"
+	"			float4 faceANormalWS = qtRotate(ornA,normal);\n"
+	"			if (dot3F4(DeltaC2,faceANormalWS)<0)\n"
+	"				faceANormalWS *= -1.f;\n"
+	"			curPlaneTests++;\n"
+	"			float d;\n"
+	"			if(!TestSepAxisLocalA( hullB, hullA, posB,ornB,posA,ornA, &faceANormalWS, verticesB,verticesA, &d))\n"
+	"				return false;\n"
+	"			if(d<*dmin)\n"
+	"			{\n"
+	"				*dmin = d;\n"
+	"				*sep = faceANormalWS;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	if((dot3F4(-DeltaC2,*sep))>0.0f)\n"
+	"	{\n"
+	"		*sep = -(*sep);\n"
+	"	}\n"
+	"	return true;\n"
+	"}\n"
+	"bool findSeparatingAxisEdgeEdgeLocalA(	const ConvexPolyhedronCL* hullA, __global const ConvexPolyhedronCL* hullB, \n"
+	"	const float4 posA1,\n"
+	"	const float4 ornA,\n"
+	"	const float4 posB1,\n"
+	"	const float4 ornB,\n"
+	"	const float4 DeltaC2,\n"
+	"	const float4* verticesA, \n"
+	"	const float4* uniqueEdgesA, \n"
+	"	const btGpuFace* facesA,\n"
+	"	const int*  indicesA,\n"
+	"	__global const float4* verticesB, \n"
+	"	__global const float4* uniqueEdgesB, \n"
+	"	__global const btGpuFace* facesB,\n"
+	"	__global const int*  indicesB,\n"
+	"		float4* sep,\n"
+	"	float* dmin)\n"
+	"{\n"
+	"	float4 posA = posA1;\n"
+	"	posA.w = 0.f;\n"
+	"	float4 posB = posB1;\n"
+	"	posB.w = 0.f;\n"
+	"	int curPlaneTests=0;\n"
+	"	int curEdgeEdge = 0;\n"
+	"	// Test edges\n"
+	"	for(int e0=0;e0<hullA->m_numUniqueEdges;e0++)\n"
+	"	{\n"
+	"		const float4 edge0 = uniqueEdgesA[hullA->m_uniqueEdgesOffset+e0];\n"
+	"		float4 edge0World = qtRotate(ornA,edge0);\n"
+	"		for(int e1=0;e1<hullB->m_numUniqueEdges;e1++)\n"
+	"		{\n"
+	"			const float4 edge1 = uniqueEdgesB[hullB->m_uniqueEdgesOffset+e1];\n"
+	"			float4 edge1World = qtRotate(ornB,edge1);\n"
+	"			float4 crossje = cross3(edge0World,edge1World);\n"
+	"			curEdgeEdge++;\n"
+	"			if(!IsAlmostZero(crossje))\n"
+	"			{\n"
+	"				crossje = normalize3(crossje);\n"
+	"				if (dot3F4(DeltaC2,crossje)<0)\n"
+	"					crossje *= -1.f;\n"
+	"				float dist;\n"
+	"				bool result = true;\n"
+	"				{\n"
+	"					float Min0,Max0;\n"
+	"					float Min1,Max1;\n"
+	"					projectLocal(hullA,posA,ornA,&crossje,verticesA, &Min0, &Max0);\n"
+	"					project(hullB,posB,ornB,&crossje,verticesB, &Min1, &Max1);\n"
+	"				\n"
+	"					if(Max0<Min1 || Max1<Min0)\n"
+	"						result = false;\n"
+	"				\n"
+	"					float d0 = Max0 - Min1;\n"
+	"					float d1 = Max1 - Min0;\n"
+	"					dist = d0<d1 ? d0:d1;\n"
+	"					result = true;\n"
+	"				}\n"
+	"				\n"
+	"				if(dist<*dmin)\n"
+	"				{\n"
+	"					*dmin = dist;\n"
+	"					*sep = crossje;\n"
+	"				}\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	\n"
+	"	if((dot3F4(-DeltaC2,*sep))>0.0f)\n"
+	"	{\n"
+	"		*sep = -(*sep);\n"
+	"	}\n"
+	"	return true;\n"
+	"}\n"
+	"inline int	findClippingFaces(const float4 separatingNormal,\n"
+	"                      const ConvexPolyhedronCL* hullA, \n"
+	"					  __global const ConvexPolyhedronCL* hullB,\n"
+	"                      const float4 posA, const Quaternion ornA,const float4 posB, const Quaternion ornB,\n"
+	"                       __global float4* worldVertsA1,\n"
+	"                      __global float4* worldNormalsA1,\n"
+	"                      __global float4* worldVertsB1,\n"
+	"                      int capacityWorldVerts,\n"
+	"                      const float minDist, float maxDist,\n"
+	"					  const float4* verticesA,\n"
+	"                      const btGpuFace* facesA,\n"
+	"                      const int* indicesA,\n"
+	"					  __global const float4* verticesB,\n"
+	"                      __global const btGpuFace* facesB,\n"
+	"                      __global const int* indicesB,\n"
+	"                      __global int4* clippingFaces, int pairIndex)\n"
+	"{\n"
+	"	int numContactsOut = 0;\n"
+	"	int numWorldVertsB1= 0;\n"
+	"    \n"
+	"    \n"
+	"	int closestFaceB=0;\n"
+	"	float dmax = -FLT_MAX;\n"
+	"    \n"
+	"	{\n"
+	"		for(int face=0;face<hullB->m_numFaces;face++)\n"
+	"		{\n"
+	"			const float4 Normal = make_float4(facesB[hullB->m_faceOffset+face].m_plane.x,\n"
+	"                                              facesB[hullB->m_faceOffset+face].m_plane.y, facesB[hullB->m_faceOffset+face].m_plane.z,0.f);\n"
+	"			const float4 WorldNormal = qtRotate(ornB, Normal);\n"
+	"			float d = dot3F4(WorldNormal,separatingNormal);\n"
+	"			if (d > dmax)\n"
+	"			{\n"
+	"				dmax = d;\n"
+	"				closestFaceB = face;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"    \n"
+	"	{\n"
+	"		const btGpuFace polyB = facesB[hullB->m_faceOffset+closestFaceB];\n"
+	"		int numVertices = polyB.m_numIndices;\n"
+	"        if (numVertices>capacityWorldVerts)\n"
+	"            numVertices = capacityWorldVerts;\n"
+	"        if (numVertices<0)\n"
+	"            numVertices = 0;\n"
+	"        \n"
+	"		for(int e0=0;e0<numVertices;e0++)\n"
+	"		{\n"
+	"            if (e0<capacityWorldVerts)\n"
+	"            {\n"
+	"                const float4 b = verticesB[hullB->m_vertexOffset+indicesB[polyB.m_indexOffset+e0]];\n"
+	"                worldVertsB1[pairIndex*capacityWorldVerts+numWorldVertsB1++] = transform(&b,&posB,&ornB);\n"
+	"            }\n"
+	"		}\n"
+	"	}\n"
+	"    \n"
+	"    int closestFaceA=0;\n"
+	"	{\n"
+	"		float dmin = FLT_MAX;\n"
+	"		for(int face=0;face<hullA->m_numFaces;face++)\n"
+	"		{\n"
+	"			const float4 Normal = make_float4(\n"
+	"                                              facesA[hullA->m_faceOffset+face].m_plane.x,\n"
+	"                                              facesA[hullA->m_faceOffset+face].m_plane.y,\n"
+	"                                              facesA[hullA->m_faceOffset+face].m_plane.z,\n"
+	"                                              0.f);\n"
+	"			const float4 faceANormalWS = qtRotate(ornA,Normal);\n"
+	"            \n"
+	"			float d = dot3F4(faceANormalWS,separatingNormal);\n"
+	"			if (d < dmin)\n"
+	"			{\n"
+	"				dmin = d;\n"
+	"				closestFaceA = face;\n"
+	"                worldNormalsA1[pairIndex] = faceANormalWS;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"    \n"
+	"    int numVerticesA = facesA[hullA->m_faceOffset+closestFaceA].m_numIndices;\n"
+	"    if (numVerticesA>capacityWorldVerts)\n"
+	"       numVerticesA = capacityWorldVerts;\n"
+	"    if (numVerticesA<0)\n"
+	"        numVerticesA=0;\n"
+	"    \n"
+	"	for(int e0=0;e0<numVerticesA;e0++)\n"
+	"	{\n"
+	"        if (e0<capacityWorldVerts)\n"
+	"        {\n"
+	"            const float4 a = verticesA[hullA->m_vertexOffset+indicesA[facesA[hullA->m_faceOffset+closestFaceA].m_indexOffset+e0]];\n"
+	"            worldVertsA1[pairIndex*capacityWorldVerts+e0] = transform(&a, &posA,&ornA);\n"
+	"        }\n"
+	"    }\n"
+	"    \n"
+	"    clippingFaces[pairIndex].x = closestFaceA;\n"
+	"    clippingFaces[pairIndex].y = closestFaceB;\n"
+	"    clippingFaces[pairIndex].z = numVerticesA;\n"
+	"    clippingFaces[pairIndex].w = numWorldVertsB1;\n"
+	"    \n"
+	"    \n"
+	"	return numContactsOut;\n"
+	"}\n"
+	"// work-in-progress\n"
+	"__kernel void   findConcaveSeparatingAxisVertexFaceKernel( __global int4* concavePairs,\n"
+	"                                                __global const BodyData* rigidBodies,\n"
+	"                                                __global const btCollidableGpu* collidables,\n"
+	"                                                __global const ConvexPolyhedronCL* convexShapes,\n"
+	"                                                __global const float4* vertices,\n"
+	"                                                __global const float4* uniqueEdges,\n"
+	"                                                __global const btGpuFace* faces,\n"
+	"                                                __global const int* indices,\n"
+	"                                                __global const btGpuChildShape* gpuChildShapes,\n"
+	"                                                __global btAabbCL* aabbs,\n"
+	"                                                __global float4* concaveSeparatingNormalsOut,\n"
+	"                                                __global int* concaveHasSeparatingNormals,\n"
+	"                                                __global int4* clippingFacesOut,\n"
+	"                                                __global float4* worldVertsA1GPU,\n"
+	"                                                __global float4*  worldNormalsAGPU,\n"
+	"                                                __global float4* worldVertsB1GPU,\n"
+	"                                                __global float* dmins,\n"
+	"                                                int vertexFaceCapacity,\n"
+	"                                                int numConcavePairs\n"
+	"                                                )\n"
+	"{\n"
+	"    \n"
+	"	int i = get_global_id(0);\n"
+	"	if (i>=numConcavePairs)\n"
+	"		return;\n"
+	"    \n"
+	"	concaveHasSeparatingNormals[i] = 0;\n"
+	"    \n"
+	"	int pairIdx = i;\n"
+	"    \n"
+	"	int bodyIndexA = concavePairs[i].x;\n"
+	"	int bodyIndexB = concavePairs[i].y;\n"
+	"    \n"
+	"	int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
+	"	int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
+	"    \n"
+	"	int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
+	"	int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
+	"    \n"
+	"	if (collidables[collidableIndexB].m_shapeType!=SHAPE_CONVEX_HULL&&\n"
+	"		collidables[collidableIndexB].m_shapeType!=SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
+	"	{\n"
+	"		concavePairs[pairIdx].w = -1;\n"
+	"		return;\n"
+	"	}\n"
+	"    \n"
+	"    \n"
+	"    \n"
+	"	int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
+	"	int numActualConcaveConvexTests = 0;\n"
+	"	\n"
+	"	int f = concavePairs[i].z;\n"
+	"	\n"
+	"	bool overlap = false;\n"
+	"	\n"
+	"	ConvexPolyhedronCL convexPolyhedronA;\n"
+	"    \n"
+	"	//add 3 vertices of the triangle\n"
+	"	convexPolyhedronA.m_numVertices = 3;\n"
+	"	convexPolyhedronA.m_vertexOffset = 0;\n"
+	"	float4	localCenter = make_float4(0.f,0.f,0.f,0.f);\n"
+	"    \n"
+	"	btGpuFace face = faces[convexShapes[shapeIndexA].m_faceOffset+f];\n"
+	"	float4 triMinAabb, triMaxAabb;\n"
+	"	btAabbCL triAabb;\n"
+	"	triAabb.m_min = make_float4(1e30f,1e30f,1e30f,0.f);\n"
+	"	triAabb.m_max = make_float4(-1e30f,-1e30f,-1e30f,0.f);\n"
+	"	\n"
+	"	float4 verticesA[3];\n"
+	"	for (int i=0;i<3;i++)\n"
+	"	{\n"
+	"		int index = indices[face.m_indexOffset+i];\n"
+	"		float4 vert = vertices[convexShapes[shapeIndexA].m_vertexOffset+index];\n"
+	"		verticesA[i] = vert;\n"
+	"		localCenter += vert;\n"
+	"        \n"
+	"		triAabb.m_min = min(triAabb.m_min,vert);\n"
+	"		triAabb.m_max = max(triAabb.m_max,vert);\n"
+	"        \n"
+	"	}\n"
+	"    \n"
+	"	overlap = true;\n"
+	"	overlap = (triAabb.m_min.x > aabbs[bodyIndexB].m_max.x || triAabb.m_max.x < aabbs[bodyIndexB].m_min.x) ? false : overlap;\n"
+	"	overlap = (triAabb.m_min.z > aabbs[bodyIndexB].m_max.z || triAabb.m_max.z < aabbs[bodyIndexB].m_min.z) ? false : overlap;\n"
+	"	overlap = (triAabb.m_min.y > aabbs[bodyIndexB].m_max.y || triAabb.m_max.y < aabbs[bodyIndexB].m_min.y) ? false : overlap;\n"
+	"    \n"
+	"	if (overlap)\n"
+	"	{\n"
+	"		float dmin = FLT_MAX;\n"
+	"		int hasSeparatingAxis=5;\n"
+	"		float4 sepAxis=make_float4(1,2,3,4);\n"
+	"        \n"
+	"		int localCC=0;\n"
+	"		numActualConcaveConvexTests++;\n"
+	"        \n"
+	"		//a triangle has 3 unique edges\n"
+	"		convexPolyhedronA.m_numUniqueEdges = 3;\n"
+	"		convexPolyhedronA.m_uniqueEdgesOffset = 0;\n"
+	"		float4 uniqueEdgesA[3];\n"
+	"		\n"
+	"		uniqueEdgesA[0] = (verticesA[1]-verticesA[0]);\n"
+	"		uniqueEdgesA[1] = (verticesA[2]-verticesA[1]);\n"
+	"		uniqueEdgesA[2] = (verticesA[0]-verticesA[2]);\n"
+	"        \n"
+	"        \n"
+	"		convexPolyhedronA.m_faceOffset = 0;\n"
+	"        \n"
+	"		float4 normal = make_float4(face.m_plane.x,face.m_plane.y,face.m_plane.z,0.f);\n"
+	"        \n"
+	"		btGpuFace facesA[TRIANGLE_NUM_CONVEX_FACES];\n"
+	"		int indicesA[3+3+2+2+2];\n"
+	"		int curUsedIndices=0;\n"
+	"		int fidx=0;\n"
+	"        \n"
+	"		//front size of triangle\n"
+	"		{\n"
+	"			facesA[fidx].m_indexOffset=curUsedIndices;\n"
+	"			indicesA[0] = 0;\n"
+	"			indicesA[1] = 1;\n"
+	"			indicesA[2] = 2;\n"
+	"			curUsedIndices+=3;\n"
+	"			float c = face.m_plane.w;\n"
+	"			facesA[fidx].m_plane.x = normal.x;\n"
+	"			facesA[fidx].m_plane.y = normal.y;\n"
+	"			facesA[fidx].m_plane.z = normal.z;\n"
+	"			facesA[fidx].m_plane.w = c;\n"
+	"			facesA[fidx].m_numIndices=3;\n"
+	"		}\n"
+	"		fidx++;\n"
+	"		//back size of triangle\n"
+	"		{\n"
+	"			facesA[fidx].m_indexOffset=curUsedIndices;\n"
+	"			indicesA[3]=2;\n"
+	"			indicesA[4]=1;\n"
+	"			indicesA[5]=0;\n"
+	"			curUsedIndices+=3;\n"
+	"			float c = dot(normal,verticesA[0]);\n"
+	"			float c1 = -face.m_plane.w;\n"
+	"			facesA[fidx].m_plane.x = -normal.x;\n"
+	"			facesA[fidx].m_plane.y = -normal.y;\n"
+	"			facesA[fidx].m_plane.z = -normal.z;\n"
+	"			facesA[fidx].m_plane.w = c;\n"
+	"			facesA[fidx].m_numIndices=3;\n"
+	"		}\n"
+	"		fidx++;\n"
+	"        \n"
+	"		bool addEdgePlanes = true;\n"
+	"		if (addEdgePlanes)\n"
+	"		{\n"
+	"			int numVertices=3;\n"
+	"			int prevVertex = numVertices-1;\n"
+	"			for (int i=0;i<numVertices;i++)\n"
+	"			{\n"
+	"				float4 v0 = verticesA[i];\n"
+	"				float4 v1 = verticesA[prevVertex];\n"
+	"                \n"
+	"				float4 edgeNormal = normalize(cross(normal,v1-v0));\n"
+	"				float c = -dot(edgeNormal,v0);\n"
+	"                \n"
+	"				facesA[fidx].m_numIndices = 2;\n"
+	"				facesA[fidx].m_indexOffset=curUsedIndices;\n"
+	"				indicesA[curUsedIndices++]=i;\n"
+	"				indicesA[curUsedIndices++]=prevVertex;\n"
+	"                \n"
+	"				facesA[fidx].m_plane.x = edgeNormal.x;\n"
+	"				facesA[fidx].m_plane.y = edgeNormal.y;\n"
+	"				facesA[fidx].m_plane.z = edgeNormal.z;\n"
+	"				facesA[fidx].m_plane.w = c;\n"
+	"				fidx++;\n"
+	"				prevVertex = i;\n"
+	"			}\n"
+	"		}\n"
+	"		convexPolyhedronA.m_numFaces = TRIANGLE_NUM_CONVEX_FACES;\n"
+	"		convexPolyhedronA.m_localCenter = localCenter*(1.f/3.f);\n"
+	"        \n"
+	"        \n"
+	"		float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
+	"		posA.w = 0.f;\n"
+	"		float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
+	"		posB.w = 0.f;\n"
+	"        \n"
+	"		float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
+	"		float4 ornB =rigidBodies[bodyIndexB].m_quat;\n"
+	"        \n"
+	"		\n"
+	"        \n"
+	"        \n"
+	"		///////////////////\n"
+	"		///compound shape support\n"
+	"        \n"
+	"		if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
+	"		{\n"
+	"			int compoundChild = concavePairs[pairIdx].w;\n"
+	"			int childShapeIndexB = compoundChild;//collidables[collidableIndexB].m_shapeIndex+compoundChild;\n"
+	"			int childColIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;\n"
+	"			float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;\n"
+	"			float4 childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;\n"
+	"			float4 newPosB = transform(&childPosB,&posB,&ornB);\n"
+	"			float4 newOrnB = qtMul(ornB,childOrnB);\n"
+	"			posB = newPosB;\n"
+	"			ornB = newOrnB;\n"
+	"			shapeIndexB = collidables[childColIndexB].m_shapeIndex;\n"
+	"		}\n"
+	"		//////////////////\n"
+	"        \n"
+	"		float4 c0local = convexPolyhedronA.m_localCenter;\n"
+	"		float4 c0 = transform(&c0local, &posA, &ornA);\n"
+	"		float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
+	"		float4 c1 = transform(&c1local,&posB,&ornB);\n"
+	"		const float4 DeltaC2 = c0 - c1;\n"
+	"        \n"
+	"        \n"
+	"		bool sepA = findSeparatingAxisLocalA(	&convexPolyhedronA, &convexShapes[shapeIndexB],\n"
+	"                                             posA,ornA,\n"
+	"                                             posB,ornB,\n"
+	"                                             DeltaC2,\n"
+	"                                             verticesA,uniqueEdgesA,facesA,indicesA,\n"
+	"                                             vertices,uniqueEdges,faces,indices,\n"
+	"                                             &sepAxis,&dmin);\n"
+	"		hasSeparatingAxis = 4;\n"
+	"		if (!sepA)\n"
+	"		{\n"
+	"			hasSeparatingAxis = 0;\n"
+	"		} else\n"
+	"		{\n"
+	"			bool sepB = findSeparatingAxisLocalB(	&convexShapes[shapeIndexB],&convexPolyhedronA,\n"
+	"                                                 posB,ornB,\n"
+	"                                                 posA,ornA,\n"
+	"                                                 DeltaC2,\n"
+	"                                                 vertices,uniqueEdges,faces,indices,\n"
+	"                                                 verticesA,uniqueEdgesA,facesA,indicesA,\n"
+	"                                                 &sepAxis,&dmin);\n"
+	"            \n"
+	"			if (!sepB)\n"
+	"			{\n"
+	"				hasSeparatingAxis = 0;\n"
+	"			} else\n"
+	"			{\n"
+	"				hasSeparatingAxis = 1;\n"
+	"			}\n"
+	"		}	\n"
+	"		\n"
+	"		if (hasSeparatingAxis)\n"
+	"		{\n"
+	"            dmins[i] = dmin;\n"
+	"			concaveSeparatingNormalsOut[pairIdx]=sepAxis;\n"
+	"			concaveHasSeparatingNormals[i]=1;\n"
+	"            \n"
+	"		} else\n"
+	"		{	\n"
+	"			//mark this pair as in-active\n"
+	"			concavePairs[pairIdx].w = -1;\n"
+	"		}\n"
+	"	}\n"
+	"	else\n"
+	"	{	\n"
+	"		//mark this pair as in-active\n"
+	"		concavePairs[pairIdx].w = -1;\n"
+	"	}\n"
+	"}\n"
+	"// work-in-progress\n"
+	"__kernel void   findConcaveSeparatingAxisEdgeEdgeKernel( __global int4* concavePairs,\n"
+	"                                                          __global const BodyData* rigidBodies,\n"
+	"                                                          __global const btCollidableGpu* collidables,\n"
+	"                                                          __global const ConvexPolyhedronCL* convexShapes,\n"
+	"                                                          __global const float4* vertices,\n"
+	"                                                          __global const float4* uniqueEdges,\n"
+	"                                                          __global const btGpuFace* faces,\n"
+	"                                                          __global const int* indices,\n"
+	"                                                          __global const btGpuChildShape* gpuChildShapes,\n"
+	"                                                          __global btAabbCL* aabbs,\n"
+	"                                                          __global float4* concaveSeparatingNormalsOut,\n"
+	"                                                          __global int* concaveHasSeparatingNormals,\n"
+	"                                                          __global int4* clippingFacesOut,\n"
+	"                                                          __global float4* worldVertsA1GPU,\n"
+	"                                                          __global float4*  worldNormalsAGPU,\n"
+	"                                                          __global float4* worldVertsB1GPU,\n"
+	"                                                          __global float* dmins,\n"
+	"                                                          int vertexFaceCapacity,\n"
+	"                                                          int numConcavePairs\n"
+	"                                                          )\n"
+	"{\n"
+	"    \n"
+	"	int i = get_global_id(0);\n"
+	"	if (i>=numConcavePairs)\n"
+	"		return;\n"
+	"    \n"
+	"	if (!concaveHasSeparatingNormals[i])\n"
+	"        return;\n"
+	"    \n"
+	"	int pairIdx = i;\n"
+	"    \n"
+	"	int bodyIndexA = concavePairs[i].x;\n"
+	"	int bodyIndexB = concavePairs[i].y;\n"
+	"    \n"
+	"	int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
+	"	int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
+	"    \n"
+	"	int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
+	"	int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
+	"    \n"
+	"    \n"
+	"	int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
+	"	int numActualConcaveConvexTests = 0;\n"
+	"	\n"
+	"	int f = concavePairs[i].z;\n"
+	"	\n"
+	"	bool overlap = false;\n"
+	"	\n"
+	"	ConvexPolyhedronCL convexPolyhedronA;\n"
+	"    \n"
+	"	//add 3 vertices of the triangle\n"
+	"	convexPolyhedronA.m_numVertices = 3;\n"
+	"	convexPolyhedronA.m_vertexOffset = 0;\n"
+	"	float4	localCenter = make_float4(0.f,0.f,0.f,0.f);\n"
+	"    \n"
+	"	btGpuFace face = faces[convexShapes[shapeIndexA].m_faceOffset+f];\n"
+	"	float4 triMinAabb, triMaxAabb;\n"
+	"	btAabbCL triAabb;\n"
+	"	triAabb.m_min = make_float4(1e30f,1e30f,1e30f,0.f);\n"
+	"	triAabb.m_max = make_float4(-1e30f,-1e30f,-1e30f,0.f);\n"
+	"	\n"
+	"	float4 verticesA[3];\n"
+	"	for (int i=0;i<3;i++)\n"
+	"	{\n"
+	"		int index = indices[face.m_indexOffset+i];\n"
+	"		float4 vert = vertices[convexShapes[shapeIndexA].m_vertexOffset+index];\n"
+	"		verticesA[i] = vert;\n"
+	"		localCenter += vert;\n"
+	"        \n"
+	"		triAabb.m_min = min(triAabb.m_min,vert);\n"
+	"		triAabb.m_max = max(triAabb.m_max,vert);\n"
+	"        \n"
+	"	}\n"
+	"    \n"
+	"	overlap = true;\n"
+	"	overlap = (triAabb.m_min.x > aabbs[bodyIndexB].m_max.x || triAabb.m_max.x < aabbs[bodyIndexB].m_min.x) ? false : overlap;\n"
+	"	overlap = (triAabb.m_min.z > aabbs[bodyIndexB].m_max.z || triAabb.m_max.z < aabbs[bodyIndexB].m_min.z) ? false : overlap;\n"
+	"	overlap = (triAabb.m_min.y > aabbs[bodyIndexB].m_max.y || triAabb.m_max.y < aabbs[bodyIndexB].m_min.y) ? false : overlap;\n"
+	"    \n"
+	"	if (overlap)\n"
+	"	{\n"
+	"		float dmin = dmins[i];\n"
+	"		int hasSeparatingAxis=5;\n"
+	"		float4 sepAxis=make_float4(1,2,3,4);\n"
+	"        sepAxis = concaveSeparatingNormalsOut[pairIdx];\n"
+	"        \n"
+	"		int localCC=0;\n"
+	"		numActualConcaveConvexTests++;\n"
+	"        \n"
+	"		//a triangle has 3 unique edges\n"
+	"		convexPolyhedronA.m_numUniqueEdges = 3;\n"
+	"		convexPolyhedronA.m_uniqueEdgesOffset = 0;\n"
+	"		float4 uniqueEdgesA[3];\n"
+	"		\n"
+	"		uniqueEdgesA[0] = (verticesA[1]-verticesA[0]);\n"
+	"		uniqueEdgesA[1] = (verticesA[2]-verticesA[1]);\n"
+	"		uniqueEdgesA[2] = (verticesA[0]-verticesA[2]);\n"
+	"        \n"
+	"        \n"
+	"		convexPolyhedronA.m_faceOffset = 0;\n"
+	"        \n"
+	"		float4 normal = make_float4(face.m_plane.x,face.m_plane.y,face.m_plane.z,0.f);\n"
+	"        \n"
+	"		btGpuFace facesA[TRIANGLE_NUM_CONVEX_FACES];\n"
+	"		int indicesA[3+3+2+2+2];\n"
+	"		int curUsedIndices=0;\n"
+	"		int fidx=0;\n"
+	"        \n"
+	"		//front size of triangle\n"
+	"		{\n"
+	"			facesA[fidx].m_indexOffset=curUsedIndices;\n"
+	"			indicesA[0] = 0;\n"
+	"			indicesA[1] = 1;\n"
+	"			indicesA[2] = 2;\n"
+	"			curUsedIndices+=3;\n"
+	"			float c = face.m_plane.w;\n"
+	"			facesA[fidx].m_plane.x = normal.x;\n"
+	"			facesA[fidx].m_plane.y = normal.y;\n"
+	"			facesA[fidx].m_plane.z = normal.z;\n"
+	"			facesA[fidx].m_plane.w = c;\n"
+	"			facesA[fidx].m_numIndices=3;\n"
+	"		}\n"
+	"		fidx++;\n"
+	"		//back size of triangle\n"
+	"		{\n"
+	"			facesA[fidx].m_indexOffset=curUsedIndices;\n"
+	"			indicesA[3]=2;\n"
+	"			indicesA[4]=1;\n"
+	"			indicesA[5]=0;\n"
+	"			curUsedIndices+=3;\n"
+	"			float c = dot(normal,verticesA[0]);\n"
+	"			float c1 = -face.m_plane.w;\n"
+	"			facesA[fidx].m_plane.x = -normal.x;\n"
+	"			facesA[fidx].m_plane.y = -normal.y;\n"
+	"			facesA[fidx].m_plane.z = -normal.z;\n"
+	"			facesA[fidx].m_plane.w = c;\n"
+	"			facesA[fidx].m_numIndices=3;\n"
+	"		}\n"
+	"		fidx++;\n"
+	"        \n"
+	"		bool addEdgePlanes = true;\n"
+	"		if (addEdgePlanes)\n"
+	"		{\n"
+	"			int numVertices=3;\n"
+	"			int prevVertex = numVertices-1;\n"
+	"			for (int i=0;i<numVertices;i++)\n"
+	"			{\n"
+	"				float4 v0 = verticesA[i];\n"
+	"				float4 v1 = verticesA[prevVertex];\n"
+	"                \n"
+	"				float4 edgeNormal = normalize(cross(normal,v1-v0));\n"
+	"				float c = -dot(edgeNormal,v0);\n"
+	"                \n"
+	"				facesA[fidx].m_numIndices = 2;\n"
+	"				facesA[fidx].m_indexOffset=curUsedIndices;\n"
+	"				indicesA[curUsedIndices++]=i;\n"
+	"				indicesA[curUsedIndices++]=prevVertex;\n"
+	"                \n"
+	"				facesA[fidx].m_plane.x = edgeNormal.x;\n"
+	"				facesA[fidx].m_plane.y = edgeNormal.y;\n"
+	"				facesA[fidx].m_plane.z = edgeNormal.z;\n"
+	"				facesA[fidx].m_plane.w = c;\n"
+	"				fidx++;\n"
+	"				prevVertex = i;\n"
+	"			}\n"
+	"		}\n"
+	"		convexPolyhedronA.m_numFaces = TRIANGLE_NUM_CONVEX_FACES;\n"
+	"		convexPolyhedronA.m_localCenter = localCenter*(1.f/3.f);\n"
+	"        \n"
+	"        \n"
+	"		float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
+	"		posA.w = 0.f;\n"
+	"		float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
+	"		posB.w = 0.f;\n"
+	"        \n"
+	"		float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
+	"		float4 ornB =rigidBodies[bodyIndexB].m_quat;\n"
+	"        \n"
+	"		\n"
+	"        \n"
+	"        \n"
+	"		///////////////////\n"
+	"		///compound shape support\n"
+	"        \n"
+	"		if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
+	"		{\n"
+	"			int compoundChild = concavePairs[pairIdx].w;\n"
+	"			int childShapeIndexB = compoundChild;//collidables[collidableIndexB].m_shapeIndex+compoundChild;\n"
+	"			int childColIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;\n"
+	"			float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;\n"
+	"			float4 childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;\n"
+	"			float4 newPosB = transform(&childPosB,&posB,&ornB);\n"
+	"			float4 newOrnB = qtMul(ornB,childOrnB);\n"
+	"			posB = newPosB;\n"
+	"			ornB = newOrnB;\n"
+	"			shapeIndexB = collidables[childColIndexB].m_shapeIndex;\n"
+	"		}\n"
+	"		//////////////////\n"
+	"        \n"
+	"		float4 c0local = convexPolyhedronA.m_localCenter;\n"
+	"		float4 c0 = transform(&c0local, &posA, &ornA);\n"
+	"		float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
+	"		float4 c1 = transform(&c1local,&posB,&ornB);\n"
+	"		const float4 DeltaC2 = c0 - c1;\n"
+	"        \n"
+	"        \n"
+	"		{\n"
+	"			bool sepEE = findSeparatingAxisEdgeEdgeLocalA(	&convexPolyhedronA, &convexShapes[shapeIndexB],\n"
+	"                                                              posA,ornA,\n"
+	"                                                              posB,ornB,\n"
+	"                                                              DeltaC2,\n"
+	"                                                              verticesA,uniqueEdgesA,facesA,indicesA,\n"
+	"                                                              vertices,uniqueEdges,faces,indices,\n"
+	"                                                              &sepAxis,&dmin);\n"
+	"                \n"
+	"			if (!sepEE)\n"
+	"			{\n"
+	"				hasSeparatingAxis = 0;\n"
+	"			} else\n"
+	"			{\n"
+	"				hasSeparatingAxis = 1;\n"
+	"			}\n"
+	"		}\n"
+	"		\n"
+	"		\n"
+	"		if (hasSeparatingAxis)\n"
+	"		{\n"
+	"			sepAxis.w = dmin;\n"
+	"            dmins[i] = dmin;\n"
+	"			concaveSeparatingNormalsOut[pairIdx]=sepAxis;\n"
+	"			concaveHasSeparatingNormals[i]=1;\n"
+	"           \n"
+	" 	float minDist = -1e30f;\n"
+	"			float maxDist = 0.02f;\n"
+	"            \n"
+	"            findClippingFaces(sepAxis,\n"
+	"                              &convexPolyhedronA,\n"
+	"                              &convexShapes[shapeIndexB],\n"
+	"                              posA,ornA,\n"
+	"                              posB,ornB,\n"
+	"                              worldVertsA1GPU,\n"
+	"                              worldNormalsAGPU,\n"
+	"                              worldVertsB1GPU,\n"
+	"                              vertexFaceCapacity,\n"
+	"                              minDist, maxDist,\n"
+	"                              verticesA,\n"
+	"                              facesA,\n"
+	"                              indicesA,\n"
+	"                              vertices,\n"
+	"                              faces,\n"
+	"                              indices,\n"
+	"                              clippingFacesOut, pairIdx);\n"
+	"	           \n"
+	"            \n"
+	"		} else\n"
+	"		{	\n"
+	"			//mark this pair as in-active\n"
+	"			concavePairs[pairIdx].w = -1;\n"
+	"		}\n"
+	"	}\n"
+	"	else\n"
+	"	{	\n"
+	"		//mark this pair as in-active\n"
+	"		concavePairs[pairIdx].w = -1;\n"
+	"	}\n"
+	"	\n"
+	"	concavePairs[i].z = -1;//for the next stage, z is used to determine existing contact points\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/satKernels.h b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/satKernels.h
index 6f8b0a90db7..e627af2799c 100644
--- a/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/satKernels.h
+++ b/thirdparty/bullet/Bullet3OpenCL/NarrowphaseCollision/kernels/satKernels.h
@@ -1,2104 +1,2103 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* satKernelsCL= \
-"//keep this enum in sync with the CPU version (in btCollidable.h)\n"
-"//written by Erwin Coumans\n"
-"#define SHAPE_CONVEX_HULL 3\n"
-"#define SHAPE_CONCAVE_TRIMESH 5\n"
-"#define TRIANGLE_NUM_CONVEX_FACES 5\n"
-"#define SHAPE_COMPOUND_OF_CONVEX_HULLS 6\n"
-"#define B3_MAX_STACK_DEPTH 256\n"
-"typedef unsigned int u32;\n"
-"///keep this in sync with btCollidable.h\n"
-"typedef struct\n"
-"{\n"
-"	union {\n"
-"		int m_numChildShapes;\n"
-"		int m_bvhIndex;\n"
-"	};\n"
-"	union\n"
-"	{\n"
-"		float m_radius;\n"
-"		int	m_compoundBvhIndex;\n"
-"	};\n"
-"	\n"
-"	int m_shapeType;\n"
-"	int m_shapeIndex;\n"
-"	\n"
-"} btCollidableGpu;\n"
-"#define MAX_NUM_PARTS_IN_BITS 10\n"
-"///b3QuantizedBvhNode is a compressed aabb node, 16 bytes.\n"
-"///Node can be used for leafnode or internal node. Leafnodes can point to 32-bit triangle index (non-negative range).\n"
-"typedef struct\n"
-"{\n"
-"	//12 bytes\n"
-"	unsigned short int	m_quantizedAabbMin[3];\n"
-"	unsigned short int	m_quantizedAabbMax[3];\n"
-"	//4 bytes\n"
-"	int	m_escapeIndexOrTriangleIndex;\n"
-"} b3QuantizedBvhNode;\n"
-"typedef struct\n"
-"{\n"
-"	float4		m_aabbMin;\n"
-"	float4		m_aabbMax;\n"
-"	float4		m_quantization;\n"
-"	int			m_numNodes;\n"
-"	int			m_numSubTrees;\n"
-"	int			m_nodeOffset;\n"
-"	int			m_subTreeOffset;\n"
-"} b3BvhInfo;\n"
-"int	getTriangleIndex(const b3QuantizedBvhNode* rootNode)\n"
-"{\n"
-"	unsigned int x=0;\n"
-"	unsigned int y = (~(x&0))<<(31-MAX_NUM_PARTS_IN_BITS);\n"
-"	// Get only the lower bits where the triangle index is stored\n"
-"	return (rootNode->m_escapeIndexOrTriangleIndex&~(y));\n"
-"}\n"
-"int	getTriangleIndexGlobal(__global const b3QuantizedBvhNode* rootNode)\n"
-"{\n"
-"	unsigned int x=0;\n"
-"	unsigned int y = (~(x&0))<<(31-MAX_NUM_PARTS_IN_BITS);\n"
-"	// Get only the lower bits where the triangle index is stored\n"
-"	return (rootNode->m_escapeIndexOrTriangleIndex&~(y));\n"
-"}\n"
-"int isLeafNode(const b3QuantizedBvhNode* rootNode)\n"
-"{\n"
-"	//skipindex is negative (internal node), triangleindex >=0 (leafnode)\n"
-"	return (rootNode->m_escapeIndexOrTriangleIndex >= 0)? 1 : 0;\n"
-"}\n"
-"int isLeafNodeGlobal(__global const b3QuantizedBvhNode* rootNode)\n"
-"{\n"
-"	//skipindex is negative (internal node), triangleindex >=0 (leafnode)\n"
-"	return (rootNode->m_escapeIndexOrTriangleIndex >= 0)? 1 : 0;\n"
-"}\n"
-"	\n"
-"int getEscapeIndex(const b3QuantizedBvhNode* rootNode)\n"
-"{\n"
-"	return -rootNode->m_escapeIndexOrTriangleIndex;\n"
-"}\n"
-"int getEscapeIndexGlobal(__global const b3QuantizedBvhNode* rootNode)\n"
-"{\n"
-"	return -rootNode->m_escapeIndexOrTriangleIndex;\n"
-"}\n"
-"typedef struct\n"
-"{\n"
-"	//12 bytes\n"
-"	unsigned short int	m_quantizedAabbMin[3];\n"
-"	unsigned short int	m_quantizedAabbMax[3];\n"
-"	//4 bytes, points to the root of the subtree\n"
-"	int			m_rootNodeIndex;\n"
-"	//4 bytes\n"
-"	int			m_subtreeSize;\n"
-"	int			m_padding[3];\n"
-"} b3BvhSubtreeInfo;\n"
-"typedef struct\n"
-"{\n"
-"	float4	m_childPosition;\n"
-"	float4	m_childOrientation;\n"
-"	int m_shapeIndex;\n"
-"	int m_unused0;\n"
-"	int m_unused1;\n"
-"	int m_unused2;\n"
-"} btGpuChildShape;\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_pos;\n"
-"	float4 m_quat;\n"
-"	float4 m_linVel;\n"
-"	float4 m_angVel;\n"
-"	u32 m_collidableIdx;\n"
-"	float m_invMass;\n"
-"	float m_restituitionCoeff;\n"
-"	float m_frictionCoeff;\n"
-"} BodyData;\n"
-"typedef struct  \n"
-"{\n"
-"	float4		m_localCenter;\n"
-"	float4		m_extents;\n"
-"	float4		mC;\n"
-"	float4		mE;\n"
-"	\n"
-"	float			m_radius;\n"
-"	int	m_faceOffset;\n"
-"	int m_numFaces;\n"
-"	int	m_numVertices;\n"
-"	int m_vertexOffset;\n"
-"	int	m_uniqueEdgesOffset;\n"
-"	int	m_numUniqueEdges;\n"
-"	int m_unused;\n"
-"} ConvexPolyhedronCL;\n"
-"typedef struct \n"
-"{\n"
-"	union\n"
-"	{\n"
-"		float4	m_min;\n"
-"		float   m_minElems[4];\n"
-"		int			m_minIndices[4];\n"
-"	};\n"
-"	union\n"
-"	{\n"
-"		float4	m_max;\n"
-"		float   m_maxElems[4];\n"
-"		int			m_maxIndices[4];\n"
-"	};\n"
-"} btAabbCL;\n"
-"#ifndef B3_AABB_H\n"
-"#define B3_AABB_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#define B3_FLOAT4_H\n"
-"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
-"#define B3_PLATFORM_DEFINITIONS_H\n"
-"struct MyTest\n"
-"{\n"
-"	int bla;\n"
-"};\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
-"#define B3_LARGE_FLOAT 1e18f\n"
-"#define B3_INFINITY 1e18f\n"
-"#define b3Assert(a)\n"
-"#define b3ConstArray(a) __global const a*\n"
-"#define b3AtomicInc atomic_inc\n"
-"#define b3AtomicAdd atomic_add\n"
-"#define b3Fabs fabs\n"
-"#define b3Sqrt native_sqrt\n"
-"#define b3Sin native_sin\n"
-"#define b3Cos native_cos\n"
-"#define B3_STATIC\n"
-"#endif\n"
-"#endif\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"	typedef float4	b3Float4;\n"
-"	#define b3Float4ConstArg const b3Float4\n"
-"	#define b3MakeFloat4 (float4)\n"
-"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return dot(a1, b1);\n"
-"	}\n"
-"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return cross(a1, b1);\n"
-"	}\n"
-"	#define b3MinFloat4 min\n"
-"	#define b3MaxFloat4 max\n"
-"	#define b3Normalized(a) normalize(a)\n"
-"#endif \n"
-"		\n"
-"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
-"{\n"
-"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
-"		return false;\n"
-"	return true;\n"
-"}\n"
-"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
-"{\n"
-"    float maxDot = -B3_INFINITY;\n"
-"    int i = 0;\n"
-"    int ptIndex = -1;\n"
-"    for( i = 0; i < vecLen; i++ )\n"
-"    {\n"
-"        float dot = b3Dot3F4(vecArray[i],vec);\n"
-"            \n"
-"        if( dot > maxDot )\n"
-"        {\n"
-"            maxDot = dot;\n"
-"            ptIndex = i;\n"
-"        }\n"
-"    }\n"
-"	b3Assert(ptIndex>=0);\n"
-"    if (ptIndex<0)\n"
-"	{\n"
-"		ptIndex = 0;\n"
-"	}\n"
-"    *dotOut = maxDot;\n"
-"    return ptIndex;\n"
-"}\n"
-"#endif //B3_FLOAT4_H\n"
-"#ifndef B3_MAT3x3_H\n"
-"#define B3_MAT3x3_H\n"
-"#ifndef B3_QUAT_H\n"
-"#define B3_QUAT_H\n"
-"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif\n"
-"#endif\n"
-"#ifndef B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"	typedef float4	b3Quat;\n"
-"	#define b3QuatConstArg const b3Quat\n"
-"	\n"
-"	\n"
-"inline float4 b3FastNormalize4(float4 v)\n"
-"{\n"
-"	v = (float4)(v.xyz,0.f);\n"
-"	return fast_normalize(v);\n"
-"}\n"
-"	\n"
-"inline b3Quat b3QuatMul(b3Quat a, b3Quat b);\n"
-"inline b3Quat b3QuatNormalized(b3QuatConstArg in);\n"
-"inline b3Quat b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec);\n"
-"inline b3Quat b3QuatInvert(b3QuatConstArg q);\n"
-"inline b3Quat b3QuatInverse(b3QuatConstArg q);\n"
-"inline b3Quat b3QuatMul(b3QuatConstArg a, b3QuatConstArg b)\n"
-"{\n"
-"	b3Quat ans;\n"
-"	ans = b3Cross3( a, b );\n"
-"	ans += a.w*b+b.w*a;\n"
-"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
-"	ans.w = a.w*b.w - b3Dot3F4(a, b);\n"
-"	return ans;\n"
-"}\n"
-"inline b3Quat b3QuatNormalized(b3QuatConstArg in)\n"
-"{\n"
-"	b3Quat q;\n"
-"	q=in;\n"
-"	//return b3FastNormalize4(in);\n"
-"	float len = native_sqrt(dot(q, q));\n"
-"	if(len > 0.f)\n"
-"	{\n"
-"		q *= 1.f / len;\n"
-"	}\n"
-"	else\n"
-"	{\n"
-"		q.x = q.y = q.z = 0.f;\n"
-"		q.w = 1.f;\n"
-"	}\n"
-"	return q;\n"
-"}\n"
-"inline float4 b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
-"{\n"
-"	b3Quat qInv = b3QuatInvert( q );\n"
-"	float4 vcpy = vec;\n"
-"	vcpy.w = 0.f;\n"
-"	float4 out = b3QuatMul(b3QuatMul(q,vcpy),qInv);\n"
-"	return out;\n"
-"}\n"
-"inline b3Quat b3QuatInverse(b3QuatConstArg q)\n"
-"{\n"
-"	return (b3Quat)(-q.xyz, q.w);\n"
-"}\n"
-"inline b3Quat b3QuatInvert(b3QuatConstArg q)\n"
-"{\n"
-"	return (b3Quat)(-q.xyz, q.w);\n"
-"}\n"
-"inline float4 b3QuatInvRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
-"{\n"
-"	return b3QuatRotate( b3QuatInvert( q ), vec );\n"
-"}\n"
-"inline b3Float4 b3TransformPoint(b3Float4ConstArg point, b3Float4ConstArg translation, b3QuatConstArg  orientation)\n"
-"{\n"
-"	return b3QuatRotate( orientation, point ) + (translation);\n"
-"}\n"
-"	\n"
-"#endif \n"
-"#endif //B3_QUAT_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"typedef struct\n"
-"{\n"
-"	b3Float4 m_row[3];\n"
-"}b3Mat3x3;\n"
-"#define b3Mat3x3ConstArg const b3Mat3x3\n"
-"#define b3GetRow(m,row) (m.m_row[row])\n"
-"inline b3Mat3x3 b3QuatGetRotationMatrix(b3Quat quat)\n"
-"{\n"
-"	b3Float4 quat2 = (b3Float4)(quat.x*quat.x, quat.y*quat.y, quat.z*quat.z, 0.f);\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0].x=1-2*quat2.y-2*quat2.z;\n"
-"	out.m_row[0].y=2*quat.x*quat.y-2*quat.w*quat.z;\n"
-"	out.m_row[0].z=2*quat.x*quat.z+2*quat.w*quat.y;\n"
-"	out.m_row[0].w = 0.f;\n"
-"	out.m_row[1].x=2*quat.x*quat.y+2*quat.w*quat.z;\n"
-"	out.m_row[1].y=1-2*quat2.x-2*quat2.z;\n"
-"	out.m_row[1].z=2*quat.y*quat.z-2*quat.w*quat.x;\n"
-"	out.m_row[1].w = 0.f;\n"
-"	out.m_row[2].x=2*quat.x*quat.z-2*quat.w*quat.y;\n"
-"	out.m_row[2].y=2*quat.y*quat.z+2*quat.w*quat.x;\n"
-"	out.m_row[2].z=1-2*quat2.x-2*quat2.y;\n"
-"	out.m_row[2].w = 0.f;\n"
-"	return out;\n"
-"}\n"
-"inline b3Mat3x3 b3AbsoluteMat3x3(b3Mat3x3ConstArg matIn)\n"
-"{\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0] = fabs(matIn.m_row[0]);\n"
-"	out.m_row[1] = fabs(matIn.m_row[1]);\n"
-"	out.m_row[2] = fabs(matIn.m_row[2]);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtZero();\n"
-"__inline\n"
-"b3Mat3x3 mtIdentity();\n"
-"__inline\n"
-"b3Mat3x3 mtTranspose(b3Mat3x3 m);\n"
-"__inline\n"
-"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b);\n"
-"__inline\n"
-"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b);\n"
-"__inline\n"
-"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b);\n"
-"__inline\n"
-"b3Mat3x3 mtZero()\n"
-"{\n"
-"	b3Mat3x3 m;\n"
-"	m.m_row[0] = (b3Float4)(0.f);\n"
-"	m.m_row[1] = (b3Float4)(0.f);\n"
-"	m.m_row[2] = (b3Float4)(0.f);\n"
-"	return m;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtIdentity()\n"
-"{\n"
-"	b3Mat3x3 m;\n"
-"	m.m_row[0] = (b3Float4)(1,0,0,0);\n"
-"	m.m_row[1] = (b3Float4)(0,1,0,0);\n"
-"	m.m_row[2] = (b3Float4)(0,0,1,0);\n"
-"	return m;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtTranspose(b3Mat3x3 m)\n"
-"{\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0] = (b3Float4)(m.m_row[0].x, m.m_row[1].x, m.m_row[2].x, 0.f);\n"
-"	out.m_row[1] = (b3Float4)(m.m_row[0].y, m.m_row[1].y, m.m_row[2].y, 0.f);\n"
-"	out.m_row[2] = (b3Float4)(m.m_row[0].z, m.m_row[1].z, m.m_row[2].z, 0.f);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b)\n"
-"{\n"
-"	b3Mat3x3 transB;\n"
-"	transB = mtTranspose( b );\n"
-"	b3Mat3x3 ans;\n"
-"	//	why this doesn't run when 0ing in the for{}\n"
-"	a.m_row[0].w = 0.f;\n"
-"	a.m_row[1].w = 0.f;\n"
-"	a.m_row[2].w = 0.f;\n"
-"	for(int i=0; i<3; i++)\n"
-"	{\n"
-"//	a.m_row[i].w = 0.f;\n"
-"		ans.m_row[i].x = b3Dot3F4(a.m_row[i],transB.m_row[0]);\n"
-"		ans.m_row[i].y = b3Dot3F4(a.m_row[i],transB.m_row[1]);\n"
-"		ans.m_row[i].z = b3Dot3F4(a.m_row[i],transB.m_row[2]);\n"
-"		ans.m_row[i].w = 0.f;\n"
-"	}\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b)\n"
-"{\n"
-"	b3Float4 ans;\n"
-"	ans.x = b3Dot3F4( a.m_row[0], b );\n"
-"	ans.y = b3Dot3F4( a.m_row[1], b );\n"
-"	ans.z = b3Dot3F4( a.m_row[2], b );\n"
-"	ans.w = 0.f;\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b)\n"
-"{\n"
-"	b3Float4 colx = b3MakeFloat4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
-"	b3Float4 coly = b3MakeFloat4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
-"	b3Float4 colz = b3MakeFloat4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
-"	b3Float4 ans;\n"
-"	ans.x = b3Dot3F4( a, colx );\n"
-"	ans.y = b3Dot3F4( a, coly );\n"
-"	ans.z = b3Dot3F4( a, colz );\n"
-"	return ans;\n"
-"}\n"
-"#endif\n"
-"#endif //B3_MAT3x3_H\n"
-"typedef struct b3Aabb b3Aabb_t;\n"
-"struct b3Aabb\n"
-"{\n"
-"	union\n"
-"	{\n"
-"		float m_min[4];\n"
-"		b3Float4 m_minVec;\n"
-"		int m_minIndices[4];\n"
-"	};\n"
-"	union\n"
-"	{\n"
-"		float	m_max[4];\n"
-"		b3Float4 m_maxVec;\n"
-"		int m_signedMaxIndices[4];\n"
-"	};\n"
-"};\n"
-"inline void b3TransformAabb2(b3Float4ConstArg localAabbMin,b3Float4ConstArg localAabbMax, float margin,\n"
-"						b3Float4ConstArg pos,\n"
-"						b3QuatConstArg orn,\n"
-"						b3Float4* aabbMinOut,b3Float4* aabbMaxOut)\n"
-"{\n"
-"		b3Float4 localHalfExtents = 0.5f*(localAabbMax-localAabbMin);\n"
-"		localHalfExtents+=b3MakeFloat4(margin,margin,margin,0.f);\n"
-"		b3Float4 localCenter = 0.5f*(localAabbMax+localAabbMin);\n"
-"		b3Mat3x3 m;\n"
-"		m = b3QuatGetRotationMatrix(orn);\n"
-"		b3Mat3x3 abs_b = b3AbsoluteMat3x3(m);\n"
-"		b3Float4 center = b3TransformPoint(localCenter,pos,orn);\n"
-"		\n"
-"		b3Float4 extent = b3MakeFloat4(b3Dot3F4(localHalfExtents,b3GetRow(abs_b,0)),\n"
-"										 b3Dot3F4(localHalfExtents,b3GetRow(abs_b,1)),\n"
-"										 b3Dot3F4(localHalfExtents,b3GetRow(abs_b,2)),\n"
-"										 0.f);\n"
-"		*aabbMinOut = center-extent;\n"
-"		*aabbMaxOut = center+extent;\n"
-"}\n"
-"/// conservative test for overlap between two aabbs\n"
-"inline bool b3TestAabbAgainstAabb(b3Float4ConstArg aabbMin1,b3Float4ConstArg aabbMax1,\n"
-"								b3Float4ConstArg aabbMin2, b3Float4ConstArg aabbMax2)\n"
-"{\n"
-"	bool overlap = true;\n"
-"	overlap = (aabbMin1.x > aabbMax2.x || aabbMax1.x < aabbMin2.x) ? false : overlap;\n"
-"	overlap = (aabbMin1.z > aabbMax2.z || aabbMax1.z < aabbMin2.z) ? false : overlap;\n"
-"	overlap = (aabbMin1.y > aabbMax2.y || aabbMax1.y < aabbMin2.y) ? false : overlap;\n"
-"	return overlap;\n"
-"}\n"
-"#endif //B3_AABB_H\n"
-"/*\n"
-"Bullet Continuous Collision Detection and Physics Library\n"
-"Copyright (c) 2003-2013 Erwin Coumans  http://bulletphysics.org\n"
-"This software is provided 'as-is', without any express or implied warranty.\n"
-"In no event will the authors be held liable for any damages arising from the use of this software.\n"
-"Permission is granted to anyone to use this software for any purpose,\n"
-"including commercial applications, and to alter it and redistribute it freely,\n"
-"subject to the following restrictions:\n"
-"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
-"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
-"3. This notice may not be removed or altered from any source distribution.\n"
-"*/\n"
-"#ifndef B3_INT2_H\n"
-"#define B3_INT2_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#define b3UnsignedInt2 uint2\n"
-"#define b3Int2 int2\n"
-"#define b3MakeInt2 (int2)\n"
-"#endif //__cplusplus\n"
-"#endif\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_plane;\n"
-"	int m_indexOffset;\n"
-"	int m_numIndices;\n"
-"} btGpuFace;\n"
-"#define make_float4 (float4)\n"
-"__inline\n"
-"float4 cross3(float4 a, float4 b)\n"
-"{\n"
-"	return cross(a,b);\n"
-"	\n"
-"//	float4 a1 = make_float4(a.xyz,0.f);\n"
-"//	float4 b1 = make_float4(b.xyz,0.f);\n"
-"//	return cross(a1,b1);\n"
-"//float4 c = make_float4(a.y*b.z - a.z*b.y,a.z*b.x - a.x*b.z,a.x*b.y - a.y*b.x,0.f);\n"
-"	\n"
-"	//	float4 c = make_float4(a.y*b.z - a.z*b.y,1.f,a.x*b.y - a.y*b.x,0.f);\n"
-"	\n"
-"	//return c;\n"
-"}\n"
-"__inline\n"
-"float dot3F4(float4 a, float4 b)\n"
-"{\n"
-"	float4 a1 = make_float4(a.xyz,0.f);\n"
-"	float4 b1 = make_float4(b.xyz,0.f);\n"
-"	return dot(a1, b1);\n"
-"}\n"
-"__inline\n"
-"float4 fastNormalize4(float4 v)\n"
-"{\n"
-"	v = make_float4(v.xyz,0.f);\n"
-"	return fast_normalize(v);\n"
-"}\n"
-"///////////////////////////////////////\n"
-"//	Quaternion\n"
-"///////////////////////////////////////\n"
-"typedef float4 Quaternion;\n"
-"__inline\n"
-"Quaternion qtMul(Quaternion a, Quaternion b);\n"
-"__inline\n"
-"Quaternion qtNormalize(Quaternion in);\n"
-"__inline\n"
-"float4 qtRotate(Quaternion q, float4 vec);\n"
-"__inline\n"
-"Quaternion qtInvert(Quaternion q);\n"
-"__inline\n"
-"Quaternion qtMul(Quaternion a, Quaternion b)\n"
-"{\n"
-"	Quaternion ans;\n"
-"	ans = cross3( a, b );\n"
-"	ans += a.w*b+b.w*a;\n"
-"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
-"	ans.w = a.w*b.w - dot3F4(a, b);\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"Quaternion qtNormalize(Quaternion in)\n"
-"{\n"
-"	return fastNormalize4(in);\n"
-"//	in /= length( in );\n"
-"//	return in;\n"
-"}\n"
-"__inline\n"
-"float4 qtRotate(Quaternion q, float4 vec)\n"
-"{\n"
-"	Quaternion qInv = qtInvert( q );\n"
-"	float4 vcpy = vec;\n"
-"	vcpy.w = 0.f;\n"
-"	float4 out = qtMul(qtMul(q,vcpy),qInv);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"Quaternion qtInvert(Quaternion q)\n"
-"{\n"
-"	return (Quaternion)(-q.xyz, q.w);\n"
-"}\n"
-"__inline\n"
-"float4 qtInvRotate(const Quaternion q, float4 vec)\n"
-"{\n"
-"	return qtRotate( qtInvert( q ), vec );\n"
-"}\n"
-"__inline\n"
-"float4 transform(const float4* p, const float4* translation, const Quaternion* orientation)\n"
-"{\n"
-"	return qtRotate( *orientation, *p ) + (*translation);\n"
-"}\n"
-"__inline\n"
-"float4 normalize3(const float4 a)\n"
-"{\n"
-"	float4 n = make_float4(a.x, a.y, a.z, 0.f);\n"
-"	return fastNormalize4( n );\n"
-"}\n"
-"inline void projectLocal(const ConvexPolyhedronCL* hull,  const float4 pos, const float4 orn, \n"
-"const float4* dir, const float4* vertices, float* min, float* max)\n"
-"{\n"
-"	min[0] = FLT_MAX;\n"
-"	max[0] = -FLT_MAX;\n"
-"	int numVerts = hull->m_numVertices;\n"
-"	const float4 localDir = qtInvRotate(orn,*dir);\n"
-"	float offset = dot(pos,*dir);\n"
-"	for(int i=0;i<numVerts;i++)\n"
-"	{\n"
-"		float dp = dot(vertices[hull->m_vertexOffset+i],localDir);\n"
-"		if(dp < min[0])	\n"
-"			min[0] = dp;\n"
-"		if(dp > max[0])	\n"
-"			max[0] = dp;\n"
-"	}\n"
-"	if(min[0]>max[0])\n"
-"	{\n"
-"		float tmp = min[0];\n"
-"		min[0] = max[0];\n"
-"		max[0] = tmp;\n"
-"	}\n"
-"	min[0] += offset;\n"
-"	max[0] += offset;\n"
-"}\n"
-"inline void project(__global const ConvexPolyhedronCL* hull,  const float4 pos, const float4 orn, \n"
-"const float4* dir, __global const float4* vertices, float* min, float* max)\n"
-"{\n"
-"	min[0] = FLT_MAX;\n"
-"	max[0] = -FLT_MAX;\n"
-"	int numVerts = hull->m_numVertices;\n"
-"	const float4 localDir = qtInvRotate(orn,*dir);\n"
-"	float offset = dot(pos,*dir);\n"
-"	for(int i=0;i<numVerts;i++)\n"
-"	{\n"
-"		float dp = dot(vertices[hull->m_vertexOffset+i],localDir);\n"
-"		if(dp < min[0])	\n"
-"			min[0] = dp;\n"
-"		if(dp > max[0])	\n"
-"			max[0] = dp;\n"
-"	}\n"
-"	if(min[0]>max[0])\n"
-"	{\n"
-"		float tmp = min[0];\n"
-"		min[0] = max[0];\n"
-"		max[0] = tmp;\n"
-"	}\n"
-"	min[0] += offset;\n"
-"	max[0] += offset;\n"
-"}\n"
-"inline bool TestSepAxisLocalA(const ConvexPolyhedronCL* hullA, __global const ConvexPolyhedronCL* hullB, \n"
-"	const float4 posA,const float4 ornA,\n"
-"	const float4 posB,const float4 ornB,\n"
-"	float4* sep_axis, const float4* verticesA, __global const float4* verticesB,float* depth)\n"
-"{\n"
-"	float Min0,Max0;\n"
-"	float Min1,Max1;\n"
-"	projectLocal(hullA,posA,ornA,sep_axis,verticesA, &Min0, &Max0);\n"
-"	project(hullB,posB,ornB, sep_axis,verticesB, &Min1, &Max1);\n"
-"	if(Max0<Min1 || Max1<Min0)\n"
-"		return false;\n"
-"	float d0 = Max0 - Min1;\n"
-"	float d1 = Max1 - Min0;\n"
-"	*depth = d0<d1 ? d0:d1;\n"
-"	return true;\n"
-"}\n"
-"inline bool IsAlmostZero(const float4 v)\n"
-"{\n"
-"	if(fabs(v.x)>1e-6f || fabs(v.y)>1e-6f || fabs(v.z)>1e-6f)\n"
-"		return false;\n"
-"	return true;\n"
-"}\n"
-"bool findSeparatingAxisLocalA(	const ConvexPolyhedronCL* hullA, __global const ConvexPolyhedronCL* hullB, \n"
-"	const float4 posA1,\n"
-"	const float4 ornA,\n"
-"	const float4 posB1,\n"
-"	const float4 ornB,\n"
-"	const float4 DeltaC2,\n"
-"	\n"
-"	const float4* verticesA, \n"
-"	const float4* uniqueEdgesA, \n"
-"	const btGpuFace* facesA,\n"
-"	const int*  indicesA,\n"
-"	__global const float4* verticesB, \n"
-"	__global const float4* uniqueEdgesB, \n"
-"	__global const btGpuFace* facesB,\n"
-"	__global const int*  indicesB,\n"
-"	float4* sep,\n"
-"	float* dmin)\n"
-"{\n"
-"	\n"
-"	float4 posA = posA1;\n"
-"	posA.w = 0.f;\n"
-"	float4 posB = posB1;\n"
-"	posB.w = 0.f;\n"
-"	int curPlaneTests=0;\n"
-"	{\n"
-"		int numFacesA = hullA->m_numFaces;\n"
-"		// Test normals from hullA\n"
-"		for(int i=0;i<numFacesA;i++)\n"
-"		{\n"
-"			const float4 normal = facesA[hullA->m_faceOffset+i].m_plane;\n"
-"			float4 faceANormalWS = qtRotate(ornA,normal);\n"
-"			if (dot3F4(DeltaC2,faceANormalWS)<0)\n"
-"				faceANormalWS*=-1.f;\n"
-"			curPlaneTests++;\n"
-"			float d;\n"
-"			if(!TestSepAxisLocalA( hullA, hullB, posA,ornA,posB,ornB,&faceANormalWS, verticesA, verticesB,&d))\n"
-"				return false;\n"
-"			if(d<*dmin)\n"
-"			{\n"
-"				*dmin = d;\n"
-"				*sep = faceANormalWS;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	if((dot3F4(-DeltaC2,*sep))>0.0f)\n"
-"	{\n"
-"		*sep = -(*sep);\n"
-"	}\n"
-"	return true;\n"
-"}\n"
-"bool findSeparatingAxisLocalB(	__global const ConvexPolyhedronCL* hullA,  const ConvexPolyhedronCL* hullB, \n"
-"	const float4 posA1,\n"
-"	const float4 ornA,\n"
-"	const float4 posB1,\n"
-"	const float4 ornB,\n"
-"	const float4 DeltaC2,\n"
-"	__global const float4* verticesA, \n"
-"	__global const float4* uniqueEdgesA, \n"
-"	__global const btGpuFace* facesA,\n"
-"	__global const int*  indicesA,\n"
-"	const float4* verticesB,\n"
-"	const float4* uniqueEdgesB, \n"
-"	const btGpuFace* facesB,\n"
-"	const int*  indicesB,\n"
-"	float4* sep,\n"
-"	float* dmin)\n"
-"{\n"
-"	float4 posA = posA1;\n"
-"	posA.w = 0.f;\n"
-"	float4 posB = posB1;\n"
-"	posB.w = 0.f;\n"
-"	int curPlaneTests=0;\n"
-"	{\n"
-"		int numFacesA = hullA->m_numFaces;\n"
-"		// Test normals from hullA\n"
-"		for(int i=0;i<numFacesA;i++)\n"
-"		{\n"
-"			const float4 normal = facesA[hullA->m_faceOffset+i].m_plane;\n"
-"			float4 faceANormalWS = qtRotate(ornA,normal);\n"
-"			if (dot3F4(DeltaC2,faceANormalWS)<0)\n"
-"				faceANormalWS *= -1.f;\n"
-"			curPlaneTests++;\n"
-"			float d;\n"
-"			if(!TestSepAxisLocalA( hullB, hullA, posB,ornB,posA,ornA, &faceANormalWS, verticesB,verticesA, &d))\n"
-"				return false;\n"
-"			if(d<*dmin)\n"
-"			{\n"
-"				*dmin = d;\n"
-"				*sep = faceANormalWS;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	if((dot3F4(-DeltaC2,*sep))>0.0f)\n"
-"	{\n"
-"		*sep = -(*sep);\n"
-"	}\n"
-"	return true;\n"
-"}\n"
-"bool findSeparatingAxisEdgeEdgeLocalA(	const ConvexPolyhedronCL* hullA, __global const ConvexPolyhedronCL* hullB, \n"
-"	const float4 posA1,\n"
-"	const float4 ornA,\n"
-"	const float4 posB1,\n"
-"	const float4 ornB,\n"
-"	const float4 DeltaC2,\n"
-"	const float4* verticesA, \n"
-"	const float4* uniqueEdgesA, \n"
-"	const btGpuFace* facesA,\n"
-"	const int*  indicesA,\n"
-"	__global const float4* verticesB, \n"
-"	__global const float4* uniqueEdgesB, \n"
-"	__global const btGpuFace* facesB,\n"
-"	__global const int*  indicesB,\n"
-"		float4* sep,\n"
-"	float* dmin)\n"
-"{\n"
-"	float4 posA = posA1;\n"
-"	posA.w = 0.f;\n"
-"	float4 posB = posB1;\n"
-"	posB.w = 0.f;\n"
-"	int curPlaneTests=0;\n"
-"	int curEdgeEdge = 0;\n"
-"	// Test edges\n"
-"	for(int e0=0;e0<hullA->m_numUniqueEdges;e0++)\n"
-"	{\n"
-"		const float4 edge0 = uniqueEdgesA[hullA->m_uniqueEdgesOffset+e0];\n"
-"		float4 edge0World = qtRotate(ornA,edge0);\n"
-"		for(int e1=0;e1<hullB->m_numUniqueEdges;e1++)\n"
-"		{\n"
-"			const float4 edge1 = uniqueEdgesB[hullB->m_uniqueEdgesOffset+e1];\n"
-"			float4 edge1World = qtRotate(ornB,edge1);\n"
-"			float4 crossje = cross3(edge0World,edge1World);\n"
-"			curEdgeEdge++;\n"
-"			if(!IsAlmostZero(crossje))\n"
-"			{\n"
-"				crossje = normalize3(crossje);\n"
-"				if (dot3F4(DeltaC2,crossje)<0)\n"
-"					crossje *= -1.f;\n"
-"				float dist;\n"
-"				bool result = true;\n"
-"				{\n"
-"					float Min0,Max0;\n"
-"					float Min1,Max1;\n"
-"					projectLocal(hullA,posA,ornA,&crossje,verticesA, &Min0, &Max0);\n"
-"					project(hullB,posB,ornB,&crossje,verticesB, &Min1, &Max1);\n"
-"				\n"
-"					if(Max0<Min1 || Max1<Min0)\n"
-"						result = false;\n"
-"				\n"
-"					float d0 = Max0 - Min1;\n"
-"					float d1 = Max1 - Min0;\n"
-"					dist = d0<d1 ? d0:d1;\n"
-"					result = true;\n"
-"				}\n"
-"				\n"
-"				if(dist<*dmin)\n"
-"				{\n"
-"					*dmin = dist;\n"
-"					*sep = crossje;\n"
-"				}\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	\n"
-"	if((dot3F4(-DeltaC2,*sep))>0.0f)\n"
-"	{\n"
-"		*sep = -(*sep);\n"
-"	}\n"
-"	return true;\n"
-"}\n"
-"inline bool TestSepAxis(__global const ConvexPolyhedronCL* hullA, __global const ConvexPolyhedronCL* hullB, \n"
-"	const float4 posA,const float4 ornA,\n"
-"	const float4 posB,const float4 ornB,\n"
-"	float4* sep_axis, __global const float4* vertices,float* depth)\n"
-"{\n"
-"	float Min0,Max0;\n"
-"	float Min1,Max1;\n"
-"	project(hullA,posA,ornA,sep_axis,vertices, &Min0, &Max0);\n"
-"	project(hullB,posB,ornB, sep_axis,vertices, &Min1, &Max1);\n"
-"	if(Max0<Min1 || Max1<Min0)\n"
-"		return false;\n"
-"	float d0 = Max0 - Min1;\n"
-"	float d1 = Max1 - Min0;\n"
-"	*depth = d0<d1 ? d0:d1;\n"
-"	return true;\n"
-"}\n"
-"bool findSeparatingAxis(	__global const ConvexPolyhedronCL* hullA, __global const ConvexPolyhedronCL* hullB, \n"
-"	const float4 posA1,\n"
-"	const float4 ornA,\n"
-"	const float4 posB1,\n"
-"	const float4 ornB,\n"
-"	const float4 DeltaC2,\n"
-"	__global const float4* vertices, \n"
-"	__global const float4* uniqueEdges, \n"
-"	__global const btGpuFace* faces,\n"
-"	__global const int*  indices,\n"
-"	float4* sep,\n"
-"	float* dmin)\n"
-"{\n"
-"	\n"
-"	float4 posA = posA1;\n"
-"	posA.w = 0.f;\n"
-"	float4 posB = posB1;\n"
-"	posB.w = 0.f;\n"
-"	\n"
-"	int curPlaneTests=0;\n"
-"	{\n"
-"		int numFacesA = hullA->m_numFaces;\n"
-"		// Test normals from hullA\n"
-"		for(int i=0;i<numFacesA;i++)\n"
-"		{\n"
-"			const float4 normal = faces[hullA->m_faceOffset+i].m_plane;\n"
-"			float4 faceANormalWS = qtRotate(ornA,normal);\n"
-"	\n"
-"			if (dot3F4(DeltaC2,faceANormalWS)<0)\n"
-"				faceANormalWS*=-1.f;\n"
-"				\n"
-"			curPlaneTests++;\n"
-"	\n"
-"			float d;\n"
-"			if(!TestSepAxis( hullA, hullB, posA,ornA,posB,ornB,&faceANormalWS, vertices,&d))\n"
-"				return false;\n"
-"	\n"
-"			if(d<*dmin)\n"
-"			{\n"
-"				*dmin = d;\n"
-"				*sep = faceANormalWS;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"		if((dot3F4(-DeltaC2,*sep))>0.0f)\n"
-"		{\n"
-"			*sep = -(*sep);\n"
-"		}\n"
-"	\n"
-"	return true;\n"
-"}\n"
-"bool findSeparatingAxisUnitSphere(	__global const ConvexPolyhedronCL* hullA, __global const ConvexPolyhedronCL* hullB, \n"
-"	const float4 posA1,\n"
-"	const float4 ornA,\n"
-"	const float4 posB1,\n"
-"	const float4 ornB,\n"
-"	const float4 DeltaC2,\n"
-"	__global const float4* vertices,\n"
-"	__global const float4* unitSphereDirections,\n"
-"	int numUnitSphereDirections,\n"
-"	float4* sep,\n"
-"	float* dmin)\n"
-"{\n"
-"	\n"
-"	float4 posA = posA1;\n"
-"	posA.w = 0.f;\n"
-"	float4 posB = posB1;\n"
-"	posB.w = 0.f;\n"
-"	int curPlaneTests=0;\n"
-"	int curEdgeEdge = 0;\n"
-"	// Test unit sphere directions\n"
-"	for (int i=0;i<numUnitSphereDirections;i++)\n"
-"	{\n"
-"		float4 crossje;\n"
-"		crossje = unitSphereDirections[i];	\n"
-"		if (dot3F4(DeltaC2,crossje)>0)\n"
-"			crossje *= -1.f;\n"
-"		{\n"
-"			float dist;\n"
-"			bool result = true;\n"
-"			float Min0,Max0;\n"
-"			float Min1,Max1;\n"
-"			project(hullA,posA,ornA,&crossje,vertices, &Min0, &Max0);\n"
-"			project(hullB,posB,ornB,&crossje,vertices, &Min1, &Max1);\n"
-"		\n"
-"			if(Max0<Min1 || Max1<Min0)\n"
-"				return false;\n"
-"		\n"
-"			float d0 = Max0 - Min1;\n"
-"			float d1 = Max1 - Min0;\n"
-"			dist = d0<d1 ? d0:d1;\n"
-"			result = true;\n"
-"	\n"
-"			if(dist<*dmin)\n"
-"			{\n"
-"				*dmin = dist;\n"
-"				*sep = crossje;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	\n"
-"	if((dot3F4(-DeltaC2,*sep))>0.0f)\n"
-"	{\n"
-"		*sep = -(*sep);\n"
-"	}\n"
-"	return true;\n"
-"}\n"
-"bool findSeparatingAxisEdgeEdge(	__global const ConvexPolyhedronCL* hullA, __global const ConvexPolyhedronCL* hullB, \n"
-"	const float4 posA1,\n"
-"	const float4 ornA,\n"
-"	const float4 posB1,\n"
-"	const float4 ornB,\n"
-"	const float4 DeltaC2,\n"
-"	__global const float4* vertices, \n"
-"	__global const float4* uniqueEdges, \n"
-"	__global const btGpuFace* faces,\n"
-"	__global const int*  indices,\n"
-"	float4* sep,\n"
-"	float* dmin)\n"
-"{\n"
-"	\n"
-"	float4 posA = posA1;\n"
-"	posA.w = 0.f;\n"
-"	float4 posB = posB1;\n"
-"	posB.w = 0.f;\n"
-"	int curPlaneTests=0;\n"
-"	int curEdgeEdge = 0;\n"
-"	// Test edges\n"
-"	for(int e0=0;e0<hullA->m_numUniqueEdges;e0++)\n"
-"	{\n"
-"		const float4 edge0 = uniqueEdges[hullA->m_uniqueEdgesOffset+e0];\n"
-"		float4 edge0World = qtRotate(ornA,edge0);\n"
-"		for(int e1=0;e1<hullB->m_numUniqueEdges;e1++)\n"
-"		{\n"
-"			const float4 edge1 = uniqueEdges[hullB->m_uniqueEdgesOffset+e1];\n"
-"			float4 edge1World = qtRotate(ornB,edge1);\n"
-"			float4 crossje = cross3(edge0World,edge1World);\n"
-"			curEdgeEdge++;\n"
-"			if(!IsAlmostZero(crossje))\n"
-"			{\n"
-"				crossje = normalize3(crossje);\n"
-"				if (dot3F4(DeltaC2,crossje)<0)\n"
-"					crossje*=-1.f;\n"
-"					\n"
-"				float dist;\n"
-"				bool result = true;\n"
-"				{\n"
-"					float Min0,Max0;\n"
-"					float Min1,Max1;\n"
-"					project(hullA,posA,ornA,&crossje,vertices, &Min0, &Max0);\n"
-"					project(hullB,posB,ornB,&crossje,vertices, &Min1, &Max1);\n"
-"				\n"
-"					if(Max0<Min1 || Max1<Min0)\n"
-"						return false;\n"
-"				\n"
-"					float d0 = Max0 - Min1;\n"
-"					float d1 = Max1 - Min0;\n"
-"					dist = d0<d1 ? d0:d1;\n"
-"					result = true;\n"
-"				}\n"
-"				\n"
-"				if(dist<*dmin)\n"
-"				{\n"
-"					*dmin = dist;\n"
-"					*sep = crossje;\n"
-"				}\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	\n"
-"	if((dot3F4(-DeltaC2,*sep))>0.0f)\n"
-"	{\n"
-"		*sep = -(*sep);\n"
-"	}\n"
-"	return true;\n"
-"}\n"
-"// work-in-progress\n"
-"__kernel void   processCompoundPairsKernel( __global const int4* gpuCompoundPairs,\n"
-"																					__global const BodyData* rigidBodies, \n"
-"																					__global const btCollidableGpu* collidables,\n"
-"																					__global const ConvexPolyhedronCL* convexShapes, \n"
-"																					__global const float4* vertices,\n"
-"																					__global const float4* uniqueEdges,\n"
-"																					__global const btGpuFace* faces,\n"
-"																					__global const int* indices,\n"
-"																					__global btAabbCL* aabbs,\n"
-"																					__global const btGpuChildShape* gpuChildShapes,\n"
-"																					__global volatile float4* gpuCompoundSepNormalsOut,\n"
-"																					__global volatile int* gpuHasCompoundSepNormalsOut,\n"
-"																					int numCompoundPairs\n"
-"																					)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	if (i<numCompoundPairs)\n"
-"	{\n"
-"		int bodyIndexA = gpuCompoundPairs[i].x;\n"
-"		int bodyIndexB = gpuCompoundPairs[i].y;\n"
-"		int childShapeIndexA = gpuCompoundPairs[i].z;\n"
-"		int childShapeIndexB = gpuCompoundPairs[i].w;\n"
-"		\n"
-"		int collidableIndexA = -1;\n"
-"		int collidableIndexB = -1;\n"
-"		\n"
-"		float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
-"		float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
-"		\n"
-"		float4 ornB = rigidBodies[bodyIndexB].m_quat;\n"
-"		float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
-"							\n"
-"		if (childShapeIndexA >= 0)\n"
-"		{\n"
-"			collidableIndexA = gpuChildShapes[childShapeIndexA].m_shapeIndex;\n"
-"			float4 childPosA = gpuChildShapes[childShapeIndexA].m_childPosition;\n"
-"			float4 childOrnA = gpuChildShapes[childShapeIndexA].m_childOrientation;\n"
-"			float4 newPosA = qtRotate(ornA,childPosA)+posA;\n"
-"			float4 newOrnA = qtMul(ornA,childOrnA);\n"
-"			posA = newPosA;\n"
-"			ornA = newOrnA;\n"
-"		} else\n"
-"		{\n"
-"			collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
-"		}\n"
-"		\n"
-"		if (childShapeIndexB>=0)\n"
-"		{\n"
-"			collidableIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;\n"
-"			float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;\n"
-"			float4 childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;\n"
-"			float4 newPosB = transform(&childPosB,&posB,&ornB);\n"
-"			float4 newOrnB = qtMul(ornB,childOrnB);\n"
-"			posB = newPosB;\n"
-"			ornB = newOrnB;\n"
-"		} else\n"
-"		{\n"
-"			collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;	\n"
-"		}\n"
-"	\n"
-"		gpuHasCompoundSepNormalsOut[i] = 0;\n"
-"	\n"
-"		int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
-"		int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
-"	\n"
-"		int shapeTypeA = collidables[collidableIndexA].m_shapeType;\n"
-"		int shapeTypeB = collidables[collidableIndexB].m_shapeType;\n"
-"	\n"
-"		if ((shapeTypeA != SHAPE_CONVEX_HULL) || (shapeTypeB != SHAPE_CONVEX_HULL))\n"
-"		{\n"
-"			return;\n"
-"		}\n"
-"		int hasSeparatingAxis = 5;\n"
-"							\n"
-"		int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
-"		float dmin = FLT_MAX;\n"
-"		posA.w = 0.f;\n"
-"		posB.w = 0.f;\n"
-"		float4 c0local = convexShapes[shapeIndexA].m_localCenter;\n"
-"		float4 c0 = transform(&c0local, &posA, &ornA);\n"
-"		float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
-"		float4 c1 = transform(&c1local,&posB,&ornB);\n"
-"		const float4 DeltaC2 = c0 - c1;\n"
-"		float4 sepNormal = make_float4(1,0,0,0);\n"
-"		bool sepA = findSeparatingAxis(	&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],posA,ornA,posB,ornB,DeltaC2,vertices,uniqueEdges,faces,indices,&sepNormal,&dmin);\n"
-"		hasSeparatingAxis = 4;\n"
-"		if (!sepA)\n"
-"		{\n"
-"			hasSeparatingAxis = 0;\n"
-"		} else\n"
-"		{\n"
-"			bool sepB = findSeparatingAxis(	&convexShapes[shapeIndexB],&convexShapes[shapeIndexA],posB,ornB,posA,ornA,DeltaC2,vertices,uniqueEdges,faces,indices,&sepNormal,&dmin);\n"
-"			if (!sepB)\n"
-"			{\n"
-"				hasSeparatingAxis = 0;\n"
-"			} else//(!sepB)\n"
-"			{\n"
-"				bool sepEE = findSeparatingAxisEdgeEdge(	&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],posA,ornA,posB,ornB,DeltaC2,vertices,uniqueEdges,faces,indices,&sepNormal,&dmin);\n"
-"				if (sepEE)\n"
-"				{\n"
-"						gpuCompoundSepNormalsOut[i] = sepNormal;//fastNormalize4(sepNormal);\n"
-"						gpuHasCompoundSepNormalsOut[i] = 1;\n"
-"				}//sepEE\n"
-"			}//(!sepB)\n"
-"		}//(!sepA)\n"
-"		\n"
-"		\n"
-"	}\n"
-"		\n"
-"}\n"
-"inline b3Float4 MyUnQuantize(const unsigned short* vecIn, b3Float4 quantization, b3Float4 bvhAabbMin)\n"
-"{\n"
-"		b3Float4 vecOut;\n"
-"		vecOut = b3MakeFloat4(\n"
-"			(float)(vecIn[0]) / (quantization.x),\n"
-"			(float)(vecIn[1]) / (quantization.y),\n"
-"			(float)(vecIn[2]) / (quantization.z),\n"
-"			0.f);\n"
-"		vecOut += bvhAabbMin;\n"
-"		return vecOut;\n"
-"}\n"
-"inline b3Float4 MyUnQuantizeGlobal(__global const unsigned short* vecIn, b3Float4 quantization, b3Float4 bvhAabbMin)\n"
-"{\n"
-"		b3Float4 vecOut;\n"
-"		vecOut = b3MakeFloat4(\n"
-"			(float)(vecIn[0]) / (quantization.x),\n"
-"			(float)(vecIn[1]) / (quantization.y),\n"
-"			(float)(vecIn[2]) / (quantization.z),\n"
-"			0.f);\n"
-"		vecOut += bvhAabbMin;\n"
-"		return vecOut;\n"
-"}\n"
-"// work-in-progress\n"
-"__kernel void   findCompoundPairsKernel( __global const int4* pairs, \n"
-"	__global const BodyData* rigidBodies, \n"
-"	__global const btCollidableGpu* collidables,\n"
-"	__global const ConvexPolyhedronCL* convexShapes, \n"
-"	__global const float4* vertices,\n"
-"	__global const float4* uniqueEdges,\n"
-"	__global const btGpuFace* faces,\n"
-"	__global const int* indices,\n"
-"	__global b3Aabb_t* aabbLocalSpace,\n"
-"	__global const btGpuChildShape* gpuChildShapes,\n"
-"	__global volatile int4* gpuCompoundPairsOut,\n"
-"	__global volatile int* numCompoundPairsOut,\n"
-"	__global const b3BvhSubtreeInfo* subtrees,\n"
-"	__global const b3QuantizedBvhNode* quantizedNodes,\n"
-"	__global const b3BvhInfo* bvhInfos,\n"
-"	int numPairs,\n"
-"	int maxNumCompoundPairsCapacity\n"
-"	)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	if (i<numPairs)\n"
-"	{\n"
-"		int bodyIndexA = pairs[i].x;\n"
-"		int bodyIndexB = pairs[i].y;\n"
-"		int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
-"		int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
-"		int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
-"		int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
-"		//once the broadphase avoids static-static pairs, we can remove this test\n"
-"		if ((rigidBodies[bodyIndexA].m_invMass==0) &&(rigidBodies[bodyIndexB].m_invMass==0))\n"
-"		{\n"
-"			return;\n"
-"		}\n"
-"		if ((collidables[collidableIndexA].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS) &&(collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS))\n"
-"		{\n"
-"			int bvhA = collidables[collidableIndexA].m_compoundBvhIndex;\n"
-"			int bvhB = collidables[collidableIndexB].m_compoundBvhIndex;\n"
-"			int numSubTreesA = bvhInfos[bvhA].m_numSubTrees;\n"
-"			int subTreesOffsetA = bvhInfos[bvhA].m_subTreeOffset;\n"
-"			int subTreesOffsetB = bvhInfos[bvhB].m_subTreeOffset;\n"
-"			int numSubTreesB = bvhInfos[bvhB].m_numSubTrees;\n"
-"			\n"
-"			float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
-"			b3Quat ornA = rigidBodies[bodyIndexA].m_quat;\n"
-"			b3Quat ornB = rigidBodies[bodyIndexB].m_quat;\n"
-"			float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
-"			\n"
-"			for (int p=0;p<numSubTreesA;p++)\n"
-"			{\n"
-"				b3BvhSubtreeInfo subtreeA = subtrees[subTreesOffsetA+p];\n"
-"				//bvhInfos[bvhA].m_quantization\n"
-"				b3Float4 treeAminLocal = MyUnQuantize(subtreeA.m_quantizedAabbMin,bvhInfos[bvhA].m_quantization,bvhInfos[bvhA].m_aabbMin);\n"
-"				b3Float4 treeAmaxLocal = MyUnQuantize(subtreeA.m_quantizedAabbMax,bvhInfos[bvhA].m_quantization,bvhInfos[bvhA].m_aabbMin);\n"
-"				b3Float4 aabbAMinOut,aabbAMaxOut;\n"
-"				float margin=0.f;\n"
-"				b3TransformAabb2(treeAminLocal,treeAmaxLocal, margin,posA,ornA,&aabbAMinOut,&aabbAMaxOut);\n"
-"				\n"
-"				for (int q=0;q<numSubTreesB;q++)\n"
-"				{\n"
-"					b3BvhSubtreeInfo subtreeB = subtrees[subTreesOffsetB+q];\n"
-"					b3Float4 treeBminLocal = MyUnQuantize(subtreeB.m_quantizedAabbMin,bvhInfos[bvhB].m_quantization,bvhInfos[bvhB].m_aabbMin);\n"
-"					b3Float4 treeBmaxLocal = MyUnQuantize(subtreeB.m_quantizedAabbMax,bvhInfos[bvhB].m_quantization,bvhInfos[bvhB].m_aabbMin);\n"
-"					b3Float4 aabbBMinOut,aabbBMaxOut;\n"
-"					float margin=0.f;\n"
-"					b3TransformAabb2(treeBminLocal,treeBmaxLocal, margin,posB,ornB,&aabbBMinOut,&aabbBMaxOut);\n"
-"					\n"
-"					\n"
-"					bool aabbOverlap = b3TestAabbAgainstAabb(aabbAMinOut,aabbAMaxOut,aabbBMinOut,aabbBMaxOut);\n"
-"					if (aabbOverlap)\n"
-"					{\n"
-"						\n"
-"						int startNodeIndexA = subtreeA.m_rootNodeIndex+bvhInfos[bvhA].m_nodeOffset;\n"
-"						int endNodeIndexA = startNodeIndexA+subtreeA.m_subtreeSize;\n"
-"						int startNodeIndexB = subtreeB.m_rootNodeIndex+bvhInfos[bvhB].m_nodeOffset;\n"
-"						int endNodeIndexB = startNodeIndexB+subtreeB.m_subtreeSize;\n"
-"						b3Int2 nodeStack[B3_MAX_STACK_DEPTH];\n"
-"						b3Int2 node0;\n"
-"						node0.x = startNodeIndexA;\n"
-"						node0.y = startNodeIndexB;\n"
-"						int maxStackDepth = B3_MAX_STACK_DEPTH;\n"
-"						int depth=0;\n"
-"						nodeStack[depth++]=node0;\n"
-"						do\n"
-"						{\n"
-"							b3Int2 node = nodeStack[--depth];\n"
-"							b3Float4 aMinLocal = MyUnQuantizeGlobal(quantizedNodes[node.x].m_quantizedAabbMin,bvhInfos[bvhA].m_quantization,bvhInfos[bvhA].m_aabbMin);\n"
-"							b3Float4 aMaxLocal = MyUnQuantizeGlobal(quantizedNodes[node.x].m_quantizedAabbMax,bvhInfos[bvhA].m_quantization,bvhInfos[bvhA].m_aabbMin);\n"
-"							b3Float4 bMinLocal = MyUnQuantizeGlobal(quantizedNodes[node.y].m_quantizedAabbMin,bvhInfos[bvhB].m_quantization,bvhInfos[bvhB].m_aabbMin);\n"
-"							b3Float4 bMaxLocal = MyUnQuantizeGlobal(quantizedNodes[node.y].m_quantizedAabbMax,bvhInfos[bvhB].m_quantization,bvhInfos[bvhB].m_aabbMin);\n"
-"							float margin=0.f;\n"
-"							b3Float4 aabbAMinOut,aabbAMaxOut;\n"
-"							b3TransformAabb2(aMinLocal,aMaxLocal, margin,posA,ornA,&aabbAMinOut,&aabbAMaxOut);\n"
-"							b3Float4 aabbBMinOut,aabbBMaxOut;\n"
-"							b3TransformAabb2(bMinLocal,bMaxLocal, margin,posB,ornB,&aabbBMinOut,&aabbBMaxOut);\n"
-"							\n"
-"							bool nodeOverlap = b3TestAabbAgainstAabb(aabbAMinOut,aabbAMaxOut,aabbBMinOut,aabbBMaxOut);\n"
-"							if (nodeOverlap)\n"
-"							{\n"
-"								bool isLeafA = isLeafNodeGlobal(&quantizedNodes[node.x]);\n"
-"								bool isLeafB = isLeafNodeGlobal(&quantizedNodes[node.y]);\n"
-"								bool isInternalA = !isLeafA;\n"
-"								bool isInternalB = !isLeafB;\n"
-"								//fail, even though it might hit two leaf nodes\n"
-"								if (depth+4>maxStackDepth && !(isLeafA && isLeafB))\n"
-"								{\n"
-"									//printf(\"Error: traversal exceeded maxStackDepth\");\n"
-"									continue;\n"
-"								}\n"
-"								if(isInternalA)\n"
-"								{\n"
-"									int nodeAleftChild = node.x+1;\n"
-"									bool isNodeALeftChildLeaf = isLeafNodeGlobal(&quantizedNodes[node.x+1]);\n"
-"									int nodeArightChild = isNodeALeftChildLeaf? node.x+2 : node.x+1 + getEscapeIndexGlobal(&quantizedNodes[node.x+1]);\n"
-"									if(isInternalB)\n"
-"									{					\n"
-"										int nodeBleftChild = node.y+1;\n"
-"										bool isNodeBLeftChildLeaf = isLeafNodeGlobal(&quantizedNodes[node.y+1]);\n"
-"										int nodeBrightChild = isNodeBLeftChildLeaf? node.y+2 : node.y+1 + getEscapeIndexGlobal(&quantizedNodes[node.y+1]);\n"
-"										nodeStack[depth++] = b3MakeInt2(nodeAleftChild, nodeBleftChild);\n"
-"										nodeStack[depth++] = b3MakeInt2(nodeArightChild, nodeBleftChild);\n"
-"										nodeStack[depth++] = b3MakeInt2(nodeAleftChild, nodeBrightChild);\n"
-"										nodeStack[depth++] = b3MakeInt2(nodeArightChild, nodeBrightChild);\n"
-"									}\n"
-"									else\n"
-"									{\n"
-"										nodeStack[depth++] = b3MakeInt2(nodeAleftChild,node.y);\n"
-"										nodeStack[depth++] = b3MakeInt2(nodeArightChild,node.y);\n"
-"									}\n"
-"								}\n"
-"								else\n"
-"								{\n"
-"									if(isInternalB)\n"
-"									{\n"
-"										int nodeBleftChild = node.y+1;\n"
-"										bool isNodeBLeftChildLeaf = isLeafNodeGlobal(&quantizedNodes[node.y+1]);\n"
-"										int nodeBrightChild = isNodeBLeftChildLeaf? node.y+2 : node.y+1 + getEscapeIndexGlobal(&quantizedNodes[node.y+1]);\n"
-"										nodeStack[depth++] = b3MakeInt2(node.x,nodeBleftChild);\n"
-"										nodeStack[depth++] = b3MakeInt2(node.x,nodeBrightChild);\n"
-"									}\n"
-"									else\n"
-"									{\n"
-"										int compoundPairIdx = atomic_inc(numCompoundPairsOut);\n"
-"										if (compoundPairIdx<maxNumCompoundPairsCapacity)\n"
-"										{\n"
-"											int childShapeIndexA = getTriangleIndexGlobal(&quantizedNodes[node.x]);\n"
-"											int childShapeIndexB = getTriangleIndexGlobal(&quantizedNodes[node.y]);\n"
-"											gpuCompoundPairsOut[compoundPairIdx]  = (int4)(bodyIndexA,bodyIndexB,childShapeIndexA,childShapeIndexB);\n"
-"										}\n"
-"									}\n"
-"								}\n"
-"							}\n"
-"						} while (depth);\n"
-"					}\n"
-"				}\n"
-"			}\n"
-"			\n"
-"			return;\n"
-"		}\n"
-"		if ((collidables[collidableIndexA].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS) ||(collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS))\n"
-"		{\n"
-"			if (collidables[collidableIndexA].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS) \n"
-"			{\n"
-"				int numChildrenA = collidables[collidableIndexA].m_numChildShapes;\n"
-"				for (int c=0;c<numChildrenA;c++)\n"
-"				{\n"
-"					int childShapeIndexA = collidables[collidableIndexA].m_shapeIndex+c;\n"
-"					int childColIndexA = gpuChildShapes[childShapeIndexA].m_shapeIndex;\n"
-"					float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
-"					float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
-"					float4 childPosA = gpuChildShapes[childShapeIndexA].m_childPosition;\n"
-"					float4 childOrnA = gpuChildShapes[childShapeIndexA].m_childOrientation;\n"
-"					float4 newPosA = qtRotate(ornA,childPosA)+posA;\n"
-"					float4 newOrnA = qtMul(ornA,childOrnA);\n"
-"					int shapeIndexA = collidables[childColIndexA].m_shapeIndex;\n"
-"					b3Aabb_t aabbAlocal = aabbLocalSpace[shapeIndexA];\n"
-"					float margin = 0.f;\n"
-"					\n"
-"					b3Float4 aabbAMinWS;\n"
-"					b3Float4 aabbAMaxWS;\n"
-"					\n"
-"					b3TransformAabb2(aabbAlocal.m_minVec,aabbAlocal.m_maxVec,margin,\n"
-"						newPosA,\n"
-"						newOrnA,\n"
-"						&aabbAMinWS,&aabbAMaxWS);\n"
-"						\n"
-"					\n"
-"					if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
-"					{\n"
-"						int numChildrenB = collidables[collidableIndexB].m_numChildShapes;\n"
-"						for (int b=0;b<numChildrenB;b++)\n"
-"						{\n"
-"							int childShapeIndexB = collidables[collidableIndexB].m_shapeIndex+b;\n"
-"							int childColIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;\n"
-"							float4 ornB = rigidBodies[bodyIndexB].m_quat;\n"
-"							float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
-"							float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;\n"
-"							float4 childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;\n"
-"							float4 newPosB = transform(&childPosB,&posB,&ornB);\n"
-"							float4 newOrnB = qtMul(ornB,childOrnB);\n"
-"							int shapeIndexB = collidables[childColIndexB].m_shapeIndex;\n"
-"							b3Aabb_t aabbBlocal = aabbLocalSpace[shapeIndexB];\n"
-"							\n"
-"							b3Float4 aabbBMinWS;\n"
-"							b3Float4 aabbBMaxWS;\n"
-"							\n"
-"							b3TransformAabb2(aabbBlocal.m_minVec,aabbBlocal.m_maxVec,margin,\n"
-"								newPosB,\n"
-"								newOrnB,\n"
-"								&aabbBMinWS,&aabbBMaxWS);\n"
-"								\n"
-"								\n"
-"							\n"
-"							bool aabbOverlap = b3TestAabbAgainstAabb(aabbAMinWS,aabbAMaxWS,aabbBMinWS,aabbBMaxWS);\n"
-"							if (aabbOverlap)\n"
-"							{\n"
-"								int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
-"								float dmin = FLT_MAX;\n"
-"								float4 posA = newPosA;\n"
-"								posA.w = 0.f;\n"
-"								float4 posB = newPosB;\n"
-"								posB.w = 0.f;\n"
-"								float4 c0local = convexShapes[shapeIndexA].m_localCenter;\n"
-"								float4 ornA = newOrnA;\n"
-"								float4 c0 = transform(&c0local, &posA, &ornA);\n"
-"								float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
-"								float4 ornB =newOrnB;\n"
-"								float4 c1 = transform(&c1local,&posB,&ornB);\n"
-"								const float4 DeltaC2 = c0 - c1;\n"
-"								{//\n"
-"									int compoundPairIdx = atomic_inc(numCompoundPairsOut);\n"
-"									if (compoundPairIdx<maxNumCompoundPairsCapacity)\n"
-"									{\n"
-"										gpuCompoundPairsOut[compoundPairIdx]  = (int4)(bodyIndexA,bodyIndexB,childShapeIndexA,childShapeIndexB);\n"
-"									}\n"
-"								}//\n"
-"							}//fi(1)\n"
-"						} //for (int b=0\n"
-"					}//if (collidables[collidableIndexB].\n"
-"					else//if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
-"					{\n"
-"						if (1)\n"
-"						{\n"
-"							int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
-"							float dmin = FLT_MAX;\n"
-"							float4 posA = newPosA;\n"
-"							posA.w = 0.f;\n"
-"							float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
-"							posB.w = 0.f;\n"
-"							float4 c0local = convexShapes[shapeIndexA].m_localCenter;\n"
-"							float4 ornA = newOrnA;\n"
-"							float4 c0 = transform(&c0local, &posA, &ornA);\n"
-"							float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
-"							float4 ornB = rigidBodies[bodyIndexB].m_quat;\n"
-"							float4 c1 = transform(&c1local,&posB,&ornB);\n"
-"							const float4 DeltaC2 = c0 - c1;\n"
-"							{\n"
-"								int compoundPairIdx = atomic_inc(numCompoundPairsOut);\n"
-"								if (compoundPairIdx<maxNumCompoundPairsCapacity)\n"
-"								{\n"
-"									gpuCompoundPairsOut[compoundPairIdx] = (int4)(bodyIndexA,bodyIndexB,childShapeIndexA,-1);\n"
-"								}//if (compoundPairIdx<maxNumCompoundPairsCapacity)\n"
-"							}//\n"
-"						}//fi (1)\n"
-"					}//if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
-"				}//for (int b=0;b<numChildrenB;b++)	\n"
-"				return;\n"
-"			}//if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
-"			if ((collidables[collidableIndexA].m_shapeType!=SHAPE_CONCAVE_TRIMESH) \n"
-"				&& (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS))\n"
-"			{\n"
-"				int numChildrenB = collidables[collidableIndexB].m_numChildShapes;\n"
-"				for (int b=0;b<numChildrenB;b++)\n"
-"				{\n"
-"					int childShapeIndexB = collidables[collidableIndexB].m_shapeIndex+b;\n"
-"					int childColIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;\n"
-"					float4 ornB = rigidBodies[bodyIndexB].m_quat;\n"
-"					float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
-"					float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;\n"
-"					float4 childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;\n"
-"					float4 newPosB = qtRotate(ornB,childPosB)+posB;\n"
-"					float4 newOrnB = qtMul(ornB,childOrnB);\n"
-"					int shapeIndexB = collidables[childColIndexB].m_shapeIndex;\n"
-"					//////////////////////////////////////\n"
-"					if (1)\n"
-"					{\n"
-"						int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
-"						float dmin = FLT_MAX;\n"
-"						float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
-"						posA.w = 0.f;\n"
-"						float4 posB = newPosB;\n"
-"						posB.w = 0.f;\n"
-"						float4 c0local = convexShapes[shapeIndexA].m_localCenter;\n"
-"						float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
-"						float4 c0 = transform(&c0local, &posA, &ornA);\n"
-"						float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
-"						float4 ornB =newOrnB;\n"
-"						float4 c1 = transform(&c1local,&posB,&ornB);\n"
-"						const float4 DeltaC2 = c0 - c1;\n"
-"						{//\n"
-"							int compoundPairIdx = atomic_inc(numCompoundPairsOut);\n"
-"							if (compoundPairIdx<maxNumCompoundPairsCapacity)\n"
-"							{\n"
-"								gpuCompoundPairsOut[compoundPairIdx] = (int4)(bodyIndexA,bodyIndexB,-1,childShapeIndexB);\n"
-"							}//fi (compoundPairIdx<maxNumCompoundPairsCapacity)\n"
-"						}//\n"
-"					}//fi (1)	\n"
-"				}//for (int b=0;b<numChildrenB;b++)\n"
-"				return;\n"
-"			}//if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
-"			return;\n"
-"		}//fi ((collidables[collidableIndexA].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS) ||(collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS))\n"
-"	}//i<numPairs\n"
-"}\n"
-"// work-in-progress\n"
-"__kernel void   findSeparatingAxisKernel( __global const int4* pairs, \n"
-"																					__global const BodyData* rigidBodies, \n"
-"																					__global const btCollidableGpu* collidables,\n"
-"																					__global const ConvexPolyhedronCL* convexShapes, \n"
-"																					__global const float4* vertices,\n"
-"																					__global const float4* uniqueEdges,\n"
-"																					__global const btGpuFace* faces,\n"
-"																					__global const int* indices,\n"
-"																					__global btAabbCL* aabbs,\n"
-"																					__global volatile float4* separatingNormals,\n"
-"																					__global volatile int* hasSeparatingAxis,\n"
-"																					int numPairs\n"
-"																					)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	\n"
-"	if (i<numPairs)\n"
-"	{\n"
-"	\n"
-"		int bodyIndexA = pairs[i].x;\n"
-"		int bodyIndexB = pairs[i].y;\n"
-"		int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
-"		int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
-"	\n"
-"		int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
-"		int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
-"		\n"
-"		\n"
-"		//once the broadphase avoids static-static pairs, we can remove this test\n"
-"		if ((rigidBodies[bodyIndexA].m_invMass==0) &&(rigidBodies[bodyIndexB].m_invMass==0))\n"
-"		{\n"
-"			hasSeparatingAxis[i] = 0;\n"
-"			return;\n"
-"		}\n"
-"		\n"
-"		if ((collidables[collidableIndexA].m_shapeType!=SHAPE_CONVEX_HULL) ||(collidables[collidableIndexB].m_shapeType!=SHAPE_CONVEX_HULL))\n"
-"		{\n"
-"			hasSeparatingAxis[i] = 0;\n"
-"			return;\n"
-"		}\n"
-"			\n"
-"		if ((collidables[collidableIndexA].m_shapeType==SHAPE_CONCAVE_TRIMESH))\n"
-"		{\n"
-"			hasSeparatingAxis[i] = 0;\n"
-"			return;\n"
-"		}\n"
-"		int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
-"		float dmin = FLT_MAX;\n"
-"		float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
-"		posA.w = 0.f;\n"
-"		float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
-"		posB.w = 0.f;\n"
-"		float4 c0local = convexShapes[shapeIndexA].m_localCenter;\n"
-"		float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
-"		float4 c0 = transform(&c0local, &posA, &ornA);\n"
-"		float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
-"		float4 ornB =rigidBodies[bodyIndexB].m_quat;\n"
-"		float4 c1 = transform(&c1local,&posB,&ornB);\n"
-"		const float4 DeltaC2 = c0 - c1;\n"
-"		float4 sepNormal;\n"
-"		\n"
-"		bool sepA = findSeparatingAxis(	&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],posA,ornA,\n"
-"																								posB,ornB,\n"
-"																								DeltaC2,\n"
-"																								vertices,uniqueEdges,faces,\n"
-"																								indices,&sepNormal,&dmin);\n"
-"		hasSeparatingAxis[i] = 4;\n"
-"		if (!sepA)\n"
-"		{\n"
-"			hasSeparatingAxis[i] = 0;\n"
-"		} else\n"
-"		{\n"
-"			bool sepB = findSeparatingAxis(	&convexShapes[shapeIndexB],&convexShapes[shapeIndexA],posB,ornB,\n"
-"																									posA,ornA,\n"
-"																									DeltaC2,\n"
-"																									vertices,uniqueEdges,faces,\n"
-"																									indices,&sepNormal,&dmin);\n"
-"			if (!sepB)\n"
-"			{\n"
-"				hasSeparatingAxis[i] = 0;\n"
-"			} else\n"
-"			{\n"
-"				bool sepEE = findSeparatingAxisEdgeEdge(	&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],posA,ornA,\n"
-"																									posB,ornB,\n"
-"																									DeltaC2,\n"
-"																									vertices,uniqueEdges,faces,\n"
-"																									indices,&sepNormal,&dmin);\n"
-"				if (!sepEE)\n"
-"				{\n"
-"					hasSeparatingAxis[i] = 0;\n"
-"				} else\n"
-"				{\n"
-"					hasSeparatingAxis[i] = 1;\n"
-"					separatingNormals[i] = sepNormal;\n"
-"				}\n"
-"			}\n"
-"		}\n"
-"		\n"
-"	}\n"
-"}\n"
-"__kernel void   findSeparatingAxisVertexFaceKernel( __global const int4* pairs, \n"
-"																					__global const BodyData* rigidBodies, \n"
-"																					__global const btCollidableGpu* collidables,\n"
-"																					__global const ConvexPolyhedronCL* convexShapes, \n"
-"																					__global const float4* vertices,\n"
-"																					__global const float4* uniqueEdges,\n"
-"																					__global const btGpuFace* faces,\n"
-"																					__global const int* indices,\n"
-"																					__global btAabbCL* aabbs,\n"
-"																					__global volatile float4* separatingNormals,\n"
-"																					__global volatile int* hasSeparatingAxis,\n"
-"																					__global  float* dmins,\n"
-"																					int numPairs\n"
-"																					)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	\n"
-"	if (i<numPairs)\n"
-"	{\n"
-"	\n"
-"		int bodyIndexA = pairs[i].x;\n"
-"		int bodyIndexB = pairs[i].y;\n"
-"		int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
-"		int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
-"	\n"
-"		int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
-"		int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
-"	\n"
-"		hasSeparatingAxis[i] = 0;	\n"
-"		\n"
-"		//once the broadphase avoids static-static pairs, we can remove this test\n"
-"		if ((rigidBodies[bodyIndexA].m_invMass==0) &&(rigidBodies[bodyIndexB].m_invMass==0))\n"
-"		{\n"
-"			return;\n"
-"		}\n"
-"		\n"
-"		if ((collidables[collidableIndexA].m_shapeType!=SHAPE_CONVEX_HULL) ||(collidables[collidableIndexB].m_shapeType!=SHAPE_CONVEX_HULL))\n"
-"		{\n"
-"			return;\n"
-"		}\n"
-"			\n"
-"		int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
-"		float dmin = FLT_MAX;\n"
-"		dmins[i] = dmin;\n"
-"		\n"
-"		float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
-"		posA.w = 0.f;\n"
-"		float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
-"		posB.w = 0.f;\n"
-"		float4 c0local = convexShapes[shapeIndexA].m_localCenter;\n"
-"		float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
-"		float4 c0 = transform(&c0local, &posA, &ornA);\n"
-"		float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
-"		float4 ornB =rigidBodies[bodyIndexB].m_quat;\n"
-"		float4 c1 = transform(&c1local,&posB,&ornB);\n"
-"		const float4 DeltaC2 = c0 - c1;\n"
-"		float4 sepNormal;\n"
-"		\n"
-"		bool sepA = findSeparatingAxis(	&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],posA,ornA,\n"
-"																								posB,ornB,\n"
-"																								DeltaC2,\n"
-"																								vertices,uniqueEdges,faces,\n"
-"																								indices,&sepNormal,&dmin);\n"
-"		hasSeparatingAxis[i] = 4;\n"
-"		if (!sepA)\n"
-"		{\n"
-"			hasSeparatingAxis[i] = 0;\n"
-"		} else\n"
-"		{\n"
-"			bool sepB = findSeparatingAxis(	&convexShapes[shapeIndexB],&convexShapes[shapeIndexA],posB,ornB,\n"
-"																									posA,ornA,\n"
-"																									DeltaC2,\n"
-"																									vertices,uniqueEdges,faces,\n"
-"																									indices,&sepNormal,&dmin);\n"
-"			if (sepB)\n"
-"			{\n"
-"				dmins[i] = dmin;\n"
-"				hasSeparatingAxis[i] = 1;\n"
-"				separatingNormals[i] = sepNormal;\n"
-"			}\n"
-"		}\n"
-"		\n"
-"	}\n"
-"}\n"
-"__kernel void   findSeparatingAxisEdgeEdgeKernel( __global const int4* pairs, \n"
-"																					__global const BodyData* rigidBodies, \n"
-"																					__global const btCollidableGpu* collidables,\n"
-"																					__global const ConvexPolyhedronCL* convexShapes, \n"
-"																					__global const float4* vertices,\n"
-"																					__global const float4* uniqueEdges,\n"
-"																					__global const btGpuFace* faces,\n"
-"																					__global const int* indices,\n"
-"																					__global btAabbCL* aabbs,\n"
-"																					__global  float4* separatingNormals,\n"
-"																					__global  int* hasSeparatingAxis,\n"
-"																					__global  float* dmins,\n"
-"																					__global const float4* unitSphereDirections,\n"
-"																					int numUnitSphereDirections,\n"
-"																					int numPairs\n"
-"																					)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	\n"
-"	if (i<numPairs)\n"
-"	{\n"
-"		if (hasSeparatingAxis[i])\n"
-"		{\n"
-"	\n"
-"			int bodyIndexA = pairs[i].x;\n"
-"			int bodyIndexB = pairs[i].y;\n"
-"	\n"
-"			int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
-"			int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
-"		\n"
-"			int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
-"			int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
-"			\n"
-"			\n"
-"			int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
-"	\n"
-"			float dmin = dmins[i];\n"
-"	\n"
-"			float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
-"			posA.w = 0.f;\n"
-"			float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
-"			posB.w = 0.f;\n"
-"			float4 c0local = convexShapes[shapeIndexA].m_localCenter;\n"
-"			float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
-"			float4 c0 = transform(&c0local, &posA, &ornA);\n"
-"			float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
-"			float4 ornB =rigidBodies[bodyIndexB].m_quat;\n"
-"			float4 c1 = transform(&c1local,&posB,&ornB);\n"
-"			const float4 DeltaC2 = c0 - c1;\n"
-"			float4 sepNormal = separatingNormals[i];\n"
-"			\n"
-"			\n"
-"			\n"
-"			bool sepEE = false;\n"
-"			int numEdgeEdgeDirections = convexShapes[shapeIndexA].m_numUniqueEdges*convexShapes[shapeIndexB].m_numUniqueEdges;\n"
-"			if (numEdgeEdgeDirections<=numUnitSphereDirections)\n"
-"			{\n"
-"				sepEE = findSeparatingAxisEdgeEdge(	&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],posA,ornA,\n"
-"																									posB,ornB,\n"
-"																									DeltaC2,\n"
-"																									vertices,uniqueEdges,faces,\n"
-"																									indices,&sepNormal,&dmin);\n"
-"																									\n"
-"					if (!sepEE)\n"
-"					{\n"
-"						hasSeparatingAxis[i] = 0;\n"
-"					} else\n"
-"					{\n"
-"						hasSeparatingAxis[i] = 1;\n"
-"						separatingNormals[i] = sepNormal;\n"
-"					}\n"
-"			}\n"
-"			/*\n"
-"			///else case is a separate kernel, to make Mac OSX OpenCL compiler happy\n"
-"			else\n"
-"			{\n"
-"				sepEE = findSeparatingAxisUnitSphere(&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],posA,ornA,\n"
-"																									posB,ornB,\n"
-"																									DeltaC2,\n"
-"																									vertices,unitSphereDirections,numUnitSphereDirections,\n"
-"																									&sepNormal,&dmin);\n"
-"					if (!sepEE)\n"
-"					{\n"
-"						hasSeparatingAxis[i] = 0;\n"
-"					} else\n"
-"					{\n"
-"						hasSeparatingAxis[i] = 1;\n"
-"						separatingNormals[i] = sepNormal;\n"
-"					}\n"
-"			}\n"
-"			*/\n"
-"		}		//if (hasSeparatingAxis[i])\n"
-"	}//(i<numPairs)\n"
-"}\n"
-"inline int	findClippingFaces(const float4 separatingNormal,\n"
-"                      const ConvexPolyhedronCL* hullA, \n"
-"					  __global const ConvexPolyhedronCL* hullB,\n"
-"                      const float4 posA, const Quaternion ornA,const float4 posB, const Quaternion ornB,\n"
-"                       __global float4* worldVertsA1,\n"
-"                      __global float4* worldNormalsA1,\n"
-"                      __global float4* worldVertsB1,\n"
-"                      int capacityWorldVerts,\n"
-"                      const float minDist, float maxDist,\n"
-"					  const float4* verticesA,\n"
-"                      const btGpuFace* facesA,\n"
-"                      const int* indicesA,\n"
-"					  __global const float4* verticesB,\n"
-"                      __global const btGpuFace* facesB,\n"
-"                      __global const int* indicesB,\n"
-"                      __global int4* clippingFaces, int pairIndex)\n"
-"{\n"
-"	int numContactsOut = 0;\n"
-"	int numWorldVertsB1= 0;\n"
-"    \n"
-"    \n"
-"	int closestFaceB=0;\n"
-"	float dmax = -FLT_MAX;\n"
-"    \n"
-"	{\n"
-"		for(int face=0;face<hullB->m_numFaces;face++)\n"
-"		{\n"
-"			const float4 Normal = make_float4(facesB[hullB->m_faceOffset+face].m_plane.x,\n"
-"                                              facesB[hullB->m_faceOffset+face].m_plane.y, facesB[hullB->m_faceOffset+face].m_plane.z,0.f);\n"
-"			const float4 WorldNormal = qtRotate(ornB, Normal);\n"
-"			float d = dot3F4(WorldNormal,separatingNormal);\n"
-"			if (d > dmax)\n"
-"			{\n"
-"				dmax = d;\n"
-"				closestFaceB = face;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"    \n"
-"	{\n"
-"		const btGpuFace polyB = facesB[hullB->m_faceOffset+closestFaceB];\n"
-"		int numVertices = polyB.m_numIndices;\n"
-"        if (numVertices>capacityWorldVerts)\n"
-"            numVertices = capacityWorldVerts;\n"
-"        \n"
-"		for(int e0=0;e0<numVertices;e0++)\n"
-"		{\n"
-"            if (e0<capacityWorldVerts)\n"
-"            {\n"
-"                const float4 b = verticesB[hullB->m_vertexOffset+indicesB[polyB.m_indexOffset+e0]];\n"
-"                worldVertsB1[pairIndex*capacityWorldVerts+numWorldVertsB1++] = transform(&b,&posB,&ornB);\n"
-"            }\n"
-"		}\n"
-"	}\n"
-"    \n"
-"    int closestFaceA=0;\n"
-"	{\n"
-"		float dmin = FLT_MAX;\n"
-"		for(int face=0;face<hullA->m_numFaces;face++)\n"
-"		{\n"
-"			const float4 Normal = make_float4(\n"
-"                                              facesA[hullA->m_faceOffset+face].m_plane.x,\n"
-"                                              facesA[hullA->m_faceOffset+face].m_plane.y,\n"
-"                                              facesA[hullA->m_faceOffset+face].m_plane.z,\n"
-"                                              0.f);\n"
-"			const float4 faceANormalWS = qtRotate(ornA,Normal);\n"
-"            \n"
-"			float d = dot3F4(faceANormalWS,separatingNormal);\n"
-"			if (d < dmin)\n"
-"			{\n"
-"				dmin = d;\n"
-"				closestFaceA = face;\n"
-"                worldNormalsA1[pairIndex] = faceANormalWS;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"    \n"
-"    int numVerticesA = facesA[hullA->m_faceOffset+closestFaceA].m_numIndices;\n"
-"    if (numVerticesA>capacityWorldVerts)\n"
-"       numVerticesA = capacityWorldVerts;\n"
-"    \n"
-"	for(int e0=0;e0<numVerticesA;e0++)\n"
-"	{\n"
-"        if (e0<capacityWorldVerts)\n"
-"        {\n"
-"            const float4 a = verticesA[hullA->m_vertexOffset+indicesA[facesA[hullA->m_faceOffset+closestFaceA].m_indexOffset+e0]];\n"
-"            worldVertsA1[pairIndex*capacityWorldVerts+e0] = transform(&a, &posA,&ornA);\n"
-"        }\n"
-"    }\n"
-"    \n"
-"    clippingFaces[pairIndex].x = closestFaceA;\n"
-"    clippingFaces[pairIndex].y = closestFaceB;\n"
-"    clippingFaces[pairIndex].z = numVerticesA;\n"
-"    clippingFaces[pairIndex].w = numWorldVertsB1;\n"
-"    \n"
-"    \n"
-"	return numContactsOut;\n"
-"}\n"
-"// work-in-progress\n"
-"__kernel void   findConcaveSeparatingAxisKernel( __global int4* concavePairs,\n"
-"																					__global const BodyData* rigidBodies,\n"
-"																					__global const btCollidableGpu* collidables,\n"
-"																					__global const ConvexPolyhedronCL* convexShapes, \n"
-"																					__global const float4* vertices,\n"
-"																					__global const float4* uniqueEdges,\n"
-"																					__global const btGpuFace* faces,\n"
-"																					__global const int* indices,\n"
-"																					__global const btGpuChildShape* gpuChildShapes,\n"
-"																					__global btAabbCL* aabbs,\n"
-"																					__global float4* concaveSeparatingNormalsOut,\n"
-"																					__global int* concaveHasSeparatingNormals,\n"
-"																					__global int4* clippingFacesOut,\n"
-"																					__global float4* worldVertsA1GPU,\n"
-"																					__global float4*  worldNormalsAGPU,\n"
-"																					__global float4* worldVertsB1GPU,\n"
-"																					int vertexFaceCapacity,\n"
-"																					int numConcavePairs\n"
-"																					)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	if (i>=numConcavePairs)\n"
-"		return;\n"
-"	concaveHasSeparatingNormals[i] = 0;\n"
-"	int pairIdx = i;\n"
-"	int bodyIndexA = concavePairs[i].x;\n"
-"	int bodyIndexB = concavePairs[i].y;\n"
-"	int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
-"	int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
-"	int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
-"	int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
-"	if (collidables[collidableIndexB].m_shapeType!=SHAPE_CONVEX_HULL&&\n"
-"		collidables[collidableIndexB].m_shapeType!=SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
-"	{\n"
-"		concavePairs[pairIdx].w = -1;\n"
-"		return;\n"
-"	}\n"
-"	int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
-"	int numActualConcaveConvexTests = 0;\n"
-"	\n"
-"	int f = concavePairs[i].z;\n"
-"	\n"
-"	bool overlap = false;\n"
-"	\n"
-"	ConvexPolyhedronCL convexPolyhedronA;\n"
-"	//add 3 vertices of the triangle\n"
-"	convexPolyhedronA.m_numVertices = 3;\n"
-"	convexPolyhedronA.m_vertexOffset = 0;\n"
-"	float4	localCenter = make_float4(0.f,0.f,0.f,0.f);\n"
-"	btGpuFace face = faces[convexShapes[shapeIndexA].m_faceOffset+f];\n"
-"	float4 triMinAabb, triMaxAabb;\n"
-"	btAabbCL triAabb;\n"
-"	triAabb.m_min = make_float4(1e30f,1e30f,1e30f,0.f);\n"
-"	triAabb.m_max = make_float4(-1e30f,-1e30f,-1e30f,0.f);\n"
-"	\n"
-"	float4 verticesA[3];\n"
-"	for (int i=0;i<3;i++)\n"
-"	{\n"
-"		int index = indices[face.m_indexOffset+i];\n"
-"		float4 vert = vertices[convexShapes[shapeIndexA].m_vertexOffset+index];\n"
-"		verticesA[i] = vert;\n"
-"		localCenter += vert;\n"
-"			\n"
-"		triAabb.m_min = min(triAabb.m_min,vert);		\n"
-"		triAabb.m_max = max(triAabb.m_max,vert);		\n"
-"	}\n"
-"	overlap = true;\n"
-"	overlap = (triAabb.m_min.x > aabbs[bodyIndexB].m_max.x || triAabb.m_max.x < aabbs[bodyIndexB].m_min.x) ? false : overlap;\n"
-"	overlap = (triAabb.m_min.z > aabbs[bodyIndexB].m_max.z || triAabb.m_max.z < aabbs[bodyIndexB].m_min.z) ? false : overlap;\n"
-"	overlap = (triAabb.m_min.y > aabbs[bodyIndexB].m_max.y || triAabb.m_max.y < aabbs[bodyIndexB].m_min.y) ? false : overlap;\n"
-"		\n"
-"	if (overlap)\n"
-"	{\n"
-"		float dmin = FLT_MAX;\n"
-"		int hasSeparatingAxis=5;\n"
-"		float4 sepAxis=make_float4(1,2,3,4);\n"
-"		int localCC=0;\n"
-"		numActualConcaveConvexTests++;\n"
-"		//a triangle has 3 unique edges\n"
-"		convexPolyhedronA.m_numUniqueEdges = 3;\n"
-"		convexPolyhedronA.m_uniqueEdgesOffset = 0;\n"
-"		float4 uniqueEdgesA[3];\n"
-"		\n"
-"		uniqueEdgesA[0] = (verticesA[1]-verticesA[0]);\n"
-"		uniqueEdgesA[1] = (verticesA[2]-verticesA[1]);\n"
-"		uniqueEdgesA[2] = (verticesA[0]-verticesA[2]);\n"
-"		convexPolyhedronA.m_faceOffset = 0;\n"
-"                                  \n"
-"		float4 normal = make_float4(face.m_plane.x,face.m_plane.y,face.m_plane.z,0.f);\n"
-"                             \n"
-"		btGpuFace facesA[TRIANGLE_NUM_CONVEX_FACES];\n"
-"		int indicesA[3+3+2+2+2];\n"
-"		int curUsedIndices=0;\n"
-"		int fidx=0;\n"
-"		//front size of triangle\n"
-"		{\n"
-"			facesA[fidx].m_indexOffset=curUsedIndices;\n"
-"			indicesA[0] = 0;\n"
-"			indicesA[1] = 1;\n"
-"			indicesA[2] = 2;\n"
-"			curUsedIndices+=3;\n"
-"			float c = face.m_plane.w;\n"
-"			facesA[fidx].m_plane.x = normal.x;\n"
-"			facesA[fidx].m_plane.y = normal.y;\n"
-"			facesA[fidx].m_plane.z = normal.z;\n"
-"			facesA[fidx].m_plane.w = c;\n"
-"			facesA[fidx].m_numIndices=3;\n"
-"		}\n"
-"		fidx++;\n"
-"		//back size of triangle\n"
-"		{\n"
-"			facesA[fidx].m_indexOffset=curUsedIndices;\n"
-"			indicesA[3]=2;\n"
-"			indicesA[4]=1;\n"
-"			indicesA[5]=0;\n"
-"			curUsedIndices+=3;\n"
-"			float c = dot(normal,verticesA[0]);\n"
-"			float c1 = -face.m_plane.w;\n"
-"			facesA[fidx].m_plane.x = -normal.x;\n"
-"			facesA[fidx].m_plane.y = -normal.y;\n"
-"			facesA[fidx].m_plane.z = -normal.z;\n"
-"			facesA[fidx].m_plane.w = c;\n"
-"			facesA[fidx].m_numIndices=3;\n"
-"		}\n"
-"		fidx++;\n"
-"		bool addEdgePlanes = true;\n"
-"		if (addEdgePlanes)\n"
-"		{\n"
-"			int numVertices=3;\n"
-"			int prevVertex = numVertices-1;\n"
-"			for (int i=0;i<numVertices;i++)\n"
-"			{\n"
-"				float4 v0 = verticesA[i];\n"
-"				float4 v1 = verticesA[prevVertex];\n"
-"                                            \n"
-"				float4 edgeNormal = normalize(cross(normal,v1-v0));\n"
-"				float c = -dot(edgeNormal,v0);\n"
-"				facesA[fidx].m_numIndices = 2;\n"
-"				facesA[fidx].m_indexOffset=curUsedIndices;\n"
-"				indicesA[curUsedIndices++]=i;\n"
-"				indicesA[curUsedIndices++]=prevVertex;\n"
-"                                            \n"
-"				facesA[fidx].m_plane.x = edgeNormal.x;\n"
-"				facesA[fidx].m_plane.y = edgeNormal.y;\n"
-"				facesA[fidx].m_plane.z = edgeNormal.z;\n"
-"				facesA[fidx].m_plane.w = c;\n"
-"				fidx++;\n"
-"				prevVertex = i;\n"
-"			}\n"
-"		}\n"
-"		convexPolyhedronA.m_numFaces = TRIANGLE_NUM_CONVEX_FACES;\n"
-"		convexPolyhedronA.m_localCenter = localCenter*(1.f/3.f);\n"
-"		float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
-"		posA.w = 0.f;\n"
-"		float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
-"		posB.w = 0.f;\n"
-"		float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
-"		float4 ornB =rigidBodies[bodyIndexB].m_quat;\n"
-"		\n"
-"		///////////////////\n"
-"		///compound shape support\n"
-"		if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
-"		{\n"
-"			int compoundChild = concavePairs[pairIdx].w;\n"
-"			int childShapeIndexB = compoundChild;//collidables[collidableIndexB].m_shapeIndex+compoundChild;\n"
-"			int childColIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;\n"
-"			float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;\n"
-"			float4 childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;\n"
-"			float4 newPosB = transform(&childPosB,&posB,&ornB);\n"
-"			float4 newOrnB = qtMul(ornB,childOrnB);\n"
-"			posB = newPosB;\n"
-"			ornB = newOrnB;\n"
-"			shapeIndexB = collidables[childColIndexB].m_shapeIndex;\n"
-"		}\n"
-"		//////////////////\n"
-"		float4 c0local = convexPolyhedronA.m_localCenter;\n"
-"		float4 c0 = transform(&c0local, &posA, &ornA);\n"
-"		float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
-"		float4 c1 = transform(&c1local,&posB,&ornB);\n"
-"		const float4 DeltaC2 = c0 - c1;\n"
-"		bool sepA = findSeparatingAxisLocalA(	&convexPolyhedronA, &convexShapes[shapeIndexB],\n"
-"												posA,ornA,\n"
-"												posB,ornB,\n"
-"												DeltaC2,\n"
-"												verticesA,uniqueEdgesA,facesA,indicesA,\n"
-"												vertices,uniqueEdges,faces,indices,\n"
-"												&sepAxis,&dmin);\n"
-"		hasSeparatingAxis = 4;\n"
-"		if (!sepA)\n"
-"		{\n"
-"			hasSeparatingAxis = 0;\n"
-"		} else\n"
-"		{\n"
-"			bool sepB = findSeparatingAxisLocalB(	&convexShapes[shapeIndexB],&convexPolyhedronA,\n"
-"												posB,ornB,\n"
-"												posA,ornA,\n"
-"												DeltaC2,\n"
-"												vertices,uniqueEdges,faces,indices,\n"
-"												verticesA,uniqueEdgesA,facesA,indicesA,\n"
-"												&sepAxis,&dmin);\n"
-"			if (!sepB)\n"
-"			{\n"
-"				hasSeparatingAxis = 0;\n"
-"			} else\n"
-"			{\n"
-"				bool sepEE = findSeparatingAxisEdgeEdgeLocalA(	&convexPolyhedronA, &convexShapes[shapeIndexB],\n"
-"															posA,ornA,\n"
-"															posB,ornB,\n"
-"															DeltaC2,\n"
-"															verticesA,uniqueEdgesA,facesA,indicesA,\n"
-"															vertices,uniqueEdges,faces,indices,\n"
-"															&sepAxis,&dmin);\n"
-"	\n"
-"				if (!sepEE)\n"
-"				{\n"
-"					hasSeparatingAxis = 0;\n"
-"				} else\n"
-"				{\n"
-"					hasSeparatingAxis = 1;\n"
-"				}\n"
-"			}\n"
-"		}	\n"
-"		\n"
-"		if (hasSeparatingAxis)\n"
-"		{\n"
-"			sepAxis.w = dmin;\n"
-"			concaveSeparatingNormalsOut[pairIdx]=sepAxis;\n"
-"			concaveHasSeparatingNormals[i]=1;\n"
-"			float minDist = -1e30f;\n"
-"			float maxDist = 0.02f;\n"
-"		\n"
-"			findClippingFaces(sepAxis,\n"
-"                     &convexPolyhedronA,\n"
-"					 &convexShapes[shapeIndexB],\n"
-"					 posA,ornA,\n"
-"					 posB,ornB,\n"
-"                      worldVertsA1GPU,\n"
-"                      worldNormalsAGPU,\n"
-"                      worldVertsB1GPU,\n"
-"					  vertexFaceCapacity,\n"
-"                      minDist, maxDist,\n"
-"                      verticesA,\n"
-"                      facesA,\n"
-"                      indicesA,\n"
-" 					  vertices,\n"
-"                      faces,\n"
-"                      indices,\n"
-"                      clippingFacesOut, pairIdx);\n"
-"		} else\n"
-"		{	\n"
-"			//mark this pair as in-active\n"
-"			concavePairs[pairIdx].w = -1;\n"
-"		}\n"
-"	}\n"
-"	else\n"
-"	{	\n"
-"		//mark this pair as in-active\n"
-"		concavePairs[pairIdx].w = -1;\n"
-"	}\n"
-"	\n"
-"	concavePairs[pairIdx].z = -1;//now z is used for existing/persistent contacts\n"
-"}\n"
-;
+static const char* satKernelsCL =
+	"//keep this enum in sync with the CPU version (in btCollidable.h)\n"
+	"//written by Erwin Coumans\n"
+	"#define SHAPE_CONVEX_HULL 3\n"
+	"#define SHAPE_CONCAVE_TRIMESH 5\n"
+	"#define TRIANGLE_NUM_CONVEX_FACES 5\n"
+	"#define SHAPE_COMPOUND_OF_CONVEX_HULLS 6\n"
+	"#define B3_MAX_STACK_DEPTH 256\n"
+	"typedef unsigned int u32;\n"
+	"///keep this in sync with btCollidable.h\n"
+	"typedef struct\n"
+	"{\n"
+	"	union {\n"
+	"		int m_numChildShapes;\n"
+	"		int m_bvhIndex;\n"
+	"	};\n"
+	"	union\n"
+	"	{\n"
+	"		float m_radius;\n"
+	"		int	m_compoundBvhIndex;\n"
+	"	};\n"
+	"	\n"
+	"	int m_shapeType;\n"
+	"	int m_shapeIndex;\n"
+	"	\n"
+	"} btCollidableGpu;\n"
+	"#define MAX_NUM_PARTS_IN_BITS 10\n"
+	"///b3QuantizedBvhNode is a compressed aabb node, 16 bytes.\n"
+	"///Node can be used for leafnode or internal node. Leafnodes can point to 32-bit triangle index (non-negative range).\n"
+	"typedef struct\n"
+	"{\n"
+	"	//12 bytes\n"
+	"	unsigned short int	m_quantizedAabbMin[3];\n"
+	"	unsigned short int	m_quantizedAabbMax[3];\n"
+	"	//4 bytes\n"
+	"	int	m_escapeIndexOrTriangleIndex;\n"
+	"} b3QuantizedBvhNode;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4		m_aabbMin;\n"
+	"	float4		m_aabbMax;\n"
+	"	float4		m_quantization;\n"
+	"	int			m_numNodes;\n"
+	"	int			m_numSubTrees;\n"
+	"	int			m_nodeOffset;\n"
+	"	int			m_subTreeOffset;\n"
+	"} b3BvhInfo;\n"
+	"int	getTriangleIndex(const b3QuantizedBvhNode* rootNode)\n"
+	"{\n"
+	"	unsigned int x=0;\n"
+	"	unsigned int y = (~(x&0))<<(31-MAX_NUM_PARTS_IN_BITS);\n"
+	"	// Get only the lower bits where the triangle index is stored\n"
+	"	return (rootNode->m_escapeIndexOrTriangleIndex&~(y));\n"
+	"}\n"
+	"int	getTriangleIndexGlobal(__global const b3QuantizedBvhNode* rootNode)\n"
+	"{\n"
+	"	unsigned int x=0;\n"
+	"	unsigned int y = (~(x&0))<<(31-MAX_NUM_PARTS_IN_BITS);\n"
+	"	// Get only the lower bits where the triangle index is stored\n"
+	"	return (rootNode->m_escapeIndexOrTriangleIndex&~(y));\n"
+	"}\n"
+	"int isLeafNode(const b3QuantizedBvhNode* rootNode)\n"
+	"{\n"
+	"	//skipindex is negative (internal node), triangleindex >=0 (leafnode)\n"
+	"	return (rootNode->m_escapeIndexOrTriangleIndex >= 0)? 1 : 0;\n"
+	"}\n"
+	"int isLeafNodeGlobal(__global const b3QuantizedBvhNode* rootNode)\n"
+	"{\n"
+	"	//skipindex is negative (internal node), triangleindex >=0 (leafnode)\n"
+	"	return (rootNode->m_escapeIndexOrTriangleIndex >= 0)? 1 : 0;\n"
+	"}\n"
+	"	\n"
+	"int getEscapeIndex(const b3QuantizedBvhNode* rootNode)\n"
+	"{\n"
+	"	return -rootNode->m_escapeIndexOrTriangleIndex;\n"
+	"}\n"
+	"int getEscapeIndexGlobal(__global const b3QuantizedBvhNode* rootNode)\n"
+	"{\n"
+	"	return -rootNode->m_escapeIndexOrTriangleIndex;\n"
+	"}\n"
+	"typedef struct\n"
+	"{\n"
+	"	//12 bytes\n"
+	"	unsigned short int	m_quantizedAabbMin[3];\n"
+	"	unsigned short int	m_quantizedAabbMax[3];\n"
+	"	//4 bytes, points to the root of the subtree\n"
+	"	int			m_rootNodeIndex;\n"
+	"	//4 bytes\n"
+	"	int			m_subtreeSize;\n"
+	"	int			m_padding[3];\n"
+	"} b3BvhSubtreeInfo;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4	m_childPosition;\n"
+	"	float4	m_childOrientation;\n"
+	"	int m_shapeIndex;\n"
+	"	int m_unused0;\n"
+	"	int m_unused1;\n"
+	"	int m_unused2;\n"
+	"} btGpuChildShape;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_pos;\n"
+	"	float4 m_quat;\n"
+	"	float4 m_linVel;\n"
+	"	float4 m_angVel;\n"
+	"	u32 m_collidableIdx;\n"
+	"	float m_invMass;\n"
+	"	float m_restituitionCoeff;\n"
+	"	float m_frictionCoeff;\n"
+	"} BodyData;\n"
+	"typedef struct  \n"
+	"{\n"
+	"	float4		m_localCenter;\n"
+	"	float4		m_extents;\n"
+	"	float4		mC;\n"
+	"	float4		mE;\n"
+	"	\n"
+	"	float			m_radius;\n"
+	"	int	m_faceOffset;\n"
+	"	int m_numFaces;\n"
+	"	int	m_numVertices;\n"
+	"	int m_vertexOffset;\n"
+	"	int	m_uniqueEdgesOffset;\n"
+	"	int	m_numUniqueEdges;\n"
+	"	int m_unused;\n"
+	"} ConvexPolyhedronCL;\n"
+	"typedef struct \n"
+	"{\n"
+	"	union\n"
+	"	{\n"
+	"		float4	m_min;\n"
+	"		float   m_minElems[4];\n"
+	"		int			m_minIndices[4];\n"
+	"	};\n"
+	"	union\n"
+	"	{\n"
+	"		float4	m_max;\n"
+	"		float   m_maxElems[4];\n"
+	"		int			m_maxIndices[4];\n"
+	"	};\n"
+	"} btAabbCL;\n"
+	"#ifndef B3_AABB_H\n"
+	"#define B3_AABB_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#define B3_FLOAT4_H\n"
+	"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
+	"#define B3_PLATFORM_DEFINITIONS_H\n"
+	"struct MyTest\n"
+	"{\n"
+	"	int bla;\n"
+	"};\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
+	"#define B3_LARGE_FLOAT 1e18f\n"
+	"#define B3_INFINITY 1e18f\n"
+	"#define b3Assert(a)\n"
+	"#define b3ConstArray(a) __global const a*\n"
+	"#define b3AtomicInc atomic_inc\n"
+	"#define b3AtomicAdd atomic_add\n"
+	"#define b3Fabs fabs\n"
+	"#define b3Sqrt native_sqrt\n"
+	"#define b3Sin native_sin\n"
+	"#define b3Cos native_cos\n"
+	"#define B3_STATIC\n"
+	"#endif\n"
+	"#endif\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"	typedef float4	b3Float4;\n"
+	"	#define b3Float4ConstArg const b3Float4\n"
+	"	#define b3MakeFloat4 (float4)\n"
+	"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return dot(a1, b1);\n"
+	"	}\n"
+	"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return cross(a1, b1);\n"
+	"	}\n"
+	"	#define b3MinFloat4 min\n"
+	"	#define b3MaxFloat4 max\n"
+	"	#define b3Normalized(a) normalize(a)\n"
+	"#endif \n"
+	"		\n"
+	"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
+	"{\n"
+	"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
+	"		return false;\n"
+	"	return true;\n"
+	"}\n"
+	"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
+	"{\n"
+	"    float maxDot = -B3_INFINITY;\n"
+	"    int i = 0;\n"
+	"    int ptIndex = -1;\n"
+	"    for( i = 0; i < vecLen; i++ )\n"
+	"    {\n"
+	"        float dot = b3Dot3F4(vecArray[i],vec);\n"
+	"            \n"
+	"        if( dot > maxDot )\n"
+	"        {\n"
+	"            maxDot = dot;\n"
+	"            ptIndex = i;\n"
+	"        }\n"
+	"    }\n"
+	"	b3Assert(ptIndex>=0);\n"
+	"    if (ptIndex<0)\n"
+	"	{\n"
+	"		ptIndex = 0;\n"
+	"	}\n"
+	"    *dotOut = maxDot;\n"
+	"    return ptIndex;\n"
+	"}\n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifndef B3_MAT3x3_H\n"
+	"#define B3_MAT3x3_H\n"
+	"#ifndef B3_QUAT_H\n"
+	"#define B3_QUAT_H\n"
+	"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif\n"
+	"#endif\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"	typedef float4	b3Quat;\n"
+	"	#define b3QuatConstArg const b3Quat\n"
+	"	\n"
+	"	\n"
+	"inline float4 b3FastNormalize4(float4 v)\n"
+	"{\n"
+	"	v = (float4)(v.xyz,0.f);\n"
+	"	return fast_normalize(v);\n"
+	"}\n"
+	"	\n"
+	"inline b3Quat b3QuatMul(b3Quat a, b3Quat b);\n"
+	"inline b3Quat b3QuatNormalized(b3QuatConstArg in);\n"
+	"inline b3Quat b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec);\n"
+	"inline b3Quat b3QuatInvert(b3QuatConstArg q);\n"
+	"inline b3Quat b3QuatInverse(b3QuatConstArg q);\n"
+	"inline b3Quat b3QuatMul(b3QuatConstArg a, b3QuatConstArg b)\n"
+	"{\n"
+	"	b3Quat ans;\n"
+	"	ans = b3Cross3( a, b );\n"
+	"	ans += a.w*b+b.w*a;\n"
+	"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
+	"	ans.w = a.w*b.w - b3Dot3F4(a, b);\n"
+	"	return ans;\n"
+	"}\n"
+	"inline b3Quat b3QuatNormalized(b3QuatConstArg in)\n"
+	"{\n"
+	"	b3Quat q;\n"
+	"	q=in;\n"
+	"	//return b3FastNormalize4(in);\n"
+	"	float len = native_sqrt(dot(q, q));\n"
+	"	if(len > 0.f)\n"
+	"	{\n"
+	"		q *= 1.f / len;\n"
+	"	}\n"
+	"	else\n"
+	"	{\n"
+	"		q.x = q.y = q.z = 0.f;\n"
+	"		q.w = 1.f;\n"
+	"	}\n"
+	"	return q;\n"
+	"}\n"
+	"inline float4 b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
+	"{\n"
+	"	b3Quat qInv = b3QuatInvert( q );\n"
+	"	float4 vcpy = vec;\n"
+	"	vcpy.w = 0.f;\n"
+	"	float4 out = b3QuatMul(b3QuatMul(q,vcpy),qInv);\n"
+	"	return out;\n"
+	"}\n"
+	"inline b3Quat b3QuatInverse(b3QuatConstArg q)\n"
+	"{\n"
+	"	return (b3Quat)(-q.xyz, q.w);\n"
+	"}\n"
+	"inline b3Quat b3QuatInvert(b3QuatConstArg q)\n"
+	"{\n"
+	"	return (b3Quat)(-q.xyz, q.w);\n"
+	"}\n"
+	"inline float4 b3QuatInvRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
+	"{\n"
+	"	return b3QuatRotate( b3QuatInvert( q ), vec );\n"
+	"}\n"
+	"inline b3Float4 b3TransformPoint(b3Float4ConstArg point, b3Float4ConstArg translation, b3QuatConstArg  orientation)\n"
+	"{\n"
+	"	return b3QuatRotate( orientation, point ) + (translation);\n"
+	"}\n"
+	"	\n"
+	"#endif \n"
+	"#endif //B3_QUAT_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"typedef struct\n"
+	"{\n"
+	"	b3Float4 m_row[3];\n"
+	"}b3Mat3x3;\n"
+	"#define b3Mat3x3ConstArg const b3Mat3x3\n"
+	"#define b3GetRow(m,row) (m.m_row[row])\n"
+	"inline b3Mat3x3 b3QuatGetRotationMatrix(b3Quat quat)\n"
+	"{\n"
+	"	b3Float4 quat2 = (b3Float4)(quat.x*quat.x, quat.y*quat.y, quat.z*quat.z, 0.f);\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0].x=1-2*quat2.y-2*quat2.z;\n"
+	"	out.m_row[0].y=2*quat.x*quat.y-2*quat.w*quat.z;\n"
+	"	out.m_row[0].z=2*quat.x*quat.z+2*quat.w*quat.y;\n"
+	"	out.m_row[0].w = 0.f;\n"
+	"	out.m_row[1].x=2*quat.x*quat.y+2*quat.w*quat.z;\n"
+	"	out.m_row[1].y=1-2*quat2.x-2*quat2.z;\n"
+	"	out.m_row[1].z=2*quat.y*quat.z-2*quat.w*quat.x;\n"
+	"	out.m_row[1].w = 0.f;\n"
+	"	out.m_row[2].x=2*quat.x*quat.z-2*quat.w*quat.y;\n"
+	"	out.m_row[2].y=2*quat.y*quat.z+2*quat.w*quat.x;\n"
+	"	out.m_row[2].z=1-2*quat2.x-2*quat2.y;\n"
+	"	out.m_row[2].w = 0.f;\n"
+	"	return out;\n"
+	"}\n"
+	"inline b3Mat3x3 b3AbsoluteMat3x3(b3Mat3x3ConstArg matIn)\n"
+	"{\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0] = fabs(matIn.m_row[0]);\n"
+	"	out.m_row[1] = fabs(matIn.m_row[1]);\n"
+	"	out.m_row[2] = fabs(matIn.m_row[2]);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtZero();\n"
+	"__inline\n"
+	"b3Mat3x3 mtIdentity();\n"
+	"__inline\n"
+	"b3Mat3x3 mtTranspose(b3Mat3x3 m);\n"
+	"__inline\n"
+	"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b);\n"
+	"__inline\n"
+	"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b);\n"
+	"__inline\n"
+	"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b);\n"
+	"__inline\n"
+	"b3Mat3x3 mtZero()\n"
+	"{\n"
+	"	b3Mat3x3 m;\n"
+	"	m.m_row[0] = (b3Float4)(0.f);\n"
+	"	m.m_row[1] = (b3Float4)(0.f);\n"
+	"	m.m_row[2] = (b3Float4)(0.f);\n"
+	"	return m;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtIdentity()\n"
+	"{\n"
+	"	b3Mat3x3 m;\n"
+	"	m.m_row[0] = (b3Float4)(1,0,0,0);\n"
+	"	m.m_row[1] = (b3Float4)(0,1,0,0);\n"
+	"	m.m_row[2] = (b3Float4)(0,0,1,0);\n"
+	"	return m;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtTranspose(b3Mat3x3 m)\n"
+	"{\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0] = (b3Float4)(m.m_row[0].x, m.m_row[1].x, m.m_row[2].x, 0.f);\n"
+	"	out.m_row[1] = (b3Float4)(m.m_row[0].y, m.m_row[1].y, m.m_row[2].y, 0.f);\n"
+	"	out.m_row[2] = (b3Float4)(m.m_row[0].z, m.m_row[1].z, m.m_row[2].z, 0.f);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b)\n"
+	"{\n"
+	"	b3Mat3x3 transB;\n"
+	"	transB = mtTranspose( b );\n"
+	"	b3Mat3x3 ans;\n"
+	"	//	why this doesn't run when 0ing in the for{}\n"
+	"	a.m_row[0].w = 0.f;\n"
+	"	a.m_row[1].w = 0.f;\n"
+	"	a.m_row[2].w = 0.f;\n"
+	"	for(int i=0; i<3; i++)\n"
+	"	{\n"
+	"//	a.m_row[i].w = 0.f;\n"
+	"		ans.m_row[i].x = b3Dot3F4(a.m_row[i],transB.m_row[0]);\n"
+	"		ans.m_row[i].y = b3Dot3F4(a.m_row[i],transB.m_row[1]);\n"
+	"		ans.m_row[i].z = b3Dot3F4(a.m_row[i],transB.m_row[2]);\n"
+	"		ans.m_row[i].w = 0.f;\n"
+	"	}\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b)\n"
+	"{\n"
+	"	b3Float4 ans;\n"
+	"	ans.x = b3Dot3F4( a.m_row[0], b );\n"
+	"	ans.y = b3Dot3F4( a.m_row[1], b );\n"
+	"	ans.z = b3Dot3F4( a.m_row[2], b );\n"
+	"	ans.w = 0.f;\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b)\n"
+	"{\n"
+	"	b3Float4 colx = b3MakeFloat4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
+	"	b3Float4 coly = b3MakeFloat4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
+	"	b3Float4 colz = b3MakeFloat4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
+	"	b3Float4 ans;\n"
+	"	ans.x = b3Dot3F4( a, colx );\n"
+	"	ans.y = b3Dot3F4( a, coly );\n"
+	"	ans.z = b3Dot3F4( a, colz );\n"
+	"	return ans;\n"
+	"}\n"
+	"#endif\n"
+	"#endif //B3_MAT3x3_H\n"
+	"typedef struct b3Aabb b3Aabb_t;\n"
+	"struct b3Aabb\n"
+	"{\n"
+	"	union\n"
+	"	{\n"
+	"		float m_min[4];\n"
+	"		b3Float4 m_minVec;\n"
+	"		int m_minIndices[4];\n"
+	"	};\n"
+	"	union\n"
+	"	{\n"
+	"		float	m_max[4];\n"
+	"		b3Float4 m_maxVec;\n"
+	"		int m_signedMaxIndices[4];\n"
+	"	};\n"
+	"};\n"
+	"inline void b3TransformAabb2(b3Float4ConstArg localAabbMin,b3Float4ConstArg localAabbMax, float margin,\n"
+	"						b3Float4ConstArg pos,\n"
+	"						b3QuatConstArg orn,\n"
+	"						b3Float4* aabbMinOut,b3Float4* aabbMaxOut)\n"
+	"{\n"
+	"		b3Float4 localHalfExtents = 0.5f*(localAabbMax-localAabbMin);\n"
+	"		localHalfExtents+=b3MakeFloat4(margin,margin,margin,0.f);\n"
+	"		b3Float4 localCenter = 0.5f*(localAabbMax+localAabbMin);\n"
+	"		b3Mat3x3 m;\n"
+	"		m = b3QuatGetRotationMatrix(orn);\n"
+	"		b3Mat3x3 abs_b = b3AbsoluteMat3x3(m);\n"
+	"		b3Float4 center = b3TransformPoint(localCenter,pos,orn);\n"
+	"		\n"
+	"		b3Float4 extent = b3MakeFloat4(b3Dot3F4(localHalfExtents,b3GetRow(abs_b,0)),\n"
+	"										 b3Dot3F4(localHalfExtents,b3GetRow(abs_b,1)),\n"
+	"										 b3Dot3F4(localHalfExtents,b3GetRow(abs_b,2)),\n"
+	"										 0.f);\n"
+	"		*aabbMinOut = center-extent;\n"
+	"		*aabbMaxOut = center+extent;\n"
+	"}\n"
+	"/// conservative test for overlap between two aabbs\n"
+	"inline bool b3TestAabbAgainstAabb(b3Float4ConstArg aabbMin1,b3Float4ConstArg aabbMax1,\n"
+	"								b3Float4ConstArg aabbMin2, b3Float4ConstArg aabbMax2)\n"
+	"{\n"
+	"	bool overlap = true;\n"
+	"	overlap = (aabbMin1.x > aabbMax2.x || aabbMax1.x < aabbMin2.x) ? false : overlap;\n"
+	"	overlap = (aabbMin1.z > aabbMax2.z || aabbMax1.z < aabbMin2.z) ? false : overlap;\n"
+	"	overlap = (aabbMin1.y > aabbMax2.y || aabbMax1.y < aabbMin2.y) ? false : overlap;\n"
+	"	return overlap;\n"
+	"}\n"
+	"#endif //B3_AABB_H\n"
+	"/*\n"
+	"Bullet Continuous Collision Detection and Physics Library\n"
+	"Copyright (c) 2003-2013 Erwin Coumans  http://bulletphysics.org\n"
+	"This software is provided 'as-is', without any express or implied warranty.\n"
+	"In no event will the authors be held liable for any damages arising from the use of this software.\n"
+	"Permission is granted to anyone to use this software for any purpose,\n"
+	"including commercial applications, and to alter it and redistribute it freely,\n"
+	"subject to the following restrictions:\n"
+	"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
+	"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
+	"3. This notice may not be removed or altered from any source distribution.\n"
+	"*/\n"
+	"#ifndef B3_INT2_H\n"
+	"#define B3_INT2_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#define b3UnsignedInt2 uint2\n"
+	"#define b3Int2 int2\n"
+	"#define b3MakeInt2 (int2)\n"
+	"#endif //__cplusplus\n"
+	"#endif\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_plane;\n"
+	"	int m_indexOffset;\n"
+	"	int m_numIndices;\n"
+	"} btGpuFace;\n"
+	"#define make_float4 (float4)\n"
+	"__inline\n"
+	"float4 cross3(float4 a, float4 b)\n"
+	"{\n"
+	"	return cross(a,b);\n"
+	"	\n"
+	"//	float4 a1 = make_float4(a.xyz,0.f);\n"
+	"//	float4 b1 = make_float4(b.xyz,0.f);\n"
+	"//	return cross(a1,b1);\n"
+	"//float4 c = make_float4(a.y*b.z - a.z*b.y,a.z*b.x - a.x*b.z,a.x*b.y - a.y*b.x,0.f);\n"
+	"	\n"
+	"	//	float4 c = make_float4(a.y*b.z - a.z*b.y,1.f,a.x*b.y - a.y*b.x,0.f);\n"
+	"	\n"
+	"	//return c;\n"
+	"}\n"
+	"__inline\n"
+	"float dot3F4(float4 a, float4 b)\n"
+	"{\n"
+	"	float4 a1 = make_float4(a.xyz,0.f);\n"
+	"	float4 b1 = make_float4(b.xyz,0.f);\n"
+	"	return dot(a1, b1);\n"
+	"}\n"
+	"__inline\n"
+	"float4 fastNormalize4(float4 v)\n"
+	"{\n"
+	"	v = make_float4(v.xyz,0.f);\n"
+	"	return fast_normalize(v);\n"
+	"}\n"
+	"///////////////////////////////////////\n"
+	"//	Quaternion\n"
+	"///////////////////////////////////////\n"
+	"typedef float4 Quaternion;\n"
+	"__inline\n"
+	"Quaternion qtMul(Quaternion a, Quaternion b);\n"
+	"__inline\n"
+	"Quaternion qtNormalize(Quaternion in);\n"
+	"__inline\n"
+	"float4 qtRotate(Quaternion q, float4 vec);\n"
+	"__inline\n"
+	"Quaternion qtInvert(Quaternion q);\n"
+	"__inline\n"
+	"Quaternion qtMul(Quaternion a, Quaternion b)\n"
+	"{\n"
+	"	Quaternion ans;\n"
+	"	ans = cross3( a, b );\n"
+	"	ans += a.w*b+b.w*a;\n"
+	"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
+	"	ans.w = a.w*b.w - dot3F4(a, b);\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"Quaternion qtNormalize(Quaternion in)\n"
+	"{\n"
+	"	return fastNormalize4(in);\n"
+	"//	in /= length( in );\n"
+	"//	return in;\n"
+	"}\n"
+	"__inline\n"
+	"float4 qtRotate(Quaternion q, float4 vec)\n"
+	"{\n"
+	"	Quaternion qInv = qtInvert( q );\n"
+	"	float4 vcpy = vec;\n"
+	"	vcpy.w = 0.f;\n"
+	"	float4 out = qtMul(qtMul(q,vcpy),qInv);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"Quaternion qtInvert(Quaternion q)\n"
+	"{\n"
+	"	return (Quaternion)(-q.xyz, q.w);\n"
+	"}\n"
+	"__inline\n"
+	"float4 qtInvRotate(const Quaternion q, float4 vec)\n"
+	"{\n"
+	"	return qtRotate( qtInvert( q ), vec );\n"
+	"}\n"
+	"__inline\n"
+	"float4 transform(const float4* p, const float4* translation, const Quaternion* orientation)\n"
+	"{\n"
+	"	return qtRotate( *orientation, *p ) + (*translation);\n"
+	"}\n"
+	"__inline\n"
+	"float4 normalize3(const float4 a)\n"
+	"{\n"
+	"	float4 n = make_float4(a.x, a.y, a.z, 0.f);\n"
+	"	return fastNormalize4( n );\n"
+	"}\n"
+	"inline void projectLocal(const ConvexPolyhedronCL* hull,  const float4 pos, const float4 orn, \n"
+	"const float4* dir, const float4* vertices, float* min, float* max)\n"
+	"{\n"
+	"	min[0] = FLT_MAX;\n"
+	"	max[0] = -FLT_MAX;\n"
+	"	int numVerts = hull->m_numVertices;\n"
+	"	const float4 localDir = qtInvRotate(orn,*dir);\n"
+	"	float offset = dot(pos,*dir);\n"
+	"	for(int i=0;i<numVerts;i++)\n"
+	"	{\n"
+	"		float dp = dot(vertices[hull->m_vertexOffset+i],localDir);\n"
+	"		if(dp < min[0])	\n"
+	"			min[0] = dp;\n"
+	"		if(dp > max[0])	\n"
+	"			max[0] = dp;\n"
+	"	}\n"
+	"	if(min[0]>max[0])\n"
+	"	{\n"
+	"		float tmp = min[0];\n"
+	"		min[0] = max[0];\n"
+	"		max[0] = tmp;\n"
+	"	}\n"
+	"	min[0] += offset;\n"
+	"	max[0] += offset;\n"
+	"}\n"
+	"inline void project(__global const ConvexPolyhedronCL* hull,  const float4 pos, const float4 orn, \n"
+	"const float4* dir, __global const float4* vertices, float* min, float* max)\n"
+	"{\n"
+	"	min[0] = FLT_MAX;\n"
+	"	max[0] = -FLT_MAX;\n"
+	"	int numVerts = hull->m_numVertices;\n"
+	"	const float4 localDir = qtInvRotate(orn,*dir);\n"
+	"	float offset = dot(pos,*dir);\n"
+	"	for(int i=0;i<numVerts;i++)\n"
+	"	{\n"
+	"		float dp = dot(vertices[hull->m_vertexOffset+i],localDir);\n"
+	"		if(dp < min[0])	\n"
+	"			min[0] = dp;\n"
+	"		if(dp > max[0])	\n"
+	"			max[0] = dp;\n"
+	"	}\n"
+	"	if(min[0]>max[0])\n"
+	"	{\n"
+	"		float tmp = min[0];\n"
+	"		min[0] = max[0];\n"
+	"		max[0] = tmp;\n"
+	"	}\n"
+	"	min[0] += offset;\n"
+	"	max[0] += offset;\n"
+	"}\n"
+	"inline bool TestSepAxisLocalA(const ConvexPolyhedronCL* hullA, __global const ConvexPolyhedronCL* hullB, \n"
+	"	const float4 posA,const float4 ornA,\n"
+	"	const float4 posB,const float4 ornB,\n"
+	"	float4* sep_axis, const float4* verticesA, __global const float4* verticesB,float* depth)\n"
+	"{\n"
+	"	float Min0,Max0;\n"
+	"	float Min1,Max1;\n"
+	"	projectLocal(hullA,posA,ornA,sep_axis,verticesA, &Min0, &Max0);\n"
+	"	project(hullB,posB,ornB, sep_axis,verticesB, &Min1, &Max1);\n"
+	"	if(Max0<Min1 || Max1<Min0)\n"
+	"		return false;\n"
+	"	float d0 = Max0 - Min1;\n"
+	"	float d1 = Max1 - Min0;\n"
+	"	*depth = d0<d1 ? d0:d1;\n"
+	"	return true;\n"
+	"}\n"
+	"inline bool IsAlmostZero(const float4 v)\n"
+	"{\n"
+	"	if(fabs(v.x)>1e-6f || fabs(v.y)>1e-6f || fabs(v.z)>1e-6f)\n"
+	"		return false;\n"
+	"	return true;\n"
+	"}\n"
+	"bool findSeparatingAxisLocalA(	const ConvexPolyhedronCL* hullA, __global const ConvexPolyhedronCL* hullB, \n"
+	"	const float4 posA1,\n"
+	"	const float4 ornA,\n"
+	"	const float4 posB1,\n"
+	"	const float4 ornB,\n"
+	"	const float4 DeltaC2,\n"
+	"	\n"
+	"	const float4* verticesA, \n"
+	"	const float4* uniqueEdgesA, \n"
+	"	const btGpuFace* facesA,\n"
+	"	const int*  indicesA,\n"
+	"	__global const float4* verticesB, \n"
+	"	__global const float4* uniqueEdgesB, \n"
+	"	__global const btGpuFace* facesB,\n"
+	"	__global const int*  indicesB,\n"
+	"	float4* sep,\n"
+	"	float* dmin)\n"
+	"{\n"
+	"	\n"
+	"	float4 posA = posA1;\n"
+	"	posA.w = 0.f;\n"
+	"	float4 posB = posB1;\n"
+	"	posB.w = 0.f;\n"
+	"	int curPlaneTests=0;\n"
+	"	{\n"
+	"		int numFacesA = hullA->m_numFaces;\n"
+	"		// Test normals from hullA\n"
+	"		for(int i=0;i<numFacesA;i++)\n"
+	"		{\n"
+	"			const float4 normal = facesA[hullA->m_faceOffset+i].m_plane;\n"
+	"			float4 faceANormalWS = qtRotate(ornA,normal);\n"
+	"			if (dot3F4(DeltaC2,faceANormalWS)<0)\n"
+	"				faceANormalWS*=-1.f;\n"
+	"			curPlaneTests++;\n"
+	"			float d;\n"
+	"			if(!TestSepAxisLocalA( hullA, hullB, posA,ornA,posB,ornB,&faceANormalWS, verticesA, verticesB,&d))\n"
+	"				return false;\n"
+	"			if(d<*dmin)\n"
+	"			{\n"
+	"				*dmin = d;\n"
+	"				*sep = faceANormalWS;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	if((dot3F4(-DeltaC2,*sep))>0.0f)\n"
+	"	{\n"
+	"		*sep = -(*sep);\n"
+	"	}\n"
+	"	return true;\n"
+	"}\n"
+	"bool findSeparatingAxisLocalB(	__global const ConvexPolyhedronCL* hullA,  const ConvexPolyhedronCL* hullB, \n"
+	"	const float4 posA1,\n"
+	"	const float4 ornA,\n"
+	"	const float4 posB1,\n"
+	"	const float4 ornB,\n"
+	"	const float4 DeltaC2,\n"
+	"	__global const float4* verticesA, \n"
+	"	__global const float4* uniqueEdgesA, \n"
+	"	__global const btGpuFace* facesA,\n"
+	"	__global const int*  indicesA,\n"
+	"	const float4* verticesB,\n"
+	"	const float4* uniqueEdgesB, \n"
+	"	const btGpuFace* facesB,\n"
+	"	const int*  indicesB,\n"
+	"	float4* sep,\n"
+	"	float* dmin)\n"
+	"{\n"
+	"	float4 posA = posA1;\n"
+	"	posA.w = 0.f;\n"
+	"	float4 posB = posB1;\n"
+	"	posB.w = 0.f;\n"
+	"	int curPlaneTests=0;\n"
+	"	{\n"
+	"		int numFacesA = hullA->m_numFaces;\n"
+	"		// Test normals from hullA\n"
+	"		for(int i=0;i<numFacesA;i++)\n"
+	"		{\n"
+	"			const float4 normal = facesA[hullA->m_faceOffset+i].m_plane;\n"
+	"			float4 faceANormalWS = qtRotate(ornA,normal);\n"
+	"			if (dot3F4(DeltaC2,faceANormalWS)<0)\n"
+	"				faceANormalWS *= -1.f;\n"
+	"			curPlaneTests++;\n"
+	"			float d;\n"
+	"			if(!TestSepAxisLocalA( hullB, hullA, posB,ornB,posA,ornA, &faceANormalWS, verticesB,verticesA, &d))\n"
+	"				return false;\n"
+	"			if(d<*dmin)\n"
+	"			{\n"
+	"				*dmin = d;\n"
+	"				*sep = faceANormalWS;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	if((dot3F4(-DeltaC2,*sep))>0.0f)\n"
+	"	{\n"
+	"		*sep = -(*sep);\n"
+	"	}\n"
+	"	return true;\n"
+	"}\n"
+	"bool findSeparatingAxisEdgeEdgeLocalA(	const ConvexPolyhedronCL* hullA, __global const ConvexPolyhedronCL* hullB, \n"
+	"	const float4 posA1,\n"
+	"	const float4 ornA,\n"
+	"	const float4 posB1,\n"
+	"	const float4 ornB,\n"
+	"	const float4 DeltaC2,\n"
+	"	const float4* verticesA, \n"
+	"	const float4* uniqueEdgesA, \n"
+	"	const btGpuFace* facesA,\n"
+	"	const int*  indicesA,\n"
+	"	__global const float4* verticesB, \n"
+	"	__global const float4* uniqueEdgesB, \n"
+	"	__global const btGpuFace* facesB,\n"
+	"	__global const int*  indicesB,\n"
+	"		float4* sep,\n"
+	"	float* dmin)\n"
+	"{\n"
+	"	float4 posA = posA1;\n"
+	"	posA.w = 0.f;\n"
+	"	float4 posB = posB1;\n"
+	"	posB.w = 0.f;\n"
+	"	int curPlaneTests=0;\n"
+	"	int curEdgeEdge = 0;\n"
+	"	// Test edges\n"
+	"	for(int e0=0;e0<hullA->m_numUniqueEdges;e0++)\n"
+	"	{\n"
+	"		const float4 edge0 = uniqueEdgesA[hullA->m_uniqueEdgesOffset+e0];\n"
+	"		float4 edge0World = qtRotate(ornA,edge0);\n"
+	"		for(int e1=0;e1<hullB->m_numUniqueEdges;e1++)\n"
+	"		{\n"
+	"			const float4 edge1 = uniqueEdgesB[hullB->m_uniqueEdgesOffset+e1];\n"
+	"			float4 edge1World = qtRotate(ornB,edge1);\n"
+	"			float4 crossje = cross3(edge0World,edge1World);\n"
+	"			curEdgeEdge++;\n"
+	"			if(!IsAlmostZero(crossje))\n"
+	"			{\n"
+	"				crossje = normalize3(crossje);\n"
+	"				if (dot3F4(DeltaC2,crossje)<0)\n"
+	"					crossje *= -1.f;\n"
+	"				float dist;\n"
+	"				bool result = true;\n"
+	"				{\n"
+	"					float Min0,Max0;\n"
+	"					float Min1,Max1;\n"
+	"					projectLocal(hullA,posA,ornA,&crossje,verticesA, &Min0, &Max0);\n"
+	"					project(hullB,posB,ornB,&crossje,verticesB, &Min1, &Max1);\n"
+	"				\n"
+	"					if(Max0<Min1 || Max1<Min0)\n"
+	"						result = false;\n"
+	"				\n"
+	"					float d0 = Max0 - Min1;\n"
+	"					float d1 = Max1 - Min0;\n"
+	"					dist = d0<d1 ? d0:d1;\n"
+	"					result = true;\n"
+	"				}\n"
+	"				\n"
+	"				if(dist<*dmin)\n"
+	"				{\n"
+	"					*dmin = dist;\n"
+	"					*sep = crossje;\n"
+	"				}\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	\n"
+	"	if((dot3F4(-DeltaC2,*sep))>0.0f)\n"
+	"	{\n"
+	"		*sep = -(*sep);\n"
+	"	}\n"
+	"	return true;\n"
+	"}\n"
+	"inline bool TestSepAxis(__global const ConvexPolyhedronCL* hullA, __global const ConvexPolyhedronCL* hullB, \n"
+	"	const float4 posA,const float4 ornA,\n"
+	"	const float4 posB,const float4 ornB,\n"
+	"	float4* sep_axis, __global const float4* vertices,float* depth)\n"
+	"{\n"
+	"	float Min0,Max0;\n"
+	"	float Min1,Max1;\n"
+	"	project(hullA,posA,ornA,sep_axis,vertices, &Min0, &Max0);\n"
+	"	project(hullB,posB,ornB, sep_axis,vertices, &Min1, &Max1);\n"
+	"	if(Max0<Min1 || Max1<Min0)\n"
+	"		return false;\n"
+	"	float d0 = Max0 - Min1;\n"
+	"	float d1 = Max1 - Min0;\n"
+	"	*depth = d0<d1 ? d0:d1;\n"
+	"	return true;\n"
+	"}\n"
+	"bool findSeparatingAxis(	__global const ConvexPolyhedronCL* hullA, __global const ConvexPolyhedronCL* hullB, \n"
+	"	const float4 posA1,\n"
+	"	const float4 ornA,\n"
+	"	const float4 posB1,\n"
+	"	const float4 ornB,\n"
+	"	const float4 DeltaC2,\n"
+	"	__global const float4* vertices, \n"
+	"	__global const float4* uniqueEdges, \n"
+	"	__global const btGpuFace* faces,\n"
+	"	__global const int*  indices,\n"
+	"	float4* sep,\n"
+	"	float* dmin)\n"
+	"{\n"
+	"	\n"
+	"	float4 posA = posA1;\n"
+	"	posA.w = 0.f;\n"
+	"	float4 posB = posB1;\n"
+	"	posB.w = 0.f;\n"
+	"	\n"
+	"	int curPlaneTests=0;\n"
+	"	{\n"
+	"		int numFacesA = hullA->m_numFaces;\n"
+	"		// Test normals from hullA\n"
+	"		for(int i=0;i<numFacesA;i++)\n"
+	"		{\n"
+	"			const float4 normal = faces[hullA->m_faceOffset+i].m_plane;\n"
+	"			float4 faceANormalWS = qtRotate(ornA,normal);\n"
+	"	\n"
+	"			if (dot3F4(DeltaC2,faceANormalWS)<0)\n"
+	"				faceANormalWS*=-1.f;\n"
+	"				\n"
+	"			curPlaneTests++;\n"
+	"	\n"
+	"			float d;\n"
+	"			if(!TestSepAxis( hullA, hullB, posA,ornA,posB,ornB,&faceANormalWS, vertices,&d))\n"
+	"				return false;\n"
+	"	\n"
+	"			if(d<*dmin)\n"
+	"			{\n"
+	"				*dmin = d;\n"
+	"				*sep = faceANormalWS;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"		if((dot3F4(-DeltaC2,*sep))>0.0f)\n"
+	"		{\n"
+	"			*sep = -(*sep);\n"
+	"		}\n"
+	"	\n"
+	"	return true;\n"
+	"}\n"
+	"bool findSeparatingAxisUnitSphere(	__global const ConvexPolyhedronCL* hullA, __global const ConvexPolyhedronCL* hullB, \n"
+	"	const float4 posA1,\n"
+	"	const float4 ornA,\n"
+	"	const float4 posB1,\n"
+	"	const float4 ornB,\n"
+	"	const float4 DeltaC2,\n"
+	"	__global const float4* vertices,\n"
+	"	__global const float4* unitSphereDirections,\n"
+	"	int numUnitSphereDirections,\n"
+	"	float4* sep,\n"
+	"	float* dmin)\n"
+	"{\n"
+	"	\n"
+	"	float4 posA = posA1;\n"
+	"	posA.w = 0.f;\n"
+	"	float4 posB = posB1;\n"
+	"	posB.w = 0.f;\n"
+	"	int curPlaneTests=0;\n"
+	"	int curEdgeEdge = 0;\n"
+	"	// Test unit sphere directions\n"
+	"	for (int i=0;i<numUnitSphereDirections;i++)\n"
+	"	{\n"
+	"		float4 crossje;\n"
+	"		crossje = unitSphereDirections[i];	\n"
+	"		if (dot3F4(DeltaC2,crossje)>0)\n"
+	"			crossje *= -1.f;\n"
+	"		{\n"
+	"			float dist;\n"
+	"			bool result = true;\n"
+	"			float Min0,Max0;\n"
+	"			float Min1,Max1;\n"
+	"			project(hullA,posA,ornA,&crossje,vertices, &Min0, &Max0);\n"
+	"			project(hullB,posB,ornB,&crossje,vertices, &Min1, &Max1);\n"
+	"		\n"
+	"			if(Max0<Min1 || Max1<Min0)\n"
+	"				return false;\n"
+	"		\n"
+	"			float d0 = Max0 - Min1;\n"
+	"			float d1 = Max1 - Min0;\n"
+	"			dist = d0<d1 ? d0:d1;\n"
+	"			result = true;\n"
+	"	\n"
+	"			if(dist<*dmin)\n"
+	"			{\n"
+	"				*dmin = dist;\n"
+	"				*sep = crossje;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	\n"
+	"	if((dot3F4(-DeltaC2,*sep))>0.0f)\n"
+	"	{\n"
+	"		*sep = -(*sep);\n"
+	"	}\n"
+	"	return true;\n"
+	"}\n"
+	"bool findSeparatingAxisEdgeEdge(	__global const ConvexPolyhedronCL* hullA, __global const ConvexPolyhedronCL* hullB, \n"
+	"	const float4 posA1,\n"
+	"	const float4 ornA,\n"
+	"	const float4 posB1,\n"
+	"	const float4 ornB,\n"
+	"	const float4 DeltaC2,\n"
+	"	__global const float4* vertices, \n"
+	"	__global const float4* uniqueEdges, \n"
+	"	__global const btGpuFace* faces,\n"
+	"	__global const int*  indices,\n"
+	"	float4* sep,\n"
+	"	float* dmin)\n"
+	"{\n"
+	"	\n"
+	"	float4 posA = posA1;\n"
+	"	posA.w = 0.f;\n"
+	"	float4 posB = posB1;\n"
+	"	posB.w = 0.f;\n"
+	"	int curPlaneTests=0;\n"
+	"	int curEdgeEdge = 0;\n"
+	"	// Test edges\n"
+	"	for(int e0=0;e0<hullA->m_numUniqueEdges;e0++)\n"
+	"	{\n"
+	"		const float4 edge0 = uniqueEdges[hullA->m_uniqueEdgesOffset+e0];\n"
+	"		float4 edge0World = qtRotate(ornA,edge0);\n"
+	"		for(int e1=0;e1<hullB->m_numUniqueEdges;e1++)\n"
+	"		{\n"
+	"			const float4 edge1 = uniqueEdges[hullB->m_uniqueEdgesOffset+e1];\n"
+	"			float4 edge1World = qtRotate(ornB,edge1);\n"
+	"			float4 crossje = cross3(edge0World,edge1World);\n"
+	"			curEdgeEdge++;\n"
+	"			if(!IsAlmostZero(crossje))\n"
+	"			{\n"
+	"				crossje = normalize3(crossje);\n"
+	"				if (dot3F4(DeltaC2,crossje)<0)\n"
+	"					crossje*=-1.f;\n"
+	"					\n"
+	"				float dist;\n"
+	"				bool result = true;\n"
+	"				{\n"
+	"					float Min0,Max0;\n"
+	"					float Min1,Max1;\n"
+	"					project(hullA,posA,ornA,&crossje,vertices, &Min0, &Max0);\n"
+	"					project(hullB,posB,ornB,&crossje,vertices, &Min1, &Max1);\n"
+	"				\n"
+	"					if(Max0<Min1 || Max1<Min0)\n"
+	"						return false;\n"
+	"				\n"
+	"					float d0 = Max0 - Min1;\n"
+	"					float d1 = Max1 - Min0;\n"
+	"					dist = d0<d1 ? d0:d1;\n"
+	"					result = true;\n"
+	"				}\n"
+	"				\n"
+	"				if(dist<*dmin)\n"
+	"				{\n"
+	"					*dmin = dist;\n"
+	"					*sep = crossje;\n"
+	"				}\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	\n"
+	"	if((dot3F4(-DeltaC2,*sep))>0.0f)\n"
+	"	{\n"
+	"		*sep = -(*sep);\n"
+	"	}\n"
+	"	return true;\n"
+	"}\n"
+	"// work-in-progress\n"
+	"__kernel void   processCompoundPairsKernel( __global const int4* gpuCompoundPairs,\n"
+	"																					__global const BodyData* rigidBodies, \n"
+	"																					__global const btCollidableGpu* collidables,\n"
+	"																					__global const ConvexPolyhedronCL* convexShapes, \n"
+	"																					__global const float4* vertices,\n"
+	"																					__global const float4* uniqueEdges,\n"
+	"																					__global const btGpuFace* faces,\n"
+	"																					__global const int* indices,\n"
+	"																					__global btAabbCL* aabbs,\n"
+	"																					__global const btGpuChildShape* gpuChildShapes,\n"
+	"																					__global volatile float4* gpuCompoundSepNormalsOut,\n"
+	"																					__global volatile int* gpuHasCompoundSepNormalsOut,\n"
+	"																					int numCompoundPairs\n"
+	"																					)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	if (i<numCompoundPairs)\n"
+	"	{\n"
+	"		int bodyIndexA = gpuCompoundPairs[i].x;\n"
+	"		int bodyIndexB = gpuCompoundPairs[i].y;\n"
+	"		int childShapeIndexA = gpuCompoundPairs[i].z;\n"
+	"		int childShapeIndexB = gpuCompoundPairs[i].w;\n"
+	"		\n"
+	"		int collidableIndexA = -1;\n"
+	"		int collidableIndexB = -1;\n"
+	"		\n"
+	"		float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
+	"		float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
+	"		\n"
+	"		float4 ornB = rigidBodies[bodyIndexB].m_quat;\n"
+	"		float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
+	"							\n"
+	"		if (childShapeIndexA >= 0)\n"
+	"		{\n"
+	"			collidableIndexA = gpuChildShapes[childShapeIndexA].m_shapeIndex;\n"
+	"			float4 childPosA = gpuChildShapes[childShapeIndexA].m_childPosition;\n"
+	"			float4 childOrnA = gpuChildShapes[childShapeIndexA].m_childOrientation;\n"
+	"			float4 newPosA = qtRotate(ornA,childPosA)+posA;\n"
+	"			float4 newOrnA = qtMul(ornA,childOrnA);\n"
+	"			posA = newPosA;\n"
+	"			ornA = newOrnA;\n"
+	"		} else\n"
+	"		{\n"
+	"			collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
+	"		}\n"
+	"		\n"
+	"		if (childShapeIndexB>=0)\n"
+	"		{\n"
+	"			collidableIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;\n"
+	"			float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;\n"
+	"			float4 childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;\n"
+	"			float4 newPosB = transform(&childPosB,&posB,&ornB);\n"
+	"			float4 newOrnB = qtMul(ornB,childOrnB);\n"
+	"			posB = newPosB;\n"
+	"			ornB = newOrnB;\n"
+	"		} else\n"
+	"		{\n"
+	"			collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;	\n"
+	"		}\n"
+	"	\n"
+	"		gpuHasCompoundSepNormalsOut[i] = 0;\n"
+	"	\n"
+	"		int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
+	"		int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
+	"	\n"
+	"		int shapeTypeA = collidables[collidableIndexA].m_shapeType;\n"
+	"		int shapeTypeB = collidables[collidableIndexB].m_shapeType;\n"
+	"	\n"
+	"		if ((shapeTypeA != SHAPE_CONVEX_HULL) || (shapeTypeB != SHAPE_CONVEX_HULL))\n"
+	"		{\n"
+	"			return;\n"
+	"		}\n"
+	"		int hasSeparatingAxis = 5;\n"
+	"							\n"
+	"		int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
+	"		float dmin = FLT_MAX;\n"
+	"		posA.w = 0.f;\n"
+	"		posB.w = 0.f;\n"
+	"		float4 c0local = convexShapes[shapeIndexA].m_localCenter;\n"
+	"		float4 c0 = transform(&c0local, &posA, &ornA);\n"
+	"		float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
+	"		float4 c1 = transform(&c1local,&posB,&ornB);\n"
+	"		const float4 DeltaC2 = c0 - c1;\n"
+	"		float4 sepNormal = make_float4(1,0,0,0);\n"
+	"		bool sepA = findSeparatingAxis(	&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],posA,ornA,posB,ornB,DeltaC2,vertices,uniqueEdges,faces,indices,&sepNormal,&dmin);\n"
+	"		hasSeparatingAxis = 4;\n"
+	"		if (!sepA)\n"
+	"		{\n"
+	"			hasSeparatingAxis = 0;\n"
+	"		} else\n"
+	"		{\n"
+	"			bool sepB = findSeparatingAxis(	&convexShapes[shapeIndexB],&convexShapes[shapeIndexA],posB,ornB,posA,ornA,DeltaC2,vertices,uniqueEdges,faces,indices,&sepNormal,&dmin);\n"
+	"			if (!sepB)\n"
+	"			{\n"
+	"				hasSeparatingAxis = 0;\n"
+	"			} else//(!sepB)\n"
+	"			{\n"
+	"				bool sepEE = findSeparatingAxisEdgeEdge(	&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],posA,ornA,posB,ornB,DeltaC2,vertices,uniqueEdges,faces,indices,&sepNormal,&dmin);\n"
+	"				if (sepEE)\n"
+	"				{\n"
+	"						gpuCompoundSepNormalsOut[i] = sepNormal;//fastNormalize4(sepNormal);\n"
+	"						gpuHasCompoundSepNormalsOut[i] = 1;\n"
+	"				}//sepEE\n"
+	"			}//(!sepB)\n"
+	"		}//(!sepA)\n"
+	"		\n"
+	"		\n"
+	"	}\n"
+	"		\n"
+	"}\n"
+	"inline b3Float4 MyUnQuantize(const unsigned short* vecIn, b3Float4 quantization, b3Float4 bvhAabbMin)\n"
+	"{\n"
+	"		b3Float4 vecOut;\n"
+	"		vecOut = b3MakeFloat4(\n"
+	"			(float)(vecIn[0]) / (quantization.x),\n"
+	"			(float)(vecIn[1]) / (quantization.y),\n"
+	"			(float)(vecIn[2]) / (quantization.z),\n"
+	"			0.f);\n"
+	"		vecOut += bvhAabbMin;\n"
+	"		return vecOut;\n"
+	"}\n"
+	"inline b3Float4 MyUnQuantizeGlobal(__global const unsigned short* vecIn, b3Float4 quantization, b3Float4 bvhAabbMin)\n"
+	"{\n"
+	"		b3Float4 vecOut;\n"
+	"		vecOut = b3MakeFloat4(\n"
+	"			(float)(vecIn[0]) / (quantization.x),\n"
+	"			(float)(vecIn[1]) / (quantization.y),\n"
+	"			(float)(vecIn[2]) / (quantization.z),\n"
+	"			0.f);\n"
+	"		vecOut += bvhAabbMin;\n"
+	"		return vecOut;\n"
+	"}\n"
+	"// work-in-progress\n"
+	"__kernel void   findCompoundPairsKernel( __global const int4* pairs, \n"
+	"	__global const BodyData* rigidBodies, \n"
+	"	__global const btCollidableGpu* collidables,\n"
+	"	__global const ConvexPolyhedronCL* convexShapes, \n"
+	"	__global const float4* vertices,\n"
+	"	__global const float4* uniqueEdges,\n"
+	"	__global const btGpuFace* faces,\n"
+	"	__global const int* indices,\n"
+	"	__global b3Aabb_t* aabbLocalSpace,\n"
+	"	__global const btGpuChildShape* gpuChildShapes,\n"
+	"	__global volatile int4* gpuCompoundPairsOut,\n"
+	"	__global volatile int* numCompoundPairsOut,\n"
+	"	__global const b3BvhSubtreeInfo* subtrees,\n"
+	"	__global const b3QuantizedBvhNode* quantizedNodes,\n"
+	"	__global const b3BvhInfo* bvhInfos,\n"
+	"	int numPairs,\n"
+	"	int maxNumCompoundPairsCapacity\n"
+	"	)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	if (i<numPairs)\n"
+	"	{\n"
+	"		int bodyIndexA = pairs[i].x;\n"
+	"		int bodyIndexB = pairs[i].y;\n"
+	"		int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
+	"		int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
+	"		int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
+	"		int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
+	"		//once the broadphase avoids static-static pairs, we can remove this test\n"
+	"		if ((rigidBodies[bodyIndexA].m_invMass==0) &&(rigidBodies[bodyIndexB].m_invMass==0))\n"
+	"		{\n"
+	"			return;\n"
+	"		}\n"
+	"		if ((collidables[collidableIndexA].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS) &&(collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS))\n"
+	"		{\n"
+	"			int bvhA = collidables[collidableIndexA].m_compoundBvhIndex;\n"
+	"			int bvhB = collidables[collidableIndexB].m_compoundBvhIndex;\n"
+	"			int numSubTreesA = bvhInfos[bvhA].m_numSubTrees;\n"
+	"			int subTreesOffsetA = bvhInfos[bvhA].m_subTreeOffset;\n"
+	"			int subTreesOffsetB = bvhInfos[bvhB].m_subTreeOffset;\n"
+	"			int numSubTreesB = bvhInfos[bvhB].m_numSubTrees;\n"
+	"			\n"
+	"			float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
+	"			b3Quat ornA = rigidBodies[bodyIndexA].m_quat;\n"
+	"			b3Quat ornB = rigidBodies[bodyIndexB].m_quat;\n"
+	"			float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
+	"			\n"
+	"			for (int p=0;p<numSubTreesA;p++)\n"
+	"			{\n"
+	"				b3BvhSubtreeInfo subtreeA = subtrees[subTreesOffsetA+p];\n"
+	"				//bvhInfos[bvhA].m_quantization\n"
+	"				b3Float4 treeAminLocal = MyUnQuantize(subtreeA.m_quantizedAabbMin,bvhInfos[bvhA].m_quantization,bvhInfos[bvhA].m_aabbMin);\n"
+	"				b3Float4 treeAmaxLocal = MyUnQuantize(subtreeA.m_quantizedAabbMax,bvhInfos[bvhA].m_quantization,bvhInfos[bvhA].m_aabbMin);\n"
+	"				b3Float4 aabbAMinOut,aabbAMaxOut;\n"
+	"				float margin=0.f;\n"
+	"				b3TransformAabb2(treeAminLocal,treeAmaxLocal, margin,posA,ornA,&aabbAMinOut,&aabbAMaxOut);\n"
+	"				\n"
+	"				for (int q=0;q<numSubTreesB;q++)\n"
+	"				{\n"
+	"					b3BvhSubtreeInfo subtreeB = subtrees[subTreesOffsetB+q];\n"
+	"					b3Float4 treeBminLocal = MyUnQuantize(subtreeB.m_quantizedAabbMin,bvhInfos[bvhB].m_quantization,bvhInfos[bvhB].m_aabbMin);\n"
+	"					b3Float4 treeBmaxLocal = MyUnQuantize(subtreeB.m_quantizedAabbMax,bvhInfos[bvhB].m_quantization,bvhInfos[bvhB].m_aabbMin);\n"
+	"					b3Float4 aabbBMinOut,aabbBMaxOut;\n"
+	"					float margin=0.f;\n"
+	"					b3TransformAabb2(treeBminLocal,treeBmaxLocal, margin,posB,ornB,&aabbBMinOut,&aabbBMaxOut);\n"
+	"					\n"
+	"					\n"
+	"					bool aabbOverlap = b3TestAabbAgainstAabb(aabbAMinOut,aabbAMaxOut,aabbBMinOut,aabbBMaxOut);\n"
+	"					if (aabbOverlap)\n"
+	"					{\n"
+	"						\n"
+	"						int startNodeIndexA = subtreeA.m_rootNodeIndex+bvhInfos[bvhA].m_nodeOffset;\n"
+	"						int endNodeIndexA = startNodeIndexA+subtreeA.m_subtreeSize;\n"
+	"						int startNodeIndexB = subtreeB.m_rootNodeIndex+bvhInfos[bvhB].m_nodeOffset;\n"
+	"						int endNodeIndexB = startNodeIndexB+subtreeB.m_subtreeSize;\n"
+	"						b3Int2 nodeStack[B3_MAX_STACK_DEPTH];\n"
+	"						b3Int2 node0;\n"
+	"						node0.x = startNodeIndexA;\n"
+	"						node0.y = startNodeIndexB;\n"
+	"						int maxStackDepth = B3_MAX_STACK_DEPTH;\n"
+	"						int depth=0;\n"
+	"						nodeStack[depth++]=node0;\n"
+	"						do\n"
+	"						{\n"
+	"							b3Int2 node = nodeStack[--depth];\n"
+	"							b3Float4 aMinLocal = MyUnQuantizeGlobal(quantizedNodes[node.x].m_quantizedAabbMin,bvhInfos[bvhA].m_quantization,bvhInfos[bvhA].m_aabbMin);\n"
+	"							b3Float4 aMaxLocal = MyUnQuantizeGlobal(quantizedNodes[node.x].m_quantizedAabbMax,bvhInfos[bvhA].m_quantization,bvhInfos[bvhA].m_aabbMin);\n"
+	"							b3Float4 bMinLocal = MyUnQuantizeGlobal(quantizedNodes[node.y].m_quantizedAabbMin,bvhInfos[bvhB].m_quantization,bvhInfos[bvhB].m_aabbMin);\n"
+	"							b3Float4 bMaxLocal = MyUnQuantizeGlobal(quantizedNodes[node.y].m_quantizedAabbMax,bvhInfos[bvhB].m_quantization,bvhInfos[bvhB].m_aabbMin);\n"
+	"							float margin=0.f;\n"
+	"							b3Float4 aabbAMinOut,aabbAMaxOut;\n"
+	"							b3TransformAabb2(aMinLocal,aMaxLocal, margin,posA,ornA,&aabbAMinOut,&aabbAMaxOut);\n"
+	"							b3Float4 aabbBMinOut,aabbBMaxOut;\n"
+	"							b3TransformAabb2(bMinLocal,bMaxLocal, margin,posB,ornB,&aabbBMinOut,&aabbBMaxOut);\n"
+	"							\n"
+	"							bool nodeOverlap = b3TestAabbAgainstAabb(aabbAMinOut,aabbAMaxOut,aabbBMinOut,aabbBMaxOut);\n"
+	"							if (nodeOverlap)\n"
+	"							{\n"
+	"								bool isLeafA = isLeafNodeGlobal(&quantizedNodes[node.x]);\n"
+	"								bool isLeafB = isLeafNodeGlobal(&quantizedNodes[node.y]);\n"
+	"								bool isInternalA = !isLeafA;\n"
+	"								bool isInternalB = !isLeafB;\n"
+	"								//fail, even though it might hit two leaf nodes\n"
+	"								if (depth+4>maxStackDepth && !(isLeafA && isLeafB))\n"
+	"								{\n"
+	"									//printf(\"Error: traversal exceeded maxStackDepth\");\n"
+	"									continue;\n"
+	"								}\n"
+	"								if(isInternalA)\n"
+	"								{\n"
+	"									int nodeAleftChild = node.x+1;\n"
+	"									bool isNodeALeftChildLeaf = isLeafNodeGlobal(&quantizedNodes[node.x+1]);\n"
+	"									int nodeArightChild = isNodeALeftChildLeaf? node.x+2 : node.x+1 + getEscapeIndexGlobal(&quantizedNodes[node.x+1]);\n"
+	"									if(isInternalB)\n"
+	"									{					\n"
+	"										int nodeBleftChild = node.y+1;\n"
+	"										bool isNodeBLeftChildLeaf = isLeafNodeGlobal(&quantizedNodes[node.y+1]);\n"
+	"										int nodeBrightChild = isNodeBLeftChildLeaf? node.y+2 : node.y+1 + getEscapeIndexGlobal(&quantizedNodes[node.y+1]);\n"
+	"										nodeStack[depth++] = b3MakeInt2(nodeAleftChild, nodeBleftChild);\n"
+	"										nodeStack[depth++] = b3MakeInt2(nodeArightChild, nodeBleftChild);\n"
+	"										nodeStack[depth++] = b3MakeInt2(nodeAleftChild, nodeBrightChild);\n"
+	"										nodeStack[depth++] = b3MakeInt2(nodeArightChild, nodeBrightChild);\n"
+	"									}\n"
+	"									else\n"
+	"									{\n"
+	"										nodeStack[depth++] = b3MakeInt2(nodeAleftChild,node.y);\n"
+	"										nodeStack[depth++] = b3MakeInt2(nodeArightChild,node.y);\n"
+	"									}\n"
+	"								}\n"
+	"								else\n"
+	"								{\n"
+	"									if(isInternalB)\n"
+	"									{\n"
+	"										int nodeBleftChild = node.y+1;\n"
+	"										bool isNodeBLeftChildLeaf = isLeafNodeGlobal(&quantizedNodes[node.y+1]);\n"
+	"										int nodeBrightChild = isNodeBLeftChildLeaf? node.y+2 : node.y+1 + getEscapeIndexGlobal(&quantizedNodes[node.y+1]);\n"
+	"										nodeStack[depth++] = b3MakeInt2(node.x,nodeBleftChild);\n"
+	"										nodeStack[depth++] = b3MakeInt2(node.x,nodeBrightChild);\n"
+	"									}\n"
+	"									else\n"
+	"									{\n"
+	"										int compoundPairIdx = atomic_inc(numCompoundPairsOut);\n"
+	"										if (compoundPairIdx<maxNumCompoundPairsCapacity)\n"
+	"										{\n"
+	"											int childShapeIndexA = getTriangleIndexGlobal(&quantizedNodes[node.x]);\n"
+	"											int childShapeIndexB = getTriangleIndexGlobal(&quantizedNodes[node.y]);\n"
+	"											gpuCompoundPairsOut[compoundPairIdx]  = (int4)(bodyIndexA,bodyIndexB,childShapeIndexA,childShapeIndexB);\n"
+	"										}\n"
+	"									}\n"
+	"								}\n"
+	"							}\n"
+	"						} while (depth);\n"
+	"					}\n"
+	"				}\n"
+	"			}\n"
+	"			\n"
+	"			return;\n"
+	"		}\n"
+	"		if ((collidables[collidableIndexA].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS) ||(collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS))\n"
+	"		{\n"
+	"			if (collidables[collidableIndexA].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS) \n"
+	"			{\n"
+	"				int numChildrenA = collidables[collidableIndexA].m_numChildShapes;\n"
+	"				for (int c=0;c<numChildrenA;c++)\n"
+	"				{\n"
+	"					int childShapeIndexA = collidables[collidableIndexA].m_shapeIndex+c;\n"
+	"					int childColIndexA = gpuChildShapes[childShapeIndexA].m_shapeIndex;\n"
+	"					float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
+	"					float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
+	"					float4 childPosA = gpuChildShapes[childShapeIndexA].m_childPosition;\n"
+	"					float4 childOrnA = gpuChildShapes[childShapeIndexA].m_childOrientation;\n"
+	"					float4 newPosA = qtRotate(ornA,childPosA)+posA;\n"
+	"					float4 newOrnA = qtMul(ornA,childOrnA);\n"
+	"					int shapeIndexA = collidables[childColIndexA].m_shapeIndex;\n"
+	"					b3Aabb_t aabbAlocal = aabbLocalSpace[shapeIndexA];\n"
+	"					float margin = 0.f;\n"
+	"					\n"
+	"					b3Float4 aabbAMinWS;\n"
+	"					b3Float4 aabbAMaxWS;\n"
+	"					\n"
+	"					b3TransformAabb2(aabbAlocal.m_minVec,aabbAlocal.m_maxVec,margin,\n"
+	"						newPosA,\n"
+	"						newOrnA,\n"
+	"						&aabbAMinWS,&aabbAMaxWS);\n"
+	"						\n"
+	"					\n"
+	"					if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
+	"					{\n"
+	"						int numChildrenB = collidables[collidableIndexB].m_numChildShapes;\n"
+	"						for (int b=0;b<numChildrenB;b++)\n"
+	"						{\n"
+	"							int childShapeIndexB = collidables[collidableIndexB].m_shapeIndex+b;\n"
+	"							int childColIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;\n"
+	"							float4 ornB = rigidBodies[bodyIndexB].m_quat;\n"
+	"							float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
+	"							float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;\n"
+	"							float4 childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;\n"
+	"							float4 newPosB = transform(&childPosB,&posB,&ornB);\n"
+	"							float4 newOrnB = qtMul(ornB,childOrnB);\n"
+	"							int shapeIndexB = collidables[childColIndexB].m_shapeIndex;\n"
+	"							b3Aabb_t aabbBlocal = aabbLocalSpace[shapeIndexB];\n"
+	"							\n"
+	"							b3Float4 aabbBMinWS;\n"
+	"							b3Float4 aabbBMaxWS;\n"
+	"							\n"
+	"							b3TransformAabb2(aabbBlocal.m_minVec,aabbBlocal.m_maxVec,margin,\n"
+	"								newPosB,\n"
+	"								newOrnB,\n"
+	"								&aabbBMinWS,&aabbBMaxWS);\n"
+	"								\n"
+	"								\n"
+	"							\n"
+	"							bool aabbOverlap = b3TestAabbAgainstAabb(aabbAMinWS,aabbAMaxWS,aabbBMinWS,aabbBMaxWS);\n"
+	"							if (aabbOverlap)\n"
+	"							{\n"
+	"								int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
+	"								float dmin = FLT_MAX;\n"
+	"								float4 posA = newPosA;\n"
+	"								posA.w = 0.f;\n"
+	"								float4 posB = newPosB;\n"
+	"								posB.w = 0.f;\n"
+	"								float4 c0local = convexShapes[shapeIndexA].m_localCenter;\n"
+	"								float4 ornA = newOrnA;\n"
+	"								float4 c0 = transform(&c0local, &posA, &ornA);\n"
+	"								float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
+	"								float4 ornB =newOrnB;\n"
+	"								float4 c1 = transform(&c1local,&posB,&ornB);\n"
+	"								const float4 DeltaC2 = c0 - c1;\n"
+	"								{//\n"
+	"									int compoundPairIdx = atomic_inc(numCompoundPairsOut);\n"
+	"									if (compoundPairIdx<maxNumCompoundPairsCapacity)\n"
+	"									{\n"
+	"										gpuCompoundPairsOut[compoundPairIdx]  = (int4)(bodyIndexA,bodyIndexB,childShapeIndexA,childShapeIndexB);\n"
+	"									}\n"
+	"								}//\n"
+	"							}//fi(1)\n"
+	"						} //for (int b=0\n"
+	"					}//if (collidables[collidableIndexB].\n"
+	"					else//if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
+	"					{\n"
+	"						if (1)\n"
+	"						{\n"
+	"							int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
+	"							float dmin = FLT_MAX;\n"
+	"							float4 posA = newPosA;\n"
+	"							posA.w = 0.f;\n"
+	"							float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
+	"							posB.w = 0.f;\n"
+	"							float4 c0local = convexShapes[shapeIndexA].m_localCenter;\n"
+	"							float4 ornA = newOrnA;\n"
+	"							float4 c0 = transform(&c0local, &posA, &ornA);\n"
+	"							float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
+	"							float4 ornB = rigidBodies[bodyIndexB].m_quat;\n"
+	"							float4 c1 = transform(&c1local,&posB,&ornB);\n"
+	"							const float4 DeltaC2 = c0 - c1;\n"
+	"							{\n"
+	"								int compoundPairIdx = atomic_inc(numCompoundPairsOut);\n"
+	"								if (compoundPairIdx<maxNumCompoundPairsCapacity)\n"
+	"								{\n"
+	"									gpuCompoundPairsOut[compoundPairIdx] = (int4)(bodyIndexA,bodyIndexB,childShapeIndexA,-1);\n"
+	"								}//if (compoundPairIdx<maxNumCompoundPairsCapacity)\n"
+	"							}//\n"
+	"						}//fi (1)\n"
+	"					}//if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
+	"				}//for (int b=0;b<numChildrenB;b++)	\n"
+	"				return;\n"
+	"			}//if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
+	"			if ((collidables[collidableIndexA].m_shapeType!=SHAPE_CONCAVE_TRIMESH) \n"
+	"				&& (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS))\n"
+	"			{\n"
+	"				int numChildrenB = collidables[collidableIndexB].m_numChildShapes;\n"
+	"				for (int b=0;b<numChildrenB;b++)\n"
+	"				{\n"
+	"					int childShapeIndexB = collidables[collidableIndexB].m_shapeIndex+b;\n"
+	"					int childColIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;\n"
+	"					float4 ornB = rigidBodies[bodyIndexB].m_quat;\n"
+	"					float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
+	"					float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;\n"
+	"					float4 childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;\n"
+	"					float4 newPosB = qtRotate(ornB,childPosB)+posB;\n"
+	"					float4 newOrnB = qtMul(ornB,childOrnB);\n"
+	"					int shapeIndexB = collidables[childColIndexB].m_shapeIndex;\n"
+	"					//////////////////////////////////////\n"
+	"					if (1)\n"
+	"					{\n"
+	"						int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
+	"						float dmin = FLT_MAX;\n"
+	"						float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
+	"						posA.w = 0.f;\n"
+	"						float4 posB = newPosB;\n"
+	"						posB.w = 0.f;\n"
+	"						float4 c0local = convexShapes[shapeIndexA].m_localCenter;\n"
+	"						float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
+	"						float4 c0 = transform(&c0local, &posA, &ornA);\n"
+	"						float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
+	"						float4 ornB =newOrnB;\n"
+	"						float4 c1 = transform(&c1local,&posB,&ornB);\n"
+	"						const float4 DeltaC2 = c0 - c1;\n"
+	"						{//\n"
+	"							int compoundPairIdx = atomic_inc(numCompoundPairsOut);\n"
+	"							if (compoundPairIdx<maxNumCompoundPairsCapacity)\n"
+	"							{\n"
+	"								gpuCompoundPairsOut[compoundPairIdx] = (int4)(bodyIndexA,bodyIndexB,-1,childShapeIndexB);\n"
+	"							}//fi (compoundPairIdx<maxNumCompoundPairsCapacity)\n"
+	"						}//\n"
+	"					}//fi (1)	\n"
+	"				}//for (int b=0;b<numChildrenB;b++)\n"
+	"				return;\n"
+	"			}//if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
+	"			return;\n"
+	"		}//fi ((collidables[collidableIndexA].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS) ||(collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS))\n"
+	"	}//i<numPairs\n"
+	"}\n"
+	"// work-in-progress\n"
+	"__kernel void   findSeparatingAxisKernel( __global const int4* pairs, \n"
+	"																					__global const BodyData* rigidBodies, \n"
+	"																					__global const btCollidableGpu* collidables,\n"
+	"																					__global const ConvexPolyhedronCL* convexShapes, \n"
+	"																					__global const float4* vertices,\n"
+	"																					__global const float4* uniqueEdges,\n"
+	"																					__global const btGpuFace* faces,\n"
+	"																					__global const int* indices,\n"
+	"																					__global btAabbCL* aabbs,\n"
+	"																					__global volatile float4* separatingNormals,\n"
+	"																					__global volatile int* hasSeparatingAxis,\n"
+	"																					int numPairs\n"
+	"																					)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	\n"
+	"	if (i<numPairs)\n"
+	"	{\n"
+	"	\n"
+	"		int bodyIndexA = pairs[i].x;\n"
+	"		int bodyIndexB = pairs[i].y;\n"
+	"		int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
+	"		int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
+	"	\n"
+	"		int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
+	"		int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
+	"		\n"
+	"		\n"
+	"		//once the broadphase avoids static-static pairs, we can remove this test\n"
+	"		if ((rigidBodies[bodyIndexA].m_invMass==0) &&(rigidBodies[bodyIndexB].m_invMass==0))\n"
+	"		{\n"
+	"			hasSeparatingAxis[i] = 0;\n"
+	"			return;\n"
+	"		}\n"
+	"		\n"
+	"		if ((collidables[collidableIndexA].m_shapeType!=SHAPE_CONVEX_HULL) ||(collidables[collidableIndexB].m_shapeType!=SHAPE_CONVEX_HULL))\n"
+	"		{\n"
+	"			hasSeparatingAxis[i] = 0;\n"
+	"			return;\n"
+	"		}\n"
+	"			\n"
+	"		if ((collidables[collidableIndexA].m_shapeType==SHAPE_CONCAVE_TRIMESH))\n"
+	"		{\n"
+	"			hasSeparatingAxis[i] = 0;\n"
+	"			return;\n"
+	"		}\n"
+	"		int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
+	"		float dmin = FLT_MAX;\n"
+	"		float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
+	"		posA.w = 0.f;\n"
+	"		float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
+	"		posB.w = 0.f;\n"
+	"		float4 c0local = convexShapes[shapeIndexA].m_localCenter;\n"
+	"		float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
+	"		float4 c0 = transform(&c0local, &posA, &ornA);\n"
+	"		float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
+	"		float4 ornB =rigidBodies[bodyIndexB].m_quat;\n"
+	"		float4 c1 = transform(&c1local,&posB,&ornB);\n"
+	"		const float4 DeltaC2 = c0 - c1;\n"
+	"		float4 sepNormal;\n"
+	"		\n"
+	"		bool sepA = findSeparatingAxis(	&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],posA,ornA,\n"
+	"																								posB,ornB,\n"
+	"																								DeltaC2,\n"
+	"																								vertices,uniqueEdges,faces,\n"
+	"																								indices,&sepNormal,&dmin);\n"
+	"		hasSeparatingAxis[i] = 4;\n"
+	"		if (!sepA)\n"
+	"		{\n"
+	"			hasSeparatingAxis[i] = 0;\n"
+	"		} else\n"
+	"		{\n"
+	"			bool sepB = findSeparatingAxis(	&convexShapes[shapeIndexB],&convexShapes[shapeIndexA],posB,ornB,\n"
+	"																									posA,ornA,\n"
+	"																									DeltaC2,\n"
+	"																									vertices,uniqueEdges,faces,\n"
+	"																									indices,&sepNormal,&dmin);\n"
+	"			if (!sepB)\n"
+	"			{\n"
+	"				hasSeparatingAxis[i] = 0;\n"
+	"			} else\n"
+	"			{\n"
+	"				bool sepEE = findSeparatingAxisEdgeEdge(	&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],posA,ornA,\n"
+	"																									posB,ornB,\n"
+	"																									DeltaC2,\n"
+	"																									vertices,uniqueEdges,faces,\n"
+	"																									indices,&sepNormal,&dmin);\n"
+	"				if (!sepEE)\n"
+	"				{\n"
+	"					hasSeparatingAxis[i] = 0;\n"
+	"				} else\n"
+	"				{\n"
+	"					hasSeparatingAxis[i] = 1;\n"
+	"					separatingNormals[i] = sepNormal;\n"
+	"				}\n"
+	"			}\n"
+	"		}\n"
+	"		\n"
+	"	}\n"
+	"}\n"
+	"__kernel void   findSeparatingAxisVertexFaceKernel( __global const int4* pairs, \n"
+	"																					__global const BodyData* rigidBodies, \n"
+	"																					__global const btCollidableGpu* collidables,\n"
+	"																					__global const ConvexPolyhedronCL* convexShapes, \n"
+	"																					__global const float4* vertices,\n"
+	"																					__global const float4* uniqueEdges,\n"
+	"																					__global const btGpuFace* faces,\n"
+	"																					__global const int* indices,\n"
+	"																					__global btAabbCL* aabbs,\n"
+	"																					__global volatile float4* separatingNormals,\n"
+	"																					__global volatile int* hasSeparatingAxis,\n"
+	"																					__global  float* dmins,\n"
+	"																					int numPairs\n"
+	"																					)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	\n"
+	"	if (i<numPairs)\n"
+	"	{\n"
+	"	\n"
+	"		int bodyIndexA = pairs[i].x;\n"
+	"		int bodyIndexB = pairs[i].y;\n"
+	"		int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
+	"		int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
+	"	\n"
+	"		int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
+	"		int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
+	"	\n"
+	"		hasSeparatingAxis[i] = 0;	\n"
+	"		\n"
+	"		//once the broadphase avoids static-static pairs, we can remove this test\n"
+	"		if ((rigidBodies[bodyIndexA].m_invMass==0) &&(rigidBodies[bodyIndexB].m_invMass==0))\n"
+	"		{\n"
+	"			return;\n"
+	"		}\n"
+	"		\n"
+	"		if ((collidables[collidableIndexA].m_shapeType!=SHAPE_CONVEX_HULL) ||(collidables[collidableIndexB].m_shapeType!=SHAPE_CONVEX_HULL))\n"
+	"		{\n"
+	"			return;\n"
+	"		}\n"
+	"			\n"
+	"		int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
+	"		float dmin = FLT_MAX;\n"
+	"		dmins[i] = dmin;\n"
+	"		\n"
+	"		float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
+	"		posA.w = 0.f;\n"
+	"		float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
+	"		posB.w = 0.f;\n"
+	"		float4 c0local = convexShapes[shapeIndexA].m_localCenter;\n"
+	"		float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
+	"		float4 c0 = transform(&c0local, &posA, &ornA);\n"
+	"		float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
+	"		float4 ornB =rigidBodies[bodyIndexB].m_quat;\n"
+	"		float4 c1 = transform(&c1local,&posB,&ornB);\n"
+	"		const float4 DeltaC2 = c0 - c1;\n"
+	"		float4 sepNormal;\n"
+	"		\n"
+	"		bool sepA = findSeparatingAxis(	&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],posA,ornA,\n"
+	"																								posB,ornB,\n"
+	"																								DeltaC2,\n"
+	"																								vertices,uniqueEdges,faces,\n"
+	"																								indices,&sepNormal,&dmin);\n"
+	"		hasSeparatingAxis[i] = 4;\n"
+	"		if (!sepA)\n"
+	"		{\n"
+	"			hasSeparatingAxis[i] = 0;\n"
+	"		} else\n"
+	"		{\n"
+	"			bool sepB = findSeparatingAxis(	&convexShapes[shapeIndexB],&convexShapes[shapeIndexA],posB,ornB,\n"
+	"																									posA,ornA,\n"
+	"																									DeltaC2,\n"
+	"																									vertices,uniqueEdges,faces,\n"
+	"																									indices,&sepNormal,&dmin);\n"
+	"			if (sepB)\n"
+	"			{\n"
+	"				dmins[i] = dmin;\n"
+	"				hasSeparatingAxis[i] = 1;\n"
+	"				separatingNormals[i] = sepNormal;\n"
+	"			}\n"
+	"		}\n"
+	"		\n"
+	"	}\n"
+	"}\n"
+	"__kernel void   findSeparatingAxisEdgeEdgeKernel( __global const int4* pairs, \n"
+	"																					__global const BodyData* rigidBodies, \n"
+	"																					__global const btCollidableGpu* collidables,\n"
+	"																					__global const ConvexPolyhedronCL* convexShapes, \n"
+	"																					__global const float4* vertices,\n"
+	"																					__global const float4* uniqueEdges,\n"
+	"																					__global const btGpuFace* faces,\n"
+	"																					__global const int* indices,\n"
+	"																					__global btAabbCL* aabbs,\n"
+	"																					__global  float4* separatingNormals,\n"
+	"																					__global  int* hasSeparatingAxis,\n"
+	"																					__global  float* dmins,\n"
+	"																					__global const float4* unitSphereDirections,\n"
+	"																					int numUnitSphereDirections,\n"
+	"																					int numPairs\n"
+	"																					)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	\n"
+	"	if (i<numPairs)\n"
+	"	{\n"
+	"		if (hasSeparatingAxis[i])\n"
+	"		{\n"
+	"	\n"
+	"			int bodyIndexA = pairs[i].x;\n"
+	"			int bodyIndexB = pairs[i].y;\n"
+	"	\n"
+	"			int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
+	"			int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
+	"		\n"
+	"			int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
+	"			int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
+	"			\n"
+	"			\n"
+	"			int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
+	"	\n"
+	"			float dmin = dmins[i];\n"
+	"	\n"
+	"			float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
+	"			posA.w = 0.f;\n"
+	"			float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
+	"			posB.w = 0.f;\n"
+	"			float4 c0local = convexShapes[shapeIndexA].m_localCenter;\n"
+	"			float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
+	"			float4 c0 = transform(&c0local, &posA, &ornA);\n"
+	"			float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
+	"			float4 ornB =rigidBodies[bodyIndexB].m_quat;\n"
+	"			float4 c1 = transform(&c1local,&posB,&ornB);\n"
+	"			const float4 DeltaC2 = c0 - c1;\n"
+	"			float4 sepNormal = separatingNormals[i];\n"
+	"			\n"
+	"			\n"
+	"			\n"
+	"			bool sepEE = false;\n"
+	"			int numEdgeEdgeDirections = convexShapes[shapeIndexA].m_numUniqueEdges*convexShapes[shapeIndexB].m_numUniqueEdges;\n"
+	"			if (numEdgeEdgeDirections<=numUnitSphereDirections)\n"
+	"			{\n"
+	"				sepEE = findSeparatingAxisEdgeEdge(	&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],posA,ornA,\n"
+	"																									posB,ornB,\n"
+	"																									DeltaC2,\n"
+	"																									vertices,uniqueEdges,faces,\n"
+	"																									indices,&sepNormal,&dmin);\n"
+	"																									\n"
+	"					if (!sepEE)\n"
+	"					{\n"
+	"						hasSeparatingAxis[i] = 0;\n"
+	"					} else\n"
+	"					{\n"
+	"						hasSeparatingAxis[i] = 1;\n"
+	"						separatingNormals[i] = sepNormal;\n"
+	"					}\n"
+	"			}\n"
+	"			/*\n"
+	"			///else case is a separate kernel, to make Mac OSX OpenCL compiler happy\n"
+	"			else\n"
+	"			{\n"
+	"				sepEE = findSeparatingAxisUnitSphere(&convexShapes[shapeIndexA], &convexShapes[shapeIndexB],posA,ornA,\n"
+	"																									posB,ornB,\n"
+	"																									DeltaC2,\n"
+	"																									vertices,unitSphereDirections,numUnitSphereDirections,\n"
+	"																									&sepNormal,&dmin);\n"
+	"					if (!sepEE)\n"
+	"					{\n"
+	"						hasSeparatingAxis[i] = 0;\n"
+	"					} else\n"
+	"					{\n"
+	"						hasSeparatingAxis[i] = 1;\n"
+	"						separatingNormals[i] = sepNormal;\n"
+	"					}\n"
+	"			}\n"
+	"			*/\n"
+	"		}		//if (hasSeparatingAxis[i])\n"
+	"	}//(i<numPairs)\n"
+	"}\n"
+	"inline int	findClippingFaces(const float4 separatingNormal,\n"
+	"                      const ConvexPolyhedronCL* hullA, \n"
+	"					  __global const ConvexPolyhedronCL* hullB,\n"
+	"                      const float4 posA, const Quaternion ornA,const float4 posB, const Quaternion ornB,\n"
+	"                       __global float4* worldVertsA1,\n"
+	"                      __global float4* worldNormalsA1,\n"
+	"                      __global float4* worldVertsB1,\n"
+	"                      int capacityWorldVerts,\n"
+	"                      const float minDist, float maxDist,\n"
+	"					  const float4* verticesA,\n"
+	"                      const btGpuFace* facesA,\n"
+	"                      const int* indicesA,\n"
+	"					  __global const float4* verticesB,\n"
+	"                      __global const btGpuFace* facesB,\n"
+	"                      __global const int* indicesB,\n"
+	"                      __global int4* clippingFaces, int pairIndex)\n"
+	"{\n"
+	"	int numContactsOut = 0;\n"
+	"	int numWorldVertsB1= 0;\n"
+	"    \n"
+	"    \n"
+	"	int closestFaceB=0;\n"
+	"	float dmax = -FLT_MAX;\n"
+	"    \n"
+	"	{\n"
+	"		for(int face=0;face<hullB->m_numFaces;face++)\n"
+	"		{\n"
+	"			const float4 Normal = make_float4(facesB[hullB->m_faceOffset+face].m_plane.x,\n"
+	"                                              facesB[hullB->m_faceOffset+face].m_plane.y, facesB[hullB->m_faceOffset+face].m_plane.z,0.f);\n"
+	"			const float4 WorldNormal = qtRotate(ornB, Normal);\n"
+	"			float d = dot3F4(WorldNormal,separatingNormal);\n"
+	"			if (d > dmax)\n"
+	"			{\n"
+	"				dmax = d;\n"
+	"				closestFaceB = face;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"    \n"
+	"	{\n"
+	"		const btGpuFace polyB = facesB[hullB->m_faceOffset+closestFaceB];\n"
+	"		int numVertices = polyB.m_numIndices;\n"
+	"        if (numVertices>capacityWorldVerts)\n"
+	"            numVertices = capacityWorldVerts;\n"
+	"        \n"
+	"		for(int e0=0;e0<numVertices;e0++)\n"
+	"		{\n"
+	"            if (e0<capacityWorldVerts)\n"
+	"            {\n"
+	"                const float4 b = verticesB[hullB->m_vertexOffset+indicesB[polyB.m_indexOffset+e0]];\n"
+	"                worldVertsB1[pairIndex*capacityWorldVerts+numWorldVertsB1++] = transform(&b,&posB,&ornB);\n"
+	"            }\n"
+	"		}\n"
+	"	}\n"
+	"    \n"
+	"    int closestFaceA=0;\n"
+	"	{\n"
+	"		float dmin = FLT_MAX;\n"
+	"		for(int face=0;face<hullA->m_numFaces;face++)\n"
+	"		{\n"
+	"			const float4 Normal = make_float4(\n"
+	"                                              facesA[hullA->m_faceOffset+face].m_plane.x,\n"
+	"                                              facesA[hullA->m_faceOffset+face].m_plane.y,\n"
+	"                                              facesA[hullA->m_faceOffset+face].m_plane.z,\n"
+	"                                              0.f);\n"
+	"			const float4 faceANormalWS = qtRotate(ornA,Normal);\n"
+	"            \n"
+	"			float d = dot3F4(faceANormalWS,separatingNormal);\n"
+	"			if (d < dmin)\n"
+	"			{\n"
+	"				dmin = d;\n"
+	"				closestFaceA = face;\n"
+	"                worldNormalsA1[pairIndex] = faceANormalWS;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"    \n"
+	"    int numVerticesA = facesA[hullA->m_faceOffset+closestFaceA].m_numIndices;\n"
+	"    if (numVerticesA>capacityWorldVerts)\n"
+	"       numVerticesA = capacityWorldVerts;\n"
+	"    \n"
+	"	for(int e0=0;e0<numVerticesA;e0++)\n"
+	"	{\n"
+	"        if (e0<capacityWorldVerts)\n"
+	"        {\n"
+	"            const float4 a = verticesA[hullA->m_vertexOffset+indicesA[facesA[hullA->m_faceOffset+closestFaceA].m_indexOffset+e0]];\n"
+	"            worldVertsA1[pairIndex*capacityWorldVerts+e0] = transform(&a, &posA,&ornA);\n"
+	"        }\n"
+	"    }\n"
+	"    \n"
+	"    clippingFaces[pairIndex].x = closestFaceA;\n"
+	"    clippingFaces[pairIndex].y = closestFaceB;\n"
+	"    clippingFaces[pairIndex].z = numVerticesA;\n"
+	"    clippingFaces[pairIndex].w = numWorldVertsB1;\n"
+	"    \n"
+	"    \n"
+	"	return numContactsOut;\n"
+	"}\n"
+	"// work-in-progress\n"
+	"__kernel void   findConcaveSeparatingAxisKernel( __global int4* concavePairs,\n"
+	"																					__global const BodyData* rigidBodies,\n"
+	"																					__global const btCollidableGpu* collidables,\n"
+	"																					__global const ConvexPolyhedronCL* convexShapes, \n"
+	"																					__global const float4* vertices,\n"
+	"																					__global const float4* uniqueEdges,\n"
+	"																					__global const btGpuFace* faces,\n"
+	"																					__global const int* indices,\n"
+	"																					__global const btGpuChildShape* gpuChildShapes,\n"
+	"																					__global btAabbCL* aabbs,\n"
+	"																					__global float4* concaveSeparatingNormalsOut,\n"
+	"																					__global int* concaveHasSeparatingNormals,\n"
+	"																					__global int4* clippingFacesOut,\n"
+	"																					__global float4* worldVertsA1GPU,\n"
+	"																					__global float4*  worldNormalsAGPU,\n"
+	"																					__global float4* worldVertsB1GPU,\n"
+	"																					int vertexFaceCapacity,\n"
+	"																					int numConcavePairs\n"
+	"																					)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	if (i>=numConcavePairs)\n"
+	"		return;\n"
+	"	concaveHasSeparatingNormals[i] = 0;\n"
+	"	int pairIdx = i;\n"
+	"	int bodyIndexA = concavePairs[i].x;\n"
+	"	int bodyIndexB = concavePairs[i].y;\n"
+	"	int collidableIndexA = rigidBodies[bodyIndexA].m_collidableIdx;\n"
+	"	int collidableIndexB = rigidBodies[bodyIndexB].m_collidableIdx;\n"
+	"	int shapeIndexA = collidables[collidableIndexA].m_shapeIndex;\n"
+	"	int shapeIndexB = collidables[collidableIndexB].m_shapeIndex;\n"
+	"	if (collidables[collidableIndexB].m_shapeType!=SHAPE_CONVEX_HULL&&\n"
+	"		collidables[collidableIndexB].m_shapeType!=SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
+	"	{\n"
+	"		concavePairs[pairIdx].w = -1;\n"
+	"		return;\n"
+	"	}\n"
+	"	int numFacesA = convexShapes[shapeIndexA].m_numFaces;\n"
+	"	int numActualConcaveConvexTests = 0;\n"
+	"	\n"
+	"	int f = concavePairs[i].z;\n"
+	"	\n"
+	"	bool overlap = false;\n"
+	"	\n"
+	"	ConvexPolyhedronCL convexPolyhedronA;\n"
+	"	//add 3 vertices of the triangle\n"
+	"	convexPolyhedronA.m_numVertices = 3;\n"
+	"	convexPolyhedronA.m_vertexOffset = 0;\n"
+	"	float4	localCenter = make_float4(0.f,0.f,0.f,0.f);\n"
+	"	btGpuFace face = faces[convexShapes[shapeIndexA].m_faceOffset+f];\n"
+	"	float4 triMinAabb, triMaxAabb;\n"
+	"	btAabbCL triAabb;\n"
+	"	triAabb.m_min = make_float4(1e30f,1e30f,1e30f,0.f);\n"
+	"	triAabb.m_max = make_float4(-1e30f,-1e30f,-1e30f,0.f);\n"
+	"	\n"
+	"	float4 verticesA[3];\n"
+	"	for (int i=0;i<3;i++)\n"
+	"	{\n"
+	"		int index = indices[face.m_indexOffset+i];\n"
+	"		float4 vert = vertices[convexShapes[shapeIndexA].m_vertexOffset+index];\n"
+	"		verticesA[i] = vert;\n"
+	"		localCenter += vert;\n"
+	"			\n"
+	"		triAabb.m_min = min(triAabb.m_min,vert);		\n"
+	"		triAabb.m_max = max(triAabb.m_max,vert);		\n"
+	"	}\n"
+	"	overlap = true;\n"
+	"	overlap = (triAabb.m_min.x > aabbs[bodyIndexB].m_max.x || triAabb.m_max.x < aabbs[bodyIndexB].m_min.x) ? false : overlap;\n"
+	"	overlap = (triAabb.m_min.z > aabbs[bodyIndexB].m_max.z || triAabb.m_max.z < aabbs[bodyIndexB].m_min.z) ? false : overlap;\n"
+	"	overlap = (triAabb.m_min.y > aabbs[bodyIndexB].m_max.y || triAabb.m_max.y < aabbs[bodyIndexB].m_min.y) ? false : overlap;\n"
+	"		\n"
+	"	if (overlap)\n"
+	"	{\n"
+	"		float dmin = FLT_MAX;\n"
+	"		int hasSeparatingAxis=5;\n"
+	"		float4 sepAxis=make_float4(1,2,3,4);\n"
+	"		int localCC=0;\n"
+	"		numActualConcaveConvexTests++;\n"
+	"		//a triangle has 3 unique edges\n"
+	"		convexPolyhedronA.m_numUniqueEdges = 3;\n"
+	"		convexPolyhedronA.m_uniqueEdgesOffset = 0;\n"
+	"		float4 uniqueEdgesA[3];\n"
+	"		\n"
+	"		uniqueEdgesA[0] = (verticesA[1]-verticesA[0]);\n"
+	"		uniqueEdgesA[1] = (verticesA[2]-verticesA[1]);\n"
+	"		uniqueEdgesA[2] = (verticesA[0]-verticesA[2]);\n"
+	"		convexPolyhedronA.m_faceOffset = 0;\n"
+	"                                  \n"
+	"		float4 normal = make_float4(face.m_plane.x,face.m_plane.y,face.m_plane.z,0.f);\n"
+	"                             \n"
+	"		btGpuFace facesA[TRIANGLE_NUM_CONVEX_FACES];\n"
+	"		int indicesA[3+3+2+2+2];\n"
+	"		int curUsedIndices=0;\n"
+	"		int fidx=0;\n"
+	"		//front size of triangle\n"
+	"		{\n"
+	"			facesA[fidx].m_indexOffset=curUsedIndices;\n"
+	"			indicesA[0] = 0;\n"
+	"			indicesA[1] = 1;\n"
+	"			indicesA[2] = 2;\n"
+	"			curUsedIndices+=3;\n"
+	"			float c = face.m_plane.w;\n"
+	"			facesA[fidx].m_plane.x = normal.x;\n"
+	"			facesA[fidx].m_plane.y = normal.y;\n"
+	"			facesA[fidx].m_plane.z = normal.z;\n"
+	"			facesA[fidx].m_plane.w = c;\n"
+	"			facesA[fidx].m_numIndices=3;\n"
+	"		}\n"
+	"		fidx++;\n"
+	"		//back size of triangle\n"
+	"		{\n"
+	"			facesA[fidx].m_indexOffset=curUsedIndices;\n"
+	"			indicesA[3]=2;\n"
+	"			indicesA[4]=1;\n"
+	"			indicesA[5]=0;\n"
+	"			curUsedIndices+=3;\n"
+	"			float c = dot(normal,verticesA[0]);\n"
+	"			float c1 = -face.m_plane.w;\n"
+	"			facesA[fidx].m_plane.x = -normal.x;\n"
+	"			facesA[fidx].m_plane.y = -normal.y;\n"
+	"			facesA[fidx].m_plane.z = -normal.z;\n"
+	"			facesA[fidx].m_plane.w = c;\n"
+	"			facesA[fidx].m_numIndices=3;\n"
+	"		}\n"
+	"		fidx++;\n"
+	"		bool addEdgePlanes = true;\n"
+	"		if (addEdgePlanes)\n"
+	"		{\n"
+	"			int numVertices=3;\n"
+	"			int prevVertex = numVertices-1;\n"
+	"			for (int i=0;i<numVertices;i++)\n"
+	"			{\n"
+	"				float4 v0 = verticesA[i];\n"
+	"				float4 v1 = verticesA[prevVertex];\n"
+	"                                            \n"
+	"				float4 edgeNormal = normalize(cross(normal,v1-v0));\n"
+	"				float c = -dot(edgeNormal,v0);\n"
+	"				facesA[fidx].m_numIndices = 2;\n"
+	"				facesA[fidx].m_indexOffset=curUsedIndices;\n"
+	"				indicesA[curUsedIndices++]=i;\n"
+	"				indicesA[curUsedIndices++]=prevVertex;\n"
+	"                                            \n"
+	"				facesA[fidx].m_plane.x = edgeNormal.x;\n"
+	"				facesA[fidx].m_plane.y = edgeNormal.y;\n"
+	"				facesA[fidx].m_plane.z = edgeNormal.z;\n"
+	"				facesA[fidx].m_plane.w = c;\n"
+	"				fidx++;\n"
+	"				prevVertex = i;\n"
+	"			}\n"
+	"		}\n"
+	"		convexPolyhedronA.m_numFaces = TRIANGLE_NUM_CONVEX_FACES;\n"
+	"		convexPolyhedronA.m_localCenter = localCenter*(1.f/3.f);\n"
+	"		float4 posA = rigidBodies[bodyIndexA].m_pos;\n"
+	"		posA.w = 0.f;\n"
+	"		float4 posB = rigidBodies[bodyIndexB].m_pos;\n"
+	"		posB.w = 0.f;\n"
+	"		float4 ornA = rigidBodies[bodyIndexA].m_quat;\n"
+	"		float4 ornB =rigidBodies[bodyIndexB].m_quat;\n"
+	"		\n"
+	"		///////////////////\n"
+	"		///compound shape support\n"
+	"		if (collidables[collidableIndexB].m_shapeType==SHAPE_COMPOUND_OF_CONVEX_HULLS)\n"
+	"		{\n"
+	"			int compoundChild = concavePairs[pairIdx].w;\n"
+	"			int childShapeIndexB = compoundChild;//collidables[collidableIndexB].m_shapeIndex+compoundChild;\n"
+	"			int childColIndexB = gpuChildShapes[childShapeIndexB].m_shapeIndex;\n"
+	"			float4 childPosB = gpuChildShapes[childShapeIndexB].m_childPosition;\n"
+	"			float4 childOrnB = gpuChildShapes[childShapeIndexB].m_childOrientation;\n"
+	"			float4 newPosB = transform(&childPosB,&posB,&ornB);\n"
+	"			float4 newOrnB = qtMul(ornB,childOrnB);\n"
+	"			posB = newPosB;\n"
+	"			ornB = newOrnB;\n"
+	"			shapeIndexB = collidables[childColIndexB].m_shapeIndex;\n"
+	"		}\n"
+	"		//////////////////\n"
+	"		float4 c0local = convexPolyhedronA.m_localCenter;\n"
+	"		float4 c0 = transform(&c0local, &posA, &ornA);\n"
+	"		float4 c1local = convexShapes[shapeIndexB].m_localCenter;\n"
+	"		float4 c1 = transform(&c1local,&posB,&ornB);\n"
+	"		const float4 DeltaC2 = c0 - c1;\n"
+	"		bool sepA = findSeparatingAxisLocalA(	&convexPolyhedronA, &convexShapes[shapeIndexB],\n"
+	"												posA,ornA,\n"
+	"												posB,ornB,\n"
+	"												DeltaC2,\n"
+	"												verticesA,uniqueEdgesA,facesA,indicesA,\n"
+	"												vertices,uniqueEdges,faces,indices,\n"
+	"												&sepAxis,&dmin);\n"
+	"		hasSeparatingAxis = 4;\n"
+	"		if (!sepA)\n"
+	"		{\n"
+	"			hasSeparatingAxis = 0;\n"
+	"		} else\n"
+	"		{\n"
+	"			bool sepB = findSeparatingAxisLocalB(	&convexShapes[shapeIndexB],&convexPolyhedronA,\n"
+	"												posB,ornB,\n"
+	"												posA,ornA,\n"
+	"												DeltaC2,\n"
+	"												vertices,uniqueEdges,faces,indices,\n"
+	"												verticesA,uniqueEdgesA,facesA,indicesA,\n"
+	"												&sepAxis,&dmin);\n"
+	"			if (!sepB)\n"
+	"			{\n"
+	"				hasSeparatingAxis = 0;\n"
+	"			} else\n"
+	"			{\n"
+	"				bool sepEE = findSeparatingAxisEdgeEdgeLocalA(	&convexPolyhedronA, &convexShapes[shapeIndexB],\n"
+	"															posA,ornA,\n"
+	"															posB,ornB,\n"
+	"															DeltaC2,\n"
+	"															verticesA,uniqueEdgesA,facesA,indicesA,\n"
+	"															vertices,uniqueEdges,faces,indices,\n"
+	"															&sepAxis,&dmin);\n"
+	"	\n"
+	"				if (!sepEE)\n"
+	"				{\n"
+	"					hasSeparatingAxis = 0;\n"
+	"				} else\n"
+	"				{\n"
+	"					hasSeparatingAxis = 1;\n"
+	"				}\n"
+	"			}\n"
+	"		}	\n"
+	"		\n"
+	"		if (hasSeparatingAxis)\n"
+	"		{\n"
+	"			sepAxis.w = dmin;\n"
+	"			concaveSeparatingNormalsOut[pairIdx]=sepAxis;\n"
+	"			concaveHasSeparatingNormals[i]=1;\n"
+	"			float minDist = -1e30f;\n"
+	"			float maxDist = 0.02f;\n"
+	"		\n"
+	"			findClippingFaces(sepAxis,\n"
+	"                     &convexPolyhedronA,\n"
+	"					 &convexShapes[shapeIndexB],\n"
+	"					 posA,ornA,\n"
+	"					 posB,ornB,\n"
+	"                      worldVertsA1GPU,\n"
+	"                      worldNormalsAGPU,\n"
+	"                      worldVertsB1GPU,\n"
+	"					  vertexFaceCapacity,\n"
+	"                      minDist, maxDist,\n"
+	"                      verticesA,\n"
+	"                      facesA,\n"
+	"                      indicesA,\n"
+	" 					  vertices,\n"
+	"                      faces,\n"
+	"                      indices,\n"
+	"                      clippingFacesOut, pairIdx);\n"
+	"		} else\n"
+	"		{	\n"
+	"			//mark this pair as in-active\n"
+	"			concavePairs[pairIdx].w = -1;\n"
+	"		}\n"
+	"	}\n"
+	"	else\n"
+	"	{	\n"
+	"		//mark this pair as in-active\n"
+	"		concavePairs[pairIdx].w = -1;\n"
+	"	}\n"
+	"	\n"
+	"	concavePairs[pairIdx].z = -1;//now z is used for existing/persistent contacts\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3BoundSearchCL.cpp b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3BoundSearchCL.cpp
index a4980f71e11..c0e11bfb26b 100644
--- a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3BoundSearchCL.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3BoundSearchCL.cpp
@@ -19,149 +19,139 @@ subject to the following restrictions:
 #define KERNEL1 "SearchSortDataUpperKernel"
 #define KERNEL2 "SubtractKernel"
 
-
 #include "b3BoundSearchCL.h"
 #include "Bullet3OpenCL/Initialize/b3OpenCLUtils.h"
 #include "b3LauncherCL.h"
 #include "kernels/BoundSearchKernelsCL.h"
 
 b3BoundSearchCL::b3BoundSearchCL(cl_context ctx, cl_device_id device, cl_command_queue queue, int maxSize)
-	:m_context(ctx),
-	m_device(device),
-	m_queue(queue)
+	: m_context(ctx),
+	  m_device(device),
+	  m_queue(queue)
 {
-
 	const char* additionalMacros = "";
 	//const char* srcFileNameForCaching="";
 
 	cl_int pErrNum;
 	const char* kernelSource = boundSearchKernelsCL;
 
-	cl_program boundSearchProg = b3OpenCLUtils::compileCLProgramFromString( ctx, device, kernelSource, &pErrNum,additionalMacros, BOUNDSEARCH_PATH);
+	cl_program boundSearchProg = b3OpenCLUtils::compileCLProgramFromString(ctx, device, kernelSource, &pErrNum, additionalMacros, BOUNDSEARCH_PATH);
 	b3Assert(boundSearchProg);
 
-	m_lowerSortDataKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, kernelSource, "SearchSortDataLowerKernel", &pErrNum, boundSearchProg,additionalMacros );
-	b3Assert(m_lowerSortDataKernel );
+	m_lowerSortDataKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, kernelSource, "SearchSortDataLowerKernel", &pErrNum, boundSearchProg, additionalMacros);
+	b3Assert(m_lowerSortDataKernel);
 
-	m_upperSortDataKernel= b3OpenCLUtils::compileCLKernelFromString( ctx, device, kernelSource, "SearchSortDataUpperKernel", &pErrNum, boundSearchProg,additionalMacros );
+	m_upperSortDataKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, kernelSource, "SearchSortDataUpperKernel", &pErrNum, boundSearchProg, additionalMacros);
 	b3Assert(m_upperSortDataKernel);
 
 	m_subtractKernel = 0;
 
-	if( maxSize )
+	if (maxSize)
 	{
-		m_subtractKernel= b3OpenCLUtils::compileCLKernelFromString( ctx, device, kernelSource, "SubtractKernel", &pErrNum, boundSearchProg,additionalMacros );
+		m_subtractKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, kernelSource, "SubtractKernel", &pErrNum, boundSearchProg, additionalMacros);
 		b3Assert(m_subtractKernel);
 	}
 
 	//m_constBuffer = new b3OpenCLArray<b3Int4>( device, 1, BufferBase::BUFFER_CONST );
-	
-	m_lower = (maxSize == 0)? 0: new b3OpenCLArray<unsigned int>(ctx,queue,maxSize );
-	m_upper = (maxSize == 0)? 0: new b3OpenCLArray<unsigned int>(ctx,queue, maxSize );
 
-	m_filler = new b3FillCL(ctx,device,queue);
+	m_lower = (maxSize == 0) ? 0 : new b3OpenCLArray<unsigned int>(ctx, queue, maxSize);
+	m_upper = (maxSize == 0) ? 0 : new b3OpenCLArray<unsigned int>(ctx, queue, maxSize);
+
+	m_filler = new b3FillCL(ctx, device, queue);
 }
 
 b3BoundSearchCL::~b3BoundSearchCL()
 {
-	
 	delete m_lower;
 	delete m_upper;
 	delete m_filler;
-			
+
 	clReleaseKernel(m_lowerSortDataKernel);
 	clReleaseKernel(m_upperSortDataKernel);
 	clReleaseKernel(m_subtractKernel);
-	
-
 }
 
-
-void b3BoundSearchCL::execute(b3OpenCLArray<b3SortData>& src, int nSrc, b3OpenCLArray<unsigned int>& dst, int nDst, Option option )
+void b3BoundSearchCL::execute(b3OpenCLArray<b3SortData>& src, int nSrc, b3OpenCLArray<unsigned int>& dst, int nDst, Option option)
 {
 	b3Int4 constBuffer;
 	constBuffer.x = nSrc;
 	constBuffer.y = nDst;
 
-	if( option == BOUND_LOWER )
+	if (option == BOUND_LOWER)
 	{
-		b3BufferInfoCL bInfo[] = { b3BufferInfoCL( src.getBufferCL(), true ), b3BufferInfoCL( dst.getBufferCL()) };
+		b3BufferInfoCL bInfo[] = {b3BufferInfoCL(src.getBufferCL(), true), b3BufferInfoCL(dst.getBufferCL())};
 
-		b3LauncherCL launcher( m_queue, m_lowerSortDataKernel,"m_lowerSortDataKernel" );
-		launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-		launcher.setConst( nSrc );
-        launcher.setConst( nDst );
-        
-		launcher.launch1D( nSrc, 64 );
+		b3LauncherCL launcher(m_queue, m_lowerSortDataKernel, "m_lowerSortDataKernel");
+		launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+		launcher.setConst(nSrc);
+		launcher.setConst(nDst);
+
+		launcher.launch1D(nSrc, 64);
 	}
-	else if( option == BOUND_UPPER )
+	else if (option == BOUND_UPPER)
 	{
-		b3BufferInfoCL bInfo[] = { b3BufferInfoCL( src.getBufferCL(), true ), b3BufferInfoCL( dst.getBufferCL() ) };
+		b3BufferInfoCL bInfo[] = {b3BufferInfoCL(src.getBufferCL(), true), b3BufferInfoCL(dst.getBufferCL())};
 
-		b3LauncherCL launcher(m_queue, m_upperSortDataKernel,"m_upperSortDataKernel" );
-		launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-        launcher.setConst( nSrc );
-        launcher.setConst( nDst );
+		b3LauncherCL launcher(m_queue, m_upperSortDataKernel, "m_upperSortDataKernel");
+		launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+		launcher.setConst(nSrc);
+		launcher.setConst(nDst);
 
-		launcher.launch1D( nSrc, 64 );
+		launcher.launch1D(nSrc, 64);
 	}
-	else if( option == COUNT )
+	else if (option == COUNT)
 	{
-		b3Assert( m_lower );
-		b3Assert( m_upper );
-		b3Assert( m_lower->capacity() <= (int)nDst );
-		b3Assert( m_upper->capacity() <= (int)nDst );
+		b3Assert(m_lower);
+		b3Assert(m_upper);
+		b3Assert(m_lower->capacity() <= (int)nDst);
+		b3Assert(m_upper->capacity() <= (int)nDst);
 
 		int zero = 0;
-		m_filler->execute( *m_lower, zero, nDst );
-		m_filler->execute( *m_upper, zero, nDst );
+		m_filler->execute(*m_lower, zero, nDst);
+		m_filler->execute(*m_upper, zero, nDst);
 
-		execute( src, nSrc, *m_lower, nDst, BOUND_LOWER );
-		execute( src, nSrc, *m_upper, nDst, BOUND_UPPER );
+		execute(src, nSrc, *m_lower, nDst, BOUND_LOWER);
+		execute(src, nSrc, *m_upper, nDst, BOUND_UPPER);
 
 		{
-			b3BufferInfoCL bInfo[] = { b3BufferInfoCL( m_upper->getBufferCL(), true ), b3BufferInfoCL( m_lower->getBufferCL(), true ), b3BufferInfoCL( dst.getBufferCL() ) };
+			b3BufferInfoCL bInfo[] = {b3BufferInfoCL(m_upper->getBufferCL(), true), b3BufferInfoCL(m_lower->getBufferCL(), true), b3BufferInfoCL(dst.getBufferCL())};
 
-			b3LauncherCL  launcher( m_queue, m_subtractKernel ,"m_subtractKernel");
-			launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-            launcher.setConst( nSrc );
-            launcher.setConst( nDst );
+			b3LauncherCL launcher(m_queue, m_subtractKernel, "m_subtractKernel");
+			launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+			launcher.setConst(nSrc);
+			launcher.setConst(nDst);
 
-			launcher.launch1D( nDst, 64 );
+			launcher.launch1D(nDst, 64);
 		}
 	}
 	else
 	{
-		b3Assert( 0 );
+		b3Assert(0);
 	}
-
 }
 
-
-void b3BoundSearchCL::executeHost( b3AlignedObjectArray<b3SortData>& src, int nSrc, 
-	b3AlignedObjectArray<unsigned int>& dst,  int nDst, Option option )
+void b3BoundSearchCL::executeHost(b3AlignedObjectArray<b3SortData>& src, int nSrc,
+								  b3AlignedObjectArray<unsigned int>& dst, int nDst, Option option)
 {
+	for (int i = 0; i < nSrc - 1; i++)
+		b3Assert(src[i].m_key <= src[i + 1].m_key);
 
-
-	for(int i=0; i<nSrc-1; i++) 
-		b3Assert( src[i].m_key <= src[i+1].m_key );
-
-	b3SortData minData,zeroData,maxData;
+	b3SortData minData, zeroData, maxData;
 	minData.m_key = -1;
 	minData.m_value = -1;
-	zeroData.m_key=0;
-	zeroData.m_value=0;
+	zeroData.m_key = 0;
+	zeroData.m_value = 0;
 	maxData.m_key = nDst;
 	maxData.m_value = nDst;
 
-	if( option == BOUND_LOWER )
+	if (option == BOUND_LOWER)
 	{
-		for(int i=0; i<nSrc; i++)
+		for (int i = 0; i < nSrc; i++)
 		{
-			b3SortData& iData = (i==0)? minData: src[i-1];
-			b3SortData& jData = (i==nSrc)? maxData: src[i];
+			b3SortData& iData = (i == 0) ? minData : src[i - 1];
+			b3SortData& jData = (i == nSrc) ? maxData : src[i];
 
-			if( iData.m_key != jData.m_key )
+			if (iData.m_key != jData.m_key)
 			{
 				int k = jData.m_key;
 				{
@@ -170,14 +160,14 @@ void b3BoundSearchCL::executeHost( b3AlignedObjectArray<b3SortData>& src, int nS
 			}
 		}
 	}
-	else if( option == BOUND_UPPER )
+	else if (option == BOUND_UPPER)
 	{
-		for(int i=1; i<nSrc+1; i++)
+		for (int i = 1; i < nSrc + 1; i++)
 		{
-			b3SortData& iData = src[i-1];
-			b3SortData& jData = (i==nSrc)? maxData: src[i];
+			b3SortData& iData = src[i - 1];
+			b3SortData& jData = (i == nSrc) ? maxData : src[i];
 
-			if( iData.m_key != jData.m_key )
+			if (iData.m_key != jData.m_key)
 			{
 				int k = iData.m_key;
 				{
@@ -186,28 +176,28 @@ void b3BoundSearchCL::executeHost( b3AlignedObjectArray<b3SortData>& src, int nS
 			}
 		}
 	}
-	else if( option == COUNT )
+	else if (option == COUNT)
 	{
 		b3AlignedObjectArray<unsigned int> lower;
-		lower.resize(nDst );
+		lower.resize(nDst);
 		b3AlignedObjectArray<unsigned int> upper;
-		upper.resize(nDst );
+		upper.resize(nDst);
 
-		for(int i=0; i<nDst; i++) 
-		{ 
-			lower[i] = upper[i] = 0; 
+		for (int i = 0; i < nDst; i++)
+		{
+			lower[i] = upper[i] = 0;
 		}
 
-		executeHost( src, nSrc, lower, nDst, BOUND_LOWER );
-		executeHost( src, nSrc, upper, nDst, BOUND_UPPER );
+		executeHost(src, nSrc, lower, nDst, BOUND_LOWER);
+		executeHost(src, nSrc, upper, nDst, BOUND_UPPER);
 
-		for( int i=0; i<nDst; i++) 
-		{ 
-			dst[i] = upper[i] - lower[i]; 
+		for (int i = 0; i < nDst; i++)
+		{
+			dst[i] = upper[i] - lower[i];
 		}
 	}
 	else
 	{
-		b3Assert( 0 );
+		b3Assert(0);
 	}
 }
diff --git a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3BoundSearchCL.h b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3BoundSearchCL.h
index 7e2940965c1..0d633e3d235 100644
--- a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3BoundSearchCL.h
+++ b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3BoundSearchCL.h
@@ -26,42 +26,39 @@ subject to the following restrictions:
 
 #include "b3OpenCLArray.h"
 #include "b3FillCL.h"
-#include "b3RadixSort32CL.h" //for b3SortData (perhaps move it?)
+#include "b3RadixSort32CL.h"  //for b3SortData (perhaps move it?)
 class b3BoundSearchCL
 {
-	public:
+public:
+	enum Option
+	{
+		BOUND_LOWER,
+		BOUND_UPPER,
+		COUNT,
+	};
 
-		enum Option
-		{
-			BOUND_LOWER,
-			BOUND_UPPER,
-			COUNT,
-		};
+	cl_context m_context;
+	cl_device_id m_device;
+	cl_command_queue m_queue;
 
-		cl_context m_context;
-		cl_device_id m_device;
-		cl_command_queue m_queue;
+	cl_kernel m_lowerSortDataKernel;
+	cl_kernel m_upperSortDataKernel;
+	cl_kernel m_subtractKernel;
 
-		
-		cl_kernel m_lowerSortDataKernel;
-		cl_kernel m_upperSortDataKernel;
-		cl_kernel m_subtractKernel;
-		
-		b3OpenCLArray<b3Int4>* m_constbtOpenCLArray;
-		b3OpenCLArray<unsigned int>* m_lower;
-		b3OpenCLArray<unsigned int>* m_upper;
-		
-		b3FillCL* m_filler;
-		
-		b3BoundSearchCL(cl_context context, cl_device_id device, cl_command_queue queue, int size);
+	b3OpenCLArray<b3Int4>* m_constbtOpenCLArray;
+	b3OpenCLArray<unsigned int>* m_lower;
+	b3OpenCLArray<unsigned int>* m_upper;
 
-		virtual ~b3BoundSearchCL();
+	b3FillCL* m_filler;
 
-		//	src has to be src[i].m_key <= src[i+1].m_key
-		void execute( b3OpenCLArray<b3SortData>& src, int nSrc, b3OpenCLArray<unsigned int>& dst, int nDst, Option option = BOUND_LOWER );
+	b3BoundSearchCL(cl_context context, cl_device_id device, cl_command_queue queue, int size);
 
-		void executeHost( b3AlignedObjectArray<b3SortData>& src, int nSrc, b3AlignedObjectArray<unsigned int>& dst, int nDst, Option option = BOUND_LOWER);
+	virtual ~b3BoundSearchCL();
+
+	//	src has to be src[i].m_key <= src[i+1].m_key
+	void execute(b3OpenCLArray<b3SortData>& src, int nSrc, b3OpenCLArray<unsigned int>& dst, int nDst, Option option = BOUND_LOWER);
+
+	void executeHost(b3AlignedObjectArray<b3SortData>& src, int nSrc, b3AlignedObjectArray<unsigned int>& dst, int nDst, Option option = BOUND_LOWER);
 };
 
-
-#endif //B3_BOUNDSEARCH_H
+#endif  //B3_BOUNDSEARCH_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3BufferInfoCL.h b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3BufferInfoCL.h
index 52f219ae3fc..35fc467b207 100644
--- a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3BufferInfoCL.h
+++ b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3BufferInfoCL.h
@@ -4,16 +4,15 @@
 
 #include "b3OpenCLArray.h"
 
-
 struct b3BufferInfoCL
 {
 	//b3BufferInfoCL(){}
 
-//	template<typename T>
-	b3BufferInfoCL(cl_mem buff, bool isReadOnly = false): m_clBuffer(buff), m_isReadOnly(isReadOnly){}
+	//	template<typename T>
+	b3BufferInfoCL(cl_mem buff, bool isReadOnly = false) : m_clBuffer(buff), m_isReadOnly(isReadOnly) {}
 
 	cl_mem m_clBuffer;
 	bool m_isReadOnly;
 };
 
-#endif //B3_BUFFER_INFO_CL_H
+#endif  //B3_BUFFER_INFO_CL_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3FillCL.cpp b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3FillCL.cpp
index f05c2648f12..bd25bb2101b 100644
--- a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3FillCL.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3FillCL.cpp
@@ -8,29 +8,26 @@
 #include "kernels/FillKernelsCL.h"
 
 b3FillCL::b3FillCL(cl_context ctx, cl_device_id device, cl_command_queue queue)
-:m_commandQueue(queue)
+	: m_commandQueue(queue)
 {
 	const char* kernelSource = fillKernelsCL;
 	cl_int pErrNum;
 	const char* additionalMacros = "";
 
-	cl_program fillProg = b3OpenCLUtils::compileCLProgramFromString( ctx, device, kernelSource, &pErrNum,additionalMacros, FILL_CL_PROGRAM_PATH);
+	cl_program fillProg = b3OpenCLUtils::compileCLProgramFromString(ctx, device, kernelSource, &pErrNum, additionalMacros, FILL_CL_PROGRAM_PATH);
 	b3Assert(fillProg);
 
-	m_fillIntKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, kernelSource, "FillIntKernel", &pErrNum, fillProg,additionalMacros );
+	m_fillIntKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, kernelSource, "FillIntKernel", &pErrNum, fillProg, additionalMacros);
 	b3Assert(m_fillIntKernel);
 
-	m_fillUnsignedIntKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, kernelSource, "FillUnsignedIntKernel", &pErrNum, fillProg,additionalMacros );
+	m_fillUnsignedIntKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, kernelSource, "FillUnsignedIntKernel", &pErrNum, fillProg, additionalMacros);
 	b3Assert(m_fillIntKernel);
 
-	m_fillFloatKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, kernelSource, "FillFloatKernel", &pErrNum, fillProg,additionalMacros );
+	m_fillFloatKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, kernelSource, "FillFloatKernel", &pErrNum, fillProg, additionalMacros);
 	b3Assert(m_fillFloatKernel);
 
-	
-
-	m_fillKernelInt2 = b3OpenCLUtils::compileCLKernelFromString( ctx, device, kernelSource, "FillInt2Kernel", &pErrNum, fillProg,additionalMacros );
+	m_fillKernelInt2 = b3OpenCLUtils::compileCLKernelFromString(ctx, device, kernelSource, "FillInt2Kernel", &pErrNum, fillProg, additionalMacros);
 	b3Assert(m_fillKernelInt2);
-	
 }
 
 b3FillCL::~b3FillCL()
@@ -39,88 +36,84 @@ b3FillCL::~b3FillCL()
 	clReleaseKernel(m_fillIntKernel);
 	clReleaseKernel(m_fillUnsignedIntKernel);
 	clReleaseKernel(m_fillFloatKernel);
-
 }
 
 void b3FillCL::execute(b3OpenCLArray<float>& src, const float value, int n, int offset)
 {
-	b3Assert( n>0 );
+	b3Assert(n > 0);
 
 	{
-		b3LauncherCL launcher( m_commandQueue, m_fillFloatKernel,"m_fillFloatKernel" );
-		launcher.setBuffer( src.getBufferCL());
-		launcher.setConst( n );
-		launcher.setConst( value );
-		launcher.setConst( offset);
+		b3LauncherCL launcher(m_commandQueue, m_fillFloatKernel, "m_fillFloatKernel");
+		launcher.setBuffer(src.getBufferCL());
+		launcher.setConst(n);
+		launcher.setConst(value);
+		launcher.setConst(offset);
 
-		launcher.launch1D( n );
+		launcher.launch1D(n);
 	}
 }
 
 void b3FillCL::execute(b3OpenCLArray<int>& src, const int value, int n, int offset)
 {
-	b3Assert( n>0 );
-	
+	b3Assert(n > 0);
 
 	{
-		b3LauncherCL launcher( m_commandQueue, m_fillIntKernel ,"m_fillIntKernel");
+		b3LauncherCL launcher(m_commandQueue, m_fillIntKernel, "m_fillIntKernel");
 		launcher.setBuffer(src.getBufferCL());
-		launcher.setConst( n);
-		launcher.setConst( value);
-		launcher.setConst( offset);
-		launcher.launch1D( n );
+		launcher.setConst(n);
+		launcher.setConst(value);
+		launcher.setConst(offset);
+		launcher.launch1D(n);
 	}
 }
 
-
 void b3FillCL::execute(b3OpenCLArray<unsigned int>& src, const unsigned int value, int n, int offset)
 {
-	b3Assert( n>0 );
+	b3Assert(n > 0);
 
 	{
-		b3BufferInfoCL bInfo[] = { b3BufferInfoCL( src.getBufferCL() ) };
+		b3BufferInfoCL bInfo[] = {b3BufferInfoCL(src.getBufferCL())};
 
-		b3LauncherCL launcher( m_commandQueue, m_fillUnsignedIntKernel,"m_fillUnsignedIntKernel" );
-		launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-		launcher.setConst( n );
-        launcher.setConst(value);
+		b3LauncherCL launcher(m_commandQueue, m_fillUnsignedIntKernel, "m_fillUnsignedIntKernel");
+		launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+		launcher.setConst(n);
+		launcher.setConst(value);
 		launcher.setConst(offset);
 
-		launcher.launch1D( n );
+		launcher.launch1D(n);
 	}
 }
 
-void b3FillCL::executeHost(b3AlignedObjectArray<b3Int2> &src, const b3Int2 &value, int n, int offset)
+void b3FillCL::executeHost(b3AlignedObjectArray<b3Int2>& src, const b3Int2& value, int n, int offset)
 {
-	for (int i=0;i<n;i++)
+	for (int i = 0; i < n; i++)
 	{
-		src[i+offset]=value;
+		src[i + offset] = value;
 	}
 }
 
-void b3FillCL::executeHost(b3AlignedObjectArray<int> &src, const int value, int n, int offset)
+void b3FillCL::executeHost(b3AlignedObjectArray<int>& src, const int value, int n, int offset)
 {
-	for (int i=0;i<n;i++)
+	for (int i = 0; i < n; i++)
 	{
-		src[i+offset]=value;
+		src[i + offset] = value;
 	}
 }
 
-void b3FillCL::execute(b3OpenCLArray<b3Int2> &src, const b3Int2 &value, int n, int offset)
+void b3FillCL::execute(b3OpenCLArray<b3Int2>& src, const b3Int2& value, int n, int offset)
 {
-	b3Assert( n>0 );
-	
+	b3Assert(n > 0);
 
 	{
-		b3BufferInfoCL bInfo[] = { b3BufferInfoCL( src.getBufferCL() ) };
+		b3BufferInfoCL bInfo[] = {b3BufferInfoCL(src.getBufferCL())};
 
-		b3LauncherCL launcher(m_commandQueue, m_fillKernelInt2,"m_fillKernelInt2");
-		launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
+		b3LauncherCL launcher(m_commandQueue, m_fillKernelInt2, "m_fillKernelInt2");
+		launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
 		launcher.setConst(n);
 		launcher.setConst(value);
 		launcher.setConst(offset);
 
 		//( constBuffer );
-		launcher.launch1D( n );
+		launcher.launch1D(n);
 	}
 }
diff --git a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3FillCL.h b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3FillCL.h
index 1609676b9dd..c92c3e51190 100644
--- a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3FillCL.h
+++ b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3FillCL.h
@@ -7,57 +7,46 @@
 #include "Bullet3Common/shared/b3Int2.h"
 #include "Bullet3Common/shared/b3Int4.h"
 
-
 class b3FillCL
 {
-	
-	cl_command_queue	m_commandQueue;
-	
-	cl_kernel			m_fillKernelInt2;
-	cl_kernel			m_fillIntKernel;
-	cl_kernel			m_fillUnsignedIntKernel;
-	cl_kernel			m_fillFloatKernel;
+	cl_command_queue m_commandQueue;
 
-	public:
-		
-		struct b3ConstData
-		{
-			union
-			{
-				b3Int4 m_data;
-				b3UnsignedInt4 m_UnsignedData;
-			};
-			int m_offset;
-			int m_n;
-			int m_padding[2];
-		};
-
-protected:
+	cl_kernel m_fillKernelInt2;
+	cl_kernel m_fillIntKernel;
+	cl_kernel m_fillUnsignedIntKernel;
+	cl_kernel m_fillFloatKernel;
 
 public:
+	struct b3ConstData
+	{
+		union {
+			b3Int4 m_data;
+			b3UnsignedInt4 m_UnsignedData;
+		};
+		int m_offset;
+		int m_n;
+		int m_padding[2];
+	};
 
-		b3FillCL(cl_context ctx, cl_device_id device, cl_command_queue queue);
+protected:
+public:
+	b3FillCL(cl_context ctx, cl_device_id device, cl_command_queue queue);
 
-		virtual ~b3FillCL();
+	virtual ~b3FillCL();
 
-		void execute(b3OpenCLArray<unsigned int>& src, const unsigned int value, int n, int offset = 0);
-	
-		void execute(b3OpenCLArray<int>& src, const int value, int n, int offset = 0);
+	void execute(b3OpenCLArray<unsigned int>& src, const unsigned int value, int n, int offset = 0);
 
-		void execute(b3OpenCLArray<float>& src, const float value, int n, int offset = 0);
+	void execute(b3OpenCLArray<int>& src, const int value, int n, int offset = 0);
 
-		void execute(b3OpenCLArray<b3Int2>& src, const b3Int2& value, int n, int offset = 0);
+	void execute(b3OpenCLArray<float>& src, const float value, int n, int offset = 0);
 
-		void executeHost(b3AlignedObjectArray<b3Int2> &src, const b3Int2 &value, int n, int offset);
+	void execute(b3OpenCLArray<b3Int2>& src, const b3Int2& value, int n, int offset = 0);
 
-		void executeHost(b3AlignedObjectArray<int> &src, const int value, int n, int offset);
+	void executeHost(b3AlignedObjectArray<b3Int2>& src, const b3Int2& value, int n, int offset);
+
+	void executeHost(b3AlignedObjectArray<int>& src, const int value, int n, int offset);
 
 	//	void execute(b3OpenCLArray<b3Int4>& src, const b3Int4& value, int n, int offset = 0);
-
 };
-		
-		
-		
-	
 
-#endif //B3_FILL_CL_H
+#endif  //B3_FILL_CL_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3LauncherCL.cpp b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3LauncherCL.cpp
index 94590d11caf..c97d02eb45a 100644
--- a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3LauncherCL.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3LauncherCL.cpp
@@ -1,13 +1,13 @@
 #include "b3LauncherCL.h"
 
 bool gDebugLauncherCL = false;
-    
+
 b3LauncherCL::b3LauncherCL(cl_command_queue queue, cl_kernel kernel, const char* name)
-:m_commandQueue(queue),
-m_kernel(kernel),
-m_idx(0),
-m_enableSerialization(false),
-m_name(name)
+	: m_commandQueue(queue),
+	  m_kernel(kernel),
+	  m_idx(0),
+	  m_enableSerialization(false),
+	  m_name(name)
 {
 	if (gDebugLauncherCL)
 	{
@@ -15,59 +15,58 @@ m_name(name)
 		printf("[%d] Prepare to launch OpenCL kernel %s\n", counter++, name);
 	}
 
-      m_serializationSizeInBytes = sizeof(int);
+	m_serializationSizeInBytes = sizeof(int);
 }
-    
+
 b3LauncherCL::~b3LauncherCL()
-  {
-      for (int i=0;i<m_arrays.size();i++)
-      {
-		  delete (m_arrays[i]);
-      }
-
-	  m_arrays.clear();
-	  if (gDebugLauncherCL)
-	  {
-		static int counter = 0;
-		printf("[%d] Finished launching OpenCL kernel %s\n", counter++,m_name);
-	  }
-  }
-
-void b3LauncherCL::setBuffer( cl_mem clBuffer)
 {
-		if (m_enableSerialization)
-		{
-			b3KernelArgData kernelArg;
-			kernelArg.m_argIndex = m_idx;
-			kernelArg.m_isBuffer = 1;
-			kernelArg.m_clBuffer = clBuffer;
-		
-			cl_mem_info param_name = CL_MEM_SIZE;
-			size_t param_value;
-			size_t sizeInBytes = sizeof(size_t);
-			size_t actualSizeInBytes;
-			cl_int err;
-			err = clGetMemObjectInfo (	kernelArg.m_clBuffer,
-									  param_name,
-									  sizeInBytes,
-									  &param_value,
-									  &actualSizeInBytes);
-			
-			b3Assert( err == CL_SUCCESS );
-			kernelArg.m_argSizeInBytes = param_value;
-			
-			m_kernelArguments.push_back(kernelArg);
-			m_serializationSizeInBytes+= sizeof(b3KernelArgData);
-			m_serializationSizeInBytes+=param_value;
-            }
-            cl_int status = clSetKernelArg( m_kernel, m_idx++, sizeof(cl_mem), &clBuffer);
-		b3Assert( status == CL_SUCCESS );
+	for (int i = 0; i < m_arrays.size(); i++)
+	{
+		delete (m_arrays[i]);
+	}
+
+	m_arrays.clear();
+	if (gDebugLauncherCL)
+	{
+		static int counter = 0;
+		printf("[%d] Finished launching OpenCL kernel %s\n", counter++, m_name);
+	}
 }
 
-
-void b3LauncherCL::setBuffers( b3BufferInfoCL* buffInfo, int n )
+void b3LauncherCL::setBuffer(cl_mem clBuffer)
 {
-	for(int i=0; i<n; i++)
+	if (m_enableSerialization)
+	{
+		b3KernelArgData kernelArg;
+		kernelArg.m_argIndex = m_idx;
+		kernelArg.m_isBuffer = 1;
+		kernelArg.m_clBuffer = clBuffer;
+
+		cl_mem_info param_name = CL_MEM_SIZE;
+		size_t param_value;
+		size_t sizeInBytes = sizeof(size_t);
+		size_t actualSizeInBytes;
+		cl_int err;
+		err = clGetMemObjectInfo(kernelArg.m_clBuffer,
+								 param_name,
+								 sizeInBytes,
+								 &param_value,
+								 &actualSizeInBytes);
+
+		b3Assert(err == CL_SUCCESS);
+		kernelArg.m_argSizeInBytes = param_value;
+
+		m_kernelArguments.push_back(kernelArg);
+		m_serializationSizeInBytes += sizeof(b3KernelArgData);
+		m_serializationSizeInBytes += param_value;
+	}
+	cl_int status = clSetKernelArg(m_kernel, m_idx++, sizeof(cl_mem), &clBuffer);
+	b3Assert(status == CL_SUCCESS);
+}
+
+void b3LauncherCL::setBuffers(b3BufferInfoCL* buffInfo, int n)
+{
+	for (int i = 0; i < n; i++)
 	{
 		if (m_enableSerialization)
 		{
@@ -75,106 +74,103 @@ void b3LauncherCL::setBuffers( b3BufferInfoCL* buffInfo, int n )
 			kernelArg.m_argIndex = m_idx;
 			kernelArg.m_isBuffer = 1;
 			kernelArg.m_clBuffer = buffInfo[i].m_clBuffer;
-		
+
 			cl_mem_info param_name = CL_MEM_SIZE;
 			size_t param_value;
 			size_t sizeInBytes = sizeof(size_t);
 			size_t actualSizeInBytes;
 			cl_int err;
-			err = clGetMemObjectInfo (	kernelArg.m_clBuffer,
-									  param_name,
-									  sizeInBytes,
-									  &param_value,
-									  &actualSizeInBytes);
-			
-			b3Assert( err == CL_SUCCESS );
+			err = clGetMemObjectInfo(kernelArg.m_clBuffer,
+									 param_name,
+									 sizeInBytes,
+									 &param_value,
+									 &actualSizeInBytes);
+
+			b3Assert(err == CL_SUCCESS);
 			kernelArg.m_argSizeInBytes = param_value;
-			
+
 			m_kernelArguments.push_back(kernelArg);
-			m_serializationSizeInBytes+= sizeof(b3KernelArgData);
-			m_serializationSizeInBytes+=param_value;
-            }
-            cl_int status = clSetKernelArg( m_kernel, m_idx++, sizeof(cl_mem), &buffInfo[i].m_clBuffer);
-		b3Assert( status == CL_SUCCESS );
-        }
+			m_serializationSizeInBytes += sizeof(b3KernelArgData);
+			m_serializationSizeInBytes += param_value;
+		}
+		cl_int status = clSetKernelArg(m_kernel, m_idx++, sizeof(cl_mem), &buffInfo[i].m_clBuffer);
+		b3Assert(status == CL_SUCCESS);
+	}
 }
 
 struct b3KernelArgDataUnaligned
 {
-    int m_isBuffer;
-    int m_argIndex;
-    int m_argSizeInBytes;
+	int m_isBuffer;
+	int m_argIndex;
+	int m_argSizeInBytes;
 	int m_unusedPadding;
-    union
-    {
-        cl_mem m_clBuffer;
-        unsigned char m_argData[B3_CL_MAX_ARG_SIZE];
-    };
-    
+	union {
+		cl_mem m_clBuffer;
+		unsigned char m_argData[B3_CL_MAX_ARG_SIZE];
+	};
 };
 #include <string.h>
 
-
-
 int b3LauncherCL::deserializeArgs(unsigned char* buf, int bufSize, cl_context ctx)
 {
-    int index=0;
-    
-    int numArguments = *(int*) &buf[index];
-    index+=sizeof(int);
-    
-    for (int i=0;i<numArguments;i++)
-    {
-        b3KernelArgDataUnaligned* arg = (b3KernelArgDataUnaligned*)&buf[index];
+	int index = 0;
 
-        index+=sizeof(b3KernelArgData);
-        if (arg->m_isBuffer)
-        {
-            b3OpenCLArray<unsigned char>* clData = new b3OpenCLArray<unsigned char>(ctx,m_commandQueue, arg->m_argSizeInBytes);
-            clData->resize(arg->m_argSizeInBytes);
-            
-            clData->copyFromHostPointer(&buf[index], arg->m_argSizeInBytes);
-            
-            arg->m_clBuffer = clData->getBufferCL();
-            
-            m_arrays.push_back(clData);
-            
-            cl_int status = clSetKernelArg( m_kernel, m_idx++, sizeof(cl_mem), &arg->m_clBuffer);
-		b3Assert( status == CL_SUCCESS );
-            index+=arg->m_argSizeInBytes;
-        } else 
-        {
-            cl_int status = clSetKernelArg( m_kernel, m_idx++, arg->m_argSizeInBytes, &arg->m_argData);
-		b3Assert( status == CL_SUCCESS );
-        }
+	int numArguments = *(int*)&buf[index];
+	index += sizeof(int);
+
+	for (int i = 0; i < numArguments; i++)
+	{
+		b3KernelArgDataUnaligned* arg = (b3KernelArgDataUnaligned*)&buf[index];
+
+		index += sizeof(b3KernelArgData);
+		if (arg->m_isBuffer)
+		{
+			b3OpenCLArray<unsigned char>* clData = new b3OpenCLArray<unsigned char>(ctx, m_commandQueue, arg->m_argSizeInBytes);
+			clData->resize(arg->m_argSizeInBytes);
+
+			clData->copyFromHostPointer(&buf[index], arg->m_argSizeInBytes);
+
+			arg->m_clBuffer = clData->getBufferCL();
+
+			m_arrays.push_back(clData);
+
+			cl_int status = clSetKernelArg(m_kernel, m_idx++, sizeof(cl_mem), &arg->m_clBuffer);
+			b3Assert(status == CL_SUCCESS);
+			index += arg->m_argSizeInBytes;
+		}
+		else
+		{
+			cl_int status = clSetKernelArg(m_kernel, m_idx++, arg->m_argSizeInBytes, &arg->m_argData);
+			b3Assert(status == CL_SUCCESS);
+		}
 		b3KernelArgData b;
-		memcpy(&b,arg,sizeof(b3KernelArgDataUnaligned));
-	m_kernelArguments.push_back(b);
-    }
-m_serializationSizeInBytes = index;
-    return index;
+		memcpy(&b, arg, sizeof(b3KernelArgDataUnaligned));
+		m_kernelArguments.push_back(b);
+	}
+	m_serializationSizeInBytes = index;
+	return index;
 }
 
 int b3LauncherCL::validateResults(unsigned char* goldBuffer, int goldBufferCapacity, cl_context ctx)
-  {
-	 int index=0;
-      
-      int numArguments = *(int*) &goldBuffer[index];
-      index+=sizeof(int);
+{
+	int index = 0;
+
+	int numArguments = *(int*)&goldBuffer[index];
+	index += sizeof(int);
 
 	if (numArguments != m_kernelArguments.size())
 	{
-		printf("failed validation: expected %d arguments, found %d\n",numArguments, m_kernelArguments.size());
+		printf("failed validation: expected %d arguments, found %d\n", numArguments, m_kernelArguments.size());
 		return -1;
 	}
-      
-      for (int ii=0;ii<numArguments;ii++)
-      {
-          b3KernelArgData* argGold = (b3KernelArgData*)&goldBuffer[index];
+
+	for (int ii = 0; ii < numArguments; ii++)
+	{
+		b3KernelArgData* argGold = (b3KernelArgData*)&goldBuffer[index];
 
 		if (m_kernelArguments[ii].m_argSizeInBytes != argGold->m_argSizeInBytes)
 		{
-			printf("failed validation: argument %d sizeInBytes expected: %d, found %d\n",ii, argGold->m_argSizeInBytes, m_kernelArguments[ii].m_argSizeInBytes);
+			printf("failed validation: argument %d sizeInBytes expected: %d, found %d\n", ii, argGold->m_argSizeInBytes, m_kernelArguments[ii].m_argSizeInBytes);
 			return -2;
 		}
 
@@ -184,125 +180,117 @@ int b3LauncherCL::validateResults(unsigned char* goldBuffer, int goldBufferCapac
 
 			if (expected != found)
 			{
-				printf("failed validation: argument %d isBuffer expected: %d, found %d\n",ii,expected, found);
+				printf("failed validation: argument %d isBuffer expected: %d, found %d\n", ii, expected, found);
 				return -3;
 			}
 		}
-		index+=sizeof(b3KernelArgData);
+		index += sizeof(b3KernelArgData);
 
 		if (argGold->m_isBuffer)
-          {
-
-			unsigned char* memBuf= (unsigned char*) malloc(m_kernelArguments[ii].m_argSizeInBytes);
+		{
+			unsigned char* memBuf = (unsigned char*)malloc(m_kernelArguments[ii].m_argSizeInBytes);
 			unsigned char* goldBuf = &goldBuffer[index];
-			for (int j=0;j<m_kernelArguments[j].m_argSizeInBytes;j++)
+			for (int j = 0; j < m_kernelArguments[j].m_argSizeInBytes; j++)
 			{
 				memBuf[j] = 0xaa;
 			}
 
 			cl_int status = 0;
-			status = clEnqueueReadBuffer( m_commandQueue, m_kernelArguments[ii].m_clBuffer, CL_TRUE, 0, m_kernelArguments[ii].m_argSizeInBytes,
-                                           memBuf, 0,0,0 );
-              b3Assert( status==CL_SUCCESS );
-              clFinish(m_commandQueue);
+			status = clEnqueueReadBuffer(m_commandQueue, m_kernelArguments[ii].m_clBuffer, CL_TRUE, 0, m_kernelArguments[ii].m_argSizeInBytes,
+										 memBuf, 0, 0, 0);
+			b3Assert(status == CL_SUCCESS);
+			clFinish(m_commandQueue);
 
-			for (int b=0;b<m_kernelArguments[ii].m_argSizeInBytes;b++)
+			for (int b = 0; b < m_kernelArguments[ii].m_argSizeInBytes; b++)
 			{
 				int expected = goldBuf[b];
 				int found = memBuf[b];
 				if (expected != found)
 				{
 					printf("failed validation: argument %d OpenCL data at byte position %d expected: %d, found %d\n",
-						ii, b, expected, found);
+						   ii, b, expected, found);
 					return -4;
 				}
 			}
 
-              
-              index+=argGold->m_argSizeInBytes;
-          } else 
-          {
-			
+			index += argGold->m_argSizeInBytes;
+		}
+		else
+		{
 			//compare content
-			for (int b=0;b<m_kernelArguments[ii].m_argSizeInBytes;b++)
+			for (int b = 0; b < m_kernelArguments[ii].m_argSizeInBytes; b++)
 			{
 				int expected = argGold->m_argData[b];
-				int found =m_kernelArguments[ii].m_argData[b];
+				int found = m_kernelArguments[ii].m_argData[b];
 				if (expected != found)
 				{
 					printf("failed validation: argument %d const data at byte position %d expected: %d, found %d\n",
-						ii, b, expected, found);
+						   ii, b, expected, found);
 					return -5;
 				}
 			}
-
-          }
-      }
-      return index;
-
+		}
+	}
+	return index;
 }
 
 int b3LauncherCL::serializeArguments(unsigned char* destBuffer, int destBufferCapacity)
 {
-//initialize to known values
-for (int i=0;i<destBufferCapacity;i++)
-	destBuffer[i] = 0xec;
+	//initialize to known values
+	for (int i = 0; i < destBufferCapacity; i++)
+		destBuffer[i] = 0xec;
 
-    assert(destBufferCapacity>=m_serializationSizeInBytes);
-    
-    //todo: use the b3Serializer for this to allow for 32/64bit, endianness etc        
-    int numArguments = m_kernelArguments.size();
-    int curBufferSize = 0;
-    int* dest = (int*)&destBuffer[curBufferSize];
-    *dest = numArguments;
-    curBufferSize += sizeof(int);
-    
-    
-    
-    for (int i=0;i<this->m_kernelArguments.size();i++)
-    {
-        b3KernelArgData* arg = (b3KernelArgData*) &destBuffer[curBufferSize];
-        *arg = m_kernelArguments[i];
-        curBufferSize+=sizeof(b3KernelArgData);
-        if (arg->m_isBuffer==1)
-        {
-            //copy the OpenCL buffer content
-            cl_int status = 0;
-            status = clEnqueueReadBuffer( m_commandQueue, arg->m_clBuffer, 0, 0, arg->m_argSizeInBytes,
-                                         &destBuffer[curBufferSize], 0,0,0 );
-            b3Assert( status==CL_SUCCESS );
-            clFinish(m_commandQueue);
-            curBufferSize+=arg->m_argSizeInBytes;
-        }
-        
-    }
-    return curBufferSize;
+	assert(destBufferCapacity >= m_serializationSizeInBytes);
+
+	//todo: use the b3Serializer for this to allow for 32/64bit, endianness etc
+	int numArguments = m_kernelArguments.size();
+	int curBufferSize = 0;
+	int* dest = (int*)&destBuffer[curBufferSize];
+	*dest = numArguments;
+	curBufferSize += sizeof(int);
+
+	for (int i = 0; i < this->m_kernelArguments.size(); i++)
+	{
+		b3KernelArgData* arg = (b3KernelArgData*)&destBuffer[curBufferSize];
+		*arg = m_kernelArguments[i];
+		curBufferSize += sizeof(b3KernelArgData);
+		if (arg->m_isBuffer == 1)
+		{
+			//copy the OpenCL buffer content
+			cl_int status = 0;
+			status = clEnqueueReadBuffer(m_commandQueue, arg->m_clBuffer, 0, 0, arg->m_argSizeInBytes,
+										 &destBuffer[curBufferSize], 0, 0, 0);
+			b3Assert(status == CL_SUCCESS);
+			clFinish(m_commandQueue);
+			curBufferSize += arg->m_argSizeInBytes;
+		}
+	}
+	return curBufferSize;
 }
 
 void b3LauncherCL::serializeToFile(const char* fileName, int numWorkItems)
 {
 	int num = numWorkItems;
 	int buffSize = getSerializationBufferSize();
-	unsigned char* buf = new unsigned char[buffSize+sizeof(int)];
-	for (int i=0;i<buffSize+1;i++)
+	unsigned char* buf = new unsigned char[buffSize + sizeof(int)];
+	for (int i = 0; i < buffSize + 1; i++)
 	{
 		unsigned char* ptr = (unsigned char*)&buf[i];
 		*ptr = 0xff;
 	}
-//	int actualWrite = serializeArguments(buf,buffSize);
-              
-//	unsigned char* cptr = (unsigned char*)&buf[buffSize];
-//            printf("buf[buffSize] = %d\n",*cptr);
-              
-	assert(buf[buffSize]==0xff);//check for buffer overrun
+	//	int actualWrite = serializeArguments(buf,buffSize);
+
+	//	unsigned char* cptr = (unsigned char*)&buf[buffSize];
+	//            printf("buf[buffSize] = %d\n",*cptr);
+
+	assert(buf[buffSize] == 0xff);  //check for buffer overrun
 	int* ptr = (int*)&buf[buffSize];
-              
+
 	*ptr = num;
-              
-	FILE* f = fopen(fileName,"wb");
-	fwrite(buf,buffSize+sizeof(int),1,f);
+
+	FILE* f = fopen(fileName, "wb");
+	fwrite(buf, buffSize + sizeof(int), 1, f);
 	fclose(f);
 
 	delete[] buf;
-}		
-
+}
diff --git a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3LauncherCL.h b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3LauncherCL.h
index 1b267b31ef9..18e9c1db2ba 100644
--- a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3LauncherCL.h
+++ b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3LauncherCL.h
@@ -9,60 +9,57 @@
 
 #define B3_DEBUG_SERIALIZE_CL
 
-
 #ifdef _WIN32
-#pragma warning(disable :4996)
+#pragma warning(disable : 4996)
 #endif
 #define B3_CL_MAX_ARG_SIZE 16
-B3_ATTRIBUTE_ALIGNED16(struct) b3KernelArgData
+B3_ATTRIBUTE_ALIGNED16(struct)
+b3KernelArgData
 {
-    int m_isBuffer;
-    int m_argIndex;
-    int m_argSizeInBytes;
+	int m_isBuffer;
+	int m_argIndex;
+	int m_argSizeInBytes;
 	int m_unusedPadding;
-    union
-    {
-        cl_mem m_clBuffer;
-        unsigned char m_argData[B3_CL_MAX_ARG_SIZE];
-    };
-    
+	union {
+		cl_mem m_clBuffer;
+		unsigned char m_argData[B3_CL_MAX_ARG_SIZE];
+	};
 };
 
 class b3LauncherCL
 {
-
 	cl_command_queue m_commandQueue;
 	cl_kernel m_kernel;
 	int m_idx;
-	
-    b3AlignedObjectArray<b3KernelArgData> m_kernelArguments;
-    int m_serializationSizeInBytes;
-	bool	m_enableSerialization;
+
+	b3AlignedObjectArray<b3KernelArgData> m_kernelArguments;
+	int m_serializationSizeInBytes;
+	bool m_enableSerialization;
 
 	const char* m_name;
-	public:
 
-     b3AlignedObjectArray<b3OpenCLArray<unsigned char>* > m_arrays;
-    
-		b3LauncherCL(cl_command_queue queue, cl_kernel kernel, const char* name);
-    
-    virtual ~b3LauncherCL();
-    
-		void setBuffer( cl_mem clBuffer);
+public:
+	b3AlignedObjectArray<b3OpenCLArray<unsigned char>*> m_arrays;
 
-		void setBuffers( b3BufferInfoCL* buffInfo, int n );
-    
-    int getSerializationBufferSize() const 
-    {
-        return m_serializationSizeInBytes;
-    }
-    
-    int deserializeArgs(unsigned char* buf, int bufSize, cl_context ctx);
+	b3LauncherCL(cl_command_queue queue, cl_kernel kernel, const char* name);
+
+	virtual ~b3LauncherCL();
+
+	void setBuffer(cl_mem clBuffer);
+
+	void setBuffers(b3BufferInfoCL* buffInfo, int n);
+
+	int getSerializationBufferSize() const
+	{
+		return m_serializationSizeInBytes;
+	}
+
+	int deserializeArgs(unsigned char* buf, int bufSize, cl_context ctx);
 
 	inline int validateResults(unsigned char* goldBuffer, int goldBufferCapacity, cl_context ctx);
 
-    int serializeArguments(unsigned char* destBuffer, int destBufferCapacity);
-    
+	int serializeArguments(unsigned char* destBuffer, int destBufferCapacity);
+
 	int getNumArguments() const
 	{
 		return m_kernelArguments.size();
@@ -75,61 +72,57 @@ class b3LauncherCL
 
 	void serializeToFile(const char* fileName, int numWorkItems);
 
-	template<typename T>
-		inline void setConst( const T& consts )
-		{
-			int sz=sizeof(T);
-			b3Assert(sz<=B3_CL_MAX_ARG_SIZE);
+	template <typename T>
+	inline void setConst(const T& consts)
+	{
+		int sz = sizeof(T);
+		b3Assert(sz <= B3_CL_MAX_ARG_SIZE);
 
-			if (m_enableSerialization)
-			{
-				b3KernelArgData kernelArg;
-				kernelArg.m_argIndex = m_idx;
-				kernelArg.m_isBuffer = 0;
-				T* destArg = (T*)kernelArg.m_argData;
-				*destArg = consts;
-				kernelArg.m_argSizeInBytes = sizeof(T);
-				m_kernelArguments.push_back(kernelArg);
-				m_serializationSizeInBytes+=sizeof(b3KernelArgData);
-			}
-            
-			cl_int status = clSetKernelArg( m_kernel, m_idx++, sz, &consts );
-			b3Assert( status == CL_SUCCESS );
+		if (m_enableSerialization)
+		{
+			b3KernelArgData kernelArg;
+			kernelArg.m_argIndex = m_idx;
+			kernelArg.m_isBuffer = 0;
+			T* destArg = (T*)kernelArg.m_argData;
+			*destArg = consts;
+			kernelArg.m_argSizeInBytes = sizeof(T);
+			m_kernelArguments.push_back(kernelArg);
+			m_serializationSizeInBytes += sizeof(b3KernelArgData);
 		}
 
-		inline void launch1D( int numThreads, int localSize = 64)
-		{
-			launch2D( numThreads, 1, localSize, 1 );
-		}
+		cl_int status = clSetKernelArg(m_kernel, m_idx++, sz, &consts);
+		b3Assert(status == CL_SUCCESS);
+	}
 
-		inline void launch2D( int numThreadsX, int numThreadsY, int localSizeX, int localSizeY )
-		{
-			size_t gRange[3] = {1,1,1};
-			size_t lRange[3] = {1,1,1};
-			lRange[0] = localSizeX;
-			lRange[1] = localSizeY;
-			gRange[0] = b3Max((size_t)1, (numThreadsX/lRange[0])+(!(numThreadsX%lRange[0])?0:1));
-			gRange[0] *= lRange[0];
-			gRange[1] = b3Max((size_t)1, (numThreadsY/lRange[1])+(!(numThreadsY%lRange[1])?0:1));
-			gRange[1] *= lRange[1];
+	inline void launch1D(int numThreads, int localSize = 64)
+	{
+		launch2D(numThreads, 1, localSize, 1);
+	}
 
-			cl_int status = clEnqueueNDRangeKernel( m_commandQueue, 
-				m_kernel, 2, NULL, gRange, lRange, 0,0,0 );
-            if (status != CL_SUCCESS)
-            {
-                printf("Error: OpenCL status = %d\n",status);
-            }
-			b3Assert( status == CL_SUCCESS );
+	inline void launch2D(int numThreadsX, int numThreadsY, int localSizeX, int localSizeY)
+	{
+		size_t gRange[3] = {1, 1, 1};
+		size_t lRange[3] = {1, 1, 1};
+		lRange[0] = localSizeX;
+		lRange[1] = localSizeY;
+		gRange[0] = b3Max((size_t)1, (numThreadsX / lRange[0]) + (!(numThreadsX % lRange[0]) ? 0 : 1));
+		gRange[0] *= lRange[0];
+		gRange[1] = b3Max((size_t)1, (numThreadsY / lRange[1]) + (!(numThreadsY % lRange[1]) ? 0 : 1));
+		gRange[1] *= lRange[1];
 
-		}
-	
-		void	enableSerialization(bool serialize)
+		cl_int status = clEnqueueNDRangeKernel(m_commandQueue,
+											   m_kernel, 2, NULL, gRange, lRange, 0, 0, 0);
+		if (status != CL_SUCCESS)
 		{
-			m_enableSerialization = serialize;
+			printf("Error: OpenCL status = %d\n", status);
 		}
-		
+		b3Assert(status == CL_SUCCESS);
+	}
+
+	void enableSerialization(bool serialize)
+	{
+		m_enableSerialization = serialize;
+	}
 };
 
-
-
-#endif //B3_LAUNCHER_CL_H
+#endif  //B3_LAUNCHER_CL_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3OpenCLArray.h b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3OpenCLArray.h
index d70c30f53f4..e837cceb665 100644
--- a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3OpenCLArray.h
+++ b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3OpenCLArray.h
@@ -7,16 +7,16 @@
 template <typename T>
 class b3OpenCLArray
 {
-	size_t	m_size;
-	size_t	m_capacity;
-	cl_mem	m_clBuffer;
+	size_t m_size;
+	size_t m_capacity;
+	cl_mem m_clBuffer;
 
-	cl_context		 m_clContext;
+	cl_context m_clContext;
 	cl_command_queue m_commandQueue;
 
-	bool	m_ownsMemory;
+	bool m_ownsMemory;
 
-	bool	m_allowGrowingCapacity;
+	bool m_allowGrowingCapacity;
 
 	void deallocate()
 	{
@@ -25,22 +25,19 @@ class b3OpenCLArray
 			clReleaseMemObject(m_clBuffer);
 		}
 		m_clBuffer = 0;
-		m_capacity=0;
+		m_capacity = 0;
 	}
 
 	b3OpenCLArray<T>& operator=(const b3OpenCLArray<T>& src);
 
-	B3_FORCE_INLINE	size_t	allocSize(size_t size)
-		{
-			return (size ? size*2 : 1);
-		}
+	B3_FORCE_INLINE size_t allocSize(size_t size)
+	{
+		return (size ? size * 2 : 1);
+	}
 
 public:
-
-	b3OpenCLArray(cl_context ctx, cl_command_queue queue, size_t initialCapacity=0, bool allowGrowingCapacity=true)
-	:m_size(0),  m_capacity(0),m_clBuffer(0),
-	m_clContext(ctx),m_commandQueue(queue),
-	m_ownsMemory(true),m_allowGrowingCapacity(true)
+	b3OpenCLArray(cl_context ctx, cl_command_queue queue, size_t initialCapacity = 0, bool allowGrowingCapacity = true)
+		: m_size(0), m_capacity(0), m_clBuffer(0), m_clContext(ctx), m_commandQueue(queue), m_ownsMemory(true), m_allowGrowingCapacity(true)
 	{
 		if (initialCapacity)
 		{
@@ -60,34 +57,32 @@ public:
 		m_capacity = sizeInElements;
 	}
 
-// we could enable this assignment, but need to make sure to avoid accidental deep copies
-//	b3OpenCLArray<T>& operator=(const b3AlignedObjectArray<T>& src)
-//	{
-//		copyFromArray(src);
-//		return *this;
-//	}
+	// we could enable this assignment, but need to make sure to avoid accidental deep copies
+	//	b3OpenCLArray<T>& operator=(const b3AlignedObjectArray<T>& src)
+	//	{
+	//		copyFromArray(src);
+	//		return *this;
+	//	}
 
-
-	cl_mem	getBufferCL() const
+	cl_mem getBufferCL() const
 	{
 		return m_clBuffer;
 	}
 
-
 	virtual ~b3OpenCLArray()
 	{
 		deallocate();
-		m_size=0;
-		m_capacity=0;
+		m_size = 0;
+		m_capacity = 0;
 	}
 
-	B3_FORCE_INLINE	bool push_back(const T& _Val,bool waitForCompletion=true)
+	B3_FORCE_INLINE bool push_back(const T& _Val, bool waitForCompletion = true)
 	{
 		bool result = true;
 		size_t sz = size();
-		if( sz == capacity() )
+		if (sz == capacity())
 		{
-			result = reserve( allocSize(size()) );
+			result = reserve(allocSize(size()));
 		}
 		copyFromHostPointer(&_Val, 1, sz, waitForCompletion);
 		m_size++;
@@ -96,23 +91,23 @@ public:
 
 	B3_FORCE_INLINE T forcedAt(size_t n) const
 	{
-		b3Assert(n>=0);
-		b3Assert(n<capacity());
+		b3Assert(n >= 0);
+		b3Assert(n < capacity());
 		T elem;
-		copyToHostPointer(&elem,1,n,true);
+		copyToHostPointer(&elem, 1, n, true);
 		return elem;
 	}
 
 	B3_FORCE_INLINE T at(size_t n) const
 	{
-		b3Assert(n>=0);
-		b3Assert(n<size());
+		b3Assert(n >= 0);
+		b3Assert(n < size());
 		T elem;
-		copyToHostPointer(&elem,1,n,true);
+		copyToHostPointer(&elem, 1, n, true);
 		return elem;
 	}
 
-	B3_FORCE_INLINE	bool resize(size_t newsize, bool copyOldContents=true)
+	B3_FORCE_INLINE bool resize(size_t newsize, bool copyOldContents = true)
 	{
 		bool result = true;
 		size_t curSize = size();
@@ -120,11 +115,12 @@ public:
 		if (newsize < curSize)
 		{
 			//leave the OpenCL memory for now
-		} else
+		}
+		else
 		{
 			if (newsize > size())
 			{
-				result = reserve(newsize,copyOldContents);
+				result = reserve(newsize, copyOldContents);
 			}
 
 			//leave new data uninitialized (init in debug mode?)
@@ -134,7 +130,8 @@ public:
 		if (result)
 		{
 			m_size = newsize;
-		} else
+		}
+		else
 		{
 			m_size = 0;
 		}
@@ -146,25 +143,25 @@ public:
 		return m_size;
 	}
 
-	B3_FORCE_INLINE	size_t capacity() const
+	B3_FORCE_INLINE size_t capacity() const
 	{
 		return m_capacity;
 	}
 
-	B3_FORCE_INLINE	bool reserve(size_t _Count, bool copyOldContents=true)
+	B3_FORCE_INLINE bool reserve(size_t _Count, bool copyOldContents = true)
 	{
-		bool result=true;
+		bool result = true;
 		// determine new minimum length of allocated storage
 		if (capacity() < _Count)
-		{	// not enough room, reallocate
+		{  // not enough room, reallocate
 
 			if (m_allowGrowingCapacity)
 			{
 				cl_int ciErrNum;
 				//create a new OpenCL buffer
-				size_t memSizeInBytes = sizeof(T)*_Count;
+				size_t memSizeInBytes = sizeof(T) * _Count;
 				cl_mem buf = clCreateBuffer(m_clContext, CL_MEM_READ_WRITE, memSizeInBytes, NULL, &ciErrNum);
-				if (ciErrNum!=CL_SUCCESS)
+				if (ciErrNum != CL_SUCCESS)
 				{
 					b3Error("OpenCL out-of-memory\n");
 					_Count = 0;
@@ -173,13 +170,13 @@ public:
 //#define B3_ALWAYS_INITIALIZE_OPENCL_BUFFERS
 #ifdef B3_ALWAYS_INITIALIZE_OPENCL_BUFFERS
 				unsigned char* src = (unsigned char*)malloc(memSizeInBytes);
-				for (size_t i=0;i<memSizeInBytes;i++)
+				for (size_t i = 0; i < memSizeInBytes; i++)
 					src[i] = 0xbb;
-				ciErrNum = clEnqueueWriteBuffer( m_commandQueue, buf, CL_TRUE, 0, memSizeInBytes, src, 0,0,0 );
-				b3Assert(ciErrNum==CL_SUCCESS);
+				ciErrNum = clEnqueueWriteBuffer(m_commandQueue, buf, CL_TRUE, 0, memSizeInBytes, src, 0, 0, 0);
+				b3Assert(ciErrNum == CL_SUCCESS);
 				clFinish(m_commandQueue);
 				free(src);
-#endif //B3_ALWAYS_INITIALIZE_OPENCL_BUFFERS
+#endif  //B3_ALWAYS_INITIALIZE_OPENCL_BUFFERS
 
 				if (result)
 				{
@@ -193,21 +190,21 @@ public:
 				m_clBuffer = buf;
 
 				m_capacity = _Count;
-			} else
+			}
+			else
 			{
 				//fail: assert and
 				b3Assert(0);
 				deallocate();
-				result=false;
+				result = false;
 			}
 		}
 		return result;
 	}
 
-
-	void copyToCL(cl_mem destination, size_t numElements, size_t firstElem=0, size_t dstOffsetInElems=0) const
+	void copyToCL(cl_mem destination, size_t numElements, size_t firstElem = 0, size_t dstOffsetInElems = 0) const
 	{
-		if (numElements<=0)
+		if (numElements <= 0)
 			return;
 
 		b3Assert(m_clBuffer);
@@ -216,75 +213,74 @@ public:
 		//likely some error, destination is same as source
 		b3Assert(m_clBuffer != destination);
 
-		b3Assert((firstElem+numElements)<=m_size);
+		b3Assert((firstElem + numElements) <= m_size);
 
 		cl_int status = 0;
 
+		b3Assert(numElements > 0);
+		b3Assert(numElements <= m_size);
 
-		b3Assert(numElements>0);
-		b3Assert(numElements<=m_size);
+		size_t srcOffsetBytes = sizeof(T) * firstElem;
+		size_t dstOffsetInBytes = sizeof(T) * dstOffsetInElems;
 
-		size_t srcOffsetBytes = sizeof(T)*firstElem;
-		size_t dstOffsetInBytes = sizeof(T)*dstOffsetInElems;
+		status = clEnqueueCopyBuffer(m_commandQueue, m_clBuffer, destination,
+									 srcOffsetBytes, dstOffsetInBytes, sizeof(T) * numElements, 0, 0, 0);
 
-		status = clEnqueueCopyBuffer( m_commandQueue, m_clBuffer, destination,
-			srcOffsetBytes, dstOffsetInBytes, sizeof(T)*numElements, 0, 0, 0 );
-
-		b3Assert( status == CL_SUCCESS );
+		b3Assert(status == CL_SUCCESS);
 	}
 
-	void copyFromHost(const b3AlignedObjectArray<T>& srcArray, bool waitForCompletion=true)
+	void copyFromHost(const b3AlignedObjectArray<T>& srcArray, bool waitForCompletion = true)
 	{
 		size_t newSize = srcArray.size();
 
 		bool copyOldContents = false;
-		resize (newSize,copyOldContents);
+		resize(newSize, copyOldContents);
 		if (newSize)
-			copyFromHostPointer(&srcArray[0],newSize,0,waitForCompletion);
-
+			copyFromHostPointer(&srcArray[0], newSize, 0, waitForCompletion);
 	}
 
-	void copyFromHostPointer(const T* src, size_t numElems, size_t destFirstElem= 0, bool waitForCompletion=true)
+	void copyFromHostPointer(const T* src, size_t numElems, size_t destFirstElem = 0, bool waitForCompletion = true)
 	{
-		b3Assert(numElems+destFirstElem <= capacity());
+		b3Assert(numElems + destFirstElem <= capacity());
 
-		if (numElems+destFirstElem)
+		if (numElems + destFirstElem)
 		{
 			cl_int status = 0;
-			size_t sizeInBytes=sizeof(T)*numElems;
-			status = clEnqueueWriteBuffer( m_commandQueue, m_clBuffer, 0, sizeof(T)*destFirstElem, sizeInBytes,
-			src, 0,0,0 );
-			b3Assert(status == CL_SUCCESS );
+			size_t sizeInBytes = sizeof(T) * numElems;
+			status = clEnqueueWriteBuffer(m_commandQueue, m_clBuffer, 0, sizeof(T) * destFirstElem, sizeInBytes,
+										  src, 0, 0, 0);
+			b3Assert(status == CL_SUCCESS);
 			if (waitForCompletion)
 				clFinish(m_commandQueue);
-		} else
+		}
+		else
 		{
 			b3Error("copyFromHostPointer invalid range\n");
 		}
 	}
 
-
-	void copyToHost(b3AlignedObjectArray<T>& destArray, bool waitForCompletion=true) const
+	void copyToHost(b3AlignedObjectArray<T>& destArray, bool waitForCompletion = true) const
 	{
 		destArray.resize(this->size());
 		if (size())
-			copyToHostPointer(&destArray[0], size(),0,waitForCompletion);
+			copyToHostPointer(&destArray[0], size(), 0, waitForCompletion);
 	}
 
-	void copyToHostPointer(T* destPtr, size_t numElem, size_t srcFirstElem=0, bool waitForCompletion=true) const
+	void copyToHostPointer(T* destPtr, size_t numElem, size_t srcFirstElem = 0, bool waitForCompletion = true) const
 	{
-		b3Assert(numElem+srcFirstElem <= capacity());
+		b3Assert(numElem + srcFirstElem <= capacity());
 
-		if(numElem+srcFirstElem <= capacity())
+		if (numElem + srcFirstElem <= capacity())
 		{
 			cl_int status = 0;
-			status = clEnqueueReadBuffer( m_commandQueue, m_clBuffer, 0, sizeof(T)*srcFirstElem, sizeof(T)*numElem,
-			destPtr, 0,0,0 );
-			b3Assert( status==CL_SUCCESS );
+			status = clEnqueueReadBuffer(m_commandQueue, m_clBuffer, 0, sizeof(T) * srcFirstElem, sizeof(T) * numElem,
+										 destPtr, 0, 0, 0);
+			b3Assert(status == CL_SUCCESS);
 
 			if (waitForCompletion)
 				clFinish(m_commandQueue);
-		} else
+		}
+		else
 		{
 			b3Error("copyToHostPointer invalid range\n");
 		}
@@ -296,11 +292,9 @@ public:
 		resize(newSize);
 		if (size())
 		{
-			src.copyToCL(m_clBuffer,size());
+			src.copyToCL(m_clBuffer, size());
 		}
 	}
-
 };
 
-
-#endif //B3_OPENCL_ARRAY_H
+#endif  //B3_OPENCL_ARRAY_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3PrefixScanCL.cpp b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3PrefixScanCL.cpp
index 42cd197740a..822b5116334 100644
--- a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3PrefixScanCL.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3PrefixScanCL.cpp
@@ -7,25 +7,24 @@
 #include "kernels/PrefixScanKernelsCL.h"
 
 b3PrefixScanCL::b3PrefixScanCL(cl_context ctx, cl_device_id device, cl_command_queue queue, int size)
-:m_commandQueue(queue)
+	: m_commandQueue(queue)
 {
 	const char* scanKernelSource = prefixScanKernelsCL;
 	cl_int pErrNum;
-	char* additionalMacros=0;
+	char* additionalMacros = 0;
 
-	m_workBuffer = new b3OpenCLArray<unsigned int>(ctx,queue,size);
-	cl_program scanProg = b3OpenCLUtils::compileCLProgramFromString( ctx, device, scanKernelSource, &pErrNum,additionalMacros, B3_PREFIXSCAN_PROG_PATH);
+	m_workBuffer = new b3OpenCLArray<unsigned int>(ctx, queue, size);
+	cl_program scanProg = b3OpenCLUtils::compileCLProgramFromString(ctx, device, scanKernelSource, &pErrNum, additionalMacros, B3_PREFIXSCAN_PROG_PATH);
 	b3Assert(scanProg);
 
-	m_localScanKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, scanKernelSource, "LocalScanKernel", &pErrNum, scanProg,additionalMacros );
-	b3Assert(m_localScanKernel );
-	m_blockSumKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, scanKernelSource, "TopLevelScanKernel", &pErrNum, scanProg,additionalMacros );
-	b3Assert(m_blockSumKernel );
-	m_propagationKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, scanKernelSource, "AddOffsetKernel", &pErrNum, scanProg,additionalMacros );
-	b3Assert(m_propagationKernel );
+	m_localScanKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, scanKernelSource, "LocalScanKernel", &pErrNum, scanProg, additionalMacros);
+	b3Assert(m_localScanKernel);
+	m_blockSumKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, scanKernelSource, "TopLevelScanKernel", &pErrNum, scanProg, additionalMacros);
+	b3Assert(m_blockSumKernel);
+	m_propagationKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, scanKernelSource, "AddOffsetKernel", &pErrNum, scanProg, additionalMacros);
+	b3Assert(m_propagationKernel);
 }
 
-
 b3PrefixScanCL::~b3PrefixScanCL()
 {
 	delete m_workBuffer;
@@ -34,20 +33,19 @@ b3PrefixScanCL::~b3PrefixScanCL()
 	clReleaseKernel(m_propagationKernel);
 }
 
-template<class T>
+template <class T>
 T b3NextPowerOf2(T n)
 {
 	n -= 1;
-	for(int i=0; i<sizeof(T)*8; i++)
-		n = n | (n>>i);
-	return n+1;
+	for (int i = 0; i < sizeof(T) * 8; i++)
+		n = n | (n >> i);
+	return n + 1;
 }
 
 void b3PrefixScanCL::execute(b3OpenCLArray<unsigned int>& src, b3OpenCLArray<unsigned int>& dst, int n, unsigned int* sum)
 {
-	
-//	b3Assert( data->m_option == EXCLUSIVE );
-	const unsigned int numBlocks = (const unsigned int)( (n+BLOCK_SIZE*2-1)/(BLOCK_SIZE*2) );
+	//	b3Assert( data->m_option == EXCLUSIVE );
+	const unsigned int numBlocks = (const unsigned int)((n + BLOCK_SIZE * 2 - 1) / (BLOCK_SIZE * 2));
 
 	dst.resize(src.size());
 	m_workBuffer->resize(src.size());
@@ -55,55 +53,51 @@ void b3PrefixScanCL::execute(b3OpenCLArray<unsigned int>& src, b3OpenCLArray<uns
 	b3Int4 constBuffer;
 	constBuffer.x = n;
 	constBuffer.y = numBlocks;
-	constBuffer.z = (int)b3NextPowerOf2( numBlocks );
+	constBuffer.z = (int)b3NextPowerOf2(numBlocks);
 
 	b3OpenCLArray<unsigned int>* srcNative = &src;
 	b3OpenCLArray<unsigned int>* dstNative = &dst;
-	
-	{
-		b3BufferInfoCL bInfo[] = { b3BufferInfoCL( dstNative->getBufferCL() ), b3BufferInfoCL( srcNative->getBufferCL() ), b3BufferInfoCL( m_workBuffer->getBufferCL() ) };
 
-		b3LauncherCL launcher( m_commandQueue, m_localScanKernel,"m_localScanKernel" );
-		launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-		launcher.setConst(  constBuffer );
-		launcher.launch1D( numBlocks*BLOCK_SIZE, BLOCK_SIZE );
+	{
+		b3BufferInfoCL bInfo[] = {b3BufferInfoCL(dstNative->getBufferCL()), b3BufferInfoCL(srcNative->getBufferCL()), b3BufferInfoCL(m_workBuffer->getBufferCL())};
+
+		b3LauncherCL launcher(m_commandQueue, m_localScanKernel, "m_localScanKernel");
+		launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+		launcher.setConst(constBuffer);
+		launcher.launch1D(numBlocks * BLOCK_SIZE, BLOCK_SIZE);
 	}
 
 	{
-		b3BufferInfoCL bInfo[] = { b3BufferInfoCL( m_workBuffer->getBufferCL() ) };
+		b3BufferInfoCL bInfo[] = {b3BufferInfoCL(m_workBuffer->getBufferCL())};
 
-		b3LauncherCL launcher( m_commandQueue, m_blockSumKernel,"m_blockSumKernel" );
-		launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-		launcher.setConst( constBuffer );
-		launcher.launch1D( BLOCK_SIZE, BLOCK_SIZE );
+		b3LauncherCL launcher(m_commandQueue, m_blockSumKernel, "m_blockSumKernel");
+		launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+		launcher.setConst(constBuffer);
+		launcher.launch1D(BLOCK_SIZE, BLOCK_SIZE);
 	}
-	
 
-	if( numBlocks > 1 )
+	if (numBlocks > 1)
 	{
-		b3BufferInfoCL bInfo[] = { b3BufferInfoCL( dstNative->getBufferCL() ), b3BufferInfoCL( m_workBuffer->getBufferCL() ) };
-		b3LauncherCL launcher( m_commandQueue, m_propagationKernel,"m_propagationKernel" );
-		launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-		launcher.setConst( constBuffer );
-		launcher.launch1D( (numBlocks-1)*BLOCK_SIZE, BLOCK_SIZE );
+		b3BufferInfoCL bInfo[] = {b3BufferInfoCL(dstNative->getBufferCL()), b3BufferInfoCL(m_workBuffer->getBufferCL())};
+		b3LauncherCL launcher(m_commandQueue, m_propagationKernel, "m_propagationKernel");
+		launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+		launcher.setConst(constBuffer);
+		launcher.launch1D((numBlocks - 1) * BLOCK_SIZE, BLOCK_SIZE);
 	}
 
-
-	if( sum )
+	if (sum)
 	{
 		clFinish(m_commandQueue);
-		dstNative->copyToHostPointer(sum,1,n-1,true);
+		dstNative->copyToHostPointer(sum, 1, n - 1, true);
 	}
-
 }
 
-
 void b3PrefixScanCL::executeHost(b3AlignedObjectArray<unsigned int>& src, b3AlignedObjectArray<unsigned int>& dst, int n, unsigned int* sum)
 {
 	unsigned int s = 0;
 	//if( data->m_option == EXCLUSIVE )
 	{
-		for(int i=0; i<n; i++)
+		for (int i = 0; i < n; i++)
 		{
 			dst[i] = s;
 			s += src[i];
@@ -119,8 +113,8 @@ void b3PrefixScanCL::executeHost(b3AlignedObjectArray<unsigned int>& src, b3Alig
 	}
 	*/
 
-	if( sum )
+	if (sum)
 	{
-		*sum = dst[n-1];
+		*sum = dst[n - 1];
 	}
 }
\ No newline at end of file
diff --git a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3PrefixScanCL.h b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3PrefixScanCL.h
index a9a2e61b9eb..346efa0c736 100644
--- a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3PrefixScanCL.h
+++ b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3PrefixScanCL.h
@@ -13,9 +13,9 @@ class b3PrefixScanCL
 		BLOCK_SIZE = 128
 	};
 
-//	Option m_option;
+	//	Option m_option;
 
-	cl_command_queue	m_commandQueue;
+	cl_command_queue m_commandQueue;
 
 	cl_kernel m_localScanKernel;
 	cl_kernel m_blockSumKernel;
@@ -23,15 +23,13 @@ class b3PrefixScanCL
 
 	b3OpenCLArray<unsigned int>* m_workBuffer;
 
-
-	public:
-		
-	b3PrefixScanCL(cl_context ctx, cl_device_id device, cl_command_queue queue,int size=0);
+public:
+	b3PrefixScanCL(cl_context ctx, cl_device_id device, cl_command_queue queue, int size = 0);
 
 	virtual ~b3PrefixScanCL();
 
 	void execute(b3OpenCLArray<unsigned int>& src, b3OpenCLArray<unsigned int>& dst, int n, unsigned int* sum = 0);
-	void executeHost(b3AlignedObjectArray<unsigned int>& src, b3AlignedObjectArray<unsigned int>& dst, int n, unsigned int* sum=0);
+	void executeHost(b3AlignedObjectArray<unsigned int>& src, b3AlignedObjectArray<unsigned int>& dst, int n, unsigned int* sum = 0);
 };
 
-#endif //B3_PREFIX_SCAN_CL_H
+#endif  //B3_PREFIX_SCAN_CL_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3PrefixScanFloat4CL.cpp b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3PrefixScanFloat4CL.cpp
index 80560d793de..1cac97c9888 100644
--- a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3PrefixScanFloat4CL.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3PrefixScanFloat4CL.cpp
@@ -7,25 +7,24 @@
 #include "kernels/PrefixScanKernelsFloat4CL.h"
 
 b3PrefixScanFloat4CL::b3PrefixScanFloat4CL(cl_context ctx, cl_device_id device, cl_command_queue queue, int size)
-:m_commandQueue(queue)
+	: m_commandQueue(queue)
 {
 	const char* scanKernelSource = prefixScanKernelsFloat4CL;
 	cl_int pErrNum;
-	char* additionalMacros=0;
+	char* additionalMacros = 0;
 
-	m_workBuffer = new b3OpenCLArray<b3Vector3>(ctx,queue,size);
-	cl_program scanProg = b3OpenCLUtils::compileCLProgramFromString( ctx, device, scanKernelSource, &pErrNum,additionalMacros, B3_PREFIXSCAN_FLOAT4_PROG_PATH);
+	m_workBuffer = new b3OpenCLArray<b3Vector3>(ctx, queue, size);
+	cl_program scanProg = b3OpenCLUtils::compileCLProgramFromString(ctx, device, scanKernelSource, &pErrNum, additionalMacros, B3_PREFIXSCAN_FLOAT4_PROG_PATH);
 	b3Assert(scanProg);
 
-	m_localScanKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, scanKernelSource, "LocalScanKernel", &pErrNum, scanProg,additionalMacros );
-	b3Assert(m_localScanKernel );
-	m_blockSumKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, scanKernelSource, "TopLevelScanKernel", &pErrNum, scanProg,additionalMacros );
-	b3Assert(m_blockSumKernel );
-	m_propagationKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, scanKernelSource, "AddOffsetKernel", &pErrNum, scanProg,additionalMacros );
-	b3Assert(m_propagationKernel );
+	m_localScanKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, scanKernelSource, "LocalScanKernel", &pErrNum, scanProg, additionalMacros);
+	b3Assert(m_localScanKernel);
+	m_blockSumKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, scanKernelSource, "TopLevelScanKernel", &pErrNum, scanProg, additionalMacros);
+	b3Assert(m_blockSumKernel);
+	m_propagationKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, scanKernelSource, "AddOffsetKernel", &pErrNum, scanProg, additionalMacros);
+	b3Assert(m_propagationKernel);
 }
 
-
 b3PrefixScanFloat4CL::~b3PrefixScanFloat4CL()
 {
 	delete m_workBuffer;
@@ -34,20 +33,19 @@ b3PrefixScanFloat4CL::~b3PrefixScanFloat4CL()
 	clReleaseKernel(m_propagationKernel);
 }
 
-template<class T>
+template <class T>
 T b3NextPowerOf2(T n)
 {
 	n -= 1;
-	for(int i=0; i<sizeof(T)*8; i++)
-		n = n | (n>>i);
-	return n+1;
+	for (int i = 0; i < sizeof(T) * 8; i++)
+		n = n | (n >> i);
+	return n + 1;
 }
 
 void b3PrefixScanFloat4CL::execute(b3OpenCLArray<b3Vector3>& src, b3OpenCLArray<b3Vector3>& dst, int n, b3Vector3* sum)
 {
-	
-//	b3Assert( data->m_option == EXCLUSIVE );
-	const unsigned int numBlocks = (const unsigned int)( (n+BLOCK_SIZE*2-1)/(BLOCK_SIZE*2) );
+	//	b3Assert( data->m_option == EXCLUSIVE );
+	const unsigned int numBlocks = (const unsigned int)((n + BLOCK_SIZE * 2 - 1) / (BLOCK_SIZE * 2));
 
 	dst.resize(src.size());
 	m_workBuffer->resize(src.size());
@@ -55,55 +53,51 @@ void b3PrefixScanFloat4CL::execute(b3OpenCLArray<b3Vector3>& src, b3OpenCLArray<
 	b3Int4 constBuffer;
 	constBuffer.x = n;
 	constBuffer.y = numBlocks;
-	constBuffer.z = (int)b3NextPowerOf2( numBlocks );
+	constBuffer.z = (int)b3NextPowerOf2(numBlocks);
 
 	b3OpenCLArray<b3Vector3>* srcNative = &src;
 	b3OpenCLArray<b3Vector3>* dstNative = &dst;
-	
-	{
-		b3BufferInfoCL bInfo[] = { b3BufferInfoCL( dstNative->getBufferCL() ), b3BufferInfoCL( srcNative->getBufferCL() ), b3BufferInfoCL( m_workBuffer->getBufferCL() ) };
 
-		b3LauncherCL launcher( m_commandQueue, m_localScanKernel ,"m_localScanKernel");
-		launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-		launcher.setConst(  constBuffer );
-		launcher.launch1D( numBlocks*BLOCK_SIZE, BLOCK_SIZE );
+	{
+		b3BufferInfoCL bInfo[] = {b3BufferInfoCL(dstNative->getBufferCL()), b3BufferInfoCL(srcNative->getBufferCL()), b3BufferInfoCL(m_workBuffer->getBufferCL())};
+
+		b3LauncherCL launcher(m_commandQueue, m_localScanKernel, "m_localScanKernel");
+		launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+		launcher.setConst(constBuffer);
+		launcher.launch1D(numBlocks * BLOCK_SIZE, BLOCK_SIZE);
 	}
 
 	{
-		b3BufferInfoCL bInfo[] = { b3BufferInfoCL( m_workBuffer->getBufferCL() ) };
+		b3BufferInfoCL bInfo[] = {b3BufferInfoCL(m_workBuffer->getBufferCL())};
 
-		b3LauncherCL launcher( m_commandQueue, m_blockSumKernel ,"m_blockSumKernel");
-		launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-		launcher.setConst( constBuffer );
-		launcher.launch1D( BLOCK_SIZE, BLOCK_SIZE );
+		b3LauncherCL launcher(m_commandQueue, m_blockSumKernel, "m_blockSumKernel");
+		launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+		launcher.setConst(constBuffer);
+		launcher.launch1D(BLOCK_SIZE, BLOCK_SIZE);
 	}
-	
 
-	if( numBlocks > 1 )
+	if (numBlocks > 1)
 	{
-		b3BufferInfoCL bInfo[] = { b3BufferInfoCL( dstNative->getBufferCL() ), b3BufferInfoCL( m_workBuffer->getBufferCL() ) };
-		b3LauncherCL launcher( m_commandQueue, m_propagationKernel ,"m_propagationKernel");
-		launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-		launcher.setConst( constBuffer );
-		launcher.launch1D( (numBlocks-1)*BLOCK_SIZE, BLOCK_SIZE );
+		b3BufferInfoCL bInfo[] = {b3BufferInfoCL(dstNative->getBufferCL()), b3BufferInfoCL(m_workBuffer->getBufferCL())};
+		b3LauncherCL launcher(m_commandQueue, m_propagationKernel, "m_propagationKernel");
+		launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+		launcher.setConst(constBuffer);
+		launcher.launch1D((numBlocks - 1) * BLOCK_SIZE, BLOCK_SIZE);
 	}
 
-
-	if( sum )
+	if (sum)
 	{
 		clFinish(m_commandQueue);
-		dstNative->copyToHostPointer(sum,1,n-1,true);
+		dstNative->copyToHostPointer(sum, 1, n - 1, true);
 	}
-
 }
 
-
 void b3PrefixScanFloat4CL::executeHost(b3AlignedObjectArray<b3Vector3>& src, b3AlignedObjectArray<b3Vector3>& dst, int n, b3Vector3* sum)
 {
-	b3Vector3 s=b3MakeVector3(0,0,0);
+	b3Vector3 s = b3MakeVector3(0, 0, 0);
 	//if( data->m_option == EXCLUSIVE )
 	{
-		for(int i=0; i<n; i++)
+		for (int i = 0; i < n; i++)
 		{
 			dst[i] = s;
 			s += src[i];
@@ -119,8 +113,8 @@ void b3PrefixScanFloat4CL::executeHost(b3AlignedObjectArray<b3Vector3>& src, b3A
 	}
 	*/
 
-	if( sum )
+	if (sum)
 	{
-		*sum = dst[n-1];
+		*sum = dst[n - 1];
 	}
 }
\ No newline at end of file
diff --git a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3PrefixScanFloat4CL.h b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3PrefixScanFloat4CL.h
index 2c8003c1bba..122b0bfd68b 100644
--- a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3PrefixScanFloat4CL.h
+++ b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3PrefixScanFloat4CL.h
@@ -14,9 +14,9 @@ class b3PrefixScanFloat4CL
 		BLOCK_SIZE = 128
 	};
 
-//	Option m_option;
+	//	Option m_option;
 
-	cl_command_queue	m_commandQueue;
+	cl_command_queue m_commandQueue;
 
 	cl_kernel m_localScanKernel;
 	cl_kernel m_blockSumKernel;
@@ -24,10 +24,8 @@ class b3PrefixScanFloat4CL
 
 	b3OpenCLArray<b3Vector3>* m_workBuffer;
 
-
-	public:
-		
-	b3PrefixScanFloat4CL(cl_context ctx, cl_device_id device, cl_command_queue queue,int size=0);
+public:
+	b3PrefixScanFloat4CL(cl_context ctx, cl_device_id device, cl_command_queue queue, int size = 0);
 
 	virtual ~b3PrefixScanFloat4CL();
 
@@ -35,4 +33,4 @@ class b3PrefixScanFloat4CL
 	void executeHost(b3AlignedObjectArray<b3Vector3>& src, b3AlignedObjectArray<b3Vector3>& dst, int n, b3Vector3* sum);
 };
 
-#endif //B3_PREFIX_SCAN_CL_H
+#endif  //B3_PREFIX_SCAN_CL_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3RadixSort32CL.cpp b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3RadixSort32CL.cpp
index f11ae4bcdb6..e86af6583f1 100644
--- a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3RadixSort32CL.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3RadixSort32CL.cpp
@@ -10,21 +10,20 @@
 #include "kernels/RadixSort32KernelsCL.h"
 
 b3RadixSort32CL::b3RadixSort32CL(cl_context ctx, cl_device_id device, cl_command_queue queue, int initialCapacity)
-:m_commandQueue(queue)
+	: m_commandQueue(queue)
 {
 	b3OpenCLDeviceInfo info;
-	b3OpenCLUtils::getDeviceInfo(device,&info);
-	m_deviceCPU = (info.m_deviceType & CL_DEVICE_TYPE_CPU)!=0;
+	b3OpenCLUtils::getDeviceInfo(device, &info);
+	m_deviceCPU = (info.m_deviceType & CL_DEVICE_TYPE_CPU) != 0;
 
-	m_workBuffer1 = new b3OpenCLArray<unsigned int>(ctx,queue);
-	m_workBuffer2 = new b3OpenCLArray<unsigned int>(ctx,queue);
-	m_workBuffer3 = new b3OpenCLArray<b3SortData>(ctx,queue);
-	m_workBuffer3a = new b3OpenCLArray<unsigned int>(ctx,queue);
-	m_workBuffer4 = new b3OpenCLArray<b3SortData>(ctx,queue);
-	m_workBuffer4a = new b3OpenCLArray<unsigned int>(ctx,queue);
+	m_workBuffer1 = new b3OpenCLArray<unsigned int>(ctx, queue);
+	m_workBuffer2 = new b3OpenCLArray<unsigned int>(ctx, queue);
+	m_workBuffer3 = new b3OpenCLArray<b3SortData>(ctx, queue);
+	m_workBuffer3a = new b3OpenCLArray<unsigned int>(ctx, queue);
+	m_workBuffer4 = new b3OpenCLArray<b3SortData>(ctx, queue);
+	m_workBuffer4a = new b3OpenCLArray<unsigned int>(ctx, queue);
 
-
-	if (initialCapacity>0)
+	if (initialCapacity > 0)
 	{
 		m_workBuffer1->resize(initialCapacity);
 		m_workBuffer3->resize(initialCapacity);
@@ -33,45 +32,40 @@ b3RadixSort32CL::b3RadixSort32CL(cl_context ctx, cl_device_id device, cl_command
 		m_workBuffer4a->resize(initialCapacity);
 	}
 
-	m_scan = new b3PrefixScanCL(ctx,device,queue);
-	m_fill = new b3FillCL(ctx,device,queue);
-	
+	m_scan = new b3PrefixScanCL(ctx, device, queue);
+	m_fill = new b3FillCL(ctx, device, queue);
+
 	const char* additionalMacros = "";
 
 	cl_int pErrNum;
 	const char* kernelSource = radixSort32KernelsCL;
-	
-	cl_program sortProg = b3OpenCLUtils::compileCLProgramFromString( ctx, device, kernelSource, &pErrNum,additionalMacros, RADIXSORT32_PATH);
+
+	cl_program sortProg = b3OpenCLUtils::compileCLProgramFromString(ctx, device, kernelSource, &pErrNum, additionalMacros, RADIXSORT32_PATH);
 	b3Assert(sortProg);
 
-	m_streamCountSortDataKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, kernelSource, "StreamCountSortDataKernel", &pErrNum, sortProg,additionalMacros );
-	b3Assert(m_streamCountSortDataKernel );
+	m_streamCountSortDataKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, kernelSource, "StreamCountSortDataKernel", &pErrNum, sortProg, additionalMacros);
+	b3Assert(m_streamCountSortDataKernel);
 
-
-	
-	m_streamCountKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, kernelSource, "StreamCountKernel", &pErrNum, sortProg,additionalMacros );
+	m_streamCountKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, kernelSource, "StreamCountKernel", &pErrNum, sortProg, additionalMacros);
 	b3Assert(m_streamCountKernel);
 
-
-	
 	if (m_deviceCPU)
 	{
-		
-		m_sortAndScatterSortDataKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, kernelSource, "SortAndScatterSortDataKernelSerial", &pErrNum, sortProg,additionalMacros );
+		m_sortAndScatterSortDataKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, kernelSource, "SortAndScatterSortDataKernelSerial", &pErrNum, sortProg, additionalMacros);
 		b3Assert(m_sortAndScatterSortDataKernel);
-		m_sortAndScatterKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, kernelSource, "SortAndScatterKernelSerial", &pErrNum, sortProg,additionalMacros );
-		b3Assert(m_sortAndScatterKernel);
-	} else
-	{
-		m_sortAndScatterSortDataKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, kernelSource, "SortAndScatterSortDataKernel", &pErrNum, sortProg,additionalMacros );
-		b3Assert(m_sortAndScatterSortDataKernel);
-		m_sortAndScatterKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, kernelSource, "SortAndScatterKernel", &pErrNum, sortProg,additionalMacros );
+		m_sortAndScatterKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, kernelSource, "SortAndScatterKernelSerial", &pErrNum, sortProg, additionalMacros);
 		b3Assert(m_sortAndScatterKernel);
 	}
-		
-	m_prefixScanKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, kernelSource, "PrefixScanKernel", &pErrNum, sortProg,additionalMacros );
+	else
+	{
+		m_sortAndScatterSortDataKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, kernelSource, "SortAndScatterSortDataKernel", &pErrNum, sortProg, additionalMacros);
+		b3Assert(m_sortAndScatterSortDataKernel);
+		m_sortAndScatterKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, kernelSource, "SortAndScatterKernel", &pErrNum, sortProg, additionalMacros);
+		b3Assert(m_sortAndScatterKernel);
+	}
+
+	m_prefixScanKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, kernelSource, "PrefixScanKernel", &pErrNum, sortProg, additionalMacros);
 	b3Assert(m_prefixScanKernel);
-		
 }
 
 b3RadixSort32CL::~b3RadixSort32CL()
@@ -96,8 +90,7 @@ void b3RadixSort32CL::executeHost(b3AlignedObjectArray<b3SortData>& inout, int s
 {
 	int n = inout.size();
 	const int BITS_PER_PASS = 8;
-	const int NUM_TABLES = (1<<BITS_PER_PASS);
-
+	const int NUM_TABLES = (1 << BITS_PER_PASS);
 
 	int tables[NUM_TABLES];
 	int counter[NUM_TABLES];
@@ -107,34 +100,33 @@ void b3RadixSort32CL::executeHost(b3AlignedObjectArray<b3SortData>& inout, int s
 	workbuffer.resize(inout.size());
 	b3SortData* dst = &workbuffer[0];
 
-	int count=0;
-	for(int startBit=0; startBit<sortBits; startBit+=BITS_PER_PASS)
+	int count = 0;
+	for (int startBit = 0; startBit < sortBits; startBit += BITS_PER_PASS)
 	{
-		for(int i=0; i<NUM_TABLES; i++)
+		for (int i = 0; i < NUM_TABLES; i++)
 		{
 			tables[i] = 0;
 		}
 
-		for(int i=0; i<n; i++)
+		for (int i = 0; i < n; i++)
 		{
-			int tableIdx = (src[i].m_key >> startBit) & (NUM_TABLES-1);
+			int tableIdx = (src[i].m_key >> startBit) & (NUM_TABLES - 1);
 			tables[tableIdx]++;
 		}
 //#define TEST
 #ifdef TEST
-		printf("histogram size=%d\n",NUM_TABLES);
-		for (int i=0;i<NUM_TABLES;i++)
+		printf("histogram size=%d\n", NUM_TABLES);
+		for (int i = 0; i < NUM_TABLES; i++)
 		{
-			if (tables[i]!=0)
+			if (tables[i] != 0)
 			{
-				printf("tables[%d]=%d]\n",i,tables[i]);
+				printf("tables[%d]=%d]\n", i, tables[i]);
 			}
-
 		}
-#endif //TEST
-		//	prefix scan
+#endif  //TEST \
+	//	prefix scan
 		int sum = 0;
-		for(int i=0; i<NUM_TABLES; i++)
+		for (int i = 0; i < NUM_TABLES; i++)
 		{
 			int iData = tables[i];
 			tables[i] = sum;
@@ -143,71 +135,65 @@ void b3RadixSort32CL::executeHost(b3AlignedObjectArray<b3SortData>& inout, int s
 		}
 
 		//	distribute
-		for(int i=0; i<n; i++)
+		for (int i = 0; i < n; i++)
 		{
-			int tableIdx = (src[i].m_key >> startBit) & (NUM_TABLES-1);
-			
+			int tableIdx = (src[i].m_key >> startBit) & (NUM_TABLES - 1);
+
 			dst[tables[tableIdx] + counter[tableIdx]] = src[i];
-			counter[tableIdx] ++;
+			counter[tableIdx]++;
 		}
 
-		b3Swap( src, dst );
+		b3Swap(src, dst);
 		count++;
 	}
 
-	if (count&1)
+	if (count & 1)
 	{
-		b3Assert(0);//need to copy 
-
+		b3Assert(0);  //need to copy
 	}
 }
 
 void b3RadixSort32CL::executeHost(b3OpenCLArray<b3SortData>& keyValuesInOut, int sortBits /* = 32 */)
 {
-
 	b3AlignedObjectArray<b3SortData> inout;
 	keyValuesInOut.copyToHost(inout);
 
-	executeHost(inout,sortBits);
+	executeHost(inout, sortBits);
 
 	keyValuesInOut.copyFromHost(inout);
 }
 
-void b3RadixSort32CL::execute(b3OpenCLArray<unsigned int>& keysIn, b3OpenCLArray<unsigned int>& keysOut, b3OpenCLArray<unsigned int>& valuesIn, 
-								b3OpenCLArray<unsigned int>& valuesOut, int n, int sortBits)
+void b3RadixSort32CL::execute(b3OpenCLArray<unsigned int>& keysIn, b3OpenCLArray<unsigned int>& keysOut, b3OpenCLArray<unsigned int>& valuesIn,
+							  b3OpenCLArray<unsigned int>& valuesOut, int n, int sortBits)
 {
-
 }
 
 //#define DEBUG_RADIXSORT
 //#define DEBUG_RADIXSORT2
 
-
 void b3RadixSort32CL::execute(b3OpenCLArray<b3SortData>& keyValuesInOut, int sortBits /* = 32 */)
 {
-	
 	int originalSize = keyValuesInOut.size();
 	int workingSize = originalSize;
-	
-			
+
 	int dataAlignment = DATA_ALIGNMENT;
 
 #ifdef DEBUG_RADIXSORT2
-    b3AlignedObjectArray<b3SortData>   test2;
-    keyValuesInOut.copyToHost(test2);
-    printf("numElem = %d\n",test2.size());
-    for (int i=0;i<test2.size();i++)
-    {
-        printf("test2[%d].m_key=%d\n",i,test2[i].m_key);
-        printf("test2[%d].m_value=%d\n",i,test2[i].m_value);
-    }
-#endif //DEBUG_RADIXSORT2
-    
+	b3AlignedObjectArray<b3SortData> test2;
+	keyValuesInOut.copyToHost(test2);
+	printf("numElem = %d\n", test2.size());
+	for (int i = 0; i < test2.size(); i++)
+	{
+		printf("test2[%d].m_key=%d\n", i, test2[i].m_key);
+		printf("test2[%d].m_value=%d\n", i, test2[i].m_value);
+	}
+#endif  //DEBUG_RADIXSORT2
+
 	b3OpenCLArray<b3SortData>* src = 0;
 
-	if (workingSize%dataAlignment)
+	if (workingSize % dataAlignment)
 	{
-		workingSize += dataAlignment-(workingSize%dataAlignment);
+		workingSize += dataAlignment - (workingSize % dataAlignment);
 		m_workBuffer4->copyFromOpenCLArray(keyValuesInOut);
 		m_workBuffer4->resize(workingSize);
 		b3SortData fillValue;
@@ -216,327 +202,301 @@ void b3RadixSort32CL::execute(b3OpenCLArray<b3SortData>& keyValuesInOut, int sor
 
 #define USE_BTFILL
 #ifdef USE_BTFILL
-		m_fill->execute((b3OpenCLArray<b3Int2>&)*m_workBuffer4,(b3Int2&)fillValue,workingSize-originalSize,originalSize);
+		m_fill->execute((b3OpenCLArray<b3Int2>&)*m_workBuffer4, (b3Int2&)fillValue, workingSize - originalSize, originalSize);
 #else
 		//fill the remaining bits (very slow way, todo: fill on GPU/OpenCL side)
-		
-		for (int i=originalSize; i<workingSize;i++)
+
+		for (int i = originalSize; i < workingSize; i++)
 		{
-			m_workBuffer4->copyFromHostPointer(&fillValue,1,i);
+			m_workBuffer4->copyFromHostPointer(&fillValue, 1, i);
 		}
-#endif//USE_BTFILL
+#endif  //USE_BTFILL
 
 		src = m_workBuffer4;
-	} else
+	}
+	else
 	{
 		src = &keyValuesInOut;
 		m_workBuffer4->resize(0);
 	}
-		
-	b3Assert( workingSize%DATA_ALIGNMENT == 0 );
-	int minCap = NUM_BUCKET*NUM_WGS;
 
+	b3Assert(workingSize % DATA_ALIGNMENT == 0);
+	int minCap = NUM_BUCKET * NUM_WGS;
 
 	int n = workingSize;
 
 	m_workBuffer1->resize(minCap);
 	m_workBuffer3->resize(workingSize);
-	
 
-//	ADLASSERT( ELEMENTS_PER_WORK_ITEM == 4 );
-	b3Assert( BITS_PER_PASS == 4 );
-	b3Assert( WG_SIZE == 64 );
-	b3Assert( (sortBits&0x3) == 0 );
+	//	ADLASSERT( ELEMENTS_PER_WORK_ITEM == 4 );
+	b3Assert(BITS_PER_PASS == 4);
+	b3Assert(WG_SIZE == 64);
+	b3Assert((sortBits & 0x3) == 0);
 
-	
-	
 	b3OpenCLArray<b3SortData>* dst = m_workBuffer3;
 
 	b3OpenCLArray<unsigned int>* srcHisto = m_workBuffer1;
 	b3OpenCLArray<unsigned int>* destHisto = m_workBuffer2;
 
-
 	int nWGs = NUM_WGS;
 	b3ConstData cdata;
 
 	{
-        int blockSize = ELEMENTS_PER_WORK_ITEM*WG_SIZE;//set at 256
-     	int nBlocks = (n+blockSize-1)/(blockSize);
+		int blockSize = ELEMENTS_PER_WORK_ITEM * WG_SIZE;  //set at 256
+		int nBlocks = (n + blockSize - 1) / (blockSize);
 		cdata.m_n = n;
 		cdata.m_nWGs = NUM_WGS;
 		cdata.m_startBit = 0;
-		cdata.m_nBlocksPerWG = (nBlocks + cdata.m_nWGs - 1)/cdata.m_nWGs;
-		if( nBlocks < NUM_WGS )
+		cdata.m_nBlocksPerWG = (nBlocks + cdata.m_nWGs - 1) / cdata.m_nWGs;
+		if (nBlocks < NUM_WGS)
 		{
 			cdata.m_nBlocksPerWG = 1;
 			nWGs = nBlocks;
 		}
 	}
 
-	int count=0;
-	for(int ib=0; ib<sortBits; ib+=4)
+	int count = 0;
+	for (int ib = 0; ib < sortBits; ib += 4)
 	{
 #ifdef DEBUG_RADIXSORT2
-        keyValuesInOut.copyToHost(test2);
-        printf("numElem = %d\n",test2.size());
-        for (int i=0;i<test2.size();i++)
-        {
-            if (test2[i].m_key != test2[i].m_value)
-            {
-                printf("test2[%d].m_key=%d\n",i,test2[i].m_key);
-                printf("test2[%d].m_value=%d\n",i,test2[i].m_value);
-            }
-        }
-#endif //DEBUG_RADIXSORT2
-        
+		keyValuesInOut.copyToHost(test2);
+		printf("numElem = %d\n", test2.size());
+		for (int i = 0; i < test2.size(); i++)
+		{
+			if (test2[i].m_key != test2[i].m_value)
+			{
+				printf("test2[%d].m_key=%d\n", i, test2[i].m_key);
+				printf("test2[%d].m_value=%d\n", i, test2[i].m_value);
+			}
+		}
+#endif  //DEBUG_RADIXSORT2
+
 		cdata.m_startBit = ib;
-		
+
 		if (src->size())
 		{
-			b3BufferInfoCL bInfo[] = { b3BufferInfoCL( src->getBufferCL(), true ), b3BufferInfoCL( srcHisto->getBufferCL() ) };
-			b3LauncherCL launcher(m_commandQueue, m_streamCountSortDataKernel,"m_streamCountSortDataKernel");
+			b3BufferInfoCL bInfo[] = {b3BufferInfoCL(src->getBufferCL(), true), b3BufferInfoCL(srcHisto->getBufferCL())};
+			b3LauncherCL launcher(m_commandQueue, m_streamCountSortDataKernel, "m_streamCountSortDataKernel");
 
-			launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-			launcher.setConst(  cdata );
-			
-			int num = NUM_WGS*WG_SIZE;
-			launcher.launch1D( num, WG_SIZE );
+			launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+			launcher.setConst(cdata);
+
+			int num = NUM_WGS * WG_SIZE;
+			launcher.launch1D(num, WG_SIZE);
 		}
 
-        
-        
 #ifdef DEBUG_RADIXSORT
 		b3AlignedObjectArray<unsigned int> testHist;
 		srcHisto->copyToHost(testHist);
-		printf("ib = %d, testHist size = %d, non zero elements:\n",ib, testHist.size());
-		for (int i=0;i<testHist.size();i++)
+		printf("ib = %d, testHist size = %d, non zero elements:\n", ib, testHist.size());
+		for (int i = 0; i < testHist.size(); i++)
 		{
-			if (testHist[i]!=0)
-				printf("testHist[%d]=%d\n",i,testHist[i]);
+			if (testHist[i] != 0)
+				printf("testHist[%d]=%d\n", i, testHist[i]);
 		}
-#endif //DEBUG_RADIXSORT
-	
-	
+#endif  //DEBUG_RADIXSORT
 
 //fast prefix scan is not working properly on Mac OSX yet
 #ifdef __APPLE__
-	bool fastScan=false;
+		bool fastScan = false;
 #else
-	bool fastScan=!m_deviceCPU;//only use fast scan on GPU
+		bool fastScan = !m_deviceCPU;  //only use fast scan on GPU
 #endif
 
 		if (fastScan)
-		{//	prefix scan group histogram
-			b3BufferInfoCL bInfo[] = { b3BufferInfoCL( srcHisto->getBufferCL() ) };
-			b3LauncherCL launcher( m_commandQueue, m_prefixScanKernel,"m_prefixScanKernel" );
-			launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-			launcher.setConst(  cdata );
-			launcher.launch1D( 128, 128 );
+		{  //	prefix scan group histogram
+			b3BufferInfoCL bInfo[] = {b3BufferInfoCL(srcHisto->getBufferCL())};
+			b3LauncherCL launcher(m_commandQueue, m_prefixScanKernel, "m_prefixScanKernel");
+			launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+			launcher.setConst(cdata);
+			launcher.launch1D(128, 128);
 			destHisto = srcHisto;
-		}else
+		}
+		else
 		{
 			//unsigned int sum; //for debugging
-            m_scan->execute(*srcHisto,*destHisto,1920,0);//,&sum);
+			m_scan->execute(*srcHisto, *destHisto, 1920, 0);  //,&sum);
 		}
 
-
 #ifdef DEBUG_RADIXSORT
 		destHisto->copyToHost(testHist);
-		printf("ib = %d, testHist size = %d, non zero elements:\n",ib, testHist.size());
-		for (int i=0;i<testHist.size();i++)
+		printf("ib = %d, testHist size = %d, non zero elements:\n", ib, testHist.size());
+		for (int i = 0; i < testHist.size(); i++)
 		{
-			if (testHist[i]!=0)
-				printf("testHist[%d]=%d\n",i,testHist[i]);
-		}
-        
-        for (int i=0;i<testHist.size();i+=NUM_WGS)
-		{
-				printf("testHist[%d]=%d\n",i/NUM_WGS,testHist[i]);
+			if (testHist[i] != 0)
+				printf("testHist[%d]=%d\n", i, testHist[i]);
 		}
 
-#endif //DEBUG_RADIXSORT
+		for (int i = 0; i < testHist.size(); i += NUM_WGS)
+		{
+			printf("testHist[%d]=%d\n", i / NUM_WGS, testHist[i]);
+		}
+
+#endif  //DEBUG_RADIXSORT
 
 #define USE_GPU
 #ifdef USE_GPU
-        
+
 		if (src->size())
-		{//	local sort and distribute
-			b3BufferInfoCL bInfo[] = { b3BufferInfoCL( src->getBufferCL(), true ), b3BufferInfoCL( destHisto->getBufferCL(), true ), b3BufferInfoCL( dst->getBufferCL() )};
-			b3LauncherCL launcher( m_commandQueue, m_sortAndScatterSortDataKernel,"m_sortAndScatterSortDataKernel" );
-			launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-			launcher.setConst(  cdata );
-			launcher.launch1D( nWGs*WG_SIZE, WG_SIZE );
-            
+		{  //	local sort and distribute
+			b3BufferInfoCL bInfo[] = {b3BufferInfoCL(src->getBufferCL(), true), b3BufferInfoCL(destHisto->getBufferCL(), true), b3BufferInfoCL(dst->getBufferCL())};
+			b3LauncherCL launcher(m_commandQueue, m_sortAndScatterSortDataKernel, "m_sortAndScatterSortDataKernel");
+			launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+			launcher.setConst(cdata);
+			launcher.launch1D(nWGs * WG_SIZE, WG_SIZE);
 		}
 #else
-        {
+		{
 #define NUM_TABLES 16
 //#define SEQUENTIAL
 #ifdef SEQUENTIAL
-            int counter2[NUM_TABLES]={0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0};
-            int tables[NUM_TABLES];
-            int startBit = ib;
-            
-            destHisto->copyToHost(testHist);
-            b3AlignedObjectArray<b3SortData> srcHost;
-            b3AlignedObjectArray<b3SortData> dstHost;
-            dstHost.resize(src->size());
-            
-            src->copyToHost(srcHost);
-            
-            for (int i=0;i<NUM_TABLES;i++)
-            {
-                tables[i] = testHist[i*NUM_WGS];
-            }
-            
-            //	distribute
-            for(int i=0; i<n; i++)
-            {
-                int tableIdx = (srcHost[i].m_key >> startBit) & (NUM_TABLES-1);
-                
-                dstHost[tables[tableIdx] + counter2[tableIdx]] = srcHost[i];
-                counter2[tableIdx] ++;
-            }
-            
-            
+			int counter2[NUM_TABLES] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
+			int tables[NUM_TABLES];
+			int startBit = ib;
+
+			destHisto->copyToHost(testHist);
+			b3AlignedObjectArray<b3SortData> srcHost;
+			b3AlignedObjectArray<b3SortData> dstHost;
+			dstHost.resize(src->size());
+
+			src->copyToHost(srcHost);
+
+			for (int i = 0; i < NUM_TABLES; i++)
+			{
+				tables[i] = testHist[i * NUM_WGS];
+			}
+
+			//	distribute
+			for (int i = 0; i < n; i++)
+			{
+				int tableIdx = (srcHost[i].m_key >> startBit) & (NUM_TABLES - 1);
+
+				dstHost[tables[tableIdx] + counter2[tableIdx]] = srcHost[i];
+				counter2[tableIdx]++;
+			}
+
 #else
-          
-            int counter2[NUM_TABLES]={0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0};
-            
-            int tables[NUM_TABLES];
-             b3AlignedObjectArray<b3SortData> dstHostOK;
-            dstHostOK.resize(src->size());
 
-            destHisto->copyToHost(testHist);
-            b3AlignedObjectArray<b3SortData> srcHost;
-            src->copyToHost(srcHost);
-        
-            int blockSize = 256;
-            int nBlocksPerWG = cdata.m_nBlocksPerWG;
-            int startBit = ib;
+			int counter2[NUM_TABLES] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
 
-            {
-                for (int i=0;i<NUM_TABLES;i++)
-                {
-                    tables[i] = testHist[i*NUM_WGS];
-                }
-                
-                //	distribute
-                for(int i=0; i<n; i++)
-                {
-                    int tableIdx = (srcHost[i].m_key >> startBit) & (NUM_TABLES-1);
-                    
-                    dstHostOK[tables[tableIdx] + counter2[tableIdx]] = srcHost[i];
-                    counter2[tableIdx] ++;
-                }
+			int tables[NUM_TABLES];
+			b3AlignedObjectArray<b3SortData> dstHostOK;
+			dstHostOK.resize(src->size());
 
-            
-            }
-            
-            
-            b3AlignedObjectArray<b3SortData> dstHost;
-            dstHost.resize(src->size());
-            
-            
-            int counter[NUM_TABLES]={0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0};
-            
-            
-            
-            for (int wgIdx=0;wgIdx<NUM_WGS;wgIdx++)
-            {
-              int counter[NUM_TABLES]={0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0};
+			destHisto->copyToHost(testHist);
+			b3AlignedObjectArray<b3SortData> srcHost;
+			src->copyToHost(srcHost);
 
-              int nBlocks = (n)/blockSize - nBlocksPerWG*wgIdx;
-                
-              for(int iblock=0; iblock<b3Min(cdata.m_nBlocksPerWG, nBlocks); iblock++)
-              {
-                for (int lIdx = 0;lIdx < 64;lIdx++)
-                {
-                    int addr = iblock*blockSize + blockSize*cdata.m_nBlocksPerWG*wgIdx + ELEMENTS_PER_WORK_ITEM*lIdx;
-                    
-                    //	MY_HISTOGRAM( localKeys.x ) ++ is much expensive than atomic add as it requires read and write while atomics can just add on AMD
-                    //	Using registers didn't perform well. It seems like use localKeys to address requires a lot of alu ops
-                    //	AMD: AtomInc performs better while NV prefers ++
-                    for(int j=0; j<ELEMENTS_PER_WORK_ITEM; j++)
-                    {
-                        if( addr+j < n )
-                        {
-                          //  printf ("addr+j=%d\n", addr+j);
-                            
-                            int i = addr+j;
-                            
-                            int tableIdx = (srcHost[i].m_key >> startBit) & (NUM_TABLES-1);
-                            
-                            int destIndex = testHist[tableIdx*NUM_WGS+wgIdx] + counter[tableIdx];
-                            
-                            b3SortData ok = dstHostOK[destIndex];
-                                                    
-                            if (ok.m_key != srcHost[i].m_key)
-                            {
-                                printf("ok.m_key = %d, srcHost[i].m_key = %d\n", ok.m_key,srcHost[i].m_key );
-                                printf("(ok.m_value = %d, srcHost[i].m_value = %d)\n", ok.m_value,srcHost[i].m_value );
-                            }
-                            if (ok.m_value != srcHost[i].m_value)
-                            {
-                                
-                               printf("ok.m_value = %d, srcHost[i].m_value = %d\n", ok.m_value,srcHost[i].m_value );
-                                printf("(ok.m_key = %d, srcHost[i].m_key = %d)\n", ok.m_key,srcHost[i].m_key );
+			int blockSize = 256;
+			int nBlocksPerWG = cdata.m_nBlocksPerWG;
+			int startBit = ib;
+
+			{
+				for (int i = 0; i < NUM_TABLES; i++)
+				{
+					tables[i] = testHist[i * NUM_WGS];
+				}
+
+				//	distribute
+				for (int i = 0; i < n; i++)
+				{
+					int tableIdx = (srcHost[i].m_key >> startBit) & (NUM_TABLES - 1);
+
+					dstHostOK[tables[tableIdx] + counter2[tableIdx]] = srcHost[i];
+					counter2[tableIdx]++;
+				}
+			}
+
+			b3AlignedObjectArray<b3SortData> dstHost;
+			dstHost.resize(src->size());
+
+			int counter[NUM_TABLES] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
+
+			for (int wgIdx = 0; wgIdx < NUM_WGS; wgIdx++)
+			{
+				int counter[NUM_TABLES] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
+
+				int nBlocks = (n) / blockSize - nBlocksPerWG * wgIdx;
+
+				for (int iblock = 0; iblock < b3Min(cdata.m_nBlocksPerWG, nBlocks); iblock++)
+				{
+					for (int lIdx = 0; lIdx < 64; lIdx++)
+					{
+						int addr = iblock * blockSize + blockSize * cdata.m_nBlocksPerWG * wgIdx + ELEMENTS_PER_WORK_ITEM * lIdx;
+
+						//	MY_HISTOGRAM( localKeys.x ) ++ is much expensive than atomic add as it requires read and write while atomics can just add on AMD
+						//	Using registers didn't perform well. It seems like use localKeys to address requires a lot of alu ops
+						//	AMD: AtomInc performs better while NV prefers ++
+						for (int j = 0; j < ELEMENTS_PER_WORK_ITEM; j++)
+						{
+							if (addr + j < n)
+							{
+								//  printf ("addr+j=%d\n", addr+j);
+
+								int i = addr + j;
+
+								int tableIdx = (srcHost[i].m_key >> startBit) & (NUM_TABLES - 1);
+
+								int destIndex = testHist[tableIdx * NUM_WGS + wgIdx] + counter[tableIdx];
+
+								b3SortData ok = dstHostOK[destIndex];
+
+								if (ok.m_key != srcHost[i].m_key)
+								{
+									printf("ok.m_key = %d, srcHost[i].m_key = %d\n", ok.m_key, srcHost[i].m_key);
+									printf("(ok.m_value = %d, srcHost[i].m_value = %d)\n", ok.m_value, srcHost[i].m_value);
+								}
+								if (ok.m_value != srcHost[i].m_value)
+								{
+									printf("ok.m_value = %d, srcHost[i].m_value = %d\n", ok.m_value, srcHost[i].m_value);
+									printf("(ok.m_key = %d, srcHost[i].m_key = %d)\n", ok.m_key, srcHost[i].m_key);
+								}
+
+								dstHost[destIndex] = srcHost[i];
+								counter[tableIdx]++;
+							}
+						}
+					}
+				}
+			}
+
+#endif  //SEQUENTIAL
+
+			dst->copyFromHost(dstHost);
+		}
+#endif  //USE_GPU
 
-                            }
-                   
-                            dstHost[destIndex] = srcHost[i];
-                            counter[tableIdx] ++;
-                            
-                        }
-                    }
-                }
-              }
-            }
-            
-         
-#endif //SEQUENTIAL
-            
-            dst->copyFromHost(dstHost);
-        }
-#endif//USE_GPU
-        
-        
-        
 #ifdef DEBUG_RADIXSORT
 		destHisto->copyToHost(testHist);
-		printf("ib = %d, testHist size = %d, non zero elements:\n",ib, testHist.size());
-		for (int i=0;i<testHist.size();i++)
+		printf("ib = %d, testHist size = %d, non zero elements:\n", ib, testHist.size());
+		for (int i = 0; i < testHist.size(); i++)
 		{
-			if (testHist[i]!=0)
-				printf("testHist[%d]=%d\n",i,testHist[i]);
+			if (testHist[i] != 0)
+				printf("testHist[%d]=%d\n", i, testHist[i]);
 		}
-#endif //DEBUG_RADIXSORT
-		b3Swap(src, dst );
-		b3Swap(srcHisto,destHisto);
+#endif  //DEBUG_RADIXSORT
+		b3Swap(src, dst);
+		b3Swap(srcHisto, destHisto);
 
 #ifdef DEBUG_RADIXSORT2
-        keyValuesInOut.copyToHost(test2);
-        printf("numElem = %d\n",test2.size());
-        for (int i=0;i<test2.size();i++)
-        {
-            if (test2[i].m_key != test2[i].m_value)
-            {
-                printf("test2[%d].m_key=%d\n",i,test2[i].m_key);
-                printf("test2[%d].m_value=%d\n",i,test2[i].m_value);
-            }
-        }
-#endif //DEBUG_RADIXSORT2
-        
-        count++;
-                
-        
+		keyValuesInOut.copyToHost(test2);
+		printf("numElem = %d\n", test2.size());
+		for (int i = 0; i < test2.size(); i++)
+		{
+			if (test2[i].m_key != test2[i].m_value)
+			{
+				printf("test2[%d].m_key=%d\n", i, test2[i].m_key);
+				printf("test2[%d].m_value=%d\n", i, test2[i].m_value);
+			}
+		}
+#endif  //DEBUG_RADIXSORT2
+
+		count++;
 	}
-	
-   
-    
-	if (count&1)
+
+	if (count & 1)
 	{
-		b3Assert(0);//need to copy from workbuffer to keyValuesInOut
+		b3Assert(0);  //need to copy from workbuffer to keyValuesInOut
 	}
 
 	if (m_workBuffer4->size())
@@ -545,153 +505,137 @@ void b3RadixSort32CL::execute(b3OpenCLArray<b3SortData>& keyValuesInOut, int sor
 		keyValuesInOut.copyFromOpenCLArray(*m_workBuffer4);
 	}
 
-
 #ifdef DEBUG_RADIXSORT
-    keyValuesInOut.copyToHost(test2);
-   
-    printf("numElem = %d\n",test2.size());
-    for (int i=0;i<test2.size();i++)
-    {
-        printf("test2[%d].m_key=%d\n",i,test2[i].m_key);
-        printf("test2[%d].m_value=%d\n",i,test2[i].m_value);
-    }
-#endif    
-	
+	keyValuesInOut.copyToHost(test2);
+
+	printf("numElem = %d\n", test2.size());
+	for (int i = 0; i < test2.size(); i++)
+	{
+		printf("test2[%d].m_key=%d\n", i, test2[i].m_key);
+		printf("test2[%d].m_value=%d\n", i, test2[i].m_value);
+	}
+#endif
 }
 
-
-
-
-
-
 void b3RadixSort32CL::execute(b3OpenCLArray<unsigned int>& keysInOut, int sortBits /* = 32 */)
 {
 	int originalSize = keysInOut.size();
 	int workingSize = originalSize;
-	
-			
+
 	int dataAlignment = DATA_ALIGNMENT;
 
 	b3OpenCLArray<unsigned int>* src = 0;
 
-	if (workingSize%dataAlignment)
+	if (workingSize % dataAlignment)
 	{
-		workingSize += dataAlignment-(workingSize%dataAlignment);
+		workingSize += dataAlignment - (workingSize % dataAlignment);
 		m_workBuffer4a->copyFromOpenCLArray(keysInOut);
 		m_workBuffer4a->resize(workingSize);
 		unsigned int fillValue = 0xffffffff;
-		
-		m_fill->execute(*m_workBuffer4a,fillValue,workingSize-originalSize,originalSize);
+
+		m_fill->execute(*m_workBuffer4a, fillValue, workingSize - originalSize, originalSize);
 
 		src = m_workBuffer4a;
-	} else
+	}
+	else
 	{
 		src = &keysInOut;
 		m_workBuffer4a->resize(0);
 	}
-	
-	
-
-	b3Assert( workingSize%DATA_ALIGNMENT == 0 );
-	int minCap = NUM_BUCKET*NUM_WGS;
 
+	b3Assert(workingSize % DATA_ALIGNMENT == 0);
+	int minCap = NUM_BUCKET * NUM_WGS;
 
 	int n = workingSize;
 
-	
 	m_workBuffer1->resize(minCap);
 	m_workBuffer3->resize(workingSize);
 	m_workBuffer3a->resize(workingSize);
 
-//	ADLASSERT( ELEMENTS_PER_WORK_ITEM == 4 );
-	b3Assert( BITS_PER_PASS == 4 );
-	b3Assert( WG_SIZE == 64 );
-	b3Assert( (sortBits&0x3) == 0 );
+	//	ADLASSERT( ELEMENTS_PER_WORK_ITEM == 4 );
+	b3Assert(BITS_PER_PASS == 4);
+	b3Assert(WG_SIZE == 64);
+	b3Assert((sortBits & 0x3) == 0);
 
-	
-	
 	b3OpenCLArray<unsigned int>* dst = m_workBuffer3a;
 
 	b3OpenCLArray<unsigned int>* srcHisto = m_workBuffer1;
 	b3OpenCLArray<unsigned int>* destHisto = m_workBuffer2;
 
-
 	int nWGs = NUM_WGS;
 	b3ConstData cdata;
 
 	{
-        int blockSize = ELEMENTS_PER_WORK_ITEM*WG_SIZE;//set at 256
-     	int nBlocks = (n+blockSize-1)/(blockSize);
+		int blockSize = ELEMENTS_PER_WORK_ITEM * WG_SIZE;  //set at 256
+		int nBlocks = (n + blockSize - 1) / (blockSize);
 		cdata.m_n = n;
 		cdata.m_nWGs = NUM_WGS;
 		cdata.m_startBit = 0;
-		cdata.m_nBlocksPerWG = (nBlocks + cdata.m_nWGs - 1)/cdata.m_nWGs;
-		if( nBlocks < NUM_WGS )
+		cdata.m_nBlocksPerWG = (nBlocks + cdata.m_nWGs - 1) / cdata.m_nWGs;
+		if (nBlocks < NUM_WGS)
 		{
 			cdata.m_nBlocksPerWG = 1;
 			nWGs = nBlocks;
 		}
 	}
 
-	int count=0;
-	for(int ib=0; ib<sortBits; ib+=4)
+	int count = 0;
+	for (int ib = 0; ib < sortBits; ib += 4)
 	{
 		cdata.m_startBit = ib;
-		
+
 		if (src->size())
 		{
-			b3BufferInfoCL bInfo[] = { b3BufferInfoCL( src->getBufferCL(), true ), b3BufferInfoCL( srcHisto->getBufferCL() ) };
-			b3LauncherCL launcher(m_commandQueue, m_streamCountKernel,"m_streamCountKernel");
+			b3BufferInfoCL bInfo[] = {b3BufferInfoCL(src->getBufferCL(), true), b3BufferInfoCL(srcHisto->getBufferCL())};
+			b3LauncherCL launcher(m_commandQueue, m_streamCountKernel, "m_streamCountKernel");
 
-			launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-			launcher.setConst(  cdata );
-			
-			int num = NUM_WGS*WG_SIZE;
-			launcher.launch1D( num, WG_SIZE );
+			launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+			launcher.setConst(cdata);
+
+			int num = NUM_WGS * WG_SIZE;
+			launcher.launch1D(num, WG_SIZE);
 		}
 
-        
-
 //fast prefix scan is not working properly on Mac OSX yet
 #ifdef __APPLE__
-	bool fastScan=false;	
+		bool fastScan = false;
 #else
-	bool fastScan=!m_deviceCPU;
+		bool fastScan = !m_deviceCPU;
 #endif
 
 		if (fastScan)
-		{//	prefix scan group histogram
-			b3BufferInfoCL bInfo[] = { b3BufferInfoCL( srcHisto->getBufferCL() ) };
-			b3LauncherCL launcher( m_commandQueue, m_prefixScanKernel,"m_prefixScanKernel" );
-			launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-			launcher.setConst(  cdata );
-			launcher.launch1D( 128, 128 );
+		{  //	prefix scan group histogram
+			b3BufferInfoCL bInfo[] = {b3BufferInfoCL(srcHisto->getBufferCL())};
+			b3LauncherCL launcher(m_commandQueue, m_prefixScanKernel, "m_prefixScanKernel");
+			launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+			launcher.setConst(cdata);
+			launcher.launch1D(128, 128);
 			destHisto = srcHisto;
-		}else
+		}
+		else
 		{
 			//unsigned int sum; //for debugging
-            m_scan->execute(*srcHisto,*destHisto,1920,0);//,&sum);
+			m_scan->execute(*srcHisto, *destHisto, 1920, 0);  //,&sum);
 		}
 
 		if (src->size())
-		{//	local sort and distribute
-			b3BufferInfoCL bInfo[] = { b3BufferInfoCL( src->getBufferCL(), true ), b3BufferInfoCL( destHisto->getBufferCL(), true ), b3BufferInfoCL( dst->getBufferCL() )};
-			b3LauncherCL launcher( m_commandQueue, m_sortAndScatterKernel ,"m_sortAndScatterKernel");
-			launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-			launcher.setConst(  cdata );
-			launcher.launch1D( nWGs*WG_SIZE, WG_SIZE );
-            
+		{  //	local sort and distribute
+			b3BufferInfoCL bInfo[] = {b3BufferInfoCL(src->getBufferCL(), true), b3BufferInfoCL(destHisto->getBufferCL(), true), b3BufferInfoCL(dst->getBufferCL())};
+			b3LauncherCL launcher(m_commandQueue, m_sortAndScatterKernel, "m_sortAndScatterKernel");
+			launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+			launcher.setConst(cdata);
+			launcher.launch1D(nWGs * WG_SIZE, WG_SIZE);
 		}
-        
-		b3Swap(src, dst );
-		b3Swap(srcHisto,destHisto);
 
-        count++;
+		b3Swap(src, dst);
+		b3Swap(srcHisto, destHisto);
+
+		count++;
 	}
-    
-	if (count&1)
+
+	if (count & 1)
 	{
-		b3Assert(0);//need to copy from workbuffer to keyValuesInOut
+		b3Assert(0);  //need to copy from workbuffer to keyValuesInOut
 	}
 
 	if (m_workBuffer4a->size())
@@ -699,12 +643,4 @@ void b3RadixSort32CL::execute(b3OpenCLArray<unsigned int>& keysInOut, int sortBi
 		m_workBuffer4a->resize(originalSize);
 		keysInOut.copyFromOpenCLArray(*m_workBuffer4a);
 	}
-	
 }
-
-
-
-
-
-
-
diff --git a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3RadixSort32CL.h b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3RadixSort32CL.h
index 975bd80e530..69caf182d7f 100644
--- a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3RadixSort32CL.h
+++ b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/b3RadixSort32CL.h
@@ -6,90 +6,79 @@
 
 struct b3SortData
 {
-	union
-	{
+	union {
 		unsigned int m_key;
 		unsigned int x;
 	};
 
-	union
-	{
+	union {
 		unsigned int m_value;
 		unsigned int y;
-		
 	};
 };
 #include "b3BufferInfoCL.h"
 
-class  b3RadixSort32CL
+class b3RadixSort32CL
 {
+	b3OpenCLArray<unsigned int>* m_workBuffer1;
+	b3OpenCLArray<unsigned int>* m_workBuffer2;
 
-		b3OpenCLArray<unsigned int>* m_workBuffer1;
-		b3OpenCLArray<unsigned int>* m_workBuffer2;
-		
-		b3OpenCLArray<b3SortData>*	m_workBuffer3;
-		b3OpenCLArray<b3SortData>*	m_workBuffer4;
+	b3OpenCLArray<b3SortData>* m_workBuffer3;
+	b3OpenCLArray<b3SortData>* m_workBuffer4;
 
-		b3OpenCLArray<unsigned int>* m_workBuffer3a;
-		b3OpenCLArray<unsigned int>* m_workBuffer4a;
+	b3OpenCLArray<unsigned int>* m_workBuffer3a;
+	b3OpenCLArray<unsigned int>* m_workBuffer4a;
 
-		cl_command_queue	m_commandQueue;
+	cl_command_queue m_commandQueue;
 
-		cl_kernel m_streamCountSortDataKernel;
-		cl_kernel m_streamCountKernel;
+	cl_kernel m_streamCountSortDataKernel;
+	cl_kernel m_streamCountKernel;
 
-		cl_kernel m_prefixScanKernel;
-		cl_kernel m_sortAndScatterSortDataKernel;
-		cl_kernel m_sortAndScatterKernel;
+	cl_kernel m_prefixScanKernel;
+	cl_kernel m_sortAndScatterSortDataKernel;
+	cl_kernel m_sortAndScatterKernel;
 
+	bool m_deviceCPU;
 
-		bool	m_deviceCPU;
-
-		class b3PrefixScanCL* m_scan;
-		class b3FillCL*	m_fill;
+	class b3PrefixScanCL* m_scan;
+	class b3FillCL* m_fill;
 
 public:
 	struct b3ConstData
-		{
-			int m_n;
-			int m_nWGs;
-			int m_startBit;
-			int m_nBlocksPerWG;
-		};
+	{
+		int m_n;
+		int m_nWGs;
+		int m_startBit;
+		int m_nBlocksPerWG;
+	};
 	enum
-		{
-			DATA_ALIGNMENT = 256,
-			WG_SIZE = 64,
-            BLOCK_SIZE = 256,
-			ELEMENTS_PER_WORK_ITEM = (BLOCK_SIZE/WG_SIZE),
-			BITS_PER_PASS = 4,
-			NUM_BUCKET=(1<<BITS_PER_PASS),
-			//	if you change this, change nPerWI in kernel as well
-			NUM_WGS = 20*6,	//	cypress
-//			NUM_WGS = 24*6,	//	cayman
-//			NUM_WGS = 32*4,	//	nv
-		};
-
+	{
+		DATA_ALIGNMENT = 256,
+		WG_SIZE = 64,
+		BLOCK_SIZE = 256,
+		ELEMENTS_PER_WORK_ITEM = (BLOCK_SIZE / WG_SIZE),
+		BITS_PER_PASS = 4,
+		NUM_BUCKET = (1 << BITS_PER_PASS),
+		//	if you change this, change nPerWI in kernel as well
+		NUM_WGS = 20 * 6,  //	cypress
+						   //			NUM_WGS = 24*6,	//	cayman
+						   //			NUM_WGS = 32*4,	//	nv
+	};
 
 private:
-		
-
 public:
+	b3RadixSort32CL(cl_context ctx, cl_device_id device, cl_command_queue queue, int initialCapacity = 0);
 
-		b3RadixSort32CL(cl_context ctx, cl_device_id device, cl_command_queue queue, int initialCapacity =0);
+	virtual ~b3RadixSort32CL();
 
-		virtual ~b3RadixSort32CL();
+	void execute(b3OpenCLArray<unsigned int>& keysIn, b3OpenCLArray<unsigned int>& keysOut, b3OpenCLArray<unsigned int>& valuesIn,
+				 b3OpenCLArray<unsigned int>& valuesOut, int n, int sortBits = 32);
 
-		void execute(b3OpenCLArray<unsigned int>& keysIn, b3OpenCLArray<unsigned int>& keysOut, b3OpenCLArray<unsigned int>& valuesIn, 
-								b3OpenCLArray<unsigned int>& valuesOut, int n, int sortBits = 32);
-
-		///keys only
-		void execute(b3OpenCLArray<unsigned int>& keysInOut, int sortBits  = 32 );
-
-		void execute(b3OpenCLArray<b3SortData>& keyValuesInOut, int sortBits  = 32 );
-		void executeHost(b3OpenCLArray<b3SortData>& keyValuesInOut, int sortBits = 32);
-		void executeHost(b3AlignedObjectArray<b3SortData>& keyValuesInOut, int sortBits = 32);
+	///keys only
+	void execute(b3OpenCLArray<unsigned int>& keysInOut, int sortBits = 32);
 
+	void execute(b3OpenCLArray<b3SortData>& keyValuesInOut, int sortBits = 32);
+	void executeHost(b3OpenCLArray<b3SortData>& keyValuesInOut, int sortBits = 32);
+	void executeHost(b3AlignedObjectArray<b3SortData>& keyValuesInOut, int sortBits = 32);
 };
-#endif //B3_RADIXSORT32_H
-
+#endif  //B3_RADIXSORT32_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/BoundSearchKernelsCL.h b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/BoundSearchKernelsCL.h
index 9c9e847138b..1758dd41e3e 100644
--- a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/BoundSearchKernelsCL.h
+++ b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/BoundSearchKernelsCL.h
@@ -1,87 +1,86 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* boundSearchKernelsCL= \
-"/*\n"
-"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
-"This software is provided 'as-is', without any express or implied warranty.\n"
-"In no event will the authors be held liable for any damages arising from the use of this software.\n"
-"Permission is granted to anyone to use this software for any purpose, \n"
-"including commercial applications, and to alter it and redistribute it freely, \n"
-"subject to the following restrictions:\n"
-"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
-"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
-"3. This notice may not be removed or altered from any source distribution.\n"
-"*/\n"
-"//Originally written by Takahiro Harada\n"
-"typedef unsigned int u32;\n"
-"#define GET_GROUP_IDX get_group_id(0)\n"
-"#define GET_LOCAL_IDX get_local_id(0)\n"
-"#define GET_GLOBAL_IDX get_global_id(0)\n"
-"#define GET_GROUP_SIZE get_local_size(0)\n"
-"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
-"typedef struct\n"
-"{\n"
-"	u32 m_key; \n"
-"	u32 m_value;\n"
-"}SortData;\n"
-"typedef struct\n"
-"{\n"
-"	u32 m_nSrc;\n"
-"	u32 m_nDst;\n"
-"	u32 m_padding[2];\n"
-"} ConstBuffer;\n"
-"__attribute__((reqd_work_group_size(64,1,1)))\n"
-"__kernel\n"
-"void SearchSortDataLowerKernel(__global SortData* src, __global u32 *dst, \n"
-"					unsigned int nSrc, unsigned int nDst)\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"	if( gIdx < nSrc )\n"
-"	{\n"
-"		SortData first; first.m_key = (u32)(-1); first.m_value = (u32)(-1);\n"
-"		SortData end; end.m_key = nDst; end.m_value = nDst;\n"
-"		SortData iData = (gIdx==0)? first: src[gIdx-1];\n"
-"		SortData jData = (gIdx==nSrc)? end: src[gIdx];\n"
-"		if( iData.m_key != jData.m_key )\n"
-"		{\n"
-"//			for(u32 k=iData.m_key+1; k<=min(jData.m_key, nDst-1); k++)\n"
-"			u32 k = jData.m_key;\n"
-"			{\n"
-"				dst[k] = gIdx;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"}\n"
-"__attribute__((reqd_work_group_size(64,1,1)))\n"
-"__kernel\n"
-"void SearchSortDataUpperKernel(__global SortData* src, __global u32 *dst, \n"
-"					unsigned int nSrc, unsigned int nDst)\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX+1;\n"
-"	if( gIdx < nSrc+1 )\n"
-"	{\n"
-"		SortData first; first.m_key = 0; first.m_value = 0;\n"
-"		SortData end; end.m_key = nDst; end.m_value = nDst;\n"
-"		SortData iData = src[gIdx-1];\n"
-"		SortData jData = (gIdx==nSrc)? end: src[gIdx];\n"
-"		if( iData.m_key != jData.m_key )\n"
-"		{\n"
-"			u32 k = iData.m_key;\n"
-"			{\n"
-"				dst[k] = gIdx;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"}\n"
-"__attribute__((reqd_work_group_size(64,1,1)))\n"
-"__kernel\n"
-"void SubtractKernel(__global u32* A, __global u32 *B, __global u32 *C, \n"
-"					unsigned int nSrc, unsigned int nDst)\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"	\n"
-"	if( gIdx < nDst )\n"
-"	{\n"
-"		C[gIdx] = A[gIdx] - B[gIdx];\n"
-"	}\n"
-"}\n"
-;
+static const char* boundSearchKernelsCL =
+	"/*\n"
+	"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
+	"This software is provided 'as-is', without any express or implied warranty.\n"
+	"In no event will the authors be held liable for any damages arising from the use of this software.\n"
+	"Permission is granted to anyone to use this software for any purpose, \n"
+	"including commercial applications, and to alter it and redistribute it freely, \n"
+	"subject to the following restrictions:\n"
+	"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
+	"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
+	"3. This notice may not be removed or altered from any source distribution.\n"
+	"*/\n"
+	"//Originally written by Takahiro Harada\n"
+	"typedef unsigned int u32;\n"
+	"#define GET_GROUP_IDX get_group_id(0)\n"
+	"#define GET_LOCAL_IDX get_local_id(0)\n"
+	"#define GET_GLOBAL_IDX get_global_id(0)\n"
+	"#define GET_GROUP_SIZE get_local_size(0)\n"
+	"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
+	"typedef struct\n"
+	"{\n"
+	"	u32 m_key; \n"
+	"	u32 m_value;\n"
+	"}SortData;\n"
+	"typedef struct\n"
+	"{\n"
+	"	u32 m_nSrc;\n"
+	"	u32 m_nDst;\n"
+	"	u32 m_padding[2];\n"
+	"} ConstBuffer;\n"
+	"__attribute__((reqd_work_group_size(64,1,1)))\n"
+	"__kernel\n"
+	"void SearchSortDataLowerKernel(__global SortData* src, __global u32 *dst, \n"
+	"					unsigned int nSrc, unsigned int nDst)\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"	if( gIdx < nSrc )\n"
+	"	{\n"
+	"		SortData first; first.m_key = (u32)(-1); first.m_value = (u32)(-1);\n"
+	"		SortData end; end.m_key = nDst; end.m_value = nDst;\n"
+	"		SortData iData = (gIdx==0)? first: src[gIdx-1];\n"
+	"		SortData jData = (gIdx==nSrc)? end: src[gIdx];\n"
+	"		if( iData.m_key != jData.m_key )\n"
+	"		{\n"
+	"//			for(u32 k=iData.m_key+1; k<=min(jData.m_key, nDst-1); k++)\n"
+	"			u32 k = jData.m_key;\n"
+	"			{\n"
+	"				dst[k] = gIdx;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"__attribute__((reqd_work_group_size(64,1,1)))\n"
+	"__kernel\n"
+	"void SearchSortDataUpperKernel(__global SortData* src, __global u32 *dst, \n"
+	"					unsigned int nSrc, unsigned int nDst)\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX+1;\n"
+	"	if( gIdx < nSrc+1 )\n"
+	"	{\n"
+	"		SortData first; first.m_key = 0; first.m_value = 0;\n"
+	"		SortData end; end.m_key = nDst; end.m_value = nDst;\n"
+	"		SortData iData = src[gIdx-1];\n"
+	"		SortData jData = (gIdx==nSrc)? end: src[gIdx];\n"
+	"		if( iData.m_key != jData.m_key )\n"
+	"		{\n"
+	"			u32 k = iData.m_key;\n"
+	"			{\n"
+	"				dst[k] = gIdx;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"__attribute__((reqd_work_group_size(64,1,1)))\n"
+	"__kernel\n"
+	"void SubtractKernel(__global u32* A, __global u32 *B, __global u32 *C, \n"
+	"					unsigned int nSrc, unsigned int nDst)\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"	\n"
+	"	if( gIdx < nDst )\n"
+	"	{\n"
+	"		C[gIdx] = A[gIdx] - B[gIdx];\n"
+	"	}\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/CopyKernelsCL.h b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/CopyKernelsCL.h
index e5670e3cd3d..33c9279462f 100644
--- a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/CopyKernelsCL.h
+++ b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/CopyKernelsCL.h
@@ -1,132 +1,131 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* copyKernelsCL= \
-"/*\n"
-"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
-"\n"
-"This software is provided 'as-is', without any express or implied warranty.\n"
-"In no event will the authors be held liable for any damages arising from the use of this software.\n"
-"Permission is granted to anyone to use this software for any purpose, \n"
-"including commercial applications, and to alter it and redistribute it freely, \n"
-"subject to the following restrictions:\n"
-"\n"
-"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
-"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
-"3. This notice may not be removed or altered from any source distribution.\n"
-"*/\n"
-"//Originally written by Takahiro Harada\n"
-"\n"
-"#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
-"\n"
-"typedef unsigned int u32;\n"
-"#define GET_GROUP_IDX get_group_id(0)\n"
-"#define GET_LOCAL_IDX get_local_id(0)\n"
-"#define GET_GLOBAL_IDX get_global_id(0)\n"
-"#define GET_GROUP_SIZE get_local_size(0)\n"
-"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
-"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
-"#define AtomInc(x) atom_inc(&(x))\n"
-"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
-"\n"
-"#define make_uint4 (uint4)\n"
-"#define make_uint2 (uint2)\n"
-"#define make_int2 (int2)\n"
-"\n"
-"typedef struct\n"
-"{\n"
-"	int m_n;\n"
-"	int m_padding[3];\n"
-"} ConstBuffer;\n"
-"\n"
-"\n"
-"\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(64,1,1)))\n"
-"void Copy1F4Kernel(__global float4* dst, __global float4* src, \n"
-"					ConstBuffer cb)\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"\n"
-"	if( gIdx < cb.m_n )\n"
-"	{\n"
-"		float4 a0 = src[gIdx];\n"
-"\n"
-"		dst[ gIdx ] = a0;\n"
-"	}\n"
-"}\n"
-"\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(64,1,1)))\n"
-"void Copy2F4Kernel(__global float4* dst, __global float4* src, \n"
-"					ConstBuffer cb)\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"\n"
-"	if( 2*gIdx <= cb.m_n )\n"
-"	{\n"
-"		float4 a0 = src[gIdx*2+0];\n"
-"		float4 a1 = src[gIdx*2+1];\n"
-"\n"
-"		dst[ gIdx*2+0 ] = a0;\n"
-"		dst[ gIdx*2+1 ] = a1;\n"
-"	}\n"
-"}\n"
-"\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(64,1,1)))\n"
-"void Copy4F4Kernel(__global float4* dst, __global float4* src, \n"
-"					ConstBuffer cb)\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"\n"
-"	if( 4*gIdx <= cb.m_n )\n"
-"	{\n"
-"		int idx0 = gIdx*4+0;\n"
-"		int idx1 = gIdx*4+1;\n"
-"		int idx2 = gIdx*4+2;\n"
-"		int idx3 = gIdx*4+3;\n"
-"\n"
-"		float4 a0 = src[idx0];\n"
-"		float4 a1 = src[idx1];\n"
-"		float4 a2 = src[idx2];\n"
-"		float4 a3 = src[idx3];\n"
-"\n"
-"		dst[ idx0 ] = a0;\n"
-"		dst[ idx1 ] = a1;\n"
-"		dst[ idx2 ] = a2;\n"
-"		dst[ idx3 ] = a3;\n"
-"	}\n"
-"}\n"
-"\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(64,1,1)))\n"
-"void CopyF1Kernel(__global float* dstF1, __global float* srcF1, \n"
-"					ConstBuffer cb)\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"\n"
-"	if( gIdx < cb.m_n )\n"
-"	{\n"
-"		float a0 = srcF1[gIdx];\n"
-"\n"
-"		dstF1[ gIdx ] = a0;\n"
-"	}\n"
-"}\n"
-"\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(64,1,1)))\n"
-"void CopyF2Kernel(__global float2* dstF2, __global float2* srcF2, \n"
-"					ConstBuffer cb)\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"\n"
-"	if( gIdx < cb.m_n )\n"
-"	{\n"
-"		float2 a0 = srcF2[gIdx];\n"
-"\n"
-"		dstF2[ gIdx ] = a0;\n"
-"	}\n"
-"}\n"
-"\n"
-"\n"
-;
+static const char* copyKernelsCL =
+	"/*\n"
+	"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
+	"\n"
+	"This software is provided 'as-is', without any express or implied warranty.\n"
+	"In no event will the authors be held liable for any damages arising from the use of this software.\n"
+	"Permission is granted to anyone to use this software for any purpose, \n"
+	"including commercial applications, and to alter it and redistribute it freely, \n"
+	"subject to the following restrictions:\n"
+	"\n"
+	"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
+	"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
+	"3. This notice may not be removed or altered from any source distribution.\n"
+	"*/\n"
+	"//Originally written by Takahiro Harada\n"
+	"\n"
+	"#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
+	"\n"
+	"typedef unsigned int u32;\n"
+	"#define GET_GROUP_IDX get_group_id(0)\n"
+	"#define GET_LOCAL_IDX get_local_id(0)\n"
+	"#define GET_GLOBAL_IDX get_global_id(0)\n"
+	"#define GET_GROUP_SIZE get_local_size(0)\n"
+	"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
+	"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
+	"#define AtomInc(x) atom_inc(&(x))\n"
+	"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
+	"\n"
+	"#define make_uint4 (uint4)\n"
+	"#define make_uint2 (uint2)\n"
+	"#define make_int2 (int2)\n"
+	"\n"
+	"typedef struct\n"
+	"{\n"
+	"	int m_n;\n"
+	"	int m_padding[3];\n"
+	"} ConstBuffer;\n"
+	"\n"
+	"\n"
+	"\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(64,1,1)))\n"
+	"void Copy1F4Kernel(__global float4* dst, __global float4* src, \n"
+	"					ConstBuffer cb)\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"\n"
+	"	if( gIdx < cb.m_n )\n"
+	"	{\n"
+	"		float4 a0 = src[gIdx];\n"
+	"\n"
+	"		dst[ gIdx ] = a0;\n"
+	"	}\n"
+	"}\n"
+	"\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(64,1,1)))\n"
+	"void Copy2F4Kernel(__global float4* dst, __global float4* src, \n"
+	"					ConstBuffer cb)\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"\n"
+	"	if( 2*gIdx <= cb.m_n )\n"
+	"	{\n"
+	"		float4 a0 = src[gIdx*2+0];\n"
+	"		float4 a1 = src[gIdx*2+1];\n"
+	"\n"
+	"		dst[ gIdx*2+0 ] = a0;\n"
+	"		dst[ gIdx*2+1 ] = a1;\n"
+	"	}\n"
+	"}\n"
+	"\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(64,1,1)))\n"
+	"void Copy4F4Kernel(__global float4* dst, __global float4* src, \n"
+	"					ConstBuffer cb)\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"\n"
+	"	if( 4*gIdx <= cb.m_n )\n"
+	"	{\n"
+	"		int idx0 = gIdx*4+0;\n"
+	"		int idx1 = gIdx*4+1;\n"
+	"		int idx2 = gIdx*4+2;\n"
+	"		int idx3 = gIdx*4+3;\n"
+	"\n"
+	"		float4 a0 = src[idx0];\n"
+	"		float4 a1 = src[idx1];\n"
+	"		float4 a2 = src[idx2];\n"
+	"		float4 a3 = src[idx3];\n"
+	"\n"
+	"		dst[ idx0 ] = a0;\n"
+	"		dst[ idx1 ] = a1;\n"
+	"		dst[ idx2 ] = a2;\n"
+	"		dst[ idx3 ] = a3;\n"
+	"	}\n"
+	"}\n"
+	"\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(64,1,1)))\n"
+	"void CopyF1Kernel(__global float* dstF1, __global float* srcF1, \n"
+	"					ConstBuffer cb)\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"\n"
+	"	if( gIdx < cb.m_n )\n"
+	"	{\n"
+	"		float a0 = srcF1[gIdx];\n"
+	"\n"
+	"		dstF1[ gIdx ] = a0;\n"
+	"	}\n"
+	"}\n"
+	"\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(64,1,1)))\n"
+	"void CopyF2Kernel(__global float2* dstF2, __global float2* srcF2, \n"
+	"					ConstBuffer cb)\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"\n"
+	"	if( gIdx < cb.m_n )\n"
+	"	{\n"
+	"		float2 a0 = srcF2[gIdx];\n"
+	"\n"
+	"		dstF2[ gIdx ] = a0;\n"
+	"	}\n"
+	"}\n"
+	"\n"
+	"\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/FillKernelsCL.h b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/FillKernelsCL.h
index 4f8b96e4895..983e6522706 100644
--- a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/FillKernelsCL.h
+++ b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/FillKernelsCL.h
@@ -1,91 +1,90 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* fillKernelsCL= \
-"/*\n"
-"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
-"This software is provided 'as-is', without any express or implied warranty.\n"
-"In no event will the authors be held liable for any damages arising from the use of this software.\n"
-"Permission is granted to anyone to use this software for any purpose, \n"
-"including commercial applications, and to alter it and redistribute it freely, \n"
-"subject to the following restrictions:\n"
-"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
-"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
-"3. This notice may not be removed or altered from any source distribution.\n"
-"*/\n"
-"//Originally written by Takahiro Harada\n"
-"#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
-"typedef unsigned int u32;\n"
-"#define GET_GROUP_IDX get_group_id(0)\n"
-"#define GET_LOCAL_IDX get_local_id(0)\n"
-"#define GET_GLOBAL_IDX get_global_id(0)\n"
-"#define GET_GROUP_SIZE get_local_size(0)\n"
-"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
-"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
-"#define AtomInc(x) atom_inc(&(x))\n"
-"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
-"#define make_uint4 (uint4)\n"
-"#define make_uint2 (uint2)\n"
-"#define make_int2 (int2)\n"
-"typedef struct\n"
-"{\n"
-"	union\n"
-"	{\n"
-"		int4 m_data;\n"
-"		uint4 m_unsignedData;\n"
-"		float	m_floatData;\n"
-"	};\n"
-"	int m_offset;\n"
-"	int m_n;\n"
-"	int m_padding[2];\n"
-"} ConstBuffer;\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(64,1,1)))\n"
-"void FillIntKernel(__global int* dstInt, 			int num_elements, int value, const int offset)\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"	if( gIdx < num_elements )\n"
-"	{\n"
-"		dstInt[ offset+gIdx ] = value;\n"
-"	}\n"
-"}\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(64,1,1)))\n"
-"void FillFloatKernel(__global float* dstFloat, 	int num_elements, float value, const int offset)\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"	if( gIdx < num_elements )\n"
-"	{\n"
-"		dstFloat[ offset+gIdx ] = value;\n"
-"	}\n"
-"}\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(64,1,1)))\n"
-"void FillUnsignedIntKernel(__global unsigned int* dstInt, const int num, const unsigned int value, const int offset)\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"	if( gIdx < num )\n"
-"	{\n"
-"		dstInt[ offset+gIdx ] = value;\n"
-"	}\n"
-"}\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(64,1,1)))\n"
-"void FillInt2Kernel(__global int2* dstInt2, 	const int num, const int2 value, const int offset)\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"	if( gIdx < num )\n"
-"	{\n"
-"		dstInt2[ gIdx + offset] = make_int2( value.x, value.y );\n"
-"	}\n"
-"}\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(64,1,1)))\n"
-"void FillInt4Kernel(__global int4* dstInt4, 		const int num, const int4 value, const int offset)\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"	if( gIdx < num )\n"
-"	{\n"
-"		dstInt4[ offset+gIdx ] = value;\n"
-"	}\n"
-"}\n"
-;
+static const char* fillKernelsCL =
+	"/*\n"
+	"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
+	"This software is provided 'as-is', without any express or implied warranty.\n"
+	"In no event will the authors be held liable for any damages arising from the use of this software.\n"
+	"Permission is granted to anyone to use this software for any purpose, \n"
+	"including commercial applications, and to alter it and redistribute it freely, \n"
+	"subject to the following restrictions:\n"
+	"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
+	"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
+	"3. This notice may not be removed or altered from any source distribution.\n"
+	"*/\n"
+	"//Originally written by Takahiro Harada\n"
+	"#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
+	"typedef unsigned int u32;\n"
+	"#define GET_GROUP_IDX get_group_id(0)\n"
+	"#define GET_LOCAL_IDX get_local_id(0)\n"
+	"#define GET_GLOBAL_IDX get_global_id(0)\n"
+	"#define GET_GROUP_SIZE get_local_size(0)\n"
+	"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
+	"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
+	"#define AtomInc(x) atom_inc(&(x))\n"
+	"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
+	"#define make_uint4 (uint4)\n"
+	"#define make_uint2 (uint2)\n"
+	"#define make_int2 (int2)\n"
+	"typedef struct\n"
+	"{\n"
+	"	union\n"
+	"	{\n"
+	"		int4 m_data;\n"
+	"		uint4 m_unsignedData;\n"
+	"		float	m_floatData;\n"
+	"	};\n"
+	"	int m_offset;\n"
+	"	int m_n;\n"
+	"	int m_padding[2];\n"
+	"} ConstBuffer;\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(64,1,1)))\n"
+	"void FillIntKernel(__global int* dstInt, 			int num_elements, int value, const int offset)\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"	if( gIdx < num_elements )\n"
+	"	{\n"
+	"		dstInt[ offset+gIdx ] = value;\n"
+	"	}\n"
+	"}\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(64,1,1)))\n"
+	"void FillFloatKernel(__global float* dstFloat, 	int num_elements, float value, const int offset)\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"	if( gIdx < num_elements )\n"
+	"	{\n"
+	"		dstFloat[ offset+gIdx ] = value;\n"
+	"	}\n"
+	"}\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(64,1,1)))\n"
+	"void FillUnsignedIntKernel(__global unsigned int* dstInt, const int num, const unsigned int value, const int offset)\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"	if( gIdx < num )\n"
+	"	{\n"
+	"		dstInt[ offset+gIdx ] = value;\n"
+	"	}\n"
+	"}\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(64,1,1)))\n"
+	"void FillInt2Kernel(__global int2* dstInt2, 	const int num, const int2 value, const int offset)\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"	if( gIdx < num )\n"
+	"	{\n"
+	"		dstInt2[ gIdx + offset] = make_int2( value.x, value.y );\n"
+	"	}\n"
+	"}\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(64,1,1)))\n"
+	"void FillInt4Kernel(__global int4* dstInt4, 		const int num, const int4 value, const int offset)\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"	if( gIdx < num )\n"
+	"	{\n"
+	"		dstInt4[ offset+gIdx ] = value;\n"
+	"	}\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/PrefixScanKernelsCL.h b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/PrefixScanKernelsCL.h
index 27baab8331a..fc5e7b865cd 100644
--- a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/PrefixScanKernelsCL.h
+++ b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/PrefixScanKernelsCL.h
@@ -1,129 +1,128 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* prefixScanKernelsCL= \
-"/*\n"
-"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
-"This software is provided 'as-is', without any express or implied warranty.\n"
-"In no event will the authors be held liable for any damages arising from the use of this software.\n"
-"Permission is granted to anyone to use this software for any purpose, \n"
-"including commercial applications, and to alter it and redistribute it freely, \n"
-"subject to the following restrictions:\n"
-"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
-"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
-"3. This notice may not be removed or altered from any source distribution.\n"
-"*/\n"
-"//Originally written by Takahiro Harada\n"
-"typedef unsigned int u32;\n"
-"#define GET_GROUP_IDX get_group_id(0)\n"
-"#define GET_LOCAL_IDX get_local_id(0)\n"
-"#define GET_GLOBAL_IDX get_global_id(0)\n"
-"#define GET_GROUP_SIZE get_local_size(0)\n"
-"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
-"// takahiro end\n"
-"#define WG_SIZE 128 \n"
-"#define m_numElems x\n"
-"#define m_numBlocks y\n"
-"#define m_numScanBlocks z\n"
-"/*typedef struct\n"
-"{\n"
-"	uint m_numElems;\n"
-"	uint m_numBlocks;\n"
-"	uint m_numScanBlocks;\n"
-"	uint m_padding[1];\n"
-"} ConstBuffer;\n"
-"*/\n"
-"u32 ScanExclusive(__local u32* data, u32 n, int lIdx, int lSize)\n"
-"{\n"
-"	u32 blocksum;\n"
-"    int offset = 1;\n"
-"    for(int nActive=n>>1; nActive>0; nActive>>=1, offset<<=1)\n"
-"    {\n"
-"        GROUP_LDS_BARRIER;\n"
-"        for(int iIdx=lIdx; iIdx<nActive; iIdx+=lSize)\n"
-"        {\n"
-"            int ai = offset*(2*iIdx+1)-1;\n"
-"            int bi = offset*(2*iIdx+2)-1;\n"
-"            data[bi] += data[ai];\n"
-"        }\n"
-"	}\n"
-"    GROUP_LDS_BARRIER;\n"
-"    if( lIdx == 0 )\n"
-"	{\n"
-"		blocksum = data[ n-1 ];\n"
-"        data[ n-1 ] = 0;\n"
-"	}\n"
-"	GROUP_LDS_BARRIER;\n"
-"	offset >>= 1;\n"
-"    for(int nActive=1; nActive<n; nActive<<=1, offset>>=1 )\n"
-"    {\n"
-"        GROUP_LDS_BARRIER;\n"
-"        for( int iIdx = lIdx; iIdx<nActive; iIdx += lSize )\n"
-"        {\n"
-"            int ai = offset*(2*iIdx+1)-1;\n"
-"            int bi = offset*(2*iIdx+2)-1;\n"
-"            u32 temp = data[ai];\n"
-"            data[ai] = data[bi];\n"
-"            data[bi] += temp;\n"
-"        }\n"
-"	}\n"
-"	GROUP_LDS_BARRIER;\n"
-"	return blocksum;\n"
-"}\n"
-"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"__kernel\n"
-"void LocalScanKernel(__global u32* dst, __global u32 *src, __global u32 *sumBuffer,\n"
-"		uint4 cb)\n"
-"{\n"
-"	__local u32 ldsData[WG_SIZE*2];\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"	int lIdx = GET_LOCAL_IDX;\n"
-"	ldsData[2*lIdx]     = ( 2*gIdx < cb.m_numElems )? src[2*gIdx]: 0;\n"
-"	ldsData[2*lIdx + 1] = ( 2*gIdx+1 < cb.m_numElems )? src[2*gIdx + 1]: 0;\n"
-"	u32 sum = ScanExclusive(ldsData, WG_SIZE*2, GET_LOCAL_IDX, GET_GROUP_SIZE);\n"
-"	if( lIdx == 0 ) sumBuffer[GET_GROUP_IDX] = sum;\n"
-"	if( (2*gIdx) < cb.m_numElems )\n"
-"    {\n"
-"        dst[2*gIdx]     = ldsData[2*lIdx];\n"
-"	}\n"
-"	if( (2*gIdx + 1) < cb.m_numElems )\n"
-"	{\n"
-"        dst[2*gIdx + 1] = ldsData[2*lIdx + 1];\n"
-"    }\n"
-"}\n"
-"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"__kernel\n"
-"void AddOffsetKernel(__global u32 *dst, __global u32 *blockSum, uint4 cb)\n"
-"{\n"
-"	const u32 blockSize = WG_SIZE*2;\n"
-"	int myIdx = GET_GROUP_IDX+1;\n"
-"	int lIdx = GET_LOCAL_IDX;\n"
-"	u32 iBlockSum = blockSum[myIdx];\n"
-"	int endValue = min((myIdx+1)*(blockSize), cb.m_numElems);\n"
-"	for(int i=myIdx*blockSize+lIdx; i<endValue; i+=GET_GROUP_SIZE)\n"
-"	{\n"
-"		dst[i] += iBlockSum;\n"
-"	}\n"
-"}\n"
-"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"__kernel\n"
-"void TopLevelScanKernel(__global u32* dst, uint4 cb)\n"
-"{\n"
-"	__local u32 ldsData[2048];\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"	int lIdx = GET_LOCAL_IDX;\n"
-"	int lSize = GET_GROUP_SIZE;\n"
-"	for(int i=lIdx; i<cb.m_numScanBlocks; i+=lSize )\n"
-"	{\n"
-"		ldsData[i] = (i<cb.m_numBlocks)? dst[i]:0;\n"
-"	}\n"
-"	GROUP_LDS_BARRIER;\n"
-"	u32 sum = ScanExclusive(ldsData, cb.m_numScanBlocks, GET_LOCAL_IDX, GET_GROUP_SIZE);\n"
-"	for(int i=lIdx; i<cb.m_numBlocks; i+=lSize )\n"
-"	{\n"
-"		dst[i] = ldsData[i];\n"
-"	}\n"
-"	if( gIdx == 0 )\n"
-"	{\n"
-"		dst[cb.m_numBlocks] = sum;\n"
-"	}\n"
-"}\n"
-;
+static const char* prefixScanKernelsCL =
+	"/*\n"
+	"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
+	"This software is provided 'as-is', without any express or implied warranty.\n"
+	"In no event will the authors be held liable for any damages arising from the use of this software.\n"
+	"Permission is granted to anyone to use this software for any purpose, \n"
+	"including commercial applications, and to alter it and redistribute it freely, \n"
+	"subject to the following restrictions:\n"
+	"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
+	"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
+	"3. This notice may not be removed or altered from any source distribution.\n"
+	"*/\n"
+	"//Originally written by Takahiro Harada\n"
+	"typedef unsigned int u32;\n"
+	"#define GET_GROUP_IDX get_group_id(0)\n"
+	"#define GET_LOCAL_IDX get_local_id(0)\n"
+	"#define GET_GLOBAL_IDX get_global_id(0)\n"
+	"#define GET_GROUP_SIZE get_local_size(0)\n"
+	"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
+	"// takahiro end\n"
+	"#define WG_SIZE 128 \n"
+	"#define m_numElems x\n"
+	"#define m_numBlocks y\n"
+	"#define m_numScanBlocks z\n"
+	"/*typedef struct\n"
+	"{\n"
+	"	uint m_numElems;\n"
+	"	uint m_numBlocks;\n"
+	"	uint m_numScanBlocks;\n"
+	"	uint m_padding[1];\n"
+	"} ConstBuffer;\n"
+	"*/\n"
+	"u32 ScanExclusive(__local u32* data, u32 n, int lIdx, int lSize)\n"
+	"{\n"
+	"	u32 blocksum;\n"
+	"    int offset = 1;\n"
+	"    for(int nActive=n>>1; nActive>0; nActive>>=1, offset<<=1)\n"
+	"    {\n"
+	"        GROUP_LDS_BARRIER;\n"
+	"        for(int iIdx=lIdx; iIdx<nActive; iIdx+=lSize)\n"
+	"        {\n"
+	"            int ai = offset*(2*iIdx+1)-1;\n"
+	"            int bi = offset*(2*iIdx+2)-1;\n"
+	"            data[bi] += data[ai];\n"
+	"        }\n"
+	"	}\n"
+	"    GROUP_LDS_BARRIER;\n"
+	"    if( lIdx == 0 )\n"
+	"	{\n"
+	"		blocksum = data[ n-1 ];\n"
+	"        data[ n-1 ] = 0;\n"
+	"	}\n"
+	"	GROUP_LDS_BARRIER;\n"
+	"	offset >>= 1;\n"
+	"    for(int nActive=1; nActive<n; nActive<<=1, offset>>=1 )\n"
+	"    {\n"
+	"        GROUP_LDS_BARRIER;\n"
+	"        for( int iIdx = lIdx; iIdx<nActive; iIdx += lSize )\n"
+	"        {\n"
+	"            int ai = offset*(2*iIdx+1)-1;\n"
+	"            int bi = offset*(2*iIdx+2)-1;\n"
+	"            u32 temp = data[ai];\n"
+	"            data[ai] = data[bi];\n"
+	"            data[bi] += temp;\n"
+	"        }\n"
+	"	}\n"
+	"	GROUP_LDS_BARRIER;\n"
+	"	return blocksum;\n"
+	"}\n"
+	"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"__kernel\n"
+	"void LocalScanKernel(__global u32* dst, __global u32 *src, __global u32 *sumBuffer,\n"
+	"		uint4 cb)\n"
+	"{\n"
+	"	__local u32 ldsData[WG_SIZE*2];\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"	int lIdx = GET_LOCAL_IDX;\n"
+	"	ldsData[2*lIdx]     = ( 2*gIdx < cb.m_numElems )? src[2*gIdx]: 0;\n"
+	"	ldsData[2*lIdx + 1] = ( 2*gIdx+1 < cb.m_numElems )? src[2*gIdx + 1]: 0;\n"
+	"	u32 sum = ScanExclusive(ldsData, WG_SIZE*2, GET_LOCAL_IDX, GET_GROUP_SIZE);\n"
+	"	if( lIdx == 0 ) sumBuffer[GET_GROUP_IDX] = sum;\n"
+	"	if( (2*gIdx) < cb.m_numElems )\n"
+	"    {\n"
+	"        dst[2*gIdx]     = ldsData[2*lIdx];\n"
+	"	}\n"
+	"	if( (2*gIdx + 1) < cb.m_numElems )\n"
+	"	{\n"
+	"        dst[2*gIdx + 1] = ldsData[2*lIdx + 1];\n"
+	"    }\n"
+	"}\n"
+	"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"__kernel\n"
+	"void AddOffsetKernel(__global u32 *dst, __global u32 *blockSum, uint4 cb)\n"
+	"{\n"
+	"	const u32 blockSize = WG_SIZE*2;\n"
+	"	int myIdx = GET_GROUP_IDX+1;\n"
+	"	int lIdx = GET_LOCAL_IDX;\n"
+	"	u32 iBlockSum = blockSum[myIdx];\n"
+	"	int endValue = min((myIdx+1)*(blockSize), cb.m_numElems);\n"
+	"	for(int i=myIdx*blockSize+lIdx; i<endValue; i+=GET_GROUP_SIZE)\n"
+	"	{\n"
+	"		dst[i] += iBlockSum;\n"
+	"	}\n"
+	"}\n"
+	"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"__kernel\n"
+	"void TopLevelScanKernel(__global u32* dst, uint4 cb)\n"
+	"{\n"
+	"	__local u32 ldsData[2048];\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"	int lIdx = GET_LOCAL_IDX;\n"
+	"	int lSize = GET_GROUP_SIZE;\n"
+	"	for(int i=lIdx; i<cb.m_numScanBlocks; i+=lSize )\n"
+	"	{\n"
+	"		ldsData[i] = (i<cb.m_numBlocks)? dst[i]:0;\n"
+	"	}\n"
+	"	GROUP_LDS_BARRIER;\n"
+	"	u32 sum = ScanExclusive(ldsData, cb.m_numScanBlocks, GET_LOCAL_IDX, GET_GROUP_SIZE);\n"
+	"	for(int i=lIdx; i<cb.m_numBlocks; i+=lSize )\n"
+	"	{\n"
+	"		dst[i] = ldsData[i];\n"
+	"	}\n"
+	"	if( gIdx == 0 )\n"
+	"	{\n"
+	"		dst[cb.m_numBlocks] = sum;\n"
+	"	}\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/PrefixScanKernelsFloat4CL.h b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/PrefixScanKernelsFloat4CL.h
index 5b132547966..15d1bc51953 100644
--- a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/PrefixScanKernelsFloat4CL.h
+++ b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/PrefixScanKernelsFloat4CL.h
@@ -1,129 +1,128 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* prefixScanKernelsFloat4CL= \
-"/*\n"
-"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
-"This software is provided 'as-is', without any express or implied warranty.\n"
-"In no event will the authors be held liable for any damages arising from the use of this software.\n"
-"Permission is granted to anyone to use this software for any purpose, \n"
-"including commercial applications, and to alter it and redistribute it freely, \n"
-"subject to the following restrictions:\n"
-"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
-"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
-"3. This notice may not be removed or altered from any source distribution.\n"
-"*/\n"
-"//Originally written by Takahiro Harada\n"
-"typedef unsigned int u32;\n"
-"#define GET_GROUP_IDX get_group_id(0)\n"
-"#define GET_LOCAL_IDX get_local_id(0)\n"
-"#define GET_GLOBAL_IDX get_global_id(0)\n"
-"#define GET_GROUP_SIZE get_local_size(0)\n"
-"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
-"// takahiro end\n"
-"#define WG_SIZE 128 \n"
-"#define m_numElems x\n"
-"#define m_numBlocks y\n"
-"#define m_numScanBlocks z\n"
-"/*typedef struct\n"
-"{\n"
-"	uint m_numElems;\n"
-"	uint m_numBlocks;\n"
-"	uint m_numScanBlocks;\n"
-"	uint m_padding[1];\n"
-"} ConstBuffer;\n"
-"*/\n"
-"float4 ScanExclusiveFloat4(__local float4* data, u32 n, int lIdx, int lSize)\n"
-"{\n"
-"	float4 blocksum;\n"
-"    int offset = 1;\n"
-"    for(int nActive=n>>1; nActive>0; nActive>>=1, offset<<=1)\n"
-"    {\n"
-"        GROUP_LDS_BARRIER;\n"
-"        for(int iIdx=lIdx; iIdx<nActive; iIdx+=lSize)\n"
-"        {\n"
-"            int ai = offset*(2*iIdx+1)-1;\n"
-"            int bi = offset*(2*iIdx+2)-1;\n"
-"            data[bi] += data[ai];\n"
-"        }\n"
-"	}\n"
-"    GROUP_LDS_BARRIER;\n"
-"    if( lIdx == 0 )\n"
-"	{\n"
-"		blocksum = data[ n-1 ];\n"
-"    data[ n-1 ] = 0;\n"
-"	}\n"
-"	GROUP_LDS_BARRIER;\n"
-"	offset >>= 1;\n"
-"    for(int nActive=1; nActive<n; nActive<<=1, offset>>=1 )\n"
-"    {\n"
-"        GROUP_LDS_BARRIER;\n"
-"        for( int iIdx = lIdx; iIdx<nActive; iIdx += lSize )\n"
-"        {\n"
-"            int ai = offset*(2*iIdx+1)-1;\n"
-"            int bi = offset*(2*iIdx+2)-1;\n"
-"            float4 temp = data[ai];\n"
-"            data[ai] = data[bi];\n"
-"            data[bi] += temp;\n"
-"        }\n"
-"	}\n"
-"	GROUP_LDS_BARRIER;\n"
-"	return blocksum;\n"
-"}\n"
-"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"__kernel\n"
-"void LocalScanKernel(__global float4* dst, __global float4* src, __global float4* sumBuffer,	uint4 cb)\n"
-"{\n"
-"	__local float4 ldsData[WG_SIZE*2];\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"	int lIdx = GET_LOCAL_IDX;\n"
-"	ldsData[2*lIdx]     = ( 2*gIdx < cb.m_numElems )? src[2*gIdx]: 0;\n"
-"	ldsData[2*lIdx + 1] = ( 2*gIdx+1 < cb.m_numElems )? src[2*gIdx + 1]: 0;\n"
-"	float4 sum = ScanExclusiveFloat4(ldsData, WG_SIZE*2, GET_LOCAL_IDX, GET_GROUP_SIZE);\n"
-"	if( lIdx == 0 ) \n"
-"		sumBuffer[GET_GROUP_IDX] = sum;\n"
-"	if( (2*gIdx) < cb.m_numElems )\n"
-"    {\n"
-"        dst[2*gIdx]     = ldsData[2*lIdx];\n"
-"	}\n"
-"	if( (2*gIdx + 1) < cb.m_numElems )\n"
-"	{\n"
-"        dst[2*gIdx + 1] = ldsData[2*lIdx + 1];\n"
-"    }\n"
-"}\n"
-"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"__kernel\n"
-"void AddOffsetKernel(__global float4* dst, __global float4* blockSum, uint4 cb)\n"
-"{\n"
-"	const u32 blockSize = WG_SIZE*2;\n"
-"	int myIdx = GET_GROUP_IDX+1;\n"
-"	int lIdx = GET_LOCAL_IDX;\n"
-"	float4 iBlockSum = blockSum[myIdx];\n"
-"	int endValue = min((myIdx+1)*(blockSize), cb.m_numElems);\n"
-"	for(int i=myIdx*blockSize+lIdx; i<endValue; i+=GET_GROUP_SIZE)\n"
-"	{\n"
-"		dst[i] += iBlockSum;\n"
-"	}\n"
-"}\n"
-"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"__kernel\n"
-"void TopLevelScanKernel(__global float4* dst, uint4 cb)\n"
-"{\n"
-"	__local float4 ldsData[2048];\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"	int lIdx = GET_LOCAL_IDX;\n"
-"	int lSize = GET_GROUP_SIZE;\n"
-"	for(int i=lIdx; i<cb.m_numScanBlocks; i+=lSize )\n"
-"	{\n"
-"		ldsData[i] = (i<cb.m_numBlocks)? dst[i]:0;\n"
-"	}\n"
-"	GROUP_LDS_BARRIER;\n"
-"	float4 sum = ScanExclusiveFloat4(ldsData, cb.m_numScanBlocks, GET_LOCAL_IDX, GET_GROUP_SIZE);\n"
-"	for(int i=lIdx; i<cb.m_numBlocks; i+=lSize )\n"
-"	{\n"
-"		dst[i] = ldsData[i];\n"
-"	}\n"
-"	if( gIdx == 0 )\n"
-"	{\n"
-"		dst[cb.m_numBlocks] = sum;\n"
-"	}\n"
-"}\n"
-;
+static const char* prefixScanKernelsFloat4CL =
+	"/*\n"
+	"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
+	"This software is provided 'as-is', without any express or implied warranty.\n"
+	"In no event will the authors be held liable for any damages arising from the use of this software.\n"
+	"Permission is granted to anyone to use this software for any purpose, \n"
+	"including commercial applications, and to alter it and redistribute it freely, \n"
+	"subject to the following restrictions:\n"
+	"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
+	"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
+	"3. This notice may not be removed or altered from any source distribution.\n"
+	"*/\n"
+	"//Originally written by Takahiro Harada\n"
+	"typedef unsigned int u32;\n"
+	"#define GET_GROUP_IDX get_group_id(0)\n"
+	"#define GET_LOCAL_IDX get_local_id(0)\n"
+	"#define GET_GLOBAL_IDX get_global_id(0)\n"
+	"#define GET_GROUP_SIZE get_local_size(0)\n"
+	"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
+	"// takahiro end\n"
+	"#define WG_SIZE 128 \n"
+	"#define m_numElems x\n"
+	"#define m_numBlocks y\n"
+	"#define m_numScanBlocks z\n"
+	"/*typedef struct\n"
+	"{\n"
+	"	uint m_numElems;\n"
+	"	uint m_numBlocks;\n"
+	"	uint m_numScanBlocks;\n"
+	"	uint m_padding[1];\n"
+	"} ConstBuffer;\n"
+	"*/\n"
+	"float4 ScanExclusiveFloat4(__local float4* data, u32 n, int lIdx, int lSize)\n"
+	"{\n"
+	"	float4 blocksum;\n"
+	"    int offset = 1;\n"
+	"    for(int nActive=n>>1; nActive>0; nActive>>=1, offset<<=1)\n"
+	"    {\n"
+	"        GROUP_LDS_BARRIER;\n"
+	"        for(int iIdx=lIdx; iIdx<nActive; iIdx+=lSize)\n"
+	"        {\n"
+	"            int ai = offset*(2*iIdx+1)-1;\n"
+	"            int bi = offset*(2*iIdx+2)-1;\n"
+	"            data[bi] += data[ai];\n"
+	"        }\n"
+	"	}\n"
+	"    GROUP_LDS_BARRIER;\n"
+	"    if( lIdx == 0 )\n"
+	"	{\n"
+	"		blocksum = data[ n-1 ];\n"
+	"    data[ n-1 ] = 0;\n"
+	"	}\n"
+	"	GROUP_LDS_BARRIER;\n"
+	"	offset >>= 1;\n"
+	"    for(int nActive=1; nActive<n; nActive<<=1, offset>>=1 )\n"
+	"    {\n"
+	"        GROUP_LDS_BARRIER;\n"
+	"        for( int iIdx = lIdx; iIdx<nActive; iIdx += lSize )\n"
+	"        {\n"
+	"            int ai = offset*(2*iIdx+1)-1;\n"
+	"            int bi = offset*(2*iIdx+2)-1;\n"
+	"            float4 temp = data[ai];\n"
+	"            data[ai] = data[bi];\n"
+	"            data[bi] += temp;\n"
+	"        }\n"
+	"	}\n"
+	"	GROUP_LDS_BARRIER;\n"
+	"	return blocksum;\n"
+	"}\n"
+	"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"__kernel\n"
+	"void LocalScanKernel(__global float4* dst, __global float4* src, __global float4* sumBuffer,	uint4 cb)\n"
+	"{\n"
+	"	__local float4 ldsData[WG_SIZE*2];\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"	int lIdx = GET_LOCAL_IDX;\n"
+	"	ldsData[2*lIdx]     = ( 2*gIdx < cb.m_numElems )? src[2*gIdx]: 0;\n"
+	"	ldsData[2*lIdx + 1] = ( 2*gIdx+1 < cb.m_numElems )? src[2*gIdx + 1]: 0;\n"
+	"	float4 sum = ScanExclusiveFloat4(ldsData, WG_SIZE*2, GET_LOCAL_IDX, GET_GROUP_SIZE);\n"
+	"	if( lIdx == 0 ) \n"
+	"		sumBuffer[GET_GROUP_IDX] = sum;\n"
+	"	if( (2*gIdx) < cb.m_numElems )\n"
+	"    {\n"
+	"        dst[2*gIdx]     = ldsData[2*lIdx];\n"
+	"	}\n"
+	"	if( (2*gIdx + 1) < cb.m_numElems )\n"
+	"	{\n"
+	"        dst[2*gIdx + 1] = ldsData[2*lIdx + 1];\n"
+	"    }\n"
+	"}\n"
+	"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"__kernel\n"
+	"void AddOffsetKernel(__global float4* dst, __global float4* blockSum, uint4 cb)\n"
+	"{\n"
+	"	const u32 blockSize = WG_SIZE*2;\n"
+	"	int myIdx = GET_GROUP_IDX+1;\n"
+	"	int lIdx = GET_LOCAL_IDX;\n"
+	"	float4 iBlockSum = blockSum[myIdx];\n"
+	"	int endValue = min((myIdx+1)*(blockSize), cb.m_numElems);\n"
+	"	for(int i=myIdx*blockSize+lIdx; i<endValue; i+=GET_GROUP_SIZE)\n"
+	"	{\n"
+	"		dst[i] += iBlockSum;\n"
+	"	}\n"
+	"}\n"
+	"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"__kernel\n"
+	"void TopLevelScanKernel(__global float4* dst, uint4 cb)\n"
+	"{\n"
+	"	__local float4 ldsData[2048];\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"	int lIdx = GET_LOCAL_IDX;\n"
+	"	int lSize = GET_GROUP_SIZE;\n"
+	"	for(int i=lIdx; i<cb.m_numScanBlocks; i+=lSize )\n"
+	"	{\n"
+	"		ldsData[i] = (i<cb.m_numBlocks)? dst[i]:0;\n"
+	"	}\n"
+	"	GROUP_LDS_BARRIER;\n"
+	"	float4 sum = ScanExclusiveFloat4(ldsData, cb.m_numScanBlocks, GET_LOCAL_IDX, GET_GROUP_SIZE);\n"
+	"	for(int i=lIdx; i<cb.m_numBlocks; i+=lSize )\n"
+	"	{\n"
+	"		dst[i] = ldsData[i];\n"
+	"	}\n"
+	"	if( gIdx == 0 )\n"
+	"	{\n"
+	"		dst[cb.m_numBlocks] = sum;\n"
+	"	}\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/RadixSort32KernelsCL.h b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/RadixSort32KernelsCL.h
index 8876c16aa64..fb4bdda3034 100644
--- a/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/RadixSort32KernelsCL.h
+++ b/thirdparty/bullet/Bullet3OpenCL/ParallelPrimitives/kernels/RadixSort32KernelsCL.h
@@ -1,910 +1,909 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* radixSort32KernelsCL= \
-"/*\n"
-"Bullet Continuous Collision Detection and Physics Library\n"
-"Copyright (c) 2011 Advanced Micro Devices, Inc.  http://bulletphysics.org\n"
-"This software is provided 'as-is', without any express or implied warranty.\n"
-"In no event will the authors be held liable for any damages arising from the use of this software.\n"
-"Permission is granted to anyone to use this software for any purpose, \n"
-"including commercial applications, and to alter it and redistribute it freely, \n"
-"subject to the following restrictions:\n"
-"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
-"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
-"3. This notice may not be removed or altered from any source distribution.\n"
-"*/\n"
-"//Author Takahiro Harada\n"
-"//#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
-"typedef unsigned int u32;\n"
-"#define GET_GROUP_IDX get_group_id(0)\n"
-"#define GET_LOCAL_IDX get_local_id(0)\n"
-"#define GET_GLOBAL_IDX get_global_id(0)\n"
-"#define GET_GROUP_SIZE get_local_size(0)\n"
-"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
-"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
-"#define AtomInc(x) atom_inc(&(x))\n"
-"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
-"#define AtomAdd(x, value) atom_add(&(x), value)\n"
-"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
-"#define make_uint4 (uint4)\n"
-"#define make_uint2 (uint2)\n"
-"#define make_int2 (int2)\n"
-"#define WG_SIZE 64\n"
-"#define ELEMENTS_PER_WORK_ITEM (256/WG_SIZE)\n"
-"#define BITS_PER_PASS 4\n"
-"#define NUM_BUCKET (1<<BITS_PER_PASS)\n"
-"typedef uchar u8;\n"
-"//	this isn't optimization for VLIW. But just reducing writes. \n"
-"#define USE_2LEVEL_REDUCE 1\n"
-"//#define CHECK_BOUNDARY 1\n"
-"//#define NV_GPU 1\n"
-"//	Cypress\n"
-"#define nPerWI 16\n"
-"//	Cayman\n"
-"//#define nPerWI 20\n"
-"#define m_n x\n"
-"#define m_nWGs y\n"
-"#define m_startBit z\n"
-"#define m_nBlocksPerWG w\n"
-"/*\n"
-"typedef struct\n"
-"{\n"
-"	int m_n;\n"
-"	int m_nWGs;\n"
-"	int m_startBit;\n"
-"	int m_nBlocksPerWG;\n"
-"} ConstBuffer;\n"
-"*/\n"
-"typedef struct\n"
-"{\n"
-"	unsigned int m_key;\n"
-"	unsigned int m_value;\n"
-"} SortDataCL;\n"
-"uint prefixScanVectorEx( uint4* data )\n"
-"{\n"
-"	u32 sum = 0;\n"
-"	u32 tmp = data[0].x;\n"
-"	data[0].x = sum;\n"
-"	sum += tmp;\n"
-"	tmp = data[0].y;\n"
-"	data[0].y = sum;\n"
-"	sum += tmp;\n"
-"	tmp = data[0].z;\n"
-"	data[0].z = sum;\n"
-"	sum += tmp;\n"
-"	tmp = data[0].w;\n"
-"	data[0].w = sum;\n"
-"	sum += tmp;\n"
-"	return sum;\n"
-"}\n"
-"u32 localPrefixSum( u32 pData, uint lIdx, uint* totalSum, __local u32* sorterSharedMemory, int wgSize /*64 or 128*/ )\n"
-"{\n"
-"	{	//	Set data\n"
-"		sorterSharedMemory[lIdx] = 0;\n"
-"		sorterSharedMemory[lIdx+wgSize] = pData;\n"
-"	}\n"
-"	GROUP_LDS_BARRIER;\n"
-"	{	//	Prefix sum\n"
-"		int idx = 2*lIdx + (wgSize+1);\n"
-"#if defined(USE_2LEVEL_REDUCE)\n"
-"		if( lIdx < 64 )\n"
-"		{\n"
-"			u32 u0, u1, u2;\n"
-"			u0 = sorterSharedMemory[idx-3];\n"
-"			u1 = sorterSharedMemory[idx-2];\n"
-"			u2 = sorterSharedMemory[idx-1];\n"
-"			AtomAdd( sorterSharedMemory[idx], u0+u1+u2 );			\n"
-"			GROUP_MEM_FENCE;\n"
-"			u0 = sorterSharedMemory[idx-12];\n"
-"			u1 = sorterSharedMemory[idx-8];\n"
-"			u2 = sorterSharedMemory[idx-4];\n"
-"			AtomAdd( sorterSharedMemory[idx], u0+u1+u2 );			\n"
-"			GROUP_MEM_FENCE;\n"
-"			u0 = sorterSharedMemory[idx-48];\n"
-"			u1 = sorterSharedMemory[idx-32];\n"
-"			u2 = sorterSharedMemory[idx-16];\n"
-"			AtomAdd( sorterSharedMemory[idx], u0+u1+u2 );			\n"
-"			GROUP_MEM_FENCE;\n"
-"			if( wgSize > 64 )\n"
-"			{\n"
-"				sorterSharedMemory[idx] += sorterSharedMemory[idx-64];\n"
-"				GROUP_MEM_FENCE;\n"
-"			}\n"
-"			sorterSharedMemory[idx-1] += sorterSharedMemory[idx-2];\n"
-"			GROUP_MEM_FENCE;\n"
-"		}\n"
-"#else\n"
-"		if( lIdx < 64 )\n"
-"		{\n"
-"			sorterSharedMemory[idx] += sorterSharedMemory[idx-1];\n"
-"			GROUP_MEM_FENCE;\n"
-"			sorterSharedMemory[idx] += sorterSharedMemory[idx-2];			\n"
-"			GROUP_MEM_FENCE;\n"
-"			sorterSharedMemory[idx] += sorterSharedMemory[idx-4];\n"
-"			GROUP_MEM_FENCE;\n"
-"			sorterSharedMemory[idx] += sorterSharedMemory[idx-8];\n"
-"			GROUP_MEM_FENCE;\n"
-"			sorterSharedMemory[idx] += sorterSharedMemory[idx-16];\n"
-"			GROUP_MEM_FENCE;\n"
-"			sorterSharedMemory[idx] += sorterSharedMemory[idx-32];\n"
-"			GROUP_MEM_FENCE;\n"
-"			if( wgSize > 64 )\n"
-"			{\n"
-"				sorterSharedMemory[idx] += sorterSharedMemory[idx-64];\n"
-"				GROUP_MEM_FENCE;\n"
-"			}\n"
-"			sorterSharedMemory[idx-1] += sorterSharedMemory[idx-2];\n"
-"			GROUP_MEM_FENCE;\n"
-"		}\n"
-"#endif\n"
-"	}\n"
-"	GROUP_LDS_BARRIER;\n"
-"	*totalSum = sorterSharedMemory[wgSize*2-1];\n"
-"	u32 addValue = sorterSharedMemory[lIdx+wgSize-1];\n"
-"	return addValue;\n"
-"}\n"
-"//__attribute__((reqd_work_group_size(128,1,1)))\n"
-"uint4 localPrefixSum128V( uint4 pData, uint lIdx, uint* totalSum, __local u32* sorterSharedMemory )\n"
-"{\n"
-"	u32 s4 = prefixScanVectorEx( &pData );\n"
-"	u32 rank = localPrefixSum( s4, lIdx, totalSum, sorterSharedMemory, 128 );\n"
-"	return pData + make_uint4( rank, rank, rank, rank );\n"
-"}\n"
-"//__attribute__((reqd_work_group_size(64,1,1)))\n"
-"uint4 localPrefixSum64V( uint4 pData, uint lIdx, uint* totalSum, __local u32* sorterSharedMemory )\n"
-"{\n"
-"	u32 s4 = prefixScanVectorEx( &pData );\n"
-"	u32 rank = localPrefixSum( s4, lIdx, totalSum, sorterSharedMemory, 64 );\n"
-"	return pData + make_uint4( rank, rank, rank, rank );\n"
-"}\n"
-"u32 unpack4Key( u32 key, int keyIdx ){ return (key>>(keyIdx*8)) & 0xff;}\n"
-"u32 bit8Scan(u32 v)\n"
-"{\n"
-"	return (v<<8) + (v<<16) + (v<<24);\n"
-"}\n"
-"//===\n"
-"#define MY_HISTOGRAM(idx) localHistogramMat[(idx)*WG_SIZE+lIdx]\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"void StreamCountKernel( __global u32* gSrc, __global u32* histogramOut, int4 cb )\n"
-"{\n"
-"	__local u32 localHistogramMat[NUM_BUCKET*WG_SIZE];\n"
-"	u32 gIdx = GET_GLOBAL_IDX;\n"
-"	u32 lIdx = GET_LOCAL_IDX;\n"
-"	u32 wgIdx = GET_GROUP_IDX;\n"
-"	u32 wgSize = GET_GROUP_SIZE;\n"
-"	const int startBit = cb.m_startBit;\n"
-"	const int n = cb.m_n;\n"
-"	const int nWGs = cb.m_nWGs;\n"
-"	const int nBlocksPerWG = cb.m_nBlocksPerWG;\n"
-"	for(int i=0; i<NUM_BUCKET; i++)\n"
-"	{\n"
-"		MY_HISTOGRAM(i) = 0;\n"
-"	}\n"
-"	GROUP_LDS_BARRIER;\n"
-"	const int blockSize = ELEMENTS_PER_WORK_ITEM*WG_SIZE;\n"
-"	u32 localKey;\n"
-"	int nBlocks = (n)/blockSize - nBlocksPerWG*wgIdx;\n"
-"	int addr = blockSize*nBlocksPerWG*wgIdx + ELEMENTS_PER_WORK_ITEM*lIdx;\n"
-"	for(int iblock=0; iblock<min(nBlocksPerWG, nBlocks); iblock++, addr+=blockSize)\n"
-"	{\n"
-"		//	MY_HISTOGRAM( localKeys.x ) ++ is much expensive than atomic add as it requires read and write while atomics can just add on AMD\n"
-"		//	Using registers didn't perform well. It seems like use localKeys to address requires a lot of alu ops\n"
-"		//	AMD: AtomInc performs better while NV prefers ++\n"
-"		for(int i=0; i<ELEMENTS_PER_WORK_ITEM; i++)\n"
-"		{\n"
-"#if defined(CHECK_BOUNDARY)\n"
-"			if( addr+i < n )\n"
-"#endif\n"
-"			{\n"
-"				localKey = (gSrc[addr+i]>>startBit) & 0xf;\n"
-"#if defined(NV_GPU)\n"
-"				MY_HISTOGRAM( localKey )++;\n"
-"#else\n"
-"				AtomInc( MY_HISTOGRAM( localKey ) );\n"
-"#endif\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	GROUP_LDS_BARRIER;\n"
-"	\n"
-"	if( lIdx < NUM_BUCKET )\n"
-"	{\n"
-"		u32 sum = 0;\n"
-"		for(int i=0; i<GET_GROUP_SIZE; i++)\n"
-"		{\n"
-"			sum += localHistogramMat[lIdx*WG_SIZE+(i+lIdx)%GET_GROUP_SIZE];\n"
-"		}\n"
-"		histogramOut[lIdx*nWGs+wgIdx] = sum;\n"
-"	}\n"
-"}\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"void StreamCountSortDataKernel( __global SortDataCL* gSrc, __global u32* histogramOut, int4  cb )\n"
-"{\n"
-"	__local u32 localHistogramMat[NUM_BUCKET*WG_SIZE];\n"
-"	u32 gIdx = GET_GLOBAL_IDX;\n"
-"	u32 lIdx = GET_LOCAL_IDX;\n"
-"	u32 wgIdx = GET_GROUP_IDX;\n"
-"	u32 wgSize = GET_GROUP_SIZE;\n"
-"	const int startBit = cb.m_startBit;\n"
-"	const int n = cb.m_n;\n"
-"	const int nWGs = cb.m_nWGs;\n"
-"	const int nBlocksPerWG = cb.m_nBlocksPerWG;\n"
-"	for(int i=0; i<NUM_BUCKET; i++)\n"
-"	{\n"
-"		MY_HISTOGRAM(i) = 0;\n"
-"	}\n"
-"	GROUP_LDS_BARRIER;\n"
-"	const int blockSize = ELEMENTS_PER_WORK_ITEM*WG_SIZE;\n"
-"	u32 localKey;\n"
-"	int nBlocks = (n)/blockSize - nBlocksPerWG*wgIdx;\n"
-"	int addr = blockSize*nBlocksPerWG*wgIdx + ELEMENTS_PER_WORK_ITEM*lIdx;\n"
-"	for(int iblock=0; iblock<min(nBlocksPerWG, nBlocks); iblock++, addr+=blockSize)\n"
-"	{\n"
-"		//	MY_HISTOGRAM( localKeys.x ) ++ is much expensive than atomic add as it requires read and write while atomics can just add on AMD\n"
-"		//	Using registers didn't perform well. It seems like use localKeys to address requires a lot of alu ops\n"
-"		//	AMD: AtomInc performs better while NV prefers ++\n"
-"		for(int i=0; i<ELEMENTS_PER_WORK_ITEM; i++)\n"
-"		{\n"
-"#if defined(CHECK_BOUNDARY)\n"
-"			if( addr+i < n )\n"
-"#endif\n"
-"			{\n"
-"				localKey = (gSrc[addr+i].m_key>>startBit) & 0xf;\n"
-"#if defined(NV_GPU)\n"
-"				MY_HISTOGRAM( localKey )++;\n"
-"#else\n"
-"				AtomInc( MY_HISTOGRAM( localKey ) );\n"
-"#endif\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	GROUP_LDS_BARRIER;\n"
-"	\n"
-"	if( lIdx < NUM_BUCKET )\n"
-"	{\n"
-"		u32 sum = 0;\n"
-"		for(int i=0; i<GET_GROUP_SIZE; i++)\n"
-"		{\n"
-"			sum += localHistogramMat[lIdx*WG_SIZE+(i+lIdx)%GET_GROUP_SIZE];\n"
-"		}\n"
-"		histogramOut[lIdx*nWGs+wgIdx] = sum;\n"
-"	}\n"
-"}\n"
-"#define nPerLane (nPerWI/4)\n"
-"//	NUM_BUCKET*nWGs < 128*nPerWI\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(128,1,1)))\n"
-"void PrefixScanKernel( __global u32* wHistogram1, int4  cb )\n"
-"{\n"
-"	__local u32 ldsTopScanData[128*2];\n"
-"	u32 lIdx = GET_LOCAL_IDX;\n"
-"	u32 wgIdx = GET_GROUP_IDX;\n"
-"	const int nWGs = cb.m_nWGs;\n"
-"	u32 data[nPerWI];\n"
-"	for(int i=0; i<nPerWI; i++)\n"
-"	{\n"
-"		data[i] = 0;\n"
-"		if( (nPerWI*lIdx+i) < NUM_BUCKET*nWGs )\n"
-"			data[i] = wHistogram1[nPerWI*lIdx+i];\n"
-"	}\n"
-"	uint4 myData = make_uint4(0,0,0,0);\n"
-"	for(int i=0; i<nPerLane; i++)\n"
-"	{\n"
-"		myData.x += data[nPerLane*0+i];\n"
-"		myData.y += data[nPerLane*1+i];\n"
-"		myData.z += data[nPerLane*2+i];\n"
-"		myData.w += data[nPerLane*3+i];\n"
-"	}\n"
-"	uint totalSum;\n"
-"	uint4 scanned = localPrefixSum128V( myData, lIdx, &totalSum, ldsTopScanData );\n"
-"//	for(int j=0; j<4; j++) //	somehow it introduces a lot of branches\n"
-"	{	int j = 0;\n"
-"		u32 sum = 0;\n"
-"		for(int i=0; i<nPerLane; i++)\n"
-"		{\n"
-"			u32 tmp = data[nPerLane*j+i];\n"
-"			data[nPerLane*j+i] = sum;\n"
-"			sum += tmp;\n"
-"		}\n"
-"	}\n"
-"	{	int j = 1;\n"
-"		u32 sum = 0;\n"
-"		for(int i=0; i<nPerLane; i++)\n"
-"		{\n"
-"			u32 tmp = data[nPerLane*j+i];\n"
-"			data[nPerLane*j+i] = sum;\n"
-"			sum += tmp;\n"
-"		}\n"
-"	}\n"
-"	{	int j = 2;\n"
-"		u32 sum = 0;\n"
-"		for(int i=0; i<nPerLane; i++)\n"
-"		{\n"
-"			u32 tmp = data[nPerLane*j+i];\n"
-"			data[nPerLane*j+i] = sum;\n"
-"			sum += tmp;\n"
-"		}\n"
-"	}\n"
-"	{	int j = 3;\n"
-"		u32 sum = 0;\n"
-"		for(int i=0; i<nPerLane; i++)\n"
-"		{\n"
-"			u32 tmp = data[nPerLane*j+i];\n"
-"			data[nPerLane*j+i] = sum;\n"
-"			sum += tmp;\n"
-"		}\n"
-"	}\n"
-"	for(int i=0; i<nPerLane; i++)\n"
-"	{\n"
-"		data[nPerLane*0+i] += scanned.x;\n"
-"		data[nPerLane*1+i] += scanned.y;\n"
-"		data[nPerLane*2+i] += scanned.z;\n"
-"		data[nPerLane*3+i] += scanned.w;\n"
-"	}\n"
-"	for(int i=0; i<nPerWI; i++)\n"
-"	{\n"
-"		int index = nPerWI*lIdx+i;\n"
-"		if (index < NUM_BUCKET*nWGs)\n"
-"			wHistogram1[nPerWI*lIdx+i] = data[i];\n"
-"	}\n"
-"}\n"
-"//	4 scan, 4 exchange\n"
-"void sort4Bits(u32 sortData[4], int startBit, int lIdx, __local u32* ldsSortData)\n"
-"{\n"
-"	for(int bitIdx=0; bitIdx<BITS_PER_PASS; bitIdx++)\n"
-"	{\n"
-"		u32 mask = (1<<bitIdx);\n"
-"		uint4 cmpResult = make_uint4( (sortData[0]>>startBit) & mask, (sortData[1]>>startBit) & mask, (sortData[2]>>startBit) & mask, (sortData[3]>>startBit) & mask );\n"
-"		uint4 prefixSum = SELECT_UINT4( make_uint4(1,1,1,1), make_uint4(0,0,0,0), cmpResult != make_uint4(0,0,0,0) );\n"
-"		u32 total;\n"
-"		prefixSum = localPrefixSum64V( prefixSum, lIdx, &total, ldsSortData );\n"
-"		{\n"
-"			uint4 localAddr = make_uint4(lIdx*4+0,lIdx*4+1,lIdx*4+2,lIdx*4+3);\n"
-"			uint4 dstAddr = localAddr - prefixSum + make_uint4( total, total, total, total );\n"
-"			dstAddr = SELECT_UINT4( prefixSum, dstAddr, cmpResult != make_uint4(0, 0, 0, 0) );\n"
-"			GROUP_LDS_BARRIER;\n"
-"			ldsSortData[dstAddr.x] = sortData[0];\n"
-"			ldsSortData[dstAddr.y] = sortData[1];\n"
-"			ldsSortData[dstAddr.z] = sortData[2];\n"
-"			ldsSortData[dstAddr.w] = sortData[3];\n"
-"			GROUP_LDS_BARRIER;\n"
-"			sortData[0] = ldsSortData[localAddr.x];\n"
-"			sortData[1] = ldsSortData[localAddr.y];\n"
-"			sortData[2] = ldsSortData[localAddr.z];\n"
-"			sortData[3] = ldsSortData[localAddr.w];\n"
-"			GROUP_LDS_BARRIER;\n"
-"		}\n"
-"	}\n"
-"}\n"
-"//	2 scan, 2 exchange\n"
-"void sort4Bits1(u32 sortData[4], int startBit, int lIdx, __local u32* ldsSortData)\n"
-"{\n"
-"	for(uint ibit=0; ibit<BITS_PER_PASS; ibit+=2)\n"
-"	{\n"
-"		uint4 b = make_uint4((sortData[0]>>(startBit+ibit)) & 0x3, \n"
-"			(sortData[1]>>(startBit+ibit)) & 0x3, \n"
-"			(sortData[2]>>(startBit+ibit)) & 0x3, \n"
-"			(sortData[3]>>(startBit+ibit)) & 0x3);\n"
-"		u32 key4;\n"
-"		u32 sKeyPacked[4] = { 0, 0, 0, 0 };\n"
-"		{\n"
-"			sKeyPacked[0] |= 1<<(8*b.x);\n"
-"			sKeyPacked[1] |= 1<<(8*b.y);\n"
-"			sKeyPacked[2] |= 1<<(8*b.z);\n"
-"			sKeyPacked[3] |= 1<<(8*b.w);\n"
-"			key4 = sKeyPacked[0] + sKeyPacked[1] + sKeyPacked[2] + sKeyPacked[3];\n"
-"		}\n"
-"		u32 rankPacked;\n"
-"		u32 sumPacked;\n"
-"		{\n"
-"			rankPacked = localPrefixSum( key4, lIdx, &sumPacked, ldsSortData, WG_SIZE );\n"
-"		}\n"
-"		GROUP_LDS_BARRIER;\n"
-"		u32 newOffset[4] = { 0,0,0,0 };\n"
-"		{\n"
-"			u32 sumScanned = bit8Scan( sumPacked );\n"
-"			u32 scannedKeys[4];\n"
-"			scannedKeys[0] = 1<<(8*b.x);\n"
-"			scannedKeys[1] = 1<<(8*b.y);\n"
-"			scannedKeys[2] = 1<<(8*b.z);\n"
-"			scannedKeys[3] = 1<<(8*b.w);\n"
-"			{	//	4 scans at once\n"
-"				u32 sum4 = 0;\n"
-"				for(int ie=0; ie<4; ie++)\n"
-"				{\n"
-"					u32 tmp = scannedKeys[ie];\n"
-"					scannedKeys[ie] = sum4;\n"
-"					sum4 += tmp;\n"
-"				}\n"
-"			}\n"
-"			{\n"
-"				u32 sumPlusRank = sumScanned + rankPacked;\n"
-"				{	u32 ie = b.x;\n"
-"					scannedKeys[0] += sumPlusRank;\n"
-"					newOffset[0] = unpack4Key( scannedKeys[0], ie );\n"
-"				}\n"
-"				{	u32 ie = b.y;\n"
-"					scannedKeys[1] += sumPlusRank;\n"
-"					newOffset[1] = unpack4Key( scannedKeys[1], ie );\n"
-"				}\n"
-"				{	u32 ie = b.z;\n"
-"					scannedKeys[2] += sumPlusRank;\n"
-"					newOffset[2] = unpack4Key( scannedKeys[2], ie );\n"
-"				}\n"
-"				{	u32 ie = b.w;\n"
-"					scannedKeys[3] += sumPlusRank;\n"
-"					newOffset[3] = unpack4Key( scannedKeys[3], ie );\n"
-"				}\n"
-"			}\n"
-"		}\n"
-"		GROUP_LDS_BARRIER;\n"
-"		{\n"
-"			ldsSortData[newOffset[0]] = sortData[0];\n"
-"			ldsSortData[newOffset[1]] = sortData[1];\n"
-"			ldsSortData[newOffset[2]] = sortData[2];\n"
-"			ldsSortData[newOffset[3]] = sortData[3];\n"
-"			GROUP_LDS_BARRIER;\n"
-"			u32 dstAddr = 4*lIdx;\n"
-"			sortData[0] = ldsSortData[dstAddr+0];\n"
-"			sortData[1] = ldsSortData[dstAddr+1];\n"
-"			sortData[2] = ldsSortData[dstAddr+2];\n"
-"			sortData[3] = ldsSortData[dstAddr+3];\n"
-"			GROUP_LDS_BARRIER;\n"
-"		}\n"
-"	}\n"
-"}\n"
-"#define SET_HISTOGRAM(setIdx, key) ldsSortData[(setIdx)*NUM_BUCKET+key]\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"void SortAndScatterKernel( __global const u32* restrict gSrc, __global const u32* rHistogram, __global u32* restrict gDst, int4  cb )\n"
-"{\n"
-"	__local u32 ldsSortData[WG_SIZE*ELEMENTS_PER_WORK_ITEM+16];\n"
-"	__local u32 localHistogramToCarry[NUM_BUCKET];\n"
-"	__local u32 localHistogram[NUM_BUCKET*2];\n"
-"	u32 gIdx = GET_GLOBAL_IDX;\n"
-"	u32 lIdx = GET_LOCAL_IDX;\n"
-"	u32 wgIdx = GET_GROUP_IDX;\n"
-"	u32 wgSize = GET_GROUP_SIZE;\n"
-"	const int n = cb.m_n;\n"
-"	const int nWGs = cb.m_nWGs;\n"
-"	const int startBit = cb.m_startBit;\n"
-"	const int nBlocksPerWG = cb.m_nBlocksPerWG;\n"
-"	if( lIdx < (NUM_BUCKET) )\n"
-"	{\n"
-"		localHistogramToCarry[lIdx] = rHistogram[lIdx*nWGs + wgIdx];\n"
-"	}\n"
-"	GROUP_LDS_BARRIER;\n"
-"	const int blockSize = ELEMENTS_PER_WORK_ITEM*WG_SIZE;\n"
-"	int nBlocks = n/blockSize - nBlocksPerWG*wgIdx;\n"
-"	int addr = blockSize*nBlocksPerWG*wgIdx + ELEMENTS_PER_WORK_ITEM*lIdx;\n"
-"	for(int iblock=0; iblock<min(nBlocksPerWG, nBlocks); iblock++, addr+=blockSize)\n"
-"	{\n"
-"		u32 myHistogram = 0;\n"
-"		u32 sortData[ELEMENTS_PER_WORK_ITEM];\n"
-"		for(int i=0; i<ELEMENTS_PER_WORK_ITEM; i++)\n"
-"#if defined(CHECK_BOUNDARY)\n"
-"			sortData[i] = ( addr+i < n )? gSrc[ addr+i ] : 0xffffffff;\n"
-"#else\n"
-"			sortData[i] = gSrc[ addr+i ];\n"
-"#endif\n"
-"		sort4Bits(sortData, startBit, lIdx, ldsSortData);\n"
-"		u32 keys[ELEMENTS_PER_WORK_ITEM];\n"
-"		for(int i=0; i<ELEMENTS_PER_WORK_ITEM; i++)\n"
-"			keys[i] = (sortData[i]>>startBit) & 0xf;\n"
-"		{	//	create histogram\n"
-"			u32 setIdx = lIdx/16;\n"
-"			if( lIdx < NUM_BUCKET )\n"
-"			{\n"
-"				localHistogram[lIdx] = 0;\n"
-"			}\n"
-"			ldsSortData[lIdx] = 0;\n"
-"			GROUP_LDS_BARRIER;\n"
-"			for(int i=0; i<ELEMENTS_PER_WORK_ITEM; i++)\n"
-"#if defined(CHECK_BOUNDARY)\n"
-"				if( addr+i < n )\n"
-"#endif\n"
-"#if defined(NV_GPU)\n"
-"				SET_HISTOGRAM( setIdx, keys[i] )++;\n"
-"#else\n"
-"				AtomInc( SET_HISTOGRAM( setIdx, keys[i] ) );\n"
-"#endif\n"
-"			\n"
-"			GROUP_LDS_BARRIER;\n"
-"			\n"
-"			uint hIdx = NUM_BUCKET+lIdx;\n"
-"			if( lIdx < NUM_BUCKET )\n"
-"			{\n"
-"				u32 sum = 0;\n"
-"				for(int i=0; i<WG_SIZE/16; i++)\n"
-"				{\n"
-"					sum += SET_HISTOGRAM( i, lIdx );\n"
-"				}\n"
-"				myHistogram = sum;\n"
-"				localHistogram[hIdx] = sum;\n"
-"			}\n"
-"			GROUP_LDS_BARRIER;\n"
-"#if defined(USE_2LEVEL_REDUCE)\n"
-"			if( lIdx < NUM_BUCKET )\n"
-"			{\n"
-"				localHistogram[hIdx] = localHistogram[hIdx-1];\n"
-"				GROUP_MEM_FENCE;\n"
-"				u32 u0, u1, u2;\n"
-"				u0 = localHistogram[hIdx-3];\n"
-"				u1 = localHistogram[hIdx-2];\n"
-"				u2 = localHistogram[hIdx-1];\n"
-"				AtomAdd( localHistogram[hIdx], u0 + u1 + u2 );\n"
-"				GROUP_MEM_FENCE;\n"
-"				u0 = localHistogram[hIdx-12];\n"
-"				u1 = localHistogram[hIdx-8];\n"
-"				u2 = localHistogram[hIdx-4];\n"
-"				AtomAdd( localHistogram[hIdx], u0 + u1 + u2 );\n"
-"				GROUP_MEM_FENCE;\n"
-"			}\n"
-"#else\n"
-"			if( lIdx < NUM_BUCKET )\n"
-"			{\n"
-"				localHistogram[hIdx] = localHistogram[hIdx-1];\n"
-"				GROUP_MEM_FENCE;\n"
-"				localHistogram[hIdx] += localHistogram[hIdx-1];\n"
-"				GROUP_MEM_FENCE;\n"
-"				localHistogram[hIdx] += localHistogram[hIdx-2];\n"
-"				GROUP_MEM_FENCE;\n"
-"				localHistogram[hIdx] += localHistogram[hIdx-4];\n"
-"				GROUP_MEM_FENCE;\n"
-"				localHistogram[hIdx] += localHistogram[hIdx-8];\n"
-"				GROUP_MEM_FENCE;\n"
-"			}\n"
-"#endif\n"
-"			GROUP_LDS_BARRIER;\n"
-"		}\n"
-"		{\n"
-"			for(int ie=0; ie<ELEMENTS_PER_WORK_ITEM; ie++)\n"
-"			{\n"
-"				int dataIdx = ELEMENTS_PER_WORK_ITEM*lIdx+ie;\n"
-"				int binIdx = keys[ie];\n"
-"				int groupOffset = localHistogramToCarry[binIdx];\n"
-"				int myIdx = dataIdx - localHistogram[NUM_BUCKET+binIdx];\n"
-"#if defined(CHECK_BOUNDARY)\n"
-"				if( addr+ie < n )\n"
-"#endif\n"
-"				gDst[ groupOffset + myIdx ] = sortData[ie];\n"
-"			}\n"
-"		}\n"
-"		GROUP_LDS_BARRIER;\n"
-"		if( lIdx < NUM_BUCKET )\n"
-"		{\n"
-"			localHistogramToCarry[lIdx] += myHistogram;\n"
-"		}\n"
-"		GROUP_LDS_BARRIER;\n"
-"	}\n"
-"}\n"
-"//	2 scan, 2 exchange\n"
-"void sort4Bits1KeyValue(u32 sortData[4], int sortVal[4], int startBit, int lIdx, __local u32* ldsSortData, __local int *ldsSortVal)\n"
-"{\n"
-"	for(uint ibit=0; ibit<BITS_PER_PASS; ibit+=2)\n"
-"	{\n"
-"		uint4 b = make_uint4((sortData[0]>>(startBit+ibit)) & 0x3, \n"
-"			(sortData[1]>>(startBit+ibit)) & 0x3, \n"
-"			(sortData[2]>>(startBit+ibit)) & 0x3, \n"
-"			(sortData[3]>>(startBit+ibit)) & 0x3);\n"
-"		u32 key4;\n"
-"		u32 sKeyPacked[4] = { 0, 0, 0, 0 };\n"
-"		{\n"
-"			sKeyPacked[0] |= 1<<(8*b.x);\n"
-"			sKeyPacked[1] |= 1<<(8*b.y);\n"
-"			sKeyPacked[2] |= 1<<(8*b.z);\n"
-"			sKeyPacked[3] |= 1<<(8*b.w);\n"
-"			key4 = sKeyPacked[0] + sKeyPacked[1] + sKeyPacked[2] + sKeyPacked[3];\n"
-"		}\n"
-"		u32 rankPacked;\n"
-"		u32 sumPacked;\n"
-"		{\n"
-"			rankPacked = localPrefixSum( key4, lIdx, &sumPacked, ldsSortData, WG_SIZE );\n"
-"		}\n"
-"		GROUP_LDS_BARRIER;\n"
-"		u32 newOffset[4] = { 0,0,0,0 };\n"
-"		{\n"
-"			u32 sumScanned = bit8Scan( sumPacked );\n"
-"			u32 scannedKeys[4];\n"
-"			scannedKeys[0] = 1<<(8*b.x);\n"
-"			scannedKeys[1] = 1<<(8*b.y);\n"
-"			scannedKeys[2] = 1<<(8*b.z);\n"
-"			scannedKeys[3] = 1<<(8*b.w);\n"
-"			{	//	4 scans at once\n"
-"				u32 sum4 = 0;\n"
-"				for(int ie=0; ie<4; ie++)\n"
-"				{\n"
-"					u32 tmp = scannedKeys[ie];\n"
-"					scannedKeys[ie] = sum4;\n"
-"					sum4 += tmp;\n"
-"				}\n"
-"			}\n"
-"			{\n"
-"				u32 sumPlusRank = sumScanned + rankPacked;\n"
-"				{	u32 ie = b.x;\n"
-"					scannedKeys[0] += sumPlusRank;\n"
-"					newOffset[0] = unpack4Key( scannedKeys[0], ie );\n"
-"				}\n"
-"				{	u32 ie = b.y;\n"
-"					scannedKeys[1] += sumPlusRank;\n"
-"					newOffset[1] = unpack4Key( scannedKeys[1], ie );\n"
-"				}\n"
-"				{	u32 ie = b.z;\n"
-"					scannedKeys[2] += sumPlusRank;\n"
-"					newOffset[2] = unpack4Key( scannedKeys[2], ie );\n"
-"				}\n"
-"				{	u32 ie = b.w;\n"
-"					scannedKeys[3] += sumPlusRank;\n"
-"					newOffset[3] = unpack4Key( scannedKeys[3], ie );\n"
-"				}\n"
-"			}\n"
-"		}\n"
-"		GROUP_LDS_BARRIER;\n"
-"		{\n"
-"			ldsSortData[newOffset[0]] = sortData[0];\n"
-"			ldsSortData[newOffset[1]] = sortData[1];\n"
-"			ldsSortData[newOffset[2]] = sortData[2];\n"
-"			ldsSortData[newOffset[3]] = sortData[3];\n"
-"			ldsSortVal[newOffset[0]] = sortVal[0];\n"
-"			ldsSortVal[newOffset[1]] = sortVal[1];\n"
-"			ldsSortVal[newOffset[2]] = sortVal[2];\n"
-"			ldsSortVal[newOffset[3]] = sortVal[3];\n"
-"			GROUP_LDS_BARRIER;\n"
-"			u32 dstAddr = 4*lIdx;\n"
-"			sortData[0] = ldsSortData[dstAddr+0];\n"
-"			sortData[1] = ldsSortData[dstAddr+1];\n"
-"			sortData[2] = ldsSortData[dstAddr+2];\n"
-"			sortData[3] = ldsSortData[dstAddr+3];\n"
-"			sortVal[0] = ldsSortVal[dstAddr+0];\n"
-"			sortVal[1] = ldsSortVal[dstAddr+1];\n"
-"			sortVal[2] = ldsSortVal[dstAddr+2];\n"
-"			sortVal[3] = ldsSortVal[dstAddr+3];\n"
-"			GROUP_LDS_BARRIER;\n"
-"		}\n"
-"	}\n"
-"}\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"void SortAndScatterSortDataKernel( __global const SortDataCL* restrict gSrc, __global const u32* rHistogram, __global SortDataCL* restrict gDst, int4 cb)\n"
-"{\n"
-"	__local int ldsSortData[WG_SIZE*ELEMENTS_PER_WORK_ITEM+16];\n"
-"	__local int ldsSortVal[WG_SIZE*ELEMENTS_PER_WORK_ITEM+16];\n"
-"	__local u32 localHistogramToCarry[NUM_BUCKET];\n"
-"	__local u32 localHistogram[NUM_BUCKET*2];\n"
-"	u32 gIdx = GET_GLOBAL_IDX;\n"
-"	u32 lIdx = GET_LOCAL_IDX;\n"
-"	u32 wgIdx = GET_GROUP_IDX;\n"
-"	u32 wgSize = GET_GROUP_SIZE;\n"
-"	const int n = cb.m_n;\n"
-"	const int nWGs = cb.m_nWGs;\n"
-"	const int startBit = cb.m_startBit;\n"
-"	const int nBlocksPerWG = cb.m_nBlocksPerWG;\n"
-"	if( lIdx < (NUM_BUCKET) )\n"
-"	{\n"
-"		localHistogramToCarry[lIdx] = rHistogram[lIdx*nWGs + wgIdx];\n"
-"	}\n"
-"	GROUP_LDS_BARRIER;\n"
-"    \n"
-"	const int blockSize = ELEMENTS_PER_WORK_ITEM*WG_SIZE;\n"
-"	int nBlocks = n/blockSize - nBlocksPerWG*wgIdx;\n"
-"	int addr = blockSize*nBlocksPerWG*wgIdx + ELEMENTS_PER_WORK_ITEM*lIdx;\n"
-"	for(int iblock=0; iblock<min(nBlocksPerWG, nBlocks); iblock++, addr+=blockSize)\n"
-"	{\n"
-"		u32 myHistogram = 0;\n"
-"		int sortData[ELEMENTS_PER_WORK_ITEM];\n"
-"		int sortVal[ELEMENTS_PER_WORK_ITEM];\n"
-"		for(int i=0; i<ELEMENTS_PER_WORK_ITEM; i++)\n"
-"#if defined(CHECK_BOUNDARY)\n"
-"		{\n"
-"			sortData[i] = ( addr+i < n )? gSrc[ addr+i ].m_key : 0xffffffff;\n"
-"			sortVal[i] = ( addr+i < n )? gSrc[ addr+i ].m_value : 0xffffffff;\n"
-"		}\n"
-"#else\n"
-"		{\n"
-"			sortData[i] = gSrc[ addr+i ].m_key;\n"
-"			sortVal[i] = gSrc[ addr+i ].m_value;\n"
-"		}\n"
-"#endif\n"
-"		sort4Bits1KeyValue(sortData, sortVal, startBit, lIdx, ldsSortData, ldsSortVal);\n"
-"		u32 keys[ELEMENTS_PER_WORK_ITEM];\n"
-"		for(int i=0; i<ELEMENTS_PER_WORK_ITEM; i++)\n"
-"			keys[i] = (sortData[i]>>startBit) & 0xf;\n"
-"		{	//	create histogram\n"
-"			u32 setIdx = lIdx/16;\n"
-"			if( lIdx < NUM_BUCKET )\n"
-"			{\n"
-"				localHistogram[lIdx] = 0;\n"
-"			}\n"
-"			ldsSortData[lIdx] = 0;\n"
-"			GROUP_LDS_BARRIER;\n"
-"			for(int i=0; i<ELEMENTS_PER_WORK_ITEM; i++)\n"
-"#if defined(CHECK_BOUNDARY)\n"
-"				if( addr+i < n )\n"
-"#endif\n"
-"#if defined(NV_GPU)\n"
-"				SET_HISTOGRAM( setIdx, keys[i] )++;\n"
-"#else\n"
-"				AtomInc( SET_HISTOGRAM( setIdx, keys[i] ) );\n"
-"#endif\n"
-"			\n"
-"			GROUP_LDS_BARRIER;\n"
-"			\n"
-"			uint hIdx = NUM_BUCKET+lIdx;\n"
-"			if( lIdx < NUM_BUCKET )\n"
-"			{\n"
-"				u32 sum = 0;\n"
-"				for(int i=0; i<WG_SIZE/16; i++)\n"
-"				{\n"
-"					sum += SET_HISTOGRAM( i, lIdx );\n"
-"				}\n"
-"				myHistogram = sum;\n"
-"				localHistogram[hIdx] = sum;\n"
-"			}\n"
-"			GROUP_LDS_BARRIER;\n"
-"#if defined(USE_2LEVEL_REDUCE)\n"
-"			if( lIdx < NUM_BUCKET )\n"
-"			{\n"
-"				localHistogram[hIdx] = localHistogram[hIdx-1];\n"
-"				GROUP_MEM_FENCE;\n"
-"				u32 u0, u1, u2;\n"
-"				u0 = localHistogram[hIdx-3];\n"
-"				u1 = localHistogram[hIdx-2];\n"
-"				u2 = localHistogram[hIdx-1];\n"
-"				AtomAdd( localHistogram[hIdx], u0 + u1 + u2 );\n"
-"				GROUP_MEM_FENCE;\n"
-"				u0 = localHistogram[hIdx-12];\n"
-"				u1 = localHistogram[hIdx-8];\n"
-"				u2 = localHistogram[hIdx-4];\n"
-"				AtomAdd( localHistogram[hIdx], u0 + u1 + u2 );\n"
-"				GROUP_MEM_FENCE;\n"
-"			}\n"
-"#else\n"
-"			if( lIdx < NUM_BUCKET )\n"
-"			{\n"
-"				localHistogram[hIdx] = localHistogram[hIdx-1];\n"
-"				GROUP_MEM_FENCE;\n"
-"				localHistogram[hIdx] += localHistogram[hIdx-1];\n"
-"				GROUP_MEM_FENCE;\n"
-"				localHistogram[hIdx] += localHistogram[hIdx-2];\n"
-"				GROUP_MEM_FENCE;\n"
-"				localHistogram[hIdx] += localHistogram[hIdx-4];\n"
-"				GROUP_MEM_FENCE;\n"
-"				localHistogram[hIdx] += localHistogram[hIdx-8];\n"
-"				GROUP_MEM_FENCE;\n"
-"			}\n"
-"#endif\n"
-"			GROUP_LDS_BARRIER;\n"
-"		}\n"
-"    	{\n"
-"			for(int ie=0; ie<ELEMENTS_PER_WORK_ITEM; ie++)\n"
-"			{\n"
-"				int dataIdx = ELEMENTS_PER_WORK_ITEM*lIdx+ie;\n"
-"				int binIdx = keys[ie];\n"
-"				int groupOffset = localHistogramToCarry[binIdx];\n"
-"				int myIdx = dataIdx - localHistogram[NUM_BUCKET+binIdx];\n"
-"#if defined(CHECK_BOUNDARY)\n"
-"				if( addr+ie < n )\n"
-"				{\n"
-"                    if ((groupOffset + myIdx)<n)\n"
-"                    {\n"
-"                        if (sortData[ie]==sortVal[ie])\n"
-"                        {\n"
-"                            \n"
-"                            SortDataCL tmp;\n"
-"                            tmp.m_key = sortData[ie];\n"
-"                            tmp.m_value = sortVal[ie];\n"
-"                            if (tmp.m_key == tmp.m_value)\n"
-"                                gDst[groupOffset + myIdx ] = tmp;\n"
-"                        }\n"
-"                        \n"
-"                    }\n"
-"				}\n"
-"#else\n"
-"                if ((groupOffset + myIdx)<n)\n"
-"                {\n"
-"                    gDst[ groupOffset + myIdx ].m_key = sortData[ie];\n"
-"                    gDst[ groupOffset + myIdx ].m_value = sortVal[ie];\n"
-"                }\n"
-"#endif\n"
-"			}\n"
-"		}\n"
-"		GROUP_LDS_BARRIER;\n"
-"		if( lIdx < NUM_BUCKET )\n"
-"		{\n"
-"			localHistogramToCarry[lIdx] += myHistogram;\n"
-"		}\n"
-"		GROUP_LDS_BARRIER;\n"
-"	}\n"
-"}\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"void SortAndScatterSortDataKernelSerial( __global const SortDataCL* restrict gSrc, __global const u32* rHistogram, __global SortDataCL* restrict gDst, int4 cb)\n"
-"{\n"
-"    \n"
-"	u32 gIdx = GET_GLOBAL_IDX;\n"
-"	u32 realLocalIdx = GET_LOCAL_IDX;\n"
-"	u32 wgIdx = GET_GROUP_IDX;\n"
-"	u32 wgSize = GET_GROUP_SIZE;\n"
-"	const int startBit = cb.m_startBit;\n"
-"	const int n = cb.m_n;\n"
-"	const int nWGs = cb.m_nWGs;\n"
-"	const int nBlocksPerWG = cb.m_nBlocksPerWG;\n"
-"    int counter[NUM_BUCKET];\n"
-"    \n"
-"    if (realLocalIdx>0)\n"
-"        return;\n"
-"    \n"
-"    for (int c=0;c<NUM_BUCKET;c++)\n"
-"        counter[c]=0;\n"
-"    const int blockSize = ELEMENTS_PER_WORK_ITEM*WG_SIZE;\n"
-"	\n"
-"	int nBlocks = (n)/blockSize - nBlocksPerWG*wgIdx;\n"
-"   for(int iblock=0; iblock<min(nBlocksPerWG, nBlocks); iblock++)\n"
-"  {\n"
-"     for (int lIdx=0;lIdx<WG_SIZE;lIdx++)\n"
-" 	{\n"
-"        int addr2 = iblock*blockSize + blockSize*nBlocksPerWG*wgIdx + ELEMENTS_PER_WORK_ITEM*lIdx;\n"
-"        \n"
-"		for(int j=0; j<ELEMENTS_PER_WORK_ITEM; j++)\n"
-"		{\n"
-"            int i = addr2+j;\n"
-"			if( i < n )\n"
-"			{\n"
-"                int tableIdx;\n"
-"				tableIdx = (gSrc[i].m_key>>startBit) & 0xf;//0xf = NUM_TABLES-1\n"
-"                gDst[rHistogram[tableIdx*nWGs+wgIdx] + counter[tableIdx]] = gSrc[i];\n"
-"                counter[tableIdx] ++;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"  }\n"
-"    \n"
-"}\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"void SortAndScatterKernelSerial( __global const u32* restrict gSrc, __global const u32* rHistogram, __global u32* restrict gDst, int4  cb )\n"
-"{\n"
-"    \n"
-"	u32 gIdx = GET_GLOBAL_IDX;\n"
-"	u32 realLocalIdx = GET_LOCAL_IDX;\n"
-"	u32 wgIdx = GET_GROUP_IDX;\n"
-"	u32 wgSize = GET_GROUP_SIZE;\n"
-"	const int startBit = cb.m_startBit;\n"
-"	const int n = cb.m_n;\n"
-"	const int nWGs = cb.m_nWGs;\n"
-"	const int nBlocksPerWG = cb.m_nBlocksPerWG;\n"
-"    int counter[NUM_BUCKET];\n"
-"    \n"
-"    if (realLocalIdx>0)\n"
-"        return;\n"
-"    \n"
-"    for (int c=0;c<NUM_BUCKET;c++)\n"
-"        counter[c]=0;\n"
-"    const int blockSize = ELEMENTS_PER_WORK_ITEM*WG_SIZE;\n"
-"	\n"
-"	int nBlocks = (n)/blockSize - nBlocksPerWG*wgIdx;\n"
-"   for(int iblock=0; iblock<min(nBlocksPerWG, nBlocks); iblock++)\n"
-"  {\n"
-"     for (int lIdx=0;lIdx<WG_SIZE;lIdx++)\n"
-" 	{\n"
-"        int addr2 = iblock*blockSize + blockSize*nBlocksPerWG*wgIdx + ELEMENTS_PER_WORK_ITEM*lIdx;\n"
-"        \n"
-"		for(int j=0; j<ELEMENTS_PER_WORK_ITEM; j++)\n"
-"		{\n"
-"            int i = addr2+j;\n"
-"			if( i < n )\n"
-"			{\n"
-"                int tableIdx;\n"
-"				tableIdx = (gSrc[i]>>startBit) & 0xf;//0xf = NUM_TABLES-1\n"
-"                gDst[rHistogram[tableIdx*nWGs+wgIdx] + counter[tableIdx]] = gSrc[i];\n"
-"                counter[tableIdx] ++;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"  }\n"
-"    \n"
-"}\n"
-;
+static const char* radixSort32KernelsCL =
+	"/*\n"
+	"Bullet Continuous Collision Detection and Physics Library\n"
+	"Copyright (c) 2011 Advanced Micro Devices, Inc.  http://bulletphysics.org\n"
+	"This software is provided 'as-is', without any express or implied warranty.\n"
+	"In no event will the authors be held liable for any damages arising from the use of this software.\n"
+	"Permission is granted to anyone to use this software for any purpose, \n"
+	"including commercial applications, and to alter it and redistribute it freely, \n"
+	"subject to the following restrictions:\n"
+	"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
+	"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
+	"3. This notice may not be removed or altered from any source distribution.\n"
+	"*/\n"
+	"//Author Takahiro Harada\n"
+	"//#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
+	"typedef unsigned int u32;\n"
+	"#define GET_GROUP_IDX get_group_id(0)\n"
+	"#define GET_LOCAL_IDX get_local_id(0)\n"
+	"#define GET_GLOBAL_IDX get_global_id(0)\n"
+	"#define GET_GROUP_SIZE get_local_size(0)\n"
+	"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
+	"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
+	"#define AtomInc(x) atom_inc(&(x))\n"
+	"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
+	"#define AtomAdd(x, value) atom_add(&(x), value)\n"
+	"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
+	"#define make_uint4 (uint4)\n"
+	"#define make_uint2 (uint2)\n"
+	"#define make_int2 (int2)\n"
+	"#define WG_SIZE 64\n"
+	"#define ELEMENTS_PER_WORK_ITEM (256/WG_SIZE)\n"
+	"#define BITS_PER_PASS 4\n"
+	"#define NUM_BUCKET (1<<BITS_PER_PASS)\n"
+	"typedef uchar u8;\n"
+	"//	this isn't optimization for VLIW. But just reducing writes. \n"
+	"#define USE_2LEVEL_REDUCE 1\n"
+	"//#define CHECK_BOUNDARY 1\n"
+	"//#define NV_GPU 1\n"
+	"//	Cypress\n"
+	"#define nPerWI 16\n"
+	"//	Cayman\n"
+	"//#define nPerWI 20\n"
+	"#define m_n x\n"
+	"#define m_nWGs y\n"
+	"#define m_startBit z\n"
+	"#define m_nBlocksPerWG w\n"
+	"/*\n"
+	"typedef struct\n"
+	"{\n"
+	"	int m_n;\n"
+	"	int m_nWGs;\n"
+	"	int m_startBit;\n"
+	"	int m_nBlocksPerWG;\n"
+	"} ConstBuffer;\n"
+	"*/\n"
+	"typedef struct\n"
+	"{\n"
+	"	unsigned int m_key;\n"
+	"	unsigned int m_value;\n"
+	"} SortDataCL;\n"
+	"uint prefixScanVectorEx( uint4* data )\n"
+	"{\n"
+	"	u32 sum = 0;\n"
+	"	u32 tmp = data[0].x;\n"
+	"	data[0].x = sum;\n"
+	"	sum += tmp;\n"
+	"	tmp = data[0].y;\n"
+	"	data[0].y = sum;\n"
+	"	sum += tmp;\n"
+	"	tmp = data[0].z;\n"
+	"	data[0].z = sum;\n"
+	"	sum += tmp;\n"
+	"	tmp = data[0].w;\n"
+	"	data[0].w = sum;\n"
+	"	sum += tmp;\n"
+	"	return sum;\n"
+	"}\n"
+	"u32 localPrefixSum( u32 pData, uint lIdx, uint* totalSum, __local u32* sorterSharedMemory, int wgSize /*64 or 128*/ )\n"
+	"{\n"
+	"	{	//	Set data\n"
+	"		sorterSharedMemory[lIdx] = 0;\n"
+	"		sorterSharedMemory[lIdx+wgSize] = pData;\n"
+	"	}\n"
+	"	GROUP_LDS_BARRIER;\n"
+	"	{	//	Prefix sum\n"
+	"		int idx = 2*lIdx + (wgSize+1);\n"
+	"#if defined(USE_2LEVEL_REDUCE)\n"
+	"		if( lIdx < 64 )\n"
+	"		{\n"
+	"			u32 u0, u1, u2;\n"
+	"			u0 = sorterSharedMemory[idx-3];\n"
+	"			u1 = sorterSharedMemory[idx-2];\n"
+	"			u2 = sorterSharedMemory[idx-1];\n"
+	"			AtomAdd( sorterSharedMemory[idx], u0+u1+u2 );			\n"
+	"			GROUP_MEM_FENCE;\n"
+	"			u0 = sorterSharedMemory[idx-12];\n"
+	"			u1 = sorterSharedMemory[idx-8];\n"
+	"			u2 = sorterSharedMemory[idx-4];\n"
+	"			AtomAdd( sorterSharedMemory[idx], u0+u1+u2 );			\n"
+	"			GROUP_MEM_FENCE;\n"
+	"			u0 = sorterSharedMemory[idx-48];\n"
+	"			u1 = sorterSharedMemory[idx-32];\n"
+	"			u2 = sorterSharedMemory[idx-16];\n"
+	"			AtomAdd( sorterSharedMemory[idx], u0+u1+u2 );			\n"
+	"			GROUP_MEM_FENCE;\n"
+	"			if( wgSize > 64 )\n"
+	"			{\n"
+	"				sorterSharedMemory[idx] += sorterSharedMemory[idx-64];\n"
+	"				GROUP_MEM_FENCE;\n"
+	"			}\n"
+	"			sorterSharedMemory[idx-1] += sorterSharedMemory[idx-2];\n"
+	"			GROUP_MEM_FENCE;\n"
+	"		}\n"
+	"#else\n"
+	"		if( lIdx < 64 )\n"
+	"		{\n"
+	"			sorterSharedMemory[idx] += sorterSharedMemory[idx-1];\n"
+	"			GROUP_MEM_FENCE;\n"
+	"			sorterSharedMemory[idx] += sorterSharedMemory[idx-2];			\n"
+	"			GROUP_MEM_FENCE;\n"
+	"			sorterSharedMemory[idx] += sorterSharedMemory[idx-4];\n"
+	"			GROUP_MEM_FENCE;\n"
+	"			sorterSharedMemory[idx] += sorterSharedMemory[idx-8];\n"
+	"			GROUP_MEM_FENCE;\n"
+	"			sorterSharedMemory[idx] += sorterSharedMemory[idx-16];\n"
+	"			GROUP_MEM_FENCE;\n"
+	"			sorterSharedMemory[idx] += sorterSharedMemory[idx-32];\n"
+	"			GROUP_MEM_FENCE;\n"
+	"			if( wgSize > 64 )\n"
+	"			{\n"
+	"				sorterSharedMemory[idx] += sorterSharedMemory[idx-64];\n"
+	"				GROUP_MEM_FENCE;\n"
+	"			}\n"
+	"			sorterSharedMemory[idx-1] += sorterSharedMemory[idx-2];\n"
+	"			GROUP_MEM_FENCE;\n"
+	"		}\n"
+	"#endif\n"
+	"	}\n"
+	"	GROUP_LDS_BARRIER;\n"
+	"	*totalSum = sorterSharedMemory[wgSize*2-1];\n"
+	"	u32 addValue = sorterSharedMemory[lIdx+wgSize-1];\n"
+	"	return addValue;\n"
+	"}\n"
+	"//__attribute__((reqd_work_group_size(128,1,1)))\n"
+	"uint4 localPrefixSum128V( uint4 pData, uint lIdx, uint* totalSum, __local u32* sorterSharedMemory )\n"
+	"{\n"
+	"	u32 s4 = prefixScanVectorEx( &pData );\n"
+	"	u32 rank = localPrefixSum( s4, lIdx, totalSum, sorterSharedMemory, 128 );\n"
+	"	return pData + make_uint4( rank, rank, rank, rank );\n"
+	"}\n"
+	"//__attribute__((reqd_work_group_size(64,1,1)))\n"
+	"uint4 localPrefixSum64V( uint4 pData, uint lIdx, uint* totalSum, __local u32* sorterSharedMemory )\n"
+	"{\n"
+	"	u32 s4 = prefixScanVectorEx( &pData );\n"
+	"	u32 rank = localPrefixSum( s4, lIdx, totalSum, sorterSharedMemory, 64 );\n"
+	"	return pData + make_uint4( rank, rank, rank, rank );\n"
+	"}\n"
+	"u32 unpack4Key( u32 key, int keyIdx ){ return (key>>(keyIdx*8)) & 0xff;}\n"
+	"u32 bit8Scan(u32 v)\n"
+	"{\n"
+	"	return (v<<8) + (v<<16) + (v<<24);\n"
+	"}\n"
+	"//===\n"
+	"#define MY_HISTOGRAM(idx) localHistogramMat[(idx)*WG_SIZE+lIdx]\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"void StreamCountKernel( __global u32* gSrc, __global u32* histogramOut, int4 cb )\n"
+	"{\n"
+	"	__local u32 localHistogramMat[NUM_BUCKET*WG_SIZE];\n"
+	"	u32 gIdx = GET_GLOBAL_IDX;\n"
+	"	u32 lIdx = GET_LOCAL_IDX;\n"
+	"	u32 wgIdx = GET_GROUP_IDX;\n"
+	"	u32 wgSize = GET_GROUP_SIZE;\n"
+	"	const int startBit = cb.m_startBit;\n"
+	"	const int n = cb.m_n;\n"
+	"	const int nWGs = cb.m_nWGs;\n"
+	"	const int nBlocksPerWG = cb.m_nBlocksPerWG;\n"
+	"	for(int i=0; i<NUM_BUCKET; i++)\n"
+	"	{\n"
+	"		MY_HISTOGRAM(i) = 0;\n"
+	"	}\n"
+	"	GROUP_LDS_BARRIER;\n"
+	"	const int blockSize = ELEMENTS_PER_WORK_ITEM*WG_SIZE;\n"
+	"	u32 localKey;\n"
+	"	int nBlocks = (n)/blockSize - nBlocksPerWG*wgIdx;\n"
+	"	int addr = blockSize*nBlocksPerWG*wgIdx + ELEMENTS_PER_WORK_ITEM*lIdx;\n"
+	"	for(int iblock=0; iblock<min(nBlocksPerWG, nBlocks); iblock++, addr+=blockSize)\n"
+	"	{\n"
+	"		//	MY_HISTOGRAM( localKeys.x ) ++ is much expensive than atomic add as it requires read and write while atomics can just add on AMD\n"
+	"		//	Using registers didn't perform well. It seems like use localKeys to address requires a lot of alu ops\n"
+	"		//	AMD: AtomInc performs better while NV prefers ++\n"
+	"		for(int i=0; i<ELEMENTS_PER_WORK_ITEM; i++)\n"
+	"		{\n"
+	"#if defined(CHECK_BOUNDARY)\n"
+	"			if( addr+i < n )\n"
+	"#endif\n"
+	"			{\n"
+	"				localKey = (gSrc[addr+i]>>startBit) & 0xf;\n"
+	"#if defined(NV_GPU)\n"
+	"				MY_HISTOGRAM( localKey )++;\n"
+	"#else\n"
+	"				AtomInc( MY_HISTOGRAM( localKey ) );\n"
+	"#endif\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	GROUP_LDS_BARRIER;\n"
+	"	\n"
+	"	if( lIdx < NUM_BUCKET )\n"
+	"	{\n"
+	"		u32 sum = 0;\n"
+	"		for(int i=0; i<GET_GROUP_SIZE; i++)\n"
+	"		{\n"
+	"			sum += localHistogramMat[lIdx*WG_SIZE+(i+lIdx)%GET_GROUP_SIZE];\n"
+	"		}\n"
+	"		histogramOut[lIdx*nWGs+wgIdx] = sum;\n"
+	"	}\n"
+	"}\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"void StreamCountSortDataKernel( __global SortDataCL* gSrc, __global u32* histogramOut, int4  cb )\n"
+	"{\n"
+	"	__local u32 localHistogramMat[NUM_BUCKET*WG_SIZE];\n"
+	"	u32 gIdx = GET_GLOBAL_IDX;\n"
+	"	u32 lIdx = GET_LOCAL_IDX;\n"
+	"	u32 wgIdx = GET_GROUP_IDX;\n"
+	"	u32 wgSize = GET_GROUP_SIZE;\n"
+	"	const int startBit = cb.m_startBit;\n"
+	"	const int n = cb.m_n;\n"
+	"	const int nWGs = cb.m_nWGs;\n"
+	"	const int nBlocksPerWG = cb.m_nBlocksPerWG;\n"
+	"	for(int i=0; i<NUM_BUCKET; i++)\n"
+	"	{\n"
+	"		MY_HISTOGRAM(i) = 0;\n"
+	"	}\n"
+	"	GROUP_LDS_BARRIER;\n"
+	"	const int blockSize = ELEMENTS_PER_WORK_ITEM*WG_SIZE;\n"
+	"	u32 localKey;\n"
+	"	int nBlocks = (n)/blockSize - nBlocksPerWG*wgIdx;\n"
+	"	int addr = blockSize*nBlocksPerWG*wgIdx + ELEMENTS_PER_WORK_ITEM*lIdx;\n"
+	"	for(int iblock=0; iblock<min(nBlocksPerWG, nBlocks); iblock++, addr+=blockSize)\n"
+	"	{\n"
+	"		//	MY_HISTOGRAM( localKeys.x ) ++ is much expensive than atomic add as it requires read and write while atomics can just add on AMD\n"
+	"		//	Using registers didn't perform well. It seems like use localKeys to address requires a lot of alu ops\n"
+	"		//	AMD: AtomInc performs better while NV prefers ++\n"
+	"		for(int i=0; i<ELEMENTS_PER_WORK_ITEM; i++)\n"
+	"		{\n"
+	"#if defined(CHECK_BOUNDARY)\n"
+	"			if( addr+i < n )\n"
+	"#endif\n"
+	"			{\n"
+	"				localKey = (gSrc[addr+i].m_key>>startBit) & 0xf;\n"
+	"#if defined(NV_GPU)\n"
+	"				MY_HISTOGRAM( localKey )++;\n"
+	"#else\n"
+	"				AtomInc( MY_HISTOGRAM( localKey ) );\n"
+	"#endif\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	GROUP_LDS_BARRIER;\n"
+	"	\n"
+	"	if( lIdx < NUM_BUCKET )\n"
+	"	{\n"
+	"		u32 sum = 0;\n"
+	"		for(int i=0; i<GET_GROUP_SIZE; i++)\n"
+	"		{\n"
+	"			sum += localHistogramMat[lIdx*WG_SIZE+(i+lIdx)%GET_GROUP_SIZE];\n"
+	"		}\n"
+	"		histogramOut[lIdx*nWGs+wgIdx] = sum;\n"
+	"	}\n"
+	"}\n"
+	"#define nPerLane (nPerWI/4)\n"
+	"//	NUM_BUCKET*nWGs < 128*nPerWI\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(128,1,1)))\n"
+	"void PrefixScanKernel( __global u32* wHistogram1, int4  cb )\n"
+	"{\n"
+	"	__local u32 ldsTopScanData[128*2];\n"
+	"	u32 lIdx = GET_LOCAL_IDX;\n"
+	"	u32 wgIdx = GET_GROUP_IDX;\n"
+	"	const int nWGs = cb.m_nWGs;\n"
+	"	u32 data[nPerWI];\n"
+	"	for(int i=0; i<nPerWI; i++)\n"
+	"	{\n"
+	"		data[i] = 0;\n"
+	"		if( (nPerWI*lIdx+i) < NUM_BUCKET*nWGs )\n"
+	"			data[i] = wHistogram1[nPerWI*lIdx+i];\n"
+	"	}\n"
+	"	uint4 myData = make_uint4(0,0,0,0);\n"
+	"	for(int i=0; i<nPerLane; i++)\n"
+	"	{\n"
+	"		myData.x += data[nPerLane*0+i];\n"
+	"		myData.y += data[nPerLane*1+i];\n"
+	"		myData.z += data[nPerLane*2+i];\n"
+	"		myData.w += data[nPerLane*3+i];\n"
+	"	}\n"
+	"	uint totalSum;\n"
+	"	uint4 scanned = localPrefixSum128V( myData, lIdx, &totalSum, ldsTopScanData );\n"
+	"//	for(int j=0; j<4; j++) //	somehow it introduces a lot of branches\n"
+	"	{	int j = 0;\n"
+	"		u32 sum = 0;\n"
+	"		for(int i=0; i<nPerLane; i++)\n"
+	"		{\n"
+	"			u32 tmp = data[nPerLane*j+i];\n"
+	"			data[nPerLane*j+i] = sum;\n"
+	"			sum += tmp;\n"
+	"		}\n"
+	"	}\n"
+	"	{	int j = 1;\n"
+	"		u32 sum = 0;\n"
+	"		for(int i=0; i<nPerLane; i++)\n"
+	"		{\n"
+	"			u32 tmp = data[nPerLane*j+i];\n"
+	"			data[nPerLane*j+i] = sum;\n"
+	"			sum += tmp;\n"
+	"		}\n"
+	"	}\n"
+	"	{	int j = 2;\n"
+	"		u32 sum = 0;\n"
+	"		for(int i=0; i<nPerLane; i++)\n"
+	"		{\n"
+	"			u32 tmp = data[nPerLane*j+i];\n"
+	"			data[nPerLane*j+i] = sum;\n"
+	"			sum += tmp;\n"
+	"		}\n"
+	"	}\n"
+	"	{	int j = 3;\n"
+	"		u32 sum = 0;\n"
+	"		for(int i=0; i<nPerLane; i++)\n"
+	"		{\n"
+	"			u32 tmp = data[nPerLane*j+i];\n"
+	"			data[nPerLane*j+i] = sum;\n"
+	"			sum += tmp;\n"
+	"		}\n"
+	"	}\n"
+	"	for(int i=0; i<nPerLane; i++)\n"
+	"	{\n"
+	"		data[nPerLane*0+i] += scanned.x;\n"
+	"		data[nPerLane*1+i] += scanned.y;\n"
+	"		data[nPerLane*2+i] += scanned.z;\n"
+	"		data[nPerLane*3+i] += scanned.w;\n"
+	"	}\n"
+	"	for(int i=0; i<nPerWI; i++)\n"
+	"	{\n"
+	"		int index = nPerWI*lIdx+i;\n"
+	"		if (index < NUM_BUCKET*nWGs)\n"
+	"			wHistogram1[nPerWI*lIdx+i] = data[i];\n"
+	"	}\n"
+	"}\n"
+	"//	4 scan, 4 exchange\n"
+	"void sort4Bits(u32 sortData[4], int startBit, int lIdx, __local u32* ldsSortData)\n"
+	"{\n"
+	"	for(int bitIdx=0; bitIdx<BITS_PER_PASS; bitIdx++)\n"
+	"	{\n"
+	"		u32 mask = (1<<bitIdx);\n"
+	"		uint4 cmpResult = make_uint4( (sortData[0]>>startBit) & mask, (sortData[1]>>startBit) & mask, (sortData[2]>>startBit) & mask, (sortData[3]>>startBit) & mask );\n"
+	"		uint4 prefixSum = SELECT_UINT4( make_uint4(1,1,1,1), make_uint4(0,0,0,0), cmpResult != make_uint4(0,0,0,0) );\n"
+	"		u32 total;\n"
+	"		prefixSum = localPrefixSum64V( prefixSum, lIdx, &total, ldsSortData );\n"
+	"		{\n"
+	"			uint4 localAddr = make_uint4(lIdx*4+0,lIdx*4+1,lIdx*4+2,lIdx*4+3);\n"
+	"			uint4 dstAddr = localAddr - prefixSum + make_uint4( total, total, total, total );\n"
+	"			dstAddr = SELECT_UINT4( prefixSum, dstAddr, cmpResult != make_uint4(0, 0, 0, 0) );\n"
+	"			GROUP_LDS_BARRIER;\n"
+	"			ldsSortData[dstAddr.x] = sortData[0];\n"
+	"			ldsSortData[dstAddr.y] = sortData[1];\n"
+	"			ldsSortData[dstAddr.z] = sortData[2];\n"
+	"			ldsSortData[dstAddr.w] = sortData[3];\n"
+	"			GROUP_LDS_BARRIER;\n"
+	"			sortData[0] = ldsSortData[localAddr.x];\n"
+	"			sortData[1] = ldsSortData[localAddr.y];\n"
+	"			sortData[2] = ldsSortData[localAddr.z];\n"
+	"			sortData[3] = ldsSortData[localAddr.w];\n"
+	"			GROUP_LDS_BARRIER;\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"//	2 scan, 2 exchange\n"
+	"void sort4Bits1(u32 sortData[4], int startBit, int lIdx, __local u32* ldsSortData)\n"
+	"{\n"
+	"	for(uint ibit=0; ibit<BITS_PER_PASS; ibit+=2)\n"
+	"	{\n"
+	"		uint4 b = make_uint4((sortData[0]>>(startBit+ibit)) & 0x3, \n"
+	"			(sortData[1]>>(startBit+ibit)) & 0x3, \n"
+	"			(sortData[2]>>(startBit+ibit)) & 0x3, \n"
+	"			(sortData[3]>>(startBit+ibit)) & 0x3);\n"
+	"		u32 key4;\n"
+	"		u32 sKeyPacked[4] = { 0, 0, 0, 0 };\n"
+	"		{\n"
+	"			sKeyPacked[0] |= 1<<(8*b.x);\n"
+	"			sKeyPacked[1] |= 1<<(8*b.y);\n"
+	"			sKeyPacked[2] |= 1<<(8*b.z);\n"
+	"			sKeyPacked[3] |= 1<<(8*b.w);\n"
+	"			key4 = sKeyPacked[0] + sKeyPacked[1] + sKeyPacked[2] + sKeyPacked[3];\n"
+	"		}\n"
+	"		u32 rankPacked;\n"
+	"		u32 sumPacked;\n"
+	"		{\n"
+	"			rankPacked = localPrefixSum( key4, lIdx, &sumPacked, ldsSortData, WG_SIZE );\n"
+	"		}\n"
+	"		GROUP_LDS_BARRIER;\n"
+	"		u32 newOffset[4] = { 0,0,0,0 };\n"
+	"		{\n"
+	"			u32 sumScanned = bit8Scan( sumPacked );\n"
+	"			u32 scannedKeys[4];\n"
+	"			scannedKeys[0] = 1<<(8*b.x);\n"
+	"			scannedKeys[1] = 1<<(8*b.y);\n"
+	"			scannedKeys[2] = 1<<(8*b.z);\n"
+	"			scannedKeys[3] = 1<<(8*b.w);\n"
+	"			{	//	4 scans at once\n"
+	"				u32 sum4 = 0;\n"
+	"				for(int ie=0; ie<4; ie++)\n"
+	"				{\n"
+	"					u32 tmp = scannedKeys[ie];\n"
+	"					scannedKeys[ie] = sum4;\n"
+	"					sum4 += tmp;\n"
+	"				}\n"
+	"			}\n"
+	"			{\n"
+	"				u32 sumPlusRank = sumScanned + rankPacked;\n"
+	"				{	u32 ie = b.x;\n"
+	"					scannedKeys[0] += sumPlusRank;\n"
+	"					newOffset[0] = unpack4Key( scannedKeys[0], ie );\n"
+	"				}\n"
+	"				{	u32 ie = b.y;\n"
+	"					scannedKeys[1] += sumPlusRank;\n"
+	"					newOffset[1] = unpack4Key( scannedKeys[1], ie );\n"
+	"				}\n"
+	"				{	u32 ie = b.z;\n"
+	"					scannedKeys[2] += sumPlusRank;\n"
+	"					newOffset[2] = unpack4Key( scannedKeys[2], ie );\n"
+	"				}\n"
+	"				{	u32 ie = b.w;\n"
+	"					scannedKeys[3] += sumPlusRank;\n"
+	"					newOffset[3] = unpack4Key( scannedKeys[3], ie );\n"
+	"				}\n"
+	"			}\n"
+	"		}\n"
+	"		GROUP_LDS_BARRIER;\n"
+	"		{\n"
+	"			ldsSortData[newOffset[0]] = sortData[0];\n"
+	"			ldsSortData[newOffset[1]] = sortData[1];\n"
+	"			ldsSortData[newOffset[2]] = sortData[2];\n"
+	"			ldsSortData[newOffset[3]] = sortData[3];\n"
+	"			GROUP_LDS_BARRIER;\n"
+	"			u32 dstAddr = 4*lIdx;\n"
+	"			sortData[0] = ldsSortData[dstAddr+0];\n"
+	"			sortData[1] = ldsSortData[dstAddr+1];\n"
+	"			sortData[2] = ldsSortData[dstAddr+2];\n"
+	"			sortData[3] = ldsSortData[dstAddr+3];\n"
+	"			GROUP_LDS_BARRIER;\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"#define SET_HISTOGRAM(setIdx, key) ldsSortData[(setIdx)*NUM_BUCKET+key]\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"void SortAndScatterKernel( __global const u32* restrict gSrc, __global const u32* rHistogram, __global u32* restrict gDst, int4  cb )\n"
+	"{\n"
+	"	__local u32 ldsSortData[WG_SIZE*ELEMENTS_PER_WORK_ITEM+16];\n"
+	"	__local u32 localHistogramToCarry[NUM_BUCKET];\n"
+	"	__local u32 localHistogram[NUM_BUCKET*2];\n"
+	"	u32 gIdx = GET_GLOBAL_IDX;\n"
+	"	u32 lIdx = GET_LOCAL_IDX;\n"
+	"	u32 wgIdx = GET_GROUP_IDX;\n"
+	"	u32 wgSize = GET_GROUP_SIZE;\n"
+	"	const int n = cb.m_n;\n"
+	"	const int nWGs = cb.m_nWGs;\n"
+	"	const int startBit = cb.m_startBit;\n"
+	"	const int nBlocksPerWG = cb.m_nBlocksPerWG;\n"
+	"	if( lIdx < (NUM_BUCKET) )\n"
+	"	{\n"
+	"		localHistogramToCarry[lIdx] = rHistogram[lIdx*nWGs + wgIdx];\n"
+	"	}\n"
+	"	GROUP_LDS_BARRIER;\n"
+	"	const int blockSize = ELEMENTS_PER_WORK_ITEM*WG_SIZE;\n"
+	"	int nBlocks = n/blockSize - nBlocksPerWG*wgIdx;\n"
+	"	int addr = blockSize*nBlocksPerWG*wgIdx + ELEMENTS_PER_WORK_ITEM*lIdx;\n"
+	"	for(int iblock=0; iblock<min(nBlocksPerWG, nBlocks); iblock++, addr+=blockSize)\n"
+	"	{\n"
+	"		u32 myHistogram = 0;\n"
+	"		u32 sortData[ELEMENTS_PER_WORK_ITEM];\n"
+	"		for(int i=0; i<ELEMENTS_PER_WORK_ITEM; i++)\n"
+	"#if defined(CHECK_BOUNDARY)\n"
+	"			sortData[i] = ( addr+i < n )? gSrc[ addr+i ] : 0xffffffff;\n"
+	"#else\n"
+	"			sortData[i] = gSrc[ addr+i ];\n"
+	"#endif\n"
+	"		sort4Bits(sortData, startBit, lIdx, ldsSortData);\n"
+	"		u32 keys[ELEMENTS_PER_WORK_ITEM];\n"
+	"		for(int i=0; i<ELEMENTS_PER_WORK_ITEM; i++)\n"
+	"			keys[i] = (sortData[i]>>startBit) & 0xf;\n"
+	"		{	//	create histogram\n"
+	"			u32 setIdx = lIdx/16;\n"
+	"			if( lIdx < NUM_BUCKET )\n"
+	"			{\n"
+	"				localHistogram[lIdx] = 0;\n"
+	"			}\n"
+	"			ldsSortData[lIdx] = 0;\n"
+	"			GROUP_LDS_BARRIER;\n"
+	"			for(int i=0; i<ELEMENTS_PER_WORK_ITEM; i++)\n"
+	"#if defined(CHECK_BOUNDARY)\n"
+	"				if( addr+i < n )\n"
+	"#endif\n"
+	"#if defined(NV_GPU)\n"
+	"				SET_HISTOGRAM( setIdx, keys[i] )++;\n"
+	"#else\n"
+	"				AtomInc( SET_HISTOGRAM( setIdx, keys[i] ) );\n"
+	"#endif\n"
+	"			\n"
+	"			GROUP_LDS_BARRIER;\n"
+	"			\n"
+	"			uint hIdx = NUM_BUCKET+lIdx;\n"
+	"			if( lIdx < NUM_BUCKET )\n"
+	"			{\n"
+	"				u32 sum = 0;\n"
+	"				for(int i=0; i<WG_SIZE/16; i++)\n"
+	"				{\n"
+	"					sum += SET_HISTOGRAM( i, lIdx );\n"
+	"				}\n"
+	"				myHistogram = sum;\n"
+	"				localHistogram[hIdx] = sum;\n"
+	"			}\n"
+	"			GROUP_LDS_BARRIER;\n"
+	"#if defined(USE_2LEVEL_REDUCE)\n"
+	"			if( lIdx < NUM_BUCKET )\n"
+	"			{\n"
+	"				localHistogram[hIdx] = localHistogram[hIdx-1];\n"
+	"				GROUP_MEM_FENCE;\n"
+	"				u32 u0, u1, u2;\n"
+	"				u0 = localHistogram[hIdx-3];\n"
+	"				u1 = localHistogram[hIdx-2];\n"
+	"				u2 = localHistogram[hIdx-1];\n"
+	"				AtomAdd( localHistogram[hIdx], u0 + u1 + u2 );\n"
+	"				GROUP_MEM_FENCE;\n"
+	"				u0 = localHistogram[hIdx-12];\n"
+	"				u1 = localHistogram[hIdx-8];\n"
+	"				u2 = localHistogram[hIdx-4];\n"
+	"				AtomAdd( localHistogram[hIdx], u0 + u1 + u2 );\n"
+	"				GROUP_MEM_FENCE;\n"
+	"			}\n"
+	"#else\n"
+	"			if( lIdx < NUM_BUCKET )\n"
+	"			{\n"
+	"				localHistogram[hIdx] = localHistogram[hIdx-1];\n"
+	"				GROUP_MEM_FENCE;\n"
+	"				localHistogram[hIdx] += localHistogram[hIdx-1];\n"
+	"				GROUP_MEM_FENCE;\n"
+	"				localHistogram[hIdx] += localHistogram[hIdx-2];\n"
+	"				GROUP_MEM_FENCE;\n"
+	"				localHistogram[hIdx] += localHistogram[hIdx-4];\n"
+	"				GROUP_MEM_FENCE;\n"
+	"				localHistogram[hIdx] += localHistogram[hIdx-8];\n"
+	"				GROUP_MEM_FENCE;\n"
+	"			}\n"
+	"#endif\n"
+	"			GROUP_LDS_BARRIER;\n"
+	"		}\n"
+	"		{\n"
+	"			for(int ie=0; ie<ELEMENTS_PER_WORK_ITEM; ie++)\n"
+	"			{\n"
+	"				int dataIdx = ELEMENTS_PER_WORK_ITEM*lIdx+ie;\n"
+	"				int binIdx = keys[ie];\n"
+	"				int groupOffset = localHistogramToCarry[binIdx];\n"
+	"				int myIdx = dataIdx - localHistogram[NUM_BUCKET+binIdx];\n"
+	"#if defined(CHECK_BOUNDARY)\n"
+	"				if( addr+ie < n )\n"
+	"#endif\n"
+	"				gDst[ groupOffset + myIdx ] = sortData[ie];\n"
+	"			}\n"
+	"		}\n"
+	"		GROUP_LDS_BARRIER;\n"
+	"		if( lIdx < NUM_BUCKET )\n"
+	"		{\n"
+	"			localHistogramToCarry[lIdx] += myHistogram;\n"
+	"		}\n"
+	"		GROUP_LDS_BARRIER;\n"
+	"	}\n"
+	"}\n"
+	"//	2 scan, 2 exchange\n"
+	"void sort4Bits1KeyValue(u32 sortData[4], int sortVal[4], int startBit, int lIdx, __local u32* ldsSortData, __local int *ldsSortVal)\n"
+	"{\n"
+	"	for(uint ibit=0; ibit<BITS_PER_PASS; ibit+=2)\n"
+	"	{\n"
+	"		uint4 b = make_uint4((sortData[0]>>(startBit+ibit)) & 0x3, \n"
+	"			(sortData[1]>>(startBit+ibit)) & 0x3, \n"
+	"			(sortData[2]>>(startBit+ibit)) & 0x3, \n"
+	"			(sortData[3]>>(startBit+ibit)) & 0x3);\n"
+	"		u32 key4;\n"
+	"		u32 sKeyPacked[4] = { 0, 0, 0, 0 };\n"
+	"		{\n"
+	"			sKeyPacked[0] |= 1<<(8*b.x);\n"
+	"			sKeyPacked[1] |= 1<<(8*b.y);\n"
+	"			sKeyPacked[2] |= 1<<(8*b.z);\n"
+	"			sKeyPacked[3] |= 1<<(8*b.w);\n"
+	"			key4 = sKeyPacked[0] + sKeyPacked[1] + sKeyPacked[2] + sKeyPacked[3];\n"
+	"		}\n"
+	"		u32 rankPacked;\n"
+	"		u32 sumPacked;\n"
+	"		{\n"
+	"			rankPacked = localPrefixSum( key4, lIdx, &sumPacked, ldsSortData, WG_SIZE );\n"
+	"		}\n"
+	"		GROUP_LDS_BARRIER;\n"
+	"		u32 newOffset[4] = { 0,0,0,0 };\n"
+	"		{\n"
+	"			u32 sumScanned = bit8Scan( sumPacked );\n"
+	"			u32 scannedKeys[4];\n"
+	"			scannedKeys[0] = 1<<(8*b.x);\n"
+	"			scannedKeys[1] = 1<<(8*b.y);\n"
+	"			scannedKeys[2] = 1<<(8*b.z);\n"
+	"			scannedKeys[3] = 1<<(8*b.w);\n"
+	"			{	//	4 scans at once\n"
+	"				u32 sum4 = 0;\n"
+	"				for(int ie=0; ie<4; ie++)\n"
+	"				{\n"
+	"					u32 tmp = scannedKeys[ie];\n"
+	"					scannedKeys[ie] = sum4;\n"
+	"					sum4 += tmp;\n"
+	"				}\n"
+	"			}\n"
+	"			{\n"
+	"				u32 sumPlusRank = sumScanned + rankPacked;\n"
+	"				{	u32 ie = b.x;\n"
+	"					scannedKeys[0] += sumPlusRank;\n"
+	"					newOffset[0] = unpack4Key( scannedKeys[0], ie );\n"
+	"				}\n"
+	"				{	u32 ie = b.y;\n"
+	"					scannedKeys[1] += sumPlusRank;\n"
+	"					newOffset[1] = unpack4Key( scannedKeys[1], ie );\n"
+	"				}\n"
+	"				{	u32 ie = b.z;\n"
+	"					scannedKeys[2] += sumPlusRank;\n"
+	"					newOffset[2] = unpack4Key( scannedKeys[2], ie );\n"
+	"				}\n"
+	"				{	u32 ie = b.w;\n"
+	"					scannedKeys[3] += sumPlusRank;\n"
+	"					newOffset[3] = unpack4Key( scannedKeys[3], ie );\n"
+	"				}\n"
+	"			}\n"
+	"		}\n"
+	"		GROUP_LDS_BARRIER;\n"
+	"		{\n"
+	"			ldsSortData[newOffset[0]] = sortData[0];\n"
+	"			ldsSortData[newOffset[1]] = sortData[1];\n"
+	"			ldsSortData[newOffset[2]] = sortData[2];\n"
+	"			ldsSortData[newOffset[3]] = sortData[3];\n"
+	"			ldsSortVal[newOffset[0]] = sortVal[0];\n"
+	"			ldsSortVal[newOffset[1]] = sortVal[1];\n"
+	"			ldsSortVal[newOffset[2]] = sortVal[2];\n"
+	"			ldsSortVal[newOffset[3]] = sortVal[3];\n"
+	"			GROUP_LDS_BARRIER;\n"
+	"			u32 dstAddr = 4*lIdx;\n"
+	"			sortData[0] = ldsSortData[dstAddr+0];\n"
+	"			sortData[1] = ldsSortData[dstAddr+1];\n"
+	"			sortData[2] = ldsSortData[dstAddr+2];\n"
+	"			sortData[3] = ldsSortData[dstAddr+3];\n"
+	"			sortVal[0] = ldsSortVal[dstAddr+0];\n"
+	"			sortVal[1] = ldsSortVal[dstAddr+1];\n"
+	"			sortVal[2] = ldsSortVal[dstAddr+2];\n"
+	"			sortVal[3] = ldsSortVal[dstAddr+3];\n"
+	"			GROUP_LDS_BARRIER;\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"void SortAndScatterSortDataKernel( __global const SortDataCL* restrict gSrc, __global const u32* rHistogram, __global SortDataCL* restrict gDst, int4 cb)\n"
+	"{\n"
+	"	__local int ldsSortData[WG_SIZE*ELEMENTS_PER_WORK_ITEM+16];\n"
+	"	__local int ldsSortVal[WG_SIZE*ELEMENTS_PER_WORK_ITEM+16];\n"
+	"	__local u32 localHistogramToCarry[NUM_BUCKET];\n"
+	"	__local u32 localHistogram[NUM_BUCKET*2];\n"
+	"	u32 gIdx = GET_GLOBAL_IDX;\n"
+	"	u32 lIdx = GET_LOCAL_IDX;\n"
+	"	u32 wgIdx = GET_GROUP_IDX;\n"
+	"	u32 wgSize = GET_GROUP_SIZE;\n"
+	"	const int n = cb.m_n;\n"
+	"	const int nWGs = cb.m_nWGs;\n"
+	"	const int startBit = cb.m_startBit;\n"
+	"	const int nBlocksPerWG = cb.m_nBlocksPerWG;\n"
+	"	if( lIdx < (NUM_BUCKET) )\n"
+	"	{\n"
+	"		localHistogramToCarry[lIdx] = rHistogram[lIdx*nWGs + wgIdx];\n"
+	"	}\n"
+	"	GROUP_LDS_BARRIER;\n"
+	"    \n"
+	"	const int blockSize = ELEMENTS_PER_WORK_ITEM*WG_SIZE;\n"
+	"	int nBlocks = n/blockSize - nBlocksPerWG*wgIdx;\n"
+	"	int addr = blockSize*nBlocksPerWG*wgIdx + ELEMENTS_PER_WORK_ITEM*lIdx;\n"
+	"	for(int iblock=0; iblock<min(nBlocksPerWG, nBlocks); iblock++, addr+=blockSize)\n"
+	"	{\n"
+	"		u32 myHistogram = 0;\n"
+	"		int sortData[ELEMENTS_PER_WORK_ITEM];\n"
+	"		int sortVal[ELEMENTS_PER_WORK_ITEM];\n"
+	"		for(int i=0; i<ELEMENTS_PER_WORK_ITEM; i++)\n"
+	"#if defined(CHECK_BOUNDARY)\n"
+	"		{\n"
+	"			sortData[i] = ( addr+i < n )? gSrc[ addr+i ].m_key : 0xffffffff;\n"
+	"			sortVal[i] = ( addr+i < n )? gSrc[ addr+i ].m_value : 0xffffffff;\n"
+	"		}\n"
+	"#else\n"
+	"		{\n"
+	"			sortData[i] = gSrc[ addr+i ].m_key;\n"
+	"			sortVal[i] = gSrc[ addr+i ].m_value;\n"
+	"		}\n"
+	"#endif\n"
+	"		sort4Bits1KeyValue(sortData, sortVal, startBit, lIdx, ldsSortData, ldsSortVal);\n"
+	"		u32 keys[ELEMENTS_PER_WORK_ITEM];\n"
+	"		for(int i=0; i<ELEMENTS_PER_WORK_ITEM; i++)\n"
+	"			keys[i] = (sortData[i]>>startBit) & 0xf;\n"
+	"		{	//	create histogram\n"
+	"			u32 setIdx = lIdx/16;\n"
+	"			if( lIdx < NUM_BUCKET )\n"
+	"			{\n"
+	"				localHistogram[lIdx] = 0;\n"
+	"			}\n"
+	"			ldsSortData[lIdx] = 0;\n"
+	"			GROUP_LDS_BARRIER;\n"
+	"			for(int i=0; i<ELEMENTS_PER_WORK_ITEM; i++)\n"
+	"#if defined(CHECK_BOUNDARY)\n"
+	"				if( addr+i < n )\n"
+	"#endif\n"
+	"#if defined(NV_GPU)\n"
+	"				SET_HISTOGRAM( setIdx, keys[i] )++;\n"
+	"#else\n"
+	"				AtomInc( SET_HISTOGRAM( setIdx, keys[i] ) );\n"
+	"#endif\n"
+	"			\n"
+	"			GROUP_LDS_BARRIER;\n"
+	"			\n"
+	"			uint hIdx = NUM_BUCKET+lIdx;\n"
+	"			if( lIdx < NUM_BUCKET )\n"
+	"			{\n"
+	"				u32 sum = 0;\n"
+	"				for(int i=0; i<WG_SIZE/16; i++)\n"
+	"				{\n"
+	"					sum += SET_HISTOGRAM( i, lIdx );\n"
+	"				}\n"
+	"				myHistogram = sum;\n"
+	"				localHistogram[hIdx] = sum;\n"
+	"			}\n"
+	"			GROUP_LDS_BARRIER;\n"
+	"#if defined(USE_2LEVEL_REDUCE)\n"
+	"			if( lIdx < NUM_BUCKET )\n"
+	"			{\n"
+	"				localHistogram[hIdx] = localHistogram[hIdx-1];\n"
+	"				GROUP_MEM_FENCE;\n"
+	"				u32 u0, u1, u2;\n"
+	"				u0 = localHistogram[hIdx-3];\n"
+	"				u1 = localHistogram[hIdx-2];\n"
+	"				u2 = localHistogram[hIdx-1];\n"
+	"				AtomAdd( localHistogram[hIdx], u0 + u1 + u2 );\n"
+	"				GROUP_MEM_FENCE;\n"
+	"				u0 = localHistogram[hIdx-12];\n"
+	"				u1 = localHistogram[hIdx-8];\n"
+	"				u2 = localHistogram[hIdx-4];\n"
+	"				AtomAdd( localHistogram[hIdx], u0 + u1 + u2 );\n"
+	"				GROUP_MEM_FENCE;\n"
+	"			}\n"
+	"#else\n"
+	"			if( lIdx < NUM_BUCKET )\n"
+	"			{\n"
+	"				localHistogram[hIdx] = localHistogram[hIdx-1];\n"
+	"				GROUP_MEM_FENCE;\n"
+	"				localHistogram[hIdx] += localHistogram[hIdx-1];\n"
+	"				GROUP_MEM_FENCE;\n"
+	"				localHistogram[hIdx] += localHistogram[hIdx-2];\n"
+	"				GROUP_MEM_FENCE;\n"
+	"				localHistogram[hIdx] += localHistogram[hIdx-4];\n"
+	"				GROUP_MEM_FENCE;\n"
+	"				localHistogram[hIdx] += localHistogram[hIdx-8];\n"
+	"				GROUP_MEM_FENCE;\n"
+	"			}\n"
+	"#endif\n"
+	"			GROUP_LDS_BARRIER;\n"
+	"		}\n"
+	"    	{\n"
+	"			for(int ie=0; ie<ELEMENTS_PER_WORK_ITEM; ie++)\n"
+	"			{\n"
+	"				int dataIdx = ELEMENTS_PER_WORK_ITEM*lIdx+ie;\n"
+	"				int binIdx = keys[ie];\n"
+	"				int groupOffset = localHistogramToCarry[binIdx];\n"
+	"				int myIdx = dataIdx - localHistogram[NUM_BUCKET+binIdx];\n"
+	"#if defined(CHECK_BOUNDARY)\n"
+	"				if( addr+ie < n )\n"
+	"				{\n"
+	"                    if ((groupOffset + myIdx)<n)\n"
+	"                    {\n"
+	"                        if (sortData[ie]==sortVal[ie])\n"
+	"                        {\n"
+	"                            \n"
+	"                            SortDataCL tmp;\n"
+	"                            tmp.m_key = sortData[ie];\n"
+	"                            tmp.m_value = sortVal[ie];\n"
+	"                            if (tmp.m_key == tmp.m_value)\n"
+	"                                gDst[groupOffset + myIdx ] = tmp;\n"
+	"                        }\n"
+	"                        \n"
+	"                    }\n"
+	"				}\n"
+	"#else\n"
+	"                if ((groupOffset + myIdx)<n)\n"
+	"                {\n"
+	"                    gDst[ groupOffset + myIdx ].m_key = sortData[ie];\n"
+	"                    gDst[ groupOffset + myIdx ].m_value = sortVal[ie];\n"
+	"                }\n"
+	"#endif\n"
+	"			}\n"
+	"		}\n"
+	"		GROUP_LDS_BARRIER;\n"
+	"		if( lIdx < NUM_BUCKET )\n"
+	"		{\n"
+	"			localHistogramToCarry[lIdx] += myHistogram;\n"
+	"		}\n"
+	"		GROUP_LDS_BARRIER;\n"
+	"	}\n"
+	"}\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"void SortAndScatterSortDataKernelSerial( __global const SortDataCL* restrict gSrc, __global const u32* rHistogram, __global SortDataCL* restrict gDst, int4 cb)\n"
+	"{\n"
+	"    \n"
+	"	u32 gIdx = GET_GLOBAL_IDX;\n"
+	"	u32 realLocalIdx = GET_LOCAL_IDX;\n"
+	"	u32 wgIdx = GET_GROUP_IDX;\n"
+	"	u32 wgSize = GET_GROUP_SIZE;\n"
+	"	const int startBit = cb.m_startBit;\n"
+	"	const int n = cb.m_n;\n"
+	"	const int nWGs = cb.m_nWGs;\n"
+	"	const int nBlocksPerWG = cb.m_nBlocksPerWG;\n"
+	"    int counter[NUM_BUCKET];\n"
+	"    \n"
+	"    if (realLocalIdx>0)\n"
+	"        return;\n"
+	"    \n"
+	"    for (int c=0;c<NUM_BUCKET;c++)\n"
+	"        counter[c]=0;\n"
+	"    const int blockSize = ELEMENTS_PER_WORK_ITEM*WG_SIZE;\n"
+	"	\n"
+	"	int nBlocks = (n)/blockSize - nBlocksPerWG*wgIdx;\n"
+	"   for(int iblock=0; iblock<min(nBlocksPerWG, nBlocks); iblock++)\n"
+	"  {\n"
+	"     for (int lIdx=0;lIdx<WG_SIZE;lIdx++)\n"
+	" 	{\n"
+	"        int addr2 = iblock*blockSize + blockSize*nBlocksPerWG*wgIdx + ELEMENTS_PER_WORK_ITEM*lIdx;\n"
+	"        \n"
+	"		for(int j=0; j<ELEMENTS_PER_WORK_ITEM; j++)\n"
+	"		{\n"
+	"            int i = addr2+j;\n"
+	"			if( i < n )\n"
+	"			{\n"
+	"                int tableIdx;\n"
+	"				tableIdx = (gSrc[i].m_key>>startBit) & 0xf;//0xf = NUM_TABLES-1\n"
+	"                gDst[rHistogram[tableIdx*nWGs+wgIdx] + counter[tableIdx]] = gSrc[i];\n"
+	"                counter[tableIdx] ++;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"  }\n"
+	"    \n"
+	"}\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"void SortAndScatterKernelSerial( __global const u32* restrict gSrc, __global const u32* rHistogram, __global u32* restrict gDst, int4  cb )\n"
+	"{\n"
+	"    \n"
+	"	u32 gIdx = GET_GLOBAL_IDX;\n"
+	"	u32 realLocalIdx = GET_LOCAL_IDX;\n"
+	"	u32 wgIdx = GET_GROUP_IDX;\n"
+	"	u32 wgSize = GET_GROUP_SIZE;\n"
+	"	const int startBit = cb.m_startBit;\n"
+	"	const int n = cb.m_n;\n"
+	"	const int nWGs = cb.m_nWGs;\n"
+	"	const int nBlocksPerWG = cb.m_nBlocksPerWG;\n"
+	"    int counter[NUM_BUCKET];\n"
+	"    \n"
+	"    if (realLocalIdx>0)\n"
+	"        return;\n"
+	"    \n"
+	"    for (int c=0;c<NUM_BUCKET;c++)\n"
+	"        counter[c]=0;\n"
+	"    const int blockSize = ELEMENTS_PER_WORK_ITEM*WG_SIZE;\n"
+	"	\n"
+	"	int nBlocks = (n)/blockSize - nBlocksPerWG*wgIdx;\n"
+	"   for(int iblock=0; iblock<min(nBlocksPerWG, nBlocks); iblock++)\n"
+	"  {\n"
+	"     for (int lIdx=0;lIdx<WG_SIZE;lIdx++)\n"
+	" 	{\n"
+	"        int addr2 = iblock*blockSize + blockSize*nBlocksPerWG*wgIdx + ELEMENTS_PER_WORK_ITEM*lIdx;\n"
+	"        \n"
+	"		for(int j=0; j<ELEMENTS_PER_WORK_ITEM; j++)\n"
+	"		{\n"
+	"            int i = addr2+j;\n"
+	"			if( i < n )\n"
+	"			{\n"
+	"                int tableIdx;\n"
+	"				tableIdx = (gSrc[i]>>startBit) & 0xf;//0xf = NUM_TABLES-1\n"
+	"                gDst[rHistogram[tableIdx*nWGs+wgIdx] + counter[tableIdx]] = gSrc[i];\n"
+	"                counter[tableIdx] ++;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"  }\n"
+	"    \n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/Raycast/b3GpuRaycast.cpp b/thirdparty/bullet/Bullet3OpenCL/Raycast/b3GpuRaycast.cpp
index 161e304f090..6571f305482 100644
--- a/thirdparty/bullet/Bullet3OpenCL/Raycast/b3GpuRaycast.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/Raycast/b3GpuRaycast.cpp
@@ -4,7 +4,6 @@
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3RigidBodyData.h"
 #include "Bullet3OpenCL/RigidBody/b3GpuNarrowPhaseInternalData.h"
 
-
 #include "Bullet3OpenCL/Initialize/b3OpenCLUtils.h"
 #include "Bullet3OpenCL/ParallelPrimitives/b3OpenCLArray.h"
 #include "Bullet3OpenCL/ParallelPrimitives/b3LauncherCL.h"
@@ -15,38 +14,35 @@
 
 #include "Bullet3OpenCL/Raycast/kernels/rayCastKernels.h"
 
-
 #define B3_RAYCAST_PATH "src/Bullet3OpenCL/Raycast/kernels/rayCastKernels.cl"
 
-
-
 struct b3GpuRaycastInternalData
 {
 	cl_context m_context;
 	cl_device_id m_device;
-	cl_command_queue  m_q;
+	cl_command_queue m_q;
 	cl_kernel m_raytraceKernel;
 	cl_kernel m_raytracePairsKernel;
 	cl_kernel m_findRayRigidPairIndexRanges;
-	
+
 	b3GpuParallelLinearBvh* m_plbvh;
 	b3RadixSort32CL* m_radixSorter;
 	b3FillCL* m_fill;
-	
+
 	//1 element per ray
 	b3OpenCLArray<b3RayInfo>* m_gpuRays;
 	b3OpenCLArray<b3RayHit>* m_gpuHitResults;
 	b3OpenCLArray<int>* m_firstRayRigidPairIndexPerRay;
 	b3OpenCLArray<int>* m_numRayRigidPairsPerRay;
-	
+
 	//1 element per (ray index, rigid index) pair, where the ray intersects with the rigid's AABB
 	b3OpenCLArray<int>* m_gpuNumRayRigidPairs;
-	b3OpenCLArray<b3Int2>* m_gpuRayRigidPairs;	//x == ray index, y == rigid index
-	
+	b3OpenCLArray<b3Int2>* m_gpuRayRigidPairs;  //x == ray index, y == rigid index
+
 	int m_test;
 };
 
-b3GpuRaycast::b3GpuRaycast(cl_context ctx,cl_device_id device, cl_command_queue  q)
+b3GpuRaycast::b3GpuRaycast(cl_context ctx, cl_device_id device, cl_command_queue q)
 {
 	m_data = new b3GpuRaycastInternalData;
 	m_data->m_context = ctx;
@@ -59,7 +55,7 @@ b3GpuRaycast::b3GpuRaycast(cl_context ctx,cl_device_id device, cl_command_queue
 	m_data->m_plbvh = new b3GpuParallelLinearBvh(ctx, device, q);
 	m_data->m_radixSorter = new b3RadixSort32CL(ctx, device, q);
 	m_data->m_fill = new b3FillCL(ctx, device, q);
-	
+
 	m_data->m_gpuRays = new b3OpenCLArray<b3RayInfo>(ctx, q);
 	m_data->m_gpuHitResults = new b3OpenCLArray<b3RayHit>(ctx, q);
 	m_data->m_firstRayRigidPairIndexPerRay = new b3OpenCLArray<int>(ctx, q);
@@ -68,19 +64,17 @@ b3GpuRaycast::b3GpuRaycast(cl_context ctx,cl_device_id device, cl_command_queue
 	m_data->m_gpuRayRigidPairs = new b3OpenCLArray<b3Int2>(ctx, q);
 
 	{
-		cl_int errNum=0;
-		cl_program prog = b3OpenCLUtils::compileCLProgramFromString(m_data->m_context,m_data->m_device,rayCastKernelCL,&errNum,"",B3_RAYCAST_PATH);
-		b3Assert(errNum==CL_SUCCESS);
-		m_data->m_raytraceKernel = b3OpenCLUtils::compileCLKernelFromString(m_data->m_context, m_data->m_device,rayCastKernelCL, "rayCastKernel",&errNum,prog);
-		b3Assert(errNum==CL_SUCCESS);
-		m_data->m_raytracePairsKernel = b3OpenCLUtils::compileCLKernelFromString(m_data->m_context, m_data->m_device,rayCastKernelCL, "rayCastPairsKernel",&errNum,prog);
-		b3Assert(errNum==CL_SUCCESS);
-		m_data->m_findRayRigidPairIndexRanges = b3OpenCLUtils::compileCLKernelFromString(m_data->m_context, m_data->m_device,rayCastKernelCL, "findRayRigidPairIndexRanges",&errNum,prog);
-		b3Assert(errNum==CL_SUCCESS);
+		cl_int errNum = 0;
+		cl_program prog = b3OpenCLUtils::compileCLProgramFromString(m_data->m_context, m_data->m_device, rayCastKernelCL, &errNum, "", B3_RAYCAST_PATH);
+		b3Assert(errNum == CL_SUCCESS);
+		m_data->m_raytraceKernel = b3OpenCLUtils::compileCLKernelFromString(m_data->m_context, m_data->m_device, rayCastKernelCL, "rayCastKernel", &errNum, prog);
+		b3Assert(errNum == CL_SUCCESS);
+		m_data->m_raytracePairsKernel = b3OpenCLUtils::compileCLKernelFromString(m_data->m_context, m_data->m_device, rayCastKernelCL, "rayCastPairsKernel", &errNum, prog);
+		b3Assert(errNum == CL_SUCCESS);
+		m_data->m_findRayRigidPairIndexRanges = b3OpenCLUtils::compileCLKernelFromString(m_data->m_context, m_data->m_device, rayCastKernelCL, "findRayRigidPairIndexRanges", &errNum, prog);
+		b3Assert(errNum == CL_SUCCESS);
 		clReleaseProgram(prog);
 	}
-
-
 }
 
 b3GpuRaycast::~b3GpuRaycast()
@@ -88,78 +82,80 @@ b3GpuRaycast::~b3GpuRaycast()
 	clReleaseKernel(m_data->m_raytraceKernel);
 	clReleaseKernel(m_data->m_raytracePairsKernel);
 	clReleaseKernel(m_data->m_findRayRigidPairIndexRanges);
-	
+
 	delete m_data->m_plbvh;
 	delete m_data->m_radixSorter;
 	delete m_data->m_fill;
-	
+
 	delete m_data->m_gpuRays;
 	delete m_data->m_gpuHitResults;
 	delete m_data->m_firstRayRigidPairIndexPerRay;
 	delete m_data->m_numRayRigidPairsPerRay;
 	delete m_data->m_gpuNumRayRigidPairs;
 	delete m_data->m_gpuRayRigidPairs;
-	
+
 	delete m_data;
 }
 
-bool sphere_intersect(const b3Vector3& spherePos,  b3Scalar radius, const b3Vector3& rayFrom, const b3Vector3& rayTo, float& hitFraction)
+bool sphere_intersect(const b3Vector3& spherePos, b3Scalar radius, const b3Vector3& rayFrom, const b3Vector3& rayTo, float& hitFraction)
 {
-    b3Vector3 rs = rayFrom - spherePos;
-	b3Vector3 rayDir = rayTo-rayFrom;
-	
-	float A = b3Dot(rayDir,rayDir);
-    float B = b3Dot(rs, rayDir);
-    float C = b3Dot(rs, rs) - (radius * radius);
-    
-	float D = B * B - A*C;
+	b3Vector3 rs = rayFrom - spherePos;
+	b3Vector3 rayDir = rayTo - rayFrom;
 
-    if (D > 0.0)
-    {
-        float t = (-B - sqrt(D))/A;
+	float A = b3Dot(rayDir, rayDir);
+	float B = b3Dot(rs, rayDir);
+	float C = b3Dot(rs, rs) - (radius * radius);
 
-        if ( (t >= 0.0f) && (t < hitFraction) )
-        {
+	float D = B * B - A * C;
+
+	if (D > 0.0)
+	{
+		float t = (-B - sqrt(D)) / A;
+
+		if ((t >= 0.0f) && (t < hitFraction))
+		{
 			hitFraction = t;
-            return true;
+			return true;
 		}
 	}
 	return false;
 }
 
 bool rayConvex(const b3Vector3& rayFromLocal, const b3Vector3& rayToLocal, const b3ConvexPolyhedronData& poly,
-	const b3AlignedObjectArray<b3GpuFace>& faces,  float& hitFraction, b3Vector3& hitNormal)
+			   const b3AlignedObjectArray<b3GpuFace>& faces, float& hitFraction, b3Vector3& hitNormal)
 {
 	float exitFraction = hitFraction;
 	float enterFraction = -0.1f;
-	b3Vector3 curHitNormal=b3MakeVector3(0,0,0);
-	for (int i=0;i<poly.m_numFaces;i++)
+	b3Vector3 curHitNormal = b3MakeVector3(0, 0, 0);
+	for (int i = 0; i < poly.m_numFaces; i++)
 	{
-		const b3GpuFace& face = faces[poly.m_faceOffset+i];
-		float fromPlaneDist = b3Dot(rayFromLocal,face.m_plane)+face.m_plane.w;
-		float toPlaneDist = b3Dot(rayToLocal,face.m_plane)+face.m_plane.w;
-		if (fromPlaneDist<0.f)
+		const b3GpuFace& face = faces[poly.m_faceOffset + i];
+		float fromPlaneDist = b3Dot(rayFromLocal, face.m_plane) + face.m_plane.w;
+		float toPlaneDist = b3Dot(rayToLocal, face.m_plane) + face.m_plane.w;
+		if (fromPlaneDist < 0.f)
 		{
 			if (toPlaneDist >= 0.f)
 			{
-				float fraction = fromPlaneDist / (fromPlaneDist-toPlaneDist);
-				if (exitFraction>fraction)
+				float fraction = fromPlaneDist / (fromPlaneDist - toPlaneDist);
+				if (exitFraction > fraction)
 				{
 					exitFraction = fraction;
 				}
-			} 			
-		} else
+			}
+		}
+		else
 		{
-			if (toPlaneDist<0.f)
+			if (toPlaneDist < 0.f)
 			{
-				float fraction = fromPlaneDist / (fromPlaneDist-toPlaneDist);
+				float fraction = fromPlaneDist / (fromPlaneDist - toPlaneDist);
 				if (enterFraction <= fraction)
 				{
 					enterFraction = fraction;
 					curHitNormal = face.m_plane;
 					curHitNormal.w = 0.f;
 				}
-			} else
+			}
+			else
 			{
 				return false;
 			}
@@ -176,44 +172,41 @@ bool rayConvex(const b3Vector3& rayFromLocal, const b3Vector3& rayToLocal, const
 	return true;
 }
 
-void b3GpuRaycast::castRaysHost(const b3AlignedObjectArray<b3RayInfo>& rays,	b3AlignedObjectArray<b3RayHit>& hitResults,
-		int numBodies,const struct b3RigidBodyData* bodies, int numCollidables,const struct b3Collidable* collidables, const struct b3GpuNarrowPhaseInternalData* narrowphaseData)
+void b3GpuRaycast::castRaysHost(const b3AlignedObjectArray<b3RayInfo>& rays, b3AlignedObjectArray<b3RayHit>& hitResults,
+								int numBodies, const struct b3RigidBodyData* bodies, int numCollidables, const struct b3Collidable* collidables, const struct b3GpuNarrowPhaseInternalData* narrowphaseData)
 {
-
-//	return castRays(rays,hitResults,numBodies,bodies,numCollidables,collidables);
+	//	return castRays(rays,hitResults,numBodies,bodies,numCollidables,collidables);
 
 	B3_PROFILE("castRaysHost");
-	for (int r=0;r<rays.size();r++)
+	for (int r = 0; r < rays.size(); r++)
 	{
 		b3Vector3 rayFrom = rays[r].m_from;
 		b3Vector3 rayTo = rays[r].m_to;
 		float hitFraction = hitResults[r].m_hitFraction;
 
-		int hitBodyIndex= -1;
+		int hitBodyIndex = -1;
 		b3Vector3 hitNormal;
 
-		for (int b=0;b<numBodies;b++)
+		for (int b = 0; b < numBodies; b++)
 		{
-				
 			const b3Vector3& pos = bodies[b].m_pos;
 			//const b3Quaternion& orn = bodies[b].m_quat;
-			
+
 			switch (collidables[bodies[b].m_collidableIdx].m_shapeType)
 			{
-			case SHAPE_SPHERE:
+				case SHAPE_SPHERE:
 				{
 					b3Scalar radius = collidables[bodies[b].m_collidableIdx].m_radius;
-					if (sphere_intersect(pos,  radius, rayFrom, rayTo,hitFraction))
+					if (sphere_intersect(pos, radius, rayFrom, rayTo, hitFraction))
 					{
 						hitBodyIndex = b;
 						b3Vector3 hitPoint;
-						hitPoint.setInterpolate3(rays[r].m_from, rays[r].m_to,hitFraction);
-						hitNormal = (hitPoint-bodies[b].m_pos).normalize();
+						hitPoint.setInterpolate3(rays[r].m_from, rays[r].m_to, hitFraction);
+						hitNormal = (hitPoint - bodies[b].m_pos).normalize();
 					}
 				}
-			case SHAPE_CONVEX_HULL:
+				case SHAPE_CONVEX_HULL:
 				{
-
 					b3Transform convexWorldTransform;
 					convexWorldTransform.setIdentity();
 					convexWorldTransform.setOrigin(bodies[b].m_pos);
@@ -222,72 +215,67 @@ void b3GpuRaycast::castRaysHost(const b3AlignedObjectArray<b3RayInfo>& rays,	b3A
 
 					b3Vector3 rayFromLocal = convexWorld2Local(rayFrom);
 					b3Vector3 rayToLocal = convexWorld2Local(rayTo);
-					
-					
+
 					int shapeIndex = collidables[bodies[b].m_collidableIdx].m_shapeIndex;
 					const b3ConvexPolyhedronData& poly = narrowphaseData->m_convexPolyhedra[shapeIndex];
-					if (rayConvex(rayFromLocal, rayToLocal,poly,narrowphaseData->m_convexFaces, hitFraction, hitNormal))
+					if (rayConvex(rayFromLocal, rayToLocal, poly, narrowphaseData->m_convexFaces, hitFraction, hitNormal))
 					{
 						hitBodyIndex = b;
 					}
 
-					
 					break;
 				}
-			default:
+				default:
 				{
-					static bool once=true;
+					static bool once = true;
 					if (once)
 					{
-						once=false;
+						once = false;
 						b3Warning("Raytest: unsupported shape type\n");
 					}
 				}
 			}
 		}
-		if (hitBodyIndex>=0)
+		if (hitBodyIndex >= 0)
 		{
-
 			hitResults[r].m_hitFraction = hitFraction;
-			hitResults[r].m_hitPoint.setInterpolate3(rays[r].m_from, rays[r].m_to,hitFraction);
+			hitResults[r].m_hitPoint.setInterpolate3(rays[r].m_from, rays[r].m_to, hitFraction);
 			hitResults[r].m_hitNormal = hitNormal;
 			hitResults[r].m_hitBody = hitBodyIndex;
 		}
-
 	}
 }
 ///todo: add some acceleration structure (AABBs, tree etc)
-void b3GpuRaycast::castRays(const b3AlignedObjectArray<b3RayInfo>& rays,	b3AlignedObjectArray<b3RayHit>& hitResults,
-		int numBodies,const struct b3RigidBodyData* bodies, int numCollidables, const struct b3Collidable* collidables, 
-		const struct b3GpuNarrowPhaseInternalData* narrowphaseData,	class b3GpuBroadphaseInterface* broadphase)
+void b3GpuRaycast::castRays(const b3AlignedObjectArray<b3RayInfo>& rays, b3AlignedObjectArray<b3RayHit>& hitResults,
+							int numBodies, const struct b3RigidBodyData* bodies, int numCollidables, const struct b3Collidable* collidables,
+							const struct b3GpuNarrowPhaseInternalData* narrowphaseData, class b3GpuBroadphaseInterface* broadphase)
 {
 	//castRaysHost(rays,hitResults,numBodies,bodies,numCollidables,collidables,narrowphaseData);
 
 	B3_PROFILE("castRaysGPU");
-	
+
 	{
 		B3_PROFILE("raycast copyFromHost");
 		m_data->m_gpuRays->copyFromHost(rays);
 		m_data->m_gpuHitResults->copyFromHost(hitResults);
-		
 	}
-	
+
 	int numRays = hitResults.size();
 	{
 		m_data->m_firstRayRigidPairIndexPerRay->resize(numRays);
 		m_data->m_numRayRigidPairsPerRay->resize(numRays);
-		
+
 		m_data->m_gpuNumRayRigidPairs->resize(1);
 		m_data->m_gpuRayRigidPairs->resize(numRays * 16);
 	}
-	
+
 	//run kernel
 	const bool USE_BRUTE_FORCE_RAYCAST = false;
-	if(USE_BRUTE_FORCE_RAYCAST)
+	if (USE_BRUTE_FORCE_RAYCAST)
 	{
 		B3_PROFILE("raycast launch1D");
 
-		b3LauncherCL launcher(m_data->m_q,m_data->m_raytraceKernel,"m_raytraceKernel");
+		b3LauncherCL launcher(m_data->m_q, m_data->m_raytraceKernel, "m_raytraceKernel");
 		int numRays = rays.size();
 		launcher.setConst(numRays);
 
@@ -299,93 +287,88 @@ void b3GpuRaycast::castRays(const b3AlignedObjectArray<b3RayInfo>& rays,	b3Align
 		launcher.setBuffer(narrowphaseData->m_collidablesGPU->getBufferCL());
 		launcher.setBuffer(narrowphaseData->m_convexFacesGPU->getBufferCL());
 		launcher.setBuffer(narrowphaseData->m_convexPolyhedraGPU->getBufferCL());
-		
+
 		launcher.launch1D(numRays);
 		clFinish(m_data->m_q);
 	}
 	else
 	{
-		m_data->m_plbvh->build( broadphase->getAllAabbsGPU(), broadphase->getSmallAabbIndicesGPU(), broadphase->getLargeAabbIndicesGPU() );
+		m_data->m_plbvh->build(broadphase->getAllAabbsGPU(), broadphase->getSmallAabbIndicesGPU(), broadphase->getLargeAabbIndicesGPU());
 
 		m_data->m_plbvh->testRaysAgainstBvhAabbs(*m_data->m_gpuRays, *m_data->m_gpuNumRayRigidPairs, *m_data->m_gpuRayRigidPairs);
-		
+
 		int numRayRigidPairs = -1;
 		m_data->m_gpuNumRayRigidPairs->copyToHostPointer(&numRayRigidPairs, 1);
-		if( numRayRigidPairs > m_data->m_gpuRayRigidPairs->size() )
+		if (numRayRigidPairs > m_data->m_gpuRayRigidPairs->size())
 		{
 			numRayRigidPairs = m_data->m_gpuRayRigidPairs->size();
 			m_data->m_gpuNumRayRigidPairs->copyFromHostPointer(&numRayRigidPairs, 1);
 		}
-		
-		m_data->m_gpuRayRigidPairs->resize(numRayRigidPairs);	//Radix sort needs b3OpenCLArray::size() to be correct
-		
+
+		m_data->m_gpuRayRigidPairs->resize(numRayRigidPairs);  //Radix sort needs b3OpenCLArray::size() to be correct
+
 		//Sort ray-rigid pairs by ray index
 		{
 			B3_PROFILE("sort ray-rigid pairs");
-			m_data->m_radixSorter->execute( *reinterpret_cast< b3OpenCLArray<b3SortData>* >(m_data->m_gpuRayRigidPairs) );
+			m_data->m_radixSorter->execute(*reinterpret_cast<b3OpenCLArray<b3SortData>*>(m_data->m_gpuRayRigidPairs));
 		}
-		
+
 		//detect start,count of each ray pair
 		{
 			B3_PROFILE("detect ray-rigid pair index ranges");
-			
+
 			{
 				B3_PROFILE("reset ray-rigid pair index ranges");
-				
-				m_data->m_fill->execute(*m_data->m_firstRayRigidPairIndexPerRay, numRayRigidPairs, numRays);	//atomic_min used to find first index
+
+				m_data->m_fill->execute(*m_data->m_firstRayRigidPairIndexPerRay, numRayRigidPairs, numRays);  //atomic_min used to find first index
 				m_data->m_fill->execute(*m_data->m_numRayRigidPairsPerRay, 0, numRays);
 				clFinish(m_data->m_q);
 			}
-			
-			b3BufferInfoCL bufferInfo[] = 
-			{
-				b3BufferInfoCL( m_data->m_gpuRayRigidPairs->getBufferCL() ),
-				
-				b3BufferInfoCL( m_data->m_firstRayRigidPairIndexPerRay->getBufferCL() ),
-				b3BufferInfoCL( m_data->m_numRayRigidPairsPerRay->getBufferCL() )
-			};
-			
+
+			b3BufferInfoCL bufferInfo[] =
+				{
+					b3BufferInfoCL(m_data->m_gpuRayRigidPairs->getBufferCL()),
+
+					b3BufferInfoCL(m_data->m_firstRayRigidPairIndexPerRay->getBufferCL()),
+					b3BufferInfoCL(m_data->m_numRayRigidPairsPerRay->getBufferCL())};
+
 			b3LauncherCL launcher(m_data->m_q, m_data->m_findRayRigidPairIndexRanges, "m_findRayRigidPairIndexRanges");
-			launcher.setBuffers( bufferInfo, sizeof(bufferInfo)/sizeof(b3BufferInfoCL) );
+			launcher.setBuffers(bufferInfo, sizeof(bufferInfo) / sizeof(b3BufferInfoCL));
 			launcher.setConst(numRayRigidPairs);
-			
+
 			launcher.launch1D(numRayRigidPairs);
 			clFinish(m_data->m_q);
 		}
-		
+
 		{
 			B3_PROFILE("ray-rigid intersection");
-			
-			b3BufferInfoCL bufferInfo[] = 
-			{
-				b3BufferInfoCL( m_data->m_gpuRays->getBufferCL() ),
-				b3BufferInfoCL( m_data->m_gpuHitResults->getBufferCL() ),
-				b3BufferInfoCL( m_data->m_firstRayRigidPairIndexPerRay->getBufferCL() ),
-				b3BufferInfoCL( m_data->m_numRayRigidPairsPerRay->getBufferCL() ),
-				
-				b3BufferInfoCL( narrowphaseData->m_bodyBufferGPU->getBufferCL() ),
-				b3BufferInfoCL( narrowphaseData->m_collidablesGPU->getBufferCL() ),
-				b3BufferInfoCL( narrowphaseData->m_convexFacesGPU->getBufferCL() ),
-				b3BufferInfoCL( narrowphaseData->m_convexPolyhedraGPU->getBufferCL() ),
-				
-				b3BufferInfoCL( m_data->m_gpuRayRigidPairs->getBufferCL() )
-			};
-			
+
+			b3BufferInfoCL bufferInfo[] =
+				{
+					b3BufferInfoCL(m_data->m_gpuRays->getBufferCL()),
+					b3BufferInfoCL(m_data->m_gpuHitResults->getBufferCL()),
+					b3BufferInfoCL(m_data->m_firstRayRigidPairIndexPerRay->getBufferCL()),
+					b3BufferInfoCL(m_data->m_numRayRigidPairsPerRay->getBufferCL()),
+
+					b3BufferInfoCL(narrowphaseData->m_bodyBufferGPU->getBufferCL()),
+					b3BufferInfoCL(narrowphaseData->m_collidablesGPU->getBufferCL()),
+					b3BufferInfoCL(narrowphaseData->m_convexFacesGPU->getBufferCL()),
+					b3BufferInfoCL(narrowphaseData->m_convexPolyhedraGPU->getBufferCL()),
+
+					b3BufferInfoCL(m_data->m_gpuRayRigidPairs->getBufferCL())};
+
 			b3LauncherCL launcher(m_data->m_q, m_data->m_raytracePairsKernel, "m_raytracePairsKernel");
-			launcher.setBuffers( bufferInfo, sizeof(bufferInfo)/sizeof(b3BufferInfoCL) );
+			launcher.setBuffers(bufferInfo, sizeof(bufferInfo) / sizeof(b3BufferInfoCL));
 			launcher.setConst(numRays);
-			
+
 			launcher.launch1D(numRays);
 			clFinish(m_data->m_q);
 		}
 	}
-	
-	
 
 	//copy results
 	{
 		B3_PROFILE("raycast copyToHost");
 		m_data->m_gpuHitResults->copyToHost(hitResults);
 	}
-
 }
\ No newline at end of file
diff --git a/thirdparty/bullet/Bullet3OpenCL/Raycast/b3GpuRaycast.h b/thirdparty/bullet/Bullet3OpenCL/Raycast/b3GpuRaycast.h
index 3a5cf44b795..f1f6ffd4020 100644
--- a/thirdparty/bullet/Bullet3OpenCL/Raycast/b3GpuRaycast.h
+++ b/thirdparty/bullet/Bullet3OpenCL/Raycast/b3GpuRaycast.h
@@ -7,26 +7,22 @@
 #include "Bullet3Common/b3AlignedObjectArray.h"
 #include "Bullet3Collision/NarrowPhaseCollision/b3RaycastInfo.h"
 
-
-
 class b3GpuRaycast
 {
 protected:
 	struct b3GpuRaycastInternalData* m_data;
+
 public:
-	b3GpuRaycast(cl_context ctx,cl_device_id device, cl_command_queue  q);
+	b3GpuRaycast(cl_context ctx, cl_device_id device, cl_command_queue q);
 	virtual ~b3GpuRaycast();
 
-	void castRaysHost(const b3AlignedObjectArray<b3RayInfo>& raysIn,	b3AlignedObjectArray<b3RayHit>& hitResults,
-		int numBodies, const struct b3RigidBodyData* bodies, int numCollidables, const struct b3Collidable* collidables,
-		const struct b3GpuNarrowPhaseInternalData* narrowphaseData);
+	void castRaysHost(const b3AlignedObjectArray<b3RayInfo>& raysIn, b3AlignedObjectArray<b3RayHit>& hitResults,
+					  int numBodies, const struct b3RigidBodyData* bodies, int numCollidables, const struct b3Collidable* collidables,
+					  const struct b3GpuNarrowPhaseInternalData* narrowphaseData);
 
-	void castRays(const b3AlignedObjectArray<b3RayInfo>& rays,	b3AlignedObjectArray<b3RayHit>& hitResults,
-		int numBodies,const struct b3RigidBodyData* bodies, int numCollidables, const struct b3Collidable* collidables,
-		const struct b3GpuNarrowPhaseInternalData* narrowphaseData, class b3GpuBroadphaseInterface* broadphase);
-	
-
-		
+	void castRays(const b3AlignedObjectArray<b3RayInfo>& rays, b3AlignedObjectArray<b3RayHit>& hitResults,
+				  int numBodies, const struct b3RigidBodyData* bodies, int numCollidables, const struct b3Collidable* collidables,
+				  const struct b3GpuNarrowPhaseInternalData* narrowphaseData, class b3GpuBroadphaseInterface* broadphase);
 };
 
-#endif //B3_GPU_RAYCAST_H
+#endif  //B3_GPU_RAYCAST_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/Raycast/kernels/rayCastKernels.h b/thirdparty/bullet/Bullet3OpenCL/Raycast/kernels/rayCastKernels.h
index 6257909a4d6..94f6a8eb9f7 100644
--- a/thirdparty/bullet/Bullet3OpenCL/Raycast/kernels/rayCastKernels.h
+++ b/thirdparty/bullet/Bullet3OpenCL/Raycast/kernels/rayCastKernels.h
@@ -1,381 +1,380 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* rayCastKernelCL= \
-"#define SHAPE_CONVEX_HULL 3\n"
-"#define SHAPE_PLANE 4\n"
-"#define SHAPE_CONCAVE_TRIMESH 5\n"
-"#define SHAPE_COMPOUND_OF_CONVEX_HULLS 6\n"
-"#define SHAPE_SPHERE 7\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_from;\n"
-"	float4 m_to;\n"
-"} b3RayInfo;\n"
-"typedef struct\n"
-"{\n"
-"	float m_hitFraction;\n"
-"	int	m_hitResult0;\n"
-"	int	m_hitResult1;\n"
-"	int	m_hitResult2;\n"
-"	float4	m_hitPoint;\n"
-"	float4	m_hitNormal;\n"
-"} b3RayHit;\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_pos;\n"
-"	float4 m_quat;\n"
-"	float4 m_linVel;\n"
-"	float4 m_angVel;\n"
-"	unsigned int m_collidableIdx;\n"
-"	float m_invMass;\n"
-"	float m_restituitionCoeff;\n"
-"	float m_frictionCoeff;\n"
-"} Body;\n"
-"typedef struct Collidable\n"
-"{\n"
-"	union {\n"
-"		int m_numChildShapes;\n"
-"		int m_bvhIndex;\n"
-"	};\n"
-"	float m_radius;\n"
-"	int m_shapeType;\n"
-"	int m_shapeIndex;\n"
-"} Collidable;\n"
-"typedef struct  \n"
-"{\n"
-"	float4		m_localCenter;\n"
-"	float4		m_extents;\n"
-"	float4		mC;\n"
-"	float4		mE;\n"
-"	float			m_radius;\n"
-"	int	m_faceOffset;\n"
-"	int m_numFaces;\n"
-"	int	m_numVertices;\n"
-"	int m_vertexOffset;\n"
-"	int	m_uniqueEdgesOffset;\n"
-"	int	m_numUniqueEdges;\n"
-"	int m_unused;\n"
-"} ConvexPolyhedronCL;\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_plane;\n"
-"	int m_indexOffset;\n"
-"	int m_numIndices;\n"
-"} b3GpuFace;\n"
-"///////////////////////////////////////\n"
-"//	Quaternion\n"
-"///////////////////////////////////////\n"
-"typedef float4 Quaternion;\n"
-"__inline\n"
-"	Quaternion qtMul(Quaternion a, Quaternion b);\n"
-"__inline\n"
-"	Quaternion qtNormalize(Quaternion in);\n"
-"__inline\n"
-"	Quaternion qtInvert(Quaternion q);\n"
-"__inline\n"
-"	float dot3F4(float4 a, float4 b)\n"
-"{\n"
-"	float4 a1 = (float4)(a.xyz,0.f);\n"
-"	float4 b1 = (float4)(b.xyz,0.f);\n"
-"	return dot(a1, b1);\n"
-"}\n"
-"__inline\n"
-"	Quaternion qtMul(Quaternion a, Quaternion b)\n"
-"{\n"
-"	Quaternion ans;\n"
-"	ans = cross( a, b );\n"
-"	ans += a.w*b+b.w*a;\n"
-"	//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
-"	ans.w = a.w*b.w - dot3F4(a, b);\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"	Quaternion qtNormalize(Quaternion in)\n"
-"{\n"
-"	return fast_normalize(in);\n"
-"	//	in /= length( in );\n"
-"	//	return in;\n"
-"}\n"
-"__inline\n"
-"	float4 qtRotate(Quaternion q, float4 vec)\n"
-"{\n"
-"	Quaternion qInv = qtInvert( q );\n"
-"	float4 vcpy = vec;\n"
-"	vcpy.w = 0.f;\n"
-"	float4 out = qtMul(q,vcpy);\n"
-"	out = qtMul(out,qInv);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"	Quaternion qtInvert(Quaternion q)\n"
-"{\n"
-"	return (Quaternion)(-q.xyz, q.w);\n"
-"}\n"
-"__inline\n"
-"	float4 qtInvRotate(const Quaternion q, float4 vec)\n"
-"{\n"
-"	return qtRotate( qtInvert( q ), vec );\n"
-"}\n"
-"void	trInverse(float4 translationIn, Quaternion orientationIn,\n"
-"	float4* translationOut, Quaternion* orientationOut)\n"
-"{\n"
-"	*orientationOut = qtInvert(orientationIn);\n"
-"	*translationOut = qtRotate(*orientationOut, -translationIn);\n"
-"}\n"
-"bool rayConvex(float4 rayFromLocal, float4 rayToLocal, int numFaces, int faceOffset,\n"
-"	__global const b3GpuFace* faces, float* hitFraction, float4* hitNormal)\n"
-"{\n"
-"	rayFromLocal.w = 0.f;\n"
-"	rayToLocal.w = 0.f;\n"
-"	bool result = true;\n"
-"	float exitFraction = hitFraction[0];\n"
-"	float enterFraction = -0.3f;\n"
-"	float4 curHitNormal = (float4)(0,0,0,0);\n"
-"	for (int i=0;i<numFaces && result;i++)\n"
-"	{\n"
-"		b3GpuFace face = faces[faceOffset+i];\n"
-"		float fromPlaneDist = dot(rayFromLocal,face.m_plane)+face.m_plane.w;\n"
-"		float toPlaneDist = dot(rayToLocal,face.m_plane)+face.m_plane.w;\n"
-"		if (fromPlaneDist<0.f)\n"
-"		{\n"
-"			if (toPlaneDist >= 0.f)\n"
-"			{\n"
-"				float fraction = fromPlaneDist / (fromPlaneDist-toPlaneDist);\n"
-"				if (exitFraction>fraction)\n"
-"				{\n"
-"					exitFraction = fraction;\n"
-"				}\n"
-"			} 			\n"
-"		} else\n"
-"		{\n"
-"			if (toPlaneDist<0.f)\n"
-"			{\n"
-"				float fraction = fromPlaneDist / (fromPlaneDist-toPlaneDist);\n"
-"				if (enterFraction <= fraction)\n"
-"				{\n"
-"					enterFraction = fraction;\n"
-"					curHitNormal = face.m_plane;\n"
-"					curHitNormal.w = 0.f;\n"
-"				}\n"
-"			} else\n"
-"			{\n"
-"				result = false;\n"
-"			}\n"
-"		}\n"
-"		if (exitFraction <= enterFraction)\n"
-"			result = false;\n"
-"	}\n"
-"	if (enterFraction < 0.f)\n"
-"	{\n"
-"		result = false;\n"
-"	}\n"
-"	if (result)\n"
-"	{	\n"
-"		hitFraction[0] = enterFraction;\n"
-"		hitNormal[0] = curHitNormal;\n"
-"	}\n"
-"	return result;\n"
-"}\n"
-"bool sphere_intersect(float4 spherePos,  float radius, float4 rayFrom, float4 rayTo, float* hitFraction)\n"
-"{\n"
-"	float4 rs = rayFrom - spherePos;\n"
-"	rs.w = 0.f;\n"
-"	float4 rayDir = rayTo-rayFrom;\n"
-"	rayDir.w = 0.f;\n"
-"	float A = dot(rayDir,rayDir);\n"
-"	float B = dot(rs, rayDir);\n"
-"	float C = dot(rs, rs) - (radius * radius);\n"
-"	float D = B * B - A*C;\n"
-"	if (D > 0.0f)\n"
-"	{\n"
-"		float t = (-B - sqrt(D))/A;\n"
-"		if ( (t >= 0.0f) && (t < (*hitFraction)) )\n"
-"		{\n"
-"			*hitFraction = t;\n"
-"			return true;\n"
-"		}\n"
-"	}\n"
-"	return false;\n"
-"}\n"
-"float4 setInterpolate3(float4 from, float4 to, float t)\n"
-"{\n"
-"	float s = 1.0f - t;\n"
-"	float4 result;\n"
-"	result = s * from + t * to;\n"
-"	result.w = 0.f;	\n"
-"	return result;	\n"
-"}\n"
-"__kernel void rayCastKernel(  \n"
-"	int numRays, \n"
-"	const __global b3RayInfo* rays, \n"
-"	__global b3RayHit* hitResults, \n"
-"	const int numBodies, \n"
-"	__global Body* bodies,\n"
-"	__global Collidable* collidables,\n"
-"	__global const b3GpuFace* faces,\n"
-"	__global const ConvexPolyhedronCL* convexShapes	)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	if (i>=numRays)\n"
-"		return;\n"
-"	hitResults[i].m_hitFraction = 1.f;\n"
-"	float4 rayFrom = rays[i].m_from;\n"
-"	float4 rayTo = rays[i].m_to;\n"
-"	float hitFraction = 1.f;\n"
-"	float4 hitPoint;\n"
-"	float4 hitNormal;\n"
-"	int hitBodyIndex= -1;\n"
-"	int cachedCollidableIndex = -1;\n"
-"	Collidable cachedCollidable;\n"
-"	for (int b=0;b<numBodies;b++)\n"
-"	{\n"
-"		if (hitResults[i].m_hitResult2==b)\n"
-"			continue;\n"
-"		Body body = bodies[b];\n"
-"		float4 pos = body.m_pos;\n"
-"		float4 orn = body.m_quat;\n"
-"		if (cachedCollidableIndex != body.m_collidableIdx)\n"
-"		{\n"
-"			cachedCollidableIndex = body.m_collidableIdx;\n"
-"			cachedCollidable = collidables[cachedCollidableIndex];\n"
-"		}\n"
-"		if (cachedCollidable.m_shapeType == SHAPE_CONVEX_HULL)\n"
-"		{\n"
-"			float4 invPos = (float4)(0,0,0,0);\n"
-"			float4 invOrn = (float4)(0,0,0,0);\n"
-"			float4 rayFromLocal = (float4)(0,0,0,0);\n"
-"			float4 rayToLocal = (float4)(0,0,0,0);\n"
-"			invOrn = qtInvert(orn);\n"
-"			invPos = qtRotate(invOrn, -pos);\n"
-"			rayFromLocal = qtRotate( invOrn, rayFrom ) + invPos;\n"
-"			rayToLocal = qtRotate( invOrn, rayTo) + invPos;\n"
-"			rayFromLocal.w = 0.f;\n"
-"			rayToLocal.w = 0.f;\n"
-"			int numFaces = convexShapes[cachedCollidable.m_shapeIndex].m_numFaces;\n"
-"			int faceOffset = convexShapes[cachedCollidable.m_shapeIndex].m_faceOffset;\n"
-"			if (numFaces)\n"
-"			{\n"
-"				if (rayConvex(rayFromLocal, rayToLocal, numFaces, faceOffset,faces, &hitFraction, &hitNormal))\n"
-"				{\n"
-"					hitBodyIndex = b;\n"
-"					\n"
-"				}\n"
-"			}\n"
-"		}\n"
-"		if (cachedCollidable.m_shapeType == SHAPE_SPHERE)\n"
-"		{\n"
-"			float radius = cachedCollidable.m_radius;\n"
-"		\n"
-"			if (sphere_intersect(pos,  radius, rayFrom, rayTo, &hitFraction))\n"
-"			{\n"
-"				hitBodyIndex = b;\n"
-"				hitNormal = (float4) (hitPoint-bodies[b].m_pos);\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	if (hitBodyIndex>=0)\n"
-"	{\n"
-"		hitPoint = setInterpolate3(rayFrom, rayTo,hitFraction);\n"
-"		hitResults[i].m_hitFraction = hitFraction;\n"
-"		hitResults[i].m_hitPoint = hitPoint;\n"
-"		hitResults[i].m_hitNormal = normalize(hitNormal);\n"
-"		hitResults[i].m_hitResult0 = hitBodyIndex;\n"
-"	}\n"
-"}\n"
-"__kernel void findRayRigidPairIndexRanges(__global int2* rayRigidPairs, \n"
-"											__global int* out_firstRayRigidPairIndexPerRay,\n"
-"											__global int* out_numRayRigidPairsPerRay,\n"
-"											int numRayRigidPairs)\n"
-"{\n"
-"	int rayRigidPairIndex = get_global_id(0);\n"
-"	if (rayRigidPairIndex >= numRayRigidPairs) return;\n"
-"	\n"
-"	int rayIndex = rayRigidPairs[rayRigidPairIndex].x;\n"
-"	\n"
-"	atomic_min(&out_firstRayRigidPairIndexPerRay[rayIndex], rayRigidPairIndex);\n"
-"	atomic_inc(&out_numRayRigidPairsPerRay[rayIndex]);\n"
-"}\n"
-"__kernel void rayCastPairsKernel(const __global b3RayInfo* rays, \n"
-"								__global b3RayHit* hitResults, \n"
-"								__global int* firstRayRigidPairIndexPerRay,\n"
-"								__global int* numRayRigidPairsPerRay,\n"
-"									\n"
-"								__global Body* bodies,\n"
-"								__global Collidable* collidables,\n"
-"								__global const b3GpuFace* faces,\n"
-"								__global const ConvexPolyhedronCL* convexShapes,\n"
-"								\n"
-"								__global int2* rayRigidPairs,\n"
-"								int numRays)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	if (i >= numRays) return;\n"
-"	\n"
-"	float4 rayFrom = rays[i].m_from;\n"
-"	float4 rayTo = rays[i].m_to;\n"
-"		\n"
-"	hitResults[i].m_hitFraction = 1.f;\n"
-"		\n"
-"	float hitFraction = 1.f;\n"
-"	float4 hitPoint;\n"
-"	float4 hitNormal;\n"
-"	int hitBodyIndex = -1;\n"
-"		\n"
-"	//\n"
-"	for(int pair = 0; pair < numRayRigidPairsPerRay[i]; ++pair)\n"
-"	{\n"
-"		int rayRigidPairIndex = pair + firstRayRigidPairIndexPerRay[i];\n"
-"		int b = rayRigidPairs[rayRigidPairIndex].y;\n"
-"		\n"
-"		if (hitResults[i].m_hitResult2 == b) continue;\n"
-"		\n"
-"		Body body = bodies[b];\n"
-"		Collidable rigidCollidable = collidables[body.m_collidableIdx];\n"
-"		\n"
-"		float4 pos = body.m_pos;\n"
-"		float4 orn = body.m_quat;\n"
-"		\n"
-"		if (rigidCollidable.m_shapeType == SHAPE_CONVEX_HULL)\n"
-"		{\n"
-"			float4 invPos = (float4)(0,0,0,0);\n"
-"			float4 invOrn = (float4)(0,0,0,0);\n"
-"			float4 rayFromLocal = (float4)(0,0,0,0);\n"
-"			float4 rayToLocal = (float4)(0,0,0,0);\n"
-"			invOrn = qtInvert(orn);\n"
-"			invPos = qtRotate(invOrn, -pos);\n"
-"			rayFromLocal = qtRotate( invOrn, rayFrom ) + invPos;\n"
-"			rayToLocal = qtRotate( invOrn, rayTo) + invPos;\n"
-"			rayFromLocal.w = 0.f;\n"
-"			rayToLocal.w = 0.f;\n"
-"			int numFaces = convexShapes[rigidCollidable.m_shapeIndex].m_numFaces;\n"
-"			int faceOffset = convexShapes[rigidCollidable.m_shapeIndex].m_faceOffset;\n"
-"			\n"
-"			if (numFaces && rayConvex(rayFromLocal, rayToLocal, numFaces, faceOffset,faces, &hitFraction, &hitNormal))\n"
-"			{\n"
-"				hitBodyIndex = b;\n"
-"				hitPoint = setInterpolate3(rayFrom, rayTo, hitFraction);\n"
-"			}\n"
-"		}\n"
-"		\n"
-"		if (rigidCollidable.m_shapeType == SHAPE_SPHERE)\n"
-"		{\n"
-"			float radius = rigidCollidable.m_radius;\n"
-"		\n"
-"			if (sphere_intersect(pos, radius, rayFrom, rayTo, &hitFraction))\n"
-"			{\n"
-"				hitBodyIndex = b;\n"
-"				hitPoint = setInterpolate3(rayFrom, rayTo, hitFraction);\n"
-"				hitNormal = (float4) (hitPoint - bodies[b].m_pos);\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"	\n"
-"	if (hitBodyIndex >= 0)\n"
-"	{\n"
-"		hitResults[i].m_hitFraction = hitFraction;\n"
-"		hitResults[i].m_hitPoint = hitPoint;\n"
-"		hitResults[i].m_hitNormal = normalize(hitNormal);\n"
-"		hitResults[i].m_hitResult0 = hitBodyIndex;\n"
-"	}\n"
-"	\n"
-"}\n"
-;
+static const char* rayCastKernelCL =
+	"#define SHAPE_CONVEX_HULL 3\n"
+	"#define SHAPE_PLANE 4\n"
+	"#define SHAPE_CONCAVE_TRIMESH 5\n"
+	"#define SHAPE_COMPOUND_OF_CONVEX_HULLS 6\n"
+	"#define SHAPE_SPHERE 7\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_from;\n"
+	"	float4 m_to;\n"
+	"} b3RayInfo;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float m_hitFraction;\n"
+	"	int	m_hitResult0;\n"
+	"	int	m_hitResult1;\n"
+	"	int	m_hitResult2;\n"
+	"	float4	m_hitPoint;\n"
+	"	float4	m_hitNormal;\n"
+	"} b3RayHit;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_pos;\n"
+	"	float4 m_quat;\n"
+	"	float4 m_linVel;\n"
+	"	float4 m_angVel;\n"
+	"	unsigned int m_collidableIdx;\n"
+	"	float m_invMass;\n"
+	"	float m_restituitionCoeff;\n"
+	"	float m_frictionCoeff;\n"
+	"} Body;\n"
+	"typedef struct Collidable\n"
+	"{\n"
+	"	union {\n"
+	"		int m_numChildShapes;\n"
+	"		int m_bvhIndex;\n"
+	"	};\n"
+	"	float m_radius;\n"
+	"	int m_shapeType;\n"
+	"	int m_shapeIndex;\n"
+	"} Collidable;\n"
+	"typedef struct  \n"
+	"{\n"
+	"	float4		m_localCenter;\n"
+	"	float4		m_extents;\n"
+	"	float4		mC;\n"
+	"	float4		mE;\n"
+	"	float			m_radius;\n"
+	"	int	m_faceOffset;\n"
+	"	int m_numFaces;\n"
+	"	int	m_numVertices;\n"
+	"	int m_vertexOffset;\n"
+	"	int	m_uniqueEdgesOffset;\n"
+	"	int	m_numUniqueEdges;\n"
+	"	int m_unused;\n"
+	"} ConvexPolyhedronCL;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_plane;\n"
+	"	int m_indexOffset;\n"
+	"	int m_numIndices;\n"
+	"} b3GpuFace;\n"
+	"///////////////////////////////////////\n"
+	"//	Quaternion\n"
+	"///////////////////////////////////////\n"
+	"typedef float4 Quaternion;\n"
+	"__inline\n"
+	"	Quaternion qtMul(Quaternion a, Quaternion b);\n"
+	"__inline\n"
+	"	Quaternion qtNormalize(Quaternion in);\n"
+	"__inline\n"
+	"	Quaternion qtInvert(Quaternion q);\n"
+	"__inline\n"
+	"	float dot3F4(float4 a, float4 b)\n"
+	"{\n"
+	"	float4 a1 = (float4)(a.xyz,0.f);\n"
+	"	float4 b1 = (float4)(b.xyz,0.f);\n"
+	"	return dot(a1, b1);\n"
+	"}\n"
+	"__inline\n"
+	"	Quaternion qtMul(Quaternion a, Quaternion b)\n"
+	"{\n"
+	"	Quaternion ans;\n"
+	"	ans = cross( a, b );\n"
+	"	ans += a.w*b+b.w*a;\n"
+	"	//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
+	"	ans.w = a.w*b.w - dot3F4(a, b);\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"	Quaternion qtNormalize(Quaternion in)\n"
+	"{\n"
+	"	return fast_normalize(in);\n"
+	"	//	in /= length( in );\n"
+	"	//	return in;\n"
+	"}\n"
+	"__inline\n"
+	"	float4 qtRotate(Quaternion q, float4 vec)\n"
+	"{\n"
+	"	Quaternion qInv = qtInvert( q );\n"
+	"	float4 vcpy = vec;\n"
+	"	vcpy.w = 0.f;\n"
+	"	float4 out = qtMul(q,vcpy);\n"
+	"	out = qtMul(out,qInv);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"	Quaternion qtInvert(Quaternion q)\n"
+	"{\n"
+	"	return (Quaternion)(-q.xyz, q.w);\n"
+	"}\n"
+	"__inline\n"
+	"	float4 qtInvRotate(const Quaternion q, float4 vec)\n"
+	"{\n"
+	"	return qtRotate( qtInvert( q ), vec );\n"
+	"}\n"
+	"void	trInverse(float4 translationIn, Quaternion orientationIn,\n"
+	"	float4* translationOut, Quaternion* orientationOut)\n"
+	"{\n"
+	"	*orientationOut = qtInvert(orientationIn);\n"
+	"	*translationOut = qtRotate(*orientationOut, -translationIn);\n"
+	"}\n"
+	"bool rayConvex(float4 rayFromLocal, float4 rayToLocal, int numFaces, int faceOffset,\n"
+	"	__global const b3GpuFace* faces, float* hitFraction, float4* hitNormal)\n"
+	"{\n"
+	"	rayFromLocal.w = 0.f;\n"
+	"	rayToLocal.w = 0.f;\n"
+	"	bool result = true;\n"
+	"	float exitFraction = hitFraction[0];\n"
+	"	float enterFraction = -0.3f;\n"
+	"	float4 curHitNormal = (float4)(0,0,0,0);\n"
+	"	for (int i=0;i<numFaces && result;i++)\n"
+	"	{\n"
+	"		b3GpuFace face = faces[faceOffset+i];\n"
+	"		float fromPlaneDist = dot(rayFromLocal,face.m_plane)+face.m_plane.w;\n"
+	"		float toPlaneDist = dot(rayToLocal,face.m_plane)+face.m_plane.w;\n"
+	"		if (fromPlaneDist<0.f)\n"
+	"		{\n"
+	"			if (toPlaneDist >= 0.f)\n"
+	"			{\n"
+	"				float fraction = fromPlaneDist / (fromPlaneDist-toPlaneDist);\n"
+	"				if (exitFraction>fraction)\n"
+	"				{\n"
+	"					exitFraction = fraction;\n"
+	"				}\n"
+	"			} 			\n"
+	"		} else\n"
+	"		{\n"
+	"			if (toPlaneDist<0.f)\n"
+	"			{\n"
+	"				float fraction = fromPlaneDist / (fromPlaneDist-toPlaneDist);\n"
+	"				if (enterFraction <= fraction)\n"
+	"				{\n"
+	"					enterFraction = fraction;\n"
+	"					curHitNormal = face.m_plane;\n"
+	"					curHitNormal.w = 0.f;\n"
+	"				}\n"
+	"			} else\n"
+	"			{\n"
+	"				result = false;\n"
+	"			}\n"
+	"		}\n"
+	"		if (exitFraction <= enterFraction)\n"
+	"			result = false;\n"
+	"	}\n"
+	"	if (enterFraction < 0.f)\n"
+	"	{\n"
+	"		result = false;\n"
+	"	}\n"
+	"	if (result)\n"
+	"	{	\n"
+	"		hitFraction[0] = enterFraction;\n"
+	"		hitNormal[0] = curHitNormal;\n"
+	"	}\n"
+	"	return result;\n"
+	"}\n"
+	"bool sphere_intersect(float4 spherePos,  float radius, float4 rayFrom, float4 rayTo, float* hitFraction)\n"
+	"{\n"
+	"	float4 rs = rayFrom - spherePos;\n"
+	"	rs.w = 0.f;\n"
+	"	float4 rayDir = rayTo-rayFrom;\n"
+	"	rayDir.w = 0.f;\n"
+	"	float A = dot(rayDir,rayDir);\n"
+	"	float B = dot(rs, rayDir);\n"
+	"	float C = dot(rs, rs) - (radius * radius);\n"
+	"	float D = B * B - A*C;\n"
+	"	if (D > 0.0f)\n"
+	"	{\n"
+	"		float t = (-B - sqrt(D))/A;\n"
+	"		if ( (t >= 0.0f) && (t < (*hitFraction)) )\n"
+	"		{\n"
+	"			*hitFraction = t;\n"
+	"			return true;\n"
+	"		}\n"
+	"	}\n"
+	"	return false;\n"
+	"}\n"
+	"float4 setInterpolate3(float4 from, float4 to, float t)\n"
+	"{\n"
+	"	float s = 1.0f - t;\n"
+	"	float4 result;\n"
+	"	result = s * from + t * to;\n"
+	"	result.w = 0.f;	\n"
+	"	return result;	\n"
+	"}\n"
+	"__kernel void rayCastKernel(  \n"
+	"	int numRays, \n"
+	"	const __global b3RayInfo* rays, \n"
+	"	__global b3RayHit* hitResults, \n"
+	"	const int numBodies, \n"
+	"	__global Body* bodies,\n"
+	"	__global Collidable* collidables,\n"
+	"	__global const b3GpuFace* faces,\n"
+	"	__global const ConvexPolyhedronCL* convexShapes	)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	if (i>=numRays)\n"
+	"		return;\n"
+	"	hitResults[i].m_hitFraction = 1.f;\n"
+	"	float4 rayFrom = rays[i].m_from;\n"
+	"	float4 rayTo = rays[i].m_to;\n"
+	"	float hitFraction = 1.f;\n"
+	"	float4 hitPoint;\n"
+	"	float4 hitNormal;\n"
+	"	int hitBodyIndex= -1;\n"
+	"	int cachedCollidableIndex = -1;\n"
+	"	Collidable cachedCollidable;\n"
+	"	for (int b=0;b<numBodies;b++)\n"
+	"	{\n"
+	"		if (hitResults[i].m_hitResult2==b)\n"
+	"			continue;\n"
+	"		Body body = bodies[b];\n"
+	"		float4 pos = body.m_pos;\n"
+	"		float4 orn = body.m_quat;\n"
+	"		if (cachedCollidableIndex != body.m_collidableIdx)\n"
+	"		{\n"
+	"			cachedCollidableIndex = body.m_collidableIdx;\n"
+	"			cachedCollidable = collidables[cachedCollidableIndex];\n"
+	"		}\n"
+	"		if (cachedCollidable.m_shapeType == SHAPE_CONVEX_HULL)\n"
+	"		{\n"
+	"			float4 invPos = (float4)(0,0,0,0);\n"
+	"			float4 invOrn = (float4)(0,0,0,0);\n"
+	"			float4 rayFromLocal = (float4)(0,0,0,0);\n"
+	"			float4 rayToLocal = (float4)(0,0,0,0);\n"
+	"			invOrn = qtInvert(orn);\n"
+	"			invPos = qtRotate(invOrn, -pos);\n"
+	"			rayFromLocal = qtRotate( invOrn, rayFrom ) + invPos;\n"
+	"			rayToLocal = qtRotate( invOrn, rayTo) + invPos;\n"
+	"			rayFromLocal.w = 0.f;\n"
+	"			rayToLocal.w = 0.f;\n"
+	"			int numFaces = convexShapes[cachedCollidable.m_shapeIndex].m_numFaces;\n"
+	"			int faceOffset = convexShapes[cachedCollidable.m_shapeIndex].m_faceOffset;\n"
+	"			if (numFaces)\n"
+	"			{\n"
+	"				if (rayConvex(rayFromLocal, rayToLocal, numFaces, faceOffset,faces, &hitFraction, &hitNormal))\n"
+	"				{\n"
+	"					hitBodyIndex = b;\n"
+	"					\n"
+	"				}\n"
+	"			}\n"
+	"		}\n"
+	"		if (cachedCollidable.m_shapeType == SHAPE_SPHERE)\n"
+	"		{\n"
+	"			float radius = cachedCollidable.m_radius;\n"
+	"		\n"
+	"			if (sphere_intersect(pos,  radius, rayFrom, rayTo, &hitFraction))\n"
+	"			{\n"
+	"				hitBodyIndex = b;\n"
+	"				hitNormal = (float4) (hitPoint-bodies[b].m_pos);\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	if (hitBodyIndex>=0)\n"
+	"	{\n"
+	"		hitPoint = setInterpolate3(rayFrom, rayTo,hitFraction);\n"
+	"		hitResults[i].m_hitFraction = hitFraction;\n"
+	"		hitResults[i].m_hitPoint = hitPoint;\n"
+	"		hitResults[i].m_hitNormal = normalize(hitNormal);\n"
+	"		hitResults[i].m_hitResult0 = hitBodyIndex;\n"
+	"	}\n"
+	"}\n"
+	"__kernel void findRayRigidPairIndexRanges(__global int2* rayRigidPairs, \n"
+	"											__global int* out_firstRayRigidPairIndexPerRay,\n"
+	"											__global int* out_numRayRigidPairsPerRay,\n"
+	"											int numRayRigidPairs)\n"
+	"{\n"
+	"	int rayRigidPairIndex = get_global_id(0);\n"
+	"	if (rayRigidPairIndex >= numRayRigidPairs) return;\n"
+	"	\n"
+	"	int rayIndex = rayRigidPairs[rayRigidPairIndex].x;\n"
+	"	\n"
+	"	atomic_min(&out_firstRayRigidPairIndexPerRay[rayIndex], rayRigidPairIndex);\n"
+	"	atomic_inc(&out_numRayRigidPairsPerRay[rayIndex]);\n"
+	"}\n"
+	"__kernel void rayCastPairsKernel(const __global b3RayInfo* rays, \n"
+	"								__global b3RayHit* hitResults, \n"
+	"								__global int* firstRayRigidPairIndexPerRay,\n"
+	"								__global int* numRayRigidPairsPerRay,\n"
+	"									\n"
+	"								__global Body* bodies,\n"
+	"								__global Collidable* collidables,\n"
+	"								__global const b3GpuFace* faces,\n"
+	"								__global const ConvexPolyhedronCL* convexShapes,\n"
+	"								\n"
+	"								__global int2* rayRigidPairs,\n"
+	"								int numRays)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	if (i >= numRays) return;\n"
+	"	\n"
+	"	float4 rayFrom = rays[i].m_from;\n"
+	"	float4 rayTo = rays[i].m_to;\n"
+	"		\n"
+	"	hitResults[i].m_hitFraction = 1.f;\n"
+	"		\n"
+	"	float hitFraction = 1.f;\n"
+	"	float4 hitPoint;\n"
+	"	float4 hitNormal;\n"
+	"	int hitBodyIndex = -1;\n"
+	"		\n"
+	"	//\n"
+	"	for(int pair = 0; pair < numRayRigidPairsPerRay[i]; ++pair)\n"
+	"	{\n"
+	"		int rayRigidPairIndex = pair + firstRayRigidPairIndexPerRay[i];\n"
+	"		int b = rayRigidPairs[rayRigidPairIndex].y;\n"
+	"		\n"
+	"		if (hitResults[i].m_hitResult2 == b) continue;\n"
+	"		\n"
+	"		Body body = bodies[b];\n"
+	"		Collidable rigidCollidable = collidables[body.m_collidableIdx];\n"
+	"		\n"
+	"		float4 pos = body.m_pos;\n"
+	"		float4 orn = body.m_quat;\n"
+	"		\n"
+	"		if (rigidCollidable.m_shapeType == SHAPE_CONVEX_HULL)\n"
+	"		{\n"
+	"			float4 invPos = (float4)(0,0,0,0);\n"
+	"			float4 invOrn = (float4)(0,0,0,0);\n"
+	"			float4 rayFromLocal = (float4)(0,0,0,0);\n"
+	"			float4 rayToLocal = (float4)(0,0,0,0);\n"
+	"			invOrn = qtInvert(orn);\n"
+	"			invPos = qtRotate(invOrn, -pos);\n"
+	"			rayFromLocal = qtRotate( invOrn, rayFrom ) + invPos;\n"
+	"			rayToLocal = qtRotate( invOrn, rayTo) + invPos;\n"
+	"			rayFromLocal.w = 0.f;\n"
+	"			rayToLocal.w = 0.f;\n"
+	"			int numFaces = convexShapes[rigidCollidable.m_shapeIndex].m_numFaces;\n"
+	"			int faceOffset = convexShapes[rigidCollidable.m_shapeIndex].m_faceOffset;\n"
+	"			\n"
+	"			if (numFaces && rayConvex(rayFromLocal, rayToLocal, numFaces, faceOffset,faces, &hitFraction, &hitNormal))\n"
+	"			{\n"
+	"				hitBodyIndex = b;\n"
+	"				hitPoint = setInterpolate3(rayFrom, rayTo, hitFraction);\n"
+	"			}\n"
+	"		}\n"
+	"		\n"
+	"		if (rigidCollidable.m_shapeType == SHAPE_SPHERE)\n"
+	"		{\n"
+	"			float radius = rigidCollidable.m_radius;\n"
+	"		\n"
+	"			if (sphere_intersect(pos, radius, rayFrom, rayTo, &hitFraction))\n"
+	"			{\n"
+	"				hitBodyIndex = b;\n"
+	"				hitPoint = setInterpolate3(rayFrom, rayTo, hitFraction);\n"
+	"				hitNormal = (float4) (hitPoint - bodies[b].m_pos);\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"	\n"
+	"	if (hitBodyIndex >= 0)\n"
+	"	{\n"
+	"		hitResults[i].m_hitFraction = hitFraction;\n"
+	"		hitResults[i].m_hitPoint = hitPoint;\n"
+	"		hitResults[i].m_hitNormal = normalize(hitNormal);\n"
+	"		hitResults[i].m_hitResult0 = hitBodyIndex;\n"
+	"	}\n"
+	"	\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuConstraint4.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuConstraint4.h
index c7478f54a1b..89c0142ab37 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuConstraint4.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuConstraint4.h
@@ -5,14 +5,13 @@
 
 #include "Bullet3Dynamics/shared/b3ContactConstraint4.h"
 
-
-B3_ATTRIBUTE_ALIGNED16(struct) b3GpuConstraint4 : public b3ContactConstraint4
+B3_ATTRIBUTE_ALIGNED16(struct)
+b3GpuConstraint4 : public b3ContactConstraint4
 {
 	B3_DECLARE_ALIGNED_ALLOCATOR();
 
-	inline	void setFrictionCoeff(float value) { m_linear[3] = value; }
-	inline	float getFrictionCoeff() const { return m_linear[3]; }
+	inline void setFrictionCoeff(float value) { m_linear[3] = value; }
+	inline float getFrictionCoeff() const { return m_linear[3]; }
 };
 
-#endif //B3_CONSTRAINT4_h
-
+#endif  //B3_CONSTRAINT4_h
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuGenericConstraint.cpp b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuGenericConstraint.cpp
index af687b54e9b..a271090af4b 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuGenericConstraint.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuGenericConstraint.cpp
@@ -19,11 +19,11 @@ subject to the following restrictions:
 #include <new>
 #include "Bullet3Common/b3Transform.h"
 
-void b3GpuGenericConstraint::getInfo1 (unsigned int* info,const b3RigidBodyData* bodies)
+void b3GpuGenericConstraint::getInfo1(unsigned int* info, const b3RigidBodyData* bodies)
 {
 	switch (m_constraintType)
 	{
-	case B3_GPU_POINT2POINT_CONSTRAINT_TYPE:
+		case B3_GPU_POINT2POINT_CONSTRAINT_TYPE:
 		{
 			*info = 3;
 			break;
@@ -35,7 +35,7 @@ void b3GpuGenericConstraint::getInfo1 (unsigned int* info,const b3RigidBodyData*
 	};
 }
 
-void getInfo2Point2Point(b3GpuGenericConstraint* constraint, b3GpuConstraintInfo2* info,  const b3RigidBodyData* bodies)
+void getInfo2Point2Point(b3GpuGenericConstraint* constraint, b3GpuConstraintInfo2* info, const b3RigidBodyData* bodies)
 {
 	b3Transform trA;
 	trA.setIdentity();
@@ -47,54 +47,52 @@ void getInfo2Point2Point(b3GpuGenericConstraint* constraint, b3GpuConstraintInfo
 	trB.setOrigin(bodies[constraint->m_rbB].m_pos);
 	trB.setRotation(bodies[constraint->m_rbB].m_quat);
 
-		// anchor points in global coordinates with respect to body PORs.
-   
-    // set jacobian
-    info->m_J1linearAxis[0] = 1;
-	info->m_J1linearAxis[info->rowskip+1] = 1;
-	info->m_J1linearAxis[2*info->rowskip+2] = 1;
+	// anchor points in global coordinates with respect to body PORs.
 
-	b3Vector3 a1 = trA.getBasis()*constraint->getPivotInA();
+	// set jacobian
+	info->m_J1linearAxis[0] = 1;
+	info->m_J1linearAxis[info->rowskip + 1] = 1;
+	info->m_J1linearAxis[2 * info->rowskip + 2] = 1;
+
+	b3Vector3 a1 = trA.getBasis() * constraint->getPivotInA();
 	//b3Vector3 a1a = b3QuatRotate(trA.getRotation(),constraint->getPivotInA());
 
 	{
 		b3Vector3* angular0 = (b3Vector3*)(info->m_J1angularAxis);
-		b3Vector3* angular1 = (b3Vector3*)(info->m_J1angularAxis+info->rowskip);
-		b3Vector3* angular2 = (b3Vector3*)(info->m_J1angularAxis+2*info->rowskip);
+		b3Vector3* angular1 = (b3Vector3*)(info->m_J1angularAxis + info->rowskip);
+		b3Vector3* angular2 = (b3Vector3*)(info->m_J1angularAxis + 2 * info->rowskip);
 		b3Vector3 a1neg = -a1;
-		a1neg.getSkewSymmetricMatrix(angular0,angular1,angular2);
+		a1neg.getSkewSymmetricMatrix(angular0, angular1, angular2);
 	}
-    
+
 	if (info->m_J2linearAxis)
 	{
 		info->m_J2linearAxis[0] = -1;
-		info->m_J2linearAxis[info->rowskip+1] = -1;
-		info->m_J2linearAxis[2*info->rowskip+2] = -1;
+		info->m_J2linearAxis[info->rowskip + 1] = -1;
+		info->m_J2linearAxis[2 * info->rowskip + 2] = -1;
 	}
-	
-	b3Vector3 a2 = trB.getBasis()*constraint->getPivotInB();
-   
+
+	b3Vector3 a2 = trB.getBasis() * constraint->getPivotInB();
+
 	{
-	//	b3Vector3 a2n = -a2;
+		//	b3Vector3 a2n = -a2;
 		b3Vector3* angular0 = (b3Vector3*)(info->m_J2angularAxis);
-		b3Vector3* angular1 = (b3Vector3*)(info->m_J2angularAxis+info->rowskip);
-		b3Vector3* angular2 = (b3Vector3*)(info->m_J2angularAxis+2*info->rowskip);
-		a2.getSkewSymmetricMatrix(angular0,angular1,angular2);
+		b3Vector3* angular1 = (b3Vector3*)(info->m_J2angularAxis + info->rowskip);
+		b3Vector3* angular2 = (b3Vector3*)(info->m_J2angularAxis + 2 * info->rowskip);
+		a2.getSkewSymmetricMatrix(angular0, angular1, angular2);
 	}
-    
 
-
-    // set right hand side
-//	b3Scalar currERP = (m_flags & B3_P2P_FLAGS_ERP) ? m_erp : info->erp;
+	// set right hand side
+	//	b3Scalar currERP = (m_flags & B3_P2P_FLAGS_ERP) ? m_erp : info->erp;
 	b3Scalar currERP = info->erp;
 
 	b3Scalar k = info->fps * currERP;
-    int j;
-	for (j=0; j<3; j++)
-    {
-        info->m_constraintError[j*info->rowskip] = k * (a2[j] + trB.getOrigin()[j] - a1[j] - trA.getOrigin()[j]);
+	int j;
+	for (j = 0; j < 3; j++)
+	{
+		info->m_constraintError[j * info->rowskip] = k * (a2[j] + trB.getOrigin()[j] - a1[j] - trA.getOrigin()[j]);
 		//printf("info->m_constraintError[%d]=%f\n",j,info->m_constraintError[j]);
-    }
+	}
 #if 0
 	if(m_flags & B3_P2P_FLAGS_CFM)
 	{
@@ -117,21 +115,20 @@ void getInfo2Point2Point(b3GpuGenericConstraint* constraint, b3GpuConstraintInfo
 	}
 	info->m_damping = m_setting.m_damping;
 #endif
-
 }
 
-void b3GpuGenericConstraint::getInfo2 (b3GpuConstraintInfo2* info,  const b3RigidBodyData* bodies)
+void b3GpuGenericConstraint::getInfo2(b3GpuConstraintInfo2* info, const b3RigidBodyData* bodies)
 {
 	switch (m_constraintType)
 	{
-	case B3_GPU_POINT2POINT_CONSTRAINT_TYPE:
+		case B3_GPU_POINT2POINT_CONSTRAINT_TYPE:
 		{
-			getInfo2Point2Point(this,info,bodies);
+			getInfo2Point2Point(this, info, bodies);
 			break;
 		};
 		default:
-			{
-				b3Assert(0);
-			}
+		{
+			b3Assert(0);
+		}
 	};
 }
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuGenericConstraint.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuGenericConstraint.h
index 14b3ba7fec7..1f163ba7d5b 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuGenericConstraint.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuGenericConstraint.h
@@ -20,37 +20,35 @@ subject to the following restrictions:
 struct b3RigidBodyData;
 enum B3_CONSTRAINT_FLAGS
 {
-	B3_CONSTRAINT_FLAG_ENABLED=1,
+	B3_CONSTRAINT_FLAG_ENABLED = 1,
 };
 
 enum b3GpuGenericConstraintType
 {
-	B3_GPU_POINT2POINT_CONSTRAINT_TYPE=3,
-	B3_GPU_FIXED_CONSTRAINT_TYPE=4,
-//	B3_HINGE_CONSTRAINT_TYPE,
-//	B3_CONETWIST_CONSTRAINT_TYPE,
-//	B3_D6_CONSTRAINT_TYPE,
-//	B3_SLIDER_CONSTRAINT_TYPE,
-//	B3_CONTACT_CONSTRAINT_TYPE,
-//	B3_D6_SPRING_CONSTRAINT_TYPE,
-//	B3_GEAR_CONSTRAINT_TYPE,
-	
+	B3_GPU_POINT2POINT_CONSTRAINT_TYPE = 3,
+	B3_GPU_FIXED_CONSTRAINT_TYPE = 4,
+	//	B3_HINGE_CONSTRAINT_TYPE,
+	//	B3_CONETWIST_CONSTRAINT_TYPE,
+	//	B3_D6_CONSTRAINT_TYPE,
+	//	B3_SLIDER_CONSTRAINT_TYPE,
+	//	B3_CONTACT_CONSTRAINT_TYPE,
+	//	B3_D6_SPRING_CONSTRAINT_TYPE,
+	//	B3_GEAR_CONSTRAINT_TYPE,
+
 	B3_GPU_MAX_CONSTRAINT_TYPE
 };
 
-
-
-struct b3GpuConstraintInfo2 
+struct b3GpuConstraintInfo2
 {
 	// integrator parameters: frames per second (1/stepsize), default error
 	// reduction parameter (0..1).
-	b3Scalar fps,erp;
+	b3Scalar fps, erp;
 
 	// for the first and second body, pointers to two (linear and angular)
 	// n*3 jacobian sub matrices, stored by rows. these matrices will have
 	// been initialized to 0 on entry. if the second body is zero then the
 	// J2xx pointers may be 0.
-	b3Scalar *m_J1linearAxis,*m_J1angularAxis,*m_J2linearAxis,*m_J2angularAxis;
+	b3Scalar *m_J1linearAxis, *m_J1angularAxis, *m_J2linearAxis, *m_J2angularAxis;
 
 	// elements to jump from one row to the next in J's
 	int rowskip;
@@ -58,44 +56,44 @@ struct b3GpuConstraintInfo2
 	// right hand sides of the equation J*v = c + cfm * lambda. cfm is the
 	// "constraint force mixing" vector. c is set to zero on entry, cfm is
 	// set to a constant value (typically very small or zero) value on entry.
-	b3Scalar *m_constraintError,*cfm;
+	b3Scalar *m_constraintError, *cfm;
 
 	// lo and hi limits for variables (set to -/+ infinity on entry).
-	b3Scalar *m_lowerLimit,*m_upperLimit;
+	b3Scalar *m_lowerLimit, *m_upperLimit;
 
 	// findex vector for variables. see the LCP solver interface for a
 	// description of what this does. this is set to -1 on entry.
 	// note that the returned indexes are relative to the first index of
 	// the constraint.
-	int *findex;
+	int* findex;
 	// number of solver iterations
 	int m_numIterations;
 
 	//damping of the velocity
-	b3Scalar	m_damping;
+	b3Scalar m_damping;
 };
 
-
-B3_ATTRIBUTE_ALIGNED16(struct) b3GpuGenericConstraint
+B3_ATTRIBUTE_ALIGNED16(struct)
+b3GpuGenericConstraint
 {
-	int				m_constraintType;
-	int				m_rbA;
-	int				m_rbB;
-	float			m_breakingImpulseThreshold;
+	int m_constraintType;
+	int m_rbA;
+	int m_rbB;
+	float m_breakingImpulseThreshold;
 
 	b3Vector3 m_pivotInA;
 	b3Vector3 m_pivotInB;
 	b3Quaternion m_relTargetAB;
 
-	int	m_flags;
+	int m_flags;
 	int m_uid;
 	int m_padding[2];
 
-	int	getRigidBodyA() const
+	int getRigidBodyA() const
 	{
 		return m_rbA;
 	}
-	int	getRigidBodyB() const
+	int getRigidBodyB() const
 	{
 		return m_rbB;
 	}
@@ -121,12 +119,10 @@ B3_ATTRIBUTE_ALIGNED16(struct) b3GpuGenericConstraint
 	}
 
 	///internal method used by the constraint solver, don't use them directly
-	void getInfo1 (unsigned int* info,const b3RigidBodyData* bodies);
+	void getInfo1(unsigned int* info, const b3RigidBodyData* bodies);
 
 	///internal method used by the constraint solver, don't use them directly
-	void getInfo2 (b3GpuConstraintInfo2* info,  const b3RigidBodyData* bodies);
-
-
+	void getInfo2(b3GpuConstraintInfo2 * info, const b3RigidBodyData* bodies);
 };
 
-#endif //B3_GPU_GENERIC_CONSTRAINT_H
\ No newline at end of file
+#endif  //B3_GPU_GENERIC_CONSTRAINT_H
\ No newline at end of file
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuJacobiContactSolver.cpp b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuJacobiContactSolver.cpp
index 179dfc4f26e..089fb1f6a60 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuJacobiContactSolver.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuJacobiContactSolver.cpp
@@ -2,7 +2,7 @@
 #include "b3GpuJacobiContactSolver.h"
 #include "Bullet3Collision/NarrowPhaseCollision/b3Contact4.h"
 #include "Bullet3Common/b3AlignedObjectArray.h"
-#include "Bullet3OpenCL/ParallelPrimitives/b3FillCL.h" //b3Int2
+#include "Bullet3OpenCL/ParallelPrimitives/b3FillCL.h"  //b3Int2
 class b3Vector3;
 #include "Bullet3OpenCL/ParallelPrimitives/b3RadixSort32CL.h"
 #include "Bullet3OpenCL/ParallelPrimitives/b3PrefixScanCL.h"
@@ -15,89 +15,78 @@ class b3Vector3;
 #include "Bullet3Common/shared/b3Int4.h"
 #define SOLVER_UTILS_KERNEL_PATH "src/Bullet3OpenCL/RigidBody/kernels/solverUtils.cl"
 
-
 struct b3GpuJacobiSolverInternalData
 {
-		//btRadixSort32CL*	m_sort32;
-		//btBoundSearchCL*	m_search;
-		b3PrefixScanCL*	m_scan;
+	//btRadixSort32CL*	m_sort32;
+	//btBoundSearchCL*	m_search;
+	b3PrefixScanCL* m_scan;
 
-		b3OpenCLArray<unsigned int>* m_bodyCount;
-		b3OpenCLArray<b3Int2>*		m_contactConstraintOffsets;
-		b3OpenCLArray<unsigned int>* m_offsetSplitBodies;
+	b3OpenCLArray<unsigned int>* m_bodyCount;
+	b3OpenCLArray<b3Int2>* m_contactConstraintOffsets;
+	b3OpenCLArray<unsigned int>* m_offsetSplitBodies;
 
-		b3OpenCLArray<b3Vector3>*	m_deltaLinearVelocities;
-		b3OpenCLArray<b3Vector3>*	m_deltaAngularVelocities;
+	b3OpenCLArray<b3Vector3>* m_deltaLinearVelocities;
+	b3OpenCLArray<b3Vector3>* m_deltaAngularVelocities;
 
-		b3AlignedObjectArray<b3Vector3>	m_deltaLinearVelocitiesCPU;
-		b3AlignedObjectArray<b3Vector3>	m_deltaAngularVelocitiesCPU;
-
-
-
-		b3OpenCLArray<b3GpuConstraint4>* m_contactConstraints;
-
-		b3FillCL*	m_filler;
-		
-
-		cl_kernel	m_countBodiesKernel;
-		cl_kernel	m_contactToConstraintSplitKernel;
-		cl_kernel	m_clearVelocitiesKernel;
-		cl_kernel	m_averageVelocitiesKernel;
-		cl_kernel	m_updateBodyVelocitiesKernel;
-		cl_kernel	m_solveContactKernel;
-		cl_kernel	m_solveFrictionKernel;
+	b3AlignedObjectArray<b3Vector3> m_deltaLinearVelocitiesCPU;
+	b3AlignedObjectArray<b3Vector3> m_deltaAngularVelocitiesCPU;
 
+	b3OpenCLArray<b3GpuConstraint4>* m_contactConstraints;
 
+	b3FillCL* m_filler;
 
+	cl_kernel m_countBodiesKernel;
+	cl_kernel m_contactToConstraintSplitKernel;
+	cl_kernel m_clearVelocitiesKernel;
+	cl_kernel m_averageVelocitiesKernel;
+	cl_kernel m_updateBodyVelocitiesKernel;
+	cl_kernel m_solveContactKernel;
+	cl_kernel m_solveFrictionKernel;
 };
 
-
 b3GpuJacobiContactSolver::b3GpuJacobiContactSolver(cl_context ctx, cl_device_id device, cl_command_queue queue, int pairCapacity)
-	:m_context(ctx),
-	m_device(device),
-	m_queue(queue)
+	: m_context(ctx),
+	  m_device(device),
+	  m_queue(queue)
 {
 	m_data = new b3GpuJacobiSolverInternalData;
-	m_data->m_scan = new b3PrefixScanCL(m_context,m_device,m_queue);
-	m_data->m_bodyCount = new b3OpenCLArray<unsigned int>(m_context,m_queue);
-	m_data->m_filler = new b3FillCL(m_context,m_device,m_queue);
-	m_data->m_contactConstraintOffsets = new b3OpenCLArray<b3Int2>(m_context,m_queue);
-	m_data->m_offsetSplitBodies = new b3OpenCLArray<unsigned int>(m_context,m_queue);
-	m_data->m_contactConstraints = new b3OpenCLArray<b3GpuConstraint4>(m_context,m_queue);
-	m_data->m_deltaLinearVelocities = new b3OpenCLArray<b3Vector3>(m_context,m_queue);
-	m_data->m_deltaAngularVelocities = new b3OpenCLArray<b3Vector3>(m_context,m_queue);
+	m_data->m_scan = new b3PrefixScanCL(m_context, m_device, m_queue);
+	m_data->m_bodyCount = new b3OpenCLArray<unsigned int>(m_context, m_queue);
+	m_data->m_filler = new b3FillCL(m_context, m_device, m_queue);
+	m_data->m_contactConstraintOffsets = new b3OpenCLArray<b3Int2>(m_context, m_queue);
+	m_data->m_offsetSplitBodies = new b3OpenCLArray<unsigned int>(m_context, m_queue);
+	m_data->m_contactConstraints = new b3OpenCLArray<b3GpuConstraint4>(m_context, m_queue);
+	m_data->m_deltaLinearVelocities = new b3OpenCLArray<b3Vector3>(m_context, m_queue);
+	m_data->m_deltaAngularVelocities = new b3OpenCLArray<b3Vector3>(m_context, m_queue);
 
 	cl_int pErrNum;
-	const char* additionalMacros="";
+	const char* additionalMacros = "";
 	const char* solverUtilsSource = solverUtilsCL;
 	{
-		cl_program solverUtilsProg= b3OpenCLUtils::compileCLProgramFromString( ctx, device, solverUtilsSource, &pErrNum,additionalMacros, SOLVER_UTILS_KERNEL_PATH);
+		cl_program solverUtilsProg = b3OpenCLUtils::compileCLProgramFromString(ctx, device, solverUtilsSource, &pErrNum, additionalMacros, SOLVER_UTILS_KERNEL_PATH);
 		b3Assert(solverUtilsProg);
-		m_data->m_countBodiesKernel =  b3OpenCLUtils::compileCLKernelFromString( ctx, device, solverUtilsSource, "CountBodiesKernel", &pErrNum, solverUtilsProg,additionalMacros );
+		m_data->m_countBodiesKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solverUtilsSource, "CountBodiesKernel", &pErrNum, solverUtilsProg, additionalMacros);
 		b3Assert(m_data->m_countBodiesKernel);
 
-		m_data->m_contactToConstraintSplitKernel  = b3OpenCLUtils::compileCLKernelFromString( ctx, device, solverUtilsSource, "ContactToConstraintSplitKernel", &pErrNum, solverUtilsProg,additionalMacros );
+		m_data->m_contactToConstraintSplitKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solverUtilsSource, "ContactToConstraintSplitKernel", &pErrNum, solverUtilsProg, additionalMacros);
 		b3Assert(m_data->m_contactToConstraintSplitKernel);
-		m_data->m_clearVelocitiesKernel  = b3OpenCLUtils::compileCLKernelFromString( ctx, device, solverUtilsSource, "ClearVelocitiesKernel", &pErrNum, solverUtilsProg,additionalMacros );
+		m_data->m_clearVelocitiesKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solverUtilsSource, "ClearVelocitiesKernel", &pErrNum, solverUtilsProg, additionalMacros);
 		b3Assert(m_data->m_clearVelocitiesKernel);
 
-		m_data->m_averageVelocitiesKernel  = b3OpenCLUtils::compileCLKernelFromString( ctx, device, solverUtilsSource, "AverageVelocitiesKernel", &pErrNum, solverUtilsProg,additionalMacros );
+		m_data->m_averageVelocitiesKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solverUtilsSource, "AverageVelocitiesKernel", &pErrNum, solverUtilsProg, additionalMacros);
 		b3Assert(m_data->m_averageVelocitiesKernel);
 
-		m_data->m_updateBodyVelocitiesKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, solverUtilsSource, "UpdateBodyVelocitiesKernel", &pErrNum, solverUtilsProg,additionalMacros );
+		m_data->m_updateBodyVelocitiesKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solverUtilsSource, "UpdateBodyVelocitiesKernel", &pErrNum, solverUtilsProg, additionalMacros);
 		b3Assert(m_data->m_updateBodyVelocitiesKernel);
 
-		
-		m_data->m_solveContactKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, solverUtilsSource, "SolveContactJacobiKernel", &pErrNum, solverUtilsProg,additionalMacros );
-		b3Assert(m_data->m_solveContactKernel );
+		m_data->m_solveContactKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solverUtilsSource, "SolveContactJacobiKernel", &pErrNum, solverUtilsProg, additionalMacros);
+		b3Assert(m_data->m_solveContactKernel);
 
-		m_data->m_solveFrictionKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, solverUtilsSource, "SolveFrictionJacobiKernel", &pErrNum, solverUtilsProg,additionalMacros );
+		m_data->m_solveFrictionKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solverUtilsSource, "SolveFrictionJacobiKernel", &pErrNum, solverUtilsProg, additionalMacros);
 		b3Assert(m_data->m_solveFrictionKernel);
 	}
-
 }
 
-
 b3GpuJacobiContactSolver::~b3GpuJacobiContactSolver()
 {
 	clReleaseKernel(m_data->m_solveContactKernel);
@@ -106,7 +95,7 @@ b3GpuJacobiContactSolver::~b3GpuJacobiContactSolver()
 	clReleaseKernel(m_data->m_contactToConstraintSplitKernel);
 	clReleaseKernel(m_data->m_averageVelocitiesKernel);
 	clReleaseKernel(m_data->m_updateBodyVelocitiesKernel);
-	clReleaseKernel(m_data->m_clearVelocitiesKernel );
+	clReleaseKernel(m_data->m_clearVelocitiesKernel);
 
 	delete m_data->m_deltaLinearVelocities;
 	delete m_data->m_deltaAngularVelocities;
@@ -119,80 +108,70 @@ b3GpuJacobiContactSolver::~b3GpuJacobiContactSolver()
 	delete m_data;
 }
 
-
-
 b3Vector3 make_float4(float v)
 {
-	return b3MakeVector3 (v,v,v);
+	return b3MakeVector3(v, v, v);
 }
 
-b3Vector4 make_float4(float x,float y, float z, float w)
+b3Vector4 make_float4(float x, float y, float z, float w)
 {
-	return b3MakeVector4 (x,y,z,w);
+	return b3MakeVector4(x, y, z, w);
 }
 
-
-	static
-	inline
-	float calcRelVel(const b3Vector3& l0, const b3Vector3& l1, const b3Vector3& a0, const b3Vector3& a1, 
-					 const b3Vector3& linVel0, const b3Vector3& angVel0, const b3Vector3& linVel1, const b3Vector3& angVel1)
-	{
-		return b3Dot(l0, linVel0) + b3Dot(a0, angVel0) + b3Dot(l1, linVel1) + b3Dot(a1, angVel1);
-	}
-
-
-	static
-	inline
-	void setLinearAndAngular(const b3Vector3& n, const b3Vector3& r0, const b3Vector3& r1,
-							 b3Vector3& linear, b3Vector3& angular0, b3Vector3& angular1)
-	{
-		linear = n;
-		angular0 = b3Cross(r0, n);
-		angular1 = -b3Cross(r1, n);
-	}
-
-
-static __inline void solveContact(b3GpuConstraint4& cs, 
-	const b3Vector3& posA, const b3Vector3& linVelARO, const b3Vector3& angVelARO, float invMassA, const b3Matrix3x3& invInertiaA,
-	const b3Vector3& posB, const b3Vector3& linVelBRO, const b3Vector3& angVelBRO, float invMassB, const b3Matrix3x3& invInertiaB, 
-	float maxRambdaDt[4], float minRambdaDt[4], b3Vector3& dLinVelA, b3Vector3& dAngVelA, b3Vector3& dLinVelB, b3Vector3& dAngVelB)
+static inline float calcRelVel(const b3Vector3& l0, const b3Vector3& l1, const b3Vector3& a0, const b3Vector3& a1,
+							   const b3Vector3& linVel0, const b3Vector3& angVel0, const b3Vector3& linVel1, const b3Vector3& angVel1)
 {
+	return b3Dot(l0, linVel0) + b3Dot(a0, angVel0) + b3Dot(l1, linVel1) + b3Dot(a1, angVel1);
+}
 
+static inline void setLinearAndAngular(const b3Vector3& n, const b3Vector3& r0, const b3Vector3& r1,
+									   b3Vector3& linear, b3Vector3& angular0, b3Vector3& angular1)
+{
+	linear = n;
+	angular0 = b3Cross(r0, n);
+	angular1 = -b3Cross(r1, n);
+}
 
-	for(int ic=0; ic<4; ic++)
+static __inline void solveContact(b3GpuConstraint4& cs,
+								  const b3Vector3& posA, const b3Vector3& linVelARO, const b3Vector3& angVelARO, float invMassA, const b3Matrix3x3& invInertiaA,
+								  const b3Vector3& posB, const b3Vector3& linVelBRO, const b3Vector3& angVelBRO, float invMassB, const b3Matrix3x3& invInertiaB,
+								  float maxRambdaDt[4], float minRambdaDt[4], b3Vector3& dLinVelA, b3Vector3& dAngVelA, b3Vector3& dLinVelB, b3Vector3& dAngVelB)
+{
+	for (int ic = 0; ic < 4; ic++)
 	{
 		//	dont necessary because this makes change to 0
-		if( cs.m_jacCoeffInv[ic] == 0.f ) continue;
+		if (cs.m_jacCoeffInv[ic] == 0.f) continue;
 
 		{
 			b3Vector3 angular0, angular1, linear;
 			b3Vector3 r0 = cs.m_worldPos[ic] - (b3Vector3&)posA;
 			b3Vector3 r1 = cs.m_worldPos[ic] - (b3Vector3&)posB;
-			setLinearAndAngular( (const b3Vector3 &)cs.m_linear, (const b3Vector3 &)r0, (const b3Vector3 &)r1, linear, angular0, angular1 );
+			setLinearAndAngular((const b3Vector3&)cs.m_linear, (const b3Vector3&)r0, (const b3Vector3&)r1, linear, angular0, angular1);
 
-			float rambdaDt = calcRelVel((const b3Vector3 &)cs.m_linear,(const b3Vector3 &) -cs.m_linear, angular0, angular1,
-				linVelARO+dLinVelA, angVelARO+dAngVelA, linVelBRO+dLinVelB, angVelBRO+dAngVelB ) + cs.m_b[ic];
+			float rambdaDt = calcRelVel((const b3Vector3&)cs.m_linear, (const b3Vector3&)-cs.m_linear, angular0, angular1,
+										linVelARO + dLinVelA, angVelARO + dAngVelA, linVelBRO + dLinVelB, angVelBRO + dAngVelB) +
+							 cs.m_b[ic];
 			rambdaDt *= cs.m_jacCoeffInv[ic];
 
 			{
 				float prevSum = cs.m_appliedRambdaDt[ic];
 				float updated = prevSum;
 				updated += rambdaDt;
-				updated = b3Max( updated, minRambdaDt[ic] );
-				updated = b3Min( updated, maxRambdaDt[ic] );
+				updated = b3Max(updated, minRambdaDt[ic]);
+				updated = b3Min(updated, maxRambdaDt[ic]);
 				rambdaDt = updated - prevSum;
 				cs.m_appliedRambdaDt[ic] = updated;
 			}
 
-			b3Vector3 linImp0 = invMassA*linear*rambdaDt;
-			b3Vector3 linImp1 = invMassB*(-linear)*rambdaDt;
-			b3Vector3 angImp0 = (invInertiaA* angular0)*rambdaDt;
-			b3Vector3 angImp1 = (invInertiaB* angular1)*rambdaDt;
+			b3Vector3 linImp0 = invMassA * linear * rambdaDt;
+			b3Vector3 linImp1 = invMassB * (-linear) * rambdaDt;
+			b3Vector3 angImp0 = (invInertiaA * angular0) * rambdaDt;
+			b3Vector3 angImp1 = (invInertiaB * angular1) * rambdaDt;
 #ifdef _WIN32
-            b3Assert(_finite(linImp0.getX()));
+			b3Assert(_finite(linImp0.getX()));
 			b3Assert(_finite(linImp1.getX()));
 #endif
-			
+
 			if (invMassA)
 			{
 				dLinVelA += linImp0;
@@ -207,43 +186,42 @@ static __inline void solveContact(b3GpuConstraint4& cs,
 	}
 }
 
-
-
 void solveContact3(b3GpuConstraint4* cs,
-			b3Vector3* posAPtr, b3Vector3* linVelA, b3Vector3* angVelA, float invMassA, const b3Matrix3x3& invInertiaA,
-			b3Vector3* posBPtr, b3Vector3* linVelB, b3Vector3* angVelB, float invMassB, const b3Matrix3x3& invInertiaB,
-			b3Vector3* dLinVelA, b3Vector3* dAngVelA, b3Vector3* dLinVelB, b3Vector3* dAngVelB)
+				   b3Vector3* posAPtr, b3Vector3* linVelA, b3Vector3* angVelA, float invMassA, const b3Matrix3x3& invInertiaA,
+				   b3Vector3* posBPtr, b3Vector3* linVelB, b3Vector3* angVelB, float invMassB, const b3Matrix3x3& invInertiaB,
+				   b3Vector3* dLinVelA, b3Vector3* dAngVelA, b3Vector3* dLinVelB, b3Vector3* dAngVelB)
 {
 	float minRambdaDt = 0;
 	float maxRambdaDt = FLT_MAX;
 
-	for(int ic=0; ic<4; ic++)
+	for (int ic = 0; ic < 4; ic++)
 	{
-		if( cs->m_jacCoeffInv[ic] == 0.f ) continue;
+		if (cs->m_jacCoeffInv[ic] == 0.f) continue;
 
 		b3Vector3 angular0, angular1, linear;
 		b3Vector3 r0 = cs->m_worldPos[ic] - *posAPtr;
 		b3Vector3 r1 = cs->m_worldPos[ic] - *posBPtr;
-		setLinearAndAngular( cs->m_linear, r0, r1, linear, angular0, angular1 );
+		setLinearAndAngular(cs->m_linear, r0, r1, linear, angular0, angular1);
 
-		float rambdaDt = calcRelVel( cs->m_linear, -cs->m_linear, angular0, angular1, 
-			*linVelA+*dLinVelA, *angVelA+*dAngVelA, *linVelB+*dLinVelB, *angVelB+*dAngVelB ) + cs->m_b[ic];
+		float rambdaDt = calcRelVel(cs->m_linear, -cs->m_linear, angular0, angular1,
+									*linVelA + *dLinVelA, *angVelA + *dAngVelA, *linVelB + *dLinVelB, *angVelB + *dAngVelB) +
+						 cs->m_b[ic];
 		rambdaDt *= cs->m_jacCoeffInv[ic];
 
 		{
 			float prevSum = cs->m_appliedRambdaDt[ic];
 			float updated = prevSum;
 			updated += rambdaDt;
-			updated = b3Max( updated, minRambdaDt );
-			updated = b3Min( updated, maxRambdaDt );
+			updated = b3Max(updated, minRambdaDt);
+			updated = b3Min(updated, maxRambdaDt);
 			rambdaDt = updated - prevSum;
 			cs->m_appliedRambdaDt[ic] = updated;
 		}
 
-		b3Vector3 linImp0 = invMassA*linear*rambdaDt;
-		b3Vector3 linImp1 = invMassB*(-linear)*rambdaDt;
-		b3Vector3 angImp0 = (invInertiaA* angular0)*rambdaDt;
-		b3Vector3 angImp1 = (invInertiaB* angular1)*rambdaDt;
+		b3Vector3 linImp0 = invMassA * linear * rambdaDt;
+		b3Vector3 linImp1 = invMassB * (-linear) * rambdaDt;
+		b3Vector3 angImp0 = (invInertiaA * angular0) * rambdaDt;
+		b3Vector3 angImp1 = (invInertiaB * angular1) * rambdaDt;
 
 		if (invMassA)
 		{
@@ -258,58 +236,56 @@ void solveContact3(b3GpuConstraint4* cs,
 	}
 }
 
-
-static inline void solveFriction(b3GpuConstraint4& cs, 
-	const b3Vector3& posA, const b3Vector3& linVelARO, const b3Vector3& angVelARO, float invMassA, const b3Matrix3x3& invInertiaA,
-	const b3Vector3& posB, const b3Vector3& linVelBRO, const b3Vector3& angVelBRO, float invMassB, const b3Matrix3x3& invInertiaB, 
-	float maxRambdaDt[4], float minRambdaDt[4], b3Vector3& dLinVelA, b3Vector3& dAngVelA, b3Vector3& dLinVelB, b3Vector3& dAngVelB)
+static inline void solveFriction(b3GpuConstraint4& cs,
+								 const b3Vector3& posA, const b3Vector3& linVelARO, const b3Vector3& angVelARO, float invMassA, const b3Matrix3x3& invInertiaA,
+								 const b3Vector3& posB, const b3Vector3& linVelBRO, const b3Vector3& angVelBRO, float invMassB, const b3Matrix3x3& invInertiaB,
+								 float maxRambdaDt[4], float minRambdaDt[4], b3Vector3& dLinVelA, b3Vector3& dAngVelA, b3Vector3& dLinVelB, b3Vector3& dAngVelB)
 {
+	b3Vector3 linVelA = linVelARO + dLinVelA;
+	b3Vector3 linVelB = linVelBRO + dLinVelB;
+	b3Vector3 angVelA = angVelARO + dAngVelA;
+	b3Vector3 angVelB = angVelBRO + dAngVelB;
 
-	b3Vector3 linVelA = linVelARO+dLinVelA;
-	b3Vector3 linVelB = linVelBRO+dLinVelB;
-	b3Vector3 angVelA = angVelARO+dAngVelA;
-	b3Vector3 angVelB = angVelBRO+dAngVelB;
-
-	if( cs.m_fJacCoeffInv[0] == 0 && cs.m_fJacCoeffInv[0] == 0 ) return;
+	if (cs.m_fJacCoeffInv[0] == 0 && cs.m_fJacCoeffInv[0] == 0) return;
 	const b3Vector3& center = (const b3Vector3&)cs.m_center;
 
 	b3Vector3 n = -(const b3Vector3&)cs.m_linear;
 
 	b3Vector3 tangent[2];
-#if 1		
-	b3PlaneSpace1 (n, tangent[0],tangent[1]);
+#if 1
+	b3PlaneSpace1(n, tangent[0], tangent[1]);
 #else
-	b3Vector3 r = cs.m_worldPos[0]-center;
-	tangent[0] = cross3( n, r );
-	tangent[1] = cross3( tangent[0], n );
-	tangent[0] = normalize3( tangent[0] );
-	tangent[1] = normalize3( tangent[1] );
+	b3Vector3 r = cs.m_worldPos[0] - center;
+	tangent[0] = cross3(n, r);
+	tangent[1] = cross3(tangent[0], n);
+	tangent[0] = normalize3(tangent[0]);
+	tangent[1] = normalize3(tangent[1]);
 #endif
 
 	b3Vector3 angular0, angular1, linear;
 	b3Vector3 r0 = center - posA;
 	b3Vector3 r1 = center - posB;
-	for(int i=0; i<2; i++)
+	for (int i = 0; i < 2; i++)
 	{
-		setLinearAndAngular( tangent[i], r0, r1, linear, angular0, angular1 );
+		setLinearAndAngular(tangent[i], r0, r1, linear, angular0, angular1);
 		float rambdaDt = calcRelVel(linear, -linear, angular0, angular1,
-			linVelA, angVelA, linVelB, angVelB );
+									linVelA, angVelA, linVelB, angVelB);
 		rambdaDt *= cs.m_fJacCoeffInv[i];
 
-			{
-				float prevSum = cs.m_fAppliedRambdaDt[i];
-				float updated = prevSum;
-				updated += rambdaDt;
-				updated = b3Max( updated, minRambdaDt[i] );
-				updated = b3Min( updated, maxRambdaDt[i] );
-				rambdaDt = updated - prevSum;
-				cs.m_fAppliedRambdaDt[i] = updated;
-			}
+		{
+			float prevSum = cs.m_fAppliedRambdaDt[i];
+			float updated = prevSum;
+			updated += rambdaDt;
+			updated = b3Max(updated, minRambdaDt[i]);
+			updated = b3Min(updated, maxRambdaDt[i]);
+			rambdaDt = updated - prevSum;
+			cs.m_fAppliedRambdaDt[i] = updated;
+		}
 
-		b3Vector3 linImp0 = invMassA*linear*rambdaDt;
-		b3Vector3 linImp1 = invMassB*(-linear)*rambdaDt;
-		b3Vector3 angImp0 = (invInertiaA* angular0)*rambdaDt;
-		b3Vector3 angImp1 = (invInertiaB* angular1)*rambdaDt;
+		b3Vector3 linImp0 = invMassA * linear * rambdaDt;
+		b3Vector3 linImp1 = invMassB * (-linear) * rambdaDt;
+		b3Vector3 angImp0 = (invInertiaA * angular0) * rambdaDt;
+		b3Vector3 angImp1 = (invInertiaB * angular1) * rambdaDt;
 #ifdef _WIN32
 		b3Assert(_finite(linImp0.getX()));
 		b3Assert(_finite(linImp1.getX()));
@@ -326,65 +302,58 @@ static inline void solveFriction(b3GpuConstraint4& cs,
 		}
 	}
 
-	{	//	angular damping for point constraint
-		b3Vector3 ab = ( posB - posA ).normalized();
-		b3Vector3 ac = ( center - posA ).normalized();
-		if( b3Dot( ab, ac ) > 0.95f || (invMassA == 0.f || invMassB == 0.f))
+	{  //	angular damping for point constraint
+		b3Vector3 ab = (posB - posA).normalized();
+		b3Vector3 ac = (center - posA).normalized();
+		if (b3Dot(ab, ac) > 0.95f || (invMassA == 0.f || invMassB == 0.f))
 		{
-			float angNA = b3Dot( n, angVelA );
-			float angNB = b3Dot( n, angVelB );
+			float angNA = b3Dot(n, angVelA);
+			float angNB = b3Dot(n, angVelB);
 
 			if (invMassA)
-				dAngVelA -= (angNA*0.1f)*n;
+				dAngVelA -= (angNA * 0.1f) * n;
 			if (invMassB)
-				dAngVelB -= (angNB*0.1f)*n;
+				dAngVelB -= (angNB * 0.1f) * n;
 		}
 	}
-
 }
 
-
-
-
 float calcJacCoeff(const b3Vector3& linear0, const b3Vector3& linear1, const b3Vector3& angular0, const b3Vector3& angular1,
-					float invMass0, const b3Matrix3x3* invInertia0, float invMass1, const b3Matrix3x3* invInertia1, float countA, float countB)
+				   float invMass0, const b3Matrix3x3* invInertia0, float invMass1, const b3Matrix3x3* invInertia1, float countA, float countB)
 {
 	//	linear0,1 are normlized
-	float jmj0 = invMass0;//dot3F4(linear0, linear0)*invMass0;
-	
-	float jmj1 = b3Dot(mtMul3(angular0,*invInertia0), angular0);
-	float jmj2 = invMass1;//dot3F4(linear1, linear1)*invMass1;
-	float jmj3 = b3Dot(mtMul3(angular1,*invInertia1), angular1);
-	return -1.f/((jmj0+jmj1)*countA+(jmj2+jmj3)*countB);
-//	return -1.f/((jmj0+jmj1)+(jmj2+jmj3));
+	float jmj0 = invMass0;  //dot3F4(linear0, linear0)*invMass0;
 
+	float jmj1 = b3Dot(mtMul3(angular0, *invInertia0), angular0);
+	float jmj2 = invMass1;  //dot3F4(linear1, linear1)*invMass1;
+	float jmj3 = b3Dot(mtMul3(angular1, *invInertia1), angular1);
+	return -1.f / ((jmj0 + jmj1) * countA + (jmj2 + jmj3) * countB);
+	//	return -1.f/((jmj0+jmj1)+(jmj2+jmj3));
 }
 
-
-void setConstraint4( const b3Vector3& posA, const b3Vector3& linVelA, const b3Vector3& angVelA, float invMassA, const b3Matrix3x3& invInertiaA,
-	const b3Vector3& posB, const b3Vector3& linVelB, const b3Vector3& angVelB, float invMassB, const b3Matrix3x3& invInertiaB, 
-	 b3Contact4* src, float dt, float positionDrift, float positionConstraintCoeff, float countA, float countB,
-	b3GpuConstraint4* dstC )
+void setConstraint4(const b3Vector3& posA, const b3Vector3& linVelA, const b3Vector3& angVelA, float invMassA, const b3Matrix3x3& invInertiaA,
+					const b3Vector3& posB, const b3Vector3& linVelB, const b3Vector3& angVelB, float invMassB, const b3Matrix3x3& invInertiaB,
+					b3Contact4* src, float dt, float positionDrift, float positionConstraintCoeff, float countA, float countB,
+					b3GpuConstraint4* dstC)
 {
 	dstC->m_bodyA = abs(src->m_bodyAPtrAndSignBit);
 	dstC->m_bodyB = abs(src->m_bodyBPtrAndSignBit);
 
-	float dtInv = 1.f/dt;
-	for(int ic=0; ic<4; ic++)
+	float dtInv = 1.f / dt;
+	for (int ic = 0; ic < 4; ic++)
 	{
 		dstC->m_appliedRambdaDt[ic] = 0.f;
 	}
 	dstC->m_fJacCoeffInv[0] = dstC->m_fJacCoeffInv[1] = 0.f;
 
-
 	dstC->m_linear = src->m_worldNormalOnB;
-	dstC->m_linear[3] = 0.7f ;//src->getFrictionCoeff() );
-	for(int ic=0; ic<4; ic++)
+	dstC->m_linear[3] = 0.7f;  //src->getFrictionCoeff() );
+	for (int ic = 0; ic < 4; ic++)
 	{
 		b3Vector3 r0 = src->m_worldPosB[ic] - posA;
 		b3Vector3 r1 = src->m_worldPosB[ic] - posB;
 
-		if( ic >= src->m_worldNormalOnB[3] )//npoints
+		if (ic >= src->m_worldNormalOnB[3])  //npoints
 		{
 			dstC->m_jacCoeffInv[ic] = 0.f;
 			continue;
@@ -396,53 +365,53 @@ void setConstraint4( const b3Vector3& posA, const b3Vector3& linVelA, const b3Ve
 			setLinearAndAngular(src->m_worldNormalOnB, r0, r1, linear, angular0, angular1);
 
 			dstC->m_jacCoeffInv[ic] = calcJacCoeff(linear, -linear, angular0, angular1,
-				invMassA, &invInertiaA, invMassB, &invInertiaB ,countA,countB);
+												   invMassA, &invInertiaA, invMassB, &invInertiaB, countA, countB);
 
 			relVelN = calcRelVel(linear, -linear, angular0, angular1,
-				linVelA, angVelA, linVelB, angVelB);
+								 linVelA, angVelA, linVelB, angVelB);
 
-			float e = 0.f;//src->getRestituitionCoeff();
-			if( relVelN*relVelN < 0.004f ) 
+			float e = 0.f;  //src->getRestituitionCoeff();
+			if (relVelN * relVelN < 0.004f)
 			{
 				e = 0.f;
 			}
 
-			dstC->m_b[ic] = e*relVelN;
+			dstC->m_b[ic] = e * relVelN;
 			//float penetration = src->m_worldPos[ic].w;
-			dstC->m_b[ic] += (src->m_worldPosB[ic][3] + positionDrift)*positionConstraintCoeff*dtInv;
+			dstC->m_b[ic] += (src->m_worldPosB[ic][3] + positionDrift) * positionConstraintCoeff * dtInv;
 			dstC->m_appliedRambdaDt[ic] = 0.f;
 		}
 	}
 
-	if( src->m_worldNormalOnB[3] > 0 )//npoints
-	{	//	prepare friction
+	if (src->m_worldNormalOnB[3] > 0)  //npoints
+	{                                  //	prepare friction
 		b3Vector3 center = make_float4(0.f);
-		for(int i=0; i<src->m_worldNormalOnB[3]; i++) 
+		for (int i = 0; i < src->m_worldNormalOnB[3]; i++)
 			center += src->m_worldPosB[i];
 		center /= (float)src->m_worldNormalOnB[3];
 
 		b3Vector3 tangent[2];
-		b3PlaneSpace1(src->m_worldNormalOnB,tangent[0],tangent[1]);
-		
+		b3PlaneSpace1(src->m_worldNormalOnB, tangent[0], tangent[1]);
+
 		b3Vector3 r[2];
 		r[0] = center - posA;
 		r[1] = center - posB;
 
-		for(int i=0; i<2; i++)
+		for (int i = 0; i < 2; i++)
 		{
 			b3Vector3 linear, angular0, angular1;
 			setLinearAndAngular(tangent[i], r[0], r[1], linear, angular0, angular1);
 
 			dstC->m_fJacCoeffInv[i] = calcJacCoeff(linear, -linear, angular0, angular1,
-				invMassA, &invInertiaA, invMassB, &invInertiaB ,countA,countB);
+												   invMassA, &invInertiaA, invMassB, &invInertiaB, countA, countB);
 			dstC->m_fAppliedRambdaDt[i] = 0.f;
 		}
 		dstC->m_center = center;
 	}
 
-	for(int i=0; i<4; i++)
+	for (int i = 0; i < 4; i++)
 	{
-		if( i<src->m_worldNormalOnB[3] )
+		if (i < src->m_worldNormalOnB[3])
 		{
 			dstC->m_worldPos[i] = src->m_worldPosB[i];
 		}
@@ -453,17 +422,14 @@ void setConstraint4( const b3Vector3& posA, const b3Vector3& linVelA, const b3Ve
 	}
 }
 
-
-
 void ContactToConstraintKernel(b3Contact4* gContact, b3RigidBodyData* gBodies, b3InertiaData* gShapes, b3GpuConstraint4* gConstraintOut, int nContacts,
-float dt,
-float positionDrift,
-float positionConstraintCoeff, int gIdx, b3AlignedObjectArray<unsigned int>& bodyCount
-)
+							   float dt,
+							   float positionDrift,
+							   float positionConstraintCoeff, int gIdx, b3AlignedObjectArray<unsigned int>& bodyCount)
 {
 	//int gIdx = 0;//GET_GLOBAL_IDX;
-	
-	if( gIdx < nContacts )
+
+	if (gIdx < nContacts)
 	{
 		int aIdx = abs(gContact[gIdx].m_bodyAPtrAndSignBit);
 		int bIdx = abs(gContact[gIdx].m_bodyBPtrAndSignBit);
@@ -472,50 +438,46 @@ float positionConstraintCoeff, int gIdx, b3AlignedObjectArray<unsigned int>& bod
 		b3Vector3 linVelA = gBodies[aIdx].m_linVel;
 		b3Vector3 angVelA = gBodies[aIdx].m_angVel;
 		float invMassA = gBodies[aIdx].m_invMass;
-		b3Matrix3x3 invInertiaA = gShapes[aIdx].m_invInertiaWorld;//.m_invInertia;
+		b3Matrix3x3 invInertiaA = gShapes[aIdx].m_invInertiaWorld;  //.m_invInertia;
 
 		b3Vector3 posB = gBodies[bIdx].m_pos;
 		b3Vector3 linVelB = gBodies[bIdx].m_linVel;
 		b3Vector3 angVelB = gBodies[bIdx].m_angVel;
 		float invMassB = gBodies[bIdx].m_invMass;
-		b3Matrix3x3 invInertiaB = gShapes[bIdx].m_invInertiaWorld;//m_invInertia;
+		b3Matrix3x3 invInertiaB = gShapes[bIdx].m_invInertiaWorld;  //m_invInertia;
 
 		b3GpuConstraint4 cs;
 		float countA = invMassA ? (float)(bodyCount[aIdx]) : 1;
 		float countB = invMassB ? (float)(bodyCount[bIdx]) : 1;
-    	setConstraint4( posA, linVelA, angVelA, invMassA, invInertiaA, posB, linVelB, angVelB, invMassB, invInertiaB,
-			&gContact[gIdx], dt, positionDrift, positionConstraintCoeff,countA,countB,
-			&cs );
-		
+		setConstraint4(posA, linVelA, angVelA, invMassA, invInertiaA, posB, linVelB, angVelB, invMassB, invInertiaB,
+					   &gContact[gIdx], dt, positionDrift, positionConstraintCoeff, countA, countB,
+					   &cs);
 
-		
 		cs.m_batchIdx = gContact[gIdx].m_batchIdx;
 
 		gConstraintOut[gIdx] = cs;
 	}
 }
 
-
-void b3GpuJacobiContactSolver::solveGroupHost(b3RigidBodyData* bodies,b3InertiaData* inertias,int numBodies,b3Contact4* manifoldPtr, int numManifolds,const b3JacobiSolverInfo& solverInfo)
+void b3GpuJacobiContactSolver::solveGroupHost(b3RigidBodyData* bodies, b3InertiaData* inertias, int numBodies, b3Contact4* manifoldPtr, int numManifolds, const b3JacobiSolverInfo& solverInfo)
 {
 	B3_PROFILE("b3GpuJacobiContactSolver::solveGroup");
 
 	b3AlignedObjectArray<unsigned int> bodyCount;
 	bodyCount.resize(numBodies);
-	for (int i=0;i<numBodies;i++)
+	for (int i = 0; i < numBodies; i++)
 		bodyCount[i] = 0;
 
 	b3AlignedObjectArray<b3Int2> contactConstraintOffsets;
 	contactConstraintOffsets.resize(numManifolds);
 
-
-	for (int i=0;i<numManifolds;i++)
+	for (int i = 0; i < numManifolds; i++)
 	{
 		int pa = manifoldPtr[i].m_bodyAPtrAndSignBit;
 		int pb = manifoldPtr[i].m_bodyBPtrAndSignBit;
 
-		bool isFixedA = (pa <0) || (pa == solverInfo.m_fixedBodyIndex);
-		bool isFixedB = (pb <0) || (pb == solverInfo.m_fixedBodyIndex);
+		bool isFixedA = (pa < 0) || (pa == solverInfo.m_fixedBodyIndex);
+		bool isFixedB = (pb < 0) || (pb == solverInfo.m_fixedBodyIndex);
 
 		int bodyIndexA = manifoldPtr[i].getBodyA();
 		int bodyIndexB = manifoldPtr[i].getBodyB();
@@ -529,71 +491,63 @@ void b3GpuJacobiContactSolver::solveGroupHost(b3RigidBodyData* bodies,b3InertiaD
 		{
 			contactConstraintOffsets[i].y = bodyCount[bodyIndexB];
 			bodyCount[bodyIndexB]++;
-		} 
+		}
 	}
 
 	b3AlignedObjectArray<unsigned int> offsetSplitBodies;
 	offsetSplitBodies.resize(numBodies);
 	unsigned int totalNumSplitBodies;
-	m_data->m_scan->executeHost(bodyCount,offsetSplitBodies,numBodies,&totalNumSplitBodies);
-	int numlastBody = bodyCount[numBodies-1];
+	m_data->m_scan->executeHost(bodyCount, offsetSplitBodies, numBodies, &totalNumSplitBodies);
+	int numlastBody = bodyCount[numBodies - 1];
 	totalNumSplitBodies += numlastBody;
-	printf("totalNumSplitBodies = %d\n",totalNumSplitBodies);
-
-	
-
-
+	printf("totalNumSplitBodies = %d\n", totalNumSplitBodies);
 
 	b3AlignedObjectArray<b3GpuConstraint4> contactConstraints;
 	contactConstraints.resize(numManifolds);
 
-	for (int i=0;i<numManifolds;i++)
+	for (int i = 0; i < numManifolds; i++)
 	{
-		ContactToConstraintKernel(&manifoldPtr[0],bodies,inertias,&contactConstraints[0],numManifolds,
-			solverInfo.m_deltaTime,
-			solverInfo.m_positionDrift,
-			solverInfo.m_positionConstraintCoeff,
-			i, bodyCount);
+		ContactToConstraintKernel(&manifoldPtr[0], bodies, inertias, &contactConstraints[0], numManifolds,
+								  solverInfo.m_deltaTime,
+								  solverInfo.m_positionDrift,
+								  solverInfo.m_positionConstraintCoeff,
+								  i, bodyCount);
 	}
 	int maxIter = solverInfo.m_numIterations;
 
-
 	b3AlignedObjectArray<b3Vector3> deltaLinearVelocities;
 	b3AlignedObjectArray<b3Vector3> deltaAngularVelocities;
 	deltaLinearVelocities.resize(totalNumSplitBodies);
 	deltaAngularVelocities.resize(totalNumSplitBodies);
-	for (unsigned int i=0;i<totalNumSplitBodies;i++)
+	for (unsigned int i = 0; i < totalNumSplitBodies; i++)
 	{
 		deltaLinearVelocities[i].setZero();
 		deltaAngularVelocities[i].setZero();
 	}
 
-
-
-	for (int iter = 0;iter<maxIter;iter++)
+	for (int iter = 0; iter < maxIter; iter++)
 	{
-		int i=0;
-		for( i=0; i<numManifolds; i++)
+		int i = 0;
+		for (i = 0; i < numManifolds; i++)
 		{
-
 			//float frictionCoeff = contactConstraints[i].getFrictionCoeff();
 			int aIdx = (int)contactConstraints[i].m_bodyA;
 			int bIdx = (int)contactConstraints[i].m_bodyB;
 			b3RigidBodyData& bodyA = bodies[aIdx];
 			b3RigidBodyData& bodyB = bodies[bIdx];
 
-			b3Vector3 zero = b3MakeVector3(0,0,0);
-			
-			b3Vector3* dlvAPtr=&zero;
-			b3Vector3* davAPtr=&zero;
-			b3Vector3* dlvBPtr=&zero;
-			b3Vector3* davBPtr=&zero;
-			
+			b3Vector3 zero = b3MakeVector3(0, 0, 0);
+
+			b3Vector3* dlvAPtr = &zero;
+			b3Vector3* davAPtr = &zero;
+			b3Vector3* dlvBPtr = &zero;
+			b3Vector3* davBPtr = &zero;
+
 			if (bodyA.m_invMass)
 			{
 				int bodyOffsetA = offsetSplitBodies[aIdx];
 				int constraintOffsetA = contactConstraintOffsets[i].x;
-				int splitIndexA = bodyOffsetA+constraintOffsetA;
+				int splitIndexA = bodyOffsetA + constraintOffsetA;
 				dlvAPtr = &deltaLinearVelocities[splitIndexA];
 				davAPtr = &deltaAngularVelocities[splitIndexA];
 			}
@@ -602,67 +556,61 @@ void b3GpuJacobiContactSolver::solveGroupHost(b3RigidBodyData* bodies,b3InertiaD
 			{
 				int bodyOffsetB = offsetSplitBodies[bIdx];
 				int constraintOffsetB = contactConstraintOffsets[i].y;
-				int splitIndexB= bodyOffsetB+constraintOffsetB;
-				dlvBPtr =&deltaLinearVelocities[splitIndexB];
+				int splitIndexB = bodyOffsetB + constraintOffsetB;
+				dlvBPtr = &deltaLinearVelocities[splitIndexB];
 				davBPtr = &deltaAngularVelocities[splitIndexB];
 			}
 
-
-
 			{
-				float maxRambdaDt[4] = {FLT_MAX,FLT_MAX,FLT_MAX,FLT_MAX};
-				float minRambdaDt[4] = {0.f,0.f,0.f,0.f};
-
-				solveContact( contactConstraints[i], (b3Vector3&)bodyA.m_pos, (b3Vector3&)bodyA.m_linVel, (b3Vector3&)bodyA.m_angVel, bodyA.m_invMass, inertias[aIdx].m_invInertiaWorld, 
-					(b3Vector3&)bodyB.m_pos, (b3Vector3&)bodyB.m_linVel, (b3Vector3&)bodyB.m_angVel, bodyB.m_invMass, inertias[bIdx].m_invInertiaWorld,
-					maxRambdaDt, minRambdaDt , *dlvAPtr,*davAPtr,*dlvBPtr,*davBPtr		);
-
+				float maxRambdaDt[4] = {FLT_MAX, FLT_MAX, FLT_MAX, FLT_MAX};
+				float minRambdaDt[4] = {0.f, 0.f, 0.f, 0.f};
 
+				solveContact(contactConstraints[i], (b3Vector3&)bodyA.m_pos, (b3Vector3&)bodyA.m_linVel, (b3Vector3&)bodyA.m_angVel, bodyA.m_invMass, inertias[aIdx].m_invInertiaWorld,
+							 (b3Vector3&)bodyB.m_pos, (b3Vector3&)bodyB.m_linVel, (b3Vector3&)bodyB.m_angVel, bodyB.m_invMass, inertias[bIdx].m_invInertiaWorld,
+							 maxRambdaDt, minRambdaDt, *dlvAPtr, *davAPtr, *dlvBPtr, *davBPtr);
 			}
-			
 		}
 
-		
 		//easy
-		for (int i=0;i<numBodies;i++)
+		for (int i = 0; i < numBodies; i++)
 		{
 			if (bodies[i].m_invMass)
 			{
 				int bodyOffset = offsetSplitBodies[i];
 				int count = bodyCount[i];
-				float factor = 1.f/float(count);
+				float factor = 1.f / float(count);
 				b3Vector3 averageLinVel;
 				averageLinVel.setZero();
 				b3Vector3 averageAngVel;
 				averageAngVel.setZero();
-				for (int j=0;j<count;j++)
+				for (int j = 0; j < count; j++)
 				{
-					averageLinVel += deltaLinearVelocities[bodyOffset+j]*factor;
-					averageAngVel += deltaAngularVelocities[bodyOffset+j]*factor;
+					averageLinVel += deltaLinearVelocities[bodyOffset + j] * factor;
+					averageAngVel += deltaAngularVelocities[bodyOffset + j] * factor;
 				}
-				for (int j=0;j<count;j++)
+				for (int j = 0; j < count; j++)
 				{
-					deltaLinearVelocities[bodyOffset+j] = averageLinVel;
-					deltaAngularVelocities[bodyOffset+j] = averageAngVel;
+					deltaLinearVelocities[bodyOffset + j] = averageLinVel;
+					deltaAngularVelocities[bodyOffset + j] = averageAngVel;
 				}
 			}
 		}
 	}
-	for (int iter = 0;iter<maxIter;iter++)
+	for (int iter = 0; iter < maxIter; iter++)
 	{
 		//int i=0;
-	
+
 		//solve friction
 
-		for(int i=0; i<numManifolds; i++)
+		for (int i = 0; i < numManifolds; i++)
 		{
-			float maxRambdaDt[4] = {FLT_MAX,FLT_MAX,FLT_MAX,FLT_MAX};
-			float minRambdaDt[4] = {0.f,0.f,0.f,0.f};
+			float maxRambdaDt[4] = {FLT_MAX, FLT_MAX, FLT_MAX, FLT_MAX};
+			float minRambdaDt[4] = {0.f, 0.f, 0.f, 0.f};
 
 			float sum = 0;
-			for(int j=0; j<4; j++)
+			for (int j = 0; j < 4; j++)
 			{
-				sum +=contactConstraints[i].m_appliedRambdaDt[j];
+				sum += contactConstraints[i].m_appliedRambdaDt[j];
 			}
 			float frictionCoeff = contactConstraints[i].getFrictionCoeff();
 			int aIdx = (int)contactConstraints[i].m_bodyA;
@@ -670,18 +618,18 @@ void b3GpuJacobiContactSolver::solveGroupHost(b3RigidBodyData* bodies,b3InertiaD
 			b3RigidBodyData& bodyA = bodies[aIdx];
 			b3RigidBodyData& bodyB = bodies[bIdx];
 
-			b3Vector3 zero = b3MakeVector3(0,0,0);
-			
-			b3Vector3* dlvAPtr=&zero;
-			b3Vector3* davAPtr=&zero;
-			b3Vector3* dlvBPtr=&zero;
-			b3Vector3* davBPtr=&zero;
-			
+			b3Vector3 zero = b3MakeVector3(0, 0, 0);
+
+			b3Vector3* dlvAPtr = &zero;
+			b3Vector3* davAPtr = &zero;
+			b3Vector3* dlvBPtr = &zero;
+			b3Vector3* davBPtr = &zero;
+
 			if (bodyA.m_invMass)
 			{
 				int bodyOffsetA = offsetSplitBodies[aIdx];
 				int constraintOffsetA = contactConstraintOffsets[i].x;
-				int splitIndexA = bodyOffsetA+constraintOffsetA;
+				int splitIndexA = bodyOffsetA + constraintOffsetA;
 				dlvAPtr = &deltaLinearVelocities[splitIndexA];
 				davAPtr = &deltaAngularVelocities[splitIndexA];
 			}
@@ -690,55 +638,50 @@ void b3GpuJacobiContactSolver::solveGroupHost(b3RigidBodyData* bodies,b3InertiaD
 			{
 				int bodyOffsetB = offsetSplitBodies[bIdx];
 				int constraintOffsetB = contactConstraintOffsets[i].y;
-				int splitIndexB= bodyOffsetB+constraintOffsetB;
-				dlvBPtr =&deltaLinearVelocities[splitIndexB];
+				int splitIndexB = bodyOffsetB + constraintOffsetB;
+				dlvBPtr = &deltaLinearVelocities[splitIndexB];
 				davBPtr = &deltaAngularVelocities[splitIndexB];
 			}
 
-			for(int j=0; j<4; j++)
+			for (int j = 0; j < 4; j++)
 			{
-				maxRambdaDt[j] = frictionCoeff*sum;
+				maxRambdaDt[j] = frictionCoeff * sum;
 				minRambdaDt[j] = -maxRambdaDt[j];
 			}
 
-			solveFriction( contactConstraints[i], (b3Vector3&)bodyA.m_pos, (b3Vector3&)bodyA.m_linVel, (b3Vector3&)bodyA.m_angVel, bodyA.m_invMass,inertias[aIdx].m_invInertiaWorld, 
-				(b3Vector3&)bodyB.m_pos, (b3Vector3&)bodyB.m_linVel, (b3Vector3&)bodyB.m_angVel, bodyB.m_invMass, inertias[bIdx].m_invInertiaWorld,
-				maxRambdaDt, minRambdaDt , *dlvAPtr,*davAPtr,*dlvBPtr,*davBPtr);
-
+			solveFriction(contactConstraints[i], (b3Vector3&)bodyA.m_pos, (b3Vector3&)bodyA.m_linVel, (b3Vector3&)bodyA.m_angVel, bodyA.m_invMass, inertias[aIdx].m_invInertiaWorld,
+						  (b3Vector3&)bodyB.m_pos, (b3Vector3&)bodyB.m_linVel, (b3Vector3&)bodyB.m_angVel, bodyB.m_invMass, inertias[bIdx].m_invInertiaWorld,
+						  maxRambdaDt, minRambdaDt, *dlvAPtr, *davAPtr, *dlvBPtr, *davBPtr);
 		}
 
 		//easy
-		for (int i=0;i<numBodies;i++)
+		for (int i = 0; i < numBodies; i++)
 		{
 			if (bodies[i].m_invMass)
 			{
 				int bodyOffset = offsetSplitBodies[i];
 				int count = bodyCount[i];
-				float factor = 1.f/float(count);
+				float factor = 1.f / float(count);
 				b3Vector3 averageLinVel;
 				averageLinVel.setZero();
 				b3Vector3 averageAngVel;
 				averageAngVel.setZero();
-				for (int j=0;j<count;j++)
+				for (int j = 0; j < count; j++)
 				{
-					averageLinVel += deltaLinearVelocities[bodyOffset+j]*factor;
-					averageAngVel += deltaAngularVelocities[bodyOffset+j]*factor;
+					averageLinVel += deltaLinearVelocities[bodyOffset + j] * factor;
+					averageAngVel += deltaAngularVelocities[bodyOffset + j] * factor;
 				}
-				for (int j=0;j<count;j++)
+				for (int j = 0; j < count; j++)
 				{
-					deltaLinearVelocities[bodyOffset+j] = averageLinVel;
-					deltaAngularVelocities[bodyOffset+j] = averageAngVel;
+					deltaLinearVelocities[bodyOffset + j] = averageLinVel;
+					deltaAngularVelocities[bodyOffset + j] = averageAngVel;
 				}
 			}
 		}
-
-
-
 	}
 
-
 	//easy
-	for (int i=0;i<numBodies;i++)
+	for (int i = 0; i < numBodies; i++)
 	{
 		if (bodies[i].m_invMass)
 		{
@@ -753,8 +696,6 @@ void b3GpuJacobiContactSolver::solveGroupHost(b3RigidBodyData* bodies,b3InertiaD
 	}
 }
 
-
-
 void b3GpuJacobiContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem inertiaBuf, int numContacts, cl_mem contactBuf, const struct b3Config& config, int static0Index)
 //
 //
@@ -762,49 +703,47 @@ void b3GpuJacobiContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_m
 {
 	b3JacobiSolverInfo solverInfo;
 	solverInfo.m_fixedBodyIndex = static0Index;
-	
 
 	B3_PROFILE("b3GpuJacobiContactSolver::solveGroup");
 
 	//int numBodies = bodies->size();
-	int numManifolds = numContacts;//manifoldPtr->size();
+	int numManifolds = numContacts;  //manifoldPtr->size();
 
 	{
 		B3_PROFILE("resize");
 		m_data->m_bodyCount->resize(numBodies);
 	}
-	
-	unsigned int val=0;
-	b3Int2 val2;
-	val2.x=0;
-	val2.y=0;
 
-	 {
+	unsigned int val = 0;
+	b3Int2 val2;
+	val2.x = 0;
+	val2.y = 0;
+
+	{
 		B3_PROFILE("m_filler");
 		m_data->m_contactConstraintOffsets->resize(numManifolds);
-		m_data->m_filler->execute(*m_data->m_bodyCount,val,numBodies);
-		
-	
-		m_data->m_filler->execute(*m_data->m_contactConstraintOffsets,val2,numManifolds);
+		m_data->m_filler->execute(*m_data->m_bodyCount, val, numBodies);
+
+		m_data->m_filler->execute(*m_data->m_contactConstraintOffsets, val2, numManifolds);
 	}
 
 	{
 		B3_PROFILE("m_countBodiesKernel");
-		b3LauncherCL launcher(this->m_queue,m_data->m_countBodiesKernel,"m_countBodiesKernel");
-		launcher.setBuffer(contactBuf);//manifoldPtr->getBufferCL());
+		b3LauncherCL launcher(this->m_queue, m_data->m_countBodiesKernel, "m_countBodiesKernel");
+		launcher.setBuffer(contactBuf);  //manifoldPtr->getBufferCL());
 		launcher.setBuffer(m_data->m_bodyCount->getBufferCL());
 		launcher.setBuffer(m_data->m_contactConstraintOffsets->getBufferCL());
 		launcher.setConst(numManifolds);
 		launcher.setConst(solverInfo.m_fixedBodyIndex);
 		launcher.launch1D(numManifolds);
 	}
-	unsigned int totalNumSplitBodies=0;
+	unsigned int totalNumSplitBodies = 0;
 	{
 		B3_PROFILE("m_scan->execute");
-		
+
 		m_data->m_offsetSplitBodies->resize(numBodies);
-		m_data->m_scan->execute(*m_data->m_bodyCount,*m_data->m_offsetSplitBodies,numBodies,&totalNumSplitBodies);
-		totalNumSplitBodies+=m_data->m_bodyCount->at(numBodies-1);
+		m_data->m_scan->execute(*m_data->m_bodyCount, *m_data->m_offsetSplitBodies, numBodies, &totalNumSplitBodies);
+		totalNumSplitBodies += m_data->m_bodyCount->at(numBodies - 1);
 	}
 
 	{
@@ -812,50 +751,45 @@ void b3GpuJacobiContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_m
 		//int numContacts = manifoldPtr->size();
 		m_data->m_contactConstraints->resize(numContacts);
 	}
-	
+
 	{
 		B3_PROFILE("contactToConstraintSplitKernel");
-		b3LauncherCL launcher( m_queue, m_data->m_contactToConstraintSplitKernel,"m_contactToConstraintSplitKernel");
+		b3LauncherCL launcher(m_queue, m_data->m_contactToConstraintSplitKernel, "m_contactToConstraintSplitKernel");
 		launcher.setBuffer(contactBuf);
 		launcher.setBuffer(bodyBuf);
 		launcher.setBuffer(inertiaBuf);
 		launcher.setBuffer(m_data->m_contactConstraints->getBufferCL());
 		launcher.setBuffer(m_data->m_bodyCount->getBufferCL());
-        launcher.setConst(numContacts);
+		launcher.setConst(numContacts);
 		launcher.setConst(solverInfo.m_deltaTime);
 		launcher.setConst(solverInfo.m_positionDrift);
 		launcher.setConst(solverInfo.m_positionConstraintCoeff);
-		launcher.launch1D( numContacts, 64 );
-		
+		launcher.launch1D(numContacts, 64);
 	}
 
-	
 	{
 		B3_PROFILE("m_data->m_deltaLinearVelocities->resize");
 		m_data->m_deltaLinearVelocities->resize(totalNumSplitBodies);
 		m_data->m_deltaAngularVelocities->resize(totalNumSplitBodies);
 	}
 
-
-	
 	{
 		B3_PROFILE("m_clearVelocitiesKernel");
-		b3LauncherCL launch(m_queue,m_data->m_clearVelocitiesKernel,"m_clearVelocitiesKernel");
+		b3LauncherCL launch(m_queue, m_data->m_clearVelocitiesKernel, "m_clearVelocitiesKernel");
 		launch.setBuffer(m_data->m_deltaAngularVelocities->getBufferCL());
 		launch.setBuffer(m_data->m_deltaLinearVelocities->getBufferCL());
 		launch.setConst(totalNumSplitBodies);
 		launch.launch1D(totalNumSplitBodies);
 		clFinish(m_queue);
 	}
-	
-	
+
 	int maxIter = solverInfo.m_numIterations;
 
-	for (int iter = 0;iter<maxIter;iter++)
+	for (int iter = 0; iter < maxIter; iter++)
 	{
 		{
 			B3_PROFILE("m_solveContactKernel");
-			b3LauncherCL launcher( m_queue, m_data->m_solveContactKernel,"m_solveContactKernel" );
+			b3LauncherCL launcher(m_queue, m_data->m_solveContactKernel, "m_solveContactKernel");
 			launcher.setBuffer(m_data->m_contactConstraints->getBufferCL());
 			launcher.setBuffer(bodyBuf);
 			launcher.setBuffer(inertiaBuf);
@@ -873,11 +807,9 @@ void b3GpuJacobiContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_m
 			clFinish(m_queue);
 		}
 
-		
-
 		{
 			B3_PROFILE("average velocities");
-			b3LauncherCL launcher( m_queue, m_data->m_averageVelocitiesKernel,"m_averageVelocitiesKernel");
+			b3LauncherCL launcher(m_queue, m_data->m_averageVelocitiesKernel, "m_averageVelocitiesKernel");
 			launcher.setBuffer(bodyBuf);
 			launcher.setBuffer(m_data->m_offsetSplitBodies->getBufferCL());
 			launcher.setBuffer(m_data->m_bodyCount->getBufferCL());
@@ -888,10 +820,9 @@ void b3GpuJacobiContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_m
 			clFinish(m_queue);
 		}
 
-		
 		{
 			B3_PROFILE("m_solveFrictionKernel");
-			b3LauncherCL launcher( m_queue, m_data->m_solveFrictionKernel,"m_solveFrictionKernel");
+			b3LauncherCL launcher(m_queue, m_data->m_solveFrictionKernel, "m_solveFrictionKernel");
 			launcher.setBuffer(m_data->m_contactConstraints->getBufferCL());
 			launcher.setBuffer(bodyBuf);
 			launcher.setBuffer(inertiaBuf);
@@ -909,10 +840,9 @@ void b3GpuJacobiContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_m
 			clFinish(m_queue);
 		}
 
-		
 		{
 			B3_PROFILE("average velocities");
-			b3LauncherCL launcher( m_queue, m_data->m_averageVelocitiesKernel,"m_averageVelocitiesKernel");
+			b3LauncherCL launcher(m_queue, m_data->m_averageVelocitiesKernel, "m_averageVelocitiesKernel");
 			launcher.setBuffer(bodyBuf);
 			launcher.setBuffer(m_data->m_offsetSplitBodies->getBufferCL());
 			launcher.setBuffer(m_data->m_bodyCount->getBufferCL());
@@ -922,27 +852,20 @@ void b3GpuJacobiContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_m
 			launcher.launch1D(numBodies);
 			clFinish(m_queue);
 		}
-
-		
-
 	}
 
-	
 	{
-			B3_PROFILE("update body velocities");
-			b3LauncherCL launcher( m_queue, m_data->m_updateBodyVelocitiesKernel,"m_updateBodyVelocitiesKernel");
-			launcher.setBuffer(bodyBuf);
-			launcher.setBuffer(m_data->m_offsetSplitBodies->getBufferCL());
-			launcher.setBuffer(m_data->m_bodyCount->getBufferCL());
-			launcher.setBuffer(m_data->m_deltaLinearVelocities->getBufferCL());
-			launcher.setBuffer(m_data->m_deltaAngularVelocities->getBufferCL());
-			launcher.setConst(numBodies);
-			launcher.launch1D(numBodies);
-			clFinish(m_queue);
-		}
-
-
-
+		B3_PROFILE("update body velocities");
+		b3LauncherCL launcher(m_queue, m_data->m_updateBodyVelocitiesKernel, "m_updateBodyVelocitiesKernel");
+		launcher.setBuffer(bodyBuf);
+		launcher.setBuffer(m_data->m_offsetSplitBodies->getBufferCL());
+		launcher.setBuffer(m_data->m_bodyCount->getBufferCL());
+		launcher.setBuffer(m_data->m_deltaLinearVelocities->getBufferCL());
+		launcher.setBuffer(m_data->m_deltaAngularVelocities->getBufferCL());
+		launcher.setConst(numBodies);
+		launcher.launch1D(numBodies);
+		clFinish(m_queue);
+	}
 }
 
 #if 0
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuJacobiContactSolver.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuJacobiContactSolver.h
index b418f29ec43..8281aee05de 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuJacobiContactSolver.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuJacobiContactSolver.h
@@ -8,7 +8,6 @@
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3Contact4Data.h"
 #include "Bullet3OpenCL/ParallelPrimitives/b3OpenCLArray.h"
 
-
 //struct b3InertiaData;
 //b3InertiaData
 
@@ -21,21 +20,20 @@ struct b3JacobiSolverInfo
 	float m_deltaTime;
 	float m_positionDrift;
 	float m_positionConstraintCoeff;
-	int	m_numIterations;
+	int m_numIterations;
 
 	b3JacobiSolverInfo()
-		:m_fixedBodyIndex(0),
-		m_deltaTime(1./60.f),
-		m_positionDrift( 0.005f ), 
-		m_positionConstraintCoeff( 0.99f ),
-		m_numIterations(7)
+		: m_fixedBodyIndex(0),
+		  m_deltaTime(1. / 60.f),
+		  m_positionDrift(0.005f),
+		  m_positionConstraintCoeff(0.99f),
+		  m_numIterations(7)
 	{
 	}
 };
 class b3GpuJacobiContactSolver
 {
 protected:
-
 	struct b3GpuJacobiSolverInternalData* m_data;
 
 	cl_context m_context;
@@ -43,20 +41,16 @@ protected:
 	cl_command_queue m_queue;
 
 public:
-
 	b3GpuJacobiContactSolver(cl_context ctx, cl_device_id device, cl_command_queue queue, int pairCapacity);
 	virtual ~b3GpuJacobiContactSolver();
 
-
 	void solveContacts(int numBodies, cl_mem bodyBuf, cl_mem inertiaBuf, int numContacts, cl_mem contactBuf, const struct b3Config& config, int static0Index);
-	void solveGroupHost(b3RigidBodyData* bodies,b3InertiaData* inertias,int numBodies,struct b3Contact4* manifoldPtr, int numManifolds,const b3JacobiSolverInfo& solverInfo);
+	void solveGroupHost(b3RigidBodyData* bodies, b3InertiaData* inertias, int numBodies, struct b3Contact4* manifoldPtr, int numManifolds, const b3JacobiSolverInfo& solverInfo);
 	//void  solveGroupHost(btRigidBodyCL* bodies,b3InertiaData* inertias,int numBodies,btContact4* manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btJacobiSolverInfo& solverInfo);
 
 	//b3Scalar solveGroup(b3OpenCLArray<b3RigidBodyData>* gpuBodies,b3OpenCLArray<b3InertiaData>* gpuInertias, int numBodies,b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints,int numConstraints,const b3ContactSolverInfo& infoGlobal);
 
 	//void  solveGroup(btOpenCLArray<btRigidBodyCL>* bodies,btOpenCLArray<btInertiaCL>* inertias,btOpenCLArray<btContact4>* manifoldPtr,const btJacobiSolverInfo& solverInfo);
 	//void  solveGroupMixed(btOpenCLArray<btRigidBodyCL>* bodies,btOpenCLArray<btInertiaCL>* inertias,btOpenCLArray<btContact4>* manifoldPtr,const btJacobiSolverInfo& solverInfo);
-
 };
-#endif //B3_GPU_JACOBI_CONTACT_SOLVER_H
-
+#endif  //B3_GPU_JACOBI_CONTACT_SOLVER_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuNarrowPhase.cpp b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuNarrowPhase.cpp
index 698fa15f961..2e4f6c1572f 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuNarrowPhase.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuNarrowPhase.cpp
@@ -1,6 +1,5 @@
 #include "b3GpuNarrowPhase.h"
 
-
 #include "Bullet3OpenCL/ParallelPrimitives/b3OpenCLArray.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3ConvexPolyhedronData.h"
 #include "Bullet3OpenCL/NarrowphaseCollision/b3ConvexHullContact.h"
@@ -16,107 +15,87 @@
 #include "Bullet3OpenCL/NarrowphaseCollision/b3QuantizedBvh.h"
 #include "Bullet3Collision/NarrowPhaseCollision/b3ConvexUtility.h"
 
-
-
-
 b3GpuNarrowPhase::b3GpuNarrowPhase(cl_context ctx, cl_device_id device, cl_command_queue queue, const b3Config& config)
-:m_data(0) ,m_planeBodyIndex(-1),m_static0Index(-1),
-m_context(ctx),
-m_device(device),
-m_queue(queue)
+	: m_data(0), m_planeBodyIndex(-1), m_static0Index(-1), m_context(ctx), m_device(device), m_queue(queue)
 {
-    
 	m_data = new b3GpuNarrowPhaseInternalData();
 	m_data->m_currentContactBuffer = 0;
 
-	memset(m_data,0,sizeof(b3GpuNarrowPhaseInternalData));
-    
+	memset(m_data, 0, sizeof(b3GpuNarrowPhaseInternalData));
 
 	m_data->m_config = config;
-	
-	m_data->m_gpuSatCollision = new GpuSatCollision(ctx,device,queue);
-	
-	
-	m_data->m_triangleConvexPairs = new b3OpenCLArray<b3Int4>(m_context,m_queue, config.m_maxTriConvexPairCapacity);
 
+	m_data->m_gpuSatCollision = new GpuSatCollision(ctx, device, queue);
+
+	m_data->m_triangleConvexPairs = new b3OpenCLArray<b3Int4>(m_context, m_queue, config.m_maxTriConvexPairCapacity);
 
 	//m_data->m_convexPairsOutGPU = new b3OpenCLArray<b3Int2>(ctx,queue,config.m_maxBroadphasePairs,false);
 	//m_data->m_planePairs = new b3OpenCLArray<b3Int2>(ctx,queue,config.m_maxBroadphasePairs,false);
-    
+
 	m_data->m_pBufContactOutCPU = new b3AlignedObjectArray<b3Contact4>();
 	m_data->m_pBufContactOutCPU->resize(config.m_maxBroadphasePairs);
 	m_data->m_bodyBufferCPU = new b3AlignedObjectArray<b3RigidBodyData>();
 	m_data->m_bodyBufferCPU->resize(config.m_maxConvexBodies);
-    
+
 	m_data->m_inertiaBufferCPU = new b3AlignedObjectArray<b3InertiaData>();
 	m_data->m_inertiaBufferCPU->resize(config.m_maxConvexBodies);
-	
-	m_data->m_pBufContactBuffersGPU[0] = new b3OpenCLArray<b3Contact4>(ctx,queue, config.m_maxContactCapacity,true);
-	m_data->m_pBufContactBuffersGPU[1] = new b3OpenCLArray<b3Contact4>(ctx,queue, config.m_maxContactCapacity,true);
-	
-	m_data->m_inertiaBufferGPU = new b3OpenCLArray<b3InertiaData>(ctx,queue,config.m_maxConvexBodies,false);
-	m_data->m_collidablesGPU = new b3OpenCLArray<b3Collidable>(ctx,queue,config.m_maxConvexShapes);
+
+	m_data->m_pBufContactBuffersGPU[0] = new b3OpenCLArray<b3Contact4>(ctx, queue, config.m_maxContactCapacity, true);
+	m_data->m_pBufContactBuffersGPU[1] = new b3OpenCLArray<b3Contact4>(ctx, queue, config.m_maxContactCapacity, true);
+
+	m_data->m_inertiaBufferGPU = new b3OpenCLArray<b3InertiaData>(ctx, queue, config.m_maxConvexBodies, false);
+	m_data->m_collidablesGPU = new b3OpenCLArray<b3Collidable>(ctx, queue, config.m_maxConvexShapes);
 	m_data->m_collidablesCPU.reserve(config.m_maxConvexShapes);
 
 	m_data->m_localShapeAABBCPU = new b3AlignedObjectArray<b3SapAabb>;
-	m_data->m_localShapeAABBGPU = new b3OpenCLArray<b3SapAabb>(ctx,queue,config.m_maxConvexShapes);
-    
-    
+	m_data->m_localShapeAABBGPU = new b3OpenCLArray<b3SapAabb>(ctx, queue, config.m_maxConvexShapes);
+
 	//m_data->m_solverDataGPU = adl::Solver<adl::TYPE_CL>::allocate(ctx,queue, config.m_maxBroadphasePairs,false);
-	m_data->m_bodyBufferGPU = new b3OpenCLArray<b3RigidBodyData>(ctx,queue, config.m_maxConvexBodies,false);
+	m_data->m_bodyBufferGPU = new b3OpenCLArray<b3RigidBodyData>(ctx, queue, config.m_maxConvexBodies, false);
 
-	m_data->m_convexFacesGPU = new b3OpenCLArray<b3GpuFace>(ctx,queue,config.m_maxConvexShapes*config.m_maxFacesPerShape,false);
-	m_data->m_convexFaces.reserve(config.m_maxConvexShapes*config.m_maxFacesPerShape);
+	m_data->m_convexFacesGPU = new b3OpenCLArray<b3GpuFace>(ctx, queue, config.m_maxConvexShapes * config.m_maxFacesPerShape, false);
+	m_data->m_convexFaces.reserve(config.m_maxConvexShapes * config.m_maxFacesPerShape);
 
-	m_data->m_gpuChildShapes = new b3OpenCLArray<b3GpuChildShape>(ctx,queue,config.m_maxCompoundChildShapes,false);
-	
-	m_data->m_convexPolyhedraGPU = new b3OpenCLArray<b3ConvexPolyhedronData>(ctx,queue,config.m_maxConvexShapes,false);
+	m_data->m_gpuChildShapes = new b3OpenCLArray<b3GpuChildShape>(ctx, queue, config.m_maxCompoundChildShapes, false);
+
+	m_data->m_convexPolyhedraGPU = new b3OpenCLArray<b3ConvexPolyhedronData>(ctx, queue, config.m_maxConvexShapes, false);
 	m_data->m_convexPolyhedra.reserve(config.m_maxConvexShapes);
 
-	m_data->m_uniqueEdgesGPU = new b3OpenCLArray<b3Vector3>(ctx,queue,config.m_maxConvexUniqueEdges,true);
+	m_data->m_uniqueEdgesGPU = new b3OpenCLArray<b3Vector3>(ctx, queue, config.m_maxConvexUniqueEdges, true);
 	m_data->m_uniqueEdges.reserve(config.m_maxConvexUniqueEdges);
-	
-	
 
-	m_data->m_convexVerticesGPU = new b3OpenCLArray<b3Vector3>(ctx,queue,config.m_maxConvexVertices,true);
+	m_data->m_convexVerticesGPU = new b3OpenCLArray<b3Vector3>(ctx, queue, config.m_maxConvexVertices, true);
 	m_data->m_convexVertices.reserve(config.m_maxConvexVertices);
 
-	m_data->m_convexIndicesGPU = new b3OpenCLArray<int>(ctx,queue,config.m_maxConvexIndices,true);
-    m_data->m_convexIndices.reserve(config.m_maxConvexIndices);
-    
-	m_data->m_worldVertsB1GPU = new b3OpenCLArray<b3Vector3>(ctx,queue,config.m_maxConvexBodies*config.m_maxVerticesPerFace);
-    m_data->m_clippingFacesOutGPU = new  b3OpenCLArray<b3Int4>(ctx,queue,config.m_maxConvexBodies);
-    m_data->m_worldNormalsAGPU = new  b3OpenCLArray<b3Vector3>(ctx,queue,config.m_maxConvexBodies);
-	m_data->m_worldVertsA1GPU = new b3OpenCLArray<b3Vector3>(ctx,queue,config.m_maxConvexBodies*config.m_maxVerticesPerFace);
-    m_data->m_worldVertsB2GPU = new  b3OpenCLArray<b3Vector3>(ctx,queue,config.m_maxConvexBodies*config.m_maxVerticesPerFace);
-    
-    
+	m_data->m_convexIndicesGPU = new b3OpenCLArray<int>(ctx, queue, config.m_maxConvexIndices, true);
+	m_data->m_convexIndices.reserve(config.m_maxConvexIndices);
 
-	m_data->m_convexData = new b3AlignedObjectArray<b3ConvexUtility* >();
+	m_data->m_worldVertsB1GPU = new b3OpenCLArray<b3Vector3>(ctx, queue, config.m_maxConvexBodies * config.m_maxVerticesPerFace);
+	m_data->m_clippingFacesOutGPU = new b3OpenCLArray<b3Int4>(ctx, queue, config.m_maxConvexBodies);
+	m_data->m_worldNormalsAGPU = new b3OpenCLArray<b3Vector3>(ctx, queue, config.m_maxConvexBodies);
+	m_data->m_worldVertsA1GPU = new b3OpenCLArray<b3Vector3>(ctx, queue, config.m_maxConvexBodies * config.m_maxVerticesPerFace);
+	m_data->m_worldVertsB2GPU = new b3OpenCLArray<b3Vector3>(ctx, queue, config.m_maxConvexBodies * config.m_maxVerticesPerFace);
+
+	m_data->m_convexData = new b3AlignedObjectArray<b3ConvexUtility*>();
 
 	m_data->m_convexData->resize(config.m_maxConvexShapes);
 	m_data->m_convexPolyhedra.resize(config.m_maxConvexShapes);
-    
+
 	m_data->m_numAcceleratedShapes = 0;
 	m_data->m_numAcceleratedRigidBodies = 0;
-    
-		
-	m_data->m_subTreesGPU = new b3OpenCLArray<b3BvhSubtreeInfo>(this->m_context,this->m_queue);
-	m_data->m_treeNodesGPU = new b3OpenCLArray<b3QuantizedBvhNode>(this->m_context,this->m_queue);
-	m_data->m_bvhInfoGPU = new b3OpenCLArray<b3BvhInfo>(this->m_context,this->m_queue);
+
+	m_data->m_subTreesGPU = new b3OpenCLArray<b3BvhSubtreeInfo>(this->m_context, this->m_queue);
+	m_data->m_treeNodesGPU = new b3OpenCLArray<b3QuantizedBvhNode>(this->m_context, this->m_queue);
+	m_data->m_bvhInfoGPU = new b3OpenCLArray<b3BvhInfo>(this->m_context, this->m_queue);
 
 	//m_data->m_contactCGPU = new b3OpenCLArray<Constraint4>(ctx,queue,config.m_maxBroadphasePairs,false);
 	//m_data->m_frictionCGPU = new b3OpenCLArray<adl::Solver<adl::TYPE_CL>::allocateFrictionConstraint( m_data->m_deviceCL, config.m_maxBroadphasePairs);
- 
-	
-
 }
 
-
 b3GpuNarrowPhase::~b3GpuNarrowPhase()
 {
 	delete m_data->m_gpuSatCollision;
-	
+
 	delete m_data->m_triangleConvexPairs;
 	//delete m_data->m_convexPairsOutGPU;
 	//delete m_data->m_planePairs;
@@ -126,7 +105,6 @@ b3GpuNarrowPhase::~b3GpuNarrowPhase()
 	delete m_data->m_pBufContactBuffersGPU[0];
 	delete m_data->m_pBufContactBuffersGPU[1];
 
-
 	delete m_data->m_inertiaBufferGPU;
 	delete m_data->m_collidablesGPU;
 	delete m_data->m_localShapeAABBCPU;
@@ -139,18 +117,18 @@ b3GpuNarrowPhase::~b3GpuNarrowPhase()
 	delete m_data->m_convexVerticesGPU;
 	delete m_data->m_convexIndicesGPU;
 	delete m_data->m_worldVertsB1GPU;
-    delete m_data->m_clippingFacesOutGPU;
-    delete m_data->m_worldNormalsAGPU;
+	delete m_data->m_clippingFacesOutGPU;
+	delete m_data->m_worldNormalsAGPU;
 	delete m_data->m_worldVertsA1GPU;
-    delete m_data->m_worldVertsB2GPU;
-    
+	delete m_data->m_worldVertsB2GPU;
+
 	delete m_data->m_bvhInfoGPU;
 
-	for (int i=0;i<m_data->m_bvhData.size();i++)
+	for (int i = 0; i < m_data->m_bvhData.size(); i++)
 	{
 		delete m_data->m_bvhData[i];
 	}
-	for (int i=0;i<m_data->m_meshInterfaces.size();i++)
+	for (int i = 0; i < m_data->m_meshInterfaces.size(); i++)
 	{
 		delete m_data->m_meshInterfaces[i];
 	}
@@ -159,198 +137,180 @@ b3GpuNarrowPhase::~b3GpuNarrowPhase()
 	delete m_data->m_treeNodesGPU;
 	delete m_data->m_subTreesGPU;
 
-    
-    delete m_data->m_convexData;
+	delete m_data->m_convexData;
 	delete m_data;
 }
 
-
-int	b3GpuNarrowPhase::allocateCollidable()
+int b3GpuNarrowPhase::allocateCollidable()
 {
 	int curSize = m_data->m_collidablesCPU.size();
-	if (curSize<m_data->m_config.m_maxConvexShapes)
+	if (curSize < m_data->m_config.m_maxConvexShapes)
 	{
 		m_data->m_collidablesCPU.expand();
 		return curSize;
 	}
 	else
 	{
-		b3Error("allocateCollidable out-of-range %d\n",m_data->m_config.m_maxConvexShapes);
+		b3Error("allocateCollidable out-of-range %d\n", m_data->m_config.m_maxConvexShapes);
 	}
 	return -1;
-
 }
 
-
-
-
-
-int		b3GpuNarrowPhase::registerSphereShape(float radius)
+int b3GpuNarrowPhase::registerSphereShape(float radius)
 {
 	int collidableIndex = allocateCollidable();
-	if (collidableIndex<0)
+	if (collidableIndex < 0)
 		return collidableIndex;
 
-
 	b3Collidable& col = getCollidableCpu(collidableIndex);
 	col.m_shapeType = SHAPE_SPHERE;
 	col.m_shapeIndex = 0;
 	col.m_radius = radius;
-	
-	if (col.m_shapeIndex>=0)
+
+	if (col.m_shapeIndex >= 0)
 	{
 		b3SapAabb aabb;
-		b3Vector3 myAabbMin=b3MakeVector3(-radius,-radius,-radius);
-		b3Vector3 myAabbMax=b3MakeVector3(radius,radius,radius);
+		b3Vector3 myAabbMin = b3MakeVector3(-radius, -radius, -radius);
+		b3Vector3 myAabbMax = b3MakeVector3(radius, radius, radius);
 
-		aabb.m_min[0] = myAabbMin[0];//s_convexHeightField->m_aabb.m_min.x;
-		aabb.m_min[1] = myAabbMin[1];//s_convexHeightField->m_aabb.m_min.y;
-		aabb.m_min[2] = myAabbMin[2];//s_convexHeightField->m_aabb.m_min.z;
+		aabb.m_min[0] = myAabbMin[0];  //s_convexHeightField->m_aabb.m_min.x;
+		aabb.m_min[1] = myAabbMin[1];  //s_convexHeightField->m_aabb.m_min.y;
+		aabb.m_min[2] = myAabbMin[2];  //s_convexHeightField->m_aabb.m_min.z;
 		aabb.m_minIndices[3] = 0;
 
-		aabb.m_max[0] = myAabbMax[0];//s_convexHeightField->m_aabb.m_max.x;
-		aabb.m_max[1] = myAabbMax[1];//s_convexHeightField->m_aabb.m_max.y;
-		aabb.m_max[2] = myAabbMax[2];//s_convexHeightField->m_aabb.m_max.z;
+		aabb.m_max[0] = myAabbMax[0];  //s_convexHeightField->m_aabb.m_max.x;
+		aabb.m_max[1] = myAabbMax[1];  //s_convexHeightField->m_aabb.m_max.y;
+		aabb.m_max[2] = myAabbMax[2];  //s_convexHeightField->m_aabb.m_max.z;
 		aabb.m_signedMaxIndices[3] = 0;
 
 		m_data->m_localShapeAABBCPU->push_back(aabb);
-//		m_data->m_localShapeAABBGPU->push_back(aabb);
+		//		m_data->m_localShapeAABBGPU->push_back(aabb);
 		clFinish(m_queue);
 	}
-	
+
 	return collidableIndex;
 }
 
-
 int b3GpuNarrowPhase::registerFace(const b3Vector3& faceNormal, float faceConstant)
 {
 	int faceOffset = m_data->m_convexFaces.size();
 	b3GpuFace& face = m_data->m_convexFaces.expand();
-	face.m_plane = b3MakeVector3(faceNormal.x,faceNormal.y,faceNormal.z,faceConstant);
+	face.m_plane = b3MakeVector3(faceNormal.x, faceNormal.y, faceNormal.z, faceConstant);
 	return faceOffset;
 }
 
-int		b3GpuNarrowPhase::registerPlaneShape(const b3Vector3& planeNormal, float planeConstant)
+int b3GpuNarrowPhase::registerPlaneShape(const b3Vector3& planeNormal, float planeConstant)
 {
 	int collidableIndex = allocateCollidable();
-	if (collidableIndex<0)
+	if (collidableIndex < 0)
 		return collidableIndex;
 
-
 	b3Collidable& col = getCollidableCpu(collidableIndex);
 	col.m_shapeType = SHAPE_PLANE;
-	col.m_shapeIndex = registerFace(planeNormal,planeConstant);
+	col.m_shapeIndex = registerFace(planeNormal, planeConstant);
 	col.m_radius = planeConstant;
-	
-	if (col.m_shapeIndex>=0)
+
+	if (col.m_shapeIndex >= 0)
 	{
 		b3SapAabb aabb;
 		aabb.m_min[0] = -1e30f;
 		aabb.m_min[1] = -1e30f;
 		aabb.m_min[2] = -1e30f;
 		aabb.m_minIndices[3] = 0;
-		
+
 		aabb.m_max[0] = 1e30f;
 		aabb.m_max[1] = 1e30f;
 		aabb.m_max[2] = 1e30f;
 		aabb.m_signedMaxIndices[3] = 0;
 
 		m_data->m_localShapeAABBCPU->push_back(aabb);
-//		m_data->m_localShapeAABBGPU->push_back(aabb);
+		//		m_data->m_localShapeAABBGPU->push_back(aabb);
 		clFinish(m_queue);
 	}
-	
+
 	return collidableIndex;
 }
 
-
-int b3GpuNarrowPhase::registerConvexHullShapeInternal(b3ConvexUtility* convexPtr,b3Collidable& col)
+int b3GpuNarrowPhase::registerConvexHullShapeInternal(b3ConvexUtility* convexPtr, b3Collidable& col)
 {
+	m_data->m_convexData->resize(m_data->m_numAcceleratedShapes + 1);
+	m_data->m_convexPolyhedra.resize(m_data->m_numAcceleratedShapes + 1);
 
-	m_data->m_convexData->resize(m_data->m_numAcceleratedShapes+1);
-	m_data->m_convexPolyhedra.resize(m_data->m_numAcceleratedShapes+1);
-	
-    
-	b3ConvexPolyhedronData& convex = m_data->m_convexPolyhedra.at(m_data->m_convexPolyhedra.size()-1);
+	b3ConvexPolyhedronData& convex = m_data->m_convexPolyhedra.at(m_data->m_convexPolyhedra.size() - 1);
 	convex.mC = convexPtr->mC;
 	convex.mE = convexPtr->mE;
-	convex.m_extents= convexPtr->m_extents;
+	convex.m_extents = convexPtr->m_extents;
 	convex.m_localCenter = convexPtr->m_localCenter;
 	convex.m_radius = convexPtr->m_radius;
-	
+
 	convex.m_numUniqueEdges = convexPtr->m_uniqueEdges.size();
 	int edgeOffset = m_data->m_uniqueEdges.size();
 	convex.m_uniqueEdgesOffset = edgeOffset;
-	
-	m_data->m_uniqueEdges.resize(edgeOffset+convex.m_numUniqueEdges);
-    
+
+	m_data->m_uniqueEdges.resize(edgeOffset + convex.m_numUniqueEdges);
+
 	//convex data here
 	int i;
-	for ( i=0;i<convexPtr->m_uniqueEdges.size();i++)
+	for (i = 0; i < convexPtr->m_uniqueEdges.size(); i++)
 	{
-		m_data->m_uniqueEdges[edgeOffset+i] = convexPtr->m_uniqueEdges[i];
+		m_data->m_uniqueEdges[edgeOffset + i] = convexPtr->m_uniqueEdges[i];
 	}
-    
+
 	int faceOffset = m_data->m_convexFaces.size();
 	convex.m_faceOffset = faceOffset;
 	convex.m_numFaces = convexPtr->m_faces.size();
 
-	m_data->m_convexFaces.resize(faceOffset+convex.m_numFaces);
-	
+	m_data->m_convexFaces.resize(faceOffset + convex.m_numFaces);
 
-	for (i=0;i<convexPtr->m_faces.size();i++)
+	for (i = 0; i < convexPtr->m_faces.size(); i++)
 	{
-		m_data->m_convexFaces[convex.m_faceOffset+i].m_plane = b3MakeVector3(convexPtr->m_faces[i].m_plane[0],
-																			convexPtr->m_faces[i].m_plane[1],
-																			convexPtr->m_faces[i].m_plane[2],
-																			convexPtr->m_faces[i].m_plane[3]);
+		m_data->m_convexFaces[convex.m_faceOffset + i].m_plane = b3MakeVector3(convexPtr->m_faces[i].m_plane[0],
+																			   convexPtr->m_faces[i].m_plane[1],
+																			   convexPtr->m_faces[i].m_plane[2],
+																			   convexPtr->m_faces[i].m_plane[3]);
 
-		
 		int indexOffset = m_data->m_convexIndices.size();
 		int numIndices = convexPtr->m_faces[i].m_indices.size();
-		m_data->m_convexFaces[convex.m_faceOffset+i].m_numIndices = numIndices;
-		m_data->m_convexFaces[convex.m_faceOffset+i].m_indexOffset = indexOffset;
-		m_data->m_convexIndices.resize(indexOffset+numIndices);
-		for (int p=0;p<numIndices;p++)
+		m_data->m_convexFaces[convex.m_faceOffset + i].m_numIndices = numIndices;
+		m_data->m_convexFaces[convex.m_faceOffset + i].m_indexOffset = indexOffset;
+		m_data->m_convexIndices.resize(indexOffset + numIndices);
+		for (int p = 0; p < numIndices; p++)
 		{
-			m_data->m_convexIndices[indexOffset+p] = convexPtr->m_faces[i].m_indices[p];
+			m_data->m_convexIndices[indexOffset + p] = convexPtr->m_faces[i].m_indices[p];
 		}
 	}
-    
+
 	convex.m_numVertices = convexPtr->m_vertices.size();
 	int vertexOffset = m_data->m_convexVertices.size();
-	convex.m_vertexOffset =vertexOffset;
-	
-	m_data->m_convexVertices.resize(vertexOffset+convex.m_numVertices);
-	for (int i=0;i<convexPtr->m_vertices.size();i++)
+	convex.m_vertexOffset = vertexOffset;
+
+	m_data->m_convexVertices.resize(vertexOffset + convex.m_numVertices);
+	for (int i = 0; i < convexPtr->m_vertices.size(); i++)
 	{
-		m_data->m_convexVertices[vertexOffset+i] = convexPtr->m_vertices[i];
+		m_data->m_convexVertices[vertexOffset + i] = convexPtr->m_vertices[i];
 	}
 
 	(*m_data->m_convexData)[m_data->m_numAcceleratedShapes] = convexPtr;
-	
-    
-    
+
 	return m_data->m_numAcceleratedShapes++;
 }
 
-
-int		b3GpuNarrowPhase::registerConvexHullShape(const float* vertices, int strideInBytes, int numVertices, const float* scaling)
+int b3GpuNarrowPhase::registerConvexHullShape(const float* vertices, int strideInBytes, int numVertices, const float* scaling)
 {
 	b3AlignedObjectArray<b3Vector3> verts;
 
-	unsigned char* vts = (unsigned char*) vertices;
-	for (int i=0;i<numVertices;i++)
+	unsigned char* vts = (unsigned char*)vertices;
+	for (int i = 0; i < numVertices; i++)
 	{
-		float* vertex = (float*) &vts[i*strideInBytes];
-		verts.push_back(b3MakeVector3(vertex[0]*scaling[0],vertex[1]*scaling[1],vertex[2]*scaling[2]));
+		float* vertex = (float*)&vts[i * strideInBytes];
+		verts.push_back(b3MakeVector3(vertex[0] * scaling[0], vertex[1] * scaling[1], vertex[2] * scaling[2]));
 	}
 
 	b3ConvexUtility* utilPtr = new b3ConvexUtility();
 	bool merge = true;
 	if (numVertices)
 	{
-		utilPtr->initializePolyhedralFeatures(&verts[0],verts.size(),merge);
+		utilPtr->initializePolyhedralFeatures(&verts[0], verts.size(), merge);
 	}
 
 	int collidableIndex = registerConvexHullShape(utilPtr);
@@ -358,35 +318,34 @@ int		b3GpuNarrowPhase::registerConvexHullShape(const float* vertices, int stride
 	return collidableIndex;
 }
 
-int		b3GpuNarrowPhase::registerConvexHullShape(b3ConvexUtility* utilPtr)
+int b3GpuNarrowPhase::registerConvexHullShape(b3ConvexUtility* utilPtr)
 {
 	int collidableIndex = allocateCollidable();
-	if (collidableIndex<0)
+	if (collidableIndex < 0)
 		return collidableIndex;
 
 	b3Collidable& col = getCollidableCpu(collidableIndex);
 	col.m_shapeType = SHAPE_CONVEX_HULL;
 	col.m_shapeIndex = -1;
-	
-	
+
 	{
-		b3Vector3 localCenter=b3MakeVector3(0,0,0);
-		for (int i=0;i<utilPtr->m_vertices.size();i++)
-			localCenter+=utilPtr->m_vertices[i];
-		localCenter*= (1.f/utilPtr->m_vertices.size());
+		b3Vector3 localCenter = b3MakeVector3(0, 0, 0);
+		for (int i = 0; i < utilPtr->m_vertices.size(); i++)
+			localCenter += utilPtr->m_vertices[i];
+		localCenter *= (1.f / utilPtr->m_vertices.size());
 		utilPtr->m_localCenter = localCenter;
 
-		col.m_shapeIndex = registerConvexHullShapeInternal(utilPtr,col);
+		col.m_shapeIndex = registerConvexHullShapeInternal(utilPtr, col);
 	}
 
-	if (col.m_shapeIndex>=0)
+	if (col.m_shapeIndex >= 0)
 	{
 		b3SapAabb aabb;
-		
-		b3Vector3 myAabbMin=b3MakeVector3(1e30f,1e30f,1e30f);
-		b3Vector3 myAabbMax=b3MakeVector3(-1e30f,-1e30f,-1e30f);
 
-		for (int i=0;i<utilPtr->m_vertices.size();i++)
+		b3Vector3 myAabbMin = b3MakeVector3(1e30f, 1e30f, 1e30f);
+		b3Vector3 myAabbMax = b3MakeVector3(-1e30f, -1e30f, -1e30f);
+
+		for (int i = 0; i < utilPtr->m_vertices.size(); i++)
 		{
 			myAabbMin.setMin(utilPtr->m_vertices[i]);
 			myAabbMax.setMax(utilPtr->m_vertices[i]);
@@ -402,18 +361,16 @@ int		b3GpuNarrowPhase::registerConvexHullShape(b3ConvexUtility* utilPtr)
 		aabb.m_signedMaxIndices[3] = 0;
 
 		m_data->m_localShapeAABBCPU->push_back(aabb);
-//		m_data->m_localShapeAABBGPU->push_back(aabb);
+		//		m_data->m_localShapeAABBGPU->push_back(aabb);
 	}
-	
-	return collidableIndex;
 
+	return collidableIndex;
 }
 
-int		b3GpuNarrowPhase::registerCompoundShape(b3AlignedObjectArray<b3GpuChildShape>* childShapes)
+int b3GpuNarrowPhase::registerCompoundShape(b3AlignedObjectArray<b3GpuChildShape>* childShapes)
 {
-	
 	int collidableIndex = allocateCollidable();
-	if (collidableIndex<0)
+	if (collidableIndex < 0)
 		return collidableIndex;
 
 	b3Collidable& col = getCollidableCpu(collidableIndex);
@@ -422,44 +379,41 @@ int		b3GpuNarrowPhase::registerCompoundShape(b3AlignedObjectArray<b3GpuChildShap
 	col.m_compoundBvhIndex = m_data->m_bvhInfoCPU.size();
 
 	{
-		b3Assert(col.m_shapeIndex+childShapes->size()<m_data->m_config.m_maxCompoundChildShapes);
-		for (int i=0;i<childShapes->size();i++)
+		b3Assert(col.m_shapeIndex + childShapes->size() < m_data->m_config.m_maxCompoundChildShapes);
+		for (int i = 0; i < childShapes->size(); i++)
 		{
 			m_data->m_cpuChildShapes.push_back(childShapes->at(i));
 		}
 	}
 
-
-
 	col.m_numChildShapes = childShapes->size();
-	
-	
+
 	b3SapAabb aabbLocalSpace;
-	b3Vector3 myAabbMin=b3MakeVector3(1e30f,1e30f,1e30f);
-	b3Vector3 myAabbMax=b3MakeVector3(-1e30f,-1e30f,-1e30f);
-	
+	b3Vector3 myAabbMin = b3MakeVector3(1e30f, 1e30f, 1e30f);
+	b3Vector3 myAabbMax = b3MakeVector3(-1e30f, -1e30f, -1e30f);
+
 	b3AlignedObjectArray<b3Aabb> childLocalAabbs;
 	childLocalAabbs.resize(childShapes->size());
 
 	//compute local AABB of the compound of all children
-	for (int i=0;i<childShapes->size();i++)
+	for (int i = 0; i < childShapes->size(); i++)
 	{
 		int childColIndex = childShapes->at(i).m_shapeIndex;
 		//b3Collidable& childCol = getCollidableCpu(childColIndex);
-		b3SapAabb aabbLoc =m_data->m_localShapeAABBCPU->at(childColIndex);
+		b3SapAabb aabbLoc = m_data->m_localShapeAABBCPU->at(childColIndex);
 
-		b3Vector3 childLocalAabbMin=b3MakeVector3(aabbLoc.m_min[0],aabbLoc.m_min[1],aabbLoc.m_min[2]);
-		b3Vector3 childLocalAabbMax=b3MakeVector3(aabbLoc.m_max[0],aabbLoc.m_max[1],aabbLoc.m_max[2]);
-		b3Vector3 aMin,aMax;
+		b3Vector3 childLocalAabbMin = b3MakeVector3(aabbLoc.m_min[0], aabbLoc.m_min[1], aabbLoc.m_min[2]);
+		b3Vector3 childLocalAabbMax = b3MakeVector3(aabbLoc.m_max[0], aabbLoc.m_max[1], aabbLoc.m_max[2]);
+		b3Vector3 aMin, aMax;
 		b3Scalar margin(0.f);
 		b3Transform childTr;
 		childTr.setIdentity();
 
 		childTr.setOrigin(childShapes->at(i).m_childPosition);
 		childTr.setRotation(b3Quaternion(childShapes->at(i).m_childOrientation));
-		b3TransformAabb(childLocalAabbMin,childLocalAabbMax,margin,childTr,aMin,aMax);
+		b3TransformAabb(childLocalAabbMin, childLocalAabbMax, margin, childTr, aMin, aMax);
 		myAabbMin.setMin(aMin);
-		myAabbMax.setMax(aMax);		
+		myAabbMax.setMax(aMax);
 		childLocalAabbs[i].m_min[0] = aMin[0];
 		childLocalAabbs[i].m_min[1] = aMin[1];
 		childLocalAabbs[i].m_min[2] = aMin[2];
@@ -469,36 +423,35 @@ int		b3GpuNarrowPhase::registerCompoundShape(b3AlignedObjectArray<b3GpuChildShap
 		childLocalAabbs[i].m_max[2] = aMax[2];
 		childLocalAabbs[i].m_max[3] = 0;
 	}
-	
-	aabbLocalSpace.m_min[0] = myAabbMin[0];//s_convexHeightField->m_aabb.m_min.x;
-	aabbLocalSpace.m_min[1]= myAabbMin[1];//s_convexHeightField->m_aabb.m_min.y;
-	aabbLocalSpace.m_min[2]= myAabbMin[2];//s_convexHeightField->m_aabb.m_min.z;
+
+	aabbLocalSpace.m_min[0] = myAabbMin[0];  //s_convexHeightField->m_aabb.m_min.x;
+	aabbLocalSpace.m_min[1] = myAabbMin[1];  //s_convexHeightField->m_aabb.m_min.y;
+	aabbLocalSpace.m_min[2] = myAabbMin[2];  //s_convexHeightField->m_aabb.m_min.z;
 	aabbLocalSpace.m_minIndices[3] = 0;
-	
-	aabbLocalSpace.m_max[0] = myAabbMax[0];//s_convexHeightField->m_aabb.m_max.x;
-	aabbLocalSpace.m_max[1]= myAabbMax[1];//s_convexHeightField->m_aabb.m_max.y;
-	aabbLocalSpace.m_max[2]= myAabbMax[2];//s_convexHeightField->m_aabb.m_max.z;
+
+	aabbLocalSpace.m_max[0] = myAabbMax[0];  //s_convexHeightField->m_aabb.m_max.x;
+	aabbLocalSpace.m_max[1] = myAabbMax[1];  //s_convexHeightField->m_aabb.m_max.y;
+	aabbLocalSpace.m_max[2] = myAabbMax[2];  //s_convexHeightField->m_aabb.m_max.z;
 	aabbLocalSpace.m_signedMaxIndices[3] = 0;
-	
+
 	m_data->m_localShapeAABBCPU->push_back(aabbLocalSpace);
 
-
 	b3QuantizedBvh* bvh = new b3QuantizedBvh;
-	bvh->setQuantizationValues(myAabbMin,myAabbMax);
-	QuantizedNodeArray&	nodes = bvh->getLeafNodeArray();
+	bvh->setQuantizationValues(myAabbMin, myAabbMax);
+	QuantizedNodeArray& nodes = bvh->getLeafNodeArray();
 	int numNodes = childShapes->size();
 
-	for (int i=0;i<numNodes;i++)
+	for (int i = 0; i < numNodes; i++)
 	{
 		b3QuantizedBvhNode node;
-		b3Vector3 aabbMin,aabbMax;
-		aabbMin = (b3Vector3&) childLocalAabbs[i].m_min;
-		aabbMax = (b3Vector3&) childLocalAabbs[i].m_max;
+		b3Vector3 aabbMin, aabbMax;
+		aabbMin = (b3Vector3&)childLocalAabbs[i].m_min;
+		aabbMax = (b3Vector3&)childLocalAabbs[i].m_max;
 
-		bvh->quantize(&node.m_quantizedAabbMin[0],aabbMin,0);
-		bvh->quantize(&node.m_quantizedAabbMax[0],aabbMax,1);
+		bvh->quantize(&node.m_quantizedAabbMin[0], aabbMin, 0);
+		bvh->quantize(&node.m_quantizedAabbMax[0], aabbMax, 1);
 		int partId = 0;
-		node.m_escapeIndexOrTriangleIndex = (partId<<(31-MAX_NUM_PARTS_IN_BITS)) | i;
+		node.m_escapeIndexOrTriangleIndex = (partId << (31 - MAX_NUM_PARTS_IN_BITS)) | i;
 		nodes.push_back(node);
 	}
 	bvh->buildInternal();
@@ -511,7 +464,7 @@ int		b3GpuNarrowPhase::registerCompoundShape(b3AlignedObjectArray<b3GpuChildShap
 	//void	buildInternal();
 
 	b3BvhInfo bvhInfo;
-	
+
 	bvhInfo.m_aabbMin = bvh->m_bvhAabbMin;
 	bvhInfo.m_aabbMax = bvh->m_bvhAabbMax;
 	bvhInfo.m_quantization = bvh->m_bvhQuantization;
@@ -520,80 +473,72 @@ int		b3GpuNarrowPhase::registerCompoundShape(b3AlignedObjectArray<b3GpuChildShap
 	bvhInfo.m_nodeOffset = m_data->m_treeNodesCPU.size();
 	bvhInfo.m_subTreeOffset = m_data->m_subTreesCPU.size();
 
-	int numNewNodes = 		bvh->getQuantizedNodeArray().size();
+	int numNewNodes = bvh->getQuantizedNodeArray().size();
 
-	for (int i=0;i<numNewNodes-1;i++)
+	for (int i = 0; i < numNewNodes - 1; i++)
 	{
-
 		if (bvh->getQuantizedNodeArray()[i].isLeafNode())
 		{
 			int orgIndex = bvh->getQuantizedNodeArray()[i].getTriangleIndex();
 
 			b3Vector3 nodeMinVec = bvh->unQuantize(bvh->getQuantizedNodeArray()[i].m_quantizedAabbMin);
 			b3Vector3 nodeMaxVec = bvh->unQuantize(bvh->getQuantizedNodeArray()[i].m_quantizedAabbMax);
-		
-			for (int c=0;c<3;c++)
+
+			for (int c = 0; c < 3; c++)
 			{
 				if (childLocalAabbs[orgIndex].m_min[c] < nodeMinVec[c])
 				{
-					printf("min org (%f) and new (%f) ? at i:%d,c:%d\n",childLocalAabbs[i].m_min[c],nodeMinVec[c],i,c);
+					printf("min org (%f) and new (%f) ? at i:%d,c:%d\n", childLocalAabbs[i].m_min[c], nodeMinVec[c], i, c);
 				}
 				if (childLocalAabbs[orgIndex].m_max[c] > nodeMaxVec[c])
 				{
-					printf("max org (%f) and new (%f) ? at i:%d,c:%d\n",childLocalAabbs[i].m_max[c],nodeMaxVec[c],i,c);
+					printf("max org (%f) and new (%f) ? at i:%d,c:%d\n", childLocalAabbs[i].m_max[c], nodeMaxVec[c], i, c);
 				}
-
 			}
 		}
-
 	}
 
 	m_data->m_bvhInfoCPU.push_back(bvhInfo);
 
 	int numNewSubtrees = bvh->getSubtreeInfoArray().size();
-	m_data->m_subTreesCPU.reserve(m_data->m_subTreesCPU.size()+numNewSubtrees);
-	for (int i=0;i<numNewSubtrees;i++)
+	m_data->m_subTreesCPU.reserve(m_data->m_subTreesCPU.size() + numNewSubtrees);
+	for (int i = 0; i < numNewSubtrees; i++)
 	{
 		m_data->m_subTreesCPU.push_back(bvh->getSubtreeInfoArray()[i]);
 	}
 	int numNewTreeNodes = bvh->getQuantizedNodeArray().size();
 
-	for (int i=0;i<numNewTreeNodes;i++)
+	for (int i = 0; i < numNewTreeNodes; i++)
 	{
 		m_data->m_treeNodesCPU.push_back(bvh->getQuantizedNodeArray()[i]);
 	}
 
-//	m_data->m_localShapeAABBGPU->push_back(aabbWS);
+	//	m_data->m_localShapeAABBGPU->push_back(aabbWS);
 	clFinish(m_queue);
 	return collidableIndex;
-	
 }
 
-
-int		b3GpuNarrowPhase::registerConcaveMesh(b3AlignedObjectArray<b3Vector3>* vertices, b3AlignedObjectArray<int>* indices,const float* scaling1)
+int b3GpuNarrowPhase::registerConcaveMesh(b3AlignedObjectArray<b3Vector3>* vertices, b3AlignedObjectArray<int>* indices, const float* scaling1)
 {
-	
-
-	b3Vector3 scaling=b3MakeVector3(scaling1[0],scaling1[1],scaling1[2]);
+	b3Vector3 scaling = b3MakeVector3(scaling1[0], scaling1[1], scaling1[2]);
 
 	int collidableIndex = allocateCollidable();
-	if (collidableIndex<0)
+	if (collidableIndex < 0)
 		return collidableIndex;
 
 	b3Collidable& col = getCollidableCpu(collidableIndex);
-	
+
 	col.m_shapeType = SHAPE_CONCAVE_TRIMESH;
-	col.m_shapeIndex = registerConcaveMeshShape(vertices,indices,col,scaling);
+	col.m_shapeIndex = registerConcaveMeshShape(vertices, indices, col, scaling);
 	col.m_bvhIndex = m_data->m_bvhInfoCPU.size();
-		
 
 	b3SapAabb aabb;
-	b3Vector3 myAabbMin=b3MakeVector3(1e30f,1e30f,1e30f);
-	b3Vector3 myAabbMax=b3MakeVector3(-1e30f,-1e30f,-1e30f);
+	b3Vector3 myAabbMin = b3MakeVector3(1e30f, 1e30f, 1e30f);
+	b3Vector3 myAabbMax = b3MakeVector3(-1e30f, -1e30f, -1e30f);
 
-	for (int i=0;i<vertices->size();i++)
+	for (int i = 0; i < vertices->size(); i++)
 	{
-		b3Vector3 vtx(vertices->at(i)*scaling);
+		b3Vector3 vtx(vertices->at(i) * scaling);
 		myAabbMin.setMin(vtx);
 		myAabbMax.setMax(vtx);
 	}
@@ -603,27 +548,27 @@ int		b3GpuNarrowPhase::registerConcaveMesh(b3AlignedObjectArray<b3Vector3>* vert
 	aabb.m_minIndices[3] = 0;
 
 	aabb.m_max[0] = myAabbMax[0];
-	aabb.m_max[1]= myAabbMax[1];
-	aabb.m_max[2]= myAabbMax[2];
-	aabb.m_signedMaxIndices[3]= 0;
+	aabb.m_max[1] = myAabbMax[1];
+	aabb.m_max[2] = myAabbMax[2];
+	aabb.m_signedMaxIndices[3] = 0;
 
 	m_data->m_localShapeAABBCPU->push_back(aabb);
-//	m_data->m_localShapeAABBGPU->push_back(aabb);
+	//	m_data->m_localShapeAABBGPU->push_back(aabb);
 
 	b3OptimizedBvh* bvh = new b3OptimizedBvh();
 	//void b3OptimizedBvh::build(b3StridingMeshInterface* triangles, bool useQuantizedAabbCompression, const b3Vector3& bvhAabbMin, const b3Vector3& bvhAabbMax)
-	
+
 	bool useQuantizedAabbCompression = true;
-	b3TriangleIndexVertexArray* meshInterface=new b3TriangleIndexVertexArray();
+	b3TriangleIndexVertexArray* meshInterface = new b3TriangleIndexVertexArray();
 	m_data->m_meshInterfaces.push_back(meshInterface);
 	b3IndexedMesh mesh;
-	mesh.m_numTriangles = indices->size()/3;
+	mesh.m_numTriangles = indices->size() / 3;
 	mesh.m_numVertices = vertices->size();
-	mesh.m_vertexBase = (const unsigned char *)&vertices->at(0).x;
+	mesh.m_vertexBase = (const unsigned char*)&vertices->at(0).x;
 	mesh.m_vertexStride = sizeof(b3Vector3);
-	mesh.m_triangleIndexStride = 3 * sizeof(int);// or sizeof(int)
-	mesh.m_triangleIndexBase = (const unsigned char *)&indices->at(0);
-	
+	mesh.m_triangleIndexStride = 3 * sizeof(int);  // or sizeof(int)
+	mesh.m_triangleIndexBase = (const unsigned char*)&indices->at(0);
+
 	meshInterface->addIndexedMesh(mesh);
 	bvh->build(meshInterface, useQuantizedAabbCompression, (b3Vector3&)aabb.m_min, (b3Vector3&)aabb.m_max);
 	m_data->m_bvhData.push_back(bvh);
@@ -632,7 +577,7 @@ int		b3GpuNarrowPhase::registerConcaveMesh(b3AlignedObjectArray<b3Vector3>* vert
 	int numSubTrees = bvh->getSubtreeInfoArray().size();
 
 	b3BvhInfo bvhInfo;
-	
+
 	bvhInfo.m_aabbMin = bvh->m_bvhAabbMin;
 	bvhInfo.m_aabbMax = bvh->m_bvhAabbMax;
 	bvhInfo.m_quantization = bvh->m_bvhQuantization;
@@ -643,97 +588,87 @@ int		b3GpuNarrowPhase::registerConcaveMesh(b3AlignedObjectArray<b3Vector3>* vert
 
 	m_data->m_bvhInfoCPU.push_back(bvhInfo);
 
-
 	int numNewSubtrees = bvh->getSubtreeInfoArray().size();
-	m_data->m_subTreesCPU.reserve(m_data->m_subTreesCPU.size()+numNewSubtrees);
-	for (int i=0;i<numNewSubtrees;i++)
+	m_data->m_subTreesCPU.reserve(m_data->m_subTreesCPU.size() + numNewSubtrees);
+	for (int i = 0; i < numNewSubtrees; i++)
 	{
 		m_data->m_subTreesCPU.push_back(bvh->getSubtreeInfoArray()[i]);
 	}
 	int numNewTreeNodes = bvh->getQuantizedNodeArray().size();
 
-	for (int i=0;i<numNewTreeNodes;i++)
+	for (int i = 0; i < numNewTreeNodes; i++)
 	{
 		m_data->m_treeNodesCPU.push_back(bvh->getQuantizedNodeArray()[i]);
 	}
 
-
-
-
 	return collidableIndex;
 }
 
-int b3GpuNarrowPhase::registerConcaveMeshShape(b3AlignedObjectArray<b3Vector3>* vertices, b3AlignedObjectArray<int>* indices,b3Collidable& col, const float* scaling1)
+int b3GpuNarrowPhase::registerConcaveMeshShape(b3AlignedObjectArray<b3Vector3>* vertices, b3AlignedObjectArray<int>* indices, b3Collidable& col, const float* scaling1)
 {
+	b3Vector3 scaling = b3MakeVector3(scaling1[0], scaling1[1], scaling1[2]);
 
+	m_data->m_convexData->resize(m_data->m_numAcceleratedShapes + 1);
+	m_data->m_convexPolyhedra.resize(m_data->m_numAcceleratedShapes + 1);
 
-	b3Vector3 scaling=b3MakeVector3(scaling1[0],scaling1[1],scaling1[2]);
-
-	m_data->m_convexData->resize(m_data->m_numAcceleratedShapes+1);
-	m_data->m_convexPolyhedra.resize(m_data->m_numAcceleratedShapes+1);
-	
-    
-	b3ConvexPolyhedronData& convex = m_data->m_convexPolyhedra.at(m_data->m_convexPolyhedra.size()-1);
-	convex.mC = b3MakeVector3(0,0,0);
-	convex.mE = b3MakeVector3(0,0,0);
-	convex.m_extents= b3MakeVector3(0,0,0);
-	convex.m_localCenter = b3MakeVector3(0,0,0);
+	b3ConvexPolyhedronData& convex = m_data->m_convexPolyhedra.at(m_data->m_convexPolyhedra.size() - 1);
+	convex.mC = b3MakeVector3(0, 0, 0);
+	convex.mE = b3MakeVector3(0, 0, 0);
+	convex.m_extents = b3MakeVector3(0, 0, 0);
+	convex.m_localCenter = b3MakeVector3(0, 0, 0);
 	convex.m_radius = 0.f;
-	
+
 	convex.m_numUniqueEdges = 0;
 	int edgeOffset = m_data->m_uniqueEdges.size();
 	convex.m_uniqueEdgesOffset = edgeOffset;
-	
+
 	int faceOffset = m_data->m_convexFaces.size();
 	convex.m_faceOffset = faceOffset;
-	
-	convex.m_numFaces = indices->size()/3;
-	m_data->m_convexFaces.resize(faceOffset+convex.m_numFaces);
-	m_data->m_convexIndices.reserve(convex.m_numFaces*3);
-	for (int i=0;i<convex.m_numFaces;i++)
+
+	convex.m_numFaces = indices->size() / 3;
+	m_data->m_convexFaces.resize(faceOffset + convex.m_numFaces);
+	m_data->m_convexIndices.reserve(convex.m_numFaces * 3);
+	for (int i = 0; i < convex.m_numFaces; i++)
 	{
-		if (i%256==0)
+		if (i % 256 == 0)
 		{
 			//printf("i=%d out of %d", i,convex.m_numFaces);
 		}
-		b3Vector3 vert0(vertices->at(indices->at(i*3))*scaling);
-		b3Vector3 vert1(vertices->at(indices->at(i*3+1))*scaling);
-		b3Vector3 vert2(vertices->at(indices->at(i*3+2))*scaling);
+		b3Vector3 vert0(vertices->at(indices->at(i * 3)) * scaling);
+		b3Vector3 vert1(vertices->at(indices->at(i * 3 + 1)) * scaling);
+		b3Vector3 vert2(vertices->at(indices->at(i * 3 + 2)) * scaling);
 
-		b3Vector3 normal = ((vert1-vert0).cross(vert2-vert0)).normalize();
+		b3Vector3 normal = ((vert1 - vert0).cross(vert2 - vert0)).normalize();
 		b3Scalar c = -(normal.dot(vert0));
 
-		m_data->m_convexFaces[convex.m_faceOffset+i].m_plane = b3MakeVector4(normal.x,normal.y,normal.z,c);
+		m_data->m_convexFaces[convex.m_faceOffset + i].m_plane = b3MakeVector4(normal.x, normal.y, normal.z, c);
 		int indexOffset = m_data->m_convexIndices.size();
 		int numIndices = 3;
-		m_data->m_convexFaces[convex.m_faceOffset+i].m_numIndices = numIndices;
-		m_data->m_convexFaces[convex.m_faceOffset+i].m_indexOffset = indexOffset;
-		m_data->m_convexIndices.resize(indexOffset+numIndices);
-		for (int p=0;p<numIndices;p++)
+		m_data->m_convexFaces[convex.m_faceOffset + i].m_numIndices = numIndices;
+		m_data->m_convexFaces[convex.m_faceOffset + i].m_indexOffset = indexOffset;
+		m_data->m_convexIndices.resize(indexOffset + numIndices);
+		for (int p = 0; p < numIndices; p++)
 		{
-			int vi = indices->at(i*3+p);
-			m_data->m_convexIndices[indexOffset+p] = vi;//convexPtr->m_faces[i].m_indices[p];
+			int vi = indices->at(i * 3 + p);
+			m_data->m_convexIndices[indexOffset + p] = vi;  //convexPtr->m_faces[i].m_indices[p];
 		}
 	}
-    
+
 	convex.m_numVertices = vertices->size();
 	int vertexOffset = m_data->m_convexVertices.size();
-	convex.m_vertexOffset =vertexOffset;
-	m_data->m_convexVertices.resize(vertexOffset+convex.m_numVertices);
-	for (int i=0;i<vertices->size();i++)
+	convex.m_vertexOffset = vertexOffset;
+	m_data->m_convexVertices.resize(vertexOffset + convex.m_numVertices);
+	for (int i = 0; i < vertices->size(); i++)
 	{
-		m_data->m_convexVertices[vertexOffset+i] = vertices->at(i)*scaling;
+		m_data->m_convexVertices[vertexOffset + i] = vertices->at(i) * scaling;
 	}
 
 	(*m_data->m_convexData)[m_data->m_numAcceleratedShapes] = 0;
-	
-  
+
 	return m_data->m_numAcceleratedShapes++;
 }
 
-
-
-cl_mem	b3GpuNarrowPhase::getBodiesGpu()
+cl_mem b3GpuNarrowPhase::getBodiesGpu()
 {
 	return (cl_mem)m_data->m_bodyBufferGPU->getBufferCL();
 }
@@ -743,25 +678,21 @@ const struct b3RigidBodyData* b3GpuNarrowPhase::getBodiesCpu() const
 	return &m_data->m_bodyBufferCPU->at(0);
 };
 
-
-
-
-int	b3GpuNarrowPhase::getNumBodiesGpu() const
+int b3GpuNarrowPhase::getNumBodiesGpu() const
 {
 	return m_data->m_bodyBufferGPU->size();
 }
 
-cl_mem	b3GpuNarrowPhase::getBodyInertiasGpu()
+cl_mem b3GpuNarrowPhase::getBodyInertiasGpu()
 {
 	return (cl_mem)m_data->m_inertiaBufferGPU->getBufferCL();
 }
 
-int	b3GpuNarrowPhase::getNumBodyInertiasGpu() const
+int b3GpuNarrowPhase::getNumBodyInertiasGpu() const
 {
 	return m_data->m_inertiaBufferGPU->size();
 }
 
-
 b3Collidable& b3GpuNarrowPhase::getCollidableCpu(int collidableIndex)
 {
 	return m_data->m_collidablesCPU[collidableIndex];
@@ -789,25 +720,20 @@ const struct b3SapAabb* b3GpuNarrowPhase::getLocalSpaceAabbsCpu() const
 	if (m_data->m_localShapeAABBCPU->size())
 	{
 		return &m_data->m_localShapeAABBCPU->at(0);
-	} 
+	}
 	return 0;
 }
 
-
-cl_mem	b3GpuNarrowPhase::getAabbLocalSpaceBufferGpu()
+cl_mem b3GpuNarrowPhase::getAabbLocalSpaceBufferGpu()
 {
 	return m_data->m_localShapeAABBGPU->getBufferCL();
 }
-int	b3GpuNarrowPhase::getNumCollidablesGpu() const
+int b3GpuNarrowPhase::getNumCollidablesGpu() const
 {
 	return m_data->m_collidablesGPU->size();
 }
 
-
-
-
-
-int	b3GpuNarrowPhase::getNumContactsGpu() const
+int b3GpuNarrowPhase::getNumContactsGpu() const
 {
 	return m_data->m_pBufContactBuffersGPU[m_data->m_currentContactBuffer]->size();
 }
@@ -824,37 +750,33 @@ const b3Contact4* b3GpuNarrowPhase::getContactsCPU() const
 
 void b3GpuNarrowPhase::computeContacts(cl_mem broadphasePairs, int numBroadphasePairs, cl_mem aabbsWorldSpace, int numObjects)
 {
-
 	cl_mem aabbsLocalSpace = m_data->m_localShapeAABBGPU->getBufferCL();
 
 	int nContactOut = 0;
 
 	//swap buffer
-	m_data->m_currentContactBuffer=1-m_data->m_currentContactBuffer;
+	m_data->m_currentContactBuffer = 1 - m_data->m_currentContactBuffer;
 
 	//int curSize = m_data->m_pBufContactBuffersGPU[m_data->m_currentContactBuffer]->size();
 
 	int maxTriConvexPairCapacity = m_data->m_config.m_maxTriConvexPairCapacity;
-	int numTriConvexPairsOut=0;
-	
-	b3OpenCLArray<b3Int4> broadphasePairsGPU(m_context,m_queue);
-	broadphasePairsGPU.setFromOpenCLBuffer(broadphasePairs,numBroadphasePairs);
+	int numTriConvexPairsOut = 0;
 
-	
+	b3OpenCLArray<b3Int4> broadphasePairsGPU(m_context, m_queue);
+	broadphasePairsGPU.setFromOpenCLBuffer(broadphasePairs, numBroadphasePairs);
 
+	b3OpenCLArray<b3Aabb> clAabbArrayWorldSpace(this->m_context, this->m_queue);
+	clAabbArrayWorldSpace.setFromOpenCLBuffer(aabbsWorldSpace, numObjects);
 
-	b3OpenCLArray<b3Aabb> clAabbArrayWorldSpace(this->m_context,this->m_queue);
-	clAabbArrayWorldSpace.setFromOpenCLBuffer(aabbsWorldSpace,numObjects);
-
-	b3OpenCLArray<b3Aabb> clAabbArrayLocalSpace(this->m_context,this->m_queue);
-	clAabbArrayLocalSpace.setFromOpenCLBuffer(aabbsLocalSpace,numObjects);
+	b3OpenCLArray<b3Aabb> clAabbArrayLocalSpace(this->m_context, this->m_queue);
+	clAabbArrayLocalSpace.setFromOpenCLBuffer(aabbsLocalSpace, numObjects);
 
 	m_data->m_gpuSatCollision->computeConvexConvexContactsGPUSAT(
 		&broadphasePairsGPU, numBroadphasePairs,
 		m_data->m_bodyBufferGPU,
 		m_data->m_pBufContactBuffersGPU[m_data->m_currentContactBuffer],
 		nContactOut,
-		m_data->m_pBufContactBuffersGPU[1-m_data->m_currentContactBuffer],
+		m_data->m_pBufContactBuffersGPU[1 - m_data->m_currentContactBuffer],
 		m_data->m_config.m_maxContactCapacity,
 		m_data->m_config.m_compoundPairCapacity,
 		*m_data->m_convexPolyhedraGPU,
@@ -878,8 +800,7 @@ void b3GpuNarrowPhase::computeContacts(cl_mem broadphasePairs, int numBroadphase
 		numObjects,
 		maxTriConvexPairCapacity,
 		*m_data->m_triangleConvexPairs,
-		numTriConvexPairsOut
-		);
+		numTriConvexPairsOut);
 
 	/*b3AlignedObjectArray<b3Int4> broadphasePairsCPU;
 	broadphasePairsGPU.copyToHost(broadphasePairsCPU);
@@ -892,105 +813,97 @@ const b3SapAabb& b3GpuNarrowPhase::getLocalSpaceAabb(int collidableIndex) const
 	return m_data->m_localShapeAABBCPU->at(collidableIndex);
 }
 
-
-
-
-
-int b3GpuNarrowPhase::registerRigidBody(int collidableIndex, float mass, const float* position, const float* orientation , const float* aabbMinPtr, const float* aabbMaxPtr,bool writeToGpu)
+int b3GpuNarrowPhase::registerRigidBody(int collidableIndex, float mass, const float* position, const float* orientation, const float* aabbMinPtr, const float* aabbMaxPtr, bool writeToGpu)
 {
-	b3Vector3 aabbMin=b3MakeVector3(aabbMinPtr[0],aabbMinPtr[1],aabbMinPtr[2]);
-	b3Vector3 aabbMax=b3MakeVector3(aabbMaxPtr[0],aabbMaxPtr[1],aabbMaxPtr[2]);
-	
+	b3Vector3 aabbMin = b3MakeVector3(aabbMinPtr[0], aabbMinPtr[1], aabbMinPtr[2]);
+	b3Vector3 aabbMax = b3MakeVector3(aabbMaxPtr[0], aabbMaxPtr[1], aabbMaxPtr[2]);
 
 	if (m_data->m_numAcceleratedRigidBodies >= (m_data->m_config.m_maxConvexBodies))
 	{
-		b3Error("registerRigidBody: exceeding the number of rigid bodies, %d > %d \n",m_data->m_numAcceleratedRigidBodies,m_data->m_config.m_maxConvexBodies);
+		b3Error("registerRigidBody: exceeding the number of rigid bodies, %d > %d \n", m_data->m_numAcceleratedRigidBodies, m_data->m_config.m_maxConvexBodies);
 		return -1;
 	}
-    
-	m_data->m_bodyBufferCPU->resize(m_data->m_numAcceleratedRigidBodies+1);
-    
+
+	m_data->m_bodyBufferCPU->resize(m_data->m_numAcceleratedRigidBodies + 1);
+
 	b3RigidBodyData& body = m_data->m_bodyBufferCPU->at(m_data->m_numAcceleratedRigidBodies);
-    
+
 	float friction = 1.f;
 	float restitution = 0.f;
-    
+
 	body.m_frictionCoeff = friction;
 	body.m_restituitionCoeff = restitution;
-	body.m_angVel = b3MakeVector3(0,0,0);
-	body.m_linVel=b3MakeVector3(0,0,0);//.setZero();
-	body.m_pos =b3MakeVector3(position[0],position[1],position[2]);
-	body.m_quat.setValue(orientation[0],orientation[1],orientation[2],orientation[3]);
+	body.m_angVel = b3MakeVector3(0, 0, 0);
+	body.m_linVel = b3MakeVector3(0, 0, 0);  //.setZero();
+	body.m_pos = b3MakeVector3(position[0], position[1], position[2]);
+	body.m_quat.setValue(orientation[0], orientation[1], orientation[2], orientation[3]);
 	body.m_collidableIdx = collidableIndex;
-	if (collidableIndex>=0)
+	if (collidableIndex >= 0)
 	{
-//		body.m_shapeType = m_data->m_collidablesCPU.at(collidableIndex).m_shapeType;
-	} else
+		//		body.m_shapeType = m_data->m_collidablesCPU.at(collidableIndex).m_shapeType;
+	}
+	else
 	{
-	//	body.m_shapeType = CollisionShape::SHAPE_PLANE;
+		//	body.m_shapeType = CollisionShape::SHAPE_PLANE;
 		m_planeBodyIndex = m_data->m_numAcceleratedRigidBodies;
 	}
 	//body.m_shapeType = shapeType;
-	
-	
-	body.m_invMass = mass? 1.f/mass : 0.f;
-    
+
+	body.m_invMass = mass ? 1.f / mass : 0.f;
+
 	if (writeToGpu)
 	{
-		m_data->m_bodyBufferGPU->copyFromHostPointer(&body,1,m_data->m_numAcceleratedRigidBodies);
+		m_data->m_bodyBufferGPU->copyFromHostPointer(&body, 1, m_data->m_numAcceleratedRigidBodies);
 	}
-    
-	b3InertiaData& shapeInfo = m_data->m_inertiaBufferCPU->at(m_data->m_numAcceleratedRigidBodies);
-    
-	if (mass==0.f)
-	{
-		if (m_data->m_numAcceleratedRigidBodies==0)
-			m_static0Index = 0;
-        
-		shapeInfo.m_initInvInertia.setValue(0,0,0,0,0,0,0,0,0);
-		shapeInfo.m_invInertiaWorld.setValue(0,0,0,0,0,0,0,0,0);
-	} else
-	{
-        
-		b3Assert(body.m_collidableIdx>=0);
-        
-		//approximate using the aabb of the shape
-        
-		//Aabb aabb = (*m_data->m_shapePointers)[shapeIndex]->m_aabb;
-		b3Vector3 halfExtents = (aabbMax-aabbMin);//*0.5f;//fake larger inertia makes demos more stable ;-)
-        
-		b3Vector3 localInertia;
-        
-		float lx=2.f*halfExtents[0];
-		float ly=2.f*halfExtents[1];
-		float lz=2.f*halfExtents[2];
-        
-		localInertia.setValue( (mass/12.0f) * (ly*ly + lz*lz),
-                                   (mass/12.0f) * (lx*lx + lz*lz),
-                                   (mass/12.0f) * (lx*lx + ly*ly));
-        
-		b3Vector3 invLocalInertia;
-		invLocalInertia[0] = 1.f/localInertia[0];
-		invLocalInertia[1] = 1.f/localInertia[1];
-		invLocalInertia[2] = 1.f/localInertia[2];
-		invLocalInertia[3] = 0.f;
-        
-		shapeInfo.m_initInvInertia.setValue(
-			invLocalInertia[0],		0,						0,
-			0,						invLocalInertia[1],		0,
-			0,						0,						invLocalInertia[2]);
 
-		b3Matrix3x3 m (body.m_quat);
+	b3InertiaData& shapeInfo = m_data->m_inertiaBufferCPU->at(m_data->m_numAcceleratedRigidBodies);
+
+	if (mass == 0.f)
+	{
+		if (m_data->m_numAcceleratedRigidBodies == 0)
+			m_static0Index = 0;
+
+		shapeInfo.m_initInvInertia.setValue(0, 0, 0, 0, 0, 0, 0, 0, 0);
+		shapeInfo.m_invInertiaWorld.setValue(0, 0, 0, 0, 0, 0, 0, 0, 0);
+	}
+	else
+	{
+		b3Assert(body.m_collidableIdx >= 0);
+
+		//approximate using the aabb of the shape
+
+		//Aabb aabb = (*m_data->m_shapePointers)[shapeIndex]->m_aabb;
+		b3Vector3 halfExtents = (aabbMax - aabbMin);  //*0.5f;//fake larger inertia makes demos more stable ;-)
+
+		b3Vector3 localInertia;
+
+		float lx = 2.f * halfExtents[0];
+		float ly = 2.f * halfExtents[1];
+		float lz = 2.f * halfExtents[2];
+
+		localInertia.setValue((mass / 12.0f) * (ly * ly + lz * lz),
+							  (mass / 12.0f) * (lx * lx + lz * lz),
+							  (mass / 12.0f) * (lx * lx + ly * ly));
+
+		b3Vector3 invLocalInertia;
+		invLocalInertia[0] = 1.f / localInertia[0];
+		invLocalInertia[1] = 1.f / localInertia[1];
+		invLocalInertia[2] = 1.f / localInertia[2];
+		invLocalInertia[3] = 0.f;
+
+		shapeInfo.m_initInvInertia.setValue(
+			invLocalInertia[0], 0, 0,
+			0, invLocalInertia[1], 0,
+			0, 0, invLocalInertia[2]);
+
+		b3Matrix3x3 m(body.m_quat);
 
 		shapeInfo.m_invInertiaWorld = m.scaled(invLocalInertia) * m.transpose();
-        
 	}
-    
+
 	if (writeToGpu)
-		m_data->m_inertiaBufferGPU->copyFromHostPointer(&shapeInfo,1,m_data->m_numAcceleratedRigidBodies);
-    
-    
-    
+		m_data->m_inertiaBufferGPU->copyFromHostPointer(&shapeInfo, 1, m_data->m_numAcceleratedRigidBodies);
+
 	return m_data->m_numAcceleratedRigidBodies++;
 }
 
@@ -999,15 +912,13 @@ int b3GpuNarrowPhase::getNumRigidBodies() const
 	return m_data->m_numAcceleratedRigidBodies;
 }
 
-void	b3GpuNarrowPhase::writeAllBodiesToGpu()
+void b3GpuNarrowPhase::writeAllBodiesToGpu()
 {
-
 	if (m_data->m_localShapeAABBCPU->size())
 	{
 		m_data->m_localShapeAABBGPU->copyFromHost(*m_data->m_localShapeAABBCPU);
 	}
-	
-	
+
 	m_data->m_gpuChildShapes->copyFromHost(m_data->m_cpuChildShapes);
 	m_data->m_convexFacesGPU->copyFromHost(m_data->m_convexFaces);
 	m_data->m_convexPolyhedraGPU->copyFromHost(m_data->m_convexPolyhedra);
@@ -1018,25 +929,21 @@ void	b3GpuNarrowPhase::writeAllBodiesToGpu()
 	m_data->m_treeNodesGPU->copyFromHost(m_data->m_treeNodesCPU);
 	m_data->m_subTreesGPU->copyFromHost(m_data->m_subTreesCPU);
 
-
 	m_data->m_bodyBufferGPU->resize(m_data->m_numAcceleratedRigidBodies);
 	m_data->m_inertiaBufferGPU->resize(m_data->m_numAcceleratedRigidBodies);
-    
+
 	if (m_data->m_numAcceleratedRigidBodies)
 	{
-		m_data->m_bodyBufferGPU->copyFromHostPointer(&m_data->m_bodyBufferCPU->at(0),m_data->m_numAcceleratedRigidBodies);
-		m_data->m_inertiaBufferGPU->copyFromHostPointer(&m_data->m_inertiaBufferCPU->at(0),m_data->m_numAcceleratedRigidBodies);
+		m_data->m_bodyBufferGPU->copyFromHostPointer(&m_data->m_bodyBufferCPU->at(0), m_data->m_numAcceleratedRigidBodies);
+		m_data->m_inertiaBufferGPU->copyFromHostPointer(&m_data->m_inertiaBufferCPU->at(0), m_data->m_numAcceleratedRigidBodies);
 	}
-    if (m_data->m_collidablesCPU.size())
+	if (m_data->m_collidablesCPU.size())
 	{
 		m_data->m_collidablesGPU->copyFromHost(m_data->m_collidablesCPU);
 	}
-	
-    
 }
 
-
-void	b3GpuNarrowPhase::reset()
+void b3GpuNarrowPhase::reset()
 {
 	m_data->m_numAcceleratedShapes = 0;
 	m_data->m_numAcceleratedRigidBodies = 0;
@@ -1053,21 +960,19 @@ void	b3GpuNarrowPhase::reset()
 	m_data->m_treeNodesCPU.resize(0);
 	m_data->m_subTreesCPU.resize(0);
 	m_data->m_bvhInfoCPU.resize(0);
-
 }
 
-
-void	b3GpuNarrowPhase::readbackAllBodiesToCpu()
+void b3GpuNarrowPhase::readbackAllBodiesToCpu()
 {
-	m_data->m_bodyBufferGPU->copyToHostPointer(&m_data->m_bodyBufferCPU->at(0),m_data->m_numAcceleratedRigidBodies);
+	m_data->m_bodyBufferGPU->copyToHostPointer(&m_data->m_bodyBufferCPU->at(0), m_data->m_numAcceleratedRigidBodies);
 }
 
-void b3GpuNarrowPhase::setObjectTransformCpu(float* position, float* orientation , int bodyIndex)
+void b3GpuNarrowPhase::setObjectTransformCpu(float* position, float* orientation, int bodyIndex)
 {
-	if (bodyIndex>=0 && bodyIndex<m_data->m_bodyBufferCPU->size())
+	if (bodyIndex >= 0 && bodyIndex < m_data->m_bodyBufferCPU->size())
 	{
-		m_data->m_bodyBufferCPU->at(bodyIndex).m_pos=b3MakeVector3(position[0],position[1],position[2]);
-		m_data->m_bodyBufferCPU->at(bodyIndex).m_quat.setValue(orientation[0],orientation[1],orientation[2],orientation[3]);
+		m_data->m_bodyBufferCPU->at(bodyIndex).m_pos = b3MakeVector3(position[0], position[1], position[2]);
+		m_data->m_bodyBufferCPU->at(bodyIndex).m_quat.setValue(orientation[0], orientation[1], orientation[2], orientation[3]);
 	}
 	else
 	{
@@ -1076,24 +981,25 @@ void b3GpuNarrowPhase::setObjectTransformCpu(float* position, float* orientation
 }
 void b3GpuNarrowPhase::setObjectVelocityCpu(float* linVel, float* angVel, int bodyIndex)
 {
-	if (bodyIndex>=0 && bodyIndex<m_data->m_bodyBufferCPU->size())
+	if (bodyIndex >= 0 && bodyIndex < m_data->m_bodyBufferCPU->size())
 	{
-		m_data->m_bodyBufferCPU->at(bodyIndex).m_linVel=b3MakeVector3(linVel[0],linVel[1],linVel[2]);
-		m_data->m_bodyBufferCPU->at(bodyIndex).m_angVel=b3MakeVector3(angVel[0],angVel[1],angVel[2]);
-	} else
+		m_data->m_bodyBufferCPU->at(bodyIndex).m_linVel = b3MakeVector3(linVel[0], linVel[1], linVel[2]);
+		m_data->m_bodyBufferCPU->at(bodyIndex).m_angVel = b3MakeVector3(angVel[0], angVel[1], angVel[2]);
+	}
+	else
 	{
 		b3Warning("setObjectVelocityCpu out of range.\n");
 	}
 }
 
-bool b3GpuNarrowPhase::getObjectTransformFromCpu(float* position, float* orientation , int bodyIndex) const
+bool b3GpuNarrowPhase::getObjectTransformFromCpu(float* position, float* orientation, int bodyIndex) const
 {
-	if (bodyIndex>=0 && bodyIndex<m_data->m_bodyBufferCPU->size())
+	if (bodyIndex >= 0 && bodyIndex < m_data->m_bodyBufferCPU->size())
 	{
 		position[0] = m_data->m_bodyBufferCPU->at(bodyIndex).m_pos.x;
 		position[1] = m_data->m_bodyBufferCPU->at(bodyIndex).m_pos.y;
 		position[2] = m_data->m_bodyBufferCPU->at(bodyIndex).m_pos.z;
-		position[3] = 1.f;//or 1
+		position[3] = 1.f;  //or 1
 
 		orientation[0] = m_data->m_bodyBufferCPU->at(bodyIndex).m_quat.x;
 		orientation[1] = m_data->m_bodyBufferCPU->at(bodyIndex).m_quat.y;
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuNarrowPhase.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuNarrowPhase.h
index 05ff3fd09e5..21a68de3433 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuNarrowPhase.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuNarrowPhase.h
@@ -9,11 +9,10 @@
 class b3GpuNarrowPhase
 {
 protected:
-
-	struct b3GpuNarrowPhaseInternalData*	m_data;
+	struct b3GpuNarrowPhaseInternalData* m_data;
 	int m_acceleratedCompanionShapeIndex;
 	int m_planeBodyIndex;
-	int	m_static0Index;
+	int m_static0Index;
 
 	cl_context m_context;
 	cl_device_id m_device;
@@ -23,64 +22,58 @@ protected:
 	int registerConcaveMeshShape(b3AlignedObjectArray<b3Vector3>* vertices, b3AlignedObjectArray<int>* indices, b3Collidable& col, const float* scaling);
 
 public:
-
-	
-
-
 	b3GpuNarrowPhase(cl_context vtx, cl_device_id dev, cl_command_queue q, const struct b3Config& config);
 
 	virtual ~b3GpuNarrowPhase(void);
 
-	int		registerSphereShape(float radius);
-	int		registerPlaneShape(const b3Vector3& planeNormal, float planeConstant);
+	int registerSphereShape(float radius);
+	int registerPlaneShape(const b3Vector3& planeNormal, float planeConstant);
 
 	int registerCompoundShape(b3AlignedObjectArray<b3GpuChildShape>* childShapes);
 	int registerFace(const b3Vector3& faceNormal, float faceConstant);
-	
-	int	registerConcaveMesh(b3AlignedObjectArray<b3Vector3>* vertices, b3AlignedObjectArray<int>* indices,const float* scaling);
-	
+
+	int registerConcaveMesh(b3AlignedObjectArray<b3Vector3>* vertices, b3AlignedObjectArray<int>* indices, const float* scaling);
+
 	//do they need to be merged?
-	
-	int	registerConvexHullShape(b3ConvexUtility* utilPtr);
-	int	registerConvexHullShape(const float* vertices, int strideInBytes, int numVertices, const float* scaling);
 
-	int registerRigidBody(int collidableIndex, float mass, const float* position, const float* orientation, const float* aabbMin, const float* aabbMax,bool writeToGpu);
-	void setObjectTransform(const float* position, const float* orientation , int bodyIndex);
+	int registerConvexHullShape(b3ConvexUtility* utilPtr);
+	int registerConvexHullShape(const float* vertices, int strideInBytes, int numVertices, const float* scaling);
 
-	void	writeAllBodiesToGpu();
-	void  reset();
-	void	readbackAllBodiesToCpu();
-	bool	getObjectTransformFromCpu(float* position, float* orientation , int bodyIndex) const;
+	int registerRigidBody(int collidableIndex, float mass, const float* position, const float* orientation, const float* aabbMin, const float* aabbMax, bool writeToGpu);
+	void setObjectTransform(const float* position, const float* orientation, int bodyIndex);
 
-	void setObjectTransformCpu(float* position, float* orientation , int bodyIndex);
+	void writeAllBodiesToGpu();
+	void reset();
+	void readbackAllBodiesToCpu();
+	bool getObjectTransformFromCpu(float* position, float* orientation, int bodyIndex) const;
+
+	void setObjectTransformCpu(float* position, float* orientation, int bodyIndex);
 	void setObjectVelocityCpu(float* linVel, float* angVel, int bodyIndex);
 
-	
 	virtual void computeContacts(cl_mem broadphasePairs, int numBroadphasePairs, cl_mem aabbsWorldSpace, int numObjects);
-	
 
-	cl_mem	getBodiesGpu();
+	cl_mem getBodiesGpu();
 	const struct b3RigidBodyData* getBodiesCpu() const;
 	//struct b3RigidBodyData* getBodiesCpu();
 
-	int	getNumBodiesGpu() const;
+	int getNumBodiesGpu() const;
 
-	cl_mem	getBodyInertiasGpu();
-	int	getNumBodyInertiasGpu() const;
+	cl_mem getBodyInertiasGpu();
+	int getNumBodyInertiasGpu() const;
 
-	cl_mem	getCollidablesGpu();
+	cl_mem getCollidablesGpu();
 	const struct b3Collidable* getCollidablesCpu() const;
-	int		getNumCollidablesGpu() const;
+	int getNumCollidablesGpu() const;
 
 	const struct b3SapAabb* getLocalSpaceAabbsCpu() const;
 
 	const struct b3Contact4* getContactsCPU() const;
 
-	cl_mem	getContactsGpu();
-	int	getNumContactsGpu() const;
+	cl_mem getContactsGpu();
+	int getNumContactsGpu() const;
+
+	cl_mem getAabbLocalSpaceBufferGpu();
 
-	cl_mem	getAabbLocalSpaceBufferGpu();
-	
 	int getNumRigidBodies() const;
 
 	int allocateCollidable();
@@ -92,18 +85,17 @@ public:
 	b3Collidable& getCollidableCpu(int collidableIndex);
 	const b3Collidable& getCollidableCpu(int collidableIndex) const;
 
-	const b3GpuNarrowPhaseInternalData*	getInternalData() const
+	const b3GpuNarrowPhaseInternalData* getInternalData() const
 	{
-			return m_data;
+		return m_data;
 	}
 
-	b3GpuNarrowPhaseInternalData*	getInternalData()
+	b3GpuNarrowPhaseInternalData* getInternalData()
 	{
-			return m_data;
+		return m_data;
 	}
 
 	const struct b3SapAabb& getLocalSpaceAabb(int collidableIndex) const;
 };
 
-#endif //B3_GPU_NARROWPHASE_H
-
+#endif  //B3_GPU_NARROWPHASE_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuNarrowPhaseInternalData.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuNarrowPhaseInternalData.h
index 8a7f1ea8590..716a5ea0fc8 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuNarrowPhaseInternalData.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuNarrowPhaseInternalData.h
@@ -20,57 +20,53 @@
 #include "Bullet3Common/shared/b3Int4.h"
 #include "Bullet3Common/shared/b3Int2.h"
 
-
 class b3ConvexUtility;
 
 struct b3GpuNarrowPhaseInternalData
 {
 	b3AlignedObjectArray<b3ConvexUtility*>* m_convexData;
-    
+
 	b3AlignedObjectArray<b3ConvexPolyhedronData> m_convexPolyhedra;
 	b3AlignedObjectArray<b3Vector3> m_uniqueEdges;
 	b3AlignedObjectArray<b3Vector3> m_convexVertices;
 	b3AlignedObjectArray<int> m_convexIndices;
-    
+
 	b3OpenCLArray<b3ConvexPolyhedronData>* m_convexPolyhedraGPU;
 	b3OpenCLArray<b3Vector3>* m_uniqueEdgesGPU;
 	b3OpenCLArray<b3Vector3>* m_convexVerticesGPU;
 	b3OpenCLArray<int>* m_convexIndicesGPU;
-    
-    b3OpenCLArray<b3Vector3>* m_worldVertsB1GPU;
-    b3OpenCLArray<b3Int4>* m_clippingFacesOutGPU;
-    b3OpenCLArray<b3Vector3>* m_worldNormalsAGPU;
-    b3OpenCLArray<b3Vector3>* m_worldVertsA1GPU;
-    b3OpenCLArray<b3Vector3>* m_worldVertsB2GPU;
-    
+
+	b3OpenCLArray<b3Vector3>* m_worldVertsB1GPU;
+	b3OpenCLArray<b3Int4>* m_clippingFacesOutGPU;
+	b3OpenCLArray<b3Vector3>* m_worldNormalsAGPU;
+	b3OpenCLArray<b3Vector3>* m_worldVertsA1GPU;
+	b3OpenCLArray<b3Vector3>* m_worldVertsB2GPU;
+
 	b3AlignedObjectArray<b3GpuChildShape> m_cpuChildShapes;
-	b3OpenCLArray<b3GpuChildShape>*	m_gpuChildShapes;
-    
+	b3OpenCLArray<b3GpuChildShape>* m_gpuChildShapes;
+
 	b3AlignedObjectArray<b3GpuFace> m_convexFaces;
 	b3OpenCLArray<b3GpuFace>* m_convexFacesGPU;
-    
-	struct GpuSatCollision*	m_gpuSatCollision;
-	    
-	
-	b3OpenCLArray<b3Int4>*			m_triangleConvexPairs;
-    
-	
+
+	struct GpuSatCollision* m_gpuSatCollision;
+
+	b3OpenCLArray<b3Int4>* m_triangleConvexPairs;
+
 	b3OpenCLArray<b3Contact4>* m_pBufContactBuffersGPU[2];
-	int	m_currentContactBuffer;
+	int m_currentContactBuffer;
 	b3AlignedObjectArray<b3Contact4>* m_pBufContactOutCPU;
-	
-    
+
 	b3AlignedObjectArray<b3RigidBodyData>* m_bodyBufferCPU;
 	b3OpenCLArray<b3RigidBodyData>* m_bodyBufferGPU;
-    
-	b3AlignedObjectArray<b3InertiaData>*	m_inertiaBufferCPU;
-	b3OpenCLArray<b3InertiaData>*	m_inertiaBufferGPU;
-    
+
+	b3AlignedObjectArray<b3InertiaData>* m_inertiaBufferCPU;
+	b3OpenCLArray<b3InertiaData>* m_inertiaBufferGPU;
+
 	int m_numAcceleratedShapes;
 	int m_numAcceleratedRigidBodies;
-    
-	b3AlignedObjectArray<b3Collidable>	m_collidablesCPU;
-	b3OpenCLArray<b3Collidable>*	m_collidablesGPU;
+
+	b3AlignedObjectArray<b3Collidable> m_collidablesCPU;
+	b3OpenCLArray<b3Collidable>* m_collidablesGPU;
 
 	b3OpenCLArray<b3SapAabb>* m_localShapeAABBGPU;
 	b3AlignedObjectArray<b3SapAabb>* m_localShapeAABBCPU;
@@ -78,18 +74,16 @@ struct b3GpuNarrowPhaseInternalData
 	b3AlignedObjectArray<class b3OptimizedBvh*> m_bvhData;
 	b3AlignedObjectArray<class b3TriangleIndexVertexArray*> m_meshInterfaces;
 
-	b3AlignedObjectArray<b3QuantizedBvhNode>	m_treeNodesCPU;
-	b3AlignedObjectArray<b3BvhSubtreeInfo>	m_subTreesCPU;
+	b3AlignedObjectArray<b3QuantizedBvhNode> m_treeNodesCPU;
+	b3AlignedObjectArray<b3BvhSubtreeInfo> m_subTreesCPU;
 
-	b3AlignedObjectArray<b3BvhInfo>	m_bvhInfoCPU;
-	b3OpenCLArray<b3BvhInfo>*			m_bvhInfoGPU;
-	
-	b3OpenCLArray<b3QuantizedBvhNode>*	m_treeNodesGPU;
-	b3OpenCLArray<b3BvhSubtreeInfo>*	m_subTreesGPU;
-	
+	b3AlignedObjectArray<b3BvhInfo> m_bvhInfoCPU;
+	b3OpenCLArray<b3BvhInfo>* m_bvhInfoGPU;
 
-	b3Config	m_config;
-    
+	b3OpenCLArray<b3QuantizedBvhNode>* m_treeNodesGPU;
+	b3OpenCLArray<b3BvhSubtreeInfo>* m_subTreesGPU;
+
+	b3Config m_config;
 };
 
-#endif //B3_GPU_NARROWPHASE_INTERNAL_DATA_H
+#endif  //B3_GPU_NARROWPHASE_INTERNAL_DATA_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuPgsConstraintSolver.cpp b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuPgsConstraintSolver.cpp
index 0d3d50c5481..bd9d6bb04b4 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuPgsConstraintSolver.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuPgsConstraintSolver.cpp
@@ -14,11 +14,10 @@ subject to the following restrictions:
 */
 //Originally written by Erwin Coumans
 
-
 bool useGpuInitSolverBodies = true;
 bool useGpuInfo1 = true;
-bool useGpuInfo2= true;
-bool useGpuSolveJointConstraintRows=true;
+bool useGpuInfo2 = true;
+bool useGpuSolveJointConstraintRows = true;
 bool useGpuWriteBackVelocities = true;
 bool gpuBreakConstraints = true;
 
@@ -29,27 +28,25 @@ bool gpuBreakConstraints = true;
 #include "Bullet3Dynamics/ConstraintSolver/b3TypedConstraint.h"
 #include <new>
 #include "Bullet3Common/b3AlignedObjectArray.h"
-#include <string.h> //for memset
+#include <string.h>  //for memset
 #include "Bullet3Collision/NarrowPhaseCollision/b3Contact4.h"
 #include "Bullet3OpenCL/ParallelPrimitives/b3OpenCLArray.h"
 #include "Bullet3OpenCL/ParallelPrimitives/b3LauncherCL.h"
 
 #include "Bullet3OpenCL/ParallelPrimitives/b3PrefixScanCL.h"
 
-#include "Bullet3OpenCL/RigidBody/kernels/jointSolver.h" //solveConstraintRowsCL
+#include "Bullet3OpenCL/RigidBody/kernels/jointSolver.h"  //solveConstraintRowsCL
 #include "Bullet3OpenCL/Initialize/b3OpenCLUtils.h"
 
 #define B3_JOINT_SOLVER_PATH "src/Bullet3OpenCL/RigidBody/kernels/jointSolver.cl"
 
-
 struct b3GpuPgsJacobiSolverInternalData
 {
-
 	cl_context m_context;
 	cl_device_id m_device;
 	cl_command_queue m_queue;
 
-	b3PrefixScanCL*	m_prefixScan;
+	b3PrefixScanCL* m_prefixScan;
 
 	cl_kernel m_solveJointConstraintRowsKernels;
 	cl_kernel m_initSolverBodiesKernel;
@@ -59,31 +56,27 @@ struct b3GpuPgsJacobiSolverInternalData
 	cl_kernel m_writeBackVelocitiesKernel;
 	cl_kernel m_breakViolatedConstraintsKernel;
 
-	b3OpenCLArray<unsigned int>*	m_gpuConstraintRowOffsets;
+	b3OpenCLArray<unsigned int>* m_gpuConstraintRowOffsets;
 
-	b3OpenCLArray<b3GpuSolverBody>*			m_gpuSolverBodies;
-	b3OpenCLArray<b3BatchConstraint>*		m_gpuBatchConstraints;
-	b3OpenCLArray<b3GpuSolverConstraint>*		m_gpuConstraintRows;
-	b3OpenCLArray<unsigned int>*			m_gpuConstraintInfo1;
+	b3OpenCLArray<b3GpuSolverBody>* m_gpuSolverBodies;
+	b3OpenCLArray<b3BatchConstraint>* m_gpuBatchConstraints;
+	b3OpenCLArray<b3GpuSolverConstraint>* m_gpuConstraintRows;
+	b3OpenCLArray<unsigned int>* m_gpuConstraintInfo1;
 
-//	b3AlignedObjectArray<b3GpuSolverBody>		m_cpuSolverBodies;
-	b3AlignedObjectArray<b3BatchConstraint>		m_cpuBatchConstraints;
-	b3AlignedObjectArray<b3GpuSolverConstraint>	m_cpuConstraintRows;
-	b3AlignedObjectArray<unsigned int>			m_cpuConstraintInfo1;
-	b3AlignedObjectArray<unsigned int>			m_cpuConstraintRowOffsets;
+	//	b3AlignedObjectArray<b3GpuSolverBody>		m_cpuSolverBodies;
+	b3AlignedObjectArray<b3BatchConstraint> m_cpuBatchConstraints;
+	b3AlignedObjectArray<b3GpuSolverConstraint> m_cpuConstraintRows;
+	b3AlignedObjectArray<unsigned int> m_cpuConstraintInfo1;
+	b3AlignedObjectArray<unsigned int> m_cpuConstraintRowOffsets;
 
-	b3AlignedObjectArray<b3RigidBodyData>			m_cpuBodies;
-	b3AlignedObjectArray<b3InertiaData>			m_cpuInertias;
+	b3AlignedObjectArray<b3RigidBodyData> m_cpuBodies;
+	b3AlignedObjectArray<b3InertiaData> m_cpuInertias;
 
-	
 	b3AlignedObjectArray<b3GpuGenericConstraint> m_cpuConstraints;
 
-	b3AlignedObjectArray<int>		m_batchSizes;
-
-
+	b3AlignedObjectArray<int> m_batchSizes;
 };
 
-
 /*
 static b3Transform	getWorldTransform(b3RigidBodyData* rb)
 {
@@ -100,12 +93,12 @@ static const b3Matrix3x3&	getInvInertiaTensorWorld(b3InertiaData* inertia)
 
 */
 
-static const b3Vector3&	getLinearVelocity(b3RigidBodyData* rb)
+static const b3Vector3& getLinearVelocity(b3RigidBodyData* rb)
 {
 	return rb->m_linVel;
 }
 
-static const b3Vector3&	getAngularVelocity(b3RigidBodyData* rb)
+static const b3Vector3& getAngularVelocity(b3RigidBodyData* rb)
 {
 	return rb->m_angVel;
 }
@@ -114,12 +107,9 @@ b3Vector3 getVelocityInLocalPoint(b3RigidBodyData* rb, const b3Vector3& rel_pos)
 {
 	//we also calculate lin/ang velocity for kinematic objects
 	return getLinearVelocity(rb) + getAngularVelocity(rb).cross(rel_pos);
-	
 }
 
-
-
-b3GpuPgsConstraintSolver::b3GpuPgsConstraintSolver (cl_context ctx, cl_device_id device, cl_command_queue queue,bool usePgs)
+b3GpuPgsConstraintSolver::b3GpuPgsConstraintSolver(cl_context ctx, cl_device_id device, cl_command_queue queue, bool usePgs)
 {
 	m_usePgs = usePgs;
 	m_gpuData = new b3GpuPgsJacobiSolverInternalData();
@@ -127,45 +117,40 @@ b3GpuPgsConstraintSolver::b3GpuPgsConstraintSolver (cl_context ctx, cl_device_id
 	m_gpuData->m_device = device;
 	m_gpuData->m_queue = queue;
 
-	m_gpuData->m_prefixScan = new b3PrefixScanCL(ctx,device,queue);
+	m_gpuData->m_prefixScan = new b3PrefixScanCL(ctx, device, queue);
 
-	m_gpuData->m_gpuConstraintRowOffsets = new b3OpenCLArray<unsigned int>(m_gpuData->m_context,m_gpuData->m_queue);
+	m_gpuData->m_gpuConstraintRowOffsets = new b3OpenCLArray<unsigned int>(m_gpuData->m_context, m_gpuData->m_queue);
 
-	m_gpuData->m_gpuSolverBodies = new b3OpenCLArray<b3GpuSolverBody>(m_gpuData->m_context,m_gpuData->m_queue);
-	m_gpuData->m_gpuBatchConstraints = new b3OpenCLArray<b3BatchConstraint>(m_gpuData->m_context,m_gpuData->m_queue);
-	m_gpuData->m_gpuConstraintRows = new b3OpenCLArray<b3GpuSolverConstraint>(m_gpuData->m_context,m_gpuData->m_queue);
-	m_gpuData->m_gpuConstraintInfo1 = new b3OpenCLArray<unsigned int>(m_gpuData->m_context,m_gpuData->m_queue);
-	cl_int errNum=0;
+	m_gpuData->m_gpuSolverBodies = new b3OpenCLArray<b3GpuSolverBody>(m_gpuData->m_context, m_gpuData->m_queue);
+	m_gpuData->m_gpuBatchConstraints = new b3OpenCLArray<b3BatchConstraint>(m_gpuData->m_context, m_gpuData->m_queue);
+	m_gpuData->m_gpuConstraintRows = new b3OpenCLArray<b3GpuSolverConstraint>(m_gpuData->m_context, m_gpuData->m_queue);
+	m_gpuData->m_gpuConstraintInfo1 = new b3OpenCLArray<unsigned int>(m_gpuData->m_context, m_gpuData->m_queue);
+	cl_int errNum = 0;
 
 	{
-		cl_program prog = b3OpenCLUtils::compileCLProgramFromString(m_gpuData->m_context,m_gpuData->m_device,solveConstraintRowsCL,&errNum,"",B3_JOINT_SOLVER_PATH);
+		cl_program prog = b3OpenCLUtils::compileCLProgramFromString(m_gpuData->m_context, m_gpuData->m_device, solveConstraintRowsCL, &errNum, "", B3_JOINT_SOLVER_PATH);
 		//cl_program prog = b3OpenCLUtils::compileCLProgramFromString(m_gpuData->m_context,m_gpuData->m_device,0,&errNum,"",B3_JOINT_SOLVER_PATH,true);
-		b3Assert(errNum==CL_SUCCESS);
-		m_gpuData->m_solveJointConstraintRowsKernels = b3OpenCLUtils::compileCLKernelFromString(m_gpuData->m_context, m_gpuData->m_device,solveConstraintRowsCL, "solveJointConstraintRows",&errNum,prog);
-		b3Assert(errNum==CL_SUCCESS);
-		m_gpuData->m_initSolverBodiesKernel = b3OpenCLUtils::compileCLKernelFromString(m_gpuData->m_context,m_gpuData->m_device,solveConstraintRowsCL,"initSolverBodies",&errNum,prog);
-		b3Assert(errNum==CL_SUCCESS);
-		m_gpuData->m_getInfo1Kernel = b3OpenCLUtils::compileCLKernelFromString(m_gpuData->m_context,m_gpuData->m_device,solveConstraintRowsCL,"getInfo1Kernel",&errNum,prog);
-		b3Assert(errNum==CL_SUCCESS);
-		m_gpuData->m_initBatchConstraintsKernel = b3OpenCLUtils::compileCLKernelFromString(m_gpuData->m_context,m_gpuData->m_device,solveConstraintRowsCL,"initBatchConstraintsKernel",&errNum,prog);
-		b3Assert(errNum==CL_SUCCESS);
-		m_gpuData->m_getInfo2Kernel= b3OpenCLUtils::compileCLKernelFromString(m_gpuData->m_context,m_gpuData->m_device,solveConstraintRowsCL,"getInfo2Kernel",&errNum,prog);
-		b3Assert(errNum==CL_SUCCESS);
-		m_gpuData->m_writeBackVelocitiesKernel = b3OpenCLUtils::compileCLKernelFromString(m_gpuData->m_context,m_gpuData->m_device,solveConstraintRowsCL,"writeBackVelocitiesKernel",&errNum,prog);
-		b3Assert(errNum==CL_SUCCESS);
-		m_gpuData->m_breakViolatedConstraintsKernel = b3OpenCLUtils::compileCLKernelFromString(m_gpuData->m_context,m_gpuData->m_device,solveConstraintRowsCL,"breakViolatedConstraintsKernel",&errNum,prog);
-		b3Assert(errNum==CL_SUCCESS);
-
-		
-		
+		b3Assert(errNum == CL_SUCCESS);
+		m_gpuData->m_solveJointConstraintRowsKernels = b3OpenCLUtils::compileCLKernelFromString(m_gpuData->m_context, m_gpuData->m_device, solveConstraintRowsCL, "solveJointConstraintRows", &errNum, prog);
+		b3Assert(errNum == CL_SUCCESS);
+		m_gpuData->m_initSolverBodiesKernel = b3OpenCLUtils::compileCLKernelFromString(m_gpuData->m_context, m_gpuData->m_device, solveConstraintRowsCL, "initSolverBodies", &errNum, prog);
+		b3Assert(errNum == CL_SUCCESS);
+		m_gpuData->m_getInfo1Kernel = b3OpenCLUtils::compileCLKernelFromString(m_gpuData->m_context, m_gpuData->m_device, solveConstraintRowsCL, "getInfo1Kernel", &errNum, prog);
+		b3Assert(errNum == CL_SUCCESS);
+		m_gpuData->m_initBatchConstraintsKernel = b3OpenCLUtils::compileCLKernelFromString(m_gpuData->m_context, m_gpuData->m_device, solveConstraintRowsCL, "initBatchConstraintsKernel", &errNum, prog);
+		b3Assert(errNum == CL_SUCCESS);
+		m_gpuData->m_getInfo2Kernel = b3OpenCLUtils::compileCLKernelFromString(m_gpuData->m_context, m_gpuData->m_device, solveConstraintRowsCL, "getInfo2Kernel", &errNum, prog);
+		b3Assert(errNum == CL_SUCCESS);
+		m_gpuData->m_writeBackVelocitiesKernel = b3OpenCLUtils::compileCLKernelFromString(m_gpuData->m_context, m_gpuData->m_device, solveConstraintRowsCL, "writeBackVelocitiesKernel", &errNum, prog);
+		b3Assert(errNum == CL_SUCCESS);
+		m_gpuData->m_breakViolatedConstraintsKernel = b3OpenCLUtils::compileCLKernelFromString(m_gpuData->m_context, m_gpuData->m_device, solveConstraintRowsCL, "breakViolatedConstraintsKernel", &errNum, prog);
+		b3Assert(errNum == CL_SUCCESS);
 
 		clReleaseProgram(prog);
 	}
-
-
 }
 
-b3GpuPgsConstraintSolver::~b3GpuPgsConstraintSolver ()
+b3GpuPgsConstraintSolver::~b3GpuPgsConstraintSolver()
 {
 	clReleaseKernel(m_gpuData->m_solveJointConstraintRowsKernels);
 	clReleaseKernel(m_gpuData->m_initSolverBodiesKernel);
@@ -195,16 +180,12 @@ struct b3BatchConstraint
 
 static b3AlignedObjectArray<b3BatchConstraint> batchConstraints;
 
-
-void	b3GpuPgsConstraintSolver::recomputeBatches()
+void b3GpuPgsConstraintSolver::recomputeBatches()
 {
 	m_gpuData->m_batchSizes.clear();
 }
 
-
-
-
-b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlySetup(b3OpenCLArray<b3RigidBodyData>* gpuBodies, b3OpenCLArray<b3InertiaData>* gpuInertias, int numBodies, b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints,int numConstraints,const b3ContactSolverInfo& infoGlobal)
+b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlySetup(b3OpenCLArray<b3RigidBodyData>* gpuBodies, b3OpenCLArray<b3InertiaData>* gpuInertias, int numBodies, b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints, int numConstraints, const b3ContactSolverInfo& infoGlobal)
 {
 	B3_PROFILE("GPU solveGroupCacheFriendlySetup");
 	batchConstraints.resize(numConstraints);
@@ -212,7 +193,6 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlySetup(b3OpenCLArray<b3
 	m_staticIdx = -1;
 	m_maxOverrideNumSolverIterations = 0;
 
-
 	/*	m_gpuData->m_gpuBodies->resize(numBodies);
 	m_gpuData->m_gpuBodies->copyFromHostPointer(bodies,numBodies);
 
@@ -223,15 +203,13 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlySetup(b3OpenCLArray<b3
 
 	m_gpuData->m_gpuSolverBodies->resize(numBodies);
 
-
 	m_tmpSolverBodyPool.resize(numBodies);
 	{
-		
 		if (useGpuInitSolverBodies)
 		{
 			B3_PROFILE("m_initSolverBodiesKernel");
 
-			b3LauncherCL launcher(m_gpuData->m_queue,m_gpuData->m_initSolverBodiesKernel,"m_initSolverBodiesKernel");
+			b3LauncherCL launcher(m_gpuData->m_queue, m_gpuData->m_initSolverBodiesKernel, "m_initSolverBodiesKernel");
 			launcher.setBuffer(m_gpuData->m_gpuSolverBodies->getBufferCL());
 			launcher.setBuffer(gpuBodies->getBufferCL());
 			launcher.setConst(numBodies);
@@ -239,48 +217,44 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlySetup(b3OpenCLArray<b3
 			clFinish(m_gpuData->m_queue);
 
 			//			m_gpuData->m_gpuSolverBodies->copyToHost(m_tmpSolverBodyPool);
-		} else
+		}
+		else
 		{
 			gpuBodies->copyToHost(m_gpuData->m_cpuBodies);
-			for (int i=0;i<numBodies;i++)
+			for (int i = 0; i < numBodies; i++)
 			{
-
 				b3RigidBodyData& body = m_gpuData->m_cpuBodies[i];
 				b3GpuSolverBody& solverBody = m_tmpSolverBodyPool[i];
-				initSolverBody(i,&solverBody,&body);
+				initSolverBody(i, &solverBody, &body);
 				solverBody.m_originalBodyIndex = i;
 			}
 			m_gpuData->m_gpuSolverBodies->copyFromHost(m_tmpSolverBodyPool);
 		}
 	}
 
-//	int totalBodies = 0;
+	//	int totalBodies = 0;
 	int totalNumRows = 0;
 	//b3RigidBody* rb0=0,*rb1=0;
 	//if (1)
 	{
 		{
-
-
 			//			int i;
 
 			m_tmpConstraintSizesPool.resizeNoInitialize(numConstraints);
 
 			//			b3OpenCLArray<b3GpuGenericConstraint> gpuConstraints(m_gpuData->m_context,m_gpuData->m_queue);
 
-
 			if (useGpuInfo1)
 			{
 				B3_PROFILE("info1 and init batchConstraint");
-				
+
 				m_gpuData->m_gpuConstraintInfo1->resize(numConstraints);
 
-				
 				if (1)
 				{
 					B3_PROFILE("getInfo1Kernel");
 
-					b3LauncherCL launcher(m_gpuData->m_queue,m_gpuData->m_getInfo1Kernel,"m_getInfo1Kernel");
+					b3LauncherCL launcher(m_gpuData->m_queue, m_gpuData->m_getInfo1Kernel, "m_getInfo1Kernel");
 					launcher.setBuffer(m_gpuData->m_gpuConstraintInfo1->getBufferCL());
 					launcher.setBuffer(gpuConstraints->getBufferCL());
 					launcher.setConst(numConstraints);
@@ -288,19 +262,19 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlySetup(b3OpenCLArray<b3
 					clFinish(m_gpuData->m_queue);
 				}
 
-				if (m_gpuData->m_batchSizes.size()==0)
+				if (m_gpuData->m_batchSizes.size() == 0)
 				{
 					B3_PROFILE("initBatchConstraintsKernel");
 
 					m_gpuData->m_gpuConstraintRowOffsets->resize(numConstraints);
-					unsigned int total=0;
-					m_gpuData->m_prefixScan->execute(*m_gpuData->m_gpuConstraintInfo1,*m_gpuData->m_gpuConstraintRowOffsets,numConstraints,&total);
-					unsigned int lastElem = m_gpuData->m_gpuConstraintInfo1->at(numConstraints-1);
-					totalNumRows = total+lastElem;
+					unsigned int total = 0;
+					m_gpuData->m_prefixScan->execute(*m_gpuData->m_gpuConstraintInfo1, *m_gpuData->m_gpuConstraintRowOffsets, numConstraints, &total);
+					unsigned int lastElem = m_gpuData->m_gpuConstraintInfo1->at(numConstraints - 1);
+					totalNumRows = total + lastElem;
 
 					{
 						B3_PROFILE("init batch constraints");
-						b3LauncherCL launcher(m_gpuData->m_queue,m_gpuData->m_initBatchConstraintsKernel,"m_initBatchConstraintsKernel");
+						b3LauncherCL launcher(m_gpuData->m_queue, m_gpuData->m_initBatchConstraintsKernel, "m_initBatchConstraintsKernel");
 						launcher.setBuffer(m_gpuData->m_gpuConstraintInfo1->getBufferCL());
 						launcher.setBuffer(m_gpuData->m_gpuConstraintRowOffsets->getBufferCL());
 						launcher.setBuffer(m_gpuData->m_gpuBatchConstraints->getBufferCL());
@@ -313,79 +287,74 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlySetup(b3OpenCLArray<b3
 					//assume the batching happens on CPU, so copy the data
 					m_gpuData->m_gpuBatchConstraints->copyToHost(batchConstraints);
 				}
-			} 
+			}
 			else
 			{
-				totalNumRows  = 0;
+				totalNumRows = 0;
 				gpuConstraints->copyToHost(m_gpuData->m_cpuConstraints);
 				//calculate the total number of contraint rows
-				for (int i=0;i<numConstraints;i++)
+				for (int i = 0; i < numConstraints; i++)
 				{
-					unsigned int& info1= m_tmpConstraintSizesPool[i];
+					unsigned int& info1 = m_tmpConstraintSizesPool[i];
 					//					unsigned int info1;
 					if (m_gpuData->m_cpuConstraints[i].isEnabled())
 					{
-
-						m_gpuData->m_cpuConstraints[i].getInfo1(&info1,&m_gpuData->m_cpuBodies[0]);
-					} else
+						m_gpuData->m_cpuConstraints[i].getInfo1(&info1, &m_gpuData->m_cpuBodies[0]);
+					}
+					else
 					{
 						info1 = 0;
 					}
-					
+
 					totalNumRows += info1;
 				}
 
 				m_gpuData->m_gpuBatchConstraints->copyFromHost(batchConstraints);
 				m_gpuData->m_gpuConstraintInfo1->copyFromHost(m_tmpConstraintSizesPool);
-
 			}
 			m_tmpSolverNonContactConstraintPool.resizeNoInitialize(totalNumRows);
 			m_gpuData->m_gpuConstraintRows->resize(totalNumRows);
-			
+
 			//			b3GpuConstraintArray		verify;
 
 			if (useGpuInfo2)
 			{
 				{
-						B3_PROFILE("getInfo2Kernel");
-						b3LauncherCL launcher(m_gpuData->m_queue,m_gpuData->m_getInfo2Kernel,"m_getInfo2Kernel");
-						launcher.setBuffer(m_gpuData->m_gpuConstraintRows->getBufferCL());
-						launcher.setBuffer(m_gpuData->m_gpuConstraintInfo1->getBufferCL());
-						launcher.setBuffer(m_gpuData->m_gpuConstraintRowOffsets->getBufferCL());
-						launcher.setBuffer(gpuConstraints->getBufferCL());
-						launcher.setBuffer(m_gpuData->m_gpuBatchConstraints->getBufferCL());
-						launcher.setBuffer(gpuBodies->getBufferCL());
-						launcher.setBuffer(gpuInertias->getBufferCL());
-						launcher.setBuffer(m_gpuData->m_gpuSolverBodies->getBufferCL());
-						launcher.setConst(infoGlobal.m_timeStep);
-						launcher.setConst(infoGlobal.m_erp);
-						launcher.setConst(infoGlobal.m_globalCfm);
-						launcher.setConst(infoGlobal.m_damping);
-						launcher.setConst(infoGlobal.m_numIterations);
-						launcher.setConst(numConstraints);
-						launcher.launch1D(numConstraints);
-						clFinish(m_gpuData->m_queue);
+					B3_PROFILE("getInfo2Kernel");
+					b3LauncherCL launcher(m_gpuData->m_queue, m_gpuData->m_getInfo2Kernel, "m_getInfo2Kernel");
+					launcher.setBuffer(m_gpuData->m_gpuConstraintRows->getBufferCL());
+					launcher.setBuffer(m_gpuData->m_gpuConstraintInfo1->getBufferCL());
+					launcher.setBuffer(m_gpuData->m_gpuConstraintRowOffsets->getBufferCL());
+					launcher.setBuffer(gpuConstraints->getBufferCL());
+					launcher.setBuffer(m_gpuData->m_gpuBatchConstraints->getBufferCL());
+					launcher.setBuffer(gpuBodies->getBufferCL());
+					launcher.setBuffer(gpuInertias->getBufferCL());
+					launcher.setBuffer(m_gpuData->m_gpuSolverBodies->getBufferCL());
+					launcher.setConst(infoGlobal.m_timeStep);
+					launcher.setConst(infoGlobal.m_erp);
+					launcher.setConst(infoGlobal.m_globalCfm);
+					launcher.setConst(infoGlobal.m_damping);
+					launcher.setConst(infoGlobal.m_numIterations);
+					launcher.setConst(numConstraints);
+					launcher.launch1D(numConstraints);
+					clFinish(m_gpuData->m_queue);
 
-						if (m_gpuData->m_batchSizes.size()==0)
-							m_gpuData->m_gpuBatchConstraints->copyToHost(batchConstraints);
-						//m_gpuData->m_gpuConstraintRows->copyToHost(verify);
-						//m_gpuData->m_gpuConstraintRows->copyToHost(m_tmpSolverNonContactConstraintPool);
-
-						
-
-					}
-			} 
+					if (m_gpuData->m_batchSizes.size() == 0)
+						m_gpuData->m_gpuBatchConstraints->copyToHost(batchConstraints);
+					//m_gpuData->m_gpuConstraintRows->copyToHost(verify);
+					//m_gpuData->m_gpuConstraintRows->copyToHost(m_tmpSolverNonContactConstraintPool);
+				}
+			}
 			else
 			{
-			
 				gpuInertias->copyToHost(m_gpuData->m_cpuInertias);
 
-					///setup the b3SolverConstraints
-			
-				for (int i=0;i<numConstraints;i++)
+				///setup the b3SolverConstraints
+
+				for (int i = 0; i < numConstraints; i++)
 				{
 					const int& info1 = m_tmpConstraintSizesPool[i];
-				
+
 					if (info1)
 					{
 						int constraintIndex = batchConstraints[i].m_originalConstraintIndex;
@@ -394,15 +363,13 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlySetup(b3OpenCLArray<b3
 						b3GpuSolverConstraint* currentConstraintRow = &m_tmpSolverNonContactConstraintPool[constraintRowOffset];
 						b3GpuGenericConstraint& constraint = m_gpuData->m_cpuConstraints[i];
 
-						b3RigidBodyData& rbA = m_gpuData->m_cpuBodies[ constraint.getRigidBodyA()];
+						b3RigidBodyData& rbA = m_gpuData->m_cpuBodies[constraint.getRigidBodyA()];
 						//b3RigidBody& rbA = constraint.getRigidBodyA();
-		//				b3RigidBody& rbB = constraint.getRigidBodyB();
-						b3RigidBodyData& rbB = m_gpuData->m_cpuBodies[ constraint.getRigidBodyB()];
-					
-					
+						//				b3RigidBody& rbB = constraint.getRigidBodyB();
+						b3RigidBodyData& rbB = m_gpuData->m_cpuBodies[constraint.getRigidBodyB()];
 
-						int solverBodyIdA = constraint.getRigidBodyA();//getOrInitSolverBody(constraint.getRigidBodyA(),bodies,inertias);
-						int solverBodyIdB = constraint.getRigidBodyB();//getOrInitSolverBody(constraint.getRigidBodyB(),bodies,inertias);
+						int solverBodyIdA = constraint.getRigidBodyA();  //getOrInitSolverBody(constraint.getRigidBodyA(),bodies,inertias);
+						int solverBodyIdB = constraint.getRigidBodyB();  //getOrInitSolverBody(constraint.getRigidBodyB(),bodies,inertias);
 
 						b3GpuSolverBody* bodyAPtr = &m_tmpSolverBodyPool[solverBodyIdA];
 						b3GpuSolverBody* bodyBPtr = &m_tmpSolverBodyPool[solverBodyIdB];
@@ -410,7 +377,8 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlySetup(b3OpenCLArray<b3
 						if (rbA.m_invMass)
 						{
 							batchConstraints[i].m_bodyAPtrAndSignBit = solverBodyIdA;
-						} else
+						}
+						else
 						{
 							if (!solverBodyIdA)
 								m_staticIdx = 0;
@@ -420,29 +388,28 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlySetup(b3OpenCLArray<b3
 						if (rbB.m_invMass)
 						{
 							batchConstraints[i].m_bodyBPtrAndSignBit = solverBodyIdB;
-						} else
+						}
+						else
 						{
 							if (!solverBodyIdB)
 								m_staticIdx = 0;
 							batchConstraints[i].m_bodyBPtrAndSignBit = -solverBodyIdB;
 						}
 
-
-						int overrideNumSolverIterations = 0;//constraint->getOverrideNumSolverIterations() > 0 ? constraint->getOverrideNumSolverIterations() : infoGlobal.m_numIterations;
-						if (overrideNumSolverIterations>m_maxOverrideNumSolverIterations)
+						int overrideNumSolverIterations = 0;  //constraint->getOverrideNumSolverIterations() > 0 ? constraint->getOverrideNumSolverIterations() : infoGlobal.m_numIterations;
+						if (overrideNumSolverIterations > m_maxOverrideNumSolverIterations)
 							m_maxOverrideNumSolverIterations = overrideNumSolverIterations;
 
-
 						int j;
-						for ( j=0;j<info1;j++)
+						for (j = 0; j < info1; j++)
 						{
-							memset(&currentConstraintRow[j],0,sizeof(b3GpuSolverConstraint));
-							currentConstraintRow[j].m_angularComponentA.setValue(0,0,0);
-							currentConstraintRow[j].m_angularComponentB.setValue(0,0,0);
+							memset(&currentConstraintRow[j], 0, sizeof(b3GpuSolverConstraint));
+							currentConstraintRow[j].m_angularComponentA.setValue(0, 0, 0);
+							currentConstraintRow[j].m_angularComponentB.setValue(0, 0, 0);
 							currentConstraintRow[j].m_appliedImpulse = 0.f;
 							currentConstraintRow[j].m_appliedPushImpulse = 0.f;
 							currentConstraintRow[j].m_cfm = 0.f;
-							currentConstraintRow[j].m_contactNormal.setValue(0,0,0);
+							currentConstraintRow[j].m_contactNormal.setValue(0, 0, 0);
 							currentConstraintRow[j].m_friction = 0.f;
 							currentConstraintRow[j].m_frictionIndex = 0;
 							currentConstraintRow[j].m_jacDiagABInv = 0.f;
@@ -451,13 +418,13 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlySetup(b3OpenCLArray<b3
 
 							currentConstraintRow[j].m_originalContactPoint = 0;
 							currentConstraintRow[j].m_overrideNumSolverIterations = 0;
-							currentConstraintRow[j].m_relpos1CrossNormal.setValue(0,0,0);
-							currentConstraintRow[j].m_relpos2CrossNormal.setValue(0,0,0);
+							currentConstraintRow[j].m_relpos1CrossNormal.setValue(0, 0, 0);
+							currentConstraintRow[j].m_relpos2CrossNormal.setValue(0, 0, 0);
 							currentConstraintRow[j].m_rhs = 0.f;
 							currentConstraintRow[j].m_rhsPenetration = 0.f;
 							currentConstraintRow[j].m_solverBodyIdA = 0;
 							currentConstraintRow[j].m_solverBodyIdB = 0;
-							
+
 							currentConstraintRow[j].m_lowerLimit = -B3_INFINITY;
 							currentConstraintRow[j].m_upperLimit = B3_INFINITY;
 							currentConstraintRow[j].m_appliedImpulse = 0.f;
@@ -467,26 +434,25 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlySetup(b3OpenCLArray<b3
 							currentConstraintRow[j].m_overrideNumSolverIterations = overrideNumSolverIterations;
 						}
 
-						bodyAPtr->internalGetDeltaLinearVelocity().setValue(0.f,0.f,0.f);
-						bodyAPtr->internalGetDeltaAngularVelocity().setValue(0.f,0.f,0.f);
-						bodyAPtr->internalGetPushVelocity().setValue(0.f,0.f,0.f);
-						bodyAPtr->internalGetTurnVelocity().setValue(0.f,0.f,0.f);
-						bodyBPtr->internalGetDeltaLinearVelocity().setValue(0.f,0.f,0.f);
-						bodyBPtr->internalGetDeltaAngularVelocity().setValue(0.f,0.f,0.f);
-						bodyBPtr->internalGetPushVelocity().setValue(0.f,0.f,0.f);
-						bodyBPtr->internalGetTurnVelocity().setValue(0.f,0.f,0.f);
-
+						bodyAPtr->internalGetDeltaLinearVelocity().setValue(0.f, 0.f, 0.f);
+						bodyAPtr->internalGetDeltaAngularVelocity().setValue(0.f, 0.f, 0.f);
+						bodyAPtr->internalGetPushVelocity().setValue(0.f, 0.f, 0.f);
+						bodyAPtr->internalGetTurnVelocity().setValue(0.f, 0.f, 0.f);
+						bodyBPtr->internalGetDeltaLinearVelocity().setValue(0.f, 0.f, 0.f);
+						bodyBPtr->internalGetDeltaAngularVelocity().setValue(0.f, 0.f, 0.f);
+						bodyBPtr->internalGetPushVelocity().setValue(0.f, 0.f, 0.f);
+						bodyBPtr->internalGetTurnVelocity().setValue(0.f, 0.f, 0.f);
 
 						b3GpuConstraintInfo2 info2;
-						info2.fps = 1.f/infoGlobal.m_timeStep;
+						info2.fps = 1.f / infoGlobal.m_timeStep;
 						info2.erp = infoGlobal.m_erp;
 						info2.m_J1linearAxis = currentConstraintRow->m_contactNormal;
 						info2.m_J1angularAxis = currentConstraintRow->m_relpos1CrossNormal;
 						info2.m_J2linearAxis = 0;
 						info2.m_J2angularAxis = currentConstraintRow->m_relpos2CrossNormal;
-						info2.rowskip = sizeof(b3GpuSolverConstraint)/sizeof(b3Scalar);//check this
+						info2.rowskip = sizeof(b3GpuSolverConstraint) / sizeof(b3Scalar);  //check this
 						///the size of b3GpuSolverConstraint needs be a multiple of b3Scalar
-						b3Assert(info2.rowskip*sizeof(b3Scalar)== sizeof(b3GpuSolverConstraint));
+						b3Assert(info2.rowskip * sizeof(b3Scalar) == sizeof(b3GpuSolverConstraint));
 						info2.m_constraintError = &currentConstraintRow->m_rhs;
 						currentConstraintRow->m_cfm = infoGlobal.m_globalCfm;
 						info2.m_damping = infoGlobal.m_damping;
@@ -494,47 +460,45 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlySetup(b3OpenCLArray<b3
 						info2.m_lowerLimit = &currentConstraintRow->m_lowerLimit;
 						info2.m_upperLimit = &currentConstraintRow->m_upperLimit;
 						info2.m_numIterations = infoGlobal.m_numIterations;
-						m_gpuData->m_cpuConstraints[i].getInfo2(&info2,&m_gpuData->m_cpuBodies[0]);
+						m_gpuData->m_cpuConstraints[i].getInfo2(&info2, &m_gpuData->m_cpuBodies[0]);
 
 						///finalize the constraint setup
-						for ( j=0;j<info1;j++)
+						for (j = 0; j < info1; j++)
 						{
 							b3GpuSolverConstraint& solverConstraint = currentConstraintRow[j];
 
-							if (solverConstraint.m_upperLimit>=m_gpuData->m_cpuConstraints[i].getBreakingImpulseThreshold())
+							if (solverConstraint.m_upperLimit >= m_gpuData->m_cpuConstraints[i].getBreakingImpulseThreshold())
 							{
 								solverConstraint.m_upperLimit = m_gpuData->m_cpuConstraints[i].getBreakingImpulseThreshold();
 							}
 
-							if (solverConstraint.m_lowerLimit<=-m_gpuData->m_cpuConstraints[i].getBreakingImpulseThreshold())
+							if (solverConstraint.m_lowerLimit <= -m_gpuData->m_cpuConstraints[i].getBreakingImpulseThreshold())
 							{
 								solverConstraint.m_lowerLimit = -m_gpuData->m_cpuConstraints[i].getBreakingImpulseThreshold();
 							}
 
-	//						solverConstraint.m_originalContactPoint = constraint;
-							
-							b3Matrix3x3& invInertiaWorldA= m_gpuData->m_cpuInertias[constraint.getRigidBodyA()].m_invInertiaWorld;
-							{
+							//						solverConstraint.m_originalContactPoint = constraint;
 
+							b3Matrix3x3& invInertiaWorldA = m_gpuData->m_cpuInertias[constraint.getRigidBodyA()].m_invInertiaWorld;
+							{
 								//b3Vector3 angularFactorA(1,1,1);
 								const b3Vector3& ftorqueAxis1 = solverConstraint.m_relpos1CrossNormal;
-								solverConstraint.m_angularComponentA = invInertiaWorldA*ftorqueAxis1;//*angularFactorA;
+								solverConstraint.m_angularComponentA = invInertiaWorldA * ftorqueAxis1;  //*angularFactorA;
 							}
-						
-							b3Matrix3x3& invInertiaWorldB= m_gpuData->m_cpuInertias[constraint.getRigidBodyB()].m_invInertiaWorld;
-							{
 
+							b3Matrix3x3& invInertiaWorldB = m_gpuData->m_cpuInertias[constraint.getRigidBodyB()].m_invInertiaWorld;
+							{
 								const b3Vector3& ftorqueAxis2 = solverConstraint.m_relpos2CrossNormal;
-								solverConstraint.m_angularComponentB = invInertiaWorldB*ftorqueAxis2;//*constraint.getRigidBodyB().getAngularFactor();
+								solverConstraint.m_angularComponentB = invInertiaWorldB * ftorqueAxis2;  //*constraint.getRigidBodyB().getAngularFactor();
 							}
 
 							{
 								//it is ok to use solverConstraint.m_contactNormal instead of -solverConstraint.m_contactNormal
 								//because it gets multiplied iMJlB
-								b3Vector3 iMJlA = solverConstraint.m_contactNormal*rbA.m_invMass;
-								b3Vector3 iMJaA = invInertiaWorldA*solverConstraint.m_relpos1CrossNormal;
-								b3Vector3 iMJlB = solverConstraint.m_contactNormal*rbB.m_invMass;//sign of normal?
-								b3Vector3 iMJaB = invInertiaWorldB*solverConstraint.m_relpos2CrossNormal;
+								b3Vector3 iMJlA = solverConstraint.m_contactNormal * rbA.m_invMass;
+								b3Vector3 iMJaA = invInertiaWorldA * solverConstraint.m_relpos1CrossNormal;
+								b3Vector3 iMJlB = solverConstraint.m_contactNormal * rbB.m_invMass;  //sign of normal?
+								b3Vector3 iMJaB = invInertiaWorldB * solverConstraint.m_relpos2CrossNormal;
 
 								b3Scalar sum = iMJlA.dot(solverConstraint.m_contactNormal);
 								sum += iMJaA.dot(solverConstraint.m_relpos1CrossNormal);
@@ -542,10 +506,9 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlySetup(b3OpenCLArray<b3
 								sum += iMJaB.dot(solverConstraint.m_relpos2CrossNormal);
 								b3Scalar fsum = b3Fabs(sum);
 								b3Assert(fsum > B3_EPSILON);
-								solverConstraint.m_jacDiagABInv = fsum>B3_EPSILON?b3Scalar(1.)/sum : 0.f;
+								solverConstraint.m_jacDiagABInv = fsum > B3_EPSILON ? b3Scalar(1.) / sum : 0.f;
 							}
 
-
 							///fix rhs
 							///todo: add force/torque accelerators
 							{
@@ -553,94 +516,80 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlySetup(b3OpenCLArray<b3
 								b3Scalar vel1Dotn = solverConstraint.m_contactNormal.dot(rbA.m_linVel) + solverConstraint.m_relpos1CrossNormal.dot(rbA.m_angVel);
 								b3Scalar vel2Dotn = -solverConstraint.m_contactNormal.dot(rbB.m_linVel) + solverConstraint.m_relpos2CrossNormal.dot(rbB.m_angVel);
 
-								rel_vel = vel1Dotn+vel2Dotn;
+								rel_vel = vel1Dotn + vel2Dotn;
 
 								b3Scalar restitution = 0.f;
-								b3Scalar positionalError = solverConstraint.m_rhs;//already filled in by getConstraintInfo2
-								b3Scalar	velocityError = restitution - rel_vel * info2.m_damping;
-								b3Scalar	penetrationImpulse = positionalError*solverConstraint.m_jacDiagABInv;
-								b3Scalar	velocityImpulse = velocityError *solverConstraint.m_jacDiagABInv;
-								solverConstraint.m_rhs = penetrationImpulse+velocityImpulse;
+								b3Scalar positionalError = solverConstraint.m_rhs;  //already filled in by getConstraintInfo2
+								b3Scalar velocityError = restitution - rel_vel * info2.m_damping;
+								b3Scalar penetrationImpulse = positionalError * solverConstraint.m_jacDiagABInv;
+								b3Scalar velocityImpulse = velocityError * solverConstraint.m_jacDiagABInv;
+								solverConstraint.m_rhs = penetrationImpulse + velocityImpulse;
 								solverConstraint.m_appliedImpulse = 0.f;
-
 							}
 						}
-
 					}
 				}
 
-
-
 				m_gpuData->m_gpuConstraintRows->copyFromHost(m_tmpSolverNonContactConstraintPool);
 				m_gpuData->m_gpuConstraintInfo1->copyFromHost(m_tmpConstraintSizesPool);
 
-				if (m_gpuData->m_batchSizes.size()==0)
+				if (m_gpuData->m_batchSizes.size() == 0)
 					m_gpuData->m_gpuBatchConstraints->copyFromHost(batchConstraints);
 				else
 					m_gpuData->m_gpuBatchConstraints->copyToHost(batchConstraints);
 
 				m_gpuData->m_gpuSolverBodies->copyFromHost(m_tmpSolverBodyPool);
 
-
-
-			}//end useGpuInfo2
-
-
+			}  //end useGpuInfo2
 		}
 
 #ifdef B3_SUPPORT_CONTACT_CONSTRAINTS
 		{
 			int i;
 
-			for (i=0;i<numManifolds;i++)
+			for (i = 0; i < numManifolds; i++)
 			{
 				b3Contact4& manifold = manifoldPtr[i];
-				convertContact(bodies,inertias,&manifold,infoGlobal);
+				convertContact(bodies, inertias, &manifold, infoGlobal);
 			}
 		}
-#endif //B3_SUPPORT_CONTACT_CONSTRAINTS
+#endif  //B3_SUPPORT_CONTACT_CONSTRAINTS
 	}
 
-//	b3ContactSolverInfo info = infoGlobal;
-
-
-//	int numNonContactPool = m_tmpSolverNonContactConstraintPool.size();
-//	int numConstraintPool = m_tmpSolverContactConstraintPool.size();
-//	int numFrictionPool = m_tmpSolverContactFrictionConstraintPool.size();
+	//	b3ContactSolverInfo info = infoGlobal;
 
+	//	int numNonContactPool = m_tmpSolverNonContactConstraintPool.size();
+	//	int numConstraintPool = m_tmpSolverContactConstraintPool.size();
+	//	int numFrictionPool = m_tmpSolverContactFrictionConstraintPool.size();
 
 	return 0.f;
-
 }
 
-
-
 ///a straight copy from GPU/OpenCL kernel, for debugging
-__inline void internalApplyImpulse( b3GpuSolverBody* body,  const b3Vector3& linearComponent, const b3Vector3& angularComponent,float impulseMagnitude)
+__inline void internalApplyImpulse(b3GpuSolverBody* body, const b3Vector3& linearComponent, const b3Vector3& angularComponent, float impulseMagnitude)
 {
-	body->m_deltaLinearVelocity += linearComponent*impulseMagnitude*body->m_linearFactor;
-	body->m_deltaAngularVelocity += angularComponent*(impulseMagnitude*body->m_angularFactor);
+	body->m_deltaLinearVelocity += linearComponent * impulseMagnitude * body->m_linearFactor;
+	body->m_deltaAngularVelocity += angularComponent * (impulseMagnitude * body->m_angularFactor);
 }
 
-
-void resolveSingleConstraintRowGeneric2( b3GpuSolverBody* body1,  b3GpuSolverBody* body2,  b3GpuSolverConstraint* c)
+void resolveSingleConstraintRowGeneric2(b3GpuSolverBody* body1, b3GpuSolverBody* body2, b3GpuSolverConstraint* c)
 {
-	float deltaImpulse = c->m_rhs-b3Scalar(c->m_appliedImpulse)*c->m_cfm;
-	float deltaVel1Dotn	=	b3Dot(c->m_contactNormal,body1->m_deltaLinearVelocity) 	+ b3Dot(c->m_relpos1CrossNormal,body1->m_deltaAngularVelocity);
-	float deltaVel2Dotn	=	-b3Dot(c->m_contactNormal,body2->m_deltaLinearVelocity) + b3Dot(c->m_relpos2CrossNormal,body2->m_deltaAngularVelocity);
+	float deltaImpulse = c->m_rhs - b3Scalar(c->m_appliedImpulse) * c->m_cfm;
+	float deltaVel1Dotn = b3Dot(c->m_contactNormal, body1->m_deltaLinearVelocity) + b3Dot(c->m_relpos1CrossNormal, body1->m_deltaAngularVelocity);
+	float deltaVel2Dotn = -b3Dot(c->m_contactNormal, body2->m_deltaLinearVelocity) + b3Dot(c->m_relpos2CrossNormal, body2->m_deltaAngularVelocity);
 
-	deltaImpulse	-=	deltaVel1Dotn*c->m_jacDiagABInv;
-	deltaImpulse	-=	deltaVel2Dotn*c->m_jacDiagABInv;
+	deltaImpulse -= deltaVel1Dotn * c->m_jacDiagABInv;
+	deltaImpulse -= deltaVel2Dotn * c->m_jacDiagABInv;
 
 	float sum = b3Scalar(c->m_appliedImpulse) + deltaImpulse;
 	if (sum < c->m_lowerLimit)
 	{
-		deltaImpulse = c->m_lowerLimit-b3Scalar(c->m_appliedImpulse);
+		deltaImpulse = c->m_lowerLimit - b3Scalar(c->m_appliedImpulse);
 		c->m_appliedImpulse = c->m_lowerLimit;
 	}
-	else if (sum > c->m_upperLimit) 
+	else if (sum > c->m_upperLimit)
 	{
-		deltaImpulse = c->m_upperLimit-b3Scalar(c->m_appliedImpulse);
+		deltaImpulse = c->m_upperLimit - b3Scalar(c->m_appliedImpulse);
 		c->m_appliedImpulse = c->m_upperLimit;
 	}
 	else
@@ -648,64 +597,56 @@ void resolveSingleConstraintRowGeneric2( b3GpuSolverBody* body1,  b3GpuSolverBod
 		c->m_appliedImpulse = sum;
 	}
 
-	internalApplyImpulse(body1,c->m_contactNormal*body1->m_invMass,c->m_angularComponentA,deltaImpulse);
-	internalApplyImpulse(body2,-c->m_contactNormal*body2->m_invMass,c->m_angularComponentB,deltaImpulse);
-
+	internalApplyImpulse(body1, c->m_contactNormal * body1->m_invMass, c->m_angularComponentA, deltaImpulse);
+	internalApplyImpulse(body2, -c->m_contactNormal * body2->m_invMass, c->m_angularComponentB, deltaImpulse);
 }
 
-
-
-void	b3GpuPgsConstraintSolver::initSolverBody(int bodyIndex, b3GpuSolverBody* solverBody, b3RigidBodyData* rb)
+void b3GpuPgsConstraintSolver::initSolverBody(int bodyIndex, b3GpuSolverBody* solverBody, b3RigidBodyData* rb)
 {
-
-	solverBody->m_deltaLinearVelocity.setValue(0.f,0.f,0.f);
-	solverBody->m_deltaAngularVelocity.setValue(0.f,0.f,0.f);
-	solverBody->internalGetPushVelocity().setValue(0.f,0.f,0.f);
-	solverBody->internalGetTurnVelocity().setValue(0.f,0.f,0.f);
+	solverBody->m_deltaLinearVelocity.setValue(0.f, 0.f, 0.f);
+	solverBody->m_deltaAngularVelocity.setValue(0.f, 0.f, 0.f);
+	solverBody->internalGetPushVelocity().setValue(0.f, 0.f, 0.f);
+	solverBody->internalGetTurnVelocity().setValue(0.f, 0.f, 0.f);
 
 	b3Assert(rb);
-//	solverBody->m_worldTransform = getWorldTransform(rb);
-	solverBody->internalSetInvMass(b3MakeVector3(rb->m_invMass,rb->m_invMass,rb->m_invMass));
+	//	solverBody->m_worldTransform = getWorldTransform(rb);
+	solverBody->internalSetInvMass(b3MakeVector3(rb->m_invMass, rb->m_invMass, rb->m_invMass));
 	solverBody->m_originalBodyIndex = bodyIndex;
-	solverBody->m_angularFactor = b3MakeVector3(1,1,1);
-	solverBody->m_linearFactor = b3MakeVector3(1,1,1);
+	solverBody->m_angularFactor = b3MakeVector3(1, 1, 1);
+	solverBody->m_linearFactor = b3MakeVector3(1, 1, 1);
 	solverBody->m_linearVelocity = getLinearVelocity(rb);
 	solverBody->m_angularVelocity = getAngularVelocity(rb);
 }
 
-
-void	b3GpuPgsConstraintSolver::averageVelocities()
+void b3GpuPgsConstraintSolver::averageVelocities()
 {
 }
 
-
-b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlyIterations(b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints1,int numConstraints,const b3ContactSolverInfo& infoGlobal)
+b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlyIterations(b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints1, int numConstraints, const b3ContactSolverInfo& infoGlobal)
 {
 	//only create the batches once.
 	//@todo: incrementally update batches when constraints are added/activated and/or removed/deactivated
 	B3_PROFILE("GpuSolveGroupCacheFriendlyIterations");
 
-	bool createBatches = m_gpuData->m_batchSizes.size()==0;
+	bool createBatches = m_gpuData->m_batchSizes.size() == 0;
 	{
-		
 		if (createBatches)
 		{
-			
 			m_gpuData->m_batchSizes.resize(0);
 
 			{
 				m_gpuData->m_gpuBatchConstraints->copyToHost(batchConstraints);
 
 				B3_PROFILE("batch joints");
-				b3Assert(batchConstraints.size()==numConstraints);
-				int simdWidth =numConstraints+1;
+				b3Assert(batchConstraints.size() == numConstraints);
+				int simdWidth = numConstraints + 1;
 				int numBodies = m_tmpSolverBodyPool.size();
-				sortConstraintByBatch3( &batchConstraints[0], numConstraints, simdWidth , m_staticIdx,  numBodies);
+				sortConstraintByBatch3(&batchConstraints[0], numConstraints, simdWidth, m_staticIdx, numBodies);
 
 				m_gpuData->m_gpuBatchConstraints->copyFromHost(batchConstraints);
-
 			}
-		} else
+		}
+		else
 		{
 			/*b3AlignedObjectArray<b3BatchConstraint> cpuCheckBatches;
 			m_gpuData->m_gpuBatchConstraints->copyToHost(cpuCheckBatches);
@@ -715,12 +656,11 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlyIterations(b3OpenCLArr
 			//>copyFromHost(batchConstraints);
 		}
 		int maxIterations = infoGlobal.m_numIterations;
-	
+
 		bool useBatching = true;
 
-		if (useBatching )
+		if (useBatching)
 		{
-			
 			if (!useGpuSolveJointConstraintRows)
 			{
 				B3_PROFILE("copy to host");
@@ -730,24 +670,21 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlyIterations(b3OpenCLArr
 				m_gpuData->m_gpuConstraintInfo1->copyToHost(m_gpuData->m_cpuConstraintInfo1);
 				m_gpuData->m_gpuConstraintRowOffsets->copyToHost(m_gpuData->m_cpuConstraintRowOffsets);
 				gpuConstraints1->copyToHost(m_gpuData->m_cpuConstraints);
-
 			}
 
-			for ( int iteration = 0 ; iteration< maxIterations ; iteration++)
+			for (int iteration = 0; iteration < maxIterations; iteration++)
 			{
-				
 				int batchOffset = 0;
-				int constraintOffset=0;
+				int constraintOffset = 0;
 				int numBatches = m_gpuData->m_batchSizes.size();
-				for (int bb=0;bb<numBatches;bb++)
+				for (int bb = 0; bb < numBatches; bb++)
 				{
 					int numConstraintsInBatch = m_gpuData->m_batchSizes[bb];
 
-					
 					if (useGpuSolveJointConstraintRows)
 					{
 						B3_PROFILE("solveJointConstraintRowsKernels");
-					
+
 						/*
 						__kernel void solveJointConstraintRows(__global b3GpuSolverBody* solverBodies,
 					  __global b3BatchConstraint* batchConstraints,
@@ -758,53 +695,48 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlyIterations(b3OpenCLArr
 						int batchOffset,
 						int numConstraintsInBatch*/
 
-
-						b3LauncherCL launcher(m_gpuData->m_queue,m_gpuData->m_solveJointConstraintRowsKernels,"m_solveJointConstraintRowsKernels");
+						b3LauncherCL launcher(m_gpuData->m_queue, m_gpuData->m_solveJointConstraintRowsKernels, "m_solveJointConstraintRowsKernels");
 						launcher.setBuffer(m_gpuData->m_gpuSolverBodies->getBufferCL());
 						launcher.setBuffer(m_gpuData->m_gpuBatchConstraints->getBufferCL());
 						launcher.setBuffer(m_gpuData->m_gpuConstraintRows->getBufferCL());
 						launcher.setBuffer(m_gpuData->m_gpuConstraintInfo1->getBufferCL());
 						launcher.setBuffer(m_gpuData->m_gpuConstraintRowOffsets->getBufferCL());
-						launcher.setBuffer(gpuConstraints1->getBufferCL());//to detect disabled constraints
+						launcher.setBuffer(gpuConstraints1->getBufferCL());  //to detect disabled constraints
 						launcher.setConst(batchOffset);
 						launcher.setConst(numConstraintsInBatch);
 
 						launcher.launch1D(numConstraintsInBatch);
-						
-
-					} else//useGpu
+					}
+					else  //useGpu
 					{
-						
-
-						
-						for (int b=0;b<numConstraintsInBatch;b++)
+						for (int b = 0; b < numConstraintsInBatch; b++)
 						{
-							const b3BatchConstraint& c = batchConstraints[batchOffset+b];
+							const b3BatchConstraint& c = batchConstraints[batchOffset + b];
 							/*printf("-----------\n");
 							printf("bb=%d\n",bb);
 							printf("c.batchId = %d\n", c.m_batchId);
 							*/
-							b3Assert(c.m_batchId==bb);
+							b3Assert(c.m_batchId == bb);
 							b3GpuGenericConstraint* constraint = &m_gpuData->m_cpuConstraints[c.m_originalConstraintIndex];
-							if (constraint->m_flags&B3_CONSTRAINT_FLAG_ENABLED)
+							if (constraint->m_flags & B3_CONSTRAINT_FLAG_ENABLED)
 							{
 								int numConstraintRows = m_gpuData->m_cpuConstraintInfo1[c.m_originalConstraintIndex];
 								int constraintOffset = m_gpuData->m_cpuConstraintRowOffsets[c.m_originalConstraintIndex];
-								
-								for (int jj=0;jj<numConstraintRows;jj++)
+
+								for (int jj = 0; jj < numConstraintRows; jj++)
 								{
-	//							
-									b3GpuSolverConstraint& constraint = m_tmpSolverNonContactConstraintPool[constraintOffset+jj];
+									//
+									b3GpuSolverConstraint& constraint = m_tmpSolverNonContactConstraintPool[constraintOffset + jj];
 									//resolveSingleConstraintRowGenericSIMD(m_tmpSolverBodyPool[constraint.m_solverBodyIdA],m_tmpSolverBodyPool[constraint.m_solverBodyIdB],constraint);
-									resolveSingleConstraintRowGeneric2(&m_tmpSolverBodyPool[constraint.m_solverBodyIdA],&m_tmpSolverBodyPool[constraint.m_solverBodyIdB],&constraint);
+									resolveSingleConstraintRowGeneric2(&m_tmpSolverBodyPool[constraint.m_solverBodyIdA], &m_tmpSolverBodyPool[constraint.m_solverBodyIdB], &constraint);
 								}
 							}
 						}
-					}//useGpu
-					batchOffset+=numConstraintsInBatch;
-					constraintOffset+=numConstraintsInBatch;
+					}  //useGpu
+					batchOffset += numConstraintsInBatch;
+					constraintOffset += numConstraintsInBatch;
 				}
-			}//for (int iteration...
+			}  //for (int iteration...
 
 			if (!useGpuSolveJointConstraintRows)
 			{
@@ -820,20 +752,16 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlyIterations(b3OpenCLArr
 			}
 			//int sz = sizeof(b3GpuSolverBody);
 			//printf("cpu sizeof(b3GpuSolverBody)=%d\n",sz);
-
-			
-			
-
-
-		} else
+		}
+		else
 		{
-			for ( int iteration = 0 ; iteration< maxIterations ; iteration++)
-			{			
-				int numJoints =			m_tmpSolverNonContactConstraintPool.size();
-				for (int j=0;j<numJoints;j++)
+			for (int iteration = 0; iteration < maxIterations; iteration++)
+			{
+				int numJoints = m_tmpSolverNonContactConstraintPool.size();
+				for (int j = 0; j < numJoints; j++)
 				{
 					b3GpuSolverConstraint& constraint = m_tmpSolverNonContactConstraintPool[j];
-					resolveSingleConstraintRowGeneric2(&m_tmpSolverBodyPool[constraint.m_solverBodyIdA],&m_tmpSolverBodyPool[constraint.m_solverBodyIdB],&constraint);
+					resolveSingleConstraintRowGeneric2(&m_tmpSolverBodyPool[constraint.m_solverBodyIdA], &m_tmpSolverBodyPool[constraint.m_solverBodyIdB], &constraint);
 				}
 
 				if (!m_usePgs)
@@ -842,212 +770,198 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlyIterations(b3OpenCLArr
 				}
 			}
 		}
-		
 	}
 	clFinish(m_gpuData->m_queue);
 	return 0.f;
 }
 
-
-
-
 static b3AlignedObjectArray<int> bodyUsed;
 static b3AlignedObjectArray<int> curUsed;
 
-
-
-inline int b3GpuPgsConstraintSolver::sortConstraintByBatch3( b3BatchConstraint* cs, int numConstraints, int simdWidth , int staticIdx, int numBodies)
+inline int b3GpuPgsConstraintSolver::sortConstraintByBatch3(b3BatchConstraint* cs, int numConstraints, int simdWidth, int staticIdx, int numBodies)
 {
 	//int sz = sizeof(b3BatchConstraint);
 
 	B3_PROFILE("sortConstraintByBatch3");
-	
+
 	static int maxSwaps = 0;
 	int numSwaps = 0;
 
-	curUsed.resize(2*simdWidth);
+	curUsed.resize(2 * simdWidth);
 
 	static int maxNumConstraints = 0;
-	if (maxNumConstraints<numConstraints)
+	if (maxNumConstraints < numConstraints)
 	{
 		maxNumConstraints = numConstraints;
 		//printf("maxNumConstraints  = %d\n",maxNumConstraints );
 	}
 
-	int numUsedArray = numBodies/32+1;
+	int numUsedArray = numBodies / 32 + 1;
 	bodyUsed.resize(numUsedArray);
 
-	for (int q=0;q<numUsedArray;q++)
-		bodyUsed[q]=0;
+	for (int q = 0; q < numUsedArray; q++)
+		bodyUsed[q] = 0;
 
-	
 	int curBodyUsed = 0;
 
 	int numIter = 0;
-    
-		
+
 #if defined(_DEBUG)
-	for(int i=0; i<numConstraints; i++)
+	for (int i = 0; i < numConstraints; i++)
 		cs[i].m_batchId = -1;
 #endif
-	
+
 	int numValidConstraints = 0;
-//	int unprocessedConstraintIndex = 0;
+	//	int unprocessedConstraintIndex = 0;
 
 	int batchIdx = 0;
-    
 
 	{
 		B3_PROFILE("cpu batch innerloop");
-		
-		while( numValidConstraints < numConstraints)
+
+		while (numValidConstraints < numConstraints)
 		{
 			numIter++;
 			int nCurrentBatch = 0;
 			//	clear flag
-			for(int i=0; i<curBodyUsed; i++) 
-				bodyUsed[curUsed[i]/32] = 0;
+			for (int i = 0; i < curBodyUsed; i++)
+				bodyUsed[curUsed[i] / 32] = 0;
 
-            curBodyUsed = 0;
+			curBodyUsed = 0;
 
-			for(int i=numValidConstraints; i<numConstraints; i++)
+			for (int i = numValidConstraints; i < numConstraints; i++)
 			{
 				int idx = i;
-				b3Assert( idx < numConstraints );
+				b3Assert(idx < numConstraints);
 				//	check if it can go
 				int bodyAS = cs[idx].m_bodyAPtrAndSignBit;
 				int bodyBS = cs[idx].m_bodyBPtrAndSignBit;
 				int bodyA = abs(bodyAS);
 				int bodyB = abs(bodyBS);
-				bool aIsStatic = (bodyAS<0) || bodyAS==staticIdx;
-				bool bIsStatic = (bodyBS<0) || bodyBS==staticIdx;
+				bool aIsStatic = (bodyAS < 0) || bodyAS == staticIdx;
+				bool bIsStatic = (bodyBS < 0) || bodyBS == staticIdx;
 				int aUnavailable = 0;
 				int bUnavailable = 0;
 				if (!aIsStatic)
 				{
-					aUnavailable = bodyUsed[ bodyA/32 ] & (1<<(bodyA&31));
+					aUnavailable = bodyUsed[bodyA / 32] & (1 << (bodyA & 31));
 				}
 				if (!aUnavailable)
-				if (!bIsStatic)
-				{
-					bUnavailable = bodyUsed[ bodyB/32 ] & (1<<(bodyB&31));
-				}
-                
-				if( aUnavailable==0 && bUnavailable==0 ) // ok
+					if (!bIsStatic)
+					{
+						bUnavailable = bodyUsed[bodyB / 32] & (1 << (bodyB & 31));
+					}
+
+				if (aUnavailable == 0 && bUnavailable == 0)  // ok
 				{
 					if (!aIsStatic)
 					{
-						bodyUsed[ bodyA/32 ] |= (1<<(bodyA&31));
-						curUsed[curBodyUsed++]=bodyA;
+						bodyUsed[bodyA / 32] |= (1 << (bodyA & 31));
+						curUsed[curBodyUsed++] = bodyA;
 					}
 					if (!bIsStatic)
 					{
-						bodyUsed[ bodyB/32 ] |= (1<<(bodyB&31));
-						curUsed[curBodyUsed++]=bodyB;
+						bodyUsed[bodyB / 32] |= (1 << (bodyB & 31));
+						curUsed[curBodyUsed++] = bodyB;
 					}
 
 					cs[idx].m_batchId = batchIdx;
 
-					if (i!=numValidConstraints)
+					if (i != numValidConstraints)
 					{
-						b3Swap(cs[i],cs[numValidConstraints]);
+						b3Swap(cs[i], cs[numValidConstraints]);
 						numSwaps++;
 					}
 
 					numValidConstraints++;
 					{
 						nCurrentBatch++;
-						if( nCurrentBatch == simdWidth )
+						if (nCurrentBatch == simdWidth)
 						{
 							nCurrentBatch = 0;
-							for(int i=0; i<curBodyUsed; i++) 
-								bodyUsed[curUsed[i]/32] = 0;
+							for (int i = 0; i < curBodyUsed; i++)
+								bodyUsed[curUsed[i] / 32] = 0;
 							curBodyUsed = 0;
 						}
 					}
 				}
 			}
 			m_gpuData->m_batchSizes.push_back(nCurrentBatch);
-			batchIdx ++;
+			batchIdx++;
 		}
 	}
-	
+
 #if defined(_DEBUG)
-    //		debugPrintf( "nBatches: %d\n", batchIdx );
-	for(int i=0; i<numConstraints; i++)
-    {
-        b3Assert( cs[i].m_batchId != -1 );
-    }
+	//		debugPrintf( "nBatches: %d\n", batchIdx );
+	for (int i = 0; i < numConstraints; i++)
+	{
+		b3Assert(cs[i].m_batchId != -1);
+	}
 #endif
 
-	if (maxSwaps<numSwaps)
+	if (maxSwaps < numSwaps)
 	{
 		maxSwaps = numSwaps;
 		//printf("maxSwaps = %d\n", maxSwaps);
 	}
-	
+
 	return batchIdx;
 }
 
-
 /// b3PgsJacobiSolver Sequentially applies impulses
-b3Scalar b3GpuPgsConstraintSolver::solveGroup(b3OpenCLArray<b3RigidBodyData>* gpuBodies, b3OpenCLArray<b3InertiaData>* gpuInertias, 
-				int numBodies, b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints,int numConstraints, const b3ContactSolverInfo& infoGlobal)
+b3Scalar b3GpuPgsConstraintSolver::solveGroup(b3OpenCLArray<b3RigidBodyData>* gpuBodies, b3OpenCLArray<b3InertiaData>* gpuInertias,
+											  int numBodies, b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints, int numConstraints, const b3ContactSolverInfo& infoGlobal)
 {
-
 	B3_PROFILE("solveJoints");
 	//you need to provide at least some bodies
-	
-	solveGroupCacheFriendlySetup( gpuBodies, gpuInertias,numBodies,gpuConstraints, numConstraints,infoGlobal);
 
-	solveGroupCacheFriendlyIterations(gpuConstraints, numConstraints,infoGlobal);
+	solveGroupCacheFriendlySetup(gpuBodies, gpuInertias, numBodies, gpuConstraints, numConstraints, infoGlobal);
+
+	solveGroupCacheFriendlyIterations(gpuConstraints, numConstraints, infoGlobal);
+
+	solveGroupCacheFriendlyFinish(gpuBodies, gpuInertias, numBodies, gpuConstraints, numConstraints, infoGlobal);
 
-	solveGroupCacheFriendlyFinish(gpuBodies, gpuInertias,numBodies, gpuConstraints, numConstraints, infoGlobal);
-	
 	return 0.f;
 }
 
-void	b3GpuPgsConstraintSolver::solveJoints(int numBodies, b3OpenCLArray<b3RigidBodyData>* gpuBodies, b3OpenCLArray<b3InertiaData>* gpuInertias, 
-				int numConstraints, b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints)
+void b3GpuPgsConstraintSolver::solveJoints(int numBodies, b3OpenCLArray<b3RigidBodyData>* gpuBodies, b3OpenCLArray<b3InertiaData>* gpuInertias,
+										   int numConstraints, b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints)
 {
 	b3ContactSolverInfo infoGlobal;
 	infoGlobal.m_splitImpulse = false;
-	infoGlobal.m_timeStep = 1.f/60.f;
-	infoGlobal.m_numIterations = 4;//4;
-//	infoGlobal.m_solverMode|=B3_SOLVER_USE_2_FRICTION_DIRECTIONS|B3_SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS|B3_SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION;
+	infoGlobal.m_timeStep = 1.f / 60.f;
+	infoGlobal.m_numIterations = 4;  //4;
+									 //	infoGlobal.m_solverMode|=B3_SOLVER_USE_2_FRICTION_DIRECTIONS|B3_SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS|B3_SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION;
 	//infoGlobal.m_solverMode|=B3_SOLVER_USE_2_FRICTION_DIRECTIONS|B3_SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS;
-	infoGlobal.m_solverMode|=B3_SOLVER_USE_2_FRICTION_DIRECTIONS;
+	infoGlobal.m_solverMode |= B3_SOLVER_USE_2_FRICTION_DIRECTIONS;
 
 	//if (infoGlobal.m_solverMode & B3_SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS)
 	//if ((infoGlobal.m_solverMode & B3_SOLVER_USE_2_FRICTION_DIRECTIONS) && (infoGlobal.m_solverMode & B3_SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION))
-				
-
-	solveGroup(gpuBodies,gpuInertias,numBodies,gpuConstraints,numConstraints,infoGlobal);
 
+	solveGroup(gpuBodies, gpuInertias, numBodies, gpuConstraints, numConstraints, infoGlobal);
 }
 
 //b3AlignedObjectArray<b3RigidBodyData> testBodies;
 
-
-b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlyFinish(b3OpenCLArray<b3RigidBodyData>* gpuBodies,b3OpenCLArray<b3InertiaData>* gpuInertias,int numBodies,b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints,int numConstraints,const b3ContactSolverInfo& infoGlobal)
+b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlyFinish(b3OpenCLArray<b3RigidBodyData>* gpuBodies, b3OpenCLArray<b3InertiaData>* gpuInertias, int numBodies, b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints, int numConstraints, const b3ContactSolverInfo& infoGlobal)
 {
 	B3_PROFILE("solveGroupCacheFriendlyFinish");
-//	int numPoolConstraints = m_tmpSolverContactConstraintPool.size();
-//	int i,j;
-
+	//	int numPoolConstraints = m_tmpSolverContactConstraintPool.size();
+	//	int i,j;
 
 	{
 		if (gpuBreakConstraints)
 		{
 			B3_PROFILE("breakViolatedConstraintsKernel");
-			b3LauncherCL launcher(m_gpuData->m_queue,m_gpuData->m_breakViolatedConstraintsKernel,"m_breakViolatedConstraintsKernel");
+			b3LauncherCL launcher(m_gpuData->m_queue, m_gpuData->m_breakViolatedConstraintsKernel, "m_breakViolatedConstraintsKernel");
 			launcher.setBuffer(gpuConstraints->getBufferCL());
 			launcher.setBuffer(m_gpuData->m_gpuConstraintInfo1->getBufferCL());
 			launcher.setBuffer(m_gpuData->m_gpuConstraintRowOffsets->getBufferCL());
 			launcher.setBuffer(m_gpuData->m_gpuConstraintRows->getBufferCL());
 			launcher.setConst(numConstraints);
 			launcher.launch1D(numConstraints);
-		} else
+		}
+		else
 		{
 			gpuConstraints->copyToHost(m_gpuData->m_cpuConstraints);
 			m_gpuData->m_gpuBatchConstraints->copyToHost(m_gpuData->m_cpuBatchConstraints);
@@ -1056,31 +970,28 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlyFinish(b3OpenCLArray<b
 			m_gpuData->m_gpuConstraintInfo1->copyToHost(m_gpuData->m_cpuConstraintInfo1);
 			m_gpuData->m_gpuConstraintRowOffsets->copyToHost(m_gpuData->m_cpuConstraintRowOffsets);
 
-			for (int cid=0;cid<numConstraints;cid++)
+			for (int cid = 0; cid < numConstraints; cid++)
 			{
 				int originalConstraintIndex = batchConstraints[cid].m_originalConstraintIndex;
 				int constraintRowOffset = m_gpuData->m_cpuConstraintRowOffsets[originalConstraintIndex];
 				int numRows = m_gpuData->m_cpuConstraintInfo1[originalConstraintIndex];
 				if (numRows)
 				{
-		
-				//	printf("cid=%d, breakingThreshold =%f\n",cid,breakingThreshold);
-					for (int i=0;i<numRows;i++)
+					//	printf("cid=%d, breakingThreshold =%f\n",cid,breakingThreshold);
+					for (int i = 0; i < numRows; i++)
 					{
-						int rowIndex =constraintRowOffset+i;
+						int rowIndex = constraintRowOffset + i;
 						int orgConstraintIndex = m_gpuData->m_cpuConstraintRows[rowIndex].m_originalConstraintIndex;
 						float breakingThreshold = m_gpuData->m_cpuConstraints[orgConstraintIndex].m_breakingImpulseThreshold;
-					//	printf("rows[%d].m_appliedImpulse=%f\n",rowIndex,rows[rowIndex].m_appliedImpulse);
+						//	printf("rows[%d].m_appliedImpulse=%f\n",rowIndex,rows[rowIndex].m_appliedImpulse);
 						if (b3Fabs(m_gpuData->m_cpuConstraintRows[rowIndex].m_appliedImpulse) >= breakingThreshold)
 						{
-
-							m_gpuData->m_cpuConstraints[orgConstraintIndex].m_flags =0;//&= ~B3_CONSTRAINT_FLAG_ENABLED;
+							m_gpuData->m_cpuConstraints[orgConstraintIndex].m_flags = 0;  //&= ~B3_CONSTRAINT_FLAG_ENABLED;
 						}
 					}
 				}
 			}
 
-
 			gpuConstraints->copyFromHost(m_gpuData->m_cpuConstraints);
 		}
 	}
@@ -1090,28 +1001,27 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlyFinish(b3OpenCLArray<b
 		{
 			B3_PROFILE("GPU write back velocities and transforms");
 
-			b3LauncherCL launcher(m_gpuData->m_queue,m_gpuData->m_writeBackVelocitiesKernel,"m_writeBackVelocitiesKernel");
+			b3LauncherCL launcher(m_gpuData->m_queue, m_gpuData->m_writeBackVelocitiesKernel, "m_writeBackVelocitiesKernel");
 			launcher.setBuffer(gpuBodies->getBufferCL());
 			launcher.setBuffer(m_gpuData->m_gpuSolverBodies->getBufferCL());
 			launcher.setConst(numBodies);
 			launcher.launch1D(numBodies);
 			clFinish(m_gpuData->m_queue);
-//			m_gpuData->m_gpuSolverBodies->copyToHost(m_tmpSolverBodyPool);
-//			m_gpuData->m_gpuBodies->copyToHostPointer(bodies,numBodies);
+			//			m_gpuData->m_gpuSolverBodies->copyToHost(m_tmpSolverBodyPool);
+			//			m_gpuData->m_gpuBodies->copyToHostPointer(bodies,numBodies);
 			//m_gpuData->m_gpuBodies->copyToHost(testBodies);
-
-		} 
+		}
 		else
 		{
 			B3_PROFILE("CPU write back velocities and transforms");
 
 			m_gpuData->m_gpuSolverBodies->copyToHost(m_tmpSolverBodyPool);
 			gpuBodies->copyToHost(m_gpuData->m_cpuBodies);
-			for ( int i=0;i<m_tmpSolverBodyPool.size();i++)
+			for (int i = 0; i < m_tmpSolverBodyPool.size(); i++)
 			{
 				int bodyIndex = m_tmpSolverBodyPool[i].m_originalBodyIndex;
 				//printf("bodyIndex=%d\n",bodyIndex);
-				b3Assert(i==bodyIndex);
+				b3Assert(i == bodyIndex);
 
 				b3RigidBodyData* body = &m_gpuData->m_cpuBodies[bodyIndex];
 				if (body->m_invMass)
@@ -1125,11 +1035,12 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlyFinish(b3OpenCLArray<b
 					{
 						body->m_linVel = m_tmpSolverBodyPool[i].m_linearVelocity;
 						body->m_angVel = m_tmpSolverBodyPool[i].m_angularVelocity;
-					} else
+					}
+					else
 					{
 						b3Assert(0);
 					}
-	/*			
+					/*			
 					if (infoGlobal.m_splitImpulse)
 					{
 						body->m_pos = m_tmpSolverBodyPool[i].m_worldTransform.getOrigin();
@@ -1139,10 +1050,9 @@ b3Scalar b3GpuPgsConstraintSolver::solveGroupCacheFriendlyFinish(b3OpenCLArray<b
 					}
 					*/
 				}
-			}//for
+			}  //for
 
 			gpuBodies->copyFromHost(m_gpuData->m_cpuBodies);
-
 		}
 	}
 
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuPgsConstraintSolver.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuPgsConstraintSolver.h
index ec0e3f73d63..00bc544f022 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuPgsConstraintSolver.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuPgsConstraintSolver.h
@@ -19,7 +19,6 @@ subject to the following restrictions:
 struct b3Contact4;
 struct b3ContactPoint;
 
-
 class b3Dispatcher;
 
 #include "Bullet3Dynamics/ConstraintSolver/b3TypedConstraint.h"
@@ -38,41 +37,40 @@ class b3GpuPgsConstraintSolver
 protected:
 	int m_staticIdx;
 	struct b3GpuPgsJacobiSolverInternalData* m_gpuData;
-	protected:
-	b3AlignedObjectArray<b3GpuSolverBody>      m_tmpSolverBodyPool;
-	b3GpuConstraintArray			m_tmpSolverContactConstraintPool;
-	b3GpuConstraintArray			m_tmpSolverNonContactConstraintPool;
-	b3GpuConstraintArray			m_tmpSolverContactFrictionConstraintPool;
-	b3GpuConstraintArray			m_tmpSolverContactRollingFrictionConstraintPool;
+
+protected:
+	b3AlignedObjectArray<b3GpuSolverBody> m_tmpSolverBodyPool;
+	b3GpuConstraintArray m_tmpSolverContactConstraintPool;
+	b3GpuConstraintArray m_tmpSolverNonContactConstraintPool;
+	b3GpuConstraintArray m_tmpSolverContactFrictionConstraintPool;
+	b3GpuConstraintArray m_tmpSolverContactRollingFrictionConstraintPool;
 
 	b3AlignedObjectArray<unsigned int> m_tmpConstraintSizesPool;
-	
 
-	bool						m_usePgs;
-	void						averageVelocities();
+	bool m_usePgs;
+	void averageVelocities();
 
-	int							m_maxOverrideNumSolverIterations;
+	int m_maxOverrideNumSolverIterations;
 
-	int							m_numSplitImpulseRecoveries;
+	int m_numSplitImpulseRecoveries;
 
-//	int	getOrInitSolverBody(int bodyIndex, b3RigidBodyData* bodies,b3InertiaData* inertias);
-	void	initSolverBody(int bodyIndex, b3GpuSolverBody* solverBody, b3RigidBodyData* rb);
+	//	int	getOrInitSolverBody(int bodyIndex, b3RigidBodyData* bodies,b3InertiaData* inertias);
+	void initSolverBody(int bodyIndex, b3GpuSolverBody* solverBody, b3RigidBodyData* rb);
 
 public:
-	b3GpuPgsConstraintSolver (cl_context ctx, cl_device_id device, cl_command_queue queue,bool usePgs);
-	virtual~b3GpuPgsConstraintSolver ();
+	b3GpuPgsConstraintSolver(cl_context ctx, cl_device_id device, cl_command_queue queue, bool usePgs);
+	virtual ~b3GpuPgsConstraintSolver();
 
-	virtual b3Scalar solveGroupCacheFriendlyIterations(b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints1,int numConstraints,const b3ContactSolverInfo& infoGlobal);
-	virtual b3Scalar solveGroupCacheFriendlySetup(b3OpenCLArray<b3RigidBodyData>* gpuBodies, b3OpenCLArray<b3InertiaData>* gpuInertias, int numBodies,b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints,int numConstraints,const b3ContactSolverInfo& infoGlobal);
-	b3Scalar solveGroupCacheFriendlyFinish(b3OpenCLArray<b3RigidBodyData>* gpuBodies,b3OpenCLArray<b3InertiaData>* gpuInertias,int numBodies,b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints,int numConstraints,const b3ContactSolverInfo& infoGlobal);
+	virtual b3Scalar solveGroupCacheFriendlyIterations(b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints1, int numConstraints, const b3ContactSolverInfo& infoGlobal);
+	virtual b3Scalar solveGroupCacheFriendlySetup(b3OpenCLArray<b3RigidBodyData>* gpuBodies, b3OpenCLArray<b3InertiaData>* gpuInertias, int numBodies, b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints, int numConstraints, const b3ContactSolverInfo& infoGlobal);
+	b3Scalar solveGroupCacheFriendlyFinish(b3OpenCLArray<b3RigidBodyData>* gpuBodies, b3OpenCLArray<b3InertiaData>* gpuInertias, int numBodies, b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints, int numConstraints, const b3ContactSolverInfo& infoGlobal);
 
+	b3Scalar solveGroup(b3OpenCLArray<b3RigidBodyData>* gpuBodies, b3OpenCLArray<b3InertiaData>* gpuInertias, int numBodies, b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints, int numConstraints, const b3ContactSolverInfo& infoGlobal);
+	void solveJoints(int numBodies, b3OpenCLArray<b3RigidBodyData>* gpuBodies, b3OpenCLArray<b3InertiaData>* gpuInertias,
+					 int numConstraints, b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints);
 
-	b3Scalar solveGroup(b3OpenCLArray<b3RigidBodyData>* gpuBodies,b3OpenCLArray<b3InertiaData>* gpuInertias, int numBodies,b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints,int numConstraints,const b3ContactSolverInfo& infoGlobal);
-	void	solveJoints(int numBodies, b3OpenCLArray<b3RigidBodyData>* gpuBodies, b3OpenCLArray<b3InertiaData>* gpuInertias, 
-				int numConstraints, b3OpenCLArray<b3GpuGenericConstraint>* gpuConstraints);
-
-	int sortConstraintByBatch3( struct b3BatchConstraint* cs, int numConstraints, int simdWidth , int staticIdx, int numBodies);
-	void	recomputeBatches();
+	int sortConstraintByBatch3(struct b3BatchConstraint* cs, int numConstraints, int simdWidth, int staticIdx, int numBodies);
+	void recomputeBatches();
 };
 
-#endif //B3_GPU_PGS_CONSTRAINT_SOLVER_H
+#endif  //B3_GPU_PGS_CONSTRAINT_SOLVER_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuPgsContactSolver.cpp b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuPgsContactSolver.cpp
index f0b0abd5e00..e3d235a4fdc 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuPgsContactSolver.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuPgsContactSolver.cpp
@@ -2,7 +2,7 @@
 bool gUseLargeBatches = false;
 bool gCpuBatchContacts = false;
 bool gCpuSolveConstraint = false;
-bool gCpuRadixSort=false;
+bool gCpuRadixSort = false;
 bool gCpuSetSortData = false;
 bool gCpuSortContactsDeterminism = false;
 bool gUseCpuCopyConstraints = false;
@@ -11,7 +11,6 @@ bool gReorderContactsOnCpu = false;
 
 bool optionalSortContactsDeterminism = true;
 
-
 #include "b3GpuPgsContactSolver.h"
 #include "Bullet3OpenCL/ParallelPrimitives/b3RadixSort32CL.h"
 
@@ -23,7 +22,6 @@ bool optionalSortContactsDeterminism = true;
 #include "Bullet3Collision/NarrowPhaseCollision/b3Config.h"
 #include "b3Solver.h"
 
-
 #define B3_SOLVER_SETUP_KERNEL_PATH "src/Bullet3OpenCL/RigidBody/kernels/solverSetup.cl"
 #define B3_SOLVER_SETUP2_KERNEL_PATH "src/Bullet3OpenCL/RigidBody/kernels/solverSetup2.cl"
 #define B3_SOLVER_CONTACT_KERNEL_PATH "src/Bullet3OpenCL/RigidBody/kernels/solveContact.cl"
@@ -38,11 +36,7 @@ bool optionalSortContactsDeterminism = true;
 #include "kernels/batchingKernels.h"
 #include "kernels/batchingKernelsNew.h"
 
-
-
-
-
-struct	b3GpuBatchingPgsSolverInternalData
+struct b3GpuBatchingPgsSolverInternalData
 {
 	cl_context m_context;
 	cl_device_id m_device;
@@ -53,9 +47,9 @@ struct	b3GpuBatchingPgsSolverInternalData
 	b3OpenCLArray<b3GpuConstraint4>* m_contactCGPU;
 	b3OpenCLArray<unsigned int>* m_numConstraints;
 	b3OpenCLArray<unsigned int>* m_offsets;
-		
-	b3Solver*		m_solverGPU;		
-	
+
+	b3Solver* m_solverGPU;
+
 	cl_kernel m_batchingKernel;
 	cl_kernel m_batchingKernelNew;
 	cl_kernel m_solveContactKernel;
@@ -67,17 +61,14 @@ struct	b3GpuBatchingPgsSolverInternalData
 	cl_kernel m_reorderContactKernel;
 	cl_kernel m_copyConstraintKernel;
 
-	cl_kernel	m_setDeterminismSortDataBodyAKernel;
-	cl_kernel	m_setDeterminismSortDataBodyBKernel;
-	cl_kernel	m_setDeterminismSortDataChildShapeAKernel;
-	cl_kernel	m_setDeterminismSortDataChildShapeBKernel;
+	cl_kernel m_setDeterminismSortDataBodyAKernel;
+	cl_kernel m_setDeterminismSortDataBodyBKernel;
+	cl_kernel m_setDeterminismSortDataChildShapeAKernel;
+	cl_kernel m_setDeterminismSortDataChildShapeBKernel;
 
-
-
-
-	class b3RadixSort32CL*	m_sort32;
-	class b3BoundSearchCL*	m_search;
-	class b3PrefixScanCL*	m_scan;
+	class b3RadixSort32CL* m_sort32;
+	class b3BoundSearchCL* m_search;
+	class b3PrefixScanCL* m_scan;
 
 	b3OpenCLArray<b3SortData>* m_sortDataBuffer;
 	b3OpenCLArray<b3Contact4>* m_contactBuffer;
@@ -85,63 +76,56 @@ struct	b3GpuBatchingPgsSolverInternalData
 	b3OpenCLArray<b3RigidBodyData>* m_bodyBufferGPU;
 	b3OpenCLArray<b3InertiaData>* m_inertiaBufferGPU;
 	b3OpenCLArray<b3Contact4>* m_pBufContactOutGPU;
-	
-	b3OpenCLArray<b3Contact4>* m_pBufContactOutGPUCopy;
-	b3OpenCLArray<b3SortData>*	m_contactKeyValues;
 
+	b3OpenCLArray<b3Contact4>* m_pBufContactOutGPUCopy;
+	b3OpenCLArray<b3SortData>* m_contactKeyValues;
 
 	b3AlignedObjectArray<unsigned int> m_idxBuffer;
 	b3AlignedObjectArray<b3SortData> m_sortData;
 	b3AlignedObjectArray<b3Contact4> m_old;
 
-	b3AlignedObjectArray<int>	m_batchSizes;
-	b3OpenCLArray<int>*	m_batchSizesGpu;
-
+	b3AlignedObjectArray<int> m_batchSizes;
+	b3OpenCLArray<int>* m_batchSizesGpu;
 };
 
-
-
-b3GpuPgsContactSolver::b3GpuPgsContactSolver(cl_context ctx,cl_device_id device, cl_command_queue  q,int pairCapacity)
+b3GpuPgsContactSolver::b3GpuPgsContactSolver(cl_context ctx, cl_device_id device, cl_command_queue q, int pairCapacity)
 {
-	m_debugOutput=0;
+	m_debugOutput = 0;
 	m_data = new b3GpuBatchingPgsSolverInternalData;
 	m_data->m_context = ctx;
 	m_data->m_device = device;
 	m_data->m_queue = q;
 	m_data->m_pairCapacity = pairCapacity;
 	m_data->m_nIterations = 4;
-	m_data->m_batchSizesGpu = new b3OpenCLArray<int>(ctx,q);
-	m_data->m_bodyBufferGPU = new b3OpenCLArray<b3RigidBodyData>(ctx,q);
-	m_data->m_inertiaBufferGPU = new b3OpenCLArray<b3InertiaData>(ctx,q);
-	m_data->m_pBufContactOutGPU = new b3OpenCLArray<b3Contact4>(ctx,q);
+	m_data->m_batchSizesGpu = new b3OpenCLArray<int>(ctx, q);
+	m_data->m_bodyBufferGPU = new b3OpenCLArray<b3RigidBodyData>(ctx, q);
+	m_data->m_inertiaBufferGPU = new b3OpenCLArray<b3InertiaData>(ctx, q);
+	m_data->m_pBufContactOutGPU = new b3OpenCLArray<b3Contact4>(ctx, q);
 
-	m_data->m_pBufContactOutGPUCopy = new b3OpenCLArray<b3Contact4>(ctx,q);
-	m_data->m_contactKeyValues = new b3OpenCLArray<b3SortData>(ctx,q);
+	m_data->m_pBufContactOutGPUCopy = new b3OpenCLArray<b3Contact4>(ctx, q);
+	m_data->m_contactKeyValues = new b3OpenCLArray<b3SortData>(ctx, q);
 
+	m_data->m_solverGPU = new b3Solver(ctx, device, q, 512 * 1024);
 
-	m_data->m_solverGPU = new b3Solver(ctx,device,q,512*1024);
+	m_data->m_sort32 = new b3RadixSort32CL(ctx, device, m_data->m_queue);
+	m_data->m_scan = new b3PrefixScanCL(ctx, device, m_data->m_queue, B3_SOLVER_N_CELLS);
+	m_data->m_search = new b3BoundSearchCL(ctx, device, m_data->m_queue, B3_SOLVER_N_CELLS);
 
-	m_data->m_sort32 = new b3RadixSort32CL(ctx,device,m_data->m_queue);
-	m_data->m_scan = new b3PrefixScanCL(ctx,device,m_data->m_queue,B3_SOLVER_N_CELLS);
-	m_data->m_search = new b3BoundSearchCL(ctx,device,m_data->m_queue,B3_SOLVER_N_CELLS);
+	const int sortSize = B3NEXTMULTIPLEOF(pairCapacity, 512);
 
-	const int sortSize = B3NEXTMULTIPLEOF( pairCapacity, 512 );
+	m_data->m_sortDataBuffer = new b3OpenCLArray<b3SortData>(ctx, m_data->m_queue, sortSize);
+	m_data->m_contactBuffer = new b3OpenCLArray<b3Contact4>(ctx, m_data->m_queue);
 
-	m_data->m_sortDataBuffer = new b3OpenCLArray<b3SortData>(ctx,m_data->m_queue,sortSize);
-	m_data->m_contactBuffer = new b3OpenCLArray<b3Contact4>(ctx,m_data->m_queue);
-
-	m_data->m_numConstraints = new b3OpenCLArray<unsigned int>(ctx,m_data->m_queue,B3_SOLVER_N_CELLS);
+	m_data->m_numConstraints = new b3OpenCLArray<unsigned int>(ctx, m_data->m_queue, B3_SOLVER_N_CELLS);
 	m_data->m_numConstraints->resize(B3_SOLVER_N_CELLS);
 
-	m_data->m_contactCGPU = new b3OpenCLArray<b3GpuConstraint4>(ctx,q,pairCapacity);
+	m_data->m_contactCGPU = new b3OpenCLArray<b3GpuConstraint4>(ctx, q, pairCapacity);
 
-	m_data->m_offsets = new b3OpenCLArray<unsigned int>( ctx,m_data->m_queue,B3_SOLVER_N_CELLS);
+	m_data->m_offsets = new b3OpenCLArray<unsigned int>(ctx, m_data->m_queue, B3_SOLVER_N_CELLS);
 	m_data->m_offsets->resize(B3_SOLVER_N_CELLS);
 	const char* additionalMacros = "";
 	//const char* srcFileNameForCaching="";
 
-
-
 	cl_int pErrNum;
 	const char* batchKernelSource = batchingKernelsCL;
 	const char* batchKernelNewSource = batchingKernelsNewCL;
@@ -149,88 +133,73 @@ b3GpuPgsContactSolver::b3GpuPgsContactSolver(cl_context ctx,cl_device_id device,
 	const char* solverSetup2Source = solverSetup2CL;
 	const char* solveContactSource = solveContactCL;
 	const char* solveFrictionSource = solveFrictionCL;
-	
-		
+
 	{
-		
-		cl_program solveContactProg= b3OpenCLUtils::compileCLProgramFromString( ctx, device, solveContactSource, &pErrNum,additionalMacros, B3_SOLVER_CONTACT_KERNEL_PATH);
+		cl_program solveContactProg = b3OpenCLUtils::compileCLProgramFromString(ctx, device, solveContactSource, &pErrNum, additionalMacros, B3_SOLVER_CONTACT_KERNEL_PATH);
 		b3Assert(solveContactProg);
-		
-		cl_program solveFrictionProg= b3OpenCLUtils::compileCLProgramFromString( ctx, device, solveFrictionSource, &pErrNum,additionalMacros, B3_SOLVER_FRICTION_KERNEL_PATH);
+
+		cl_program solveFrictionProg = b3OpenCLUtils::compileCLProgramFromString(ctx, device, solveFrictionSource, &pErrNum, additionalMacros, B3_SOLVER_FRICTION_KERNEL_PATH);
 		b3Assert(solveFrictionProg);
 
-		cl_program solverSetup2Prog= b3OpenCLUtils::compileCLProgramFromString( ctx, device, solverSetup2Source, &pErrNum,additionalMacros, B3_SOLVER_SETUP2_KERNEL_PATH);
-		
-		
+		cl_program solverSetup2Prog = b3OpenCLUtils::compileCLProgramFromString(ctx, device, solverSetup2Source, &pErrNum, additionalMacros, B3_SOLVER_SETUP2_KERNEL_PATH);
+
 		b3Assert(solverSetup2Prog);
 
-		
-		cl_program solverSetupProg= b3OpenCLUtils::compileCLProgramFromString( ctx, device, solverSetupSource, &pErrNum,additionalMacros, B3_SOLVER_SETUP_KERNEL_PATH);
+		cl_program solverSetupProg = b3OpenCLUtils::compileCLProgramFromString(ctx, device, solverSetupSource, &pErrNum, additionalMacros, B3_SOLVER_SETUP_KERNEL_PATH);
 		b3Assert(solverSetupProg);
-		
-		
-		m_data->m_solveFrictionKernel= b3OpenCLUtils::compileCLKernelFromString( ctx, device, solveFrictionSource, "BatchSolveKernelFriction", &pErrNum, solveFrictionProg,additionalMacros );
+
+		m_data->m_solveFrictionKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solveFrictionSource, "BatchSolveKernelFriction", &pErrNum, solveFrictionProg, additionalMacros);
 		b3Assert(m_data->m_solveFrictionKernel);
 
-		m_data->m_solveContactKernel= b3OpenCLUtils::compileCLKernelFromString( ctx, device, solveContactSource, "BatchSolveKernelContact", &pErrNum, solveContactProg,additionalMacros );
+		m_data->m_solveContactKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solveContactSource, "BatchSolveKernelContact", &pErrNum, solveContactProg, additionalMacros);
 		b3Assert(m_data->m_solveContactKernel);
 
-		m_data->m_solveSingleContactKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, solveContactSource, "solveSingleContactKernel", &pErrNum, solveContactProg,additionalMacros );
+		m_data->m_solveSingleContactKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solveContactSource, "solveSingleContactKernel", &pErrNum, solveContactProg, additionalMacros);
 		b3Assert(m_data->m_solveSingleContactKernel);
 
-		m_data->m_solveSingleFrictionKernel =b3OpenCLUtils::compileCLKernelFromString( ctx, device, solveFrictionSource, "solveSingleFrictionKernel", &pErrNum, solveFrictionProg,additionalMacros );
+		m_data->m_solveSingleFrictionKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solveFrictionSource, "solveSingleFrictionKernel", &pErrNum, solveFrictionProg, additionalMacros);
 		b3Assert(m_data->m_solveSingleFrictionKernel);
-		
-		m_data->m_contactToConstraintKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, solverSetupSource, "ContactToConstraintKernel", &pErrNum, solverSetupProg,additionalMacros );
+
+		m_data->m_contactToConstraintKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solverSetupSource, "ContactToConstraintKernel", &pErrNum, solverSetupProg, additionalMacros);
 		b3Assert(m_data->m_contactToConstraintKernel);
-			
-		m_data->m_setSortDataKernel =  b3OpenCLUtils::compileCLKernelFromString( ctx, device, solverSetup2Source, "SetSortDataKernel", &pErrNum, solverSetup2Prog,additionalMacros );
+
+		m_data->m_setSortDataKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solverSetup2Source, "SetSortDataKernel", &pErrNum, solverSetup2Prog, additionalMacros);
 		b3Assert(m_data->m_setSortDataKernel);
 
-		m_data->m_setDeterminismSortDataBodyAKernel =  b3OpenCLUtils::compileCLKernelFromString( ctx, device, solverSetup2Source, "SetDeterminismSortDataBodyA", &pErrNum, solverSetup2Prog,additionalMacros );
+		m_data->m_setDeterminismSortDataBodyAKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solverSetup2Source, "SetDeterminismSortDataBodyA", &pErrNum, solverSetup2Prog, additionalMacros);
 		b3Assert(m_data->m_setDeterminismSortDataBodyAKernel);
 
-		m_data->m_setDeterminismSortDataBodyBKernel =  b3OpenCLUtils::compileCLKernelFromString( ctx, device, solverSetup2Source, "SetDeterminismSortDataBodyB", &pErrNum, solverSetup2Prog,additionalMacros );
+		m_data->m_setDeterminismSortDataBodyBKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solverSetup2Source, "SetDeterminismSortDataBodyB", &pErrNum, solverSetup2Prog, additionalMacros);
 		b3Assert(m_data->m_setDeterminismSortDataBodyBKernel);
 
-		m_data->m_setDeterminismSortDataChildShapeAKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, solverSetup2Source, "SetDeterminismSortDataChildShapeA", &pErrNum, solverSetup2Prog,additionalMacros );
+		m_data->m_setDeterminismSortDataChildShapeAKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solverSetup2Source, "SetDeterminismSortDataChildShapeA", &pErrNum, solverSetup2Prog, additionalMacros);
 		b3Assert(m_data->m_setDeterminismSortDataChildShapeAKernel);
 
-		m_data->m_setDeterminismSortDataChildShapeBKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, solverSetup2Source, "SetDeterminismSortDataChildShapeB", &pErrNum, solverSetup2Prog,additionalMacros );
+		m_data->m_setDeterminismSortDataChildShapeBKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solverSetup2Source, "SetDeterminismSortDataChildShapeB", &pErrNum, solverSetup2Prog, additionalMacros);
 		b3Assert(m_data->m_setDeterminismSortDataChildShapeBKernel);
 
-		
-		m_data->m_reorderContactKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, solverSetup2Source, "ReorderContactKernel", &pErrNum, solverSetup2Prog,additionalMacros );
+		m_data->m_reorderContactKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solverSetup2Source, "ReorderContactKernel", &pErrNum, solverSetup2Prog, additionalMacros);
 		b3Assert(m_data->m_reorderContactKernel);
-		
 
-		m_data->m_copyConstraintKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, solverSetup2Source, "CopyConstraintKernel", &pErrNum, solverSetup2Prog,additionalMacros );
+		m_data->m_copyConstraintKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solverSetup2Source, "CopyConstraintKernel", &pErrNum, solverSetup2Prog, additionalMacros);
 		b3Assert(m_data->m_copyConstraintKernel);
-		
 	}
 
 	{
-		cl_program batchingProg = b3OpenCLUtils::compileCLProgramFromString( ctx, device, batchKernelSource, &pErrNum,additionalMacros, B3_BATCHING_PATH);
+		cl_program batchingProg = b3OpenCLUtils::compileCLProgramFromString(ctx, device, batchKernelSource, &pErrNum, additionalMacros, B3_BATCHING_PATH);
 		b3Assert(batchingProg);
-		
-		m_data->m_batchingKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, batchKernelSource, "CreateBatches", &pErrNum, batchingProg,additionalMacros );
+
+		m_data->m_batchingKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, batchKernelSource, "CreateBatches", &pErrNum, batchingProg, additionalMacros);
 		b3Assert(m_data->m_batchingKernel);
 	}
-			
+
 	{
-		cl_program batchingNewProg = b3OpenCLUtils::compileCLProgramFromString( ctx, device, batchKernelNewSource, &pErrNum,additionalMacros, B3_BATCHING_NEW_PATH);
+		cl_program batchingNewProg = b3OpenCLUtils::compileCLProgramFromString(ctx, device, batchKernelNewSource, &pErrNum, additionalMacros, B3_BATCHING_NEW_PATH);
 		b3Assert(batchingNewProg);
-		
-		m_data->m_batchingKernelNew = b3OpenCLUtils::compileCLKernelFromString( ctx, device, batchKernelNewSource, "CreateBatchesNew", &pErrNum, batchingNewProg,additionalMacros );
+
+		m_data->m_batchingKernelNew = b3OpenCLUtils::compileCLKernelFromString(ctx, device, batchKernelNewSource, "CreateBatchesNew", &pErrNum, batchingNewProg, additionalMacros);
 		b3Assert(m_data->m_batchingKernelNew);
 	}
-		
-
-
-
-
-
-
 }
 
 b3GpuPgsContactSolver::~b3GpuPgsContactSolver()
@@ -242,8 +211,6 @@ b3GpuPgsContactSolver::~b3GpuPgsContactSolver()
 	delete m_data->m_pBufContactOutGPUCopy;
 	delete m_data->m_contactKeyValues;
 
-
-
 	delete m_data->m_contactCGPU;
 	delete m_data->m_numConstraints;
 	delete m_data->m_offsets;
@@ -259,29 +226,25 @@ b3GpuPgsContactSolver::~b3GpuPgsContactSolver()
 	clReleaseKernel(m_data->m_batchingKernelNew);
 	clReleaseKernel(m_data->m_solveSingleContactKernel);
 	clReleaseKernel(m_data->m_solveSingleFrictionKernel);
-	clReleaseKernel( m_data->m_solveContactKernel);
-	clReleaseKernel( m_data->m_solveFrictionKernel);
+	clReleaseKernel(m_data->m_solveContactKernel);
+	clReleaseKernel(m_data->m_solveFrictionKernel);
 
-	clReleaseKernel( m_data->m_contactToConstraintKernel);
-	clReleaseKernel( m_data->m_setSortDataKernel);
-	clReleaseKernel( m_data->m_reorderContactKernel);
-	clReleaseKernel( m_data->m_copyConstraintKernel);
+	clReleaseKernel(m_data->m_contactToConstraintKernel);
+	clReleaseKernel(m_data->m_setSortDataKernel);
+	clReleaseKernel(m_data->m_reorderContactKernel);
+	clReleaseKernel(m_data->m_copyConstraintKernel);
 
 	clReleaseKernel(m_data->m_setDeterminismSortDataBodyAKernel);
 	clReleaseKernel(m_data->m_setDeterminismSortDataBodyBKernel);
 	clReleaseKernel(m_data->m_setDeterminismSortDataChildShapeAKernel);
 	clReleaseKernel(m_data->m_setDeterminismSortDataChildShapeBKernel);
 
-
-
 	delete m_data;
 }
 
-
-
 struct b3ConstraintCfg
 {
-	b3ConstraintCfg( float dt = 0.f ): m_positionDrift( 0.005f ), m_positionConstraintCoeff( 0.2f ), m_dt(dt), m_staticIdx(0) {}
+	b3ConstraintCfg(float dt = 0.f) : m_positionDrift(0.005f), m_positionConstraintCoeff(0.2f), m_dt(dt), m_staticIdx(0) {}
 
 	float m_positionDrift;
 	float m_positionConstraintCoeff;
@@ -291,354 +254,306 @@ struct b3ConstraintCfg
 	int m_staticIdx;
 };
 
-
-
-void b3GpuPgsContactSolver::solveContactConstraintBatchSizes(  const b3OpenCLArray<b3RigidBodyData>* bodyBuf, const b3OpenCLArray<b3InertiaData>* shapeBuf, 
-			b3OpenCLArray<b3GpuConstraint4>* constraint, void* additionalData, int n ,int maxNumBatches,int numIterations, const b3AlignedObjectArray<int>* batchSizes)//const b3OpenCLArray<int>* gpuBatchSizes)
+void b3GpuPgsContactSolver::solveContactConstraintBatchSizes(const b3OpenCLArray<b3RigidBodyData>* bodyBuf, const b3OpenCLArray<b3InertiaData>* shapeBuf,
+															 b3OpenCLArray<b3GpuConstraint4>* constraint, void* additionalData, int n, int maxNumBatches, int numIterations, const b3AlignedObjectArray<int>* batchSizes)  //const b3OpenCLArray<int>* gpuBatchSizes)
 {
 	B3_PROFILE("solveContactConstraintBatchSizes");
-	int numBatches = batchSizes->size()/B3_MAX_NUM_BATCHES;
-	for(int iter=0; iter<numIterations; iter++)
+	int numBatches = batchSizes->size() / B3_MAX_NUM_BATCHES;
+	for (int iter = 0; iter < numIterations; iter++)
 	{
-		
-		for (int cellId=0;cellId<numBatches;cellId++)
+		for (int cellId = 0; cellId < numBatches; cellId++)
 		{
 			int offset = 0;
-			for (int ii=0;ii<B3_MAX_NUM_BATCHES;ii++)
+			for (int ii = 0; ii < B3_MAX_NUM_BATCHES; ii++)
 			{
-				int numInBatch = batchSizes->at(cellId*B3_MAX_NUM_BATCHES+ii);
+				int numInBatch = batchSizes->at(cellId * B3_MAX_NUM_BATCHES + ii);
 				if (!numInBatch)
 					break;
 
 				{
-					b3LauncherCL launcher( m_data->m_queue, m_data->m_solveSingleContactKernel,"m_solveSingleContactKernel" );
-					launcher.setBuffer(bodyBuf->getBufferCL() );
-					launcher.setBuffer(shapeBuf->getBufferCL() );
-					launcher.setBuffer(	constraint->getBufferCL() );
+					b3LauncherCL launcher(m_data->m_queue, m_data->m_solveSingleContactKernel, "m_solveSingleContactKernel");
+					launcher.setBuffer(bodyBuf->getBufferCL());
+					launcher.setBuffer(shapeBuf->getBufferCL());
+					launcher.setBuffer(constraint->getBufferCL());
 					launcher.setConst(cellId);
 					launcher.setConst(offset);
 					launcher.setConst(numInBatch);
 					launcher.launch1D(numInBatch);
-					offset+=numInBatch;
+					offset += numInBatch;
 				}
 			}
 		}
 	}
 
-
-	for(int iter=0; iter<numIterations; iter++)
+	for (int iter = 0; iter < numIterations; iter++)
 	{
-		for (int cellId=0;cellId<numBatches;cellId++)
+		for (int cellId = 0; cellId < numBatches; cellId++)
 		{
 			int offset = 0;
-			for (int ii=0;ii<B3_MAX_NUM_BATCHES;ii++)
+			for (int ii = 0; ii < B3_MAX_NUM_BATCHES; ii++)
 			{
-				int numInBatch = batchSizes->at(cellId*B3_MAX_NUM_BATCHES+ii);
+				int numInBatch = batchSizes->at(cellId * B3_MAX_NUM_BATCHES + ii);
 				if (!numInBatch)
 					break;
 
 				{
-					b3LauncherCL launcher( m_data->m_queue, m_data->m_solveSingleFrictionKernel,"m_solveSingleFrictionKernel" );
-					launcher.setBuffer(bodyBuf->getBufferCL() );
-					launcher.setBuffer(shapeBuf->getBufferCL() );
-					launcher.setBuffer(	constraint->getBufferCL() );
+					b3LauncherCL launcher(m_data->m_queue, m_data->m_solveSingleFrictionKernel, "m_solveSingleFrictionKernel");
+					launcher.setBuffer(bodyBuf->getBufferCL());
+					launcher.setBuffer(shapeBuf->getBufferCL());
+					launcher.setBuffer(constraint->getBufferCL());
 					launcher.setConst(cellId);
 					launcher.setConst(offset);
 					launcher.setConst(numInBatch);
 					launcher.launch1D(numInBatch);
-					offset+=numInBatch;
+					offset += numInBatch;
 				}
 			}
 		}
 	}
 }
 
-void b3GpuPgsContactSolver::solveContactConstraint(  const b3OpenCLArray<b3RigidBodyData>* bodyBuf, const b3OpenCLArray<b3InertiaData>* shapeBuf, 
-			b3OpenCLArray<b3GpuConstraint4>* constraint, void* additionalData, int n ,int maxNumBatches,int numIterations, const b3AlignedObjectArray<int>* batchSizes)//,const b3OpenCLArray<int>* gpuBatchSizes)
+void b3GpuPgsContactSolver::solveContactConstraint(const b3OpenCLArray<b3RigidBodyData>* bodyBuf, const b3OpenCLArray<b3InertiaData>* shapeBuf,
+												   b3OpenCLArray<b3GpuConstraint4>* constraint, void* additionalData, int n, int maxNumBatches, int numIterations, const b3AlignedObjectArray<int>* batchSizes)  //,const b3OpenCLArray<int>* gpuBatchSizes)
 {
-	
 	//sort the contacts
 
-	
-	b3Int4 cdata = b3MakeInt4( n, 0, 0, 0 );
+	b3Int4 cdata = b3MakeInt4(n, 0, 0, 0);
 	{
-		
 		const int nn = B3_SOLVER_N_CELLS;
 
 		cdata.x = 0;
-		cdata.y = maxNumBatches;//250;
+		cdata.y = maxNumBatches;  //250;
 
-
-		int numWorkItems = 64*nn/B3_SOLVER_N_BATCHES;
+		int numWorkItems = 64 * nn / B3_SOLVER_N_BATCHES;
 #ifdef DEBUG_ME
-		SolverDebugInfo* debugInfo = new  SolverDebugInfo[numWorkItems];
-		adl::b3OpenCLArray<SolverDebugInfo> gpuDebugInfo(data->m_device,numWorkItems);
+		SolverDebugInfo* debugInfo = new SolverDebugInfo[numWorkItems];
+		adl::b3OpenCLArray<SolverDebugInfo> gpuDebugInfo(data->m_device, numWorkItems);
 #endif
 
-
-
 		{
-
 			B3_PROFILE("m_batchSolveKernel iterations");
-			for(int iter=0; iter<numIterations; iter++)
+			for (int iter = 0; iter < numIterations; iter++)
 			{
-				for(int ib=0; ib<B3_SOLVER_N_BATCHES; ib++)
+				for (int ib = 0; ib < B3_SOLVER_N_BATCHES; ib++)
 				{
 #ifdef DEBUG_ME
-					memset(debugInfo,0,sizeof(SolverDebugInfo)*numWorkItems);
-					gpuDebugInfo.write(debugInfo,numWorkItems);
+					memset(debugInfo, 0, sizeof(SolverDebugInfo) * numWorkItems);
+					gpuDebugInfo.write(debugInfo, numWorkItems);
 #endif
 
-
 					cdata.z = ib;
-					
 
-				b3LauncherCL launcher( m_data->m_queue, m_data->m_solveContactKernel,"m_solveContactKernel" );
+					b3LauncherCL launcher(m_data->m_queue, m_data->m_solveContactKernel, "m_solveContactKernel");
 #if 1
-                    
-					b3BufferInfoCL bInfo[] = { 
 
-						b3BufferInfoCL( bodyBuf->getBufferCL() ), 
-						b3BufferInfoCL( shapeBuf->getBufferCL() ), 
-						b3BufferInfoCL( constraint->getBufferCL() ),
-						b3BufferInfoCL( m_data->m_solverGPU->m_numConstraints->getBufferCL() ), 
-						b3BufferInfoCL( m_data->m_solverGPU->m_offsets->getBufferCL() ) 
+					b3BufferInfoCL bInfo[] = {
+
+						b3BufferInfoCL(bodyBuf->getBufferCL()),
+						b3BufferInfoCL(shapeBuf->getBufferCL()),
+						b3BufferInfoCL(constraint->getBufferCL()),
+						b3BufferInfoCL(m_data->m_solverGPU->m_numConstraints->getBufferCL()),
+						b3BufferInfoCL(m_data->m_solverGPU->m_offsets->getBufferCL())
 #ifdef DEBUG_ME
-						,	b3BufferInfoCL(&gpuDebugInfo)
+							,
+						b3BufferInfoCL(&gpuDebugInfo)
 #endif
-						};
+					};
 
-					
-
-                    launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-					launcher.setBuffer( m_data->m_solverGPU->m_batchSizes.getBufferCL());
+					launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+					launcher.setBuffer(m_data->m_solverGPU->m_batchSizes.getBufferCL());
 					//launcher.setConst(  cdata.x );
-                    launcher.setConst(  cdata.y );
-                    launcher.setConst(  cdata.z );
+					launcher.setConst(cdata.y);
+					launcher.setConst(cdata.z);
 					b3Int4 nSplit;
 					nSplit.x = B3_SOLVER_N_SPLIT_X;
 					nSplit.y = B3_SOLVER_N_SPLIT_Y;
 					nSplit.z = B3_SOLVER_N_SPLIT_Z;
 
-                    launcher.setConst(  nSplit );
-                    launcher.launch1D( numWorkItems, 64 );
+					launcher.setConst(nSplit);
+					launcher.launch1D(numWorkItems, 64);
 
-                    
 #else
-                    const char* fileName = "m_batchSolveKernel.bin";
-                    FILE* f = fopen(fileName,"rb");
-                    if (f)
-                    {
-                        int sizeInBytes=0;
-                        if (fseek(f, 0, SEEK_END) || (sizeInBytes = ftell(f)) == EOF || fseek(f, 0, SEEK_SET))
-                        {
-                            printf("error, cannot get file size\n");
-                            exit(0);
-                        }
-                        
-                        unsigned char* buf = (unsigned char*) malloc(sizeInBytes);
-                        fread(buf,sizeInBytes,1,f);
-                        int serializedBytes = launcher.deserializeArgs(buf, sizeInBytes,m_context);
-                        int num = *(int*)&buf[serializedBytes];
-                        
-                        launcher.launch1D( num);
+					const char* fileName = "m_batchSolveKernel.bin";
+					FILE* f = fopen(fileName, "rb");
+					if (f)
+					{
+						int sizeInBytes = 0;
+						if (fseek(f, 0, SEEK_END) || (sizeInBytes = ftell(f)) == EOF || fseek(f, 0, SEEK_SET))
+						{
+							printf("error, cannot get file size\n");
+							exit(0);
+						}
 
-                        //this clFinish is for testing on errors
-                        clFinish(m_queue);
-                    }
+						unsigned char* buf = (unsigned char*)malloc(sizeInBytes);
+						fread(buf, sizeInBytes, 1, f);
+						int serializedBytes = launcher.deserializeArgs(buf, sizeInBytes, m_context);
+						int num = *(int*)&buf[serializedBytes];
+
+						launcher.launch1D(num);
+
+						//this clFinish is for testing on errors
+						clFinish(m_queue);
+					}
 
 #endif
-					
 
 #ifdef DEBUG_ME
 					clFinish(m_queue);
-					gpuDebugInfo.read(debugInfo,numWorkItems);
+					gpuDebugInfo.read(debugInfo, numWorkItems);
 					clFinish(m_queue);
-					for (int i=0;i<numWorkItems;i++)
+					for (int i = 0; i < numWorkItems; i++)
 					{
-						if (debugInfo[i].m_valInt2>0)
+						if (debugInfo[i].m_valInt2 > 0)
 						{
-							printf("debugInfo[i].m_valInt2 = %d\n",i,debugInfo[i].m_valInt2);
+							printf("debugInfo[i].m_valInt2 = %d\n", i, debugInfo[i].m_valInt2);
 						}
 
-						if (debugInfo[i].m_valInt3>0)
+						if (debugInfo[i].m_valInt3 > 0)
 						{
-							printf("debugInfo[i].m_valInt3 = %d\n",i,debugInfo[i].m_valInt3);
+							printf("debugInfo[i].m_valInt3 = %d\n", i, debugInfo[i].m_valInt3);
 						}
 					}
-#endif //DEBUG_ME
-
-
+#endif  //DEBUG_ME
 				}
 			}
-		
+
 			clFinish(m_data->m_queue);
-
-
 		}
 
 		cdata.x = 1;
-		bool applyFriction=true;
+		bool applyFriction = true;
 		if (applyFriction)
-    	{
+		{
 			B3_PROFILE("m_batchSolveKernel iterations2");
-			for(int iter=0; iter<numIterations; iter++)
+			for (int iter = 0; iter < numIterations; iter++)
 			{
-				for(int ib=0; ib<B3_SOLVER_N_BATCHES; ib++)
+				for (int ib = 0; ib < B3_SOLVER_N_BATCHES; ib++)
 				{
 					cdata.z = ib;
-					
 
-					b3BufferInfoCL bInfo[] = { 
-						b3BufferInfoCL( bodyBuf->getBufferCL() ), 
-						b3BufferInfoCL( shapeBuf->getBufferCL() ), 
-						b3BufferInfoCL( constraint->getBufferCL() ),
-						b3BufferInfoCL( m_data->m_solverGPU->m_numConstraints->getBufferCL() ), 
-						b3BufferInfoCL( m_data->m_solverGPU->m_offsets->getBufferCL() )
+					b3BufferInfoCL bInfo[] = {
+						b3BufferInfoCL(bodyBuf->getBufferCL()),
+						b3BufferInfoCL(shapeBuf->getBufferCL()),
+						b3BufferInfoCL(constraint->getBufferCL()),
+						b3BufferInfoCL(m_data->m_solverGPU->m_numConstraints->getBufferCL()),
+						b3BufferInfoCL(m_data->m_solverGPU->m_offsets->getBufferCL())
 #ifdef DEBUG_ME
-						,b3BufferInfoCL(&gpuDebugInfo)
-#endif //DEBUG_ME
+							,
+						b3BufferInfoCL(&gpuDebugInfo)
+#endif  //DEBUG_ME
 					};
-					b3LauncherCL launcher( m_data->m_queue, m_data->m_solveFrictionKernel,"m_solveFrictionKernel" );
-					launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-					launcher.setBuffer( m_data->m_solverGPU->m_batchSizes.getBufferCL());
+					b3LauncherCL launcher(m_data->m_queue, m_data->m_solveFrictionKernel, "m_solveFrictionKernel");
+					launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+					launcher.setBuffer(m_data->m_solverGPU->m_batchSizes.getBufferCL());
 					//launcher.setConst(  cdata.x );
-                    launcher.setConst(  cdata.y );
-                    launcher.setConst(  cdata.z );
+					launcher.setConst(cdata.y);
+					launcher.setConst(cdata.z);
 
-                    b3Int4 nSplit;
+					b3Int4 nSplit;
 					nSplit.x = B3_SOLVER_N_SPLIT_X;
 					nSplit.y = B3_SOLVER_N_SPLIT_Y;
 					nSplit.z = B3_SOLVER_N_SPLIT_Z;
 
-                    launcher.setConst(  nSplit );
-                    
-					launcher.launch1D( 64*nn/B3_SOLVER_N_BATCHES, 64 );
+					launcher.setConst(nSplit);
+
+					launcher.launch1D(64 * nn / B3_SOLVER_N_BATCHES, 64);
 				}
 			}
 			clFinish(m_data->m_queue);
-			
 		}
 #ifdef DEBUG_ME
 		delete[] debugInfo;
-#endif //DEBUG_ME
+#endif  //DEBUG_ME
 	}
-
-	
 }
 
-
-
-
-
-
-
-
-
-
-
-static bool sortfnc(const b3SortData& a,const b3SortData& b)
+static bool sortfnc(const b3SortData& a, const b3SortData& b)
 {
-	return (a.m_key<b.m_key);
+	return (a.m_key < b.m_key);
 }
 
 static bool b3ContactCmp(const b3Contact4& p, const b3Contact4& q)
 {
-	return ((p.m_bodyAPtrAndSignBit<q.m_bodyAPtrAndSignBit) ||
-		((p.m_bodyAPtrAndSignBit==q.m_bodyAPtrAndSignBit) && (p.m_bodyBPtrAndSignBit<q.m_bodyBPtrAndSignBit)) ||
-		((p.m_bodyAPtrAndSignBit==q.m_bodyAPtrAndSignBit) && (p.m_bodyBPtrAndSignBit==q.m_bodyBPtrAndSignBit)  &&     p.m_childIndexA<q.m_childIndexA ) ||
-		((p.m_bodyAPtrAndSignBit==q.m_bodyAPtrAndSignBit) && (p.m_bodyBPtrAndSignBit==q.m_bodyBPtrAndSignBit)  &&     p.m_childIndexA<q.m_childIndexA ) ||
-		((p.m_bodyAPtrAndSignBit==q.m_bodyAPtrAndSignBit) && (p.m_bodyBPtrAndSignBit==q.m_bodyBPtrAndSignBit)  &&     p.m_childIndexA==q.m_childIndexA  && p.m_childIndexB<q.m_childIndexB)
-		);
+	return ((p.m_bodyAPtrAndSignBit < q.m_bodyAPtrAndSignBit) ||
+			((p.m_bodyAPtrAndSignBit == q.m_bodyAPtrAndSignBit) && (p.m_bodyBPtrAndSignBit < q.m_bodyBPtrAndSignBit)) ||
+			((p.m_bodyAPtrAndSignBit == q.m_bodyAPtrAndSignBit) && (p.m_bodyBPtrAndSignBit == q.m_bodyBPtrAndSignBit) && p.m_childIndexA < q.m_childIndexA) ||
+			((p.m_bodyAPtrAndSignBit == q.m_bodyAPtrAndSignBit) && (p.m_bodyBPtrAndSignBit == q.m_bodyBPtrAndSignBit) && p.m_childIndexA < q.m_childIndexA) ||
+			((p.m_bodyAPtrAndSignBit == q.m_bodyAPtrAndSignBit) && (p.m_bodyBPtrAndSignBit == q.m_bodyBPtrAndSignBit) && p.m_childIndexA == q.m_childIndexA && p.m_childIndexB < q.m_childIndexB));
 }
 
-
-
-
-
-
-
-
-
-
-
 #define USE_SPATIAL_BATCHING 1
 #define USE_4x4_GRID 1
 
 #ifndef USE_SPATIAL_BATCHING
-static const int gridTable4x4[] = 
-{
-	0,1,17,16,
-	1,2,18,19,
-	17,18,32,3,
-	16,19,3,34
-};
-static const int gridTable8x8[] = 
-{
-	0,  2,  3, 16, 17, 18, 19,  1,
-	66, 64, 80, 67, 82, 81, 65, 83,
-	131,144,128,130,147,129,145,146,
-	208,195,194,192,193,211,210,209,
-	21, 22, 23,  5,  4,  6,  7, 20,
-	86, 85, 69, 87, 70, 68, 84, 71,
-	151,133,149,150,135,148,132,134,
-	197,27,214,213,212,199,198,196
-	
-};
+static const int gridTable4x4[] =
+	{
+		0, 1, 17, 16,
+		1, 2, 18, 19,
+		17, 18, 32, 3,
+		16, 19, 3, 34};
+static const int gridTable8x8[] =
+	{
+		0, 2, 3, 16, 17, 18, 19, 1,
+		66, 64, 80, 67, 82, 81, 65, 83,
+		131, 144, 128, 130, 147, 129, 145, 146,
+		208, 195, 194, 192, 193, 211, 210, 209,
+		21, 22, 23, 5, 4, 6, 7, 20,
+		86, 85, 69, 87, 70, 68, 84, 71,
+		151, 133, 149, 150, 135, 148, 132, 134,
+		197, 27, 214, 213, 212, 199, 198, 196
 
+};
 
 #endif
 
-
-void SetSortDataCPU(b3Contact4* gContact, b3RigidBodyData* gBodies, b3SortData* gSortDataOut, int nContacts,float scale,const b3Int4& nSplit,int staticIdx)
+void SetSortDataCPU(b3Contact4* gContact, b3RigidBodyData* gBodies, b3SortData* gSortDataOut, int nContacts, float scale, const b3Int4& nSplit, int staticIdx)
 {
-	for (int gIdx=0;gIdx<nContacts;gIdx++)
+	for (int gIdx = 0; gIdx < nContacts; gIdx++)
 	{
-		if( gIdx < nContacts )
+		if (gIdx < nContacts)
 		{
-			int aPtrAndSignBit  = gContact[gIdx].m_bodyAPtrAndSignBit;
-			int bPtrAndSignBit  = gContact[gIdx].m_bodyBPtrAndSignBit;
+			int aPtrAndSignBit = gContact[gIdx].m_bodyAPtrAndSignBit;
+			int bPtrAndSignBit = gContact[gIdx].m_bodyBPtrAndSignBit;
 
-			int aIdx = abs(aPtrAndSignBit );
+			int aIdx = abs(aPtrAndSignBit);
 			int bIdx = abs(bPtrAndSignBit);
 
-			bool aStatic = (aPtrAndSignBit<0) ||(aPtrAndSignBit==staticIdx);
+			bool aStatic = (aPtrAndSignBit < 0) || (aPtrAndSignBit == staticIdx);
 
-	#if USE_SPATIAL_BATCHING		
-			int idx = (aStatic)? bIdx: aIdx;
+#if USE_SPATIAL_BATCHING
+			int idx = (aStatic) ? bIdx : aIdx;
 			b3Vector3 p = gBodies[idx].m_pos;
-			int xIdx = (int)((p.x-((p.x<0.f)?1.f:0.f))*scale) & (nSplit.x-1);
-			int yIdx = (int)((p.y-((p.y<0.f)?1.f:0.f))*scale) & (nSplit.y-1);
-			int zIdx = (int)((p.z-((p.z<0.f)?1.f:0.f))*scale) & (nSplit.z-1);
-			
-			int newIndex = (xIdx+yIdx*nSplit.x+zIdx*nSplit.x*nSplit.y);
-		
-	#else//USE_SPATIAL_BATCHING
-			bool bStatic = (bPtrAndSignBit<0) ||(bPtrAndSignBit==staticIdx);
+			int xIdx = (int)((p.x - ((p.x < 0.f) ? 1.f : 0.f)) * scale) & (nSplit.x - 1);
+			int yIdx = (int)((p.y - ((p.y < 0.f) ? 1.f : 0.f)) * scale) & (nSplit.y - 1);
+			int zIdx = (int)((p.z - ((p.z < 0.f) ? 1.f : 0.f)) * scale) & (nSplit.z - 1);
 
-		#if USE_4x4_GRID
-			int aa = aIdx&3;
-			int bb = bIdx&3;
+			int newIndex = (xIdx + yIdx * nSplit.x + zIdx * nSplit.x * nSplit.y);
+
+#else  //USE_SPATIAL_BATCHING
+			bool bStatic = (bPtrAndSignBit < 0) || (bPtrAndSignBit == staticIdx);
+
+#if USE_4x4_GRID
+			int aa = aIdx & 3;
+			int bb = bIdx & 3;
 			if (aStatic)
 				aa = bb;
 			if (bStatic)
 				bb = aa;
 
-			int gridIndex = aa + bb*4;
+			int gridIndex = aa + bb * 4;
 			int newIndex = gridTable4x4[gridIndex];
-		#else//USE_4x4_GRID
-			int aa = aIdx&7;
-			int bb = bIdx&7;
+#else   //USE_4x4_GRID
+			int aa = aIdx & 7;
+			int bb = bIdx & 7;
 			if (aStatic)
 				aa = bb;
 			if (bStatic)
 				bb = aa;
 
-			int gridIndex = aa + bb*8;
+			int gridIndex = aa + bb * 8;
 			int newIndex = gridTable8x8[gridIndex];
-		#endif//USE_4x4_GRID
-	#endif//USE_SPATIAL_BATCHING
-
+#endif  //USE_4x4_GRID
+#endif  //USE_SPATIAL_BATCHING
 
 			gSortDataOut[gIdx].x = newIndex;
 			gSortDataOut[gIdx].y = gIdx;
@@ -650,17 +565,12 @@ void SetSortDataCPU(b3Contact4* gContact, b3RigidBodyData* gBodies, b3SortData*
 	}
 }
 
-
-
-
-
-
 void b3GpuPgsContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem inertiaBuf, int numContacts, cl_mem contactBuf, const b3Config& config, int static0Index)
 {
 	B3_PROFILE("solveContacts");
-	m_data->m_bodyBufferGPU->setFromOpenCLBuffer(bodyBuf,numBodies);
-	m_data->m_inertiaBufferGPU->setFromOpenCLBuffer(inertiaBuf,numBodies);
-	m_data->m_pBufContactOutGPU->setFromOpenCLBuffer(contactBuf,numContacts);
+	m_data->m_bodyBufferGPU->setFromOpenCLBuffer(bodyBuf, numBodies);
+	m_data->m_inertiaBufferGPU->setFromOpenCLBuffer(inertiaBuf, numBodies);
+	m_data->m_pBufContactOutGPU->setFromOpenCLBuffer(contactBuf, numContacts);
 
 	if (optionalSortContactsDeterminism)
 	{
@@ -671,61 +581,61 @@ void b3GpuPgsContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem
 			m_data->m_pBufContactOutGPUCopy->resize(numContacts);
 			m_data->m_contactKeyValues->resize(numContacts);
 
-			m_data->m_pBufContactOutGPU->copyToCL(m_data->m_pBufContactOutGPUCopy->getBufferCL(),numContacts,0,0);
+			m_data->m_pBufContactOutGPU->copyToCL(m_data->m_pBufContactOutGPUCopy->getBufferCL(), numContacts, 0, 0);
 
 			{
-				b3LauncherCL launcher(m_data->m_queue, m_data->m_setDeterminismSortDataChildShapeBKernel,"m_setDeterminismSortDataChildShapeBKernel");
+				b3LauncherCL launcher(m_data->m_queue, m_data->m_setDeterminismSortDataChildShapeBKernel, "m_setDeterminismSortDataChildShapeBKernel");
 				launcher.setBuffer(m_data->m_pBufContactOutGPUCopy->getBufferCL());
 				launcher.setBuffer(m_data->m_contactKeyValues->getBufferCL());
 				launcher.setConst(numContacts);
-				launcher.launch1D( numContacts, 64 );
+				launcher.launch1D(numContacts, 64);
 			}
 			m_data->m_solverGPU->m_sort32->execute(*m_data->m_contactKeyValues);
 			{
-				b3LauncherCL launcher(m_data->m_queue, m_data->m_setDeterminismSortDataChildShapeAKernel,"m_setDeterminismSortDataChildShapeAKernel");
+				b3LauncherCL launcher(m_data->m_queue, m_data->m_setDeterminismSortDataChildShapeAKernel, "m_setDeterminismSortDataChildShapeAKernel");
 				launcher.setBuffer(m_data->m_pBufContactOutGPUCopy->getBufferCL());
 				launcher.setBuffer(m_data->m_contactKeyValues->getBufferCL());
 				launcher.setConst(numContacts);
-				launcher.launch1D( numContacts, 64 );
+				launcher.launch1D(numContacts, 64);
 			}
 			m_data->m_solverGPU->m_sort32->execute(*m_data->m_contactKeyValues);
 			{
-				b3LauncherCL launcher(m_data->m_queue, m_data->m_setDeterminismSortDataBodyBKernel,"m_setDeterminismSortDataBodyBKernel");
+				b3LauncherCL launcher(m_data->m_queue, m_data->m_setDeterminismSortDataBodyBKernel, "m_setDeterminismSortDataBodyBKernel");
 				launcher.setBuffer(m_data->m_pBufContactOutGPUCopy->getBufferCL());
 				launcher.setBuffer(m_data->m_contactKeyValues->getBufferCL());
 				launcher.setConst(numContacts);
-				launcher.launch1D( numContacts, 64 );
+				launcher.launch1D(numContacts, 64);
 			}
-						
+
 			m_data->m_solverGPU->m_sort32->execute(*m_data->m_contactKeyValues);
-			
+
 			{
-				b3LauncherCL launcher(m_data->m_queue, m_data->m_setDeterminismSortDataBodyAKernel,"m_setDeterminismSortDataBodyAKernel");
+				b3LauncherCL launcher(m_data->m_queue, m_data->m_setDeterminismSortDataBodyAKernel, "m_setDeterminismSortDataBodyAKernel");
 				launcher.setBuffer(m_data->m_pBufContactOutGPUCopy->getBufferCL());
 				launcher.setBuffer(m_data->m_contactKeyValues->getBufferCL());
 				launcher.setConst(numContacts);
-				launcher.launch1D( numContacts, 64 );
+				launcher.launch1D(numContacts, 64);
 			}
 
 			m_data->m_solverGPU->m_sort32->execute(*m_data->m_contactKeyValues);
 
 			{
 				B3_PROFILE("gpu reorderContactKernel (determinism)");
-                                
+
 				b3Int4 cdata;
 				cdata.x = numContacts;
-                                
+
 				//b3BufferInfoCL bInfo[] = { b3BufferInfoCL( m_data->m_pBufContactOutGPU->getBufferCL() ), b3BufferInfoCL( m_data->m_solverGPU->m_contactBuffer2->getBufferCL())
 				//	, b3BufferInfoCL( m_data->m_solverGPU->m_sortDataBuffer->getBufferCL()) };
-				b3LauncherCL launcher(m_data->m_queue,m_data->m_solverGPU->m_reorderContactKernel,"m_reorderContactKernel");
+				b3LauncherCL launcher(m_data->m_queue, m_data->m_solverGPU->m_reorderContactKernel, "m_reorderContactKernel");
 				launcher.setBuffer(m_data->m_pBufContactOutGPUCopy->getBufferCL());
 				launcher.setBuffer(m_data->m_pBufContactOutGPU->getBufferCL());
 				launcher.setBuffer(m_data->m_contactKeyValues->getBufferCL());
-				launcher.setConst( cdata );
-				launcher.launch1D( numContacts, 64 );
-            }
-
-		} else
+				launcher.setConst(cdata);
+				launcher.launch1D(numContacts, 64);
+			}
+		}
+		else
 		{
 			B3_PROFILE("CPU Sort contact constraints (determinism)");
 			b3AlignedObjectArray<b3Contact4> cpuConstraints;
@@ -735,96 +645,80 @@ void b3GpuPgsContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem
 			{
 				cpuConstraints.quickSort(b3ContactCmp);
 
-				for (int i=0;i<cpuConstraints.size();i++)
+				for (int i = 0; i < cpuConstraints.size(); i++)
 				{
 					cpuConstraints[i].m_batchIdx = i;
 				}
 			}
 			m_data->m_pBufContactOutGPU->copyFromHost(cpuConstraints);
-			if (m_debugOutput==100)
+			if (m_debugOutput == 100)
 			{
-				for (int i=0;i<cpuConstraints.size();i++)
+				for (int i = 0; i < cpuConstraints.size(); i++)
 				{
-					printf("c[%d].m_bodyA = %d, m_bodyB = %d, batchId = %d\n",i,cpuConstraints[i].m_bodyAPtrAndSignBit,cpuConstraints[i].m_bodyBPtrAndSignBit, cpuConstraints[i].m_batchIdx);
+					printf("c[%d].m_bodyA = %d, m_bodyB = %d, batchId = %d\n", i, cpuConstraints[i].m_bodyAPtrAndSignBit, cpuConstraints[i].m_bodyBPtrAndSignBit, cpuConstraints[i].m_batchIdx);
 				}
 			}
 
 			m_debugOutput++;
 		}
 	}
-	
-
-
 
 	int nContactOut = m_data->m_pBufContactOutGPU->size();
 
 	bool useSolver = true;
-	
 
-    if (useSolver)
-    {
-        float dt=1./60.;
-        b3ConstraintCfg csCfg( dt );
-        csCfg.m_enableParallelSolve = true;
-        csCfg.m_batchCellSize = 6;
-        csCfg.m_staticIdx = static0Index;
-        
-        
-        b3OpenCLArray<b3RigidBodyData>* bodyBuf = m_data->m_bodyBufferGPU;
+	if (useSolver)
+	{
+		float dt = 1. / 60.;
+		b3ConstraintCfg csCfg(dt);
+		csCfg.m_enableParallelSolve = true;
+		csCfg.m_batchCellSize = 6;
+		csCfg.m_staticIdx = static0Index;
+
+		b3OpenCLArray<b3RigidBodyData>* bodyBuf = m_data->m_bodyBufferGPU;
+
+		void* additionalData = 0;  //m_data->m_frictionCGPU;
+		const b3OpenCLArray<b3InertiaData>* shapeBuf = m_data->m_inertiaBufferGPU;
+		b3OpenCLArray<b3GpuConstraint4>* contactConstraintOut = m_data->m_contactCGPU;
+		int nContacts = nContactOut;
 
-        void* additionalData = 0;//m_data->m_frictionCGPU;
-        const b3OpenCLArray<b3InertiaData>* shapeBuf = m_data->m_inertiaBufferGPU;
-        b3OpenCLArray<b3GpuConstraint4>* contactConstraintOut = m_data->m_contactCGPU;
-        int nContacts = nContactOut;
-        
-        
 		int maxNumBatches = 0;
- 
+
 		if (!gUseLargeBatches)
-        {
-            
-            if( m_data->m_solverGPU->m_contactBuffer2)
-            {
-                m_data->m_solverGPU->m_contactBuffer2->resize(nContacts);
-            }
-            
-            if( m_data->m_solverGPU->m_contactBuffer2 == 0 )
-            {
-				m_data->m_solverGPU->m_contactBuffer2 = new b3OpenCLArray<b3Contact4>(m_data->m_context,m_data->m_queue, nContacts );
-                m_data->m_solverGPU->m_contactBuffer2->resize(nContacts);
-            }
-			
-            //clFinish(m_data->m_queue);
-            
-            
-            
+		{
+			if (m_data->m_solverGPU->m_contactBuffer2)
+			{
+				m_data->m_solverGPU->m_contactBuffer2->resize(nContacts);
+			}
+
+			if (m_data->m_solverGPU->m_contactBuffer2 == 0)
+			{
+				m_data->m_solverGPU->m_contactBuffer2 = new b3OpenCLArray<b3Contact4>(m_data->m_context, m_data->m_queue, nContacts);
+				m_data->m_solverGPU->m_contactBuffer2->resize(nContacts);
+			}
+
+			//clFinish(m_data->m_queue);
+
 			{
 				B3_PROFILE("batching");
 				//@todo: just reserve it, without copy of original contact (unless we use warmstarting)
 
-
-
 				//const b3OpenCLArray<b3RigidBodyData>* bodyNative = bodyBuf;
 
-
 				{
-
 					//b3OpenCLArray<b3RigidBodyData>* bodyNative = b3OpenCLArrayUtils::map<adl::TYPE_CL, true>( data->m_device, bodyBuf );
 					//b3OpenCLArray<b3Contact4>* contactNative = b3OpenCLArrayUtils::map<adl::TYPE_CL, true>( data->m_device, contactsIn );
 
-					const int sortAlignment = 512; // todo. get this out of sort
-					if( csCfg.m_enableParallelSolve )
+					const int sortAlignment = 512;  // todo. get this out of sort
+					if (csCfg.m_enableParallelSolve)
 					{
-
-
-						int sortSize = B3NEXTMULTIPLEOF( nContacts, sortAlignment );
+						int sortSize = B3NEXTMULTIPLEOF(nContacts, sortAlignment);
 
 						b3OpenCLArray<unsigned int>* countsNative = m_data->m_solverGPU->m_numConstraints;
 						b3OpenCLArray<unsigned int>* offsetsNative = m_data->m_solverGPU->m_offsets;
 
-
 						if (!gCpuSetSortData)
-						{	//	2. set cell idx
+						{  //	2. set cell idx
 							B3_PROFILE("GPU set cell idx");
 							struct CB
 							{
@@ -834,29 +728,28 @@ void b3GpuPgsContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem
 								b3Int4 m_nSplit;
 							};
 
-							b3Assert( sortSize%64 == 0 );
+							b3Assert(sortSize % 64 == 0);
 							CB cdata;
 							cdata.m_nContacts = nContacts;
 							cdata.m_staticIdx = csCfg.m_staticIdx;
-							cdata.m_scale = 1.f/csCfg.m_batchCellSize;
+							cdata.m_scale = 1.f / csCfg.m_batchCellSize;
 							cdata.m_nSplit.x = B3_SOLVER_N_SPLIT_X;
 							cdata.m_nSplit.y = B3_SOLVER_N_SPLIT_Y;
 							cdata.m_nSplit.z = B3_SOLVER_N_SPLIT_Z;
 
 							m_data->m_solverGPU->m_sortDataBuffer->resize(nContacts);
 
-
-							b3BufferInfoCL bInfo[] = { b3BufferInfoCL( m_data->m_pBufContactOutGPU->getBufferCL() ), b3BufferInfoCL( bodyBuf->getBufferCL()), b3BufferInfoCL( m_data->m_solverGPU->m_sortDataBuffer->getBufferCL()) };
-							b3LauncherCL launcher(m_data->m_queue, m_data->m_solverGPU->m_setSortDataKernel,"m_setSortDataKernel" );
-							launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-							launcher.setConst( cdata.m_nContacts );
-							launcher.setConst( cdata.m_scale );
+							b3BufferInfoCL bInfo[] = {b3BufferInfoCL(m_data->m_pBufContactOutGPU->getBufferCL()), b3BufferInfoCL(bodyBuf->getBufferCL()), b3BufferInfoCL(m_data->m_solverGPU->m_sortDataBuffer->getBufferCL())};
+							b3LauncherCL launcher(m_data->m_queue, m_data->m_solverGPU->m_setSortDataKernel, "m_setSortDataKernel");
+							launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+							launcher.setConst(cdata.m_nContacts);
+							launcher.setConst(cdata.m_scale);
 							launcher.setConst(cdata.m_nSplit);
 							launcher.setConst(cdata.m_staticIdx);
 
-
-							launcher.launch1D( sortSize, 64 );
-						} else
+							launcher.launch1D(sortSize, 64);
+						}
+						else
 						{
 							m_data->m_solverGPU->m_sortDataBuffer->resize(nContacts);
 							b3AlignedObjectArray<b3SortData> sortDataCPU;
@@ -866,22 +759,19 @@ void b3GpuPgsContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem
 							m_data->m_pBufContactOutGPU->copyToHost(contactCPU);
 							b3AlignedObjectArray<b3RigidBodyData> bodiesCPU;
 							bodyBuf->copyToHost(bodiesCPU);
-							float scale = 1.f/csCfg.m_batchCellSize;
+							float scale = 1.f / csCfg.m_batchCellSize;
 							b3Int4 nSplit;
 							nSplit.x = B3_SOLVER_N_SPLIT_X;
 							nSplit.y = B3_SOLVER_N_SPLIT_Y;
 							nSplit.z = B3_SOLVER_N_SPLIT_Z;
 
-							SetSortDataCPU(&contactCPU[0],  &bodiesCPU[0], &sortDataCPU[0], nContacts,scale,nSplit,csCfg.m_staticIdx);
-
+							SetSortDataCPU(&contactCPU[0], &bodiesCPU[0], &sortDataCPU[0], nContacts, scale, nSplit, csCfg.m_staticIdx);
 
 							m_data->m_solverGPU->m_sortDataBuffer->copyFromHost(sortDataCPU);
 						}
 
-
-
 						if (!gCpuRadixSort)
-						{	//	3. sort by cell idx
+						{  //	3. sort by cell idx
 							B3_PROFILE("gpuRadixSort");
 							//int n = B3_SOLVER_N_SPLIT*B3_SOLVER_N_SPLIT;
 							//int sortBit = 32;
@@ -891,10 +781,8 @@ void b3GpuPgsContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem
 							//adl::RadixSort32<adl::TYPE_CL>::execute( data->m_sort32, *data->m_sortDataBuffer, sortSize );
 							b3OpenCLArray<b3SortData>& keyValuesInOut = *(m_data->m_solverGPU->m_sortDataBuffer);
 							this->m_data->m_solverGPU->m_sort32->execute(keyValuesInOut);
-
-
-
-						} else
+						}
+						else
 						{
 							b3OpenCLArray<b3SortData>& keyValuesInOut = *(m_data->m_solverGPU->m_sortDataBuffer);
 							b3AlignedObjectArray<b3SortData> hostValues;
@@ -903,7 +791,6 @@ void b3GpuPgsContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem
 							keyValuesInOut.copyFromHost(hostValues);
 						}
 
-
 						if (gUseScanHost)
 						{
 							//	4. find entries
@@ -914,13 +801,11 @@ void b3GpuPgsContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem
 							b3AlignedObjectArray<b3SortData> sortDataHost;
 							m_data->m_solverGPU->m_sortDataBuffer->copyToHost(sortDataHost);
 
-
 							//m_data->m_solverGPU->m_search->executeHost(*m_data->m_solverGPU->m_sortDataBuffer,nContacts,*countsNative,B3_SOLVER_N_CELLS,b3BoundSearchCL::COUNT);
-							m_data->m_solverGPU->m_search->executeHost(sortDataHost,nContacts,countsHost,B3_SOLVER_N_CELLS,b3BoundSearchCL::COUNT);
+							m_data->m_solverGPU->m_search->executeHost(sortDataHost, nContacts, countsHost, B3_SOLVER_N_CELLS, b3BoundSearchCL::COUNT);
 
 							countsNative->copyFromHost(countsHost);
 
-
 							//adl::BoundSearch<adl::TYPE_CL>::execute( data->m_search, *data->m_sortDataBuffer, nContacts, *countsNative,
 							//	B3_SOLVER_N_SPLIT*B3_SOLVER_N_SPLIT, adl::BoundSearchBase::COUNT );
 
@@ -929,24 +814,21 @@ void b3GpuPgsContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem
 							b3AlignedObjectArray<unsigned int> offsetsHost;
 							offsetsHost.resize(offsetsNative->size());
 
-
-							m_data->m_solverGPU->m_scan->executeHost(countsHost,offsetsHost, B3_SOLVER_N_CELLS);//,&sum );
+							m_data->m_solverGPU->m_scan->executeHost(countsHost, offsetsHost, B3_SOLVER_N_CELLS);  //,&sum );
 							offsetsNative->copyFromHost(offsetsHost);
 
 							//printf("sum = %d\n",sum);
-						}  else
+						}
+						else
 						{
 							//	4. find entries
 							B3_PROFILE("gpuBoundSearch");
-							m_data->m_solverGPU->m_search->execute(*m_data->m_solverGPU->m_sortDataBuffer,nContacts,*countsNative,B3_SOLVER_N_CELLS,b3BoundSearchCL::COUNT);
-							m_data->m_solverGPU->m_scan->execute(*countsNative,*offsetsNative, B3_SOLVER_N_CELLS);//,&sum );
-						} 
-
-
-
+							m_data->m_solverGPU->m_search->execute(*m_data->m_solverGPU->m_sortDataBuffer, nContacts, *countsNative, B3_SOLVER_N_CELLS, b3BoundSearchCL::COUNT);
+							m_data->m_solverGPU->m_scan->execute(*countsNative, *offsetsNative, B3_SOLVER_N_CELLS);  //,&sum );
+						}
 
 						if (nContacts)
-						{	//	5. sort constraints by cellIdx
+						{  //	5. sort constraints by cellIdx
 							if (gReorderContactsOnCpu)
 							{
 								B3_PROFILE("cpu m_reorderContactKernel");
@@ -956,7 +838,7 @@ void b3GpuPgsContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem
 								b3AlignedObjectArray<b3Contact4> outContacts;
 								m_data->m_pBufContactOutGPU->copyToHost(inContacts);
 								outContacts.resize(inContacts.size());
-								for (int i=0;i<nContacts;i++)
+								for (int i = 0; i < nContacts; i++)
 								{
 									int srcIdx = sortDataHost[i].y;
 									outContacts[i] = inContacts[srcIdx];
@@ -974,30 +856,25 @@ void b3GpuPgsContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem
 								"	}\n"
 								"}\n"
 								*/
-							} else
+							}
+							else
 							{
 								B3_PROFILE("gpu m_reorderContactKernel");
 
 								b3Int4 cdata;
 								cdata.x = nContacts;
 
-								b3BufferInfoCL bInfo[] = { 
-									b3BufferInfoCL( m_data->m_pBufContactOutGPU->getBufferCL() ), 
-									b3BufferInfoCL( m_data->m_solverGPU->m_contactBuffer2->getBufferCL())
-									, b3BufferInfoCL( m_data->m_solverGPU->m_sortDataBuffer->getBufferCL()) };
+								b3BufferInfoCL bInfo[] = {
+									b3BufferInfoCL(m_data->m_pBufContactOutGPU->getBufferCL()),
+									b3BufferInfoCL(m_data->m_solverGPU->m_contactBuffer2->getBufferCL()), b3BufferInfoCL(m_data->m_solverGPU->m_sortDataBuffer->getBufferCL())};
 
-									b3LauncherCL launcher(m_data->m_queue,m_data->m_solverGPU->m_reorderContactKernel,"m_reorderContactKernel");
-									launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-									launcher.setConst( cdata );
-									launcher.launch1D( nContacts, 64 );
+								b3LauncherCL launcher(m_data->m_queue, m_data->m_solverGPU->m_reorderContactKernel, "m_reorderContactKernel");
+								launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+								launcher.setConst(cdata);
+								launcher.launch1D(nContacts, 64);
 							}
 						}
-
-
-
-
 					}
-
 				}
 
 				//clFinish(m_data->m_queue);
@@ -1008,48 +885,46 @@ void b3GpuPgsContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem
 				//				printf(",,,\n");
 				//				}
 
-
 				if (nContacts)
 				{
-
 					if (gUseCpuCopyConstraints)
 					{
-						for (int i=0;i<nContacts;i++)
+						for (int i = 0; i < nContacts; i++)
 						{
 							m_data->m_pBufContactOutGPU->copyFromOpenCLArray(*m_data->m_solverGPU->m_contactBuffer2);
-							//							m_data->m_solverGPU->m_contactBuffer2->getBufferCL(); 
-							//						m_data->m_pBufContactOutGPU->getBufferCL() 
+							//							m_data->m_solverGPU->m_contactBuffer2->getBufferCL();
+							//						m_data->m_pBufContactOutGPU->getBufferCL()
 						}
-
-					} else
+					}
+					else
 					{
 						B3_PROFILE("gpu m_copyConstraintKernel");
-						b3Int4 cdata; cdata.x = nContacts;
-						b3BufferInfoCL bInfo[] = { 
-							b3BufferInfoCL(  m_data->m_solverGPU->m_contactBuffer2->getBufferCL() ), 
-							b3BufferInfoCL( m_data->m_pBufContactOutGPU->getBufferCL() ) 
-						};
+						b3Int4 cdata;
+						cdata.x = nContacts;
+						b3BufferInfoCL bInfo[] = {
+							b3BufferInfoCL(m_data->m_solverGPU->m_contactBuffer2->getBufferCL()),
+							b3BufferInfoCL(m_data->m_pBufContactOutGPU->getBufferCL())};
 
-						b3LauncherCL launcher(m_data->m_queue, m_data->m_solverGPU->m_copyConstraintKernel,"m_copyConstraintKernel" );
-						launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
-						launcher.setConst(  cdata );
-						launcher.launch1D( nContacts, 64 );
+						b3LauncherCL launcher(m_data->m_queue, m_data->m_solverGPU->m_copyConstraintKernel, "m_copyConstraintKernel");
+						launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
+						launcher.setConst(cdata);
+						launcher.launch1D(nContacts, 64);
 						//we use the clFinish for proper benchmark/profile
 						clFinish(m_data->m_queue);
 					}
 				}
 
-
-//				bool compareGPU = false;
+				//				bool compareGPU = false;
 				if (nContacts)
 				{
 					if (!gCpuBatchContacts)
 					{
 						B3_PROFILE("gpu batchContacts");
-						maxNumBatches = 250;//250;
-						m_data->m_solverGPU->batchContacts( m_data->m_pBufContactOutGPU, nContacts, m_data->m_solverGPU->m_numConstraints, m_data->m_solverGPU->m_offsets, csCfg.m_staticIdx );
+						maxNumBatches = 250;  //250;
+						m_data->m_solverGPU->batchContacts(m_data->m_pBufContactOutGPU, nContacts, m_data->m_solverGPU->m_numConstraints, m_data->m_solverGPU->m_offsets, csCfg.m_staticIdx);
 						clFinish(m_data->m_queue);
-					} else
+					}
+					else
 					{
 						B3_PROFILE("cpu batchContacts");
 						static b3AlignedObjectArray<b3Contact4> cpuContacts;
@@ -1070,45 +945,43 @@ void b3GpuPgsContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem
 							offsetsNative->copyToHost(offsetsNativeHost);
 						}
 
-
-						int numNonzeroGrid=0;
+						int numNonzeroGrid = 0;
 
 						if (gUseLargeBatches)
 						{
 							m_data->m_batchSizes.resize(B3_MAX_NUM_BATCHES);
 							int totalNumConstraints = cpuContacts.size();
 							//int simdWidth =numBodies+1;//-1;//64;//-1;//32;
-							int numBatches = sortConstraintByBatch3( &cpuContacts[0], totalNumConstraints, totalNumConstraints+1,csCfg.m_staticIdx ,numBodies,&m_data->m_batchSizes[0]);	//	on GPU
-							maxNumBatches = b3Max(numBatches,maxNumBatches);
+							int numBatches = sortConstraintByBatch3(&cpuContacts[0], totalNumConstraints, totalNumConstraints + 1, csCfg.m_staticIdx, numBodies, &m_data->m_batchSizes[0]);  //	on GPU
+							maxNumBatches = b3Max(numBatches, maxNumBatches);
 							static int globalMaxBatch = 0;
-							if (maxNumBatches>globalMaxBatch )
+							if (maxNumBatches > globalMaxBatch)
 							{
-								globalMaxBatch  = maxNumBatches;
-								b3Printf("maxNumBatches = %d\n",maxNumBatches);
+								globalMaxBatch = maxNumBatches;
+								b3Printf("maxNumBatches = %d\n", maxNumBatches);
 							}
-								
-						} else
+						}
+						else
 						{
-							m_data->m_batchSizes.resize(B3_SOLVER_N_CELLS*B3_MAX_NUM_BATCHES);
+							m_data->m_batchSizes.resize(B3_SOLVER_N_CELLS * B3_MAX_NUM_BATCHES);
 							B3_PROFILE("cpu batch grid");
-							for(int i=0; i<B3_SOLVER_N_CELLS; i++)
+							for (int i = 0; i < B3_SOLVER_N_CELLS; i++)
 							{
 								int n = (nNativeHost)[i];
 								int offset = (offsetsNativeHost)[i];
-								if( n )
+								if (n)
 								{
 									numNonzeroGrid++;
-									int simdWidth =numBodies+1;//-1;//64;//-1;//32;
-									int numBatches = sortConstraintByBatch3( &cpuContacts[0]+offset, n, simdWidth,csCfg.m_staticIdx ,numBodies,&m_data->m_batchSizes[i*B3_MAX_NUM_BATCHES]);	//	on GPU
-									maxNumBatches = b3Max(numBatches,maxNumBatches);
+									int simdWidth = numBodies + 1;                                                                                                                                 //-1;//64;//-1;//32;
+									int numBatches = sortConstraintByBatch3(&cpuContacts[0] + offset, n, simdWidth, csCfg.m_staticIdx, numBodies, &m_data->m_batchSizes[i * B3_MAX_NUM_BATCHES]);  //	on GPU
+									maxNumBatches = b3Max(numBatches, maxNumBatches);
 									static int globalMaxBatch = 0;
-									if (maxNumBatches>globalMaxBatch )
+									if (maxNumBatches > globalMaxBatch)
 									{
-										globalMaxBatch  = maxNumBatches;
-										b3Printf("maxNumBatches = %d\n",maxNumBatches);
+										globalMaxBatch = maxNumBatches;
+										b3Printf("maxNumBatches = %d\n", maxNumBatches);
 									}
 									//we use the clFinish for proper benchmark/profile
-									
 								}
 							}
 							//clFinish(m_data->m_queue);
@@ -1117,22 +990,12 @@ void b3GpuPgsContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem
 							B3_PROFILE("m_contactBuffer->copyFromHost");
 							m_data->m_solverGPU->m_contactBuffer2->copyFromHost((b3AlignedObjectArray<b3Contact4>&)cpuContacts);
 						}
-
-					} 
-
+					}
 				}
+			}
+		}
 
-
-			
-
-
-			} 
-
-
-		} 
-
-
-			//printf("maxNumBatches = %d\n", maxNumBatches);
+		//printf("maxNumBatches = %d\n", maxNumBatches);
 
 		if (gUseLargeBatches)
 		{
@@ -1140,58 +1003,52 @@ void b3GpuPgsContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem
 			{
 				B3_PROFILE("cpu batchContacts");
 				static b3AlignedObjectArray<b3Contact4> cpuContacts;
-//				b3OpenCLArray<b3Contact4>* contactsIn = m_data->m_solverGPU->m_contactBuffer2;
+				//				b3OpenCLArray<b3Contact4>* contactsIn = m_data->m_solverGPU->m_contactBuffer2;
 				{
 					B3_PROFILE("copyToHost");
 					m_data->m_pBufContactOutGPU->copyToHost(cpuContacts);
 				}
-//				b3OpenCLArray<unsigned int>* countsNative = m_data->m_solverGPU->m_numConstraints;
-//				b3OpenCLArray<unsigned int>* offsetsNative = m_data->m_solverGPU->m_offsets;
+				//				b3OpenCLArray<unsigned int>* countsNative = m_data->m_solverGPU->m_numConstraints;
+				//				b3OpenCLArray<unsigned int>* offsetsNative = m_data->m_solverGPU->m_offsets;
 
-
-
-//				int numNonzeroGrid=0;
+				//				int numNonzeroGrid=0;
 
 				{
 					m_data->m_batchSizes.resize(B3_MAX_NUM_BATCHES);
 					int totalNumConstraints = cpuContacts.size();
-	//				int simdWidth =numBodies+1;//-1;//64;//-1;//32;
-					int numBatches = sortConstraintByBatch3( &cpuContacts[0], totalNumConstraints, totalNumConstraints+1,csCfg.m_staticIdx ,numBodies,&m_data->m_batchSizes[0]);	//	on GPU
-					maxNumBatches = b3Max(numBatches,maxNumBatches);
+					//				int simdWidth =numBodies+1;//-1;//64;//-1;//32;
+					int numBatches = sortConstraintByBatch3(&cpuContacts[0], totalNumConstraints, totalNumConstraints + 1, csCfg.m_staticIdx, numBodies, &m_data->m_batchSizes[0]);  //	on GPU
+					maxNumBatches = b3Max(numBatches, maxNumBatches);
 					static int globalMaxBatch = 0;
-					if (maxNumBatches>globalMaxBatch )
+					if (maxNumBatches > globalMaxBatch)
 					{
-						globalMaxBatch  = maxNumBatches;
-						b3Printf("maxNumBatches = %d\n",maxNumBatches);
+						globalMaxBatch = maxNumBatches;
+						b3Printf("maxNumBatches = %d\n", maxNumBatches);
 					}
-								
 				}
 				{
 					B3_PROFILE("m_contactBuffer->copyFromHost");
 					m_data->m_solverGPU->m_contactBuffer2->copyFromHost((b3AlignedObjectArray<b3Contact4>&)cpuContacts);
 				}
-
-			} 
-
+			}
 		}
 
 		if (nContacts)
 		{
 			B3_PROFILE("gpu convertToConstraints");
-			m_data->m_solverGPU->convertToConstraints( bodyBuf, 
-				shapeBuf, m_data->m_solverGPU->m_contactBuffer2,
-				contactConstraintOut, 
-				additionalData, nContacts, 
-				(b3SolverBase::ConstraintCfg&) csCfg );
+			m_data->m_solverGPU->convertToConstraints(bodyBuf,
+													  shapeBuf, m_data->m_solverGPU->m_contactBuffer2,
+													  contactConstraintOut,
+													  additionalData, nContacts,
+													  (b3SolverBase::ConstraintCfg&)csCfg);
 			clFinish(m_data->m_queue);
 		}
 
-
 		if (1)
 		{
 			int numIter = 4;
 
-			m_data->m_solverGPU->m_nIterations = numIter;//10
+			m_data->m_solverGPU->m_nIterations = numIter;  //10
 			if (!gCpuSolveConstraint)
 			{
 				B3_PROFILE("GPU solveContactConstraint");
@@ -1208,32 +1065,30 @@ void b3GpuPgsContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem
 
 				if (gUseLargeBatches)
 				{
-					solveContactConstraintBatchSizes(m_data->m_bodyBufferGPU, 
-						m_data->m_inertiaBufferGPU,
-						m_data->m_contactCGPU,0,
-						nContactOut ,
-						maxNumBatches,numIter,&m_data->m_batchSizes);
-				} else
+					solveContactConstraintBatchSizes(m_data->m_bodyBufferGPU,
+													 m_data->m_inertiaBufferGPU,
+													 m_data->m_contactCGPU, 0,
+													 nContactOut,
+													 maxNumBatches, numIter, &m_data->m_batchSizes);
+				}
+				else
 				{
 					solveContactConstraint(
-						m_data->m_bodyBufferGPU, 
+						m_data->m_bodyBufferGPU,
 						m_data->m_inertiaBufferGPU,
-						m_data->m_contactCGPU,0,
-						nContactOut ,
-						maxNumBatches,numIter,&m_data->m_batchSizes);//m_data->m_batchSizesGpu);
+						m_data->m_contactCGPU, 0,
+						nContactOut,
+						maxNumBatches, numIter, &m_data->m_batchSizes);  //m_data->m_batchSizesGpu);
 				}
 			}
 			else
 			{
 				B3_PROFILE("Host solveContactConstraint");
 
-				m_data->m_solverGPU->solveContactConstraintHost(m_data->m_bodyBufferGPU, m_data->m_inertiaBufferGPU, m_data->m_contactCGPU,0, nContactOut ,maxNumBatches,&m_data->m_batchSizes);
+				m_data->m_solverGPU->solveContactConstraintHost(m_data->m_bodyBufferGPU, m_data->m_inertiaBufferGPU, m_data->m_contactCGPU, 0, nContactOut, maxNumBatches, &m_data->m_batchSizes);
 			}
-            
-            
-        }
-        
-        
+		}
+
 #if 0
         if (0)
         {
@@ -1244,114 +1099,96 @@ void b3GpuPgsContactSolver::solveContacts(int numBodies, cl_mem bodyBuf, cl_mem
             adl::DeviceUtils::waitForCompletion( m_data->m_deviceCL );
         }
 #endif
-        
-    }
-
+	}
 }
 
-
-void b3GpuPgsContactSolver::batchContacts( b3OpenCLArray<b3Contact4>* contacts, int nContacts, b3OpenCLArray<unsigned int>* n, b3OpenCLArray<unsigned int>* offsets, int staticIdx )
+void b3GpuPgsContactSolver::batchContacts(b3OpenCLArray<b3Contact4>* contacts, int nContacts, b3OpenCLArray<unsigned int>* n, b3OpenCLArray<unsigned int>* offsets, int staticIdx)
 {
 }
 
-
-
-
-
-
-
-
-
-
-
 b3AlignedObjectArray<unsigned int> idxBuffer;
 b3AlignedObjectArray<b3SortData> sortData;
 b3AlignedObjectArray<b3Contact4> old;
 
-
-inline int b3GpuPgsContactSolver::sortConstraintByBatch( b3Contact4* cs, int n, int simdWidth , int staticIdx, int numBodies)
+inline int b3GpuPgsContactSolver::sortConstraintByBatch(b3Contact4* cs, int n, int simdWidth, int staticIdx, int numBodies)
 {
-	
 	B3_PROFILE("sortConstraintByBatch");
 	int numIter = 0;
-    
+
 	sortData.resize(n);
 	idxBuffer.resize(n);
 	old.resize(n);
-	
+
 	unsigned int* idxSrc = &idxBuffer[0];
 	unsigned int* idxDst = &idxBuffer[0];
 	int nIdxSrc, nIdxDst;
-    
+
 	const int N_FLG = 256;
-	const int FLG_MASK = N_FLG-1;
-	unsigned int flg[N_FLG/32];
+	const int FLG_MASK = N_FLG - 1;
+	unsigned int flg[N_FLG / 32];
 #if defined(_DEBUG)
-	for(int i=0; i<n; i++)
+	for (int i = 0; i < n; i++)
 		cs[i].getBatchIdx() = -1;
 #endif
-	for(int i=0; i<n; i++) 
+	for (int i = 0; i < n; i++)
 		idxSrc[i] = i;
 	nIdxSrc = n;
-    
+
 	int batchIdx = 0;
-    
+
 	{
 		B3_PROFILE("cpu batch innerloop");
-		while( nIdxSrc )
+		while (nIdxSrc)
 		{
 			numIter++;
 			nIdxDst = 0;
 			int nCurrentBatch = 0;
-            
+
 			//	clear flag
-			for(int i=0; i<N_FLG/32; i++) flg[i] = 0;
-            
-			for(int i=0; i<nIdxSrc; i++)
+			for (int i = 0; i < N_FLG / 32; i++) flg[i] = 0;
+
+			for (int i = 0; i < nIdxSrc; i++)
 			{
 				int idx = idxSrc[i];
-				
 
-				b3Assert( idx < n );
+				b3Assert(idx < n);
 				//	check if it can go
 				int bodyAS = cs[idx].m_bodyAPtrAndSignBit;
 				int bodyBS = cs[idx].m_bodyBPtrAndSignBit;
-                
-				
-                
+
 				int bodyA = abs(bodyAS);
 				int bodyB = abs(bodyBS);
-                
+
 				int aIdx = bodyA & FLG_MASK;
 				int bIdx = bodyB & FLG_MASK;
-                
-				unsigned int aUnavailable = flg[ aIdx/32 ] & (1<<(aIdx&31));
-				unsigned int bUnavailable = flg[ bIdx/32 ] & (1<<(bIdx&31));
-                
-				bool aIsStatic = (bodyAS<0) || bodyAS==staticIdx;
-				bool bIsStatic = (bodyBS<0) || bodyBS==staticIdx;
 
-                //use inv_mass!
-				aUnavailable = !aIsStatic? aUnavailable:0;//
-				bUnavailable = !bIsStatic? bUnavailable:0;
-                
-				if( aUnavailable==0 && bUnavailable==0 ) // ok
+				unsigned int aUnavailable = flg[aIdx / 32] & (1 << (aIdx & 31));
+				unsigned int bUnavailable = flg[bIdx / 32] & (1 << (bIdx & 31));
+
+				bool aIsStatic = (bodyAS < 0) || bodyAS == staticIdx;
+				bool bIsStatic = (bodyBS < 0) || bodyBS == staticIdx;
+
+				//use inv_mass!
+				aUnavailable = !aIsStatic ? aUnavailable : 0;  //
+				bUnavailable = !bIsStatic ? bUnavailable : 0;
+
+				if (aUnavailable == 0 && bUnavailable == 0)  // ok
 				{
 					if (!aIsStatic)
-						flg[ aIdx/32 ] |= (1<<(aIdx&31));
+						flg[aIdx / 32] |= (1 << (aIdx & 31));
 					if (!bIsStatic)
-						flg[ bIdx/32 ] |= (1<<(bIdx&31));
+						flg[bIdx / 32] |= (1 << (bIdx & 31));
 
 					cs[idx].getBatchIdx() = batchIdx;
 					sortData[idx].m_key = batchIdx;
 					sortData[idx].m_value = idx;
-                    
+
 					{
 						nCurrentBatch++;
-						if( nCurrentBatch == simdWidth )
+						if (nCurrentBatch == simdWidth)
 						{
 							nCurrentBatch = 0;
-							for(int i=0; i<N_FLG/32; i++) flg[i] = 0;
+							for (int i = 0; i < N_FLG / 32; i++) flg[i] = 0;
 						}
 					}
 				}
@@ -1360,128 +1197,121 @@ inline int b3GpuPgsContactSolver::sortConstraintByBatch( b3Contact4* cs, int n,
 					idxDst[nIdxDst++] = idx;
 				}
 			}
-			b3Swap( idxSrc, idxDst );
-			b3Swap( nIdxSrc, nIdxDst );
-			batchIdx ++;
+			b3Swap(idxSrc, idxDst);
+			b3Swap(nIdxSrc, nIdxDst);
+			batchIdx++;
 		}
 	}
 	{
 		B3_PROFILE("quickSort");
 		sortData.quickSort(sortfnc);
 	}
-	
-	
+
 	{
-        B3_PROFILE("reorder");
+		B3_PROFILE("reorder");
 		//	reorder
-		
-		memcpy( &old[0], cs, sizeof(b3Contact4)*n);
-		for(int i=0; i<n; i++)
+
+		memcpy(&old[0], cs, sizeof(b3Contact4) * n);
+		for (int i = 0; i < n; i++)
 		{
 			int idx = sortData[i].m_value;
 			cs[i] = old[idx];
 		}
 	}
-    
-	
+
 #if defined(_DEBUG)
-    //		debugPrintf( "nBatches: %d\n", batchIdx );
-	for(int i=0; i<n; i++)
-    {
-        b3Assert( cs[i].getBatchIdx() != -1 );
-    }
+	//		debugPrintf( "nBatches: %d\n", batchIdx );
+	for (int i = 0; i < n; i++)
+	{
+		b3Assert(cs[i].getBatchIdx() != -1);
+	}
 #endif
 	return batchIdx;
 }
 
-
 b3AlignedObjectArray<int> bodyUsed2;
 
-inline int b3GpuPgsContactSolver::sortConstraintByBatch2( b3Contact4* cs, int numConstraints, int simdWidth , int staticIdx, int numBodies)
+inline int b3GpuPgsContactSolver::sortConstraintByBatch2(b3Contact4* cs, int numConstraints, int simdWidth, int staticIdx, int numBodies)
 {
-	
 	B3_PROFILE("sortConstraintByBatch2");
-	
 
-	
-	bodyUsed2.resize(2*simdWidth);
+	bodyUsed2.resize(2 * simdWidth);
 
-	for (int q=0;q<2*simdWidth;q++)
-		bodyUsed2[q]=0;
+	for (int q = 0; q < 2 * simdWidth; q++)
+		bodyUsed2[q] = 0;
 
 	int curBodyUsed = 0;
 
 	int numIter = 0;
-    
+
 	m_data->m_sortData.resize(numConstraints);
 	m_data->m_idxBuffer.resize(numConstraints);
 	m_data->m_old.resize(numConstraints);
-	
+
 	unsigned int* idxSrc = &m_data->m_idxBuffer[0];
-		
+
 #if defined(_DEBUG)
-	for(int i=0; i<numConstraints; i++)
+	for (int i = 0; i < numConstraints; i++)
 		cs[i].getBatchIdx() = -1;
 #endif
-	for(int i=0; i<numConstraints; i++) 
+	for (int i = 0; i < numConstraints; i++)
 		idxSrc[i] = i;
-    
+
 	int numValidConstraints = 0;
-//	int unprocessedConstraintIndex = 0;
+	//	int unprocessedConstraintIndex = 0;
 
 	int batchIdx = 0;
-    
 
 	{
 		B3_PROFILE("cpu batch innerloop");
-		
-		while( numValidConstraints < numConstraints)
+
+		while (numValidConstraints < numConstraints)
 		{
 			numIter++;
 			int nCurrentBatch = 0;
 			//	clear flag
-			for(int i=0; i<curBodyUsed; i++) 
+			for (int i = 0; i < curBodyUsed; i++)
 				bodyUsed2[i] = 0;
-            curBodyUsed = 0;
+			curBodyUsed = 0;
 
-			for(int i=numValidConstraints; i<numConstraints; i++)
+			for (int i = numValidConstraints; i < numConstraints; i++)
 			{
 				int idx = idxSrc[i];
-				b3Assert( idx < numConstraints );
+				b3Assert(idx < numConstraints);
 				//	check if it can go
 				int bodyAS = cs[idx].m_bodyAPtrAndSignBit;
 				int bodyBS = cs[idx].m_bodyBPtrAndSignBit;
 				int bodyA = abs(bodyAS);
 				int bodyB = abs(bodyBS);
-				bool aIsStatic = (bodyAS<0) || bodyAS==staticIdx;
-				bool bIsStatic = (bodyBS<0) || bodyBS==staticIdx;
+				bool aIsStatic = (bodyAS < 0) || bodyAS == staticIdx;
+				bool bIsStatic = (bodyBS < 0) || bodyBS == staticIdx;
 				int aUnavailable = 0;
 				int bUnavailable = 0;
 				if (!aIsStatic)
 				{
-					for (int j=0;j<curBodyUsed;j++)
+					for (int j = 0; j < curBodyUsed; j++)
 					{
 						if (bodyA == bodyUsed2[j])
 						{
-							aUnavailable=1;
+							aUnavailable = 1;
 							break;
 						}
 					}
 				}
 				if (!aUnavailable)
-				if (!bIsStatic)
-				{
-					for (int j=0;j<curBodyUsed;j++)
+					if (!bIsStatic)
 					{
-						if (bodyB == bodyUsed2[j])
+						for (int j = 0; j < curBodyUsed; j++)
 						{
-							bUnavailable=1;
-							break;
+							if (bodyB == bodyUsed2[j])
+							{
+								bUnavailable = 1;
+								break;
+							}
 						}
 					}
-				}
-                
-				if( aUnavailable==0 && bUnavailable==0 ) // ok
+
+				if (aUnavailable == 0 && bUnavailable == 0)  // ok
 				{
 					if (!aIsStatic)
 					{
@@ -1496,7 +1326,7 @@ inline int b3GpuPgsContactSolver::sortConstraintByBatch2( b3Contact4* cs, int nu
 					m_data->m_sortData[idx].m_key = batchIdx;
 					m_data->m_sortData[idx].m_value = idx;
 
-					if (i!=numValidConstraints)
+					if (i != numValidConstraints)
 					{
 						b3Swap(idxSrc[i], idxSrc[numValidConstraints]);
 					}
@@ -1504,20 +1334,19 @@ inline int b3GpuPgsContactSolver::sortConstraintByBatch2( b3Contact4* cs, int nu
 					numValidConstraints++;
 					{
 						nCurrentBatch++;
-						if( nCurrentBatch == simdWidth )
+						if (nCurrentBatch == simdWidth)
 						{
 							nCurrentBatch = 0;
-							for(int i=0; i<curBodyUsed; i++) 
+							for (int i = 0; i < curBodyUsed; i++)
 								bodyUsed2[i] = 0;
 
-							
 							curBodyUsed = 0;
 						}
 					}
 				}
 			}
-			
-			batchIdx ++;
+
+			batchIdx++;
 		}
 	}
 	{
@@ -1526,155 +1355,148 @@ inline int b3GpuPgsContactSolver::sortConstraintByBatch2( b3Contact4* cs, int nu
 	}
 
 	{
-        B3_PROFILE("reorder");
+		B3_PROFILE("reorder");
 		//	reorder
-		
-		memcpy( &m_data->m_old[0], cs, sizeof(b3Contact4)*numConstraints);
 
-		for(int i=0; i<numConstraints; i++)
+		memcpy(&m_data->m_old[0], cs, sizeof(b3Contact4) * numConstraints);
+
+		for (int i = 0; i < numConstraints; i++)
 		{
 			b3Assert(m_data->m_sortData[idxSrc[i]].m_value == idxSrc[i]);
 			int idx = m_data->m_sortData[idxSrc[i]].m_value;
 			cs[i] = m_data->m_old[idx];
 		}
 	}
-	
+
 #if defined(_DEBUG)
-    //		debugPrintf( "nBatches: %d\n", batchIdx );
-	for(int i=0; i<numConstraints; i++)
-    {
-        b3Assert( cs[i].getBatchIdx() != -1 );
-    }
+	//		debugPrintf( "nBatches: %d\n", batchIdx );
+	for (int i = 0; i < numConstraints; i++)
+	{
+		b3Assert(cs[i].getBatchIdx() != -1);
+	}
 #endif
 
-	
 	return batchIdx;
 }
 
-
 b3AlignedObjectArray<int> bodyUsed;
 b3AlignedObjectArray<int> curUsed;
 
-
-inline int b3GpuPgsContactSolver::sortConstraintByBatch3( b3Contact4* cs, int numConstraints, int simdWidth , int staticIdx, int numBodies, int* batchSizes)
+inline int b3GpuPgsContactSolver::sortConstraintByBatch3(b3Contact4* cs, int numConstraints, int simdWidth, int staticIdx, int numBodies, int* batchSizes)
 {
-	
 	B3_PROFILE("sortConstraintByBatch3");
-	
+
 	static int maxSwaps = 0;
 	int numSwaps = 0;
 
-	curUsed.resize(2*simdWidth);
+	curUsed.resize(2 * simdWidth);
 
 	static int maxNumConstraints = 0;
-	if (maxNumConstraints<numConstraints)
+	if (maxNumConstraints < numConstraints)
 	{
 		maxNumConstraints = numConstraints;
 		//printf("maxNumConstraints  = %d\n",maxNumConstraints );
 	}
 
-	int numUsedArray = numBodies/32+1;
+	int numUsedArray = numBodies / 32 + 1;
 	bodyUsed.resize(numUsedArray);
 
-	for (int q=0;q<numUsedArray;q++)
-		bodyUsed[q]=0;
+	for (int q = 0; q < numUsedArray; q++)
+		bodyUsed[q] = 0;
 
-	
 	int curBodyUsed = 0;
 
 	int numIter = 0;
-    
+
 	m_data->m_sortData.resize(0);
 	m_data->m_idxBuffer.resize(0);
 	m_data->m_old.resize(0);
-	
-		
+
 #if defined(_DEBUG)
-	for(int i=0; i<numConstraints; i++)
+	for (int i = 0; i < numConstraints; i++)
 		cs[i].getBatchIdx() = -1;
 #endif
-	
+
 	int numValidConstraints = 0;
-//	int unprocessedConstraintIndex = 0;
+	//	int unprocessedConstraintIndex = 0;
 
 	int batchIdx = 0;
-    
 
 	{
 		B3_PROFILE("cpu batch innerloop");
-		
-		while( numValidConstraints < numConstraints)
+
+		while (numValidConstraints < numConstraints)
 		{
 			numIter++;
 			int nCurrentBatch = 0;
 			batchSizes[batchIdx] = 0;
 
 			//	clear flag
-			for(int i=0; i<curBodyUsed; i++) 
-				bodyUsed[curUsed[i]/32] = 0;
+			for (int i = 0; i < curBodyUsed; i++)
+				bodyUsed[curUsed[i] / 32] = 0;
 
-            curBodyUsed = 0;
+			curBodyUsed = 0;
 
-			for(int i=numValidConstraints; i<numConstraints; i++)
+			for (int i = numValidConstraints; i < numConstraints; i++)
 			{
 				int idx = i;
-				b3Assert( idx < numConstraints );
+				b3Assert(idx < numConstraints);
 				//	check if it can go
 				int bodyAS = cs[idx].m_bodyAPtrAndSignBit;
 				int bodyBS = cs[idx].m_bodyBPtrAndSignBit;
 				int bodyA = abs(bodyAS);
 				int bodyB = abs(bodyBS);
-				bool aIsStatic = (bodyAS<0) || bodyAS==staticIdx;
-				bool bIsStatic = (bodyBS<0) || bodyBS==staticIdx;
+				bool aIsStatic = (bodyAS < 0) || bodyAS == staticIdx;
+				bool bIsStatic = (bodyBS < 0) || bodyBS == staticIdx;
 				int aUnavailable = 0;
 				int bUnavailable = 0;
 				if (!aIsStatic)
 				{
-					aUnavailable = bodyUsed[ bodyA/32 ] & (1<<(bodyA&31));
+					aUnavailable = bodyUsed[bodyA / 32] & (1 << (bodyA & 31));
 				}
 				if (!aUnavailable)
-				if (!bIsStatic)
-				{
-					bUnavailable = bodyUsed[ bodyB/32 ] & (1<<(bodyB&31));
-				}
-                
-				if( aUnavailable==0 && bUnavailable==0 ) // ok
+					if (!bIsStatic)
+					{
+						bUnavailable = bodyUsed[bodyB / 32] & (1 << (bodyB & 31));
+					}
+
+				if (aUnavailable == 0 && bUnavailable == 0)  // ok
 				{
 					if (!aIsStatic)
 					{
-						bodyUsed[ bodyA/32 ] |= (1<<(bodyA&31));
-						curUsed[curBodyUsed++]=bodyA;
+						bodyUsed[bodyA / 32] |= (1 << (bodyA & 31));
+						curUsed[curBodyUsed++] = bodyA;
 					}
 					if (!bIsStatic)
 					{
-						bodyUsed[ bodyB/32 ] |= (1<<(bodyB&31));
-						curUsed[curBodyUsed++]=bodyB;
+						bodyUsed[bodyB / 32] |= (1 << (bodyB & 31));
+						curUsed[curBodyUsed++] = bodyB;
 					}
 
 					cs[idx].getBatchIdx() = batchIdx;
 
-					if (i!=numValidConstraints)
+					if (i != numValidConstraints)
 					{
-						b3Swap(cs[i],cs[numValidConstraints]);
+						b3Swap(cs[i], cs[numValidConstraints]);
 						numSwaps++;
 					}
 
 					numValidConstraints++;
 					{
 						nCurrentBatch++;
-						if( nCurrentBatch == simdWidth )
+						if (nCurrentBatch == simdWidth)
 						{
 							batchSizes[batchIdx] += simdWidth;
 							nCurrentBatch = 0;
-							for(int i=0; i<curBodyUsed; i++) 
-								bodyUsed[curUsed[i]/32] = 0;
+							for (int i = 0; i < curBodyUsed; i++)
+								bodyUsed[curUsed[i] / 32] = 0;
 							curBodyUsed = 0;
 						}
 					}
 				}
 			}
 
-			if (batchIdx>=B3_MAX_NUM_BATCHES)
+			if (batchIdx >= B3_MAX_NUM_BATCHES)
 			{
 				b3Error("batchIdx>=B3_MAX_NUM_BATCHES");
 				b3Assert(0);
@@ -1683,26 +1505,25 @@ inline int b3GpuPgsContactSolver::sortConstraintByBatch3( b3Contact4* cs, int nu
 
 			batchSizes[batchIdx] += nCurrentBatch;
 
-			batchIdx ++;
-			
+			batchIdx++;
 		}
 	}
-	
+
 #if defined(_DEBUG)
-    //		debugPrintf( "nBatches: %d\n", batchIdx );
-	for(int i=0; i<numConstraints; i++)
-    {
-        b3Assert( cs[i].getBatchIdx() != -1 );
-    }
+	//		debugPrintf( "nBatches: %d\n", batchIdx );
+	for (int i = 0; i < numConstraints; i++)
+	{
+		b3Assert(cs[i].getBatchIdx() != -1);
+	}
 #endif
 
-	batchSizes[batchIdx] =0;
-	
-	if (maxSwaps<numSwaps)
+	batchSizes[batchIdx] = 0;
+
+	if (maxSwaps < numSwaps)
 	{
 		maxSwaps = numSwaps;
 		//printf("maxSwaps = %d\n", maxSwaps);
 	}
-	
+
 	return batchIdx;
 }
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuPgsContactSolver.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuPgsContactSolver.h
index 98e2a5b8c46..6ab7502af3d 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuPgsContactSolver.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuPgsContactSolver.h
@@ -11,33 +11,27 @@
 class b3GpuPgsContactSolver
 {
 protected:
-
 	int m_debugOutput;
 
-	struct b3GpuBatchingPgsSolverInternalData*		m_data;
+	struct b3GpuBatchingPgsSolverInternalData* m_data;
 
-	void batchContacts( b3OpenCLArray<b3Contact4>* contacts, int nContacts, b3OpenCLArray<unsigned int>* n, b3OpenCLArray<unsigned int>* offsets, int staticIdx );
-	
-	inline int sortConstraintByBatch( b3Contact4* cs, int n, int simdWidth , int staticIdx, int numBodies);
-	inline int sortConstraintByBatch2( b3Contact4* cs, int n, int simdWidth , int staticIdx, int numBodies);
-	inline int sortConstraintByBatch3( b3Contact4* cs, int n, int simdWidth , int staticIdx, int numBodies, int* batchSizes);
-	
+	void batchContacts(b3OpenCLArray<b3Contact4>* contacts, int nContacts, b3OpenCLArray<unsigned int>* n, b3OpenCLArray<unsigned int>* offsets, int staticIdx);
 
-	
-	void solveContactConstraintBatchSizes(  const b3OpenCLArray<b3RigidBodyData>* bodyBuf, const b3OpenCLArray<b3InertiaData>* shapeBuf, 
-			b3OpenCLArray<b3GpuConstraint4>* constraint, void* additionalData, int n ,int maxNumBatches, int numIterations, const b3AlignedObjectArray<int>* batchSizes);//const b3OpenCLArray<int>* gpuBatchSizes);
+	inline int sortConstraintByBatch(b3Contact4* cs, int n, int simdWidth, int staticIdx, int numBodies);
+	inline int sortConstraintByBatch2(b3Contact4* cs, int n, int simdWidth, int staticIdx, int numBodies);
+	inline int sortConstraintByBatch3(b3Contact4* cs, int n, int simdWidth, int staticIdx, int numBodies, int* batchSizes);
 
-		void solveContactConstraint(  const b3OpenCLArray<b3RigidBodyData>* bodyBuf, const b3OpenCLArray<b3InertiaData>* shapeBuf, 
-			b3OpenCLArray<b3GpuConstraint4>* constraint, void* additionalData, int n ,int maxNumBatches, int numIterations, const b3AlignedObjectArray<int>* batchSizes);//const b3OpenCLArray<int>* gpuBatchSizes);
+	void solveContactConstraintBatchSizes(const b3OpenCLArray<b3RigidBodyData>* bodyBuf, const b3OpenCLArray<b3InertiaData>* shapeBuf,
+										  b3OpenCLArray<b3GpuConstraint4>* constraint, void* additionalData, int n, int maxNumBatches, int numIterations, const b3AlignedObjectArray<int>* batchSizes);  //const b3OpenCLArray<int>* gpuBatchSizes);
+
+	void solveContactConstraint(const b3OpenCLArray<b3RigidBodyData>* bodyBuf, const b3OpenCLArray<b3InertiaData>* shapeBuf,
+								b3OpenCLArray<b3GpuConstraint4>* constraint, void* additionalData, int n, int maxNumBatches, int numIterations, const b3AlignedObjectArray<int>* batchSizes);  //const b3OpenCLArray<int>* gpuBatchSizes);
 
 public:
-	
-	b3GpuPgsContactSolver(cl_context ctx,cl_device_id device, cl_command_queue  q,int pairCapacity);
+	b3GpuPgsContactSolver(cl_context ctx, cl_device_id device, cl_command_queue q, int pairCapacity);
 	virtual ~b3GpuPgsContactSolver();
 
 	void solveContacts(int numBodies, cl_mem bodyBuf, cl_mem inertiaBuf, int numContacts, cl_mem contactBuf, const struct b3Config& config, int static0Index);
-
 };
 
-#endif //B3_GPU_BATCHING_PGS_SOLVER_H
-
+#endif  //B3_GPU_BATCHING_PGS_SOLVER_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuRigidBodyPipeline.cpp b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuRigidBodyPipeline.cpp
index 783e4430604..fef33ad1cdb 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuRigidBodyPipeline.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuRigidBodyPipeline.cpp
@@ -47,7 +47,7 @@ bool gClearPairsOnGpu = true;
 #define TEST_OTHER_GPU_SOLVER 1
 #ifdef TEST_OTHER_GPU_SOLVER
 #include "b3GpuJacobiContactSolver.h"
-#endif //TEST_OTHER_GPU_SOLVER
+#endif  //TEST_OTHER_GPU_SOLVER
 
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3RigidBodyData.h"
 #include "Bullet3Collision/NarrowPhaseCollision/b3Contact4.h"
@@ -59,73 +59,68 @@ bool gClearPairsOnGpu = true;
 #include "Bullet3Collision/NarrowPhaseCollision/b3Config.h"
 #include "Bullet3OpenCL/Raycast/b3GpuRaycast.h"
 
-	
 #include "Bullet3Dynamics/shared/b3IntegrateTransforms.h"
 #include "Bullet3OpenCL/RigidBody/b3GpuNarrowPhaseInternalData.h"
 
-b3GpuRigidBodyPipeline::b3GpuRigidBodyPipeline(cl_context ctx,cl_device_id device, cl_command_queue  q,class b3GpuNarrowPhase* narrowphase, class b3GpuBroadphaseInterface* broadphaseSap , struct b3DynamicBvhBroadphase* broadphaseDbvt, const b3Config& config)
+b3GpuRigidBodyPipeline::b3GpuRigidBodyPipeline(cl_context ctx, cl_device_id device, cl_command_queue q, class b3GpuNarrowPhase* narrowphase, class b3GpuBroadphaseInterface* broadphaseSap, struct b3DynamicBvhBroadphase* broadphaseDbvt, const b3Config& config)
 {
 	m_data = new b3GpuRigidBodyPipelineInternalData;
-	m_data->m_constraintUid=0;
+	m_data->m_constraintUid = 0;
 	m_data->m_config = config;
 	m_data->m_context = ctx;
 	m_data->m_device = device;
 	m_data->m_queue = q;
 
-	m_data->m_solver = new b3PgsJacobiSolver(true);//new b3PgsJacobiSolver(true);
-	m_data->m_gpuSolver = new b3GpuPgsConstraintSolver(ctx,device,q,true);//new b3PgsJacobiSolver(true);
-	
-	m_data->m_allAabbsGPU = new b3OpenCLArray<b3SapAabb>(ctx,q,config.m_maxConvexBodies);
-	m_data->m_overlappingPairsGPU = new b3OpenCLArray<b3BroadphasePair>(ctx,q,config.m_maxBroadphasePairs);
+	m_data->m_solver = new b3PgsJacobiSolver(true);                            //new b3PgsJacobiSolver(true);
+	m_data->m_gpuSolver = new b3GpuPgsConstraintSolver(ctx, device, q, true);  //new b3PgsJacobiSolver(true);
 
-	m_data->m_gpuConstraints = new b3OpenCLArray<b3GpuGenericConstraint>(ctx,q);
+	m_data->m_allAabbsGPU = new b3OpenCLArray<b3SapAabb>(ctx, q, config.m_maxConvexBodies);
+	m_data->m_overlappingPairsGPU = new b3OpenCLArray<b3BroadphasePair>(ctx, q, config.m_maxBroadphasePairs);
+
+	m_data->m_gpuConstraints = new b3OpenCLArray<b3GpuGenericConstraint>(ctx, q);
 #ifdef TEST_OTHER_GPU_SOLVER
-	m_data->m_solver3 = new b3GpuJacobiContactSolver(ctx,device,q,config.m_maxBroadphasePairs);	
-#endif //	TEST_OTHER_GPU_SOLVER
-	
-	m_data->m_solver2 = new b3GpuPgsContactSolver(ctx,device,q,config.m_maxBroadphasePairs);
+	m_data->m_solver3 = new b3GpuJacobiContactSolver(ctx, device, q, config.m_maxBroadphasePairs);
+#endif  //	TEST_OTHER_GPU_SOLVER
 
-	m_data->m_raycaster = new b3GpuRaycast(ctx,device,q);
+	m_data->m_solver2 = new b3GpuPgsContactSolver(ctx, device, q, config.m_maxBroadphasePairs);
+
+	m_data->m_raycaster = new b3GpuRaycast(ctx, device, q);
 
-	
 	m_data->m_broadphaseDbvt = broadphaseDbvt;
 	m_data->m_broadphaseSap = broadphaseSap;
 	m_data->m_narrowphase = narrowphase;
-	m_data->m_gravity.setValue(0.f,-9.8f,0.f);
+	m_data->m_gravity.setValue(0.f, -9.8f, 0.f);
 
-	cl_int errNum=0;
+	cl_int errNum = 0;
 
 	{
-		cl_program prog = b3OpenCLUtils::compileCLProgramFromString(m_data->m_context,m_data->m_device,integrateKernelCL,&errNum,"",B3_RIGIDBODY_INTEGRATE_PATH);
-		b3Assert(errNum==CL_SUCCESS);
-		m_data->m_integrateTransformsKernel = b3OpenCLUtils::compileCLKernelFromString(m_data->m_context, m_data->m_device,integrateKernelCL, "integrateTransformsKernel",&errNum,prog);
-		b3Assert(errNum==CL_SUCCESS);
+		cl_program prog = b3OpenCLUtils::compileCLProgramFromString(m_data->m_context, m_data->m_device, integrateKernelCL, &errNum, "", B3_RIGIDBODY_INTEGRATE_PATH);
+		b3Assert(errNum == CL_SUCCESS);
+		m_data->m_integrateTransformsKernel = b3OpenCLUtils::compileCLKernelFromString(m_data->m_context, m_data->m_device, integrateKernelCL, "integrateTransformsKernel", &errNum, prog);
+		b3Assert(errNum == CL_SUCCESS);
 		clReleaseProgram(prog);
 	}
 	{
-		cl_program prog = b3OpenCLUtils::compileCLProgramFromString(m_data->m_context,m_data->m_device,updateAabbsKernelCL,&errNum,"",B3_RIGIDBODY_UPDATEAABB_PATH);
-		b3Assert(errNum==CL_SUCCESS);
-		m_data->m_updateAabbsKernel = b3OpenCLUtils::compileCLKernelFromString(m_data->m_context, m_data->m_device,updateAabbsKernelCL, "initializeGpuAabbsFull",&errNum,prog);
-		b3Assert(errNum==CL_SUCCESS);
+		cl_program prog = b3OpenCLUtils::compileCLProgramFromString(m_data->m_context, m_data->m_device, updateAabbsKernelCL, &errNum, "", B3_RIGIDBODY_UPDATEAABB_PATH);
+		b3Assert(errNum == CL_SUCCESS);
+		m_data->m_updateAabbsKernel = b3OpenCLUtils::compileCLKernelFromString(m_data->m_context, m_data->m_device, updateAabbsKernelCL, "initializeGpuAabbsFull", &errNum, prog);
+		b3Assert(errNum == CL_SUCCESS);
 
-
-		m_data->m_clearOverlappingPairsKernel = b3OpenCLUtils::compileCLKernelFromString(m_data->m_context, m_data->m_device,updateAabbsKernelCL, "clearOverlappingPairsKernel",&errNum,prog);
-		b3Assert(errNum==CL_SUCCESS);
+		m_data->m_clearOverlappingPairsKernel = b3OpenCLUtils::compileCLKernelFromString(m_data->m_context, m_data->m_device, updateAabbsKernelCL, "clearOverlappingPairsKernel", &errNum, prog);
+		b3Assert(errNum == CL_SUCCESS);
 
 		clReleaseProgram(prog);
 	}
-
-
 }
 
 b3GpuRigidBodyPipeline::~b3GpuRigidBodyPipeline()
 {
 	if (m_data->m_integrateTransformsKernel)
 		clReleaseKernel(m_data->m_integrateTransformsKernel);
-	
+
 	if (m_data->m_updateAabbsKernel)
 		clReleaseKernel(m_data->m_updateAabbsKernel);
-	
+
 	if (m_data->m_clearOverlappingPairsKernel)
 		clReleaseKernel(m_data->m_clearOverlappingPairsKernel);
 	delete m_data->m_raycaster;
@@ -136,15 +131,14 @@ b3GpuRigidBodyPipeline::~b3GpuRigidBodyPipeline()
 
 #ifdef TEST_OTHER_GPU_SOLVER
 	delete m_data->m_solver3;
-#endif //TEST_OTHER_GPU_SOLVER
-	
+#endif  //TEST_OTHER_GPU_SOLVER
+
 	delete m_data->m_solver2;
-	
-	
+
 	delete m_data;
 }
 
-void	b3GpuRigidBodyPipeline::reset()
+void b3GpuRigidBodyPipeline::reset()
 {
 	m_data->m_gpuConstraints->resize(0);
 	m_data->m_cpuConstraints.resize(0);
@@ -152,30 +146,28 @@ void	b3GpuRigidBodyPipeline::reset()
 	m_data->m_allAabbsCPU.resize(0);
 }
 
-void	b3GpuRigidBodyPipeline::addConstraint(b3TypedConstraint* constraint)
+void b3GpuRigidBodyPipeline::addConstraint(b3TypedConstraint* constraint)
 {
 	m_data->m_joints.push_back(constraint);
 }
 
-void	b3GpuRigidBodyPipeline::removeConstraint(b3TypedConstraint* constraint)
+void b3GpuRigidBodyPipeline::removeConstraint(b3TypedConstraint* constraint)
 {
 	m_data->m_joints.remove(constraint);
 }
 
-
-
-void  b3GpuRigidBodyPipeline::removeConstraintByUid(int uid)
+void b3GpuRigidBodyPipeline::removeConstraintByUid(int uid)
 {
 	m_data->m_gpuSolver->recomputeBatches();
 	//slow linear search
 	m_data->m_gpuConstraints->copyToHost(m_data->m_cpuConstraints);
 	//remove
-	for (int i=0;i<m_data->m_cpuConstraints.size();i++)
+	for (int i = 0; i < m_data->m_cpuConstraints.size(); i++)
 	{
 		if (m_data->m_cpuConstraints[i].m_uid == uid)
 		{
 			//m_data->m_cpuConstraints.remove(m_data->m_cpuConstraints[i]);
-			m_data->m_cpuConstraints.swap(i,m_data->m_cpuConstraints.size()-1);
+			m_data->m_cpuConstraints.swap(i, m_data->m_cpuConstraints.size() - 1);
 			m_data->m_cpuConstraints.pop_back();
 
 			break;
@@ -185,13 +177,13 @@ void  b3GpuRigidBodyPipeline::removeConstraintByUid(int uid)
 	if (m_data->m_cpuConstraints.size())
 	{
 		m_data->m_gpuConstraints->copyFromHost(m_data->m_cpuConstraints);
-	} else
+	}
+	else
 	{
 		m_data->m_gpuConstraints->resize(0);
 	}
-
 }
-int b3GpuRigidBodyPipeline::createPoint2PointConstraint(int bodyA, int bodyB, const float* pivotInA, const float* pivotInB,float breakingThreshold)
+int b3GpuRigidBodyPipeline::createPoint2PointConstraint(int bodyA, int bodyB, const float* pivotInA, const float* pivotInB, float breakingThreshold)
 {
 	m_data->m_gpuSolver->recomputeBatches();
 	b3GpuGenericConstraint c;
@@ -200,14 +192,14 @@ int b3GpuRigidBodyPipeline::createPoint2PointConstraint(int bodyA, int bodyB, co
 	c.m_flags = B3_CONSTRAINT_FLAG_ENABLED;
 	c.m_rbA = bodyA;
 	c.m_rbB = bodyB;
-	c.m_pivotInA.setValue(pivotInA[0],pivotInA[1],pivotInA[2]);
-	c.m_pivotInB.setValue(pivotInB[0],pivotInB[1],pivotInB[2]);
+	c.m_pivotInA.setValue(pivotInA[0], pivotInA[1], pivotInA[2]);
+	c.m_pivotInB.setValue(pivotInB[0], pivotInB[1], pivotInB[2]);
 	c.m_breakingImpulseThreshold = breakingThreshold;
 	c.m_constraintType = B3_GPU_POINT2POINT_CONSTRAINT_TYPE;
 	m_data->m_cpuConstraints.push_back(c);
 	return c.m_uid;
 }
-int b3GpuRigidBodyPipeline::createFixedConstraint(int bodyA, int bodyB, const float* pivotInA, const float* pivotInB, const float* relTargetAB,float breakingThreshold)
+int b3GpuRigidBodyPipeline::createFixedConstraint(int bodyA, int bodyB, const float* pivotInA, const float* pivotInB, const float* relTargetAB, float breakingThreshold)
 {
 	m_data->m_gpuSolver->recomputeBatches();
 	b3GpuGenericConstraint c;
@@ -216,9 +208,9 @@ int b3GpuRigidBodyPipeline::createFixedConstraint(int bodyA, int bodyB, const fl
 	c.m_flags = B3_CONSTRAINT_FLAG_ENABLED;
 	c.m_rbA = bodyA;
 	c.m_rbB = bodyB;
-	c.m_pivotInA.setValue(pivotInA[0],pivotInA[1],pivotInA[2]);
-	c.m_pivotInB.setValue(pivotInB[0],pivotInB[1],pivotInB[2]);
-	c.m_relTargetAB.setValue(relTargetAB[0],relTargetAB[1],relTargetAB[2],relTargetAB[3]);
+	c.m_pivotInA.setValue(pivotInA[0], pivotInA[1], pivotInA[2]);
+	c.m_pivotInB.setValue(pivotInB[0], pivotInB[1], pivotInB[2]);
+	c.m_relTargetAB.setValue(relTargetAB[0], relTargetAB[1], relTargetAB[2], relTargetAB[3]);
 	c.m_breakingImpulseThreshold = breakingThreshold;
 	c.m_constraintType = B3_GPU_FIXED_CONSTRAINT_TYPE;
 
@@ -226,31 +218,28 @@ int b3GpuRigidBodyPipeline::createFixedConstraint(int bodyA, int bodyB, const fl
 	return c.m_uid;
 }
 
-
-void	b3GpuRigidBodyPipeline::stepSimulation(float deltaTime)
+void b3GpuRigidBodyPipeline::stepSimulation(float deltaTime)
 {
-
 	//update worldspace AABBs from local AABB/worldtransform
 	{
 		B3_PROFILE("setupGpuAabbs");
 		setupGpuAabbsFull();
 	}
 
-	int numPairs =0;
+	int numPairs = 0;
 
 	//compute overlapping pairs
 	{
-
 		if (gUseDbvt)
 		{
 			{
 				B3_PROFILE("setAabb");
 				m_data->m_allAabbsGPU->copyToHost(m_data->m_allAabbsCPU);
-				for (int i=0;i<m_data->m_allAabbsCPU.size();i++)
+				for (int i = 0; i < m_data->m_allAabbsCPU.size(); i++)
 				{
-					b3Vector3 aabbMin=b3MakeVector3(m_data->m_allAabbsCPU[i].m_min[0],m_data->m_allAabbsCPU[i].m_min[1],m_data->m_allAabbsCPU[i].m_min[2]);
-					b3Vector3 aabbMax=b3MakeVector3(m_data->m_allAabbsCPU[i].m_max[0],m_data->m_allAabbsCPU[i].m_max[1],m_data->m_allAabbsCPU[i].m_max[2]);
-					m_data->m_broadphaseDbvt->setAabb(i,aabbMin,aabbMax,0);
+					b3Vector3 aabbMin = b3MakeVector3(m_data->m_allAabbsCPU[i].m_min[0], m_data->m_allAabbsCPU[i].m_min[1], m_data->m_allAabbsCPU[i].m_min[2]);
+					b3Vector3 aabbMax = b3MakeVector3(m_data->m_allAabbsCPU[i].m_max[0], m_data->m_allAabbsCPU[i].m_max[1], m_data->m_allAabbsCPU[i].m_max[2]);
+					m_data->m_broadphaseDbvt->setAabb(i, aabbMin, aabbMax, 0);
 				}
 			}
 
@@ -259,13 +248,14 @@ void	b3GpuRigidBodyPipeline::stepSimulation(float deltaTime)
 				m_data->m_broadphaseDbvt->calculateOverlappingPairs();
 			}
 			numPairs = m_data->m_broadphaseDbvt->getOverlappingPairCache()->getNumOverlappingPairs();
-
-		} else
+		}
+		else
 		{
 			if (gUseCalculateOverlappingPairsHost)
 			{
 				m_data->m_broadphaseSap->calculateOverlappingPairsHost(m_data->m_config.m_maxBroadphasePairs);
-			} else
+			}
+			else
 			{
 				m_data->m_broadphaseSap->calculateOverlappingPairs(m_data->m_config.m_maxBroadphasePairs);
 			}
@@ -274,24 +264,24 @@ void	b3GpuRigidBodyPipeline::stepSimulation(float deltaTime)
 	}
 
 	//compute contact points
-//	printf("numPairs=%d\n",numPairs);
-	
-	int numContacts  = 0;
+	//	printf("numPairs=%d\n",numPairs);
 
+	int numContacts = 0;
 
 	int numBodies = m_data->m_narrowphase->getNumRigidBodies();
 
 	if (numPairs)
 	{
-		cl_mem pairs =0;
-		cl_mem aabbsWS =0;
+		cl_mem pairs = 0;
+		cl_mem aabbsWS = 0;
 		if (gUseDbvt)
 		{
 			B3_PROFILE("m_overlappingPairsGPU->copyFromHost");
 			m_data->m_overlappingPairsGPU->copyFromHost(m_data->m_broadphaseDbvt->getOverlappingPairCache()->getOverlappingPairArray());
 			pairs = m_data->m_overlappingPairsGPU->getBufferCL();
 			aabbsWS = m_data->m_allAabbsGPU->getBufferCL();
-		} else
+		}
+		else
 		{
 			pairs = m_data->m_broadphaseSap->getOverlappingPairBuffer();
 			aabbsWS = m_data->m_broadphaseSap->getAabbBufferWS();
@@ -302,31 +292,27 @@ void	b3GpuRigidBodyPipeline::stepSimulation(float deltaTime)
 		//mark the contacts for each pair as 'unused'
 		if (numPairs)
 		{
-			b3OpenCLArray<b3BroadphasePair> gpuPairs(this->m_data->m_context,m_data->m_queue);
-			gpuPairs.setFromOpenCLBuffer(pairs,numPairs);
+			b3OpenCLArray<b3BroadphasePair> gpuPairs(this->m_data->m_context, m_data->m_queue);
+			gpuPairs.setFromOpenCLBuffer(pairs, numPairs);
 
 			if (gClearPairsOnGpu)
 			{
-				
-
 				//b3AlignedObjectArray<b3BroadphasePair> hostPairs;//just for debugging
 				//gpuPairs.copyToHost(hostPairs);
 
-				b3LauncherCL launcher(m_data->m_queue,m_data->m_clearOverlappingPairsKernel,"clearOverlappingPairsKernel");
+				b3LauncherCL launcher(m_data->m_queue, m_data->m_clearOverlappingPairsKernel, "clearOverlappingPairsKernel");
 				launcher.setBuffer(pairs);
 				launcher.setConst(numPairs);
 				launcher.launch1D(numPairs);
 
-
 				//gpuPairs.copyToHost(hostPairs);
-			
-
-			} else
+			}
+			else
 			{
 				b3AlignedObjectArray<b3BroadphasePair> hostPairs;
 				gpuPairs.copyToHost(hostPairs);
 
-				for (int i=0;i<hostPairs.size();i++)
+				for (int i = 0; i < hostPairs.size(); i++)
 				{
 					hostPairs[i].z = 0xffffffff;
 				}
@@ -335,7 +321,7 @@ void	b3GpuRigidBodyPipeline::stepSimulation(float deltaTime)
 			}
 		}
 
-		m_data->m_narrowphase->computeContacts(pairs,numPairs,aabbsWS,numBodies);
+		m_data->m_narrowphase->computeContacts(pairs, numPairs, aabbsWS, numBodies);
 		numContacts = m_data->m_narrowphase->getNumContactsGpu();
 
 		if (gUseDbvt)
@@ -347,56 +333,54 @@ void	b3GpuRigidBodyPipeline::stepSimulation(float deltaTime)
 		if (gDumpContactStats && numContacts)
 		{
 			m_data->m_narrowphase->getContactsGpu();
-			
+
 			printf("numContacts = %d\n", numContacts);
 
-			int totalPoints  = 0;
+			int totalPoints = 0;
 			const b3Contact4* contacts = m_data->m_narrowphase->getContactsCPU();
 
-			for (int i=0;i<numContacts;i++)
+			for (int i = 0; i < numContacts; i++)
 			{
 				totalPoints += contacts->getNPoints();
 			}
-			printf("totalPoints=%d\n",totalPoints);
-
+			printf("totalPoints=%d\n", totalPoints);
 		}
 	}
-	
 
 	//convert contact points to contact constraints
-	
+
 	//solve constraints
 
-	b3OpenCLArray<b3RigidBodyData> gpuBodies(m_data->m_context,m_data->m_queue,0,true);
-	gpuBodies.setFromOpenCLBuffer(m_data->m_narrowphase->getBodiesGpu(),m_data->m_narrowphase->getNumRigidBodies());
-	b3OpenCLArray<b3InertiaData> gpuInertias(m_data->m_context,m_data->m_queue,0,true);
-	gpuInertias.setFromOpenCLBuffer(m_data->m_narrowphase->getBodyInertiasGpu(),m_data->m_narrowphase->getNumRigidBodies());
-	b3OpenCLArray<b3Contact4> gpuContacts(m_data->m_context,m_data->m_queue,0,true);
-	gpuContacts.setFromOpenCLBuffer(m_data->m_narrowphase->getContactsGpu(),m_data->m_narrowphase->getNumContactsGpu());
+	b3OpenCLArray<b3RigidBodyData> gpuBodies(m_data->m_context, m_data->m_queue, 0, true);
+	gpuBodies.setFromOpenCLBuffer(m_data->m_narrowphase->getBodiesGpu(), m_data->m_narrowphase->getNumRigidBodies());
+	b3OpenCLArray<b3InertiaData> gpuInertias(m_data->m_context, m_data->m_queue, 0, true);
+	gpuInertias.setFromOpenCLBuffer(m_data->m_narrowphase->getBodyInertiasGpu(), m_data->m_narrowphase->getNumRigidBodies());
+	b3OpenCLArray<b3Contact4> gpuContacts(m_data->m_context, m_data->m_queue, 0, true);
+	gpuContacts.setFromOpenCLBuffer(m_data->m_narrowphase->getContactsGpu(), m_data->m_narrowphase->getNumContactsGpu());
 
-	int numJoints =  m_data->m_joints.size() ?  m_data->m_joints.size() : m_data->m_cpuConstraints.size();
+	int numJoints = m_data->m_joints.size() ? m_data->m_joints.size() : m_data->m_cpuConstraints.size();
 	if (useBullet2CpuSolver && numJoints)
 	{
-
-	//	b3AlignedObjectArray<b3Contact4> hostContacts;
+		//	b3AlignedObjectArray<b3Contact4> hostContacts;
 		//gpuContacts.copyToHost(hostContacts);
 		{
-			bool useGpu = m_data->m_joints.size()==0;
+			bool useGpu = m_data->m_joints.size() == 0;
 
-//			b3Contact4* contacts = numContacts? &hostContacts[0]: 0;
+			//			b3Contact4* contacts = numContacts? &hostContacts[0]: 0;
 			//m_data->m_solver->solveContacts(m_data->m_narrowphase->getNumBodiesGpu(),&hostBodies[0],&hostInertias[0],numContacts,contacts,numJoints, joints);
 			if (useGpu)
 			{
-				m_data->m_gpuSolver->solveJoints(m_data->m_narrowphase->getNumRigidBodies(),&gpuBodies,&gpuInertias,numJoints, m_data->m_gpuConstraints);
-			} else
+				m_data->m_gpuSolver->solveJoints(m_data->m_narrowphase->getNumRigidBodies(), &gpuBodies, &gpuInertias, numJoints, m_data->m_gpuConstraints);
+			}
+			else
 			{
 				b3AlignedObjectArray<b3RigidBodyData> hostBodies;
 				gpuBodies.copyToHost(hostBodies);
 				b3AlignedObjectArray<b3InertiaData> hostInertias;
 				gpuInertias.copyToHost(hostInertias);
 
-				b3TypedConstraint** joints = numJoints? &m_data->m_joints[0] : 0;
-				m_data->m_solver->solveContacts(m_data->m_narrowphase->getNumRigidBodies(),&hostBodies[0],&hostInertias[0],0,0,numJoints, joints);
+				b3TypedConstraint** joints = numJoints ? &m_data->m_joints[0] : 0;
+				m_data->m_solver->solveContacts(m_data->m_narrowphase->getNumRigidBodies(), &hostBodies[0], &hostInertias[0], 0, 0, numJoints, joints);
 				gpuBodies.copyFromHost(hostBodies);
 			}
 		}
@@ -404,22 +388,20 @@ void	b3GpuRigidBodyPipeline::stepSimulation(float deltaTime)
 
 	if (numContacts)
 	{
-
 #ifdef TEST_OTHER_GPU_SOLVER
-		
+
 		if (gUseJacobi)
 		{
 			bool useGpu = true;
 			if (useGpu)
 			{
-
 				bool forceHost = false;
 				if (forceHost)
 				{
 					b3AlignedObjectArray<b3RigidBodyData> hostBodies;
 					b3AlignedObjectArray<b3InertiaData> hostInertias;
 					b3AlignedObjectArray<b3Contact4> hostContacts;
-				
+
 					{
 						B3_PROFILE("copyToHost");
 						gpuBodies.copyToHost(hostBodies);
@@ -429,25 +411,24 @@ void	b3GpuRigidBodyPipeline::stepSimulation(float deltaTime)
 
 					{
 						b3JacobiSolverInfo solverInfo;
-						m_data->m_solver3->solveGroupHost(&hostBodies[0], &hostInertias[0], hostBodies.size(),&hostContacts[0],hostContacts.size(),solverInfo);
-
-						
+						m_data->m_solver3->solveGroupHost(&hostBodies[0], &hostInertias[0], hostBodies.size(), &hostContacts[0], hostContacts.size(), solverInfo);
 					}
 					{
 						B3_PROFILE("copyFromHost");
 						gpuBodies.copyFromHost(hostBodies);
 					}
-				} else
-
+				}
+				else
 
 				{
 					int static0Index = m_data->m_narrowphase->getStatic0Index();
 					b3JacobiSolverInfo solverInfo;
 					//m_data->m_solver3->solveContacts(    >solveGroup(&gpuBodies, &gpuInertias, &gpuContacts,solverInfo);
 					//m_data->m_solver3->solveContacts(m_data->m_narrowphase->getNumBodiesGpu(),&hostBodies[0],&hostInertias[0],numContacts,&hostContacts[0]);
-					m_data->m_solver3->solveContacts(numBodies, gpuBodies.getBufferCL(),gpuInertias.getBufferCL(),numContacts, gpuContacts.getBufferCL(),m_data->m_config, static0Index);
+					m_data->m_solver3->solveContacts(numBodies, gpuBodies.getBufferCL(), gpuInertias.getBufferCL(), numContacts, gpuContacts.getBufferCL(), m_data->m_config, static0Index);
 				}
-			} else
+			}
+			else
 			{
 				b3AlignedObjectArray<b3RigidBodyData> hostBodies;
 				gpuBodies.copyToHost(hostBodies);
@@ -460,17 +441,15 @@ void	b3GpuRigidBodyPipeline::stepSimulation(float deltaTime)
 				}
 				gpuBodies.copyFromHost(hostBodies);
 			}
-		
-		} else
-#endif //TEST_OTHER_GPU_SOLVER
+		}
+		else
+#endif  //TEST_OTHER_GPU_SOLVER
 		{
-			
 			int static0Index = m_data->m_narrowphase->getStatic0Index();
-			m_data->m_solver2->solveContacts(numBodies, gpuBodies.getBufferCL(),gpuInertias.getBufferCL(),numContacts, gpuContacts.getBufferCL(),m_data->m_config, static0Index);
-			
+			m_data->m_solver2->solveContacts(numBodies, gpuBodies.getBufferCL(), gpuInertias.getBufferCL(), numContacts, gpuContacts.getBufferCL(), m_data->m_config, static0Index);
+
 			//m_data->m_solver4->solveContacts(m_data->m_narrowphase->getNumBodiesGpu(), gpuBodies.getBufferCL(), gpuInertias.getBufferCL(), numContacts, gpuContacts.getBufferCL());
-			
-			
+
 			/*m_data->m_solver3->solveContactConstraintHost(
 			(b3OpenCLArray<RigidBodyBase::Body>*)&gpuBodies,
 			(b3OpenCLArray<RigidBodyBase::Inertia>*)&gpuInertias,
@@ -481,11 +460,9 @@ void	b3GpuRigidBodyPipeline::stepSimulation(float deltaTime)
 	}
 
 	integrate(deltaTime);
-
 }
 
-
-void	b3GpuRigidBodyPipeline::integrate(float timeStep)
+void b3GpuRigidBodyPipeline::integrate(float timeStep)
 {
 	//integrate
 	int numBodies = m_data->m_narrowphase->getNumRigidBodies();
@@ -493,24 +470,25 @@ void	b3GpuRigidBodyPipeline::integrate(float timeStep)
 
 	if (gIntegrateOnCpu)
 	{
-		if(numBodies)
+		if (numBodies)
 		{
-			b3GpuNarrowPhaseInternalData*	npData = m_data->m_narrowphase->getInternalData();
+			b3GpuNarrowPhaseInternalData* npData = m_data->m_narrowphase->getInternalData();
 			npData->m_bodyBufferGPU->copyToHost(*npData->m_bodyBufferCPU);
 
 			b3RigidBodyData_t* bodies = &npData->m_bodyBufferCPU->at(0);
 
-			for (int nodeID=0;nodeID<numBodies;nodeID++)
+			for (int nodeID = 0; nodeID < numBodies; nodeID++)
 			{
-				integrateSingleTransform( bodies,nodeID, timeStep, angularDamp, m_data->m_gravity);
+				integrateSingleTransform(bodies, nodeID, timeStep, angularDamp, m_data->m_gravity);
 			}
 			npData->m_bodyBufferGPU->copyFromHost(*npData->m_bodyBufferCPU);
 		}
-	} else
+	}
+	else
 	{
-		b3LauncherCL launcher(m_data->m_queue,m_data->m_integrateTransformsKernel,"m_integrateTransformsKernel");
+		b3LauncherCL launcher(m_data->m_queue, m_data->m_integrateTransformsKernel, "m_integrateTransformsKernel");
 		launcher.setBuffer(m_data->m_narrowphase->getBodiesGpu());
-		
+
 		launcher.setConst(numBodies);
 		launcher.setConst(timeStep);
 		launcher.setConst(angularDamp);
@@ -519,12 +497,9 @@ void	b3GpuRigidBodyPipeline::integrate(float timeStep)
 	}
 }
 
-
-
-
-void	b3GpuRigidBodyPipeline::setupGpuAabbsFull()
+void b3GpuRigidBodyPipeline::setupGpuAabbsFull()
 {
-	cl_int ciErrNum=0;
+	cl_int ciErrNum = 0;
 
 	int numBodies = m_data->m_narrowphase->getNumRigidBodies();
 	if (!numBodies)
@@ -532,34 +507,35 @@ void	b3GpuRigidBodyPipeline::setupGpuAabbsFull()
 
 	if (gCalcWorldSpaceAabbOnCpu)
 	{
-		
 		if (numBodies)
 		{
 			if (gUseDbvt)
 			{
 				m_data->m_allAabbsCPU.resize(numBodies);
 				m_data->m_narrowphase->readbackAllBodiesToCpu();
-				for (int i=0;i<numBodies;i++)
+				for (int i = 0; i < numBodies; i++)
 				{
-					b3ComputeWorldAabb(  i, m_data->m_narrowphase->getBodiesCpu(), m_data->m_narrowphase->getCollidablesCpu(), m_data->m_narrowphase->getLocalSpaceAabbsCpu(),&m_data->m_allAabbsCPU[0]);
+					b3ComputeWorldAabb(i, m_data->m_narrowphase->getBodiesCpu(), m_data->m_narrowphase->getCollidablesCpu(), m_data->m_narrowphase->getLocalSpaceAabbsCpu(), &m_data->m_allAabbsCPU[0]);
 				}
 				m_data->m_allAabbsGPU->copyFromHost(m_data->m_allAabbsCPU);
-			} else
+			}
+			else
 			{
 				m_data->m_broadphaseSap->getAllAabbsCPU().resize(numBodies);
 				m_data->m_narrowphase->readbackAllBodiesToCpu();
-				for (int i=0;i<numBodies;i++)
+				for (int i = 0; i < numBodies; i++)
 				{
-					b3ComputeWorldAabb(  i, m_data->m_narrowphase->getBodiesCpu(), m_data->m_narrowphase->getCollidablesCpu(), m_data->m_narrowphase->getLocalSpaceAabbsCpu(),&m_data->m_broadphaseSap->getAllAabbsCPU()[0]);
+					b3ComputeWorldAabb(i, m_data->m_narrowphase->getBodiesCpu(), m_data->m_narrowphase->getCollidablesCpu(), m_data->m_narrowphase->getLocalSpaceAabbsCpu(), &m_data->m_broadphaseSap->getAllAabbsCPU()[0]);
 				}
 				m_data->m_broadphaseSap->getAllAabbsGPU().copyFromHost(m_data->m_broadphaseSap->getAllAabbsCPU());
 				//m_data->m_broadphaseSap->writeAabbsToGpu();
 			}
 		}
-	} else
+	}
+	else
 	{
 		//__kernel void initializeGpuAabbsFull(  const int numNodes, __global Body* gBodies,__global Collidable* collidables, __global b3AABBCL* plocalShapeAABB, __global b3AABBCL* pAABB)
-		b3LauncherCL launcher(m_data->m_queue,m_data->m_updateAabbsKernel,"m_updateAabbsKernel");
+		b3LauncherCL launcher(m_data->m_queue, m_data->m_updateAabbsKernel, "m_updateAabbsKernel");
 		launcher.setConst(numBodies);
 		cl_mem bodies = m_data->m_narrowphase->getBodiesGpu();
 		launcher.setBuffer(bodies);
@@ -568,17 +544,18 @@ void	b3GpuRigidBodyPipeline::setupGpuAabbsFull()
 		cl_mem localAabbs = m_data->m_narrowphase->getAabbLocalSpaceBufferGpu();
 		launcher.setBuffer(localAabbs);
 
-		cl_mem worldAabbs =0;
+		cl_mem worldAabbs = 0;
 		if (gUseDbvt)
 		{
 			worldAabbs = m_data->m_allAabbsGPU->getBufferCL();
-		} else
+		}
+		else
 		{
 			worldAabbs = m_data->m_broadphaseSap->getAabbBufferWS();
 		}
 		launcher.setBuffer(worldAabbs);
 		launcher.launch1D(numBodies);
-	
+
 		oclCHECKERROR(ciErrNum, CL_SUCCESS);
 	}
 
@@ -595,78 +572,68 @@ void	b3GpuRigidBodyPipeline::setupGpuAabbsFull()
 
 	};
 	*/
-
-	
-
-
-
 }
 
-
-
-cl_mem	b3GpuRigidBodyPipeline::getBodyBuffer()
+cl_mem b3GpuRigidBodyPipeline::getBodyBuffer()
 {
 	return m_data->m_narrowphase->getBodiesGpu();
 }
 
-int	b3GpuRigidBodyPipeline::getNumBodies() const
+int b3GpuRigidBodyPipeline::getNumBodies() const
 {
 	return m_data->m_narrowphase->getNumRigidBodies();
 }
 
-void	b3GpuRigidBodyPipeline::setGravity(const float* grav)
+void b3GpuRigidBodyPipeline::setGravity(const float* grav)
 {
-	m_data->m_gravity.setValue(grav[0],grav[1],grav[2]);
+	m_data->m_gravity.setValue(grav[0], grav[1], grav[2]);
 }
 
-void 		b3GpuRigidBodyPipeline::copyConstraintsToHost()
+void b3GpuRigidBodyPipeline::copyConstraintsToHost()
 {
 	m_data->m_gpuConstraints->copyToHost(m_data->m_cpuConstraints);
 }
 
-void 		b3GpuRigidBodyPipeline::writeAllInstancesToGpu()
+void b3GpuRigidBodyPipeline::writeAllInstancesToGpu()
 {
 	m_data->m_allAabbsGPU->copyFromHost(m_data->m_allAabbsCPU);
 	m_data->m_gpuConstraints->copyFromHost(m_data->m_cpuConstraints);
 }
 
-
-int		b3GpuRigidBodyPipeline::registerPhysicsInstance(float mass, const float* position, const float* orientation, int collidableIndex, int userIndex, bool writeInstanceToGpu)
+int b3GpuRigidBodyPipeline::registerPhysicsInstance(float mass, const float* position, const float* orientation, int collidableIndex, int userIndex, bool writeInstanceToGpu)
 {
-	
-	b3Vector3 aabbMin=b3MakeVector3(0,0,0),aabbMax=b3MakeVector3(0,0,0);
+	b3Vector3 aabbMin = b3MakeVector3(0, 0, 0), aabbMax = b3MakeVector3(0, 0, 0);
 
-	
-	if (collidableIndex>=0)
+	if (collidableIndex >= 0)
 	{
 		b3SapAabb localAabb = m_data->m_narrowphase->getLocalSpaceAabb(collidableIndex);
-		b3Vector3 localAabbMin=b3MakeVector3(localAabb.m_min[0],localAabb.m_min[1],localAabb.m_min[2]);
-		b3Vector3 localAabbMax=b3MakeVector3(localAabb.m_max[0],localAabb.m_max[1],localAabb.m_max[2]);
-		
+		b3Vector3 localAabbMin = b3MakeVector3(localAabb.m_min[0], localAabb.m_min[1], localAabb.m_min[2]);
+		b3Vector3 localAabbMax = b3MakeVector3(localAabb.m_max[0], localAabb.m_max[1], localAabb.m_max[2]);
+
 		b3Scalar margin = 0.01f;
 		b3Transform t;
 		t.setIdentity();
-		t.setOrigin(b3MakeVector3(position[0],position[1],position[2]));
-		t.setRotation(b3Quaternion(orientation[0],orientation[1],orientation[2],orientation[3]));
-		b3TransformAabb(localAabbMin,localAabbMax, margin,t,aabbMin,aabbMax);
-	} else
+		t.setOrigin(b3MakeVector3(position[0], position[1], position[2]));
+		t.setRotation(b3Quaternion(orientation[0], orientation[1], orientation[2], orientation[3]));
+		b3TransformAabb(localAabbMin, localAabbMax, margin, t, aabbMin, aabbMax);
+	}
+	else
 	{
 		b3Error("registerPhysicsInstance using invalid collidableIndex\n");
 		return -1;
 	}
-			
-	
+
 	bool writeToGpu = false;
 	int bodyIndex = m_data->m_narrowphase->getNumRigidBodies();
-	bodyIndex = m_data->m_narrowphase->registerRigidBody(collidableIndex,mass,position,orientation,&aabbMin.getX(),&aabbMax.getX(),writeToGpu);
+	bodyIndex = m_data->m_narrowphase->registerRigidBody(collidableIndex, mass, position, orientation, &aabbMin.getX(), &aabbMax.getX(), writeToGpu);
 
-	if (bodyIndex>=0)
+	if (bodyIndex >= 0)
 	{
 		if (gUseDbvt)
 		{
-			m_data->m_broadphaseDbvt->createProxy(aabbMin,aabbMax,bodyIndex,0,1,1);
+			m_data->m_broadphaseDbvt->createProxy(aabbMin, aabbMax, bodyIndex, 0, 1, 1);
 			b3SapAabb aabb;
-			for (int i=0;i<3;i++)
+			for (int i = 0; i < 3; i++)
 			{
 				aabb.m_min[i] = aabbMin[i];
 				aabb.m_max[i] = aabbMax[i];
@@ -677,14 +644,16 @@ int		b3GpuRigidBodyPipeline::registerPhysicsInstance(float mass, const float* po
 			{
 				m_data->m_allAabbsGPU->copyFromHost(m_data->m_allAabbsCPU);
 			}
-		} else
+		}
+		else
 		{
 			if (mass)
 			{
-				m_data->m_broadphaseSap->createProxy(aabbMin,aabbMax,bodyIndex,1,1);//m_dispatcher);
-			} else
+				m_data->m_broadphaseSap->createProxy(aabbMin, aabbMax, bodyIndex, 1, 1);  //m_dispatcher);
+			}
+			else
 			{
-				m_data->m_broadphaseSap->createLargeProxy(aabbMin,aabbMax,bodyIndex,1,1);//m_dispatcher);	
+				m_data->m_broadphaseSap->createLargeProxy(aabbMin, aabbMax, bodyIndex, 1, 1);  //m_dispatcher);
 			}
 		}
 	}
@@ -699,10 +668,10 @@ int		b3GpuRigidBodyPipeline::registerPhysicsInstance(float mass, const float* po
 	return bodyIndex;
 }
 
-void	b3GpuRigidBodyPipeline::castRays(const b3AlignedObjectArray<b3RayInfo>& rays,	b3AlignedObjectArray<b3RayHit>& hitResults)
+void b3GpuRigidBodyPipeline::castRays(const b3AlignedObjectArray<b3RayInfo>& rays, b3AlignedObjectArray<b3RayHit>& hitResults)
 {
-	this->m_data->m_raycaster->castRays(rays,hitResults,
-		getNumBodies(),this->m_data->m_narrowphase->getBodiesCpu(),
-		m_data->m_narrowphase->getNumCollidablesGpu(), m_data->m_narrowphase->getCollidablesCpu(),
-		m_data->m_narrowphase->getInternalData(), m_data->m_broadphaseSap);
+	this->m_data->m_raycaster->castRays(rays, hitResults,
+										getNumBodies(), this->m_data->m_narrowphase->getBodiesCpu(),
+										m_data->m_narrowphase->getNumCollidablesGpu(), m_data->m_narrowphase->getCollidablesCpu(),
+										m_data->m_narrowphase->getInternalData(), m_data->m_broadphaseSap);
 }
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuRigidBodyPipeline.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuRigidBodyPipeline.h
index b4eac6841a1..0e5c6fec123 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuRigidBodyPipeline.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuRigidBodyPipeline.h
@@ -25,50 +25,46 @@ subject to the following restrictions:
 class b3GpuRigidBodyPipeline
 {
 protected:
-	struct b3GpuRigidBodyPipelineInternalData*	m_data;
+	struct b3GpuRigidBodyPipelineInternalData* m_data;
 
 	int allocateCollidable();
 
 public:
-
-
-	b3GpuRigidBodyPipeline(cl_context ctx,cl_device_id device, cl_command_queue  q , class b3GpuNarrowPhase* narrowphase, class b3GpuBroadphaseInterface* broadphaseSap, struct b3DynamicBvhBroadphase* broadphaseDbvt, const b3Config& config);
+	b3GpuRigidBodyPipeline(cl_context ctx, cl_device_id device, cl_command_queue q, class b3GpuNarrowPhase* narrowphase, class b3GpuBroadphaseInterface* broadphaseSap, struct b3DynamicBvhBroadphase* broadphaseDbvt, const b3Config& config);
 	virtual ~b3GpuRigidBodyPipeline();
 
-	void	stepSimulation(float deltaTime);
-	void	integrate(float timeStep);
-	void	setupGpuAabbsFull();
+	void stepSimulation(float deltaTime);
+	void integrate(float timeStep);
+	void setupGpuAabbsFull();
 
-	int		registerConvexPolyhedron(class b3ConvexUtility* convex);
+	int registerConvexPolyhedron(class b3ConvexUtility* convex);
 
 	//int		registerConvexPolyhedron(const float* vertices, int strideInBytes, int numVertices, const float* scaling);
 	//int		registerSphereShape(float radius);
 	//int		registerPlaneShape(const b3Vector3& planeNormal, float planeConstant);
-	
+
 	//int		registerConcaveMesh(b3AlignedObjectArray<b3Vector3>* vertices, b3AlignedObjectArray<int>* indices, const float* scaling);
 	//int		registerCompoundShape(b3AlignedObjectArray<b3GpuChildShape>* childShapes);
 
-	
-	int		registerPhysicsInstance(float mass, const float* position, const float* orientation, int collisionShapeIndex, int userData, bool writeInstanceToGpu);
+	int registerPhysicsInstance(float mass, const float* position, const float* orientation, int collisionShapeIndex, int userData, bool writeInstanceToGpu);
 	//if you passed "writeInstanceToGpu" false in the registerPhysicsInstance method (for performance) you need to call writeAllInstancesToGpu after all instances are registered
-	void	writeAllInstancesToGpu();
-	void	copyConstraintsToHost();
-	void	setGravity(const float* grav);
+	void writeAllInstancesToGpu();
+	void copyConstraintsToHost();
+	void setGravity(const float* grav);
 	void reset();
-	
-	int createPoint2PointConstraint(int bodyA, int bodyB, const float* pivotInA, const float* pivotInB,float breakingThreshold);
+
+	int createPoint2PointConstraint(int bodyA, int bodyB, const float* pivotInA, const float* pivotInB, float breakingThreshold);
 	int createFixedConstraint(int bodyA, int bodyB, const float* pivotInA, const float* pivotInB, const float* relTargetAB, float breakingThreshold);
 	void removeConstraintByUid(int uid);
 
-	void	addConstraint(class b3TypedConstraint* constraint);
-	void	removeConstraint(b3TypedConstraint* constraint);
+	void addConstraint(class b3TypedConstraint* constraint);
+	void removeConstraint(b3TypedConstraint* constraint);
 
-	void	castRays(const b3AlignedObjectArray<b3RayInfo>& rays,	b3AlignedObjectArray<b3RayHit>& hitResults);
+	void castRays(const b3AlignedObjectArray<b3RayInfo>& rays, b3AlignedObjectArray<b3RayHit>& hitResults);
 
-	cl_mem	getBodyBuffer();
-
-	int	getNumBodies() const;
+	cl_mem getBodyBuffer();
 
+	int getNumBodies() const;
 };
 
-#endif //B3_GPU_RIGIDBODY_PIPELINE_H
\ No newline at end of file
+#endif  //B3_GPU_RIGIDBODY_PIPELINE_H
\ No newline at end of file
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuRigidBodyPipelineInternalData.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuRigidBodyPipelineInternalData.h
index 5ac92f97d61..e0a26fda17c 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuRigidBodyPipelineInternalData.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuRigidBodyPipelineInternalData.h
@@ -22,52 +22,47 @@ subject to the following restrictions:
 #include "Bullet3OpenCL/ParallelPrimitives/b3OpenCLArray.h"
 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3Collidable.h"
 
-
 #include "Bullet3OpenCL/BroadphaseCollision/b3SapAabb.h"
 #include "Bullet3Dynamics/ConstraintSolver/b3TypedConstraint.h"
 #include "Bullet3Collision/NarrowPhaseCollision/b3Config.h"
 
-
-
 #include "Bullet3Collision/BroadPhaseCollision/b3OverlappingPair.h"
 #include "Bullet3OpenCL/RigidBody/b3GpuGenericConstraint.h"
 
 struct b3GpuRigidBodyPipelineInternalData
 {
+	cl_context m_context;
+	cl_device_id m_device;
+	cl_command_queue m_queue;
 
-	cl_context			m_context;
-	cl_device_id		m_device;
-	cl_command_queue	m_queue;
+	cl_kernel m_integrateTransformsKernel;
+	cl_kernel m_updateAabbsKernel;
+	cl_kernel m_clearOverlappingPairsKernel;
 
-	cl_kernel	m_integrateTransformsKernel;
-	cl_kernel	m_updateAabbsKernel;
-	cl_kernel	m_clearOverlappingPairsKernel;
-	
 	class b3PgsJacobiSolver* m_solver;
-	
+
 	class b3GpuPgsConstraintSolver* m_gpuSolver;
 
 	class b3GpuPgsContactSolver* m_solver2;
 	class b3GpuJacobiContactSolver* m_solver3;
 	class b3GpuRaycast* m_raycaster;
-	
+
 	class b3GpuBroadphaseInterface* m_broadphaseSap;
-	
+
 	struct b3DynamicBvhBroadphase* m_broadphaseDbvt;
-	b3OpenCLArray<b3SapAabb>*	m_allAabbsGPU;
-	b3AlignedObjectArray<b3SapAabb>	m_allAabbsCPU;
-	b3OpenCLArray<b3BroadphasePair>*		m_overlappingPairsGPU;
+	b3OpenCLArray<b3SapAabb>* m_allAabbsGPU;
+	b3AlignedObjectArray<b3SapAabb> m_allAabbsCPU;
+	b3OpenCLArray<b3BroadphasePair>* m_overlappingPairsGPU;
 
 	b3OpenCLArray<b3GpuGenericConstraint>* m_gpuConstraints;
 	b3AlignedObjectArray<b3GpuGenericConstraint> m_cpuConstraints;
 
 	b3AlignedObjectArray<b3TypedConstraint*> m_joints;
-	int	m_constraintUid;
-	class b3GpuNarrowPhase*	m_narrowphase;
-	b3Vector3	m_gravity;
+	int m_constraintUid;
+	class b3GpuNarrowPhase* m_narrowphase;
+	b3Vector3 m_gravity;
 
-	b3Config	m_config;
+	b3Config m_config;
 };
 
-#endif //B3_GPU_RIGIDBODY_PIPELINE_INTERNAL_DATA_H
-
+#endif  //B3_GPU_RIGIDBODY_PIPELINE_INTERNAL_DATA_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuSolverBody.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuSolverBody.h
index f2a61801ac9..db815d9b31d 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuSolverBody.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuSolverBody.h
@@ -13,11 +13,9 @@ subject to the following restrictions:
 */
 //Originally written by Erwin Coumans
 
-
 #ifndef B3_GPU_SOLVER_BODY_H
 #define B3_GPU_SOLVER_BODY_H
 
-
 #include "Bullet3Common/b3Vector3.h"
 #include "Bullet3Common/b3Matrix3x3.h"
 
@@ -27,29 +25,27 @@ subject to the following restrictions:
 ///Until we get other contributions, only use SIMD on Windows, when using Visual Studio 2008 or later, and not double precision
 #ifdef B3_USE_SSE
 #define USE_SIMD 1
-#endif //
-
-
+#endif  //
 
 ///The b3SolverBody is an internal datastructure for the constraint solver. Only necessary data is packed to increase cache coherence/performance.
-B3_ATTRIBUTE_ALIGNED16 (struct)	b3GpuSolverBody
+B3_ATTRIBUTE_ALIGNED16(struct)
+b3GpuSolverBody
 {
 	B3_DECLARE_ALIGNED_ALLOCATOR();
-//	b3Transform		m_worldTransformUnused;
-	b3Vector3		m_deltaLinearVelocity;
-	b3Vector3		m_deltaAngularVelocity;
-	b3Vector3		m_angularFactor;
-	b3Vector3		m_linearFactor;
-	b3Vector3		m_invMass;
-	b3Vector3		m_pushVelocity;
-	b3Vector3		m_turnVelocity;
-	b3Vector3		m_linearVelocity;
-	b3Vector3		m_angularVelocity;
+	//	b3Transform		m_worldTransformUnused;
+	b3Vector3 m_deltaLinearVelocity;
+	b3Vector3 m_deltaAngularVelocity;
+	b3Vector3 m_angularFactor;
+	b3Vector3 m_linearFactor;
+	b3Vector3 m_invMass;
+	b3Vector3 m_pushVelocity;
+	b3Vector3 m_turnVelocity;
+	b3Vector3 m_linearVelocity;
+	b3Vector3 m_angularVelocity;
 
-	union 
-	{
-		void*	m_originalBody;
-		int		m_originalBodyIndex;
+	union {
+		void* m_originalBody;
+		int m_originalBodyIndex;
 	};
 
 	int padding[3];
@@ -65,44 +61,41 @@ B3_ATTRIBUTE_ALIGNED16 (struct)	b3GpuSolverBody
 		return m_worldTransform;
 	}
 	*/
-	B3_FORCE_INLINE void	getVelocityInLocalPointObsolete(const b3Vector3& rel_pos, b3Vector3& velocity ) const
+	B3_FORCE_INLINE void getVelocityInLocalPointObsolete(const b3Vector3& rel_pos, b3Vector3& velocity) const
 	{
 		if (m_originalBody)
-			velocity = m_linearVelocity+m_deltaLinearVelocity + (m_angularVelocity+m_deltaAngularVelocity).cross(rel_pos);
+			velocity = m_linearVelocity + m_deltaLinearVelocity + (m_angularVelocity + m_deltaAngularVelocity).cross(rel_pos);
 		else
-			velocity.setValue(0,0,0);
+			velocity.setValue(0, 0, 0);
 	}
 
-	B3_FORCE_INLINE void	getAngularVelocity(b3Vector3& angVel) const
+	B3_FORCE_INLINE void getAngularVelocity(b3Vector3 & angVel) const
 	{
 		if (m_originalBody)
-			angVel =m_angularVelocity+m_deltaAngularVelocity;
+			angVel = m_angularVelocity + m_deltaAngularVelocity;
 		else
-			angVel.setValue(0,0,0);
+			angVel.setValue(0, 0, 0);
 	}
 
-
 	//Optimization for the iterative solver: avoid calculating constant terms involving inertia, normal, relative position
-	B3_FORCE_INLINE void applyImpulse(const b3Vector3& linearComponent, const b3Vector3& angularComponent,const b3Scalar impulseMagnitude)
+	B3_FORCE_INLINE void applyImpulse(const b3Vector3& linearComponent, const b3Vector3& angularComponent, const b3Scalar impulseMagnitude)
 	{
 		if (m_originalBody)
 		{
-			m_deltaLinearVelocity += linearComponent*impulseMagnitude*m_linearFactor;
-			m_deltaAngularVelocity += angularComponent*(impulseMagnitude*m_angularFactor);
+			m_deltaLinearVelocity += linearComponent * impulseMagnitude * m_linearFactor;
+			m_deltaAngularVelocity += angularComponent * (impulseMagnitude * m_angularFactor);
 		}
 	}
 
-	B3_FORCE_INLINE void internalApplyPushImpulse(const b3Vector3& linearComponent, const b3Vector3& angularComponent,b3Scalar impulseMagnitude)
+	B3_FORCE_INLINE void internalApplyPushImpulse(const b3Vector3& linearComponent, const b3Vector3& angularComponent, b3Scalar impulseMagnitude)
 	{
 		if (m_originalBody)
 		{
-			m_pushVelocity += linearComponent*impulseMagnitude*m_linearFactor;
-			m_turnVelocity += angularComponent*(impulseMagnitude*m_angularFactor);
+			m_pushVelocity += linearComponent * impulseMagnitude * m_linearFactor;
+			m_turnVelocity += angularComponent * (impulseMagnitude * m_angularFactor);
 		}
 	}
 
-
-
 	const b3Vector3& getDeltaLinearVelocity() const
 	{
 		return m_deltaLinearVelocity;
@@ -113,20 +106,19 @@ B3_ATTRIBUTE_ALIGNED16 (struct)	b3GpuSolverBody
 		return m_deltaAngularVelocity;
 	}
 
-	const b3Vector3& getPushVelocity() const 
+	const b3Vector3& getPushVelocity() const
 	{
 		return m_pushVelocity;
 	}
 
-	const b3Vector3& getTurnVelocity() const 
+	const b3Vector3& getTurnVelocity() const
 	{
 		return m_turnVelocity;
 	}
 
-
 	////////////////////////////////////////////////
 	///some internal methods, don't use them
-		
+
 	b3Vector3& internalGetDeltaLinearVelocity()
 	{
 		return m_deltaLinearVelocity;
@@ -151,7 +143,7 @@ B3_ATTRIBUTE_ALIGNED16 (struct)	b3GpuSolverBody
 	{
 		m_invMass = invMass;
 	}
-	
+
 	b3Vector3& internalGetPushVelocity()
 	{
 		return m_pushVelocity;
@@ -162,67 +154,57 @@ B3_ATTRIBUTE_ALIGNED16 (struct)	b3GpuSolverBody
 		return m_turnVelocity;
 	}
 
-	B3_FORCE_INLINE void	internalGetVelocityInLocalPointObsolete(const b3Vector3& rel_pos, b3Vector3& velocity ) const
+	B3_FORCE_INLINE void internalGetVelocityInLocalPointObsolete(const b3Vector3& rel_pos, b3Vector3& velocity) const
 	{
-		velocity = m_linearVelocity+m_deltaLinearVelocity + (m_angularVelocity+m_deltaAngularVelocity).cross(rel_pos);
+		velocity = m_linearVelocity + m_deltaLinearVelocity + (m_angularVelocity + m_deltaAngularVelocity).cross(rel_pos);
 	}
 
-	B3_FORCE_INLINE void	internalGetAngularVelocity(b3Vector3& angVel) const
+	B3_FORCE_INLINE void internalGetAngularVelocity(b3Vector3 & angVel) const
 	{
-		angVel = m_angularVelocity+m_deltaAngularVelocity;
+		angVel = m_angularVelocity + m_deltaAngularVelocity;
 	}
 
-
 	//Optimization for the iterative solver: avoid calculating constant terms involving inertia, normal, relative position
-	B3_FORCE_INLINE void internalApplyImpulse(const b3Vector3& linearComponent, const b3Vector3& angularComponent,const b3Scalar impulseMagnitude)
+	B3_FORCE_INLINE void internalApplyImpulse(const b3Vector3& linearComponent, const b3Vector3& angularComponent, const b3Scalar impulseMagnitude)
 	{
 		//if (m_originalBody)
 		{
-			m_deltaLinearVelocity += linearComponent*impulseMagnitude*m_linearFactor;
-			m_deltaAngularVelocity += angularComponent*(impulseMagnitude*m_angularFactor);
+			m_deltaLinearVelocity += linearComponent * impulseMagnitude * m_linearFactor;
+			m_deltaAngularVelocity += angularComponent * (impulseMagnitude * m_angularFactor);
 		}
 	}
-		
-	
-	
 
-	void	writebackVelocity()
+	void writebackVelocity()
 	{
 		//if (m_originalBody>=0)
 		{
-			m_linearVelocity +=m_deltaLinearVelocity;
+			m_linearVelocity += m_deltaLinearVelocity;
 			m_angularVelocity += m_deltaAngularVelocity;
-			
+
 			//m_originalBody->setCompanionId(-1);
 		}
 	}
 
-
-	void	writebackVelocityAndTransform(b3Scalar timeStep, b3Scalar splitImpulseTurnErp)
+	void writebackVelocityAndTransform(b3Scalar timeStep, b3Scalar splitImpulseTurnErp)
 	{
-        (void) timeStep;
+		(void)timeStep;
 		if (m_originalBody)
 		{
 			m_linearVelocity += m_deltaLinearVelocity;
 			m_angularVelocity += m_deltaAngularVelocity;
-			
+
 			//correct the position/orientation based on push/turn recovery
 			b3Transform newTransform;
-			if (m_pushVelocity[0]!=0.f || m_pushVelocity[1]!=0 || m_pushVelocity[2]!=0 || m_turnVelocity[0]!=0.f || m_turnVelocity[1]!=0 || m_turnVelocity[2]!=0)
+			if (m_pushVelocity[0] != 0.f || m_pushVelocity[1] != 0 || m_pushVelocity[2] != 0 || m_turnVelocity[0] != 0.f || m_turnVelocity[1] != 0 || m_turnVelocity[2] != 0)
 			{
-			//	b3Quaternion orn = m_worldTransform.getRotation();
-//				b3TransformUtil::integrateTransform(m_worldTransform,m_pushVelocity,m_turnVelocity*splitImpulseTurnErp,timeStep,newTransform);
-//				m_worldTransform = newTransform;
+				//	b3Quaternion orn = m_worldTransform.getRotation();
+				//				b3TransformUtil::integrateTransform(m_worldTransform,m_pushVelocity,m_turnVelocity*splitImpulseTurnErp,timeStep,newTransform);
+				//				m_worldTransform = newTransform;
 			}
 			//m_worldTransform.setRotation(orn);
 			//m_originalBody->setCompanionId(-1);
 		}
 	}
-	
-
-
 };
 
-#endif //B3_SOLVER_BODY_H
-
-
+#endif  //B3_SOLVER_BODY_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuSolverConstraint.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuSolverConstraint.h
index 60d235baab0..7d9eea243ab 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuSolverConstraint.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3GpuSolverConstraint.h
@@ -13,11 +13,9 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef B3_GPU_SOLVER_CONSTRAINT_H
 #define B3_GPU_SOLVER_CONSTRAINT_H
 
-
 #include "Bullet3Common/b3Vector3.h"
 #include "Bullet3Common/b3Matrix3x3.h"
 //#include "b3JacobianEntry.h"
@@ -25,58 +23,51 @@ subject to the following restrictions:
 
 //#define NO_FRICTION_TANGENTIALS 1
 
-
-
 ///1D constraint along a normal axis between bodyA and bodyB. It can be combined to solve contact and friction constraints.
-B3_ATTRIBUTE_ALIGNED16 (struct)	b3GpuSolverConstraint
+B3_ATTRIBUTE_ALIGNED16(struct)
+b3GpuSolverConstraint
 {
 	B3_DECLARE_ALIGNED_ALLOCATOR();
 
-	b3Vector3		m_relpos1CrossNormal;
-	b3Vector3		m_contactNormal;
+	b3Vector3 m_relpos1CrossNormal;
+	b3Vector3 m_contactNormal;
 
-	b3Vector3		m_relpos2CrossNormal;
+	b3Vector3 m_relpos2CrossNormal;
 	//b3Vector3		m_contactNormal2;//usually m_contactNormal2 == -m_contactNormal
 
-	b3Vector3		m_angularComponentA;
-	b3Vector3		m_angularComponentB;
-	
-	mutable b3Scalar	m_appliedPushImpulse;
-	mutable b3Scalar	m_appliedImpulse;
+	b3Vector3 m_angularComponentA;
+	b3Vector3 m_angularComponentB;
+
+	mutable b3Scalar m_appliedPushImpulse;
+	mutable b3Scalar m_appliedImpulse;
 	int m_padding1;
 	int m_padding2;
-	b3Scalar	m_friction;
-	b3Scalar	m_jacDiagABInv;
-	b3Scalar		m_rhs;
-	b3Scalar		m_cfm;
-	
-    b3Scalar		m_lowerLimit;
-	b3Scalar		m_upperLimit;
-	b3Scalar		m_rhsPenetration;
-    union
-	{
-		void*		m_originalContactPoint;
-		int		m_originalConstraintIndex;
-		b3Scalar	m_unusedPadding4;
+	b3Scalar m_friction;
+	b3Scalar m_jacDiagABInv;
+	b3Scalar m_rhs;
+	b3Scalar m_cfm;
+
+	b3Scalar m_lowerLimit;
+	b3Scalar m_upperLimit;
+	b3Scalar m_rhsPenetration;
+	union {
+		void* m_originalContactPoint;
+		int m_originalConstraintIndex;
+		b3Scalar m_unusedPadding4;
 	};
 
-	int	m_overrideNumSolverIterations;
-    int			m_frictionIndex;
+	int m_overrideNumSolverIterations;
+	int m_frictionIndex;
 	int m_solverBodyIdA;
 	int m_solverBodyIdB;
 
-    
-	enum		b3SolverConstraintType
+	enum b3SolverConstraintType
 	{
 		B3_SOLVER_CONTACT_1D = 0,
 		B3_SOLVER_FRICTION_1D
 	};
 };
 
-typedef b3AlignedObjectArray<b3GpuSolverConstraint>	b3GpuConstraintArray;
-
-
-#endif //B3_GPU_SOLVER_CONSTRAINT_H
-
-
+typedef b3AlignedObjectArray<b3GpuSolverConstraint> b3GpuConstraintArray;
 
+#endif  //B3_GPU_SOLVER_CONSTRAINT_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3Solver.cpp b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3Solver.cpp
index 20bf6d47c5e..ccf67da1a82 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3Solver.cpp
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3Solver.cpp
@@ -13,7 +13,6 @@ subject to the following restrictions:
 */
 //Originally written by Takahiro Harada
 
-
 #include "b3Solver.h"
 
 ///useNewBatchingKernel  is a rewritten kernel using just a single thread of the warp, for experiments
@@ -38,7 +37,6 @@ bool gConvertConstraintOnCpu = false;
 #include "kernels/batchingKernels.h"
 #include "kernels/batchingKernelsNew.h"
 
-
 #include "Bullet3OpenCL/ParallelPrimitives/b3LauncherCL.h"
 #include "Bullet3Common/b3Vector3.h"
 
@@ -48,7 +46,7 @@ struct SolverDebugInfo
 	int m_valInt1;
 	int m_valInt2;
 	int m_valInt3;
-	
+
 	int m_valInt4;
 	int m_valInt5;
 	int m_valInt6;
@@ -59,11 +57,10 @@ struct SolverDebugInfo
 	int m_valInt10;
 	int m_valInt11;
 
-	int	m_valInt12;
-	int	m_valInt13;
-	int	m_valInt14;
-	int	m_valInt15;
-
+	int m_valInt12;
+	int m_valInt13;
+	int m_valInt14;
+	int m_valInt15;
 
 	float m_val0;
 	float m_val1;
@@ -71,9 +68,6 @@ struct SolverDebugInfo
 	float m_val3;
 };
 
-
-
-
 class SolverDeviceInl
 {
 public:
@@ -84,101 +78,89 @@ public:
 	};
 };
 
-
-
 b3Solver::b3Solver(cl_context ctx, cl_device_id device, cl_command_queue queue, int pairCapacity)
-			:
-			m_context(ctx),
-			m_device(device),
-			m_queue(queue),
-			m_batchSizes(ctx,queue),
-			m_nIterations(4)
+	: m_context(ctx),
+	  m_device(device),
+	  m_queue(queue),
+	  m_batchSizes(ctx, queue),
+	  m_nIterations(4)
 {
-	m_sort32 = new b3RadixSort32CL(ctx,device,queue);
-	m_scan = new b3PrefixScanCL(ctx,device,queue,B3_SOLVER_N_CELLS);
-	m_search = new b3BoundSearchCL(ctx,device,queue,B3_SOLVER_N_CELLS);
+	m_sort32 = new b3RadixSort32CL(ctx, device, queue);
+	m_scan = new b3PrefixScanCL(ctx, device, queue, B3_SOLVER_N_CELLS);
+	m_search = new b3BoundSearchCL(ctx, device, queue, B3_SOLVER_N_CELLS);
 
-	const int sortSize = B3NEXTMULTIPLEOF( pairCapacity, 512 );
+	const int sortSize = B3NEXTMULTIPLEOF(pairCapacity, 512);
 
-	m_sortDataBuffer = new b3OpenCLArray<b3SortData>(ctx,queue,sortSize);
-	m_contactBuffer2 = new b3OpenCLArray<b3Contact4>(ctx,queue);
+	m_sortDataBuffer = new b3OpenCLArray<b3SortData>(ctx, queue, sortSize);
+	m_contactBuffer2 = new b3OpenCLArray<b3Contact4>(ctx, queue);
 
-	m_numConstraints = new b3OpenCLArray<unsigned int>(ctx,queue,B3_SOLVER_N_CELLS );
+	m_numConstraints = new b3OpenCLArray<unsigned int>(ctx, queue, B3_SOLVER_N_CELLS);
 	m_numConstraints->resize(B3_SOLVER_N_CELLS);
 
-	m_offsets = new b3OpenCLArray<unsigned int>( ctx,queue,B3_SOLVER_N_CELLS);
+	m_offsets = new b3OpenCLArray<unsigned int>(ctx, queue, B3_SOLVER_N_CELLS);
 	m_offsets->resize(B3_SOLVER_N_CELLS);
 	const char* additionalMacros = "";
-//	const char* srcFileNameForCaching="";
-
-
+	//	const char* srcFileNameForCaching="";
 
 	cl_int pErrNum;
 	const char* batchKernelSource = batchingKernelsCL;
 	const char* batchKernelNewSource = batchingKernelsNewCL;
-	
+
 	const char* solverSetupSource = solverSetupCL;
 	const char* solverSetup2Source = solverSetup2CL;
 	const char* solveContactSource = solveContactCL;
 	const char* solveFrictionSource = solveFrictionCL;
-	
-	
-	
+
 	{
-		
-		cl_program solveContactProg= b3OpenCLUtils::compileCLProgramFromString( ctx, device, solveContactSource, &pErrNum,additionalMacros, B3_SOLVER_CONTACT_KERNEL_PATH);
+		cl_program solveContactProg = b3OpenCLUtils::compileCLProgramFromString(ctx, device, solveContactSource, &pErrNum, additionalMacros, B3_SOLVER_CONTACT_KERNEL_PATH);
 		b3Assert(solveContactProg);
-		
-		cl_program solveFrictionProg= b3OpenCLUtils::compileCLProgramFromString( ctx, device, solveFrictionSource, &pErrNum,additionalMacros, B3_SOLVER_FRICTION_KERNEL_PATH);
+
+		cl_program solveFrictionProg = b3OpenCLUtils::compileCLProgramFromString(ctx, device, solveFrictionSource, &pErrNum, additionalMacros, B3_SOLVER_FRICTION_KERNEL_PATH);
 		b3Assert(solveFrictionProg);
 
-		cl_program solverSetup2Prog= b3OpenCLUtils::compileCLProgramFromString( ctx, device, solverSetup2Source, &pErrNum,additionalMacros, B3_SOLVER_SETUP2_KERNEL_PATH);
+		cl_program solverSetup2Prog = b3OpenCLUtils::compileCLProgramFromString(ctx, device, solverSetup2Source, &pErrNum, additionalMacros, B3_SOLVER_SETUP2_KERNEL_PATH);
 		b3Assert(solverSetup2Prog);
 
-		
-		cl_program solverSetupProg= b3OpenCLUtils::compileCLProgramFromString( ctx, device, solverSetupSource, &pErrNum,additionalMacros, B3_SOLVER_SETUP_KERNEL_PATH);
+		cl_program solverSetupProg = b3OpenCLUtils::compileCLProgramFromString(ctx, device, solverSetupSource, &pErrNum, additionalMacros, B3_SOLVER_SETUP_KERNEL_PATH);
 		b3Assert(solverSetupProg);
-		
-		
-		m_solveFrictionKernel= b3OpenCLUtils::compileCLKernelFromString( ctx, device, solveFrictionSource, "BatchSolveKernelFriction", &pErrNum, solveFrictionProg,additionalMacros );
+
+		m_solveFrictionKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solveFrictionSource, "BatchSolveKernelFriction", &pErrNum, solveFrictionProg, additionalMacros);
 		b3Assert(m_solveFrictionKernel);
 
-		m_solveContactKernel= b3OpenCLUtils::compileCLKernelFromString( ctx, device, solveContactSource, "BatchSolveKernelContact", &pErrNum, solveContactProg,additionalMacros );
+		m_solveContactKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solveContactSource, "BatchSolveKernelContact", &pErrNum, solveContactProg, additionalMacros);
 		b3Assert(m_solveContactKernel);
-		
-		m_contactToConstraintKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, solverSetupSource, "ContactToConstraintKernel", &pErrNum, solverSetupProg,additionalMacros );
-		b3Assert(m_contactToConstraintKernel);
-			
-		m_setSortDataKernel =  b3OpenCLUtils::compileCLKernelFromString( ctx, device, solverSetup2Source, "SetSortDataKernel", &pErrNum, solverSetup2Prog,additionalMacros );
-		b3Assert(m_setSortDataKernel);
-				
-		m_reorderContactKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, solverSetup2Source, "ReorderContactKernel", &pErrNum, solverSetup2Prog,additionalMacros );
-		b3Assert(m_reorderContactKernel);
-		
 
-		m_copyConstraintKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, solverSetup2Source, "CopyConstraintKernel", &pErrNum, solverSetup2Prog,additionalMacros );
+		m_contactToConstraintKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solverSetupSource, "ContactToConstraintKernel", &pErrNum, solverSetupProg, additionalMacros);
+		b3Assert(m_contactToConstraintKernel);
+
+		m_setSortDataKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solverSetup2Source, "SetSortDataKernel", &pErrNum, solverSetup2Prog, additionalMacros);
+		b3Assert(m_setSortDataKernel);
+
+		m_reorderContactKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solverSetup2Source, "ReorderContactKernel", &pErrNum, solverSetup2Prog, additionalMacros);
+		b3Assert(m_reorderContactKernel);
+
+		m_copyConstraintKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, solverSetup2Source, "CopyConstraintKernel", &pErrNum, solverSetup2Prog, additionalMacros);
 		b3Assert(m_copyConstraintKernel);
-		
 	}
 
 	{
-		cl_program batchingProg = b3OpenCLUtils::compileCLProgramFromString( ctx, device, batchKernelSource, &pErrNum,additionalMacros, B3_BATCHING_PATH);
+		cl_program batchingProg = b3OpenCLUtils::compileCLProgramFromString(ctx, device, batchKernelSource, &pErrNum, additionalMacros, B3_BATCHING_PATH);
 		//cl_program batchingProg = b3OpenCLUtils::compileCLProgramFromString( ctx, device, 0, &pErrNum,additionalMacros, B3_BATCHING_PATH,true);
 		b3Assert(batchingProg);
-		
-		m_batchingKernel = b3OpenCLUtils::compileCLKernelFromString( ctx, device, batchKernelSource, "CreateBatches", &pErrNum, batchingProg,additionalMacros );
+
+		m_batchingKernel = b3OpenCLUtils::compileCLKernelFromString(ctx, device, batchKernelSource, "CreateBatches", &pErrNum, batchingProg, additionalMacros);
 		b3Assert(m_batchingKernel);
 	}
 	{
-		cl_program batchingNewProg = b3OpenCLUtils::compileCLProgramFromString( ctx, device, batchKernelNewSource, &pErrNum,additionalMacros, B3_BATCHING_NEW_PATH);
+		cl_program batchingNewProg = b3OpenCLUtils::compileCLProgramFromString(ctx, device, batchKernelNewSource, &pErrNum, additionalMacros, B3_BATCHING_NEW_PATH);
 		b3Assert(batchingNewProg);
 
-		m_batchingKernelNew = b3OpenCLUtils::compileCLKernelFromString( ctx, device, batchKernelNewSource, "CreateBatchesNew", &pErrNum, batchingNewProg,additionalMacros );
+		m_batchingKernelNew = b3OpenCLUtils::compileCLKernelFromString(ctx, device, batchKernelNewSource, "CreateBatchesNew", &pErrNum, batchingNewProg, additionalMacros);
 		//m_batchingKernelNew = b3OpenCLUtils::compileCLKernelFromString( ctx, device, batchKernelNewSource, "CreateBatchesBruteForce", &pErrNum, batchingNewProg,additionalMacros );
 		b3Assert(m_batchingKernelNew);
 	}
 }
-		
+
 b3Solver::~b3Solver()
 {
 	delete m_offsets;
@@ -190,71 +172,68 @@ b3Solver::~b3Solver()
 	delete m_scan;
 	delete m_search;
 
-
 	clReleaseKernel(m_batchingKernel);
 	clReleaseKernel(m_batchingKernelNew);
-	
-	clReleaseKernel( m_solveContactKernel);
-	clReleaseKernel( m_solveFrictionKernel);
 
-	clReleaseKernel( m_contactToConstraintKernel);
-	clReleaseKernel( m_setSortDataKernel);
-	clReleaseKernel( m_reorderContactKernel);
-	clReleaseKernel( m_copyConstraintKernel);
-			
+	clReleaseKernel(m_solveContactKernel);
+	clReleaseKernel(m_solveFrictionKernel);
+
+	clReleaseKernel(m_contactToConstraintKernel);
+	clReleaseKernel(m_setSortDataKernel);
+	clReleaseKernel(m_reorderContactKernel);
+	clReleaseKernel(m_copyConstraintKernel);
 }
 
-
- 
-
-template<bool JACOBI>
-static
-__inline
-void solveContact(b3GpuConstraint4& cs, 
-	const b3Vector3& posA, b3Vector3& linVelA, b3Vector3& angVelA, float invMassA, const b3Matrix3x3& invInertiaA,
-	const b3Vector3& posB, b3Vector3& linVelB, b3Vector3& angVelB, float invMassB, const b3Matrix3x3& invInertiaB, 
-	float maxRambdaDt[4], float minRambdaDt[4])
+template <bool JACOBI>
+static __inline void solveContact(b3GpuConstraint4& cs,
+								  const b3Vector3& posA, b3Vector3& linVelA, b3Vector3& angVelA, float invMassA, const b3Matrix3x3& invInertiaA,
+								  const b3Vector3& posB, b3Vector3& linVelB, b3Vector3& angVelB, float invMassB, const b3Matrix3x3& invInertiaB,
+								  float maxRambdaDt[4], float minRambdaDt[4])
 {
+	b3Vector3 dLinVelA;
+	dLinVelA.setZero();
+	b3Vector3 dAngVelA;
+	dAngVelA.setZero();
+	b3Vector3 dLinVelB;
+	dLinVelB.setZero();
+	b3Vector3 dAngVelB;
+	dAngVelB.setZero();
 
-	b3Vector3 dLinVelA; dLinVelA.setZero();
-	b3Vector3 dAngVelA; dAngVelA.setZero();
-	b3Vector3 dLinVelB; dLinVelB.setZero();
-	b3Vector3 dAngVelB; dAngVelB.setZero();
-
-	for(int ic=0; ic<4; ic++)
+	for (int ic = 0; ic < 4; ic++)
 	{
 		//	dont necessary because this makes change to 0
-		if( cs.m_jacCoeffInv[ic] == 0.f ) continue;
+		if (cs.m_jacCoeffInv[ic] == 0.f) continue;
 
 		{
 			b3Vector3 angular0, angular1, linear;
 			b3Vector3 r0 = cs.m_worldPos[ic] - (b3Vector3&)posA;
 			b3Vector3 r1 = cs.m_worldPos[ic] - (b3Vector3&)posB;
-			setLinearAndAngular( (const b3Vector3 &)cs.m_linear, (const b3Vector3 &)r0, (const b3Vector3 &)r1, &linear, &angular0, &angular1 );
+			setLinearAndAngular((const b3Vector3&)cs.m_linear, (const b3Vector3&)r0, (const b3Vector3&)r1, &linear, &angular0, &angular1);
 
-			float rambdaDt = calcRelVel((const b3Vector3 &)cs.m_linear,(const b3Vector3 &) -cs.m_linear, angular0, angular1,
-				linVelA, angVelA, linVelB, angVelB ) + cs.m_b[ic];
+			float rambdaDt = calcRelVel((const b3Vector3&)cs.m_linear, (const b3Vector3&)-cs.m_linear, angular0, angular1,
+										linVelA, angVelA, linVelB, angVelB) +
+							 cs.m_b[ic];
 			rambdaDt *= cs.m_jacCoeffInv[ic];
 
 			{
 				float prevSum = cs.m_appliedRambdaDt[ic];
 				float updated = prevSum;
 				updated += rambdaDt;
-				updated = b3Max( updated, minRambdaDt[ic] );
-				updated = b3Min( updated, maxRambdaDt[ic] );
+				updated = b3Max(updated, minRambdaDt[ic]);
+				updated = b3Min(updated, maxRambdaDt[ic]);
 				rambdaDt = updated - prevSum;
 				cs.m_appliedRambdaDt[ic] = updated;
 			}
 
-			b3Vector3 linImp0 = invMassA*linear*rambdaDt;
-			b3Vector3 linImp1 = invMassB*(-linear)*rambdaDt;
-			b3Vector3 angImp0 = (invInertiaA* angular0)*rambdaDt;
-			b3Vector3 angImp1 = (invInertiaB* angular1)*rambdaDt;
+			b3Vector3 linImp0 = invMassA * linear * rambdaDt;
+			b3Vector3 linImp1 = invMassB * (-linear) * rambdaDt;
+			b3Vector3 angImp0 = (invInertiaA * angular0) * rambdaDt;
+			b3Vector3 angImp1 = (invInertiaB * angular1) * rambdaDt;
 #ifdef _WIN32
-            b3Assert(_finite(linImp0.getX()));
+			b3Assert(_finite(linImp0.getX()));
 			b3Assert(_finite(linImp1.getX()));
 #endif
-			if( JACOBI )
+			if (JACOBI)
 			{
 				dLinVelA += linImp0;
 				dAngVelA += angImp0;
@@ -271,92 +250,83 @@ void solveContact(b3GpuConstraint4& cs,
 		}
 	}
 
-	if( JACOBI )
+	if (JACOBI)
 	{
 		linVelA += dLinVelA;
 		angVelA += dAngVelA;
 		linVelB += dLinVelB;
 		angVelB += dAngVelB;
 	}
-
 }
 
+static __inline void solveFriction(b3GpuConstraint4& cs,
+								   const b3Vector3& posA, b3Vector3& linVelA, b3Vector3& angVelA, float invMassA, const b3Matrix3x3& invInertiaA,
+								   const b3Vector3& posB, b3Vector3& linVelB, b3Vector3& angVelB, float invMassB, const b3Matrix3x3& invInertiaB,
+								   float maxRambdaDt[4], float minRambdaDt[4])
+{
+	if (cs.m_fJacCoeffInv[0] == 0 && cs.m_fJacCoeffInv[0] == 0) return;
+	const b3Vector3& center = (const b3Vector3&)cs.m_center;
 
+	b3Vector3 n = -(const b3Vector3&)cs.m_linear;
 
-
-
-	static
-	__inline
-	void solveFriction(b3GpuConstraint4& cs, 
-		const b3Vector3& posA, b3Vector3& linVelA, b3Vector3& angVelA, float invMassA, const b3Matrix3x3& invInertiaA,
-		const b3Vector3& posB, b3Vector3& linVelB, b3Vector3& angVelB, float invMassB, const b3Matrix3x3& invInertiaB, 
-		float maxRambdaDt[4], float minRambdaDt[4])
-	{
-
-		if( cs.m_fJacCoeffInv[0] == 0 && cs.m_fJacCoeffInv[0] == 0 ) return;
-		const b3Vector3& center = (const b3Vector3&)cs.m_center;
-
-		b3Vector3 n = -(const b3Vector3&)cs.m_linear;
-
-		b3Vector3 tangent[2];
-#if 1		
-		b3PlaneSpace1 (n, tangent[0],tangent[1]);
+	b3Vector3 tangent[2];
+#if 1
+	b3PlaneSpace1(n, tangent[0], tangent[1]);
 #else
-		b3Vector3 r = cs.m_worldPos[0]-center;
-		tangent[0] = cross3( n, r );
-		tangent[1] = cross3( tangent[0], n );
-		tangent[0] = normalize3( tangent[0] );
-		tangent[1] = normalize3( tangent[1] );
+	b3Vector3 r = cs.m_worldPos[0] - center;
+	tangent[0] = cross3(n, r);
+	tangent[1] = cross3(tangent[0], n);
+	tangent[0] = normalize3(tangent[0]);
+	tangent[1] = normalize3(tangent[1]);
 #endif
 
-		b3Vector3 angular0, angular1, linear;
-		b3Vector3 r0 = center - posA;
-		b3Vector3 r1 = center - posB;
-		for(int i=0; i<2; i++)
+	b3Vector3 angular0, angular1, linear;
+	b3Vector3 r0 = center - posA;
+	b3Vector3 r1 = center - posB;
+	for (int i = 0; i < 2; i++)
+	{
+		setLinearAndAngular(tangent[i], r0, r1, &linear, &angular0, &angular1);
+		float rambdaDt = calcRelVel(linear, -linear, angular0, angular1,
+									linVelA, angVelA, linVelB, angVelB);
+		rambdaDt *= cs.m_fJacCoeffInv[i];
+
 		{
-			setLinearAndAngular( tangent[i], r0, r1, &linear, &angular0, &angular1 );
-			float rambdaDt = calcRelVel(linear, -linear, angular0, angular1,
-				linVelA, angVelA, linVelB, angVelB );
-			rambdaDt *= cs.m_fJacCoeffInv[i];
+			float prevSum = cs.m_fAppliedRambdaDt[i];
+			float updated = prevSum;
+			updated += rambdaDt;
+			updated = b3Max(updated, minRambdaDt[i]);
+			updated = b3Min(updated, maxRambdaDt[i]);
+			rambdaDt = updated - prevSum;
+			cs.m_fAppliedRambdaDt[i] = updated;
+		}
 
-				{
-					float prevSum = cs.m_fAppliedRambdaDt[i];
-					float updated = prevSum;
-					updated += rambdaDt;
-					updated = b3Max( updated, minRambdaDt[i] );
-					updated = b3Min( updated, maxRambdaDt[i] );
-					rambdaDt = updated - prevSum;
-					cs.m_fAppliedRambdaDt[i] = updated;
-				}
-
-			b3Vector3 linImp0 = invMassA*linear*rambdaDt;
-			b3Vector3 linImp1 = invMassB*(-linear)*rambdaDt;
-			b3Vector3 angImp0 = (invInertiaA* angular0)*rambdaDt;
-			b3Vector3 angImp1 = (invInertiaB* angular1)*rambdaDt;
+		b3Vector3 linImp0 = invMassA * linear * rambdaDt;
+		b3Vector3 linImp1 = invMassB * (-linear) * rambdaDt;
+		b3Vector3 angImp0 = (invInertiaA * angular0) * rambdaDt;
+		b3Vector3 angImp1 = (invInertiaB * angular1) * rambdaDt;
 #ifdef _WIN32
-			b3Assert(_finite(linImp0.getX()));
-			b3Assert(_finite(linImp1.getX()));
+		b3Assert(_finite(linImp0.getX()));
+		b3Assert(_finite(linImp1.getX()));
 #endif
-			linVelA += linImp0;
-			angVelA += angImp0;
-			linVelB += linImp1;
-			angVelB += angImp1;
-		}
-
-		{	//	angular damping for point constraint
-			b3Vector3 ab = ( posB - posA ).normalized();
-			b3Vector3 ac = ( center - posA ).normalized();
-			if( b3Dot( ab, ac ) > 0.95f || (invMassA == 0.f || invMassB == 0.f))
-			{
-				float angNA = b3Dot( n, angVelA );
-				float angNB = b3Dot( n, angVelB );
-
-				angVelA -= (angNA*0.1f)*n;
-				angVelB -= (angNB*0.1f)*n;
-			}
-		}
-
+		linVelA += linImp0;
+		angVelA += angImp0;
+		linVelB += linImp1;
+		angVelB += angImp1;
 	}
+
+	{  //	angular damping for point constraint
+		b3Vector3 ab = (posB - posA).normalized();
+		b3Vector3 ac = (center - posA).normalized();
+		if (b3Dot(ab, ac) > 0.95f || (invMassA == 0.f || invMassB == 0.f))
+		{
+			float angNA = b3Dot(n, angVelA);
+			float angNB = b3Dot(n, angVelB);
+
+			angVelA -= (angNA * 0.1f) * n;
+			angVelB -= (angNB * 0.1f) * n;
+		}
+	}
+}
 /*
  b3AlignedObjectArray<b3RigidBodyData>& m_bodies;
 	b3AlignedObjectArray<b3InertiaData>& m_shapes;
@@ -370,79 +340,69 @@ void solveContact(b3GpuConstraint4& cs,
 	int m_maxNumBatches;
  */
 
-struct SolveTask// : public ThreadPool::Task
+struct SolveTask  // : public ThreadPool::Task
 {
-	SolveTask(b3AlignedObjectArray<b3RigidBodyData>& bodies,  b3AlignedObjectArray<b3InertiaData>& shapes, b3AlignedObjectArray<b3GpuConstraint4>& constraints,
-		int start, int nConstraints,int maxNumBatches,b3AlignedObjectArray<int>* wgUsedBodies, int curWgidx, b3AlignedObjectArray<int>* batchSizes, int cellIndex)
-		: m_bodies( bodies ), m_shapes( shapes ), 
-		m_constraints( constraints ), 
-		m_batchSizes(batchSizes),
-		m_cellIndex(cellIndex),
-		m_curWgidx(curWgidx),
-		m_start( start ), 
-		m_nConstraints( nConstraints ),
-		m_solveFriction( true ),
-		m_maxNumBatches(maxNumBatches)
-	{}
+	SolveTask(b3AlignedObjectArray<b3RigidBodyData>& bodies, b3AlignedObjectArray<b3InertiaData>& shapes, b3AlignedObjectArray<b3GpuConstraint4>& constraints,
+			  int start, int nConstraints, int maxNumBatches, b3AlignedObjectArray<int>* wgUsedBodies, int curWgidx, b3AlignedObjectArray<int>* batchSizes, int cellIndex)
+		: m_bodies(bodies), m_shapes(shapes), m_constraints(constraints), m_batchSizes(batchSizes), m_cellIndex(cellIndex), m_curWgidx(curWgidx), m_start(start), m_nConstraints(nConstraints), m_solveFriction(true), m_maxNumBatches(maxNumBatches)
+	{
+	}
 
-	unsigned short int getType(){ return 0; }
+	unsigned short int getType() { return 0; }
 
 	void run(int tIdx)
 	{
 		int offset = 0;
-		for (int ii=0;ii<B3_MAX_NUM_BATCHES;ii++)
+		for (int ii = 0; ii < B3_MAX_NUM_BATCHES; ii++)
 		{
-			int numInBatch = m_batchSizes->at(m_cellIndex*B3_MAX_NUM_BATCHES+ii);
+			int numInBatch = m_batchSizes->at(m_cellIndex * B3_MAX_NUM_BATCHES + ii);
 			if (!numInBatch)
 				break;
 
-			for (int jj=0;jj<numInBatch;jj++)
+			for (int jj = 0; jj < numInBatch; jj++)
 			{
-				int i = m_start + offset+jj;
+				int i = m_start + offset + jj;
 				int batchId = m_constraints[i].m_batchIdx;
-				b3Assert(batchId==ii);
+				b3Assert(batchId == ii);
 				float frictionCoeff = m_constraints[i].getFrictionCoeff();
 				int aIdx = (int)m_constraints[i].m_bodyA;
 				int bIdx = (int)m_constraints[i].m_bodyB;
-//				int localBatch = m_constraints[i].m_batchIdx;
+				//				int localBatch = m_constraints[i].m_batchIdx;
 				b3RigidBodyData& bodyA = m_bodies[aIdx];
 				b3RigidBodyData& bodyB = m_bodies[bIdx];
 
-				if( !m_solveFriction )
+				if (!m_solveFriction)
 				{
-					float maxRambdaDt[4] = {FLT_MAX,FLT_MAX,FLT_MAX,FLT_MAX};
-					float minRambdaDt[4] = {0.f,0.f,0.f,0.f};
+					float maxRambdaDt[4] = {FLT_MAX, FLT_MAX, FLT_MAX, FLT_MAX};
+					float minRambdaDt[4] = {0.f, 0.f, 0.f, 0.f};
 
-					solveContact<false>( m_constraints[i], (b3Vector3&)bodyA.m_pos, (b3Vector3&)bodyA.m_linVel, (b3Vector3&)bodyA.m_angVel, bodyA.m_invMass, (const b3Matrix3x3 &)m_shapes[aIdx].m_invInertiaWorld, 
-							(b3Vector3&)bodyB.m_pos, (b3Vector3&)bodyB.m_linVel, (b3Vector3&)bodyB.m_angVel, bodyB.m_invMass, (const b3Matrix3x3 &)m_shapes[bIdx].m_invInertiaWorld,
-						maxRambdaDt, minRambdaDt );
+					solveContact<false>(m_constraints[i], (b3Vector3&)bodyA.m_pos, (b3Vector3&)bodyA.m_linVel, (b3Vector3&)bodyA.m_angVel, bodyA.m_invMass, (const b3Matrix3x3&)m_shapes[aIdx].m_invInertiaWorld,
+										(b3Vector3&)bodyB.m_pos, (b3Vector3&)bodyB.m_linVel, (b3Vector3&)bodyB.m_angVel, bodyB.m_invMass, (const b3Matrix3x3&)m_shapes[bIdx].m_invInertiaWorld,
+										maxRambdaDt, minRambdaDt);
 				}
 				else
 				{
-					float maxRambdaDt[4] = {FLT_MAX,FLT_MAX,FLT_MAX,FLT_MAX};
-					float minRambdaDt[4] = {0.f,0.f,0.f,0.f};
+					float maxRambdaDt[4] = {FLT_MAX, FLT_MAX, FLT_MAX, FLT_MAX};
+					float minRambdaDt[4] = {0.f, 0.f, 0.f, 0.f};
 					float sum = 0;
-					for(int j=0; j<4; j++)
+					for (int j = 0; j < 4; j++)
 					{
-						sum +=m_constraints[i].m_appliedRambdaDt[j];
+						sum += m_constraints[i].m_appliedRambdaDt[j];
 					}
 					frictionCoeff = 0.7f;
-					for(int j=0; j<4; j++)
+					for (int j = 0; j < 4; j++)
 					{
-						maxRambdaDt[j] = frictionCoeff*sum;
+						maxRambdaDt[j] = frictionCoeff * sum;
 						minRambdaDt[j] = -maxRambdaDt[j];
 					}
-					solveFriction( m_constraints[i], (b3Vector3&)bodyA.m_pos, (b3Vector3&)bodyA.m_linVel, (b3Vector3&)bodyA.m_angVel, bodyA.m_invMass,(const b3Matrix3x3 &) m_shapes[aIdx].m_invInertiaWorld, 
-						(b3Vector3&)bodyB.m_pos, (b3Vector3&)bodyB.m_linVel, (b3Vector3&)bodyB.m_angVel, bodyB.m_invMass,(const b3Matrix3x3 &) m_shapes[bIdx].m_invInertiaWorld,
-						maxRambdaDt, minRambdaDt );
-			
+					solveFriction(m_constraints[i], (b3Vector3&)bodyA.m_pos, (b3Vector3&)bodyA.m_linVel, (b3Vector3&)bodyA.m_angVel, bodyA.m_invMass, (const b3Matrix3x3&)m_shapes[aIdx].m_invInertiaWorld,
+								  (b3Vector3&)bodyB.m_pos, (b3Vector3&)bodyB.m_linVel, (b3Vector3&)bodyB.m_angVel, bodyB.m_invMass, (const b3Matrix3x3&)m_shapes[bIdx].m_invInertiaWorld,
+								  maxRambdaDt, minRambdaDt);
 				}
 			}
-			offset+=numInBatch;
-
-
+			offset += numInBatch;
 		}
-/*		for (int bb=0;bb<m_maxNumBatches;bb++)
+		/*		for (int bb=0;bb<m_maxNumBatches;bb++)
 		{
 			//for(int ic=m_nConstraints-1; ic>=0; ic--)
 			for(int ic=0; ic<m_nConstraints; ic++)
@@ -491,9 +451,6 @@ struct SolveTask// : public ThreadPool::Task
 			}
 		}
 		*/
-
-
-		
 	}
 
 	b3AlignedObjectArray<b3RigidBodyData>& m_bodies;
@@ -508,11 +465,9 @@ struct SolveTask// : public ThreadPool::Task
 	int m_maxNumBatches;
 };
 
-
-void b3Solver::solveContactConstraintHost(  b3OpenCLArray<b3RigidBodyData>* bodyBuf, b3OpenCLArray<b3InertiaData>* shapeBuf, 
-			b3OpenCLArray<b3GpuConstraint4>* constraint, void* additionalData, int n ,int maxNumBatches,b3AlignedObjectArray<int>* batchSizes)
+void b3Solver::solveContactConstraintHost(b3OpenCLArray<b3RigidBodyData>* bodyBuf, b3OpenCLArray<b3InertiaData>* shapeBuf,
+										  b3OpenCLArray<b3GpuConstraint4>* constraint, void* additionalData, int n, int maxNumBatches, b3AlignedObjectArray<int>* batchSizes)
 {
-
 #if 0
 	{	
 		int nSplitX = B3_SOLVER_N_SPLIT_X;
@@ -571,114 +526,105 @@ void b3Solver::solveContactConstraintHost(  b3OpenCLArray<b3RigidBodyData>* body
 	//printf("------------------------\n");
 	b3AlignedObjectArray<unsigned int> offsetsHost;
 	m_offsets->copyToHost(offsetsHost);
-	static int frame=0;
-	bool useBatches=true;
+	static int frame = 0;
+	bool useBatches = true;
 	if (useBatches)
 	{
-		for(int iter=0; iter<m_nIterations; iter++)
+		for (int iter = 0; iter < m_nIterations; iter++)
 		{
-			for (int cellBatch=0;cellBatch<B3_SOLVER_N_BATCHES;cellBatch++)
+			for (int cellBatch = 0; cellBatch < B3_SOLVER_N_BATCHES; cellBatch++)
 			{
-				
 				int nSplitX = B3_SOLVER_N_SPLIT_X;
 				int nSplitY = B3_SOLVER_N_SPLIT_Y;
-				int numWorkgroups = B3_SOLVER_N_CELLS/B3_SOLVER_N_BATCHES;
+				int numWorkgroups = B3_SOLVER_N_CELLS / B3_SOLVER_N_BATCHES;
 				//printf("cell Batch %d\n",cellBatch);
 				b3AlignedObjectArray<int> usedBodies[B3_SOLVER_N_CELLS];
-				for (int i=0;i<B3_SOLVER_N_CELLS;i++)
+				for (int i = 0; i < B3_SOLVER_N_CELLS; i++)
 				{
 					usedBodies[i].resize(0);
 				}
 
-				
-
-
 				//for (int wgIdx=numWorkgroups-1;wgIdx>=0;wgIdx--)
-				for (int wgIdx=0;wgIdx<numWorkgroups;wgIdx++)
+				for (int wgIdx = 0; wgIdx < numWorkgroups; wgIdx++)
 				{
-					int zIdx = (wgIdx/((nSplitX*nSplitY)/4))*2+((cellBatch&4)>>2);
-					int remain= (wgIdx%((nSplitX*nSplitY)/4));
-					int yIdx = (remain/(nSplitX/2))*2 + ((cellBatch&2)>>1);
-					int xIdx = (remain%(nSplitX/2))*2 + (cellBatch&1);
-					int cellIdx = xIdx+yIdx*nSplitX+zIdx*(nSplitX*nSplitY);
-					
-	
-					if( numConstraintsHost[cellIdx] == 0 ) 
+					int zIdx = (wgIdx / ((nSplitX * nSplitY) / 4)) * 2 + ((cellBatch & 4) >> 2);
+					int remain = (wgIdx % ((nSplitX * nSplitY) / 4));
+					int yIdx = (remain / (nSplitX / 2)) * 2 + ((cellBatch & 2) >> 1);
+					int xIdx = (remain % (nSplitX / 2)) * 2 + (cellBatch & 1);
+					int cellIdx = xIdx + yIdx * nSplitX + zIdx * (nSplitX * nSplitY);
+
+					if (numConstraintsHost[cellIdx] == 0)
 						continue;
 
 					//printf("wgIdx %d: xIdx=%d, yIdx=%d, zIdx=%d, cellIdx=%d, cell Batch %d\n",wgIdx,xIdx,yIdx,zIdx,cellIdx,cellBatch);
 					//printf("cell %d has %d constraints\n", cellIdx,numConstraintsHost[cellIdx]);
 					if (zIdx)
 					{
-					//printf("?\n");
+						//printf("?\n");
 					}
 
-					if (iter==0)
+					if (iter == 0)
 					{
 						//printf("frame=%d, Cell xIdx=%x, yIdx=%d ",frame, xIdx,yIdx);
 						//printf("cellBatch=%d, wgIdx=%d, #constraints in cell=%d\n",cellBatch,wgIdx,numConstraintsHost[cellIdx]);
 					}
 					const int start = offsetsHost[cellIdx];
 					int numConstraintsInCell = numConstraintsHost[cellIdx];
-	//				const int end = start + numConstraintsInCell;
+					//				const int end = start + numConstraintsInCell;
 
-					SolveTask task( bodyNative, shapeNative, constraintNative, start, numConstraintsInCell ,maxNumBatches,usedBodies,wgIdx,batchSizes,cellIdx);
+					SolveTask task(bodyNative, shapeNative, constraintNative, start, numConstraintsInCell, maxNumBatches, usedBodies, wgIdx, batchSizes, cellIdx);
 					task.m_solveFriction = false;
 					task.run(0);
-				
 				}
 			}
 		}
 
-		for(int iter=0; iter<m_nIterations; iter++)
+		for (int iter = 0; iter < m_nIterations; iter++)
 		{
-			for (int cellBatch=0;cellBatch<B3_SOLVER_N_BATCHES;cellBatch++)
+			for (int cellBatch = 0; cellBatch < B3_SOLVER_N_BATCHES; cellBatch++)
 			{
 				int nSplitX = B3_SOLVER_N_SPLIT_X;
 				int nSplitY = B3_SOLVER_N_SPLIT_Y;
-				
 
-				int numWorkgroups = B3_SOLVER_N_CELLS/B3_SOLVER_N_BATCHES;
+				int numWorkgroups = B3_SOLVER_N_CELLS / B3_SOLVER_N_BATCHES;
 
-				for (int wgIdx=0;wgIdx<numWorkgroups;wgIdx++)
+				for (int wgIdx = 0; wgIdx < numWorkgroups; wgIdx++)
 				{
-					int zIdx = (wgIdx/((nSplitX*nSplitY)/4))*2+((cellBatch&4)>>2);
-					int remain= (wgIdx%((nSplitX*nSplitY)/4));
-					int yIdx = (remain/(nSplitX/2))*2 + ((cellBatch&2)>>1);
-					int xIdx = (remain%(nSplitX/2))*2 + (cellBatch&1);
-					
-					int cellIdx = xIdx+yIdx*nSplitX+zIdx*(nSplitX*nSplitY);
-	
-					if( numConstraintsHost[cellIdx] == 0 ) 
+					int zIdx = (wgIdx / ((nSplitX * nSplitY) / 4)) * 2 + ((cellBatch & 4) >> 2);
+					int remain = (wgIdx % ((nSplitX * nSplitY) / 4));
+					int yIdx = (remain / (nSplitX / 2)) * 2 + ((cellBatch & 2) >> 1);
+					int xIdx = (remain % (nSplitX / 2)) * 2 + (cellBatch & 1);
+
+					int cellIdx = xIdx + yIdx * nSplitX + zIdx * (nSplitX * nSplitY);
+
+					if (numConstraintsHost[cellIdx] == 0)
 						continue;
-	
+
 					//printf("yIdx=%d\n",yIdx);
-					
+
 					const int start = offsetsHost[cellIdx];
 					int numConstraintsInCell = numConstraintsHost[cellIdx];
-	//				const int end = start + numConstraintsInCell;
+					//				const int end = start + numConstraintsInCell;
 
-					SolveTask task( bodyNative, shapeNative, constraintNative, start, numConstraintsInCell,maxNumBatches, 0,0,batchSizes,cellIdx);
+					SolveTask task(bodyNative, shapeNative, constraintNative, start, numConstraintsInCell, maxNumBatches, 0, 0, batchSizes, cellIdx);
 					task.m_solveFriction = true;
 					task.run(0);
-					
 				}
 			}
 		}
-
-
-	} else
+	}
+	else
 	{
-		for(int iter=0; iter<m_nIterations; iter++)
+		for (int iter = 0; iter < m_nIterations; iter++)
 		{
-			SolveTask task( bodyNative, shapeNative, constraintNative, 0, n ,maxNumBatches,0,0,0,0);
+			SolveTask task(bodyNative, shapeNative, constraintNative, 0, n, maxNumBatches, 0, 0, 0, 0);
 			task.m_solveFriction = false;
 			task.run(0);
 		}
 
-		for(int iter=0; iter<m_nIterations; iter++)
+		for (int iter = 0; iter < m_nIterations; iter++)
 		{
-			SolveTask task( bodyNative, shapeNative, constraintNative, 0, n ,maxNumBatches,0,0,0,0);
+			SolveTask task(bodyNative, shapeNative, constraintNative, 0, n, maxNumBatches, 0, 0, 0, 0);
 			task.m_solveFriction = true;
 			task.run(0);
 		}
@@ -688,23 +634,21 @@ void b3Solver::solveContactConstraintHost(  b3OpenCLArray<b3RigidBodyData>* body
 	shapeBuf->copyFromHost(shapeNative);
 	constraint->copyFromHost(constraintNative);
 	frame++;
-	
 }
 
 void checkConstraintBatch(const b3OpenCLArray<b3RigidBodyData>* bodyBuf,
-					const b3OpenCLArray<b3InertiaData>* shapeBuf,
-					b3OpenCLArray<b3GpuConstraint4>* constraint, 
-					b3OpenCLArray<unsigned int>* m_numConstraints,
-					b3OpenCLArray<unsigned int>* m_offsets,
-					int batchId
-					)
+						  const b3OpenCLArray<b3InertiaData>* shapeBuf,
+						  b3OpenCLArray<b3GpuConstraint4>* constraint,
+						  b3OpenCLArray<unsigned int>* m_numConstraints,
+						  b3OpenCLArray<unsigned int>* m_offsets,
+						  int batchId)
 {
-//						b3BufferInfoCL( m_numConstraints->getBufferCL() ), 
-//						b3BufferInfoCL( m_offsets->getBufferCL() ) 
-	
+	//						b3BufferInfoCL( m_numConstraints->getBufferCL() ),
+	//						b3BufferInfoCL( m_offsets->getBufferCL() )
+
 	int cellBatch = batchId;
 	const int nn = B3_SOLVER_N_CELLS;
-//	int numWorkItems = 64*nn/B3_SOLVER_N_BATCHES;
+	//	int numWorkItems = 64*nn/B3_SOLVER_N_BATCHES;
 
 	b3AlignedObjectArray<unsigned int> gN;
 	m_numConstraints->copyToHost(gN);
@@ -712,243 +656,220 @@ void checkConstraintBatch(const b3OpenCLArray<b3RigidBodyData>* bodyBuf,
 	m_offsets->copyToHost(gOffsets);
 	int nSplitX = B3_SOLVER_N_SPLIT_X;
 	int nSplitY = B3_SOLVER_N_SPLIT_Y;
-	
-//	int bIdx = batchId;
+
+	//	int bIdx = batchId;
 
 	b3AlignedObjectArray<b3GpuConstraint4> cpuConstraints;
 	constraint->copyToHost(cpuConstraints);
 
 	printf("batch = %d\n", batchId);
 
-	int numWorkgroups = nn/B3_SOLVER_N_BATCHES;
+	int numWorkgroups = nn / B3_SOLVER_N_BATCHES;
 	b3AlignedObjectArray<int> usedBodies;
 
-
-	for (int wgIdx=0;wgIdx<numWorkgroups;wgIdx++)
+	for (int wgIdx = 0; wgIdx < numWorkgroups; wgIdx++)
 	{
 		printf("wgIdx = %d           ", wgIdx);
 
-		int zIdx = (wgIdx/((nSplitX*nSplitY))/2)*2+((cellBatch&4)>>2);					
-		int remain = wgIdx%((nSplitX*nSplitY));
-		int yIdx = (remain%(nSplitX/2))*2 + ((cellBatch&2)>>1);
-		int xIdx = (remain/(nSplitX/2))*2 + (cellBatch&1);
+		int zIdx = (wgIdx / ((nSplitX * nSplitY)) / 2) * 2 + ((cellBatch & 4) >> 2);
+		int remain = wgIdx % ((nSplitX * nSplitY));
+		int yIdx = (remain % (nSplitX / 2)) * 2 + ((cellBatch & 2) >> 1);
+		int xIdx = (remain / (nSplitX / 2)) * 2 + (cellBatch & 1);
 
-		
-		int cellIdx = xIdx+yIdx*nSplitX+zIdx*(nSplitX*nSplitY);
-		printf("cellIdx=%d\n",cellIdx);
-		if( gN[cellIdx] == 0 ) 
+		int cellIdx = xIdx + yIdx * nSplitX + zIdx * (nSplitX * nSplitY);
+		printf("cellIdx=%d\n", cellIdx);
+		if (gN[cellIdx] == 0)
 			continue;
 
 		const int start = gOffsets[cellIdx];
 		const int end = start + gN[cellIdx];
 
-		for (int c=start;c<end;c++)
+		for (int c = start; c < end; c++)
 		{
 			b3GpuConstraint4& constraint = cpuConstraints[c];
 			//printf("constraint (%d,%d)\n", constraint.m_bodyA,constraint.m_bodyB);
-			if (usedBodies.findLinearSearch(constraint.m_bodyA)< usedBodies.size())
+			if (usedBodies.findLinearSearch(constraint.m_bodyA) < usedBodies.size())
 			{
 				printf("error?\n");
 			}
-			if (usedBodies.findLinearSearch(constraint.m_bodyB)< usedBodies.size())
+			if (usedBodies.findLinearSearch(constraint.m_bodyB) < usedBodies.size())
 			{
 				printf("error?\n");
 			}
 		}
 
-		for (int c=start;c<end;c++)
+		for (int c = start; c < end; c++)
 		{
 			b3GpuConstraint4& constraint = cpuConstraints[c];
 			usedBodies.push_back(constraint.m_bodyA);
 			usedBodies.push_back(constraint.m_bodyB);
 		}
-
 	}
 }
 
-static bool verify=false;
+static bool verify = false;
 
-void b3Solver::solveContactConstraint(  const b3OpenCLArray<b3RigidBodyData>* bodyBuf, const b3OpenCLArray<b3InertiaData>* shapeBuf, 
-			b3OpenCLArray<b3GpuConstraint4>* constraint, void* additionalData, int n ,int maxNumBatches)
+void b3Solver::solveContactConstraint(const b3OpenCLArray<b3RigidBodyData>* bodyBuf, const b3OpenCLArray<b3InertiaData>* shapeBuf,
+									  b3OpenCLArray<b3GpuConstraint4>* constraint, void* additionalData, int n, int maxNumBatches)
 {
-	
-	
-	b3Int4 cdata = b3MakeInt4( n, 0, 0, 0 );
+	b3Int4 cdata = b3MakeInt4(n, 0, 0, 0);
 	{
-		
 		const int nn = B3_SOLVER_N_CELLS;
 
 		cdata.x = 0;
-		cdata.y = maxNumBatches;//250;
+		cdata.y = maxNumBatches;  //250;
 
-
-		int numWorkItems = 64*nn/B3_SOLVER_N_BATCHES;
+		int numWorkItems = 64 * nn / B3_SOLVER_N_BATCHES;
 #ifdef DEBUG_ME
-		SolverDebugInfo* debugInfo = new  SolverDebugInfo[numWorkItems];
-		adl::b3OpenCLArray<SolverDebugInfo> gpuDebugInfo(data->m_device,numWorkItems);
+		SolverDebugInfo* debugInfo = new SolverDebugInfo[numWorkItems];
+		adl::b3OpenCLArray<SolverDebugInfo> gpuDebugInfo(data->m_device, numWorkItems);
 #endif
 
-
-
 		{
-
 			B3_PROFILE("m_batchSolveKernel iterations");
-			for(int iter=0; iter<m_nIterations; iter++)
+			for (int iter = 0; iter < m_nIterations; iter++)
 			{
-				for(int ib=0; ib<B3_SOLVER_N_BATCHES; ib++)
+				for (int ib = 0; ib < B3_SOLVER_N_BATCHES; ib++)
 				{
-					
 					if (verify)
 					{
-						checkConstraintBatch(bodyBuf,shapeBuf,constraint,m_numConstraints,m_offsets,ib);
+						checkConstraintBatch(bodyBuf, shapeBuf, constraint, m_numConstraints, m_offsets, ib);
 					}
 
 #ifdef DEBUG_ME
-					memset(debugInfo,0,sizeof(SolverDebugInfo)*numWorkItems);
-					gpuDebugInfo.write(debugInfo,numWorkItems);
+					memset(debugInfo, 0, sizeof(SolverDebugInfo) * numWorkItems);
+					gpuDebugInfo.write(debugInfo, numWorkItems);
 #endif
 
-
 					cdata.z = ib;
-					
 
-				b3LauncherCL launcher( m_queue, m_solveContactKernel ,"m_solveContactKernel");
+					b3LauncherCL launcher(m_queue, m_solveContactKernel, "m_solveContactKernel");
 #if 1
-                    
-					b3BufferInfoCL bInfo[] = { 
 
-						b3BufferInfoCL( bodyBuf->getBufferCL() ), 
-						b3BufferInfoCL( shapeBuf->getBufferCL() ), 
-						b3BufferInfoCL( constraint->getBufferCL() ),
-						b3BufferInfoCL( m_numConstraints->getBufferCL() ), 
-						b3BufferInfoCL( m_offsets->getBufferCL() ) 
+					b3BufferInfoCL bInfo[] = {
+
+						b3BufferInfoCL(bodyBuf->getBufferCL()),
+						b3BufferInfoCL(shapeBuf->getBufferCL()),
+						b3BufferInfoCL(constraint->getBufferCL()),
+						b3BufferInfoCL(m_numConstraints->getBufferCL()),
+						b3BufferInfoCL(m_offsets->getBufferCL())
 #ifdef DEBUG_ME
-						,	b3BufferInfoCL(&gpuDebugInfo)
+							,
+						b3BufferInfoCL(&gpuDebugInfo)
 #endif
-						};
+					};
 
-					
-
-                    launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
+					launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
 					//launcher.setConst(  cdata.x );
-                    launcher.setConst(  cdata.y );
-                    launcher.setConst(  cdata.z );
-                    b3Int4 nSplit;
+					launcher.setConst(cdata.y);
+					launcher.setConst(cdata.z);
+					b3Int4 nSplit;
 					nSplit.x = B3_SOLVER_N_SPLIT_X;
 					nSplit.y = B3_SOLVER_N_SPLIT_Y;
 					nSplit.z = B3_SOLVER_N_SPLIT_Z;
 
-                    launcher.setConst(  nSplit );
-                    launcher.launch1D( numWorkItems, 64 );
+					launcher.setConst(nSplit);
+					launcher.launch1D(numWorkItems, 64);
 
-                    
 #else
-                    const char* fileName = "m_batchSolveKernel.bin";
-                    FILE* f = fopen(fileName,"rb");
-                    if (f)
-                    {
-                        int sizeInBytes=0;
-                        if (fseek(f, 0, SEEK_END) || (sizeInBytes = ftell(f)) == EOF || fseek(f, 0, SEEK_SET))
-                        {
-                            printf("error, cannot get file size\n");
-                            exit(0);
-                        }
-                        
-                        unsigned char* buf = (unsigned char*) malloc(sizeInBytes);
-                        fread(buf,sizeInBytes,1,f);
-                        int serializedBytes = launcher.deserializeArgs(buf, sizeInBytes,m_context);
-                        int num = *(int*)&buf[serializedBytes];
-                        
-                        launcher.launch1D( num);
+					const char* fileName = "m_batchSolveKernel.bin";
+					FILE* f = fopen(fileName, "rb");
+					if (f)
+					{
+						int sizeInBytes = 0;
+						if (fseek(f, 0, SEEK_END) || (sizeInBytes = ftell(f)) == EOF || fseek(f, 0, SEEK_SET))
+						{
+							printf("error, cannot get file size\n");
+							exit(0);
+						}
 
-                        //this clFinish is for testing on errors
-                        clFinish(m_queue);
-                    }
+						unsigned char* buf = (unsigned char*)malloc(sizeInBytes);
+						fread(buf, sizeInBytes, 1, f);
+						int serializedBytes = launcher.deserializeArgs(buf, sizeInBytes, m_context);
+						int num = *(int*)&buf[serializedBytes];
+
+						launcher.launch1D(num);
+
+						//this clFinish is for testing on errors
+						clFinish(m_queue);
+					}
 
 #endif
-					
 
 #ifdef DEBUG_ME
 					clFinish(m_queue);
-					gpuDebugInfo.read(debugInfo,numWorkItems);
+					gpuDebugInfo.read(debugInfo, numWorkItems);
 					clFinish(m_queue);
-					for (int i=0;i<numWorkItems;i++)
+					for (int i = 0; i < numWorkItems; i++)
 					{
-						if (debugInfo[i].m_valInt2>0)
+						if (debugInfo[i].m_valInt2 > 0)
 						{
-							printf("debugInfo[i].m_valInt2 = %d\n",i,debugInfo[i].m_valInt2);
+							printf("debugInfo[i].m_valInt2 = %d\n", i, debugInfo[i].m_valInt2);
 						}
 
-						if (debugInfo[i].m_valInt3>0)
+						if (debugInfo[i].m_valInt3 > 0)
 						{
-							printf("debugInfo[i].m_valInt3 = %d\n",i,debugInfo[i].m_valInt3);
+							printf("debugInfo[i].m_valInt3 = %d\n", i, debugInfo[i].m_valInt3);
 						}
 					}
-#endif //DEBUG_ME
-
-
+#endif  //DEBUG_ME
 				}
 			}
-		
+
 			clFinish(m_queue);
-
-
 		}
 
 		cdata.x = 1;
-		bool applyFriction=true;
+		bool applyFriction = true;
 		if (applyFriction)
-    	{
+		{
 			B3_PROFILE("m_batchSolveKernel iterations2");
-			for(int iter=0; iter<m_nIterations; iter++)
+			for (int iter = 0; iter < m_nIterations; iter++)
 			{
-				for(int ib=0; ib<B3_SOLVER_N_BATCHES; ib++)
+				for (int ib = 0; ib < B3_SOLVER_N_BATCHES; ib++)
 				{
 					cdata.z = ib;
-					
 
-					b3BufferInfoCL bInfo[] = { 
-						b3BufferInfoCL( bodyBuf->getBufferCL() ), 
-						b3BufferInfoCL( shapeBuf->getBufferCL() ), 
-						b3BufferInfoCL( constraint->getBufferCL() ),
-						b3BufferInfoCL( m_numConstraints->getBufferCL() ), 
-						b3BufferInfoCL( m_offsets->getBufferCL() )
+					b3BufferInfoCL bInfo[] = {
+						b3BufferInfoCL(bodyBuf->getBufferCL()),
+						b3BufferInfoCL(shapeBuf->getBufferCL()),
+						b3BufferInfoCL(constraint->getBufferCL()),
+						b3BufferInfoCL(m_numConstraints->getBufferCL()),
+						b3BufferInfoCL(m_offsets->getBufferCL())
 #ifdef DEBUG_ME
-						,b3BufferInfoCL(&gpuDebugInfo)
-#endif //DEBUG_ME
+							,
+						b3BufferInfoCL(&gpuDebugInfo)
+#endif  //DEBUG_ME
 					};
-					b3LauncherCL launcher( m_queue, m_solveFrictionKernel,"m_solveFrictionKernel" );
-					launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
+					b3LauncherCL launcher(m_queue, m_solveFrictionKernel, "m_solveFrictionKernel");
+					launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
 					//launcher.setConst(  cdata.x );
-                    launcher.setConst(  cdata.y );
-                    launcher.setConst(  cdata.z );
-                    b3Int4 nSplit;
+					launcher.setConst(cdata.y);
+					launcher.setConst(cdata.z);
+					b3Int4 nSplit;
 					nSplit.x = B3_SOLVER_N_SPLIT_X;
 					nSplit.y = B3_SOLVER_N_SPLIT_Y;
 					nSplit.z = B3_SOLVER_N_SPLIT_Z;
 
-                    launcher.setConst(  nSplit );
-                    
-					launcher.launch1D( 64*nn/B3_SOLVER_N_BATCHES, 64 );
+					launcher.setConst(nSplit);
+
+					launcher.launch1D(64 * nn / B3_SOLVER_N_BATCHES, 64);
 				}
 			}
 			clFinish(m_queue);
-			
 		}
 #ifdef DEBUG_ME
 		delete[] debugInfo;
-#endif //DEBUG_ME
+#endif  //DEBUG_ME
 	}
-
-	
 }
 
-void b3Solver::convertToConstraints( const b3OpenCLArray<b3RigidBodyData>* bodyBuf, 
-	const b3OpenCLArray<b3InertiaData>* shapeBuf, 
-	b3OpenCLArray<b3Contact4>* contactsIn, b3OpenCLArray<b3GpuConstraint4>* contactCOut, void* additionalData, 
-	int nContacts, const ConstraintCfg& cfg )
+void b3Solver::convertToConstraints(const b3OpenCLArray<b3RigidBodyData>* bodyBuf,
+									const b3OpenCLArray<b3InertiaData>* shapeBuf,
+									b3OpenCLArray<b3Contact4>* contactsIn, b3OpenCLArray<b3GpuConstraint4>* contactCOut, void* additionalData,
+									int nContacts, const ConstraintCfg& cfg)
 {
-//	b3OpenCLArray<b3GpuConstraint4>* constraintNative =0;
+	//	b3OpenCLArray<b3GpuConstraint4>* constraintNative =0;
 	contactCOut->resize(nContacts);
 	struct CB
 	{
@@ -959,30 +880,28 @@ void b3Solver::convertToConstraints( const b3OpenCLArray<b3RigidBodyData>* bodyB
 	};
 
 	{
-
 		CB cdata;
 		cdata.m_nContacts = nContacts;
 		cdata.m_dt = cfg.m_dt;
 		cdata.m_positionDrift = cfg.m_positionDrift;
 		cdata.m_positionConstraintCoeff = cfg.m_positionConstraintCoeff;
 
-		
 		if (gConvertConstraintOnCpu)
 		{
 			b3AlignedObjectArray<b3RigidBodyData> gBodies;
-		bodyBuf->copyToHost(gBodies);
+			bodyBuf->copyToHost(gBodies);
 
-		b3AlignedObjectArray<b3Contact4> gContact;
-		contactsIn->copyToHost(gContact);
+			b3AlignedObjectArray<b3Contact4> gContact;
+			contactsIn->copyToHost(gContact);
+
+			b3AlignedObjectArray<b3InertiaData> gShapes;
+			shapeBuf->copyToHost(gShapes);
+
+			b3AlignedObjectArray<b3GpuConstraint4> gConstraintOut;
+			gConstraintOut.resize(nContacts);
 
-		b3AlignedObjectArray<b3InertiaData> gShapes;
-		shapeBuf->copyToHost(gShapes);
-		
-		b3AlignedObjectArray<b3GpuConstraint4> gConstraintOut;
-		gConstraintOut.resize(nContacts);
-		
 			B3_PROFILE("cpu contactToConstraintKernel");
-			for (int gIdx=0;gIdx<nContacts;gIdx++)
+			for (int gIdx = 0; gIdx < nContacts; gIdx++)
 			{
 				int aIdx = abs(gContact[gIdx].m_bodyAPtrAndSignBit);
 				int bIdx = abs(gContact[gIdx].m_bodyBPtrAndSignBit);
@@ -1001,40 +920,36 @@ void b3Solver::convertToConstraints( const b3OpenCLArray<b3RigidBodyData>* bodyB
 
 				b3ContactConstraint4_t cs;
 
-    			setConstraint4( posA, linVelA, angVelA, invMassA, invInertiaA, posB, linVelB, angVelB, invMassB, invInertiaB,
-					&gContact[gIdx], cdata.m_dt, cdata.m_positionDrift, cdata.m_positionConstraintCoeff,
-					&cs );
-		
+				setConstraint4(posA, linVelA, angVelA, invMassA, invInertiaA, posB, linVelB, angVelB, invMassB, invInertiaB,
+							   &gContact[gIdx], cdata.m_dt, cdata.m_positionDrift, cdata.m_positionConstraintCoeff,
+							   &cs);
+
 				cs.m_batchIdx = gContact[gIdx].m_batchIdx;
 
 				gConstraintOut[gIdx] = (b3GpuConstraint4&)cs;
 			}
 
 			contactCOut->copyFromHost(gConstraintOut);
-
-		} else
+		}
+		else
 		{
 			B3_PROFILE("gpu m_contactToConstraintKernel");
 
-		
-			b3BufferInfoCL bInfo[] = { b3BufferInfoCL( contactsIn->getBufferCL() ), b3BufferInfoCL( bodyBuf->getBufferCL() ), b3BufferInfoCL( shapeBuf->getBufferCL()),
-				b3BufferInfoCL( contactCOut->getBufferCL() )};
-			b3LauncherCL launcher( m_queue, m_contactToConstraintKernel,"m_contactToConstraintKernel" );
-			launcher.setBuffers( bInfo, sizeof(bInfo)/sizeof(b3BufferInfoCL) );
+			b3BufferInfoCL bInfo[] = {b3BufferInfoCL(contactsIn->getBufferCL()), b3BufferInfoCL(bodyBuf->getBufferCL()), b3BufferInfoCL(shapeBuf->getBufferCL()),
+									  b3BufferInfoCL(contactCOut->getBufferCL())};
+			b3LauncherCL launcher(m_queue, m_contactToConstraintKernel, "m_contactToConstraintKernel");
+			launcher.setBuffers(bInfo, sizeof(bInfo) / sizeof(b3BufferInfoCL));
 			//launcher.setConst(  cdata );
-        
+
 			launcher.setConst(cdata.m_nContacts);
 			launcher.setConst(cdata.m_dt);
 			launcher.setConst(cdata.m_positionDrift);
 			launcher.setConst(cdata.m_positionConstraintCoeff);
-        
-			launcher.launch1D( nContacts, 64 );	
-			clFinish(m_queue);
 
+			launcher.launch1D(nContacts, 64);
+			clFinish(m_queue);
 		}
 	}
-
-	
 }
 
 /*
@@ -1115,28 +1030,24 @@ void b3Solver::sortContacts(  const b3OpenCLArray<b3RigidBodyData>* bodyBuf,
 }
 
 */
-void	b3Solver::batchContacts(  b3OpenCLArray<b3Contact4>* contacts, int nContacts, b3OpenCLArray<unsigned int>* nNative, b3OpenCLArray<unsigned int>* offsetsNative, int staticIdx )
+void b3Solver::batchContacts(b3OpenCLArray<b3Contact4>* contacts, int nContacts, b3OpenCLArray<unsigned int>* nNative, b3OpenCLArray<unsigned int>* offsetsNative, int staticIdx)
 {
-	
-	int numWorkItems = 64*B3_SOLVER_N_CELLS;
+	int numWorkItems = 64 * B3_SOLVER_N_CELLS;
 	{
 		B3_PROFILE("batch generation");
-		
+
 		b3Int4 cdata;
 		cdata.x = nContacts;
 		cdata.y = 0;
 		cdata.z = staticIdx;
 
-		
 #ifdef BATCH_DEBUG
-		SolverDebugInfo* debugInfo = new  SolverDebugInfo[numWorkItems];
-		adl::b3OpenCLArray<SolverDebugInfo> gpuDebugInfo(data->m_device,numWorkItems);
-		memset(debugInfo,0,sizeof(SolverDebugInfo)*numWorkItems);
-		gpuDebugInfo.write(debugInfo,numWorkItems);
+		SolverDebugInfo* debugInfo = new SolverDebugInfo[numWorkItems];
+		adl::b3OpenCLArray<SolverDebugInfo> gpuDebugInfo(data->m_device, numWorkItems);
+		memset(debugInfo, 0, sizeof(SolverDebugInfo) * numWorkItems);
+		gpuDebugInfo.write(debugInfo, numWorkItems);
 #endif
 
-		
-
 #if 0
 		b3BufferInfoCL bInfo[] = { 
 			b3BufferInfoCL( contacts->getBufferCL() ), 
@@ -1148,8 +1059,6 @@ void	b3Solver::batchContacts(  b3OpenCLArray<b3Contact4>* contacts, int nContact
 #endif
 		};
 #endif
-		
-		
 
 		{
 			m_batchSizes.resize(nNative->size());
@@ -1157,22 +1066,21 @@ void	b3Solver::batchContacts(  b3OpenCLArray<b3Contact4>* contacts, int nContact
 			//b3LauncherCL launcher( m_queue, m_batchingKernel);
 			cl_kernel k = useNewBatchingKernel ? m_batchingKernelNew : m_batchingKernel;
 
-			b3LauncherCL launcher( m_queue, k,"*batchingKernel");
-			if (!useNewBatchingKernel )
+			b3LauncherCL launcher(m_queue, k, "*batchingKernel");
+			if (!useNewBatchingKernel)
 			{
-				launcher.setBuffer( contacts->getBufferCL() );
+				launcher.setBuffer(contacts->getBufferCL());
 			}
-			launcher.setBuffer( m_contactBuffer2->getBufferCL() );
-			launcher.setBuffer( nNative->getBufferCL());
-			launcher.setBuffer( offsetsNative->getBufferCL());
-			
+			launcher.setBuffer(m_contactBuffer2->getBufferCL());
+			launcher.setBuffer(nNative->getBufferCL());
+			launcher.setBuffer(offsetsNative->getBufferCL());
+
 			launcher.setBuffer(m_batchSizes.getBufferCL());
-			
 
 			//launcher.setConst(  cdata );
-            launcher.setConst(staticIdx);
-            
-			launcher.launch1D( numWorkItems, 64 );
+			launcher.setConst(staticIdx);
+
+			launcher.launch1D(numWorkItems, 64);
 			//clFinish(m_queue);
 			//b3AlignedObjectArray<int> batchSizesCPU;
 			//m_batchSizes.copyToHost(batchSizesCPU);
@@ -1180,46 +1088,41 @@ void	b3Solver::batchContacts(  b3OpenCLArray<b3Contact4>* contacts, int nContact
 		}
 
 #ifdef BATCH_DEBUG
-	aaaa
-		b3Contact4* hostContacts = new b3Contact4[nContacts];
-		m_contactBuffer->read(hostContacts,nContacts);
+		aaaa
+			b3Contact4* hostContacts = new b3Contact4[nContacts];
+		m_contactBuffer->read(hostContacts, nContacts);
 		clFinish(m_queue);
 
-		gpuDebugInfo.read(debugInfo,numWorkItems);
+		gpuDebugInfo.read(debugInfo, numWorkItems);
 		clFinish(m_queue);
 
-		for (int i=0;i<numWorkItems;i++)
+		for (int i = 0; i < numWorkItems; i++)
 		{
-			if (debugInfo[i].m_valInt1>0)
+			if (debugInfo[i].m_valInt1 > 0)
 			{
 				printf("catch\n");
 			}
-			if (debugInfo[i].m_valInt2>0)
+			if (debugInfo[i].m_valInt2 > 0)
 			{
 				printf("catch22\n");
 			}
 
-			if (debugInfo[i].m_valInt3>0)
+			if (debugInfo[i].m_valInt3 > 0)
 			{
 				printf("catch666\n");
 			}
 
-			if (debugInfo[i].m_valInt4>0)
+			if (debugInfo[i].m_valInt4 > 0)
 			{
 				printf("catch777\n");
 			}
 		}
 		delete[] debugInfo;
-#endif //BATCH_DEBUG
-
+#endif  //BATCH_DEBUG
 	}
 
-//	copy buffer to buffer
+	//	copy buffer to buffer
 	//b3Assert(m_contactBuffer->size()==nContacts);
 	//contacts->copyFromOpenCLArray( *m_contactBuffer);
 	//clFinish(m_queue);//needed?
-	
-	
-	
 }
-
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3Solver.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3Solver.h
index b37f2f1bece..ee63531d78d 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3Solver.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/b3Solver.h
@@ -13,7 +13,6 @@ subject to the following restrictions:
 */
 //Originally written by Takahiro Harada
 
-
 #ifndef __ADL_SOLVER_H
 #define __ADL_SOLVER_H
 
@@ -29,98 +28,83 @@ subject to the following restrictions:
 
 #include "Bullet3OpenCL/Initialize/b3OpenCLUtils.h"
 
-
-#define B3NEXTMULTIPLEOF(num, alignment) (((num)/(alignment) + (((num)%(alignment)==0)?0:1))*(alignment))
+#define B3NEXTMULTIPLEOF(num, alignment) (((num) / (alignment) + (((num) % (alignment) == 0) ? 0 : 1)) * (alignment))
 
 enum
 {
-	B3_SOLVER_N_SPLIT_X = 8,//16,//4,
-	B3_SOLVER_N_SPLIT_Y = 4,//16,//4,
-	B3_SOLVER_N_SPLIT_Z = 8,//,
-	B3_SOLVER_N_CELLS = B3_SOLVER_N_SPLIT_X*B3_SOLVER_N_SPLIT_Y*B3_SOLVER_N_SPLIT_Z,
-	B3_SOLVER_N_BATCHES = 8,//4,//8,//4,
+	B3_SOLVER_N_SPLIT_X = 8,  //16,//4,
+	B3_SOLVER_N_SPLIT_Y = 4,  //16,//4,
+	B3_SOLVER_N_SPLIT_Z = 8,  //,
+	B3_SOLVER_N_CELLS = B3_SOLVER_N_SPLIT_X * B3_SOLVER_N_SPLIT_Y * B3_SOLVER_N_SPLIT_Z,
+	B3_SOLVER_N_BATCHES = 8,  //4,//8,//4,
 	B3_MAX_NUM_BATCHES = 128,
 };
 
 class b3SolverBase
 {
-	public:
-		
+public:
+	struct ConstraintCfg
+	{
+		ConstraintCfg(float dt = 0.f) : m_positionDrift(0.005f), m_positionConstraintCoeff(0.2f), m_dt(dt), m_staticIdx(-1) {}
 
-		struct ConstraintCfg
-		{
-			ConstraintCfg( float dt = 0.f ): m_positionDrift( 0.005f ), m_positionConstraintCoeff( 0.2f ), m_dt(dt), m_staticIdx(-1) {}
-
-			float m_positionDrift;
-			float m_positionConstraintCoeff;
-			float m_dt;
-			bool m_enableParallelSolve;
-			float m_batchCellSize;
-			int m_staticIdx;
-		};
-		
+		float m_positionDrift;
+		float m_positionConstraintCoeff;
+		float m_dt;
+		bool m_enableParallelSolve;
+		float m_batchCellSize;
+		int m_staticIdx;
+	};
 };
 
 class b3Solver : public b3SolverBase
 {
-	public:
+public:
+	cl_context m_context;
+	cl_device_id m_device;
+	cl_command_queue m_queue;
 
-		cl_context m_context;
-		cl_device_id m_device;
-		cl_command_queue m_queue;
-				
+	b3OpenCLArray<unsigned int>* m_numConstraints;
+	b3OpenCLArray<unsigned int>* m_offsets;
+	b3OpenCLArray<int> m_batchSizes;
 
-		b3OpenCLArray<unsigned int>* m_numConstraints;
-		b3OpenCLArray<unsigned int>* m_offsets;
-		b3OpenCLArray<int> m_batchSizes;
-		
-		
-		int m_nIterations;
-		cl_kernel m_batchingKernel;
-		cl_kernel m_batchingKernelNew;
-		cl_kernel m_solveContactKernel;
-		cl_kernel m_solveFrictionKernel;
-		cl_kernel m_contactToConstraintKernel;
-		cl_kernel m_setSortDataKernel;
-		cl_kernel m_reorderContactKernel;
-		cl_kernel m_copyConstraintKernel;
+	int m_nIterations;
+	cl_kernel m_batchingKernel;
+	cl_kernel m_batchingKernelNew;
+	cl_kernel m_solveContactKernel;
+	cl_kernel m_solveFrictionKernel;
+	cl_kernel m_contactToConstraintKernel;
+	cl_kernel m_setSortDataKernel;
+	cl_kernel m_reorderContactKernel;
+	cl_kernel m_copyConstraintKernel;
 
-		class b3RadixSort32CL*	m_sort32;
-		class b3BoundSearchCL*	m_search;
-		class b3PrefixScanCL*	m_scan;
+	class b3RadixSort32CL* m_sort32;
+	class b3BoundSearchCL* m_search;
+	class b3PrefixScanCL* m_scan;
 
-		b3OpenCLArray<b3SortData>* m_sortDataBuffer;
-		b3OpenCLArray<b3Contact4>* m_contactBuffer2;
+	b3OpenCLArray<b3SortData>* m_sortDataBuffer;
+	b3OpenCLArray<b3Contact4>* m_contactBuffer2;
 
-		enum
-		{
-			DYNAMIC_CONTACT_ALLOCATION_THRESHOLD = 2000000,
-		};
+	enum
+	{
+		DYNAMIC_CONTACT_ALLOCATION_THRESHOLD = 2000000,
+	};
 
-		
+	b3Solver(cl_context ctx, cl_device_id device, cl_command_queue queue, int pairCapacity);
 
-		
-		b3Solver(cl_context ctx, cl_device_id device, cl_command_queue queue, int pairCapacity);
+	virtual ~b3Solver();
 
-		virtual ~b3Solver();
-		
-		void solveContactConstraint( const b3OpenCLArray<b3RigidBodyData>* bodyBuf, const b3OpenCLArray<b3InertiaData>* inertiaBuf, 
-			b3OpenCLArray<b3GpuConstraint4>* constraint, void* additionalData, int n ,int maxNumBatches);
+	void solveContactConstraint(const b3OpenCLArray<b3RigidBodyData>* bodyBuf, const b3OpenCLArray<b3InertiaData>* inertiaBuf,
+								b3OpenCLArray<b3GpuConstraint4>* constraint, void* additionalData, int n, int maxNumBatches);
 
-		void solveContactConstraintHost(  b3OpenCLArray<b3RigidBodyData>* bodyBuf, b3OpenCLArray<b3InertiaData>* shapeBuf, 
-			b3OpenCLArray<b3GpuConstraint4>* constraint, void* additionalData, int n ,int maxNumBatches, b3AlignedObjectArray<int>* batchSizes);
+	void solveContactConstraintHost(b3OpenCLArray<b3RigidBodyData>* bodyBuf, b3OpenCLArray<b3InertiaData>* shapeBuf,
+									b3OpenCLArray<b3GpuConstraint4>* constraint, void* additionalData, int n, int maxNumBatches, b3AlignedObjectArray<int>* batchSizes);
 
+	void convertToConstraints(const b3OpenCLArray<b3RigidBodyData>* bodyBuf,
+							  const b3OpenCLArray<b3InertiaData>* shapeBuf,
+							  b3OpenCLArray<b3Contact4>* contactsIn, b3OpenCLArray<b3GpuConstraint4>* contactCOut, void* additionalData,
+							  int nContacts, const ConstraintCfg& cfg);
 
-		void convertToConstraints( const b3OpenCLArray<b3RigidBodyData>* bodyBuf, 
-			const b3OpenCLArray<b3InertiaData>* shapeBuf, 
-			b3OpenCLArray<b3Contact4>* contactsIn, b3OpenCLArray<b3GpuConstraint4>* contactCOut, void* additionalData, 
-			int nContacts, const ConstraintCfg& cfg );
-
-		void	batchContacts( b3OpenCLArray<b3Contact4>* contacts, int nContacts, b3OpenCLArray<unsigned int>* n, b3OpenCLArray<unsigned int>* offsets, int staticIdx );
-
+	void batchContacts(b3OpenCLArray<b3Contact4>* contacts, int nContacts, b3OpenCLArray<unsigned int>* n, b3OpenCLArray<unsigned int>* offsets, int staticIdx);
 };
 
-
-
-
-#endif //__ADL_SOLVER_H
+#endif  //__ADL_SOLVER_H
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/batchingKernels.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/batchingKernels.h
index 150eedc94b1..7c73c96baa0 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/batchingKernels.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/batchingKernels.h
@@ -1,388 +1,387 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* batchingKernelsCL= \
-"/*\n"
-"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
-"This software is provided 'as-is', without any express or implied warranty.\n"
-"In no event will the authors be held liable for any damages arising from the use of this software.\n"
-"Permission is granted to anyone to use this software for any purpose, \n"
-"including commercial applications, and to alter it and redistribute it freely, \n"
-"subject to the following restrictions:\n"
-"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
-"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
-"3. This notice may not be removed or altered from any source distribution.\n"
-"*/\n"
-"//Originally written by Takahiro Harada\n"
-"#ifndef B3_CONTACT4DATA_H\n"
-"#define B3_CONTACT4DATA_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#define B3_FLOAT4_H\n"
-"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
-"#define B3_PLATFORM_DEFINITIONS_H\n"
-"struct MyTest\n"
-"{\n"
-"	int bla;\n"
-"};\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
-"#define B3_LARGE_FLOAT 1e18f\n"
-"#define B3_INFINITY 1e18f\n"
-"#define b3Assert(a)\n"
-"#define b3ConstArray(a) __global const a*\n"
-"#define b3AtomicInc atomic_inc\n"
-"#define b3AtomicAdd atomic_add\n"
-"#define b3Fabs fabs\n"
-"#define b3Sqrt native_sqrt\n"
-"#define b3Sin native_sin\n"
-"#define b3Cos native_cos\n"
-"#define B3_STATIC\n"
-"#endif\n"
-"#endif\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"	typedef float4	b3Float4;\n"
-"	#define b3Float4ConstArg const b3Float4\n"
-"	#define b3MakeFloat4 (float4)\n"
-"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return dot(a1, b1);\n"
-"	}\n"
-"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return cross(a1, b1);\n"
-"	}\n"
-"	#define b3MinFloat4 min\n"
-"	#define b3MaxFloat4 max\n"
-"	#define b3Normalized(a) normalize(a)\n"
-"#endif \n"
-"		\n"
-"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
-"{\n"
-"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
-"		return false;\n"
-"	return true;\n"
-"}\n"
-"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
-"{\n"
-"    float maxDot = -B3_INFINITY;\n"
-"    int i = 0;\n"
-"    int ptIndex = -1;\n"
-"    for( i = 0; i < vecLen; i++ )\n"
-"    {\n"
-"        float dot = b3Dot3F4(vecArray[i],vec);\n"
-"            \n"
-"        if( dot > maxDot )\n"
-"        {\n"
-"            maxDot = dot;\n"
-"            ptIndex = i;\n"
-"        }\n"
-"    }\n"
-"	b3Assert(ptIndex>=0);\n"
-"    if (ptIndex<0)\n"
-"	{\n"
-"		ptIndex = 0;\n"
-"	}\n"
-"    *dotOut = maxDot;\n"
-"    return ptIndex;\n"
-"}\n"
-"#endif //B3_FLOAT4_H\n"
-"typedef  struct b3Contact4Data b3Contact4Data_t;\n"
-"struct b3Contact4Data\n"
-"{\n"
-"	b3Float4	m_worldPosB[4];\n"
-"//	b3Float4	m_localPosA[4];\n"
-"//	b3Float4	m_localPosB[4];\n"
-"	b3Float4	m_worldNormalOnB;	//	w: m_nPoints\n"
-"	unsigned short  m_restituitionCoeffCmp;\n"
-"	unsigned short  m_frictionCoeffCmp;\n"
-"	int m_batchIdx;\n"
-"	int m_bodyAPtrAndSignBit;//x:m_bodyAPtr, y:m_bodyBPtr\n"
-"	int m_bodyBPtrAndSignBit;\n"
-"	int	m_childIndexA;\n"
-"	int	m_childIndexB;\n"
-"	int m_unused1;\n"
-"	int m_unused2;\n"
-"};\n"
-"inline int b3Contact4Data_getNumPoints(const struct b3Contact4Data* contact)\n"
-"{\n"
-"	return (int)contact->m_worldNormalOnB.w;\n"
-"};\n"
-"inline void b3Contact4Data_setNumPoints(struct b3Contact4Data* contact, int numPoints)\n"
-"{\n"
-"	contact->m_worldNormalOnB.w = (float)numPoints;\n"
-"};\n"
-"#endif //B3_CONTACT4DATA_H\n"
-"#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_extended_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics : enable\n"
-"#ifdef cl_ext_atomic_counters_32\n"
-"#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
-"#else\n"
-"#define counter32_t volatile __global int*\n"
-"#endif\n"
-"typedef unsigned int u32;\n"
-"typedef unsigned short u16;\n"
-"typedef unsigned char u8;\n"
-"#define GET_GROUP_IDX get_group_id(0)\n"
-"#define GET_LOCAL_IDX get_local_id(0)\n"
-"#define GET_GLOBAL_IDX get_global_id(0)\n"
-"#define GET_GROUP_SIZE get_local_size(0)\n"
-"#define GET_NUM_GROUPS get_num_groups(0)\n"
-"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
-"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
-"#define AtomInc(x) atom_inc(&(x))\n"
-"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
-"#define AppendInc(x, out) out = atomic_inc(x)\n"
-"#define AtomAdd(x, value) atom_add(&(x), value)\n"
-"#define AtomCmpxhg(x, cmp, value) atom_cmpxchg( &(x), cmp, value )\n"
-"#define AtomXhg(x, value) atom_xchg ( &(x), value )\n"
-"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
-"#define make_float4 (float4)\n"
-"#define make_float2 (float2)\n"
-"#define make_uint4 (uint4)\n"
-"#define make_int4 (int4)\n"
-"#define make_uint2 (uint2)\n"
-"#define make_int2 (int2)\n"
-"#define max2 max\n"
-"#define min2 min\n"
-"#define WG_SIZE 64\n"
-"typedef struct \n"
-"{\n"
-"	int m_n;\n"
-"	int m_start;\n"
-"	int m_staticIdx;\n"
-"	int m_paddings[1];\n"
-"} ConstBuffer;\n"
-"typedef struct \n"
-"{\n"
-"	int m_a;\n"
-"	int m_b;\n"
-"	u32 m_idx;\n"
-"}Elem;\n"
-"#define STACK_SIZE (WG_SIZE*10)\n"
-"//#define STACK_SIZE (WG_SIZE)\n"
-"#define RING_SIZE 1024\n"
-"#define RING_SIZE_MASK (RING_SIZE-1)\n"
-"#define CHECK_SIZE (WG_SIZE)\n"
-"#define GET_RING_CAPACITY (RING_SIZE - ldsRingEnd)\n"
-"#define RING_END ldsTmp\n"
-"u32 readBuf(__local u32* buff, int idx)\n"
-"{\n"
-"	idx = idx % (32*CHECK_SIZE);\n"
-"	int bitIdx = idx%32;\n"
-"	int bufIdx = idx/32;\n"
-"	return buff[bufIdx] & (1<<bitIdx);\n"
-"}\n"
-"void writeBuf(__local u32* buff, int idx)\n"
-"{\n"
-"	idx = idx % (32*CHECK_SIZE);\n"
-"	int bitIdx = idx%32;\n"
-"	int bufIdx = idx/32;\n"
-"//	buff[bufIdx] |= (1<<bitIdx);\n"
-"	atom_or( &buff[bufIdx], (1<<bitIdx) );\n"
-"}\n"
-"u32 tryWrite(__local u32* buff, int idx)\n"
-"{\n"
-"	idx = idx % (32*CHECK_SIZE);\n"
-"	int bitIdx = idx%32;\n"
-"	int bufIdx = idx/32;\n"
-"	u32 ans = (u32)atom_or( &buff[bufIdx], (1<<bitIdx) );\n"
-"	return ((ans >> bitIdx)&1) == 0;\n"
-"}\n"
-"//	batching on the GPU\n"
-"__kernel void CreateBatches( __global const struct b3Contact4Data* gConstraints, __global struct b3Contact4Data* gConstraintsOut,\n"
-"		__global const u32* gN, __global const u32* gStart, __global int* batchSizes, \n"
-"		int m_staticIdx )\n"
-"{\n"
-"	__local u32 ldsStackIdx[STACK_SIZE];\n"
-"	__local u32 ldsStackEnd;\n"
-"	__local Elem ldsRingElem[RING_SIZE];\n"
-"	__local u32 ldsRingEnd;\n"
-"	__local u32 ldsTmp;\n"
-"	__local u32 ldsCheckBuffer[CHECK_SIZE];\n"
-"	__local u32 ldsFixedBuffer[CHECK_SIZE];\n"
-"	__local u32 ldsGEnd;\n"
-"	__local u32 ldsDstEnd;\n"
-"	int wgIdx = GET_GROUP_IDX;\n"
-"	int lIdx = GET_LOCAL_IDX;\n"
-"	\n"
-"	const int m_n = gN[wgIdx];\n"
-"	const int m_start = gStart[wgIdx];\n"
-"		\n"
-"	if( lIdx == 0 )\n"
-"	{\n"
-"		ldsRingEnd = 0;\n"
-"		ldsGEnd = 0;\n"
-"		ldsStackEnd = 0;\n"
-"		ldsDstEnd = m_start;\n"
-"	}\n"
-"	\n"
-"	\n"
-"	\n"
-"//	while(1)\n"
-"//was 250\n"
-"	int ie=0;\n"
-"	int maxBatch = 0;\n"
-"	for(ie=0; ie<50; ie++)\n"
-"	{\n"
-"		ldsFixedBuffer[lIdx] = 0;\n"
-"		for(int giter=0; giter<4; giter++)\n"
-"		{\n"
-"			int ringCap = GET_RING_CAPACITY;\n"
-"		\n"
-"			//	1. fill ring\n"
-"			if( ldsGEnd < m_n )\n"
-"			{\n"
-"				while( ringCap > WG_SIZE )\n"
-"				{\n"
-"					if( ldsGEnd >= m_n ) break;\n"
-"					if( lIdx < ringCap - WG_SIZE )\n"
-"					{\n"
-"						int srcIdx;\n"
-"						AtomInc1( ldsGEnd, srcIdx );\n"
-"						if( srcIdx < m_n )\n"
-"						{\n"
-"							int dstIdx;\n"
-"							AtomInc1( ldsRingEnd, dstIdx );\n"
-"							\n"
-"							int a = gConstraints[m_start+srcIdx].m_bodyAPtrAndSignBit;\n"
-"							int b = gConstraints[m_start+srcIdx].m_bodyBPtrAndSignBit;\n"
-"							ldsRingElem[dstIdx].m_a = (a>b)? b:a;\n"
-"							ldsRingElem[dstIdx].m_b = (a>b)? a:b;\n"
-"							ldsRingElem[dstIdx].m_idx = srcIdx;\n"
-"						}\n"
-"					}\n"
-"					ringCap = GET_RING_CAPACITY;\n"
-"				}\n"
-"			}\n"
-"			GROUP_LDS_BARRIER;\n"
-"	\n"
-"			//	2. fill stack\n"
-"			__local Elem* dst = ldsRingElem;\n"
-"			if( lIdx == 0 ) RING_END = 0;\n"
-"			int srcIdx=lIdx;\n"
-"			int end = ldsRingEnd;\n"
-"			{\n"
-"				for(int ii=0; ii<end; ii+=WG_SIZE, srcIdx+=WG_SIZE)\n"
-"				{\n"
-"					Elem e;\n"
-"					if(srcIdx<end) e = ldsRingElem[srcIdx];\n"
-"					bool done = (srcIdx<end)?false:true;\n"
-"					for(int i=lIdx; i<CHECK_SIZE; i+=WG_SIZE) ldsCheckBuffer[lIdx] = 0;\n"
-"					\n"
-"					if( !done )\n"
-"					{\n"
-"						int aUsed = readBuf( ldsFixedBuffer, abs(e.m_a));\n"
-"						int bUsed = readBuf( ldsFixedBuffer, abs(e.m_b));\n"
-"						if( aUsed==0 && bUsed==0 )\n"
-"						{\n"
-"							int aAvailable=1;\n"
-"							int bAvailable=1;\n"
-"							int ea = abs(e.m_a);\n"
-"							int eb = abs(e.m_b);\n"
-"							bool aStatic = (e.m_a<0) ||(ea==m_staticIdx);\n"
-"							bool bStatic = (e.m_b<0) ||(eb==m_staticIdx);\n"
-"							\n"
-"							if (!aStatic)\n"
-"								aAvailable = tryWrite( ldsCheckBuffer, ea );\n"
-"							if (!bStatic)\n"
-"								bAvailable = tryWrite( ldsCheckBuffer, eb );\n"
-"							\n"
-"							//aAvailable = aStatic? 1: aAvailable;\n"
-"							//bAvailable = bStatic? 1: bAvailable;\n"
-"							bool success = (aAvailable && bAvailable);\n"
-"							if(success)\n"
-"							{\n"
-"							\n"
-"								if (!aStatic)\n"
-"									writeBuf( ldsFixedBuffer, ea );\n"
-"								if (!bStatic)\n"
-"									writeBuf( ldsFixedBuffer, eb );\n"
-"							}\n"
-"							done = success;\n"
-"						}\n"
-"					}\n"
-"					//	put it aside\n"
-"					if(srcIdx<end)\n"
-"					{\n"
-"						if( done )\n"
-"						{\n"
-"							int dstIdx; AtomInc1( ldsStackEnd, dstIdx );\n"
-"							if( dstIdx < STACK_SIZE )\n"
-"								ldsStackIdx[dstIdx] = e.m_idx;\n"
-"							else{\n"
-"								done = false;\n"
-"								AtomAdd( ldsStackEnd, -1 );\n"
-"							}\n"
-"						}\n"
-"						if( !done )\n"
-"						{\n"
-"							int dstIdx; AtomInc1( RING_END, dstIdx );\n"
-"							dst[dstIdx] = e;\n"
-"						}\n"
-"					}\n"
-"					//	if filled, flush\n"
-"					if( ldsStackEnd == STACK_SIZE )\n"
-"					{\n"
-"						for(int i=lIdx; i<STACK_SIZE; i+=WG_SIZE)\n"
-"						{\n"
-"							int idx = m_start + ldsStackIdx[i];\n"
-"							int dstIdx; AtomInc1( ldsDstEnd, dstIdx );\n"
-"							gConstraintsOut[ dstIdx ] = gConstraints[ idx ];\n"
-"							gConstraintsOut[ dstIdx ].m_batchIdx = ie;\n"
-"						}\n"
-"						if( lIdx == 0 ) ldsStackEnd = 0;\n"
-"						//for(int i=lIdx; i<CHECK_SIZE; i+=WG_SIZE) \n"
-"						ldsFixedBuffer[lIdx] = 0;\n"
-"					}\n"
-"				}\n"
-"			}\n"
-"			if( lIdx == 0 ) ldsRingEnd = RING_END;\n"
-"		}\n"
-"		GROUP_LDS_BARRIER;\n"
-"		for(int i=lIdx; i<ldsStackEnd; i+=WG_SIZE)\n"
-"		{\n"
-"			int idx = m_start + ldsStackIdx[i];\n"
-"			int dstIdx; AtomInc1( ldsDstEnd, dstIdx );\n"
-"			gConstraintsOut[ dstIdx ] = gConstraints[ idx ];\n"
-"			gConstraintsOut[ dstIdx ].m_batchIdx = ie;\n"
-"		}\n"
-"		//	in case it couldn't consume any pair. Flush them\n"
-"		//	todo. Serial batch worth while?\n"
-"		if( ldsStackEnd == 0 )\n"
-"		{\n"
-"			for(int i=lIdx; i<ldsRingEnd; i+=WG_SIZE)\n"
-"			{\n"
-"				int idx = m_start + ldsRingElem[i].m_idx;\n"
-"				int dstIdx; AtomInc1( ldsDstEnd, dstIdx );\n"
-"				gConstraintsOut[ dstIdx ] = gConstraints[ idx ];\n"
-"				int curBatch = 100+i;\n"
-"				if (maxBatch < curBatch)\n"
-"					maxBatch = curBatch;\n"
-"				\n"
-"				gConstraintsOut[ dstIdx ].m_batchIdx = curBatch;\n"
-"				\n"
-"			}\n"
-"			GROUP_LDS_BARRIER;\n"
-"			if( lIdx == 0 ) ldsRingEnd = 0;\n"
-"		}\n"
-"		if( lIdx == 0 ) ldsStackEnd = 0;\n"
-"		GROUP_LDS_BARRIER;\n"
-"		//	termination\n"
-"		if( ldsGEnd == m_n && ldsRingEnd == 0 )\n"
-"			break;\n"
-"	}\n"
-"	if( lIdx == 0 )\n"
-"	{\n"
-"		if (maxBatch < ie)\n"
-"			maxBatch=ie;\n"
-"		batchSizes[wgIdx]=maxBatch;\n"
-"	}\n"
-"}\n"
-;
+static const char* batchingKernelsCL =
+	"/*\n"
+	"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
+	"This software is provided 'as-is', without any express or implied warranty.\n"
+	"In no event will the authors be held liable for any damages arising from the use of this software.\n"
+	"Permission is granted to anyone to use this software for any purpose, \n"
+	"including commercial applications, and to alter it and redistribute it freely, \n"
+	"subject to the following restrictions:\n"
+	"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
+	"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
+	"3. This notice may not be removed or altered from any source distribution.\n"
+	"*/\n"
+	"//Originally written by Takahiro Harada\n"
+	"#ifndef B3_CONTACT4DATA_H\n"
+	"#define B3_CONTACT4DATA_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#define B3_FLOAT4_H\n"
+	"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
+	"#define B3_PLATFORM_DEFINITIONS_H\n"
+	"struct MyTest\n"
+	"{\n"
+	"	int bla;\n"
+	"};\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
+	"#define B3_LARGE_FLOAT 1e18f\n"
+	"#define B3_INFINITY 1e18f\n"
+	"#define b3Assert(a)\n"
+	"#define b3ConstArray(a) __global const a*\n"
+	"#define b3AtomicInc atomic_inc\n"
+	"#define b3AtomicAdd atomic_add\n"
+	"#define b3Fabs fabs\n"
+	"#define b3Sqrt native_sqrt\n"
+	"#define b3Sin native_sin\n"
+	"#define b3Cos native_cos\n"
+	"#define B3_STATIC\n"
+	"#endif\n"
+	"#endif\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"	typedef float4	b3Float4;\n"
+	"	#define b3Float4ConstArg const b3Float4\n"
+	"	#define b3MakeFloat4 (float4)\n"
+	"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return dot(a1, b1);\n"
+	"	}\n"
+	"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return cross(a1, b1);\n"
+	"	}\n"
+	"	#define b3MinFloat4 min\n"
+	"	#define b3MaxFloat4 max\n"
+	"	#define b3Normalized(a) normalize(a)\n"
+	"#endif \n"
+	"		\n"
+	"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
+	"{\n"
+	"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
+	"		return false;\n"
+	"	return true;\n"
+	"}\n"
+	"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
+	"{\n"
+	"    float maxDot = -B3_INFINITY;\n"
+	"    int i = 0;\n"
+	"    int ptIndex = -1;\n"
+	"    for( i = 0; i < vecLen; i++ )\n"
+	"    {\n"
+	"        float dot = b3Dot3F4(vecArray[i],vec);\n"
+	"            \n"
+	"        if( dot > maxDot )\n"
+	"        {\n"
+	"            maxDot = dot;\n"
+	"            ptIndex = i;\n"
+	"        }\n"
+	"    }\n"
+	"	b3Assert(ptIndex>=0);\n"
+	"    if (ptIndex<0)\n"
+	"	{\n"
+	"		ptIndex = 0;\n"
+	"	}\n"
+	"    *dotOut = maxDot;\n"
+	"    return ptIndex;\n"
+	"}\n"
+	"#endif //B3_FLOAT4_H\n"
+	"typedef  struct b3Contact4Data b3Contact4Data_t;\n"
+	"struct b3Contact4Data\n"
+	"{\n"
+	"	b3Float4	m_worldPosB[4];\n"
+	"//	b3Float4	m_localPosA[4];\n"
+	"//	b3Float4	m_localPosB[4];\n"
+	"	b3Float4	m_worldNormalOnB;	//	w: m_nPoints\n"
+	"	unsigned short  m_restituitionCoeffCmp;\n"
+	"	unsigned short  m_frictionCoeffCmp;\n"
+	"	int m_batchIdx;\n"
+	"	int m_bodyAPtrAndSignBit;//x:m_bodyAPtr, y:m_bodyBPtr\n"
+	"	int m_bodyBPtrAndSignBit;\n"
+	"	int	m_childIndexA;\n"
+	"	int	m_childIndexB;\n"
+	"	int m_unused1;\n"
+	"	int m_unused2;\n"
+	"};\n"
+	"inline int b3Contact4Data_getNumPoints(const struct b3Contact4Data* contact)\n"
+	"{\n"
+	"	return (int)contact->m_worldNormalOnB.w;\n"
+	"};\n"
+	"inline void b3Contact4Data_setNumPoints(struct b3Contact4Data* contact, int numPoints)\n"
+	"{\n"
+	"	contact->m_worldNormalOnB.w = (float)numPoints;\n"
+	"};\n"
+	"#endif //B3_CONTACT4DATA_H\n"
+	"#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_extended_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics : enable\n"
+	"#ifdef cl_ext_atomic_counters_32\n"
+	"#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
+	"#else\n"
+	"#define counter32_t volatile __global int*\n"
+	"#endif\n"
+	"typedef unsigned int u32;\n"
+	"typedef unsigned short u16;\n"
+	"typedef unsigned char u8;\n"
+	"#define GET_GROUP_IDX get_group_id(0)\n"
+	"#define GET_LOCAL_IDX get_local_id(0)\n"
+	"#define GET_GLOBAL_IDX get_global_id(0)\n"
+	"#define GET_GROUP_SIZE get_local_size(0)\n"
+	"#define GET_NUM_GROUPS get_num_groups(0)\n"
+	"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
+	"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
+	"#define AtomInc(x) atom_inc(&(x))\n"
+	"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
+	"#define AppendInc(x, out) out = atomic_inc(x)\n"
+	"#define AtomAdd(x, value) atom_add(&(x), value)\n"
+	"#define AtomCmpxhg(x, cmp, value) atom_cmpxchg( &(x), cmp, value )\n"
+	"#define AtomXhg(x, value) atom_xchg ( &(x), value )\n"
+	"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
+	"#define make_float4 (float4)\n"
+	"#define make_float2 (float2)\n"
+	"#define make_uint4 (uint4)\n"
+	"#define make_int4 (int4)\n"
+	"#define make_uint2 (uint2)\n"
+	"#define make_int2 (int2)\n"
+	"#define max2 max\n"
+	"#define min2 min\n"
+	"#define WG_SIZE 64\n"
+	"typedef struct \n"
+	"{\n"
+	"	int m_n;\n"
+	"	int m_start;\n"
+	"	int m_staticIdx;\n"
+	"	int m_paddings[1];\n"
+	"} ConstBuffer;\n"
+	"typedef struct \n"
+	"{\n"
+	"	int m_a;\n"
+	"	int m_b;\n"
+	"	u32 m_idx;\n"
+	"}Elem;\n"
+	"#define STACK_SIZE (WG_SIZE*10)\n"
+	"//#define STACK_SIZE (WG_SIZE)\n"
+	"#define RING_SIZE 1024\n"
+	"#define RING_SIZE_MASK (RING_SIZE-1)\n"
+	"#define CHECK_SIZE (WG_SIZE)\n"
+	"#define GET_RING_CAPACITY (RING_SIZE - ldsRingEnd)\n"
+	"#define RING_END ldsTmp\n"
+	"u32 readBuf(__local u32* buff, int idx)\n"
+	"{\n"
+	"	idx = idx % (32*CHECK_SIZE);\n"
+	"	int bitIdx = idx%32;\n"
+	"	int bufIdx = idx/32;\n"
+	"	return buff[bufIdx] & (1<<bitIdx);\n"
+	"}\n"
+	"void writeBuf(__local u32* buff, int idx)\n"
+	"{\n"
+	"	idx = idx % (32*CHECK_SIZE);\n"
+	"	int bitIdx = idx%32;\n"
+	"	int bufIdx = idx/32;\n"
+	"//	buff[bufIdx] |= (1<<bitIdx);\n"
+	"	atom_or( &buff[bufIdx], (1<<bitIdx) );\n"
+	"}\n"
+	"u32 tryWrite(__local u32* buff, int idx)\n"
+	"{\n"
+	"	idx = idx % (32*CHECK_SIZE);\n"
+	"	int bitIdx = idx%32;\n"
+	"	int bufIdx = idx/32;\n"
+	"	u32 ans = (u32)atom_or( &buff[bufIdx], (1<<bitIdx) );\n"
+	"	return ((ans >> bitIdx)&1) == 0;\n"
+	"}\n"
+	"//	batching on the GPU\n"
+	"__kernel void CreateBatches( __global const struct b3Contact4Data* gConstraints, __global struct b3Contact4Data* gConstraintsOut,\n"
+	"		__global const u32* gN, __global const u32* gStart, __global int* batchSizes, \n"
+	"		int m_staticIdx )\n"
+	"{\n"
+	"	__local u32 ldsStackIdx[STACK_SIZE];\n"
+	"	__local u32 ldsStackEnd;\n"
+	"	__local Elem ldsRingElem[RING_SIZE];\n"
+	"	__local u32 ldsRingEnd;\n"
+	"	__local u32 ldsTmp;\n"
+	"	__local u32 ldsCheckBuffer[CHECK_SIZE];\n"
+	"	__local u32 ldsFixedBuffer[CHECK_SIZE];\n"
+	"	__local u32 ldsGEnd;\n"
+	"	__local u32 ldsDstEnd;\n"
+	"	int wgIdx = GET_GROUP_IDX;\n"
+	"	int lIdx = GET_LOCAL_IDX;\n"
+	"	\n"
+	"	const int m_n = gN[wgIdx];\n"
+	"	const int m_start = gStart[wgIdx];\n"
+	"		\n"
+	"	if( lIdx == 0 )\n"
+	"	{\n"
+	"		ldsRingEnd = 0;\n"
+	"		ldsGEnd = 0;\n"
+	"		ldsStackEnd = 0;\n"
+	"		ldsDstEnd = m_start;\n"
+	"	}\n"
+	"	\n"
+	"	\n"
+	"	\n"
+	"//	while(1)\n"
+	"//was 250\n"
+	"	int ie=0;\n"
+	"	int maxBatch = 0;\n"
+	"	for(ie=0; ie<50; ie++)\n"
+	"	{\n"
+	"		ldsFixedBuffer[lIdx] = 0;\n"
+	"		for(int giter=0; giter<4; giter++)\n"
+	"		{\n"
+	"			int ringCap = GET_RING_CAPACITY;\n"
+	"		\n"
+	"			//	1. fill ring\n"
+	"			if( ldsGEnd < m_n )\n"
+	"			{\n"
+	"				while( ringCap > WG_SIZE )\n"
+	"				{\n"
+	"					if( ldsGEnd >= m_n ) break;\n"
+	"					if( lIdx < ringCap - WG_SIZE )\n"
+	"					{\n"
+	"						int srcIdx;\n"
+	"						AtomInc1( ldsGEnd, srcIdx );\n"
+	"						if( srcIdx < m_n )\n"
+	"						{\n"
+	"							int dstIdx;\n"
+	"							AtomInc1( ldsRingEnd, dstIdx );\n"
+	"							\n"
+	"							int a = gConstraints[m_start+srcIdx].m_bodyAPtrAndSignBit;\n"
+	"							int b = gConstraints[m_start+srcIdx].m_bodyBPtrAndSignBit;\n"
+	"							ldsRingElem[dstIdx].m_a = (a>b)? b:a;\n"
+	"							ldsRingElem[dstIdx].m_b = (a>b)? a:b;\n"
+	"							ldsRingElem[dstIdx].m_idx = srcIdx;\n"
+	"						}\n"
+	"					}\n"
+	"					ringCap = GET_RING_CAPACITY;\n"
+	"				}\n"
+	"			}\n"
+	"			GROUP_LDS_BARRIER;\n"
+	"	\n"
+	"			//	2. fill stack\n"
+	"			__local Elem* dst = ldsRingElem;\n"
+	"			if( lIdx == 0 ) RING_END = 0;\n"
+	"			int srcIdx=lIdx;\n"
+	"			int end = ldsRingEnd;\n"
+	"			{\n"
+	"				for(int ii=0; ii<end; ii+=WG_SIZE, srcIdx+=WG_SIZE)\n"
+	"				{\n"
+	"					Elem e;\n"
+	"					if(srcIdx<end) e = ldsRingElem[srcIdx];\n"
+	"					bool done = (srcIdx<end)?false:true;\n"
+	"					for(int i=lIdx; i<CHECK_SIZE; i+=WG_SIZE) ldsCheckBuffer[lIdx] = 0;\n"
+	"					\n"
+	"					if( !done )\n"
+	"					{\n"
+	"						int aUsed = readBuf( ldsFixedBuffer, abs(e.m_a));\n"
+	"						int bUsed = readBuf( ldsFixedBuffer, abs(e.m_b));\n"
+	"						if( aUsed==0 && bUsed==0 )\n"
+	"						{\n"
+	"							int aAvailable=1;\n"
+	"							int bAvailable=1;\n"
+	"							int ea = abs(e.m_a);\n"
+	"							int eb = abs(e.m_b);\n"
+	"							bool aStatic = (e.m_a<0) ||(ea==m_staticIdx);\n"
+	"							bool bStatic = (e.m_b<0) ||(eb==m_staticIdx);\n"
+	"							\n"
+	"							if (!aStatic)\n"
+	"								aAvailable = tryWrite( ldsCheckBuffer, ea );\n"
+	"							if (!bStatic)\n"
+	"								bAvailable = tryWrite( ldsCheckBuffer, eb );\n"
+	"							\n"
+	"							//aAvailable = aStatic? 1: aAvailable;\n"
+	"							//bAvailable = bStatic? 1: bAvailable;\n"
+	"							bool success = (aAvailable && bAvailable);\n"
+	"							if(success)\n"
+	"							{\n"
+	"							\n"
+	"								if (!aStatic)\n"
+	"									writeBuf( ldsFixedBuffer, ea );\n"
+	"								if (!bStatic)\n"
+	"									writeBuf( ldsFixedBuffer, eb );\n"
+	"							}\n"
+	"							done = success;\n"
+	"						}\n"
+	"					}\n"
+	"					//	put it aside\n"
+	"					if(srcIdx<end)\n"
+	"					{\n"
+	"						if( done )\n"
+	"						{\n"
+	"							int dstIdx; AtomInc1( ldsStackEnd, dstIdx );\n"
+	"							if( dstIdx < STACK_SIZE )\n"
+	"								ldsStackIdx[dstIdx] = e.m_idx;\n"
+	"							else{\n"
+	"								done = false;\n"
+	"								AtomAdd( ldsStackEnd, -1 );\n"
+	"							}\n"
+	"						}\n"
+	"						if( !done )\n"
+	"						{\n"
+	"							int dstIdx; AtomInc1( RING_END, dstIdx );\n"
+	"							dst[dstIdx] = e;\n"
+	"						}\n"
+	"					}\n"
+	"					//	if filled, flush\n"
+	"					if( ldsStackEnd == STACK_SIZE )\n"
+	"					{\n"
+	"						for(int i=lIdx; i<STACK_SIZE; i+=WG_SIZE)\n"
+	"						{\n"
+	"							int idx = m_start + ldsStackIdx[i];\n"
+	"							int dstIdx; AtomInc1( ldsDstEnd, dstIdx );\n"
+	"							gConstraintsOut[ dstIdx ] = gConstraints[ idx ];\n"
+	"							gConstraintsOut[ dstIdx ].m_batchIdx = ie;\n"
+	"						}\n"
+	"						if( lIdx == 0 ) ldsStackEnd = 0;\n"
+	"						//for(int i=lIdx; i<CHECK_SIZE; i+=WG_SIZE) \n"
+	"						ldsFixedBuffer[lIdx] = 0;\n"
+	"					}\n"
+	"				}\n"
+	"			}\n"
+	"			if( lIdx == 0 ) ldsRingEnd = RING_END;\n"
+	"		}\n"
+	"		GROUP_LDS_BARRIER;\n"
+	"		for(int i=lIdx; i<ldsStackEnd; i+=WG_SIZE)\n"
+	"		{\n"
+	"			int idx = m_start + ldsStackIdx[i];\n"
+	"			int dstIdx; AtomInc1( ldsDstEnd, dstIdx );\n"
+	"			gConstraintsOut[ dstIdx ] = gConstraints[ idx ];\n"
+	"			gConstraintsOut[ dstIdx ].m_batchIdx = ie;\n"
+	"		}\n"
+	"		//	in case it couldn't consume any pair. Flush them\n"
+	"		//	todo. Serial batch worth while?\n"
+	"		if( ldsStackEnd == 0 )\n"
+	"		{\n"
+	"			for(int i=lIdx; i<ldsRingEnd; i+=WG_SIZE)\n"
+	"			{\n"
+	"				int idx = m_start + ldsRingElem[i].m_idx;\n"
+	"				int dstIdx; AtomInc1( ldsDstEnd, dstIdx );\n"
+	"				gConstraintsOut[ dstIdx ] = gConstraints[ idx ];\n"
+	"				int curBatch = 100+i;\n"
+	"				if (maxBatch < curBatch)\n"
+	"					maxBatch = curBatch;\n"
+	"				\n"
+	"				gConstraintsOut[ dstIdx ].m_batchIdx = curBatch;\n"
+	"				\n"
+	"			}\n"
+	"			GROUP_LDS_BARRIER;\n"
+	"			if( lIdx == 0 ) ldsRingEnd = 0;\n"
+	"		}\n"
+	"		if( lIdx == 0 ) ldsStackEnd = 0;\n"
+	"		GROUP_LDS_BARRIER;\n"
+	"		//	termination\n"
+	"		if( ldsGEnd == m_n && ldsRingEnd == 0 )\n"
+	"			break;\n"
+	"	}\n"
+	"	if( lIdx == 0 )\n"
+	"	{\n"
+	"		if (maxBatch < ie)\n"
+	"			maxBatch=ie;\n"
+	"		batchSizes[wgIdx]=maxBatch;\n"
+	"	}\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/batchingKernelsNew.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/batchingKernelsNew.h
index 1e5957adae5..05800656cb1 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/batchingKernelsNew.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/batchingKernelsNew.h
@@ -1,291 +1,290 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* batchingKernelsNewCL= \
-"/*\n"
-"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
-"This software is provided 'as-is', without any express or implied warranty.\n"
-"In no event will the authors be held liable for any damages arising from the use of this software.\n"
-"Permission is granted to anyone to use this software for any purpose, \n"
-"including commercial applications, and to alter it and redistribute it freely, \n"
-"subject to the following restrictions:\n"
-"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
-"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
-"3. This notice may not be removed or altered from any source distribution.\n"
-"*/\n"
-"//Originally written by Erwin Coumans\n"
-"#ifndef B3_CONTACT4DATA_H\n"
-"#define B3_CONTACT4DATA_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#define B3_FLOAT4_H\n"
-"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
-"#define B3_PLATFORM_DEFINITIONS_H\n"
-"struct MyTest\n"
-"{\n"
-"	int bla;\n"
-"};\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
-"#define B3_LARGE_FLOAT 1e18f\n"
-"#define B3_INFINITY 1e18f\n"
-"#define b3Assert(a)\n"
-"#define b3ConstArray(a) __global const a*\n"
-"#define b3AtomicInc atomic_inc\n"
-"#define b3AtomicAdd atomic_add\n"
-"#define b3Fabs fabs\n"
-"#define b3Sqrt native_sqrt\n"
-"#define b3Sin native_sin\n"
-"#define b3Cos native_cos\n"
-"#define B3_STATIC\n"
-"#endif\n"
-"#endif\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"	typedef float4	b3Float4;\n"
-"	#define b3Float4ConstArg const b3Float4\n"
-"	#define b3MakeFloat4 (float4)\n"
-"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return dot(a1, b1);\n"
-"	}\n"
-"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return cross(a1, b1);\n"
-"	}\n"
-"	#define b3MinFloat4 min\n"
-"	#define b3MaxFloat4 max\n"
-"	#define b3Normalized(a) normalize(a)\n"
-"#endif \n"
-"		\n"
-"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
-"{\n"
-"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
-"		return false;\n"
-"	return true;\n"
-"}\n"
-"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
-"{\n"
-"    float maxDot = -B3_INFINITY;\n"
-"    int i = 0;\n"
-"    int ptIndex = -1;\n"
-"    for( i = 0; i < vecLen; i++ )\n"
-"    {\n"
-"        float dot = b3Dot3F4(vecArray[i],vec);\n"
-"            \n"
-"        if( dot > maxDot )\n"
-"        {\n"
-"            maxDot = dot;\n"
-"            ptIndex = i;\n"
-"        }\n"
-"    }\n"
-"	b3Assert(ptIndex>=0);\n"
-"    if (ptIndex<0)\n"
-"	{\n"
-"		ptIndex = 0;\n"
-"	}\n"
-"    *dotOut = maxDot;\n"
-"    return ptIndex;\n"
-"}\n"
-"#endif //B3_FLOAT4_H\n"
-"typedef  struct b3Contact4Data b3Contact4Data_t;\n"
-"struct b3Contact4Data\n"
-"{\n"
-"	b3Float4	m_worldPosB[4];\n"
-"//	b3Float4	m_localPosA[4];\n"
-"//	b3Float4	m_localPosB[4];\n"
-"	b3Float4	m_worldNormalOnB;	//	w: m_nPoints\n"
-"	unsigned short  m_restituitionCoeffCmp;\n"
-"	unsigned short  m_frictionCoeffCmp;\n"
-"	int m_batchIdx;\n"
-"	int m_bodyAPtrAndSignBit;//x:m_bodyAPtr, y:m_bodyBPtr\n"
-"	int m_bodyBPtrAndSignBit;\n"
-"	int	m_childIndexA;\n"
-"	int	m_childIndexB;\n"
-"	int m_unused1;\n"
-"	int m_unused2;\n"
-"};\n"
-"inline int b3Contact4Data_getNumPoints(const struct b3Contact4Data* contact)\n"
-"{\n"
-"	return (int)contact->m_worldNormalOnB.w;\n"
-"};\n"
-"inline void b3Contact4Data_setNumPoints(struct b3Contact4Data* contact, int numPoints)\n"
-"{\n"
-"	contact->m_worldNormalOnB.w = (float)numPoints;\n"
-"};\n"
-"#endif //B3_CONTACT4DATA_H\n"
-"#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_extended_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics : enable\n"
-"#ifdef cl_ext_atomic_counters_32\n"
-"#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
-"#else\n"
-"#define counter32_t volatile __global int*\n"
-"#endif\n"
-"#define SIMD_WIDTH 64\n"
-"typedef unsigned int u32;\n"
-"typedef unsigned short u16;\n"
-"typedef unsigned char u8;\n"
-"#define GET_GROUP_IDX get_group_id(0)\n"
-"#define GET_LOCAL_IDX get_local_id(0)\n"
-"#define GET_GLOBAL_IDX get_global_id(0)\n"
-"#define GET_GROUP_SIZE get_local_size(0)\n"
-"#define GET_NUM_GROUPS get_num_groups(0)\n"
-"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
-"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
-"#define AtomInc(x) atom_inc(&(x))\n"
-"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
-"#define AppendInc(x, out) out = atomic_inc(x)\n"
-"#define AtomAdd(x, value) atom_add(&(x), value)\n"
-"#define AtomCmpxhg(x, cmp, value) atom_cmpxchg( &(x), cmp, value )\n"
-"#define AtomXhg(x, value) atom_xchg ( &(x), value )\n"
-"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
-"#define make_float4 (float4)\n"
-"#define make_float2 (float2)\n"
-"#define make_uint4 (uint4)\n"
-"#define make_int4 (int4)\n"
-"#define make_uint2 (uint2)\n"
-"#define make_int2 (int2)\n"
-"#define max2 max\n"
-"#define min2 min\n"
-"#define WG_SIZE 64\n"
-"typedef struct \n"
-"{\n"
-"	int m_n;\n"
-"	int m_start;\n"
-"	int m_staticIdx;\n"
-"	int m_paddings[1];\n"
-"} ConstBuffer;\n"
-"typedef struct \n"
-"{\n"
-"	int m_a;\n"
-"	int m_b;\n"
-"	u32 m_idx;\n"
-"}Elem;\n"
-"//	batching on the GPU\n"
-"__kernel void CreateBatchesBruteForce( __global struct b3Contact4Data* gConstraints, 	__global const u32* gN, __global const u32* gStart, int m_staticIdx )\n"
-"{\n"
-"	int wgIdx = GET_GROUP_IDX;\n"
-"	int lIdx = GET_LOCAL_IDX;\n"
-"	\n"
-"	const int m_n = gN[wgIdx];\n"
-"	const int m_start = gStart[wgIdx];\n"
-"		\n"
-"	if( lIdx == 0 )\n"
-"	{\n"
-"		for (int i=0;i<m_n;i++)\n"
-"		{\n"
-"			int srcIdx = i+m_start;\n"
-"			int batchIndex = i;\n"
-"			gConstraints[ srcIdx ].m_batchIdx = batchIndex;	\n"
-"		}\n"
-"	}\n"
-"}\n"
-"#define CHECK_SIZE (WG_SIZE)\n"
-"u32 readBuf(__local u32* buff, int idx)\n"
-"{\n"
-"	idx = idx % (32*CHECK_SIZE);\n"
-"	int bitIdx = idx%32;\n"
-"	int bufIdx = idx/32;\n"
-"	return buff[bufIdx] & (1<<bitIdx);\n"
-"}\n"
-"void writeBuf(__local u32* buff, int idx)\n"
-"{\n"
-"	idx = idx % (32*CHECK_SIZE);\n"
-"	int bitIdx = idx%32;\n"
-"	int bufIdx = idx/32;\n"
-"	buff[bufIdx] |= (1<<bitIdx);\n"
-"	//atom_or( &buff[bufIdx], (1<<bitIdx) );\n"
-"}\n"
-"u32 tryWrite(__local u32* buff, int idx)\n"
-"{\n"
-"	idx = idx % (32*CHECK_SIZE);\n"
-"	int bitIdx = idx%32;\n"
-"	int bufIdx = idx/32;\n"
-"	u32 ans = (u32)atom_or( &buff[bufIdx], (1<<bitIdx) );\n"
-"	return ((ans >> bitIdx)&1) == 0;\n"
-"}\n"
-"//	batching on the GPU\n"
-"__kernel void CreateBatchesNew( __global struct b3Contact4Data* gConstraints, __global const u32* gN, __global const u32* gStart, __global int* batchSizes, int staticIdx )\n"
-"{\n"
-"	int wgIdx = GET_GROUP_IDX;\n"
-"	int lIdx = GET_LOCAL_IDX;\n"
-"	const int numConstraints = gN[wgIdx];\n"
-"	const int m_start = gStart[wgIdx];\n"
-"	b3Contact4Data_t tmp;\n"
-"	\n"
-"	__local u32 ldsFixedBuffer[CHECK_SIZE];\n"
-"		\n"
-"	\n"
-"	\n"
-"	\n"
-"	\n"
-"	if( lIdx == 0 )\n"
-"	{\n"
-"	\n"
-"		\n"
-"		__global struct b3Contact4Data* cs = &gConstraints[m_start];	\n"
-"	\n"
-"		\n"
-"		int numValidConstraints = 0;\n"
-"		int batchIdx = 0;\n"
-"		while( numValidConstraints < numConstraints)\n"
-"		{\n"
-"			int nCurrentBatch = 0;\n"
-"			//	clear flag\n"
-"	\n"
-"			for(int i=0; i<CHECK_SIZE; i++) \n"
-"				ldsFixedBuffer[i] = 0;		\n"
-"			for(int i=numValidConstraints; i<numConstraints; i++)\n"
-"			{\n"
-"				int bodyAS = cs[i].m_bodyAPtrAndSignBit;\n"
-"				int bodyBS = cs[i].m_bodyBPtrAndSignBit;\n"
-"				int bodyA = abs(bodyAS);\n"
-"				int bodyB = abs(bodyBS);\n"
-"				bool aIsStatic = (bodyAS<0) || bodyAS==staticIdx;\n"
-"				bool bIsStatic = (bodyBS<0) || bodyBS==staticIdx;\n"
-"				int aUnavailable = aIsStatic ? 0 : readBuf( ldsFixedBuffer, bodyA);\n"
-"				int bUnavailable = bIsStatic ? 0 : readBuf( ldsFixedBuffer, bodyB);\n"
-"				\n"
-"				if( aUnavailable==0 && bUnavailable==0 ) // ok\n"
-"				{\n"
-"					if (!aIsStatic)\n"
-"					{\n"
-"						writeBuf( ldsFixedBuffer, bodyA );\n"
-"					}\n"
-"					if (!bIsStatic)\n"
-"					{\n"
-"						writeBuf( ldsFixedBuffer, bodyB );\n"
-"					}\n"
-"					cs[i].m_batchIdx = batchIdx;\n"
-"					if (i!=numValidConstraints)\n"
-"					{\n"
-"						tmp = cs[i];\n"
-"						cs[i] = cs[numValidConstraints];\n"
-"						cs[numValidConstraints]  = tmp;\n"
-"					}\n"
-"					numValidConstraints++;\n"
-"					\n"
-"					nCurrentBatch++;\n"
-"					if( nCurrentBatch == SIMD_WIDTH)\n"
-"					{\n"
-"						nCurrentBatch = 0;\n"
-"						for(int i=0; i<CHECK_SIZE; i++) \n"
-"							ldsFixedBuffer[i] = 0;\n"
-"						\n"
-"					}\n"
-"				}\n"
-"			}//for\n"
-"			batchIdx ++;\n"
-"		}//while\n"
-"		\n"
-"		batchSizes[wgIdx] = batchIdx;\n"
-"	}//if( lIdx == 0 )\n"
-"	\n"
-"	//return batchIdx;\n"
-"}\n"
-;
+static const char* batchingKernelsNewCL =
+	"/*\n"
+	"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
+	"This software is provided 'as-is', without any express or implied warranty.\n"
+	"In no event will the authors be held liable for any damages arising from the use of this software.\n"
+	"Permission is granted to anyone to use this software for any purpose, \n"
+	"including commercial applications, and to alter it and redistribute it freely, \n"
+	"subject to the following restrictions:\n"
+	"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
+	"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
+	"3. This notice may not be removed or altered from any source distribution.\n"
+	"*/\n"
+	"//Originally written by Erwin Coumans\n"
+	"#ifndef B3_CONTACT4DATA_H\n"
+	"#define B3_CONTACT4DATA_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#define B3_FLOAT4_H\n"
+	"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
+	"#define B3_PLATFORM_DEFINITIONS_H\n"
+	"struct MyTest\n"
+	"{\n"
+	"	int bla;\n"
+	"};\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
+	"#define B3_LARGE_FLOAT 1e18f\n"
+	"#define B3_INFINITY 1e18f\n"
+	"#define b3Assert(a)\n"
+	"#define b3ConstArray(a) __global const a*\n"
+	"#define b3AtomicInc atomic_inc\n"
+	"#define b3AtomicAdd atomic_add\n"
+	"#define b3Fabs fabs\n"
+	"#define b3Sqrt native_sqrt\n"
+	"#define b3Sin native_sin\n"
+	"#define b3Cos native_cos\n"
+	"#define B3_STATIC\n"
+	"#endif\n"
+	"#endif\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"	typedef float4	b3Float4;\n"
+	"	#define b3Float4ConstArg const b3Float4\n"
+	"	#define b3MakeFloat4 (float4)\n"
+	"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return dot(a1, b1);\n"
+	"	}\n"
+	"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return cross(a1, b1);\n"
+	"	}\n"
+	"	#define b3MinFloat4 min\n"
+	"	#define b3MaxFloat4 max\n"
+	"	#define b3Normalized(a) normalize(a)\n"
+	"#endif \n"
+	"		\n"
+	"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
+	"{\n"
+	"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
+	"		return false;\n"
+	"	return true;\n"
+	"}\n"
+	"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
+	"{\n"
+	"    float maxDot = -B3_INFINITY;\n"
+	"    int i = 0;\n"
+	"    int ptIndex = -1;\n"
+	"    for( i = 0; i < vecLen; i++ )\n"
+	"    {\n"
+	"        float dot = b3Dot3F4(vecArray[i],vec);\n"
+	"            \n"
+	"        if( dot > maxDot )\n"
+	"        {\n"
+	"            maxDot = dot;\n"
+	"            ptIndex = i;\n"
+	"        }\n"
+	"    }\n"
+	"	b3Assert(ptIndex>=0);\n"
+	"    if (ptIndex<0)\n"
+	"	{\n"
+	"		ptIndex = 0;\n"
+	"	}\n"
+	"    *dotOut = maxDot;\n"
+	"    return ptIndex;\n"
+	"}\n"
+	"#endif //B3_FLOAT4_H\n"
+	"typedef  struct b3Contact4Data b3Contact4Data_t;\n"
+	"struct b3Contact4Data\n"
+	"{\n"
+	"	b3Float4	m_worldPosB[4];\n"
+	"//	b3Float4	m_localPosA[4];\n"
+	"//	b3Float4	m_localPosB[4];\n"
+	"	b3Float4	m_worldNormalOnB;	//	w: m_nPoints\n"
+	"	unsigned short  m_restituitionCoeffCmp;\n"
+	"	unsigned short  m_frictionCoeffCmp;\n"
+	"	int m_batchIdx;\n"
+	"	int m_bodyAPtrAndSignBit;//x:m_bodyAPtr, y:m_bodyBPtr\n"
+	"	int m_bodyBPtrAndSignBit;\n"
+	"	int	m_childIndexA;\n"
+	"	int	m_childIndexB;\n"
+	"	int m_unused1;\n"
+	"	int m_unused2;\n"
+	"};\n"
+	"inline int b3Contact4Data_getNumPoints(const struct b3Contact4Data* contact)\n"
+	"{\n"
+	"	return (int)contact->m_worldNormalOnB.w;\n"
+	"};\n"
+	"inline void b3Contact4Data_setNumPoints(struct b3Contact4Data* contact, int numPoints)\n"
+	"{\n"
+	"	contact->m_worldNormalOnB.w = (float)numPoints;\n"
+	"};\n"
+	"#endif //B3_CONTACT4DATA_H\n"
+	"#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_extended_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics : enable\n"
+	"#ifdef cl_ext_atomic_counters_32\n"
+	"#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
+	"#else\n"
+	"#define counter32_t volatile __global int*\n"
+	"#endif\n"
+	"#define SIMD_WIDTH 64\n"
+	"typedef unsigned int u32;\n"
+	"typedef unsigned short u16;\n"
+	"typedef unsigned char u8;\n"
+	"#define GET_GROUP_IDX get_group_id(0)\n"
+	"#define GET_LOCAL_IDX get_local_id(0)\n"
+	"#define GET_GLOBAL_IDX get_global_id(0)\n"
+	"#define GET_GROUP_SIZE get_local_size(0)\n"
+	"#define GET_NUM_GROUPS get_num_groups(0)\n"
+	"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
+	"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
+	"#define AtomInc(x) atom_inc(&(x))\n"
+	"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
+	"#define AppendInc(x, out) out = atomic_inc(x)\n"
+	"#define AtomAdd(x, value) atom_add(&(x), value)\n"
+	"#define AtomCmpxhg(x, cmp, value) atom_cmpxchg( &(x), cmp, value )\n"
+	"#define AtomXhg(x, value) atom_xchg ( &(x), value )\n"
+	"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
+	"#define make_float4 (float4)\n"
+	"#define make_float2 (float2)\n"
+	"#define make_uint4 (uint4)\n"
+	"#define make_int4 (int4)\n"
+	"#define make_uint2 (uint2)\n"
+	"#define make_int2 (int2)\n"
+	"#define max2 max\n"
+	"#define min2 min\n"
+	"#define WG_SIZE 64\n"
+	"typedef struct \n"
+	"{\n"
+	"	int m_n;\n"
+	"	int m_start;\n"
+	"	int m_staticIdx;\n"
+	"	int m_paddings[1];\n"
+	"} ConstBuffer;\n"
+	"typedef struct \n"
+	"{\n"
+	"	int m_a;\n"
+	"	int m_b;\n"
+	"	u32 m_idx;\n"
+	"}Elem;\n"
+	"//	batching on the GPU\n"
+	"__kernel void CreateBatchesBruteForce( __global struct b3Contact4Data* gConstraints, 	__global const u32* gN, __global const u32* gStart, int m_staticIdx )\n"
+	"{\n"
+	"	int wgIdx = GET_GROUP_IDX;\n"
+	"	int lIdx = GET_LOCAL_IDX;\n"
+	"	\n"
+	"	const int m_n = gN[wgIdx];\n"
+	"	const int m_start = gStart[wgIdx];\n"
+	"		\n"
+	"	if( lIdx == 0 )\n"
+	"	{\n"
+	"		for (int i=0;i<m_n;i++)\n"
+	"		{\n"
+	"			int srcIdx = i+m_start;\n"
+	"			int batchIndex = i;\n"
+	"			gConstraints[ srcIdx ].m_batchIdx = batchIndex;	\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"#define CHECK_SIZE (WG_SIZE)\n"
+	"u32 readBuf(__local u32* buff, int idx)\n"
+	"{\n"
+	"	idx = idx % (32*CHECK_SIZE);\n"
+	"	int bitIdx = idx%32;\n"
+	"	int bufIdx = idx/32;\n"
+	"	return buff[bufIdx] & (1<<bitIdx);\n"
+	"}\n"
+	"void writeBuf(__local u32* buff, int idx)\n"
+	"{\n"
+	"	idx = idx % (32*CHECK_SIZE);\n"
+	"	int bitIdx = idx%32;\n"
+	"	int bufIdx = idx/32;\n"
+	"	buff[bufIdx] |= (1<<bitIdx);\n"
+	"	//atom_or( &buff[bufIdx], (1<<bitIdx) );\n"
+	"}\n"
+	"u32 tryWrite(__local u32* buff, int idx)\n"
+	"{\n"
+	"	idx = idx % (32*CHECK_SIZE);\n"
+	"	int bitIdx = idx%32;\n"
+	"	int bufIdx = idx/32;\n"
+	"	u32 ans = (u32)atom_or( &buff[bufIdx], (1<<bitIdx) );\n"
+	"	return ((ans >> bitIdx)&1) == 0;\n"
+	"}\n"
+	"//	batching on the GPU\n"
+	"__kernel void CreateBatchesNew( __global struct b3Contact4Data* gConstraints, __global const u32* gN, __global const u32* gStart, __global int* batchSizes, int staticIdx )\n"
+	"{\n"
+	"	int wgIdx = GET_GROUP_IDX;\n"
+	"	int lIdx = GET_LOCAL_IDX;\n"
+	"	const int numConstraints = gN[wgIdx];\n"
+	"	const int m_start = gStart[wgIdx];\n"
+	"	b3Contact4Data_t tmp;\n"
+	"	\n"
+	"	__local u32 ldsFixedBuffer[CHECK_SIZE];\n"
+	"		\n"
+	"	\n"
+	"	\n"
+	"	\n"
+	"	\n"
+	"	if( lIdx == 0 )\n"
+	"	{\n"
+	"	\n"
+	"		\n"
+	"		__global struct b3Contact4Data* cs = &gConstraints[m_start];	\n"
+	"	\n"
+	"		\n"
+	"		int numValidConstraints = 0;\n"
+	"		int batchIdx = 0;\n"
+	"		while( numValidConstraints < numConstraints)\n"
+	"		{\n"
+	"			int nCurrentBatch = 0;\n"
+	"			//	clear flag\n"
+	"	\n"
+	"			for(int i=0; i<CHECK_SIZE; i++) \n"
+	"				ldsFixedBuffer[i] = 0;		\n"
+	"			for(int i=numValidConstraints; i<numConstraints; i++)\n"
+	"			{\n"
+	"				int bodyAS = cs[i].m_bodyAPtrAndSignBit;\n"
+	"				int bodyBS = cs[i].m_bodyBPtrAndSignBit;\n"
+	"				int bodyA = abs(bodyAS);\n"
+	"				int bodyB = abs(bodyBS);\n"
+	"				bool aIsStatic = (bodyAS<0) || bodyAS==staticIdx;\n"
+	"				bool bIsStatic = (bodyBS<0) || bodyBS==staticIdx;\n"
+	"				int aUnavailable = aIsStatic ? 0 : readBuf( ldsFixedBuffer, bodyA);\n"
+	"				int bUnavailable = bIsStatic ? 0 : readBuf( ldsFixedBuffer, bodyB);\n"
+	"				\n"
+	"				if( aUnavailable==0 && bUnavailable==0 ) // ok\n"
+	"				{\n"
+	"					if (!aIsStatic)\n"
+	"					{\n"
+	"						writeBuf( ldsFixedBuffer, bodyA );\n"
+	"					}\n"
+	"					if (!bIsStatic)\n"
+	"					{\n"
+	"						writeBuf( ldsFixedBuffer, bodyB );\n"
+	"					}\n"
+	"					cs[i].m_batchIdx = batchIdx;\n"
+	"					if (i!=numValidConstraints)\n"
+	"					{\n"
+	"						tmp = cs[i];\n"
+	"						cs[i] = cs[numValidConstraints];\n"
+	"						cs[numValidConstraints]  = tmp;\n"
+	"					}\n"
+	"					numValidConstraints++;\n"
+	"					\n"
+	"					nCurrentBatch++;\n"
+	"					if( nCurrentBatch == SIMD_WIDTH)\n"
+	"					{\n"
+	"						nCurrentBatch = 0;\n"
+	"						for(int i=0; i<CHECK_SIZE; i++) \n"
+	"							ldsFixedBuffer[i] = 0;\n"
+	"						\n"
+	"					}\n"
+	"				}\n"
+	"			}//for\n"
+	"			batchIdx ++;\n"
+	"		}//while\n"
+	"		\n"
+	"		batchSizes[wgIdx] = batchIdx;\n"
+	"	}//if( lIdx == 0 )\n"
+	"	\n"
+	"	//return batchIdx;\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/integrateKernel.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/integrateKernel.h
index a5a432947c1..6e9c53e1617 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/integrateKernel.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/integrateKernel.h
@@ -1,433 +1,432 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* integrateKernelCL= \
-"/*\n"
-"Copyright (c) 2013 Advanced Micro Devices, Inc.  \n"
-"This software is provided 'as-is', without any express or implied warranty.\n"
-"In no event will the authors be held liable for any damages arising from the use of this software.\n"
-"Permission is granted to anyone to use this software for any purpose, \n"
-"including commercial applications, and to alter it and redistribute it freely, \n"
-"subject to the following restrictions:\n"
-"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
-"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
-"3. This notice may not be removed or altered from any source distribution.\n"
-"*/\n"
-"//Originally written by Erwin Coumans\n"
-"#ifndef B3_RIGIDBODY_DATA_H\n"
-"#define B3_RIGIDBODY_DATA_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#define B3_FLOAT4_H\n"
-"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
-"#define B3_PLATFORM_DEFINITIONS_H\n"
-"struct MyTest\n"
-"{\n"
-"	int bla;\n"
-"};\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
-"#define B3_LARGE_FLOAT 1e18f\n"
-"#define B3_INFINITY 1e18f\n"
-"#define b3Assert(a)\n"
-"#define b3ConstArray(a) __global const a*\n"
-"#define b3AtomicInc atomic_inc\n"
-"#define b3AtomicAdd atomic_add\n"
-"#define b3Fabs fabs\n"
-"#define b3Sqrt native_sqrt\n"
-"#define b3Sin native_sin\n"
-"#define b3Cos native_cos\n"
-"#define B3_STATIC\n"
-"#endif\n"
-"#endif\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"	typedef float4	b3Float4;\n"
-"	#define b3Float4ConstArg const b3Float4\n"
-"	#define b3MakeFloat4 (float4)\n"
-"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return dot(a1, b1);\n"
-"	}\n"
-"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return cross(a1, b1);\n"
-"	}\n"
-"	#define b3MinFloat4 min\n"
-"	#define b3MaxFloat4 max\n"
-"	#define b3Normalized(a) normalize(a)\n"
-"#endif \n"
-"		\n"
-"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
-"{\n"
-"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
-"		return false;\n"
-"	return true;\n"
-"}\n"
-"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
-"{\n"
-"    float maxDot = -B3_INFINITY;\n"
-"    int i = 0;\n"
-"    int ptIndex = -1;\n"
-"    for( i = 0; i < vecLen; i++ )\n"
-"    {\n"
-"        float dot = b3Dot3F4(vecArray[i],vec);\n"
-"            \n"
-"        if( dot > maxDot )\n"
-"        {\n"
-"            maxDot = dot;\n"
-"            ptIndex = i;\n"
-"        }\n"
-"    }\n"
-"	b3Assert(ptIndex>=0);\n"
-"    if (ptIndex<0)\n"
-"	{\n"
-"		ptIndex = 0;\n"
-"	}\n"
-"    *dotOut = maxDot;\n"
-"    return ptIndex;\n"
-"}\n"
-"#endif //B3_FLOAT4_H\n"
-"#ifndef B3_QUAT_H\n"
-"#define B3_QUAT_H\n"
-"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif\n"
-"#endif\n"
-"#ifndef B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"	typedef float4	b3Quat;\n"
-"	#define b3QuatConstArg const b3Quat\n"
-"	\n"
-"	\n"
-"inline float4 b3FastNormalize4(float4 v)\n"
-"{\n"
-"	v = (float4)(v.xyz,0.f);\n"
-"	return fast_normalize(v);\n"
-"}\n"
-"	\n"
-"inline b3Quat b3QuatMul(b3Quat a, b3Quat b);\n"
-"inline b3Quat b3QuatNormalized(b3QuatConstArg in);\n"
-"inline b3Quat b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec);\n"
-"inline b3Quat b3QuatInvert(b3QuatConstArg q);\n"
-"inline b3Quat b3QuatInverse(b3QuatConstArg q);\n"
-"inline b3Quat b3QuatMul(b3QuatConstArg a, b3QuatConstArg b)\n"
-"{\n"
-"	b3Quat ans;\n"
-"	ans = b3Cross3( a, b );\n"
-"	ans += a.w*b+b.w*a;\n"
-"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
-"	ans.w = a.w*b.w - b3Dot3F4(a, b);\n"
-"	return ans;\n"
-"}\n"
-"inline b3Quat b3QuatNormalized(b3QuatConstArg in)\n"
-"{\n"
-"	b3Quat q;\n"
-"	q=in;\n"
-"	//return b3FastNormalize4(in);\n"
-"	float len = native_sqrt(dot(q, q));\n"
-"	if(len > 0.f)\n"
-"	{\n"
-"		q *= 1.f / len;\n"
-"	}\n"
-"	else\n"
-"	{\n"
-"		q.x = q.y = q.z = 0.f;\n"
-"		q.w = 1.f;\n"
-"	}\n"
-"	return q;\n"
-"}\n"
-"inline float4 b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
-"{\n"
-"	b3Quat qInv = b3QuatInvert( q );\n"
-"	float4 vcpy = vec;\n"
-"	vcpy.w = 0.f;\n"
-"	float4 out = b3QuatMul(b3QuatMul(q,vcpy),qInv);\n"
-"	return out;\n"
-"}\n"
-"inline b3Quat b3QuatInverse(b3QuatConstArg q)\n"
-"{\n"
-"	return (b3Quat)(-q.xyz, q.w);\n"
-"}\n"
-"inline b3Quat b3QuatInvert(b3QuatConstArg q)\n"
-"{\n"
-"	return (b3Quat)(-q.xyz, q.w);\n"
-"}\n"
-"inline float4 b3QuatInvRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
-"{\n"
-"	return b3QuatRotate( b3QuatInvert( q ), vec );\n"
-"}\n"
-"inline b3Float4 b3TransformPoint(b3Float4ConstArg point, b3Float4ConstArg translation, b3QuatConstArg  orientation)\n"
-"{\n"
-"	return b3QuatRotate( orientation, point ) + (translation);\n"
-"}\n"
-"	\n"
-"#endif \n"
-"#endif //B3_QUAT_H\n"
-"#ifndef B3_MAT3x3_H\n"
-"#define B3_MAT3x3_H\n"
-"#ifndef B3_QUAT_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_QUAT_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"typedef struct\n"
-"{\n"
-"	b3Float4 m_row[3];\n"
-"}b3Mat3x3;\n"
-"#define b3Mat3x3ConstArg const b3Mat3x3\n"
-"#define b3GetRow(m,row) (m.m_row[row])\n"
-"inline b3Mat3x3 b3QuatGetRotationMatrix(b3Quat quat)\n"
-"{\n"
-"	b3Float4 quat2 = (b3Float4)(quat.x*quat.x, quat.y*quat.y, quat.z*quat.z, 0.f);\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0].x=1-2*quat2.y-2*quat2.z;\n"
-"	out.m_row[0].y=2*quat.x*quat.y-2*quat.w*quat.z;\n"
-"	out.m_row[0].z=2*quat.x*quat.z+2*quat.w*quat.y;\n"
-"	out.m_row[0].w = 0.f;\n"
-"	out.m_row[1].x=2*quat.x*quat.y+2*quat.w*quat.z;\n"
-"	out.m_row[1].y=1-2*quat2.x-2*quat2.z;\n"
-"	out.m_row[1].z=2*quat.y*quat.z-2*quat.w*quat.x;\n"
-"	out.m_row[1].w = 0.f;\n"
-"	out.m_row[2].x=2*quat.x*quat.z-2*quat.w*quat.y;\n"
-"	out.m_row[2].y=2*quat.y*quat.z+2*quat.w*quat.x;\n"
-"	out.m_row[2].z=1-2*quat2.x-2*quat2.y;\n"
-"	out.m_row[2].w = 0.f;\n"
-"	return out;\n"
-"}\n"
-"inline b3Mat3x3 b3AbsoluteMat3x3(b3Mat3x3ConstArg matIn)\n"
-"{\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0] = fabs(matIn.m_row[0]);\n"
-"	out.m_row[1] = fabs(matIn.m_row[1]);\n"
-"	out.m_row[2] = fabs(matIn.m_row[2]);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtZero();\n"
-"__inline\n"
-"b3Mat3x3 mtIdentity();\n"
-"__inline\n"
-"b3Mat3x3 mtTranspose(b3Mat3x3 m);\n"
-"__inline\n"
-"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b);\n"
-"__inline\n"
-"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b);\n"
-"__inline\n"
-"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b);\n"
-"__inline\n"
-"b3Mat3x3 mtZero()\n"
-"{\n"
-"	b3Mat3x3 m;\n"
-"	m.m_row[0] = (b3Float4)(0.f);\n"
-"	m.m_row[1] = (b3Float4)(0.f);\n"
-"	m.m_row[2] = (b3Float4)(0.f);\n"
-"	return m;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtIdentity()\n"
-"{\n"
-"	b3Mat3x3 m;\n"
-"	m.m_row[0] = (b3Float4)(1,0,0,0);\n"
-"	m.m_row[1] = (b3Float4)(0,1,0,0);\n"
-"	m.m_row[2] = (b3Float4)(0,0,1,0);\n"
-"	return m;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtTranspose(b3Mat3x3 m)\n"
-"{\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0] = (b3Float4)(m.m_row[0].x, m.m_row[1].x, m.m_row[2].x, 0.f);\n"
-"	out.m_row[1] = (b3Float4)(m.m_row[0].y, m.m_row[1].y, m.m_row[2].y, 0.f);\n"
-"	out.m_row[2] = (b3Float4)(m.m_row[0].z, m.m_row[1].z, m.m_row[2].z, 0.f);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b)\n"
-"{\n"
-"	b3Mat3x3 transB;\n"
-"	transB = mtTranspose( b );\n"
-"	b3Mat3x3 ans;\n"
-"	//	why this doesn't run when 0ing in the for{}\n"
-"	a.m_row[0].w = 0.f;\n"
-"	a.m_row[1].w = 0.f;\n"
-"	a.m_row[2].w = 0.f;\n"
-"	for(int i=0; i<3; i++)\n"
-"	{\n"
-"//	a.m_row[i].w = 0.f;\n"
-"		ans.m_row[i].x = b3Dot3F4(a.m_row[i],transB.m_row[0]);\n"
-"		ans.m_row[i].y = b3Dot3F4(a.m_row[i],transB.m_row[1]);\n"
-"		ans.m_row[i].z = b3Dot3F4(a.m_row[i],transB.m_row[2]);\n"
-"		ans.m_row[i].w = 0.f;\n"
-"	}\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b)\n"
-"{\n"
-"	b3Float4 ans;\n"
-"	ans.x = b3Dot3F4( a.m_row[0], b );\n"
-"	ans.y = b3Dot3F4( a.m_row[1], b );\n"
-"	ans.z = b3Dot3F4( a.m_row[2], b );\n"
-"	ans.w = 0.f;\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b)\n"
-"{\n"
-"	b3Float4 colx = b3MakeFloat4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
-"	b3Float4 coly = b3MakeFloat4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
-"	b3Float4 colz = b3MakeFloat4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
-"	b3Float4 ans;\n"
-"	ans.x = b3Dot3F4( a, colx );\n"
-"	ans.y = b3Dot3F4( a, coly );\n"
-"	ans.z = b3Dot3F4( a, colz );\n"
-"	return ans;\n"
-"}\n"
-"#endif\n"
-"#endif //B3_MAT3x3_H\n"
-"typedef struct b3RigidBodyData b3RigidBodyData_t;\n"
-"struct b3RigidBodyData\n"
-"{\n"
-"	b3Float4				m_pos;\n"
-"	b3Quat					m_quat;\n"
-"	b3Float4				m_linVel;\n"
-"	b3Float4				m_angVel;\n"
-"	int 					m_collidableIdx;\n"
-"	float 				m_invMass;\n"
-"	float 				m_restituitionCoeff;\n"
-"	float 				m_frictionCoeff;\n"
-"};\n"
-"typedef struct b3InertiaData b3InertiaData_t;\n"
-"struct b3InertiaData\n"
-"{\n"
-"	b3Mat3x3 m_invInertiaWorld;\n"
-"	b3Mat3x3 m_initInvInertia;\n"
-"};\n"
-"#endif //B3_RIGIDBODY_DATA_H\n"
-"	\n"
-"#ifndef B3_RIGIDBODY_DATA_H\n"
-"#endif //B3_RIGIDBODY_DATA_H\n"
-"	\n"
-"inline void integrateSingleTransform( __global b3RigidBodyData_t* bodies,int nodeID, float timeStep, float angularDamping, b3Float4ConstArg gravityAcceleration)\n"
-"{\n"
-"	\n"
-"	if (bodies[nodeID].m_invMass != 0.f)\n"
-"	{\n"
-"		float BT_GPU_ANGULAR_MOTION_THRESHOLD = (0.25f * 3.14159254f);\n"
-"		//angular velocity\n"
-"		{\n"
-"			b3Float4 axis;\n"
-"			//add some hardcoded angular damping\n"
-"			bodies[nodeID].m_angVel.x *= angularDamping;\n"
-"			bodies[nodeID].m_angVel.y *= angularDamping;\n"
-"			bodies[nodeID].m_angVel.z *= angularDamping;\n"
-"			\n"
-"			b3Float4 angvel = bodies[nodeID].m_angVel;\n"
-"			float fAngle = b3Sqrt(b3Dot3F4(angvel, angvel));\n"
-"			\n"
-"			//limit the angular motion\n"
-"			if(fAngle*timeStep > BT_GPU_ANGULAR_MOTION_THRESHOLD)\n"
-"			{\n"
-"				fAngle = BT_GPU_ANGULAR_MOTION_THRESHOLD / timeStep;\n"
-"			}\n"
-"			if(fAngle < 0.001f)\n"
-"			{\n"
-"				// use Taylor's expansions of sync function\n"
-"				axis = angvel * (0.5f*timeStep-(timeStep*timeStep*timeStep)*0.020833333333f * fAngle * fAngle);\n"
-"			}\n"
-"			else\n"
-"			{\n"
-"				// sync(fAngle) = sin(c*fAngle)/t\n"
-"				axis = angvel * ( b3Sin(0.5f * fAngle * timeStep) / fAngle);\n"
-"			}\n"
-"			\n"
-"			b3Quat dorn;\n"
-"			dorn.x = axis.x;\n"
-"			dorn.y = axis.y;\n"
-"			dorn.z = axis.z;\n"
-"			dorn.w = b3Cos(fAngle * timeStep * 0.5f);\n"
-"			b3Quat orn0 = bodies[nodeID].m_quat;\n"
-"			b3Quat predictedOrn = b3QuatMul(dorn, orn0);\n"
-"			predictedOrn = b3QuatNormalized(predictedOrn);\n"
-"			bodies[nodeID].m_quat=predictedOrn;\n"
-"		}\n"
-"		//linear velocity		\n"
-"		bodies[nodeID].m_pos +=  bodies[nodeID].m_linVel * timeStep;\n"
-"		\n"
-"		//apply gravity\n"
-"		bodies[nodeID].m_linVel += gravityAcceleration * timeStep;\n"
-"		\n"
-"	}\n"
-"	\n"
-"}\n"
-"inline void b3IntegrateTransform( __global b3RigidBodyData_t* body, float timeStep, float angularDamping, b3Float4ConstArg gravityAcceleration)\n"
-"{\n"
-"	float BT_GPU_ANGULAR_MOTION_THRESHOLD = (0.25f * 3.14159254f);\n"
-"	\n"
-"	if( (body->m_invMass != 0.f))\n"
-"	{\n"
-"		//angular velocity\n"
-"		{\n"
-"			b3Float4 axis;\n"
-"			//add some hardcoded angular damping\n"
-"			body->m_angVel.x *= angularDamping;\n"
-"			body->m_angVel.y *= angularDamping;\n"
-"			body->m_angVel.z *= angularDamping;\n"
-"			\n"
-"			b3Float4 angvel = body->m_angVel;\n"
-"			float fAngle = b3Sqrt(b3Dot3F4(angvel, angvel));\n"
-"			//limit the angular motion\n"
-"			if(fAngle*timeStep > BT_GPU_ANGULAR_MOTION_THRESHOLD)\n"
-"			{\n"
-"				fAngle = BT_GPU_ANGULAR_MOTION_THRESHOLD / timeStep;\n"
-"			}\n"
-"			if(fAngle < 0.001f)\n"
-"			{\n"
-"				// use Taylor's expansions of sync function\n"
-"				axis = angvel * (0.5f*timeStep-(timeStep*timeStep*timeStep)*0.020833333333f * fAngle * fAngle);\n"
-"			}\n"
-"			else\n"
-"			{\n"
-"				// sync(fAngle) = sin(c*fAngle)/t\n"
-"				axis = angvel * ( b3Sin(0.5f * fAngle * timeStep) / fAngle);\n"
-"			}\n"
-"			b3Quat dorn;\n"
-"			dorn.x = axis.x;\n"
-"			dorn.y = axis.y;\n"
-"			dorn.z = axis.z;\n"
-"			dorn.w = b3Cos(fAngle * timeStep * 0.5f);\n"
-"			b3Quat orn0 = body->m_quat;\n"
-"			b3Quat predictedOrn = b3QuatMul(dorn, orn0);\n"
-"			predictedOrn = b3QuatNormalized(predictedOrn);\n"
-"			body->m_quat=predictedOrn;\n"
-"		}\n"
-"		//apply gravity\n"
-"		body->m_linVel += gravityAcceleration * timeStep;\n"
-"		//linear velocity		\n"
-"		body->m_pos +=  body->m_linVel * timeStep;\n"
-"		\n"
-"	}\n"
-"	\n"
-"}\n"
-"__kernel void \n"
-"  integrateTransformsKernel( __global b3RigidBodyData_t* bodies,const int numNodes, float timeStep, float angularDamping, float4 gravityAcceleration)\n"
-"{\n"
-"	int nodeID = get_global_id(0);\n"
-"	\n"
-"	if( nodeID < numNodes)\n"
-"	{\n"
-"		integrateSingleTransform(bodies,nodeID, timeStep, angularDamping,gravityAcceleration);\n"
-"	}\n"
-"}\n"
-;
+static const char* integrateKernelCL =
+	"/*\n"
+	"Copyright (c) 2013 Advanced Micro Devices, Inc.  \n"
+	"This software is provided 'as-is', without any express or implied warranty.\n"
+	"In no event will the authors be held liable for any damages arising from the use of this software.\n"
+	"Permission is granted to anyone to use this software for any purpose, \n"
+	"including commercial applications, and to alter it and redistribute it freely, \n"
+	"subject to the following restrictions:\n"
+	"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
+	"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
+	"3. This notice may not be removed or altered from any source distribution.\n"
+	"*/\n"
+	"//Originally written by Erwin Coumans\n"
+	"#ifndef B3_RIGIDBODY_DATA_H\n"
+	"#define B3_RIGIDBODY_DATA_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#define B3_FLOAT4_H\n"
+	"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
+	"#define B3_PLATFORM_DEFINITIONS_H\n"
+	"struct MyTest\n"
+	"{\n"
+	"	int bla;\n"
+	"};\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
+	"#define B3_LARGE_FLOAT 1e18f\n"
+	"#define B3_INFINITY 1e18f\n"
+	"#define b3Assert(a)\n"
+	"#define b3ConstArray(a) __global const a*\n"
+	"#define b3AtomicInc atomic_inc\n"
+	"#define b3AtomicAdd atomic_add\n"
+	"#define b3Fabs fabs\n"
+	"#define b3Sqrt native_sqrt\n"
+	"#define b3Sin native_sin\n"
+	"#define b3Cos native_cos\n"
+	"#define B3_STATIC\n"
+	"#endif\n"
+	"#endif\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"	typedef float4	b3Float4;\n"
+	"	#define b3Float4ConstArg const b3Float4\n"
+	"	#define b3MakeFloat4 (float4)\n"
+	"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return dot(a1, b1);\n"
+	"	}\n"
+	"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return cross(a1, b1);\n"
+	"	}\n"
+	"	#define b3MinFloat4 min\n"
+	"	#define b3MaxFloat4 max\n"
+	"	#define b3Normalized(a) normalize(a)\n"
+	"#endif \n"
+	"		\n"
+	"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
+	"{\n"
+	"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
+	"		return false;\n"
+	"	return true;\n"
+	"}\n"
+	"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
+	"{\n"
+	"    float maxDot = -B3_INFINITY;\n"
+	"    int i = 0;\n"
+	"    int ptIndex = -1;\n"
+	"    for( i = 0; i < vecLen; i++ )\n"
+	"    {\n"
+	"        float dot = b3Dot3F4(vecArray[i],vec);\n"
+	"            \n"
+	"        if( dot > maxDot )\n"
+	"        {\n"
+	"            maxDot = dot;\n"
+	"            ptIndex = i;\n"
+	"        }\n"
+	"    }\n"
+	"	b3Assert(ptIndex>=0);\n"
+	"    if (ptIndex<0)\n"
+	"	{\n"
+	"		ptIndex = 0;\n"
+	"	}\n"
+	"    *dotOut = maxDot;\n"
+	"    return ptIndex;\n"
+	"}\n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifndef B3_QUAT_H\n"
+	"#define B3_QUAT_H\n"
+	"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif\n"
+	"#endif\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"	typedef float4	b3Quat;\n"
+	"	#define b3QuatConstArg const b3Quat\n"
+	"	\n"
+	"	\n"
+	"inline float4 b3FastNormalize4(float4 v)\n"
+	"{\n"
+	"	v = (float4)(v.xyz,0.f);\n"
+	"	return fast_normalize(v);\n"
+	"}\n"
+	"	\n"
+	"inline b3Quat b3QuatMul(b3Quat a, b3Quat b);\n"
+	"inline b3Quat b3QuatNormalized(b3QuatConstArg in);\n"
+	"inline b3Quat b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec);\n"
+	"inline b3Quat b3QuatInvert(b3QuatConstArg q);\n"
+	"inline b3Quat b3QuatInverse(b3QuatConstArg q);\n"
+	"inline b3Quat b3QuatMul(b3QuatConstArg a, b3QuatConstArg b)\n"
+	"{\n"
+	"	b3Quat ans;\n"
+	"	ans = b3Cross3( a, b );\n"
+	"	ans += a.w*b+b.w*a;\n"
+	"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
+	"	ans.w = a.w*b.w - b3Dot3F4(a, b);\n"
+	"	return ans;\n"
+	"}\n"
+	"inline b3Quat b3QuatNormalized(b3QuatConstArg in)\n"
+	"{\n"
+	"	b3Quat q;\n"
+	"	q=in;\n"
+	"	//return b3FastNormalize4(in);\n"
+	"	float len = native_sqrt(dot(q, q));\n"
+	"	if(len > 0.f)\n"
+	"	{\n"
+	"		q *= 1.f / len;\n"
+	"	}\n"
+	"	else\n"
+	"	{\n"
+	"		q.x = q.y = q.z = 0.f;\n"
+	"		q.w = 1.f;\n"
+	"	}\n"
+	"	return q;\n"
+	"}\n"
+	"inline float4 b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
+	"{\n"
+	"	b3Quat qInv = b3QuatInvert( q );\n"
+	"	float4 vcpy = vec;\n"
+	"	vcpy.w = 0.f;\n"
+	"	float4 out = b3QuatMul(b3QuatMul(q,vcpy),qInv);\n"
+	"	return out;\n"
+	"}\n"
+	"inline b3Quat b3QuatInverse(b3QuatConstArg q)\n"
+	"{\n"
+	"	return (b3Quat)(-q.xyz, q.w);\n"
+	"}\n"
+	"inline b3Quat b3QuatInvert(b3QuatConstArg q)\n"
+	"{\n"
+	"	return (b3Quat)(-q.xyz, q.w);\n"
+	"}\n"
+	"inline float4 b3QuatInvRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
+	"{\n"
+	"	return b3QuatRotate( b3QuatInvert( q ), vec );\n"
+	"}\n"
+	"inline b3Float4 b3TransformPoint(b3Float4ConstArg point, b3Float4ConstArg translation, b3QuatConstArg  orientation)\n"
+	"{\n"
+	"	return b3QuatRotate( orientation, point ) + (translation);\n"
+	"}\n"
+	"	\n"
+	"#endif \n"
+	"#endif //B3_QUAT_H\n"
+	"#ifndef B3_MAT3x3_H\n"
+	"#define B3_MAT3x3_H\n"
+	"#ifndef B3_QUAT_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_QUAT_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"typedef struct\n"
+	"{\n"
+	"	b3Float4 m_row[3];\n"
+	"}b3Mat3x3;\n"
+	"#define b3Mat3x3ConstArg const b3Mat3x3\n"
+	"#define b3GetRow(m,row) (m.m_row[row])\n"
+	"inline b3Mat3x3 b3QuatGetRotationMatrix(b3Quat quat)\n"
+	"{\n"
+	"	b3Float4 quat2 = (b3Float4)(quat.x*quat.x, quat.y*quat.y, quat.z*quat.z, 0.f);\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0].x=1-2*quat2.y-2*quat2.z;\n"
+	"	out.m_row[0].y=2*quat.x*quat.y-2*quat.w*quat.z;\n"
+	"	out.m_row[0].z=2*quat.x*quat.z+2*quat.w*quat.y;\n"
+	"	out.m_row[0].w = 0.f;\n"
+	"	out.m_row[1].x=2*quat.x*quat.y+2*quat.w*quat.z;\n"
+	"	out.m_row[1].y=1-2*quat2.x-2*quat2.z;\n"
+	"	out.m_row[1].z=2*quat.y*quat.z-2*quat.w*quat.x;\n"
+	"	out.m_row[1].w = 0.f;\n"
+	"	out.m_row[2].x=2*quat.x*quat.z-2*quat.w*quat.y;\n"
+	"	out.m_row[2].y=2*quat.y*quat.z+2*quat.w*quat.x;\n"
+	"	out.m_row[2].z=1-2*quat2.x-2*quat2.y;\n"
+	"	out.m_row[2].w = 0.f;\n"
+	"	return out;\n"
+	"}\n"
+	"inline b3Mat3x3 b3AbsoluteMat3x3(b3Mat3x3ConstArg matIn)\n"
+	"{\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0] = fabs(matIn.m_row[0]);\n"
+	"	out.m_row[1] = fabs(matIn.m_row[1]);\n"
+	"	out.m_row[2] = fabs(matIn.m_row[2]);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtZero();\n"
+	"__inline\n"
+	"b3Mat3x3 mtIdentity();\n"
+	"__inline\n"
+	"b3Mat3x3 mtTranspose(b3Mat3x3 m);\n"
+	"__inline\n"
+	"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b);\n"
+	"__inline\n"
+	"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b);\n"
+	"__inline\n"
+	"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b);\n"
+	"__inline\n"
+	"b3Mat3x3 mtZero()\n"
+	"{\n"
+	"	b3Mat3x3 m;\n"
+	"	m.m_row[0] = (b3Float4)(0.f);\n"
+	"	m.m_row[1] = (b3Float4)(0.f);\n"
+	"	m.m_row[2] = (b3Float4)(0.f);\n"
+	"	return m;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtIdentity()\n"
+	"{\n"
+	"	b3Mat3x3 m;\n"
+	"	m.m_row[0] = (b3Float4)(1,0,0,0);\n"
+	"	m.m_row[1] = (b3Float4)(0,1,0,0);\n"
+	"	m.m_row[2] = (b3Float4)(0,0,1,0);\n"
+	"	return m;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtTranspose(b3Mat3x3 m)\n"
+	"{\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0] = (b3Float4)(m.m_row[0].x, m.m_row[1].x, m.m_row[2].x, 0.f);\n"
+	"	out.m_row[1] = (b3Float4)(m.m_row[0].y, m.m_row[1].y, m.m_row[2].y, 0.f);\n"
+	"	out.m_row[2] = (b3Float4)(m.m_row[0].z, m.m_row[1].z, m.m_row[2].z, 0.f);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b)\n"
+	"{\n"
+	"	b3Mat3x3 transB;\n"
+	"	transB = mtTranspose( b );\n"
+	"	b3Mat3x3 ans;\n"
+	"	//	why this doesn't run when 0ing in the for{}\n"
+	"	a.m_row[0].w = 0.f;\n"
+	"	a.m_row[1].w = 0.f;\n"
+	"	a.m_row[2].w = 0.f;\n"
+	"	for(int i=0; i<3; i++)\n"
+	"	{\n"
+	"//	a.m_row[i].w = 0.f;\n"
+	"		ans.m_row[i].x = b3Dot3F4(a.m_row[i],transB.m_row[0]);\n"
+	"		ans.m_row[i].y = b3Dot3F4(a.m_row[i],transB.m_row[1]);\n"
+	"		ans.m_row[i].z = b3Dot3F4(a.m_row[i],transB.m_row[2]);\n"
+	"		ans.m_row[i].w = 0.f;\n"
+	"	}\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b)\n"
+	"{\n"
+	"	b3Float4 ans;\n"
+	"	ans.x = b3Dot3F4( a.m_row[0], b );\n"
+	"	ans.y = b3Dot3F4( a.m_row[1], b );\n"
+	"	ans.z = b3Dot3F4( a.m_row[2], b );\n"
+	"	ans.w = 0.f;\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b)\n"
+	"{\n"
+	"	b3Float4 colx = b3MakeFloat4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
+	"	b3Float4 coly = b3MakeFloat4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
+	"	b3Float4 colz = b3MakeFloat4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
+	"	b3Float4 ans;\n"
+	"	ans.x = b3Dot3F4( a, colx );\n"
+	"	ans.y = b3Dot3F4( a, coly );\n"
+	"	ans.z = b3Dot3F4( a, colz );\n"
+	"	return ans;\n"
+	"}\n"
+	"#endif\n"
+	"#endif //B3_MAT3x3_H\n"
+	"typedef struct b3RigidBodyData b3RigidBodyData_t;\n"
+	"struct b3RigidBodyData\n"
+	"{\n"
+	"	b3Float4				m_pos;\n"
+	"	b3Quat					m_quat;\n"
+	"	b3Float4				m_linVel;\n"
+	"	b3Float4				m_angVel;\n"
+	"	int 					m_collidableIdx;\n"
+	"	float 				m_invMass;\n"
+	"	float 				m_restituitionCoeff;\n"
+	"	float 				m_frictionCoeff;\n"
+	"};\n"
+	"typedef struct b3InertiaData b3InertiaData_t;\n"
+	"struct b3InertiaData\n"
+	"{\n"
+	"	b3Mat3x3 m_invInertiaWorld;\n"
+	"	b3Mat3x3 m_initInvInertia;\n"
+	"};\n"
+	"#endif //B3_RIGIDBODY_DATA_H\n"
+	"	\n"
+	"#ifndef B3_RIGIDBODY_DATA_H\n"
+	"#endif //B3_RIGIDBODY_DATA_H\n"
+	"	\n"
+	"inline void integrateSingleTransform( __global b3RigidBodyData_t* bodies,int nodeID, float timeStep, float angularDamping, b3Float4ConstArg gravityAcceleration)\n"
+	"{\n"
+	"	\n"
+	"	if (bodies[nodeID].m_invMass != 0.f)\n"
+	"	{\n"
+	"		float BT_GPU_ANGULAR_MOTION_THRESHOLD = (0.25f * 3.14159254f);\n"
+	"		//angular velocity\n"
+	"		{\n"
+	"			b3Float4 axis;\n"
+	"			//add some hardcoded angular damping\n"
+	"			bodies[nodeID].m_angVel.x *= angularDamping;\n"
+	"			bodies[nodeID].m_angVel.y *= angularDamping;\n"
+	"			bodies[nodeID].m_angVel.z *= angularDamping;\n"
+	"			\n"
+	"			b3Float4 angvel = bodies[nodeID].m_angVel;\n"
+	"			float fAngle = b3Sqrt(b3Dot3F4(angvel, angvel));\n"
+	"			\n"
+	"			//limit the angular motion\n"
+	"			if(fAngle*timeStep > BT_GPU_ANGULAR_MOTION_THRESHOLD)\n"
+	"			{\n"
+	"				fAngle = BT_GPU_ANGULAR_MOTION_THRESHOLD / timeStep;\n"
+	"			}\n"
+	"			if(fAngle < 0.001f)\n"
+	"			{\n"
+	"				// use Taylor's expansions of sync function\n"
+	"				axis = angvel * (0.5f*timeStep-(timeStep*timeStep*timeStep)*0.020833333333f * fAngle * fAngle);\n"
+	"			}\n"
+	"			else\n"
+	"			{\n"
+	"				// sync(fAngle) = sin(c*fAngle)/t\n"
+	"				axis = angvel * ( b3Sin(0.5f * fAngle * timeStep) / fAngle);\n"
+	"			}\n"
+	"			\n"
+	"			b3Quat dorn;\n"
+	"			dorn.x = axis.x;\n"
+	"			dorn.y = axis.y;\n"
+	"			dorn.z = axis.z;\n"
+	"			dorn.w = b3Cos(fAngle * timeStep * 0.5f);\n"
+	"			b3Quat orn0 = bodies[nodeID].m_quat;\n"
+	"			b3Quat predictedOrn = b3QuatMul(dorn, orn0);\n"
+	"			predictedOrn = b3QuatNormalized(predictedOrn);\n"
+	"			bodies[nodeID].m_quat=predictedOrn;\n"
+	"		}\n"
+	"		//linear velocity		\n"
+	"		bodies[nodeID].m_pos +=  bodies[nodeID].m_linVel * timeStep;\n"
+	"		\n"
+	"		//apply gravity\n"
+	"		bodies[nodeID].m_linVel += gravityAcceleration * timeStep;\n"
+	"		\n"
+	"	}\n"
+	"	\n"
+	"}\n"
+	"inline void b3IntegrateTransform( __global b3RigidBodyData_t* body, float timeStep, float angularDamping, b3Float4ConstArg gravityAcceleration)\n"
+	"{\n"
+	"	float BT_GPU_ANGULAR_MOTION_THRESHOLD = (0.25f * 3.14159254f);\n"
+	"	\n"
+	"	if( (body->m_invMass != 0.f))\n"
+	"	{\n"
+	"		//angular velocity\n"
+	"		{\n"
+	"			b3Float4 axis;\n"
+	"			//add some hardcoded angular damping\n"
+	"			body->m_angVel.x *= angularDamping;\n"
+	"			body->m_angVel.y *= angularDamping;\n"
+	"			body->m_angVel.z *= angularDamping;\n"
+	"			\n"
+	"			b3Float4 angvel = body->m_angVel;\n"
+	"			float fAngle = b3Sqrt(b3Dot3F4(angvel, angvel));\n"
+	"			//limit the angular motion\n"
+	"			if(fAngle*timeStep > BT_GPU_ANGULAR_MOTION_THRESHOLD)\n"
+	"			{\n"
+	"				fAngle = BT_GPU_ANGULAR_MOTION_THRESHOLD / timeStep;\n"
+	"			}\n"
+	"			if(fAngle < 0.001f)\n"
+	"			{\n"
+	"				// use Taylor's expansions of sync function\n"
+	"				axis = angvel * (0.5f*timeStep-(timeStep*timeStep*timeStep)*0.020833333333f * fAngle * fAngle);\n"
+	"			}\n"
+	"			else\n"
+	"			{\n"
+	"				// sync(fAngle) = sin(c*fAngle)/t\n"
+	"				axis = angvel * ( b3Sin(0.5f * fAngle * timeStep) / fAngle);\n"
+	"			}\n"
+	"			b3Quat dorn;\n"
+	"			dorn.x = axis.x;\n"
+	"			dorn.y = axis.y;\n"
+	"			dorn.z = axis.z;\n"
+	"			dorn.w = b3Cos(fAngle * timeStep * 0.5f);\n"
+	"			b3Quat orn0 = body->m_quat;\n"
+	"			b3Quat predictedOrn = b3QuatMul(dorn, orn0);\n"
+	"			predictedOrn = b3QuatNormalized(predictedOrn);\n"
+	"			body->m_quat=predictedOrn;\n"
+	"		}\n"
+	"		//apply gravity\n"
+	"		body->m_linVel += gravityAcceleration * timeStep;\n"
+	"		//linear velocity		\n"
+	"		body->m_pos +=  body->m_linVel * timeStep;\n"
+	"		\n"
+	"	}\n"
+	"	\n"
+	"}\n"
+	"__kernel void \n"
+	"  integrateTransformsKernel( __global b3RigidBodyData_t* bodies,const int numNodes, float timeStep, float angularDamping, float4 gravityAcceleration)\n"
+	"{\n"
+	"	int nodeID = get_global_id(0);\n"
+	"	\n"
+	"	if( nodeID < numNodes)\n"
+	"	{\n"
+	"		integrateSingleTransform(bodies,nodeID, timeStep, angularDamping,gravityAcceleration);\n"
+	"	}\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/jointSolver.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/jointSolver.h
index d48ecf6ea66..c94b55851e1 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/jointSolver.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/jointSolver.h
@@ -1,721 +1,720 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* solveConstraintRowsCL= \
-"/*\n"
-"Copyright (c) 2013 Advanced Micro Devices, Inc.  \n"
-"This software is provided 'as-is', without any express or implied warranty.\n"
-"In no event will the authors be held liable for any damages arising from the use of this software.\n"
-"Permission is granted to anyone to use this software for any purpose, \n"
-"including commercial applications, and to alter it and redistribute it freely, \n"
-"subject to the following restrictions:\n"
-"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
-"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
-"3. This notice may not be removed or altered from any source distribution.\n"
-"*/\n"
-"//Originally written by Erwin Coumans\n"
-"#define B3_CONSTRAINT_FLAG_ENABLED 1\n"
-"#define B3_GPU_POINT2POINT_CONSTRAINT_TYPE 3\n"
-"#define B3_GPU_FIXED_CONSTRAINT_TYPE 4\n"
-"#define MOTIONCLAMP 100000 //unused, for debugging/safety in case constraint solver fails\n"
-"#define B3_INFINITY 1e30f\n"
-"#define mymake_float4 (float4)\n"
-"__inline float dot3F4(float4 a, float4 b)\n"
-"{\n"
-"	float4 a1 = mymake_float4(a.xyz,0.f);\n"
-"	float4 b1 = mymake_float4(b.xyz,0.f);\n"
-"	return dot(a1, b1);\n"
-"}\n"
-"typedef float4 Quaternion;\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_row[3];\n"
-"}Matrix3x3;\n"
-"__inline\n"
-"float4 mtMul1(Matrix3x3 a, float4 b);\n"
-"__inline\n"
-"float4 mtMul3(float4 a, Matrix3x3 b);\n"
-"__inline\n"
-"float4 mtMul1(Matrix3x3 a, float4 b)\n"
-"{\n"
-"	float4 ans;\n"
-"	ans.x = dot3F4( a.m_row[0], b );\n"
-"	ans.y = dot3F4( a.m_row[1], b );\n"
-"	ans.z = dot3F4( a.m_row[2], b );\n"
-"	ans.w = 0.f;\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"float4 mtMul3(float4 a, Matrix3x3 b)\n"
-"{\n"
-"	float4 colx = mymake_float4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
-"	float4 coly = mymake_float4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
-"	float4 colz = mymake_float4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
-"	float4 ans;\n"
-"	ans.x = dot3F4( a, colx );\n"
-"	ans.y = dot3F4( a, coly );\n"
-"	ans.z = dot3F4( a, colz );\n"
-"	return ans;\n"
-"}\n"
-"typedef struct\n"
-"{\n"
-"	Matrix3x3 m_invInertiaWorld;\n"
-"	Matrix3x3 m_initInvInertia;\n"
-"} BodyInertia;\n"
-"typedef struct\n"
-"{\n"
-"	Matrix3x3 m_basis;//orientation\n"
-"	float4	m_origin;//transform\n"
-"}b3Transform;\n"
-"typedef struct\n"
-"{\n"
-"//	b3Transform		m_worldTransformUnused;\n"
-"	float4		m_deltaLinearVelocity;\n"
-"	float4		m_deltaAngularVelocity;\n"
-"	float4		m_angularFactor;\n"
-"	float4		m_linearFactor;\n"
-"	float4		m_invMass;\n"
-"	float4		m_pushVelocity;\n"
-"	float4		m_turnVelocity;\n"
-"	float4		m_linearVelocity;\n"
-"	float4		m_angularVelocity;\n"
-"	union \n"
-"	{\n"
-"		void*	m_originalBody;\n"
-"		int		m_originalBodyIndex;\n"
-"	};\n"
-"	int padding[3];\n"
-"} b3GpuSolverBody;\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_pos;\n"
-"	Quaternion m_quat;\n"
-"	float4 m_linVel;\n"
-"	float4 m_angVel;\n"
-"	unsigned int m_shapeIdx;\n"
-"	float m_invMass;\n"
-"	float m_restituitionCoeff;\n"
-"	float m_frictionCoeff;\n"
-"} b3RigidBodyCL;\n"
-"typedef struct\n"
-"{\n"
-"	float4		m_relpos1CrossNormal;\n"
-"	float4		m_contactNormal;\n"
-"	float4		m_relpos2CrossNormal;\n"
-"	//float4		m_contactNormal2;//usually m_contactNormal2 == -m_contactNormal\n"
-"	float4		m_angularComponentA;\n"
-"	float4		m_angularComponentB;\n"
-"	\n"
-"	float	m_appliedPushImpulse;\n"
-"	float	m_appliedImpulse;\n"
-"	int	m_padding1;\n"
-"	int	m_padding2;\n"
-"	float	m_friction;\n"
-"	float	m_jacDiagABInv;\n"
-"	float		m_rhs;\n"
-"	float		m_cfm;\n"
-"	\n"
-"    float		m_lowerLimit;\n"
-"	float		m_upperLimit;\n"
-"	float		m_rhsPenetration;\n"
-"	int			m_originalConstraint;\n"
-"	int	m_overrideNumSolverIterations;\n"
-"    int			m_frictionIndex;\n"
-"	int m_solverBodyIdA;\n"
-"	int m_solverBodyIdB;\n"
-"} b3SolverConstraint;\n"
-"typedef struct \n"
-"{\n"
-"	int m_bodyAPtrAndSignBit;\n"
-"	int m_bodyBPtrAndSignBit;\n"
-"	int m_originalConstraintIndex;\n"
-"	int m_batchId;\n"
-"} b3BatchConstraint;\n"
-"typedef struct \n"
-"{\n"
-"	int				m_constraintType;\n"
-"	int				m_rbA;\n"
-"	int				m_rbB;\n"
-"	float			m_breakingImpulseThreshold;\n"
-"	float4 m_pivotInA;\n"
-"	float4 m_pivotInB;\n"
-"	Quaternion m_relTargetAB;\n"
-"	int	m_flags;\n"
-"	int m_padding[3];\n"
-"} b3GpuGenericConstraint;\n"
-"/*b3Transform	getWorldTransform(b3RigidBodyCL* rb)\n"
-"{\n"
-"	b3Transform newTrans;\n"
-"	newTrans.setOrigin(rb->m_pos);\n"
-"	newTrans.setRotation(rb->m_quat);\n"
-"	return newTrans;\n"
-"}*/\n"
-"__inline\n"
-"float4 cross3(float4 a, float4 b)\n"
-"{\n"
-"	return cross(a,b);\n"
-"}\n"
-"__inline\n"
-"float4 fastNormalize4(float4 v)\n"
-"{\n"
-"	v = mymake_float4(v.xyz,0.f);\n"
-"	return fast_normalize(v);\n"
-"}\n"
-"__inline\n"
-"Quaternion qtMul(Quaternion a, Quaternion b);\n"
-"__inline\n"
-"Quaternion qtNormalize(Quaternion in);\n"
-"__inline\n"
-"float4 qtRotate(Quaternion q, float4 vec);\n"
-"__inline\n"
-"Quaternion qtInvert(Quaternion q);\n"
-"__inline\n"
-"Quaternion qtMul(Quaternion a, Quaternion b)\n"
-"{\n"
-"	Quaternion ans;\n"
-"	ans = cross3( a, b );\n"
-"	ans += a.w*b+b.w*a;\n"
-"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
-"	ans.w = a.w*b.w - dot3F4(a, b);\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"Quaternion qtNormalize(Quaternion in)\n"
-"{\n"
-"	return fastNormalize4(in);\n"
-"//	in /= length( in );\n"
-"//	return in;\n"
-"}\n"
-"__inline\n"
-"float4 qtRotate(Quaternion q, float4 vec)\n"
-"{\n"
-"	Quaternion qInv = qtInvert( q );\n"
-"	float4 vcpy = vec;\n"
-"	vcpy.w = 0.f;\n"
-"	float4 out = qtMul(qtMul(q,vcpy),qInv);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"Quaternion qtInvert(Quaternion q)\n"
-"{\n"
-"	return (Quaternion)(-q.xyz, q.w);\n"
-"}\n"
-"__inline void internalApplyImpulse(__global b3GpuSolverBody* body,  float4 linearComponent, float4 angularComponent,float impulseMagnitude)\n"
-"{\n"
-"	body->m_deltaLinearVelocity += linearComponent*impulseMagnitude*body->m_linearFactor;\n"
-"	body->m_deltaAngularVelocity += angularComponent*(impulseMagnitude*body->m_angularFactor);\n"
-"}\n"
-"void resolveSingleConstraintRowGeneric(__global b3GpuSolverBody* body1, __global b3GpuSolverBody* body2, __global b3SolverConstraint* c)\n"
-"{\n"
-"	float deltaImpulse = c->m_rhs-c->m_appliedImpulse*c->m_cfm;\n"
-"	float deltaVel1Dotn	=	dot3F4(c->m_contactNormal,body1->m_deltaLinearVelocity) 	+ dot3F4(c->m_relpos1CrossNormal,body1->m_deltaAngularVelocity);\n"
-"	float deltaVel2Dotn	=	-dot3F4(c->m_contactNormal,body2->m_deltaLinearVelocity) + dot3F4(c->m_relpos2CrossNormal,body2->m_deltaAngularVelocity);\n"
-"	deltaImpulse	-=	deltaVel1Dotn*c->m_jacDiagABInv;\n"
-"	deltaImpulse	-=	deltaVel2Dotn*c->m_jacDiagABInv;\n"
-"	float sum = c->m_appliedImpulse + deltaImpulse;\n"
-"	if (sum < c->m_lowerLimit)\n"
-"	{\n"
-"		deltaImpulse = c->m_lowerLimit-c->m_appliedImpulse;\n"
-"		c->m_appliedImpulse = c->m_lowerLimit;\n"
-"	}\n"
-"	else if (sum > c->m_upperLimit) \n"
-"	{\n"
-"		deltaImpulse = c->m_upperLimit-c->m_appliedImpulse;\n"
-"		c->m_appliedImpulse = c->m_upperLimit;\n"
-"	}\n"
-"	else\n"
-"	{\n"
-"		c->m_appliedImpulse = sum;\n"
-"	}\n"
-"	internalApplyImpulse(body1,c->m_contactNormal*body1->m_invMass,c->m_angularComponentA,deltaImpulse);\n"
-"	internalApplyImpulse(body2,-c->m_contactNormal*body2->m_invMass,c->m_angularComponentB,deltaImpulse);\n"
-"}\n"
-"__kernel void solveJointConstraintRows(__global b3GpuSolverBody* solverBodies,\n"
-"					  __global b3BatchConstraint* batchConstraints,\n"
-"					  	__global b3SolverConstraint* rows,\n"
-"						__global unsigned int* numConstraintRowsInfo1, \n"
-"						__global unsigned int* rowOffsets,\n"
-"						__global b3GpuGenericConstraint* constraints,\n"
-"						int batchOffset,\n"
-"						int numConstraintsInBatch\n"
-"                      )\n"
-"{\n"
-"	int b = get_global_id(0);\n"
-"	if (b>=numConstraintsInBatch)\n"
-"		return;\n"
-"	__global b3BatchConstraint* c = &batchConstraints[b+batchOffset];\n"
-"	int originalConstraintIndex = c->m_originalConstraintIndex;\n"
-"	if (constraints[originalConstraintIndex].m_flags&B3_CONSTRAINT_FLAG_ENABLED)\n"
-"	{\n"
-"		int numConstraintRows = numConstraintRowsInfo1[originalConstraintIndex];\n"
-"		int rowOffset = rowOffsets[originalConstraintIndex];\n"
-"		for (int jj=0;jj<numConstraintRows;jj++)\n"
-"		{\n"
-"			__global b3SolverConstraint* constraint = &rows[rowOffset+jj];\n"
-"			resolveSingleConstraintRowGeneric(&solverBodies[constraint->m_solverBodyIdA],&solverBodies[constraint->m_solverBodyIdB],constraint);\n"
-"		}\n"
-"	}\n"
-"};\n"
-"__kernel void initSolverBodies(__global b3GpuSolverBody* solverBodies,__global b3RigidBodyCL* bodiesCL, int numBodies)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	if (i>=numBodies)\n"
-"		return;\n"
-"	__global b3GpuSolverBody* solverBody = &solverBodies[i];\n"
-"	__global b3RigidBodyCL* bodyCL = &bodiesCL[i];\n"
-"	solverBody->m_deltaLinearVelocity = (float4)(0.f,0.f,0.f,0.f);\n"
-"	solverBody->m_deltaAngularVelocity  = (float4)(0.f,0.f,0.f,0.f);\n"
-"	solverBody->m_pushVelocity = (float4)(0.f,0.f,0.f,0.f);\n"
-"	solverBody->m_pushVelocity = (float4)(0.f,0.f,0.f,0.f);\n"
-"	solverBody->m_invMass = (float4)(bodyCL->m_invMass,bodyCL->m_invMass,bodyCL->m_invMass,0.f);\n"
-"	solverBody->m_originalBodyIndex = i;\n"
-"	solverBody->m_angularFactor = (float4)(1,1,1,0);\n"
-"	solverBody->m_linearFactor = (float4) (1,1,1,0);\n"
-"	solverBody->m_linearVelocity = bodyCL->m_linVel;\n"
-"	solverBody->m_angularVelocity = bodyCL->m_angVel;\n"
-"}\n"
-"__kernel void breakViolatedConstraintsKernel(__global b3GpuGenericConstraint* constraints, __global unsigned int* numConstraintRows, __global unsigned int* rowOffsets, __global b3SolverConstraint* rows, int numConstraints)\n"
-"{\n"
-"	int cid = get_global_id(0);\n"
-"	if (cid>=numConstraints)\n"
-"		return;\n"
-"	int numRows = numConstraintRows[cid];\n"
-"	if (numRows)\n"
-"	{\n"
-"		for (int i=0;i<numRows;i++)\n"
-"		{\n"
-"			int rowIndex = rowOffsets[cid]+i;\n"
-"			float breakingThreshold = constraints[cid].m_breakingImpulseThreshold;\n"
-"			if (fabs(rows[rowIndex].m_appliedImpulse) >= breakingThreshold)\n"
-"			{\n"
-"				constraints[cid].m_flags =0;//&= ~B3_CONSTRAINT_FLAG_ENABLED;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"}\n"
-"__kernel void getInfo1Kernel(__global unsigned int* infos, __global b3GpuGenericConstraint* constraints, int numConstraints)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	if (i>=numConstraints)\n"
-"		return;\n"
-"	__global b3GpuGenericConstraint* constraint = &constraints[i];\n"
-"	switch (constraint->m_constraintType)\n"
-"	{\n"
-"		case B3_GPU_POINT2POINT_CONSTRAINT_TYPE:\n"
-"		{\n"
-"			infos[i] = 3;\n"
-"			break;\n"
-"		}\n"
-"		case B3_GPU_FIXED_CONSTRAINT_TYPE:\n"
-"		{\n"
-"			infos[i] = 6;\n"
-"			break;\n"
-"		}\n"
-"		default:\n"
-"		{\n"
-"		}\n"
-"	}\n"
-"}\n"
-"__kernel void initBatchConstraintsKernel(__global unsigned int* numConstraintRows, __global unsigned int* rowOffsets, \n"
-"										__global b3BatchConstraint* batchConstraints, \n"
-"										__global b3GpuGenericConstraint* constraints,\n"
-"										__global b3RigidBodyCL* bodies,\n"
-"										int numConstraints)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	if (i>=numConstraints)\n"
-"		return;\n"
-"	int rbA = constraints[i].m_rbA;\n"
-"	int rbB = constraints[i].m_rbB;\n"
-"	batchConstraints[i].m_bodyAPtrAndSignBit = bodies[rbA].m_invMass != 0.f ? rbA : -rbA;\n"
-"	batchConstraints[i].m_bodyBPtrAndSignBit = bodies[rbB].m_invMass != 0.f ? rbB : -rbB;\n"
-"	batchConstraints[i].m_batchId = -1;\n"
-"	batchConstraints[i].m_originalConstraintIndex = i;\n"
-"}\n"
-"typedef struct\n"
-"{\n"
-"	// integrator parameters: frames per second (1/stepsize), default error\n"
-"	// reduction parameter (0..1).\n"
-"	float fps,erp;\n"
-"	// for the first and second body, pointers to two (linear and angular)\n"
-"	// n*3 jacobian sub matrices, stored by rows. these matrices will have\n"
-"	// been initialized to 0 on entry. if the second body is zero then the\n"
-"	// J2xx pointers may be 0.\n"
-"	union \n"
-"	{\n"
-"		__global float4* m_J1linearAxisFloat4;\n"
-"		__global float* m_J1linearAxis;\n"
-"	};\n"
-"	union\n"
-"	{\n"
-"		__global float4* m_J1angularAxisFloat4;\n"
-"		__global float* m_J1angularAxis;\n"
-"	};\n"
-"	union\n"
-"	{\n"
-"	__global float4* m_J2linearAxisFloat4;\n"
-"	__global float* m_J2linearAxis;\n"
-"	};\n"
-"	union\n"
-"	{\n"
-"		__global float4* m_J2angularAxisFloat4;\n"
-"		__global float* m_J2angularAxis;\n"
-"	};\n"
-"	// elements to jump from one row to the next in J's\n"
-"	int rowskip;\n"
-"	// right hand sides of the equation J*v = c + cfm * lambda. cfm is the\n"
-"	// \"constraint force mixing\" vector. c is set to zero on entry, cfm is\n"
-"	// set to a constant value (typically very small or zero) value on entry.\n"
-"	__global float* m_constraintError;\n"
-"	__global float* cfm;\n"
-"	// lo and hi limits for variables (set to -/+ infinity on entry).\n"
-"	__global float* m_lowerLimit;\n"
-"	__global float* m_upperLimit;\n"
-"	// findex vector for variables. see the LCP solver interface for a\n"
-"	// description of what this does. this is set to -1 on entry.\n"
-"	// note that the returned indexes are relative to the first index of\n"
-"	// the constraint.\n"
-"	__global int *findex;\n"
-"	// number of solver iterations\n"
-"	int m_numIterations;\n"
-"	//damping of the velocity\n"
-"	float	m_damping;\n"
-"} b3GpuConstraintInfo2;\n"
-"void	getSkewSymmetricMatrix(float4 vecIn, __global float4* v0,__global float4* v1,__global float4* v2)\n"
-"{\n"
-"	*v0 = (float4)(0.		,-vecIn.z		,vecIn.y,0.f);\n"
-"	*v1 = (float4)(vecIn.z	,0.			,-vecIn.x,0.f);\n"
-"	*v2 = (float4)(-vecIn.y	,vecIn.x	,0.f,0.f);\n"
-"}\n"
-"void getInfo2Point2Point(__global b3GpuGenericConstraint* constraint,b3GpuConstraintInfo2* info,__global b3RigidBodyCL* bodies)\n"
-"{\n"
-"	float4 posA = bodies[constraint->m_rbA].m_pos;\n"
-"	Quaternion rotA = bodies[constraint->m_rbA].m_quat;\n"
-"	float4 posB = bodies[constraint->m_rbB].m_pos;\n"
-"	Quaternion rotB = bodies[constraint->m_rbB].m_quat;\n"
-"		// anchor points in global coordinates with respect to body PORs.\n"
-"   \n"
-"    // set jacobian\n"
-"    info->m_J1linearAxis[0] = 1;\n"
-"	info->m_J1linearAxis[info->rowskip+1] = 1;\n"
-"	info->m_J1linearAxis[2*info->rowskip+2] = 1;\n"
-"	float4 a1 = qtRotate(rotA,constraint->m_pivotInA);\n"
-"	{\n"
-"		__global float4* angular0 = (__global float4*)(info->m_J1angularAxis);\n"
-"		__global float4* angular1 = (__global float4*)(info->m_J1angularAxis+info->rowskip);\n"
-"		__global float4* angular2 = (__global float4*)(info->m_J1angularAxis+2*info->rowskip);\n"
-"		float4 a1neg = -a1;\n"
-"		getSkewSymmetricMatrix(a1neg,angular0,angular1,angular2);\n"
-"	}\n"
-"	if (info->m_J2linearAxis)\n"
-"	{\n"
-"		info->m_J2linearAxis[0] = -1;\n"
-"		info->m_J2linearAxis[info->rowskip+1] = -1;\n"
-"		info->m_J2linearAxis[2*info->rowskip+2] = -1;\n"
-"	}\n"
-"	\n"
-"	float4 a2 = qtRotate(rotB,constraint->m_pivotInB);\n"
-"   \n"
-"	{\n"
-"	//	float4 a2n = -a2;\n"
-"		__global float4* angular0 = (__global float4*)(info->m_J2angularAxis);\n"
-"		__global float4* angular1 = (__global float4*)(info->m_J2angularAxis+info->rowskip);\n"
-"		__global float4* angular2 = (__global float4*)(info->m_J2angularAxis+2*info->rowskip);\n"
-"		getSkewSymmetricMatrix(a2,angular0,angular1,angular2);\n"
-"	}\n"
-"    \n"
-"    // set right hand side\n"
-"//	float currERP = (m_flags & B3_P2P_FLAGS_ERP) ? m_erp : info->erp;\n"
-"	float currERP = info->erp;\n"
-"	float k = info->fps * currERP;\n"
-"    int j;\n"
-"	float4 result = a2 + posB - a1 - posA;\n"
-"	float* resultPtr = &result;\n"
-"	for (j=0; j<3; j++)\n"
-"    {\n"
-"        info->m_constraintError[j*info->rowskip] = k * (resultPtr[j]);\n"
-"    }\n"
-"}\n"
-"Quaternion nearest( Quaternion first, Quaternion qd)\n"
-"{\n"
-"	Quaternion diff,sum;\n"
-"	diff = first- qd;\n"
-"	sum = first + qd;\n"
-"	\n"
-"	if( dot(diff,diff) < dot(sum,sum) )\n"
-"		return qd;\n"
-"	return (-qd);\n"
-"}\n"
-"float b3Acos(float x) \n"
-"{ \n"
-"	if (x<-1)	\n"
-"		x=-1; \n"
-"	if (x>1)	\n"
-"		x=1;\n"
-"	return acos(x); \n"
-"}\n"
-"float getAngle(Quaternion orn)\n"
-"{\n"
-"	if (orn.w>=1.f)\n"
-"		orn.w=1.f;\n"
-"	float s = 2.f * b3Acos(orn.w);\n"
-"	return s;\n"
-"}\n"
-"void calculateDiffAxisAngleQuaternion( Quaternion orn0,Quaternion orn1a,float4* axis,float* angle)\n"
-"{\n"
-"	Quaternion orn1 = nearest(orn0,orn1a);\n"
-"	\n"
-"	Quaternion dorn = qtMul(orn1,qtInvert(orn0));\n"
-"	*angle = getAngle(dorn);\n"
-"	*axis = (float4)(dorn.x,dorn.y,dorn.z,0.f);\n"
-"	\n"
-"	//check for axis length\n"
-"	float len = dot3F4(*axis,*axis);\n"
-"	if (len < FLT_EPSILON*FLT_EPSILON)\n"
-"		*axis = (float4)(1,0,0,0);\n"
-"	else\n"
-"		*axis /= sqrt(len);\n"
-"}\n"
-"void getInfo2FixedOrientation(__global b3GpuGenericConstraint* constraint,b3GpuConstraintInfo2* info,__global b3RigidBodyCL* bodies, int start_row)\n"
-"{\n"
-"	Quaternion worldOrnA = bodies[constraint->m_rbA].m_quat;\n"
-"	Quaternion worldOrnB = bodies[constraint->m_rbB].m_quat;\n"
-"	int s = info->rowskip;\n"
-"	int start_index = start_row * s;\n"
-"	// 3 rows to make body rotations equal\n"
-"	info->m_J1angularAxis[start_index] = 1;\n"
-"	info->m_J1angularAxis[start_index + s + 1] = 1;\n"
-"	info->m_J1angularAxis[start_index + s*2+2] = 1;\n"
-"	if ( info->m_J2angularAxis)\n"
-"	{\n"
-"		info->m_J2angularAxis[start_index] = -1;\n"
-"		info->m_J2angularAxis[start_index + s+1] = -1;\n"
-"		info->m_J2angularAxis[start_index + s*2+2] = -1;\n"
-"	}\n"
-"	\n"
-"	float currERP = info->erp;\n"
-"	float k = info->fps * currERP;\n"
-"	float4 diff;\n"
-"	float angle;\n"
-"	float4 qrelCur = qtMul(worldOrnA,qtInvert(worldOrnB));\n"
-"	\n"
-"	calculateDiffAxisAngleQuaternion(constraint->m_relTargetAB,qrelCur,&diff,&angle);\n"
-"	diff*=-angle;\n"
-"		\n"
-"	float* resultPtr = &diff;\n"
-"	\n"
-"	for (int j=0; j<3; j++)\n"
-"    {\n"
-"        info->m_constraintError[(3+j)*info->rowskip] = k * resultPtr[j];\n"
-"    }\n"
-"	\n"
-"}\n"
-"__kernel void writeBackVelocitiesKernel(__global b3RigidBodyCL* bodies,__global b3GpuSolverBody* solverBodies,int numBodies)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	if (i>=numBodies)\n"
-"		return;\n"
-"	if (bodies[i].m_invMass)\n"
-"	{\n"
-"//		if (length(solverBodies[i].m_deltaLinearVelocity)<MOTIONCLAMP)\n"
-"		{\n"
-"			bodies[i].m_linVel += solverBodies[i].m_deltaLinearVelocity;\n"
-"		}\n"
-"//		if (length(solverBodies[i].m_deltaAngularVelocity)<MOTIONCLAMP)\n"
-"		{\n"
-"			bodies[i].m_angVel += solverBodies[i].m_deltaAngularVelocity;\n"
-"		} \n"
-"	}\n"
-"}\n"
-"__kernel void getInfo2Kernel(__global b3SolverConstraint* solverConstraintRows, \n"
-"							__global unsigned int* infos, \n"
-"							__global unsigned int* constraintRowOffsets, \n"
-"							__global b3GpuGenericConstraint* constraints, \n"
-"							__global b3BatchConstraint* batchConstraints, \n"
-"							__global b3RigidBodyCL* bodies,\n"
-"							__global BodyInertia* inertias,\n"
-"							__global b3GpuSolverBody* solverBodies,\n"
-"							float timeStep,\n"
-"							float globalErp,\n"
-"							float globalCfm,\n"
-"							float globalDamping,\n"
-"							int globalNumIterations,\n"
-"							int numConstraints)\n"
-"{\n"
-"	int i = get_global_id(0);\n"
-"	if (i>=numConstraints)\n"
-"		return;\n"
-"		\n"
-"	//for now, always initialize the batch info\n"
-"	int info1 = infos[i];\n"
-"			\n"
-"	__global b3SolverConstraint* currentConstraintRow = &solverConstraintRows[constraintRowOffsets[i]];\n"
-"	__global b3GpuGenericConstraint* constraint = &constraints[i];\n"
-"	__global b3RigidBodyCL* rbA = &bodies[ constraint->m_rbA];\n"
-"	__global b3RigidBodyCL* rbB = &bodies[ constraint->m_rbB];\n"
-"	int solverBodyIdA = constraint->m_rbA;\n"
-"	int solverBodyIdB = constraint->m_rbB;\n"
-"	__global b3GpuSolverBody* bodyAPtr = &solverBodies[solverBodyIdA];\n"
-"	__global b3GpuSolverBody* bodyBPtr = &solverBodies[solverBodyIdB];\n"
-"	if (rbA->m_invMass)\n"
-"	{\n"
-"		batchConstraints[i].m_bodyAPtrAndSignBit = solverBodyIdA;\n"
-"	} else\n"
-"	{\n"
-"//			if (!solverBodyIdA)\n"
-"//				m_staticIdx = 0;\n"
-"		batchConstraints[i].m_bodyAPtrAndSignBit = -solverBodyIdA;\n"
-"	}\n"
-"	if (rbB->m_invMass)\n"
-"	{\n"
-"		batchConstraints[i].m_bodyBPtrAndSignBit = solverBodyIdB;\n"
-"	} else\n"
-"	{\n"
-"//			if (!solverBodyIdB)\n"
-"//				m_staticIdx = 0;\n"
-"		batchConstraints[i].m_bodyBPtrAndSignBit = -solverBodyIdB;\n"
-"	}\n"
-"	if (info1)\n"
-"	{\n"
-"		int overrideNumSolverIterations = 0;//constraint->getOverrideNumSolverIterations() > 0 ? constraint->getOverrideNumSolverIterations() : infoGlobal.m_numIterations;\n"
-"//		if (overrideNumSolverIterations>m_maxOverrideNumSolverIterations)\n"
-"	//		m_maxOverrideNumSolverIterations = overrideNumSolverIterations;\n"
-"		int j;\n"
-"		for ( j=0;j<info1;j++)\n"
-"		{\n"
-"//			memset(&currentConstraintRow[j],0,sizeof(b3SolverConstraint));\n"
-"			currentConstraintRow[j].m_angularComponentA = (float4)(0,0,0,0);\n"
-"			currentConstraintRow[j].m_angularComponentB = (float4)(0,0,0,0);\n"
-"			currentConstraintRow[j].m_appliedImpulse = 0.f;\n"
-"			currentConstraintRow[j].m_appliedPushImpulse = 0.f;\n"
-"			currentConstraintRow[j].m_cfm = 0.f;\n"
-"			currentConstraintRow[j].m_contactNormal = (float4)(0,0,0,0);\n"
-"			currentConstraintRow[j].m_friction = 0.f;\n"
-"			currentConstraintRow[j].m_frictionIndex = 0;\n"
-"			currentConstraintRow[j].m_jacDiagABInv = 0.f;\n"
-"			currentConstraintRow[j].m_lowerLimit = 0.f;\n"
-"			currentConstraintRow[j].m_upperLimit = 0.f;\n"
-"			currentConstraintRow[j].m_originalConstraint = i;\n"
-"			currentConstraintRow[j].m_overrideNumSolverIterations = 0;\n"
-"			currentConstraintRow[j].m_relpos1CrossNormal = (float4)(0,0,0,0);\n"
-"			currentConstraintRow[j].m_relpos2CrossNormal = (float4)(0,0,0,0);\n"
-"			currentConstraintRow[j].m_rhs = 0.f;\n"
-"			currentConstraintRow[j].m_rhsPenetration = 0.f;\n"
-"			currentConstraintRow[j].m_solverBodyIdA = 0;\n"
-"			currentConstraintRow[j].m_solverBodyIdB = 0;\n"
-"							\n"
-"			currentConstraintRow[j].m_lowerLimit = -B3_INFINITY;\n"
-"			currentConstraintRow[j].m_upperLimit = B3_INFINITY;\n"
-"			currentConstraintRow[j].m_appliedImpulse = 0.f;\n"
-"			currentConstraintRow[j].m_appliedPushImpulse = 0.f;\n"
-"			currentConstraintRow[j].m_solverBodyIdA = solverBodyIdA;\n"
-"			currentConstraintRow[j].m_solverBodyIdB = solverBodyIdB;\n"
-"			currentConstraintRow[j].m_overrideNumSolverIterations = overrideNumSolverIterations;		\n"
-"		}\n"
-"		bodyAPtr->m_deltaLinearVelocity = (float4)(0,0,0,0);\n"
-"		bodyAPtr->m_deltaAngularVelocity = (float4)(0,0,0,0);\n"
-"		bodyAPtr->m_pushVelocity = (float4)(0,0,0,0);\n"
-"		bodyAPtr->m_turnVelocity = (float4)(0,0,0,0);\n"
-"		bodyBPtr->m_deltaLinearVelocity = (float4)(0,0,0,0);\n"
-"		bodyBPtr->m_deltaAngularVelocity = (float4)(0,0,0,0);\n"
-"		bodyBPtr->m_pushVelocity = (float4)(0,0,0,0);\n"
-"		bodyBPtr->m_turnVelocity  = (float4)(0,0,0,0);\n"
-"		int rowskip = sizeof(b3SolverConstraint)/sizeof(float);//check this\n"
-"		\n"
-"		b3GpuConstraintInfo2 info2;\n"
-"		info2.fps = 1.f/timeStep;\n"
-"		info2.erp = globalErp;\n"
-"		info2.m_J1linearAxisFloat4 = &currentConstraintRow->m_contactNormal;\n"
-"		info2.m_J1angularAxisFloat4 = &currentConstraintRow->m_relpos1CrossNormal;\n"
-"		info2.m_J2linearAxisFloat4 = 0;\n"
-"		info2.m_J2angularAxisFloat4 = &currentConstraintRow->m_relpos2CrossNormal;\n"
-"		info2.rowskip = sizeof(b3SolverConstraint)/sizeof(float);//check this\n"
-"		///the size of b3SolverConstraint needs be a multiple of float\n"
-"//		b3Assert(info2.rowskip*sizeof(float)== sizeof(b3SolverConstraint));\n"
-"		info2.m_constraintError = &currentConstraintRow->m_rhs;\n"
-"		currentConstraintRow->m_cfm = globalCfm;\n"
-"		info2.m_damping = globalDamping;\n"
-"		info2.cfm = &currentConstraintRow->m_cfm;\n"
-"		info2.m_lowerLimit = &currentConstraintRow->m_lowerLimit;\n"
-"		info2.m_upperLimit = &currentConstraintRow->m_upperLimit;\n"
-"		info2.m_numIterations = globalNumIterations;\n"
-"		switch (constraint->m_constraintType)\n"
-"		{\n"
-"			case B3_GPU_POINT2POINT_CONSTRAINT_TYPE:\n"
-"			{\n"
-"				getInfo2Point2Point(constraint,&info2,bodies);\n"
-"				break;\n"
-"			}\n"
-"			case B3_GPU_FIXED_CONSTRAINT_TYPE:\n"
-"			{\n"
-"				getInfo2Point2Point(constraint,&info2,bodies);\n"
-"				getInfo2FixedOrientation(constraint,&info2,bodies,3);\n"
-"				break;\n"
-"			}\n"
-"			default:\n"
-"			{\n"
-"			}\n"
-"		}\n"
-"		///finalize the constraint setup\n"
-"		for ( j=0;j<info1;j++)\n"
-"		{\n"
-"			__global b3SolverConstraint* solverConstraint = &currentConstraintRow[j];\n"
-"			if (solverConstraint->m_upperLimit>=constraint->m_breakingImpulseThreshold)\n"
-"			{\n"
-"				solverConstraint->m_upperLimit = constraint->m_breakingImpulseThreshold;\n"
-"			}\n"
-"			if (solverConstraint->m_lowerLimit<=-constraint->m_breakingImpulseThreshold)\n"
-"			{\n"
-"				solverConstraint->m_lowerLimit = -constraint->m_breakingImpulseThreshold;\n"
-"			}\n"
-"//						solverConstraint->m_originalContactPoint = constraint;\n"
-"							\n"
-"			Matrix3x3 invInertiaWorldA= inertias[constraint->m_rbA].m_invInertiaWorld;\n"
-"			{\n"
-"				//float4 angularFactorA(1,1,1);\n"
-"				float4 ftorqueAxis1 = solverConstraint->m_relpos1CrossNormal;\n"
-"				solverConstraint->m_angularComponentA = mtMul1(invInertiaWorldA,ftorqueAxis1);//*angularFactorA;\n"
-"			}\n"
-"						\n"
-"			Matrix3x3 invInertiaWorldB= inertias[constraint->m_rbB].m_invInertiaWorld;\n"
-"			{\n"
-"				float4 ftorqueAxis2 = solverConstraint->m_relpos2CrossNormal;\n"
-"				solverConstraint->m_angularComponentB = mtMul1(invInertiaWorldB,ftorqueAxis2);//*constraint->m_rbB.getAngularFactor();\n"
-"			}\n"
-"			{\n"
-"				//it is ok to use solverConstraint->m_contactNormal instead of -solverConstraint->m_contactNormal\n"
-"				//because it gets multiplied iMJlB\n"
-"				float4 iMJlA = solverConstraint->m_contactNormal*rbA->m_invMass;\n"
-"				float4 iMJaA = mtMul3(solverConstraint->m_relpos1CrossNormal,invInertiaWorldA);\n"
-"				float4 iMJlB = solverConstraint->m_contactNormal*rbB->m_invMass;//sign of normal?\n"
-"				float4 iMJaB = mtMul3(solverConstraint->m_relpos2CrossNormal,invInertiaWorldB);\n"
-"				float sum = dot3F4(iMJlA,solverConstraint->m_contactNormal);\n"
-"				sum += dot3F4(iMJaA,solverConstraint->m_relpos1CrossNormal);\n"
-"				sum += dot3F4(iMJlB,solverConstraint->m_contactNormal);\n"
-"				sum += dot3F4(iMJaB,solverConstraint->m_relpos2CrossNormal);\n"
-"				float fsum = fabs(sum);\n"
-"				if (fsum>FLT_EPSILON)\n"
-"				{\n"
-"					solverConstraint->m_jacDiagABInv = 1.f/sum;\n"
-"				} else\n"
-"				{\n"
-"					solverConstraint->m_jacDiagABInv = 0.f;\n"
-"				}\n"
-"			}\n"
-"			///fix rhs\n"
-"			///todo: add force/torque accelerators\n"
-"			{\n"
-"				float rel_vel;\n"
-"				float vel1Dotn = dot3F4(solverConstraint->m_contactNormal,rbA->m_linVel) + dot3F4(solverConstraint->m_relpos1CrossNormal,rbA->m_angVel);\n"
-"				float vel2Dotn = -dot3F4(solverConstraint->m_contactNormal,rbB->m_linVel) + dot3F4(solverConstraint->m_relpos2CrossNormal,rbB->m_angVel);\n"
-"				rel_vel = vel1Dotn+vel2Dotn;\n"
-"				float restitution = 0.f;\n"
-"				float positionalError = solverConstraint->m_rhs;//already filled in by getConstraintInfo2\n"
-"				float	velocityError = restitution - rel_vel * info2.m_damping;\n"
-"				float	penetrationImpulse = positionalError*solverConstraint->m_jacDiagABInv;\n"
-"				float	velocityImpulse = velocityError *solverConstraint->m_jacDiagABInv;\n"
-"				solverConstraint->m_rhs = penetrationImpulse+velocityImpulse;\n"
-"				solverConstraint->m_appliedImpulse = 0.f;\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"}\n"
-;
+static const char* solveConstraintRowsCL =
+	"/*\n"
+	"Copyright (c) 2013 Advanced Micro Devices, Inc.  \n"
+	"This software is provided 'as-is', without any express or implied warranty.\n"
+	"In no event will the authors be held liable for any damages arising from the use of this software.\n"
+	"Permission is granted to anyone to use this software for any purpose, \n"
+	"including commercial applications, and to alter it and redistribute it freely, \n"
+	"subject to the following restrictions:\n"
+	"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
+	"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
+	"3. This notice may not be removed or altered from any source distribution.\n"
+	"*/\n"
+	"//Originally written by Erwin Coumans\n"
+	"#define B3_CONSTRAINT_FLAG_ENABLED 1\n"
+	"#define B3_GPU_POINT2POINT_CONSTRAINT_TYPE 3\n"
+	"#define B3_GPU_FIXED_CONSTRAINT_TYPE 4\n"
+	"#define MOTIONCLAMP 100000 //unused, for debugging/safety in case constraint solver fails\n"
+	"#define B3_INFINITY 1e30f\n"
+	"#define mymake_float4 (float4)\n"
+	"__inline float dot3F4(float4 a, float4 b)\n"
+	"{\n"
+	"	float4 a1 = mymake_float4(a.xyz,0.f);\n"
+	"	float4 b1 = mymake_float4(b.xyz,0.f);\n"
+	"	return dot(a1, b1);\n"
+	"}\n"
+	"typedef float4 Quaternion;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_row[3];\n"
+	"}Matrix3x3;\n"
+	"__inline\n"
+	"float4 mtMul1(Matrix3x3 a, float4 b);\n"
+	"__inline\n"
+	"float4 mtMul3(float4 a, Matrix3x3 b);\n"
+	"__inline\n"
+	"float4 mtMul1(Matrix3x3 a, float4 b)\n"
+	"{\n"
+	"	float4 ans;\n"
+	"	ans.x = dot3F4( a.m_row[0], b );\n"
+	"	ans.y = dot3F4( a.m_row[1], b );\n"
+	"	ans.z = dot3F4( a.m_row[2], b );\n"
+	"	ans.w = 0.f;\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"float4 mtMul3(float4 a, Matrix3x3 b)\n"
+	"{\n"
+	"	float4 colx = mymake_float4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
+	"	float4 coly = mymake_float4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
+	"	float4 colz = mymake_float4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
+	"	float4 ans;\n"
+	"	ans.x = dot3F4( a, colx );\n"
+	"	ans.y = dot3F4( a, coly );\n"
+	"	ans.z = dot3F4( a, colz );\n"
+	"	return ans;\n"
+	"}\n"
+	"typedef struct\n"
+	"{\n"
+	"	Matrix3x3 m_invInertiaWorld;\n"
+	"	Matrix3x3 m_initInvInertia;\n"
+	"} BodyInertia;\n"
+	"typedef struct\n"
+	"{\n"
+	"	Matrix3x3 m_basis;//orientation\n"
+	"	float4	m_origin;//transform\n"
+	"}b3Transform;\n"
+	"typedef struct\n"
+	"{\n"
+	"//	b3Transform		m_worldTransformUnused;\n"
+	"	float4		m_deltaLinearVelocity;\n"
+	"	float4		m_deltaAngularVelocity;\n"
+	"	float4		m_angularFactor;\n"
+	"	float4		m_linearFactor;\n"
+	"	float4		m_invMass;\n"
+	"	float4		m_pushVelocity;\n"
+	"	float4		m_turnVelocity;\n"
+	"	float4		m_linearVelocity;\n"
+	"	float4		m_angularVelocity;\n"
+	"	union \n"
+	"	{\n"
+	"		void*	m_originalBody;\n"
+	"		int		m_originalBodyIndex;\n"
+	"	};\n"
+	"	int padding[3];\n"
+	"} b3GpuSolverBody;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_pos;\n"
+	"	Quaternion m_quat;\n"
+	"	float4 m_linVel;\n"
+	"	float4 m_angVel;\n"
+	"	unsigned int m_shapeIdx;\n"
+	"	float m_invMass;\n"
+	"	float m_restituitionCoeff;\n"
+	"	float m_frictionCoeff;\n"
+	"} b3RigidBodyCL;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4		m_relpos1CrossNormal;\n"
+	"	float4		m_contactNormal;\n"
+	"	float4		m_relpos2CrossNormal;\n"
+	"	//float4		m_contactNormal2;//usually m_contactNormal2 == -m_contactNormal\n"
+	"	float4		m_angularComponentA;\n"
+	"	float4		m_angularComponentB;\n"
+	"	\n"
+	"	float	m_appliedPushImpulse;\n"
+	"	float	m_appliedImpulse;\n"
+	"	int	m_padding1;\n"
+	"	int	m_padding2;\n"
+	"	float	m_friction;\n"
+	"	float	m_jacDiagABInv;\n"
+	"	float		m_rhs;\n"
+	"	float		m_cfm;\n"
+	"	\n"
+	"    float		m_lowerLimit;\n"
+	"	float		m_upperLimit;\n"
+	"	float		m_rhsPenetration;\n"
+	"	int			m_originalConstraint;\n"
+	"	int	m_overrideNumSolverIterations;\n"
+	"    int			m_frictionIndex;\n"
+	"	int m_solverBodyIdA;\n"
+	"	int m_solverBodyIdB;\n"
+	"} b3SolverConstraint;\n"
+	"typedef struct \n"
+	"{\n"
+	"	int m_bodyAPtrAndSignBit;\n"
+	"	int m_bodyBPtrAndSignBit;\n"
+	"	int m_originalConstraintIndex;\n"
+	"	int m_batchId;\n"
+	"} b3BatchConstraint;\n"
+	"typedef struct \n"
+	"{\n"
+	"	int				m_constraintType;\n"
+	"	int				m_rbA;\n"
+	"	int				m_rbB;\n"
+	"	float			m_breakingImpulseThreshold;\n"
+	"	float4 m_pivotInA;\n"
+	"	float4 m_pivotInB;\n"
+	"	Quaternion m_relTargetAB;\n"
+	"	int	m_flags;\n"
+	"	int m_padding[3];\n"
+	"} b3GpuGenericConstraint;\n"
+	"/*b3Transform	getWorldTransform(b3RigidBodyCL* rb)\n"
+	"{\n"
+	"	b3Transform newTrans;\n"
+	"	newTrans.setOrigin(rb->m_pos);\n"
+	"	newTrans.setRotation(rb->m_quat);\n"
+	"	return newTrans;\n"
+	"}*/\n"
+	"__inline\n"
+	"float4 cross3(float4 a, float4 b)\n"
+	"{\n"
+	"	return cross(a,b);\n"
+	"}\n"
+	"__inline\n"
+	"float4 fastNormalize4(float4 v)\n"
+	"{\n"
+	"	v = mymake_float4(v.xyz,0.f);\n"
+	"	return fast_normalize(v);\n"
+	"}\n"
+	"__inline\n"
+	"Quaternion qtMul(Quaternion a, Quaternion b);\n"
+	"__inline\n"
+	"Quaternion qtNormalize(Quaternion in);\n"
+	"__inline\n"
+	"float4 qtRotate(Quaternion q, float4 vec);\n"
+	"__inline\n"
+	"Quaternion qtInvert(Quaternion q);\n"
+	"__inline\n"
+	"Quaternion qtMul(Quaternion a, Quaternion b)\n"
+	"{\n"
+	"	Quaternion ans;\n"
+	"	ans = cross3( a, b );\n"
+	"	ans += a.w*b+b.w*a;\n"
+	"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
+	"	ans.w = a.w*b.w - dot3F4(a, b);\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"Quaternion qtNormalize(Quaternion in)\n"
+	"{\n"
+	"	return fastNormalize4(in);\n"
+	"//	in /= length( in );\n"
+	"//	return in;\n"
+	"}\n"
+	"__inline\n"
+	"float4 qtRotate(Quaternion q, float4 vec)\n"
+	"{\n"
+	"	Quaternion qInv = qtInvert( q );\n"
+	"	float4 vcpy = vec;\n"
+	"	vcpy.w = 0.f;\n"
+	"	float4 out = qtMul(qtMul(q,vcpy),qInv);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"Quaternion qtInvert(Quaternion q)\n"
+	"{\n"
+	"	return (Quaternion)(-q.xyz, q.w);\n"
+	"}\n"
+	"__inline void internalApplyImpulse(__global b3GpuSolverBody* body,  float4 linearComponent, float4 angularComponent,float impulseMagnitude)\n"
+	"{\n"
+	"	body->m_deltaLinearVelocity += linearComponent*impulseMagnitude*body->m_linearFactor;\n"
+	"	body->m_deltaAngularVelocity += angularComponent*(impulseMagnitude*body->m_angularFactor);\n"
+	"}\n"
+	"void resolveSingleConstraintRowGeneric(__global b3GpuSolverBody* body1, __global b3GpuSolverBody* body2, __global b3SolverConstraint* c)\n"
+	"{\n"
+	"	float deltaImpulse = c->m_rhs-c->m_appliedImpulse*c->m_cfm;\n"
+	"	float deltaVel1Dotn	=	dot3F4(c->m_contactNormal,body1->m_deltaLinearVelocity) 	+ dot3F4(c->m_relpos1CrossNormal,body1->m_deltaAngularVelocity);\n"
+	"	float deltaVel2Dotn	=	-dot3F4(c->m_contactNormal,body2->m_deltaLinearVelocity) + dot3F4(c->m_relpos2CrossNormal,body2->m_deltaAngularVelocity);\n"
+	"	deltaImpulse	-=	deltaVel1Dotn*c->m_jacDiagABInv;\n"
+	"	deltaImpulse	-=	deltaVel2Dotn*c->m_jacDiagABInv;\n"
+	"	float sum = c->m_appliedImpulse + deltaImpulse;\n"
+	"	if (sum < c->m_lowerLimit)\n"
+	"	{\n"
+	"		deltaImpulse = c->m_lowerLimit-c->m_appliedImpulse;\n"
+	"		c->m_appliedImpulse = c->m_lowerLimit;\n"
+	"	}\n"
+	"	else if (sum > c->m_upperLimit) \n"
+	"	{\n"
+	"		deltaImpulse = c->m_upperLimit-c->m_appliedImpulse;\n"
+	"		c->m_appliedImpulse = c->m_upperLimit;\n"
+	"	}\n"
+	"	else\n"
+	"	{\n"
+	"		c->m_appliedImpulse = sum;\n"
+	"	}\n"
+	"	internalApplyImpulse(body1,c->m_contactNormal*body1->m_invMass,c->m_angularComponentA,deltaImpulse);\n"
+	"	internalApplyImpulse(body2,-c->m_contactNormal*body2->m_invMass,c->m_angularComponentB,deltaImpulse);\n"
+	"}\n"
+	"__kernel void solveJointConstraintRows(__global b3GpuSolverBody* solverBodies,\n"
+	"					  __global b3BatchConstraint* batchConstraints,\n"
+	"					  	__global b3SolverConstraint* rows,\n"
+	"						__global unsigned int* numConstraintRowsInfo1, \n"
+	"						__global unsigned int* rowOffsets,\n"
+	"						__global b3GpuGenericConstraint* constraints,\n"
+	"						int batchOffset,\n"
+	"						int numConstraintsInBatch\n"
+	"                      )\n"
+	"{\n"
+	"	int b = get_global_id(0);\n"
+	"	if (b>=numConstraintsInBatch)\n"
+	"		return;\n"
+	"	__global b3BatchConstraint* c = &batchConstraints[b+batchOffset];\n"
+	"	int originalConstraintIndex = c->m_originalConstraintIndex;\n"
+	"	if (constraints[originalConstraintIndex].m_flags&B3_CONSTRAINT_FLAG_ENABLED)\n"
+	"	{\n"
+	"		int numConstraintRows = numConstraintRowsInfo1[originalConstraintIndex];\n"
+	"		int rowOffset = rowOffsets[originalConstraintIndex];\n"
+	"		for (int jj=0;jj<numConstraintRows;jj++)\n"
+	"		{\n"
+	"			__global b3SolverConstraint* constraint = &rows[rowOffset+jj];\n"
+	"			resolveSingleConstraintRowGeneric(&solverBodies[constraint->m_solverBodyIdA],&solverBodies[constraint->m_solverBodyIdB],constraint);\n"
+	"		}\n"
+	"	}\n"
+	"};\n"
+	"__kernel void initSolverBodies(__global b3GpuSolverBody* solverBodies,__global b3RigidBodyCL* bodiesCL, int numBodies)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	if (i>=numBodies)\n"
+	"		return;\n"
+	"	__global b3GpuSolverBody* solverBody = &solverBodies[i];\n"
+	"	__global b3RigidBodyCL* bodyCL = &bodiesCL[i];\n"
+	"	solverBody->m_deltaLinearVelocity = (float4)(0.f,0.f,0.f,0.f);\n"
+	"	solverBody->m_deltaAngularVelocity  = (float4)(0.f,0.f,0.f,0.f);\n"
+	"	solverBody->m_pushVelocity = (float4)(0.f,0.f,0.f,0.f);\n"
+	"	solverBody->m_pushVelocity = (float4)(0.f,0.f,0.f,0.f);\n"
+	"	solverBody->m_invMass = (float4)(bodyCL->m_invMass,bodyCL->m_invMass,bodyCL->m_invMass,0.f);\n"
+	"	solverBody->m_originalBodyIndex = i;\n"
+	"	solverBody->m_angularFactor = (float4)(1,1,1,0);\n"
+	"	solverBody->m_linearFactor = (float4) (1,1,1,0);\n"
+	"	solverBody->m_linearVelocity = bodyCL->m_linVel;\n"
+	"	solverBody->m_angularVelocity = bodyCL->m_angVel;\n"
+	"}\n"
+	"__kernel void breakViolatedConstraintsKernel(__global b3GpuGenericConstraint* constraints, __global unsigned int* numConstraintRows, __global unsigned int* rowOffsets, __global b3SolverConstraint* rows, int numConstraints)\n"
+	"{\n"
+	"	int cid = get_global_id(0);\n"
+	"	if (cid>=numConstraints)\n"
+	"		return;\n"
+	"	int numRows = numConstraintRows[cid];\n"
+	"	if (numRows)\n"
+	"	{\n"
+	"		for (int i=0;i<numRows;i++)\n"
+	"		{\n"
+	"			int rowIndex = rowOffsets[cid]+i;\n"
+	"			float breakingThreshold = constraints[cid].m_breakingImpulseThreshold;\n"
+	"			if (fabs(rows[rowIndex].m_appliedImpulse) >= breakingThreshold)\n"
+	"			{\n"
+	"				constraints[cid].m_flags =0;//&= ~B3_CONSTRAINT_FLAG_ENABLED;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"__kernel void getInfo1Kernel(__global unsigned int* infos, __global b3GpuGenericConstraint* constraints, int numConstraints)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	if (i>=numConstraints)\n"
+	"		return;\n"
+	"	__global b3GpuGenericConstraint* constraint = &constraints[i];\n"
+	"	switch (constraint->m_constraintType)\n"
+	"	{\n"
+	"		case B3_GPU_POINT2POINT_CONSTRAINT_TYPE:\n"
+	"		{\n"
+	"			infos[i] = 3;\n"
+	"			break;\n"
+	"		}\n"
+	"		case B3_GPU_FIXED_CONSTRAINT_TYPE:\n"
+	"		{\n"
+	"			infos[i] = 6;\n"
+	"			break;\n"
+	"		}\n"
+	"		default:\n"
+	"		{\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"__kernel void initBatchConstraintsKernel(__global unsigned int* numConstraintRows, __global unsigned int* rowOffsets, \n"
+	"										__global b3BatchConstraint* batchConstraints, \n"
+	"										__global b3GpuGenericConstraint* constraints,\n"
+	"										__global b3RigidBodyCL* bodies,\n"
+	"										int numConstraints)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	if (i>=numConstraints)\n"
+	"		return;\n"
+	"	int rbA = constraints[i].m_rbA;\n"
+	"	int rbB = constraints[i].m_rbB;\n"
+	"	batchConstraints[i].m_bodyAPtrAndSignBit = bodies[rbA].m_invMass != 0.f ? rbA : -rbA;\n"
+	"	batchConstraints[i].m_bodyBPtrAndSignBit = bodies[rbB].m_invMass != 0.f ? rbB : -rbB;\n"
+	"	batchConstraints[i].m_batchId = -1;\n"
+	"	batchConstraints[i].m_originalConstraintIndex = i;\n"
+	"}\n"
+	"typedef struct\n"
+	"{\n"
+	"	// integrator parameters: frames per second (1/stepsize), default error\n"
+	"	// reduction parameter (0..1).\n"
+	"	float fps,erp;\n"
+	"	// for the first and second body, pointers to two (linear and angular)\n"
+	"	// n*3 jacobian sub matrices, stored by rows. these matrices will have\n"
+	"	// been initialized to 0 on entry. if the second body is zero then the\n"
+	"	// J2xx pointers may be 0.\n"
+	"	union \n"
+	"	{\n"
+	"		__global float4* m_J1linearAxisFloat4;\n"
+	"		__global float* m_J1linearAxis;\n"
+	"	};\n"
+	"	union\n"
+	"	{\n"
+	"		__global float4* m_J1angularAxisFloat4;\n"
+	"		__global float* m_J1angularAxis;\n"
+	"	};\n"
+	"	union\n"
+	"	{\n"
+	"	__global float4* m_J2linearAxisFloat4;\n"
+	"	__global float* m_J2linearAxis;\n"
+	"	};\n"
+	"	union\n"
+	"	{\n"
+	"		__global float4* m_J2angularAxisFloat4;\n"
+	"		__global float* m_J2angularAxis;\n"
+	"	};\n"
+	"	// elements to jump from one row to the next in J's\n"
+	"	int rowskip;\n"
+	"	// right hand sides of the equation J*v = c + cfm * lambda. cfm is the\n"
+	"	// \"constraint force mixing\" vector. c is set to zero on entry, cfm is\n"
+	"	// set to a constant value (typically very small or zero) value on entry.\n"
+	"	__global float* m_constraintError;\n"
+	"	__global float* cfm;\n"
+	"	// lo and hi limits for variables (set to -/+ infinity on entry).\n"
+	"	__global float* m_lowerLimit;\n"
+	"	__global float* m_upperLimit;\n"
+	"	// findex vector for variables. see the LCP solver interface for a\n"
+	"	// description of what this does. this is set to -1 on entry.\n"
+	"	// note that the returned indexes are relative to the first index of\n"
+	"	// the constraint.\n"
+	"	__global int *findex;\n"
+	"	// number of solver iterations\n"
+	"	int m_numIterations;\n"
+	"	//damping of the velocity\n"
+	"	float	m_damping;\n"
+	"} b3GpuConstraintInfo2;\n"
+	"void	getSkewSymmetricMatrix(float4 vecIn, __global float4* v0,__global float4* v1,__global float4* v2)\n"
+	"{\n"
+	"	*v0 = (float4)(0.		,-vecIn.z		,vecIn.y,0.f);\n"
+	"	*v1 = (float4)(vecIn.z	,0.			,-vecIn.x,0.f);\n"
+	"	*v2 = (float4)(-vecIn.y	,vecIn.x	,0.f,0.f);\n"
+	"}\n"
+	"void getInfo2Point2Point(__global b3GpuGenericConstraint* constraint,b3GpuConstraintInfo2* info,__global b3RigidBodyCL* bodies)\n"
+	"{\n"
+	"	float4 posA = bodies[constraint->m_rbA].m_pos;\n"
+	"	Quaternion rotA = bodies[constraint->m_rbA].m_quat;\n"
+	"	float4 posB = bodies[constraint->m_rbB].m_pos;\n"
+	"	Quaternion rotB = bodies[constraint->m_rbB].m_quat;\n"
+	"		// anchor points in global coordinates with respect to body PORs.\n"
+	"   \n"
+	"    // set jacobian\n"
+	"    info->m_J1linearAxis[0] = 1;\n"
+	"	info->m_J1linearAxis[info->rowskip+1] = 1;\n"
+	"	info->m_J1linearAxis[2*info->rowskip+2] = 1;\n"
+	"	float4 a1 = qtRotate(rotA,constraint->m_pivotInA);\n"
+	"	{\n"
+	"		__global float4* angular0 = (__global float4*)(info->m_J1angularAxis);\n"
+	"		__global float4* angular1 = (__global float4*)(info->m_J1angularAxis+info->rowskip);\n"
+	"		__global float4* angular2 = (__global float4*)(info->m_J1angularAxis+2*info->rowskip);\n"
+	"		float4 a1neg = -a1;\n"
+	"		getSkewSymmetricMatrix(a1neg,angular0,angular1,angular2);\n"
+	"	}\n"
+	"	if (info->m_J2linearAxis)\n"
+	"	{\n"
+	"		info->m_J2linearAxis[0] = -1;\n"
+	"		info->m_J2linearAxis[info->rowskip+1] = -1;\n"
+	"		info->m_J2linearAxis[2*info->rowskip+2] = -1;\n"
+	"	}\n"
+	"	\n"
+	"	float4 a2 = qtRotate(rotB,constraint->m_pivotInB);\n"
+	"   \n"
+	"	{\n"
+	"	//	float4 a2n = -a2;\n"
+	"		__global float4* angular0 = (__global float4*)(info->m_J2angularAxis);\n"
+	"		__global float4* angular1 = (__global float4*)(info->m_J2angularAxis+info->rowskip);\n"
+	"		__global float4* angular2 = (__global float4*)(info->m_J2angularAxis+2*info->rowskip);\n"
+	"		getSkewSymmetricMatrix(a2,angular0,angular1,angular2);\n"
+	"	}\n"
+	"    \n"
+	"    // set right hand side\n"
+	"//	float currERP = (m_flags & B3_P2P_FLAGS_ERP) ? m_erp : info->erp;\n"
+	"	float currERP = info->erp;\n"
+	"	float k = info->fps * currERP;\n"
+	"    int j;\n"
+	"	float4 result = a2 + posB - a1 - posA;\n"
+	"	float* resultPtr = &result;\n"
+	"	for (j=0; j<3; j++)\n"
+	"    {\n"
+	"        info->m_constraintError[j*info->rowskip] = k * (resultPtr[j]);\n"
+	"    }\n"
+	"}\n"
+	"Quaternion nearest( Quaternion first, Quaternion qd)\n"
+	"{\n"
+	"	Quaternion diff,sum;\n"
+	"	diff = first- qd;\n"
+	"	sum = first + qd;\n"
+	"	\n"
+	"	if( dot(diff,diff) < dot(sum,sum) )\n"
+	"		return qd;\n"
+	"	return (-qd);\n"
+	"}\n"
+	"float b3Acos(float x) \n"
+	"{ \n"
+	"	if (x<-1)	\n"
+	"		x=-1; \n"
+	"	if (x>1)	\n"
+	"		x=1;\n"
+	"	return acos(x); \n"
+	"}\n"
+	"float getAngle(Quaternion orn)\n"
+	"{\n"
+	"	if (orn.w>=1.f)\n"
+	"		orn.w=1.f;\n"
+	"	float s = 2.f * b3Acos(orn.w);\n"
+	"	return s;\n"
+	"}\n"
+	"void calculateDiffAxisAngleQuaternion( Quaternion orn0,Quaternion orn1a,float4* axis,float* angle)\n"
+	"{\n"
+	"	Quaternion orn1 = nearest(orn0,orn1a);\n"
+	"	\n"
+	"	Quaternion dorn = qtMul(orn1,qtInvert(orn0));\n"
+	"	*angle = getAngle(dorn);\n"
+	"	*axis = (float4)(dorn.x,dorn.y,dorn.z,0.f);\n"
+	"	\n"
+	"	//check for axis length\n"
+	"	float len = dot3F4(*axis,*axis);\n"
+	"	if (len < FLT_EPSILON*FLT_EPSILON)\n"
+	"		*axis = (float4)(1,0,0,0);\n"
+	"	else\n"
+	"		*axis /= sqrt(len);\n"
+	"}\n"
+	"void getInfo2FixedOrientation(__global b3GpuGenericConstraint* constraint,b3GpuConstraintInfo2* info,__global b3RigidBodyCL* bodies, int start_row)\n"
+	"{\n"
+	"	Quaternion worldOrnA = bodies[constraint->m_rbA].m_quat;\n"
+	"	Quaternion worldOrnB = bodies[constraint->m_rbB].m_quat;\n"
+	"	int s = info->rowskip;\n"
+	"	int start_index = start_row * s;\n"
+	"	// 3 rows to make body rotations equal\n"
+	"	info->m_J1angularAxis[start_index] = 1;\n"
+	"	info->m_J1angularAxis[start_index + s + 1] = 1;\n"
+	"	info->m_J1angularAxis[start_index + s*2+2] = 1;\n"
+	"	if ( info->m_J2angularAxis)\n"
+	"	{\n"
+	"		info->m_J2angularAxis[start_index] = -1;\n"
+	"		info->m_J2angularAxis[start_index + s+1] = -1;\n"
+	"		info->m_J2angularAxis[start_index + s*2+2] = -1;\n"
+	"	}\n"
+	"	\n"
+	"	float currERP = info->erp;\n"
+	"	float k = info->fps * currERP;\n"
+	"	float4 diff;\n"
+	"	float angle;\n"
+	"	float4 qrelCur = qtMul(worldOrnA,qtInvert(worldOrnB));\n"
+	"	\n"
+	"	calculateDiffAxisAngleQuaternion(constraint->m_relTargetAB,qrelCur,&diff,&angle);\n"
+	"	diff*=-angle;\n"
+	"		\n"
+	"	float* resultPtr = &diff;\n"
+	"	\n"
+	"	for (int j=0; j<3; j++)\n"
+	"    {\n"
+	"        info->m_constraintError[(3+j)*info->rowskip] = k * resultPtr[j];\n"
+	"    }\n"
+	"	\n"
+	"}\n"
+	"__kernel void writeBackVelocitiesKernel(__global b3RigidBodyCL* bodies,__global b3GpuSolverBody* solverBodies,int numBodies)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	if (i>=numBodies)\n"
+	"		return;\n"
+	"	if (bodies[i].m_invMass)\n"
+	"	{\n"
+	"//		if (length(solverBodies[i].m_deltaLinearVelocity)<MOTIONCLAMP)\n"
+	"		{\n"
+	"			bodies[i].m_linVel += solverBodies[i].m_deltaLinearVelocity;\n"
+	"		}\n"
+	"//		if (length(solverBodies[i].m_deltaAngularVelocity)<MOTIONCLAMP)\n"
+	"		{\n"
+	"			bodies[i].m_angVel += solverBodies[i].m_deltaAngularVelocity;\n"
+	"		} \n"
+	"	}\n"
+	"}\n"
+	"__kernel void getInfo2Kernel(__global b3SolverConstraint* solverConstraintRows, \n"
+	"							__global unsigned int* infos, \n"
+	"							__global unsigned int* constraintRowOffsets, \n"
+	"							__global b3GpuGenericConstraint* constraints, \n"
+	"							__global b3BatchConstraint* batchConstraints, \n"
+	"							__global b3RigidBodyCL* bodies,\n"
+	"							__global BodyInertia* inertias,\n"
+	"							__global b3GpuSolverBody* solverBodies,\n"
+	"							float timeStep,\n"
+	"							float globalErp,\n"
+	"							float globalCfm,\n"
+	"							float globalDamping,\n"
+	"							int globalNumIterations,\n"
+	"							int numConstraints)\n"
+	"{\n"
+	"	int i = get_global_id(0);\n"
+	"	if (i>=numConstraints)\n"
+	"		return;\n"
+	"		\n"
+	"	//for now, always initialize the batch info\n"
+	"	int info1 = infos[i];\n"
+	"			\n"
+	"	__global b3SolverConstraint* currentConstraintRow = &solverConstraintRows[constraintRowOffsets[i]];\n"
+	"	__global b3GpuGenericConstraint* constraint = &constraints[i];\n"
+	"	__global b3RigidBodyCL* rbA = &bodies[ constraint->m_rbA];\n"
+	"	__global b3RigidBodyCL* rbB = &bodies[ constraint->m_rbB];\n"
+	"	int solverBodyIdA = constraint->m_rbA;\n"
+	"	int solverBodyIdB = constraint->m_rbB;\n"
+	"	__global b3GpuSolverBody* bodyAPtr = &solverBodies[solverBodyIdA];\n"
+	"	__global b3GpuSolverBody* bodyBPtr = &solverBodies[solverBodyIdB];\n"
+	"	if (rbA->m_invMass)\n"
+	"	{\n"
+	"		batchConstraints[i].m_bodyAPtrAndSignBit = solverBodyIdA;\n"
+	"	} else\n"
+	"	{\n"
+	"//			if (!solverBodyIdA)\n"
+	"//				m_staticIdx = 0;\n"
+	"		batchConstraints[i].m_bodyAPtrAndSignBit = -solverBodyIdA;\n"
+	"	}\n"
+	"	if (rbB->m_invMass)\n"
+	"	{\n"
+	"		batchConstraints[i].m_bodyBPtrAndSignBit = solverBodyIdB;\n"
+	"	} else\n"
+	"	{\n"
+	"//			if (!solverBodyIdB)\n"
+	"//				m_staticIdx = 0;\n"
+	"		batchConstraints[i].m_bodyBPtrAndSignBit = -solverBodyIdB;\n"
+	"	}\n"
+	"	if (info1)\n"
+	"	{\n"
+	"		int overrideNumSolverIterations = 0;//constraint->getOverrideNumSolverIterations() > 0 ? constraint->getOverrideNumSolverIterations() : infoGlobal.m_numIterations;\n"
+	"//		if (overrideNumSolverIterations>m_maxOverrideNumSolverIterations)\n"
+	"	//		m_maxOverrideNumSolverIterations = overrideNumSolverIterations;\n"
+	"		int j;\n"
+	"		for ( j=0;j<info1;j++)\n"
+	"		{\n"
+	"//			memset(&currentConstraintRow[j],0,sizeof(b3SolverConstraint));\n"
+	"			currentConstraintRow[j].m_angularComponentA = (float4)(0,0,0,0);\n"
+	"			currentConstraintRow[j].m_angularComponentB = (float4)(0,0,0,0);\n"
+	"			currentConstraintRow[j].m_appliedImpulse = 0.f;\n"
+	"			currentConstraintRow[j].m_appliedPushImpulse = 0.f;\n"
+	"			currentConstraintRow[j].m_cfm = 0.f;\n"
+	"			currentConstraintRow[j].m_contactNormal = (float4)(0,0,0,0);\n"
+	"			currentConstraintRow[j].m_friction = 0.f;\n"
+	"			currentConstraintRow[j].m_frictionIndex = 0;\n"
+	"			currentConstraintRow[j].m_jacDiagABInv = 0.f;\n"
+	"			currentConstraintRow[j].m_lowerLimit = 0.f;\n"
+	"			currentConstraintRow[j].m_upperLimit = 0.f;\n"
+	"			currentConstraintRow[j].m_originalConstraint = i;\n"
+	"			currentConstraintRow[j].m_overrideNumSolverIterations = 0;\n"
+	"			currentConstraintRow[j].m_relpos1CrossNormal = (float4)(0,0,0,0);\n"
+	"			currentConstraintRow[j].m_relpos2CrossNormal = (float4)(0,0,0,0);\n"
+	"			currentConstraintRow[j].m_rhs = 0.f;\n"
+	"			currentConstraintRow[j].m_rhsPenetration = 0.f;\n"
+	"			currentConstraintRow[j].m_solverBodyIdA = 0;\n"
+	"			currentConstraintRow[j].m_solverBodyIdB = 0;\n"
+	"							\n"
+	"			currentConstraintRow[j].m_lowerLimit = -B3_INFINITY;\n"
+	"			currentConstraintRow[j].m_upperLimit = B3_INFINITY;\n"
+	"			currentConstraintRow[j].m_appliedImpulse = 0.f;\n"
+	"			currentConstraintRow[j].m_appliedPushImpulse = 0.f;\n"
+	"			currentConstraintRow[j].m_solverBodyIdA = solverBodyIdA;\n"
+	"			currentConstraintRow[j].m_solverBodyIdB = solverBodyIdB;\n"
+	"			currentConstraintRow[j].m_overrideNumSolverIterations = overrideNumSolverIterations;		\n"
+	"		}\n"
+	"		bodyAPtr->m_deltaLinearVelocity = (float4)(0,0,0,0);\n"
+	"		bodyAPtr->m_deltaAngularVelocity = (float4)(0,0,0,0);\n"
+	"		bodyAPtr->m_pushVelocity = (float4)(0,0,0,0);\n"
+	"		bodyAPtr->m_turnVelocity = (float4)(0,0,0,0);\n"
+	"		bodyBPtr->m_deltaLinearVelocity = (float4)(0,0,0,0);\n"
+	"		bodyBPtr->m_deltaAngularVelocity = (float4)(0,0,0,0);\n"
+	"		bodyBPtr->m_pushVelocity = (float4)(0,0,0,0);\n"
+	"		bodyBPtr->m_turnVelocity  = (float4)(0,0,0,0);\n"
+	"		int rowskip = sizeof(b3SolverConstraint)/sizeof(float);//check this\n"
+	"		\n"
+	"		b3GpuConstraintInfo2 info2;\n"
+	"		info2.fps = 1.f/timeStep;\n"
+	"		info2.erp = globalErp;\n"
+	"		info2.m_J1linearAxisFloat4 = &currentConstraintRow->m_contactNormal;\n"
+	"		info2.m_J1angularAxisFloat4 = &currentConstraintRow->m_relpos1CrossNormal;\n"
+	"		info2.m_J2linearAxisFloat4 = 0;\n"
+	"		info2.m_J2angularAxisFloat4 = &currentConstraintRow->m_relpos2CrossNormal;\n"
+	"		info2.rowskip = sizeof(b3SolverConstraint)/sizeof(float);//check this\n"
+	"		///the size of b3SolverConstraint needs be a multiple of float\n"
+	"//		b3Assert(info2.rowskip*sizeof(float)== sizeof(b3SolverConstraint));\n"
+	"		info2.m_constraintError = &currentConstraintRow->m_rhs;\n"
+	"		currentConstraintRow->m_cfm = globalCfm;\n"
+	"		info2.m_damping = globalDamping;\n"
+	"		info2.cfm = &currentConstraintRow->m_cfm;\n"
+	"		info2.m_lowerLimit = &currentConstraintRow->m_lowerLimit;\n"
+	"		info2.m_upperLimit = &currentConstraintRow->m_upperLimit;\n"
+	"		info2.m_numIterations = globalNumIterations;\n"
+	"		switch (constraint->m_constraintType)\n"
+	"		{\n"
+	"			case B3_GPU_POINT2POINT_CONSTRAINT_TYPE:\n"
+	"			{\n"
+	"				getInfo2Point2Point(constraint,&info2,bodies);\n"
+	"				break;\n"
+	"			}\n"
+	"			case B3_GPU_FIXED_CONSTRAINT_TYPE:\n"
+	"			{\n"
+	"				getInfo2Point2Point(constraint,&info2,bodies);\n"
+	"				getInfo2FixedOrientation(constraint,&info2,bodies,3);\n"
+	"				break;\n"
+	"			}\n"
+	"			default:\n"
+	"			{\n"
+	"			}\n"
+	"		}\n"
+	"		///finalize the constraint setup\n"
+	"		for ( j=0;j<info1;j++)\n"
+	"		{\n"
+	"			__global b3SolverConstraint* solverConstraint = &currentConstraintRow[j];\n"
+	"			if (solverConstraint->m_upperLimit>=constraint->m_breakingImpulseThreshold)\n"
+	"			{\n"
+	"				solverConstraint->m_upperLimit = constraint->m_breakingImpulseThreshold;\n"
+	"			}\n"
+	"			if (solverConstraint->m_lowerLimit<=-constraint->m_breakingImpulseThreshold)\n"
+	"			{\n"
+	"				solverConstraint->m_lowerLimit = -constraint->m_breakingImpulseThreshold;\n"
+	"			}\n"
+	"//						solverConstraint->m_originalContactPoint = constraint;\n"
+	"							\n"
+	"			Matrix3x3 invInertiaWorldA= inertias[constraint->m_rbA].m_invInertiaWorld;\n"
+	"			{\n"
+	"				//float4 angularFactorA(1,1,1);\n"
+	"				float4 ftorqueAxis1 = solverConstraint->m_relpos1CrossNormal;\n"
+	"				solverConstraint->m_angularComponentA = mtMul1(invInertiaWorldA,ftorqueAxis1);//*angularFactorA;\n"
+	"			}\n"
+	"						\n"
+	"			Matrix3x3 invInertiaWorldB= inertias[constraint->m_rbB].m_invInertiaWorld;\n"
+	"			{\n"
+	"				float4 ftorqueAxis2 = solverConstraint->m_relpos2CrossNormal;\n"
+	"				solverConstraint->m_angularComponentB = mtMul1(invInertiaWorldB,ftorqueAxis2);//*constraint->m_rbB.getAngularFactor();\n"
+	"			}\n"
+	"			{\n"
+	"				//it is ok to use solverConstraint->m_contactNormal instead of -solverConstraint->m_contactNormal\n"
+	"				//because it gets multiplied iMJlB\n"
+	"				float4 iMJlA = solverConstraint->m_contactNormal*rbA->m_invMass;\n"
+	"				float4 iMJaA = mtMul3(solverConstraint->m_relpos1CrossNormal,invInertiaWorldA);\n"
+	"				float4 iMJlB = solverConstraint->m_contactNormal*rbB->m_invMass;//sign of normal?\n"
+	"				float4 iMJaB = mtMul3(solverConstraint->m_relpos2CrossNormal,invInertiaWorldB);\n"
+	"				float sum = dot3F4(iMJlA,solverConstraint->m_contactNormal);\n"
+	"				sum += dot3F4(iMJaA,solverConstraint->m_relpos1CrossNormal);\n"
+	"				sum += dot3F4(iMJlB,solverConstraint->m_contactNormal);\n"
+	"				sum += dot3F4(iMJaB,solverConstraint->m_relpos2CrossNormal);\n"
+	"				float fsum = fabs(sum);\n"
+	"				if (fsum>FLT_EPSILON)\n"
+	"				{\n"
+	"					solverConstraint->m_jacDiagABInv = 1.f/sum;\n"
+	"				} else\n"
+	"				{\n"
+	"					solverConstraint->m_jacDiagABInv = 0.f;\n"
+	"				}\n"
+	"			}\n"
+	"			///fix rhs\n"
+	"			///todo: add force/torque accelerators\n"
+	"			{\n"
+	"				float rel_vel;\n"
+	"				float vel1Dotn = dot3F4(solverConstraint->m_contactNormal,rbA->m_linVel) + dot3F4(solverConstraint->m_relpos1CrossNormal,rbA->m_angVel);\n"
+	"				float vel2Dotn = -dot3F4(solverConstraint->m_contactNormal,rbB->m_linVel) + dot3F4(solverConstraint->m_relpos2CrossNormal,rbB->m_angVel);\n"
+	"				rel_vel = vel1Dotn+vel2Dotn;\n"
+	"				float restitution = 0.f;\n"
+	"				float positionalError = solverConstraint->m_rhs;//already filled in by getConstraintInfo2\n"
+	"				float	velocityError = restitution - rel_vel * info2.m_damping;\n"
+	"				float	penetrationImpulse = positionalError*solverConstraint->m_jacDiagABInv;\n"
+	"				float	velocityImpulse = velocityError *solverConstraint->m_jacDiagABInv;\n"
+	"				solverConstraint->m_rhs = penetrationImpulse+velocityImpulse;\n"
+	"				solverConstraint->m_appliedImpulse = 0.f;\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/solveContact.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/solveContact.h
index 15a049992b1..6e14ad51fc2 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/solveContact.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/solveContact.h
@@ -1,393 +1,392 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* solveContactCL= \
-"/*\n"
-"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
-"This software is provided 'as-is', without any express or implied warranty.\n"
-"In no event will the authors be held liable for any damages arising from the use of this software.\n"
-"Permission is granted to anyone to use this software for any purpose, \n"
-"including commercial applications, and to alter it and redistribute it freely, \n"
-"subject to the following restrictions:\n"
-"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
-"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
-"3. This notice may not be removed or altered from any source distribution.\n"
-"*/\n"
-"//Originally written by Takahiro Harada\n"
-"//#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_extended_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics : enable\n"
-"#ifdef cl_ext_atomic_counters_32\n"
-"#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
-"#else\n"
-"#define counter32_t volatile global int*\n"
-"#endif\n"
-"typedef unsigned int u32;\n"
-"typedef unsigned short u16;\n"
-"typedef unsigned char u8;\n"
-"#define GET_GROUP_IDX get_group_id(0)\n"
-"#define GET_LOCAL_IDX get_local_id(0)\n"
-"#define GET_GLOBAL_IDX get_global_id(0)\n"
-"#define GET_GROUP_SIZE get_local_size(0)\n"
-"#define GET_NUM_GROUPS get_num_groups(0)\n"
-"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
-"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
-"#define AtomInc(x) atom_inc(&(x))\n"
-"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
-"#define AppendInc(x, out) out = atomic_inc(x)\n"
-"#define AtomAdd(x, value) atom_add(&(x), value)\n"
-"#define AtomCmpxhg(x, cmp, value) atom_cmpxchg( &(x), cmp, value )\n"
-"#define AtomXhg(x, value) atom_xchg ( &(x), value )\n"
-"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
-"#define mymake_float4 (float4)\n"
-"//#define make_float2 (float2)\n"
-"//#define make_uint4 (uint4)\n"
-"//#define make_int4 (int4)\n"
-"//#define make_uint2 (uint2)\n"
-"//#define make_int2 (int2)\n"
-"#define max2 max\n"
-"#define min2 min\n"
-"///////////////////////////////////////\n"
-"//	Vector\n"
-"///////////////////////////////////////\n"
-"__inline\n"
-"float4 fastNormalize4(float4 v)\n"
-"{\n"
-"	return fast_normalize(v);\n"
-"}\n"
-"__inline\n"
-"float4 cross3(float4 a, float4 b)\n"
-"{\n"
-"	return cross(a,b);\n"
-"}\n"
-"__inline\n"
-"float dot3F4(float4 a, float4 b)\n"
-"{\n"
-"	float4 a1 = mymake_float4(a.xyz,0.f);\n"
-"	float4 b1 = mymake_float4(b.xyz,0.f);\n"
-"	return dot(a1, b1);\n"
-"}\n"
-"__inline\n"
-"float4 normalize3(const float4 a)\n"
-"{\n"
-"	float4 n = mymake_float4(a.x, a.y, a.z, 0.f);\n"
-"	return fastNormalize4( n );\n"
-"//	float length = sqrtf(dot3F4(a, a));\n"
-"//	return 1.f/length * a;\n"
-"}\n"
-"///////////////////////////////////////\n"
-"//	Matrix3x3\n"
-"///////////////////////////////////////\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_row[3];\n"
-"}Matrix3x3;\n"
-"__inline\n"
-"float4 mtMul1(Matrix3x3 a, float4 b);\n"
-"__inline\n"
-"float4 mtMul3(float4 a, Matrix3x3 b);\n"
-"__inline\n"
-"float4 mtMul1(Matrix3x3 a, float4 b)\n"
-"{\n"
-"	float4 ans;\n"
-"	ans.x = dot3F4( a.m_row[0], b );\n"
-"	ans.y = dot3F4( a.m_row[1], b );\n"
-"	ans.z = dot3F4( a.m_row[2], b );\n"
-"	ans.w = 0.f;\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"float4 mtMul3(float4 a, Matrix3x3 b)\n"
-"{\n"
-"	float4 colx = mymake_float4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
-"	float4 coly = mymake_float4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
-"	float4 colz = mymake_float4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
-"	float4 ans;\n"
-"	ans.x = dot3F4( a, colx );\n"
-"	ans.y = dot3F4( a, coly );\n"
-"	ans.z = dot3F4( a, colz );\n"
-"	return ans;\n"
-"}\n"
-"///////////////////////////////////////\n"
-"//	Quaternion\n"
-"///////////////////////////////////////\n"
-"typedef float4 Quaternion;\n"
-"#define WG_SIZE 64\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_pos;\n"
-"	Quaternion m_quat;\n"
-"	float4 m_linVel;\n"
-"	float4 m_angVel;\n"
-"	u32 m_shapeIdx;\n"
-"	float m_invMass;\n"
-"	float m_restituitionCoeff;\n"
-"	float m_frictionCoeff;\n"
-"} Body;\n"
-"typedef struct\n"
-"{\n"
-"	Matrix3x3 m_invInertia;\n"
-"	Matrix3x3 m_initInvInertia;\n"
-"} Shape;\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_linear;\n"
-"	float4 m_worldPos[4];\n"
-"	float4 m_center;	\n"
-"	float m_jacCoeffInv[4];\n"
-"	float m_b[4];\n"
-"	float m_appliedRambdaDt[4];\n"
-"	float m_fJacCoeffInv[2];	\n"
-"	float m_fAppliedRambdaDt[2];	\n"
-"	u32 m_bodyA;\n"
-"	u32 m_bodyB;\n"
-"	int m_batchIdx;\n"
-"	u32 m_paddings[1];\n"
-"} Constraint4;\n"
-"typedef struct\n"
-"{\n"
-"	int m_nConstraints;\n"
-"	int m_start;\n"
-"	int m_batchIdx;\n"
-"	int m_nSplit;\n"
-"//	int m_paddings[1];\n"
-"} ConstBuffer;\n"
-"typedef struct\n"
-"{\n"
-"	int m_solveFriction;\n"
-"	int m_maxBatch;	//	long batch really kills the performance\n"
-"	int m_batchIdx;\n"
-"	int m_nSplit;\n"
-"//	int m_paddings[1];\n"
-"} ConstBufferBatchSolve;\n"
-"void setLinearAndAngular( float4 n, float4 r0, float4 r1, float4* linear, float4* angular0, float4* angular1);\n"
-"void setLinearAndAngular( float4 n, float4 r0, float4 r1, float4* linear, float4* angular0, float4* angular1)\n"
-"{\n"
-"	*linear = mymake_float4(-n.xyz,0.f);\n"
-"	*angular0 = -cross3(r0, n);\n"
-"	*angular1 = cross3(r1, n);\n"
-"}\n"
-"float calcRelVel( float4 l0, float4 l1, float4 a0, float4 a1, float4 linVel0, float4 angVel0, float4 linVel1, float4 angVel1 );\n"
-"float calcRelVel( float4 l0, float4 l1, float4 a0, float4 a1, float4 linVel0, float4 angVel0, float4 linVel1, float4 angVel1 )\n"
-"{\n"
-"	return dot3F4(l0, linVel0) + dot3F4(a0, angVel0) + dot3F4(l1, linVel1) + dot3F4(a1, angVel1);\n"
-"}\n"
-"float calcJacCoeff(const float4 linear0, const float4 linear1, const float4 angular0, const float4 angular1,\n"
-"				   float invMass0, const Matrix3x3* invInertia0, float invMass1, const Matrix3x3* invInertia1);\n"
-"float calcJacCoeff(const float4 linear0, const float4 linear1, const float4 angular0, const float4 angular1,\n"
-"					float invMass0, const Matrix3x3* invInertia0, float invMass1, const Matrix3x3* invInertia1)\n"
-"{\n"
-"	//	linear0,1 are normlized\n"
-"	float jmj0 = invMass0;//dot3F4(linear0, linear0)*invMass0;\n"
-"	float jmj1 = dot3F4(mtMul3(angular0,*invInertia0), angular0);\n"
-"	float jmj2 = invMass1;//dot3F4(linear1, linear1)*invMass1;\n"
-"	float jmj3 = dot3F4(mtMul3(angular1,*invInertia1), angular1);\n"
-"	return -1.f/(jmj0+jmj1+jmj2+jmj3);\n"
-"}\n"
-"void solveContact(__global Constraint4* cs,\n"
-"				  float4 posA, float4* linVelA, float4* angVelA, float invMassA, Matrix3x3 invInertiaA,\n"
-"				  float4 posB, float4* linVelB, float4* angVelB, float invMassB, Matrix3x3 invInertiaB);\n"
-"void solveContact(__global Constraint4* cs,\n"
-"			float4 posA, float4* linVelA, float4* angVelA, float invMassA, Matrix3x3 invInertiaA,\n"
-"			float4 posB, float4* linVelB, float4* angVelB, float invMassB, Matrix3x3 invInertiaB)\n"
-"{\n"
-"	float minRambdaDt = 0;\n"
-"	float maxRambdaDt = FLT_MAX;\n"
-"	for(int ic=0; ic<4; ic++)\n"
-"	{\n"
-"		if( cs->m_jacCoeffInv[ic] == 0.f ) continue;\n"
-"		float4 angular0, angular1, linear;\n"
-"		float4 r0 = cs->m_worldPos[ic] - posA;\n"
-"		float4 r1 = cs->m_worldPos[ic] - posB;\n"
-"		setLinearAndAngular( -cs->m_linear, r0, r1, &linear, &angular0, &angular1 );\n"
-"		float rambdaDt = calcRelVel( cs->m_linear, -cs->m_linear, angular0, angular1, \n"
-"			*linVelA, *angVelA, *linVelB, *angVelB ) + cs->m_b[ic];\n"
-"		rambdaDt *= cs->m_jacCoeffInv[ic];\n"
-"		{\n"
-"			float prevSum = cs->m_appliedRambdaDt[ic];\n"
-"			float updated = prevSum;\n"
-"			updated += rambdaDt;\n"
-"			updated = max2( updated, minRambdaDt );\n"
-"			updated = min2( updated, maxRambdaDt );\n"
-"			rambdaDt = updated - prevSum;\n"
-"			cs->m_appliedRambdaDt[ic] = updated;\n"
-"		}\n"
-"		float4 linImp0 = invMassA*linear*rambdaDt;\n"
-"		float4 linImp1 = invMassB*(-linear)*rambdaDt;\n"
-"		float4 angImp0 = mtMul1(invInertiaA, angular0)*rambdaDt;\n"
-"		float4 angImp1 = mtMul1(invInertiaB, angular1)*rambdaDt;\n"
-"		*linVelA += linImp0;\n"
-"		*angVelA += angImp0;\n"
-"		*linVelB += linImp1;\n"
-"		*angVelB += angImp1;\n"
-"	}\n"
-"}\n"
-"void btPlaneSpace1 (const float4* n, float4* p, float4* q);\n"
-" void btPlaneSpace1 (const float4* n, float4* p, float4* q)\n"
-"{\n"
-"  if (fabs(n[0].z) > 0.70710678f) {\n"
-"    // choose p in y-z plane\n"
-"    float a = n[0].y*n[0].y + n[0].z*n[0].z;\n"
-"    float k = 1.f/sqrt(a);\n"
-"    p[0].x = 0;\n"
-"	p[0].y = -n[0].z*k;\n"
-"	p[0].z = n[0].y*k;\n"
-"    // set q = n x p\n"
-"    q[0].x = a*k;\n"
-"	q[0].y = -n[0].x*p[0].z;\n"
-"	q[0].z = n[0].x*p[0].y;\n"
-"  }\n"
-"  else {\n"
-"    // choose p in x-y plane\n"
-"    float a = n[0].x*n[0].x + n[0].y*n[0].y;\n"
-"    float k = 1.f/sqrt(a);\n"
-"    p[0].x = -n[0].y*k;\n"
-"	p[0].y = n[0].x*k;\n"
-"	p[0].z = 0;\n"
-"    // set q = n x p\n"
-"    q[0].x = -n[0].z*p[0].y;\n"
-"	q[0].y = n[0].z*p[0].x;\n"
-"	q[0].z = a*k;\n"
-"  }\n"
-"}\n"
-"void solveContactConstraint(__global Body* gBodies, __global Shape* gShapes, __global Constraint4* ldsCs);\n"
-"void solveContactConstraint(__global Body* gBodies, __global Shape* gShapes, __global Constraint4* ldsCs)\n"
-"{\n"
-"	//float frictionCoeff = ldsCs[0].m_linear.w;\n"
-"	int aIdx = ldsCs[0].m_bodyA;\n"
-"	int bIdx = ldsCs[0].m_bodyB;\n"
-"	float4 posA = gBodies[aIdx].m_pos;\n"
-"	float4 linVelA = gBodies[aIdx].m_linVel;\n"
-"	float4 angVelA = gBodies[aIdx].m_angVel;\n"
-"	float invMassA = gBodies[aIdx].m_invMass;\n"
-"	Matrix3x3 invInertiaA = gShapes[aIdx].m_invInertia;\n"
-"	float4 posB = gBodies[bIdx].m_pos;\n"
-"	float4 linVelB = gBodies[bIdx].m_linVel;\n"
-"	float4 angVelB = gBodies[bIdx].m_angVel;\n"
-"	float invMassB = gBodies[bIdx].m_invMass;\n"
-"	Matrix3x3 invInertiaB = gShapes[bIdx].m_invInertia;\n"
-"	solveContact( ldsCs, posA, &linVelA, &angVelA, invMassA, invInertiaA,\n"
-"			posB, &linVelB, &angVelB, invMassB, invInertiaB );\n"
-"  if (gBodies[aIdx].m_invMass)\n"
-"  {\n"
-"		gBodies[aIdx].m_linVel = linVelA;\n"
-"		gBodies[aIdx].m_angVel = angVelA;\n"
-"	} else\n"
-"	{\n"
-"		gBodies[aIdx].m_linVel = mymake_float4(0,0,0,0);\n"
-"		gBodies[aIdx].m_angVel = mymake_float4(0,0,0,0);\n"
-"	\n"
-"	}\n"
-"	if (gBodies[bIdx].m_invMass)\n"
-"  {\n"
-"		gBodies[bIdx].m_linVel = linVelB;\n"
-"		gBodies[bIdx].m_angVel = angVelB;\n"
-"	} else\n"
-"	{\n"
-"		gBodies[bIdx].m_linVel = mymake_float4(0,0,0,0);\n"
-"		gBodies[bIdx].m_angVel = mymake_float4(0,0,0,0);\n"
-"	\n"
-"	}\n"
-"}\n"
-"typedef struct \n"
-"{\n"
-"	int m_valInt0;\n"
-"	int m_valInt1;\n"
-"	int m_valInt2;\n"
-"	int m_valInt3;\n"
-"	float m_val0;\n"
-"	float m_val1;\n"
-"	float m_val2;\n"
-"	float m_val3;\n"
-"} SolverDebugInfo;\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"void BatchSolveKernelContact(__global Body* gBodies,\n"
-"                      __global Shape* gShapes,\n"
-"                      __global Constraint4* gConstraints,\n"
-"                      __global int* gN,\n"
-"                      __global int* gOffsets,\n"
-"                      __global	int* batchSizes,\n"
-"                       int maxBatch1,\n"
-"                       int cellBatch,\n"
-"                       int4 nSplit\n"
-"                      )\n"
-"{\n"
-"	//__local int ldsBatchIdx[WG_SIZE+1];\n"
-"	__local int ldsCurBatch;\n"
-"	__local int ldsNextBatch;\n"
-"	__local int ldsStart;\n"
-"	int lIdx = GET_LOCAL_IDX;\n"
-"	int wgIdx = GET_GROUP_IDX;\n"
-"//	int gIdx = GET_GLOBAL_IDX;\n"
-"//	debugInfo[gIdx].m_valInt0 = gIdx;\n"
-"	//debugInfo[gIdx].m_valInt1 = GET_GROUP_SIZE;\n"
-"	\n"
-"	\n"
-"	int zIdx = (wgIdx/((nSplit.x*nSplit.y)/4))*2+((cellBatch&4)>>2);\n"
-"	int remain= (wgIdx%((nSplit.x*nSplit.y)/4));\n"
-"	int yIdx = (remain/(nSplit.x/2))*2 + ((cellBatch&2)>>1);\n"
-"	int xIdx = (remain%(nSplit.x/2))*2 + (cellBatch&1);\n"
-"	int cellIdx = xIdx+yIdx*nSplit.x+zIdx*(nSplit.x*nSplit.y);\n"
-"	//int xIdx = (wgIdx/(nSplit/2))*2 + (bIdx&1);\n"
-"	//int yIdx = (wgIdx%(nSplit/2))*2 + (bIdx>>1);\n"
-"	//int cellIdx = xIdx+yIdx*nSplit;\n"
-"	\n"
-"	if( gN[cellIdx] == 0 ) \n"
-"		return;\n"
-"	int maxBatch = batchSizes[cellIdx];\n"
-"	\n"
-"	\n"
-"	const int start = gOffsets[cellIdx];\n"
-"	const int end = start + gN[cellIdx];\n"
-"	\n"
-"	\n"
-"	\n"
-"	if( lIdx == 0 )\n"
-"	{\n"
-"		ldsCurBatch = 0;\n"
-"		ldsNextBatch = 0;\n"
-"		ldsStart = start;\n"
-"	}\n"
-"	GROUP_LDS_BARRIER;\n"
-"	int idx=ldsStart+lIdx;\n"
-"	while (ldsCurBatch < maxBatch)\n"
-"	{\n"
-"		for(; idx<end; )\n"
-"		{\n"
-"			if (gConstraints[idx].m_batchIdx == ldsCurBatch)\n"
-"			{\n"
-"					solveContactConstraint( gBodies, gShapes, &gConstraints[idx] );\n"
-"				 idx+=64;\n"
-"			} else\n"
-"			{\n"
-"				break;\n"
-"			}\n"
-"		}\n"
-"		GROUP_LDS_BARRIER;\n"
-"	\n"
-"		if( lIdx == 0 )\n"
-"		{\n"
-"			ldsCurBatch++;\n"
-"		}\n"
-"		GROUP_LDS_BARRIER;\n"
-"	}\n"
-"	\n"
-"    \n"
-"}\n"
-"__kernel void solveSingleContactKernel(__global Body* gBodies,\n"
-"                      __global Shape* gShapes,\n"
-"                      __global Constraint4* gConstraints,\n"
-"                       int cellIdx,\n"
-"                       int batchOffset,\n"
-"                       int numConstraintsInBatch\n"
-"                      )\n"
-"{\n"
-"	int index = get_global_id(0);\n"
-"	if (index < numConstraintsInBatch)\n"
-"	{\n"
-"		int idx=batchOffset+index;\n"
-"		solveContactConstraint( gBodies, gShapes, &gConstraints[idx] );\n"
-"	}    \n"
-"}\n"
-;
+static const char* solveContactCL =
+	"/*\n"
+	"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
+	"This software is provided 'as-is', without any express or implied warranty.\n"
+	"In no event will the authors be held liable for any damages arising from the use of this software.\n"
+	"Permission is granted to anyone to use this software for any purpose, \n"
+	"including commercial applications, and to alter it and redistribute it freely, \n"
+	"subject to the following restrictions:\n"
+	"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
+	"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
+	"3. This notice may not be removed or altered from any source distribution.\n"
+	"*/\n"
+	"//Originally written by Takahiro Harada\n"
+	"//#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_extended_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics : enable\n"
+	"#ifdef cl_ext_atomic_counters_32\n"
+	"#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
+	"#else\n"
+	"#define counter32_t volatile global int*\n"
+	"#endif\n"
+	"typedef unsigned int u32;\n"
+	"typedef unsigned short u16;\n"
+	"typedef unsigned char u8;\n"
+	"#define GET_GROUP_IDX get_group_id(0)\n"
+	"#define GET_LOCAL_IDX get_local_id(0)\n"
+	"#define GET_GLOBAL_IDX get_global_id(0)\n"
+	"#define GET_GROUP_SIZE get_local_size(0)\n"
+	"#define GET_NUM_GROUPS get_num_groups(0)\n"
+	"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
+	"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
+	"#define AtomInc(x) atom_inc(&(x))\n"
+	"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
+	"#define AppendInc(x, out) out = atomic_inc(x)\n"
+	"#define AtomAdd(x, value) atom_add(&(x), value)\n"
+	"#define AtomCmpxhg(x, cmp, value) atom_cmpxchg( &(x), cmp, value )\n"
+	"#define AtomXhg(x, value) atom_xchg ( &(x), value )\n"
+	"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
+	"#define mymake_float4 (float4)\n"
+	"//#define make_float2 (float2)\n"
+	"//#define make_uint4 (uint4)\n"
+	"//#define make_int4 (int4)\n"
+	"//#define make_uint2 (uint2)\n"
+	"//#define make_int2 (int2)\n"
+	"#define max2 max\n"
+	"#define min2 min\n"
+	"///////////////////////////////////////\n"
+	"//	Vector\n"
+	"///////////////////////////////////////\n"
+	"__inline\n"
+	"float4 fastNormalize4(float4 v)\n"
+	"{\n"
+	"	return fast_normalize(v);\n"
+	"}\n"
+	"__inline\n"
+	"float4 cross3(float4 a, float4 b)\n"
+	"{\n"
+	"	return cross(a,b);\n"
+	"}\n"
+	"__inline\n"
+	"float dot3F4(float4 a, float4 b)\n"
+	"{\n"
+	"	float4 a1 = mymake_float4(a.xyz,0.f);\n"
+	"	float4 b1 = mymake_float4(b.xyz,0.f);\n"
+	"	return dot(a1, b1);\n"
+	"}\n"
+	"__inline\n"
+	"float4 normalize3(const float4 a)\n"
+	"{\n"
+	"	float4 n = mymake_float4(a.x, a.y, a.z, 0.f);\n"
+	"	return fastNormalize4( n );\n"
+	"//	float length = sqrtf(dot3F4(a, a));\n"
+	"//	return 1.f/length * a;\n"
+	"}\n"
+	"///////////////////////////////////////\n"
+	"//	Matrix3x3\n"
+	"///////////////////////////////////////\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_row[3];\n"
+	"}Matrix3x3;\n"
+	"__inline\n"
+	"float4 mtMul1(Matrix3x3 a, float4 b);\n"
+	"__inline\n"
+	"float4 mtMul3(float4 a, Matrix3x3 b);\n"
+	"__inline\n"
+	"float4 mtMul1(Matrix3x3 a, float4 b)\n"
+	"{\n"
+	"	float4 ans;\n"
+	"	ans.x = dot3F4( a.m_row[0], b );\n"
+	"	ans.y = dot3F4( a.m_row[1], b );\n"
+	"	ans.z = dot3F4( a.m_row[2], b );\n"
+	"	ans.w = 0.f;\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"float4 mtMul3(float4 a, Matrix3x3 b)\n"
+	"{\n"
+	"	float4 colx = mymake_float4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
+	"	float4 coly = mymake_float4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
+	"	float4 colz = mymake_float4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
+	"	float4 ans;\n"
+	"	ans.x = dot3F4( a, colx );\n"
+	"	ans.y = dot3F4( a, coly );\n"
+	"	ans.z = dot3F4( a, colz );\n"
+	"	return ans;\n"
+	"}\n"
+	"///////////////////////////////////////\n"
+	"//	Quaternion\n"
+	"///////////////////////////////////////\n"
+	"typedef float4 Quaternion;\n"
+	"#define WG_SIZE 64\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_pos;\n"
+	"	Quaternion m_quat;\n"
+	"	float4 m_linVel;\n"
+	"	float4 m_angVel;\n"
+	"	u32 m_shapeIdx;\n"
+	"	float m_invMass;\n"
+	"	float m_restituitionCoeff;\n"
+	"	float m_frictionCoeff;\n"
+	"} Body;\n"
+	"typedef struct\n"
+	"{\n"
+	"	Matrix3x3 m_invInertia;\n"
+	"	Matrix3x3 m_initInvInertia;\n"
+	"} Shape;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_linear;\n"
+	"	float4 m_worldPos[4];\n"
+	"	float4 m_center;	\n"
+	"	float m_jacCoeffInv[4];\n"
+	"	float m_b[4];\n"
+	"	float m_appliedRambdaDt[4];\n"
+	"	float m_fJacCoeffInv[2];	\n"
+	"	float m_fAppliedRambdaDt[2];	\n"
+	"	u32 m_bodyA;\n"
+	"	u32 m_bodyB;\n"
+	"	int m_batchIdx;\n"
+	"	u32 m_paddings[1];\n"
+	"} Constraint4;\n"
+	"typedef struct\n"
+	"{\n"
+	"	int m_nConstraints;\n"
+	"	int m_start;\n"
+	"	int m_batchIdx;\n"
+	"	int m_nSplit;\n"
+	"//	int m_paddings[1];\n"
+	"} ConstBuffer;\n"
+	"typedef struct\n"
+	"{\n"
+	"	int m_solveFriction;\n"
+	"	int m_maxBatch;	//	long batch really kills the performance\n"
+	"	int m_batchIdx;\n"
+	"	int m_nSplit;\n"
+	"//	int m_paddings[1];\n"
+	"} ConstBufferBatchSolve;\n"
+	"void setLinearAndAngular( float4 n, float4 r0, float4 r1, float4* linear, float4* angular0, float4* angular1);\n"
+	"void setLinearAndAngular( float4 n, float4 r0, float4 r1, float4* linear, float4* angular0, float4* angular1)\n"
+	"{\n"
+	"	*linear = mymake_float4(-n.xyz,0.f);\n"
+	"	*angular0 = -cross3(r0, n);\n"
+	"	*angular1 = cross3(r1, n);\n"
+	"}\n"
+	"float calcRelVel( float4 l0, float4 l1, float4 a0, float4 a1, float4 linVel0, float4 angVel0, float4 linVel1, float4 angVel1 );\n"
+	"float calcRelVel( float4 l0, float4 l1, float4 a0, float4 a1, float4 linVel0, float4 angVel0, float4 linVel1, float4 angVel1 )\n"
+	"{\n"
+	"	return dot3F4(l0, linVel0) + dot3F4(a0, angVel0) + dot3F4(l1, linVel1) + dot3F4(a1, angVel1);\n"
+	"}\n"
+	"float calcJacCoeff(const float4 linear0, const float4 linear1, const float4 angular0, const float4 angular1,\n"
+	"				   float invMass0, const Matrix3x3* invInertia0, float invMass1, const Matrix3x3* invInertia1);\n"
+	"float calcJacCoeff(const float4 linear0, const float4 linear1, const float4 angular0, const float4 angular1,\n"
+	"					float invMass0, const Matrix3x3* invInertia0, float invMass1, const Matrix3x3* invInertia1)\n"
+	"{\n"
+	"	//	linear0,1 are normlized\n"
+	"	float jmj0 = invMass0;//dot3F4(linear0, linear0)*invMass0;\n"
+	"	float jmj1 = dot3F4(mtMul3(angular0,*invInertia0), angular0);\n"
+	"	float jmj2 = invMass1;//dot3F4(linear1, linear1)*invMass1;\n"
+	"	float jmj3 = dot3F4(mtMul3(angular1,*invInertia1), angular1);\n"
+	"	return -1.f/(jmj0+jmj1+jmj2+jmj3);\n"
+	"}\n"
+	"void solveContact(__global Constraint4* cs,\n"
+	"				  float4 posA, float4* linVelA, float4* angVelA, float invMassA, Matrix3x3 invInertiaA,\n"
+	"				  float4 posB, float4* linVelB, float4* angVelB, float invMassB, Matrix3x3 invInertiaB);\n"
+	"void solveContact(__global Constraint4* cs,\n"
+	"			float4 posA, float4* linVelA, float4* angVelA, float invMassA, Matrix3x3 invInertiaA,\n"
+	"			float4 posB, float4* linVelB, float4* angVelB, float invMassB, Matrix3x3 invInertiaB)\n"
+	"{\n"
+	"	float minRambdaDt = 0;\n"
+	"	float maxRambdaDt = FLT_MAX;\n"
+	"	for(int ic=0; ic<4; ic++)\n"
+	"	{\n"
+	"		if( cs->m_jacCoeffInv[ic] == 0.f ) continue;\n"
+	"		float4 angular0, angular1, linear;\n"
+	"		float4 r0 = cs->m_worldPos[ic] - posA;\n"
+	"		float4 r1 = cs->m_worldPos[ic] - posB;\n"
+	"		setLinearAndAngular( -cs->m_linear, r0, r1, &linear, &angular0, &angular1 );\n"
+	"		float rambdaDt = calcRelVel( cs->m_linear, -cs->m_linear, angular0, angular1, \n"
+	"			*linVelA, *angVelA, *linVelB, *angVelB ) + cs->m_b[ic];\n"
+	"		rambdaDt *= cs->m_jacCoeffInv[ic];\n"
+	"		{\n"
+	"			float prevSum = cs->m_appliedRambdaDt[ic];\n"
+	"			float updated = prevSum;\n"
+	"			updated += rambdaDt;\n"
+	"			updated = max2( updated, minRambdaDt );\n"
+	"			updated = min2( updated, maxRambdaDt );\n"
+	"			rambdaDt = updated - prevSum;\n"
+	"			cs->m_appliedRambdaDt[ic] = updated;\n"
+	"		}\n"
+	"		float4 linImp0 = invMassA*linear*rambdaDt;\n"
+	"		float4 linImp1 = invMassB*(-linear)*rambdaDt;\n"
+	"		float4 angImp0 = mtMul1(invInertiaA, angular0)*rambdaDt;\n"
+	"		float4 angImp1 = mtMul1(invInertiaB, angular1)*rambdaDt;\n"
+	"		*linVelA += linImp0;\n"
+	"		*angVelA += angImp0;\n"
+	"		*linVelB += linImp1;\n"
+	"		*angVelB += angImp1;\n"
+	"	}\n"
+	"}\n"
+	"void btPlaneSpace1 (const float4* n, float4* p, float4* q);\n"
+	" void btPlaneSpace1 (const float4* n, float4* p, float4* q)\n"
+	"{\n"
+	"  if (fabs(n[0].z) > 0.70710678f) {\n"
+	"    // choose p in y-z plane\n"
+	"    float a = n[0].y*n[0].y + n[0].z*n[0].z;\n"
+	"    float k = 1.f/sqrt(a);\n"
+	"    p[0].x = 0;\n"
+	"	p[0].y = -n[0].z*k;\n"
+	"	p[0].z = n[0].y*k;\n"
+	"    // set q = n x p\n"
+	"    q[0].x = a*k;\n"
+	"	q[0].y = -n[0].x*p[0].z;\n"
+	"	q[0].z = n[0].x*p[0].y;\n"
+	"  }\n"
+	"  else {\n"
+	"    // choose p in x-y plane\n"
+	"    float a = n[0].x*n[0].x + n[0].y*n[0].y;\n"
+	"    float k = 1.f/sqrt(a);\n"
+	"    p[0].x = -n[0].y*k;\n"
+	"	p[0].y = n[0].x*k;\n"
+	"	p[0].z = 0;\n"
+	"    // set q = n x p\n"
+	"    q[0].x = -n[0].z*p[0].y;\n"
+	"	q[0].y = n[0].z*p[0].x;\n"
+	"	q[0].z = a*k;\n"
+	"  }\n"
+	"}\n"
+	"void solveContactConstraint(__global Body* gBodies, __global Shape* gShapes, __global Constraint4* ldsCs);\n"
+	"void solveContactConstraint(__global Body* gBodies, __global Shape* gShapes, __global Constraint4* ldsCs)\n"
+	"{\n"
+	"	//float frictionCoeff = ldsCs[0].m_linear.w;\n"
+	"	int aIdx = ldsCs[0].m_bodyA;\n"
+	"	int bIdx = ldsCs[0].m_bodyB;\n"
+	"	float4 posA = gBodies[aIdx].m_pos;\n"
+	"	float4 linVelA = gBodies[aIdx].m_linVel;\n"
+	"	float4 angVelA = gBodies[aIdx].m_angVel;\n"
+	"	float invMassA = gBodies[aIdx].m_invMass;\n"
+	"	Matrix3x3 invInertiaA = gShapes[aIdx].m_invInertia;\n"
+	"	float4 posB = gBodies[bIdx].m_pos;\n"
+	"	float4 linVelB = gBodies[bIdx].m_linVel;\n"
+	"	float4 angVelB = gBodies[bIdx].m_angVel;\n"
+	"	float invMassB = gBodies[bIdx].m_invMass;\n"
+	"	Matrix3x3 invInertiaB = gShapes[bIdx].m_invInertia;\n"
+	"	solveContact( ldsCs, posA, &linVelA, &angVelA, invMassA, invInertiaA,\n"
+	"			posB, &linVelB, &angVelB, invMassB, invInertiaB );\n"
+	"  if (gBodies[aIdx].m_invMass)\n"
+	"  {\n"
+	"		gBodies[aIdx].m_linVel = linVelA;\n"
+	"		gBodies[aIdx].m_angVel = angVelA;\n"
+	"	} else\n"
+	"	{\n"
+	"		gBodies[aIdx].m_linVel = mymake_float4(0,0,0,0);\n"
+	"		gBodies[aIdx].m_angVel = mymake_float4(0,0,0,0);\n"
+	"	\n"
+	"	}\n"
+	"	if (gBodies[bIdx].m_invMass)\n"
+	"  {\n"
+	"		gBodies[bIdx].m_linVel = linVelB;\n"
+	"		gBodies[bIdx].m_angVel = angVelB;\n"
+	"	} else\n"
+	"	{\n"
+	"		gBodies[bIdx].m_linVel = mymake_float4(0,0,0,0);\n"
+	"		gBodies[bIdx].m_angVel = mymake_float4(0,0,0,0);\n"
+	"	\n"
+	"	}\n"
+	"}\n"
+	"typedef struct \n"
+	"{\n"
+	"	int m_valInt0;\n"
+	"	int m_valInt1;\n"
+	"	int m_valInt2;\n"
+	"	int m_valInt3;\n"
+	"	float m_val0;\n"
+	"	float m_val1;\n"
+	"	float m_val2;\n"
+	"	float m_val3;\n"
+	"} SolverDebugInfo;\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"void BatchSolveKernelContact(__global Body* gBodies,\n"
+	"                      __global Shape* gShapes,\n"
+	"                      __global Constraint4* gConstraints,\n"
+	"                      __global int* gN,\n"
+	"                      __global int* gOffsets,\n"
+	"                      __global	int* batchSizes,\n"
+	"                       int maxBatch1,\n"
+	"                       int cellBatch,\n"
+	"                       int4 nSplit\n"
+	"                      )\n"
+	"{\n"
+	"	//__local int ldsBatchIdx[WG_SIZE+1];\n"
+	"	__local int ldsCurBatch;\n"
+	"	__local int ldsNextBatch;\n"
+	"	__local int ldsStart;\n"
+	"	int lIdx = GET_LOCAL_IDX;\n"
+	"	int wgIdx = GET_GROUP_IDX;\n"
+	"//	int gIdx = GET_GLOBAL_IDX;\n"
+	"//	debugInfo[gIdx].m_valInt0 = gIdx;\n"
+	"	//debugInfo[gIdx].m_valInt1 = GET_GROUP_SIZE;\n"
+	"	\n"
+	"	\n"
+	"	int zIdx = (wgIdx/((nSplit.x*nSplit.y)/4))*2+((cellBatch&4)>>2);\n"
+	"	int remain= (wgIdx%((nSplit.x*nSplit.y)/4));\n"
+	"	int yIdx = (remain/(nSplit.x/2))*2 + ((cellBatch&2)>>1);\n"
+	"	int xIdx = (remain%(nSplit.x/2))*2 + (cellBatch&1);\n"
+	"	int cellIdx = xIdx+yIdx*nSplit.x+zIdx*(nSplit.x*nSplit.y);\n"
+	"	//int xIdx = (wgIdx/(nSplit/2))*2 + (bIdx&1);\n"
+	"	//int yIdx = (wgIdx%(nSplit/2))*2 + (bIdx>>1);\n"
+	"	//int cellIdx = xIdx+yIdx*nSplit;\n"
+	"	\n"
+	"	if( gN[cellIdx] == 0 ) \n"
+	"		return;\n"
+	"	int maxBatch = batchSizes[cellIdx];\n"
+	"	\n"
+	"	\n"
+	"	const int start = gOffsets[cellIdx];\n"
+	"	const int end = start + gN[cellIdx];\n"
+	"	\n"
+	"	\n"
+	"	\n"
+	"	if( lIdx == 0 )\n"
+	"	{\n"
+	"		ldsCurBatch = 0;\n"
+	"		ldsNextBatch = 0;\n"
+	"		ldsStart = start;\n"
+	"	}\n"
+	"	GROUP_LDS_BARRIER;\n"
+	"	int idx=ldsStart+lIdx;\n"
+	"	while (ldsCurBatch < maxBatch)\n"
+	"	{\n"
+	"		for(; idx<end; )\n"
+	"		{\n"
+	"			if (gConstraints[idx].m_batchIdx == ldsCurBatch)\n"
+	"			{\n"
+	"					solveContactConstraint( gBodies, gShapes, &gConstraints[idx] );\n"
+	"				 idx+=64;\n"
+	"			} else\n"
+	"			{\n"
+	"				break;\n"
+	"			}\n"
+	"		}\n"
+	"		GROUP_LDS_BARRIER;\n"
+	"	\n"
+	"		if( lIdx == 0 )\n"
+	"		{\n"
+	"			ldsCurBatch++;\n"
+	"		}\n"
+	"		GROUP_LDS_BARRIER;\n"
+	"	}\n"
+	"	\n"
+	"    \n"
+	"}\n"
+	"__kernel void solveSingleContactKernel(__global Body* gBodies,\n"
+	"                      __global Shape* gShapes,\n"
+	"                      __global Constraint4* gConstraints,\n"
+	"                       int cellIdx,\n"
+	"                       int batchOffset,\n"
+	"                       int numConstraintsInBatch\n"
+	"                      )\n"
+	"{\n"
+	"	int index = get_global_id(0);\n"
+	"	if (index < numConstraintsInBatch)\n"
+	"	{\n"
+	"		int idx=batchOffset+index;\n"
+	"		solveContactConstraint( gBodies, gShapes, &gConstraints[idx] );\n"
+	"	}    \n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/solveFriction.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/solveFriction.h
index eb58674f22f..9707cdb25dd 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/solveFriction.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/solveFriction.h
@@ -1,421 +1,420 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* solveFrictionCL= \
-"/*\n"
-"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
-"This software is provided 'as-is', without any express or implied warranty.\n"
-"In no event will the authors be held liable for any damages arising from the use of this software.\n"
-"Permission is granted to anyone to use this software for any purpose, \n"
-"including commercial applications, and to alter it and redistribute it freely, \n"
-"subject to the following restrictions:\n"
-"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
-"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
-"3. This notice may not be removed or altered from any source distribution.\n"
-"*/\n"
-"//Originally written by Takahiro Harada\n"
-"//#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_extended_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics : enable\n"
-"#ifdef cl_ext_atomic_counters_32\n"
-"#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
-"#else\n"
-"#define counter32_t volatile global int*\n"
-"#endif\n"
-"typedef unsigned int u32;\n"
-"typedef unsigned short u16;\n"
-"typedef unsigned char u8;\n"
-"#define GET_GROUP_IDX get_group_id(0)\n"
-"#define GET_LOCAL_IDX get_local_id(0)\n"
-"#define GET_GLOBAL_IDX get_global_id(0)\n"
-"#define GET_GROUP_SIZE get_local_size(0)\n"
-"#define GET_NUM_GROUPS get_num_groups(0)\n"
-"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
-"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
-"#define AtomInc(x) atom_inc(&(x))\n"
-"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
-"#define AppendInc(x, out) out = atomic_inc(x)\n"
-"#define AtomAdd(x, value) atom_add(&(x), value)\n"
-"#define AtomCmpxhg(x, cmp, value) atom_cmpxchg( &(x), cmp, value )\n"
-"#define AtomXhg(x, value) atom_xchg ( &(x), value )\n"
-"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
-"#define mymake_float4 (float4)\n"
-"//#define make_float2 (float2)\n"
-"//#define make_uint4 (uint4)\n"
-"//#define make_int4 (int4)\n"
-"//#define make_uint2 (uint2)\n"
-"//#define make_int2 (int2)\n"
-"#define max2 max\n"
-"#define min2 min\n"
-"///////////////////////////////////////\n"
-"//	Vector\n"
-"///////////////////////////////////////\n"
-"__inline\n"
-"float4 fastNormalize4(float4 v)\n"
-"{\n"
-"	return fast_normalize(v);\n"
-"}\n"
-"__inline\n"
-"float4 cross3(float4 a, float4 b)\n"
-"{\n"
-"	return cross(a,b);\n"
-"}\n"
-"__inline\n"
-"float dot3F4(float4 a, float4 b)\n"
-"{\n"
-"	float4 a1 = mymake_float4(a.xyz,0.f);\n"
-"	float4 b1 = mymake_float4(b.xyz,0.f);\n"
-"	return dot(a1, b1);\n"
-"}\n"
-"__inline\n"
-"float4 normalize3(const float4 a)\n"
-"{\n"
-"	float4 n = mymake_float4(a.x, a.y, a.z, 0.f);\n"
-"	return fastNormalize4( n );\n"
-"//	float length = sqrtf(dot3F4(a, a));\n"
-"//	return 1.f/length * a;\n"
-"}\n"
-"///////////////////////////////////////\n"
-"//	Matrix3x3\n"
-"///////////////////////////////////////\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_row[3];\n"
-"}Matrix3x3;\n"
-"__inline\n"
-"float4 mtMul1(Matrix3x3 a, float4 b);\n"
-"__inline\n"
-"float4 mtMul3(float4 a, Matrix3x3 b);\n"
-"__inline\n"
-"float4 mtMul1(Matrix3x3 a, float4 b)\n"
-"{\n"
-"	float4 ans;\n"
-"	ans.x = dot3F4( a.m_row[0], b );\n"
-"	ans.y = dot3F4( a.m_row[1], b );\n"
-"	ans.z = dot3F4( a.m_row[2], b );\n"
-"	ans.w = 0.f;\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"float4 mtMul3(float4 a, Matrix3x3 b)\n"
-"{\n"
-"	float4 colx = mymake_float4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
-"	float4 coly = mymake_float4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
-"	float4 colz = mymake_float4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
-"	float4 ans;\n"
-"	ans.x = dot3F4( a, colx );\n"
-"	ans.y = dot3F4( a, coly );\n"
-"	ans.z = dot3F4( a, colz );\n"
-"	return ans;\n"
-"}\n"
-"///////////////////////////////////////\n"
-"//	Quaternion\n"
-"///////////////////////////////////////\n"
-"typedef float4 Quaternion;\n"
-"#define WG_SIZE 64\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_pos;\n"
-"	Quaternion m_quat;\n"
-"	float4 m_linVel;\n"
-"	float4 m_angVel;\n"
-"	u32 m_shapeIdx;\n"
-"	float m_invMass;\n"
-"	float m_restituitionCoeff;\n"
-"	float m_frictionCoeff;\n"
-"} Body;\n"
-"typedef struct\n"
-"{\n"
-"	Matrix3x3 m_invInertia;\n"
-"	Matrix3x3 m_initInvInertia;\n"
-"} Shape;\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_linear;\n"
-"	float4 m_worldPos[4];\n"
-"	float4 m_center;	\n"
-"	float m_jacCoeffInv[4];\n"
-"	float m_b[4];\n"
-"	float m_appliedRambdaDt[4];\n"
-"	float m_fJacCoeffInv[2];	\n"
-"	float m_fAppliedRambdaDt[2];	\n"
-"	u32 m_bodyA;\n"
-"	u32 m_bodyB;\n"
-"	int m_batchIdx;\n"
-"	u32 m_paddings[1];\n"
-"} Constraint4;\n"
-"typedef struct\n"
-"{\n"
-"	int m_nConstraints;\n"
-"	int m_start;\n"
-"	int m_batchIdx;\n"
-"	int m_nSplit;\n"
-"//	int m_paddings[1];\n"
-"} ConstBuffer;\n"
-"typedef struct\n"
-"{\n"
-"	int m_solveFriction;\n"
-"	int m_maxBatch;	//	long batch really kills the performance\n"
-"	int m_batchIdx;\n"
-"	int m_nSplit;\n"
-"//	int m_paddings[1];\n"
-"} ConstBufferBatchSolve;\n"
-"void setLinearAndAngular( float4 n, float4 r0, float4 r1, float4* linear, float4* angular0, float4* angular1);\n"
-"void setLinearAndAngular( float4 n, float4 r0, float4 r1, float4* linear, float4* angular0, float4* angular1)\n"
-"{\n"
-"	*linear = mymake_float4(-n.xyz,0.f);\n"
-"	*angular0 = -cross3(r0, n);\n"
-"	*angular1 = cross3(r1, n);\n"
-"}\n"
-"float calcRelVel( float4 l0, float4 l1, float4 a0, float4 a1, float4 linVel0, float4 angVel0, float4 linVel1, float4 angVel1 );\n"
-"float calcRelVel( float4 l0, float4 l1, float4 a0, float4 a1, float4 linVel0, float4 angVel0, float4 linVel1, float4 angVel1 )\n"
-"{\n"
-"	return dot3F4(l0, linVel0) + dot3F4(a0, angVel0) + dot3F4(l1, linVel1) + dot3F4(a1, angVel1);\n"
-"}\n"
-"float calcJacCoeff(const float4 linear0, const float4 linear1, const float4 angular0, const float4 angular1,\n"
-"				   float invMass0, const Matrix3x3* invInertia0, float invMass1, const Matrix3x3* invInertia1);\n"
-"float calcJacCoeff(const float4 linear0, const float4 linear1, const float4 angular0, const float4 angular1,\n"
-"					float invMass0, const Matrix3x3* invInertia0, float invMass1, const Matrix3x3* invInertia1)\n"
-"{\n"
-"	//	linear0,1 are normlized\n"
-"	float jmj0 = invMass0;//dot3F4(linear0, linear0)*invMass0;\n"
-"	float jmj1 = dot3F4(mtMul3(angular0,*invInertia0), angular0);\n"
-"	float jmj2 = invMass1;//dot3F4(linear1, linear1)*invMass1;\n"
-"	float jmj3 = dot3F4(mtMul3(angular1,*invInertia1), angular1);\n"
-"	return -1.f/(jmj0+jmj1+jmj2+jmj3);\n"
-"}\n"
-"void btPlaneSpace1 (const float4* n, float4* p, float4* q);\n"
-" void btPlaneSpace1 (const float4* n, float4* p, float4* q)\n"
-"{\n"
-"  if (fabs(n[0].z) > 0.70710678f) {\n"
-"    // choose p in y-z plane\n"
-"    float a = n[0].y*n[0].y + n[0].z*n[0].z;\n"
-"    float k = 1.f/sqrt(a);\n"
-"    p[0].x = 0;\n"
-"	p[0].y = -n[0].z*k;\n"
-"	p[0].z = n[0].y*k;\n"
-"    // set q = n x p\n"
-"    q[0].x = a*k;\n"
-"	q[0].y = -n[0].x*p[0].z;\n"
-"	q[0].z = n[0].x*p[0].y;\n"
-"  }\n"
-"  else {\n"
-"    // choose p in x-y plane\n"
-"    float a = n[0].x*n[0].x + n[0].y*n[0].y;\n"
-"    float k = 1.f/sqrt(a);\n"
-"    p[0].x = -n[0].y*k;\n"
-"	p[0].y = n[0].x*k;\n"
-"	p[0].z = 0;\n"
-"    // set q = n x p\n"
-"    q[0].x = -n[0].z*p[0].y;\n"
-"	q[0].y = n[0].z*p[0].x;\n"
-"	q[0].z = a*k;\n"
-"  }\n"
-"}\n"
-"void solveFrictionConstraint(__global Body* gBodies, __global Shape* gShapes, __global Constraint4* ldsCs);\n"
-"void solveFrictionConstraint(__global Body* gBodies, __global Shape* gShapes, __global Constraint4* ldsCs)\n"
-"{\n"
-"	float frictionCoeff = ldsCs[0].m_linear.w;\n"
-"	int aIdx = ldsCs[0].m_bodyA;\n"
-"	int bIdx = ldsCs[0].m_bodyB;\n"
-"	float4 posA = gBodies[aIdx].m_pos;\n"
-"	float4 linVelA = gBodies[aIdx].m_linVel;\n"
-"	float4 angVelA = gBodies[aIdx].m_angVel;\n"
-"	float invMassA = gBodies[aIdx].m_invMass;\n"
-"	Matrix3x3 invInertiaA = gShapes[aIdx].m_invInertia;\n"
-"	float4 posB = gBodies[bIdx].m_pos;\n"
-"	float4 linVelB = gBodies[bIdx].m_linVel;\n"
-"	float4 angVelB = gBodies[bIdx].m_angVel;\n"
-"	float invMassB = gBodies[bIdx].m_invMass;\n"
-"	Matrix3x3 invInertiaB = gShapes[bIdx].m_invInertia;\n"
-"	\n"
-"	{\n"
-"		float maxRambdaDt[4] = {FLT_MAX,FLT_MAX,FLT_MAX,FLT_MAX};\n"
-"		float minRambdaDt[4] = {0.f,0.f,0.f,0.f};\n"
-"		float sum = 0;\n"
-"		for(int j=0; j<4; j++)\n"
-"		{\n"
-"			sum +=ldsCs[0].m_appliedRambdaDt[j];\n"
-"		}\n"
-"		frictionCoeff = 0.7f;\n"
-"		for(int j=0; j<4; j++)\n"
-"		{\n"
-"			maxRambdaDt[j] = frictionCoeff*sum;\n"
-"			minRambdaDt[j] = -maxRambdaDt[j];\n"
-"		}\n"
-"		\n"
-"//		solveFriction( ldsCs, posA, &linVelA, &angVelA, invMassA, invInertiaA,\n"
-"//			posB, &linVelB, &angVelB, invMassB, invInertiaB, maxRambdaDt, minRambdaDt );\n"
-"		\n"
-"		\n"
-"		{\n"
-"			\n"
-"			__global Constraint4* cs = ldsCs;\n"
-"			\n"
-"			if( cs->m_fJacCoeffInv[0] == 0 && cs->m_fJacCoeffInv[0] == 0 ) return;\n"
-"			const float4 center = cs->m_center;\n"
-"			\n"
-"			float4 n = -cs->m_linear;\n"
-"			\n"
-"			float4 tangent[2];\n"
-"			btPlaneSpace1(&n,&tangent[0],&tangent[1]);\n"
-"			float4 angular0, angular1, linear;\n"
-"			float4 r0 = center - posA;\n"
-"			float4 r1 = center - posB;\n"
-"			for(int i=0; i<2; i++)\n"
-"			{\n"
-"				setLinearAndAngular( tangent[i], r0, r1, &linear, &angular0, &angular1 );\n"
-"				float rambdaDt = calcRelVel(linear, -linear, angular0, angular1,\n"
-"											linVelA, angVelA, linVelB, angVelB );\n"
-"				rambdaDt *= cs->m_fJacCoeffInv[i];\n"
-"				\n"
-"				{\n"
-"					float prevSum = cs->m_fAppliedRambdaDt[i];\n"
-"					float updated = prevSum;\n"
-"					updated += rambdaDt;\n"
-"					updated = max2( updated, minRambdaDt[i] );\n"
-"					updated = min2( updated, maxRambdaDt[i] );\n"
-"					rambdaDt = updated - prevSum;\n"
-"					cs->m_fAppliedRambdaDt[i] = updated;\n"
-"				}\n"
-"				\n"
-"				float4 linImp0 = invMassA*linear*rambdaDt;\n"
-"				float4 linImp1 = invMassB*(-linear)*rambdaDt;\n"
-"				float4 angImp0 = mtMul1(invInertiaA, angular0)*rambdaDt;\n"
-"				float4 angImp1 = mtMul1(invInertiaB, angular1)*rambdaDt;\n"
-"				\n"
-"				linVelA += linImp0;\n"
-"				angVelA += angImp0;\n"
-"				linVelB += linImp1;\n"
-"				angVelB += angImp1;\n"
-"			}\n"
-"			{	//	angular damping for point constraint\n"
-"				float4 ab = normalize3( posB - posA );\n"
-"				float4 ac = normalize3( center - posA );\n"
-"				if( dot3F4( ab, ac ) > 0.95f  || (invMassA == 0.f || invMassB == 0.f))\n"
-"				{\n"
-"					float angNA = dot3F4( n, angVelA );\n"
-"					float angNB = dot3F4( n, angVelB );\n"
-"					\n"
-"					angVelA -= (angNA*0.1f)*n;\n"
-"					angVelB -= (angNB*0.1f)*n;\n"
-"				}\n"
-"			}\n"
-"		}\n"
-"		\n"
-"		\n"
-"	}\n"
-"	if (gBodies[aIdx].m_invMass)\n"
-"	{\n"
-"		gBodies[aIdx].m_linVel = linVelA;\n"
-"		gBodies[aIdx].m_angVel = angVelA;\n"
-"	} else\n"
-"	{\n"
-"		gBodies[aIdx].m_linVel = mymake_float4(0,0,0,0);\n"
-"		gBodies[aIdx].m_angVel = mymake_float4(0,0,0,0);\n"
-"	}\n"
-"	if (gBodies[bIdx].m_invMass)\n"
-"	{\n"
-"		gBodies[bIdx].m_linVel = linVelB;\n"
-"		gBodies[bIdx].m_angVel = angVelB;\n"
-"	} else\n"
-"	{\n"
-"		gBodies[bIdx].m_linVel = mymake_float4(0,0,0,0);\n"
-"		gBodies[bIdx].m_angVel = mymake_float4(0,0,0,0);\n"
-"	}\n"
-" \n"
-"}\n"
-"typedef struct \n"
-"{\n"
-"	int m_valInt0;\n"
-"	int m_valInt1;\n"
-"	int m_valInt2;\n"
-"	int m_valInt3;\n"
-"	float m_val0;\n"
-"	float m_val1;\n"
-"	float m_val2;\n"
-"	float m_val3;\n"
-"} SolverDebugInfo;\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"void BatchSolveKernelFriction(__global Body* gBodies,\n"
-"                      __global Shape* gShapes,\n"
-"                      __global Constraint4* gConstraints,\n"
-"                      __global int* gN,\n"
-"                      __global int* gOffsets,\n"
-"                      __global int* batchSizes,\n"
-"                       int maxBatch1,\n"
-"                       int cellBatch,\n"
-"                       int4 nSplit\n"
-"                      )\n"
-"{\n"
-"	//__local int ldsBatchIdx[WG_SIZE+1];\n"
-"	__local int ldsCurBatch;\n"
-"	__local int ldsNextBatch;\n"
-"	__local int ldsStart;\n"
-"	int lIdx = GET_LOCAL_IDX;\n"
-"	int wgIdx = GET_GROUP_IDX;\n"
-"//	int gIdx = GET_GLOBAL_IDX;\n"
-"//	debugInfo[gIdx].m_valInt0 = gIdx;\n"
-"	//debugInfo[gIdx].m_valInt1 = GET_GROUP_SIZE;\n"
-"	int zIdx = (wgIdx/((nSplit.x*nSplit.y)/4))*2+((cellBatch&4)>>2);\n"
-"	int remain= (wgIdx%((nSplit.x*nSplit.y)/4));\n"
-"	int yIdx = (remain/(nSplit.x/2))*2 + ((cellBatch&2)>>1);\n"
-"	int xIdx = (remain%(nSplit.x/2))*2 + (cellBatch&1);\n"
-"	int cellIdx = xIdx+yIdx*nSplit.x+zIdx*(nSplit.x*nSplit.y);\n"
-"	\n"
-"	if( gN[cellIdx] == 0 ) \n"
-"		return;\n"
-"	int maxBatch = batchSizes[cellIdx];\n"
-"	const int start = gOffsets[cellIdx];\n"
-"	const int end = start + gN[cellIdx];\n"
-"	\n"
-"	if( lIdx == 0 )\n"
-"	{\n"
-"		ldsCurBatch = 0;\n"
-"		ldsNextBatch = 0;\n"
-"		ldsStart = start;\n"
-"	}\n"
-"	GROUP_LDS_BARRIER;\n"
-"	int idx=ldsStart+lIdx;\n"
-"	while (ldsCurBatch < maxBatch)\n"
-"	{\n"
-"		for(; idx<end; )\n"
-"		{\n"
-"			if (gConstraints[idx].m_batchIdx == ldsCurBatch)\n"
-"			{\n"
-"					solveFrictionConstraint( gBodies, gShapes, &gConstraints[idx] );\n"
-"				 idx+=64;\n"
-"			} else\n"
-"			{\n"
-"				break;\n"
-"			}\n"
-"		}\n"
-"		GROUP_LDS_BARRIER;\n"
-"		if( lIdx == 0 )\n"
-"		{\n"
-"			ldsCurBatch++;\n"
-"		}\n"
-"		GROUP_LDS_BARRIER;\n"
-"	}\n"
-"	\n"
-"    \n"
-"}\n"
-"__kernel void solveSingleFrictionKernel(__global Body* gBodies,\n"
-"                      __global Shape* gShapes,\n"
-"                      __global Constraint4* gConstraints,\n"
-"                       int cellIdx,\n"
-"                       int batchOffset,\n"
-"                       int numConstraintsInBatch\n"
-"                      )\n"
-"{\n"
-"	int index = get_global_id(0);\n"
-"	if (index < numConstraintsInBatch)\n"
-"	{\n"
-"		\n"
-"		int idx=batchOffset+index;\n"
-"	\n"
-"		solveFrictionConstraint( gBodies, gShapes, &gConstraints[idx] );\n"
-"	}    \n"
-"}\n"
-;
+static const char* solveFrictionCL =
+	"/*\n"
+	"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
+	"This software is provided 'as-is', without any express or implied warranty.\n"
+	"In no event will the authors be held liable for any damages arising from the use of this software.\n"
+	"Permission is granted to anyone to use this software for any purpose, \n"
+	"including commercial applications, and to alter it and redistribute it freely, \n"
+	"subject to the following restrictions:\n"
+	"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
+	"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
+	"3. This notice may not be removed or altered from any source distribution.\n"
+	"*/\n"
+	"//Originally written by Takahiro Harada\n"
+	"//#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_extended_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics : enable\n"
+	"#ifdef cl_ext_atomic_counters_32\n"
+	"#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
+	"#else\n"
+	"#define counter32_t volatile global int*\n"
+	"#endif\n"
+	"typedef unsigned int u32;\n"
+	"typedef unsigned short u16;\n"
+	"typedef unsigned char u8;\n"
+	"#define GET_GROUP_IDX get_group_id(0)\n"
+	"#define GET_LOCAL_IDX get_local_id(0)\n"
+	"#define GET_GLOBAL_IDX get_global_id(0)\n"
+	"#define GET_GROUP_SIZE get_local_size(0)\n"
+	"#define GET_NUM_GROUPS get_num_groups(0)\n"
+	"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
+	"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
+	"#define AtomInc(x) atom_inc(&(x))\n"
+	"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
+	"#define AppendInc(x, out) out = atomic_inc(x)\n"
+	"#define AtomAdd(x, value) atom_add(&(x), value)\n"
+	"#define AtomCmpxhg(x, cmp, value) atom_cmpxchg( &(x), cmp, value )\n"
+	"#define AtomXhg(x, value) atom_xchg ( &(x), value )\n"
+	"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
+	"#define mymake_float4 (float4)\n"
+	"//#define make_float2 (float2)\n"
+	"//#define make_uint4 (uint4)\n"
+	"//#define make_int4 (int4)\n"
+	"//#define make_uint2 (uint2)\n"
+	"//#define make_int2 (int2)\n"
+	"#define max2 max\n"
+	"#define min2 min\n"
+	"///////////////////////////////////////\n"
+	"//	Vector\n"
+	"///////////////////////////////////////\n"
+	"__inline\n"
+	"float4 fastNormalize4(float4 v)\n"
+	"{\n"
+	"	return fast_normalize(v);\n"
+	"}\n"
+	"__inline\n"
+	"float4 cross3(float4 a, float4 b)\n"
+	"{\n"
+	"	return cross(a,b);\n"
+	"}\n"
+	"__inline\n"
+	"float dot3F4(float4 a, float4 b)\n"
+	"{\n"
+	"	float4 a1 = mymake_float4(a.xyz,0.f);\n"
+	"	float4 b1 = mymake_float4(b.xyz,0.f);\n"
+	"	return dot(a1, b1);\n"
+	"}\n"
+	"__inline\n"
+	"float4 normalize3(const float4 a)\n"
+	"{\n"
+	"	float4 n = mymake_float4(a.x, a.y, a.z, 0.f);\n"
+	"	return fastNormalize4( n );\n"
+	"//	float length = sqrtf(dot3F4(a, a));\n"
+	"//	return 1.f/length * a;\n"
+	"}\n"
+	"///////////////////////////////////////\n"
+	"//	Matrix3x3\n"
+	"///////////////////////////////////////\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_row[3];\n"
+	"}Matrix3x3;\n"
+	"__inline\n"
+	"float4 mtMul1(Matrix3x3 a, float4 b);\n"
+	"__inline\n"
+	"float4 mtMul3(float4 a, Matrix3x3 b);\n"
+	"__inline\n"
+	"float4 mtMul1(Matrix3x3 a, float4 b)\n"
+	"{\n"
+	"	float4 ans;\n"
+	"	ans.x = dot3F4( a.m_row[0], b );\n"
+	"	ans.y = dot3F4( a.m_row[1], b );\n"
+	"	ans.z = dot3F4( a.m_row[2], b );\n"
+	"	ans.w = 0.f;\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"float4 mtMul3(float4 a, Matrix3x3 b)\n"
+	"{\n"
+	"	float4 colx = mymake_float4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
+	"	float4 coly = mymake_float4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
+	"	float4 colz = mymake_float4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
+	"	float4 ans;\n"
+	"	ans.x = dot3F4( a, colx );\n"
+	"	ans.y = dot3F4( a, coly );\n"
+	"	ans.z = dot3F4( a, colz );\n"
+	"	return ans;\n"
+	"}\n"
+	"///////////////////////////////////////\n"
+	"//	Quaternion\n"
+	"///////////////////////////////////////\n"
+	"typedef float4 Quaternion;\n"
+	"#define WG_SIZE 64\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_pos;\n"
+	"	Quaternion m_quat;\n"
+	"	float4 m_linVel;\n"
+	"	float4 m_angVel;\n"
+	"	u32 m_shapeIdx;\n"
+	"	float m_invMass;\n"
+	"	float m_restituitionCoeff;\n"
+	"	float m_frictionCoeff;\n"
+	"} Body;\n"
+	"typedef struct\n"
+	"{\n"
+	"	Matrix3x3 m_invInertia;\n"
+	"	Matrix3x3 m_initInvInertia;\n"
+	"} Shape;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_linear;\n"
+	"	float4 m_worldPos[4];\n"
+	"	float4 m_center;	\n"
+	"	float m_jacCoeffInv[4];\n"
+	"	float m_b[4];\n"
+	"	float m_appliedRambdaDt[4];\n"
+	"	float m_fJacCoeffInv[2];	\n"
+	"	float m_fAppliedRambdaDt[2];	\n"
+	"	u32 m_bodyA;\n"
+	"	u32 m_bodyB;\n"
+	"	int m_batchIdx;\n"
+	"	u32 m_paddings[1];\n"
+	"} Constraint4;\n"
+	"typedef struct\n"
+	"{\n"
+	"	int m_nConstraints;\n"
+	"	int m_start;\n"
+	"	int m_batchIdx;\n"
+	"	int m_nSplit;\n"
+	"//	int m_paddings[1];\n"
+	"} ConstBuffer;\n"
+	"typedef struct\n"
+	"{\n"
+	"	int m_solveFriction;\n"
+	"	int m_maxBatch;	//	long batch really kills the performance\n"
+	"	int m_batchIdx;\n"
+	"	int m_nSplit;\n"
+	"//	int m_paddings[1];\n"
+	"} ConstBufferBatchSolve;\n"
+	"void setLinearAndAngular( float4 n, float4 r0, float4 r1, float4* linear, float4* angular0, float4* angular1);\n"
+	"void setLinearAndAngular( float4 n, float4 r0, float4 r1, float4* linear, float4* angular0, float4* angular1)\n"
+	"{\n"
+	"	*linear = mymake_float4(-n.xyz,0.f);\n"
+	"	*angular0 = -cross3(r0, n);\n"
+	"	*angular1 = cross3(r1, n);\n"
+	"}\n"
+	"float calcRelVel( float4 l0, float4 l1, float4 a0, float4 a1, float4 linVel0, float4 angVel0, float4 linVel1, float4 angVel1 );\n"
+	"float calcRelVel( float4 l0, float4 l1, float4 a0, float4 a1, float4 linVel0, float4 angVel0, float4 linVel1, float4 angVel1 )\n"
+	"{\n"
+	"	return dot3F4(l0, linVel0) + dot3F4(a0, angVel0) + dot3F4(l1, linVel1) + dot3F4(a1, angVel1);\n"
+	"}\n"
+	"float calcJacCoeff(const float4 linear0, const float4 linear1, const float4 angular0, const float4 angular1,\n"
+	"				   float invMass0, const Matrix3x3* invInertia0, float invMass1, const Matrix3x3* invInertia1);\n"
+	"float calcJacCoeff(const float4 linear0, const float4 linear1, const float4 angular0, const float4 angular1,\n"
+	"					float invMass0, const Matrix3x3* invInertia0, float invMass1, const Matrix3x3* invInertia1)\n"
+	"{\n"
+	"	//	linear0,1 are normlized\n"
+	"	float jmj0 = invMass0;//dot3F4(linear0, linear0)*invMass0;\n"
+	"	float jmj1 = dot3F4(mtMul3(angular0,*invInertia0), angular0);\n"
+	"	float jmj2 = invMass1;//dot3F4(linear1, linear1)*invMass1;\n"
+	"	float jmj3 = dot3F4(mtMul3(angular1,*invInertia1), angular1);\n"
+	"	return -1.f/(jmj0+jmj1+jmj2+jmj3);\n"
+	"}\n"
+	"void btPlaneSpace1 (const float4* n, float4* p, float4* q);\n"
+	" void btPlaneSpace1 (const float4* n, float4* p, float4* q)\n"
+	"{\n"
+	"  if (fabs(n[0].z) > 0.70710678f) {\n"
+	"    // choose p in y-z plane\n"
+	"    float a = n[0].y*n[0].y + n[0].z*n[0].z;\n"
+	"    float k = 1.f/sqrt(a);\n"
+	"    p[0].x = 0;\n"
+	"	p[0].y = -n[0].z*k;\n"
+	"	p[0].z = n[0].y*k;\n"
+	"    // set q = n x p\n"
+	"    q[0].x = a*k;\n"
+	"	q[0].y = -n[0].x*p[0].z;\n"
+	"	q[0].z = n[0].x*p[0].y;\n"
+	"  }\n"
+	"  else {\n"
+	"    // choose p in x-y plane\n"
+	"    float a = n[0].x*n[0].x + n[0].y*n[0].y;\n"
+	"    float k = 1.f/sqrt(a);\n"
+	"    p[0].x = -n[0].y*k;\n"
+	"	p[0].y = n[0].x*k;\n"
+	"	p[0].z = 0;\n"
+	"    // set q = n x p\n"
+	"    q[0].x = -n[0].z*p[0].y;\n"
+	"	q[0].y = n[0].z*p[0].x;\n"
+	"	q[0].z = a*k;\n"
+	"  }\n"
+	"}\n"
+	"void solveFrictionConstraint(__global Body* gBodies, __global Shape* gShapes, __global Constraint4* ldsCs);\n"
+	"void solveFrictionConstraint(__global Body* gBodies, __global Shape* gShapes, __global Constraint4* ldsCs)\n"
+	"{\n"
+	"	float frictionCoeff = ldsCs[0].m_linear.w;\n"
+	"	int aIdx = ldsCs[0].m_bodyA;\n"
+	"	int bIdx = ldsCs[0].m_bodyB;\n"
+	"	float4 posA = gBodies[aIdx].m_pos;\n"
+	"	float4 linVelA = gBodies[aIdx].m_linVel;\n"
+	"	float4 angVelA = gBodies[aIdx].m_angVel;\n"
+	"	float invMassA = gBodies[aIdx].m_invMass;\n"
+	"	Matrix3x3 invInertiaA = gShapes[aIdx].m_invInertia;\n"
+	"	float4 posB = gBodies[bIdx].m_pos;\n"
+	"	float4 linVelB = gBodies[bIdx].m_linVel;\n"
+	"	float4 angVelB = gBodies[bIdx].m_angVel;\n"
+	"	float invMassB = gBodies[bIdx].m_invMass;\n"
+	"	Matrix3x3 invInertiaB = gShapes[bIdx].m_invInertia;\n"
+	"	\n"
+	"	{\n"
+	"		float maxRambdaDt[4] = {FLT_MAX,FLT_MAX,FLT_MAX,FLT_MAX};\n"
+	"		float minRambdaDt[4] = {0.f,0.f,0.f,0.f};\n"
+	"		float sum = 0;\n"
+	"		for(int j=0; j<4; j++)\n"
+	"		{\n"
+	"			sum +=ldsCs[0].m_appliedRambdaDt[j];\n"
+	"		}\n"
+	"		frictionCoeff = 0.7f;\n"
+	"		for(int j=0; j<4; j++)\n"
+	"		{\n"
+	"			maxRambdaDt[j] = frictionCoeff*sum;\n"
+	"			minRambdaDt[j] = -maxRambdaDt[j];\n"
+	"		}\n"
+	"		\n"
+	"//		solveFriction( ldsCs, posA, &linVelA, &angVelA, invMassA, invInertiaA,\n"
+	"//			posB, &linVelB, &angVelB, invMassB, invInertiaB, maxRambdaDt, minRambdaDt );\n"
+	"		\n"
+	"		\n"
+	"		{\n"
+	"			\n"
+	"			__global Constraint4* cs = ldsCs;\n"
+	"			\n"
+	"			if( cs->m_fJacCoeffInv[0] == 0 && cs->m_fJacCoeffInv[0] == 0 ) return;\n"
+	"			const float4 center = cs->m_center;\n"
+	"			\n"
+	"			float4 n = -cs->m_linear;\n"
+	"			\n"
+	"			float4 tangent[2];\n"
+	"			btPlaneSpace1(&n,&tangent[0],&tangent[1]);\n"
+	"			float4 angular0, angular1, linear;\n"
+	"			float4 r0 = center - posA;\n"
+	"			float4 r1 = center - posB;\n"
+	"			for(int i=0; i<2; i++)\n"
+	"			{\n"
+	"				setLinearAndAngular( tangent[i], r0, r1, &linear, &angular0, &angular1 );\n"
+	"				float rambdaDt = calcRelVel(linear, -linear, angular0, angular1,\n"
+	"											linVelA, angVelA, linVelB, angVelB );\n"
+	"				rambdaDt *= cs->m_fJacCoeffInv[i];\n"
+	"				\n"
+	"				{\n"
+	"					float prevSum = cs->m_fAppliedRambdaDt[i];\n"
+	"					float updated = prevSum;\n"
+	"					updated += rambdaDt;\n"
+	"					updated = max2( updated, minRambdaDt[i] );\n"
+	"					updated = min2( updated, maxRambdaDt[i] );\n"
+	"					rambdaDt = updated - prevSum;\n"
+	"					cs->m_fAppliedRambdaDt[i] = updated;\n"
+	"				}\n"
+	"				\n"
+	"				float4 linImp0 = invMassA*linear*rambdaDt;\n"
+	"				float4 linImp1 = invMassB*(-linear)*rambdaDt;\n"
+	"				float4 angImp0 = mtMul1(invInertiaA, angular0)*rambdaDt;\n"
+	"				float4 angImp1 = mtMul1(invInertiaB, angular1)*rambdaDt;\n"
+	"				\n"
+	"				linVelA += linImp0;\n"
+	"				angVelA += angImp0;\n"
+	"				linVelB += linImp1;\n"
+	"				angVelB += angImp1;\n"
+	"			}\n"
+	"			{	//	angular damping for point constraint\n"
+	"				float4 ab = normalize3( posB - posA );\n"
+	"				float4 ac = normalize3( center - posA );\n"
+	"				if( dot3F4( ab, ac ) > 0.95f  || (invMassA == 0.f || invMassB == 0.f))\n"
+	"				{\n"
+	"					float angNA = dot3F4( n, angVelA );\n"
+	"					float angNB = dot3F4( n, angVelB );\n"
+	"					\n"
+	"					angVelA -= (angNA*0.1f)*n;\n"
+	"					angVelB -= (angNB*0.1f)*n;\n"
+	"				}\n"
+	"			}\n"
+	"		}\n"
+	"		\n"
+	"		\n"
+	"	}\n"
+	"	if (gBodies[aIdx].m_invMass)\n"
+	"	{\n"
+	"		gBodies[aIdx].m_linVel = linVelA;\n"
+	"		gBodies[aIdx].m_angVel = angVelA;\n"
+	"	} else\n"
+	"	{\n"
+	"		gBodies[aIdx].m_linVel = mymake_float4(0,0,0,0);\n"
+	"		gBodies[aIdx].m_angVel = mymake_float4(0,0,0,0);\n"
+	"	}\n"
+	"	if (gBodies[bIdx].m_invMass)\n"
+	"	{\n"
+	"		gBodies[bIdx].m_linVel = linVelB;\n"
+	"		gBodies[bIdx].m_angVel = angVelB;\n"
+	"	} else\n"
+	"	{\n"
+	"		gBodies[bIdx].m_linVel = mymake_float4(0,0,0,0);\n"
+	"		gBodies[bIdx].m_angVel = mymake_float4(0,0,0,0);\n"
+	"	}\n"
+	" \n"
+	"}\n"
+	"typedef struct \n"
+	"{\n"
+	"	int m_valInt0;\n"
+	"	int m_valInt1;\n"
+	"	int m_valInt2;\n"
+	"	int m_valInt3;\n"
+	"	float m_val0;\n"
+	"	float m_val1;\n"
+	"	float m_val2;\n"
+	"	float m_val3;\n"
+	"} SolverDebugInfo;\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"void BatchSolveKernelFriction(__global Body* gBodies,\n"
+	"                      __global Shape* gShapes,\n"
+	"                      __global Constraint4* gConstraints,\n"
+	"                      __global int* gN,\n"
+	"                      __global int* gOffsets,\n"
+	"                      __global int* batchSizes,\n"
+	"                       int maxBatch1,\n"
+	"                       int cellBatch,\n"
+	"                       int4 nSplit\n"
+	"                      )\n"
+	"{\n"
+	"	//__local int ldsBatchIdx[WG_SIZE+1];\n"
+	"	__local int ldsCurBatch;\n"
+	"	__local int ldsNextBatch;\n"
+	"	__local int ldsStart;\n"
+	"	int lIdx = GET_LOCAL_IDX;\n"
+	"	int wgIdx = GET_GROUP_IDX;\n"
+	"//	int gIdx = GET_GLOBAL_IDX;\n"
+	"//	debugInfo[gIdx].m_valInt0 = gIdx;\n"
+	"	//debugInfo[gIdx].m_valInt1 = GET_GROUP_SIZE;\n"
+	"	int zIdx = (wgIdx/((nSplit.x*nSplit.y)/4))*2+((cellBatch&4)>>2);\n"
+	"	int remain= (wgIdx%((nSplit.x*nSplit.y)/4));\n"
+	"	int yIdx = (remain/(nSplit.x/2))*2 + ((cellBatch&2)>>1);\n"
+	"	int xIdx = (remain%(nSplit.x/2))*2 + (cellBatch&1);\n"
+	"	int cellIdx = xIdx+yIdx*nSplit.x+zIdx*(nSplit.x*nSplit.y);\n"
+	"	\n"
+	"	if( gN[cellIdx] == 0 ) \n"
+	"		return;\n"
+	"	int maxBatch = batchSizes[cellIdx];\n"
+	"	const int start = gOffsets[cellIdx];\n"
+	"	const int end = start + gN[cellIdx];\n"
+	"	\n"
+	"	if( lIdx == 0 )\n"
+	"	{\n"
+	"		ldsCurBatch = 0;\n"
+	"		ldsNextBatch = 0;\n"
+	"		ldsStart = start;\n"
+	"	}\n"
+	"	GROUP_LDS_BARRIER;\n"
+	"	int idx=ldsStart+lIdx;\n"
+	"	while (ldsCurBatch < maxBatch)\n"
+	"	{\n"
+	"		for(; idx<end; )\n"
+	"		{\n"
+	"			if (gConstraints[idx].m_batchIdx == ldsCurBatch)\n"
+	"			{\n"
+	"					solveFrictionConstraint( gBodies, gShapes, &gConstraints[idx] );\n"
+	"				 idx+=64;\n"
+	"			} else\n"
+	"			{\n"
+	"				break;\n"
+	"			}\n"
+	"		}\n"
+	"		GROUP_LDS_BARRIER;\n"
+	"		if( lIdx == 0 )\n"
+	"		{\n"
+	"			ldsCurBatch++;\n"
+	"		}\n"
+	"		GROUP_LDS_BARRIER;\n"
+	"	}\n"
+	"	\n"
+	"    \n"
+	"}\n"
+	"__kernel void solveSingleFrictionKernel(__global Body* gBodies,\n"
+	"                      __global Shape* gShapes,\n"
+	"                      __global Constraint4* gConstraints,\n"
+	"                       int cellIdx,\n"
+	"                       int batchOffset,\n"
+	"                       int numConstraintsInBatch\n"
+	"                      )\n"
+	"{\n"
+	"	int index = get_global_id(0);\n"
+	"	if (index < numConstraintsInBatch)\n"
+	"	{\n"
+	"		\n"
+	"		int idx=batchOffset+index;\n"
+	"	\n"
+	"		solveFrictionConstraint( gBodies, gShapes, &gConstraints[idx] );\n"
+	"	}    \n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/solverSetup.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/solverSetup.h
index eb1834ee00e..d53db031817 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/solverSetup.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/solverSetup.h
@@ -1,703 +1,702 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* solverSetupCL= \
-"/*\n"
-"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
-"This software is provided 'as-is', without any express or implied warranty.\n"
-"In no event will the authors be held liable for any damages arising from the use of this software.\n"
-"Permission is granted to anyone to use this software for any purpose, \n"
-"including commercial applications, and to alter it and redistribute it freely, \n"
-"subject to the following restrictions:\n"
-"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
-"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
-"3. This notice may not be removed or altered from any source distribution.\n"
-"*/\n"
-"//Originally written by Takahiro Harada\n"
-"#ifndef B3_CONTACT4DATA_H\n"
-"#define B3_CONTACT4DATA_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#define B3_FLOAT4_H\n"
-"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
-"#define B3_PLATFORM_DEFINITIONS_H\n"
-"struct MyTest\n"
-"{\n"
-"	int bla;\n"
-"};\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
-"#define B3_LARGE_FLOAT 1e18f\n"
-"#define B3_INFINITY 1e18f\n"
-"#define b3Assert(a)\n"
-"#define b3ConstArray(a) __global const a*\n"
-"#define b3AtomicInc atomic_inc\n"
-"#define b3AtomicAdd atomic_add\n"
-"#define b3Fabs fabs\n"
-"#define b3Sqrt native_sqrt\n"
-"#define b3Sin native_sin\n"
-"#define b3Cos native_cos\n"
-"#define B3_STATIC\n"
-"#endif\n"
-"#endif\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"	typedef float4	b3Float4;\n"
-"	#define b3Float4ConstArg const b3Float4\n"
-"	#define b3MakeFloat4 (float4)\n"
-"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return dot(a1, b1);\n"
-"	}\n"
-"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return cross(a1, b1);\n"
-"	}\n"
-"	#define b3MinFloat4 min\n"
-"	#define b3MaxFloat4 max\n"
-"	#define b3Normalized(a) normalize(a)\n"
-"#endif \n"
-"		\n"
-"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
-"{\n"
-"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
-"		return false;\n"
-"	return true;\n"
-"}\n"
-"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
-"{\n"
-"    float maxDot = -B3_INFINITY;\n"
-"    int i = 0;\n"
-"    int ptIndex = -1;\n"
-"    for( i = 0; i < vecLen; i++ )\n"
-"    {\n"
-"        float dot = b3Dot3F4(vecArray[i],vec);\n"
-"            \n"
-"        if( dot > maxDot )\n"
-"        {\n"
-"            maxDot = dot;\n"
-"            ptIndex = i;\n"
-"        }\n"
-"    }\n"
-"	b3Assert(ptIndex>=0);\n"
-"    if (ptIndex<0)\n"
-"	{\n"
-"		ptIndex = 0;\n"
-"	}\n"
-"    *dotOut = maxDot;\n"
-"    return ptIndex;\n"
-"}\n"
-"#endif //B3_FLOAT4_H\n"
-"typedef  struct b3Contact4Data b3Contact4Data_t;\n"
-"struct b3Contact4Data\n"
-"{\n"
-"	b3Float4	m_worldPosB[4];\n"
-"//	b3Float4	m_localPosA[4];\n"
-"//	b3Float4	m_localPosB[4];\n"
-"	b3Float4	m_worldNormalOnB;	//	w: m_nPoints\n"
-"	unsigned short  m_restituitionCoeffCmp;\n"
-"	unsigned short  m_frictionCoeffCmp;\n"
-"	int m_batchIdx;\n"
-"	int m_bodyAPtrAndSignBit;//x:m_bodyAPtr, y:m_bodyBPtr\n"
-"	int m_bodyBPtrAndSignBit;\n"
-"	int	m_childIndexA;\n"
-"	int	m_childIndexB;\n"
-"	int m_unused1;\n"
-"	int m_unused2;\n"
-"};\n"
-"inline int b3Contact4Data_getNumPoints(const struct b3Contact4Data* contact)\n"
-"{\n"
-"	return (int)contact->m_worldNormalOnB.w;\n"
-"};\n"
-"inline void b3Contact4Data_setNumPoints(struct b3Contact4Data* contact, int numPoints)\n"
-"{\n"
-"	contact->m_worldNormalOnB.w = (float)numPoints;\n"
-"};\n"
-"#endif //B3_CONTACT4DATA_H\n"
-"#ifndef B3_CONTACT_CONSTRAINT5_H\n"
-"#define B3_CONTACT_CONSTRAINT5_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_FLOAT4_H\n"
-"typedef struct b3ContactConstraint4 b3ContactConstraint4_t;\n"
-"struct b3ContactConstraint4\n"
-"{\n"
-"	b3Float4 m_linear;//normal?\n"
-"	b3Float4 m_worldPos[4];\n"
-"	b3Float4 m_center;	//	friction\n"
-"	float m_jacCoeffInv[4];\n"
-"	float m_b[4];\n"
-"	float m_appliedRambdaDt[4];\n"
-"	float m_fJacCoeffInv[2];	//	friction\n"
-"	float m_fAppliedRambdaDt[2];	//	friction\n"
-"	unsigned int m_bodyA;\n"
-"	unsigned int m_bodyB;\n"
-"	int			m_batchIdx;\n"
-"	unsigned int m_paddings;\n"
-"};\n"
-"//inline	void setFrictionCoeff(float value) { m_linear[3] = value; }\n"
-"inline	float b3GetFrictionCoeff(b3ContactConstraint4_t* constraint) \n"
-"{\n"
-"	return constraint->m_linear.w; \n"
-"}\n"
-"#endif //B3_CONTACT_CONSTRAINT5_H\n"
-"#ifndef B3_RIGIDBODY_DATA_H\n"
-"#define B3_RIGIDBODY_DATA_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_FLOAT4_H\n"
-"#ifndef B3_QUAT_H\n"
-"#define B3_QUAT_H\n"
-"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif\n"
-"#endif\n"
-"#ifndef B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"	typedef float4	b3Quat;\n"
-"	#define b3QuatConstArg const b3Quat\n"
-"	\n"
-"	\n"
-"inline float4 b3FastNormalize4(float4 v)\n"
-"{\n"
-"	v = (float4)(v.xyz,0.f);\n"
-"	return fast_normalize(v);\n"
-"}\n"
-"	\n"
-"inline b3Quat b3QuatMul(b3Quat a, b3Quat b);\n"
-"inline b3Quat b3QuatNormalized(b3QuatConstArg in);\n"
-"inline b3Quat b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec);\n"
-"inline b3Quat b3QuatInvert(b3QuatConstArg q);\n"
-"inline b3Quat b3QuatInverse(b3QuatConstArg q);\n"
-"inline b3Quat b3QuatMul(b3QuatConstArg a, b3QuatConstArg b)\n"
-"{\n"
-"	b3Quat ans;\n"
-"	ans = b3Cross3( a, b );\n"
-"	ans += a.w*b+b.w*a;\n"
-"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
-"	ans.w = a.w*b.w - b3Dot3F4(a, b);\n"
-"	return ans;\n"
-"}\n"
-"inline b3Quat b3QuatNormalized(b3QuatConstArg in)\n"
-"{\n"
-"	b3Quat q;\n"
-"	q=in;\n"
-"	//return b3FastNormalize4(in);\n"
-"	float len = native_sqrt(dot(q, q));\n"
-"	if(len > 0.f)\n"
-"	{\n"
-"		q *= 1.f / len;\n"
-"	}\n"
-"	else\n"
-"	{\n"
-"		q.x = q.y = q.z = 0.f;\n"
-"		q.w = 1.f;\n"
-"	}\n"
-"	return q;\n"
-"}\n"
-"inline float4 b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
-"{\n"
-"	b3Quat qInv = b3QuatInvert( q );\n"
-"	float4 vcpy = vec;\n"
-"	vcpy.w = 0.f;\n"
-"	float4 out = b3QuatMul(b3QuatMul(q,vcpy),qInv);\n"
-"	return out;\n"
-"}\n"
-"inline b3Quat b3QuatInverse(b3QuatConstArg q)\n"
-"{\n"
-"	return (b3Quat)(-q.xyz, q.w);\n"
-"}\n"
-"inline b3Quat b3QuatInvert(b3QuatConstArg q)\n"
-"{\n"
-"	return (b3Quat)(-q.xyz, q.w);\n"
-"}\n"
-"inline float4 b3QuatInvRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
-"{\n"
-"	return b3QuatRotate( b3QuatInvert( q ), vec );\n"
-"}\n"
-"inline b3Float4 b3TransformPoint(b3Float4ConstArg point, b3Float4ConstArg translation, b3QuatConstArg  orientation)\n"
-"{\n"
-"	return b3QuatRotate( orientation, point ) + (translation);\n"
-"}\n"
-"	\n"
-"#endif \n"
-"#endif //B3_QUAT_H\n"
-"#ifndef B3_MAT3x3_H\n"
-"#define B3_MAT3x3_H\n"
-"#ifndef B3_QUAT_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_QUAT_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"typedef struct\n"
-"{\n"
-"	b3Float4 m_row[3];\n"
-"}b3Mat3x3;\n"
-"#define b3Mat3x3ConstArg const b3Mat3x3\n"
-"#define b3GetRow(m,row) (m.m_row[row])\n"
-"inline b3Mat3x3 b3QuatGetRotationMatrix(b3Quat quat)\n"
-"{\n"
-"	b3Float4 quat2 = (b3Float4)(quat.x*quat.x, quat.y*quat.y, quat.z*quat.z, 0.f);\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0].x=1-2*quat2.y-2*quat2.z;\n"
-"	out.m_row[0].y=2*quat.x*quat.y-2*quat.w*quat.z;\n"
-"	out.m_row[0].z=2*quat.x*quat.z+2*quat.w*quat.y;\n"
-"	out.m_row[0].w = 0.f;\n"
-"	out.m_row[1].x=2*quat.x*quat.y+2*quat.w*quat.z;\n"
-"	out.m_row[1].y=1-2*quat2.x-2*quat2.z;\n"
-"	out.m_row[1].z=2*quat.y*quat.z-2*quat.w*quat.x;\n"
-"	out.m_row[1].w = 0.f;\n"
-"	out.m_row[2].x=2*quat.x*quat.z-2*quat.w*quat.y;\n"
-"	out.m_row[2].y=2*quat.y*quat.z+2*quat.w*quat.x;\n"
-"	out.m_row[2].z=1-2*quat2.x-2*quat2.y;\n"
-"	out.m_row[2].w = 0.f;\n"
-"	return out;\n"
-"}\n"
-"inline b3Mat3x3 b3AbsoluteMat3x3(b3Mat3x3ConstArg matIn)\n"
-"{\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0] = fabs(matIn.m_row[0]);\n"
-"	out.m_row[1] = fabs(matIn.m_row[1]);\n"
-"	out.m_row[2] = fabs(matIn.m_row[2]);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtZero();\n"
-"__inline\n"
-"b3Mat3x3 mtIdentity();\n"
-"__inline\n"
-"b3Mat3x3 mtTranspose(b3Mat3x3 m);\n"
-"__inline\n"
-"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b);\n"
-"__inline\n"
-"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b);\n"
-"__inline\n"
-"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b);\n"
-"__inline\n"
-"b3Mat3x3 mtZero()\n"
-"{\n"
-"	b3Mat3x3 m;\n"
-"	m.m_row[0] = (b3Float4)(0.f);\n"
-"	m.m_row[1] = (b3Float4)(0.f);\n"
-"	m.m_row[2] = (b3Float4)(0.f);\n"
-"	return m;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtIdentity()\n"
-"{\n"
-"	b3Mat3x3 m;\n"
-"	m.m_row[0] = (b3Float4)(1,0,0,0);\n"
-"	m.m_row[1] = (b3Float4)(0,1,0,0);\n"
-"	m.m_row[2] = (b3Float4)(0,0,1,0);\n"
-"	return m;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtTranspose(b3Mat3x3 m)\n"
-"{\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0] = (b3Float4)(m.m_row[0].x, m.m_row[1].x, m.m_row[2].x, 0.f);\n"
-"	out.m_row[1] = (b3Float4)(m.m_row[0].y, m.m_row[1].y, m.m_row[2].y, 0.f);\n"
-"	out.m_row[2] = (b3Float4)(m.m_row[0].z, m.m_row[1].z, m.m_row[2].z, 0.f);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b)\n"
-"{\n"
-"	b3Mat3x3 transB;\n"
-"	transB = mtTranspose( b );\n"
-"	b3Mat3x3 ans;\n"
-"	//	why this doesn't run when 0ing in the for{}\n"
-"	a.m_row[0].w = 0.f;\n"
-"	a.m_row[1].w = 0.f;\n"
-"	a.m_row[2].w = 0.f;\n"
-"	for(int i=0; i<3; i++)\n"
-"	{\n"
-"//	a.m_row[i].w = 0.f;\n"
-"		ans.m_row[i].x = b3Dot3F4(a.m_row[i],transB.m_row[0]);\n"
-"		ans.m_row[i].y = b3Dot3F4(a.m_row[i],transB.m_row[1]);\n"
-"		ans.m_row[i].z = b3Dot3F4(a.m_row[i],transB.m_row[2]);\n"
-"		ans.m_row[i].w = 0.f;\n"
-"	}\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b)\n"
-"{\n"
-"	b3Float4 ans;\n"
-"	ans.x = b3Dot3F4( a.m_row[0], b );\n"
-"	ans.y = b3Dot3F4( a.m_row[1], b );\n"
-"	ans.z = b3Dot3F4( a.m_row[2], b );\n"
-"	ans.w = 0.f;\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b)\n"
-"{\n"
-"	b3Float4 colx = b3MakeFloat4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
-"	b3Float4 coly = b3MakeFloat4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
-"	b3Float4 colz = b3MakeFloat4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
-"	b3Float4 ans;\n"
-"	ans.x = b3Dot3F4( a, colx );\n"
-"	ans.y = b3Dot3F4( a, coly );\n"
-"	ans.z = b3Dot3F4( a, colz );\n"
-"	return ans;\n"
-"}\n"
-"#endif\n"
-"#endif //B3_MAT3x3_H\n"
-"typedef struct b3RigidBodyData b3RigidBodyData_t;\n"
-"struct b3RigidBodyData\n"
-"{\n"
-"	b3Float4				m_pos;\n"
-"	b3Quat					m_quat;\n"
-"	b3Float4				m_linVel;\n"
-"	b3Float4				m_angVel;\n"
-"	int 					m_collidableIdx;\n"
-"	float 				m_invMass;\n"
-"	float 				m_restituitionCoeff;\n"
-"	float 				m_frictionCoeff;\n"
-"};\n"
-"typedef struct b3InertiaData b3InertiaData_t;\n"
-"struct b3InertiaData\n"
-"{\n"
-"	b3Mat3x3 m_invInertiaWorld;\n"
-"	b3Mat3x3 m_initInvInertia;\n"
-"};\n"
-"#endif //B3_RIGIDBODY_DATA_H\n"
-"	\n"
-"void b3PlaneSpace1 (b3Float4ConstArg n, b3Float4* p, b3Float4* q);\n"
-" void b3PlaneSpace1 (b3Float4ConstArg n, b3Float4* p, b3Float4* q)\n"
-"{\n"
-"  if (b3Fabs(n.z) > 0.70710678f) {\n"
-"    // choose p in y-z plane\n"
-"    float a = n.y*n.y + n.z*n.z;\n"
-"    float k = 1.f/sqrt(a);\n"
-"    p[0].x = 0;\n"
-"	p[0].y = -n.z*k;\n"
-"	p[0].z = n.y*k;\n"
-"    // set q = n x p\n"
-"    q[0].x = a*k;\n"
-"	q[0].y = -n.x*p[0].z;\n"
-"	q[0].z = n.x*p[0].y;\n"
-"  }\n"
-"  else {\n"
-"    // choose p in x-y plane\n"
-"    float a = n.x*n.x + n.y*n.y;\n"
-"    float k = 1.f/sqrt(a);\n"
-"    p[0].x = -n.y*k;\n"
-"	p[0].y = n.x*k;\n"
-"	p[0].z = 0;\n"
-"    // set q = n x p\n"
-"    q[0].x = -n.z*p[0].y;\n"
-"	q[0].y = n.z*p[0].x;\n"
-"	q[0].z = a*k;\n"
-"  }\n"
-"}\n"
-" \n"
-"void setLinearAndAngular( b3Float4ConstArg n, b3Float4ConstArg r0, b3Float4ConstArg r1, b3Float4* linear, b3Float4* angular0, b3Float4* angular1)\n"
-"{\n"
-"	*linear = b3MakeFloat4(n.x,n.y,n.z,0.f);\n"
-"	*angular0 = b3Cross3(r0, n);\n"
-"	*angular1 = -b3Cross3(r1, n);\n"
-"}\n"
-"float calcRelVel( b3Float4ConstArg l0, b3Float4ConstArg l1, b3Float4ConstArg a0, b3Float4ConstArg a1, b3Float4ConstArg linVel0,\n"
-"	b3Float4ConstArg angVel0, b3Float4ConstArg linVel1, b3Float4ConstArg angVel1 )\n"
-"{\n"
-"	return b3Dot3F4(l0, linVel0) + b3Dot3F4(a0, angVel0) + b3Dot3F4(l1, linVel1) + b3Dot3F4(a1, angVel1);\n"
-"}\n"
-"float calcJacCoeff(b3Float4ConstArg linear0, b3Float4ConstArg linear1, b3Float4ConstArg angular0, b3Float4ConstArg angular1,\n"
-"					float invMass0, const b3Mat3x3* invInertia0, float invMass1, const b3Mat3x3* invInertia1)\n"
-"{\n"
-"	//	linear0,1 are normlized\n"
-"	float jmj0 = invMass0;//b3Dot3F4(linear0, linear0)*invMass0;\n"
-"	float jmj1 = b3Dot3F4(mtMul3(angular0,*invInertia0), angular0);\n"
-"	float jmj2 = invMass1;//b3Dot3F4(linear1, linear1)*invMass1;\n"
-"	float jmj3 = b3Dot3F4(mtMul3(angular1,*invInertia1), angular1);\n"
-"	return -1.f/(jmj0+jmj1+jmj2+jmj3);\n"
-"}\n"
-"void setConstraint4( b3Float4ConstArg posA, b3Float4ConstArg linVelA, b3Float4ConstArg angVelA, float invMassA, b3Mat3x3ConstArg invInertiaA,\n"
-"	b3Float4ConstArg posB, b3Float4ConstArg linVelB, b3Float4ConstArg angVelB, float invMassB, b3Mat3x3ConstArg invInertiaB, \n"
-"	__global struct b3Contact4Data* src, float dt, float positionDrift, float positionConstraintCoeff,\n"
-"	b3ContactConstraint4_t* dstC )\n"
-"{\n"
-"	dstC->m_bodyA = abs(src->m_bodyAPtrAndSignBit);\n"
-"	dstC->m_bodyB = abs(src->m_bodyBPtrAndSignBit);\n"
-"	float dtInv = 1.f/dt;\n"
-"	for(int ic=0; ic<4; ic++)\n"
-"	{\n"
-"		dstC->m_appliedRambdaDt[ic] = 0.f;\n"
-"	}\n"
-"	dstC->m_fJacCoeffInv[0] = dstC->m_fJacCoeffInv[1] = 0.f;\n"
-"	dstC->m_linear = src->m_worldNormalOnB;\n"
-"	dstC->m_linear.w = 0.7f ;//src->getFrictionCoeff() );\n"
-"	for(int ic=0; ic<4; ic++)\n"
-"	{\n"
-"		b3Float4 r0 = src->m_worldPosB[ic] - posA;\n"
-"		b3Float4 r1 = src->m_worldPosB[ic] - posB;\n"
-"		if( ic >= src->m_worldNormalOnB.w )//npoints\n"
-"		{\n"
-"			dstC->m_jacCoeffInv[ic] = 0.f;\n"
-"			continue;\n"
-"		}\n"
-"		float relVelN;\n"
-"		{\n"
-"			b3Float4 linear, angular0, angular1;\n"
-"			setLinearAndAngular(src->m_worldNormalOnB, r0, r1, &linear, &angular0, &angular1);\n"
-"			dstC->m_jacCoeffInv[ic] = calcJacCoeff(linear, -linear, angular0, angular1,\n"
-"				invMassA, &invInertiaA, invMassB, &invInertiaB );\n"
-"			relVelN = calcRelVel(linear, -linear, angular0, angular1,\n"
-"				linVelA, angVelA, linVelB, angVelB);\n"
-"			float e = 0.f;//src->getRestituitionCoeff();\n"
-"			if( relVelN*relVelN < 0.004f ) e = 0.f;\n"
-"			dstC->m_b[ic] = e*relVelN;\n"
-"			//float penetration = src->m_worldPosB[ic].w;\n"
-"			dstC->m_b[ic] += (src->m_worldPosB[ic].w + positionDrift)*positionConstraintCoeff*dtInv;\n"
-"			dstC->m_appliedRambdaDt[ic] = 0.f;\n"
-"		}\n"
-"	}\n"
-"	if( src->m_worldNormalOnB.w > 0 )//npoints\n"
-"	{	//	prepare friction\n"
-"		b3Float4 center = b3MakeFloat4(0.f,0.f,0.f,0.f);\n"
-"		for(int i=0; i<src->m_worldNormalOnB.w; i++) \n"
-"			center += src->m_worldPosB[i];\n"
-"		center /= (float)src->m_worldNormalOnB.w;\n"
-"		b3Float4 tangent[2];\n"
-"		b3PlaneSpace1(src->m_worldNormalOnB,&tangent[0],&tangent[1]);\n"
-"		\n"
-"		b3Float4 r[2];\n"
-"		r[0] = center - posA;\n"
-"		r[1] = center - posB;\n"
-"		for(int i=0; i<2; i++)\n"
-"		{\n"
-"			b3Float4 linear, angular0, angular1;\n"
-"			setLinearAndAngular(tangent[i], r[0], r[1], &linear, &angular0, &angular1);\n"
-"			dstC->m_fJacCoeffInv[i] = calcJacCoeff(linear, -linear, angular0, angular1,\n"
-"				invMassA, &invInertiaA, invMassB, &invInertiaB );\n"
-"			dstC->m_fAppliedRambdaDt[i] = 0.f;\n"
-"		}\n"
-"		dstC->m_center = center;\n"
-"	}\n"
-"	for(int i=0; i<4; i++)\n"
-"	{\n"
-"		if( i<src->m_worldNormalOnB.w )\n"
-"		{\n"
-"			dstC->m_worldPos[i] = src->m_worldPosB[i];\n"
-"		}\n"
-"		else\n"
-"		{\n"
-"			dstC->m_worldPos[i] = b3MakeFloat4(0.f,0.f,0.f,0.f);\n"
-"		}\n"
-"	}\n"
-"}\n"
-"#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_extended_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics : enable\n"
-"#ifdef cl_ext_atomic_counters_32\n"
-"#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
-"#else\n"
-"#define counter32_t volatile global int*\n"
-"#endif\n"
-"typedef unsigned int u32;\n"
-"typedef unsigned short u16;\n"
-"typedef unsigned char u8;\n"
-"#define GET_GROUP_IDX get_group_id(0)\n"
-"#define GET_LOCAL_IDX get_local_id(0)\n"
-"#define GET_GLOBAL_IDX get_global_id(0)\n"
-"#define GET_GROUP_SIZE get_local_size(0)\n"
-"#define GET_NUM_GROUPS get_num_groups(0)\n"
-"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
-"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
-"#define AtomInc(x) atom_inc(&(x))\n"
-"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
-"#define AppendInc(x, out) out = atomic_inc(x)\n"
-"#define AtomAdd(x, value) atom_add(&(x), value)\n"
-"#define AtomCmpxhg(x, cmp, value) atom_cmpxchg( &(x), cmp, value )\n"
-"#define AtomXhg(x, value) atom_xchg ( &(x), value )\n"
-"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
-"#define make_float4 (float4)\n"
-"#define make_float2 (float2)\n"
-"#define make_uint4 (uint4)\n"
-"#define make_int4 (int4)\n"
-"#define make_uint2 (uint2)\n"
-"#define make_int2 (int2)\n"
-"#define max2 max\n"
-"#define min2 min\n"
-"///////////////////////////////////////\n"
-"//	Vector\n"
-"///////////////////////////////////////\n"
-"__inline\n"
-"float fastDiv(float numerator, float denominator)\n"
-"{\n"
-"	return native_divide(numerator, denominator);	\n"
-"//	return numerator/denominator;	\n"
-"}\n"
-"__inline\n"
-"float4 fastDiv4(float4 numerator, float4 denominator)\n"
-"{\n"
-"	return native_divide(numerator, denominator);	\n"
-"}\n"
-"__inline\n"
-"float fastSqrtf(float f2)\n"
-"{\n"
-"	return native_sqrt(f2);\n"
-"//	return sqrt(f2);\n"
-"}\n"
-"__inline\n"
-"float fastRSqrt(float f2)\n"
-"{\n"
-"	return native_rsqrt(f2);\n"
-"}\n"
-"__inline\n"
-"float fastLength4(float4 v)\n"
-"{\n"
-"	return fast_length(v);\n"
-"}\n"
-"__inline\n"
-"float4 fastNormalize4(float4 v)\n"
-"{\n"
-"	return fast_normalize(v);\n"
-"}\n"
-"__inline\n"
-"float sqrtf(float a)\n"
-"{\n"
-"//	return sqrt(a);\n"
-"	return native_sqrt(a);\n"
-"}\n"
-"__inline\n"
-"float4 cross3(float4 a, float4 b)\n"
-"{\n"
-"	return cross(a,b);\n"
-"}\n"
-"__inline\n"
-"float dot3F4(float4 a, float4 b)\n"
-"{\n"
-"	float4 a1 = make_float4(a.xyz,0.f);\n"
-"	float4 b1 = make_float4(b.xyz,0.f);\n"
-"	return dot(a1, b1);\n"
-"}\n"
-"__inline\n"
-"float length3(const float4 a)\n"
-"{\n"
-"	return sqrtf(dot3F4(a,a));\n"
-"}\n"
-"__inline\n"
-"float dot4(const float4 a, const float4 b)\n"
-"{\n"
-"	return dot( a, b );\n"
-"}\n"
-"//	for height\n"
-"__inline\n"
-"float dot3w1(const float4 point, const float4 eqn)\n"
-"{\n"
-"	return dot3F4(point,eqn) + eqn.w;\n"
-"}\n"
-"__inline\n"
-"float4 normalize3(const float4 a)\n"
-"{\n"
-"	float4 n = make_float4(a.x, a.y, a.z, 0.f);\n"
-"	return fastNormalize4( n );\n"
-"//	float length = sqrtf(dot3F4(a, a));\n"
-"//	return 1.f/length * a;\n"
-"}\n"
-"__inline\n"
-"float4 normalize4(const float4 a)\n"
-"{\n"
-"	float length = sqrtf(dot4(a, a));\n"
-"	return 1.f/length * a;\n"
-"}\n"
-"__inline\n"
-"float4 createEquation(const float4 a, const float4 b, const float4 c)\n"
-"{\n"
-"	float4 eqn;\n"
-"	float4 ab = b-a;\n"
-"	float4 ac = c-a;\n"
-"	eqn = normalize3( cross3(ab, ac) );\n"
-"	eqn.w = -dot3F4(eqn,a);\n"
-"	return eqn;\n"
-"}\n"
-"#define WG_SIZE 64\n"
-"typedef struct\n"
-"{\n"
-"	int m_nConstraints;\n"
-"	int m_start;\n"
-"	int m_batchIdx;\n"
-"	int m_nSplit;\n"
-"//	int m_paddings[1];\n"
-"} ConstBuffer;\n"
-"typedef struct\n"
-"{\n"
-"	int m_solveFriction;\n"
-"	int m_maxBatch;	//	long batch really kills the performance\n"
-"	int m_batchIdx;\n"
-"	int m_nSplit;\n"
-"//	int m_paddings[1];\n"
-"} ConstBufferBatchSolve;\n"
-" \n"
-"typedef struct \n"
-"{\n"
-"	int m_valInt0;\n"
-"	int m_valInt1;\n"
-"	int m_valInt2;\n"
-"	int m_valInt3;\n"
-"	float m_val0;\n"
-"	float m_val1;\n"
-"	float m_val2;\n"
-"	float m_val3;\n"
-"} SolverDebugInfo;\n"
-"typedef struct\n"
-"{\n"
-"	int m_nContacts;\n"
-"	float m_dt;\n"
-"	float m_positionDrift;\n"
-"	float m_positionConstraintCoeff;\n"
-"} ConstBufferCTC;\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"void ContactToConstraintKernel(__global struct b3Contact4Data* gContact, __global b3RigidBodyData_t* gBodies, __global b3InertiaData_t* gShapes, __global b3ContactConstraint4_t* gConstraintOut, \n"
-"int nContacts,\n"
-"float dt,\n"
-"float positionDrift,\n"
-"float positionConstraintCoeff\n"
-")\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"	\n"
-"	if( gIdx < nContacts )\n"
-"	{\n"
-"		int aIdx = abs(gContact[gIdx].m_bodyAPtrAndSignBit);\n"
-"		int bIdx = abs(gContact[gIdx].m_bodyBPtrAndSignBit);\n"
-"		float4 posA = gBodies[aIdx].m_pos;\n"
-"		float4 linVelA = gBodies[aIdx].m_linVel;\n"
-"		float4 angVelA = gBodies[aIdx].m_angVel;\n"
-"		float invMassA = gBodies[aIdx].m_invMass;\n"
-"		b3Mat3x3 invInertiaA = gShapes[aIdx].m_initInvInertia;\n"
-"		float4 posB = gBodies[bIdx].m_pos;\n"
-"		float4 linVelB = gBodies[bIdx].m_linVel;\n"
-"		float4 angVelB = gBodies[bIdx].m_angVel;\n"
-"		float invMassB = gBodies[bIdx].m_invMass;\n"
-"		b3Mat3x3 invInertiaB = gShapes[bIdx].m_initInvInertia;\n"
-"		b3ContactConstraint4_t cs;\n"
-"    	setConstraint4( posA, linVelA, angVelA, invMassA, invInertiaA, posB, linVelB, angVelB, invMassB, invInertiaB,\n"
-"			&gContact[gIdx], dt, positionDrift, positionConstraintCoeff,\n"
-"			&cs );\n"
-"		\n"
-"		cs.m_batchIdx = gContact[gIdx].m_batchIdx;\n"
-"		gConstraintOut[gIdx] = cs;\n"
-"	}\n"
-"}\n"
-;
+static const char* solverSetupCL =
+	"/*\n"
+	"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
+	"This software is provided 'as-is', without any express or implied warranty.\n"
+	"In no event will the authors be held liable for any damages arising from the use of this software.\n"
+	"Permission is granted to anyone to use this software for any purpose, \n"
+	"including commercial applications, and to alter it and redistribute it freely, \n"
+	"subject to the following restrictions:\n"
+	"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
+	"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
+	"3. This notice may not be removed or altered from any source distribution.\n"
+	"*/\n"
+	"//Originally written by Takahiro Harada\n"
+	"#ifndef B3_CONTACT4DATA_H\n"
+	"#define B3_CONTACT4DATA_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#define B3_FLOAT4_H\n"
+	"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
+	"#define B3_PLATFORM_DEFINITIONS_H\n"
+	"struct MyTest\n"
+	"{\n"
+	"	int bla;\n"
+	"};\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
+	"#define B3_LARGE_FLOAT 1e18f\n"
+	"#define B3_INFINITY 1e18f\n"
+	"#define b3Assert(a)\n"
+	"#define b3ConstArray(a) __global const a*\n"
+	"#define b3AtomicInc atomic_inc\n"
+	"#define b3AtomicAdd atomic_add\n"
+	"#define b3Fabs fabs\n"
+	"#define b3Sqrt native_sqrt\n"
+	"#define b3Sin native_sin\n"
+	"#define b3Cos native_cos\n"
+	"#define B3_STATIC\n"
+	"#endif\n"
+	"#endif\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"	typedef float4	b3Float4;\n"
+	"	#define b3Float4ConstArg const b3Float4\n"
+	"	#define b3MakeFloat4 (float4)\n"
+	"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return dot(a1, b1);\n"
+	"	}\n"
+	"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return cross(a1, b1);\n"
+	"	}\n"
+	"	#define b3MinFloat4 min\n"
+	"	#define b3MaxFloat4 max\n"
+	"	#define b3Normalized(a) normalize(a)\n"
+	"#endif \n"
+	"		\n"
+	"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
+	"{\n"
+	"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
+	"		return false;\n"
+	"	return true;\n"
+	"}\n"
+	"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
+	"{\n"
+	"    float maxDot = -B3_INFINITY;\n"
+	"    int i = 0;\n"
+	"    int ptIndex = -1;\n"
+	"    for( i = 0; i < vecLen; i++ )\n"
+	"    {\n"
+	"        float dot = b3Dot3F4(vecArray[i],vec);\n"
+	"            \n"
+	"        if( dot > maxDot )\n"
+	"        {\n"
+	"            maxDot = dot;\n"
+	"            ptIndex = i;\n"
+	"        }\n"
+	"    }\n"
+	"	b3Assert(ptIndex>=0);\n"
+	"    if (ptIndex<0)\n"
+	"	{\n"
+	"		ptIndex = 0;\n"
+	"	}\n"
+	"    *dotOut = maxDot;\n"
+	"    return ptIndex;\n"
+	"}\n"
+	"#endif //B3_FLOAT4_H\n"
+	"typedef  struct b3Contact4Data b3Contact4Data_t;\n"
+	"struct b3Contact4Data\n"
+	"{\n"
+	"	b3Float4	m_worldPosB[4];\n"
+	"//	b3Float4	m_localPosA[4];\n"
+	"//	b3Float4	m_localPosB[4];\n"
+	"	b3Float4	m_worldNormalOnB;	//	w: m_nPoints\n"
+	"	unsigned short  m_restituitionCoeffCmp;\n"
+	"	unsigned short  m_frictionCoeffCmp;\n"
+	"	int m_batchIdx;\n"
+	"	int m_bodyAPtrAndSignBit;//x:m_bodyAPtr, y:m_bodyBPtr\n"
+	"	int m_bodyBPtrAndSignBit;\n"
+	"	int	m_childIndexA;\n"
+	"	int	m_childIndexB;\n"
+	"	int m_unused1;\n"
+	"	int m_unused2;\n"
+	"};\n"
+	"inline int b3Contact4Data_getNumPoints(const struct b3Contact4Data* contact)\n"
+	"{\n"
+	"	return (int)contact->m_worldNormalOnB.w;\n"
+	"};\n"
+	"inline void b3Contact4Data_setNumPoints(struct b3Contact4Data* contact, int numPoints)\n"
+	"{\n"
+	"	contact->m_worldNormalOnB.w = (float)numPoints;\n"
+	"};\n"
+	"#endif //B3_CONTACT4DATA_H\n"
+	"#ifndef B3_CONTACT_CONSTRAINT5_H\n"
+	"#define B3_CONTACT_CONSTRAINT5_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_FLOAT4_H\n"
+	"typedef struct b3ContactConstraint4 b3ContactConstraint4_t;\n"
+	"struct b3ContactConstraint4\n"
+	"{\n"
+	"	b3Float4 m_linear;//normal?\n"
+	"	b3Float4 m_worldPos[4];\n"
+	"	b3Float4 m_center;	//	friction\n"
+	"	float m_jacCoeffInv[4];\n"
+	"	float m_b[4];\n"
+	"	float m_appliedRambdaDt[4];\n"
+	"	float m_fJacCoeffInv[2];	//	friction\n"
+	"	float m_fAppliedRambdaDt[2];	//	friction\n"
+	"	unsigned int m_bodyA;\n"
+	"	unsigned int m_bodyB;\n"
+	"	int			m_batchIdx;\n"
+	"	unsigned int m_paddings;\n"
+	"};\n"
+	"//inline	void setFrictionCoeff(float value) { m_linear[3] = value; }\n"
+	"inline	float b3GetFrictionCoeff(b3ContactConstraint4_t* constraint) \n"
+	"{\n"
+	"	return constraint->m_linear.w; \n"
+	"}\n"
+	"#endif //B3_CONTACT_CONSTRAINT5_H\n"
+	"#ifndef B3_RIGIDBODY_DATA_H\n"
+	"#define B3_RIGIDBODY_DATA_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifndef B3_QUAT_H\n"
+	"#define B3_QUAT_H\n"
+	"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif\n"
+	"#endif\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"	typedef float4	b3Quat;\n"
+	"	#define b3QuatConstArg const b3Quat\n"
+	"	\n"
+	"	\n"
+	"inline float4 b3FastNormalize4(float4 v)\n"
+	"{\n"
+	"	v = (float4)(v.xyz,0.f);\n"
+	"	return fast_normalize(v);\n"
+	"}\n"
+	"	\n"
+	"inline b3Quat b3QuatMul(b3Quat a, b3Quat b);\n"
+	"inline b3Quat b3QuatNormalized(b3QuatConstArg in);\n"
+	"inline b3Quat b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec);\n"
+	"inline b3Quat b3QuatInvert(b3QuatConstArg q);\n"
+	"inline b3Quat b3QuatInverse(b3QuatConstArg q);\n"
+	"inline b3Quat b3QuatMul(b3QuatConstArg a, b3QuatConstArg b)\n"
+	"{\n"
+	"	b3Quat ans;\n"
+	"	ans = b3Cross3( a, b );\n"
+	"	ans += a.w*b+b.w*a;\n"
+	"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
+	"	ans.w = a.w*b.w - b3Dot3F4(a, b);\n"
+	"	return ans;\n"
+	"}\n"
+	"inline b3Quat b3QuatNormalized(b3QuatConstArg in)\n"
+	"{\n"
+	"	b3Quat q;\n"
+	"	q=in;\n"
+	"	//return b3FastNormalize4(in);\n"
+	"	float len = native_sqrt(dot(q, q));\n"
+	"	if(len > 0.f)\n"
+	"	{\n"
+	"		q *= 1.f / len;\n"
+	"	}\n"
+	"	else\n"
+	"	{\n"
+	"		q.x = q.y = q.z = 0.f;\n"
+	"		q.w = 1.f;\n"
+	"	}\n"
+	"	return q;\n"
+	"}\n"
+	"inline float4 b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
+	"{\n"
+	"	b3Quat qInv = b3QuatInvert( q );\n"
+	"	float4 vcpy = vec;\n"
+	"	vcpy.w = 0.f;\n"
+	"	float4 out = b3QuatMul(b3QuatMul(q,vcpy),qInv);\n"
+	"	return out;\n"
+	"}\n"
+	"inline b3Quat b3QuatInverse(b3QuatConstArg q)\n"
+	"{\n"
+	"	return (b3Quat)(-q.xyz, q.w);\n"
+	"}\n"
+	"inline b3Quat b3QuatInvert(b3QuatConstArg q)\n"
+	"{\n"
+	"	return (b3Quat)(-q.xyz, q.w);\n"
+	"}\n"
+	"inline float4 b3QuatInvRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
+	"{\n"
+	"	return b3QuatRotate( b3QuatInvert( q ), vec );\n"
+	"}\n"
+	"inline b3Float4 b3TransformPoint(b3Float4ConstArg point, b3Float4ConstArg translation, b3QuatConstArg  orientation)\n"
+	"{\n"
+	"	return b3QuatRotate( orientation, point ) + (translation);\n"
+	"}\n"
+	"	\n"
+	"#endif \n"
+	"#endif //B3_QUAT_H\n"
+	"#ifndef B3_MAT3x3_H\n"
+	"#define B3_MAT3x3_H\n"
+	"#ifndef B3_QUAT_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_QUAT_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"typedef struct\n"
+	"{\n"
+	"	b3Float4 m_row[3];\n"
+	"}b3Mat3x3;\n"
+	"#define b3Mat3x3ConstArg const b3Mat3x3\n"
+	"#define b3GetRow(m,row) (m.m_row[row])\n"
+	"inline b3Mat3x3 b3QuatGetRotationMatrix(b3Quat quat)\n"
+	"{\n"
+	"	b3Float4 quat2 = (b3Float4)(quat.x*quat.x, quat.y*quat.y, quat.z*quat.z, 0.f);\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0].x=1-2*quat2.y-2*quat2.z;\n"
+	"	out.m_row[0].y=2*quat.x*quat.y-2*quat.w*quat.z;\n"
+	"	out.m_row[0].z=2*quat.x*quat.z+2*quat.w*quat.y;\n"
+	"	out.m_row[0].w = 0.f;\n"
+	"	out.m_row[1].x=2*quat.x*quat.y+2*quat.w*quat.z;\n"
+	"	out.m_row[1].y=1-2*quat2.x-2*quat2.z;\n"
+	"	out.m_row[1].z=2*quat.y*quat.z-2*quat.w*quat.x;\n"
+	"	out.m_row[1].w = 0.f;\n"
+	"	out.m_row[2].x=2*quat.x*quat.z-2*quat.w*quat.y;\n"
+	"	out.m_row[2].y=2*quat.y*quat.z+2*quat.w*quat.x;\n"
+	"	out.m_row[2].z=1-2*quat2.x-2*quat2.y;\n"
+	"	out.m_row[2].w = 0.f;\n"
+	"	return out;\n"
+	"}\n"
+	"inline b3Mat3x3 b3AbsoluteMat3x3(b3Mat3x3ConstArg matIn)\n"
+	"{\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0] = fabs(matIn.m_row[0]);\n"
+	"	out.m_row[1] = fabs(matIn.m_row[1]);\n"
+	"	out.m_row[2] = fabs(matIn.m_row[2]);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtZero();\n"
+	"__inline\n"
+	"b3Mat3x3 mtIdentity();\n"
+	"__inline\n"
+	"b3Mat3x3 mtTranspose(b3Mat3x3 m);\n"
+	"__inline\n"
+	"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b);\n"
+	"__inline\n"
+	"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b);\n"
+	"__inline\n"
+	"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b);\n"
+	"__inline\n"
+	"b3Mat3x3 mtZero()\n"
+	"{\n"
+	"	b3Mat3x3 m;\n"
+	"	m.m_row[0] = (b3Float4)(0.f);\n"
+	"	m.m_row[1] = (b3Float4)(0.f);\n"
+	"	m.m_row[2] = (b3Float4)(0.f);\n"
+	"	return m;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtIdentity()\n"
+	"{\n"
+	"	b3Mat3x3 m;\n"
+	"	m.m_row[0] = (b3Float4)(1,0,0,0);\n"
+	"	m.m_row[1] = (b3Float4)(0,1,0,0);\n"
+	"	m.m_row[2] = (b3Float4)(0,0,1,0);\n"
+	"	return m;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtTranspose(b3Mat3x3 m)\n"
+	"{\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0] = (b3Float4)(m.m_row[0].x, m.m_row[1].x, m.m_row[2].x, 0.f);\n"
+	"	out.m_row[1] = (b3Float4)(m.m_row[0].y, m.m_row[1].y, m.m_row[2].y, 0.f);\n"
+	"	out.m_row[2] = (b3Float4)(m.m_row[0].z, m.m_row[1].z, m.m_row[2].z, 0.f);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b)\n"
+	"{\n"
+	"	b3Mat3x3 transB;\n"
+	"	transB = mtTranspose( b );\n"
+	"	b3Mat3x3 ans;\n"
+	"	//	why this doesn't run when 0ing in the for{}\n"
+	"	a.m_row[0].w = 0.f;\n"
+	"	a.m_row[1].w = 0.f;\n"
+	"	a.m_row[2].w = 0.f;\n"
+	"	for(int i=0; i<3; i++)\n"
+	"	{\n"
+	"//	a.m_row[i].w = 0.f;\n"
+	"		ans.m_row[i].x = b3Dot3F4(a.m_row[i],transB.m_row[0]);\n"
+	"		ans.m_row[i].y = b3Dot3F4(a.m_row[i],transB.m_row[1]);\n"
+	"		ans.m_row[i].z = b3Dot3F4(a.m_row[i],transB.m_row[2]);\n"
+	"		ans.m_row[i].w = 0.f;\n"
+	"	}\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b)\n"
+	"{\n"
+	"	b3Float4 ans;\n"
+	"	ans.x = b3Dot3F4( a.m_row[0], b );\n"
+	"	ans.y = b3Dot3F4( a.m_row[1], b );\n"
+	"	ans.z = b3Dot3F4( a.m_row[2], b );\n"
+	"	ans.w = 0.f;\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b)\n"
+	"{\n"
+	"	b3Float4 colx = b3MakeFloat4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
+	"	b3Float4 coly = b3MakeFloat4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
+	"	b3Float4 colz = b3MakeFloat4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
+	"	b3Float4 ans;\n"
+	"	ans.x = b3Dot3F4( a, colx );\n"
+	"	ans.y = b3Dot3F4( a, coly );\n"
+	"	ans.z = b3Dot3F4( a, colz );\n"
+	"	return ans;\n"
+	"}\n"
+	"#endif\n"
+	"#endif //B3_MAT3x3_H\n"
+	"typedef struct b3RigidBodyData b3RigidBodyData_t;\n"
+	"struct b3RigidBodyData\n"
+	"{\n"
+	"	b3Float4				m_pos;\n"
+	"	b3Quat					m_quat;\n"
+	"	b3Float4				m_linVel;\n"
+	"	b3Float4				m_angVel;\n"
+	"	int 					m_collidableIdx;\n"
+	"	float 				m_invMass;\n"
+	"	float 				m_restituitionCoeff;\n"
+	"	float 				m_frictionCoeff;\n"
+	"};\n"
+	"typedef struct b3InertiaData b3InertiaData_t;\n"
+	"struct b3InertiaData\n"
+	"{\n"
+	"	b3Mat3x3 m_invInertiaWorld;\n"
+	"	b3Mat3x3 m_initInvInertia;\n"
+	"};\n"
+	"#endif //B3_RIGIDBODY_DATA_H\n"
+	"	\n"
+	"void b3PlaneSpace1 (b3Float4ConstArg n, b3Float4* p, b3Float4* q);\n"
+	" void b3PlaneSpace1 (b3Float4ConstArg n, b3Float4* p, b3Float4* q)\n"
+	"{\n"
+	"  if (b3Fabs(n.z) > 0.70710678f) {\n"
+	"    // choose p in y-z plane\n"
+	"    float a = n.y*n.y + n.z*n.z;\n"
+	"    float k = 1.f/sqrt(a);\n"
+	"    p[0].x = 0;\n"
+	"	p[0].y = -n.z*k;\n"
+	"	p[0].z = n.y*k;\n"
+	"    // set q = n x p\n"
+	"    q[0].x = a*k;\n"
+	"	q[0].y = -n.x*p[0].z;\n"
+	"	q[0].z = n.x*p[0].y;\n"
+	"  }\n"
+	"  else {\n"
+	"    // choose p in x-y plane\n"
+	"    float a = n.x*n.x + n.y*n.y;\n"
+	"    float k = 1.f/sqrt(a);\n"
+	"    p[0].x = -n.y*k;\n"
+	"	p[0].y = n.x*k;\n"
+	"	p[0].z = 0;\n"
+	"    // set q = n x p\n"
+	"    q[0].x = -n.z*p[0].y;\n"
+	"	q[0].y = n.z*p[0].x;\n"
+	"	q[0].z = a*k;\n"
+	"  }\n"
+	"}\n"
+	" \n"
+	"void setLinearAndAngular( b3Float4ConstArg n, b3Float4ConstArg r0, b3Float4ConstArg r1, b3Float4* linear, b3Float4* angular0, b3Float4* angular1)\n"
+	"{\n"
+	"	*linear = b3MakeFloat4(n.x,n.y,n.z,0.f);\n"
+	"	*angular0 = b3Cross3(r0, n);\n"
+	"	*angular1 = -b3Cross3(r1, n);\n"
+	"}\n"
+	"float calcRelVel( b3Float4ConstArg l0, b3Float4ConstArg l1, b3Float4ConstArg a0, b3Float4ConstArg a1, b3Float4ConstArg linVel0,\n"
+	"	b3Float4ConstArg angVel0, b3Float4ConstArg linVel1, b3Float4ConstArg angVel1 )\n"
+	"{\n"
+	"	return b3Dot3F4(l0, linVel0) + b3Dot3F4(a0, angVel0) + b3Dot3F4(l1, linVel1) + b3Dot3F4(a1, angVel1);\n"
+	"}\n"
+	"float calcJacCoeff(b3Float4ConstArg linear0, b3Float4ConstArg linear1, b3Float4ConstArg angular0, b3Float4ConstArg angular1,\n"
+	"					float invMass0, const b3Mat3x3* invInertia0, float invMass1, const b3Mat3x3* invInertia1)\n"
+	"{\n"
+	"	//	linear0,1 are normlized\n"
+	"	float jmj0 = invMass0;//b3Dot3F4(linear0, linear0)*invMass0;\n"
+	"	float jmj1 = b3Dot3F4(mtMul3(angular0,*invInertia0), angular0);\n"
+	"	float jmj2 = invMass1;//b3Dot3F4(linear1, linear1)*invMass1;\n"
+	"	float jmj3 = b3Dot3F4(mtMul3(angular1,*invInertia1), angular1);\n"
+	"	return -1.f/(jmj0+jmj1+jmj2+jmj3);\n"
+	"}\n"
+	"void setConstraint4( b3Float4ConstArg posA, b3Float4ConstArg linVelA, b3Float4ConstArg angVelA, float invMassA, b3Mat3x3ConstArg invInertiaA,\n"
+	"	b3Float4ConstArg posB, b3Float4ConstArg linVelB, b3Float4ConstArg angVelB, float invMassB, b3Mat3x3ConstArg invInertiaB, \n"
+	"	__global struct b3Contact4Data* src, float dt, float positionDrift, float positionConstraintCoeff,\n"
+	"	b3ContactConstraint4_t* dstC )\n"
+	"{\n"
+	"	dstC->m_bodyA = abs(src->m_bodyAPtrAndSignBit);\n"
+	"	dstC->m_bodyB = abs(src->m_bodyBPtrAndSignBit);\n"
+	"	float dtInv = 1.f/dt;\n"
+	"	for(int ic=0; ic<4; ic++)\n"
+	"	{\n"
+	"		dstC->m_appliedRambdaDt[ic] = 0.f;\n"
+	"	}\n"
+	"	dstC->m_fJacCoeffInv[0] = dstC->m_fJacCoeffInv[1] = 0.f;\n"
+	"	dstC->m_linear = src->m_worldNormalOnB;\n"
+	"	dstC->m_linear.w = 0.7f ;//src->getFrictionCoeff() );\n"
+	"	for(int ic=0; ic<4; ic++)\n"
+	"	{\n"
+	"		b3Float4 r0 = src->m_worldPosB[ic] - posA;\n"
+	"		b3Float4 r1 = src->m_worldPosB[ic] - posB;\n"
+	"		if( ic >= src->m_worldNormalOnB.w )//npoints\n"
+	"		{\n"
+	"			dstC->m_jacCoeffInv[ic] = 0.f;\n"
+	"			continue;\n"
+	"		}\n"
+	"		float relVelN;\n"
+	"		{\n"
+	"			b3Float4 linear, angular0, angular1;\n"
+	"			setLinearAndAngular(src->m_worldNormalOnB, r0, r1, &linear, &angular0, &angular1);\n"
+	"			dstC->m_jacCoeffInv[ic] = calcJacCoeff(linear, -linear, angular0, angular1,\n"
+	"				invMassA, &invInertiaA, invMassB, &invInertiaB );\n"
+	"			relVelN = calcRelVel(linear, -linear, angular0, angular1,\n"
+	"				linVelA, angVelA, linVelB, angVelB);\n"
+	"			float e = 0.f;//src->getRestituitionCoeff();\n"
+	"			if( relVelN*relVelN < 0.004f ) e = 0.f;\n"
+	"			dstC->m_b[ic] = e*relVelN;\n"
+	"			//float penetration = src->m_worldPosB[ic].w;\n"
+	"			dstC->m_b[ic] += (src->m_worldPosB[ic].w + positionDrift)*positionConstraintCoeff*dtInv;\n"
+	"			dstC->m_appliedRambdaDt[ic] = 0.f;\n"
+	"		}\n"
+	"	}\n"
+	"	if( src->m_worldNormalOnB.w > 0 )//npoints\n"
+	"	{	//	prepare friction\n"
+	"		b3Float4 center = b3MakeFloat4(0.f,0.f,0.f,0.f);\n"
+	"		for(int i=0; i<src->m_worldNormalOnB.w; i++) \n"
+	"			center += src->m_worldPosB[i];\n"
+	"		center /= (float)src->m_worldNormalOnB.w;\n"
+	"		b3Float4 tangent[2];\n"
+	"		b3PlaneSpace1(src->m_worldNormalOnB,&tangent[0],&tangent[1]);\n"
+	"		\n"
+	"		b3Float4 r[2];\n"
+	"		r[0] = center - posA;\n"
+	"		r[1] = center - posB;\n"
+	"		for(int i=0; i<2; i++)\n"
+	"		{\n"
+	"			b3Float4 linear, angular0, angular1;\n"
+	"			setLinearAndAngular(tangent[i], r[0], r[1], &linear, &angular0, &angular1);\n"
+	"			dstC->m_fJacCoeffInv[i] = calcJacCoeff(linear, -linear, angular0, angular1,\n"
+	"				invMassA, &invInertiaA, invMassB, &invInertiaB );\n"
+	"			dstC->m_fAppliedRambdaDt[i] = 0.f;\n"
+	"		}\n"
+	"		dstC->m_center = center;\n"
+	"	}\n"
+	"	for(int i=0; i<4; i++)\n"
+	"	{\n"
+	"		if( i<src->m_worldNormalOnB.w )\n"
+	"		{\n"
+	"			dstC->m_worldPos[i] = src->m_worldPosB[i];\n"
+	"		}\n"
+	"		else\n"
+	"		{\n"
+	"			dstC->m_worldPos[i] = b3MakeFloat4(0.f,0.f,0.f,0.f);\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_extended_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics : enable\n"
+	"#ifdef cl_ext_atomic_counters_32\n"
+	"#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
+	"#else\n"
+	"#define counter32_t volatile global int*\n"
+	"#endif\n"
+	"typedef unsigned int u32;\n"
+	"typedef unsigned short u16;\n"
+	"typedef unsigned char u8;\n"
+	"#define GET_GROUP_IDX get_group_id(0)\n"
+	"#define GET_LOCAL_IDX get_local_id(0)\n"
+	"#define GET_GLOBAL_IDX get_global_id(0)\n"
+	"#define GET_GROUP_SIZE get_local_size(0)\n"
+	"#define GET_NUM_GROUPS get_num_groups(0)\n"
+	"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
+	"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
+	"#define AtomInc(x) atom_inc(&(x))\n"
+	"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
+	"#define AppendInc(x, out) out = atomic_inc(x)\n"
+	"#define AtomAdd(x, value) atom_add(&(x), value)\n"
+	"#define AtomCmpxhg(x, cmp, value) atom_cmpxchg( &(x), cmp, value )\n"
+	"#define AtomXhg(x, value) atom_xchg ( &(x), value )\n"
+	"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
+	"#define make_float4 (float4)\n"
+	"#define make_float2 (float2)\n"
+	"#define make_uint4 (uint4)\n"
+	"#define make_int4 (int4)\n"
+	"#define make_uint2 (uint2)\n"
+	"#define make_int2 (int2)\n"
+	"#define max2 max\n"
+	"#define min2 min\n"
+	"///////////////////////////////////////\n"
+	"//	Vector\n"
+	"///////////////////////////////////////\n"
+	"__inline\n"
+	"float fastDiv(float numerator, float denominator)\n"
+	"{\n"
+	"	return native_divide(numerator, denominator);	\n"
+	"//	return numerator/denominator;	\n"
+	"}\n"
+	"__inline\n"
+	"float4 fastDiv4(float4 numerator, float4 denominator)\n"
+	"{\n"
+	"	return native_divide(numerator, denominator);	\n"
+	"}\n"
+	"__inline\n"
+	"float fastSqrtf(float f2)\n"
+	"{\n"
+	"	return native_sqrt(f2);\n"
+	"//	return sqrt(f2);\n"
+	"}\n"
+	"__inline\n"
+	"float fastRSqrt(float f2)\n"
+	"{\n"
+	"	return native_rsqrt(f2);\n"
+	"}\n"
+	"__inline\n"
+	"float fastLength4(float4 v)\n"
+	"{\n"
+	"	return fast_length(v);\n"
+	"}\n"
+	"__inline\n"
+	"float4 fastNormalize4(float4 v)\n"
+	"{\n"
+	"	return fast_normalize(v);\n"
+	"}\n"
+	"__inline\n"
+	"float sqrtf(float a)\n"
+	"{\n"
+	"//	return sqrt(a);\n"
+	"	return native_sqrt(a);\n"
+	"}\n"
+	"__inline\n"
+	"float4 cross3(float4 a, float4 b)\n"
+	"{\n"
+	"	return cross(a,b);\n"
+	"}\n"
+	"__inline\n"
+	"float dot3F4(float4 a, float4 b)\n"
+	"{\n"
+	"	float4 a1 = make_float4(a.xyz,0.f);\n"
+	"	float4 b1 = make_float4(b.xyz,0.f);\n"
+	"	return dot(a1, b1);\n"
+	"}\n"
+	"__inline\n"
+	"float length3(const float4 a)\n"
+	"{\n"
+	"	return sqrtf(dot3F4(a,a));\n"
+	"}\n"
+	"__inline\n"
+	"float dot4(const float4 a, const float4 b)\n"
+	"{\n"
+	"	return dot( a, b );\n"
+	"}\n"
+	"//	for height\n"
+	"__inline\n"
+	"float dot3w1(const float4 point, const float4 eqn)\n"
+	"{\n"
+	"	return dot3F4(point,eqn) + eqn.w;\n"
+	"}\n"
+	"__inline\n"
+	"float4 normalize3(const float4 a)\n"
+	"{\n"
+	"	float4 n = make_float4(a.x, a.y, a.z, 0.f);\n"
+	"	return fastNormalize4( n );\n"
+	"//	float length = sqrtf(dot3F4(a, a));\n"
+	"//	return 1.f/length * a;\n"
+	"}\n"
+	"__inline\n"
+	"float4 normalize4(const float4 a)\n"
+	"{\n"
+	"	float length = sqrtf(dot4(a, a));\n"
+	"	return 1.f/length * a;\n"
+	"}\n"
+	"__inline\n"
+	"float4 createEquation(const float4 a, const float4 b, const float4 c)\n"
+	"{\n"
+	"	float4 eqn;\n"
+	"	float4 ab = b-a;\n"
+	"	float4 ac = c-a;\n"
+	"	eqn = normalize3( cross3(ab, ac) );\n"
+	"	eqn.w = -dot3F4(eqn,a);\n"
+	"	return eqn;\n"
+	"}\n"
+	"#define WG_SIZE 64\n"
+	"typedef struct\n"
+	"{\n"
+	"	int m_nConstraints;\n"
+	"	int m_start;\n"
+	"	int m_batchIdx;\n"
+	"	int m_nSplit;\n"
+	"//	int m_paddings[1];\n"
+	"} ConstBuffer;\n"
+	"typedef struct\n"
+	"{\n"
+	"	int m_solveFriction;\n"
+	"	int m_maxBatch;	//	long batch really kills the performance\n"
+	"	int m_batchIdx;\n"
+	"	int m_nSplit;\n"
+	"//	int m_paddings[1];\n"
+	"} ConstBufferBatchSolve;\n"
+	" \n"
+	"typedef struct \n"
+	"{\n"
+	"	int m_valInt0;\n"
+	"	int m_valInt1;\n"
+	"	int m_valInt2;\n"
+	"	int m_valInt3;\n"
+	"	float m_val0;\n"
+	"	float m_val1;\n"
+	"	float m_val2;\n"
+	"	float m_val3;\n"
+	"} SolverDebugInfo;\n"
+	"typedef struct\n"
+	"{\n"
+	"	int m_nContacts;\n"
+	"	float m_dt;\n"
+	"	float m_positionDrift;\n"
+	"	float m_positionConstraintCoeff;\n"
+	"} ConstBufferCTC;\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"void ContactToConstraintKernel(__global struct b3Contact4Data* gContact, __global b3RigidBodyData_t* gBodies, __global b3InertiaData_t* gShapes, __global b3ContactConstraint4_t* gConstraintOut, \n"
+	"int nContacts,\n"
+	"float dt,\n"
+	"float positionDrift,\n"
+	"float positionConstraintCoeff\n"
+	")\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"	\n"
+	"	if( gIdx < nContacts )\n"
+	"	{\n"
+	"		int aIdx = abs(gContact[gIdx].m_bodyAPtrAndSignBit);\n"
+	"		int bIdx = abs(gContact[gIdx].m_bodyBPtrAndSignBit);\n"
+	"		float4 posA = gBodies[aIdx].m_pos;\n"
+	"		float4 linVelA = gBodies[aIdx].m_linVel;\n"
+	"		float4 angVelA = gBodies[aIdx].m_angVel;\n"
+	"		float invMassA = gBodies[aIdx].m_invMass;\n"
+	"		b3Mat3x3 invInertiaA = gShapes[aIdx].m_initInvInertia;\n"
+	"		float4 posB = gBodies[bIdx].m_pos;\n"
+	"		float4 linVelB = gBodies[bIdx].m_linVel;\n"
+	"		float4 angVelB = gBodies[bIdx].m_angVel;\n"
+	"		float invMassB = gBodies[bIdx].m_invMass;\n"
+	"		b3Mat3x3 invInertiaB = gShapes[bIdx].m_initInvInertia;\n"
+	"		b3ContactConstraint4_t cs;\n"
+	"    	setConstraint4( posA, linVelA, angVelA, invMassA, invInertiaA, posB, linVelB, angVelB, invMassB, invInertiaB,\n"
+	"			&gContact[gIdx], dt, positionDrift, positionConstraintCoeff,\n"
+	"			&cs );\n"
+	"		\n"
+	"		cs.m_batchIdx = gContact[gIdx].m_batchIdx;\n"
+	"		gConstraintOut[gIdx] = cs;\n"
+	"	}\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/solverSetup2.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/solverSetup2.h
index 1b5819f6cf3..1e6e3579b68 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/solverSetup2.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/solverSetup2.h
@@ -1,601 +1,600 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* solverSetup2CL= \
-"/*\n"
-"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
-"This software is provided 'as-is', without any express or implied warranty.\n"
-"In no event will the authors be held liable for any damages arising from the use of this software.\n"
-"Permission is granted to anyone to use this software for any purpose, \n"
-"including commercial applications, and to alter it and redistribute it freely, \n"
-"subject to the following restrictions:\n"
-"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
-"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
-"3. This notice may not be removed or altered from any source distribution.\n"
-"*/\n"
-"//Originally written by Takahiro Harada\n"
-"#ifndef B3_CONTACT4DATA_H\n"
-"#define B3_CONTACT4DATA_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#define B3_FLOAT4_H\n"
-"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
-"#define B3_PLATFORM_DEFINITIONS_H\n"
-"struct MyTest\n"
-"{\n"
-"	int bla;\n"
-"};\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
-"#define B3_LARGE_FLOAT 1e18f\n"
-"#define B3_INFINITY 1e18f\n"
-"#define b3Assert(a)\n"
-"#define b3ConstArray(a) __global const a*\n"
-"#define b3AtomicInc atomic_inc\n"
-"#define b3AtomicAdd atomic_add\n"
-"#define b3Fabs fabs\n"
-"#define b3Sqrt native_sqrt\n"
-"#define b3Sin native_sin\n"
-"#define b3Cos native_cos\n"
-"#define B3_STATIC\n"
-"#endif\n"
-"#endif\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"	typedef float4	b3Float4;\n"
-"	#define b3Float4ConstArg const b3Float4\n"
-"	#define b3MakeFloat4 (float4)\n"
-"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return dot(a1, b1);\n"
-"	}\n"
-"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return cross(a1, b1);\n"
-"	}\n"
-"	#define b3MinFloat4 min\n"
-"	#define b3MaxFloat4 max\n"
-"	#define b3Normalized(a) normalize(a)\n"
-"#endif \n"
-"		\n"
-"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
-"{\n"
-"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
-"		return false;\n"
-"	return true;\n"
-"}\n"
-"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
-"{\n"
-"    float maxDot = -B3_INFINITY;\n"
-"    int i = 0;\n"
-"    int ptIndex = -1;\n"
-"    for( i = 0; i < vecLen; i++ )\n"
-"    {\n"
-"        float dot = b3Dot3F4(vecArray[i],vec);\n"
-"            \n"
-"        if( dot > maxDot )\n"
-"        {\n"
-"            maxDot = dot;\n"
-"            ptIndex = i;\n"
-"        }\n"
-"    }\n"
-"	b3Assert(ptIndex>=0);\n"
-"    if (ptIndex<0)\n"
-"	{\n"
-"		ptIndex = 0;\n"
-"	}\n"
-"    *dotOut = maxDot;\n"
-"    return ptIndex;\n"
-"}\n"
-"#endif //B3_FLOAT4_H\n"
-"typedef  struct b3Contact4Data b3Contact4Data_t;\n"
-"struct b3Contact4Data\n"
-"{\n"
-"	b3Float4	m_worldPosB[4];\n"
-"//	b3Float4	m_localPosA[4];\n"
-"//	b3Float4	m_localPosB[4];\n"
-"	b3Float4	m_worldNormalOnB;	//	w: m_nPoints\n"
-"	unsigned short  m_restituitionCoeffCmp;\n"
-"	unsigned short  m_frictionCoeffCmp;\n"
-"	int m_batchIdx;\n"
-"	int m_bodyAPtrAndSignBit;//x:m_bodyAPtr, y:m_bodyBPtr\n"
-"	int m_bodyBPtrAndSignBit;\n"
-"	int	m_childIndexA;\n"
-"	int	m_childIndexB;\n"
-"	int m_unused1;\n"
-"	int m_unused2;\n"
-"};\n"
-"inline int b3Contact4Data_getNumPoints(const struct b3Contact4Data* contact)\n"
-"{\n"
-"	return (int)contact->m_worldNormalOnB.w;\n"
-"};\n"
-"inline void b3Contact4Data_setNumPoints(struct b3Contact4Data* contact, int numPoints)\n"
-"{\n"
-"	contact->m_worldNormalOnB.w = (float)numPoints;\n"
-"};\n"
-"#endif //B3_CONTACT4DATA_H\n"
-"#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_extended_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics : enable\n"
-"#ifdef cl_ext_atomic_counters_32\n"
-"#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
-"#else\n"
-"#define counter32_t volatile global int*\n"
-"#endif\n"
-"typedef unsigned int u32;\n"
-"typedef unsigned short u16;\n"
-"typedef unsigned char u8;\n"
-"#define GET_GROUP_IDX get_group_id(0)\n"
-"#define GET_LOCAL_IDX get_local_id(0)\n"
-"#define GET_GLOBAL_IDX get_global_id(0)\n"
-"#define GET_GROUP_SIZE get_local_size(0)\n"
-"#define GET_NUM_GROUPS get_num_groups(0)\n"
-"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
-"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
-"#define AtomInc(x) atom_inc(&(x))\n"
-"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
-"#define AppendInc(x, out) out = atomic_inc(x)\n"
-"#define AtomAdd(x, value) atom_add(&(x), value)\n"
-"#define AtomCmpxhg(x, cmp, value) atom_cmpxchg( &(x), cmp, value )\n"
-"#define AtomXhg(x, value) atom_xchg ( &(x), value )\n"
-"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
-"#define make_float4 (float4)\n"
-"#define make_float2 (float2)\n"
-"#define make_uint4 (uint4)\n"
-"#define make_int4 (int4)\n"
-"#define make_uint2 (uint2)\n"
-"#define make_int2 (int2)\n"
-"#define max2 max\n"
-"#define min2 min\n"
-"///////////////////////////////////////\n"
-"//	Vector\n"
-"///////////////////////////////////////\n"
-"__inline\n"
-"float fastDiv(float numerator, float denominator)\n"
-"{\n"
-"	return native_divide(numerator, denominator);	\n"
-"//	return numerator/denominator;	\n"
-"}\n"
-"__inline\n"
-"float4 fastDiv4(float4 numerator, float4 denominator)\n"
-"{\n"
-"	return native_divide(numerator, denominator);	\n"
-"}\n"
-"__inline\n"
-"float fastSqrtf(float f2)\n"
-"{\n"
-"	return native_sqrt(f2);\n"
-"//	return sqrt(f2);\n"
-"}\n"
-"__inline\n"
-"float fastRSqrt(float f2)\n"
-"{\n"
-"	return native_rsqrt(f2);\n"
-"}\n"
-"__inline\n"
-"float fastLength4(float4 v)\n"
-"{\n"
-"	return fast_length(v);\n"
-"}\n"
-"__inline\n"
-"float4 fastNormalize4(float4 v)\n"
-"{\n"
-"	return fast_normalize(v);\n"
-"}\n"
-"__inline\n"
-"float sqrtf(float a)\n"
-"{\n"
-"//	return sqrt(a);\n"
-"	return native_sqrt(a);\n"
-"}\n"
-"__inline\n"
-"float4 cross3(float4 a, float4 b)\n"
-"{\n"
-"	return cross(a,b);\n"
-"}\n"
-"__inline\n"
-"float dot3F4(float4 a, float4 b)\n"
-"{\n"
-"	float4 a1 = make_float4(a.xyz,0.f);\n"
-"	float4 b1 = make_float4(b.xyz,0.f);\n"
-"	return dot(a1, b1);\n"
-"}\n"
-"__inline\n"
-"float length3(const float4 a)\n"
-"{\n"
-"	return sqrtf(dot3F4(a,a));\n"
-"}\n"
-"__inline\n"
-"float dot4(const float4 a, const float4 b)\n"
-"{\n"
-"	return dot( a, b );\n"
-"}\n"
-"//	for height\n"
-"__inline\n"
-"float dot3w1(const float4 point, const float4 eqn)\n"
-"{\n"
-"	return dot3F4(point,eqn) + eqn.w;\n"
-"}\n"
-"__inline\n"
-"float4 normalize3(const float4 a)\n"
-"{\n"
-"	float4 n = make_float4(a.x, a.y, a.z, 0.f);\n"
-"	return fastNormalize4( n );\n"
-"//	float length = sqrtf(dot3F4(a, a));\n"
-"//	return 1.f/length * a;\n"
-"}\n"
-"__inline\n"
-"float4 normalize4(const float4 a)\n"
-"{\n"
-"	float length = sqrtf(dot4(a, a));\n"
-"	return 1.f/length * a;\n"
-"}\n"
-"__inline\n"
-"float4 createEquation(const float4 a, const float4 b, const float4 c)\n"
-"{\n"
-"	float4 eqn;\n"
-"	float4 ab = b-a;\n"
-"	float4 ac = c-a;\n"
-"	eqn = normalize3( cross3(ab, ac) );\n"
-"	eqn.w = -dot3F4(eqn,a);\n"
-"	return eqn;\n"
-"}\n"
-"///////////////////////////////////////\n"
-"//	Matrix3x3\n"
-"///////////////////////////////////////\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_row[3];\n"
-"}Matrix3x3;\n"
-"__inline\n"
-"Matrix3x3 mtZero();\n"
-"__inline\n"
-"Matrix3x3 mtIdentity();\n"
-"__inline\n"
-"Matrix3x3 mtTranspose(Matrix3x3 m);\n"
-"__inline\n"
-"Matrix3x3 mtMul(Matrix3x3 a, Matrix3x3 b);\n"
-"__inline\n"
-"float4 mtMul1(Matrix3x3 a, float4 b);\n"
-"__inline\n"
-"float4 mtMul3(float4 a, Matrix3x3 b);\n"
-"__inline\n"
-"Matrix3x3 mtZero()\n"
-"{\n"
-"	Matrix3x3 m;\n"
-"	m.m_row[0] = (float4)(0.f);\n"
-"	m.m_row[1] = (float4)(0.f);\n"
-"	m.m_row[2] = (float4)(0.f);\n"
-"	return m;\n"
-"}\n"
-"__inline\n"
-"Matrix3x3 mtIdentity()\n"
-"{\n"
-"	Matrix3x3 m;\n"
-"	m.m_row[0] = (float4)(1,0,0,0);\n"
-"	m.m_row[1] = (float4)(0,1,0,0);\n"
-"	m.m_row[2] = (float4)(0,0,1,0);\n"
-"	return m;\n"
-"}\n"
-"__inline\n"
-"Matrix3x3 mtTranspose(Matrix3x3 m)\n"
-"{\n"
-"	Matrix3x3 out;\n"
-"	out.m_row[0] = (float4)(m.m_row[0].x, m.m_row[1].x, m.m_row[2].x, 0.f);\n"
-"	out.m_row[1] = (float4)(m.m_row[0].y, m.m_row[1].y, m.m_row[2].y, 0.f);\n"
-"	out.m_row[2] = (float4)(m.m_row[0].z, m.m_row[1].z, m.m_row[2].z, 0.f);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"Matrix3x3 mtMul(Matrix3x3 a, Matrix3x3 b)\n"
-"{\n"
-"	Matrix3x3 transB;\n"
-"	transB = mtTranspose( b );\n"
-"	Matrix3x3 ans;\n"
-"	//	why this doesn't run when 0ing in the for{}\n"
-"	a.m_row[0].w = 0.f;\n"
-"	a.m_row[1].w = 0.f;\n"
-"	a.m_row[2].w = 0.f;\n"
-"	for(int i=0; i<3; i++)\n"
-"	{\n"
-"//	a.m_row[i].w = 0.f;\n"
-"		ans.m_row[i].x = dot3F4(a.m_row[i],transB.m_row[0]);\n"
-"		ans.m_row[i].y = dot3F4(a.m_row[i],transB.m_row[1]);\n"
-"		ans.m_row[i].z = dot3F4(a.m_row[i],transB.m_row[2]);\n"
-"		ans.m_row[i].w = 0.f;\n"
-"	}\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"float4 mtMul1(Matrix3x3 a, float4 b)\n"
-"{\n"
-"	float4 ans;\n"
-"	ans.x = dot3F4( a.m_row[0], b );\n"
-"	ans.y = dot3F4( a.m_row[1], b );\n"
-"	ans.z = dot3F4( a.m_row[2], b );\n"
-"	ans.w = 0.f;\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"float4 mtMul3(float4 a, Matrix3x3 b)\n"
-"{\n"
-"	float4 colx = make_float4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
-"	float4 coly = make_float4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
-"	float4 colz = make_float4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
-"	float4 ans;\n"
-"	ans.x = dot3F4( a, colx );\n"
-"	ans.y = dot3F4( a, coly );\n"
-"	ans.z = dot3F4( a, colz );\n"
-"	return ans;\n"
-"}\n"
-"///////////////////////////////////////\n"
-"//	Quaternion\n"
-"///////////////////////////////////////\n"
-"typedef float4 Quaternion;\n"
-"__inline\n"
-"Quaternion qtMul(Quaternion a, Quaternion b);\n"
-"__inline\n"
-"Quaternion qtNormalize(Quaternion in);\n"
-"__inline\n"
-"float4 qtRotate(Quaternion q, float4 vec);\n"
-"__inline\n"
-"Quaternion qtInvert(Quaternion q);\n"
-"__inline\n"
-"Quaternion qtMul(Quaternion a, Quaternion b)\n"
-"{\n"
-"	Quaternion ans;\n"
-"	ans = cross3( a, b );\n"
-"	ans += a.w*b+b.w*a;\n"
-"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
-"	ans.w = a.w*b.w - dot3F4(a, b);\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"Quaternion qtNormalize(Quaternion in)\n"
-"{\n"
-"	return fastNormalize4(in);\n"
-"//	in /= length( in );\n"
-"//	return in;\n"
-"}\n"
-"__inline\n"
-"float4 qtRotate(Quaternion q, float4 vec)\n"
-"{\n"
-"	Quaternion qInv = qtInvert( q );\n"
-"	float4 vcpy = vec;\n"
-"	vcpy.w = 0.f;\n"
-"	float4 out = qtMul(qtMul(q,vcpy),qInv);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"Quaternion qtInvert(Quaternion q)\n"
-"{\n"
-"	return (Quaternion)(-q.xyz, q.w);\n"
-"}\n"
-"__inline\n"
-"float4 qtInvRotate(const Quaternion q, float4 vec)\n"
-"{\n"
-"	return qtRotate( qtInvert( q ), vec );\n"
-"}\n"
-"#define WG_SIZE 64\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_pos;\n"
-"	Quaternion m_quat;\n"
-"	float4 m_linVel;\n"
-"	float4 m_angVel;\n"
-"	u32 m_shapeIdx;\n"
-"	float m_invMass;\n"
-"	float m_restituitionCoeff;\n"
-"	float m_frictionCoeff;\n"
-"} Body;\n"
-"typedef struct\n"
-"{\n"
-"	Matrix3x3 m_invInertia;\n"
-"	Matrix3x3 m_initInvInertia;\n"
-"} Shape;\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_linear;\n"
-"	float4 m_worldPos[4];\n"
-"	float4 m_center;	\n"
-"	float m_jacCoeffInv[4];\n"
-"	float m_b[4];\n"
-"	float m_appliedRambdaDt[4];\n"
-"	float m_fJacCoeffInv[2];	\n"
-"	float m_fAppliedRambdaDt[2];	\n"
-"	u32 m_bodyA;\n"
-"	u32 m_bodyB;\n"
-"	int m_batchIdx;\n"
-"	u32 m_paddings[1];\n"
-"} Constraint4;\n"
-"typedef struct\n"
-"{\n"
-"	int m_nConstraints;\n"
-"	int m_start;\n"
-"	int m_batchIdx;\n"
-"	int m_nSplit;\n"
-"//	int m_paddings[1];\n"
-"} ConstBuffer;\n"
-"typedef struct\n"
-"{\n"
-"	int m_solveFriction;\n"
-"	int m_maxBatch;	//	long batch really kills the performance\n"
-"	int m_batchIdx;\n"
-"	int m_nSplit;\n"
-"//	int m_paddings[1];\n"
-"} ConstBufferBatchSolve;\n"
-" \n"
-"typedef struct \n"
-"{\n"
-"	int m_valInt0;\n"
-"	int m_valInt1;\n"
-"	int m_valInt2;\n"
-"	int m_valInt3;\n"
-"	float m_val0;\n"
-"	float m_val1;\n"
-"	float m_val2;\n"
-"	float m_val3;\n"
-"} SolverDebugInfo;\n"
-"//	others\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"void ReorderContactKernel(__global struct b3Contact4Data* in, __global struct b3Contact4Data* out, __global int2* sortData, int4 cb )\n"
-"{\n"
-"	int nContacts = cb.x;\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"	if( gIdx < nContacts )\n"
-"	{\n"
-"		int srcIdx = sortData[gIdx].y;\n"
-"		out[gIdx] = in[srcIdx];\n"
-"	}\n"
-"}\n"
-"__kernel __attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"void SetDeterminismSortDataChildShapeB(__global struct b3Contact4Data* contactsIn, __global int2* sortDataOut, int nContacts)\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"	if( gIdx < nContacts )\n"
-"	{\n"
-"		int2 sd;\n"
-"		sd.x = contactsIn[gIdx].m_childIndexB;\n"
-"		sd.y = gIdx;\n"
-"		sortDataOut[gIdx] = sd;\n"
-"	}\n"
-"}\n"
-"__kernel __attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"void SetDeterminismSortDataChildShapeA(__global struct b3Contact4Data* contactsIn, __global int2* sortDataInOut, int nContacts)\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"	if( gIdx < nContacts )\n"
-"	{\n"
-"		int2 sdIn;\n"
-"		sdIn = sortDataInOut[gIdx];\n"
-"		int2 sdOut;\n"
-"		sdOut.x = contactsIn[sdIn.y].m_childIndexA;\n"
-"		sdOut.y = sdIn.y;\n"
-"		sortDataInOut[gIdx] = sdOut;\n"
-"	}\n"
-"}\n"
-"__kernel __attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"void SetDeterminismSortDataBodyA(__global struct b3Contact4Data* contactsIn, __global int2* sortDataInOut, int nContacts)\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"	if( gIdx < nContacts )\n"
-"	{\n"
-"		int2 sdIn;\n"
-"		sdIn = sortDataInOut[gIdx];\n"
-"		int2 sdOut;\n"
-"		sdOut.x = contactsIn[sdIn.y].m_bodyAPtrAndSignBit;\n"
-"		sdOut.y = sdIn.y;\n"
-"		sortDataInOut[gIdx] = sdOut;\n"
-"	}\n"
-"}\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"void SetDeterminismSortDataBodyB(__global struct b3Contact4Data* contactsIn, __global int2* sortDataInOut, int nContacts)\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"	if( gIdx < nContacts )\n"
-"	{\n"
-"		int2 sdIn;\n"
-"		sdIn = sortDataInOut[gIdx];\n"
-"		int2 sdOut;\n"
-"		sdOut.x = contactsIn[sdIn.y].m_bodyBPtrAndSignBit;\n"
-"		sdOut.y = sdIn.y;\n"
-"		sortDataInOut[gIdx] = sdOut;\n"
-"	}\n"
-"}\n"
-"typedef struct\n"
-"{\n"
-"	int m_nContacts;\n"
-"	int m_staticIdx;\n"
-"	float m_scale;\n"
-"	int m_nSplit;\n"
-"} ConstBufferSSD;\n"
-"__constant const int gridTable4x4[] = \n"
-"{\n"
-"    0,1,17,16,\n"
-"	1,2,18,19,\n"
-"	17,18,32,3,\n"
-"	16,19,3,34\n"
-"};\n"
-"__constant const int gridTable8x8[] = \n"
-"{\n"
-"	  0,  2,  3, 16, 17, 18, 19,  1,\n"
-"	 66, 64, 80, 67, 82, 81, 65, 83,\n"
-"	131,144,128,130,147,129,145,146,\n"
-"	208,195,194,192,193,211,210,209,\n"
-"	 21, 22, 23,  5,  4,  6,  7, 20,\n"
-"	 86, 85, 69, 87, 70, 68, 84, 71,\n"
-"	151,133,149,150,135,148,132,134,\n"
-"	197,27,214,213,212,199,198,196\n"
-"	\n"
-"};\n"
-"#define USE_SPATIAL_BATCHING 1\n"
-"#define USE_4x4_GRID 1\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"void SetSortDataKernel(__global struct b3Contact4Data* gContact, __global Body* gBodies, __global int2* gSortDataOut, \n"
-"int nContacts,float scale,int4 nSplit,int staticIdx)\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"	\n"
-"	if( gIdx < nContacts )\n"
-"	{\n"
-"		int aPtrAndSignBit  = gContact[gIdx].m_bodyAPtrAndSignBit;\n"
-"		int bPtrAndSignBit  = gContact[gIdx].m_bodyBPtrAndSignBit;\n"
-"		int aIdx = abs(aPtrAndSignBit );\n"
-"		int bIdx = abs(bPtrAndSignBit);\n"
-"		bool aStatic = (aPtrAndSignBit<0) ||(aPtrAndSignBit==staticIdx);\n"
-"		bool bStatic = (bPtrAndSignBit<0) ||(bPtrAndSignBit==staticIdx);\n"
-"#if USE_SPATIAL_BATCHING		\n"
-"		int idx = (aStatic)? bIdx: aIdx;\n"
-"		float4 p = gBodies[idx].m_pos;\n"
-"		int xIdx = (int)((p.x-((p.x<0.f)?1.f:0.f))*scale) & (nSplit.x-1);\n"
-"		int yIdx = (int)((p.y-((p.y<0.f)?1.f:0.f))*scale) & (nSplit.y-1);\n"
-"		int zIdx = (int)((p.z-((p.z<0.f)?1.f:0.f))*scale) & (nSplit.z-1);\n"
-"		int newIndex = (xIdx+yIdx*nSplit.x+zIdx*nSplit.x*nSplit.y);\n"
-"		\n"
-"#else//USE_SPATIAL_BATCHING\n"
-"	#if USE_4x4_GRID\n"
-"		int aa = aIdx&3;\n"
-"		int bb = bIdx&3;\n"
-"		if (aStatic)\n"
-"			aa = bb;\n"
-"		if (bStatic)\n"
-"			bb = aa;\n"
-"		int gridIndex = aa + bb*4;\n"
-"		int newIndex = gridTable4x4[gridIndex];\n"
-"	#else//USE_4x4_GRID\n"
-"		int aa = aIdx&7;\n"
-"		int bb = bIdx&7;\n"
-"		if (aStatic)\n"
-"			aa = bb;\n"
-"		if (bStatic)\n"
-"			bb = aa;\n"
-"		int gridIndex = aa + bb*8;\n"
-"		int newIndex = gridTable8x8[gridIndex];\n"
-"	#endif//USE_4x4_GRID\n"
-"#endif//USE_SPATIAL_BATCHING\n"
-"		gSortDataOut[gIdx].x = newIndex;\n"
-"		gSortDataOut[gIdx].y = gIdx;\n"
-"	}\n"
-"	else\n"
-"	{\n"
-"		gSortDataOut[gIdx].x = 0xffffffff;\n"
-"	}\n"
-"}\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"void CopyConstraintKernel(__global struct b3Contact4Data* gIn, __global struct b3Contact4Data* gOut, int4 cb )\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"	if( gIdx < cb.x )\n"
-"	{\n"
-"		gOut[gIdx] = gIn[gIdx];\n"
-"	}\n"
-"}\n"
-;
+static const char* solverSetup2CL =
+	"/*\n"
+	"Copyright (c) 2012 Advanced Micro Devices, Inc.  \n"
+	"This software is provided 'as-is', without any express or implied warranty.\n"
+	"In no event will the authors be held liable for any damages arising from the use of this software.\n"
+	"Permission is granted to anyone to use this software for any purpose, \n"
+	"including commercial applications, and to alter it and redistribute it freely, \n"
+	"subject to the following restrictions:\n"
+	"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
+	"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
+	"3. This notice may not be removed or altered from any source distribution.\n"
+	"*/\n"
+	"//Originally written by Takahiro Harada\n"
+	"#ifndef B3_CONTACT4DATA_H\n"
+	"#define B3_CONTACT4DATA_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#define B3_FLOAT4_H\n"
+	"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
+	"#define B3_PLATFORM_DEFINITIONS_H\n"
+	"struct MyTest\n"
+	"{\n"
+	"	int bla;\n"
+	"};\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
+	"#define B3_LARGE_FLOAT 1e18f\n"
+	"#define B3_INFINITY 1e18f\n"
+	"#define b3Assert(a)\n"
+	"#define b3ConstArray(a) __global const a*\n"
+	"#define b3AtomicInc atomic_inc\n"
+	"#define b3AtomicAdd atomic_add\n"
+	"#define b3Fabs fabs\n"
+	"#define b3Sqrt native_sqrt\n"
+	"#define b3Sin native_sin\n"
+	"#define b3Cos native_cos\n"
+	"#define B3_STATIC\n"
+	"#endif\n"
+	"#endif\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"	typedef float4	b3Float4;\n"
+	"	#define b3Float4ConstArg const b3Float4\n"
+	"	#define b3MakeFloat4 (float4)\n"
+	"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return dot(a1, b1);\n"
+	"	}\n"
+	"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return cross(a1, b1);\n"
+	"	}\n"
+	"	#define b3MinFloat4 min\n"
+	"	#define b3MaxFloat4 max\n"
+	"	#define b3Normalized(a) normalize(a)\n"
+	"#endif \n"
+	"		\n"
+	"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
+	"{\n"
+	"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
+	"		return false;\n"
+	"	return true;\n"
+	"}\n"
+	"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
+	"{\n"
+	"    float maxDot = -B3_INFINITY;\n"
+	"    int i = 0;\n"
+	"    int ptIndex = -1;\n"
+	"    for( i = 0; i < vecLen; i++ )\n"
+	"    {\n"
+	"        float dot = b3Dot3F4(vecArray[i],vec);\n"
+	"            \n"
+	"        if( dot > maxDot )\n"
+	"        {\n"
+	"            maxDot = dot;\n"
+	"            ptIndex = i;\n"
+	"        }\n"
+	"    }\n"
+	"	b3Assert(ptIndex>=0);\n"
+	"    if (ptIndex<0)\n"
+	"	{\n"
+	"		ptIndex = 0;\n"
+	"	}\n"
+	"    *dotOut = maxDot;\n"
+	"    return ptIndex;\n"
+	"}\n"
+	"#endif //B3_FLOAT4_H\n"
+	"typedef  struct b3Contact4Data b3Contact4Data_t;\n"
+	"struct b3Contact4Data\n"
+	"{\n"
+	"	b3Float4	m_worldPosB[4];\n"
+	"//	b3Float4	m_localPosA[4];\n"
+	"//	b3Float4	m_localPosB[4];\n"
+	"	b3Float4	m_worldNormalOnB;	//	w: m_nPoints\n"
+	"	unsigned short  m_restituitionCoeffCmp;\n"
+	"	unsigned short  m_frictionCoeffCmp;\n"
+	"	int m_batchIdx;\n"
+	"	int m_bodyAPtrAndSignBit;//x:m_bodyAPtr, y:m_bodyBPtr\n"
+	"	int m_bodyBPtrAndSignBit;\n"
+	"	int	m_childIndexA;\n"
+	"	int	m_childIndexB;\n"
+	"	int m_unused1;\n"
+	"	int m_unused2;\n"
+	"};\n"
+	"inline int b3Contact4Data_getNumPoints(const struct b3Contact4Data* contact)\n"
+	"{\n"
+	"	return (int)contact->m_worldNormalOnB.w;\n"
+	"};\n"
+	"inline void b3Contact4Data_setNumPoints(struct b3Contact4Data* contact, int numPoints)\n"
+	"{\n"
+	"	contact->m_worldNormalOnB.w = (float)numPoints;\n"
+	"};\n"
+	"#endif //B3_CONTACT4DATA_H\n"
+	"#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_extended_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics : enable\n"
+	"#ifdef cl_ext_atomic_counters_32\n"
+	"#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
+	"#else\n"
+	"#define counter32_t volatile global int*\n"
+	"#endif\n"
+	"typedef unsigned int u32;\n"
+	"typedef unsigned short u16;\n"
+	"typedef unsigned char u8;\n"
+	"#define GET_GROUP_IDX get_group_id(0)\n"
+	"#define GET_LOCAL_IDX get_local_id(0)\n"
+	"#define GET_GLOBAL_IDX get_global_id(0)\n"
+	"#define GET_GROUP_SIZE get_local_size(0)\n"
+	"#define GET_NUM_GROUPS get_num_groups(0)\n"
+	"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
+	"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
+	"#define AtomInc(x) atom_inc(&(x))\n"
+	"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
+	"#define AppendInc(x, out) out = atomic_inc(x)\n"
+	"#define AtomAdd(x, value) atom_add(&(x), value)\n"
+	"#define AtomCmpxhg(x, cmp, value) atom_cmpxchg( &(x), cmp, value )\n"
+	"#define AtomXhg(x, value) atom_xchg ( &(x), value )\n"
+	"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
+	"#define make_float4 (float4)\n"
+	"#define make_float2 (float2)\n"
+	"#define make_uint4 (uint4)\n"
+	"#define make_int4 (int4)\n"
+	"#define make_uint2 (uint2)\n"
+	"#define make_int2 (int2)\n"
+	"#define max2 max\n"
+	"#define min2 min\n"
+	"///////////////////////////////////////\n"
+	"//	Vector\n"
+	"///////////////////////////////////////\n"
+	"__inline\n"
+	"float fastDiv(float numerator, float denominator)\n"
+	"{\n"
+	"	return native_divide(numerator, denominator);	\n"
+	"//	return numerator/denominator;	\n"
+	"}\n"
+	"__inline\n"
+	"float4 fastDiv4(float4 numerator, float4 denominator)\n"
+	"{\n"
+	"	return native_divide(numerator, denominator);	\n"
+	"}\n"
+	"__inline\n"
+	"float fastSqrtf(float f2)\n"
+	"{\n"
+	"	return native_sqrt(f2);\n"
+	"//	return sqrt(f2);\n"
+	"}\n"
+	"__inline\n"
+	"float fastRSqrt(float f2)\n"
+	"{\n"
+	"	return native_rsqrt(f2);\n"
+	"}\n"
+	"__inline\n"
+	"float fastLength4(float4 v)\n"
+	"{\n"
+	"	return fast_length(v);\n"
+	"}\n"
+	"__inline\n"
+	"float4 fastNormalize4(float4 v)\n"
+	"{\n"
+	"	return fast_normalize(v);\n"
+	"}\n"
+	"__inline\n"
+	"float sqrtf(float a)\n"
+	"{\n"
+	"//	return sqrt(a);\n"
+	"	return native_sqrt(a);\n"
+	"}\n"
+	"__inline\n"
+	"float4 cross3(float4 a, float4 b)\n"
+	"{\n"
+	"	return cross(a,b);\n"
+	"}\n"
+	"__inline\n"
+	"float dot3F4(float4 a, float4 b)\n"
+	"{\n"
+	"	float4 a1 = make_float4(a.xyz,0.f);\n"
+	"	float4 b1 = make_float4(b.xyz,0.f);\n"
+	"	return dot(a1, b1);\n"
+	"}\n"
+	"__inline\n"
+	"float length3(const float4 a)\n"
+	"{\n"
+	"	return sqrtf(dot3F4(a,a));\n"
+	"}\n"
+	"__inline\n"
+	"float dot4(const float4 a, const float4 b)\n"
+	"{\n"
+	"	return dot( a, b );\n"
+	"}\n"
+	"//	for height\n"
+	"__inline\n"
+	"float dot3w1(const float4 point, const float4 eqn)\n"
+	"{\n"
+	"	return dot3F4(point,eqn) + eqn.w;\n"
+	"}\n"
+	"__inline\n"
+	"float4 normalize3(const float4 a)\n"
+	"{\n"
+	"	float4 n = make_float4(a.x, a.y, a.z, 0.f);\n"
+	"	return fastNormalize4( n );\n"
+	"//	float length = sqrtf(dot3F4(a, a));\n"
+	"//	return 1.f/length * a;\n"
+	"}\n"
+	"__inline\n"
+	"float4 normalize4(const float4 a)\n"
+	"{\n"
+	"	float length = sqrtf(dot4(a, a));\n"
+	"	return 1.f/length * a;\n"
+	"}\n"
+	"__inline\n"
+	"float4 createEquation(const float4 a, const float4 b, const float4 c)\n"
+	"{\n"
+	"	float4 eqn;\n"
+	"	float4 ab = b-a;\n"
+	"	float4 ac = c-a;\n"
+	"	eqn = normalize3( cross3(ab, ac) );\n"
+	"	eqn.w = -dot3F4(eqn,a);\n"
+	"	return eqn;\n"
+	"}\n"
+	"///////////////////////////////////////\n"
+	"//	Matrix3x3\n"
+	"///////////////////////////////////////\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_row[3];\n"
+	"}Matrix3x3;\n"
+	"__inline\n"
+	"Matrix3x3 mtZero();\n"
+	"__inline\n"
+	"Matrix3x3 mtIdentity();\n"
+	"__inline\n"
+	"Matrix3x3 mtTranspose(Matrix3x3 m);\n"
+	"__inline\n"
+	"Matrix3x3 mtMul(Matrix3x3 a, Matrix3x3 b);\n"
+	"__inline\n"
+	"float4 mtMul1(Matrix3x3 a, float4 b);\n"
+	"__inline\n"
+	"float4 mtMul3(float4 a, Matrix3x3 b);\n"
+	"__inline\n"
+	"Matrix3x3 mtZero()\n"
+	"{\n"
+	"	Matrix3x3 m;\n"
+	"	m.m_row[0] = (float4)(0.f);\n"
+	"	m.m_row[1] = (float4)(0.f);\n"
+	"	m.m_row[2] = (float4)(0.f);\n"
+	"	return m;\n"
+	"}\n"
+	"__inline\n"
+	"Matrix3x3 mtIdentity()\n"
+	"{\n"
+	"	Matrix3x3 m;\n"
+	"	m.m_row[0] = (float4)(1,0,0,0);\n"
+	"	m.m_row[1] = (float4)(0,1,0,0);\n"
+	"	m.m_row[2] = (float4)(0,0,1,0);\n"
+	"	return m;\n"
+	"}\n"
+	"__inline\n"
+	"Matrix3x3 mtTranspose(Matrix3x3 m)\n"
+	"{\n"
+	"	Matrix3x3 out;\n"
+	"	out.m_row[0] = (float4)(m.m_row[0].x, m.m_row[1].x, m.m_row[2].x, 0.f);\n"
+	"	out.m_row[1] = (float4)(m.m_row[0].y, m.m_row[1].y, m.m_row[2].y, 0.f);\n"
+	"	out.m_row[2] = (float4)(m.m_row[0].z, m.m_row[1].z, m.m_row[2].z, 0.f);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"Matrix3x3 mtMul(Matrix3x3 a, Matrix3x3 b)\n"
+	"{\n"
+	"	Matrix3x3 transB;\n"
+	"	transB = mtTranspose( b );\n"
+	"	Matrix3x3 ans;\n"
+	"	//	why this doesn't run when 0ing in the for{}\n"
+	"	a.m_row[0].w = 0.f;\n"
+	"	a.m_row[1].w = 0.f;\n"
+	"	a.m_row[2].w = 0.f;\n"
+	"	for(int i=0; i<3; i++)\n"
+	"	{\n"
+	"//	a.m_row[i].w = 0.f;\n"
+	"		ans.m_row[i].x = dot3F4(a.m_row[i],transB.m_row[0]);\n"
+	"		ans.m_row[i].y = dot3F4(a.m_row[i],transB.m_row[1]);\n"
+	"		ans.m_row[i].z = dot3F4(a.m_row[i],transB.m_row[2]);\n"
+	"		ans.m_row[i].w = 0.f;\n"
+	"	}\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"float4 mtMul1(Matrix3x3 a, float4 b)\n"
+	"{\n"
+	"	float4 ans;\n"
+	"	ans.x = dot3F4( a.m_row[0], b );\n"
+	"	ans.y = dot3F4( a.m_row[1], b );\n"
+	"	ans.z = dot3F4( a.m_row[2], b );\n"
+	"	ans.w = 0.f;\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"float4 mtMul3(float4 a, Matrix3x3 b)\n"
+	"{\n"
+	"	float4 colx = make_float4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
+	"	float4 coly = make_float4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
+	"	float4 colz = make_float4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
+	"	float4 ans;\n"
+	"	ans.x = dot3F4( a, colx );\n"
+	"	ans.y = dot3F4( a, coly );\n"
+	"	ans.z = dot3F4( a, colz );\n"
+	"	return ans;\n"
+	"}\n"
+	"///////////////////////////////////////\n"
+	"//	Quaternion\n"
+	"///////////////////////////////////////\n"
+	"typedef float4 Quaternion;\n"
+	"__inline\n"
+	"Quaternion qtMul(Quaternion a, Quaternion b);\n"
+	"__inline\n"
+	"Quaternion qtNormalize(Quaternion in);\n"
+	"__inline\n"
+	"float4 qtRotate(Quaternion q, float4 vec);\n"
+	"__inline\n"
+	"Quaternion qtInvert(Quaternion q);\n"
+	"__inline\n"
+	"Quaternion qtMul(Quaternion a, Quaternion b)\n"
+	"{\n"
+	"	Quaternion ans;\n"
+	"	ans = cross3( a, b );\n"
+	"	ans += a.w*b+b.w*a;\n"
+	"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
+	"	ans.w = a.w*b.w - dot3F4(a, b);\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"Quaternion qtNormalize(Quaternion in)\n"
+	"{\n"
+	"	return fastNormalize4(in);\n"
+	"//	in /= length( in );\n"
+	"//	return in;\n"
+	"}\n"
+	"__inline\n"
+	"float4 qtRotate(Quaternion q, float4 vec)\n"
+	"{\n"
+	"	Quaternion qInv = qtInvert( q );\n"
+	"	float4 vcpy = vec;\n"
+	"	vcpy.w = 0.f;\n"
+	"	float4 out = qtMul(qtMul(q,vcpy),qInv);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"Quaternion qtInvert(Quaternion q)\n"
+	"{\n"
+	"	return (Quaternion)(-q.xyz, q.w);\n"
+	"}\n"
+	"__inline\n"
+	"float4 qtInvRotate(const Quaternion q, float4 vec)\n"
+	"{\n"
+	"	return qtRotate( qtInvert( q ), vec );\n"
+	"}\n"
+	"#define WG_SIZE 64\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_pos;\n"
+	"	Quaternion m_quat;\n"
+	"	float4 m_linVel;\n"
+	"	float4 m_angVel;\n"
+	"	u32 m_shapeIdx;\n"
+	"	float m_invMass;\n"
+	"	float m_restituitionCoeff;\n"
+	"	float m_frictionCoeff;\n"
+	"} Body;\n"
+	"typedef struct\n"
+	"{\n"
+	"	Matrix3x3 m_invInertia;\n"
+	"	Matrix3x3 m_initInvInertia;\n"
+	"} Shape;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_linear;\n"
+	"	float4 m_worldPos[4];\n"
+	"	float4 m_center;	\n"
+	"	float m_jacCoeffInv[4];\n"
+	"	float m_b[4];\n"
+	"	float m_appliedRambdaDt[4];\n"
+	"	float m_fJacCoeffInv[2];	\n"
+	"	float m_fAppliedRambdaDt[2];	\n"
+	"	u32 m_bodyA;\n"
+	"	u32 m_bodyB;\n"
+	"	int m_batchIdx;\n"
+	"	u32 m_paddings[1];\n"
+	"} Constraint4;\n"
+	"typedef struct\n"
+	"{\n"
+	"	int m_nConstraints;\n"
+	"	int m_start;\n"
+	"	int m_batchIdx;\n"
+	"	int m_nSplit;\n"
+	"//	int m_paddings[1];\n"
+	"} ConstBuffer;\n"
+	"typedef struct\n"
+	"{\n"
+	"	int m_solveFriction;\n"
+	"	int m_maxBatch;	//	long batch really kills the performance\n"
+	"	int m_batchIdx;\n"
+	"	int m_nSplit;\n"
+	"//	int m_paddings[1];\n"
+	"} ConstBufferBatchSolve;\n"
+	" \n"
+	"typedef struct \n"
+	"{\n"
+	"	int m_valInt0;\n"
+	"	int m_valInt1;\n"
+	"	int m_valInt2;\n"
+	"	int m_valInt3;\n"
+	"	float m_val0;\n"
+	"	float m_val1;\n"
+	"	float m_val2;\n"
+	"	float m_val3;\n"
+	"} SolverDebugInfo;\n"
+	"//	others\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"void ReorderContactKernel(__global struct b3Contact4Data* in, __global struct b3Contact4Data* out, __global int2* sortData, int4 cb )\n"
+	"{\n"
+	"	int nContacts = cb.x;\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"	if( gIdx < nContacts )\n"
+	"	{\n"
+	"		int srcIdx = sortData[gIdx].y;\n"
+	"		out[gIdx] = in[srcIdx];\n"
+	"	}\n"
+	"}\n"
+	"__kernel __attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"void SetDeterminismSortDataChildShapeB(__global struct b3Contact4Data* contactsIn, __global int2* sortDataOut, int nContacts)\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"	if( gIdx < nContacts )\n"
+	"	{\n"
+	"		int2 sd;\n"
+	"		sd.x = contactsIn[gIdx].m_childIndexB;\n"
+	"		sd.y = gIdx;\n"
+	"		sortDataOut[gIdx] = sd;\n"
+	"	}\n"
+	"}\n"
+	"__kernel __attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"void SetDeterminismSortDataChildShapeA(__global struct b3Contact4Data* contactsIn, __global int2* sortDataInOut, int nContacts)\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"	if( gIdx < nContacts )\n"
+	"	{\n"
+	"		int2 sdIn;\n"
+	"		sdIn = sortDataInOut[gIdx];\n"
+	"		int2 sdOut;\n"
+	"		sdOut.x = contactsIn[sdIn.y].m_childIndexA;\n"
+	"		sdOut.y = sdIn.y;\n"
+	"		sortDataInOut[gIdx] = sdOut;\n"
+	"	}\n"
+	"}\n"
+	"__kernel __attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"void SetDeterminismSortDataBodyA(__global struct b3Contact4Data* contactsIn, __global int2* sortDataInOut, int nContacts)\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"	if( gIdx < nContacts )\n"
+	"	{\n"
+	"		int2 sdIn;\n"
+	"		sdIn = sortDataInOut[gIdx];\n"
+	"		int2 sdOut;\n"
+	"		sdOut.x = contactsIn[sdIn.y].m_bodyAPtrAndSignBit;\n"
+	"		sdOut.y = sdIn.y;\n"
+	"		sortDataInOut[gIdx] = sdOut;\n"
+	"	}\n"
+	"}\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"void SetDeterminismSortDataBodyB(__global struct b3Contact4Data* contactsIn, __global int2* sortDataInOut, int nContacts)\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"	if( gIdx < nContacts )\n"
+	"	{\n"
+	"		int2 sdIn;\n"
+	"		sdIn = sortDataInOut[gIdx];\n"
+	"		int2 sdOut;\n"
+	"		sdOut.x = contactsIn[sdIn.y].m_bodyBPtrAndSignBit;\n"
+	"		sdOut.y = sdIn.y;\n"
+	"		sortDataInOut[gIdx] = sdOut;\n"
+	"	}\n"
+	"}\n"
+	"typedef struct\n"
+	"{\n"
+	"	int m_nContacts;\n"
+	"	int m_staticIdx;\n"
+	"	float m_scale;\n"
+	"	int m_nSplit;\n"
+	"} ConstBufferSSD;\n"
+	"__constant const int gridTable4x4[] = \n"
+	"{\n"
+	"    0,1,17,16,\n"
+	"	1,2,18,19,\n"
+	"	17,18,32,3,\n"
+	"	16,19,3,34\n"
+	"};\n"
+	"__constant const int gridTable8x8[] = \n"
+	"{\n"
+	"	  0,  2,  3, 16, 17, 18, 19,  1,\n"
+	"	 66, 64, 80, 67, 82, 81, 65, 83,\n"
+	"	131,144,128,130,147,129,145,146,\n"
+	"	208,195,194,192,193,211,210,209,\n"
+	"	 21, 22, 23,  5,  4,  6,  7, 20,\n"
+	"	 86, 85, 69, 87, 70, 68, 84, 71,\n"
+	"	151,133,149,150,135,148,132,134,\n"
+	"	197,27,214,213,212,199,198,196\n"
+	"	\n"
+	"};\n"
+	"#define USE_SPATIAL_BATCHING 1\n"
+	"#define USE_4x4_GRID 1\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"void SetSortDataKernel(__global struct b3Contact4Data* gContact, __global Body* gBodies, __global int2* gSortDataOut, \n"
+	"int nContacts,float scale,int4 nSplit,int staticIdx)\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"	\n"
+	"	if( gIdx < nContacts )\n"
+	"	{\n"
+	"		int aPtrAndSignBit  = gContact[gIdx].m_bodyAPtrAndSignBit;\n"
+	"		int bPtrAndSignBit  = gContact[gIdx].m_bodyBPtrAndSignBit;\n"
+	"		int aIdx = abs(aPtrAndSignBit );\n"
+	"		int bIdx = abs(bPtrAndSignBit);\n"
+	"		bool aStatic = (aPtrAndSignBit<0) ||(aPtrAndSignBit==staticIdx);\n"
+	"		bool bStatic = (bPtrAndSignBit<0) ||(bPtrAndSignBit==staticIdx);\n"
+	"#if USE_SPATIAL_BATCHING		\n"
+	"		int idx = (aStatic)? bIdx: aIdx;\n"
+	"		float4 p = gBodies[idx].m_pos;\n"
+	"		int xIdx = (int)((p.x-((p.x<0.f)?1.f:0.f))*scale) & (nSplit.x-1);\n"
+	"		int yIdx = (int)((p.y-((p.y<0.f)?1.f:0.f))*scale) & (nSplit.y-1);\n"
+	"		int zIdx = (int)((p.z-((p.z<0.f)?1.f:0.f))*scale) & (nSplit.z-1);\n"
+	"		int newIndex = (xIdx+yIdx*nSplit.x+zIdx*nSplit.x*nSplit.y);\n"
+	"		\n"
+	"#else//USE_SPATIAL_BATCHING\n"
+	"	#if USE_4x4_GRID\n"
+	"		int aa = aIdx&3;\n"
+	"		int bb = bIdx&3;\n"
+	"		if (aStatic)\n"
+	"			aa = bb;\n"
+	"		if (bStatic)\n"
+	"			bb = aa;\n"
+	"		int gridIndex = aa + bb*4;\n"
+	"		int newIndex = gridTable4x4[gridIndex];\n"
+	"	#else//USE_4x4_GRID\n"
+	"		int aa = aIdx&7;\n"
+	"		int bb = bIdx&7;\n"
+	"		if (aStatic)\n"
+	"			aa = bb;\n"
+	"		if (bStatic)\n"
+	"			bb = aa;\n"
+	"		int gridIndex = aa + bb*8;\n"
+	"		int newIndex = gridTable8x8[gridIndex];\n"
+	"	#endif//USE_4x4_GRID\n"
+	"#endif//USE_SPATIAL_BATCHING\n"
+	"		gSortDataOut[gIdx].x = newIndex;\n"
+	"		gSortDataOut[gIdx].y = gIdx;\n"
+	"	}\n"
+	"	else\n"
+	"	{\n"
+	"		gSortDataOut[gIdx].x = 0xffffffff;\n"
+	"	}\n"
+	"}\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"void CopyConstraintKernel(__global struct b3Contact4Data* gIn, __global struct b3Contact4Data* gOut, int4 cb )\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"	if( gIdx < cb.x )\n"
+	"	{\n"
+	"		gOut[gIdx] = gIn[gIdx];\n"
+	"	}\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/solverUtils.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/solverUtils.h
index c0173ad9f42..f4d98d99415 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/solverUtils.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/solverUtils.h
@@ -1,909 +1,908 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* solverUtilsCL= \
-"/*\n"
-"Copyright (c) 2013 Advanced Micro Devices, Inc.  \n"
-"This software is provided 'as-is', without any express or implied warranty.\n"
-"In no event will the authors be held liable for any damages arising from the use of this software.\n"
-"Permission is granted to anyone to use this software for any purpose, \n"
-"including commercial applications, and to alter it and redistribute it freely, \n"
-"subject to the following restrictions:\n"
-"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
-"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
-"3. This notice may not be removed or altered from any source distribution.\n"
-"*/\n"
-"//Originally written by Erwin Coumans\n"
-"#ifndef B3_CONTACT4DATA_H\n"
-"#define B3_CONTACT4DATA_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#define B3_FLOAT4_H\n"
-"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
-"#define B3_PLATFORM_DEFINITIONS_H\n"
-"struct MyTest\n"
-"{\n"
-"	int bla;\n"
-"};\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
-"#define B3_LARGE_FLOAT 1e18f\n"
-"#define B3_INFINITY 1e18f\n"
-"#define b3Assert(a)\n"
-"#define b3ConstArray(a) __global const a*\n"
-"#define b3AtomicInc atomic_inc\n"
-"#define b3AtomicAdd atomic_add\n"
-"#define b3Fabs fabs\n"
-"#define b3Sqrt native_sqrt\n"
-"#define b3Sin native_sin\n"
-"#define b3Cos native_cos\n"
-"#define B3_STATIC\n"
-"#endif\n"
-"#endif\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"	typedef float4	b3Float4;\n"
-"	#define b3Float4ConstArg const b3Float4\n"
-"	#define b3MakeFloat4 (float4)\n"
-"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return dot(a1, b1);\n"
-"	}\n"
-"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return cross(a1, b1);\n"
-"	}\n"
-"	#define b3MinFloat4 min\n"
-"	#define b3MaxFloat4 max\n"
-"	#define b3Normalized(a) normalize(a)\n"
-"#endif \n"
-"		\n"
-"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
-"{\n"
-"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
-"		return false;\n"
-"	return true;\n"
-"}\n"
-"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
-"{\n"
-"    float maxDot = -B3_INFINITY;\n"
-"    int i = 0;\n"
-"    int ptIndex = -1;\n"
-"    for( i = 0; i < vecLen; i++ )\n"
-"    {\n"
-"        float dot = b3Dot3F4(vecArray[i],vec);\n"
-"            \n"
-"        if( dot > maxDot )\n"
-"        {\n"
-"            maxDot = dot;\n"
-"            ptIndex = i;\n"
-"        }\n"
-"    }\n"
-"	b3Assert(ptIndex>=0);\n"
-"    if (ptIndex<0)\n"
-"	{\n"
-"		ptIndex = 0;\n"
-"	}\n"
-"    *dotOut = maxDot;\n"
-"    return ptIndex;\n"
-"}\n"
-"#endif //B3_FLOAT4_H\n"
-"typedef  struct b3Contact4Data b3Contact4Data_t;\n"
-"struct b3Contact4Data\n"
-"{\n"
-"	b3Float4	m_worldPosB[4];\n"
-"//	b3Float4	m_localPosA[4];\n"
-"//	b3Float4	m_localPosB[4];\n"
-"	b3Float4	m_worldNormalOnB;	//	w: m_nPoints\n"
-"	unsigned short  m_restituitionCoeffCmp;\n"
-"	unsigned short  m_frictionCoeffCmp;\n"
-"	int m_batchIdx;\n"
-"	int m_bodyAPtrAndSignBit;//x:m_bodyAPtr, y:m_bodyBPtr\n"
-"	int m_bodyBPtrAndSignBit;\n"
-"	int	m_childIndexA;\n"
-"	int	m_childIndexB;\n"
-"	int m_unused1;\n"
-"	int m_unused2;\n"
-"};\n"
-"inline int b3Contact4Data_getNumPoints(const struct b3Contact4Data* contact)\n"
-"{\n"
-"	return (int)contact->m_worldNormalOnB.w;\n"
-"};\n"
-"inline void b3Contact4Data_setNumPoints(struct b3Contact4Data* contact, int numPoints)\n"
-"{\n"
-"	contact->m_worldNormalOnB.w = (float)numPoints;\n"
-"};\n"
-"#endif //B3_CONTACT4DATA_H\n"
-"#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_local_int32_extended_atomics : enable\n"
-"#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics : enable\n"
-"#ifdef cl_ext_atomic_counters_32\n"
-"#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
-"#else\n"
-"#define counter32_t volatile global int*\n"
-"#endif\n"
-"typedef unsigned int u32;\n"
-"typedef unsigned short u16;\n"
-"typedef unsigned char u8;\n"
-"#define GET_GROUP_IDX get_group_id(0)\n"
-"#define GET_LOCAL_IDX get_local_id(0)\n"
-"#define GET_GLOBAL_IDX get_global_id(0)\n"
-"#define GET_GROUP_SIZE get_local_size(0)\n"
-"#define GET_NUM_GROUPS get_num_groups(0)\n"
-"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
-"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
-"#define AtomInc(x) atom_inc(&(x))\n"
-"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
-"#define AppendInc(x, out) out = atomic_inc(x)\n"
-"#define AtomAdd(x, value) atom_add(&(x), value)\n"
-"#define AtomCmpxhg(x, cmp, value) atom_cmpxchg( &(x), cmp, value )\n"
-"#define AtomXhg(x, value) atom_xchg ( &(x), value )\n"
-"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
-"#define make_float4 (float4)\n"
-"#define make_float2 (float2)\n"
-"#define make_uint4 (uint4)\n"
-"#define make_int4 (int4)\n"
-"#define make_uint2 (uint2)\n"
-"#define make_int2 (int2)\n"
-"#define max2 max\n"
-"#define min2 min\n"
-"///////////////////////////////////////\n"
-"//	Vector\n"
-"///////////////////////////////////////\n"
-"__inline\n"
-"float fastDiv(float numerator, float denominator)\n"
-"{\n"
-"	return native_divide(numerator, denominator);	\n"
-"//	return numerator/denominator;	\n"
-"}\n"
-"__inline\n"
-"float4 fastDiv4(float4 numerator, float4 denominator)\n"
-"{\n"
-"	return native_divide(numerator, denominator);	\n"
-"}\n"
-"__inline\n"
-"float fastSqrtf(float f2)\n"
-"{\n"
-"	return native_sqrt(f2);\n"
-"//	return sqrt(f2);\n"
-"}\n"
-"__inline\n"
-"float fastRSqrt(float f2)\n"
-"{\n"
-"	return native_rsqrt(f2);\n"
-"}\n"
-"__inline\n"
-"float fastLength4(float4 v)\n"
-"{\n"
-"	return fast_length(v);\n"
-"}\n"
-"__inline\n"
-"float4 fastNormalize4(float4 v)\n"
-"{\n"
-"	return fast_normalize(v);\n"
-"}\n"
-"__inline\n"
-"float sqrtf(float a)\n"
-"{\n"
-"//	return sqrt(a);\n"
-"	return native_sqrt(a);\n"
-"}\n"
-"__inline\n"
-"float4 cross3(float4 a1, float4 b1)\n"
-"{\n"
-"	float4 	a=make_float4(a1.xyz,0.f);\n"
-"	float4 	b=make_float4(b1.xyz,0.f);\n"
-"	//float4 	a=a1;\n"
-"	//float4 	b=b1;\n"
-"	return cross(a,b);\n"
-"}\n"
-"__inline\n"
-"float dot3F4(float4 a, float4 b)\n"
-"{\n"
-"	float4 a1 = make_float4(a.xyz,0.f);\n"
-"	float4 b1 = make_float4(b.xyz,0.f);\n"
-"	return dot(a1, b1);\n"
-"}\n"
-"__inline\n"
-"float length3(const float4 a)\n"
-"{\n"
-"	return sqrtf(dot3F4(a,a));\n"
-"}\n"
-"__inline\n"
-"float dot4(const float4 a, const float4 b)\n"
-"{\n"
-"	return dot( a, b );\n"
-"}\n"
-"//	for height\n"
-"__inline\n"
-"float dot3w1(const float4 point, const float4 eqn)\n"
-"{\n"
-"	return dot3F4(point,eqn) + eqn.w;\n"
-"}\n"
-"__inline\n"
-"float4 normalize3(const float4 a)\n"
-"{\n"
-"	float4 n = make_float4(a.x, a.y, a.z, 0.f);\n"
-"	return fastNormalize4( n );\n"
-"//	float length = sqrtf(dot3F4(a, a));\n"
-"//	return 1.f/length * a;\n"
-"}\n"
-"__inline\n"
-"float4 normalize4(const float4 a)\n"
-"{\n"
-"	float length = sqrtf(dot4(a, a));\n"
-"	return 1.f/length * a;\n"
-"}\n"
-"__inline\n"
-"float4 createEquation(const float4 a, const float4 b, const float4 c)\n"
-"{\n"
-"	float4 eqn;\n"
-"	float4 ab = b-a;\n"
-"	float4 ac = c-a;\n"
-"	eqn = normalize3( cross3(ab, ac) );\n"
-"	eqn.w = -dot3F4(eqn,a);\n"
-"	return eqn;\n"
-"}\n"
-"///////////////////////////////////////\n"
-"//	Matrix3x3\n"
-"///////////////////////////////////////\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_row[3];\n"
-"}Matrix3x3;\n"
-"__inline\n"
-"Matrix3x3 mtZero();\n"
-"__inline\n"
-"Matrix3x3 mtIdentity();\n"
-"__inline\n"
-"Matrix3x3 mtTranspose(Matrix3x3 m);\n"
-"__inline\n"
-"Matrix3x3 mtMul(Matrix3x3 a, Matrix3x3 b);\n"
-"__inline\n"
-"float4 mtMul1(Matrix3x3 a, float4 b);\n"
-"__inline\n"
-"float4 mtMul3(float4 a, Matrix3x3 b);\n"
-"__inline\n"
-"Matrix3x3 mtZero()\n"
-"{\n"
-"	Matrix3x3 m;\n"
-"	m.m_row[0] = (float4)(0.f);\n"
-"	m.m_row[1] = (float4)(0.f);\n"
-"	m.m_row[2] = (float4)(0.f);\n"
-"	return m;\n"
-"}\n"
-"__inline\n"
-"Matrix3x3 mtIdentity()\n"
-"{\n"
-"	Matrix3x3 m;\n"
-"	m.m_row[0] = (float4)(1,0,0,0);\n"
-"	m.m_row[1] = (float4)(0,1,0,0);\n"
-"	m.m_row[2] = (float4)(0,0,1,0);\n"
-"	return m;\n"
-"}\n"
-"__inline\n"
-"Matrix3x3 mtTranspose(Matrix3x3 m)\n"
-"{\n"
-"	Matrix3x3 out;\n"
-"	out.m_row[0] = (float4)(m.m_row[0].x, m.m_row[1].x, m.m_row[2].x, 0.f);\n"
-"	out.m_row[1] = (float4)(m.m_row[0].y, m.m_row[1].y, m.m_row[2].y, 0.f);\n"
-"	out.m_row[2] = (float4)(m.m_row[0].z, m.m_row[1].z, m.m_row[2].z, 0.f);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"Matrix3x3 mtMul(Matrix3x3 a, Matrix3x3 b)\n"
-"{\n"
-"	Matrix3x3 transB;\n"
-"	transB = mtTranspose( b );\n"
-"	Matrix3x3 ans;\n"
-"	//	why this doesn't run when 0ing in the for{}\n"
-"	a.m_row[0].w = 0.f;\n"
-"	a.m_row[1].w = 0.f;\n"
-"	a.m_row[2].w = 0.f;\n"
-"	for(int i=0; i<3; i++)\n"
-"	{\n"
-"//	a.m_row[i].w = 0.f;\n"
-"		ans.m_row[i].x = dot3F4(a.m_row[i],transB.m_row[0]);\n"
-"		ans.m_row[i].y = dot3F4(a.m_row[i],transB.m_row[1]);\n"
-"		ans.m_row[i].z = dot3F4(a.m_row[i],transB.m_row[2]);\n"
-"		ans.m_row[i].w = 0.f;\n"
-"	}\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"float4 mtMul1(Matrix3x3 a, float4 b)\n"
-"{\n"
-"	float4 ans;\n"
-"	ans.x = dot3F4( a.m_row[0], b );\n"
-"	ans.y = dot3F4( a.m_row[1], b );\n"
-"	ans.z = dot3F4( a.m_row[2], b );\n"
-"	ans.w = 0.f;\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"float4 mtMul3(float4 a, Matrix3x3 b)\n"
-"{\n"
-"	float4 colx = make_float4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
-"	float4 coly = make_float4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
-"	float4 colz = make_float4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
-"	float4 ans;\n"
-"	ans.x = dot3F4( a, colx );\n"
-"	ans.y = dot3F4( a, coly );\n"
-"	ans.z = dot3F4( a, colz );\n"
-"	return ans;\n"
-"}\n"
-"///////////////////////////////////////\n"
-"//	Quaternion\n"
-"///////////////////////////////////////\n"
-"typedef float4 Quaternion;\n"
-"__inline\n"
-"Quaternion qtMul(Quaternion a, Quaternion b);\n"
-"__inline\n"
-"Quaternion qtNormalize(Quaternion in);\n"
-"__inline\n"
-"float4 qtRotate(Quaternion q, float4 vec);\n"
-"__inline\n"
-"Quaternion qtInvert(Quaternion q);\n"
-"__inline\n"
-"Quaternion qtMul(Quaternion a, Quaternion b)\n"
-"{\n"
-"	Quaternion ans;\n"
-"	ans = cross3( a, b );\n"
-"	ans += a.w*b+b.w*a;\n"
-"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
-"	ans.w = a.w*b.w - dot3F4(a, b);\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"Quaternion qtNormalize(Quaternion in)\n"
-"{\n"
-"	return fastNormalize4(in);\n"
-"//	in /= length( in );\n"
-"//	return in;\n"
-"}\n"
-"__inline\n"
-"float4 qtRotate(Quaternion q, float4 vec)\n"
-"{\n"
-"	Quaternion qInv = qtInvert( q );\n"
-"	float4 vcpy = vec;\n"
-"	vcpy.w = 0.f;\n"
-"	float4 out = qtMul(qtMul(q,vcpy),qInv);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"Quaternion qtInvert(Quaternion q)\n"
-"{\n"
-"	return (Quaternion)(-q.xyz, q.w);\n"
-"}\n"
-"__inline\n"
-"float4 qtInvRotate(const Quaternion q, float4 vec)\n"
-"{\n"
-"	return qtRotate( qtInvert( q ), vec );\n"
-"}\n"
-"#define WG_SIZE 64\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_pos;\n"
-"	Quaternion m_quat;\n"
-"	float4 m_linVel;\n"
-"	float4 m_angVel;\n"
-"	u32 m_shapeIdx;\n"
-"	float m_invMass;\n"
-"	float m_restituitionCoeff;\n"
-"	float m_frictionCoeff;\n"
-"} Body;\n"
-"typedef struct\n"
-"{\n"
-"	Matrix3x3 m_invInertia;\n"
-"	Matrix3x3 m_initInvInertia;\n"
-"} Shape;\n"
-"typedef struct\n"
-"{\n"
-"	float4 m_linear;\n"
-"	float4 m_worldPos[4];\n"
-"	float4 m_center;	\n"
-"	float m_jacCoeffInv[4];\n"
-"	float m_b[4];\n"
-"	float m_appliedRambdaDt[4];\n"
-"	float m_fJacCoeffInv[2];	\n"
-"	float m_fAppliedRambdaDt[2];	\n"
-"	u32 m_bodyA;\n"
-"	u32 m_bodyB;\n"
-"	int m_batchIdx;\n"
-"	u32 m_paddings;\n"
-"} Constraint4;\n"
-"__kernel void CountBodiesKernel(__global struct b3Contact4Data* manifoldPtr, __global unsigned int* bodyCount, __global int2* contactConstraintOffsets, int numContactManifolds, int fixedBodyIndex)\n"
-"{\n"
-"	int i = GET_GLOBAL_IDX;\n"
-"	\n"
-"	if( i < numContactManifolds)\n"
-"	{\n"
-"		int pa = manifoldPtr[i].m_bodyAPtrAndSignBit;\n"
-"		bool isFixedA = (pa <0) || (pa == fixedBodyIndex);\n"
-"		int bodyIndexA = abs(pa);\n"
-"		if (!isFixedA)\n"
-"		{\n"
-"			 AtomInc1(bodyCount[bodyIndexA],contactConstraintOffsets[i].x);\n"
-"		}\n"
-"		barrier(CLK_GLOBAL_MEM_FENCE);\n"
-"		int pb = manifoldPtr[i].m_bodyBPtrAndSignBit;\n"
-"		bool isFixedB = (pb <0) || (pb == fixedBodyIndex);\n"
-"		int bodyIndexB = abs(pb);\n"
-"		if (!isFixedB)\n"
-"		{\n"
-"			AtomInc1(bodyCount[bodyIndexB],contactConstraintOffsets[i].y);\n"
-"		} \n"
-"	}\n"
-"}\n"
-"__kernel void ClearVelocitiesKernel(__global float4* linearVelocities,__global float4* angularVelocities, int numSplitBodies)\n"
-"{\n"
-"	int i = GET_GLOBAL_IDX;\n"
-"	\n"
-"	if( i < numSplitBodies)\n"
-"	{\n"
-"		linearVelocities[i] = make_float4(0);\n"
-"		angularVelocities[i] = make_float4(0);\n"
-"	}\n"
-"}\n"
-"__kernel void AverageVelocitiesKernel(__global Body* gBodies,__global int* offsetSplitBodies,__global const unsigned int* bodyCount,\n"
-"__global float4* deltaLinearVelocities, __global float4* deltaAngularVelocities, int numBodies)\n"
-"{\n"
-"	int i = GET_GLOBAL_IDX;\n"
-"	if (i<numBodies)\n"
-"	{\n"
-"		if (gBodies[i].m_invMass)\n"
-"		{\n"
-"			int bodyOffset = offsetSplitBodies[i];\n"
-"			int count = bodyCount[i];\n"
-"			float factor = 1.f/((float)count);\n"
-"			float4 averageLinVel = make_float4(0.f);\n"
-"			float4 averageAngVel = make_float4(0.f);\n"
-"			\n"
-"			for (int j=0;j<count;j++)\n"
-"			{\n"
-"				averageLinVel += deltaLinearVelocities[bodyOffset+j]*factor;\n"
-"				averageAngVel += deltaAngularVelocities[bodyOffset+j]*factor;\n"
-"			}\n"
-"			\n"
-"			for (int j=0;j<count;j++)\n"
-"			{\n"
-"				deltaLinearVelocities[bodyOffset+j] = averageLinVel;\n"
-"				deltaAngularVelocities[bodyOffset+j] = averageAngVel;\n"
-"			}\n"
-"			\n"
-"		}//bodies[i].m_invMass\n"
-"	}//i<numBodies\n"
-"}\n"
-"void setLinearAndAngular( float4 n, float4 r0, float4 r1, float4* linear, float4* angular0, float4* angular1)\n"
-"{\n"
-"	*linear = make_float4(n.xyz,0.f);\n"
-"	*angular0 = cross3(r0, n);\n"
-"	*angular1 = -cross3(r1, n);\n"
-"}\n"
-"float calcRelVel( float4 l0, float4 l1, float4 a0, float4 a1, float4 linVel0, float4 angVel0, float4 linVel1, float4 angVel1 )\n"
-"{\n"
-"	return dot3F4(l0, linVel0) + dot3F4(a0, angVel0) + dot3F4(l1, linVel1) + dot3F4(a1, angVel1);\n"
-"}\n"
-"float calcJacCoeff(const float4 linear0, const float4 linear1, const float4 angular0, const float4 angular1,\n"
-"					float invMass0, const Matrix3x3* invInertia0, float invMass1, const Matrix3x3* invInertia1, float countA, float countB)\n"
-"{\n"
-"	//	linear0,1 are normlized\n"
-"	float jmj0 = invMass0;//dot3F4(linear0, linear0)*invMass0;\n"
-"	float jmj1 = dot3F4(mtMul3(angular0,*invInertia0), angular0);\n"
-"	float jmj2 = invMass1;//dot3F4(linear1, linear1)*invMass1;\n"
-"	float jmj3 = dot3F4(mtMul3(angular1,*invInertia1), angular1);\n"
-"	return -1.f/((jmj0+jmj1)*countA+(jmj2+jmj3)*countB);\n"
-"}\n"
-"void btPlaneSpace1 (float4 n, float4* p, float4* q);\n"
-" void btPlaneSpace1 (float4 n, float4* p, float4* q)\n"
-"{\n"
-"  if (fabs(n.z) > 0.70710678f) {\n"
-"    // choose p in y-z plane\n"
-"    float a = n.y*n.y + n.z*n.z;\n"
-"    float k = 1.f/sqrt(a);\n"
-"    p[0].x = 0;\n"
-"	p[0].y = -n.z*k;\n"
-"	p[0].z = n.y*k;\n"
-"    // set q = n x p\n"
-"    q[0].x = a*k;\n"
-"	q[0].y = -n.x*p[0].z;\n"
-"	q[0].z = n.x*p[0].y;\n"
-"  }\n"
-"  else {\n"
-"    // choose p in x-y plane\n"
-"    float a = n.x*n.x + n.y*n.y;\n"
-"    float k = 1.f/sqrt(a);\n"
-"    p[0].x = -n.y*k;\n"
-"	p[0].y = n.x*k;\n"
-"	p[0].z = 0;\n"
-"    // set q = n x p\n"
-"    q[0].x = -n.z*p[0].y;\n"
-"	q[0].y = n.z*p[0].x;\n"
-"	q[0].z = a*k;\n"
-"  }\n"
-"}\n"
-"void solveContact(__global Constraint4* cs,\n"
-"			float4 posA, float4* linVelA, float4* angVelA, float invMassA, Matrix3x3 invInertiaA,\n"
-"			float4 posB, float4* linVelB, float4* angVelB, float invMassB, Matrix3x3 invInertiaB,\n"
-"			float4* dLinVelA, float4* dAngVelA, float4* dLinVelB, float4* dAngVelB)\n"
-"{\n"
-"	float minRambdaDt = 0;\n"
-"	float maxRambdaDt = FLT_MAX;\n"
-"	for(int ic=0; ic<4; ic++)\n"
-"	{\n"
-"		if( cs->m_jacCoeffInv[ic] == 0.f ) continue;\n"
-"		float4 angular0, angular1, linear;\n"
-"		float4 r0 = cs->m_worldPos[ic] - posA;\n"
-"		float4 r1 = cs->m_worldPos[ic] - posB;\n"
-"		setLinearAndAngular( cs->m_linear, r0, r1, &linear, &angular0, &angular1 );\n"
-"	\n"
-"		float rambdaDt = calcRelVel( cs->m_linear, -cs->m_linear, angular0, angular1, \n"
-"			*linVelA+*dLinVelA, *angVelA+*dAngVelA, *linVelB+*dLinVelB, *angVelB+*dAngVelB ) + cs->m_b[ic];\n"
-"		rambdaDt *= cs->m_jacCoeffInv[ic];\n"
-"		\n"
-"		{\n"
-"			float prevSum = cs->m_appliedRambdaDt[ic];\n"
-"			float updated = prevSum;\n"
-"			updated += rambdaDt;\n"
-"			updated = max2( updated, minRambdaDt );\n"
-"			updated = min2( updated, maxRambdaDt );\n"
-"			rambdaDt = updated - prevSum;\n"
-"			cs->m_appliedRambdaDt[ic] = updated;\n"
-"		}\n"
-"			\n"
-"		float4 linImp0 = invMassA*linear*rambdaDt;\n"
-"		float4 linImp1 = invMassB*(-linear)*rambdaDt;\n"
-"		float4 angImp0 = mtMul1(invInertiaA, angular0)*rambdaDt;\n"
-"		float4 angImp1 = mtMul1(invInertiaB, angular1)*rambdaDt;\n"
-"		\n"
-"		if (invMassA)\n"
-"		{\n"
-"			*dLinVelA += linImp0;\n"
-"			*dAngVelA += angImp0;\n"
-"		}\n"
-"		if (invMassB)\n"
-"		{\n"
-"			*dLinVelB += linImp1;\n"
-"			*dAngVelB += angImp1;\n"
-"		}\n"
-"	}\n"
-"}\n"
-"//	solveContactConstraint( gBodies, gShapes, &gConstraints[i] ,contactConstraintOffsets,offsetSplitBodies, deltaLinearVelocities, deltaAngularVelocities);\n"
-"void solveContactConstraint(__global Body* gBodies, __global Shape* gShapes, __global Constraint4* ldsCs, \n"
-"__global int2* contactConstraintOffsets,__global unsigned int* offsetSplitBodies,\n"
-"__global float4* deltaLinearVelocities, __global float4* deltaAngularVelocities)\n"
-"{\n"
-"	//float frictionCoeff = ldsCs[0].m_linear.w;\n"
-"	int aIdx = ldsCs[0].m_bodyA;\n"
-"	int bIdx = ldsCs[0].m_bodyB;\n"
-"	float4 posA = gBodies[aIdx].m_pos;\n"
-"	float4 linVelA = gBodies[aIdx].m_linVel;\n"
-"	float4 angVelA = gBodies[aIdx].m_angVel;\n"
-"	float invMassA = gBodies[aIdx].m_invMass;\n"
-"	Matrix3x3 invInertiaA = gShapes[aIdx].m_invInertia;\n"
-"	float4 posB = gBodies[bIdx].m_pos;\n"
-"	float4 linVelB = gBodies[bIdx].m_linVel;\n"
-"	float4 angVelB = gBodies[bIdx].m_angVel;\n"
-"	float invMassB = gBodies[bIdx].m_invMass;\n"
-"	Matrix3x3 invInertiaB = gShapes[bIdx].m_invInertia;\n"
-"			\n"
-"	float4 dLinVelA = make_float4(0,0,0,0);\n"
-"	float4 dAngVelA = make_float4(0,0,0,0);\n"
-"	float4 dLinVelB = make_float4(0,0,0,0);\n"
-"	float4 dAngVelB = make_float4(0,0,0,0);\n"
-"			\n"
-"	int bodyOffsetA = offsetSplitBodies[aIdx];\n"
-"	int constraintOffsetA = contactConstraintOffsets[0].x;\n"
-"	int splitIndexA = bodyOffsetA+constraintOffsetA;\n"
-"	\n"
-"	if (invMassA)\n"
-"	{\n"
-"		dLinVelA = deltaLinearVelocities[splitIndexA];\n"
-"		dAngVelA = deltaAngularVelocities[splitIndexA];\n"
-"	}\n"
-"	int bodyOffsetB = offsetSplitBodies[bIdx];\n"
-"	int constraintOffsetB = contactConstraintOffsets[0].y;\n"
-"	int splitIndexB= bodyOffsetB+constraintOffsetB;\n"
-"	if (invMassB)\n"
-"	{\n"
-"		dLinVelB = deltaLinearVelocities[splitIndexB];\n"
-"		dAngVelB = deltaAngularVelocities[splitIndexB];\n"
-"	}\n"
-"	solveContact( ldsCs, posA, &linVelA, &angVelA, invMassA, invInertiaA,\n"
-"			posB, &linVelB, &angVelB, invMassB, invInertiaB ,&dLinVelA, &dAngVelA, &dLinVelB, &dAngVelB);\n"
-"	if (invMassA)\n"
-"	{\n"
-"		deltaLinearVelocities[splitIndexA] = dLinVelA;\n"
-"		deltaAngularVelocities[splitIndexA] = dAngVelA;\n"
-"	} \n"
-"	if (invMassB)\n"
-"	{\n"
-"		deltaLinearVelocities[splitIndexB] = dLinVelB;\n"
-"		deltaAngularVelocities[splitIndexB] = dAngVelB;\n"
-"	}\n"
-"}\n"
-"__kernel void SolveContactJacobiKernel(__global Constraint4* gConstraints, __global Body* gBodies, __global Shape* gShapes ,\n"
-"__global int2* contactConstraintOffsets,__global unsigned int* offsetSplitBodies,__global float4* deltaLinearVelocities, __global float4* deltaAngularVelocities,\n"
-"float deltaTime, float positionDrift, float positionConstraintCoeff, int fixedBodyIndex, int numManifolds\n"
-")\n"
-"{\n"
-"	int i = GET_GLOBAL_IDX;\n"
-"	if (i<numManifolds)\n"
-"	{\n"
-"		solveContactConstraint( gBodies, gShapes, &gConstraints[i] ,&contactConstraintOffsets[i],offsetSplitBodies, deltaLinearVelocities, deltaAngularVelocities);\n"
-"	}\n"
-"}\n"
-"void solveFrictionConstraint(__global Body* gBodies, __global Shape* gShapes, __global Constraint4* ldsCs,\n"
-"							__global int2* contactConstraintOffsets,__global unsigned int* offsetSplitBodies,\n"
-"							__global float4* deltaLinearVelocities, __global float4* deltaAngularVelocities)\n"
-"{\n"
-"	float frictionCoeff = 0.7f;//ldsCs[0].m_linear.w;\n"
-"	int aIdx = ldsCs[0].m_bodyA;\n"
-"	int bIdx = ldsCs[0].m_bodyB;\n"
-"	float4 posA = gBodies[aIdx].m_pos;\n"
-"	float4 linVelA = gBodies[aIdx].m_linVel;\n"
-"	float4 angVelA = gBodies[aIdx].m_angVel;\n"
-"	float invMassA = gBodies[aIdx].m_invMass;\n"
-"	Matrix3x3 invInertiaA = gShapes[aIdx].m_invInertia;\n"
-"	float4 posB = gBodies[bIdx].m_pos;\n"
-"	float4 linVelB = gBodies[bIdx].m_linVel;\n"
-"	float4 angVelB = gBodies[bIdx].m_angVel;\n"
-"	float invMassB = gBodies[bIdx].m_invMass;\n"
-"	Matrix3x3 invInertiaB = gShapes[bIdx].m_invInertia;\n"
-"	\n"
-"	float4 dLinVelA = make_float4(0,0,0,0);\n"
-"	float4 dAngVelA = make_float4(0,0,0,0);\n"
-"	float4 dLinVelB = make_float4(0,0,0,0);\n"
-"	float4 dAngVelB = make_float4(0,0,0,0);\n"
-"			\n"
-"	int bodyOffsetA = offsetSplitBodies[aIdx];\n"
-"	int constraintOffsetA = contactConstraintOffsets[0].x;\n"
-"	int splitIndexA = bodyOffsetA+constraintOffsetA;\n"
-"	\n"
-"	if (invMassA)\n"
-"	{\n"
-"		dLinVelA = deltaLinearVelocities[splitIndexA];\n"
-"		dAngVelA = deltaAngularVelocities[splitIndexA];\n"
-"	}\n"
-"	int bodyOffsetB = offsetSplitBodies[bIdx];\n"
-"	int constraintOffsetB = contactConstraintOffsets[0].y;\n"
-"	int splitIndexB= bodyOffsetB+constraintOffsetB;\n"
-"	if (invMassB)\n"
-"	{\n"
-"		dLinVelB = deltaLinearVelocities[splitIndexB];\n"
-"		dAngVelB = deltaAngularVelocities[splitIndexB];\n"
-"	}\n"
-"	{\n"
-"		float maxRambdaDt[4] = {FLT_MAX,FLT_MAX,FLT_MAX,FLT_MAX};\n"
-"		float minRambdaDt[4] = {0.f,0.f,0.f,0.f};\n"
-"		float sum = 0;\n"
-"		for(int j=0; j<4; j++)\n"
-"		{\n"
-"			sum +=ldsCs[0].m_appliedRambdaDt[j];\n"
-"		}\n"
-"		frictionCoeff = 0.7f;\n"
-"		for(int j=0; j<4; j++)\n"
-"		{\n"
-"			maxRambdaDt[j] = frictionCoeff*sum;\n"
-"			minRambdaDt[j] = -maxRambdaDt[j];\n"
-"		}\n"
-"		\n"
-"//		solveFriction( ldsCs, posA, &linVelA, &angVelA, invMassA, invInertiaA,\n"
-"//			posB, &linVelB, &angVelB, invMassB, invInertiaB, maxRambdaDt, minRambdaDt );\n"
-"		\n"
-"		\n"
-"		{\n"
-"			\n"
-"			__global Constraint4* cs = ldsCs;\n"
-"			\n"
-"			if( cs->m_fJacCoeffInv[0] == 0 && cs->m_fJacCoeffInv[0] == 0 ) return;\n"
-"			const float4 center = cs->m_center;\n"
-"			\n"
-"			float4 n = -cs->m_linear;\n"
-"			\n"
-"			float4 tangent[2];\n"
-"			btPlaneSpace1(n,&tangent[0],&tangent[1]);\n"
-"			float4 angular0, angular1, linear;\n"
-"			float4 r0 = center - posA;\n"
-"			float4 r1 = center - posB;\n"
-"			for(int i=0; i<2; i++)\n"
-"			{\n"
-"				setLinearAndAngular( tangent[i], r0, r1, &linear, &angular0, &angular1 );\n"
-"				float rambdaDt = calcRelVel(linear, -linear, angular0, angular1,\n"
-"											linVelA+dLinVelA, angVelA+dAngVelA, linVelB+dLinVelB, angVelB+dAngVelB );\n"
-"				rambdaDt *= cs->m_fJacCoeffInv[i];\n"
-"				\n"
-"				{\n"
-"					float prevSum = cs->m_fAppliedRambdaDt[i];\n"
-"					float updated = prevSum;\n"
-"					updated += rambdaDt;\n"
-"					updated = max2( updated, minRambdaDt[i] );\n"
-"					updated = min2( updated, maxRambdaDt[i] );\n"
-"					rambdaDt = updated - prevSum;\n"
-"					cs->m_fAppliedRambdaDt[i] = updated;\n"
-"				}\n"
-"				\n"
-"				float4 linImp0 = invMassA*linear*rambdaDt;\n"
-"				float4 linImp1 = invMassB*(-linear)*rambdaDt;\n"
-"				float4 angImp0 = mtMul1(invInertiaA, angular0)*rambdaDt;\n"
-"				float4 angImp1 = mtMul1(invInertiaB, angular1)*rambdaDt;\n"
-"				\n"
-"				dLinVelA += linImp0;\n"
-"				dAngVelA += angImp0;\n"
-"				dLinVelB += linImp1;\n"
-"				dAngVelB += angImp1;\n"
-"			}\n"
-"			{	//	angular damping for point constraint\n"
-"				float4 ab = normalize3( posB - posA );\n"
-"				float4 ac = normalize3( center - posA );\n"
-"				if( dot3F4( ab, ac ) > 0.95f  || (invMassA == 0.f || invMassB == 0.f))\n"
-"				{\n"
-"					float angNA = dot3F4( n, angVelA );\n"
-"					float angNB = dot3F4( n, angVelB );\n"
-"					\n"
-"					dAngVelA -= (angNA*0.1f)*n;\n"
-"					dAngVelB -= (angNB*0.1f)*n;\n"
-"				}\n"
-"			}\n"
-"		}\n"
-"		\n"
-"		\n"
-"	}\n"
-"	if (invMassA)\n"
-"	{\n"
-"		deltaLinearVelocities[splitIndexA] = dLinVelA;\n"
-"		deltaAngularVelocities[splitIndexA] = dAngVelA;\n"
-"	} \n"
-"	if (invMassB)\n"
-"	{\n"
-"		deltaLinearVelocities[splitIndexB] = dLinVelB;\n"
-"		deltaAngularVelocities[splitIndexB] = dAngVelB;\n"
-"	}\n"
-" \n"
-"}\n"
-"__kernel void SolveFrictionJacobiKernel(__global Constraint4* gConstraints, __global Body* gBodies, __global Shape* gShapes ,\n"
-"										__global int2* contactConstraintOffsets,__global unsigned int* offsetSplitBodies,\n"
-"										__global float4* deltaLinearVelocities, __global float4* deltaAngularVelocities,\n"
-"										float deltaTime, float positionDrift, float positionConstraintCoeff, int fixedBodyIndex, int numManifolds\n"
-")\n"
-"{\n"
-"	int i = GET_GLOBAL_IDX;\n"
-"	if (i<numManifolds)\n"
-"	{\n"
-"		solveFrictionConstraint( gBodies, gShapes, &gConstraints[i] ,&contactConstraintOffsets[i],offsetSplitBodies, deltaLinearVelocities, deltaAngularVelocities);\n"
-"	}\n"
-"}\n"
-"__kernel void UpdateBodyVelocitiesKernel(__global Body* gBodies,__global int* offsetSplitBodies,__global const unsigned int* bodyCount,\n"
-"									__global float4* deltaLinearVelocities, __global float4* deltaAngularVelocities, int numBodies)\n"
-"{\n"
-"	int i = GET_GLOBAL_IDX;\n"
-"	if (i<numBodies)\n"
-"	{\n"
-"		if (gBodies[i].m_invMass)\n"
-"		{\n"
-"			int bodyOffset = offsetSplitBodies[i];\n"
-"			int count = bodyCount[i];\n"
-"			if (count)\n"
-"			{\n"
-"				gBodies[i].m_linVel += deltaLinearVelocities[bodyOffset];\n"
-"				gBodies[i].m_angVel += deltaAngularVelocities[bodyOffset];\n"
-"			}\n"
-"		}\n"
-"	}\n"
-"}\n"
-"void setConstraint4( const float4 posA, const float4 linVelA, const float4 angVelA, float invMassA, const Matrix3x3 invInertiaA,\n"
-"	const float4 posB, const float4 linVelB, const float4 angVelB, float invMassB, const Matrix3x3 invInertiaB, \n"
-"	__global struct b3Contact4Data* src, float dt, float positionDrift, float positionConstraintCoeff,float countA, float countB,\n"
-"	Constraint4* dstC )\n"
-"{\n"
-"	dstC->m_bodyA = abs(src->m_bodyAPtrAndSignBit);\n"
-"	dstC->m_bodyB = abs(src->m_bodyBPtrAndSignBit);\n"
-"	float dtInv = 1.f/dt;\n"
-"	for(int ic=0; ic<4; ic++)\n"
-"	{\n"
-"		dstC->m_appliedRambdaDt[ic] = 0.f;\n"
-"	}\n"
-"	dstC->m_fJacCoeffInv[0] = dstC->m_fJacCoeffInv[1] = 0.f;\n"
-"	dstC->m_linear = src->m_worldNormalOnB;\n"
-"	dstC->m_linear.w = 0.7f ;//src->getFrictionCoeff() );\n"
-"	for(int ic=0; ic<4; ic++)\n"
-"	{\n"
-"		float4 r0 = src->m_worldPosB[ic] - posA;\n"
-"		float4 r1 = src->m_worldPosB[ic] - posB;\n"
-"		if( ic >= src->m_worldNormalOnB.w )//npoints\n"
-"		{\n"
-"			dstC->m_jacCoeffInv[ic] = 0.f;\n"
-"			continue;\n"
-"		}\n"
-"		float relVelN;\n"
-"		{\n"
-"			float4 linear, angular0, angular1;\n"
-"			setLinearAndAngular(src->m_worldNormalOnB, r0, r1, &linear, &angular0, &angular1);\n"
-"			dstC->m_jacCoeffInv[ic] = calcJacCoeff(linear, -linear, angular0, angular1,\n"
-"				invMassA, &invInertiaA, invMassB, &invInertiaB , countA, countB);\n"
-"			relVelN = calcRelVel(linear, -linear, angular0, angular1,\n"
-"				linVelA, angVelA, linVelB, angVelB);\n"
-"			float e = 0.f;//src->getRestituitionCoeff();\n"
-"			if( relVelN*relVelN < 0.004f ) e = 0.f;\n"
-"			dstC->m_b[ic] = e*relVelN;\n"
-"			//float penetration = src->m_worldPosB[ic].w;\n"
-"			dstC->m_b[ic] += (src->m_worldPosB[ic].w + positionDrift)*positionConstraintCoeff*dtInv;\n"
-"			dstC->m_appliedRambdaDt[ic] = 0.f;\n"
-"		}\n"
-"	}\n"
-"	if( src->m_worldNormalOnB.w > 0 )//npoints\n"
-"	{	//	prepare friction\n"
-"		float4 center = make_float4(0.f);\n"
-"		for(int i=0; i<src->m_worldNormalOnB.w; i++) \n"
-"			center += src->m_worldPosB[i];\n"
-"		center /= (float)src->m_worldNormalOnB.w;\n"
-"		float4 tangent[2];\n"
-"		btPlaneSpace1(-src->m_worldNormalOnB,&tangent[0],&tangent[1]);\n"
-"		\n"
-"		float4 r[2];\n"
-"		r[0] = center - posA;\n"
-"		r[1] = center - posB;\n"
-"		for(int i=0; i<2; i++)\n"
-"		{\n"
-"			float4 linear, angular0, angular1;\n"
-"			setLinearAndAngular(tangent[i], r[0], r[1], &linear, &angular0, &angular1);\n"
-"			dstC->m_fJacCoeffInv[i] = calcJacCoeff(linear, -linear, angular0, angular1,\n"
-"				invMassA, &invInertiaA, invMassB, &invInertiaB ,countA, countB);\n"
-"			dstC->m_fAppliedRambdaDt[i] = 0.f;\n"
-"		}\n"
-"		dstC->m_center = center;\n"
-"	}\n"
-"	for(int i=0; i<4; i++)\n"
-"	{\n"
-"		if( i<src->m_worldNormalOnB.w )\n"
-"		{\n"
-"			dstC->m_worldPos[i] = src->m_worldPosB[i];\n"
-"		}\n"
-"		else\n"
-"		{\n"
-"			dstC->m_worldPos[i] = make_float4(0.f);\n"
-"		}\n"
-"	}\n"
-"}\n"
-"__kernel\n"
-"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
-"void ContactToConstraintSplitKernel(__global const struct b3Contact4Data* gContact, __global const Body* gBodies, __global const Shape* gShapes, __global Constraint4* gConstraintOut, \n"
-"__global const unsigned int* bodyCount,\n"
-"int nContacts,\n"
-"float dt,\n"
-"float positionDrift,\n"
-"float positionConstraintCoeff\n"
-")\n"
-"{\n"
-"	int gIdx = GET_GLOBAL_IDX;\n"
-"	\n"
-"	if( gIdx < nContacts )\n"
-"	{\n"
-"		int aIdx = abs(gContact[gIdx].m_bodyAPtrAndSignBit);\n"
-"		int bIdx = abs(gContact[gIdx].m_bodyBPtrAndSignBit);\n"
-"		float4 posA = gBodies[aIdx].m_pos;\n"
-"		float4 linVelA = gBodies[aIdx].m_linVel;\n"
-"		float4 angVelA = gBodies[aIdx].m_angVel;\n"
-"		float invMassA = gBodies[aIdx].m_invMass;\n"
-"		Matrix3x3 invInertiaA = gShapes[aIdx].m_invInertia;\n"
-"		float4 posB = gBodies[bIdx].m_pos;\n"
-"		float4 linVelB = gBodies[bIdx].m_linVel;\n"
-"		float4 angVelB = gBodies[bIdx].m_angVel;\n"
-"		float invMassB = gBodies[bIdx].m_invMass;\n"
-"		Matrix3x3 invInertiaB = gShapes[bIdx].m_invInertia;\n"
-"		Constraint4 cs;\n"
-"		float countA = invMassA != 0.f ? (float)bodyCount[aIdx] : 1;\n"
-"		float countB = invMassB != 0.f ? (float)bodyCount[bIdx] : 1;\n"
-"    	setConstraint4( posA, linVelA, angVelA, invMassA, invInertiaA, posB, linVelB, angVelB, invMassB, invInertiaB,\n"
-"			&gContact[gIdx], dt, positionDrift, positionConstraintCoeff,countA,countB,\n"
-"			&cs  );\n"
-"		\n"
-"		cs.m_batchIdx = gContact[gIdx].m_batchIdx;\n"
-"		gConstraintOut[gIdx] = cs;\n"
-"	}\n"
-"}\n"
-;
+static const char* solverUtilsCL =
+	"/*\n"
+	"Copyright (c) 2013 Advanced Micro Devices, Inc.  \n"
+	"This software is provided 'as-is', without any express or implied warranty.\n"
+	"In no event will the authors be held liable for any damages arising from the use of this software.\n"
+	"Permission is granted to anyone to use this software for any purpose, \n"
+	"including commercial applications, and to alter it and redistribute it freely, \n"
+	"subject to the following restrictions:\n"
+	"1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.\n"
+	"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
+	"3. This notice may not be removed or altered from any source distribution.\n"
+	"*/\n"
+	"//Originally written by Erwin Coumans\n"
+	"#ifndef B3_CONTACT4DATA_H\n"
+	"#define B3_CONTACT4DATA_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#define B3_FLOAT4_H\n"
+	"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
+	"#define B3_PLATFORM_DEFINITIONS_H\n"
+	"struct MyTest\n"
+	"{\n"
+	"	int bla;\n"
+	"};\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
+	"#define B3_LARGE_FLOAT 1e18f\n"
+	"#define B3_INFINITY 1e18f\n"
+	"#define b3Assert(a)\n"
+	"#define b3ConstArray(a) __global const a*\n"
+	"#define b3AtomicInc atomic_inc\n"
+	"#define b3AtomicAdd atomic_add\n"
+	"#define b3Fabs fabs\n"
+	"#define b3Sqrt native_sqrt\n"
+	"#define b3Sin native_sin\n"
+	"#define b3Cos native_cos\n"
+	"#define B3_STATIC\n"
+	"#endif\n"
+	"#endif\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"	typedef float4	b3Float4;\n"
+	"	#define b3Float4ConstArg const b3Float4\n"
+	"	#define b3MakeFloat4 (float4)\n"
+	"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return dot(a1, b1);\n"
+	"	}\n"
+	"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return cross(a1, b1);\n"
+	"	}\n"
+	"	#define b3MinFloat4 min\n"
+	"	#define b3MaxFloat4 max\n"
+	"	#define b3Normalized(a) normalize(a)\n"
+	"#endif \n"
+	"		\n"
+	"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
+	"{\n"
+	"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
+	"		return false;\n"
+	"	return true;\n"
+	"}\n"
+	"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
+	"{\n"
+	"    float maxDot = -B3_INFINITY;\n"
+	"    int i = 0;\n"
+	"    int ptIndex = -1;\n"
+	"    for( i = 0; i < vecLen; i++ )\n"
+	"    {\n"
+	"        float dot = b3Dot3F4(vecArray[i],vec);\n"
+	"            \n"
+	"        if( dot > maxDot )\n"
+	"        {\n"
+	"            maxDot = dot;\n"
+	"            ptIndex = i;\n"
+	"        }\n"
+	"    }\n"
+	"	b3Assert(ptIndex>=0);\n"
+	"    if (ptIndex<0)\n"
+	"	{\n"
+	"		ptIndex = 0;\n"
+	"	}\n"
+	"    *dotOut = maxDot;\n"
+	"    return ptIndex;\n"
+	"}\n"
+	"#endif //B3_FLOAT4_H\n"
+	"typedef  struct b3Contact4Data b3Contact4Data_t;\n"
+	"struct b3Contact4Data\n"
+	"{\n"
+	"	b3Float4	m_worldPosB[4];\n"
+	"//	b3Float4	m_localPosA[4];\n"
+	"//	b3Float4	m_localPosB[4];\n"
+	"	b3Float4	m_worldNormalOnB;	//	w: m_nPoints\n"
+	"	unsigned short  m_restituitionCoeffCmp;\n"
+	"	unsigned short  m_frictionCoeffCmp;\n"
+	"	int m_batchIdx;\n"
+	"	int m_bodyAPtrAndSignBit;//x:m_bodyAPtr, y:m_bodyBPtr\n"
+	"	int m_bodyBPtrAndSignBit;\n"
+	"	int	m_childIndexA;\n"
+	"	int	m_childIndexB;\n"
+	"	int m_unused1;\n"
+	"	int m_unused2;\n"
+	"};\n"
+	"inline int b3Contact4Data_getNumPoints(const struct b3Contact4Data* contact)\n"
+	"{\n"
+	"	return (int)contact->m_worldNormalOnB.w;\n"
+	"};\n"
+	"inline void b3Contact4Data_setNumPoints(struct b3Contact4Data* contact, int numPoints)\n"
+	"{\n"
+	"	contact->m_worldNormalOnB.w = (float)numPoints;\n"
+	"};\n"
+	"#endif //B3_CONTACT4DATA_H\n"
+	"#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_local_int32_extended_atomics : enable\n"
+	"#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics : enable\n"
+	"#ifdef cl_ext_atomic_counters_32\n"
+	"#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
+	"#else\n"
+	"#define counter32_t volatile global int*\n"
+	"#endif\n"
+	"typedef unsigned int u32;\n"
+	"typedef unsigned short u16;\n"
+	"typedef unsigned char u8;\n"
+	"#define GET_GROUP_IDX get_group_id(0)\n"
+	"#define GET_LOCAL_IDX get_local_id(0)\n"
+	"#define GET_GLOBAL_IDX get_global_id(0)\n"
+	"#define GET_GROUP_SIZE get_local_size(0)\n"
+	"#define GET_NUM_GROUPS get_num_groups(0)\n"
+	"#define GROUP_LDS_BARRIER barrier(CLK_LOCAL_MEM_FENCE)\n"
+	"#define GROUP_MEM_FENCE mem_fence(CLK_LOCAL_MEM_FENCE)\n"
+	"#define AtomInc(x) atom_inc(&(x))\n"
+	"#define AtomInc1(x, out) out = atom_inc(&(x))\n"
+	"#define AppendInc(x, out) out = atomic_inc(x)\n"
+	"#define AtomAdd(x, value) atom_add(&(x), value)\n"
+	"#define AtomCmpxhg(x, cmp, value) atom_cmpxchg( &(x), cmp, value )\n"
+	"#define AtomXhg(x, value) atom_xchg ( &(x), value )\n"
+	"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
+	"#define make_float4 (float4)\n"
+	"#define make_float2 (float2)\n"
+	"#define make_uint4 (uint4)\n"
+	"#define make_int4 (int4)\n"
+	"#define make_uint2 (uint2)\n"
+	"#define make_int2 (int2)\n"
+	"#define max2 max\n"
+	"#define min2 min\n"
+	"///////////////////////////////////////\n"
+	"//	Vector\n"
+	"///////////////////////////////////////\n"
+	"__inline\n"
+	"float fastDiv(float numerator, float denominator)\n"
+	"{\n"
+	"	return native_divide(numerator, denominator);	\n"
+	"//	return numerator/denominator;	\n"
+	"}\n"
+	"__inline\n"
+	"float4 fastDiv4(float4 numerator, float4 denominator)\n"
+	"{\n"
+	"	return native_divide(numerator, denominator);	\n"
+	"}\n"
+	"__inline\n"
+	"float fastSqrtf(float f2)\n"
+	"{\n"
+	"	return native_sqrt(f2);\n"
+	"//	return sqrt(f2);\n"
+	"}\n"
+	"__inline\n"
+	"float fastRSqrt(float f2)\n"
+	"{\n"
+	"	return native_rsqrt(f2);\n"
+	"}\n"
+	"__inline\n"
+	"float fastLength4(float4 v)\n"
+	"{\n"
+	"	return fast_length(v);\n"
+	"}\n"
+	"__inline\n"
+	"float4 fastNormalize4(float4 v)\n"
+	"{\n"
+	"	return fast_normalize(v);\n"
+	"}\n"
+	"__inline\n"
+	"float sqrtf(float a)\n"
+	"{\n"
+	"//	return sqrt(a);\n"
+	"	return native_sqrt(a);\n"
+	"}\n"
+	"__inline\n"
+	"float4 cross3(float4 a1, float4 b1)\n"
+	"{\n"
+	"	float4 	a=make_float4(a1.xyz,0.f);\n"
+	"	float4 	b=make_float4(b1.xyz,0.f);\n"
+	"	//float4 	a=a1;\n"
+	"	//float4 	b=b1;\n"
+	"	return cross(a,b);\n"
+	"}\n"
+	"__inline\n"
+	"float dot3F4(float4 a, float4 b)\n"
+	"{\n"
+	"	float4 a1 = make_float4(a.xyz,0.f);\n"
+	"	float4 b1 = make_float4(b.xyz,0.f);\n"
+	"	return dot(a1, b1);\n"
+	"}\n"
+	"__inline\n"
+	"float length3(const float4 a)\n"
+	"{\n"
+	"	return sqrtf(dot3F4(a,a));\n"
+	"}\n"
+	"__inline\n"
+	"float dot4(const float4 a, const float4 b)\n"
+	"{\n"
+	"	return dot( a, b );\n"
+	"}\n"
+	"//	for height\n"
+	"__inline\n"
+	"float dot3w1(const float4 point, const float4 eqn)\n"
+	"{\n"
+	"	return dot3F4(point,eqn) + eqn.w;\n"
+	"}\n"
+	"__inline\n"
+	"float4 normalize3(const float4 a)\n"
+	"{\n"
+	"	float4 n = make_float4(a.x, a.y, a.z, 0.f);\n"
+	"	return fastNormalize4( n );\n"
+	"//	float length = sqrtf(dot3F4(a, a));\n"
+	"//	return 1.f/length * a;\n"
+	"}\n"
+	"__inline\n"
+	"float4 normalize4(const float4 a)\n"
+	"{\n"
+	"	float length = sqrtf(dot4(a, a));\n"
+	"	return 1.f/length * a;\n"
+	"}\n"
+	"__inline\n"
+	"float4 createEquation(const float4 a, const float4 b, const float4 c)\n"
+	"{\n"
+	"	float4 eqn;\n"
+	"	float4 ab = b-a;\n"
+	"	float4 ac = c-a;\n"
+	"	eqn = normalize3( cross3(ab, ac) );\n"
+	"	eqn.w = -dot3F4(eqn,a);\n"
+	"	return eqn;\n"
+	"}\n"
+	"///////////////////////////////////////\n"
+	"//	Matrix3x3\n"
+	"///////////////////////////////////////\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_row[3];\n"
+	"}Matrix3x3;\n"
+	"__inline\n"
+	"Matrix3x3 mtZero();\n"
+	"__inline\n"
+	"Matrix3x3 mtIdentity();\n"
+	"__inline\n"
+	"Matrix3x3 mtTranspose(Matrix3x3 m);\n"
+	"__inline\n"
+	"Matrix3x3 mtMul(Matrix3x3 a, Matrix3x3 b);\n"
+	"__inline\n"
+	"float4 mtMul1(Matrix3x3 a, float4 b);\n"
+	"__inline\n"
+	"float4 mtMul3(float4 a, Matrix3x3 b);\n"
+	"__inline\n"
+	"Matrix3x3 mtZero()\n"
+	"{\n"
+	"	Matrix3x3 m;\n"
+	"	m.m_row[0] = (float4)(0.f);\n"
+	"	m.m_row[1] = (float4)(0.f);\n"
+	"	m.m_row[2] = (float4)(0.f);\n"
+	"	return m;\n"
+	"}\n"
+	"__inline\n"
+	"Matrix3x3 mtIdentity()\n"
+	"{\n"
+	"	Matrix3x3 m;\n"
+	"	m.m_row[0] = (float4)(1,0,0,0);\n"
+	"	m.m_row[1] = (float4)(0,1,0,0);\n"
+	"	m.m_row[2] = (float4)(0,0,1,0);\n"
+	"	return m;\n"
+	"}\n"
+	"__inline\n"
+	"Matrix3x3 mtTranspose(Matrix3x3 m)\n"
+	"{\n"
+	"	Matrix3x3 out;\n"
+	"	out.m_row[0] = (float4)(m.m_row[0].x, m.m_row[1].x, m.m_row[2].x, 0.f);\n"
+	"	out.m_row[1] = (float4)(m.m_row[0].y, m.m_row[1].y, m.m_row[2].y, 0.f);\n"
+	"	out.m_row[2] = (float4)(m.m_row[0].z, m.m_row[1].z, m.m_row[2].z, 0.f);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"Matrix3x3 mtMul(Matrix3x3 a, Matrix3x3 b)\n"
+	"{\n"
+	"	Matrix3x3 transB;\n"
+	"	transB = mtTranspose( b );\n"
+	"	Matrix3x3 ans;\n"
+	"	//	why this doesn't run when 0ing in the for{}\n"
+	"	a.m_row[0].w = 0.f;\n"
+	"	a.m_row[1].w = 0.f;\n"
+	"	a.m_row[2].w = 0.f;\n"
+	"	for(int i=0; i<3; i++)\n"
+	"	{\n"
+	"//	a.m_row[i].w = 0.f;\n"
+	"		ans.m_row[i].x = dot3F4(a.m_row[i],transB.m_row[0]);\n"
+	"		ans.m_row[i].y = dot3F4(a.m_row[i],transB.m_row[1]);\n"
+	"		ans.m_row[i].z = dot3F4(a.m_row[i],transB.m_row[2]);\n"
+	"		ans.m_row[i].w = 0.f;\n"
+	"	}\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"float4 mtMul1(Matrix3x3 a, float4 b)\n"
+	"{\n"
+	"	float4 ans;\n"
+	"	ans.x = dot3F4( a.m_row[0], b );\n"
+	"	ans.y = dot3F4( a.m_row[1], b );\n"
+	"	ans.z = dot3F4( a.m_row[2], b );\n"
+	"	ans.w = 0.f;\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"float4 mtMul3(float4 a, Matrix3x3 b)\n"
+	"{\n"
+	"	float4 colx = make_float4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
+	"	float4 coly = make_float4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
+	"	float4 colz = make_float4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
+	"	float4 ans;\n"
+	"	ans.x = dot3F4( a, colx );\n"
+	"	ans.y = dot3F4( a, coly );\n"
+	"	ans.z = dot3F4( a, colz );\n"
+	"	return ans;\n"
+	"}\n"
+	"///////////////////////////////////////\n"
+	"//	Quaternion\n"
+	"///////////////////////////////////////\n"
+	"typedef float4 Quaternion;\n"
+	"__inline\n"
+	"Quaternion qtMul(Quaternion a, Quaternion b);\n"
+	"__inline\n"
+	"Quaternion qtNormalize(Quaternion in);\n"
+	"__inline\n"
+	"float4 qtRotate(Quaternion q, float4 vec);\n"
+	"__inline\n"
+	"Quaternion qtInvert(Quaternion q);\n"
+	"__inline\n"
+	"Quaternion qtMul(Quaternion a, Quaternion b)\n"
+	"{\n"
+	"	Quaternion ans;\n"
+	"	ans = cross3( a, b );\n"
+	"	ans += a.w*b+b.w*a;\n"
+	"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
+	"	ans.w = a.w*b.w - dot3F4(a, b);\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"Quaternion qtNormalize(Quaternion in)\n"
+	"{\n"
+	"	return fastNormalize4(in);\n"
+	"//	in /= length( in );\n"
+	"//	return in;\n"
+	"}\n"
+	"__inline\n"
+	"float4 qtRotate(Quaternion q, float4 vec)\n"
+	"{\n"
+	"	Quaternion qInv = qtInvert( q );\n"
+	"	float4 vcpy = vec;\n"
+	"	vcpy.w = 0.f;\n"
+	"	float4 out = qtMul(qtMul(q,vcpy),qInv);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"Quaternion qtInvert(Quaternion q)\n"
+	"{\n"
+	"	return (Quaternion)(-q.xyz, q.w);\n"
+	"}\n"
+	"__inline\n"
+	"float4 qtInvRotate(const Quaternion q, float4 vec)\n"
+	"{\n"
+	"	return qtRotate( qtInvert( q ), vec );\n"
+	"}\n"
+	"#define WG_SIZE 64\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_pos;\n"
+	"	Quaternion m_quat;\n"
+	"	float4 m_linVel;\n"
+	"	float4 m_angVel;\n"
+	"	u32 m_shapeIdx;\n"
+	"	float m_invMass;\n"
+	"	float m_restituitionCoeff;\n"
+	"	float m_frictionCoeff;\n"
+	"} Body;\n"
+	"typedef struct\n"
+	"{\n"
+	"	Matrix3x3 m_invInertia;\n"
+	"	Matrix3x3 m_initInvInertia;\n"
+	"} Shape;\n"
+	"typedef struct\n"
+	"{\n"
+	"	float4 m_linear;\n"
+	"	float4 m_worldPos[4];\n"
+	"	float4 m_center;	\n"
+	"	float m_jacCoeffInv[4];\n"
+	"	float m_b[4];\n"
+	"	float m_appliedRambdaDt[4];\n"
+	"	float m_fJacCoeffInv[2];	\n"
+	"	float m_fAppliedRambdaDt[2];	\n"
+	"	u32 m_bodyA;\n"
+	"	u32 m_bodyB;\n"
+	"	int m_batchIdx;\n"
+	"	u32 m_paddings;\n"
+	"} Constraint4;\n"
+	"__kernel void CountBodiesKernel(__global struct b3Contact4Data* manifoldPtr, __global unsigned int* bodyCount, __global int2* contactConstraintOffsets, int numContactManifolds, int fixedBodyIndex)\n"
+	"{\n"
+	"	int i = GET_GLOBAL_IDX;\n"
+	"	\n"
+	"	if( i < numContactManifolds)\n"
+	"	{\n"
+	"		int pa = manifoldPtr[i].m_bodyAPtrAndSignBit;\n"
+	"		bool isFixedA = (pa <0) || (pa == fixedBodyIndex);\n"
+	"		int bodyIndexA = abs(pa);\n"
+	"		if (!isFixedA)\n"
+	"		{\n"
+	"			 AtomInc1(bodyCount[bodyIndexA],contactConstraintOffsets[i].x);\n"
+	"		}\n"
+	"		barrier(CLK_GLOBAL_MEM_FENCE);\n"
+	"		int pb = manifoldPtr[i].m_bodyBPtrAndSignBit;\n"
+	"		bool isFixedB = (pb <0) || (pb == fixedBodyIndex);\n"
+	"		int bodyIndexB = abs(pb);\n"
+	"		if (!isFixedB)\n"
+	"		{\n"
+	"			AtomInc1(bodyCount[bodyIndexB],contactConstraintOffsets[i].y);\n"
+	"		} \n"
+	"	}\n"
+	"}\n"
+	"__kernel void ClearVelocitiesKernel(__global float4* linearVelocities,__global float4* angularVelocities, int numSplitBodies)\n"
+	"{\n"
+	"	int i = GET_GLOBAL_IDX;\n"
+	"	\n"
+	"	if( i < numSplitBodies)\n"
+	"	{\n"
+	"		linearVelocities[i] = make_float4(0);\n"
+	"		angularVelocities[i] = make_float4(0);\n"
+	"	}\n"
+	"}\n"
+	"__kernel void AverageVelocitiesKernel(__global Body* gBodies,__global int* offsetSplitBodies,__global const unsigned int* bodyCount,\n"
+	"__global float4* deltaLinearVelocities, __global float4* deltaAngularVelocities, int numBodies)\n"
+	"{\n"
+	"	int i = GET_GLOBAL_IDX;\n"
+	"	if (i<numBodies)\n"
+	"	{\n"
+	"		if (gBodies[i].m_invMass)\n"
+	"		{\n"
+	"			int bodyOffset = offsetSplitBodies[i];\n"
+	"			int count = bodyCount[i];\n"
+	"			float factor = 1.f/((float)count);\n"
+	"			float4 averageLinVel = make_float4(0.f);\n"
+	"			float4 averageAngVel = make_float4(0.f);\n"
+	"			\n"
+	"			for (int j=0;j<count;j++)\n"
+	"			{\n"
+	"				averageLinVel += deltaLinearVelocities[bodyOffset+j]*factor;\n"
+	"				averageAngVel += deltaAngularVelocities[bodyOffset+j]*factor;\n"
+	"			}\n"
+	"			\n"
+	"			for (int j=0;j<count;j++)\n"
+	"			{\n"
+	"				deltaLinearVelocities[bodyOffset+j] = averageLinVel;\n"
+	"				deltaAngularVelocities[bodyOffset+j] = averageAngVel;\n"
+	"			}\n"
+	"			\n"
+	"		}//bodies[i].m_invMass\n"
+	"	}//i<numBodies\n"
+	"}\n"
+	"void setLinearAndAngular( float4 n, float4 r0, float4 r1, float4* linear, float4* angular0, float4* angular1)\n"
+	"{\n"
+	"	*linear = make_float4(n.xyz,0.f);\n"
+	"	*angular0 = cross3(r0, n);\n"
+	"	*angular1 = -cross3(r1, n);\n"
+	"}\n"
+	"float calcRelVel( float4 l0, float4 l1, float4 a0, float4 a1, float4 linVel0, float4 angVel0, float4 linVel1, float4 angVel1 )\n"
+	"{\n"
+	"	return dot3F4(l0, linVel0) + dot3F4(a0, angVel0) + dot3F4(l1, linVel1) + dot3F4(a1, angVel1);\n"
+	"}\n"
+	"float calcJacCoeff(const float4 linear0, const float4 linear1, const float4 angular0, const float4 angular1,\n"
+	"					float invMass0, const Matrix3x3* invInertia0, float invMass1, const Matrix3x3* invInertia1, float countA, float countB)\n"
+	"{\n"
+	"	//	linear0,1 are normlized\n"
+	"	float jmj0 = invMass0;//dot3F4(linear0, linear0)*invMass0;\n"
+	"	float jmj1 = dot3F4(mtMul3(angular0,*invInertia0), angular0);\n"
+	"	float jmj2 = invMass1;//dot3F4(linear1, linear1)*invMass1;\n"
+	"	float jmj3 = dot3F4(mtMul3(angular1,*invInertia1), angular1);\n"
+	"	return -1.f/((jmj0+jmj1)*countA+(jmj2+jmj3)*countB);\n"
+	"}\n"
+	"void btPlaneSpace1 (float4 n, float4* p, float4* q);\n"
+	" void btPlaneSpace1 (float4 n, float4* p, float4* q)\n"
+	"{\n"
+	"  if (fabs(n.z) > 0.70710678f) {\n"
+	"    // choose p in y-z plane\n"
+	"    float a = n.y*n.y + n.z*n.z;\n"
+	"    float k = 1.f/sqrt(a);\n"
+	"    p[0].x = 0;\n"
+	"	p[0].y = -n.z*k;\n"
+	"	p[0].z = n.y*k;\n"
+	"    // set q = n x p\n"
+	"    q[0].x = a*k;\n"
+	"	q[0].y = -n.x*p[0].z;\n"
+	"	q[0].z = n.x*p[0].y;\n"
+	"  }\n"
+	"  else {\n"
+	"    // choose p in x-y plane\n"
+	"    float a = n.x*n.x + n.y*n.y;\n"
+	"    float k = 1.f/sqrt(a);\n"
+	"    p[0].x = -n.y*k;\n"
+	"	p[0].y = n.x*k;\n"
+	"	p[0].z = 0;\n"
+	"    // set q = n x p\n"
+	"    q[0].x = -n.z*p[0].y;\n"
+	"	q[0].y = n.z*p[0].x;\n"
+	"	q[0].z = a*k;\n"
+	"  }\n"
+	"}\n"
+	"void solveContact(__global Constraint4* cs,\n"
+	"			float4 posA, float4* linVelA, float4* angVelA, float invMassA, Matrix3x3 invInertiaA,\n"
+	"			float4 posB, float4* linVelB, float4* angVelB, float invMassB, Matrix3x3 invInertiaB,\n"
+	"			float4* dLinVelA, float4* dAngVelA, float4* dLinVelB, float4* dAngVelB)\n"
+	"{\n"
+	"	float minRambdaDt = 0;\n"
+	"	float maxRambdaDt = FLT_MAX;\n"
+	"	for(int ic=0; ic<4; ic++)\n"
+	"	{\n"
+	"		if( cs->m_jacCoeffInv[ic] == 0.f ) continue;\n"
+	"		float4 angular0, angular1, linear;\n"
+	"		float4 r0 = cs->m_worldPos[ic] - posA;\n"
+	"		float4 r1 = cs->m_worldPos[ic] - posB;\n"
+	"		setLinearAndAngular( cs->m_linear, r0, r1, &linear, &angular0, &angular1 );\n"
+	"	\n"
+	"		float rambdaDt = calcRelVel( cs->m_linear, -cs->m_linear, angular0, angular1, \n"
+	"			*linVelA+*dLinVelA, *angVelA+*dAngVelA, *linVelB+*dLinVelB, *angVelB+*dAngVelB ) + cs->m_b[ic];\n"
+	"		rambdaDt *= cs->m_jacCoeffInv[ic];\n"
+	"		\n"
+	"		{\n"
+	"			float prevSum = cs->m_appliedRambdaDt[ic];\n"
+	"			float updated = prevSum;\n"
+	"			updated += rambdaDt;\n"
+	"			updated = max2( updated, minRambdaDt );\n"
+	"			updated = min2( updated, maxRambdaDt );\n"
+	"			rambdaDt = updated - prevSum;\n"
+	"			cs->m_appliedRambdaDt[ic] = updated;\n"
+	"		}\n"
+	"			\n"
+	"		float4 linImp0 = invMassA*linear*rambdaDt;\n"
+	"		float4 linImp1 = invMassB*(-linear)*rambdaDt;\n"
+	"		float4 angImp0 = mtMul1(invInertiaA, angular0)*rambdaDt;\n"
+	"		float4 angImp1 = mtMul1(invInertiaB, angular1)*rambdaDt;\n"
+	"		\n"
+	"		if (invMassA)\n"
+	"		{\n"
+	"			*dLinVelA += linImp0;\n"
+	"			*dAngVelA += angImp0;\n"
+	"		}\n"
+	"		if (invMassB)\n"
+	"		{\n"
+	"			*dLinVelB += linImp1;\n"
+	"			*dAngVelB += angImp1;\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"//	solveContactConstraint( gBodies, gShapes, &gConstraints[i] ,contactConstraintOffsets,offsetSplitBodies, deltaLinearVelocities, deltaAngularVelocities);\n"
+	"void solveContactConstraint(__global Body* gBodies, __global Shape* gShapes, __global Constraint4* ldsCs, \n"
+	"__global int2* contactConstraintOffsets,__global unsigned int* offsetSplitBodies,\n"
+	"__global float4* deltaLinearVelocities, __global float4* deltaAngularVelocities)\n"
+	"{\n"
+	"	//float frictionCoeff = ldsCs[0].m_linear.w;\n"
+	"	int aIdx = ldsCs[0].m_bodyA;\n"
+	"	int bIdx = ldsCs[0].m_bodyB;\n"
+	"	float4 posA = gBodies[aIdx].m_pos;\n"
+	"	float4 linVelA = gBodies[aIdx].m_linVel;\n"
+	"	float4 angVelA = gBodies[aIdx].m_angVel;\n"
+	"	float invMassA = gBodies[aIdx].m_invMass;\n"
+	"	Matrix3x3 invInertiaA = gShapes[aIdx].m_invInertia;\n"
+	"	float4 posB = gBodies[bIdx].m_pos;\n"
+	"	float4 linVelB = gBodies[bIdx].m_linVel;\n"
+	"	float4 angVelB = gBodies[bIdx].m_angVel;\n"
+	"	float invMassB = gBodies[bIdx].m_invMass;\n"
+	"	Matrix3x3 invInertiaB = gShapes[bIdx].m_invInertia;\n"
+	"			\n"
+	"	float4 dLinVelA = make_float4(0,0,0,0);\n"
+	"	float4 dAngVelA = make_float4(0,0,0,0);\n"
+	"	float4 dLinVelB = make_float4(0,0,0,0);\n"
+	"	float4 dAngVelB = make_float4(0,0,0,0);\n"
+	"			\n"
+	"	int bodyOffsetA = offsetSplitBodies[aIdx];\n"
+	"	int constraintOffsetA = contactConstraintOffsets[0].x;\n"
+	"	int splitIndexA = bodyOffsetA+constraintOffsetA;\n"
+	"	\n"
+	"	if (invMassA)\n"
+	"	{\n"
+	"		dLinVelA = deltaLinearVelocities[splitIndexA];\n"
+	"		dAngVelA = deltaAngularVelocities[splitIndexA];\n"
+	"	}\n"
+	"	int bodyOffsetB = offsetSplitBodies[bIdx];\n"
+	"	int constraintOffsetB = contactConstraintOffsets[0].y;\n"
+	"	int splitIndexB= bodyOffsetB+constraintOffsetB;\n"
+	"	if (invMassB)\n"
+	"	{\n"
+	"		dLinVelB = deltaLinearVelocities[splitIndexB];\n"
+	"		dAngVelB = deltaAngularVelocities[splitIndexB];\n"
+	"	}\n"
+	"	solveContact( ldsCs, posA, &linVelA, &angVelA, invMassA, invInertiaA,\n"
+	"			posB, &linVelB, &angVelB, invMassB, invInertiaB ,&dLinVelA, &dAngVelA, &dLinVelB, &dAngVelB);\n"
+	"	if (invMassA)\n"
+	"	{\n"
+	"		deltaLinearVelocities[splitIndexA] = dLinVelA;\n"
+	"		deltaAngularVelocities[splitIndexA] = dAngVelA;\n"
+	"	} \n"
+	"	if (invMassB)\n"
+	"	{\n"
+	"		deltaLinearVelocities[splitIndexB] = dLinVelB;\n"
+	"		deltaAngularVelocities[splitIndexB] = dAngVelB;\n"
+	"	}\n"
+	"}\n"
+	"__kernel void SolveContactJacobiKernel(__global Constraint4* gConstraints, __global Body* gBodies, __global Shape* gShapes ,\n"
+	"__global int2* contactConstraintOffsets,__global unsigned int* offsetSplitBodies,__global float4* deltaLinearVelocities, __global float4* deltaAngularVelocities,\n"
+	"float deltaTime, float positionDrift, float positionConstraintCoeff, int fixedBodyIndex, int numManifolds\n"
+	")\n"
+	"{\n"
+	"	int i = GET_GLOBAL_IDX;\n"
+	"	if (i<numManifolds)\n"
+	"	{\n"
+	"		solveContactConstraint( gBodies, gShapes, &gConstraints[i] ,&contactConstraintOffsets[i],offsetSplitBodies, deltaLinearVelocities, deltaAngularVelocities);\n"
+	"	}\n"
+	"}\n"
+	"void solveFrictionConstraint(__global Body* gBodies, __global Shape* gShapes, __global Constraint4* ldsCs,\n"
+	"							__global int2* contactConstraintOffsets,__global unsigned int* offsetSplitBodies,\n"
+	"							__global float4* deltaLinearVelocities, __global float4* deltaAngularVelocities)\n"
+	"{\n"
+	"	float frictionCoeff = 0.7f;//ldsCs[0].m_linear.w;\n"
+	"	int aIdx = ldsCs[0].m_bodyA;\n"
+	"	int bIdx = ldsCs[0].m_bodyB;\n"
+	"	float4 posA = gBodies[aIdx].m_pos;\n"
+	"	float4 linVelA = gBodies[aIdx].m_linVel;\n"
+	"	float4 angVelA = gBodies[aIdx].m_angVel;\n"
+	"	float invMassA = gBodies[aIdx].m_invMass;\n"
+	"	Matrix3x3 invInertiaA = gShapes[aIdx].m_invInertia;\n"
+	"	float4 posB = gBodies[bIdx].m_pos;\n"
+	"	float4 linVelB = gBodies[bIdx].m_linVel;\n"
+	"	float4 angVelB = gBodies[bIdx].m_angVel;\n"
+	"	float invMassB = gBodies[bIdx].m_invMass;\n"
+	"	Matrix3x3 invInertiaB = gShapes[bIdx].m_invInertia;\n"
+	"	\n"
+	"	float4 dLinVelA = make_float4(0,0,0,0);\n"
+	"	float4 dAngVelA = make_float4(0,0,0,0);\n"
+	"	float4 dLinVelB = make_float4(0,0,0,0);\n"
+	"	float4 dAngVelB = make_float4(0,0,0,0);\n"
+	"			\n"
+	"	int bodyOffsetA = offsetSplitBodies[aIdx];\n"
+	"	int constraintOffsetA = contactConstraintOffsets[0].x;\n"
+	"	int splitIndexA = bodyOffsetA+constraintOffsetA;\n"
+	"	\n"
+	"	if (invMassA)\n"
+	"	{\n"
+	"		dLinVelA = deltaLinearVelocities[splitIndexA];\n"
+	"		dAngVelA = deltaAngularVelocities[splitIndexA];\n"
+	"	}\n"
+	"	int bodyOffsetB = offsetSplitBodies[bIdx];\n"
+	"	int constraintOffsetB = contactConstraintOffsets[0].y;\n"
+	"	int splitIndexB= bodyOffsetB+constraintOffsetB;\n"
+	"	if (invMassB)\n"
+	"	{\n"
+	"		dLinVelB = deltaLinearVelocities[splitIndexB];\n"
+	"		dAngVelB = deltaAngularVelocities[splitIndexB];\n"
+	"	}\n"
+	"	{\n"
+	"		float maxRambdaDt[4] = {FLT_MAX,FLT_MAX,FLT_MAX,FLT_MAX};\n"
+	"		float minRambdaDt[4] = {0.f,0.f,0.f,0.f};\n"
+	"		float sum = 0;\n"
+	"		for(int j=0; j<4; j++)\n"
+	"		{\n"
+	"			sum +=ldsCs[0].m_appliedRambdaDt[j];\n"
+	"		}\n"
+	"		frictionCoeff = 0.7f;\n"
+	"		for(int j=0; j<4; j++)\n"
+	"		{\n"
+	"			maxRambdaDt[j] = frictionCoeff*sum;\n"
+	"			minRambdaDt[j] = -maxRambdaDt[j];\n"
+	"		}\n"
+	"		\n"
+	"//		solveFriction( ldsCs, posA, &linVelA, &angVelA, invMassA, invInertiaA,\n"
+	"//			posB, &linVelB, &angVelB, invMassB, invInertiaB, maxRambdaDt, minRambdaDt );\n"
+	"		\n"
+	"		\n"
+	"		{\n"
+	"			\n"
+	"			__global Constraint4* cs = ldsCs;\n"
+	"			\n"
+	"			if( cs->m_fJacCoeffInv[0] == 0 && cs->m_fJacCoeffInv[0] == 0 ) return;\n"
+	"			const float4 center = cs->m_center;\n"
+	"			\n"
+	"			float4 n = -cs->m_linear;\n"
+	"			\n"
+	"			float4 tangent[2];\n"
+	"			btPlaneSpace1(n,&tangent[0],&tangent[1]);\n"
+	"			float4 angular0, angular1, linear;\n"
+	"			float4 r0 = center - posA;\n"
+	"			float4 r1 = center - posB;\n"
+	"			for(int i=0; i<2; i++)\n"
+	"			{\n"
+	"				setLinearAndAngular( tangent[i], r0, r1, &linear, &angular0, &angular1 );\n"
+	"				float rambdaDt = calcRelVel(linear, -linear, angular0, angular1,\n"
+	"											linVelA+dLinVelA, angVelA+dAngVelA, linVelB+dLinVelB, angVelB+dAngVelB );\n"
+	"				rambdaDt *= cs->m_fJacCoeffInv[i];\n"
+	"				\n"
+	"				{\n"
+	"					float prevSum = cs->m_fAppliedRambdaDt[i];\n"
+	"					float updated = prevSum;\n"
+	"					updated += rambdaDt;\n"
+	"					updated = max2( updated, minRambdaDt[i] );\n"
+	"					updated = min2( updated, maxRambdaDt[i] );\n"
+	"					rambdaDt = updated - prevSum;\n"
+	"					cs->m_fAppliedRambdaDt[i] = updated;\n"
+	"				}\n"
+	"				\n"
+	"				float4 linImp0 = invMassA*linear*rambdaDt;\n"
+	"				float4 linImp1 = invMassB*(-linear)*rambdaDt;\n"
+	"				float4 angImp0 = mtMul1(invInertiaA, angular0)*rambdaDt;\n"
+	"				float4 angImp1 = mtMul1(invInertiaB, angular1)*rambdaDt;\n"
+	"				\n"
+	"				dLinVelA += linImp0;\n"
+	"				dAngVelA += angImp0;\n"
+	"				dLinVelB += linImp1;\n"
+	"				dAngVelB += angImp1;\n"
+	"			}\n"
+	"			{	//	angular damping for point constraint\n"
+	"				float4 ab = normalize3( posB - posA );\n"
+	"				float4 ac = normalize3( center - posA );\n"
+	"				if( dot3F4( ab, ac ) > 0.95f  || (invMassA == 0.f || invMassB == 0.f))\n"
+	"				{\n"
+	"					float angNA = dot3F4( n, angVelA );\n"
+	"					float angNB = dot3F4( n, angVelB );\n"
+	"					\n"
+	"					dAngVelA -= (angNA*0.1f)*n;\n"
+	"					dAngVelB -= (angNB*0.1f)*n;\n"
+	"				}\n"
+	"			}\n"
+	"		}\n"
+	"		\n"
+	"		\n"
+	"	}\n"
+	"	if (invMassA)\n"
+	"	{\n"
+	"		deltaLinearVelocities[splitIndexA] = dLinVelA;\n"
+	"		deltaAngularVelocities[splitIndexA] = dAngVelA;\n"
+	"	} \n"
+	"	if (invMassB)\n"
+	"	{\n"
+	"		deltaLinearVelocities[splitIndexB] = dLinVelB;\n"
+	"		deltaAngularVelocities[splitIndexB] = dAngVelB;\n"
+	"	}\n"
+	" \n"
+	"}\n"
+	"__kernel void SolveFrictionJacobiKernel(__global Constraint4* gConstraints, __global Body* gBodies, __global Shape* gShapes ,\n"
+	"										__global int2* contactConstraintOffsets,__global unsigned int* offsetSplitBodies,\n"
+	"										__global float4* deltaLinearVelocities, __global float4* deltaAngularVelocities,\n"
+	"										float deltaTime, float positionDrift, float positionConstraintCoeff, int fixedBodyIndex, int numManifolds\n"
+	")\n"
+	"{\n"
+	"	int i = GET_GLOBAL_IDX;\n"
+	"	if (i<numManifolds)\n"
+	"	{\n"
+	"		solveFrictionConstraint( gBodies, gShapes, &gConstraints[i] ,&contactConstraintOffsets[i],offsetSplitBodies, deltaLinearVelocities, deltaAngularVelocities);\n"
+	"	}\n"
+	"}\n"
+	"__kernel void UpdateBodyVelocitiesKernel(__global Body* gBodies,__global int* offsetSplitBodies,__global const unsigned int* bodyCount,\n"
+	"									__global float4* deltaLinearVelocities, __global float4* deltaAngularVelocities, int numBodies)\n"
+	"{\n"
+	"	int i = GET_GLOBAL_IDX;\n"
+	"	if (i<numBodies)\n"
+	"	{\n"
+	"		if (gBodies[i].m_invMass)\n"
+	"		{\n"
+	"			int bodyOffset = offsetSplitBodies[i];\n"
+	"			int count = bodyCount[i];\n"
+	"			if (count)\n"
+	"			{\n"
+	"				gBodies[i].m_linVel += deltaLinearVelocities[bodyOffset];\n"
+	"				gBodies[i].m_angVel += deltaAngularVelocities[bodyOffset];\n"
+	"			}\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"void setConstraint4( const float4 posA, const float4 linVelA, const float4 angVelA, float invMassA, const Matrix3x3 invInertiaA,\n"
+	"	const float4 posB, const float4 linVelB, const float4 angVelB, float invMassB, const Matrix3x3 invInertiaB, \n"
+	"	__global struct b3Contact4Data* src, float dt, float positionDrift, float positionConstraintCoeff,float countA, float countB,\n"
+	"	Constraint4* dstC )\n"
+	"{\n"
+	"	dstC->m_bodyA = abs(src->m_bodyAPtrAndSignBit);\n"
+	"	dstC->m_bodyB = abs(src->m_bodyBPtrAndSignBit);\n"
+	"	float dtInv = 1.f/dt;\n"
+	"	for(int ic=0; ic<4; ic++)\n"
+	"	{\n"
+	"		dstC->m_appliedRambdaDt[ic] = 0.f;\n"
+	"	}\n"
+	"	dstC->m_fJacCoeffInv[0] = dstC->m_fJacCoeffInv[1] = 0.f;\n"
+	"	dstC->m_linear = src->m_worldNormalOnB;\n"
+	"	dstC->m_linear.w = 0.7f ;//src->getFrictionCoeff() );\n"
+	"	for(int ic=0; ic<4; ic++)\n"
+	"	{\n"
+	"		float4 r0 = src->m_worldPosB[ic] - posA;\n"
+	"		float4 r1 = src->m_worldPosB[ic] - posB;\n"
+	"		if( ic >= src->m_worldNormalOnB.w )//npoints\n"
+	"		{\n"
+	"			dstC->m_jacCoeffInv[ic] = 0.f;\n"
+	"			continue;\n"
+	"		}\n"
+	"		float relVelN;\n"
+	"		{\n"
+	"			float4 linear, angular0, angular1;\n"
+	"			setLinearAndAngular(src->m_worldNormalOnB, r0, r1, &linear, &angular0, &angular1);\n"
+	"			dstC->m_jacCoeffInv[ic] = calcJacCoeff(linear, -linear, angular0, angular1,\n"
+	"				invMassA, &invInertiaA, invMassB, &invInertiaB , countA, countB);\n"
+	"			relVelN = calcRelVel(linear, -linear, angular0, angular1,\n"
+	"				linVelA, angVelA, linVelB, angVelB);\n"
+	"			float e = 0.f;//src->getRestituitionCoeff();\n"
+	"			if( relVelN*relVelN < 0.004f ) e = 0.f;\n"
+	"			dstC->m_b[ic] = e*relVelN;\n"
+	"			//float penetration = src->m_worldPosB[ic].w;\n"
+	"			dstC->m_b[ic] += (src->m_worldPosB[ic].w + positionDrift)*positionConstraintCoeff*dtInv;\n"
+	"			dstC->m_appliedRambdaDt[ic] = 0.f;\n"
+	"		}\n"
+	"	}\n"
+	"	if( src->m_worldNormalOnB.w > 0 )//npoints\n"
+	"	{	//	prepare friction\n"
+	"		float4 center = make_float4(0.f);\n"
+	"		for(int i=0; i<src->m_worldNormalOnB.w; i++) \n"
+	"			center += src->m_worldPosB[i];\n"
+	"		center /= (float)src->m_worldNormalOnB.w;\n"
+	"		float4 tangent[2];\n"
+	"		btPlaneSpace1(-src->m_worldNormalOnB,&tangent[0],&tangent[1]);\n"
+	"		\n"
+	"		float4 r[2];\n"
+	"		r[0] = center - posA;\n"
+	"		r[1] = center - posB;\n"
+	"		for(int i=0; i<2; i++)\n"
+	"		{\n"
+	"			float4 linear, angular0, angular1;\n"
+	"			setLinearAndAngular(tangent[i], r[0], r[1], &linear, &angular0, &angular1);\n"
+	"			dstC->m_fJacCoeffInv[i] = calcJacCoeff(linear, -linear, angular0, angular1,\n"
+	"				invMassA, &invInertiaA, invMassB, &invInertiaB ,countA, countB);\n"
+	"			dstC->m_fAppliedRambdaDt[i] = 0.f;\n"
+	"		}\n"
+	"		dstC->m_center = center;\n"
+	"	}\n"
+	"	for(int i=0; i<4; i++)\n"
+	"	{\n"
+	"		if( i<src->m_worldNormalOnB.w )\n"
+	"		{\n"
+	"			dstC->m_worldPos[i] = src->m_worldPosB[i];\n"
+	"		}\n"
+	"		else\n"
+	"		{\n"
+	"			dstC->m_worldPos[i] = make_float4(0.f);\n"
+	"		}\n"
+	"	}\n"
+	"}\n"
+	"__kernel\n"
+	"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
+	"void ContactToConstraintSplitKernel(__global const struct b3Contact4Data* gContact, __global const Body* gBodies, __global const Shape* gShapes, __global Constraint4* gConstraintOut, \n"
+	"__global const unsigned int* bodyCount,\n"
+	"int nContacts,\n"
+	"float dt,\n"
+	"float positionDrift,\n"
+	"float positionConstraintCoeff\n"
+	")\n"
+	"{\n"
+	"	int gIdx = GET_GLOBAL_IDX;\n"
+	"	\n"
+	"	if( gIdx < nContacts )\n"
+	"	{\n"
+	"		int aIdx = abs(gContact[gIdx].m_bodyAPtrAndSignBit);\n"
+	"		int bIdx = abs(gContact[gIdx].m_bodyBPtrAndSignBit);\n"
+	"		float4 posA = gBodies[aIdx].m_pos;\n"
+	"		float4 linVelA = gBodies[aIdx].m_linVel;\n"
+	"		float4 angVelA = gBodies[aIdx].m_angVel;\n"
+	"		float invMassA = gBodies[aIdx].m_invMass;\n"
+	"		Matrix3x3 invInertiaA = gShapes[aIdx].m_invInertia;\n"
+	"		float4 posB = gBodies[bIdx].m_pos;\n"
+	"		float4 linVelB = gBodies[bIdx].m_linVel;\n"
+	"		float4 angVelB = gBodies[bIdx].m_angVel;\n"
+	"		float invMassB = gBodies[bIdx].m_invMass;\n"
+	"		Matrix3x3 invInertiaB = gShapes[bIdx].m_invInertia;\n"
+	"		Constraint4 cs;\n"
+	"		float countA = invMassA != 0.f ? (float)bodyCount[aIdx] : 1;\n"
+	"		float countB = invMassB != 0.f ? (float)bodyCount[bIdx] : 1;\n"
+	"    	setConstraint4( posA, linVelA, angVelA, invMassA, invInertiaA, posB, linVelB, angVelB, invMassB, invInertiaB,\n"
+	"			&gContact[gIdx], dt, positionDrift, positionConstraintCoeff,countA,countB,\n"
+	"			&cs  );\n"
+	"		\n"
+	"		cs.m_batchIdx = gContact[gIdx].m_batchIdx;\n"
+	"		gConstraintOut[gIdx] = cs;\n"
+	"	}\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/updateAabbsKernel.h b/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/updateAabbsKernel.h
index d70e74017aa..bb949b20275 100644
--- a/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/updateAabbsKernel.h
+++ b/thirdparty/bullet/Bullet3OpenCL/RigidBody/kernels/updateAabbsKernel.h
@@ -1,483 +1,482 @@
 //this file is autogenerated using stringify.bat (premake --stringify) in the build folder of this project
-static const char* updateAabbsKernelCL= \
-"#ifndef B3_UPDATE_AABBS_H\n"
-"#define B3_UPDATE_AABBS_H\n"
-"#ifndef B3_AABB_H\n"
-"#define B3_AABB_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#define B3_FLOAT4_H\n"
-"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
-"#define B3_PLATFORM_DEFINITIONS_H\n"
-"struct MyTest\n"
-"{\n"
-"	int bla;\n"
-"};\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
-"#define B3_LARGE_FLOAT 1e18f\n"
-"#define B3_INFINITY 1e18f\n"
-"#define b3Assert(a)\n"
-"#define b3ConstArray(a) __global const a*\n"
-"#define b3AtomicInc atomic_inc\n"
-"#define b3AtomicAdd atomic_add\n"
-"#define b3Fabs fabs\n"
-"#define b3Sqrt native_sqrt\n"
-"#define b3Sin native_sin\n"
-"#define b3Cos native_cos\n"
-"#define B3_STATIC\n"
-"#endif\n"
-"#endif\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"	typedef float4	b3Float4;\n"
-"	#define b3Float4ConstArg const b3Float4\n"
-"	#define b3MakeFloat4 (float4)\n"
-"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return dot(a1, b1);\n"
-"	}\n"
-"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
-"	{\n"
-"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
-"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
-"		return cross(a1, b1);\n"
-"	}\n"
-"	#define b3MinFloat4 min\n"
-"	#define b3MaxFloat4 max\n"
-"	#define b3Normalized(a) normalize(a)\n"
-"#endif \n"
-"		\n"
-"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
-"{\n"
-"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
-"		return false;\n"
-"	return true;\n"
-"}\n"
-"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
-"{\n"
-"    float maxDot = -B3_INFINITY;\n"
-"    int i = 0;\n"
-"    int ptIndex = -1;\n"
-"    for( i = 0; i < vecLen; i++ )\n"
-"    {\n"
-"        float dot = b3Dot3F4(vecArray[i],vec);\n"
-"            \n"
-"        if( dot > maxDot )\n"
-"        {\n"
-"            maxDot = dot;\n"
-"            ptIndex = i;\n"
-"        }\n"
-"    }\n"
-"	b3Assert(ptIndex>=0);\n"
-"    if (ptIndex<0)\n"
-"	{\n"
-"		ptIndex = 0;\n"
-"	}\n"
-"    *dotOut = maxDot;\n"
-"    return ptIndex;\n"
-"}\n"
-"#endif //B3_FLOAT4_H\n"
-"#ifndef B3_MAT3x3_H\n"
-"#define B3_MAT3x3_H\n"
-"#ifndef B3_QUAT_H\n"
-"#define B3_QUAT_H\n"
-"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif\n"
-"#endif\n"
-"#ifndef B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"	typedef float4	b3Quat;\n"
-"	#define b3QuatConstArg const b3Quat\n"
-"	\n"
-"	\n"
-"inline float4 b3FastNormalize4(float4 v)\n"
-"{\n"
-"	v = (float4)(v.xyz,0.f);\n"
-"	return fast_normalize(v);\n"
-"}\n"
-"	\n"
-"inline b3Quat b3QuatMul(b3Quat a, b3Quat b);\n"
-"inline b3Quat b3QuatNormalized(b3QuatConstArg in);\n"
-"inline b3Quat b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec);\n"
-"inline b3Quat b3QuatInvert(b3QuatConstArg q);\n"
-"inline b3Quat b3QuatInverse(b3QuatConstArg q);\n"
-"inline b3Quat b3QuatMul(b3QuatConstArg a, b3QuatConstArg b)\n"
-"{\n"
-"	b3Quat ans;\n"
-"	ans = b3Cross3( a, b );\n"
-"	ans += a.w*b+b.w*a;\n"
-"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
-"	ans.w = a.w*b.w - b3Dot3F4(a, b);\n"
-"	return ans;\n"
-"}\n"
-"inline b3Quat b3QuatNormalized(b3QuatConstArg in)\n"
-"{\n"
-"	b3Quat q;\n"
-"	q=in;\n"
-"	//return b3FastNormalize4(in);\n"
-"	float len = native_sqrt(dot(q, q));\n"
-"	if(len > 0.f)\n"
-"	{\n"
-"		q *= 1.f / len;\n"
-"	}\n"
-"	else\n"
-"	{\n"
-"		q.x = q.y = q.z = 0.f;\n"
-"		q.w = 1.f;\n"
-"	}\n"
-"	return q;\n"
-"}\n"
-"inline float4 b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
-"{\n"
-"	b3Quat qInv = b3QuatInvert( q );\n"
-"	float4 vcpy = vec;\n"
-"	vcpy.w = 0.f;\n"
-"	float4 out = b3QuatMul(b3QuatMul(q,vcpy),qInv);\n"
-"	return out;\n"
-"}\n"
-"inline b3Quat b3QuatInverse(b3QuatConstArg q)\n"
-"{\n"
-"	return (b3Quat)(-q.xyz, q.w);\n"
-"}\n"
-"inline b3Quat b3QuatInvert(b3QuatConstArg q)\n"
-"{\n"
-"	return (b3Quat)(-q.xyz, q.w);\n"
-"}\n"
-"inline float4 b3QuatInvRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
-"{\n"
-"	return b3QuatRotate( b3QuatInvert( q ), vec );\n"
-"}\n"
-"inline b3Float4 b3TransformPoint(b3Float4ConstArg point, b3Float4ConstArg translation, b3QuatConstArg  orientation)\n"
-"{\n"
-"	return b3QuatRotate( orientation, point ) + (translation);\n"
-"}\n"
-"	\n"
-"#endif \n"
-"#endif //B3_QUAT_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"typedef struct\n"
-"{\n"
-"	b3Float4 m_row[3];\n"
-"}b3Mat3x3;\n"
-"#define b3Mat3x3ConstArg const b3Mat3x3\n"
-"#define b3GetRow(m,row) (m.m_row[row])\n"
-"inline b3Mat3x3 b3QuatGetRotationMatrix(b3Quat quat)\n"
-"{\n"
-"	b3Float4 quat2 = (b3Float4)(quat.x*quat.x, quat.y*quat.y, quat.z*quat.z, 0.f);\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0].x=1-2*quat2.y-2*quat2.z;\n"
-"	out.m_row[0].y=2*quat.x*quat.y-2*quat.w*quat.z;\n"
-"	out.m_row[0].z=2*quat.x*quat.z+2*quat.w*quat.y;\n"
-"	out.m_row[0].w = 0.f;\n"
-"	out.m_row[1].x=2*quat.x*quat.y+2*quat.w*quat.z;\n"
-"	out.m_row[1].y=1-2*quat2.x-2*quat2.z;\n"
-"	out.m_row[1].z=2*quat.y*quat.z-2*quat.w*quat.x;\n"
-"	out.m_row[1].w = 0.f;\n"
-"	out.m_row[2].x=2*quat.x*quat.z-2*quat.w*quat.y;\n"
-"	out.m_row[2].y=2*quat.y*quat.z+2*quat.w*quat.x;\n"
-"	out.m_row[2].z=1-2*quat2.x-2*quat2.y;\n"
-"	out.m_row[2].w = 0.f;\n"
-"	return out;\n"
-"}\n"
-"inline b3Mat3x3 b3AbsoluteMat3x3(b3Mat3x3ConstArg matIn)\n"
-"{\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0] = fabs(matIn.m_row[0]);\n"
-"	out.m_row[1] = fabs(matIn.m_row[1]);\n"
-"	out.m_row[2] = fabs(matIn.m_row[2]);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtZero();\n"
-"__inline\n"
-"b3Mat3x3 mtIdentity();\n"
-"__inline\n"
-"b3Mat3x3 mtTranspose(b3Mat3x3 m);\n"
-"__inline\n"
-"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b);\n"
-"__inline\n"
-"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b);\n"
-"__inline\n"
-"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b);\n"
-"__inline\n"
-"b3Mat3x3 mtZero()\n"
-"{\n"
-"	b3Mat3x3 m;\n"
-"	m.m_row[0] = (b3Float4)(0.f);\n"
-"	m.m_row[1] = (b3Float4)(0.f);\n"
-"	m.m_row[2] = (b3Float4)(0.f);\n"
-"	return m;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtIdentity()\n"
-"{\n"
-"	b3Mat3x3 m;\n"
-"	m.m_row[0] = (b3Float4)(1,0,0,0);\n"
-"	m.m_row[1] = (b3Float4)(0,1,0,0);\n"
-"	m.m_row[2] = (b3Float4)(0,0,1,0);\n"
-"	return m;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtTranspose(b3Mat3x3 m)\n"
-"{\n"
-"	b3Mat3x3 out;\n"
-"	out.m_row[0] = (b3Float4)(m.m_row[0].x, m.m_row[1].x, m.m_row[2].x, 0.f);\n"
-"	out.m_row[1] = (b3Float4)(m.m_row[0].y, m.m_row[1].y, m.m_row[2].y, 0.f);\n"
-"	out.m_row[2] = (b3Float4)(m.m_row[0].z, m.m_row[1].z, m.m_row[2].z, 0.f);\n"
-"	return out;\n"
-"}\n"
-"__inline\n"
-"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b)\n"
-"{\n"
-"	b3Mat3x3 transB;\n"
-"	transB = mtTranspose( b );\n"
-"	b3Mat3x3 ans;\n"
-"	//	why this doesn't run when 0ing in the for{}\n"
-"	a.m_row[0].w = 0.f;\n"
-"	a.m_row[1].w = 0.f;\n"
-"	a.m_row[2].w = 0.f;\n"
-"	for(int i=0; i<3; i++)\n"
-"	{\n"
-"//	a.m_row[i].w = 0.f;\n"
-"		ans.m_row[i].x = b3Dot3F4(a.m_row[i],transB.m_row[0]);\n"
-"		ans.m_row[i].y = b3Dot3F4(a.m_row[i],transB.m_row[1]);\n"
-"		ans.m_row[i].z = b3Dot3F4(a.m_row[i],transB.m_row[2]);\n"
-"		ans.m_row[i].w = 0.f;\n"
-"	}\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b)\n"
-"{\n"
-"	b3Float4 ans;\n"
-"	ans.x = b3Dot3F4( a.m_row[0], b );\n"
-"	ans.y = b3Dot3F4( a.m_row[1], b );\n"
-"	ans.z = b3Dot3F4( a.m_row[2], b );\n"
-"	ans.w = 0.f;\n"
-"	return ans;\n"
-"}\n"
-"__inline\n"
-"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b)\n"
-"{\n"
-"	b3Float4 colx = b3MakeFloat4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
-"	b3Float4 coly = b3MakeFloat4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
-"	b3Float4 colz = b3MakeFloat4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
-"	b3Float4 ans;\n"
-"	ans.x = b3Dot3F4( a, colx );\n"
-"	ans.y = b3Dot3F4( a, coly );\n"
-"	ans.z = b3Dot3F4( a, colz );\n"
-"	return ans;\n"
-"}\n"
-"#endif\n"
-"#endif //B3_MAT3x3_H\n"
-"typedef struct b3Aabb b3Aabb_t;\n"
-"struct b3Aabb\n"
-"{\n"
-"	union\n"
-"	{\n"
-"		float m_min[4];\n"
-"		b3Float4 m_minVec;\n"
-"		int m_minIndices[4];\n"
-"	};\n"
-"	union\n"
-"	{\n"
-"		float	m_max[4];\n"
-"		b3Float4 m_maxVec;\n"
-"		int m_signedMaxIndices[4];\n"
-"	};\n"
-"};\n"
-"inline void b3TransformAabb2(b3Float4ConstArg localAabbMin,b3Float4ConstArg localAabbMax, float margin,\n"
-"						b3Float4ConstArg pos,\n"
-"						b3QuatConstArg orn,\n"
-"						b3Float4* aabbMinOut,b3Float4* aabbMaxOut)\n"
-"{\n"
-"		b3Float4 localHalfExtents = 0.5f*(localAabbMax-localAabbMin);\n"
-"		localHalfExtents+=b3MakeFloat4(margin,margin,margin,0.f);\n"
-"		b3Float4 localCenter = 0.5f*(localAabbMax+localAabbMin);\n"
-"		b3Mat3x3 m;\n"
-"		m = b3QuatGetRotationMatrix(orn);\n"
-"		b3Mat3x3 abs_b = b3AbsoluteMat3x3(m);\n"
-"		b3Float4 center = b3TransformPoint(localCenter,pos,orn);\n"
-"		\n"
-"		b3Float4 extent = b3MakeFloat4(b3Dot3F4(localHalfExtents,b3GetRow(abs_b,0)),\n"
-"										 b3Dot3F4(localHalfExtents,b3GetRow(abs_b,1)),\n"
-"										 b3Dot3F4(localHalfExtents,b3GetRow(abs_b,2)),\n"
-"										 0.f);\n"
-"		*aabbMinOut = center-extent;\n"
-"		*aabbMaxOut = center+extent;\n"
-"}\n"
-"/// conservative test for overlap between two aabbs\n"
-"inline bool b3TestAabbAgainstAabb(b3Float4ConstArg aabbMin1,b3Float4ConstArg aabbMax1,\n"
-"								b3Float4ConstArg aabbMin2, b3Float4ConstArg aabbMax2)\n"
-"{\n"
-"	bool overlap = true;\n"
-"	overlap = (aabbMin1.x > aabbMax2.x || aabbMax1.x < aabbMin2.x) ? false : overlap;\n"
-"	overlap = (aabbMin1.z > aabbMax2.z || aabbMax1.z < aabbMin2.z) ? false : overlap;\n"
-"	overlap = (aabbMin1.y > aabbMax2.y || aabbMax1.y < aabbMin2.y) ? false : overlap;\n"
-"	return overlap;\n"
-"}\n"
-"#endif //B3_AABB_H\n"
-"#ifndef B3_COLLIDABLE_H\n"
-"#define B3_COLLIDABLE_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_FLOAT4_H\n"
-"#ifndef B3_QUAT_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_QUAT_H\n"
-"enum b3ShapeTypes\n"
-"{\n"
-"	SHAPE_HEIGHT_FIELD=1,\n"
-"	SHAPE_CONVEX_HULL=3,\n"
-"	SHAPE_PLANE=4,\n"
-"	SHAPE_CONCAVE_TRIMESH=5,\n"
-"	SHAPE_COMPOUND_OF_CONVEX_HULLS=6,\n"
-"	SHAPE_SPHERE=7,\n"
-"	MAX_NUM_SHAPE_TYPES,\n"
-"};\n"
-"typedef struct b3Collidable b3Collidable_t;\n"
-"struct b3Collidable\n"
-"{\n"
-"	union {\n"
-"		int m_numChildShapes;\n"
-"		int m_bvhIndex;\n"
-"	};\n"
-"	union\n"
-"	{\n"
-"		float m_radius;\n"
-"		int	m_compoundBvhIndex;\n"
-"	};\n"
-"	int m_shapeType;\n"
-"	union\n"
-"	{\n"
-"		int m_shapeIndex;\n"
-"		float m_height;\n"
-"	};\n"
-"};\n"
-"typedef struct b3GpuChildShape b3GpuChildShape_t;\n"
-"struct b3GpuChildShape\n"
-"{\n"
-"	b3Float4	m_childPosition;\n"
-"	b3Quat		m_childOrientation;\n"
-"	union\n"
-"	{\n"
-"		int			m_shapeIndex;//used for SHAPE_COMPOUND_OF_CONVEX_HULLS\n"
-"		int			m_capsuleAxis;\n"
-"	};\n"
-"	union \n"
-"	{\n"
-"		float		m_radius;//used for childshape of SHAPE_COMPOUND_OF_SPHERES or SHAPE_COMPOUND_OF_CAPSULES\n"
-"		int			m_numChildShapes;//used for compound shape\n"
-"	};\n"
-"	union \n"
-"	{\n"
-"		float		m_height;//used for childshape of SHAPE_COMPOUND_OF_CAPSULES\n"
-"		int	m_collidableShapeIndex;\n"
-"	};\n"
-"	int			m_shapeType;\n"
-"};\n"
-"struct b3CompoundOverlappingPair\n"
-"{\n"
-"	int m_bodyIndexA;\n"
-"	int m_bodyIndexB;\n"
-"//	int	m_pairType;\n"
-"	int m_childShapeIndexA;\n"
-"	int m_childShapeIndexB;\n"
-"};\n"
-"#endif //B3_COLLIDABLE_H\n"
-"#ifndef B3_RIGIDBODY_DATA_H\n"
-"#define B3_RIGIDBODY_DATA_H\n"
-"#ifndef B3_FLOAT4_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_FLOAT4_H\n"
-"#ifndef B3_QUAT_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif \n"
-"#endif //B3_QUAT_H\n"
-"#ifndef B3_MAT3x3_H\n"
-"#ifdef __cplusplus\n"
-"#else\n"
-"#endif\n"
-"#endif //B3_MAT3x3_H\n"
-"typedef struct b3RigidBodyData b3RigidBodyData_t;\n"
-"struct b3RigidBodyData\n"
-"{\n"
-"	b3Float4				m_pos;\n"
-"	b3Quat					m_quat;\n"
-"	b3Float4				m_linVel;\n"
-"	b3Float4				m_angVel;\n"
-"	int 					m_collidableIdx;\n"
-"	float 				m_invMass;\n"
-"	float 				m_restituitionCoeff;\n"
-"	float 				m_frictionCoeff;\n"
-"};\n"
-"typedef struct b3InertiaData b3InertiaData_t;\n"
-"struct b3InertiaData\n"
-"{\n"
-"	b3Mat3x3 m_invInertiaWorld;\n"
-"	b3Mat3x3 m_initInvInertia;\n"
-"};\n"
-"#endif //B3_RIGIDBODY_DATA_H\n"
-"	\n"
-"void b3ComputeWorldAabb(  int bodyId, __global const b3RigidBodyData_t* bodies, __global const  b3Collidable_t* collidables, __global const  b3Aabb_t* localShapeAABB, __global b3Aabb_t* worldAabbs)\n"
-"{\n"
-"	__global const b3RigidBodyData_t* body = &bodies[bodyId];\n"
-"	b3Float4 position = body->m_pos;\n"
-"	b3Quat	orientation = body->m_quat;\n"
-"	\n"
-"	int collidableIndex = body->m_collidableIdx;\n"
-"	int shapeIndex = collidables[collidableIndex].m_shapeIndex;\n"
-"		\n"
-"	if (shapeIndex>=0)\n"
-"	{\n"
-"				\n"
-"		b3Aabb_t localAabb = localShapeAABB[collidableIndex];\n"
-"		b3Aabb_t worldAabb;\n"
-"		\n"
-"		b3Float4 aabbAMinOut,aabbAMaxOut;	\n"
-"		float margin = 0.f;\n"
-"		b3TransformAabb2(localAabb.m_minVec,localAabb.m_maxVec,margin,position,orientation,&aabbAMinOut,&aabbAMaxOut);\n"
-"		\n"
-"		worldAabb.m_minVec =aabbAMinOut;\n"
-"		worldAabb.m_minIndices[3] = bodyId;\n"
-"		worldAabb.m_maxVec = aabbAMaxOut;\n"
-"		worldAabb.m_signedMaxIndices[3] = body[bodyId].m_invMass==0.f? 0 : 1;\n"
-"		worldAabbs[bodyId] = worldAabb;\n"
-"	}\n"
-"}\n"
-"#endif //B3_UPDATE_AABBS_H\n"
-"__kernel void initializeGpuAabbsFull(  const int numNodes, __global b3RigidBodyData_t* gBodies,__global b3Collidable_t* collidables, __global b3Aabb_t* plocalShapeAABB, __global b3Aabb_t* pAABB)\n"
-"{\n"
-"	int nodeID = get_global_id(0);\n"
-"	if( nodeID < numNodes )\n"
-"	{\n"
-"		b3ComputeWorldAabb(nodeID, gBodies, collidables, plocalShapeAABB,pAABB);\n"
-"	}\n"
-"}\n"
-"__kernel void clearOverlappingPairsKernel(  __global int4* pairs, int numPairs)\n"
-"{\n"
-"	int pairId = get_global_id(0);\n"
-"	if( pairId< numPairs )\n"
-"	{\n"
-"		pairs[pairId].z = 0xffffffff;\n"
-"	}\n"
-"}\n"
-;
+static const char* updateAabbsKernelCL =
+	"#ifndef B3_UPDATE_AABBS_H\n"
+	"#define B3_UPDATE_AABBS_H\n"
+	"#ifndef B3_AABB_H\n"
+	"#define B3_AABB_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#define B3_FLOAT4_H\n"
+	"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
+	"#define B3_PLATFORM_DEFINITIONS_H\n"
+	"struct MyTest\n"
+	"{\n"
+	"	int bla;\n"
+	"};\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"//keep B3_LARGE_FLOAT*B3_LARGE_FLOAT < FLT_MAX\n"
+	"#define B3_LARGE_FLOAT 1e18f\n"
+	"#define B3_INFINITY 1e18f\n"
+	"#define b3Assert(a)\n"
+	"#define b3ConstArray(a) __global const a*\n"
+	"#define b3AtomicInc atomic_inc\n"
+	"#define b3AtomicAdd atomic_add\n"
+	"#define b3Fabs fabs\n"
+	"#define b3Sqrt native_sqrt\n"
+	"#define b3Sin native_sin\n"
+	"#define b3Cos native_cos\n"
+	"#define B3_STATIC\n"
+	"#endif\n"
+	"#endif\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"	typedef float4	b3Float4;\n"
+	"	#define b3Float4ConstArg const b3Float4\n"
+	"	#define b3MakeFloat4 (float4)\n"
+	"	float b3Dot3F4(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return dot(a1, b1);\n"
+	"	}\n"
+	"	b3Float4 b3Cross3(b3Float4ConstArg v0,b3Float4ConstArg v1)\n"
+	"	{\n"
+	"		float4 a1 = b3MakeFloat4(v0.xyz,0.f);\n"
+	"		float4 b1 = b3MakeFloat4(v1.xyz,0.f);\n"
+	"		return cross(a1, b1);\n"
+	"	}\n"
+	"	#define b3MinFloat4 min\n"
+	"	#define b3MaxFloat4 max\n"
+	"	#define b3Normalized(a) normalize(a)\n"
+	"#endif \n"
+	"		\n"
+	"inline bool b3IsAlmostZero(b3Float4ConstArg v)\n"
+	"{\n"
+	"	if(b3Fabs(v.x)>1e-6 || b3Fabs(v.y)>1e-6 || b3Fabs(v.z)>1e-6)	\n"
+	"		return false;\n"
+	"	return true;\n"
+	"}\n"
+	"inline int    b3MaxDot( b3Float4ConstArg vec, __global const b3Float4* vecArray, int vecLen, float* dotOut )\n"
+	"{\n"
+	"    float maxDot = -B3_INFINITY;\n"
+	"    int i = 0;\n"
+	"    int ptIndex = -1;\n"
+	"    for( i = 0; i < vecLen; i++ )\n"
+	"    {\n"
+	"        float dot = b3Dot3F4(vecArray[i],vec);\n"
+	"            \n"
+	"        if( dot > maxDot )\n"
+	"        {\n"
+	"            maxDot = dot;\n"
+	"            ptIndex = i;\n"
+	"        }\n"
+	"    }\n"
+	"	b3Assert(ptIndex>=0);\n"
+	"    if (ptIndex<0)\n"
+	"	{\n"
+	"		ptIndex = 0;\n"
+	"	}\n"
+	"    *dotOut = maxDot;\n"
+	"    return ptIndex;\n"
+	"}\n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifndef B3_MAT3x3_H\n"
+	"#define B3_MAT3x3_H\n"
+	"#ifndef B3_QUAT_H\n"
+	"#define B3_QUAT_H\n"
+	"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif\n"
+	"#endif\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"	typedef float4	b3Quat;\n"
+	"	#define b3QuatConstArg const b3Quat\n"
+	"	\n"
+	"	\n"
+	"inline float4 b3FastNormalize4(float4 v)\n"
+	"{\n"
+	"	v = (float4)(v.xyz,0.f);\n"
+	"	return fast_normalize(v);\n"
+	"}\n"
+	"	\n"
+	"inline b3Quat b3QuatMul(b3Quat a, b3Quat b);\n"
+	"inline b3Quat b3QuatNormalized(b3QuatConstArg in);\n"
+	"inline b3Quat b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec);\n"
+	"inline b3Quat b3QuatInvert(b3QuatConstArg q);\n"
+	"inline b3Quat b3QuatInverse(b3QuatConstArg q);\n"
+	"inline b3Quat b3QuatMul(b3QuatConstArg a, b3QuatConstArg b)\n"
+	"{\n"
+	"	b3Quat ans;\n"
+	"	ans = b3Cross3( a, b );\n"
+	"	ans += a.w*b+b.w*a;\n"
+	"//	ans.w = a.w*b.w - (a.x*b.x+a.y*b.y+a.z*b.z);\n"
+	"	ans.w = a.w*b.w - b3Dot3F4(a, b);\n"
+	"	return ans;\n"
+	"}\n"
+	"inline b3Quat b3QuatNormalized(b3QuatConstArg in)\n"
+	"{\n"
+	"	b3Quat q;\n"
+	"	q=in;\n"
+	"	//return b3FastNormalize4(in);\n"
+	"	float len = native_sqrt(dot(q, q));\n"
+	"	if(len > 0.f)\n"
+	"	{\n"
+	"		q *= 1.f / len;\n"
+	"	}\n"
+	"	else\n"
+	"	{\n"
+	"		q.x = q.y = q.z = 0.f;\n"
+	"		q.w = 1.f;\n"
+	"	}\n"
+	"	return q;\n"
+	"}\n"
+	"inline float4 b3QuatRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
+	"{\n"
+	"	b3Quat qInv = b3QuatInvert( q );\n"
+	"	float4 vcpy = vec;\n"
+	"	vcpy.w = 0.f;\n"
+	"	float4 out = b3QuatMul(b3QuatMul(q,vcpy),qInv);\n"
+	"	return out;\n"
+	"}\n"
+	"inline b3Quat b3QuatInverse(b3QuatConstArg q)\n"
+	"{\n"
+	"	return (b3Quat)(-q.xyz, q.w);\n"
+	"}\n"
+	"inline b3Quat b3QuatInvert(b3QuatConstArg q)\n"
+	"{\n"
+	"	return (b3Quat)(-q.xyz, q.w);\n"
+	"}\n"
+	"inline float4 b3QuatInvRotate(b3QuatConstArg q, b3QuatConstArg vec)\n"
+	"{\n"
+	"	return b3QuatRotate( b3QuatInvert( q ), vec );\n"
+	"}\n"
+	"inline b3Float4 b3TransformPoint(b3Float4ConstArg point, b3Float4ConstArg translation, b3QuatConstArg  orientation)\n"
+	"{\n"
+	"	return b3QuatRotate( orientation, point ) + (translation);\n"
+	"}\n"
+	"	\n"
+	"#endif \n"
+	"#endif //B3_QUAT_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"typedef struct\n"
+	"{\n"
+	"	b3Float4 m_row[3];\n"
+	"}b3Mat3x3;\n"
+	"#define b3Mat3x3ConstArg const b3Mat3x3\n"
+	"#define b3GetRow(m,row) (m.m_row[row])\n"
+	"inline b3Mat3x3 b3QuatGetRotationMatrix(b3Quat quat)\n"
+	"{\n"
+	"	b3Float4 quat2 = (b3Float4)(quat.x*quat.x, quat.y*quat.y, quat.z*quat.z, 0.f);\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0].x=1-2*quat2.y-2*quat2.z;\n"
+	"	out.m_row[0].y=2*quat.x*quat.y-2*quat.w*quat.z;\n"
+	"	out.m_row[0].z=2*quat.x*quat.z+2*quat.w*quat.y;\n"
+	"	out.m_row[0].w = 0.f;\n"
+	"	out.m_row[1].x=2*quat.x*quat.y+2*quat.w*quat.z;\n"
+	"	out.m_row[1].y=1-2*quat2.x-2*quat2.z;\n"
+	"	out.m_row[1].z=2*quat.y*quat.z-2*quat.w*quat.x;\n"
+	"	out.m_row[1].w = 0.f;\n"
+	"	out.m_row[2].x=2*quat.x*quat.z-2*quat.w*quat.y;\n"
+	"	out.m_row[2].y=2*quat.y*quat.z+2*quat.w*quat.x;\n"
+	"	out.m_row[2].z=1-2*quat2.x-2*quat2.y;\n"
+	"	out.m_row[2].w = 0.f;\n"
+	"	return out;\n"
+	"}\n"
+	"inline b3Mat3x3 b3AbsoluteMat3x3(b3Mat3x3ConstArg matIn)\n"
+	"{\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0] = fabs(matIn.m_row[0]);\n"
+	"	out.m_row[1] = fabs(matIn.m_row[1]);\n"
+	"	out.m_row[2] = fabs(matIn.m_row[2]);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtZero();\n"
+	"__inline\n"
+	"b3Mat3x3 mtIdentity();\n"
+	"__inline\n"
+	"b3Mat3x3 mtTranspose(b3Mat3x3 m);\n"
+	"__inline\n"
+	"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b);\n"
+	"__inline\n"
+	"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b);\n"
+	"__inline\n"
+	"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b);\n"
+	"__inline\n"
+	"b3Mat3x3 mtZero()\n"
+	"{\n"
+	"	b3Mat3x3 m;\n"
+	"	m.m_row[0] = (b3Float4)(0.f);\n"
+	"	m.m_row[1] = (b3Float4)(0.f);\n"
+	"	m.m_row[2] = (b3Float4)(0.f);\n"
+	"	return m;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtIdentity()\n"
+	"{\n"
+	"	b3Mat3x3 m;\n"
+	"	m.m_row[0] = (b3Float4)(1,0,0,0);\n"
+	"	m.m_row[1] = (b3Float4)(0,1,0,0);\n"
+	"	m.m_row[2] = (b3Float4)(0,0,1,0);\n"
+	"	return m;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtTranspose(b3Mat3x3 m)\n"
+	"{\n"
+	"	b3Mat3x3 out;\n"
+	"	out.m_row[0] = (b3Float4)(m.m_row[0].x, m.m_row[1].x, m.m_row[2].x, 0.f);\n"
+	"	out.m_row[1] = (b3Float4)(m.m_row[0].y, m.m_row[1].y, m.m_row[2].y, 0.f);\n"
+	"	out.m_row[2] = (b3Float4)(m.m_row[0].z, m.m_row[1].z, m.m_row[2].z, 0.f);\n"
+	"	return out;\n"
+	"}\n"
+	"__inline\n"
+	"b3Mat3x3 mtMul(b3Mat3x3 a, b3Mat3x3 b)\n"
+	"{\n"
+	"	b3Mat3x3 transB;\n"
+	"	transB = mtTranspose( b );\n"
+	"	b3Mat3x3 ans;\n"
+	"	//	why this doesn't run when 0ing in the for{}\n"
+	"	a.m_row[0].w = 0.f;\n"
+	"	a.m_row[1].w = 0.f;\n"
+	"	a.m_row[2].w = 0.f;\n"
+	"	for(int i=0; i<3; i++)\n"
+	"	{\n"
+	"//	a.m_row[i].w = 0.f;\n"
+	"		ans.m_row[i].x = b3Dot3F4(a.m_row[i],transB.m_row[0]);\n"
+	"		ans.m_row[i].y = b3Dot3F4(a.m_row[i],transB.m_row[1]);\n"
+	"		ans.m_row[i].z = b3Dot3F4(a.m_row[i],transB.m_row[2]);\n"
+	"		ans.m_row[i].w = 0.f;\n"
+	"	}\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"b3Float4 mtMul1(b3Mat3x3 a, b3Float4 b)\n"
+	"{\n"
+	"	b3Float4 ans;\n"
+	"	ans.x = b3Dot3F4( a.m_row[0], b );\n"
+	"	ans.y = b3Dot3F4( a.m_row[1], b );\n"
+	"	ans.z = b3Dot3F4( a.m_row[2], b );\n"
+	"	ans.w = 0.f;\n"
+	"	return ans;\n"
+	"}\n"
+	"__inline\n"
+	"b3Float4 mtMul3(b3Float4 a, b3Mat3x3 b)\n"
+	"{\n"
+	"	b3Float4 colx = b3MakeFloat4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
+	"	b3Float4 coly = b3MakeFloat4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
+	"	b3Float4 colz = b3MakeFloat4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
+	"	b3Float4 ans;\n"
+	"	ans.x = b3Dot3F4( a, colx );\n"
+	"	ans.y = b3Dot3F4( a, coly );\n"
+	"	ans.z = b3Dot3F4( a, colz );\n"
+	"	return ans;\n"
+	"}\n"
+	"#endif\n"
+	"#endif //B3_MAT3x3_H\n"
+	"typedef struct b3Aabb b3Aabb_t;\n"
+	"struct b3Aabb\n"
+	"{\n"
+	"	union\n"
+	"	{\n"
+	"		float m_min[4];\n"
+	"		b3Float4 m_minVec;\n"
+	"		int m_minIndices[4];\n"
+	"	};\n"
+	"	union\n"
+	"	{\n"
+	"		float	m_max[4];\n"
+	"		b3Float4 m_maxVec;\n"
+	"		int m_signedMaxIndices[4];\n"
+	"	};\n"
+	"};\n"
+	"inline void b3TransformAabb2(b3Float4ConstArg localAabbMin,b3Float4ConstArg localAabbMax, float margin,\n"
+	"						b3Float4ConstArg pos,\n"
+	"						b3QuatConstArg orn,\n"
+	"						b3Float4* aabbMinOut,b3Float4* aabbMaxOut)\n"
+	"{\n"
+	"		b3Float4 localHalfExtents = 0.5f*(localAabbMax-localAabbMin);\n"
+	"		localHalfExtents+=b3MakeFloat4(margin,margin,margin,0.f);\n"
+	"		b3Float4 localCenter = 0.5f*(localAabbMax+localAabbMin);\n"
+	"		b3Mat3x3 m;\n"
+	"		m = b3QuatGetRotationMatrix(orn);\n"
+	"		b3Mat3x3 abs_b = b3AbsoluteMat3x3(m);\n"
+	"		b3Float4 center = b3TransformPoint(localCenter,pos,orn);\n"
+	"		\n"
+	"		b3Float4 extent = b3MakeFloat4(b3Dot3F4(localHalfExtents,b3GetRow(abs_b,0)),\n"
+	"										 b3Dot3F4(localHalfExtents,b3GetRow(abs_b,1)),\n"
+	"										 b3Dot3F4(localHalfExtents,b3GetRow(abs_b,2)),\n"
+	"										 0.f);\n"
+	"		*aabbMinOut = center-extent;\n"
+	"		*aabbMaxOut = center+extent;\n"
+	"}\n"
+	"/// conservative test for overlap between two aabbs\n"
+	"inline bool b3TestAabbAgainstAabb(b3Float4ConstArg aabbMin1,b3Float4ConstArg aabbMax1,\n"
+	"								b3Float4ConstArg aabbMin2, b3Float4ConstArg aabbMax2)\n"
+	"{\n"
+	"	bool overlap = true;\n"
+	"	overlap = (aabbMin1.x > aabbMax2.x || aabbMax1.x < aabbMin2.x) ? false : overlap;\n"
+	"	overlap = (aabbMin1.z > aabbMax2.z || aabbMax1.z < aabbMin2.z) ? false : overlap;\n"
+	"	overlap = (aabbMin1.y > aabbMax2.y || aabbMax1.y < aabbMin2.y) ? false : overlap;\n"
+	"	return overlap;\n"
+	"}\n"
+	"#endif //B3_AABB_H\n"
+	"#ifndef B3_COLLIDABLE_H\n"
+	"#define B3_COLLIDABLE_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifndef B3_QUAT_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_QUAT_H\n"
+	"enum b3ShapeTypes\n"
+	"{\n"
+	"	SHAPE_HEIGHT_FIELD=1,\n"
+	"	SHAPE_CONVEX_HULL=3,\n"
+	"	SHAPE_PLANE=4,\n"
+	"	SHAPE_CONCAVE_TRIMESH=5,\n"
+	"	SHAPE_COMPOUND_OF_CONVEX_HULLS=6,\n"
+	"	SHAPE_SPHERE=7,\n"
+	"	MAX_NUM_SHAPE_TYPES,\n"
+	"};\n"
+	"typedef struct b3Collidable b3Collidable_t;\n"
+	"struct b3Collidable\n"
+	"{\n"
+	"	union {\n"
+	"		int m_numChildShapes;\n"
+	"		int m_bvhIndex;\n"
+	"	};\n"
+	"	union\n"
+	"	{\n"
+	"		float m_radius;\n"
+	"		int	m_compoundBvhIndex;\n"
+	"	};\n"
+	"	int m_shapeType;\n"
+	"	union\n"
+	"	{\n"
+	"		int m_shapeIndex;\n"
+	"		float m_height;\n"
+	"	};\n"
+	"};\n"
+	"typedef struct b3GpuChildShape b3GpuChildShape_t;\n"
+	"struct b3GpuChildShape\n"
+	"{\n"
+	"	b3Float4	m_childPosition;\n"
+	"	b3Quat		m_childOrientation;\n"
+	"	union\n"
+	"	{\n"
+	"		int			m_shapeIndex;//used for SHAPE_COMPOUND_OF_CONVEX_HULLS\n"
+	"		int			m_capsuleAxis;\n"
+	"	};\n"
+	"	union \n"
+	"	{\n"
+	"		float		m_radius;//used for childshape of SHAPE_COMPOUND_OF_SPHERES or SHAPE_COMPOUND_OF_CAPSULES\n"
+	"		int			m_numChildShapes;//used for compound shape\n"
+	"	};\n"
+	"	union \n"
+	"	{\n"
+	"		float		m_height;//used for childshape of SHAPE_COMPOUND_OF_CAPSULES\n"
+	"		int	m_collidableShapeIndex;\n"
+	"	};\n"
+	"	int			m_shapeType;\n"
+	"};\n"
+	"struct b3CompoundOverlappingPair\n"
+	"{\n"
+	"	int m_bodyIndexA;\n"
+	"	int m_bodyIndexB;\n"
+	"//	int	m_pairType;\n"
+	"	int m_childShapeIndexA;\n"
+	"	int m_childShapeIndexB;\n"
+	"};\n"
+	"#endif //B3_COLLIDABLE_H\n"
+	"#ifndef B3_RIGIDBODY_DATA_H\n"
+	"#define B3_RIGIDBODY_DATA_H\n"
+	"#ifndef B3_FLOAT4_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_FLOAT4_H\n"
+	"#ifndef B3_QUAT_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif \n"
+	"#endif //B3_QUAT_H\n"
+	"#ifndef B3_MAT3x3_H\n"
+	"#ifdef __cplusplus\n"
+	"#else\n"
+	"#endif\n"
+	"#endif //B3_MAT3x3_H\n"
+	"typedef struct b3RigidBodyData b3RigidBodyData_t;\n"
+	"struct b3RigidBodyData\n"
+	"{\n"
+	"	b3Float4				m_pos;\n"
+	"	b3Quat					m_quat;\n"
+	"	b3Float4				m_linVel;\n"
+	"	b3Float4				m_angVel;\n"
+	"	int 					m_collidableIdx;\n"
+	"	float 				m_invMass;\n"
+	"	float 				m_restituitionCoeff;\n"
+	"	float 				m_frictionCoeff;\n"
+	"};\n"
+	"typedef struct b3InertiaData b3InertiaData_t;\n"
+	"struct b3InertiaData\n"
+	"{\n"
+	"	b3Mat3x3 m_invInertiaWorld;\n"
+	"	b3Mat3x3 m_initInvInertia;\n"
+	"};\n"
+	"#endif //B3_RIGIDBODY_DATA_H\n"
+	"	\n"
+	"void b3ComputeWorldAabb(  int bodyId, __global const b3RigidBodyData_t* bodies, __global const  b3Collidable_t* collidables, __global const  b3Aabb_t* localShapeAABB, __global b3Aabb_t* worldAabbs)\n"
+	"{\n"
+	"	__global const b3RigidBodyData_t* body = &bodies[bodyId];\n"
+	"	b3Float4 position = body->m_pos;\n"
+	"	b3Quat	orientation = body->m_quat;\n"
+	"	\n"
+	"	int collidableIndex = body->m_collidableIdx;\n"
+	"	int shapeIndex = collidables[collidableIndex].m_shapeIndex;\n"
+	"		\n"
+	"	if (shapeIndex>=0)\n"
+	"	{\n"
+	"				\n"
+	"		b3Aabb_t localAabb = localShapeAABB[collidableIndex];\n"
+	"		b3Aabb_t worldAabb;\n"
+	"		\n"
+	"		b3Float4 aabbAMinOut,aabbAMaxOut;	\n"
+	"		float margin = 0.f;\n"
+	"		b3TransformAabb2(localAabb.m_minVec,localAabb.m_maxVec,margin,position,orientation,&aabbAMinOut,&aabbAMaxOut);\n"
+	"		\n"
+	"		worldAabb.m_minVec =aabbAMinOut;\n"
+	"		worldAabb.m_minIndices[3] = bodyId;\n"
+	"		worldAabb.m_maxVec = aabbAMaxOut;\n"
+	"		worldAabb.m_signedMaxIndices[3] = body[bodyId].m_invMass==0.f? 0 : 1;\n"
+	"		worldAabbs[bodyId] = worldAabb;\n"
+	"	}\n"
+	"}\n"
+	"#endif //B3_UPDATE_AABBS_H\n"
+	"__kernel void initializeGpuAabbsFull(  const int numNodes, __global b3RigidBodyData_t* gBodies,__global b3Collidable_t* collidables, __global b3Aabb_t* plocalShapeAABB, __global b3Aabb_t* pAABB)\n"
+	"{\n"
+	"	int nodeID = get_global_id(0);\n"
+	"	if( nodeID < numNodes )\n"
+	"	{\n"
+	"		b3ComputeWorldAabb(nodeID, gBodies, collidables, plocalShapeAABB,pAABB);\n"
+	"	}\n"
+	"}\n"
+	"__kernel void clearOverlappingPairsKernel(  __global int4* pairs, int numPairs)\n"
+	"{\n"
+	"	int pairId = get_global_id(0);\n"
+	"	if( pairId< numPairs )\n"
+	"	{\n"
+	"		pairs[pairId].z = 0xffffffff;\n"
+	"	}\n"
+	"}\n";
diff --git a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/autogenerated/bullet2.h b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/autogenerated/bullet2.h
index a6b57b1a128..eaa27dfe8f7 100644
--- a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/autogenerated/bullet2.h
+++ b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/autogenerated/bullet2.h
@@ -19,1035 +19,969 @@
 // Auto generated from Bullet/Extras/HeaderGenerator/bulletGenerate.py
 #ifndef __BULLET2_H__
 #define __BULLET2_H__
-namespace Bullet3SerializeBullet2 {
-
+namespace Bullet3SerializeBullet2
+{
 // put an empty struct in the case
-typedef struct bInvalidHandle {
+typedef struct bInvalidHandle
+{
 	int unused;
-}bInvalidHandle;
+} bInvalidHandle;
 
-    class PointerArray;
-    class b3PhysicsSystem;
-    class ListBase;
-    class b3Vector3FloatData;
-    class b3Vector3DoubleData;
-    class b3Matrix3x3FloatData;
-    class b3Matrix3x3DoubleData;
-    class b3TransformFloatData;
-    class b3TransformDoubleData;
-    class b3BvhSubtreeInfoData;
-    class b3OptimizedBvhNodeFloatData;
-    class b3OptimizedBvhNodeDoubleData;
-    class b3QuantizedBvhNodeData;
-    class b3QuantizedBvhFloatData;
-    class b3QuantizedBvhDoubleData;
-    class b3CollisionShapeData;
-    class b3StaticPlaneShapeData;
-    class b3ConvexInternalShapeData;
-    class b3PositionAndRadius;
-    class b3MultiSphereShapeData;
-    class b3IntIndexData;
-    class b3ShortIntIndexData;
-    class b3ShortIntIndexTripletData;
-    class b3CharIndexTripletData;
-    class b3MeshPartData;
-    class b3StridingMeshInterfaceData;
-    class b3TriangleMeshShapeData;
-    class b3ScaledTriangleMeshShapeData;
-    class b3CompoundShapeChildData;
-    class b3CompoundShapeData;
-    class b3CylinderShapeData;
-    class b3CapsuleShapeData;
-    class b3TriangleInfoData;
-    class b3TriangleInfoMapData;
-    class b3GImpactMeshShapeData;
-    class b3ConvexHullShapeData;
-    class b3CollisionObjectDoubleData;
-    class b3CollisionObjectFloatData;
-    class b3DynamicsWorldDoubleData;
-    class b3DynamicsWorldFloatData;
-    class b3RigidBodyFloatData;
-    class b3RigidBodyDoubleData;
-    class b3ConstraintInfo1;
-    class b3TypedConstraintData;
-    class b3Point2PointConstraintFloatData;
-    class b3Point2PointConstraintDoubleData;
-    class b3HingeConstraintDoubleData;
-    class b3HingeConstraintFloatData;
-    class b3ConeTwistConstraintData;
-    class b3Generic6DofConstraintData;
-    class b3Generic6DofSpringConstraintData;
-    class b3SliderConstraintData;
-    class b3ContactSolverInfoDoubleData;
-    class b3ContactSolverInfoFloatData;
-    class SoftBodyMaterialData;
-    class SoftBodyNodeData;
-    class SoftBodyLinkData;
-    class SoftBodyFaceData;
-    class SoftBodyTetraData;
-    class SoftRigidAnchorData;
-    class SoftBodyConfigData;
-    class SoftBodyPoseData;
-    class SoftBodyClusterData;
-    class b3SoftBodyJointData;
-    class b3SoftBodyFloatData;
+class PointerArray;
+class b3PhysicsSystem;
+class ListBase;
+class b3Vector3FloatData;
+class b3Vector3DoubleData;
+class b3Matrix3x3FloatData;
+class b3Matrix3x3DoubleData;
+class b3TransformFloatData;
+class b3TransformDoubleData;
+class b3BvhSubtreeInfoData;
+class b3OptimizedBvhNodeFloatData;
+class b3OptimizedBvhNodeDoubleData;
+class b3QuantizedBvhNodeData;
+class b3QuantizedBvhFloatData;
+class b3QuantizedBvhDoubleData;
+class b3CollisionShapeData;
+class b3StaticPlaneShapeData;
+class b3ConvexInternalShapeData;
+class b3PositionAndRadius;
+class b3MultiSphereShapeData;
+class b3IntIndexData;
+class b3ShortIntIndexData;
+class b3ShortIntIndexTripletData;
+class b3CharIndexTripletData;
+class b3MeshPartData;
+class b3StridingMeshInterfaceData;
+class b3TriangleMeshShapeData;
+class b3ScaledTriangleMeshShapeData;
+class b3CompoundShapeChildData;
+class b3CompoundShapeData;
+class b3CylinderShapeData;
+class b3CapsuleShapeData;
+class b3TriangleInfoData;
+class b3TriangleInfoMapData;
+class b3GImpactMeshShapeData;
+class b3ConvexHullShapeData;
+class b3CollisionObjectDoubleData;
+class b3CollisionObjectFloatData;
+class b3DynamicsWorldDoubleData;
+class b3DynamicsWorldFloatData;
+class b3RigidBodyFloatData;
+class b3RigidBodyDoubleData;
+class b3ConstraintInfo1;
+class b3TypedConstraintData;
+class b3Point2PointConstraintFloatData;
+class b3Point2PointConstraintDoubleData;
+class b3HingeConstraintDoubleData;
+class b3HingeConstraintFloatData;
+class b3ConeTwistConstraintData;
+class b3Generic6DofConstraintData;
+class b3Generic6DofSpringConstraintData;
+class b3SliderConstraintData;
+class b3ContactSolverInfoDoubleData;
+class b3ContactSolverInfoFloatData;
+class SoftBodyMaterialData;
+class SoftBodyNodeData;
+class SoftBodyLinkData;
+class SoftBodyFaceData;
+class SoftBodyTetraData;
+class SoftRigidAnchorData;
+class SoftBodyConfigData;
+class SoftBodyPoseData;
+class SoftBodyClusterData;
+class b3SoftBodyJointData;
+class b3SoftBodyFloatData;
 // -------------------------------------------------- //
-    class PointerArray
-    {
-    public:
-        int m_size;
-        int m_capacity;
-        void *m_data;
-    };
-
+class PointerArray
+{
+public:
+	int m_size;
+	int m_capacity;
+	void *m_data;
+};
 
 // -------------------------------------------------- //
-    class b3PhysicsSystem
-    {
-    public:
-        PointerArray m_collisionShapes;
-        PointerArray m_collisionObjects;
-        PointerArray m_constraints;
-    };
-
+class b3PhysicsSystem
+{
+public:
+	PointerArray m_collisionShapes;
+	PointerArray m_collisionObjects;
+	PointerArray m_constraints;
+};
 
 // -------------------------------------------------- //
-    class ListBase
-    {
-    public:
-        void *first;
-        void *last;
-    };
-
+class ListBase
+{
+public:
+	void *first;
+	void *last;
+};
 
 // -------------------------------------------------- //
-    class b3Vector3FloatData
-    {
-    public:
-        float m_floats[4];
-    };
-
+class b3Vector3FloatData
+{
+public:
+	float m_floats[4];
+};
 
 // -------------------------------------------------- //
-    class b3Vector3DoubleData
-    {
-    public:
-        double m_floats[4];
-    };
-
+class b3Vector3DoubleData
+{
+public:
+	double m_floats[4];
+};
 
 // -------------------------------------------------- //
-    class b3Matrix3x3FloatData
-    {
-    public:
-        b3Vector3FloatData m_el[3];
-    };
-
+class b3Matrix3x3FloatData
+{
+public:
+	b3Vector3FloatData m_el[3];
+};
 
 // -------------------------------------------------- //
-    class b3Matrix3x3DoubleData
-    {
-    public:
-        b3Vector3DoubleData m_el[3];
-    };
-
+class b3Matrix3x3DoubleData
+{
+public:
+	b3Vector3DoubleData m_el[3];
+};
 
 // -------------------------------------------------- //
-    class b3TransformFloatData
-    {
-    public:
-        b3Matrix3x3FloatData m_basis;
-        b3Vector3FloatData m_origin;
-    };
-
+class b3TransformFloatData
+{
+public:
+	b3Matrix3x3FloatData m_basis;
+	b3Vector3FloatData m_origin;
+};
 
 // -------------------------------------------------- //
-    class b3TransformDoubleData
-    {
-    public:
-        b3Matrix3x3DoubleData m_basis;
-        b3Vector3DoubleData m_origin;
-    };
-
+class b3TransformDoubleData
+{
+public:
+	b3Matrix3x3DoubleData m_basis;
+	b3Vector3DoubleData m_origin;
+};
 
 // -------------------------------------------------- //
-    class b3BvhSubtreeInfoData
-    {
-    public:
-        int m_rootNodeIndex;
-        int m_subtreeSize;
-        short m_quantizedAabbMin[3];
-        short m_quantizedAabbMax[3];
-    };
-
+class b3BvhSubtreeInfoData
+{
+public:
+	int m_rootNodeIndex;
+	int m_subtreeSize;
+	short m_quantizedAabbMin[3];
+	short m_quantizedAabbMax[3];
+};
 
 // -------------------------------------------------- //
-    class b3OptimizedBvhNodeFloatData
-    {
-    public:
-        b3Vector3FloatData m_aabbMinOrg;
-        b3Vector3FloatData m_aabbMaxOrg;
-        int m_escapeIndex;
-        int m_subPart;
-        int m_triangleIndex;
-        char m_pad[4];
-    };
-
+class b3OptimizedBvhNodeFloatData
+{
+public:
+	b3Vector3FloatData m_aabbMinOrg;
+	b3Vector3FloatData m_aabbMaxOrg;
+	int m_escapeIndex;
+	int m_subPart;
+	int m_triangleIndex;
+	char m_pad[4];
+};
 
 // -------------------------------------------------- //
-    class b3OptimizedBvhNodeDoubleData
-    {
-    public:
-        b3Vector3DoubleData m_aabbMinOrg;
-        b3Vector3DoubleData m_aabbMaxOrg;
-        int m_escapeIndex;
-        int m_subPart;
-        int m_triangleIndex;
-        char m_pad[4];
-    };
-
+class b3OptimizedBvhNodeDoubleData
+{
+public:
+	b3Vector3DoubleData m_aabbMinOrg;
+	b3Vector3DoubleData m_aabbMaxOrg;
+	int m_escapeIndex;
+	int m_subPart;
+	int m_triangleIndex;
+	char m_pad[4];
+};
 
 // -------------------------------------------------- //
-    class b3QuantizedBvhNodeData
-    {
-    public:
-        short m_quantizedAabbMin[3];
-        short m_quantizedAabbMax[3];
-        int m_escapeIndexOrTriangleIndex;
-    };
-
+class b3QuantizedBvhNodeData
+{
+public:
+	short m_quantizedAabbMin[3];
+	short m_quantizedAabbMax[3];
+	int m_escapeIndexOrTriangleIndex;
+};
 
 // -------------------------------------------------- //
-    class b3QuantizedBvhFloatData
-    {
-    public:
-        b3Vector3FloatData m_bvhAabbMin;
-        b3Vector3FloatData m_bvhAabbMax;
-        b3Vector3FloatData m_bvhQuantization;
-        int m_curNodeIndex;
-        int m_useQuantization;
-        int m_numContiguousLeafNodes;
-        int m_numQuantizedContiguousNodes;
-        b3OptimizedBvhNodeFloatData *m_contiguousNodesPtr;
-        b3QuantizedBvhNodeData *m_quantizedContiguousNodesPtr;
-        b3BvhSubtreeInfoData *m_subTreeInfoPtr;
-        int m_traversalMode;
-        int m_numSubtreeHeaders;
-    };
-
+class b3QuantizedBvhFloatData
+{
+public:
+	b3Vector3FloatData m_bvhAabbMin;
+	b3Vector3FloatData m_bvhAabbMax;
+	b3Vector3FloatData m_bvhQuantization;
+	int m_curNodeIndex;
+	int m_useQuantization;
+	int m_numContiguousLeafNodes;
+	int m_numQuantizedContiguousNodes;
+	b3OptimizedBvhNodeFloatData *m_contiguousNodesPtr;
+	b3QuantizedBvhNodeData *m_quantizedContiguousNodesPtr;
+	b3BvhSubtreeInfoData *m_subTreeInfoPtr;
+	int m_traversalMode;
+	int m_numSubtreeHeaders;
+};
 
 // -------------------------------------------------- //
-    class b3QuantizedBvhDoubleData
-    {
-    public:
-        b3Vector3DoubleData m_bvhAabbMin;
-        b3Vector3DoubleData m_bvhAabbMax;
-        b3Vector3DoubleData m_bvhQuantization;
-        int m_curNodeIndex;
-        int m_useQuantization;
-        int m_numContiguousLeafNodes;
-        int m_numQuantizedContiguousNodes;
-        b3OptimizedBvhNodeDoubleData *m_contiguousNodesPtr;
-        b3QuantizedBvhNodeData *m_quantizedContiguousNodesPtr;
-        int m_traversalMode;
-        int m_numSubtreeHeaders;
-        b3BvhSubtreeInfoData *m_subTreeInfoPtr;
-    };
-
+class b3QuantizedBvhDoubleData
+{
+public:
+	b3Vector3DoubleData m_bvhAabbMin;
+	b3Vector3DoubleData m_bvhAabbMax;
+	b3Vector3DoubleData m_bvhQuantization;
+	int m_curNodeIndex;
+	int m_useQuantization;
+	int m_numContiguousLeafNodes;
+	int m_numQuantizedContiguousNodes;
+	b3OptimizedBvhNodeDoubleData *m_contiguousNodesPtr;
+	b3QuantizedBvhNodeData *m_quantizedContiguousNodesPtr;
+	int m_traversalMode;
+	int m_numSubtreeHeaders;
+	b3BvhSubtreeInfoData *m_subTreeInfoPtr;
+};
 
 // -------------------------------------------------- //
-    class b3CollisionShapeData
-    {
-    public:
-        char *m_name;
-        int m_shapeType;
-        char m_padding[4];
-    };
-
+class b3CollisionShapeData
+{
+public:
+	char *m_name;
+	int m_shapeType;
+	char m_padding[4];
+};
 
 // -------------------------------------------------- //
-    class b3StaticPlaneShapeData
-    {
-    public:
-        b3CollisionShapeData m_collisionShapeData;
-        b3Vector3FloatData m_localScaling;
-        b3Vector3FloatData m_planeNormal;
-        float m_planeConstant;
-        char m_pad[4];
-    };
-
+class b3StaticPlaneShapeData
+{
+public:
+	b3CollisionShapeData m_collisionShapeData;
+	b3Vector3FloatData m_localScaling;
+	b3Vector3FloatData m_planeNormal;
+	float m_planeConstant;
+	char m_pad[4];
+};
 
 // -------------------------------------------------- //
-    class b3ConvexInternalShapeData
-    {
-    public:
-        b3CollisionShapeData m_collisionShapeData;
-        b3Vector3FloatData m_localScaling;
-        b3Vector3FloatData m_implicitShapeDimensions;
-        float m_collisionMargin;
-        int m_padding;
-    };
-
+class b3ConvexInternalShapeData
+{
+public:
+	b3CollisionShapeData m_collisionShapeData;
+	b3Vector3FloatData m_localScaling;
+	b3Vector3FloatData m_implicitShapeDimensions;
+	float m_collisionMargin;
+	int m_padding;
+};
 
 // -------------------------------------------------- //
-    class b3PositionAndRadius
-    {
-    public:
-        b3Vector3FloatData m_pos;
-        float m_radius;
-    };
-
+class b3PositionAndRadius
+{
+public:
+	b3Vector3FloatData m_pos;
+	float m_radius;
+};
 
 // -------------------------------------------------- //
-    class b3MultiSphereShapeData
-    {
-    public:
-        b3ConvexInternalShapeData m_convexInternalShapeData;
-        b3PositionAndRadius *m_localPositionArrayPtr;
-        int m_localPositionArraySize;
-        char m_padding[4];
-    };
-
+class b3MultiSphereShapeData
+{
+public:
+	b3ConvexInternalShapeData m_convexInternalShapeData;
+	b3PositionAndRadius *m_localPositionArrayPtr;
+	int m_localPositionArraySize;
+	char m_padding[4];
+};
 
 // -------------------------------------------------- //
-    class b3IntIndexData
-    {
-    public:
-        int m_value;
-    };
-
+class b3IntIndexData
+{
+public:
+	int m_value;
+};
 
 // -------------------------------------------------- //
-    class b3ShortIntIndexData
-    {
-    public:
-        short m_value;
-        char m_pad[2];
-    };
-
+class b3ShortIntIndexData
+{
+public:
+	short m_value;
+	char m_pad[2];
+};
 
 // -------------------------------------------------- //
-    class b3ShortIntIndexTripletData
-    {
-    public:
-        short m_values[3];
-        char m_pad[2];
-    };
-
+class b3ShortIntIndexTripletData
+{
+public:
+	short m_values[3];
+	char m_pad[2];
+};
 
 // -------------------------------------------------- //
-    class b3CharIndexTripletData
-    {
-    public:
-        char m_values[3];
-        char m_pad;
-    };
-
+class b3CharIndexTripletData
+{
+public:
+	char m_values[3];
+	char m_pad;
+};
 
 // -------------------------------------------------- //
-    class b3MeshPartData
-    {
-    public:
-        b3Vector3FloatData *m_vertices3f;
-        b3Vector3DoubleData *m_vertices3d;
-        b3IntIndexData *m_indices32;
-        b3ShortIntIndexTripletData *m_3indices16;
-        b3CharIndexTripletData *m_3indices8;
-        b3ShortIntIndexData *m_indices16;
-        int m_numTriangles;
-        int m_numVertices;
-    };
-
+class b3MeshPartData
+{
+public:
+	b3Vector3FloatData *m_vertices3f;
+	b3Vector3DoubleData *m_vertices3d;
+	b3IntIndexData *m_indices32;
+	b3ShortIntIndexTripletData *m_3indices16;
+	b3CharIndexTripletData *m_3indices8;
+	b3ShortIntIndexData *m_indices16;
+	int m_numTriangles;
+	int m_numVertices;
+};
 
 // -------------------------------------------------- //
-    class b3StridingMeshInterfaceData
-    {
-    public:
-        b3MeshPartData *m_meshPartsPtr;
-        b3Vector3FloatData m_scaling;
-        int m_numMeshParts;
-        char m_padding[4];
-    };
-
+class b3StridingMeshInterfaceData
+{
+public:
+	b3MeshPartData *m_meshPartsPtr;
+	b3Vector3FloatData m_scaling;
+	int m_numMeshParts;
+	char m_padding[4];
+};
 
 // -------------------------------------------------- //
-    class b3TriangleMeshShapeData
-    {
-    public:
-        b3CollisionShapeData m_collisionShapeData;
-        b3StridingMeshInterfaceData m_meshInterface;
-        b3QuantizedBvhFloatData *m_quantizedFloatBvh;
-        b3QuantizedBvhDoubleData *m_quantizedDoubleBvh;
-        b3TriangleInfoMapData *m_triangleInfoMap;
-        float m_collisionMargin;
-        char m_pad3[4];
-    };
-
+class b3TriangleMeshShapeData
+{
+public:
+	b3CollisionShapeData m_collisionShapeData;
+	b3StridingMeshInterfaceData m_meshInterface;
+	b3QuantizedBvhFloatData *m_quantizedFloatBvh;
+	b3QuantizedBvhDoubleData *m_quantizedDoubleBvh;
+	b3TriangleInfoMapData *m_triangleInfoMap;
+	float m_collisionMargin;
+	char m_pad3[4];
+};
 
 // -------------------------------------------------- //
-    class b3ScaledTriangleMeshShapeData
-    {
-    public:
-        b3TriangleMeshShapeData m_trimeshShapeData;
-        b3Vector3FloatData m_localScaling;
-    };
-
+class b3ScaledTriangleMeshShapeData
+{
+public:
+	b3TriangleMeshShapeData m_trimeshShapeData;
+	b3Vector3FloatData m_localScaling;
+};
 
 // -------------------------------------------------- //
-    class b3CompoundShapeChildData
-    {
-    public:
-        b3TransformFloatData m_transform;
-        b3CollisionShapeData *m_childShape;
-        int m_childShapeType;
-        float m_childMargin;
-    };
-
+class b3CompoundShapeChildData
+{
+public:
+	b3TransformFloatData m_transform;
+	b3CollisionShapeData *m_childShape;
+	int m_childShapeType;
+	float m_childMargin;
+};
 
 // -------------------------------------------------- //
-    class b3CompoundShapeData
-    {
-    public:
-        b3CollisionShapeData m_collisionShapeData;
-        b3CompoundShapeChildData *m_childShapePtr;
-        int m_numChildShapes;
-        float m_collisionMargin;
-    };
-
+class b3CompoundShapeData
+{
+public:
+	b3CollisionShapeData m_collisionShapeData;
+	b3CompoundShapeChildData *m_childShapePtr;
+	int m_numChildShapes;
+	float m_collisionMargin;
+};
 
 // -------------------------------------------------- //
-    class b3CylinderShapeData
-    {
-    public:
-        b3ConvexInternalShapeData m_convexInternalShapeData;
-        int m_upAxis;
-        char m_padding[4];
-    };
-
+class b3CylinderShapeData
+{
+public:
+	b3ConvexInternalShapeData m_convexInternalShapeData;
+	int m_upAxis;
+	char m_padding[4];
+};
 
 // -------------------------------------------------- //
-    class b3CapsuleShapeData
-    {
-    public:
-        b3ConvexInternalShapeData m_convexInternalShapeData;
-        int m_upAxis;
-        char m_padding[4];
-    };
-
+class b3CapsuleShapeData
+{
+public:
+	b3ConvexInternalShapeData m_convexInternalShapeData;
+	int m_upAxis;
+	char m_padding[4];
+};
 
 // -------------------------------------------------- //
-    class b3TriangleInfoData
-    {
-    public:
-        int m_flags;
-        float m_edgeV0V1Angle;
-        float m_edgeV1V2Angle;
-        float m_edgeV2V0Angle;
-    };
-
+class b3TriangleInfoData
+{
+public:
+	int m_flags;
+	float m_edgeV0V1Angle;
+	float m_edgeV1V2Angle;
+	float m_edgeV2V0Angle;
+};
 
 // -------------------------------------------------- //
-    class b3TriangleInfoMapData
-    {
-    public:
-        int *m_hashTablePtr;
-        int *m_nextPtr;
-        b3TriangleInfoData *m_valueArrayPtr;
-        int *m_keyArrayPtr;
-        float m_convexEpsilon;
-        float m_planarEpsilon;
-        float m_equalVertexThreshold;
-        float m_edgeDistanceThreshold;
-        float m_zeroAreaThreshold;
-        int m_nextSize;
-        int m_hashTableSize;
-        int m_numValues;
-        int m_numKeys;
-        char m_padding[4];
-    };
-
+class b3TriangleInfoMapData
+{
+public:
+	int *m_hashTablePtr;
+	int *m_nextPtr;
+	b3TriangleInfoData *m_valueArrayPtr;
+	int *m_keyArrayPtr;
+	float m_convexEpsilon;
+	float m_planarEpsilon;
+	float m_equalVertexThreshold;
+	float m_edgeDistanceThreshold;
+	float m_zeroAreaThreshold;
+	int m_nextSize;
+	int m_hashTableSize;
+	int m_numValues;
+	int m_numKeys;
+	char m_padding[4];
+};
 
 // -------------------------------------------------- //
-    class b3GImpactMeshShapeData
-    {
-    public:
-        b3CollisionShapeData m_collisionShapeData;
-        b3StridingMeshInterfaceData m_meshInterface;
-        b3Vector3FloatData m_localScaling;
-        float m_collisionMargin;
-        int m_gimpactSubType;
-    };
-
+class b3GImpactMeshShapeData
+{
+public:
+	b3CollisionShapeData m_collisionShapeData;
+	b3StridingMeshInterfaceData m_meshInterface;
+	b3Vector3FloatData m_localScaling;
+	float m_collisionMargin;
+	int m_gimpactSubType;
+};
 
 // -------------------------------------------------- //
-    class b3ConvexHullShapeData
-    {
-    public:
-        b3ConvexInternalShapeData m_convexInternalShapeData;
-        b3Vector3FloatData *m_unscaledPointsFloatPtr;
-        b3Vector3DoubleData *m_unscaledPointsDoublePtr;
-        int m_numUnscaledPoints;
-        char m_padding3[4];
-    };
-
+class b3ConvexHullShapeData
+{
+public:
+	b3ConvexInternalShapeData m_convexInternalShapeData;
+	b3Vector3FloatData *m_unscaledPointsFloatPtr;
+	b3Vector3DoubleData *m_unscaledPointsDoublePtr;
+	int m_numUnscaledPoints;
+	char m_padding3[4];
+};
 
 // -------------------------------------------------- //
-    class b3CollisionObjectDoubleData
-    {
-    public:
-        void *m_broadphaseHandle;
-        void *m_collisionShape;
-        b3CollisionShapeData *m_rootCollisionShape;
-        char *m_name;
-        b3TransformDoubleData m_worldTransform;
-        b3TransformDoubleData m_interpolationWorldTransform;
-        b3Vector3DoubleData m_interpolationLinearVelocity;
-        b3Vector3DoubleData m_interpolationAngularVelocity;
-        b3Vector3DoubleData m_anisotropicFriction;
-        double m_contactProcessingThreshold;
-        double m_deactivationTime;
-        double m_friction;
-        double m_rollingFriction;
-        double m_restitution;
-        double m_hitFraction;
-        double m_ccdSweptSphereRadius;
-        double m_ccdMotionThreshold;
-        int m_hasAnisotropicFriction;
-        int m_collisionFlags;
-        int m_islandTag1;
-        int m_companionId;
-        int m_activationState1;
-        int m_internalType;
-        int m_checkCollideWith;
-        char m_padding[4];
-    };
-
+class b3CollisionObjectDoubleData
+{
+public:
+	void *m_broadphaseHandle;
+	void *m_collisionShape;
+	b3CollisionShapeData *m_rootCollisionShape;
+	char *m_name;
+	b3TransformDoubleData m_worldTransform;
+	b3TransformDoubleData m_interpolationWorldTransform;
+	b3Vector3DoubleData m_interpolationLinearVelocity;
+	b3Vector3DoubleData m_interpolationAngularVelocity;
+	b3Vector3DoubleData m_anisotropicFriction;
+	double m_contactProcessingThreshold;
+	double m_deactivationTime;
+	double m_friction;
+	double m_rollingFriction;
+	double m_restitution;
+	double m_hitFraction;
+	double m_ccdSweptSphereRadius;
+	double m_ccdMotionThreshold;
+	int m_hasAnisotropicFriction;
+	int m_collisionFlags;
+	int m_islandTag1;
+	int m_companionId;
+	int m_activationState1;
+	int m_internalType;
+	int m_checkCollideWith;
+	char m_padding[4];
+};
 
 // -------------------------------------------------- //
-    class b3CollisionObjectFloatData
-    {
-    public:
-        void *m_broadphaseHandle;
-        void *m_collisionShape;
-        b3CollisionShapeData *m_rootCollisionShape;
-        char *m_name;
-        b3TransformFloatData m_worldTransform;
-        b3TransformFloatData m_interpolationWorldTransform;
-        b3Vector3FloatData m_interpolationLinearVelocity;
-        b3Vector3FloatData m_interpolationAngularVelocity;
-        b3Vector3FloatData m_anisotropicFriction;
-        float m_contactProcessingThreshold;
-        float m_deactivationTime;
-        float m_friction;
-        float m_rollingFriction;
-        float m_restitution;
-        float m_hitFraction;
-        float m_ccdSweptSphereRadius;
-        float m_ccdMotionThreshold;
-        int m_hasAnisotropicFriction;
-        int m_collisionFlags;
-        int m_islandTag1;
-        int m_companionId;
-        int m_activationState1;
-        int m_internalType;
-        int m_checkCollideWith;
-        char m_padding[4];
-    };
-
-
+class b3CollisionObjectFloatData
+{
+public:
+	void *m_broadphaseHandle;
+	void *m_collisionShape;
+	b3CollisionShapeData *m_rootCollisionShape;
+	char *m_name;
+	b3TransformFloatData m_worldTransform;
+	b3TransformFloatData m_interpolationWorldTransform;
+	b3Vector3FloatData m_interpolationLinearVelocity;
+	b3Vector3FloatData m_interpolationAngularVelocity;
+	b3Vector3FloatData m_anisotropicFriction;
+	float m_contactProcessingThreshold;
+	float m_deactivationTime;
+	float m_friction;
+	float m_rollingFriction;
+	float m_restitution;
+	float m_hitFraction;
+	float m_ccdSweptSphereRadius;
+	float m_ccdMotionThreshold;
+	int m_hasAnisotropicFriction;
+	int m_collisionFlags;
+	int m_islandTag1;
+	int m_companionId;
+	int m_activationState1;
+	int m_internalType;
+	int m_checkCollideWith;
+	char m_padding[4];
+};
 
 // -------------------------------------------------- //
-    class b3RigidBodyFloatData
-    {
-    public:
-        b3CollisionObjectFloatData m_collisionObjectData;
-        b3Matrix3x3FloatData m_invInertiaTensorWorld;
-        b3Vector3FloatData m_linearVelocity;
-        b3Vector3FloatData m_angularVelocity;
-        b3Vector3FloatData m_angularFactor;
-        b3Vector3FloatData m_linearFactor;
-        b3Vector3FloatData m_gravity;
-        b3Vector3FloatData m_gravity_acceleration;
-        b3Vector3FloatData m_invInertiaLocal;
-        b3Vector3FloatData m_totalForce;
-        b3Vector3FloatData m_totalTorque;
-        float m_inverseMass;
-        float m_linearDamping;
-        float m_angularDamping;
-        float m_additionalDampingFactor;
-        float m_additionalLinearDampingThresholdSqr;
-        float m_additionalAngularDampingThresholdSqr;
-        float m_additionalAngularDampingFactor;
-        float m_linearSleepingThreshold;
-        float m_angularSleepingThreshold;
-        int m_additionalDamping;
-    };
-
+class b3RigidBodyFloatData
+{
+public:
+	b3CollisionObjectFloatData m_collisionObjectData;
+	b3Matrix3x3FloatData m_invInertiaTensorWorld;
+	b3Vector3FloatData m_linearVelocity;
+	b3Vector3FloatData m_angularVelocity;
+	b3Vector3FloatData m_angularFactor;
+	b3Vector3FloatData m_linearFactor;
+	b3Vector3FloatData m_gravity;
+	b3Vector3FloatData m_gravity_acceleration;
+	b3Vector3FloatData m_invInertiaLocal;
+	b3Vector3FloatData m_totalForce;
+	b3Vector3FloatData m_totalTorque;
+	float m_inverseMass;
+	float m_linearDamping;
+	float m_angularDamping;
+	float m_additionalDampingFactor;
+	float m_additionalLinearDampingThresholdSqr;
+	float m_additionalAngularDampingThresholdSqr;
+	float m_additionalAngularDampingFactor;
+	float m_linearSleepingThreshold;
+	float m_angularSleepingThreshold;
+	int m_additionalDamping;
+};
 
 // -------------------------------------------------- //
-    class b3RigidBodyDoubleData
-    {
-    public:
-        b3CollisionObjectDoubleData m_collisionObjectData;
-        b3Matrix3x3DoubleData m_invInertiaTensorWorld;
-        b3Vector3DoubleData m_linearVelocity;
-        b3Vector3DoubleData m_angularVelocity;
-        b3Vector3DoubleData m_angularFactor;
-        b3Vector3DoubleData m_linearFactor;
-        b3Vector3DoubleData m_gravity;
-        b3Vector3DoubleData m_gravity_acceleration;
-        b3Vector3DoubleData m_invInertiaLocal;
-        b3Vector3DoubleData m_totalForce;
-        b3Vector3DoubleData m_totalTorque;
-        double m_inverseMass;
-        double m_linearDamping;
-        double m_angularDamping;
-        double m_additionalDampingFactor;
-        double m_additionalLinearDampingThresholdSqr;
-        double m_additionalAngularDampingThresholdSqr;
-        double m_additionalAngularDampingFactor;
-        double m_linearSleepingThreshold;
-        double m_angularSleepingThreshold;
-        int m_additionalDamping;
-        char m_padding[4];
-    };
-
+class b3RigidBodyDoubleData
+{
+public:
+	b3CollisionObjectDoubleData m_collisionObjectData;
+	b3Matrix3x3DoubleData m_invInertiaTensorWorld;
+	b3Vector3DoubleData m_linearVelocity;
+	b3Vector3DoubleData m_angularVelocity;
+	b3Vector3DoubleData m_angularFactor;
+	b3Vector3DoubleData m_linearFactor;
+	b3Vector3DoubleData m_gravity;
+	b3Vector3DoubleData m_gravity_acceleration;
+	b3Vector3DoubleData m_invInertiaLocal;
+	b3Vector3DoubleData m_totalForce;
+	b3Vector3DoubleData m_totalTorque;
+	double m_inverseMass;
+	double m_linearDamping;
+	double m_angularDamping;
+	double m_additionalDampingFactor;
+	double m_additionalLinearDampingThresholdSqr;
+	double m_additionalAngularDampingThresholdSqr;
+	double m_additionalAngularDampingFactor;
+	double m_linearSleepingThreshold;
+	double m_angularSleepingThreshold;
+	int m_additionalDamping;
+	char m_padding[4];
+};
 
 // -------------------------------------------------- //
-    class b3ConstraintInfo1
-    {
-    public:
-        int m_numConstraintRows;
-        int nub;
-    };
-
+class b3ConstraintInfo1
+{
+public:
+	int m_numConstraintRows;
+	int nub;
+};
 
 // -------------------------------------------------- //
-    class b3TypedConstraintData
-    {
-    public:
-        bInvalidHandle *m_rbA;
-        bInvalidHandle *m_rbB;
-        char *m_name;
-        int m_objectType;
-        int m_userConstraintType;
-        int m_userConstraintId;
-        int m_needsFeedback;
-        float m_appliedImpulse;
-        float m_dbgDrawSize;
-        int m_disableCollisionsBetweenLinkedBodies;
-        int m_overrideNumSolverIterations;
-        float m_breakingImpulseThreshold;
-        int m_isEnabled;
-    };
-
+class b3TypedConstraintData
+{
+public:
+	bInvalidHandle *m_rbA;
+	bInvalidHandle *m_rbB;
+	char *m_name;
+	int m_objectType;
+	int m_userConstraintType;
+	int m_userConstraintId;
+	int m_needsFeedback;
+	float m_appliedImpulse;
+	float m_dbgDrawSize;
+	int m_disableCollisionsBetweenLinkedBodies;
+	int m_overrideNumSolverIterations;
+	float m_breakingImpulseThreshold;
+	int m_isEnabled;
+};
 
 // -------------------------------------------------- //
-    class b3Point2PointConstraintFloatData
-    {
-    public:
-        b3TypedConstraintData m_typeConstraintData;
-        b3Vector3FloatData m_pivotInA;
-        b3Vector3FloatData m_pivotInB;
-    };
-
+class b3Point2PointConstraintFloatData
+{
+public:
+	b3TypedConstraintData m_typeConstraintData;
+	b3Vector3FloatData m_pivotInA;
+	b3Vector3FloatData m_pivotInB;
+};
 
 // -------------------------------------------------- //
-    class b3Point2PointConstraintDoubleData
-    {
-    public:
-        b3TypedConstraintData m_typeConstraintData;
-        b3Vector3DoubleData m_pivotInA;
-        b3Vector3DoubleData m_pivotInB;
-    };
-
+class b3Point2PointConstraintDoubleData
+{
+public:
+	b3TypedConstraintData m_typeConstraintData;
+	b3Vector3DoubleData m_pivotInA;
+	b3Vector3DoubleData m_pivotInB;
+};
 
 // -------------------------------------------------- //
-    class b3HingeConstraintDoubleData
-    {
-    public:
-        b3TypedConstraintData m_typeConstraintData;
-        b3TransformDoubleData m_rbAFrame;
-        b3TransformDoubleData m_rbBFrame;
-        int m_useReferenceFrameA;
-        int m_angularOnly;
-        int m_enableAngularMotor;
-        float m_motorTargetVelocity;
-        float m_maxMotorImpulse;
-        float m_lowerLimit;
-        float m_upperLimit;
-        float m_limitSoftness;
-        float m_biasFactor;
-        float m_relaxationFactor;
-    };
-
+class b3HingeConstraintDoubleData
+{
+public:
+	b3TypedConstraintData m_typeConstraintData;
+	b3TransformDoubleData m_rbAFrame;
+	b3TransformDoubleData m_rbBFrame;
+	int m_useReferenceFrameA;
+	int m_angularOnly;
+	int m_enableAngularMotor;
+	float m_motorTargetVelocity;
+	float m_maxMotorImpulse;
+	float m_lowerLimit;
+	float m_upperLimit;
+	float m_limitSoftness;
+	float m_biasFactor;
+	float m_relaxationFactor;
+};
 
 // -------------------------------------------------- //
-    class b3HingeConstraintFloatData
-    {
-    public:
-        b3TypedConstraintData m_typeConstraintData;
-        b3TransformFloatData m_rbAFrame;
-        b3TransformFloatData m_rbBFrame;
-        int m_useReferenceFrameA;
-        int m_angularOnly;
-        int m_enableAngularMotor;
-        float m_motorTargetVelocity;
-        float m_maxMotorImpulse;
-        float m_lowerLimit;
-        float m_upperLimit;
-        float m_limitSoftness;
-        float m_biasFactor;
-        float m_relaxationFactor;
-    };
-
+class b3HingeConstraintFloatData
+{
+public:
+	b3TypedConstraintData m_typeConstraintData;
+	b3TransformFloatData m_rbAFrame;
+	b3TransformFloatData m_rbBFrame;
+	int m_useReferenceFrameA;
+	int m_angularOnly;
+	int m_enableAngularMotor;
+	float m_motorTargetVelocity;
+	float m_maxMotorImpulse;
+	float m_lowerLimit;
+	float m_upperLimit;
+	float m_limitSoftness;
+	float m_biasFactor;
+	float m_relaxationFactor;
+};
 
 // -------------------------------------------------- //
-    class b3ConeTwistConstraintData
-    {
-    public:
-        b3TypedConstraintData m_typeConstraintData;
-        b3TransformFloatData m_rbAFrame;
-        b3TransformFloatData m_rbBFrame;
-        float m_swingSpan1;
-        float m_swingSpan2;
-        float m_twistSpan;
-        float m_limitSoftness;
-        float m_biasFactor;
-        float m_relaxationFactor;
-        float m_damping;
-        char m_pad[4];
-    };
-
+class b3ConeTwistConstraintData
+{
+public:
+	b3TypedConstraintData m_typeConstraintData;
+	b3TransformFloatData m_rbAFrame;
+	b3TransformFloatData m_rbBFrame;
+	float m_swingSpan1;
+	float m_swingSpan2;
+	float m_twistSpan;
+	float m_limitSoftness;
+	float m_biasFactor;
+	float m_relaxationFactor;
+	float m_damping;
+	char m_pad[4];
+};
 
 // -------------------------------------------------- //
-    class b3Generic6DofConstraintData
-    {
-    public:
-        b3TypedConstraintData m_typeConstraintData;
-        b3TransformFloatData m_rbAFrame;
-        b3TransformFloatData m_rbBFrame;
-        b3Vector3FloatData m_linearUpperLimit;
-        b3Vector3FloatData m_linearLowerLimit;
-        b3Vector3FloatData m_angularUpperLimit;
-        b3Vector3FloatData m_angularLowerLimit;
-        int m_useLinearReferenceFrameA;
-        int m_useOffsetForConstraintFrame;
-    };
-
+class b3Generic6DofConstraintData
+{
+public:
+	b3TypedConstraintData m_typeConstraintData;
+	b3TransformFloatData m_rbAFrame;
+	b3TransformFloatData m_rbBFrame;
+	b3Vector3FloatData m_linearUpperLimit;
+	b3Vector3FloatData m_linearLowerLimit;
+	b3Vector3FloatData m_angularUpperLimit;
+	b3Vector3FloatData m_angularLowerLimit;
+	int m_useLinearReferenceFrameA;
+	int m_useOffsetForConstraintFrame;
+};
 
 // -------------------------------------------------- //
-    class b3Generic6DofSpringConstraintData
-    {
-    public:
-        b3Generic6DofConstraintData m_6dofData;
-        int m_springEnabled[6];
-        float m_equilibriumPoint[6];
-        float m_springStiffness[6];
-        float m_springDamping[6];
-    };
-
+class b3Generic6DofSpringConstraintData
+{
+public:
+	b3Generic6DofConstraintData m_6dofData;
+	int m_springEnabled[6];
+	float m_equilibriumPoint[6];
+	float m_springStiffness[6];
+	float m_springDamping[6];
+};
 
 // -------------------------------------------------- //
-    class b3SliderConstraintData
-    {
-    public:
-        b3TypedConstraintData m_typeConstraintData;
-        b3TransformFloatData m_rbAFrame;
-        b3TransformFloatData m_rbBFrame;
-        float m_linearUpperLimit;
-        float m_linearLowerLimit;
-        float m_angularUpperLimit;
-        float m_angularLowerLimit;
-        int m_useLinearReferenceFrameA;
-        int m_useOffsetForConstraintFrame;
-    };
-
+class b3SliderConstraintData
+{
+public:
+	b3TypedConstraintData m_typeConstraintData;
+	b3TransformFloatData m_rbAFrame;
+	b3TransformFloatData m_rbBFrame;
+	float m_linearUpperLimit;
+	float m_linearLowerLimit;
+	float m_angularUpperLimit;
+	float m_angularLowerLimit;
+	int m_useLinearReferenceFrameA;
+	int m_useOffsetForConstraintFrame;
+};
 
 // -------------------------------------------------- //
-    class b3ContactSolverInfoDoubleData
-    {
-    public:
-        double m_tau;
-        double m_damping;
-        double m_friction;
-        double m_timeStep;
-        double m_restitution;
-        double m_maxErrorReduction;
-        double m_sor;
-        double m_erp;
-        double m_erp2;
-        double m_globalCfm;
-        double m_splitImpulsePenetrationThreshold;
-        double m_splitImpulseTurnErp;
-        double m_linearSlop;
-        double m_warmstartingFactor;
-        double m_maxGyroscopicForce;
-        double m_singleAxisRollingFrictionThreshold;
-        int m_numIterations;
-        int m_solverMode;
-        int m_restingContactRestitutionThreshold;
-        int m_minimumSolverBatchSize;
-        int m_splitImpulse;
-        char m_padding[4];
-    };
-
+class b3ContactSolverInfoDoubleData
+{
+public:
+	double m_tau;
+	double m_damping;
+	double m_friction;
+	double m_timeStep;
+	double m_restitution;
+	double m_maxErrorReduction;
+	double m_sor;
+	double m_erp;
+	double m_erp2;
+	double m_globalCfm;
+	double m_splitImpulsePenetrationThreshold;
+	double m_splitImpulseTurnErp;
+	double m_linearSlop;
+	double m_warmstartingFactor;
+	double m_maxGyroscopicForce;
+	double m_singleAxisRollingFrictionThreshold;
+	int m_numIterations;
+	int m_solverMode;
+	int m_restingContactRestitutionThreshold;
+	int m_minimumSolverBatchSize;
+	int m_splitImpulse;
+	char m_padding[4];
+};
 
 // -------------------------------------------------- //
-    class b3ContactSolverInfoFloatData
-    {
-    public:
-        float m_tau;
-        float m_damping;
-        float m_friction;
-        float m_timeStep;
-        float m_restitution;
-        float m_maxErrorReduction;
-        float m_sor;
-        float m_erp;
-        float m_erp2;
-        float m_globalCfm;
-        float m_splitImpulsePenetrationThreshold;
-        float m_splitImpulseTurnErp;
-        float m_linearSlop;
-        float m_warmstartingFactor;
-        float m_maxGyroscopicForce;
-        float m_singleAxisRollingFrictionThreshold;
-        int m_numIterations;
-        int m_solverMode;
-        int m_restingContactRestitutionThreshold;
-        int m_minimumSolverBatchSize;
-        int m_splitImpulse;
-        char m_padding[4];
-    };
-
-
-	// -------------------------------------------------- //
-    class b3DynamicsWorldDoubleData
-    {
-    public:
-        b3ContactSolverInfoDoubleData m_solverInfo;
-        b3Vector3DoubleData m_gravity;
-    };
-
+class b3ContactSolverInfoFloatData
+{
+public:
+	float m_tau;
+	float m_damping;
+	float m_friction;
+	float m_timeStep;
+	float m_restitution;
+	float m_maxErrorReduction;
+	float m_sor;
+	float m_erp;
+	float m_erp2;
+	float m_globalCfm;
+	float m_splitImpulsePenetrationThreshold;
+	float m_splitImpulseTurnErp;
+	float m_linearSlop;
+	float m_warmstartingFactor;
+	float m_maxGyroscopicForce;
+	float m_singleAxisRollingFrictionThreshold;
+	int m_numIterations;
+	int m_solverMode;
+	int m_restingContactRestitutionThreshold;
+	int m_minimumSolverBatchSize;
+	int m_splitImpulse;
+	char m_padding[4];
+};
 
 // -------------------------------------------------- //
-    class b3DynamicsWorldFloatData
-    {
-    public:
-        b3ContactSolverInfoFloatData m_solverInfo;
-        b3Vector3FloatData m_gravity;
-    };
-
-
+class b3DynamicsWorldDoubleData
+{
+public:
+	b3ContactSolverInfoDoubleData m_solverInfo;
+	b3Vector3DoubleData m_gravity;
+};
 
 // -------------------------------------------------- //
-    class SoftBodyMaterialData
-    {
-    public:
-        float m_linearStiffness;
-        float m_angularStiffness;
-        float m_volumeStiffness;
-        int m_flags;
-    };
-
+class b3DynamicsWorldFloatData
+{
+public:
+	b3ContactSolverInfoFloatData m_solverInfo;
+	b3Vector3FloatData m_gravity;
+};
 
 // -------------------------------------------------- //
-    class SoftBodyNodeData
-    {
-    public:
-        SoftBodyMaterialData *m_material;
-        b3Vector3FloatData m_position;
-        b3Vector3FloatData m_previousPosition;
-        b3Vector3FloatData m_velocity;
-        b3Vector3FloatData m_accumulatedForce;
-        b3Vector3FloatData m_normal;
-        float m_inverseMass;
-        float m_area;
-        int m_attach;
-        int m_pad;
-    };
-
+class SoftBodyMaterialData
+{
+public:
+	float m_linearStiffness;
+	float m_angularStiffness;
+	float m_volumeStiffness;
+	int m_flags;
+};
 
 // -------------------------------------------------- //
-    class SoftBodyLinkData
-    {
-    public:
-        SoftBodyMaterialData *m_material;
-        int m_nodeIndices[2];
-        float m_restLength;
-        int m_bbending;
-    };
-
+class SoftBodyNodeData
+{
+public:
+	SoftBodyMaterialData *m_material;
+	b3Vector3FloatData m_position;
+	b3Vector3FloatData m_previousPosition;
+	b3Vector3FloatData m_velocity;
+	b3Vector3FloatData m_accumulatedForce;
+	b3Vector3FloatData m_normal;
+	float m_inverseMass;
+	float m_area;
+	int m_attach;
+	int m_pad;
+};
 
 // -------------------------------------------------- //
-    class SoftBodyFaceData
-    {
-    public:
-        b3Vector3FloatData m_normal;
-        SoftBodyMaterialData *m_material;
-        int m_nodeIndices[3];
-        float m_restArea;
-    };
-
+class SoftBodyLinkData
+{
+public:
+	SoftBodyMaterialData *m_material;
+	int m_nodeIndices[2];
+	float m_restLength;
+	int m_bbending;
+};
 
 // -------------------------------------------------- //
-    class SoftBodyTetraData
-    {
-    public:
-        b3Vector3FloatData m_c0[4];
-        SoftBodyMaterialData *m_material;
-        int m_nodeIndices[4];
-        float m_restVolume;
-        float m_c1;
-        float m_c2;
-        int m_pad;
-    };
-
+class SoftBodyFaceData
+{
+public:
+	b3Vector3FloatData m_normal;
+	SoftBodyMaterialData *m_material;
+	int m_nodeIndices[3];
+	float m_restArea;
+};
 
 // -------------------------------------------------- //
-    class SoftRigidAnchorData
-    {
-    public:
-        b3Matrix3x3FloatData m_c0;
-        b3Vector3FloatData m_c1;
-        b3Vector3FloatData m_localFrame;
-        bInvalidHandle *m_rigidBody;
-        int m_nodeIndex;
-        float m_c2;
-    };
-
+class SoftBodyTetraData
+{
+public:
+	b3Vector3FloatData m_c0[4];
+	SoftBodyMaterialData *m_material;
+	int m_nodeIndices[4];
+	float m_restVolume;
+	float m_c1;
+	float m_c2;
+	int m_pad;
+};
 
 // -------------------------------------------------- //
-    class SoftBodyConfigData
-    {
-    public:
-        int m_aeroModel;
-        float m_baumgarte;
-        float m_damping;
-        float m_drag;
-        float m_lift;
-        float m_pressure;
-        float m_volume;
-        float m_dynamicFriction;
-        float m_poseMatch;
-        float m_rigidContactHardness;
-        float m_kineticContactHardness;
-        float m_softContactHardness;
-        float m_anchorHardness;
-        float m_softRigidClusterHardness;
-        float m_softKineticClusterHardness;
-        float m_softSoftClusterHardness;
-        float m_softRigidClusterImpulseSplit;
-        float m_softKineticClusterImpulseSplit;
-        float m_softSoftClusterImpulseSplit;
-        float m_maxVolume;
-        float m_timeScale;
-        int m_velocityIterations;
-        int m_positionIterations;
-        int m_driftIterations;
-        int m_clusterIterations;
-        int m_collisionFlags;
-    };
-
+class SoftRigidAnchorData
+{
+public:
+	b3Matrix3x3FloatData m_c0;
+	b3Vector3FloatData m_c1;
+	b3Vector3FloatData m_localFrame;
+	bInvalidHandle *m_rigidBody;
+	int m_nodeIndex;
+	float m_c2;
+};
 
 // -------------------------------------------------- //
-    class SoftBodyPoseData
-    {
-    public:
-        b3Matrix3x3FloatData m_rot;
-        b3Matrix3x3FloatData m_scale;
-        b3Matrix3x3FloatData m_aqq;
-        b3Vector3FloatData m_com;
-        b3Vector3FloatData *m_positions;
-        float *m_weights;
-        int m_numPositions;
-        int m_numWeigts;
-        int m_bvolume;
-        int m_bframe;
-        float m_restVolume;
-        int m_pad;
-    };
-
+class SoftBodyConfigData
+{
+public:
+	int m_aeroModel;
+	float m_baumgarte;
+	float m_damping;
+	float m_drag;
+	float m_lift;
+	float m_pressure;
+	float m_volume;
+	float m_dynamicFriction;
+	float m_poseMatch;
+	float m_rigidContactHardness;
+	float m_kineticContactHardness;
+	float m_softContactHardness;
+	float m_anchorHardness;
+	float m_softRigidClusterHardness;
+	float m_softKineticClusterHardness;
+	float m_softSoftClusterHardness;
+	float m_softRigidClusterImpulseSplit;
+	float m_softKineticClusterImpulseSplit;
+	float m_softSoftClusterImpulseSplit;
+	float m_maxVolume;
+	float m_timeScale;
+	int m_velocityIterations;
+	int m_positionIterations;
+	int m_driftIterations;
+	int m_clusterIterations;
+	int m_collisionFlags;
+};
 
 // -------------------------------------------------- //
-    class SoftBodyClusterData
-    {
-    public:
-        b3TransformFloatData m_framexform;
-        b3Matrix3x3FloatData m_locii;
-        b3Matrix3x3FloatData m_invwi;
-        b3Vector3FloatData m_com;
-        b3Vector3FloatData m_vimpulses[2];
-        b3Vector3FloatData m_dimpulses[2];
-        b3Vector3FloatData m_lv;
-        b3Vector3FloatData m_av;
-        b3Vector3FloatData *m_framerefs;
-        int *m_nodeIndices;
-        float *m_masses;
-        int m_numFrameRefs;
-        int m_numNodes;
-        int m_numMasses;
-        float m_idmass;
-        float m_imass;
-        int m_nvimpulses;
-        int m_ndimpulses;
-        float m_ndamping;
-        float m_ldamping;
-        float m_adamping;
-        float m_matching;
-        float m_maxSelfCollisionImpulse;
-        float m_selfCollisionImpulseFactor;
-        int m_containsAnchor;
-        int m_collide;
-        int m_clusterIndex;
-    };
-
+class SoftBodyPoseData
+{
+public:
+	b3Matrix3x3FloatData m_rot;
+	b3Matrix3x3FloatData m_scale;
+	b3Matrix3x3FloatData m_aqq;
+	b3Vector3FloatData m_com;
+	b3Vector3FloatData *m_positions;
+	float *m_weights;
+	int m_numPositions;
+	int m_numWeigts;
+	int m_bvolume;
+	int m_bframe;
+	float m_restVolume;
+	int m_pad;
+};
 
 // -------------------------------------------------- //
-    class b3SoftBodyJointData
-    {
-    public:
-        void *m_bodyA;
-        void *m_bodyB;
-        b3Vector3FloatData m_refs[2];
-        float m_cfm;
-        float m_erp;
-        float m_split;
-        int m_delete;
-        b3Vector3FloatData m_relPosition[2];
-        int m_bodyAtype;
-        int m_bodyBtype;
-        int m_jointType;
-        int m_pad;
-    };
-
+class SoftBodyClusterData
+{
+public:
+	b3TransformFloatData m_framexform;
+	b3Matrix3x3FloatData m_locii;
+	b3Matrix3x3FloatData m_invwi;
+	b3Vector3FloatData m_com;
+	b3Vector3FloatData m_vimpulses[2];
+	b3Vector3FloatData m_dimpulses[2];
+	b3Vector3FloatData m_lv;
+	b3Vector3FloatData m_av;
+	b3Vector3FloatData *m_framerefs;
+	int *m_nodeIndices;
+	float *m_masses;
+	int m_numFrameRefs;
+	int m_numNodes;
+	int m_numMasses;
+	float m_idmass;
+	float m_imass;
+	int m_nvimpulses;
+	int m_ndimpulses;
+	float m_ndamping;
+	float m_ldamping;
+	float m_adamping;
+	float m_matching;
+	float m_maxSelfCollisionImpulse;
+	float m_selfCollisionImpulseFactor;
+	int m_containsAnchor;
+	int m_collide;
+	int m_clusterIndex;
+};
 
 // -------------------------------------------------- //
-    class b3SoftBodyFloatData
-    {
-    public:
-        b3CollisionObjectFloatData m_collisionObjectData;
-        SoftBodyPoseData *m_pose;
-        SoftBodyMaterialData **m_materials;
-        SoftBodyNodeData *m_nodes;
-        SoftBodyLinkData *m_links;
-        SoftBodyFaceData *m_faces;
-        SoftBodyTetraData *m_tetrahedra;
-        SoftRigidAnchorData *m_anchors;
-        SoftBodyClusterData *m_clusters;
-        b3SoftBodyJointData *m_joints;
-        int m_numMaterials;
-        int m_numNodes;
-        int m_numLinks;
-        int m_numFaces;
-        int m_numTetrahedra;
-        int m_numAnchors;
-        int m_numClusters;
-        int m_numJoints;
-        SoftBodyConfigData m_config;
-    };
+class b3SoftBodyJointData
+{
+public:
+	void *m_bodyA;
+	void *m_bodyB;
+	b3Vector3FloatData m_refs[2];
+	float m_cfm;
+	float m_erp;
+	float m_split;
+	int m_delete;
+	b3Vector3FloatData m_relPosition[2];
+	int m_bodyAtype;
+	int m_bodyBtype;
+	int m_jointType;
+	int m_pad;
+};
 
+// -------------------------------------------------- //
+class b3SoftBodyFloatData
+{
+public:
+	b3CollisionObjectFloatData m_collisionObjectData;
+	SoftBodyPoseData *m_pose;
+	SoftBodyMaterialData **m_materials;
+	SoftBodyNodeData *m_nodes;
+	SoftBodyLinkData *m_links;
+	SoftBodyFaceData *m_faces;
+	SoftBodyTetraData *m_tetrahedra;
+	SoftRigidAnchorData *m_anchors;
+	SoftBodyClusterData *m_clusters;
+	b3SoftBodyJointData *m_joints;
+	int m_numMaterials;
+	int m_numNodes;
+	int m_numLinks;
+	int m_numFaces;
+	int m_numTetrahedra;
+	int m_numAnchors;
+	int m_numClusters;
+	int m_numJoints;
+	SoftBodyConfigData m_config;
+};
 
-}
-#endif//__BULLET2_H__
\ No newline at end of file
+}  // namespace Bullet3SerializeBullet2
+#endif  //__BULLET2_H__
\ No newline at end of file
diff --git a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3BulletFile.cpp b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3BulletFile.cpp
index c3ceb8388ca..d2a71636708 100644
--- a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3BulletFile.cpp
+++ b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3BulletFile.cpp
@@ -17,12 +17,11 @@ subject to the following restrictions:
 #include "b3Defines.h"
 #include "b3DNA.h"
 
-#if !defined( __CELLOS_LV2__) && !defined(__MWERKS__)
+#if !defined(__CELLOS_LV2__) && !defined(__MWERKS__)
 #include <memory.h>
 #endif
 #include <string.h>
 
-
 // 32 && 64 bit versions
 #ifdef B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
 #ifdef _WIN64
@@ -31,130 +30,116 @@ extern int b3s_bulletDNAlen64;
 #else
 extern char b3s_bulletDNAstr[];
 extern int b3s_bulletDNAlen;
-#endif //_WIN64
-#else//B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
+#endif  //_WIN64
+#else   //B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
 
 extern char b3s_bulletDNAstr64[];
 extern int b3s_bulletDNAlen64;
 extern char b3s_bulletDNAstr[];
 extern int b3s_bulletDNAlen;
 
-#endif //B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
+#endif  //B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
 
 using namespace bParse;
 
 b3BulletFile::b3BulletFile()
-:bFile("", "BULLET ")
+	: bFile("", "BULLET ")
 {
-	mMemoryDNA = new bDNA(); //this memory gets released in the bFile::~bFile destructor,@todo not consistent with the rule 'who allocates it, has to deallocate it"
+	mMemoryDNA = new bDNA();  //this memory gets released in the bFile::~bFile destructor,@todo not consistent with the rule 'who allocates it, has to deallocate it"
 
 	m_DnaCopy = 0;
 
-
 #ifdef B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
 #ifdef _WIN64
-		m_DnaCopy = (char*)b3AlignedAlloc(b3s_bulletDNAlen64,16);
-		memcpy(m_DnaCopy,b3s_bulletDNAstr64,b3s_bulletDNAlen64);
-		mMemoryDNA->init(m_DnaCopy,b3s_bulletDNAlen64);
-#else//_WIN64
-		m_DnaCopy = (char*)b3AlignedAlloc(b3s_bulletDNAlen,16);
-		memcpy(m_DnaCopy,b3s_bulletDNAstr,b3s_bulletDNAlen);
-		mMemoryDNA->init(m_DnaCopy,b3s_bulletDNAlen);
-#endif//_WIN64
-#else//B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
+	m_DnaCopy = (char*)b3AlignedAlloc(b3s_bulletDNAlen64, 16);
+	memcpy(m_DnaCopy, b3s_bulletDNAstr64, b3s_bulletDNAlen64);
+	mMemoryDNA->init(m_DnaCopy, b3s_bulletDNAlen64);
+#else   //_WIN64
+	m_DnaCopy = (char*)b3AlignedAlloc(b3s_bulletDNAlen, 16);
+	memcpy(m_DnaCopy, b3s_bulletDNAstr, b3s_bulletDNAlen);
+	mMemoryDNA->init(m_DnaCopy, b3s_bulletDNAlen);
+#endif  //_WIN64
+#else   //B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
 	if (VOID_IS_8)
 	{
-		m_DnaCopy = (char*) b3AlignedAlloc(b3s_bulletDNAlen64,16);
-		memcpy(m_DnaCopy,b3s_bulletDNAstr64,b3s_bulletDNAlen64);
-		mMemoryDNA->init(m_DnaCopy,b3s_bulletDNAlen64);
+		m_DnaCopy = (char*)b3AlignedAlloc(b3s_bulletDNAlen64, 16);
+		memcpy(m_DnaCopy, b3s_bulletDNAstr64, b3s_bulletDNAlen64);
+		mMemoryDNA->init(m_DnaCopy, b3s_bulletDNAlen64);
 	}
 	else
 	{
-		m_DnaCopy =(char*) b3AlignedAlloc(b3s_bulletDNAlen,16);
-		memcpy(m_DnaCopy,b3s_bulletDNAstr,b3s_bulletDNAlen);
-		mMemoryDNA->init(m_DnaCopy,b3s_bulletDNAlen);
+		m_DnaCopy = (char*)b3AlignedAlloc(b3s_bulletDNAlen, 16);
+		memcpy(m_DnaCopy, b3s_bulletDNAstr, b3s_bulletDNAlen);
+		mMemoryDNA->init(m_DnaCopy, b3s_bulletDNAlen);
 	}
-#endif//B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
+#endif  //B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
 }
 
-
-
 b3BulletFile::b3BulletFile(const char* fileName)
-:bFile(fileName, "BULLET ")
+	: bFile(fileName, "BULLET ")
 {
 	m_DnaCopy = 0;
 }
 
-
-
-b3BulletFile::b3BulletFile(char *memoryBuffer, int len)
-:bFile(memoryBuffer,len, "BULLET ")
+b3BulletFile::b3BulletFile(char* memoryBuffer, int len)
+	: bFile(memoryBuffer, len, "BULLET ")
 {
 	m_DnaCopy = 0;
 }
 
-
 b3BulletFile::~b3BulletFile()
 {
 	if (m_DnaCopy)
 		b3AlignedFree(m_DnaCopy);
 
-	
 	while (m_dataBlocks.size())
 	{
-		char* dataBlock = m_dataBlocks[m_dataBlocks.size()-1];
+		char* dataBlock = m_dataBlocks[m_dataBlocks.size() - 1];
 		delete[] dataBlock;
 		m_dataBlocks.pop_back();
 	}
-
 }
 
-
-
 // ----------------------------------------------------- //
 void b3BulletFile::parseData()
 {
-//	printf ("Building datablocks");
-//	printf ("Chunk size = %d",CHUNK_HEADER_LEN);
-//	printf ("File chunk size = %d",ChunkUtils::getOffset(mFlags));
+	//	printf ("Building datablocks");
+	//	printf ("Chunk size = %d",CHUNK_HEADER_LEN);
+	//	printf ("File chunk size = %d",ChunkUtils::getOffset(mFlags));
 
-	const bool brokenDNA = (mFlags&FD_BROKEN_DNA)!=0;
+	const bool brokenDNA = (mFlags & FD_BROKEN_DNA) != 0;
 
 	//const bool swap = (mFlags&FD_ENDIAN_SWAP)!=0;
-	
 
 	mDataStart = 12;
 
-	char *dataPtr = mFileBuffer+mDataStart;
+	char* dataPtr = mFileBuffer + mDataStart;
 
 	bChunkInd dataChunk;
 	dataChunk.code = 0;
 
-
 	//dataPtr += ChunkUtils::getNextBlock(&dataChunk, dataPtr, mFlags);
 	int seek = getNextBlock(&dataChunk, dataPtr, mFlags);
-	
-	
-	if (mFlags &FD_ENDIAN_SWAP) 
+
+	if (mFlags & FD_ENDIAN_SWAP)
 		swapLen(dataPtr);
 
 	//dataPtr += ChunkUtils::getOffset(mFlags);
-	char *dataPtrHead = 0;
+	char* dataPtrHead = 0;
 
 	while (dataChunk.code != B3_DNA1)
 	{
-		if (!brokenDNA || (dataChunk.code != B3_QUANTIZED_BVH_CODE) )
+		if (!brokenDNA || (dataChunk.code != B3_QUANTIZED_BVH_CODE))
 		{
-
 			// one behind
 			if (dataChunk.code == B3_SDNA) break;
 			//if (dataChunk.code == DNA1) break;
 
 			// same as (BHEAD+DATA dependency)
-			dataPtrHead = dataPtr+ChunkUtils::getOffset(mFlags);
-			if (dataChunk.dna_nr>=0)
+			dataPtrHead = dataPtr + ChunkUtils::getOffset(mFlags);
+			if (dataChunk.dna_nr >= 0)
 			{
-				char *id = readStruct(dataPtrHead, dataChunk);
+				char* id = readStruct(dataPtrHead, dataChunk);
 
 				// lookup maps
 				if (id)
@@ -171,96 +156,91 @@ void b3BulletFile::parseData()
 
 				if (dataChunk.code == B3_SOFTBODY_CODE)
 				{
-					m_softBodies.push_back((bStructHandle*) id);
+					m_softBodies.push_back((bStructHandle*)id);
 				}
-				
+
 				if (dataChunk.code == B3_RIGIDBODY_CODE)
 				{
-					m_rigidBodies.push_back((bStructHandle*) id);
+					m_rigidBodies.push_back((bStructHandle*)id);
 				}
 
 				if (dataChunk.code == B3_DYNAMICSWORLD_CODE)
 				{
-					m_dynamicsWorldInfo.push_back((bStructHandle*) id);
+					m_dynamicsWorldInfo.push_back((bStructHandle*)id);
 				}
 
 				if (dataChunk.code == B3_CONSTRAINT_CODE)
 				{
-					m_constraints.push_back((bStructHandle*) id);
+					m_constraints.push_back((bStructHandle*)id);
 				}
 
 				if (dataChunk.code == B3_QUANTIZED_BVH_CODE)
 				{
-					m_bvhs.push_back((bStructHandle*) id);
+					m_bvhs.push_back((bStructHandle*)id);
 				}
 
 				if (dataChunk.code == B3_TRIANLGE_INFO_MAP)
 				{
-					m_triangleInfoMaps.push_back((bStructHandle*) id);
+					m_triangleInfoMaps.push_back((bStructHandle*)id);
 				}
 
 				if (dataChunk.code == B3_COLLISIONOBJECT_CODE)
 				{
-					m_collisionObjects.push_back((bStructHandle*) id);
+					m_collisionObjects.push_back((bStructHandle*)id);
 				}
 
 				if (dataChunk.code == B3_SHAPE_CODE)
 				{
-					m_collisionShapes.push_back((bStructHandle*) id);
+					m_collisionShapes.push_back((bStructHandle*)id);
 				}
 
-		//		if (dataChunk.code == GLOB)
-		//		{
-		//			m_glob = (bStructHandle*) id;
-		//		}
-			} else
+				//		if (dataChunk.code == GLOB)
+				//		{
+				//			m_glob = (bStructHandle*) id;
+				//		}
+			}
+			else
 			{
 				//printf("unknown chunk\n");
 
 				mLibPointers.insert(dataChunk.oldPtr, (bStructHandle*)dataPtrHead);
 			}
-		} else
+		}
+		else
 		{
 			printf("skipping B3_QUANTIZED_BVH_CODE due to broken DNA\n");
 		}
 
-		
 		dataPtr += seek;
 
-		seek =  getNextBlock(&dataChunk, dataPtr, mFlags);
-		if (mFlags &FD_ENDIAN_SWAP) 
+		seek = getNextBlock(&dataChunk, dataPtr, mFlags);
+		if (mFlags & FD_ENDIAN_SWAP)
 			swapLen(dataPtr);
 
 		if (seek < 0)
 			break;
 	}
-
 }
 
-void	b3BulletFile::addDataBlock(char* dataBlock)
+void b3BulletFile::addDataBlock(char* dataBlock)
 {
 	m_dataBlocks.push_back(dataBlock);
-
 }
 
-
-
-
-void	b3BulletFile::writeDNA(FILE* fp)
+void b3BulletFile::writeDNA(FILE* fp)
 {
-
 	bChunkInd dataChunk;
 	dataChunk.code = B3_DNA1;
 	dataChunk.dna_nr = 0;
 	dataChunk.nr = 1;
-#ifdef B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES	
+#ifdef B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
 	if (VOID_IS_8)
 	{
 #ifdef _WIN64
 		dataChunk.len = b3s_bulletDNAlen64;
 		dataChunk.oldPtr = b3s_bulletDNAstr64;
-		fwrite(&dataChunk,sizeof(bChunkInd),1,fp);
-		fwrite(b3s_bulletDNAstr64, b3s_bulletDNAlen64,1,fp);
+		fwrite(&dataChunk, sizeof(bChunkInd), 1, fp);
+		fwrite(b3s_bulletDNAstr64, b3s_bulletDNAlen64, 1, fp);
 #else
 		b3Assert(0);
 #endif
@@ -270,43 +250,42 @@ void	b3BulletFile::writeDNA(FILE* fp)
 #ifndef _WIN64
 		dataChunk.len = b3s_bulletDNAlen;
 		dataChunk.oldPtr = b3s_bulletDNAstr;
-		fwrite(&dataChunk,sizeof(bChunkInd),1,fp);
-		fwrite(b3s_bulletDNAstr, b3s_bulletDNAlen,1,fp);
-#else//_WIN64
+		fwrite(&dataChunk, sizeof(bChunkInd), 1, fp);
+		fwrite(b3s_bulletDNAstr, b3s_bulletDNAlen, 1, fp);
+#else   //_WIN64
 		b3Assert(0);
-#endif//_WIN64
+#endif  //_WIN64
 	}
-#else//B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
+#else   //B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
 	if (VOID_IS_8)
 	{
 		dataChunk.len = b3s_bulletDNAlen64;
 		dataChunk.oldPtr = b3s_bulletDNAstr64;
-		fwrite(&dataChunk,sizeof(bChunkInd),1,fp);
-		fwrite(b3s_bulletDNAstr64, b3s_bulletDNAlen64,1,fp);
+		fwrite(&dataChunk, sizeof(bChunkInd), 1, fp);
+		fwrite(b3s_bulletDNAstr64, b3s_bulletDNAlen64, 1, fp);
 	}
 	else
 	{
 		dataChunk.len = b3s_bulletDNAlen;
 		dataChunk.oldPtr = b3s_bulletDNAstr;
-		fwrite(&dataChunk,sizeof(bChunkInd),1,fp);
-		fwrite(b3s_bulletDNAstr, b3s_bulletDNAlen,1,fp);
+		fwrite(&dataChunk, sizeof(bChunkInd), 1, fp);
+		fwrite(b3s_bulletDNAstr, b3s_bulletDNAlen, 1, fp);
 	}
-#endif//B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
+#endif  //B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
 }
 
-
-void	b3BulletFile::parse(int verboseMode)
+void b3BulletFile::parse(int verboseMode)
 {
 #ifdef B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
 	if (VOID_IS_8)
 	{
 #ifdef _WIN64
-		
+
 		if (m_DnaCopy)
 			delete m_DnaCopy;
-		m_DnaCopy = (char*)b3AlignedAlloc(b3s_bulletDNAlen64,16);
-		memcpy(m_DnaCopy,b3s_bulletDNAstr64,b3s_bulletDNAlen64);
-		parseInternal(verboseMode,(char*)b3s_bulletDNAstr64,b3s_bulletDNAlen64);
+		m_DnaCopy = (char*)b3AlignedAlloc(b3s_bulletDNAlen64, 16);
+		memcpy(m_DnaCopy, b3s_bulletDNAstr64, b3s_bulletDNAlen64);
+		parseInternal(verboseMode, (char*)b3s_bulletDNAstr64, b3s_bulletDNAlen64);
 #else
 		b3Assert(0);
 #endif
@@ -317,93 +296,91 @@ void	b3BulletFile::parse(int verboseMode)
 
 		if (m_DnaCopy)
 			delete m_DnaCopy;
-		m_DnaCopy = (char*)b3AlignedAlloc(b3s_bulletDNAlen,16);
-		memcpy(m_DnaCopy,b3s_bulletDNAstr,b3s_bulletDNAlen);
-		parseInternal(verboseMode,m_DnaCopy,b3s_bulletDNAlen);
+		m_DnaCopy = (char*)b3AlignedAlloc(b3s_bulletDNAlen, 16);
+		memcpy(m_DnaCopy, b3s_bulletDNAstr, b3s_bulletDNAlen);
+		parseInternal(verboseMode, m_DnaCopy, b3s_bulletDNAlen);
 #else
 		b3Assert(0);
 #endif
 	}
-#else//B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
+#else   //B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
 	if (VOID_IS_8)
 	{
 		if (m_DnaCopy)
 			delete m_DnaCopy;
-		m_DnaCopy = (char*)b3AlignedAlloc(b3s_bulletDNAlen64,16);
-		memcpy(m_DnaCopy,b3s_bulletDNAstr64,b3s_bulletDNAlen64);
-		parseInternal(verboseMode,m_DnaCopy,b3s_bulletDNAlen64);
+		m_DnaCopy = (char*)b3AlignedAlloc(b3s_bulletDNAlen64, 16);
+		memcpy(m_DnaCopy, b3s_bulletDNAstr64, b3s_bulletDNAlen64);
+		parseInternal(verboseMode, m_DnaCopy, b3s_bulletDNAlen64);
 	}
 	else
 	{
 		if (m_DnaCopy)
 			delete m_DnaCopy;
-		m_DnaCopy = (char*)b3AlignedAlloc(b3s_bulletDNAlen,16);
-		memcpy(m_DnaCopy,b3s_bulletDNAstr,b3s_bulletDNAlen);
-		parseInternal(verboseMode,m_DnaCopy,b3s_bulletDNAlen);
+		m_DnaCopy = (char*)b3AlignedAlloc(b3s_bulletDNAlen, 16);
+		memcpy(m_DnaCopy, b3s_bulletDNAstr, b3s_bulletDNAlen);
+		parseInternal(verboseMode, m_DnaCopy, b3s_bulletDNAlen);
 	}
-#endif//B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
-	
-	//the parsing will convert to cpu endian
-	mFlags &=~FD_ENDIAN_SWAP;
+#endif  //B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
 
-	int littleEndian= 1;
-	littleEndian= ((char*)&littleEndian)[0];
-	
-	mFileBuffer[8] = littleEndian?'v':'V';
-	
+	//the parsing will convert to cpu endian
+	mFlags &= ~FD_ENDIAN_SWAP;
+
+	int littleEndian = 1;
+	littleEndian = ((char*)&littleEndian)[0];
+
+	mFileBuffer[8] = littleEndian ? 'v' : 'V';
 }
 
 // experimental
-int		b3BulletFile::write(const char* fileName, bool fixupPointers)
+int b3BulletFile::write(const char* fileName, bool fixupPointers)
 {
-	FILE *fp = fopen(fileName, "wb");
+	FILE* fp = fopen(fileName, "wb");
 	if (fp)
 	{
-		char header[B3_SIZEOFBLENDERHEADER] ;
+		char header[B3_SIZEOFBLENDERHEADER];
 		memcpy(header, m_headerString, 7);
-		int endian= 1;
-		endian= ((char*)&endian)[0];
+		int endian = 1;
+		endian = ((char*)&endian)[0];
 
 		if (endian)
 		{
 			header[7] = '_';
-		} else
+		}
+		else
 		{
 			header[7] = '-';
 		}
 		if (VOID_IS_8)
 		{
-			header[8]='V';
-		} else
+			header[8] = 'V';
+		}
+		else
 		{
-			header[8]='v';
+			header[8] = 'v';
 		}
 
 		header[9] = '2';
 		header[10] = '7';
 		header[11] = '5';
-		
-		fwrite(header,B3_SIZEOFBLENDERHEADER,1,fp);
+
+		fwrite(header, B3_SIZEOFBLENDERHEADER, 1, fp);
 
 		writeChunks(fp, fixupPointers);
 
 		writeDNA(fp);
 
 		fclose(fp);
-		
-	} else
+	}
+	else
 	{
-		printf("Error: cannot open file %s for writing\n",fileName);
+		printf("Error: cannot open file %s for writing\n", fileName);
 		return 0;
 	}
 	return 1;
 }
 
-
-
-void	b3BulletFile::addStruct(const	char* structType,void* data, int len, void* oldPtr, int code)
+void b3BulletFile::addStruct(const char* structType, void* data, int len, void* oldPtr, int code)
 {
-	
 	bParse::bChunkInd dataChunk;
 	dataChunk.code = code;
 	dataChunk.nr = 1;
@@ -412,11 +389,11 @@ void	b3BulletFile::addStruct(const	char* structType,void* data, int len, void* o
 	dataChunk.oldPtr = oldPtr;
 
 	///Perform structure size validation
-	short* structInfo= mMemoryDNA->getStruct(dataChunk.dna_nr);
+	short* structInfo = mMemoryDNA->getStruct(dataChunk.dna_nr);
 	int elemBytes;
-	elemBytes= mMemoryDNA->getLength(structInfo[0]);
-//	int elemBytes = mMemoryDNA->getElementSize(structInfo[0],structInfo[1]);
-	assert(len==elemBytes);
+	elemBytes = mMemoryDNA->getLength(structInfo[0]);
+	//	int elemBytes = mMemoryDNA->getElementSize(structInfo[0],structInfo[1]);
+	assert(len == elemBytes);
 
 	mLibPointers.insert(dataChunk.oldPtr, (bStructHandle*)data);
 	m_chunks.push_back(dataChunk);
diff --git a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3BulletFile.h b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3BulletFile.h
index fb1b9b0dde4..ede1d378ae0 100644
--- a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3BulletFile.h
+++ b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3BulletFile.h
@@ -16,68 +16,59 @@ subject to the following restrictions:
 #ifndef B3_BULLET_FILE_H
 #define B3_BULLET_FILE_H
 
-
 #include "b3File.h"
 #include "Bullet3Common/b3AlignedObjectArray.h"
 #include "b3Defines.h"
 
 #include "Bullet3Serialize/Bullet2FileLoader/b3Serializer.h"
 
+namespace bParse
+{
+// ----------------------------------------------------- //
+class b3BulletFile : public bFile
+{
+protected:
+	char* m_DnaCopy;
 
+public:
+	b3AlignedObjectArray<bStructHandle*> m_softBodies;
 
-namespace bParse {
+	b3AlignedObjectArray<bStructHandle*> m_rigidBodies;
 
-	// ----------------------------------------------------- //
-	class b3BulletFile : public bFile
-	{
-		
+	b3AlignedObjectArray<bStructHandle*> m_collisionObjects;
 
-	protected:
-	
-		char*	m_DnaCopy;
-				
-	public:
+	b3AlignedObjectArray<bStructHandle*> m_collisionShapes;
 
-		b3AlignedObjectArray<bStructHandle*>	m_softBodies;
+	b3AlignedObjectArray<bStructHandle*> m_constraints;
 
-		b3AlignedObjectArray<bStructHandle*>	m_rigidBodies;
+	b3AlignedObjectArray<bStructHandle*> m_bvhs;
 
-		b3AlignedObjectArray<bStructHandle*>	m_collisionObjects;
+	b3AlignedObjectArray<bStructHandle*> m_triangleInfoMaps;
 
-		b3AlignedObjectArray<bStructHandle*>	m_collisionShapes;
+	b3AlignedObjectArray<bStructHandle*> m_dynamicsWorldInfo;
 
-		b3AlignedObjectArray<bStructHandle*>	m_constraints;
+	b3AlignedObjectArray<char*> m_dataBlocks;
+	b3BulletFile();
 
-		b3AlignedObjectArray<bStructHandle*>	m_bvhs;
+	b3BulletFile(const char* fileName);
 
-		b3AlignedObjectArray<bStructHandle*>	m_triangleInfoMaps;
+	b3BulletFile(char* memoryBuffer, int len);
 
-		b3AlignedObjectArray<bStructHandle*>	m_dynamicsWorldInfo;
+	virtual ~b3BulletFile();
 
-		b3AlignedObjectArray<char*>				m_dataBlocks;
-		b3BulletFile();
+	virtual void addDataBlock(char* dataBlock);
 
-		b3BulletFile(const char* fileName);
+	// experimental
+	virtual int write(const char* fileName, bool fixupPointers = false);
 
-		b3BulletFile(char *memoryBuffer, int len);
+	virtual void parse(int verboseMode);
 
-		virtual ~b3BulletFile();
+	virtual void parseData();
 
-		virtual	void	addDataBlock(char* dataBlock);
-	
+	virtual void writeDNA(FILE* fp);
 
-		// experimental
-		virtual int		write(const char* fileName, bool fixupPointers=false);
-
-		virtual	void	parse(int verboseMode);
-
-		virtual	void parseData();
-
-		virtual	void	writeDNA(FILE* fp);
-
-		void	addStruct(const char* structType,void* data, int len, void* oldPtr, int code);
-
-	};
+	void addStruct(const char* structType, void* data, int len, void* oldPtr, int code);
 };
+};  // namespace bParse
 
-#endif //B3_BULLET_FILE_H
+#endif  //B3_BULLET_FILE_H
diff --git a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Chunk.cpp b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Chunk.cpp
index c0e1bb708ce..ff75ff8cc4c 100644
--- a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Chunk.cpp
+++ b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Chunk.cpp
@@ -17,15 +17,13 @@ subject to the following restrictions:
 #include "b3Defines.h"
 #include "b3File.h"
 
-#if !defined( __CELLOS_LV2__) && !defined(__MWERKS__)
+#if !defined(__CELLOS_LV2__) && !defined(__MWERKS__)
 #include <memory.h>
 #endif
 #include <string.h>
 
-
 using namespace bParse;
 
-
 // ----------------------------------------------------- //
 short ChunkUtils::swapShort(short sht)
 {
@@ -57,19 +55,15 @@ int ChunkUtils::getOffset(int flags)
 
 	if (VOID_IS_8)
 	{
-		if (flags &FD_BITS_VARIES)
+		if (flags & FD_BITS_VARIES)
 			res = sizeof(bChunkPtr4);
 	}
 	else
 	{
-		if (flags &FD_BITS_VARIES)
+		if (flags & FD_BITS_VARIES)
 			res = sizeof(bChunkPtr8);
 	}
 	return res;
 }
 
-
-
-
-
 //eof
diff --git a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Chunk.h b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Chunk.h
index 03ecb6b4faf..c9d0f37d9ec 100644
--- a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Chunk.h
+++ b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Chunk.h
@@ -16,77 +16,69 @@ subject to the following restrictions:
 #ifndef __BCHUNK_H__
 #define __BCHUNK_H__
 
-#if defined (_WIN32) && ! defined (__MINGW32__)
-	#define b3Long64 __int64
-#elif defined (__MINGW32__)	
-	#include <stdint.h>
-	#define b3Long64 int64_t
+#if defined(_WIN32) && !defined(__MINGW32__)
+#define b3Long64 __int64
+#elif defined(__MINGW32__)
+#include <stdint.h>
+#define b3Long64 int64_t
 #else
-	#define b3Long64 long long
+#define b3Long64 long long
 #endif
 
-
-namespace bParse {
-
-
-	// ----------------------------------------------------- //
-	class bChunkPtr4
-	{
-	public:
-		bChunkPtr4(){}
-		int code;
-		int len;
-		union
-		{
-			int m_uniqueInt;
-		};
-		int dna_nr;
-		int nr;
+namespace bParse
+{
+// ----------------------------------------------------- //
+class bChunkPtr4
+{
+public:
+	bChunkPtr4() {}
+	int code;
+	int len;
+	union {
+		int m_uniqueInt;
 	};
+	int dna_nr;
+	int nr;
+};
 
-	// ----------------------------------------------------- //
-	class bChunkPtr8
-	{
-	public:
-		bChunkPtr8(){}
-		int code,  len;
-		union
-		{
-			b3Long64 oldPrev;
-			int	m_uniqueInts[2];
-		};
-		int dna_nr, nr;
+// ----------------------------------------------------- //
+class bChunkPtr8
+{
+public:
+	bChunkPtr8() {}
+	int code, len;
+	union {
+		b3Long64 oldPrev;
+		int m_uniqueInts[2];
 	};
+	int dna_nr, nr;
+};
 
-	// ----------------------------------------------------- //
-	class bChunkInd
-	{
-	public:
-		bChunkInd(){}
-		int code, len;
-		void *oldPtr;
-		int dna_nr, nr;
-	};
+// ----------------------------------------------------- //
+class bChunkInd
+{
+public:
+	bChunkInd() {}
+	int code, len;
+	void *oldPtr;
+	int dna_nr, nr;
+};
 
+// ----------------------------------------------------- //
+class ChunkUtils
+{
+public:
+	// file chunk offset
+	static int getOffset(int flags);
 
-	// ----------------------------------------------------- //
-	class ChunkUtils
-	{
-	public:
-		
-		// file chunk offset
-		static int getOffset(int flags);
+	// endian utils
+	static short swapShort(short sht);
+	static int swapInt(int inte);
+	static b3Long64 swapLong64(b3Long64 lng);
+};
 
-		// endian utils
-		static short swapShort(short sht);
-		static int swapInt(int inte);
-		static b3Long64 swapLong64(b3Long64 lng);
+const int CHUNK_HEADER_LEN = ((sizeof(bChunkInd)));
+const bool VOID_IS_8 = ((sizeof(void *) == 8));
+}  // namespace bParse
 
-	};
-
-
-	const int CHUNK_HEADER_LEN = ((sizeof(bChunkInd)));
-	const bool VOID_IS_8 = ((sizeof(void*)==8));
-}
-
-#endif//__BCHUNK_H__
+#endif  //__BCHUNK_H__
diff --git a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Common.h b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Common.h
index 2792d84033f..5884fad4d6f 100644
--- a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Common.h
+++ b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Common.h
@@ -16,24 +16,25 @@ subject to the following restrictions:
 #ifndef __BCOMMON_H__
 #define __BCOMMON_H__
 
-
 #include <assert.h>
 //#include "bLog.h"
 #include "Bullet3Common/b3AlignedObjectArray.h"
 #include "Bullet3Common/b3HashMap.h"
 
-namespace bParse {
+namespace bParse
+{
+class bMain;
+class bFileData;
+class bFile;
+class bDNA;
 
-	class bMain;
-	class bFileData;
-	class bFile;
-	class bDNA;
+// delete void* undefined
+typedef struct bStructHandle
+{
+	int unused;
+} bStructHandle;
+typedef b3AlignedObjectArray<bStructHandle*> bListBasePtr;
+typedef b3HashMap<b3HashPtr, bStructHandle*> bPtrMap;
+}  // namespace bParse
 
-	// delete void* undefined
-	typedef struct bStructHandle {int unused;}bStructHandle;
-	typedef b3AlignedObjectArray<bStructHandle*>	bListBasePtr;
-	typedef b3HashMap<b3HashPtr, bStructHandle*> bPtrMap;
-}
-
-
-#endif//__BCOMMON_H__
+#endif  //__BCOMMON_H__
diff --git a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3DNA.cpp b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3DNA.cpp
index 0fe50569221..09c8f23859a 100644
--- a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3DNA.cpp
+++ b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3DNA.cpp
@@ -23,13 +23,11 @@ subject to the following restrictions:
 //this define will force traversal of structures, to check backward (and forward) compatibility
 //#define TEST_BACKWARD_FORWARD_COMPATIBILITY
 
-
 using namespace bParse;
 
-
 // ----------------------------------------------------- //
 bDNA::bDNA()
-	:	mPtrLen(0)
+	: mPtrLen(0)
 {
 	// --
 }
@@ -43,7 +41,7 @@ bDNA::~bDNA()
 // ----------------------------------------------------- //
 bool bDNA::lessThan(bDNA *file)
 {
-	return ( m_Names.size() < file->m_Names.size());
+	return (m_Names.size() < file->m_Names.size());
 }
 
 // ----------------------------------------------------- //
@@ -53,36 +51,31 @@ char *bDNA::getName(int ind)
 	return m_Names[ind].m_name;
 }
 
-
 // ----------------------------------------------------- //
 char *bDNA::getType(int ind)
 {
-	assert(ind<=  (int)mTypes.size());
+	assert(ind <= (int)mTypes.size());
 	return mTypes[ind];
 }
 
-
 // ----------------------------------------------------- //
 short *bDNA::getStruct(int ind)
 {
-	assert(ind <=  (int)mStructs.size());
+	assert(ind <= (int)mStructs.size());
 	return mStructs[ind];
 }
 
-
 // ----------------------------------------------------- //
 short bDNA::getLength(int ind)
 {
-	assert(ind <=  (int)mTlens.size());
+	assert(ind <= (int)mTlens.size());
 	return mTlens[ind];
 }
 
-
 // ----------------------------------------------------- //
 int bDNA::getReverseType(short type)
 {
-
-	int* intPtr = mStructReverse.find(type);
+	int *intPtr = mStructReverse.find(type);
 	if (intPtr)
 		return *intPtr;
 
@@ -92,12 +85,11 @@ int bDNA::getReverseType(short type)
 // ----------------------------------------------------- //
 int bDNA::getReverseType(const char *type)
 {
-
 	b3HashString key(type);
-	int* valuePtr = mTypeLookup.find(key);
+	int *valuePtr = mTypeLookup.find(key);
 	if (valuePtr)
 		return *valuePtr;
-	
+
 	return -1;
 }
 
@@ -110,22 +102,22 @@ int bDNA::getNumStructs()
 // ----------------------------------------------------- //
 bool bDNA::flagNotEqual(int dna_nr)
 {
-	assert(dna_nr <=	(int)mCMPFlags.size());
+	assert(dna_nr <= (int)mCMPFlags.size());
 	return mCMPFlags[dna_nr] == FDF_STRUCT_NEQU;
 }
 
 // ----------------------------------------------------- //
 bool bDNA::flagEqual(int dna_nr)
 {
-	assert(dna_nr <=	(int)mCMPFlags.size());
+	assert(dna_nr <= (int)mCMPFlags.size());
 	int flag = mCMPFlags[dna_nr];
-	return  flag == FDF_STRUCT_EQU;
+	return flag == FDF_STRUCT_EQU;
 }
 
 // ----------------------------------------------------- //
 bool bDNA::flagNone(int dna_nr)
 {
-	assert(dna_nr <=	(int)mCMPFlags.size());
+	assert(dna_nr <= (int)mCMPFlags.size());
 	return mCMPFlags[dna_nr] == FDF_NONE;
 }
 
@@ -143,15 +135,15 @@ void bDNA::initRecurseCmpFlags(int iter)
 	short *oldStrc = mStructs[iter];
 	short type = oldStrc[0];
 
-	for (int i=0; i<(int)mStructs.size(); i++)
+	for (int i = 0; i < (int)mStructs.size(); i++)
 	{
-		if (i != iter && mCMPFlags[i] == FDF_STRUCT_EQU )
+		if (i != iter && mCMPFlags[i] == FDF_STRUCT_EQU)
 		{
 			short *curStruct = mStructs[i];
 			int eleLen = curStruct[1];
-			curStruct+=2;
+			curStruct += 2;
 
-			for (int j=0; j<eleLen; j++, curStruct+=2)
+			for (int j = 0; j < eleLen; j++, curStruct += 2)
 			{
 				if (curStruct[0] == type)
 				{
@@ -171,18 +163,14 @@ void bDNA::initRecurseCmpFlags(int iter)
 // ----------------------------------------------------- //
 void bDNA::initCmpFlags(bDNA *memDNA)
 {
-
-    // compare the file to memory
+	// compare the file to memory
 	// this ptr should be the file data
 
-
-	assert(!(m_Names.size() == 0));// && "SDNA empty!");
+	assert(!(m_Names.size() == 0));  // && "SDNA empty!");
 	mCMPFlags.resize(mStructs.size(), FDF_NONE);
 
-
-
 	int i;
-	for ( i=0; i<(int)mStructs.size(); i++)
+	for (i = 0; i < (int)mStructs.size(); i++)
 	{
 		short *oldStruct = mStructs[i];
 
@@ -196,7 +184,7 @@ void bDNA::initCmpFlags(bDNA *memDNA)
 
 //#define SLOW_FORWARD_COMPATIBLE 1
 #ifdef SLOW_FORWARD_COMPATIBLE
-		char* typeName = mTypes[oldLookup];
+		char *typeName = mTypes[oldLookup];
 		int newLookup = memDNA->getReverseType(typeName);
 		if (newLookup == -1)
 		{
@@ -210,71 +198,61 @@ void bDNA::initCmpFlags(bDNA *memDNA)
 		if (oldLookup < memDNA->mStructs.size())
 		{
 			short *curStruct = memDNA->mStructs[oldLookup];
-#endif	
-	
-		
+#endif
 
-			// rebuild...
-			mCMPFlags[i] = FDF_STRUCT_NEQU;
+		// rebuild...
+		mCMPFlags[i] = FDF_STRUCT_NEQU;
 
 #ifndef TEST_BACKWARD_FORWARD_COMPATIBILITY
 
-			if (curStruct[1] == oldStruct[1])
+		if (curStruct[1] == oldStruct[1])
+		{
+			// type len same ...
+			if (mTlens[oldStruct[0]] == memDNA->mTlens[curStruct[0]])
 			{
-				// type len same ...
-				if (mTlens[oldStruct[0]] == memDNA->mTlens[curStruct[0]])
+				bool isSame = true;
+				int elementLength = oldStruct[1];
+
+				curStruct += 2;
+				oldStruct += 2;
+
+				for (int j = 0; j < elementLength; j++, curStruct += 2, oldStruct += 2)
 				{
-					bool isSame = true;
-					int elementLength = oldStruct[1];
-
-
-					curStruct+=2;
-					oldStruct+=2;
-
-
-					for (int j=0; j<elementLength; j++, curStruct+=2, oldStruct+=2)
+					// type the same
+					//const char* typeFileDNA = mTypes[oldStruct[0]];
+					//const char* typeMemDNA = mTypes[curStruct[0]];
+					if (strcmp(mTypes[oldStruct[0]], memDNA->mTypes[curStruct[0]]) != 0)
 					{
-						// type the same
-						//const char* typeFileDNA = mTypes[oldStruct[0]];
-						//const char* typeMemDNA = mTypes[curStruct[0]];
-						if (strcmp(mTypes[oldStruct[0]], memDNA->mTypes[curStruct[0]])!=0)
-						{
-							isSame=false;
-							break;
-						}
-
-						// name the same
-						if (strcmp(m_Names[oldStruct[1]].m_name, memDNA->m_Names[curStruct[1]].m_name)!=0)
-						{
-							isSame=false;
-							break;
-						}
+						isSame = false;
+						break;
+					}
+
+					// name the same
+					if (strcmp(m_Names[oldStruct[1]].m_name, memDNA->m_Names[curStruct[1]].m_name) != 0)
+					{
+						isSame = false;
+						break;
 					}
-					// flag valid ==
-					if (isSame)
-						mCMPFlags[i] = FDF_STRUCT_EQU;
 				}
+				// flag valid ==
+				if (isSame)
+					mCMPFlags[i] = FDF_STRUCT_EQU;
 			}
-#endif
 		}
-	}
-
-
-
-
-
-	// recurse in
-	for ( i=0; i<(int)mStructs.size(); i++)
-	{
-		if (mCMPFlags[i] == FDF_STRUCT_NEQU)
-			initRecurseCmpFlags(i);
+#endif
 	}
 }
 
+// recurse in
+for (i = 0; i < (int)mStructs.size(); i++)
+{
+	if (mCMPFlags[i] == FDF_STRUCT_NEQU)
+		initRecurseCmpFlags(i);
+}
+}
 
-
-
-static int name_is_array(char* name, int* dim1, int* dim2) {
+static int name_is_array(char *name, int *dim1, int *dim2)
+{
 	int len = strlen(name);
 	/*fprintf(stderr,"[%s]",name);*/
 	/*if (len >= 1) {
@@ -284,58 +262,77 @@ static int name_is_array(char* name, int* dim1, int* dim2) {
 	return 0;*/
 	char *bp;
 	int num;
-	if (dim1) {
+	if (dim1)
+	{
 		*dim1 = 1;
 	}
-	if (dim2) {
+	if (dim2)
+	{
 		*dim2 = 1;
 	}
 	bp = strchr(name, '[');
-	if (!bp) {
+	if (!bp)
+	{
 		return 0;
 	}
 	num = 0;
-	while (++bp < name+len-1) {
+	while (++bp < name + len - 1)
+	{
 		const char c = *bp;
-		if (c == ']') {
+		if (c == ']')
+		{
 			break;
 		}
-		if (c <= '9' && c >= '0') {
+		if (c <= '9' && c >= '0')
+		{
 			num *= 10;
 			num += (c - '0');
-		} else {
+		}
+		else
+		{
 			printf("array parse error.\n");
 			return 0;
 		}
 	}
-	if (dim2) {
+	if (dim2)
+	{
 		*dim2 = num;
 	}
 
 	/* find second dim, if any. */
 	bp = strchr(bp, '[');
-	if (!bp) {
+	if (!bp)
+	{
 		return 1; /* at least we got the first dim. */
 	}
 	num = 0;
-	while (++bp < name+len-1) {
+	while (++bp < name + len - 1)
+	{
 		const char c = *bp;
-		if (c == ']') {
+		if (c == ']')
+		{
 			break;
 		}
-		if (c <= '9' && c >= '0') {
+		if (c <= '9' && c >= '0')
+		{
 			num *= 10;
 			num += (c - '0');
-		} else {
+		}
+		else
+		{
 			printf("array2 parse error.\n");
 			return 1;
 		}
 	}
-	if (dim1) {
-		if (dim2) {
+	if (dim1)
+	{
+		if (dim2)
+		{
 			*dim1 = *dim2;
 			*dim2 = num;
-		} else {
+		}
+		else
+		{
 			*dim1 = num;
 		}
 	}
@@ -343,14 +340,15 @@ static int name_is_array(char* name, int* dim1, int* dim2) {
 	return 1;
 }
 
-
 // ----------------------------------------------------- //
 void bDNA::init(char *data, int len, bool swap)
 {
-	int *intPtr=0;short *shtPtr=0;
-	char *cp = 0;int dataLen =0;
+	int *intPtr = 0;
+	short *shtPtr = 0;
+	char *cp = 0;
+	int dataLen = 0;
 	//long nr=0;
-	intPtr = (int*)data;
+	intPtr = (int *)data;
 
 	/*
 		SDNA (4 bytes) (magic number)
@@ -360,38 +358,35 @@ void bDNA::init(char *data, int len, bool swap)
 		<string>
 	*/
 
-	if (strncmp(data, "SDNA", 4)==0)
+	if (strncmp(data, "SDNA", 4) == 0)
 	{
 		// skip ++ NAME
-		intPtr++; intPtr++;
+		intPtr++;
+		intPtr++;
 	}
 
-
-
 	// Parse names
-	if (swap) 
+	if (swap)
 	{
 		*intPtr = ChunkUtils::swapInt(*intPtr);
 	}
 	dataLen = *intPtr;
 	intPtr++;
 
-	cp = (char*)intPtr;
+	cp = (char *)intPtr;
 	int i;
-	for ( i=0; i<dataLen; i++)
+	for (i = 0; i < dataLen; i++)
 	{
 		bNameInfo info;
 		info.m_name = cp;
 		info.m_isPointer = (info.m_name[0] == '*') || (info.m_name[1] == '*');
-		name_is_array(info.m_name,&info.m_dim0,&info.m_dim1);
+		name_is_array(info.m_name, &info.m_dim0, &info.m_dim1);
 		m_Names.push_back(info);
-		while (*cp)cp++;
+		while (*cp) cp++;
 		cp++;
 	}
 
-	
-	cp = b3AlignPointer(cp,4);
-	
+	cp = b3AlignPointer(cp, 4);
 
 	/*
 		TYPE (4 bytes)
@@ -400,26 +395,26 @@ void bDNA::init(char *data, int len, bool swap)
 		<string>
 	*/
 
-	intPtr = (int*)cp;
-	assert(strncmp(cp, "TYPE", 4)==0); intPtr++;
+	intPtr = (int *)cp;
+	assert(strncmp(cp, "TYPE", 4) == 0);
+	intPtr++;
 
-	if (swap) 
+	if (swap)
 	{
 		*intPtr = ChunkUtils::swapInt(*intPtr);
 	}
 	dataLen = *intPtr;
 	intPtr++;
 
-	cp = (char*)intPtr;
-	for ( i=0; i<dataLen; i++)
+	cp = (char *)intPtr;
+	for (i = 0; i < dataLen; i++)
 	{
 		mTypes.push_back(cp);
-		while (*cp)cp++;
+		while (*cp) cp++;
 		cp++;
 	}
 
-
-	cp = b3AlignPointer(cp,4);
+	cp = b3AlignPointer(cp, 4);
 
 	/*
 		TLEN (4 bytes)
@@ -428,13 +423,14 @@ void bDNA::init(char *data, int len, bool swap)
 	*/
 
 	// Parse type lens
-	intPtr = (int*)cp;
-	assert(strncmp(cp, "TLEN", 4)==0); intPtr++;
+	intPtr = (int *)cp;
+	assert(strncmp(cp, "TLEN", 4) == 0);
+	intPtr++;
 
 	dataLen = (int)mTypes.size();
 
-	shtPtr = (short*)intPtr;
-	for ( i=0; i<dataLen; i++, shtPtr++)
+	shtPtr = (short *)intPtr;
+	for (i = 0; i < dataLen; i++, shtPtr++)
 	{
 		if (swap)
 			shtPtr[0] = ChunkUtils::swapShort(shtPtr[0]);
@@ -454,94 +450,89 @@ void bDNA::init(char *data, int len, bool swap)
 		<namenr>
 	*/
 
-	intPtr = (int*)shtPtr;
-	cp = (char*)intPtr;
-	assert(strncmp(cp, "STRC", 4)==0); intPtr++;
+	intPtr = (int *)shtPtr;
+	cp = (char *)intPtr;
+	assert(strncmp(cp, "STRC", 4) == 0);
+	intPtr++;
 
-	if (swap) 
+	if (swap)
 	{
 		*intPtr = ChunkUtils::swapInt(*intPtr);
 	}
 	dataLen = *intPtr;
 	intPtr++;
 
-
-	shtPtr = (short*)intPtr;
-	for ( i=0; i<dataLen; i++)
+	shtPtr = (short *)intPtr;
+	for (i = 0; i < dataLen; i++)
 	{
-		mStructs.push_back (shtPtr);
+		mStructs.push_back(shtPtr);
 		if (swap)
 		{
-			shtPtr[0]= ChunkUtils::swapShort(shtPtr[0]);
-			shtPtr[1]= ChunkUtils::swapShort(shtPtr[1]);
+			shtPtr[0] = ChunkUtils::swapShort(shtPtr[0]);
+			shtPtr[1] = ChunkUtils::swapShort(shtPtr[1]);
 
 			int len = shtPtr[1];
-			shtPtr+= 2;
+			shtPtr += 2;
 
-			for (int a=0; a<len; a++, shtPtr+=2)
+			for (int a = 0; a < len; a++, shtPtr += 2)
 			{
-				shtPtr[0]= ChunkUtils::swapShort(shtPtr[0]);
-				shtPtr[1]= ChunkUtils::swapShort(shtPtr[1]);
+				shtPtr[0] = ChunkUtils::swapShort(shtPtr[0]);
+				shtPtr[1] = ChunkUtils::swapShort(shtPtr[1]);
 			}
 		}
 		else
-			shtPtr+= (2*shtPtr[1])+2;
+			shtPtr += (2 * shtPtr[1]) + 2;
 	}
 
-
 	// build reverse lookups
-	for ( i=0; i<(int)mStructs.size(); i++)
+	for (i = 0; i < (int)mStructs.size(); i++)
 	{
 		short *strc = mStructs.at(i);
-		if (!mPtrLen && strcmp(mTypes[strc[0]],"ListBase")==0)
+		if (!mPtrLen && strcmp(mTypes[strc[0]], "ListBase") == 0)
 		{
-			mPtrLen = mTlens[strc[0]]/2;
+			mPtrLen = mTlens[strc[0]] / 2;
 		}
 
 		mStructReverse.insert(strc[0], i);
-		mTypeLookup.insert(b3HashString(mTypes[strc[0]]),i);
+		mTypeLookup.insert(b3HashString(mTypes[strc[0]]), i);
 	}
 }
 
-
 // ----------------------------------------------------- //
-int bDNA::getArraySize(char* string)
+int bDNA::getArraySize(char *string)
 {
 	int ret = 1;
 	int len = strlen(string);
 
-	
-	char* next = 0;
-	for (int i=0; i<len; i++)
+	char *next = 0;
+	for (int i = 0; i < len; i++)
 	{
 		char c = string[i];
 
 		if (c == '[')
-			next = &string[i+1];
-		else if (c==']')
+			next = &string[i + 1];
+		else if (c == ']')
 			if (next)
 				ret *= atoi(next);
 	}
 
-//	print (string << ' ' << ret);
+	//	print (string << ' ' << ret);
 	return ret;
 }
 
-
 void bDNA::dumpTypeDefinitions()
 {
 	int i;
 
 	int numTypes = mTypes.size();
-	
-	for (i=0;i<numTypes;i++)
-	{
 
+	for (i = 0; i < numTypes; i++)
+	{
 	}
 
-	for ( i=0; i<(int)mStructs.size(); i++)
+	for (i = 0; i < (int)mStructs.size(); i++)
 	{
-		int totalBytes=0;
+		int totalBytes = 0;
 		short *oldStruct = mStructs[i];
 
 		int oldLookup = getReverseType(oldStruct[0]);
@@ -551,44 +542,46 @@ void bDNA::dumpTypeDefinitions()
 			continue;
 		}
 
-		short* newStruct = mStructs[oldLookup];
-		char* typeName = mTypes[newStruct[0]];
-		printf("%3d: %s ",i,typeName);
-		
+		short *newStruct = mStructs[oldLookup];
+		char *typeName = mTypes[newStruct[0]];
+		printf("%3d: %s ", i, typeName);
+
 		//char *name = mNames[oldStruct[1]];
 		int len = oldStruct[1];
-		printf(" (%d fields) ",len);
-		oldStruct+=2;
+		printf(" (%d fields) ", len);
+		oldStruct += 2;
 
 		printf("{");
 		int j;
-		for (j=0; j<len; ++j,oldStruct+=2) {
-			const char* name = m_Names[oldStruct[1]].m_name;
-			printf("%s %s",	mTypes[oldStruct[0]],name);
-			int elemNumBytes= 0;
+		for (j = 0; j < len; ++j, oldStruct += 2)
+		{
+			const char *name = m_Names[oldStruct[1]].m_name;
+			printf("%s %s", mTypes[oldStruct[0]], name);
+			int elemNumBytes = 0;
 			int arrayDimensions = getArraySizeNew(oldStruct[1]);
 
 			if (m_Names[oldStruct[1]].m_isPointer)
 			{
 				elemNumBytes = VOID_IS_8 ? 8 : 4;
-			} else
+			}
+			else
 			{
 				elemNumBytes = getLength(oldStruct[0]);
 			}
-			printf(" /* %d bytes */",elemNumBytes*arrayDimensions);
-			
-			if (j == len-1) {
+			printf(" /* %d bytes */", elemNumBytes * arrayDimensions);
+
+			if (j == len - 1)
+			{
 				printf(";}");
-			} else {
+			}
+			else
+			{
 				printf("; ");
 			}
-			totalBytes+=elemNumBytes*arrayDimensions;
+			totalBytes += elemNumBytes * arrayDimensions;
 		}
-		printf("\ntotalBytes=%d\n\n",totalBytes);
-
+		printf("\ntotalBytes=%d\n\n", totalBytes);
 	}
-	
-
 
 #if 0
 	/* dump out display of types and their sizes */
@@ -618,12 +611,6 @@ void bDNA::dumpTypeDefinitions()
 		}
 	}
 #endif
-
 }
 
-
-
-
 //eof
-
-
diff --git a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3DNA.h b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3DNA.h
index 6e60087cce3..ca6004d960d 100644
--- a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3DNA.h
+++ b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3DNA.h
@@ -16,95 +16,86 @@ subject to the following restrictions:
 #ifndef __BDNA_H__
 #define __BDNA_H__
 
-
 #include "b3Common.h"
 
-namespace bParse {
+namespace bParse
+{
+struct bNameInfo
+{
+	char *m_name;
+	bool m_isPointer;
+	int m_dim0;
+	int m_dim1;
+};
 
-	struct	bNameInfo
+class bDNA
+{
+public:
+	bDNA();
+	~bDNA();
+
+	void init(char *data, int len, bool swap = false);
+
+	int getArraySize(char *str);
+	int getArraySizeNew(short name)
 	{
-		char*	m_name;
-		bool	m_isPointer;
-		int		m_dim0;
-		int		m_dim1;
+		const bNameInfo &nameInfo = m_Names[name];
+		return nameInfo.m_dim0 * nameInfo.m_dim1;
+	}
+	int getElementSize(short type, short name)
+	{
+		const bNameInfo &nameInfo = m_Names[name];
+		int size = nameInfo.m_isPointer ? mPtrLen * nameInfo.m_dim0 * nameInfo.m_dim1 : mTlens[type] * nameInfo.m_dim0 * nameInfo.m_dim1;
+		return size;
+	}
+
+	int getNumNames() const
+	{
+		return m_Names.size();
+	}
+
+	char *getName(int ind);
+	char *getType(int ind);
+	short *getStruct(int ind);
+	short getLength(int ind);
+	int getReverseType(short type);
+	int getReverseType(const char *type);
+
+	int getNumStructs();
+
+	//
+	bool lessThan(bDNA *other);
+
+	void initCmpFlags(bDNA *memDNA);
+	bool flagNotEqual(int dna_nr);
+	bool flagEqual(int dna_nr);
+	bool flagNone(int dna_nr);
+
+	int getPointerSize();
+
+	void dumpTypeDefinitions();
+
+private:
+	enum FileDNAFlags
+	{
+		FDF_NONE = 0,
+		FDF_STRUCT_NEQU,
+		FDF_STRUCT_EQU
 	};
 
-	class bDNA
-	{
-	public:
-		bDNA();
-		~bDNA();
+	void initRecurseCmpFlags(int i);
 
-		void init(char *data, int len, bool swap=false);
+	b3AlignedObjectArray<int> mCMPFlags;
 
-		int getArraySize(char* str);
-		int getArraySizeNew(short name)
-		{
-			const bNameInfo& nameInfo = m_Names[name];
-			return nameInfo.m_dim0*nameInfo.m_dim1;
-		}
-		int getElementSize(short type, short name)
-		{
-			const bNameInfo& nameInfo = m_Names[name];
-			int size = nameInfo.m_isPointer ? mPtrLen*nameInfo.m_dim0*nameInfo.m_dim1 : mTlens[type]*nameInfo.m_dim0*nameInfo.m_dim1;
-			return size;
-		}
+	b3AlignedObjectArray<bNameInfo> m_Names;
+	b3AlignedObjectArray<char *> mTypes;
+	b3AlignedObjectArray<short *> mStructs;
+	b3AlignedObjectArray<short> mTlens;
+	b3HashMap<b3HashInt, int> mStructReverse;
+	b3HashMap<b3HashString, int> mTypeLookup;
 
-		int	getNumNames() const
-		{
-			return m_Names.size();
-		}
+	int mPtrLen;
+};
+}  // namespace bParse
 
-		char *getName(int ind);
-		char *getType(int ind);
-		short *getStruct(int ind);
-		short getLength(int ind);
-		int getReverseType(short type);
-		int getReverseType(const char *type);
-
-
-		int getNumStructs();
-
-		//
-		bool lessThan(bDNA* other);
-
-		void initCmpFlags(bDNA *memDNA);
-		bool flagNotEqual(int dna_nr);
-		bool flagEqual(int dna_nr);
-		bool flagNone(int dna_nr);
-
-
-		int getPointerSize();
-
-		void	dumpTypeDefinitions();
-
-	
-	private:
-		enum FileDNAFlags
-		{
-			FDF_NONE=0,
-			FDF_STRUCT_NEQU,
-			FDF_STRUCT_EQU
-		};
-
-		void initRecurseCmpFlags(int i);
-
-		b3AlignedObjectArray<int>			mCMPFlags;
-
-		b3AlignedObjectArray<bNameInfo>			m_Names;
-		b3AlignedObjectArray<char*>			mTypes;
-		b3AlignedObjectArray<short*>			mStructs;
-		b3AlignedObjectArray<short>			mTlens;
-		b3HashMap<b3HashInt, int>			mStructReverse;
-		b3HashMap<b3HashString,int>	mTypeLookup;
-
-		int							mPtrLen;
-		
-
-
-
-	};
-}
-
-
-#endif//__BDNA_H__
+#endif  //__BDNA_H__
diff --git a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Defines.h b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Defines.h
index 8f28d3c4417..0524c94db18 100644
--- a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Defines.h
+++ b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Defines.h
@@ -19,118 +19,131 @@
 #ifndef __B_DEFINES_H__
 #define __B_DEFINES_H__
 
-
 // MISC defines, see BKE_global.h, BKE_utildefines.h
 #define B3_SIZEOFBLENDERHEADER 12
 
-
 // ------------------------------------------------------------
-#if defined(__sgi) || defined (__sparc) || defined (__sparc__) || defined (__PPC__) || defined (__ppc__) || defined (__BIG_ENDIAN__)
-#	define B3_MAKE_ID(a,b,c,d) ( (int)(a)<<24 | (int)(b)<<16 | (c)<<8 | (d) )
+#if defined(__sgi) || defined(__sparc) || defined(__sparc__) || defined(__PPC__) || defined(__ppc__) || defined(__BIG_ENDIAN__)
+#define B3_MAKE_ID(a, b, c, d) ((int)(a) << 24 | (int)(b) << 16 | (c) << 8 | (d))
 #else
-#	define B3_MAKE_ID(a,b,c,d) ( (int)(d)<<24 | (int)(c)<<16 | (b)<<8 | (a) )
-#endif
-
-
-// ------------------------------------------------------------
-#if defined(__sgi) || defined(__sparc) || defined(__sparc__) || defined (__PPC__) || defined (__ppc__) || defined (__BIG_ENDIAN__)
-#	define B3_MAKE_ID2(c, d) ( (c)<<8 | (d) )
-#else
-#	define B3_MAKE_ID2(c, d) ( (d)<<8 | (c) )
+#define B3_MAKE_ID(a, b, c, d) ((int)(d) << 24 | (int)(c) << 16 | (b) << 8 | (a))
 #endif
 
 // ------------------------------------------------------------
-#define B3_ID_SCE		B3_MAKE_ID2('S', 'C')
-#define B3_ID_LI		B3_MAKE_ID2('L', 'I')
-#define B3_ID_OB		B3_MAKE_ID2('O', 'B')
-#define B3_ID_ME		B3_MAKE_ID2('M', 'E')
-#define B3_ID_CU		B3_MAKE_ID2('C', 'U')
-#define B3_ID_MB		B3_MAKE_ID2('M', 'B')
-#define B3_ID_MA		B3_MAKE_ID2('M', 'A')
-#define B3_ID_TE		B3_MAKE_ID2('T', 'E')
-#define B3_ID_IM		B3_MAKE_ID2('I', 'M')
-#define B3_ID_IK		B3_MAKE_ID2('I', 'K')
-#define B3_ID_WV		B3_MAKE_ID2('W', 'V')
-#define B3_ID_LT		B3_MAKE_ID2('L', 'T')
-#define B3_ID_SE		B3_MAKE_ID2('S', 'E')
-#define B3_ID_LF		B3_MAKE_ID2('L', 'F')
-#define B3_ID_LA		B3_MAKE_ID2('L', 'A')
-#define B3_ID_CA		B3_MAKE_ID2('C', 'A')
-#define B3_ID_IP		B3_MAKE_ID2('I', 'P')
-#define B3_ID_KE		B3_MAKE_ID2('K', 'E')
-#define B3_ID_WO		B3_MAKE_ID2('W', 'O')
-#define B3_ID_SCR		B3_MAKE_ID2('S', 'R')
-#define B3_ID_VF		B3_MAKE_ID2('V', 'F')
-#define B3_ID_TXT		B3_MAKE_ID2('T', 'X')
-#define B3_ID_SO		B3_MAKE_ID2('S', 'O')
-#define B3_ID_SAMPLE	B3_MAKE_ID2('S', 'A')
-#define B3_ID_GR		B3_MAKE_ID2('G', 'R')
-#define B3_ID_ID		B3_MAKE_ID2('I', 'D')
-#define B3_ID_AR		B3_MAKE_ID2('A', 'R')
-#define B3_ID_AC		B3_MAKE_ID2('A', 'C')
-#define B3_ID_SCRIPT	B3_MAKE_ID2('P', 'Y')
-#define B3_ID_FLUIDSIM	B3_MAKE_ID2('F', 'S')
-#define B3_ID_NT		B3_MAKE_ID2('N', 'T')
-#define B3_ID_BR		B3_MAKE_ID2('B', 'R')
-
-
-#define B3_ID_SEQ		B3_MAKE_ID2('S', 'Q')
-#define B3_ID_CO		B3_MAKE_ID2('C', 'O')
-#define B3_ID_PO		B3_MAKE_ID2('A', 'C')
-#define B3_ID_NLA		B3_MAKE_ID2('N', 'L')
-
-#define B3_ID_VS		B3_MAKE_ID2('V', 'S')
-#define B3_ID_VN		B3_MAKE_ID2('V', 'N')
-
+#if defined(__sgi) || defined(__sparc) || defined(__sparc__) || defined(__PPC__) || defined(__ppc__) || defined(__BIG_ENDIAN__)
+#define B3_MAKE_ID2(c, d) ((c) << 8 | (d))
+#else
+#define B3_MAKE_ID2(c, d) ((d) << 8 | (c))
+#endif
 
 // ------------------------------------------------------------
-#define B3_FORM B3_MAKE_ID('F','O','R','M')
-#define B3_DDG1 B3_MAKE_ID('3','D','G','1')
-#define B3_DDG2 B3_MAKE_ID('3','D','G','2')
-#define B3_DDG3 B3_MAKE_ID('3','D','G','3')
-#define B3_DDG4 B3_MAKE_ID('3','D','G','4')
-#define B3_GOUR B3_MAKE_ID('G','O','U','R')
-#define B3_BLEN B3_MAKE_ID('B','L','E','N')
-#define B3_DER_ B3_MAKE_ID('D','E','R','_')
-#define B3_V100 B3_MAKE_ID('V','1','0','0')
-#define B3_DATA B3_MAKE_ID('D','A','T','A')
-#define B3_GLOB B3_MAKE_ID('G','L','O','B')
-#define B3_IMAG B3_MAKE_ID('I','M','A','G')
-#define B3_TEST B3_MAKE_ID('T','E','S','T')
-#define B3_USER B3_MAKE_ID('U','S','E','R')
+#define B3_ID_SCE B3_MAKE_ID2('S', 'C')
+#define B3_ID_LI B3_MAKE_ID2('L', 'I')
+#define B3_ID_OB B3_MAKE_ID2('O', 'B')
+#define B3_ID_ME B3_MAKE_ID2('M', 'E')
+#define B3_ID_CU B3_MAKE_ID2('C', 'U')
+#define B3_ID_MB B3_MAKE_ID2('M', 'B')
+#define B3_ID_MA B3_MAKE_ID2('M', 'A')
+#define B3_ID_TE B3_MAKE_ID2('T', 'E')
+#define B3_ID_IM B3_MAKE_ID2('I', 'M')
+#define B3_ID_IK B3_MAKE_ID2('I', 'K')
+#define B3_ID_WV B3_MAKE_ID2('W', 'V')
+#define B3_ID_LT B3_MAKE_ID2('L', 'T')
+#define B3_ID_SE B3_MAKE_ID2('S', 'E')
+#define B3_ID_LF B3_MAKE_ID2('L', 'F')
+#define B3_ID_LA B3_MAKE_ID2('L', 'A')
+#define B3_ID_CA B3_MAKE_ID2('C', 'A')
+#define B3_ID_IP B3_MAKE_ID2('I', 'P')
+#define B3_ID_KE B3_MAKE_ID2('K', 'E')
+#define B3_ID_WO B3_MAKE_ID2('W', 'O')
+#define B3_ID_SCR B3_MAKE_ID2('S', 'R')
+#define B3_ID_VF B3_MAKE_ID2('V', 'F')
+#define B3_ID_TXT B3_MAKE_ID2('T', 'X')
+#define B3_ID_SO B3_MAKE_ID2('S', 'O')
+#define B3_ID_SAMPLE B3_MAKE_ID2('S', 'A')
+#define B3_ID_GR B3_MAKE_ID2('G', 'R')
+#define B3_ID_ID B3_MAKE_ID2('I', 'D')
+#define B3_ID_AR B3_MAKE_ID2('A', 'R')
+#define B3_ID_AC B3_MAKE_ID2('A', 'C')
+#define B3_ID_SCRIPT B3_MAKE_ID2('P', 'Y')
+#define B3_ID_FLUIDSIM B3_MAKE_ID2('F', 'S')
+#define B3_ID_NT B3_MAKE_ID2('N', 'T')
+#define B3_ID_BR B3_MAKE_ID2('B', 'R')
 
+#define B3_ID_SEQ B3_MAKE_ID2('S', 'Q')
+#define B3_ID_CO B3_MAKE_ID2('C', 'O')
+#define B3_ID_PO B3_MAKE_ID2('A', 'C')
+#define B3_ID_NLA B3_MAKE_ID2('N', 'L')
+
+#define B3_ID_VS B3_MAKE_ID2('V', 'S')
+#define B3_ID_VN B3_MAKE_ID2('V', 'N')
 
 // ------------------------------------------------------------
-#define B3_DNA1 B3_MAKE_ID('D','N','A','1')
-#define B3_REND B3_MAKE_ID('R','E','N','D')
-#define B3_ENDB B3_MAKE_ID('E','N','D','B')
-#define B3_NAME B3_MAKE_ID('N','A','M','E')
-#define B3_SDNA B3_MAKE_ID('S','D','N','A')
-#define B3_TYPE B3_MAKE_ID('T','Y','P','E')
-#define B3_TLEN B3_MAKE_ID('T','L','E','N')
-#define B3_STRC B3_MAKE_ID('S','T','R','C')
-
+#define B3_FORM B3_MAKE_ID('F', 'O', 'R', 'M')
+#define B3_DDG1 B3_MAKE_ID('3', 'D', 'G', '1')
+#define B3_DDG2 B3_MAKE_ID('3', 'D', 'G', '2')
+#define B3_DDG3 B3_MAKE_ID('3', 'D', 'G', '3')
+#define B3_DDG4 B3_MAKE_ID('3', 'D', 'G', '4')
+#define B3_GOUR B3_MAKE_ID('G', 'O', 'U', 'R')
+#define B3_BLEN B3_MAKE_ID('B', 'L', 'E', 'N')
+#define B3_DER_ B3_MAKE_ID('D', 'E', 'R', '_')
+#define B3_V100 B3_MAKE_ID('V', '1', '0', '0')
+#define B3_DATA B3_MAKE_ID('D', 'A', 'T', 'A')
+#define B3_GLOB B3_MAKE_ID('G', 'L', 'O', 'B')
+#define B3_IMAG B3_MAKE_ID('I', 'M', 'A', 'G')
+#define B3_TEST B3_MAKE_ID('T', 'E', 'S', 'T')
+#define B3_USER B3_MAKE_ID('U', 'S', 'E', 'R')
 
 // ------------------------------------------------------------
-#define B3_SWITCH_INT(a) { \
-    char s_i, *p_i; \
-    p_i= (char *)&(a); \
-    s_i=p_i[0]; p_i[0]=p_i[3]; p_i[3]=s_i; \
-    s_i=p_i[1]; p_i[1]=p_i[2]; p_i[2]=s_i; }
+#define B3_DNA1 B3_MAKE_ID('D', 'N', 'A', '1')
+#define B3_REND B3_MAKE_ID('R', 'E', 'N', 'D')
+#define B3_ENDB B3_MAKE_ID('E', 'N', 'D', 'B')
+#define B3_NAME B3_MAKE_ID('N', 'A', 'M', 'E')
+#define B3_SDNA B3_MAKE_ID('S', 'D', 'N', 'A')
+#define B3_TYPE B3_MAKE_ID('T', 'Y', 'P', 'E')
+#define B3_TLEN B3_MAKE_ID('T', 'L', 'E', 'N')
+#define B3_STRC B3_MAKE_ID('S', 'T', 'R', 'C')
 
 // ------------------------------------------------------------
-#define B3_SWITCH_SHORT(a)	{ \
-    char s_i, *p_i; \
-	p_i= (char *)&(a); \
-	s_i=p_i[0]; p_i[0]=p_i[1]; p_i[1]=s_i; }
+#define B3_SWITCH_INT(a)    \
+	{                       \
+		char s_i, *p_i;     \
+		p_i = (char *)&(a); \
+		s_i = p_i[0];       \
+		p_i[0] = p_i[3];    \
+		p_i[3] = s_i;       \
+		s_i = p_i[1];       \
+		p_i[1] = p_i[2];    \
+		p_i[2] = s_i;       \
+	}
 
 // ------------------------------------------------------------
-#define B3_SWITCH_LONGINT(a) { \
-    char s_i, *p_i; \
-    p_i= (char *)&(a);  \
-    s_i=p_i[0]; p_i[0]=p_i[7]; p_i[7]=s_i; \
-    s_i=p_i[1]; p_i[1]=p_i[6]; p_i[6]=s_i; \
-    s_i=p_i[2]; p_i[2]=p_i[5]; p_i[5]=s_i; \
-    s_i=p_i[3]; p_i[3]=p_i[4]; p_i[4]=s_i; }
+#define B3_SWITCH_SHORT(a)  \
+	{                       \
+		char s_i, *p_i;     \
+		p_i = (char *)&(a); \
+		s_i = p_i[0];       \
+		p_i[0] = p_i[1];    \
+		p_i[1] = s_i;       \
+	}
 
-#endif//__B_DEFINES_H__
+// ------------------------------------------------------------
+#define B3_SWITCH_LONGINT(a) \
+	{                        \
+		char s_i, *p_i;      \
+		p_i = (char *)&(a);  \
+		s_i = p_i[0];        \
+		p_i[0] = p_i[7];     \
+		p_i[7] = s_i;        \
+		s_i = p_i[1];        \
+		p_i[1] = p_i[6];     \
+		p_i[6] = s_i;        \
+		s_i = p_i[2];        \
+		p_i[2] = p_i[5];     \
+		p_i[5] = s_i;        \
+		s_i = p_i[3];        \
+		p_i[3] = p_i[4];     \
+		p_i[4] = s_i;        \
+	}
+
+#endif  //__B_DEFINES_H__
diff --git a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3File.cpp b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3File.cpp
index 432f7fc2b4b..145de62db32 100644
--- a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3File.cpp
+++ b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3File.cpp
@@ -29,39 +29,38 @@ subject to the following restrictions:
 using namespace bParse;
 #define MAX_STRLEN 1024
 
-const char* getCleanName(const char* memName, char* buffer)
+const char *getCleanName(const char *memName, char *buffer)
 {
 	int slen = strlen(memName);
-	assert(slen<MAX_STRLEN);
-	slen=b3Min(slen,MAX_STRLEN);
-	for (int i=0;i<slen;i++)
+	assert(slen < MAX_STRLEN);
+	slen = b3Min(slen, MAX_STRLEN);
+	for (int i = 0; i < slen; i++)
 	{
-		if (memName[i]==']'||memName[i]=='[')
+		if (memName[i] == ']' || memName[i] == '[')
 		{
-			buffer[i] = 0;//'_';
-		} else
+			buffer[i] = 0;  //'_';
+		}
+		else
 		{
 			buffer[i] = memName[i];
 		}
 	}
-	buffer[slen]=0;
+	buffer[slen] = 0;
 	return buffer;
 }
 
-
-
 // ----------------------------------------------------- //
 bFile::bFile(const char *filename, const char headerString[7])
-	:	mOwnsBuffer(true),
-		mFileBuffer(0),
-		mFileLen(0),
-		mVersion(0),
-		mDataStart(0),
-		mFileDNA(0),
-		mMemoryDNA(0),
-		mFlags(FD_INVALID)
+	: mOwnsBuffer(true),
+	  mFileBuffer(0),
+	  mFileLen(0),
+	  mVersion(0),
+	  mDataStart(0),
+	  mFileDNA(0),
+	  mMemoryDNA(0),
+	  mFlags(FD_INVALID)
 {
-	for (int i=0;i<7;i++)
+	for (int i = 0; i < 7; i++)
 	{
 		m_headerString[i] = headerString[i];
 	}
@@ -73,7 +72,7 @@ bFile::bFile(const char *filename, const char headerString[7])
 		mFileLen = ftell(fp);
 		fseek(fp, 0L, SEEK_SET);
 
-		mFileBuffer = (char*)malloc(mFileLen+1);
+		mFileBuffer = (char *)malloc(mFileLen + 1);
 		int bytesRead;
 		bytesRead = fread(mFileBuffer, mFileLen, 1, fp);
 
@@ -81,22 +80,21 @@ bFile::bFile(const char *filename, const char headerString[7])
 
 		//
 		parseHeader();
-
 	}
 }
 
 // ----------------------------------------------------- //
-bFile::bFile( char *memoryBuffer, int len, const char headerString[7])
-:	mOwnsBuffer(false),
-	mFileBuffer(0),
-		mFileLen(0),
-		mVersion(0),
-		mDataStart(0),
-		mFileDNA(0),
-		mMemoryDNA(0),
-		mFlags(FD_INVALID)
+bFile::bFile(char *memoryBuffer, int len, const char headerString[7])
+	: mOwnsBuffer(false),
+	  mFileBuffer(0),
+	  mFileLen(0),
+	  mVersion(0),
+	  mDataStart(0),
+	  mFileDNA(0),
+	  mMemoryDNA(0),
+	  mFlags(FD_INVALID)
 {
-	for (int i=0;i<7;i++)
+	for (int i = 0; i < 7; i++)
 	{
 		m_headerString[i] = headerString[i];
 	}
@@ -104,10 +102,8 @@ bFile::bFile( char *memoryBuffer, int len, const char headerString[7])
 	mFileLen = len;
 
 	parseHeader();
-
 }
 
-
 // ----------------------------------------------------- //
 bFile::~bFile()
 {
@@ -117,15 +113,10 @@ bFile::~bFile()
 		mFileBuffer = 0;
 	}
 
-
 	delete mMemoryDNA;
 	delete mFileDNA;
 }
 
-
-
-
-
 // ----------------------------------------------------- //
 void bFile::parseHeader()
 {
@@ -133,11 +124,11 @@ void bFile::parseHeader()
 		return;
 
 	char *blenderBuf = mFileBuffer;
-	char header[B3_SIZEOFBLENDERHEADER+1] ;
+	char header[B3_SIZEOFBLENDERHEADER + 1];
 	memcpy(header, blenderBuf, B3_SIZEOFBLENDERHEADER);
-	header[B3_SIZEOFBLENDERHEADER]='\0';
+	header[B3_SIZEOFBLENDERHEADER] = '\0';
 
-	if (strncmp(header, m_headerString, 6)!=0)
+	if (strncmp(header, m_headerString, 6) != 0)
 	{
 		memcpy(header, m_headerString, B3_SIZEOFBLENDERHEADER);
 		return;
@@ -148,35 +139,34 @@ void bFile::parseHeader()
 		mFlags |= FD_DOUBLE_PRECISION;
 	}
 
-	char *ver = header+9;
+	char *ver = header + 9;
 	mVersion = atoi(ver);
 	if (mVersion <= 241)
 	{
 		//printf("Warning, %d not fully tested : <= 242\n", mVersion);
 	}
 
-	int littleEndian= 1;
-	littleEndian= ((char*)&littleEndian)[0];
+	int littleEndian = 1;
+	littleEndian = ((char *)&littleEndian)[0];
 
 	// swap ptr sizes...
-	if (header[7]=='-')
+	if (header[7] == '-')
 	{
 		mFlags |= FD_FILE_64;
 		if (!VOID_IS_8)
 			mFlags |= FD_BITS_VARIES;
 	}
-	else if (VOID_IS_8) mFlags |= FD_BITS_VARIES;
+	else if (VOID_IS_8)
+		mFlags |= FD_BITS_VARIES;
 
 	// swap endian...
-	if (header[8]=='V')
+	if (header[8] == 'V')
 	{
-		if (littleEndian ==1)
+		if (littleEndian == 1)
 			mFlags |= FD_ENDIAN_SWAP;
 	}
-	else
-		if (littleEndian==0)
-			mFlags |= FD_ENDIAN_SWAP;
-
+	else if (littleEndian == 0)
+		mFlags |= FD_ENDIAN_SWAP;
 
 	mFlags |= FD_OK;
 }
@@ -184,13 +174,13 @@ void bFile::parseHeader()
 // ----------------------------------------------------- //
 bool bFile::ok()
 {
-	return (mFlags &FD_OK)!=0;
+	return (mFlags & FD_OK) != 0;
 }
 
 // ----------------------------------------------------- //
-void bFile::parseInternal(int verboseMode, char* memDna,int memDnaLength)
+void bFile::parseInternal(int verboseMode, char *memDna, int memDnaLength)
 {
-	if ( (mFlags &FD_OK) ==0)
+	if ((mFlags & FD_OK) == 0)
 		return;
 
 	char *blenderData = mFileBuffer;
@@ -198,38 +188,40 @@ void bFile::parseInternal(int verboseMode, char* memDna,int memDnaLength)
 	dna.oldPtr = 0;
 
 	char *tempBuffer = blenderData;
-	for (int i=0; i<mFileLen; i++)
+	for (int i = 0; i < mFileLen; i++)
 	{
 		// looking for the data's starting position
 		// and the start of SDNA decls
 
-		if (!mDataStart && strncmp(tempBuffer, "REND", 4)==0)
+		if (!mDataStart && strncmp(tempBuffer, "REND", 4) == 0)
 			mDataStart = i;
 
-		if (strncmp(tempBuffer, "DNA1", 4)==0)
+		if (strncmp(tempBuffer, "DNA1", 4) == 0)
 		{
 			// read the DNA1 block and extract SDNA
 			if (getNextBlock(&dna, tempBuffer, mFlags) > 0)
 			{
-				if (strncmp((tempBuffer + ChunkUtils::getOffset(mFlags)), "SDNANAME", 8) ==0)
+				if (strncmp((tempBuffer + ChunkUtils::getOffset(mFlags)), "SDNANAME", 8) == 0)
 					dna.oldPtr = (tempBuffer + ChunkUtils::getOffset(mFlags));
-				else dna.oldPtr = 0;
+				else
+					dna.oldPtr = 0;
 			}
-			else dna.oldPtr = 0;
+			else
+				dna.oldPtr = 0;
 		}
 		// Some Bullet files are missing the DNA1 block
 		// In Blender it's DNA1 + ChunkUtils::getOffset() + SDNA + NAME
 		// In Bullet tests its SDNA + NAME
-		else if (strncmp(tempBuffer, "SDNANAME", 8) ==0)
+		else if (strncmp(tempBuffer, "SDNANAME", 8) == 0)
 		{
 			dna.oldPtr = blenderData + i;
-			dna.len = mFileLen-i;
+			dna.len = mFileLen - i;
 
 			// Also no REND block, so exit now.
-			if (mVersion==276) break;
+			if (mVersion == 276) break;
 		}
 
-        if (mDataStart && dna.oldPtr) break;
+		if (mDataStart && dna.oldPtr) break;
 		tempBuffer++;
 	}
 	if (!dna.oldPtr || !dna.len)
@@ -239,43 +231,35 @@ void bFile::parseInternal(int verboseMode, char* memDna,int memDnaLength)
 		return;
 	}
 
-
 	mFileDNA = new bDNA();
 
-
 	///mFileDNA->init will convert part of DNA file endianness to current CPU endianness if necessary
-	mFileDNA->init((char*)dna.oldPtr, dna.len, (mFlags & FD_ENDIAN_SWAP)!=0);
+	mFileDNA->init((char *)dna.oldPtr, dna.len, (mFlags & FD_ENDIAN_SWAP) != 0);
 
-
-	if (mVersion==276)
+	if (mVersion == 276)
 	{
 		int i;
-		for (i=0;i<mFileDNA->getNumNames();i++)
+		for (i = 0; i < mFileDNA->getNumNames(); i++)
 		{
-			if (strcmp(mFileDNA->getName(i),"int")==0)
+			if (strcmp(mFileDNA->getName(i), "int") == 0)
 			{
 				mFlags |= FD_BROKEN_DNA;
 			}
 		}
-		if ((mFlags&FD_BROKEN_DNA)!=0)
+		if ((mFlags & FD_BROKEN_DNA) != 0)
 		{
 			//printf("warning: fixing some broken DNA version\n");
 		}
 	}
 
-
-
 	if (verboseMode & FD_VERBOSE_DUMP_DNA_TYPE_DEFINITIONS)
 		mFileDNA->dumpTypeDefinitions();
 
 	mMemoryDNA = new bDNA();
-	int littleEndian= 1;
-	littleEndian= ((char*)&littleEndian)[0];
-
-	mMemoryDNA->init(memDna,memDnaLength,littleEndian==0);
-
-
+	int littleEndian = 1;
+	littleEndian = ((char *)&littleEndian)[0];
 
+	mMemoryDNA->init(memDna, memDnaLength, littleEndian == 0);
 
 	///@todo we need a better version check, add version/sub version info from FileGlobal into memory DNA/header files
 	if (mMemoryDNA->getNumNames() != mFileDNA->getNumNames())
@@ -290,7 +274,6 @@ void bFile::parseInternal(int verboseMode, char* memDna,int memDnaLength)
 		//printf ("Warning, file DNA is newer than built in.");
 	}
 
-
 	mFileDNA->initCmpFlags(mMemoryDNA);
 
 	parseData();
@@ -298,102 +281,95 @@ void bFile::parseInternal(int verboseMode, char* memDna,int memDnaLength)
 	resolvePointers(verboseMode);
 
 	updateOldPointers();
-
-
 }
 
-
-
 // ----------------------------------------------------- //
-void bFile::swap(char *head, bChunkInd& dataChunk, bool ignoreEndianFlag)
+void bFile::swap(char *head, bChunkInd &dataChunk, bool ignoreEndianFlag)
 {
 	char *data = head;
 	short *strc = mFileDNA->getStruct(dataChunk.dna_nr);
 
-
-
 	const char s[] = "SoftBodyMaterialData";
 	int szs = sizeof(s);
-	if (strncmp((char*)&dataChunk.code,"ARAY",4)==0)
+	if (strncmp((char *)&dataChunk.code, "ARAY", 4) == 0)
 	{
 		short *oldStruct = mFileDNA->getStruct(dataChunk.dna_nr);
 		char *oldType = mFileDNA->getType(oldStruct[0]);
-		if (strncmp(oldType,s,szs)==0)
+		if (strncmp(oldType, s, szs) == 0)
 		{
 			return;
 		}
 	}
 
-
 	int len = mFileDNA->getLength(strc[0]);
 
-	for (int i=0; i<dataChunk.nr; i++)
+	for (int i = 0; i < dataChunk.nr; i++)
 	{
-		swapStruct(dataChunk.dna_nr, data,ignoreEndianFlag);
-		data+=len;
+		swapStruct(dataChunk.dna_nr, data, ignoreEndianFlag);
+		data += len;
 	}
 }
 
 void bFile::swapLen(char *dataPtr)
 {
-	const bool VOID_IS_8 = ((sizeof(void*)==8));
+	const bool VOID_IS_8 = ((sizeof(void *) == 8));
 	if (VOID_IS_8)
 	{
-		if (mFlags &FD_BITS_VARIES)
+		if (mFlags & FD_BITS_VARIES)
 		{
-			bChunkPtr4*c = (bChunkPtr4*) dataPtr;
-			if ((c->code & 0xFFFF)==0)
-					c->code >>=16;
+			bChunkPtr4 *c = (bChunkPtr4 *)dataPtr;
+			if ((c->code & 0xFFFF) == 0)
+				c->code >>= 16;
 			B3_SWITCH_INT(c->len);
 			B3_SWITCH_INT(c->dna_nr);
 			B3_SWITCH_INT(c->nr);
-		} else
-		{
-			bChunkPtr8* c = (bChunkPtr8*) dataPtr;
-			if ((c->code & 0xFFFF)==0)
-				c->code >>=16;
-			B3_SWITCH_INT(c->len);
-			B3_SWITCH_INT(c->dna_nr);
-			B3_SWITCH_INT(c->nr);
-
 		}
-	} else
-	{
-		if (mFlags &FD_BITS_VARIES)
+		else
 		{
-			bChunkPtr8*c = (bChunkPtr8*) dataPtr;
-			if ((c->code & 0xFFFF)==0)
-				c->code >>=16;
+			bChunkPtr8 *c = (bChunkPtr8 *)dataPtr;
+			if ((c->code & 0xFFFF) == 0)
+				c->code >>= 16;
 			B3_SWITCH_INT(c->len);
 			B3_SWITCH_INT(c->dna_nr);
 			B3_SWITCH_INT(c->nr);
-
-		} else
-		{
-			bChunkPtr4* c = (bChunkPtr4*) dataPtr;
-			if ((c->code & 0xFFFF)==0)
-				c->code >>=16;
-			B3_SWITCH_INT(c->len);
-
-			B3_SWITCH_INT(c->dna_nr);
-			B3_SWITCH_INT(c->nr);
-
 		}
 	}
+	else
+	{
+		if (mFlags & FD_BITS_VARIES)
+		{
+			bChunkPtr8 *c = (bChunkPtr8 *)dataPtr;
+			if ((c->code & 0xFFFF) == 0)
+				c->code >>= 16;
+			B3_SWITCH_INT(c->len);
+			B3_SWITCH_INT(c->dna_nr);
+			B3_SWITCH_INT(c->nr);
+		}
+		else
+		{
+			bChunkPtr4 *c = (bChunkPtr4 *)dataPtr;
+			if ((c->code & 0xFFFF) == 0)
+				c->code >>= 16;
+			B3_SWITCH_INT(c->len);
 
+			B3_SWITCH_INT(c->dna_nr);
+			B3_SWITCH_INT(c->nr);
+		}
+	}
 }
 
-
-void bFile::swapDNA(char* ptr)
+void bFile::swapDNA(char *ptr)
 {
-	bool swap = ((mFlags & FD_ENDIAN_SWAP)!=0);
+	bool swap = ((mFlags & FD_ENDIAN_SWAP) != 0);
 
-	char* data = &ptr[20];
-//	void bDNA::init(char *data, int len, bool swap)
-	int *intPtr=0;short *shtPtr=0;
-	char *cp = 0;int dataLen =0;
+	char *data = &ptr[20];
+	//	void bDNA::init(char *data, int len, bool swap)
+	int *intPtr = 0;
+	short *shtPtr = 0;
+	char *cp = 0;
+	int dataLen = 0;
 	//long nr=0;
-	intPtr = (int*)data;
+	intPtr = (int *)data;
 
 	/*
 		SDNA (4 bytes) (magic number)
@@ -403,14 +379,13 @@ void bFile::swapDNA(char* ptr)
 		<string>
 	*/
 
-	if (strncmp(data, "SDNA", 4)==0)
+	if (strncmp(data, "SDNA", 4) == 0)
 	{
 		// skip ++ NAME
-		intPtr++; intPtr++;
+		intPtr++;
+		intPtr++;
 	}
 
-
-
 	// Parse names
 	if (swap)
 		dataLen = ChunkUtils::swapInt(*intPtr);
@@ -420,16 +395,15 @@ void bFile::swapDNA(char* ptr)
 	*intPtr = ChunkUtils::swapInt(*intPtr);
 	intPtr++;
 
-	cp = (char*)intPtr;
+	cp = (char *)intPtr;
 	int i;
-	for ( i=0; i<dataLen; i++)
+	for (i = 0; i < dataLen; i++)
 	{
-		while (*cp)cp++;
+		while (*cp) cp++;
 		cp++;
 	}
 
-
-	cp = b3AlignPointer(cp,4);
+	cp = b3AlignPointer(cp, 4);
 
 	/*
 		TYPE (4 bytes)
@@ -438,8 +412,9 @@ void bFile::swapDNA(char* ptr)
 		<string>
 	*/
 
-	intPtr = (int*)cp;
-	assert(strncmp(cp, "TYPE", 4)==0); intPtr++;
+	intPtr = (int *)cp;
+	assert(strncmp(cp, "TYPE", 4) == 0);
+	intPtr++;
 
 	if (swap)
 		dataLen = ChunkUtils::swapInt(*intPtr);
@@ -450,14 +425,14 @@ void bFile::swapDNA(char* ptr)
 
 	intPtr++;
 
-	cp = (char*)intPtr;
-	for ( i=0; i<dataLen; i++)
+	cp = (char *)intPtr;
+	for (i = 0; i < dataLen; i++)
 	{
-		while (*cp)cp++;
+		while (*cp) cp++;
 		cp++;
 	}
 
-	cp = b3AlignPointer(cp,4);
+	cp = b3AlignPointer(cp, 4);
 
 	/*
 		TLEN (4 bytes)
@@ -466,15 +441,15 @@ void bFile::swapDNA(char* ptr)
 	*/
 
 	// Parse type lens
-	intPtr = (int*)cp;
-	assert(strncmp(cp, "TLEN", 4)==0); intPtr++;
+	intPtr = (int *)cp;
+	assert(strncmp(cp, "TLEN", 4) == 0);
+	intPtr++;
 
-
-	shtPtr = (short*)intPtr;
-	for ( i=0; i<dataLen; i++, shtPtr++)
+	shtPtr = (short *)intPtr;
+	for (i = 0; i < dataLen; i++, shtPtr++)
 	{
 		//??????if (swap)
-			shtPtr[0] = ChunkUtils::swapShort(shtPtr[0]);
+		shtPtr[0] = ChunkUtils::swapShort(shtPtr[0]);
 	}
 
 	if (dataLen & 1)
@@ -491,9 +466,9 @@ void bFile::swapDNA(char* ptr)
 		<namenr>
 	*/
 
-	intPtr = (int*)shtPtr;
-	cp = (char*)intPtr;
-	assert(strncmp(cp, "STRC", 4)==0);
+	intPtr = (int *)shtPtr;
+	cp = (char *)intPtr;
+	assert(strncmp(cp, "STRC", 4) == 0);
 	intPtr++;
 
 	if (swap)
@@ -505,66 +480,56 @@ void bFile::swapDNA(char* ptr)
 
 	intPtr++;
 
-
-	shtPtr = (short*)intPtr;
-	for ( i=0; i<dataLen; i++)
+	shtPtr = (short *)intPtr;
+	for (i = 0; i < dataLen; i++)
 	{
-
 		//if (swap)
 		{
 			int len = shtPtr[1];
 
-			shtPtr[0]= ChunkUtils::swapShort(shtPtr[0]);
-			shtPtr[1]= ChunkUtils::swapShort(shtPtr[1]);
+			shtPtr[0] = ChunkUtils::swapShort(shtPtr[0]);
+			shtPtr[1] = ChunkUtils::swapShort(shtPtr[1]);
 
-			shtPtr+= 2;
+			shtPtr += 2;
 
-			for (int a=0; a<len; a++, shtPtr+=2)
+			for (int a = 0; a < len; a++, shtPtr += 2)
 			{
-				shtPtr[0]= ChunkUtils::swapShort(shtPtr[0]);
-				shtPtr[1]= ChunkUtils::swapShort(shtPtr[1]);
+				shtPtr[0] = ChunkUtils::swapShort(shtPtr[0]);
+				shtPtr[1] = ChunkUtils::swapShort(shtPtr[1]);
 			}
 		}
-//		else
-//			shtPtr+= (2*shtPtr[1])+2;
+		//		else
+		//			shtPtr+= (2*shtPtr[1])+2;
 	}
-
 }
 
-void bFile::writeFile(const char* fileName)
+void bFile::writeFile(const char *fileName)
 {
-	FILE* f = fopen(fileName,"wb");
-	fwrite(mFileBuffer,1,mFileLen,f);
+	FILE *f = fopen(fileName, "wb");
+	fwrite(mFileBuffer, 1, mFileLen, f);
 	fclose(f);
 }
 
 void bFile::preSwap()
 {
-
 	//const bool brokenDNA = (mFlags&FD_BROKEN_DNA)!=0;
 	//FD_ENDIAN_SWAP
 	//byte 8 determines the endianness of the file, little (v) versus big (V)
-	int littleEndian= 1;
-	littleEndian= ((char*)&littleEndian)[0];
+	int littleEndian = 1;
+	littleEndian = ((char *)&littleEndian)[0];
 
-
-	if (mFileBuffer[8]=='V')
+	if (mFileBuffer[8] == 'V')
 	{
-		mFileBuffer[8]='v';
+		mFileBuffer[8] = 'v';
 	}
 	else
 	{
-		mFileBuffer[8]='V';
+		mFileBuffer[8] = 'V';
 	}
 
-
-
-
-
-
 	mDataStart = 12;
 
-	char *dataPtr = mFileBuffer+mDataStart;
+	char *dataPtr = mFileBuffer + mDataStart;
 
 	bChunkInd dataChunk;
 	dataChunk.code = 0;
@@ -579,21 +544,22 @@ void bFile::preSwap()
 	while (1)
 	{
 		// one behind
-		if (dataChunk.code == B3_SDNA || dataChunk.code==B3_DNA1 || dataChunk.code == B3_TYPE || dataChunk.code == B3_TLEN || dataChunk.code==B3_STRC)
+		if (dataChunk.code == B3_SDNA || dataChunk.code == B3_DNA1 || dataChunk.code == B3_TYPE || dataChunk.code == B3_TLEN || dataChunk.code == B3_STRC)
 		{
-
 			swapDNA(dataPtr);
 			break;
-		} else
+		}
+		else
 		{
 			//if (dataChunk.code == DNA1) break;
-			dataPtrHead = dataPtr+ChunkUtils::getOffset(mFlags);
+			dataPtrHead = dataPtr + ChunkUtils::getOffset(mFlags);
 
 			swapLen(dataPtr);
-			if (dataChunk.dna_nr>=0)
+			if (dataChunk.dna_nr >= 0)
 			{
-				swap(dataPtrHead, dataChunk,ignoreEndianFlag);
-			} else
+				swap(dataPtrHead, dataChunk, ignoreEndianFlag);
+			}
+			else
 			{
 				//printf("unknown chunk\n");
 			}
@@ -602,7 +568,7 @@ void bFile::preSwap()
 		// next please!
 		dataPtr += seek;
 
-		seek =  getNextBlock(&dataChunk, dataPtr, mFlags);
+		seek = getNextBlock(&dataChunk, dataPtr, mFlags);
 		if (seek < 0)
 			break;
 	}
@@ -610,56 +576,50 @@ void bFile::preSwap()
 	if (mFlags & FD_ENDIAN_SWAP)
 	{
 		mFlags &= ~FD_ENDIAN_SWAP;
-	} else
+	}
+	else
 	{
 		mFlags |= FD_ENDIAN_SWAP;
 	}
-
-
-
 }
 
-
 // ----------------------------------------------------- //
-char* bFile::readStruct(char *head, bChunkInd&  dataChunk)
+char *bFile::readStruct(char *head, bChunkInd &dataChunk)
 {
 	bool ignoreEndianFlag = false;
 
 	if (mFlags & FD_ENDIAN_SWAP)
 		swap(head, dataChunk, ignoreEndianFlag);
 
-
-
 	if (!mFileDNA->flagEqual(dataChunk.dna_nr))
 	{
 		// Ouch! need to rebuild the struct
-		short *oldStruct,*curStruct;
+		short *oldStruct, *curStruct;
 		char *oldType, *newType;
 		int oldLen, curLen, reverseOld;
 
-
 		oldStruct = mFileDNA->getStruct(dataChunk.dna_nr);
 		oldType = mFileDNA->getType(oldStruct[0]);
 
 		oldLen = mFileDNA->getLength(oldStruct[0]);
 
-		if ((mFlags&FD_BROKEN_DNA)!=0)
+		if ((mFlags & FD_BROKEN_DNA) != 0)
 		{
-			if ((strcmp(oldType,"b3QuantizedBvhNodeData")==0)&&oldLen==20)
+			if ((strcmp(oldType, "b3QuantizedBvhNodeData") == 0) && oldLen == 20)
 			{
 				return 0;
 			}
-			if ((strcmp(oldType,"b3ShortIntIndexData")==0))
+			if ((strcmp(oldType, "b3ShortIntIndexData") == 0))
 			{
 				int allocLen = 2;
-	    		char *dataAlloc = new char[(dataChunk.nr*allocLen)+1];
-				memset(dataAlloc, 0, (dataChunk.nr*allocLen)+1);
-				short* dest = (short*) dataAlloc;
-				const short* src = (short*) head;
-				for (int i=0;i<dataChunk.nr;i++)
+				char *dataAlloc = new char[(dataChunk.nr * allocLen) + 1];
+				memset(dataAlloc, 0, (dataChunk.nr * allocLen) + 1);
+				short *dest = (short *)dataAlloc;
+				const short *src = (short *)head;
+				for (int i = 0; i < dataChunk.nr; i++)
 				{
 					dest[i] = src[i];
-					if (mFlags &FD_ENDIAN_SWAP)
+					if (mFlags & FD_ENDIAN_SWAP)
 					{
 						B3_SWITCH_SHORT(dest[i]);
 					}
@@ -669,14 +629,12 @@ char* bFile::readStruct(char *head, bChunkInd&  dataChunk)
 			}
 		}
 
-
-
 		///don't try to convert Link block data, just memcpy it. Other data can be converted.
-		if (strcmp("Link",oldType)!=0)
+		if (strcmp("Link", oldType) != 0)
 		{
 			reverseOld = mMemoryDNA->getReverseType(oldType);
 
-			if ((reverseOld!=-1))
+			if ((reverseOld != -1))
 			{
 				// make sure it's here
 				//assert(reverseOld!= -1 && "getReverseType() returned -1, struct required!");
@@ -686,39 +644,38 @@ char* bFile::readStruct(char *head, bChunkInd&  dataChunk)
 				newType = mMemoryDNA->getType(curStruct[0]);
 				curLen = mMemoryDNA->getLength(curStruct[0]);
 
-
-
 				// make sure it's the same
-				assert((strcmp(oldType, newType)==0) && "internal error, struct mismatch!");
-
+				assert((strcmp(oldType, newType) == 0) && "internal error, struct mismatch!");
 
 				// numBlocks * length
 
-                int allocLen = (curLen);
-    			char *dataAlloc = new char[(dataChunk.nr*allocLen)+1];
-				memset(dataAlloc, 0, (dataChunk.nr*allocLen));
+				int allocLen = (curLen);
+				char *dataAlloc = new char[(dataChunk.nr * allocLen) + 1];
+				memset(dataAlloc, 0, (dataChunk.nr * allocLen));
 
 				// track allocated
 				addDataBlock(dataAlloc);
 
 				char *cur = dataAlloc;
 				char *old = head;
-				for (int block=0; block<dataChunk.nr; block++)
+				for (int block = 0; block < dataChunk.nr; block++)
 				{
 					bool fixupPointers = true;
 					parseStruct(cur, old, dataChunk.dna_nr, reverseOld, fixupPointers);
-					mLibPointers.insert(old,(bStructHandle*)cur);
+					mLibPointers.insert(old, (bStructHandle *)cur);
 
 					cur += curLen;
 					old += oldLen;
 				}
 				return dataAlloc;
 			}
-		} else
+		}
+		else
 		{
 			//printf("Link found\n");
 		}
-	} else
+	}
+	else
 	{
 //#define DEBUG_EQUAL_STRUCTS
 #ifdef DEBUG_EQUAL_STRUCTS
@@ -726,24 +683,20 @@ char* bFile::readStruct(char *head, bChunkInd&  dataChunk)
 		char *oldType;
 		oldStruct = mFileDNA->getStruct(dataChunk.dna_nr);
 		oldType = mFileDNA->getType(oldStruct[0]);
-		printf("%s equal structure, just memcpy\n",oldType);
-#endif //
+		printf("%s equal structure, just memcpy\n", oldType);
+#endif  //
 	}
 
-
-	char *dataAlloc = new char[(dataChunk.len)+1];
-	memset(dataAlloc, 0, dataChunk.len+1);
-
+	char *dataAlloc = new char[(dataChunk.len) + 1];
+	memset(dataAlloc, 0, dataChunk.len + 1);
 
 	// track allocated
 	addDataBlock(dataAlloc);
 
 	memcpy(dataAlloc, head, dataChunk.len);
 	return dataAlloc;
-
 }
 
-
 // ----------------------------------------------------- //
 void bFile::parseStruct(char *strcPtr, char *dtPtr, int old_dna, int new_dna, bool fixupPointers)
 {
@@ -751,7 +704,7 @@ void bFile::parseStruct(char *strcPtr, char *dtPtr, int old_dna, int new_dna, bo
 	if (new_dna == -1) return;
 
 	//disable this, because we need to fixup pointers/ListBase
-	if (0)//mFileDNA->flagEqual(old_dna))
+	if (0)  //mFileDNA->flagEqual(old_dna))
 	{
 		short *strc = mFileDNA->getStruct(old_dna);
 		int len = mFileDNA->getLength(strc[0]);
@@ -766,31 +719,29 @@ void bFile::parseStruct(char *strcPtr, char *dtPtr, int old_dna, int new_dna, bo
 	int elementLength, size, revType, old_nr, new_nr, fpLen;
 	short firstStructType;
 
-
 	// File to memory lookup
 	memoryStruct = mMemoryDNA->getStruct(new_dna);
 	fileStruct = mFileDNA->getStruct(old_dna);
 	firstStruct = fileStruct;
 
-
 	filePtrOld = fileStruct;
 	firstStructType = mMemoryDNA->getStruct(0)[0];
 
 	// Get number of elements
 	elementLength = memoryStruct[1];
-	memoryStruct+=2;
+	memoryStruct += 2;
 
-	cpc = strcPtr; cpo = 0;
-	for (int ele=0; ele<elementLength; ele++, memoryStruct+=2)
+	cpc = strcPtr;
+	cpo = 0;
+	for (int ele = 0; ele < elementLength; ele++, memoryStruct += 2)
 	{
 		memType = mMemoryDNA->getType(memoryStruct[0]);
 		memName = mMemoryDNA->getName(memoryStruct[1]);
 
-
 		size = mMemoryDNA->getElementSize(memoryStruct[0], memoryStruct[1]);
 		revType = mMemoryDNA->getReverseType(memoryStruct[0]);
 
-		if (revType != -1 && memoryStruct[0]>=firstStructType && memName[0] != '*')
+		if (revType != -1 && memoryStruct[0] >= firstStructType && memName[0] != '*')
 		{
 			cpo = getFileElement(firstStruct, memName, memType, dtPtr, &filePtrOld);
 			if (cpo)
@@ -799,94 +750,92 @@ void bFile::parseStruct(char *strcPtr, char *dtPtr, int old_dna, int new_dna, bo
 				old_nr = mFileDNA->getReverseType(memType);
 				new_nr = revType;
 				fpLen = mFileDNA->getElementSize(filePtrOld[0], filePtrOld[1]);
-				if (arrayLen==1)
+				if (arrayLen == 1)
 				{
-					parseStruct(cpc, cpo, old_nr, new_nr,fixupPointers);
-				} else
+					parseStruct(cpc, cpo, old_nr, new_nr, fixupPointers);
+				}
+				else
 				{
-					char* tmpCpc = cpc;
-					char* tmpCpo = cpo;
+					char *tmpCpc = cpc;
+					char *tmpCpo = cpo;
 
-					for (int i=0;i<arrayLen;i++)
+					for (int i = 0; i < arrayLen; i++)
 					{
-						parseStruct(tmpCpc, tmpCpo, old_nr, new_nr,fixupPointers);
-						tmpCpc += size/arrayLen;
-						tmpCpo += fpLen/arrayLen;
+						parseStruct(tmpCpc, tmpCpo, old_nr, new_nr, fixupPointers);
+						tmpCpc += size / arrayLen;
+						tmpCpo += fpLen / arrayLen;
 					}
 				}
-				cpc+=size;
-				cpo+=fpLen;
+				cpc += size;
+				cpo += fpLen;
 			}
 			else
-				cpc+=size;
+				cpc += size;
 		}
 		else
 		{
-			getMatchingFileDNA(fileStruct, memName, memType, cpc, dtPtr,fixupPointers);
-			cpc+=size;
+			getMatchingFileDNA(fileStruct, memName, memType, cpc, dtPtr, fixupPointers);
+			cpc += size;
 		}
-
 	}
 }
 
-
 // ----------------------------------------------------- //
 static void getElement(int arrayLen, const char *cur, const char *old, char *oldPtr, char *curData)
 {
-#define b3GetEle(value, current, type, cast, size, ptr)\
-	if (strcmp(current, type)==0)\
-	{\
-		value = (*(cast*)ptr);\
-		ptr += size;\
+#define b3GetEle(value, current, type, cast, size, ptr) \
+	if (strcmp(current, type) == 0)                     \
+	{                                                   \
+		value = (*(cast *)ptr);                         \
+		ptr += size;                                    \
 	}
 
-#define b3SetEle(value, current, type, cast, size, ptr)\
-	if (strcmp(current, type)==0)\
-	{\
-		(*(cast*)ptr) = (cast)value;\
-		ptr += size;\
+#define b3SetEle(value, current, type, cast, size, ptr) \
+	if (strcmp(current, type) == 0)                     \
+	{                                                   \
+		(*(cast *)ptr) = (cast)value;                   \
+		ptr += size;                                    \
 	}
 	double value = 0.0;
 
-	for (int i=0; i<arrayLen; i++)
+	for (int i = 0; i < arrayLen; i++)
 	{
-		b3GetEle(value, old, "char",   char,   sizeof(char),   oldPtr);
-		b3SetEle(value, cur, "char",   char,   sizeof(char),   curData);
-		b3GetEle(value, old, "short",  short,  sizeof(short),  oldPtr);
-		b3SetEle(value, cur, "short",  short,  sizeof(short),  curData);
-		b3GetEle(value, old, "ushort",  unsigned short,  sizeof(unsigned short),  oldPtr);
-		b3SetEle(value, cur, "ushort",  unsigned short,  sizeof(unsigned short),  curData);
-		b3GetEle(value, old, "int",    int,    sizeof(int),    oldPtr);
-		b3SetEle(value, cur, "int",    int,    sizeof(int),    curData);
-		b3GetEle(value, old, "long",   int,    sizeof(int),    oldPtr);
-		b3SetEle(value, cur, "long",   int,    sizeof(int),    curData);
-		b3GetEle(value, old, "float",  float,  sizeof(float),  oldPtr);
-		b3SetEle(value, cur, "float",  float,  sizeof(float),  curData);
+		b3GetEle(value, old, "char", char, sizeof(char), oldPtr);
+		b3SetEle(value, cur, "char", char, sizeof(char), curData);
+		b3GetEle(value, old, "short", short, sizeof(short), oldPtr);
+		b3SetEle(value, cur, "short", short, sizeof(short), curData);
+		b3GetEle(value, old, "ushort", unsigned short, sizeof(unsigned short), oldPtr);
+		b3SetEle(value, cur, "ushort", unsigned short, sizeof(unsigned short), curData);
+		b3GetEle(value, old, "int", int, sizeof(int), oldPtr);
+		b3SetEle(value, cur, "int", int, sizeof(int), curData);
+		b3GetEle(value, old, "long", int, sizeof(int), oldPtr);
+		b3SetEle(value, cur, "long", int, sizeof(int), curData);
+		b3GetEle(value, old, "float", float, sizeof(float), oldPtr);
+		b3SetEle(value, cur, "float", float, sizeof(float), curData);
 		b3GetEle(value, old, "double", double, sizeof(double), oldPtr);
 		b3SetEle(value, cur, "double", double, sizeof(double), curData);
 	}
 }
 
-
 // ----------------------------------------------------- //
-void bFile::swapData(char *data, short type, int arraySize,bool ignoreEndianFlag)
+void bFile::swapData(char *data, short type, int arraySize, bool ignoreEndianFlag)
 {
-	if (ignoreEndianFlag || (mFlags &FD_ENDIAN_SWAP))
+	if (ignoreEndianFlag || (mFlags & FD_ENDIAN_SWAP))
 	{
 		if (type == 2 || type == 3)
 		{
-			short *sp = (short*)data;
-			for (int i=0; i<arraySize; i++)
+			short *sp = (short *)data;
+			for (int i = 0; i < arraySize; i++)
 			{
 				sp[0] = ChunkUtils::swapShort(sp[0]);
 				sp++;
 			}
 		}
-		if (type>3 && type <8)
+		if (type > 3 && type < 8)
 		{
 			char c;
 			char *cp = data;
-			for (int i=0; i<arraySize; i++)
+			for (int i = 0; i < arraySize; i++)
 			{
 				c = cp[0];
 				cp[0] = cp[3];
@@ -894,14 +843,12 @@ void bFile::swapData(char *data, short type, int arraySize,bool ignoreEndianFlag
 				c = cp[1];
 				cp[1] = cp[2];
 				cp[2] = c;
-				cp+=4;
+				cp += 4;
 			}
 		}
 	}
 }
 
-
-
 void bFile::safeSwapPtr(char *dst, const char *src)
 {
 	int ptrFile = mFileDNA->getPointerSize();
@@ -910,84 +857,79 @@ void bFile::safeSwapPtr(char *dst, const char *src)
 	if (!src && !dst)
 		return;
 
-
 	if (ptrFile == ptrMem)
 	{
 		memcpy(dst, src, ptrMem);
 	}
-	else if (ptrMem==4 && ptrFile==8)
+	else if (ptrMem == 4 && ptrFile == 8)
 	{
-		b3PointerUid* oldPtr = (b3PointerUid*)src;
-		b3PointerUid* newPtr = (b3PointerUid*)dst;
+		b3PointerUid *oldPtr = (b3PointerUid *)src;
+		b3PointerUid *newPtr = (b3PointerUid *)dst;
 
 		if (oldPtr->m_uniqueIds[0] == oldPtr->m_uniqueIds[1])
 		{
 			//Bullet stores the 32bit unique ID in both upper and lower part of 64bit pointers
 			//so it can be used to distinguish between .blend and .bullet
 			newPtr->m_uniqueIds[0] = oldPtr->m_uniqueIds[0];
-		} else
+		}
+		else
 		{
 			//deal with pointers the Blender .blend style way, see
 			//readfile.c in the Blender source tree
-			b3Long64 longValue = *((b3Long64*)src);
+			b3Long64 longValue = *((b3Long64 *)src);
 			//endian swap for 64bit pointer otherwise truncation will fail due to trailing zeros
 			if (mFlags & FD_ENDIAN_SWAP)
 				B3_SWITCH_LONGINT(longValue);
-			*((int*)dst) = (int)(longValue>>3);
+			*((int *)dst) = (int)(longValue >> 3);
 		}
-
 	}
-	else if (ptrMem==8 && ptrFile==4)
+	else if (ptrMem == 8 && ptrFile == 4)
 	{
-		b3PointerUid* oldPtr = (b3PointerUid*)src;
-		b3PointerUid* newPtr = (b3PointerUid*)dst;
+		b3PointerUid *oldPtr = (b3PointerUid *)src;
+		b3PointerUid *newPtr = (b3PointerUid *)dst;
 		if (oldPtr->m_uniqueIds[0] == oldPtr->m_uniqueIds[1])
 		{
 			newPtr->m_uniqueIds[0] = oldPtr->m_uniqueIds[0];
 			newPtr->m_uniqueIds[1] = 0;
-		} else
+		}
+		else
 		{
-			*((b3Long64*)dst)= *((int*)src);
+			*((b3Long64 *)dst) = *((int *)src);
 		}
 	}
 	else
 	{
-		printf ("%d %d\n", ptrFile,ptrMem);
+		printf("%d %d\n", ptrFile, ptrMem);
 		assert(0 && "Invalid pointer len");
 	}
-
-
 }
 
-
 // ----------------------------------------------------- //
-void bFile::getMatchingFileDNA(short* dna_addr, const char* lookupName,  const char* lookupType, char *strcData, char *data, bool fixupPointers)
+void bFile::getMatchingFileDNA(short *dna_addr, const char *lookupName, const char *lookupType, char *strcData, char *data, bool fixupPointers)
 {
 	// find the matching memory dna data
 	// to the file being loaded. Fill the
 	// memory with the file data...
 
 	int len = dna_addr[1];
-	dna_addr+=2;
+	dna_addr += 2;
 
-	for (int i=0; i<len; i++, dna_addr+=2)
+	for (int i = 0; i < len; i++, dna_addr += 2)
 	{
-		const char* type = mFileDNA->getType(dna_addr[0]);
-		const char* name = mFileDNA->getName(dna_addr[1]);
-
-
+		const char *type = mFileDNA->getType(dna_addr[0]);
+		const char *name = mFileDNA->getName(dna_addr[1]);
 
 		int eleLen = mFileDNA->getElementSize(dna_addr[0], dna_addr[1]);
 
-		if ((mFlags&FD_BROKEN_DNA)!=0)
+		if ((mFlags & FD_BROKEN_DNA) != 0)
 		{
-			if ((strcmp(type,"short")==0)&&(strcmp(name,"int")==0))
+			if ((strcmp(type, "short") == 0) && (strcmp(name, "int") == 0))
 			{
 				eleLen = 0;
 			}
 		}
 
-		if (strcmp(lookupName, name)==0)
+		if (strcmp(lookupName, name) == 0)
 		{
 			//int arrayLenold = mFileDNA->getArraySize((char*)name.c_str());
 			int arrayLen = mFileDNA->getArraySizeNew(dna_addr[1]);
@@ -998,7 +940,7 @@ void bFile::getMatchingFileDNA(short* dna_addr, const char* lookupName,  const c
 				// cast pointers
 				int ptrFile = mFileDNA->getPointerSize();
 				int ptrMem = mMemoryDNA->getPointerSize();
-				safeSwapPtr(strcData,data);
+				safeSwapPtr(strcData, data);
 
 				if (fixupPointers)
 				{
@@ -1007,11 +949,11 @@ void bFile::getMatchingFileDNA(short* dna_addr, const char* lookupName,  const c
 						//void **sarray = (void**)strcData;
 						//void **darray = (void**)data;
 
-                        char *cpc, *cpo;
-						cpc = (char*)strcData;
-						cpo = (char*)data;
+						char *cpc, *cpo;
+						cpc = (char *)strcData;
+						cpo = (char *)data;
 
-						for (int a=0; a<arrayLen; a++)
+						for (int a = 0; a < arrayLen; a++)
 						{
 							safeSwapPtr(cpc, cpo);
 							m_pointerFixupArray.push_back(cpc);
@@ -1029,12 +971,11 @@ void bFile::getMatchingFileDNA(short* dna_addr, const char* lookupName,  const c
 				}
 				else
 				{
-//					printf("skipped %s %s : %x\n",type.c_str(),name.c_str(),strcData);
+					//					printf("skipped %s %s : %x\n",type.c_str(),name.c_str(),strcData);
 				}
-
 			}
 
-			else if (strcmp(type, lookupType)==0)
+			else if (strcmp(type, lookupType) == 0)
 				memcpy(strcData, data, eleLen);
 			else
 				getElement(arrayLen, lookupType, type, data, strcData);
@@ -1042,27 +983,26 @@ void bFile::getMatchingFileDNA(short* dna_addr, const char* lookupName,  const c
 			// --
 			return;
 		}
-		data+=eleLen;
+		data += eleLen;
 	}
 }
 
-
 // ----------------------------------------------------- //
-char* bFile::getFileElement(short *firstStruct, char *lookupName, char *lookupType, char *data, short **foundPos)
+char *bFile::getFileElement(short *firstStruct, char *lookupName, char *lookupType, char *data, short **foundPos)
 {
-	short *old = firstStruct;//mFileDNA->getStruct(old_nr);
+	short *old = firstStruct;  //mFileDNA->getStruct(old_nr);
 	int elementLength = old[1];
-	old+=2;
+	old += 2;
 
-	for (int i=0; i<elementLength; i++, old+=2)
+	for (int i = 0; i < elementLength; i++, old += 2)
 	{
-		char* type = mFileDNA->getType(old[0]);
-		char* name = mFileDNA->getName(old[1]);
+		char *type = mFileDNA->getType(old[0]);
+		char *name = mFileDNA->getName(old[1]);
 		int len = mFileDNA->getElementSize(old[0], old[1]);
 
-		if (strcmp(lookupName, name)==0)
+		if (strcmp(lookupName, name) == 0)
 		{
-			if (strcmp(type, lookupType)==0)
+			if (strcmp(type, lookupType) == 0)
 			{
 				if (foundPos)
 					*foundPos = old;
@@ -1070,46 +1010,46 @@ char* bFile::getFileElement(short *firstStruct, char *lookupName, char *lookupTy
 			}
 			return 0;
 		}
-		data+=len;
+		data += len;
 	}
 	return 0;
 }
 
-
 // ----------------------------------------------------- //
-void bFile::swapStruct(int dna_nr, char *data,bool ignoreEndianFlag)
+void bFile::swapStruct(int dna_nr, char *data, bool ignoreEndianFlag)
 {
 	if (dna_nr == -1) return;
 
 	short *strc = mFileDNA->getStruct(dna_nr);
 	//short *firstStrc = strc;
 
-	int elementLen= strc[1];
-	strc+=2;
+	int elementLen = strc[1];
+	strc += 2;
 
 	short first = mFileDNA->getStruct(0)[0];
 
 	char *buf = data;
-	for (int i=0; i<elementLen; i++, strc+=2)
+	for (int i = 0; i < elementLen; i++, strc += 2)
 	{
 		char *type = mFileDNA->getType(strc[0]);
 		char *name = mFileDNA->getName(strc[1]);
 
 		int size = mFileDNA->getElementSize(strc[0], strc[1]);
-		if (strc[0] >= first && name[0]!='*')
+		if (strc[0] >= first && name[0] != '*')
 		{
 			int old_nr = mFileDNA->getReverseType(type);
 			int arrayLen = mFileDNA->getArraySizeNew(strc[1]);
-			if (arrayLen==1)
+			if (arrayLen == 1)
 			{
-				swapStruct(old_nr,buf,ignoreEndianFlag);
-			} else
+				swapStruct(old_nr, buf, ignoreEndianFlag);
+			}
+			else
 			{
-				char* tmpBuf = buf;
-				for (int i=0;i<arrayLen;i++)
+				char *tmpBuf = buf;
+				for (int i = 0; i < arrayLen; i++)
 				{
-					swapStruct(old_nr,tmpBuf,ignoreEndianFlag);
-					tmpBuf+=size/arrayLen;
+					swapStruct(old_nr, tmpBuf, ignoreEndianFlag);
+					tmpBuf += size / arrayLen;
 				}
 			}
 		}
@@ -1118,39 +1058,39 @@ void bFile::swapStruct(int dna_nr, char *data,bool ignoreEndianFlag)
 			//int arrayLenOld = mFileDNA->getArraySize(name);
 			int arrayLen = mFileDNA->getArraySizeNew(strc[1]);
 			//assert(arrayLenOld == arrayLen);
-			swapData(buf, strc[0], arrayLen,ignoreEndianFlag);
+			swapData(buf, strc[0], arrayLen, ignoreEndianFlag);
 		}
-		buf+=size;
+		buf += size;
 	}
 }
 
 void bFile::resolvePointersMismatch()
 {
-//	printf("resolvePointersStructMismatch\n");
+	//	printf("resolvePointersStructMismatch\n");
 
 	int i;
 
-	for (i=0;i<	m_pointerFixupArray.size();i++)
+	for (i = 0; i < m_pointerFixupArray.size(); i++)
 	{
-		char* cur = m_pointerFixupArray.at(i);
-		void** ptrptr = (void**) cur;
-		void* ptr = *ptrptr;
+		char *cur = m_pointerFixupArray.at(i);
+		void **ptrptr = (void **)cur;
+		void *ptr = *ptrptr;
 		ptr = findLibPointer(ptr);
 		if (ptr)
 		{
 			//printf("Fixup pointer!\n");
 			*(ptrptr) = ptr;
-		} else
+		}
+		else
 		{
-//			printf("pointer not found: %x\n",cur);
+			//			printf("pointer not found: %x\n",cur);
 		}
 	}
 
-
-	for (i=0; i<m_pointerPtrFixupArray.size(); i++)
+	for (i = 0; i < m_pointerPtrFixupArray.size(); i++)
 	{
-		char* cur= m_pointerPtrFixupArray.at(i);
-		void** ptrptr = (void**)cur;
+		char *cur = m_pointerPtrFixupArray.at(i);
+		void **ptrptr = (void **)cur;
 
 		bChunkInd *block = m_chunkPtrPtrMap.find(*ptrptr);
 		if (block)
@@ -1158,7 +1098,6 @@ void bFile::resolvePointersMismatch()
 			int ptrMem = mMemoryDNA->getPointerSize();
 			int ptrFile = mFileDNA->getPointerSize();
 
-
 			int blockLen = block->len / ptrFile;
 
 			void *onptr = findLibPointer(*ptrptr);
@@ -1168,16 +1107,16 @@ void bFile::resolvePointersMismatch()
 				addDataBlock(newPtr);
 				memset(newPtr, 0, blockLen * ptrMem);
 
-				void **onarray = (void**)onptr;
-				char *oldPtr = (char*)onarray;
+				void **onarray = (void **)onptr;
+				char *oldPtr = (char *)onarray;
 
 				int p = 0;
 				while (blockLen-- > 0)
 				{
 					b3PointerUid dp = {{0}};
-					safeSwapPtr((char*)dp.m_uniqueIds, oldPtr);
+					safeSwapPtr((char *)dp.m_uniqueIds, oldPtr);
 
-					void **tptr = (void**)(newPtr + p * ptrMem);
+					void **tptr = (void **)(newPtr + p * ptrMem);
 					*tptr = findLibPointer(dp.m_ptr);
 
 					oldPtr += ptrFile;
@@ -1190,70 +1129,63 @@ void bFile::resolvePointersMismatch()
 	}
 }
 
-
 ///this loop only works fine if the Blender DNA structure of the file matches the headerfiles
-void bFile::resolvePointersChunk(const bChunkInd& dataChunk, int verboseMode)
+void bFile::resolvePointersChunk(const bChunkInd &dataChunk, int verboseMode)
 {
-	bParse::bDNA* fileDna = mFileDNA ? mFileDNA : mMemoryDNA;
+	bParse::bDNA *fileDna = mFileDNA ? mFileDNA : mMemoryDNA;
 
-	short int* oldStruct = fileDna->getStruct(dataChunk.dna_nr);
+	short int *oldStruct = fileDna->getStruct(dataChunk.dna_nr);
 	short oldLen = fileDna->getLength(oldStruct[0]);
 	//char* structType = fileDna->getType(oldStruct[0]);
 
-	char* cur	= (char*)findLibPointer(dataChunk.oldPtr);
-	for (int block=0; block<dataChunk.nr; block++)
+	char *cur = (char *)findLibPointer(dataChunk.oldPtr);
+	for (int block = 0; block < dataChunk.nr; block++)
 	{
-		resolvePointersStructRecursive(cur,dataChunk.dna_nr, verboseMode,1);
+		resolvePointersStructRecursive(cur, dataChunk.dna_nr, verboseMode, 1);
 		cur += oldLen;
 	}
 }
 
-
-int bFile::resolvePointersStructRecursive(char *strcPtr, int dna_nr, int verboseMode,int recursion)
+int bFile::resolvePointersStructRecursive(char *strcPtr, int dna_nr, int verboseMode, int recursion)
 {
+	bParse::bDNA *fileDna = mFileDNA ? mFileDNA : mMemoryDNA;
 
-	bParse::bDNA* fileDna = mFileDNA ? mFileDNA : mMemoryDNA;
+	char *memType;
+	char *memName;
+	short firstStructType = fileDna->getStruct(0)[0];
 
-	char* memType;
-	char* memName;
-	short	firstStructType = fileDna->getStruct(0)[0];
+	char *elemPtr = strcPtr;
 
-
-	char* elemPtr= strcPtr;
-
-	short int* oldStruct = fileDna->getStruct(dna_nr);
+	short int *oldStruct = fileDna->getStruct(dna_nr);
 
 	int elementLength = oldStruct[1];
-	oldStruct+=2;
+	oldStruct += 2;
 
 	int totalSize = 0;
 
-	for (int ele=0; ele<elementLength; ele++, oldStruct+=2)
+	for (int ele = 0; ele < elementLength; ele++, oldStruct += 2)
 	{
-
 		memType = fileDna->getType(oldStruct[0]);
 		memName = fileDna->getName(oldStruct[1]);
 
-
-
 		int arrayLen = fileDna->getArraySizeNew(oldStruct[1]);
 		if (memName[0] == '*')
 		{
 			if (arrayLen > 1)
 			{
-				void **array= (void**)elemPtr;
-				for (int a=0; a<arrayLen; a++)
+				void **array = (void **)elemPtr;
+				for (int a = 0; a < arrayLen; a++)
 				{
 					if (verboseMode & FD_VERBOSE_EXPORT_XML)
 					{
-						for (int i=0;i<recursion;i++)
+						for (int i = 0; i < recursion; i++)
 						{
 							printf("  ");
 						}
 						//skip the *
-						printf("<%s type=\"pointer\"> ",&memName[1]);
+						printf("<%s type=\"pointer\"> ", &memName[1]);
 						printf("%p ", array[a]);
-						printf("</%s>\n",&memName[1]);
+						printf("</%s>\n", &memName[1]);
 					}
 
 					array[a] = findLibPointer(array[a]);
@@ -1261,266 +1193,259 @@ int bFile::resolvePointersStructRecursive(char *strcPtr, int dna_nr, int verbose
 			}
 			else
 			{
-				void** ptrptr = (void**) elemPtr;
-				void* ptr = *ptrptr;
+				void **ptrptr = (void **)elemPtr;
+				void *ptr = *ptrptr;
 				if (verboseMode & FD_VERBOSE_EXPORT_XML)
 				{
-					for (int i=0;i<recursion;i++)
+					for (int i = 0; i < recursion; i++)
 					{
 						printf("  ");
 					}
-					printf("<%s type=\"pointer\"> ",&memName[1]);
+					printf("<%s type=\"pointer\"> ", &memName[1]);
 					printf("%p ", ptr);
-					printf("</%s>\n",&memName[1]);
+					printf("</%s>\n", &memName[1]);
 				}
 				ptr = findLibPointer(ptr);
 
 				if (ptr)
 				{
-	//				printf("Fixup pointer at 0x%x from 0x%x to 0x%x!\n",ptrptr,*ptrptr,ptr);
+					//				printf("Fixup pointer at 0x%x from 0x%x to 0x%x!\n",ptrptr,*ptrptr,ptr);
 					*(ptrptr) = ptr;
 					if (memName[1] == '*' && ptrptr && *ptrptr)
 					{
 						// This	will only work if the given	**array	is continuous
-						void **array= (void**)*(ptrptr);
-						void *np= array[0];
-						int	n=0;
+						void **array = (void **)*(ptrptr);
+						void *np = array[0];
+						int n = 0;
 						while (np)
 						{
-							np= findLibPointer(array[n]);
-							if (np) array[n]= np;
+							np = findLibPointer(array[n]);
+							if (np) array[n] = np;
 							n++;
 						}
 					}
-				} else
+				}
+				else
 				{
-	//				printf("Cannot fixup pointer at 0x%x from 0x%x to 0x%x!\n",ptrptr,*ptrptr,ptr);
+					//				printf("Cannot fixup pointer at 0x%x from 0x%x to 0x%x!\n",ptrptr,*ptrptr,ptr);
 				}
 			}
-		} else
+		}
+		else
 		{
 			int revType = fileDna->getReverseType(oldStruct[0]);
-			if (oldStruct[0]>=firstStructType) //revType != -1 &&
+			if (oldStruct[0] >= firstStructType)  //revType != -1 &&
 			{
 				char cleanName[MAX_STRLEN];
-				getCleanName(memName,cleanName);
+				getCleanName(memName, cleanName);
 
 				int arrayLen = fileDna->getArraySizeNew(oldStruct[1]);
 				int byteOffset = 0;
 
 				if (verboseMode & FD_VERBOSE_EXPORT_XML)
 				{
-					for (int i=0;i<recursion;i++)
+					for (int i = 0; i < recursion; i++)
 					{
 						printf("  ");
 					}
 
-					if (arrayLen>1)
+					if (arrayLen > 1)
 					{
-						printf("<%s type=\"%s\" count=%d>\n",cleanName,memType, arrayLen);
-					} else
+						printf("<%s type=\"%s\" count=%d>\n", cleanName, memType, arrayLen);
+					}
+					else
 					{
-						printf("<%s type=\"%s\">\n",cleanName,memType);
+						printf("<%s type=\"%s\">\n", cleanName, memType);
 					}
 				}
 
-				for (int i=0;i<arrayLen;i++)
+				for (int i = 0; i < arrayLen; i++)
 				{
-					byteOffset += resolvePointersStructRecursive(elemPtr+byteOffset,revType, verboseMode,recursion+1);
+					byteOffset += resolvePointersStructRecursive(elemPtr + byteOffset, revType, verboseMode, recursion + 1);
 				}
 				if (verboseMode & FD_VERBOSE_EXPORT_XML)
 				{
-					for (int i=0;i<recursion;i++)
+					for (int i = 0; i < recursion; i++)
 					{
 						printf("  ");
 					}
-					printf("</%s>\n",cleanName);
+					printf("</%s>\n", cleanName);
 				}
-			} else
+			}
+			else
 			{
 				//export a simple type
 				if (verboseMode & FD_VERBOSE_EXPORT_XML)
 				{
-
-					if (arrayLen>MAX_ARRAY_LENGTH)
+					if (arrayLen > MAX_ARRAY_LENGTH)
 					{
 						printf("too long\n");
-					} else
+					}
+					else
 					{
 						//printf("%s %s\n",memType,memName);
 
-						bool isIntegerType = (strcmp(memType,"char")==0) || (strcmp(memType,"int")==0) || (strcmp(memType,"short")==0);
+						bool isIntegerType = (strcmp(memType, "char") == 0) || (strcmp(memType, "int") == 0) || (strcmp(memType, "short") == 0);
 
 						if (isIntegerType)
 						{
-							const char* newtype="int";
+							const char *newtype = "int";
 							int dbarray[MAX_ARRAY_LENGTH];
-							int* dbPtr = 0;
-							char* tmp = elemPtr;
+							int *dbPtr = 0;
+							char *tmp = elemPtr;
 							dbPtr = &dbarray[0];
 							if (dbPtr)
 							{
 								char cleanName[MAX_STRLEN];
-								getCleanName(memName,cleanName);
+								getCleanName(memName, cleanName);
 
 								int i;
-								getElement(arrayLen, newtype,memType, tmp, (char*)dbPtr);
-								for (i=0;i<recursion;i++)
+								getElement(arrayLen, newtype, memType, tmp, (char *)dbPtr);
+								for (i = 0; i < recursion; i++)
 									printf("  ");
-								if (arrayLen==1)
-									printf("<%s type=\"%s\">",cleanName,memType);
+								if (arrayLen == 1)
+									printf("<%s type=\"%s\">", cleanName, memType);
 								else
-									printf("<%s type=\"%s\" count=%d>",cleanName,memType,arrayLen);
-								for (i=0;i<arrayLen;i++)
-									printf(" %d ",dbPtr[i]);
-								printf("</%s>\n",cleanName);
+									printf("<%s type=\"%s\" count=%d>", cleanName, memType, arrayLen);
+								for (i = 0; i < arrayLen; i++)
+									printf(" %d ", dbPtr[i]);
+								printf("</%s>\n", cleanName);
 							}
-						} else
+						}
+						else
 						{
-							const char* newtype="double";
+							const char *newtype = "double";
 							double dbarray[MAX_ARRAY_LENGTH];
-					 		double* dbPtr = 0;
-							char* tmp = elemPtr;
+							double *dbPtr = 0;
+							char *tmp = elemPtr;
 							dbPtr = &dbarray[0];
 							if (dbPtr)
 							{
 								int i;
-								getElement(arrayLen, newtype,memType, tmp, (char*)dbPtr);
-								for (i=0;i<recursion;i++)
+								getElement(arrayLen, newtype, memType, tmp, (char *)dbPtr);
+								for (i = 0; i < recursion; i++)
 									printf("  ");
 								char cleanName[MAX_STRLEN];
-								getCleanName(memName,cleanName);
+								getCleanName(memName, cleanName);
 
-								if (arrayLen==1)
+								if (arrayLen == 1)
 								{
-									printf("<%s type=\"%s\">",memName,memType);
+									printf("<%s type=\"%s\">", memName, memType);
 								}
 								else
 								{
-									printf("<%s type=\"%s\" count=%d>",cleanName,memType,arrayLen);
+									printf("<%s type=\"%s\" count=%d>", cleanName, memType, arrayLen);
 								}
-								for (i=0;i<arrayLen;i++)
-									printf(" %f ",dbPtr[i]);
-								printf("</%s>\n",cleanName);
+								for (i = 0; i < arrayLen; i++)
+									printf(" %f ", dbPtr[i]);
+								printf("</%s>\n", cleanName);
 							}
 						}
 					}
-
 				}
 			}
 		}
 
 		int size = fileDna->getElementSize(oldStruct[0], oldStruct[1]);
 		totalSize += size;
-		elemPtr+=size;
-
+		elemPtr += size;
 	}
 
 	return totalSize;
 }
 
-
 ///Resolve pointers replaces the original pointers in structures, and linked lists by the new in-memory structures
 void bFile::resolvePointers(int verboseMode)
 {
-	bParse::bDNA* fileDna = mFileDNA ? mFileDNA : mMemoryDNA;
+	bParse::bDNA *fileDna = mFileDNA ? mFileDNA : mMemoryDNA;
 
 	//char *dataPtr = mFileBuffer+mDataStart;
 
-	if (1) //mFlags & (FD_BITS_VARIES | FD_VERSION_VARIES))
+	if (1)  //mFlags & (FD_BITS_VARIES | FD_VERSION_VARIES))
 	{
 		resolvePointersMismatch();
 	}
 
 	{
-
 		if (verboseMode & FD_VERBOSE_EXPORT_XML)
 		{
 			printf("<?xml version=\"1.0\" encoding=\"utf-8\"?>\n");
 			int numitems = m_chunks.size();
 			printf("<bullet_physics version=%d itemcount = %d>\n", b3GetVersion(), numitems);
 		}
-		for (int i=0;i<m_chunks.size();i++)
+		for (int i = 0; i < m_chunks.size(); i++)
 		{
-			const bChunkInd& dataChunk = m_chunks.at(i);
+			const bChunkInd &dataChunk = m_chunks.at(i);
 
 			if (!mFileDNA || fileDna->flagEqual(dataChunk.dna_nr))
 			{
 				//dataChunk.len
-				short int* oldStruct = fileDna->getStruct(dataChunk.dna_nr);
-				char* oldType = fileDna->getType(oldStruct[0]);
+				short int *oldStruct = fileDna->getStruct(dataChunk.dna_nr);
+				char *oldType = fileDna->getType(oldStruct[0]);
 
 				if (verboseMode & FD_VERBOSE_EXPORT_XML)
-					printf(" <%s pointer=%p>\n",oldType,dataChunk.oldPtr);
+					printf(" <%s pointer=%p>\n", oldType, dataChunk.oldPtr);
 
 				resolvePointersChunk(dataChunk, verboseMode);
 
 				if (verboseMode & FD_VERBOSE_EXPORT_XML)
-					printf(" </%s>\n",oldType);
-			} else
+					printf(" </%s>\n", oldType);
+			}
+			else
 			{
 				//printf("skipping mStruct\n");
 			}
 		}
-			if (verboseMode & FD_VERBOSE_EXPORT_XML)
-			{
-				printf("</bullet_physics>\n");
-			}
+		if (verboseMode & FD_VERBOSE_EXPORT_XML)
+		{
+			printf("</bullet_physics>\n");
+		}
 	}
-
-
 }
 
-
 // ----------------------------------------------------- //
-void* bFile::findLibPointer(void *ptr)
+void *bFile::findLibPointer(void *ptr)
 {
-
-	bStructHandle** ptrptr = getLibPointers().find(ptr);
+	bStructHandle **ptrptr = getLibPointers().find(ptr);
 	if (ptrptr)
 		return *ptrptr;
 	return 0;
 }
 
-
-void	bFile::updateOldPointers()
+void bFile::updateOldPointers()
 {
 	int i;
 
-	for (i=0;i<m_chunks.size();i++)
+	for (i = 0; i < m_chunks.size(); i++)
 	{
-		bChunkInd& dataChunk = m_chunks[i];
+		bChunkInd &dataChunk = m_chunks[i];
 		dataChunk.oldPtr = findLibPointer(dataChunk.oldPtr);
 	}
 }
-void	bFile::dumpChunks(bParse::bDNA* dna)
+void bFile::dumpChunks(bParse::bDNA *dna)
 {
 	int i;
 
-	for (i=0;i<m_chunks.size();i++)
+	for (i = 0; i < m_chunks.size(); i++)
 	{
-		bChunkInd& dataChunk = m_chunks[i];
-		char* codeptr = (char*)&dataChunk.code;
-		char codestr[5] = {codeptr[0],codeptr[1],codeptr[2],codeptr[3],0};
+		bChunkInd &dataChunk = m_chunks[i];
+		char *codeptr = (char *)&dataChunk.code;
+		char codestr[5] = {codeptr[0], codeptr[1], codeptr[2], codeptr[3], 0};
 
-		short* newStruct = dna->getStruct(dataChunk.dna_nr);
-		char* typeName = dna->getType(newStruct[0]);
-		printf("%3d: %s  ",i,typeName);
+		short *newStruct = dna->getStruct(dataChunk.dna_nr);
+		char *typeName = dna->getType(newStruct[0]);
+		printf("%3d: %s  ", i, typeName);
 
-		printf("code=%s  ",codestr);
+		printf("code=%s  ", codestr);
 
-		printf("ptr=%p  ",dataChunk.oldPtr);
-		printf("len=%d  ",dataChunk.len);
-		printf("nr=%d  ",dataChunk.nr);
-		if (dataChunk.nr!=1)
+		printf("ptr=%p  ", dataChunk.oldPtr);
+		printf("len=%d  ", dataChunk.len);
+		printf("nr=%d  ", dataChunk.nr);
+		if (dataChunk.nr != 1)
 		{
 			printf("not 1\n");
 		}
 		printf("\n");
-
-
-
-
 	}
 
 #if 0
@@ -1536,20 +1461,18 @@ void	bFile::dumpChunks(bParse::bDNA* dna)
 		printf("\n");
 	}
 #endif
-
 }
 
-
-void	bFile::writeChunks(FILE* fp, bool fixupPointers)
+void bFile::writeChunks(FILE *fp, bool fixupPointers)
 {
-	bParse::bDNA* fileDna = mFileDNA ? mFileDNA : mMemoryDNA;
+	bParse::bDNA *fileDna = mFileDNA ? mFileDNA : mMemoryDNA;
 
-	for (int i=0;i<m_chunks.size();i++)
+	for (int i = 0; i < m_chunks.size(); i++)
 	{
-		bChunkInd& dataChunk = m_chunks.at(i);
+		bChunkInd &dataChunk = m_chunks.at(i);
 
 		// Ouch! need to rebuild the struct
-		short *oldStruct,*curStruct;
+		short *oldStruct, *curStruct;
 		char *oldType, *newType;
 		int oldLen, curLen, reverseOld;
 
@@ -1559,8 +1482,7 @@ void	bFile::writeChunks(FILE* fp, bool fixupPointers)
 		///don't try to convert Link block data, just memcpy it. Other data can be converted.
 		reverseOld = mMemoryDNA->getReverseType(oldType);
 
-
-		if ((reverseOld!=-1))
+		if ((reverseOld != -1))
 		{
 			// make sure it's here
 			//assert(reverseOld!= -1 && "getReverseType() returned -1, struct required!");
@@ -1568,50 +1490,47 @@ void	bFile::writeChunks(FILE* fp, bool fixupPointers)
 			curStruct = mMemoryDNA->getStruct(reverseOld);
 			newType = mMemoryDNA->getType(curStruct[0]);
 			// make sure it's the same
-			assert((strcmp(oldType, newType)==0) && "internal error, struct mismatch!");
-
+			assert((strcmp(oldType, newType) == 0) && "internal error, struct mismatch!");
 
 			curLen = mMemoryDNA->getLength(curStruct[0]);
 			dataChunk.dna_nr = reverseOld;
-			if (strcmp("Link",oldType)!=0)
+			if (strcmp("Link", oldType) != 0)
 			{
 				dataChunk.len = curLen * dataChunk.nr;
-			} else
+			}
+			else
 			{
-//				printf("keep length of link = %d\n",dataChunk.len);
+				//				printf("keep length of link = %d\n",dataChunk.len);
 			}
 
 			//write the structure header
-			fwrite(&dataChunk,sizeof(bChunkInd),1,fp);
+			fwrite(&dataChunk, sizeof(bChunkInd), 1, fp);
 
-
-
-			short int* curStruct1;
+			short int *curStruct1;
 			curStruct1 = mMemoryDNA->getStruct(dataChunk.dna_nr);
 			assert(curStruct1 == curStruct);
 
-			char* cur	= fixupPointers  ?  (char*)findLibPointer(dataChunk.oldPtr) : (char*)dataChunk.oldPtr;
+			char *cur = fixupPointers ? (char *)findLibPointer(dataChunk.oldPtr) : (char *)dataChunk.oldPtr;
 
 			//write the actual contents of the structure(s)
-			fwrite(cur,dataChunk.len,1,fp);
-		} else
+			fwrite(cur, dataChunk.len, 1, fp);
+		}
+		else
 		{
 			printf("serious error, struct mismatch: don't write\n");
 		}
 	}
-
 }
 
-
 // ----------------------------------------------------- //
-int bFile::getNextBlock(bChunkInd *dataChunk,  const char *dataPtr, const int flags)
+int bFile::getNextBlock(bChunkInd *dataChunk, const char *dataPtr, const int flags)
 {
 	bool swap = false;
 	bool varies = false;
 
-	if (flags &FD_ENDIAN_SWAP)
+	if (flags & FD_ENDIAN_SWAP)
 		swap = true;
-	if (flags &FD_BITS_VARIES)
+	if (flags & FD_BITS_VARIES)
 		varies = true;
 
 	if (VOID_IS_8)
@@ -1621,27 +1540,25 @@ int bFile::getNextBlock(bChunkInd *dataChunk,  const char *dataPtr, const int fl
 			bChunkPtr4 head;
 			memcpy(&head, dataPtr, sizeof(bChunkPtr4));
 
-
 			bChunkPtr8 chunk;
 
-			chunk.code		= head.code;
-			chunk.len		= head.len;
+			chunk.code = head.code;
+			chunk.len = head.len;
 			chunk.m_uniqueInts[0] = head.m_uniqueInt;
 			chunk.m_uniqueInts[1] = 0;
-			chunk.dna_nr	= head.dna_nr;
-			chunk.nr		= head.nr;
+			chunk.dna_nr = head.dna_nr;
+			chunk.nr = head.nr;
 
 			if (swap)
 			{
-				if ((chunk.code & 0xFFFF)==0)
-					chunk.code >>=16;
+				if ((chunk.code & 0xFFFF) == 0)
+					chunk.code >>= 16;
 
 				B3_SWITCH_INT(chunk.len);
 				B3_SWITCH_INT(chunk.dna_nr);
 				B3_SWITCH_INT(chunk.nr);
 			}
 
-
 			memcpy(dataChunk, &chunk, sizeof(bChunkInd));
 		}
 		else
@@ -1651,8 +1568,8 @@ int bFile::getNextBlock(bChunkInd *dataChunk,  const char *dataPtr, const int fl
 
 			if (swap)
 			{
-				if ((c.code & 0xFFFF)==0)
-					c.code >>=16;
+				if ((c.code & 0xFFFF) == 0)
+					c.code >>= 16;
 
 				B3_SWITCH_INT(c.len);
 				B3_SWITCH_INT(c.dna_nr);
@@ -1669,31 +1586,30 @@ int bFile::getNextBlock(bChunkInd *dataChunk,  const char *dataPtr, const int fl
 			bChunkPtr8 head;
 			memcpy(&head, dataPtr, sizeof(bChunkPtr8));
 
-
 			bChunkPtr4 chunk;
 			chunk.code = head.code;
 			chunk.len = head.len;
 
-			if (head.m_uniqueInts[0]==head.m_uniqueInts[1])
+			if (head.m_uniqueInts[0] == head.m_uniqueInts[1])
 			{
 				chunk.m_uniqueInt = head.m_uniqueInts[0];
-			} else
+			}
+			else
 			{
-				b3Long64 oldPtr =0;
+				b3Long64 oldPtr = 0;
 				memcpy(&oldPtr, &head.m_uniqueInts[0], 8);
 				if (swap)
 					B3_SWITCH_LONGINT(oldPtr);
 				chunk.m_uniqueInt = (int)(oldPtr >> 3);
 			}
 
-
 			chunk.dna_nr = head.dna_nr;
 			chunk.nr = head.nr;
 
 			if (swap)
 			{
-				if ((chunk.code & 0xFFFF)==0)
-					chunk.code >>=16;
+				if ((chunk.code & 0xFFFF) == 0)
+					chunk.code >>= 16;
 
 				B3_SWITCH_INT(chunk.len);
 				B3_SWITCH_INT(chunk.dna_nr);
@@ -1709,8 +1625,8 @@ int bFile::getNextBlock(bChunkInd *dataChunk,  const char *dataPtr, const int fl
 
 			if (swap)
 			{
-				if ((c.code & 0xFFFF)==0)
-					c.code >>=16;
+				if ((c.code & 0xFFFF) == 0)
+					c.code >>= 16;
 
 				B3_SWITCH_INT(c.len);
 				B3_SWITCH_INT(c.dna_nr);
@@ -1731,9 +1647,7 @@ int bFile::getNextBlock(bChunkInd *dataChunk,  const char *dataPtr, const int fl
 	print (dataChunk->dna_nr);
 	print (dataChunk->nr);
 #endif
-	return (dataChunk->len+ChunkUtils::getOffset(flags));
+	return (dataChunk->len + ChunkUtils::getOffset(flags));
 }
 
-
-
 //eof
diff --git a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3File.h b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3File.h
index 861056806d7..bda229cfbd7 100644
--- a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3File.h
+++ b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3File.h
@@ -20,146 +20,139 @@ subject to the following restrictions:
 #include "b3Chunk.h"
 #include <stdio.h>
 
-namespace bParse {
+namespace bParse
+{
+// ----------------------------------------------------- //
+enum bFileFlags
+{
+	FD_INVALID = 0,
+	FD_OK = 1,
+	FD_VOID_IS_8 = 2,
+	FD_ENDIAN_SWAP = 4,
+	FD_FILE_64 = 8,
+	FD_BITS_VARIES = 16,
+	FD_VERSION_VARIES = 32,
+	FD_DOUBLE_PRECISION = 64,
+	FD_BROKEN_DNA = 128
+};
 
-	// ----------------------------------------------------- //
-	enum bFileFlags
+enum bFileVerboseMode
+{
+	FD_VERBOSE_EXPORT_XML = 1,
+	FD_VERBOSE_DUMP_DNA_TYPE_DEFINITIONS = 2,
+	FD_VERBOSE_DUMP_CHUNKS = 4,
+	FD_VERBOSE_DUMP_FILE_INFO = 8,
+};
+// ----------------------------------------------------- //
+class bFile
+{
+protected:
+	char m_headerString[7];
+
+	bool mOwnsBuffer;
+	char *mFileBuffer;
+	int mFileLen;
+	int mVersion;
+
+	bPtrMap mLibPointers;
+
+	int mDataStart;
+	bDNA *mFileDNA;
+	bDNA *mMemoryDNA;
+
+	b3AlignedObjectArray<char *> m_pointerFixupArray;
+	b3AlignedObjectArray<char *> m_pointerPtrFixupArray;
+
+	b3AlignedObjectArray<bChunkInd> m_chunks;
+	b3HashMap<b3HashPtr, bChunkInd> m_chunkPtrPtrMap;
+
+	//
+
+	bPtrMap mDataPointers;
+
+	int mFlags;
+
+	// ////////////////////////////////////////////////////////////////////////////
+
+	// buffer offset util
+	int getNextBlock(bChunkInd *dataChunk, const char *dataPtr, const int flags);
+	void safeSwapPtr(char *dst, const char *src);
+
+	virtual void parseHeader();
+
+	virtual void parseData() = 0;
+
+	void resolvePointersMismatch();
+	void resolvePointersChunk(const bChunkInd &dataChunk, int verboseMode);
+
+	int resolvePointersStructRecursive(char *strcPtr, int old_dna, int verboseMode, int recursion);
+	//void swapPtr(char *dst, char *src);
+
+	void parseStruct(char *strcPtr, char *dtPtr, int old_dna, int new_dna, bool fixupPointers);
+	void getMatchingFileDNA(short *old, const char *lookupName, const char *lookupType, char *strcData, char *data, bool fixupPointers);
+	char *getFileElement(short *firstStruct, char *lookupName, char *lookupType, char *data, short **foundPos);
+
+	void swap(char *head, class bChunkInd &ch, bool ignoreEndianFlag);
+	void swapData(char *data, short type, int arraySize, bool ignoreEndianFlag);
+	void swapStruct(int dna_nr, char *data, bool ignoreEndianFlag);
+	void swapLen(char *dataPtr);
+	void swapDNA(char *ptr);
+
+	char *readStruct(char *head, class bChunkInd &chunk);
+	char *getAsString(int code);
+
+	void parseInternal(int verboseMode, char *memDna, int memDnaLength);
+
+public:
+	bFile(const char *filename, const char headerString[7]);
+
+	//todo: make memoryBuffer const char
+	//bFile( const char *memoryBuffer, int len);
+	bFile(char *memoryBuffer, int len, const char headerString[7]);
+	virtual ~bFile();
+
+	bDNA *getFileDNA()
 	{
-		FD_INVALID   =0,
-		FD_OK        =1,
-		FD_VOID_IS_8 =2,
-		FD_ENDIAN_SWAP      =4,
-		FD_FILE_64   =8,
-		FD_BITS_VARIES    =16,
-		FD_VERSION_VARIES = 32,
-		FD_DOUBLE_PRECISION =64,
-		FD_BROKEN_DNA = 128
-	};
+		return mFileDNA;
+	}
 
-	enum bFileVerboseMode
+	virtual void addDataBlock(char *dataBlock) = 0;
+
+	int getFlags() const
 	{
-		FD_VERBOSE_EXPORT_XML = 1,
-		FD_VERBOSE_DUMP_DNA_TYPE_DEFINITIONS = 2,
-		FD_VERBOSE_DUMP_CHUNKS = 4,
-		FD_VERBOSE_DUMP_FILE_INFO=8,
-	};
-	// ----------------------------------------------------- //
-	class bFile
+		return mFlags;
+	}
+
+	bPtrMap &getLibPointers()
 	{
-	protected:
-		
-		char				m_headerString[7];
+		return mLibPointers;
+	}
 
-		bool				mOwnsBuffer;
-		char*				mFileBuffer;
-		int					mFileLen;
-		int					mVersion;
+	void *findLibPointer(void *ptr);
 
+	bool ok();
 
-		bPtrMap				mLibPointers;
+	virtual void parse(int verboseMode) = 0;
 
-		int					mDataStart;
-		bDNA*				mFileDNA;
-		bDNA*				mMemoryDNA;
+	virtual int write(const char *fileName, bool fixupPointers = false) = 0;
 
-		b3AlignedObjectArray<char*>	m_pointerFixupArray;
-		b3AlignedObjectArray<char*>	m_pointerPtrFixupArray;
-		
-		b3AlignedObjectArray<bChunkInd>	m_chunks;
-        b3HashMap<b3HashPtr, bChunkInd> m_chunkPtrPtrMap;
+	virtual void writeChunks(FILE *fp, bool fixupPointers);
 
-        // 
-	
-		bPtrMap				mDataPointers;
+	virtual void writeDNA(FILE *fp) = 0;
 
-		
-		int					mFlags;
+	void updateOldPointers();
+	void resolvePointers(int verboseMode);
 
-		// ////////////////////////////////////////////////////////////////////////////
+	void dumpChunks(bDNA *dna);
 
-			// buffer offset util
-		int getNextBlock(bChunkInd *dataChunk,  const char *dataPtr, const int flags);
-		void safeSwapPtr(char *dst, const char *src);
+	int getVersion() const
+	{
+		return mVersion;
+	}
+	//pre-swap the endianness, so that data loaded on a target with different endianness doesn't need to be swapped
+	void preSwap();
+	void writeFile(const char *fileName);
+};
+}  // namespace bParse
 
-		virtual	void parseHeader();
-		
-		virtual	void parseData() = 0;
-
-		void resolvePointersMismatch();
-		void resolvePointersChunk(const bChunkInd& dataChunk, int verboseMode);
-
-		int resolvePointersStructRecursive(char *strcPtr, int old_dna, int verboseMode, int recursion);
-		//void swapPtr(char *dst, char *src);
-
-		void parseStruct(char *strcPtr, char *dtPtr, int old_dna, int new_dna, bool fixupPointers);
-		void getMatchingFileDNA(short* old, const char* lookupName, const char* lookupType, char *strcData, char *data, bool fixupPointers);
-		char* getFileElement(short *firstStruct, char *lookupName, char *lookupType, char *data, short **foundPos);
-
-
-		void swap(char *head, class bChunkInd& ch, bool ignoreEndianFlag);
-		void swapData(char *data, short type, int arraySize, bool ignoreEndianFlag);
-		void swapStruct(int dna_nr, char *data, bool ignoreEndianFlag);
-		void swapLen(char *dataPtr);
-		void swapDNA(char* ptr);
-
-
-		char* readStruct(char *head, class bChunkInd& chunk);
-		char *getAsString(int code);
-
-		void	parseInternal(int verboseMode, char* memDna,int memDnaLength);
-
-	public:
-		bFile(const char *filename, const char headerString[7]);
-		
-		//todo: make memoryBuffer const char
-		//bFile( const char *memoryBuffer, int len);
-		bFile( char *memoryBuffer, int len, const char headerString[7]);
-		virtual ~bFile();
-
-		bDNA*				getFileDNA()
-		{
-			return mFileDNA;
-		}
-
-		virtual	void	addDataBlock(char* dataBlock) = 0;
-
-		int	getFlags() const
-		{
-			return mFlags;
-		}
-
-		bPtrMap&		getLibPointers()
-		{
-			return mLibPointers;
-		}
-		
-		void* findLibPointer(void *ptr);
-
-		bool ok();
-
-		virtual	void parse(int verboseMode) = 0;
-
-		virtual	int	write(const char* fileName, bool fixupPointers=false) = 0;
-
-		virtual	void	writeChunks(FILE* fp, bool fixupPointers );
-
-		virtual	void	writeDNA(FILE* fp) = 0;
-
-		void	updateOldPointers();
-		void	resolvePointers(int verboseMode);
-
-		void	dumpChunks(bDNA* dna);
-		
-		int		getVersion() const
-		{
-			return mVersion;
-		}
-		//pre-swap the endianness, so that data loaded on a target with different endianness doesn't need to be swapped
-		void preSwap();
-		void writeFile(const char* fileName);
-
-	};
-}
-
-
-#endif//__BFILE_H__
+#endif  //__BFILE_H__
diff --git a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Serializer.cpp b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Serializer.cpp
index c6a2a832ad9..ea4a8e20073 100644
--- a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Serializer.cpp
+++ b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Serializer.cpp
@@ -1,908 +1,18062 @@
-char b3s_bulletDNAstr[]= {
-char(83),char(68),char(78),char(65),char(78),char(65),char(77),char(69),char(63),char(1),char(0),char(0),char(109),char(95),char(115),char(105),char(122),char(101),char(0),char(109),
-char(95),char(99),char(97),char(112),char(97),char(99),char(105),char(116),char(121),char(0),char(42),char(109),char(95),char(100),char(97),char(116),char(97),char(0),char(109),char(95),
-char(99),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(83),char(104),char(97),char(112),char(101),char(115),char(0),char(109),char(95),char(99),char(111),
-char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(79),char(98),char(106),char(101),char(99),char(116),char(115),char(0),char(109),char(95),char(99),char(111),char(110),
-char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(115),char(0),char(42),char(102),char(105),char(114),char(115),char(116),char(0),char(42),char(108),char(97),char(115),
-char(116),char(0),char(109),char(95),char(102),char(108),char(111),char(97),char(116),char(115),char(91),char(52),char(93),char(0),char(109),char(95),char(101),char(108),char(91),char(51),
-char(93),char(0),char(109),char(95),char(98),char(97),char(115),char(105),char(115),char(0),char(109),char(95),char(111),char(114),char(105),char(103),char(105),char(110),char(0),char(109),
-char(95),char(114),char(111),char(111),char(116),char(78),char(111),char(100),char(101),char(73),char(110),char(100),char(101),char(120),char(0),char(109),char(95),char(115),char(117),char(98),
-char(116),char(114),char(101),char(101),char(83),char(105),char(122),char(101),char(0),char(109),char(95),char(113),char(117),char(97),char(110),char(116),char(105),char(122),char(101),char(100),
-char(65),char(97),char(98),char(98),char(77),char(105),char(110),char(91),char(51),char(93),char(0),char(109),char(95),char(113),char(117),char(97),char(110),char(116),char(105),char(122),
-char(101),char(100),char(65),char(97),char(98),char(98),char(77),char(97),char(120),char(91),char(51),char(93),char(0),char(109),char(95),char(97),char(97),char(98),char(98),char(77),
-char(105),char(110),char(79),char(114),char(103),char(0),char(109),char(95),char(97),char(97),char(98),char(98),char(77),char(97),char(120),char(79),char(114),char(103),char(0),char(109),
-char(95),char(101),char(115),char(99),char(97),char(112),char(101),char(73),char(110),char(100),char(101),char(120),char(0),char(109),char(95),char(115),char(117),char(98),char(80),char(97),
-char(114),char(116),char(0),char(109),char(95),char(116),char(114),char(105),char(97),char(110),char(103),char(108),char(101),char(73),char(110),char(100),char(101),char(120),char(0),char(109),
-char(95),char(112),char(97),char(100),char(91),char(52),char(93),char(0),char(109),char(95),char(101),char(115),char(99),char(97),char(112),char(101),char(73),char(110),char(100),char(101),
-char(120),char(79),char(114),char(84),char(114),char(105),char(97),char(110),char(103),char(108),char(101),char(73),char(110),char(100),char(101),char(120),char(0),char(109),char(95),char(98),
-char(118),char(104),char(65),char(97),char(98),char(98),char(77),char(105),char(110),char(0),char(109),char(95),char(98),char(118),char(104),char(65),char(97),char(98),char(98),char(77),
-char(97),char(120),char(0),char(109),char(95),char(98),char(118),char(104),char(81),char(117),char(97),char(110),char(116),char(105),char(122),char(97),char(116),char(105),char(111),char(110),
-char(0),char(109),char(95),char(99),char(117),char(114),char(78),char(111),char(100),char(101),char(73),char(110),char(100),char(101),char(120),char(0),char(109),char(95),char(117),char(115),
-char(101),char(81),char(117),char(97),char(110),char(116),char(105),char(122),char(97),char(116),char(105),char(111),char(110),char(0),char(109),char(95),char(110),char(117),char(109),char(67),
-char(111),char(110),char(116),char(105),char(103),char(117),char(111),char(117),char(115),char(76),char(101),char(97),char(102),char(78),char(111),char(100),char(101),char(115),char(0),char(109),
-char(95),char(110),char(117),char(109),char(81),char(117),char(97),char(110),char(116),char(105),char(122),char(101),char(100),char(67),char(111),char(110),char(116),char(105),char(103),char(117),
-char(111),char(117),char(115),char(78),char(111),char(100),char(101),char(115),char(0),char(42),char(109),char(95),char(99),char(111),char(110),char(116),char(105),char(103),char(117),char(111),
-char(117),char(115),char(78),char(111),char(100),char(101),char(115),char(80),char(116),char(114),char(0),char(42),char(109),char(95),char(113),char(117),char(97),char(110),char(116),char(105),
-char(122),char(101),char(100),char(67),char(111),char(110),char(116),char(105),char(103),char(117),char(111),char(117),char(115),char(78),char(111),char(100),char(101),char(115),char(80),char(116),
-char(114),char(0),char(42),char(109),char(95),char(115),char(117),char(98),char(84),char(114),char(101),char(101),char(73),char(110),char(102),char(111),char(80),char(116),char(114),char(0),
-char(109),char(95),char(116),char(114),char(97),char(118),char(101),char(114),char(115),char(97),char(108),char(77),char(111),char(100),char(101),char(0),char(109),char(95),char(110),char(117),
-char(109),char(83),char(117),char(98),char(116),char(114),char(101),char(101),char(72),char(101),char(97),char(100),char(101),char(114),char(115),char(0),char(42),char(109),char(95),char(110),
-char(97),char(109),char(101),char(0),char(109),char(95),char(115),char(104),char(97),char(112),char(101),char(84),char(121),char(112),char(101),char(0),char(109),char(95),char(112),char(97),
-char(100),char(100),char(105),char(110),char(103),char(91),char(52),char(93),char(0),char(109),char(95),char(99),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),
-char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(109),char(95),char(108),char(111),char(99),char(97),char(108),char(83),char(99),char(97),
-char(108),char(105),char(110),char(103),char(0),char(109),char(95),char(112),char(108),char(97),char(110),char(101),char(78),char(111),char(114),char(109),char(97),char(108),char(0),char(109),
-char(95),char(112),char(108),char(97),char(110),char(101),char(67),char(111),char(110),char(115),char(116),char(97),char(110),char(116),char(0),char(109),char(95),char(105),char(109),char(112),
-char(108),char(105),char(99),char(105),char(116),char(83),char(104),char(97),char(112),char(101),char(68),char(105),char(109),char(101),char(110),char(115),char(105),char(111),char(110),char(115),
-char(0),char(109),char(95),char(99),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(77),char(97),char(114),char(103),char(105),char(110),char(0),char(109),
-char(95),char(112),char(97),char(100),char(100),char(105),char(110),char(103),char(0),char(109),char(95),char(112),char(111),char(115),char(0),char(109),char(95),char(114),char(97),char(100),
-char(105),char(117),char(115),char(0),char(109),char(95),char(99),char(111),char(110),char(118),char(101),char(120),char(73),char(110),char(116),char(101),char(114),char(110),char(97),char(108),
-char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(42),char(109),char(95),char(108),char(111),char(99),char(97),char(108),char(80),char(111),
-char(115),char(105),char(116),char(105),char(111),char(110),char(65),char(114),char(114),char(97),char(121),char(80),char(116),char(114),char(0),char(109),char(95),char(108),char(111),char(99),
-char(97),char(108),char(80),char(111),char(115),char(105),char(116),char(105),char(111),char(110),char(65),char(114),char(114),char(97),char(121),char(83),char(105),char(122),char(101),char(0),
-char(109),char(95),char(118),char(97),char(108),char(117),char(101),char(0),char(109),char(95),char(112),char(97),char(100),char(91),char(50),char(93),char(0),char(109),char(95),char(118),
-char(97),char(108),char(117),char(101),char(115),char(91),char(51),char(93),char(0),char(109),char(95),char(112),char(97),char(100),char(0),char(42),char(109),char(95),char(118),char(101),
-char(114),char(116),char(105),char(99),char(101),char(115),char(51),char(102),char(0),char(42),char(109),char(95),char(118),char(101),char(114),char(116),char(105),char(99),char(101),char(115),
-char(51),char(100),char(0),char(42),char(109),char(95),char(105),char(110),char(100),char(105),char(99),char(101),char(115),char(51),char(50),char(0),char(42),char(109),char(95),char(51),
-char(105),char(110),char(100),char(105),char(99),char(101),char(115),char(49),char(54),char(0),char(42),char(109),char(95),char(51),char(105),char(110),char(100),char(105),char(99),char(101),
-char(115),char(56),char(0),char(42),char(109),char(95),char(105),char(110),char(100),char(105),char(99),char(101),char(115),char(49),char(54),char(0),char(109),char(95),char(110),char(117),
-char(109),char(84),char(114),char(105),char(97),char(110),char(103),char(108),char(101),char(115),char(0),char(109),char(95),char(110),char(117),char(109),char(86),char(101),char(114),char(116),
-char(105),char(99),char(101),char(115),char(0),char(42),char(109),char(95),char(109),char(101),char(115),char(104),char(80),char(97),char(114),char(116),char(115),char(80),char(116),char(114),
-char(0),char(109),char(95),char(115),char(99),char(97),char(108),char(105),char(110),char(103),char(0),char(109),char(95),char(110),char(117),char(109),char(77),char(101),char(115),char(104),
-char(80),char(97),char(114),char(116),char(115),char(0),char(109),char(95),char(109),char(101),char(115),char(104),char(73),char(110),char(116),char(101),char(114),char(102),char(97),char(99),
-char(101),char(0),char(42),char(109),char(95),char(113),char(117),char(97),char(110),char(116),char(105),char(122),char(101),char(100),char(70),char(108),char(111),char(97),char(116),char(66),
-char(118),char(104),char(0),char(42),char(109),char(95),char(113),char(117),char(97),char(110),char(116),char(105),char(122),char(101),char(100),char(68),char(111),char(117),char(98),char(108),
-char(101),char(66),char(118),char(104),char(0),char(42),char(109),char(95),char(116),char(114),char(105),char(97),char(110),char(103),char(108),char(101),char(73),char(110),char(102),char(111),
-char(77),char(97),char(112),char(0),char(109),char(95),char(112),char(97),char(100),char(51),char(91),char(52),char(93),char(0),char(109),char(95),char(116),char(114),char(105),char(109),
-char(101),char(115),char(104),char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(109),char(95),char(116),char(114),char(97),char(110),char(115),
-char(102),char(111),char(114),char(109),char(0),char(42),char(109),char(95),char(99),char(104),char(105),char(108),char(100),char(83),char(104),char(97),char(112),char(101),char(0),char(109),
-char(95),char(99),char(104),char(105),char(108),char(100),char(83),char(104),char(97),char(112),char(101),char(84),char(121),char(112),char(101),char(0),char(109),char(95),char(99),char(104),
-char(105),char(108),char(100),char(77),char(97),char(114),char(103),char(105),char(110),char(0),char(42),char(109),char(95),char(99),char(104),char(105),char(108),char(100),char(83),char(104),
-char(97),char(112),char(101),char(80),char(116),char(114),char(0),char(109),char(95),char(110),char(117),char(109),char(67),char(104),char(105),char(108),char(100),char(83),char(104),char(97),
-char(112),char(101),char(115),char(0),char(109),char(95),char(117),char(112),char(65),char(120),char(105),char(115),char(0),char(109),char(95),char(102),char(108),char(97),char(103),char(115),
-char(0),char(109),char(95),char(101),char(100),char(103),char(101),char(86),char(48),char(86),char(49),char(65),char(110),char(103),char(108),char(101),char(0),char(109),char(95),char(101),
-char(100),char(103),char(101),char(86),char(49),char(86),char(50),char(65),char(110),char(103),char(108),char(101),char(0),char(109),char(95),char(101),char(100),char(103),char(101),char(86),
-char(50),char(86),char(48),char(65),char(110),char(103),char(108),char(101),char(0),char(42),char(109),char(95),char(104),char(97),char(115),char(104),char(84),char(97),char(98),char(108),
-char(101),char(80),char(116),char(114),char(0),char(42),char(109),char(95),char(110),char(101),char(120),char(116),char(80),char(116),char(114),char(0),char(42),char(109),char(95),char(118),
-char(97),char(108),char(117),char(101),char(65),char(114),char(114),char(97),char(121),char(80),char(116),char(114),char(0),char(42),char(109),char(95),char(107),char(101),char(121),char(65),
-char(114),char(114),char(97),char(121),char(80),char(116),char(114),char(0),char(109),char(95),char(99),char(111),char(110),char(118),char(101),char(120),char(69),char(112),char(115),char(105),
-char(108),char(111),char(110),char(0),char(109),char(95),char(112),char(108),char(97),char(110),char(97),char(114),char(69),char(112),char(115),char(105),char(108),char(111),char(110),char(0),
-char(109),char(95),char(101),char(113),char(117),char(97),char(108),char(86),char(101),char(114),char(116),char(101),char(120),char(84),char(104),char(114),char(101),char(115),char(104),char(111),
-char(108),char(100),char(0),char(109),char(95),char(101),char(100),char(103),char(101),char(68),char(105),char(115),char(116),char(97),char(110),char(99),char(101),char(84),char(104),char(114),
-char(101),char(115),char(104),char(111),char(108),char(100),char(0),char(109),char(95),char(122),char(101),char(114),char(111),char(65),char(114),char(101),char(97),char(84),char(104),char(114),
-char(101),char(115),char(104),char(111),char(108),char(100),char(0),char(109),char(95),char(110),char(101),char(120),char(116),char(83),char(105),char(122),char(101),char(0),char(109),char(95),
-char(104),char(97),char(115),char(104),char(84),char(97),char(98),char(108),char(101),char(83),char(105),char(122),char(101),char(0),char(109),char(95),char(110),char(117),char(109),char(86),
-char(97),char(108),char(117),char(101),char(115),char(0),char(109),char(95),char(110),char(117),char(109),char(75),char(101),char(121),char(115),char(0),char(109),char(95),char(103),char(105),
-char(109),char(112),char(97),char(99),char(116),char(83),char(117),char(98),char(84),char(121),char(112),char(101),char(0),char(42),char(109),char(95),char(117),char(110),char(115),char(99),
-char(97),char(108),char(101),char(100),char(80),char(111),char(105),char(110),char(116),char(115),char(70),char(108),char(111),char(97),char(116),char(80),char(116),char(114),char(0),char(42),
-char(109),char(95),char(117),char(110),char(115),char(99),char(97),char(108),char(101),char(100),char(80),char(111),char(105),char(110),char(116),char(115),char(68),char(111),char(117),char(98),
-char(108),char(101),char(80),char(116),char(114),char(0),char(109),char(95),char(110),char(117),char(109),char(85),char(110),char(115),char(99),char(97),char(108),char(101),char(100),char(80),
-char(111),char(105),char(110),char(116),char(115),char(0),char(109),char(95),char(112),char(97),char(100),char(100),char(105),char(110),char(103),char(51),char(91),char(52),char(93),char(0),
-char(42),char(109),char(95),char(98),char(114),char(111),char(97),char(100),char(112),char(104),char(97),char(115),char(101),char(72),char(97),char(110),char(100),char(108),char(101),char(0),
-char(42),char(109),char(95),char(99),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(83),char(104),char(97),char(112),char(101),char(0),char(42),char(109),
-char(95),char(114),char(111),char(111),char(116),char(67),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(83),char(104),char(97),char(112),char(101),char(0),
-char(109),char(95),char(119),char(111),char(114),char(108),char(100),char(84),char(114),char(97),char(110),char(115),char(102),char(111),char(114),char(109),char(0),char(109),char(95),char(105),
-char(110),char(116),char(101),char(114),char(112),char(111),char(108),char(97),char(116),char(105),char(111),char(110),char(87),char(111),char(114),char(108),char(100),char(84),char(114),char(97),
-char(110),char(115),char(102),char(111),char(114),char(109),char(0),char(109),char(95),char(105),char(110),char(116),char(101),char(114),char(112),char(111),char(108),char(97),char(116),char(105),
-char(111),char(110),char(76),char(105),char(110),char(101),char(97),char(114),char(86),char(101),char(108),char(111),char(99),char(105),char(116),char(121),char(0),char(109),char(95),char(105),
-char(110),char(116),char(101),char(114),char(112),char(111),char(108),char(97),char(116),char(105),char(111),char(110),char(65),char(110),char(103),char(117),char(108),char(97),char(114),char(86),
-char(101),char(108),char(111),char(99),char(105),char(116),char(121),char(0),char(109),char(95),char(97),char(110),char(105),char(115),char(111),char(116),char(114),char(111),char(112),char(105),
-char(99),char(70),char(114),char(105),char(99),char(116),char(105),char(111),char(110),char(0),char(109),char(95),char(99),char(111),char(110),char(116),char(97),char(99),char(116),char(80),
-char(114),char(111),char(99),char(101),char(115),char(115),char(105),char(110),char(103),char(84),char(104),char(114),char(101),char(115),char(104),char(111),char(108),char(100),char(0),char(109),
-char(95),char(100),char(101),char(97),char(99),char(116),char(105),char(118),char(97),char(116),char(105),char(111),char(110),char(84),char(105),char(109),char(101),char(0),char(109),char(95),
-char(102),char(114),char(105),char(99),char(116),char(105),char(111),char(110),char(0),char(109),char(95),char(114),char(111),char(108),char(108),char(105),char(110),char(103),char(70),char(114),
-char(105),char(99),char(116),char(105),char(111),char(110),char(0),char(109),char(95),char(114),char(101),char(115),char(116),char(105),char(116),char(117),char(116),char(105),char(111),char(110),
-char(0),char(109),char(95),char(104),char(105),char(116),char(70),char(114),char(97),char(99),char(116),char(105),char(111),char(110),char(0),char(109),char(95),char(99),char(99),char(100),
-char(83),char(119),char(101),char(112),char(116),char(83),char(112),char(104),char(101),char(114),char(101),char(82),char(97),char(100),char(105),char(117),char(115),char(0),char(109),char(95),
-char(99),char(99),char(100),char(77),char(111),char(116),char(105),char(111),char(110),char(84),char(104),char(114),char(101),char(115),char(104),char(111),char(108),char(100),char(0),char(109),
-char(95),char(104),char(97),char(115),char(65),char(110),char(105),char(115),char(111),char(116),char(114),char(111),char(112),char(105),char(99),char(70),char(114),char(105),char(99),char(116),
-char(105),char(111),char(110),char(0),char(109),char(95),char(99),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(70),char(108),char(97),char(103),char(115),
-char(0),char(109),char(95),char(105),char(115),char(108),char(97),char(110),char(100),char(84),char(97),char(103),char(49),char(0),char(109),char(95),char(99),char(111),char(109),char(112),
-char(97),char(110),char(105),char(111),char(110),char(73),char(100),char(0),char(109),char(95),char(97),char(99),char(116),char(105),char(118),char(97),char(116),char(105),char(111),char(110),
-char(83),char(116),char(97),char(116),char(101),char(49),char(0),char(109),char(95),char(105),char(110),char(116),char(101),char(114),char(110),char(97),char(108),char(84),char(121),char(112),
-char(101),char(0),char(109),char(95),char(99),char(104),char(101),char(99),char(107),char(67),char(111),char(108),char(108),char(105),char(100),char(101),char(87),char(105),char(116),char(104),
-char(0),char(109),char(95),char(115),char(111),char(108),char(118),char(101),char(114),char(73),char(110),char(102),char(111),char(0),char(109),char(95),char(103),char(114),char(97),char(118),
-char(105),char(116),char(121),char(0),char(109),char(95),char(99),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(79),char(98),char(106),char(101),char(99),
-char(116),char(68),char(97),char(116),char(97),char(0),char(109),char(95),char(105),char(110),char(118),char(73),char(110),char(101),char(114),char(116),char(105),char(97),char(84),char(101),
-char(110),char(115),char(111),char(114),char(87),char(111),char(114),char(108),char(100),char(0),char(109),char(95),char(108),char(105),char(110),char(101),char(97),char(114),char(86),char(101),
-char(108),char(111),char(99),char(105),char(116),char(121),char(0),char(109),char(95),char(97),char(110),char(103),char(117),char(108),char(97),char(114),char(86),char(101),char(108),char(111),
-char(99),char(105),char(116),char(121),char(0),char(109),char(95),char(97),char(110),char(103),char(117),char(108),char(97),char(114),char(70),char(97),char(99),char(116),char(111),char(114),
-char(0),char(109),char(95),char(108),char(105),char(110),char(101),char(97),char(114),char(70),char(97),char(99),char(116),char(111),char(114),char(0),char(109),char(95),char(103),char(114),
-char(97),char(118),char(105),char(116),char(121),char(95),char(97),char(99),char(99),char(101),char(108),char(101),char(114),char(97),char(116),char(105),char(111),char(110),char(0),char(109),
-char(95),char(105),char(110),char(118),char(73),char(110),char(101),char(114),char(116),char(105),char(97),char(76),char(111),char(99),char(97),char(108),char(0),char(109),char(95),char(116),
-char(111),char(116),char(97),char(108),char(70),char(111),char(114),char(99),char(101),char(0),char(109),char(95),char(116),char(111),char(116),char(97),char(108),char(84),char(111),char(114),
-char(113),char(117),char(101),char(0),char(109),char(95),char(105),char(110),char(118),char(101),char(114),char(115),char(101),char(77),char(97),char(115),char(115),char(0),char(109),char(95),
-char(108),char(105),char(110),char(101),char(97),char(114),char(68),char(97),char(109),char(112),char(105),char(110),char(103),char(0),char(109),char(95),char(97),char(110),char(103),char(117),
-char(108),char(97),char(114),char(68),char(97),char(109),char(112),char(105),char(110),char(103),char(0),char(109),char(95),char(97),char(100),char(100),char(105),char(116),char(105),char(111),
-char(110),char(97),char(108),char(68),char(97),char(109),char(112),char(105),char(110),char(103),char(70),char(97),char(99),char(116),char(111),char(114),char(0),char(109),char(95),char(97),
-char(100),char(100),char(105),char(116),char(105),char(111),char(110),char(97),char(108),char(76),char(105),char(110),char(101),char(97),char(114),char(68),char(97),char(109),char(112),char(105),
-char(110),char(103),char(84),char(104),char(114),char(101),char(115),char(104),char(111),char(108),char(100),char(83),char(113),char(114),char(0),char(109),char(95),char(97),char(100),char(100),
-char(105),char(116),char(105),char(111),char(110),char(97),char(108),char(65),char(110),char(103),char(117),char(108),char(97),char(114),char(68),char(97),char(109),char(112),char(105),char(110),
-char(103),char(84),char(104),char(114),char(101),char(115),char(104),char(111),char(108),char(100),char(83),char(113),char(114),char(0),char(109),char(95),char(97),char(100),char(100),char(105),
-char(116),char(105),char(111),char(110),char(97),char(108),char(65),char(110),char(103),char(117),char(108),char(97),char(114),char(68),char(97),char(109),char(112),char(105),char(110),char(103),
-char(70),char(97),char(99),char(116),char(111),char(114),char(0),char(109),char(95),char(108),char(105),char(110),char(101),char(97),char(114),char(83),char(108),char(101),char(101),char(112),
-char(105),char(110),char(103),char(84),char(104),char(114),char(101),char(115),char(104),char(111),char(108),char(100),char(0),char(109),char(95),char(97),char(110),char(103),char(117),char(108),
-char(97),char(114),char(83),char(108),char(101),char(101),char(112),char(105),char(110),char(103),char(84),char(104),char(114),char(101),char(115),char(104),char(111),char(108),char(100),char(0),
-char(109),char(95),char(97),char(100),char(100),char(105),char(116),char(105),char(111),char(110),char(97),char(108),char(68),char(97),char(109),char(112),char(105),char(110),char(103),char(0),
-char(109),char(95),char(110),char(117),char(109),char(67),char(111),char(110),char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(82),char(111),char(119),char(115),char(0),
-char(110),char(117),char(98),char(0),char(42),char(109),char(95),char(114),char(98),char(65),char(0),char(42),char(109),char(95),char(114),char(98),char(66),char(0),char(109),char(95),
-char(111),char(98),char(106),char(101),char(99),char(116),char(84),char(121),char(112),char(101),char(0),char(109),char(95),char(117),char(115),char(101),char(114),char(67),char(111),char(110),
-char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(84),char(121),char(112),char(101),char(0),char(109),char(95),char(117),char(115),char(101),char(114),char(67),char(111),
-char(110),char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(73),char(100),char(0),char(109),char(95),char(110),char(101),char(101),char(100),char(115),char(70),char(101),
-char(101),char(100),char(98),char(97),char(99),char(107),char(0),char(109),char(95),char(97),char(112),char(112),char(108),char(105),char(101),char(100),char(73),char(109),char(112),char(117),
-char(108),char(115),char(101),char(0),char(109),char(95),char(100),char(98),char(103),char(68),char(114),char(97),char(119),char(83),char(105),char(122),char(101),char(0),char(109),char(95),
-char(100),char(105),char(115),char(97),char(98),char(108),char(101),char(67),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(115),char(66),char(101),char(116),
-char(119),char(101),char(101),char(110),char(76),char(105),char(110),char(107),char(101),char(100),char(66),char(111),char(100),char(105),char(101),char(115),char(0),char(109),char(95),char(111),
-char(118),char(101),char(114),char(114),char(105),char(100),char(101),char(78),char(117),char(109),char(83),char(111),char(108),char(118),char(101),char(114),char(73),char(116),char(101),char(114),
-char(97),char(116),char(105),char(111),char(110),char(115),char(0),char(109),char(95),char(98),char(114),char(101),char(97),char(107),char(105),char(110),char(103),char(73),char(109),char(112),
-char(117),char(108),char(115),char(101),char(84),char(104),char(114),char(101),char(115),char(104),char(111),char(108),char(100),char(0),char(109),char(95),char(105),char(115),char(69),char(110),
-char(97),char(98),char(108),char(101),char(100),char(0),char(109),char(95),char(116),char(121),char(112),char(101),char(67),char(111),char(110),char(115),char(116),char(114),char(97),char(105),
-char(110),char(116),char(68),char(97),char(116),char(97),char(0),char(109),char(95),char(112),char(105),char(118),char(111),char(116),char(73),char(110),char(65),char(0),char(109),char(95),
-char(112),char(105),char(118),char(111),char(116),char(73),char(110),char(66),char(0),char(109),char(95),char(114),char(98),char(65),char(70),char(114),char(97),char(109),char(101),char(0),
-char(109),char(95),char(114),char(98),char(66),char(70),char(114),char(97),char(109),char(101),char(0),char(109),char(95),char(117),char(115),char(101),char(82),char(101),char(102),char(101),
-char(114),char(101),char(110),char(99),char(101),char(70),char(114),char(97),char(109),char(101),char(65),char(0),char(109),char(95),char(97),char(110),char(103),char(117),char(108),char(97),
-char(114),char(79),char(110),char(108),char(121),char(0),char(109),char(95),char(101),char(110),char(97),char(98),char(108),char(101),char(65),char(110),char(103),char(117),char(108),char(97),
-char(114),char(77),char(111),char(116),char(111),char(114),char(0),char(109),char(95),char(109),char(111),char(116),char(111),char(114),char(84),char(97),char(114),char(103),char(101),char(116),
-char(86),char(101),char(108),char(111),char(99),char(105),char(116),char(121),char(0),char(109),char(95),char(109),char(97),char(120),char(77),char(111),char(116),char(111),char(114),char(73),
-char(109),char(112),char(117),char(108),char(115),char(101),char(0),char(109),char(95),char(108),char(111),char(119),char(101),char(114),char(76),char(105),char(109),char(105),char(116),char(0),
-char(109),char(95),char(117),char(112),char(112),char(101),char(114),char(76),char(105),char(109),char(105),char(116),char(0),char(109),char(95),char(108),char(105),char(109),char(105),char(116),
-char(83),char(111),char(102),char(116),char(110),char(101),char(115),char(115),char(0),char(109),char(95),char(98),char(105),char(97),char(115),char(70),char(97),char(99),char(116),char(111),
-char(114),char(0),char(109),char(95),char(114),char(101),char(108),char(97),char(120),char(97),char(116),char(105),char(111),char(110),char(70),char(97),char(99),char(116),char(111),char(114),
-char(0),char(109),char(95),char(115),char(119),char(105),char(110),char(103),char(83),char(112),char(97),char(110),char(49),char(0),char(109),char(95),char(115),char(119),char(105),char(110),
-char(103),char(83),char(112),char(97),char(110),char(50),char(0),char(109),char(95),char(116),char(119),char(105),char(115),char(116),char(83),char(112),char(97),char(110),char(0),char(109),
-char(95),char(100),char(97),char(109),char(112),char(105),char(110),char(103),char(0),char(109),char(95),char(108),char(105),char(110),char(101),char(97),char(114),char(85),char(112),char(112),
-char(101),char(114),char(76),char(105),char(109),char(105),char(116),char(0),char(109),char(95),char(108),char(105),char(110),char(101),char(97),char(114),char(76),char(111),char(119),char(101),
-char(114),char(76),char(105),char(109),char(105),char(116),char(0),char(109),char(95),char(97),char(110),char(103),char(117),char(108),char(97),char(114),char(85),char(112),char(112),char(101),
-char(114),char(76),char(105),char(109),char(105),char(116),char(0),char(109),char(95),char(97),char(110),char(103),char(117),char(108),char(97),char(114),char(76),char(111),char(119),char(101),
-char(114),char(76),char(105),char(109),char(105),char(116),char(0),char(109),char(95),char(117),char(115),char(101),char(76),char(105),char(110),char(101),char(97),char(114),char(82),char(101),
-char(102),char(101),char(114),char(101),char(110),char(99),char(101),char(70),char(114),char(97),char(109),char(101),char(65),char(0),char(109),char(95),char(117),char(115),char(101),char(79),
-char(102),char(102),char(115),char(101),char(116),char(70),char(111),char(114),char(67),char(111),char(110),char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(70),char(114),
-char(97),char(109),char(101),char(0),char(109),char(95),char(54),char(100),char(111),char(102),char(68),char(97),char(116),char(97),char(0),char(109),char(95),char(115),char(112),char(114),
-char(105),char(110),char(103),char(69),char(110),char(97),char(98),char(108),char(101),char(100),char(91),char(54),char(93),char(0),char(109),char(95),char(101),char(113),char(117),char(105),
-char(108),char(105),char(98),char(114),char(105),char(117),char(109),char(80),char(111),char(105),char(110),char(116),char(91),char(54),char(93),char(0),char(109),char(95),char(115),char(112),
-char(114),char(105),char(110),char(103),char(83),char(116),char(105),char(102),char(102),char(110),char(101),char(115),char(115),char(91),char(54),char(93),char(0),char(109),char(95),char(115),
-char(112),char(114),char(105),char(110),char(103),char(68),char(97),char(109),char(112),char(105),char(110),char(103),char(91),char(54),char(93),char(0),char(109),char(95),char(116),char(97),
-char(117),char(0),char(109),char(95),char(116),char(105),char(109),char(101),char(83),char(116),char(101),char(112),char(0),char(109),char(95),char(109),char(97),char(120),char(69),char(114),
-char(114),char(111),char(114),char(82),char(101),char(100),char(117),char(99),char(116),char(105),char(111),char(110),char(0),char(109),char(95),char(115),char(111),char(114),char(0),char(109),
-char(95),char(101),char(114),char(112),char(0),char(109),char(95),char(101),char(114),char(112),char(50),char(0),char(109),char(95),char(103),char(108),char(111),char(98),char(97),char(108),
-char(67),char(102),char(109),char(0),char(109),char(95),char(115),char(112),char(108),char(105),char(116),char(73),char(109),char(112),char(117),char(108),char(115),char(101),char(80),char(101),
-char(110),char(101),char(116),char(114),char(97),char(116),char(105),char(111),char(110),char(84),char(104),char(114),char(101),char(115),char(104),char(111),char(108),char(100),char(0),char(109),
-char(95),char(115),char(112),char(108),char(105),char(116),char(73),char(109),char(112),char(117),char(108),char(115),char(101),char(84),char(117),char(114),char(110),char(69),char(114),char(112),
-char(0),char(109),char(95),char(108),char(105),char(110),char(101),char(97),char(114),char(83),char(108),char(111),char(112),char(0),char(109),char(95),char(119),char(97),char(114),char(109),
-char(115),char(116),char(97),char(114),char(116),char(105),char(110),char(103),char(70),char(97),char(99),char(116),char(111),char(114),char(0),char(109),char(95),char(109),char(97),char(120),
-char(71),char(121),char(114),char(111),char(115),char(99),char(111),char(112),char(105),char(99),char(70),char(111),char(114),char(99),char(101),char(0),char(109),char(95),char(115),char(105),
-char(110),char(103),char(108),char(101),char(65),char(120),char(105),char(115),char(82),char(111),char(108),char(108),char(105),char(110),char(103),char(70),char(114),char(105),char(99),char(116),
-char(105),char(111),char(110),char(84),char(104),char(114),char(101),char(115),char(104),char(111),char(108),char(100),char(0),char(109),char(95),char(110),char(117),char(109),char(73),char(116),
-char(101),char(114),char(97),char(116),char(105),char(111),char(110),char(115),char(0),char(109),char(95),char(115),char(111),char(108),char(118),char(101),char(114),char(77),char(111),char(100),
-char(101),char(0),char(109),char(95),char(114),char(101),char(115),char(116),char(105),char(110),char(103),char(67),char(111),char(110),char(116),char(97),char(99),char(116),char(82),char(101),
-char(115),char(116),char(105),char(116),char(117),char(116),char(105),char(111),char(110),char(84),char(104),char(114),char(101),char(115),char(104),char(111),char(108),char(100),char(0),char(109),
-char(95),char(109),char(105),char(110),char(105),char(109),char(117),char(109),char(83),char(111),char(108),char(118),char(101),char(114),char(66),char(97),char(116),char(99),char(104),char(83),
-char(105),char(122),char(101),char(0),char(109),char(95),char(115),char(112),char(108),char(105),char(116),char(73),char(109),char(112),char(117),char(108),char(115),char(101),char(0),char(109),
-char(95),char(108),char(105),char(110),char(101),char(97),char(114),char(83),char(116),char(105),char(102),char(102),char(110),char(101),char(115),char(115),char(0),char(109),char(95),char(97),
-char(110),char(103),char(117),char(108),char(97),char(114),char(83),char(116),char(105),char(102),char(102),char(110),char(101),char(115),char(115),char(0),char(109),char(95),char(118),char(111),
-char(108),char(117),char(109),char(101),char(83),char(116),char(105),char(102),char(102),char(110),char(101),char(115),char(115),char(0),char(42),char(109),char(95),char(109),char(97),char(116),
-char(101),char(114),char(105),char(97),char(108),char(0),char(109),char(95),char(112),char(111),char(115),char(105),char(116),char(105),char(111),char(110),char(0),char(109),char(95),char(112),
-char(114),char(101),char(118),char(105),char(111),char(117),char(115),char(80),char(111),char(115),char(105),char(116),char(105),char(111),char(110),char(0),char(109),char(95),char(118),char(101),
-char(108),char(111),char(99),char(105),char(116),char(121),char(0),char(109),char(95),char(97),char(99),char(99),char(117),char(109),char(117),char(108),char(97),char(116),char(101),char(100),
-char(70),char(111),char(114),char(99),char(101),char(0),char(109),char(95),char(110),char(111),char(114),char(109),char(97),char(108),char(0),char(109),char(95),char(97),char(114),char(101),
-char(97),char(0),char(109),char(95),char(97),char(116),char(116),char(97),char(99),char(104),char(0),char(109),char(95),char(110),char(111),char(100),char(101),char(73),char(110),char(100),
-char(105),char(99),char(101),char(115),char(91),char(50),char(93),char(0),char(109),char(95),char(114),char(101),char(115),char(116),char(76),char(101),char(110),char(103),char(116),char(104),
-char(0),char(109),char(95),char(98),char(98),char(101),char(110),char(100),char(105),char(110),char(103),char(0),char(109),char(95),char(110),char(111),char(100),char(101),char(73),char(110),
-char(100),char(105),char(99),char(101),char(115),char(91),char(51),char(93),char(0),char(109),char(95),char(114),char(101),char(115),char(116),char(65),char(114),char(101),char(97),char(0),
-char(109),char(95),char(99),char(48),char(91),char(52),char(93),char(0),char(109),char(95),char(110),char(111),char(100),char(101),char(73),char(110),char(100),char(105),char(99),char(101),
-char(115),char(91),char(52),char(93),char(0),char(109),char(95),char(114),char(101),char(115),char(116),char(86),char(111),char(108),char(117),char(109),char(101),char(0),char(109),char(95),
-char(99),char(49),char(0),char(109),char(95),char(99),char(50),char(0),char(109),char(95),char(99),char(48),char(0),char(109),char(95),char(108),char(111),char(99),char(97),char(108),
-char(70),char(114),char(97),char(109),char(101),char(0),char(42),char(109),char(95),char(114),char(105),char(103),char(105),char(100),char(66),char(111),char(100),char(121),char(0),char(109),
-char(95),char(110),char(111),char(100),char(101),char(73),char(110),char(100),char(101),char(120),char(0),char(109),char(95),char(97),char(101),char(114),char(111),char(77),char(111),char(100),
-char(101),char(108),char(0),char(109),char(95),char(98),char(97),char(117),char(109),char(103),char(97),char(114),char(116),char(101),char(0),char(109),char(95),char(100),char(114),char(97),
-char(103),char(0),char(109),char(95),char(108),char(105),char(102),char(116),char(0),char(109),char(95),char(112),char(114),char(101),char(115),char(115),char(117),char(114),char(101),char(0),
-char(109),char(95),char(118),char(111),char(108),char(117),char(109),char(101),char(0),char(109),char(95),char(100),char(121),char(110),char(97),char(109),char(105),char(99),char(70),char(114),
-char(105),char(99),char(116),char(105),char(111),char(110),char(0),char(109),char(95),char(112),char(111),char(115),char(101),char(77),char(97),char(116),char(99),char(104),char(0),char(109),
-char(95),char(114),char(105),char(103),char(105),char(100),char(67),char(111),char(110),char(116),char(97),char(99),char(116),char(72),char(97),char(114),char(100),char(110),char(101),char(115),
-char(115),char(0),char(109),char(95),char(107),char(105),char(110),char(101),char(116),char(105),char(99),char(67),char(111),char(110),char(116),char(97),char(99),char(116),char(72),char(97),
-char(114),char(100),char(110),char(101),char(115),char(115),char(0),char(109),char(95),char(115),char(111),char(102),char(116),char(67),char(111),char(110),char(116),char(97),char(99),char(116),
-char(72),char(97),char(114),char(100),char(110),char(101),char(115),char(115),char(0),char(109),char(95),char(97),char(110),char(99),char(104),char(111),char(114),char(72),char(97),char(114),
-char(100),char(110),char(101),char(115),char(115),char(0),char(109),char(95),char(115),char(111),char(102),char(116),char(82),char(105),char(103),char(105),char(100),char(67),char(108),char(117),
-char(115),char(116),char(101),char(114),char(72),char(97),char(114),char(100),char(110),char(101),char(115),char(115),char(0),char(109),char(95),char(115),char(111),char(102),char(116),char(75),
-char(105),char(110),char(101),char(116),char(105),char(99),char(67),char(108),char(117),char(115),char(116),char(101),char(114),char(72),char(97),char(114),char(100),char(110),char(101),char(115),
-char(115),char(0),char(109),char(95),char(115),char(111),char(102),char(116),char(83),char(111),char(102),char(116),char(67),char(108),char(117),char(115),char(116),char(101),char(114),char(72),
-char(97),char(114),char(100),char(110),char(101),char(115),char(115),char(0),char(109),char(95),char(115),char(111),char(102),char(116),char(82),char(105),char(103),char(105),char(100),char(67),
-char(108),char(117),char(115),char(116),char(101),char(114),char(73),char(109),char(112),char(117),char(108),char(115),char(101),char(83),char(112),char(108),char(105),char(116),char(0),char(109),
-char(95),char(115),char(111),char(102),char(116),char(75),char(105),char(110),char(101),char(116),char(105),char(99),char(67),char(108),char(117),char(115),char(116),char(101),char(114),char(73),
-char(109),char(112),char(117),char(108),char(115),char(101),char(83),char(112),char(108),char(105),char(116),char(0),char(109),char(95),char(115),char(111),char(102),char(116),char(83),char(111),
-char(102),char(116),char(67),char(108),char(117),char(115),char(116),char(101),char(114),char(73),char(109),char(112),char(117),char(108),char(115),char(101),char(83),char(112),char(108),char(105),
-char(116),char(0),char(109),char(95),char(109),char(97),char(120),char(86),char(111),char(108),char(117),char(109),char(101),char(0),char(109),char(95),char(116),char(105),char(109),char(101),
-char(83),char(99),char(97),char(108),char(101),char(0),char(109),char(95),char(118),char(101),char(108),char(111),char(99),char(105),char(116),char(121),char(73),char(116),char(101),char(114),
-char(97),char(116),char(105),char(111),char(110),char(115),char(0),char(109),char(95),char(112),char(111),char(115),char(105),char(116),char(105),char(111),char(110),char(73),char(116),char(101),
-char(114),char(97),char(116),char(105),char(111),char(110),char(115),char(0),char(109),char(95),char(100),char(114),char(105),char(102),char(116),char(73),char(116),char(101),char(114),char(97),
-char(116),char(105),char(111),char(110),char(115),char(0),char(109),char(95),char(99),char(108),char(117),char(115),char(116),char(101),char(114),char(73),char(116),char(101),char(114),char(97),
-char(116),char(105),char(111),char(110),char(115),char(0),char(109),char(95),char(114),char(111),char(116),char(0),char(109),char(95),char(115),char(99),char(97),char(108),char(101),char(0),
-char(109),char(95),char(97),char(113),char(113),char(0),char(109),char(95),char(99),char(111),char(109),char(0),char(42),char(109),char(95),char(112),char(111),char(115),char(105),char(116),
-char(105),char(111),char(110),char(115),char(0),char(42),char(109),char(95),char(119),char(101),char(105),char(103),char(104),char(116),char(115),char(0),char(109),char(95),char(110),char(117),
-char(109),char(80),char(111),char(115),char(105),char(116),char(105),char(111),char(110),char(115),char(0),char(109),char(95),char(110),char(117),char(109),char(87),char(101),char(105),char(103),
-char(116),char(115),char(0),char(109),char(95),char(98),char(118),char(111),char(108),char(117),char(109),char(101),char(0),char(109),char(95),char(98),char(102),char(114),char(97),char(109),
-char(101),char(0),char(109),char(95),char(102),char(114),char(97),char(109),char(101),char(120),char(102),char(111),char(114),char(109),char(0),char(109),char(95),char(108),char(111),char(99),
-char(105),char(105),char(0),char(109),char(95),char(105),char(110),char(118),char(119),char(105),char(0),char(109),char(95),char(118),char(105),char(109),char(112),char(117),char(108),char(115),
-char(101),char(115),char(91),char(50),char(93),char(0),char(109),char(95),char(100),char(105),char(109),char(112),char(117),char(108),char(115),char(101),char(115),char(91),char(50),char(93),
-char(0),char(109),char(95),char(108),char(118),char(0),char(109),char(95),char(97),char(118),char(0),char(42),char(109),char(95),char(102),char(114),char(97),char(109),char(101),char(114),
-char(101),char(102),char(115),char(0),char(42),char(109),char(95),char(110),char(111),char(100),char(101),char(73),char(110),char(100),char(105),char(99),char(101),char(115),char(0),char(42),
-char(109),char(95),char(109),char(97),char(115),char(115),char(101),char(115),char(0),char(109),char(95),char(110),char(117),char(109),char(70),char(114),char(97),char(109),char(101),char(82),
-char(101),char(102),char(115),char(0),char(109),char(95),char(110),char(117),char(109),char(78),char(111),char(100),char(101),char(115),char(0),char(109),char(95),char(110),char(117),char(109),
-char(77),char(97),char(115),char(115),char(101),char(115),char(0),char(109),char(95),char(105),char(100),char(109),char(97),char(115),char(115),char(0),char(109),char(95),char(105),char(109),
-char(97),char(115),char(115),char(0),char(109),char(95),char(110),char(118),char(105),char(109),char(112),char(117),char(108),char(115),char(101),char(115),char(0),char(109),char(95),char(110),
-char(100),char(105),char(109),char(112),char(117),char(108),char(115),char(101),char(115),char(0),char(109),char(95),char(110),char(100),char(97),char(109),char(112),char(105),char(110),char(103),
-char(0),char(109),char(95),char(108),char(100),char(97),char(109),char(112),char(105),char(110),char(103),char(0),char(109),char(95),char(97),char(100),char(97),char(109),char(112),char(105),
-char(110),char(103),char(0),char(109),char(95),char(109),char(97),char(116),char(99),char(104),char(105),char(110),char(103),char(0),char(109),char(95),char(109),char(97),char(120),char(83),
-char(101),char(108),char(102),char(67),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(73),char(109),char(112),char(117),char(108),char(115),char(101),char(0),
-char(109),char(95),char(115),char(101),char(108),char(102),char(67),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(73),char(109),char(112),char(117),char(108),
-char(115),char(101),char(70),char(97),char(99),char(116),char(111),char(114),char(0),char(109),char(95),char(99),char(111),char(110),char(116),char(97),char(105),char(110),char(115),char(65),
-char(110),char(99),char(104),char(111),char(114),char(0),char(109),char(95),char(99),char(111),char(108),char(108),char(105),char(100),char(101),char(0),char(109),char(95),char(99),char(108),
-char(117),char(115),char(116),char(101),char(114),char(73),char(110),char(100),char(101),char(120),char(0),char(42),char(109),char(95),char(98),char(111),char(100),char(121),char(65),char(0),
-char(42),char(109),char(95),char(98),char(111),char(100),char(121),char(66),char(0),char(109),char(95),char(114),char(101),char(102),char(115),char(91),char(50),char(93),char(0),char(109),
-char(95),char(99),char(102),char(109),char(0),char(109),char(95),char(115),char(112),char(108),char(105),char(116),char(0),char(109),char(95),char(100),char(101),char(108),char(101),char(116),
-char(101),char(0),char(109),char(95),char(114),char(101),char(108),char(80),char(111),char(115),char(105),char(116),char(105),char(111),char(110),char(91),char(50),char(93),char(0),char(109),
-char(95),char(98),char(111),char(100),char(121),char(65),char(116),char(121),char(112),char(101),char(0),char(109),char(95),char(98),char(111),char(100),char(121),char(66),char(116),char(121),
-char(112),char(101),char(0),char(109),char(95),char(106),char(111),char(105),char(110),char(116),char(84),char(121),char(112),char(101),char(0),char(42),char(109),char(95),char(112),char(111),
-char(115),char(101),char(0),char(42),char(42),char(109),char(95),char(109),char(97),char(116),char(101),char(114),char(105),char(97),char(108),char(115),char(0),char(42),char(109),char(95),
-char(110),char(111),char(100),char(101),char(115),char(0),char(42),char(109),char(95),char(108),char(105),char(110),char(107),char(115),char(0),char(42),char(109),char(95),char(102),char(97),
-char(99),char(101),char(115),char(0),char(42),char(109),char(95),char(116),char(101),char(116),char(114),char(97),char(104),char(101),char(100),char(114),char(97),char(0),char(42),char(109),
-char(95),char(97),char(110),char(99),char(104),char(111),char(114),char(115),char(0),char(42),char(109),char(95),char(99),char(108),char(117),char(115),char(116),char(101),char(114),char(115),
-char(0),char(42),char(109),char(95),char(106),char(111),char(105),char(110),char(116),char(115),char(0),char(109),char(95),char(110),char(117),char(109),char(77),char(97),char(116),char(101),
-char(114),char(105),char(97),char(108),char(115),char(0),char(109),char(95),char(110),char(117),char(109),char(76),char(105),char(110),char(107),char(115),char(0),char(109),char(95),char(110),
-char(117),char(109),char(70),char(97),char(99),char(101),char(115),char(0),char(109),char(95),char(110),char(117),char(109),char(84),char(101),char(116),char(114),char(97),char(104),char(101),
-char(100),char(114),char(97),char(0),char(109),char(95),char(110),char(117),char(109),char(65),char(110),char(99),char(104),char(111),char(114),char(115),char(0),char(109),char(95),char(110),
-char(117),char(109),char(67),char(108),char(117),char(115),char(116),char(101),char(114),char(115),char(0),char(109),char(95),char(110),char(117),char(109),char(74),char(111),char(105),char(110),
-char(116),char(115),char(0),char(109),char(95),char(99),char(111),char(110),char(102),char(105),char(103),char(0),char(84),char(89),char(80),char(69),char(76),char(0),char(0),char(0),
-char(99),char(104),char(97),char(114),char(0),char(117),char(99),char(104),char(97),char(114),char(0),char(115),char(104),char(111),char(114),char(116),char(0),char(117),char(115),char(104),
-char(111),char(114),char(116),char(0),char(105),char(110),char(116),char(0),char(108),char(111),char(110),char(103),char(0),char(117),char(108),char(111),char(110),char(103),char(0),char(102),
-char(108),char(111),char(97),char(116),char(0),char(100),char(111),char(117),char(98),char(108),char(101),char(0),char(118),char(111),char(105),char(100),char(0),char(80),char(111),char(105),
-char(110),char(116),char(101),char(114),char(65),char(114),char(114),char(97),char(121),char(0),char(98),char(116),char(80),char(104),char(121),char(115),char(105),char(99),char(115),char(83),
-char(121),char(115),char(116),char(101),char(109),char(0),char(76),char(105),char(115),char(116),char(66),char(97),char(115),char(101),char(0),char(98),char(116),char(86),char(101),char(99),
-char(116),char(111),char(114),char(51),char(70),char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(86),char(101),char(99),char(116),
-char(111),char(114),char(51),char(68),char(111),char(117),char(98),char(108),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(77),char(97),char(116),char(114),
-char(105),char(120),char(51),char(120),char(51),char(70),char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(77),char(97),char(116),
-char(114),char(105),char(120),char(51),char(120),char(51),char(68),char(111),char(117),char(98),char(108),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(84),
-char(114),char(97),char(110),char(115),char(102),char(111),char(114),char(109),char(70),char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),
-char(84),char(114),char(97),char(110),char(115),char(102),char(111),char(114),char(109),char(68),char(111),char(117),char(98),char(108),char(101),char(68),char(97),char(116),char(97),char(0),
-char(98),char(116),char(66),char(118),char(104),char(83),char(117),char(98),char(116),char(114),char(101),char(101),char(73),char(110),char(102),char(111),char(68),char(97),char(116),char(97),
-char(0),char(98),char(116),char(79),char(112),char(116),char(105),char(109),char(105),char(122),char(101),char(100),char(66),char(118),char(104),char(78),char(111),char(100),char(101),char(70),
-char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(79),char(112),char(116),char(105),char(109),char(105),char(122),char(101),char(100),
-char(66),char(118),char(104),char(78),char(111),char(100),char(101),char(68),char(111),char(117),char(98),char(108),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),
-char(81),char(117),char(97),char(110),char(116),char(105),char(122),char(101),char(100),char(66),char(118),char(104),char(78),char(111),char(100),char(101),char(68),char(97),char(116),char(97),
-char(0),char(98),char(116),char(81),char(117),char(97),char(110),char(116),char(105),char(122),char(101),char(100),char(66),char(118),char(104),char(70),char(108),char(111),char(97),char(116),
-char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(81),char(117),char(97),char(110),char(116),char(105),char(122),char(101),char(100),char(66),char(118),char(104),char(68),
-char(111),char(117),char(98),char(108),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(111),char(108),char(108),char(105),char(115),char(105),char(111),
-char(110),char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(83),char(116),char(97),char(116),char(105),char(99),char(80),
-char(108),char(97),char(110),char(101),char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(111),char(110),char(118),
-char(101),char(120),char(73),char(110),char(116),char(101),char(114),char(110),char(97),char(108),char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),
-char(98),char(116),char(80),char(111),char(115),char(105),char(116),char(105),char(111),char(110),char(65),char(110),char(100),char(82),char(97),char(100),char(105),char(117),char(115),char(0),
-char(98),char(116),char(77),char(117),char(108),char(116),char(105),char(83),char(112),char(104),char(101),char(114),char(101),char(83),char(104),char(97),char(112),char(101),char(68),char(97),
-char(116),char(97),char(0),char(98),char(116),char(73),char(110),char(116),char(73),char(110),char(100),char(101),char(120),char(68),char(97),char(116),char(97),char(0),char(98),char(116),
-char(83),char(104),char(111),char(114),char(116),char(73),char(110),char(116),char(73),char(110),char(100),char(101),char(120),char(68),char(97),char(116),char(97),char(0),char(98),char(116),
-char(83),char(104),char(111),char(114),char(116),char(73),char(110),char(116),char(73),char(110),char(100),char(101),char(120),char(84),char(114),char(105),char(112),char(108),char(101),char(116),
-char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(104),char(97),char(114),char(73),char(110),char(100),char(101),char(120),char(84),char(114),char(105),char(112),
-char(108),char(101),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(77),char(101),char(115),char(104),char(80),char(97),char(114),char(116),char(68),char(97),
-char(116),char(97),char(0),char(98),char(116),char(83),char(116),char(114),char(105),char(100),char(105),char(110),char(103),char(77),char(101),char(115),char(104),char(73),char(110),char(116),
-char(101),char(114),char(102),char(97),char(99),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(84),char(114),char(105),char(97),char(110),char(103),char(108),
-char(101),char(77),char(101),char(115),char(104),char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(84),char(114),char(105),
-char(97),char(110),char(103),char(108),char(101),char(73),char(110),char(102),char(111),char(77),char(97),char(112),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(83),
-char(99),char(97),char(108),char(101),char(100),char(84),char(114),char(105),char(97),char(110),char(103),char(108),char(101),char(77),char(101),char(115),char(104),char(83),char(104),char(97),
-char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(111),char(109),char(112),char(111),char(117),char(110),char(100),char(83),char(104),char(97),
-char(112),char(101),char(67),char(104),char(105),char(108),char(100),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(111),char(109),char(112),char(111),char(117),
-char(110),char(100),char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(121),char(108),char(105),char(110),char(100),
-char(101),char(114),char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(97),char(112),char(115),char(117),char(108),
-char(101),char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(84),char(114),char(105),char(97),char(110),char(103),char(108),
-char(101),char(73),char(110),char(102),char(111),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(71),char(73),char(109),char(112),char(97),char(99),char(116),char(77),
-char(101),char(115),char(104),char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(111),char(110),char(118),char(101),
-char(120),char(72),char(117),char(108),char(108),char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(111),char(108),
-char(108),char(105),char(115),char(105),char(111),char(110),char(79),char(98),char(106),char(101),char(99),char(116),char(68),char(111),char(117),char(98),char(108),char(101),char(68),char(97),
-char(116),char(97),char(0),char(98),char(116),char(67),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(79),char(98),char(106),char(101),char(99),char(116),
-char(70),char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(68),char(121),char(110),char(97),char(109),char(105),char(99),char(115),
-char(87),char(111),char(114),char(108),char(100),char(68),char(111),char(117),char(98),char(108),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(111),
-char(110),char(116),char(97),char(99),char(116),char(83),char(111),char(108),char(118),char(101),char(114),char(73),char(110),char(102),char(111),char(68),char(111),char(117),char(98),char(108),
-char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(68),char(121),char(110),char(97),char(109),char(105),char(99),char(115),char(87),char(111),char(114),char(108),
-char(100),char(70),char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(111),char(110),char(116),char(97),char(99),char(116),
-char(83),char(111),char(108),char(118),char(101),char(114),char(73),char(110),char(102),char(111),char(70),char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),char(0),
-char(98),char(116),char(82),char(105),char(103),char(105),char(100),char(66),char(111),char(100),char(121),char(70),char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),
-char(0),char(98),char(116),char(82),char(105),char(103),char(105),char(100),char(66),char(111),char(100),char(121),char(68),char(111),char(117),char(98),char(108),char(101),char(68),char(97),
-char(116),char(97),char(0),char(98),char(116),char(67),char(111),char(110),char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(73),char(110),char(102),char(111),char(49),
-char(0),char(98),char(116),char(84),char(121),char(112),char(101),char(100),char(67),char(111),char(110),char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(68),char(97),
-char(116),char(97),char(0),char(98),char(116),char(82),char(105),char(103),char(105),char(100),char(66),char(111),char(100),char(121),char(68),char(97),char(116),char(97),char(0),char(98),
-char(116),char(80),char(111),char(105),char(110),char(116),char(50),char(80),char(111),char(105),char(110),char(116),char(67),char(111),char(110),char(115),char(116),char(114),char(97),char(105),
-char(110),char(116),char(70),char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(80),char(111),char(105),char(110),char(116),char(50),
-char(80),char(111),char(105),char(110),char(116),char(67),char(111),char(110),char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(68),char(111),char(117),char(98),char(108),
-char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(72),char(105),char(110),char(103),char(101),char(67),char(111),char(110),char(115),char(116),char(114),char(97),
-char(105),char(110),char(116),char(68),char(111),char(117),char(98),char(108),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(72),char(105),char(110),char(103),
-char(101),char(67),char(111),char(110),char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(70),char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),
-char(0),char(98),char(116),char(67),char(111),char(110),char(101),char(84),char(119),char(105),char(115),char(116),char(67),char(111),char(110),char(115),char(116),char(114),char(97),char(105),
-char(110),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(71),char(101),char(110),char(101),char(114),char(105),char(99),char(54),char(68),char(111),char(102),
-char(67),char(111),char(110),char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(71),char(101),char(110),
-char(101),char(114),char(105),char(99),char(54),char(68),char(111),char(102),char(83),char(112),char(114),char(105),char(110),char(103),char(67),char(111),char(110),char(115),char(116),char(114),
-char(97),char(105),char(110),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(83),char(108),char(105),char(100),char(101),char(114),char(67),char(111),char(110),
-char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(68),char(97),char(116),char(97),char(0),char(83),char(111),char(102),char(116),char(66),char(111),char(100),char(121),
-char(77),char(97),char(116),char(101),char(114),char(105),char(97),char(108),char(68),char(97),char(116),char(97),char(0),char(83),char(111),char(102),char(116),char(66),char(111),char(100),
-char(121),char(78),char(111),char(100),char(101),char(68),char(97),char(116),char(97),char(0),char(83),char(111),char(102),char(116),char(66),char(111),char(100),char(121),char(76),char(105),
-char(110),char(107),char(68),char(97),char(116),char(97),char(0),char(83),char(111),char(102),char(116),char(66),char(111),char(100),char(121),char(70),char(97),char(99),char(101),char(68),
-char(97),char(116),char(97),char(0),char(83),char(111),char(102),char(116),char(66),char(111),char(100),char(121),char(84),char(101),char(116),char(114),char(97),char(68),char(97),char(116),
-char(97),char(0),char(83),char(111),char(102),char(116),char(82),char(105),char(103),char(105),char(100),char(65),char(110),char(99),char(104),char(111),char(114),char(68),char(97),char(116),
-char(97),char(0),char(83),char(111),char(102),char(116),char(66),char(111),char(100),char(121),char(67),char(111),char(110),char(102),char(105),char(103),char(68),char(97),char(116),char(97),
-char(0),char(83),char(111),char(102),char(116),char(66),char(111),char(100),char(121),char(80),char(111),char(115),char(101),char(68),char(97),char(116),char(97),char(0),char(83),char(111),
-char(102),char(116),char(66),char(111),char(100),char(121),char(67),char(108),char(117),char(115),char(116),char(101),char(114),char(68),char(97),char(116),char(97),char(0),char(98),char(116),
-char(83),char(111),char(102),char(116),char(66),char(111),char(100),char(121),char(74),char(111),char(105),char(110),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),
-char(83),char(111),char(102),char(116),char(66),char(111),char(100),char(121),char(70),char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),char(0),char(0),char(0),
-char(84),char(76),char(69),char(78),char(1),char(0),char(1),char(0),char(2),char(0),char(2),char(0),char(4),char(0),char(4),char(0),char(4),char(0),char(4),char(0),
-char(8),char(0),char(0),char(0),char(12),char(0),char(36),char(0),char(8),char(0),char(16),char(0),char(32),char(0),char(48),char(0),char(96),char(0),char(64),char(0),
-char(-128),char(0),char(20),char(0),char(48),char(0),char(80),char(0),char(16),char(0),char(84),char(0),char(-124),char(0),char(12),char(0),char(52),char(0),char(52),char(0),
-char(20),char(0),char(64),char(0),char(4),char(0),char(4),char(0),char(8),char(0),char(4),char(0),char(32),char(0),char(28),char(0),char(60),char(0),char(56),char(0),
-char(76),char(0),char(76),char(0),char(24),char(0),char(60),char(0),char(60),char(0),char(16),char(0),char(64),char(0),char(68),char(0),char(-48),char(1),char(0),char(1),
-char(-72),char(0),char(-104),char(0),char(104),char(0),char(88),char(0),char(-24),char(1),char(-96),char(3),char(8),char(0),char(52),char(0),char(0),char(0),char(84),char(0),
-char(116),char(0),char(92),char(1),char(-36),char(0),char(-44),char(0),char(-4),char(0),char(92),char(1),char(-52),char(0),char(16),char(0),char(100),char(0),char(20),char(0),
-char(36),char(0),char(100),char(0),char(92),char(0),char(104),char(0),char(-64),char(0),char(92),char(1),char(104),char(0),char(-84),char(1),char(83),char(84),char(82),char(67),
-char(65),char(0),char(0),char(0),char(10),char(0),char(3),char(0),char(4),char(0),char(0),char(0),char(4),char(0),char(1),char(0),char(9),char(0),char(2),char(0),
-char(11),char(0),char(3),char(0),char(10),char(0),char(3),char(0),char(10),char(0),char(4),char(0),char(10),char(0),char(5),char(0),char(12),char(0),char(2),char(0),
-char(9),char(0),char(6),char(0),char(9),char(0),char(7),char(0),char(13),char(0),char(1),char(0),char(7),char(0),char(8),char(0),char(14),char(0),char(1),char(0),
-char(8),char(0),char(8),char(0),char(15),char(0),char(1),char(0),char(13),char(0),char(9),char(0),char(16),char(0),char(1),char(0),char(14),char(0),char(9),char(0),
-char(17),char(0),char(2),char(0),char(15),char(0),char(10),char(0),char(13),char(0),char(11),char(0),char(18),char(0),char(2),char(0),char(16),char(0),char(10),char(0),
-char(14),char(0),char(11),char(0),char(19),char(0),char(4),char(0),char(4),char(0),char(12),char(0),char(4),char(0),char(13),char(0),char(2),char(0),char(14),char(0),
-char(2),char(0),char(15),char(0),char(20),char(0),char(6),char(0),char(13),char(0),char(16),char(0),char(13),char(0),char(17),char(0),char(4),char(0),char(18),char(0),
-char(4),char(0),char(19),char(0),char(4),char(0),char(20),char(0),char(0),char(0),char(21),char(0),char(21),char(0),char(6),char(0),char(14),char(0),char(16),char(0),
-char(14),char(0),char(17),char(0),char(4),char(0),char(18),char(0),char(4),char(0),char(19),char(0),char(4),char(0),char(20),char(0),char(0),char(0),char(21),char(0),
-char(22),char(0),char(3),char(0),char(2),char(0),char(14),char(0),char(2),char(0),char(15),char(0),char(4),char(0),char(22),char(0),char(23),char(0),char(12),char(0),
-char(13),char(0),char(23),char(0),char(13),char(0),char(24),char(0),char(13),char(0),char(25),char(0),char(4),char(0),char(26),char(0),char(4),char(0),char(27),char(0),
-char(4),char(0),char(28),char(0),char(4),char(0),char(29),char(0),char(20),char(0),char(30),char(0),char(22),char(0),char(31),char(0),char(19),char(0),char(32),char(0),
-char(4),char(0),char(33),char(0),char(4),char(0),char(34),char(0),char(24),char(0),char(12),char(0),char(14),char(0),char(23),char(0),char(14),char(0),char(24),char(0),
-char(14),char(0),char(25),char(0),char(4),char(0),char(26),char(0),char(4),char(0),char(27),char(0),char(4),char(0),char(28),char(0),char(4),char(0),char(29),char(0),
-char(21),char(0),char(30),char(0),char(22),char(0),char(31),char(0),char(4),char(0),char(33),char(0),char(4),char(0),char(34),char(0),char(19),char(0),char(32),char(0),
-char(25),char(0),char(3),char(0),char(0),char(0),char(35),char(0),char(4),char(0),char(36),char(0),char(0),char(0),char(37),char(0),char(26),char(0),char(5),char(0),
-char(25),char(0),char(38),char(0),char(13),char(0),char(39),char(0),char(13),char(0),char(40),char(0),char(7),char(0),char(41),char(0),char(0),char(0),char(21),char(0),
-char(27),char(0),char(5),char(0),char(25),char(0),char(38),char(0),char(13),char(0),char(39),char(0),char(13),char(0),char(42),char(0),char(7),char(0),char(43),char(0),
-char(4),char(0),char(44),char(0),char(28),char(0),char(2),char(0),char(13),char(0),char(45),char(0),char(7),char(0),char(46),char(0),char(29),char(0),char(4),char(0),
-char(27),char(0),char(47),char(0),char(28),char(0),char(48),char(0),char(4),char(0),char(49),char(0),char(0),char(0),char(37),char(0),char(30),char(0),char(1),char(0),
-char(4),char(0),char(50),char(0),char(31),char(0),char(2),char(0),char(2),char(0),char(50),char(0),char(0),char(0),char(51),char(0),char(32),char(0),char(2),char(0),
-char(2),char(0),char(52),char(0),char(0),char(0),char(51),char(0),char(33),char(0),char(2),char(0),char(0),char(0),char(52),char(0),char(0),char(0),char(53),char(0),
-char(34),char(0),char(8),char(0),char(13),char(0),char(54),char(0),char(14),char(0),char(55),char(0),char(30),char(0),char(56),char(0),char(32),char(0),char(57),char(0),
-char(33),char(0),char(58),char(0),char(31),char(0),char(59),char(0),char(4),char(0),char(60),char(0),char(4),char(0),char(61),char(0),char(35),char(0),char(4),char(0),
-char(34),char(0),char(62),char(0),char(13),char(0),char(63),char(0),char(4),char(0),char(64),char(0),char(0),char(0),char(37),char(0),char(36),char(0),char(7),char(0),
-char(25),char(0),char(38),char(0),char(35),char(0),char(65),char(0),char(23),char(0),char(66),char(0),char(24),char(0),char(67),char(0),char(37),char(0),char(68),char(0),
-char(7),char(0),char(43),char(0),char(0),char(0),char(69),char(0),char(38),char(0),char(2),char(0),char(36),char(0),char(70),char(0),char(13),char(0),char(39),char(0),
-char(39),char(0),char(4),char(0),char(17),char(0),char(71),char(0),char(25),char(0),char(72),char(0),char(4),char(0),char(73),char(0),char(7),char(0),char(74),char(0),
-char(40),char(0),char(4),char(0),char(25),char(0),char(38),char(0),char(39),char(0),char(75),char(0),char(4),char(0),char(76),char(0),char(7),char(0),char(43),char(0),
-char(41),char(0),char(3),char(0),char(27),char(0),char(47),char(0),char(4),char(0),char(77),char(0),char(0),char(0),char(37),char(0),char(42),char(0),char(3),char(0),
-char(27),char(0),char(47),char(0),char(4),char(0),char(77),char(0),char(0),char(0),char(37),char(0),char(43),char(0),char(4),char(0),char(4),char(0),char(78),char(0),
-char(7),char(0),char(79),char(0),char(7),char(0),char(80),char(0),char(7),char(0),char(81),char(0),char(37),char(0),char(14),char(0),char(4),char(0),char(82),char(0),
-char(4),char(0),char(83),char(0),char(43),char(0),char(84),char(0),char(4),char(0),char(85),char(0),char(7),char(0),char(86),char(0),char(7),char(0),char(87),char(0),
-char(7),char(0),char(88),char(0),char(7),char(0),char(89),char(0),char(7),char(0),char(90),char(0),char(4),char(0),char(91),char(0),char(4),char(0),char(92),char(0),
-char(4),char(0),char(93),char(0),char(4),char(0),char(94),char(0),char(0),char(0),char(37),char(0),char(44),char(0),char(5),char(0),char(25),char(0),char(38),char(0),
-char(35),char(0),char(65),char(0),char(13),char(0),char(39),char(0),char(7),char(0),char(43),char(0),char(4),char(0),char(95),char(0),char(45),char(0),char(5),char(0),
-char(27),char(0),char(47),char(0),char(13),char(0),char(96),char(0),char(14),char(0),char(97),char(0),char(4),char(0),char(98),char(0),char(0),char(0),char(99),char(0),
-char(46),char(0),char(25),char(0),char(9),char(0),char(100),char(0),char(9),char(0),char(101),char(0),char(25),char(0),char(102),char(0),char(0),char(0),char(35),char(0),
-char(18),char(0),char(103),char(0),char(18),char(0),char(104),char(0),char(14),char(0),char(105),char(0),char(14),char(0),char(106),char(0),char(14),char(0),char(107),char(0),
-char(8),char(0),char(108),char(0),char(8),char(0),char(109),char(0),char(8),char(0),char(110),char(0),char(8),char(0),char(111),char(0),char(8),char(0),char(112),char(0),
-char(8),char(0),char(113),char(0),char(8),char(0),char(114),char(0),char(8),char(0),char(115),char(0),char(4),char(0),char(116),char(0),char(4),char(0),char(117),char(0),
-char(4),char(0),char(118),char(0),char(4),char(0),char(119),char(0),char(4),char(0),char(120),char(0),char(4),char(0),char(121),char(0),char(4),char(0),char(122),char(0),
-char(0),char(0),char(37),char(0),char(47),char(0),char(25),char(0),char(9),char(0),char(100),char(0),char(9),char(0),char(101),char(0),char(25),char(0),char(102),char(0),
-char(0),char(0),char(35),char(0),char(17),char(0),char(103),char(0),char(17),char(0),char(104),char(0),char(13),char(0),char(105),char(0),char(13),char(0),char(106),char(0),
-char(13),char(0),char(107),char(0),char(7),char(0),char(108),char(0),char(7),char(0),char(109),char(0),char(7),char(0),char(110),char(0),char(7),char(0),char(111),char(0),
-char(7),char(0),char(112),char(0),char(7),char(0),char(113),char(0),char(7),char(0),char(114),char(0),char(7),char(0),char(115),char(0),char(4),char(0),char(116),char(0),
-char(4),char(0),char(117),char(0),char(4),char(0),char(118),char(0),char(4),char(0),char(119),char(0),char(4),char(0),char(120),char(0),char(4),char(0),char(121),char(0),
-char(4),char(0),char(122),char(0),char(0),char(0),char(37),char(0),char(48),char(0),char(2),char(0),char(49),char(0),char(123),char(0),char(14),char(0),char(124),char(0),
-char(50),char(0),char(2),char(0),char(51),char(0),char(123),char(0),char(13),char(0),char(124),char(0),char(52),char(0),char(21),char(0),char(47),char(0),char(125),char(0),
-char(15),char(0),char(126),char(0),char(13),char(0),char(127),char(0),char(13),char(0),char(-128),char(0),char(13),char(0),char(-127),char(0),char(13),char(0),char(-126),char(0),
-char(13),char(0),char(124),char(0),char(13),char(0),char(-125),char(0),char(13),char(0),char(-124),char(0),char(13),char(0),char(-123),char(0),char(13),char(0),char(-122),char(0),
-char(7),char(0),char(-121),char(0),char(7),char(0),char(-120),char(0),char(7),char(0),char(-119),char(0),char(7),char(0),char(-118),char(0),char(7),char(0),char(-117),char(0),
-char(7),char(0),char(-116),char(0),char(7),char(0),char(-115),char(0),char(7),char(0),char(-114),char(0),char(7),char(0),char(-113),char(0),char(4),char(0),char(-112),char(0),
-char(53),char(0),char(22),char(0),char(46),char(0),char(125),char(0),char(16),char(0),char(126),char(0),char(14),char(0),char(127),char(0),char(14),char(0),char(-128),char(0),
-char(14),char(0),char(-127),char(0),char(14),char(0),char(-126),char(0),char(14),char(0),char(124),char(0),char(14),char(0),char(-125),char(0),char(14),char(0),char(-124),char(0),
-char(14),char(0),char(-123),char(0),char(14),char(0),char(-122),char(0),char(8),char(0),char(-121),char(0),char(8),char(0),char(-120),char(0),char(8),char(0),char(-119),char(0),
-char(8),char(0),char(-118),char(0),char(8),char(0),char(-117),char(0),char(8),char(0),char(-116),char(0),char(8),char(0),char(-115),char(0),char(8),char(0),char(-114),char(0),
-char(8),char(0),char(-113),char(0),char(4),char(0),char(-112),char(0),char(0),char(0),char(37),char(0),char(54),char(0),char(2),char(0),char(4),char(0),char(-111),char(0),
-char(4),char(0),char(-110),char(0),char(55),char(0),char(13),char(0),char(56),char(0),char(-109),char(0),char(56),char(0),char(-108),char(0),char(0),char(0),char(35),char(0),
-char(4),char(0),char(-107),char(0),char(4),char(0),char(-106),char(0),char(4),char(0),char(-105),char(0),char(4),char(0),char(-104),char(0),char(7),char(0),char(-103),char(0),
-char(7),char(0),char(-102),char(0),char(4),char(0),char(-101),char(0),char(4),char(0),char(-100),char(0),char(7),char(0),char(-99),char(0),char(4),char(0),char(-98),char(0),
-char(57),char(0),char(3),char(0),char(55),char(0),char(-97),char(0),char(13),char(0),char(-96),char(0),char(13),char(0),char(-95),char(0),char(58),char(0),char(3),char(0),
-char(55),char(0),char(-97),char(0),char(14),char(0),char(-96),char(0),char(14),char(0),char(-95),char(0),char(59),char(0),char(13),char(0),char(55),char(0),char(-97),char(0),
-char(18),char(0),char(-94),char(0),char(18),char(0),char(-93),char(0),char(4),char(0),char(-92),char(0),char(4),char(0),char(-91),char(0),char(4),char(0),char(-90),char(0),
-char(7),char(0),char(-89),char(0),char(7),char(0),char(-88),char(0),char(7),char(0),char(-87),char(0),char(7),char(0),char(-86),char(0),char(7),char(0),char(-85),char(0),
-char(7),char(0),char(-84),char(0),char(7),char(0),char(-83),char(0),char(60),char(0),char(13),char(0),char(55),char(0),char(-97),char(0),char(17),char(0),char(-94),char(0),
-char(17),char(0),char(-93),char(0),char(4),char(0),char(-92),char(0),char(4),char(0),char(-91),char(0),char(4),char(0),char(-90),char(0),char(7),char(0),char(-89),char(0),
-char(7),char(0),char(-88),char(0),char(7),char(0),char(-87),char(0),char(7),char(0),char(-86),char(0),char(7),char(0),char(-85),char(0),char(7),char(0),char(-84),char(0),
-char(7),char(0),char(-83),char(0),char(61),char(0),char(11),char(0),char(55),char(0),char(-97),char(0),char(17),char(0),char(-94),char(0),char(17),char(0),char(-93),char(0),
-char(7),char(0),char(-82),char(0),char(7),char(0),char(-81),char(0),char(7),char(0),char(-80),char(0),char(7),char(0),char(-85),char(0),char(7),char(0),char(-84),char(0),
-char(7),char(0),char(-83),char(0),char(7),char(0),char(-79),char(0),char(0),char(0),char(21),char(0),char(62),char(0),char(9),char(0),char(55),char(0),char(-97),char(0),
-char(17),char(0),char(-94),char(0),char(17),char(0),char(-93),char(0),char(13),char(0),char(-78),char(0),char(13),char(0),char(-77),char(0),char(13),char(0),char(-76),char(0),
-char(13),char(0),char(-75),char(0),char(4),char(0),char(-74),char(0),char(4),char(0),char(-73),char(0),char(63),char(0),char(5),char(0),char(62),char(0),char(-72),char(0),
-char(4),char(0),char(-71),char(0),char(7),char(0),char(-70),char(0),char(7),char(0),char(-69),char(0),char(7),char(0),char(-68),char(0),char(64),char(0),char(9),char(0),
-char(55),char(0),char(-97),char(0),char(17),char(0),char(-94),char(0),char(17),char(0),char(-93),char(0),char(7),char(0),char(-78),char(0),char(7),char(0),char(-77),char(0),
-char(7),char(0),char(-76),char(0),char(7),char(0),char(-75),char(0),char(4),char(0),char(-74),char(0),char(4),char(0),char(-73),char(0),char(49),char(0),char(22),char(0),
-char(8),char(0),char(-67),char(0),char(8),char(0),char(-79),char(0),char(8),char(0),char(110),char(0),char(8),char(0),char(-66),char(0),char(8),char(0),char(112),char(0),
-char(8),char(0),char(-65),char(0),char(8),char(0),char(-64),char(0),char(8),char(0),char(-63),char(0),char(8),char(0),char(-62),char(0),char(8),char(0),char(-61),char(0),
-char(8),char(0),char(-60),char(0),char(8),char(0),char(-59),char(0),char(8),char(0),char(-58),char(0),char(8),char(0),char(-57),char(0),char(8),char(0),char(-56),char(0),
-char(8),char(0),char(-55),char(0),char(4),char(0),char(-54),char(0),char(4),char(0),char(-53),char(0),char(4),char(0),char(-52),char(0),char(4),char(0),char(-51),char(0),
-char(4),char(0),char(-50),char(0),char(0),char(0),char(37),char(0),char(51),char(0),char(22),char(0),char(7),char(0),char(-67),char(0),char(7),char(0),char(-79),char(0),
-char(7),char(0),char(110),char(0),char(7),char(0),char(-66),char(0),char(7),char(0),char(112),char(0),char(7),char(0),char(-65),char(0),char(7),char(0),char(-64),char(0),
-char(7),char(0),char(-63),char(0),char(7),char(0),char(-62),char(0),char(7),char(0),char(-61),char(0),char(7),char(0),char(-60),char(0),char(7),char(0),char(-59),char(0),
-char(7),char(0),char(-58),char(0),char(7),char(0),char(-57),char(0),char(7),char(0),char(-56),char(0),char(7),char(0),char(-55),char(0),char(4),char(0),char(-54),char(0),
-char(4),char(0),char(-53),char(0),char(4),char(0),char(-52),char(0),char(4),char(0),char(-51),char(0),char(4),char(0),char(-50),char(0),char(0),char(0),char(37),char(0),
-char(65),char(0),char(4),char(0),char(7),char(0),char(-49),char(0),char(7),char(0),char(-48),char(0),char(7),char(0),char(-47),char(0),char(4),char(0),char(78),char(0),
-char(66),char(0),char(10),char(0),char(65),char(0),char(-46),char(0),char(13),char(0),char(-45),char(0),char(13),char(0),char(-44),char(0),char(13),char(0),char(-43),char(0),
-char(13),char(0),char(-42),char(0),char(13),char(0),char(-41),char(0),char(7),char(0),char(-121),char(0),char(7),char(0),char(-40),char(0),char(4),char(0),char(-39),char(0),
-char(4),char(0),char(53),char(0),char(67),char(0),char(4),char(0),char(65),char(0),char(-46),char(0),char(4),char(0),char(-38),char(0),char(7),char(0),char(-37),char(0),
-char(4),char(0),char(-36),char(0),char(68),char(0),char(4),char(0),char(13),char(0),char(-41),char(0),char(65),char(0),char(-46),char(0),char(4),char(0),char(-35),char(0),
-char(7),char(0),char(-34),char(0),char(69),char(0),char(7),char(0),char(13),char(0),char(-33),char(0),char(65),char(0),char(-46),char(0),char(4),char(0),char(-32),char(0),
-char(7),char(0),char(-31),char(0),char(7),char(0),char(-30),char(0),char(7),char(0),char(-29),char(0),char(4),char(0),char(53),char(0),char(70),char(0),char(6),char(0),
-char(15),char(0),char(-28),char(0),char(13),char(0),char(-30),char(0),char(13),char(0),char(-27),char(0),char(56),char(0),char(-26),char(0),char(4),char(0),char(-25),char(0),
-char(7),char(0),char(-29),char(0),char(71),char(0),char(26),char(0),char(4),char(0),char(-24),char(0),char(7),char(0),char(-23),char(0),char(7),char(0),char(-79),char(0),
-char(7),char(0),char(-22),char(0),char(7),char(0),char(-21),char(0),char(7),char(0),char(-20),char(0),char(7),char(0),char(-19),char(0),char(7),char(0),char(-18),char(0),
-char(7),char(0),char(-17),char(0),char(7),char(0),char(-16),char(0),char(7),char(0),char(-15),char(0),char(7),char(0),char(-14),char(0),char(7),char(0),char(-13),char(0),
-char(7),char(0),char(-12),char(0),char(7),char(0),char(-11),char(0),char(7),char(0),char(-10),char(0),char(7),char(0),char(-9),char(0),char(7),char(0),char(-8),char(0),
-char(7),char(0),char(-7),char(0),char(7),char(0),char(-6),char(0),char(7),char(0),char(-5),char(0),char(4),char(0),char(-4),char(0),char(4),char(0),char(-3),char(0),
-char(4),char(0),char(-2),char(0),char(4),char(0),char(-1),char(0),char(4),char(0),char(117),char(0),char(72),char(0),char(12),char(0),char(15),char(0),char(0),char(1),
-char(15),char(0),char(1),char(1),char(15),char(0),char(2),char(1),char(13),char(0),char(3),char(1),char(13),char(0),char(4),char(1),char(7),char(0),char(5),char(1),
-char(4),char(0),char(6),char(1),char(4),char(0),char(7),char(1),char(4),char(0),char(8),char(1),char(4),char(0),char(9),char(1),char(7),char(0),char(-31),char(0),
-char(4),char(0),char(53),char(0),char(73),char(0),char(27),char(0),char(17),char(0),char(10),char(1),char(15),char(0),char(11),char(1),char(15),char(0),char(12),char(1),
-char(13),char(0),char(3),char(1),char(13),char(0),char(13),char(1),char(13),char(0),char(14),char(1),char(13),char(0),char(15),char(1),char(13),char(0),char(16),char(1),
-char(13),char(0),char(17),char(1),char(4),char(0),char(18),char(1),char(7),char(0),char(19),char(1),char(4),char(0),char(20),char(1),char(4),char(0),char(21),char(1),
-char(4),char(0),char(22),char(1),char(7),char(0),char(23),char(1),char(7),char(0),char(24),char(1),char(4),char(0),char(25),char(1),char(4),char(0),char(26),char(1),
-char(7),char(0),char(27),char(1),char(7),char(0),char(28),char(1),char(7),char(0),char(29),char(1),char(7),char(0),char(30),char(1),char(7),char(0),char(31),char(1),
-char(7),char(0),char(32),char(1),char(4),char(0),char(33),char(1),char(4),char(0),char(34),char(1),char(4),char(0),char(35),char(1),char(74),char(0),char(12),char(0),
-char(9),char(0),char(36),char(1),char(9),char(0),char(37),char(1),char(13),char(0),char(38),char(1),char(7),char(0),char(39),char(1),char(7),char(0),char(-63),char(0),
-char(7),char(0),char(40),char(1),char(4),char(0),char(41),char(1),char(13),char(0),char(42),char(1),char(4),char(0),char(43),char(1),char(4),char(0),char(44),char(1),
-char(4),char(0),char(45),char(1),char(4),char(0),char(53),char(0),char(75),char(0),char(19),char(0),char(47),char(0),char(125),char(0),char(72),char(0),char(46),char(1),
-char(65),char(0),char(47),char(1),char(66),char(0),char(48),char(1),char(67),char(0),char(49),char(1),char(68),char(0),char(50),char(1),char(69),char(0),char(51),char(1),
-char(70),char(0),char(52),char(1),char(73),char(0),char(53),char(1),char(74),char(0),char(54),char(1),char(4),char(0),char(55),char(1),char(4),char(0),char(21),char(1),
-char(4),char(0),char(56),char(1),char(4),char(0),char(57),char(1),char(4),char(0),char(58),char(1),char(4),char(0),char(59),char(1),char(4),char(0),char(60),char(1),
-char(4),char(0),char(61),char(1),char(71),char(0),char(62),char(1),};
-int b3s_bulletDNAlen= sizeof(b3s_bulletDNAstr);
-char b3s_bulletDNAstr64[]= {
-char(83),char(68),char(78),char(65),char(78),char(65),char(77),char(69),char(63),char(1),char(0),char(0),char(109),char(95),char(115),char(105),char(122),char(101),char(0),char(109),
-char(95),char(99),char(97),char(112),char(97),char(99),char(105),char(116),char(121),char(0),char(42),char(109),char(95),char(100),char(97),char(116),char(97),char(0),char(109),char(95),
-char(99),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(83),char(104),char(97),char(112),char(101),char(115),char(0),char(109),char(95),char(99),char(111),
-char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(79),char(98),char(106),char(101),char(99),char(116),char(115),char(0),char(109),char(95),char(99),char(111),char(110),
-char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(115),char(0),char(42),char(102),char(105),char(114),char(115),char(116),char(0),char(42),char(108),char(97),char(115),
-char(116),char(0),char(109),char(95),char(102),char(108),char(111),char(97),char(116),char(115),char(91),char(52),char(93),char(0),char(109),char(95),char(101),char(108),char(91),char(51),
-char(93),char(0),char(109),char(95),char(98),char(97),char(115),char(105),char(115),char(0),char(109),char(95),char(111),char(114),char(105),char(103),char(105),char(110),char(0),char(109),
-char(95),char(114),char(111),char(111),char(116),char(78),char(111),char(100),char(101),char(73),char(110),char(100),char(101),char(120),char(0),char(109),char(95),char(115),char(117),char(98),
-char(116),char(114),char(101),char(101),char(83),char(105),char(122),char(101),char(0),char(109),char(95),char(113),char(117),char(97),char(110),char(116),char(105),char(122),char(101),char(100),
-char(65),char(97),char(98),char(98),char(77),char(105),char(110),char(91),char(51),char(93),char(0),char(109),char(95),char(113),char(117),char(97),char(110),char(116),char(105),char(122),
-char(101),char(100),char(65),char(97),char(98),char(98),char(77),char(97),char(120),char(91),char(51),char(93),char(0),char(109),char(95),char(97),char(97),char(98),char(98),char(77),
-char(105),char(110),char(79),char(114),char(103),char(0),char(109),char(95),char(97),char(97),char(98),char(98),char(77),char(97),char(120),char(79),char(114),char(103),char(0),char(109),
-char(95),char(101),char(115),char(99),char(97),char(112),char(101),char(73),char(110),char(100),char(101),char(120),char(0),char(109),char(95),char(115),char(117),char(98),char(80),char(97),
-char(114),char(116),char(0),char(109),char(95),char(116),char(114),char(105),char(97),char(110),char(103),char(108),char(101),char(73),char(110),char(100),char(101),char(120),char(0),char(109),
-char(95),char(112),char(97),char(100),char(91),char(52),char(93),char(0),char(109),char(95),char(101),char(115),char(99),char(97),char(112),char(101),char(73),char(110),char(100),char(101),
-char(120),char(79),char(114),char(84),char(114),char(105),char(97),char(110),char(103),char(108),char(101),char(73),char(110),char(100),char(101),char(120),char(0),char(109),char(95),char(98),
-char(118),char(104),char(65),char(97),char(98),char(98),char(77),char(105),char(110),char(0),char(109),char(95),char(98),char(118),char(104),char(65),char(97),char(98),char(98),char(77),
-char(97),char(120),char(0),char(109),char(95),char(98),char(118),char(104),char(81),char(117),char(97),char(110),char(116),char(105),char(122),char(97),char(116),char(105),char(111),char(110),
-char(0),char(109),char(95),char(99),char(117),char(114),char(78),char(111),char(100),char(101),char(73),char(110),char(100),char(101),char(120),char(0),char(109),char(95),char(117),char(115),
-char(101),char(81),char(117),char(97),char(110),char(116),char(105),char(122),char(97),char(116),char(105),char(111),char(110),char(0),char(109),char(95),char(110),char(117),char(109),char(67),
-char(111),char(110),char(116),char(105),char(103),char(117),char(111),char(117),char(115),char(76),char(101),char(97),char(102),char(78),char(111),char(100),char(101),char(115),char(0),char(109),
-char(95),char(110),char(117),char(109),char(81),char(117),char(97),char(110),char(116),char(105),char(122),char(101),char(100),char(67),char(111),char(110),char(116),char(105),char(103),char(117),
-char(111),char(117),char(115),char(78),char(111),char(100),char(101),char(115),char(0),char(42),char(109),char(95),char(99),char(111),char(110),char(116),char(105),char(103),char(117),char(111),
-char(117),char(115),char(78),char(111),char(100),char(101),char(115),char(80),char(116),char(114),char(0),char(42),char(109),char(95),char(113),char(117),char(97),char(110),char(116),char(105),
-char(122),char(101),char(100),char(67),char(111),char(110),char(116),char(105),char(103),char(117),char(111),char(117),char(115),char(78),char(111),char(100),char(101),char(115),char(80),char(116),
-char(114),char(0),char(42),char(109),char(95),char(115),char(117),char(98),char(84),char(114),char(101),char(101),char(73),char(110),char(102),char(111),char(80),char(116),char(114),char(0),
-char(109),char(95),char(116),char(114),char(97),char(118),char(101),char(114),char(115),char(97),char(108),char(77),char(111),char(100),char(101),char(0),char(109),char(95),char(110),char(117),
-char(109),char(83),char(117),char(98),char(116),char(114),char(101),char(101),char(72),char(101),char(97),char(100),char(101),char(114),char(115),char(0),char(42),char(109),char(95),char(110),
-char(97),char(109),char(101),char(0),char(109),char(95),char(115),char(104),char(97),char(112),char(101),char(84),char(121),char(112),char(101),char(0),char(109),char(95),char(112),char(97),
-char(100),char(100),char(105),char(110),char(103),char(91),char(52),char(93),char(0),char(109),char(95),char(99),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),
-char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(109),char(95),char(108),char(111),char(99),char(97),char(108),char(83),char(99),char(97),
-char(108),char(105),char(110),char(103),char(0),char(109),char(95),char(112),char(108),char(97),char(110),char(101),char(78),char(111),char(114),char(109),char(97),char(108),char(0),char(109),
-char(95),char(112),char(108),char(97),char(110),char(101),char(67),char(111),char(110),char(115),char(116),char(97),char(110),char(116),char(0),char(109),char(95),char(105),char(109),char(112),
-char(108),char(105),char(99),char(105),char(116),char(83),char(104),char(97),char(112),char(101),char(68),char(105),char(109),char(101),char(110),char(115),char(105),char(111),char(110),char(115),
-char(0),char(109),char(95),char(99),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(77),char(97),char(114),char(103),char(105),char(110),char(0),char(109),
-char(95),char(112),char(97),char(100),char(100),char(105),char(110),char(103),char(0),char(109),char(95),char(112),char(111),char(115),char(0),char(109),char(95),char(114),char(97),char(100),
-char(105),char(117),char(115),char(0),char(109),char(95),char(99),char(111),char(110),char(118),char(101),char(120),char(73),char(110),char(116),char(101),char(114),char(110),char(97),char(108),
-char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(42),char(109),char(95),char(108),char(111),char(99),char(97),char(108),char(80),char(111),
-char(115),char(105),char(116),char(105),char(111),char(110),char(65),char(114),char(114),char(97),char(121),char(80),char(116),char(114),char(0),char(109),char(95),char(108),char(111),char(99),
-char(97),char(108),char(80),char(111),char(115),char(105),char(116),char(105),char(111),char(110),char(65),char(114),char(114),char(97),char(121),char(83),char(105),char(122),char(101),char(0),
-char(109),char(95),char(118),char(97),char(108),char(117),char(101),char(0),char(109),char(95),char(112),char(97),char(100),char(91),char(50),char(93),char(0),char(109),char(95),char(118),
-char(97),char(108),char(117),char(101),char(115),char(91),char(51),char(93),char(0),char(109),char(95),char(112),char(97),char(100),char(0),char(42),char(109),char(95),char(118),char(101),
-char(114),char(116),char(105),char(99),char(101),char(115),char(51),char(102),char(0),char(42),char(109),char(95),char(118),char(101),char(114),char(116),char(105),char(99),char(101),char(115),
-char(51),char(100),char(0),char(42),char(109),char(95),char(105),char(110),char(100),char(105),char(99),char(101),char(115),char(51),char(50),char(0),char(42),char(109),char(95),char(51),
-char(105),char(110),char(100),char(105),char(99),char(101),char(115),char(49),char(54),char(0),char(42),char(109),char(95),char(51),char(105),char(110),char(100),char(105),char(99),char(101),
-char(115),char(56),char(0),char(42),char(109),char(95),char(105),char(110),char(100),char(105),char(99),char(101),char(115),char(49),char(54),char(0),char(109),char(95),char(110),char(117),
-char(109),char(84),char(114),char(105),char(97),char(110),char(103),char(108),char(101),char(115),char(0),char(109),char(95),char(110),char(117),char(109),char(86),char(101),char(114),char(116),
-char(105),char(99),char(101),char(115),char(0),char(42),char(109),char(95),char(109),char(101),char(115),char(104),char(80),char(97),char(114),char(116),char(115),char(80),char(116),char(114),
-char(0),char(109),char(95),char(115),char(99),char(97),char(108),char(105),char(110),char(103),char(0),char(109),char(95),char(110),char(117),char(109),char(77),char(101),char(115),char(104),
-char(80),char(97),char(114),char(116),char(115),char(0),char(109),char(95),char(109),char(101),char(115),char(104),char(73),char(110),char(116),char(101),char(114),char(102),char(97),char(99),
-char(101),char(0),char(42),char(109),char(95),char(113),char(117),char(97),char(110),char(116),char(105),char(122),char(101),char(100),char(70),char(108),char(111),char(97),char(116),char(66),
-char(118),char(104),char(0),char(42),char(109),char(95),char(113),char(117),char(97),char(110),char(116),char(105),char(122),char(101),char(100),char(68),char(111),char(117),char(98),char(108),
-char(101),char(66),char(118),char(104),char(0),char(42),char(109),char(95),char(116),char(114),char(105),char(97),char(110),char(103),char(108),char(101),char(73),char(110),char(102),char(111),
-char(77),char(97),char(112),char(0),char(109),char(95),char(112),char(97),char(100),char(51),char(91),char(52),char(93),char(0),char(109),char(95),char(116),char(114),char(105),char(109),
-char(101),char(115),char(104),char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(109),char(95),char(116),char(114),char(97),char(110),char(115),
-char(102),char(111),char(114),char(109),char(0),char(42),char(109),char(95),char(99),char(104),char(105),char(108),char(100),char(83),char(104),char(97),char(112),char(101),char(0),char(109),
-char(95),char(99),char(104),char(105),char(108),char(100),char(83),char(104),char(97),char(112),char(101),char(84),char(121),char(112),char(101),char(0),char(109),char(95),char(99),char(104),
-char(105),char(108),char(100),char(77),char(97),char(114),char(103),char(105),char(110),char(0),char(42),char(109),char(95),char(99),char(104),char(105),char(108),char(100),char(83),char(104),
-char(97),char(112),char(101),char(80),char(116),char(114),char(0),char(109),char(95),char(110),char(117),char(109),char(67),char(104),char(105),char(108),char(100),char(83),char(104),char(97),
-char(112),char(101),char(115),char(0),char(109),char(95),char(117),char(112),char(65),char(120),char(105),char(115),char(0),char(109),char(95),char(102),char(108),char(97),char(103),char(115),
-char(0),char(109),char(95),char(101),char(100),char(103),char(101),char(86),char(48),char(86),char(49),char(65),char(110),char(103),char(108),char(101),char(0),char(109),char(95),char(101),
-char(100),char(103),char(101),char(86),char(49),char(86),char(50),char(65),char(110),char(103),char(108),char(101),char(0),char(109),char(95),char(101),char(100),char(103),char(101),char(86),
-char(50),char(86),char(48),char(65),char(110),char(103),char(108),char(101),char(0),char(42),char(109),char(95),char(104),char(97),char(115),char(104),char(84),char(97),char(98),char(108),
-char(101),char(80),char(116),char(114),char(0),char(42),char(109),char(95),char(110),char(101),char(120),char(116),char(80),char(116),char(114),char(0),char(42),char(109),char(95),char(118),
-char(97),char(108),char(117),char(101),char(65),char(114),char(114),char(97),char(121),char(80),char(116),char(114),char(0),char(42),char(109),char(95),char(107),char(101),char(121),char(65),
-char(114),char(114),char(97),char(121),char(80),char(116),char(114),char(0),char(109),char(95),char(99),char(111),char(110),char(118),char(101),char(120),char(69),char(112),char(115),char(105),
-char(108),char(111),char(110),char(0),char(109),char(95),char(112),char(108),char(97),char(110),char(97),char(114),char(69),char(112),char(115),char(105),char(108),char(111),char(110),char(0),
-char(109),char(95),char(101),char(113),char(117),char(97),char(108),char(86),char(101),char(114),char(116),char(101),char(120),char(84),char(104),char(114),char(101),char(115),char(104),char(111),
-char(108),char(100),char(0),char(109),char(95),char(101),char(100),char(103),char(101),char(68),char(105),char(115),char(116),char(97),char(110),char(99),char(101),char(84),char(104),char(114),
-char(101),char(115),char(104),char(111),char(108),char(100),char(0),char(109),char(95),char(122),char(101),char(114),char(111),char(65),char(114),char(101),char(97),char(84),char(104),char(114),
-char(101),char(115),char(104),char(111),char(108),char(100),char(0),char(109),char(95),char(110),char(101),char(120),char(116),char(83),char(105),char(122),char(101),char(0),char(109),char(95),
-char(104),char(97),char(115),char(104),char(84),char(97),char(98),char(108),char(101),char(83),char(105),char(122),char(101),char(0),char(109),char(95),char(110),char(117),char(109),char(86),
-char(97),char(108),char(117),char(101),char(115),char(0),char(109),char(95),char(110),char(117),char(109),char(75),char(101),char(121),char(115),char(0),char(109),char(95),char(103),char(105),
-char(109),char(112),char(97),char(99),char(116),char(83),char(117),char(98),char(84),char(121),char(112),char(101),char(0),char(42),char(109),char(95),char(117),char(110),char(115),char(99),
-char(97),char(108),char(101),char(100),char(80),char(111),char(105),char(110),char(116),char(115),char(70),char(108),char(111),char(97),char(116),char(80),char(116),char(114),char(0),char(42),
-char(109),char(95),char(117),char(110),char(115),char(99),char(97),char(108),char(101),char(100),char(80),char(111),char(105),char(110),char(116),char(115),char(68),char(111),char(117),char(98),
-char(108),char(101),char(80),char(116),char(114),char(0),char(109),char(95),char(110),char(117),char(109),char(85),char(110),char(115),char(99),char(97),char(108),char(101),char(100),char(80),
-char(111),char(105),char(110),char(116),char(115),char(0),char(109),char(95),char(112),char(97),char(100),char(100),char(105),char(110),char(103),char(51),char(91),char(52),char(93),char(0),
-char(42),char(109),char(95),char(98),char(114),char(111),char(97),char(100),char(112),char(104),char(97),char(115),char(101),char(72),char(97),char(110),char(100),char(108),char(101),char(0),
-char(42),char(109),char(95),char(99),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(83),char(104),char(97),char(112),char(101),char(0),char(42),char(109),
-char(95),char(114),char(111),char(111),char(116),char(67),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(83),char(104),char(97),char(112),char(101),char(0),
-char(109),char(95),char(119),char(111),char(114),char(108),char(100),char(84),char(114),char(97),char(110),char(115),char(102),char(111),char(114),char(109),char(0),char(109),char(95),char(105),
-char(110),char(116),char(101),char(114),char(112),char(111),char(108),char(97),char(116),char(105),char(111),char(110),char(87),char(111),char(114),char(108),char(100),char(84),char(114),char(97),
-char(110),char(115),char(102),char(111),char(114),char(109),char(0),char(109),char(95),char(105),char(110),char(116),char(101),char(114),char(112),char(111),char(108),char(97),char(116),char(105),
-char(111),char(110),char(76),char(105),char(110),char(101),char(97),char(114),char(86),char(101),char(108),char(111),char(99),char(105),char(116),char(121),char(0),char(109),char(95),char(105),
-char(110),char(116),char(101),char(114),char(112),char(111),char(108),char(97),char(116),char(105),char(111),char(110),char(65),char(110),char(103),char(117),char(108),char(97),char(114),char(86),
-char(101),char(108),char(111),char(99),char(105),char(116),char(121),char(0),char(109),char(95),char(97),char(110),char(105),char(115),char(111),char(116),char(114),char(111),char(112),char(105),
-char(99),char(70),char(114),char(105),char(99),char(116),char(105),char(111),char(110),char(0),char(109),char(95),char(99),char(111),char(110),char(116),char(97),char(99),char(116),char(80),
-char(114),char(111),char(99),char(101),char(115),char(115),char(105),char(110),char(103),char(84),char(104),char(114),char(101),char(115),char(104),char(111),char(108),char(100),char(0),char(109),
-char(95),char(100),char(101),char(97),char(99),char(116),char(105),char(118),char(97),char(116),char(105),char(111),char(110),char(84),char(105),char(109),char(101),char(0),char(109),char(95),
-char(102),char(114),char(105),char(99),char(116),char(105),char(111),char(110),char(0),char(109),char(95),char(114),char(111),char(108),char(108),char(105),char(110),char(103),char(70),char(114),
-char(105),char(99),char(116),char(105),char(111),char(110),char(0),char(109),char(95),char(114),char(101),char(115),char(116),char(105),char(116),char(117),char(116),char(105),char(111),char(110),
-char(0),char(109),char(95),char(104),char(105),char(116),char(70),char(114),char(97),char(99),char(116),char(105),char(111),char(110),char(0),char(109),char(95),char(99),char(99),char(100),
-char(83),char(119),char(101),char(112),char(116),char(83),char(112),char(104),char(101),char(114),char(101),char(82),char(97),char(100),char(105),char(117),char(115),char(0),char(109),char(95),
-char(99),char(99),char(100),char(77),char(111),char(116),char(105),char(111),char(110),char(84),char(104),char(114),char(101),char(115),char(104),char(111),char(108),char(100),char(0),char(109),
-char(95),char(104),char(97),char(115),char(65),char(110),char(105),char(115),char(111),char(116),char(114),char(111),char(112),char(105),char(99),char(70),char(114),char(105),char(99),char(116),
-char(105),char(111),char(110),char(0),char(109),char(95),char(99),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(70),char(108),char(97),char(103),char(115),
-char(0),char(109),char(95),char(105),char(115),char(108),char(97),char(110),char(100),char(84),char(97),char(103),char(49),char(0),char(109),char(95),char(99),char(111),char(109),char(112),
-char(97),char(110),char(105),char(111),char(110),char(73),char(100),char(0),char(109),char(95),char(97),char(99),char(116),char(105),char(118),char(97),char(116),char(105),char(111),char(110),
-char(83),char(116),char(97),char(116),char(101),char(49),char(0),char(109),char(95),char(105),char(110),char(116),char(101),char(114),char(110),char(97),char(108),char(84),char(121),char(112),
-char(101),char(0),char(109),char(95),char(99),char(104),char(101),char(99),char(107),char(67),char(111),char(108),char(108),char(105),char(100),char(101),char(87),char(105),char(116),char(104),
-char(0),char(109),char(95),char(115),char(111),char(108),char(118),char(101),char(114),char(73),char(110),char(102),char(111),char(0),char(109),char(95),char(103),char(114),char(97),char(118),
-char(105),char(116),char(121),char(0),char(109),char(95),char(99),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(79),char(98),char(106),char(101),char(99),
-char(116),char(68),char(97),char(116),char(97),char(0),char(109),char(95),char(105),char(110),char(118),char(73),char(110),char(101),char(114),char(116),char(105),char(97),char(84),char(101),
-char(110),char(115),char(111),char(114),char(87),char(111),char(114),char(108),char(100),char(0),char(109),char(95),char(108),char(105),char(110),char(101),char(97),char(114),char(86),char(101),
-char(108),char(111),char(99),char(105),char(116),char(121),char(0),char(109),char(95),char(97),char(110),char(103),char(117),char(108),char(97),char(114),char(86),char(101),char(108),char(111),
-char(99),char(105),char(116),char(121),char(0),char(109),char(95),char(97),char(110),char(103),char(117),char(108),char(97),char(114),char(70),char(97),char(99),char(116),char(111),char(114),
-char(0),char(109),char(95),char(108),char(105),char(110),char(101),char(97),char(114),char(70),char(97),char(99),char(116),char(111),char(114),char(0),char(109),char(95),char(103),char(114),
-char(97),char(118),char(105),char(116),char(121),char(95),char(97),char(99),char(99),char(101),char(108),char(101),char(114),char(97),char(116),char(105),char(111),char(110),char(0),char(109),
-char(95),char(105),char(110),char(118),char(73),char(110),char(101),char(114),char(116),char(105),char(97),char(76),char(111),char(99),char(97),char(108),char(0),char(109),char(95),char(116),
-char(111),char(116),char(97),char(108),char(70),char(111),char(114),char(99),char(101),char(0),char(109),char(95),char(116),char(111),char(116),char(97),char(108),char(84),char(111),char(114),
-char(113),char(117),char(101),char(0),char(109),char(95),char(105),char(110),char(118),char(101),char(114),char(115),char(101),char(77),char(97),char(115),char(115),char(0),char(109),char(95),
-char(108),char(105),char(110),char(101),char(97),char(114),char(68),char(97),char(109),char(112),char(105),char(110),char(103),char(0),char(109),char(95),char(97),char(110),char(103),char(117),
-char(108),char(97),char(114),char(68),char(97),char(109),char(112),char(105),char(110),char(103),char(0),char(109),char(95),char(97),char(100),char(100),char(105),char(116),char(105),char(111),
-char(110),char(97),char(108),char(68),char(97),char(109),char(112),char(105),char(110),char(103),char(70),char(97),char(99),char(116),char(111),char(114),char(0),char(109),char(95),char(97),
-char(100),char(100),char(105),char(116),char(105),char(111),char(110),char(97),char(108),char(76),char(105),char(110),char(101),char(97),char(114),char(68),char(97),char(109),char(112),char(105),
-char(110),char(103),char(84),char(104),char(114),char(101),char(115),char(104),char(111),char(108),char(100),char(83),char(113),char(114),char(0),char(109),char(95),char(97),char(100),char(100),
-char(105),char(116),char(105),char(111),char(110),char(97),char(108),char(65),char(110),char(103),char(117),char(108),char(97),char(114),char(68),char(97),char(109),char(112),char(105),char(110),
-char(103),char(84),char(104),char(114),char(101),char(115),char(104),char(111),char(108),char(100),char(83),char(113),char(114),char(0),char(109),char(95),char(97),char(100),char(100),char(105),
-char(116),char(105),char(111),char(110),char(97),char(108),char(65),char(110),char(103),char(117),char(108),char(97),char(114),char(68),char(97),char(109),char(112),char(105),char(110),char(103),
-char(70),char(97),char(99),char(116),char(111),char(114),char(0),char(109),char(95),char(108),char(105),char(110),char(101),char(97),char(114),char(83),char(108),char(101),char(101),char(112),
-char(105),char(110),char(103),char(84),char(104),char(114),char(101),char(115),char(104),char(111),char(108),char(100),char(0),char(109),char(95),char(97),char(110),char(103),char(117),char(108),
-char(97),char(114),char(83),char(108),char(101),char(101),char(112),char(105),char(110),char(103),char(84),char(104),char(114),char(101),char(115),char(104),char(111),char(108),char(100),char(0),
-char(109),char(95),char(97),char(100),char(100),char(105),char(116),char(105),char(111),char(110),char(97),char(108),char(68),char(97),char(109),char(112),char(105),char(110),char(103),char(0),
-char(109),char(95),char(110),char(117),char(109),char(67),char(111),char(110),char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(82),char(111),char(119),char(115),char(0),
-char(110),char(117),char(98),char(0),char(42),char(109),char(95),char(114),char(98),char(65),char(0),char(42),char(109),char(95),char(114),char(98),char(66),char(0),char(109),char(95),
-char(111),char(98),char(106),char(101),char(99),char(116),char(84),char(121),char(112),char(101),char(0),char(109),char(95),char(117),char(115),char(101),char(114),char(67),char(111),char(110),
-char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(84),char(121),char(112),char(101),char(0),char(109),char(95),char(117),char(115),char(101),char(114),char(67),char(111),
-char(110),char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(73),char(100),char(0),char(109),char(95),char(110),char(101),char(101),char(100),char(115),char(70),char(101),
-char(101),char(100),char(98),char(97),char(99),char(107),char(0),char(109),char(95),char(97),char(112),char(112),char(108),char(105),char(101),char(100),char(73),char(109),char(112),char(117),
-char(108),char(115),char(101),char(0),char(109),char(95),char(100),char(98),char(103),char(68),char(114),char(97),char(119),char(83),char(105),char(122),char(101),char(0),char(109),char(95),
-char(100),char(105),char(115),char(97),char(98),char(108),char(101),char(67),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(115),char(66),char(101),char(116),
-char(119),char(101),char(101),char(110),char(76),char(105),char(110),char(107),char(101),char(100),char(66),char(111),char(100),char(105),char(101),char(115),char(0),char(109),char(95),char(111),
-char(118),char(101),char(114),char(114),char(105),char(100),char(101),char(78),char(117),char(109),char(83),char(111),char(108),char(118),char(101),char(114),char(73),char(116),char(101),char(114),
-char(97),char(116),char(105),char(111),char(110),char(115),char(0),char(109),char(95),char(98),char(114),char(101),char(97),char(107),char(105),char(110),char(103),char(73),char(109),char(112),
-char(117),char(108),char(115),char(101),char(84),char(104),char(114),char(101),char(115),char(104),char(111),char(108),char(100),char(0),char(109),char(95),char(105),char(115),char(69),char(110),
-char(97),char(98),char(108),char(101),char(100),char(0),char(109),char(95),char(116),char(121),char(112),char(101),char(67),char(111),char(110),char(115),char(116),char(114),char(97),char(105),
-char(110),char(116),char(68),char(97),char(116),char(97),char(0),char(109),char(95),char(112),char(105),char(118),char(111),char(116),char(73),char(110),char(65),char(0),char(109),char(95),
-char(112),char(105),char(118),char(111),char(116),char(73),char(110),char(66),char(0),char(109),char(95),char(114),char(98),char(65),char(70),char(114),char(97),char(109),char(101),char(0),
-char(109),char(95),char(114),char(98),char(66),char(70),char(114),char(97),char(109),char(101),char(0),char(109),char(95),char(117),char(115),char(101),char(82),char(101),char(102),char(101),
-char(114),char(101),char(110),char(99),char(101),char(70),char(114),char(97),char(109),char(101),char(65),char(0),char(109),char(95),char(97),char(110),char(103),char(117),char(108),char(97),
-char(114),char(79),char(110),char(108),char(121),char(0),char(109),char(95),char(101),char(110),char(97),char(98),char(108),char(101),char(65),char(110),char(103),char(117),char(108),char(97),
-char(114),char(77),char(111),char(116),char(111),char(114),char(0),char(109),char(95),char(109),char(111),char(116),char(111),char(114),char(84),char(97),char(114),char(103),char(101),char(116),
-char(86),char(101),char(108),char(111),char(99),char(105),char(116),char(121),char(0),char(109),char(95),char(109),char(97),char(120),char(77),char(111),char(116),char(111),char(114),char(73),
-char(109),char(112),char(117),char(108),char(115),char(101),char(0),char(109),char(95),char(108),char(111),char(119),char(101),char(114),char(76),char(105),char(109),char(105),char(116),char(0),
-char(109),char(95),char(117),char(112),char(112),char(101),char(114),char(76),char(105),char(109),char(105),char(116),char(0),char(109),char(95),char(108),char(105),char(109),char(105),char(116),
-char(83),char(111),char(102),char(116),char(110),char(101),char(115),char(115),char(0),char(109),char(95),char(98),char(105),char(97),char(115),char(70),char(97),char(99),char(116),char(111),
-char(114),char(0),char(109),char(95),char(114),char(101),char(108),char(97),char(120),char(97),char(116),char(105),char(111),char(110),char(70),char(97),char(99),char(116),char(111),char(114),
-char(0),char(109),char(95),char(115),char(119),char(105),char(110),char(103),char(83),char(112),char(97),char(110),char(49),char(0),char(109),char(95),char(115),char(119),char(105),char(110),
-char(103),char(83),char(112),char(97),char(110),char(50),char(0),char(109),char(95),char(116),char(119),char(105),char(115),char(116),char(83),char(112),char(97),char(110),char(0),char(109),
-char(95),char(100),char(97),char(109),char(112),char(105),char(110),char(103),char(0),char(109),char(95),char(108),char(105),char(110),char(101),char(97),char(114),char(85),char(112),char(112),
-char(101),char(114),char(76),char(105),char(109),char(105),char(116),char(0),char(109),char(95),char(108),char(105),char(110),char(101),char(97),char(114),char(76),char(111),char(119),char(101),
-char(114),char(76),char(105),char(109),char(105),char(116),char(0),char(109),char(95),char(97),char(110),char(103),char(117),char(108),char(97),char(114),char(85),char(112),char(112),char(101),
-char(114),char(76),char(105),char(109),char(105),char(116),char(0),char(109),char(95),char(97),char(110),char(103),char(117),char(108),char(97),char(114),char(76),char(111),char(119),char(101),
-char(114),char(76),char(105),char(109),char(105),char(116),char(0),char(109),char(95),char(117),char(115),char(101),char(76),char(105),char(110),char(101),char(97),char(114),char(82),char(101),
-char(102),char(101),char(114),char(101),char(110),char(99),char(101),char(70),char(114),char(97),char(109),char(101),char(65),char(0),char(109),char(95),char(117),char(115),char(101),char(79),
-char(102),char(102),char(115),char(101),char(116),char(70),char(111),char(114),char(67),char(111),char(110),char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(70),char(114),
-char(97),char(109),char(101),char(0),char(109),char(95),char(54),char(100),char(111),char(102),char(68),char(97),char(116),char(97),char(0),char(109),char(95),char(115),char(112),char(114),
-char(105),char(110),char(103),char(69),char(110),char(97),char(98),char(108),char(101),char(100),char(91),char(54),char(93),char(0),char(109),char(95),char(101),char(113),char(117),char(105),
-char(108),char(105),char(98),char(114),char(105),char(117),char(109),char(80),char(111),char(105),char(110),char(116),char(91),char(54),char(93),char(0),char(109),char(95),char(115),char(112),
-char(114),char(105),char(110),char(103),char(83),char(116),char(105),char(102),char(102),char(110),char(101),char(115),char(115),char(91),char(54),char(93),char(0),char(109),char(95),char(115),
-char(112),char(114),char(105),char(110),char(103),char(68),char(97),char(109),char(112),char(105),char(110),char(103),char(91),char(54),char(93),char(0),char(109),char(95),char(116),char(97),
-char(117),char(0),char(109),char(95),char(116),char(105),char(109),char(101),char(83),char(116),char(101),char(112),char(0),char(109),char(95),char(109),char(97),char(120),char(69),char(114),
-char(114),char(111),char(114),char(82),char(101),char(100),char(117),char(99),char(116),char(105),char(111),char(110),char(0),char(109),char(95),char(115),char(111),char(114),char(0),char(109),
-char(95),char(101),char(114),char(112),char(0),char(109),char(95),char(101),char(114),char(112),char(50),char(0),char(109),char(95),char(103),char(108),char(111),char(98),char(97),char(108),
-char(67),char(102),char(109),char(0),char(109),char(95),char(115),char(112),char(108),char(105),char(116),char(73),char(109),char(112),char(117),char(108),char(115),char(101),char(80),char(101),
-char(110),char(101),char(116),char(114),char(97),char(116),char(105),char(111),char(110),char(84),char(104),char(114),char(101),char(115),char(104),char(111),char(108),char(100),char(0),char(109),
-char(95),char(115),char(112),char(108),char(105),char(116),char(73),char(109),char(112),char(117),char(108),char(115),char(101),char(84),char(117),char(114),char(110),char(69),char(114),char(112),
-char(0),char(109),char(95),char(108),char(105),char(110),char(101),char(97),char(114),char(83),char(108),char(111),char(112),char(0),char(109),char(95),char(119),char(97),char(114),char(109),
-char(115),char(116),char(97),char(114),char(116),char(105),char(110),char(103),char(70),char(97),char(99),char(116),char(111),char(114),char(0),char(109),char(95),char(109),char(97),char(120),
-char(71),char(121),char(114),char(111),char(115),char(99),char(111),char(112),char(105),char(99),char(70),char(111),char(114),char(99),char(101),char(0),char(109),char(95),char(115),char(105),
-char(110),char(103),char(108),char(101),char(65),char(120),char(105),char(115),char(82),char(111),char(108),char(108),char(105),char(110),char(103),char(70),char(114),char(105),char(99),char(116),
-char(105),char(111),char(110),char(84),char(104),char(114),char(101),char(115),char(104),char(111),char(108),char(100),char(0),char(109),char(95),char(110),char(117),char(109),char(73),char(116),
-char(101),char(114),char(97),char(116),char(105),char(111),char(110),char(115),char(0),char(109),char(95),char(115),char(111),char(108),char(118),char(101),char(114),char(77),char(111),char(100),
-char(101),char(0),char(109),char(95),char(114),char(101),char(115),char(116),char(105),char(110),char(103),char(67),char(111),char(110),char(116),char(97),char(99),char(116),char(82),char(101),
-char(115),char(116),char(105),char(116),char(117),char(116),char(105),char(111),char(110),char(84),char(104),char(114),char(101),char(115),char(104),char(111),char(108),char(100),char(0),char(109),
-char(95),char(109),char(105),char(110),char(105),char(109),char(117),char(109),char(83),char(111),char(108),char(118),char(101),char(114),char(66),char(97),char(116),char(99),char(104),char(83),
-char(105),char(122),char(101),char(0),char(109),char(95),char(115),char(112),char(108),char(105),char(116),char(73),char(109),char(112),char(117),char(108),char(115),char(101),char(0),char(109),
-char(95),char(108),char(105),char(110),char(101),char(97),char(114),char(83),char(116),char(105),char(102),char(102),char(110),char(101),char(115),char(115),char(0),char(109),char(95),char(97),
-char(110),char(103),char(117),char(108),char(97),char(114),char(83),char(116),char(105),char(102),char(102),char(110),char(101),char(115),char(115),char(0),char(109),char(95),char(118),char(111),
-char(108),char(117),char(109),char(101),char(83),char(116),char(105),char(102),char(102),char(110),char(101),char(115),char(115),char(0),char(42),char(109),char(95),char(109),char(97),char(116),
-char(101),char(114),char(105),char(97),char(108),char(0),char(109),char(95),char(112),char(111),char(115),char(105),char(116),char(105),char(111),char(110),char(0),char(109),char(95),char(112),
-char(114),char(101),char(118),char(105),char(111),char(117),char(115),char(80),char(111),char(115),char(105),char(116),char(105),char(111),char(110),char(0),char(109),char(95),char(118),char(101),
-char(108),char(111),char(99),char(105),char(116),char(121),char(0),char(109),char(95),char(97),char(99),char(99),char(117),char(109),char(117),char(108),char(97),char(116),char(101),char(100),
-char(70),char(111),char(114),char(99),char(101),char(0),char(109),char(95),char(110),char(111),char(114),char(109),char(97),char(108),char(0),char(109),char(95),char(97),char(114),char(101),
-char(97),char(0),char(109),char(95),char(97),char(116),char(116),char(97),char(99),char(104),char(0),char(109),char(95),char(110),char(111),char(100),char(101),char(73),char(110),char(100),
-char(105),char(99),char(101),char(115),char(91),char(50),char(93),char(0),char(109),char(95),char(114),char(101),char(115),char(116),char(76),char(101),char(110),char(103),char(116),char(104),
-char(0),char(109),char(95),char(98),char(98),char(101),char(110),char(100),char(105),char(110),char(103),char(0),char(109),char(95),char(110),char(111),char(100),char(101),char(73),char(110),
-char(100),char(105),char(99),char(101),char(115),char(91),char(51),char(93),char(0),char(109),char(95),char(114),char(101),char(115),char(116),char(65),char(114),char(101),char(97),char(0),
-char(109),char(95),char(99),char(48),char(91),char(52),char(93),char(0),char(109),char(95),char(110),char(111),char(100),char(101),char(73),char(110),char(100),char(105),char(99),char(101),
-char(115),char(91),char(52),char(93),char(0),char(109),char(95),char(114),char(101),char(115),char(116),char(86),char(111),char(108),char(117),char(109),char(101),char(0),char(109),char(95),
-char(99),char(49),char(0),char(109),char(95),char(99),char(50),char(0),char(109),char(95),char(99),char(48),char(0),char(109),char(95),char(108),char(111),char(99),char(97),char(108),
-char(70),char(114),char(97),char(109),char(101),char(0),char(42),char(109),char(95),char(114),char(105),char(103),char(105),char(100),char(66),char(111),char(100),char(121),char(0),char(109),
-char(95),char(110),char(111),char(100),char(101),char(73),char(110),char(100),char(101),char(120),char(0),char(109),char(95),char(97),char(101),char(114),char(111),char(77),char(111),char(100),
-char(101),char(108),char(0),char(109),char(95),char(98),char(97),char(117),char(109),char(103),char(97),char(114),char(116),char(101),char(0),char(109),char(95),char(100),char(114),char(97),
-char(103),char(0),char(109),char(95),char(108),char(105),char(102),char(116),char(0),char(109),char(95),char(112),char(114),char(101),char(115),char(115),char(117),char(114),char(101),char(0),
-char(109),char(95),char(118),char(111),char(108),char(117),char(109),char(101),char(0),char(109),char(95),char(100),char(121),char(110),char(97),char(109),char(105),char(99),char(70),char(114),
-char(105),char(99),char(116),char(105),char(111),char(110),char(0),char(109),char(95),char(112),char(111),char(115),char(101),char(77),char(97),char(116),char(99),char(104),char(0),char(109),
-char(95),char(114),char(105),char(103),char(105),char(100),char(67),char(111),char(110),char(116),char(97),char(99),char(116),char(72),char(97),char(114),char(100),char(110),char(101),char(115),
-char(115),char(0),char(109),char(95),char(107),char(105),char(110),char(101),char(116),char(105),char(99),char(67),char(111),char(110),char(116),char(97),char(99),char(116),char(72),char(97),
-char(114),char(100),char(110),char(101),char(115),char(115),char(0),char(109),char(95),char(115),char(111),char(102),char(116),char(67),char(111),char(110),char(116),char(97),char(99),char(116),
-char(72),char(97),char(114),char(100),char(110),char(101),char(115),char(115),char(0),char(109),char(95),char(97),char(110),char(99),char(104),char(111),char(114),char(72),char(97),char(114),
-char(100),char(110),char(101),char(115),char(115),char(0),char(109),char(95),char(115),char(111),char(102),char(116),char(82),char(105),char(103),char(105),char(100),char(67),char(108),char(117),
-char(115),char(116),char(101),char(114),char(72),char(97),char(114),char(100),char(110),char(101),char(115),char(115),char(0),char(109),char(95),char(115),char(111),char(102),char(116),char(75),
-char(105),char(110),char(101),char(116),char(105),char(99),char(67),char(108),char(117),char(115),char(116),char(101),char(114),char(72),char(97),char(114),char(100),char(110),char(101),char(115),
-char(115),char(0),char(109),char(95),char(115),char(111),char(102),char(116),char(83),char(111),char(102),char(116),char(67),char(108),char(117),char(115),char(116),char(101),char(114),char(72),
-char(97),char(114),char(100),char(110),char(101),char(115),char(115),char(0),char(109),char(95),char(115),char(111),char(102),char(116),char(82),char(105),char(103),char(105),char(100),char(67),
-char(108),char(117),char(115),char(116),char(101),char(114),char(73),char(109),char(112),char(117),char(108),char(115),char(101),char(83),char(112),char(108),char(105),char(116),char(0),char(109),
-char(95),char(115),char(111),char(102),char(116),char(75),char(105),char(110),char(101),char(116),char(105),char(99),char(67),char(108),char(117),char(115),char(116),char(101),char(114),char(73),
-char(109),char(112),char(117),char(108),char(115),char(101),char(83),char(112),char(108),char(105),char(116),char(0),char(109),char(95),char(115),char(111),char(102),char(116),char(83),char(111),
-char(102),char(116),char(67),char(108),char(117),char(115),char(116),char(101),char(114),char(73),char(109),char(112),char(117),char(108),char(115),char(101),char(83),char(112),char(108),char(105),
-char(116),char(0),char(109),char(95),char(109),char(97),char(120),char(86),char(111),char(108),char(117),char(109),char(101),char(0),char(109),char(95),char(116),char(105),char(109),char(101),
-char(83),char(99),char(97),char(108),char(101),char(0),char(109),char(95),char(118),char(101),char(108),char(111),char(99),char(105),char(116),char(121),char(73),char(116),char(101),char(114),
-char(97),char(116),char(105),char(111),char(110),char(115),char(0),char(109),char(95),char(112),char(111),char(115),char(105),char(116),char(105),char(111),char(110),char(73),char(116),char(101),
-char(114),char(97),char(116),char(105),char(111),char(110),char(115),char(0),char(109),char(95),char(100),char(114),char(105),char(102),char(116),char(73),char(116),char(101),char(114),char(97),
-char(116),char(105),char(111),char(110),char(115),char(0),char(109),char(95),char(99),char(108),char(117),char(115),char(116),char(101),char(114),char(73),char(116),char(101),char(114),char(97),
-char(116),char(105),char(111),char(110),char(115),char(0),char(109),char(95),char(114),char(111),char(116),char(0),char(109),char(95),char(115),char(99),char(97),char(108),char(101),char(0),
-char(109),char(95),char(97),char(113),char(113),char(0),char(109),char(95),char(99),char(111),char(109),char(0),char(42),char(109),char(95),char(112),char(111),char(115),char(105),char(116),
-char(105),char(111),char(110),char(115),char(0),char(42),char(109),char(95),char(119),char(101),char(105),char(103),char(104),char(116),char(115),char(0),char(109),char(95),char(110),char(117),
-char(109),char(80),char(111),char(115),char(105),char(116),char(105),char(111),char(110),char(115),char(0),char(109),char(95),char(110),char(117),char(109),char(87),char(101),char(105),char(103),
-char(116),char(115),char(0),char(109),char(95),char(98),char(118),char(111),char(108),char(117),char(109),char(101),char(0),char(109),char(95),char(98),char(102),char(114),char(97),char(109),
-char(101),char(0),char(109),char(95),char(102),char(114),char(97),char(109),char(101),char(120),char(102),char(111),char(114),char(109),char(0),char(109),char(95),char(108),char(111),char(99),
-char(105),char(105),char(0),char(109),char(95),char(105),char(110),char(118),char(119),char(105),char(0),char(109),char(95),char(118),char(105),char(109),char(112),char(117),char(108),char(115),
-char(101),char(115),char(91),char(50),char(93),char(0),char(109),char(95),char(100),char(105),char(109),char(112),char(117),char(108),char(115),char(101),char(115),char(91),char(50),char(93),
-char(0),char(109),char(95),char(108),char(118),char(0),char(109),char(95),char(97),char(118),char(0),char(42),char(109),char(95),char(102),char(114),char(97),char(109),char(101),char(114),
-char(101),char(102),char(115),char(0),char(42),char(109),char(95),char(110),char(111),char(100),char(101),char(73),char(110),char(100),char(105),char(99),char(101),char(115),char(0),char(42),
-char(109),char(95),char(109),char(97),char(115),char(115),char(101),char(115),char(0),char(109),char(95),char(110),char(117),char(109),char(70),char(114),char(97),char(109),char(101),char(82),
-char(101),char(102),char(115),char(0),char(109),char(95),char(110),char(117),char(109),char(78),char(111),char(100),char(101),char(115),char(0),char(109),char(95),char(110),char(117),char(109),
-char(77),char(97),char(115),char(115),char(101),char(115),char(0),char(109),char(95),char(105),char(100),char(109),char(97),char(115),char(115),char(0),char(109),char(95),char(105),char(109),
-char(97),char(115),char(115),char(0),char(109),char(95),char(110),char(118),char(105),char(109),char(112),char(117),char(108),char(115),char(101),char(115),char(0),char(109),char(95),char(110),
-char(100),char(105),char(109),char(112),char(117),char(108),char(115),char(101),char(115),char(0),char(109),char(95),char(110),char(100),char(97),char(109),char(112),char(105),char(110),char(103),
-char(0),char(109),char(95),char(108),char(100),char(97),char(109),char(112),char(105),char(110),char(103),char(0),char(109),char(95),char(97),char(100),char(97),char(109),char(112),char(105),
-char(110),char(103),char(0),char(109),char(95),char(109),char(97),char(116),char(99),char(104),char(105),char(110),char(103),char(0),char(109),char(95),char(109),char(97),char(120),char(83),
-char(101),char(108),char(102),char(67),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(73),char(109),char(112),char(117),char(108),char(115),char(101),char(0),
-char(109),char(95),char(115),char(101),char(108),char(102),char(67),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(73),char(109),char(112),char(117),char(108),
-char(115),char(101),char(70),char(97),char(99),char(116),char(111),char(114),char(0),char(109),char(95),char(99),char(111),char(110),char(116),char(97),char(105),char(110),char(115),char(65),
-char(110),char(99),char(104),char(111),char(114),char(0),char(109),char(95),char(99),char(111),char(108),char(108),char(105),char(100),char(101),char(0),char(109),char(95),char(99),char(108),
-char(117),char(115),char(116),char(101),char(114),char(73),char(110),char(100),char(101),char(120),char(0),char(42),char(109),char(95),char(98),char(111),char(100),char(121),char(65),char(0),
-char(42),char(109),char(95),char(98),char(111),char(100),char(121),char(66),char(0),char(109),char(95),char(114),char(101),char(102),char(115),char(91),char(50),char(93),char(0),char(109),
-char(95),char(99),char(102),char(109),char(0),char(109),char(95),char(115),char(112),char(108),char(105),char(116),char(0),char(109),char(95),char(100),char(101),char(108),char(101),char(116),
-char(101),char(0),char(109),char(95),char(114),char(101),char(108),char(80),char(111),char(115),char(105),char(116),char(105),char(111),char(110),char(91),char(50),char(93),char(0),char(109),
-char(95),char(98),char(111),char(100),char(121),char(65),char(116),char(121),char(112),char(101),char(0),char(109),char(95),char(98),char(111),char(100),char(121),char(66),char(116),char(121),
-char(112),char(101),char(0),char(109),char(95),char(106),char(111),char(105),char(110),char(116),char(84),char(121),char(112),char(101),char(0),char(42),char(109),char(95),char(112),char(111),
-char(115),char(101),char(0),char(42),char(42),char(109),char(95),char(109),char(97),char(116),char(101),char(114),char(105),char(97),char(108),char(115),char(0),char(42),char(109),char(95),
-char(110),char(111),char(100),char(101),char(115),char(0),char(42),char(109),char(95),char(108),char(105),char(110),char(107),char(115),char(0),char(42),char(109),char(95),char(102),char(97),
-char(99),char(101),char(115),char(0),char(42),char(109),char(95),char(116),char(101),char(116),char(114),char(97),char(104),char(101),char(100),char(114),char(97),char(0),char(42),char(109),
-char(95),char(97),char(110),char(99),char(104),char(111),char(114),char(115),char(0),char(42),char(109),char(95),char(99),char(108),char(117),char(115),char(116),char(101),char(114),char(115),
-char(0),char(42),char(109),char(95),char(106),char(111),char(105),char(110),char(116),char(115),char(0),char(109),char(95),char(110),char(117),char(109),char(77),char(97),char(116),char(101),
-char(114),char(105),char(97),char(108),char(115),char(0),char(109),char(95),char(110),char(117),char(109),char(76),char(105),char(110),char(107),char(115),char(0),char(109),char(95),char(110),
-char(117),char(109),char(70),char(97),char(99),char(101),char(115),char(0),char(109),char(95),char(110),char(117),char(109),char(84),char(101),char(116),char(114),char(97),char(104),char(101),
-char(100),char(114),char(97),char(0),char(109),char(95),char(110),char(117),char(109),char(65),char(110),char(99),char(104),char(111),char(114),char(115),char(0),char(109),char(95),char(110),
-char(117),char(109),char(67),char(108),char(117),char(115),char(116),char(101),char(114),char(115),char(0),char(109),char(95),char(110),char(117),char(109),char(74),char(111),char(105),char(110),
-char(116),char(115),char(0),char(109),char(95),char(99),char(111),char(110),char(102),char(105),char(103),char(0),char(84),char(89),char(80),char(69),char(76),char(0),char(0),char(0),
-char(99),char(104),char(97),char(114),char(0),char(117),char(99),char(104),char(97),char(114),char(0),char(115),char(104),char(111),char(114),char(116),char(0),char(117),char(115),char(104),
-char(111),char(114),char(116),char(0),char(105),char(110),char(116),char(0),char(108),char(111),char(110),char(103),char(0),char(117),char(108),char(111),char(110),char(103),char(0),char(102),
-char(108),char(111),char(97),char(116),char(0),char(100),char(111),char(117),char(98),char(108),char(101),char(0),char(118),char(111),char(105),char(100),char(0),char(80),char(111),char(105),
-char(110),char(116),char(101),char(114),char(65),char(114),char(114),char(97),char(121),char(0),char(98),char(116),char(80),char(104),char(121),char(115),char(105),char(99),char(115),char(83),
-char(121),char(115),char(116),char(101),char(109),char(0),char(76),char(105),char(115),char(116),char(66),char(97),char(115),char(101),char(0),char(98),char(116),char(86),char(101),char(99),
-char(116),char(111),char(114),char(51),char(70),char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(86),char(101),char(99),char(116),
-char(111),char(114),char(51),char(68),char(111),char(117),char(98),char(108),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(77),char(97),char(116),char(114),
-char(105),char(120),char(51),char(120),char(51),char(70),char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(77),char(97),char(116),
-char(114),char(105),char(120),char(51),char(120),char(51),char(68),char(111),char(117),char(98),char(108),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(84),
-char(114),char(97),char(110),char(115),char(102),char(111),char(114),char(109),char(70),char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),
-char(84),char(114),char(97),char(110),char(115),char(102),char(111),char(114),char(109),char(68),char(111),char(117),char(98),char(108),char(101),char(68),char(97),char(116),char(97),char(0),
-char(98),char(116),char(66),char(118),char(104),char(83),char(117),char(98),char(116),char(114),char(101),char(101),char(73),char(110),char(102),char(111),char(68),char(97),char(116),char(97),
-char(0),char(98),char(116),char(79),char(112),char(116),char(105),char(109),char(105),char(122),char(101),char(100),char(66),char(118),char(104),char(78),char(111),char(100),char(101),char(70),
-char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(79),char(112),char(116),char(105),char(109),char(105),char(122),char(101),char(100),
-char(66),char(118),char(104),char(78),char(111),char(100),char(101),char(68),char(111),char(117),char(98),char(108),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),
-char(81),char(117),char(97),char(110),char(116),char(105),char(122),char(101),char(100),char(66),char(118),char(104),char(78),char(111),char(100),char(101),char(68),char(97),char(116),char(97),
-char(0),char(98),char(116),char(81),char(117),char(97),char(110),char(116),char(105),char(122),char(101),char(100),char(66),char(118),char(104),char(70),char(108),char(111),char(97),char(116),
-char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(81),char(117),char(97),char(110),char(116),char(105),char(122),char(101),char(100),char(66),char(118),char(104),char(68),
-char(111),char(117),char(98),char(108),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(111),char(108),char(108),char(105),char(115),char(105),char(111),
-char(110),char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(83),char(116),char(97),char(116),char(105),char(99),char(80),
-char(108),char(97),char(110),char(101),char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(111),char(110),char(118),
-char(101),char(120),char(73),char(110),char(116),char(101),char(114),char(110),char(97),char(108),char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),
-char(98),char(116),char(80),char(111),char(115),char(105),char(116),char(105),char(111),char(110),char(65),char(110),char(100),char(82),char(97),char(100),char(105),char(117),char(115),char(0),
-char(98),char(116),char(77),char(117),char(108),char(116),char(105),char(83),char(112),char(104),char(101),char(114),char(101),char(83),char(104),char(97),char(112),char(101),char(68),char(97),
-char(116),char(97),char(0),char(98),char(116),char(73),char(110),char(116),char(73),char(110),char(100),char(101),char(120),char(68),char(97),char(116),char(97),char(0),char(98),char(116),
-char(83),char(104),char(111),char(114),char(116),char(73),char(110),char(116),char(73),char(110),char(100),char(101),char(120),char(68),char(97),char(116),char(97),char(0),char(98),char(116),
-char(83),char(104),char(111),char(114),char(116),char(73),char(110),char(116),char(73),char(110),char(100),char(101),char(120),char(84),char(114),char(105),char(112),char(108),char(101),char(116),
-char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(104),char(97),char(114),char(73),char(110),char(100),char(101),char(120),char(84),char(114),char(105),char(112),
-char(108),char(101),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(77),char(101),char(115),char(104),char(80),char(97),char(114),char(116),char(68),char(97),
-char(116),char(97),char(0),char(98),char(116),char(83),char(116),char(114),char(105),char(100),char(105),char(110),char(103),char(77),char(101),char(115),char(104),char(73),char(110),char(116),
-char(101),char(114),char(102),char(97),char(99),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(84),char(114),char(105),char(97),char(110),char(103),char(108),
-char(101),char(77),char(101),char(115),char(104),char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(84),char(114),char(105),
-char(97),char(110),char(103),char(108),char(101),char(73),char(110),char(102),char(111),char(77),char(97),char(112),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(83),
-char(99),char(97),char(108),char(101),char(100),char(84),char(114),char(105),char(97),char(110),char(103),char(108),char(101),char(77),char(101),char(115),char(104),char(83),char(104),char(97),
-char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(111),char(109),char(112),char(111),char(117),char(110),char(100),char(83),char(104),char(97),
-char(112),char(101),char(67),char(104),char(105),char(108),char(100),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(111),char(109),char(112),char(111),char(117),
-char(110),char(100),char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(121),char(108),char(105),char(110),char(100),
-char(101),char(114),char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(97),char(112),char(115),char(117),char(108),
-char(101),char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(84),char(114),char(105),char(97),char(110),char(103),char(108),
-char(101),char(73),char(110),char(102),char(111),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(71),char(73),char(109),char(112),char(97),char(99),char(116),char(77),
-char(101),char(115),char(104),char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(111),char(110),char(118),char(101),
-char(120),char(72),char(117),char(108),char(108),char(83),char(104),char(97),char(112),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(111),char(108),
-char(108),char(105),char(115),char(105),char(111),char(110),char(79),char(98),char(106),char(101),char(99),char(116),char(68),char(111),char(117),char(98),char(108),char(101),char(68),char(97),
-char(116),char(97),char(0),char(98),char(116),char(67),char(111),char(108),char(108),char(105),char(115),char(105),char(111),char(110),char(79),char(98),char(106),char(101),char(99),char(116),
-char(70),char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(68),char(121),char(110),char(97),char(109),char(105),char(99),char(115),
-char(87),char(111),char(114),char(108),char(100),char(68),char(111),char(117),char(98),char(108),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(111),
-char(110),char(116),char(97),char(99),char(116),char(83),char(111),char(108),char(118),char(101),char(114),char(73),char(110),char(102),char(111),char(68),char(111),char(117),char(98),char(108),
-char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(68),char(121),char(110),char(97),char(109),char(105),char(99),char(115),char(87),char(111),char(114),char(108),
-char(100),char(70),char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(67),char(111),char(110),char(116),char(97),char(99),char(116),
-char(83),char(111),char(108),char(118),char(101),char(114),char(73),char(110),char(102),char(111),char(70),char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),char(0),
-char(98),char(116),char(82),char(105),char(103),char(105),char(100),char(66),char(111),char(100),char(121),char(70),char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),
-char(0),char(98),char(116),char(82),char(105),char(103),char(105),char(100),char(66),char(111),char(100),char(121),char(68),char(111),char(117),char(98),char(108),char(101),char(68),char(97),
-char(116),char(97),char(0),char(98),char(116),char(67),char(111),char(110),char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(73),char(110),char(102),char(111),char(49),
-char(0),char(98),char(116),char(84),char(121),char(112),char(101),char(100),char(67),char(111),char(110),char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(68),char(97),
-char(116),char(97),char(0),char(98),char(116),char(82),char(105),char(103),char(105),char(100),char(66),char(111),char(100),char(121),char(68),char(97),char(116),char(97),char(0),char(98),
-char(116),char(80),char(111),char(105),char(110),char(116),char(50),char(80),char(111),char(105),char(110),char(116),char(67),char(111),char(110),char(115),char(116),char(114),char(97),char(105),
-char(110),char(116),char(70),char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(80),char(111),char(105),char(110),char(116),char(50),
-char(80),char(111),char(105),char(110),char(116),char(67),char(111),char(110),char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(68),char(111),char(117),char(98),char(108),
-char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(72),char(105),char(110),char(103),char(101),char(67),char(111),char(110),char(115),char(116),char(114),char(97),
-char(105),char(110),char(116),char(68),char(111),char(117),char(98),char(108),char(101),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(72),char(105),char(110),char(103),
-char(101),char(67),char(111),char(110),char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(70),char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),
-char(0),char(98),char(116),char(67),char(111),char(110),char(101),char(84),char(119),char(105),char(115),char(116),char(67),char(111),char(110),char(115),char(116),char(114),char(97),char(105),
-char(110),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(71),char(101),char(110),char(101),char(114),char(105),char(99),char(54),char(68),char(111),char(102),
-char(67),char(111),char(110),char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(71),char(101),char(110),
-char(101),char(114),char(105),char(99),char(54),char(68),char(111),char(102),char(83),char(112),char(114),char(105),char(110),char(103),char(67),char(111),char(110),char(115),char(116),char(114),
-char(97),char(105),char(110),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),char(83),char(108),char(105),char(100),char(101),char(114),char(67),char(111),char(110),
-char(115),char(116),char(114),char(97),char(105),char(110),char(116),char(68),char(97),char(116),char(97),char(0),char(83),char(111),char(102),char(116),char(66),char(111),char(100),char(121),
-char(77),char(97),char(116),char(101),char(114),char(105),char(97),char(108),char(68),char(97),char(116),char(97),char(0),char(83),char(111),char(102),char(116),char(66),char(111),char(100),
-char(121),char(78),char(111),char(100),char(101),char(68),char(97),char(116),char(97),char(0),char(83),char(111),char(102),char(116),char(66),char(111),char(100),char(121),char(76),char(105),
-char(110),char(107),char(68),char(97),char(116),char(97),char(0),char(83),char(111),char(102),char(116),char(66),char(111),char(100),char(121),char(70),char(97),char(99),char(101),char(68),
-char(97),char(116),char(97),char(0),char(83),char(111),char(102),char(116),char(66),char(111),char(100),char(121),char(84),char(101),char(116),char(114),char(97),char(68),char(97),char(116),
-char(97),char(0),char(83),char(111),char(102),char(116),char(82),char(105),char(103),char(105),char(100),char(65),char(110),char(99),char(104),char(111),char(114),char(68),char(97),char(116),
-char(97),char(0),char(83),char(111),char(102),char(116),char(66),char(111),char(100),char(121),char(67),char(111),char(110),char(102),char(105),char(103),char(68),char(97),char(116),char(97),
-char(0),char(83),char(111),char(102),char(116),char(66),char(111),char(100),char(121),char(80),char(111),char(115),char(101),char(68),char(97),char(116),char(97),char(0),char(83),char(111),
-char(102),char(116),char(66),char(111),char(100),char(121),char(67),char(108),char(117),char(115),char(116),char(101),char(114),char(68),char(97),char(116),char(97),char(0),char(98),char(116),
-char(83),char(111),char(102),char(116),char(66),char(111),char(100),char(121),char(74),char(111),char(105),char(110),char(116),char(68),char(97),char(116),char(97),char(0),char(98),char(116),
-char(83),char(111),char(102),char(116),char(66),char(111),char(100),char(121),char(70),char(108),char(111),char(97),char(116),char(68),char(97),char(116),char(97),char(0),char(0),char(0),
-char(84),char(76),char(69),char(78),char(1),char(0),char(1),char(0),char(2),char(0),char(2),char(0),char(4),char(0),char(4),char(0),char(4),char(0),char(4),char(0),
-char(8),char(0),char(0),char(0),char(16),char(0),char(48),char(0),char(16),char(0),char(16),char(0),char(32),char(0),char(48),char(0),char(96),char(0),char(64),char(0),
-char(-128),char(0),char(20),char(0),char(48),char(0),char(80),char(0),char(16),char(0),char(96),char(0),char(-112),char(0),char(16),char(0),char(56),char(0),char(56),char(0),
-char(20),char(0),char(72),char(0),char(4),char(0),char(4),char(0),char(8),char(0),char(4),char(0),char(56),char(0),char(32),char(0),char(80),char(0),char(72),char(0),
-char(96),char(0),char(80),char(0),char(32),char(0),char(64),char(0),char(64),char(0),char(16),char(0),char(72),char(0),char(80),char(0),char(-32),char(1),char(16),char(1),
-char(-72),char(0),char(-104),char(0),char(104),char(0),char(88),char(0),char(-8),char(1),char(-80),char(3),char(8),char(0),char(64),char(0),char(0),char(0),char(96),char(0),
-char(-128),char(0),char(104),char(1),char(-24),char(0),char(-32),char(0),char(8),char(1),char(104),char(1),char(-40),char(0),char(16),char(0),char(104),char(0),char(24),char(0),
-char(40),char(0),char(104),char(0),char(96),char(0),char(104),char(0),char(-56),char(0),char(104),char(1),char(112),char(0),char(-32),char(1),char(83),char(84),char(82),char(67),
-char(65),char(0),char(0),char(0),char(10),char(0),char(3),char(0),char(4),char(0),char(0),char(0),char(4),char(0),char(1),char(0),char(9),char(0),char(2),char(0),
-char(11),char(0),char(3),char(0),char(10),char(0),char(3),char(0),char(10),char(0),char(4),char(0),char(10),char(0),char(5),char(0),char(12),char(0),char(2),char(0),
-char(9),char(0),char(6),char(0),char(9),char(0),char(7),char(0),char(13),char(0),char(1),char(0),char(7),char(0),char(8),char(0),char(14),char(0),char(1),char(0),
-char(8),char(0),char(8),char(0),char(15),char(0),char(1),char(0),char(13),char(0),char(9),char(0),char(16),char(0),char(1),char(0),char(14),char(0),char(9),char(0),
-char(17),char(0),char(2),char(0),char(15),char(0),char(10),char(0),char(13),char(0),char(11),char(0),char(18),char(0),char(2),char(0),char(16),char(0),char(10),char(0),
-char(14),char(0),char(11),char(0),char(19),char(0),char(4),char(0),char(4),char(0),char(12),char(0),char(4),char(0),char(13),char(0),char(2),char(0),char(14),char(0),
-char(2),char(0),char(15),char(0),char(20),char(0),char(6),char(0),char(13),char(0),char(16),char(0),char(13),char(0),char(17),char(0),char(4),char(0),char(18),char(0),
-char(4),char(0),char(19),char(0),char(4),char(0),char(20),char(0),char(0),char(0),char(21),char(0),char(21),char(0),char(6),char(0),char(14),char(0),char(16),char(0),
-char(14),char(0),char(17),char(0),char(4),char(0),char(18),char(0),char(4),char(0),char(19),char(0),char(4),char(0),char(20),char(0),char(0),char(0),char(21),char(0),
-char(22),char(0),char(3),char(0),char(2),char(0),char(14),char(0),char(2),char(0),char(15),char(0),char(4),char(0),char(22),char(0),char(23),char(0),char(12),char(0),
-char(13),char(0),char(23),char(0),char(13),char(0),char(24),char(0),char(13),char(0),char(25),char(0),char(4),char(0),char(26),char(0),char(4),char(0),char(27),char(0),
-char(4),char(0),char(28),char(0),char(4),char(0),char(29),char(0),char(20),char(0),char(30),char(0),char(22),char(0),char(31),char(0),char(19),char(0),char(32),char(0),
-char(4),char(0),char(33),char(0),char(4),char(0),char(34),char(0),char(24),char(0),char(12),char(0),char(14),char(0),char(23),char(0),char(14),char(0),char(24),char(0),
-char(14),char(0),char(25),char(0),char(4),char(0),char(26),char(0),char(4),char(0),char(27),char(0),char(4),char(0),char(28),char(0),char(4),char(0),char(29),char(0),
-char(21),char(0),char(30),char(0),char(22),char(0),char(31),char(0),char(4),char(0),char(33),char(0),char(4),char(0),char(34),char(0),char(19),char(0),char(32),char(0),
-char(25),char(0),char(3),char(0),char(0),char(0),char(35),char(0),char(4),char(0),char(36),char(0),char(0),char(0),char(37),char(0),char(26),char(0),char(5),char(0),
-char(25),char(0),char(38),char(0),char(13),char(0),char(39),char(0),char(13),char(0),char(40),char(0),char(7),char(0),char(41),char(0),char(0),char(0),char(21),char(0),
-char(27),char(0),char(5),char(0),char(25),char(0),char(38),char(0),char(13),char(0),char(39),char(0),char(13),char(0),char(42),char(0),char(7),char(0),char(43),char(0),
-char(4),char(0),char(44),char(0),char(28),char(0),char(2),char(0),char(13),char(0),char(45),char(0),char(7),char(0),char(46),char(0),char(29),char(0),char(4),char(0),
-char(27),char(0),char(47),char(0),char(28),char(0),char(48),char(0),char(4),char(0),char(49),char(0),char(0),char(0),char(37),char(0),char(30),char(0),char(1),char(0),
-char(4),char(0),char(50),char(0),char(31),char(0),char(2),char(0),char(2),char(0),char(50),char(0),char(0),char(0),char(51),char(0),char(32),char(0),char(2),char(0),
-char(2),char(0),char(52),char(0),char(0),char(0),char(51),char(0),char(33),char(0),char(2),char(0),char(0),char(0),char(52),char(0),char(0),char(0),char(53),char(0),
-char(34),char(0),char(8),char(0),char(13),char(0),char(54),char(0),char(14),char(0),char(55),char(0),char(30),char(0),char(56),char(0),char(32),char(0),char(57),char(0),
-char(33),char(0),char(58),char(0),char(31),char(0),char(59),char(0),char(4),char(0),char(60),char(0),char(4),char(0),char(61),char(0),char(35),char(0),char(4),char(0),
-char(34),char(0),char(62),char(0),char(13),char(0),char(63),char(0),char(4),char(0),char(64),char(0),char(0),char(0),char(37),char(0),char(36),char(0),char(7),char(0),
-char(25),char(0),char(38),char(0),char(35),char(0),char(65),char(0),char(23),char(0),char(66),char(0),char(24),char(0),char(67),char(0),char(37),char(0),char(68),char(0),
-char(7),char(0),char(43),char(0),char(0),char(0),char(69),char(0),char(38),char(0),char(2),char(0),char(36),char(0),char(70),char(0),char(13),char(0),char(39),char(0),
-char(39),char(0),char(4),char(0),char(17),char(0),char(71),char(0),char(25),char(0),char(72),char(0),char(4),char(0),char(73),char(0),char(7),char(0),char(74),char(0),
-char(40),char(0),char(4),char(0),char(25),char(0),char(38),char(0),char(39),char(0),char(75),char(0),char(4),char(0),char(76),char(0),char(7),char(0),char(43),char(0),
-char(41),char(0),char(3),char(0),char(27),char(0),char(47),char(0),char(4),char(0),char(77),char(0),char(0),char(0),char(37),char(0),char(42),char(0),char(3),char(0),
-char(27),char(0),char(47),char(0),char(4),char(0),char(77),char(0),char(0),char(0),char(37),char(0),char(43),char(0),char(4),char(0),char(4),char(0),char(78),char(0),
-char(7),char(0),char(79),char(0),char(7),char(0),char(80),char(0),char(7),char(0),char(81),char(0),char(37),char(0),char(14),char(0),char(4),char(0),char(82),char(0),
-char(4),char(0),char(83),char(0),char(43),char(0),char(84),char(0),char(4),char(0),char(85),char(0),char(7),char(0),char(86),char(0),char(7),char(0),char(87),char(0),
-char(7),char(0),char(88),char(0),char(7),char(0),char(89),char(0),char(7),char(0),char(90),char(0),char(4),char(0),char(91),char(0),char(4),char(0),char(92),char(0),
-char(4),char(0),char(93),char(0),char(4),char(0),char(94),char(0),char(0),char(0),char(37),char(0),char(44),char(0),char(5),char(0),char(25),char(0),char(38),char(0),
-char(35),char(0),char(65),char(0),char(13),char(0),char(39),char(0),char(7),char(0),char(43),char(0),char(4),char(0),char(95),char(0),char(45),char(0),char(5),char(0),
-char(27),char(0),char(47),char(0),char(13),char(0),char(96),char(0),char(14),char(0),char(97),char(0),char(4),char(0),char(98),char(0),char(0),char(0),char(99),char(0),
-char(46),char(0),char(25),char(0),char(9),char(0),char(100),char(0),char(9),char(0),char(101),char(0),char(25),char(0),char(102),char(0),char(0),char(0),char(35),char(0),
-char(18),char(0),char(103),char(0),char(18),char(0),char(104),char(0),char(14),char(0),char(105),char(0),char(14),char(0),char(106),char(0),char(14),char(0),char(107),char(0),
-char(8),char(0),char(108),char(0),char(8),char(0),char(109),char(0),char(8),char(0),char(110),char(0),char(8),char(0),char(111),char(0),char(8),char(0),char(112),char(0),
-char(8),char(0),char(113),char(0),char(8),char(0),char(114),char(0),char(8),char(0),char(115),char(0),char(4),char(0),char(116),char(0),char(4),char(0),char(117),char(0),
-char(4),char(0),char(118),char(0),char(4),char(0),char(119),char(0),char(4),char(0),char(120),char(0),char(4),char(0),char(121),char(0),char(4),char(0),char(122),char(0),
-char(0),char(0),char(37),char(0),char(47),char(0),char(25),char(0),char(9),char(0),char(100),char(0),char(9),char(0),char(101),char(0),char(25),char(0),char(102),char(0),
-char(0),char(0),char(35),char(0),char(17),char(0),char(103),char(0),char(17),char(0),char(104),char(0),char(13),char(0),char(105),char(0),char(13),char(0),char(106),char(0),
-char(13),char(0),char(107),char(0),char(7),char(0),char(108),char(0),char(7),char(0),char(109),char(0),char(7),char(0),char(110),char(0),char(7),char(0),char(111),char(0),
-char(7),char(0),char(112),char(0),char(7),char(0),char(113),char(0),char(7),char(0),char(114),char(0),char(7),char(0),char(115),char(0),char(4),char(0),char(116),char(0),
-char(4),char(0),char(117),char(0),char(4),char(0),char(118),char(0),char(4),char(0),char(119),char(0),char(4),char(0),char(120),char(0),char(4),char(0),char(121),char(0),
-char(4),char(0),char(122),char(0),char(0),char(0),char(37),char(0),char(48),char(0),char(2),char(0),char(49),char(0),char(123),char(0),char(14),char(0),char(124),char(0),
-char(50),char(0),char(2),char(0),char(51),char(0),char(123),char(0),char(13),char(0),char(124),char(0),char(52),char(0),char(21),char(0),char(47),char(0),char(125),char(0),
-char(15),char(0),char(126),char(0),char(13),char(0),char(127),char(0),char(13),char(0),char(-128),char(0),char(13),char(0),char(-127),char(0),char(13),char(0),char(-126),char(0),
-char(13),char(0),char(124),char(0),char(13),char(0),char(-125),char(0),char(13),char(0),char(-124),char(0),char(13),char(0),char(-123),char(0),char(13),char(0),char(-122),char(0),
-char(7),char(0),char(-121),char(0),char(7),char(0),char(-120),char(0),char(7),char(0),char(-119),char(0),char(7),char(0),char(-118),char(0),char(7),char(0),char(-117),char(0),
-char(7),char(0),char(-116),char(0),char(7),char(0),char(-115),char(0),char(7),char(0),char(-114),char(0),char(7),char(0),char(-113),char(0),char(4),char(0),char(-112),char(0),
-char(53),char(0),char(22),char(0),char(46),char(0),char(125),char(0),char(16),char(0),char(126),char(0),char(14),char(0),char(127),char(0),char(14),char(0),char(-128),char(0),
-char(14),char(0),char(-127),char(0),char(14),char(0),char(-126),char(0),char(14),char(0),char(124),char(0),char(14),char(0),char(-125),char(0),char(14),char(0),char(-124),char(0),
-char(14),char(0),char(-123),char(0),char(14),char(0),char(-122),char(0),char(8),char(0),char(-121),char(0),char(8),char(0),char(-120),char(0),char(8),char(0),char(-119),char(0),
-char(8),char(0),char(-118),char(0),char(8),char(0),char(-117),char(0),char(8),char(0),char(-116),char(0),char(8),char(0),char(-115),char(0),char(8),char(0),char(-114),char(0),
-char(8),char(0),char(-113),char(0),char(4),char(0),char(-112),char(0),char(0),char(0),char(37),char(0),char(54),char(0),char(2),char(0),char(4),char(0),char(-111),char(0),
-char(4),char(0),char(-110),char(0),char(55),char(0),char(13),char(0),char(56),char(0),char(-109),char(0),char(56),char(0),char(-108),char(0),char(0),char(0),char(35),char(0),
-char(4),char(0),char(-107),char(0),char(4),char(0),char(-106),char(0),char(4),char(0),char(-105),char(0),char(4),char(0),char(-104),char(0),char(7),char(0),char(-103),char(0),
-char(7),char(0),char(-102),char(0),char(4),char(0),char(-101),char(0),char(4),char(0),char(-100),char(0),char(7),char(0),char(-99),char(0),char(4),char(0),char(-98),char(0),
-char(57),char(0),char(3),char(0),char(55),char(0),char(-97),char(0),char(13),char(0),char(-96),char(0),char(13),char(0),char(-95),char(0),char(58),char(0),char(3),char(0),
-char(55),char(0),char(-97),char(0),char(14),char(0),char(-96),char(0),char(14),char(0),char(-95),char(0),char(59),char(0),char(13),char(0),char(55),char(0),char(-97),char(0),
-char(18),char(0),char(-94),char(0),char(18),char(0),char(-93),char(0),char(4),char(0),char(-92),char(0),char(4),char(0),char(-91),char(0),char(4),char(0),char(-90),char(0),
-char(7),char(0),char(-89),char(0),char(7),char(0),char(-88),char(0),char(7),char(0),char(-87),char(0),char(7),char(0),char(-86),char(0),char(7),char(0),char(-85),char(0),
-char(7),char(0),char(-84),char(0),char(7),char(0),char(-83),char(0),char(60),char(0),char(13),char(0),char(55),char(0),char(-97),char(0),char(17),char(0),char(-94),char(0),
-char(17),char(0),char(-93),char(0),char(4),char(0),char(-92),char(0),char(4),char(0),char(-91),char(0),char(4),char(0),char(-90),char(0),char(7),char(0),char(-89),char(0),
-char(7),char(0),char(-88),char(0),char(7),char(0),char(-87),char(0),char(7),char(0),char(-86),char(0),char(7),char(0),char(-85),char(0),char(7),char(0),char(-84),char(0),
-char(7),char(0),char(-83),char(0),char(61),char(0),char(11),char(0),char(55),char(0),char(-97),char(0),char(17),char(0),char(-94),char(0),char(17),char(0),char(-93),char(0),
-char(7),char(0),char(-82),char(0),char(7),char(0),char(-81),char(0),char(7),char(0),char(-80),char(0),char(7),char(0),char(-85),char(0),char(7),char(0),char(-84),char(0),
-char(7),char(0),char(-83),char(0),char(7),char(0),char(-79),char(0),char(0),char(0),char(21),char(0),char(62),char(0),char(9),char(0),char(55),char(0),char(-97),char(0),
-char(17),char(0),char(-94),char(0),char(17),char(0),char(-93),char(0),char(13),char(0),char(-78),char(0),char(13),char(0),char(-77),char(0),char(13),char(0),char(-76),char(0),
-char(13),char(0),char(-75),char(0),char(4),char(0),char(-74),char(0),char(4),char(0),char(-73),char(0),char(63),char(0),char(5),char(0),char(62),char(0),char(-72),char(0),
-char(4),char(0),char(-71),char(0),char(7),char(0),char(-70),char(0),char(7),char(0),char(-69),char(0),char(7),char(0),char(-68),char(0),char(64),char(0),char(9),char(0),
-char(55),char(0),char(-97),char(0),char(17),char(0),char(-94),char(0),char(17),char(0),char(-93),char(0),char(7),char(0),char(-78),char(0),char(7),char(0),char(-77),char(0),
-char(7),char(0),char(-76),char(0),char(7),char(0),char(-75),char(0),char(4),char(0),char(-74),char(0),char(4),char(0),char(-73),char(0),char(49),char(0),char(22),char(0),
-char(8),char(0),char(-67),char(0),char(8),char(0),char(-79),char(0),char(8),char(0),char(110),char(0),char(8),char(0),char(-66),char(0),char(8),char(0),char(112),char(0),
-char(8),char(0),char(-65),char(0),char(8),char(0),char(-64),char(0),char(8),char(0),char(-63),char(0),char(8),char(0),char(-62),char(0),char(8),char(0),char(-61),char(0),
-char(8),char(0),char(-60),char(0),char(8),char(0),char(-59),char(0),char(8),char(0),char(-58),char(0),char(8),char(0),char(-57),char(0),char(8),char(0),char(-56),char(0),
-char(8),char(0),char(-55),char(0),char(4),char(0),char(-54),char(0),char(4),char(0),char(-53),char(0),char(4),char(0),char(-52),char(0),char(4),char(0),char(-51),char(0),
-char(4),char(0),char(-50),char(0),char(0),char(0),char(37),char(0),char(51),char(0),char(22),char(0),char(7),char(0),char(-67),char(0),char(7),char(0),char(-79),char(0),
-char(7),char(0),char(110),char(0),char(7),char(0),char(-66),char(0),char(7),char(0),char(112),char(0),char(7),char(0),char(-65),char(0),char(7),char(0),char(-64),char(0),
-char(7),char(0),char(-63),char(0),char(7),char(0),char(-62),char(0),char(7),char(0),char(-61),char(0),char(7),char(0),char(-60),char(0),char(7),char(0),char(-59),char(0),
-char(7),char(0),char(-58),char(0),char(7),char(0),char(-57),char(0),char(7),char(0),char(-56),char(0),char(7),char(0),char(-55),char(0),char(4),char(0),char(-54),char(0),
-char(4),char(0),char(-53),char(0),char(4),char(0),char(-52),char(0),char(4),char(0),char(-51),char(0),char(4),char(0),char(-50),char(0),char(0),char(0),char(37),char(0),
-char(65),char(0),char(4),char(0),char(7),char(0),char(-49),char(0),char(7),char(0),char(-48),char(0),char(7),char(0),char(-47),char(0),char(4),char(0),char(78),char(0),
-char(66),char(0),char(10),char(0),char(65),char(0),char(-46),char(0),char(13),char(0),char(-45),char(0),char(13),char(0),char(-44),char(0),char(13),char(0),char(-43),char(0),
-char(13),char(0),char(-42),char(0),char(13),char(0),char(-41),char(0),char(7),char(0),char(-121),char(0),char(7),char(0),char(-40),char(0),char(4),char(0),char(-39),char(0),
-char(4),char(0),char(53),char(0),char(67),char(0),char(4),char(0),char(65),char(0),char(-46),char(0),char(4),char(0),char(-38),char(0),char(7),char(0),char(-37),char(0),
-char(4),char(0),char(-36),char(0),char(68),char(0),char(4),char(0),char(13),char(0),char(-41),char(0),char(65),char(0),char(-46),char(0),char(4),char(0),char(-35),char(0),
-char(7),char(0),char(-34),char(0),char(69),char(0),char(7),char(0),char(13),char(0),char(-33),char(0),char(65),char(0),char(-46),char(0),char(4),char(0),char(-32),char(0),
-char(7),char(0),char(-31),char(0),char(7),char(0),char(-30),char(0),char(7),char(0),char(-29),char(0),char(4),char(0),char(53),char(0),char(70),char(0),char(6),char(0),
-char(15),char(0),char(-28),char(0),char(13),char(0),char(-30),char(0),char(13),char(0),char(-27),char(0),char(56),char(0),char(-26),char(0),char(4),char(0),char(-25),char(0),
-char(7),char(0),char(-29),char(0),char(71),char(0),char(26),char(0),char(4),char(0),char(-24),char(0),char(7),char(0),char(-23),char(0),char(7),char(0),char(-79),char(0),
-char(7),char(0),char(-22),char(0),char(7),char(0),char(-21),char(0),char(7),char(0),char(-20),char(0),char(7),char(0),char(-19),char(0),char(7),char(0),char(-18),char(0),
-char(7),char(0),char(-17),char(0),char(7),char(0),char(-16),char(0),char(7),char(0),char(-15),char(0),char(7),char(0),char(-14),char(0),char(7),char(0),char(-13),char(0),
-char(7),char(0),char(-12),char(0),char(7),char(0),char(-11),char(0),char(7),char(0),char(-10),char(0),char(7),char(0),char(-9),char(0),char(7),char(0),char(-8),char(0),
-char(7),char(0),char(-7),char(0),char(7),char(0),char(-6),char(0),char(7),char(0),char(-5),char(0),char(4),char(0),char(-4),char(0),char(4),char(0),char(-3),char(0),
-char(4),char(0),char(-2),char(0),char(4),char(0),char(-1),char(0),char(4),char(0),char(117),char(0),char(72),char(0),char(12),char(0),char(15),char(0),char(0),char(1),
-char(15),char(0),char(1),char(1),char(15),char(0),char(2),char(1),char(13),char(0),char(3),char(1),char(13),char(0),char(4),char(1),char(7),char(0),char(5),char(1),
-char(4),char(0),char(6),char(1),char(4),char(0),char(7),char(1),char(4),char(0),char(8),char(1),char(4),char(0),char(9),char(1),char(7),char(0),char(-31),char(0),
-char(4),char(0),char(53),char(0),char(73),char(0),char(27),char(0),char(17),char(0),char(10),char(1),char(15),char(0),char(11),char(1),char(15),char(0),char(12),char(1),
-char(13),char(0),char(3),char(1),char(13),char(0),char(13),char(1),char(13),char(0),char(14),char(1),char(13),char(0),char(15),char(1),char(13),char(0),char(16),char(1),
-char(13),char(0),char(17),char(1),char(4),char(0),char(18),char(1),char(7),char(0),char(19),char(1),char(4),char(0),char(20),char(1),char(4),char(0),char(21),char(1),
-char(4),char(0),char(22),char(1),char(7),char(0),char(23),char(1),char(7),char(0),char(24),char(1),char(4),char(0),char(25),char(1),char(4),char(0),char(26),char(1),
-char(7),char(0),char(27),char(1),char(7),char(0),char(28),char(1),char(7),char(0),char(29),char(1),char(7),char(0),char(30),char(1),char(7),char(0),char(31),char(1),
-char(7),char(0),char(32),char(1),char(4),char(0),char(33),char(1),char(4),char(0),char(34),char(1),char(4),char(0),char(35),char(1),char(74),char(0),char(12),char(0),
-char(9),char(0),char(36),char(1),char(9),char(0),char(37),char(1),char(13),char(0),char(38),char(1),char(7),char(0),char(39),char(1),char(7),char(0),char(-63),char(0),
-char(7),char(0),char(40),char(1),char(4),char(0),char(41),char(1),char(13),char(0),char(42),char(1),char(4),char(0),char(43),char(1),char(4),char(0),char(44),char(1),
-char(4),char(0),char(45),char(1),char(4),char(0),char(53),char(0),char(75),char(0),char(19),char(0),char(47),char(0),char(125),char(0),char(72),char(0),char(46),char(1),
-char(65),char(0),char(47),char(1),char(66),char(0),char(48),char(1),char(67),char(0),char(49),char(1),char(68),char(0),char(50),char(1),char(69),char(0),char(51),char(1),
-char(70),char(0),char(52),char(1),char(73),char(0),char(53),char(1),char(74),char(0),char(54),char(1),char(4),char(0),char(55),char(1),char(4),char(0),char(21),char(1),
-char(4),char(0),char(56),char(1),char(4),char(0),char(57),char(1),char(4),char(0),char(58),char(1),char(4),char(0),char(59),char(1),char(4),char(0),char(60),char(1),
-char(4),char(0),char(61),char(1),char(71),char(0),char(62),char(1),};
-int b3s_bulletDNAlen64= sizeof(b3s_bulletDNAstr64);
+char b3s_bulletDNAstr[] = {
+	char(83),
+	char(68),
+	char(78),
+	char(65),
+	char(78),
+	char(65),
+	char(77),
+	char(69),
+	char(63),
+	char(1),
+	char(0),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(105),
+	char(122),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(97),
+	char(112),
+	char(97),
+	char(99),
+	char(105),
+	char(116),
+	char(121),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(100),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(79),
+	char(98),
+	char(106),
+	char(101),
+	char(99),
+	char(116),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(115),
+	char(0),
+	char(42),
+	char(102),
+	char(105),
+	char(114),
+	char(115),
+	char(116),
+	char(0),
+	char(42),
+	char(108),
+	char(97),
+	char(115),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(102),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(115),
+	char(91),
+	char(52),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(108),
+	char(91),
+	char(51),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(97),
+	char(115),
+	char(105),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(111),
+	char(114),
+	char(105),
+	char(103),
+	char(105),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(111),
+	char(111),
+	char(116),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(117),
+	char(98),
+	char(116),
+	char(114),
+	char(101),
+	char(101),
+	char(83),
+	char(105),
+	char(122),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(113),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(65),
+	char(97),
+	char(98),
+	char(98),
+	char(77),
+	char(105),
+	char(110),
+	char(91),
+	char(51),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(113),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(65),
+	char(97),
+	char(98),
+	char(98),
+	char(77),
+	char(97),
+	char(120),
+	char(91),
+	char(51),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(97),
+	char(98),
+	char(98),
+	char(77),
+	char(105),
+	char(110),
+	char(79),
+	char(114),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(97),
+	char(98),
+	char(98),
+	char(77),
+	char(97),
+	char(120),
+	char(79),
+	char(114),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(115),
+	char(99),
+	char(97),
+	char(112),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(117),
+	char(98),
+	char(80),
+	char(97),
+	char(114),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(114),
+	char(105),
+	char(97),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(97),
+	char(100),
+	char(91),
+	char(52),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(115),
+	char(99),
+	char(97),
+	char(112),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(79),
+	char(114),
+	char(84),
+	char(114),
+	char(105),
+	char(97),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(118),
+	char(104),
+	char(65),
+	char(97),
+	char(98),
+	char(98),
+	char(77),
+	char(105),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(118),
+	char(104),
+	char(65),
+	char(97),
+	char(98),
+	char(98),
+	char(77),
+	char(97),
+	char(120),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(118),
+	char(104),
+	char(81),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(117),
+	char(114),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(0),
+	char(109),
+	char(95),
+	char(117),
+	char(115),
+	char(101),
+	char(81),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(67),
+	char(111),
+	char(110),
+	char(116),
+	char(105),
+	char(103),
+	char(117),
+	char(111),
+	char(117),
+	char(115),
+	char(76),
+	char(101),
+	char(97),
+	char(102),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(81),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(67),
+	char(111),
+	char(110),
+	char(116),
+	char(105),
+	char(103),
+	char(117),
+	char(111),
+	char(117),
+	char(115),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(110),
+	char(116),
+	char(105),
+	char(103),
+	char(117),
+	char(111),
+	char(117),
+	char(115),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(115),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(113),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(67),
+	char(111),
+	char(110),
+	char(116),
+	char(105),
+	char(103),
+	char(117),
+	char(111),
+	char(117),
+	char(115),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(115),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(115),
+	char(117),
+	char(98),
+	char(84),
+	char(114),
+	char(101),
+	char(101),
+	char(73),
+	char(110),
+	char(102),
+	char(111),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(114),
+	char(97),
+	char(118),
+	char(101),
+	char(114),
+	char(115),
+	char(97),
+	char(108),
+	char(77),
+	char(111),
+	char(100),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(83),
+	char(117),
+	char(98),
+	char(116),
+	char(114),
+	char(101),
+	char(101),
+	char(72),
+	char(101),
+	char(97),
+	char(100),
+	char(101),
+	char(114),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(110),
+	char(97),
+	char(109),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(84),
+	char(121),
+	char(112),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(97),
+	char(100),
+	char(100),
+	char(105),
+	char(110),
+	char(103),
+	char(91),
+	char(52),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(111),
+	char(99),
+	char(97),
+	char(108),
+	char(83),
+	char(99),
+	char(97),
+	char(108),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(108),
+	char(97),
+	char(110),
+	char(101),
+	char(78),
+	char(111),
+	char(114),
+	char(109),
+	char(97),
+	char(108),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(108),
+	char(97),
+	char(110),
+	char(101),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(97),
+	char(110),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(109),
+	char(112),
+	char(108),
+	char(105),
+	char(99),
+	char(105),
+	char(116),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(105),
+	char(109),
+	char(101),
+	char(110),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(77),
+	char(97),
+	char(114),
+	char(103),
+	char(105),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(97),
+	char(100),
+	char(100),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(111),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(97),
+	char(100),
+	char(105),
+	char(117),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(110),
+	char(118),
+	char(101),
+	char(120),
+	char(73),
+	char(110),
+	char(116),
+	char(101),
+	char(114),
+	char(110),
+	char(97),
+	char(108),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(108),
+	char(111),
+	char(99),
+	char(97),
+	char(108),
+	char(80),
+	char(111),
+	char(115),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(65),
+	char(114),
+	char(114),
+	char(97),
+	char(121),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(111),
+	char(99),
+	char(97),
+	char(108),
+	char(80),
+	char(111),
+	char(115),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(65),
+	char(114),
+	char(114),
+	char(97),
+	char(121),
+	char(83),
+	char(105),
+	char(122),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(118),
+	char(97),
+	char(108),
+	char(117),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(97),
+	char(100),
+	char(91),
+	char(50),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(118),
+	char(97),
+	char(108),
+	char(117),
+	char(101),
+	char(115),
+	char(91),
+	char(51),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(97),
+	char(100),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(118),
+	char(101),
+	char(114),
+	char(116),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(51),
+	char(102),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(118),
+	char(101),
+	char(114),
+	char(116),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(51),
+	char(100),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(105),
+	char(110),
+	char(100),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(51),
+	char(50),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(51),
+	char(105),
+	char(110),
+	char(100),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(49),
+	char(54),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(51),
+	char(105),
+	char(110),
+	char(100),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(56),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(105),
+	char(110),
+	char(100),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(49),
+	char(54),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(84),
+	char(114),
+	char(105),
+	char(97),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(86),
+	char(101),
+	char(114),
+	char(116),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(109),
+	char(101),
+	char(115),
+	char(104),
+	char(80),
+	char(97),
+	char(114),
+	char(116),
+	char(115),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(99),
+	char(97),
+	char(108),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(77),
+	char(101),
+	char(115),
+	char(104),
+	char(80),
+	char(97),
+	char(114),
+	char(116),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(109),
+	char(101),
+	char(115),
+	char(104),
+	char(73),
+	char(110),
+	char(116),
+	char(101),
+	char(114),
+	char(102),
+	char(97),
+	char(99),
+	char(101),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(113),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(66),
+	char(118),
+	char(104),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(113),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(66),
+	char(118),
+	char(104),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(116),
+	char(114),
+	char(105),
+	char(97),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(73),
+	char(110),
+	char(102),
+	char(111),
+	char(77),
+	char(97),
+	char(112),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(97),
+	char(100),
+	char(51),
+	char(91),
+	char(52),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(114),
+	char(105),
+	char(109),
+	char(101),
+	char(115),
+	char(104),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(114),
+	char(97),
+	char(110),
+	char(115),
+	char(102),
+	char(111),
+	char(114),
+	char(109),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(99),
+	char(104),
+	char(105),
+	char(108),
+	char(100),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(104),
+	char(105),
+	char(108),
+	char(100),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(84),
+	char(121),
+	char(112),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(104),
+	char(105),
+	char(108),
+	char(100),
+	char(77),
+	char(97),
+	char(114),
+	char(103),
+	char(105),
+	char(110),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(99),
+	char(104),
+	char(105),
+	char(108),
+	char(100),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(67),
+	char(104),
+	char(105),
+	char(108),
+	char(100),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(117),
+	char(112),
+	char(65),
+	char(120),
+	char(105),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(102),
+	char(108),
+	char(97),
+	char(103),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(100),
+	char(103),
+	char(101),
+	char(86),
+	char(48),
+	char(86),
+	char(49),
+	char(65),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(100),
+	char(103),
+	char(101),
+	char(86),
+	char(49),
+	char(86),
+	char(50),
+	char(65),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(100),
+	char(103),
+	char(101),
+	char(86),
+	char(50),
+	char(86),
+	char(48),
+	char(65),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(104),
+	char(97),
+	char(115),
+	char(104),
+	char(84),
+	char(97),
+	char(98),
+	char(108),
+	char(101),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(110),
+	char(101),
+	char(120),
+	char(116),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(118),
+	char(97),
+	char(108),
+	char(117),
+	char(101),
+	char(65),
+	char(114),
+	char(114),
+	char(97),
+	char(121),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(107),
+	char(101),
+	char(121),
+	char(65),
+	char(114),
+	char(114),
+	char(97),
+	char(121),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(110),
+	char(118),
+	char(101),
+	char(120),
+	char(69),
+	char(112),
+	char(115),
+	char(105),
+	char(108),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(108),
+	char(97),
+	char(110),
+	char(97),
+	char(114),
+	char(69),
+	char(112),
+	char(115),
+	char(105),
+	char(108),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(113),
+	char(117),
+	char(97),
+	char(108),
+	char(86),
+	char(101),
+	char(114),
+	char(116),
+	char(101),
+	char(120),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(100),
+	char(103),
+	char(101),
+	char(68),
+	char(105),
+	char(115),
+	char(116),
+	char(97),
+	char(110),
+	char(99),
+	char(101),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(122),
+	char(101),
+	char(114),
+	char(111),
+	char(65),
+	char(114),
+	char(101),
+	char(97),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(101),
+	char(120),
+	char(116),
+	char(83),
+	char(105),
+	char(122),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(104),
+	char(97),
+	char(115),
+	char(104),
+	char(84),
+	char(97),
+	char(98),
+	char(108),
+	char(101),
+	char(83),
+	char(105),
+	char(122),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(86),
+	char(97),
+	char(108),
+	char(117),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(75),
+	char(101),
+	char(121),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(103),
+	char(105),
+	char(109),
+	char(112),
+	char(97),
+	char(99),
+	char(116),
+	char(83),
+	char(117),
+	char(98),
+	char(84),
+	char(121),
+	char(112),
+	char(101),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(117),
+	char(110),
+	char(115),
+	char(99),
+	char(97),
+	char(108),
+	char(101),
+	char(100),
+	char(80),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(115),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(117),
+	char(110),
+	char(115),
+	char(99),
+	char(97),
+	char(108),
+	char(101),
+	char(100),
+	char(80),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(115),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(85),
+	char(110),
+	char(115),
+	char(99),
+	char(97),
+	char(108),
+	char(101),
+	char(100),
+	char(80),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(97),
+	char(100),
+	char(100),
+	char(105),
+	char(110),
+	char(103),
+	char(51),
+	char(91),
+	char(52),
+	char(93),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(98),
+	char(114),
+	char(111),
+	char(97),
+	char(100),
+	char(112),
+	char(104),
+	char(97),
+	char(115),
+	char(101),
+	char(72),
+	char(97),
+	char(110),
+	char(100),
+	char(108),
+	char(101),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(114),
+	char(111),
+	char(111),
+	char(116),
+	char(67),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(119),
+	char(111),
+	char(114),
+	char(108),
+	char(100),
+	char(84),
+	char(114),
+	char(97),
+	char(110),
+	char(115),
+	char(102),
+	char(111),
+	char(114),
+	char(109),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(110),
+	char(116),
+	char(101),
+	char(114),
+	char(112),
+	char(111),
+	char(108),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(87),
+	char(111),
+	char(114),
+	char(108),
+	char(100),
+	char(84),
+	char(114),
+	char(97),
+	char(110),
+	char(115),
+	char(102),
+	char(111),
+	char(114),
+	char(109),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(110),
+	char(116),
+	char(101),
+	char(114),
+	char(112),
+	char(111),
+	char(108),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(76),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(86),
+	char(101),
+	char(108),
+	char(111),
+	char(99),
+	char(105),
+	char(116),
+	char(121),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(110),
+	char(116),
+	char(101),
+	char(114),
+	char(112),
+	char(111),
+	char(108),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(65),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(86),
+	char(101),
+	char(108),
+	char(111),
+	char(99),
+	char(105),
+	char(116),
+	char(121),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(105),
+	char(115),
+	char(111),
+	char(116),
+	char(114),
+	char(111),
+	char(112),
+	char(105),
+	char(99),
+	char(70),
+	char(114),
+	char(105),
+	char(99),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(110),
+	char(116),
+	char(97),
+	char(99),
+	char(116),
+	char(80),
+	char(114),
+	char(111),
+	char(99),
+	char(101),
+	char(115),
+	char(115),
+	char(105),
+	char(110),
+	char(103),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(100),
+	char(101),
+	char(97),
+	char(99),
+	char(116),
+	char(105),
+	char(118),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(84),
+	char(105),
+	char(109),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(102),
+	char(114),
+	char(105),
+	char(99),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(110),
+	char(103),
+	char(70),
+	char(114),
+	char(105),
+	char(99),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(101),
+	char(115),
+	char(116),
+	char(105),
+	char(116),
+	char(117),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(104),
+	char(105),
+	char(116),
+	char(70),
+	char(114),
+	char(97),
+	char(99),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(99),
+	char(100),
+	char(83),
+	char(119),
+	char(101),
+	char(112),
+	char(116),
+	char(83),
+	char(112),
+	char(104),
+	char(101),
+	char(114),
+	char(101),
+	char(82),
+	char(97),
+	char(100),
+	char(105),
+	char(117),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(99),
+	char(100),
+	char(77),
+	char(111),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(104),
+	char(97),
+	char(115),
+	char(65),
+	char(110),
+	char(105),
+	char(115),
+	char(111),
+	char(116),
+	char(114),
+	char(111),
+	char(112),
+	char(105),
+	char(99),
+	char(70),
+	char(114),
+	char(105),
+	char(99),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(70),
+	char(108),
+	char(97),
+	char(103),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(115),
+	char(108),
+	char(97),
+	char(110),
+	char(100),
+	char(84),
+	char(97),
+	char(103),
+	char(49),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(109),
+	char(112),
+	char(97),
+	char(110),
+	char(105),
+	char(111),
+	char(110),
+	char(73),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(99),
+	char(116),
+	char(105),
+	char(118),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(83),
+	char(116),
+	char(97),
+	char(116),
+	char(101),
+	char(49),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(110),
+	char(116),
+	char(101),
+	char(114),
+	char(110),
+	char(97),
+	char(108),
+	char(84),
+	char(121),
+	char(112),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(104),
+	char(101),
+	char(99),
+	char(107),
+	char(67),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(100),
+	char(101),
+	char(87),
+	char(105),
+	char(116),
+	char(104),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(111),
+	char(108),
+	char(118),
+	char(101),
+	char(114),
+	char(73),
+	char(110),
+	char(102),
+	char(111),
+	char(0),
+	char(109),
+	char(95),
+	char(103),
+	char(114),
+	char(97),
+	char(118),
+	char(105),
+	char(116),
+	char(121),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(79),
+	char(98),
+	char(106),
+	char(101),
+	char(99),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(110),
+	char(118),
+	char(73),
+	char(110),
+	char(101),
+	char(114),
+	char(116),
+	char(105),
+	char(97),
+	char(84),
+	char(101),
+	char(110),
+	char(115),
+	char(111),
+	char(114),
+	char(87),
+	char(111),
+	char(114),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(86),
+	char(101),
+	char(108),
+	char(111),
+	char(99),
+	char(105),
+	char(116),
+	char(121),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(86),
+	char(101),
+	char(108),
+	char(111),
+	char(99),
+	char(105),
+	char(116),
+	char(121),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(70),
+	char(97),
+	char(99),
+	char(116),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(70),
+	char(97),
+	char(99),
+	char(116),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(103),
+	char(114),
+	char(97),
+	char(118),
+	char(105),
+	char(116),
+	char(121),
+	char(95),
+	char(97),
+	char(99),
+	char(99),
+	char(101),
+	char(108),
+	char(101),
+	char(114),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(110),
+	char(118),
+	char(73),
+	char(110),
+	char(101),
+	char(114),
+	char(116),
+	char(105),
+	char(97),
+	char(76),
+	char(111),
+	char(99),
+	char(97),
+	char(108),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(111),
+	char(116),
+	char(97),
+	char(108),
+	char(70),
+	char(111),
+	char(114),
+	char(99),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(111),
+	char(116),
+	char(97),
+	char(108),
+	char(84),
+	char(111),
+	char(114),
+	char(113),
+	char(117),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(110),
+	char(118),
+	char(101),
+	char(114),
+	char(115),
+	char(101),
+	char(77),
+	char(97),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(68),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(68),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(100),
+	char(100),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(97),
+	char(108),
+	char(68),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(70),
+	char(97),
+	char(99),
+	char(116),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(100),
+	char(100),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(97),
+	char(108),
+	char(76),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(68),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(83),
+	char(113),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(100),
+	char(100),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(97),
+	char(108),
+	char(65),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(68),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(83),
+	char(113),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(100),
+	char(100),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(97),
+	char(108),
+	char(65),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(68),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(70),
+	char(97),
+	char(99),
+	char(116),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(83),
+	char(108),
+	char(101),
+	char(101),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(83),
+	char(108),
+	char(101),
+	char(101),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(100),
+	char(100),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(97),
+	char(108),
+	char(68),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(82),
+	char(111),
+	char(119),
+	char(115),
+	char(0),
+	char(110),
+	char(117),
+	char(98),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(114),
+	char(98),
+	char(65),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(114),
+	char(98),
+	char(66),
+	char(0),
+	char(109),
+	char(95),
+	char(111),
+	char(98),
+	char(106),
+	char(101),
+	char(99),
+	char(116),
+	char(84),
+	char(121),
+	char(112),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(117),
+	char(115),
+	char(101),
+	char(114),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(84),
+	char(121),
+	char(112),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(117),
+	char(115),
+	char(101),
+	char(114),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(73),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(101),
+	char(101),
+	char(100),
+	char(115),
+	char(70),
+	char(101),
+	char(101),
+	char(100),
+	char(98),
+	char(97),
+	char(99),
+	char(107),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(112),
+	char(112),
+	char(108),
+	char(105),
+	char(101),
+	char(100),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(100),
+	char(98),
+	char(103),
+	char(68),
+	char(114),
+	char(97),
+	char(119),
+	char(83),
+	char(105),
+	char(122),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(100),
+	char(105),
+	char(115),
+	char(97),
+	char(98),
+	char(108),
+	char(101),
+	char(67),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(115),
+	char(66),
+	char(101),
+	char(116),
+	char(119),
+	char(101),
+	char(101),
+	char(110),
+	char(76),
+	char(105),
+	char(110),
+	char(107),
+	char(101),
+	char(100),
+	char(66),
+	char(111),
+	char(100),
+	char(105),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(111),
+	char(118),
+	char(101),
+	char(114),
+	char(114),
+	char(105),
+	char(100),
+	char(101),
+	char(78),
+	char(117),
+	char(109),
+	char(83),
+	char(111),
+	char(108),
+	char(118),
+	char(101),
+	char(114),
+	char(73),
+	char(116),
+	char(101),
+	char(114),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(114),
+	char(101),
+	char(97),
+	char(107),
+	char(105),
+	char(110),
+	char(103),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(115),
+	char(69),
+	char(110),
+	char(97),
+	char(98),
+	char(108),
+	char(101),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(121),
+	char(112),
+	char(101),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(105),
+	char(118),
+	char(111),
+	char(116),
+	char(73),
+	char(110),
+	char(65),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(105),
+	char(118),
+	char(111),
+	char(116),
+	char(73),
+	char(110),
+	char(66),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(98),
+	char(65),
+	char(70),
+	char(114),
+	char(97),
+	char(109),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(98),
+	char(66),
+	char(70),
+	char(114),
+	char(97),
+	char(109),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(117),
+	char(115),
+	char(101),
+	char(82),
+	char(101),
+	char(102),
+	char(101),
+	char(114),
+	char(101),
+	char(110),
+	char(99),
+	char(101),
+	char(70),
+	char(114),
+	char(97),
+	char(109),
+	char(101),
+	char(65),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(79),
+	char(110),
+	char(108),
+	char(121),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(110),
+	char(97),
+	char(98),
+	char(108),
+	char(101),
+	char(65),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(77),
+	char(111),
+	char(116),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(109),
+	char(111),
+	char(116),
+	char(111),
+	char(114),
+	char(84),
+	char(97),
+	char(114),
+	char(103),
+	char(101),
+	char(116),
+	char(86),
+	char(101),
+	char(108),
+	char(111),
+	char(99),
+	char(105),
+	char(116),
+	char(121),
+	char(0),
+	char(109),
+	char(95),
+	char(109),
+	char(97),
+	char(120),
+	char(77),
+	char(111),
+	char(116),
+	char(111),
+	char(114),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(111),
+	char(119),
+	char(101),
+	char(114),
+	char(76),
+	char(105),
+	char(109),
+	char(105),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(117),
+	char(112),
+	char(112),
+	char(101),
+	char(114),
+	char(76),
+	char(105),
+	char(109),
+	char(105),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(109),
+	char(105),
+	char(116),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(105),
+	char(97),
+	char(115),
+	char(70),
+	char(97),
+	char(99),
+	char(116),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(101),
+	char(108),
+	char(97),
+	char(120),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(70),
+	char(97),
+	char(99),
+	char(116),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(119),
+	char(105),
+	char(110),
+	char(103),
+	char(83),
+	char(112),
+	char(97),
+	char(110),
+	char(49),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(119),
+	char(105),
+	char(110),
+	char(103),
+	char(83),
+	char(112),
+	char(97),
+	char(110),
+	char(50),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(119),
+	char(105),
+	char(115),
+	char(116),
+	char(83),
+	char(112),
+	char(97),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(100),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(85),
+	char(112),
+	char(112),
+	char(101),
+	char(114),
+	char(76),
+	char(105),
+	char(109),
+	char(105),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(76),
+	char(111),
+	char(119),
+	char(101),
+	char(114),
+	char(76),
+	char(105),
+	char(109),
+	char(105),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(85),
+	char(112),
+	char(112),
+	char(101),
+	char(114),
+	char(76),
+	char(105),
+	char(109),
+	char(105),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(76),
+	char(111),
+	char(119),
+	char(101),
+	char(114),
+	char(76),
+	char(105),
+	char(109),
+	char(105),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(117),
+	char(115),
+	char(101),
+	char(76),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(82),
+	char(101),
+	char(102),
+	char(101),
+	char(114),
+	char(101),
+	char(110),
+	char(99),
+	char(101),
+	char(70),
+	char(114),
+	char(97),
+	char(109),
+	char(101),
+	char(65),
+	char(0),
+	char(109),
+	char(95),
+	char(117),
+	char(115),
+	char(101),
+	char(79),
+	char(102),
+	char(102),
+	char(115),
+	char(101),
+	char(116),
+	char(70),
+	char(111),
+	char(114),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(70),
+	char(114),
+	char(97),
+	char(109),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(54),
+	char(100),
+	char(111),
+	char(102),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(112),
+	char(114),
+	char(105),
+	char(110),
+	char(103),
+	char(69),
+	char(110),
+	char(97),
+	char(98),
+	char(108),
+	char(101),
+	char(100),
+	char(91),
+	char(54),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(113),
+	char(117),
+	char(105),
+	char(108),
+	char(105),
+	char(98),
+	char(114),
+	char(105),
+	char(117),
+	char(109),
+	char(80),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(91),
+	char(54),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(112),
+	char(114),
+	char(105),
+	char(110),
+	char(103),
+	char(83),
+	char(116),
+	char(105),
+	char(102),
+	char(102),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(91),
+	char(54),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(112),
+	char(114),
+	char(105),
+	char(110),
+	char(103),
+	char(68),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(91),
+	char(54),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(97),
+	char(117),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(105),
+	char(109),
+	char(101),
+	char(83),
+	char(116),
+	char(101),
+	char(112),
+	char(0),
+	char(109),
+	char(95),
+	char(109),
+	char(97),
+	char(120),
+	char(69),
+	char(114),
+	char(114),
+	char(111),
+	char(114),
+	char(82),
+	char(101),
+	char(100),
+	char(117),
+	char(99),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(114),
+	char(112),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(114),
+	char(112),
+	char(50),
+	char(0),
+	char(109),
+	char(95),
+	char(103),
+	char(108),
+	char(111),
+	char(98),
+	char(97),
+	char(108),
+	char(67),
+	char(102),
+	char(109),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(112),
+	char(108),
+	char(105),
+	char(116),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(80),
+	char(101),
+	char(110),
+	char(101),
+	char(116),
+	char(114),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(112),
+	char(108),
+	char(105),
+	char(116),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(84),
+	char(117),
+	char(114),
+	char(110),
+	char(69),
+	char(114),
+	char(112),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(83),
+	char(108),
+	char(111),
+	char(112),
+	char(0),
+	char(109),
+	char(95),
+	char(119),
+	char(97),
+	char(114),
+	char(109),
+	char(115),
+	char(116),
+	char(97),
+	char(114),
+	char(116),
+	char(105),
+	char(110),
+	char(103),
+	char(70),
+	char(97),
+	char(99),
+	char(116),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(109),
+	char(97),
+	char(120),
+	char(71),
+	char(121),
+	char(114),
+	char(111),
+	char(115),
+	char(99),
+	char(111),
+	char(112),
+	char(105),
+	char(99),
+	char(70),
+	char(111),
+	char(114),
+	char(99),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(105),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(65),
+	char(120),
+	char(105),
+	char(115),
+	char(82),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(110),
+	char(103),
+	char(70),
+	char(114),
+	char(105),
+	char(99),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(73),
+	char(116),
+	char(101),
+	char(114),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(111),
+	char(108),
+	char(118),
+	char(101),
+	char(114),
+	char(77),
+	char(111),
+	char(100),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(101),
+	char(115),
+	char(116),
+	char(105),
+	char(110),
+	char(103),
+	char(67),
+	char(111),
+	char(110),
+	char(116),
+	char(97),
+	char(99),
+	char(116),
+	char(82),
+	char(101),
+	char(115),
+	char(116),
+	char(105),
+	char(116),
+	char(117),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(109),
+	char(105),
+	char(110),
+	char(105),
+	char(109),
+	char(117),
+	char(109),
+	char(83),
+	char(111),
+	char(108),
+	char(118),
+	char(101),
+	char(114),
+	char(66),
+	char(97),
+	char(116),
+	char(99),
+	char(104),
+	char(83),
+	char(105),
+	char(122),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(112),
+	char(108),
+	char(105),
+	char(116),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(83),
+	char(116),
+	char(105),
+	char(102),
+	char(102),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(83),
+	char(116),
+	char(105),
+	char(102),
+	char(102),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(118),
+	char(111),
+	char(108),
+	char(117),
+	char(109),
+	char(101),
+	char(83),
+	char(116),
+	char(105),
+	char(102),
+	char(102),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(109),
+	char(97),
+	char(116),
+	char(101),
+	char(114),
+	char(105),
+	char(97),
+	char(108),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(111),
+	char(115),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(114),
+	char(101),
+	char(118),
+	char(105),
+	char(111),
+	char(117),
+	char(115),
+	char(80),
+	char(111),
+	char(115),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(118),
+	char(101),
+	char(108),
+	char(111),
+	char(99),
+	char(105),
+	char(116),
+	char(121),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(99),
+	char(99),
+	char(117),
+	char(109),
+	char(117),
+	char(108),
+	char(97),
+	char(116),
+	char(101),
+	char(100),
+	char(70),
+	char(111),
+	char(114),
+	char(99),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(111),
+	char(114),
+	char(109),
+	char(97),
+	char(108),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(114),
+	char(101),
+	char(97),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(116),
+	char(116),
+	char(97),
+	char(99),
+	char(104),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(111),
+	char(100),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(91),
+	char(50),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(101),
+	char(115),
+	char(116),
+	char(76),
+	char(101),
+	char(110),
+	char(103),
+	char(116),
+	char(104),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(98),
+	char(101),
+	char(110),
+	char(100),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(111),
+	char(100),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(91),
+	char(51),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(101),
+	char(115),
+	char(116),
+	char(65),
+	char(114),
+	char(101),
+	char(97),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(48),
+	char(91),
+	char(52),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(111),
+	char(100),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(91),
+	char(52),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(101),
+	char(115),
+	char(116),
+	char(86),
+	char(111),
+	char(108),
+	char(117),
+	char(109),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(49),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(50),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(48),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(111),
+	char(99),
+	char(97),
+	char(108),
+	char(70),
+	char(114),
+	char(97),
+	char(109),
+	char(101),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(114),
+	char(105),
+	char(103),
+	char(105),
+	char(100),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(111),
+	char(100),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(101),
+	char(114),
+	char(111),
+	char(77),
+	char(111),
+	char(100),
+	char(101),
+	char(108),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(97),
+	char(117),
+	char(109),
+	char(103),
+	char(97),
+	char(114),
+	char(116),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(100),
+	char(114),
+	char(97),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(102),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(114),
+	char(101),
+	char(115),
+	char(115),
+	char(117),
+	char(114),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(118),
+	char(111),
+	char(108),
+	char(117),
+	char(109),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(100),
+	char(121),
+	char(110),
+	char(97),
+	char(109),
+	char(105),
+	char(99),
+	char(70),
+	char(114),
+	char(105),
+	char(99),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(111),
+	char(115),
+	char(101),
+	char(77),
+	char(97),
+	char(116),
+	char(99),
+	char(104),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(105),
+	char(103),
+	char(105),
+	char(100),
+	char(67),
+	char(111),
+	char(110),
+	char(116),
+	char(97),
+	char(99),
+	char(116),
+	char(72),
+	char(97),
+	char(114),
+	char(100),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(107),
+	char(105),
+	char(110),
+	char(101),
+	char(116),
+	char(105),
+	char(99),
+	char(67),
+	char(111),
+	char(110),
+	char(116),
+	char(97),
+	char(99),
+	char(116),
+	char(72),
+	char(97),
+	char(114),
+	char(100),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(111),
+	char(102),
+	char(116),
+	char(67),
+	char(111),
+	char(110),
+	char(116),
+	char(97),
+	char(99),
+	char(116),
+	char(72),
+	char(97),
+	char(114),
+	char(100),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(99),
+	char(104),
+	char(111),
+	char(114),
+	char(72),
+	char(97),
+	char(114),
+	char(100),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(111),
+	char(102),
+	char(116),
+	char(82),
+	char(105),
+	char(103),
+	char(105),
+	char(100),
+	char(67),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(72),
+	char(97),
+	char(114),
+	char(100),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(111),
+	char(102),
+	char(116),
+	char(75),
+	char(105),
+	char(110),
+	char(101),
+	char(116),
+	char(105),
+	char(99),
+	char(67),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(72),
+	char(97),
+	char(114),
+	char(100),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(111),
+	char(102),
+	char(116),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(67),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(72),
+	char(97),
+	char(114),
+	char(100),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(111),
+	char(102),
+	char(116),
+	char(82),
+	char(105),
+	char(103),
+	char(105),
+	char(100),
+	char(67),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(83),
+	char(112),
+	char(108),
+	char(105),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(111),
+	char(102),
+	char(116),
+	char(75),
+	char(105),
+	char(110),
+	char(101),
+	char(116),
+	char(105),
+	char(99),
+	char(67),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(83),
+	char(112),
+	char(108),
+	char(105),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(111),
+	char(102),
+	char(116),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(67),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(83),
+	char(112),
+	char(108),
+	char(105),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(109),
+	char(97),
+	char(120),
+	char(86),
+	char(111),
+	char(108),
+	char(117),
+	char(109),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(105),
+	char(109),
+	char(101),
+	char(83),
+	char(99),
+	char(97),
+	char(108),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(118),
+	char(101),
+	char(108),
+	char(111),
+	char(99),
+	char(105),
+	char(116),
+	char(121),
+	char(73),
+	char(116),
+	char(101),
+	char(114),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(111),
+	char(115),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(73),
+	char(116),
+	char(101),
+	char(114),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(100),
+	char(114),
+	char(105),
+	char(102),
+	char(116),
+	char(73),
+	char(116),
+	char(101),
+	char(114),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(73),
+	char(116),
+	char(101),
+	char(114),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(111),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(99),
+	char(97),
+	char(108),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(113),
+	char(113),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(109),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(112),
+	char(111),
+	char(115),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(119),
+	char(101),
+	char(105),
+	char(103),
+	char(104),
+	char(116),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(80),
+	char(111),
+	char(115),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(87),
+	char(101),
+	char(105),
+	char(103),
+	char(116),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(118),
+	char(111),
+	char(108),
+	char(117),
+	char(109),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(102),
+	char(114),
+	char(97),
+	char(109),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(102),
+	char(114),
+	char(97),
+	char(109),
+	char(101),
+	char(120),
+	char(102),
+	char(111),
+	char(114),
+	char(109),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(111),
+	char(99),
+	char(105),
+	char(105),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(110),
+	char(118),
+	char(119),
+	char(105),
+	char(0),
+	char(109),
+	char(95),
+	char(118),
+	char(105),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(115),
+	char(91),
+	char(50),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(100),
+	char(105),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(115),
+	char(91),
+	char(50),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(118),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(118),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(102),
+	char(114),
+	char(97),
+	char(109),
+	char(101),
+	char(114),
+	char(101),
+	char(102),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(110),
+	char(111),
+	char(100),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(109),
+	char(97),
+	char(115),
+	char(115),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(70),
+	char(114),
+	char(97),
+	char(109),
+	char(101),
+	char(82),
+	char(101),
+	char(102),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(77),
+	char(97),
+	char(115),
+	char(115),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(100),
+	char(109),
+	char(97),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(109),
+	char(97),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(118),
+	char(105),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(100),
+	char(105),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(100),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(100),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(100),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(109),
+	char(97),
+	char(116),
+	char(99),
+	char(104),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(109),
+	char(97),
+	char(120),
+	char(83),
+	char(101),
+	char(108),
+	char(102),
+	char(67),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(101),
+	char(108),
+	char(102),
+	char(67),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(70),
+	char(97),
+	char(99),
+	char(116),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(110),
+	char(116),
+	char(97),
+	char(105),
+	char(110),
+	char(115),
+	char(65),
+	char(110),
+	char(99),
+	char(104),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(100),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(98),
+	char(111),
+	char(100),
+	char(121),
+	char(65),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(98),
+	char(111),
+	char(100),
+	char(121),
+	char(66),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(101),
+	char(102),
+	char(115),
+	char(91),
+	char(50),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(102),
+	char(109),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(112),
+	char(108),
+	char(105),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(100),
+	char(101),
+	char(108),
+	char(101),
+	char(116),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(101),
+	char(108),
+	char(80),
+	char(111),
+	char(115),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(91),
+	char(50),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(111),
+	char(100),
+	char(121),
+	char(65),
+	char(116),
+	char(121),
+	char(112),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(111),
+	char(100),
+	char(121),
+	char(66),
+	char(116),
+	char(121),
+	char(112),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(106),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(84),
+	char(121),
+	char(112),
+	char(101),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(112),
+	char(111),
+	char(115),
+	char(101),
+	char(0),
+	char(42),
+	char(42),
+	char(109),
+	char(95),
+	char(109),
+	char(97),
+	char(116),
+	char(101),
+	char(114),
+	char(105),
+	char(97),
+	char(108),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(110),
+	char(111),
+	char(100),
+	char(101),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(110),
+	char(107),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(102),
+	char(97),
+	char(99),
+	char(101),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(116),
+	char(101),
+	char(116),
+	char(114),
+	char(97),
+	char(104),
+	char(101),
+	char(100),
+	char(114),
+	char(97),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(99),
+	char(104),
+	char(111),
+	char(114),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(99),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(106),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(77),
+	char(97),
+	char(116),
+	char(101),
+	char(114),
+	char(105),
+	char(97),
+	char(108),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(76),
+	char(105),
+	char(110),
+	char(107),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(70),
+	char(97),
+	char(99),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(84),
+	char(101),
+	char(116),
+	char(114),
+	char(97),
+	char(104),
+	char(101),
+	char(100),
+	char(114),
+	char(97),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(65),
+	char(110),
+	char(99),
+	char(104),
+	char(111),
+	char(114),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(67),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(74),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(110),
+	char(102),
+	char(105),
+	char(103),
+	char(0),
+	char(84),
+	char(89),
+	char(80),
+	char(69),
+	char(76),
+	char(0),
+	char(0),
+	char(0),
+	char(99),
+	char(104),
+	char(97),
+	char(114),
+	char(0),
+	char(117),
+	char(99),
+	char(104),
+	char(97),
+	char(114),
+	char(0),
+	char(115),
+	char(104),
+	char(111),
+	char(114),
+	char(116),
+	char(0),
+	char(117),
+	char(115),
+	char(104),
+	char(111),
+	char(114),
+	char(116),
+	char(0),
+	char(105),
+	char(110),
+	char(116),
+	char(0),
+	char(108),
+	char(111),
+	char(110),
+	char(103),
+	char(0),
+	char(117),
+	char(108),
+	char(111),
+	char(110),
+	char(103),
+	char(0),
+	char(102),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(0),
+	char(100),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(0),
+	char(118),
+	char(111),
+	char(105),
+	char(100),
+	char(0),
+	char(80),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(101),
+	char(114),
+	char(65),
+	char(114),
+	char(114),
+	char(97),
+	char(121),
+	char(0),
+	char(98),
+	char(116),
+	char(80),
+	char(104),
+	char(121),
+	char(115),
+	char(105),
+	char(99),
+	char(115),
+	char(83),
+	char(121),
+	char(115),
+	char(116),
+	char(101),
+	char(109),
+	char(0),
+	char(76),
+	char(105),
+	char(115),
+	char(116),
+	char(66),
+	char(97),
+	char(115),
+	char(101),
+	char(0),
+	char(98),
+	char(116),
+	char(86),
+	char(101),
+	char(99),
+	char(116),
+	char(111),
+	char(114),
+	char(51),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(86),
+	char(101),
+	char(99),
+	char(116),
+	char(111),
+	char(114),
+	char(51),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(77),
+	char(97),
+	char(116),
+	char(114),
+	char(105),
+	char(120),
+	char(51),
+	char(120),
+	char(51),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(77),
+	char(97),
+	char(116),
+	char(114),
+	char(105),
+	char(120),
+	char(51),
+	char(120),
+	char(51),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(84),
+	char(114),
+	char(97),
+	char(110),
+	char(115),
+	char(102),
+	char(111),
+	char(114),
+	char(109),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(84),
+	char(114),
+	char(97),
+	char(110),
+	char(115),
+	char(102),
+	char(111),
+	char(114),
+	char(109),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(66),
+	char(118),
+	char(104),
+	char(83),
+	char(117),
+	char(98),
+	char(116),
+	char(114),
+	char(101),
+	char(101),
+	char(73),
+	char(110),
+	char(102),
+	char(111),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(79),
+	char(112),
+	char(116),
+	char(105),
+	char(109),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(66),
+	char(118),
+	char(104),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(79),
+	char(112),
+	char(116),
+	char(105),
+	char(109),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(66),
+	char(118),
+	char(104),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(81),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(66),
+	char(118),
+	char(104),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(81),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(66),
+	char(118),
+	char(104),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(81),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(66),
+	char(118),
+	char(104),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(83),
+	char(116),
+	char(97),
+	char(116),
+	char(105),
+	char(99),
+	char(80),
+	char(108),
+	char(97),
+	char(110),
+	char(101),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(110),
+	char(118),
+	char(101),
+	char(120),
+	char(73),
+	char(110),
+	char(116),
+	char(101),
+	char(114),
+	char(110),
+	char(97),
+	char(108),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(80),
+	char(111),
+	char(115),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(65),
+	char(110),
+	char(100),
+	char(82),
+	char(97),
+	char(100),
+	char(105),
+	char(117),
+	char(115),
+	char(0),
+	char(98),
+	char(116),
+	char(77),
+	char(117),
+	char(108),
+	char(116),
+	char(105),
+	char(83),
+	char(112),
+	char(104),
+	char(101),
+	char(114),
+	char(101),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(73),
+	char(110),
+	char(116),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(83),
+	char(104),
+	char(111),
+	char(114),
+	char(116),
+	char(73),
+	char(110),
+	char(116),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(83),
+	char(104),
+	char(111),
+	char(114),
+	char(116),
+	char(73),
+	char(110),
+	char(116),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(84),
+	char(114),
+	char(105),
+	char(112),
+	char(108),
+	char(101),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(104),
+	char(97),
+	char(114),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(84),
+	char(114),
+	char(105),
+	char(112),
+	char(108),
+	char(101),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(77),
+	char(101),
+	char(115),
+	char(104),
+	char(80),
+	char(97),
+	char(114),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(83),
+	char(116),
+	char(114),
+	char(105),
+	char(100),
+	char(105),
+	char(110),
+	char(103),
+	char(77),
+	char(101),
+	char(115),
+	char(104),
+	char(73),
+	char(110),
+	char(116),
+	char(101),
+	char(114),
+	char(102),
+	char(97),
+	char(99),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(84),
+	char(114),
+	char(105),
+	char(97),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(77),
+	char(101),
+	char(115),
+	char(104),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(84),
+	char(114),
+	char(105),
+	char(97),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(73),
+	char(110),
+	char(102),
+	char(111),
+	char(77),
+	char(97),
+	char(112),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(83),
+	char(99),
+	char(97),
+	char(108),
+	char(101),
+	char(100),
+	char(84),
+	char(114),
+	char(105),
+	char(97),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(77),
+	char(101),
+	char(115),
+	char(104),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(109),
+	char(112),
+	char(111),
+	char(117),
+	char(110),
+	char(100),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(67),
+	char(104),
+	char(105),
+	char(108),
+	char(100),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(109),
+	char(112),
+	char(111),
+	char(117),
+	char(110),
+	char(100),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(121),
+	char(108),
+	char(105),
+	char(110),
+	char(100),
+	char(101),
+	char(114),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(97),
+	char(112),
+	char(115),
+	char(117),
+	char(108),
+	char(101),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(84),
+	char(114),
+	char(105),
+	char(97),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(73),
+	char(110),
+	char(102),
+	char(111),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(71),
+	char(73),
+	char(109),
+	char(112),
+	char(97),
+	char(99),
+	char(116),
+	char(77),
+	char(101),
+	char(115),
+	char(104),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(110),
+	char(118),
+	char(101),
+	char(120),
+	char(72),
+	char(117),
+	char(108),
+	char(108),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(79),
+	char(98),
+	char(106),
+	char(101),
+	char(99),
+	char(116),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(79),
+	char(98),
+	char(106),
+	char(101),
+	char(99),
+	char(116),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(68),
+	char(121),
+	char(110),
+	char(97),
+	char(109),
+	char(105),
+	char(99),
+	char(115),
+	char(87),
+	char(111),
+	char(114),
+	char(108),
+	char(100),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(110),
+	char(116),
+	char(97),
+	char(99),
+	char(116),
+	char(83),
+	char(111),
+	char(108),
+	char(118),
+	char(101),
+	char(114),
+	char(73),
+	char(110),
+	char(102),
+	char(111),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(68),
+	char(121),
+	char(110),
+	char(97),
+	char(109),
+	char(105),
+	char(99),
+	char(115),
+	char(87),
+	char(111),
+	char(114),
+	char(108),
+	char(100),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(110),
+	char(116),
+	char(97),
+	char(99),
+	char(116),
+	char(83),
+	char(111),
+	char(108),
+	char(118),
+	char(101),
+	char(114),
+	char(73),
+	char(110),
+	char(102),
+	char(111),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(82),
+	char(105),
+	char(103),
+	char(105),
+	char(100),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(82),
+	char(105),
+	char(103),
+	char(105),
+	char(100),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(73),
+	char(110),
+	char(102),
+	char(111),
+	char(49),
+	char(0),
+	char(98),
+	char(116),
+	char(84),
+	char(121),
+	char(112),
+	char(101),
+	char(100),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(82),
+	char(105),
+	char(103),
+	char(105),
+	char(100),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(80),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(50),
+	char(80),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(80),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(50),
+	char(80),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(72),
+	char(105),
+	char(110),
+	char(103),
+	char(101),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(72),
+	char(105),
+	char(110),
+	char(103),
+	char(101),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(110),
+	char(101),
+	char(84),
+	char(119),
+	char(105),
+	char(115),
+	char(116),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(71),
+	char(101),
+	char(110),
+	char(101),
+	char(114),
+	char(105),
+	char(99),
+	char(54),
+	char(68),
+	char(111),
+	char(102),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(71),
+	char(101),
+	char(110),
+	char(101),
+	char(114),
+	char(105),
+	char(99),
+	char(54),
+	char(68),
+	char(111),
+	char(102),
+	char(83),
+	char(112),
+	char(114),
+	char(105),
+	char(110),
+	char(103),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(83),
+	char(108),
+	char(105),
+	char(100),
+	char(101),
+	char(114),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(77),
+	char(97),
+	char(116),
+	char(101),
+	char(114),
+	char(105),
+	char(97),
+	char(108),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(76),
+	char(105),
+	char(110),
+	char(107),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(70),
+	char(97),
+	char(99),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(84),
+	char(101),
+	char(116),
+	char(114),
+	char(97),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(82),
+	char(105),
+	char(103),
+	char(105),
+	char(100),
+	char(65),
+	char(110),
+	char(99),
+	char(104),
+	char(111),
+	char(114),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(67),
+	char(111),
+	char(110),
+	char(102),
+	char(105),
+	char(103),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(80),
+	char(111),
+	char(115),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(67),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(74),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(0),
+	char(0),
+	char(84),
+	char(76),
+	char(69),
+	char(78),
+	char(1),
+	char(0),
+	char(1),
+	char(0),
+	char(2),
+	char(0),
+	char(2),
+	char(0),
+	char(4),
+	char(0),
+	char(4),
+	char(0),
+	char(4),
+	char(0),
+	char(4),
+	char(0),
+	char(8),
+	char(0),
+	char(0),
+	char(0),
+	char(12),
+	char(0),
+	char(36),
+	char(0),
+	char(8),
+	char(0),
+	char(16),
+	char(0),
+	char(32),
+	char(0),
+	char(48),
+	char(0),
+	char(96),
+	char(0),
+	char(64),
+	char(0),
+	char(-128),
+	char(0),
+	char(20),
+	char(0),
+	char(48),
+	char(0),
+	char(80),
+	char(0),
+	char(16),
+	char(0),
+	char(84),
+	char(0),
+	char(-124),
+	char(0),
+	char(12),
+	char(0),
+	char(52),
+	char(0),
+	char(52),
+	char(0),
+	char(20),
+	char(0),
+	char(64),
+	char(0),
+	char(4),
+	char(0),
+	char(4),
+	char(0),
+	char(8),
+	char(0),
+	char(4),
+	char(0),
+	char(32),
+	char(0),
+	char(28),
+	char(0),
+	char(60),
+	char(0),
+	char(56),
+	char(0),
+	char(76),
+	char(0),
+	char(76),
+	char(0),
+	char(24),
+	char(0),
+	char(60),
+	char(0),
+	char(60),
+	char(0),
+	char(16),
+	char(0),
+	char(64),
+	char(0),
+	char(68),
+	char(0),
+	char(-48),
+	char(1),
+	char(0),
+	char(1),
+	char(-72),
+	char(0),
+	char(-104),
+	char(0),
+	char(104),
+	char(0),
+	char(88),
+	char(0),
+	char(-24),
+	char(1),
+	char(-96),
+	char(3),
+	char(8),
+	char(0),
+	char(52),
+	char(0),
+	char(0),
+	char(0),
+	char(84),
+	char(0),
+	char(116),
+	char(0),
+	char(92),
+	char(1),
+	char(-36),
+	char(0),
+	char(-44),
+	char(0),
+	char(-4),
+	char(0),
+	char(92),
+	char(1),
+	char(-52),
+	char(0),
+	char(16),
+	char(0),
+	char(100),
+	char(0),
+	char(20),
+	char(0),
+	char(36),
+	char(0),
+	char(100),
+	char(0),
+	char(92),
+	char(0),
+	char(104),
+	char(0),
+	char(-64),
+	char(0),
+	char(92),
+	char(1),
+	char(104),
+	char(0),
+	char(-84),
+	char(1),
+	char(83),
+	char(84),
+	char(82),
+	char(67),
+	char(65),
+	char(0),
+	char(0),
+	char(0),
+	char(10),
+	char(0),
+	char(3),
+	char(0),
+	char(4),
+	char(0),
+	char(0),
+	char(0),
+	char(4),
+	char(0),
+	char(1),
+	char(0),
+	char(9),
+	char(0),
+	char(2),
+	char(0),
+	char(11),
+	char(0),
+	char(3),
+	char(0),
+	char(10),
+	char(0),
+	char(3),
+	char(0),
+	char(10),
+	char(0),
+	char(4),
+	char(0),
+	char(10),
+	char(0),
+	char(5),
+	char(0),
+	char(12),
+	char(0),
+	char(2),
+	char(0),
+	char(9),
+	char(0),
+	char(6),
+	char(0),
+	char(9),
+	char(0),
+	char(7),
+	char(0),
+	char(13),
+	char(0),
+	char(1),
+	char(0),
+	char(7),
+	char(0),
+	char(8),
+	char(0),
+	char(14),
+	char(0),
+	char(1),
+	char(0),
+	char(8),
+	char(0),
+	char(8),
+	char(0),
+	char(15),
+	char(0),
+	char(1),
+	char(0),
+	char(13),
+	char(0),
+	char(9),
+	char(0),
+	char(16),
+	char(0),
+	char(1),
+	char(0),
+	char(14),
+	char(0),
+	char(9),
+	char(0),
+	char(17),
+	char(0),
+	char(2),
+	char(0),
+	char(15),
+	char(0),
+	char(10),
+	char(0),
+	char(13),
+	char(0),
+	char(11),
+	char(0),
+	char(18),
+	char(0),
+	char(2),
+	char(0),
+	char(16),
+	char(0),
+	char(10),
+	char(0),
+	char(14),
+	char(0),
+	char(11),
+	char(0),
+	char(19),
+	char(0),
+	char(4),
+	char(0),
+	char(4),
+	char(0),
+	char(12),
+	char(0),
+	char(4),
+	char(0),
+	char(13),
+	char(0),
+	char(2),
+	char(0),
+	char(14),
+	char(0),
+	char(2),
+	char(0),
+	char(15),
+	char(0),
+	char(20),
+	char(0),
+	char(6),
+	char(0),
+	char(13),
+	char(0),
+	char(16),
+	char(0),
+	char(13),
+	char(0),
+	char(17),
+	char(0),
+	char(4),
+	char(0),
+	char(18),
+	char(0),
+	char(4),
+	char(0),
+	char(19),
+	char(0),
+	char(4),
+	char(0),
+	char(20),
+	char(0),
+	char(0),
+	char(0),
+	char(21),
+	char(0),
+	char(21),
+	char(0),
+	char(6),
+	char(0),
+	char(14),
+	char(0),
+	char(16),
+	char(0),
+	char(14),
+	char(0),
+	char(17),
+	char(0),
+	char(4),
+	char(0),
+	char(18),
+	char(0),
+	char(4),
+	char(0),
+	char(19),
+	char(0),
+	char(4),
+	char(0),
+	char(20),
+	char(0),
+	char(0),
+	char(0),
+	char(21),
+	char(0),
+	char(22),
+	char(0),
+	char(3),
+	char(0),
+	char(2),
+	char(0),
+	char(14),
+	char(0),
+	char(2),
+	char(0),
+	char(15),
+	char(0),
+	char(4),
+	char(0),
+	char(22),
+	char(0),
+	char(23),
+	char(0),
+	char(12),
+	char(0),
+	char(13),
+	char(0),
+	char(23),
+	char(0),
+	char(13),
+	char(0),
+	char(24),
+	char(0),
+	char(13),
+	char(0),
+	char(25),
+	char(0),
+	char(4),
+	char(0),
+	char(26),
+	char(0),
+	char(4),
+	char(0),
+	char(27),
+	char(0),
+	char(4),
+	char(0),
+	char(28),
+	char(0),
+	char(4),
+	char(0),
+	char(29),
+	char(0),
+	char(20),
+	char(0),
+	char(30),
+	char(0),
+	char(22),
+	char(0),
+	char(31),
+	char(0),
+	char(19),
+	char(0),
+	char(32),
+	char(0),
+	char(4),
+	char(0),
+	char(33),
+	char(0),
+	char(4),
+	char(0),
+	char(34),
+	char(0),
+	char(24),
+	char(0),
+	char(12),
+	char(0),
+	char(14),
+	char(0),
+	char(23),
+	char(0),
+	char(14),
+	char(0),
+	char(24),
+	char(0),
+	char(14),
+	char(0),
+	char(25),
+	char(0),
+	char(4),
+	char(0),
+	char(26),
+	char(0),
+	char(4),
+	char(0),
+	char(27),
+	char(0),
+	char(4),
+	char(0),
+	char(28),
+	char(0),
+	char(4),
+	char(0),
+	char(29),
+	char(0),
+	char(21),
+	char(0),
+	char(30),
+	char(0),
+	char(22),
+	char(0),
+	char(31),
+	char(0),
+	char(4),
+	char(0),
+	char(33),
+	char(0),
+	char(4),
+	char(0),
+	char(34),
+	char(0),
+	char(19),
+	char(0),
+	char(32),
+	char(0),
+	char(25),
+	char(0),
+	char(3),
+	char(0),
+	char(0),
+	char(0),
+	char(35),
+	char(0),
+	char(4),
+	char(0),
+	char(36),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(26),
+	char(0),
+	char(5),
+	char(0),
+	char(25),
+	char(0),
+	char(38),
+	char(0),
+	char(13),
+	char(0),
+	char(39),
+	char(0),
+	char(13),
+	char(0),
+	char(40),
+	char(0),
+	char(7),
+	char(0),
+	char(41),
+	char(0),
+	char(0),
+	char(0),
+	char(21),
+	char(0),
+	char(27),
+	char(0),
+	char(5),
+	char(0),
+	char(25),
+	char(0),
+	char(38),
+	char(0),
+	char(13),
+	char(0),
+	char(39),
+	char(0),
+	char(13),
+	char(0),
+	char(42),
+	char(0),
+	char(7),
+	char(0),
+	char(43),
+	char(0),
+	char(4),
+	char(0),
+	char(44),
+	char(0),
+	char(28),
+	char(0),
+	char(2),
+	char(0),
+	char(13),
+	char(0),
+	char(45),
+	char(0),
+	char(7),
+	char(0),
+	char(46),
+	char(0),
+	char(29),
+	char(0),
+	char(4),
+	char(0),
+	char(27),
+	char(0),
+	char(47),
+	char(0),
+	char(28),
+	char(0),
+	char(48),
+	char(0),
+	char(4),
+	char(0),
+	char(49),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(30),
+	char(0),
+	char(1),
+	char(0),
+	char(4),
+	char(0),
+	char(50),
+	char(0),
+	char(31),
+	char(0),
+	char(2),
+	char(0),
+	char(2),
+	char(0),
+	char(50),
+	char(0),
+	char(0),
+	char(0),
+	char(51),
+	char(0),
+	char(32),
+	char(0),
+	char(2),
+	char(0),
+	char(2),
+	char(0),
+	char(52),
+	char(0),
+	char(0),
+	char(0),
+	char(51),
+	char(0),
+	char(33),
+	char(0),
+	char(2),
+	char(0),
+	char(0),
+	char(0),
+	char(52),
+	char(0),
+	char(0),
+	char(0),
+	char(53),
+	char(0),
+	char(34),
+	char(0),
+	char(8),
+	char(0),
+	char(13),
+	char(0),
+	char(54),
+	char(0),
+	char(14),
+	char(0),
+	char(55),
+	char(0),
+	char(30),
+	char(0),
+	char(56),
+	char(0),
+	char(32),
+	char(0),
+	char(57),
+	char(0),
+	char(33),
+	char(0),
+	char(58),
+	char(0),
+	char(31),
+	char(0),
+	char(59),
+	char(0),
+	char(4),
+	char(0),
+	char(60),
+	char(0),
+	char(4),
+	char(0),
+	char(61),
+	char(0),
+	char(35),
+	char(0),
+	char(4),
+	char(0),
+	char(34),
+	char(0),
+	char(62),
+	char(0),
+	char(13),
+	char(0),
+	char(63),
+	char(0),
+	char(4),
+	char(0),
+	char(64),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(36),
+	char(0),
+	char(7),
+	char(0),
+	char(25),
+	char(0),
+	char(38),
+	char(0),
+	char(35),
+	char(0),
+	char(65),
+	char(0),
+	char(23),
+	char(0),
+	char(66),
+	char(0),
+	char(24),
+	char(0),
+	char(67),
+	char(0),
+	char(37),
+	char(0),
+	char(68),
+	char(0),
+	char(7),
+	char(0),
+	char(43),
+	char(0),
+	char(0),
+	char(0),
+	char(69),
+	char(0),
+	char(38),
+	char(0),
+	char(2),
+	char(0),
+	char(36),
+	char(0),
+	char(70),
+	char(0),
+	char(13),
+	char(0),
+	char(39),
+	char(0),
+	char(39),
+	char(0),
+	char(4),
+	char(0),
+	char(17),
+	char(0),
+	char(71),
+	char(0),
+	char(25),
+	char(0),
+	char(72),
+	char(0),
+	char(4),
+	char(0),
+	char(73),
+	char(0),
+	char(7),
+	char(0),
+	char(74),
+	char(0),
+	char(40),
+	char(0),
+	char(4),
+	char(0),
+	char(25),
+	char(0),
+	char(38),
+	char(0),
+	char(39),
+	char(0),
+	char(75),
+	char(0),
+	char(4),
+	char(0),
+	char(76),
+	char(0),
+	char(7),
+	char(0),
+	char(43),
+	char(0),
+	char(41),
+	char(0),
+	char(3),
+	char(0),
+	char(27),
+	char(0),
+	char(47),
+	char(0),
+	char(4),
+	char(0),
+	char(77),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(42),
+	char(0),
+	char(3),
+	char(0),
+	char(27),
+	char(0),
+	char(47),
+	char(0),
+	char(4),
+	char(0),
+	char(77),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(43),
+	char(0),
+	char(4),
+	char(0),
+	char(4),
+	char(0),
+	char(78),
+	char(0),
+	char(7),
+	char(0),
+	char(79),
+	char(0),
+	char(7),
+	char(0),
+	char(80),
+	char(0),
+	char(7),
+	char(0),
+	char(81),
+	char(0),
+	char(37),
+	char(0),
+	char(14),
+	char(0),
+	char(4),
+	char(0),
+	char(82),
+	char(0),
+	char(4),
+	char(0),
+	char(83),
+	char(0),
+	char(43),
+	char(0),
+	char(84),
+	char(0),
+	char(4),
+	char(0),
+	char(85),
+	char(0),
+	char(7),
+	char(0),
+	char(86),
+	char(0),
+	char(7),
+	char(0),
+	char(87),
+	char(0),
+	char(7),
+	char(0),
+	char(88),
+	char(0),
+	char(7),
+	char(0),
+	char(89),
+	char(0),
+	char(7),
+	char(0),
+	char(90),
+	char(0),
+	char(4),
+	char(0),
+	char(91),
+	char(0),
+	char(4),
+	char(0),
+	char(92),
+	char(0),
+	char(4),
+	char(0),
+	char(93),
+	char(0),
+	char(4),
+	char(0),
+	char(94),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(44),
+	char(0),
+	char(5),
+	char(0),
+	char(25),
+	char(0),
+	char(38),
+	char(0),
+	char(35),
+	char(0),
+	char(65),
+	char(0),
+	char(13),
+	char(0),
+	char(39),
+	char(0),
+	char(7),
+	char(0),
+	char(43),
+	char(0),
+	char(4),
+	char(0),
+	char(95),
+	char(0),
+	char(45),
+	char(0),
+	char(5),
+	char(0),
+	char(27),
+	char(0),
+	char(47),
+	char(0),
+	char(13),
+	char(0),
+	char(96),
+	char(0),
+	char(14),
+	char(0),
+	char(97),
+	char(0),
+	char(4),
+	char(0),
+	char(98),
+	char(0),
+	char(0),
+	char(0),
+	char(99),
+	char(0),
+	char(46),
+	char(0),
+	char(25),
+	char(0),
+	char(9),
+	char(0),
+	char(100),
+	char(0),
+	char(9),
+	char(0),
+	char(101),
+	char(0),
+	char(25),
+	char(0),
+	char(102),
+	char(0),
+	char(0),
+	char(0),
+	char(35),
+	char(0),
+	char(18),
+	char(0),
+	char(103),
+	char(0),
+	char(18),
+	char(0),
+	char(104),
+	char(0),
+	char(14),
+	char(0),
+	char(105),
+	char(0),
+	char(14),
+	char(0),
+	char(106),
+	char(0),
+	char(14),
+	char(0),
+	char(107),
+	char(0),
+	char(8),
+	char(0),
+	char(108),
+	char(0),
+	char(8),
+	char(0),
+	char(109),
+	char(0),
+	char(8),
+	char(0),
+	char(110),
+	char(0),
+	char(8),
+	char(0),
+	char(111),
+	char(0),
+	char(8),
+	char(0),
+	char(112),
+	char(0),
+	char(8),
+	char(0),
+	char(113),
+	char(0),
+	char(8),
+	char(0),
+	char(114),
+	char(0),
+	char(8),
+	char(0),
+	char(115),
+	char(0),
+	char(4),
+	char(0),
+	char(116),
+	char(0),
+	char(4),
+	char(0),
+	char(117),
+	char(0),
+	char(4),
+	char(0),
+	char(118),
+	char(0),
+	char(4),
+	char(0),
+	char(119),
+	char(0),
+	char(4),
+	char(0),
+	char(120),
+	char(0),
+	char(4),
+	char(0),
+	char(121),
+	char(0),
+	char(4),
+	char(0),
+	char(122),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(47),
+	char(0),
+	char(25),
+	char(0),
+	char(9),
+	char(0),
+	char(100),
+	char(0),
+	char(9),
+	char(0),
+	char(101),
+	char(0),
+	char(25),
+	char(0),
+	char(102),
+	char(0),
+	char(0),
+	char(0),
+	char(35),
+	char(0),
+	char(17),
+	char(0),
+	char(103),
+	char(0),
+	char(17),
+	char(0),
+	char(104),
+	char(0),
+	char(13),
+	char(0),
+	char(105),
+	char(0),
+	char(13),
+	char(0),
+	char(106),
+	char(0),
+	char(13),
+	char(0),
+	char(107),
+	char(0),
+	char(7),
+	char(0),
+	char(108),
+	char(0),
+	char(7),
+	char(0),
+	char(109),
+	char(0),
+	char(7),
+	char(0),
+	char(110),
+	char(0),
+	char(7),
+	char(0),
+	char(111),
+	char(0),
+	char(7),
+	char(0),
+	char(112),
+	char(0),
+	char(7),
+	char(0),
+	char(113),
+	char(0),
+	char(7),
+	char(0),
+	char(114),
+	char(0),
+	char(7),
+	char(0),
+	char(115),
+	char(0),
+	char(4),
+	char(0),
+	char(116),
+	char(0),
+	char(4),
+	char(0),
+	char(117),
+	char(0),
+	char(4),
+	char(0),
+	char(118),
+	char(0),
+	char(4),
+	char(0),
+	char(119),
+	char(0),
+	char(4),
+	char(0),
+	char(120),
+	char(0),
+	char(4),
+	char(0),
+	char(121),
+	char(0),
+	char(4),
+	char(0),
+	char(122),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(48),
+	char(0),
+	char(2),
+	char(0),
+	char(49),
+	char(0),
+	char(123),
+	char(0),
+	char(14),
+	char(0),
+	char(124),
+	char(0),
+	char(50),
+	char(0),
+	char(2),
+	char(0),
+	char(51),
+	char(0),
+	char(123),
+	char(0),
+	char(13),
+	char(0),
+	char(124),
+	char(0),
+	char(52),
+	char(0),
+	char(21),
+	char(0),
+	char(47),
+	char(0),
+	char(125),
+	char(0),
+	char(15),
+	char(0),
+	char(126),
+	char(0),
+	char(13),
+	char(0),
+	char(127),
+	char(0),
+	char(13),
+	char(0),
+	char(-128),
+	char(0),
+	char(13),
+	char(0),
+	char(-127),
+	char(0),
+	char(13),
+	char(0),
+	char(-126),
+	char(0),
+	char(13),
+	char(0),
+	char(124),
+	char(0),
+	char(13),
+	char(0),
+	char(-125),
+	char(0),
+	char(13),
+	char(0),
+	char(-124),
+	char(0),
+	char(13),
+	char(0),
+	char(-123),
+	char(0),
+	char(13),
+	char(0),
+	char(-122),
+	char(0),
+	char(7),
+	char(0),
+	char(-121),
+	char(0),
+	char(7),
+	char(0),
+	char(-120),
+	char(0),
+	char(7),
+	char(0),
+	char(-119),
+	char(0),
+	char(7),
+	char(0),
+	char(-118),
+	char(0),
+	char(7),
+	char(0),
+	char(-117),
+	char(0),
+	char(7),
+	char(0),
+	char(-116),
+	char(0),
+	char(7),
+	char(0),
+	char(-115),
+	char(0),
+	char(7),
+	char(0),
+	char(-114),
+	char(0),
+	char(7),
+	char(0),
+	char(-113),
+	char(0),
+	char(4),
+	char(0),
+	char(-112),
+	char(0),
+	char(53),
+	char(0),
+	char(22),
+	char(0),
+	char(46),
+	char(0),
+	char(125),
+	char(0),
+	char(16),
+	char(0),
+	char(126),
+	char(0),
+	char(14),
+	char(0),
+	char(127),
+	char(0),
+	char(14),
+	char(0),
+	char(-128),
+	char(0),
+	char(14),
+	char(0),
+	char(-127),
+	char(0),
+	char(14),
+	char(0),
+	char(-126),
+	char(0),
+	char(14),
+	char(0),
+	char(124),
+	char(0),
+	char(14),
+	char(0),
+	char(-125),
+	char(0),
+	char(14),
+	char(0),
+	char(-124),
+	char(0),
+	char(14),
+	char(0),
+	char(-123),
+	char(0),
+	char(14),
+	char(0),
+	char(-122),
+	char(0),
+	char(8),
+	char(0),
+	char(-121),
+	char(0),
+	char(8),
+	char(0),
+	char(-120),
+	char(0),
+	char(8),
+	char(0),
+	char(-119),
+	char(0),
+	char(8),
+	char(0),
+	char(-118),
+	char(0),
+	char(8),
+	char(0),
+	char(-117),
+	char(0),
+	char(8),
+	char(0),
+	char(-116),
+	char(0),
+	char(8),
+	char(0),
+	char(-115),
+	char(0),
+	char(8),
+	char(0),
+	char(-114),
+	char(0),
+	char(8),
+	char(0),
+	char(-113),
+	char(0),
+	char(4),
+	char(0),
+	char(-112),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(54),
+	char(0),
+	char(2),
+	char(0),
+	char(4),
+	char(0),
+	char(-111),
+	char(0),
+	char(4),
+	char(0),
+	char(-110),
+	char(0),
+	char(55),
+	char(0),
+	char(13),
+	char(0),
+	char(56),
+	char(0),
+	char(-109),
+	char(0),
+	char(56),
+	char(0),
+	char(-108),
+	char(0),
+	char(0),
+	char(0),
+	char(35),
+	char(0),
+	char(4),
+	char(0),
+	char(-107),
+	char(0),
+	char(4),
+	char(0),
+	char(-106),
+	char(0),
+	char(4),
+	char(0),
+	char(-105),
+	char(0),
+	char(4),
+	char(0),
+	char(-104),
+	char(0),
+	char(7),
+	char(0),
+	char(-103),
+	char(0),
+	char(7),
+	char(0),
+	char(-102),
+	char(0),
+	char(4),
+	char(0),
+	char(-101),
+	char(0),
+	char(4),
+	char(0),
+	char(-100),
+	char(0),
+	char(7),
+	char(0),
+	char(-99),
+	char(0),
+	char(4),
+	char(0),
+	char(-98),
+	char(0),
+	char(57),
+	char(0),
+	char(3),
+	char(0),
+	char(55),
+	char(0),
+	char(-97),
+	char(0),
+	char(13),
+	char(0),
+	char(-96),
+	char(0),
+	char(13),
+	char(0),
+	char(-95),
+	char(0),
+	char(58),
+	char(0),
+	char(3),
+	char(0),
+	char(55),
+	char(0),
+	char(-97),
+	char(0),
+	char(14),
+	char(0),
+	char(-96),
+	char(0),
+	char(14),
+	char(0),
+	char(-95),
+	char(0),
+	char(59),
+	char(0),
+	char(13),
+	char(0),
+	char(55),
+	char(0),
+	char(-97),
+	char(0),
+	char(18),
+	char(0),
+	char(-94),
+	char(0),
+	char(18),
+	char(0),
+	char(-93),
+	char(0),
+	char(4),
+	char(0),
+	char(-92),
+	char(0),
+	char(4),
+	char(0),
+	char(-91),
+	char(0),
+	char(4),
+	char(0),
+	char(-90),
+	char(0),
+	char(7),
+	char(0),
+	char(-89),
+	char(0),
+	char(7),
+	char(0),
+	char(-88),
+	char(0),
+	char(7),
+	char(0),
+	char(-87),
+	char(0),
+	char(7),
+	char(0),
+	char(-86),
+	char(0),
+	char(7),
+	char(0),
+	char(-85),
+	char(0),
+	char(7),
+	char(0),
+	char(-84),
+	char(0),
+	char(7),
+	char(0),
+	char(-83),
+	char(0),
+	char(60),
+	char(0),
+	char(13),
+	char(0),
+	char(55),
+	char(0),
+	char(-97),
+	char(0),
+	char(17),
+	char(0),
+	char(-94),
+	char(0),
+	char(17),
+	char(0),
+	char(-93),
+	char(0),
+	char(4),
+	char(0),
+	char(-92),
+	char(0),
+	char(4),
+	char(0),
+	char(-91),
+	char(0),
+	char(4),
+	char(0),
+	char(-90),
+	char(0),
+	char(7),
+	char(0),
+	char(-89),
+	char(0),
+	char(7),
+	char(0),
+	char(-88),
+	char(0),
+	char(7),
+	char(0),
+	char(-87),
+	char(0),
+	char(7),
+	char(0),
+	char(-86),
+	char(0),
+	char(7),
+	char(0),
+	char(-85),
+	char(0),
+	char(7),
+	char(0),
+	char(-84),
+	char(0),
+	char(7),
+	char(0),
+	char(-83),
+	char(0),
+	char(61),
+	char(0),
+	char(11),
+	char(0),
+	char(55),
+	char(0),
+	char(-97),
+	char(0),
+	char(17),
+	char(0),
+	char(-94),
+	char(0),
+	char(17),
+	char(0),
+	char(-93),
+	char(0),
+	char(7),
+	char(0),
+	char(-82),
+	char(0),
+	char(7),
+	char(0),
+	char(-81),
+	char(0),
+	char(7),
+	char(0),
+	char(-80),
+	char(0),
+	char(7),
+	char(0),
+	char(-85),
+	char(0),
+	char(7),
+	char(0),
+	char(-84),
+	char(0),
+	char(7),
+	char(0),
+	char(-83),
+	char(0),
+	char(7),
+	char(0),
+	char(-79),
+	char(0),
+	char(0),
+	char(0),
+	char(21),
+	char(0),
+	char(62),
+	char(0),
+	char(9),
+	char(0),
+	char(55),
+	char(0),
+	char(-97),
+	char(0),
+	char(17),
+	char(0),
+	char(-94),
+	char(0),
+	char(17),
+	char(0),
+	char(-93),
+	char(0),
+	char(13),
+	char(0),
+	char(-78),
+	char(0),
+	char(13),
+	char(0),
+	char(-77),
+	char(0),
+	char(13),
+	char(0),
+	char(-76),
+	char(0),
+	char(13),
+	char(0),
+	char(-75),
+	char(0),
+	char(4),
+	char(0),
+	char(-74),
+	char(0),
+	char(4),
+	char(0),
+	char(-73),
+	char(0),
+	char(63),
+	char(0),
+	char(5),
+	char(0),
+	char(62),
+	char(0),
+	char(-72),
+	char(0),
+	char(4),
+	char(0),
+	char(-71),
+	char(0),
+	char(7),
+	char(0),
+	char(-70),
+	char(0),
+	char(7),
+	char(0),
+	char(-69),
+	char(0),
+	char(7),
+	char(0),
+	char(-68),
+	char(0),
+	char(64),
+	char(0),
+	char(9),
+	char(0),
+	char(55),
+	char(0),
+	char(-97),
+	char(0),
+	char(17),
+	char(0),
+	char(-94),
+	char(0),
+	char(17),
+	char(0),
+	char(-93),
+	char(0),
+	char(7),
+	char(0),
+	char(-78),
+	char(0),
+	char(7),
+	char(0),
+	char(-77),
+	char(0),
+	char(7),
+	char(0),
+	char(-76),
+	char(0),
+	char(7),
+	char(0),
+	char(-75),
+	char(0),
+	char(4),
+	char(0),
+	char(-74),
+	char(0),
+	char(4),
+	char(0),
+	char(-73),
+	char(0),
+	char(49),
+	char(0),
+	char(22),
+	char(0),
+	char(8),
+	char(0),
+	char(-67),
+	char(0),
+	char(8),
+	char(0),
+	char(-79),
+	char(0),
+	char(8),
+	char(0),
+	char(110),
+	char(0),
+	char(8),
+	char(0),
+	char(-66),
+	char(0),
+	char(8),
+	char(0),
+	char(112),
+	char(0),
+	char(8),
+	char(0),
+	char(-65),
+	char(0),
+	char(8),
+	char(0),
+	char(-64),
+	char(0),
+	char(8),
+	char(0),
+	char(-63),
+	char(0),
+	char(8),
+	char(0),
+	char(-62),
+	char(0),
+	char(8),
+	char(0),
+	char(-61),
+	char(0),
+	char(8),
+	char(0),
+	char(-60),
+	char(0),
+	char(8),
+	char(0),
+	char(-59),
+	char(0),
+	char(8),
+	char(0),
+	char(-58),
+	char(0),
+	char(8),
+	char(0),
+	char(-57),
+	char(0),
+	char(8),
+	char(0),
+	char(-56),
+	char(0),
+	char(8),
+	char(0),
+	char(-55),
+	char(0),
+	char(4),
+	char(0),
+	char(-54),
+	char(0),
+	char(4),
+	char(0),
+	char(-53),
+	char(0),
+	char(4),
+	char(0),
+	char(-52),
+	char(0),
+	char(4),
+	char(0),
+	char(-51),
+	char(0),
+	char(4),
+	char(0),
+	char(-50),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(51),
+	char(0),
+	char(22),
+	char(0),
+	char(7),
+	char(0),
+	char(-67),
+	char(0),
+	char(7),
+	char(0),
+	char(-79),
+	char(0),
+	char(7),
+	char(0),
+	char(110),
+	char(0),
+	char(7),
+	char(0),
+	char(-66),
+	char(0),
+	char(7),
+	char(0),
+	char(112),
+	char(0),
+	char(7),
+	char(0),
+	char(-65),
+	char(0),
+	char(7),
+	char(0),
+	char(-64),
+	char(0),
+	char(7),
+	char(0),
+	char(-63),
+	char(0),
+	char(7),
+	char(0),
+	char(-62),
+	char(0),
+	char(7),
+	char(0),
+	char(-61),
+	char(0),
+	char(7),
+	char(0),
+	char(-60),
+	char(0),
+	char(7),
+	char(0),
+	char(-59),
+	char(0),
+	char(7),
+	char(0),
+	char(-58),
+	char(0),
+	char(7),
+	char(0),
+	char(-57),
+	char(0),
+	char(7),
+	char(0),
+	char(-56),
+	char(0),
+	char(7),
+	char(0),
+	char(-55),
+	char(0),
+	char(4),
+	char(0),
+	char(-54),
+	char(0),
+	char(4),
+	char(0),
+	char(-53),
+	char(0),
+	char(4),
+	char(0),
+	char(-52),
+	char(0),
+	char(4),
+	char(0),
+	char(-51),
+	char(0),
+	char(4),
+	char(0),
+	char(-50),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(65),
+	char(0),
+	char(4),
+	char(0),
+	char(7),
+	char(0),
+	char(-49),
+	char(0),
+	char(7),
+	char(0),
+	char(-48),
+	char(0),
+	char(7),
+	char(0),
+	char(-47),
+	char(0),
+	char(4),
+	char(0),
+	char(78),
+	char(0),
+	char(66),
+	char(0),
+	char(10),
+	char(0),
+	char(65),
+	char(0),
+	char(-46),
+	char(0),
+	char(13),
+	char(0),
+	char(-45),
+	char(0),
+	char(13),
+	char(0),
+	char(-44),
+	char(0),
+	char(13),
+	char(0),
+	char(-43),
+	char(0),
+	char(13),
+	char(0),
+	char(-42),
+	char(0),
+	char(13),
+	char(0),
+	char(-41),
+	char(0),
+	char(7),
+	char(0),
+	char(-121),
+	char(0),
+	char(7),
+	char(0),
+	char(-40),
+	char(0),
+	char(4),
+	char(0),
+	char(-39),
+	char(0),
+	char(4),
+	char(0),
+	char(53),
+	char(0),
+	char(67),
+	char(0),
+	char(4),
+	char(0),
+	char(65),
+	char(0),
+	char(-46),
+	char(0),
+	char(4),
+	char(0),
+	char(-38),
+	char(0),
+	char(7),
+	char(0),
+	char(-37),
+	char(0),
+	char(4),
+	char(0),
+	char(-36),
+	char(0),
+	char(68),
+	char(0),
+	char(4),
+	char(0),
+	char(13),
+	char(0),
+	char(-41),
+	char(0),
+	char(65),
+	char(0),
+	char(-46),
+	char(0),
+	char(4),
+	char(0),
+	char(-35),
+	char(0),
+	char(7),
+	char(0),
+	char(-34),
+	char(0),
+	char(69),
+	char(0),
+	char(7),
+	char(0),
+	char(13),
+	char(0),
+	char(-33),
+	char(0),
+	char(65),
+	char(0),
+	char(-46),
+	char(0),
+	char(4),
+	char(0),
+	char(-32),
+	char(0),
+	char(7),
+	char(0),
+	char(-31),
+	char(0),
+	char(7),
+	char(0),
+	char(-30),
+	char(0),
+	char(7),
+	char(0),
+	char(-29),
+	char(0),
+	char(4),
+	char(0),
+	char(53),
+	char(0),
+	char(70),
+	char(0),
+	char(6),
+	char(0),
+	char(15),
+	char(0),
+	char(-28),
+	char(0),
+	char(13),
+	char(0),
+	char(-30),
+	char(0),
+	char(13),
+	char(0),
+	char(-27),
+	char(0),
+	char(56),
+	char(0),
+	char(-26),
+	char(0),
+	char(4),
+	char(0),
+	char(-25),
+	char(0),
+	char(7),
+	char(0),
+	char(-29),
+	char(0),
+	char(71),
+	char(0),
+	char(26),
+	char(0),
+	char(4),
+	char(0),
+	char(-24),
+	char(0),
+	char(7),
+	char(0),
+	char(-23),
+	char(0),
+	char(7),
+	char(0),
+	char(-79),
+	char(0),
+	char(7),
+	char(0),
+	char(-22),
+	char(0),
+	char(7),
+	char(0),
+	char(-21),
+	char(0),
+	char(7),
+	char(0),
+	char(-20),
+	char(0),
+	char(7),
+	char(0),
+	char(-19),
+	char(0),
+	char(7),
+	char(0),
+	char(-18),
+	char(0),
+	char(7),
+	char(0),
+	char(-17),
+	char(0),
+	char(7),
+	char(0),
+	char(-16),
+	char(0),
+	char(7),
+	char(0),
+	char(-15),
+	char(0),
+	char(7),
+	char(0),
+	char(-14),
+	char(0),
+	char(7),
+	char(0),
+	char(-13),
+	char(0),
+	char(7),
+	char(0),
+	char(-12),
+	char(0),
+	char(7),
+	char(0),
+	char(-11),
+	char(0),
+	char(7),
+	char(0),
+	char(-10),
+	char(0),
+	char(7),
+	char(0),
+	char(-9),
+	char(0),
+	char(7),
+	char(0),
+	char(-8),
+	char(0),
+	char(7),
+	char(0),
+	char(-7),
+	char(0),
+	char(7),
+	char(0),
+	char(-6),
+	char(0),
+	char(7),
+	char(0),
+	char(-5),
+	char(0),
+	char(4),
+	char(0),
+	char(-4),
+	char(0),
+	char(4),
+	char(0),
+	char(-3),
+	char(0),
+	char(4),
+	char(0),
+	char(-2),
+	char(0),
+	char(4),
+	char(0),
+	char(-1),
+	char(0),
+	char(4),
+	char(0),
+	char(117),
+	char(0),
+	char(72),
+	char(0),
+	char(12),
+	char(0),
+	char(15),
+	char(0),
+	char(0),
+	char(1),
+	char(15),
+	char(0),
+	char(1),
+	char(1),
+	char(15),
+	char(0),
+	char(2),
+	char(1),
+	char(13),
+	char(0),
+	char(3),
+	char(1),
+	char(13),
+	char(0),
+	char(4),
+	char(1),
+	char(7),
+	char(0),
+	char(5),
+	char(1),
+	char(4),
+	char(0),
+	char(6),
+	char(1),
+	char(4),
+	char(0),
+	char(7),
+	char(1),
+	char(4),
+	char(0),
+	char(8),
+	char(1),
+	char(4),
+	char(0),
+	char(9),
+	char(1),
+	char(7),
+	char(0),
+	char(-31),
+	char(0),
+	char(4),
+	char(0),
+	char(53),
+	char(0),
+	char(73),
+	char(0),
+	char(27),
+	char(0),
+	char(17),
+	char(0),
+	char(10),
+	char(1),
+	char(15),
+	char(0),
+	char(11),
+	char(1),
+	char(15),
+	char(0),
+	char(12),
+	char(1),
+	char(13),
+	char(0),
+	char(3),
+	char(1),
+	char(13),
+	char(0),
+	char(13),
+	char(1),
+	char(13),
+	char(0),
+	char(14),
+	char(1),
+	char(13),
+	char(0),
+	char(15),
+	char(1),
+	char(13),
+	char(0),
+	char(16),
+	char(1),
+	char(13),
+	char(0),
+	char(17),
+	char(1),
+	char(4),
+	char(0),
+	char(18),
+	char(1),
+	char(7),
+	char(0),
+	char(19),
+	char(1),
+	char(4),
+	char(0),
+	char(20),
+	char(1),
+	char(4),
+	char(0),
+	char(21),
+	char(1),
+	char(4),
+	char(0),
+	char(22),
+	char(1),
+	char(7),
+	char(0),
+	char(23),
+	char(1),
+	char(7),
+	char(0),
+	char(24),
+	char(1),
+	char(4),
+	char(0),
+	char(25),
+	char(1),
+	char(4),
+	char(0),
+	char(26),
+	char(1),
+	char(7),
+	char(0),
+	char(27),
+	char(1),
+	char(7),
+	char(0),
+	char(28),
+	char(1),
+	char(7),
+	char(0),
+	char(29),
+	char(1),
+	char(7),
+	char(0),
+	char(30),
+	char(1),
+	char(7),
+	char(0),
+	char(31),
+	char(1),
+	char(7),
+	char(0),
+	char(32),
+	char(1),
+	char(4),
+	char(0),
+	char(33),
+	char(1),
+	char(4),
+	char(0),
+	char(34),
+	char(1),
+	char(4),
+	char(0),
+	char(35),
+	char(1),
+	char(74),
+	char(0),
+	char(12),
+	char(0),
+	char(9),
+	char(0),
+	char(36),
+	char(1),
+	char(9),
+	char(0),
+	char(37),
+	char(1),
+	char(13),
+	char(0),
+	char(38),
+	char(1),
+	char(7),
+	char(0),
+	char(39),
+	char(1),
+	char(7),
+	char(0),
+	char(-63),
+	char(0),
+	char(7),
+	char(0),
+	char(40),
+	char(1),
+	char(4),
+	char(0),
+	char(41),
+	char(1),
+	char(13),
+	char(0),
+	char(42),
+	char(1),
+	char(4),
+	char(0),
+	char(43),
+	char(1),
+	char(4),
+	char(0),
+	char(44),
+	char(1),
+	char(4),
+	char(0),
+	char(45),
+	char(1),
+	char(4),
+	char(0),
+	char(53),
+	char(0),
+	char(75),
+	char(0),
+	char(19),
+	char(0),
+	char(47),
+	char(0),
+	char(125),
+	char(0),
+	char(72),
+	char(0),
+	char(46),
+	char(1),
+	char(65),
+	char(0),
+	char(47),
+	char(1),
+	char(66),
+	char(0),
+	char(48),
+	char(1),
+	char(67),
+	char(0),
+	char(49),
+	char(1),
+	char(68),
+	char(0),
+	char(50),
+	char(1),
+	char(69),
+	char(0),
+	char(51),
+	char(1),
+	char(70),
+	char(0),
+	char(52),
+	char(1),
+	char(73),
+	char(0),
+	char(53),
+	char(1),
+	char(74),
+	char(0),
+	char(54),
+	char(1),
+	char(4),
+	char(0),
+	char(55),
+	char(1),
+	char(4),
+	char(0),
+	char(21),
+	char(1),
+	char(4),
+	char(0),
+	char(56),
+	char(1),
+	char(4),
+	char(0),
+	char(57),
+	char(1),
+	char(4),
+	char(0),
+	char(58),
+	char(1),
+	char(4),
+	char(0),
+	char(59),
+	char(1),
+	char(4),
+	char(0),
+	char(60),
+	char(1),
+	char(4),
+	char(0),
+	char(61),
+	char(1),
+	char(71),
+	char(0),
+	char(62),
+	char(1),
+};
+int b3s_bulletDNAlen = sizeof(b3s_bulletDNAstr);
+char b3s_bulletDNAstr64[] = {
+	char(83),
+	char(68),
+	char(78),
+	char(65),
+	char(78),
+	char(65),
+	char(77),
+	char(69),
+	char(63),
+	char(1),
+	char(0),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(105),
+	char(122),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(97),
+	char(112),
+	char(97),
+	char(99),
+	char(105),
+	char(116),
+	char(121),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(100),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(79),
+	char(98),
+	char(106),
+	char(101),
+	char(99),
+	char(116),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(115),
+	char(0),
+	char(42),
+	char(102),
+	char(105),
+	char(114),
+	char(115),
+	char(116),
+	char(0),
+	char(42),
+	char(108),
+	char(97),
+	char(115),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(102),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(115),
+	char(91),
+	char(52),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(108),
+	char(91),
+	char(51),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(97),
+	char(115),
+	char(105),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(111),
+	char(114),
+	char(105),
+	char(103),
+	char(105),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(111),
+	char(111),
+	char(116),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(117),
+	char(98),
+	char(116),
+	char(114),
+	char(101),
+	char(101),
+	char(83),
+	char(105),
+	char(122),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(113),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(65),
+	char(97),
+	char(98),
+	char(98),
+	char(77),
+	char(105),
+	char(110),
+	char(91),
+	char(51),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(113),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(65),
+	char(97),
+	char(98),
+	char(98),
+	char(77),
+	char(97),
+	char(120),
+	char(91),
+	char(51),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(97),
+	char(98),
+	char(98),
+	char(77),
+	char(105),
+	char(110),
+	char(79),
+	char(114),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(97),
+	char(98),
+	char(98),
+	char(77),
+	char(97),
+	char(120),
+	char(79),
+	char(114),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(115),
+	char(99),
+	char(97),
+	char(112),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(117),
+	char(98),
+	char(80),
+	char(97),
+	char(114),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(114),
+	char(105),
+	char(97),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(97),
+	char(100),
+	char(91),
+	char(52),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(115),
+	char(99),
+	char(97),
+	char(112),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(79),
+	char(114),
+	char(84),
+	char(114),
+	char(105),
+	char(97),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(118),
+	char(104),
+	char(65),
+	char(97),
+	char(98),
+	char(98),
+	char(77),
+	char(105),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(118),
+	char(104),
+	char(65),
+	char(97),
+	char(98),
+	char(98),
+	char(77),
+	char(97),
+	char(120),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(118),
+	char(104),
+	char(81),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(117),
+	char(114),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(0),
+	char(109),
+	char(95),
+	char(117),
+	char(115),
+	char(101),
+	char(81),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(67),
+	char(111),
+	char(110),
+	char(116),
+	char(105),
+	char(103),
+	char(117),
+	char(111),
+	char(117),
+	char(115),
+	char(76),
+	char(101),
+	char(97),
+	char(102),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(81),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(67),
+	char(111),
+	char(110),
+	char(116),
+	char(105),
+	char(103),
+	char(117),
+	char(111),
+	char(117),
+	char(115),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(110),
+	char(116),
+	char(105),
+	char(103),
+	char(117),
+	char(111),
+	char(117),
+	char(115),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(115),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(113),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(67),
+	char(111),
+	char(110),
+	char(116),
+	char(105),
+	char(103),
+	char(117),
+	char(111),
+	char(117),
+	char(115),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(115),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(115),
+	char(117),
+	char(98),
+	char(84),
+	char(114),
+	char(101),
+	char(101),
+	char(73),
+	char(110),
+	char(102),
+	char(111),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(114),
+	char(97),
+	char(118),
+	char(101),
+	char(114),
+	char(115),
+	char(97),
+	char(108),
+	char(77),
+	char(111),
+	char(100),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(83),
+	char(117),
+	char(98),
+	char(116),
+	char(114),
+	char(101),
+	char(101),
+	char(72),
+	char(101),
+	char(97),
+	char(100),
+	char(101),
+	char(114),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(110),
+	char(97),
+	char(109),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(84),
+	char(121),
+	char(112),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(97),
+	char(100),
+	char(100),
+	char(105),
+	char(110),
+	char(103),
+	char(91),
+	char(52),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(111),
+	char(99),
+	char(97),
+	char(108),
+	char(83),
+	char(99),
+	char(97),
+	char(108),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(108),
+	char(97),
+	char(110),
+	char(101),
+	char(78),
+	char(111),
+	char(114),
+	char(109),
+	char(97),
+	char(108),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(108),
+	char(97),
+	char(110),
+	char(101),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(97),
+	char(110),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(109),
+	char(112),
+	char(108),
+	char(105),
+	char(99),
+	char(105),
+	char(116),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(105),
+	char(109),
+	char(101),
+	char(110),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(77),
+	char(97),
+	char(114),
+	char(103),
+	char(105),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(97),
+	char(100),
+	char(100),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(111),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(97),
+	char(100),
+	char(105),
+	char(117),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(110),
+	char(118),
+	char(101),
+	char(120),
+	char(73),
+	char(110),
+	char(116),
+	char(101),
+	char(114),
+	char(110),
+	char(97),
+	char(108),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(108),
+	char(111),
+	char(99),
+	char(97),
+	char(108),
+	char(80),
+	char(111),
+	char(115),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(65),
+	char(114),
+	char(114),
+	char(97),
+	char(121),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(111),
+	char(99),
+	char(97),
+	char(108),
+	char(80),
+	char(111),
+	char(115),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(65),
+	char(114),
+	char(114),
+	char(97),
+	char(121),
+	char(83),
+	char(105),
+	char(122),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(118),
+	char(97),
+	char(108),
+	char(117),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(97),
+	char(100),
+	char(91),
+	char(50),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(118),
+	char(97),
+	char(108),
+	char(117),
+	char(101),
+	char(115),
+	char(91),
+	char(51),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(97),
+	char(100),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(118),
+	char(101),
+	char(114),
+	char(116),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(51),
+	char(102),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(118),
+	char(101),
+	char(114),
+	char(116),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(51),
+	char(100),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(105),
+	char(110),
+	char(100),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(51),
+	char(50),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(51),
+	char(105),
+	char(110),
+	char(100),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(49),
+	char(54),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(51),
+	char(105),
+	char(110),
+	char(100),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(56),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(105),
+	char(110),
+	char(100),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(49),
+	char(54),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(84),
+	char(114),
+	char(105),
+	char(97),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(86),
+	char(101),
+	char(114),
+	char(116),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(109),
+	char(101),
+	char(115),
+	char(104),
+	char(80),
+	char(97),
+	char(114),
+	char(116),
+	char(115),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(99),
+	char(97),
+	char(108),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(77),
+	char(101),
+	char(115),
+	char(104),
+	char(80),
+	char(97),
+	char(114),
+	char(116),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(109),
+	char(101),
+	char(115),
+	char(104),
+	char(73),
+	char(110),
+	char(116),
+	char(101),
+	char(114),
+	char(102),
+	char(97),
+	char(99),
+	char(101),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(113),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(66),
+	char(118),
+	char(104),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(113),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(66),
+	char(118),
+	char(104),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(116),
+	char(114),
+	char(105),
+	char(97),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(73),
+	char(110),
+	char(102),
+	char(111),
+	char(77),
+	char(97),
+	char(112),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(97),
+	char(100),
+	char(51),
+	char(91),
+	char(52),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(114),
+	char(105),
+	char(109),
+	char(101),
+	char(115),
+	char(104),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(114),
+	char(97),
+	char(110),
+	char(115),
+	char(102),
+	char(111),
+	char(114),
+	char(109),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(99),
+	char(104),
+	char(105),
+	char(108),
+	char(100),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(104),
+	char(105),
+	char(108),
+	char(100),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(84),
+	char(121),
+	char(112),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(104),
+	char(105),
+	char(108),
+	char(100),
+	char(77),
+	char(97),
+	char(114),
+	char(103),
+	char(105),
+	char(110),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(99),
+	char(104),
+	char(105),
+	char(108),
+	char(100),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(67),
+	char(104),
+	char(105),
+	char(108),
+	char(100),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(117),
+	char(112),
+	char(65),
+	char(120),
+	char(105),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(102),
+	char(108),
+	char(97),
+	char(103),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(100),
+	char(103),
+	char(101),
+	char(86),
+	char(48),
+	char(86),
+	char(49),
+	char(65),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(100),
+	char(103),
+	char(101),
+	char(86),
+	char(49),
+	char(86),
+	char(50),
+	char(65),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(100),
+	char(103),
+	char(101),
+	char(86),
+	char(50),
+	char(86),
+	char(48),
+	char(65),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(104),
+	char(97),
+	char(115),
+	char(104),
+	char(84),
+	char(97),
+	char(98),
+	char(108),
+	char(101),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(110),
+	char(101),
+	char(120),
+	char(116),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(118),
+	char(97),
+	char(108),
+	char(117),
+	char(101),
+	char(65),
+	char(114),
+	char(114),
+	char(97),
+	char(121),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(107),
+	char(101),
+	char(121),
+	char(65),
+	char(114),
+	char(114),
+	char(97),
+	char(121),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(110),
+	char(118),
+	char(101),
+	char(120),
+	char(69),
+	char(112),
+	char(115),
+	char(105),
+	char(108),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(108),
+	char(97),
+	char(110),
+	char(97),
+	char(114),
+	char(69),
+	char(112),
+	char(115),
+	char(105),
+	char(108),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(113),
+	char(117),
+	char(97),
+	char(108),
+	char(86),
+	char(101),
+	char(114),
+	char(116),
+	char(101),
+	char(120),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(100),
+	char(103),
+	char(101),
+	char(68),
+	char(105),
+	char(115),
+	char(116),
+	char(97),
+	char(110),
+	char(99),
+	char(101),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(122),
+	char(101),
+	char(114),
+	char(111),
+	char(65),
+	char(114),
+	char(101),
+	char(97),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(101),
+	char(120),
+	char(116),
+	char(83),
+	char(105),
+	char(122),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(104),
+	char(97),
+	char(115),
+	char(104),
+	char(84),
+	char(97),
+	char(98),
+	char(108),
+	char(101),
+	char(83),
+	char(105),
+	char(122),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(86),
+	char(97),
+	char(108),
+	char(117),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(75),
+	char(101),
+	char(121),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(103),
+	char(105),
+	char(109),
+	char(112),
+	char(97),
+	char(99),
+	char(116),
+	char(83),
+	char(117),
+	char(98),
+	char(84),
+	char(121),
+	char(112),
+	char(101),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(117),
+	char(110),
+	char(115),
+	char(99),
+	char(97),
+	char(108),
+	char(101),
+	char(100),
+	char(80),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(115),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(117),
+	char(110),
+	char(115),
+	char(99),
+	char(97),
+	char(108),
+	char(101),
+	char(100),
+	char(80),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(115),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(80),
+	char(116),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(85),
+	char(110),
+	char(115),
+	char(99),
+	char(97),
+	char(108),
+	char(101),
+	char(100),
+	char(80),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(97),
+	char(100),
+	char(100),
+	char(105),
+	char(110),
+	char(103),
+	char(51),
+	char(91),
+	char(52),
+	char(93),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(98),
+	char(114),
+	char(111),
+	char(97),
+	char(100),
+	char(112),
+	char(104),
+	char(97),
+	char(115),
+	char(101),
+	char(72),
+	char(97),
+	char(110),
+	char(100),
+	char(108),
+	char(101),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(114),
+	char(111),
+	char(111),
+	char(116),
+	char(67),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(119),
+	char(111),
+	char(114),
+	char(108),
+	char(100),
+	char(84),
+	char(114),
+	char(97),
+	char(110),
+	char(115),
+	char(102),
+	char(111),
+	char(114),
+	char(109),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(110),
+	char(116),
+	char(101),
+	char(114),
+	char(112),
+	char(111),
+	char(108),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(87),
+	char(111),
+	char(114),
+	char(108),
+	char(100),
+	char(84),
+	char(114),
+	char(97),
+	char(110),
+	char(115),
+	char(102),
+	char(111),
+	char(114),
+	char(109),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(110),
+	char(116),
+	char(101),
+	char(114),
+	char(112),
+	char(111),
+	char(108),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(76),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(86),
+	char(101),
+	char(108),
+	char(111),
+	char(99),
+	char(105),
+	char(116),
+	char(121),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(110),
+	char(116),
+	char(101),
+	char(114),
+	char(112),
+	char(111),
+	char(108),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(65),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(86),
+	char(101),
+	char(108),
+	char(111),
+	char(99),
+	char(105),
+	char(116),
+	char(121),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(105),
+	char(115),
+	char(111),
+	char(116),
+	char(114),
+	char(111),
+	char(112),
+	char(105),
+	char(99),
+	char(70),
+	char(114),
+	char(105),
+	char(99),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(110),
+	char(116),
+	char(97),
+	char(99),
+	char(116),
+	char(80),
+	char(114),
+	char(111),
+	char(99),
+	char(101),
+	char(115),
+	char(115),
+	char(105),
+	char(110),
+	char(103),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(100),
+	char(101),
+	char(97),
+	char(99),
+	char(116),
+	char(105),
+	char(118),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(84),
+	char(105),
+	char(109),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(102),
+	char(114),
+	char(105),
+	char(99),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(110),
+	char(103),
+	char(70),
+	char(114),
+	char(105),
+	char(99),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(101),
+	char(115),
+	char(116),
+	char(105),
+	char(116),
+	char(117),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(104),
+	char(105),
+	char(116),
+	char(70),
+	char(114),
+	char(97),
+	char(99),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(99),
+	char(100),
+	char(83),
+	char(119),
+	char(101),
+	char(112),
+	char(116),
+	char(83),
+	char(112),
+	char(104),
+	char(101),
+	char(114),
+	char(101),
+	char(82),
+	char(97),
+	char(100),
+	char(105),
+	char(117),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(99),
+	char(100),
+	char(77),
+	char(111),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(104),
+	char(97),
+	char(115),
+	char(65),
+	char(110),
+	char(105),
+	char(115),
+	char(111),
+	char(116),
+	char(114),
+	char(111),
+	char(112),
+	char(105),
+	char(99),
+	char(70),
+	char(114),
+	char(105),
+	char(99),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(70),
+	char(108),
+	char(97),
+	char(103),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(115),
+	char(108),
+	char(97),
+	char(110),
+	char(100),
+	char(84),
+	char(97),
+	char(103),
+	char(49),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(109),
+	char(112),
+	char(97),
+	char(110),
+	char(105),
+	char(111),
+	char(110),
+	char(73),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(99),
+	char(116),
+	char(105),
+	char(118),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(83),
+	char(116),
+	char(97),
+	char(116),
+	char(101),
+	char(49),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(110),
+	char(116),
+	char(101),
+	char(114),
+	char(110),
+	char(97),
+	char(108),
+	char(84),
+	char(121),
+	char(112),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(104),
+	char(101),
+	char(99),
+	char(107),
+	char(67),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(100),
+	char(101),
+	char(87),
+	char(105),
+	char(116),
+	char(104),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(111),
+	char(108),
+	char(118),
+	char(101),
+	char(114),
+	char(73),
+	char(110),
+	char(102),
+	char(111),
+	char(0),
+	char(109),
+	char(95),
+	char(103),
+	char(114),
+	char(97),
+	char(118),
+	char(105),
+	char(116),
+	char(121),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(79),
+	char(98),
+	char(106),
+	char(101),
+	char(99),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(110),
+	char(118),
+	char(73),
+	char(110),
+	char(101),
+	char(114),
+	char(116),
+	char(105),
+	char(97),
+	char(84),
+	char(101),
+	char(110),
+	char(115),
+	char(111),
+	char(114),
+	char(87),
+	char(111),
+	char(114),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(86),
+	char(101),
+	char(108),
+	char(111),
+	char(99),
+	char(105),
+	char(116),
+	char(121),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(86),
+	char(101),
+	char(108),
+	char(111),
+	char(99),
+	char(105),
+	char(116),
+	char(121),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(70),
+	char(97),
+	char(99),
+	char(116),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(70),
+	char(97),
+	char(99),
+	char(116),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(103),
+	char(114),
+	char(97),
+	char(118),
+	char(105),
+	char(116),
+	char(121),
+	char(95),
+	char(97),
+	char(99),
+	char(99),
+	char(101),
+	char(108),
+	char(101),
+	char(114),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(110),
+	char(118),
+	char(73),
+	char(110),
+	char(101),
+	char(114),
+	char(116),
+	char(105),
+	char(97),
+	char(76),
+	char(111),
+	char(99),
+	char(97),
+	char(108),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(111),
+	char(116),
+	char(97),
+	char(108),
+	char(70),
+	char(111),
+	char(114),
+	char(99),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(111),
+	char(116),
+	char(97),
+	char(108),
+	char(84),
+	char(111),
+	char(114),
+	char(113),
+	char(117),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(110),
+	char(118),
+	char(101),
+	char(114),
+	char(115),
+	char(101),
+	char(77),
+	char(97),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(68),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(68),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(100),
+	char(100),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(97),
+	char(108),
+	char(68),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(70),
+	char(97),
+	char(99),
+	char(116),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(100),
+	char(100),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(97),
+	char(108),
+	char(76),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(68),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(83),
+	char(113),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(100),
+	char(100),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(97),
+	char(108),
+	char(65),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(68),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(83),
+	char(113),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(100),
+	char(100),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(97),
+	char(108),
+	char(65),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(68),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(70),
+	char(97),
+	char(99),
+	char(116),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(83),
+	char(108),
+	char(101),
+	char(101),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(83),
+	char(108),
+	char(101),
+	char(101),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(100),
+	char(100),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(97),
+	char(108),
+	char(68),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(82),
+	char(111),
+	char(119),
+	char(115),
+	char(0),
+	char(110),
+	char(117),
+	char(98),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(114),
+	char(98),
+	char(65),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(114),
+	char(98),
+	char(66),
+	char(0),
+	char(109),
+	char(95),
+	char(111),
+	char(98),
+	char(106),
+	char(101),
+	char(99),
+	char(116),
+	char(84),
+	char(121),
+	char(112),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(117),
+	char(115),
+	char(101),
+	char(114),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(84),
+	char(121),
+	char(112),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(117),
+	char(115),
+	char(101),
+	char(114),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(73),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(101),
+	char(101),
+	char(100),
+	char(115),
+	char(70),
+	char(101),
+	char(101),
+	char(100),
+	char(98),
+	char(97),
+	char(99),
+	char(107),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(112),
+	char(112),
+	char(108),
+	char(105),
+	char(101),
+	char(100),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(100),
+	char(98),
+	char(103),
+	char(68),
+	char(114),
+	char(97),
+	char(119),
+	char(83),
+	char(105),
+	char(122),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(100),
+	char(105),
+	char(115),
+	char(97),
+	char(98),
+	char(108),
+	char(101),
+	char(67),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(115),
+	char(66),
+	char(101),
+	char(116),
+	char(119),
+	char(101),
+	char(101),
+	char(110),
+	char(76),
+	char(105),
+	char(110),
+	char(107),
+	char(101),
+	char(100),
+	char(66),
+	char(111),
+	char(100),
+	char(105),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(111),
+	char(118),
+	char(101),
+	char(114),
+	char(114),
+	char(105),
+	char(100),
+	char(101),
+	char(78),
+	char(117),
+	char(109),
+	char(83),
+	char(111),
+	char(108),
+	char(118),
+	char(101),
+	char(114),
+	char(73),
+	char(116),
+	char(101),
+	char(114),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(114),
+	char(101),
+	char(97),
+	char(107),
+	char(105),
+	char(110),
+	char(103),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(115),
+	char(69),
+	char(110),
+	char(97),
+	char(98),
+	char(108),
+	char(101),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(121),
+	char(112),
+	char(101),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(105),
+	char(118),
+	char(111),
+	char(116),
+	char(73),
+	char(110),
+	char(65),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(105),
+	char(118),
+	char(111),
+	char(116),
+	char(73),
+	char(110),
+	char(66),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(98),
+	char(65),
+	char(70),
+	char(114),
+	char(97),
+	char(109),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(98),
+	char(66),
+	char(70),
+	char(114),
+	char(97),
+	char(109),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(117),
+	char(115),
+	char(101),
+	char(82),
+	char(101),
+	char(102),
+	char(101),
+	char(114),
+	char(101),
+	char(110),
+	char(99),
+	char(101),
+	char(70),
+	char(114),
+	char(97),
+	char(109),
+	char(101),
+	char(65),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(79),
+	char(110),
+	char(108),
+	char(121),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(110),
+	char(97),
+	char(98),
+	char(108),
+	char(101),
+	char(65),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(77),
+	char(111),
+	char(116),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(109),
+	char(111),
+	char(116),
+	char(111),
+	char(114),
+	char(84),
+	char(97),
+	char(114),
+	char(103),
+	char(101),
+	char(116),
+	char(86),
+	char(101),
+	char(108),
+	char(111),
+	char(99),
+	char(105),
+	char(116),
+	char(121),
+	char(0),
+	char(109),
+	char(95),
+	char(109),
+	char(97),
+	char(120),
+	char(77),
+	char(111),
+	char(116),
+	char(111),
+	char(114),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(111),
+	char(119),
+	char(101),
+	char(114),
+	char(76),
+	char(105),
+	char(109),
+	char(105),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(117),
+	char(112),
+	char(112),
+	char(101),
+	char(114),
+	char(76),
+	char(105),
+	char(109),
+	char(105),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(109),
+	char(105),
+	char(116),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(105),
+	char(97),
+	char(115),
+	char(70),
+	char(97),
+	char(99),
+	char(116),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(101),
+	char(108),
+	char(97),
+	char(120),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(70),
+	char(97),
+	char(99),
+	char(116),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(119),
+	char(105),
+	char(110),
+	char(103),
+	char(83),
+	char(112),
+	char(97),
+	char(110),
+	char(49),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(119),
+	char(105),
+	char(110),
+	char(103),
+	char(83),
+	char(112),
+	char(97),
+	char(110),
+	char(50),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(119),
+	char(105),
+	char(115),
+	char(116),
+	char(83),
+	char(112),
+	char(97),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(100),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(85),
+	char(112),
+	char(112),
+	char(101),
+	char(114),
+	char(76),
+	char(105),
+	char(109),
+	char(105),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(76),
+	char(111),
+	char(119),
+	char(101),
+	char(114),
+	char(76),
+	char(105),
+	char(109),
+	char(105),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(85),
+	char(112),
+	char(112),
+	char(101),
+	char(114),
+	char(76),
+	char(105),
+	char(109),
+	char(105),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(76),
+	char(111),
+	char(119),
+	char(101),
+	char(114),
+	char(76),
+	char(105),
+	char(109),
+	char(105),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(117),
+	char(115),
+	char(101),
+	char(76),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(82),
+	char(101),
+	char(102),
+	char(101),
+	char(114),
+	char(101),
+	char(110),
+	char(99),
+	char(101),
+	char(70),
+	char(114),
+	char(97),
+	char(109),
+	char(101),
+	char(65),
+	char(0),
+	char(109),
+	char(95),
+	char(117),
+	char(115),
+	char(101),
+	char(79),
+	char(102),
+	char(102),
+	char(115),
+	char(101),
+	char(116),
+	char(70),
+	char(111),
+	char(114),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(70),
+	char(114),
+	char(97),
+	char(109),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(54),
+	char(100),
+	char(111),
+	char(102),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(112),
+	char(114),
+	char(105),
+	char(110),
+	char(103),
+	char(69),
+	char(110),
+	char(97),
+	char(98),
+	char(108),
+	char(101),
+	char(100),
+	char(91),
+	char(54),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(113),
+	char(117),
+	char(105),
+	char(108),
+	char(105),
+	char(98),
+	char(114),
+	char(105),
+	char(117),
+	char(109),
+	char(80),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(91),
+	char(54),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(112),
+	char(114),
+	char(105),
+	char(110),
+	char(103),
+	char(83),
+	char(116),
+	char(105),
+	char(102),
+	char(102),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(91),
+	char(54),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(112),
+	char(114),
+	char(105),
+	char(110),
+	char(103),
+	char(68),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(91),
+	char(54),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(97),
+	char(117),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(105),
+	char(109),
+	char(101),
+	char(83),
+	char(116),
+	char(101),
+	char(112),
+	char(0),
+	char(109),
+	char(95),
+	char(109),
+	char(97),
+	char(120),
+	char(69),
+	char(114),
+	char(114),
+	char(111),
+	char(114),
+	char(82),
+	char(101),
+	char(100),
+	char(117),
+	char(99),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(114),
+	char(112),
+	char(0),
+	char(109),
+	char(95),
+	char(101),
+	char(114),
+	char(112),
+	char(50),
+	char(0),
+	char(109),
+	char(95),
+	char(103),
+	char(108),
+	char(111),
+	char(98),
+	char(97),
+	char(108),
+	char(67),
+	char(102),
+	char(109),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(112),
+	char(108),
+	char(105),
+	char(116),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(80),
+	char(101),
+	char(110),
+	char(101),
+	char(116),
+	char(114),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(112),
+	char(108),
+	char(105),
+	char(116),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(84),
+	char(117),
+	char(114),
+	char(110),
+	char(69),
+	char(114),
+	char(112),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(83),
+	char(108),
+	char(111),
+	char(112),
+	char(0),
+	char(109),
+	char(95),
+	char(119),
+	char(97),
+	char(114),
+	char(109),
+	char(115),
+	char(116),
+	char(97),
+	char(114),
+	char(116),
+	char(105),
+	char(110),
+	char(103),
+	char(70),
+	char(97),
+	char(99),
+	char(116),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(109),
+	char(97),
+	char(120),
+	char(71),
+	char(121),
+	char(114),
+	char(111),
+	char(115),
+	char(99),
+	char(111),
+	char(112),
+	char(105),
+	char(99),
+	char(70),
+	char(111),
+	char(114),
+	char(99),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(105),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(65),
+	char(120),
+	char(105),
+	char(115),
+	char(82),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(110),
+	char(103),
+	char(70),
+	char(114),
+	char(105),
+	char(99),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(73),
+	char(116),
+	char(101),
+	char(114),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(111),
+	char(108),
+	char(118),
+	char(101),
+	char(114),
+	char(77),
+	char(111),
+	char(100),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(101),
+	char(115),
+	char(116),
+	char(105),
+	char(110),
+	char(103),
+	char(67),
+	char(111),
+	char(110),
+	char(116),
+	char(97),
+	char(99),
+	char(116),
+	char(82),
+	char(101),
+	char(115),
+	char(116),
+	char(105),
+	char(116),
+	char(117),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(84),
+	char(104),
+	char(114),
+	char(101),
+	char(115),
+	char(104),
+	char(111),
+	char(108),
+	char(100),
+	char(0),
+	char(109),
+	char(95),
+	char(109),
+	char(105),
+	char(110),
+	char(105),
+	char(109),
+	char(117),
+	char(109),
+	char(83),
+	char(111),
+	char(108),
+	char(118),
+	char(101),
+	char(114),
+	char(66),
+	char(97),
+	char(116),
+	char(99),
+	char(104),
+	char(83),
+	char(105),
+	char(122),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(112),
+	char(108),
+	char(105),
+	char(116),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(110),
+	char(101),
+	char(97),
+	char(114),
+	char(83),
+	char(116),
+	char(105),
+	char(102),
+	char(102),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(103),
+	char(117),
+	char(108),
+	char(97),
+	char(114),
+	char(83),
+	char(116),
+	char(105),
+	char(102),
+	char(102),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(118),
+	char(111),
+	char(108),
+	char(117),
+	char(109),
+	char(101),
+	char(83),
+	char(116),
+	char(105),
+	char(102),
+	char(102),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(109),
+	char(97),
+	char(116),
+	char(101),
+	char(114),
+	char(105),
+	char(97),
+	char(108),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(111),
+	char(115),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(114),
+	char(101),
+	char(118),
+	char(105),
+	char(111),
+	char(117),
+	char(115),
+	char(80),
+	char(111),
+	char(115),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(118),
+	char(101),
+	char(108),
+	char(111),
+	char(99),
+	char(105),
+	char(116),
+	char(121),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(99),
+	char(99),
+	char(117),
+	char(109),
+	char(117),
+	char(108),
+	char(97),
+	char(116),
+	char(101),
+	char(100),
+	char(70),
+	char(111),
+	char(114),
+	char(99),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(111),
+	char(114),
+	char(109),
+	char(97),
+	char(108),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(114),
+	char(101),
+	char(97),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(116),
+	char(116),
+	char(97),
+	char(99),
+	char(104),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(111),
+	char(100),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(91),
+	char(50),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(101),
+	char(115),
+	char(116),
+	char(76),
+	char(101),
+	char(110),
+	char(103),
+	char(116),
+	char(104),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(98),
+	char(101),
+	char(110),
+	char(100),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(111),
+	char(100),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(91),
+	char(51),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(101),
+	char(115),
+	char(116),
+	char(65),
+	char(114),
+	char(101),
+	char(97),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(48),
+	char(91),
+	char(52),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(111),
+	char(100),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(91),
+	char(52),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(101),
+	char(115),
+	char(116),
+	char(86),
+	char(111),
+	char(108),
+	char(117),
+	char(109),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(49),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(50),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(48),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(111),
+	char(99),
+	char(97),
+	char(108),
+	char(70),
+	char(114),
+	char(97),
+	char(109),
+	char(101),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(114),
+	char(105),
+	char(103),
+	char(105),
+	char(100),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(111),
+	char(100),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(101),
+	char(114),
+	char(111),
+	char(77),
+	char(111),
+	char(100),
+	char(101),
+	char(108),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(97),
+	char(117),
+	char(109),
+	char(103),
+	char(97),
+	char(114),
+	char(116),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(100),
+	char(114),
+	char(97),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(102),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(114),
+	char(101),
+	char(115),
+	char(115),
+	char(117),
+	char(114),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(118),
+	char(111),
+	char(108),
+	char(117),
+	char(109),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(100),
+	char(121),
+	char(110),
+	char(97),
+	char(109),
+	char(105),
+	char(99),
+	char(70),
+	char(114),
+	char(105),
+	char(99),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(111),
+	char(115),
+	char(101),
+	char(77),
+	char(97),
+	char(116),
+	char(99),
+	char(104),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(105),
+	char(103),
+	char(105),
+	char(100),
+	char(67),
+	char(111),
+	char(110),
+	char(116),
+	char(97),
+	char(99),
+	char(116),
+	char(72),
+	char(97),
+	char(114),
+	char(100),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(107),
+	char(105),
+	char(110),
+	char(101),
+	char(116),
+	char(105),
+	char(99),
+	char(67),
+	char(111),
+	char(110),
+	char(116),
+	char(97),
+	char(99),
+	char(116),
+	char(72),
+	char(97),
+	char(114),
+	char(100),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(111),
+	char(102),
+	char(116),
+	char(67),
+	char(111),
+	char(110),
+	char(116),
+	char(97),
+	char(99),
+	char(116),
+	char(72),
+	char(97),
+	char(114),
+	char(100),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(99),
+	char(104),
+	char(111),
+	char(114),
+	char(72),
+	char(97),
+	char(114),
+	char(100),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(111),
+	char(102),
+	char(116),
+	char(82),
+	char(105),
+	char(103),
+	char(105),
+	char(100),
+	char(67),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(72),
+	char(97),
+	char(114),
+	char(100),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(111),
+	char(102),
+	char(116),
+	char(75),
+	char(105),
+	char(110),
+	char(101),
+	char(116),
+	char(105),
+	char(99),
+	char(67),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(72),
+	char(97),
+	char(114),
+	char(100),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(111),
+	char(102),
+	char(116),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(67),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(72),
+	char(97),
+	char(114),
+	char(100),
+	char(110),
+	char(101),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(111),
+	char(102),
+	char(116),
+	char(82),
+	char(105),
+	char(103),
+	char(105),
+	char(100),
+	char(67),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(83),
+	char(112),
+	char(108),
+	char(105),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(111),
+	char(102),
+	char(116),
+	char(75),
+	char(105),
+	char(110),
+	char(101),
+	char(116),
+	char(105),
+	char(99),
+	char(67),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(83),
+	char(112),
+	char(108),
+	char(105),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(111),
+	char(102),
+	char(116),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(67),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(83),
+	char(112),
+	char(108),
+	char(105),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(109),
+	char(97),
+	char(120),
+	char(86),
+	char(111),
+	char(108),
+	char(117),
+	char(109),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(116),
+	char(105),
+	char(109),
+	char(101),
+	char(83),
+	char(99),
+	char(97),
+	char(108),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(118),
+	char(101),
+	char(108),
+	char(111),
+	char(99),
+	char(105),
+	char(116),
+	char(121),
+	char(73),
+	char(116),
+	char(101),
+	char(114),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(112),
+	char(111),
+	char(115),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(73),
+	char(116),
+	char(101),
+	char(114),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(100),
+	char(114),
+	char(105),
+	char(102),
+	char(116),
+	char(73),
+	char(116),
+	char(101),
+	char(114),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(73),
+	char(116),
+	char(101),
+	char(114),
+	char(97),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(111),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(99),
+	char(97),
+	char(108),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(113),
+	char(113),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(109),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(112),
+	char(111),
+	char(115),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(119),
+	char(101),
+	char(105),
+	char(103),
+	char(104),
+	char(116),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(80),
+	char(111),
+	char(115),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(87),
+	char(101),
+	char(105),
+	char(103),
+	char(116),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(118),
+	char(111),
+	char(108),
+	char(117),
+	char(109),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(102),
+	char(114),
+	char(97),
+	char(109),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(102),
+	char(114),
+	char(97),
+	char(109),
+	char(101),
+	char(120),
+	char(102),
+	char(111),
+	char(114),
+	char(109),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(111),
+	char(99),
+	char(105),
+	char(105),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(110),
+	char(118),
+	char(119),
+	char(105),
+	char(0),
+	char(109),
+	char(95),
+	char(118),
+	char(105),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(115),
+	char(91),
+	char(50),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(100),
+	char(105),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(115),
+	char(91),
+	char(50),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(118),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(118),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(102),
+	char(114),
+	char(97),
+	char(109),
+	char(101),
+	char(114),
+	char(101),
+	char(102),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(110),
+	char(111),
+	char(100),
+	char(101),
+	char(73),
+	char(110),
+	char(100),
+	char(105),
+	char(99),
+	char(101),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(109),
+	char(97),
+	char(115),
+	char(115),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(70),
+	char(114),
+	char(97),
+	char(109),
+	char(101),
+	char(82),
+	char(101),
+	char(102),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(77),
+	char(97),
+	char(115),
+	char(115),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(100),
+	char(109),
+	char(97),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(105),
+	char(109),
+	char(97),
+	char(115),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(118),
+	char(105),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(100),
+	char(105),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(100),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(108),
+	char(100),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(97),
+	char(100),
+	char(97),
+	char(109),
+	char(112),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(109),
+	char(97),
+	char(116),
+	char(99),
+	char(104),
+	char(105),
+	char(110),
+	char(103),
+	char(0),
+	char(109),
+	char(95),
+	char(109),
+	char(97),
+	char(120),
+	char(83),
+	char(101),
+	char(108),
+	char(102),
+	char(67),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(101),
+	char(108),
+	char(102),
+	char(67),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(73),
+	char(109),
+	char(112),
+	char(117),
+	char(108),
+	char(115),
+	char(101),
+	char(70),
+	char(97),
+	char(99),
+	char(116),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(110),
+	char(116),
+	char(97),
+	char(105),
+	char(110),
+	char(115),
+	char(65),
+	char(110),
+	char(99),
+	char(104),
+	char(111),
+	char(114),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(100),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(98),
+	char(111),
+	char(100),
+	char(121),
+	char(65),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(98),
+	char(111),
+	char(100),
+	char(121),
+	char(66),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(101),
+	char(102),
+	char(115),
+	char(91),
+	char(50),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(102),
+	char(109),
+	char(0),
+	char(109),
+	char(95),
+	char(115),
+	char(112),
+	char(108),
+	char(105),
+	char(116),
+	char(0),
+	char(109),
+	char(95),
+	char(100),
+	char(101),
+	char(108),
+	char(101),
+	char(116),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(114),
+	char(101),
+	char(108),
+	char(80),
+	char(111),
+	char(115),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(91),
+	char(50),
+	char(93),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(111),
+	char(100),
+	char(121),
+	char(65),
+	char(116),
+	char(121),
+	char(112),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(98),
+	char(111),
+	char(100),
+	char(121),
+	char(66),
+	char(116),
+	char(121),
+	char(112),
+	char(101),
+	char(0),
+	char(109),
+	char(95),
+	char(106),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(84),
+	char(121),
+	char(112),
+	char(101),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(112),
+	char(111),
+	char(115),
+	char(101),
+	char(0),
+	char(42),
+	char(42),
+	char(109),
+	char(95),
+	char(109),
+	char(97),
+	char(116),
+	char(101),
+	char(114),
+	char(105),
+	char(97),
+	char(108),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(110),
+	char(111),
+	char(100),
+	char(101),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(108),
+	char(105),
+	char(110),
+	char(107),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(102),
+	char(97),
+	char(99),
+	char(101),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(116),
+	char(101),
+	char(116),
+	char(114),
+	char(97),
+	char(104),
+	char(101),
+	char(100),
+	char(114),
+	char(97),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(97),
+	char(110),
+	char(99),
+	char(104),
+	char(111),
+	char(114),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(99),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(115),
+	char(0),
+	char(42),
+	char(109),
+	char(95),
+	char(106),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(77),
+	char(97),
+	char(116),
+	char(101),
+	char(114),
+	char(105),
+	char(97),
+	char(108),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(76),
+	char(105),
+	char(110),
+	char(107),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(70),
+	char(97),
+	char(99),
+	char(101),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(84),
+	char(101),
+	char(116),
+	char(114),
+	char(97),
+	char(104),
+	char(101),
+	char(100),
+	char(114),
+	char(97),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(65),
+	char(110),
+	char(99),
+	char(104),
+	char(111),
+	char(114),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(67),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(110),
+	char(117),
+	char(109),
+	char(74),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(115),
+	char(0),
+	char(109),
+	char(95),
+	char(99),
+	char(111),
+	char(110),
+	char(102),
+	char(105),
+	char(103),
+	char(0),
+	char(84),
+	char(89),
+	char(80),
+	char(69),
+	char(76),
+	char(0),
+	char(0),
+	char(0),
+	char(99),
+	char(104),
+	char(97),
+	char(114),
+	char(0),
+	char(117),
+	char(99),
+	char(104),
+	char(97),
+	char(114),
+	char(0),
+	char(115),
+	char(104),
+	char(111),
+	char(114),
+	char(116),
+	char(0),
+	char(117),
+	char(115),
+	char(104),
+	char(111),
+	char(114),
+	char(116),
+	char(0),
+	char(105),
+	char(110),
+	char(116),
+	char(0),
+	char(108),
+	char(111),
+	char(110),
+	char(103),
+	char(0),
+	char(117),
+	char(108),
+	char(111),
+	char(110),
+	char(103),
+	char(0),
+	char(102),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(0),
+	char(100),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(0),
+	char(118),
+	char(111),
+	char(105),
+	char(100),
+	char(0),
+	char(80),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(101),
+	char(114),
+	char(65),
+	char(114),
+	char(114),
+	char(97),
+	char(121),
+	char(0),
+	char(98),
+	char(116),
+	char(80),
+	char(104),
+	char(121),
+	char(115),
+	char(105),
+	char(99),
+	char(115),
+	char(83),
+	char(121),
+	char(115),
+	char(116),
+	char(101),
+	char(109),
+	char(0),
+	char(76),
+	char(105),
+	char(115),
+	char(116),
+	char(66),
+	char(97),
+	char(115),
+	char(101),
+	char(0),
+	char(98),
+	char(116),
+	char(86),
+	char(101),
+	char(99),
+	char(116),
+	char(111),
+	char(114),
+	char(51),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(86),
+	char(101),
+	char(99),
+	char(116),
+	char(111),
+	char(114),
+	char(51),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(77),
+	char(97),
+	char(116),
+	char(114),
+	char(105),
+	char(120),
+	char(51),
+	char(120),
+	char(51),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(77),
+	char(97),
+	char(116),
+	char(114),
+	char(105),
+	char(120),
+	char(51),
+	char(120),
+	char(51),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(84),
+	char(114),
+	char(97),
+	char(110),
+	char(115),
+	char(102),
+	char(111),
+	char(114),
+	char(109),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(84),
+	char(114),
+	char(97),
+	char(110),
+	char(115),
+	char(102),
+	char(111),
+	char(114),
+	char(109),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(66),
+	char(118),
+	char(104),
+	char(83),
+	char(117),
+	char(98),
+	char(116),
+	char(114),
+	char(101),
+	char(101),
+	char(73),
+	char(110),
+	char(102),
+	char(111),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(79),
+	char(112),
+	char(116),
+	char(105),
+	char(109),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(66),
+	char(118),
+	char(104),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(79),
+	char(112),
+	char(116),
+	char(105),
+	char(109),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(66),
+	char(118),
+	char(104),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(81),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(66),
+	char(118),
+	char(104),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(81),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(66),
+	char(118),
+	char(104),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(81),
+	char(117),
+	char(97),
+	char(110),
+	char(116),
+	char(105),
+	char(122),
+	char(101),
+	char(100),
+	char(66),
+	char(118),
+	char(104),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(83),
+	char(116),
+	char(97),
+	char(116),
+	char(105),
+	char(99),
+	char(80),
+	char(108),
+	char(97),
+	char(110),
+	char(101),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(110),
+	char(118),
+	char(101),
+	char(120),
+	char(73),
+	char(110),
+	char(116),
+	char(101),
+	char(114),
+	char(110),
+	char(97),
+	char(108),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(80),
+	char(111),
+	char(115),
+	char(105),
+	char(116),
+	char(105),
+	char(111),
+	char(110),
+	char(65),
+	char(110),
+	char(100),
+	char(82),
+	char(97),
+	char(100),
+	char(105),
+	char(117),
+	char(115),
+	char(0),
+	char(98),
+	char(116),
+	char(77),
+	char(117),
+	char(108),
+	char(116),
+	char(105),
+	char(83),
+	char(112),
+	char(104),
+	char(101),
+	char(114),
+	char(101),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(73),
+	char(110),
+	char(116),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(83),
+	char(104),
+	char(111),
+	char(114),
+	char(116),
+	char(73),
+	char(110),
+	char(116),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(83),
+	char(104),
+	char(111),
+	char(114),
+	char(116),
+	char(73),
+	char(110),
+	char(116),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(84),
+	char(114),
+	char(105),
+	char(112),
+	char(108),
+	char(101),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(104),
+	char(97),
+	char(114),
+	char(73),
+	char(110),
+	char(100),
+	char(101),
+	char(120),
+	char(84),
+	char(114),
+	char(105),
+	char(112),
+	char(108),
+	char(101),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(77),
+	char(101),
+	char(115),
+	char(104),
+	char(80),
+	char(97),
+	char(114),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(83),
+	char(116),
+	char(114),
+	char(105),
+	char(100),
+	char(105),
+	char(110),
+	char(103),
+	char(77),
+	char(101),
+	char(115),
+	char(104),
+	char(73),
+	char(110),
+	char(116),
+	char(101),
+	char(114),
+	char(102),
+	char(97),
+	char(99),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(84),
+	char(114),
+	char(105),
+	char(97),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(77),
+	char(101),
+	char(115),
+	char(104),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(84),
+	char(114),
+	char(105),
+	char(97),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(73),
+	char(110),
+	char(102),
+	char(111),
+	char(77),
+	char(97),
+	char(112),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(83),
+	char(99),
+	char(97),
+	char(108),
+	char(101),
+	char(100),
+	char(84),
+	char(114),
+	char(105),
+	char(97),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(77),
+	char(101),
+	char(115),
+	char(104),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(109),
+	char(112),
+	char(111),
+	char(117),
+	char(110),
+	char(100),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(67),
+	char(104),
+	char(105),
+	char(108),
+	char(100),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(109),
+	char(112),
+	char(111),
+	char(117),
+	char(110),
+	char(100),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(121),
+	char(108),
+	char(105),
+	char(110),
+	char(100),
+	char(101),
+	char(114),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(97),
+	char(112),
+	char(115),
+	char(117),
+	char(108),
+	char(101),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(84),
+	char(114),
+	char(105),
+	char(97),
+	char(110),
+	char(103),
+	char(108),
+	char(101),
+	char(73),
+	char(110),
+	char(102),
+	char(111),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(71),
+	char(73),
+	char(109),
+	char(112),
+	char(97),
+	char(99),
+	char(116),
+	char(77),
+	char(101),
+	char(115),
+	char(104),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(110),
+	char(118),
+	char(101),
+	char(120),
+	char(72),
+	char(117),
+	char(108),
+	char(108),
+	char(83),
+	char(104),
+	char(97),
+	char(112),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(79),
+	char(98),
+	char(106),
+	char(101),
+	char(99),
+	char(116),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(108),
+	char(108),
+	char(105),
+	char(115),
+	char(105),
+	char(111),
+	char(110),
+	char(79),
+	char(98),
+	char(106),
+	char(101),
+	char(99),
+	char(116),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(68),
+	char(121),
+	char(110),
+	char(97),
+	char(109),
+	char(105),
+	char(99),
+	char(115),
+	char(87),
+	char(111),
+	char(114),
+	char(108),
+	char(100),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(110),
+	char(116),
+	char(97),
+	char(99),
+	char(116),
+	char(83),
+	char(111),
+	char(108),
+	char(118),
+	char(101),
+	char(114),
+	char(73),
+	char(110),
+	char(102),
+	char(111),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(68),
+	char(121),
+	char(110),
+	char(97),
+	char(109),
+	char(105),
+	char(99),
+	char(115),
+	char(87),
+	char(111),
+	char(114),
+	char(108),
+	char(100),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(110),
+	char(116),
+	char(97),
+	char(99),
+	char(116),
+	char(83),
+	char(111),
+	char(108),
+	char(118),
+	char(101),
+	char(114),
+	char(73),
+	char(110),
+	char(102),
+	char(111),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(82),
+	char(105),
+	char(103),
+	char(105),
+	char(100),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(82),
+	char(105),
+	char(103),
+	char(105),
+	char(100),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(73),
+	char(110),
+	char(102),
+	char(111),
+	char(49),
+	char(0),
+	char(98),
+	char(116),
+	char(84),
+	char(121),
+	char(112),
+	char(101),
+	char(100),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(82),
+	char(105),
+	char(103),
+	char(105),
+	char(100),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(80),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(50),
+	char(80),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(80),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(50),
+	char(80),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(72),
+	char(105),
+	char(110),
+	char(103),
+	char(101),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(68),
+	char(111),
+	char(117),
+	char(98),
+	char(108),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(72),
+	char(105),
+	char(110),
+	char(103),
+	char(101),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(67),
+	char(111),
+	char(110),
+	char(101),
+	char(84),
+	char(119),
+	char(105),
+	char(115),
+	char(116),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(71),
+	char(101),
+	char(110),
+	char(101),
+	char(114),
+	char(105),
+	char(99),
+	char(54),
+	char(68),
+	char(111),
+	char(102),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(71),
+	char(101),
+	char(110),
+	char(101),
+	char(114),
+	char(105),
+	char(99),
+	char(54),
+	char(68),
+	char(111),
+	char(102),
+	char(83),
+	char(112),
+	char(114),
+	char(105),
+	char(110),
+	char(103),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(83),
+	char(108),
+	char(105),
+	char(100),
+	char(101),
+	char(114),
+	char(67),
+	char(111),
+	char(110),
+	char(115),
+	char(116),
+	char(114),
+	char(97),
+	char(105),
+	char(110),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(77),
+	char(97),
+	char(116),
+	char(101),
+	char(114),
+	char(105),
+	char(97),
+	char(108),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(78),
+	char(111),
+	char(100),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(76),
+	char(105),
+	char(110),
+	char(107),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(70),
+	char(97),
+	char(99),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(84),
+	char(101),
+	char(116),
+	char(114),
+	char(97),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(82),
+	char(105),
+	char(103),
+	char(105),
+	char(100),
+	char(65),
+	char(110),
+	char(99),
+	char(104),
+	char(111),
+	char(114),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(67),
+	char(111),
+	char(110),
+	char(102),
+	char(105),
+	char(103),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(80),
+	char(111),
+	char(115),
+	char(101),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(67),
+	char(108),
+	char(117),
+	char(115),
+	char(116),
+	char(101),
+	char(114),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(74),
+	char(111),
+	char(105),
+	char(110),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(98),
+	char(116),
+	char(83),
+	char(111),
+	char(102),
+	char(116),
+	char(66),
+	char(111),
+	char(100),
+	char(121),
+	char(70),
+	char(108),
+	char(111),
+	char(97),
+	char(116),
+	char(68),
+	char(97),
+	char(116),
+	char(97),
+	char(0),
+	char(0),
+	char(0),
+	char(84),
+	char(76),
+	char(69),
+	char(78),
+	char(1),
+	char(0),
+	char(1),
+	char(0),
+	char(2),
+	char(0),
+	char(2),
+	char(0),
+	char(4),
+	char(0),
+	char(4),
+	char(0),
+	char(4),
+	char(0),
+	char(4),
+	char(0),
+	char(8),
+	char(0),
+	char(0),
+	char(0),
+	char(16),
+	char(0),
+	char(48),
+	char(0),
+	char(16),
+	char(0),
+	char(16),
+	char(0),
+	char(32),
+	char(0),
+	char(48),
+	char(0),
+	char(96),
+	char(0),
+	char(64),
+	char(0),
+	char(-128),
+	char(0),
+	char(20),
+	char(0),
+	char(48),
+	char(0),
+	char(80),
+	char(0),
+	char(16),
+	char(0),
+	char(96),
+	char(0),
+	char(-112),
+	char(0),
+	char(16),
+	char(0),
+	char(56),
+	char(0),
+	char(56),
+	char(0),
+	char(20),
+	char(0),
+	char(72),
+	char(0),
+	char(4),
+	char(0),
+	char(4),
+	char(0),
+	char(8),
+	char(0),
+	char(4),
+	char(0),
+	char(56),
+	char(0),
+	char(32),
+	char(0),
+	char(80),
+	char(0),
+	char(72),
+	char(0),
+	char(96),
+	char(0),
+	char(80),
+	char(0),
+	char(32),
+	char(0),
+	char(64),
+	char(0),
+	char(64),
+	char(0),
+	char(16),
+	char(0),
+	char(72),
+	char(0),
+	char(80),
+	char(0),
+	char(-32),
+	char(1),
+	char(16),
+	char(1),
+	char(-72),
+	char(0),
+	char(-104),
+	char(0),
+	char(104),
+	char(0),
+	char(88),
+	char(0),
+	char(-8),
+	char(1),
+	char(-80),
+	char(3),
+	char(8),
+	char(0),
+	char(64),
+	char(0),
+	char(0),
+	char(0),
+	char(96),
+	char(0),
+	char(-128),
+	char(0),
+	char(104),
+	char(1),
+	char(-24),
+	char(0),
+	char(-32),
+	char(0),
+	char(8),
+	char(1),
+	char(104),
+	char(1),
+	char(-40),
+	char(0),
+	char(16),
+	char(0),
+	char(104),
+	char(0),
+	char(24),
+	char(0),
+	char(40),
+	char(0),
+	char(104),
+	char(0),
+	char(96),
+	char(0),
+	char(104),
+	char(0),
+	char(-56),
+	char(0),
+	char(104),
+	char(1),
+	char(112),
+	char(0),
+	char(-32),
+	char(1),
+	char(83),
+	char(84),
+	char(82),
+	char(67),
+	char(65),
+	char(0),
+	char(0),
+	char(0),
+	char(10),
+	char(0),
+	char(3),
+	char(0),
+	char(4),
+	char(0),
+	char(0),
+	char(0),
+	char(4),
+	char(0),
+	char(1),
+	char(0),
+	char(9),
+	char(0),
+	char(2),
+	char(0),
+	char(11),
+	char(0),
+	char(3),
+	char(0),
+	char(10),
+	char(0),
+	char(3),
+	char(0),
+	char(10),
+	char(0),
+	char(4),
+	char(0),
+	char(10),
+	char(0),
+	char(5),
+	char(0),
+	char(12),
+	char(0),
+	char(2),
+	char(0),
+	char(9),
+	char(0),
+	char(6),
+	char(0),
+	char(9),
+	char(0),
+	char(7),
+	char(0),
+	char(13),
+	char(0),
+	char(1),
+	char(0),
+	char(7),
+	char(0),
+	char(8),
+	char(0),
+	char(14),
+	char(0),
+	char(1),
+	char(0),
+	char(8),
+	char(0),
+	char(8),
+	char(0),
+	char(15),
+	char(0),
+	char(1),
+	char(0),
+	char(13),
+	char(0),
+	char(9),
+	char(0),
+	char(16),
+	char(0),
+	char(1),
+	char(0),
+	char(14),
+	char(0),
+	char(9),
+	char(0),
+	char(17),
+	char(0),
+	char(2),
+	char(0),
+	char(15),
+	char(0),
+	char(10),
+	char(0),
+	char(13),
+	char(0),
+	char(11),
+	char(0),
+	char(18),
+	char(0),
+	char(2),
+	char(0),
+	char(16),
+	char(0),
+	char(10),
+	char(0),
+	char(14),
+	char(0),
+	char(11),
+	char(0),
+	char(19),
+	char(0),
+	char(4),
+	char(0),
+	char(4),
+	char(0),
+	char(12),
+	char(0),
+	char(4),
+	char(0),
+	char(13),
+	char(0),
+	char(2),
+	char(0),
+	char(14),
+	char(0),
+	char(2),
+	char(0),
+	char(15),
+	char(0),
+	char(20),
+	char(0),
+	char(6),
+	char(0),
+	char(13),
+	char(0),
+	char(16),
+	char(0),
+	char(13),
+	char(0),
+	char(17),
+	char(0),
+	char(4),
+	char(0),
+	char(18),
+	char(0),
+	char(4),
+	char(0),
+	char(19),
+	char(0),
+	char(4),
+	char(0),
+	char(20),
+	char(0),
+	char(0),
+	char(0),
+	char(21),
+	char(0),
+	char(21),
+	char(0),
+	char(6),
+	char(0),
+	char(14),
+	char(0),
+	char(16),
+	char(0),
+	char(14),
+	char(0),
+	char(17),
+	char(0),
+	char(4),
+	char(0),
+	char(18),
+	char(0),
+	char(4),
+	char(0),
+	char(19),
+	char(0),
+	char(4),
+	char(0),
+	char(20),
+	char(0),
+	char(0),
+	char(0),
+	char(21),
+	char(0),
+	char(22),
+	char(0),
+	char(3),
+	char(0),
+	char(2),
+	char(0),
+	char(14),
+	char(0),
+	char(2),
+	char(0),
+	char(15),
+	char(0),
+	char(4),
+	char(0),
+	char(22),
+	char(0),
+	char(23),
+	char(0),
+	char(12),
+	char(0),
+	char(13),
+	char(0),
+	char(23),
+	char(0),
+	char(13),
+	char(0),
+	char(24),
+	char(0),
+	char(13),
+	char(0),
+	char(25),
+	char(0),
+	char(4),
+	char(0),
+	char(26),
+	char(0),
+	char(4),
+	char(0),
+	char(27),
+	char(0),
+	char(4),
+	char(0),
+	char(28),
+	char(0),
+	char(4),
+	char(0),
+	char(29),
+	char(0),
+	char(20),
+	char(0),
+	char(30),
+	char(0),
+	char(22),
+	char(0),
+	char(31),
+	char(0),
+	char(19),
+	char(0),
+	char(32),
+	char(0),
+	char(4),
+	char(0),
+	char(33),
+	char(0),
+	char(4),
+	char(0),
+	char(34),
+	char(0),
+	char(24),
+	char(0),
+	char(12),
+	char(0),
+	char(14),
+	char(0),
+	char(23),
+	char(0),
+	char(14),
+	char(0),
+	char(24),
+	char(0),
+	char(14),
+	char(0),
+	char(25),
+	char(0),
+	char(4),
+	char(0),
+	char(26),
+	char(0),
+	char(4),
+	char(0),
+	char(27),
+	char(0),
+	char(4),
+	char(0),
+	char(28),
+	char(0),
+	char(4),
+	char(0),
+	char(29),
+	char(0),
+	char(21),
+	char(0),
+	char(30),
+	char(0),
+	char(22),
+	char(0),
+	char(31),
+	char(0),
+	char(4),
+	char(0),
+	char(33),
+	char(0),
+	char(4),
+	char(0),
+	char(34),
+	char(0),
+	char(19),
+	char(0),
+	char(32),
+	char(0),
+	char(25),
+	char(0),
+	char(3),
+	char(0),
+	char(0),
+	char(0),
+	char(35),
+	char(0),
+	char(4),
+	char(0),
+	char(36),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(26),
+	char(0),
+	char(5),
+	char(0),
+	char(25),
+	char(0),
+	char(38),
+	char(0),
+	char(13),
+	char(0),
+	char(39),
+	char(0),
+	char(13),
+	char(0),
+	char(40),
+	char(0),
+	char(7),
+	char(0),
+	char(41),
+	char(0),
+	char(0),
+	char(0),
+	char(21),
+	char(0),
+	char(27),
+	char(0),
+	char(5),
+	char(0),
+	char(25),
+	char(0),
+	char(38),
+	char(0),
+	char(13),
+	char(0),
+	char(39),
+	char(0),
+	char(13),
+	char(0),
+	char(42),
+	char(0),
+	char(7),
+	char(0),
+	char(43),
+	char(0),
+	char(4),
+	char(0),
+	char(44),
+	char(0),
+	char(28),
+	char(0),
+	char(2),
+	char(0),
+	char(13),
+	char(0),
+	char(45),
+	char(0),
+	char(7),
+	char(0),
+	char(46),
+	char(0),
+	char(29),
+	char(0),
+	char(4),
+	char(0),
+	char(27),
+	char(0),
+	char(47),
+	char(0),
+	char(28),
+	char(0),
+	char(48),
+	char(0),
+	char(4),
+	char(0),
+	char(49),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(30),
+	char(0),
+	char(1),
+	char(0),
+	char(4),
+	char(0),
+	char(50),
+	char(0),
+	char(31),
+	char(0),
+	char(2),
+	char(0),
+	char(2),
+	char(0),
+	char(50),
+	char(0),
+	char(0),
+	char(0),
+	char(51),
+	char(0),
+	char(32),
+	char(0),
+	char(2),
+	char(0),
+	char(2),
+	char(0),
+	char(52),
+	char(0),
+	char(0),
+	char(0),
+	char(51),
+	char(0),
+	char(33),
+	char(0),
+	char(2),
+	char(0),
+	char(0),
+	char(0),
+	char(52),
+	char(0),
+	char(0),
+	char(0),
+	char(53),
+	char(0),
+	char(34),
+	char(0),
+	char(8),
+	char(0),
+	char(13),
+	char(0),
+	char(54),
+	char(0),
+	char(14),
+	char(0),
+	char(55),
+	char(0),
+	char(30),
+	char(0),
+	char(56),
+	char(0),
+	char(32),
+	char(0),
+	char(57),
+	char(0),
+	char(33),
+	char(0),
+	char(58),
+	char(0),
+	char(31),
+	char(0),
+	char(59),
+	char(0),
+	char(4),
+	char(0),
+	char(60),
+	char(0),
+	char(4),
+	char(0),
+	char(61),
+	char(0),
+	char(35),
+	char(0),
+	char(4),
+	char(0),
+	char(34),
+	char(0),
+	char(62),
+	char(0),
+	char(13),
+	char(0),
+	char(63),
+	char(0),
+	char(4),
+	char(0),
+	char(64),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(36),
+	char(0),
+	char(7),
+	char(0),
+	char(25),
+	char(0),
+	char(38),
+	char(0),
+	char(35),
+	char(0),
+	char(65),
+	char(0),
+	char(23),
+	char(0),
+	char(66),
+	char(0),
+	char(24),
+	char(0),
+	char(67),
+	char(0),
+	char(37),
+	char(0),
+	char(68),
+	char(0),
+	char(7),
+	char(0),
+	char(43),
+	char(0),
+	char(0),
+	char(0),
+	char(69),
+	char(0),
+	char(38),
+	char(0),
+	char(2),
+	char(0),
+	char(36),
+	char(0),
+	char(70),
+	char(0),
+	char(13),
+	char(0),
+	char(39),
+	char(0),
+	char(39),
+	char(0),
+	char(4),
+	char(0),
+	char(17),
+	char(0),
+	char(71),
+	char(0),
+	char(25),
+	char(0),
+	char(72),
+	char(0),
+	char(4),
+	char(0),
+	char(73),
+	char(0),
+	char(7),
+	char(0),
+	char(74),
+	char(0),
+	char(40),
+	char(0),
+	char(4),
+	char(0),
+	char(25),
+	char(0),
+	char(38),
+	char(0),
+	char(39),
+	char(0),
+	char(75),
+	char(0),
+	char(4),
+	char(0),
+	char(76),
+	char(0),
+	char(7),
+	char(0),
+	char(43),
+	char(0),
+	char(41),
+	char(0),
+	char(3),
+	char(0),
+	char(27),
+	char(0),
+	char(47),
+	char(0),
+	char(4),
+	char(0),
+	char(77),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(42),
+	char(0),
+	char(3),
+	char(0),
+	char(27),
+	char(0),
+	char(47),
+	char(0),
+	char(4),
+	char(0),
+	char(77),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(43),
+	char(0),
+	char(4),
+	char(0),
+	char(4),
+	char(0),
+	char(78),
+	char(0),
+	char(7),
+	char(0),
+	char(79),
+	char(0),
+	char(7),
+	char(0),
+	char(80),
+	char(0),
+	char(7),
+	char(0),
+	char(81),
+	char(0),
+	char(37),
+	char(0),
+	char(14),
+	char(0),
+	char(4),
+	char(0),
+	char(82),
+	char(0),
+	char(4),
+	char(0),
+	char(83),
+	char(0),
+	char(43),
+	char(0),
+	char(84),
+	char(0),
+	char(4),
+	char(0),
+	char(85),
+	char(0),
+	char(7),
+	char(0),
+	char(86),
+	char(0),
+	char(7),
+	char(0),
+	char(87),
+	char(0),
+	char(7),
+	char(0),
+	char(88),
+	char(0),
+	char(7),
+	char(0),
+	char(89),
+	char(0),
+	char(7),
+	char(0),
+	char(90),
+	char(0),
+	char(4),
+	char(0),
+	char(91),
+	char(0),
+	char(4),
+	char(0),
+	char(92),
+	char(0),
+	char(4),
+	char(0),
+	char(93),
+	char(0),
+	char(4),
+	char(0),
+	char(94),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(44),
+	char(0),
+	char(5),
+	char(0),
+	char(25),
+	char(0),
+	char(38),
+	char(0),
+	char(35),
+	char(0),
+	char(65),
+	char(0),
+	char(13),
+	char(0),
+	char(39),
+	char(0),
+	char(7),
+	char(0),
+	char(43),
+	char(0),
+	char(4),
+	char(0),
+	char(95),
+	char(0),
+	char(45),
+	char(0),
+	char(5),
+	char(0),
+	char(27),
+	char(0),
+	char(47),
+	char(0),
+	char(13),
+	char(0),
+	char(96),
+	char(0),
+	char(14),
+	char(0),
+	char(97),
+	char(0),
+	char(4),
+	char(0),
+	char(98),
+	char(0),
+	char(0),
+	char(0),
+	char(99),
+	char(0),
+	char(46),
+	char(0),
+	char(25),
+	char(0),
+	char(9),
+	char(0),
+	char(100),
+	char(0),
+	char(9),
+	char(0),
+	char(101),
+	char(0),
+	char(25),
+	char(0),
+	char(102),
+	char(0),
+	char(0),
+	char(0),
+	char(35),
+	char(0),
+	char(18),
+	char(0),
+	char(103),
+	char(0),
+	char(18),
+	char(0),
+	char(104),
+	char(0),
+	char(14),
+	char(0),
+	char(105),
+	char(0),
+	char(14),
+	char(0),
+	char(106),
+	char(0),
+	char(14),
+	char(0),
+	char(107),
+	char(0),
+	char(8),
+	char(0),
+	char(108),
+	char(0),
+	char(8),
+	char(0),
+	char(109),
+	char(0),
+	char(8),
+	char(0),
+	char(110),
+	char(0),
+	char(8),
+	char(0),
+	char(111),
+	char(0),
+	char(8),
+	char(0),
+	char(112),
+	char(0),
+	char(8),
+	char(0),
+	char(113),
+	char(0),
+	char(8),
+	char(0),
+	char(114),
+	char(0),
+	char(8),
+	char(0),
+	char(115),
+	char(0),
+	char(4),
+	char(0),
+	char(116),
+	char(0),
+	char(4),
+	char(0),
+	char(117),
+	char(0),
+	char(4),
+	char(0),
+	char(118),
+	char(0),
+	char(4),
+	char(0),
+	char(119),
+	char(0),
+	char(4),
+	char(0),
+	char(120),
+	char(0),
+	char(4),
+	char(0),
+	char(121),
+	char(0),
+	char(4),
+	char(0),
+	char(122),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(47),
+	char(0),
+	char(25),
+	char(0),
+	char(9),
+	char(0),
+	char(100),
+	char(0),
+	char(9),
+	char(0),
+	char(101),
+	char(0),
+	char(25),
+	char(0),
+	char(102),
+	char(0),
+	char(0),
+	char(0),
+	char(35),
+	char(0),
+	char(17),
+	char(0),
+	char(103),
+	char(0),
+	char(17),
+	char(0),
+	char(104),
+	char(0),
+	char(13),
+	char(0),
+	char(105),
+	char(0),
+	char(13),
+	char(0),
+	char(106),
+	char(0),
+	char(13),
+	char(0),
+	char(107),
+	char(0),
+	char(7),
+	char(0),
+	char(108),
+	char(0),
+	char(7),
+	char(0),
+	char(109),
+	char(0),
+	char(7),
+	char(0),
+	char(110),
+	char(0),
+	char(7),
+	char(0),
+	char(111),
+	char(0),
+	char(7),
+	char(0),
+	char(112),
+	char(0),
+	char(7),
+	char(0),
+	char(113),
+	char(0),
+	char(7),
+	char(0),
+	char(114),
+	char(0),
+	char(7),
+	char(0),
+	char(115),
+	char(0),
+	char(4),
+	char(0),
+	char(116),
+	char(0),
+	char(4),
+	char(0),
+	char(117),
+	char(0),
+	char(4),
+	char(0),
+	char(118),
+	char(0),
+	char(4),
+	char(0),
+	char(119),
+	char(0),
+	char(4),
+	char(0),
+	char(120),
+	char(0),
+	char(4),
+	char(0),
+	char(121),
+	char(0),
+	char(4),
+	char(0),
+	char(122),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(48),
+	char(0),
+	char(2),
+	char(0),
+	char(49),
+	char(0),
+	char(123),
+	char(0),
+	char(14),
+	char(0),
+	char(124),
+	char(0),
+	char(50),
+	char(0),
+	char(2),
+	char(0),
+	char(51),
+	char(0),
+	char(123),
+	char(0),
+	char(13),
+	char(0),
+	char(124),
+	char(0),
+	char(52),
+	char(0),
+	char(21),
+	char(0),
+	char(47),
+	char(0),
+	char(125),
+	char(0),
+	char(15),
+	char(0),
+	char(126),
+	char(0),
+	char(13),
+	char(0),
+	char(127),
+	char(0),
+	char(13),
+	char(0),
+	char(-128),
+	char(0),
+	char(13),
+	char(0),
+	char(-127),
+	char(0),
+	char(13),
+	char(0),
+	char(-126),
+	char(0),
+	char(13),
+	char(0),
+	char(124),
+	char(0),
+	char(13),
+	char(0),
+	char(-125),
+	char(0),
+	char(13),
+	char(0),
+	char(-124),
+	char(0),
+	char(13),
+	char(0),
+	char(-123),
+	char(0),
+	char(13),
+	char(0),
+	char(-122),
+	char(0),
+	char(7),
+	char(0),
+	char(-121),
+	char(0),
+	char(7),
+	char(0),
+	char(-120),
+	char(0),
+	char(7),
+	char(0),
+	char(-119),
+	char(0),
+	char(7),
+	char(0),
+	char(-118),
+	char(0),
+	char(7),
+	char(0),
+	char(-117),
+	char(0),
+	char(7),
+	char(0),
+	char(-116),
+	char(0),
+	char(7),
+	char(0),
+	char(-115),
+	char(0),
+	char(7),
+	char(0),
+	char(-114),
+	char(0),
+	char(7),
+	char(0),
+	char(-113),
+	char(0),
+	char(4),
+	char(0),
+	char(-112),
+	char(0),
+	char(53),
+	char(0),
+	char(22),
+	char(0),
+	char(46),
+	char(0),
+	char(125),
+	char(0),
+	char(16),
+	char(0),
+	char(126),
+	char(0),
+	char(14),
+	char(0),
+	char(127),
+	char(0),
+	char(14),
+	char(0),
+	char(-128),
+	char(0),
+	char(14),
+	char(0),
+	char(-127),
+	char(0),
+	char(14),
+	char(0),
+	char(-126),
+	char(0),
+	char(14),
+	char(0),
+	char(124),
+	char(0),
+	char(14),
+	char(0),
+	char(-125),
+	char(0),
+	char(14),
+	char(0),
+	char(-124),
+	char(0),
+	char(14),
+	char(0),
+	char(-123),
+	char(0),
+	char(14),
+	char(0),
+	char(-122),
+	char(0),
+	char(8),
+	char(0),
+	char(-121),
+	char(0),
+	char(8),
+	char(0),
+	char(-120),
+	char(0),
+	char(8),
+	char(0),
+	char(-119),
+	char(0),
+	char(8),
+	char(0),
+	char(-118),
+	char(0),
+	char(8),
+	char(0),
+	char(-117),
+	char(0),
+	char(8),
+	char(0),
+	char(-116),
+	char(0),
+	char(8),
+	char(0),
+	char(-115),
+	char(0),
+	char(8),
+	char(0),
+	char(-114),
+	char(0),
+	char(8),
+	char(0),
+	char(-113),
+	char(0),
+	char(4),
+	char(0),
+	char(-112),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(54),
+	char(0),
+	char(2),
+	char(0),
+	char(4),
+	char(0),
+	char(-111),
+	char(0),
+	char(4),
+	char(0),
+	char(-110),
+	char(0),
+	char(55),
+	char(0),
+	char(13),
+	char(0),
+	char(56),
+	char(0),
+	char(-109),
+	char(0),
+	char(56),
+	char(0),
+	char(-108),
+	char(0),
+	char(0),
+	char(0),
+	char(35),
+	char(0),
+	char(4),
+	char(0),
+	char(-107),
+	char(0),
+	char(4),
+	char(0),
+	char(-106),
+	char(0),
+	char(4),
+	char(0),
+	char(-105),
+	char(0),
+	char(4),
+	char(0),
+	char(-104),
+	char(0),
+	char(7),
+	char(0),
+	char(-103),
+	char(0),
+	char(7),
+	char(0),
+	char(-102),
+	char(0),
+	char(4),
+	char(0),
+	char(-101),
+	char(0),
+	char(4),
+	char(0),
+	char(-100),
+	char(0),
+	char(7),
+	char(0),
+	char(-99),
+	char(0),
+	char(4),
+	char(0),
+	char(-98),
+	char(0),
+	char(57),
+	char(0),
+	char(3),
+	char(0),
+	char(55),
+	char(0),
+	char(-97),
+	char(0),
+	char(13),
+	char(0),
+	char(-96),
+	char(0),
+	char(13),
+	char(0),
+	char(-95),
+	char(0),
+	char(58),
+	char(0),
+	char(3),
+	char(0),
+	char(55),
+	char(0),
+	char(-97),
+	char(0),
+	char(14),
+	char(0),
+	char(-96),
+	char(0),
+	char(14),
+	char(0),
+	char(-95),
+	char(0),
+	char(59),
+	char(0),
+	char(13),
+	char(0),
+	char(55),
+	char(0),
+	char(-97),
+	char(0),
+	char(18),
+	char(0),
+	char(-94),
+	char(0),
+	char(18),
+	char(0),
+	char(-93),
+	char(0),
+	char(4),
+	char(0),
+	char(-92),
+	char(0),
+	char(4),
+	char(0),
+	char(-91),
+	char(0),
+	char(4),
+	char(0),
+	char(-90),
+	char(0),
+	char(7),
+	char(0),
+	char(-89),
+	char(0),
+	char(7),
+	char(0),
+	char(-88),
+	char(0),
+	char(7),
+	char(0),
+	char(-87),
+	char(0),
+	char(7),
+	char(0),
+	char(-86),
+	char(0),
+	char(7),
+	char(0),
+	char(-85),
+	char(0),
+	char(7),
+	char(0),
+	char(-84),
+	char(0),
+	char(7),
+	char(0),
+	char(-83),
+	char(0),
+	char(60),
+	char(0),
+	char(13),
+	char(0),
+	char(55),
+	char(0),
+	char(-97),
+	char(0),
+	char(17),
+	char(0),
+	char(-94),
+	char(0),
+	char(17),
+	char(0),
+	char(-93),
+	char(0),
+	char(4),
+	char(0),
+	char(-92),
+	char(0),
+	char(4),
+	char(0),
+	char(-91),
+	char(0),
+	char(4),
+	char(0),
+	char(-90),
+	char(0),
+	char(7),
+	char(0),
+	char(-89),
+	char(0),
+	char(7),
+	char(0),
+	char(-88),
+	char(0),
+	char(7),
+	char(0),
+	char(-87),
+	char(0),
+	char(7),
+	char(0),
+	char(-86),
+	char(0),
+	char(7),
+	char(0),
+	char(-85),
+	char(0),
+	char(7),
+	char(0),
+	char(-84),
+	char(0),
+	char(7),
+	char(0),
+	char(-83),
+	char(0),
+	char(61),
+	char(0),
+	char(11),
+	char(0),
+	char(55),
+	char(0),
+	char(-97),
+	char(0),
+	char(17),
+	char(0),
+	char(-94),
+	char(0),
+	char(17),
+	char(0),
+	char(-93),
+	char(0),
+	char(7),
+	char(0),
+	char(-82),
+	char(0),
+	char(7),
+	char(0),
+	char(-81),
+	char(0),
+	char(7),
+	char(0),
+	char(-80),
+	char(0),
+	char(7),
+	char(0),
+	char(-85),
+	char(0),
+	char(7),
+	char(0),
+	char(-84),
+	char(0),
+	char(7),
+	char(0),
+	char(-83),
+	char(0),
+	char(7),
+	char(0),
+	char(-79),
+	char(0),
+	char(0),
+	char(0),
+	char(21),
+	char(0),
+	char(62),
+	char(0),
+	char(9),
+	char(0),
+	char(55),
+	char(0),
+	char(-97),
+	char(0),
+	char(17),
+	char(0),
+	char(-94),
+	char(0),
+	char(17),
+	char(0),
+	char(-93),
+	char(0),
+	char(13),
+	char(0),
+	char(-78),
+	char(0),
+	char(13),
+	char(0),
+	char(-77),
+	char(0),
+	char(13),
+	char(0),
+	char(-76),
+	char(0),
+	char(13),
+	char(0),
+	char(-75),
+	char(0),
+	char(4),
+	char(0),
+	char(-74),
+	char(0),
+	char(4),
+	char(0),
+	char(-73),
+	char(0),
+	char(63),
+	char(0),
+	char(5),
+	char(0),
+	char(62),
+	char(0),
+	char(-72),
+	char(0),
+	char(4),
+	char(0),
+	char(-71),
+	char(0),
+	char(7),
+	char(0),
+	char(-70),
+	char(0),
+	char(7),
+	char(0),
+	char(-69),
+	char(0),
+	char(7),
+	char(0),
+	char(-68),
+	char(0),
+	char(64),
+	char(0),
+	char(9),
+	char(0),
+	char(55),
+	char(0),
+	char(-97),
+	char(0),
+	char(17),
+	char(0),
+	char(-94),
+	char(0),
+	char(17),
+	char(0),
+	char(-93),
+	char(0),
+	char(7),
+	char(0),
+	char(-78),
+	char(0),
+	char(7),
+	char(0),
+	char(-77),
+	char(0),
+	char(7),
+	char(0),
+	char(-76),
+	char(0),
+	char(7),
+	char(0),
+	char(-75),
+	char(0),
+	char(4),
+	char(0),
+	char(-74),
+	char(0),
+	char(4),
+	char(0),
+	char(-73),
+	char(0),
+	char(49),
+	char(0),
+	char(22),
+	char(0),
+	char(8),
+	char(0),
+	char(-67),
+	char(0),
+	char(8),
+	char(0),
+	char(-79),
+	char(0),
+	char(8),
+	char(0),
+	char(110),
+	char(0),
+	char(8),
+	char(0),
+	char(-66),
+	char(0),
+	char(8),
+	char(0),
+	char(112),
+	char(0),
+	char(8),
+	char(0),
+	char(-65),
+	char(0),
+	char(8),
+	char(0),
+	char(-64),
+	char(0),
+	char(8),
+	char(0),
+	char(-63),
+	char(0),
+	char(8),
+	char(0),
+	char(-62),
+	char(0),
+	char(8),
+	char(0),
+	char(-61),
+	char(0),
+	char(8),
+	char(0),
+	char(-60),
+	char(0),
+	char(8),
+	char(0),
+	char(-59),
+	char(0),
+	char(8),
+	char(0),
+	char(-58),
+	char(0),
+	char(8),
+	char(0),
+	char(-57),
+	char(0),
+	char(8),
+	char(0),
+	char(-56),
+	char(0),
+	char(8),
+	char(0),
+	char(-55),
+	char(0),
+	char(4),
+	char(0),
+	char(-54),
+	char(0),
+	char(4),
+	char(0),
+	char(-53),
+	char(0),
+	char(4),
+	char(0),
+	char(-52),
+	char(0),
+	char(4),
+	char(0),
+	char(-51),
+	char(0),
+	char(4),
+	char(0),
+	char(-50),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(51),
+	char(0),
+	char(22),
+	char(0),
+	char(7),
+	char(0),
+	char(-67),
+	char(0),
+	char(7),
+	char(0),
+	char(-79),
+	char(0),
+	char(7),
+	char(0),
+	char(110),
+	char(0),
+	char(7),
+	char(0),
+	char(-66),
+	char(0),
+	char(7),
+	char(0),
+	char(112),
+	char(0),
+	char(7),
+	char(0),
+	char(-65),
+	char(0),
+	char(7),
+	char(0),
+	char(-64),
+	char(0),
+	char(7),
+	char(0),
+	char(-63),
+	char(0),
+	char(7),
+	char(0),
+	char(-62),
+	char(0),
+	char(7),
+	char(0),
+	char(-61),
+	char(0),
+	char(7),
+	char(0),
+	char(-60),
+	char(0),
+	char(7),
+	char(0),
+	char(-59),
+	char(0),
+	char(7),
+	char(0),
+	char(-58),
+	char(0),
+	char(7),
+	char(0),
+	char(-57),
+	char(0),
+	char(7),
+	char(0),
+	char(-56),
+	char(0),
+	char(7),
+	char(0),
+	char(-55),
+	char(0),
+	char(4),
+	char(0),
+	char(-54),
+	char(0),
+	char(4),
+	char(0),
+	char(-53),
+	char(0),
+	char(4),
+	char(0),
+	char(-52),
+	char(0),
+	char(4),
+	char(0),
+	char(-51),
+	char(0),
+	char(4),
+	char(0),
+	char(-50),
+	char(0),
+	char(0),
+	char(0),
+	char(37),
+	char(0),
+	char(65),
+	char(0),
+	char(4),
+	char(0),
+	char(7),
+	char(0),
+	char(-49),
+	char(0),
+	char(7),
+	char(0),
+	char(-48),
+	char(0),
+	char(7),
+	char(0),
+	char(-47),
+	char(0),
+	char(4),
+	char(0),
+	char(78),
+	char(0),
+	char(66),
+	char(0),
+	char(10),
+	char(0),
+	char(65),
+	char(0),
+	char(-46),
+	char(0),
+	char(13),
+	char(0),
+	char(-45),
+	char(0),
+	char(13),
+	char(0),
+	char(-44),
+	char(0),
+	char(13),
+	char(0),
+	char(-43),
+	char(0),
+	char(13),
+	char(0),
+	char(-42),
+	char(0),
+	char(13),
+	char(0),
+	char(-41),
+	char(0),
+	char(7),
+	char(0),
+	char(-121),
+	char(0),
+	char(7),
+	char(0),
+	char(-40),
+	char(0),
+	char(4),
+	char(0),
+	char(-39),
+	char(0),
+	char(4),
+	char(0),
+	char(53),
+	char(0),
+	char(67),
+	char(0),
+	char(4),
+	char(0),
+	char(65),
+	char(0),
+	char(-46),
+	char(0),
+	char(4),
+	char(0),
+	char(-38),
+	char(0),
+	char(7),
+	char(0),
+	char(-37),
+	char(0),
+	char(4),
+	char(0),
+	char(-36),
+	char(0),
+	char(68),
+	char(0),
+	char(4),
+	char(0),
+	char(13),
+	char(0),
+	char(-41),
+	char(0),
+	char(65),
+	char(0),
+	char(-46),
+	char(0),
+	char(4),
+	char(0),
+	char(-35),
+	char(0),
+	char(7),
+	char(0),
+	char(-34),
+	char(0),
+	char(69),
+	char(0),
+	char(7),
+	char(0),
+	char(13),
+	char(0),
+	char(-33),
+	char(0),
+	char(65),
+	char(0),
+	char(-46),
+	char(0),
+	char(4),
+	char(0),
+	char(-32),
+	char(0),
+	char(7),
+	char(0),
+	char(-31),
+	char(0),
+	char(7),
+	char(0),
+	char(-30),
+	char(0),
+	char(7),
+	char(0),
+	char(-29),
+	char(0),
+	char(4),
+	char(0),
+	char(53),
+	char(0),
+	char(70),
+	char(0),
+	char(6),
+	char(0),
+	char(15),
+	char(0),
+	char(-28),
+	char(0),
+	char(13),
+	char(0),
+	char(-30),
+	char(0),
+	char(13),
+	char(0),
+	char(-27),
+	char(0),
+	char(56),
+	char(0),
+	char(-26),
+	char(0),
+	char(4),
+	char(0),
+	char(-25),
+	char(0),
+	char(7),
+	char(0),
+	char(-29),
+	char(0),
+	char(71),
+	char(0),
+	char(26),
+	char(0),
+	char(4),
+	char(0),
+	char(-24),
+	char(0),
+	char(7),
+	char(0),
+	char(-23),
+	char(0),
+	char(7),
+	char(0),
+	char(-79),
+	char(0),
+	char(7),
+	char(0),
+	char(-22),
+	char(0),
+	char(7),
+	char(0),
+	char(-21),
+	char(0),
+	char(7),
+	char(0),
+	char(-20),
+	char(0),
+	char(7),
+	char(0),
+	char(-19),
+	char(0),
+	char(7),
+	char(0),
+	char(-18),
+	char(0),
+	char(7),
+	char(0),
+	char(-17),
+	char(0),
+	char(7),
+	char(0),
+	char(-16),
+	char(0),
+	char(7),
+	char(0),
+	char(-15),
+	char(0),
+	char(7),
+	char(0),
+	char(-14),
+	char(0),
+	char(7),
+	char(0),
+	char(-13),
+	char(0),
+	char(7),
+	char(0),
+	char(-12),
+	char(0),
+	char(7),
+	char(0),
+	char(-11),
+	char(0),
+	char(7),
+	char(0),
+	char(-10),
+	char(0),
+	char(7),
+	char(0),
+	char(-9),
+	char(0),
+	char(7),
+	char(0),
+	char(-8),
+	char(0),
+	char(7),
+	char(0),
+	char(-7),
+	char(0),
+	char(7),
+	char(0),
+	char(-6),
+	char(0),
+	char(7),
+	char(0),
+	char(-5),
+	char(0),
+	char(4),
+	char(0),
+	char(-4),
+	char(0),
+	char(4),
+	char(0),
+	char(-3),
+	char(0),
+	char(4),
+	char(0),
+	char(-2),
+	char(0),
+	char(4),
+	char(0),
+	char(-1),
+	char(0),
+	char(4),
+	char(0),
+	char(117),
+	char(0),
+	char(72),
+	char(0),
+	char(12),
+	char(0),
+	char(15),
+	char(0),
+	char(0),
+	char(1),
+	char(15),
+	char(0),
+	char(1),
+	char(1),
+	char(15),
+	char(0),
+	char(2),
+	char(1),
+	char(13),
+	char(0),
+	char(3),
+	char(1),
+	char(13),
+	char(0),
+	char(4),
+	char(1),
+	char(7),
+	char(0),
+	char(5),
+	char(1),
+	char(4),
+	char(0),
+	char(6),
+	char(1),
+	char(4),
+	char(0),
+	char(7),
+	char(1),
+	char(4),
+	char(0),
+	char(8),
+	char(1),
+	char(4),
+	char(0),
+	char(9),
+	char(1),
+	char(7),
+	char(0),
+	char(-31),
+	char(0),
+	char(4),
+	char(0),
+	char(53),
+	char(0),
+	char(73),
+	char(0),
+	char(27),
+	char(0),
+	char(17),
+	char(0),
+	char(10),
+	char(1),
+	char(15),
+	char(0),
+	char(11),
+	char(1),
+	char(15),
+	char(0),
+	char(12),
+	char(1),
+	char(13),
+	char(0),
+	char(3),
+	char(1),
+	char(13),
+	char(0),
+	char(13),
+	char(1),
+	char(13),
+	char(0),
+	char(14),
+	char(1),
+	char(13),
+	char(0),
+	char(15),
+	char(1),
+	char(13),
+	char(0),
+	char(16),
+	char(1),
+	char(13),
+	char(0),
+	char(17),
+	char(1),
+	char(4),
+	char(0),
+	char(18),
+	char(1),
+	char(7),
+	char(0),
+	char(19),
+	char(1),
+	char(4),
+	char(0),
+	char(20),
+	char(1),
+	char(4),
+	char(0),
+	char(21),
+	char(1),
+	char(4),
+	char(0),
+	char(22),
+	char(1),
+	char(7),
+	char(0),
+	char(23),
+	char(1),
+	char(7),
+	char(0),
+	char(24),
+	char(1),
+	char(4),
+	char(0),
+	char(25),
+	char(1),
+	char(4),
+	char(0),
+	char(26),
+	char(1),
+	char(7),
+	char(0),
+	char(27),
+	char(1),
+	char(7),
+	char(0),
+	char(28),
+	char(1),
+	char(7),
+	char(0),
+	char(29),
+	char(1),
+	char(7),
+	char(0),
+	char(30),
+	char(1),
+	char(7),
+	char(0),
+	char(31),
+	char(1),
+	char(7),
+	char(0),
+	char(32),
+	char(1),
+	char(4),
+	char(0),
+	char(33),
+	char(1),
+	char(4),
+	char(0),
+	char(34),
+	char(1),
+	char(4),
+	char(0),
+	char(35),
+	char(1),
+	char(74),
+	char(0),
+	char(12),
+	char(0),
+	char(9),
+	char(0),
+	char(36),
+	char(1),
+	char(9),
+	char(0),
+	char(37),
+	char(1),
+	char(13),
+	char(0),
+	char(38),
+	char(1),
+	char(7),
+	char(0),
+	char(39),
+	char(1),
+	char(7),
+	char(0),
+	char(-63),
+	char(0),
+	char(7),
+	char(0),
+	char(40),
+	char(1),
+	char(4),
+	char(0),
+	char(41),
+	char(1),
+	char(13),
+	char(0),
+	char(42),
+	char(1),
+	char(4),
+	char(0),
+	char(43),
+	char(1),
+	char(4),
+	char(0),
+	char(44),
+	char(1),
+	char(4),
+	char(0),
+	char(45),
+	char(1),
+	char(4),
+	char(0),
+	char(53),
+	char(0),
+	char(75),
+	char(0),
+	char(19),
+	char(0),
+	char(47),
+	char(0),
+	char(125),
+	char(0),
+	char(72),
+	char(0),
+	char(46),
+	char(1),
+	char(65),
+	char(0),
+	char(47),
+	char(1),
+	char(66),
+	char(0),
+	char(48),
+	char(1),
+	char(67),
+	char(0),
+	char(49),
+	char(1),
+	char(68),
+	char(0),
+	char(50),
+	char(1),
+	char(69),
+	char(0),
+	char(51),
+	char(1),
+	char(70),
+	char(0),
+	char(52),
+	char(1),
+	char(73),
+	char(0),
+	char(53),
+	char(1),
+	char(74),
+	char(0),
+	char(54),
+	char(1),
+	char(4),
+	char(0),
+	char(55),
+	char(1),
+	char(4),
+	char(0),
+	char(21),
+	char(1),
+	char(4),
+	char(0),
+	char(56),
+	char(1),
+	char(4),
+	char(0),
+	char(57),
+	char(1),
+	char(4),
+	char(0),
+	char(58),
+	char(1),
+	char(4),
+	char(0),
+	char(59),
+	char(1),
+	char(4),
+	char(0),
+	char(60),
+	char(1),
+	char(4),
+	char(0),
+	char(61),
+	char(1),
+	char(71),
+	char(0),
+	char(62),
+	char(1),
+};
+int b3s_bulletDNAlen64 = sizeof(b3s_bulletDNAstr64);
diff --git a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Serializer.h b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Serializer.h
index 1c1ce437647..d9e153e238f 100644
--- a/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Serializer.h
+++ b/thirdparty/bullet/Bullet3Serialize/Bullet2FileLoader/b3Serializer.h
@@ -16,158 +16,142 @@ subject to the following restrictions:
 #ifndef B3_SERIALIZER_H
 #define B3_SERIALIZER_H
 
-#include "Bullet3Common/b3Scalar.h" // has definitions like B3_FORCE_INLINE
+#include "Bullet3Common/b3Scalar.h"  // has definitions like B3_FORCE_INLINE
 #include "Bullet3Common/b3StackAlloc.h"
 #include "Bullet3Common/b3HashMap.h"
 
-#if !defined( __CELLOS_LV2__) && !defined(__MWERKS__)
+#if !defined(__CELLOS_LV2__) && !defined(__MWERKS__)
 #include <memory.h>
 #endif
 #include <string.h>
 
-
-
 extern char b3s_bulletDNAstr[];
 extern int b3s_bulletDNAlen;
 extern char b3s_bulletDNAstr64[];
 extern int b3s_bulletDNAlen64;
 
-B3_FORCE_INLINE	int b3StrLen(const char* str) 
+B3_FORCE_INLINE int b3StrLen(const char* str)
 {
-    if (!str) 
-		return(0);
+	if (!str)
+		return (0);
 	int len = 0;
-    
+
 	while (*str != 0)
 	{
-        str++;
-        len++;
-    }
+		str++;
+		len++;
+	}
 
-    return len;
+	return len;
 }
 
-
 class b3Chunk
 {
 public:
-	int		m_chunkCode;
-	int		m_length;
-	void	*m_oldPtr;
-	int		m_dna_nr;
-	int		m_number;
+	int m_chunkCode;
+	int m_length;
+	void* m_oldPtr;
+	int m_dna_nr;
+	int m_number;
 };
 
-enum	b3SerializationFlags
+enum b3SerializationFlags
 {
 	B3_SERIALIZE_NO_BVH = 1,
 	B3_SERIALIZE_NO_TRIANGLEINFOMAP = 2,
 	B3_SERIALIZE_NO_DUPLICATE_ASSERT = 4
 };
 
-class	b3Serializer
+class b3Serializer
 {
-
 public:
-
 	virtual ~b3Serializer() {}
 
-	virtual	const unsigned char*		getBufferPointer() const = 0;
+	virtual const unsigned char* getBufferPointer() const = 0;
 
-	virtual	int		getCurrentBufferSize() const = 0;
+	virtual int getCurrentBufferSize() const = 0;
 
-	virtual	b3Chunk*	allocate(size_t size, int numElements) = 0;
+	virtual b3Chunk* allocate(size_t size, int numElements) = 0;
 
-	virtual	void	finalizeChunk(b3Chunk* chunk, const char* structType, int chunkCode,void* oldPtr)= 0;
+	virtual void finalizeChunk(b3Chunk* chunk, const char* structType, int chunkCode, void* oldPtr) = 0;
 
-	virtual	 void*	findPointer(void* oldPtr)  = 0;
+	virtual void* findPointer(void* oldPtr) = 0;
 
-	virtual	void*	getUniquePointer(void*oldPtr) = 0;
+	virtual void* getUniquePointer(void* oldPtr) = 0;
 
-	virtual	void	startSerialization() = 0;
-	
-	virtual	void	finishSerialization() = 0;
+	virtual void startSerialization() = 0;
 
-	virtual	const char*	findNameForPointer(const void* ptr) const = 0;
+	virtual void finishSerialization() = 0;
 
-	virtual	void	registerNameForPointer(const void* ptr, const char* name) = 0;
+	virtual const char* findNameForPointer(const void* ptr) const = 0;
 
-	virtual void	serializeName(const char* ptr) = 0;
+	virtual void registerNameForPointer(const void* ptr, const char* name) = 0;
 
-	virtual int		getSerializationFlags() const = 0;
-
-	virtual void	setSerializationFlags(int flags) = 0;
+	virtual void serializeName(const char* ptr) = 0;
 
+	virtual int getSerializationFlags() const = 0;
 
+	virtual void setSerializationFlags(int flags) = 0;
 };
 
-
-
 #define B3_HEADER_LENGTH 12
-#if defined(__sgi) || defined (__sparc) || defined (__sparc__) || defined (__PPC__) || defined (__ppc__) || defined (__BIG_ENDIAN__)
-#	define B3_MAKE_ID(a,b,c,d) ( (int)(a)<<24 | (int)(b)<<16 | (c)<<8 | (d) )
+#if defined(__sgi) || defined(__sparc) || defined(__sparc__) || defined(__PPC__) || defined(__ppc__) || defined(__BIG_ENDIAN__)
+#define B3_MAKE_ID(a, b, c, d) ((int)(a) << 24 | (int)(b) << 16 | (c) << 8 | (d))
 #else
-#	define B3_MAKE_ID(a,b,c,d) ( (int)(d)<<24 | (int)(c)<<16 | (b)<<8 | (a) )
+#define B3_MAKE_ID(a, b, c, d) ((int)(d) << 24 | (int)(c) << 16 | (b) << 8 | (a))
 #endif
 
-#define B3_SOFTBODY_CODE		B3_MAKE_ID('S','B','D','Y')
-#define B3_COLLISIONOBJECT_CODE B3_MAKE_ID('C','O','B','J')
-#define B3_RIGIDBODY_CODE		B3_MAKE_ID('R','B','D','Y')
-#define B3_CONSTRAINT_CODE		B3_MAKE_ID('C','O','N','S')
-#define B3_BOXSHAPE_CODE		B3_MAKE_ID('B','O','X','S')
-#define B3_QUANTIZED_BVH_CODE	B3_MAKE_ID('Q','B','V','H')
-#define B3_TRIANLGE_INFO_MAP	B3_MAKE_ID('T','M','A','P')
-#define B3_SHAPE_CODE			B3_MAKE_ID('S','H','A','P')
-#define B3_ARRAY_CODE			B3_MAKE_ID('A','R','A','Y')
-#define B3_SBMATERIAL_CODE		B3_MAKE_ID('S','B','M','T')
-#define B3_SBNODE_CODE			B3_MAKE_ID('S','B','N','D')
-#define B3_DYNAMICSWORLD_CODE	B3_MAKE_ID('D','W','L','D')
-#define B3_DNA_CODE				B3_MAKE_ID('D','N','A','1')
+#define B3_SOFTBODY_CODE B3_MAKE_ID('S', 'B', 'D', 'Y')
+#define B3_COLLISIONOBJECT_CODE B3_MAKE_ID('C', 'O', 'B', 'J')
+#define B3_RIGIDBODY_CODE B3_MAKE_ID('R', 'B', 'D', 'Y')
+#define B3_CONSTRAINT_CODE B3_MAKE_ID('C', 'O', 'N', 'S')
+#define B3_BOXSHAPE_CODE B3_MAKE_ID('B', 'O', 'X', 'S')
+#define B3_QUANTIZED_BVH_CODE B3_MAKE_ID('Q', 'B', 'V', 'H')
+#define B3_TRIANLGE_INFO_MAP B3_MAKE_ID('T', 'M', 'A', 'P')
+#define B3_SHAPE_CODE B3_MAKE_ID('S', 'H', 'A', 'P')
+#define B3_ARRAY_CODE B3_MAKE_ID('A', 'R', 'A', 'Y')
+#define B3_SBMATERIAL_CODE B3_MAKE_ID('S', 'B', 'M', 'T')
+#define B3_SBNODE_CODE B3_MAKE_ID('S', 'B', 'N', 'D')
+#define B3_DYNAMICSWORLD_CODE B3_MAKE_ID('D', 'W', 'L', 'D')
+#define B3_DNA_CODE B3_MAKE_ID('D', 'N', 'A', '1')
 
-
-struct	b3PointerUid
+struct b3PointerUid
 {
-	union
-	{
-		void*	m_ptr;
-		int		m_uniqueIds[2];
+	union {
+		void* m_ptr;
+		int m_uniqueIds[2];
 	};
 };
 
 ///The b3DefaultSerializer is the main Bullet serialization class.
 ///The constructor takes an optional argument for backwards compatibility, it is recommended to leave this empty/zero.
-class b3DefaultSerializer	:	public b3Serializer
+class b3DefaultSerializer : public b3Serializer
 {
+	b3AlignedObjectArray<char*> mTypes;
+	b3AlignedObjectArray<short*> mStructs;
+	b3AlignedObjectArray<short> mTlens;
+	b3HashMap<b3HashInt, int> mStructReverse;
+	b3HashMap<b3HashString, int> mTypeLookup;
 
+	b3HashMap<b3HashPtr, void*> m_chunkP;
 
-	b3AlignedObjectArray<char*>			mTypes;
-	b3AlignedObjectArray<short*>			mStructs;
-	b3AlignedObjectArray<short>			mTlens;
-	b3HashMap<b3HashInt, int>			mStructReverse;
-	b3HashMap<b3HashString,int>	mTypeLookup;
+	b3HashMap<b3HashPtr, const char*> m_nameMap;
 
-	
-	b3HashMap<b3HashPtr,void*>	m_chunkP;
-	
-	b3HashMap<b3HashPtr,const char*>	m_nameMap;
+	b3HashMap<b3HashPtr, b3PointerUid> m_uniquePointers;
+	int m_uniqueIdGenerator;
 
-	b3HashMap<b3HashPtr,b3PointerUid>	m_uniquePointers;
-	int	m_uniqueIdGenerator;
+	int m_totalSize;
+	unsigned char* m_buffer;
+	int m_currentSize;
+	void* m_dna;
+	int m_dnaLength;
 
-	int					m_totalSize;
-	unsigned char*		m_buffer;
-	int					m_currentSize;
-	void*				m_dna;
-	int					m_dnaLength;
+	int m_serializationFlags;
 
-	int					m_serializationFlags;
+	b3AlignedObjectArray<b3Chunk*> m_chunkPtrs;
 
-
-	b3AlignedObjectArray<b3Chunk*>	m_chunkPtrs;
-	
 protected:
-
-	virtual	void*	findPointer(void* oldPtr) 
+	virtual void* findPointer(void* oldPtr)
 	{
 		void** ptr = m_chunkP.find(oldPtr);
 		if (ptr && *ptr)
@@ -175,48 +159,43 @@ protected:
 		return 0;
 	}
 
-	
+	void writeDNA()
+	{
+		b3Chunk* dnaChunk = allocate(m_dnaLength, 1);
+		memcpy(dnaChunk->m_oldPtr, m_dna, m_dnaLength);
+		finalizeChunk(dnaChunk, "DNA1", B3_DNA_CODE, m_dna);
+	}
 
+	int getReverseType(const char* type) const
+	{
+		b3HashString key(type);
+		const int* valuePtr = mTypeLookup.find(key);
+		if (valuePtr)
+			return *valuePtr;
 
+		return -1;
+	}
 
-		void	writeDNA()
-		{
-			b3Chunk* dnaChunk = allocate(m_dnaLength,1);
-			memcpy(dnaChunk->m_oldPtr,m_dna,m_dnaLength);
-			finalizeChunk(dnaChunk,"DNA1",B3_DNA_CODE, m_dna);
-		}
+	void initDNA(const char* bdnaOrg, int dnalen)
+	{
+		///was already initialized
+		if (m_dna)
+			return;
 
-		int getReverseType(const char *type) const
-		{
+		int littleEndian = 1;
+		littleEndian = ((char*)&littleEndian)[0];
 
-			b3HashString key(type);
-			const int* valuePtr = mTypeLookup.find(key);
-			if (valuePtr)
-				return *valuePtr;
-			
-			return -1;
-		}
+		m_dna = b3AlignedAlloc(dnalen, 16);
+		memcpy(m_dna, bdnaOrg, dnalen);
+		m_dnaLength = dnalen;
 
-		void initDNA(const char* bdnaOrg,int dnalen)
-		{
-			///was already initialized
-			if (m_dna)
-				return;
+		int* intPtr = 0;
+		short* shtPtr = 0;
+		char* cp = 0;
+		int dataLen = 0;
+		intPtr = (int*)m_dna;
 
-			int littleEndian= 1;
-			littleEndian= ((char*)&littleEndian)[0];
-			
-
-			m_dna = b3AlignedAlloc(dnalen,16);
-			memcpy(m_dna,bdnaOrg,dnalen);
-			m_dnaLength = dnalen;
-
-			int *intPtr=0;
-			short *shtPtr=0;
-			char *cp = 0;int dataLen =0;
-			intPtr = (int*)m_dna;
-
-			/*
+		/*
 				SDNA (4 bytes) (magic number)
 				NAME (4 bytes)
 				<nr> (4 bytes) amount of names (int)
@@ -224,81 +203,81 @@ protected:
 				<string>
 			*/
 
-			if (strncmp((const char*)m_dna, "SDNA", 4)==0)
-			{
-				// skip ++ NAME
-				intPtr++; intPtr++;
-			}
-
-			// Parse names
-			if (!littleEndian)
-				*intPtr = b3SwapEndian(*intPtr);
-				
-			dataLen = *intPtr;
-			
+		if (strncmp((const char*)m_dna, "SDNA", 4) == 0)
+		{
+			// skip ++ NAME
 			intPtr++;
+			intPtr++;
+		}
 
-			cp = (char*)intPtr;
-			int i;
-			for ( i=0; i<dataLen; i++)
-			{
-				
-				while (*cp)cp++;
-				cp++;
-			}
-			cp = b3AlignPointer(cp,4);
+		// Parse names
+		if (!littleEndian)
+			*intPtr = b3SwapEndian(*intPtr);
 
-			/*
+		dataLen = *intPtr;
+
+		intPtr++;
+
+		cp = (char*)intPtr;
+		int i;
+		for (i = 0; i < dataLen; i++)
+		{
+			while (*cp) cp++;
+			cp++;
+		}
+		cp = b3AlignPointer(cp, 4);
+
+		/*
 				TYPE (4 bytes)
 				<nr> amount of types (int)
 				<string>
 				<string>
 			*/
 
-			intPtr = (int*)cp;
-			b3Assert(strncmp(cp, "TYPE", 4)==0); intPtr++;
+		intPtr = (int*)cp;
+		b3Assert(strncmp(cp, "TYPE", 4) == 0);
+		intPtr++;
 
-			if (!littleEndian)
-				*intPtr =  b3SwapEndian(*intPtr);
-			
-			dataLen = *intPtr;
-			intPtr++;
+		if (!littleEndian)
+			*intPtr = b3SwapEndian(*intPtr);
 
-			
-			cp = (char*)intPtr;
-			for (i=0; i<dataLen; i++)
-			{
-				mTypes.push_back(cp);
-				while (*cp)cp++;
-				cp++;
-			}
+		dataLen = *intPtr;
+		intPtr++;
 
-			cp = b3AlignPointer(cp,4);
+		cp = (char*)intPtr;
+		for (i = 0; i < dataLen; i++)
+		{
+			mTypes.push_back(cp);
+			while (*cp) cp++;
+			cp++;
+		}
 
+		cp = b3AlignPointer(cp, 4);
 
-			/*
+		/*
 				TLEN (4 bytes)
 				<len> (short) the lengths of types
 				<len>
 			*/
 
-			// Parse type lens
-			intPtr = (int*)cp;
-			b3Assert(strncmp(cp, "TLEN", 4)==0); intPtr++;
+		// Parse type lens
+		intPtr = (int*)cp;
+		b3Assert(strncmp(cp, "TLEN", 4) == 0);
+		intPtr++;
 
-			dataLen = (int)mTypes.size();
+		dataLen = (int)mTypes.size();
 
-			shtPtr = (short*)intPtr;
-			for (i=0; i<dataLen; i++, shtPtr++)
-			{
-				if (!littleEndian)
-					shtPtr[0] = b3SwapEndian(shtPtr[0]);
-				mTlens.push_back(shtPtr[0]);
-			}
+		shtPtr = (short*)intPtr;
+		for (i = 0; i < dataLen; i++, shtPtr++)
+		{
+			if (!littleEndian)
+				shtPtr[0] = b3SwapEndian(shtPtr[0]);
+			mTlens.push_back(shtPtr[0]);
+		}
 
-			if (dataLen & 1) shtPtr++;
+		if (dataLen & 1) shtPtr++;
 
-			/*
+		/*
 				STRC (4 bytes)
 				<nr> amount of structs (int)
 				<typenr>
@@ -309,331 +288,314 @@ protected:
 				<namenr>
 			*/
 
-			intPtr = (int*)shtPtr;
-			cp = (char*)intPtr;
-			b3Assert(strncmp(cp, "STRC", 4)==0); intPtr++;
+		intPtr = (int*)shtPtr;
+		cp = (char*)intPtr;
+		b3Assert(strncmp(cp, "STRC", 4) == 0);
+		intPtr++;
+
+		if (!littleEndian)
+			*intPtr = b3SwapEndian(*intPtr);
+		dataLen = *intPtr;
+		intPtr++;
+
+		shtPtr = (short*)intPtr;
+		for (i = 0; i < dataLen; i++)
+		{
+			mStructs.push_back(shtPtr);
 
 			if (!littleEndian)
-				*intPtr = b3SwapEndian(*intPtr);
-			dataLen = *intPtr ; 
-			intPtr++;
-
-
-			shtPtr = (short*)intPtr;
-			for (i=0; i<dataLen; i++)
 			{
-				mStructs.push_back (shtPtr);
-				
-				if (!littleEndian)
+				shtPtr[0] = b3SwapEndian(shtPtr[0]);
+				shtPtr[1] = b3SwapEndian(shtPtr[1]);
+
+				int len = shtPtr[1];
+				shtPtr += 2;
+
+				for (int a = 0; a < len; a++, shtPtr += 2)
 				{
-					shtPtr[0]= b3SwapEndian(shtPtr[0]);
-					shtPtr[1]= b3SwapEndian(shtPtr[1]);
-
-					int len = shtPtr[1];
-					shtPtr+= 2;
-
-					for (int a=0; a<len; a++, shtPtr+=2)
-					{
-							shtPtr[0]= b3SwapEndian(shtPtr[0]);
-							shtPtr[1]= b3SwapEndian(shtPtr[1]);
-					}
-
-				} else
-				{
-					shtPtr+= (2*shtPtr[1])+2;
+					shtPtr[0] = b3SwapEndian(shtPtr[0]);
+					shtPtr[1] = b3SwapEndian(shtPtr[1]);
 				}
 			}
-
-			// build reverse lookups
-			for (i=0; i<(int)mStructs.size(); i++)
+			else
 			{
-				short *strc = mStructs.at(i);
-				mStructReverse.insert(strc[0], i);
-				mTypeLookup.insert(b3HashString(mTypes[strc[0]]),i);
+				shtPtr += (2 * shtPtr[1]) + 2;
 			}
 		}
 
-public:	
-	
-
-	
-
-		b3DefaultSerializer(int totalSize=0)
-			:m_totalSize(totalSize),
-			m_currentSize(0),
-			m_dna(0),
-			m_dnaLength(0),
-			m_serializationFlags(0)
+		// build reverse lookups
+		for (i = 0; i < (int)mStructs.size(); i++)
 		{
-			m_buffer = m_totalSize?(unsigned char*)b3AlignedAlloc(totalSize,16):0;
-			
-			const bool VOID_IS_8 = ((sizeof(void*)==8));
+			short* strc = mStructs.at(i);
+			mStructReverse.insert(strc[0], i);
+			mTypeLookup.insert(b3HashString(mTypes[strc[0]]), i);
+		}
+	}
+
+public:
+	b3DefaultSerializer(int totalSize = 0)
+		: m_totalSize(totalSize),
+		  m_currentSize(0),
+		  m_dna(0),
+		  m_dnaLength(0),
+		  m_serializationFlags(0)
+	{
+		m_buffer = m_totalSize ? (unsigned char*)b3AlignedAlloc(totalSize, 16) : 0;
+
+		const bool VOID_IS_8 = ((sizeof(void*) == 8));
 
 #ifdef B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
-			if (VOID_IS_8)
-			{
+		if (VOID_IS_8)
+		{
 #if _WIN64
-				initDNA((const char*)b3s_bulletDNAstr64,b3s_bulletDNAlen64);
+			initDNA((const char*)b3s_bulletDNAstr64, b3s_bulletDNAlen64);
 #else
-				b3Assert(0);
+			b3Assert(0);
 #endif
-			} else
-			{
+		}
+		else
+		{
 #ifndef _WIN64
-				initDNA((const char*)b3s_bulletDNAstr,b3s_bulletDNAlen);
+			initDNA((const char*)b3s_bulletDNAstr, b3s_bulletDNAlen);
 #else
-				b3Assert(0);
+			b3Assert(0);
 #endif
-			}
-	
-#else //B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
-			if (VOID_IS_8)
-			{
-				initDNA((const char*)b3s_bulletDNAstr64,b3s_bulletDNAlen64);
-			} else
-			{
-				initDNA((const char*)b3s_bulletDNAstr,b3s_bulletDNAlen);
-			}
-#endif //B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
-	
 		}
 
-		virtual ~b3DefaultSerializer() 
+#else   //B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
+		if (VOID_IS_8)
+		{
+			initDNA((const char*)b3s_bulletDNAstr64, b3s_bulletDNAlen64);
+		}
+		else
+		{
+			initDNA((const char*)b3s_bulletDNAstr, b3s_bulletDNAlen);
+		}
+#endif  //B3_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
+	}
+
+	virtual ~b3DefaultSerializer()
+	{
+		if (m_buffer)
+			b3AlignedFree(m_buffer);
+		if (m_dna)
+			b3AlignedFree(m_dna);
+	}
+
+	void writeHeader(unsigned char* buffer) const
+	{
+#ifdef B3_USE_DOUBLE_PRECISION
+		memcpy(buffer, "BULLETd", 7);
+#else
+		memcpy(buffer, "BULLETf", 7);
+#endif  //B3_USE_DOUBLE_PRECISION
+
+		int littleEndian = 1;
+		littleEndian = ((char*)&littleEndian)[0];
+
+		if (sizeof(void*) == 8)
+		{
+			buffer[7] = '-';
+		}
+		else
+		{
+			buffer[7] = '_';
+		}
+
+		if (littleEndian)
+		{
+			buffer[8] = 'v';
+		}
+		else
+		{
+			buffer[8] = 'V';
+		}
+
+		buffer[9] = '2';
+		buffer[10] = '8';
+		buffer[11] = '1';
+	}
+
+	virtual void startSerialization()
+	{
+		m_uniqueIdGenerator = 1;
+		if (m_totalSize)
+		{
+			unsigned char* buffer = internalAlloc(B3_HEADER_LENGTH);
+			writeHeader(buffer);
+		}
+	}
+
+	virtual void finishSerialization()
+	{
+		writeDNA();
+
+		//if we didn't pre-allocate a buffer, we need to create a contiguous buffer now
+		int mysize = 0;
+		if (!m_totalSize)
 		{
 			if (m_buffer)
 				b3AlignedFree(m_buffer);
-			if (m_dna)
-				b3AlignedFree(m_dna);
-		}
 
-		void	writeHeader(unsigned char* buffer) const
-		{
-			
+			m_currentSize += B3_HEADER_LENGTH;
+			m_buffer = (unsigned char*)b3AlignedAlloc(m_currentSize, 16);
 
-#ifdef  B3_USE_DOUBLE_PRECISION
-			memcpy(buffer, "BULLETd", 7);
-#else
-			memcpy(buffer, "BULLETf", 7);
-#endif //B3_USE_DOUBLE_PRECISION
-	
-			int littleEndian= 1;
-			littleEndian= ((char*)&littleEndian)[0];
-
-			if (sizeof(void*)==8)
+			unsigned char* currentPtr = m_buffer;
+			writeHeader(m_buffer);
+			currentPtr += B3_HEADER_LENGTH;
+			mysize += B3_HEADER_LENGTH;
+			for (int i = 0; i < m_chunkPtrs.size(); i++)
 			{
-				buffer[7] = '-';
-			} else
-			{
-				buffer[7] = '_';
+				int curLength = sizeof(b3Chunk) + m_chunkPtrs[i]->m_length;
+				memcpy(currentPtr, m_chunkPtrs[i], curLength);
+				b3AlignedFree(m_chunkPtrs[i]);
+				currentPtr += curLength;
+				mysize += curLength;
 			}
-
-			if (littleEndian)
-			{
-				buffer[8]='v';				
-			} else
-			{
-				buffer[8]='V';
-			}
-
-
-			buffer[9] = '2';
-			buffer[10] = '8';
-			buffer[11] = '1';
-
 		}
 
-		virtual	void	startSerialization()
-		{
-			m_uniqueIdGenerator= 1;
-			if (m_totalSize)
-			{
-				unsigned char* buffer = internalAlloc(B3_HEADER_LENGTH);
-				writeHeader(buffer);
-			}
-			
-		}
+		mTypes.clear();
+		mStructs.clear();
+		mTlens.clear();
+		mStructReverse.clear();
+		mTypeLookup.clear();
+		m_chunkP.clear();
+		m_nameMap.clear();
+		m_uniquePointers.clear();
+		m_chunkPtrs.clear();
+	}
 
-		virtual	void	finishSerialization()
-		{
-			writeDNA();
-
-			//if we didn't pre-allocate a buffer, we need to create a contiguous buffer now
-			int mysize = 0;
-			if (!m_totalSize)
-			{
-				if (m_buffer)
-					b3AlignedFree(m_buffer);
-
-				m_currentSize += B3_HEADER_LENGTH;
-				m_buffer = (unsigned char*)b3AlignedAlloc(m_currentSize,16);
-
-				unsigned char* currentPtr = m_buffer;
-				writeHeader(m_buffer);
-				currentPtr += B3_HEADER_LENGTH;
-				mysize+=B3_HEADER_LENGTH;
-				for (int i=0;i<	m_chunkPtrs.size();i++)
-				{
-					int curLength = sizeof(b3Chunk)+m_chunkPtrs[i]->m_length;
-					memcpy(currentPtr,m_chunkPtrs[i], curLength);
-					b3AlignedFree(m_chunkPtrs[i]);
-					currentPtr+=curLength;
-					mysize+=curLength;
-				}
-			}
-
-			mTypes.clear();
-			mStructs.clear();
-			mTlens.clear();
-			mStructReverse.clear();
-			mTypeLookup.clear();
-			m_chunkP.clear();
-			m_nameMap.clear();
-			m_uniquePointers.clear();
-			m_chunkPtrs.clear();
-		}
-
-		virtual	void*	getUniquePointer(void*oldPtr)
-		{
-			if (!oldPtr)
-				return 0;
-
-			b3PointerUid* uptr = (b3PointerUid*)m_uniquePointers.find(oldPtr);
-			if (uptr)
-			{
-				return uptr->m_ptr;
-			}
-			m_uniqueIdGenerator++;
-			
-			b3PointerUid uid;
-			uid.m_uniqueIds[0] = m_uniqueIdGenerator;
-			uid.m_uniqueIds[1] = m_uniqueIdGenerator;
-			m_uniquePointers.insert(oldPtr,uid);
-			return uid.m_ptr;
-
-		}
-
-		virtual	const unsigned char*		getBufferPointer() const
-		{
-			return m_buffer;
-		}
-
-		virtual	int					getCurrentBufferSize() const
-		{
-			return	m_currentSize;
-		}
-
-		virtual	void	finalizeChunk(b3Chunk* chunk, const char* structType, int chunkCode,void* oldPtr)
-		{
-			if (!(m_serializationFlags&B3_SERIALIZE_NO_DUPLICATE_ASSERT))
-			{
-				b3Assert(!findPointer(oldPtr));
-			}
-
-			chunk->m_dna_nr = getReverseType(structType);
-			
-			chunk->m_chunkCode = chunkCode;
-			
-			void* uniquePtr = getUniquePointer(oldPtr);
-			
-			m_chunkP.insert(oldPtr,uniquePtr);//chunk->m_oldPtr);
-			chunk->m_oldPtr = uniquePtr;//oldPtr;
-			
-		}
-
-		
-		virtual unsigned char* internalAlloc(size_t size)
-		{
-			unsigned char* ptr = 0;
-
-			if (m_totalSize)
-			{
-				ptr = m_buffer+m_currentSize;
-				m_currentSize += int(size);
-				b3Assert(m_currentSize<m_totalSize);
-			} else
-			{
-				ptr = (unsigned char*)b3AlignedAlloc(size,16);
-				m_currentSize += int(size);
-			}
-			return ptr;
-		}
-
-		
-
-		virtual	b3Chunk*	allocate(size_t size, int numElements)
-		{
-
-			unsigned char* ptr = internalAlloc(int(size)*numElements+sizeof(b3Chunk));
-
-			unsigned char* data = ptr + sizeof(b3Chunk);
-			
-			b3Chunk* chunk = (b3Chunk*)ptr;
-			chunk->m_chunkCode = 0;
-			chunk->m_oldPtr = data;
-			chunk->m_length = int(size)*numElements;
-			chunk->m_number = numElements;
-			
-			m_chunkPtrs.push_back(chunk);
-			
-
-			return chunk;
-		}
-
-		virtual	const char*	findNameForPointer(const void* ptr) const
-		{
-			const char*const * namePtr = m_nameMap.find(ptr);
-			if (namePtr && *namePtr)
-				return *namePtr;
+	virtual void* getUniquePointer(void* oldPtr)
+	{
+		if (!oldPtr)
 			return 0;
 
+		b3PointerUid* uptr = (b3PointerUid*)m_uniquePointers.find(oldPtr);
+		if (uptr)
+		{
+			return uptr->m_ptr;
+		}
+		m_uniqueIdGenerator++;
+
+		b3PointerUid uid;
+		uid.m_uniqueIds[0] = m_uniqueIdGenerator;
+		uid.m_uniqueIds[1] = m_uniqueIdGenerator;
+		m_uniquePointers.insert(oldPtr, uid);
+		return uid.m_ptr;
+	}
+
+	virtual const unsigned char* getBufferPointer() const
+	{
+		return m_buffer;
+	}
+
+	virtual int getCurrentBufferSize() const
+	{
+		return m_currentSize;
+	}
+
+	virtual void finalizeChunk(b3Chunk* chunk, const char* structType, int chunkCode, void* oldPtr)
+	{
+		if (!(m_serializationFlags & B3_SERIALIZE_NO_DUPLICATE_ASSERT))
+		{
+			b3Assert(!findPointer(oldPtr));
 		}
 
-		virtual	void	registerNameForPointer(const void* ptr, const char* name)
-		{
-			m_nameMap.insert(ptr,name);
-		}
+		chunk->m_dna_nr = getReverseType(structType);
 
-		virtual void	serializeName(const char* name)
+		chunk->m_chunkCode = chunkCode;
+
+		void* uniquePtr = getUniquePointer(oldPtr);
+
+		m_chunkP.insert(oldPtr, uniquePtr);  //chunk->m_oldPtr);
+		chunk->m_oldPtr = uniquePtr;         //oldPtr;
+	}
+
+	virtual unsigned char* internalAlloc(size_t size)
+	{
+		unsigned char* ptr = 0;
+
+		if (m_totalSize)
 		{
-			if (name)
+			ptr = m_buffer + m_currentSize;
+			m_currentSize += int(size);
+			b3Assert(m_currentSize < m_totalSize);
+		}
+		else
+		{
+			ptr = (unsigned char*)b3AlignedAlloc(size, 16);
+			m_currentSize += int(size);
+		}
+		return ptr;
+	}
+
+	virtual b3Chunk* allocate(size_t size, int numElements)
+	{
+		unsigned char* ptr = internalAlloc(int(size) * numElements + sizeof(b3Chunk));
+
+		unsigned char* data = ptr + sizeof(b3Chunk);
+
+		b3Chunk* chunk = (b3Chunk*)ptr;
+		chunk->m_chunkCode = 0;
+		chunk->m_oldPtr = data;
+		chunk->m_length = int(size) * numElements;
+		chunk->m_number = numElements;
+
+		m_chunkPtrs.push_back(chunk);
+
+		return chunk;
+	}
+
+	virtual const char* findNameForPointer(const void* ptr) const
+	{
+		const char* const* namePtr = m_nameMap.find(ptr);
+		if (namePtr && *namePtr)
+			return *namePtr;
+		return 0;
+	}
+
+	virtual void registerNameForPointer(const void* ptr, const char* name)
+	{
+		m_nameMap.insert(ptr, name);
+	}
+
+	virtual void serializeName(const char* name)
+	{
+		if (name)
+		{
+			//don't serialize name twice
+			if (findPointer((void*)name))
+				return;
+
+			int len = b3StrLen(name);
+			if (len)
 			{
-				//don't serialize name twice
-				if (findPointer((void*)name))
-					return;
+				int newLen = len + 1;
+				int padding = ((newLen + 3) & ~3) - newLen;
+				newLen += padding;
 
-				int len = b3StrLen(name);
-				if (len)
+				//serialize name string now
+				b3Chunk* chunk = allocate(sizeof(char), newLen);
+				char* destinationName = (char*)chunk->m_oldPtr;
+				for (int i = 0; i < len; i++)
 				{
-
-					int newLen = len+1;
-					int padding = ((newLen+3)&~3)-newLen;
-					newLen += padding;
-
-					//serialize name string now
-					b3Chunk* chunk = allocate(sizeof(char),newLen);
-					char* destinationName = (char*)chunk->m_oldPtr;
-					for (int i=0;i<len;i++)
-					{
-						destinationName[i] = name[i];
-					}
-					destinationName[len] = 0;
-					finalizeChunk(chunk,"char",B3_ARRAY_CODE,(void*)name);
+					destinationName[i] = name[i];
 				}
+				destinationName[len] = 0;
+				finalizeChunk(chunk, "char", B3_ARRAY_CODE, (void*)name);
 			}
 		}
+	}
 
-		virtual int		getSerializationFlags() const
-		{
-			return m_serializationFlags;
-		}
-
-		virtual void	setSerializationFlags(int flags)
-		{
-			m_serializationFlags = flags;
-		}
+	virtual int getSerializationFlags() const
+	{
+		return m_serializationFlags;
+	}
 
+	virtual void setSerializationFlags(int flags)
+	{
+		m_serializationFlags = flags;
+	}
 };
 
-
-#endif //B3_SERIALIZER_H
-
+#endif  //B3_SERIALIZER_H
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btAxisSweep3.cpp b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btAxisSweep3.cpp
index 77763305b1b..ec6fe9f4d8a 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btAxisSweep3.cpp
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btAxisSweep3.cpp
@@ -2,7 +2,6 @@
 //Bullet Continuous Collision Detection and Physics Library
 //Copyright (c) 2003-2006 Erwin Coumans  http://continuousphysics.com/Bullet/
 
-
 //
 // btAxisSweep3
 //
@@ -19,18 +18,15 @@
 // 3. This notice may not be removed or altered from any source distribution.
 #include "btAxisSweep3.h"
 
-
-btAxisSweep3::btAxisSweep3(const btVector3& worldAabbMin,const btVector3& worldAabbMax, unsigned short int maxHandles, btOverlappingPairCache* pairCache, bool disableRaycastAccelerator)
-:btAxisSweep3Internal<unsigned short int>(worldAabbMin,worldAabbMax,0xfffe,0xffff,maxHandles,pairCache,disableRaycastAccelerator)
+btAxisSweep3::btAxisSweep3(const btVector3& worldAabbMin, const btVector3& worldAabbMax, unsigned short int maxHandles, btOverlappingPairCache* pairCache, bool disableRaycastAccelerator)
+	: btAxisSweep3Internal<unsigned short int>(worldAabbMin, worldAabbMax, 0xfffe, 0xffff, maxHandles, pairCache, disableRaycastAccelerator)
 {
 	// 1 handle is reserved as sentinel
 	btAssert(maxHandles > 1 && maxHandles < 32767);
-
 }
 
-
-bt32BitAxisSweep3::bt32BitAxisSweep3(const btVector3& worldAabbMin,const btVector3& worldAabbMax, unsigned int maxHandles , btOverlappingPairCache* pairCache , bool disableRaycastAccelerator)
-:btAxisSweep3Internal<unsigned int>(worldAabbMin,worldAabbMax,0xfffffffe,0x7fffffff,maxHandles,pairCache,disableRaycastAccelerator)
+bt32BitAxisSweep3::bt32BitAxisSweep3(const btVector3& worldAabbMin, const btVector3& worldAabbMax, unsigned int maxHandles, btOverlappingPairCache* pairCache, bool disableRaycastAccelerator)
+	: btAxisSweep3Internal<unsigned int>(worldAabbMin, worldAabbMax, 0xfffffffe, 0x7fffffff, maxHandles, pairCache, disableRaycastAccelerator)
 {
 	// 1 handle is reserved as sentinel
 	btAssert(maxHandles > 1 && maxHandles < 2147483647);
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btAxisSweep3.h b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btAxisSweep3.h
index a3648df1afc..1e42f25f3bd 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btAxisSweep3.h
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btAxisSweep3.h
@@ -33,9 +33,7 @@
 class btAxisSweep3 : public btAxisSweep3Internal<unsigned short int>
 {
 public:
-
-	btAxisSweep3(const btVector3& worldAabbMin,const btVector3& worldAabbMax, unsigned short int maxHandles = 16384, btOverlappingPairCache* pairCache = 0, bool disableRaycastAccelerator = false);
-
+	btAxisSweep3(const btVector3& worldAabbMin, const btVector3& worldAabbMax, unsigned short int maxHandles = 16384, btOverlappingPairCache* pairCache = 0, bool disableRaycastAccelerator = false);
 };
 
 /// The bt32BitAxisSweep3 allows higher precision quantization and more objects compared to the btAxisSweep3 sweep and prune.
@@ -44,9 +42,7 @@ public:
 class bt32BitAxisSweep3 : public btAxisSweep3Internal<unsigned int>
 {
 public:
-
-	bt32BitAxisSweep3(const btVector3& worldAabbMin,const btVector3& worldAabbMax, unsigned int maxHandles = 1500000, btOverlappingPairCache* pairCache = 0, bool disableRaycastAccelerator = false);
-
+	bt32BitAxisSweep3(const btVector3& worldAabbMin, const btVector3& worldAabbMax, unsigned int maxHandles = 1500000, btOverlappingPairCache* pairCache = 0, bool disableRaycastAccelerator = false);
 };
 
 #endif
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btAxisSweep3Internal.h b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btAxisSweep3Internal.h
index 323aa96dcac..2ee35528fd9 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btAxisSweep3Internal.h
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btAxisSweep3Internal.h
@@ -36,172 +36,158 @@ template <typename BP_FP_INT_TYPE>
 class btAxisSweep3Internal : public btBroadphaseInterface
 {
 protected:
-
-	BP_FP_INT_TYPE	m_bpHandleMask;
-	BP_FP_INT_TYPE	m_handleSentinel;
+	BP_FP_INT_TYPE m_bpHandleMask;
+	BP_FP_INT_TYPE m_handleSentinel;
 
 public:
-	
- BT_DECLARE_ALIGNED_ALLOCATOR();
+	BT_DECLARE_ALIGNED_ALLOCATOR();
 
 	class Edge
 	{
 	public:
-		BP_FP_INT_TYPE m_pos;			// low bit is min/max
+		BP_FP_INT_TYPE m_pos;  // low bit is min/max
 		BP_FP_INT_TYPE m_handle;
 
-		BP_FP_INT_TYPE IsMax() const {return static_cast<BP_FP_INT_TYPE>(m_pos & 1);}
+		BP_FP_INT_TYPE IsMax() const { return static_cast<BP_FP_INT_TYPE>(m_pos & 1); }
 	};
 
 public:
-	class	Handle : public btBroadphaseProxy
+	class Handle : public btBroadphaseProxy
 	{
 	public:
-	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
+		BT_DECLARE_ALIGNED_ALLOCATOR();
+
 		// indexes into the edge arrays
-		BP_FP_INT_TYPE m_minEdges[3], m_maxEdges[3];		// 6 * 2 = 12
-//		BP_FP_INT_TYPE m_uniqueId;
-		btBroadphaseProxy*	m_dbvtProxy;//for faster raycast
+		BP_FP_INT_TYPE m_minEdges[3], m_maxEdges[3];  // 6 * 2 = 12
+													  //		BP_FP_INT_TYPE m_uniqueId;
+		btBroadphaseProxy* m_dbvtProxy;               //for faster raycast
 		//void* m_pOwner; this is now in btBroadphaseProxy.m_clientObject
-	
-		SIMD_FORCE_INLINE void SetNextFree(BP_FP_INT_TYPE next) {m_minEdges[0] = next;}
-		SIMD_FORCE_INLINE BP_FP_INT_TYPE GetNextFree() const {return m_minEdges[0];}
-	};		// 24 bytes + 24 for Edge structures = 44 bytes total per entry
 
-	
+		SIMD_FORCE_INLINE void SetNextFree(BP_FP_INT_TYPE next) { m_minEdges[0] = next; }
+		SIMD_FORCE_INLINE BP_FP_INT_TYPE GetNextFree() const { return m_minEdges[0]; }
+	};  // 24 bytes + 24 for Edge structures = 44 bytes total per entry
+
 protected:
-	btVector3 m_worldAabbMin;						// overall system bounds
-	btVector3 m_worldAabbMax;						// overall system bounds
+	btVector3 m_worldAabbMin;  // overall system bounds
+	btVector3 m_worldAabbMax;  // overall system bounds
 
-	btVector3 m_quantize;						// scaling factor for quantization
+	btVector3 m_quantize;  // scaling factor for quantization
 
-	BP_FP_INT_TYPE m_numHandles;						// number of active handles
-	BP_FP_INT_TYPE m_maxHandles;						// max number of handles
-	Handle* m_pHandles;						// handles pool
-	
-	BP_FP_INT_TYPE m_firstFreeHandle;		// free handles list
+	BP_FP_INT_TYPE m_numHandles;  // number of active handles
+	BP_FP_INT_TYPE m_maxHandles;  // max number of handles
+	Handle* m_pHandles;           // handles pool
 
-	Edge* m_pEdges[3];						// edge arrays for the 3 axes (each array has m_maxHandles * 2 + 2 sentinel entries)
+	BP_FP_INT_TYPE m_firstFreeHandle;  // free handles list
+
+	Edge* m_pEdges[3];  // edge arrays for the 3 axes (each array has m_maxHandles * 2 + 2 sentinel entries)
 	void* m_pEdgesRawPtr[3];
 
 	btOverlappingPairCache* m_pairCache;
 
 	///btOverlappingPairCallback is an additional optional user callback for adding/removing overlapping pairs, similar interface to btOverlappingPairCache.
 	btOverlappingPairCallback* m_userPairCallback;
-	
-	bool	m_ownsPairCache;
 
-	int	m_invalidPair;
+	bool m_ownsPairCache;
+
+	int m_invalidPair;
 
 	///additional dynamic aabb structure, used to accelerate ray cast queries.
 	///can be disabled using a optional argument in the constructor
-	btDbvtBroadphase*	m_raycastAccelerator;
-	btOverlappingPairCache*	m_nullPairCache;
-
+	btDbvtBroadphase* m_raycastAccelerator;
+	btOverlappingPairCache* m_nullPairCache;
 
 	// allocation/deallocation
 	BP_FP_INT_TYPE allocHandle();
 	void freeHandle(BP_FP_INT_TYPE handle);
-	
 
-	bool testOverlap2D(const Handle* pHandleA, const Handle* pHandleB,int axis0,int axis1);
+	bool testOverlap2D(const Handle* pHandleA, const Handle* pHandleB, int axis0, int axis1);
 
 #ifdef DEBUG_BROADPHASE
-	void debugPrintAxis(int axis,bool checkCardinality=true);
-#endif //DEBUG_BROADPHASE
+	void debugPrintAxis(int axis, bool checkCardinality = true);
+#endif  //DEBUG_BROADPHASE
 
 	//Overlap* AddOverlap(BP_FP_INT_TYPE handleA, BP_FP_INT_TYPE handleB);
 	//void RemoveOverlap(BP_FP_INT_TYPE handleA, BP_FP_INT_TYPE handleB);
 
-	
-
-	void sortMinDown(int axis, BP_FP_INT_TYPE edge, btDispatcher* dispatcher, bool updateOverlaps );
-	void sortMinUp(int axis, BP_FP_INT_TYPE edge, btDispatcher* dispatcher, bool updateOverlaps );
-	void sortMaxDown(int axis, BP_FP_INT_TYPE edge, btDispatcher* dispatcher, bool updateOverlaps );
-	void sortMaxUp(int axis, BP_FP_INT_TYPE edge, btDispatcher* dispatcher, bool updateOverlaps );
+	void sortMinDown(int axis, BP_FP_INT_TYPE edge, btDispatcher* dispatcher, bool updateOverlaps);
+	void sortMinUp(int axis, BP_FP_INT_TYPE edge, btDispatcher* dispatcher, bool updateOverlaps);
+	void sortMaxDown(int axis, BP_FP_INT_TYPE edge, btDispatcher* dispatcher, bool updateOverlaps);
+	void sortMaxUp(int axis, BP_FP_INT_TYPE edge, btDispatcher* dispatcher, bool updateOverlaps);
 
 public:
+	btAxisSweep3Internal(const btVector3& worldAabbMin, const btVector3& worldAabbMax, BP_FP_INT_TYPE handleMask, BP_FP_INT_TYPE handleSentinel, BP_FP_INT_TYPE maxHandles = 16384, btOverlappingPairCache* pairCache = 0, bool disableRaycastAccelerator = false);
 
-	btAxisSweep3Internal(const btVector3& worldAabbMin,const btVector3& worldAabbMax, BP_FP_INT_TYPE handleMask, BP_FP_INT_TYPE handleSentinel, BP_FP_INT_TYPE maxHandles = 16384, btOverlappingPairCache* pairCache=0,bool disableRaycastAccelerator = false);
-
-	virtual	~btAxisSweep3Internal();
+	virtual ~btAxisSweep3Internal();
 
 	BP_FP_INT_TYPE getNumHandles() const
 	{
 		return m_numHandles;
 	}
 
-	virtual void	calculateOverlappingPairs(btDispatcher* dispatcher);
-	
-	BP_FP_INT_TYPE addHandle(const btVector3& aabbMin,const btVector3& aabbMax, void* pOwner, int collisionFilterGroup, int collisionFilterMask,btDispatcher* dispatcher);
-	void removeHandle(BP_FP_INT_TYPE handle,btDispatcher* dispatcher);
-	void updateHandle(BP_FP_INT_TYPE handle, const btVector3& aabbMin,const btVector3& aabbMax,btDispatcher* dispatcher);
-	SIMD_FORCE_INLINE Handle* getHandle(BP_FP_INT_TYPE index) const {return m_pHandles + index;}
+	virtual void calculateOverlappingPairs(btDispatcher* dispatcher);
+
+	BP_FP_INT_TYPE addHandle(const btVector3& aabbMin, const btVector3& aabbMax, void* pOwner, int collisionFilterGroup, int collisionFilterMask, btDispatcher* dispatcher);
+	void removeHandle(BP_FP_INT_TYPE handle, btDispatcher* dispatcher);
+	void updateHandle(BP_FP_INT_TYPE handle, const btVector3& aabbMin, const btVector3& aabbMax, btDispatcher* dispatcher);
+	SIMD_FORCE_INLINE Handle* getHandle(BP_FP_INT_TYPE index) const { return m_pHandles + index; }
 
 	virtual void resetPool(btDispatcher* dispatcher);
 
-	void	processAllOverlappingPairs(btOverlapCallback* callback);
+	void processAllOverlappingPairs(btOverlapCallback* callback);
 
 	//Broadphase Interface
-	virtual btBroadphaseProxy*	createProxy(  const btVector3& aabbMin,  const btVector3& aabbMax,int shapeType,void* userPtr , int collisionFilterGroup, int collisionFilterMask,btDispatcher* dispatcher);
-	virtual void	destroyProxy(btBroadphaseProxy* proxy,btDispatcher* dispatcher);
-	virtual void	setAabb(btBroadphaseProxy* proxy,const btVector3& aabbMin,const btVector3& aabbMax,btDispatcher* dispatcher);
-	virtual void  getAabb(btBroadphaseProxy* proxy,btVector3& aabbMin, btVector3& aabbMax ) const;
-	
-	virtual void	rayTest(const btVector3& rayFrom,const btVector3& rayTo, btBroadphaseRayCallback& rayCallback, const btVector3& aabbMin=btVector3(0,0,0), const btVector3& aabbMax = btVector3(0,0,0));
-	virtual void	aabbTest(const btVector3& aabbMin, const btVector3& aabbMax, btBroadphaseAabbCallback& callback);
+	virtual btBroadphaseProxy* createProxy(const btVector3& aabbMin, const btVector3& aabbMax, int shapeType, void* userPtr, int collisionFilterGroup, int collisionFilterMask, btDispatcher* dispatcher);
+	virtual void destroyProxy(btBroadphaseProxy* proxy, btDispatcher* dispatcher);
+	virtual void setAabb(btBroadphaseProxy* proxy, const btVector3& aabbMin, const btVector3& aabbMax, btDispatcher* dispatcher);
+	virtual void getAabb(btBroadphaseProxy* proxy, btVector3& aabbMin, btVector3& aabbMax) const;
+
+	virtual void rayTest(const btVector3& rayFrom, const btVector3& rayTo, btBroadphaseRayCallback& rayCallback, const btVector3& aabbMin = btVector3(0, 0, 0), const btVector3& aabbMax = btVector3(0, 0, 0));
+	virtual void aabbTest(const btVector3& aabbMin, const btVector3& aabbMax, btBroadphaseAabbCallback& callback);
 
-	
 	void quantize(BP_FP_INT_TYPE* out, const btVector3& point, int isMax) const;
 	///unQuantize should be conservative: aabbMin/aabbMax should be larger then 'getAabb' result
-	void unQuantize(btBroadphaseProxy* proxy,btVector3& aabbMin, btVector3& aabbMax ) const;
-	
-	bool	testAabbOverlap(btBroadphaseProxy* proxy0,btBroadphaseProxy* proxy1);
+	void unQuantize(btBroadphaseProxy* proxy, btVector3& aabbMin, btVector3& aabbMax) const;
 
-	btOverlappingPairCache*	getOverlappingPairCache()
+	bool testAabbOverlap(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1);
+
+	btOverlappingPairCache* getOverlappingPairCache()
 	{
 		return m_pairCache;
 	}
-	const btOverlappingPairCache*	getOverlappingPairCache() const
+	const btOverlappingPairCache* getOverlappingPairCache() const
 	{
 		return m_pairCache;
 	}
 
-	void	setOverlappingPairUserCallback(btOverlappingPairCallback* pairCallback)
+	void setOverlappingPairUserCallback(btOverlappingPairCallback* pairCallback)
 	{
 		m_userPairCallback = pairCallback;
 	}
-	const btOverlappingPairCallback*	getOverlappingPairUserCallback() const
+	const btOverlappingPairCallback* getOverlappingPairUserCallback() const
 	{
 		return m_userPairCallback;
 	}
 
 	///getAabb returns the axis aligned bounding box in the 'global' coordinate frame
 	///will add some transform later
-	virtual void getBroadphaseAabb(btVector3& aabbMin,btVector3& aabbMax) const
+	virtual void getBroadphaseAabb(btVector3& aabbMin, btVector3& aabbMax) const
 	{
 		aabbMin = m_worldAabbMin;
 		aabbMax = m_worldAabbMax;
 	}
 
-	virtual void	printStats()
+	virtual void printStats()
 	{
-/*		printf("btAxisSweep3.h\n");
+		/*		printf("btAxisSweep3.h\n");
 		printf("numHandles = %d, maxHandles = %d\n",m_numHandles,m_maxHandles);
 		printf("aabbMin=%f,%f,%f,aabbMax=%f,%f,%f\n",m_worldAabbMin.getX(),m_worldAabbMin.getY(),m_worldAabbMin.getZ(),
 			m_worldAabbMax.getX(),m_worldAabbMax.getY(),m_worldAabbMax.getZ());
 			*/
-
 	}
-
 };
 
 ////////////////////////////////////////////////////////////////////
 
-
-
-
 #ifdef DEBUG_BROADPHASE
 #include <stdio.h>
 
@@ -209,75 +195,73 @@ template <typename BP_FP_INT_TYPE>
 void btAxisSweep3<BP_FP_INT_TYPE>::debugPrintAxis(int axis, bool checkCardinality)
 {
 	int numEdges = m_pHandles[0].m_maxEdges[axis];
-	printf("SAP Axis %d, numEdges=%d\n",axis,numEdges);
+	printf("SAP Axis %d, numEdges=%d\n", axis, numEdges);
 
 	int i;
-	for (i=0;i<numEdges+1;i++)
+	for (i = 0; i < numEdges + 1; i++)
 	{
 		Edge* pEdge = m_pEdges[axis] + i;
 		Handle* pHandlePrev = getHandle(pEdge->m_handle);
-		int handleIndex = pEdge->IsMax()? pHandlePrev->m_maxEdges[axis] : pHandlePrev->m_minEdges[axis];
+		int handleIndex = pEdge->IsMax() ? pHandlePrev->m_maxEdges[axis] : pHandlePrev->m_minEdges[axis];
 		char beginOrEnd;
-		beginOrEnd=pEdge->IsMax()?'E':'B';
-		printf("	[%c,h=%d,p=%x,i=%d]\n",beginOrEnd,pEdge->m_handle,pEdge->m_pos,handleIndex);
+		beginOrEnd = pEdge->IsMax() ? 'E' : 'B';
+		printf("	[%c,h=%d,p=%x,i=%d]\n", beginOrEnd, pEdge->m_handle, pEdge->m_pos, handleIndex);
 	}
 
 	if (checkCardinality)
-		btAssert(numEdges == m_numHandles*2+1);
+		btAssert(numEdges == m_numHandles * 2 + 1);
 }
-#endif //DEBUG_BROADPHASE
+#endif  //DEBUG_BROADPHASE
 
 template <typename BP_FP_INT_TYPE>
-btBroadphaseProxy*	btAxisSweep3Internal<BP_FP_INT_TYPE>::createProxy(  const btVector3& aabbMin,  const btVector3& aabbMax,int shapeType,void* userPtr, int collisionFilterGroup, int collisionFilterMask,btDispatcher* dispatcher)
+btBroadphaseProxy* btAxisSweep3Internal<BP_FP_INT_TYPE>::createProxy(const btVector3& aabbMin, const btVector3& aabbMax, int shapeType, void* userPtr, int collisionFilterGroup, int collisionFilterMask, btDispatcher* dispatcher)
 {
-		(void)shapeType;
-		BP_FP_INT_TYPE handleId = addHandle(aabbMin,aabbMax, userPtr,collisionFilterGroup,collisionFilterMask,dispatcher);
-		
-		Handle* handle = getHandle(handleId);
-		
-		if (m_raycastAccelerator)
-		{
-			btBroadphaseProxy* rayProxy = m_raycastAccelerator->createProxy(aabbMin,aabbMax,shapeType,userPtr,collisionFilterGroup,collisionFilterMask,dispatcher);
-			handle->m_dbvtProxy = rayProxy;
-		}
-		return handle;
+	(void)shapeType;
+	BP_FP_INT_TYPE handleId = addHandle(aabbMin, aabbMax, userPtr, collisionFilterGroup, collisionFilterMask, dispatcher);
+
+	Handle* handle = getHandle(handleId);
+
+	if (m_raycastAccelerator)
+	{
+		btBroadphaseProxy* rayProxy = m_raycastAccelerator->createProxy(aabbMin, aabbMax, shapeType, userPtr, collisionFilterGroup, collisionFilterMask, dispatcher);
+		handle->m_dbvtProxy = rayProxy;
+	}
+	return handle;
 }
 
-
-
 template <typename BP_FP_INT_TYPE>
-void	btAxisSweep3Internal<BP_FP_INT_TYPE>::destroyProxy(btBroadphaseProxy* proxy,btDispatcher* dispatcher)
+void btAxisSweep3Internal<BP_FP_INT_TYPE>::destroyProxy(btBroadphaseProxy* proxy, btDispatcher* dispatcher)
 {
 	Handle* handle = static_cast<Handle*>(proxy);
 	if (m_raycastAccelerator)
-		m_raycastAccelerator->destroyProxy(handle->m_dbvtProxy,dispatcher);
+		m_raycastAccelerator->destroyProxy(handle->m_dbvtProxy, dispatcher);
 	removeHandle(static_cast<BP_FP_INT_TYPE>(handle->m_uniqueId), dispatcher);
 }
 
 template <typename BP_FP_INT_TYPE>
-void	btAxisSweep3Internal<BP_FP_INT_TYPE>::setAabb(btBroadphaseProxy* proxy,const btVector3& aabbMin,const btVector3& aabbMax,btDispatcher* dispatcher)
+void btAxisSweep3Internal<BP_FP_INT_TYPE>::setAabb(btBroadphaseProxy* proxy, const btVector3& aabbMin, const btVector3& aabbMax, btDispatcher* dispatcher)
 {
 	Handle* handle = static_cast<Handle*>(proxy);
 	handle->m_aabbMin = aabbMin;
 	handle->m_aabbMax = aabbMax;
-	updateHandle(static_cast<BP_FP_INT_TYPE>(handle->m_uniqueId), aabbMin, aabbMax,dispatcher);
+	updateHandle(static_cast<BP_FP_INT_TYPE>(handle->m_uniqueId), aabbMin, aabbMax, dispatcher);
 	if (m_raycastAccelerator)
-		m_raycastAccelerator->setAabb(handle->m_dbvtProxy,aabbMin,aabbMax,dispatcher);
-
+		m_raycastAccelerator->setAabb(handle->m_dbvtProxy, aabbMin, aabbMax, dispatcher);
 }
 
 template <typename BP_FP_INT_TYPE>
-void	btAxisSweep3Internal<BP_FP_INT_TYPE>::rayTest(const btVector3& rayFrom,const btVector3& rayTo, btBroadphaseRayCallback& rayCallback,const btVector3& aabbMin,const btVector3& aabbMax)
+void btAxisSweep3Internal<BP_FP_INT_TYPE>::rayTest(const btVector3& rayFrom, const btVector3& rayTo, btBroadphaseRayCallback& rayCallback, const btVector3& aabbMin, const btVector3& aabbMax)
 {
 	if (m_raycastAccelerator)
 	{
-		m_raycastAccelerator->rayTest(rayFrom,rayTo,rayCallback,aabbMin,aabbMax);
-	} else
+		m_raycastAccelerator->rayTest(rayFrom, rayTo, rayCallback, aabbMin, aabbMax);
+	}
+	else
 	{
 		//choose axis?
 		BP_FP_INT_TYPE axis = 0;
 		//for each proxy
-		for (BP_FP_INT_TYPE i=1;i<m_numHandles*2+1;i++)
+		for (BP_FP_INT_TYPE i = 1; i < m_numHandles * 2 + 1; i++)
 		{
 			if (m_pEdges[axis][i].IsMax())
 			{
@@ -288,22 +272,23 @@ void	btAxisSweep3Internal<BP_FP_INT_TYPE>::rayTest(const btVector3& rayFrom,cons
 }
 
 template <typename BP_FP_INT_TYPE>
-void	btAxisSweep3Internal<BP_FP_INT_TYPE>::aabbTest(const btVector3& aabbMin, const btVector3& aabbMax, btBroadphaseAabbCallback& callback)
+void btAxisSweep3Internal<BP_FP_INT_TYPE>::aabbTest(const btVector3& aabbMin, const btVector3& aabbMax, btBroadphaseAabbCallback& callback)
 {
 	if (m_raycastAccelerator)
 	{
-		m_raycastAccelerator->aabbTest(aabbMin,aabbMax,callback);
-	} else
+		m_raycastAccelerator->aabbTest(aabbMin, aabbMax, callback);
+	}
+	else
 	{
 		//choose axis?
 		BP_FP_INT_TYPE axis = 0;
 		//for each proxy
-		for (BP_FP_INT_TYPE i=1;i<m_numHandles*2+1;i++)
+		for (BP_FP_INT_TYPE i = 1; i < m_numHandles * 2 + 1; i++)
 		{
 			if (m_pEdges[axis][i].IsMax())
 			{
 				Handle* handle = getHandle(m_pEdges[axis][i].m_handle);
-				if (TestAabbAgainstAabb2(aabbMin,aabbMax,handle->m_aabbMin,handle->m_aabbMax))
+				if (TestAabbAgainstAabb2(aabbMin, aabbMax, handle->m_aabbMin, handle->m_aabbMax))
 				{
 					callback.process(handle);
 				}
@@ -312,66 +297,60 @@ void	btAxisSweep3Internal<BP_FP_INT_TYPE>::aabbTest(const btVector3& aabbMin, co
 	}
 }
 
-
-
 template <typename BP_FP_INT_TYPE>
-void btAxisSweep3Internal<BP_FP_INT_TYPE>::getAabb(btBroadphaseProxy* proxy,btVector3& aabbMin, btVector3& aabbMax ) const
+void btAxisSweep3Internal<BP_FP_INT_TYPE>::getAabb(btBroadphaseProxy* proxy, btVector3& aabbMin, btVector3& aabbMax) const
 {
 	Handle* pHandle = static_cast<Handle*>(proxy);
 	aabbMin = pHandle->m_aabbMin;
 	aabbMax = pHandle->m_aabbMax;
 }
 
-
 template <typename BP_FP_INT_TYPE>
-void btAxisSweep3Internal<BP_FP_INT_TYPE>::unQuantize(btBroadphaseProxy* proxy,btVector3& aabbMin, btVector3& aabbMax ) const
+void btAxisSweep3Internal<BP_FP_INT_TYPE>::unQuantize(btBroadphaseProxy* proxy, btVector3& aabbMin, btVector3& aabbMax) const
 {
 	Handle* pHandle = static_cast<Handle*>(proxy);
 
 	unsigned short vecInMin[3];
 	unsigned short vecInMax[3];
 
-	vecInMin[0] = m_pEdges[0][pHandle->m_minEdges[0]].m_pos ;
-	vecInMax[0] = m_pEdges[0][pHandle->m_maxEdges[0]].m_pos +1 ;
-	vecInMin[1] = m_pEdges[1][pHandle->m_minEdges[1]].m_pos ;
-	vecInMax[1] = m_pEdges[1][pHandle->m_maxEdges[1]].m_pos +1 ;
-	vecInMin[2] = m_pEdges[2][pHandle->m_minEdges[2]].m_pos ;
-	vecInMax[2] = m_pEdges[2][pHandle->m_maxEdges[2]].m_pos +1 ;
-	
-	aabbMin.setValue((btScalar)(vecInMin[0]) / (m_quantize.getX()),(btScalar)(vecInMin[1]) / (m_quantize.getY()),(btScalar)(vecInMin[2]) / (m_quantize.getZ()));
+	vecInMin[0] = m_pEdges[0][pHandle->m_minEdges[0]].m_pos;
+	vecInMax[0] = m_pEdges[0][pHandle->m_maxEdges[0]].m_pos + 1;
+	vecInMin[1] = m_pEdges[1][pHandle->m_minEdges[1]].m_pos;
+	vecInMax[1] = m_pEdges[1][pHandle->m_maxEdges[1]].m_pos + 1;
+	vecInMin[2] = m_pEdges[2][pHandle->m_minEdges[2]].m_pos;
+	vecInMax[2] = m_pEdges[2][pHandle->m_maxEdges[2]].m_pos + 1;
+
+	aabbMin.setValue((btScalar)(vecInMin[0]) / (m_quantize.getX()), (btScalar)(vecInMin[1]) / (m_quantize.getY()), (btScalar)(vecInMin[2]) / (m_quantize.getZ()));
 	aabbMin += m_worldAabbMin;
-	
-	aabbMax.setValue((btScalar)(vecInMax[0]) / (m_quantize.getX()),(btScalar)(vecInMax[1]) / (m_quantize.getY()),(btScalar)(vecInMax[2]) / (m_quantize.getZ()));
+
+	aabbMax.setValue((btScalar)(vecInMax[0]) / (m_quantize.getX()), (btScalar)(vecInMax[1]) / (m_quantize.getY()), (btScalar)(vecInMax[2]) / (m_quantize.getZ()));
 	aabbMax += m_worldAabbMin;
 }
 
-
-
-
 template <typename BP_FP_INT_TYPE>
-btAxisSweep3Internal<BP_FP_INT_TYPE>::btAxisSweep3Internal(const btVector3& worldAabbMin,const btVector3& worldAabbMax, BP_FP_INT_TYPE handleMask, BP_FP_INT_TYPE handleSentinel,BP_FP_INT_TYPE userMaxHandles, btOverlappingPairCache* pairCache , bool disableRaycastAccelerator)
-:m_bpHandleMask(handleMask),
-m_handleSentinel(handleSentinel),
-m_pairCache(pairCache),
-m_userPairCallback(0),
-m_ownsPairCache(false),
-m_invalidPair(0),
-m_raycastAccelerator(0)
+btAxisSweep3Internal<BP_FP_INT_TYPE>::btAxisSweep3Internal(const btVector3& worldAabbMin, const btVector3& worldAabbMax, BP_FP_INT_TYPE handleMask, BP_FP_INT_TYPE handleSentinel, BP_FP_INT_TYPE userMaxHandles, btOverlappingPairCache* pairCache, bool disableRaycastAccelerator)
+	: m_bpHandleMask(handleMask),
+	  m_handleSentinel(handleSentinel),
+	  m_pairCache(pairCache),
+	  m_userPairCallback(0),
+	  m_ownsPairCache(false),
+	  m_invalidPair(0),
+	  m_raycastAccelerator(0)
 {
-	BP_FP_INT_TYPE maxHandles = static_cast<BP_FP_INT_TYPE>(userMaxHandles+1);//need to add one sentinel handle
+	BP_FP_INT_TYPE maxHandles = static_cast<BP_FP_INT_TYPE>(userMaxHandles + 1);  //need to add one sentinel handle
 
 	if (!m_pairCache)
 	{
-		void* ptr = btAlignedAlloc(sizeof(btHashedOverlappingPairCache),16);
-		m_pairCache = new(ptr) btHashedOverlappingPairCache();
+		void* ptr = btAlignedAlloc(sizeof(btHashedOverlappingPairCache), 16);
+		m_pairCache = new (ptr) btHashedOverlappingPairCache();
 		m_ownsPairCache = true;
 	}
 
 	if (!disableRaycastAccelerator)
 	{
-		m_nullPairCache = new (btAlignedAlloc(sizeof(btNullPairCache),16)) btNullPairCache();
-		m_raycastAccelerator = new (btAlignedAlloc(sizeof(btDbvtBroadphase),16)) btDbvtBroadphase(m_nullPairCache);//m_pairCache);
-		m_raycastAccelerator->m_deferedcollide = true;//don't add/remove pairs
+		m_nullPairCache = new (btAlignedAlloc(sizeof(btNullPairCache), 16)) btNullPairCache();
+		m_raycastAccelerator = new (btAlignedAlloc(sizeof(btDbvtBroadphase), 16)) btDbvtBroadphase(m_nullPairCache);  //m_pairCache);
+		m_raycastAccelerator->m_deferedcollide = true;                                                                //don't add/remove pairs
 	}
 
 	//btAssert(bounds.HasVolume());
@@ -382,13 +361,13 @@ m_raycastAccelerator(0)
 
 	btVector3 aabbSize = m_worldAabbMax - m_worldAabbMin;
 
-	BP_FP_INT_TYPE	maxInt = m_handleSentinel;
+	BP_FP_INT_TYPE maxInt = m_handleSentinel;
 
-	m_quantize = btVector3(btScalar(maxInt),btScalar(maxInt),btScalar(maxInt)) / aabbSize;
+	m_quantize = btVector3(btScalar(maxInt), btScalar(maxInt), btScalar(maxInt)) / aabbSize;
 
 	// allocate handles buffer, using btAlignedAlloc, and put all handles on free list
 	m_pHandles = new Handle[maxHandles];
-	
+
 	m_maxHandles = maxHandles;
 	m_numHandles = 0;
 
@@ -404,14 +383,14 @@ m_raycastAccelerator(0)
 		// allocate edge buffers
 		for (int i = 0; i < 3; i++)
 		{
-			m_pEdgesRawPtr[i] = btAlignedAlloc(sizeof(Edge)*maxHandles*2,16);
-			m_pEdges[i] = new(m_pEdgesRawPtr[i]) Edge[maxHandles * 2];
+			m_pEdgesRawPtr[i] = btAlignedAlloc(sizeof(Edge) * maxHandles * 2, 16);
+			m_pEdges[i] = new (m_pEdgesRawPtr[i]) Edge[maxHandles * 2];
 		}
 	}
 	//removed overlap management
 
 	// make boundary sentinels
-	
+
 	m_pHandles[0].m_clientObject = 0;
 
 	for (int axis = 0; axis < 3; axis++)
@@ -425,10 +404,8 @@ m_raycastAccelerator(0)
 		m_pEdges[axis][1].m_handle = 0;
 #ifdef DEBUG_BROADPHASE
 		debugPrintAxis(axis);
-#endif //DEBUG_BROADPHASE
-
+#endif  //DEBUG_BROADPHASE
 	}
-
 }
 
 template <typename BP_FP_INT_TYPE>
@@ -439,14 +416,14 @@ btAxisSweep3Internal<BP_FP_INT_TYPE>::~btAxisSweep3Internal()
 		m_nullPairCache->~btOverlappingPairCache();
 		btAlignedFree(m_nullPairCache);
 		m_raycastAccelerator->~btDbvtBroadphase();
-		btAlignedFree (m_raycastAccelerator);
+		btAlignedFree(m_raycastAccelerator);
 	}
 
 	for (int i = 2; i >= 0; i--)
 	{
 		btAlignedFree(m_pEdgesRawPtr[i]);
 	}
-	delete [] m_pHandles;
+	delete[] m_pHandles;
 
 	if (m_ownsPairCache)
 	{
@@ -470,13 +447,12 @@ void btAxisSweep3Internal<BP_FP_INT_TYPE>::quantize(BP_FP_INT_TYPE* out, const b
 	out[2] = (BP_FP_INT_TYPE)(((BP_FP_INT_TYPE)v.getZ() & m_bpHandleMask) | isMax);
 #else
 	btVector3 v = (point - m_worldAabbMin) * m_quantize;
-	out[0]=(v[0]<=0)?(BP_FP_INT_TYPE)isMax:(v[0]>=m_handleSentinel)?(BP_FP_INT_TYPE)((m_handleSentinel&m_bpHandleMask)|isMax):(BP_FP_INT_TYPE)(((BP_FP_INT_TYPE)v[0]&m_bpHandleMask)|isMax);
-	out[1]=(v[1]<=0)?(BP_FP_INT_TYPE)isMax:(v[1]>=m_handleSentinel)?(BP_FP_INT_TYPE)((m_handleSentinel&m_bpHandleMask)|isMax):(BP_FP_INT_TYPE)(((BP_FP_INT_TYPE)v[1]&m_bpHandleMask)|isMax);
-	out[2]=(v[2]<=0)?(BP_FP_INT_TYPE)isMax:(v[2]>=m_handleSentinel)?(BP_FP_INT_TYPE)((m_handleSentinel&m_bpHandleMask)|isMax):(BP_FP_INT_TYPE)(((BP_FP_INT_TYPE)v[2]&m_bpHandleMask)|isMax);
-#endif //OLD_CLAMPING_METHOD
+	out[0] = (v[0] <= 0) ? (BP_FP_INT_TYPE)isMax : (v[0] >= m_handleSentinel) ? (BP_FP_INT_TYPE)((m_handleSentinel & m_bpHandleMask) | isMax) : (BP_FP_INT_TYPE)(((BP_FP_INT_TYPE)v[0] & m_bpHandleMask) | isMax);
+	out[1] = (v[1] <= 0) ? (BP_FP_INT_TYPE)isMax : (v[1] >= m_handleSentinel) ? (BP_FP_INT_TYPE)((m_handleSentinel & m_bpHandleMask) | isMax) : (BP_FP_INT_TYPE)(((BP_FP_INT_TYPE)v[1] & m_bpHandleMask) | isMax);
+	out[2] = (v[2] <= 0) ? (BP_FP_INT_TYPE)isMax : (v[2] >= m_handleSentinel) ? (BP_FP_INT_TYPE)((m_handleSentinel & m_bpHandleMask) | isMax) : (BP_FP_INT_TYPE)(((BP_FP_INT_TYPE)v[2] & m_bpHandleMask) | isMax);
+#endif  //OLD_CLAMPING_METHOD
 }
 
-
 template <typename BP_FP_INT_TYPE>
 BP_FP_INT_TYPE btAxisSweep3Internal<BP_FP_INT_TYPE>::allocHandle()
 {
@@ -500,9 +476,8 @@ void btAxisSweep3Internal<BP_FP_INT_TYPE>::freeHandle(BP_FP_INT_TYPE handle)
 	m_numHandles--;
 }
 
-
 template <typename BP_FP_INT_TYPE>
-BP_FP_INT_TYPE btAxisSweep3Internal<BP_FP_INT_TYPE>::addHandle(const btVector3& aabbMin,const btVector3& aabbMax, void* pOwner, int collisionFilterGroup, int collisionFilterMask,btDispatcher* dispatcher)
+BP_FP_INT_TYPE btAxisSweep3Internal<BP_FP_INT_TYPE>::addHandle(const btVector3& aabbMin, const btVector3& aabbMax, void* pOwner, int collisionFilterGroup, int collisionFilterMask, btDispatcher* dispatcher)
 {
 	// quantize the bounds
 	BP_FP_INT_TYPE min[3], max[3];
@@ -511,10 +486,9 @@ BP_FP_INT_TYPE btAxisSweep3Internal<BP_FP_INT_TYPE>::addHandle(const btVector3&
 
 	// allocate a handle
 	BP_FP_INT_TYPE handle = allocHandle();
-	
 
 	Handle* pHandle = getHandle(handle);
-	
+
 	pHandle->m_uniqueId = static_cast<int>(handle);
 	//pHandle->m_pOverlaps = 0;
 	pHandle->m_clientObject = pOwner;
@@ -524,11 +498,9 @@ BP_FP_INT_TYPE btAxisSweep3Internal<BP_FP_INT_TYPE>::addHandle(const btVector3&
 	// compute current limit of edge arrays
 	BP_FP_INT_TYPE limit = static_cast<BP_FP_INT_TYPE>(m_numHandles * 2);
 
-	
 	// insert new edges just inside the max boundary edge
 	for (BP_FP_INT_TYPE axis = 0; axis < 3; axis++)
 	{
-
 		m_pHandles[0].m_maxEdges[axis] += 2;
 
 		m_pEdges[axis][limit + 1] = m_pEdges[axis][limit - 1];
@@ -544,22 +516,19 @@ BP_FP_INT_TYPE btAxisSweep3Internal<BP_FP_INT_TYPE>::addHandle(const btVector3&
 	}
 
 	// now sort the new edges to their correct position
-	sortMinDown(0, pHandle->m_minEdges[0], dispatcher,false);
-	sortMaxDown(0, pHandle->m_maxEdges[0], dispatcher,false);
-	sortMinDown(1, pHandle->m_minEdges[1], dispatcher,false);
-	sortMaxDown(1, pHandle->m_maxEdges[1], dispatcher,false);
-	sortMinDown(2, pHandle->m_minEdges[2], dispatcher,true);
-	sortMaxDown(2, pHandle->m_maxEdges[2], dispatcher,true);
-
+	sortMinDown(0, pHandle->m_minEdges[0], dispatcher, false);
+	sortMaxDown(0, pHandle->m_maxEdges[0], dispatcher, false);
+	sortMinDown(1, pHandle->m_minEdges[1], dispatcher, false);
+	sortMaxDown(1, pHandle->m_maxEdges[1], dispatcher, false);
+	sortMinDown(2, pHandle->m_minEdges[2], dispatcher, true);
+	sortMaxDown(2, pHandle->m_maxEdges[2], dispatcher, true);
 
 	return handle;
 }
 
-
 template <typename BP_FP_INT_TYPE>
-void btAxisSweep3Internal<BP_FP_INT_TYPE>::removeHandle(BP_FP_INT_TYPE handle,btDispatcher* dispatcher)
+void btAxisSweep3Internal<BP_FP_INT_TYPE>::removeHandle(BP_FP_INT_TYPE handle, btDispatcher* dispatcher)
 {
-
 	Handle* pHandle = getHandle(handle);
 
 	//explicitly remove the pairs containing the proxy
@@ -567,50 +536,43 @@ void btAxisSweep3Internal<BP_FP_INT_TYPE>::removeHandle(BP_FP_INT_TYPE handle,bt
 	///@todo: compare performance
 	if (!m_pairCache->hasDeferredRemoval())
 	{
-		m_pairCache->removeOverlappingPairsContainingProxy(pHandle,dispatcher);
+		m_pairCache->removeOverlappingPairsContainingProxy(pHandle, dispatcher);
 	}
 
 	// compute current limit of edge arrays
 	int limit = static_cast<int>(m_numHandles * 2);
-	
+
 	int axis;
 
-	for (axis = 0;axis<3;axis++)
+	for (axis = 0; axis < 3; axis++)
 	{
 		m_pHandles[0].m_maxEdges[axis] -= 2;
 	}
 
 	// remove the edges by sorting them up to the end of the list
-	for ( axis = 0; axis < 3; axis++)
+	for (axis = 0; axis < 3; axis++)
 	{
 		Edge* pEdges = m_pEdges[axis];
 		BP_FP_INT_TYPE max = pHandle->m_maxEdges[axis];
 		pEdges[max].m_pos = m_handleSentinel;
 
-		sortMaxUp(axis,max,dispatcher,false);
-
+		sortMaxUp(axis, max, dispatcher, false);
 
 		BP_FP_INT_TYPE i = pHandle->m_minEdges[axis];
 		pEdges[i].m_pos = m_handleSentinel;
 
+		sortMinUp(axis, i, dispatcher, false);
 
-		sortMinUp(axis,i,dispatcher,false);
+		pEdges[limit - 1].m_handle = 0;
+		pEdges[limit - 1].m_pos = m_handleSentinel;
 
-		pEdges[limit-1].m_handle = 0;
-		pEdges[limit-1].m_pos = m_handleSentinel;
-		
 #ifdef DEBUG_BROADPHASE
-			debugPrintAxis(axis,false);
-#endif //DEBUG_BROADPHASE
-
-
+		debugPrintAxis(axis, false);
+#endif  //DEBUG_BROADPHASE
 	}
 
-
 	// free the handle
 	freeHandle(handle);
-
-	
 }
 
 template <typename BP_FP_INT_TYPE>
@@ -625,19 +587,16 @@ void btAxisSweep3Internal<BP_FP_INT_TYPE>::resetPool(btDispatcher* /*dispatcher*
 			m_pHandles[m_maxHandles - 1].SetNextFree(0);
 		}
 	}
-}       
-
+}
 
 //#include <stdio.h>
 
 template <typename BP_FP_INT_TYPE>
-void	btAxisSweep3Internal<BP_FP_INT_TYPE>::calculateOverlappingPairs(btDispatcher* dispatcher)
+void btAxisSweep3Internal<BP_FP_INT_TYPE>::calculateOverlappingPairs(btDispatcher* dispatcher)
 {
-
 	if (m_pairCache->hasDeferredRemoval())
 	{
-	
-		btBroadphasePairArray&	overlappingPairArray = m_pairCache->getOverlappingPairArray();
+		btBroadphasePairArray& overlappingPairArray = m_pairCache->getOverlappingPairArray();
 
 		//perform a sort, to find duplicates and to sort 'invalid' pairs to the end
 		overlappingPairArray.quickSort(btBroadphasePairSortPredicate());
@@ -645,18 +604,15 @@ void	btAxisSweep3Internal<BP_FP_INT_TYPE>::calculateOverlappingPairs(btDispatche
 		overlappingPairArray.resize(overlappingPairArray.size() - m_invalidPair);
 		m_invalidPair = 0;
 
-		
 		int i;
 
 		btBroadphasePair previousPair;
 		previousPair.m_pProxy0 = 0;
 		previousPair.m_pProxy1 = 0;
 		previousPair.m_algorithm = 0;
-		
-		
-		for (i=0;i<overlappingPairArray.size();i++)
+
+		for (i = 0; i < overlappingPairArray.size(); i++)
 		{
-		
 			btBroadphasePair& pair = overlappingPairArray[i];
 
 			bool isDuplicate = (pair == previousPair);
@@ -668,91 +624,90 @@ void	btAxisSweep3Internal<BP_FP_INT_TYPE>::calculateOverlappingPairs(btDispatche
 			if (!isDuplicate)
 			{
 				///important to use an AABB test that is consistent with the broadphase
-				bool hasOverlap = testAabbOverlap(pair.m_pProxy0,pair.m_pProxy1);
+				bool hasOverlap = testAabbOverlap(pair.m_pProxy0, pair.m_pProxy1);
 
 				if (hasOverlap)
 				{
-					needsRemoval = false;//callback->processOverlap(pair);
-				} else
+					needsRemoval = false;  //callback->processOverlap(pair);
+				}
+				else
 				{
 					needsRemoval = true;
 				}
-			} else
+			}
+			else
 			{
 				//remove duplicate
 				needsRemoval = true;
 				//should have no algorithm
 				btAssert(!pair.m_algorithm);
 			}
-			
+
 			if (needsRemoval)
 			{
-				m_pairCache->cleanOverlappingPair(pair,dispatcher);
+				m_pairCache->cleanOverlappingPair(pair, dispatcher);
 
-		//		m_overlappingPairArray.swap(i,m_overlappingPairArray.size()-1);
-		//		m_overlappingPairArray.pop_back();
+				//		m_overlappingPairArray.swap(i,m_overlappingPairArray.size()-1);
+				//		m_overlappingPairArray.pop_back();
 				pair.m_pProxy0 = 0;
 				pair.m_pProxy1 = 0;
 				m_invalidPair++;
-      }
+			}
+		}
 
-    }
-
-	///if you don't like to skip the invalid pairs in the array, execute following code:
-	#define CLEAN_INVALID_PAIRS 1
-	#ifdef CLEAN_INVALID_PAIRS
+///if you don't like to skip the invalid pairs in the array, execute following code:
+#define CLEAN_INVALID_PAIRS 1
+#ifdef CLEAN_INVALID_PAIRS
 
 		//perform a sort, to sort 'invalid' pairs to the end
 		overlappingPairArray.quickSort(btBroadphasePairSortPredicate());
 
 		overlappingPairArray.resize(overlappingPairArray.size() - m_invalidPair);
 		m_invalidPair = 0;
-	#endif//CLEAN_INVALID_PAIRS
-		
+#endif  //CLEAN_INVALID_PAIRS
+
 		//printf("overlappingPairArray.size()=%d\n",overlappingPairArray.size());
 	}
-
 }
 
-
 template <typename BP_FP_INT_TYPE>
-bool btAxisSweep3Internal<BP_FP_INT_TYPE>::testAabbOverlap(btBroadphaseProxy* proxy0,btBroadphaseProxy* proxy1)
+bool btAxisSweep3Internal<BP_FP_INT_TYPE>::testAabbOverlap(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1)
 {
 	const Handle* pHandleA = static_cast<Handle*>(proxy0);
 	const Handle* pHandleB = static_cast<Handle*>(proxy1);
-	
+
 	//optimization 1: check the array index (memory address), instead of the m_pos
 
 	for (int axis = 0; axis < 3; axis++)
-	{ 
-		if (pHandleA->m_maxEdges[axis] < pHandleB->m_minEdges[axis] || 
-			pHandleB->m_maxEdges[axis] < pHandleA->m_minEdges[axis]) 
-		{ 
-			return false; 
-		} 
-	} 
+	{
+		if (pHandleA->m_maxEdges[axis] < pHandleB->m_minEdges[axis] ||
+			pHandleB->m_maxEdges[axis] < pHandleA->m_minEdges[axis])
+		{
+			return false;
+		}
+	}
 	return true;
 }
 
 template <typename BP_FP_INT_TYPE>
-bool btAxisSweep3Internal<BP_FP_INT_TYPE>::testOverlap2D(const Handle* pHandleA, const Handle* pHandleB,int axis0,int axis1)
+bool btAxisSweep3Internal<BP_FP_INT_TYPE>::testOverlap2D(const Handle* pHandleA, const Handle* pHandleB, int axis0, int axis1)
 {
 	//optimization 1: check the array index (memory address), instead of the m_pos
 
-	if (pHandleA->m_maxEdges[axis0] < pHandleB->m_minEdges[axis0] || 
+	if (pHandleA->m_maxEdges[axis0] < pHandleB->m_minEdges[axis0] ||
 		pHandleB->m_maxEdges[axis0] < pHandleA->m_minEdges[axis0] ||
 		pHandleA->m_maxEdges[axis1] < pHandleB->m_minEdges[axis1] ||
-		pHandleB->m_maxEdges[axis1] < pHandleA->m_minEdges[axis1]) 
-	{ 
-		return false; 
-	} 
+		pHandleB->m_maxEdges[axis1] < pHandleA->m_minEdges[axis1])
+	{
+		return false;
+	}
 	return true;
 }
 
 template <typename BP_FP_INT_TYPE>
-void btAxisSweep3Internal<BP_FP_INT_TYPE>::updateHandle(BP_FP_INT_TYPE handle, const btVector3& aabbMin,const btVector3& aabbMax,btDispatcher* dispatcher)
+void btAxisSweep3Internal<BP_FP_INT_TYPE>::updateHandle(BP_FP_INT_TYPE handle, const btVector3& aabbMin, const btVector3& aabbMax, btDispatcher* dispatcher)
 {
-//	btAssert(bounds.IsFinite());
+	//	btAssert(bounds.IsFinite());
 	//btAssert(bounds.HasVolume());
 
 	Handle* pHandle = getHandle(handle);
@@ -776,34 +731,28 @@ void btAxisSweep3Internal<BP_FP_INT_TYPE>::updateHandle(BP_FP_INT_TYPE handle, c
 
 		// expand (only adds overlaps)
 		if (dmin < 0)
-			sortMinDown(axis, emin,dispatcher,true);
+			sortMinDown(axis, emin, dispatcher, true);
 
 		if (dmax > 0)
-			sortMaxUp(axis, emax,dispatcher,true);
+			sortMaxUp(axis, emax, dispatcher, true);
 
 		// shrink (only removes overlaps)
 		if (dmin > 0)
-			sortMinUp(axis, emin,dispatcher,true);
+			sortMinUp(axis, emin, dispatcher, true);
 
 		if (dmax < 0)
-			sortMaxDown(axis, emax,dispatcher,true);
+			sortMaxDown(axis, emax, dispatcher, true);
 
 #ifdef DEBUG_BROADPHASE
-	debugPrintAxis(axis);
-#endif //DEBUG_BROADPHASE
+		debugPrintAxis(axis);
+#endif  //DEBUG_BROADPHASE
 	}
-
-	
 }
 
-
-
-
 // sorting a min edge downwards can only ever *add* overlaps
 template <typename BP_FP_INT_TYPE>
 void btAxisSweep3Internal<BP_FP_INT_TYPE>::sortMinDown(int axis, BP_FP_INT_TYPE edge, btDispatcher* /* dispatcher */, bool updateOverlaps)
 {
-
 	Edge* pEdge = m_pEdges[axis] + edge;
 	Edge* pPrev = pEdge - 1;
 	Handle* pHandleEdge = getHandle(pEdge->m_handle);
@@ -815,16 +764,15 @@ void btAxisSweep3Internal<BP_FP_INT_TYPE>::sortMinDown(int axis, BP_FP_INT_TYPE
 		if (pPrev->IsMax())
 		{
 			// if previous edge is a maximum check the bounds and add an overlap if necessary
-			const int axis1 = (1  << axis) & 3;
-			const int axis2 = (1  << axis1) & 3;
-			if (updateOverlaps && testOverlap2D(pHandleEdge, pHandlePrev,axis1,axis2))
+			const int axis1 = (1 << axis) & 3;
+			const int axis2 = (1 << axis1) & 3;
+			if (updateOverlaps && testOverlap2D(pHandleEdge, pHandlePrev, axis1, axis2))
 			{
-				m_pairCache->addOverlappingPair(pHandleEdge,pHandlePrev);
+				m_pairCache->addOverlappingPair(pHandleEdge, pHandlePrev);
 				if (m_userPairCallback)
-					m_userPairCallback->addOverlappingPair(pHandleEdge,pHandlePrev);
+					m_userPairCallback->addOverlappingPair(pHandleEdge, pHandlePrev);
 
 				//AddOverlap(pEdge->m_handle, pPrev->m_handle);
-
 			}
 
 			// update edge reference in other handle
@@ -847,8 +795,7 @@ void btAxisSweep3Internal<BP_FP_INT_TYPE>::sortMinDown(int axis, BP_FP_INT_TYPE
 
 #ifdef DEBUG_BROADPHASE
 	debugPrintAxis(axis);
-#endif //DEBUG_BROADPHASE
-
+#endif  //DEBUG_BROADPHASE
 }
 
 // sorting a min edge upwards can only ever *remove* overlaps
@@ -867,25 +814,21 @@ void btAxisSweep3Internal<BP_FP_INT_TYPE>::sortMinUp(int axis, BP_FP_INT_TYPE ed
 		{
 			Handle* handle0 = getHandle(pEdge->m_handle);
 			Handle* handle1 = getHandle(pNext->m_handle);
-			const int axis1 = (1  << axis) & 3;
-			const int axis2 = (1  << axis1) & 3;
-			
+			const int axis1 = (1 << axis) & 3;
+			const int axis2 = (1 << axis1) & 3;
+
 			// if next edge is maximum remove any overlap between the two handles
-			if (updateOverlaps 
+			if (updateOverlaps
 #ifdef USE_OVERLAP_TEST_ON_REMOVES
-				&& testOverlap2D(handle0,handle1,axis1,axis2)
-#endif //USE_OVERLAP_TEST_ON_REMOVES
-				)
+				&& testOverlap2D(handle0, handle1, axis1, axis2)
+#endif  //USE_OVERLAP_TEST_ON_REMOVES
+			)
 			{
-				
-
-				m_pairCache->removeOverlappingPair(handle0,handle1,dispatcher);	
+				m_pairCache->removeOverlappingPair(handle0, handle1, dispatcher);
 				if (m_userPairCallback)
-					m_userPairCallback->removeOverlappingPair(handle0,handle1,dispatcher);
-				
+					m_userPairCallback->removeOverlappingPair(handle0, handle1, dispatcher);
 			}
 
-
 			// update edge reference in other handle
 			pHandleNext->m_maxEdges[axis]--;
 		}
@@ -903,15 +846,12 @@ void btAxisSweep3Internal<BP_FP_INT_TYPE>::sortMinUp(int axis, BP_FP_INT_TYPE ed
 		pEdge++;
 		pNext++;
 	}
-
-
 }
 
 // sorting a max edge downwards can only ever *remove* overlaps
 template <typename BP_FP_INT_TYPE>
 void btAxisSweep3Internal<BP_FP_INT_TYPE>::sortMaxDown(int axis, BP_FP_INT_TYPE edge, btDispatcher* dispatcher, bool updateOverlaps)
 {
-
 	Edge* pEdge = m_pEdges[axis] + edge;
 	Edge* pPrev = pEdge - 1;
 	Handle* pHandleEdge = getHandle(pEdge->m_handle);
@@ -925,28 +865,25 @@ void btAxisSweep3Internal<BP_FP_INT_TYPE>::sortMaxDown(int axis, BP_FP_INT_TYPE
 			// if previous edge was a minimum remove any overlap between the two handles
 			Handle* handle0 = getHandle(pEdge->m_handle);
 			Handle* handle1 = getHandle(pPrev->m_handle);
-			const int axis1 = (1  << axis) & 3;
-			const int axis2 = (1  << axis1) & 3;
+			const int axis1 = (1 << axis) & 3;
+			const int axis2 = (1 << axis1) & 3;
 
-			if (updateOverlaps  
+			if (updateOverlaps
 #ifdef USE_OVERLAP_TEST_ON_REMOVES
-				&& testOverlap2D(handle0,handle1,axis1,axis2)
-#endif //USE_OVERLAP_TEST_ON_REMOVES
-				)
+				&& testOverlap2D(handle0, handle1, axis1, axis2)
+#endif  //USE_OVERLAP_TEST_ON_REMOVES
+			)
 			{
 				//this is done during the overlappingpairarray iteration/narrowphase collision
 
-				
-				m_pairCache->removeOverlappingPair(handle0,handle1,dispatcher);
+				m_pairCache->removeOverlappingPair(handle0, handle1, dispatcher);
 				if (m_userPairCallback)
-					m_userPairCallback->removeOverlappingPair(handle0,handle1,dispatcher);
-			
-
-
+					m_userPairCallback->removeOverlappingPair(handle0, handle1, dispatcher);
 			}
 
 			// update edge reference in other handle
-			pHandlePrev->m_minEdges[axis]++;;
+			pHandlePrev->m_minEdges[axis]++;
+			;
 		}
 		else
 			pHandlePrev->m_maxEdges[axis]++;
@@ -963,11 +900,9 @@ void btAxisSweep3Internal<BP_FP_INT_TYPE>::sortMaxDown(int axis, BP_FP_INT_TYPE
 		pPrev--;
 	}
 
-	
 #ifdef DEBUG_BROADPHASE
 	debugPrintAxis(axis);
-#endif //DEBUG_BROADPHASE
-
+#endif  //DEBUG_BROADPHASE
 }
 
 // sorting a max edge upwards can only ever *add* overlaps
@@ -982,19 +917,19 @@ void btAxisSweep3Internal<BP_FP_INT_TYPE>::sortMaxUp(int axis, BP_FP_INT_TYPE ed
 	{
 		Handle* pHandleNext = getHandle(pNext->m_handle);
 
-		const int axis1 = (1  << axis) & 3;
-		const int axis2 = (1  << axis1) & 3;
+		const int axis1 = (1 << axis) & 3;
+		const int axis2 = (1 << axis1) & 3;
 
 		if (!pNext->IsMax())
 		{
 			// if next edge is a minimum check the bounds and add an overlap if necessary
-			if (updateOverlaps && testOverlap2D(pHandleEdge, pHandleNext,axis1,axis2))
+			if (updateOverlaps && testOverlap2D(pHandleEdge, pHandleNext, axis1, axis2))
 			{
 				Handle* handle0 = getHandle(pEdge->m_handle);
 				Handle* handle1 = getHandle(pNext->m_handle);
-				m_pairCache->addOverlappingPair(handle0,handle1);
+				m_pairCache->addOverlappingPair(handle0, handle1);
 				if (m_userPairCallback)
-					m_userPairCallback->addOverlappingPair(handle0,handle1);
+					m_userPairCallback->addOverlappingPair(handle0, handle1);
 			}
 
 			// update edge reference in other handle
@@ -1014,7 +949,6 @@ void btAxisSweep3Internal<BP_FP_INT_TYPE>::sortMaxUp(int axis, BP_FP_INT_TYPE ed
 		pEdge++;
 		pNext++;
 	}
-	
 }
 
 #endif
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btBroadphaseInterface.h b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btBroadphaseInterface.h
index fb68e0024e0..b097eca5f5b 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btBroadphaseInterface.h
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btBroadphaseInterface.h
@@ -13,10 +13,8 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-#ifndef		BT_BROADPHASE_INTERFACE_H
-#define 	BT_BROADPHASE_INTERFACE_H
-
-
+#ifndef BT_BROADPHASE_INTERFACE_H
+#define BT_BROADPHASE_INTERFACE_H
 
 struct btDispatcherInfo;
 class btDispatcher;
@@ -24,27 +22,23 @@ class btDispatcher;
 
 class btOverlappingPairCache;
 
-
-
-struct	btBroadphaseAabbCallback
+struct btBroadphaseAabbCallback
 {
 	virtual ~btBroadphaseAabbCallback() {}
-	virtual bool	process(const btBroadphaseProxy* proxy) = 0;
+	virtual bool process(const btBroadphaseProxy* proxy) = 0;
 };
 
-
-struct	btBroadphaseRayCallback : public btBroadphaseAabbCallback
+struct btBroadphaseRayCallback : public btBroadphaseAabbCallback
 {
 	///added some cached data to accelerate ray-AABB tests
-	btVector3		m_rayDirectionInverse;
-	unsigned int	m_signs[3];
-	btScalar		m_lambda_max;
+	btVector3 m_rayDirectionInverse;
+	unsigned int m_signs[3];
+	btScalar m_lambda_max;
 
 	virtual ~btBroadphaseRayCallback() {}
-	
+
 protected:
-    
-    btBroadphaseRayCallback() {}
+	btBroadphaseRayCallback() {}
 };
 
 #include "LinearMath/btVector3.h"
@@ -57,30 +51,29 @@ class btBroadphaseInterface
 public:
 	virtual ~btBroadphaseInterface() {}
 
-	virtual btBroadphaseProxy*	createProxy(  const btVector3& aabbMin,  const btVector3& aabbMax,int shapeType,void* userPtr,  int collisionFilterGroup, int collisionFilterMask, btDispatcher* dispatcher) =0;
-	virtual void	destroyProxy(btBroadphaseProxy* proxy,btDispatcher* dispatcher)=0;
-	virtual void	setAabb(btBroadphaseProxy* proxy,const btVector3& aabbMin,const btVector3& aabbMax, btDispatcher* dispatcher)=0;
-	virtual void	getAabb(btBroadphaseProxy* proxy,btVector3& aabbMin, btVector3& aabbMax ) const =0;
+	virtual btBroadphaseProxy* createProxy(const btVector3& aabbMin, const btVector3& aabbMax, int shapeType, void* userPtr, int collisionFilterGroup, int collisionFilterMask, btDispatcher* dispatcher) = 0;
+	virtual void destroyProxy(btBroadphaseProxy* proxy, btDispatcher* dispatcher) = 0;
+	virtual void setAabb(btBroadphaseProxy* proxy, const btVector3& aabbMin, const btVector3& aabbMax, btDispatcher* dispatcher) = 0;
+	virtual void getAabb(btBroadphaseProxy* proxy, btVector3& aabbMin, btVector3& aabbMax) const = 0;
 
-	virtual void	rayTest(const btVector3& rayFrom,const btVector3& rayTo, btBroadphaseRayCallback& rayCallback, const btVector3& aabbMin=btVector3(0,0,0), const btVector3& aabbMax = btVector3(0,0,0)) = 0;
+	virtual void rayTest(const btVector3& rayFrom, const btVector3& rayTo, btBroadphaseRayCallback& rayCallback, const btVector3& aabbMin = btVector3(0, 0, 0), const btVector3& aabbMax = btVector3(0, 0, 0)) = 0;
 
-	virtual void	aabbTest(const btVector3& aabbMin, const btVector3& aabbMax, btBroadphaseAabbCallback& callback) = 0;
+	virtual void aabbTest(const btVector3& aabbMin, const btVector3& aabbMax, btBroadphaseAabbCallback& callback) = 0;
 
 	///calculateOverlappingPairs is optional: incremental algorithms (sweep and prune) might do it during the set aabb
-	virtual void	calculateOverlappingPairs(btDispatcher* dispatcher)=0;
+	virtual void calculateOverlappingPairs(btDispatcher* dispatcher) = 0;
 
-	virtual	btOverlappingPairCache*	getOverlappingPairCache()=0;
-	virtual	const btOverlappingPairCache*	getOverlappingPairCache() const =0;
+	virtual btOverlappingPairCache* getOverlappingPairCache() = 0;
+	virtual const btOverlappingPairCache* getOverlappingPairCache() const = 0;
 
 	///getAabb returns the axis aligned bounding box in the 'global' coordinate frame
 	///will add some transform later
-	virtual void getBroadphaseAabb(btVector3& aabbMin,btVector3& aabbMax) const =0;
+	virtual void getBroadphaseAabb(btVector3& aabbMin, btVector3& aabbMax) const = 0;
 
 	///reset broadphase internal structures, to ensure determinism/reproducability
-	virtual void resetPool(btDispatcher* dispatcher) { (void) dispatcher; };
-
-	virtual void	printStats() = 0;
+	virtual void resetPool(btDispatcher* dispatcher) { (void)dispatcher; };
 
+	virtual void printStats() = 0;
 };
 
-#endif //BT_BROADPHASE_INTERFACE_H
+#endif  //BT_BROADPHASE_INTERFACE_H
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btBroadphaseProxy.cpp b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btBroadphaseProxy.cpp
index 0fd4ef46be5..7ee065aac32 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btBroadphaseProxy.cpp
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btBroadphaseProxy.cpp
@@ -15,4 +15,4 @@ subject to the following restrictions:
 
 #include "btBroadphaseProxy.h"
 
-BT_NOT_EMPTY_FILE // fix warning LNK4221: This object file does not define any previously undefined public symbols, so it will not be used by any link operation that consumes this library
+BT_NOT_EMPTY_FILE  // fix warning LNK4221: This object file does not define any previously undefined public symbols, so it will not be used by any link operation that consumes this library
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btBroadphaseProxy.h b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btBroadphaseProxy.h
index f6e1202a697..825caeef566 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btBroadphaseProxy.h
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btBroadphaseProxy.h
@@ -16,11 +16,10 @@ subject to the following restrictions:
 #ifndef BT_BROADPHASE_PROXY_H
 #define BT_BROADPHASE_PROXY_H
 
-#include "LinearMath/btScalar.h" //for SIMD_FORCE_INLINE
+#include "LinearMath/btScalar.h"  //for SIMD_FORCE_INLINE
 #include "LinearMath/btVector3.h"
 #include "LinearMath/btAlignedAllocator.h"
 
-
 /// btDispatcher uses these types
 /// IMPORTANT NOTE:The types are ordered polyhedral, implicit convex and concave
 /// to facilitate type checking
@@ -35,8 +34,8 @@ enum BroadphaseNativeTypes
 	CONVEX_HULL_SHAPE_PROXYTYPE,
 	CONVEX_POINT_CLOUD_SHAPE_PROXYTYPE,
 	CUSTOM_POLYHEDRAL_SHAPE_TYPE,
-//implicit convex shapes
-IMPLICIT_CONVEX_SHAPES_START_HERE,
+	//implicit convex shapes
+	IMPLICIT_CONVEX_SHAPES_START_HERE,
 	SPHERE_SHAPE_PROXYTYPE,
 	MULTI_SPHERE_SHAPE_PROXYTYPE,
 	CAPSULE_SHAPE_PROXYTYPE,
@@ -49,8 +48,8 @@ IMPLICIT_CONVEX_SHAPES_START_HERE,
 	BOX_2D_SHAPE_PROXYTYPE,
 	CONVEX_2D_SHAPE_PROXYTYPE,
 	CUSTOM_CONVEX_SHAPE_TYPE,
-//concave shapes
-CONCAVE_SHAPES_START_HERE,
+	//concave shapes
+	CONCAVE_SHAPES_START_HERE,
 	//keep all the convex shapetype below here, for the check IsConvexShape in broadphase proxy!
 	TRIANGLE_MESH_SHAPE_PROXYTYPE,
 	SCALED_TRIANGLE_MESH_SHAPE_PROXYTYPE,
@@ -58,16 +57,16 @@ CONCAVE_SHAPES_START_HERE,
 	FAST_CONCAVE_MESH_PROXYTYPE,
 	//terrain
 	TERRAIN_SHAPE_PROXYTYPE,
-///Used for GIMPACT Trimesh integration
+	///Used for GIMPACT Trimesh integration
 	GIMPACT_SHAPE_PROXYTYPE,
-///Multimaterial mesh
-    MULTIMATERIAL_TRIANGLE_MESH_PROXYTYPE,
-	
+	///Multimaterial mesh
+	MULTIMATERIAL_TRIANGLE_MESH_PROXYTYPE,
+
 	EMPTY_SHAPE_PROXYTYPE,
 	STATIC_PLANE_PROXYTYPE,
 	CUSTOM_CONCAVE_SHAPE_TYPE,
-	SDF_SHAPE_PROXYTYPE=CUSTOM_CONCAVE_SHAPE_TYPE,
-CONCAVE_SHAPES_END_HERE,
+	SDF_SHAPE_PROXYTYPE = CUSTOM_CONCAVE_SHAPE_TYPE,
+	CONCAVE_SHAPES_END_HERE,
 
 	COMPOUND_SHAPE_PROXYTYPE,
 
@@ -77,38 +76,37 @@ CONCAVE_SHAPES_END_HERE,
 	INVALID_SHAPE_PROXYTYPE,
 
 	MAX_BROADPHASE_COLLISION_TYPES
-	
+
 };
 
-
-///The btBroadphaseProxy is the main class that can be used with the Bullet broadphases. 
+///The btBroadphaseProxy is the main class that can be used with the Bullet broadphases.
 ///It stores collision shape type information, collision filter information and a client object, typically a btCollisionObject or btRigidBody.
-ATTRIBUTE_ALIGNED16(struct) btBroadphaseProxy
+ATTRIBUTE_ALIGNED16(struct)
+btBroadphaseProxy
 {
+	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-BT_DECLARE_ALIGNED_ALLOCATOR();
-	
 	///optional filtering to cull potential collisions
 	enum CollisionFilterGroups
 	{
-	        DefaultFilter = 1,
-	        StaticFilter = 2,
-	        KinematicFilter = 4,
-	        DebrisFilter = 8,
-			SensorTrigger = 16,
-			CharacterFilter = 32,
-	        AllFilter = -1 //all bits sets: DefaultFilter | StaticFilter | KinematicFilter | DebrisFilter | SensorTrigger
+		DefaultFilter = 1,
+		StaticFilter = 2,
+		KinematicFilter = 4,
+		DebrisFilter = 8,
+		SensorTrigger = 16,
+		CharacterFilter = 32,
+		AllFilter = -1  //all bits sets: DefaultFilter | StaticFilter | KinematicFilter | DebrisFilter | SensorTrigger
 	};
 
 	//Usually the client btCollisionObject or Rigidbody class
-	void*	m_clientObject;
-	int		m_collisionFilterGroup;
-	int		m_collisionFilterMask;
+	void* m_clientObject;
+	int m_collisionFilterGroup;
+	int m_collisionFilterMask;
 
-	int		m_uniqueId;//m_uniqueId is introduced for paircache. could get rid of this, by calculating the address offset etc.
+	int m_uniqueId;  //m_uniqueId is introduced for paircache. could get rid of this, by calculating the address offset etc.
 
-	btVector3	m_aabbMin;
-	btVector3	m_aabbMax;
+	btVector3 m_aabbMin;
+	btVector3 m_aabbMax;
 
 	SIMD_FORCE_INLINE int getUid() const
 	{
@@ -116,47 +114,45 @@ BT_DECLARE_ALIGNED_ALLOCATOR();
 	}
 
 	//used for memory pools
-	btBroadphaseProxy() :m_clientObject(0)
+	btBroadphaseProxy() : m_clientObject(0)
 	{
 	}
 
-	btBroadphaseProxy(const btVector3& aabbMin,const btVector3& aabbMax,void* userPtr, int collisionFilterGroup,  int collisionFilterMask)
-		:m_clientObject(userPtr),
-		m_collisionFilterGroup(collisionFilterGroup),
-		m_collisionFilterMask(collisionFilterMask),
-		m_aabbMin(aabbMin),
-		m_aabbMax(aabbMax)
+	btBroadphaseProxy(const btVector3& aabbMin, const btVector3& aabbMax, void* userPtr, int collisionFilterGroup, int collisionFilterMask)
+		: m_clientObject(userPtr),
+		  m_collisionFilterGroup(collisionFilterGroup),
+		  m_collisionFilterMask(collisionFilterMask),
+		  m_aabbMin(aabbMin),
+		  m_aabbMax(aabbMax)
 	{
 	}
 
-	
-
 	static SIMD_FORCE_INLINE bool isPolyhedral(int proxyType)
 	{
-		return (proxyType  < IMPLICIT_CONVEX_SHAPES_START_HERE);
+		return (proxyType < IMPLICIT_CONVEX_SHAPES_START_HERE);
 	}
 
-	static SIMD_FORCE_INLINE bool	isConvex(int proxyType)
+	static SIMD_FORCE_INLINE bool isConvex(int proxyType)
 	{
 		return (proxyType < CONCAVE_SHAPES_START_HERE);
 	}
 
-	static SIMD_FORCE_INLINE bool	isNonMoving(int proxyType)
+	static SIMD_FORCE_INLINE bool isNonMoving(int proxyType)
 	{
-		return (isConcave(proxyType) && !(proxyType==GIMPACT_SHAPE_PROXYTYPE));
+		return (isConcave(proxyType) && !(proxyType == GIMPACT_SHAPE_PROXYTYPE));
 	}
 
-	static SIMD_FORCE_INLINE bool	isConcave(int proxyType)
+	static SIMD_FORCE_INLINE bool isConcave(int proxyType)
 	{
 		return ((proxyType > CONCAVE_SHAPES_START_HERE) &&
-			(proxyType < CONCAVE_SHAPES_END_HERE));
+				(proxyType < CONCAVE_SHAPES_END_HERE));
 	}
-	static SIMD_FORCE_INLINE bool	isCompound(int proxyType)
+	static SIMD_FORCE_INLINE bool isCompound(int proxyType)
 	{
 		return (proxyType == COMPOUND_SHAPE_PROXYTYPE);
 	}
 
-	static SIMD_FORCE_INLINE bool	isSoftBody(int proxyType)
+	static SIMD_FORCE_INLINE bool isSoftBody(int proxyType)
 	{
 		return (proxyType == SOFTBODY_SHAPE_PROXYTYPE);
 	}
@@ -168,67 +164,62 @@ BT_DECLARE_ALIGNED_ALLOCATOR();
 
 	static SIMD_FORCE_INLINE bool isConvex2d(int proxyType)
 	{
-		return (proxyType == BOX_2D_SHAPE_PROXYTYPE) ||	(proxyType == CONVEX_2D_SHAPE_PROXYTYPE);
+		return (proxyType == BOX_2D_SHAPE_PROXYTYPE) || (proxyType == CONVEX_2D_SHAPE_PROXYTYPE);
 	}
-
-	
-}
-;
+};
 
 class btCollisionAlgorithm;
 
 struct btBroadphaseProxy;
 
-
-
 ///The btBroadphasePair class contains a pair of aabb-overlapping objects.
 ///A btDispatcher can search a btCollisionAlgorithm that performs exact/narrowphase collision detection on the actual collision shapes.
-ATTRIBUTE_ALIGNED16(struct) btBroadphasePair
+ATTRIBUTE_ALIGNED16(struct)
+btBroadphasePair
 {
-	btBroadphasePair ()
-		:
-	m_pProxy0(0),
-		m_pProxy1(0),
-		m_algorithm(0),
-		m_internalInfo1(0)
+	btBroadphasePair()
+		: m_pProxy0(0),
+		  m_pProxy1(0),
+		  m_algorithm(0),
+		  m_internalInfo1(0)
 	{
 	}
 
-BT_DECLARE_ALIGNED_ALLOCATOR();
+	BT_DECLARE_ALIGNED_ALLOCATOR();
 
 	btBroadphasePair(const btBroadphasePair& other)
-		:		m_pProxy0(other.m_pProxy0),
-				m_pProxy1(other.m_pProxy1),
-				m_algorithm(other.m_algorithm),
-				m_internalInfo1(other.m_internalInfo1)
+		: m_pProxy0(other.m_pProxy0),
+		  m_pProxy1(other.m_pProxy1),
+		  m_algorithm(other.m_algorithm),
+		  m_internalInfo1(other.m_internalInfo1)
 	{
 	}
-	btBroadphasePair(btBroadphaseProxy& proxy0,btBroadphaseProxy& proxy1)
+	btBroadphasePair(btBroadphaseProxy & proxy0, btBroadphaseProxy & proxy1)
 	{
-
 		//keep them sorted, so the std::set operations work
 		if (proxy0.m_uniqueId < proxy1.m_uniqueId)
-        { 
-            m_pProxy0 = &proxy0; 
-            m_pProxy1 = &proxy1; 
-        }
-        else 
-        { 
-			m_pProxy0 = &proxy1; 
-            m_pProxy1 = &proxy0; 
-        }
+		{
+			m_pProxy0 = &proxy0;
+			m_pProxy1 = &proxy1;
+		}
+		else
+		{
+			m_pProxy0 = &proxy1;
+			m_pProxy1 = &proxy0;
+		}
 
 		m_algorithm = 0;
 		m_internalInfo1 = 0;
-
 	}
-	
+
 	btBroadphaseProxy* m_pProxy0;
 	btBroadphaseProxy* m_pProxy1;
-	
-	mutable btCollisionAlgorithm* m_algorithm;
-	union { void* m_internalInfo1; int m_internalTmpValue;};//don't use this data, it will be removed in future version.
 
+	mutable btCollisionAlgorithm* m_algorithm;
+	union {
+		void* m_internalInfo1;
+		int m_internalTmpValue;
+	};  //don't use this data, it will be removed in future version.
 };
 
 /*
@@ -240,31 +231,25 @@ SIMD_FORCE_INLINE bool operator<(const btBroadphasePair& a, const btBroadphasePa
 }
 */
 
-
-
 class btBroadphasePairSortPredicate
 {
-	public:
+public:
+	bool operator()(const btBroadphasePair& a, const btBroadphasePair& b) const
+	{
+		const int uidA0 = a.m_pProxy0 ? a.m_pProxy0->m_uniqueId : -1;
+		const int uidB0 = b.m_pProxy0 ? b.m_pProxy0->m_uniqueId : -1;
+		const int uidA1 = a.m_pProxy1 ? a.m_pProxy1->m_uniqueId : -1;
+		const int uidB1 = b.m_pProxy1 ? b.m_pProxy1->m_uniqueId : -1;
 
-		bool operator() ( const btBroadphasePair& a, const btBroadphasePair& b ) const
-		{
-			const int uidA0 = a.m_pProxy0 ? a.m_pProxy0->m_uniqueId : -1;
-			const int uidB0 = b.m_pProxy0 ? b.m_pProxy0->m_uniqueId : -1;
-			const int uidA1 = a.m_pProxy1 ? a.m_pProxy1->m_uniqueId : -1;
-			const int uidB1 = b.m_pProxy1 ? b.m_pProxy1->m_uniqueId : -1;
-
-			 return uidA0 > uidB0 || 
-				(a.m_pProxy0 == b.m_pProxy0 && uidA1 > uidB1) ||
-				(a.m_pProxy0 == b.m_pProxy0 && a.m_pProxy1 == b.m_pProxy1 && a.m_algorithm > b.m_algorithm); 
-		}
+		return uidA0 > uidB0 ||
+			   (a.m_pProxy0 == b.m_pProxy0 && uidA1 > uidB1) ||
+			   (a.m_pProxy0 == b.m_pProxy0 && a.m_pProxy1 == b.m_pProxy1 && a.m_algorithm > b.m_algorithm);
+	}
 };
 
-
-SIMD_FORCE_INLINE bool operator==(const btBroadphasePair& a, const btBroadphasePair& b) 
+SIMD_FORCE_INLINE bool operator==(const btBroadphasePair& a, const btBroadphasePair& b)
 {
-	 return (a.m_pProxy0 == b.m_pProxy0) && (a.m_pProxy1 == b.m_pProxy1);
+	return (a.m_pProxy0 == b.m_pProxy0) && (a.m_pProxy1 == b.m_pProxy1);
 }
 
-
-#endif //BT_BROADPHASE_PROXY_H
-
+#endif  //BT_BROADPHASE_PROXY_H
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btCollisionAlgorithm.cpp b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btCollisionAlgorithm.cpp
index c95d1be0f2c..6e36d3bd730 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btCollisionAlgorithm.cpp
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btCollisionAlgorithm.cpp
@@ -20,4 +20,3 @@ btCollisionAlgorithm::btCollisionAlgorithm(const btCollisionAlgorithmConstructio
 {
 	m_dispatcher = ci.m_dispatcher1;
 }
-
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btCollisionAlgorithm.h b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btCollisionAlgorithm.h
index 405656236bd..b00c0b1b4bb 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btCollisionAlgorithm.h
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btCollisionAlgorithm.h
@@ -25,57 +25,51 @@ class btManifoldResult;
 class btCollisionObject;
 struct btCollisionObjectWrapper;
 struct btDispatcherInfo;
-class	btPersistentManifold;
+class btPersistentManifold;
 
-typedef btAlignedObjectArray<btPersistentManifold*>	btManifoldArray;
+typedef btAlignedObjectArray<btPersistentManifold*> btManifoldArray;
 
 struct btCollisionAlgorithmConstructionInfo
 {
 	btCollisionAlgorithmConstructionInfo()
-		:m_dispatcher1(0),
-		m_manifold(0)
+		: m_dispatcher1(0),
+		  m_manifold(0)
 	{
 	}
-	btCollisionAlgorithmConstructionInfo(btDispatcher* dispatcher,int temp)
-		:m_dispatcher1(dispatcher)
+	btCollisionAlgorithmConstructionInfo(btDispatcher* dispatcher, int temp)
+		: m_dispatcher1(dispatcher)
 	{
 		(void)temp;
 	}
 
-	btDispatcher*	m_dispatcher1;
-	btPersistentManifold*	m_manifold;
-
-//	int	getDispatcherId();
+	btDispatcher* m_dispatcher1;
+	btPersistentManifold* m_manifold;
 
+	//	int	getDispatcherId();
 };
 
-
 ///btCollisionAlgorithm is an collision interface that is compatible with the Broadphase and btDispatcher.
 ///It is persistent over frames
 class btCollisionAlgorithm
 {
+protected:
+	btDispatcher* m_dispatcher;
 
 protected:
+	//	int	getDispatcherId();
 
-	btDispatcher*	m_dispatcher;
-
-protected:
-//	int	getDispatcherId();
-	
 public:
-
-	btCollisionAlgorithm() {};
+	btCollisionAlgorithm(){};
 
 	btCollisionAlgorithm(const btCollisionAlgorithmConstructionInfo& ci);
 
-	virtual ~btCollisionAlgorithm() {};
+	virtual ~btCollisionAlgorithm(){};
 
-	virtual void processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut) = 0;
+	virtual void processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut) = 0;
 
-	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut) = 0;
+	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut) = 0;
 
-	virtual	void	getAllContactManifolds(btManifoldArray&	manifoldArray) = 0;
+	virtual void getAllContactManifolds(btManifoldArray& manifoldArray) = 0;
 };
 
-
-#endif //BT_COLLISION_ALGORITHM_H
+#endif  //BT_COLLISION_ALGORITHM_H
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDbvt.cpp b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDbvt.cpp
index d791d074182..37156fd589e 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDbvt.cpp
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDbvt.cpp
@@ -17,210 +17,225 @@ subject to the following restrictions:
 #include "btDbvt.h"
 
 //
-typedef btAlignedObjectArray<btDbvtNode*>			tNodeArray;
-typedef btAlignedObjectArray<const btDbvtNode*>	tConstNodeArray;
+typedef btAlignedObjectArray<btDbvtNode*> tNodeArray;
+typedef btAlignedObjectArray<const btDbvtNode*> tConstNodeArray;
 
 //
 struct btDbvtNodeEnumerator : btDbvt::ICollide
 {
-	tConstNodeArray	nodes;
+	tConstNodeArray nodes;
 	void Process(const btDbvtNode* n) { nodes.push_back(n); }
 };
 
 //
-static DBVT_INLINE int			indexof(const btDbvtNode* node)
+static DBVT_INLINE int indexof(const btDbvtNode* node)
 {
-	return(node->parent->childs[1]==node);
+	return (node->parent->childs[1] == node);
 }
 
 //
-static DBVT_INLINE btDbvtVolume	merge(	const btDbvtVolume& a,
+static DBVT_INLINE btDbvtVolume merge(const btDbvtVolume& a,
 									  const btDbvtVolume& b)
 {
-#if (DBVT_MERGE_IMPL==DBVT_IMPL_SSE)
-	ATTRIBUTE_ALIGNED16( char locals[sizeof(btDbvtAabbMm)]);
-	btDbvtVolume* ptr = (btDbvtVolume*) locals;
-	btDbvtVolume&	res=*ptr;
+#if (DBVT_MERGE_IMPL == DBVT_IMPL_SSE)
+	ATTRIBUTE_ALIGNED16(char locals[sizeof(btDbvtAabbMm)]);
+	btDbvtVolume* ptr = (btDbvtVolume*)locals;
+	btDbvtVolume& res = *ptr;
 #else
-		btDbvtVolume	res;
+	btDbvtVolume res;
 #endif
-	Merge(a,b,res);
-	return(res);
+	Merge(a, b, res);
+	return (res);
 }
 
 // volume+edge lengths
-static DBVT_INLINE btScalar		size(const btDbvtVolume& a)
+static DBVT_INLINE btScalar size(const btDbvtVolume& a)
 {
-	const btVector3	edges=a.Lengths();
-	return(	edges.x()*edges.y()*edges.z()+
-		edges.x()+edges.y()+edges.z());
+	const btVector3 edges = a.Lengths();
+	return (edges.x() * edges.y() * edges.z() +
+			edges.x() + edges.y() + edges.z());
 }
 
 //
-static void						getmaxdepth(const btDbvtNode* node,int depth,int& maxdepth)
+static void getmaxdepth(const btDbvtNode* node, int depth, int& maxdepth)
 {
-	if(node->isinternal())
+	if (node->isinternal())
 	{
-		getmaxdepth(node->childs[0],depth+1,maxdepth);
-		getmaxdepth(node->childs[1],depth+1,maxdepth);
-	} else maxdepth=btMax(maxdepth,depth);
+		getmaxdepth(node->childs[0], depth + 1, maxdepth);
+		getmaxdepth(node->childs[1], depth + 1, maxdepth);
+	}
+	else
+		maxdepth = btMax(maxdepth, depth);
 }
 
 //
-static DBVT_INLINE void			deletenode(	btDbvt* pdbvt,
-										   btDbvtNode* node)
+static DBVT_INLINE void deletenode(btDbvt* pdbvt,
+								   btDbvtNode* node)
 {
 	btAlignedFree(pdbvt->m_free);
-	pdbvt->m_free=node;
+	pdbvt->m_free = node;
 }
 
 //
-static void						recursedeletenode(	btDbvt* pdbvt,
-												  btDbvtNode* node)
+static void recursedeletenode(btDbvt* pdbvt,
+							  btDbvtNode* node)
 {
-	if(!node->isleaf())
+	if (!node->isleaf())
 	{
-		recursedeletenode(pdbvt,node->childs[0]);
-		recursedeletenode(pdbvt,node->childs[1]);
+		recursedeletenode(pdbvt, node->childs[0]);
+		recursedeletenode(pdbvt, node->childs[1]);
 	}
-	if(node==pdbvt->m_root) pdbvt->m_root=0;
-	deletenode(pdbvt,node);
+	if (node == pdbvt->m_root) pdbvt->m_root = 0;
+	deletenode(pdbvt, node);
 }
 
 //
-static DBVT_INLINE btDbvtNode*	createnode(	btDbvt* pdbvt,
-										   btDbvtNode* parent,
-										   void* data)
+static DBVT_INLINE btDbvtNode* createnode(btDbvt* pdbvt,
+										  btDbvtNode* parent,
+										  void* data)
 {
-	btDbvtNode*	node;
-	if(pdbvt->m_free)
-	{ node=pdbvt->m_free;pdbvt->m_free=0; }
-	else
-	{ node=new(btAlignedAlloc(sizeof(btDbvtNode),16)) btDbvtNode(); }
-	node->parent	=	parent;
-	node->data		=	data;
-	node->childs[1]	=	0;
-	return(node);
-}
-
-//
-static DBVT_INLINE btDbvtNode*	createnode(	btDbvt* pdbvt,
-										   btDbvtNode* parent,
-										   const btDbvtVolume& volume,
-										   void* data)
-{
-	btDbvtNode*	node=createnode(pdbvt,parent,data);
-	node->volume=volume;
-	return(node);
-}
-
-//
-static DBVT_INLINE btDbvtNode*	createnode(	btDbvt* pdbvt,
-										   btDbvtNode* parent,
-										   const btDbvtVolume& volume0,
-										   const btDbvtVolume& volume1,
-										   void* data)
-{
-	btDbvtNode*	node=createnode(pdbvt,parent,data);
-	Merge(volume0,volume1,node->volume);
-	return(node);
-}
-
-//
-static void						insertleaf(	btDbvt* pdbvt,
-										   btDbvtNode* root,
-										   btDbvtNode* leaf)
-{
-	if(!pdbvt->m_root)
+	btDbvtNode* node;
+	if (pdbvt->m_free)
 	{
-		pdbvt->m_root	=	leaf;
-		leaf->parent	=	0;
+		node = pdbvt->m_free;
+		pdbvt->m_free = 0;
 	}
 	else
 	{
-		if(!root->isleaf())
+		node = new (btAlignedAlloc(sizeof(btDbvtNode), 16)) btDbvtNode();
+	}
+	node->parent = parent;
+	node->data = data;
+	node->childs[1] = 0;
+	return (node);
+}
+
+//
+static DBVT_INLINE btDbvtNode* createnode(btDbvt* pdbvt,
+										  btDbvtNode* parent,
+										  const btDbvtVolume& volume,
+										  void* data)
+{
+	btDbvtNode* node = createnode(pdbvt, parent, data);
+	node->volume = volume;
+	return (node);
+}
+
+//
+static DBVT_INLINE btDbvtNode* createnode(btDbvt* pdbvt,
+										  btDbvtNode* parent,
+										  const btDbvtVolume& volume0,
+										  const btDbvtVolume& volume1,
+										  void* data)
+{
+	btDbvtNode* node = createnode(pdbvt, parent, data);
+	Merge(volume0, volume1, node->volume);
+	return (node);
+}
+
+//
+static void insertleaf(btDbvt* pdbvt,
+					   btDbvtNode* root,
+					   btDbvtNode* leaf)
+{
+	if (!pdbvt->m_root)
+	{
+		pdbvt->m_root = leaf;
+		leaf->parent = 0;
+	}
+	else
+	{
+		if (!root->isleaf())
 		{
-			do	{
-				root=root->childs[Select(	leaf->volume,
-					root->childs[0]->volume,
-					root->childs[1]->volume)];
-			} while(!root->isleaf());
+			do
+			{
+				root = root->childs[Select(leaf->volume,
+										   root->childs[0]->volume,
+										   root->childs[1]->volume)];
+			} while (!root->isleaf());
 		}
-		btDbvtNode*	prev=root->parent;
-		btDbvtNode*	node=createnode(pdbvt,prev,leaf->volume,root->volume,0);
-		if(prev)
+		btDbvtNode* prev = root->parent;
+		btDbvtNode* node = createnode(pdbvt, prev, leaf->volume, root->volume, 0);
+		if (prev)
 		{
-			prev->childs[indexof(root)]	=	node;
-			node->childs[0]				=	root;root->parent=node;
-			node->childs[1]				=	leaf;leaf->parent=node;
-			do	{
-				if(!prev->volume.Contain(node->volume))
-					Merge(prev->childs[0]->volume,prev->childs[1]->volume,prev->volume);
+			prev->childs[indexof(root)] = node;
+			node->childs[0] = root;
+			root->parent = node;
+			node->childs[1] = leaf;
+			leaf->parent = node;
+			do
+			{
+				if (!prev->volume.Contain(node->volume))
+					Merge(prev->childs[0]->volume, prev->childs[1]->volume, prev->volume);
 				else
 					break;
-				node=prev;
-			} while(0!=(prev=node->parent));
+				node = prev;
+			} while (0 != (prev = node->parent));
 		}
 		else
 		{
-			node->childs[0]	=	root;root->parent=node;
-			node->childs[1]	=	leaf;leaf->parent=node;
-			pdbvt->m_root	=	node;
+			node->childs[0] = root;
+			root->parent = node;
+			node->childs[1] = leaf;
+			leaf->parent = node;
+			pdbvt->m_root = node;
 		}
 	}
 }
 
 //
-static btDbvtNode*				removeleaf(	btDbvt* pdbvt,
-										   btDbvtNode* leaf)
+static btDbvtNode* removeleaf(btDbvt* pdbvt,
+							  btDbvtNode* leaf)
 {
-	if(leaf==pdbvt->m_root)
+	if (leaf == pdbvt->m_root)
 	{
-		pdbvt->m_root=0;
-		return(0);
+		pdbvt->m_root = 0;
+		return (0);
 	}
 	else
 	{
-		btDbvtNode*	parent=leaf->parent;
-		btDbvtNode*	prev=parent->parent;
-		btDbvtNode*	sibling=parent->childs[1-indexof(leaf)];			
-		if(prev)
+		btDbvtNode* parent = leaf->parent;
+		btDbvtNode* prev = parent->parent;
+		btDbvtNode* sibling = parent->childs[1 - indexof(leaf)];
+		if (prev)
 		{
-			prev->childs[indexof(parent)]=sibling;
-			sibling->parent=prev;
-			deletenode(pdbvt,parent);
-			while(prev)
+			prev->childs[indexof(parent)] = sibling;
+			sibling->parent = prev;
+			deletenode(pdbvt, parent);
+			while (prev)
 			{
-				const btDbvtVolume	pb=prev->volume;
-				Merge(prev->childs[0]->volume,prev->childs[1]->volume,prev->volume);
-				if(NotEqual(pb,prev->volume))
+				const btDbvtVolume pb = prev->volume;
+				Merge(prev->childs[0]->volume, prev->childs[1]->volume, prev->volume);
+				if (NotEqual(pb, prev->volume))
 				{
-					prev=prev->parent;
-				} else break;
+					prev = prev->parent;
+				}
+				else
+					break;
 			}
-			return(prev?prev:pdbvt->m_root);
+			return (prev ? prev : pdbvt->m_root);
 		}
 		else
-		{								
-			pdbvt->m_root=sibling;
-			sibling->parent=0;
-			deletenode(pdbvt,parent);
-			return(pdbvt->m_root);
-		}			
+		{
+			pdbvt->m_root = sibling;
+			sibling->parent = 0;
+			deletenode(pdbvt, parent);
+			return (pdbvt->m_root);
+		}
 	}
 }
 
 //
-static void						fetchleaves(btDbvt* pdbvt,
-											btDbvtNode* root,
-											tNodeArray& leaves,
-											int depth=-1)
+static void fetchleaves(btDbvt* pdbvt,
+						btDbvtNode* root,
+						tNodeArray& leaves,
+						int depth = -1)
 {
-	if(root->isinternal()&&depth)
+	if (root->isinternal() && depth)
 	{
-		fetchleaves(pdbvt,root->childs[0],leaves,depth-1);
-		fetchleaves(pdbvt,root->childs[1],leaves,depth-1);
-		deletenode(pdbvt,root);
+		fetchleaves(pdbvt, root->childs[0], leaves, depth - 1);
+		fetchleaves(pdbvt, root->childs[1], leaves, depth - 1);
+		deletenode(pdbvt, root);
 	}
 	else
 	{
@@ -229,51 +244,50 @@ static void						fetchleaves(btDbvt* pdbvt,
 }
 
 //
-static bool						leftOfAxis(	const btDbvtNode* node,
-										   const btVector3& org,
-										   const btVector3& axis)
+static bool leftOfAxis(const btDbvtNode* node,
+					   const btVector3& org,
+					   const btVector3& axis)
 {
 	return btDot(axis, node->volume.Center() - org) <= 0;
 }
 
-
 // Partitions leaves such that leaves[0, n) are on the
 // left of axis, and leaves[n, count) are on the right
 // of axis. returns N.
-static int						split(	btDbvtNode** leaves,
-									  int count,
-									  const btVector3& org,
-									  const btVector3& axis)
+static int split(btDbvtNode** leaves,
+				 int count,
+				 const btVector3& org,
+				 const btVector3& axis)
 {
-	int begin=0;
-	int end=count;
-	for(;;)
+	int begin = 0;
+	int end = count;
+	for (;;)
 	{
-		while(begin!=end && leftOfAxis(leaves[begin],org,axis))
+		while (begin != end && leftOfAxis(leaves[begin], org, axis))
 		{
 			++begin;
 		}
 
-		if(begin==end)
+		if (begin == end)
 		{
 			break;
 		}
 
-		while(begin!=end && !leftOfAxis(leaves[end-1],org,axis))
+		while (begin != end && !leftOfAxis(leaves[end - 1], org, axis))
 		{
 			--end;
 		}
 
-		if(begin==end)
+		if (begin == end)
 		{
 			break;
 		}
 
 		// swap out of place nodes
 		--end;
-		btDbvtNode* temp=leaves[begin];
-		leaves[begin]=leaves[end];
-		leaves[end]=temp;
+		btDbvtNode* temp = leaves[begin];
+		leaves[begin] = leaves[end];
+		leaves[end] = temp;
 		++begin;
 	}
 
@@ -281,150 +295,153 @@ static int						split(	btDbvtNode** leaves,
 }
 
 //
-static btDbvtVolume				bounds(	btDbvtNode** leaves,
-									   int count)
+static btDbvtVolume bounds(btDbvtNode** leaves,
+						   int count)
 {
-#if DBVT_MERGE_IMPL==DBVT_IMPL_SSE
-	ATTRIBUTE_ALIGNED16(char	locals[sizeof(btDbvtVolume)]);
-	btDbvtVolume* ptr = (btDbvtVolume*) locals;
-	btDbvtVolume&	volume=*ptr;
-	volume=leaves[0]->volume;
+#if DBVT_MERGE_IMPL == DBVT_IMPL_SSE
+	ATTRIBUTE_ALIGNED16(char locals[sizeof(btDbvtVolume)]);
+	btDbvtVolume* ptr = (btDbvtVolume*)locals;
+	btDbvtVolume& volume = *ptr;
+	volume = leaves[0]->volume;
 #else
-	btDbvtVolume volume=leaves[0]->volume;
+	btDbvtVolume volume = leaves[0]->volume;
 #endif
-	for(int i=1,ni=count;i<ni;++i)
+	for (int i = 1, ni = count; i < ni; ++i)
 	{
-		Merge(volume,leaves[i]->volume,volume);
+		Merge(volume, leaves[i]->volume, volume);
 	}
-	return(volume);
+	return (volume);
 }
 
 //
-static void						bottomup(	btDbvt* pdbvt,
-										 btDbvtNode** leaves,
-										 int count)
+static void bottomup(btDbvt* pdbvt,
+					 btDbvtNode** leaves,
+					 int count)
 {
-	while(count>1)
+	while (count > 1)
 	{
-		btScalar	minsize=SIMD_INFINITY;
-		int			minidx[2]={-1,-1};
-		for(int i=0;i<count;++i)
+		btScalar minsize = SIMD_INFINITY;
+		int minidx[2] = {-1, -1};
+		for (int i = 0; i < count; ++i)
 		{
-			for(int j=i+1;j<count;++j)
+			for (int j = i + 1; j < count; ++j)
 			{
-				const btScalar	sz=size(merge(leaves[i]->volume,leaves[j]->volume));
-				if(sz<minsize)
+				const btScalar sz = size(merge(leaves[i]->volume, leaves[j]->volume));
+				if (sz < minsize)
 				{
-					minsize		=	sz;
-					minidx[0]	=	i;
-					minidx[1]	=	j;
+					minsize = sz;
+					minidx[0] = i;
+					minidx[1] = j;
 				}
 			}
 		}
-		btDbvtNode*	n[]	=	{leaves[minidx[0]],leaves[minidx[1]]};
-		btDbvtNode*	p	=	createnode(pdbvt,0,n[0]->volume,n[1]->volume,0);
-		p->childs[0]		=	n[0];
-		p->childs[1]		=	n[1];
-		n[0]->parent		=	p;
-		n[1]->parent		=	p;
-		leaves[minidx[0]]	=	p;
-		leaves[minidx[1]]	=	leaves[count-1];
+		btDbvtNode* n[] = {leaves[minidx[0]], leaves[minidx[1]]};
+		btDbvtNode* p = createnode(pdbvt, 0, n[0]->volume, n[1]->volume, 0);
+		p->childs[0] = n[0];
+		p->childs[1] = n[1];
+		n[0]->parent = p;
+		n[1]->parent = p;
+		leaves[minidx[0]] = p;
+		leaves[minidx[1]] = leaves[count - 1];
 		--count;
 	}
 }
 
 //
-static btDbvtNode*			topdown(btDbvt* pdbvt,
-									btDbvtNode** leaves,
-									int count,
-									int bu_treshold)
+static btDbvtNode* topdown(btDbvt* pdbvt,
+						   btDbvtNode** leaves,
+						   int count,
+						   int bu_treshold)
 {
-	static const btVector3	axis[]={btVector3(1,0,0),
-		btVector3(0,1,0),
-		btVector3(0,0,1)};
-	btAssert(bu_treshold>2);
-	if(count>1)
+	static const btVector3 axis[] = {btVector3(1, 0, 0),
+									 btVector3(0, 1, 0),
+									 btVector3(0, 0, 1)};
+	btAssert(bu_treshold > 2);
+	if (count > 1)
 	{
-		if(count>bu_treshold)
+		if (count > bu_treshold)
 		{
-			const btDbvtVolume	vol=bounds(leaves,count);
-			const btVector3			org=vol.Center();
-			int						partition;
-			int						bestaxis=-1;
-			int						bestmidp=count;
-			int						splitcount[3][2]={{0,0},{0,0},{0,0}};
+			const btDbvtVolume vol = bounds(leaves, count);
+			const btVector3 org = vol.Center();
+			int partition;
+			int bestaxis = -1;
+			int bestmidp = count;
+			int splitcount[3][2] = {{0, 0}, {0, 0}, {0, 0}};
 			int i;
-			for( i=0;i<count;++i)
+			for (i = 0; i < count; ++i)
 			{
-				const btVector3	x=leaves[i]->volume.Center()-org;
-				for(int j=0;j<3;++j)
+				const btVector3 x = leaves[i]->volume.Center() - org;
+				for (int j = 0; j < 3; ++j)
 				{
-					++splitcount[j][btDot(x,axis[j])>0?1:0];
+					++splitcount[j][btDot(x, axis[j]) > 0 ? 1 : 0];
 				}
 			}
-			for( i=0;i<3;++i)
+			for (i = 0; i < 3; ++i)
 			{
-				if((splitcount[i][0]>0)&&(splitcount[i][1]>0))
+				if ((splitcount[i][0] > 0) && (splitcount[i][1] > 0))
 				{
-					const int	midp=(int)btFabs(btScalar(splitcount[i][0]-splitcount[i][1]));
-					if(midp<bestmidp)
+					const int midp = (int)btFabs(btScalar(splitcount[i][0] - splitcount[i][1]));
+					if (midp < bestmidp)
 					{
-						bestaxis=i;
-						bestmidp=midp;
+						bestaxis = i;
+						bestmidp = midp;
 					}
 				}
 			}
-			if(bestaxis>=0)
+			if (bestaxis >= 0)
 			{
-				partition=split(leaves,count,org,axis[bestaxis]);
-				btAssert(partition!=0 && partition!=count);
+				partition = split(leaves, count, org, axis[bestaxis]);
+				btAssert(partition != 0 && partition != count);
 			}
 			else
 			{
-				partition=count/2+1;
+				partition = count / 2 + 1;
 			}
-			btDbvtNode*	node=createnode(pdbvt,0,vol,0);
-			node->childs[0]=topdown(pdbvt,&leaves[0],partition,bu_treshold);
-			node->childs[1]=topdown(pdbvt,&leaves[partition],count-partition,bu_treshold);
-			node->childs[0]->parent=node;
-			node->childs[1]->parent=node;
-			return(node);
+			btDbvtNode* node = createnode(pdbvt, 0, vol, 0);
+			node->childs[0] = topdown(pdbvt, &leaves[0], partition, bu_treshold);
+			node->childs[1] = topdown(pdbvt, &leaves[partition], count - partition, bu_treshold);
+			node->childs[0]->parent = node;
+			node->childs[1]->parent = node;
+			return (node);
 		}
 		else
 		{
-			bottomup(pdbvt,leaves,count);
-			return(leaves[0]);
+			bottomup(pdbvt, leaves, count);
+			return (leaves[0]);
 		}
 	}
-	return(leaves[0]);
+	return (leaves[0]);
 }
 
 //
-static DBVT_INLINE btDbvtNode*	sort(btDbvtNode* n,btDbvtNode*& r)
+static DBVT_INLINE btDbvtNode* sort(btDbvtNode* n, btDbvtNode*& r)
 {
-	btDbvtNode*	p=n->parent;
+	btDbvtNode* p = n->parent;
 	btAssert(n->isinternal());
-	if(p>n)
+	if (p > n)
 	{
-		const int		i=indexof(n);
-		const int		j=1-i;
-		btDbvtNode*	s=p->childs[j];
-		btDbvtNode*	q=p->parent;
-		btAssert(n==p->childs[i]);
-		if(q) q->childs[indexof(p)]=n; else r=n;
-		s->parent=n;
-		p->parent=n;
-		n->parent=q;
-		p->childs[0]=n->childs[0];
-		p->childs[1]=n->childs[1];
-		n->childs[0]->parent=p;
-		n->childs[1]->parent=p;
-		n->childs[i]=p;
-		n->childs[j]=s;
-		btSwap(p->volume,n->volume);
-		return(p);
+		const int i = indexof(n);
+		const int j = 1 - i;
+		btDbvtNode* s = p->childs[j];
+		btDbvtNode* q = p->parent;
+		btAssert(n == p->childs[i]);
+		if (q)
+			q->childs[indexof(p)] = n;
+		else
+			r = n;
+		s->parent = n;
+		p->parent = n;
+		n->parent = q;
+		p->childs[0] = n->childs[0];
+		p->childs[1] = n->childs[1];
+		n->childs[0]->parent = p;
+		n->childs[1]->parent = p;
+		n->childs[i] = p;
+		n->childs[j] = s;
+		btSwap(p->volume, n->volume);
+		return (p);
 	}
-	return(n);
+	return (n);
 }
 
 #if 0
@@ -442,11 +459,11 @@ static DBVT_INLINE btDbvtNode*	walkup(btDbvtNode* n,int count)
 //
 btDbvt::btDbvt()
 {
-	m_root		=	0;
-	m_free		=	0;
-	m_lkhd		=	-1;
-	m_leaves	=	0;
-	m_opath		=	0;
+	m_root = 0;
+	m_free = 0;
+	m_lkhd = -1;
+	m_leaves = 0;
+	m_opath = 0;
 }
 
 //
@@ -456,228 +473,233 @@ btDbvt::~btDbvt()
 }
 
 //
-void			btDbvt::clear()
+void btDbvt::clear()
 {
-	if(m_root)	
-		recursedeletenode(this,m_root);
+	if (m_root)
+		recursedeletenode(this, m_root);
 	btAlignedFree(m_free);
-	m_free=0;
-	m_lkhd		=	-1;
+	m_free = 0;
+	m_lkhd = -1;
 	m_stkStack.clear();
-	m_opath		=	0;
-	
+	m_opath = 0;
 }
 
 //
-void			btDbvt::optimizeBottomUp()
+void btDbvt::optimizeBottomUp()
 {
-	if(m_root)
+	if (m_root)
 	{
 		tNodeArray leaves;
 		leaves.reserve(m_leaves);
-		fetchleaves(this,m_root,leaves);
-		bottomup(this,&leaves[0],leaves.size());
-		m_root=leaves[0];
+		fetchleaves(this, m_root, leaves);
+		bottomup(this, &leaves[0], leaves.size());
+		m_root = leaves[0];
 	}
 }
 
 //
-void			btDbvt::optimizeTopDown(int bu_treshold)
+void btDbvt::optimizeTopDown(int bu_treshold)
 {
-	if(m_root)
+	if (m_root)
 	{
-		tNodeArray	leaves;
+		tNodeArray leaves;
 		leaves.reserve(m_leaves);
-		fetchleaves(this,m_root,leaves);
-		m_root=topdown(this,&leaves[0],leaves.size(),bu_treshold);
+		fetchleaves(this, m_root, leaves);
+		m_root = topdown(this, &leaves[0], leaves.size(), bu_treshold);
 	}
 }
 
 //
-void			btDbvt::optimizeIncremental(int passes)
+void btDbvt::optimizeIncremental(int passes)
 {
-	if(passes<0) passes=m_leaves;
-	if(m_root&&(passes>0))
+	if (passes < 0) passes = m_leaves;
+	if (m_root && (passes > 0))
 	{
-		do	{
-			btDbvtNode*		node=m_root;
-			unsigned	bit=0;
-			while(node->isinternal())
+		do
+		{
+			btDbvtNode* node = m_root;
+			unsigned bit = 0;
+			while (node->isinternal())
 			{
-				node=sort(node,m_root)->childs[(m_opath>>bit)&1];
-				bit=(bit+1)&(sizeof(unsigned)*8-1);
+				node = sort(node, m_root)->childs[(m_opath >> bit) & 1];
+				bit = (bit + 1) & (sizeof(unsigned) * 8 - 1);
 			}
 			update(node);
 			++m_opath;
-		} while(--passes);
+		} while (--passes);
 	}
 }
 
 //
-btDbvtNode*	btDbvt::insert(const btDbvtVolume& volume,void* data)
+btDbvtNode* btDbvt::insert(const btDbvtVolume& volume, void* data)
 {
-	btDbvtNode*	leaf=createnode(this,0,volume,data);
-	insertleaf(this,m_root,leaf);
+	btDbvtNode* leaf = createnode(this, 0, volume, data);
+	insertleaf(this, m_root, leaf);
 	++m_leaves;
-	return(leaf);
+	return (leaf);
 }
 
 //
-void			btDbvt::update(btDbvtNode* leaf,int lookahead)
+void btDbvt::update(btDbvtNode* leaf, int lookahead)
 {
-	btDbvtNode*	root=removeleaf(this,leaf);
-	if(root)
+	btDbvtNode* root = removeleaf(this, leaf);
+	if (root)
 	{
-		if(lookahead>=0)
+		if (lookahead >= 0)
 		{
-			for(int i=0;(i<lookahead)&&root->parent;++i)
+			for (int i = 0; (i < lookahead) && root->parent; ++i)
 			{
-				root=root->parent;
+				root = root->parent;
 			}
-		} else root=m_root;
+		}
+		else
+			root = m_root;
 	}
-	insertleaf(this,root,leaf);
+	insertleaf(this, root, leaf);
 }
 
 //
-void			btDbvt::update(btDbvtNode* leaf,btDbvtVolume& volume)
+void btDbvt::update(btDbvtNode* leaf, btDbvtVolume& volume)
 {
-	btDbvtNode*	root=removeleaf(this,leaf);
-	if(root)
+	btDbvtNode* root = removeleaf(this, leaf);
+	if (root)
 	{
-		if(m_lkhd>=0)
+		if (m_lkhd >= 0)
 		{
-			for(int i=0;(i<m_lkhd)&&root->parent;++i)
+			for (int i = 0; (i < m_lkhd) && root->parent; ++i)
 			{
-				root=root->parent;
+				root = root->parent;
 			}
-		} else root=m_root;
+		}
+		else
+			root = m_root;
 	}
-	leaf->volume=volume;
-	insertleaf(this,root,leaf);
+	leaf->volume = volume;
+	insertleaf(this, root, leaf);
 }
 
 //
-bool			btDbvt::update(btDbvtNode* leaf,btDbvtVolume& volume,const btVector3& velocity,btScalar margin)
+bool btDbvt::update(btDbvtNode* leaf, btDbvtVolume& volume, const btVector3& velocity, btScalar margin)
 {
-	if(leaf->volume.Contain(volume)) return(false);
-	volume.Expand(btVector3(margin,margin,margin));
+	if (leaf->volume.Contain(volume)) return (false);
+	volume.Expand(btVector3(margin, margin, margin));
 	volume.SignedExpand(velocity);
-	update(leaf,volume);
-	return(true);
+	update(leaf, volume);
+	return (true);
 }
 
 //
-bool			btDbvt::update(btDbvtNode* leaf,btDbvtVolume& volume,const btVector3& velocity)
+bool btDbvt::update(btDbvtNode* leaf, btDbvtVolume& volume, const btVector3& velocity)
 {
-	if(leaf->volume.Contain(volume)) return(false);
+	if (leaf->volume.Contain(volume)) return (false);
 	volume.SignedExpand(velocity);
-	update(leaf,volume);
-	return(true);
+	update(leaf, volume);
+	return (true);
 }
 
 //
-bool			btDbvt::update(btDbvtNode* leaf,btDbvtVolume& volume,btScalar margin)
+bool btDbvt::update(btDbvtNode* leaf, btDbvtVolume& volume, btScalar margin)
 {
-	if(leaf->volume.Contain(volume)) return(false);
-	volume.Expand(btVector3(margin,margin,margin));
-	update(leaf,volume);
-	return(true);
+	if (leaf->volume.Contain(volume)) return (false);
+	volume.Expand(btVector3(margin, margin, margin));
+	update(leaf, volume);
+	return (true);
 }
 
 //
-void			btDbvt::remove(btDbvtNode* leaf)
+void btDbvt::remove(btDbvtNode* leaf)
 {
-	removeleaf(this,leaf);
-	deletenode(this,leaf);
+	removeleaf(this, leaf);
+	deletenode(this, leaf);
 	--m_leaves;
 }
 
 //
-void			btDbvt::write(IWriter* iwriter) const
+void btDbvt::write(IWriter* iwriter) const
 {
-	btDbvtNodeEnumerator	nodes;
-	nodes.nodes.reserve(m_leaves*2);
-	enumNodes(m_root,nodes);
-	iwriter->Prepare(m_root,nodes.nodes.size());
-	for(int i=0;i<nodes.nodes.size();++i)
+	btDbvtNodeEnumerator nodes;
+	nodes.nodes.reserve(m_leaves * 2);
+	enumNodes(m_root, nodes);
+	iwriter->Prepare(m_root, nodes.nodes.size());
+	for (int i = 0; i < nodes.nodes.size(); ++i)
 	{
-		const btDbvtNode* n=nodes.nodes[i];
-		int			p=-1;
-		if(n->parent) p=nodes.nodes.findLinearSearch(n->parent);
-		if(n->isinternal())
+		const btDbvtNode* n = nodes.nodes[i];
+		int p = -1;
+		if (n->parent) p = nodes.nodes.findLinearSearch(n->parent);
+		if (n->isinternal())
 		{
-			const int	c0=nodes.nodes.findLinearSearch(n->childs[0]);
-			const int	c1=nodes.nodes.findLinearSearch(n->childs[1]);
-			iwriter->WriteNode(n,i,p,c0,c1);
+			const int c0 = nodes.nodes.findLinearSearch(n->childs[0]);
+			const int c1 = nodes.nodes.findLinearSearch(n->childs[1]);
+			iwriter->WriteNode(n, i, p, c0, c1);
 		}
 		else
 		{
-			iwriter->WriteLeaf(n,i,p);
-		}	
+			iwriter->WriteLeaf(n, i, p);
+		}
 	}
 }
 
 //
-void			btDbvt::clone(btDbvt& dest,IClone* iclone) const
+void btDbvt::clone(btDbvt& dest, IClone* iclone) const
 {
 	dest.clear();
-	if(m_root!=0)
-	{	
-		btAlignedObjectArray<sStkCLN>	stack;
+	if (m_root != 0)
+	{
+		btAlignedObjectArray<sStkCLN> stack;
 		stack.reserve(m_leaves);
-		stack.push_back(sStkCLN(m_root,0));
-		do	{
-			const int		i=stack.size()-1;
-			const sStkCLN	e=stack[i];
-			btDbvtNode*			n=createnode(&dest,e.parent,e.node->volume,e.node->data);
+		stack.push_back(sStkCLN(m_root, 0));
+		do
+		{
+			const int i = stack.size() - 1;
+			const sStkCLN e = stack[i];
+			btDbvtNode* n = createnode(&dest, e.parent, e.node->volume, e.node->data);
 			stack.pop_back();
-			if(e.parent!=0)
-				e.parent->childs[i&1]=n;
+			if (e.parent != 0)
+				e.parent->childs[i & 1] = n;
 			else
-				dest.m_root=n;
-			if(e.node->isinternal())
+				dest.m_root = n;
+			if (e.node->isinternal())
 			{
-				stack.push_back(sStkCLN(e.node->childs[0],n));
-				stack.push_back(sStkCLN(e.node->childs[1],n));
+				stack.push_back(sStkCLN(e.node->childs[0], n));
+				stack.push_back(sStkCLN(e.node->childs[1], n));
 			}
 			else
 			{
 				iclone->CloneLeaf(n);
 			}
-		} while(stack.size()>0);
+		} while (stack.size() > 0);
 	}
 }
 
 //
-int				btDbvt::maxdepth(const btDbvtNode* node)
+int btDbvt::maxdepth(const btDbvtNode* node)
 {
-	int	depth=0;
-	if(node) getmaxdepth(node,1,depth);
-	return(depth);
+	int depth = 0;
+	if (node) getmaxdepth(node, 1, depth);
+	return (depth);
 }
 
 //
-int				btDbvt::countLeaves(const btDbvtNode* node)
+int btDbvt::countLeaves(const btDbvtNode* node)
 {
-	if(node->isinternal())
-		return(countLeaves(node->childs[0])+countLeaves(node->childs[1]));
+	if (node->isinternal())
+		return (countLeaves(node->childs[0]) + countLeaves(node->childs[1]));
 	else
-		return(1);
+		return (1);
 }
 
 //
-void			btDbvt::extractLeaves(const btDbvtNode* node,btAlignedObjectArray<const btDbvtNode*>& leaves)
+void btDbvt::extractLeaves(const btDbvtNode* node, btAlignedObjectArray<const btDbvtNode*>& leaves)
 {
-	if(node->isinternal())
+	if (node->isinternal())
 	{
-		extractLeaves(node->childs[0],leaves);
-		extractLeaves(node->childs[1],leaves);
+		extractLeaves(node->childs[0], leaves);
+		extractLeaves(node->childs[1], leaves);
 	}
 	else
 	{
 		leaves.push_back(node);
-	}	
+	}
 }
 
 //
@@ -726,603 +748,608 @@ struct btDbvtBenchmark
 {
 	struct NilPolicy : btDbvt::ICollide
 	{
-		NilPolicy() : m_pcount(0),m_depth(-SIMD_INFINITY),m_checksort(true)		{}
-		void	Process(const btDbvtNode*,const btDbvtNode*)				{ ++m_pcount; }
-		void	Process(const btDbvtNode*)									{ ++m_pcount; }
-		void	Process(const btDbvtNode*,btScalar depth)
+		NilPolicy() : m_pcount(0), m_depth(-SIMD_INFINITY), m_checksort(true) {}
+		void Process(const btDbvtNode*, const btDbvtNode*) { ++m_pcount; }
+		void Process(const btDbvtNode*) { ++m_pcount; }
+		void Process(const btDbvtNode*, btScalar depth)
 		{
 			++m_pcount;
-			if(m_checksort)
-			{ if(depth>=m_depth) m_depth=depth; else printf("wrong depth: %f (should be >= %f)\r\n",depth,m_depth); }
+			if (m_checksort)
+			{
+				if (depth >= m_depth)
+					m_depth = depth;
+				else
+					printf("wrong depth: %f (should be >= %f)\r\n", depth, m_depth);
+			}
 		}
-		int			m_pcount;
-		btScalar	m_depth;
-		bool		m_checksort;
+		int m_pcount;
+		btScalar m_depth;
+		bool m_checksort;
 	};
 	struct P14 : btDbvt::ICollide
 	{
 		struct Node
 		{
-			const btDbvtNode*	leaf;
-			btScalar			depth;
+			const btDbvtNode* leaf;
+			btScalar depth;
 		};
-		void Process(const btDbvtNode* leaf,btScalar depth)
+		void Process(const btDbvtNode* leaf, btScalar depth)
 		{
-			Node	n;
-			n.leaf	=	leaf;
-			n.depth	=	depth;
+			Node n;
+			n.leaf = leaf;
+			n.depth = depth;
 		}
-		static int sortfnc(const Node& a,const Node& b)
+		static int sortfnc(const Node& a, const Node& b)
 		{
-			if(a.depth<b.depth) return(+1);
-			if(a.depth>b.depth) return(-1);
-			return(0);
+			if (a.depth < b.depth) return (+1);
+			if (a.depth > b.depth) return (-1);
+			return (0);
 		}
-		btAlignedObjectArray<Node>		m_nodes;
+		btAlignedObjectArray<Node> m_nodes;
 	};
 	struct P15 : btDbvt::ICollide
 	{
 		struct Node
 		{
-			const btDbvtNode*	leaf;
-			btScalar			depth;
+			const btDbvtNode* leaf;
+			btScalar depth;
 		};
 		void Process(const btDbvtNode* leaf)
 		{
-			Node	n;
-			n.leaf	=	leaf;
-			n.depth	=	dot(leaf->volume.Center(),m_axis);
+			Node n;
+			n.leaf = leaf;
+			n.depth = dot(leaf->volume.Center(), m_axis);
 		}
-		static int sortfnc(const Node& a,const Node& b)
+		static int sortfnc(const Node& a, const Node& b)
 		{
-			if(a.depth<b.depth) return(+1);
-			if(a.depth>b.depth) return(-1);
-			return(0);
+			if (a.depth < b.depth) return (+1);
+			if (a.depth > b.depth) return (-1);
+			return (0);
 		}
-		btAlignedObjectArray<Node>		m_nodes;
-		btVector3						m_axis;
+		btAlignedObjectArray<Node> m_nodes;
+		btVector3 m_axis;
 	};
-	static btScalar			RandUnit()
+	static btScalar RandUnit()
 	{
-		return(rand()/(btScalar)RAND_MAX);
+		return (rand() / (btScalar)RAND_MAX);
 	}
-	static btVector3		RandVector3()
+	static btVector3 RandVector3()
 	{
-		return(btVector3(RandUnit(),RandUnit(),RandUnit()));
+		return (btVector3(RandUnit(), RandUnit(), RandUnit()));
 	}
-	static btVector3		RandVector3(btScalar cs)
+	static btVector3 RandVector3(btScalar cs)
 	{
-		return(RandVector3()*cs-btVector3(cs,cs,cs)/2);
+		return (RandVector3() * cs - btVector3(cs, cs, cs) / 2);
 	}
-	static btDbvtVolume	RandVolume(btScalar cs,btScalar eb,btScalar es)
+	static btDbvtVolume RandVolume(btScalar cs, btScalar eb, btScalar es)
 	{
-		return(btDbvtVolume::FromCE(RandVector3(cs),btVector3(eb,eb,eb)+RandVector3()*es));
+		return (btDbvtVolume::FromCE(RandVector3(cs), btVector3(eb, eb, eb) + RandVector3() * es));
 	}
-	static btTransform		RandTransform(btScalar cs)
+	static btTransform RandTransform(btScalar cs)
 	{
-		btTransform	t;
+		btTransform t;
 		t.setOrigin(RandVector3(cs));
-		t.setRotation(btQuaternion(RandUnit()*SIMD_PI*2,RandUnit()*SIMD_PI*2,RandUnit()*SIMD_PI*2).normalized());
-		return(t);
+		t.setRotation(btQuaternion(RandUnit() * SIMD_PI * 2, RandUnit() * SIMD_PI * 2, RandUnit() * SIMD_PI * 2).normalized());
+		return (t);
 	}
-	static void				RandTree(btScalar cs,btScalar eb,btScalar es,int leaves,btDbvt& dbvt)
+	static void RandTree(btScalar cs, btScalar eb, btScalar es, int leaves, btDbvt& dbvt)
 	{
 		dbvt.clear();
-		for(int i=0;i<leaves;++i)
+		for (int i = 0; i < leaves; ++i)
 		{
-			dbvt.insert(RandVolume(cs,eb,es),0);
+			dbvt.insert(RandVolume(cs, eb, es), 0);
 		}
 	}
 };
 
-void			btDbvt::benchmark()
+void btDbvt::benchmark()
 {
-	static const btScalar	cfgVolumeCenterScale		=	100;
-	static const btScalar	cfgVolumeExentsBase			=	1;
-	static const btScalar	cfgVolumeExentsScale		=	4;
-	static const int		cfgLeaves					=	8192;
-	static const bool		cfgEnable					=	true;
+	static const btScalar cfgVolumeCenterScale = 100;
+	static const btScalar cfgVolumeExentsBase = 1;
+	static const btScalar cfgVolumeExentsScale = 4;
+	static const int cfgLeaves = 8192;
+	static const bool cfgEnable = true;
 
 	//[1] btDbvtVolume intersections
-	bool					cfgBenchmark1_Enable		=	cfgEnable;
-	static const int		cfgBenchmark1_Iterations	=	8;
-	static const int		cfgBenchmark1_Reference		=	3499;
+	bool cfgBenchmark1_Enable = cfgEnable;
+	static const int cfgBenchmark1_Iterations = 8;
+	static const int cfgBenchmark1_Reference = 3499;
 	//[2] btDbvtVolume merges
-	bool					cfgBenchmark2_Enable		=	cfgEnable;
-	static const int		cfgBenchmark2_Iterations	=	4;
-	static const int		cfgBenchmark2_Reference		=	1945;
+	bool cfgBenchmark2_Enable = cfgEnable;
+	static const int cfgBenchmark2_Iterations = 4;
+	static const int cfgBenchmark2_Reference = 1945;
 	//[3] btDbvt::collideTT
-	bool					cfgBenchmark3_Enable		=	cfgEnable;
-	static const int		cfgBenchmark3_Iterations	=	512;
-	static const int		cfgBenchmark3_Reference		=	5485;
+	bool cfgBenchmark3_Enable = cfgEnable;
+	static const int cfgBenchmark3_Iterations = 512;
+	static const int cfgBenchmark3_Reference = 5485;
 	//[4] btDbvt::collideTT self
-	bool					cfgBenchmark4_Enable		=	cfgEnable;
-	static const int		cfgBenchmark4_Iterations	=	512;
-	static const int		cfgBenchmark4_Reference		=	2814;
+	bool cfgBenchmark4_Enable = cfgEnable;
+	static const int cfgBenchmark4_Iterations = 512;
+	static const int cfgBenchmark4_Reference = 2814;
 	//[5] btDbvt::collideTT xform
-	bool					cfgBenchmark5_Enable		=	cfgEnable;
-	static const int		cfgBenchmark5_Iterations	=	512;
-	static const btScalar	cfgBenchmark5_OffsetScale	=	2;
-	static const int		cfgBenchmark5_Reference		=	7379;
+	bool cfgBenchmark5_Enable = cfgEnable;
+	static const int cfgBenchmark5_Iterations = 512;
+	static const btScalar cfgBenchmark5_OffsetScale = 2;
+	static const int cfgBenchmark5_Reference = 7379;
 	//[6] btDbvt::collideTT xform,self
-	bool					cfgBenchmark6_Enable		=	cfgEnable;
-	static const int		cfgBenchmark6_Iterations	=	512;
-	static const btScalar	cfgBenchmark6_OffsetScale	=	2;
-	static const int		cfgBenchmark6_Reference		=	7270;
+	bool cfgBenchmark6_Enable = cfgEnable;
+	static const int cfgBenchmark6_Iterations = 512;
+	static const btScalar cfgBenchmark6_OffsetScale = 2;
+	static const int cfgBenchmark6_Reference = 7270;
 	//[7] btDbvt::rayTest
-	bool					cfgBenchmark7_Enable		=	cfgEnable;
-	static const int		cfgBenchmark7_Passes		=	32;
-	static const int		cfgBenchmark7_Iterations	=	65536;
-	static const int		cfgBenchmark7_Reference		=	6307;
+	bool cfgBenchmark7_Enable = cfgEnable;
+	static const int cfgBenchmark7_Passes = 32;
+	static const int cfgBenchmark7_Iterations = 65536;
+	static const int cfgBenchmark7_Reference = 6307;
 	//[8] insert/remove
-	bool					cfgBenchmark8_Enable		=	cfgEnable;
-	static const int		cfgBenchmark8_Passes		=	32;
-	static const int		cfgBenchmark8_Iterations	=	65536;
-	static const int		cfgBenchmark8_Reference		=	2105;
+	bool cfgBenchmark8_Enable = cfgEnable;
+	static const int cfgBenchmark8_Passes = 32;
+	static const int cfgBenchmark8_Iterations = 65536;
+	static const int cfgBenchmark8_Reference = 2105;
 	//[9] updates (teleport)
-	bool					cfgBenchmark9_Enable		=	cfgEnable;
-	static const int		cfgBenchmark9_Passes		=	32;
-	static const int		cfgBenchmark9_Iterations	=	65536;
-	static const int		cfgBenchmark9_Reference		=	1879;
+	bool cfgBenchmark9_Enable = cfgEnable;
+	static const int cfgBenchmark9_Passes = 32;
+	static const int cfgBenchmark9_Iterations = 65536;
+	static const int cfgBenchmark9_Reference = 1879;
 	//[10] updates (jitter)
-	bool					cfgBenchmark10_Enable		=	cfgEnable;
-	static const btScalar	cfgBenchmark10_Scale		=	cfgVolumeCenterScale/10000;
-	static const int		cfgBenchmark10_Passes		=	32;
-	static const int		cfgBenchmark10_Iterations	=	65536;
-	static const int		cfgBenchmark10_Reference	=	1244;
+	bool cfgBenchmark10_Enable = cfgEnable;
+	static const btScalar cfgBenchmark10_Scale = cfgVolumeCenterScale / 10000;
+	static const int cfgBenchmark10_Passes = 32;
+	static const int cfgBenchmark10_Iterations = 65536;
+	static const int cfgBenchmark10_Reference = 1244;
 	//[11] optimize (incremental)
-	bool					cfgBenchmark11_Enable		=	cfgEnable;
-	static const int		cfgBenchmark11_Passes		=	64;
-	static const int		cfgBenchmark11_Iterations	=	65536;
-	static const int		cfgBenchmark11_Reference	=	2510;
+	bool cfgBenchmark11_Enable = cfgEnable;
+	static const int cfgBenchmark11_Passes = 64;
+	static const int cfgBenchmark11_Iterations = 65536;
+	static const int cfgBenchmark11_Reference = 2510;
 	//[12] btDbvtVolume notequal
-	bool					cfgBenchmark12_Enable		=	cfgEnable;
-	static const int		cfgBenchmark12_Iterations	=	32;
-	static const int		cfgBenchmark12_Reference	=	3677;
+	bool cfgBenchmark12_Enable = cfgEnable;
+	static const int cfgBenchmark12_Iterations = 32;
+	static const int cfgBenchmark12_Reference = 3677;
 	//[13] culling(OCL+fullsort)
-	bool					cfgBenchmark13_Enable		=	cfgEnable;
-	static const int		cfgBenchmark13_Iterations	=	1024;
-	static const int		cfgBenchmark13_Reference	=	2231;
+	bool cfgBenchmark13_Enable = cfgEnable;
+	static const int cfgBenchmark13_Iterations = 1024;
+	static const int cfgBenchmark13_Reference = 2231;
 	//[14] culling(OCL+qsort)
-	bool					cfgBenchmark14_Enable		=	cfgEnable;
-	static const int		cfgBenchmark14_Iterations	=	8192;
-	static const int		cfgBenchmark14_Reference	=	3500;
+	bool cfgBenchmark14_Enable = cfgEnable;
+	static const int cfgBenchmark14_Iterations = 8192;
+	static const int cfgBenchmark14_Reference = 3500;
 	//[15] culling(KDOP+qsort)
-	bool					cfgBenchmark15_Enable		=	cfgEnable;
-	static const int		cfgBenchmark15_Iterations	=	8192;
-	static const int		cfgBenchmark15_Reference	=	1151;
+	bool cfgBenchmark15_Enable = cfgEnable;
+	static const int cfgBenchmark15_Iterations = 8192;
+	static const int cfgBenchmark15_Reference = 1151;
 	//[16] insert/remove batch
-	bool					cfgBenchmark16_Enable		=	cfgEnable;
-	static const int		cfgBenchmark16_BatchCount	=	256;
-	static const int		cfgBenchmark16_Passes		=	16384;
-	static const int		cfgBenchmark16_Reference	=	5138;
+	bool cfgBenchmark16_Enable = cfgEnable;
+	static const int cfgBenchmark16_BatchCount = 256;
+	static const int cfgBenchmark16_Passes = 16384;
+	static const int cfgBenchmark16_Reference = 5138;
 	//[17] select
-	bool					cfgBenchmark17_Enable		=	cfgEnable;
-	static const int		cfgBenchmark17_Iterations	=	4;
-	static const int		cfgBenchmark17_Reference	=	3390;
+	bool cfgBenchmark17_Enable = cfgEnable;
+	static const int cfgBenchmark17_Iterations = 4;
+	static const int cfgBenchmark17_Reference = 3390;
 
-	btClock					wallclock;
+	btClock wallclock;
 	printf("Benchmarking dbvt...\r\n");
-	printf("\tWorld scale: %f\r\n",cfgVolumeCenterScale);
-	printf("\tExtents base: %f\r\n",cfgVolumeExentsBase);
-	printf("\tExtents range: %f\r\n",cfgVolumeExentsScale);
-	printf("\tLeaves: %u\r\n",cfgLeaves);
-	printf("\tsizeof(btDbvtVolume): %u bytes\r\n",sizeof(btDbvtVolume));
-	printf("\tsizeof(btDbvtNode):   %u bytes\r\n",sizeof(btDbvtNode));
-	if(cfgBenchmark1_Enable)
-	{// Benchmark 1	
+	printf("\tWorld scale: %f\r\n", cfgVolumeCenterScale);
+	printf("\tExtents base: %f\r\n", cfgVolumeExentsBase);
+	printf("\tExtents range: %f\r\n", cfgVolumeExentsScale);
+	printf("\tLeaves: %u\r\n", cfgLeaves);
+	printf("\tsizeof(btDbvtVolume): %u bytes\r\n", sizeof(btDbvtVolume));
+	printf("\tsizeof(btDbvtNode):   %u bytes\r\n", sizeof(btDbvtNode));
+	if (cfgBenchmark1_Enable)
+	{  // Benchmark 1
 		srand(380843);
-		btAlignedObjectArray<btDbvtVolume>	volumes;
-		btAlignedObjectArray<bool>			results;
+		btAlignedObjectArray<btDbvtVolume> volumes;
+		btAlignedObjectArray<bool> results;
 		volumes.resize(cfgLeaves);
 		results.resize(cfgLeaves);
-		for(int i=0;i<cfgLeaves;++i)
+		for (int i = 0; i < cfgLeaves; ++i)
 		{
-			volumes[i]=btDbvtBenchmark::RandVolume(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale);
+			volumes[i] = btDbvtBenchmark::RandVolume(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale);
 		}
 		printf("[1] btDbvtVolume intersections: ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark1_Iterations;++i)
+		for (int i = 0; i < cfgBenchmark1_Iterations; ++i)
 		{
-			for(int j=0;j<cfgLeaves;++j)
+			for (int j = 0; j < cfgLeaves; ++j)
 			{
-				for(int k=0;k<cfgLeaves;++k)
+				for (int k = 0; k < cfgLeaves; ++k)
 				{
-					results[k]=Intersect(volumes[j],volumes[k]);
+					results[k] = Intersect(volumes[j], volumes[k]);
 				}
 			}
 		}
-		const int time=(int)wallclock.getTimeMilliseconds();
-		printf("%u ms (%i%%)\r\n",time,(time-cfgBenchmark1_Reference)*100/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		printf("%u ms (%i%%)\r\n", time, (time - cfgBenchmark1_Reference) * 100 / time);
 	}
-	if(cfgBenchmark2_Enable)
-	{// Benchmark 2	
+	if (cfgBenchmark2_Enable)
+	{  // Benchmark 2
 		srand(380843);
-		btAlignedObjectArray<btDbvtVolume>	volumes;
-		btAlignedObjectArray<btDbvtVolume>	results;
+		btAlignedObjectArray<btDbvtVolume> volumes;
+		btAlignedObjectArray<btDbvtVolume> results;
 		volumes.resize(cfgLeaves);
 		results.resize(cfgLeaves);
-		for(int i=0;i<cfgLeaves;++i)
+		for (int i = 0; i < cfgLeaves; ++i)
 		{
-			volumes[i]=btDbvtBenchmark::RandVolume(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale);
+			volumes[i] = btDbvtBenchmark::RandVolume(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale);
 		}
 		printf("[2] btDbvtVolume merges: ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark2_Iterations;++i)
+		for (int i = 0; i < cfgBenchmark2_Iterations; ++i)
 		{
-			for(int j=0;j<cfgLeaves;++j)
+			for (int j = 0; j < cfgLeaves; ++j)
 			{
-				for(int k=0;k<cfgLeaves;++k)
+				for (int k = 0; k < cfgLeaves; ++k)
 				{
-					Merge(volumes[j],volumes[k],results[k]);
+					Merge(volumes[j], volumes[k], results[k]);
 				}
 			}
 		}
-		const int time=(int)wallclock.getTimeMilliseconds();
-		printf("%u ms (%i%%)\r\n",time,(time-cfgBenchmark2_Reference)*100/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		printf("%u ms (%i%%)\r\n", time, (time - cfgBenchmark2_Reference) * 100 / time);
 	}
-	if(cfgBenchmark3_Enable)
-	{// Benchmark 3	
+	if (cfgBenchmark3_Enable)
+	{  // Benchmark 3
 		srand(380843);
-		btDbvt						dbvt[2];
-		btDbvtBenchmark::NilPolicy	policy;
-		btDbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt[0]);
-		btDbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt[1]);
+		btDbvt dbvt[2];
+		btDbvtBenchmark::NilPolicy policy;
+		btDbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt[0]);
+		btDbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt[1]);
 		dbvt[0].optimizeTopDown();
 		dbvt[1].optimizeTopDown();
 		printf("[3] btDbvt::collideTT: ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark3_Iterations;++i)
+		for (int i = 0; i < cfgBenchmark3_Iterations; ++i)
 		{
-			btDbvt::collideTT(dbvt[0].m_root,dbvt[1].m_root,policy);
+			btDbvt::collideTT(dbvt[0].m_root, dbvt[1].m_root, policy);
 		}
-		const int time=(int)wallclock.getTimeMilliseconds();
-		printf("%u ms (%i%%)\r\n",time,(time-cfgBenchmark3_Reference)*100/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		printf("%u ms (%i%%)\r\n", time, (time - cfgBenchmark3_Reference) * 100 / time);
 	}
-	if(cfgBenchmark4_Enable)
-	{// Benchmark 4
+	if (cfgBenchmark4_Enable)
+	{  // Benchmark 4
 		srand(380843);
-		btDbvt						dbvt;
-		btDbvtBenchmark::NilPolicy	policy;
-		btDbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		btDbvt dbvt;
+		btDbvtBenchmark::NilPolicy policy;
+		btDbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
 		printf("[4] btDbvt::collideTT self: ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark4_Iterations;++i)
+		for (int i = 0; i < cfgBenchmark4_Iterations; ++i)
 		{
-			btDbvt::collideTT(dbvt.m_root,dbvt.m_root,policy);
+			btDbvt::collideTT(dbvt.m_root, dbvt.m_root, policy);
 		}
-		const int time=(int)wallclock.getTimeMilliseconds();
-		printf("%u ms (%i%%)\r\n",time,(time-cfgBenchmark4_Reference)*100/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		printf("%u ms (%i%%)\r\n", time, (time - cfgBenchmark4_Reference) * 100 / time);
 	}
-	if(cfgBenchmark5_Enable)
-	{// Benchmark 5	
+	if (cfgBenchmark5_Enable)
+	{  // Benchmark 5
 		srand(380843);
-		btDbvt								dbvt[2];
-		btAlignedObjectArray<btTransform>	transforms;
-		btDbvtBenchmark::NilPolicy			policy;
+		btDbvt dbvt[2];
+		btAlignedObjectArray<btTransform> transforms;
+		btDbvtBenchmark::NilPolicy policy;
 		transforms.resize(cfgBenchmark5_Iterations);
-		for(int i=0;i<transforms.size();++i)
+		for (int i = 0; i < transforms.size(); ++i)
 		{
-			transforms[i]=btDbvtBenchmark::RandTransform(cfgVolumeCenterScale*cfgBenchmark5_OffsetScale);
+			transforms[i] = btDbvtBenchmark::RandTransform(cfgVolumeCenterScale * cfgBenchmark5_OffsetScale);
 		}
-		btDbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt[0]);
-		btDbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt[1]);
+		btDbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt[0]);
+		btDbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt[1]);
 		dbvt[0].optimizeTopDown();
 		dbvt[1].optimizeTopDown();
 		printf("[5] btDbvt::collideTT xform: ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark5_Iterations;++i)
+		for (int i = 0; i < cfgBenchmark5_Iterations; ++i)
 		{
-			btDbvt::collideTT(dbvt[0].m_root,dbvt[1].m_root,transforms[i],policy);
+			btDbvt::collideTT(dbvt[0].m_root, dbvt[1].m_root, transforms[i], policy);
 		}
-		const int time=(int)wallclock.getTimeMilliseconds();
-		printf("%u ms (%i%%)\r\n",time,(time-cfgBenchmark5_Reference)*100/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		printf("%u ms (%i%%)\r\n", time, (time - cfgBenchmark5_Reference) * 100 / time);
 	}
-	if(cfgBenchmark6_Enable)
-	{// Benchmark 6	
+	if (cfgBenchmark6_Enable)
+	{  // Benchmark 6
 		srand(380843);
-		btDbvt								dbvt;
-		btAlignedObjectArray<btTransform>	transforms;
-		btDbvtBenchmark::NilPolicy			policy;
+		btDbvt dbvt;
+		btAlignedObjectArray<btTransform> transforms;
+		btDbvtBenchmark::NilPolicy policy;
 		transforms.resize(cfgBenchmark6_Iterations);
-		for(int i=0;i<transforms.size();++i)
+		for (int i = 0; i < transforms.size(); ++i)
 		{
-			transforms[i]=btDbvtBenchmark::RandTransform(cfgVolumeCenterScale*cfgBenchmark6_OffsetScale);
+			transforms[i] = btDbvtBenchmark::RandTransform(cfgVolumeCenterScale * cfgBenchmark6_OffsetScale);
 		}
-		btDbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		btDbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
 		printf("[6] btDbvt::collideTT xform,self: ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark6_Iterations;++i)
+		for (int i = 0; i < cfgBenchmark6_Iterations; ++i)
 		{
-			btDbvt::collideTT(dbvt.m_root,dbvt.m_root,transforms[i],policy);		
+			btDbvt::collideTT(dbvt.m_root, dbvt.m_root, transforms[i], policy);
 		}
-		const int time=(int)wallclock.getTimeMilliseconds();
-		printf("%u ms (%i%%)\r\n",time,(time-cfgBenchmark6_Reference)*100/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		printf("%u ms (%i%%)\r\n", time, (time - cfgBenchmark6_Reference) * 100 / time);
 	}
-	if(cfgBenchmark7_Enable)
-	{// Benchmark 7	
+	if (cfgBenchmark7_Enable)
+	{  // Benchmark 7
 		srand(380843);
-		btDbvt								dbvt;
-		btAlignedObjectArray<btVector3>		rayorg;
-		btAlignedObjectArray<btVector3>		raydir;
-		btDbvtBenchmark::NilPolicy			policy;
+		btDbvt dbvt;
+		btAlignedObjectArray<btVector3> rayorg;
+		btAlignedObjectArray<btVector3> raydir;
+		btDbvtBenchmark::NilPolicy policy;
 		rayorg.resize(cfgBenchmark7_Iterations);
 		raydir.resize(cfgBenchmark7_Iterations);
-		for(int i=0;i<rayorg.size();++i)
+		for (int i = 0; i < rayorg.size(); ++i)
 		{
-			rayorg[i]=btDbvtBenchmark::RandVector3(cfgVolumeCenterScale*2);
-			raydir[i]=btDbvtBenchmark::RandVector3(cfgVolumeCenterScale*2);
+			rayorg[i] = btDbvtBenchmark::RandVector3(cfgVolumeCenterScale * 2);
+			raydir[i] = btDbvtBenchmark::RandVector3(cfgVolumeCenterScale * 2);
 		}
-		btDbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		btDbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
 		printf("[7] btDbvt::rayTest: ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark7_Passes;++i)
+		for (int i = 0; i < cfgBenchmark7_Passes; ++i)
 		{
-			for(int j=0;j<cfgBenchmark7_Iterations;++j)
+			for (int j = 0; j < cfgBenchmark7_Iterations; ++j)
 			{
-				btDbvt::rayTest(dbvt.m_root,rayorg[j],rayorg[j]+raydir[j],policy);
+				btDbvt::rayTest(dbvt.m_root, rayorg[j], rayorg[j] + raydir[j], policy);
 			}
 		}
-		const int	time=(int)wallclock.getTimeMilliseconds();
-		unsigned	rays=cfgBenchmark7_Passes*cfgBenchmark7_Iterations;
-		printf("%u ms (%i%%),(%u r/s)\r\n",time,(time-cfgBenchmark7_Reference)*100/time,(rays*1000)/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		unsigned rays = cfgBenchmark7_Passes * cfgBenchmark7_Iterations;
+		printf("%u ms (%i%%),(%u r/s)\r\n", time, (time - cfgBenchmark7_Reference) * 100 / time, (rays * 1000) / time);
 	}
-	if(cfgBenchmark8_Enable)
-	{// Benchmark 8	
+	if (cfgBenchmark8_Enable)
+	{  // Benchmark 8
 		srand(380843);
-		btDbvt								dbvt;
-		btDbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		btDbvt dbvt;
+		btDbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
 		printf("[8] insert/remove: ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark8_Passes;++i)
+		for (int i = 0; i < cfgBenchmark8_Passes; ++i)
 		{
-			for(int j=0;j<cfgBenchmark8_Iterations;++j)
+			for (int j = 0; j < cfgBenchmark8_Iterations; ++j)
 			{
-				dbvt.remove(dbvt.insert(btDbvtBenchmark::RandVolume(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale),0));
+				dbvt.remove(dbvt.insert(btDbvtBenchmark::RandVolume(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale), 0));
 			}
 		}
-		const int	time=(int)wallclock.getTimeMilliseconds();
-		const int	ir=cfgBenchmark8_Passes*cfgBenchmark8_Iterations;
-		printf("%u ms (%i%%),(%u ir/s)\r\n",time,(time-cfgBenchmark8_Reference)*100/time,ir*1000/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		const int ir = cfgBenchmark8_Passes * cfgBenchmark8_Iterations;
+		printf("%u ms (%i%%),(%u ir/s)\r\n", time, (time - cfgBenchmark8_Reference) * 100 / time, ir * 1000 / time);
 	}
-	if(cfgBenchmark9_Enable)
-	{// Benchmark 9	
+	if (cfgBenchmark9_Enable)
+	{  // Benchmark 9
 		srand(380843);
-		btDbvt										dbvt;
-		btAlignedObjectArray<const btDbvtNode*>	leaves;
-		btDbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		btDbvt dbvt;
+		btAlignedObjectArray<const btDbvtNode*> leaves;
+		btDbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
-		dbvt.extractLeaves(dbvt.m_root,leaves);
+		dbvt.extractLeaves(dbvt.m_root, leaves);
 		printf("[9] updates (teleport): ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark9_Passes;++i)
+		for (int i = 0; i < cfgBenchmark9_Passes; ++i)
 		{
-			for(int j=0;j<cfgBenchmark9_Iterations;++j)
+			for (int j = 0; j < cfgBenchmark9_Iterations; ++j)
 			{
-				dbvt.update(const_cast<btDbvtNode*>(leaves[rand()%cfgLeaves]),
-					btDbvtBenchmark::RandVolume(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale));
+				dbvt.update(const_cast<btDbvtNode*>(leaves[rand() % cfgLeaves]),
+							btDbvtBenchmark::RandVolume(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale));
 			}
 		}
-		const int	time=(int)wallclock.getTimeMilliseconds();
-		const int	up=cfgBenchmark9_Passes*cfgBenchmark9_Iterations;
-		printf("%u ms (%i%%),(%u u/s)\r\n",time,(time-cfgBenchmark9_Reference)*100/time,up*1000/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		const int up = cfgBenchmark9_Passes * cfgBenchmark9_Iterations;
+		printf("%u ms (%i%%),(%u u/s)\r\n", time, (time - cfgBenchmark9_Reference) * 100 / time, up * 1000 / time);
 	}
-	if(cfgBenchmark10_Enable)
-	{// Benchmark 10	
+	if (cfgBenchmark10_Enable)
+	{  // Benchmark 10
 		srand(380843);
-		btDbvt										dbvt;
-		btAlignedObjectArray<const btDbvtNode*>	leaves;
-		btAlignedObjectArray<btVector3>				vectors;
+		btDbvt dbvt;
+		btAlignedObjectArray<const btDbvtNode*> leaves;
+		btAlignedObjectArray<btVector3> vectors;
 		vectors.resize(cfgBenchmark10_Iterations);
-		for(int i=0;i<vectors.size();++i)
+		for (int i = 0; i < vectors.size(); ++i)
 		{
-			vectors[i]=(btDbvtBenchmark::RandVector3()*2-btVector3(1,1,1))*cfgBenchmark10_Scale;
+			vectors[i] = (btDbvtBenchmark::RandVector3() * 2 - btVector3(1, 1, 1)) * cfgBenchmark10_Scale;
 		}
-		btDbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		btDbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
-		dbvt.extractLeaves(dbvt.m_root,leaves);
+		dbvt.extractLeaves(dbvt.m_root, leaves);
 		printf("[10] updates (jitter): ");
 		wallclock.reset();
 
-		for(int i=0;i<cfgBenchmark10_Passes;++i)
+		for (int i = 0; i < cfgBenchmark10_Passes; ++i)
 		{
-			for(int j=0;j<cfgBenchmark10_Iterations;++j)
-			{			
-				const btVector3&	d=vectors[j];
-				btDbvtNode*		l=const_cast<btDbvtNode*>(leaves[rand()%cfgLeaves]);
-				btDbvtVolume		v=btDbvtVolume::FromMM(l->volume.Mins()+d,l->volume.Maxs()+d);
-				dbvt.update(l,v);
+			for (int j = 0; j < cfgBenchmark10_Iterations; ++j)
+			{
+				const btVector3& d = vectors[j];
+				btDbvtNode* l = const_cast<btDbvtNode*>(leaves[rand() % cfgLeaves]);
+				btDbvtVolume v = btDbvtVolume::FromMM(l->volume.Mins() + d, l->volume.Maxs() + d);
+				dbvt.update(l, v);
 			}
 		}
-		const int	time=(int)wallclock.getTimeMilliseconds();
-		const int	up=cfgBenchmark10_Passes*cfgBenchmark10_Iterations;
-		printf("%u ms (%i%%),(%u u/s)\r\n",time,(time-cfgBenchmark10_Reference)*100/time,up*1000/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		const int up = cfgBenchmark10_Passes * cfgBenchmark10_Iterations;
+		printf("%u ms (%i%%),(%u u/s)\r\n", time, (time - cfgBenchmark10_Reference) * 100 / time, up * 1000 / time);
 	}
-	if(cfgBenchmark11_Enable)
-	{// Benchmark 11	
+	if (cfgBenchmark11_Enable)
+	{  // Benchmark 11
 		srand(380843);
-		btDbvt										dbvt;
-		btDbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		btDbvt dbvt;
+		btDbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
 		printf("[11] optimize (incremental): ");
-		wallclock.reset();	
-		for(int i=0;i<cfgBenchmark11_Passes;++i)
+		wallclock.reset();
+		for (int i = 0; i < cfgBenchmark11_Passes; ++i)
 		{
 			dbvt.optimizeIncremental(cfgBenchmark11_Iterations);
 		}
-		const int	time=(int)wallclock.getTimeMilliseconds();
-		const int	op=cfgBenchmark11_Passes*cfgBenchmark11_Iterations;
-		printf("%u ms (%i%%),(%u o/s)\r\n",time,(time-cfgBenchmark11_Reference)*100/time,op/time*1000);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		const int op = cfgBenchmark11_Passes * cfgBenchmark11_Iterations;
+		printf("%u ms (%i%%),(%u o/s)\r\n", time, (time - cfgBenchmark11_Reference) * 100 / time, op / time * 1000);
 	}
-	if(cfgBenchmark12_Enable)
-	{// Benchmark 12	
+	if (cfgBenchmark12_Enable)
+	{  // Benchmark 12
 		srand(380843);
-		btAlignedObjectArray<btDbvtVolume>	volumes;
-		btAlignedObjectArray<bool>				results;
+		btAlignedObjectArray<btDbvtVolume> volumes;
+		btAlignedObjectArray<bool> results;
 		volumes.resize(cfgLeaves);
 		results.resize(cfgLeaves);
-		for(int i=0;i<cfgLeaves;++i)
+		for (int i = 0; i < cfgLeaves; ++i)
 		{
-			volumes[i]=btDbvtBenchmark::RandVolume(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale);
+			volumes[i] = btDbvtBenchmark::RandVolume(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale);
 		}
 		printf("[12] btDbvtVolume notequal: ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark12_Iterations;++i)
+		for (int i = 0; i < cfgBenchmark12_Iterations; ++i)
 		{
-			for(int j=0;j<cfgLeaves;++j)
+			for (int j = 0; j < cfgLeaves; ++j)
 			{
-				for(int k=0;k<cfgLeaves;++k)
+				for (int k = 0; k < cfgLeaves; ++k)
 				{
-					results[k]=NotEqual(volumes[j],volumes[k]);
+					results[k] = NotEqual(volumes[j], volumes[k]);
 				}
 			}
 		}
-		const int time=(int)wallclock.getTimeMilliseconds();
-		printf("%u ms (%i%%)\r\n",time,(time-cfgBenchmark12_Reference)*100/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		printf("%u ms (%i%%)\r\n", time, (time - cfgBenchmark12_Reference) * 100 / time);
 	}
-	if(cfgBenchmark13_Enable)
-	{// Benchmark 13	
+	if (cfgBenchmark13_Enable)
+	{  // Benchmark 13
 		srand(380843);
-		btDbvt								dbvt;
-		btAlignedObjectArray<btVector3>		vectors;
-		btDbvtBenchmark::NilPolicy			policy;
+		btDbvt dbvt;
+		btAlignedObjectArray<btVector3> vectors;
+		btDbvtBenchmark::NilPolicy policy;
 		vectors.resize(cfgBenchmark13_Iterations);
-		for(int i=0;i<vectors.size();++i)
+		for (int i = 0; i < vectors.size(); ++i)
 		{
-			vectors[i]=(btDbvtBenchmark::RandVector3()*2-btVector3(1,1,1)).normalized();
+			vectors[i] = (btDbvtBenchmark::RandVector3() * 2 - btVector3(1, 1, 1)).normalized();
 		}
-		btDbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		btDbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
 		printf("[13] culling(OCL+fullsort): ");
-		wallclock.reset();	
-		for(int i=0;i<cfgBenchmark13_Iterations;++i)
+		wallclock.reset();
+		for (int i = 0; i < cfgBenchmark13_Iterations; ++i)
 		{
-			static const btScalar	offset=0;
-			policy.m_depth=-SIMD_INFINITY;
-			dbvt.collideOCL(dbvt.m_root,&vectors[i],&offset,vectors[i],1,policy);
+			static const btScalar offset = 0;
+			policy.m_depth = -SIMD_INFINITY;
+			dbvt.collideOCL(dbvt.m_root, &vectors[i], &offset, vectors[i], 1, policy);
 		}
-		const int	time=(int)wallclock.getTimeMilliseconds();
-		const int	t=cfgBenchmark13_Iterations;
-		printf("%u ms (%i%%),(%u t/s)\r\n",time,(time-cfgBenchmark13_Reference)*100/time,(t*1000)/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		const int t = cfgBenchmark13_Iterations;
+		printf("%u ms (%i%%),(%u t/s)\r\n", time, (time - cfgBenchmark13_Reference) * 100 / time, (t * 1000) / time);
 	}
-	if(cfgBenchmark14_Enable)
-	{// Benchmark 14	
+	if (cfgBenchmark14_Enable)
+	{  // Benchmark 14
 		srand(380843);
-		btDbvt								dbvt;
-		btAlignedObjectArray<btVector3>		vectors;
-		btDbvtBenchmark::P14				policy;
+		btDbvt dbvt;
+		btAlignedObjectArray<btVector3> vectors;
+		btDbvtBenchmark::P14 policy;
 		vectors.resize(cfgBenchmark14_Iterations);
-		for(int i=0;i<vectors.size();++i)
+		for (int i = 0; i < vectors.size(); ++i)
 		{
-			vectors[i]=(btDbvtBenchmark::RandVector3()*2-btVector3(1,1,1)).normalized();
+			vectors[i] = (btDbvtBenchmark::RandVector3() * 2 - btVector3(1, 1, 1)).normalized();
 		}
-		btDbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		btDbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
 		policy.m_nodes.reserve(cfgLeaves);
 		printf("[14] culling(OCL+qsort): ");
-		wallclock.reset();	
-		for(int i=0;i<cfgBenchmark14_Iterations;++i)
+		wallclock.reset();
+		for (int i = 0; i < cfgBenchmark14_Iterations; ++i)
 		{
-			static const btScalar	offset=0;
+			static const btScalar offset = 0;
 			policy.m_nodes.resize(0);
-			dbvt.collideOCL(dbvt.m_root,&vectors[i],&offset,vectors[i],1,policy,false);
+			dbvt.collideOCL(dbvt.m_root, &vectors[i], &offset, vectors[i], 1, policy, false);
 			policy.m_nodes.quickSort(btDbvtBenchmark::P14::sortfnc);
 		}
-		const int	time=(int)wallclock.getTimeMilliseconds();
-		const int	t=cfgBenchmark14_Iterations;
-		printf("%u ms (%i%%),(%u t/s)\r\n",time,(time-cfgBenchmark14_Reference)*100/time,(t*1000)/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		const int t = cfgBenchmark14_Iterations;
+		printf("%u ms (%i%%),(%u t/s)\r\n", time, (time - cfgBenchmark14_Reference) * 100 / time, (t * 1000) / time);
 	}
-	if(cfgBenchmark15_Enable)
-	{// Benchmark 15	
+	if (cfgBenchmark15_Enable)
+	{  // Benchmark 15
 		srand(380843);
-		btDbvt								dbvt;
-		btAlignedObjectArray<btVector3>		vectors;
-		btDbvtBenchmark::P15				policy;
+		btDbvt dbvt;
+		btAlignedObjectArray<btVector3> vectors;
+		btDbvtBenchmark::P15 policy;
 		vectors.resize(cfgBenchmark15_Iterations);
-		for(int i=0;i<vectors.size();++i)
+		for (int i = 0; i < vectors.size(); ++i)
 		{
-			vectors[i]=(btDbvtBenchmark::RandVector3()*2-btVector3(1,1,1)).normalized();
+			vectors[i] = (btDbvtBenchmark::RandVector3() * 2 - btVector3(1, 1, 1)).normalized();
 		}
-		btDbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		btDbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
 		policy.m_nodes.reserve(cfgLeaves);
 		printf("[15] culling(KDOP+qsort): ");
-		wallclock.reset();	
-		for(int i=0;i<cfgBenchmark15_Iterations;++i)
+		wallclock.reset();
+		for (int i = 0; i < cfgBenchmark15_Iterations; ++i)
 		{
-			static const btScalar	offset=0;
+			static const btScalar offset = 0;
 			policy.m_nodes.resize(0);
-			policy.m_axis=vectors[i];
-			dbvt.collideKDOP(dbvt.m_root,&vectors[i],&offset,1,policy);
+			policy.m_axis = vectors[i];
+			dbvt.collideKDOP(dbvt.m_root, &vectors[i], &offset, 1, policy);
 			policy.m_nodes.quickSort(btDbvtBenchmark::P15::sortfnc);
 		}
-		const int	time=(int)wallclock.getTimeMilliseconds();
-		const int	t=cfgBenchmark15_Iterations;
-		printf("%u ms (%i%%),(%u t/s)\r\n",time,(time-cfgBenchmark15_Reference)*100/time,(t*1000)/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		const int t = cfgBenchmark15_Iterations;
+		printf("%u ms (%i%%),(%u t/s)\r\n", time, (time - cfgBenchmark15_Reference) * 100 / time, (t * 1000) / time);
 	}
-	if(cfgBenchmark16_Enable)
-	{// Benchmark 16	
+	if (cfgBenchmark16_Enable)
+	{  // Benchmark 16
 		srand(380843);
-		btDbvt								dbvt;
-		btAlignedObjectArray<btDbvtNode*>	batch;
-		btDbvtBenchmark::RandTree(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale,cfgLeaves,dbvt);
+		btDbvt dbvt;
+		btAlignedObjectArray<btDbvtNode*> batch;
+		btDbvtBenchmark::RandTree(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale, cfgLeaves, dbvt);
 		dbvt.optimizeTopDown();
 		batch.reserve(cfgBenchmark16_BatchCount);
-		printf("[16] insert/remove batch(%u): ",cfgBenchmark16_BatchCount);
+		printf("[16] insert/remove batch(%u): ", cfgBenchmark16_BatchCount);
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark16_Passes;++i)
+		for (int i = 0; i < cfgBenchmark16_Passes; ++i)
 		{
-			for(int j=0;j<cfgBenchmark16_BatchCount;++j)
+			for (int j = 0; j < cfgBenchmark16_BatchCount; ++j)
 			{
-				batch.push_back(dbvt.insert(btDbvtBenchmark::RandVolume(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale),0));
+				batch.push_back(dbvt.insert(btDbvtBenchmark::RandVolume(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale), 0));
 			}
-			for(int j=0;j<cfgBenchmark16_BatchCount;++j)
+			for (int j = 0; j < cfgBenchmark16_BatchCount; ++j)
 			{
 				dbvt.remove(batch[j]);
 			}
 			batch.resize(0);
 		}
-		const int	time=(int)wallclock.getTimeMilliseconds();
-		const int	ir=cfgBenchmark16_Passes*cfgBenchmark16_BatchCount;
-		printf("%u ms (%i%%),(%u bir/s)\r\n",time,(time-cfgBenchmark16_Reference)*100/time,int(ir*1000.0/time));
+		const int time = (int)wallclock.getTimeMilliseconds();
+		const int ir = cfgBenchmark16_Passes * cfgBenchmark16_BatchCount;
+		printf("%u ms (%i%%),(%u bir/s)\r\n", time, (time - cfgBenchmark16_Reference) * 100 / time, int(ir * 1000.0 / time));
 	}
-	if(cfgBenchmark17_Enable)
-	{// Benchmark 17
+	if (cfgBenchmark17_Enable)
+	{  // Benchmark 17
 		srand(380843);
-		btAlignedObjectArray<btDbvtVolume>	volumes;
-		btAlignedObjectArray<int>			results;
-		btAlignedObjectArray<int>			indices;
+		btAlignedObjectArray<btDbvtVolume> volumes;
+		btAlignedObjectArray<int> results;
+		btAlignedObjectArray<int> indices;
 		volumes.resize(cfgLeaves);
 		results.resize(cfgLeaves);
 		indices.resize(cfgLeaves);
-		for(int i=0;i<cfgLeaves;++i)
+		for (int i = 0; i < cfgLeaves; ++i)
 		{
-			indices[i]=i;
-			volumes[i]=btDbvtBenchmark::RandVolume(cfgVolumeCenterScale,cfgVolumeExentsBase,cfgVolumeExentsScale);
+			indices[i] = i;
+			volumes[i] = btDbvtBenchmark::RandVolume(cfgVolumeCenterScale, cfgVolumeExentsBase, cfgVolumeExentsScale);
 		}
-		for(int i=0;i<cfgLeaves;++i)
+		for (int i = 0; i < cfgLeaves; ++i)
 		{
-			btSwap(indices[i],indices[rand()%cfgLeaves]);
+			btSwap(indices[i], indices[rand() % cfgLeaves]);
 		}
 		printf("[17] btDbvtVolume select: ");
 		wallclock.reset();
-		for(int i=0;i<cfgBenchmark17_Iterations;++i)
+		for (int i = 0; i < cfgBenchmark17_Iterations; ++i)
 		{
-			for(int j=0;j<cfgLeaves;++j)
+			for (int j = 0; j < cfgLeaves; ++j)
 			{
-				for(int k=0;k<cfgLeaves;++k)
+				for (int k = 0; k < cfgLeaves; ++k)
 				{
-					const int idx=indices[k];
-					results[idx]=Select(volumes[idx],volumes[j],volumes[k]);
+					const int idx = indices[k];
+					results[idx] = Select(volumes[idx], volumes[j], volumes[k]);
 				}
 			}
 		}
-		const int time=(int)wallclock.getTimeMilliseconds();
-		printf("%u ms (%i%%)\r\n",time,(time-cfgBenchmark17_Reference)*100/time);
+		const int time = (int)wallclock.getTimeMilliseconds();
+		printf("%u ms (%i%%)\r\n", time, (time - cfgBenchmark17_Reference) * 100 / time);
 	}
 	printf("\r\n\r\n");
 }
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDbvt.h b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDbvt.h
index b5a0014580d..a316dbf2071 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDbvt.h
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDbvt.h
@@ -26,50 +26,49 @@ subject to the following restrictions:
 // Compile time configuration
 //
 
-
 // Implementation profiles
-#define DBVT_IMPL_GENERIC		0	// Generic implementation	
-#define DBVT_IMPL_SSE			1	// SSE
+#define DBVT_IMPL_GENERIC 0  // Generic implementation
+#define DBVT_IMPL_SSE 1      // SSE
 
 // Template implementation of ICollide
 #ifdef _WIN32
-#if (defined (_MSC_VER) && _MSC_VER >= 1400)
-#define	DBVT_USE_TEMPLATE		1
+#if (defined(_MSC_VER) && _MSC_VER >= 1400)
+#define DBVT_USE_TEMPLATE 1
 #else
-#define	DBVT_USE_TEMPLATE		0
+#define DBVT_USE_TEMPLATE 0
 #endif
 #else
-#define	DBVT_USE_TEMPLATE		0
+#define DBVT_USE_TEMPLATE 0
 #endif
 
 // Use only intrinsics instead of inline asm
-#define DBVT_USE_INTRINSIC_SSE	1
+#define DBVT_USE_INTRINSIC_SSE 1
 
 // Using memmov for collideOCL
-#define DBVT_USE_MEMMOVE		1
+#define DBVT_USE_MEMMOVE 1
 
 // Enable benchmarking code
-#define	DBVT_ENABLE_BENCHMARK	0
+#define DBVT_ENABLE_BENCHMARK 0
 
 // Inlining
-#define DBVT_INLINE				SIMD_FORCE_INLINE
+#define DBVT_INLINE SIMD_FORCE_INLINE
 
 // Specific methods implementation
 
 //SSE gives errors on a MSVC 7.1
-#if defined (BT_USE_SSE) //&& defined (_WIN32)
-#define DBVT_SELECT_IMPL		DBVT_IMPL_SSE
-#define DBVT_MERGE_IMPL			DBVT_IMPL_SSE
-#define DBVT_INT0_IMPL			DBVT_IMPL_SSE
+#if defined(BT_USE_SSE)  //&& defined (_WIN32)
+#define DBVT_SELECT_IMPL DBVT_IMPL_SSE
+#define DBVT_MERGE_IMPL DBVT_IMPL_SSE
+#define DBVT_INT0_IMPL DBVT_IMPL_SSE
 #else
-#define DBVT_SELECT_IMPL		DBVT_IMPL_GENERIC
-#define DBVT_MERGE_IMPL			DBVT_IMPL_GENERIC
-#define DBVT_INT0_IMPL			DBVT_IMPL_GENERIC
+#define DBVT_SELECT_IMPL DBVT_IMPL_GENERIC
+#define DBVT_MERGE_IMPL DBVT_IMPL_GENERIC
+#define DBVT_INT0_IMPL DBVT_IMPL_GENERIC
 #endif
 
-#if	(DBVT_SELECT_IMPL==DBVT_IMPL_SSE)||	\
-	(DBVT_MERGE_IMPL==DBVT_IMPL_SSE)||	\
-	(DBVT_INT0_IMPL==DBVT_IMPL_SSE)
+#if (DBVT_SELECT_IMPL == DBVT_IMPL_SSE) || \
+	(DBVT_MERGE_IMPL == DBVT_IMPL_SSE) ||  \
+	(DBVT_INT0_IMPL == DBVT_IMPL_SSE)
 #include <emmintrin.h>
 #endif
 
@@ -78,21 +77,24 @@ subject to the following restrictions:
 //
 
 #if DBVT_USE_TEMPLATE
-#define	DBVT_VIRTUAL
+#define DBVT_VIRTUAL
 #define DBVT_VIRTUAL_DTOR(a)
-#define DBVT_PREFIX					template <typename T>
-#define DBVT_IPOLICY				T& policy
-#define DBVT_CHECKTYPE				static const ICollide&	typechecker=*(T*)1;(void)typechecker;
+#define DBVT_PREFIX template <typename T>
+#define DBVT_IPOLICY T& policy
+#define DBVT_CHECKTYPE                           \
+	static const ICollide& typechecker = *(T*)1; \
+	(void)typechecker;
 #else
-#define	DBVT_VIRTUAL_DTOR(a)		virtual ~a() {}
-#define DBVT_VIRTUAL				virtual
+#define DBVT_VIRTUAL_DTOR(a) \
+	virtual ~a() {}
+#define DBVT_VIRTUAL virtual
 #define DBVT_PREFIX
-#define DBVT_IPOLICY				ICollide& policy
+#define DBVT_IPOLICY ICollide& policy
 #define DBVT_CHECKTYPE
 #endif
 
 #if DBVT_USE_MEMMOVE
-#if !defined( __CELLOS_LV2__) && !defined(__MWERKS__)
+#if !defined(__CELLOS_LV2__) && !defined(__MWERKS__)
 #include <memory.h>
 #endif
 #include <string.h>
@@ -122,194 +124,193 @@ subject to the following restrictions:
 #error "DBVT_INT0_IMPL undefined"
 #endif
 
-
 //
 // Defaults volumes
 //
 
-/* btDbvtAabbMm			*/ 
-struct	btDbvtAabbMm
+/* btDbvtAabbMm			*/
+struct btDbvtAabbMm
 {
-	DBVT_INLINE btVector3			Center() const	{ return((mi+mx)/2); }
-	DBVT_INLINE btVector3			Lengths() const	{ return(mx-mi); }
-	DBVT_INLINE btVector3			Extents() const	{ return((mx-mi)/2); }
-	DBVT_INLINE const btVector3&	Mins() const	{ return(mi); }
-	DBVT_INLINE const btVector3&	Maxs() const	{ return(mx); }
-	static inline btDbvtAabbMm		FromCE(const btVector3& c,const btVector3& e);
-	static inline btDbvtAabbMm		FromCR(const btVector3& c,btScalar r);
-	static inline btDbvtAabbMm		FromMM(const btVector3& mi,const btVector3& mx);
-	static inline btDbvtAabbMm		FromPoints(const btVector3* pts,int n);
-	static inline btDbvtAabbMm		FromPoints(const btVector3** ppts,int n);
-	DBVT_INLINE void				Expand(const btVector3& e);
-	DBVT_INLINE void				SignedExpand(const btVector3& e);
-	DBVT_INLINE bool				Contain(const btDbvtAabbMm& a) const;
-	DBVT_INLINE int					Classify(const btVector3& n,btScalar o,int s) const;
-	DBVT_INLINE btScalar			ProjectMinimum(const btVector3& v,unsigned signs) const;
-	DBVT_INLINE friend bool			Intersect(	const btDbvtAabbMm& a,
-		const btDbvtAabbMm& b);
-	
-	DBVT_INLINE friend bool			Intersect(	const btDbvtAabbMm& a,
-		const btVector3& b);
+	DBVT_INLINE btVector3 Center() const { return ((mi + mx) / 2); }
+	DBVT_INLINE btVector3 Lengths() const { return (mx - mi); }
+	DBVT_INLINE btVector3 Extents() const { return ((mx - mi) / 2); }
+	DBVT_INLINE const btVector3& Mins() const { return (mi); }
+	DBVT_INLINE const btVector3& Maxs() const { return (mx); }
+	static inline btDbvtAabbMm FromCE(const btVector3& c, const btVector3& e);
+	static inline btDbvtAabbMm FromCR(const btVector3& c, btScalar r);
+	static inline btDbvtAabbMm FromMM(const btVector3& mi, const btVector3& mx);
+	static inline btDbvtAabbMm FromPoints(const btVector3* pts, int n);
+	static inline btDbvtAabbMm FromPoints(const btVector3** ppts, int n);
+	DBVT_INLINE void Expand(const btVector3& e);
+	DBVT_INLINE void SignedExpand(const btVector3& e);
+	DBVT_INLINE bool Contain(const btDbvtAabbMm& a) const;
+	DBVT_INLINE int Classify(const btVector3& n, btScalar o, int s) const;
+	DBVT_INLINE btScalar ProjectMinimum(const btVector3& v, unsigned signs) const;
+	DBVT_INLINE friend bool Intersect(const btDbvtAabbMm& a,
+									  const btDbvtAabbMm& b);
+
+	DBVT_INLINE friend bool Intersect(const btDbvtAabbMm& a,
+									  const btVector3& b);
+
+	DBVT_INLINE friend btScalar Proximity(const btDbvtAabbMm& a,
+										  const btDbvtAabbMm& b);
+	DBVT_INLINE friend int Select(const btDbvtAabbMm& o,
+								  const btDbvtAabbMm& a,
+								  const btDbvtAabbMm& b);
+	DBVT_INLINE friend void Merge(const btDbvtAabbMm& a,
+								  const btDbvtAabbMm& b,
+								  btDbvtAabbMm& r);
+	DBVT_INLINE friend bool NotEqual(const btDbvtAabbMm& a,
+									 const btDbvtAabbMm& b);
+
+	DBVT_INLINE btVector3& tMins() { return (mi); }
+	DBVT_INLINE btVector3& tMaxs() { return (mx); }
 
-	DBVT_INLINE friend btScalar		Proximity(	const btDbvtAabbMm& a,
-		const btDbvtAabbMm& b);
-	DBVT_INLINE friend int			Select(		const btDbvtAabbMm& o,
-		const btDbvtAabbMm& a,
-		const btDbvtAabbMm& b);
-	DBVT_INLINE friend void			Merge(		const btDbvtAabbMm& a,
-		const btDbvtAabbMm& b,
-		btDbvtAabbMm& r);
-	DBVT_INLINE friend bool			NotEqual(	const btDbvtAabbMm& a,
-		const btDbvtAabbMm& b);
-    
-    DBVT_INLINE btVector3&	tMins()	{ return(mi); }
-	DBVT_INLINE btVector3&	tMaxs()	{ return(mx); }
-    
 private:
-	DBVT_INLINE void				AddSpan(const btVector3& d,btScalar& smi,btScalar& smx) const;
+	DBVT_INLINE void AddSpan(const btVector3& d, btScalar& smi, btScalar& smx) const;
+
 private:
-	btVector3	mi,mx;
+	btVector3 mi, mx;
 };
 
-// Types	
-typedef	btDbvtAabbMm	btDbvtVolume;
+// Types
+typedef btDbvtAabbMm btDbvtVolume;
 
-/* btDbvtNode				*/ 
-struct	btDbvtNode
+/* btDbvtNode				*/
+struct btDbvtNode
 {
-	btDbvtVolume	volume;
-	btDbvtNode*		parent;
-	DBVT_INLINE bool	isleaf() const		{ return(childs[1]==0); }
-	DBVT_INLINE bool	isinternal() const	{ return(!isleaf()); }
-	union
-	{
-		btDbvtNode*	childs[2];
-		void*	data;
-		int		dataAsInt;
+	btDbvtVolume volume;
+	btDbvtNode* parent;
+	DBVT_INLINE bool isleaf() const { return (childs[1] == 0); }
+	DBVT_INLINE bool isinternal() const { return (!isleaf()); }
+	union {
+		btDbvtNode* childs[2];
+		void* data;
+		int dataAsInt;
 	};
 };
 
 typedef btAlignedObjectArray<const btDbvtNode*> btNodeStack;
 
-
 ///The btDbvt class implements a fast dynamic bounding volume tree based on axis aligned bounding boxes (aabb tree).
 ///This btDbvt is used for soft body collision detection and for the btDbvtBroadphase. It has a fast insert, remove and update of nodes.
 ///Unlike the btQuantizedBvh, nodes can be dynamically moved around, which allows for change in topology of the underlying data structure.
-struct	btDbvt
+struct btDbvt
 {
-	/* Stack element	*/ 
-	struct	sStkNN
+	/* Stack element	*/
+	struct sStkNN
 	{
-		const btDbvtNode*	a;
-		const btDbvtNode*	b;
+		const btDbvtNode* a;
+		const btDbvtNode* b;
 		sStkNN() {}
-		sStkNN(const btDbvtNode* na,const btDbvtNode* nb) : a(na),b(nb) {}
+		sStkNN(const btDbvtNode* na, const btDbvtNode* nb) : a(na), b(nb) {}
 	};
-	struct	sStkNP
+	struct sStkNP
 	{
-		const btDbvtNode*	node;
-		int			mask;
-		sStkNP(const btDbvtNode* n,unsigned m) : node(n),mask(m) {}
+		const btDbvtNode* node;
+		int mask;
+		sStkNP(const btDbvtNode* n, unsigned m) : node(n), mask(m) {}
 	};
-	struct	sStkNPS
+	struct sStkNPS
 	{
-		const btDbvtNode*	node;
-		int			mask;
-		btScalar	value;
+		const btDbvtNode* node;
+		int mask;
+		btScalar value;
 		sStkNPS() {}
-		sStkNPS(const btDbvtNode* n,unsigned m,btScalar v) : node(n),mask(m),value(v) {}
+		sStkNPS(const btDbvtNode* n, unsigned m, btScalar v) : node(n), mask(m), value(v) {}
 	};
-	struct	sStkCLN
+	struct sStkCLN
 	{
-		const btDbvtNode*	node;
-		btDbvtNode*		parent;
-		sStkCLN(const btDbvtNode* n,btDbvtNode* p) : node(n),parent(p) {}
+		const btDbvtNode* node;
+		btDbvtNode* parent;
+		sStkCLN(const btDbvtNode* n, btDbvtNode* p) : node(n), parent(p) {}
 	};
 	// Policies/Interfaces
 
-	/* ICollide	*/ 
-	struct	ICollide
-	{		
+	/* ICollide	*/
+	struct ICollide
+	{
 		DBVT_VIRTUAL_DTOR(ICollide)
-			DBVT_VIRTUAL void	Process(const btDbvtNode*,const btDbvtNode*)		{}
-		DBVT_VIRTUAL void	Process(const btDbvtNode*)					{}
-		DBVT_VIRTUAL void	Process(const btDbvtNode* n,btScalar)			{ Process(n); }
-		DBVT_VIRTUAL bool	Descent(const btDbvtNode*)					{ return(true); }
-		DBVT_VIRTUAL bool	AllLeaves(const btDbvtNode*)					{ return(true); }
+		DBVT_VIRTUAL void Process(const btDbvtNode*, const btDbvtNode*) {}
+		DBVT_VIRTUAL void Process(const btDbvtNode*) {}
+		DBVT_VIRTUAL void Process(const btDbvtNode* n, btScalar) { Process(n); }
+		DBVT_VIRTUAL bool Descent(const btDbvtNode*) { return (true); }
+		DBVT_VIRTUAL bool AllLeaves(const btDbvtNode*) { return (true); }
 	};
-	/* IWriter	*/ 
-	struct	IWriter
+	/* IWriter	*/
+	struct IWriter
 	{
 		virtual ~IWriter() {}
-		virtual void		Prepare(const btDbvtNode* root,int numnodes)=0;
-		virtual void		WriteNode(const btDbvtNode*,int index,int parent,int child0,int child1)=0;
-		virtual void		WriteLeaf(const btDbvtNode*,int index,int parent)=0;
+		virtual void Prepare(const btDbvtNode* root, int numnodes) = 0;
+		virtual void WriteNode(const btDbvtNode*, int index, int parent, int child0, int child1) = 0;
+		virtual void WriteLeaf(const btDbvtNode*, int index, int parent) = 0;
 	};
-	/* IClone	*/ 
-	struct	IClone
+	/* IClone	*/
+	struct IClone
 	{
-		virtual ~IClone()	{}
-		virtual void		CloneLeaf(btDbvtNode*) {}
+		virtual ~IClone() {}
+		virtual void CloneLeaf(btDbvtNode*) {}
 	};
 
 	// Constants
-	enum	{
-		SIMPLE_STACKSIZE	=	64,
-		DOUBLE_STACKSIZE	=	SIMPLE_STACKSIZE*2
+	enum
+	{
+		SIMPLE_STACKSIZE = 64,
+		DOUBLE_STACKSIZE = SIMPLE_STACKSIZE * 2
 	};
 
 	// Fields
-	btDbvtNode*		m_root;
-	btDbvtNode*		m_free;
-	int				m_lkhd;
-	int				m_leaves;
-	unsigned		m_opath;
-
-	
-	btAlignedObjectArray<sStkNN>	m_stkStack;
+	btDbvtNode* m_root;
+	btDbvtNode* m_free;
+	int m_lkhd;
+	int m_leaves;
+	unsigned m_opath;
 
+	btAlignedObjectArray<sStkNN> m_stkStack;
 
 	// Methods
 	btDbvt();
 	~btDbvt();
-	void			clear();
-	bool			empty() const { return(0==m_root); }
-	void			optimizeBottomUp();
-	void			optimizeTopDown(int bu_treshold=128);
-	void			optimizeIncremental(int passes);
-	btDbvtNode*		insert(const btDbvtVolume& box,void* data);
-	void			update(btDbvtNode* leaf,int lookahead=-1);
-	void			update(btDbvtNode* leaf,btDbvtVolume& volume);
-	bool			update(btDbvtNode* leaf,btDbvtVolume& volume,const btVector3& velocity,btScalar margin);
-	bool			update(btDbvtNode* leaf,btDbvtVolume& volume,const btVector3& velocity);
-	bool			update(btDbvtNode* leaf,btDbvtVolume& volume,btScalar margin);	
-	void			remove(btDbvtNode* leaf);
-	void			write(IWriter* iwriter) const;
-	void			clone(btDbvt& dest,IClone* iclone=0) const;
-	static int		maxdepth(const btDbvtNode* node);
-	static int		countLeaves(const btDbvtNode* node);
-	static void		extractLeaves(const btDbvtNode* node,btAlignedObjectArray<const btDbvtNode*>& leaves);
+	void clear();
+	bool empty() const { return (0 == m_root); }
+	void optimizeBottomUp();
+	void optimizeTopDown(int bu_treshold = 128);
+	void optimizeIncremental(int passes);
+	btDbvtNode* insert(const btDbvtVolume& box, void* data);
+	void update(btDbvtNode* leaf, int lookahead = -1);
+	void update(btDbvtNode* leaf, btDbvtVolume& volume);
+	bool update(btDbvtNode* leaf, btDbvtVolume& volume, const btVector3& velocity, btScalar margin);
+	bool update(btDbvtNode* leaf, btDbvtVolume& volume, const btVector3& velocity);
+	bool update(btDbvtNode* leaf, btDbvtVolume& volume, btScalar margin);
+	void remove(btDbvtNode* leaf);
+	void write(IWriter* iwriter) const;
+	void clone(btDbvt& dest, IClone* iclone = 0) const;
+	static int maxdepth(const btDbvtNode* node);
+	static int countLeaves(const btDbvtNode* node);
+	static void extractLeaves(const btDbvtNode* node, btAlignedObjectArray<const btDbvtNode*>& leaves);
 #if DBVT_ENABLE_BENCHMARK
-	static void		benchmark();
+	static void benchmark();
 #else
-	static void		benchmark(){}
+	static void benchmark()
+	{
+	}
 #endif
 	// DBVT_IPOLICY must support ICollide policy/interface
 	DBVT_PREFIX
-		static void		enumNodes(	const btDbvtNode* root,
-		DBVT_IPOLICY);
+	static void enumNodes(const btDbvtNode* root,
+						  DBVT_IPOLICY);
 	DBVT_PREFIX
-		static void		enumLeaves(	const btDbvtNode* root,
-		DBVT_IPOLICY);
+	static void enumLeaves(const btDbvtNode* root,
+						   DBVT_IPOLICY);
 	DBVT_PREFIX
-		void		collideTT(	const btDbvtNode* root0,
-		const btDbvtNode* root1,
-		DBVT_IPOLICY);
+	void collideTT(const btDbvtNode* root0,
+				   const btDbvtNode* root1,
+				   DBVT_IPOLICY);
 
 	DBVT_PREFIX
-		void		collideTTpersistentStack(	const btDbvtNode* root0,
-		  const btDbvtNode* root1,
-		  DBVT_IPOLICY);
+	void collideTTpersistentStack(const btDbvtNode* root0,
+								  const btDbvtNode* root1,
+								  DBVT_IPOLICY);
 #if 0
 	DBVT_PREFIX
 		void		collideTT(	const btDbvtNode* root0,
@@ -325,82 +326,89 @@ struct	btDbvt
 #endif
 
 	DBVT_PREFIX
-		void		collideTV(	const btDbvtNode* root,
-		const btDbvtVolume& volume,
-		DBVT_IPOLICY) const;
-	
+	void collideTV(const btDbvtNode* root,
+				   const btDbvtVolume& volume,
+				   DBVT_IPOLICY) const;
+
 	DBVT_PREFIX
-	void		collideTVNoStackAlloc(	const btDbvtNode* root,
-						  const btDbvtVolume& volume,
-						  btNodeStack& stack,
-						  DBVT_IPOLICY) const;
-	
-	
-	
-	
+	void collideTVNoStackAlloc(const btDbvtNode* root,
+							   const btDbvtVolume& volume,
+							   btNodeStack& stack,
+							   DBVT_IPOLICY) const;
+
 	///rayTest is a re-entrant ray test, and can be called in parallel as long as the btAlignedAlloc is thread-safe (uses locking etc)
 	///rayTest is slower than rayTestInternal, because it builds a local stack, using memory allocations, and it recomputes signs/rayDirectionInverses each time
 	DBVT_PREFIX
-		static void		rayTest(	const btDbvtNode* root,
-		const btVector3& rayFrom,
-		const btVector3& rayTo,
-		DBVT_IPOLICY);
+	static void rayTest(const btDbvtNode* root,
+						const btVector3& rayFrom,
+						const btVector3& rayTo,
+						DBVT_IPOLICY);
 	///rayTestInternal is faster than rayTest, because it uses a persistent stack (to reduce dynamic memory allocations to a minimum) and it uses precomputed signs/rayInverseDirections
 	///rayTestInternal is used by btDbvtBroadphase to accelerate world ray casts
 	DBVT_PREFIX
-		void		rayTestInternal(	const btDbvtNode* root,
-								const btVector3& rayFrom,
-								const btVector3& rayTo,
-								const btVector3& rayDirectionInverse,
-								unsigned int signs[3],
-								btScalar lambda_max,
-								const btVector3& aabbMin,
-								const btVector3& aabbMax,
-                                btAlignedObjectArray<const btDbvtNode*>& stack,
-								DBVT_IPOLICY) const;
+	void rayTestInternal(const btDbvtNode* root,
+						 const btVector3& rayFrom,
+						 const btVector3& rayTo,
+						 const btVector3& rayDirectionInverse,
+						 unsigned int signs[3],
+						 btScalar lambda_max,
+						 const btVector3& aabbMin,
+						 const btVector3& aabbMax,
+						 btAlignedObjectArray<const btDbvtNode*>& stack,
+						 DBVT_IPOLICY) const;
 
 	DBVT_PREFIX
-		static void		collideKDOP(const btDbvtNode* root,
-		const btVector3* normals,
-		const btScalar* offsets,
-		int count,
-		DBVT_IPOLICY);
+	static void collideKDOP(const btDbvtNode* root,
+							const btVector3* normals,
+							const btScalar* offsets,
+							int count,
+							DBVT_IPOLICY);
 	DBVT_PREFIX
-		static void		collideOCL(	const btDbvtNode* root,
-		const btVector3* normals,
-		const btScalar* offsets,
-		const btVector3& sortaxis,
-		int count,								
-		DBVT_IPOLICY,
-		bool fullsort=true);
+	static void collideOCL(const btDbvtNode* root,
+						   const btVector3* normals,
+						   const btScalar* offsets,
+						   const btVector3& sortaxis,
+						   int count,
+						   DBVT_IPOLICY,
+						   bool fullsort = true);
 	DBVT_PREFIX
-		static void		collideTU(	const btDbvtNode* root,
-		DBVT_IPOLICY);
-	// Helpers	
-	static DBVT_INLINE int	nearest(const int* i,const btDbvt::sStkNPS* a,btScalar v,int l,int h)
+	static void collideTU(const btDbvtNode* root,
+						  DBVT_IPOLICY);
+	// Helpers
+	static DBVT_INLINE int nearest(const int* i, const btDbvt::sStkNPS* a, btScalar v, int l, int h)
 	{
-		int	m=0;
-		while(l<h)
+		int m = 0;
+		while (l < h)
 		{
-			m=(l+h)>>1;
-			if(a[i[m]].value>=v) l=m+1; else h=m;
+			m = (l + h) >> 1;
+			if (a[i[m]].value >= v)
+				l = m + 1;
+			else
+				h = m;
 		}
-		return(h);
+		return (h);
 	}
-	static DBVT_INLINE int	allocate(	btAlignedObjectArray<int>& ifree,
-		btAlignedObjectArray<sStkNPS>& stock,
-		const sStkNPS& value)
+	static DBVT_INLINE int allocate(btAlignedObjectArray<int>& ifree,
+									btAlignedObjectArray<sStkNPS>& stock,
+									const sStkNPS& value)
 	{
-		int	i;
-		if(ifree.size()>0)
-		{ i=ifree[ifree.size()-1];ifree.pop_back();stock[i]=value; }
+		int i;
+		if (ifree.size() > 0)
+		{
+			i = ifree[ifree.size() - 1];
+			ifree.pop_back();
+			stock[i] = value;
+		}
 		else
-		{ i=stock.size();stock.push_back(value); }
-		return(i); 
+		{
+			i = stock.size();
+			stock.push_back(value);
+		}
+		return (i);
 	}
 	//
 private:
-	btDbvt(const btDbvt&)	{}	
+	btDbvt(const btDbvt&) {}
 };
 
 //
@@ -408,227 +416,252 @@ private:
 //
 
 //
-inline btDbvtAabbMm			btDbvtAabbMm::FromCE(const btVector3& c,const btVector3& e)
+inline btDbvtAabbMm btDbvtAabbMm::FromCE(const btVector3& c, const btVector3& e)
 {
 	btDbvtAabbMm box;
-	box.mi=c-e;box.mx=c+e;
-	return(box);
+	box.mi = c - e;
+	box.mx = c + e;
+	return (box);
 }
 
 //
-inline btDbvtAabbMm			btDbvtAabbMm::FromCR(const btVector3& c,btScalar r)
+inline btDbvtAabbMm btDbvtAabbMm::FromCR(const btVector3& c, btScalar r)
 {
-	return(FromCE(c,btVector3(r,r,r)));
+	return (FromCE(c, btVector3(r, r, r)));
 }
 
 //
-inline btDbvtAabbMm			btDbvtAabbMm::FromMM(const btVector3& mi,const btVector3& mx)
+inline btDbvtAabbMm btDbvtAabbMm::FromMM(const btVector3& mi, const btVector3& mx)
 {
 	btDbvtAabbMm box;
-	box.mi=mi;box.mx=mx;
-	return(box);
+	box.mi = mi;
+	box.mx = mx;
+	return (box);
 }
 
 //
-inline btDbvtAabbMm			btDbvtAabbMm::FromPoints(const btVector3* pts,int n)
+inline btDbvtAabbMm btDbvtAabbMm::FromPoints(const btVector3* pts, int n)
 {
 	btDbvtAabbMm box;
-	box.mi=box.mx=pts[0];
-	for(int i=1;i<n;++i)
+	box.mi = box.mx = pts[0];
+	for (int i = 1; i < n; ++i)
 	{
 		box.mi.setMin(pts[i]);
 		box.mx.setMax(pts[i]);
 	}
-	return(box);
+	return (box);
 }
 
 //
-inline btDbvtAabbMm			btDbvtAabbMm::FromPoints(const btVector3** ppts,int n)
+inline btDbvtAabbMm btDbvtAabbMm::FromPoints(const btVector3** ppts, int n)
 {
 	btDbvtAabbMm box;
-	box.mi=box.mx=*ppts[0];
-	for(int i=1;i<n;++i)
+	box.mi = box.mx = *ppts[0];
+	for (int i = 1; i < n; ++i)
 	{
 		box.mi.setMin(*ppts[i]);
 		box.mx.setMax(*ppts[i]);
 	}
-	return(box);
+	return (box);
 }
 
 //
-DBVT_INLINE void		btDbvtAabbMm::Expand(const btVector3& e)
+DBVT_INLINE void btDbvtAabbMm::Expand(const btVector3& e)
 {
-	mi-=e;mx+=e;
+	mi -= e;
+	mx += e;
 }
 
 //
-DBVT_INLINE void		btDbvtAabbMm::SignedExpand(const btVector3& e)
+DBVT_INLINE void btDbvtAabbMm::SignedExpand(const btVector3& e)
 {
-	if(e.x()>0) mx.setX(mx.x()+e[0]); else mi.setX(mi.x()+e[0]);
-	if(e.y()>0) mx.setY(mx.y()+e[1]); else mi.setY(mi.y()+e[1]);
-	if(e.z()>0) mx.setZ(mx.z()+e[2]); else mi.setZ(mi.z()+e[2]);
+	if (e.x() > 0)
+		mx.setX(mx.x() + e[0]);
+	else
+		mi.setX(mi.x() + e[0]);
+	if (e.y() > 0)
+		mx.setY(mx.y() + e[1]);
+	else
+		mi.setY(mi.y() + e[1]);
+	if (e.z() > 0)
+		mx.setZ(mx.z() + e[2]);
+	else
+		mi.setZ(mi.z() + e[2]);
 }
 
 //
-DBVT_INLINE bool		btDbvtAabbMm::Contain(const btDbvtAabbMm& a) const
+DBVT_INLINE bool btDbvtAabbMm::Contain(const btDbvtAabbMm& a) const
 {
-	return(	(mi.x()<=a.mi.x())&&
-		(mi.y()<=a.mi.y())&&
-		(mi.z()<=a.mi.z())&&
-		(mx.x()>=a.mx.x())&&
-		(mx.y()>=a.mx.y())&&
-		(mx.z()>=a.mx.z()));
+	return ((mi.x() <= a.mi.x()) &&
+			(mi.y() <= a.mi.y()) &&
+			(mi.z() <= a.mi.z()) &&
+			(mx.x() >= a.mx.x()) &&
+			(mx.y() >= a.mx.y()) &&
+			(mx.z() >= a.mx.z()));
 }
 
 //
-DBVT_INLINE int		btDbvtAabbMm::Classify(const btVector3& n,btScalar o,int s) const
+DBVT_INLINE int btDbvtAabbMm::Classify(const btVector3& n, btScalar o, int s) const
 {
-	btVector3			pi,px;
-	switch(s)
+	btVector3 pi, px;
+	switch (s)
 	{
-	case	(0+0+0):	px=btVector3(mi.x(),mi.y(),mi.z());
-		pi=btVector3(mx.x(),mx.y(),mx.z());break;
-	case	(1+0+0):	px=btVector3(mx.x(),mi.y(),mi.z());
-		pi=btVector3(mi.x(),mx.y(),mx.z());break;
-	case	(0+2+0):	px=btVector3(mi.x(),mx.y(),mi.z());
-		pi=btVector3(mx.x(),mi.y(),mx.z());break;
-	case	(1+2+0):	px=btVector3(mx.x(),mx.y(),mi.z());
-		pi=btVector3(mi.x(),mi.y(),mx.z());break;
-	case	(0+0+4):	px=btVector3(mi.x(),mi.y(),mx.z());
-		pi=btVector3(mx.x(),mx.y(),mi.z());break;
-	case	(1+0+4):	px=btVector3(mx.x(),mi.y(),mx.z());
-		pi=btVector3(mi.x(),mx.y(),mi.z());break;
-	case	(0+2+4):	px=btVector3(mi.x(),mx.y(),mx.z());
-		pi=btVector3(mx.x(),mi.y(),mi.z());break;
-	case	(1+2+4):	px=btVector3(mx.x(),mx.y(),mx.z());
-		pi=btVector3(mi.x(),mi.y(),mi.z());break;
+		case (0 + 0 + 0):
+			px = btVector3(mi.x(), mi.y(), mi.z());
+			pi = btVector3(mx.x(), mx.y(), mx.z());
+			break;
+		case (1 + 0 + 0):
+			px = btVector3(mx.x(), mi.y(), mi.z());
+			pi = btVector3(mi.x(), mx.y(), mx.z());
+			break;
+		case (0 + 2 + 0):
+			px = btVector3(mi.x(), mx.y(), mi.z());
+			pi = btVector3(mx.x(), mi.y(), mx.z());
+			break;
+		case (1 + 2 + 0):
+			px = btVector3(mx.x(), mx.y(), mi.z());
+			pi = btVector3(mi.x(), mi.y(), mx.z());
+			break;
+		case (0 + 0 + 4):
+			px = btVector3(mi.x(), mi.y(), mx.z());
+			pi = btVector3(mx.x(), mx.y(), mi.z());
+			break;
+		case (1 + 0 + 4):
+			px = btVector3(mx.x(), mi.y(), mx.z());
+			pi = btVector3(mi.x(), mx.y(), mi.z());
+			break;
+		case (0 + 2 + 4):
+			px = btVector3(mi.x(), mx.y(), mx.z());
+			pi = btVector3(mx.x(), mi.y(), mi.z());
+			break;
+		case (1 + 2 + 4):
+			px = btVector3(mx.x(), mx.y(), mx.z());
+			pi = btVector3(mi.x(), mi.y(), mi.z());
+			break;
 	}
-	if((btDot(n,px)+o)<0)		return(-1);
-	if((btDot(n,pi)+o)>=0)	return(+1);
-	return(0);
+	if ((btDot(n, px) + o) < 0) return (-1);
+	if ((btDot(n, pi) + o) >= 0) return (+1);
+	return (0);
 }
 
 //
-DBVT_INLINE btScalar	btDbvtAabbMm::ProjectMinimum(const btVector3& v,unsigned signs) const
+DBVT_INLINE btScalar btDbvtAabbMm::ProjectMinimum(const btVector3& v, unsigned signs) const
 {
-	const btVector3*	b[]={&mx,&mi};
-	const btVector3		p(	b[(signs>>0)&1]->x(),
-		b[(signs>>1)&1]->y(),
-		b[(signs>>2)&1]->z());
-	return(btDot(p,v));
+	const btVector3* b[] = {&mx, &mi};
+	const btVector3 p(b[(signs >> 0) & 1]->x(),
+					  b[(signs >> 1) & 1]->y(),
+					  b[(signs >> 2) & 1]->z());
+	return (btDot(p, v));
 }
 
 //
-DBVT_INLINE void		btDbvtAabbMm::AddSpan(const btVector3& d,btScalar& smi,btScalar& smx) const
+DBVT_INLINE void btDbvtAabbMm::AddSpan(const btVector3& d, btScalar& smi, btScalar& smx) const
 {
-	for(int i=0;i<3;++i)
+	for (int i = 0; i < 3; ++i)
 	{
-		if(d[i]<0)
-		{ smi+=mx[i]*d[i];smx+=mi[i]*d[i]; }
+		if (d[i] < 0)
+		{
+			smi += mx[i] * d[i];
+			smx += mi[i] * d[i];
+		}
 		else
-		{ smi+=mi[i]*d[i];smx+=mx[i]*d[i]; }
+		{
+			smi += mi[i] * d[i];
+			smx += mx[i] * d[i];
+		}
 	}
 }
 
 //
-DBVT_INLINE bool		Intersect(	const btDbvtAabbMm& a,
-								  const btDbvtAabbMm& b)
+DBVT_INLINE bool Intersect(const btDbvtAabbMm& a,
+						   const btDbvtAabbMm& b)
 {
-#if	DBVT_INT0_IMPL == DBVT_IMPL_SSE
-	const __m128	rt(_mm_or_ps(	_mm_cmplt_ps(_mm_load_ps(b.mx),_mm_load_ps(a.mi)),
-		_mm_cmplt_ps(_mm_load_ps(a.mx),_mm_load_ps(b.mi))));
-#if defined (_WIN32)
-	const __int32*	pu((const __int32*)&rt);
+#if DBVT_INT0_IMPL == DBVT_IMPL_SSE
+	const __m128 rt(_mm_or_ps(_mm_cmplt_ps(_mm_load_ps(b.mx), _mm_load_ps(a.mi)),
+							  _mm_cmplt_ps(_mm_load_ps(a.mx), _mm_load_ps(b.mi))));
+#if defined(_WIN32)
+	const __int32* pu((const __int32*)&rt);
 #else
-    const int*	pu((const int*)&rt);
+	const int* pu((const int*)&rt);
 #endif
-	return((pu[0]|pu[1]|pu[2])==0);
+	return ((pu[0] | pu[1] | pu[2]) == 0);
 #else
-	return(	(a.mi.x()<=b.mx.x())&&
-		(a.mx.x()>=b.mi.x())&&
-		(a.mi.y()<=b.mx.y())&&
-		(a.mx.y()>=b.mi.y())&&
-		(a.mi.z()<=b.mx.z())&&		
-		(a.mx.z()>=b.mi.z()));
+	return ((a.mi.x() <= b.mx.x()) &&
+			(a.mx.x() >= b.mi.x()) &&
+			(a.mi.y() <= b.mx.y()) &&
+			(a.mx.y() >= b.mi.y()) &&
+			(a.mi.z() <= b.mx.z()) &&
+			(a.mx.z() >= b.mi.z()));
 #endif
 }
 
-
-
 //
-DBVT_INLINE bool		Intersect(	const btDbvtAabbMm& a,
-								  const btVector3& b)
+DBVT_INLINE bool Intersect(const btDbvtAabbMm& a,
+						   const btVector3& b)
 {
-	return(	(b.x()>=a.mi.x())&&
-		(b.y()>=a.mi.y())&&
-		(b.z()>=a.mi.z())&&
-		(b.x()<=a.mx.x())&&
-		(b.y()<=a.mx.y())&&
-		(b.z()<=a.mx.z()));
+	return ((b.x() >= a.mi.x()) &&
+			(b.y() >= a.mi.y()) &&
+			(b.z() >= a.mi.z()) &&
+			(b.x() <= a.mx.x()) &&
+			(b.y() <= a.mx.y()) &&
+			(b.z() <= a.mx.z()));
 }
 
-
-
-
-
 //////////////////////////////////////
 
-
 //
-DBVT_INLINE btScalar	Proximity(	const btDbvtAabbMm& a,
-								  const btDbvtAabbMm& b)
-{
-	const btVector3	d=(a.mi+a.mx)-(b.mi+b.mx);
-	return(btFabs(d.x())+btFabs(d.y())+btFabs(d.z()));
-}
-
-
-
-//
-DBVT_INLINE int			Select(	const btDbvtAabbMm& o,
-							   const btDbvtAabbMm& a,
+DBVT_INLINE btScalar Proximity(const btDbvtAabbMm& a,
 							   const btDbvtAabbMm& b)
 {
-#if	DBVT_SELECT_IMPL == DBVT_IMPL_SSE
-    
-#if defined (_WIN32)
-	static ATTRIBUTE_ALIGNED16(const unsigned __int32)	mask[]={0x7fffffff,0x7fffffff,0x7fffffff,0x7fffffff};
+	const btVector3 d = (a.mi + a.mx) - (b.mi + b.mx);
+	return (btFabs(d.x()) + btFabs(d.y()) + btFabs(d.z()));
+}
+
+//
+DBVT_INLINE int Select(const btDbvtAabbMm& o,
+					   const btDbvtAabbMm& a,
+					   const btDbvtAabbMm& b)
+{
+#if DBVT_SELECT_IMPL == DBVT_IMPL_SSE
+
+#if defined(_WIN32)
+	static ATTRIBUTE_ALIGNED16(const unsigned __int32) mask[] = {0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff};
 #else
-    static ATTRIBUTE_ALIGNED16(const unsigned int)	mask[]={0x7fffffff,0x7fffffff,0x7fffffff,0x00000000 /*0x7fffffff*/};
+	static ATTRIBUTE_ALIGNED16(const unsigned int) mask[] = {0x7fffffff, 0x7fffffff, 0x7fffffff, 0x00000000 /*0x7fffffff*/};
 #endif
 	///@todo: the intrinsic version is 11% slower
 #if DBVT_USE_INTRINSIC_SSE
 
-	union btSSEUnion ///NOTE: if we use more intrinsics, move btSSEUnion into the LinearMath directory
+	union btSSEUnion  ///NOTE: if we use more intrinsics, move btSSEUnion into the LinearMath directory
 	{
-	   __m128		ssereg;
-	   float		floats[4];
-	   int			ints[4];
+		__m128 ssereg;
+		float floats[4];
+		int ints[4];
 	};
 
-	__m128	omi(_mm_load_ps(o.mi));
-	omi=_mm_add_ps(omi,_mm_load_ps(o.mx));
-	__m128	ami(_mm_load_ps(a.mi));
-	ami=_mm_add_ps(ami,_mm_load_ps(a.mx));
-	ami=_mm_sub_ps(ami,omi);
-	ami=_mm_and_ps(ami,_mm_load_ps((const float*)mask));
-	__m128	bmi(_mm_load_ps(b.mi));
-	bmi=_mm_add_ps(bmi,_mm_load_ps(b.mx));
-	bmi=_mm_sub_ps(bmi,omi);
-	bmi=_mm_and_ps(bmi,_mm_load_ps((const float*)mask));
-	__m128	t0(_mm_movehl_ps(ami,ami));
-	ami=_mm_add_ps(ami,t0);
-	ami=_mm_add_ss(ami,_mm_shuffle_ps(ami,ami,1));
-	__m128 t1(_mm_movehl_ps(bmi,bmi));
-	bmi=_mm_add_ps(bmi,t1);
-	bmi=_mm_add_ss(bmi,_mm_shuffle_ps(bmi,bmi,1));
-	
+	__m128 omi(_mm_load_ps(o.mi));
+	omi = _mm_add_ps(omi, _mm_load_ps(o.mx));
+	__m128 ami(_mm_load_ps(a.mi));
+	ami = _mm_add_ps(ami, _mm_load_ps(a.mx));
+	ami = _mm_sub_ps(ami, omi);
+	ami = _mm_and_ps(ami, _mm_load_ps((const float*)mask));
+	__m128 bmi(_mm_load_ps(b.mi));
+	bmi = _mm_add_ps(bmi, _mm_load_ps(b.mx));
+	bmi = _mm_sub_ps(bmi, omi);
+	bmi = _mm_and_ps(bmi, _mm_load_ps((const float*)mask));
+	__m128 t0(_mm_movehl_ps(ami, ami));
+	ami = _mm_add_ps(ami, t0);
+	ami = _mm_add_ss(ami, _mm_shuffle_ps(ami, ami, 1));
+	__m128 t1(_mm_movehl_ps(bmi, bmi));
+	bmi = _mm_add_ps(bmi, t1);
+	bmi = _mm_add_ss(bmi, _mm_shuffle_ps(bmi, bmi, 1));
+
 	btSSEUnion tmp;
-	tmp.ssereg = _mm_cmple_ss(bmi,ami);
-	return tmp.ints[0]&1;
+	tmp.ssereg = _mm_cmple_ss(bmi, ami);
+	return tmp.ints[0] & 1;
 
 #else
-	ATTRIBUTE_ALIGNED16(__int32	r[1]);
+	ATTRIBUTE_ALIGNED16(__int32 r[1]);
 	__asm
 	{
 		mov		eax,o
@@ -656,46 +689,52 @@ DBVT_INLINE int			Select(	const btDbvtAabbMm& o,
 			cmpless	xmm2,xmm1
 			movss	r,xmm2
 	}
-	return(r[0]&1);
+	return (r[0] & 1);
 #endif
 #else
-	return(Proximity(o,a)<Proximity(o,b)?0:1);
+	return (Proximity(o, a) < Proximity(o, b) ? 0 : 1);
 #endif
 }
 
 //
-DBVT_INLINE void		Merge(	const btDbvtAabbMm& a,
-							  const btDbvtAabbMm& b,
-							  btDbvtAabbMm& r)
+DBVT_INLINE void Merge(const btDbvtAabbMm& a,
+					   const btDbvtAabbMm& b,
+					   btDbvtAabbMm& r)
 {
-#if DBVT_MERGE_IMPL==DBVT_IMPL_SSE
-	__m128	ami(_mm_load_ps(a.mi));
-	__m128	amx(_mm_load_ps(a.mx));
-	__m128	bmi(_mm_load_ps(b.mi));
-	__m128	bmx(_mm_load_ps(b.mx));
-	ami=_mm_min_ps(ami,bmi);
-	amx=_mm_max_ps(amx,bmx);
-	_mm_store_ps(r.mi,ami);
-	_mm_store_ps(r.mx,amx);
+#if DBVT_MERGE_IMPL == DBVT_IMPL_SSE
+	__m128 ami(_mm_load_ps(a.mi));
+	__m128 amx(_mm_load_ps(a.mx));
+	__m128 bmi(_mm_load_ps(b.mi));
+	__m128 bmx(_mm_load_ps(b.mx));
+	ami = _mm_min_ps(ami, bmi);
+	amx = _mm_max_ps(amx, bmx);
+	_mm_store_ps(r.mi, ami);
+	_mm_store_ps(r.mx, amx);
 #else
-	for(int i=0;i<3;++i)
+	for (int i = 0; i < 3; ++i)
 	{
-		if(a.mi[i]<b.mi[i]) r.mi[i]=a.mi[i]; else r.mi[i]=b.mi[i];
-		if(a.mx[i]>b.mx[i]) r.mx[i]=a.mx[i]; else r.mx[i]=b.mx[i];
+		if (a.mi[i] < b.mi[i])
+			r.mi[i] = a.mi[i];
+		else
+			r.mi[i] = b.mi[i];
+		if (a.mx[i] > b.mx[i])
+			r.mx[i] = a.mx[i];
+		else
+			r.mx[i] = b.mx[i];
 	}
 #endif
 }
 
 //
-DBVT_INLINE bool		NotEqual(	const btDbvtAabbMm& a,
-								 const btDbvtAabbMm& b)
+DBVT_INLINE bool NotEqual(const btDbvtAabbMm& a,
+						  const btDbvtAabbMm& b)
 {
-	return(	(a.mi.x()!=b.mi.x())||
-		(a.mi.y()!=b.mi.y())||
-		(a.mi.z()!=b.mi.z())||
-		(a.mx.x()!=b.mx.x())||
-		(a.mx.y()!=b.mx.y())||
-		(a.mx.z()!=b.mx.z()));
+	return ((a.mi.x() != b.mi.x()) ||
+			(a.mi.y() != b.mi.y()) ||
+			(a.mi.z() != b.mi.z()) ||
+			(a.mx.x() != b.mx.x()) ||
+			(a.mx.y() != b.mx.y()) ||
+			(a.mx.z() != b.mx.z()));
 }
 
 //
@@ -704,162 +743,162 @@ DBVT_INLINE bool		NotEqual(	const btDbvtAabbMm& a,
 
 //
 DBVT_PREFIX
-inline void		btDbvt::enumNodes(	const btDbvtNode* root,
-								  DBVT_IPOLICY)
+inline void btDbvt::enumNodes(const btDbvtNode* root,
+							  DBVT_IPOLICY)
 {
 	DBVT_CHECKTYPE
-		policy.Process(root);
-	if(root->isinternal())
+	policy.Process(root);
+	if (root->isinternal())
 	{
-		enumNodes(root->childs[0],policy);
-		enumNodes(root->childs[1],policy);
+		enumNodes(root->childs[0], policy);
+		enumNodes(root->childs[1], policy);
 	}
 }
 
 //
 DBVT_PREFIX
-inline void		btDbvt::enumLeaves(	const btDbvtNode* root,
-								   DBVT_IPOLICY)
+inline void btDbvt::enumLeaves(const btDbvtNode* root,
+							   DBVT_IPOLICY)
 {
 	DBVT_CHECKTYPE
-		if(root->isinternal())
-		{
-			enumLeaves(root->childs[0],policy);
-			enumLeaves(root->childs[1],policy);
-		}
-		else
-		{
-			policy.Process(root);
-		}
+	if (root->isinternal())
+	{
+		enumLeaves(root->childs[0], policy);
+		enumLeaves(root->childs[1], policy);
+	}
+	else
+	{
+		policy.Process(root);
+	}
 }
 
 //
 DBVT_PREFIX
-inline void		btDbvt::collideTT(	const btDbvtNode* root0,
-								  const btDbvtNode* root1,
-								  DBVT_IPOLICY)
+inline void btDbvt::collideTT(const btDbvtNode* root0,
+							  const btDbvtNode* root1,
+							  DBVT_IPOLICY)
 {
 	DBVT_CHECKTYPE
-		if(root0&&root1)
+	if (root0 && root1)
+	{
+		int depth = 1;
+		int treshold = DOUBLE_STACKSIZE - 4;
+		btAlignedObjectArray<sStkNN> stkStack;
+		stkStack.resize(DOUBLE_STACKSIZE);
+		stkStack[0] = sStkNN(root0, root1);
+		do
 		{
-			int								depth=1;
-			int								treshold=DOUBLE_STACKSIZE-4;
-			btAlignedObjectArray<sStkNN>	stkStack;
-			stkStack.resize(DOUBLE_STACKSIZE);
-			stkStack[0]=sStkNN(root0,root1);
-			do	{		
-				sStkNN	p=stkStack[--depth];
-				if(depth>treshold)
+			sStkNN p = stkStack[--depth];
+			if (depth > treshold)
+			{
+				stkStack.resize(stkStack.size() * 2);
+				treshold = stkStack.size() - 4;
+			}
+			if (p.a == p.b)
+			{
+				if (p.a->isinternal())
 				{
-					stkStack.resize(stkStack.size()*2);
-					treshold=stkStack.size()-4;
+					stkStack[depth++] = sStkNN(p.a->childs[0], p.a->childs[0]);
+					stkStack[depth++] = sStkNN(p.a->childs[1], p.a->childs[1]);
+					stkStack[depth++] = sStkNN(p.a->childs[0], p.a->childs[1]);
 				}
-				if(p.a==p.b)
+			}
+			else if (Intersect(p.a->volume, p.b->volume))
+			{
+				if (p.a->isinternal())
 				{
-					if(p.a->isinternal())
+					if (p.b->isinternal())
 					{
-						stkStack[depth++]=sStkNN(p.a->childs[0],p.a->childs[0]);
-						stkStack[depth++]=sStkNN(p.a->childs[1],p.a->childs[1]);
-						stkStack[depth++]=sStkNN(p.a->childs[0],p.a->childs[1]);
-					}
-				}
-				else if(Intersect(p.a->volume,p.b->volume))
-				{
-					if(p.a->isinternal())
-					{
-						if(p.b->isinternal())
-						{
-							stkStack[depth++]=sStkNN(p.a->childs[0],p.b->childs[0]);
-							stkStack[depth++]=sStkNN(p.a->childs[1],p.b->childs[0]);
-							stkStack[depth++]=sStkNN(p.a->childs[0],p.b->childs[1]);
-							stkStack[depth++]=sStkNN(p.a->childs[1],p.b->childs[1]);
-						}
-						else
-						{
-							stkStack[depth++]=sStkNN(p.a->childs[0],p.b);
-							stkStack[depth++]=sStkNN(p.a->childs[1],p.b);
-						}
+						stkStack[depth++] = sStkNN(p.a->childs[0], p.b->childs[0]);
+						stkStack[depth++] = sStkNN(p.a->childs[1], p.b->childs[0]);
+						stkStack[depth++] = sStkNN(p.a->childs[0], p.b->childs[1]);
+						stkStack[depth++] = sStkNN(p.a->childs[1], p.b->childs[1]);
 					}
 					else
 					{
-						if(p.b->isinternal())
-						{
-							stkStack[depth++]=sStkNN(p.a,p.b->childs[0]);
-							stkStack[depth++]=sStkNN(p.a,p.b->childs[1]);
-						}
-						else
-						{
-							policy.Process(p.a,p.b);
-						}
+						stkStack[depth++] = sStkNN(p.a->childs[0], p.b);
+						stkStack[depth++] = sStkNN(p.a->childs[1], p.b);
 					}
 				}
-			} while(depth);
-		}
+				else
+				{
+					if (p.b->isinternal())
+					{
+						stkStack[depth++] = sStkNN(p.a, p.b->childs[0]);
+						stkStack[depth++] = sStkNN(p.a, p.b->childs[1]);
+					}
+					else
+					{
+						policy.Process(p.a, p.b);
+					}
+				}
+			}
+		} while (depth);
+	}
 }
 
-
-
 DBVT_PREFIX
-inline void		btDbvt::collideTTpersistentStack(	const btDbvtNode* root0,
-								  const btDbvtNode* root1,
-								  DBVT_IPOLICY)
+inline void btDbvt::collideTTpersistentStack(const btDbvtNode* root0,
+											 const btDbvtNode* root1,
+											 DBVT_IPOLICY)
 {
 	DBVT_CHECKTYPE
-		if(root0&&root1)
+	if (root0 && root1)
+	{
+		int depth = 1;
+		int treshold = DOUBLE_STACKSIZE - 4;
+
+		m_stkStack.resize(DOUBLE_STACKSIZE);
+		m_stkStack[0] = sStkNN(root0, root1);
+		do
 		{
-			int								depth=1;
-			int								treshold=DOUBLE_STACKSIZE-4;
-			
-			m_stkStack.resize(DOUBLE_STACKSIZE);
-			m_stkStack[0]=sStkNN(root0,root1);
-			do	{		
-				sStkNN	p=m_stkStack[--depth];
-				if(depth>treshold)
+			sStkNN p = m_stkStack[--depth];
+			if (depth > treshold)
+			{
+				m_stkStack.resize(m_stkStack.size() * 2);
+				treshold = m_stkStack.size() - 4;
+			}
+			if (p.a == p.b)
+			{
+				if (p.a->isinternal())
 				{
-					m_stkStack.resize(m_stkStack.size()*2);
-					treshold=m_stkStack.size()-4;
+					m_stkStack[depth++] = sStkNN(p.a->childs[0], p.a->childs[0]);
+					m_stkStack[depth++] = sStkNN(p.a->childs[1], p.a->childs[1]);
+					m_stkStack[depth++] = sStkNN(p.a->childs[0], p.a->childs[1]);
 				}
-				if(p.a==p.b)
+			}
+			else if (Intersect(p.a->volume, p.b->volume))
+			{
+				if (p.a->isinternal())
 				{
-					if(p.a->isinternal())
+					if (p.b->isinternal())
 					{
-						m_stkStack[depth++]=sStkNN(p.a->childs[0],p.a->childs[0]);
-						m_stkStack[depth++]=sStkNN(p.a->childs[1],p.a->childs[1]);
-						m_stkStack[depth++]=sStkNN(p.a->childs[0],p.a->childs[1]);
-					}
-				}
-				else if(Intersect(p.a->volume,p.b->volume))
-				{
-					if(p.a->isinternal())
-					{
-						if(p.b->isinternal())
-						{
-							m_stkStack[depth++]=sStkNN(p.a->childs[0],p.b->childs[0]);
-							m_stkStack[depth++]=sStkNN(p.a->childs[1],p.b->childs[0]);
-							m_stkStack[depth++]=sStkNN(p.a->childs[0],p.b->childs[1]);
-							m_stkStack[depth++]=sStkNN(p.a->childs[1],p.b->childs[1]);
-						}
-						else
-						{
-							m_stkStack[depth++]=sStkNN(p.a->childs[0],p.b);
-							m_stkStack[depth++]=sStkNN(p.a->childs[1],p.b);
-						}
+						m_stkStack[depth++] = sStkNN(p.a->childs[0], p.b->childs[0]);
+						m_stkStack[depth++] = sStkNN(p.a->childs[1], p.b->childs[0]);
+						m_stkStack[depth++] = sStkNN(p.a->childs[0], p.b->childs[1]);
+						m_stkStack[depth++] = sStkNN(p.a->childs[1], p.b->childs[1]);
 					}
 					else
 					{
-						if(p.b->isinternal())
-						{
-							m_stkStack[depth++]=sStkNN(p.a,p.b->childs[0]);
-							m_stkStack[depth++]=sStkNN(p.a,p.b->childs[1]);
-						}
-						else
-						{
-							policy.Process(p.a,p.b);
-						}
+						m_stkStack[depth++] = sStkNN(p.a->childs[0], p.b);
+						m_stkStack[depth++] = sStkNN(p.a->childs[1], p.b);
 					}
 				}
-			} while(depth);
-		}
+				else
+				{
+					if (p.b->isinternal())
+					{
+						m_stkStack[depth++] = sStkNN(p.a, p.b->childs[0]);
+						m_stkStack[depth++] = sStkNN(p.a, p.b->childs[1]);
+					}
+					else
+					{
+						policy.Process(p.a, p.b);
+					}
+				}
+			}
+		} while (depth);
+	}
 }
 
 #if 0
@@ -929,33 +968,35 @@ inline void		btDbvt::collideTT(	const btDbvtNode* root0,
 	const btTransform	xform=xform0.inverse()*xform1;
 	collideTT(root0,root1,xform,policy);
 }
-#endif 
+#endif
 
 DBVT_PREFIX
-inline void		btDbvt::collideTV(	const btDbvtNode* root,
-								  const btDbvtVolume& vol,
-								  DBVT_IPOLICY) const
+inline void btDbvt::collideTV(const btDbvtNode* root,
+							  const btDbvtVolume& vol,
+							  DBVT_IPOLICY) const
 {
 	DBVT_CHECKTYPE
-	if(root)
+	if (root)
 	{
-		ATTRIBUTE_ALIGNED16(btDbvtVolume)		volume(vol);
-		btAlignedObjectArray<const btDbvtNode*>	stack;
+		ATTRIBUTE_ALIGNED16(btDbvtVolume)
+		volume(vol);
+		btAlignedObjectArray<const btDbvtNode*> stack;
 		stack.resize(0);
 #ifndef BT_DISABLE_STACK_TEMP_MEMORY
-		char tempmemory[SIMPLE_STACKSIZE*sizeof(const btDbvtNode*)];
+		char tempmemory[SIMPLE_STACKSIZE * sizeof(const btDbvtNode*)];
 		stack.initializeFromBuffer(tempmemory, 0, SIMPLE_STACKSIZE);
 #else
 		stack.reserve(SIMPLE_STACKSIZE);
-#endif //BT_DISABLE_STACK_TEMP_MEMORY
+#endif  //BT_DISABLE_STACK_TEMP_MEMORY
 
 		stack.push_back(root);
-		do	{
-			const btDbvtNode*	n=stack[stack.size()-1];
+		do
+		{
+			const btDbvtNode* n = stack[stack.size() - 1];
 			stack.pop_back();
-			if(Intersect(n->volume,volume))
+			if (Intersect(n->volume, volume))
 			{
-				if(n->isinternal())
+				if (n->isinternal())
 				{
 					stack.push_back(n->childs[0]);
 					stack.push_back(n->childs[1]);
@@ -965,30 +1006,32 @@ inline void		btDbvt::collideTV(	const btDbvtNode* root,
 					policy.Process(n);
 				}
 			}
-		} while(stack.size()>0);
+		} while (stack.size() > 0);
 	}
 }
 
 //
 DBVT_PREFIX
-inline void		btDbvt::collideTVNoStackAlloc(	const btDbvtNode* root,
-											 const btDbvtVolume& vol,
-											 btNodeStack& stack,
-											 DBVT_IPOLICY) const
+inline void btDbvt::collideTVNoStackAlloc(const btDbvtNode* root,
+										  const btDbvtVolume& vol,
+										  btNodeStack& stack,
+										  DBVT_IPOLICY) const
 {
 	DBVT_CHECKTYPE
-	if(root)
+	if (root)
 	{
-		ATTRIBUTE_ALIGNED16(btDbvtVolume)		volume(vol);
+		ATTRIBUTE_ALIGNED16(btDbvtVolume)
+		volume(vol);
 		stack.resize(0);
 		stack.reserve(SIMPLE_STACKSIZE);
 		stack.push_back(root);
-		do	{
-			const btDbvtNode*	n=stack[stack.size()-1];
+		do
+		{
+			const btDbvtNode* n = stack[stack.size() - 1];
 			stack.pop_back();
-			if(Intersect(n->volume,volume))
+			if (Intersect(n->volume, volume))
 			{
-				if(n->isinternal())
+				if (n->isinternal())
 				{
 					stack.push_back(n->childs[0]);
 					stack.push_back(n->childs[1]);
@@ -998,328 +1041,346 @@ inline void		btDbvt::collideTVNoStackAlloc(	const btDbvtNode* root,
 					policy.Process(n);
 				}
 			}
-		} while(stack.size()>0);
+		} while (stack.size() > 0);
 	}
 }
 
-
 DBVT_PREFIX
-inline void		btDbvt::rayTestInternal(	const btDbvtNode* root,
-								const btVector3& rayFrom,
-								const btVector3& rayTo,
-								const btVector3& rayDirectionInverse,
-								unsigned int signs[3],
-								btScalar lambda_max,
-								const btVector3& aabbMin,
-								const btVector3& aabbMax,
-                                btAlignedObjectArray<const btDbvtNode*>& stack,
-                                DBVT_IPOLICY ) const
+inline void btDbvt::rayTestInternal(const btDbvtNode* root,
+									const btVector3& rayFrom,
+									const btVector3& rayTo,
+									const btVector3& rayDirectionInverse,
+									unsigned int signs[3],
+									btScalar lambda_max,
+									const btVector3& aabbMin,
+									const btVector3& aabbMax,
+									btAlignedObjectArray<const btDbvtNode*>& stack,
+									DBVT_IPOLICY) const
 {
-        (void) rayTo;
+	(void)rayTo;
 	DBVT_CHECKTYPE
-	if(root)
+	if (root)
 	{
 		btVector3 resultNormal;
 
-		int								depth=1;
-		int								treshold=DOUBLE_STACKSIZE-2;
+		int depth = 1;
+		int treshold = DOUBLE_STACKSIZE - 2;
 		stack.resize(DOUBLE_STACKSIZE);
-		stack[0]=root;
+		stack[0] = root;
 		btVector3 bounds[2];
-		do	
+		do
 		{
-			const btDbvtNode*	node=stack[--depth];
-			bounds[0] = node->volume.Mins()-aabbMax;
-			bounds[1] = node->volume.Maxs()-aabbMin;
-			btScalar tmin=1.f,lambda_min=0.f;
-			unsigned int result1=false;
-			result1 = btRayAabb2(rayFrom,rayDirectionInverse,signs,bounds,tmin,lambda_min,lambda_max);
-			if(result1)
+			const btDbvtNode* node = stack[--depth];
+			bounds[0] = node->volume.Mins() - aabbMax;
+			bounds[1] = node->volume.Maxs() - aabbMin;
+			btScalar tmin = 1.f, lambda_min = 0.f;
+			unsigned int result1 = false;
+			result1 = btRayAabb2(rayFrom, rayDirectionInverse, signs, bounds, tmin, lambda_min, lambda_max);
+			if (result1)
 			{
-				if(node->isinternal())
+				if (node->isinternal())
 				{
-					if(depth>treshold)
+					if (depth > treshold)
 					{
-						stack.resize(stack.size()*2);
-						treshold=stack.size()-2;
+						stack.resize(stack.size() * 2);
+						treshold = stack.size() - 2;
 					}
-					stack[depth++]=node->childs[0];
-					stack[depth++]=node->childs[1];
+					stack[depth++] = node->childs[0];
+					stack[depth++] = node->childs[1];
 				}
 				else
 				{
 					policy.Process(node);
 				}
 			}
-		} while(depth);
+		} while (depth);
 	}
 }
 
 //
 DBVT_PREFIX
-inline void		btDbvt::rayTest(	const btDbvtNode* root,
-								const btVector3& rayFrom,
-								const btVector3& rayTo,
+inline void btDbvt::rayTest(const btDbvtNode* root,
+							const btVector3& rayFrom,
+							const btVector3& rayTo,
+							DBVT_IPOLICY)
+{
+	DBVT_CHECKTYPE
+	if (root)
+	{
+		btVector3 rayDir = (rayTo - rayFrom);
+		rayDir.normalize();
+
+		///what about division by zero? --> just set rayDirection[i] to INF/BT_LARGE_FLOAT
+		btVector3 rayDirectionInverse;
+		rayDirectionInverse[0] = rayDir[0] == btScalar(0.0) ? btScalar(BT_LARGE_FLOAT) : btScalar(1.0) / rayDir[0];
+		rayDirectionInverse[1] = rayDir[1] == btScalar(0.0) ? btScalar(BT_LARGE_FLOAT) : btScalar(1.0) / rayDir[1];
+		rayDirectionInverse[2] = rayDir[2] == btScalar(0.0) ? btScalar(BT_LARGE_FLOAT) : btScalar(1.0) / rayDir[2];
+		unsigned int signs[3] = {rayDirectionInverse[0] < 0.0, rayDirectionInverse[1] < 0.0, rayDirectionInverse[2] < 0.0};
+
+		btScalar lambda_max = rayDir.dot(rayTo - rayFrom);
+
+		btVector3 resultNormal;
+
+		btAlignedObjectArray<const btDbvtNode*> stack;
+
+		int depth = 1;
+		int treshold = DOUBLE_STACKSIZE - 2;
+
+		char tempmemory[DOUBLE_STACKSIZE * sizeof(const btDbvtNode*)];
+#ifndef BT_DISABLE_STACK_TEMP_MEMORY
+		stack.initializeFromBuffer(tempmemory, DOUBLE_STACKSIZE, DOUBLE_STACKSIZE);
+#else   //BT_DISABLE_STACK_TEMP_MEMORY
+		stack.resize(DOUBLE_STACKSIZE);
+#endif  //BT_DISABLE_STACK_TEMP_MEMORY
+		stack[0] = root;
+		btVector3 bounds[2];
+		do
+		{
+			const btDbvtNode* node = stack[--depth];
+
+			bounds[0] = node->volume.Mins();
+			bounds[1] = node->volume.Maxs();
+
+			btScalar tmin = 1.f, lambda_min = 0.f;
+			unsigned int result1 = btRayAabb2(rayFrom, rayDirectionInverse, signs, bounds, tmin, lambda_min, lambda_max);
+
+#ifdef COMPARE_BTRAY_AABB2
+			btScalar param = 1.f;
+			bool result2 = btRayAabb(rayFrom, rayTo, node->volume.Mins(), node->volume.Maxs(), param, resultNormal);
+			btAssert(result1 == result2);
+#endif  //TEST_BTRAY_AABB2
+
+			if (result1)
+			{
+				if (node->isinternal())
+				{
+					if (depth > treshold)
+					{
+						stack.resize(stack.size() * 2);
+						treshold = stack.size() - 2;
+					}
+					stack[depth++] = node->childs[0];
+					stack[depth++] = node->childs[1];
+				}
+				else
+				{
+					policy.Process(node);
+				}
+			}
+		} while (depth);
+	}
+}
+
+//
+DBVT_PREFIX
+inline void btDbvt::collideKDOP(const btDbvtNode* root,
+								const btVector3* normals,
+								const btScalar* offsets,
+								int count,
 								DBVT_IPOLICY)
 {
 	DBVT_CHECKTYPE
-		if(root)
+	if (root)
+	{
+		const int inside = (1 << count) - 1;
+		btAlignedObjectArray<sStkNP> stack;
+		int signs[sizeof(unsigned) * 8];
+		btAssert(count < int(sizeof(signs) / sizeof(signs[0])));
+		for (int i = 0; i < count; ++i)
 		{
-			btVector3 rayDir = (rayTo-rayFrom);
-			rayDir.normalize ();
-
-			///what about division by zero? --> just set rayDirection[i] to INF/BT_LARGE_FLOAT
-			btVector3 rayDirectionInverse;
-			rayDirectionInverse[0] = rayDir[0] == btScalar(0.0) ? btScalar(BT_LARGE_FLOAT) : btScalar(1.0) / rayDir[0];
-			rayDirectionInverse[1] = rayDir[1] == btScalar(0.0) ? btScalar(BT_LARGE_FLOAT) : btScalar(1.0) / rayDir[1];
-			rayDirectionInverse[2] = rayDir[2] == btScalar(0.0) ? btScalar(BT_LARGE_FLOAT) : btScalar(1.0) / rayDir[2];
-			unsigned int signs[3] = { rayDirectionInverse[0] < 0.0, rayDirectionInverse[1] < 0.0, rayDirectionInverse[2] < 0.0};
-
-			btScalar lambda_max = rayDir.dot(rayTo-rayFrom);
-
-			btVector3 resultNormal;
-
-			btAlignedObjectArray<const btDbvtNode*>	stack;
-
-			int								depth=1;
-			int								treshold=DOUBLE_STACKSIZE-2;
-
-			char tempmemory[DOUBLE_STACKSIZE * sizeof(const btDbvtNode*)];
-#ifndef BT_DISABLE_STACK_TEMP_MEMORY
-			stack.initializeFromBuffer(tempmemory, DOUBLE_STACKSIZE, DOUBLE_STACKSIZE);
-#else//BT_DISABLE_STACK_TEMP_MEMORY
-			stack.resize(DOUBLE_STACKSIZE);
-#endif //BT_DISABLE_STACK_TEMP_MEMORY
-			stack[0]=root;
-			btVector3 bounds[2];
-			do	{
-				const btDbvtNode*	node=stack[--depth];
-
-				bounds[0] = node->volume.Mins();
-				bounds[1] = node->volume.Maxs();
-				
-				btScalar tmin=1.f,lambda_min=0.f;
-				unsigned int result1 = btRayAabb2(rayFrom,rayDirectionInverse,signs,bounds,tmin,lambda_min,lambda_max);
-
-#ifdef COMPARE_BTRAY_AABB2
-				btScalar param=1.f;
-				bool result2 = btRayAabb(rayFrom,rayTo,node->volume.Mins(),node->volume.Maxs(),param,resultNormal);
-				btAssert(result1 == result2);
-#endif //TEST_BTRAY_AABB2
-
-				if(result1)
-				{
-					if(node->isinternal())
-					{
-						if(depth>treshold)
-						{
-							stack.resize(stack.size()*2);
-							treshold=stack.size()-2;
-						}
-						stack[depth++]=node->childs[0];
-						stack[depth++]=node->childs[1];
-					}
-					else
-					{
-						policy.Process(node);
-					}
-				}
-			} while(depth);
-
+			signs[i] = ((normals[i].x() >= 0) ? 1 : 0) +
+					   ((normals[i].y() >= 0) ? 2 : 0) +
+					   ((normals[i].z() >= 0) ? 4 : 0);
 		}
-}
-
-//
-DBVT_PREFIX
-inline void		btDbvt::collideKDOP(const btDbvtNode* root,
-									const btVector3* normals,
-									const btScalar* offsets,
-									int count,
-									DBVT_IPOLICY)
-{
-	DBVT_CHECKTYPE
-		if(root)
+		stack.reserve(SIMPLE_STACKSIZE);
+		stack.push_back(sStkNP(root, 0));
+		do
 		{
-			const int						inside=(1<<count)-1;
-			btAlignedObjectArray<sStkNP>	stack;
-			int								signs[sizeof(unsigned)*8];
-			btAssert(count<int (sizeof(signs)/sizeof(signs[0])));
-			for(int i=0;i<count;++i)
+			sStkNP se = stack[stack.size() - 1];
+			bool out = false;
+			stack.pop_back();
+			for (int i = 0, j = 1; (!out) && (i < count); ++i, j <<= 1)
 			{
-				signs[i]=	((normals[i].x()>=0)?1:0)+
-					((normals[i].y()>=0)?2:0)+
-					((normals[i].z()>=0)?4:0);
+				if (0 == (se.mask & j))
+				{
+					const int side = se.node->volume.Classify(normals[i], offsets[i], signs[i]);
+					switch (side)
+					{
+						case -1:
+							out = true;
+							break;
+						case +1:
+							se.mask |= j;
+							break;
+					}
+				}
 			}
-			stack.reserve(SIMPLE_STACKSIZE);
-			stack.push_back(sStkNP(root,0));
-			do	{
-				sStkNP	se=stack[stack.size()-1];
-				bool	out=false;
-				stack.pop_back();
-				for(int i=0,j=1;(!out)&&(i<count);++i,j<<=1)
-				{
-					if(0==(se.mask&j))
-					{
-						const int	side=se.node->volume.Classify(normals[i],offsets[i],signs[i]);
-						switch(side)
-						{
-						case	-1:	out=true;break;
-						case	+1:	se.mask|=j;break;
-						}
-					}
-				}
-				if(!out)
-				{
-					if((se.mask!=inside)&&(se.node->isinternal()))
-					{
-						stack.push_back(sStkNP(se.node->childs[0],se.mask));
-						stack.push_back(sStkNP(se.node->childs[1],se.mask));
-					}
-					else
-					{
-						if(policy.AllLeaves(se.node)) enumLeaves(se.node,policy);
-					}
-				}
-			} while(stack.size());
-		}
-}
-
-//
-DBVT_PREFIX
-inline void		btDbvt::collideOCL(	const btDbvtNode* root,
-								   const btVector3* normals,
-								   const btScalar* offsets,
-								   const btVector3& sortaxis,
-								   int count,
-								   DBVT_IPOLICY,
-								   bool fsort)
-{
-	DBVT_CHECKTYPE
-		if(root)
-		{
-			const unsigned					srtsgns=(sortaxis[0]>=0?1:0)+
-				(sortaxis[1]>=0?2:0)+
-				(sortaxis[2]>=0?4:0);
-			const int						inside=(1<<count)-1;
-			btAlignedObjectArray<sStkNPS>	stock;
-			btAlignedObjectArray<int>		ifree;
-			btAlignedObjectArray<int>		stack;
-			int								signs[sizeof(unsigned)*8];
-			btAssert(count<int (sizeof(signs)/sizeof(signs[0])));
-			for(int i=0;i<count;++i)
+			if (!out)
 			{
-				signs[i]=	((normals[i].x()>=0)?1:0)+
-					((normals[i].y()>=0)?2:0)+
-					((normals[i].z()>=0)?4:0);
+				if ((se.mask != inside) && (se.node->isinternal()))
+				{
+					stack.push_back(sStkNP(se.node->childs[0], se.mask));
+					stack.push_back(sStkNP(se.node->childs[1], se.mask));
+				}
+				else
+				{
+					if (policy.AllLeaves(se.node)) enumLeaves(se.node, policy);
+				}
 			}
-			stock.reserve(SIMPLE_STACKSIZE);
-			stack.reserve(SIMPLE_STACKSIZE);
-			ifree.reserve(SIMPLE_STACKSIZE);
-			stack.push_back(allocate(ifree,stock,sStkNPS(root,0,root->volume.ProjectMinimum(sortaxis,srtsgns))));
-			do	{
-				const int	id=stack[stack.size()-1];
-				sStkNPS		se=stock[id];
-				stack.pop_back();ifree.push_back(id);
-				if(se.mask!=inside)
-				{
-					bool	out=false;
-					for(int i=0,j=1;(!out)&&(i<count);++i,j<<=1)
-					{
-						if(0==(se.mask&j))
-						{
-							const int	side=se.node->volume.Classify(normals[i],offsets[i],signs[i]);
-							switch(side)
-							{
-							case	-1:	out=true;break;
-							case	+1:	se.mask|=j;break;
-							}
-						}
-					}
-					if(out) continue;
-				}
-				if(policy.Descent(se.node))
-				{
-					if(se.node->isinternal())
-					{
-						const btDbvtNode* pns[]={	se.node->childs[0],se.node->childs[1]};
-						sStkNPS		nes[]={	sStkNPS(pns[0],se.mask,pns[0]->volume.ProjectMinimum(sortaxis,srtsgns)),
-							sStkNPS(pns[1],se.mask,pns[1]->volume.ProjectMinimum(sortaxis,srtsgns))};
-						const int	q=nes[0].value<nes[1].value?1:0;				
-						int			j=stack.size();
-						if(fsort&&(j>0))
-						{
-							/* Insert 0	*/ 
-							j=nearest(&stack[0],&stock[0],nes[q].value,0,stack.size());
-							stack.push_back(0);
-							
-							//void * memmove ( void * destination, const void * source, size_t num );
-							
-#if DBVT_USE_MEMMOVE
-                     {
-                     int num_items_to_move = stack.size()-1-j;
-                     if(num_items_to_move > 0)
-                        memmove(&stack[j+1],&stack[j],sizeof(int)*num_items_to_move);
-                     }
-#else
-                     for(int k=stack.size()-1;k>j;--k) {
-								stack[k]=stack[k-1];
-                     }
-#endif
-							stack[j]=allocate(ifree,stock,nes[q]);
-							/* Insert 1	*/ 
-							j=nearest(&stack[0],&stock[0],nes[1-q].value,j,stack.size());
-							stack.push_back(0);
-#if DBVT_USE_MEMMOVE
-                     {
-                     int num_items_to_move = stack.size()-1-j;
-                     if(num_items_to_move > 0)
-                        memmove(&stack[j+1],&stack[j],sizeof(int)*num_items_to_move);
-                     }
-#else
-                     for(int k=stack.size()-1;k>j;--k) {
-                        stack[k]=stack[k-1];
-                     }
-#endif
-							stack[j]=allocate(ifree,stock,nes[1-q]);
-						}
-						else
-						{
-							stack.push_back(allocate(ifree,stock,nes[q]));
-							stack.push_back(allocate(ifree,stock,nes[1-q]));
-						}
-					}
-					else
-					{
-						policy.Process(se.node,se.value);
-					}
-				}
-			} while(stack.size());
-		}
+		} while (stack.size());
+	}
 }
 
 //
 DBVT_PREFIX
-inline void		btDbvt::collideTU(	const btDbvtNode* root,
-								  DBVT_IPOLICY)
+inline void btDbvt::collideOCL(const btDbvtNode* root,
+							   const btVector3* normals,
+							   const btScalar* offsets,
+							   const btVector3& sortaxis,
+							   int count,
+							   DBVT_IPOLICY,
+							   bool fsort)
 {
 	DBVT_CHECKTYPE
-		if(root)
+	if (root)
+	{
+		const unsigned srtsgns = (sortaxis[0] >= 0 ? 1 : 0) +
+								 (sortaxis[1] >= 0 ? 2 : 0) +
+								 (sortaxis[2] >= 0 ? 4 : 0);
+		const int inside = (1 << count) - 1;
+		btAlignedObjectArray<sStkNPS> stock;
+		btAlignedObjectArray<int> ifree;
+		btAlignedObjectArray<int> stack;
+		int signs[sizeof(unsigned) * 8];
+		btAssert(count < int(sizeof(signs) / sizeof(signs[0])));
+		for (int i = 0; i < count; ++i)
 		{
-			btAlignedObjectArray<const btDbvtNode*>	stack;
-			stack.reserve(SIMPLE_STACKSIZE);
-			stack.push_back(root);
-			do	{
-				const btDbvtNode*	n=stack[stack.size()-1];
-				stack.pop_back();
-				if(policy.Descent(n))
-				{
-					if(n->isinternal())
-					{ stack.push_back(n->childs[0]);stack.push_back(n->childs[1]); }
-					else
-					{ policy.Process(n); }
-				}
-			} while(stack.size()>0);
+			signs[i] = ((normals[i].x() >= 0) ? 1 : 0) +
+					   ((normals[i].y() >= 0) ? 2 : 0) +
+					   ((normals[i].z() >= 0) ? 4 : 0);
 		}
+		stock.reserve(SIMPLE_STACKSIZE);
+		stack.reserve(SIMPLE_STACKSIZE);
+		ifree.reserve(SIMPLE_STACKSIZE);
+		stack.push_back(allocate(ifree, stock, sStkNPS(root, 0, root->volume.ProjectMinimum(sortaxis, srtsgns))));
+		do
+		{
+			const int id = stack[stack.size() - 1];
+			sStkNPS se = stock[id];
+			stack.pop_back();
+			ifree.push_back(id);
+			if (se.mask != inside)
+			{
+				bool out = false;
+				for (int i = 0, j = 1; (!out) && (i < count); ++i, j <<= 1)
+				{
+					if (0 == (se.mask & j))
+					{
+						const int side = se.node->volume.Classify(normals[i], offsets[i], signs[i]);
+						switch (side)
+						{
+							case -1:
+								out = true;
+								break;
+							case +1:
+								se.mask |= j;
+								break;
+						}
+					}
+				}
+				if (out) continue;
+			}
+			if (policy.Descent(se.node))
+			{
+				if (se.node->isinternal())
+				{
+					const btDbvtNode* pns[] = {se.node->childs[0], se.node->childs[1]};
+					sStkNPS nes[] = {sStkNPS(pns[0], se.mask, pns[0]->volume.ProjectMinimum(sortaxis, srtsgns)),
+									 sStkNPS(pns[1], se.mask, pns[1]->volume.ProjectMinimum(sortaxis, srtsgns))};
+					const int q = nes[0].value < nes[1].value ? 1 : 0;
+					int j = stack.size();
+					if (fsort && (j > 0))
+					{
+						/* Insert 0	*/
+						j = nearest(&stack[0], &stock[0], nes[q].value, 0, stack.size());
+						stack.push_back(0);
+
+						//void * memmove ( void * destination, const void * source, size_t num );
+
+#if DBVT_USE_MEMMOVE
+						{
+							int num_items_to_move = stack.size() - 1 - j;
+							if (num_items_to_move > 0)
+								memmove(&stack[j + 1], &stack[j], sizeof(int) * num_items_to_move);
+						}
+#else
+						for (int k = stack.size() - 1; k > j; --k)
+						{
+							stack[k] = stack[k - 1];
+						}
+#endif
+						stack[j] = allocate(ifree, stock, nes[q]);
+						/* Insert 1	*/
+						j = nearest(&stack[0], &stock[0], nes[1 - q].value, j, stack.size());
+						stack.push_back(0);
+#if DBVT_USE_MEMMOVE
+						{
+							int num_items_to_move = stack.size() - 1 - j;
+							if (num_items_to_move > 0)
+								memmove(&stack[j + 1], &stack[j], sizeof(int) * num_items_to_move);
+						}
+#else
+						for (int k = stack.size() - 1; k > j; --k)
+						{
+							stack[k] = stack[k - 1];
+						}
+#endif
+						stack[j] = allocate(ifree, stock, nes[1 - q]);
+					}
+					else
+					{
+						stack.push_back(allocate(ifree, stock, nes[q]));
+						stack.push_back(allocate(ifree, stock, nes[1 - q]));
+					}
+				}
+				else
+				{
+					policy.Process(se.node, se.value);
+				}
+			}
+		} while (stack.size());
+	}
+}
+
+//
+DBVT_PREFIX
+inline void btDbvt::collideTU(const btDbvtNode* root,
+							  DBVT_IPOLICY)
+{
+	DBVT_CHECKTYPE
+	if (root)
+	{
+		btAlignedObjectArray<const btDbvtNode*> stack;
+		stack.reserve(SIMPLE_STACKSIZE);
+		stack.push_back(root);
+		do
+		{
+			const btDbvtNode* n = stack[stack.size() - 1];
+			stack.pop_back();
+			if (policy.Descent(n))
+			{
+				if (n->isinternal())
+				{
+					stack.push_back(n->childs[0]);
+					stack.push_back(n->childs[1]);
+				}
+				else
+				{
+					policy.Process(n);
+				}
+			}
+		} while (stack.size() > 0);
+	}
 }
 
 //
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDbvtBroadphase.cpp b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDbvtBroadphase.cpp
index 14cd1a31ea1..7b39dbdc0fe 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDbvtBroadphase.cpp
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDbvtBroadphase.cpp
@@ -22,28 +22,27 @@ btScalar gDbvtMargin = btScalar(0.05);
 // Profiling
 //
 
-#if DBVT_BP_PROFILE||DBVT_BP_ENABLE_BENCHMARK
+#if DBVT_BP_PROFILE || DBVT_BP_ENABLE_BENCHMARK
 #include <stdio.h>
 #endif
 
 #if DBVT_BP_PROFILE
-struct	ProfileScope
+struct ProfileScope
 {
-	__forceinline ProfileScope(btClock& clock,unsigned long& value) :
-	m_clock(&clock),m_value(&value),m_base(clock.getTimeMicroseconds())
+	__forceinline ProfileScope(btClock& clock, unsigned long& value) : m_clock(&clock), m_value(&value), m_base(clock.getTimeMicroseconds())
 	{
 	}
 	__forceinline ~ProfileScope()
 	{
-		(*m_value)+=m_clock->getTimeMicroseconds()-m_base;
+		(*m_value) += m_clock->getTimeMicroseconds() - m_base;
 	}
-	btClock*		m_clock;
-	unsigned long*	m_value;
-	unsigned long	m_base;
+	btClock* m_clock;
+	unsigned long* m_value;
+	unsigned long m_base;
 };
-#define	SPC(_value_)	ProfileScope	spc_scope(m_clock,_value_)
+#define SPC(_value_) ProfileScope spc_scope(m_clock, _value_)
 #else
-#define	SPC(_value_)
+#define SPC(_value_)
 #endif
 
 //
@@ -52,66 +51,75 @@ struct	ProfileScope
 
 //
 template <typename T>
-static inline void	listappend(T* item,T*& list)
+static inline void listappend(T* item, T*& list)
 {
-	item->links[0]=0;
-	item->links[1]=list;
-	if(list) list->links[0]=item;
-	list=item;
+	item->links[0] = 0;
+	item->links[1] = list;
+	if (list) list->links[0] = item;
+	list = item;
 }
 
 //
 template <typename T>
-static inline void	listremove(T* item,T*& list)
+static inline void listremove(T* item, T*& list)
 {
-	if(item->links[0]) item->links[0]->links[1]=item->links[1]; else list=item->links[1];
-	if(item->links[1]) item->links[1]->links[0]=item->links[0];
+	if (item->links[0])
+		item->links[0]->links[1] = item->links[1];
+	else
+		list = item->links[1];
+	if (item->links[1]) item->links[1]->links[0] = item->links[0];
 }
 
 //
 template <typename T>
-static inline int	listcount(T* root)
+static inline int listcount(T* root)
 {
-	int	n=0;
-	while(root) { ++n;root=root->links[1]; }
-	return(n);
+	int n = 0;
+	while (root)
+	{
+		++n;
+		root = root->links[1];
+	}
+	return (n);
 }
 
 //
 template <typename T>
-static inline void	clear(T& value)
+static inline void clear(T& value)
 {
-	static const struct ZeroDummy : T {} zerodummy;
-	value=zerodummy;
+	static const struct ZeroDummy : T
+	{
+	} zerodummy;
+	value = zerodummy;
 }
 
 //
 // Colliders
 //
 
-/* Tree collider	*/ 
-struct	btDbvtTreeCollider : btDbvt::ICollide
+/* Tree collider	*/
+struct btDbvtTreeCollider : btDbvt::ICollide
 {
-	btDbvtBroadphase*	pbp;
-	btDbvtProxy*		proxy;
+	btDbvtBroadphase* pbp;
+	btDbvtProxy* proxy;
 	btDbvtTreeCollider(btDbvtBroadphase* p) : pbp(p) {}
-	void	Process(const btDbvtNode* na,const btDbvtNode* nb)
+	void Process(const btDbvtNode* na, const btDbvtNode* nb)
 	{
-		if(na!=nb)
+		if (na != nb)
 		{
-			btDbvtProxy*	pa=(btDbvtProxy*)na->data;
-			btDbvtProxy*	pb=(btDbvtProxy*)nb->data;
+			btDbvtProxy* pa = (btDbvtProxy*)na->data;
+			btDbvtProxy* pb = (btDbvtProxy*)nb->data;
 #if DBVT_BP_SORTPAIRS
-			if(pa->m_uniqueId>pb->m_uniqueId) 
-				btSwap(pa,pb);
+			if (pa->m_uniqueId > pb->m_uniqueId)
+				btSwap(pa, pb);
 #endif
-			pbp->m_paircache->addOverlappingPair(pa,pb);
+			pbp->m_paircache->addOverlappingPair(pa, pb);
 			++pbp->m_newpairs;
 		}
 	}
-	void	Process(const btDbvtNode* n)
+	void Process(const btDbvtNode* n)
 	{
-		Process(n,proxy->leaf);
+		Process(n, proxy->leaf);
 	}
 };
 
@@ -122,31 +130,31 @@ struct	btDbvtTreeCollider : btDbvt::ICollide
 //
 btDbvtBroadphase::btDbvtBroadphase(btOverlappingPairCache* paircache)
 {
-	m_deferedcollide	=	false;
-	m_needcleanup		=	true;
-	m_releasepaircache	=	(paircache!=0)?false:true;
-	m_prediction		=	0;
-	m_stageCurrent		=	0;
-	m_fixedleft			=	0;
-	m_fupdates			=	1;
-	m_dupdates			=	0;
-	m_cupdates			=	10;
-	m_newpairs			=	1;
-	m_updates_call		=	0;
-	m_updates_done		=	0;
-	m_updates_ratio		=	0;
-	m_paircache			=	paircache? paircache	: new(btAlignedAlloc(sizeof(btHashedOverlappingPairCache),16)) btHashedOverlappingPairCache();
-	m_gid				=	0;
-	m_pid				=	0;
-	m_cid				=	0;
-	for(int i=0;i<=STAGECOUNT;++i)
+	m_deferedcollide = false;
+	m_needcleanup = true;
+	m_releasepaircache = (paircache != 0) ? false : true;
+	m_prediction = 0;
+	m_stageCurrent = 0;
+	m_fixedleft = 0;
+	m_fupdates = 1;
+	m_dupdates = 0;
+	m_cupdates = 10;
+	m_newpairs = 1;
+	m_updates_call = 0;
+	m_updates_done = 0;
+	m_updates_ratio = 0;
+	m_paircache = paircache ? paircache : new (btAlignedAlloc(sizeof(btHashedOverlappingPairCache), 16)) btHashedOverlappingPairCache();
+	m_gid = 0;
+	m_pid = 0;
+	m_cid = 0;
+	for (int i = 0; i <= STAGECOUNT; ++i)
 	{
-		m_stageRoots[i]=0;
+		m_stageRoots[i] = 0;
 	}
 #if BT_THREADSAFE
-    m_rayTestStacks.resize(BT_MAX_THREAD_COUNT);
+	m_rayTestStacks.resize(BT_MAX_THREAD_COUNT);
 #else
-    m_rayTestStacks.resize(1);
+	m_rayTestStacks.resize(1);
 #endif
 #if DBVT_BP_PROFILE
 	clear(m_profiling);
@@ -156,7 +164,7 @@ btDbvtBroadphase::btDbvtBroadphase(btOverlappingPairCache* paircache)
 //
 btDbvtBroadphase::~btDbvtBroadphase()
 {
-	if(m_releasepaircache) 
+	if (m_releasepaircache)
 	{
 		m_paircache->~btOverlappingPairCache();
 		btAlignedFree(m_paircache);
@@ -164,302 +172,294 @@ btDbvtBroadphase::~btDbvtBroadphase()
 }
 
 //
-btBroadphaseProxy*				btDbvtBroadphase::createProxy(	const btVector3& aabbMin,
-															  const btVector3& aabbMax,
-															  int /*shapeType*/,
-															  void* userPtr,
-															  int collisionFilterGroup,
-															  int collisionFilterMask,
-															  btDispatcher* /*dispatcher*/)
+btBroadphaseProxy* btDbvtBroadphase::createProxy(const btVector3& aabbMin,
+												 const btVector3& aabbMax,
+												 int /*shapeType*/,
+												 void* userPtr,
+												 int collisionFilterGroup,
+												 int collisionFilterMask,
+												 btDispatcher* /*dispatcher*/)
 {
-	btDbvtProxy*		proxy=new(btAlignedAlloc(sizeof(btDbvtProxy),16)) btDbvtProxy(	aabbMin,aabbMax,userPtr,
-		collisionFilterGroup,
-		collisionFilterMask);
+	btDbvtProxy* proxy = new (btAlignedAlloc(sizeof(btDbvtProxy), 16)) btDbvtProxy(aabbMin, aabbMax, userPtr,
+																				   collisionFilterGroup,
+																				   collisionFilterMask);
 
-	btDbvtAabbMm aabb = btDbvtVolume::FromMM(aabbMin,aabbMax);
+	btDbvtAabbMm aabb = btDbvtVolume::FromMM(aabbMin, aabbMax);
 
 	//bproxy->aabb			=	btDbvtVolume::FromMM(aabbMin,aabbMax);
-	proxy->stage		=	m_stageCurrent;
-	proxy->m_uniqueId	=	++m_gid;
-	proxy->leaf			=	m_sets[0].insert(aabb,proxy);
-	listappend(proxy,m_stageRoots[m_stageCurrent]);
-	if(!m_deferedcollide)
+	proxy->stage = m_stageCurrent;
+	proxy->m_uniqueId = ++m_gid;
+	proxy->leaf = m_sets[0].insert(aabb, proxy);
+	listappend(proxy, m_stageRoots[m_stageCurrent]);
+	if (!m_deferedcollide)
 	{
-		btDbvtTreeCollider	collider(this);
-		collider.proxy=proxy;
-		m_sets[0].collideTV(m_sets[0].m_root,aabb,collider);
-		m_sets[1].collideTV(m_sets[1].m_root,aabb,collider);
+		btDbvtTreeCollider collider(this);
+		collider.proxy = proxy;
+		m_sets[0].collideTV(m_sets[0].m_root, aabb, collider);
+		m_sets[1].collideTV(m_sets[1].m_root, aabb, collider);
 	}
-	return(proxy);
+	return (proxy);
 }
 
 //
-void							btDbvtBroadphase::destroyProxy(	btBroadphaseProxy* absproxy,
-															   btDispatcher* dispatcher)
+void btDbvtBroadphase::destroyProxy(btBroadphaseProxy* absproxy,
+									btDispatcher* dispatcher)
 {
-	btDbvtProxy*	proxy=(btDbvtProxy*)absproxy;
-	if(proxy->stage==STAGECOUNT)
+	btDbvtProxy* proxy = (btDbvtProxy*)absproxy;
+	if (proxy->stage == STAGECOUNT)
 		m_sets[1].remove(proxy->leaf);
 	else
 		m_sets[0].remove(proxy->leaf);
-	listremove(proxy,m_stageRoots[proxy->stage]);
-	m_paircache->removeOverlappingPairsContainingProxy(proxy,dispatcher);
+	listremove(proxy, m_stageRoots[proxy->stage]);
+	m_paircache->removeOverlappingPairsContainingProxy(proxy, dispatcher);
 	btAlignedFree(proxy);
-	m_needcleanup=true;
+	m_needcleanup = true;
 }
 
-void	btDbvtBroadphase::getAabb(btBroadphaseProxy* absproxy,btVector3& aabbMin, btVector3& aabbMax ) const
+void btDbvtBroadphase::getAabb(btBroadphaseProxy* absproxy, btVector3& aabbMin, btVector3& aabbMax) const
 {
-	btDbvtProxy*						proxy=(btDbvtProxy*)absproxy;
+	btDbvtProxy* proxy = (btDbvtProxy*)absproxy;
 	aabbMin = proxy->m_aabbMin;
 	aabbMax = proxy->m_aabbMax;
 }
 
-struct	BroadphaseRayTester : btDbvt::ICollide
+struct BroadphaseRayTester : btDbvt::ICollide
 {
 	btBroadphaseRayCallback& m_rayCallback;
 	BroadphaseRayTester(btBroadphaseRayCallback& orgCallback)
-		:m_rayCallback(orgCallback)
+		: m_rayCallback(orgCallback)
 	{
 	}
-	void					Process(const btDbvtNode* leaf)
+	void Process(const btDbvtNode* leaf)
 	{
-		btDbvtProxy*	proxy=(btDbvtProxy*)leaf->data;
+		btDbvtProxy* proxy = (btDbvtProxy*)leaf->data;
 		m_rayCallback.process(proxy);
 	}
-};	
+};
 
-void	btDbvtBroadphase::rayTest(const btVector3& rayFrom,const btVector3& rayTo, btBroadphaseRayCallback& rayCallback,const btVector3& aabbMin,const btVector3& aabbMax)
+void btDbvtBroadphase::rayTest(const btVector3& rayFrom, const btVector3& rayTo, btBroadphaseRayCallback& rayCallback, const btVector3& aabbMin, const btVector3& aabbMax)
 {
 	BroadphaseRayTester callback(rayCallback);
-    btAlignedObjectArray<const btDbvtNode*>* stack = &m_rayTestStacks[0];
+	btAlignedObjectArray<const btDbvtNode*>* stack = &m_rayTestStacks[0];
 #if BT_THREADSAFE
-    // for this function to be threadsafe, each thread must have a separate copy
-    // of this stack.  This could be thread-local static to avoid dynamic allocations,
-    // instead of just a local.
-    int threadIndex = btGetCurrentThreadIndex();
-    btAlignedObjectArray<const btDbvtNode*> localStack;
-    if (threadIndex < m_rayTestStacks.size())
-    {
-        // use per-thread preallocated stack if possible to avoid dynamic allocations
-        stack = &m_rayTestStacks[threadIndex];
-    }
-    else
-    {
-        stack = &localStack;
-    }
+	// for this function to be threadsafe, each thread must have a separate copy
+	// of this stack.  This could be thread-local static to avoid dynamic allocations,
+	// instead of just a local.
+	int threadIndex = btGetCurrentThreadIndex();
+	btAlignedObjectArray<const btDbvtNode*> localStack;
+	//todo(erwincoumans, "why do we get tsan issue here?")
+	if (0)//threadIndex < m_rayTestStacks.size())
+	//if (threadIndex < m_rayTestStacks.size())
+	{
+		// use per-thread preallocated stack if possible to avoid dynamic allocations
+		stack = &m_rayTestStacks[threadIndex];
+	}
+	else
+	{
+		stack = &localStack;
+	}
 #endif
 
-	m_sets[0].rayTestInternal(	m_sets[0].m_root,
-		rayFrom,
-		rayTo,
-		rayCallback.m_rayDirectionInverse,
-		rayCallback.m_signs,
-		rayCallback.m_lambda_max,
-		aabbMin,
-		aabbMax,
-        *stack,
-		callback);
-
-	m_sets[1].rayTestInternal(	m_sets[1].m_root,
-		rayFrom,
-		rayTo,
-		rayCallback.m_rayDirectionInverse,
-		rayCallback.m_signs,
-		rayCallback.m_lambda_max,
-		aabbMin,
-		aabbMax,
-        *stack,
-		callback);
+	m_sets[0].rayTestInternal(m_sets[0].m_root,
+							  rayFrom,
+							  rayTo,
+							  rayCallback.m_rayDirectionInverse,
+							  rayCallback.m_signs,
+							  rayCallback.m_lambda_max,
+							  aabbMin,
+							  aabbMax,
+							  *stack,
+							  callback);
 
+	m_sets[1].rayTestInternal(m_sets[1].m_root,
+							  rayFrom,
+							  rayTo,
+							  rayCallback.m_rayDirectionInverse,
+							  rayCallback.m_signs,
+							  rayCallback.m_lambda_max,
+							  aabbMin,
+							  aabbMax,
+							  *stack,
+							  callback);
 }
 
-
-struct	BroadphaseAabbTester : btDbvt::ICollide
+struct BroadphaseAabbTester : btDbvt::ICollide
 {
 	btBroadphaseAabbCallback& m_aabbCallback;
 	BroadphaseAabbTester(btBroadphaseAabbCallback& orgCallback)
-		:m_aabbCallback(orgCallback)
+		: m_aabbCallback(orgCallback)
 	{
 	}
-	void					Process(const btDbvtNode* leaf)
+	void Process(const btDbvtNode* leaf)
 	{
-		btDbvtProxy*	proxy=(btDbvtProxy*)leaf->data;
+		btDbvtProxy* proxy = (btDbvtProxy*)leaf->data;
 		m_aabbCallback.process(proxy);
 	}
-};	
+};
 
-void	btDbvtBroadphase::aabbTest(const btVector3& aabbMin,const btVector3& aabbMax,btBroadphaseAabbCallback& aabbCallback)
+void btDbvtBroadphase::aabbTest(const btVector3& aabbMin, const btVector3& aabbMax, btBroadphaseAabbCallback& aabbCallback)
 {
 	BroadphaseAabbTester callback(aabbCallback);
 
-	const ATTRIBUTE_ALIGNED16(btDbvtVolume)	bounds=btDbvtVolume::FromMM(aabbMin,aabbMax);
-		//process all children, that overlap with  the given AABB bounds
-	m_sets[0].collideTV(m_sets[0].m_root,bounds,callback);
-	m_sets[1].collideTV(m_sets[1].m_root,bounds,callback);
-
+	const ATTRIBUTE_ALIGNED16(btDbvtVolume) bounds = btDbvtVolume::FromMM(aabbMin, aabbMax);
+	//process all children, that overlap with  the given AABB bounds
+	m_sets[0].collideTV(m_sets[0].m_root, bounds, callback);
+	m_sets[1].collideTV(m_sets[1].m_root, bounds, callback);
 }
 
-
-
 //
-void							btDbvtBroadphase::setAabb(		btBroadphaseProxy* absproxy,
-														  const btVector3& aabbMin,
-														  const btVector3& aabbMax,
-														  btDispatcher* /*dispatcher*/)
+void btDbvtBroadphase::setAabb(btBroadphaseProxy* absproxy,
+							   const btVector3& aabbMin,
+							   const btVector3& aabbMax,
+							   btDispatcher* /*dispatcher*/)
 {
-	btDbvtProxy*						proxy=(btDbvtProxy*)absproxy;
-	ATTRIBUTE_ALIGNED16(btDbvtVolume)	aabb=btDbvtVolume::FromMM(aabbMin,aabbMax);
+	btDbvtProxy* proxy = (btDbvtProxy*)absproxy;
+	ATTRIBUTE_ALIGNED16(btDbvtVolume)
+	aabb = btDbvtVolume::FromMM(aabbMin, aabbMax);
 #if DBVT_BP_PREVENTFALSEUPDATE
-	if(NotEqual(aabb,proxy->leaf->volume))
+	if (NotEqual(aabb, proxy->leaf->volume))
 #endif
 	{
-		bool	docollide=false;
-		if(proxy->stage==STAGECOUNT)
-		{/* fixed -> dynamic set	*/ 
+		bool docollide = false;
+		if (proxy->stage == STAGECOUNT)
+		{ /* fixed -> dynamic set	*/
 			m_sets[1].remove(proxy->leaf);
-			proxy->leaf=m_sets[0].insert(aabb,proxy);
-			docollide=true;
+			proxy->leaf = m_sets[0].insert(aabb, proxy);
+			docollide = true;
 		}
 		else
-		{/* dynamic set				*/ 
+		{ /* dynamic set				*/
 			++m_updates_call;
-			if(Intersect(proxy->leaf->volume,aabb))
-			{/* Moving				*/ 
+			if (Intersect(proxy->leaf->volume, aabb))
+			{ /* Moving				*/
 
-				const btVector3	delta=aabbMin-proxy->m_aabbMin;
-				btVector3		velocity(((proxy->m_aabbMax-proxy->m_aabbMin)/2)*m_prediction);
-				if(delta[0]<0) velocity[0]=-velocity[0];
-				if(delta[1]<0) velocity[1]=-velocity[1];
-				if(delta[2]<0) velocity[2]=-velocity[2];
+				const btVector3 delta = aabbMin - proxy->m_aabbMin;
+				btVector3 velocity(((proxy->m_aabbMax - proxy->m_aabbMin) / 2) * m_prediction);
+				if (delta[0] < 0) velocity[0] = -velocity[0];
+				if (delta[1] < 0) velocity[1] = -velocity[1];
+				if (delta[2] < 0) velocity[2] = -velocity[2];
 				if (
 					m_sets[0].update(proxy->leaf, aabb, velocity, gDbvtMargin)
 
-					)
+				)
 				{
 					++m_updates_done;
-					docollide=true;
+					docollide = true;
 				}
 			}
 			else
-			{/* Teleporting			*/ 
-				m_sets[0].update(proxy->leaf,aabb);
+			{ /* Teleporting			*/
+				m_sets[0].update(proxy->leaf, aabb);
 				++m_updates_done;
-				docollide=true;
-			}	
+				docollide = true;
+			}
 		}
-		listremove(proxy,m_stageRoots[proxy->stage]);
+		listremove(proxy, m_stageRoots[proxy->stage]);
 		proxy->m_aabbMin = aabbMin;
 		proxy->m_aabbMax = aabbMax;
-		proxy->stage	=	m_stageCurrent;
-		listappend(proxy,m_stageRoots[m_stageCurrent]);
-		if(docollide)
+		proxy->stage = m_stageCurrent;
+		listappend(proxy, m_stageRoots[m_stageCurrent]);
+		if (docollide)
 		{
-			m_needcleanup=true;
-			if(!m_deferedcollide)
+			m_needcleanup = true;
+			if (!m_deferedcollide)
 			{
-				btDbvtTreeCollider	collider(this);
-				m_sets[1].collideTTpersistentStack(m_sets[1].m_root,proxy->leaf,collider);
-				m_sets[0].collideTTpersistentStack(m_sets[0].m_root,proxy->leaf,collider);
+				btDbvtTreeCollider collider(this);
+				m_sets[1].collideTTpersistentStack(m_sets[1].m_root, proxy->leaf, collider);
+				m_sets[0].collideTTpersistentStack(m_sets[0].m_root, proxy->leaf, collider);
 			}
-		}	
+		}
 	}
 }
 
-
 //
-void							btDbvtBroadphase::setAabbForceUpdate(		btBroadphaseProxy* absproxy,
-														  const btVector3& aabbMin,
-														  const btVector3& aabbMax,
-														  btDispatcher* /*dispatcher*/)
+void btDbvtBroadphase::setAabbForceUpdate(btBroadphaseProxy* absproxy,
+										  const btVector3& aabbMin,
+										  const btVector3& aabbMax,
+										  btDispatcher* /*dispatcher*/)
 {
-	btDbvtProxy*						proxy=(btDbvtProxy*)absproxy;
-	ATTRIBUTE_ALIGNED16(btDbvtVolume)	aabb=btDbvtVolume::FromMM(aabbMin,aabbMax);
-	bool	docollide=false;
-	if(proxy->stage==STAGECOUNT)
-	{/* fixed -> dynamic set	*/ 
+	btDbvtProxy* proxy = (btDbvtProxy*)absproxy;
+	ATTRIBUTE_ALIGNED16(btDbvtVolume)
+	aabb = btDbvtVolume::FromMM(aabbMin, aabbMax);
+	bool docollide = false;
+	if (proxy->stage == STAGECOUNT)
+	{ /* fixed -> dynamic set	*/
 		m_sets[1].remove(proxy->leaf);
-		proxy->leaf=m_sets[0].insert(aabb,proxy);
-		docollide=true;
+		proxy->leaf = m_sets[0].insert(aabb, proxy);
+		docollide = true;
 	}
 	else
-	{/* dynamic set				*/ 
+	{ /* dynamic set				*/
 		++m_updates_call;
-		/* Teleporting			*/ 
-		m_sets[0].update(proxy->leaf,aabb);
+		/* Teleporting			*/
+		m_sets[0].update(proxy->leaf, aabb);
 		++m_updates_done;
-		docollide=true;
+		docollide = true;
 	}
-	listremove(proxy,m_stageRoots[proxy->stage]);
+	listremove(proxy, m_stageRoots[proxy->stage]);
 	proxy->m_aabbMin = aabbMin;
 	proxy->m_aabbMax = aabbMax;
-	proxy->stage	=	m_stageCurrent;
-	listappend(proxy,m_stageRoots[m_stageCurrent]);
-	if(docollide)
+	proxy->stage = m_stageCurrent;
+	listappend(proxy, m_stageRoots[m_stageCurrent]);
+	if (docollide)
 	{
-		m_needcleanup=true;
-		if(!m_deferedcollide)
+		m_needcleanup = true;
+		if (!m_deferedcollide)
 		{
-			btDbvtTreeCollider	collider(this);
-			m_sets[1].collideTTpersistentStack(m_sets[1].m_root,proxy->leaf,collider);
-			m_sets[0].collideTTpersistentStack(m_sets[0].m_root,proxy->leaf,collider);
+			btDbvtTreeCollider collider(this);
+			m_sets[1].collideTTpersistentStack(m_sets[1].m_root, proxy->leaf, collider);
+			m_sets[0].collideTTpersistentStack(m_sets[0].m_root, proxy->leaf, collider);
 		}
-	}	
+	}
 }
 
 //
-void							btDbvtBroadphase::calculateOverlappingPairs(btDispatcher* dispatcher)
+void btDbvtBroadphase::calculateOverlappingPairs(btDispatcher* dispatcher)
 {
 	collide(dispatcher);
 #if DBVT_BP_PROFILE
-	if(0==(m_pid%DBVT_BP_PROFILING_RATE))
-	{	
-		printf("fixed(%u) dynamics(%u) pairs(%u)\r\n",m_sets[1].m_leaves,m_sets[0].m_leaves,m_paircache->getNumOverlappingPairs());
-		unsigned int	total=m_profiling.m_total;
-		if(total<=0) total=1;
-		printf("ddcollide: %u%% (%uus)\r\n",(50+m_profiling.m_ddcollide*100)/total,m_profiling.m_ddcollide/DBVT_BP_PROFILING_RATE);
-		printf("fdcollide: %u%% (%uus)\r\n",(50+m_profiling.m_fdcollide*100)/total,m_profiling.m_fdcollide/DBVT_BP_PROFILING_RATE);
-		printf("cleanup:   %u%% (%uus)\r\n",(50+m_profiling.m_cleanup*100)/total,m_profiling.m_cleanup/DBVT_BP_PROFILING_RATE);
-		printf("total:     %uus\r\n",total/DBVT_BP_PROFILING_RATE);
-		const unsigned long	sum=m_profiling.m_ddcollide+
-			m_profiling.m_fdcollide+
-			m_profiling.m_cleanup;
-		printf("leaked: %u%% (%uus)\r\n",100-((50+sum*100)/total),(total-sum)/DBVT_BP_PROFILING_RATE);
-		printf("job counts: %u%%\r\n",(m_profiling.m_jobcount*100)/((m_sets[0].m_leaves+m_sets[1].m_leaves)*DBVT_BP_PROFILING_RATE));
+	if (0 == (m_pid % DBVT_BP_PROFILING_RATE))
+	{
+		printf("fixed(%u) dynamics(%u) pairs(%u)\r\n", m_sets[1].m_leaves, m_sets[0].m_leaves, m_paircache->getNumOverlappingPairs());
+		unsigned int total = m_profiling.m_total;
+		if (total <= 0) total = 1;
+		printf("ddcollide: %u%% (%uus)\r\n", (50 + m_profiling.m_ddcollide * 100) / total, m_profiling.m_ddcollide / DBVT_BP_PROFILING_RATE);
+		printf("fdcollide: %u%% (%uus)\r\n", (50 + m_profiling.m_fdcollide * 100) / total, m_profiling.m_fdcollide / DBVT_BP_PROFILING_RATE);
+		printf("cleanup:   %u%% (%uus)\r\n", (50 + m_profiling.m_cleanup * 100) / total, m_profiling.m_cleanup / DBVT_BP_PROFILING_RATE);
+		printf("total:     %uus\r\n", total / DBVT_BP_PROFILING_RATE);
+		const unsigned long sum = m_profiling.m_ddcollide +
+								  m_profiling.m_fdcollide +
+								  m_profiling.m_cleanup;
+		printf("leaked: %u%% (%uus)\r\n", 100 - ((50 + sum * 100) / total), (total - sum) / DBVT_BP_PROFILING_RATE);
+		printf("job counts: %u%%\r\n", (m_profiling.m_jobcount * 100) / ((m_sets[0].m_leaves + m_sets[1].m_leaves) * DBVT_BP_PROFILING_RATE));
 		clear(m_profiling);
 		m_clock.reset();
 	}
 #endif
 
 	performDeferredRemoval(dispatcher);
-
 }
 
 void btDbvtBroadphase::performDeferredRemoval(btDispatcher* dispatcher)
 {
-
 	if (m_paircache->hasDeferredRemoval())
 	{
-
-		btBroadphasePairArray&	overlappingPairArray = m_paircache->getOverlappingPairArray();
+		btBroadphasePairArray& overlappingPairArray = m_paircache->getOverlappingPairArray();
 
 		//perform a sort, to find duplicates and to sort 'invalid' pairs to the end
 		overlappingPairArray.quickSort(btBroadphasePairSortPredicate());
 
 		int invalidPair = 0;
 
-		
 		int i;
 
 		btBroadphasePair previousPair;
 		previousPair.m_pProxy0 = 0;
 		previousPair.m_pProxy1 = 0;
 		previousPair.m_algorithm = 0;
-		
-		
-		for (i=0;i<overlappingPairArray.size();i++)
+
+		for (i = 0; i < overlappingPairArray.size(); i++)
 		{
-		
 			btBroadphasePair& pair = overlappingPairArray[i];
 
 			bool isDuplicate = (pair == previousPair);
@@ -471,34 +471,35 @@ void btDbvtBroadphase::performDeferredRemoval(btDispatcher* dispatcher)
 			if (!isDuplicate)
 			{
 				//important to perform AABB check that is consistent with the broadphase
-				btDbvtProxy*		pa=(btDbvtProxy*)pair.m_pProxy0;
-				btDbvtProxy*		pb=(btDbvtProxy*)pair.m_pProxy1;
-				bool hasOverlap = Intersect(pa->leaf->volume,pb->leaf->volume);
+				btDbvtProxy* pa = (btDbvtProxy*)pair.m_pProxy0;
+				btDbvtProxy* pb = (btDbvtProxy*)pair.m_pProxy1;
+				bool hasOverlap = Intersect(pa->leaf->volume, pb->leaf->volume);
 
 				if (hasOverlap)
 				{
 					needsRemoval = false;
-				} else
+				}
+				else
 				{
 					needsRemoval = true;
 				}
-			} else
+			}
+			else
 			{
 				//remove duplicate
 				needsRemoval = true;
 				//should have no algorithm
 				btAssert(!pair.m_algorithm);
 			}
-			
+
 			if (needsRemoval)
 			{
-				m_paircache->cleanOverlappingPair(pair,dispatcher);
+				m_paircache->cleanOverlappingPair(pair, dispatcher);
 
 				pair.m_pProxy0 = 0;
 				pair.m_pProxy1 = 0;
 				invalidPair++;
-			} 
-			
+			}
 		}
 
 		//perform a sort, to sort 'invalid' pairs to the end
@@ -508,7 +509,7 @@ void btDbvtBroadphase::performDeferredRemoval(btDispatcher* dispatcher)
 }
 
 //
-void							btDbvtBroadphase::collide(btDispatcher* dispatcher)
+void btDbvtBroadphase::collide(btDispatcher* dispatcher)
 {
 	/*printf("---------------------------------------------------------\n");
 	printf("m_sets[0].m_leaves=%d\n",m_sets[0].m_leaves);
@@ -525,295 +526,303 @@ void							btDbvtBroadphase::collide(btDispatcher* dispatcher)
 	}
 */
 
-
-
 	SPC(m_profiling.m_total);
-	/* optimize				*/ 
-	m_sets[0].optimizeIncremental(1+(m_sets[0].m_leaves*m_dupdates)/100);
-	if(m_fixedleft)
+	/* optimize				*/
+	m_sets[0].optimizeIncremental(1 + (m_sets[0].m_leaves * m_dupdates) / 100);
+	if (m_fixedleft)
 	{
-		const int count=1+(m_sets[1].m_leaves*m_fupdates)/100;
-		m_sets[1].optimizeIncremental(1+(m_sets[1].m_leaves*m_fupdates)/100);
-		m_fixedleft=btMax<int>(0,m_fixedleft-count);
+		const int count = 1 + (m_sets[1].m_leaves * m_fupdates) / 100;
+		m_sets[1].optimizeIncremental(1 + (m_sets[1].m_leaves * m_fupdates) / 100);
+		m_fixedleft = btMax<int>(0, m_fixedleft - count);
 	}
-	/* dynamic -> fixed set	*/ 
-	m_stageCurrent=(m_stageCurrent+1)%STAGECOUNT;
-	btDbvtProxy*	current=m_stageRoots[m_stageCurrent];
-	if(current)
+	/* dynamic -> fixed set	*/
+	m_stageCurrent = (m_stageCurrent + 1) % STAGECOUNT;
+	btDbvtProxy* current = m_stageRoots[m_stageCurrent];
+	if (current)
 	{
 #if DBVT_BP_ACCURATESLEEPING
-		btDbvtTreeCollider	collider(this);
+		btDbvtTreeCollider collider(this);
 #endif
-		do	{
-			btDbvtProxy*	next=current->links[1];
-			listremove(current,m_stageRoots[current->stage]);
-			listappend(current,m_stageRoots[STAGECOUNT]);
+		do
+		{
+			btDbvtProxy* next = current->links[1];
+			listremove(current, m_stageRoots[current->stage]);
+			listappend(current, m_stageRoots[STAGECOUNT]);
 #if DBVT_BP_ACCURATESLEEPING
-			m_paircache->removeOverlappingPairsContainingProxy(current,dispatcher);
-			collider.proxy=current;
-			btDbvt::collideTV(m_sets[0].m_root,current->aabb,collider);
-			btDbvt::collideTV(m_sets[1].m_root,current->aabb,collider);
+			m_paircache->removeOverlappingPairsContainingProxy(current, dispatcher);
+			collider.proxy = current;
+			btDbvt::collideTV(m_sets[0].m_root, current->aabb, collider);
+			btDbvt::collideTV(m_sets[1].m_root, current->aabb, collider);
 #endif
 			m_sets[0].remove(current->leaf);
-			ATTRIBUTE_ALIGNED16(btDbvtVolume)	curAabb=btDbvtVolume::FromMM(current->m_aabbMin,current->m_aabbMax);
-			current->leaf	=	m_sets[1].insert(curAabb,current);
-			current->stage	=	STAGECOUNT;	
-			current			=	next;
-		} while(current);
-		m_fixedleft=m_sets[1].m_leaves;
-		m_needcleanup=true;
+			ATTRIBUTE_ALIGNED16(btDbvtVolume)
+			curAabb = btDbvtVolume::FromMM(current->m_aabbMin, current->m_aabbMax);
+			current->leaf = m_sets[1].insert(curAabb, current);
+			current->stage = STAGECOUNT;
+			current = next;
+		} while (current);
+		m_fixedleft = m_sets[1].m_leaves;
+		m_needcleanup = true;
 	}
-	/* collide dynamics		*/ 
+	/* collide dynamics		*/
 	{
-		btDbvtTreeCollider	collider(this);
-		if(m_deferedcollide)
+		btDbvtTreeCollider collider(this);
+		if (m_deferedcollide)
 		{
 			SPC(m_profiling.m_fdcollide);
-			m_sets[0].collideTTpersistentStack(m_sets[0].m_root,m_sets[1].m_root,collider);
+			m_sets[0].collideTTpersistentStack(m_sets[0].m_root, m_sets[1].m_root, collider);
 		}
-		if(m_deferedcollide)
+		if (m_deferedcollide)
 		{
 			SPC(m_profiling.m_ddcollide);
-			m_sets[0].collideTTpersistentStack(m_sets[0].m_root,m_sets[0].m_root,collider);
+			m_sets[0].collideTTpersistentStack(m_sets[0].m_root, m_sets[0].m_root, collider);
 		}
 	}
-	/* clean up				*/ 
-	if(m_needcleanup)
+	/* clean up				*/
+	if (m_needcleanup)
 	{
 		SPC(m_profiling.m_cleanup);
-		btBroadphasePairArray&	pairs=m_paircache->getOverlappingPairArray();
-		if(pairs.size()>0)
+		btBroadphasePairArray& pairs = m_paircache->getOverlappingPairArray();
+		if (pairs.size() > 0)
 		{
-
-			int			ni=btMin(pairs.size(),btMax<int>(m_newpairs,(pairs.size()*m_cupdates)/100));
-			for(int i=0;i<ni;++i)
+			int ni = btMin(pairs.size(), btMax<int>(m_newpairs, (pairs.size() * m_cupdates) / 100));
+			for (int i = 0; i < ni; ++i)
 			{
-				btBroadphasePair&	p=pairs[(m_cid+i)%pairs.size()];
-				btDbvtProxy*		pa=(btDbvtProxy*)p.m_pProxy0;
-				btDbvtProxy*		pb=(btDbvtProxy*)p.m_pProxy1;
-				if(!Intersect(pa->leaf->volume,pb->leaf->volume))
+				btBroadphasePair& p = pairs[(m_cid + i) % pairs.size()];
+				btDbvtProxy* pa = (btDbvtProxy*)p.m_pProxy0;
+				btDbvtProxy* pb = (btDbvtProxy*)p.m_pProxy1;
+				if (!Intersect(pa->leaf->volume, pb->leaf->volume))
 				{
 #if DBVT_BP_SORTPAIRS
-					if(pa->m_uniqueId>pb->m_uniqueId) 
-						btSwap(pa,pb);
+					if (pa->m_uniqueId > pb->m_uniqueId)
+						btSwap(pa, pb);
 #endif
-					m_paircache->removeOverlappingPair(pa,pb,dispatcher);
-					--ni;--i;
+					m_paircache->removeOverlappingPair(pa, pb, dispatcher);
+					--ni;
+					--i;
 				}
 			}
-			if(pairs.size()>0) m_cid=(m_cid+ni)%pairs.size(); else m_cid=0;
+			if (pairs.size() > 0)
+				m_cid = (m_cid + ni) % pairs.size();
+			else
+				m_cid = 0;
 		}
 	}
 	++m_pid;
-	m_newpairs=1;
-	m_needcleanup=false;
-	if(m_updates_call>0)
-	{ m_updates_ratio=m_updates_done/(btScalar)m_updates_call; }
+	m_newpairs = 1;
+	m_needcleanup = false;
+	if (m_updates_call > 0)
+	{
+		m_updates_ratio = m_updates_done / (btScalar)m_updates_call;
+	}
 	else
-	{ m_updates_ratio=0; }
-	m_updates_done/=2;
-	m_updates_call/=2;
+	{
+		m_updates_ratio = 0;
+	}
+	m_updates_done /= 2;
+	m_updates_call /= 2;
 }
 
 //
-void							btDbvtBroadphase::optimize()
+void btDbvtBroadphase::optimize()
 {
 	m_sets[0].optimizeTopDown();
 	m_sets[1].optimizeTopDown();
 }
 
 //
-btOverlappingPairCache*			btDbvtBroadphase::getOverlappingPairCache()
+btOverlappingPairCache* btDbvtBroadphase::getOverlappingPairCache()
 {
-	return(m_paircache);
+	return (m_paircache);
 }
 
 //
-const btOverlappingPairCache*	btDbvtBroadphase::getOverlappingPairCache() const
+const btOverlappingPairCache* btDbvtBroadphase::getOverlappingPairCache() const
 {
-	return(m_paircache);
+	return (m_paircache);
 }
 
 //
-void							btDbvtBroadphase::getBroadphaseAabb(btVector3& aabbMin,btVector3& aabbMax) const
+void btDbvtBroadphase::getBroadphaseAabb(btVector3& aabbMin, btVector3& aabbMax) const
 {
+	ATTRIBUTE_ALIGNED16(btDbvtVolume)
+	bounds;
 
-	ATTRIBUTE_ALIGNED16(btDbvtVolume)	bounds;
-
-	if(!m_sets[0].empty())
-		if(!m_sets[1].empty())	Merge(	m_sets[0].m_root->volume,
-			m_sets[1].m_root->volume,bounds);
+	if (!m_sets[0].empty())
+		if (!m_sets[1].empty())
+			Merge(m_sets[0].m_root->volume,
+				  m_sets[1].m_root->volume, bounds);
 		else
-			bounds=m_sets[0].m_root->volume;
-	else if(!m_sets[1].empty())	bounds=m_sets[1].m_root->volume;
+			bounds = m_sets[0].m_root->volume;
+	else if (!m_sets[1].empty())
+		bounds = m_sets[1].m_root->volume;
 	else
-		bounds=btDbvtVolume::FromCR(btVector3(0,0,0),0);
-	aabbMin=bounds.Mins();
-	aabbMax=bounds.Maxs();
+		bounds = btDbvtVolume::FromCR(btVector3(0, 0, 0), 0);
+	aabbMin = bounds.Mins();
+	aabbMax = bounds.Maxs();
 }
 
 void btDbvtBroadphase::resetPool(btDispatcher* dispatcher)
 {
-	
 	int totalObjects = m_sets[0].m_leaves + m_sets[1].m_leaves;
 	if (!totalObjects)
 	{
 		//reset internal dynamic tree data structures
 		m_sets[0].clear();
 		m_sets[1].clear();
-		
-		m_deferedcollide	=	false;
-		m_needcleanup		=	true;
-		m_stageCurrent		=	0;
-		m_fixedleft			=	0;
-		m_fupdates			=	1;
-		m_dupdates			=	0;
-		m_cupdates			=	10;
-		m_newpairs			=	1;
-		m_updates_call		=	0;
-		m_updates_done		=	0;
-		m_updates_ratio		=	0;
-		
-		m_gid				=	0;
-		m_pid				=	0;
-		m_cid				=	0;
-		for(int i=0;i<=STAGECOUNT;++i)
+
+		m_deferedcollide = false;
+		m_needcleanup = true;
+		m_stageCurrent = 0;
+		m_fixedleft = 0;
+		m_fupdates = 1;
+		m_dupdates = 0;
+		m_cupdates = 10;
+		m_newpairs = 1;
+		m_updates_call = 0;
+		m_updates_done = 0;
+		m_updates_ratio = 0;
+
+		m_gid = 0;
+		m_pid = 0;
+		m_cid = 0;
+		for (int i = 0; i <= STAGECOUNT; ++i)
 		{
-			m_stageRoots[i]=0;
+			m_stageRoots[i] = 0;
 		}
 	}
 }
 
 //
-void							btDbvtBroadphase::printStats()
-{}
+void btDbvtBroadphase::printStats()
+{
+}
 
 //
 #if DBVT_BP_ENABLE_BENCHMARK
 
-struct	btBroadphaseBenchmark
+struct btBroadphaseBenchmark
 {
-	struct	Experiment
+	struct Experiment
 	{
-		const char*			name;
-		int					object_count;
-		int					update_count;
-		int					spawn_count;
-		int					iterations;
-		btScalar			speed;
-		btScalar			amplitude;
+		const char* name;
+		int object_count;
+		int update_count;
+		int spawn_count;
+		int iterations;
+		btScalar speed;
+		btScalar amplitude;
 	};
-	struct	Object
+	struct Object
 	{
-		btVector3			center;
-		btVector3			extents;
-		btBroadphaseProxy*	proxy;
-		btScalar			time;
-		void				update(btScalar speed,btScalar amplitude,btBroadphaseInterface* pbi)
+		btVector3 center;
+		btVector3 extents;
+		btBroadphaseProxy* proxy;
+		btScalar time;
+		void update(btScalar speed, btScalar amplitude, btBroadphaseInterface* pbi)
 		{
-			time		+=	speed;
-			center[0]	=	btCos(time*(btScalar)2.17)*amplitude+
-				btSin(time)*amplitude/2;
-			center[1]	=	btCos(time*(btScalar)1.38)*amplitude+
-				btSin(time)*amplitude;
-			center[2]	=	btSin(time*(btScalar)0.777)*amplitude;
-			pbi->setAabb(proxy,center-extents,center+extents,0);
+			time += speed;
+			center[0] = btCos(time * (btScalar)2.17) * amplitude +
+						btSin(time) * amplitude / 2;
+			center[1] = btCos(time * (btScalar)1.38) * amplitude +
+						btSin(time) * amplitude;
+			center[2] = btSin(time * (btScalar)0.777) * amplitude;
+			pbi->setAabb(proxy, center - extents, center + extents, 0);
 		}
 	};
-	static int		UnsignedRand(int range=RAND_MAX-1)	{ return(rand()%(range+1)); }
-	static btScalar	UnitRand()							{ return(UnsignedRand(16384)/(btScalar)16384); }
-	static void		OutputTime(const char* name,btClock& c,unsigned count=0)
+	static int UnsignedRand(int range = RAND_MAX - 1) { return (rand() % (range + 1)); }
+	static btScalar UnitRand() { return (UnsignedRand(16384) / (btScalar)16384); }
+	static void OutputTime(const char* name, btClock& c, unsigned count = 0)
 	{
-		const unsigned long	us=c.getTimeMicroseconds();
-		const unsigned long	ms=(us+500)/1000;
-		const btScalar		sec=us/(btScalar)(1000*1000);
-		if(count>0)
-			printf("%s : %u us (%u ms), %.2f/s\r\n",name,us,ms,count/sec);
+		const unsigned long us = c.getTimeMicroseconds();
+		const unsigned long ms = (us + 500) / 1000;
+		const btScalar sec = us / (btScalar)(1000 * 1000);
+		if (count > 0)
+			printf("%s : %u us (%u ms), %.2f/s\r\n", name, us, ms, count / sec);
 		else
-			printf("%s : %u us (%u ms)\r\n",name,us,ms);
+			printf("%s : %u us (%u ms)\r\n", name, us, ms);
 	}
 };
 
-void							btDbvtBroadphase::benchmark(btBroadphaseInterface* pbi)
+void btDbvtBroadphase::benchmark(btBroadphaseInterface* pbi)
 {
-	static const btBroadphaseBenchmark::Experiment		experiments[]=
-	{
-		{"1024o.10%",1024,10,0,8192,(btScalar)0.005,(btScalar)100},
-		/*{"4096o.10%",4096,10,0,8192,(btScalar)0.005,(btScalar)100},
+	static const btBroadphaseBenchmark::Experiment experiments[] =
+		{
+			{"1024o.10%", 1024, 10, 0, 8192, (btScalar)0.005, (btScalar)100},
+			/*{"4096o.10%",4096,10,0,8192,(btScalar)0.005,(btScalar)100},
 		{"8192o.10%",8192,10,0,8192,(btScalar)0.005,(btScalar)100},*/
-	};
-	static const int										nexperiments=sizeof(experiments)/sizeof(experiments[0]);
-	btAlignedObjectArray<btBroadphaseBenchmark::Object*>	objects;
-	btClock													wallclock;
-	/* Begin			*/ 
-	for(int iexp=0;iexp<nexperiments;++iexp)
+		};
+	static const int nexperiments = sizeof(experiments) / sizeof(experiments[0]);
+	btAlignedObjectArray<btBroadphaseBenchmark::Object*> objects;
+	btClock wallclock;
+	/* Begin			*/
+	for (int iexp = 0; iexp < nexperiments; ++iexp)
 	{
-		const btBroadphaseBenchmark::Experiment&	experiment=experiments[iexp];
-		const int									object_count=experiment.object_count;
-		const int									update_count=(object_count*experiment.update_count)/100;
-		const int									spawn_count=(object_count*experiment.spawn_count)/100;
-		const btScalar								speed=experiment.speed;	
-		const btScalar								amplitude=experiment.amplitude;
-		printf("Experiment #%u '%s':\r\n",iexp,experiment.name);
-		printf("\tObjects: %u\r\n",object_count);
-		printf("\tUpdate: %u\r\n",update_count);
-		printf("\tSpawn: %u\r\n",spawn_count);
-		printf("\tSpeed: %f\r\n",speed);
-		printf("\tAmplitude: %f\r\n",amplitude);
+		const btBroadphaseBenchmark::Experiment& experiment = experiments[iexp];
+		const int object_count = experiment.object_count;
+		const int update_count = (object_count * experiment.update_count) / 100;
+		const int spawn_count = (object_count * experiment.spawn_count) / 100;
+		const btScalar speed = experiment.speed;
+		const btScalar amplitude = experiment.amplitude;
+		printf("Experiment #%u '%s':\r\n", iexp, experiment.name);
+		printf("\tObjects: %u\r\n", object_count);
+		printf("\tUpdate: %u\r\n", update_count);
+		printf("\tSpawn: %u\r\n", spawn_count);
+		printf("\tSpeed: %f\r\n", speed);
+		printf("\tAmplitude: %f\r\n", amplitude);
 		srand(180673);
-		/* Create objects	*/ 
+		/* Create objects	*/
 		wallclock.reset();
 		objects.reserve(object_count);
-		for(int i=0;i<object_count;++i)
+		for (int i = 0; i < object_count; ++i)
 		{
-			btBroadphaseBenchmark::Object*	po=new btBroadphaseBenchmark::Object();
-			po->center[0]=btBroadphaseBenchmark::UnitRand()*50;
-			po->center[1]=btBroadphaseBenchmark::UnitRand()*50;
-			po->center[2]=btBroadphaseBenchmark::UnitRand()*50;
-			po->extents[0]=btBroadphaseBenchmark::UnitRand()*2+2;
-			po->extents[1]=btBroadphaseBenchmark::UnitRand()*2+2;
-			po->extents[2]=btBroadphaseBenchmark::UnitRand()*2+2;
-			po->time=btBroadphaseBenchmark::UnitRand()*2000;
-			po->proxy=pbi->createProxy(po->center-po->extents,po->center+po->extents,0,po,1,1,0,0);
+			btBroadphaseBenchmark::Object* po = new btBroadphaseBenchmark::Object();
+			po->center[0] = btBroadphaseBenchmark::UnitRand() * 50;
+			po->center[1] = btBroadphaseBenchmark::UnitRand() * 50;
+			po->center[2] = btBroadphaseBenchmark::UnitRand() * 50;
+			po->extents[0] = btBroadphaseBenchmark::UnitRand() * 2 + 2;
+			po->extents[1] = btBroadphaseBenchmark::UnitRand() * 2 + 2;
+			po->extents[2] = btBroadphaseBenchmark::UnitRand() * 2 + 2;
+			po->time = btBroadphaseBenchmark::UnitRand() * 2000;
+			po->proxy = pbi->createProxy(po->center - po->extents, po->center + po->extents, 0, po, 1, 1, 0, 0);
 			objects.push_back(po);
 		}
-		btBroadphaseBenchmark::OutputTime("\tInitialization",wallclock);
-		/* First update		*/ 
+		btBroadphaseBenchmark::OutputTime("\tInitialization", wallclock);
+		/* First update		*/
 		wallclock.reset();
-		for(int i=0;i<objects.size();++i)
+		for (int i = 0; i < objects.size(); ++i)
 		{
-			objects[i]->update(speed,amplitude,pbi);
+			objects[i]->update(speed, amplitude, pbi);
 		}
-		btBroadphaseBenchmark::OutputTime("\tFirst update",wallclock);
-		/* Updates			*/ 
+		btBroadphaseBenchmark::OutputTime("\tFirst update", wallclock);
+		/* Updates			*/
 		wallclock.reset();
-		for(int i=0;i<experiment.iterations;++i)
+		for (int i = 0; i < experiment.iterations; ++i)
 		{
-			for(int j=0;j<update_count;++j)
-			{				
-				objects[j]->update(speed,amplitude,pbi);
+			for (int j = 0; j < update_count; ++j)
+			{
+				objects[j]->update(speed, amplitude, pbi);
 			}
 			pbi->calculateOverlappingPairs(0);
 		}
-		btBroadphaseBenchmark::OutputTime("\tUpdate",wallclock,experiment.iterations);
-		/* Clean up			*/ 
+		btBroadphaseBenchmark::OutputTime("\tUpdate", wallclock, experiment.iterations);
+		/* Clean up			*/
 		wallclock.reset();
-		for(int i=0;i<objects.size();++i)
+		for (int i = 0; i < objects.size(); ++i)
 		{
-			pbi->destroyProxy(objects[i]->proxy,0);
+			pbi->destroyProxy(objects[i]->proxy, 0);
 			delete objects[i];
 		}
 		objects.resize(0);
-		btBroadphaseBenchmark::OutputTime("\tRelease",wallclock);
+		btBroadphaseBenchmark::OutputTime("\tRelease", wallclock);
 	}
-
 }
 #else
-void							btDbvtBroadphase::benchmark(btBroadphaseInterface*)
-{}
+void btDbvtBroadphase::benchmark(btBroadphaseInterface*)
+{
+}
 #endif
 
 #if DBVT_BP_PROFILE
-#undef	SPC
+#undef SPC
 #endif
-
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDbvtBroadphase.h b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDbvtBroadphase.h
index 90b333d846a..a71feef53bb 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDbvtBroadphase.h
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDbvtBroadphase.h
@@ -24,16 +24,16 @@ subject to the following restrictions:
 // Compile time config
 //
 
-#define	DBVT_BP_PROFILE					0
+#define DBVT_BP_PROFILE 0
 //#define DBVT_BP_SORTPAIRS				1
-#define DBVT_BP_PREVENTFALSEUPDATE		0
-#define DBVT_BP_ACCURATESLEEPING		0
-#define DBVT_BP_ENABLE_BENCHMARK		0
+#define DBVT_BP_PREVENTFALSEUPDATE 0
+#define DBVT_BP_ACCURATESLEEPING 0
+#define DBVT_BP_ENABLE_BENCHMARK 0
 //#define DBVT_BP_MARGIN					(btScalar)0.05
 extern btScalar gDbvtMargin;
 
 #if DBVT_BP_PROFILE
-#define	DBVT_BP_PROFILING_RATE	256
+#define DBVT_BP_PROFILING_RATE 256
 #include "LinearMath/btQuickprof.h"
 #endif
 
@@ -42,90 +42,90 @@ extern btScalar gDbvtMargin;
 //
 struct btDbvtProxy : btBroadphaseProxy
 {
-	/* Fields		*/ 
+	/* Fields		*/
 	//btDbvtAabbMm	aabb;
-	btDbvtNode*		leaf;
-	btDbvtProxy*	links[2];
-	int				stage;
-	/* ctor			*/ 
-	btDbvtProxy(const btVector3& aabbMin,const btVector3& aabbMax,void* userPtr, int collisionFilterGroup,  int collisionFilterMask) :
-	btBroadphaseProxy(aabbMin,aabbMax,userPtr,collisionFilterGroup,collisionFilterMask)
+	btDbvtNode* leaf;
+	btDbvtProxy* links[2];
+	int stage;
+	/* ctor			*/
+	btDbvtProxy(const btVector3& aabbMin, const btVector3& aabbMax, void* userPtr, int collisionFilterGroup, int collisionFilterMask) : btBroadphaseProxy(aabbMin, aabbMax, userPtr, collisionFilterGroup, collisionFilterMask)
 	{
-		links[0]=links[1]=0;
+		links[0] = links[1] = 0;
 	}
 };
 
-typedef btAlignedObjectArray<btDbvtProxy*>	btDbvtProxyArray;
+typedef btAlignedObjectArray<btDbvtProxy*> btDbvtProxyArray;
 
 ///The btDbvtBroadphase implements a broadphase using two dynamic AABB bounding volume hierarchies/trees (see btDbvt).
 ///One tree is used for static/non-moving objects, and another tree is used for dynamic objects. Objects can move from one tree to the other.
 ///This is a very fast broadphase, especially for very dynamic worlds where many objects are moving. Its insert/add and remove of objects is generally faster than the sweep and prune broadphases btAxisSweep3 and bt32BitAxisSweep3.
-struct	btDbvtBroadphase : btBroadphaseInterface
+struct btDbvtBroadphase : btBroadphaseInterface
 {
-	/* Config		*/ 
-	enum	{
-		DYNAMIC_SET			=	0,	/* Dynamic set index	*/ 
-		FIXED_SET			=	1,	/* Fixed set index		*/ 
-		STAGECOUNT			=	2	/* Number of stages		*/ 
+	/* Config		*/
+	enum
+	{
+		DYNAMIC_SET = 0, /* Dynamic set index	*/
+		FIXED_SET = 1,   /* Fixed set index		*/
+		STAGECOUNT = 2   /* Number of stages		*/
 	};
-	/* Fields		*/ 
-	btDbvt					m_sets[2];					// Dbvt sets
-	btDbvtProxy*			m_stageRoots[STAGECOUNT+1];	// Stages list
-	btOverlappingPairCache*	m_paircache;				// Pair cache
-	btScalar				m_prediction;				// Velocity prediction
-	int						m_stageCurrent;				// Current stage
-	int						m_fupdates;					// % of fixed updates per frame
-	int						m_dupdates;					// % of dynamic updates per frame
-	int						m_cupdates;					// % of cleanup updates per frame
-	int						m_newpairs;					// Number of pairs created
-	int						m_fixedleft;				// Fixed optimization left
-	unsigned				m_updates_call;				// Number of updates call
-	unsigned				m_updates_done;				// Number of updates done
-	btScalar				m_updates_ratio;			// m_updates_done/m_updates_call
-	int						m_pid;						// Parse id
-	int						m_cid;						// Cleanup index
-	int						m_gid;						// Gen id
-	bool					m_releasepaircache;			// Release pair cache on delete
-	bool					m_deferedcollide;			// Defere dynamic/static collision to collide call
-	bool					m_needcleanup;				// Need to run cleanup?
-    btAlignedObjectArray< btAlignedObjectArray<const btDbvtNode*> > m_rayTestStacks;
+	/* Fields		*/
+	btDbvt m_sets[2];                           // Dbvt sets
+	btDbvtProxy* m_stageRoots[STAGECOUNT + 1];  // Stages list
+	btOverlappingPairCache* m_paircache;        // Pair cache
+	btScalar m_prediction;                      // Velocity prediction
+	int m_stageCurrent;                         // Current stage
+	int m_fupdates;                             // % of fixed updates per frame
+	int m_dupdates;                             // % of dynamic updates per frame
+	int m_cupdates;                             // % of cleanup updates per frame
+	int m_newpairs;                             // Number of pairs created
+	int m_fixedleft;                            // Fixed optimization left
+	unsigned m_updates_call;                    // Number of updates call
+	unsigned m_updates_done;                    // Number of updates done
+	btScalar m_updates_ratio;                   // m_updates_done/m_updates_call
+	int m_pid;                                  // Parse id
+	int m_cid;                                  // Cleanup index
+	int m_gid;                                  // Gen id
+	bool m_releasepaircache;                    // Release pair cache on delete
+	bool m_deferedcollide;                      // Defere dynamic/static collision to collide call
+	bool m_needcleanup;                         // Need to run cleanup?
+	btAlignedObjectArray<btAlignedObjectArray<const btDbvtNode*> > m_rayTestStacks;
 #if DBVT_BP_PROFILE
-	btClock					m_clock;
-	struct	{
-		unsigned long		m_total;
-		unsigned long		m_ddcollide;
-		unsigned long		m_fdcollide;
-		unsigned long		m_cleanup;
-		unsigned long		m_jobcount;
-	}				m_profiling;
+	btClock m_clock;
+	struct
+	{
+		unsigned long m_total;
+		unsigned long m_ddcollide;
+		unsigned long m_fdcollide;
+		unsigned long m_cleanup;
+		unsigned long m_jobcount;
+	} m_profiling;
 #endif
-	/* Methods		*/ 
-	btDbvtBroadphase(btOverlappingPairCache* paircache=0);
+	/* Methods		*/
+	btDbvtBroadphase(btOverlappingPairCache* paircache = 0);
 	~btDbvtBroadphase();
-	void							collide(btDispatcher* dispatcher);
-	void							optimize();
-	
+	void collide(btDispatcher* dispatcher);
+	void optimize();
+
 	/* btBroadphaseInterface Implementation	*/
-	btBroadphaseProxy*				createProxy(const btVector3& aabbMin,const btVector3& aabbMax,int shapeType,void* userPtr, int collisionFilterGroup, int collisionFilterMask,btDispatcher* dispatcher);
-	virtual void					destroyProxy(btBroadphaseProxy* proxy,btDispatcher* dispatcher);
-	virtual void					setAabb(btBroadphaseProxy* proxy,const btVector3& aabbMin,const btVector3& aabbMax,btDispatcher* dispatcher);
-	virtual void					rayTest(const btVector3& rayFrom,const btVector3& rayTo, btBroadphaseRayCallback& rayCallback, const btVector3& aabbMin=btVector3(0,0,0), const btVector3& aabbMax = btVector3(0,0,0));
-	virtual void					aabbTest(const btVector3& aabbMin, const btVector3& aabbMax, btBroadphaseAabbCallback& callback);
-
-	virtual void					getAabb(btBroadphaseProxy* proxy,btVector3& aabbMin, btVector3& aabbMax ) const;
-	virtual	void					calculateOverlappingPairs(btDispatcher* dispatcher);
-	virtual	btOverlappingPairCache*	getOverlappingPairCache();
-	virtual	const btOverlappingPairCache*	getOverlappingPairCache() const;
-	virtual	void					getBroadphaseAabb(btVector3& aabbMin,btVector3& aabbMax) const;
-	virtual	void					printStats();
+	btBroadphaseProxy* createProxy(const btVector3& aabbMin, const btVector3& aabbMax, int shapeType, void* userPtr, int collisionFilterGroup, int collisionFilterMask, btDispatcher* dispatcher);
+	virtual void destroyProxy(btBroadphaseProxy* proxy, btDispatcher* dispatcher);
+	virtual void setAabb(btBroadphaseProxy* proxy, const btVector3& aabbMin, const btVector3& aabbMax, btDispatcher* dispatcher);
+	virtual void rayTest(const btVector3& rayFrom, const btVector3& rayTo, btBroadphaseRayCallback& rayCallback, const btVector3& aabbMin = btVector3(0, 0, 0), const btVector3& aabbMax = btVector3(0, 0, 0));
+	virtual void aabbTest(const btVector3& aabbMin, const btVector3& aabbMax, btBroadphaseAabbCallback& callback);
 
+	virtual void getAabb(btBroadphaseProxy* proxy, btVector3& aabbMin, btVector3& aabbMax) const;
+	virtual void calculateOverlappingPairs(btDispatcher* dispatcher);
+	virtual btOverlappingPairCache* getOverlappingPairCache();
+	virtual const btOverlappingPairCache* getOverlappingPairCache() const;
+	virtual void getBroadphaseAabb(btVector3& aabbMin, btVector3& aabbMax) const;
+	virtual void printStats();
 
 	///reset broadphase internal structures, to ensure determinism/reproducability
 	virtual void resetPool(btDispatcher* dispatcher);
 
-	void	performDeferredRemoval(btDispatcher* dispatcher);
-	
-	void	setVelocityPrediction(btScalar prediction)
+	void performDeferredRemoval(btDispatcher* dispatcher);
+
+	void setVelocityPrediction(btScalar prediction)
 	{
 		m_prediction = prediction;
 	}
@@ -134,15 +134,13 @@ struct	btDbvtBroadphase : btBroadphaseInterface
 		return m_prediction;
 	}
 
-	///this setAabbForceUpdate is similar to setAabb but always forces the aabb update. 
+	///this setAabbForceUpdate is similar to setAabb but always forces the aabb update.
 	///it is not part of the btBroadphaseInterface but specific to btDbvtBroadphase.
 	///it bypasses certain optimizations that prevent aabb updates (when the aabb shrinks), see
 	///http://code.google.com/p/bullet/issues/detail?id=223
-	void							setAabbForceUpdate(		btBroadphaseProxy* absproxy,const btVector3& aabbMin,const btVector3& aabbMax,btDispatcher* /*dispatcher*/);
-
-	static void						benchmark(btBroadphaseInterface*);
-
+	void setAabbForceUpdate(btBroadphaseProxy* absproxy, const btVector3& aabbMin, const btVector3& aabbMax, btDispatcher* /*dispatcher*/);
 
+	static void benchmark(btBroadphaseInterface*);
 };
 
 #endif
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDispatcher.cpp b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDispatcher.cpp
index 20768225b3a..d76d408aa65 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDispatcher.cpp
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDispatcher.cpp
@@ -17,6 +17,4 @@ subject to the following restrictions:
 
 btDispatcher::~btDispatcher()
 {
-
 }
-
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDispatcher.h b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDispatcher.h
index a0e4c189270..b09b7d4d422 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDispatcher.h
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btDispatcher.h
@@ -20,7 +20,7 @@ subject to the following restrictions:
 class btCollisionAlgorithm;
 struct btBroadphaseProxy;
 class btRigidBody;
-class	btCollisionObject;
+class btCollisionObject;
 class btOverlappingPairCache;
 struct btCollisionObjectWrapper;
 
@@ -35,35 +35,34 @@ struct btDispatcherInfo
 		DISPATCH_CONTINUOUS
 	};
 	btDispatcherInfo()
-		:m_timeStep(btScalar(0.)),
-		m_stepCount(0),
-		m_dispatchFunc(DISPATCH_DISCRETE),
-		m_timeOfImpact(btScalar(1.)),
-		m_useContinuous(true),
-		m_debugDraw(0),
-		m_enableSatConvex(false),
-		m_enableSPU(true),
-		m_useEpa(true),
-		m_allowedCcdPenetration(btScalar(0.04)),
-		m_useConvexConservativeDistanceUtil(false),
-		m_convexConservativeDistanceThreshold(0.0f),
-		m_deterministicOverlappingPairs(false)
+		: m_timeStep(btScalar(0.)),
+		  m_stepCount(0),
+		  m_dispatchFunc(DISPATCH_DISCRETE),
+		  m_timeOfImpact(btScalar(1.)),
+		  m_useContinuous(true),
+		  m_debugDraw(0),
+		  m_enableSatConvex(false),
+		  m_enableSPU(true),
+		  m_useEpa(true),
+		  m_allowedCcdPenetration(btScalar(0.04)),
+		  m_useConvexConservativeDistanceUtil(false),
+		  m_convexConservativeDistanceThreshold(0.0f),
+		  m_deterministicOverlappingPairs(false)
 	{
-
 	}
-	btScalar	m_timeStep;
-	int			m_stepCount;
-	int			m_dispatchFunc;
-	mutable btScalar	m_timeOfImpact;
-	bool		m_useContinuous;
-	class btIDebugDraw*	m_debugDraw;
-	bool		m_enableSatConvex;
-	bool		m_enableSPU;
-	bool		m_useEpa;
-	btScalar	m_allowedCcdPenetration;
-	bool		m_useConvexConservativeDistanceUtil;
-	btScalar	m_convexConservativeDistanceThreshold;
-	bool		m_deterministicOverlappingPairs;
+	btScalar m_timeStep;
+	int m_stepCount;
+	int m_dispatchFunc;
+	mutable btScalar m_timeOfImpact;
+	bool m_useContinuous;
+	class btIDebugDraw* m_debugDraw;
+	bool m_enableSatConvex;
+	bool m_enableSPU;
+	bool m_useEpa;
+	btScalar m_allowedCcdPenetration;
+	bool m_useConvexConservativeDistanceUtil;
+	btScalar m_convexConservativeDistanceThreshold;
+	bool m_deterministicOverlappingPairs;
 };
 
 enum ebtDispatcherQueryType
@@ -76,40 +75,36 @@ enum ebtDispatcherQueryType
 ///For example for pairwise collision detection, calculating contact points stored in btPersistentManifold or user callbacks (game logic).
 class btDispatcher
 {
-
-
 public:
-	virtual ~btDispatcher() ;
+	virtual ~btDispatcher();
 
-	virtual btCollisionAlgorithm* findAlgorithm(const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,btPersistentManifold* sharedManifold, ebtDispatcherQueryType queryType) = 0;
+	virtual btCollisionAlgorithm* findAlgorithm(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, btPersistentManifold* sharedManifold, ebtDispatcherQueryType queryType) = 0;
 
-	virtual btPersistentManifold*	getNewManifold(const btCollisionObject* b0,const btCollisionObject* b1)=0;
+	virtual btPersistentManifold* getNewManifold(const btCollisionObject* b0, const btCollisionObject* b1) = 0;
 
-	virtual void releaseManifold(btPersistentManifold* manifold)=0;
+	virtual void releaseManifold(btPersistentManifold* manifold) = 0;
 
-	virtual void clearManifold(btPersistentManifold* manifold)=0;
+	virtual void clearManifold(btPersistentManifold* manifold) = 0;
 
-	virtual bool	needsCollision(const btCollisionObject* body0,const btCollisionObject* body1) = 0;
+	virtual bool needsCollision(const btCollisionObject* body0, const btCollisionObject* body1) = 0;
 
-	virtual bool	needsResponse(const btCollisionObject* body0,const btCollisionObject* body1)=0;
+	virtual bool needsResponse(const btCollisionObject* body0, const btCollisionObject* body1) = 0;
 
-	virtual void	dispatchAllCollisionPairs(btOverlappingPairCache* pairCache,const btDispatcherInfo& dispatchInfo,btDispatcher* dispatcher)  =0;
+	virtual void dispatchAllCollisionPairs(btOverlappingPairCache* pairCache, const btDispatcherInfo& dispatchInfo, btDispatcher* dispatcher) = 0;
 
 	virtual int getNumManifolds() const = 0;
 
 	virtual btPersistentManifold* getManifoldByIndexInternal(int index) = 0;
 
-	virtual	btPersistentManifold**	getInternalManifoldPointer() = 0;
+	virtual btPersistentManifold** getInternalManifoldPointer() = 0;
 
-	virtual	btPoolAllocator*	getInternalManifoldPool() = 0;
+	virtual btPoolAllocator* getInternalManifoldPool() = 0;
 
-	virtual	const btPoolAllocator*	getInternalManifoldPool() const = 0;
+	virtual const btPoolAllocator* getInternalManifoldPool() const = 0;
 
-	virtual	void* allocateCollisionAlgorithm(int size)  = 0;
-
-	virtual	void freeCollisionAlgorithm(void* ptr) = 0;
+	virtual void* allocateCollisionAlgorithm(int size) = 0;
 
+	virtual void freeCollisionAlgorithm(void* ptr) = 0;
 };
 
-
-#endif //BT_DISPATCHER_H
+#endif  //BT_DISPATCHER_H
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btOverlappingPairCache.cpp b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btOverlappingPairCache.cpp
index 9e3337c5f65..8ce1087c9f1 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btOverlappingPairCache.cpp
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btOverlappingPairCache.cpp
@@ -13,8 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #include "btOverlappingPairCache.h"
 
 #include "btDispatcher.h"
@@ -23,121 +21,95 @@ subject to the following restrictions:
 
 #include <stdio.h>
 
-
-
-
-
-btHashedOverlappingPairCache::btHashedOverlappingPairCache():
-	m_overlapFilterCallback(0),
-	m_ghostPairCallback(0)
+btHashedOverlappingPairCache::btHashedOverlappingPairCache() : m_overlapFilterCallback(0),
+															   m_ghostPairCallback(0)
 {
-	int initialAllocatedSize= 2;
+	int initialAllocatedSize = 2;
 	m_overlappingPairArray.reserve(initialAllocatedSize);
 	growTables();
 }
 
-
-
-
 btHashedOverlappingPairCache::~btHashedOverlappingPairCache()
 {
 }
 
-
-
-void	btHashedOverlappingPairCache::cleanOverlappingPair(btBroadphasePair& pair,btDispatcher* dispatcher)
+void btHashedOverlappingPairCache::cleanOverlappingPair(btBroadphasePair& pair, btDispatcher* dispatcher)
 {
 	if (pair.m_algorithm && dispatcher)
 	{
 		{
 			pair.m_algorithm->~btCollisionAlgorithm();
 			dispatcher->freeCollisionAlgorithm(pair.m_algorithm);
-			pair.m_algorithm=0;
+			pair.m_algorithm = 0;
 		}
 	}
 }
 
-
-
-
-void	btHashedOverlappingPairCache::cleanProxyFromPairs(btBroadphaseProxy* proxy,btDispatcher* dispatcher)
+void btHashedOverlappingPairCache::cleanProxyFromPairs(btBroadphaseProxy* proxy, btDispatcher* dispatcher)
 {
-
-	class	CleanPairCallback : public btOverlapCallback
+	class CleanPairCallback : public btOverlapCallback
 	{
 		btBroadphaseProxy* m_cleanProxy;
-		btOverlappingPairCache*	m_pairCache;
+		btOverlappingPairCache* m_pairCache;
 		btDispatcher* m_dispatcher;
 
 	public:
-		CleanPairCallback(btBroadphaseProxy* cleanProxy,btOverlappingPairCache* pairCache,btDispatcher* dispatcher)
-			:m_cleanProxy(cleanProxy),
-			m_pairCache(pairCache),
-			m_dispatcher(dispatcher)
+		CleanPairCallback(btBroadphaseProxy* cleanProxy, btOverlappingPairCache* pairCache, btDispatcher* dispatcher)
+			: m_cleanProxy(cleanProxy),
+			  m_pairCache(pairCache),
+			  m_dispatcher(dispatcher)
 		{
 		}
-		virtual	bool	processOverlap(btBroadphasePair& pair)
+		virtual bool processOverlap(btBroadphasePair& pair)
 		{
 			if ((pair.m_pProxy0 == m_cleanProxy) ||
 				(pair.m_pProxy1 == m_cleanProxy))
 			{
-				m_pairCache->cleanOverlappingPair(pair,m_dispatcher);
+				m_pairCache->cleanOverlappingPair(pair, m_dispatcher);
 			}
 			return false;
 		}
-		
 	};
 
-	CleanPairCallback cleanPairs(proxy,this,dispatcher);
-
-	processAllOverlappingPairs(&cleanPairs,dispatcher);
+	CleanPairCallback cleanPairs(proxy, this, dispatcher);
 
+	processAllOverlappingPairs(&cleanPairs, dispatcher);
 }
 
-
-
-
-void	btHashedOverlappingPairCache::removeOverlappingPairsContainingProxy(btBroadphaseProxy* proxy,btDispatcher* dispatcher)
+void btHashedOverlappingPairCache::removeOverlappingPairsContainingProxy(btBroadphaseProxy* proxy, btDispatcher* dispatcher)
 {
-
-	class	RemovePairCallback : public btOverlapCallback
+	class RemovePairCallback : public btOverlapCallback
 	{
 		btBroadphaseProxy* m_obsoleteProxy;
 
 	public:
 		RemovePairCallback(btBroadphaseProxy* obsoleteProxy)
-			:m_obsoleteProxy(obsoleteProxy)
+			: m_obsoleteProxy(obsoleteProxy)
 		{
 		}
-		virtual	bool	processOverlap(btBroadphasePair& pair)
+		virtual bool processOverlap(btBroadphasePair& pair)
 		{
 			return ((pair.m_pProxy0 == m_obsoleteProxy) ||
-				(pair.m_pProxy1 == m_obsoleteProxy));
+					(pair.m_pProxy1 == m_obsoleteProxy));
 		}
-		
 	};
 
-
 	RemovePairCallback removeCallback(proxy);
 
-	processAllOverlappingPairs(&removeCallback,dispatcher);
+	processAllOverlappingPairs(&removeCallback, dispatcher);
 }
 
-
-
-
-
 btBroadphasePair* btHashedOverlappingPairCache::findPair(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1)
 {
-	if(proxy0->m_uniqueId>proxy1->m_uniqueId)
-		btSwap(proxy0,proxy1);
+	if (proxy0->m_uniqueId > proxy1->m_uniqueId)
+		btSwap(proxy0, proxy1);
 	int proxyId1 = proxy0->getUid();
 	int proxyId2 = proxy1->getUid();
 
 	/*if (proxyId1 > proxyId2)
 		btSwap(proxyId1, proxyId2);*/
 
-	int hash = static_cast<int>(getHash(static_cast<unsigned int>(proxyId1), static_cast<unsigned int>(proxyId2)) & (m_overlappingPairArray.capacity()-1));
+	int hash = static_cast<int>(getHash(static_cast<unsigned int>(proxyId1), static_cast<unsigned int>(proxyId2)) & (m_overlappingPairArray.capacity() - 1));
 
 	if (hash >= m_hashTable.size())
 	{
@@ -162,9 +134,8 @@ btBroadphasePair* btHashedOverlappingPairCache::findPair(btBroadphaseProxy* prox
 
 //#include <stdio.h>
 
-void	btHashedOverlappingPairCache::growTables()
+void btHashedOverlappingPairCache::growTables()
 {
-
 	int newCapacity = m_overlappingPairArray.capacity();
 
 	if (m_hashTable.size() < newCapacity)
@@ -175,10 +146,9 @@ void	btHashedOverlappingPairCache::growTables()
 		m_hashTable.resize(newCapacity);
 		m_next.resize(newCapacity);
 
-
 		int i;
 
-		for (i= 0; i < newCapacity; ++i)
+		for (i = 0; i < newCapacity; ++i)
 		{
 			m_hashTable[i] = BT_NULL_PAIR;
 		}
@@ -187,35 +157,31 @@ void	btHashedOverlappingPairCache::growTables()
 			m_next[i] = BT_NULL_PAIR;
 		}
 
-		for(i=0;i<curHashtableSize;i++)
+		for (i = 0; i < curHashtableSize; i++)
 		{
-	
 			const btBroadphasePair& pair = m_overlappingPairArray[i];
 			int proxyId1 = pair.m_pProxy0->getUid();
 			int proxyId2 = pair.m_pProxy1->getUid();
 			/*if (proxyId1 > proxyId2) 
 				btSwap(proxyId1, proxyId2);*/
-			int	hashValue = static_cast<int>(getHash(static_cast<unsigned int>(proxyId1),static_cast<unsigned int>(proxyId2)) & (m_overlappingPairArray.capacity()-1));	// New hash value with new mask
+			int hashValue = static_cast<int>(getHash(static_cast<unsigned int>(proxyId1), static_cast<unsigned int>(proxyId2)) & (m_overlappingPairArray.capacity() - 1));  // New hash value with new mask
 			m_next[i] = m_hashTable[hashValue];
 			m_hashTable[hashValue] = i;
 		}
-
-
 	}
 }
 
 btBroadphasePair* btHashedOverlappingPairCache::internalAddPair(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1)
 {
-	if(proxy0->m_uniqueId>proxy1->m_uniqueId) 
-		btSwap(proxy0,proxy1);
+	if (proxy0->m_uniqueId > proxy1->m_uniqueId)
+		btSwap(proxy0, proxy1);
 	int proxyId1 = proxy0->getUid();
 	int proxyId2 = proxy1->getUid();
 
 	/*if (proxyId1 > proxyId2) 
 		btSwap(proxyId1, proxyId2);*/
 
-	int	hash = static_cast<int>(getHash(static_cast<unsigned int>(proxyId1),static_cast<unsigned int>(proxyId2)) & (m_overlappingPairArray.capacity()-1));	// New hash value with new mask
-
+	int hash = static_cast<int>(getHash(static_cast<unsigned int>(proxyId1), static_cast<unsigned int>(proxyId2)) & (m_overlappingPairArray.capacity() - 1));  // New hash value with new mask
 
 	btBroadphasePair* pair = internalFindPair(proxy0, proxy1, hash);
 	if (pair != NULL)
@@ -237,7 +203,7 @@ btBroadphasePair* btHashedOverlappingPairCache::internalAddPair(btBroadphaseProx
 
 	//this is where we add an actual pair, so also call the 'ghost'
 	if (m_ghostPairCallback)
-		m_ghostPairCallback->addOverlappingPair(proxy0,proxy1);
+		m_ghostPairCallback->addOverlappingPair(proxy0, proxy1);
 
 	int newCapacity = m_overlappingPairArray.capacity();
 
@@ -245,15 +211,14 @@ btBroadphasePair* btHashedOverlappingPairCache::internalAddPair(btBroadphaseProx
 	{
 		growTables();
 		//hash with new capacity
-		hash = static_cast<int>(getHash(static_cast<unsigned int>(proxyId1),static_cast<unsigned int>(proxyId2)) & (m_overlappingPairArray.capacity()-1));
+		hash = static_cast<int>(getHash(static_cast<unsigned int>(proxyId1), static_cast<unsigned int>(proxyId2)) & (m_overlappingPairArray.capacity() - 1));
 	}
-	
-	pair = new (mem) btBroadphasePair(*proxy0,*proxy1);
-//	pair->m_pProxy0 = proxy0;
-//	pair->m_pProxy1 = proxy1;
+
+	pair = new (mem) btBroadphasePair(*proxy0, *proxy1);
+	//	pair->m_pProxy0 = proxy0;
+	//	pair->m_pProxy1 = proxy1;
 	pair->m_algorithm = 0;
 	pair->m_internalTmpValue = 0;
-	
 
 	m_next[count] = m_hashTable[hash];
 	m_hashTable[hash] = count;
@@ -261,19 +226,17 @@ btBroadphasePair* btHashedOverlappingPairCache::internalAddPair(btBroadphaseProx
 	return pair;
 }
 
-
-
-void* btHashedOverlappingPairCache::removeOverlappingPair(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1,btDispatcher* dispatcher)
+void* btHashedOverlappingPairCache::removeOverlappingPair(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1, btDispatcher* dispatcher)
 {
-	if(proxy0->m_uniqueId>proxy1->m_uniqueId)
-		btSwap(proxy0,proxy1);
+	if (proxy0->m_uniqueId > proxy1->m_uniqueId)
+		btSwap(proxy0, proxy1);
 	int proxyId1 = proxy0->getUid();
 	int proxyId2 = proxy1->getUid();
 
 	/*if (proxyId1 > proxyId2)
 		btSwap(proxyId1, proxyId2);*/
 
-	int	hash = static_cast<int>(getHash(static_cast<unsigned int>(proxyId1),static_cast<unsigned int>(proxyId2)) & (m_overlappingPairArray.capacity()-1));
+	int hash = static_cast<int>(getHash(static_cast<unsigned int>(proxyId1), static_cast<unsigned int>(proxyId2)) & (m_overlappingPairArray.capacity() - 1));
 
 	btBroadphasePair* pair = internalFindPair(proxy0, proxy1, hash);
 	if (pair == NULL)
@@ -281,7 +244,7 @@ void* btHashedOverlappingPairCache::removeOverlappingPair(btBroadphaseProxy* pro
 		return 0;
 	}
 
-	cleanOverlappingPair(*pair,dispatcher);
+	cleanOverlappingPair(*pair, dispatcher);
 
 	void* userData = pair->m_internalInfo1;
 
@@ -319,7 +282,7 @@ void* btHashedOverlappingPairCache::removeOverlappingPair(btBroadphaseProxy* pro
 	int lastPairIndex = m_overlappingPairArray.size() - 1;
 
 	if (m_ghostPairCallback)
-		m_ghostPairCallback->removeOverlappingPair(proxy0, proxy1,dispatcher);
+		m_ghostPairCallback->removeOverlappingPair(proxy0, proxy1, dispatcher);
 
 	// If the removed pair is the last pair, we are done.
 	if (lastPairIndex == pairIndex)
@@ -330,8 +293,8 @@ void* btHashedOverlappingPairCache::removeOverlappingPair(btBroadphaseProxy* pro
 
 	// Remove the last pair from the hash table.
 	const btBroadphasePair* last = &m_overlappingPairArray[lastPairIndex];
-		/* missing swap here too, Nat. */ 
-	int lastHash = static_cast<int>(getHash(static_cast<unsigned int>(last->m_pProxy0->getUid()), static_cast<unsigned int>(last->m_pProxy1->getUid())) & (m_overlappingPairArray.capacity()-1));
+	/* missing swap here too, Nat. */
+	int lastHash = static_cast<int>(getHash(static_cast<unsigned int>(last->m_pProxy0->getUid()), static_cast<unsigned int>(last->m_pProxy1->getUid())) & (m_overlappingPairArray.capacity() - 1));
 
 	index = m_hashTable[lastHash];
 	btAssert(index != BT_NULL_PAIR);
@@ -366,20 +329,20 @@ void* btHashedOverlappingPairCache::removeOverlappingPair(btBroadphaseProxy* pro
 }
 //#include <stdio.h>
 #include "LinearMath/btQuickprof.h"
-void	btHashedOverlappingPairCache::processAllOverlappingPairs(btOverlapCallback* callback,btDispatcher* dispatcher)
+void btHashedOverlappingPairCache::processAllOverlappingPairs(btOverlapCallback* callback, btDispatcher* dispatcher)
 {
 	BT_PROFILE("btHashedOverlappingPairCache::processAllOverlappingPairs");
 	int i;
 
-//	printf("m_overlappingPairArray.size()=%d\n",m_overlappingPairArray.size());
-	for (i=0;i<m_overlappingPairArray.size();)
+	//	printf("m_overlappingPairArray.size()=%d\n",m_overlappingPairArray.size());
+	for (i = 0; i < m_overlappingPairArray.size();)
 	{
-	
 		btBroadphasePair* pair = &m_overlappingPairArray[i];
 		if (callback->processOverlap(*pair))
 		{
-			removeOverlappingPair(pair->m_pProxy0,pair->m_pProxy1,dispatcher);
-		} else
+			removeOverlappingPair(pair->m_pProxy0, pair->m_pProxy1, dispatcher);
+		}
+		else
 		{
 			i++;
 		}
@@ -388,83 +351,83 @@ void	btHashedOverlappingPairCache::processAllOverlappingPairs(btOverlapCallback*
 
 struct MyPairIndex
 {
-    int m_orgIndex;
-    int m_uidA0;
-    int m_uidA1;
+	int m_orgIndex;
+	int m_uidA0;
+	int m_uidA1;
 };
 
 class MyPairIndeSortPredicate
 {
 public:
-    
-    bool operator() ( const MyPairIndex& a, const MyPairIndex& b ) const
-    {
-        const int uidA0 = a.m_uidA0;
-        const int uidB0 = b.m_uidA0;
-        const int uidA1 = a.m_uidA1;
-        const int uidB1 = b.m_uidA1;
-        return uidA0 > uidB0 || (uidA0 == uidB0 && uidA1 > uidB1);
-    }
+	bool operator()(const MyPairIndex& a, const MyPairIndex& b) const
+	{
+		const int uidA0 = a.m_uidA0;
+		const int uidB0 = b.m_uidA0;
+		const int uidA1 = a.m_uidA1;
+		const int uidB1 = b.m_uidA1;
+		return uidA0 > uidB0 || (uidA0 == uidB0 && uidA1 > uidB1);
+	}
 };
 
-void    btHashedOverlappingPairCache::processAllOverlappingPairs(btOverlapCallback* callback,btDispatcher* dispatcher, const struct btDispatcherInfo& dispatchInfo)
+void btHashedOverlappingPairCache::processAllOverlappingPairs(btOverlapCallback* callback, btDispatcher* dispatcher, const struct btDispatcherInfo& dispatchInfo)
 {
-    if (dispatchInfo.m_deterministicOverlappingPairs)
-    {
-        btBroadphasePairArray& pa = getOverlappingPairArray();
-        btAlignedObjectArray<MyPairIndex> indices;
-        {
-            BT_PROFILE("sortOverlappingPairs");
-            indices.resize(pa.size());
-            for (int i=0;i<indices.size();i++)
-            {
-                const btBroadphasePair& p = pa[i];
-                const int uidA0 = p.m_pProxy0 ? p.m_pProxy0->m_uniqueId : -1;
-                const int uidA1 = p.m_pProxy1 ? p.m_pProxy1->m_uniqueId : -1;
-                
-                indices[i].m_uidA0 = uidA0;
-                indices[i].m_uidA1 = uidA1;
-                indices[i].m_orgIndex = i;
-            }
-            indices.quickSort(MyPairIndeSortPredicate());
-        }
-        {
-            BT_PROFILE("btHashedOverlappingPairCache::processAllOverlappingPairs");
-            int i;
-            for (i=0;i<indices.size();)
-            {
-                btBroadphasePair* pair = &pa[indices[i].m_orgIndex];
-                if (callback->processOverlap(*pair))
-                {
-                    removeOverlappingPair(pair->m_pProxy0,pair->m_pProxy1,dispatcher);
-                } else
-                {
-                    i++;
-                }
-            }
-        }
-    } else
-    {
-        processAllOverlappingPairs(callback, dispatcher);
-    }
+	if (dispatchInfo.m_deterministicOverlappingPairs)
+	{
+		btBroadphasePairArray& pa = getOverlappingPairArray();
+		btAlignedObjectArray<MyPairIndex> indices;
+		{
+			BT_PROFILE("sortOverlappingPairs");
+			indices.resize(pa.size());
+			for (int i = 0; i < indices.size(); i++)
+			{
+				const btBroadphasePair& p = pa[i];
+				const int uidA0 = p.m_pProxy0 ? p.m_pProxy0->m_uniqueId : -1;
+				const int uidA1 = p.m_pProxy1 ? p.m_pProxy1->m_uniqueId : -1;
+
+				indices[i].m_uidA0 = uidA0;
+				indices[i].m_uidA1 = uidA1;
+				indices[i].m_orgIndex = i;
+			}
+			indices.quickSort(MyPairIndeSortPredicate());
+		}
+		{
+			BT_PROFILE("btHashedOverlappingPairCache::processAllOverlappingPairs");
+			int i;
+			for (i = 0; i < indices.size();)
+			{
+				btBroadphasePair* pair = &pa[indices[i].m_orgIndex];
+				if (callback->processOverlap(*pair))
+				{
+					removeOverlappingPair(pair->m_pProxy0, pair->m_pProxy1, dispatcher);
+				}
+				else
+				{
+					i++;
+				}
+			}
+		}
+	}
+	else
+	{
+		processAllOverlappingPairs(callback, dispatcher);
+	}
 }
 
-
-void	btHashedOverlappingPairCache::sortOverlappingPairs(btDispatcher* dispatcher)
+void btHashedOverlappingPairCache::sortOverlappingPairs(btDispatcher* dispatcher)
 {
 	///need to keep hashmap in sync with pair address, so rebuild all
 	btBroadphasePairArray tmpPairs;
 	int i;
-	for (i=0;i<m_overlappingPairArray.size();i++)
+	for (i = 0; i < m_overlappingPairArray.size(); i++)
 	{
 		tmpPairs.push_back(m_overlappingPairArray[i]);
 	}
 
-	for (i=0;i<tmpPairs.size();i++)
+	for (i = 0; i < tmpPairs.size(); i++)
 	{
-		removeOverlappingPair(tmpPairs[i].m_pProxy0,tmpPairs[i].m_pProxy1,dispatcher);
+		removeOverlappingPair(tmpPairs[i].m_pProxy0, tmpPairs[i].m_pProxy1, dispatcher);
 	}
-	
+
 	for (i = 0; i < m_next.size(); i++)
 	{
 		m_next[i] = BT_NULL_PAIR;
@@ -472,31 +435,28 @@ void	btHashedOverlappingPairCache::sortOverlappingPairs(btDispatcher* dispatcher
 
 	tmpPairs.quickSort(btBroadphasePairSortPredicate());
 
-	for (i=0;i<tmpPairs.size();i++)
+	for (i = 0; i < tmpPairs.size(); i++)
 	{
-		addOverlappingPair(tmpPairs[i].m_pProxy0,tmpPairs[i].m_pProxy1);
+		addOverlappingPair(tmpPairs[i].m_pProxy0, tmpPairs[i].m_pProxy1);
 	}
-
-	
 }
 
-
-void*	btSortedOverlappingPairCache::removeOverlappingPair(btBroadphaseProxy* proxy0,btBroadphaseProxy* proxy1, btDispatcher* dispatcher )
+void* btSortedOverlappingPairCache::removeOverlappingPair(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1, btDispatcher* dispatcher)
 {
 	if (!hasDeferredRemoval())
 	{
-		btBroadphasePair findPair(*proxy0,*proxy1);
+		btBroadphasePair findPair(*proxy0, *proxy1);
 
 		int findIndex = m_overlappingPairArray.findLinearSearch(findPair);
 		if (findIndex < m_overlappingPairArray.size())
 		{
 			btBroadphasePair& pair = m_overlappingPairArray[findIndex];
 			void* userData = pair.m_internalInfo1;
-			cleanOverlappingPair(pair,dispatcher);
+			cleanOverlappingPair(pair, dispatcher);
 			if (m_ghostPairCallback)
-				m_ghostPairCallback->removeOverlappingPair(proxy0, proxy1,dispatcher);
-			
-			m_overlappingPairArray.swap(findIndex,m_overlappingPairArray.capacity()-1);
+				m_ghostPairCallback->removeOverlappingPair(proxy0, proxy1, dispatcher);
+
+			m_overlappingPairArray.swap(findIndex, m_overlappingPairArray.capacity() - 1);
 			m_overlappingPairArray.pop_back();
 			return userData;
 		}
@@ -505,95 +465,73 @@ void*	btSortedOverlappingPairCache::removeOverlappingPair(btBroadphaseProxy* pro
 	return 0;
 }
 
-
-
-
-
-
-
-
-btBroadphasePair*	btSortedOverlappingPairCache::addOverlappingPair(btBroadphaseProxy* proxy0,btBroadphaseProxy* proxy1)
+btBroadphasePair* btSortedOverlappingPairCache::addOverlappingPair(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1)
 {
 	//don't add overlap with own
 	btAssert(proxy0 != proxy1);
 
-	if (!needsBroadphaseCollision(proxy0,proxy1))
+	if (!needsBroadphaseCollision(proxy0, proxy1))
 		return 0;
-	
-	void* mem = &m_overlappingPairArray.expandNonInitializing();
-	btBroadphasePair* pair = new (mem) btBroadphasePair(*proxy0,*proxy1);
 
-  if (m_ghostPairCallback)
+	void* mem = &m_overlappingPairArray.expandNonInitializing();
+	btBroadphasePair* pair = new (mem) btBroadphasePair(*proxy0, *proxy1);
+
+	if (m_ghostPairCallback)
 		m_ghostPairCallback->addOverlappingPair(proxy0, proxy1);
 	return pair;
-	
 }
 
 ///this findPair becomes really slow. Either sort the list to speedup the query, or
 ///use a different solution. It is mainly used for Removing overlapping pairs. Removal could be delayed.
 ///we could keep a linked list in each proxy, and store pair in one of the proxies (with lowest memory address)
 ///Also we can use a 2D bitmap, which can be useful for a future GPU implementation
- btBroadphasePair*	btSortedOverlappingPairCache::findPair(btBroadphaseProxy* proxy0,btBroadphaseProxy* proxy1)
+btBroadphasePair* btSortedOverlappingPairCache::findPair(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1)
 {
-	if (!needsBroadphaseCollision(proxy0,proxy1))
+	if (!needsBroadphaseCollision(proxy0, proxy1))
 		return 0;
 
-	btBroadphasePair tmpPair(*proxy0,*proxy1);
+	btBroadphasePair tmpPair(*proxy0, *proxy1);
 	int findIndex = m_overlappingPairArray.findLinearSearch(tmpPair);
 
 	if (findIndex < m_overlappingPairArray.size())
 	{
 		//btAssert(it != m_overlappingPairSet.end());
-		 btBroadphasePair* pair = &m_overlappingPairArray[findIndex];
+		btBroadphasePair* pair = &m_overlappingPairArray[findIndex];
 		return pair;
 	}
 	return 0;
 }
 
-
-
-
-
-
-
-
-
-
 //#include <stdio.h>
 
-void	btSortedOverlappingPairCache::processAllOverlappingPairs(btOverlapCallback* callback,btDispatcher* dispatcher)
+void btSortedOverlappingPairCache::processAllOverlappingPairs(btOverlapCallback* callback, btDispatcher* dispatcher)
 {
-
 	int i;
 
-	for (i=0;i<m_overlappingPairArray.size();)
+	for (i = 0; i < m_overlappingPairArray.size();)
 	{
-	
 		btBroadphasePair* pair = &m_overlappingPairArray[i];
 		if (callback->processOverlap(*pair))
 		{
-			cleanOverlappingPair(*pair,dispatcher);
+			cleanOverlappingPair(*pair, dispatcher);
 			pair->m_pProxy0 = 0;
 			pair->m_pProxy1 = 0;
-			m_overlappingPairArray.swap(i,m_overlappingPairArray.size()-1);
+			m_overlappingPairArray.swap(i, m_overlappingPairArray.size() - 1);
 			m_overlappingPairArray.pop_back();
-		} else
+		}
+		else
 		{
 			i++;
 		}
 	}
 }
 
-
-
-
-btSortedOverlappingPairCache::btSortedOverlappingPairCache():
-	m_blockedForChanges(false),
-	m_hasDeferredRemoval(true),
-	m_overlapFilterCallback(0),
-	m_ghostPairCallback(0)
+btSortedOverlappingPairCache::btSortedOverlappingPairCache() : m_blockedForChanges(false),
+															   m_hasDeferredRemoval(true),
+															   m_overlapFilterCallback(0),
+															   m_ghostPairCallback(0)
 {
-	int initialAllocatedSize= 2;
+	int initialAllocatedSize = 2;
 	m_overlappingPairArray.reserve(initialAllocatedSize);
 }
 
@@ -601,81 +539,73 @@ btSortedOverlappingPairCache::~btSortedOverlappingPairCache()
 {
 }
 
-void	btSortedOverlappingPairCache::cleanOverlappingPair(btBroadphasePair& pair,btDispatcher* dispatcher)
+void btSortedOverlappingPairCache::cleanOverlappingPair(btBroadphasePair& pair, btDispatcher* dispatcher)
 {
 	if (pair.m_algorithm)
 	{
 		{
 			pair.m_algorithm->~btCollisionAlgorithm();
 			dispatcher->freeCollisionAlgorithm(pair.m_algorithm);
-			pair.m_algorithm=0;
+			pair.m_algorithm = 0;
 		}
 	}
 }
 
-
-void	btSortedOverlappingPairCache::cleanProxyFromPairs(btBroadphaseProxy* proxy,btDispatcher* dispatcher)
+void btSortedOverlappingPairCache::cleanProxyFromPairs(btBroadphaseProxy* proxy, btDispatcher* dispatcher)
 {
-
-	class	CleanPairCallback : public btOverlapCallback
+	class CleanPairCallback : public btOverlapCallback
 	{
 		btBroadphaseProxy* m_cleanProxy;
-		btOverlappingPairCache*	m_pairCache;
+		btOverlappingPairCache* m_pairCache;
 		btDispatcher* m_dispatcher;
 
 	public:
-		CleanPairCallback(btBroadphaseProxy* cleanProxy,btOverlappingPairCache* pairCache,btDispatcher* dispatcher)
-			:m_cleanProxy(cleanProxy),
-			m_pairCache(pairCache),
-			m_dispatcher(dispatcher)
+		CleanPairCallback(btBroadphaseProxy* cleanProxy, btOverlappingPairCache* pairCache, btDispatcher* dispatcher)
+			: m_cleanProxy(cleanProxy),
+			  m_pairCache(pairCache),
+			  m_dispatcher(dispatcher)
 		{
 		}
-		virtual	bool	processOverlap(btBroadphasePair& pair)
+		virtual bool processOverlap(btBroadphasePair& pair)
 		{
 			if ((pair.m_pProxy0 == m_cleanProxy) ||
 				(pair.m_pProxy1 == m_cleanProxy))
 			{
-				m_pairCache->cleanOverlappingPair(pair,m_dispatcher);
+				m_pairCache->cleanOverlappingPair(pair, m_dispatcher);
 			}
 			return false;
 		}
-		
 	};
 
-	CleanPairCallback cleanPairs(proxy,this,dispatcher);
-
-	processAllOverlappingPairs(&cleanPairs,dispatcher);
+	CleanPairCallback cleanPairs(proxy, this, dispatcher);
 
+	processAllOverlappingPairs(&cleanPairs, dispatcher);
 }
 
-
-void	btSortedOverlappingPairCache::removeOverlappingPairsContainingProxy(btBroadphaseProxy* proxy,btDispatcher* dispatcher)
+void btSortedOverlappingPairCache::removeOverlappingPairsContainingProxy(btBroadphaseProxy* proxy, btDispatcher* dispatcher)
 {
-
-	class	RemovePairCallback : public btOverlapCallback
+	class RemovePairCallback : public btOverlapCallback
 	{
 		btBroadphaseProxy* m_obsoleteProxy;
 
 	public:
 		RemovePairCallback(btBroadphaseProxy* obsoleteProxy)
-			:m_obsoleteProxy(obsoleteProxy)
+			: m_obsoleteProxy(obsoleteProxy)
 		{
 		}
-		virtual	bool	processOverlap(btBroadphasePair& pair)
+		virtual bool processOverlap(btBroadphasePair& pair)
 		{
 			return ((pair.m_pProxy0 == m_obsoleteProxy) ||
-				(pair.m_pProxy1 == m_obsoleteProxy));
+					(pair.m_pProxy1 == m_obsoleteProxy));
 		}
-		
 	};
 
 	RemovePairCallback removeCallback(proxy);
 
-	processAllOverlappingPairs(&removeCallback,dispatcher);
+	processAllOverlappingPairs(&removeCallback, dispatcher);
 }
 
-void	btSortedOverlappingPairCache::sortOverlappingPairs(btDispatcher* dispatcher)
+void btSortedOverlappingPairCache::sortOverlappingPairs(btDispatcher* dispatcher)
 {
 	//should already be sorted
 }
-
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btOverlappingPairCache.h b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btOverlappingPairCache.h
index 7a38d34f052..a85782bc8ac 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btOverlappingPairCache.h
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btOverlappingPairCache.h
@@ -16,7 +16,6 @@ subject to the following restrictions:
 #ifndef BT_OVERLAPPING_PAIR_CACHE_H
 #define BT_OVERLAPPING_PAIR_CACHE_H
 
-
 #include "btBroadphaseInterface.h"
 #include "btBroadphaseProxy.h"
 #include "btOverlappingPairCallback.h"
@@ -24,177 +23,163 @@ subject to the following restrictions:
 #include "LinearMath/btAlignedObjectArray.h"
 class btDispatcher;
 
-typedef btAlignedObjectArray<btBroadphasePair>	btBroadphasePairArray;
+typedef btAlignedObjectArray<btBroadphasePair> btBroadphasePairArray;
 
-struct	btOverlapCallback
+struct btOverlapCallback
 {
 	virtual ~btOverlapCallback()
-	{}
+	{
+	}
 	//return true for deletion of the pair
-	virtual bool	processOverlap(btBroadphasePair& pair) = 0;
-
+	virtual bool processOverlap(btBroadphasePair& pair) = 0;
 };
 
 struct btOverlapFilterCallback
 {
 	virtual ~btOverlapFilterCallback()
-	{}
+	{
+	}
 	// return true when pairs need collision
-	virtual bool	needBroadphaseCollision(btBroadphaseProxy* proxy0,btBroadphaseProxy* proxy1) const = 0;
+	virtual bool needBroadphaseCollision(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1) const = 0;
 };
 
-
-
-
-
-
-
-const int BT_NULL_PAIR=0xffffffff;
+const int BT_NULL_PAIR = 0xffffffff;
 
 ///The btOverlappingPairCache provides an interface for overlapping pair management (add, remove, storage), used by the btBroadphaseInterface broadphases.
 ///The btHashedOverlappingPairCache and btSortedOverlappingPairCache classes are two implementations.
 class btOverlappingPairCache : public btOverlappingPairCallback
 {
 public:
-	virtual ~btOverlappingPairCache() {} // this is needed so we can get to the derived class destructor
+	virtual ~btOverlappingPairCache() {}  // this is needed so we can get to the derived class destructor
 
-	virtual btBroadphasePair*	getOverlappingPairArrayPtr() = 0;
-	
-	virtual const btBroadphasePair*	getOverlappingPairArrayPtr() const = 0;
+	virtual btBroadphasePair* getOverlappingPairArrayPtr() = 0;
 
-	virtual btBroadphasePairArray&	getOverlappingPairArray() = 0;
+	virtual const btBroadphasePair* getOverlappingPairArrayPtr() const = 0;
 
-	virtual	void	cleanOverlappingPair(btBroadphasePair& pair,btDispatcher* dispatcher) = 0;
+	virtual btBroadphasePairArray& getOverlappingPairArray() = 0;
+
+	virtual void cleanOverlappingPair(btBroadphasePair& pair, btDispatcher* dispatcher) = 0;
 
 	virtual int getNumOverlappingPairs() const = 0;
 
-	virtual void	cleanProxyFromPairs(btBroadphaseProxy* proxy,btDispatcher* dispatcher) = 0;
+	virtual void cleanProxyFromPairs(btBroadphaseProxy* proxy, btDispatcher* dispatcher) = 0;
 
-	virtual	void setOverlapFilterCallback(btOverlapFilterCallback* callback) = 0;
+	virtual void setOverlapFilterCallback(btOverlapFilterCallback* callback) = 0;
 
-	virtual void	processAllOverlappingPairs(btOverlapCallback*,btDispatcher* dispatcher) = 0;
+	virtual void processAllOverlappingPairs(btOverlapCallback*, btDispatcher* dispatcher) = 0;
 
-	virtual void    processAllOverlappingPairs(btOverlapCallback* callback,btDispatcher* dispatcher, const struct btDispatcherInfo& dispatchInfo)
+	virtual void processAllOverlappingPairs(btOverlapCallback* callback, btDispatcher* dispatcher, const struct btDispatcherInfo& dispatchInfo)
 	{
 		processAllOverlappingPairs(callback, dispatcher);
 	}
 	virtual btBroadphasePair* findPair(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1) = 0;
 
-	virtual bool	hasDeferredRemoval() = 0;
-
-	virtual	void	setInternalGhostPairCallback(btOverlappingPairCallback* ghostPairCallback)=0;
-
-	virtual void	sortOverlappingPairs(btDispatcher* dispatcher) = 0;
+	virtual bool hasDeferredRemoval() = 0;
 
+	virtual void setInternalGhostPairCallback(btOverlappingPairCallback* ghostPairCallback) = 0;
 
+	virtual void sortOverlappingPairs(btDispatcher* dispatcher) = 0;
 };
 
 /// Hash-space based Pair Cache, thanks to Erin Catto, Box2D, http://www.box2d.org, and Pierre Terdiman, Codercorner, http://codercorner.com
 
-ATTRIBUTE_ALIGNED16(class) btHashedOverlappingPairCache : public btOverlappingPairCache
+ATTRIBUTE_ALIGNED16(class)
+btHashedOverlappingPairCache : public btOverlappingPairCache
 {
-	btBroadphasePairArray	m_overlappingPairArray;
+	btBroadphasePairArray m_overlappingPairArray;
 	btOverlapFilterCallback* m_overlapFilterCallback;
 
 protected:
-	
-	btAlignedObjectArray<int>	m_hashTable;
-	btAlignedObjectArray<int>	m_next;
-	btOverlappingPairCallback*	m_ghostPairCallback;
-
+	btAlignedObjectArray<int> m_hashTable;
+	btAlignedObjectArray<int> m_next;
+	btOverlappingPairCallback* m_ghostPairCallback;
 
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
+
 	btHashedOverlappingPairCache();
 	virtual ~btHashedOverlappingPairCache();
 
-	
-	void	removeOverlappingPairsContainingProxy(btBroadphaseProxy* proxy,btDispatcher* dispatcher);
+	void removeOverlappingPairsContainingProxy(btBroadphaseProxy * proxy, btDispatcher * dispatcher);
 
-	virtual void*	removeOverlappingPair(btBroadphaseProxy* proxy0,btBroadphaseProxy* proxy1,btDispatcher* dispatcher);
-	
-	SIMD_FORCE_INLINE bool needsBroadphaseCollision(btBroadphaseProxy* proxy0,btBroadphaseProxy* proxy1) const
+	virtual void* removeOverlappingPair(btBroadphaseProxy * proxy0, btBroadphaseProxy * proxy1, btDispatcher * dispatcher);
+
+	SIMD_FORCE_INLINE bool needsBroadphaseCollision(btBroadphaseProxy * proxy0, btBroadphaseProxy * proxy1) const
 	{
 		if (m_overlapFilterCallback)
-			return m_overlapFilterCallback->needBroadphaseCollision(proxy0,proxy1);
+			return m_overlapFilterCallback->needBroadphaseCollision(proxy0, proxy1);
 
 		bool collides = (proxy0->m_collisionFilterGroup & proxy1->m_collisionFilterMask) != 0;
 		collides = collides && (proxy1->m_collisionFilterGroup & proxy0->m_collisionFilterMask);
-		
+
 		return collides;
 	}
 
 	// Add a pair and return the new pair. If the pair already exists,
 	// no new pair is created and the old one is returned.
-	virtual btBroadphasePair* 	addOverlappingPair(btBroadphaseProxy* proxy0,btBroadphaseProxy* proxy1)
+	virtual btBroadphasePair* addOverlappingPair(btBroadphaseProxy * proxy0, btBroadphaseProxy * proxy1)
 	{
-		if (!needsBroadphaseCollision(proxy0,proxy1))
+		if (!needsBroadphaseCollision(proxy0, proxy1))
 			return 0;
 
-		return internalAddPair(proxy0,proxy1);
+		return internalAddPair(proxy0, proxy1);
 	}
 
-	
+	void cleanProxyFromPairs(btBroadphaseProxy * proxy, btDispatcher * dispatcher);
 
-	void	cleanProxyFromPairs(btBroadphaseProxy* proxy,btDispatcher* dispatcher);
+	virtual void processAllOverlappingPairs(btOverlapCallback*, btDispatcher * dispatcher);
 
-	
-	virtual void	processAllOverlappingPairs(btOverlapCallback*,btDispatcher* dispatcher);
+	virtual void processAllOverlappingPairs(btOverlapCallback * callback, btDispatcher * dispatcher, const struct btDispatcherInfo& dispatchInfo);
 
-    virtual void    processAllOverlappingPairs(btOverlapCallback* callback,btDispatcher* dispatcher, const struct btDispatcherInfo& dispatchInfo);
-
-	virtual btBroadphasePair*	getOverlappingPairArrayPtr()
+	virtual btBroadphasePair* getOverlappingPairArrayPtr()
 	{
 		return &m_overlappingPairArray[0];
 	}
 
-	const btBroadphasePair*	getOverlappingPairArrayPtr() const
+	const btBroadphasePair* getOverlappingPairArrayPtr() const
 	{
 		return &m_overlappingPairArray[0];
 	}
 
-	btBroadphasePairArray&	getOverlappingPairArray()
+	btBroadphasePairArray& getOverlappingPairArray()
 	{
 		return m_overlappingPairArray;
 	}
 
-	const btBroadphasePairArray&	getOverlappingPairArray() const
+	const btBroadphasePairArray& getOverlappingPairArray() const
 	{
 		return m_overlappingPairArray;
 	}
 
-	void	cleanOverlappingPair(btBroadphasePair& pair,btDispatcher* dispatcher);
+	void cleanOverlappingPair(btBroadphasePair & pair, btDispatcher * dispatcher);
 
-
-
-	btBroadphasePair* findPair(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1);
+	btBroadphasePair* findPair(btBroadphaseProxy * proxy0, btBroadphaseProxy * proxy1);
 
 	int GetCount() const { return m_overlappingPairArray.size(); }
-//	btBroadphasePair* GetPairs() { return m_pairs; }
+	//	btBroadphasePair* GetPairs() { return m_pairs; }
 
 	btOverlapFilterCallback* getOverlapFilterCallback()
 	{
 		return m_overlapFilterCallback;
 	}
 
-	void setOverlapFilterCallback(btOverlapFilterCallback* callback)
+	void setOverlapFilterCallback(btOverlapFilterCallback * callback)
 	{
 		m_overlapFilterCallback = callback;
 	}
 
-	int	getNumOverlappingPairs() const
+	int getNumOverlappingPairs() const
 	{
 		return m_overlappingPairArray.size();
 	}
-private:
-	
-	btBroadphasePair* 	internalAddPair(btBroadphaseProxy* proxy0,btBroadphaseProxy* proxy1);
 
-	void	growTables();
+private:
+	btBroadphasePair* internalAddPair(btBroadphaseProxy * proxy0, btBroadphaseProxy * proxy1);
+
+	void growTables();
 
 	SIMD_FORCE_INLINE bool equalsPair(const btBroadphasePair& pair, int proxyId1, int proxyId2)
-	{	
+	{
 		return pair.m_pProxy0->getUid() == proxyId1 && pair.m_pProxy1->getUid() == proxyId2;
 	}
 
@@ -214,40 +199,37 @@ private:
 	}
 	*/
 
-
 	SIMD_FORCE_INLINE unsigned int getHash(unsigned int proxyId1, unsigned int proxyId2)
 	{
 		unsigned int key = proxyId1 | (proxyId2 << 16);
 		// Thomas Wang's hash
 
 		key += ~(key << 15);
-		key ^=  (key >> 10);
-		key +=  (key << 3);
-		key ^=  (key >> 6);
+		key ^= (key >> 10);
+		key += (key << 3);
+		key ^= (key >> 6);
 		key += ~(key << 11);
-		key ^=  (key >> 16);
+		key ^= (key >> 16);
 		return key;
 	}
-	
 
-
-	SIMD_FORCE_INLINE btBroadphasePair* internalFindPair(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1, int hash)
+	SIMD_FORCE_INLINE btBroadphasePair* internalFindPair(btBroadphaseProxy * proxy0, btBroadphaseProxy * proxy1, int hash)
 	{
 		int proxyId1 = proxy0->getUid();
 		int proxyId2 = proxy1->getUid();
-		#if 0 // wrong, 'equalsPair' use unsorted uids, copy-past devil striked again. Nat.
+#if 0  // wrong, 'equalsPair' use unsorted uids, copy-past devil striked again. Nat.
 		if (proxyId1 > proxyId2) 
 			btSwap(proxyId1, proxyId2);
-		#endif
+#endif
 
 		int index = m_hashTable[hash];
-		
-		while( index != BT_NULL_PAIR && equalsPair(m_overlappingPairArray[index], proxyId1, proxyId2) == false)
+
+		while (index != BT_NULL_PAIR && equalsPair(m_overlappingPairArray[index], proxyId1, proxyId2) == false)
 		{
 			index = m_next[index];
 		}
 
-		if ( index == BT_NULL_PAIR )
+		if (index == BT_NULL_PAIR)
 		{
 			return NULL;
 		}
@@ -257,155 +239,136 @@ private:
 		return &m_overlappingPairArray[index];
 	}
 
-	virtual bool	hasDeferredRemoval()
+	virtual bool hasDeferredRemoval()
 	{
 		return false;
 	}
 
-	virtual	void	setInternalGhostPairCallback(btOverlappingPairCallback* ghostPairCallback)
+	virtual void setInternalGhostPairCallback(btOverlappingPairCallback * ghostPairCallback)
 	{
 		m_ghostPairCallback = ghostPairCallback;
 	}
 
-	virtual void	sortOverlappingPairs(btDispatcher* dispatcher);
-	
-
-	
+	virtual void sortOverlappingPairs(btDispatcher * dispatcher);
 };
 
-
-
-
 ///btSortedOverlappingPairCache maintains the objects with overlapping AABB
 ///Typically managed by the Broadphase, Axis3Sweep or btSimpleBroadphase
-class	btSortedOverlappingPairCache : public btOverlappingPairCache
+class btSortedOverlappingPairCache : public btOverlappingPairCache
 {
-	protected:
-		//avoid brute-force finding all the time
-		btBroadphasePairArray	m_overlappingPairArray;
+protected:
+	//avoid brute-force finding all the time
+	btBroadphasePairArray m_overlappingPairArray;
 
-		//during the dispatch, check that user doesn't destroy/create proxy
-		bool		m_blockedForChanges;
+	//during the dispatch, check that user doesn't destroy/create proxy
+	bool m_blockedForChanges;
 
-		///by default, do the removal during the pair traversal
-		bool		m_hasDeferredRemoval;
-		
-		//if set, use the callback instead of the built in filter in needBroadphaseCollision
-		btOverlapFilterCallback* m_overlapFilterCallback;
+	///by default, do the removal during the pair traversal
+	bool m_hasDeferredRemoval;
 
-		btOverlappingPairCallback*	m_ghostPairCallback;
+	//if set, use the callback instead of the built in filter in needBroadphaseCollision
+	btOverlapFilterCallback* m_overlapFilterCallback;
 
-	public:
-			
-		btSortedOverlappingPairCache();	
-		virtual ~btSortedOverlappingPairCache();
+	btOverlappingPairCallback* m_ghostPairCallback;
 
-		virtual void	processAllOverlappingPairs(btOverlapCallback*,btDispatcher* dispatcher);
+public:
+	btSortedOverlappingPairCache();
+	virtual ~btSortedOverlappingPairCache();
 
-		void*	removeOverlappingPair(btBroadphaseProxy* proxy0,btBroadphaseProxy* proxy1,btDispatcher* dispatcher);
+	virtual void processAllOverlappingPairs(btOverlapCallback*, btDispatcher* dispatcher);
 
-		void	cleanOverlappingPair(btBroadphasePair& pair,btDispatcher* dispatcher);
-		
-		btBroadphasePair*	addOverlappingPair(btBroadphaseProxy* proxy0,btBroadphaseProxy* proxy1);
+	void* removeOverlappingPair(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1, btDispatcher* dispatcher);
 
-		btBroadphasePair*	findPair(btBroadphaseProxy* proxy0,btBroadphaseProxy* proxy1);
-			
-		
-		void	cleanProxyFromPairs(btBroadphaseProxy* proxy,btDispatcher* dispatcher);
+	void cleanOverlappingPair(btBroadphasePair& pair, btDispatcher* dispatcher);
 
-		void	removeOverlappingPairsContainingProxy(btBroadphaseProxy* proxy,btDispatcher* dispatcher);
+	btBroadphasePair* addOverlappingPair(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1);
 
+	btBroadphasePair* findPair(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1);
 
-		inline bool needsBroadphaseCollision(btBroadphaseProxy* proxy0,btBroadphaseProxy* proxy1) const
-		{
-			if (m_overlapFilterCallback)
-				return m_overlapFilterCallback->needBroadphaseCollision(proxy0,proxy1);
+	void cleanProxyFromPairs(btBroadphaseProxy* proxy, btDispatcher* dispatcher);
 
-			bool collides = (proxy0->m_collisionFilterGroup & proxy1->m_collisionFilterMask) != 0;
-			collides = collides && (proxy1->m_collisionFilterGroup & proxy0->m_collisionFilterMask);
-			
-			return collides;
-		}
-		
-		btBroadphasePairArray&	getOverlappingPairArray()
-		{
-			return m_overlappingPairArray;
-		}
+	void removeOverlappingPairsContainingProxy(btBroadphaseProxy* proxy, btDispatcher* dispatcher);
 
-		const btBroadphasePairArray&	getOverlappingPairArray() const
-		{
-			return m_overlappingPairArray;
-		}
+	inline bool needsBroadphaseCollision(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1) const
+	{
+		if (m_overlapFilterCallback)
+			return m_overlapFilterCallback->needBroadphaseCollision(proxy0, proxy1);
 
-		
+		bool collides = (proxy0->m_collisionFilterGroup & proxy1->m_collisionFilterMask) != 0;
+		collides = collides && (proxy1->m_collisionFilterGroup & proxy0->m_collisionFilterMask);
 
+		return collides;
+	}
 
-		btBroadphasePair*	getOverlappingPairArrayPtr()
-		{
-			return &m_overlappingPairArray[0];
-		}
+	btBroadphasePairArray& getOverlappingPairArray()
+	{
+		return m_overlappingPairArray;
+	}
 
-		const btBroadphasePair*	getOverlappingPairArrayPtr() const
-		{
-			return &m_overlappingPairArray[0];
-		}
+	const btBroadphasePairArray& getOverlappingPairArray() const
+	{
+		return m_overlappingPairArray;
+	}
 
-		int	getNumOverlappingPairs() const
-		{
-			return m_overlappingPairArray.size();
-		}
-		
-		btOverlapFilterCallback* getOverlapFilterCallback()
-		{
-			return m_overlapFilterCallback;
-		}
+	btBroadphasePair* getOverlappingPairArrayPtr()
+	{
+		return &m_overlappingPairArray[0];
+	}
 
-		void setOverlapFilterCallback(btOverlapFilterCallback* callback)
-		{
-			m_overlapFilterCallback = callback;
-		}
+	const btBroadphasePair* getOverlappingPairArrayPtr() const
+	{
+		return &m_overlappingPairArray[0];
+	}
 
-		virtual bool	hasDeferredRemoval()
-		{
-			return m_hasDeferredRemoval;
-		}
+	int getNumOverlappingPairs() const
+	{
+		return m_overlappingPairArray.size();
+	}
 
-		virtual	void	setInternalGhostPairCallback(btOverlappingPairCallback* ghostPairCallback)
-		{
-			m_ghostPairCallback = ghostPairCallback;
-		}
+	btOverlapFilterCallback* getOverlapFilterCallback()
+	{
+		return m_overlapFilterCallback;
+	}
 
-		virtual void	sortOverlappingPairs(btDispatcher* dispatcher);
-		
+	void setOverlapFilterCallback(btOverlapFilterCallback* callback)
+	{
+		m_overlapFilterCallback = callback;
+	}
 
+	virtual bool hasDeferredRemoval()
+	{
+		return m_hasDeferredRemoval;
+	}
+
+	virtual void setInternalGhostPairCallback(btOverlappingPairCallback* ghostPairCallback)
+	{
+		m_ghostPairCallback = ghostPairCallback;
+	}
+
+	virtual void sortOverlappingPairs(btDispatcher* dispatcher);
 };
 
-
-
 ///btNullPairCache skips add/removal of overlapping pairs. Userful for benchmarking and unit testing.
 class btNullPairCache : public btOverlappingPairCache
 {
-
-	btBroadphasePairArray	m_overlappingPairArray;
+	btBroadphasePairArray m_overlappingPairArray;
 
 public:
-
-	virtual btBroadphasePair*	getOverlappingPairArrayPtr()
+	virtual btBroadphasePair* getOverlappingPairArrayPtr()
 	{
 		return &m_overlappingPairArray[0];
 	}
-	const btBroadphasePair*	getOverlappingPairArrayPtr() const
+	const btBroadphasePair* getOverlappingPairArrayPtr() const
 	{
 		return &m_overlappingPairArray[0];
 	}
-	btBroadphasePairArray&	getOverlappingPairArray()
+	btBroadphasePairArray& getOverlappingPairArray()
 	{
 		return m_overlappingPairArray;
 	}
-	
-	virtual	void	cleanOverlappingPair(btBroadphasePair& /*pair*/,btDispatcher* /*dispatcher*/)
-	{
 
+	virtual void cleanOverlappingPair(btBroadphasePair& /*pair*/, btDispatcher* /*dispatcher*/)
+	{
 	}
 
 	virtual int getNumOverlappingPairs() const
@@ -413,16 +376,15 @@ public:
 		return 0;
 	}
 
-	virtual void	cleanProxyFromPairs(btBroadphaseProxy* /*proxy*/,btDispatcher* /*dispatcher*/)
-	{
-
-	}
-
-	virtual	void setOverlapFilterCallback(btOverlapFilterCallback* /*callback*/)
+	virtual void cleanProxyFromPairs(btBroadphaseProxy* /*proxy*/, btDispatcher* /*dispatcher*/)
 	{
 	}
 
-	virtual void	processAllOverlappingPairs(btOverlapCallback*,btDispatcher* /*dispatcher*/)
+	virtual void setOverlapFilterCallback(btOverlapFilterCallback* /*callback*/)
+	{
+	}
+
+	virtual void processAllOverlappingPairs(btOverlapCallback*, btDispatcher* /*dispatcher*/)
 	{
 	}
 
@@ -431,39 +393,33 @@ public:
 		return 0;
 	}
 
-	virtual bool	hasDeferredRemoval()
+	virtual bool hasDeferredRemoval()
 	{
 		return true;
 	}
 
-	virtual	void	setInternalGhostPairCallback(btOverlappingPairCallback* /* ghostPairCallback */)
+	virtual void setInternalGhostPairCallback(btOverlappingPairCallback* /* ghostPairCallback */)
 	{
-
 	}
 
-	virtual btBroadphasePair*	addOverlappingPair(btBroadphaseProxy* /*proxy0*/,btBroadphaseProxy* /*proxy1*/)
+	virtual btBroadphasePair* addOverlappingPair(btBroadphaseProxy* /*proxy0*/, btBroadphaseProxy* /*proxy1*/)
 	{
 		return 0;
 	}
 
-	virtual void*	removeOverlappingPair(btBroadphaseProxy* /*proxy0*/,btBroadphaseProxy* /*proxy1*/,btDispatcher* /*dispatcher*/)
+	virtual void* removeOverlappingPair(btBroadphaseProxy* /*proxy0*/, btBroadphaseProxy* /*proxy1*/, btDispatcher* /*dispatcher*/)
 	{
 		return 0;
 	}
 
-	virtual void	removeOverlappingPairsContainingProxy(btBroadphaseProxy* /*proxy0*/,btDispatcher* /*dispatcher*/)
+	virtual void removeOverlappingPairsContainingProxy(btBroadphaseProxy* /*proxy0*/, btDispatcher* /*dispatcher*/)
 	{
 	}
-	
-	virtual void	sortOverlappingPairs(btDispatcher* dispatcher)
+
+	virtual void sortOverlappingPairs(btDispatcher* dispatcher)
 	{
-        (void) dispatcher;
+		(void)dispatcher;
 	}
-
-
 };
 
-
-#endif //BT_OVERLAPPING_PAIR_CACHE_H
-
-
+#endif  //BT_OVERLAPPING_PAIR_CACHE_H
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btOverlappingPairCallback.h b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btOverlappingPairCallback.h
index 3e069fa5e28..d16c72542f1 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btOverlappingPairCallback.h
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btOverlappingPairCallback.h
@@ -18,26 +18,24 @@ subject to the following restrictions:
 #define OVERLAPPING_PAIR_CALLBACK_H
 
 class btDispatcher;
-struct  btBroadphasePair;
+struct btBroadphasePair;
 
 ///The btOverlappingPairCallback class is an additional optional broadphase user callback for adding/removing overlapping pairs, similar interface to btOverlappingPairCache.
 class btOverlappingPairCallback
 {
 protected:
-    btOverlappingPairCallback() {}
-    
+	btOverlappingPairCallback() {}
+
 public:
 	virtual ~btOverlappingPairCallback()
 	{
-
 	}
-	
-	virtual btBroadphasePair*	addOverlappingPair(btBroadphaseProxy* proxy0,btBroadphaseProxy* proxy1) = 0;
 
-	virtual void*	removeOverlappingPair(btBroadphaseProxy* proxy0,btBroadphaseProxy* proxy1,btDispatcher* dispatcher) = 0;
+	virtual btBroadphasePair* addOverlappingPair(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1) = 0;
 
-	virtual void	removeOverlappingPairsContainingProxy(btBroadphaseProxy* proxy0,btDispatcher* dispatcher) = 0;
+	virtual void* removeOverlappingPair(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1, btDispatcher* dispatcher) = 0;
 
+	virtual void removeOverlappingPairsContainingProxy(btBroadphaseProxy* proxy0, btDispatcher* dispatcher) = 0;
 };
 
-#endif //OVERLAPPING_PAIR_CALLBACK_H
+#endif  //OVERLAPPING_PAIR_CALLBACK_H
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btQuantizedBvh.cpp b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btQuantizedBvh.cpp
index 875d89c53ec..b814fd84d85 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btQuantizedBvh.cpp
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btQuantizedBvh.cpp
@@ -21,43 +21,38 @@ subject to the following restrictions:
 
 #define RAYAABB2
 
-btQuantizedBvh::btQuantizedBvh() : 
-					m_bulletVersion(BT_BULLET_VERSION),
-					m_useQuantization(false), 
-					//m_traversalMode(TRAVERSAL_STACKLESS_CACHE_FRIENDLY)
-					m_traversalMode(TRAVERSAL_STACKLESS)
-					//m_traversalMode(TRAVERSAL_RECURSIVE)
-					,m_subtreeHeaderCount(0) //PCK: add this line
+btQuantizedBvh::btQuantizedBvh() : m_bulletVersion(BT_BULLET_VERSION),
+								   m_useQuantization(false),
+								   //m_traversalMode(TRAVERSAL_STACKLESS_CACHE_FRIENDLY)
+								   m_traversalMode(TRAVERSAL_STACKLESS)
+								   //m_traversalMode(TRAVERSAL_RECURSIVE)
+								   ,
+								   m_subtreeHeaderCount(0)  //PCK: add this line
 {
-	m_bvhAabbMin.setValue(-SIMD_INFINITY,-SIMD_INFINITY,-SIMD_INFINITY);
-	m_bvhAabbMax.setValue(SIMD_INFINITY,SIMD_INFINITY,SIMD_INFINITY);
+	m_bvhAabbMin.setValue(-SIMD_INFINITY, -SIMD_INFINITY, -SIMD_INFINITY);
+	m_bvhAabbMax.setValue(SIMD_INFINITY, SIMD_INFINITY, SIMD_INFINITY);
 }
 
-
-
-
-
 void btQuantizedBvh::buildInternal()
 {
 	///assumes that caller filled in the m_quantizedLeafNodes
 	m_useQuantization = true;
 	int numLeafNodes = 0;
-	
+
 	if (m_useQuantization)
 	{
 		//now we have an array of leafnodes in m_leafNodes
 		numLeafNodes = m_quantizedLeafNodes.size();
 
-		m_quantizedContiguousNodes.resize(2*numLeafNodes);
-
+		m_quantizedContiguousNodes.resize(2 * numLeafNodes);
 	}
 
 	m_curNodeIndex = 0;
 
-	buildTree(0,numLeafNodes);
+	buildTree(0, numLeafNodes);
 
 	///if the entire tree is small then subtree size, we need to create a header info for the tree
-	if(m_useQuantization && !m_SubtreeHeaders.size())
+	if (m_useQuantization && !m_SubtreeHeaders.size())
 	{
 		btBvhSubtreeInfo& subtree = m_SubtreeHeaders.expand();
 		subtree.setAabbFromQuantizeNode(m_quantizedContiguousNodes[0]);
@@ -73,29 +68,24 @@ void btQuantizedBvh::buildInternal()
 	m_leafNodes.clear();
 }
 
-
-
 ///just for debugging, to visualize the individual patches/subtrees
 #ifdef DEBUG_PATCH_COLORS
-btVector3 color[4]=
-{
-	btVector3(1,0,0),
-	btVector3(0,1,0),
-	btVector3(0,0,1),
-	btVector3(0,1,1)
-};
-#endif //DEBUG_PATCH_COLORS
+btVector3 color[4] =
+	{
+		btVector3(1, 0, 0),
+		btVector3(0, 1, 0),
+		btVector3(0, 0, 1),
+		btVector3(0, 1, 1)};
+#endif  //DEBUG_PATCH_COLORS
 
-
-
-void	btQuantizedBvh::setQuantizationValues(const btVector3& bvhAabbMin,const btVector3& bvhAabbMax,btScalar quantizationMargin)
+void btQuantizedBvh::setQuantizationValues(const btVector3& bvhAabbMin, const btVector3& bvhAabbMax, btScalar quantizationMargin)
 {
 	//enlarge the AABB to avoid division by zero when initializing the quantization values
-	btVector3 clampValue(quantizationMargin,quantizationMargin,quantizationMargin);
+	btVector3 clampValue(quantizationMargin, quantizationMargin, quantizationMargin);
 	m_bvhAabbMin = bvhAabbMin - clampValue;
 	m_bvhAabbMax = bvhAabbMax + clampValue;
 	btVector3 aabbSize = m_bvhAabbMax - m_bvhAabbMin;
-	m_bvhQuantization = btVector3(btScalar(65533.0),btScalar(65533.0),btScalar(65533.0)) / aabbSize;
+	m_bvhQuantization = btVector3(btScalar(65533.0), btScalar(65533.0), btScalar(65533.0)) / aabbSize;
 
 	m_useQuantization = true;
 
@@ -103,25 +93,22 @@ void	btQuantizedBvh::setQuantizationValues(const btVector3& bvhAabbMin,const btV
 		unsigned short vecIn[3];
 		btVector3 v;
 		{
-			quantize(vecIn,m_bvhAabbMin,false);
+			quantize(vecIn, m_bvhAabbMin, false);
 			v = unQuantize(vecIn);
-			m_bvhAabbMin.setMin(v-clampValue);
-		}
-        aabbSize = m_bvhAabbMax - m_bvhAabbMin;
-        m_bvhQuantization = btVector3(btScalar(65533.0),btScalar(65533.0),btScalar(65533.0)) / aabbSize;
-		{
-			quantize(vecIn,m_bvhAabbMax,true);
-			v = unQuantize(vecIn);
-			m_bvhAabbMax.setMax(v+clampValue);
+			m_bvhAabbMin.setMin(v - clampValue);
 		}
 		aabbSize = m_bvhAabbMax - m_bvhAabbMin;
-		m_bvhQuantization = btVector3(btScalar(65533.0),btScalar(65533.0),btScalar(65533.0)) / aabbSize;
+		m_bvhQuantization = btVector3(btScalar(65533.0), btScalar(65533.0), btScalar(65533.0)) / aabbSize;
+		{
+			quantize(vecIn, m_bvhAabbMax, true);
+			v = unQuantize(vecIn);
+			m_bvhAabbMax.setMax(v + clampValue);
+		}
+		aabbSize = m_bvhAabbMax - m_bvhAabbMin;
+		m_bvhQuantization = btVector3(btScalar(65533.0), btScalar(65533.0), btScalar(65533.0)) / aabbSize;
 	}
 }
 
-
-
-
 btQuantizedBvh::~btQuantizedBvh()
 {
 }
@@ -129,104 +116,100 @@ btQuantizedBvh::~btQuantizedBvh()
 #ifdef DEBUG_TREE_BUILDING
 int gStackDepth = 0;
 int gMaxStackDepth = 0;
-#endif //DEBUG_TREE_BUILDING
+#endif  //DEBUG_TREE_BUILDING
 
-void	btQuantizedBvh::buildTree	(int startIndex,int endIndex)
+void btQuantizedBvh::buildTree(int startIndex, int endIndex)
 {
 #ifdef DEBUG_TREE_BUILDING
 	gStackDepth++;
 	if (gStackDepth > gMaxStackDepth)
 		gMaxStackDepth = gStackDepth;
-#endif //DEBUG_TREE_BUILDING
-
+#endif  //DEBUG_TREE_BUILDING
 
 	int splitAxis, splitIndex, i;
-	int numIndices =endIndex-startIndex;
+	int numIndices = endIndex - startIndex;
 	int curIndex = m_curNodeIndex;
 
-	btAssert(numIndices>0);
+	btAssert(numIndices > 0);
 
-	if (numIndices==1)
+	if (numIndices == 1)
 	{
 #ifdef DEBUG_TREE_BUILDING
 		gStackDepth--;
-#endif //DEBUG_TREE_BUILDING
-		
-		assignInternalNodeFromLeafNode(m_curNodeIndex,startIndex);
+#endif  //DEBUG_TREE_BUILDING
+
+		assignInternalNodeFromLeafNode(m_curNodeIndex, startIndex);
 
 		m_curNodeIndex++;
-		return;	
+		return;
 	}
 	//calculate Best Splitting Axis and where to split it. Sort the incoming 'leafNodes' array within range 'startIndex/endIndex'.
-	
-	splitAxis = calcSplittingAxis(startIndex,endIndex);
 
-	splitIndex = sortAndCalcSplittingIndex(startIndex,endIndex,splitAxis);
+	splitAxis = calcSplittingAxis(startIndex, endIndex);
+
+	splitIndex = sortAndCalcSplittingIndex(startIndex, endIndex, splitAxis);
 
 	int internalNodeIndex = m_curNodeIndex;
-	
+
 	//set the min aabb to 'inf' or a max value, and set the max aabb to a -inf/minimum value.
 	//the aabb will be expanded during buildTree/mergeInternalNodeAabb with actual node values
-	setInternalNodeAabbMin(m_curNodeIndex,m_bvhAabbMax);//can't use btVector3(SIMD_INFINITY,SIMD_INFINITY,SIMD_INFINITY)) because of quantization
-	setInternalNodeAabbMax(m_curNodeIndex,m_bvhAabbMin);//can't use btVector3(-SIMD_INFINITY,-SIMD_INFINITY,-SIMD_INFINITY)) because of quantization
-	
-	
-	for (i=startIndex;i<endIndex;i++)
+	setInternalNodeAabbMin(m_curNodeIndex, m_bvhAabbMax);  //can't use btVector3(SIMD_INFINITY,SIMD_INFINITY,SIMD_INFINITY)) because of quantization
+	setInternalNodeAabbMax(m_curNodeIndex, m_bvhAabbMin);  //can't use btVector3(-SIMD_INFINITY,-SIMD_INFINITY,-SIMD_INFINITY)) because of quantization
+
+	for (i = startIndex; i < endIndex; i++)
 	{
-		mergeInternalNodeAabb(m_curNodeIndex,getAabbMin(i),getAabbMax(i));
+		mergeInternalNodeAabb(m_curNodeIndex, getAabbMin(i), getAabbMax(i));
 	}
 
 	m_curNodeIndex++;
-	
 
 	//internalNode->m_escapeIndex;
-	
+
 	int leftChildNodexIndex = m_curNodeIndex;
 
 	//build left child tree
-	buildTree(startIndex,splitIndex);
+	buildTree(startIndex, splitIndex);
 
 	int rightChildNodexIndex = m_curNodeIndex;
 	//build right child tree
-	buildTree(splitIndex,endIndex);
+	buildTree(splitIndex, endIndex);
 
 #ifdef DEBUG_TREE_BUILDING
 	gStackDepth--;
-#endif //DEBUG_TREE_BUILDING
+#endif  //DEBUG_TREE_BUILDING
 
 	int escapeIndex = m_curNodeIndex - curIndex;
 
 	if (m_useQuantization)
 	{
 		//escapeIndex is the number of nodes of this subtree
-		const int sizeQuantizedNode =sizeof(btQuantizedBvhNode);
+		const int sizeQuantizedNode = sizeof(btQuantizedBvhNode);
 		const int treeSizeInBytes = escapeIndex * sizeQuantizedNode;
 		if (treeSizeInBytes > MAX_SUBTREE_SIZE_IN_BYTES)
 		{
-			updateSubtreeHeaders(leftChildNodexIndex,rightChildNodexIndex);
+			updateSubtreeHeaders(leftChildNodexIndex, rightChildNodexIndex);
 		}
-	} else
+	}
+	else
 	{
-
 	}
 
-	setInternalNodeEscapeIndex(internalNodeIndex,escapeIndex);
-
+	setInternalNodeEscapeIndex(internalNodeIndex, escapeIndex);
 }
 
-void	btQuantizedBvh::updateSubtreeHeaders(int leftChildNodexIndex,int rightChildNodexIndex)
+void btQuantizedBvh::updateSubtreeHeaders(int leftChildNodexIndex, int rightChildNodexIndex)
 {
 	btAssert(m_useQuantization);
 
 	btQuantizedBvhNode& leftChildNode = m_quantizedContiguousNodes[leftChildNodexIndex];
 	int leftSubTreeSize = leftChildNode.isLeafNode() ? 1 : leftChildNode.getEscapeIndex();
-	int leftSubTreeSizeInBytes =  leftSubTreeSize * static_cast<int>(sizeof(btQuantizedBvhNode));
-	
+	int leftSubTreeSizeInBytes = leftSubTreeSize * static_cast<int>(sizeof(btQuantizedBvhNode));
+
 	btQuantizedBvhNode& rightChildNode = m_quantizedContiguousNodes[rightChildNodexIndex];
 	int rightSubTreeSize = rightChildNode.isLeafNode() ? 1 : rightChildNode.getEscapeIndex();
-	int rightSubTreeSizeInBytes =  rightSubTreeSize *  static_cast<int>(sizeof(btQuantizedBvhNode));
+	int rightSubTreeSizeInBytes = rightSubTreeSize * static_cast<int>(sizeof(btQuantizedBvhNode));
 
-	if(leftSubTreeSizeInBytes <= MAX_SUBTREE_SIZE_IN_BYTES)
+	if (leftSubTreeSizeInBytes <= MAX_SUBTREE_SIZE_IN_BYTES)
 	{
 		btBvhSubtreeInfo& subtree = m_SubtreeHeaders.expand();
 		subtree.setAabbFromQuantizeNode(leftChildNode);
@@ -234,7 +217,7 @@ void	btQuantizedBvh::updateSubtreeHeaders(int leftChildNodexIndex,int rightChild
 		subtree.m_subtreeSize = leftSubTreeSize;
 	}
 
-	if(rightSubTreeSizeInBytes <= MAX_SUBTREE_SIZE_IN_BYTES)
+	if (rightSubTreeSizeInBytes <= MAX_SUBTREE_SIZE_IN_BYTES)
 	{
 		btBvhSubtreeInfo& subtree = m_SubtreeHeaders.expand();
 		subtree.setAabbFromQuantizeNode(rightChildNode);
@@ -246,32 +229,31 @@ void	btQuantizedBvh::updateSubtreeHeaders(int leftChildNodexIndex,int rightChild
 	m_subtreeHeaderCount = m_SubtreeHeaders.size();
 }
 
-
-int	btQuantizedBvh::sortAndCalcSplittingIndex(int startIndex,int endIndex,int splitAxis)
+int btQuantizedBvh::sortAndCalcSplittingIndex(int startIndex, int endIndex, int splitAxis)
 {
 	int i;
-	int splitIndex =startIndex;
+	int splitIndex = startIndex;
 	int numIndices = endIndex - startIndex;
 	btScalar splitValue;
 
-	btVector3 means(btScalar(0.),btScalar(0.),btScalar(0.));
-	for (i=startIndex;i<endIndex;i++)
+	btVector3 means(btScalar(0.), btScalar(0.), btScalar(0.));
+	for (i = startIndex; i < endIndex; i++)
 	{
-		btVector3 center = btScalar(0.5)*(getAabbMax(i)+getAabbMin(i));
-		means+=center;
+		btVector3 center = btScalar(0.5) * (getAabbMax(i) + getAabbMin(i));
+		means += center;
 	}
-	means *= (btScalar(1.)/(btScalar)numIndices);
-	
+	means *= (btScalar(1.) / (btScalar)numIndices);
+
 	splitValue = means[splitAxis];
-	
+
 	//sort leafNodes so all values larger then splitValue comes first, and smaller values start from 'splitIndex'.
-	for (i=startIndex;i<endIndex;i++)
+	for (i = startIndex; i < endIndex; i++)
 	{
-		btVector3 center = btScalar(0.5)*(getAabbMax(i)+getAabbMin(i));
+		btVector3 center = btScalar(0.5) * (getAabbMax(i) + getAabbMin(i));
 		if (center[splitAxis] > splitValue)
 		{
 			//swap
-			swapLeafNodes(i,splitIndex);
+			swapLeafNodes(i, splitIndex);
 			splitIndex++;
 		}
 	}
@@ -281,56 +263,53 @@ int	btQuantizedBvh::sortAndCalcSplittingIndex(int startIndex,int endIndex,int sp
 	//unbalanced1 is unsafe: it can cause stack overflows
 	//bool unbalanced1 = ((splitIndex==startIndex) || (splitIndex == (endIndex-1)));
 
-	//unbalanced2 should work too: always use center (perfect balanced trees)	
+	//unbalanced2 should work too: always use center (perfect balanced trees)
 	//bool unbalanced2 = true;
 
 	//this should be safe too:
-	int rangeBalancedIndices = numIndices/3;
-	bool unbalanced = ((splitIndex<=(startIndex+rangeBalancedIndices)) || (splitIndex >=(endIndex-1-rangeBalancedIndices)));
-	
+	int rangeBalancedIndices = numIndices / 3;
+	bool unbalanced = ((splitIndex <= (startIndex + rangeBalancedIndices)) || (splitIndex >= (endIndex - 1 - rangeBalancedIndices)));
+
 	if (unbalanced)
 	{
-		splitIndex = startIndex+ (numIndices>>1);
+		splitIndex = startIndex + (numIndices >> 1);
 	}
 
-	bool unbal = (splitIndex==startIndex) || (splitIndex == (endIndex));
+	bool unbal = (splitIndex == startIndex) || (splitIndex == (endIndex));
 	(void)unbal;
 	btAssert(!unbal);
 
 	return splitIndex;
 }
 
-
-int	btQuantizedBvh::calcSplittingAxis(int startIndex,int endIndex)
+int btQuantizedBvh::calcSplittingAxis(int startIndex, int endIndex)
 {
 	int i;
 
-	btVector3 means(btScalar(0.),btScalar(0.),btScalar(0.));
-	btVector3 variance(btScalar(0.),btScalar(0.),btScalar(0.));
-	int numIndices = endIndex-startIndex;
+	btVector3 means(btScalar(0.), btScalar(0.), btScalar(0.));
+	btVector3 variance(btScalar(0.), btScalar(0.), btScalar(0.));
+	int numIndices = endIndex - startIndex;
 
-	for (i=startIndex;i<endIndex;i++)
+	for (i = startIndex; i < endIndex; i++)
 	{
-		btVector3 center = btScalar(0.5)*(getAabbMax(i)+getAabbMin(i));
-		means+=center;
+		btVector3 center = btScalar(0.5) * (getAabbMax(i) + getAabbMin(i));
+		means += center;
 	}
-	means *= (btScalar(1.)/(btScalar)numIndices);
-		
-	for (i=startIndex;i<endIndex;i++)
+	means *= (btScalar(1.) / (btScalar)numIndices);
+
+	for (i = startIndex; i < endIndex; i++)
 	{
-		btVector3 center = btScalar(0.5)*(getAabbMax(i)+getAabbMin(i));
-		btVector3 diff2 = center-means;
+		btVector3 center = btScalar(0.5) * (getAabbMax(i) + getAabbMin(i));
+		btVector3 diff2 = center - means;
 		diff2 = diff2 * diff2;
 		variance += diff2;
 	}
-	variance *= (btScalar(1.)/	((btScalar)numIndices-1)	);
-	
+	variance *= (btScalar(1.) / ((btScalar)numIndices - 1));
+
 	return variance.maxAxis();
 }
 
-
-
-void	btQuantizedBvh::reportAabbOverlappingNodex(btNodeOverlapCallback* nodeCallback,const btVector3& aabbMin,const btVector3& aabbMax) const
+void btQuantizedBvh::reportAabbOverlappingNodex(btNodeOverlapCallback* nodeCallback, const btVector3& aabbMin, const btVector3& aabbMax) const
 {
 	//either choose recursive traversal (walkTree) or stackless (walkStacklessTree)
 
@@ -339,38 +318,37 @@ void	btQuantizedBvh::reportAabbOverlappingNodex(btNodeOverlapCallback* nodeCallb
 		///quantize query AABB
 		unsigned short int quantizedQueryAabbMin[3];
 		unsigned short int quantizedQueryAabbMax[3];
-		quantizeWithClamp(quantizedQueryAabbMin,aabbMin,0);
-		quantizeWithClamp(quantizedQueryAabbMax,aabbMax,1);
+		quantizeWithClamp(quantizedQueryAabbMin, aabbMin, 0);
+		quantizeWithClamp(quantizedQueryAabbMax, aabbMax, 1);
 
 		switch (m_traversalMode)
 		{
-		case TRAVERSAL_STACKLESS:
-				walkStacklessQuantizedTree(nodeCallback,quantizedQueryAabbMin,quantizedQueryAabbMax,0,m_curNodeIndex);
-			break;
-		case TRAVERSAL_STACKLESS_CACHE_FRIENDLY:
-				walkStacklessQuantizedTreeCacheFriendly(nodeCallback,quantizedQueryAabbMin,quantizedQueryAabbMax);
-			break;
-		case TRAVERSAL_RECURSIVE:
+			case TRAVERSAL_STACKLESS:
+				walkStacklessQuantizedTree(nodeCallback, quantizedQueryAabbMin, quantizedQueryAabbMax, 0, m_curNodeIndex);
+				break;
+			case TRAVERSAL_STACKLESS_CACHE_FRIENDLY:
+				walkStacklessQuantizedTreeCacheFriendly(nodeCallback, quantizedQueryAabbMin, quantizedQueryAabbMax);
+				break;
+			case TRAVERSAL_RECURSIVE:
 			{
 				const btQuantizedBvhNode* rootNode = &m_quantizedContiguousNodes[0];
-				walkRecursiveQuantizedTreeAgainstQueryAabb(rootNode,nodeCallback,quantizedQueryAabbMin,quantizedQueryAabbMax);
+				walkRecursiveQuantizedTreeAgainstQueryAabb(rootNode, nodeCallback, quantizedQueryAabbMin, quantizedQueryAabbMax);
 			}
 			break;
-		default:
-			//unsupported
-			btAssert(0);
+			default:
+				//unsupported
+				btAssert(0);
 		}
-	} else
+	}
+	else
 	{
-		walkStacklessTree(nodeCallback,aabbMin,aabbMax);
+		walkStacklessTree(nodeCallback, aabbMin, aabbMax);
 	}
 }
 
-
 int maxIterations = 0;
 
-
-void	btQuantizedBvh::walkStacklessTree(btNodeOverlapCallback* nodeCallback,const btVector3& aabbMin,const btVector3& aabbMax) const
+void btQuantizedBvh::walkStacklessTree(btNodeOverlapCallback* nodeCallback, const btVector3& aabbMin, const btVector3& aabbMax) const
 {
 	btAssert(!m_useQuantization);
 
@@ -384,24 +362,25 @@ void	btQuantizedBvh::walkStacklessTree(btNodeOverlapCallback* nodeCallback,const
 	while (curIndex < m_curNodeIndex)
 	{
 		//catch bugs in tree data
-		btAssert (walkIterations < m_curNodeIndex);
+		btAssert(walkIterations < m_curNodeIndex);
 
 		walkIterations++;
-		aabbOverlap = TestAabbAgainstAabb2(aabbMin,aabbMax,rootNode->m_aabbMinOrg,rootNode->m_aabbMaxOrg);
+		aabbOverlap = TestAabbAgainstAabb2(aabbMin, aabbMax, rootNode->m_aabbMinOrg, rootNode->m_aabbMaxOrg);
 		isLeafNode = rootNode->m_escapeIndex == -1;
-		
+
 		//PCK: unsigned instead of bool
 		if (isLeafNode && (aabbOverlap != 0))
 		{
-			nodeCallback->processNode(rootNode->m_subPart,rootNode->m_triangleIndex);
-		} 
-		
+			nodeCallback->processNode(rootNode->m_subPart, rootNode->m_triangleIndex);
+		}
+
 		//PCK: unsigned instead of bool
 		if ((aabbOverlap != 0) || isLeafNode)
 		{
 			rootNode++;
 			curIndex++;
-		} else
+		}
+		else
 		{
 			escapeIndex = rootNode->m_escapeIndex;
 			rootNode += escapeIndex;
@@ -410,7 +389,6 @@ void	btQuantizedBvh::walkStacklessTree(btNodeOverlapCallback* nodeCallback,const
 	}
 	if (maxIterations < walkIterations)
 		maxIterations = walkIterations;
-
 }
 
 /*
@@ -434,39 +412,38 @@ void	btQuantizedBvh::walkTree(btOptimizedBvhNode* rootNode,btNodeOverlapCallback
 }
 */
 
-void btQuantizedBvh::walkRecursiveQuantizedTreeAgainstQueryAabb(const btQuantizedBvhNode* currentNode,btNodeOverlapCallback* nodeCallback,unsigned short int* quantizedQueryAabbMin,unsigned short int* quantizedQueryAabbMax) const
+void btQuantizedBvh::walkRecursiveQuantizedTreeAgainstQueryAabb(const btQuantizedBvhNode* currentNode, btNodeOverlapCallback* nodeCallback, unsigned short int* quantizedQueryAabbMin, unsigned short int* quantizedQueryAabbMax) const
 {
 	btAssert(m_useQuantization);
-	
+
 	bool isLeafNode;
 	//PCK: unsigned instead of bool
 	unsigned aabbOverlap;
 
 	//PCK: unsigned instead of bool
-	aabbOverlap = testQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin,quantizedQueryAabbMax,currentNode->m_quantizedAabbMin,currentNode->m_quantizedAabbMax);
+	aabbOverlap = testQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin, quantizedQueryAabbMax, currentNode->m_quantizedAabbMin, currentNode->m_quantizedAabbMax);
 	isLeafNode = currentNode->isLeafNode();
-		
+
 	//PCK: unsigned instead of bool
 	if (aabbOverlap != 0)
 	{
 		if (isLeafNode)
 		{
-			nodeCallback->processNode(currentNode->getPartId(),currentNode->getTriangleIndex());
-		} else
+			nodeCallback->processNode(currentNode->getPartId(), currentNode->getTriangleIndex());
+		}
+		else
 		{
 			//process left and right children
-			const btQuantizedBvhNode* leftChildNode = currentNode+1;
-			walkRecursiveQuantizedTreeAgainstQueryAabb(leftChildNode,nodeCallback,quantizedQueryAabbMin,quantizedQueryAabbMax);
+			const btQuantizedBvhNode* leftChildNode = currentNode + 1;
+			walkRecursiveQuantizedTreeAgainstQueryAabb(leftChildNode, nodeCallback, quantizedQueryAabbMin, quantizedQueryAabbMax);
 
-			const btQuantizedBvhNode* rightChildNode = leftChildNode->isLeafNode() ? leftChildNode+1:leftChildNode+leftChildNode->getEscapeIndex();
-			walkRecursiveQuantizedTreeAgainstQueryAabb(rightChildNode,nodeCallback,quantizedQueryAabbMin,quantizedQueryAabbMax);
+			const btQuantizedBvhNode* rightChildNode = leftChildNode->isLeafNode() ? leftChildNode + 1 : leftChildNode + leftChildNode->getEscapeIndex();
+			walkRecursiveQuantizedTreeAgainstQueryAabb(rightChildNode, nodeCallback, quantizedQueryAabbMin, quantizedQueryAabbMax);
 		}
-	}		
+	}
 }
 
-
-
-void	btQuantizedBvh::walkStacklessTreeAgainstRay(btNodeOverlapCallback* nodeCallback, const btVector3& raySource, const btVector3& rayTarget, const btVector3& aabbMin, const btVector3& aabbMax, int startNodeIndex,int endNodeIndex) const
+void btQuantizedBvh::walkStacklessTreeAgainstRay(btNodeOverlapCallback* nodeCallback, const btVector3& raySource, const btVector3& rayTarget, const btVector3& aabbMin, const btVector3& aabbMax, int startNodeIndex, int endNodeIndex) const
 {
 	btAssert(!m_useQuantization);
 
@@ -475,11 +452,11 @@ void	btQuantizedBvh::walkStacklessTreeAgainstRay(btNodeOverlapCallback* nodeCall
 	int walkIterations = 0;
 	bool isLeafNode;
 	//PCK: unsigned instead of bool
-	unsigned aabbOverlap=0;
-	unsigned rayBoxOverlap=0;
+	unsigned aabbOverlap = 0;
+	unsigned rayBoxOverlap = 0;
 	btScalar lambda_max = 1.0;
-	
-		/* Quick pruning by quantized box */
+
+	/* Quick pruning by quantized box */
 	btVector3 rayAabbMin = raySource;
 	btVector3 rayAabbMax = raySource;
 	rayAabbMin.setMin(rayTarget);
@@ -490,15 +467,15 @@ void	btQuantizedBvh::walkStacklessTreeAgainstRay(btNodeOverlapCallback* nodeCall
 	rayAabbMax += aabbMax;
 
 #ifdef RAYAABB2
-	btVector3 rayDir = (rayTarget-raySource);
-	rayDir.normalize ();
-	lambda_max = rayDir.dot(rayTarget-raySource);
+	btVector3 rayDir = (rayTarget - raySource);
+	rayDir.normalize();
+	lambda_max = rayDir.dot(rayTarget - raySource);
 	///what about division by zero? --> just set rayDirection[i] to 1.0
 	btVector3 rayDirectionInverse;
 	rayDirectionInverse[0] = rayDir[0] == btScalar(0.0) ? btScalar(BT_LARGE_FLOAT) : btScalar(1.0) / rayDir[0];
 	rayDirectionInverse[1] = rayDir[1] == btScalar(0.0) ? btScalar(BT_LARGE_FLOAT) : btScalar(1.0) / rayDir[1];
 	rayDirectionInverse[2] = rayDir[2] == btScalar(0.0) ? btScalar(BT_LARGE_FLOAT) : btScalar(1.0) / rayDir[2];
-	unsigned int sign[3] = { rayDirectionInverse[0] < 0.0, rayDirectionInverse[1] < 0.0, rayDirectionInverse[2] < 0.0};
+	unsigned int sign[3] = {rayDirectionInverse[0] < 0.0, rayDirectionInverse[1] < 0.0, rayDirectionInverse[2] < 0.0};
 #endif
 
 	btVector3 bounds[2];
@@ -507,7 +484,7 @@ void	btQuantizedBvh::walkStacklessTreeAgainstRay(btNodeOverlapCallback* nodeCall
 	{
 		btScalar param = 1.0;
 		//catch bugs in tree data
-		btAssert (walkIterations < m_curNodeIndex);
+		btAssert(walkIterations < m_curNodeIndex);
 
 		walkIterations++;
 
@@ -517,34 +494,35 @@ void	btQuantizedBvh::walkStacklessTreeAgainstRay(btNodeOverlapCallback* nodeCall
 		bounds[0] -= aabbMax;
 		bounds[1] -= aabbMin;
 
-		aabbOverlap = TestAabbAgainstAabb2(rayAabbMin,rayAabbMax,rootNode->m_aabbMinOrg,rootNode->m_aabbMaxOrg);
+		aabbOverlap = TestAabbAgainstAabb2(rayAabbMin, rayAabbMax, rootNode->m_aabbMinOrg, rootNode->m_aabbMaxOrg);
 		//perhaps profile if it is worth doing the aabbOverlap test first
 
 #ifdef RAYAABB2
-			///careful with this check: need to check division by zero (above) and fix the unQuantize method
-			///thanks Joerg/hiker for the reproduction case!
-			///http://www.bulletphysics.com/Bullet/phpBB3/viewtopic.php?f=9&t=1858
-		rayBoxOverlap = aabbOverlap ? btRayAabb2 (raySource, rayDirectionInverse, sign, bounds, param, 0.0f, lambda_max) : false;
+		///careful with this check: need to check division by zero (above) and fix the unQuantize method
+		///thanks Joerg/hiker for the reproduction case!
+		///http://www.bulletphysics.com/Bullet/phpBB3/viewtopic.php?f=9&t=1858
+		rayBoxOverlap = aabbOverlap ? btRayAabb2(raySource, rayDirectionInverse, sign, bounds, param, 0.0f, lambda_max) : false;
 
 #else
 		btVector3 normal;
-		rayBoxOverlap = btRayAabb(raySource, rayTarget,bounds[0],bounds[1],param, normal);
+		rayBoxOverlap = btRayAabb(raySource, rayTarget, bounds[0], bounds[1], param, normal);
 #endif
 
 		isLeafNode = rootNode->m_escapeIndex == -1;
-		
+
 		//PCK: unsigned instead of bool
 		if (isLeafNode && (rayBoxOverlap != 0))
 		{
-			nodeCallback->processNode(rootNode->m_subPart,rootNode->m_triangleIndex);
-		} 
-		
+			nodeCallback->processNode(rootNode->m_subPart, rootNode->m_triangleIndex);
+		}
+
 		//PCK: unsigned instead of bool
 		if ((rayBoxOverlap != 0) || isLeafNode)
 		{
 			rootNode++;
 			curIndex++;
-		} else
+		}
+		else
 		{
 			escapeIndex = rootNode->m_escapeIndex;
 			rootNode += escapeIndex;
@@ -553,15 +531,12 @@ void	btQuantizedBvh::walkStacklessTreeAgainstRay(btNodeOverlapCallback* nodeCall
 	}
 	if (maxIterations < walkIterations)
 		maxIterations = walkIterations;
-
 }
 
-
-
-void	btQuantizedBvh::walkStacklessQuantizedTreeAgainstRay(btNodeOverlapCallback* nodeCallback, const btVector3& raySource, const btVector3& rayTarget, const btVector3& aabbMin, const btVector3& aabbMax, int startNodeIndex,int endNodeIndex) const
+void btQuantizedBvh::walkStacklessQuantizedTreeAgainstRay(btNodeOverlapCallback* nodeCallback, const btVector3& raySource, const btVector3& rayTarget, const btVector3& aabbMin, const btVector3& aabbMax, int startNodeIndex, int endNodeIndex) const
 {
 	btAssert(m_useQuantization);
-	
+
 	int curIndex = startNodeIndex;
 	int walkIterations = 0;
 	int subTreeSize = endNodeIndex - startNodeIndex;
@@ -569,7 +544,7 @@ void	btQuantizedBvh::walkStacklessQuantizedTreeAgainstRay(btNodeOverlapCallback*
 
 	const btQuantizedBvhNode* rootNode = &m_quantizedContiguousNodes[startNodeIndex];
 	int escapeIndex;
-	
+
 	bool isLeafNode;
 	//PCK: unsigned instead of bool
 	unsigned boxBoxOverlap = 0;
@@ -578,14 +553,14 @@ void	btQuantizedBvh::walkStacklessQuantizedTreeAgainstRay(btNodeOverlapCallback*
 	btScalar lambda_max = 1.0;
 
 #ifdef RAYAABB2
-	btVector3 rayDirection = (rayTarget-raySource);
-	rayDirection.normalize ();
-	lambda_max = rayDirection.dot(rayTarget-raySource);
+	btVector3 rayDirection = (rayTarget - raySource);
+	rayDirection.normalize();
+	lambda_max = rayDirection.dot(rayTarget - raySource);
 	///what about division by zero? --> just set rayDirection[i] to 1.0
 	rayDirection[0] = rayDirection[0] == btScalar(0.0) ? btScalar(BT_LARGE_FLOAT) : btScalar(1.0) / rayDirection[0];
 	rayDirection[1] = rayDirection[1] == btScalar(0.0) ? btScalar(BT_LARGE_FLOAT) : btScalar(1.0) / rayDirection[1];
 	rayDirection[2] = rayDirection[2] == btScalar(0.0) ? btScalar(BT_LARGE_FLOAT) : btScalar(1.0) / rayDirection[2];
-	unsigned int sign[3] = { rayDirection[0] < 0.0, rayDirection[1] < 0.0, rayDirection[2] < 0.0};
+	unsigned int sign[3] = {rayDirection[0] < 0.0, rayDirection[1] < 0.0, rayDirection[2] < 0.0};
 #endif
 
 	/* Quick pruning by quantized box */
@@ -600,37 +575,36 @@ void	btQuantizedBvh::walkStacklessQuantizedTreeAgainstRay(btNodeOverlapCallback*
 
 	unsigned short int quantizedQueryAabbMin[3];
 	unsigned short int quantizedQueryAabbMax[3];
-	quantizeWithClamp(quantizedQueryAabbMin,rayAabbMin,0);
-	quantizeWithClamp(quantizedQueryAabbMax,rayAabbMax,1);
+	quantizeWithClamp(quantizedQueryAabbMin, rayAabbMin, 0);
+	quantizeWithClamp(quantizedQueryAabbMax, rayAabbMax, 1);
 
 	while (curIndex < endNodeIndex)
 	{
-
 //#define VISUALLY_ANALYZE_BVH 1
 #ifdef VISUALLY_ANALYZE_BVH
 		//some code snippet to debugDraw aabb, to visually analyze bvh structure
 		static int drawPatch = 0;
 		//need some global access to a debugDrawer
 		extern btIDebugDraw* debugDrawerPtr;
-		if (curIndex==drawPatch)
+		if (curIndex == drawPatch)
 		{
-			btVector3 aabbMin,aabbMax;
+			btVector3 aabbMin, aabbMax;
 			aabbMin = unQuantize(rootNode->m_quantizedAabbMin);
 			aabbMax = unQuantize(rootNode->m_quantizedAabbMax);
-			btVector3	color(1,0,0);
-			debugDrawerPtr->drawAabb(aabbMin,aabbMax,color);
+			btVector3 color(1, 0, 0);
+			debugDrawerPtr->drawAabb(aabbMin, aabbMax, color);
 		}
-#endif//VISUALLY_ANALYZE_BVH
+#endif  //VISUALLY_ANALYZE_BVH
 
 		//catch bugs in tree data
-		btAssert (walkIterations < subTreeSize);
+		btAssert(walkIterations < subTreeSize);
 
 		walkIterations++;
 		//PCK: unsigned instead of bool
 		// only interested if this is closer than any previous hit
 		btScalar param = 1.0;
 		rayBoxOverlap = 0;
-		boxBoxOverlap = testQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin,quantizedQueryAabbMax,rootNode->m_quantizedAabbMin,rootNode->m_quantizedAabbMax);
+		boxBoxOverlap = testQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin, quantizedQueryAabbMax, rootNode->m_quantizedAabbMin, rootNode->m_quantizedAabbMax);
 		isLeafNode = rootNode->isLeafNode();
 		if (boxBoxOverlap)
 		{
@@ -655,24 +629,25 @@ void	btQuantizedBvh::walkStacklessQuantizedTreeAgainstRay(btNodeOverlapCallback*
 			///http://www.bulletphysics.com/Bullet/phpBB3/viewtopic.php?f=9&t=1858
 
 			//BT_PROFILE("btRayAabb2");
-			rayBoxOverlap = btRayAabb2 (raySource, rayDirection, sign, bounds, param, 0.0f, lambda_max);
-			
+			rayBoxOverlap = btRayAabb2(raySource, rayDirection, sign, bounds, param, 0.0f, lambda_max);
+
 #else
-			rayBoxOverlap = true;//btRayAabb(raySource, rayTarget, bounds[0], bounds[1], param, normal);
+			rayBoxOverlap = true;  //btRayAabb(raySource, rayTarget, bounds[0], bounds[1], param, normal);
 #endif
 		}
-		
+
 		if (isLeafNode && rayBoxOverlap)
 		{
-			nodeCallback->processNode(rootNode->getPartId(),rootNode->getTriangleIndex());
+			nodeCallback->processNode(rootNode->getPartId(), rootNode->getTriangleIndex());
 		}
-		
+
 		//PCK: unsigned instead of bool
 		if ((rayBoxOverlap != 0) || isLeafNode)
 		{
 			rootNode++;
 			curIndex++;
-		} else
+		}
+		else
 		{
 			escapeIndex = rootNode->getEscapeIndex();
 			rootNode += escapeIndex;
@@ -681,13 +656,12 @@ void	btQuantizedBvh::walkStacklessQuantizedTreeAgainstRay(btNodeOverlapCallback*
 	}
 	if (maxIterations < walkIterations)
 		maxIterations = walkIterations;
-
 }
 
-void	btQuantizedBvh::walkStacklessQuantizedTree(btNodeOverlapCallback* nodeCallback,unsigned short int* quantizedQueryAabbMin,unsigned short int* quantizedQueryAabbMax,int startNodeIndex,int endNodeIndex) const
+void btQuantizedBvh::walkStacklessQuantizedTree(btNodeOverlapCallback* nodeCallback, unsigned short int* quantizedQueryAabbMin, unsigned short int* quantizedQueryAabbMax, int startNodeIndex, int endNodeIndex) const
 {
 	btAssert(m_useQuantization);
-	
+
 	int curIndex = startNodeIndex;
 	int walkIterations = 0;
 	int subTreeSize = endNodeIndex - startNodeIndex;
@@ -695,49 +669,49 @@ void	btQuantizedBvh::walkStacklessQuantizedTree(btNodeOverlapCallback* nodeCallb
 
 	const btQuantizedBvhNode* rootNode = &m_quantizedContiguousNodes[startNodeIndex];
 	int escapeIndex;
-	
+
 	bool isLeafNode;
 	//PCK: unsigned instead of bool
 	unsigned aabbOverlap;
 
 	while (curIndex < endNodeIndex)
 	{
-
 //#define VISUALLY_ANALYZE_BVH 1
 #ifdef VISUALLY_ANALYZE_BVH
 		//some code snippet to debugDraw aabb, to visually analyze bvh structure
 		static int drawPatch = 0;
 		//need some global access to a debugDrawer
 		extern btIDebugDraw* debugDrawerPtr;
-		if (curIndex==drawPatch)
+		if (curIndex == drawPatch)
 		{
-			btVector3 aabbMin,aabbMax;
+			btVector3 aabbMin, aabbMax;
 			aabbMin = unQuantize(rootNode->m_quantizedAabbMin);
 			aabbMax = unQuantize(rootNode->m_quantizedAabbMax);
-			btVector3	color(1,0,0);
-			debugDrawerPtr->drawAabb(aabbMin,aabbMax,color);
+			btVector3 color(1, 0, 0);
+			debugDrawerPtr->drawAabb(aabbMin, aabbMax, color);
 		}
-#endif//VISUALLY_ANALYZE_BVH
+#endif  //VISUALLY_ANALYZE_BVH
 
 		//catch bugs in tree data
-		btAssert (walkIterations < subTreeSize);
+		btAssert(walkIterations < subTreeSize);
 
 		walkIterations++;
 		//PCK: unsigned instead of bool
-		aabbOverlap = testQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin,quantizedQueryAabbMax,rootNode->m_quantizedAabbMin,rootNode->m_quantizedAabbMax);
+		aabbOverlap = testQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin, quantizedQueryAabbMax, rootNode->m_quantizedAabbMin, rootNode->m_quantizedAabbMax);
 		isLeafNode = rootNode->isLeafNode();
-		
+
 		if (isLeafNode && aabbOverlap)
 		{
-			nodeCallback->processNode(rootNode->getPartId(),rootNode->getTriangleIndex());
-		} 
-		
+			nodeCallback->processNode(rootNode->getPartId(), rootNode->getTriangleIndex());
+		}
+
 		//PCK: unsigned instead of bool
 		if ((aabbOverlap != 0) || isLeafNode)
 		{
 			rootNode++;
 			curIndex++;
-		} else
+		}
+		else
 		{
 			escapeIndex = rootNode->getEscapeIndex();
 			rootNode += escapeIndex;
@@ -746,40 +720,36 @@ void	btQuantizedBvh::walkStacklessQuantizedTree(btNodeOverlapCallback* nodeCallb
 	}
 	if (maxIterations < walkIterations)
 		maxIterations = walkIterations;
-
 }
 
 //This traversal can be called from Playstation 3 SPU
-void	btQuantizedBvh::walkStacklessQuantizedTreeCacheFriendly(btNodeOverlapCallback* nodeCallback,unsigned short int* quantizedQueryAabbMin,unsigned short int* quantizedQueryAabbMax) const
+void btQuantizedBvh::walkStacklessQuantizedTreeCacheFriendly(btNodeOverlapCallback* nodeCallback, unsigned short int* quantizedQueryAabbMin, unsigned short int* quantizedQueryAabbMax) const
 {
 	btAssert(m_useQuantization);
 
 	int i;
 
-
-	for (i=0;i<this->m_SubtreeHeaders.size();i++)
+	for (i = 0; i < this->m_SubtreeHeaders.size(); i++)
 	{
 		const btBvhSubtreeInfo& subtree = m_SubtreeHeaders[i];
 
 		//PCK: unsigned instead of bool
-		unsigned overlap = testQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin,quantizedQueryAabbMax,subtree.m_quantizedAabbMin,subtree.m_quantizedAabbMax);
+		unsigned overlap = testQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin, quantizedQueryAabbMax, subtree.m_quantizedAabbMin, subtree.m_quantizedAabbMax);
 		if (overlap != 0)
 		{
-			walkStacklessQuantizedTree(nodeCallback,quantizedQueryAabbMin,quantizedQueryAabbMax,
-				subtree.m_rootNodeIndex,
-				subtree.m_rootNodeIndex+subtree.m_subtreeSize);
+			walkStacklessQuantizedTree(nodeCallback, quantizedQueryAabbMin, quantizedQueryAabbMax,
+									   subtree.m_rootNodeIndex,
+									   subtree.m_rootNodeIndex + subtree.m_subtreeSize);
 		}
 	}
 }
 
-
-void	btQuantizedBvh::reportRayOverlappingNodex (btNodeOverlapCallback* nodeCallback, const btVector3& raySource, const btVector3& rayTarget) const
+void btQuantizedBvh::reportRayOverlappingNodex(btNodeOverlapCallback* nodeCallback, const btVector3& raySource, const btVector3& rayTarget) const
 {
-	reportBoxCastOverlappingNodex(nodeCallback,raySource,rayTarget,btVector3(0,0,0),btVector3(0,0,0));
+	reportBoxCastOverlappingNodex(nodeCallback, raySource, rayTarget, btVector3(0, 0, 0), btVector3(0, 0, 0));
 }
 
-
-void	btQuantizedBvh::reportBoxCastOverlappingNodex(btNodeOverlapCallback* nodeCallback, const btVector3& raySource, const btVector3& rayTarget, const btVector3& aabbMin,const btVector3& aabbMax) const
+void btQuantizedBvh::reportBoxCastOverlappingNodex(btNodeOverlapCallback* nodeCallback, const btVector3& raySource, const btVector3& rayTarget, const btVector3& aabbMin, const btVector3& aabbMax) const
 {
 	//always use stackless
 
@@ -803,31 +773,31 @@ void	btQuantizedBvh::reportBoxCastOverlappingNodex(btNodeOverlapCallback* nodeCa
 		reportAabbOverlappingNodex(nodeCallback,qaabbMin,qaabbMax);
 	}
 	*/
-
 }
 
-
-void	btQuantizedBvh::swapLeafNodes(int i,int splitIndex)
+void btQuantizedBvh::swapLeafNodes(int i, int splitIndex)
 {
 	if (m_useQuantization)
 	{
-			btQuantizedBvhNode tmp = m_quantizedLeafNodes[i];
-			m_quantizedLeafNodes[i] = m_quantizedLeafNodes[splitIndex];
-			m_quantizedLeafNodes[splitIndex] = tmp;
-	} else
+		btQuantizedBvhNode tmp = m_quantizedLeafNodes[i];
+		m_quantizedLeafNodes[i] = m_quantizedLeafNodes[splitIndex];
+		m_quantizedLeafNodes[splitIndex] = tmp;
+	}
+	else
 	{
-			btOptimizedBvhNode tmp = m_leafNodes[i];
-			m_leafNodes[i] = m_leafNodes[splitIndex];
-			m_leafNodes[splitIndex] = tmp;
+		btOptimizedBvhNode tmp = m_leafNodes[i];
+		m_leafNodes[i] = m_leafNodes[splitIndex];
+		m_leafNodes[splitIndex] = tmp;
 	}
 }
 
-void	btQuantizedBvh::assignInternalNodeFromLeafNode(int internalNode,int leafNodeIndex)
+void btQuantizedBvh::assignInternalNodeFromLeafNode(int internalNode, int leafNodeIndex)
 {
 	if (m_useQuantization)
 	{
 		m_quantizedContiguousNodes[internalNode] = m_quantizedLeafNodes[leafNodeIndex];
-	} else
+	}
+	else
 	{
 		m_contiguousNodes[internalNode] = m_leafNodes[leafNodeIndex];
 	}
@@ -844,11 +814,10 @@ static const unsigned BVH_ALIGNMENT_MASK = BVH_ALIGNMENT-1;
 static const unsigned BVH_ALIGNMENT_BLOCKS = 2;
 #endif
 
-
 unsigned int btQuantizedBvh::getAlignmentSerializationPadding()
 {
 	// I changed this to 0 since the extra padding is not needed or used.
-	return 0;//BVH_ALIGNMENT_BLOCKS * BVH_ALIGNMENT;
+	return 0;  //BVH_ALIGNMENT_BLOCKS * BVH_ALIGNMENT;
 }
 
 unsigned btQuantizedBvh::calculateSerializeBufferSize() const
@@ -862,12 +831,12 @@ unsigned btQuantizedBvh::calculateSerializeBufferSize() const
 	return baseSize + m_curNodeIndex * sizeof(btOptimizedBvhNode);
 }
 
-bool btQuantizedBvh::serialize(void *o_alignedDataBuffer, unsigned /*i_dataBufferSize */, bool i_swapEndian) const
+bool btQuantizedBvh::serialize(void* o_alignedDataBuffer, unsigned /*i_dataBufferSize */, bool i_swapEndian) const
 {
 	btAssert(m_subtreeHeaderCount == m_SubtreeHeaders.size());
 	m_subtreeHeaderCount = m_SubtreeHeaders.size();
 
-/*	if (i_dataBufferSize < calculateSerializeBufferSize() || o_alignedDataBuffer == NULL || (((unsigned)o_alignedDataBuffer & BVH_ALIGNMENT_MASK) != 0))
+	/*	if (i_dataBufferSize < calculateSerializeBufferSize() || o_alignedDataBuffer == NULL || (((unsigned)o_alignedDataBuffer & BVH_ALIGNMENT_MASK) != 0))
 	{
 		///check alignedment for buffer?
 		btAssert(0);
@@ -875,7 +844,7 @@ bool btQuantizedBvh::serialize(void *o_alignedDataBuffer, unsigned /*i_dataBuffe
 	}
 */
 
-	btQuantizedBvh *targetBvh = (btQuantizedBvh *)o_alignedDataBuffer;
+	btQuantizedBvh* targetBvh = (btQuantizedBvh*)o_alignedDataBuffer;
 
 	// construct the class so the virtual function table, etc will be set up
 	// Also, m_leafNodes and m_quantizedLeafNodes will be initialized to default values by the constructor
@@ -885,10 +854,9 @@ bool btQuantizedBvh::serialize(void *o_alignedDataBuffer, unsigned /*i_dataBuffe
 	{
 		targetBvh->m_curNodeIndex = static_cast<int>(btSwapEndian(m_curNodeIndex));
 
-
-		btSwapVector3Endian(m_bvhAabbMin,targetBvh->m_bvhAabbMin);
-		btSwapVector3Endian(m_bvhAabbMax,targetBvh->m_bvhAabbMax);
-		btSwapVector3Endian(m_bvhQuantization,targetBvh->m_bvhQuantization);
+		btSwapVector3Endian(m_bvhAabbMin, targetBvh->m_bvhAabbMin);
+		btSwapVector3Endian(m_bvhAabbMax, targetBvh->m_bvhAabbMax);
+		btSwapVector3Endian(m_bvhQuantization, targetBvh->m_bvhQuantization);
 
 		targetBvh->m_traversalMode = (btTraversalMode)btSwapEndian(m_traversalMode);
 		targetBvh->m_subtreeHeaderCount = static_cast<int>(btSwapEndian(m_subtreeHeaderCount));
@@ -905,12 +873,12 @@ bool btQuantizedBvh::serialize(void *o_alignedDataBuffer, unsigned /*i_dataBuffe
 
 	targetBvh->m_useQuantization = m_useQuantization;
 
-	unsigned char *nodeData = (unsigned char *)targetBvh;
+	unsigned char* nodeData = (unsigned char*)targetBvh;
 	nodeData += sizeof(btQuantizedBvh);
-	
-	unsigned sizeToAdd = 0;//(BVH_ALIGNMENT-((unsigned)nodeData & BVH_ALIGNMENT_MASK))&BVH_ALIGNMENT_MASK;
+
+	unsigned sizeToAdd = 0;  //(BVH_ALIGNMENT-((unsigned)nodeData & BVH_ALIGNMENT_MASK))&BVH_ALIGNMENT_MASK;
 	nodeData += sizeToAdd;
-	
+
 	int nodeCount = m_curNodeIndex;
 
 	if (m_useQuantization)
@@ -936,7 +904,6 @@ bool btQuantizedBvh::serialize(void *o_alignedDataBuffer, unsigned /*i_dataBuffe
 		{
 			for (int nodeIndex = 0; nodeIndex < nodeCount; nodeIndex++)
 			{
-	
 				targetBvh->m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMin[0] = m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMin[0];
 				targetBvh->m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMin[1] = m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMin[1];
 				targetBvh->m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMin[2] = m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMin[2];
@@ -946,8 +913,6 @@ bool btQuantizedBvh::serialize(void *o_alignedDataBuffer, unsigned /*i_dataBuffe
 				targetBvh->m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMax[2] = m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMax[2];
 
 				targetBvh->m_quantizedContiguousNodes[nodeIndex].m_escapeIndexOrTriangleIndex = m_quantizedContiguousNodes[nodeIndex].m_escapeIndexOrTriangleIndex;
-
-
 			}
 		}
 		nodeData += sizeof(btQuantizedBvhNode) * nodeCount;
@@ -993,7 +958,7 @@ bool btQuantizedBvh::serialize(void *o_alignedDataBuffer, unsigned /*i_dataBuffe
 		targetBvh->m_contiguousNodes.initializeFromBuffer(NULL, 0, 0);
 	}
 
-	sizeToAdd = 0;//(BVH_ALIGNMENT-((unsigned)nodeData & BVH_ALIGNMENT_MASK))&BVH_ALIGNMENT_MASK;
+	sizeToAdd = 0;  //(BVH_ALIGNMENT-((unsigned)nodeData & BVH_ALIGNMENT_MASK))&BVH_ALIGNMENT_MASK;
 	nodeData += sizeToAdd;
 
 	// Now serialize the subtree headers
@@ -1048,14 +1013,13 @@ bool btQuantizedBvh::serialize(void *o_alignedDataBuffer, unsigned /*i_dataBuffe
 	return true;
 }
 
-btQuantizedBvh *btQuantizedBvh::deSerializeInPlace(void *i_alignedDataBuffer, unsigned int i_dataBufferSize, bool i_swapEndian)
+btQuantizedBvh* btQuantizedBvh::deSerializeInPlace(void* i_alignedDataBuffer, unsigned int i_dataBufferSize, bool i_swapEndian)
 {
-
-	if (i_alignedDataBuffer == NULL)// || (((unsigned)i_alignedDataBuffer & BVH_ALIGNMENT_MASK) != 0))
+	if (i_alignedDataBuffer == NULL)  // || (((unsigned)i_alignedDataBuffer & BVH_ALIGNMENT_MASK) != 0))
 	{
 		return NULL;
 	}
-	btQuantizedBvh *bvh = (btQuantizedBvh *)i_alignedDataBuffer;
+	btQuantizedBvh* bvh = (btQuantizedBvh*)i_alignedDataBuffer;
 
 	if (i_swapEndian)
 	{
@@ -1077,12 +1041,12 @@ btQuantizedBvh *btQuantizedBvh::deSerializeInPlace(void *i_alignedDataBuffer, un
 		return NULL;
 	}
 
-	unsigned char *nodeData = (unsigned char *)bvh;
+	unsigned char* nodeData = (unsigned char*)bvh;
 	nodeData += sizeof(btQuantizedBvh);
-	
-	unsigned sizeToAdd = 0;//(BVH_ALIGNMENT-((unsigned)nodeData & BVH_ALIGNMENT_MASK))&BVH_ALIGNMENT_MASK;
+
+	unsigned sizeToAdd = 0;  //(BVH_ALIGNMENT-((unsigned)nodeData & BVH_ALIGNMENT_MASK))&BVH_ALIGNMENT_MASK;
 	nodeData += sizeToAdd;
-	
+
 	int nodeCount = bvh->m_curNodeIndex;
 
 	// Must call placement new to fill in virtual function table, etc, but we don't want to overwrite most data, so call a special version of the constructor
@@ -1120,7 +1084,7 @@ btQuantizedBvh *btQuantizedBvh::deSerializeInPlace(void *i_alignedDataBuffer, un
 			{
 				btUnSwapVector3Endian(bvh->m_contiguousNodes[nodeIndex].m_aabbMinOrg);
 				btUnSwapVector3Endian(bvh->m_contiguousNodes[nodeIndex].m_aabbMaxOrg);
-				
+
 				bvh->m_contiguousNodes[nodeIndex].m_escapeIndex = static_cast<int>(btSwapEndian(bvh->m_contiguousNodes[nodeIndex].m_escapeIndex));
 				bvh->m_contiguousNodes[nodeIndex].m_subPart = static_cast<int>(btSwapEndian(bvh->m_contiguousNodes[nodeIndex].m_subPart));
 				bvh->m_contiguousNodes[nodeIndex].m_triangleIndex = static_cast<int>(btSwapEndian(bvh->m_contiguousNodes[nodeIndex].m_triangleIndex));
@@ -1129,7 +1093,7 @@ btQuantizedBvh *btQuantizedBvh::deSerializeInPlace(void *i_alignedDataBuffer, un
 		nodeData += sizeof(btOptimizedBvhNode) * nodeCount;
 	}
 
-	sizeToAdd = 0;//(BVH_ALIGNMENT-((unsigned)nodeData & BVH_ALIGNMENT_MASK))&BVH_ALIGNMENT_MASK;
+	sizeToAdd = 0;  //(BVH_ALIGNMENT-((unsigned)nodeData & BVH_ALIGNMENT_MASK))&BVH_ALIGNMENT_MASK;
 	nodeData += sizeToAdd;
 
 	// Now serialize the subtree headers
@@ -1155,13 +1119,11 @@ btQuantizedBvh *btQuantizedBvh::deSerializeInPlace(void *i_alignedDataBuffer, un
 }
 
 // Constructor that prevents btVector3's default constructor from being called
-btQuantizedBvh::btQuantizedBvh(btQuantizedBvh &self, bool /* ownsMemory */) :
-m_bvhAabbMin(self.m_bvhAabbMin),
-m_bvhAabbMax(self.m_bvhAabbMax),
-m_bvhQuantization(self.m_bvhQuantization),
-m_bulletVersion(BT_BULLET_VERSION)
+btQuantizedBvh::btQuantizedBvh(btQuantizedBvh& self, bool /* ownsMemory */) : m_bvhAabbMin(self.m_bvhAabbMin),
+																			  m_bvhAabbMax(self.m_bvhAabbMax),
+																			  m_bvhQuantization(self.m_bvhQuantization),
+																			  m_bulletVersion(BT_BULLET_VERSION)
 {
-
 }
 
 void btQuantizedBvh::deSerializeFloat(struct btQuantizedBvhFloatData& quantizedBvhFloatData)
@@ -1171,8 +1133,8 @@ void btQuantizedBvh::deSerializeFloat(struct btQuantizedBvhFloatData& quantizedB
 	m_bvhQuantization.deSerializeFloat(quantizedBvhFloatData.m_bvhQuantization);
 
 	m_curNodeIndex = quantizedBvhFloatData.m_curNodeIndex;
-	m_useQuantization = quantizedBvhFloatData.m_useQuantization!=0;
-	
+	m_useQuantization = quantizedBvhFloatData.m_useQuantization != 0;
+
 	{
 		int numElem = quantizedBvhFloatData.m_numContiguousLeafNodes;
 		m_contiguousNodes.resize(numElem);
@@ -1181,7 +1143,7 @@ void btQuantizedBvh::deSerializeFloat(struct btQuantizedBvhFloatData& quantizedB
 		{
 			btOptimizedBvhNodeFloatData* memPtr = quantizedBvhFloatData.m_contiguousNodesPtr;
 
-			for (int i=0;i<numElem;i++,memPtr++)
+			for (int i = 0; i < numElem; i++, memPtr++)
 			{
 				m_contiguousNodes[i].m_aabbMaxOrg.deSerializeFloat(memPtr->m_aabbMaxOrg);
 				m_contiguousNodes[i].m_aabbMinOrg.deSerializeFloat(memPtr->m_aabbMinOrg);
@@ -1195,11 +1157,11 @@ void btQuantizedBvh::deSerializeFloat(struct btQuantizedBvhFloatData& quantizedB
 	{
 		int numElem = quantizedBvhFloatData.m_numQuantizedContiguousNodes;
 		m_quantizedContiguousNodes.resize(numElem);
-		
+
 		if (numElem)
 		{
 			btQuantizedBvhNodeData* memPtr = quantizedBvhFloatData.m_quantizedContiguousNodesPtr;
-			for (int i=0;i<numElem;i++,memPtr++)
+			for (int i = 0; i < numElem; i++, memPtr++)
 			{
 				m_quantizedContiguousNodes[i].m_escapeIndexOrTriangleIndex = memPtr->m_escapeIndexOrTriangleIndex;
 				m_quantizedContiguousNodes[i].m_quantizedAabbMax[0] = memPtr->m_quantizedAabbMax[0];
@@ -1213,16 +1175,16 @@ void btQuantizedBvh::deSerializeFloat(struct btQuantizedBvhFloatData& quantizedB
 	}
 
 	m_traversalMode = btTraversalMode(quantizedBvhFloatData.m_traversalMode);
-	
+
 	{
 		int numElem = quantizedBvhFloatData.m_numSubtreeHeaders;
 		m_SubtreeHeaders.resize(numElem);
 		if (numElem)
 		{
 			btBvhSubtreeInfoData* memPtr = quantizedBvhFloatData.m_subTreeInfoPtr;
-			for (int i=0;i<numElem;i++,memPtr++)
+			for (int i = 0; i < numElem; i++, memPtr++)
 			{
-				m_SubtreeHeaders[i].m_quantizedAabbMax[0] = memPtr->m_quantizedAabbMax[0] ;
+				m_SubtreeHeaders[i].m_quantizedAabbMax[0] = memPtr->m_quantizedAabbMax[0];
 				m_SubtreeHeaders[i].m_quantizedAabbMax[1] = memPtr->m_quantizedAabbMax[1];
 				m_SubtreeHeaders[i].m_quantizedAabbMax[2] = memPtr->m_quantizedAabbMax[2];
 				m_SubtreeHeaders[i].m_quantizedAabbMin[0] = memPtr->m_quantizedAabbMin[0];
@@ -1242,8 +1204,8 @@ void btQuantizedBvh::deSerializeDouble(struct btQuantizedBvhDoubleData& quantize
 	m_bvhQuantization.deSerializeDouble(quantizedBvhDoubleData.m_bvhQuantization);
 
 	m_curNodeIndex = quantizedBvhDoubleData.m_curNodeIndex;
-	m_useQuantization = quantizedBvhDoubleData.m_useQuantization!=0;
-	
+	m_useQuantization = quantizedBvhDoubleData.m_useQuantization != 0;
+
 	{
 		int numElem = quantizedBvhDoubleData.m_numContiguousLeafNodes;
 		m_contiguousNodes.resize(numElem);
@@ -1252,7 +1214,7 @@ void btQuantizedBvh::deSerializeDouble(struct btQuantizedBvhDoubleData& quantize
 		{
 			btOptimizedBvhNodeDoubleData* memPtr = quantizedBvhDoubleData.m_contiguousNodesPtr;
 
-			for (int i=0;i<numElem;i++,memPtr++)
+			for (int i = 0; i < numElem; i++, memPtr++)
 			{
 				m_contiguousNodes[i].m_aabbMaxOrg.deSerializeDouble(memPtr->m_aabbMaxOrg);
 				m_contiguousNodes[i].m_aabbMinOrg.deSerializeDouble(memPtr->m_aabbMinOrg);
@@ -1266,11 +1228,11 @@ void btQuantizedBvh::deSerializeDouble(struct btQuantizedBvhDoubleData& quantize
 	{
 		int numElem = quantizedBvhDoubleData.m_numQuantizedContiguousNodes;
 		m_quantizedContiguousNodes.resize(numElem);
-		
+
 		if (numElem)
 		{
 			btQuantizedBvhNodeData* memPtr = quantizedBvhDoubleData.m_quantizedContiguousNodesPtr;
-			for (int i=0;i<numElem;i++,memPtr++)
+			for (int i = 0; i < numElem; i++, memPtr++)
 			{
 				m_quantizedContiguousNodes[i].m_escapeIndexOrTriangleIndex = memPtr->m_escapeIndexOrTriangleIndex;
 				m_quantizedContiguousNodes[i].m_quantizedAabbMax[0] = memPtr->m_quantizedAabbMax[0];
@@ -1284,16 +1246,16 @@ void btQuantizedBvh::deSerializeDouble(struct btQuantizedBvhDoubleData& quantize
 	}
 
 	m_traversalMode = btTraversalMode(quantizedBvhDoubleData.m_traversalMode);
-	
+
 	{
 		int numElem = quantizedBvhDoubleData.m_numSubtreeHeaders;
 		m_SubtreeHeaders.resize(numElem);
 		if (numElem)
 		{
 			btBvhSubtreeInfoData* memPtr = quantizedBvhDoubleData.m_subTreeInfoPtr;
-			for (int i=0;i<numElem;i++,memPtr++)
+			for (int i = 0; i < numElem; i++, memPtr++)
 			{
-				m_SubtreeHeaders[i].m_quantizedAabbMax[0] = memPtr->m_quantizedAabbMax[0] ;
+				m_SubtreeHeaders[i].m_quantizedAabbMax[0] = memPtr->m_quantizedAabbMax[0];
 				m_SubtreeHeaders[i].m_quantizedAabbMax[1] = memPtr->m_quantizedAabbMax[1];
 				m_SubtreeHeaders[i].m_quantizedAabbMax[2] = memPtr->m_quantizedAabbMax[2];
 				m_SubtreeHeaders[i].m_quantizedAabbMin[0] = memPtr->m_quantizedAabbMin[0];
@@ -1304,32 +1266,29 @@ void btQuantizedBvh::deSerializeDouble(struct btQuantizedBvhDoubleData& quantize
 			}
 		}
 	}
-
 }
 
-
-
 ///fills the dataBuffer and returns the struct name (and 0 on failure)
-const char*	btQuantizedBvh::serialize(void* dataBuffer, btSerializer* serializer) const
+const char* btQuantizedBvh::serialize(void* dataBuffer, btSerializer* serializer) const
 {
 	btQuantizedBvhData* quantizedData = (btQuantizedBvhData*)dataBuffer;
-	
+
 	m_bvhAabbMax.serialize(quantizedData->m_bvhAabbMax);
 	m_bvhAabbMin.serialize(quantizedData->m_bvhAabbMin);
 	m_bvhQuantization.serialize(quantizedData->m_bvhQuantization);
 
 	quantizedData->m_curNodeIndex = m_curNodeIndex;
 	quantizedData->m_useQuantization = m_useQuantization;
-	
+
 	quantizedData->m_numContiguousLeafNodes = m_contiguousNodes.size();
-	quantizedData->m_contiguousNodesPtr = (btOptimizedBvhNodeData*) (m_contiguousNodes.size() ? serializer->getUniquePointer((void*)&m_contiguousNodes[0]) : 0);
+	quantizedData->m_contiguousNodesPtr = (btOptimizedBvhNodeData*)(m_contiguousNodes.size() ? serializer->getUniquePointer((void*)&m_contiguousNodes[0]) : 0);
 	if (quantizedData->m_contiguousNodesPtr)
 	{
 		int sz = sizeof(btOptimizedBvhNodeData);
 		int numElem = m_contiguousNodes.size();
-		btChunk* chunk = serializer->allocate(sz,numElem);
+		btChunk* chunk = serializer->allocate(sz, numElem);
 		btOptimizedBvhNodeData* memPtr = (btOptimizedBvhNodeData*)chunk->m_oldPtr;
-		for (int i=0;i<numElem;i++,memPtr++)
+		for (int i = 0; i < numElem; i++, memPtr++)
 		{
 			m_contiguousNodes[i].m_aabbMaxOrg.serialize(memPtr->m_aabbMaxOrg);
 			m_contiguousNodes[i].m_aabbMinOrg.serialize(memPtr->m_aabbMinOrg);
@@ -1339,19 +1298,19 @@ const char*	btQuantizedBvh::serialize(void* dataBuffer, btSerializer* serializer
 			// Fill padding with zeros to appease msan.
 			memset(memPtr->m_pad, 0, sizeof(memPtr->m_pad));
 		}
-		serializer->finalizeChunk(chunk,"btOptimizedBvhNodeData",BT_ARRAY_CODE,(void*)&m_contiguousNodes[0]);
+		serializer->finalizeChunk(chunk, "btOptimizedBvhNodeData", BT_ARRAY_CODE, (void*)&m_contiguousNodes[0]);
 	}
 
 	quantizedData->m_numQuantizedContiguousNodes = m_quantizedContiguousNodes.size();
-//	printf("quantizedData->m_numQuantizedContiguousNodes=%d\n",quantizedData->m_numQuantizedContiguousNodes);
-	quantizedData->m_quantizedContiguousNodesPtr =(btQuantizedBvhNodeData*) (m_quantizedContiguousNodes.size() ? serializer->getUniquePointer((void*)&m_quantizedContiguousNodes[0]) : 0);
+	//	printf("quantizedData->m_numQuantizedContiguousNodes=%d\n",quantizedData->m_numQuantizedContiguousNodes);
+	quantizedData->m_quantizedContiguousNodesPtr = (btQuantizedBvhNodeData*)(m_quantizedContiguousNodes.size() ? serializer->getUniquePointer((void*)&m_quantizedContiguousNodes[0]) : 0);
 	if (quantizedData->m_quantizedContiguousNodesPtr)
 	{
 		int sz = sizeof(btQuantizedBvhNodeData);
 		int numElem = m_quantizedContiguousNodes.size();
-		btChunk* chunk = serializer->allocate(sz,numElem);
+		btChunk* chunk = serializer->allocate(sz, numElem);
 		btQuantizedBvhNodeData* memPtr = (btQuantizedBvhNodeData*)chunk->m_oldPtr;
-		for (int i=0;i<numElem;i++,memPtr++)
+		for (int i = 0; i < numElem; i++, memPtr++)
 		{
 			memPtr->m_escapeIndexOrTriangleIndex = m_quantizedContiguousNodes[i].m_escapeIndexOrTriangleIndex;
 			memPtr->m_quantizedAabbMax[0] = m_quantizedContiguousNodes[i].m_quantizedAabbMax[0];
@@ -1361,20 +1320,20 @@ const char*	btQuantizedBvh::serialize(void* dataBuffer, btSerializer* serializer
 			memPtr->m_quantizedAabbMin[1] = m_quantizedContiguousNodes[i].m_quantizedAabbMin[1];
 			memPtr->m_quantizedAabbMin[2] = m_quantizedContiguousNodes[i].m_quantizedAabbMin[2];
 		}
-		serializer->finalizeChunk(chunk,"btQuantizedBvhNodeData",BT_ARRAY_CODE,(void*)&m_quantizedContiguousNodes[0]);
+		serializer->finalizeChunk(chunk, "btQuantizedBvhNodeData", BT_ARRAY_CODE, (void*)&m_quantizedContiguousNodes[0]);
 	}
 
 	quantizedData->m_traversalMode = int(m_traversalMode);
 	quantizedData->m_numSubtreeHeaders = m_SubtreeHeaders.size();
 
-	quantizedData->m_subTreeInfoPtr = (btBvhSubtreeInfoData*) (m_SubtreeHeaders.size() ? serializer->getUniquePointer((void*)&m_SubtreeHeaders[0]) : 0);
+	quantizedData->m_subTreeInfoPtr = (btBvhSubtreeInfoData*)(m_SubtreeHeaders.size() ? serializer->getUniquePointer((void*)&m_SubtreeHeaders[0]) : 0);
 	if (quantizedData->m_subTreeInfoPtr)
 	{
 		int sz = sizeof(btBvhSubtreeInfoData);
 		int numElem = m_SubtreeHeaders.size();
-		btChunk* chunk = serializer->allocate(sz,numElem);
+		btChunk* chunk = serializer->allocate(sz, numElem);
 		btBvhSubtreeInfoData* memPtr = (btBvhSubtreeInfoData*)chunk->m_oldPtr;
-		for (int i=0;i<numElem;i++,memPtr++)
+		for (int i = 0; i < numElem; i++, memPtr++)
 		{
 			memPtr->m_quantizedAabbMax[0] = m_SubtreeHeaders[i].m_quantizedAabbMax[0];
 			memPtr->m_quantizedAabbMax[1] = m_SubtreeHeaders[i].m_quantizedAabbMax[1];
@@ -1386,12 +1345,7 @@ const char*	btQuantizedBvh::serialize(void* dataBuffer, btSerializer* serializer
 			memPtr->m_rootNodeIndex = m_SubtreeHeaders[i].m_rootNodeIndex;
 			memPtr->m_subtreeSize = m_SubtreeHeaders[i].m_subtreeSize;
 		}
-		serializer->finalizeChunk(chunk,"btBvhSubtreeInfoData",BT_ARRAY_CODE,(void*)&m_SubtreeHeaders[0]);
+		serializer->finalizeChunk(chunk, "btBvhSubtreeInfoData", BT_ARRAY_CODE, (void*)&m_SubtreeHeaders[0]);
 	}
 	return btQuantizedBvhDataName;
 }
-
-
-
-
-
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btQuantizedBvh.h b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btQuantizedBvh.h
index 3dd5ac9bb65..1c47b9ccf29 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btQuantizedBvh.h
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btQuantizedBvh.h
@@ -22,11 +22,11 @@ class btSerializer;
 #ifdef DEBUG_CHECK_DEQUANTIZATION
 #ifdef __SPU__
 #define printf spu_printf
-#endif //__SPU__
+#endif  //__SPU__
 
 #include <stdio.h>
 #include <stdlib.h>
-#endif //DEBUG_CHECK_DEQUANTIZATION
+#endif  //DEBUG_CHECK_DEQUANTIZATION
 
 #include "LinearMath/btVector3.h"
 #include "LinearMath/btAlignedAllocator.h"
@@ -41,13 +41,10 @@ class btSerializer;
 #define btQuantizedBvhDataName "btQuantizedBvhFloatData"
 #endif
 
-
-
 //http://msdn.microsoft.com/library/default.asp?url=/library/en-us/vclang/html/vclrf__m128.asp
 
-
 //Note: currently we have 16 bytes per quantized node
-#define MAX_SUBTREE_SIZE_IN_BYTES  2048
+#define MAX_SUBTREE_SIZE_IN_BYTES 2048
 
 // 10 gives the potential for 1024 parts, with at most 2^21 (2097152) (minus one
 // actually) triangles each (since the sign bit is reserved
@@ -55,15 +52,16 @@ class btSerializer;
 
 ///btQuantizedBvhNode is a compressed aabb node, 16 bytes.
 ///Node can be used for leafnode or internal node. Leafnodes can point to 32-bit triangle index (non-negative range).
-ATTRIBUTE_ALIGNED16	(struct) btQuantizedBvhNode
+ATTRIBUTE_ALIGNED16(struct)
+btQuantizedBvhNode
 {
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
 	//12 bytes
-	unsigned short int	m_quantizedAabbMin[3];
-	unsigned short int	m_quantizedAabbMax[3];
+	unsigned short int m_quantizedAabbMin[3];
+	unsigned short int m_quantizedAabbMax[3];
 	//4 bytes
-	int	m_escapeIndexOrTriangleIndex;
+	int m_escapeIndexOrTriangleIndex;
 
 	bool isLeafNode() const
 	{
@@ -75,68 +73,67 @@ ATTRIBUTE_ALIGNED16	(struct) btQuantizedBvhNode
 		btAssert(!isLeafNode());
 		return -m_escapeIndexOrTriangleIndex;
 	}
-	int	getTriangleIndex() const
+	int getTriangleIndex() const
 	{
 		btAssert(isLeafNode());
-		unsigned int x=0;
-		unsigned int y = (~(x&0))<<(31-MAX_NUM_PARTS_IN_BITS);
+		unsigned int x = 0;
+		unsigned int y = (~(x & 0)) << (31 - MAX_NUM_PARTS_IN_BITS);
 		// Get only the lower bits where the triangle index is stored
-		return (m_escapeIndexOrTriangleIndex&~(y));
+		return (m_escapeIndexOrTriangleIndex & ~(y));
 	}
-	int	getPartId() const
+	int getPartId() const
 	{
 		btAssert(isLeafNode());
 		// Get only the highest bits where the part index is stored
-		return (m_escapeIndexOrTriangleIndex>>(31-MAX_NUM_PARTS_IN_BITS));
+		return (m_escapeIndexOrTriangleIndex >> (31 - MAX_NUM_PARTS_IN_BITS));
 	}
-}
-;
+};
 
 /// btOptimizedBvhNode contains both internal and leaf node information.
 /// Total node size is 44 bytes / node. You can use the compressed version of 16 bytes.
-ATTRIBUTE_ALIGNED16 (struct) btOptimizedBvhNode
+ATTRIBUTE_ALIGNED16(struct)
+btOptimizedBvhNode
 {
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
 	//32 bytes
-	btVector3	m_aabbMinOrg;
-	btVector3	m_aabbMaxOrg;
+	btVector3 m_aabbMinOrg;
+	btVector3 m_aabbMaxOrg;
 
 	//4
-	int	m_escapeIndex;
+	int m_escapeIndex;
 
 	//8
 	//for child nodes
-	int	m_subPart;
-	int	m_triangleIndex;
+	int m_subPart;
+	int m_triangleIndex;
 
-//pad the size to 64 bytes
-	char	m_padding[20];
+	//pad the size to 64 bytes
+	char m_padding[20];
 };
 
-
 ///btBvhSubtreeInfo provides info to gather a subtree of limited size
-ATTRIBUTE_ALIGNED16(class) btBvhSubtreeInfo
+ATTRIBUTE_ALIGNED16(class)
+btBvhSubtreeInfo
 {
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
 	//12 bytes
-	unsigned short int	m_quantizedAabbMin[3];
-	unsigned short int	m_quantizedAabbMax[3];
+	unsigned short int m_quantizedAabbMin[3];
+	unsigned short int m_quantizedAabbMax[3];
 	//4 bytes, points to the root of the subtree
-	int			m_rootNodeIndex;
+	int m_rootNodeIndex;
 	//4 bytes
-	int			m_subtreeSize;
-	int			m_padding[3];
+	int m_subtreeSize;
+	int m_padding[3];
 
 	btBvhSubtreeInfo()
 	{
 		//memset(&m_padding[0], 0, sizeof(m_padding));
 	}
 
-
-	void	setAabbFromQuantizeNode(const btQuantizedBvhNode& quantizedNode)
+	void setAabbFromQuantizeNode(const btQuantizedBvhNode& quantizedNode)
 	{
 		m_quantizedAabbMin[0] = quantizedNode.m_quantizedAabbMin[0];
 		m_quantizedAabbMin[1] = quantizedNode.m_quantizedAabbMin[1];
@@ -145,14 +142,12 @@ public:
 		m_quantizedAabbMax[1] = quantizedNode.m_quantizedAabbMax[1];
 		m_quantizedAabbMax[2] = quantizedNode.m_quantizedAabbMax[2];
 	}
-}
-;
-
+};
 
 class btNodeOverlapCallback
 {
 public:
-	virtual ~btNodeOverlapCallback() {};
+	virtual ~btNodeOverlapCallback(){};
 
 	virtual void processNode(int subPart, int triangleIndex) = 0;
 };
@@ -160,18 +155,16 @@ public:
 #include "LinearMath/btAlignedAllocator.h"
 #include "LinearMath/btAlignedObjectArray.h"
 
-
-
 ///for code readability:
-typedef btAlignedObjectArray<btOptimizedBvhNode>	NodeArray;
-typedef btAlignedObjectArray<btQuantizedBvhNode>	QuantizedNodeArray;
-typedef btAlignedObjectArray<btBvhSubtreeInfo>		BvhSubtreeInfoArray;
-
+typedef btAlignedObjectArray<btOptimizedBvhNode> NodeArray;
+typedef btAlignedObjectArray<btQuantizedBvhNode> QuantizedNodeArray;
+typedef btAlignedObjectArray<btBvhSubtreeInfo> BvhSubtreeInfoArray;
 
 ///The btQuantizedBvh class stores an AABB tree that can be quickly traversed on CPU and Cell SPU.
 ///It is used by the btBvhTriangleMeshShape as midphase.
 ///It is recommended to use quantization for better performance and lower memory requirements.
-ATTRIBUTE_ALIGNED16(class) btQuantizedBvh
+ATTRIBUTE_ALIGNED16(class)
+btQuantizedBvh
 {
 public:
 	enum btTraversalMode
@@ -182,54 +175,47 @@ public:
 	};
 
 protected:
+	btVector3 m_bvhAabbMin;
+	btVector3 m_bvhAabbMax;
+	btVector3 m_bvhQuantization;
 
+	int m_bulletVersion;  //for serialization versioning. It could also be used to detect endianess.
 
-	btVector3			m_bvhAabbMin;
-	btVector3			m_bvhAabbMax;
-	btVector3			m_bvhQuantization;
-
-	int					m_bulletVersion;	//for serialization versioning. It could also be used to detect endianess.
-
-	int					m_curNodeIndex;
+	int m_curNodeIndex;
 	//quantization data
-	bool				m_useQuantization;
+	bool m_useQuantization;
 
+	NodeArray m_leafNodes;
+	NodeArray m_contiguousNodes;
+	QuantizedNodeArray m_quantizedLeafNodes;
+	QuantizedNodeArray m_quantizedContiguousNodes;
 
-
-	NodeArray			m_leafNodes;
-	NodeArray			m_contiguousNodes;
-	QuantizedNodeArray	m_quantizedLeafNodes;
-	QuantizedNodeArray	m_quantizedContiguousNodes;
-	
-	btTraversalMode	m_traversalMode;
-	BvhSubtreeInfoArray		m_SubtreeHeaders;
+	btTraversalMode m_traversalMode;
+	BvhSubtreeInfoArray m_SubtreeHeaders;
 
 	//This is only used for serialization so we don't have to add serialization directly to btAlignedObjectArray
 	mutable int m_subtreeHeaderCount;
 
-	
-
-
-
 	///two versions, one for quantized and normal nodes. This allows code-reuse while maintaining readability (no template/macro!)
 	///this might be refactored into a virtual, it is usually not calculated at run-time
-	void	setInternalNodeAabbMin(int nodeIndex, const btVector3& aabbMin)
+	void setInternalNodeAabbMin(int nodeIndex, const btVector3& aabbMin)
 	{
 		if (m_useQuantization)
 		{
-			quantize(&m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMin[0] ,aabbMin,0);
-		} else
+			quantize(&m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMin[0], aabbMin, 0);
+		}
+		else
 		{
 			m_contiguousNodes[nodeIndex].m_aabbMinOrg = aabbMin;
-
 		}
 	}
-	void	setInternalNodeAabbMax(int nodeIndex,const btVector3& aabbMax)
+	void setInternalNodeAabbMax(int nodeIndex, const btVector3& aabbMax)
 	{
 		if (m_useQuantization)
 		{
-			quantize(&m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMax[0],aabbMax,1);
-		} else
+			quantize(&m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMax[0], aabbMax, 1);
+		}
+		else
 		{
 			m_contiguousNodes[nodeIndex].m_aabbMaxOrg = aabbMax;
 		}
@@ -243,115 +229,102 @@ protected:
 		}
 		//non-quantized
 		return m_leafNodes[nodeIndex].m_aabbMinOrg;
-
 	}
 	btVector3 getAabbMax(int nodeIndex) const
 	{
 		if (m_useQuantization)
 		{
 			return unQuantize(&m_quantizedLeafNodes[nodeIndex].m_quantizedAabbMax[0]);
-		} 
+		}
 		//non-quantized
 		return m_leafNodes[nodeIndex].m_aabbMaxOrg;
-		
 	}
 
-	
-	void	setInternalNodeEscapeIndex(int nodeIndex, int escapeIndex)
+	void setInternalNodeEscapeIndex(int nodeIndex, int escapeIndex)
 	{
 		if (m_useQuantization)
 		{
 			m_quantizedContiguousNodes[nodeIndex].m_escapeIndexOrTriangleIndex = -escapeIndex;
-		} 
+		}
 		else
 		{
 			m_contiguousNodes[nodeIndex].m_escapeIndex = escapeIndex;
 		}
-
 	}
 
-	void mergeInternalNodeAabb(int nodeIndex,const btVector3& newAabbMin,const btVector3& newAabbMax) 
+	void mergeInternalNodeAabb(int nodeIndex, const btVector3& newAabbMin, const btVector3& newAabbMax)
 	{
 		if (m_useQuantization)
 		{
 			unsigned short int quantizedAabbMin[3];
 			unsigned short int quantizedAabbMax[3];
-			quantize(quantizedAabbMin,newAabbMin,0);
-			quantize(quantizedAabbMax,newAabbMax,1);
-			for (int i=0;i<3;i++)
+			quantize(quantizedAabbMin, newAabbMin, 0);
+			quantize(quantizedAabbMax, newAabbMax, 1);
+			for (int i = 0; i < 3; i++)
 			{
 				if (m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMin[i] > quantizedAabbMin[i])
 					m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMin[i] = quantizedAabbMin[i];
 
 				if (m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMax[i] < quantizedAabbMax[i])
 					m_quantizedContiguousNodes[nodeIndex].m_quantizedAabbMax[i] = quantizedAabbMax[i];
-
 			}
-		} else
+		}
+		else
 		{
 			//non-quantized
 			m_contiguousNodes[nodeIndex].m_aabbMinOrg.setMin(newAabbMin);
-			m_contiguousNodes[nodeIndex].m_aabbMaxOrg.setMax(newAabbMax);		
+			m_contiguousNodes[nodeIndex].m_aabbMaxOrg.setMax(newAabbMax);
 		}
 	}
 
-	void	swapLeafNodes(int firstIndex,int secondIndex);
+	void swapLeafNodes(int firstIndex, int secondIndex);
 
-	void	assignInternalNodeFromLeafNode(int internalNode,int leafNodeIndex);
+	void assignInternalNodeFromLeafNode(int internalNode, int leafNodeIndex);
 
 protected:
+	void buildTree(int startIndex, int endIndex);
 
-	
+	int calcSplittingAxis(int startIndex, int endIndex);
 
-	void	buildTree	(int startIndex,int endIndex);
+	int sortAndCalcSplittingIndex(int startIndex, int endIndex, int splitAxis);
 
-	int	calcSplittingAxis(int startIndex,int endIndex);
+	void walkStacklessTree(btNodeOverlapCallback * nodeCallback, const btVector3& aabbMin, const btVector3& aabbMax) const;
 
-	int	sortAndCalcSplittingIndex(int startIndex,int endIndex,int splitAxis);
-	
-	void	walkStacklessTree(btNodeOverlapCallback* nodeCallback,const btVector3& aabbMin,const btVector3& aabbMax) const;
-
-	void	walkStacklessQuantizedTreeAgainstRay(btNodeOverlapCallback* nodeCallback, const btVector3& raySource, const btVector3& rayTarget, const btVector3& aabbMin, const btVector3& aabbMax, int startNodeIndex,int endNodeIndex) const;
-	void	walkStacklessQuantizedTree(btNodeOverlapCallback* nodeCallback,unsigned short int* quantizedQueryAabbMin,unsigned short int* quantizedQueryAabbMax,int startNodeIndex,int endNodeIndex) const;
-	void	walkStacklessTreeAgainstRay(btNodeOverlapCallback* nodeCallback, const btVector3& raySource, const btVector3& rayTarget, const btVector3& aabbMin, const btVector3& aabbMax, int startNodeIndex,int endNodeIndex) const;
+	void walkStacklessQuantizedTreeAgainstRay(btNodeOverlapCallback * nodeCallback, const btVector3& raySource, const btVector3& rayTarget, const btVector3& aabbMin, const btVector3& aabbMax, int startNodeIndex, int endNodeIndex) const;
+	void walkStacklessQuantizedTree(btNodeOverlapCallback * nodeCallback, unsigned short int* quantizedQueryAabbMin, unsigned short int* quantizedQueryAabbMax, int startNodeIndex, int endNodeIndex) const;
+	void walkStacklessTreeAgainstRay(btNodeOverlapCallback * nodeCallback, const btVector3& raySource, const btVector3& rayTarget, const btVector3& aabbMin, const btVector3& aabbMax, int startNodeIndex, int endNodeIndex) const;
 
 	///tree traversal designed for small-memory processors like PS3 SPU
-	void	walkStacklessQuantizedTreeCacheFriendly(btNodeOverlapCallback* nodeCallback,unsigned short int* quantizedQueryAabbMin,unsigned short int* quantizedQueryAabbMax) const;
+	void walkStacklessQuantizedTreeCacheFriendly(btNodeOverlapCallback * nodeCallback, unsigned short int* quantizedQueryAabbMin, unsigned short int* quantizedQueryAabbMax) const;
 
 	///use the 16-byte stackless 'skipindex' node tree to do a recursive traversal
-	void	walkRecursiveQuantizedTreeAgainstQueryAabb(const btQuantizedBvhNode* currentNode,btNodeOverlapCallback* nodeCallback,unsigned short int* quantizedQueryAabbMin,unsigned short int* quantizedQueryAabbMax) const;
+	void walkRecursiveQuantizedTreeAgainstQueryAabb(const btQuantizedBvhNode* currentNode, btNodeOverlapCallback* nodeCallback, unsigned short int* quantizedQueryAabbMin, unsigned short int* quantizedQueryAabbMax) const;
 
 	///use the 16-byte stackless 'skipindex' node tree to do a recursive traversal
-	void	walkRecursiveQuantizedTreeAgainstQuantizedTree(const btQuantizedBvhNode* treeNodeA,const btQuantizedBvhNode* treeNodeB,btNodeOverlapCallback* nodeCallback) const;
-	
+	void walkRecursiveQuantizedTreeAgainstQuantizedTree(const btQuantizedBvhNode* treeNodeA, const btQuantizedBvhNode* treeNodeB, btNodeOverlapCallback* nodeCallback) const;
 
-
-
-	void	updateSubtreeHeaders(int leftChildNodexIndex,int rightChildNodexIndex);
+	void updateSubtreeHeaders(int leftChildNodexIndex, int rightChildNodexIndex);
 
 public:
-	
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
 	btQuantizedBvh();
 
 	virtual ~btQuantizedBvh();
 
-	
 	///***************************************** expert/internal use only *************************
-	void	setQuantizationValues(const btVector3& bvhAabbMin,const btVector3& bvhAabbMax,btScalar quantizationMargin=btScalar(1.0));
-	QuantizedNodeArray&	getLeafNodeArray() {			return	m_quantizedLeafNodes;	}
+	void setQuantizationValues(const btVector3& bvhAabbMin, const btVector3& bvhAabbMax, btScalar quantizationMargin = btScalar(1.0));
+	QuantizedNodeArray& getLeafNodeArray() { return m_quantizedLeafNodes; }
 	///buildInternal is expert use only: assumes that setQuantizationValues and LeafNodeArray are initialized
-	void	buildInternal();
+	void buildInternal();
 	///***************************************** expert/internal use only *************************
 
-	void	reportAabbOverlappingNodex(btNodeOverlapCallback* nodeCallback,const btVector3& aabbMin,const btVector3& aabbMax) const;
-	void	reportRayOverlappingNodex (btNodeOverlapCallback* nodeCallback, const btVector3& raySource, const btVector3& rayTarget) const;
-	void	reportBoxCastOverlappingNodex(btNodeOverlapCallback* nodeCallback, const btVector3& raySource, const btVector3& rayTarget, const btVector3& aabbMin,const btVector3& aabbMax) const;
+	void reportAabbOverlappingNodex(btNodeOverlapCallback * nodeCallback, const btVector3& aabbMin, const btVector3& aabbMax) const;
+	void reportRayOverlappingNodex(btNodeOverlapCallback * nodeCallback, const btVector3& raySource, const btVector3& rayTarget) const;
+	void reportBoxCastOverlappingNodex(btNodeOverlapCallback * nodeCallback, const btVector3& raySource, const btVector3& rayTarget, const btVector3& aabbMin, const btVector3& aabbMax) const;
 
-		SIMD_FORCE_INLINE void quantize(unsigned short* out, const btVector3& point,int isMax) const
+	SIMD_FORCE_INLINE void quantize(unsigned short* out, const btVector3& point, int isMax) const
 	{
-
 		btAssert(m_useQuantization);
 
 		btAssert(point.getX() <= m_bvhAabbMax.getX());
@@ -368,16 +341,16 @@ public:
 		///@todo: double-check this
 		if (isMax)
 		{
-			out[0] = (unsigned short) (((unsigned short)(v.getX()+btScalar(1.)) | 1));
-			out[1] = (unsigned short) (((unsigned short)(v.getY()+btScalar(1.)) | 1));
-			out[2] = (unsigned short) (((unsigned short)(v.getZ()+btScalar(1.)) | 1));
-		} else
-		{
-			out[0] = (unsigned short) (((unsigned short)(v.getX()) & 0xfffe));
-			out[1] = (unsigned short) (((unsigned short)(v.getY()) & 0xfffe));
-			out[2] = (unsigned short) (((unsigned short)(v.getZ()) & 0xfffe));
+			out[0] = (unsigned short)(((unsigned short)(v.getX() + btScalar(1.)) | 1));
+			out[1] = (unsigned short)(((unsigned short)(v.getY() + btScalar(1.)) | 1));
+			out[2] = (unsigned short)(((unsigned short)(v.getZ() + btScalar(1.)) | 1));
+		}
+		else
+		{
+			out[0] = (unsigned short)(((unsigned short)(v.getX()) & 0xfffe));
+			out[1] = (unsigned short)(((unsigned short)(v.getY()) & 0xfffe));
+			out[2] = (unsigned short)(((unsigned short)(v.getZ()) & 0xfffe));
 		}
-
 
 #ifdef DEBUG_CHECK_DEQUANTIZATION
 		btVector3 newPoint = unQuantize(out);
@@ -385,105 +358,97 @@ public:
 		{
 			if (newPoint.getX() < point.getX())
 			{
-				printf("unconservative X, diffX = %f, oldX=%f,newX=%f\n",newPoint.getX()-point.getX(), newPoint.getX(),point.getX());
+				printf("unconservative X, diffX = %f, oldX=%f,newX=%f\n", newPoint.getX() - point.getX(), newPoint.getX(), point.getX());
 			}
 			if (newPoint.getY() < point.getY())
 			{
-				printf("unconservative Y, diffY = %f, oldY=%f,newY=%f\n",newPoint.getY()-point.getY(), newPoint.getY(),point.getY());
+				printf("unconservative Y, diffY = %f, oldY=%f,newY=%f\n", newPoint.getY() - point.getY(), newPoint.getY(), point.getY());
 			}
 			if (newPoint.getZ() < point.getZ())
 			{
-
-				printf("unconservative Z, diffZ = %f, oldZ=%f,newZ=%f\n",newPoint.getZ()-point.getZ(), newPoint.getZ(),point.getZ());
+				printf("unconservative Z, diffZ = %f, oldZ=%f,newZ=%f\n", newPoint.getZ() - point.getZ(), newPoint.getZ(), point.getZ());
 			}
-		} else
+		}
+		else
 		{
 			if (newPoint.getX() > point.getX())
 			{
-				printf("unconservative X, diffX = %f, oldX=%f,newX=%f\n",newPoint.getX()-point.getX(), newPoint.getX(),point.getX());
+				printf("unconservative X, diffX = %f, oldX=%f,newX=%f\n", newPoint.getX() - point.getX(), newPoint.getX(), point.getX());
 			}
 			if (newPoint.getY() > point.getY())
 			{
-				printf("unconservative Y, diffY = %f, oldY=%f,newY=%f\n",newPoint.getY()-point.getY(), newPoint.getY(),point.getY());
+				printf("unconservative Y, diffY = %f, oldY=%f,newY=%f\n", newPoint.getY() - point.getY(), newPoint.getY(), point.getY());
 			}
 			if (newPoint.getZ() > point.getZ())
 			{
-				printf("unconservative Z, diffZ = %f, oldZ=%f,newZ=%f\n",newPoint.getZ()-point.getZ(), newPoint.getZ(),point.getZ());
+				printf("unconservative Z, diffZ = %f, oldZ=%f,newZ=%f\n", newPoint.getZ() - point.getZ(), newPoint.getZ(), point.getZ());
 			}
 		}
-#endif //DEBUG_CHECK_DEQUANTIZATION
-
+#endif  //DEBUG_CHECK_DEQUANTIZATION
 	}
 
-
-	SIMD_FORCE_INLINE void quantizeWithClamp(unsigned short* out, const btVector3& point2,int isMax) const
+	SIMD_FORCE_INLINE void quantizeWithClamp(unsigned short* out, const btVector3& point2, int isMax) const
 	{
-
 		btAssert(m_useQuantization);
 
 		btVector3 clampedPoint(point2);
 		clampedPoint.setMax(m_bvhAabbMin);
 		clampedPoint.setMin(m_bvhAabbMax);
 
-		quantize(out,clampedPoint,isMax);
-
+		quantize(out, clampedPoint, isMax);
 	}
-	
-	SIMD_FORCE_INLINE btVector3	unQuantize(const unsigned short* vecIn) const
+
+	SIMD_FORCE_INLINE btVector3 unQuantize(const unsigned short* vecIn) const
 	{
-			btVector3	vecOut;
-			vecOut.setValue(
+		btVector3 vecOut;
+		vecOut.setValue(
 			(btScalar)(vecIn[0]) / (m_bvhQuantization.getX()),
 			(btScalar)(vecIn[1]) / (m_bvhQuantization.getY()),
 			(btScalar)(vecIn[2]) / (m_bvhQuantization.getZ()));
-			vecOut += m_bvhAabbMin;
-			return vecOut;
+		vecOut += m_bvhAabbMin;
+		return vecOut;
 	}
 
 	///setTraversalMode let's you choose between stackless, recursive or stackless cache friendly tree traversal. Note this is only implemented for quantized trees.
-	void	setTraversalMode(btTraversalMode	traversalMode)
+	void setTraversalMode(btTraversalMode traversalMode)
 	{
 		m_traversalMode = traversalMode;
 	}
 
-
-	SIMD_FORCE_INLINE QuantizedNodeArray&	getQuantizedNodeArray()
-	{	
-		return	m_quantizedContiguousNodes;
+	SIMD_FORCE_INLINE QuantizedNodeArray& getQuantizedNodeArray()
+	{
+		return m_quantizedContiguousNodes;
 	}
 
-
-	SIMD_FORCE_INLINE BvhSubtreeInfoArray&	getSubtreeInfoArray()
+	SIMD_FORCE_INLINE BvhSubtreeInfoArray& getSubtreeInfoArray()
 	{
 		return m_SubtreeHeaders;
 	}
 
-////////////////////////////////////////////////////////////////////
+	////////////////////////////////////////////////////////////////////
 
 	/////Calculate space needed to store BVH for serialization
 	unsigned calculateSerializeBufferSize() const;
 
 	/// Data buffer MUST be 16 byte aligned
-	virtual bool serialize(void *o_alignedDataBuffer, unsigned i_dataBufferSize, bool i_swapEndian) const;
+	virtual bool serialize(void* o_alignedDataBuffer, unsigned i_dataBufferSize, bool i_swapEndian) const;
 
 	///deSerializeInPlace loads and initializes a BVH from a buffer in memory 'in place'
-	static btQuantizedBvh *deSerializeInPlace(void *i_alignedDataBuffer, unsigned int i_dataBufferSize, bool i_swapEndian);
+	static btQuantizedBvh* deSerializeInPlace(void* i_alignedDataBuffer, unsigned int i_dataBufferSize, bool i_swapEndian);
 
 	static unsigned int getAlignmentSerializationPadding();
-//////////////////////////////////////////////////////////////////////
+	//////////////////////////////////////////////////////////////////////
 
-	
-	virtual	int	calculateSerializeBufferSizeNew() const;
+	virtual int calculateSerializeBufferSizeNew() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 
-	virtual	void deSerializeFloat(struct btQuantizedBvhFloatData& quantizedBvhFloatData);
+	virtual void deSerializeFloat(struct btQuantizedBvhFloatData & quantizedBvhFloatData);
 
-	virtual	void deSerializeDouble(struct btQuantizedBvhDoubleData& quantizedBvhDoubleData);
+	virtual void deSerializeDouble(struct btQuantizedBvhDoubleData & quantizedBvhDoubleData);
 
-
-////////////////////////////////////////////////////////////////////
+	////////////////////////////////////////////////////////////////////
 
 	SIMD_FORCE_INLINE bool isQuantized()
 	{
@@ -494,38 +459,37 @@ private:
 	// Special "copy" constructor that allows for in-place deserialization
 	// Prevents btVector3's default constructor from being called, but doesn't inialize much else
 	// ownsMemory should most likely be false if deserializing, and if you are not, don't call this (it also changes the function signature, which we need)
-	btQuantizedBvh(btQuantizedBvh &other, bool ownsMemory);
+	btQuantizedBvh(btQuantizedBvh & other, bool ownsMemory);
+};
 
-}
-;
-
-
-struct	btBvhSubtreeInfoData
+// clang-format off
+// parser needs * with the name
+struct btBvhSubtreeInfoData
 {
-	int			m_rootNodeIndex;
-	int			m_subtreeSize;
+	int m_rootNodeIndex;
+	int m_subtreeSize;
 	unsigned short m_quantizedAabbMin[3];
 	unsigned short m_quantizedAabbMax[3];
 };
 
 struct btOptimizedBvhNodeFloatData
 {
-	btVector3FloatData	m_aabbMinOrg;
-	btVector3FloatData	m_aabbMaxOrg;
-	int	m_escapeIndex;
-	int	m_subPart;
-	int	m_triangleIndex;
+	btVector3FloatData m_aabbMinOrg;
+	btVector3FloatData m_aabbMaxOrg;
+	int m_escapeIndex;
+	int m_subPart;
+	int m_triangleIndex;
 	char m_pad[4];
 };
 
 struct btOptimizedBvhNodeDoubleData
 {
-	btVector3DoubleData	m_aabbMinOrg;
-	btVector3DoubleData	m_aabbMaxOrg;
-	int	m_escapeIndex;
-	int	m_subPart;
-	int	m_triangleIndex;
-	char	m_pad[4];
+	btVector3DoubleData m_aabbMinOrg;
+	btVector3DoubleData m_aabbMaxOrg;
+	int m_escapeIndex;
+	int m_subPart;
+	int m_triangleIndex;
+	char m_pad[4];
 };
 
 
@@ -569,13 +533,11 @@ struct	btQuantizedBvhDoubleData
 	int							m_numSubtreeHeaders;
 	btBvhSubtreeInfoData		*m_subTreeInfoPtr;
 };
+// clang-format on
 
-
-SIMD_FORCE_INLINE	int	btQuantizedBvh::calculateSerializeBufferSizeNew() const
+SIMD_FORCE_INLINE int btQuantizedBvh::calculateSerializeBufferSizeNew() const
 {
 	return sizeof(btQuantizedBvhData);
 }
 
-
-
-#endif //BT_QUANTIZED_BVH_H
+#endif  //BT_QUANTIZED_BVH_H
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btSimpleBroadphase.cpp b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btSimpleBroadphase.cpp
index 5f89f960e81..166cf771fe5 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btSimpleBroadphase.cpp
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btSimpleBroadphase.cpp
@@ -24,50 +24,45 @@ subject to the following restrictions:
 
 #include <new>
 
-void	btSimpleBroadphase::validate()
+void btSimpleBroadphase::validate()
 {
-	for (int i=0;i<m_numHandles;i++)
+	for (int i = 0; i < m_numHandles; i++)
 	{
-		for (int j=i+1;j<m_numHandles;j++)
+		for (int j = i + 1; j < m_numHandles; j++)
 		{
 			btAssert(&m_pHandles[i] != &m_pHandles[j]);
 		}
 	}
-	
 }
 
 btSimpleBroadphase::btSimpleBroadphase(int maxProxies, btOverlappingPairCache* overlappingPairCache)
-	:m_pairCache(overlappingPairCache),
-	m_ownsPairCache(false),
-	m_invalidPair(0)
+	: m_pairCache(overlappingPairCache),
+	  m_ownsPairCache(false),
+	  m_invalidPair(0)
 {
-
 	if (!overlappingPairCache)
 	{
-		void* mem = btAlignedAlloc(sizeof(btHashedOverlappingPairCache),16);
-		m_pairCache = new (mem)btHashedOverlappingPairCache();
+		void* mem = btAlignedAlloc(sizeof(btHashedOverlappingPairCache), 16);
+		m_pairCache = new (mem) btHashedOverlappingPairCache();
 		m_ownsPairCache = true;
 	}
 
 	// allocate handles buffer and put all handles on free list
-	m_pHandlesRawPtr = btAlignedAlloc(sizeof(btSimpleBroadphaseProxy)*maxProxies,16);
-	m_pHandles = new(m_pHandlesRawPtr) btSimpleBroadphaseProxy[maxProxies];
+	m_pHandlesRawPtr = btAlignedAlloc(sizeof(btSimpleBroadphaseProxy) * maxProxies, 16);
+	m_pHandles = new (m_pHandlesRawPtr) btSimpleBroadphaseProxy[maxProxies];
 	m_maxHandles = maxProxies;
 	m_numHandles = 0;
 	m_firstFreeHandle = 0;
 	m_LastHandleIndex = -1;
-	
 
 	{
 		for (int i = m_firstFreeHandle; i < maxProxies; i++)
 		{
 			m_pHandles[i].SetNextFree(i + 1);
-			m_pHandles[i].m_uniqueId = i+2;//any UID will do, we just avoid too trivial values (0,1) for debugging purposes
+			m_pHandles[i].m_uniqueId = i + 2;  //any UID will do, we just avoid too trivial values (0,1) for debugging purposes
 		}
 		m_pHandles[maxProxies - 1].SetNextFree(0);
-	
 	}
-
 }
 
 btSimpleBroadphase::~btSimpleBroadphase()
@@ -81,26 +76,25 @@ btSimpleBroadphase::~btSimpleBroadphase()
 	}
 }
 
-
-btBroadphaseProxy*	btSimpleBroadphase::createProxy(  const btVector3& aabbMin,  const btVector3& aabbMax,int shapeType,void* userPtr , int collisionFilterGroup, int collisionFilterMask, btDispatcher* /*dispatcher*/)
+btBroadphaseProxy* btSimpleBroadphase::createProxy(const btVector3& aabbMin, const btVector3& aabbMax, int shapeType, void* userPtr, int collisionFilterGroup, int collisionFilterMask, btDispatcher* /*dispatcher*/)
 {
 	if (m_numHandles >= m_maxHandles)
 	{
 		btAssert(0);
-		return 0; //should never happen, but don't let the game crash ;-)
+		return 0;  //should never happen, but don't let the game crash ;-)
 	}
-	btAssert(aabbMin[0]<= aabbMax[0] && aabbMin[1]<= aabbMax[1] && aabbMin[2]<= aabbMax[2]);
+	btAssert(aabbMin[0] <= aabbMax[0] && aabbMin[1] <= aabbMax[1] && aabbMin[2] <= aabbMax[2]);
 
 	int newHandleIndex = allocHandle();
-	btSimpleBroadphaseProxy* proxy = new (&m_pHandles[newHandleIndex])btSimpleBroadphaseProxy(aabbMin,aabbMax,shapeType,userPtr,collisionFilterGroup,collisionFilterMask);
+	btSimpleBroadphaseProxy* proxy = new (&m_pHandles[newHandleIndex]) btSimpleBroadphaseProxy(aabbMin, aabbMax, shapeType, userPtr, collisionFilterGroup, collisionFilterMask);
 
 	return proxy;
 }
 
-class	RemovingOverlapCallback : public btOverlapCallback
+class RemovingOverlapCallback : public btOverlapCallback
 {
 protected:
-	virtual bool	processOverlap(btBroadphasePair& pair)
+	virtual bool processOverlap(btBroadphasePair& pair)
 	{
 		(void)pair;
 		btAssert(0);
@@ -110,12 +104,13 @@ protected:
 
 class RemovePairContainingProxy
 {
+	btBroadphaseProxy* m_targetProxy;
 
-	btBroadphaseProxy*	m_targetProxy;
-	public:
+public:
 	virtual ~RemovePairContainingProxy()
 	{
 	}
+
 protected:
 	virtual bool processOverlap(btBroadphasePair& pair)
 	{
@@ -126,38 +121,36 @@ protected:
 	};
 };
 
-void	btSimpleBroadphase::destroyProxy(btBroadphaseProxy* proxyOrg,btDispatcher* dispatcher)
+void btSimpleBroadphase::destroyProxy(btBroadphaseProxy* proxyOrg, btDispatcher* dispatcher)
 {
-		
-		btSimpleBroadphaseProxy* proxy0 = static_cast<btSimpleBroadphaseProxy*>(proxyOrg);
-		freeHandle(proxy0);
+	btSimpleBroadphaseProxy* proxy0 = static_cast<btSimpleBroadphaseProxy*>(proxyOrg);
+	freeHandle(proxy0);
 
-		m_pairCache->removeOverlappingPairsContainingProxy(proxyOrg,dispatcher);
+	m_pairCache->removeOverlappingPairsContainingProxy(proxyOrg, dispatcher);
 
-		//validate();
-		
+	//validate();
 }
 
-void	btSimpleBroadphase::getAabb(btBroadphaseProxy* proxy,btVector3& aabbMin, btVector3& aabbMax ) const
+void btSimpleBroadphase::getAabb(btBroadphaseProxy* proxy, btVector3& aabbMin, btVector3& aabbMax) const
 {
 	const btSimpleBroadphaseProxy* sbp = getSimpleProxyFromProxy(proxy);
 	aabbMin = sbp->m_aabbMin;
 	aabbMax = sbp->m_aabbMax;
 }
 
-void	btSimpleBroadphase::setAabb(btBroadphaseProxy* proxy,const btVector3& aabbMin,const btVector3& aabbMax, btDispatcher* /*dispatcher*/)
+void btSimpleBroadphase::setAabb(btBroadphaseProxy* proxy, const btVector3& aabbMin, const btVector3& aabbMax, btDispatcher* /*dispatcher*/)
 {
 	btSimpleBroadphaseProxy* sbp = getSimpleProxyFromProxy(proxy);
 	sbp->m_aabbMin = aabbMin;
 	sbp->m_aabbMax = aabbMax;
 }
 
-void	btSimpleBroadphase::rayTest(const btVector3& rayFrom,const btVector3& rayTo, btBroadphaseRayCallback& rayCallback, const btVector3& aabbMin,const btVector3& aabbMax)
+void btSimpleBroadphase::rayTest(const btVector3& rayFrom, const btVector3& rayTo, btBroadphaseRayCallback& rayCallback, const btVector3& aabbMin, const btVector3& aabbMax)
 {
-	for (int i=0; i <= m_LastHandleIndex; i++)
+	for (int i = 0; i <= m_LastHandleIndex; i++)
 	{
 		btSimpleBroadphaseProxy* proxy = &m_pHandles[i];
-		if(!proxy->m_clientObject)
+		if (!proxy->m_clientObject)
 		{
 			continue;
 		}
@@ -165,69 +158,59 @@ void	btSimpleBroadphase::rayTest(const btVector3& rayFrom,const btVector3& rayTo
 	}
 }
 
-
-void	btSimpleBroadphase::aabbTest(const btVector3& aabbMin, const btVector3& aabbMax, btBroadphaseAabbCallback& callback)
+void btSimpleBroadphase::aabbTest(const btVector3& aabbMin, const btVector3& aabbMax, btBroadphaseAabbCallback& callback)
 {
-	for (int i=0; i <= m_LastHandleIndex; i++)
+	for (int i = 0; i <= m_LastHandleIndex; i++)
 	{
 		btSimpleBroadphaseProxy* proxy = &m_pHandles[i];
-		if(!proxy->m_clientObject)
+		if (!proxy->m_clientObject)
 		{
 			continue;
 		}
-		if (TestAabbAgainstAabb2(aabbMin,aabbMax,proxy->m_aabbMin,proxy->m_aabbMax))
+		if (TestAabbAgainstAabb2(aabbMin, aabbMax, proxy->m_aabbMin, proxy->m_aabbMax))
 		{
 			callback.process(proxy);
 		}
 	}
 }
 
-
-
-	
-
-
-
-bool	btSimpleBroadphase::aabbOverlap(btSimpleBroadphaseProxy* proxy0,btSimpleBroadphaseProxy* proxy1)
+bool btSimpleBroadphase::aabbOverlap(btSimpleBroadphaseProxy* proxy0, btSimpleBroadphaseProxy* proxy1)
 {
-	return proxy0->m_aabbMin[0] <= proxy1->m_aabbMax[0] && proxy1->m_aabbMin[0] <= proxy0->m_aabbMax[0] && 
+	return proxy0->m_aabbMin[0] <= proxy1->m_aabbMax[0] && proxy1->m_aabbMin[0] <= proxy0->m_aabbMax[0] &&
 		   proxy0->m_aabbMin[1] <= proxy1->m_aabbMax[1] && proxy1->m_aabbMin[1] <= proxy0->m_aabbMax[1] &&
 		   proxy0->m_aabbMin[2] <= proxy1->m_aabbMax[2] && proxy1->m_aabbMin[2] <= proxy0->m_aabbMax[2];
-
 }
 
-
-
 //then remove non-overlapping ones
 class CheckOverlapCallback : public btOverlapCallback
 {
 public:
 	virtual bool processOverlap(btBroadphasePair& pair)
 	{
-		return (!btSimpleBroadphase::aabbOverlap(static_cast<btSimpleBroadphaseProxy*>(pair.m_pProxy0),static_cast<btSimpleBroadphaseProxy*>(pair.m_pProxy1)));
+		return (!btSimpleBroadphase::aabbOverlap(static_cast<btSimpleBroadphaseProxy*>(pair.m_pProxy0), static_cast<btSimpleBroadphaseProxy*>(pair.m_pProxy1)));
 	}
 };
 
-void	btSimpleBroadphase::calculateOverlappingPairs(btDispatcher* dispatcher)
+void btSimpleBroadphase::calculateOverlappingPairs(btDispatcher* dispatcher)
 {
 	//first check for new overlapping pairs
-	int i,j;
+	int i, j;
 	if (m_numHandles >= 0)
 	{
 		int new_largest_index = -1;
-		for (i=0; i <= m_LastHandleIndex; i++)
+		for (i = 0; i <= m_LastHandleIndex; i++)
 		{
 			btSimpleBroadphaseProxy* proxy0 = &m_pHandles[i];
-			if(!proxy0->m_clientObject)
+			if (!proxy0->m_clientObject)
 			{
 				continue;
 			}
 			new_largest_index = i;
-			for (j=i+1; j <= m_LastHandleIndex; j++)
+			for (j = i + 1; j <= m_LastHandleIndex; j++)
 			{
 				btSimpleBroadphaseProxy* proxy1 = &m_pHandles[j];
 				btAssert(proxy0 != proxy1);
-				if(!proxy1->m_clientObject)
+				if (!proxy1->m_clientObject)
 				{
 					continue;
 				}
@@ -235,19 +218,20 @@ void	btSimpleBroadphase::calculateOverlappingPairs(btDispatcher* dispatcher)
 				btSimpleBroadphaseProxy* p0 = getSimpleProxyFromProxy(proxy0);
 				btSimpleBroadphaseProxy* p1 = getSimpleProxyFromProxy(proxy1);
 
-				if (aabbOverlap(p0,p1))
+				if (aabbOverlap(p0, p1))
 				{
-					if ( !m_pairCache->findPair(proxy0,proxy1))
+					if (!m_pairCache->findPair(proxy0, proxy1))
 					{
-						m_pairCache->addOverlappingPair(proxy0,proxy1);
+						m_pairCache->addOverlappingPair(proxy0, proxy1);
 					}
-				} else
+				}
+				else
 				{
 					if (!m_pairCache->hasDeferredRemoval())
 					{
-						if ( m_pairCache->findPair(proxy0,proxy1))
+						if (m_pairCache->findPair(proxy0, proxy1))
 						{
-							m_pairCache->removeOverlappingPair(proxy0,proxy1,dispatcher);
+							m_pairCache->removeOverlappingPair(proxy0, proxy1, dispatcher);
 						}
 					}
 				}
@@ -258,8 +242,7 @@ void	btSimpleBroadphase::calculateOverlappingPairs(btDispatcher* dispatcher)
 
 		if (m_ownsPairCache && m_pairCache->hasDeferredRemoval())
 		{
-
-			btBroadphasePairArray&	overlappingPairArray = m_pairCache->getOverlappingPairArray();
+			btBroadphasePairArray& overlappingPairArray = m_pairCache->getOverlappingPairArray();
 
 			//perform a sort, to find duplicates and to sort 'invalid' pairs to the end
 			overlappingPairArray.quickSort(btBroadphasePairSortPredicate());
@@ -267,16 +250,13 @@ void	btSimpleBroadphase::calculateOverlappingPairs(btDispatcher* dispatcher)
 			overlappingPairArray.resize(overlappingPairArray.size() - m_invalidPair);
 			m_invalidPair = 0;
 
-
 			btBroadphasePair previousPair;
 			previousPair.m_pProxy0 = 0;
 			previousPair.m_pProxy1 = 0;
 			previousPair.m_algorithm = 0;
 
-
-			for (i=0;i<overlappingPairArray.size();i++)
+			for (i = 0; i < overlappingPairArray.size(); i++)
 			{
-
 				btBroadphasePair& pair = overlappingPairArray[i];
 
 				bool isDuplicate = (pair == previousPair);
@@ -287,16 +267,18 @@ void	btSimpleBroadphase::calculateOverlappingPairs(btDispatcher* dispatcher)
 
 				if (!isDuplicate)
 				{
-					bool hasOverlap = testAabbOverlap(pair.m_pProxy0,pair.m_pProxy1);
+					bool hasOverlap = testAabbOverlap(pair.m_pProxy0, pair.m_pProxy1);
 
 					if (hasOverlap)
 					{
-						needsRemoval = false;//callback->processOverlap(pair);
-					} else
+						needsRemoval = false;  //callback->processOverlap(pair);
+					}
+					else
 					{
 						needsRemoval = true;
 					}
-				} else
+				}
+				else
 				{
 					//remove duplicate
 					needsRemoval = true;
@@ -306,7 +288,7 @@ void	btSimpleBroadphase::calculateOverlappingPairs(btDispatcher* dispatcher)
 
 				if (needsRemoval)
 				{
-					m_pairCache->cleanOverlappingPair(pair,dispatcher);
+					m_pairCache->cleanOverlappingPair(pair, dispatcher);
 
 					//		m_overlappingPairArray.swap(i,m_overlappingPairArray.size()-1);
 					//		m_overlappingPairArray.pop_back();
@@ -314,7 +296,6 @@ void	btSimpleBroadphase::calculateOverlappingPairs(btDispatcher* dispatcher)
 					pair.m_pProxy1 = 0;
 					m_invalidPair++;
 				}
-
 			}
 
 			///if you don't like to skip the invalid pairs in the array, execute following code:
@@ -326,21 +307,19 @@ void	btSimpleBroadphase::calculateOverlappingPairs(btDispatcher* dispatcher)
 
 			overlappingPairArray.resize(overlappingPairArray.size() - m_invalidPair);
 			m_invalidPair = 0;
-#endif//CLEAN_INVALID_PAIRS
-
+#endif  //CLEAN_INVALID_PAIRS
 		}
 	}
 }
 
-
-bool btSimpleBroadphase::testAabbOverlap(btBroadphaseProxy* proxy0,btBroadphaseProxy* proxy1)
+bool btSimpleBroadphase::testAabbOverlap(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1)
 {
 	btSimpleBroadphaseProxy* p0 = getSimpleProxyFromProxy(proxy0);
 	btSimpleBroadphaseProxy* p1 = getSimpleProxyFromProxy(proxy1);
-	return aabbOverlap(p0,p1);
+	return aabbOverlap(p0, p1);
 }
 
-void	btSimpleBroadphase::resetPool(btDispatcher* dispatcher)
+void btSimpleBroadphase::resetPool(btDispatcher* dispatcher)
 {
 	//not yet
 }
diff --git a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btSimpleBroadphase.h b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btSimpleBroadphase.h
index d7a18e400a4..3e02fdc0036 100644
--- a/thirdparty/bullet/BulletCollision/BroadphaseCollision/btSimpleBroadphase.h
+++ b/thirdparty/bullet/BulletCollision/BroadphaseCollision/btSimpleBroadphase.h
@@ -16,57 +16,47 @@ subject to the following restrictions:
 #ifndef BT_SIMPLE_BROADPHASE_H
 #define BT_SIMPLE_BROADPHASE_H
 
-
 #include "btOverlappingPairCache.h"
 
-
 struct btSimpleBroadphaseProxy : public btBroadphaseProxy
 {
-	int			m_nextFree;
-	
-//	int			m_handleId;
+	int m_nextFree;
 
-	
-	btSimpleBroadphaseProxy() {};
+	//	int			m_handleId;
 
-	btSimpleBroadphaseProxy(const btVector3& minpt,const btVector3& maxpt,int shapeType,void* userPtr, int collisionFilterGroup, int collisionFilterMask)
-	:btBroadphaseProxy(minpt,maxpt,userPtr,collisionFilterGroup,collisionFilterMask)
+	btSimpleBroadphaseProxy(){};
+
+	btSimpleBroadphaseProxy(const btVector3& minpt, const btVector3& maxpt, int shapeType, void* userPtr, int collisionFilterGroup, int collisionFilterMask)
+		: btBroadphaseProxy(minpt, maxpt, userPtr, collisionFilterGroup, collisionFilterMask)
 	{
 		(void)shapeType;
 	}
-	
-	
-	SIMD_FORCE_INLINE void SetNextFree(int next) {m_nextFree = next;}
-	SIMD_FORCE_INLINE int GetNextFree() const {return m_nextFree;}
-
-	
-
 
+	SIMD_FORCE_INLINE void SetNextFree(int next) { m_nextFree = next; }
+	SIMD_FORCE_INLINE int GetNextFree() const { return m_nextFree; }
 };
 
 ///The SimpleBroadphase is just a unit-test for btAxisSweep3, bt32BitAxisSweep3, or btDbvtBroadphase, so use those classes instead.
 ///It is a brute force aabb culling broadphase based on O(n^2) aabb checks
 class btSimpleBroadphase : public btBroadphaseInterface
 {
-
 protected:
+	int m_numHandles;  // number of active handles
+	int m_maxHandles;  // max number of handles
+	int m_LastHandleIndex;
 
-	int		m_numHandles;						// number of active handles
-	int		m_maxHandles;						// max number of handles
-	int		m_LastHandleIndex;							
-	
-	btSimpleBroadphaseProxy* m_pHandles;						// handles pool
+	btSimpleBroadphaseProxy* m_pHandles;  // handles pool
 
 	void* m_pHandlesRawPtr;
-	int		m_firstFreeHandle;		// free handles list
-	
+	int m_firstFreeHandle;  // free handles list
+
 	int allocHandle()
 	{
 		btAssert(m_numHandles < m_maxHandles);
 		int freeHandle = m_firstFreeHandle;
 		m_firstFreeHandle = m_pHandles[freeHandle].GetNextFree();
 		m_numHandles++;
-		if(freeHandle > m_LastHandleIndex)
+		if (freeHandle > m_LastHandleIndex)
 		{
 			m_LastHandleIndex = freeHandle;
 		}
@@ -75,9 +65,9 @@ protected:
 
 	void freeHandle(btSimpleBroadphaseProxy* proxy)
 	{
-		int handle = int(proxy-m_pHandles);
+		int handle = int(proxy - m_pHandles);
 		btAssert(handle >= 0 && handle < m_maxHandles);
-		if(handle == m_LastHandleIndex)
+		if (handle == m_LastHandleIndex)
 		{
 			m_LastHandleIndex--;
 		}
@@ -89,20 +79,18 @@ protected:
 		m_numHandles--;
 	}
 
-	btOverlappingPairCache*	m_pairCache;
-	bool	m_ownsPairCache;
+	btOverlappingPairCache* m_pairCache;
+	bool m_ownsPairCache;
 
-	int	m_invalidPair;
+	int m_invalidPair;
 
-	
-	
-	inline btSimpleBroadphaseProxy*	getSimpleProxyFromProxy(btBroadphaseProxy* proxy)
+	inline btSimpleBroadphaseProxy* getSimpleProxyFromProxy(btBroadphaseProxy* proxy)
 	{
 		btSimpleBroadphaseProxy* proxy0 = static_cast<btSimpleBroadphaseProxy*>(proxy);
 		return proxy0;
 	}
 
-	inline const btSimpleBroadphaseProxy*	getSimpleProxyFromProxy(btBroadphaseProxy* proxy) const
+	inline const btSimpleBroadphaseProxy* getSimpleProxyFromProxy(btBroadphaseProxy* proxy) const
 	{
 		const btSimpleBroadphaseProxy* proxy0 = static_cast<const btSimpleBroadphaseProxy*>(proxy);
 		return proxy0;
@@ -111,61 +99,50 @@ protected:
 	///reset broadphase internal structures, to ensure determinism/reproducability
 	virtual void resetPool(btDispatcher* dispatcher);
 
-
-	void	validate();
+	void validate();
 
 protected:
-
-
-	
-
 public:
-	btSimpleBroadphase(int maxProxies=16384,btOverlappingPairCache* overlappingPairCache=0);
+	btSimpleBroadphase(int maxProxies = 16384, btOverlappingPairCache* overlappingPairCache = 0);
 	virtual ~btSimpleBroadphase();
 
+	static bool aabbOverlap(btSimpleBroadphaseProxy* proxy0, btSimpleBroadphaseProxy* proxy1);
 
-		static bool	aabbOverlap(btSimpleBroadphaseProxy* proxy0,btSimpleBroadphaseProxy* proxy1);
+	virtual btBroadphaseProxy* createProxy(const btVector3& aabbMin, const btVector3& aabbMax, int shapeType, void* userPtr, int collisionFilterGroup, int collisionFilterMask, btDispatcher* dispatcher);
 
+	virtual void calculateOverlappingPairs(btDispatcher* dispatcher);
 
-	virtual btBroadphaseProxy*	createProxy(  const btVector3& aabbMin,  const btVector3& aabbMax,int shapeType,void* userPtr , int collisionFilterGroup, int collisionFilterMask, btDispatcher* dispatcher);
+	virtual void destroyProxy(btBroadphaseProxy* proxy, btDispatcher* dispatcher);
+	virtual void setAabb(btBroadphaseProxy* proxy, const btVector3& aabbMin, const btVector3& aabbMax, btDispatcher* dispatcher);
+	virtual void getAabb(btBroadphaseProxy* proxy, btVector3& aabbMin, btVector3& aabbMax) const;
 
-	virtual void	calculateOverlappingPairs(btDispatcher* dispatcher);
+	virtual void rayTest(const btVector3& rayFrom, const btVector3& rayTo, btBroadphaseRayCallback& rayCallback, const btVector3& aabbMin = btVector3(0, 0, 0), const btVector3& aabbMax = btVector3(0, 0, 0));
+	virtual void aabbTest(const btVector3& aabbMin, const btVector3& aabbMax, btBroadphaseAabbCallback& callback);
 
-	virtual void	destroyProxy(btBroadphaseProxy* proxy,btDispatcher* dispatcher);
-	virtual void	setAabb(btBroadphaseProxy* proxy,const btVector3& aabbMin,const btVector3& aabbMax, btDispatcher* dispatcher);
-	virtual void	getAabb(btBroadphaseProxy* proxy,btVector3& aabbMin, btVector3& aabbMax ) const;
-
-	virtual void	rayTest(const btVector3& rayFrom,const btVector3& rayTo, btBroadphaseRayCallback& rayCallback, const btVector3& aabbMin=btVector3(0,0,0),const btVector3& aabbMax=btVector3(0,0,0));
-	virtual void	aabbTest(const btVector3& aabbMin, const btVector3& aabbMax, btBroadphaseAabbCallback& callback);
-		
-	btOverlappingPairCache*	getOverlappingPairCache()
+	btOverlappingPairCache* getOverlappingPairCache()
 	{
 		return m_pairCache;
 	}
-	const btOverlappingPairCache*	getOverlappingPairCache() const
+	const btOverlappingPairCache* getOverlappingPairCache() const
 	{
 		return m_pairCache;
 	}
 
-	bool	testAabbOverlap(btBroadphaseProxy* proxy0,btBroadphaseProxy* proxy1);
-
+	bool testAabbOverlap(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1);
 
 	///getAabb returns the axis aligned bounding box in the 'global' coordinate frame
 	///will add some transform later
-	virtual void getBroadphaseAabb(btVector3& aabbMin,btVector3& aabbMax) const
+	virtual void getBroadphaseAabb(btVector3& aabbMin, btVector3& aabbMax) const
 	{
-		aabbMin.setValue(-BT_LARGE_FLOAT,-BT_LARGE_FLOAT,-BT_LARGE_FLOAT);
-		aabbMax.setValue(BT_LARGE_FLOAT,BT_LARGE_FLOAT,BT_LARGE_FLOAT);
+		aabbMin.setValue(-BT_LARGE_FLOAT, -BT_LARGE_FLOAT, -BT_LARGE_FLOAT);
+		aabbMax.setValue(BT_LARGE_FLOAT, BT_LARGE_FLOAT, BT_LARGE_FLOAT);
 	}
 
-	virtual void	printStats()
+	virtual void printStats()
 	{
-//		printf("btSimpleBroadphase.h\n");
-//		printf("numHandles = %d, maxHandles = %d\n",m_numHandles,m_maxHandles);
+		//		printf("btSimpleBroadphase.h\n");
+		//		printf("numHandles = %d, maxHandles = %d\n",m_numHandles,m_maxHandles);
 	}
 };
 
-
-
-#endif //BT_SIMPLE_BROADPHASE_H
-
+#endif  //BT_SIMPLE_BROADPHASE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/SphereTriangleDetector.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/SphereTriangleDetector.cpp
index e5bac8438ed..7647f673600 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/SphereTriangleDetector.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/SphereTriangleDetector.cpp
@@ -18,94 +18,95 @@ subject to the following restrictions:
 #include "BulletCollision/CollisionShapes/btTriangleShape.h"
 #include "BulletCollision/CollisionShapes/btSphereShape.h"
 
-
-SphereTriangleDetector::SphereTriangleDetector(btSphereShape* sphere,btTriangleShape* triangle,btScalar contactBreakingThreshold)
-:m_sphere(sphere),
-m_triangle(triangle),
-m_contactBreakingThreshold(contactBreakingThreshold)
+SphereTriangleDetector::SphereTriangleDetector(btSphereShape* sphere, btTriangleShape* triangle, btScalar contactBreakingThreshold)
+	: m_sphere(sphere),
+	  m_triangle(triangle),
+	  m_contactBreakingThreshold(contactBreakingThreshold)
 {
-
 }
 
-void	SphereTriangleDetector::getClosestPoints(const ClosestPointInput& input,Result& output,class btIDebugDraw* debugDraw,bool swapResults)
+void SphereTriangleDetector::getClosestPoints(const ClosestPointInput& input, Result& output, class btIDebugDraw* debugDraw, bool swapResults)
 {
-
 	(void)debugDraw;
 	const btTransform& transformA = input.m_transformA;
 	const btTransform& transformB = input.m_transformB;
 
-	btVector3 point,normal;
+	btVector3 point, normal;
 	btScalar timeOfImpact = btScalar(1.);
 	btScalar depth = btScalar(0.);
-//	output.m_distance = btScalar(BT_LARGE_FLOAT);
+	//	output.m_distance = btScalar(BT_LARGE_FLOAT);
 	//move sphere into triangle space
-	btTransform	sphereInTr = transformB.inverseTimes(transformA);
+	btTransform sphereInTr = transformB.inverseTimes(transformA);
 
-	if (collide(sphereInTr.getOrigin(),point,normal,depth,timeOfImpact,m_contactBreakingThreshold))
+	if (collide(sphereInTr.getOrigin(), point, normal, depth, timeOfImpact, m_contactBreakingThreshold))
 	{
 		if (swapResults)
 		{
-			btVector3 normalOnB = transformB.getBasis()*normal;
+			btVector3 normalOnB = transformB.getBasis() * normal;
 			btVector3 normalOnA = -normalOnB;
-			btVector3 pointOnA = transformB*point+normalOnB*depth;
-			output.addContactPoint(normalOnA,pointOnA,depth);
-		} else
+			btVector3 pointOnA = transformB * point + normalOnB * depth;
+			output.addContactPoint(normalOnA, pointOnA, depth);
+		}
+		else
 		{
-			output.addContactPoint(transformB.getBasis()*normal,transformB*point,depth);
+			output.addContactPoint(transformB.getBasis() * normal, transformB * point, depth);
 		}
 	}
-
 }
 
-
-
 // See also geometrictools.com
 // Basic idea: D = |p - (lo + t0*lv)| where t0 = lv . (p - lo) / lv . lv
-btScalar SegmentSqrDistance(const btVector3& from, const btVector3& to,const btVector3 &p, btVector3 &nearest);
+btScalar SegmentSqrDistance(const btVector3& from, const btVector3& to, const btVector3& p, btVector3& nearest);
 
-btScalar SegmentSqrDistance(const btVector3& from, const btVector3& to,const btVector3 &p, btVector3 &nearest) {
+btScalar SegmentSqrDistance(const btVector3& from, const btVector3& to, const btVector3& p, btVector3& nearest)
+{
 	btVector3 diff = p - from;
 	btVector3 v = to - from;
 	btScalar t = v.dot(diff);
-	
-	if (t > 0) {
+
+	if (t > 0)
+	{
 		btScalar dotVV = v.dot(v);
-		if (t < dotVV) {
+		if (t < dotVV)
+		{
 			t /= dotVV;
-			diff -= t*v;
-		} else {
+			diff -= t * v;
+		}
+		else
+		{
 			t = 1;
 			diff -= v;
 		}
-	} else
+	}
+	else
 		t = 0;
 
-	nearest = from + t*v;
-	return diff.dot(diff);	
+	nearest = from + t * v;
+	return diff.dot(diff);
 }
 
-bool SphereTriangleDetector::facecontains(const btVector3 &p,const btVector3* vertices,btVector3& normal)  {
+bool SphereTriangleDetector::facecontains(const btVector3& p, const btVector3* vertices, btVector3& normal)
+{
 	btVector3 lp(p);
 	btVector3 lnormal(normal);
-	
+
 	return pointInTriangle(vertices, lnormal, &lp);
 }
 
-bool SphereTriangleDetector::collide(const btVector3& sphereCenter,btVector3 &point, btVector3& resultNormal, btScalar& depth, btScalar &timeOfImpact, btScalar contactBreakingThreshold)
+bool SphereTriangleDetector::collide(const btVector3& sphereCenter, btVector3& point, btVector3& resultNormal, btScalar& depth, btScalar& timeOfImpact, btScalar contactBreakingThreshold)
 {
-
 	const btVector3* vertices = &m_triangle->getVertexPtr(0);
-	
+
 	btScalar radius = m_sphere->getRadius();
 	btScalar radiusWithThreshold = radius + contactBreakingThreshold;
 
-	btVector3 normal = (vertices[1]-vertices[0]).cross(vertices[2]-vertices[0]);
+	btVector3 normal = (vertices[1] - vertices[0]).cross(vertices[2] - vertices[0]);
 
 	btScalar l2 = normal.length2();
 	bool hasContact = false;
 	btVector3 contactPoint;
 
-	if (l2 >= SIMD_EPSILON*SIMD_EPSILON)
+	if (l2 >= SIMD_EPSILON * SIMD_EPSILON)
 	{
 		normal /= btSqrt(l2);
 
@@ -120,54 +121,59 @@ bool SphereTriangleDetector::collide(const btVector3& sphereCenter,btVector3 &po
 		}
 
 		bool isInsideContactPlane = distanceFromPlane < radiusWithThreshold;
-	
+
 		// Check for contact / intersection
-	
-		if (isInsideContactPlane) {
-			if (facecontains(sphereCenter, vertices, normal)) {
+
+		if (isInsideContactPlane)
+		{
+			if (facecontains(sphereCenter, vertices, normal))
+			{
 				// Inside the contact wedge - touches a point on the shell plane
 				hasContact = true;
-				contactPoint = sphereCenter - normal*distanceFromPlane;
+				contactPoint = sphereCenter - normal * distanceFromPlane;
 			}
-			else {
+			else
+			{
 				// Could be inside one of the contact capsules
-				btScalar contactCapsuleRadiusSqr = radiusWithThreshold*radiusWithThreshold;
+				btScalar contactCapsuleRadiusSqr = radiusWithThreshold * radiusWithThreshold;
 				btScalar minDistSqr = contactCapsuleRadiusSqr;
 				btVector3 nearestOnEdge;
-				for (int i = 0; i < m_triangle->getNumEdges(); i++) {
-
+				for (int i = 0; i < m_triangle->getNumEdges(); i++)
+				{
 					btVector3 pa;
 					btVector3 pb;
 
 					m_triangle->getEdge(i, pa, pb);
 
 					btScalar distanceSqr = SegmentSqrDistance(pa, pb, sphereCenter, nearestOnEdge);
-					if (distanceSqr < minDistSqr) {
+					if (distanceSqr < minDistSqr)
+					{
 						// Yep, we're inside a capsule, and record the capsule with smallest distance
 						minDistSqr = distanceSqr;
 						hasContact = true;
 						contactPoint = nearestOnEdge;
 					}
-
 				}
 			}
 		}
 	}
 
-	if (hasContact) {
+	if (hasContact)
+	{
 		btVector3 contactToCentre = sphereCenter - contactPoint;
 		btScalar distanceSqr = contactToCentre.length2();
 
-		if (distanceSqr < radiusWithThreshold*radiusWithThreshold)
+		if (distanceSqr < radiusWithThreshold * radiusWithThreshold)
 		{
-			if (distanceSqr>SIMD_EPSILON)
+			if (distanceSqr > SIMD_EPSILON)
 			{
 				btScalar distance = btSqrt(distanceSqr);
 				resultNormal = contactToCentre;
 				resultNormal.normalize();
 				point = contactPoint;
-				depth = -(radius-distance);
-			} else
+				depth = -(radius - distance);
+			}
+			else
 			{
 				resultNormal = normal;
 				point = contactPoint;
@@ -176,36 +182,34 @@ bool SphereTriangleDetector::collide(const btVector3& sphereCenter,btVector3 &po
 			return true;
 		}
 	}
-	
+
 	return false;
 }
 
-
-bool SphereTriangleDetector::pointInTriangle(const btVector3 vertices[], const btVector3 &normal, btVector3 *p )
+bool SphereTriangleDetector::pointInTriangle(const btVector3 vertices[], const btVector3& normal, btVector3* p)
 {
 	const btVector3* p1 = &vertices[0];
 	const btVector3* p2 = &vertices[1];
 	const btVector3* p3 = &vertices[2];
 
-	btVector3 edge1( *p2 - *p1 );
-	btVector3 edge2( *p3 - *p2 );
-	btVector3 edge3( *p1 - *p3 );
+	btVector3 edge1(*p2 - *p1);
+	btVector3 edge2(*p3 - *p2);
+	btVector3 edge3(*p1 - *p3);
 
-	btVector3 p1_to_p( *p - *p1 );
-	btVector3 p2_to_p( *p - *p2 );
-	btVector3 p3_to_p( *p - *p3 );
+	btVector3 p1_to_p(*p - *p1);
+	btVector3 p2_to_p(*p - *p2);
+	btVector3 p3_to_p(*p - *p3);
+
+	btVector3 edge1_normal(edge1.cross(normal));
+	btVector3 edge2_normal(edge2.cross(normal));
+	btVector3 edge3_normal(edge3.cross(normal));
 
-	btVector3 edge1_normal( edge1.cross(normal));
-	btVector3 edge2_normal( edge2.cross(normal));
-	btVector3 edge3_normal( edge3.cross(normal));
-	
 	btScalar r1, r2, r3;
-	r1 = edge1_normal.dot( p1_to_p );
-	r2 = edge2_normal.dot( p2_to_p );
-	r3 = edge3_normal.dot( p3_to_p );
-	if ( ( r1 > 0 && r2 > 0 && r3 > 0 ) ||
-	     ( r1 <= 0 && r2 <= 0 && r3 <= 0 ) )
+	r1 = edge1_normal.dot(p1_to_p);
+	r2 = edge2_normal.dot(p2_to_p);
+	r3 = edge3_normal.dot(p3_to_p);
+	if ((r1 > 0 && r2 > 0 && r3 > 0) ||
+		(r1 <= 0 && r2 <= 0 && r3 <= 0))
 		return true;
 	return false;
-
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/SphereTriangleDetector.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/SphereTriangleDetector.h
index 22953af43fd..d47e47530d4 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/SphereTriangleDetector.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/SphereTriangleDetector.h
@@ -18,34 +18,26 @@ subject to the following restrictions:
 
 #include "BulletCollision/NarrowPhaseCollision/btDiscreteCollisionDetectorInterface.h"
 
-
-
 class btSphereShape;
 class btTriangleShape;
 
-
-
 /// sphere-triangle to match the btDiscreteCollisionDetectorInterface
 struct SphereTriangleDetector : public btDiscreteCollisionDetectorInterface
 {
-	virtual void	getClosestPoints(const ClosestPointInput& input,Result& output,class btIDebugDraw* debugDraw,bool swapResults=false);
+	virtual void getClosestPoints(const ClosestPointInput& input, Result& output, class btIDebugDraw* debugDraw, bool swapResults = false);
 
-	SphereTriangleDetector(btSphereShape* sphere,btTriangleShape* triangle, btScalar contactBreakingThreshold);
+	SphereTriangleDetector(btSphereShape* sphere, btTriangleShape* triangle, btScalar contactBreakingThreshold);
 
-	virtual ~SphereTriangleDetector() {};
+	virtual ~SphereTriangleDetector(){};
 
-	bool collide(const btVector3& sphereCenter,btVector3 &point, btVector3& resultNormal, btScalar& depth, btScalar &timeOfImpact, btScalar	contactBreakingThreshold);
+	bool collide(const btVector3& sphereCenter, btVector3& point, btVector3& resultNormal, btScalar& depth, btScalar& timeOfImpact, btScalar contactBreakingThreshold);
 
 private:
-
-	
-	bool pointInTriangle(const btVector3 vertices[], const btVector3 &normal, btVector3 *p );
-	bool facecontains(const btVector3 &p,const btVector3* vertices,btVector3& normal);
+	bool pointInTriangle(const btVector3 vertices[], const btVector3& normal, btVector3* p);
+	bool facecontains(const btVector3& p, const btVector3* vertices, btVector3& normal);
 
 	btSphereShape* m_sphere;
 	btTriangleShape* m_triangle;
-	btScalar	m_contactBreakingThreshold;
-	
+	btScalar m_contactBreakingThreshold;
 };
-#endif //BT_SPHERE_TRIANGLE_DETECTOR_H
-
+#endif  //BT_SPHERE_TRIANGLE_DETECTOR_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btActivatingCollisionAlgorithm.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btActivatingCollisionAlgorithm.cpp
index 57f14649353..ac5de45d271 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btActivatingCollisionAlgorithm.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btActivatingCollisionAlgorithm.cpp
@@ -17,31 +17,31 @@ subject to the following restrictions:
 #include "btCollisionDispatcher.h"
 #include "btCollisionObject.h"
 
-btActivatingCollisionAlgorithm::btActivatingCollisionAlgorithm (const btCollisionAlgorithmConstructionInfo& ci)
-:btCollisionAlgorithm(ci)
+btActivatingCollisionAlgorithm::btActivatingCollisionAlgorithm(const btCollisionAlgorithmConstructionInfo& ci)
+	: btCollisionAlgorithm(ci)
 //,
 //m_colObj0(0),
 //m_colObj1(0)
 {
 }
-btActivatingCollisionAlgorithm::btActivatingCollisionAlgorithm (const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* ,const btCollisionObjectWrapper* )
-:btCollisionAlgorithm(ci)
+btActivatingCollisionAlgorithm::btActivatingCollisionAlgorithm(const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper*, const btCollisionObjectWrapper*)
+	: btCollisionAlgorithm(ci)
 //,
 //m_colObj0(0),
 //m_colObj1(0)
 {
-//	if (ci.m_dispatcher1->needsCollision(colObj0,colObj1))
-//	{
-//		m_colObj0 = colObj0;
-//		m_colObj1 = colObj1;
-//		
-//		m_colObj0->activate();
-//		m_colObj1->activate();
-//	}
+	//	if (ci.m_dispatcher1->needsCollision(colObj0,colObj1))
+	//	{
+	//		m_colObj0 = colObj0;
+	//		m_colObj1 = colObj1;
+	//
+	//		m_colObj0->activate();
+	//		m_colObj1->activate();
+	//	}
 }
 
 btActivatingCollisionAlgorithm::~btActivatingCollisionAlgorithm()
 {
-//		m_colObj0->activate();
-//		m_colObj1->activate();
+	//		m_colObj0->activate();
+	//		m_colObj1->activate();
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btActivatingCollisionAlgorithm.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btActivatingCollisionAlgorithm.h
index 0e19f1ea356..862060571b5 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btActivatingCollisionAlgorithm.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btActivatingCollisionAlgorithm.h
@@ -21,17 +21,15 @@ subject to the following restrictions:
 ///This class is not enabled yet (work-in-progress) to more aggressively activate objects.
 class btActivatingCollisionAlgorithm : public btCollisionAlgorithm
 {
-//	btCollisionObject* m_colObj0;
-//	btCollisionObject* m_colObj1;
+	//	btCollisionObject* m_colObj0;
+	//	btCollisionObject* m_colObj1;
 
 protected:
+	btActivatingCollisionAlgorithm(const btCollisionAlgorithmConstructionInfo& ci);
 
-	btActivatingCollisionAlgorithm (const btCollisionAlgorithmConstructionInfo& ci);
-
-	btActivatingCollisionAlgorithm (const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap);
+	btActivatingCollisionAlgorithm(const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap);
 
 public:
 	virtual ~btActivatingCollisionAlgorithm();
-
 };
-#endif //__BT_ACTIVATING_COLLISION_ALGORITHM_H
+#endif  //__BT_ACTIVATING_COLLISION_ALGORITHM_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btBox2dBox2dCollisionAlgorithm.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btBox2dBox2dCollisionAlgorithm.cpp
index cabbb0bf6a0..6873a95d90a 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btBox2dBox2dCollisionAlgorithm.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btBox2dBox2dCollisionAlgorithm.cpp
@@ -26,61 +26,55 @@ subject to the following restrictions:
 
 #define USE_PERSISTENT_CONTACTS 1
 
-btBox2dBox2dCollisionAlgorithm::btBox2dBox2dCollisionAlgorithm(btPersistentManifold* mf,const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* obj0Wrap,const btCollisionObjectWrapper* obj1Wrap)
-: btActivatingCollisionAlgorithm(ci,obj0Wrap,obj1Wrap),
-m_ownManifold(false),
-m_manifoldPtr(mf)
+btBox2dBox2dCollisionAlgorithm::btBox2dBox2dCollisionAlgorithm(btPersistentManifold* mf, const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* obj0Wrap, const btCollisionObjectWrapper* obj1Wrap)
+	: btActivatingCollisionAlgorithm(ci, obj0Wrap, obj1Wrap),
+	  m_ownManifold(false),
+	  m_manifoldPtr(mf)
 {
-	if (!m_manifoldPtr && m_dispatcher->needsCollision(obj0Wrap->getCollisionObject(),obj1Wrap->getCollisionObject()))
+	if (!m_manifoldPtr && m_dispatcher->needsCollision(obj0Wrap->getCollisionObject(), obj1Wrap->getCollisionObject()))
 	{
-		m_manifoldPtr = m_dispatcher->getNewManifold(obj0Wrap->getCollisionObject(),obj1Wrap->getCollisionObject());
+		m_manifoldPtr = m_dispatcher->getNewManifold(obj0Wrap->getCollisionObject(), obj1Wrap->getCollisionObject());
 		m_ownManifold = true;
 	}
 }
 
 btBox2dBox2dCollisionAlgorithm::~btBox2dBox2dCollisionAlgorithm()
 {
-	
 	if (m_ownManifold)
 	{
 		if (m_manifoldPtr)
 			m_dispatcher->releaseManifold(m_manifoldPtr);
 	}
-	
 }
 
-
-void b2CollidePolygons(btManifoldResult* manifold,  const btBox2dShape* polyA, const btTransform& xfA, const btBox2dShape* polyB, const btTransform& xfB);
+void b2CollidePolygons(btManifoldResult* manifold, const btBox2dShape* polyA, const btTransform& xfA, const btBox2dShape* polyB, const btTransform& xfB);
 
 //#include <stdio.h>
-void btBox2dBox2dCollisionAlgorithm::processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+void btBox2dBox2dCollisionAlgorithm::processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	if (!m_manifoldPtr)
 		return;
 
-	
 	const btBox2dShape* box0 = (const btBox2dShape*)body0Wrap->getCollisionShape();
 	const btBox2dShape* box1 = (const btBox2dShape*)body1Wrap->getCollisionShape();
 
 	resultOut->setPersistentManifold(m_manifoldPtr);
 
-	b2CollidePolygons(resultOut,box0,body0Wrap->getWorldTransform(),box1,body1Wrap->getWorldTransform());
+	b2CollidePolygons(resultOut, box0, body0Wrap->getWorldTransform(), box1, body1Wrap->getWorldTransform());
 
 	//  refreshContactPoints is only necessary when using persistent contact points. otherwise all points are newly added
 	if (m_ownManifold)
 	{
 		resultOut->refreshContactPoints();
 	}
-
 }
 
-btScalar btBox2dBox2dCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* /*body0*/,btCollisionObject* /*body1*/,const btDispatcherInfo& /*dispatchInfo*/,btManifoldResult* /*resultOut*/)
+btScalar btBox2dBox2dCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* /*body0*/, btCollisionObject* /*body1*/, const btDispatcherInfo& /*dispatchInfo*/, btManifoldResult* /*resultOut*/)
 {
 	//not yet
 	return 1.f;
 }
 
-
 struct ClipVertex
 {
 	btVector3 v;
@@ -89,16 +83,16 @@ struct ClipVertex
 	//b2ContactID id;
 };
 
-#define b2Dot(a,b) (a).dot(b)
-#define b2Mul(a,b) (a)*(b)
-#define b2MulT(a,b) (a).transpose()*(b)
-#define b2Cross(a,b) (a).cross(b)
-#define btCrossS(a,s) btVector3(s * a.getY(), -s * a.getX(),0.f)
+#define b2Dot(a, b) (a).dot(b)
+#define b2Mul(a, b) (a) * (b)
+#define b2MulT(a, b) (a).transpose() * (b)
+#define b2Cross(a, b) (a).cross(b)
+#define btCrossS(a, s) btVector3(s* a.getY(), -s* a.getX(), 0.f)
 
-int b2_maxManifoldPoints =2;
+int b2_maxManifoldPoints = 2;
 
 static int ClipSegmentToLine(ClipVertex vOut[2], ClipVertex vIn[2],
-					  const btVector3& normal, btScalar offset)
+							 const btVector3& normal, btScalar offset)
 {
 	// Start with no output points
 	int numOut = 0;
@@ -133,7 +127,7 @@ static int ClipSegmentToLine(ClipVertex vOut[2], ClipVertex vIn[2],
 
 // Find the separation between poly1 and poly2 for a give edge normal on poly1.
 static btScalar EdgeSeparation(const btBox2dShape* poly1, const btTransform& xf1, int edge1,
-							  const btBox2dShape* poly2, const btTransform& xf2)
+							   const btBox2dShape* poly2, const btTransform& xf2)
 {
 	const btVector3* vertices1 = poly1->getVertices();
 	const btVector3* normals1 = poly1->getNormals();
@@ -151,8 +145,8 @@ static btScalar EdgeSeparation(const btBox2dShape* poly1, const btTransform& xf1
 	int index = 0;
 	btScalar minDot = BT_LARGE_FLOAT;
 
-	if( count2 > 0 )
-		index = (int) normal1.minDot( vertices2, count2, minDot);
+	if (count2 > 0)
+		index = (int)normal1.minDot(vertices2, count2, minDot);
 
 	btVector3 v1 = b2Mul(xf1, vertices1[edge1]);
 	btVector3 v2 = b2Mul(xf2, vertices2[index]);
@@ -162,8 +156,8 @@ static btScalar EdgeSeparation(const btBox2dShape* poly1, const btTransform& xf1
 
 // Find the max separation between poly1 and poly2 using edge normals from poly1.
 static btScalar FindMaxSeparation(int* edgeIndex,
-								 const btBox2dShape* poly1, const btTransform& xf1,
-								 const btBox2dShape* poly2, const btTransform& xf2)
+								  const btBox2dShape* poly1, const btTransform& xf1,
+								  const btBox2dShape* poly2, const btTransform& xf2)
 {
 	int count1 = poly1->getVertexCount();
 	const btVector3* normals1 = poly1->getNormals();
@@ -174,9 +168,9 @@ static btScalar FindMaxSeparation(int* edgeIndex,
 
 	// Find edge normal on poly1 that has the largest projection onto d.
 	int edge = 0;
-    btScalar maxDot;
-	if( count1 > 0 )
-		edge = (int) dLocal1.maxDot( normals1, count1, maxDot);
+	btScalar maxDot;
+	if (count1 > 0)
+		edge = (int)dLocal1.maxDot(normals1, count1, maxDot);
 
 	// Get the separation for the edge normal.
 	btScalar s = EdgeSeparation(poly1, xf1, edge, poly2, xf2);
@@ -224,7 +218,7 @@ static btScalar FindMaxSeparation(int* edgeIndex,
 	}
 
 	// Perform a local search for the best edge normal.
-	for ( ; ; )
+	for (;;)
 	{
 		if (increment == -1)
 			edge = bestEdge - 1 >= 0 ? bestEdge - 1 : count1 - 1;
@@ -285,14 +279,14 @@ static void FindIncidentEdge(ClipVertex c[2],
 	int i2 = i1 + 1 < count2 ? i1 + 1 : 0;
 
 	c[0].v = b2Mul(xf2, vertices2[i1]);
-//	c[0].id.features.referenceEdge = (unsigned char)edge1;
-//	c[0].id.features.incidentEdge = (unsigned char)i1;
-//	c[0].id.features.incidentVertex = 0;
+	//	c[0].id.features.referenceEdge = (unsigned char)edge1;
+	//	c[0].id.features.incidentEdge = (unsigned char)i1;
+	//	c[0].id.features.incidentVertex = 0;
 
 	c[1].v = b2Mul(xf2, vertices2[i2]);
-//	c[1].id.features.referenceEdge = (unsigned char)edge1;
-//	c[1].id.features.incidentEdge = (unsigned char)i2;
-//	c[1].id.features.incidentVertex = 1;
+	//	c[1].id.features.referenceEdge = (unsigned char)edge1;
+	//	c[1].id.features.incidentEdge = (unsigned char)i2;
+	//	c[1].id.features.incidentVertex = 1;
 }
 
 // Find edge normal of max separation on A - return if separating axis is found
@@ -303,10 +297,9 @@ static void FindIncidentEdge(ClipVertex c[2],
 
 // The normal points from 1 to 2
 void b2CollidePolygons(btManifoldResult* manifold,
-					  const btBox2dShape* polyA, const btTransform& xfA,
-					  const btBox2dShape* polyB, const btTransform& xfB)
+					   const btBox2dShape* polyA, const btTransform& xfA,
+					   const btBox2dShape* polyB, const btTransform& xfB)
 {
-
 	int edgeA = 0;
 	btScalar separationA = FindMaxSeparation(&edgeA, polyA, xfA, polyB, xfB);
 	if (separationA > 0.0f)
@@ -317,10 +310,10 @@ void b2CollidePolygons(btManifoldResult* manifold,
 	if (separationB > 0.0f)
 		return;
 
-	const btBox2dShape* poly1;	// reference poly
-	const btBox2dShape* poly2;	// incident poly
+	const btBox2dShape* poly1;  // reference poly
+	const btBox2dShape* poly2;  // incident poly
 	btTransform xf1, xf2;
-	int edge1;		// reference edge
+	int edge1;  // reference edge
 	unsigned char flip;
 	const btScalar k_relativeTol = 0.98f;
 	const btScalar k_absoluteTol = 0.001f;
@@ -352,14 +345,13 @@ void b2CollidePolygons(btManifoldResult* manifold,
 	const btVector3* vertices1 = poly1->getVertices();
 
 	btVector3 v11 = vertices1[edge1];
-	btVector3 v12 = edge1 + 1 < count1 ? vertices1[edge1+1] : vertices1[0];
+	btVector3 v12 = edge1 + 1 < count1 ? vertices1[edge1 + 1] : vertices1[0];
 
 	//btVector3 dv = v12 - v11;
 	btVector3 sideNormal = b2Mul(xf1.getBasis(), v12 - v11);
 	sideNormal.normalize();
 	btVector3 frontNormal = btCrossS(sideNormal, 1.0f);
-	
-	
+
 	v11 = b2Mul(xf1, v11);
 	v12 = b2Mul(xf1, v12);
 
@@ -369,13 +361,12 @@ void b2CollidePolygons(btManifoldResult* manifold,
 
 	// Clip incident edge against extruded edge1 side edges.
 	ClipVertex clipPoints1[2];
-	clipPoints1[0].v.setValue(0,0,0);
-	clipPoints1[1].v.setValue(0,0,0);
+	clipPoints1[0].v.setValue(0, 0, 0);
+	clipPoints1[1].v.setValue(0, 0, 0);
 
 	ClipVertex clipPoints2[2];
-	clipPoints2[0].v.setValue(0,0,0);
-	clipPoints2[1].v.setValue(0,0,0);
-
+	clipPoints2[0].v.setValue(0, 0, 0);
+	clipPoints2[1].v.setValue(0, 0, 0);
 
 	int np;
 
@@ -386,7 +377,7 @@ void b2CollidePolygons(btManifoldResult* manifold,
 		return;
 
 	// Clip to negative box side 1
-	np = ClipSegmentToLine(clipPoints2, clipPoints1,  sideNormal, sideOffset2);
+	np = ClipSegmentToLine(clipPoints2, clipPoints1, sideNormal, sideOffset2);
 
 	if (np < 2)
 	{
@@ -403,19 +394,18 @@ void b2CollidePolygons(btManifoldResult* manifold,
 
 		if (separation <= 0.0f)
 		{
-			
 			//b2ManifoldPoint* cp = manifold->points + pointCount;
 			//btScalar separation = separation;
 			//cp->localPoint1 = b2MulT(xfA, clipPoints2[i].v);
 			//cp->localPoint2 = b2MulT(xfB, clipPoints2[i].v);
 
-			manifold->addContactPoint(-manifoldNormal,clipPoints2[i].v,separation);
+			manifold->addContactPoint(-manifoldNormal, clipPoints2[i].v, separation);
 
-//			cp->id = clipPoints2[i].id;
-//			cp->id.features.flip = flip;
+			//			cp->id = clipPoints2[i].id;
+			//			cp->id.features.flip = flip;
 			++pointCount;
 		}
 	}
 
-//	manifold->pointCount = pointCount;}
+	//	manifold->pointCount = pointCount;}
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btBox2dBox2dCollisionAlgorithm.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btBox2dBox2dCollisionAlgorithm.h
index 6ea6e89bda6..3b66d1fd0b9 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btBox2dBox2dCollisionAlgorithm.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btBox2dBox2dCollisionAlgorithm.h
@@ -26,22 +26,22 @@ class btPersistentManifold;
 ///box-box collision detection
 class btBox2dBox2dCollisionAlgorithm : public btActivatingCollisionAlgorithm
 {
-	bool	m_ownManifold;
-	btPersistentManifold*	m_manifoldPtr;
-	
+	bool m_ownManifold;
+	btPersistentManifold* m_manifoldPtr;
+
 public:
 	btBox2dBox2dCollisionAlgorithm(const btCollisionAlgorithmConstructionInfo& ci)
 		: btActivatingCollisionAlgorithm(ci) {}
 
-	virtual void processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual void processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	btBox2dBox2dCollisionAlgorithm(btPersistentManifold* mf,const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap);
+	btBox2dBox2dCollisionAlgorithm(btPersistentManifold* mf, const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap);
 
 	virtual ~btBox2dBox2dCollisionAlgorithm();
 
-	virtual	void	getAllContactManifolds(btManifoldArray&	manifoldArray)
+	virtual void getAllContactManifolds(btManifoldArray& manifoldArray)
 	{
 		if (m_manifoldPtr && m_ownManifold)
 		{
@@ -49,18 +49,15 @@ public:
 		}
 	}
 
-
-	struct CreateFunc :public 	btCollisionAlgorithmCreateFunc
+	struct CreateFunc : public btCollisionAlgorithmCreateFunc
 	{
-		virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+		virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 		{
 			int bbsize = sizeof(btBox2dBox2dCollisionAlgorithm);
 			void* ptr = ci.m_dispatcher1->allocateCollisionAlgorithm(bbsize);
-			return new(ptr) btBox2dBox2dCollisionAlgorithm(0,ci,body0Wrap,body1Wrap);
+			return new (ptr) btBox2dBox2dCollisionAlgorithm(0, ci, body0Wrap, body1Wrap);
 		}
 	};
-
 };
 
-#endif //BT_BOX_2D_BOX_2D__COLLISION_ALGORITHM_H
-
+#endif  //BT_BOX_2D_BOX_2D__COLLISION_ALGORITHM_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btBoxBoxCollisionAlgorithm.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btBoxBoxCollisionAlgorithm.cpp
index ac68968f590..7a391e059a7 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btBoxBoxCollisionAlgorithm.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btBoxBoxCollisionAlgorithm.cpp
@@ -21,14 +21,14 @@ subject to the following restrictions:
 #include "BulletCollision/CollisionDispatch/btCollisionObjectWrapper.h"
 #define USE_PERSISTENT_CONTACTS 1
 
-btBoxBoxCollisionAlgorithm::btBoxBoxCollisionAlgorithm(btPersistentManifold* mf,const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
-: btActivatingCollisionAlgorithm(ci,body0Wrap,body1Wrap),
-m_ownManifold(false),
-m_manifoldPtr(mf)
+btBoxBoxCollisionAlgorithm::btBoxBoxCollisionAlgorithm(btPersistentManifold* mf, const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
+	: btActivatingCollisionAlgorithm(ci, body0Wrap, body1Wrap),
+	  m_ownManifold(false),
+	  m_manifoldPtr(mf)
 {
-	if (!m_manifoldPtr && m_dispatcher->needsCollision(body0Wrap->getCollisionObject(),body1Wrap->getCollisionObject()))
+	if (!m_manifoldPtr && m_dispatcher->needsCollision(body0Wrap->getCollisionObject(), body1Wrap->getCollisionObject()))
 	{
-		m_manifoldPtr = m_dispatcher->getNewManifold(body0Wrap->getCollisionObject(),body1Wrap->getCollisionObject());
+		m_manifoldPtr = m_dispatcher->getNewManifold(body0Wrap->getCollisionObject(), body1Wrap->getCollisionObject());
 		m_ownManifold = true;
 	}
 }
@@ -42,30 +42,27 @@ btBoxBoxCollisionAlgorithm::~btBoxBoxCollisionAlgorithm()
 	}
 }
 
-void btBoxBoxCollisionAlgorithm::processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+void btBoxBoxCollisionAlgorithm::processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	if (!m_manifoldPtr)
 		return;
 
-	
 	const btBoxShape* box0 = (btBoxShape*)body0Wrap->getCollisionShape();
 	const btBoxShape* box1 = (btBoxShape*)body1Wrap->getCollisionShape();
 
-
-
 	/// report a contact. internally this will be kept persistent, and contact reduction is done
 	resultOut->setPersistentManifold(m_manifoldPtr);
-#ifndef USE_PERSISTENT_CONTACTS	
+#ifndef USE_PERSISTENT_CONTACTS
 	m_manifoldPtr->clearManifold();
-#endif //USE_PERSISTENT_CONTACTS
+#endif  //USE_PERSISTENT_CONTACTS
 
 	btDiscreteCollisionDetectorInterface::ClosestPointInput input;
 	input.m_maximumDistanceSquared = BT_LARGE_FLOAT;
 	input.m_transformA = body0Wrap->getWorldTransform();
 	input.m_transformB = body1Wrap->getWorldTransform();
 
-	btBoxBoxDetector detector(box0,box1);
-	detector.getClosestPoints(input,*resultOut,dispatchInfo.m_debugDraw);
+	btBoxBoxDetector detector(box0, box1);
+	detector.getClosestPoints(input, *resultOut, dispatchInfo.m_debugDraw);
 
 #ifdef USE_PERSISTENT_CONTACTS
 	//  refreshContactPoints is only necessary when using persistent contact points. otherwise all points are newly added
@@ -73,11 +70,10 @@ void btBoxBoxCollisionAlgorithm::processCollision (const btCollisionObjectWrappe
 	{
 		resultOut->refreshContactPoints();
 	}
-#endif //USE_PERSISTENT_CONTACTS
-
+#endif  //USE_PERSISTENT_CONTACTS
 }
 
-btScalar btBoxBoxCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* /*body0*/,btCollisionObject* /*body1*/,const btDispatcherInfo& /*dispatchInfo*/,btManifoldResult* /*resultOut*/)
+btScalar btBoxBoxCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* /*body0*/, btCollisionObject* /*body1*/, const btDispatcherInfo& /*dispatchInfo*/, btManifoldResult* /*resultOut*/)
 {
 	//not yet
 	return 1.f;
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btBoxBoxCollisionAlgorithm.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btBoxBoxCollisionAlgorithm.h
index 59808df5a9d..eb210657659 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btBoxBoxCollisionAlgorithm.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btBoxBoxCollisionAlgorithm.h
@@ -26,22 +26,22 @@ class btPersistentManifold;
 ///box-box collision detection
 class btBoxBoxCollisionAlgorithm : public btActivatingCollisionAlgorithm
 {
-	bool	m_ownManifold;
-	btPersistentManifold*	m_manifoldPtr;
-	
+	bool m_ownManifold;
+	btPersistentManifold* m_manifoldPtr;
+
 public:
 	btBoxBoxCollisionAlgorithm(const btCollisionAlgorithmConstructionInfo& ci)
 		: btActivatingCollisionAlgorithm(ci) {}
 
-	virtual void processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual void processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	btBoxBoxCollisionAlgorithm(btPersistentManifold* mf,const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap);
+	btBoxBoxCollisionAlgorithm(btPersistentManifold* mf, const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap);
 
 	virtual ~btBoxBoxCollisionAlgorithm();
 
-	virtual	void	getAllContactManifolds(btManifoldArray&	manifoldArray)
+	virtual void getAllContactManifolds(btManifoldArray& manifoldArray)
 	{
 		if (m_manifoldPtr && m_ownManifold)
 		{
@@ -49,18 +49,15 @@ public:
 		}
 	}
 
-
-	struct CreateFunc :public 	btCollisionAlgorithmCreateFunc
+	struct CreateFunc : public btCollisionAlgorithmCreateFunc
 	{
-		virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+		virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 		{
 			int bbsize = sizeof(btBoxBoxCollisionAlgorithm);
 			void* ptr = ci.m_dispatcher1->allocateCollisionAlgorithm(bbsize);
-			return new(ptr) btBoxBoxCollisionAlgorithm(0,ci,body0Wrap,body1Wrap);
+			return new (ptr) btBoxBoxCollisionAlgorithm(0, ci, body0Wrap, body1Wrap);
 		}
 	};
-
 };
 
-#endif //BT_BOX_BOX__COLLISION_ALGORITHM_H
-
+#endif  //BT_BOX_BOX__COLLISION_ALGORITHM_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btBoxBoxDetector.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btBoxBoxDetector.cpp
index 7043bde34f5..202039956e7 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btBoxBoxDetector.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btBoxBoxDetector.cpp
@@ -24,14 +24,12 @@ subject to the following restrictions:
 #include <float.h>
 #include <string.h>
 
-btBoxBoxDetector::btBoxBoxDetector(const btBoxShape* box1,const btBoxShape* box2)
-: m_box1(box1),
-m_box2(box2)
+btBoxBoxDetector::btBoxBoxDetector(const btBoxShape* box1, const btBoxShape* box2)
+	: m_box1(box1),
+	  m_box2(box2)
 {
-
 }
 
-
 // given two boxes (p1,R1,side1) and (p2,R2,side2), collide them together and
 // generate contact points. this returns 0 if there is no contact otherwise
 // it returns the number of contacts generated.
@@ -48,67 +46,66 @@ m_box2(box2)
 // collision functions. this function only fills in the position and depth
 // fields.
 struct dContactGeom;
-#define dDOTpq(a,b,p,q) ((a)[0]*(b)[0] + (a)[p]*(b)[q] + (a)[2*(p)]*(b)[2*(q)])
+#define dDOTpq(a, b, p, q) ((a)[0] * (b)[0] + (a)[p] * (b)[q] + (a)[2 * (p)] * (b)[2 * (q)])
 #define dInfinity FLT_MAX
 
-
 /*PURE_INLINE btScalar dDOT   (const btScalar *a, const btScalar *b) { return dDOTpq(a,b,1,1); }
 PURE_INLINE btScalar dDOT13 (const btScalar *a, const btScalar *b) { return dDOTpq(a,b,1,3); }
 PURE_INLINE btScalar dDOT31 (const btScalar *a, const btScalar *b) { return dDOTpq(a,b,3,1); }
 PURE_INLINE btScalar dDOT33 (const btScalar *a, const btScalar *b) { return dDOTpq(a,b,3,3); }
 */
-static btScalar dDOT   (const btScalar *a, const btScalar *b) { return dDOTpq(a,b,1,1); }
-static btScalar dDOT44 (const btScalar *a, const btScalar *b) { return dDOTpq(a,b,4,4); }
-static btScalar dDOT41 (const btScalar *a, const btScalar *b) { return dDOTpq(a,b,4,1); }
-static btScalar dDOT14 (const btScalar *a, const btScalar *b) { return dDOTpq(a,b,1,4); }
-#define dMULTIPLYOP1_331(A,op,B,C) \
-{\
-  (A)[0] op dDOT41((B),(C)); \
-  (A)[1] op dDOT41((B+1),(C)); \
-  (A)[2] op dDOT41((B+2),(C)); \
-}
+static btScalar dDOT(const btScalar* a, const btScalar* b) { return dDOTpq(a, b, 1, 1); }
+static btScalar dDOT44(const btScalar* a, const btScalar* b) { return dDOTpq(a, b, 4, 4); }
+static btScalar dDOT41(const btScalar* a, const btScalar* b) { return dDOTpq(a, b, 4, 1); }
+static btScalar dDOT14(const btScalar* a, const btScalar* b) { return dDOTpq(a, b, 1, 4); }
+#define dMULTIPLYOP1_331(A, op, B, C)   \
+	{                                   \
+		(A)[0] op dDOT41((B), (C));     \
+		(A)[1] op dDOT41((B + 1), (C)); \
+		(A)[2] op dDOT41((B + 2), (C)); \
+	}
 
-#define dMULTIPLYOP0_331(A,op,B,C) \
-{ \
-  (A)[0] op dDOT((B),(C)); \
-  (A)[1] op dDOT((B+4),(C)); \
-  (A)[2] op dDOT((B+8),(C)); \
-} 
+#define dMULTIPLYOP0_331(A, op, B, C) \
+	{                                 \
+		(A)[0] op dDOT((B), (C));     \
+		(A)[1] op dDOT((B + 4), (C)); \
+		(A)[2] op dDOT((B + 8), (C)); \
+	}
 
-#define dMULTIPLY1_331(A,B,C) dMULTIPLYOP1_331(A,=,B,C)
-#define dMULTIPLY0_331(A,B,C) dMULTIPLYOP0_331(A,=,B,C)
+#define dMULTIPLY1_331(A, B, C) dMULTIPLYOP1_331(A, =, B, C)
+#define dMULTIPLY0_331(A, B, C) dMULTIPLYOP0_331(A, =, B, C)
 
-typedef btScalar dMatrix3[4*3];
+typedef btScalar dMatrix3[4 * 3];
 
-void dLineClosestApproach (const btVector3& pa, const btVector3& ua,
-			   const btVector3& pb, const btVector3& ub,
-			   btScalar *alpha, btScalar *beta);
-void dLineClosestApproach (const btVector3& pa, const btVector3& ua,
-			   const btVector3& pb, const btVector3& ub,
-			   btScalar *alpha, btScalar *beta)
+void dLineClosestApproach(const btVector3& pa, const btVector3& ua,
+						  const btVector3& pb, const btVector3& ub,
+						  btScalar* alpha, btScalar* beta);
+void dLineClosestApproach(const btVector3& pa, const btVector3& ua,
+						  const btVector3& pb, const btVector3& ub,
+						  btScalar* alpha, btScalar* beta)
 {
-  btVector3 p;
-  p[0] = pb[0] - pa[0];
-  p[1] = pb[1] - pa[1];
-  p[2] = pb[2] - pa[2];
-  btScalar uaub = dDOT(ua,ub);
-  btScalar q1 =  dDOT(ua,p);
-  btScalar q2 = -dDOT(ub,p);
-  btScalar d = 1-uaub*uaub;
-  if (d <= btScalar(0.0001f)) {
-    // @@@ this needs to be made more robust
-    *alpha = 0;
-    *beta  = 0;
-  }
-  else {
-    d = 1.f/d;
-    *alpha = (q1 + uaub*q2)*d;
-    *beta  = (uaub*q1 + q2)*d;
-  }
+	btVector3 p;
+	p[0] = pb[0] - pa[0];
+	p[1] = pb[1] - pa[1];
+	p[2] = pb[2] - pa[2];
+	btScalar uaub = dDOT(ua, ub);
+	btScalar q1 = dDOT(ua, p);
+	btScalar q2 = -dDOT(ub, p);
+	btScalar d = 1 - uaub * uaub;
+	if (d <= btScalar(0.0001f))
+	{
+		// @@@ this needs to be made more robust
+		*alpha = 0;
+		*beta = 0;
+	}
+	else
+	{
+		d = 1.f / d;
+		*alpha = (q1 + uaub * q2) * d;
+		*beta = (uaub * q1 + q2) * d;
+	}
 }
 
-
-
 // find all the intersection points between the 2D rectangle with vertices
 // at (+/-h[0],+/-h[1]) and the 2D quadrilateral with vertices (p[0],p[1]),
 // (p[2],p[3]),(p[4],p[5]),(p[6],p[7]).
@@ -117,60 +114,66 @@ void dLineClosestApproach (const btVector3& pa, const btVector3& ua,
 // the number of intersection points is returned by the function (this will
 // be in the range 0 to 8).
 
-static int intersectRectQuad2 (btScalar h[2], btScalar p[8], btScalar ret[16])
+static int intersectRectQuad2(btScalar h[2], btScalar p[8], btScalar ret[16])
 {
-  // q (and r) contain nq (and nr) coordinate points for the current (and
-  // chopped) polygons
-  int nq=4,nr=0;
-  btScalar buffer[16];
-  btScalar *q = p;
-  btScalar *r = ret;
-  for (int dir=0; dir <= 1; dir++) {
-    // direction notation: xy[0] = x axis, xy[1] = y axis
-    for (int sign=-1; sign <= 1; sign += 2) {
-      // chop q along the line xy[dir] = sign*h[dir]
-      btScalar *pq = q;
-      btScalar *pr = r;
-      nr = 0;
-      for (int i=nq; i > 0; i--) {
-	// go through all points in q and all lines between adjacent points
-	if (sign*pq[dir] < h[dir]) {
-	  // this point is inside the chopping line
-	  pr[0] = pq[0];
-	  pr[1] = pq[1];
-	  pr += 2;
-	  nr++;
-	  if (nr & 8) {
-	    q = r;
-	    goto done;
-	  }
+	// q (and r) contain nq (and nr) coordinate points for the current (and
+	// chopped) polygons
+	int nq = 4, nr = 0;
+	btScalar buffer[16];
+	btScalar* q = p;
+	btScalar* r = ret;
+	for (int dir = 0; dir <= 1; dir++)
+	{
+		// direction notation: xy[0] = x axis, xy[1] = y axis
+		for (int sign = -1; sign <= 1; sign += 2)
+		{
+			// chop q along the line xy[dir] = sign*h[dir]
+			btScalar* pq = q;
+			btScalar* pr = r;
+			nr = 0;
+			for (int i = nq; i > 0; i--)
+			{
+				// go through all points in q and all lines between adjacent points
+				if (sign * pq[dir] < h[dir])
+				{
+					// this point is inside the chopping line
+					pr[0] = pq[0];
+					pr[1] = pq[1];
+					pr += 2;
+					nr++;
+					if (nr & 8)
+					{
+						q = r;
+						goto done;
+					}
+				}
+				btScalar* nextq = (i > 1) ? pq + 2 : q;
+				if ((sign * pq[dir] < h[dir]) ^ (sign * nextq[dir] < h[dir]))
+				{
+					// this line crosses the chopping line
+					pr[1 - dir] = pq[1 - dir] + (nextq[1 - dir] - pq[1 - dir]) /
+													(nextq[dir] - pq[dir]) * (sign * h[dir] - pq[dir]);
+					pr[dir] = sign * h[dir];
+					pr += 2;
+					nr++;
+					if (nr & 8)
+					{
+						q = r;
+						goto done;
+					}
+				}
+				pq += 2;
+			}
+			q = r;
+			r = (q == ret) ? buffer : ret;
+			nq = nr;
+		}
 	}
-	btScalar *nextq = (i > 1) ? pq+2 : q;
-	if ((sign*pq[dir] < h[dir]) ^ (sign*nextq[dir] < h[dir])) {
-	  // this line crosses the chopping line
-	  pr[1-dir] = pq[1-dir] + (nextq[1-dir]-pq[1-dir]) /
-	    (nextq[dir]-pq[dir]) * (sign*h[dir]-pq[dir]);
-	  pr[dir] = sign*h[dir];
-	  pr += 2;
-	  nr++;
-	  if (nr & 8) {
-	    q = r;
-	    goto done;
-	  }
-	}
-	pq += 2;
-      }
-      q = r;
-      r = (q==ret) ? buffer : ret;
-      nq = nr;
-    }
-  }
- done:
-  if (q != ret) memcpy (ret,q,nr*2*sizeof(btScalar));
-  return nr;
+done:
+	if (q != ret) memcpy(ret, q, nr * 2 * sizeof(btScalar));
+	return nr;
 }
 
-
 #define M__PI 3.14159265f
 
 // given n points in the plane (array p, of size 2*n), generate m points that
@@ -181,538 +184,584 @@ static int intersectRectQuad2 (btScalar h[2], btScalar p[8], btScalar ret[16])
 // n must be in the range [1..8]. m must be in the range [1..n]. i0 must be
 // in the range [0..n-1].
 
-void cullPoints2 (int n, btScalar p[], int m, int i0, int iret[]);
-void cullPoints2 (int n, btScalar p[], int m, int i0, int iret[])
+void cullPoints2(int n, btScalar p[], int m, int i0, int iret[]);
+void cullPoints2(int n, btScalar p[], int m, int i0, int iret[])
 {
-  // compute the centroid of the polygon in cx,cy
-  int i,j;
-  btScalar a,cx,cy,q;
-  if (n==1) {
-    cx = p[0];
-    cy = p[1];
-  }
-  else if (n==2) {
-    cx = btScalar(0.5)*(p[0] + p[2]);
-    cy = btScalar(0.5)*(p[1] + p[3]);
-  }
-  else {
-    a = 0;
-    cx = 0;
-    cy = 0;
-    for (i=0; i<(n-1); i++) {
-      q = p[i*2]*p[i*2+3] - p[i*2+2]*p[i*2+1];
-      a += q;
-      cx += q*(p[i*2]+p[i*2+2]);
-      cy += q*(p[i*2+1]+p[i*2+3]);
-    }
-    q = p[n*2-2]*p[1] - p[0]*p[n*2-1];
-	if (btFabs(a+q) > SIMD_EPSILON)
+	// compute the centroid of the polygon in cx,cy
+	int i, j;
+	btScalar a, cx, cy, q;
+	if (n == 1)
 	{
-		a = 1.f/(btScalar(3.0)*(a+q));
-	} else
+		cx = p[0];
+		cy = p[1];
+	}
+	else if (n == 2)
 	{
-		a=BT_LARGE_FLOAT;
+		cx = btScalar(0.5) * (p[0] + p[2]);
+		cy = btScalar(0.5) * (p[1] + p[3]);
 	}
-    cx = a*(cx + q*(p[n*2-2]+p[0]));
-    cy = a*(cy + q*(p[n*2-1]+p[1]));
-  }
-
-  // compute the angle of each point w.r.t. the centroid
-  btScalar A[8];
-  for (i=0; i<n; i++) A[i] = btAtan2(p[i*2+1]-cy,p[i*2]-cx);
-
-  // search for points that have angles closest to A[i0] + i*(2*pi/m).
-  int avail[8];
-  for (i=0; i<n; i++) avail[i] = 1;
-  avail[i0] = 0;
-  iret[0] = i0;
-  iret++;
-  for (j=1; j<m; j++) {
-    a = btScalar(j)*(2*M__PI/m) + A[i0];
-    if (a > M__PI) a -= 2*M__PI;
-    btScalar maxdiff=1e9,diff;
-
-    *iret = i0;			// iret is not allowed to keep this value, but it sometimes does, when diff=#QNAN0
-
-    for (i=0; i<n; i++) {
-      if (avail[i]) {
-	diff = btFabs (A[i]-a);
-	if (diff > M__PI) diff = 2*M__PI - diff;
-	if (diff < maxdiff) {
-	  maxdiff = diff;
-	  *iret = i;
+	else
+	{
+		a = 0;
+		cx = 0;
+		cy = 0;
+		for (i = 0; i < (n - 1); i++)
+		{
+			q = p[i * 2] * p[i * 2 + 3] - p[i * 2 + 2] * p[i * 2 + 1];
+			a += q;
+			cx += q * (p[i * 2] + p[i * 2 + 2]);
+			cy += q * (p[i * 2 + 1] + p[i * 2 + 3]);
+		}
+		q = p[n * 2 - 2] * p[1] - p[0] * p[n * 2 - 1];
+		if (btFabs(a + q) > SIMD_EPSILON)
+		{
+			a = 1.f / (btScalar(3.0) * (a + q));
+		}
+		else
+		{
+			a = BT_LARGE_FLOAT;
+		}
+		cx = a * (cx + q * (p[n * 2 - 2] + p[0]));
+		cy = a * (cy + q * (p[n * 2 - 1] + p[1]));
 	}
-      }
-    }
-#if defined(DEBUG) || defined (_DEBUG)
-    btAssert (*iret != i0);	// ensure iret got set
+
+	// compute the angle of each point w.r.t. the centroid
+	btScalar A[8];
+	for (i = 0; i < n; i++) A[i] = btAtan2(p[i * 2 + 1] - cy, p[i * 2] - cx);
+
+	// search for points that have angles closest to A[i0] + i*(2*pi/m).
+	int avail[8];
+	for (i = 0; i < n; i++) avail[i] = 1;
+	avail[i0] = 0;
+	iret[0] = i0;
+	iret++;
+	for (j = 1; j < m; j++)
+	{
+		a = btScalar(j) * (2 * M__PI / m) + A[i0];
+		if (a > M__PI) a -= 2 * M__PI;
+		btScalar maxdiff = 1e9, diff;
+
+		*iret = i0;  // iret is not allowed to keep this value, but it sometimes does, when diff=#QNAN0
+
+		for (i = 0; i < n; i++)
+		{
+			if (avail[i])
+			{
+				diff = btFabs(A[i] - a);
+				if (diff > M__PI) diff = 2 * M__PI - diff;
+				if (diff < maxdiff)
+				{
+					maxdiff = diff;
+					*iret = i;
+				}
+			}
+		}
+#if defined(DEBUG) || defined(_DEBUG)
+		btAssert(*iret != i0);  // ensure iret got set
 #endif
-    avail[*iret] = 0;
-    iret++;
-  }
+		avail[*iret] = 0;
+		iret++;
+	}
 }
 
-
-
-int dBoxBox2 (const btVector3& p1, const dMatrix3 R1,
-	     const btVector3& side1, const btVector3& p2,
-	     const dMatrix3 R2, const btVector3& side2,
-	     btVector3& normal, btScalar *depth, int *return_code,
-		 int maxc, dContactGeom * /*contact*/, int /*skip*/,btDiscreteCollisionDetectorInterface::Result& output);
-int dBoxBox2 (const btVector3& p1, const dMatrix3 R1,
-	     const btVector3& side1, const btVector3& p2,
-	     const dMatrix3 R2, const btVector3& side2,
-	     btVector3& normal, btScalar *depth, int *return_code,
-		 int maxc, dContactGeom * /*contact*/, int /*skip*/,btDiscreteCollisionDetectorInterface::Result& output)
+int dBoxBox2(const btVector3& p1, const dMatrix3 R1,
+			 const btVector3& side1, const btVector3& p2,
+			 const dMatrix3 R2, const btVector3& side2,
+			 btVector3& normal, btScalar* depth, int* return_code,
+			 int maxc, dContactGeom* /*contact*/, int /*skip*/, btDiscreteCollisionDetectorInterface::Result& output);
+int dBoxBox2(const btVector3& p1, const dMatrix3 R1,
+			 const btVector3& side1, const btVector3& p2,
+			 const dMatrix3 R2, const btVector3& side2,
+			 btVector3& normal, btScalar* depth, int* return_code,
+			 int maxc, dContactGeom* /*contact*/, int /*skip*/, btDiscreteCollisionDetectorInterface::Result& output)
 {
-  const btScalar fudge_factor = btScalar(1.05);
-  btVector3 p,pp,normalC(0.f,0.f,0.f);
-  const btScalar *normalR = 0;
-  btScalar A[3],B[3],R11,R12,R13,R21,R22,R23,R31,R32,R33,
-    Q11,Q12,Q13,Q21,Q22,Q23,Q31,Q32,Q33,s,s2,l;
-  int i,j,invert_normal,code;
+	const btScalar fudge_factor = btScalar(1.05);
+	btVector3 p, pp, normalC(0.f, 0.f, 0.f);
+	const btScalar* normalR = 0;
+	btScalar A[3], B[3], R11, R12, R13, R21, R22, R23, R31, R32, R33,
+		Q11, Q12, Q13, Q21, Q22, Q23, Q31, Q32, Q33, s, s2, l;
+	int i, j, invert_normal, code;
 
-  // get vector from centers of box 1 to box 2, relative to box 1
-  p = p2 - p1;
-  dMULTIPLY1_331 (pp,R1,p);		// get pp = p relative to body 1
+	// get vector from centers of box 1 to box 2, relative to box 1
+	p = p2 - p1;
+	dMULTIPLY1_331(pp, R1, p);  // get pp = p relative to body 1
 
-  // get side lengths / 2
-  A[0] = side1[0]*btScalar(0.5);
-  A[1] = side1[1]*btScalar(0.5);
-  A[2] = side1[2]*btScalar(0.5);
-  B[0] = side2[0]*btScalar(0.5);
-  B[1] = side2[1]*btScalar(0.5);
-  B[2] = side2[2]*btScalar(0.5);
+	// get side lengths / 2
+	A[0] = side1[0] * btScalar(0.5);
+	A[1] = side1[1] * btScalar(0.5);
+	A[2] = side1[2] * btScalar(0.5);
+	B[0] = side2[0] * btScalar(0.5);
+	B[1] = side2[1] * btScalar(0.5);
+	B[2] = side2[2] * btScalar(0.5);
 
-  // Rij is R1'*R2, i.e. the relative rotation between R1 and R2
-  R11 = dDOT44(R1+0,R2+0); R12 = dDOT44(R1+0,R2+1); R13 = dDOT44(R1+0,R2+2);
-  R21 = dDOT44(R1+1,R2+0); R22 = dDOT44(R1+1,R2+1); R23 = dDOT44(R1+1,R2+2);
-  R31 = dDOT44(R1+2,R2+0); R32 = dDOT44(R1+2,R2+1); R33 = dDOT44(R1+2,R2+2);
+	// Rij is R1'*R2, i.e. the relative rotation between R1 and R2
+	R11 = dDOT44(R1 + 0, R2 + 0);
+	R12 = dDOT44(R1 + 0, R2 + 1);
+	R13 = dDOT44(R1 + 0, R2 + 2);
+	R21 = dDOT44(R1 + 1, R2 + 0);
+	R22 = dDOT44(R1 + 1, R2 + 1);
+	R23 = dDOT44(R1 + 1, R2 + 2);
+	R31 = dDOT44(R1 + 2, R2 + 0);
+	R32 = dDOT44(R1 + 2, R2 + 1);
+	R33 = dDOT44(R1 + 2, R2 + 2);
 
-  Q11 = btFabs(R11); Q12 = btFabs(R12); Q13 = btFabs(R13);
-  Q21 = btFabs(R21); Q22 = btFabs(R22); Q23 = btFabs(R23);
-  Q31 = btFabs(R31); Q32 = btFabs(R32); Q33 = btFabs(R33);
+	Q11 = btFabs(R11);
+	Q12 = btFabs(R12);
+	Q13 = btFabs(R13);
+	Q21 = btFabs(R21);
+	Q22 = btFabs(R22);
+	Q23 = btFabs(R23);
+	Q31 = btFabs(R31);
+	Q32 = btFabs(R32);
+	Q33 = btFabs(R33);
 
-  // for all 15 possible separating axes:
-  //   * see if the axis separates the boxes. if so, return 0.
-  //   * find the depth of the penetration along the separating axis (s2)
-  //   * if this is the largest depth so far, record it.
-  // the normal vector will be set to the separating axis with the smallest
-  // depth. note: normalR is set to point to a column of R1 or R2 if that is
-  // the smallest depth normal so far. otherwise normalR is 0 and normalC is
-  // set to a vector relative to body 1. invert_normal is 1 if the sign of
-  // the normal should be flipped.
+	// for all 15 possible separating axes:
+	//   * see if the axis separates the boxes. if so, return 0.
+	//   * find the depth of the penetration along the separating axis (s2)
+	//   * if this is the largest depth so far, record it.
+	// the normal vector will be set to the separating axis with the smallest
+	// depth. note: normalR is set to point to a column of R1 or R2 if that is
+	// the smallest depth normal so far. otherwise normalR is 0 and normalC is
+	// set to a vector relative to body 1. invert_normal is 1 if the sign of
+	// the normal should be flipped.
 
-#define TST(expr1,expr2,norm,cc) \
-  s2 = btFabs(expr1) - (expr2); \
-  if (s2 > 0) return 0; \
-  if (s2 > s) { \
-    s = s2; \
-    normalR = norm; \
-    invert_normal = ((expr1) < 0); \
-    code = (cc); \
-  }
+#define TST(expr1, expr2, norm, cc)    \
+	s2 = btFabs(expr1) - (expr2);      \
+	if (s2 > 0) return 0;              \
+	if (s2 > s)                        \
+	{                                  \
+		s = s2;                        \
+		normalR = norm;                \
+		invert_normal = ((expr1) < 0); \
+		code = (cc);                   \
+	}
 
-  s = -dInfinity;
-  invert_normal = 0;
-  code = 0;
+	s = -dInfinity;
+	invert_normal = 0;
+	code = 0;
 
-  // separating axis = u1,u2,u3
-  TST (pp[0],(A[0] + B[0]*Q11 + B[1]*Q12 + B[2]*Q13),R1+0,1);
-  TST (pp[1],(A[1] + B[0]*Q21 + B[1]*Q22 + B[2]*Q23),R1+1,2);
-  TST (pp[2],(A[2] + B[0]*Q31 + B[1]*Q32 + B[2]*Q33),R1+2,3);
+	// separating axis = u1,u2,u3
+	TST(pp[0], (A[0] + B[0] * Q11 + B[1] * Q12 + B[2] * Q13), R1 + 0, 1);
+	TST(pp[1], (A[1] + B[0] * Q21 + B[1] * Q22 + B[2] * Q23), R1 + 1, 2);
+	TST(pp[2], (A[2] + B[0] * Q31 + B[1] * Q32 + B[2] * Q33), R1 + 2, 3);
 
-  // separating axis = v1,v2,v3
-  TST (dDOT41(R2+0,p),(A[0]*Q11 + A[1]*Q21 + A[2]*Q31 + B[0]),R2+0,4);
-  TST (dDOT41(R2+1,p),(A[0]*Q12 + A[1]*Q22 + A[2]*Q32 + B[1]),R2+1,5);
-  TST (dDOT41(R2+2,p),(A[0]*Q13 + A[1]*Q23 + A[2]*Q33 + B[2]),R2+2,6);
+	// separating axis = v1,v2,v3
+	TST(dDOT41(R2 + 0, p), (A[0] * Q11 + A[1] * Q21 + A[2] * Q31 + B[0]), R2 + 0, 4);
+	TST(dDOT41(R2 + 1, p), (A[0] * Q12 + A[1] * Q22 + A[2] * Q32 + B[1]), R2 + 1, 5);
+	TST(dDOT41(R2 + 2, p), (A[0] * Q13 + A[1] * Q23 + A[2] * Q33 + B[2]), R2 + 2, 6);
 
-  // note: cross product axes need to be scaled when s is computed.
-  // normal (n1,n2,n3) is relative to box 1.
+	// note: cross product axes need to be scaled when s is computed.
+	// normal (n1,n2,n3) is relative to box 1.
 #undef TST
-#define TST(expr1,expr2,n1,n2,n3,cc) \
-  s2 = btFabs(expr1) - (expr2); \
-  if (s2 > SIMD_EPSILON) return 0; \
-  l = btSqrt((n1)*(n1) + (n2)*(n2) + (n3)*(n3)); \
-  if (l > SIMD_EPSILON) { \
-    s2 /= l; \
-    if (s2*fudge_factor > s) { \
-      s = s2; \
-      normalR = 0; \
-      normalC[0] = (n1)/l; normalC[1] = (n2)/l; normalC[2] = (n3)/l; \
-      invert_normal = ((expr1) < 0); \
-      code = (cc); \
-    } \
-  }
+#define TST(expr1, expr2, n1, n2, n3, cc)                \
+	s2 = btFabs(expr1) - (expr2);                        \
+	if (s2 > SIMD_EPSILON) return 0;                     \
+	l = btSqrt((n1) * (n1) + (n2) * (n2) + (n3) * (n3)); \
+	if (l > SIMD_EPSILON)                                \
+	{                                                    \
+		s2 /= l;                                         \
+		if (s2 * fudge_factor > s)                       \
+		{                                                \
+			s = s2;                                      \
+			normalR = 0;                                 \
+			normalC[0] = (n1) / l;                       \
+			normalC[1] = (n2) / l;                       \
+			normalC[2] = (n3) / l;                       \
+			invert_normal = ((expr1) < 0);               \
+			code = (cc);                                 \
+		}                                                \
+	}
 
-  btScalar fudge2 (1.0e-5f);
+	btScalar fudge2(1.0e-5f);
 
-  Q11 += fudge2;
-  Q12 += fudge2;
-  Q13 += fudge2;
+	Q11 += fudge2;
+	Q12 += fudge2;
+	Q13 += fudge2;
 
-  Q21 += fudge2;
-  Q22 += fudge2;
-  Q23 += fudge2;
+	Q21 += fudge2;
+	Q22 += fudge2;
+	Q23 += fudge2;
 
-  Q31 += fudge2;
-  Q32 += fudge2;
-  Q33 += fudge2;
+	Q31 += fudge2;
+	Q32 += fudge2;
+	Q33 += fudge2;
 
-  // separating axis = u1 x (v1,v2,v3)
-  TST(pp[2]*R21-pp[1]*R31,(A[1]*Q31+A[2]*Q21+B[1]*Q13+B[2]*Q12),0,-R31,R21,7);
-  TST(pp[2]*R22-pp[1]*R32,(A[1]*Q32+A[2]*Q22+B[0]*Q13+B[2]*Q11),0,-R32,R22,8);
-  TST(pp[2]*R23-pp[1]*R33,(A[1]*Q33+A[2]*Q23+B[0]*Q12+B[1]*Q11),0,-R33,R23,9);
+	// separating axis = u1 x (v1,v2,v3)
+	TST(pp[2] * R21 - pp[1] * R31, (A[1] * Q31 + A[2] * Q21 + B[1] * Q13 + B[2] * Q12), 0, -R31, R21, 7);
+	TST(pp[2] * R22 - pp[1] * R32, (A[1] * Q32 + A[2] * Q22 + B[0] * Q13 + B[2] * Q11), 0, -R32, R22, 8);
+	TST(pp[2] * R23 - pp[1] * R33, (A[1] * Q33 + A[2] * Q23 + B[0] * Q12 + B[1] * Q11), 0, -R33, R23, 9);
 
-  // separating axis = u2 x (v1,v2,v3)
-  TST(pp[0]*R31-pp[2]*R11,(A[0]*Q31+A[2]*Q11+B[1]*Q23+B[2]*Q22),R31,0,-R11,10);
-  TST(pp[0]*R32-pp[2]*R12,(A[0]*Q32+A[2]*Q12+B[0]*Q23+B[2]*Q21),R32,0,-R12,11);
-  TST(pp[0]*R33-pp[2]*R13,(A[0]*Q33+A[2]*Q13+B[0]*Q22+B[1]*Q21),R33,0,-R13,12);
+	// separating axis = u2 x (v1,v2,v3)
+	TST(pp[0] * R31 - pp[2] * R11, (A[0] * Q31 + A[2] * Q11 + B[1] * Q23 + B[2] * Q22), R31, 0, -R11, 10);
+	TST(pp[0] * R32 - pp[2] * R12, (A[0] * Q32 + A[2] * Q12 + B[0] * Q23 + B[2] * Q21), R32, 0, -R12, 11);
+	TST(pp[0] * R33 - pp[2] * R13, (A[0] * Q33 + A[2] * Q13 + B[0] * Q22 + B[1] * Q21), R33, 0, -R13, 12);
 
-  // separating axis = u3 x (v1,v2,v3)
-  TST(pp[1]*R11-pp[0]*R21,(A[0]*Q21+A[1]*Q11+B[1]*Q33+B[2]*Q32),-R21,R11,0,13);
-  TST(pp[1]*R12-pp[0]*R22,(A[0]*Q22+A[1]*Q12+B[0]*Q33+B[2]*Q31),-R22,R12,0,14);
-  TST(pp[1]*R13-pp[0]*R23,(A[0]*Q23+A[1]*Q13+B[0]*Q32+B[1]*Q31),-R23,R13,0,15);
+	// separating axis = u3 x (v1,v2,v3)
+	TST(pp[1] * R11 - pp[0] * R21, (A[0] * Q21 + A[1] * Q11 + B[1] * Q33 + B[2] * Q32), -R21, R11, 0, 13);
+	TST(pp[1] * R12 - pp[0] * R22, (A[0] * Q22 + A[1] * Q12 + B[0] * Q33 + B[2] * Q31), -R22, R12, 0, 14);
+	TST(pp[1] * R13 - pp[0] * R23, (A[0] * Q23 + A[1] * Q13 + B[0] * Q32 + B[1] * Q31), -R23, R13, 0, 15);
 
 #undef TST
 
-  if (!code) return 0;
-
-  // if we get to this point, the boxes interpenetrate. compute the normal
-  // in global coordinates.
-  if (normalR) {
-    normal[0] = normalR[0];
-    normal[1] = normalR[4];
-    normal[2] = normalR[8];
-  }
-  else {
-    dMULTIPLY0_331 (normal,R1,normalC);
-  }
-  if (invert_normal) {
-    normal[0] = -normal[0];
-    normal[1] = -normal[1];
-    normal[2] = -normal[2];
-  }
-  *depth = -s;
-
-  // compute contact point(s)
-
-  if (code > 6) {
-    // an edge from box 1 touches an edge from box 2.
-    // find a point pa on the intersecting edge of box 1
-    btVector3 pa;
-    btScalar sign;
-    for (i=0; i<3; i++) pa[i] = p1[i];
-    for (j=0; j<3; j++) {
-      sign = (dDOT14(normal,R1+j) > 0) ? btScalar(1.0) : btScalar(-1.0);
-      for (i=0; i<3; i++) pa[i] += sign * A[j] * R1[i*4+j];
-    }
-
-    // find a point pb on the intersecting edge of box 2
-    btVector3 pb;
-    for (i=0; i<3; i++) pb[i] = p2[i];
-    for (j=0; j<3; j++) {
-      sign = (dDOT14(normal,R2+j) > 0) ? btScalar(-1.0) : btScalar(1.0);
-      for (i=0; i<3; i++) pb[i] += sign * B[j] * R2[i*4+j];
-    }
-
-    btScalar alpha,beta;
-    btVector3 ua,ub;
-    for (i=0; i<3; i++) ua[i] = R1[((code)-7)/3 + i*4];
-    for (i=0; i<3; i++) ub[i] = R2[((code)-7)%3 + i*4];
-
-    dLineClosestApproach (pa,ua,pb,ub,&alpha,&beta);
-    for (i=0; i<3; i++) pa[i] += ua[i]*alpha;
-    for (i=0; i<3; i++) pb[i] += ub[i]*beta;
+	if (!code) return 0;
 
+	// if we get to this point, the boxes interpenetrate. compute the normal
+	// in global coordinates.
+	if (normalR)
 	{
-		
-		//contact[0].pos[i] = btScalar(0.5)*(pa[i]+pb[i]);
-		//contact[0].depth = *depth;
-		btVector3 pointInWorld;
+		normal[0] = normalR[0];
+		normal[1] = normalR[4];
+		normal[2] = normalR[8];
+	}
+	else
+	{
+		dMULTIPLY0_331(normal, R1, normalC);
+	}
+	if (invert_normal)
+	{
+		normal[0] = -normal[0];
+		normal[1] = -normal[1];
+		normal[2] = -normal[2];
+	}
+	*depth = -s;
+
+	// compute contact point(s)
+
+	if (code > 6)
+	{
+		// an edge from box 1 touches an edge from box 2.
+		// find a point pa on the intersecting edge of box 1
+		btVector3 pa;
+		btScalar sign;
+		for (i = 0; i < 3; i++) pa[i] = p1[i];
+		for (j = 0; j < 3; j++)
+		{
+			sign = (dDOT14(normal, R1 + j) > 0) ? btScalar(1.0) : btScalar(-1.0);
+			for (i = 0; i < 3; i++) pa[i] += sign * A[j] * R1[i * 4 + j];
+		}
+
+		// find a point pb on the intersecting edge of box 2
+		btVector3 pb;
+		for (i = 0; i < 3; i++) pb[i] = p2[i];
+		for (j = 0; j < 3; j++)
+		{
+			sign = (dDOT14(normal, R2 + j) > 0) ? btScalar(-1.0) : btScalar(1.0);
+			for (i = 0; i < 3; i++) pb[i] += sign * B[j] * R2[i * 4 + j];
+		}
+
+		btScalar alpha, beta;
+		btVector3 ua, ub;
+		for (i = 0; i < 3; i++) ua[i] = R1[((code)-7) / 3 + i * 4];
+		for (i = 0; i < 3; i++) ub[i] = R2[((code)-7) % 3 + i * 4];
+
+		dLineClosestApproach(pa, ua, pb, ub, &alpha, &beta);
+		for (i = 0; i < 3; i++) pa[i] += ua[i] * alpha;
+		for (i = 0; i < 3; i++) pb[i] += ub[i] * beta;
+
+		{
+			//contact[0].pos[i] = btScalar(0.5)*(pa[i]+pb[i]);
+			//contact[0].depth = *depth;
+			btVector3 pointInWorld;
 
 #ifdef USE_CENTER_POINT
-	    for (i=0; i<3; i++) 
-			pointInWorld[i] = (pa[i]+pb[i])*btScalar(0.5);
-		output.addContactPoint(-normal,pointInWorld,-*depth);
+			for (i = 0; i < 3; i++)
+				pointInWorld[i] = (pa[i] + pb[i]) * btScalar(0.5);
+			output.addContactPoint(-normal, pointInWorld, -*depth);
 #else
-		output.addContactPoint(-normal,pb,-*depth);
+			output.addContactPoint(-normal, pb, -*depth);
 
-#endif //
-		*return_code = code;
+#endif  //
+			*return_code = code;
+		}
+		return 1;
 	}
-    return 1;
-  }
 
-  // okay, we have a face-something intersection (because the separating
-  // axis is perpendicular to a face). define face 'a' to be the reference
-  // face (i.e. the normal vector is perpendicular to this) and face 'b' to be
-  // the incident face (the closest face of the other box).
+	// okay, we have a face-something intersection (because the separating
+	// axis is perpendicular to a face). define face 'a' to be the reference
+	// face (i.e. the normal vector is perpendicular to this) and face 'b' to be
+	// the incident face (the closest face of the other box).
 
-  const btScalar *Ra,*Rb,*pa,*pb,*Sa,*Sb;
-  if (code <= 3) {
-    Ra = R1;
-    Rb = R2;
-    pa = p1;
-    pb = p2;
-    Sa = A;
-    Sb = B;
-  }
-  else {
-    Ra = R2;
-    Rb = R1;
-    pa = p2;
-    pb = p1;
-    Sa = B;
-    Sb = A;
-  }
-
-  // nr = normal vector of reference face dotted with axes of incident box.
-  // anr = absolute values of nr.
-  btVector3 normal2,nr,anr;
-  if (code <= 3) {
-    normal2[0] = normal[0];
-    normal2[1] = normal[1];
-    normal2[2] = normal[2];
-  }
-  else {
-    normal2[0] = -normal[0];
-    normal2[1] = -normal[1];
-    normal2[2] = -normal[2];
-  }
-  dMULTIPLY1_331 (nr,Rb,normal2);
-  anr[0] = btFabs (nr[0]);
-  anr[1] = btFabs (nr[1]);
-  anr[2] = btFabs (nr[2]);
-
-  // find the largest compontent of anr: this corresponds to the normal
-  // for the indident face. the other axis numbers of the indicent face
-  // are stored in a1,a2.
-  int lanr,a1,a2;
-  if (anr[1] > anr[0]) {
-    if (anr[1] > anr[2]) {
-      a1 = 0;
-      lanr = 1;
-      a2 = 2;
-    }
-    else {
-      a1 = 0;
-      a2 = 1;
-      lanr = 2;
-    }
-  }
-  else {
-    if (anr[0] > anr[2]) {
-      lanr = 0;
-      a1 = 1;
-      a2 = 2;
-    }
-    else {
-      a1 = 0;
-      a2 = 1;
-      lanr = 2;
-    }
-  }
-
-  // compute center point of incident face, in reference-face coordinates
-  btVector3 center;
-  if (nr[lanr] < 0) {
-    for (i=0; i<3; i++) center[i] = pb[i] - pa[i] + Sb[lanr] * Rb[i*4+lanr];
-  }
-  else {
-    for (i=0; i<3; i++) center[i] = pb[i] - pa[i] - Sb[lanr] * Rb[i*4+lanr];
-  }
-
-  // find the normal and non-normal axis numbers of the reference box
-  int codeN,code1,code2;
-  if (code <= 3) codeN = code-1; else codeN = code-4;
-  if (codeN==0) {
-    code1 = 1;
-    code2 = 2;
-  }
-  else if (codeN==1) {
-    code1 = 0;
-    code2 = 2;
-  }
-  else {
-    code1 = 0;
-    code2 = 1;
-  }
-
-  // find the four corners of the incident face, in reference-face coordinates
-  btScalar quad[8];	// 2D coordinate of incident face (x,y pairs)
-  btScalar c1,c2,m11,m12,m21,m22;
-  c1 = dDOT14 (center,Ra+code1);
-  c2 = dDOT14 (center,Ra+code2);
-  // optimize this? - we have already computed this data above, but it is not
-  // stored in an easy-to-index format. for now it's quicker just to recompute
-  // the four dot products.
-  m11 = dDOT44 (Ra+code1,Rb+a1);
-  m12 = dDOT44 (Ra+code1,Rb+a2);
-  m21 = dDOT44 (Ra+code2,Rb+a1);
-  m22 = dDOT44 (Ra+code2,Rb+a2);
-  {
-    btScalar k1 = m11*Sb[a1];
-    btScalar k2 = m21*Sb[a1];
-    btScalar k3 = m12*Sb[a2];
-    btScalar k4 = m22*Sb[a2];
-    quad[0] = c1 - k1 - k3;
-    quad[1] = c2 - k2 - k4;
-    quad[2] = c1 - k1 + k3;
-    quad[3] = c2 - k2 + k4;
-    quad[4] = c1 + k1 + k3;
-    quad[5] = c2 + k2 + k4;
-    quad[6] = c1 + k1 - k3;
-    quad[7] = c2 + k2 - k4;
-  }
-
-  // find the size of the reference face
-  btScalar rect[2];
-  rect[0] = Sa[code1];
-  rect[1] = Sa[code2];
-
-  // intersect the incident and reference faces
-  btScalar ret[16];
-  int n = intersectRectQuad2 (rect,quad,ret);
-  if (n < 1) return 0;		// this should never happen
-
-  // convert the intersection points into reference-face coordinates,
-  // and compute the contact position and depth for each point. only keep
-  // those points that have a positive (penetrating) depth. delete points in
-  // the 'ret' array as necessary so that 'point' and 'ret' correspond.
-  btScalar point[3*8];		// penetrating contact points
-  btScalar dep[8];			// depths for those points
-  btScalar det1 = 1.f/(m11*m22 - m12*m21);
-  m11 *= det1;
-  m12 *= det1;
-  m21 *= det1;
-  m22 *= det1;
-  int cnum = 0;			// number of penetrating contact points found
-  for (j=0; j < n; j++) {
-    btScalar k1 =  m22*(ret[j*2]-c1) - m12*(ret[j*2+1]-c2);
-    btScalar k2 = -m21*(ret[j*2]-c1) + m11*(ret[j*2+1]-c2);
-    for (i=0; i<3; i++) point[cnum*3+i] =
-			  center[i] + k1*Rb[i*4+a1] + k2*Rb[i*4+a2];
-    dep[cnum] = Sa[codeN] - dDOT(normal2,point+cnum*3);
-    if (dep[cnum] >= 0) {
-      ret[cnum*2] = ret[j*2];
-      ret[cnum*2+1] = ret[j*2+1];
-      cnum++;
-    }
-  }
-  if (cnum < 1) return 0;	// this should never happen
-
-  // we can't generate more contacts than we actually have
-  if (maxc > cnum) maxc = cnum;
-  if (maxc < 1) maxc = 1;
-
-  if (cnum <= maxc) {
-
-	  if (code<4) 
-	  {
-    // we have less contacts than we need, so we use them all
-    for (j=0; j < cnum; j++) 
+	const btScalar *Ra, *Rb, *pa, *pb, *Sa, *Sb;
+	if (code <= 3)
 	{
-		btVector3 pointInWorld;
-		for (i=0; i<3; i++) 
-			pointInWorld[i] = point[j*3+i] + pa[i];
-		output.addContactPoint(-normal,pointInWorld,-dep[j]);
+		Ra = R1;
+		Rb = R2;
+		pa = p1;
+		pb = p2;
+		Sa = A;
+		Sb = B;
+	}
+	else
+	{
+		Ra = R2;
+		Rb = R1;
+		pa = p2;
+		pb = p1;
+		Sa = B;
+		Sb = A;
+	}
 
-    }
-	  } else
-	  {
-		  // we have less contacts than we need, so we use them all
-		for (j=0; j < cnum; j++) 
+	// nr = normal vector of reference face dotted with axes of incident box.
+	// anr = absolute values of nr.
+	btVector3 normal2, nr, anr;
+	if (code <= 3)
+	{
+		normal2[0] = normal[0];
+		normal2[1] = normal[1];
+		normal2[2] = normal[2];
+	}
+	else
+	{
+		normal2[0] = -normal[0];
+		normal2[1] = -normal[1];
+		normal2[2] = -normal[2];
+	}
+	dMULTIPLY1_331(nr, Rb, normal2);
+	anr[0] = btFabs(nr[0]);
+	anr[1] = btFabs(nr[1]);
+	anr[2] = btFabs(nr[2]);
+
+	// find the largest compontent of anr: this corresponds to the normal
+	// for the indident face. the other axis numbers of the indicent face
+	// are stored in a1,a2.
+	int lanr, a1, a2;
+	if (anr[1] > anr[0])
+	{
+		if (anr[1] > anr[2])
 		{
-			btVector3 pointInWorld;
-			for (i=0; i<3; i++) 
-				pointInWorld[i] = point[j*3+i] + pa[i]-normal[i]*dep[j];
+			a1 = 0;
+			lanr = 1;
+			a2 = 2;
+		}
+		else
+		{
+			a1 = 0;
+			a2 = 1;
+			lanr = 2;
+		}
+	}
+	else
+	{
+		if (anr[0] > anr[2])
+		{
+			lanr = 0;
+			a1 = 1;
+			a2 = 2;
+		}
+		else
+		{
+			a1 = 0;
+			a2 = 1;
+			lanr = 2;
+		}
+	}
+
+	// compute center point of incident face, in reference-face coordinates
+	btVector3 center;
+	if (nr[lanr] < 0)
+	{
+		for (i = 0; i < 3; i++) center[i] = pb[i] - pa[i] + Sb[lanr] * Rb[i * 4 + lanr];
+	}
+	else
+	{
+		for (i = 0; i < 3; i++) center[i] = pb[i] - pa[i] - Sb[lanr] * Rb[i * 4 + lanr];
+	}
+
+	// find the normal and non-normal axis numbers of the reference box
+	int codeN, code1, code2;
+	if (code <= 3)
+		codeN = code - 1;
+	else
+		codeN = code - 4;
+	if (codeN == 0)
+	{
+		code1 = 1;
+		code2 = 2;
+	}
+	else if (codeN == 1)
+	{
+		code1 = 0;
+		code2 = 2;
+	}
+	else
+	{
+		code1 = 0;
+		code2 = 1;
+	}
+
+	// find the four corners of the incident face, in reference-face coordinates
+	btScalar quad[8];  // 2D coordinate of incident face (x,y pairs)
+	btScalar c1, c2, m11, m12, m21, m22;
+	c1 = dDOT14(center, Ra + code1);
+	c2 = dDOT14(center, Ra + code2);
+	// optimize this? - we have already computed this data above, but it is not
+	// stored in an easy-to-index format. for now it's quicker just to recompute
+	// the four dot products.
+	m11 = dDOT44(Ra + code1, Rb + a1);
+	m12 = dDOT44(Ra + code1, Rb + a2);
+	m21 = dDOT44(Ra + code2, Rb + a1);
+	m22 = dDOT44(Ra + code2, Rb + a2);
+	{
+		btScalar k1 = m11 * Sb[a1];
+		btScalar k2 = m21 * Sb[a1];
+		btScalar k3 = m12 * Sb[a2];
+		btScalar k4 = m22 * Sb[a2];
+		quad[0] = c1 - k1 - k3;
+		quad[1] = c2 - k2 - k4;
+		quad[2] = c1 - k1 + k3;
+		quad[3] = c2 - k2 + k4;
+		quad[4] = c1 + k1 + k3;
+		quad[5] = c2 + k2 + k4;
+		quad[6] = c1 + k1 - k3;
+		quad[7] = c2 + k2 - k4;
+	}
+
+	// find the size of the reference face
+	btScalar rect[2];
+	rect[0] = Sa[code1];
+	rect[1] = Sa[code2];
+
+	// intersect the incident and reference faces
+	btScalar ret[16];
+	int n = intersectRectQuad2(rect, quad, ret);
+	if (n < 1) return 0;  // this should never happen
+
+	// convert the intersection points into reference-face coordinates,
+	// and compute the contact position and depth for each point. only keep
+	// those points that have a positive (penetrating) depth. delete points in
+	// the 'ret' array as necessary so that 'point' and 'ret' correspond.
+	btScalar point[3 * 8];  // penetrating contact points
+	btScalar dep[8];        // depths for those points
+	btScalar det1 = 1.f / (m11 * m22 - m12 * m21);
+	m11 *= det1;
+	m12 *= det1;
+	m21 *= det1;
+	m22 *= det1;
+	int cnum = 0;  // number of penetrating contact points found
+	for (j = 0; j < n; j++)
+	{
+		btScalar k1 = m22 * (ret[j * 2] - c1) - m12 * (ret[j * 2 + 1] - c2);
+		btScalar k2 = -m21 * (ret[j * 2] - c1) + m11 * (ret[j * 2 + 1] - c2);
+		for (i = 0; i < 3; i++) point[cnum * 3 + i] =
+									center[i] + k1 * Rb[i * 4 + a1] + k2 * Rb[i * 4 + a2];
+		dep[cnum] = Sa[codeN] - dDOT(normal2, point + cnum * 3);
+		if (dep[cnum] >= 0)
+		{
+			ret[cnum * 2] = ret[j * 2];
+			ret[cnum * 2 + 1] = ret[j * 2 + 1];
+			cnum++;
+		}
+	}
+	if (cnum < 1) return 0;  // this should never happen
+
+	// we can't generate more contacts than we actually have
+	if (maxc > cnum) maxc = cnum;
+	if (maxc < 1) maxc = 1;
+
+	if (cnum <= maxc)
+	{
+		if (code < 4)
+		{
+			// we have less contacts than we need, so we use them all
+			for (j = 0; j < cnum; j++)
+			{
+				btVector3 pointInWorld;
+				for (i = 0; i < 3; i++)
+					pointInWorld[i] = point[j * 3 + i] + pa[i];
+				output.addContactPoint(-normal, pointInWorld, -dep[j]);
+			}
+		}
+		else
+		{
+			// we have less contacts than we need, so we use them all
+			for (j = 0; j < cnum; j++)
+			{
+				btVector3 pointInWorld;
+				for (i = 0; i < 3; i++)
+					pointInWorld[i] = point[j * 3 + i] + pa[i] - normal[i] * dep[j];
 				//pointInWorld[i] = point[j*3+i] + pa[i];
-			output.addContactPoint(-normal,pointInWorld,-dep[j]);
+				output.addContactPoint(-normal, pointInWorld, -dep[j]);
+			}
 		}
-	  }
-  }
-  else {
-    // we have more contacts than are wanted, some of them must be culled.
-    // find the deepest point, it is always the first contact.
-    int i1 = 0;
-    btScalar maxdepth = dep[0];
-    for (i=1; i<cnum; i++) {
-      if (dep[i] > maxdepth) {
-	maxdepth = dep[i];
-	i1 = i;
-      }
-    }
-
-    int iret[8];
-    cullPoints2 (cnum,ret,maxc,i1,iret);
-
-    for (j=0; j < maxc; j++) {
-//      dContactGeom *con = CONTACT(contact,skip*j);
-  //    for (i=0; i<3; i++) con->pos[i] = point[iret[j]*3+i] + pa[i];
-    //  con->depth = dep[iret[j]];
-
-		btVector3 posInWorld;
-		for (i=0; i<3; i++) 
-			posInWorld[i] = point[iret[j]*3+i] + pa[i];
-		if (code<4) 
-	   {
-			output.addContactPoint(-normal,posInWorld,-dep[iret[j]]);
-		} else
+	}
+	else
+	{
+		// we have more contacts than are wanted, some of them must be culled.
+		// find the deepest point, it is always the first contact.
+		int i1 = 0;
+		btScalar maxdepth = dep[0];
+		for (i = 1; i < cnum; i++)
 		{
-			output.addContactPoint(-normal,posInWorld-normal*dep[iret[j]],-dep[iret[j]]);
+			if (dep[i] > maxdepth)
+			{
+				maxdepth = dep[i];
+				i1 = i;
+			}
 		}
-    }
-    cnum = maxc;
-  }
 
-  *return_code = code;
-  return cnum;
+		int iret[8];
+		cullPoints2(cnum, ret, maxc, i1, iret);
+
+		for (j = 0; j < maxc; j++)
+		{
+			//      dContactGeom *con = CONTACT(contact,skip*j);
+			//    for (i=0; i<3; i++) con->pos[i] = point[iret[j]*3+i] + pa[i];
+			//  con->depth = dep[iret[j]];
+
+			btVector3 posInWorld;
+			for (i = 0; i < 3; i++)
+				posInWorld[i] = point[iret[j] * 3 + i] + pa[i];
+			if (code < 4)
+			{
+				output.addContactPoint(-normal, posInWorld, -dep[iret[j]]);
+			}
+			else
+			{
+				output.addContactPoint(-normal, posInWorld - normal * dep[iret[j]], -dep[iret[j]]);
+			}
+		}
+		cnum = maxc;
+	}
+
+	*return_code = code;
+	return cnum;
 }
 
-void	btBoxBoxDetector::getClosestPoints(const ClosestPointInput& input,Result& output,class btIDebugDraw* /*debugDraw*/,bool /*swapResults*/)
+void btBoxBoxDetector::getClosestPoints(const ClosestPointInput& input, Result& output, class btIDebugDraw* /*debugDraw*/, bool /*swapResults*/)
 {
-	
 	const btTransform& transformA = input.m_transformA;
 	const btTransform& transformB = input.m_transformB;
-	
+
 	int skip = 0;
-	dContactGeom *contact = 0;
+	dContactGeom* contact = 0;
 
 	dMatrix3 R1;
 	dMatrix3 R2;
 
-	for (int j=0;j<3;j++)
+	for (int j = 0; j < 3; j++)
 	{
-		R1[0+4*j] = transformA.getBasis()[j].x();
-		R2[0+4*j] = transformB.getBasis()[j].x();
+		R1[0 + 4 * j] = transformA.getBasis()[j].x();
+		R2[0 + 4 * j] = transformB.getBasis()[j].x();
 
-		R1[1+4*j] = transformA.getBasis()[j].y();
-		R2[1+4*j] = transformB.getBasis()[j].y();
-
-
-		R1[2+4*j] = transformA.getBasis()[j].z();
-		R2[2+4*j] = transformB.getBasis()[j].z();
+		R1[1 + 4 * j] = transformA.getBasis()[j].y();
+		R2[1 + 4 * j] = transformB.getBasis()[j].y();
 
+		R1[2 + 4 * j] = transformA.getBasis()[j].z();
+		R2[2 + 4 * j] = transformB.getBasis()[j].z();
 	}
 
-	
-
 	btVector3 normal;
 	btScalar depth;
 	int return_code;
 	int maxc = 4;
 
-
-	dBoxBox2 (transformA.getOrigin(), 
-	R1,
-	2.f*m_box1->getHalfExtentsWithMargin(),
-	transformB.getOrigin(),
-	R2, 
-	2.f*m_box2->getHalfExtentsWithMargin(),
-	normal, &depth, &return_code,
-	maxc, contact, skip,
-	output
-	);
-
+	dBoxBox2(transformA.getOrigin(),
+			 R1,
+			 2.f * m_box1->getHalfExtentsWithMargin(),
+			 transformB.getOrigin(),
+			 R2,
+			 2.f * m_box2->getHalfExtentsWithMargin(),
+			 normal, &depth, &return_code,
+			 maxc, contact, skip,
+			 output);
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btBoxBoxDetector.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btBoxBoxDetector.h
index 39243777051..9f7d988fc11 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btBoxBoxDetector.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btBoxBoxDetector.h
@@ -19,11 +19,9 @@ subject to the following restrictions:
 #ifndef BT_BOX_BOX_DETECTOR_H
 #define BT_BOX_BOX_DETECTOR_H
 
-
 class btBoxShape;
 #include "BulletCollision/NarrowPhaseCollision/btDiscreteCollisionDetectorInterface.h"
 
-
 /// btBoxBoxDetector wraps the ODE box-box collision detector
 /// re-distributed under the Zlib license with permission from Russell L. Smith
 struct btBoxBoxDetector : public btDiscreteCollisionDetectorInterface
@@ -32,13 +30,11 @@ struct btBoxBoxDetector : public btDiscreteCollisionDetectorInterface
 	const btBoxShape* m_box2;
 
 public:
+	btBoxBoxDetector(const btBoxShape* box1, const btBoxShape* box2);
 
-	btBoxBoxDetector(const btBoxShape* box1,const btBoxShape* box2);
-
-	virtual ~btBoxBoxDetector() {};
-
-	virtual void	getClosestPoints(const ClosestPointInput& input,Result& output,class btIDebugDraw* debugDraw,bool swapResults=false);
+	virtual ~btBoxBoxDetector(){};
 
+	virtual void getClosestPoints(const ClosestPointInput& input, Result& output, class btIDebugDraw* debugDraw, bool swapResults = false);
 };
 
-#endif //BT_BOX_BOX_DETECTOR_H
+#endif  //BT_BOX_BOX_DETECTOR_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionConfiguration.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionConfiguration.h
index 35f77d4e654..d6e15f555bf 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionConfiguration.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionConfiguration.h
@@ -23,11 +23,9 @@ class btPoolAllocator;
 ///btCollisionConfiguration allows to configure Bullet collision detection
 ///stack allocator size, default collision algorithms and persistent manifold pool size
 ///@todo: describe the meaning
-class	btCollisionConfiguration
+class btCollisionConfiguration
 {
-
 public:
-
 	virtual ~btCollisionConfiguration()
 	{
 	}
@@ -37,13 +35,9 @@ public:
 
 	virtual btPoolAllocator* getCollisionAlgorithmPool() = 0;
 
-
-	virtual btCollisionAlgorithmCreateFunc* getCollisionAlgorithmCreateFunc(int proxyType0,int proxyType1) =0;
+	virtual btCollisionAlgorithmCreateFunc* getCollisionAlgorithmCreateFunc(int proxyType0, int proxyType1) = 0;
 
 	virtual btCollisionAlgorithmCreateFunc* getClosestPointsAlgorithmCreateFunc(int proxyType0, int proxyType1) = 0;
-
-
 };
 
-#endif //BT_COLLISION_CONFIGURATION
-
+#endif  //BT_COLLISION_CONFIGURATION
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionCreateFunc.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionCreateFunc.h
index 62ee66c4e9a..bd812849393 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionCreateFunc.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionCreateFunc.h
@@ -26,20 +26,18 @@ struct btCollisionAlgorithmConstructionInfo;
 struct btCollisionAlgorithmCreateFunc
 {
 	bool m_swapped;
-	
+
 	btCollisionAlgorithmCreateFunc()
-		:m_swapped(false)
+		: m_swapped(false)
 	{
 	}
 	virtual ~btCollisionAlgorithmCreateFunc(){};
 
-	virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& , const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+	virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo&, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 	{
-		
 		(void)body0Wrap;
 		(void)body1Wrap;
 		return 0;
 	}
 };
-#endif //BT_COLLISION_CREATE_FUNC
-
+#endif  //BT_COLLISION_CREATE_FUNC
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionDispatcher.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionDispatcher.cpp
index f8794dec47c..25b2b1ea464 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionDispatcher.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionDispatcher.cpp
@@ -13,8 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #include "btCollisionDispatcher.h"
 #include "LinearMath/btQuickprof.h"
 
@@ -31,40 +29,34 @@ subject to the following restrictions:
 #include <stdio.h>
 #endif
 
-
-btCollisionDispatcher::btCollisionDispatcher (btCollisionConfiguration* collisionConfiguration): 
-m_dispatcherFlags(btCollisionDispatcher::CD_USE_RELATIVE_CONTACT_BREAKING_THRESHOLD),
-	m_collisionConfiguration(collisionConfiguration)
+btCollisionDispatcher::btCollisionDispatcher(btCollisionConfiguration* collisionConfiguration) : m_dispatcherFlags(btCollisionDispatcher::CD_USE_RELATIVE_CONTACT_BREAKING_THRESHOLD),
+																								 m_collisionConfiguration(collisionConfiguration)
 {
 	int i;
 
 	setNearCallback(defaultNearCallback);
-	
+
 	m_collisionAlgorithmPoolAllocator = collisionConfiguration->getCollisionAlgorithmPool();
 
 	m_persistentManifoldPoolAllocator = collisionConfiguration->getPersistentManifoldPool();
 
-	for (i=0;i<MAX_BROADPHASE_COLLISION_TYPES;i++)
+	for (i = 0; i < MAX_BROADPHASE_COLLISION_TYPES; i++)
 	{
-		for (int j=0;j<MAX_BROADPHASE_COLLISION_TYPES;j++)
+		for (int j = 0; j < MAX_BROADPHASE_COLLISION_TYPES; j++)
 		{
-			m_doubleDispatchContactPoints[i][j] = m_collisionConfiguration->getCollisionAlgorithmCreateFunc(i,j);
+			m_doubleDispatchContactPoints[i][j] = m_collisionConfiguration->getCollisionAlgorithmCreateFunc(i, j);
 			btAssert(m_doubleDispatchContactPoints[i][j]);
 			m_doubleDispatchClosestPoints[i][j] = m_collisionConfiguration->getClosestPointsAlgorithmCreateFunc(i, j);
-
 		}
 	}
-	
-	
 }
 
-
-void btCollisionDispatcher::registerCollisionCreateFunc(int proxyType0, int proxyType1, btCollisionAlgorithmCreateFunc *createFunc)
+void btCollisionDispatcher::registerCollisionCreateFunc(int proxyType0, int proxyType1, btCollisionAlgorithmCreateFunc* createFunc)
 {
 	m_doubleDispatchContactPoints[proxyType0][proxyType1] = createFunc;
 }
 
-void btCollisionDispatcher::registerClosestPointsCreateFunc(int proxyType0, int proxyType1, btCollisionAlgorithmCreateFunc *createFunc)
+void btCollisionDispatcher::registerClosestPointsCreateFunc(int proxyType0, int proxyType1, btCollisionAlgorithmCreateFunc* createFunc)
 {
 	m_doubleDispatchClosestPoints[proxyType0][proxyType1] = createFunc;
 }
@@ -73,35 +65,33 @@ btCollisionDispatcher::~btCollisionDispatcher()
 {
 }
 
-btPersistentManifold*	btCollisionDispatcher::getNewManifold(const btCollisionObject* body0,const btCollisionObject* body1) 
-{ 
+btPersistentManifold* btCollisionDispatcher::getNewManifold(const btCollisionObject* body0, const btCollisionObject* body1)
+{
 	//btAssert(gNumManifold < 65535);
-	
-
 
 	//optional relative contact breaking threshold, turned on by default (use setDispatcherFlags to switch off feature for improved performance)
-	
-	btScalar contactBreakingThreshold =  (m_dispatcherFlags & btCollisionDispatcher::CD_USE_RELATIVE_CONTACT_BREAKING_THRESHOLD) ? 
-		btMin(body0->getCollisionShape()->getContactBreakingThreshold(gContactBreakingThreshold) , body1->getCollisionShape()->getContactBreakingThreshold(gContactBreakingThreshold))
-		: gContactBreakingThreshold ;
 
-	btScalar contactProcessingThreshold = btMin(body0->getContactProcessingThreshold(),body1->getContactProcessingThreshold());
-		
- 	void* mem = m_persistentManifoldPoolAllocator->allocate( sizeof( btPersistentManifold ) );
-    if (NULL == mem)
+	btScalar contactBreakingThreshold = (m_dispatcherFlags & btCollisionDispatcher::CD_USE_RELATIVE_CONTACT_BREAKING_THRESHOLD) ? btMin(body0->getCollisionShape()->getContactBreakingThreshold(gContactBreakingThreshold), body1->getCollisionShape()->getContactBreakingThreshold(gContactBreakingThreshold))
+																																: gContactBreakingThreshold;
+
+	btScalar contactProcessingThreshold = btMin(body0->getContactProcessingThreshold(), body1->getContactProcessingThreshold());
+
+	void* mem = m_persistentManifoldPoolAllocator->allocate(sizeof(btPersistentManifold));
+	if (NULL == mem)
 	{
-        //we got a pool memory overflow, by default we fallback to dynamically allocate memory. If we require a contiguous contact pool then assert.
-		if ((m_dispatcherFlags&CD_DISABLE_CONTACTPOOL_DYNAMIC_ALLOCATION)==0)
+		//we got a pool memory overflow, by default we fallback to dynamically allocate memory. If we require a contiguous contact pool then assert.
+		if ((m_dispatcherFlags & CD_DISABLE_CONTACTPOOL_DYNAMIC_ALLOCATION) == 0)
 		{
-			mem = btAlignedAlloc(sizeof(btPersistentManifold),16);
-		} else
+			mem = btAlignedAlloc(sizeof(btPersistentManifold), 16);
+		}
+		else
 		{
 			btAssert(0);
 			//make sure to increase the m_defaultMaxPersistentManifoldPoolSize in the btDefaultCollisionConstructionInfo/btDefaultCollisionConfiguration
 			return 0;
 		}
 	}
-	btPersistentManifold* manifold = new(mem) btPersistentManifold (body0,body1,0,contactBreakingThreshold,contactProcessingThreshold);
+	btPersistentManifold* manifold = new (mem) btPersistentManifold(body0, body1, 0, contactBreakingThreshold, contactProcessingThreshold);
 	manifold->m_index1a = m_manifoldsPtr.size();
 	m_manifoldsPtr.push_back(manifold);
 
@@ -113,17 +103,14 @@ void btCollisionDispatcher::clearManifold(btPersistentManifold* manifold)
 	manifold->clearManifold();
 }
 
-	
 void btCollisionDispatcher::releaseManifold(btPersistentManifold* manifold)
 {
-	
-
 	//printf("releaseManifold: gNumManifold %d\n",gNumManifold);
 	clearManifold(manifold);
 
 	int findIndex = manifold->m_index1a;
 	btAssert(findIndex < m_manifoldsPtr.size());
-	m_manifoldsPtr.swap(findIndex,m_manifoldsPtr.size()-1);
+	m_manifoldsPtr.swap(findIndex, m_manifoldsPtr.size() - 1);
 	m_manifoldsPtr[findIndex]->m_index1a = findIndex;
 	m_manifoldsPtr.pop_back();
 
@@ -131,19 +118,15 @@ void btCollisionDispatcher::releaseManifold(btPersistentManifold* manifold)
 	if (m_persistentManifoldPoolAllocator->validPtr(manifold))
 	{
 		m_persistentManifoldPoolAllocator->freeMemory(manifold);
-	} else
+	}
+	else
 	{
 		btAlignedFree(manifold);
 	}
-	
 }
 
-	
-
-
-btCollisionAlgorithm* btCollisionDispatcher::findAlgorithm(const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,btPersistentManifold* sharedManifold, ebtDispatcherQueryType algoType)
+btCollisionAlgorithm* btCollisionDispatcher::findAlgorithm(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, btPersistentManifold* sharedManifold, ebtDispatcherQueryType algoType)
 {
-	
 	btCollisionAlgorithmConstructionInfo ci;
 
 	ci.m_dispatcher1 = this;
@@ -161,21 +144,18 @@ btCollisionAlgorithm* btCollisionDispatcher::findAlgorithm(const btCollisionObje
 	return algo;
 }
 
-
-
-
-bool	btCollisionDispatcher::needsResponse(const btCollisionObject* body0,const btCollisionObject* body1)
+bool btCollisionDispatcher::needsResponse(const btCollisionObject* body0, const btCollisionObject* body1)
 {
 	//here you can do filtering
-	bool hasResponse = 
+	bool hasResponse =
 		(body0->hasContactResponse() && body1->hasContactResponse());
 	//no response between two static/kinematic bodies:
 	hasResponse = hasResponse &&
-		((!body0->isStaticOrKinematicObject()) ||(! body1->isStaticOrKinematicObject()));
+				  ((!body0->isStaticOrKinematicObject()) || (!body1->isStaticOrKinematicObject()));
 	return hasResponse;
 }
 
-bool	btCollisionDispatcher::needsCollision(const btCollisionObject* body0,const btCollisionObject* body1)
+bool btCollisionDispatcher::needsCollision(const btCollisionObject* body0, const btCollisionObject* body1)
 {
 	btAssert(body0);
 	btAssert(body1);
@@ -192,31 +172,27 @@ bool	btCollisionDispatcher::needsCollision(const btCollisionObject* body0,const
 			printf("warning btCollisionDispatcher::needsCollision: static-static collision!\n");
 		}
 	}
-#endif //BT_DEBUG
+#endif  //BT_DEBUG
 
 	if ((!body0->isActive()) && (!body1->isActive()))
 		needsCollision = false;
 	else if ((!body0->checkCollideWith(body1)) || (!body1->checkCollideWith(body0)))
 		needsCollision = false;
-	
-	return needsCollision ;
 
+	return needsCollision;
 }
 
-
-
 ///interface for iterating all overlapping collision pairs, no matter how those pairs are stored (array, set, map etc)
 ///this is useful for the collision dispatcher.
 class btCollisionPairCallback : public btOverlapCallback
 {
 	const btDispatcherInfo& m_dispatchInfo;
-	btCollisionDispatcher*	m_dispatcher;
+	btCollisionDispatcher* m_dispatcher;
 
 public:
-
-	btCollisionPairCallback(const btDispatcherInfo& dispatchInfo,btCollisionDispatcher*	dispatcher)
-	:m_dispatchInfo(dispatchInfo),
-	m_dispatcher(dispatcher)
+	btCollisionPairCallback(const btDispatcherInfo& dispatchInfo, btCollisionDispatcher* dispatcher)
+		: m_dispatchInfo(dispatchInfo),
+		  m_dispatcher(dispatcher)
 	{
 	}
 
@@ -228,87 +204,76 @@ public:
 	}
 	*/
 
-
 	virtual ~btCollisionPairCallback() {}
 
-
-	virtual bool	processOverlap(btBroadphasePair& pair)
+	virtual bool processOverlap(btBroadphasePair& pair)
 	{
-		(*m_dispatcher->getNearCallback())(pair,*m_dispatcher,m_dispatchInfo);
+		(*m_dispatcher->getNearCallback())(pair, *m_dispatcher, m_dispatchInfo);
 		return false;
 	}
 };
 
-
-
-
-void	btCollisionDispatcher::dispatchAllCollisionPairs(btOverlappingPairCache* pairCache,const btDispatcherInfo& dispatchInfo,btDispatcher* dispatcher) 
+void btCollisionDispatcher::dispatchAllCollisionPairs(btOverlappingPairCache* pairCache, const btDispatcherInfo& dispatchInfo, btDispatcher* dispatcher)
 {
 	//m_blockedForChanges = true;
 
-	btCollisionPairCallback	collisionCallback(dispatchInfo,this);
+	btCollisionPairCallback collisionCallback(dispatchInfo, this);
 
-    {
+	{
 		BT_PROFILE("processAllOverlappingPairs");
-		pairCache->processAllOverlappingPairs(&collisionCallback,dispatcher, dispatchInfo);
+		pairCache->processAllOverlappingPairs(&collisionCallback, dispatcher, dispatchInfo);
 	}
 
 	//m_blockedForChanges = false;
-
 }
 
-
-
 //by default, Bullet will use this near callback
 void btCollisionDispatcher::defaultNearCallback(btBroadphasePair& collisionPair, btCollisionDispatcher& dispatcher, const btDispatcherInfo& dispatchInfo)
 {
-		btCollisionObject* colObj0 = (btCollisionObject*)collisionPair.m_pProxy0->m_clientObject;
-		btCollisionObject* colObj1 = (btCollisionObject*)collisionPair.m_pProxy1->m_clientObject;
+	btCollisionObject* colObj0 = (btCollisionObject*)collisionPair.m_pProxy0->m_clientObject;
+	btCollisionObject* colObj1 = (btCollisionObject*)collisionPair.m_pProxy1->m_clientObject;
 
-		if (dispatcher.needsCollision(colObj0,colObj1))
+	if (dispatcher.needsCollision(colObj0, colObj1))
+	{
+		btCollisionObjectWrapper obj0Wrap(0, colObj0->getCollisionShape(), colObj0, colObj0->getWorldTransform(), -1, -1);
+		btCollisionObjectWrapper obj1Wrap(0, colObj1->getCollisionShape(), colObj1, colObj1->getWorldTransform(), -1, -1);
+
+		//dispatcher will keep algorithms persistent in the collision pair
+		if (!collisionPair.m_algorithm)
 		{
-			btCollisionObjectWrapper obj0Wrap(0,colObj0->getCollisionShape(),colObj0,colObj0->getWorldTransform(),-1,-1);
-			btCollisionObjectWrapper obj1Wrap(0,colObj1->getCollisionShape(),colObj1,colObj1->getWorldTransform(),-1,-1);
-
-
-			//dispatcher will keep algorithms persistent in the collision pair
-			if (!collisionPair.m_algorithm)
-			{
-				collisionPair.m_algorithm = dispatcher.findAlgorithm(&obj0Wrap,&obj1Wrap,0, BT_CONTACT_POINT_ALGORITHMS);
-			}
-
-			if (collisionPair.m_algorithm)
-			{
-				btManifoldResult contactPointResult(&obj0Wrap,&obj1Wrap);
-				
-				if (dispatchInfo.m_dispatchFunc == 		btDispatcherInfo::DISPATCH_DISCRETE)
-				{
-					//discrete collision detection query
-					
-					collisionPair.m_algorithm->processCollision(&obj0Wrap,&obj1Wrap,dispatchInfo,&contactPointResult);
-				} else
-				{
-					//continuous collision detection query, time of impact (toi)
-					btScalar toi = collisionPair.m_algorithm->calculateTimeOfImpact(colObj0,colObj1,dispatchInfo,&contactPointResult);
-					if (dispatchInfo.m_timeOfImpact > toi)
-						dispatchInfo.m_timeOfImpact = toi;
-
-				}
-			}
+			collisionPair.m_algorithm = dispatcher.findAlgorithm(&obj0Wrap, &obj1Wrap, 0, BT_CONTACT_POINT_ALGORITHMS);
 		}
 
-}
+		if (collisionPair.m_algorithm)
+		{
+			btManifoldResult contactPointResult(&obj0Wrap, &obj1Wrap);
 
+			if (dispatchInfo.m_dispatchFunc == btDispatcherInfo::DISPATCH_DISCRETE)
+			{
+				//discrete collision detection query
+
+				collisionPair.m_algorithm->processCollision(&obj0Wrap, &obj1Wrap, dispatchInfo, &contactPointResult);
+			}
+			else
+			{
+				//continuous collision detection query, time of impact (toi)
+				btScalar toi = collisionPair.m_algorithm->calculateTimeOfImpact(colObj0, colObj1, dispatchInfo, &contactPointResult);
+				if (dispatchInfo.m_timeOfImpact > toi)
+					dispatchInfo.m_timeOfImpact = toi;
+			}
+		}
+	}
+}
 
 void* btCollisionDispatcher::allocateCollisionAlgorithm(int size)
 {
-    void* mem = m_collisionAlgorithmPoolAllocator->allocate( size );
-    if (NULL == mem)
-    {
-	    //warn user for overflow?
-	    return btAlignedAlloc(static_cast<size_t>(size), 16);
-    }
-    return mem;
+	void* mem = m_collisionAlgorithmPoolAllocator->allocate(size);
+	if (NULL == mem)
+	{
+		//warn user for overflow?
+		return btAlignedAlloc(static_cast<size_t>(size), 16);
+	}
+	return mem;
 }
 
 void btCollisionDispatcher::freeCollisionAlgorithm(void* ptr)
@@ -316,7 +281,8 @@ void btCollisionDispatcher::freeCollisionAlgorithm(void* ptr)
 	if (m_collisionAlgorithmPoolAllocator->validPtr(ptr))
 	{
 		m_collisionAlgorithmPoolAllocator->freeMemory(ptr);
-	} else
+	}
+	else
 	{
 		btAlignedFree(ptr);
 	}
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionDispatcher.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionDispatcher.h
index b97ee3c1ba6..6b9f5e23a58 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionDispatcher.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionDispatcher.h
@@ -37,35 +37,30 @@ class btCollisionDispatcher;
 ///user can override this nearcallback for collision filtering and more finegrained control over collision detection
 typedef void (*btNearCallback)(btBroadphasePair& collisionPair, btCollisionDispatcher& dispatcher, const btDispatcherInfo& dispatchInfo);
 
-
 ///btCollisionDispatcher supports algorithms that handle ConvexConvex and ConvexConcave collision pairs.
 ///Time of Impact, Closest Points and Penetration Depth.
 class btCollisionDispatcher : public btDispatcher
 {
-
 protected:
+	int m_dispatcherFlags;
 
-	int		m_dispatcherFlags;
+	btAlignedObjectArray<btPersistentManifold*> m_manifoldsPtr;
 
-	btAlignedObjectArray<btPersistentManifold*>	m_manifoldsPtr;
+	btManifoldResult m_defaultManifoldResult;
 
-	btManifoldResult	m_defaultManifoldResult;
+	btNearCallback m_nearCallback;
 
-	btNearCallback		m_nearCallback;
-	
-	btPoolAllocator*	m_collisionAlgorithmPoolAllocator;
+	btPoolAllocator* m_collisionAlgorithmPoolAllocator;
 
-	btPoolAllocator*	m_persistentManifoldPoolAllocator;
+	btPoolAllocator* m_persistentManifoldPoolAllocator;
 
 	btCollisionAlgorithmCreateFunc* m_doubleDispatchContactPoints[MAX_BROADPHASE_COLLISION_TYPES][MAX_BROADPHASE_COLLISION_TYPES];
 
 	btCollisionAlgorithmCreateFunc* m_doubleDispatchClosestPoints[MAX_BROADPHASE_COLLISION_TYPES][MAX_BROADPHASE_COLLISION_TYPES];
 
-	btCollisionConfiguration*	m_collisionConfiguration;
-
+	btCollisionConfiguration* m_collisionConfiguration;
 
 public:
-
 	enum DispatcherFlags
 	{
 		CD_STATIC_STATIC_REPORTED = 1,
@@ -73,103 +68,100 @@ public:
 		CD_DISABLE_CONTACTPOOL_DYNAMIC_ALLOCATION = 4
 	};
 
-	int	getDispatcherFlags() const
+	int getDispatcherFlags() const
 	{
 		return m_dispatcherFlags;
 	}
 
-	void	setDispatcherFlags(int flags)
+	void setDispatcherFlags(int flags)
 	{
 		m_dispatcherFlags = flags;
 	}
 
 	///registerCollisionCreateFunc allows registration of custom/alternative collision create functions
-	void	registerCollisionCreateFunc(int proxyType0,int proxyType1, btCollisionAlgorithmCreateFunc* createFunc);
+	void registerCollisionCreateFunc(int proxyType0, int proxyType1, btCollisionAlgorithmCreateFunc* createFunc);
 
-	void	registerClosestPointsCreateFunc(int proxyType0, int proxyType1, btCollisionAlgorithmCreateFunc *createFunc);
+	void registerClosestPointsCreateFunc(int proxyType0, int proxyType1, btCollisionAlgorithmCreateFunc* createFunc);
 
-	int	getNumManifolds() const
-	{ 
-		return int( m_manifoldsPtr.size());
-	}
-
-	btPersistentManifold**	getInternalManifoldPointer()
+	int getNumManifolds() const
 	{
-		return m_manifoldsPtr.size()? &m_manifoldsPtr[0] : 0;
+		return int(m_manifoldsPtr.size());
 	}
 
-	 btPersistentManifold* getManifoldByIndexInternal(int index)
+	btPersistentManifold** getInternalManifoldPointer()
+	{
+		return m_manifoldsPtr.size() ? &m_manifoldsPtr[0] : 0;
+	}
+
+	btPersistentManifold* getManifoldByIndexInternal(int index)
 	{
 		return m_manifoldsPtr[index];
 	}
 
-	 const btPersistentManifold* getManifoldByIndexInternal(int index) const
+	const btPersistentManifold* getManifoldByIndexInternal(int index) const
 	{
 		return m_manifoldsPtr[index];
 	}
 
-	btCollisionDispatcher (btCollisionConfiguration* collisionConfiguration);
+	btCollisionDispatcher(btCollisionConfiguration* collisionConfiguration);
 
 	virtual ~btCollisionDispatcher();
 
-	virtual btPersistentManifold*	getNewManifold(const btCollisionObject* b0,const btCollisionObject* b1);
-	
-	virtual void releaseManifold(btPersistentManifold* manifold);
+	virtual btPersistentManifold* getNewManifold(const btCollisionObject* b0, const btCollisionObject* b1);
 
+	virtual void releaseManifold(btPersistentManifold* manifold);
 
 	virtual void clearManifold(btPersistentManifold* manifold);
 
-	btCollisionAlgorithm* findAlgorithm(const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,btPersistentManifold* sharedManifold, ebtDispatcherQueryType queryType);
-		
-	virtual bool	needsCollision(const btCollisionObject* body0,const btCollisionObject* body1);
-	
-	virtual bool	needsResponse(const btCollisionObject* body0,const btCollisionObject* body1);
-	
-	virtual void	dispatchAllCollisionPairs(btOverlappingPairCache* pairCache,const btDispatcherInfo& dispatchInfo,btDispatcher* dispatcher) ;
+	btCollisionAlgorithm* findAlgorithm(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, btPersistentManifold* sharedManifold, ebtDispatcherQueryType queryType);
 
-	void	setNearCallback(btNearCallback	nearCallback)
+	virtual bool needsCollision(const btCollisionObject* body0, const btCollisionObject* body1);
+
+	virtual bool needsResponse(const btCollisionObject* body0, const btCollisionObject* body1);
+
+	virtual void dispatchAllCollisionPairs(btOverlappingPairCache* pairCache, const btDispatcherInfo& dispatchInfo, btDispatcher* dispatcher);
+
+	void setNearCallback(btNearCallback nearCallback)
 	{
-		m_nearCallback = nearCallback; 
+		m_nearCallback = nearCallback;
 	}
 
-	btNearCallback	getNearCallback() const
+	btNearCallback getNearCallback() const
 	{
 		return m_nearCallback;
 	}
 
 	//by default, Bullet will use this near callback
-	static void  defaultNearCallback(btBroadphasePair& collisionPair, btCollisionDispatcher& dispatcher, const btDispatcherInfo& dispatchInfo);
+	static void defaultNearCallback(btBroadphasePair& collisionPair, btCollisionDispatcher& dispatcher, const btDispatcherInfo& dispatchInfo);
 
-	virtual	void* allocateCollisionAlgorithm(int size);
+	virtual void* allocateCollisionAlgorithm(int size);
 
-	virtual	void freeCollisionAlgorithm(void* ptr);
+	virtual void freeCollisionAlgorithm(void* ptr);
 
-	btCollisionConfiguration*	getCollisionConfiguration()
+	btCollisionConfiguration* getCollisionConfiguration()
 	{
 		return m_collisionConfiguration;
 	}
 
-	const btCollisionConfiguration*	getCollisionConfiguration() const
+	const btCollisionConfiguration* getCollisionConfiguration() const
 	{
 		return m_collisionConfiguration;
 	}
 
-	void	setCollisionConfiguration(btCollisionConfiguration* config)
+	void setCollisionConfiguration(btCollisionConfiguration* config)
 	{
 		m_collisionConfiguration = config;
 	}
 
-	virtual	btPoolAllocator*	getInternalManifoldPool()
+	virtual btPoolAllocator* getInternalManifoldPool()
 	{
 		return m_persistentManifoldPoolAllocator;
 	}
 
-	virtual	const btPoolAllocator*	getInternalManifoldPool() const
+	virtual const btPoolAllocator* getInternalManifoldPool() const
 	{
 		return m_persistentManifoldPoolAllocator;
 	}
-
 };
 
-#endif //BT_COLLISION__DISPATCHER_H
-
+#endif  //BT_COLLISION__DISPATCHER_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionDispatcherMt.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionDispatcherMt.cpp
index 075860c503d..6fe56538d2c 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionDispatcherMt.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionDispatcherMt.cpp
@@ -13,8 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #include "btCollisionDispatcherMt.h"
 #include "LinearMath/btQuickprof.h"
 
@@ -27,138 +25,132 @@ subject to the following restrictions:
 #include "BulletCollision/CollisionDispatch/btCollisionConfiguration.h"
 #include "BulletCollision/CollisionDispatch/btCollisionObjectWrapper.h"
 
-
-btCollisionDispatcherMt::btCollisionDispatcherMt( btCollisionConfiguration* config, int grainSize )
-    : btCollisionDispatcher( config )
+btCollisionDispatcherMt::btCollisionDispatcherMt(btCollisionConfiguration* config, int grainSize)
+	: btCollisionDispatcher(config)
 {
-    m_batchUpdating = false;
-    m_grainSize = grainSize;  // iterations per task
+	m_batchUpdating = false;
+	m_grainSize = grainSize;  // iterations per task
 }
 
-
-btPersistentManifold* btCollisionDispatcherMt::getNewManifold( const btCollisionObject* body0, const btCollisionObject* body1 )
+btPersistentManifold* btCollisionDispatcherMt::getNewManifold(const btCollisionObject* body0, const btCollisionObject* body1)
 {
-    //optional relative contact breaking threshold, turned on by default (use setDispatcherFlags to switch off feature for improved performance)
+	//optional relative contact breaking threshold, turned on by default (use setDispatcherFlags to switch off feature for improved performance)
 
-    btScalar contactBreakingThreshold = ( m_dispatcherFlags & btCollisionDispatcher::CD_USE_RELATIVE_CONTACT_BREAKING_THRESHOLD ) ?
-        btMin( body0->getCollisionShape()->getContactBreakingThreshold( gContactBreakingThreshold ), body1->getCollisionShape()->getContactBreakingThreshold( gContactBreakingThreshold ) )
-        : gContactBreakingThreshold;
+	btScalar contactBreakingThreshold = (m_dispatcherFlags & btCollisionDispatcher::CD_USE_RELATIVE_CONTACT_BREAKING_THRESHOLD) ? btMin(body0->getCollisionShape()->getContactBreakingThreshold(gContactBreakingThreshold), body1->getCollisionShape()->getContactBreakingThreshold(gContactBreakingThreshold))
+																																: gContactBreakingThreshold;
 
-    btScalar contactProcessingThreshold = btMin( body0->getContactProcessingThreshold(), body1->getContactProcessingThreshold() );
+	btScalar contactProcessingThreshold = btMin(body0->getContactProcessingThreshold(), body1->getContactProcessingThreshold());
 
-    void* mem = m_persistentManifoldPoolAllocator->allocate( sizeof( btPersistentManifold ) );
-    if ( NULL == mem )
-    {
-        //we got a pool memory overflow, by default we fallback to dynamically allocate memory. If we require a contiguous contact pool then assert.
-        if ( ( m_dispatcherFlags&CD_DISABLE_CONTACTPOOL_DYNAMIC_ALLOCATION ) == 0 )
-        {
-            mem = btAlignedAlloc( sizeof( btPersistentManifold ), 16 );
-        }
-        else
-        {
-            btAssert( 0 );
-            //make sure to increase the m_defaultMaxPersistentManifoldPoolSize in the btDefaultCollisionConstructionInfo/btDefaultCollisionConfiguration
-            return 0;
-        }
-    }
-    btPersistentManifold* manifold = new( mem ) btPersistentManifold( body0, body1, 0, contactBreakingThreshold, contactProcessingThreshold );
-    if ( !m_batchUpdating )
-    {
-        // batch updater will update manifold pointers array after finishing, so
-        // only need to update array when not batch-updating
-        //btAssert( !btThreadsAreRunning() );
-        manifold->m_index1a = m_manifoldsPtr.size();
-        m_manifoldsPtr.push_back( manifold );
-    }
+	void* mem = m_persistentManifoldPoolAllocator->allocate(sizeof(btPersistentManifold));
+	if (NULL == mem)
+	{
+		//we got a pool memory overflow, by default we fallback to dynamically allocate memory. If we require a contiguous contact pool then assert.
+		if ((m_dispatcherFlags & CD_DISABLE_CONTACTPOOL_DYNAMIC_ALLOCATION) == 0)
+		{
+			mem = btAlignedAlloc(sizeof(btPersistentManifold), 16);
+		}
+		else
+		{
+			btAssert(0);
+			//make sure to increase the m_defaultMaxPersistentManifoldPoolSize in the btDefaultCollisionConstructionInfo/btDefaultCollisionConfiguration
+			return 0;
+		}
+	}
+	btPersistentManifold* manifold = new (mem) btPersistentManifold(body0, body1, 0, contactBreakingThreshold, contactProcessingThreshold);
+	if (!m_batchUpdating)
+	{
+		// batch updater will update manifold pointers array after finishing, so
+		// only need to update array when not batch-updating
+		//btAssert( !btThreadsAreRunning() );
+		manifold->m_index1a = m_manifoldsPtr.size();
+		m_manifoldsPtr.push_back(manifold);
+	}
 
-    return manifold;
+	return manifold;
 }
 
-void btCollisionDispatcherMt::releaseManifold( btPersistentManifold* manifold )
+void btCollisionDispatcherMt::releaseManifold(btPersistentManifold* manifold)
 {
-    clearManifold( manifold );
-    //btAssert( !btThreadsAreRunning() );
-    if ( !m_batchUpdating )
-    {
-        // batch updater will update manifold pointers array after finishing, so
-        // only need to update array when not batch-updating
-        int findIndex = manifold->m_index1a;
-        btAssert( findIndex < m_manifoldsPtr.size() );
-        m_manifoldsPtr.swap( findIndex, m_manifoldsPtr.size() - 1 );
-        m_manifoldsPtr[ findIndex ]->m_index1a = findIndex;
-        m_manifoldsPtr.pop_back();
-    }
+	clearManifold(manifold);
+	//btAssert( !btThreadsAreRunning() );
+	if (!m_batchUpdating)
+	{
+		// batch updater will update manifold pointers array after finishing, so
+		// only need to update array when not batch-updating
+		int findIndex = manifold->m_index1a;
+		btAssert(findIndex < m_manifoldsPtr.size());
+		m_manifoldsPtr.swap(findIndex, m_manifoldsPtr.size() - 1);
+		m_manifoldsPtr[findIndex]->m_index1a = findIndex;
+		m_manifoldsPtr.pop_back();
+	}
 
-    manifold->~btPersistentManifold();
-    if ( m_persistentManifoldPoolAllocator->validPtr( manifold ) )
-    {
-        m_persistentManifoldPoolAllocator->freeMemory( manifold );
-    }
-    else
-    {
-        btAlignedFree( manifold );
-    }
+	manifold->~btPersistentManifold();
+	if (m_persistentManifoldPoolAllocator->validPtr(manifold))
+	{
+		m_persistentManifoldPoolAllocator->freeMemory(manifold);
+	}
+	else
+	{
+		btAlignedFree(manifold);
+	}
 }
 
 struct CollisionDispatcherUpdater : public btIParallelForBody
 {
-    btBroadphasePair* mPairArray;
-    btNearCallback mCallback;
-    btCollisionDispatcher* mDispatcher;
-    const btDispatcherInfo* mInfo;
+	btBroadphasePair* mPairArray;
+	btNearCallback mCallback;
+	btCollisionDispatcher* mDispatcher;
+	const btDispatcherInfo* mInfo;
 
-    CollisionDispatcherUpdater()
-    {
-        mPairArray = NULL;
-        mCallback = NULL;
-        mDispatcher = NULL;
-        mInfo = NULL;
-    }
-    void forLoop( int iBegin, int iEnd ) const
-    {
-        for ( int i = iBegin; i < iEnd; ++i )
-        {
-            btBroadphasePair* pair = &mPairArray[ i ];
-            mCallback( *pair, *mDispatcher, *mInfo );
-        }
-    }
+	CollisionDispatcherUpdater()
+	{
+		mPairArray = NULL;
+		mCallback = NULL;
+		mDispatcher = NULL;
+		mInfo = NULL;
+	}
+	void forLoop(int iBegin, int iEnd) const
+	{
+		for (int i = iBegin; i < iEnd; ++i)
+		{
+			btBroadphasePair* pair = &mPairArray[i];
+			mCallback(*pair, *mDispatcher, *mInfo);
+		}
+	}
 };
 
-
-void btCollisionDispatcherMt::dispatchAllCollisionPairs( btOverlappingPairCache* pairCache, const btDispatcherInfo& info, btDispatcher* dispatcher )
+void btCollisionDispatcherMt::dispatchAllCollisionPairs(btOverlappingPairCache* pairCache, const btDispatcherInfo& info, btDispatcher* dispatcher)
 {
-    int pairCount = pairCache->getNumOverlappingPairs();
-    if ( pairCount == 0 )
-    {
-        return;
-    }
-    CollisionDispatcherUpdater updater;
-    updater.mCallback = getNearCallback();
-    updater.mPairArray = pairCache->getOverlappingPairArrayPtr();
-    updater.mDispatcher = this;
-    updater.mInfo = &info;
+	int pairCount = pairCache->getNumOverlappingPairs();
+	if (pairCount == 0)
+	{
+		return;
+	}
+	CollisionDispatcherUpdater updater;
+	updater.mCallback = getNearCallback();
+	updater.mPairArray = pairCache->getOverlappingPairArrayPtr();
+	updater.mDispatcher = this;
+	updater.mInfo = &info;
 
-    m_batchUpdating = true;
-    btParallelFor( 0, pairCount, m_grainSize, updater );
-    m_batchUpdating = false;
+	m_batchUpdating = true;
+	btParallelFor(0, pairCount, m_grainSize, updater);
+	m_batchUpdating = false;
 
-    // reconstruct the manifolds array to ensure determinism
-    m_manifoldsPtr.resizeNoInitialize( 0 );
+	// reconstruct the manifolds array to ensure determinism
+	m_manifoldsPtr.resizeNoInitialize(0);
 
-    btBroadphasePair* pairs = pairCache->getOverlappingPairArrayPtr();
-    for ( int i = 0; i < pairCount; ++i )
-    {
-        if (btCollisionAlgorithm* algo = pairs[ i ].m_algorithm)
-        {
-            algo->getAllContactManifolds( m_manifoldsPtr );
-        }
-    }
+	btBroadphasePair* pairs = pairCache->getOverlappingPairArrayPtr();
+	for (int i = 0; i < pairCount; ++i)
+	{
+		if (btCollisionAlgorithm* algo = pairs[i].m_algorithm)
+		{
+			algo->getAllContactManifolds(m_manifoldsPtr);
+		}
+	}
 
-    // update the indices (used when releasing manifolds)
-    for ( int i = 0; i < m_manifoldsPtr.size(); ++i )
-    {
-        m_manifoldsPtr[ i ]->m_index1a = i;
-    }
+	// update the indices (used when releasing manifolds)
+	for (int i = 0; i < m_manifoldsPtr.size(); ++i)
+	{
+		m_manifoldsPtr[i]->m_index1a = i;
+	}
 }
-
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionDispatcherMt.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionDispatcherMt.h
index f1d7eafdc96..28eba7f32a1 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionDispatcherMt.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionDispatcherMt.h
@@ -19,21 +19,19 @@ subject to the following restrictions:
 #include "BulletCollision/CollisionDispatch/btCollisionDispatcher.h"
 #include "LinearMath/btThreads.h"
 
-
 class btCollisionDispatcherMt : public btCollisionDispatcher
 {
 public:
-    btCollisionDispatcherMt( btCollisionConfiguration* config, int grainSize = 40 );
+	btCollisionDispatcherMt(btCollisionConfiguration* config, int grainSize = 40);
 
-    virtual btPersistentManifold* getNewManifold( const btCollisionObject* body0, const btCollisionObject* body1 ) BT_OVERRIDE;
-    virtual void releaseManifold( btPersistentManifold* manifold ) BT_OVERRIDE;
+	virtual btPersistentManifold* getNewManifold(const btCollisionObject* body0, const btCollisionObject* body1) BT_OVERRIDE;
+	virtual void releaseManifold(btPersistentManifold* manifold) BT_OVERRIDE;
 
-    virtual void dispatchAllCollisionPairs( btOverlappingPairCache* pairCache, const btDispatcherInfo& info, btDispatcher* dispatcher ) BT_OVERRIDE;
+	virtual void dispatchAllCollisionPairs(btOverlappingPairCache* pairCache, const btDispatcherInfo& info, btDispatcher* dispatcher) BT_OVERRIDE;
 
 protected:
-    bool m_batchUpdating;
-    int m_grainSize;
+	bool m_batchUpdating;
+	int m_grainSize;
 };
 
-#endif //BT_COLLISION_DISPATCHER_MT_H
-
+#endif  //BT_COLLISION_DISPATCHER_MT_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionObject.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionObject.cpp
index 05f96a14bc8..98a02d0c452 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionObject.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionObject.cpp
@@ -13,42 +13,41 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btCollisionObject.h"
 #include "LinearMath/btSerializer.h"
 #include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h"
 
 btCollisionObject::btCollisionObject()
-	:	m_interpolationLinearVelocity(0.f, 0.f, 0.f),
-		m_interpolationAngularVelocity(0.f, 0.f, 0.f),
-		m_anisotropicFriction(1.f,1.f,1.f),
-		m_hasAnisotropicFriction(false),
-		m_contactProcessingThreshold(BT_LARGE_FLOAT),
-		m_broadphaseHandle(0),
-		m_collisionShape(0),
-		m_extensionPointer(0),
-		m_rootCollisionShape(0),
-		m_collisionFlags(btCollisionObject::CF_STATIC_OBJECT),
-		m_islandTag1(-1),
-		m_companionId(-1),
-        m_worldArrayIndex(-1),
-		m_activationState1(1),
-		m_deactivationTime(btScalar(0.)),
-		m_friction(btScalar(0.5)),
-		m_restitution(btScalar(0.)),
-		m_rollingFriction(0.0f),
-        m_spinningFriction(0.f),
-		m_contactDamping(.1),
-		m_contactStiffness(BT_LARGE_FLOAT),
-		m_internalType(CO_COLLISION_OBJECT),
-		m_userObjectPointer(0),
-		m_userIndex2(-1),
-		m_userIndex(-1),
-		m_hitFraction(btScalar(1.)),
-		m_ccdSweptSphereRadius(btScalar(0.)),
-		m_ccdMotionThreshold(btScalar(0.)),
-		m_checkCollideWith(false),
-		m_updateRevision(0)
+	: m_interpolationLinearVelocity(0.f, 0.f, 0.f),
+	  m_interpolationAngularVelocity(0.f, 0.f, 0.f),
+	  m_anisotropicFriction(1.f, 1.f, 1.f),
+	  m_hasAnisotropicFriction(false),
+	  m_contactProcessingThreshold(BT_LARGE_FLOAT),
+	  m_broadphaseHandle(0),
+	  m_collisionShape(0),
+	  m_extensionPointer(0),
+	  m_rootCollisionShape(0),
+	  m_collisionFlags(btCollisionObject::CF_STATIC_OBJECT),
+	  m_islandTag1(-1),
+	  m_companionId(-1),
+	  m_worldArrayIndex(-1),
+	  m_activationState1(1),
+	  m_deactivationTime(btScalar(0.)),
+	  m_friction(btScalar(0.5)),
+	  m_restitution(btScalar(0.)),
+	  m_rollingFriction(0.0f),
+	  m_spinningFriction(0.f),
+	  m_contactDamping(.1),
+	  m_contactStiffness(BT_LARGE_FLOAT),
+	  m_internalType(CO_COLLISION_OBJECT),
+	  m_userObjectPointer(0),
+	  m_userIndex2(-1),
+	  m_userIndex(-1),
+	  m_hitFraction(btScalar(1.)),
+	  m_ccdSweptSphereRadius(btScalar(0.)),
+	  m_ccdMotionThreshold(btScalar(0.)),
+	  m_checkCollideWith(false),
+	  m_updateRevision(0)
 {
 	m_worldTransform.setIdentity();
 	m_interpolationWorldTransform.setIdentity();
@@ -59,8 +58,8 @@ btCollisionObject::~btCollisionObject()
 }
 
 void btCollisionObject::setActivationState(int newState) const
-{ 
-	if ( (m_activationState1 != DISABLE_DEACTIVATION) && (m_activationState1 != DISABLE_SIMULATION))
+{
+	if ((m_activationState1 != DISABLE_DEACTIVATION) && (m_activationState1 != DISABLE_SIMULATION))
 		m_activationState1 = newState;
 }
 
@@ -71,7 +70,7 @@ void btCollisionObject::forceActivationState(int newState) const
 
 void btCollisionObject::activate(bool forceActivation) const
 {
-	if (forceActivation || !(m_collisionFlags & (CF_STATIC_OBJECT|CF_KINEMATIC_OBJECT)))
+	if (forceActivation || !(m_collisionFlags & (CF_STATIC_OBJECT | CF_KINEMATIC_OBJECT)))
 	{
 		setActivationState(ACTIVE_TAG);
 		m_deactivationTime = btScalar(0.);
@@ -80,7 +79,6 @@ void btCollisionObject::activate(bool forceActivation) const
 
 const char* btCollisionObject::serialize(void* dataBuffer, btSerializer* serializer) const
 {
-
 	btCollisionObjectData* dataOut = (btCollisionObjectData*)dataBuffer;
 
 	m_worldTransform.serialize(dataOut->m_worldTransform);
@@ -92,7 +90,7 @@ const char* btCollisionObject::serialize(void* dataBuffer, btSerializer* seriali
 	dataOut->m_contactProcessingThreshold = m_contactProcessingThreshold;
 	dataOut->m_broadphaseHandle = 0;
 	dataOut->m_collisionShape = serializer->getUniquePointer(m_collisionShape);
-	dataOut->m_rootCollisionShape = 0;//@todo
+	dataOut->m_rootCollisionShape = 0;  //@todo
 	dataOut->m_collisionFlags = m_collisionFlags;
 	dataOut->m_islandTag1 = m_islandTag1;
 	dataOut->m_companionId = m_companionId;
@@ -104,8 +102,8 @@ const char* btCollisionObject::serialize(void* dataBuffer, btSerializer* seriali
 	dataOut->m_contactStiffness = m_contactStiffness;
 	dataOut->m_restitution = m_restitution;
 	dataOut->m_internalType = m_internalType;
-	
-	char* name = (char*) serializer->findNameForPointer(this);
+
+	char* name = (char*)serializer->findNameForPointer(this);
 	dataOut->m_name = (char*)serializer->getUniquePointer(name);
 	if (dataOut->m_name)
 	{
@@ -130,11 +128,10 @@ const char* btCollisionObject::serialize(void* dataBuffer, btSerializer* seriali
 	return btCollisionObjectDataName;
 }
 
-
 void btCollisionObject::serializeSingleObject(class btSerializer* serializer) const
 {
 	int len = calculateSerializeBufferSize();
-	btChunk* chunk = serializer->allocate(len,1);
+	btChunk* chunk = serializer->allocate(len, 1);
 	const char* structType = serialize(chunk->m_oldPtr, serializer);
-	serializer->finalizeChunk(chunk,structType,BT_COLLISIONOBJECT_CODE,(void*)this);
+	serializer->finalizeChunk(chunk, structType, BT_COLLISIONOBJECT_CODE, (void*)this);
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionObject.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionObject.h
index 135f8a033c3..56b3d89e569 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionObject.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionObject.h
@@ -25,8 +25,8 @@ subject to the following restrictions:
 #define DISABLE_DEACTIVATION 4
 #define DISABLE_SIMULATION 5
 
-struct	btBroadphaseProxy;
-class	btCollisionShape;
+struct btBroadphaseProxy;
+class btCollisionShape;
 struct btCollisionShapeData;
 #include "LinearMath/btMotionState.h"
 #include "LinearMath/btAlignedAllocator.h"
@@ -42,123 +42,118 @@ typedef btAlignedObjectArray<class btCollisionObject*> btCollisionObjectArray;
 #define btCollisionObjectDataName "btCollisionObjectFloatData"
 #endif
 
-
-/// btCollisionObject can be used to manage collision detection objects. 
+/// btCollisionObject can be used to manage collision detection objects.
 /// btCollisionObject maintains all information that is needed for a collision detection: Shape, Transform and AABB proxy.
 /// They can be added to the btCollisionWorld.
-ATTRIBUTE_ALIGNED16(class)	btCollisionObject
+ATTRIBUTE_ALIGNED16(class)
+btCollisionObject
 {
-
 protected:
-
-	btTransform	m_worldTransform;
+	btTransform m_worldTransform;
 
 	///m_interpolationWorldTransform is used for CCD and interpolation
 	///it can be either previous or future (predicted) transform
-	btTransform	m_interpolationWorldTransform;
-	//those two are experimental: just added for bullet time effect, so you can still apply impulses (directly modifying velocities) 
+	btTransform m_interpolationWorldTransform;
+	//those two are experimental: just added for bullet time effect, so you can still apply impulses (directly modifying velocities)
 	//without destroying the continuous interpolated motion (which uses this interpolation velocities)
-	btVector3	m_interpolationLinearVelocity;
-	btVector3	m_interpolationAngularVelocity;
-	
-	btVector3	m_anisotropicFriction;
-	int			m_hasAnisotropicFriction;
-	btScalar	m_contactProcessingThreshold;	
+	btVector3 m_interpolationLinearVelocity;
+	btVector3 m_interpolationAngularVelocity;
 
-	btBroadphaseProxy*		m_broadphaseHandle;
-	btCollisionShape*		m_collisionShape;
+	btVector3 m_anisotropicFriction;
+	int m_hasAnisotropicFriction;
+	btScalar m_contactProcessingThreshold;
+
+	btBroadphaseProxy* m_broadphaseHandle;
+	btCollisionShape* m_collisionShape;
 	///m_extensionPointer is used by some internal low-level Bullet extensions.
-	void*					m_extensionPointer;
-	
+	void* m_extensionPointer;
+
 	///m_rootCollisionShape is temporarily used to store the original collision shape
 	///The m_collisionShape might be temporarily replaced by a child collision shape during collision detection purposes
 	///If it is NULL, the m_collisionShape is not temporarily replaced.
-	btCollisionShape*		m_rootCollisionShape;
+	btCollisionShape* m_rootCollisionShape;
 
-	int				m_collisionFlags;
+	int m_collisionFlags;
 
-	int				m_islandTag1;
-	int				m_companionId;
-    int             m_worldArrayIndex;  // index of object in world's collisionObjects array
+	int m_islandTag1;
+	int m_companionId;
+	int m_worldArrayIndex;  // index of object in world's collisionObjects array
 
-	mutable int				m_activationState1;
-	mutable btScalar			m_deactivationTime;
+	mutable int m_activationState1;
+	mutable btScalar m_deactivationTime;
 
-	btScalar		m_friction;
-	btScalar		m_restitution;
-	btScalar		m_rollingFriction;//torsional friction orthogonal to contact normal (useful to stop spheres rolling forever)
-    btScalar        m_spinningFriction; // torsional friction around the contact normal (useful for grasping)
-	btScalar		m_contactDamping;
-	btScalar		m_contactStiffness;
-	
-	
+	btScalar m_friction;
+	btScalar m_restitution;
+	btScalar m_rollingFriction;   //torsional friction orthogonal to contact normal (useful to stop spheres rolling forever)
+	btScalar m_spinningFriction;  // torsional friction around the contact normal (useful for grasping)
+	btScalar m_contactDamping;
+	btScalar m_contactStiffness;
 
 	///m_internalType is reserved to distinguish Bullet's btCollisionObject, btRigidBody, btSoftBody, btGhostObject etc.
 	///do not assign your own m_internalType unless you write a new dynamics object class.
-	int				m_internalType;
+	int m_internalType;
 
 	///users can point to their objects, m_userPointer is not used by Bullet, see setUserPointer/getUserPointer
 
-	void*			m_userObjectPointer;
+	void* m_userObjectPointer;
 
-	int				m_userIndex2;
-	
-    int	m_userIndex;
+	int m_userIndex2;
+
+	int m_userIndex;
 
 	///time of impact calculation
-	btScalar		m_hitFraction; 
-	
+	btScalar m_hitFraction;
+
 	///Swept sphere radius (0.0 by default), see btConvexConvexAlgorithm::
-	btScalar		m_ccdSweptSphereRadius;
+	btScalar m_ccdSweptSphereRadius;
 
 	/// Don't do continuous collision detection if the motion (in one step) is less then m_ccdMotionThreshold
-	btScalar		m_ccdMotionThreshold;
-	
+	btScalar m_ccdMotionThreshold;
+
 	/// If some object should have elaborate collision filtering by sub-classes
-	int			m_checkCollideWith;
+	int m_checkCollideWith;
 
 	btAlignedObjectArray<const btCollisionObject*> m_objectsWithoutCollisionCheck;
 
 	///internal update revision number. It will be increased when the object changes. This allows some subsystems to perform lazy evaluation.
-	int			m_updateRevision;
+	int m_updateRevision;
 
-	btVector3	m_customDebugColorRGB;
+	btVector3 m_customDebugColorRGB;
 
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
 	enum CollisionFlags
 	{
-		CF_STATIC_OBJECT= 1,
-		CF_KINEMATIC_OBJECT= 2,
+		CF_STATIC_OBJECT = 1,
+		CF_KINEMATIC_OBJECT = 2,
 		CF_NO_CONTACT_RESPONSE = 4,
-		CF_CUSTOM_MATERIAL_CALLBACK = 8,//this allows per-triangle material (friction/restitution)
+		CF_CUSTOM_MATERIAL_CALLBACK = 8,  //this allows per-triangle material (friction/restitution)
 		CF_CHARACTER_OBJECT = 16,
-		CF_DISABLE_VISUALIZE_OBJECT = 32, //disable debug drawing
-		CF_DISABLE_SPU_COLLISION_PROCESSING = 64,//disable parallel/SPU processing
+		CF_DISABLE_VISUALIZE_OBJECT = 32,          //disable debug drawing
+		CF_DISABLE_SPU_COLLISION_PROCESSING = 64,  //disable parallel/SPU processing
 		CF_HAS_CONTACT_STIFFNESS_DAMPING = 128,
 		CF_HAS_CUSTOM_DEBUG_RENDERING_COLOR = 256,
 		CF_HAS_FRICTION_ANCHOR = 512,
 		CF_HAS_COLLISION_SOUND_TRIGGER = 1024
 	};
 
-	enum	CollisionObjectTypes
+	enum CollisionObjectTypes
 	{
-		CO_COLLISION_OBJECT =1,
-		CO_RIGID_BODY=2,
+		CO_COLLISION_OBJECT = 1,
+		CO_RIGID_BODY = 2,
 		///CO_GHOST_OBJECT keeps track of all objects overlapping its AABB and that pass its collision filter
 		///It is useful for collision sensors, explosion objects, character controller etc.
-		CO_GHOST_OBJECT=4,
-		CO_SOFT_BODY=8,
-		CO_HF_FLUID=16,
-		CO_USER_TYPE=32,
-		CO_FEATHERSTONE_LINK=64
+		CO_GHOST_OBJECT = 4,
+		CO_SOFT_BODY = 8,
+		CO_HF_FLUID = 16,
+		CO_USER_TYPE = 32,
+		CO_FEATHERSTONE_LINK = 64
 	};
 
 	enum AnisotropicFrictionFlags
 	{
-		CF_ANISOTROPIC_FRICTION_DISABLED=0,
+		CF_ANISOTROPIC_FRICTION_DISABLED = 0,
 		CF_ANISOTROPIC_FRICTION = 1,
 		CF_ANISOTROPIC_ROLLING_FRICTION = 2
 	};
@@ -166,76 +161,77 @@ public:
 	SIMD_FORCE_INLINE bool mergesSimulationIslands() const
 	{
 		///static objects, kinematic and object without contact response don't merge islands
-		return  ((m_collisionFlags & (CF_STATIC_OBJECT | CF_KINEMATIC_OBJECT | CF_NO_CONTACT_RESPONSE) )==0);
+		return ((m_collisionFlags & (CF_STATIC_OBJECT | CF_KINEMATIC_OBJECT | CF_NO_CONTACT_RESPONSE)) == 0);
 	}
 
 	const btVector3& getAnisotropicFriction() const
 	{
 		return m_anisotropicFriction;
 	}
-	void	setAnisotropicFriction(const btVector3& anisotropicFriction, int frictionMode = CF_ANISOTROPIC_FRICTION)
+	void setAnisotropicFriction(const btVector3& anisotropicFriction, int frictionMode = CF_ANISOTROPIC_FRICTION)
 	{
 		m_anisotropicFriction = anisotropicFriction;
-		bool isUnity = (anisotropicFriction[0]!=1.f) || (anisotropicFriction[1]!=1.f) || (anisotropicFriction[2]!=1.f);
-		m_hasAnisotropicFriction = isUnity?frictionMode : 0;
+		bool isUnity = (anisotropicFriction[0] != 1.f) || (anisotropicFriction[1] != 1.f) || (anisotropicFriction[2] != 1.f);
+		m_hasAnisotropicFriction = isUnity ? frictionMode : 0;
 	}
-	bool	hasAnisotropicFriction(int frictionMode = CF_ANISOTROPIC_FRICTION) const
+	bool hasAnisotropicFriction(int frictionMode = CF_ANISOTROPIC_FRICTION) const
 	{
-		return (m_hasAnisotropicFriction&frictionMode)!=0;
+		return (m_hasAnisotropicFriction & frictionMode) != 0;
 	}
 
 	///the constraint solver can discard solving contacts, if the distance is above this threshold. 0 by default.
 	///Note that using contacts with positive distance can improve stability. It increases, however, the chance of colliding with degerate contacts, such as 'interior' triangle edges
-	void	setContactProcessingThreshold( btScalar contactProcessingThreshold)
+	void setContactProcessingThreshold(btScalar contactProcessingThreshold)
 	{
 		m_contactProcessingThreshold = contactProcessingThreshold;
 	}
-	btScalar	getContactProcessingThreshold() const
+	btScalar getContactProcessingThreshold() const
 	{
 		return m_contactProcessingThreshold;
 	}
 
-	SIMD_FORCE_INLINE bool		isStaticObject() const {
+	SIMD_FORCE_INLINE bool isStaticObject() const
+	{
 		return (m_collisionFlags & CF_STATIC_OBJECT) != 0;
 	}
 
-	SIMD_FORCE_INLINE bool		isKinematicObject() const
+	SIMD_FORCE_INLINE bool isKinematicObject() const
 	{
 		return (m_collisionFlags & CF_KINEMATIC_OBJECT) != 0;
 	}
 
-	SIMD_FORCE_INLINE bool		isStaticOrKinematicObject() const
+	SIMD_FORCE_INLINE bool isStaticOrKinematicObject() const
 	{
-		return (m_collisionFlags & (CF_KINEMATIC_OBJECT | CF_STATIC_OBJECT)) != 0 ;
+		return (m_collisionFlags & (CF_KINEMATIC_OBJECT | CF_STATIC_OBJECT)) != 0;
 	}
 
-	SIMD_FORCE_INLINE bool		hasContactResponse() const {
-		return (m_collisionFlags & CF_NO_CONTACT_RESPONSE)==0;
+	SIMD_FORCE_INLINE bool hasContactResponse() const
+	{
+		return (m_collisionFlags & CF_NO_CONTACT_RESPONSE) == 0;
 	}
 
-	
 	btCollisionObject();
 
 	virtual ~btCollisionObject();
 
-	virtual void	setCollisionShape(btCollisionShape* collisionShape)
+	virtual void setCollisionShape(btCollisionShape * collisionShape)
 	{
 		m_updateRevision++;
 		m_collisionShape = collisionShape;
 		m_rootCollisionShape = collisionShape;
 	}
 
-	SIMD_FORCE_INLINE const btCollisionShape*	getCollisionShape() const
+	SIMD_FORCE_INLINE const btCollisionShape* getCollisionShape() const
 	{
 		return m_collisionShape;
 	}
 
-	SIMD_FORCE_INLINE btCollisionShape*	getCollisionShape()
+	SIMD_FORCE_INLINE btCollisionShape* getCollisionShape()
 	{
 		return m_collisionShape;
 	}
 
-	void	setIgnoreCollisionCheck(const btCollisionObject* co, bool ignoreCollisionCheck)
+	void setIgnoreCollisionCheck(const btCollisionObject* co, bool ignoreCollisionCheck)
 	{
 		if (ignoreCollisionCheck)
 		{
@@ -253,7 +249,7 @@ public:
 		m_checkCollideWith = m_objectsWithoutCollisionCheck.size() > 0;
 	}
 
-	virtual bool	checkCollideWithOverride(const btCollisionObject*  co) const
+	virtual bool checkCollideWithOverride(const btCollisionObject* co) const
 	{
 		int index = m_objectsWithoutCollisionCheck.findLinearSearch(co);
 		if (index < m_objectsWithoutCollisionCheck.size())
@@ -263,317 +259,309 @@ public:
 		return true;
 	}
 
-
-	
-
-	///Avoid using this internal API call, the extension pointer is used by some Bullet extensions. 
+	///Avoid using this internal API call, the extension pointer is used by some Bullet extensions.
 	///If you need to store your own user pointer, use 'setUserPointer/getUserPointer' instead.
-	void*		internalGetExtensionPointer() const
+	void* internalGetExtensionPointer() const
 	{
 		return m_extensionPointer;
 	}
 	///Avoid using this internal API call, the extension pointer is used by some Bullet extensions
 	///If you need to store your own user pointer, use 'setUserPointer/getUserPointer' instead.
-	void	internalSetExtensionPointer(void* pointer)
+	void internalSetExtensionPointer(void* pointer)
 	{
 		m_extensionPointer = pointer;
 	}
 
-	SIMD_FORCE_INLINE	int	getActivationState() const { return m_activationState1;}
-	
+	SIMD_FORCE_INLINE int getActivationState() const { return m_activationState1; }
+
 	void setActivationState(int newState) const;
 
-	void	setDeactivationTime(btScalar time)
+	void setDeactivationTime(btScalar time)
 	{
 		m_deactivationTime = time;
 	}
-	btScalar	getDeactivationTime() const
+	btScalar getDeactivationTime() const
 	{
 		return m_deactivationTime;
 	}
 
 	void forceActivationState(int newState) const;
 
-	void	activate(bool forceActivation = false) const;
+	void activate(bool forceActivation = false) const;
 
 	SIMD_FORCE_INLINE bool isActive() const
 	{
 		return ((getActivationState() != ISLAND_SLEEPING) && (getActivationState() != DISABLE_SIMULATION));
 	}
 
-	void	setRestitution(btScalar rest)
+	void setRestitution(btScalar rest)
 	{
 		m_updateRevision++;
 		m_restitution = rest;
 	}
-	btScalar	getRestitution() const
+	btScalar getRestitution() const
 	{
 		return m_restitution;
 	}
-	void	setFriction(btScalar frict)
+	void setFriction(btScalar frict)
 	{
 		m_updateRevision++;
 		m_friction = frict;
 	}
-	btScalar	getFriction() const
+	btScalar getFriction() const
 	{
 		return m_friction;
 	}
 
-	void	setRollingFriction(btScalar frict)
+	void setRollingFriction(btScalar frict)
 	{
 		m_updateRevision++;
 		m_rollingFriction = frict;
 	}
-	btScalar	getRollingFriction() const
+	btScalar getRollingFriction() const
 	{
 		return m_rollingFriction;
 	}
-    void	setSpinningFriction(btScalar frict)
-    {
-        m_updateRevision++;
-        m_spinningFriction = frict;
-    }
-    btScalar	getSpinningFriction() const
-    {
-        return m_spinningFriction;
-    }
-    void	setContactStiffnessAndDamping(btScalar stiffness, btScalar damping)
+	void setSpinningFriction(btScalar frict)
+	{
+		m_updateRevision++;
+		m_spinningFriction = frict;
+	}
+	btScalar getSpinningFriction() const
+	{
+		return m_spinningFriction;
+	}
+	void setContactStiffnessAndDamping(btScalar stiffness, btScalar damping)
 	{
 		m_updateRevision++;
 		m_contactStiffness = stiffness;
 		m_contactDamping = damping;
-		
-		m_collisionFlags |=CF_HAS_CONTACT_STIFFNESS_DAMPING;
-		
-        //avoid divisions by zero...
-		if (m_contactStiffness< SIMD_EPSILON)
-        {
-            m_contactStiffness = SIMD_EPSILON;
-        }
+
+		m_collisionFlags |= CF_HAS_CONTACT_STIFFNESS_DAMPING;
+
+		//avoid divisions by zero...
+		if (m_contactStiffness < SIMD_EPSILON)
+		{
+			m_contactStiffness = SIMD_EPSILON;
+		}
 	}
-	
-	btScalar	getContactStiffness() const
+
+	btScalar getContactStiffness() const
 	{
 		return m_contactStiffness;
 	}
-	
-	btScalar	getContactDamping() const
+
+	btScalar getContactDamping() const
 	{
 		return m_contactDamping;
 	}
-    
+
 	///reserved for Bullet internal usage
-	int	getInternalType() const
+	int getInternalType() const
 	{
 		return m_internalType;
 	}
 
-	btTransform&	getWorldTransform()
+	btTransform& getWorldTransform()
 	{
 		return m_worldTransform;
 	}
 
-	const btTransform&	getWorldTransform() const
+	const btTransform& getWorldTransform() const
 	{
 		return m_worldTransform;
 	}
 
-	void	setWorldTransform(const btTransform& worldTrans)
+	void setWorldTransform(const btTransform& worldTrans)
 	{
 		m_updateRevision++;
 		m_worldTransform = worldTrans;
 	}
 
-
-	SIMD_FORCE_INLINE btBroadphaseProxy*	getBroadphaseHandle()
+	SIMD_FORCE_INLINE btBroadphaseProxy* getBroadphaseHandle()
 	{
 		return m_broadphaseHandle;
 	}
 
-	SIMD_FORCE_INLINE const btBroadphaseProxy*	getBroadphaseHandle() const
+	SIMD_FORCE_INLINE const btBroadphaseProxy* getBroadphaseHandle() const
 	{
 		return m_broadphaseHandle;
 	}
 
-	void	setBroadphaseHandle(btBroadphaseProxy* handle)
+	void setBroadphaseHandle(btBroadphaseProxy * handle)
 	{
 		m_broadphaseHandle = handle;
 	}
 
-
-	const btTransform&	getInterpolationWorldTransform() const
+	const btTransform& getInterpolationWorldTransform() const
 	{
 		return m_interpolationWorldTransform;
 	}
 
-	btTransform&	getInterpolationWorldTransform()
+	btTransform& getInterpolationWorldTransform()
 	{
 		return m_interpolationWorldTransform;
 	}
 
-	void	setInterpolationWorldTransform(const btTransform&	trans)
+	void setInterpolationWorldTransform(const btTransform& trans)
 	{
 		m_updateRevision++;
 		m_interpolationWorldTransform = trans;
 	}
 
-	void	setInterpolationLinearVelocity(const btVector3& linvel)
+	void setInterpolationLinearVelocity(const btVector3& linvel)
 	{
 		m_updateRevision++;
 		m_interpolationLinearVelocity = linvel;
 	}
 
-	void	setInterpolationAngularVelocity(const btVector3& angvel)
+	void setInterpolationAngularVelocity(const btVector3& angvel)
 	{
 		m_updateRevision++;
 		m_interpolationAngularVelocity = angvel;
 	}
 
-	const btVector3&	getInterpolationLinearVelocity() const
+	const btVector3& getInterpolationLinearVelocity() const
 	{
 		return m_interpolationLinearVelocity;
 	}
 
-	const btVector3&	getInterpolationAngularVelocity() const
+	const btVector3& getInterpolationAngularVelocity() const
 	{
 		return m_interpolationAngularVelocity;
 	}
 
 	SIMD_FORCE_INLINE int getIslandTag() const
 	{
-		return	m_islandTag1;
+		return m_islandTag1;
 	}
 
-	void	setIslandTag(int tag)
+	void setIslandTag(int tag)
 	{
 		m_islandTag1 = tag;
 	}
 
 	SIMD_FORCE_INLINE int getCompanionId() const
 	{
-		return	m_companionId;
+		return m_companionId;
 	}
 
-	void	setCompanionId(int id)
+	void setCompanionId(int id)
 	{
 		m_companionId = id;
 	}
 
-    SIMD_FORCE_INLINE int getWorldArrayIndex() const
-    {
-        return	m_worldArrayIndex;
-    }
-
-    // only should be called by CollisionWorld
-    void setWorldArrayIndex(int ix)
-    {
-        m_worldArrayIndex = ix;
-    }
-
-    SIMD_FORCE_INLINE btScalar			getHitFraction() const
+	SIMD_FORCE_INLINE int getWorldArrayIndex() const
 	{
-		return m_hitFraction; 
+		return m_worldArrayIndex;
 	}
 
-	void	setHitFraction(btScalar hitFraction)
+	// only should be called by CollisionWorld
+	void setWorldArrayIndex(int ix)
+	{
+		m_worldArrayIndex = ix;
+	}
+
+	SIMD_FORCE_INLINE btScalar getHitFraction() const
+	{
+		return m_hitFraction;
+	}
+
+	void setHitFraction(btScalar hitFraction)
 	{
 		m_hitFraction = hitFraction;
 	}
 
-	
-	SIMD_FORCE_INLINE int	getCollisionFlags() const
+	SIMD_FORCE_INLINE int getCollisionFlags() const
 	{
 		return m_collisionFlags;
 	}
 
-	void	setCollisionFlags(int flags)
+	void setCollisionFlags(int flags)
 	{
 		m_collisionFlags = flags;
 	}
-	
+
 	///Swept sphere radius (0.0 by default), see btConvexConvexAlgorithm::
-	btScalar			getCcdSweptSphereRadius() const
+	btScalar getCcdSweptSphereRadius() const
 	{
 		return m_ccdSweptSphereRadius;
 	}
 
 	///Swept sphere radius (0.0 by default), see btConvexConvexAlgorithm::
-	void	setCcdSweptSphereRadius(btScalar radius)
+	void setCcdSweptSphereRadius(btScalar radius)
 	{
 		m_ccdSweptSphereRadius = radius;
 	}
 
-	btScalar 	getCcdMotionThreshold() const
+	btScalar getCcdMotionThreshold() const
 	{
 		return m_ccdMotionThreshold;
 	}
 
-	btScalar 	getCcdSquareMotionThreshold() const
+	btScalar getCcdSquareMotionThreshold() const
 	{
-		return m_ccdMotionThreshold*m_ccdMotionThreshold;
+		return m_ccdMotionThreshold * m_ccdMotionThreshold;
 	}
 
-
-
 	/// Don't do continuous collision detection if the motion (in one step) is less then m_ccdMotionThreshold
-	void	setCcdMotionThreshold(btScalar ccdMotionThreshold)
+	void setCcdMotionThreshold(btScalar ccdMotionThreshold)
 	{
 		m_ccdMotionThreshold = ccdMotionThreshold;
 	}
 
 	///users can point to their objects, userPointer is not used by Bullet
-	void*	getUserPointer() const
+	void* getUserPointer() const
 	{
 		return m_userObjectPointer;
 	}
 
-	int	getUserIndex() const
+	int getUserIndex() const
 	{
 		return m_userIndex;
 	}
-	
-	int	getUserIndex2() const
+
+	int getUserIndex2() const
 	{
 		return m_userIndex2;
 	}
-	
+
 	///users can point to their objects, userPointer is not used by Bullet
-	void	setUserPointer(void* userPointer)
+	void setUserPointer(void* userPointer)
 	{
 		m_userObjectPointer = userPointer;
 	}
 
 	///users can point to their objects, userPointer is not used by Bullet
-	void	setUserIndex(int index)
+	void setUserIndex(int index)
 	{
 		m_userIndex = index;
 	}
-	
-	void	setUserIndex2(int index)
+
+	void setUserIndex2(int index)
 	{
 		m_userIndex2 = index;
 	}
 
-	int	getUpdateRevisionInternal() const
+	int getUpdateRevisionInternal() const
 	{
 		return m_updateRevision;
 	}
 
-	void	setCustomDebugColor(const btVector3& colorRGB)
+	void setCustomDebugColor(const btVector3& colorRGB)
 	{
 		m_customDebugColorRGB = colorRGB;
 		m_collisionFlags |= CF_HAS_CUSTOM_DEBUG_RENDERING_COLOR;
 	}
 
-	void	removeCustomDebugColor()
+	void removeCustomDebugColor()
 	{
 		m_collisionFlags &= ~CF_HAS_CUSTOM_DEBUG_RENDERING_COLOR;
 	}
 
-	bool getCustomDebugColor(btVector3& colorRGB) const
+	bool getCustomDebugColor(btVector3 & colorRGB) const
 	{
-		bool hasCustomColor = (0!=(m_collisionFlags&CF_HAS_CUSTOM_DEBUG_RENDERING_COLOR));
+		bool hasCustomColor = (0 != (m_collisionFlags & CF_HAS_CUSTOM_DEBUG_RENDERING_COLOR));
 		if (hasCustomColor)
 		{
 			colorRGB = m_customDebugColorRGB;
@@ -589,15 +577,16 @@ public:
 		return true;
 	}
 
-	virtual	int	calculateSerializeBufferSize()	const;
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, class btSerializer* serializer) const;
-
-	virtual void serializeSingleObject(class btSerializer* serializer) const;
+	virtual const char* serialize(void* dataBuffer, class btSerializer* serializer) const;
 
+	virtual void serializeSingleObject(class btSerializer * serializer) const;
 };
 
+// clang-format off
+
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
 struct	btCollisionObjectDoubleData
 {
@@ -667,14 +656,11 @@ struct	btCollisionObjectFloatData
 	int						m_collisionFilterMask;
 	int						m_uniqueId;
 };
+// clang-format on
 
-
-
-SIMD_FORCE_INLINE	int	btCollisionObject::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btCollisionObject::calculateSerializeBufferSize() const
 {
 	return sizeof(btCollisionObjectData);
 }
 
-
-
-#endif //BT_COLLISION_OBJECT_H
+#endif  //BT_COLLISION_OBJECT_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionObjectWrapper.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionObjectWrapper.h
index 952440b7dee..1cc4a5ac5f6 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionObjectWrapper.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionObjectWrapper.h
@@ -1,25 +1,25 @@
 #ifndef BT_COLLISION_OBJECT_WRAPPER_H
 #define BT_COLLISION_OBJECT_WRAPPER_H
 
-///btCollisionObjectWrapperis an internal data structure. 
+///btCollisionObjectWrapperis an internal data structure.
 ///Most users can ignore this and use btCollisionObject and btCollisionShape instead
 class btCollisionShape;
 class btCollisionObject;
 class btTransform;
-#include "LinearMath/btScalar.h" // for SIMD_FORCE_INLINE definition
+#include "LinearMath/btScalar.h"  // for SIMD_FORCE_INLINE definition
 
 #define BT_DECLARE_STACK_ONLY_OBJECT \
-	private: \
-		void* operator new(size_t size); \
-		void operator delete(void*);
+private:                             \
+	void* operator new(size_t size); \
+	void operator delete(void*);
 
 struct btCollisionObjectWrapper;
 struct btCollisionObjectWrapper
 {
-BT_DECLARE_STACK_ONLY_OBJECT
+	BT_DECLARE_STACK_ONLY_OBJECT
 
 private:
-	btCollisionObjectWrapper(const btCollisionObjectWrapper&); // not implemented. Not allowed.
+	btCollisionObjectWrapper(const btCollisionObjectWrapper&);  // not implemented. Not allowed.
 	btCollisionObjectWrapper* operator=(const btCollisionObjectWrapper&);
 
 public:
@@ -27,17 +27,17 @@ public:
 	const btCollisionShape* m_shape;
 	const btCollisionObject* m_collisionObject;
 	const btTransform& m_worldTransform;
-	int		m_partId;
-	int		m_index;
+	int m_partId;
+	int m_index;
 
 	btCollisionObjectWrapper(const btCollisionObjectWrapper* parent, const btCollisionShape* shape, const btCollisionObject* collisionObject, const btTransform& worldTransform, int partId, int index)
-	: m_parent(parent), m_shape(shape), m_collisionObject(collisionObject), m_worldTransform(worldTransform),
-	m_partId(partId), m_index(index)
-	{}
+		: m_parent(parent), m_shape(shape), m_collisionObject(collisionObject), m_worldTransform(worldTransform), m_partId(partId), m_index(index)
+	{
+	}
 
 	SIMD_FORCE_INLINE const btTransform& getWorldTransform() const { return m_worldTransform; }
 	SIMD_FORCE_INLINE const btCollisionObject* getCollisionObject() const { return m_collisionObject; }
 	SIMD_FORCE_INLINE const btCollisionShape* getCollisionShape() const { return m_shape; }
 };
 
-#endif //BT_COLLISION_OBJECT_WRAPPER_H
+#endif  //BT_COLLISION_OBJECT_WRAPPER_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionWorld.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionWorld.cpp
index 3de8d6995ed..782e9efaf1c 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionWorld.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionWorld.cpp
@@ -19,9 +19,9 @@ subject to the following restrictions:
 #include "BulletCollision/CollisionShapes/btCollisionShape.h"
 #include "BulletCollision/CollisionShapes/btConvexShape.h"
 #include "BulletCollision/NarrowPhaseCollision/btGjkEpaPenetrationDepthSolver.h"
-#include "BulletCollision/CollisionShapes/btSphereShape.h" //for raycasting
-#include "BulletCollision/CollisionShapes/btBvhTriangleMeshShape.h" //for raycasting
-#include "BulletCollision/CollisionShapes/btScaledBvhTriangleMeshShape.h" //for raycasting
+#include "BulletCollision/CollisionShapes/btSphereShape.h"                 //for raycasting
+#include "BulletCollision/CollisionShapes/btBvhTriangleMeshShape.h"        //for raycasting
+#include "BulletCollision/CollisionShapes/btScaledBvhTriangleMeshShape.h"  //for raycasting
 #include "BulletCollision/NarrowPhaseCollision/btRaycastCallback.h"
 #include "BulletCollision/CollisionShapes/btCompoundShape.h"
 #include "BulletCollision/NarrowPhaseCollision/btSubSimplexConvexCast.h"
@@ -38,7 +38,6 @@ subject to the following restrictions:
 
 //#define DISABLE_DBVT_COMPOUNDSHAPE_RAYCAST_ACCELERATION
 
-
 //#define USE_BRUTEFORCE_RAYBROADPHASE 1
 //RECALCULATE_AABB is slower, but benefit is that you don't need to call 'stepSimulation'  or 'updateAabbs' before using a rayTest
 //#define RECALCULATE_AABB_RAYCAST 1
@@ -48,7 +47,6 @@ subject to the following restrictions:
 #include "BulletCollision/BroadphaseCollision/btSimpleBroadphase.h"
 #include "BulletCollision/CollisionDispatch/btCollisionConfiguration.h"
 
-
 ///for debug drawing
 
 //for debug rendering
@@ -65,25 +63,21 @@ subject to the following restrictions:
 #include "BulletCollision/CollisionShapes/btTriangleMeshShape.h"
 #include "BulletCollision/CollisionShapes/btStaticPlaneShape.h"
 
-
-
-btCollisionWorld::btCollisionWorld(btDispatcher* dispatcher,btBroadphaseInterface* pairCache, btCollisionConfiguration* collisionConfiguration)
-:m_dispatcher1(dispatcher),
-m_broadphasePairCache(pairCache),
-m_debugDrawer(0),
-m_forceUpdateAllAabbs(true)
+btCollisionWorld::btCollisionWorld(btDispatcher* dispatcher, btBroadphaseInterface* pairCache, btCollisionConfiguration* collisionConfiguration)
+	: m_dispatcher1(dispatcher),
+	  m_broadphasePairCache(pairCache),
+	  m_debugDrawer(0),
+	  m_forceUpdateAllAabbs(true)
 {
 }
 
-
 btCollisionWorld::~btCollisionWorld()
 {
-
 	//clean up remaining objects
 	int i;
-	for (i=0;i<m_collisionObjects.size();i++)
+	for (i = 0; i < m_collisionObjects.size(); i++)
 	{
-		btCollisionObject* collisionObject= m_collisionObjects[i];
+		btCollisionObject* collisionObject = m_collisionObjects[i];
 
 		btBroadphaseProxy* bp = collisionObject->getBroadphaseHandle();
 		if (bp)
@@ -91,74 +85,83 @@ btCollisionWorld::~btCollisionWorld()
 			//
 			// only clear the cached algorithms
 			//
-			getBroadphase()->getOverlappingPairCache()->cleanProxyFromPairs(bp,m_dispatcher1);
-			getBroadphase()->destroyProxy(bp,m_dispatcher1);
+			getBroadphase()->getOverlappingPairCache()->cleanProxyFromPairs(bp, m_dispatcher1);
+			getBroadphase()->destroyProxy(bp, m_dispatcher1);
 			collisionObject->setBroadphaseHandle(0);
 		}
 	}
-
-
 }
 
-
-
-
-
-
-
-
-
-
-void	btCollisionWorld::addCollisionObject(btCollisionObject* collisionObject, int collisionFilterGroup, int collisionFilterMask)
+void btCollisionWorld::refreshBroadphaseProxy(btCollisionObject* collisionObject)
 {
+	if (collisionObject->getBroadphaseHandle())
+	{
+		int collisionFilterGroup = collisionObject->getBroadphaseHandle()->m_collisionFilterGroup;
+		int collisionFilterMask = collisionObject->getBroadphaseHandle()->m_collisionFilterMask;
 
+		getBroadphase()->destroyProxy(collisionObject->getBroadphaseHandle(), getDispatcher());
+
+		//calculate new AABB
+		btTransform trans = collisionObject->getWorldTransform();
+
+		btVector3 minAabb;
+		btVector3 maxAabb;
+		collisionObject->getCollisionShape()->getAabb(trans, minAabb, maxAabb);
+
+		int type = collisionObject->getCollisionShape()->getShapeType();
+		collisionObject->setBroadphaseHandle(getBroadphase()->createProxy(
+			minAabb,
+			maxAabb,
+			type,
+			collisionObject,
+			collisionFilterGroup,
+			collisionFilterMask,
+			m_dispatcher1));
+	}
+}
+
+void btCollisionWorld::addCollisionObject(btCollisionObject* collisionObject, int collisionFilterGroup, int collisionFilterMask)
+{
 	btAssert(collisionObject);
 
 	//check that the object isn't already added
-	btAssert( m_collisionObjects.findLinearSearch(collisionObject)  == m_collisionObjects.size());
-    btAssert(collisionObject->getWorldArrayIndex() == -1);  // do not add the same object to more than one collision world
+	btAssert(m_collisionObjects.findLinearSearch(collisionObject) == m_collisionObjects.size());
+	btAssert(collisionObject->getWorldArrayIndex() == -1);  // do not add the same object to more than one collision world
 
-    collisionObject->setWorldArrayIndex(m_collisionObjects.size());
+	collisionObject->setWorldArrayIndex(m_collisionObjects.size());
 	m_collisionObjects.push_back(collisionObject);
 
 	//calculate new AABB
 	btTransform trans = collisionObject->getWorldTransform();
 
-	btVector3	minAabb;
-	btVector3	maxAabb;
-	collisionObject->getCollisionShape()->getAabb(trans,minAabb,maxAabb);
+	btVector3 minAabb;
+	btVector3 maxAabb;
+	collisionObject->getCollisionShape()->getAabb(trans, minAabb, maxAabb);
 
 	int type = collisionObject->getCollisionShape()->getShapeType();
-	collisionObject->setBroadphaseHandle( getBroadphase()->createProxy(
+	collisionObject->setBroadphaseHandle(getBroadphase()->createProxy(
 		minAabb,
 		maxAabb,
 		type,
 		collisionObject,
 		collisionFilterGroup,
 		collisionFilterMask,
-		m_dispatcher1))	;
-
-
-
-
-
+		m_dispatcher1));
 }
 
-
-
-void	btCollisionWorld::updateSingleAabb(btCollisionObject* colObj)
+void btCollisionWorld::updateSingleAabb(btCollisionObject* colObj)
 {
-	btVector3 minAabb,maxAabb;
-	colObj->getCollisionShape()->getAabb(colObj->getWorldTransform(), minAabb,maxAabb);
+	btVector3 minAabb, maxAabb;
+	colObj->getCollisionShape()->getAabb(colObj->getWorldTransform(), minAabb, maxAabb);
 	//need to increase the aabb for contact thresholds
-	btVector3 contactThreshold(gContactBreakingThreshold,gContactBreakingThreshold,gContactBreakingThreshold);
+	btVector3 contactThreshold(gContactBreakingThreshold, gContactBreakingThreshold, gContactBreakingThreshold);
 	minAabb -= contactThreshold;
 	maxAabb += contactThreshold;
 
-	if(getDispatchInfo().m_useContinuous && colObj->getInternalType()==btCollisionObject::CO_RIGID_BODY && !colObj->isStaticOrKinematicObject())
+	if (getDispatchInfo().m_useContinuous && colObj->getInternalType() == btCollisionObject::CO_RIGID_BODY && !colObj->isStaticOrKinematicObject())
 	{
-		btVector3 minAabb2,maxAabb2;
-		colObj->getCollisionShape()->getAabb(colObj->getInterpolationWorldTransform(),minAabb2,maxAabb2);
+		btVector3 minAabb2, maxAabb2;
+		colObj->getCollisionShape()->getAabb(colObj->getInterpolationWorldTransform(), minAabb2, maxAabb2);
 		minAabb2 -= contactThreshold;
 		maxAabb2 += contactThreshold;
 		minAabb.setMin(minAabb2);
@@ -168,10 +171,11 @@ void	btCollisionWorld::updateSingleAabb(btCollisionObject* colObj)
 	btBroadphaseInterface* bp = (btBroadphaseInterface*)m_broadphasePairCache;
 
 	//moving objects should be moderately sized, probably something wrong if not
-	if ( colObj->isStaticObject() || ((maxAabb-minAabb).length2() < btScalar(1e12)))
+	if (colObj->isStaticObject() || ((maxAabb - minAabb).length2() < btScalar(1e12)))
 	{
-		bp->setAabb(colObj->getBroadphaseHandle(),minAabb,maxAabb, m_dispatcher1);
-	} else
+		bp->setAabb(colObj->getBroadphaseHandle(), minAabb, maxAabb, m_dispatcher1);
+	}
+	else
 	{
 		//something went wrong, investigate
 		//this assert is unwanted in 3D modelers (danger of loosing work)
@@ -189,15 +193,15 @@ void	btCollisionWorld::updateSingleAabb(btCollisionObject* colObj)
 	}
 }
 
-void	btCollisionWorld::updateAabbs()
+void btCollisionWorld::updateAabbs()
 {
 	BT_PROFILE("updateAabbs");
 
 	btTransform predictedTrans;
-	for ( int i=0;i<m_collisionObjects.size();i++)
+	for (int i = 0; i < m_collisionObjects.size(); i++)
 	{
 		btCollisionObject* colObj = m_collisionObjects[i];
-        btAssert(colObj->getWorldArrayIndex() == i);
+		btAssert(colObj->getWorldArrayIndex() == i);
 
 		//only update aabb of active objects
 		if (m_forceUpdateAllAabbs || colObj->isActive())
@@ -207,14 +211,13 @@ void	btCollisionWorld::updateAabbs()
 	}
 }
 
-
-void	btCollisionWorld::computeOverlappingPairs()
+void btCollisionWorld::computeOverlappingPairs()
 {
 	BT_PROFILE("calculateOverlappingPairs");
 	m_broadphasePairCache->calculateOverlappingPairs(m_dispatcher1);
 }
 
-void	btCollisionWorld::performDiscreteCollisionDetection()
+void btCollisionWorld::performDiscreteCollisionDetection()
 {
 	BT_PROFILE("performDiscreteCollisionDetection");
 
@@ -228,69 +231,61 @@ void	btCollisionWorld::performDiscreteCollisionDetection()
 	{
 		BT_PROFILE("dispatchAllCollisionPairs");
 		if (dispatcher)
-			dispatcher->dispatchAllCollisionPairs(m_broadphasePairCache->getOverlappingPairCache(),dispatchInfo,m_dispatcher1);
+			dispatcher->dispatchAllCollisionPairs(m_broadphasePairCache->getOverlappingPairCache(), dispatchInfo, m_dispatcher1);
 	}
-
 }
 
-
-
-void	btCollisionWorld::removeCollisionObject(btCollisionObject* collisionObject)
+void btCollisionWorld::removeCollisionObject(btCollisionObject* collisionObject)
 {
-
-
 	//bool removeFromBroadphase = false;
 
 	{
-
 		btBroadphaseProxy* bp = collisionObject->getBroadphaseHandle();
 		if (bp)
 		{
 			//
 			// only clear the cached algorithms
 			//
-			getBroadphase()->getOverlappingPairCache()->cleanProxyFromPairs(bp,m_dispatcher1);
-			getBroadphase()->destroyProxy(bp,m_dispatcher1);
+			getBroadphase()->getOverlappingPairCache()->cleanProxyFromPairs(bp, m_dispatcher1);
+			getBroadphase()->destroyProxy(bp, m_dispatcher1);
 			collisionObject->setBroadphaseHandle(0);
 		}
 	}
 
-
-    int iObj = collisionObject->getWorldArrayIndex();
-//    btAssert(iObj >= 0 && iObj < m_collisionObjects.size()); // trying to remove an object that was never added or already removed previously?
-    if (iObj >= 0 && iObj < m_collisionObjects.size())
-    {
-        btAssert(collisionObject == m_collisionObjects[iObj]);
-        m_collisionObjects.swap(iObj, m_collisionObjects.size()-1);
-        m_collisionObjects.pop_back();
-        if (iObj < m_collisionObjects.size())
-        {
-            m_collisionObjects[iObj]->setWorldArrayIndex(iObj);
-        }
-    }
-    else
-    {
-        // slow linear search
-        //swapremove
-        m_collisionObjects.remove(collisionObject);
-    }
-    collisionObject->setWorldArrayIndex(-1);
+	int iObj = collisionObject->getWorldArrayIndex();
+	//    btAssert(iObj >= 0 && iObj < m_collisionObjects.size()); // trying to remove an object that was never added or already removed previously?
+	if (iObj >= 0 && iObj < m_collisionObjects.size())
+	{
+		btAssert(collisionObject == m_collisionObjects[iObj]);
+		m_collisionObjects.swap(iObj, m_collisionObjects.size() - 1);
+		m_collisionObjects.pop_back();
+		if (iObj < m_collisionObjects.size())
+		{
+			m_collisionObjects[iObj]->setWorldArrayIndex(iObj);
+		}
+	}
+	else
+	{
+		// slow linear search
+		//swapremove
+		m_collisionObjects.remove(collisionObject);
+	}
+	collisionObject->setWorldArrayIndex(-1);
 }
 
-
-void	btCollisionWorld::rayTestSingle(const btTransform& rayFromTrans,const btTransform& rayToTrans,
-										btCollisionObject* collisionObject,
-										const btCollisionShape* collisionShape,
-										const btTransform& colObjWorldTransform,
-										RayResultCallback& resultCallback)
+void btCollisionWorld::rayTestSingle(const btTransform& rayFromTrans, const btTransform& rayToTrans,
+									 btCollisionObject* collisionObject,
+									 const btCollisionShape* collisionShape,
+									 const btTransform& colObjWorldTransform,
+									 RayResultCallback& resultCallback)
 {
-	btCollisionObjectWrapper colObWrap(0,collisionShape,collisionObject,colObjWorldTransform,-1,-1);
-	btCollisionWorld::rayTestSingleInternal(rayFromTrans,rayToTrans,&colObWrap,resultCallback);
+	btCollisionObjectWrapper colObWrap(0, collisionShape, collisionObject, colObjWorldTransform, -1, -1);
+	btCollisionWorld::rayTestSingleInternal(rayFromTrans, rayToTrans, &colObWrap, resultCallback);
 }
 
-void	btCollisionWorld::rayTestSingleInternal(const btTransform& rayFromTrans,const btTransform& rayToTrans,
-										const btCollisionObjectWrapper* collisionObjectWrap,
-										RayResultCallback& resultCallback)
+void btCollisionWorld::rayTestSingleInternal(const btTransform& rayFromTrans, const btTransform& rayToTrans,
+											 const btCollisionObjectWrapper* collisionObjectWrap,
+											 RayResultCallback& resultCallback)
 {
 	btSphereShape pointShape(btScalar(0.0));
 	pointShape.setMargin(0.f);
@@ -304,12 +299,12 @@ void	btCollisionWorld::rayTestSingleInternal(const btTransform& rayFromTrans,con
 		btConvexCast::CastResult castResult;
 		castResult.m_fraction = resultCallback.m_closestHitFraction;
 
-		btConvexShape* convexShape = (btConvexShape*) collisionShape;
-		btVoronoiSimplexSolver	simplexSolver;
-		btSubsimplexConvexCast subSimplexConvexCaster(castShape,convexShape,&simplexSolver);
-		
-		btGjkConvexCast	gjkConvexCaster(castShape,convexShape,&simplexSolver);
-		
+		btConvexShape* convexShape = (btConvexShape*)collisionShape;
+		btVoronoiSimplexSolver simplexSolver;
+		btSubsimplexConvexCast subSimplexConvexCaster(castShape, convexShape, &simplexSolver);
+
+		btGjkConvexCast gjkConvexCaster(castShape, convexShape, &simplexSolver);
+
 		//btContinuousConvexCollision convexCaster(castShape,convexShape,&simplexSolver,0);
 
 		btConvexCast* convexCasterPtr = 0;
@@ -318,10 +313,10 @@ void	btCollisionWorld::rayTestSingleInternal(const btTransform& rayFromTrans,con
 			convexCasterPtr = &gjkConvexCaster;
 		else
 			convexCasterPtr = &subSimplexConvexCaster;
-		
+
 		btConvexCast& convexCaster = *convexCasterPtr;
 
-		if (convexCaster.calcTimeOfImpact(rayFromTrans,rayToTrans,colObjWorldTransform,colObjWorldTransform,castResult))
+		if (convexCaster.calcTimeOfImpact(rayFromTrans, rayToTrans, colObjWorldTransform, colObjWorldTransform, castResult))
 		{
 			//add hit
 			if (castResult.m_normal.length2() > btScalar(0.0001))
@@ -332,81 +327,75 @@ void	btCollisionWorld::rayTestSingleInternal(const btTransform& rayFromTrans,con
 #ifdef USE_SUBSIMPLEX_CONVEX_CAST
 					//rotate normal into worldspace
 					castResult.m_normal = rayFromTrans.getBasis() * castResult.m_normal;
-#endif //USE_SUBSIMPLEX_CONVEX_CAST
+#endif  //USE_SUBSIMPLEX_CONVEX_CAST
 
 					castResult.m_normal.normalize();
-					btCollisionWorld::LocalRayResult localRayResult
-						(
+					btCollisionWorld::LocalRayResult localRayResult(
 						collisionObjectWrap->getCollisionObject(),
 						0,
 						castResult.m_normal,
-						castResult.m_fraction
-						);
+						castResult.m_fraction);
 
 					bool normalInWorldSpace = true;
 					resultCallback.addSingleResult(localRayResult, normalInWorldSpace);
-
 				}
 			}
 		}
-	} else {
+	}
+	else
+	{
 		if (collisionShape->isConcave())
 		{
-
 			//ConvexCast::CastResult
-				struct BridgeTriangleRaycastCallback : public btTriangleRaycastCallback
+			struct BridgeTriangleRaycastCallback : public btTriangleRaycastCallback
+			{
+				btCollisionWorld::RayResultCallback* m_resultCallback;
+				const btCollisionObject* m_collisionObject;
+				const btConcaveShape* m_triangleMesh;
+
+				btTransform m_colObjWorldTransform;
+
+				BridgeTriangleRaycastCallback(const btVector3& from, const btVector3& to,
+											  btCollisionWorld::RayResultCallback* resultCallback, const btCollisionObject* collisionObject, const btConcaveShape* triangleMesh, const btTransform& colObjWorldTransform) :  //@BP Mod
+																																																							btTriangleRaycastCallback(from, to, resultCallback->m_flags),
+																																																							m_resultCallback(resultCallback),
+																																																							m_collisionObject(collisionObject),
+																																																							m_triangleMesh(triangleMesh),
+																																																							m_colObjWorldTransform(colObjWorldTransform)
 				{
-					btCollisionWorld::RayResultCallback* m_resultCallback;
-					const btCollisionObject*	m_collisionObject;
-					const btConcaveShape*	m_triangleMesh;
+				}
 
-					btTransform m_colObjWorldTransform;
+				virtual btScalar reportHit(const btVector3& hitNormalLocal, btScalar hitFraction, int partId, int triangleIndex)
+				{
+					btCollisionWorld::LocalShapeInfo shapeInfo;
+					shapeInfo.m_shapePart = partId;
+					shapeInfo.m_triangleIndex = triangleIndex;
 
-					BridgeTriangleRaycastCallback( const btVector3& from,const btVector3& to,
-					btCollisionWorld::RayResultCallback* resultCallback, const btCollisionObject* collisionObject,const btConcaveShape*	triangleMesh,const btTransform& colObjWorldTransform):
-						//@BP Mod
-						btTriangleRaycastCallback(from,to, resultCallback->m_flags),
-							m_resultCallback(resultCallback),
-							m_collisionObject(collisionObject),
-							m_triangleMesh(triangleMesh),
-							m_colObjWorldTransform(colObjWorldTransform)
-						{
-						}
+					btVector3 hitNormalWorld = m_colObjWorldTransform.getBasis() * hitNormalLocal;
 
+					btCollisionWorld::LocalRayResult rayResult(m_collisionObject,
+															   &shapeInfo,
+															   hitNormalWorld,
+															   hitFraction);
 
-					virtual btScalar reportHit(const btVector3& hitNormalLocal, btScalar hitFraction, int partId, int triangleIndex )
-					{
-						btCollisionWorld::LocalShapeInfo	shapeInfo;
-						shapeInfo.m_shapePart = partId;
-						shapeInfo.m_triangleIndex = triangleIndex;
-
-						btVector3 hitNormalWorld = m_colObjWorldTransform.getBasis() * hitNormalLocal;
-
-						btCollisionWorld::LocalRayResult rayResult
-							(m_collisionObject,
-							&shapeInfo,
-							hitNormalWorld,
-							hitFraction);
-
-						bool	normalInWorldSpace = true;
-						return m_resultCallback->addSingleResult(rayResult,normalInWorldSpace);
-					}
-
-				};
+					bool normalInWorldSpace = true;
+					return m_resultCallback->addSingleResult(rayResult, normalInWorldSpace);
+				}
+			};
 
 			btTransform worldTocollisionObject = colObjWorldTransform.inverse();
 			btVector3 rayFromLocal = worldTocollisionObject * rayFromTrans.getOrigin();
 			btVector3 rayToLocal = worldTocollisionObject * rayToTrans.getOrigin();
 
 			//			BT_PROFILE("rayTestConcave");
-			if (collisionShape->getShapeType()==TRIANGLE_MESH_SHAPE_PROXYTYPE)
+			if (collisionShape->getShapeType() == TRIANGLE_MESH_SHAPE_PROXYTYPE)
 			{
 				///optimized version for btBvhTriangleMeshShape
 				btBvhTriangleMeshShape* triangleMesh = (btBvhTriangleMeshShape*)collisionShape;
-				
-				BridgeTriangleRaycastCallback rcb(rayFromLocal,rayToLocal,&resultCallback,collisionObjectWrap->getCollisionObject(),triangleMesh,colObjWorldTransform);
+
+				BridgeTriangleRaycastCallback rcb(rayFromLocal, rayToLocal, &resultCallback, collisionObjectWrap->getCollisionObject(), triangleMesh, colObjWorldTransform);
 				rcb.m_hitFraction = resultCallback.m_closestHitFraction;
-				triangleMesh->performRaycast(&rcb,rayFromLocal,rayToLocal);
+				triangleMesh->performRaycast(&rcb, rayFromLocal, rayToLocal);
 			}
 			else if (collisionShape->getShapeType() == SCALED_TRIANGLE_MESH_SHAPE_PROXYTYPE)
 			{
@@ -418,7 +407,7 @@ void	btCollisionWorld::rayTestSingleInternal(const btTransform& rayFromTrans,con
 				btVector3 scale = scaledTriangleMesh->getLocalScaling();
 				btVector3 rayFromLocalScaled = rayFromLocal / scale;
 				btVector3 rayToLocalScaled = rayToLocal / scale;
-				
+
 				//perform raycast in the underlying btBvhTriangleMeshShape
 				BridgeTriangleRaycastCallback rcb(rayFromLocalScaled, rayToLocalScaled, &resultCallback, collisionObjectWrap->getCollisionObject(), triangleMesh, colObjWorldTransform);
 				rcb.m_hitFraction = resultCallback.m_closestHitFraction;
@@ -439,45 +428,40 @@ void	btCollisionWorld::rayTestSingleInternal(const btTransform& rayFromTrans,con
 				struct BridgeTriangleRaycastCallback : public btTriangleRaycastCallback
 				{
 					btCollisionWorld::RayResultCallback* m_resultCallback;
-					const btCollisionObject*	m_collisionObject;
-					btConcaveShape*	m_triangleMesh;
+					const btCollisionObject* m_collisionObject;
+					btConcaveShape* m_triangleMesh;
 
 					btTransform m_colObjWorldTransform;
 
-					BridgeTriangleRaycastCallback( const btVector3& from,const btVector3& to,
-						btCollisionWorld::RayResultCallback* resultCallback, const btCollisionObject* collisionObject,btConcaveShape*	triangleMesh, const btTransform& colObjWorldTransform):
-					//@BP Mod
-					btTriangleRaycastCallback(from,to, resultCallback->m_flags),
-						m_resultCallback(resultCallback),
-						m_collisionObject(collisionObject),
-						m_triangleMesh(triangleMesh),
-						m_colObjWorldTransform(colObjWorldTransform)
+					BridgeTriangleRaycastCallback(const btVector3& from, const btVector3& to,
+												  btCollisionWorld::RayResultCallback* resultCallback, const btCollisionObject* collisionObject, btConcaveShape* triangleMesh, const btTransform& colObjWorldTransform) :  //@BP Mod
+																																																						  btTriangleRaycastCallback(from, to, resultCallback->m_flags),
+																																																						  m_resultCallback(resultCallback),
+																																																						  m_collisionObject(collisionObject),
+																																																						  m_triangleMesh(triangleMesh),
+																																																						  m_colObjWorldTransform(colObjWorldTransform)
 					{
 					}
 
-
-					virtual btScalar reportHit(const btVector3& hitNormalLocal, btScalar hitFraction, int partId, int triangleIndex )
+					virtual btScalar reportHit(const btVector3& hitNormalLocal, btScalar hitFraction, int partId, int triangleIndex)
 					{
-						btCollisionWorld::LocalShapeInfo	shapeInfo;
+						btCollisionWorld::LocalShapeInfo shapeInfo;
 						shapeInfo.m_shapePart = partId;
 						shapeInfo.m_triangleIndex = triangleIndex;
 
 						btVector3 hitNormalWorld = m_colObjWorldTransform.getBasis() * hitNormalLocal;
 
-						btCollisionWorld::LocalRayResult rayResult
-							(m_collisionObject,
-							&shapeInfo,
-							hitNormalWorld,
-							hitFraction);
+						btCollisionWorld::LocalRayResult rayResult(m_collisionObject,
+																   &shapeInfo,
+																   hitNormalWorld,
+																   hitFraction);
 
-						bool	normalInWorldSpace = true;
-						return m_resultCallback->addSingleResult(rayResult,normalInWorldSpace);
+						bool normalInWorldSpace = true;
+						return m_resultCallback->addSingleResult(rayResult, normalInWorldSpace);
 					}
-
 				};
 
-
-				BridgeTriangleRaycastCallback	rcb(rayFromLocal,rayToLocal,&resultCallback,collisionObjectWrap->getCollisionObject(),concaveShape, colObjWorldTransform);
+				BridgeTriangleRaycastCallback rcb(rayFromLocal, rayToLocal, &resultCallback, collisionObjectWrap->getCollisionObject(), concaveShape, colObjWorldTransform);
 				rcb.m_hitFraction = resultCallback.m_closestHitFraction;
 
 				btVector3 rayAabbMinLocal = rayFromLocal;
@@ -485,9 +469,11 @@ void	btCollisionWorld::rayTestSingleInternal(const btTransform& rayFromTrans,con
 				btVector3 rayAabbMaxLocal = rayFromLocal;
 				rayAabbMaxLocal.setMax(rayToLocal);
 
-				concaveShape->processAllTriangles(&rcb,rayAabbMinLocal,rayAabbMaxLocal);
+				concaveShape->processAllTriangles(&rcb, rayAabbMinLocal, rayAabbMaxLocal);
 			}
-		} else {
+		}
+		else
+		{
 			//			BT_PROFILE("rayTestCompound");
 			if (collisionShape->isCompound())
 			{
@@ -495,10 +481,10 @@ void	btCollisionWorld::rayTestSingleInternal(const btTransform& rayFromTrans,con
 				{
 					RayResultCallback* m_userCallback;
 					int m_i;
-					
-					LocalInfoAdder2 (int i, RayResultCallback *user)
+
+					LocalInfoAdder2(int i, RayResultCallback* user)
 						: m_userCallback(user), m_i(i)
-					{ 
+					{
 						m_closestHitFraction = m_userCallback->m_closestHitFraction;
 						m_flags = m_userCallback->m_flags;
 					}
@@ -507,7 +493,7 @@ void	btCollisionWorld::rayTestSingleInternal(const btTransform& rayFromTrans,con
 						return m_userCallback->needsCollision(p);
 					}
 
-					virtual btScalar addSingleResult (btCollisionWorld::LocalRayResult &r, bool b)
+					virtual btScalar addSingleResult(btCollisionWorld::LocalRayResult& r, bool b)
 					{
 						btCollisionWorld::LocalShapeInfo shapeInfo;
 						shapeInfo.m_shapePart = -1;
@@ -520,7 +506,7 @@ void	btCollisionWorld::rayTestSingleInternal(const btTransform& rayFromTrans,con
 						return result;
 					}
 				};
-				
+
 				struct RayTester : btDbvt::ICollide
 				{
 					const btCollisionObject* m_collisionObject;
@@ -529,33 +515,29 @@ void	btCollisionWorld::rayTestSingleInternal(const btTransform& rayFromTrans,con
 					const btTransform& m_rayFromTrans;
 					const btTransform& m_rayToTrans;
 					RayResultCallback& m_resultCallback;
-					
+
 					RayTester(const btCollisionObject* collisionObject,
-							const btCompoundShape* compoundShape,
-							const btTransform& colObjWorldTransform,
-							const btTransform& rayFromTrans,
-							const btTransform& rayToTrans,
-							RayResultCallback& resultCallback):
-						m_collisionObject(collisionObject),
-						m_compoundShape(compoundShape),
-						m_colObjWorldTransform(colObjWorldTransform),
-						m_rayFromTrans(rayFromTrans),
-						m_rayToTrans(rayToTrans),
-						m_resultCallback(resultCallback)
+							  const btCompoundShape* compoundShape,
+							  const btTransform& colObjWorldTransform,
+							  const btTransform& rayFromTrans,
+							  const btTransform& rayToTrans,
+							  RayResultCallback& resultCallback) : m_collisionObject(collisionObject),
+																   m_compoundShape(compoundShape),
+																   m_colObjWorldTransform(colObjWorldTransform),
+																   m_rayFromTrans(rayFromTrans),
+																   m_rayToTrans(rayToTrans),
+																   m_resultCallback(resultCallback)
 					{
-						
 					}
-					
+
 					void ProcessLeaf(int i)
 					{
 						const btCollisionShape* childCollisionShape = m_compoundShape->getChildShape(i);
 						const btTransform& childTrans = m_compoundShape->getChildTransform(i);
 						btTransform childWorldTrans = m_colObjWorldTransform * childTrans;
-						
-						btCollisionObjectWrapper tmpOb(0,childCollisionShape,m_collisionObject,childWorldTrans,-1,i);
-						// replace collision shape so that callback can determine the triangle
 
-						
+						btCollisionObjectWrapper tmpOb(0, childCollisionShape, m_collisionObject, childWorldTrans, -1, i);
+						// replace collision shape so that callback can determine the triangle
 
 						LocalInfoAdder2 my_cb(i, &m_resultCallback);
 
@@ -564,19 +546,17 @@ void	btCollisionWorld::rayTestSingleInternal(const btTransform& rayFromTrans,con
 							m_rayToTrans,
 							&tmpOb,
 							my_cb);
-						
 					}
-				
+
 					void Process(const btDbvtNode* leaf)
 					{
 						ProcessLeaf(leaf->dataAsInt);
 					}
 				};
-				
+
 				const btCompoundShape* compoundShape = static_cast<const btCompoundShape*>(collisionShape);
 				const btDbvt* dbvt = compoundShape->getDynamicAabbTree();
 
-
 				RayTester rayCB(
 					collisionObjectWrap->getCollisionObject(),
 					compoundShape,
@@ -584,39 +564,39 @@ void	btCollisionWorld::rayTestSingleInternal(const btTransform& rayFromTrans,con
 					rayFromTrans,
 					rayToTrans,
 					resultCallback);
-#ifndef	DISABLE_DBVT_COMPOUNDSHAPE_RAYCAST_ACCELERATION
+#ifndef DISABLE_DBVT_COMPOUNDSHAPE_RAYCAST_ACCELERATION
 				if (dbvt)
 				{
 					btVector3 localRayFrom = colObjWorldTransform.inverseTimes(rayFromTrans).getOrigin();
 					btVector3 localRayTo = colObjWorldTransform.inverseTimes(rayToTrans).getOrigin();
-					btDbvt::rayTest(dbvt->m_root, localRayFrom , localRayTo, rayCB);
+					btDbvt::rayTest(dbvt->m_root, localRayFrom, localRayTo, rayCB);
 				}
 				else
-#endif //DISABLE_DBVT_COMPOUNDSHAPE_RAYCAST_ACCELERATION
+#endif  //DISABLE_DBVT_COMPOUNDSHAPE_RAYCAST_ACCELERATION
 				{
 					for (int i = 0, n = compoundShape->getNumChildShapes(); i < n; ++i)
 					{
 						rayCB.ProcessLeaf(i);
-					}	
+					}
 				}
 			}
 		}
 	}
 }
 
-void	btCollisionWorld::objectQuerySingle(const btConvexShape* castShape,const btTransform& convexFromTrans,const btTransform& convexToTrans,
-											btCollisionObject* collisionObject,
-											const btCollisionShape* collisionShape,
-											const btTransform& colObjWorldTransform,
-											ConvexResultCallback& resultCallback, btScalar allowedPenetration)
+void btCollisionWorld::objectQuerySingle(const btConvexShape* castShape, const btTransform& convexFromTrans, const btTransform& convexToTrans,
+										 btCollisionObject* collisionObject,
+										 const btCollisionShape* collisionShape,
+										 const btTransform& colObjWorldTransform,
+										 ConvexResultCallback& resultCallback, btScalar allowedPenetration)
 {
-	btCollisionObjectWrapper tmpOb(0,collisionShape,collisionObject,colObjWorldTransform,-1,-1);
-	btCollisionWorld::objectQuerySingleInternal(castShape,convexFromTrans,convexToTrans,&tmpOb,resultCallback,allowedPenetration);
+	btCollisionObjectWrapper tmpOb(0, collisionShape, collisionObject, colObjWorldTransform, -1, -1);
+	btCollisionWorld::objectQuerySingleInternal(castShape, convexFromTrans, convexToTrans, &tmpOb, resultCallback, allowedPenetration);
 }
 
-void	btCollisionWorld::objectQuerySingleInternal(const btConvexShape* castShape,const btTransform& convexFromTrans,const btTransform& convexToTrans,
-											const btCollisionObjectWrapper* colObjWrap,
-											ConvexResultCallback& resultCallback, btScalar allowedPenetration)
+void btCollisionWorld::objectQuerySingleInternal(const btConvexShape* castShape, const btTransform& convexFromTrans, const btTransform& convexToTrans,
+												 const btCollisionObjectWrapper* colObjWrap,
+												 ConvexResultCallback& resultCallback, btScalar allowedPenetration)
 {
 	const btCollisionShape* collisionShape = colObjWrap->getCollisionShape();
 	const btTransform& colObjWorldTransform = colObjWrap->getWorldTransform();
@@ -626,21 +606,19 @@ void	btCollisionWorld::objectQuerySingleInternal(const btConvexShape* castShape,
 		//BT_PROFILE("convexSweepConvex");
 		btConvexCast::CastResult castResult;
 		castResult.m_allowedPenetration = allowedPenetration;
-		castResult.m_fraction = resultCallback.m_closestHitFraction;//btScalar(1.);//??
+		castResult.m_fraction = resultCallback.m_closestHitFraction;  //btScalar(1.);//??
 
-		btConvexShape* convexShape = (btConvexShape*) collisionShape;
-		btVoronoiSimplexSolver	simplexSolver;
-		btGjkEpaPenetrationDepthSolver	gjkEpaPenetrationSolver;
+		btConvexShape* convexShape = (btConvexShape*)collisionShape;
+		btVoronoiSimplexSolver simplexSolver;
+		btGjkEpaPenetrationDepthSolver gjkEpaPenetrationSolver;
 
-		btContinuousConvexCollision convexCaster1(castShape,convexShape,&simplexSolver,&gjkEpaPenetrationSolver);
+		btContinuousConvexCollision convexCaster1(castShape, convexShape, &simplexSolver, &gjkEpaPenetrationSolver);
 		//btGjkConvexCast convexCaster2(castShape,convexShape,&simplexSolver);
 		//btSubsimplexConvexCast convexCaster3(castShape,convexShape,&simplexSolver);
 
 		btConvexCast* castPtr = &convexCaster1;
 
-
-
-		if (castPtr->calcTimeOfImpact(convexFromTrans,convexToTrans,colObjWorldTransform,colObjWorldTransform,castResult))
+		if (castPtr->calcTimeOfImpact(convexFromTrans, convexToTrans, colObjWorldTransform, colObjWorldTransform, castResult))
 		{
 			//add hit
 			if (castResult.m_normal.length2() > btScalar(0.0001))
@@ -648,25 +626,24 @@ void	btCollisionWorld::objectQuerySingleInternal(const btConvexShape* castShape,
 				if (castResult.m_fraction < resultCallback.m_closestHitFraction)
 				{
 					castResult.m_normal.normalize();
-					btCollisionWorld::LocalConvexResult localConvexResult
-						(
+					btCollisionWorld::LocalConvexResult localConvexResult(
 						colObjWrap->getCollisionObject(),
 						0,
 						castResult.m_normal,
 						castResult.m_hitPoint,
-						castResult.m_fraction
-						);
+						castResult.m_fraction);
 
 					bool normalInWorldSpace = true;
 					resultCallback.addSingleResult(localConvexResult, normalInWorldSpace);
-
 				}
 			}
 		}
-	} else {
+	}
+	else
+	{
 		if (collisionShape->isConcave())
 		{
-			if (collisionShape->getShapeType()==TRIANGLE_MESH_SHAPE_PROXYTYPE)
+			if (collisionShape->getShapeType() == TRIANGLE_MESH_SHAPE_PROXYTYPE)
 			{
 				//BT_PROFILE("convexSweepbtBvhTriangleMesh");
 				btBvhTriangleMeshShape* triangleMesh = (btBvhTriangleMeshShape*)collisionShape;
@@ -680,62 +657,57 @@ void	btCollisionWorld::objectQuerySingleInternal(const btConvexShape* castShape,
 				struct BridgeTriangleConvexcastCallback : public btTriangleConvexcastCallback
 				{
 					btCollisionWorld::ConvexResultCallback* m_resultCallback;
-					const btCollisionObject*	m_collisionObject;
-					btTriangleMeshShape*	m_triangleMesh;
+					const btCollisionObject* m_collisionObject;
+					btTriangleMeshShape* m_triangleMesh;
 
-					BridgeTriangleConvexcastCallback(const btConvexShape* castShape, const btTransform& from,const btTransform& to,
-						btCollisionWorld::ConvexResultCallback* resultCallback, const btCollisionObject* collisionObject,btTriangleMeshShape*	triangleMesh, const btTransform& triangleToWorld):
-					btTriangleConvexcastCallback(castShape, from,to, triangleToWorld, triangleMesh->getMargin()),
-						m_resultCallback(resultCallback),
-						m_collisionObject(collisionObject),
-						m_triangleMesh(triangleMesh)
+					BridgeTriangleConvexcastCallback(const btConvexShape* castShape, const btTransform& from, const btTransform& to,
+													 btCollisionWorld::ConvexResultCallback* resultCallback, const btCollisionObject* collisionObject, btTriangleMeshShape* triangleMesh, const btTransform& triangleToWorld) : btTriangleConvexcastCallback(castShape, from, to, triangleToWorld, triangleMesh->getMargin()),
+																																																								m_resultCallback(resultCallback),
+																																																								m_collisionObject(collisionObject),
+																																																								m_triangleMesh(triangleMesh)
 					{
 					}
 
-
-					virtual btScalar reportHit(const btVector3& hitNormalLocal, const btVector3& hitPointLocal, btScalar hitFraction, int partId, int triangleIndex )
+					virtual btScalar reportHit(const btVector3& hitNormalLocal, const btVector3& hitPointLocal, btScalar hitFraction, int partId, int triangleIndex)
 					{
-						btCollisionWorld::LocalShapeInfo	shapeInfo;
+						btCollisionWorld::LocalShapeInfo shapeInfo;
 						shapeInfo.m_shapePart = partId;
 						shapeInfo.m_triangleIndex = triangleIndex;
 						if (hitFraction <= m_resultCallback->m_closestHitFraction)
 						{
+							btCollisionWorld::LocalConvexResult convexResult(m_collisionObject,
+																			 &shapeInfo,
+																			 hitNormalLocal,
+																			 hitPointLocal,
+																			 hitFraction);
 
-							btCollisionWorld::LocalConvexResult convexResult
-								(m_collisionObject,
-								&shapeInfo,
-								hitNormalLocal,
-								hitPointLocal,
-								hitFraction);
+							bool normalInWorldSpace = true;
 
-							bool	normalInWorldSpace = true;
-
-
-							return m_resultCallback->addSingleResult(convexResult,normalInWorldSpace);
+							return m_resultCallback->addSingleResult(convexResult, normalInWorldSpace);
 						}
 						return hitFraction;
 					}
-
 				};
 
-				BridgeTriangleConvexcastCallback tccb(castShape, convexFromTrans,convexToTrans,&resultCallback,colObjWrap->getCollisionObject(),triangleMesh, colObjWorldTransform);
+				BridgeTriangleConvexcastCallback tccb(castShape, convexFromTrans, convexToTrans, &resultCallback, colObjWrap->getCollisionObject(), triangleMesh, colObjWorldTransform);
 				tccb.m_hitFraction = resultCallback.m_closestHitFraction;
 				tccb.m_allowedPenetration = allowedPenetration;
 				btVector3 boxMinLocal, boxMaxLocal;
 				castShape->getAabb(rotationXform, boxMinLocal, boxMaxLocal);
-				triangleMesh->performConvexcast(&tccb,convexFromLocal,convexToLocal,boxMinLocal, boxMaxLocal);
-			} else
+				triangleMesh->performConvexcast(&tccb, convexFromLocal, convexToLocal, boxMinLocal, boxMaxLocal);
+			}
+			else
 			{
-				if (collisionShape->getShapeType()==STATIC_PLANE_PROXYTYPE)
+				if (collisionShape->getShapeType() == STATIC_PLANE_PROXYTYPE)
 				{
 					btConvexCast::CastResult castResult;
 					castResult.m_allowedPenetration = allowedPenetration;
 					castResult.m_fraction = resultCallback.m_closestHitFraction;
-					btStaticPlaneShape* planeShape = (btStaticPlaneShape*) collisionShape;
-					btContinuousConvexCollision convexCaster1(castShape,planeShape);
+					btStaticPlaneShape* planeShape = (btStaticPlaneShape*)collisionShape;
+					btContinuousConvexCollision convexCaster1(castShape, planeShape);
 					btConvexCast* castPtr = &convexCaster1;
 
-					if (castPtr->calcTimeOfImpact(convexFromTrans,convexToTrans,colObjWorldTransform,colObjWorldTransform,castResult))
+					if (castPtr->calcTimeOfImpact(convexFromTrans, convexToTrans, colObjWorldTransform, colObjWorldTransform, castResult))
 					{
 						//add hit
 						if (castResult.m_normal.length2() > btScalar(0.0001))
@@ -743,22 +715,20 @@ void	btCollisionWorld::objectQuerySingleInternal(const btConvexShape* castShape,
 							if (castResult.m_fraction < resultCallback.m_closestHitFraction)
 							{
 								castResult.m_normal.normalize();
-								btCollisionWorld::LocalConvexResult localConvexResult
-									(
+								btCollisionWorld::LocalConvexResult localConvexResult(
 									colObjWrap->getCollisionObject(),
 									0,
 									castResult.m_normal,
 									castResult.m_hitPoint,
-									castResult.m_fraction
-									);
+									castResult.m_fraction);
 
 								bool normalInWorldSpace = true;
 								resultCallback.addSingleResult(localConvexResult, normalInWorldSpace);
 							}
 						}
 					}
-
-				} else
+				}
+				else
 				{
 					//BT_PROFILE("convexSweepConcave");
 					btConcaveShape* concaveShape = (btConcaveShape*)collisionShape;
@@ -772,44 +742,39 @@ void	btCollisionWorld::objectQuerySingleInternal(const btConvexShape* castShape,
 					struct BridgeTriangleConvexcastCallback : public btTriangleConvexcastCallback
 					{
 						btCollisionWorld::ConvexResultCallback* m_resultCallback;
-						const btCollisionObject*	m_collisionObject;
-						btConcaveShape*	m_triangleMesh;
+						const btCollisionObject* m_collisionObject;
+						btConcaveShape* m_triangleMesh;
 
-						BridgeTriangleConvexcastCallback(const btConvexShape* castShape, const btTransform& from,const btTransform& to,
-							btCollisionWorld::ConvexResultCallback* resultCallback, const btCollisionObject* collisionObject,btConcaveShape*	triangleMesh, const btTransform& triangleToWorld):
-						btTriangleConvexcastCallback(castShape, from,to, triangleToWorld, triangleMesh->getMargin()),
-							m_resultCallback(resultCallback),
-							m_collisionObject(collisionObject),
-							m_triangleMesh(triangleMesh)
+						BridgeTriangleConvexcastCallback(const btConvexShape* castShape, const btTransform& from, const btTransform& to,
+														 btCollisionWorld::ConvexResultCallback* resultCallback, const btCollisionObject* collisionObject, btConcaveShape* triangleMesh, const btTransform& triangleToWorld) : btTriangleConvexcastCallback(castShape, from, to, triangleToWorld, triangleMesh->getMargin()),
+																																																							   m_resultCallback(resultCallback),
+																																																							   m_collisionObject(collisionObject),
+																																																							   m_triangleMesh(triangleMesh)
 						{
 						}
 
-
-						virtual btScalar reportHit(const btVector3& hitNormalLocal, const btVector3& hitPointLocal, btScalar hitFraction, int partId, int triangleIndex )
+						virtual btScalar reportHit(const btVector3& hitNormalLocal, const btVector3& hitPointLocal, btScalar hitFraction, int partId, int triangleIndex)
 						{
-							btCollisionWorld::LocalShapeInfo	shapeInfo;
+							btCollisionWorld::LocalShapeInfo shapeInfo;
 							shapeInfo.m_shapePart = partId;
 							shapeInfo.m_triangleIndex = triangleIndex;
 							if (hitFraction <= m_resultCallback->m_closestHitFraction)
 							{
+								btCollisionWorld::LocalConvexResult convexResult(m_collisionObject,
+																				 &shapeInfo,
+																				 hitNormalLocal,
+																				 hitPointLocal,
+																				 hitFraction);
 
-								btCollisionWorld::LocalConvexResult convexResult
-									(m_collisionObject,
-									&shapeInfo,
-									hitNormalLocal,
-									hitPointLocal,
-									hitFraction);
+								bool normalInWorldSpace = true;
 
-								bool	normalInWorldSpace = true;
-
-								return m_resultCallback->addSingleResult(convexResult,normalInWorldSpace);
+								return m_resultCallback->addSingleResult(convexResult, normalInWorldSpace);
 							}
 							return hitFraction;
 						}
-
 					};
 
-					BridgeTriangleConvexcastCallback tccb(castShape, convexFromTrans,convexToTrans,&resultCallback,colObjWrap->getCollisionObject(),concaveShape, colObjWorldTransform);
+					BridgeTriangleConvexcastCallback tccb(castShape, convexFromTrans, convexToTrans, &resultCallback, colObjWrap->getCollisionObject(), concaveShape, colObjWorldTransform);
 					tccb.m_hitFraction = resultCallback.m_closestHitFraction;
 					tccb.m_allowedPenetration = allowedPenetration;
 					btVector3 boxMinLocal, boxMaxLocal;
@@ -821,35 +786,37 @@ void	btCollisionWorld::objectQuerySingleInternal(const btConvexShape* castShape,
 					rayAabbMaxLocal.setMax(convexToLocal);
 					rayAabbMinLocal += boxMinLocal;
 					rayAabbMaxLocal += boxMaxLocal;
-					concaveShape->processAllTriangles(&tccb,rayAabbMinLocal,rayAabbMaxLocal);
+					concaveShape->processAllTriangles(&tccb, rayAabbMinLocal, rayAabbMaxLocal);
 				}
 			}
-		} else {
+		}
+		else
+		{
 			if (collisionShape->isCompound())
 			{
-				struct	btCompoundLeafCallback : btDbvt::ICollide
+				struct btCompoundLeafCallback : btDbvt::ICollide
 				{
 					btCompoundLeafCallback(
-										   const btCollisionObjectWrapper* colObjWrap,
-										   const btConvexShape* castShape,
-										   const btTransform& convexFromTrans,
-										   const btTransform& convexToTrans,
-										   btScalar allowedPenetration,
-										   const btCompoundShape* compoundShape,
-										   const btTransform& colObjWorldTransform,
-										   ConvexResultCallback& resultCallback)
-					: 
-					  m_colObjWrap(colObjWrap),
-						m_castShape(castShape),
-						m_convexFromTrans(convexFromTrans),
-						m_convexToTrans(convexToTrans),
-						m_allowedPenetration(allowedPenetration),
-						m_compoundShape(compoundShape),
-						m_colObjWorldTransform(colObjWorldTransform),
-						m_resultCallback(resultCallback) {
+						const btCollisionObjectWrapper* colObjWrap,
+						const btConvexShape* castShape,
+						const btTransform& convexFromTrans,
+						const btTransform& convexToTrans,
+						btScalar allowedPenetration,
+						const btCompoundShape* compoundShape,
+						const btTransform& colObjWorldTransform,
+						ConvexResultCallback& resultCallback)
+						: m_colObjWrap(colObjWrap),
+						  m_castShape(castShape),
+						  m_convexFromTrans(convexFromTrans),
+						  m_convexToTrans(convexToTrans),
+						  m_allowedPenetration(allowedPenetration),
+						  m_compoundShape(compoundShape),
+						  m_colObjWorldTransform(colObjWorldTransform),
+						  m_resultCallback(resultCallback)
+					{
 					}
 
-				  const btCollisionObjectWrapper* m_colObjWrap;
+					const btCollisionObjectWrapper* m_colObjWrap;
 					const btConvexShape* m_castShape;
 					const btTransform& m_convexFromTrans;
 					const btTransform& m_convexToTrans;
@@ -859,16 +826,16 @@ void	btCollisionWorld::objectQuerySingleInternal(const btConvexShape* castShape,
 					ConvexResultCallback& m_resultCallback;
 
 				public:
-
-					void		ProcessChild(int index, const btTransform& childTrans, const btCollisionShape* childCollisionShape)
+					void ProcessChild(int index, const btTransform& childTrans, const btCollisionShape* childCollisionShape)
 					{
 						btTransform childWorldTrans = m_colObjWorldTransform * childTrans;
 
-						struct	LocalInfoAdder : public ConvexResultCallback {
+						struct LocalInfoAdder : public ConvexResultCallback
+						{
 							ConvexResultCallback* m_userCallback;
 							int m_i;
 
-							LocalInfoAdder(int i, ConvexResultCallback *user)
+							LocalInfoAdder(int i, ConvexResultCallback* user)
 								: m_userCallback(user), m_i(i)
 							{
 								m_closestHitFraction = m_userCallback->m_closestHitFraction;
@@ -877,9 +844,9 @@ void	btCollisionWorld::objectQuerySingleInternal(const btConvexShape* castShape,
 							{
 								return m_userCallback->needsCollision(p);
 							}
-							virtual btScalar addSingleResult(btCollisionWorld::LocalConvexResult&	r, bool b)
+							virtual btScalar addSingleResult(btCollisionWorld::LocalConvexResult& r, bool b)
 							{
-								btCollisionWorld::LocalShapeInfo	shapeInfo;
+								btCollisionWorld::LocalShapeInfo shapeInfo;
 								shapeInfo.m_shapePart = -1;
 								shapeInfo.m_triangleIndex = m_i;
 								if (r.m_localShapeInfo == NULL)
@@ -887,7 +854,6 @@ void	btCollisionWorld::objectQuerySingleInternal(const btConvexShape* castShape,
 								const btScalar result = m_userCallback->addSingleResult(r, b);
 								m_closestHitFraction = m_userCallback->m_closestHitFraction;
 								return result;
-
 							}
 						};
 
@@ -898,7 +864,7 @@ void	btCollisionWorld::objectQuerySingleInternal(const btConvexShape* castShape,
 						objectQuerySingleInternal(m_castShape, m_convexFromTrans, m_convexToTrans, &tmpObj, my_cb, m_allowedPenetration);
 					}
 
-					void		Process(const btDbvtNode* leaf)
+					void Process(const btDbvtNode* leaf)
 					{
 						// Processing leaf node
 						int index = leaf->dataAsInt;
@@ -923,15 +889,18 @@ void	btCollisionWorld::objectQuerySingleInternal(const btConvexShape* castShape,
 				fromLocalAabbMax.setMax(toLocalAabbMax);
 
 				btCompoundLeafCallback callback(colObjWrap, castShape, convexFromTrans, convexToTrans,
-					  allowedPenetration, compoundShape, colObjWorldTransform, resultCallback);
+												allowedPenetration, compoundShape, colObjWorldTransform, resultCallback);
 
 				const btDbvt* tree = compoundShape->getDynamicAabbTree();
-				if (tree) {
-					const ATTRIBUTE_ALIGNED16(btDbvtVolume)	bounds = btDbvtVolume::FromMM(fromLocalAabbMin, fromLocalAabbMax);
+				if (tree)
+				{
+					const ATTRIBUTE_ALIGNED16(btDbvtVolume) bounds = btDbvtVolume::FromMM(fromLocalAabbMin, fromLocalAabbMax);
 					tree->collideTV(tree->m_root, bounds, callback);
-				} else {
+				}
+				else
+				{
 					int i;
-					for (i=0;i<compoundShape->getNumChildShapes();i++)
+					for (i = 0; i < compoundShape->getNumChildShapes(); i++)
 					{
 						const btCollisionShape* childCollisionShape = compoundShape->getChildShape(i);
 						btTransform childTrans = compoundShape->getChildTransform(i);
@@ -943,33 +912,31 @@ void	btCollisionWorld::objectQuerySingleInternal(const btConvexShape* castShape,
 	}
 }
 
-
 struct btSingleRayCallback : public btBroadphaseRayCallback
 {
+	btVector3 m_rayFromWorld;
+	btVector3 m_rayToWorld;
+	btTransform m_rayFromTrans;
+	btTransform m_rayToTrans;
+	btVector3 m_hitNormal;
 
-	btVector3	m_rayFromWorld;
-	btVector3	m_rayToWorld;
-	btTransform	m_rayFromTrans;
-	btTransform	m_rayToTrans;
-	btVector3	m_hitNormal;
+	const btCollisionWorld* m_world;
+	btCollisionWorld::RayResultCallback& m_resultCallback;
 
-	const btCollisionWorld*	m_world;
-	btCollisionWorld::RayResultCallback&	m_resultCallback;
-
-	btSingleRayCallback(const btVector3& rayFromWorld,const btVector3& rayToWorld,const btCollisionWorld* world,btCollisionWorld::RayResultCallback& resultCallback)
-		:m_rayFromWorld(rayFromWorld),
-		m_rayToWorld(rayToWorld),
-		m_world(world),
-		m_resultCallback(resultCallback)
+	btSingleRayCallback(const btVector3& rayFromWorld, const btVector3& rayToWorld, const btCollisionWorld* world, btCollisionWorld::RayResultCallback& resultCallback)
+		: m_rayFromWorld(rayFromWorld),
+		  m_rayToWorld(rayToWorld),
+		  m_world(world),
+		  m_resultCallback(resultCallback)
 	{
 		m_rayFromTrans.setIdentity();
 		m_rayFromTrans.setOrigin(m_rayFromWorld);
 		m_rayToTrans.setIdentity();
 		m_rayToTrans.setOrigin(m_rayToWorld);
 
-		btVector3 rayDir = (rayToWorld-rayFromWorld);
+		btVector3 rayDir = (rayToWorld - rayFromWorld);
 
-		rayDir.normalize ();
+		rayDir.normalize();
 		///what about division by zero? --> just set rayDirection[i] to INF/BT_LARGE_FLOAT
 		m_rayDirectionInverse[0] = rayDir[0] == btScalar(0.0) ? btScalar(BT_LARGE_FLOAT) : btScalar(1.0) / rayDir[0];
 		m_rayDirectionInverse[1] = rayDir[1] == btScalar(0.0) ? btScalar(BT_LARGE_FLOAT) : btScalar(1.0) / rayDir[1];
@@ -978,22 +945,19 @@ struct btSingleRayCallback : public btBroadphaseRayCallback
 		m_signs[1] = m_rayDirectionInverse[1] < 0.0;
 		m_signs[2] = m_rayDirectionInverse[2] < 0.0;
 
-		m_lambda_max = rayDir.dot(m_rayToWorld-m_rayFromWorld);
-
+		m_lambda_max = rayDir.dot(m_rayToWorld - m_rayFromWorld);
 	}
 
-
-
-	virtual bool	process(const btBroadphaseProxy* proxy)
+	virtual bool process(const btBroadphaseProxy* proxy)
 	{
 		///terminate further ray tests, once the closestHitFraction reached zero
 		if (m_resultCallback.m_closestHitFraction == btScalar(0.f))
 			return false;
 
-		btCollisionObject*	collisionObject = (btCollisionObject*)proxy->m_clientObject;
+		btCollisionObject* collisionObject = (btCollisionObject*)proxy->m_clientObject;
 
 		//only perform raycast if filterMask matches
-		if(m_resultCallback.needsCollision(collisionObject->getBroadphaseHandle())) 
+		if (m_resultCallback.needsCollision(collisionObject->getBroadphaseHandle()))
 		{
 			//RigidcollisionObject* collisionObject = ctrl->GetRigidcollisionObject();
 			//btVector3 collisionObjectAabbMin,collisionObjectAabbMax;
@@ -1011,57 +975,53 @@ struct btSingleRayCallback : public btBroadphaseRayCallback
 			//culling already done by broadphase
 			//if (btRayAabb(m_rayFromWorld,m_rayToWorld,collisionObjectAabbMin,collisionObjectAabbMax,hitLambda,m_hitNormal))
 			{
-				m_world->rayTestSingle(m_rayFromTrans,m_rayToTrans,
-					collisionObject,
-					collisionObject->getCollisionShape(),
-					collisionObject->getWorldTransform(),
-					m_resultCallback);
+				m_world->rayTestSingle(m_rayFromTrans, m_rayToTrans,
+									   collisionObject,
+									   collisionObject->getCollisionShape(),
+									   collisionObject->getWorldTransform(),
+									   m_resultCallback);
 			}
 		}
 		return true;
 	}
 };
 
-void	btCollisionWorld::rayTest(const btVector3& rayFromWorld, const btVector3& rayToWorld, RayResultCallback& resultCallback) const
+void btCollisionWorld::rayTest(const btVector3& rayFromWorld, const btVector3& rayToWorld, RayResultCallback& resultCallback) const
 {
 	//BT_PROFILE("rayTest");
 	/// use the broadphase to accelerate the search for objects, based on their aabb
 	/// and for each object with ray-aabb overlap, perform an exact ray test
-	btSingleRayCallback rayCB(rayFromWorld,rayToWorld,this,resultCallback);
+	btSingleRayCallback rayCB(rayFromWorld, rayToWorld, this, resultCallback);
 
 #ifndef USE_BRUTEFORCE_RAYBROADPHASE
-	m_broadphasePairCache->rayTest(rayFromWorld,rayToWorld,rayCB);
+	m_broadphasePairCache->rayTest(rayFromWorld, rayToWorld, rayCB);
 #else
-	for (int i=0;i<this->getNumCollisionObjects();i++)
+	for (int i = 0; i < this->getNumCollisionObjects(); i++)
 	{
 		rayCB.process(m_collisionObjects[i]->getBroadphaseHandle());
-	}	
-#endif //USE_BRUTEFORCE_RAYBROADPHASE
-
+	}
+#endif  //USE_BRUTEFORCE_RAYBROADPHASE
 }
 
-
 struct btSingleSweepCallback : public btBroadphaseRayCallback
 {
-
-	btTransform	m_convexFromTrans;
-	btTransform	m_convexToTrans;
-	btVector3	m_hitNormal;
-	const btCollisionWorld*	m_world;
-	btCollisionWorld::ConvexResultCallback&	m_resultCallback;
-	btScalar	m_allowedCcdPenetration;
+	btTransform m_convexFromTrans;
+	btTransform m_convexToTrans;
+	btVector3 m_hitNormal;
+	const btCollisionWorld* m_world;
+	btCollisionWorld::ConvexResultCallback& m_resultCallback;
+	btScalar m_allowedCcdPenetration;
 	const btConvexShape* m_castShape;
 
-
-	btSingleSweepCallback(const btConvexShape* castShape, const btTransform& convexFromTrans,const btTransform& convexToTrans,const btCollisionWorld* world,btCollisionWorld::ConvexResultCallback& resultCallback,btScalar allowedPenetration)
-		:m_convexFromTrans(convexFromTrans),
-		m_convexToTrans(convexToTrans),
-		m_world(world),
-		m_resultCallback(resultCallback),
-		m_allowedCcdPenetration(allowedPenetration),
-		m_castShape(castShape)
+	btSingleSweepCallback(const btConvexShape* castShape, const btTransform& convexFromTrans, const btTransform& convexToTrans, const btCollisionWorld* world, btCollisionWorld::ConvexResultCallback& resultCallback, btScalar allowedPenetration)
+		: m_convexFromTrans(convexFromTrans),
+		  m_convexToTrans(convexToTrans),
+		  m_world(world),
+		  m_resultCallback(resultCallback),
+		  m_allowedCcdPenetration(allowedPenetration),
+		  m_castShape(castShape)
 	{
-		btVector3 unnormalizedRayDir = (m_convexToTrans.getOrigin()-m_convexFromTrans.getOrigin());
+		btVector3 unnormalizedRayDir = (m_convexToTrans.getOrigin() - m_convexFromTrans.getOrigin());
 		btVector3 rayDir = unnormalizedRayDir.normalized();
 		///what about division by zero? --> just set rayDirection[i] to INF/BT_LARGE_FLOAT
 		m_rayDirectionInverse[0] = rayDir[0] == btScalar(0.0) ? btScalar(BT_LARGE_FLOAT) : btScalar(1.0) / rayDir[0];
@@ -1072,109 +1032,102 @@ struct btSingleSweepCallback : public btBroadphaseRayCallback
 		m_signs[2] = m_rayDirectionInverse[2] < 0.0;
 
 		m_lambda_max = rayDir.dot(unnormalizedRayDir);
-
 	}
 
-	virtual bool	process(const btBroadphaseProxy* proxy)
+	virtual bool process(const btBroadphaseProxy* proxy)
 	{
 		///terminate further convex sweep tests, once the closestHitFraction reached zero
 		if (m_resultCallback.m_closestHitFraction == btScalar(0.f))
 			return false;
 
-		btCollisionObject*	collisionObject = (btCollisionObject*)proxy->m_clientObject;
+		btCollisionObject* collisionObject = (btCollisionObject*)proxy->m_clientObject;
 
 		//only perform raycast if filterMask matches
-		if(m_resultCallback.needsCollision(collisionObject->getBroadphaseHandle())) {
+		if (m_resultCallback.needsCollision(collisionObject->getBroadphaseHandle()))
+		{
 			//RigidcollisionObject* collisionObject = ctrl->GetRigidcollisionObject();
-			m_world->objectQuerySingle(m_castShape, m_convexFromTrans,m_convexToTrans,
-				collisionObject,
-				collisionObject->getCollisionShape(),
-				collisionObject->getWorldTransform(),
-				m_resultCallback,
-				m_allowedCcdPenetration);
+			m_world->objectQuerySingle(m_castShape, m_convexFromTrans, m_convexToTrans,
+									   collisionObject,
+									   collisionObject->getCollisionShape(),
+									   collisionObject->getWorldTransform(),
+									   m_resultCallback,
+									   m_allowedCcdPenetration);
 		}
 
 		return true;
 	}
 };
 
-
-
-void	btCollisionWorld::convexSweepTest(const btConvexShape* castShape, const btTransform& convexFromWorld, const btTransform& convexToWorld, ConvexResultCallback& resultCallback, btScalar allowedCcdPenetration) const
+void btCollisionWorld::convexSweepTest(const btConvexShape* castShape, const btTransform& convexFromWorld, const btTransform& convexToWorld, ConvexResultCallback& resultCallback, btScalar allowedCcdPenetration) const
 {
-
 	BT_PROFILE("convexSweepTest");
 	/// use the broadphase to accelerate the search for objects, based on their aabb
 	/// and for each object with ray-aabb overlap, perform an exact ray test
 	/// unfortunately the implementation for rayTest and convexSweepTest duplicated, albeit practically identical
 
-
-
-	btTransform	convexFromTrans,convexToTrans;
+	btTransform convexFromTrans, convexToTrans;
 	convexFromTrans = convexFromWorld;
 	convexToTrans = convexToWorld;
 	btVector3 castShapeAabbMin, castShapeAabbMax;
 	/* Compute AABB that encompasses angular movement */
 	{
 		btVector3 linVel, angVel;
-		btTransformUtil::calculateVelocity (convexFromTrans, convexToTrans, 1.0f, linVel, angVel);
+		btTransformUtil::calculateVelocity(convexFromTrans, convexToTrans, 1.0f, linVel, angVel);
 		btVector3 zeroLinVel;
-		zeroLinVel.setValue(0,0,0);
+		zeroLinVel.setValue(0, 0, 0);
 		btTransform R;
-		R.setIdentity ();
-		R.setRotation (convexFromTrans.getRotation());
-		castShape->calculateTemporalAabb (R, zeroLinVel, angVel, 1.0f, castShapeAabbMin, castShapeAabbMax);
+		R.setIdentity();
+		R.setRotation(convexFromTrans.getRotation());
+		castShape->calculateTemporalAabb(R, zeroLinVel, angVel, 1.0f, castShapeAabbMin, castShapeAabbMax);
 	}
 
 #ifndef USE_BRUTEFORCE_RAYBROADPHASE
 
-	btSingleSweepCallback	convexCB(castShape,convexFromWorld,convexToWorld,this,resultCallback,allowedCcdPenetration);
+	btSingleSweepCallback convexCB(castShape, convexFromWorld, convexToWorld, this, resultCallback, allowedCcdPenetration);
 
-	m_broadphasePairCache->rayTest(convexFromTrans.getOrigin(),convexToTrans.getOrigin(),convexCB,castShapeAabbMin,castShapeAabbMax);
+	m_broadphasePairCache->rayTest(convexFromTrans.getOrigin(), convexToTrans.getOrigin(), convexCB, castShapeAabbMin, castShapeAabbMax);
 
 #else
 	/// go over all objects, and if the ray intersects their aabb + cast shape aabb,
 	// do a ray-shape query using convexCaster (CCD)
 	int i;
-	for (i=0;i<m_collisionObjects.size();i++)
+	for (i = 0; i < m_collisionObjects.size(); i++)
 	{
-		btCollisionObject*	collisionObject= m_collisionObjects[i];
+		btCollisionObject* collisionObject = m_collisionObjects[i];
 		//only perform raycast if filterMask matches
-		if(resultCallback.needsCollision(collisionObject->getBroadphaseHandle())) {
+		if (resultCallback.needsCollision(collisionObject->getBroadphaseHandle()))
+		{
 			//RigidcollisionObject* collisionObject = ctrl->GetRigidcollisionObject();
-			btVector3 collisionObjectAabbMin,collisionObjectAabbMax;
-			collisionObject->getCollisionShape()->getAabb(collisionObject->getWorldTransform(),collisionObjectAabbMin,collisionObjectAabbMax);
-			AabbExpand (collisionObjectAabbMin, collisionObjectAabbMax, castShapeAabbMin, castShapeAabbMax);
-			btScalar hitLambda = btScalar(1.); //could use resultCallback.m_closestHitFraction, but needs testing
+			btVector3 collisionObjectAabbMin, collisionObjectAabbMax;
+			collisionObject->getCollisionShape()->getAabb(collisionObject->getWorldTransform(), collisionObjectAabbMin, collisionObjectAabbMax);
+			AabbExpand(collisionObjectAabbMin, collisionObjectAabbMax, castShapeAabbMin, castShapeAabbMax);
+			btScalar hitLambda = btScalar(1.);  //could use resultCallback.m_closestHitFraction, but needs testing
 			btVector3 hitNormal;
-			if (btRayAabb(convexFromWorld.getOrigin(),convexToWorld.getOrigin(),collisionObjectAabbMin,collisionObjectAabbMax,hitLambda,hitNormal))
+			if (btRayAabb(convexFromWorld.getOrigin(), convexToWorld.getOrigin(), collisionObjectAabbMin, collisionObjectAabbMax, hitLambda, hitNormal))
 			{
-				objectQuerySingle(castShape, convexFromTrans,convexToTrans,
-					collisionObject,
-					collisionObject->getCollisionShape(),
-					collisionObject->getWorldTransform(),
-					resultCallback,
-					allowedCcdPenetration);
+				objectQuerySingle(castShape, convexFromTrans, convexToTrans,
+								  collisionObject,
+								  collisionObject->getCollisionShape(),
+								  collisionObject->getWorldTransform(),
+								  resultCallback,
+								  allowedCcdPenetration);
 			}
 		}
 	}
-#endif //USE_BRUTEFORCE_RAYBROADPHASE
+#endif  //USE_BRUTEFORCE_RAYBROADPHASE
 }
 
-
-
 struct btBridgedManifoldResult : public btManifoldResult
 {
+	btCollisionWorld::ContactResultCallback& m_resultCallback;
 
-	btCollisionWorld::ContactResultCallback&	m_resultCallback;
-
-	btBridgedManifoldResult( const btCollisionObjectWrapper* obj0Wrap,const btCollisionObjectWrapper* obj1Wrap,btCollisionWorld::ContactResultCallback& resultCallback )
-		:btManifoldResult(obj0Wrap,obj1Wrap),
-		m_resultCallback(resultCallback)
+	btBridgedManifoldResult(const btCollisionObjectWrapper* obj0Wrap, const btCollisionObjectWrapper* obj1Wrap, btCollisionWorld::ContactResultCallback& resultCallback)
+		: btManifoldResult(obj0Wrap, obj1Wrap),
+		  m_resultCallback(resultCallback)
 	{
 	}
 
-	virtual void addContactPoint(const btVector3& normalOnBInWorld,const btVector3& pointInWorld,btScalar depth)
+	virtual void addContactPoint(const btVector3& normalOnBInWorld, const btVector3& pointInWorld, btScalar depth)
 	{
 		bool isSwapped = m_manifoldPtr->getBody0() != m_body0Wrap->getCollisionObject();
 		btVector3 pointA = pointInWorld + normalOnBInWorld * depth;
@@ -1182,78 +1135,74 @@ struct btBridgedManifoldResult : public btManifoldResult
 		btVector3 localB;
 		if (isSwapped)
 		{
-			localA = m_body1Wrap->getCollisionObject()->getWorldTransform().invXform(pointA );
+			localA = m_body1Wrap->getCollisionObject()->getWorldTransform().invXform(pointA);
 			localB = m_body0Wrap->getCollisionObject()->getWorldTransform().invXform(pointInWorld);
-		} else
+		}
+		else
 		{
-			localA = m_body0Wrap->getCollisionObject()->getWorldTransform().invXform(pointA );
+			localA = m_body0Wrap->getCollisionObject()->getWorldTransform().invXform(pointA);
 			localB = m_body1Wrap->getCollisionObject()->getWorldTransform().invXform(pointInWorld);
 		}
-		
-		btManifoldPoint newPt(localA,localB,normalOnBInWorld,depth);
+
+		btManifoldPoint newPt(localA, localB, normalOnBInWorld, depth);
 		newPt.m_positionWorldOnA = pointA;
 		newPt.m_positionWorldOnB = pointInWorld;
-		
-	   //BP mod, store contact triangles.
+
+		//BP mod, store contact triangles.
 		if (isSwapped)
 		{
 			newPt.m_partId0 = m_partId1;
 			newPt.m_partId1 = m_partId0;
-			newPt.m_index0  = m_index1;
-			newPt.m_index1  = m_index0;
-		} else
+			newPt.m_index0 = m_index1;
+			newPt.m_index1 = m_index0;
+		}
+		else
 		{
 			newPt.m_partId0 = m_partId0;
 			newPt.m_partId1 = m_partId1;
-			newPt.m_index0  = m_index0;
-			newPt.m_index1  = m_index1;
+			newPt.m_index0 = m_index0;
+			newPt.m_index1 = m_index1;
 		}
 
 		//experimental feature info, for per-triangle material etc.
-		const btCollisionObjectWrapper* obj0Wrap = isSwapped? m_body1Wrap : m_body0Wrap;
-		const btCollisionObjectWrapper* obj1Wrap = isSwapped? m_body0Wrap : m_body1Wrap;
-		m_resultCallback.addSingleResult(newPt,obj0Wrap,newPt.m_partId0,newPt.m_index0,obj1Wrap,newPt.m_partId1,newPt.m_index1);
-
+		const btCollisionObjectWrapper* obj0Wrap = isSwapped ? m_body1Wrap : m_body0Wrap;
+		const btCollisionObjectWrapper* obj1Wrap = isSwapped ? m_body0Wrap : m_body1Wrap;
+		m_resultCallback.addSingleResult(newPt, obj0Wrap, newPt.m_partId0, newPt.m_index0, obj1Wrap, newPt.m_partId1, newPt.m_index1);
 	}
-	
 };
 
-
-
 struct btSingleContactCallback : public btBroadphaseAabbCallback
 {
-
 	btCollisionObject* m_collisionObject;
-	btCollisionWorld*	m_world;
-	btCollisionWorld::ContactResultCallback&	m_resultCallback;
-	
-	
-	btSingleContactCallback(btCollisionObject* collisionObject, btCollisionWorld* world,btCollisionWorld::ContactResultCallback& resultCallback)
-		:m_collisionObject(collisionObject),
-		m_world(world),
-		m_resultCallback(resultCallback)
+	btCollisionWorld* m_world;
+	btCollisionWorld::ContactResultCallback& m_resultCallback;
+
+	btSingleContactCallback(btCollisionObject* collisionObject, btCollisionWorld* world, btCollisionWorld::ContactResultCallback& resultCallback)
+		: m_collisionObject(collisionObject),
+		  m_world(world),
+		  m_resultCallback(resultCallback)
 	{
 	}
 
-	virtual bool	process(const btBroadphaseProxy* proxy)
+	virtual bool process(const btBroadphaseProxy* proxy)
 	{
-		btCollisionObject*	collisionObject = (btCollisionObject*)proxy->m_clientObject;
+		btCollisionObject* collisionObject = (btCollisionObject*)proxy->m_clientObject;
 		if (collisionObject == m_collisionObject)
 			return true;
 
 		//only perform raycast if filterMask matches
-		if(m_resultCallback.needsCollision(collisionObject->getBroadphaseHandle())) 
+		if (m_resultCallback.needsCollision(collisionObject->getBroadphaseHandle()))
 		{
-			btCollisionObjectWrapper ob0(0,m_collisionObject->getCollisionShape(),m_collisionObject,m_collisionObject->getWorldTransform(),-1,-1);
-			btCollisionObjectWrapper ob1(0,collisionObject->getCollisionShape(),collisionObject,collisionObject->getWorldTransform(),-1,-1);
+			btCollisionObjectWrapper ob0(0, m_collisionObject->getCollisionShape(), m_collisionObject, m_collisionObject->getWorldTransform(), -1, -1);
+			btCollisionObjectWrapper ob1(0, collisionObject->getCollisionShape(), collisionObject, collisionObject->getWorldTransform(), -1, -1);
 
-			btCollisionAlgorithm* algorithm = m_world->getDispatcher()->findAlgorithm(&ob0,&ob1,0, BT_CLOSEST_POINT_ALGORITHMS);
+			btCollisionAlgorithm* algorithm = m_world->getDispatcher()->findAlgorithm(&ob0, &ob1, 0, BT_CLOSEST_POINT_ALGORITHMS);
 			if (algorithm)
 			{
-				btBridgedManifoldResult contactPointResult(&ob0,&ob1, m_resultCallback);
+				btBridgedManifoldResult contactPointResult(&ob0, &ob1, m_resultCallback);
 				//discrete collision detection query
-				
-				algorithm->processCollision(&ob0,&ob1, m_world->getDispatchInfo(),&contactPointResult);
+
+				algorithm->processCollision(&ob0, &ob1, m_world->getDispatchInfo(), &contactPointResult);
 
 				algorithm->~btCollisionAlgorithm();
 				m_world->getDispatcher()->freeCollisionAlgorithm(algorithm);
@@ -1263,271 +1212,247 @@ struct btSingleContactCallback : public btBroadphaseAabbCallback
 	}
 };
 
-
 ///contactTest performs a discrete collision test against all objects in the btCollisionWorld, and calls the resultCallback.
 ///it reports one or more contact points for every overlapping object (including the one with deepest penetration)
-void	btCollisionWorld::contactTest( btCollisionObject* colObj, ContactResultCallback& resultCallback)
+void btCollisionWorld::contactTest(btCollisionObject* colObj, ContactResultCallback& resultCallback)
 {
-	btVector3 aabbMin,aabbMax;
-	colObj->getCollisionShape()->getAabb(colObj->getWorldTransform(),aabbMin,aabbMax);
-	btSingleContactCallback	contactCB(colObj,this,resultCallback);
-	
-	m_broadphasePairCache->aabbTest(aabbMin,aabbMax,contactCB);
-}
+	btVector3 aabbMin, aabbMax;
+	colObj->getCollisionShape()->getAabb(colObj->getWorldTransform(), aabbMin, aabbMax);
+	btSingleContactCallback contactCB(colObj, this, resultCallback);
 
+	m_broadphasePairCache->aabbTest(aabbMin, aabbMax, contactCB);
+}
 
 ///contactTest performs a discrete collision test between two collision objects and calls the resultCallback if overlap if detected.
 ///it reports one or more contact points (including the one with deepest penetration)
-void	btCollisionWorld::contactPairTest(btCollisionObject* colObjA, btCollisionObject* colObjB, ContactResultCallback& resultCallback)
+void btCollisionWorld::contactPairTest(btCollisionObject* colObjA, btCollisionObject* colObjB, ContactResultCallback& resultCallback)
 {
-	btCollisionObjectWrapper obA(0,colObjA->getCollisionShape(),colObjA,colObjA->getWorldTransform(),-1,-1);
-	btCollisionObjectWrapper obB(0,colObjB->getCollisionShape(),colObjB,colObjB->getWorldTransform(),-1,-1);
+	btCollisionObjectWrapper obA(0, colObjA->getCollisionShape(), colObjA, colObjA->getWorldTransform(), -1, -1);
+	btCollisionObjectWrapper obB(0, colObjB->getCollisionShape(), colObjB, colObjB->getWorldTransform(), -1, -1);
 
-	btCollisionAlgorithm* algorithm = getDispatcher()->findAlgorithm(&obA,&obB, 0, BT_CLOSEST_POINT_ALGORITHMS);
+	btCollisionAlgorithm* algorithm = getDispatcher()->findAlgorithm(&obA, &obB, 0, BT_CLOSEST_POINT_ALGORITHMS);
 	if (algorithm)
 	{
-		btBridgedManifoldResult contactPointResult(&obA,&obB, resultCallback);
+		btBridgedManifoldResult contactPointResult(&obA, &obB, resultCallback);
 		contactPointResult.m_closestPointDistanceThreshold = resultCallback.m_closestDistanceThreshold;
 		//discrete collision detection query
-		algorithm->processCollision(&obA,&obB, getDispatchInfo(),&contactPointResult);
+		algorithm->processCollision(&obA, &obB, getDispatchInfo(), &contactPointResult);
 
 		algorithm->~btCollisionAlgorithm();
 		getDispatcher()->freeCollisionAlgorithm(algorithm);
 	}
-
 }
 
-
-
-
 class DebugDrawcallback : public btTriangleCallback, public btInternalTriangleIndexCallback
 {
-	btIDebugDraw*	m_debugDrawer;
-	btVector3	m_color;
-	btTransform	m_worldTrans;
+	btIDebugDraw* m_debugDrawer;
+	btVector3 m_color;
+	btTransform m_worldTrans;
 
 public:
+	DebugDrawcallback(btIDebugDraw* debugDrawer, const btTransform& worldTrans, const btVector3& color) : m_debugDrawer(debugDrawer),
+																										  m_color(color),
+																										  m_worldTrans(worldTrans)
+	{
+	}
 
-	DebugDrawcallback(btIDebugDraw*	debugDrawer,const btTransform& worldTrans,const btVector3& color) :
-	  m_debugDrawer(debugDrawer),
-		  m_color(color),
-		  m_worldTrans(worldTrans)
-	  {
-	  }
+	virtual void internalProcessTriangleIndex(btVector3* triangle, int partId, int triangleIndex)
+	{
+		processTriangle(triangle, partId, triangleIndex);
+	}
 
-	  virtual void internalProcessTriangleIndex(btVector3* triangle,int partId,int  triangleIndex)
-	  {
-		  processTriangle(triangle,partId,triangleIndex);
-	  }
+	virtual void processTriangle(btVector3* triangle, int partId, int triangleIndex)
+	{
+		(void)partId;
+		(void)triangleIndex;
 
-	  virtual void processTriangle(btVector3* triangle,int partId, int triangleIndex)
-	  {
-		  (void)partId;
-		  (void)triangleIndex;
+		btVector3 wv0, wv1, wv2;
+		wv0 = m_worldTrans * triangle[0];
+		wv1 = m_worldTrans * triangle[1];
+		wv2 = m_worldTrans * triangle[2];
+		btVector3 center = (wv0 + wv1 + wv2) * btScalar(1. / 3.);
 
-		  btVector3 wv0,wv1,wv2;
-		  wv0 = m_worldTrans*triangle[0];
-		  wv1 = m_worldTrans*triangle[1];
-		  wv2 = m_worldTrans*triangle[2];
-		  btVector3 center = (wv0+wv1+wv2)*btScalar(1./3.);
-          
-          if (m_debugDrawer->getDebugMode() & btIDebugDraw::DBG_DrawNormals )
-          {
-		    btVector3 normal = (wv1-wv0).cross(wv2-wv0);
-		    normal.normalize();
-		    btVector3 normalColor(1,1,0);
-		    m_debugDrawer->drawLine(center,center+normal,normalColor);
-          }
-		  m_debugDrawer->drawLine(wv0,wv1,m_color);
-		  m_debugDrawer->drawLine(wv1,wv2,m_color);
-		  m_debugDrawer->drawLine(wv2,wv0,m_color);
-	  }
+		if (m_debugDrawer->getDebugMode() & btIDebugDraw::DBG_DrawNormals)
+		{
+			btVector3 normal = (wv1 - wv0).cross(wv2 - wv0);
+			normal.normalize();
+			btVector3 normalColor(1, 1, 0);
+			m_debugDrawer->drawLine(center, center + normal, normalColor);
+		}
+		m_debugDrawer->drawLine(wv0, wv1, m_color);
+		m_debugDrawer->drawLine(wv1, wv2, m_color);
+		m_debugDrawer->drawLine(wv2, wv0, m_color);
+	}
 };
 
-
 void btCollisionWorld::debugDrawObject(const btTransform& worldTransform, const btCollisionShape* shape, const btVector3& color)
 {
 	// Draw a small simplex at the center of the object
 	if (getDebugDrawer() && getDebugDrawer()->getDebugMode() & btIDebugDraw::DBG_DrawFrames)
 	{
-		getDebugDrawer()->drawTransform(worldTransform,.1);
+		getDebugDrawer()->drawTransform(worldTransform, .1);
 	}
 
 	if (shape->getShapeType() == COMPOUND_SHAPE_PROXYTYPE)
 	{
 		const btCompoundShape* compoundShape = static_cast<const btCompoundShape*>(shape);
-		for (int i=compoundShape->getNumChildShapes()-1;i>=0;i--)
+		for (int i = compoundShape->getNumChildShapes() - 1; i >= 0; i--)
 		{
 			btTransform childTrans = compoundShape->getChildTransform(i);
 			const btCollisionShape* colShape = compoundShape->getChildShape(i);
-			debugDrawObject(worldTransform*childTrans,colShape,color);
+			debugDrawObject(worldTransform * childTrans, colShape, color);
 		}
-
-	} else
+	}
+	else
 	{
+		switch (shape->getShapeType())
+		{
+			case BOX_SHAPE_PROXYTYPE:
+			{
+				const btBoxShape* boxShape = static_cast<const btBoxShape*>(shape);
+				btVector3 halfExtents = boxShape->getHalfExtentsWithMargin();
+				getDebugDrawer()->drawBox(-halfExtents, halfExtents, worldTransform, color);
+				break;
+			}
 
-        switch (shape->getShapeType())
-        {
+			case SPHERE_SHAPE_PROXYTYPE:
+			{
+				const btSphereShape* sphereShape = static_cast<const btSphereShape*>(shape);
+				btScalar radius = sphereShape->getMargin();  //radius doesn't include the margin, so draw with margin
 
-        case BOX_SHAPE_PROXYTYPE:
-            {
-                const btBoxShape* boxShape = static_cast<const btBoxShape*>(shape);
-                btVector3 halfExtents = boxShape->getHalfExtentsWithMargin();
-                getDebugDrawer()->drawBox(-halfExtents,halfExtents,worldTransform,color);
-                break;
-            }
+				getDebugDrawer()->drawSphere(radius, worldTransform, color);
+				break;
+			}
+			case MULTI_SPHERE_SHAPE_PROXYTYPE:
+			{
+				const btMultiSphereShape* multiSphereShape = static_cast<const btMultiSphereShape*>(shape);
 
-        case SPHERE_SHAPE_PROXYTYPE:
-            {
-                const btSphereShape* sphereShape = static_cast<const btSphereShape*>(shape);
-                btScalar radius = sphereShape->getMargin();//radius doesn't include the margin, so draw with margin
+				btTransform childTransform;
+				childTransform.setIdentity();
 
-                getDebugDrawer()->drawSphere(radius, worldTransform, color);
-                break;
-            }
-        case MULTI_SPHERE_SHAPE_PROXYTYPE:
-            {
-                const btMultiSphereShape* multiSphereShape = static_cast<const btMultiSphereShape*>(shape);
+				for (int i = multiSphereShape->getSphereCount() - 1; i >= 0; i--)
+				{
+					childTransform.setOrigin(multiSphereShape->getSpherePosition(i));
+					getDebugDrawer()->drawSphere(multiSphereShape->getSphereRadius(i), worldTransform * childTransform, color);
+				}
 
-                btTransform childTransform;
-                childTransform.setIdentity();
+				break;
+			}
+			case CAPSULE_SHAPE_PROXYTYPE:
+			{
+				const btCapsuleShape* capsuleShape = static_cast<const btCapsuleShape*>(shape);
 
-                for (int i = multiSphereShape->getSphereCount()-1; i>=0;i--)
-                {
-                    childTransform.setOrigin(multiSphereShape->getSpherePosition(i));
-                    getDebugDrawer()->drawSphere(multiSphereShape->getSphereRadius(i), worldTransform*childTransform, color);
-                }
+				btScalar radius = capsuleShape->getRadius();
+				btScalar halfHeight = capsuleShape->getHalfHeight();
 
-                break;
-            }
-        case CAPSULE_SHAPE_PROXYTYPE:
-            {
-                const btCapsuleShape* capsuleShape = static_cast<const btCapsuleShape*>(shape);
+				int upAxis = capsuleShape->getUpAxis();
+				getDebugDrawer()->drawCapsule(radius, halfHeight, upAxis, worldTransform, color);
+				break;
+			}
+			case CONE_SHAPE_PROXYTYPE:
+			{
+				const btConeShape* coneShape = static_cast<const btConeShape*>(shape);
+				btScalar radius = coneShape->getRadius();  //+coneShape->getMargin();
+				btScalar height = coneShape->getHeight();  //+coneShape->getMargin();
 
-                btScalar radius = capsuleShape->getRadius();
-                btScalar halfHeight = capsuleShape->getHalfHeight();
+				int upAxis = coneShape->getConeUpIndex();
+				getDebugDrawer()->drawCone(radius, height, upAxis, worldTransform, color);
+				break;
+			}
+			case CYLINDER_SHAPE_PROXYTYPE:
+			{
+				const btCylinderShape* cylinder = static_cast<const btCylinderShape*>(shape);
+				int upAxis = cylinder->getUpAxis();
+				btScalar radius = cylinder->getRadius();
+				btScalar halfHeight = cylinder->getHalfExtentsWithMargin()[upAxis];
+				getDebugDrawer()->drawCylinder(radius, halfHeight, upAxis, worldTransform, color);
+				break;
+			}
 
-                int upAxis = capsuleShape->getUpAxis();
-                getDebugDrawer()->drawCapsule(radius, halfHeight, upAxis, worldTransform, color);
-                break;
-            }
-        case CONE_SHAPE_PROXYTYPE:
-            {
-                const btConeShape* coneShape = static_cast<const btConeShape*>(shape);
-                btScalar radius = coneShape->getRadius();//+coneShape->getMargin();
-                btScalar height = coneShape->getHeight();//+coneShape->getMargin();
+			case STATIC_PLANE_PROXYTYPE:
+			{
+				const btStaticPlaneShape* staticPlaneShape = static_cast<const btStaticPlaneShape*>(shape);
+				btScalar planeConst = staticPlaneShape->getPlaneConstant();
+				const btVector3& planeNormal = staticPlaneShape->getPlaneNormal();
+				getDebugDrawer()->drawPlane(planeNormal, planeConst, worldTransform, color);
+				break;
+			}
+			default:
+			{
+				/// for polyhedral shapes
+				if (shape->isPolyhedral())
+				{
+					btPolyhedralConvexShape* polyshape = (btPolyhedralConvexShape*)shape;
 
-                int upAxis= coneShape->getConeUpIndex();
-                getDebugDrawer()->drawCone(radius, height, upAxis, worldTransform, color);
-                break;
+					int i;
+					if (polyshape->getConvexPolyhedron())
+					{
+						const btConvexPolyhedron* poly = polyshape->getConvexPolyhedron();
+						for (i = 0; i < poly->m_faces.size(); i++)
+						{
+							btVector3 centroid(0, 0, 0);
+							int numVerts = poly->m_faces[i].m_indices.size();
+							if (numVerts)
+							{
+								int lastV = poly->m_faces[i].m_indices[numVerts - 1];
+								for (int v = 0; v < poly->m_faces[i].m_indices.size(); v++)
+								{
+									int curVert = poly->m_faces[i].m_indices[v];
+									centroid += poly->m_vertices[curVert];
+									getDebugDrawer()->drawLine(worldTransform * poly->m_vertices[lastV], worldTransform * poly->m_vertices[curVert], color);
+									lastV = curVert;
+								}
+							}
+							centroid *= btScalar(1.f) / btScalar(numVerts);
+							if (getDebugDrawer()->getDebugMode() & btIDebugDraw::DBG_DrawNormals)
+							{
+								btVector3 normalColor(1, 1, 0);
+								btVector3 faceNormal(poly->m_faces[i].m_plane[0], poly->m_faces[i].m_plane[1], poly->m_faces[i].m_plane[2]);
+								getDebugDrawer()->drawLine(worldTransform * centroid, worldTransform * (centroid + faceNormal), normalColor);
+							}
+						}
+					}
+					else
+					{
+						for (i = 0; i < polyshape->getNumEdges(); i++)
+						{
+							btVector3 a, b;
+							polyshape->getEdge(i, a, b);
+							btVector3 wa = worldTransform * a;
+							btVector3 wb = worldTransform * b;
+							getDebugDrawer()->drawLine(wa, wb, color);
+						}
+					}
+				}
 
-            }
-        case CYLINDER_SHAPE_PROXYTYPE:
-            {
-                const btCylinderShape* cylinder = static_cast<const btCylinderShape*>(shape);
-                int upAxis = cylinder->getUpAxis();
-                btScalar radius = cylinder->getRadius();
-                btScalar halfHeight = cylinder->getHalfExtentsWithMargin()[upAxis];
-                getDebugDrawer()->drawCylinder(radius, halfHeight, upAxis, worldTransform, color);
-                break;
-            }
+				if (shape->isConcave())
+				{
+					btConcaveShape* concaveMesh = (btConcaveShape*)shape;
 
-        case STATIC_PLANE_PROXYTYPE:
-            {
-                const btStaticPlaneShape* staticPlaneShape = static_cast<const btStaticPlaneShape*>(shape);
-                btScalar planeConst = staticPlaneShape->getPlaneConstant();
-                const btVector3& planeNormal = staticPlaneShape->getPlaneNormal();
-                getDebugDrawer()->drawPlane(planeNormal, planeConst,worldTransform, color);
-                break;
+					///@todo pass camera, for some culling? no -> we are not a graphics lib
+					btVector3 aabbMax(btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT));
+					btVector3 aabbMin(btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT));
 
-            }
-        default:
-            {
+					DebugDrawcallback drawCallback(getDebugDrawer(), worldTransform, color);
+					concaveMesh->processAllTriangles(&drawCallback, aabbMin, aabbMax);
+				}
 
-                /// for polyhedral shapes
-                if (shape->isPolyhedral())
-                {
-                    btPolyhedralConvexShape* polyshape = (btPolyhedralConvexShape*) shape;
-                    
-                    int i;
-                    if (polyshape->getConvexPolyhedron())
-                    {
-                        const btConvexPolyhedron* poly = polyshape->getConvexPolyhedron();
-                        for (i=0;i<poly->m_faces.size();i++)
-                        {
-                            btVector3 centroid(0,0,0);
-                            int numVerts = poly->m_faces[i].m_indices.size();
-                            if (numVerts)
-                            {
-                                int lastV = poly->m_faces[i].m_indices[numVerts-1];
-                                for (int v=0;v<poly->m_faces[i].m_indices.size();v++)
-                                {
-                                    int curVert = poly->m_faces[i].m_indices[v];
-                                    centroid+=poly->m_vertices[curVert];
-                                    getDebugDrawer()->drawLine(worldTransform*poly->m_vertices[lastV],worldTransform*poly->m_vertices[curVert],color);
-                                    lastV = curVert;
-                                }
-                            }
-                            centroid*= btScalar(1.f)/btScalar(numVerts);
-                            if (getDebugDrawer()->getDebugMode() & btIDebugDraw::DBG_DrawNormals)
-                            {
-                                btVector3 normalColor(1,1,0);
-                                btVector3 faceNormal(poly->m_faces[i].m_plane[0],poly->m_faces[i].m_plane[1],poly->m_faces[i].m_plane[2]);
-                                getDebugDrawer()->drawLine(worldTransform*centroid,worldTransform*(centroid+faceNormal),normalColor);
-                            }
-                            
-                        }
-                        
-                        
-                    } else
-                    {
-                        for (i=0;i<polyshape->getNumEdges();i++)
-                        {
-                            btVector3 a,b;
-                            polyshape->getEdge(i,a,b);
-                            btVector3 wa = worldTransform * a;
-                            btVector3 wb = worldTransform * b;
-                            getDebugDrawer()->drawLine(wa,wb,color);
-                        }
-                    }
-                    
-                    
-                }
-                    
-                if (shape->isConcave())
-                {
-                    btConcaveShape* concaveMesh = (btConcaveShape*) shape;
-
-                    ///@todo pass camera, for some culling? no -> we are not a graphics lib
-                    btVector3 aabbMax(btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT));
-                    btVector3 aabbMin(btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT));
-
-                    DebugDrawcallback drawCallback(getDebugDrawer(),worldTransform,color);
-                    concaveMesh->processAllTriangles(&drawCallback,aabbMin,aabbMax);
-
-                }
-
-                if (shape->getShapeType() == CONVEX_TRIANGLEMESH_SHAPE_PROXYTYPE)
-                {
-                    btConvexTriangleMeshShape* convexMesh = (btConvexTriangleMeshShape*) shape;
-                    //todo: pass camera for some culling			
-                    btVector3 aabbMax(btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT));
-                    btVector3 aabbMin(btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT));
-                    //DebugDrawcallback drawCallback;
-                    DebugDrawcallback drawCallback(getDebugDrawer(),worldTransform,color);
-                    convexMesh->getMeshInterface()->InternalProcessAllTriangles(&drawCallback,aabbMin,aabbMax);
-                }
-
-
-                
-            }
-       
+				if (shape->getShapeType() == CONVEX_TRIANGLEMESH_SHAPE_PROXYTYPE)
+				{
+					btConvexTriangleMeshShape* convexMesh = (btConvexTriangleMeshShape*)shape;
+					//todo: pass camera for some culling
+					btVector3 aabbMax(btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT));
+					btVector3 aabbMin(btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT));
+					//DebugDrawcallback drawCallback;
+					DebugDrawcallback drawCallback(getDebugDrawer(), worldTransform, color);
+					convexMesh->getMeshInterface()->InternalProcessAllTriangles(&drawCallback, aabbMin, aabbMax);
+				}
+			}
 		}
 	}
 }
 
-
-void	btCollisionWorld::debugDrawWorld()
+void btCollisionWorld::debugDrawWorld()
 {
 	if (getDebugDrawer())
 	{
@@ -1535,25 +1460,23 @@ void	btCollisionWorld::debugDrawWorld()
 
 		btIDebugDraw::DefaultColors defaultColors = getDebugDrawer()->getDefaultColors();
 
-		if ( getDebugDrawer()->getDebugMode() & btIDebugDraw::DBG_DrawContactPoints)
+		if (getDebugDrawer()->getDebugMode() & btIDebugDraw::DBG_DrawContactPoints)
 		{
-		
-
 			if (getDispatcher())
 			{
 				int numManifolds = getDispatcher()->getNumManifolds();
-			
-				for (int i=0;i<numManifolds;i++)
+
+				for (int i = 0; i < numManifolds; i++)
 				{
 					btPersistentManifold* contactManifold = getDispatcher()->getManifoldByIndexInternal(i);
 					//btCollisionObject* obA = static_cast<btCollisionObject*>(contactManifold->getBody0());
 					//btCollisionObject* obB = static_cast<btCollisionObject*>(contactManifold->getBody1());
 
 					int numContacts = contactManifold->getNumContacts();
-					for (int j=0;j<numContacts;j++)
+					for (int j = 0; j < numContacts; j++)
 					{
 						btManifoldPoint& cp = contactManifold->getContactPoint(j);
-						getDebugDrawer()->drawContactPoint(cp.m_positionWorldOnB,cp.m_normalWorldOnB,cp.getDistance(),cp.getLifeTime(),defaultColors.m_contactPoint);
+						getDebugDrawer()->drawContactPoint(cp.m_positionWorldOnB, cp.m_normalWorldOnB, cp.getDistance(), cp.getLifeTime(), defaultColors.m_contactPoint);
 					}
 				}
 			}
@@ -1563,58 +1486,63 @@ void	btCollisionWorld::debugDrawWorld()
 		{
 			int i;
 
-			for (  i=0;i<m_collisionObjects.size();i++)
+			for (i = 0; i < m_collisionObjects.size(); i++)
 			{
 				btCollisionObject* colObj = m_collisionObjects[i];
-				if ((colObj->getCollisionFlags() & btCollisionObject::CF_DISABLE_VISUALIZE_OBJECT)==0)
+				if ((colObj->getCollisionFlags() & btCollisionObject::CF_DISABLE_VISUALIZE_OBJECT) == 0)
 				{
 					if (getDebugDrawer() && (getDebugDrawer()->getDebugMode() & btIDebugDraw::DBG_DrawWireframe))
 					{
-						btVector3 color(btScalar(0.4),btScalar(0.4),btScalar(0.4));
+						btVector3 color(btScalar(0.4), btScalar(0.4), btScalar(0.4));
 
-						switch(colObj->getActivationState())
+						switch (colObj->getActivationState())
 						{
-						case  ACTIVE_TAG:
-							color = defaultColors.m_activeObject; break;
-						case ISLAND_SLEEPING:
-							color =  defaultColors.m_deactivatedObject;break;
-						case WANTS_DEACTIVATION:
-							color = defaultColors.m_wantsDeactivationObject;break;
-						case DISABLE_DEACTIVATION:
-							color = defaultColors.m_disabledDeactivationObject;break;
-						case DISABLE_SIMULATION:
-							color = defaultColors.m_disabledSimulationObject;break;
-						default:
+							case ACTIVE_TAG:
+								color = defaultColors.m_activeObject;
+								break;
+							case ISLAND_SLEEPING:
+								color = defaultColors.m_deactivatedObject;
+								break;
+							case WANTS_DEACTIVATION:
+								color = defaultColors.m_wantsDeactivationObject;
+								break;
+							case DISABLE_DEACTIVATION:
+								color = defaultColors.m_disabledDeactivationObject;
+								break;
+							case DISABLE_SIMULATION:
+								color = defaultColors.m_disabledSimulationObject;
+								break;
+							default:
 							{
-								color = btVector3(btScalar(.3),btScalar(0.3),btScalar(0.3));
+								color = btVector3(btScalar(.3), btScalar(0.3), btScalar(0.3));
 							}
 						};
 
 						colObj->getCustomDebugColor(color);
 
-						debugDrawObject(colObj->getWorldTransform(),colObj->getCollisionShape(),color);
+						debugDrawObject(colObj->getWorldTransform(), colObj->getCollisionShape(), color);
 					}
 					if (m_debugDrawer && (m_debugDrawer->getDebugMode() & btIDebugDraw::DBG_DrawAabb))
 					{
-						btVector3 minAabb,maxAabb;
+						btVector3 minAabb, maxAabb;
 						btVector3 colorvec = defaultColors.m_aabb;
-						colObj->getCollisionShape()->getAabb(colObj->getWorldTransform(), minAabb,maxAabb);
-						btVector3 contactThreshold(gContactBreakingThreshold,gContactBreakingThreshold,gContactBreakingThreshold);
+						colObj->getCollisionShape()->getAabb(colObj->getWorldTransform(), minAabb, maxAabb);
+						btVector3 contactThreshold(gContactBreakingThreshold, gContactBreakingThreshold, gContactBreakingThreshold);
 						minAabb -= contactThreshold;
 						maxAabb += contactThreshold;
 
-						btVector3 minAabb2,maxAabb2;
+						btVector3 minAabb2, maxAabb2;
 
-						if(getDispatchInfo().m_useContinuous && colObj->getInternalType()==btCollisionObject::CO_RIGID_BODY && !colObj->isStaticOrKinematicObject())
+						if (getDispatchInfo().m_useContinuous && colObj->getInternalType() == btCollisionObject::CO_RIGID_BODY && !colObj->isStaticOrKinematicObject())
 						{
-							colObj->getCollisionShape()->getAabb(colObj->getInterpolationWorldTransform(),minAabb2,maxAabb2);
+							colObj->getCollisionShape()->getAabb(colObj->getInterpolationWorldTransform(), minAabb2, maxAabb2);
 							minAabb2 -= contactThreshold;
 							maxAabb2 += contactThreshold;
 							minAabb.setMin(minAabb2);
 							maxAabb.setMax(maxAabb2);
 						}
 
-						m_debugDrawer->drawAabb(minAabb,maxAabb,colorvec);
+						m_debugDrawer->drawAabb(minAabb, maxAabb, colorvec);
 					}
 				}
 			}
@@ -1622,28 +1550,27 @@ void	btCollisionWorld::debugDrawWorld()
 	}
 }
 
-
-void	btCollisionWorld::serializeCollisionObjects(btSerializer* serializer)
+void btCollisionWorld::serializeCollisionObjects(btSerializer* serializer)
 {
 	int i;
 
 	///keep track of shapes already serialized
-	btHashMap<btHashPtr,btCollisionShape*>	serializedShapes;
+	btHashMap<btHashPtr, btCollisionShape*> serializedShapes;
 
-	for (i=0;i<m_collisionObjects.size();i++)
+	for (i = 0; i < m_collisionObjects.size(); i++)
 	{
 		btCollisionObject* colObj = m_collisionObjects[i];
 		btCollisionShape* shape = colObj->getCollisionShape();
 
 		if (!serializedShapes.find(shape))
 		{
-			serializedShapes.insert(shape,shape);
+			serializedShapes.insert(shape, shape);
 			shape->serializeSingleShape(serializer);
 		}
 	}
 
 	//serialize all collision objects
-	for (i=0;i<m_collisionObjects.size();i++)
+	for (i = 0; i < m_collisionObjects.size(); i++)
 	{
 		btCollisionObject* colObj = m_collisionObjects[i];
 		if (colObj->getInternalType() == btCollisionObject::CO_COLLISION_OBJECT)
@@ -1653,8 +1580,6 @@ void	btCollisionWorld::serializeCollisionObjects(btSerializer* serializer)
 	}
 }
 
-
-
 void btCollisionWorld::serializeContactManifolds(btSerializer* serializer)
 {
 	if (serializer->getSerializationFlags() & BT_SERIALIZE_CONTACT_MANIFOLDS)
@@ -1663,7 +1588,7 @@ void btCollisionWorld::serializeContactManifolds(btSerializer* serializer)
 		for (int i = 0; i < numManifolds; i++)
 		{
 			const btPersistentManifold* manifold = getDispatcher()->getInternalManifoldPointer()[i];
-			//don't serialize empty manifolds, they just take space 
+			//don't serialize empty manifolds, they just take space
 			//(may have to do it anyway if it destroys determinism)
 			if (manifold->getNumContacts() == 0)
 				continue;
@@ -1675,16 +1600,13 @@ void btCollisionWorld::serializeContactManifolds(btSerializer* serializer)
 	}
 }
 
-
-void	btCollisionWorld::serialize(btSerializer* serializer)
+void btCollisionWorld::serialize(btSerializer* serializer)
 {
-
 	serializer->startSerialization();
-	
+
 	serializeCollisionObjects(serializer);
 
 	serializeContactManifolds(serializer);
-	
+
 	serializer->finishSerialization();
 }
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionWorld.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionWorld.h
index 886476e8ad0..fd0e5b9bbdd 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionWorld.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionWorld.h
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 /**
  * @mainpage Bullet Documentation
  *
@@ -66,8 +65,6 @@ subject to the following restrictions:
  * For up-to-data information and copyright and contributors list check out the Bullet_User_Manual.pdf
  * 
  */
- 
- 
 
 #ifndef BT_COLLISION_WORLD_H
 #define BT_COLLISION_WORLD_H
@@ -87,147 +84,138 @@ class btSerializer;
 ///CollisionWorld is interface and container for the collision detection
 class btCollisionWorld
 {
-
-	
 protected:
+	btAlignedObjectArray<btCollisionObject*> m_collisionObjects;
 
-	btAlignedObjectArray<btCollisionObject*>	m_collisionObjects;
-	
-	btDispatcher*	m_dispatcher1;
+	btDispatcher* m_dispatcher1;
 
-	btDispatcherInfo	m_dispatchInfo;
+	btDispatcherInfo m_dispatchInfo;
 
-	btBroadphaseInterface*	m_broadphasePairCache;
+	btBroadphaseInterface* m_broadphasePairCache;
 
-	btIDebugDraw*	m_debugDrawer;
+	btIDebugDraw* m_debugDrawer;
 
 	///m_forceUpdateAllAabbs can be set to false as an optimization to only update active object AABBs
 	///it is true by default, because it is error-prone (setting the position of static objects wouldn't update their AABB)
 	bool m_forceUpdateAllAabbs;
 
-	void	serializeCollisionObjects(btSerializer* serializer);
+	void serializeCollisionObjects(btSerializer* serializer);
 
 	void serializeContactManifolds(btSerializer* serializer);
 
-
 public:
-
 	//this constructor doesn't own the dispatcher and paircache/broadphase
-	btCollisionWorld(btDispatcher* dispatcher,btBroadphaseInterface* broadphasePairCache, btCollisionConfiguration* collisionConfiguration);
+	btCollisionWorld(btDispatcher* dispatcher, btBroadphaseInterface* broadphasePairCache, btCollisionConfiguration* collisionConfiguration);
 
 	virtual ~btCollisionWorld();
 
-	void	setBroadphase(btBroadphaseInterface*	pairCache)
+	void setBroadphase(btBroadphaseInterface* pairCache)
 	{
 		m_broadphasePairCache = pairCache;
 	}
 
-	const btBroadphaseInterface*	getBroadphase() const
+	const btBroadphaseInterface* getBroadphase() const
 	{
 		return m_broadphasePairCache;
 	}
 
-	btBroadphaseInterface*	getBroadphase()
+	btBroadphaseInterface* getBroadphase()
 	{
 		return m_broadphasePairCache;
 	}
 
-	btOverlappingPairCache*	getPairCache()
+	btOverlappingPairCache* getPairCache()
 	{
 		return m_broadphasePairCache->getOverlappingPairCache();
 	}
 
-
-	btDispatcher*	getDispatcher()
+	btDispatcher* getDispatcher()
 	{
 		return m_dispatcher1;
 	}
 
-	const btDispatcher*	getDispatcher() const
+	const btDispatcher* getDispatcher() const
 	{
 		return m_dispatcher1;
 	}
 
-	void	updateSingleAabb(btCollisionObject* colObj);
+	void updateSingleAabb(btCollisionObject* colObj);
 
-	virtual void	updateAabbs();
+	virtual void updateAabbs();
 
 	///the computeOverlappingPairs is usually already called by performDiscreteCollisionDetection (or stepSimulation)
 	///it can be useful to use if you perform ray tests without collision detection/simulation
-	virtual void	computeOverlappingPairs();
+	virtual void computeOverlappingPairs();
 
-	
-	virtual void	setDebugDrawer(btIDebugDraw*	debugDrawer)
+	virtual void setDebugDrawer(btIDebugDraw* debugDrawer)
 	{
-			m_debugDrawer = debugDrawer;
+		m_debugDrawer = debugDrawer;
 	}
 
-	virtual btIDebugDraw*	getDebugDrawer()
+	virtual btIDebugDraw* getDebugDrawer()
 	{
 		return m_debugDrawer;
 	}
 
-	virtual void	debugDrawWorld();
+	virtual void debugDrawWorld();
 
 	virtual void debugDrawObject(const btTransform& worldTransform, const btCollisionShape* shape, const btVector3& color);
 
-
 	///LocalShapeInfo gives extra information for complex shapes
 	///Currently, only btTriangleMeshShape is available, so it just contains triangleIndex and subpart
-	struct	LocalShapeInfo
+	struct LocalShapeInfo
 	{
-		int	m_shapePart;
-		int	m_triangleIndex;
-		
+		int m_shapePart;
+		int m_triangleIndex;
+
 		//const btCollisionShape*	m_shapeTemp;
 		//const btTransform*	m_shapeLocalTransform;
 	};
 
-	struct	LocalRayResult
+	struct LocalRayResult
 	{
-		LocalRayResult(const btCollisionObject*	collisionObject, 
-			LocalShapeInfo*	localShapeInfo,
-			const btVector3&		hitNormalLocal,
-			btScalar hitFraction)
-		:m_collisionObject(collisionObject),
-		m_localShapeInfo(localShapeInfo),
-		m_hitNormalLocal(hitNormalLocal),
-		m_hitFraction(hitFraction)
+		LocalRayResult(const btCollisionObject* collisionObject,
+					   LocalShapeInfo* localShapeInfo,
+					   const btVector3& hitNormalLocal,
+					   btScalar hitFraction)
+			: m_collisionObject(collisionObject),
+			  m_localShapeInfo(localShapeInfo),
+			  m_hitNormalLocal(hitNormalLocal),
+			  m_hitFraction(hitFraction)
 		{
 		}
 
-		const btCollisionObject*		m_collisionObject;
-		LocalShapeInfo*			m_localShapeInfo;
-		btVector3				m_hitNormalLocal;
-		btScalar				m_hitFraction;
-
+		const btCollisionObject* m_collisionObject;
+		LocalShapeInfo* m_localShapeInfo;
+		btVector3 m_hitNormalLocal;
+		btScalar m_hitFraction;
 	};
 
 	///RayResultCallback is used to report new raycast results
-	struct	RayResultCallback
+	struct RayResultCallback
 	{
-		btScalar	m_closestHitFraction;
-		const btCollisionObject*		m_collisionObject;
-		int	m_collisionFilterGroup;
-		int	m_collisionFilterMask;
+		btScalar m_closestHitFraction;
+		const btCollisionObject* m_collisionObject;
+		int m_collisionFilterGroup;
+		int m_collisionFilterMask;
 		//@BP Mod - Custom flags, currently used to enable backface culling on tri-meshes, see btRaycastCallback.h. Apply any of the EFlags defined there on m_flags here to invoke.
 		unsigned int m_flags;
 
 		virtual ~RayResultCallback()
 		{
 		}
-		bool	hasHit() const
+		bool hasHit() const
 		{
 			return (m_collisionObject != 0);
 		}
 
 		RayResultCallback()
-			:m_closestHitFraction(btScalar(1.)),
-			m_collisionObject(0),
-			m_collisionFilterGroup(btBroadphaseProxy::DefaultFilter),
-			m_collisionFilterMask(btBroadphaseProxy::AllFilter),
-			//@BP Mod
-			m_flags(0)
+			: m_closestHitFraction(btScalar(1.)),
+			  m_collisionObject(0),
+			  m_collisionFilterGroup(btBroadphaseProxy::DefaultFilter),
+			  m_collisionFilterMask(btBroadphaseProxy::AllFilter),
+			  //@BP Mod
+			  m_flags(0)
 		{
 		}
 
@@ -238,62 +226,62 @@ public:
 			return collides;
 		}
 
-
-		virtual	btScalar	addSingleResult(LocalRayResult& rayResult,bool normalInWorldSpace) = 0;
+		virtual btScalar addSingleResult(LocalRayResult& rayResult, bool normalInWorldSpace) = 0;
 	};
 
-	struct	ClosestRayResultCallback : public RayResultCallback
+	struct ClosestRayResultCallback : public RayResultCallback
 	{
-		ClosestRayResultCallback(const btVector3&	rayFromWorld,const btVector3&	rayToWorld)
-		:m_rayFromWorld(rayFromWorld),
-		m_rayToWorld(rayToWorld)
+		ClosestRayResultCallback(const btVector3& rayFromWorld, const btVector3& rayToWorld)
+			: m_rayFromWorld(rayFromWorld),
+			  m_rayToWorld(rayToWorld)
 		{
 		}
 
-		btVector3	m_rayFromWorld;//used to calculate hitPointWorld from hitFraction
-		btVector3	m_rayToWorld;
+		btVector3 m_rayFromWorld;  //used to calculate hitPointWorld from hitFraction
+		btVector3 m_rayToWorld;
 
-		btVector3	m_hitNormalWorld;
-		btVector3	m_hitPointWorld;
-			
-		virtual	btScalar	addSingleResult(LocalRayResult& rayResult,bool normalInWorldSpace)
+		btVector3 m_hitNormalWorld;
+		btVector3 m_hitPointWorld;
+
+		virtual btScalar addSingleResult(LocalRayResult& rayResult, bool normalInWorldSpace)
 		{
 			//caller already does the filter on the m_closestHitFraction
 			btAssert(rayResult.m_hitFraction <= m_closestHitFraction);
-			
+
 			m_closestHitFraction = rayResult.m_hitFraction;
 			m_collisionObject = rayResult.m_collisionObject;
 			if (normalInWorldSpace)
 			{
 				m_hitNormalWorld = rayResult.m_hitNormalLocal;
-			} else
+			}
+			else
 			{
 				///need to transform normal into worldspace
-				m_hitNormalWorld = m_collisionObject->getWorldTransform().getBasis()*rayResult.m_hitNormalLocal;
+				m_hitNormalWorld = m_collisionObject->getWorldTransform().getBasis() * rayResult.m_hitNormalLocal;
 			}
-			m_hitPointWorld.setInterpolate3(m_rayFromWorld,m_rayToWorld,rayResult.m_hitFraction);
+			m_hitPointWorld.setInterpolate3(m_rayFromWorld, m_rayToWorld, rayResult.m_hitFraction);
 			return rayResult.m_hitFraction;
 		}
 	};
 
-	struct	AllHitsRayResultCallback : public RayResultCallback
+	struct AllHitsRayResultCallback : public RayResultCallback
 	{
-		AllHitsRayResultCallback(const btVector3&	rayFromWorld,const btVector3&	rayToWorld)
-		:m_rayFromWorld(rayFromWorld),
-		m_rayToWorld(rayToWorld)
+		AllHitsRayResultCallback(const btVector3& rayFromWorld, const btVector3& rayToWorld)
+			: m_rayFromWorld(rayFromWorld),
+			  m_rayToWorld(rayToWorld)
 		{
 		}
 
-		btAlignedObjectArray<const btCollisionObject*>		m_collisionObjects;
+		btAlignedObjectArray<const btCollisionObject*> m_collisionObjects;
 
-		btVector3	m_rayFromWorld;//used to calculate hitPointWorld from hitFraction
-		btVector3	m_rayToWorld;
+		btVector3 m_rayFromWorld;  //used to calculate hitPointWorld from hitFraction
+		btVector3 m_rayToWorld;
 
-		btAlignedObjectArray<btVector3>	m_hitNormalWorld;
-		btAlignedObjectArray<btVector3>	m_hitPointWorld;
+		btAlignedObjectArray<btVector3> m_hitNormalWorld;
+		btAlignedObjectArray<btVector3> m_hitPointWorld;
 		btAlignedObjectArray<btScalar> m_hitFractions;
-			
-		virtual	btScalar	addSingleResult(LocalRayResult& rayResult,bool normalInWorldSpace)
+
+		virtual btScalar addSingleResult(LocalRayResult& rayResult, bool normalInWorldSpace)
 		{
 			m_collisionObject = rayResult.m_collisionObject;
 			m_collisionObjects.push_back(rayResult.m_collisionObject);
@@ -301,69 +289,66 @@ public:
 			if (normalInWorldSpace)
 			{
 				hitNormalWorld = rayResult.m_hitNormalLocal;
-			} else
+			}
+			else
 			{
 				///need to transform normal into worldspace
-				hitNormalWorld = m_collisionObject->getWorldTransform().getBasis()*rayResult.m_hitNormalLocal;
+				hitNormalWorld = m_collisionObject->getWorldTransform().getBasis() * rayResult.m_hitNormalLocal;
 			}
 			m_hitNormalWorld.push_back(hitNormalWorld);
 			btVector3 hitPointWorld;
-			hitPointWorld.setInterpolate3(m_rayFromWorld,m_rayToWorld,rayResult.m_hitFraction);
+			hitPointWorld.setInterpolate3(m_rayFromWorld, m_rayToWorld, rayResult.m_hitFraction);
 			m_hitPointWorld.push_back(hitPointWorld);
 			m_hitFractions.push_back(rayResult.m_hitFraction);
 			return m_closestHitFraction;
 		}
 	};
 
-
 	struct LocalConvexResult
 	{
-		LocalConvexResult(const btCollisionObject*	hitCollisionObject, 
-			LocalShapeInfo*	localShapeInfo,
-			const btVector3&		hitNormalLocal,
-			const btVector3&		hitPointLocal,
-			btScalar hitFraction
-			)
-		:m_hitCollisionObject(hitCollisionObject),
-		m_localShapeInfo(localShapeInfo),
-		m_hitNormalLocal(hitNormalLocal),
-		m_hitPointLocal(hitPointLocal),
-		m_hitFraction(hitFraction)
+		LocalConvexResult(const btCollisionObject* hitCollisionObject,
+						  LocalShapeInfo* localShapeInfo,
+						  const btVector3& hitNormalLocal,
+						  const btVector3& hitPointLocal,
+						  btScalar hitFraction)
+			: m_hitCollisionObject(hitCollisionObject),
+			  m_localShapeInfo(localShapeInfo),
+			  m_hitNormalLocal(hitNormalLocal),
+			  m_hitPointLocal(hitPointLocal),
+			  m_hitFraction(hitFraction)
 		{
 		}
 
-		const btCollisionObject*		m_hitCollisionObject;
-		LocalShapeInfo*			m_localShapeInfo;
-		btVector3				m_hitNormalLocal;
-		btVector3				m_hitPointLocal;
-		btScalar				m_hitFraction;
+		const btCollisionObject* m_hitCollisionObject;
+		LocalShapeInfo* m_localShapeInfo;
+		btVector3 m_hitNormalLocal;
+		btVector3 m_hitPointLocal;
+		btScalar m_hitFraction;
 	};
 
 	///RayResultCallback is used to report new raycast results
-	struct	ConvexResultCallback
+	struct ConvexResultCallback
 	{
-		btScalar	m_closestHitFraction;
-		int	m_collisionFilterGroup;
-		int	m_collisionFilterMask;
-		
+		btScalar m_closestHitFraction;
+		int m_collisionFilterGroup;
+		int m_collisionFilterMask;
+
 		ConvexResultCallback()
-			:m_closestHitFraction(btScalar(1.)),
-			m_collisionFilterGroup(btBroadphaseProxy::DefaultFilter),
-			m_collisionFilterMask(btBroadphaseProxy::AllFilter)
+			: m_closestHitFraction(btScalar(1.)),
+			  m_collisionFilterGroup(btBroadphaseProxy::DefaultFilter),
+			  m_collisionFilterMask(btBroadphaseProxy::AllFilter)
 		{
 		}
 
 		virtual ~ConvexResultCallback()
 		{
 		}
-		
-		bool	hasHit() const
+
+		bool hasHit() const
 		{
 			return (m_closestHitFraction < btScalar(1.));
 		}
 
-		
-
 		virtual bool needsCollision(btBroadphaseProxy* proxy0) const
 		{
 			bool collides = (proxy0->m_collisionFilterGroup & m_collisionFilterMask) != 0;
@@ -371,39 +356,40 @@ public:
 			return collides;
 		}
 
-		virtual	btScalar	addSingleResult(LocalConvexResult& convexResult,bool normalInWorldSpace) = 0;
+		virtual btScalar addSingleResult(LocalConvexResult& convexResult, bool normalInWorldSpace) = 0;
 	};
 
-	struct	ClosestConvexResultCallback : public ConvexResultCallback
+	struct ClosestConvexResultCallback : public ConvexResultCallback
 	{
-		ClosestConvexResultCallback(const btVector3&	convexFromWorld,const btVector3&	convexToWorld)
-		:m_convexFromWorld(convexFromWorld),
-		m_convexToWorld(convexToWorld),
-		m_hitCollisionObject(0)
+		ClosestConvexResultCallback(const btVector3& convexFromWorld, const btVector3& convexToWorld)
+			: m_convexFromWorld(convexFromWorld),
+			  m_convexToWorld(convexToWorld),
+			  m_hitCollisionObject(0)
 		{
 		}
 
-		btVector3	m_convexFromWorld;//used to calculate hitPointWorld from hitFraction
-		btVector3	m_convexToWorld;
+		btVector3 m_convexFromWorld;  //used to calculate hitPointWorld from hitFraction
+		btVector3 m_convexToWorld;
 
-		btVector3	m_hitNormalWorld;
-		btVector3	m_hitPointWorld;
-		const btCollisionObject*	m_hitCollisionObject;
-		
-		virtual	btScalar	addSingleResult(LocalConvexResult& convexResult,bool normalInWorldSpace)
+		btVector3 m_hitNormalWorld;
+		btVector3 m_hitPointWorld;
+		const btCollisionObject* m_hitCollisionObject;
+
+		virtual btScalar addSingleResult(LocalConvexResult& convexResult, bool normalInWorldSpace)
 		{
-//caller already does the filter on the m_closestHitFraction
+			//caller already does the filter on the m_closestHitFraction
 			btAssert(convexResult.m_hitFraction <= m_closestHitFraction);
-						
+
 			m_closestHitFraction = convexResult.m_hitFraction;
 			m_hitCollisionObject = convexResult.m_hitCollisionObject;
 			if (normalInWorldSpace)
 			{
 				m_hitNormalWorld = convexResult.m_hitNormalLocal;
-			} else
+			}
+			else
 			{
 				///need to transform normal into worldspace
-				m_hitNormalWorld = m_hitCollisionObject->getWorldTransform().getBasis()*convexResult.m_hitNormalLocal;
+				m_hitNormalWorld = m_hitCollisionObject->getWorldTransform().getBasis() * convexResult.m_hitNormalLocal;
 			}
 			m_hitPointWorld = convexResult.m_hitPointLocal;
 			return convexResult.m_hitFraction;
@@ -411,23 +397,23 @@ public:
 	};
 
 	///ContactResultCallback is used to report contact points
-	struct	ContactResultCallback
+	struct ContactResultCallback
 	{
-		int	m_collisionFilterGroup;
-		int	m_collisionFilterMask;
-		btScalar	m_closestDistanceThreshold;
+		int m_collisionFilterGroup;
+		int m_collisionFilterMask;
+		btScalar m_closestDistanceThreshold;
 
 		ContactResultCallback()
-			:m_collisionFilterGroup(btBroadphaseProxy::DefaultFilter),
-			m_collisionFilterMask(btBroadphaseProxy::AllFilter),
-			m_closestDistanceThreshold(0)
+			: m_collisionFilterGroup(btBroadphaseProxy::DefaultFilter),
+			  m_collisionFilterMask(btBroadphaseProxy::AllFilter),
+			  m_closestDistanceThreshold(0)
 		{
 		}
 
 		virtual ~ContactResultCallback()
 		{
 		}
-		
+
 		virtual bool needsCollision(btBroadphaseProxy* proxy0) const
 		{
 			bool collides = (proxy0->m_collisionFilterGroup & m_collisionFilterMask) != 0;
@@ -435,58 +421,57 @@ public:
 			return collides;
 		}
 
-		virtual	btScalar	addSingleResult(btManifoldPoint& cp,	const btCollisionObjectWrapper* colObj0Wrap,int partId0,int index0,const btCollisionObjectWrapper* colObj1Wrap,int partId1,int index1) = 0;
+		virtual btScalar addSingleResult(btManifoldPoint& cp, const btCollisionObjectWrapper* colObj0Wrap, int partId0, int index0, const btCollisionObjectWrapper* colObj1Wrap, int partId1, int index1) = 0;
 	};
 
-
-
-	int	getNumCollisionObjects() const
+	int getNumCollisionObjects() const
 	{
 		return int(m_collisionObjects.size());
 	}
 
 	/// rayTest performs a raycast on all objects in the btCollisionWorld, and calls the resultCallback
 	/// This allows for several queries: first hit, all hits, any hit, dependent on the value returned by the callback.
-	virtual void rayTest(const btVector3& rayFromWorld, const btVector3& rayToWorld, RayResultCallback& resultCallback) const; 
+	virtual void rayTest(const btVector3& rayFromWorld, const btVector3& rayToWorld, RayResultCallback& resultCallback) const;
 
 	/// convexTest performs a swept convex cast on all objects in the btCollisionWorld, and calls the resultCallback
 	/// This allows for several queries: first hit, all hits, any hit, dependent on the value return by the callback.
-	void    convexSweepTest (const btConvexShape* castShape, const btTransform& from, const btTransform& to, ConvexResultCallback& resultCallback,  btScalar allowedCcdPenetration = btScalar(0.)) const;
+	void convexSweepTest(const btConvexShape* castShape, const btTransform& from, const btTransform& to, ConvexResultCallback& resultCallback, btScalar allowedCcdPenetration = btScalar(0.)) const;
 
 	///contactTest performs a discrete collision test between colObj against all objects in the btCollisionWorld, and calls the resultCallback.
 	///it reports one or more contact points for every overlapping object (including the one with deepest penetration)
-	void	contactTest(btCollisionObject* colObj, ContactResultCallback& resultCallback);
+	void contactTest(btCollisionObject* colObj, ContactResultCallback& resultCallback);
 
 	///contactTest performs a discrete collision test between two collision objects and calls the resultCallback if overlap if detected.
 	///it reports one or more contact points (including the one with deepest penetration)
-	void	contactPairTest(btCollisionObject* colObjA, btCollisionObject* colObjB, ContactResultCallback& resultCallback);
-
+	void contactPairTest(btCollisionObject* colObjA, btCollisionObject* colObjB, ContactResultCallback& resultCallback);
 
 	/// rayTestSingle performs a raycast call and calls the resultCallback. It is used internally by rayTest.
 	/// In a future implementation, we consider moving the ray test as a virtual method in btCollisionShape.
 	/// This allows more customization.
-	static void	rayTestSingle(const btTransform& rayFromTrans,const btTransform& rayToTrans,
-					  btCollisionObject* collisionObject,
-					  const btCollisionShape* collisionShape,
-					  const btTransform& colObjWorldTransform,
-					  RayResultCallback& resultCallback);
+	static void rayTestSingle(const btTransform& rayFromTrans, const btTransform& rayToTrans,
+							  btCollisionObject* collisionObject,
+							  const btCollisionShape* collisionShape,
+							  const btTransform& colObjWorldTransform,
+							  RayResultCallback& resultCallback);
 
-	static void	rayTestSingleInternal(const btTransform& rayFromTrans,const btTransform& rayToTrans,
-					  const btCollisionObjectWrapper* collisionObjectWrap,
-					  RayResultCallback& resultCallback);
+	static void rayTestSingleInternal(const btTransform& rayFromTrans, const btTransform& rayToTrans,
+									  const btCollisionObjectWrapper* collisionObjectWrap,
+									  RayResultCallback& resultCallback);
 
 	/// objectQuerySingle performs a collision detection query and calls the resultCallback. It is used internally by rayTest.
-	static void	objectQuerySingle(const btConvexShape* castShape, const btTransform& rayFromTrans,const btTransform& rayToTrans,
-					  btCollisionObject* collisionObject,
-					  const btCollisionShape* collisionShape,
-					  const btTransform& colObjWorldTransform,
-					  ConvexResultCallback& resultCallback, btScalar	allowedPenetration);
+	static void objectQuerySingle(const btConvexShape* castShape, const btTransform& rayFromTrans, const btTransform& rayToTrans,
+								  btCollisionObject* collisionObject,
+								  const btCollisionShape* collisionShape,
+								  const btTransform& colObjWorldTransform,
+								  ConvexResultCallback& resultCallback, btScalar allowedPenetration);
 
-	static void	objectQuerySingleInternal(const btConvexShape* castShape,const btTransform& convexFromTrans,const btTransform& convexToTrans,
-											const btCollisionObjectWrapper* colObjWrap,
-											ConvexResultCallback& resultCallback, btScalar allowedPenetration);
+	static void objectQuerySingleInternal(const btConvexShape* castShape, const btTransform& convexFromTrans, const btTransform& convexToTrans,
+										  const btCollisionObjectWrapper* colObjWrap,
+										  ConvexResultCallback& resultCallback, btScalar allowedPenetration);
 
-	virtual void	addCollisionObject(btCollisionObject* collisionObject, int collisionFilterGroup=btBroadphaseProxy::DefaultFilter, int collisionFilterMask=btBroadphaseProxy::AllFilter);
+	virtual void addCollisionObject(btCollisionObject* collisionObject, int collisionFilterGroup = btBroadphaseProxy::DefaultFilter, int collisionFilterMask = btBroadphaseProxy::AllFilter);
+
+	virtual void refreshBroadphaseProxy(btCollisionObject* collisionObject);
 
 	btCollisionObjectArray& getCollisionObjectArray()
 	{
@@ -498,10 +483,9 @@ public:
 		return m_collisionObjects;
 	}
 
+	virtual void removeCollisionObject(btCollisionObject* collisionObject);
 
-	virtual void	removeCollisionObject(btCollisionObject* collisionObject);
-
-	virtual void	performDiscreteCollisionDetection();
+	virtual void performDiscreteCollisionDetection();
 
 	btDispatcherInfo& getDispatchInfo()
 	{
@@ -512,20 +496,18 @@ public:
 	{
 		return m_dispatchInfo;
 	}
-	
-	bool	getForceUpdateAllAabbs() const
+
+	bool getForceUpdateAllAabbs() const
 	{
 		return m_forceUpdateAllAabbs;
 	}
-	void setForceUpdateAllAabbs( bool forceUpdateAllAabbs)
+	void setForceUpdateAllAabbs(bool forceUpdateAllAabbs)
 	{
 		m_forceUpdateAllAabbs = forceUpdateAllAabbs;
 	}
 
 	///Preliminary serialization test for Bullet 2.76. Loading those files requires a separate parser (Bullet/Demos/SerializeDemo)
-	virtual	void	serialize(btSerializer* serializer);
-
+	virtual void serialize(btSerializer* serializer);
 };
 
-
-#endif //BT_COLLISION_WORLD_H
+#endif  //BT_COLLISION_WORLD_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionWorldImporter.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionWorldImporter.cpp
index f2b08378086..9f422dc157a 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionWorldImporter.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionWorldImporter.cpp
@@ -15,269 +15,251 @@ subject to the following restrictions:
 
 #include "btCollisionWorldImporter.h"
 #include "btBulletCollisionCommon.h"
-#include "LinearMath/btSerializer.h" //for btBulletSerializedArrays definition
+#include "LinearMath/btSerializer.h"  //for btBulletSerializedArrays definition
 
 #ifdef SUPPORT_GIMPACT_SHAPE_IMPORT
 #include "BulletCollision/Gimpact/btGImpactShape.h"
-#endif //SUPPORT_GIMPACT_SHAPE_IMPORT
+#endif  //SUPPORT_GIMPACT_SHAPE_IMPORT
 
 btCollisionWorldImporter::btCollisionWorldImporter(btCollisionWorld* world)
-:m_collisionWorld(world),
-m_verboseMode(0)
+	: m_collisionWorld(world),
+	  m_verboseMode(0)
 {
-
 }
 
 btCollisionWorldImporter::~btCollisionWorldImporter()
 {
 }
 
-
-
-
-
-bool	btCollisionWorldImporter::convertAllObjects( btBulletSerializedArrays* arrays)
+bool btCollisionWorldImporter::convertAllObjects(btBulletSerializedArrays* arrays)
 {
-
 	m_shapeMap.clear();
 	m_bodyMap.clear();
 
 	int i;
 
-	for (i=0;i<arrays->m_bvhsDouble.size();i++)
+	for (i = 0; i < arrays->m_bvhsDouble.size(); i++)
 	{
 		btOptimizedBvh* bvh = createOptimizedBvh();
 		btQuantizedBvhDoubleData* bvhData = arrays->m_bvhsDouble[i];
 		bvh->deSerializeDouble(*bvhData);
-		m_bvhMap.insert(arrays->m_bvhsDouble[i],bvh);
+		m_bvhMap.insert(arrays->m_bvhsDouble[i], bvh);
 	}
-	for (i=0;i<arrays->m_bvhsFloat.size();i++)
-    {
-        btOptimizedBvh* bvh = createOptimizedBvh();
-   		btQuantizedBvhFloatData* bvhData = arrays->m_bvhsFloat[i];
+	for (i = 0; i < arrays->m_bvhsFloat.size(); i++)
+	{
+		btOptimizedBvh* bvh = createOptimizedBvh();
+		btQuantizedBvhFloatData* bvhData = arrays->m_bvhsFloat[i];
 		bvh->deSerializeFloat(*bvhData);
-		m_bvhMap.insert(arrays->m_bvhsFloat[i],bvh);
+		m_bvhMap.insert(arrays->m_bvhsFloat[i], bvh);
 	}
 
-
-
-
-
-	for (i=0;i<arrays->m_colShapeData.size();i++)
+	for (i = 0; i < arrays->m_colShapeData.size(); i++)
 	{
 		btCollisionShapeData* shapeData = arrays->m_colShapeData[i];
 		btCollisionShape* shape = convertCollisionShape(shapeData);
 		if (shape)
 		{
-	//		printf("shapeMap.insert(%x,%x)\n",shapeData,shape);
-			m_shapeMap.insert(shapeData,shape);
+			//		printf("shapeMap.insert(%x,%x)\n",shapeData,shape);
+			m_shapeMap.insert(shapeData, shape);
 		}
 
-		if (shape&& shapeData->m_name)
+		if (shape && shapeData->m_name)
 		{
 			char* newname = duplicateName(shapeData->m_name);
-			m_objectNameMap.insert(shape,newname);
-			m_nameShapeMap.insert(newname,shape);
+			m_objectNameMap.insert(shape, newname);
+			m_nameShapeMap.insert(newname, shape);
 		}
 	}
 
-
-	for (i=0;i<arrays->m_collisionObjectDataDouble.size();i++)
+	for (i = 0; i < arrays->m_collisionObjectDataDouble.size(); i++)
 	{
-        btCollisionObjectDoubleData* colObjData = arrays->m_collisionObjectDataDouble[i];
-        btCollisionShape** shapePtr = m_shapeMap.find(colObjData->m_collisionShape);
-        if (shapePtr && *shapePtr)
-        {
-            btTransform startTransform;
-            colObjData->m_worldTransform.m_origin.m_floats[3] = 0.f;
-            startTransform.deSerializeDouble(colObjData->m_worldTransform);
+		btCollisionObjectDoubleData* colObjData = arrays->m_collisionObjectDataDouble[i];
+		btCollisionShape** shapePtr = m_shapeMap.find(colObjData->m_collisionShape);
+		if (shapePtr && *shapePtr)
+		{
+			btTransform startTransform;
+			colObjData->m_worldTransform.m_origin.m_floats[3] = 0.f;
+			startTransform.deSerializeDouble(colObjData->m_worldTransform);
 
-            btCollisionShape* shape = (btCollisionShape*)*shapePtr;
-            btCollisionObject* body = createCollisionObject(startTransform,shape,colObjData->m_name);
-            body->setFriction(btScalar(colObjData->m_friction));
-            body->setRestitution(btScalar(colObjData->m_restitution));
+			btCollisionShape* shape = (btCollisionShape*)*shapePtr;
+			btCollisionObject* body = createCollisionObject(startTransform, shape, colObjData->m_name);
+			body->setFriction(btScalar(colObjData->m_friction));
+			body->setRestitution(btScalar(colObjData->m_restitution));
 
 #ifdef USE_INTERNAL_EDGE_UTILITY
-            if (shape->getShapeType() == TRIANGLE_MESH_SHAPE_PROXYTYPE)
-            {
-                btBvhTriangleMeshShape* trimesh = (btBvhTriangleMeshShape*)shape;
-                if (trimesh->getTriangleInfoMap())
-                {
-                    body->setCollisionFlags(body->getCollisionFlags()  | btCollisionObject::CF_CUSTOM_MATERIAL_CALLBACK);
-                }
-            }
-#endif //USE_INTERNAL_EDGE_UTILITY
-            m_bodyMap.insert(colObjData,body);
-        } else
-        {
-            printf("error: no shape found\n");
-        }
+			if (shape->getShapeType() == TRIANGLE_MESH_SHAPE_PROXYTYPE)
+			{
+				btBvhTriangleMeshShape* trimesh = (btBvhTriangleMeshShape*)shape;
+				if (trimesh->getTriangleInfoMap())
+				{
+					body->setCollisionFlags(body->getCollisionFlags() | btCollisionObject::CF_CUSTOM_MATERIAL_CALLBACK);
+				}
+			}
+#endif  //USE_INTERNAL_EDGE_UTILITY
+			m_bodyMap.insert(colObjData, body);
+		}
+		else
+		{
+			printf("error: no shape found\n");
+		}
 	}
-	for (i=0;i<arrays->m_collisionObjectDataFloat.size();i++)
+	for (i = 0; i < arrays->m_collisionObjectDataFloat.size(); i++)
 	{
-        btCollisionObjectFloatData* colObjData = arrays->m_collisionObjectDataFloat[i];
-        btCollisionShape** shapePtr = m_shapeMap.find(colObjData->m_collisionShape);
-        if (shapePtr && *shapePtr)
-        {
-            btTransform startTransform;
-            colObjData->m_worldTransform.m_origin.m_floats[3] = 0.f;
-            startTransform.deSerializeFloat(colObjData->m_worldTransform);
+		btCollisionObjectFloatData* colObjData = arrays->m_collisionObjectDataFloat[i];
+		btCollisionShape** shapePtr = m_shapeMap.find(colObjData->m_collisionShape);
+		if (shapePtr && *shapePtr)
+		{
+			btTransform startTransform;
+			colObjData->m_worldTransform.m_origin.m_floats[3] = 0.f;
+			startTransform.deSerializeFloat(colObjData->m_worldTransform);
 
-            btCollisionShape* shape = (btCollisionShape*)*shapePtr;
-            btCollisionObject* body = createCollisionObject(startTransform,shape,colObjData->m_name);
+			btCollisionShape* shape = (btCollisionShape*)*shapePtr;
+			btCollisionObject* body = createCollisionObject(startTransform, shape, colObjData->m_name);
 
 #ifdef USE_INTERNAL_EDGE_UTILITY
-            if (shape->getShapeType() == TRIANGLE_MESH_SHAPE_PROXYTYPE)
-            {
-                btBvhTriangleMeshShape* trimesh = (btBvhTriangleMeshShape*)shape;
-                if (trimesh->getTriangleInfoMap())
-                {
-                    body->setCollisionFlags(body->getCollisionFlags()  | btCollisionObject::CF_CUSTOM_MATERIAL_CALLBACK);
-                }
-            }
-#endif //USE_INTERNAL_EDGE_UTILITY
-            m_bodyMap.insert(colObjData,body);
-        } else
-        {
-            printf("error: no shape found\n");
-        }
-    }
+			if (shape->getShapeType() == TRIANGLE_MESH_SHAPE_PROXYTYPE)
+			{
+				btBvhTriangleMeshShape* trimesh = (btBvhTriangleMeshShape*)shape;
+				if (trimesh->getTriangleInfoMap())
+				{
+					body->setCollisionFlags(body->getCollisionFlags() | btCollisionObject::CF_CUSTOM_MATERIAL_CALLBACK);
+				}
+			}
+#endif  //USE_INTERNAL_EDGE_UTILITY
+			m_bodyMap.insert(colObjData, body);
+		}
+		else
+		{
+			printf("error: no shape found\n");
+		}
+	}
 
 	return true;
 }
 
-
-
 void btCollisionWorldImporter::deleteAllData()
 {
 	int i;
 
-	for (i=0;i<m_allocatedCollisionObjects.size();i++)
+	for (i = 0; i < m_allocatedCollisionObjects.size(); i++)
 	{
-		if(m_collisionWorld)
+		if (m_collisionWorld)
 			m_collisionWorld->removeCollisionObject(m_allocatedCollisionObjects[i]);
 		delete m_allocatedCollisionObjects[i];
 	}
 
 	m_allocatedCollisionObjects.clear();
 
-
-	for (i=0;i<m_allocatedCollisionShapes.size();i++)
+	for (i = 0; i < m_allocatedCollisionShapes.size(); i++)
 	{
 		delete m_allocatedCollisionShapes[i];
 	}
 	m_allocatedCollisionShapes.clear();
 
-
-	for (i=0;i<m_allocatedBvhs.size();i++)
+	for (i = 0; i < m_allocatedBvhs.size(); i++)
 	{
 		delete m_allocatedBvhs[i];
 	}
 	m_allocatedBvhs.clear();
 
-	for (i=0;i<m_allocatedTriangleInfoMaps.size();i++)
+	for (i = 0; i < m_allocatedTriangleInfoMaps.size(); i++)
 	{
 		delete m_allocatedTriangleInfoMaps[i];
 	}
 	m_allocatedTriangleInfoMaps.clear();
-	for (i=0;i<m_allocatedTriangleIndexArrays.size();i++)
+	for (i = 0; i < m_allocatedTriangleIndexArrays.size(); i++)
 	{
 		delete m_allocatedTriangleIndexArrays[i];
 	}
 	m_allocatedTriangleIndexArrays.clear();
-	for (i=0;i<m_allocatedNames.size();i++)
+	for (i = 0; i < m_allocatedNames.size(); i++)
 	{
 		delete[] m_allocatedNames[i];
 	}
 	m_allocatedNames.clear();
 
-	for (i=0;i<m_allocatedbtStridingMeshInterfaceDatas.size();i++)
+	for (i = 0; i < m_allocatedbtStridingMeshInterfaceDatas.size(); i++)
 	{
 		btStridingMeshInterfaceData* curData = m_allocatedbtStridingMeshInterfaceDatas[i];
 
-		for(int a = 0;a < curData->m_numMeshParts;a++)
+		for (int a = 0; a < curData->m_numMeshParts; a++)
 		{
 			btMeshPartData* curPart = &curData->m_meshPartsPtr[a];
-			if(curPart->m_vertices3f)
-				delete [] curPart->m_vertices3f;
+			if (curPart->m_vertices3f)
+				delete[] curPart->m_vertices3f;
 
-			if(curPart->m_vertices3d)
-				delete [] curPart->m_vertices3d;
+			if (curPart->m_vertices3d)
+				delete[] curPart->m_vertices3d;
 
-			if(curPart->m_indices32)
-				delete [] curPart->m_indices32;
+			if (curPart->m_indices32)
+				delete[] curPart->m_indices32;
 
-			if(curPart->m_3indices16)
-				delete [] curPart->m_3indices16;
+			if (curPart->m_3indices16)
+				delete[] curPart->m_3indices16;
 
-			if(curPart->m_indices16)
-				delete [] curPart->m_indices16;
+			if (curPart->m_indices16)
+				delete[] curPart->m_indices16;
 
 			if (curPart->m_3indices8)
-				delete [] curPart->m_3indices8;
-
+				delete[] curPart->m_3indices8;
 		}
-		delete [] curData->m_meshPartsPtr;
+		delete[] curData->m_meshPartsPtr;
 		delete curData;
 	}
 	m_allocatedbtStridingMeshInterfaceDatas.clear();
 
-	for (i=0;i<m_indexArrays.size();i++)
+	for (i = 0; i < m_indexArrays.size(); i++)
 	{
 		btAlignedFree(m_indexArrays[i]);
 	}
-  m_indexArrays.clear();
+	m_indexArrays.clear();
 
-	for (i=0;i<m_shortIndexArrays.size();i++)
+	for (i = 0; i < m_shortIndexArrays.size(); i++)
 	{
 		btAlignedFree(m_shortIndexArrays[i]);
 	}
-  m_shortIndexArrays.clear();
+	m_shortIndexArrays.clear();
 
-	for (i=0;i<m_charIndexArrays.size();i++)
+	for (i = 0; i < m_charIndexArrays.size(); i++)
 	{
 		btAlignedFree(m_charIndexArrays[i]);
 	}
-  m_charIndexArrays.clear();
+	m_charIndexArrays.clear();
 
-	for (i=0;i<m_floatVertexArrays.size();i++)
+	for (i = 0; i < m_floatVertexArrays.size(); i++)
 	{
 		btAlignedFree(m_floatVertexArrays[i]);
 	}
-  m_floatVertexArrays.clear();
+	m_floatVertexArrays.clear();
 
-	for (i=0;i<m_doubleVertexArrays.size();i++)
+	for (i = 0; i < m_doubleVertexArrays.size(); i++)
 	{
 		btAlignedFree(m_doubleVertexArrays[i]);
 	}
-   m_doubleVertexArrays.clear();
-
-
+	m_doubleVertexArrays.clear();
 }
 
-
-
-btCollisionShape* btCollisionWorldImporter::convertCollisionShape(  btCollisionShapeData* shapeData  )
+btCollisionShape* btCollisionWorldImporter::convertCollisionShape(btCollisionShapeData* shapeData)
 {
 	btCollisionShape* shape = 0;
 
 	switch (shapeData->m_shapeType)
-		{
-	case STATIC_PLANE_PROXYTYPE:
+	{
+		case STATIC_PLANE_PROXYTYPE:
 		{
 			btStaticPlaneShapeData* planeData = (btStaticPlaneShapeData*)shapeData;
-			btVector3 planeNormal,localScaling;
+			btVector3 planeNormal, localScaling;
 			planeNormal.deSerializeFloat(planeData->m_planeNormal);
 			localScaling.deSerializeFloat(planeData->m_localScaling);
-			shape = createPlaneShape(planeNormal,planeData->m_planeConstant);
+			shape = createPlaneShape(planeNormal, planeData->m_planeConstant);
 			shape->setLocalScaling(localScaling);
 
 			break;
 		}
-	case SCALED_TRIANGLE_MESH_SHAPE_PROXYTYPE:
+		case SCALED_TRIANGLE_MESH_SHAPE_PROXYTYPE:
 		{
-			btScaledTriangleMeshShapeData* scaledMesh = (btScaledTriangleMeshShapeData*) shapeData;
-			btCollisionShapeData* colShapeData = (btCollisionShapeData*) &scaledMesh->m_trimeshShapeData;
+			btScaledTriangleMeshShapeData* scaledMesh = (btScaledTriangleMeshShapeData*)shapeData;
+			btCollisionShapeData* colShapeData = (btCollisionShapeData*)&scaledMesh->m_trimeshShapeData;
 			colShapeData->m_shapeType = TRIANGLE_MESH_SHAPE_PROXYTYPE;
 			btCollisionShape* childShape = convertCollisionShape(colShapeData);
 			btBvhTriangleMeshShape* meshShape = (btBvhTriangleMeshShape*)childShape;
@@ -288,15 +270,14 @@ btCollisionShape* btCollisionWorldImporter::convertCollisionShape(  btCollisionS
 			break;
 		}
 #ifdef SUPPORT_GIMPACT_SHAPE_IMPORT
-	case GIMPACT_SHAPE_PROXYTYPE:
+		case GIMPACT_SHAPE_PROXYTYPE:
 		{
-			btGImpactMeshShapeData* gimpactData = (btGImpactMeshShapeData*) shapeData;
+			btGImpactMeshShapeData* gimpactData = (btGImpactMeshShapeData*)shapeData;
 			if (gimpactData->m_gimpactSubType == CONST_GIMPACT_TRIMESH_SHAPE)
 			{
 				btStridingMeshInterfaceData* interfaceData = createStridingMeshInterfaceData(&gimpactData->m_meshInterface);
 				btTriangleIndexVertexArray* meshInterface = createMeshInterface(*interfaceData);
 
-
 				btGImpactMeshShape* gimpactShape = createGimpactShape(meshInterface);
 				btVector3 localScaling;
 				localScaling.deSerializeFloat(gimpactData->m_localScaling);
@@ -304,47 +285,45 @@ btCollisionShape* btCollisionWorldImporter::convertCollisionShape(  btCollisionS
 				gimpactShape->setMargin(btScalar(gimpactData->m_collisionMargin));
 				gimpactShape->updateBound();
 				shape = gimpactShape;
-			} else
+			}
+			else
 			{
 				printf("unsupported gimpact sub type\n");
 			}
 			break;
 		}
-#endif //SUPPORT_GIMPACT_SHAPE_IMPORT
-	//The btCapsuleShape* API has issue passing the margin/scaling/halfextents unmodified through the API
-	//so deal with this
+#endif  //SUPPORT_GIMPACT_SHAPE_IMPORT                                                                        \
+		//The btCapsuleShape* API has issue passing the margin/scaling/halfextents unmodified through the API \
+		//so deal with this
 		case CAPSULE_SHAPE_PROXYTYPE:
 		{
 			btCapsuleShapeData* capData = (btCapsuleShapeData*)shapeData;
 
-
 			switch (capData->m_upAxis)
 			{
-			case 0:
+				case 0:
 				{
-					shape = createCapsuleShapeX(1,1);
+					shape = createCapsuleShapeX(1, 1);
 					break;
 				}
-			case 1:
+				case 1:
 				{
-					shape = createCapsuleShapeY(1,1);
+					shape = createCapsuleShapeY(1, 1);
 					break;
 				}
-			case 2:
+				case 2:
 				{
-					shape = createCapsuleShapeZ(1,1);
+					shape = createCapsuleShapeZ(1, 1);
 					break;
 				}
-			default:
+				default:
 				{
 					printf("error: wrong up axis for btCapsuleShape\n");
 				}
-
-
 			};
 			if (shape)
 			{
-				btCapsuleShape* cap = (btCapsuleShape*) shape;
+				btCapsuleShape* cap = (btCapsuleShape*)shape;
 				cap->deSerializeFloat(capData);
 			}
 			break;
@@ -355,163 +334,156 @@ btCollisionShape* btCollisionWorldImporter::convertCollisionShape(  btCollisionS
 		case SPHERE_SHAPE_PROXYTYPE:
 		case MULTI_SPHERE_SHAPE_PROXYTYPE:
 		case CONVEX_HULL_SHAPE_PROXYTYPE:
+		{
+			btConvexInternalShapeData* bsd = (btConvexInternalShapeData*)shapeData;
+			btVector3 implicitShapeDimensions;
+			implicitShapeDimensions.deSerializeFloat(bsd->m_implicitShapeDimensions);
+			btVector3 localScaling;
+			localScaling.deSerializeFloat(bsd->m_localScaling);
+			btVector3 margin(bsd->m_collisionMargin, bsd->m_collisionMargin, bsd->m_collisionMargin);
+			switch (shapeData->m_shapeType)
 			{
-				btConvexInternalShapeData* bsd = (btConvexInternalShapeData*)shapeData;
-				btVector3 implicitShapeDimensions;
-				implicitShapeDimensions.deSerializeFloat(bsd->m_implicitShapeDimensions);
+				case BOX_SHAPE_PROXYTYPE:
+				{
+					btBoxShape* box = (btBoxShape*)createBoxShape(implicitShapeDimensions / localScaling + margin);
+					//box->initializePolyhedralFeatures();
+					shape = box;
+
+					break;
+				}
+				case SPHERE_SHAPE_PROXYTYPE:
+				{
+					shape = createSphereShape(implicitShapeDimensions.getX());
+					break;
+				}
+
+				case CYLINDER_SHAPE_PROXYTYPE:
+				{
+					btCylinderShapeData* cylData = (btCylinderShapeData*)shapeData;
+					btVector3 halfExtents = implicitShapeDimensions + margin;
+					switch (cylData->m_upAxis)
+					{
+						case 0:
+						{
+							shape = createCylinderShapeX(halfExtents.getY(), halfExtents.getX());
+							break;
+						}
+						case 1:
+						{
+							shape = createCylinderShapeY(halfExtents.getX(), halfExtents.getY());
+							break;
+						}
+						case 2:
+						{
+							shape = createCylinderShapeZ(halfExtents.getX(), halfExtents.getZ());
+							break;
+						}
+						default:
+						{
+							printf("unknown Cylinder up axis\n");
+						}
+					};
+
+					break;
+				}
+				case CONE_SHAPE_PROXYTYPE:
+				{
+					btConeShapeData* conData = (btConeShapeData*)shapeData;
+					btVector3 halfExtents = implicitShapeDimensions;  //+margin;
+					switch (conData->m_upIndex)
+					{
+						case 0:
+						{
+							shape = createConeShapeX(halfExtents.getY(), halfExtents.getX());
+							break;
+						}
+						case 1:
+						{
+							shape = createConeShapeY(halfExtents.getX(), halfExtents.getY());
+							break;
+						}
+						case 2:
+						{
+							shape = createConeShapeZ(halfExtents.getX(), halfExtents.getZ());
+							break;
+						}
+						default:
+						{
+							printf("unknown Cone up axis\n");
+						}
+					};
+
+					break;
+				}
+				case MULTI_SPHERE_SHAPE_PROXYTYPE:
+				{
+					btMultiSphereShapeData* mss = (btMultiSphereShapeData*)bsd;
+					int numSpheres = mss->m_localPositionArraySize;
+
+					btAlignedObjectArray<btVector3> tmpPos;
+					btAlignedObjectArray<btScalar> radii;
+					radii.resize(numSpheres);
+					tmpPos.resize(numSpheres);
+					int i;
+					for (i = 0; i < numSpheres; i++)
+					{
+						tmpPos[i].deSerializeFloat(mss->m_localPositionArrayPtr[i].m_pos);
+						radii[i] = mss->m_localPositionArrayPtr[i].m_radius;
+					}
+					shape = createMultiSphereShape(&tmpPos[0], &radii[0], numSpheres);
+					break;
+				}
+				case CONVEX_HULL_SHAPE_PROXYTYPE:
+				{
+					//	int sz = sizeof(btConvexHullShapeData);
+					//	int sz2 = sizeof(btConvexInternalShapeData);
+					//	int sz3 = sizeof(btCollisionShapeData);
+					btConvexHullShapeData* convexData = (btConvexHullShapeData*)bsd;
+					int numPoints = convexData->m_numUnscaledPoints;
+
+					btAlignedObjectArray<btVector3> tmpPoints;
+					tmpPoints.resize(numPoints);
+					int i;
+					for (i = 0; i < numPoints; i++)
+					{
+#ifdef BT_USE_DOUBLE_PRECISION
+						if (convexData->m_unscaledPointsDoublePtr)
+							tmpPoints[i].deSerialize(convexData->m_unscaledPointsDoublePtr[i]);
+						if (convexData->m_unscaledPointsFloatPtr)
+							tmpPoints[i].deSerializeFloat(convexData->m_unscaledPointsFloatPtr[i]);
+#else
+						if (convexData->m_unscaledPointsFloatPtr)
+							tmpPoints[i].deSerialize(convexData->m_unscaledPointsFloatPtr[i]);
+						if (convexData->m_unscaledPointsDoublePtr)
+							tmpPoints[i].deSerializeDouble(convexData->m_unscaledPointsDoublePtr[i]);
+#endif  //BT_USE_DOUBLE_PRECISION
+					}
+					btConvexHullShape* hullShape = createConvexHullShape();
+					for (i = 0; i < numPoints; i++)
+					{
+						hullShape->addPoint(tmpPoints[i]);
+					}
+					hullShape->setMargin(bsd->m_collisionMargin);
+					//hullShape->initializePolyhedralFeatures();
+					shape = hullShape;
+					break;
+				}
+				default:
+				{
+					printf("error: cannot create shape type (%d)\n", shapeData->m_shapeType);
+				}
+			}
+
+			if (shape)
+			{
+				shape->setMargin(bsd->m_collisionMargin);
+
 				btVector3 localScaling;
 				localScaling.deSerializeFloat(bsd->m_localScaling);
-				btVector3 margin(bsd->m_collisionMargin,bsd->m_collisionMargin,bsd->m_collisionMargin);
-				switch (shapeData->m_shapeType)
-				{
-					case BOX_SHAPE_PROXYTYPE:
-						{
-							btBoxShape* box= (btBoxShape*)createBoxShape(implicitShapeDimensions/localScaling+margin);
-							//box->initializePolyhedralFeatures();
-							shape = box;
-
-							break;
-						}
-					case SPHERE_SHAPE_PROXYTYPE:
-						{
-							shape = createSphereShape(implicitShapeDimensions.getX());
-							break;
-						}
-
-					case CYLINDER_SHAPE_PROXYTYPE:
-						{
-							btCylinderShapeData* cylData = (btCylinderShapeData*) shapeData;
-							btVector3 halfExtents = implicitShapeDimensions+margin;
-							switch (cylData->m_upAxis)
-							{
-							case 0:
-								{
-									shape = createCylinderShapeX(halfExtents.getY(),halfExtents.getX());
-									break;
-								}
-							case 1:
-								{
-									shape = createCylinderShapeY(halfExtents.getX(),halfExtents.getY());
-									break;
-								}
-							case 2:
-								{
-									shape = createCylinderShapeZ(halfExtents.getX(),halfExtents.getZ());
-									break;
-								}
-							default:
-								{
-									printf("unknown Cylinder up axis\n");
-								}
-
-							};
-
-
-
-							break;
-						}
-					case CONE_SHAPE_PROXYTYPE:
-						{
-							btConeShapeData* conData = (btConeShapeData*) shapeData;
-							btVector3 halfExtents = implicitShapeDimensions;//+margin;
-							switch (conData->m_upIndex)
-							{
-							case 0:
-								{
-									shape = createConeShapeX(halfExtents.getY(),halfExtents.getX());
-									break;
-								}
-							case 1:
-								{
-									shape = createConeShapeY(halfExtents.getX(),halfExtents.getY());
-									break;
-								}
-							case 2:
-								{
-									shape = createConeShapeZ(halfExtents.getX(),halfExtents.getZ());
-									break;
-								}
-							default:
-								{
-									printf("unknown Cone up axis\n");
-								}
-
-							};
-
-
-
-							break;
-						}
-					case MULTI_SPHERE_SHAPE_PROXYTYPE:
-						{
-							btMultiSphereShapeData* mss = (btMultiSphereShapeData*)bsd;
-							int numSpheres = mss->m_localPositionArraySize;
-
-							btAlignedObjectArray<btVector3> tmpPos;
-							btAlignedObjectArray<btScalar> radii;
-							radii.resize(numSpheres);
-							tmpPos.resize(numSpheres);
-							int i;
-							for ( i=0;i<numSpheres;i++)
-							{
-								tmpPos[i].deSerializeFloat(mss->m_localPositionArrayPtr[i].m_pos);
-								radii[i] = mss->m_localPositionArrayPtr[i].m_radius;
-							}
-							shape = createMultiSphereShape(&tmpPos[0],&radii[0],numSpheres);
-							break;
-						}
-					case CONVEX_HULL_SHAPE_PROXYTYPE:
-						{
-						//	int sz = sizeof(btConvexHullShapeData);
-						//	int sz2 = sizeof(btConvexInternalShapeData);
-						//	int sz3 = sizeof(btCollisionShapeData);
-							btConvexHullShapeData* convexData = (btConvexHullShapeData*)bsd;
-							int numPoints = convexData->m_numUnscaledPoints;
-
-							btAlignedObjectArray<btVector3> tmpPoints;
-							tmpPoints.resize(numPoints);
-							int i;
-							for ( i=0;i<numPoints;i++)
-							{
-#ifdef BT_USE_DOUBLE_PRECISION
-							if (convexData->m_unscaledPointsDoublePtr)
-								tmpPoints[i].deSerialize(convexData->m_unscaledPointsDoublePtr[i]);
-							if (convexData->m_unscaledPointsFloatPtr)
-								tmpPoints[i].deSerializeFloat(convexData->m_unscaledPointsFloatPtr[i]);
-#else
-							if (convexData->m_unscaledPointsFloatPtr)
-								tmpPoints[i].deSerialize(convexData->m_unscaledPointsFloatPtr[i]);
-							if (convexData->m_unscaledPointsDoublePtr)
-								tmpPoints[i].deSerializeDouble(convexData->m_unscaledPointsDoublePtr[i]);
-#endif //BT_USE_DOUBLE_PRECISION
-							}
-							btConvexHullShape* hullShape = createConvexHullShape();
-							for (i=0;i<numPoints;i++)
-							{
-								hullShape->addPoint(tmpPoints[i]);
-							}
-							hullShape->setMargin(bsd->m_collisionMargin);
-							//hullShape->initializePolyhedralFeatures();
-							shape = hullShape;
-							break;
-						}
-					default:
-						{
-							printf("error: cannot create shape type (%d)\n",shapeData->m_shapeType);
-						}
-				}
-
-				if (shape)
-				{
-					shape->setMargin(bsd->m_collisionMargin);
-
-					btVector3 localScaling;
-					localScaling.deSerializeFloat(bsd->m_localScaling);
-					shape->setLocalScaling(localScaling);
-
-				}
-				break;
+				shape->setLocalScaling(localScaling);
 			}
+			break;
+		}
 		case TRIANGLE_MESH_SHAPE_PROXYTYPE:
 		{
 			btTriangleMeshShapeData* trimesh = (btTriangleMeshShapeData*)shapeData;
@@ -522,10 +494,10 @@ btCollisionShape* btCollisionWorldImporter::convertCollisionShape(  btCollisionS
 				return 0;
 			}
 
-			btVector3 scaling; scaling.deSerializeFloat(trimesh->m_meshInterface.m_scaling);
+			btVector3 scaling;
+			scaling.deSerializeFloat(trimesh->m_meshInterface.m_scaling);
 			meshInterface->setScaling(scaling);
 
-
 			btOptimizedBvh* bvh = 0;
 #if 1
 			if (trimesh->m_quantizedFloatBvh)
@@ -534,7 +506,8 @@ btCollisionShape* btCollisionWorldImporter::convertCollisionShape(  btCollisionS
 				if (bvhPtr && *bvhPtr)
 				{
 					bvh = *bvhPtr;
-				} else
+				}
+				else
 				{
 					bvh = createOptimizedBvh();
 					bvh->deSerializeFloat(*trimesh->m_quantizedFloatBvh);
@@ -546,7 +519,8 @@ btCollisionShape* btCollisionWorldImporter::convertCollisionShape(  btCollisionS
 				if (bvhPtr && *bvhPtr)
 				{
 					bvh = *bvhPtr;
-				} else
+				}
+				else
 				{
 					bvh = createOptimizedBvh();
 					bvh->deSerializeDouble(*trimesh->m_quantizedDoubleBvh);
@@ -554,8 +528,7 @@ btCollisionShape* btCollisionWorldImporter::convertCollisionShape(  btCollisionS
 			}
 #endif
 
-
-			btBvhTriangleMeshShape* trimeshShape = createBvhTriangleMeshShape(meshInterface,bvh);
+			btBvhTriangleMeshShape* trimeshShape = createBvhTriangleMeshShape(meshInterface, bvh);
 			trimeshShape->setMargin(trimesh->m_collisionMargin);
 			shape = trimeshShape;
 
@@ -567,71 +540,66 @@ btCollisionShape* btCollisionWorldImporter::convertCollisionShape(  btCollisionS
 
 #ifdef USE_INTERNAL_EDGE_UTILITY
 				gContactAddedCallback = btAdjustInternalEdgeContactsCallback;
-#endif //USE_INTERNAL_EDGE_UTILITY
-
+#endif  //USE_INTERNAL_EDGE_UTILITY
 			}
 
 			//printf("trimesh->m_collisionMargin=%f\n",trimesh->m_collisionMargin);
 			break;
 		}
 		case COMPOUND_SHAPE_PROXYTYPE:
+		{
+			btCompoundShapeData* compoundData = (btCompoundShapeData*)shapeData;
+			btCompoundShape* compoundShape = createCompoundShape();
+
+			//btCompoundShapeChildData* childShapeDataArray = &compoundData->m_childShapePtr[0];
+
+			btAlignedObjectArray<btCollisionShape*> childShapes;
+			for (int i = 0; i < compoundData->m_numChildShapes; i++)
 			{
-				btCompoundShapeData* compoundData = (btCompoundShapeData*)shapeData;
-				btCompoundShape* compoundShape = createCompoundShape();
+				//btCompoundShapeChildData* ptr = &compoundData->m_childShapePtr[i];
 
-				//btCompoundShapeChildData* childShapeDataArray = &compoundData->m_childShapePtr[0];
+				btCollisionShapeData* cd = compoundData->m_childShapePtr[i].m_childShape;
 
-
-				btAlignedObjectArray<btCollisionShape*> childShapes;
-				for (int i=0;i<compoundData->m_numChildShapes;i++)
+				btCollisionShape* childShape = convertCollisionShape(cd);
+				if (childShape)
 				{
-					//btCompoundShapeChildData* ptr = &compoundData->m_childShapePtr[i];
-
-					btCollisionShapeData* cd = compoundData->m_childShapePtr[i].m_childShape;
-
-					btCollisionShape* childShape = convertCollisionShape(cd);
-					if (childShape)
-					{
-						btTransform localTransform;
-						localTransform.deSerializeFloat(compoundData->m_childShapePtr[i].m_transform);
-						compoundShape->addChildShape(localTransform,childShape);
-					} else
-					{
-#ifdef _DEBUG
-						printf("error: couldn't create childShape for compoundShape\n");
-#endif
-					}
-
+					btTransform localTransform;
+					localTransform.deSerializeFloat(compoundData->m_childShapePtr[i].m_transform);
+					compoundShape->addChildShape(localTransform, childShape);
 				}
-				shape = compoundShape;
-
-				break;
-			}
-		case SOFTBODY_SHAPE_PROXYTYPE:
-			{
-				return 0;
-			}
-		default:
-			{
+				else
+				{
 #ifdef _DEBUG
-				printf("unsupported shape type (%d)\n",shapeData->m_shapeType);
+					printf("error: couldn't create childShape for compoundShape\n");
 #endif
+				}
 			}
+			shape = compoundShape;
+
+			break;
 		}
+		case SOFTBODY_SHAPE_PROXYTYPE:
+		{
+			return 0;
+		}
+		default:
+		{
+#ifdef _DEBUG
+			printf("unsupported shape type (%d)\n", shapeData->m_shapeType);
+#endif
+		}
+	}
 
-		return shape;
-
+	return shape;
 }
 
-
-
 char* btCollisionWorldImporter::duplicateName(const char* name)
 {
 	if (name)
 	{
 		int l = (int)strlen(name);
-		char* newName = new char[l+1];
-		memcpy(newName,name,l);
+		char* newName = new char[l + 1];
+		memcpy(newName, name, l);
 		newName[l] = 0;
 		m_allocatedNames.push_back(newName);
 		return newName;
@@ -639,53 +607,43 @@ char* btCollisionWorldImporter::duplicateName(const char* name)
 	return 0;
 }
 
-
-
-
-
-
-
-
-
-
-
-btTriangleIndexVertexArray* btCollisionWorldImporter::createMeshInterface(btStridingMeshInterfaceData&  meshData)
+btTriangleIndexVertexArray* btCollisionWorldImporter::createMeshInterface(btStridingMeshInterfaceData& meshData)
 {
 	btTriangleIndexVertexArray* meshInterface = createTriangleMeshContainer();
 
-	for (int i=0;i<meshData.m_numMeshParts;i++)
+	for (int i = 0; i < meshData.m_numMeshParts; i++)
 	{
 		btIndexedMesh meshPart;
 		meshPart.m_numTriangles = meshData.m_meshPartsPtr[i].m_numTriangles;
 		meshPart.m_numVertices = meshData.m_meshPartsPtr[i].m_numVertices;
 
-
 		if (meshData.m_meshPartsPtr[i].m_indices32)
 		{
 			meshPart.m_indexType = PHY_INTEGER;
-			meshPart.m_triangleIndexStride = 3*sizeof(int);
-			int* indexArray = (int*)btAlignedAlloc(sizeof(int)*3*meshPart.m_numTriangles,16);
+			meshPart.m_triangleIndexStride = 3 * sizeof(int);
+			int* indexArray = (int*)btAlignedAlloc(sizeof(int) * 3 * meshPart.m_numTriangles, 16);
 			m_indexArrays.push_back(indexArray);
-			for (int j=0;j<3*meshPart.m_numTriangles;j++)
+			for (int j = 0; j < 3 * meshPart.m_numTriangles; j++)
 			{
 				indexArray[j] = meshData.m_meshPartsPtr[i].m_indices32[j].m_value;
 			}
 			meshPart.m_triangleIndexBase = (const unsigned char*)indexArray;
-		} else
+		}
+		else
 		{
 			if (meshData.m_meshPartsPtr[i].m_3indices16)
 			{
 				meshPart.m_indexType = PHY_SHORT;
-				meshPart.m_triangleIndexStride = sizeof(short int)*3;//sizeof(btShortIntIndexTripletData);
+				meshPart.m_triangleIndexStride = sizeof(short int) * 3;  //sizeof(btShortIntIndexTripletData);
 
-				short int* indexArray = (short int*)btAlignedAlloc(sizeof(short int)*3*meshPart.m_numTriangles,16);
+				short int* indexArray = (short int*)btAlignedAlloc(sizeof(short int) * 3 * meshPart.m_numTriangles, 16);
 				m_shortIndexArrays.push_back(indexArray);
 
-				for (int j=0;j<meshPart.m_numTriangles;j++)
+				for (int j = 0; j < meshPart.m_numTriangles; j++)
 				{
-					indexArray[3*j] = meshData.m_meshPartsPtr[i].m_3indices16[j].m_values[0];
-					indexArray[3*j+1] = meshData.m_meshPartsPtr[i].m_3indices16[j].m_values[1];
-					indexArray[3*j+2] = meshData.m_meshPartsPtr[i].m_3indices16[j].m_values[2];
+					indexArray[3 * j] = meshData.m_meshPartsPtr[i].m_3indices16[j].m_values[0];
+					indexArray[3 * j + 1] = meshData.m_meshPartsPtr[i].m_3indices16[j].m_values[1];
+					indexArray[3 * j + 2] = meshData.m_meshPartsPtr[i].m_3indices16[j].m_values[2];
 				}
 
 				meshPart.m_triangleIndexBase = (const unsigned char*)indexArray;
@@ -693,10 +651,10 @@ btTriangleIndexVertexArray* btCollisionWorldImporter::createMeshInterface(btStri
 			if (meshData.m_meshPartsPtr[i].m_indices16)
 			{
 				meshPart.m_indexType = PHY_SHORT;
-				meshPart.m_triangleIndexStride = 3*sizeof(short int);
-				short int* indexArray = (short int*)btAlignedAlloc(sizeof(short int)*3*meshPart.m_numTriangles,16);
+				meshPart.m_triangleIndexStride = 3 * sizeof(short int);
+				short int* indexArray = (short int*)btAlignedAlloc(sizeof(short int) * 3 * meshPart.m_numTriangles, 16);
 				m_shortIndexArrays.push_back(indexArray);
-				for (int j=0;j<3*meshPart.m_numTriangles;j++)
+				for (int j = 0; j < 3 * meshPart.m_numTriangles; j++)
 				{
 					indexArray[j] = meshData.m_meshPartsPtr[i].m_indices16[j].m_value;
 				}
@@ -707,16 +665,16 @@ btTriangleIndexVertexArray* btCollisionWorldImporter::createMeshInterface(btStri
 			if (meshData.m_meshPartsPtr[i].m_3indices8)
 			{
 				meshPart.m_indexType = PHY_UCHAR;
-				meshPart.m_triangleIndexStride = sizeof(unsigned char)*3;
+				meshPart.m_triangleIndexStride = sizeof(unsigned char) * 3;
 
-				unsigned char* indexArray = (unsigned char*)btAlignedAlloc(sizeof(unsigned char)*3*meshPart.m_numTriangles,16);
+				unsigned char* indexArray = (unsigned char*)btAlignedAlloc(sizeof(unsigned char) * 3 * meshPart.m_numTriangles, 16);
 				m_charIndexArrays.push_back(indexArray);
 
-				for (int j=0;j<meshPart.m_numTriangles;j++)
+				for (int j = 0; j < meshPart.m_numTriangles; j++)
 				{
-					indexArray[3*j] = meshData.m_meshPartsPtr[i].m_3indices8[j].m_values[0];
-					indexArray[3*j+1] = meshData.m_meshPartsPtr[i].m_3indices8[j].m_values[1];
-					indexArray[3*j+2] = meshData.m_meshPartsPtr[i].m_3indices8[j].m_values[2];
+					indexArray[3 * j] = meshData.m_meshPartsPtr[i].m_3indices8[j].m_values[0];
+					indexArray[3 * j + 1] = meshData.m_meshPartsPtr[i].m_3indices8[j].m_values[1];
+					indexArray[3 * j + 2] = meshData.m_meshPartsPtr[i].m_3indices8[j].m_values[2];
 				}
 
 				meshPart.m_triangleIndexBase = (const unsigned char*)indexArray;
@@ -727,10 +685,10 @@ btTriangleIndexVertexArray* btCollisionWorldImporter::createMeshInterface(btStri
 		{
 			meshPart.m_vertexType = PHY_FLOAT;
 			meshPart.m_vertexStride = sizeof(btVector3FloatData);
-			btVector3FloatData* vertices = (btVector3FloatData*) btAlignedAlloc(sizeof(btVector3FloatData)*meshPart.m_numVertices,16);
+			btVector3FloatData* vertices = (btVector3FloatData*)btAlignedAlloc(sizeof(btVector3FloatData) * meshPart.m_numVertices, 16);
 			m_floatVertexArrays.push_back(vertices);
 
-			for (int j=0;j<meshPart.m_numVertices;j++)
+			for (int j = 0; j < meshPart.m_numVertices; j++)
 			{
 				vertices[j].m_floats[0] = meshData.m_meshPartsPtr[i].m_vertices3f[j].m_floats[0];
 				vertices[j].m_floats[1] = meshData.m_meshPartsPtr[i].m_vertices3f[j].m_floats[1];
@@ -738,16 +696,16 @@ btTriangleIndexVertexArray* btCollisionWorldImporter::createMeshInterface(btStri
 				vertices[j].m_floats[3] = meshData.m_meshPartsPtr[i].m_vertices3f[j].m_floats[3];
 			}
 			meshPart.m_vertexBase = (const unsigned char*)vertices;
-		} else
+		}
+		else
 		{
 			meshPart.m_vertexType = PHY_DOUBLE;
 			meshPart.m_vertexStride = sizeof(btVector3DoubleData);
 
-
-			btVector3DoubleData* vertices = (btVector3DoubleData*) btAlignedAlloc(sizeof(btVector3DoubleData)*meshPart.m_numVertices,16);
+			btVector3DoubleData* vertices = (btVector3DoubleData*)btAlignedAlloc(sizeof(btVector3DoubleData) * meshPart.m_numVertices, 16);
 			m_doubleVertexArrays.push_back(vertices);
 
-			for (int j=0;j<meshPart.m_numVertices;j++)
+			for (int j = 0; j < meshPart.m_numVertices; j++)
 			{
 				vertices[j].m_floats[0] = meshData.m_meshPartsPtr[i].m_vertices3d[j].m_floats[0];
 				vertices[j].m_floats[1] = meshData.m_meshPartsPtr[i].m_vertices3d[j].m_floats[1];
@@ -759,14 +717,13 @@ btTriangleIndexVertexArray* btCollisionWorldImporter::createMeshInterface(btStri
 
 		if (meshPart.m_triangleIndexBase && meshPart.m_vertexBase)
 		{
-			meshInterface->addIndexedMesh(meshPart,meshPart.m_indexType);
+			meshInterface->addIndexedMesh(meshPart, meshPart.m_indexType);
 		}
 	}
 
 	return meshInterface;
 }
 
-
 btStridingMeshInterfaceData* btCollisionWorldImporter::createStridingMeshInterfaceData(btStridingMeshInterfaceData* interfaceData)
 {
 	//create a new btStridingMeshInterfaceData that is an exact copy of shapedata and store it in the WorldImporter
@@ -776,7 +733,7 @@ btStridingMeshInterfaceData* btCollisionWorldImporter::createStridingMeshInterfa
 	newData->m_numMeshParts = interfaceData->m_numMeshParts;
 	newData->m_meshPartsPtr = new btMeshPartData[newData->m_numMeshParts];
 
-	for(int i = 0;i < newData->m_numMeshParts;i++)
+	for (int i = 0; i < newData->m_numMeshParts; i++)
 	{
 		btMeshPartData* curPart = &interfaceData->m_meshPartsPtr[i];
 		btMeshPartData* curNewPart = &newData->m_meshPartsPtr[i];
@@ -784,18 +741,18 @@ btStridingMeshInterfaceData* btCollisionWorldImporter::createStridingMeshInterfa
 		curNewPart->m_numTriangles = curPart->m_numTriangles;
 		curNewPart->m_numVertices = curPart->m_numVertices;
 
-		if(curPart->m_vertices3f)
+		if (curPart->m_vertices3f)
 		{
 			curNewPart->m_vertices3f = new btVector3FloatData[curNewPart->m_numVertices];
-			memcpy(curNewPart->m_vertices3f,curPart->m_vertices3f,sizeof(btVector3FloatData) * curNewPart->m_numVertices);
+			memcpy(curNewPart->m_vertices3f, curPart->m_vertices3f, sizeof(btVector3FloatData) * curNewPart->m_numVertices);
 		}
 		else
 			curNewPart->m_vertices3f = NULL;
 
-		if(curPart->m_vertices3d)
+		if (curPart->m_vertices3d)
 		{
 			curNewPart->m_vertices3d = new btVector3DoubleData[curNewPart->m_numVertices];
-			memcpy(curNewPart->m_vertices3d,curPart->m_vertices3d,sizeof(btVector3DoubleData) * curNewPart->m_numVertices);
+			memcpy(curNewPart->m_vertices3d, curPart->m_vertices3d, sizeof(btVector3DoubleData) * curNewPart->m_numVertices);
 		}
 		else
 			curNewPart->m_vertices3d = NULL;
@@ -803,63 +760,60 @@ btStridingMeshInterfaceData* btCollisionWorldImporter::createStridingMeshInterfa
 		int numIndices = curNewPart->m_numTriangles * 3;
 		///the m_3indices8 was not initialized in some Bullet versions, this can cause crashes at loading time
 		///we catch it by only dealing with m_3indices8 if none of the other indices are initialized
-		bool uninitialized3indices8Workaround =false;
+		bool uninitialized3indices8Workaround = false;
 
-		if(curPart->m_indices32)
+		if (curPart->m_indices32)
 		{
-			uninitialized3indices8Workaround=true;
+			uninitialized3indices8Workaround = true;
 			curNewPart->m_indices32 = new btIntIndexData[numIndices];
-			memcpy(curNewPart->m_indices32,curPart->m_indices32,sizeof(btIntIndexData) * numIndices);
+			memcpy(curNewPart->m_indices32, curPart->m_indices32, sizeof(btIntIndexData) * numIndices);
 		}
 		else
 			curNewPart->m_indices32 = NULL;
 
-		if(curPart->m_3indices16)
+		if (curPart->m_3indices16)
 		{
-			uninitialized3indices8Workaround=true;
+			uninitialized3indices8Workaround = true;
 			curNewPart->m_3indices16 = new btShortIntIndexTripletData[curNewPart->m_numTriangles];
-			memcpy(curNewPart->m_3indices16,curPart->m_3indices16,sizeof(btShortIntIndexTripletData) * curNewPart->m_numTriangles);
+			memcpy(curNewPart->m_3indices16, curPart->m_3indices16, sizeof(btShortIntIndexTripletData) * curNewPart->m_numTriangles);
 		}
 		else
 			curNewPart->m_3indices16 = NULL;
 
-		if(curPart->m_indices16)
+		if (curPart->m_indices16)
 		{
-			uninitialized3indices8Workaround=true;
+			uninitialized3indices8Workaround = true;
 			curNewPart->m_indices16 = new btShortIntIndexData[numIndices];
-			memcpy(curNewPart->m_indices16,curPart->m_indices16,sizeof(btShortIntIndexData) * numIndices);
+			memcpy(curNewPart->m_indices16, curPart->m_indices16, sizeof(btShortIntIndexData) * numIndices);
 		}
 		else
 			curNewPart->m_indices16 = NULL;
 
-		if(!uninitialized3indices8Workaround && curPart->m_3indices8)
+		if (!uninitialized3indices8Workaround && curPart->m_3indices8)
 		{
 			curNewPart->m_3indices8 = new btCharIndexTripletData[curNewPart->m_numTriangles];
-			memcpy(curNewPart->m_3indices8,curPart->m_3indices8,sizeof(btCharIndexTripletData) * curNewPart->m_numTriangles);
+			memcpy(curNewPart->m_3indices8, curPart->m_3indices8, sizeof(btCharIndexTripletData) * curNewPart->m_numTriangles);
 		}
 		else
 			curNewPart->m_3indices8 = NULL;
-
 	}
 
 	m_allocatedbtStridingMeshInterfaceDatas.push_back(newData);
 
-	return(newData);
+	return (newData);
 }
 
 #ifdef USE_INTERNAL_EDGE_UTILITY
-extern ContactAddedCallback		gContactAddedCallback;
+extern ContactAddedCallback gContactAddedCallback;
 
-static bool btAdjustInternalEdgeContactsCallback(btManifoldPoint& cp,	const btCollisionObject* colObj0,int partId0,int index0,const btCollisionObject* colObj1,int partId1,int index1)
+static bool btAdjustInternalEdgeContactsCallback(btManifoldPoint& cp, const btCollisionObject* colObj0, int partId0, int index0, const btCollisionObject* colObj1, int partId1, int index1)
 {
-
-	btAdjustInternalEdgeContacts(cp,colObj1,colObj0, partId1,index1);
-		//btAdjustInternalEdgeContacts(cp,colObj1,colObj0, partId1,index1, BT_TRIANGLE_CONVEX_BACKFACE_MODE);
-		//btAdjustInternalEdgeContacts(cp,colObj1,colObj0, partId1,index1, BT_TRIANGLE_CONVEX_DOUBLE_SIDED+BT_TRIANGLE_CONCAVE_DOUBLE_SIDED);
+	btAdjustInternalEdgeContacts(cp, colObj1, colObj0, partId1, index1);
+	//btAdjustInternalEdgeContacts(cp,colObj1,colObj0, partId1,index1, BT_TRIANGLE_CONVEX_BACKFACE_MODE);
+	//btAdjustInternalEdgeContacts(cp,colObj1,colObj0, partId1,index1, BT_TRIANGLE_CONVEX_DOUBLE_SIDED+BT_TRIANGLE_CONCAVE_DOUBLE_SIDED);
 	return true;
 }
-#endif //USE_INTERNAL_EDGE_UTILITY
-
+#endif  //USE_INTERNAL_EDGE_UTILITY
 
 /*
 btRigidBody*  btWorldImporter::createRigidBody(bool isDynamic, btScalar mass, const btTransform& startTransform,btCollisionShape* shape,const char* bodyName)
@@ -898,29 +852,27 @@ btCollisionObject* btCollisionWorldImporter::getCollisionObjectByName(const char
 	return 0;
 }
 
-btCollisionObject* btCollisionWorldImporter::createCollisionObject(const btTransform& startTransform,btCollisionShape* shape, const char* bodyName)
+btCollisionObject* btCollisionWorldImporter::createCollisionObject(const btTransform& startTransform, btCollisionShape* shape, const char* bodyName)
 {
 	btCollisionObject* colObj = new btCollisionObject();
 	colObj->setWorldTransform(startTransform);
 	colObj->setCollisionShape(shape);
-	m_collisionWorld->addCollisionObject(colObj);//todo: flags etc
+	m_collisionWorld->addCollisionObject(colObj);  //todo: flags etc
 
 	if (bodyName)
 	{
 		char* newname = duplicateName(bodyName);
-		m_objectNameMap.insert(colObj,newname);
-		m_nameColObjMap.insert(newname,colObj);
+		m_objectNameMap.insert(colObj, newname);
+		m_nameColObjMap.insert(newname, colObj);
 	}
 	m_allocatedCollisionObjects.push_back(colObj);
 
 	return colObj;
 }
 
-
-
-btCollisionShape* btCollisionWorldImporter::createPlaneShape(const btVector3& planeNormal,btScalar planeConstant)
+btCollisionShape* btCollisionWorldImporter::createPlaneShape(const btVector3& planeNormal, btScalar planeConstant)
 {
-	btStaticPlaneShape* shape = new btStaticPlaneShape(planeNormal,planeConstant);
+	btStaticPlaneShape* shape = new btStaticPlaneShape(planeNormal, planeConstant);
 	m_allocatedCollisionShapes.push_back(shape);
 	return shape;
 }
@@ -937,85 +889,83 @@ btCollisionShape* btCollisionWorldImporter::createSphereShape(btScalar radius)
 	return shape;
 }
 
-
 btCollisionShape* btCollisionWorldImporter::createCapsuleShapeX(btScalar radius, btScalar height)
 {
-	btCapsuleShapeX* shape = new btCapsuleShapeX(radius,height);
+	btCapsuleShapeX* shape = new btCapsuleShapeX(radius, height);
 	m_allocatedCollisionShapes.push_back(shape);
 	return shape;
 }
 
 btCollisionShape* btCollisionWorldImporter::createCapsuleShapeY(btScalar radius, btScalar height)
 {
-	btCapsuleShape* shape = new btCapsuleShape(radius,height);
+	btCapsuleShape* shape = new btCapsuleShape(radius, height);
 	m_allocatedCollisionShapes.push_back(shape);
 	return shape;
 }
 
 btCollisionShape* btCollisionWorldImporter::createCapsuleShapeZ(btScalar radius, btScalar height)
 {
-	btCapsuleShapeZ* shape = new btCapsuleShapeZ(radius,height);
+	btCapsuleShapeZ* shape = new btCapsuleShapeZ(radius, height);
 	m_allocatedCollisionShapes.push_back(shape);
 	return shape;
 }
 
-btCollisionShape* btCollisionWorldImporter::createCylinderShapeX(btScalar radius,btScalar height)
+btCollisionShape* btCollisionWorldImporter::createCylinderShapeX(btScalar radius, btScalar height)
 {
-	btCylinderShapeX* shape = new btCylinderShapeX(btVector3(height,radius,radius));
+	btCylinderShapeX* shape = new btCylinderShapeX(btVector3(height, radius, radius));
 	m_allocatedCollisionShapes.push_back(shape);
 	return shape;
 }
 
-btCollisionShape* btCollisionWorldImporter::createCylinderShapeY(btScalar radius,btScalar height)
+btCollisionShape* btCollisionWorldImporter::createCylinderShapeY(btScalar radius, btScalar height)
 {
-	btCylinderShape* shape = new btCylinderShape(btVector3(radius,height,radius));
+	btCylinderShape* shape = new btCylinderShape(btVector3(radius, height, radius));
 	m_allocatedCollisionShapes.push_back(shape);
 	return shape;
 }
 
-btCollisionShape* btCollisionWorldImporter::createCylinderShapeZ(btScalar radius,btScalar height)
+btCollisionShape* btCollisionWorldImporter::createCylinderShapeZ(btScalar radius, btScalar height)
 {
-	btCylinderShapeZ* shape = new btCylinderShapeZ(btVector3(radius,radius,height));
+	btCylinderShapeZ* shape = new btCylinderShapeZ(btVector3(radius, radius, height));
 	m_allocatedCollisionShapes.push_back(shape);
 	return shape;
 }
 
-btCollisionShape* btCollisionWorldImporter::createConeShapeX(btScalar radius,btScalar height)
+btCollisionShape* btCollisionWorldImporter::createConeShapeX(btScalar radius, btScalar height)
 {
-	btConeShapeX* shape = new btConeShapeX(radius,height);
+	btConeShapeX* shape = new btConeShapeX(radius, height);
 	m_allocatedCollisionShapes.push_back(shape);
 	return shape;
 }
 
-btCollisionShape* btCollisionWorldImporter::createConeShapeY(btScalar radius,btScalar height)
+btCollisionShape* btCollisionWorldImporter::createConeShapeY(btScalar radius, btScalar height)
 {
-	btConeShape* shape = new btConeShape(radius,height);
+	btConeShape* shape = new btConeShape(radius, height);
 	m_allocatedCollisionShapes.push_back(shape);
 	return shape;
 }
 
-btCollisionShape* btCollisionWorldImporter::createConeShapeZ(btScalar radius,btScalar height)
+btCollisionShape* btCollisionWorldImporter::createConeShapeZ(btScalar radius, btScalar height)
 {
-	btConeShapeZ* shape = new btConeShapeZ(radius,height);
+	btConeShapeZ* shape = new btConeShapeZ(radius, height);
 	m_allocatedCollisionShapes.push_back(shape);
 	return shape;
 }
 
-btTriangleIndexVertexArray*	btCollisionWorldImporter::createTriangleMeshContainer()
+btTriangleIndexVertexArray* btCollisionWorldImporter::createTriangleMeshContainer()
 {
 	btTriangleIndexVertexArray* in = new btTriangleIndexVertexArray();
 	m_allocatedTriangleIndexArrays.push_back(in);
 	return in;
 }
 
-btOptimizedBvh*	btCollisionWorldImporter::createOptimizedBvh()
+btOptimizedBvh* btCollisionWorldImporter::createOptimizedBvh()
 {
 	btOptimizedBvh* bvh = new btOptimizedBvh();
 	m_allocatedBvhs.push_back(bvh);
 	return bvh;
 }
 
-
 btTriangleInfoMap* btCollisionWorldImporter::createTriangleInfoMap()
 {
 	btTriangleInfoMap* tim = new btTriangleInfoMap();
@@ -1027,16 +977,15 @@ btBvhTriangleMeshShape* btCollisionWorldImporter::createBvhTriangleMeshShape(btS
 {
 	if (bvh)
 	{
-		btBvhTriangleMeshShape* bvhTriMesh = new btBvhTriangleMeshShape(trimesh,bvh->isQuantized(), false);
+		btBvhTriangleMeshShape* bvhTriMesh = new btBvhTriangleMeshShape(trimesh, bvh->isQuantized(), false);
 		bvhTriMesh->setOptimizedBvh(bvh);
 		m_allocatedCollisionShapes.push_back(bvhTriMesh);
 		return bvhTriMesh;
 	}
 
-	btBvhTriangleMeshShape* ts = new btBvhTriangleMeshShape(trimesh,true);
+	btBvhTriangleMeshShape* ts = new btBvhTriangleMeshShape(trimesh, true);
 	m_allocatedCollisionShapes.push_back(ts);
 	return ts;
-
 }
 btCollisionShape* btCollisionWorldImporter::createConvexTriangleMeshShape(btStridingMeshInterface* trimesh)
 {
@@ -1048,9 +997,8 @@ btGImpactMeshShape* btCollisionWorldImporter::createGimpactShape(btStridingMeshI
 	btGImpactMeshShape* shape = new btGImpactMeshShape(trimesh);
 	m_allocatedCollisionShapes.push_back(shape);
 	return shape;
-
 }
-#endif //SUPPORT_GIMPACT_SHAPE_IMPORT
+#endif  //SUPPORT_GIMPACT_SHAPE_IMPORT
 
 btConvexHullShape* btCollisionWorldImporter::createConvexHullShape()
 {
@@ -1066,25 +1014,22 @@ btCompoundShape* btCollisionWorldImporter::createCompoundShape()
 	return shape;
 }
 
-
-btScaledBvhTriangleMeshShape* btCollisionWorldImporter::createScaledTrangleMeshShape(btBvhTriangleMeshShape* meshShape,const btVector3& localScaling)
+btScaledBvhTriangleMeshShape* btCollisionWorldImporter::createScaledTrangleMeshShape(btBvhTriangleMeshShape* meshShape, const btVector3& localScaling)
 {
-	btScaledBvhTriangleMeshShape* shape = new btScaledBvhTriangleMeshShape(meshShape,localScaling);
+	btScaledBvhTriangleMeshShape* shape = new btScaledBvhTriangleMeshShape(meshShape, localScaling);
 	m_allocatedCollisionShapes.push_back(shape);
 	return shape;
 }
 
-btMultiSphereShape* btCollisionWorldImporter::createMultiSphereShape(const btVector3* positions,const btScalar* radi,int numSpheres)
+btMultiSphereShape* btCollisionWorldImporter::createMultiSphereShape(const btVector3* positions, const btScalar* radi, int numSpheres)
 {
 	btMultiSphereShape* shape = new btMultiSphereShape(positions, radi, numSpheres);
 	m_allocatedCollisionShapes.push_back(shape);
 	return shape;
 }
 
-
-
-	// query for data
-int	btCollisionWorldImporter::getNumCollisionShapes() const
+// query for data
+int btCollisionWorldImporter::getNumCollisionShapes() const
 {
 	return m_allocatedCollisionShapes.size();
 }
@@ -1097,23 +1042,21 @@ btCollisionShape* btCollisionWorldImporter::getCollisionShapeByIndex(int index)
 btCollisionShape* btCollisionWorldImporter::getCollisionShapeByName(const char* name)
 {
 	btCollisionShape** shapePtr = m_nameShapeMap.find(name);
-	if (shapePtr&& *shapePtr)
+	if (shapePtr && *shapePtr)
 	{
 		return *shapePtr;
 	}
 	return 0;
 }
 
-
-const char*	btCollisionWorldImporter::getNameForPointer(const void* ptr) const
+const char* btCollisionWorldImporter::getNameForPointer(const void* ptr) const
 {
-	const char*const * namePtr = m_objectNameMap.find(ptr);
+	const char* const* namePtr = m_objectNameMap.find(ptr);
 	if (namePtr && *namePtr)
 		return *namePtr;
 	return 0;
 }
 
-
 int btCollisionWorldImporter::getNumRigidBodies() const
 {
 	return m_allocatedRigidBodies.size();
@@ -1124,12 +1067,11 @@ btCollisionObject* btCollisionWorldImporter::getRigidBodyByIndex(int index) cons
 	return m_allocatedRigidBodies[index];
 }
 
-
 int btCollisionWorldImporter::getNumBvhs() const
 {
 	return m_allocatedBvhs.size();
 }
- btOptimizedBvh* btCollisionWorldImporter::getBvhByIndex(int index) const
+btOptimizedBvh* btCollisionWorldImporter::getBvhByIndex(int index) const
 {
 	return m_allocatedBvhs[index];
 }
@@ -1143,5 +1085,3 @@ btTriangleInfoMap* btCollisionWorldImporter::getTriangleInfoMapByIndex(int index
 {
 	return m_allocatedTriangleInfoMaps[index];
 }
-
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionWorldImporter.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionWorldImporter.h
index 81c61427267..5e8bc95341d 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionWorldImporter.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCollisionWorldImporter.h
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef BT_COLLISION_WORLD_IMPORTER_H
 #define BT_COLLISION_WORLD_IMPORTER_H
 
@@ -26,7 +25,6 @@ class btCollisionShape;
 class btCollisionObject;
 struct btBulletSerializedArrays;
 
-
 struct ConstraintInput;
 class btCollisionWorld;
 struct btCollisionShapeData;
@@ -46,9 +44,6 @@ class btSliderConstraint;
 class btGearConstraint;
 struct btContactSolverInfo;
 
-
-
-
 class btCollisionWorldImporter
 {
 protected:
@@ -56,60 +51,53 @@ protected:
 
 	int m_verboseMode;
 
-	btAlignedObjectArray<btCollisionShape*>  m_allocatedCollisionShapes;
+	btAlignedObjectArray<btCollisionShape*> m_allocatedCollisionShapes;
 	btAlignedObjectArray<btCollisionObject*> m_allocatedRigidBodies;
 
-	btAlignedObjectArray<btOptimizedBvh*>	 m_allocatedBvhs;
+	btAlignedObjectArray<btOptimizedBvh*> m_allocatedBvhs;
 	btAlignedObjectArray<btTriangleInfoMap*> m_allocatedTriangleInfoMaps;
 	btAlignedObjectArray<btTriangleIndexVertexArray*> m_allocatedTriangleIndexArrays;
 	btAlignedObjectArray<btStridingMeshInterfaceData*> m_allocatedbtStridingMeshInterfaceDatas;
 	btAlignedObjectArray<btCollisionObject*> m_allocatedCollisionObjects;
 
+	btAlignedObjectArray<char*> m_allocatedNames;
 
-	btAlignedObjectArray<char*>				m_allocatedNames;
+	btAlignedObjectArray<int*> m_indexArrays;
+	btAlignedObjectArray<short int*> m_shortIndexArrays;
+	btAlignedObjectArray<unsigned char*> m_charIndexArrays;
 
-	btAlignedObjectArray<int*>				m_indexArrays;
-	btAlignedObjectArray<short int*>		m_shortIndexArrays;
-	btAlignedObjectArray<unsigned char*>	m_charIndexArrays;
+	btAlignedObjectArray<btVector3FloatData*> m_floatVertexArrays;
+	btAlignedObjectArray<btVector3DoubleData*> m_doubleVertexArrays;
 
-	btAlignedObjectArray<btVector3FloatData*>	m_floatVertexArrays;
-	btAlignedObjectArray<btVector3DoubleData*>	m_doubleVertexArrays;
+	btHashMap<btHashPtr, btOptimizedBvh*> m_bvhMap;
+	btHashMap<btHashPtr, btTriangleInfoMap*> m_timMap;
 
+	btHashMap<btHashString, btCollisionShape*> m_nameShapeMap;
+	btHashMap<btHashString, btCollisionObject*> m_nameColObjMap;
 
-	btHashMap<btHashPtr,btOptimizedBvh*>	m_bvhMap;
-	btHashMap<btHashPtr,btTriangleInfoMap*>	m_timMap;
-
-	btHashMap<btHashString,btCollisionShape*>	m_nameShapeMap;
-	btHashMap<btHashString,btCollisionObject*>	m_nameColObjMap;
-
-	btHashMap<btHashPtr,const char*>	m_objectNameMap;
-
-	btHashMap<btHashPtr,btCollisionShape*>	m_shapeMap;
-	btHashMap<btHashPtr,btCollisionObject*>	m_bodyMap;
+	btHashMap<btHashPtr, const char*> m_objectNameMap;
 
+	btHashMap<btHashPtr, btCollisionShape*> m_shapeMap;
+	btHashMap<btHashPtr, btCollisionObject*> m_bodyMap;
 
 	//methods
 
+	char* duplicateName(const char* name);
 
-
-	char*	duplicateName(const char* name);
-
-	btCollisionShape* convertCollisionShape(  btCollisionShapeData* shapeData  );
-
+	btCollisionShape* convertCollisionShape(btCollisionShapeData* shapeData);
 
 public:
-
 	btCollisionWorldImporter(btCollisionWorld* world);
 
 	virtual ~btCollisionWorldImporter();
 
-    bool	convertAllObjects( btBulletSerializedArrays* arrays);
+	bool convertAllObjects(btBulletSerializedArrays* arrays);
 
-		///delete all memory collision shapes, rigid bodies, constraints etc. allocated during the load.
+	///delete all memory collision shapes, rigid bodies, constraints etc. allocated during the load.
 	///make sure you don't use the dynamics world containing objects after you call this method
 	virtual void deleteAllData();
 
-	void	setVerboseMode(int verboseMode)
+	void setVerboseMode(int verboseMode)
 	{
 		m_verboseMode = verboseMode;
 	}
@@ -119,14 +107,14 @@ public:
 		return m_verboseMode;
 	}
 
-		// query for data
-	int	getNumCollisionShapes() const;
+	// query for data
+	int getNumCollisionShapes() const;
 	btCollisionShape* getCollisionShapeByIndex(int index);
 	int getNumRigidBodies() const;
 	btCollisionObject* getRigidBodyByIndex(int index) const;
 
 	int getNumBvhs() const;
-	btOptimizedBvh*  getBvhByIndex(int index) const;
+	btOptimizedBvh* getBvhByIndex(int index) const;
 	int getNumTriangleInfoMaps() const;
 	btTriangleInfoMap* getTriangleInfoMapByIndex(int index) const;
 
@@ -134,56 +122,48 @@ public:
 	btCollisionShape* getCollisionShapeByName(const char* name);
 	btCollisionObject* getCollisionObjectByName(const char* name);
 
-
-	const char*	getNameForPointer(const void* ptr) const;
+	const char* getNameForPointer(const void* ptr) const;
 
 	///those virtuals are called by load and can be overridden by the user
 
-
-
 	//bodies
 
-	virtual btCollisionObject*  createCollisionObject(	const btTransform& startTransform,	btCollisionShape* shape,const char* bodyName);
+	virtual btCollisionObject* createCollisionObject(const btTransform& startTransform, btCollisionShape* shape, const char* bodyName);
 
 	///shapes
 
-	virtual btCollisionShape* createPlaneShape(const btVector3& planeNormal,btScalar planeConstant);
+	virtual btCollisionShape* createPlaneShape(const btVector3& planeNormal, btScalar planeConstant);
 	virtual btCollisionShape* createBoxShape(const btVector3& halfExtents);
 	virtual btCollisionShape* createSphereShape(btScalar radius);
 	virtual btCollisionShape* createCapsuleShapeX(btScalar radius, btScalar height);
 	virtual btCollisionShape* createCapsuleShapeY(btScalar radius, btScalar height);
 	virtual btCollisionShape* createCapsuleShapeZ(btScalar radius, btScalar height);
 
-	virtual btCollisionShape* createCylinderShapeX(btScalar radius,btScalar height);
-	virtual btCollisionShape* createCylinderShapeY(btScalar radius,btScalar height);
-	virtual btCollisionShape* createCylinderShapeZ(btScalar radius,btScalar height);
-	virtual btCollisionShape* createConeShapeX(btScalar radius,btScalar height);
-	virtual btCollisionShape* createConeShapeY(btScalar radius,btScalar height);
-	virtual btCollisionShape* createConeShapeZ(btScalar radius,btScalar height);
-	virtual class btTriangleIndexVertexArray*	createTriangleMeshContainer();
-	virtual	btBvhTriangleMeshShape* createBvhTriangleMeshShape(btStridingMeshInterface* trimesh, btOptimizedBvh* bvh);
+	virtual btCollisionShape* createCylinderShapeX(btScalar radius, btScalar height);
+	virtual btCollisionShape* createCylinderShapeY(btScalar radius, btScalar height);
+	virtual btCollisionShape* createCylinderShapeZ(btScalar radius, btScalar height);
+	virtual btCollisionShape* createConeShapeX(btScalar radius, btScalar height);
+	virtual btCollisionShape* createConeShapeY(btScalar radius, btScalar height);
+	virtual btCollisionShape* createConeShapeZ(btScalar radius, btScalar height);
+	virtual class btTriangleIndexVertexArray* createTriangleMeshContainer();
+	virtual btBvhTriangleMeshShape* createBvhTriangleMeshShape(btStridingMeshInterface* trimesh, btOptimizedBvh* bvh);
 	virtual btCollisionShape* createConvexTriangleMeshShape(btStridingMeshInterface* trimesh);
 #ifdef SUPPORT_GIMPACT_SHAPE_IMPORT
 	virtual btGImpactMeshShape* createGimpactShape(btStridingMeshInterface* trimesh);
-#endif //SUPPORT_GIMPACT_SHAPE_IMPORT
+#endif  //SUPPORT_GIMPACT_SHAPE_IMPORT
 	virtual btStridingMeshInterfaceData* createStridingMeshInterfaceData(btStridingMeshInterfaceData* interfaceData);
 
 	virtual class btConvexHullShape* createConvexHullShape();
 	virtual class btCompoundShape* createCompoundShape();
-	virtual class btScaledBvhTriangleMeshShape* createScaledTrangleMeshShape(btBvhTriangleMeshShape* meshShape,const btVector3& localScalingbtBvhTriangleMeshShape);
+	virtual class btScaledBvhTriangleMeshShape* createScaledTrangleMeshShape(btBvhTriangleMeshShape* meshShape, const btVector3& localScalingbtBvhTriangleMeshShape);
 
-	virtual class btMultiSphereShape* createMultiSphereShape(const btVector3* positions,const btScalar* radi,int numSpheres);
+	virtual class btMultiSphereShape* createMultiSphereShape(const btVector3* positions, const btScalar* radi, int numSpheres);
 
 	virtual btTriangleIndexVertexArray* createMeshInterface(btStridingMeshInterfaceData& meshData);
 
 	///acceleration and connectivity structures
-	virtual btOptimizedBvh*	createOptimizedBvh();
+	virtual btOptimizedBvh* createOptimizedBvh();
 	virtual btTriangleInfoMap* createTriangleInfoMap();
-
-
-
-
 };
 
-
-#endif //BT_WORLD_IMPORTER_H
+#endif  //BT_WORLD_IMPORTER_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCompoundCollisionAlgorithm.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCompoundCollisionAlgorithm.cpp
index 91b7809c17b..633bee48253 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCompoundCollisionAlgorithm.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCompoundCollisionAlgorithm.cpp
@@ -25,62 +25,58 @@ subject to the following restrictions:
 
 btShapePairCallback gCompoundChildShapePairCallback = 0;
 
-btCompoundCollisionAlgorithm::btCompoundCollisionAlgorithm( const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,bool isSwapped)
-:btActivatingCollisionAlgorithm(ci,body0Wrap,body1Wrap),
-m_isSwapped(isSwapped),
-m_sharedManifold(ci.m_manifold)
+btCompoundCollisionAlgorithm::btCompoundCollisionAlgorithm(const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, bool isSwapped)
+	: btActivatingCollisionAlgorithm(ci, body0Wrap, body1Wrap),
+	  m_isSwapped(isSwapped),
+	  m_sharedManifold(ci.m_manifold)
 {
 	m_ownsManifold = false;
 
-	const btCollisionObjectWrapper* colObjWrap = m_isSwapped? body1Wrap : body0Wrap;
-	btAssert (colObjWrap->getCollisionShape()->isCompound());
-	
+	const btCollisionObjectWrapper* colObjWrap = m_isSwapped ? body1Wrap : body0Wrap;
+	btAssert(colObjWrap->getCollisionShape()->isCompound());
+
 	const btCompoundShape* compoundShape = static_cast<const btCompoundShape*>(colObjWrap->getCollisionShape());
 	m_compoundShapeRevision = compoundShape->getUpdateRevision();
-	
-	
-	preallocateChildAlgorithms(body0Wrap,body1Wrap);
+
+	preallocateChildAlgorithms(body0Wrap, body1Wrap);
 }
 
-void	btCompoundCollisionAlgorithm::preallocateChildAlgorithms(const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+void btCompoundCollisionAlgorithm::preallocateChildAlgorithms(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 {
-	const btCollisionObjectWrapper* colObjWrap = m_isSwapped? body1Wrap : body0Wrap;
-	const btCollisionObjectWrapper* otherObjWrap = m_isSwapped? body0Wrap : body1Wrap;
-	btAssert (colObjWrap->getCollisionShape()->isCompound());
-	
+	const btCollisionObjectWrapper* colObjWrap = m_isSwapped ? body1Wrap : body0Wrap;
+	const btCollisionObjectWrapper* otherObjWrap = m_isSwapped ? body0Wrap : body1Wrap;
+	btAssert(colObjWrap->getCollisionShape()->isCompound());
+
 	const btCompoundShape* compoundShape = static_cast<const btCompoundShape*>(colObjWrap->getCollisionShape());
 
 	int numChildren = compoundShape->getNumChildShapes();
 	int i;
-	
+
 	m_childCollisionAlgorithms.resize(numChildren);
-	for (i=0;i<numChildren;i++)
+	for (i = 0; i < numChildren; i++)
 	{
 		if (compoundShape->getDynamicAabbTree())
 		{
 			m_childCollisionAlgorithms[i] = 0;
-		} else
+		}
+		else
 		{
-			
 			const btCollisionShape* childShape = compoundShape->getChildShape(i);
 
-			btCollisionObjectWrapper childWrap(colObjWrap,childShape,colObjWrap->getCollisionObject(),colObjWrap->getWorldTransform(),-1,i);//wrong child trans, but unused (hopefully)
-			m_childCollisionAlgorithms[i] = m_dispatcher->findAlgorithm(&childWrap,otherObjWrap,m_sharedManifold, BT_CONTACT_POINT_ALGORITHMS);
-
+			btCollisionObjectWrapper childWrap(colObjWrap, childShape, colObjWrap->getCollisionObject(), colObjWrap->getWorldTransform(), -1, i);  //wrong child trans, but unused (hopefully)
+			m_childCollisionAlgorithms[i] = m_dispatcher->findAlgorithm(&childWrap, otherObjWrap, m_sharedManifold, BT_CONTACT_POINT_ALGORITHMS);
 
 			btAlignedObjectArray<btCollisionAlgorithm*> m_childCollisionAlgorithmsContact;
 			btAlignedObjectArray<btCollisionAlgorithm*> m_childCollisionAlgorithmsClosestPoints;
-
-
 		}
 	}
 }
 
-void	btCompoundCollisionAlgorithm::removeChildAlgorithms()
+void btCompoundCollisionAlgorithm::removeChildAlgorithms()
 {
 	int numChildren = m_childCollisionAlgorithms.size();
 	int i;
-	for (i=0;i<numChildren;i++)
+	for (i = 0; i < numChildren; i++)
 	{
 		if (m_childCollisionAlgorithms[i])
 		{
@@ -95,54 +91,27 @@ btCompoundCollisionAlgorithm::~btCompoundCollisionAlgorithm()
 	removeChildAlgorithms();
 }
 
-
-
-
-struct	btCompoundLeafCallback : btDbvt::ICollide
+struct btCompoundLeafCallback : btDbvt::ICollide
 {
-
 public:
-
 	const btCollisionObjectWrapper* m_compoundColObjWrap;
 	const btCollisionObjectWrapper* m_otherObjWrap;
 	btDispatcher* m_dispatcher;
 	const btDispatcherInfo& m_dispatchInfo;
-	btManifoldResult*	m_resultOut;
-	btCollisionAlgorithm**	m_childCollisionAlgorithms;
-	btPersistentManifold*	m_sharedManifold;
-	
-	btCompoundLeafCallback (const btCollisionObjectWrapper* compoundObjWrap,const btCollisionObjectWrapper* otherObjWrap,btDispatcher* dispatcher,const btDispatcherInfo& dispatchInfo,btManifoldResult*	resultOut,btCollisionAlgorithm**	childCollisionAlgorithms,btPersistentManifold*	sharedManifold)
-		:m_compoundColObjWrap(compoundObjWrap),m_otherObjWrap(otherObjWrap),m_dispatcher(dispatcher),m_dispatchInfo(dispatchInfo),m_resultOut(resultOut),
-		m_childCollisionAlgorithms(childCollisionAlgorithms),
-		m_sharedManifold(sharedManifold)
-	{
+	btManifoldResult* m_resultOut;
+	btCollisionAlgorithm** m_childCollisionAlgorithms;
+	btPersistentManifold* m_sharedManifold;
 
+	btCompoundLeafCallback(const btCollisionObjectWrapper* compoundObjWrap, const btCollisionObjectWrapper* otherObjWrap, btDispatcher* dispatcher, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut, btCollisionAlgorithm** childCollisionAlgorithms, btPersistentManifold* sharedManifold)
+		: m_compoundColObjWrap(compoundObjWrap), m_otherObjWrap(otherObjWrap), m_dispatcher(dispatcher), m_dispatchInfo(dispatchInfo), m_resultOut(resultOut), m_childCollisionAlgorithms(childCollisionAlgorithms), m_sharedManifold(sharedManifold)
+	{
 	}
 
-
-	void	ProcessChildShape(const btCollisionShape* childShape,int index)
+	void ProcessChildShape(const btCollisionShape* childShape, int index)
 	{
-		btAssert(index>=0);
+		btAssert(index >= 0);
 		const btCompoundShape* compoundShape = static_cast<const btCompoundShape*>(m_compoundColObjWrap->getCollisionShape());
-		btAssert(index<compoundShape->getNumChildShapes());
-
-
-		//backup
-		btTransform	orgTrans = m_compoundColObjWrap->getWorldTransform();
-		
-		const btTransform& childTrans = compoundShape->getChildTransform(index);
-		btTransform	newChildWorldTrans = orgTrans*childTrans ;
-
-		//perform an AABB check first
-		btVector3 aabbMin0,aabbMax0;
-		childShape->getAabb(newChildWorldTrans,aabbMin0,aabbMax0);
-
-		btVector3 extendAabb(m_resultOut->m_closestPointDistanceThreshold, m_resultOut->m_closestPointDistanceThreshold, m_resultOut->m_closestPointDistanceThreshold);
-		aabbMin0 -= extendAabb;
-		aabbMax0 += extendAabb;
-
-		btVector3 aabbMin1, aabbMax1;
-		m_otherObjWrap->getCollisionShape()->getAabb(m_otherObjWrap->getWorldTransform(),aabbMin1,aabbMax1);
+		btAssert(index < compoundShape->getNumChildShapes());
 
 		if (gCompoundChildShapePairCallback)
 		{
@@ -150,11 +119,28 @@ public:
 				return;
 		}
 
-		if (TestAabbAgainstAabb2(aabbMin0,aabbMax0,aabbMin1,aabbMax1))
-		{
+		//backup
+		btTransform orgTrans = m_compoundColObjWrap->getWorldTransform();
+
+		const btTransform& childTrans = compoundShape->getChildTransform(index);
+		btTransform newChildWorldTrans = orgTrans * childTrans;
+
+		//perform an AABB check first
+		btVector3 aabbMin0, aabbMax0;
+		childShape->getAabb(newChildWorldTrans, aabbMin0, aabbMax0);
+
+		btVector3 extendAabb(m_resultOut->m_closestPointDistanceThreshold, m_resultOut->m_closestPointDistanceThreshold, m_resultOut->m_closestPointDistanceThreshold);
+		aabbMin0 -= extendAabb;
+		aabbMax0 += extendAabb;
+
+		btVector3 aabbMin1, aabbMax1;
+		m_otherObjWrap->getCollisionShape()->getAabb(m_otherObjWrap->getWorldTransform(), aabbMin1, aabbMax1);
+
+
+		if (TestAabbAgainstAabb2(aabbMin0, aabbMax0, aabbMin1, aabbMax1))
+		{
+			btCollisionObjectWrapper compoundWrap(this->m_compoundColObjWrap, childShape, m_compoundColObjWrap->getCollisionObject(), newChildWorldTrans, -1, index);
 
-			btCollisionObjectWrapper compoundWrap(this->m_compoundColObjWrap,childShape,m_compoundColObjWrap->getCollisionObject(),newChildWorldTrans,-1,index);
-			
 			btCollisionAlgorithm* algo = 0;
 			bool allocatedAlgorithm = false;
 
@@ -172,7 +158,7 @@ public:
 				}
 				algo = m_childCollisionAlgorithms[index];
 			}
-			
+
 			const btCollisionObjectWrapper* tmpWrap = 0;
 
 			///detect swapping case
@@ -180,15 +166,16 @@ public:
 			{
 				tmpWrap = m_resultOut->getBody0Wrap();
 				m_resultOut->setBody0Wrap(&compoundWrap);
-				m_resultOut->setShapeIdentifiersA(-1,index);
-			} else
+				m_resultOut->setShapeIdentifiersA(-1, index);
+			}
+			else
 			{
 				tmpWrap = m_resultOut->getBody1Wrap();
 				m_resultOut->setBody1Wrap(&compoundWrap);
-				m_resultOut->setShapeIdentifiersB(-1,index);
+				m_resultOut->setShapeIdentifiersB(-1, index);
 			}
 
-			algo->processCollision(&compoundWrap,m_otherObjWrap,m_dispatchInfo,m_resultOut);
+			algo->processCollision(&compoundWrap, m_otherObjWrap, m_dispatchInfo, m_resultOut);
 
 #if 0
 			if (m_dispatchInfo.m_debugDraw && (m_dispatchInfo.m_debugDraw->getDebugMode() & btIDebugDraw::DBG_DrawAabb))
@@ -202,18 +189,19 @@ public:
 			if (m_resultOut->getBody0Internal() == m_compoundColObjWrap->getCollisionObject())
 			{
 				m_resultOut->setBody0Wrap(tmpWrap);
-			} else
+			}
+			else
 			{
 				m_resultOut->setBody1Wrap(tmpWrap);
 			}
-			if(allocatedAlgorithm)
+			if (allocatedAlgorithm)
 			{
 				algo->~btCollisionAlgorithm();
 				m_dispatcher->freeCollisionAlgorithm(algo);
- 			}
+			}
 		}
 	}
-	void		Process(const btDbvtNode* leaf)
+	void Process(const btDbvtNode* leaf)
 	{
 		int index = leaf->dataAsInt;
 
@@ -230,22 +218,16 @@ public:
 		}
 #endif
 
-		ProcessChildShape(childShape,index);
-
+		ProcessChildShape(childShape, index);
 	}
 };
 
-
-
-
-
-
-void btCompoundCollisionAlgorithm::processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+void btCompoundCollisionAlgorithm::processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
-	const btCollisionObjectWrapper* colObjWrap = m_isSwapped? body1Wrap : body0Wrap;
-	const btCollisionObjectWrapper* otherObjWrap = m_isSwapped? body0Wrap : body1Wrap;
+	const btCollisionObjectWrapper* colObjWrap = m_isSwapped ? body1Wrap : body0Wrap;
+	const btCollisionObjectWrapper* otherObjWrap = m_isSwapped ? body0Wrap : body1Wrap;
 
-	btAssert (colObjWrap->getCollisionShape()->isCompound());
+	btAssert(colObjWrap->getCollisionShape()->isCompound());
 	const btCompoundShape* compoundShape = static_cast<const btCompoundShape*>(colObjWrap->getCollisionShape());
 
 	///btCompoundShape might have changed:
@@ -254,17 +236,17 @@ void btCompoundCollisionAlgorithm::processCollision (const btCollisionObjectWrap
 	{
 		///clear and update all
 		removeChildAlgorithms();
-		
-		preallocateChildAlgorithms(body0Wrap,body1Wrap);
+
+		preallocateChildAlgorithms(body0Wrap, body1Wrap);
 		m_compoundShapeRevision = compoundShape->getUpdateRevision();
 	}
 
-	if (m_childCollisionAlgorithms.size()==0)
+	if (m_childCollisionAlgorithms.size() == 0)
 		return;
 
 	const btDbvt* tree = compoundShape->getDynamicAabbTree();
 	//use a dynamic aabb tree to cull potential child-overlaps
-	btCompoundLeafCallback  callback(colObjWrap,otherObjWrap,m_dispatcher,dispatchInfo,resultOut,&m_childCollisionAlgorithms[0],m_sharedManifold);
+	btCompoundLeafCallback callback(colObjWrap, otherObjWrap, m_dispatcher, dispatchInfo, resultOut, &m_childCollisionAlgorithms[0], m_sharedManifold);
 
 	///we need to refresh all contact manifolds
 	///note that we should actually recursively traverse all children, btCompoundShape can nested more then 1 level deep
@@ -272,18 +254,18 @@ void btCompoundCollisionAlgorithm::processCollision (const btCollisionObjectWrap
 	{
 		int i;
 		manifoldArray.resize(0);
-		for (i=0;i<m_childCollisionAlgorithms.size();i++)
+		for (i = 0; i < m_childCollisionAlgorithms.size(); i++)
 		{
 			if (m_childCollisionAlgorithms[i])
 			{
 				m_childCollisionAlgorithms[i]->getAllContactManifolds(manifoldArray);
-				for (int m=0;m<manifoldArray.size();m++)
+				for (int m = 0; m < manifoldArray.size(); m++)
 				{
 					if (manifoldArray[m]->getNumContacts())
 					{
 						resultOut->setPersistentManifold(manifoldArray[m]);
 						resultOut->refreshContactPoints();
-						resultOut->setPersistentManifold(0);//??necessary?
+						resultOut->setPersistentManifold(0);  //??necessary?
 					}
 				}
 				manifoldArray.resize(0);
@@ -293,57 +275,56 @@ void btCompoundCollisionAlgorithm::processCollision (const btCollisionObjectWrap
 
 	if (tree)
 	{
-
-		btVector3 localAabbMin,localAabbMax;
+		btVector3 localAabbMin, localAabbMax;
 		btTransform otherInCompoundSpace;
 		otherInCompoundSpace = colObjWrap->getWorldTransform().inverse() * otherObjWrap->getWorldTransform();
-		otherObjWrap->getCollisionShape()->getAabb(otherInCompoundSpace,localAabbMin,localAabbMax);
+		otherObjWrap->getCollisionShape()->getAabb(otherInCompoundSpace, localAabbMin, localAabbMax);
 		btVector3 extraExtends(resultOut->m_closestPointDistanceThreshold, resultOut->m_closestPointDistanceThreshold, resultOut->m_closestPointDistanceThreshold);
 		localAabbMin -= extraExtends;
 		localAabbMax += extraExtends;
 
-		const ATTRIBUTE_ALIGNED16(btDbvtVolume)	bounds=btDbvtVolume::FromMM(localAabbMin,localAabbMax);
+		const ATTRIBUTE_ALIGNED16(btDbvtVolume) bounds = btDbvtVolume::FromMM(localAabbMin, localAabbMax);
 		//process all children, that overlap with  the given AABB bounds
-		tree->collideTVNoStackAlloc(tree->m_root,bounds,stack2,callback);
-
-	} else
+		tree->collideTVNoStackAlloc(tree->m_root, bounds, stack2, callback);
+	}
+	else
 	{
 		//iterate over all children, perform an AABB check inside ProcessChildShape
 		int numChildren = m_childCollisionAlgorithms.size();
 		int i;
-		for (i=0;i<numChildren;i++)
+		for (i = 0; i < numChildren; i++)
 		{
-			callback.ProcessChildShape(compoundShape->getChildShape(i),i);
+			callback.ProcessChildShape(compoundShape->getChildShape(i), i);
 		}
 	}
 
 	{
-				//iterate over all children, perform an AABB check inside ProcessChildShape
+		//iterate over all children, perform an AABB check inside ProcessChildShape
 		int numChildren = m_childCollisionAlgorithms.size();
 		int i;
 		manifoldArray.resize(0);
-        const btCollisionShape* childShape = 0;
-        btTransform	orgTrans;
-        
-        btTransform	newChildWorldTrans;
-        btVector3 aabbMin0,aabbMax0,aabbMin1,aabbMax1;        
-        
-		for (i=0;i<numChildren;i++)
+		const btCollisionShape* childShape = 0;
+		btTransform orgTrans;
+
+		btTransform newChildWorldTrans;
+		btVector3 aabbMin0, aabbMax0, aabbMin1, aabbMax1;
+
+		for (i = 0; i < numChildren; i++)
 		{
 			if (m_childCollisionAlgorithms[i])
 			{
 				childShape = compoundShape->getChildShape(i);
-			//if not longer overlapping, remove the algorithm
+				//if not longer overlapping, remove the algorithm
 				orgTrans = colObjWrap->getWorldTransform();
-                
+
 				const btTransform& childTrans = compoundShape->getChildTransform(i);
-                newChildWorldTrans = orgTrans*childTrans ;
+				newChildWorldTrans = orgTrans * childTrans;
 
 				//perform an AABB check first
-				childShape->getAabb(newChildWorldTrans,aabbMin0,aabbMax0);
-				otherObjWrap->getCollisionShape()->getAabb(otherObjWrap->getWorldTransform(),aabbMin1,aabbMax1);
+				childShape->getAabb(newChildWorldTrans, aabbMin0, aabbMax0);
+				otherObjWrap->getCollisionShape()->getAabb(otherObjWrap->getWorldTransform(), aabbMin1, aabbMax1);
 
-				if (!TestAabbAgainstAabb2(aabbMin0,aabbMax0,aabbMin1,aabbMax1))
+				if (!TestAabbAgainstAabb2(aabbMin0, aabbMax0, aabbMin1, aabbMax1))
 				{
 					m_childCollisionAlgorithms[i]->~btCollisionAlgorithm();
 					m_dispatcher->freeCollisionAlgorithm(m_childCollisionAlgorithms[i]);
@@ -354,15 +335,15 @@ void btCompoundCollisionAlgorithm::processCollision (const btCollisionObjectWrap
 	}
 }
 
-btScalar	btCompoundCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+btScalar btCompoundCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	btAssert(0);
 	//needs to be fixed, using btCollisionObjectWrapper and NOT modifying internal data structures
-	btCollisionObject* colObj = m_isSwapped? body1 : body0;
-	btCollisionObject* otherObj = m_isSwapped? body0 : body1;
+	btCollisionObject* colObj = m_isSwapped ? body1 : body0;
+	btCollisionObject* otherObj = m_isSwapped ? body0 : body1;
+
+	btAssert(colObj->getCollisionShape()->isCompound());
 
-	btAssert (colObj->getCollisionShape()->isCompound());
-	
 	btCompoundShape* compoundShape = static_cast<btCompoundShape*>(colObj->getCollisionShape());
 
 	//We will use the OptimizedBVH, AABB tree to cull potential child-overlaps
@@ -376,33 +357,29 @@ btScalar	btCompoundCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject*
 
 	int numChildren = m_childCollisionAlgorithms.size();
 	int i;
-    btTransform	orgTrans;
-    btScalar frac;
-	for (i=0;i<numChildren;i++)
+	btTransform orgTrans;
+	btScalar frac;
+	for (i = 0; i < numChildren; i++)
 	{
 		//btCollisionShape* childShape = compoundShape->getChildShape(i);
 
 		//backup
-        orgTrans = colObj->getWorldTransform();
-	
+		orgTrans = colObj->getWorldTransform();
+
 		const btTransform& childTrans = compoundShape->getChildTransform(i);
 		//btTransform	newChildWorldTrans = orgTrans*childTrans ;
-		colObj->setWorldTransform( orgTrans*childTrans );
+		colObj->setWorldTransform(orgTrans * childTrans);
 
 		//btCollisionShape* tmpShape = colObj->getCollisionShape();
 		//colObj->internalSetTemporaryCollisionShape( childShape );
-        frac = m_childCollisionAlgorithms[i]->calculateTimeOfImpact(colObj,otherObj,dispatchInfo,resultOut);
-		if (frac<hitFraction)
+		frac = m_childCollisionAlgorithms[i]->calculateTimeOfImpact(colObj, otherObj, dispatchInfo, resultOut);
+		if (frac < hitFraction)
 		{
 			hitFraction = frac;
 		}
 		//revert back
 		//colObj->internalSetTemporaryCollisionShape( tmpShape);
-		colObj->setWorldTransform( orgTrans);
+		colObj->setWorldTransform(orgTrans);
 	}
 	return hitFraction;
-
 }
-
-
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCompoundCollisionAlgorithm.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCompoundCollisionAlgorithm.h
index d2086fbc026..4ea5e77185b 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCompoundCollisionAlgorithm.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCompoundCollisionAlgorithm.h
@@ -35,7 +35,7 @@ typedef bool (*btShapePairCallback)(const btCollisionShape* pShape0, const btCol
 extern btShapePairCallback gCompoundChildShapePairCallback;
 
 /// btCompoundCollisionAlgorithm  supports collision between CompoundCollisionShapes and other collision shapes
-class btCompoundCollisionAlgorithm  : public btActivatingCollisionAlgorithm
+class btCompoundCollisionAlgorithm : public btActivatingCollisionAlgorithm
 {
 	btNodeStack stack2;
 	btManifoldArray manifoldArray;
@@ -44,61 +44,56 @@ protected:
 	btAlignedObjectArray<btCollisionAlgorithm*> m_childCollisionAlgorithms;
 	bool m_isSwapped;
 
-	class btPersistentManifold*	m_sharedManifold;
-	bool					m_ownsManifold;
+	class btPersistentManifold* m_sharedManifold;
+	bool m_ownsManifold;
 
+	int m_compoundShapeRevision;  //to keep track of changes, so that childAlgorithm array can be updated
 
-	int	m_compoundShapeRevision;//to keep track of changes, so that childAlgorithm array can be updated
-	
-	void	removeChildAlgorithms();
-	
-	void	preallocateChildAlgorithms(const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap);
+	void removeChildAlgorithms();
+
+	void preallocateChildAlgorithms(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap);
 
 public:
-
-	btCompoundCollisionAlgorithm( const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,bool isSwapped);
+	btCompoundCollisionAlgorithm(const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, bool isSwapped);
 
 	virtual ~btCompoundCollisionAlgorithm();
 
-	btCollisionAlgorithm* getChildAlgorithm (int n) const
+	btCollisionAlgorithm* getChildAlgorithm(int n) const
 	{
 		return m_childCollisionAlgorithms[n];
 	}
 
+	virtual void processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual void processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	btScalar calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	btScalar	calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
-
-	virtual	void	getAllContactManifolds(btManifoldArray&	manifoldArray)
+	virtual void getAllContactManifolds(btManifoldArray& manifoldArray)
 	{
 		int i;
-		for (i=0;i<m_childCollisionAlgorithms.size();i++)
+		for (i = 0; i < m_childCollisionAlgorithms.size(); i++)
 		{
 			if (m_childCollisionAlgorithms[i])
 				m_childCollisionAlgorithms[i]->getAllContactManifolds(manifoldArray);
 		}
 	}
 
-	
-	struct CreateFunc :public 	btCollisionAlgorithmCreateFunc
+	struct CreateFunc : public btCollisionAlgorithmCreateFunc
 	{
-		virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+		virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 		{
 			void* mem = ci.m_dispatcher1->allocateCollisionAlgorithm(sizeof(btCompoundCollisionAlgorithm));
-			return new(mem) btCompoundCollisionAlgorithm(ci,body0Wrap,body1Wrap,false);
+			return new (mem) btCompoundCollisionAlgorithm(ci, body0Wrap, body1Wrap, false);
 		}
 	};
 
-	struct SwappedCreateFunc :public 	btCollisionAlgorithmCreateFunc
+	struct SwappedCreateFunc : public btCollisionAlgorithmCreateFunc
 	{
-		virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+		virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 		{
 			void* mem = ci.m_dispatcher1->allocateCollisionAlgorithm(sizeof(btCompoundCollisionAlgorithm));
-			return new(mem) btCompoundCollisionAlgorithm(ci,body0Wrap,body1Wrap,true);
+			return new (mem) btCompoundCollisionAlgorithm(ci, body0Wrap, body1Wrap, true);
 		}
 	};
-
 };
 
-#endif //BT_COMPOUND_COLLISION_ALGORITHM_H
+#endif  //BT_COMPOUND_COLLISION_ALGORITHM_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCompoundCompoundCollisionAlgorithm.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCompoundCompoundCollisionAlgorithm.cpp
index 20b542f6706..044b60dbb1d 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCompoundCompoundCollisionAlgorithm.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCompoundCompoundCollisionAlgorithm.cpp
@@ -29,29 +29,25 @@ subject to the following restrictions:
 
 btShapePairCallback gCompoundCompoundChildShapePairCallback = 0;
 
-btCompoundCompoundCollisionAlgorithm::btCompoundCompoundCollisionAlgorithm( const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,bool isSwapped)
-:btCompoundCollisionAlgorithm(ci,body0Wrap,body1Wrap,isSwapped)
+btCompoundCompoundCollisionAlgorithm::btCompoundCompoundCollisionAlgorithm(const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, bool isSwapped)
+	: btCompoundCollisionAlgorithm(ci, body0Wrap, body1Wrap, isSwapped)
 {
-
-	void* ptr = btAlignedAlloc(sizeof(btHashedSimplePairCache),16);
-	m_childCollisionAlgorithmCache= new(ptr) btHashedSimplePairCache();
+	void* ptr = btAlignedAlloc(sizeof(btHashedSimplePairCache), 16);
+	m_childCollisionAlgorithmCache = new (ptr) btHashedSimplePairCache();
 
 	const btCollisionObjectWrapper* col0ObjWrap = body0Wrap;
-	btAssert (col0ObjWrap->getCollisionShape()->isCompound());
+	btAssert(col0ObjWrap->getCollisionShape()->isCompound());
 
 	const btCollisionObjectWrapper* col1ObjWrap = body1Wrap;
-	btAssert (col1ObjWrap->getCollisionShape()->isCompound());
-	
+	btAssert(col1ObjWrap->getCollisionShape()->isCompound());
+
 	const btCompoundShape* compoundShape0 = static_cast<const btCompoundShape*>(col0ObjWrap->getCollisionShape());
 	m_compoundShapeRevision0 = compoundShape0->getUpdateRevision();
 
 	const btCompoundShape* compoundShape1 = static_cast<const btCompoundShape*>(col1ObjWrap->getCollisionShape());
 	m_compoundShapeRevision1 = compoundShape1->getUpdateRevision();
-	
-	
 }
 
-
 btCompoundCompoundCollisionAlgorithm::~btCompoundCompoundCollisionAlgorithm()
 {
 	removeChildAlgorithms();
@@ -59,32 +55,30 @@ btCompoundCompoundCollisionAlgorithm::~btCompoundCompoundCollisionAlgorithm()
 	btAlignedFree(m_childCollisionAlgorithmCache);
 }
 
-void	btCompoundCompoundCollisionAlgorithm::getAllContactManifolds(btManifoldArray&	manifoldArray)
+void btCompoundCompoundCollisionAlgorithm::getAllContactManifolds(btManifoldArray& manifoldArray)
 {
 	int i;
 	btSimplePairArray& pairs = m_childCollisionAlgorithmCache->getOverlappingPairArray();
-	for (i=0;i<pairs.size();i++)
+	for (i = 0; i < pairs.size(); i++)
 	{
 		if (pairs[i].m_userPointer)
 		{
-			
 			((btCollisionAlgorithm*)pairs[i].m_userPointer)->getAllContactManifolds(manifoldArray);
 		}
 	}
 }
 
-
-void	btCompoundCompoundCollisionAlgorithm::removeChildAlgorithms()
+void btCompoundCompoundCollisionAlgorithm::removeChildAlgorithms()
 {
 	btSimplePairArray& pairs = m_childCollisionAlgorithmCache->getOverlappingPairArray();
 
 	int numChildren = pairs.size();
 	int i;
-	for (i=0;i<numChildren;i++)
+	for (i = 0; i < numChildren; i++)
 	{
 		if (pairs[i].m_userPointer)
 		{
-			btCollisionAlgorithm* algo = (btCollisionAlgorithm*) pairs[i].m_userPointer;
+			btCollisionAlgorithm* algo = (btCollisionAlgorithm*)pairs[i].m_userPointer;
 			algo->~btCollisionAlgorithm();
 			m_dispatcher->freeCollisionAlgorithm(algo);
 		}
@@ -92,77 +86,65 @@ void	btCompoundCompoundCollisionAlgorithm::removeChildAlgorithms()
 	m_childCollisionAlgorithmCache->removeAllPairs();
 }
 
-struct	btCompoundCompoundLeafCallback : btDbvt::ICollide
+struct btCompoundCompoundLeafCallback : btDbvt::ICollide
 {
 	int m_numOverlapPairs;
 
-
 	const btCollisionObjectWrapper* m_compound0ColObjWrap;
 	const btCollisionObjectWrapper* m_compound1ColObjWrap;
 	btDispatcher* m_dispatcher;
 	const btDispatcherInfo& m_dispatchInfo;
-	btManifoldResult*	m_resultOut;
-	
-	
-	class btHashedSimplePairCache*	m_childCollisionAlgorithmCache;
-	
-	btPersistentManifold*	m_sharedManifold;
-	
-	btCompoundCompoundLeafCallback (const btCollisionObjectWrapper* compound1ObjWrap,
-									const btCollisionObjectWrapper* compound0ObjWrap,
-									btDispatcher* dispatcher,
-									const btDispatcherInfo& dispatchInfo,
-									btManifoldResult*	resultOut,
-									btHashedSimplePairCache* childAlgorithmsCache,
-									btPersistentManifold*	sharedManifold)
-		:m_numOverlapPairs(0),m_compound0ColObjWrap(compound1ObjWrap),m_compound1ColObjWrap(compound0ObjWrap),m_dispatcher(dispatcher),m_dispatchInfo(dispatchInfo),m_resultOut(resultOut),
-		m_childCollisionAlgorithmCache(childAlgorithmsCache),
-		m_sharedManifold(sharedManifold)
-	{
+	btManifoldResult* m_resultOut;
 
+	class btHashedSimplePairCache* m_childCollisionAlgorithmCache;
+
+	btPersistentManifold* m_sharedManifold;
+
+	btCompoundCompoundLeafCallback(const btCollisionObjectWrapper* compound1ObjWrap,
+								   const btCollisionObjectWrapper* compound0ObjWrap,
+								   btDispatcher* dispatcher,
+								   const btDispatcherInfo& dispatchInfo,
+								   btManifoldResult* resultOut,
+								   btHashedSimplePairCache* childAlgorithmsCache,
+								   btPersistentManifold* sharedManifold)
+		: m_numOverlapPairs(0), m_compound0ColObjWrap(compound1ObjWrap), m_compound1ColObjWrap(compound0ObjWrap), m_dispatcher(dispatcher), m_dispatchInfo(dispatchInfo), m_resultOut(resultOut), m_childCollisionAlgorithmCache(childAlgorithmsCache), m_sharedManifold(sharedManifold)
+	{
 	}
 
-
-
-	
-	void		Process(const btDbvtNode* leaf0,const btDbvtNode* leaf1)
+	void Process(const btDbvtNode* leaf0, const btDbvtNode* leaf1)
 	{
 		BT_PROFILE("btCompoundCompoundLeafCallback::Process");
 		m_numOverlapPairs++;
 
-
 		int childIndex0 = leaf0->dataAsInt;
 		int childIndex1 = leaf1->dataAsInt;
-		
-
-		btAssert(childIndex0>=0);
-		btAssert(childIndex1>=0);
 
+		btAssert(childIndex0 >= 0);
+		btAssert(childIndex1 >= 0);
 
 		const btCompoundShape* compoundShape0 = static_cast<const btCompoundShape*>(m_compound0ColObjWrap->getCollisionShape());
-		btAssert(childIndex0<compoundShape0->getNumChildShapes());
+		btAssert(childIndex0 < compoundShape0->getNumChildShapes());
 
 		const btCompoundShape* compoundShape1 = static_cast<const btCompoundShape*>(m_compound1ColObjWrap->getCollisionShape());
-		btAssert(childIndex1<compoundShape1->getNumChildShapes());
+		btAssert(childIndex1 < compoundShape1->getNumChildShapes());
 
 		const btCollisionShape* childShape0 = compoundShape0->getChildShape(childIndex0);
 		const btCollisionShape* childShape1 = compoundShape1->getChildShape(childIndex1);
 
 		//backup
-		btTransform	orgTrans0 = m_compound0ColObjWrap->getWorldTransform();
+		btTransform orgTrans0 = m_compound0ColObjWrap->getWorldTransform();
 		const btTransform& childTrans0 = compoundShape0->getChildTransform(childIndex0);
-		btTransform	newChildWorldTrans0 = orgTrans0*childTrans0 ;
-		
-		btTransform	orgTrans1 = m_compound1ColObjWrap->getWorldTransform();
+		btTransform newChildWorldTrans0 = orgTrans0 * childTrans0;
+
+		btTransform orgTrans1 = m_compound1ColObjWrap->getWorldTransform();
 		const btTransform& childTrans1 = compoundShape1->getChildTransform(childIndex1);
-		btTransform	newChildWorldTrans1 = orgTrans1*childTrans1 ;
-		
+		btTransform newChildWorldTrans1 = orgTrans1 * childTrans1;
 
 		//perform an AABB check first
-		btVector3 aabbMin0,aabbMax0,aabbMin1,aabbMax1;
-		childShape0->getAabb(newChildWorldTrans0,aabbMin0,aabbMax0);
-		childShape1->getAabb(newChildWorldTrans1,aabbMin1,aabbMax1);
-		
+		btVector3 aabbMin0, aabbMax0, aabbMin1, aabbMax1;
+		childShape0->getAabb(newChildWorldTrans0, aabbMin0, aabbMax0);
+		childShape1->getAabb(newChildWorldTrans1, aabbMin1, aabbMax1);
+
 		btVector3 thresholdVec(m_resultOut->m_closestPointDistanceThreshold, m_resultOut->m_closestPointDistanceThreshold, m_resultOut->m_closestPointDistanceThreshold);
 
 		aabbMin0 -= thresholdVec;
@@ -170,17 +152,16 @@ struct	btCompoundCompoundLeafCallback : btDbvt::ICollide
 
 		if (gCompoundCompoundChildShapePairCallback)
 		{
-			if (!gCompoundCompoundChildShapePairCallback(childShape0,childShape1))
+			if (!gCompoundCompoundChildShapePairCallback(childShape0, childShape1))
 				return;
 		}
 
-		if (TestAabbAgainstAabb2(aabbMin0,aabbMax0,aabbMin1,aabbMax1))
+		if (TestAabbAgainstAabb2(aabbMin0, aabbMax0, aabbMin1, aabbMax1))
 		{
-			btCollisionObjectWrapper compoundWrap0(this->m_compound0ColObjWrap,childShape0, m_compound0ColObjWrap->getCollisionObject(),newChildWorldTrans0,-1,childIndex0);
-			btCollisionObjectWrapper compoundWrap1(this->m_compound1ColObjWrap,childShape1,m_compound1ColObjWrap->getCollisionObject(),newChildWorldTrans1,-1,childIndex1);
-			
+			btCollisionObjectWrapper compoundWrap0(this->m_compound0ColObjWrap, childShape0, m_compound0ColObjWrap->getCollisionObject(), newChildWorldTrans0, -1, childIndex0);
+			btCollisionObjectWrapper compoundWrap1(this->m_compound1ColObjWrap, childShape1, m_compound1ColObjWrap->getCollisionObject(), newChildWorldTrans1, -1, childIndex1);
 
-			btSimplePair* pair = m_childCollisionAlgorithmCache->findPair(childIndex0,childIndex1);
+			btSimplePair* pair = m_childCollisionAlgorithmCache->findPair(childIndex0, childIndex1);
 			bool removePair = false;
 			btCollisionAlgorithm* colAlgo = 0;
 			if (m_resultOut->m_closestPointDistanceThreshold > 0)
@@ -193,7 +174,6 @@ struct	btCompoundCompoundLeafCallback : btDbvt::ICollide
 				if (pair)
 				{
 					colAlgo = (btCollisionAlgorithm*)pair->m_userPointer;
-
 				}
 				else
 				{
@@ -205,7 +185,7 @@ struct	btCompoundCompoundLeafCallback : btDbvt::ICollide
 			}
 
 			btAssert(colAlgo);
-						
+
 			const btCollisionObjectWrapper* tmpWrap0 = 0;
 			const btCollisionObjectWrapper* tmpWrap1 = 0;
 
@@ -215,105 +195,100 @@ struct	btCompoundCompoundLeafCallback : btDbvt::ICollide
 			m_resultOut->setBody0Wrap(&compoundWrap0);
 			m_resultOut->setBody1Wrap(&compoundWrap1);
 
-			m_resultOut->setShapeIdentifiersA(-1,childIndex0);
-			m_resultOut->setShapeIdentifiersB(-1,childIndex1);
+			m_resultOut->setShapeIdentifiersA(-1, childIndex0);
+			m_resultOut->setShapeIdentifiersB(-1, childIndex1);
 
+			colAlgo->processCollision(&compoundWrap0, &compoundWrap1, m_dispatchInfo, m_resultOut);
 
-			colAlgo->processCollision(&compoundWrap0,&compoundWrap1,m_dispatchInfo,m_resultOut);
-			
 			m_resultOut->setBody0Wrap(tmpWrap0);
 			m_resultOut->setBody1Wrap(tmpWrap1);
-			
+
 			if (removePair)
 			{
 				colAlgo->~btCollisionAlgorithm();
 				m_dispatcher->freeCollisionAlgorithm(colAlgo);
 			}
-
 		}
 	}
 };
 
-
-static DBVT_INLINE bool		MyIntersect(	const btDbvtAabbMm& a,
-								  const btDbvtAabbMm& b, const btTransform& xform, btScalar distanceThreshold)
+static DBVT_INLINE bool MyIntersect(const btDbvtAabbMm& a,
+									const btDbvtAabbMm& b, const btTransform& xform, btScalar distanceThreshold)
 {
-	btVector3 newmin,newmax;
-	btTransformAabb(b.Mins(),b.Maxs(),0.f,xform,newmin,newmax);
+	btVector3 newmin, newmax;
+	btTransformAabb(b.Mins(), b.Maxs(), 0.f, xform, newmin, newmax);
 	newmin -= btVector3(distanceThreshold, distanceThreshold, distanceThreshold);
 	newmax += btVector3(distanceThreshold, distanceThreshold, distanceThreshold);
-	btDbvtAabbMm newb = btDbvtAabbMm::FromMM(newmin,newmax);
-	return Intersect(a,newb);
+	btDbvtAabbMm newb = btDbvtAabbMm::FromMM(newmin, newmax);
+	return Intersect(a, newb);
 }
 
-
-static inline void		MycollideTT(	const btDbvtNode* root0,
-								  const btDbvtNode* root1,
-								  const btTransform& xform,
-								  btCompoundCompoundLeafCallback* callback, btScalar distanceThreshold)
+static inline void MycollideTT(const btDbvtNode* root0,
+							   const btDbvtNode* root1,
+							   const btTransform& xform,
+							   btCompoundCompoundLeafCallback* callback, btScalar distanceThreshold)
 {
-
-		if(root0&&root1)
-		{
-			int								depth=1;
-			int								treshold=btDbvt::DOUBLE_STACKSIZE-4;
-			btAlignedObjectArray<btDbvt::sStkNN>	stkStack;
+	if (root0 && root1)
+	{
+		int depth = 1;
+		int treshold = btDbvt::DOUBLE_STACKSIZE - 4;
+		btAlignedObjectArray<btDbvt::sStkNN> stkStack;
 #ifdef USE_LOCAL_STACK
-			ATTRIBUTE_ALIGNED16(btDbvt::sStkNN localStack[btDbvt::DOUBLE_STACKSIZE]);
-			stkStack.initializeFromBuffer(&localStack,btDbvt::DOUBLE_STACKSIZE,btDbvt::DOUBLE_STACKSIZE);
+		ATTRIBUTE_ALIGNED16(btDbvt::sStkNN localStack[btDbvt::DOUBLE_STACKSIZE]);
+		stkStack.initializeFromBuffer(&localStack, btDbvt::DOUBLE_STACKSIZE, btDbvt::DOUBLE_STACKSIZE);
 #else
-			stkStack.resize(btDbvt::DOUBLE_STACKSIZE);
+		stkStack.resize(btDbvt::DOUBLE_STACKSIZE);
 #endif
-			stkStack[0]=btDbvt::sStkNN(root0,root1);
-			do	{
-				btDbvt::sStkNN	p=stkStack[--depth];
-				if(MyIntersect(p.a->volume,p.b->volume,xform, distanceThreshold))
+		stkStack[0] = btDbvt::sStkNN(root0, root1);
+		do
+		{
+			btDbvt::sStkNN p = stkStack[--depth];
+			if (MyIntersect(p.a->volume, p.b->volume, xform, distanceThreshold))
+			{
+				if (depth > treshold)
 				{
-					if(depth>treshold)
+					stkStack.resize(stkStack.size() * 2);
+					treshold = stkStack.size() - 4;
+				}
+				if (p.a->isinternal())
+				{
+					if (p.b->isinternal())
 					{
-						stkStack.resize(stkStack.size()*2);
-						treshold=stkStack.size()-4;
-					}
-					if(p.a->isinternal())
-					{
-						if(p.b->isinternal())
-						{					
-							stkStack[depth++]=btDbvt::sStkNN(p.a->childs[0],p.b->childs[0]);
-							stkStack[depth++]=btDbvt::sStkNN(p.a->childs[1],p.b->childs[0]);
-							stkStack[depth++]=btDbvt::sStkNN(p.a->childs[0],p.b->childs[1]);
-							stkStack[depth++]=btDbvt::sStkNN(p.a->childs[1],p.b->childs[1]);
-						}
-						else
-						{
-							stkStack[depth++]=btDbvt::sStkNN(p.a->childs[0],p.b);
-							stkStack[depth++]=btDbvt::sStkNN(p.a->childs[1],p.b);
-						}
+						stkStack[depth++] = btDbvt::sStkNN(p.a->childs[0], p.b->childs[0]);
+						stkStack[depth++] = btDbvt::sStkNN(p.a->childs[1], p.b->childs[0]);
+						stkStack[depth++] = btDbvt::sStkNN(p.a->childs[0], p.b->childs[1]);
+						stkStack[depth++] = btDbvt::sStkNN(p.a->childs[1], p.b->childs[1]);
 					}
 					else
 					{
-						if(p.b->isinternal())
-						{
-							stkStack[depth++]=btDbvt::sStkNN(p.a,p.b->childs[0]);
-							stkStack[depth++]=btDbvt::sStkNN(p.a,p.b->childs[1]);
-						}
-						else
-						{
-							callback->Process(p.a,p.b);
-						}
+						stkStack[depth++] = btDbvt::sStkNN(p.a->childs[0], p.b);
+						stkStack[depth++] = btDbvt::sStkNN(p.a->childs[1], p.b);
 					}
 				}
-			} while(depth);
-		}
+				else
+				{
+					if (p.b->isinternal())
+					{
+						stkStack[depth++] = btDbvt::sStkNN(p.a, p.b->childs[0]);
+						stkStack[depth++] = btDbvt::sStkNN(p.a, p.b->childs[1]);
+					}
+					else
+					{
+						callback->Process(p.a, p.b);
+					}
+				}
+			}
+		} while (depth);
+	}
 }
 
-void btCompoundCompoundCollisionAlgorithm::processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+void btCompoundCompoundCollisionAlgorithm::processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
-
 	const btCollisionObjectWrapper* col0ObjWrap = body0Wrap;
-	const btCollisionObjectWrapper* col1ObjWrap= body1Wrap;
+	const btCollisionObjectWrapper* col1ObjWrap = body1Wrap;
 
-	btAssert (col0ObjWrap->getCollisionShape()->isCompound());
-	btAssert (col1ObjWrap->getCollisionShape()->isCompound());
+	btAssert(col0ObjWrap->getCollisionShape()->isCompound());
+	btAssert(col1ObjWrap->getCollisionShape()->isCompound());
 	const btCompoundShape* compoundShape0 = static_cast<const btCompoundShape*>(col0ObjWrap->getCollisionShape());
 	const btCompoundShape* compoundShape1 = static_cast<const btCompoundShape*>(col1ObjWrap->getCollisionShape());
 
@@ -321,7 +296,7 @@ void btCompoundCompoundCollisionAlgorithm::processCollision (const btCollisionOb
 	const btDbvt* tree1 = compoundShape1->getDynamicAabbTree();
 	if (!tree0 || !tree1)
 	{
-		return btCompoundCollisionAlgorithm::processCollision(body0Wrap,body1Wrap,dispatchInfo,resultOut);
+		return btCompoundCollisionAlgorithm::processCollision(body0Wrap, body1Wrap, dispatchInfo, resultOut);
 	}
 	///btCompoundShape might have changed:
 	////make sure the internal child collision algorithm caches are still valid
@@ -331,28 +306,26 @@ void btCompoundCompoundCollisionAlgorithm::processCollision (const btCollisionOb
 		removeChildAlgorithms();
 		m_compoundShapeRevision0 = compoundShape0->getUpdateRevision();
 		m_compoundShapeRevision1 = compoundShape1->getUpdateRevision();
-
 	}
 
-
 	///we need to refresh all contact manifolds
 	///note that we should actually recursively traverse all children, btCompoundShape can nested more then 1 level deep
 	///so we should add a 'refreshManifolds' in the btCollisionAlgorithm
 	{
 		int i;
 		btManifoldArray manifoldArray;
-#ifdef USE_LOCAL_STACK 
+#ifdef USE_LOCAL_STACK
 		btPersistentManifold localManifolds[4];
-		manifoldArray.initializeFromBuffer(&localManifolds,0,4);
+		manifoldArray.initializeFromBuffer(&localManifolds, 0, 4);
 #endif
 		btSimplePairArray& pairs = m_childCollisionAlgorithmCache->getOverlappingPairArray();
-		for (i=0;i<pairs.size();i++)
+		for (i = 0; i < pairs.size(); i++)
 		{
 			if (pairs[i].m_userPointer)
 			{
-				btCollisionAlgorithm* algo = (btCollisionAlgorithm*) pairs[i].m_userPointer;
+				btCollisionAlgorithm* algo = (btCollisionAlgorithm*)pairs[i].m_userPointer;
 				algo->getAllContactManifolds(manifoldArray);
-				for (int m=0;m<manifoldArray.size();m++)
+				for (int m = 0; m < manifoldArray.size(); m++)
 				{
 					if (manifoldArray[m]->getNumContacts())
 					{
@@ -366,35 +339,27 @@ void btCompoundCompoundCollisionAlgorithm::processCollision (const btCollisionOb
 		}
 	}
 
+	btCompoundCompoundLeafCallback callback(col0ObjWrap, col1ObjWrap, this->m_dispatcher, dispatchInfo, resultOut, this->m_childCollisionAlgorithmCache, m_sharedManifold);
 
-	
-
-	btCompoundCompoundLeafCallback callback(col0ObjWrap,col1ObjWrap,this->m_dispatcher,dispatchInfo,resultOut,this->m_childCollisionAlgorithmCache,m_sharedManifold);
-
-
-	const btTransform	xform=col0ObjWrap->getWorldTransform().inverse()*col1ObjWrap->getWorldTransform();
-	MycollideTT(tree0->m_root,tree1->m_root,xform,&callback, resultOut->m_closestPointDistanceThreshold);
+	const btTransform xform = col0ObjWrap->getWorldTransform().inverse() * col1ObjWrap->getWorldTransform();
+	MycollideTT(tree0->m_root, tree1->m_root, xform, &callback, resultOut->m_closestPointDistanceThreshold);
 
 	//printf("#compound-compound child/leaf overlap =%d                      \r",callback.m_numOverlapPairs);
 
 	//remove non-overlapping child pairs
 
 	{
-		btAssert(m_removePairs.size()==0);
+		btAssert(m_removePairs.size() == 0);
 
 		//iterate over all children, perform an AABB check inside ProcessChildShape
 		btSimplePairArray& pairs = m_childCollisionAlgorithmCache->getOverlappingPairArray();
-		
-		int i;
-		btManifoldArray	manifoldArray;
-        
-		
 
-        
-        
-        btVector3 aabbMin0,aabbMax0,aabbMin1,aabbMax1;        
-        
-		for (i=0;i<pairs.size();i++)
+		int i;
+		btManifoldArray manifoldArray;
+
+		btVector3 aabbMin0, aabbMax0, aabbMin1, aabbMax1;
+
+		for (i = 0; i < pairs.size(); i++)
 		{
 			if (pairs[i].m_userPointer)
 			{
@@ -402,52 +367,47 @@ void btCompoundCompoundCollisionAlgorithm::processCollision (const btCollisionOb
 
 				{
 					const btCollisionShape* childShape0 = 0;
-					
-					btTransform	newChildWorldTrans0;
+
+					btTransform newChildWorldTrans0;
 					childShape0 = compoundShape0->getChildShape(pairs[i].m_indexA);
 					const btTransform& childTrans0 = compoundShape0->getChildTransform(pairs[i].m_indexA);
-					newChildWorldTrans0 = col0ObjWrap->getWorldTransform()*childTrans0 ;
-					childShape0->getAabb(newChildWorldTrans0,aabbMin0,aabbMax0);
+					newChildWorldTrans0 = col0ObjWrap->getWorldTransform() * childTrans0;
+					childShape0->getAabb(newChildWorldTrans0, aabbMin0, aabbMax0);
 				}
 				btVector3 thresholdVec(resultOut->m_closestPointDistanceThreshold, resultOut->m_closestPointDistanceThreshold, resultOut->m_closestPointDistanceThreshold);
 				aabbMin0 -= thresholdVec;
 				aabbMax0 += thresholdVec;
 				{
 					const btCollisionShape* childShape1 = 0;
-					btTransform	newChildWorldTrans1;
+					btTransform newChildWorldTrans1;
 
 					childShape1 = compoundShape1->getChildShape(pairs[i].m_indexB);
 					const btTransform& childTrans1 = compoundShape1->getChildTransform(pairs[i].m_indexB);
-					newChildWorldTrans1 = col1ObjWrap->getWorldTransform()*childTrans1 ;
-					childShape1->getAabb(newChildWorldTrans1,aabbMin1,aabbMax1);
+					newChildWorldTrans1 = col1ObjWrap->getWorldTransform() * childTrans1;
+					childShape1->getAabb(newChildWorldTrans1, aabbMin1, aabbMax1);
 				}
-				
+
 				aabbMin1 -= thresholdVec;
 				aabbMax1 += thresholdVec;
 
-				if (!TestAabbAgainstAabb2(aabbMin0,aabbMax0,aabbMin1,aabbMax1))
+				if (!TestAabbAgainstAabb2(aabbMin0, aabbMax0, aabbMin1, aabbMax1))
 				{
 					algo->~btCollisionAlgorithm();
 					m_dispatcher->freeCollisionAlgorithm(algo);
-					m_removePairs.push_back(btSimplePair(pairs[i].m_indexA,pairs[i].m_indexB));
+					m_removePairs.push_back(btSimplePair(pairs[i].m_indexA, pairs[i].m_indexB));
 				}
 			}
 		}
-		for (int i=0;i<m_removePairs.size();i++)
+		for (int i = 0; i < m_removePairs.size(); i++)
 		{
-			m_childCollisionAlgorithmCache->removeOverlappingPair(m_removePairs[i].m_indexA,m_removePairs[i].m_indexB);
+			m_childCollisionAlgorithmCache->removeOverlappingPair(m_removePairs[i].m_indexA, m_removePairs[i].m_indexB);
 		}
 		m_removePairs.clear();
 	}
-
 }
 
-btScalar	btCompoundCompoundCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+btScalar btCompoundCompoundCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	btAssert(0);
 	return 0.f;
-
 }
-
-
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCompoundCompoundCollisionAlgorithm.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCompoundCompoundCollisionAlgorithm.h
index f29f7a709a4..a940d840e0c 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btCompoundCompoundCollisionAlgorithm.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btCompoundCompoundCollisionAlgorithm.h
@@ -34,54 +34,49 @@ class btCollisionObject;
 
 class btCollisionShape;
 
-/// btCompoundCompoundCollisionAlgorithm  supports collision between two btCompoundCollisionShape shapes
-class btCompoundCompoundCollisionAlgorithm  : public btCompoundCollisionAlgorithm
-{
+extern btShapePairCallback gCompoundCompoundChildShapePairCallback;
 
-	class btHashedSimplePairCache*	m_childCollisionAlgorithmCache;
+/// btCompoundCompoundCollisionAlgorithm  supports collision between two btCompoundCollisionShape shapes
+class btCompoundCompoundCollisionAlgorithm : public btCompoundCollisionAlgorithm
+{
+	class btHashedSimplePairCache* m_childCollisionAlgorithmCache;
 	btSimplePairArray m_removePairs;
 
+	int m_compoundShapeRevision0;  //to keep track of changes, so that childAlgorithm array can be updated
+	int m_compoundShapeRevision1;
 
-	int	m_compoundShapeRevision0;//to keep track of changes, so that childAlgorithm array can be updated
-	int	m_compoundShapeRevision1;
-	
-	void	removeChildAlgorithms();
-	
-//	void	preallocateChildAlgorithms(const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap);
+	void removeChildAlgorithms();
+
+	//	void	preallocateChildAlgorithms(const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap);
 
 public:
-
-	btCompoundCompoundCollisionAlgorithm( const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,bool isSwapped);
+	btCompoundCompoundCollisionAlgorithm(const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, bool isSwapped);
 
 	virtual ~btCompoundCompoundCollisionAlgorithm();
 
-	
+	virtual void processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual void processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	btScalar calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	btScalar	calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual void getAllContactManifolds(btManifoldArray& manifoldArray);
 
-	virtual	void	getAllContactManifolds(btManifoldArray&	manifoldArray);
-	
-	
-	struct CreateFunc :public 	btCollisionAlgorithmCreateFunc
+	struct CreateFunc : public btCollisionAlgorithmCreateFunc
 	{
-		virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+		virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 		{
 			void* mem = ci.m_dispatcher1->allocateCollisionAlgorithm(sizeof(btCompoundCompoundCollisionAlgorithm));
-			return new(mem) btCompoundCompoundCollisionAlgorithm(ci,body0Wrap,body1Wrap,false);
+			return new (mem) btCompoundCompoundCollisionAlgorithm(ci, body0Wrap, body1Wrap, false);
 		}
 	};
 
-	struct SwappedCreateFunc :public 	btCollisionAlgorithmCreateFunc
+	struct SwappedCreateFunc : public btCollisionAlgorithmCreateFunc
 	{
-		virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+		virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 		{
 			void* mem = ci.m_dispatcher1->allocateCollisionAlgorithm(sizeof(btCompoundCompoundCollisionAlgorithm));
-			return new(mem) btCompoundCompoundCollisionAlgorithm(ci,body0Wrap,body1Wrap,true);
+			return new (mem) btCompoundCompoundCollisionAlgorithm(ci, body0Wrap, body1Wrap, true);
 		}
 	};
-
 };
 
-#endif //BT_COMPOUND_COMPOUND_COLLISION_ALGORITHM_H
+#endif  //BT_COMPOUND_COMPOUND_COLLISION_ALGORITHM_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvex2dConvex2dAlgorithm.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvex2dConvex2dAlgorithm.cpp
index 1cb3d2e7a14..9087f843985 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvex2dConvex2dAlgorithm.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvex2dConvex2dAlgorithm.cpp
@@ -22,7 +22,6 @@ subject to the following restrictions:
 #include "BulletCollision/CollisionShapes/btConvexShape.h"
 #include "BulletCollision/CollisionShapes/btCapsuleShape.h"
 
-
 #include "BulletCollision/NarrowPhaseCollision/btGjkPairDetector.h"
 #include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h"
 #include "BulletCollision/CollisionDispatch/btCollisionDispatcher.h"
@@ -34,8 +33,6 @@ subject to the following restrictions:
 #include "BulletCollision/NarrowPhaseCollision/btSubSimplexConvexCast.h"
 #include "BulletCollision/NarrowPhaseCollision/btGjkConvexCast.h"
 
-
-
 #include "BulletCollision/NarrowPhaseCollision/btVoronoiSimplexSolver.h"
 #include "BulletCollision/CollisionShapes/btSphereShape.h"
 
@@ -45,31 +42,28 @@ subject to the following restrictions:
 #include "BulletCollision/NarrowPhaseCollision/btGjkEpaPenetrationDepthSolver.h"
 #include "BulletCollision/CollisionDispatch/btCollisionObjectWrapper.h"
 
-btConvex2dConvex2dAlgorithm::CreateFunc::CreateFunc(btSimplexSolverInterface*			simplexSolver, btConvexPenetrationDepthSolver* pdSolver)
+btConvex2dConvex2dAlgorithm::CreateFunc::CreateFunc(btSimplexSolverInterface* simplexSolver, btConvexPenetrationDepthSolver* pdSolver)
 {
 	m_simplexSolver = simplexSolver;
 	m_pdSolver = pdSolver;
 }
 
-btConvex2dConvex2dAlgorithm::CreateFunc::~CreateFunc() 
-{ 
+btConvex2dConvex2dAlgorithm::CreateFunc::~CreateFunc()
+{
 }
 
-btConvex2dConvex2dAlgorithm::btConvex2dConvex2dAlgorithm(btPersistentManifold* mf,const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,btSimplexSolverInterface* simplexSolver, btConvexPenetrationDepthSolver* pdSolver,int /* numPerturbationIterations */, int /* minimumPointsPerturbationThreshold */)
-: btActivatingCollisionAlgorithm(ci,body0Wrap,body1Wrap),
-m_simplexSolver(simplexSolver),
-m_pdSolver(pdSolver),
-m_ownManifold (false),
-m_manifoldPtr(mf),
-m_lowLevelOfDetail(false)
+btConvex2dConvex2dAlgorithm::btConvex2dConvex2dAlgorithm(btPersistentManifold* mf, const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, btSimplexSolverInterface* simplexSolver, btConvexPenetrationDepthSolver* pdSolver, int /* numPerturbationIterations */, int /* minimumPointsPerturbationThreshold */)
+	: btActivatingCollisionAlgorithm(ci, body0Wrap, body1Wrap),
+	  m_simplexSolver(simplexSolver),
+	  m_pdSolver(pdSolver),
+	  m_ownManifold(false),
+	  m_manifoldPtr(mf),
+	  m_lowLevelOfDetail(false)
 {
 	(void)body0Wrap;
 	(void)body1Wrap;
 }
 
-
-
-
 btConvex2dConvex2dAlgorithm::~btConvex2dConvex2dAlgorithm()
 {
 	if (m_ownManifold)
@@ -79,26 +73,22 @@ btConvex2dConvex2dAlgorithm::~btConvex2dConvex2dAlgorithm()
 	}
 }
 
-void	btConvex2dConvex2dAlgorithm ::setLowLevelOfDetail(bool useLowLevel)
+void btConvex2dConvex2dAlgorithm ::setLowLevelOfDetail(bool useLowLevel)
 {
 	m_lowLevelOfDetail = useLowLevel;
 }
 
-
-
 extern btScalar gContactBreakingThreshold;
 
-
 //
 // Convex-Convex collision algorithm
 //
-void btConvex2dConvex2dAlgorithm ::processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+void btConvex2dConvex2dAlgorithm ::processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
-
 	if (!m_manifoldPtr)
 	{
 		//swapped?
-		m_manifoldPtr = m_dispatcher->getNewManifold(body0Wrap->getCollisionObject(),body1Wrap->getCollisionObject());
+		m_manifoldPtr = m_dispatcher->getNewManifold(body0Wrap->getCollisionObject(), body1Wrap->getCollisionObject());
 		m_ownManifold = true;
 	}
 	resultOut->setPersistentManifold(m_manifoldPtr);
@@ -106,49 +96,41 @@ void btConvex2dConvex2dAlgorithm ::processCollision (const btCollisionObjectWrap
 	//comment-out next line to test multi-contact generation
 	//resultOut->getPersistentManifold()->clearManifold();
 
-
 	const btConvexShape* min0 = static_cast<const btConvexShape*>(body0Wrap->getCollisionShape());
 	const btConvexShape* min1 = static_cast<const btConvexShape*>(body1Wrap->getCollisionShape());
 
-	btVector3  normalOnB;
-	btVector3  pointOnBWorld;
+	btVector3 normalOnB;
+	btVector3 pointOnBWorld;
 
 	{
-
-
 		btGjkPairDetector::ClosestPointInput input;
 
-		btGjkPairDetector	gjkPairDetector(min0,min1,m_simplexSolver,m_pdSolver);
+		btGjkPairDetector gjkPairDetector(min0, min1, m_simplexSolver, m_pdSolver);
 		//TODO: if (dispatchInfo.m_useContinuous)
 		gjkPairDetector.setMinkowskiA(min0);
 		gjkPairDetector.setMinkowskiB(min1);
 
 		{
 			input.m_maximumDistanceSquared = min0->getMargin() + min1->getMargin() + m_manifoldPtr->getContactBreakingThreshold();
-			input.m_maximumDistanceSquared*= input.m_maximumDistanceSquared;
+			input.m_maximumDistanceSquared *= input.m_maximumDistanceSquared;
 		}
 
 		input.m_transformA = body0Wrap->getWorldTransform();
 		input.m_transformB = body1Wrap->getWorldTransform();
 
-		gjkPairDetector.getClosestPoints(input,*resultOut,dispatchInfo.m_debugDraw);
+		gjkPairDetector.getClosestPoints(input, *resultOut, dispatchInfo.m_debugDraw);
 
-		btVector3 v0,v1;
+		btVector3 v0, v1;
 		btVector3 sepNormalWorldSpace;
-
 	}
 
 	if (m_ownManifold)
 	{
 		resultOut->refreshContactPoints();
 	}
-
 }
 
-
-
-
-btScalar	btConvex2dConvex2dAlgorithm::calculateTimeOfImpact(btCollisionObject* col0,btCollisionObject* col1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+btScalar btConvex2dConvex2dAlgorithm::calculateTimeOfImpact(btCollisionObject* col0, btCollisionObject* col1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	(void)resultOut;
 	(void)dispatchInfo;
@@ -158,7 +140,6 @@ btScalar	btConvex2dConvex2dAlgorithm::calculateTimeOfImpact(btCollisionObject* c
 	///col0->m_worldTransform,
 	btScalar resultFraction = btScalar(1.);
 
-
 	btScalar squareMot0 = (col0->getInterpolationWorldTransform().getOrigin() - col0->getWorldTransform().getOrigin()).length2();
 	btScalar squareMot1 = (col1->getInterpolationWorldTransform().getOrigin() - col1->getWorldTransform().getOrigin()).length2();
 
@@ -166,77 +147,65 @@ btScalar	btConvex2dConvex2dAlgorithm::calculateTimeOfImpact(btCollisionObject* c
 		squareMot1 < col1->getCcdSquareMotionThreshold())
 		return resultFraction;
 
-
 	//An adhoc way of testing the Continuous Collision Detection algorithms
 	//One object is approximated as a sphere, to simplify things
 	//Starting in penetration should report no time of impact
 	//For proper CCD, better accuracy and handling of 'allowed' penetration should be added
 	//also the mainloop of the physics should have a kind of toi queue (something like Brian Mirtich's application of Timewarp for Rigidbodies)
 
-
 	/// Convex0 against sphere for Convex1
 	{
 		btConvexShape* convex0 = static_cast<btConvexShape*>(col0->getCollisionShape());
 
-		btSphereShape	sphere1(col1->getCcdSweptSphereRadius()); //todo: allow non-zero sphere sizes, for better approximation
+		btSphereShape sphere1(col1->getCcdSweptSphereRadius());  //todo: allow non-zero sphere sizes, for better approximation
 		btConvexCast::CastResult result;
 		btVoronoiSimplexSolver voronoiSimplex;
 		//SubsimplexConvexCast ccd0(&sphere,min0,&voronoiSimplex);
 		///Simplification, one object is simplified as a sphere
-		btGjkConvexCast ccd1( convex0 ,&sphere1,&voronoiSimplex);
+		btGjkConvexCast ccd1(convex0, &sphere1, &voronoiSimplex);
 		//ContinuousConvexCollision ccd(min0,min1,&voronoiSimplex,0);
-		if (ccd1.calcTimeOfImpact(col0->getWorldTransform(),col0->getInterpolationWorldTransform(),
-			col1->getWorldTransform(),col1->getInterpolationWorldTransform(),result))
+		if (ccd1.calcTimeOfImpact(col0->getWorldTransform(), col0->getInterpolationWorldTransform(),
+								  col1->getWorldTransform(), col1->getInterpolationWorldTransform(), result))
 		{
-
 			//store result.m_fraction in both bodies
 
-			if (col0->getHitFraction()> result.m_fraction)
-				col0->setHitFraction( result.m_fraction );
+			if (col0->getHitFraction() > result.m_fraction)
+				col0->setHitFraction(result.m_fraction);
 
 			if (col1->getHitFraction() > result.m_fraction)
-				col1->setHitFraction( result.m_fraction);
+				col1->setHitFraction(result.m_fraction);
 
 			if (resultFraction > result.m_fraction)
 				resultFraction = result.m_fraction;
-
 		}
-
-
-
-
 	}
 
 	/// Sphere (for convex0) against Convex1
 	{
 		btConvexShape* convex1 = static_cast<btConvexShape*>(col1->getCollisionShape());
 
-		btSphereShape	sphere0(col0->getCcdSweptSphereRadius()); //todo: allow non-zero sphere sizes, for better approximation
+		btSphereShape sphere0(col0->getCcdSweptSphereRadius());  //todo: allow non-zero sphere sizes, for better approximation
 		btConvexCast::CastResult result;
 		btVoronoiSimplexSolver voronoiSimplex;
 		//SubsimplexConvexCast ccd0(&sphere,min0,&voronoiSimplex);
 		///Simplification, one object is simplified as a sphere
-		btGjkConvexCast ccd1(&sphere0,convex1,&voronoiSimplex);
+		btGjkConvexCast ccd1(&sphere0, convex1, &voronoiSimplex);
 		//ContinuousConvexCollision ccd(min0,min1,&voronoiSimplex,0);
-		if (ccd1.calcTimeOfImpact(col0->getWorldTransform(),col0->getInterpolationWorldTransform(),
-			col1->getWorldTransform(),col1->getInterpolationWorldTransform(),result))
+		if (ccd1.calcTimeOfImpact(col0->getWorldTransform(), col0->getInterpolationWorldTransform(),
+								  col1->getWorldTransform(), col1->getInterpolationWorldTransform(), result))
 		{
-
 			//store result.m_fraction in both bodies
 
-			if (col0->getHitFraction()	> result.m_fraction)
-				col0->setHitFraction( result.m_fraction);
+			if (col0->getHitFraction() > result.m_fraction)
+				col0->setHitFraction(result.m_fraction);
 
 			if (col1->getHitFraction() > result.m_fraction)
-				col1->setHitFraction( result.m_fraction);
+				col1->setHitFraction(result.m_fraction);
 
 			if (resultFraction > result.m_fraction)
 				resultFraction = result.m_fraction;
-
 		}
 	}
 
 	return resultFraction;
-
 }
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvex2dConvex2dAlgorithm.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvex2dConvex2dAlgorithm.h
index 24d13367786..9fca463fbe4 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvex2dConvex2dAlgorithm.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvex2dConvex2dAlgorithm.h
@@ -23,70 +23,61 @@ subject to the following restrictions:
 #include "BulletCollision/NarrowPhaseCollision/btVoronoiSimplexSolver.h"
 #include "BulletCollision/CollisionDispatch/btCollisionCreateFunc.h"
 #include "BulletCollision/CollisionDispatch/btCollisionDispatcher.h"
-#include "LinearMath/btTransformUtil.h" //for btConvexSeparatingDistanceUtil
+#include "LinearMath/btTransformUtil.h"  //for btConvexSeparatingDistanceUtil
 
 class btConvexPenetrationDepthSolver;
 
-
 ///The convex2dConvex2dAlgorithm collision algorithm support 2d collision detection for btConvex2dShape
 ///Currently it requires the btMinkowskiPenetrationDepthSolver, it has support for 2d penetration depth computation
 class btConvex2dConvex2dAlgorithm : public btActivatingCollisionAlgorithm
 {
-	btSimplexSolverInterface*		m_simplexSolver;
+	btSimplexSolverInterface* m_simplexSolver;
 	btConvexPenetrationDepthSolver* m_pdSolver;
 
-	
-	bool	m_ownManifold;
-	btPersistentManifold*	m_manifoldPtr;
-	bool			m_lowLevelOfDetail;
-	
+	bool m_ownManifold;
+	btPersistentManifold* m_manifoldPtr;
+	bool m_lowLevelOfDetail;
+
 public:
-
-	btConvex2dConvex2dAlgorithm(btPersistentManifold* mf,const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap, btSimplexSolverInterface* simplexSolver, btConvexPenetrationDepthSolver* pdSolver, int numPerturbationIterations, int minimumPointsPerturbationThreshold);
-
+	btConvex2dConvex2dAlgorithm(btPersistentManifold* mf, const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, btSimplexSolverInterface* simplexSolver, btConvexPenetrationDepthSolver* pdSolver, int numPerturbationIterations, int minimumPointsPerturbationThreshold);
 
 	virtual ~btConvex2dConvex2dAlgorithm();
 
-	virtual void processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual void processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual	void	getAllContactManifolds(btManifoldArray&	manifoldArray)
+	virtual void getAllContactManifolds(btManifoldArray& manifoldArray)
 	{
 		///should we use m_ownManifold to avoid adding duplicates?
 		if (m_manifoldPtr && m_ownManifold)
 			manifoldArray.push_back(m_manifoldPtr);
 	}
 
+	void setLowLevelOfDetail(bool useLowLevel);
 
-	void	setLowLevelOfDetail(bool useLowLevel);
-
-
-	const btPersistentManifold*	getManifold()
+	const btPersistentManifold* getManifold()
 	{
 		return m_manifoldPtr;
 	}
 
-	struct CreateFunc :public 	btCollisionAlgorithmCreateFunc
+	struct CreateFunc : public btCollisionAlgorithmCreateFunc
 	{
-
-		btConvexPenetrationDepthSolver*		m_pdSolver;
-		btSimplexSolverInterface*			m_simplexSolver;
+		btConvexPenetrationDepthSolver* m_pdSolver;
+		btSimplexSolverInterface* m_simplexSolver;
 		int m_numPerturbationIterations;
 		int m_minimumPointsPerturbationThreshold;
 
-		CreateFunc(btSimplexSolverInterface*			simplexSolver, btConvexPenetrationDepthSolver* pdSolver);
-		
+		CreateFunc(btSimplexSolverInterface* simplexSolver, btConvexPenetrationDepthSolver* pdSolver);
+
 		virtual ~CreateFunc();
 
-		virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+		virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 		{
 			void* mem = ci.m_dispatcher1->allocateCollisionAlgorithm(sizeof(btConvex2dConvex2dAlgorithm));
-			return new(mem) btConvex2dConvex2dAlgorithm(ci.m_manifold,ci,body0Wrap,body1Wrap,m_simplexSolver,m_pdSolver,m_numPerturbationIterations,m_minimumPointsPerturbationThreshold);
+			return new (mem) btConvex2dConvex2dAlgorithm(ci.m_manifold, ci, body0Wrap, body1Wrap, m_simplexSolver, m_pdSolver, m_numPerturbationIterations, m_minimumPointsPerturbationThreshold);
 		}
 	};
-
-
 };
 
-#endif //BT_CONVEX_2D_CONVEX_2D_ALGORITHM_H
+#endif  //BT_CONVEX_2D_CONVEX_2D_ALGORITHM_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexConcaveCollisionAlgorithm.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexConcaveCollisionAlgorithm.cpp
index d8cbe961429..e50f85e2bbe 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexConcaveCollisionAlgorithm.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexConcaveCollisionAlgorithm.cpp
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btConvexConcaveCollisionAlgorithm.h"
 #include "LinearMath/btQuickprof.h"
 #include "BulletCollision/CollisionDispatch/btCollisionObject.h"
@@ -29,10 +28,10 @@ subject to the following restrictions:
 #include "BulletCollision/CollisionDispatch/btCollisionObjectWrapper.h"
 #include "BulletCollision/CollisionShapes/btSdfCollisionShape.h"
 
-btConvexConcaveCollisionAlgorithm::btConvexConcaveCollisionAlgorithm( const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,bool isSwapped)
-: btActivatingCollisionAlgorithm(ci,body0Wrap,body1Wrap),
-m_btConvexTriangleCallback(ci.m_dispatcher1,body0Wrap,body1Wrap,isSwapped),
-m_isSwapped(isSwapped)
+btConvexConcaveCollisionAlgorithm::btConvexConcaveCollisionAlgorithm(const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, bool isSwapped)
+	: btActivatingCollisionAlgorithm(ci, body0Wrap, body1Wrap),
+	  m_btConvexTriangleCallback(ci.m_dispatcher1, body0Wrap, body1Wrap, isSwapped),
+	  m_isSwapped(isSwapped)
 {
 }
 
@@ -40,7 +39,7 @@ btConvexConcaveCollisionAlgorithm::~btConvexConcaveCollisionAlgorithm()
 {
 }
 
-void	btConvexConcaveCollisionAlgorithm::getAllContactManifolds(btManifoldArray&	manifoldArray)
+void btConvexConcaveCollisionAlgorithm::getAllContactManifolds(btManifoldArray& manifoldArray)
 {
 	if (m_btConvexTriangleCallback.m_manifoldPtr)
 	{
@@ -48,38 +47,32 @@ void	btConvexConcaveCollisionAlgorithm::getAllContactManifolds(btManifoldArray&
 	}
 }
 
-
-btConvexTriangleCallback::btConvexTriangleCallback(btDispatcher*  dispatcher,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,bool isSwapped):
-	  m_dispatcher(dispatcher),
-	m_dispatchInfoPtr(0)
+btConvexTriangleCallback::btConvexTriangleCallback(btDispatcher* dispatcher, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, bool isSwapped) : m_dispatcher(dispatcher),
+																																													 m_dispatchInfoPtr(0)
 {
-	m_convexBodyWrap = isSwapped? body1Wrap:body0Wrap;
-	m_triBodyWrap = isSwapped? body0Wrap:body1Wrap;
-	
-	  //
-	  // create the manifold from the dispatcher 'manifold pool'
-	  //
-	  m_manifoldPtr = m_dispatcher->getNewManifold(m_convexBodyWrap->getCollisionObject(),m_triBodyWrap->getCollisionObject());
+	m_convexBodyWrap = isSwapped ? body1Wrap : body0Wrap;
+	m_triBodyWrap = isSwapped ? body0Wrap : body1Wrap;
 
-  	  clearCache();
+	//
+	// create the manifold from the dispatcher 'manifold pool'
+	//
+	m_manifoldPtr = m_dispatcher->getNewManifold(m_convexBodyWrap->getCollisionObject(), m_triBodyWrap->getCollisionObject());
+
+	clearCache();
 }
 
 btConvexTriangleCallback::~btConvexTriangleCallback()
 {
 	clearCache();
-	m_dispatcher->releaseManifold( m_manifoldPtr );
-  
+	m_dispatcher->releaseManifold(m_manifoldPtr);
 }
-  
 
-void	btConvexTriangleCallback::clearCache()
+void btConvexTriangleCallback::clearCache()
 {
 	m_dispatcher->clearManifold(m_manifoldPtr);
 }
 
-
-void btConvexTriangleCallback::processTriangle(btVector3* triangle,int
-partId, int triangleIndex)
+void btConvexTriangleCallback::processTriangle(btVector3* triangle, int partId, int triangleIndex)
 {
 	BT_PROFILE("btConvexTriangleCallback::processTriangle");
 
@@ -88,16 +81,12 @@ partId, int triangleIndex)
 		return;
 	}
 
-        //just for debugging purposes
-        //printf("triangle %d",m_triangleCount++);
-
-
+	//just for debugging purposes
+	//printf("triangle %d",m_triangleCount++);
 
 	btCollisionAlgorithmConstructionInfo ci;
 	ci.m_dispatcher1 = m_dispatcher;
 
-
-
 #if 0	
 	
 	///debug drawing of the overlapping triangles
@@ -111,16 +100,15 @@ partId, int triangleIndex)
 		m_dispatchInfoPtr->m_debugDraw->drawLine(tr(triangle[2]),tr(triangle[0]),color);
 	}
 #endif
-	
+
 	if (m_convexBodyWrap->getCollisionShape()->isConvex())
 	{
-		btTriangleShape tm(triangle[0],triangle[1],triangle[2]);	
+		btTriangleShape tm(triangle[0], triangle[1], triangle[2]);
 		tm.setMargin(m_collisionMarginTriangle);
-		
-		
-		btCollisionObjectWrapper triObWrap(m_triBodyWrap,&tm,m_triBodyWrap->getCollisionObject(),m_triBodyWrap->getWorldTransform(),partId,triangleIndex);//correct transform?
+
+		btCollisionObjectWrapper triObWrap(m_triBodyWrap, &tm, m_triBodyWrap->getCollisionObject(), m_triBodyWrap->getWorldTransform(), partId, triangleIndex);  //correct transform?
 		btCollisionAlgorithm* colAlgo = 0;
-		
+
 		if (m_resultOut->m_closestPointDistanceThreshold > 0)
 		{
 			colAlgo = ci.m_dispatcher1->findAlgorithm(m_convexBodyWrap, &triObWrap, 0, BT_CLOSEST_POINT_ALGORITHMS);
@@ -135,36 +123,32 @@ partId, int triangleIndex)
 		{
 			tmpWrap = m_resultOut->getBody0Wrap();
 			m_resultOut->setBody0Wrap(&triObWrap);
-			m_resultOut->setShapeIdentifiersA(partId,triangleIndex);
+			m_resultOut->setShapeIdentifiersA(partId, triangleIndex);
 		}
 		else
 		{
 			tmpWrap = m_resultOut->getBody1Wrap();
 			m_resultOut->setBody1Wrap(&triObWrap);
-			m_resultOut->setShapeIdentifiersB(partId,triangleIndex);
+			m_resultOut->setShapeIdentifiersB(partId, triangleIndex);
 		}
-	
-		colAlgo->processCollision(m_convexBodyWrap,&triObWrap,*m_dispatchInfoPtr,m_resultOut);
+
+		colAlgo->processCollision(m_convexBodyWrap, &triObWrap, *m_dispatchInfoPtr, m_resultOut);
 
 		if (m_resultOut->getBody0Internal() == m_triBodyWrap->getCollisionObject())
 		{
 			m_resultOut->setBody0Wrap(tmpWrap);
-		} else
+		}
+		else
 		{
 			m_resultOut->setBody1Wrap(tmpWrap);
 		}
 
-
-
 		colAlgo->~btCollisionAlgorithm();
 		ci.m_dispatcher1->freeCollisionAlgorithm(colAlgo);
 	}
-
 }
 
-
-
-void	btConvexTriangleCallback::setTimeStepAndCounters(btScalar collisionMarginTriangle, const btDispatcherInfo& dispatchInfo, const btCollisionObjectWrapper* convexBodyWrap, const btCollisionObjectWrapper* triBodyWrap, btManifoldResult* resultOut)
+void btConvexTriangleCallback::setTimeStepAndCounters(btScalar collisionMarginTriangle, const btDispatcherInfo& dispatchInfo, const btCollisionObjectWrapper* convexBodyWrap, const btCollisionObjectWrapper* triBodyWrap, btManifoldResult* resultOut)
 {
 	m_convexBodyWrap = convexBodyWrap;
 	m_triBodyWrap = triBodyWrap;
@@ -185,16 +169,14 @@ void	btConvexTriangleCallback::setTimeStepAndCounters(btScalar collisionMarginTr
 
 	m_aabbMax += extra;
 	m_aabbMin -= extra;
-
 }
 
 void btConvexConcaveCollisionAlgorithm::clearCache()
 {
 	m_btConvexTriangleCallback.clearCache();
-
 }
 
-void btConvexConcaveCollisionAlgorithm::processCollision (const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
+void btConvexConcaveCollisionAlgorithm::processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	BT_PROFILE("btConvexConcaveCollisionAlgorithm::processCollision");
 
@@ -208,7 +190,6 @@ void btConvexConcaveCollisionAlgorithm::processCollision (const btCollisionObjec
 			btSdfCollisionShape* sdfShape = (btSdfCollisionShape*)triBodyWrap->getCollisionShape();
 			if (convexBodyWrap->getCollisionShape()->isConvex())
 			{
-
 				btConvexShape* convex = (btConvexShape*)convexBodyWrap->getCollisionShape();
 				btAlignedObjectArray<btVector3> queryVertices;
 
@@ -229,7 +210,6 @@ void btConvexConcaveCollisionAlgorithm::processCollision (const btCollisionObjec
 					queryVertices.push_back(btVector3(0, 0, 0));
 					btSphereShape* sphere = (btSphereShape*)convex;
 					maxDist = sphere->getRadius() + SIMD_EPSILON;
-
 				}
 				if (queryVertices.size())
 				{
@@ -240,7 +220,7 @@ void btConvexConcaveCollisionAlgorithm::processCollision (const btCollisionObjec
 					for (int v = 0; v < queryVertices.size(); v++)
 					{
 						const btVector3& vtx = queryVertices[v];
-						btVector3 vtxWorldSpace = convexBodyWrap->getWorldTransform()*vtx;
+						btVector3 vtxWorldSpace = convexBodyWrap->getWorldTransform() * vtx;
 						btVector3 vtxInSdf = triBodyWrap->getWorldTransform().invXform(vtxWorldSpace);
 
 						btVector3 normalLocal;
@@ -250,58 +230,52 @@ void btConvexConcaveCollisionAlgorithm::processCollision (const btCollisionObjec
 							if (dist <= maxDist)
 							{
 								normalLocal.safeNormalize();
-								btVector3 normal = triBodyWrap->getWorldTransform().getBasis()*normalLocal;
+								btVector3 normal = triBodyWrap->getWorldTransform().getBasis() * normalLocal;
 
 								if (convex->getShapeType() == SPHERE_SHAPE_PROXYTYPE)
 								{
 									btSphereShape* sphere = (btSphereShape*)convex;
 									dist -= sphere->getRadius();
-									vtxWorldSpace -= sphere->getRadius()*normal;
-
+									vtxWorldSpace -= sphere->getRadius() * normal;
 								}
-								resultOut->addContactPoint(normal,vtxWorldSpace-normal*dist, dist);
+								resultOut->addContactPoint(normal, vtxWorldSpace - normal * dist, dist);
 							}
 						}
 					}
 					resultOut->refreshContactPoints();
 				}
-
 			}
-		} else
+		}
+		else
 		{
-			const btConcaveShape* concaveShape = static_cast<const btConcaveShape*>( triBodyWrap->getCollisionShape());
-		
+			const btConcaveShape* concaveShape = static_cast<const btConcaveShape*>(triBodyWrap->getCollisionShape());
+
 			if (convexBodyWrap->getCollisionShape()->isConvex())
 			{
 				btScalar collisionMarginTriangle = concaveShape->getMargin();
-					
+
 				resultOut->setPersistentManifold(m_btConvexTriangleCallback.m_manifoldPtr);
-				m_btConvexTriangleCallback.setTimeStepAndCounters(collisionMarginTriangle,dispatchInfo,convexBodyWrap,triBodyWrap,resultOut);
+				m_btConvexTriangleCallback.setTimeStepAndCounters(collisionMarginTriangle, dispatchInfo, convexBodyWrap, triBodyWrap, resultOut);
 
-				m_btConvexTriangleCallback.m_manifoldPtr->setBodies(convexBodyWrap->getCollisionObject(),triBodyWrap->getCollisionObject());
+				m_btConvexTriangleCallback.m_manifoldPtr->setBodies(convexBodyWrap->getCollisionObject(), triBodyWrap->getCollisionObject());
+
+				concaveShape->processAllTriangles(&m_btConvexTriangleCallback, m_btConvexTriangleCallback.getAabbMin(), m_btConvexTriangleCallback.getAabbMax());
 
-				concaveShape->processAllTriangles( &m_btConvexTriangleCallback,m_btConvexTriangleCallback.getAabbMin(),m_btConvexTriangleCallback.getAabbMax());
-			
 				resultOut->refreshContactPoints();
 
 				m_btConvexTriangleCallback.clearWrapperData();
-	
 			}
 		}
-	
 	}
-
 }
 
-
-btScalar btConvexConcaveCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+btScalar btConvexConcaveCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	(void)resultOut;
 	(void)dispatchInfo;
 	btCollisionObject* convexbody = m_isSwapped ? body1 : body0;
 	btCollisionObject* triBody = m_isSwapped ? body0 : body1;
 
-
 	//quick approximation using raycast, todo: hook up to the continuous collision detection (one of the btConvexCast)
 
 	//only perform CCD above a certain threshold, this prevents blocking on the long run
@@ -320,25 +294,23 @@ btScalar btConvexConcaveCollisionAlgorithm::calculateTimeOfImpact(btCollisionObj
 	btTransform convexFromLocal = triInv * convexbody->getWorldTransform();
 	btTransform convexToLocal = triInv * convexbody->getInterpolationWorldTransform();
 
-	struct LocalTriangleSphereCastCallback	: public btTriangleCallback
+	struct LocalTriangleSphereCastCallback : public btTriangleCallback
 	{
 		btTransform m_ccdSphereFromTrans;
 		btTransform m_ccdSphereToTrans;
-		btTransform	m_meshTransform;
+		btTransform m_meshTransform;
 
-		btScalar	m_ccdSphereRadius;
-		btScalar	m_hitFraction;
-	
+		btScalar m_ccdSphereRadius;
+		btScalar m_hitFraction;
 
-		LocalTriangleSphereCastCallback(const btTransform& from,const btTransform& to,btScalar ccdSphereRadius,btScalar hitFraction)
-			:m_ccdSphereFromTrans(from),
-			m_ccdSphereToTrans(to),
-			m_ccdSphereRadius(ccdSphereRadius),
-			m_hitFraction(hitFraction)
-		{			
+		LocalTriangleSphereCastCallback(const btTransform& from, const btTransform& to, btScalar ccdSphereRadius, btScalar hitFraction)
+			: m_ccdSphereFromTrans(from),
+			  m_ccdSphereToTrans(to),
+			  m_ccdSphereRadius(ccdSphereRadius),
+			  m_hitFraction(hitFraction)
+		{
 		}
-		
-		
+
 		virtual void processTriangle(btVector3* triangle, int partId, int triangleIndex)
 		{
 			BT_PROFILE("processTriangle");
@@ -349,29 +321,23 @@ btScalar btConvexConcaveCollisionAlgorithm::calculateTimeOfImpact(btCollisionObj
 			ident.setIdentity();
 			btConvexCast::CastResult castResult;
 			castResult.m_fraction = m_hitFraction;
-			btSphereShape	pointShape(m_ccdSphereRadius);
-			btTriangleShape	triShape(triangle[0],triangle[1],triangle[2]);
-			btVoronoiSimplexSolver	simplexSolver;
-			btSubsimplexConvexCast convexCaster(&pointShape,&triShape,&simplexSolver);
+			btSphereShape pointShape(m_ccdSphereRadius);
+			btTriangleShape triShape(triangle[0], triangle[1], triangle[2]);
+			btVoronoiSimplexSolver simplexSolver;
+			btSubsimplexConvexCast convexCaster(&pointShape, &triShape, &simplexSolver);
 			//GjkConvexCast	convexCaster(&pointShape,convexShape,&simplexSolver);
 			//ContinuousConvexCollision convexCaster(&pointShape,convexShape,&simplexSolver,0);
 			//local space?
 
-			if (convexCaster.calcTimeOfImpact(m_ccdSphereFromTrans,m_ccdSphereToTrans,
-				ident,ident,castResult))
+			if (convexCaster.calcTimeOfImpact(m_ccdSphereFromTrans, m_ccdSphereToTrans,
+											  ident, ident, castResult))
 			{
 				if (m_hitFraction > castResult.m_fraction)
 					m_hitFraction = castResult.m_fraction;
 			}
-
 		}
-
 	};
 
-
-	
-
-	
 	if (triBody->getCollisionShape()->isConcave())
 	{
 		btVector3 rayAabbMin = convexFromLocal.getOrigin();
@@ -379,33 +345,30 @@ btScalar btConvexConcaveCollisionAlgorithm::calculateTimeOfImpact(btCollisionObj
 		btVector3 rayAabbMax = convexFromLocal.getOrigin();
 		rayAabbMax.setMax(convexToLocal.getOrigin());
 		btScalar ccdRadius0 = convexbody->getCcdSweptSphereRadius();
-		rayAabbMin -= btVector3(ccdRadius0,ccdRadius0,ccdRadius0);
-		rayAabbMax += btVector3(ccdRadius0,ccdRadius0,ccdRadius0);
+		rayAabbMin -= btVector3(ccdRadius0, ccdRadius0, ccdRadius0);
+		rayAabbMax += btVector3(ccdRadius0, ccdRadius0, ccdRadius0);
 
-		btScalar curHitFraction = btScalar(1.); //is this available?
-		LocalTriangleSphereCastCallback raycastCallback(convexFromLocal,convexToLocal,
-			convexbody->getCcdSweptSphereRadius(),curHitFraction);
+		btScalar curHitFraction = btScalar(1.);  //is this available?
+		LocalTriangleSphereCastCallback raycastCallback(convexFromLocal, convexToLocal,
+														convexbody->getCcdSweptSphereRadius(), curHitFraction);
 
 		raycastCallback.m_hitFraction = convexbody->getHitFraction();
 
 		btCollisionObject* concavebody = triBody;
 
-		btConcaveShape* triangleMesh = (btConcaveShape*) concavebody->getCollisionShape();
-		
+		btConcaveShape* triangleMesh = (btConcaveShape*)concavebody->getCollisionShape();
+
 		if (triangleMesh)
 		{
-			triangleMesh->processAllTriangles(&raycastCallback,rayAabbMin,rayAabbMax);
+			triangleMesh->processAllTriangles(&raycastCallback, rayAabbMin, rayAabbMax);
 		}
-	
-
 
 		if (raycastCallback.m_hitFraction < convexbody->getHitFraction())
 		{
-			convexbody->setHitFraction( raycastCallback.m_hitFraction);
+			convexbody->setHitFraction(raycastCallback.m_hitFraction);
 			return raycastCallback.m_hitFraction;
 		}
 	}
 
 	return btScalar(1.);
-
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexConcaveCollisionAlgorithm.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexConcaveCollisionAlgorithm.h
index 93d842ef501..b72e4029811 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexConcaveCollisionAlgorithm.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexConcaveCollisionAlgorithm.h
@@ -26,42 +26,40 @@ class btDispatcher;
 #include "btCollisionCreateFunc.h"
 
 ///For each triangle in the concave mesh that overlaps with the AABB of a convex (m_convexProxy), processTriangle is called.
-ATTRIBUTE_ALIGNED16(class)  btConvexTriangleCallback : public btTriangleCallback
+ATTRIBUTE_ALIGNED16(class)
+btConvexTriangleCallback : public btTriangleCallback
 {
-
-	btVector3	m_aabbMin;
-	btVector3	m_aabbMax ;
+	btVector3 m_aabbMin;
+	btVector3 m_aabbMax;
 
 	const btCollisionObjectWrapper* m_convexBodyWrap;
 	const btCollisionObjectWrapper* m_triBodyWrap;
 
-
-
 	btManifoldResult* m_resultOut;
-	btDispatcher*	m_dispatcher;
+	btDispatcher* m_dispatcher;
 	const btDispatcherInfo* m_dispatchInfoPtr;
 	btScalar m_collisionMarginTriangle;
-	
+
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
-int	m_triangleCount;
-	
-	btPersistentManifold*	m_manifoldPtr;
 
-	btConvexTriangleCallback(btDispatcher* dispatcher,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,bool isSwapped);
+	int m_triangleCount;
 
-	void	setTimeStepAndCounters(btScalar collisionMarginTriangle,const btDispatcherInfo& dispatchInfo,const btCollisionObjectWrapper* convexBodyWrap, const btCollisionObjectWrapper* triBodyWrap, btManifoldResult* resultOut);
+	btPersistentManifold* m_manifoldPtr;
 
-	void	clearWrapperData()
+	btConvexTriangleCallback(btDispatcher * dispatcher, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, bool isSwapped);
+
+	void setTimeStepAndCounters(btScalar collisionMarginTriangle, const btDispatcherInfo& dispatchInfo, const btCollisionObjectWrapper* convexBodyWrap, const btCollisionObjectWrapper* triBodyWrap, btManifoldResult* resultOut);
+
+	void clearWrapperData()
 	{
 		m_convexBodyWrap = 0;
 		m_triBodyWrap = 0;
 	}
 	virtual ~btConvexTriangleCallback();
 
-	virtual void processTriangle(btVector3* triangle, int partId, int triangleIndex);
-	
+	virtual void processTriangle(btVector3 * triangle, int partId, int triangleIndex);
+
 	void clearCache();
 
 	SIMD_FORCE_INLINE const btVector3& getAabbMin() const
@@ -72,56 +70,48 @@ int	m_triangleCount;
 	{
 		return m_aabbMax;
 	}
-
 };
 
-
-
-
 /// btConvexConcaveCollisionAlgorithm  supports collision between convex shapes and (concave) trianges meshes.
-ATTRIBUTE_ALIGNED16(class)  btConvexConcaveCollisionAlgorithm  : public btActivatingCollisionAlgorithm
+ATTRIBUTE_ALIGNED16(class)
+btConvexConcaveCollisionAlgorithm : public btActivatingCollisionAlgorithm
 {
-
 	btConvexTriangleCallback m_btConvexTriangleCallback;
 
-	bool	m_isSwapped;
-
-
+	bool m_isSwapped;
 
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
-	btConvexConcaveCollisionAlgorithm( const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,bool isSwapped);
+
+	btConvexConcaveCollisionAlgorithm(const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, bool isSwapped);
 
 	virtual ~btConvexConcaveCollisionAlgorithm();
 
-	virtual void processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual void processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	btScalar	calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	btScalar calculateTimeOfImpact(btCollisionObject * body0, btCollisionObject * body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual	void	getAllContactManifolds(btManifoldArray&	manifoldArray);
-	
-	void	clearCache();
+	virtual void getAllContactManifolds(btManifoldArray & manifoldArray);
 
-	struct CreateFunc :public 	btCollisionAlgorithmCreateFunc
+	void clearCache();
+
+	struct CreateFunc : public btCollisionAlgorithmCreateFunc
 	{
-		virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+		virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 		{
 			void* mem = ci.m_dispatcher1->allocateCollisionAlgorithm(sizeof(btConvexConcaveCollisionAlgorithm));
-			return new(mem) btConvexConcaveCollisionAlgorithm(ci,body0Wrap,body1Wrap,false);
+			return new (mem) btConvexConcaveCollisionAlgorithm(ci, body0Wrap, body1Wrap, false);
 		}
 	};
 
-	struct SwappedCreateFunc :public 	btCollisionAlgorithmCreateFunc
+	struct SwappedCreateFunc : public btCollisionAlgorithmCreateFunc
 	{
-		virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+		virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 		{
 			void* mem = ci.m_dispatcher1->allocateCollisionAlgorithm(sizeof(btConvexConcaveCollisionAlgorithm));
-			return new(mem) btConvexConcaveCollisionAlgorithm(ci,body0Wrap,body1Wrap,true);
+			return new (mem) btConvexConcaveCollisionAlgorithm(ci, body0Wrap, body1Wrap, true);
 		}
 	};
-
 };
 
-#endif //BT_CONVEX_CONCAVE_COLLISION_ALGORITHM_H
+#endif  //BT_CONVEX_CONCAVE_COLLISION_ALGORITHM_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexConvexAlgorithm.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexConvexAlgorithm.cpp
index 3e8bc6e426e..44dd3c553e6 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexConvexAlgorithm.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexConvexAlgorithm.cpp
@@ -30,8 +30,6 @@ subject to the following restrictions:
 #include "BulletCollision/CollisionShapes/btTriangleShape.h"
 #include "BulletCollision/CollisionShapes/btConvexPolyhedron.h"
 
-
-
 #include "BulletCollision/NarrowPhaseCollision/btGjkPairDetector.h"
 #include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h"
 #include "BulletCollision/CollisionDispatch/btCollisionDispatcher.h"
@@ -43,8 +41,6 @@ subject to the following restrictions:
 #include "BulletCollision/NarrowPhaseCollision/btSubSimplexConvexCast.h"
 #include "BulletCollision/NarrowPhaseCollision/btGjkConvexCast.h"
 
-
-
 #include "BulletCollision/NarrowPhaseCollision/btVoronoiSimplexSolver.h"
 #include "BulletCollision/CollisionShapes/btSphereShape.h"
 
@@ -57,8 +53,6 @@ subject to the following restrictions:
 
 ///////////
 
-
-
 static SIMD_FORCE_INLINE void segmentsClosestPoints(
 	btVector3& ptsVector,
 	btVector3& offsetA,
@@ -66,43 +60,49 @@ static SIMD_FORCE_INLINE void segmentsClosestPoints(
 	btScalar& tA, btScalar& tB,
 	const btVector3& translation,
 	const btVector3& dirA, btScalar hlenA,
-	const btVector3& dirB, btScalar hlenB )
+	const btVector3& dirB, btScalar hlenB)
 {
 	// compute the parameters of the closest points on each line segment
 
-	btScalar dirA_dot_dirB = btDot(dirA,dirB);
-	btScalar dirA_dot_trans = btDot(dirA,translation);
-	btScalar dirB_dot_trans = btDot(dirB,translation);
+	btScalar dirA_dot_dirB = btDot(dirA, dirB);
+	btScalar dirA_dot_trans = btDot(dirA, translation);
+	btScalar dirB_dot_trans = btDot(dirB, translation);
 
 	btScalar denom = 1.0f - dirA_dot_dirB * dirA_dot_dirB;
 
-	if ( denom == 0.0f ) {
+	if (denom == 0.0f)
+	{
 		tA = 0.0f;
-	} else {
-		tA = ( dirA_dot_trans - dirB_dot_trans * dirA_dot_dirB ) / denom;
-		if ( tA < -hlenA )
+	}
+	else
+	{
+		tA = (dirA_dot_trans - dirB_dot_trans * dirA_dot_dirB) / denom;
+		if (tA < -hlenA)
 			tA = -hlenA;
-		else if ( tA > hlenA )
+		else if (tA > hlenA)
 			tA = hlenA;
 	}
 
 	tB = tA * dirA_dot_dirB - dirB_dot_trans;
 
-	if ( tB < -hlenB ) {
+	if (tB < -hlenB)
+	{
 		tB = -hlenB;
 		tA = tB * dirA_dot_dirB + dirA_dot_trans;
 
-		if ( tA < -hlenA )
+		if (tA < -hlenA)
 			tA = -hlenA;
-		else if ( tA > hlenA )
+		else if (tA > hlenA)
 			tA = hlenA;
-	} else if ( tB > hlenB ) {
+	}
+	else if (tB > hlenB)
+	{
 		tB = hlenB;
 		tA = tB * dirA_dot_dirB + dirA_dot_trans;
 
-		if ( tA < -hlenA )
+		if (tA < -hlenA)
 			tA = -hlenA;
-		else if ( tA > hlenA )
+		else if (tA > hlenA)
 			tA = hlenA;
 	}
 
@@ -114,19 +114,18 @@ static SIMD_FORCE_INLINE void segmentsClosestPoints(
 	ptsVector = translation - offsetA + offsetB;
 }
 
-
 static SIMD_FORCE_INLINE btScalar capsuleCapsuleDistance(
 	btVector3& normalOnB,
 	btVector3& pointOnB,
 	btScalar capsuleLengthA,
-	btScalar	capsuleRadiusA,
+	btScalar capsuleRadiusA,
 	btScalar capsuleLengthB,
-	btScalar	capsuleRadiusB,
+	btScalar capsuleRadiusB,
 	int capsuleAxisA,
 	int capsuleAxisB,
 	const btTransform& transformA,
 	const btTransform& transformB,
-	btScalar distanceThreshold )
+	btScalar distanceThreshold)
 {
 	btVector3 directionA = transformA.getBasis().getColumn(capsuleAxisA);
 	btVector3 translationA = transformA.getOrigin();
@@ -139,47 +138,38 @@ static SIMD_FORCE_INLINE btScalar capsuleCapsuleDistance(
 
 	// compute the closest points of the capsule line segments
 
-	btVector3 ptsVector;           // the vector between the closest points
-	
-	btVector3 offsetA, offsetB;    // offsets from segment centers to their closest points
-	btScalar tA, tB;              // parameters on line segment
+	btVector3 ptsVector;  // the vector between the closest points
 
-	segmentsClosestPoints( ptsVector, offsetA, offsetB, tA, tB, translation,
-						   directionA, capsuleLengthA, directionB, capsuleLengthB );
+	btVector3 offsetA, offsetB;  // offsets from segment centers to their closest points
+	btScalar tA, tB;             // parameters on line segment
+
+	segmentsClosestPoints(ptsVector, offsetA, offsetB, tA, tB, translation,
+						  directionA, capsuleLengthA, directionB, capsuleLengthB);
 
 	btScalar distance = ptsVector.length() - capsuleRadiusA - capsuleRadiusB;
 
-	if ( distance > distanceThreshold )
+	if (distance > distanceThreshold)
 		return distance;
 
 	btScalar lenSqr = ptsVector.length2();
-	if (lenSqr<= (SIMD_EPSILON*SIMD_EPSILON))
+	if (lenSqr <= (SIMD_EPSILON * SIMD_EPSILON))
 	{
 		//degenerate case where 2 capsules are likely at the same location: take a vector tangential to 'directionA'
 		btVector3 q;
-		btPlaneSpace1(directionA,normalOnB,q);
-	} else
+		btPlaneSpace1(directionA, normalOnB, q);
+	}
+	else
 	{
 		// compute the contact normal
-		normalOnB = ptsVector*-btRecipSqrt(lenSqr);
+		normalOnB = ptsVector * -btRecipSqrt(lenSqr);
 	}
-	pointOnB = transformB.getOrigin()+offsetB + normalOnB * capsuleRadiusB;
+	pointOnB = transformB.getOrigin() + offsetB + normalOnB * capsuleRadiusB;
 
 	return distance;
 }
 
-
-
-
-
-
-
 //////////
 
-
-
-
-
 btConvexConvexAlgorithm::CreateFunc::CreateFunc(btConvexPenetrationDepthSolver* pdSolver)
 {
 	m_numPerturbationIterations = 0;
@@ -187,30 +177,27 @@ btConvexConvexAlgorithm::CreateFunc::CreateFunc(btConvexPenetrationDepthSolver*
 	m_pdSolver = pdSolver;
 }
 
-btConvexConvexAlgorithm::CreateFunc::~CreateFunc() 
-{ 
+btConvexConvexAlgorithm::CreateFunc::~CreateFunc()
+{
 }
 
-btConvexConvexAlgorithm::btConvexConvexAlgorithm(btPersistentManifold* mf,const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,btConvexPenetrationDepthSolver* pdSolver,int numPerturbationIterations, int minimumPointsPerturbationThreshold)
-: btActivatingCollisionAlgorithm(ci,body0Wrap,body1Wrap),
-m_pdSolver(pdSolver),
-m_ownManifold (false),
-m_manifoldPtr(mf),
-m_lowLevelOfDetail(false),
+btConvexConvexAlgorithm::btConvexConvexAlgorithm(btPersistentManifold* mf, const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, btConvexPenetrationDepthSolver* pdSolver, int numPerturbationIterations, int minimumPointsPerturbationThreshold)
+	: btActivatingCollisionAlgorithm(ci, body0Wrap, body1Wrap),
+	  m_pdSolver(pdSolver),
+	  m_ownManifold(false),
+	  m_manifoldPtr(mf),
+	  m_lowLevelOfDetail(false),
 #ifdef USE_SEPDISTANCE_UTIL2
-m_sepDistance((static_cast<btConvexShape*>(body0->getCollisionShape()))->getAngularMotionDisc(),
-			  (static_cast<btConvexShape*>(body1->getCollisionShape()))->getAngularMotionDisc()),
+	  m_sepDistance((static_cast<btConvexShape*>(body0->getCollisionShape()))->getAngularMotionDisc(),
+					(static_cast<btConvexShape*>(body1->getCollisionShape()))->getAngularMotionDisc()),
 #endif
-m_numPerturbationIterations(numPerturbationIterations),
-m_minimumPointsPerturbationThreshold(minimumPointsPerturbationThreshold)
+	  m_numPerturbationIterations(numPerturbationIterations),
+	  m_minimumPointsPerturbationThreshold(minimumPointsPerturbationThreshold)
 {
 	(void)body0Wrap;
 	(void)body1Wrap;
 }
 
-
-
-
 btConvexConvexAlgorithm::~btConvexConvexAlgorithm()
 {
 	if (m_ownManifold)
@@ -220,112 +207,105 @@ btConvexConvexAlgorithm::~btConvexConvexAlgorithm()
 	}
 }
 
-void	btConvexConvexAlgorithm ::setLowLevelOfDetail(bool useLowLevel)
+void btConvexConvexAlgorithm ::setLowLevelOfDetail(bool useLowLevel)
 {
 	m_lowLevelOfDetail = useLowLevel;
 }
 
-
 struct btPerturbedContactResult : public btManifoldResult
 {
 	btManifoldResult* m_originalManifoldResult;
 	btTransform m_transformA;
 	btTransform m_transformB;
-	btTransform	m_unPerturbedTransform;
-	bool	m_perturbA;
-	btIDebugDraw*	m_debugDrawer;
+	btTransform m_unPerturbedTransform;
+	bool m_perturbA;
+	btIDebugDraw* m_debugDrawer;
 
-
-	btPerturbedContactResult(btManifoldResult* originalResult,const btTransform& transformA,const btTransform& transformB,const btTransform& unPerturbedTransform,bool perturbA,btIDebugDraw* debugDrawer)
-		:m_originalManifoldResult(originalResult),
-		m_transformA(transformA),
-		m_transformB(transformB),
-		m_unPerturbedTransform(unPerturbedTransform),
-		m_perturbA(perturbA),
-		m_debugDrawer(debugDrawer)
+	btPerturbedContactResult(btManifoldResult* originalResult, const btTransform& transformA, const btTransform& transformB, const btTransform& unPerturbedTransform, bool perturbA, btIDebugDraw* debugDrawer)
+		: m_originalManifoldResult(originalResult),
+		  m_transformA(transformA),
+		  m_transformB(transformB),
+		  m_unPerturbedTransform(unPerturbedTransform),
+		  m_perturbA(perturbA),
+		  m_debugDrawer(debugDrawer)
 	{
 	}
-	virtual ~ btPerturbedContactResult()
+	virtual ~btPerturbedContactResult()
 	{
 	}
 
-	virtual void addContactPoint(const btVector3& normalOnBInWorld,const btVector3& pointInWorld,btScalar orgDepth)
+	virtual void addContactPoint(const btVector3& normalOnBInWorld, const btVector3& pointInWorld, btScalar orgDepth)
 	{
-		btVector3 endPt,startPt;
+		btVector3 endPt, startPt;
 		btScalar newDepth;
 		btVector3 newNormal;
 
 		if (m_perturbA)
 		{
-			btVector3 endPtOrg = pointInWorld + normalOnBInWorld*orgDepth;
-			endPt = (m_unPerturbedTransform*m_transformA.inverse())(endPtOrg);
-			newDepth = (endPt -  pointInWorld).dot(normalOnBInWorld);
-			startPt = endPt - normalOnBInWorld*newDepth;
-		} else
+			btVector3 endPtOrg = pointInWorld + normalOnBInWorld * orgDepth;
+			endPt = (m_unPerturbedTransform * m_transformA.inverse())(endPtOrg);
+			newDepth = (endPt - pointInWorld).dot(normalOnBInWorld);
+			startPt = endPt - normalOnBInWorld * newDepth;
+		}
+		else
 		{
-			endPt = pointInWorld + normalOnBInWorld*orgDepth;
-			startPt = (m_unPerturbedTransform*m_transformB.inverse())(pointInWorld);
-			newDepth = (endPt -  startPt).dot(normalOnBInWorld);
-			
+			endPt = pointInWorld + normalOnBInWorld * orgDepth;
+			startPt = (m_unPerturbedTransform * m_transformB.inverse())(pointInWorld);
+			newDepth = (endPt - startPt).dot(normalOnBInWorld);
 		}
 
 //#define DEBUG_CONTACTS 1
 #ifdef DEBUG_CONTACTS
-		m_debugDrawer->drawLine(startPt,endPt,btVector3(1,0,0));
-		m_debugDrawer->drawSphere(startPt,0.05,btVector3(0,1,0));
-		m_debugDrawer->drawSphere(endPt,0.05,btVector3(0,0,1));
-#endif //DEBUG_CONTACTS
+		m_debugDrawer->drawLine(startPt, endPt, btVector3(1, 0, 0));
+		m_debugDrawer->drawSphere(startPt, 0.05, btVector3(0, 1, 0));
+		m_debugDrawer->drawSphere(endPt, 0.05, btVector3(0, 0, 1));
+#endif  //DEBUG_CONTACTS
 
-		
-		m_originalManifoldResult->addContactPoint(normalOnBInWorld,startPt,newDepth);
+		m_originalManifoldResult->addContactPoint(normalOnBInWorld, startPt, newDepth);
 	}
-
 };
 
 extern btScalar gContactBreakingThreshold;
 
-
 //
 // Convex-Convex collision algorithm
 //
-void btConvexConvexAlgorithm ::processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+void btConvexConvexAlgorithm ::processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
-
 	if (!m_manifoldPtr)
 	{
 		//swapped?
-		m_manifoldPtr = m_dispatcher->getNewManifold(body0Wrap->getCollisionObject(),body1Wrap->getCollisionObject());
+		m_manifoldPtr = m_dispatcher->getNewManifold(body0Wrap->getCollisionObject(), body1Wrap->getCollisionObject());
 		m_ownManifold = true;
 	}
 	resultOut->setPersistentManifold(m_manifoldPtr);
 
 	//comment-out next line to test multi-contact generation
 	//resultOut->getPersistentManifold()->clearManifold();
-	
 
 	const btConvexShape* min0 = static_cast<const btConvexShape*>(body0Wrap->getCollisionShape());
 	const btConvexShape* min1 = static_cast<const btConvexShape*>(body1Wrap->getCollisionShape());
 
-	btVector3  normalOnB;
-		btVector3  pointOnBWorld;
+	btVector3 normalOnB;
+	btVector3 pointOnBWorld;
 #ifndef BT_DISABLE_CAPSULE_CAPSULE_COLLIDER
 	if ((min0->getShapeType() == CAPSULE_SHAPE_PROXYTYPE) && (min1->getShapeType() == CAPSULE_SHAPE_PROXYTYPE))
 	{
 		//m_manifoldPtr->clearManifold();
 
-		btCapsuleShape* capsuleA = (btCapsuleShape*) min0;
-		btCapsuleShape* capsuleB = (btCapsuleShape*) min1;
-		
-		btScalar threshold = m_manifoldPtr->getContactBreakingThreshold();
+		btCapsuleShape* capsuleA = (btCapsuleShape*)min0;
+		btCapsuleShape* capsuleB = (btCapsuleShape*)min1;
 
-		btScalar dist = capsuleCapsuleDistance(normalOnB,	pointOnBWorld,capsuleA->getHalfHeight(),capsuleA->getRadius(),
-			capsuleB->getHalfHeight(),capsuleB->getRadius(),capsuleA->getUpAxis(),capsuleB->getUpAxis(),
-			body0Wrap->getWorldTransform(),body1Wrap->getWorldTransform(),threshold);
+		btScalar threshold = m_manifoldPtr->getContactBreakingThreshold()+ resultOut->m_closestPointDistanceThreshold;
 
-		if (dist<threshold)
+		btScalar dist = capsuleCapsuleDistance(normalOnB, pointOnBWorld, capsuleA->getHalfHeight(), capsuleA->getRadius(),
+											   capsuleB->getHalfHeight(), capsuleB->getRadius(), capsuleA->getUpAxis(), capsuleB->getUpAxis(),
+											   body0Wrap->getWorldTransform(), body1Wrap->getWorldTransform(), threshold);
+
+		if (dist < threshold)
 		{
-			btAssert(normalOnB.length2()>=(SIMD_EPSILON*SIMD_EPSILON));
-			resultOut->addContactPoint(normalOnB,pointOnBWorld,dist);	
+			btAssert(normalOnB.length2() >= (SIMD_EPSILON * SIMD_EPSILON));
+			resultOut->addContactPoint(normalOnB, pointOnBWorld, dist);
 		}
 		resultOut->refreshContactPoints();
 		return;
@@ -335,19 +315,19 @@ void btConvexConvexAlgorithm ::processCollision (const btCollisionObjectWrapper*
 	{
 		//m_manifoldPtr->clearManifold();
 
-		btCapsuleShape* capsuleA = (btCapsuleShape*) min0;
-		btSphereShape* capsuleB = (btSphereShape*) min1;
-		
-		btScalar threshold = m_manifoldPtr->getContactBreakingThreshold();
+		btCapsuleShape* capsuleA = (btCapsuleShape*)min0;
+		btSphereShape* capsuleB = (btSphereShape*)min1;
 
-		btScalar dist = capsuleCapsuleDistance(normalOnB,	pointOnBWorld,capsuleA->getHalfHeight(),capsuleA->getRadius(),
-			0.,capsuleB->getRadius(),capsuleA->getUpAxis(),1,
-			body0Wrap->getWorldTransform(),body1Wrap->getWorldTransform(),threshold);
+		btScalar threshold = m_manifoldPtr->getContactBreakingThreshold()+ resultOut->m_closestPointDistanceThreshold;
 
-		if (dist<threshold)
+		btScalar dist = capsuleCapsuleDistance(normalOnB, pointOnBWorld, capsuleA->getHalfHeight(), capsuleA->getRadius(),
+											   0., capsuleB->getRadius(), capsuleA->getUpAxis(), 1,
+											   body0Wrap->getWorldTransform(), body1Wrap->getWorldTransform(), threshold);
+
+		if (dist < threshold)
 		{
-			btAssert(normalOnB.length2()>=(SIMD_EPSILON*SIMD_EPSILON));
-			resultOut->addContactPoint(normalOnB,pointOnBWorld,dist);	
+			btAssert(normalOnB.length2() >= (SIMD_EPSILON * SIMD_EPSILON));
+			resultOut->addContactPoint(normalOnB, pointOnBWorld, dist);
 		}
 		resultOut->refreshContactPoints();
 		return;
@@ -357,252 +337,227 @@ void btConvexConvexAlgorithm ::processCollision (const btCollisionObjectWrapper*
 	{
 		//m_manifoldPtr->clearManifold();
 
-		btSphereShape* capsuleA = (btSphereShape*) min0;
-		btCapsuleShape* capsuleB = (btCapsuleShape*) min1;
-		
-		btScalar threshold = m_manifoldPtr->getContactBreakingThreshold();
+		btSphereShape* capsuleA = (btSphereShape*)min0;
+		btCapsuleShape* capsuleB = (btCapsuleShape*)min1;
 
-		btScalar dist = capsuleCapsuleDistance(normalOnB,	pointOnBWorld,0.,capsuleA->getRadius(),
-			capsuleB->getHalfHeight(),capsuleB->getRadius(),1,capsuleB->getUpAxis(),
-			body0Wrap->getWorldTransform(),body1Wrap->getWorldTransform(),threshold);
+		btScalar threshold = m_manifoldPtr->getContactBreakingThreshold()+ resultOut->m_closestPointDistanceThreshold;
 
-		if (dist<threshold)
+		btScalar dist = capsuleCapsuleDistance(normalOnB, pointOnBWorld, 0., capsuleA->getRadius(),
+											   capsuleB->getHalfHeight(), capsuleB->getRadius(), 1, capsuleB->getUpAxis(),
+											   body0Wrap->getWorldTransform(), body1Wrap->getWorldTransform(), threshold);
+
+		if (dist < threshold)
 		{
-			btAssert(normalOnB.length2()>=(SIMD_EPSILON*SIMD_EPSILON));
-			resultOut->addContactPoint(normalOnB,pointOnBWorld,dist);	
+			btAssert(normalOnB.length2() >= (SIMD_EPSILON * SIMD_EPSILON));
+			resultOut->addContactPoint(normalOnB, pointOnBWorld, dist);
 		}
 		resultOut->refreshContactPoints();
 		return;
 	}
-#endif //BT_DISABLE_CAPSULE_CAPSULE_COLLIDER
-
-
-
+#endif  //BT_DISABLE_CAPSULE_CAPSULE_COLLIDER
 
 #ifdef USE_SEPDISTANCE_UTIL2
 	if (dispatchInfo.m_useConvexConservativeDistanceUtil)
 	{
-		m_sepDistance.updateSeparatingDistance(body0->getWorldTransform(),body1->getWorldTransform());
+		m_sepDistance.updateSeparatingDistance(body0->getWorldTransform(), body1->getWorldTransform());
 	}
 
-	if (!dispatchInfo.m_useConvexConservativeDistanceUtil || m_sepDistance.getConservativeSeparatingDistance()<=0.f)
-#endif //USE_SEPDISTANCE_UTIL2
+	if (!dispatchInfo.m_useConvexConservativeDistanceUtil || m_sepDistance.getConservativeSeparatingDistance() <= 0.f)
+#endif  //USE_SEPDISTANCE_UTIL2
 
 	{
-
-	
-	btGjkPairDetector::ClosestPointInput input;
-    btVoronoiSimplexSolver simplexSolver;
-    btGjkPairDetector	gjkPairDetector( min0, min1, &simplexSolver, m_pdSolver );
-	//TODO: if (dispatchInfo.m_useContinuous)
-	gjkPairDetector.setMinkowskiA(min0);
-	gjkPairDetector.setMinkowskiB(min1);
+		btGjkPairDetector::ClosestPointInput input;
+		btVoronoiSimplexSolver simplexSolver;
+		btGjkPairDetector gjkPairDetector(min0, min1, &simplexSolver, m_pdSolver);
+		//TODO: if (dispatchInfo.m_useContinuous)
+		gjkPairDetector.setMinkowskiA(min0);
+		gjkPairDetector.setMinkowskiB(min1);
 
 #ifdef USE_SEPDISTANCE_UTIL2
-	if (dispatchInfo.m_useConvexConservativeDistanceUtil)
-	{
-		input.m_maximumDistanceSquared = BT_LARGE_FLOAT;
-	} else
-#endif //USE_SEPDISTANCE_UTIL2
-	{
-		//if (dispatchInfo.m_convexMaxDistanceUseCPT)
-		//{
-		//	input.m_maximumDistanceSquared = min0->getMargin() + min1->getMargin() + m_manifoldPtr->getContactProcessingThreshold();
-		//} else
-		//{
-		input.m_maximumDistanceSquared = min0->getMargin() + min1->getMargin() + m_manifoldPtr->getContactBreakingThreshold()+resultOut->m_closestPointDistanceThreshold;
-//		}
-
-		input.m_maximumDistanceSquared*= input.m_maximumDistanceSquared;
-	}
-
-	input.m_transformA = body0Wrap->getWorldTransform();
-	input.m_transformB = body1Wrap->getWorldTransform();
-
-
-
-	
-
-#ifdef USE_SEPDISTANCE_UTIL2
-	btScalar sepDist = 0.f;
-	if (dispatchInfo.m_useConvexConservativeDistanceUtil)
-	{
-		sepDist = gjkPairDetector.getCachedSeparatingDistance();
-		if (sepDist>SIMD_EPSILON)
+		if (dispatchInfo.m_useConvexConservativeDistanceUtil)
 		{
-			sepDist += dispatchInfo.m_convexConservativeDistanceThreshold;
-			//now perturbe directions to get multiple contact points
-			
+			input.m_maximumDistanceSquared = BT_LARGE_FLOAT;
 		}
-	}
-#endif //USE_SEPDISTANCE_UTIL2
-
-	if (min0->isPolyhedral() && min1->isPolyhedral())
-	{
-
-
-		struct btDummyResult : public btDiscreteCollisionDetectorInterface::Result
+		else
+#endif  //USE_SEPDISTANCE_UTIL2
 		{
-			btVector3 m_normalOnBInWorld;
-			btVector3 m_pointInWorld;
-			btScalar m_depth;
-			bool m_hasContact;
-			
+			//if (dispatchInfo.m_convexMaxDistanceUseCPT)
+			//{
+			//	input.m_maximumDistanceSquared = min0->getMargin() + min1->getMargin() + m_manifoldPtr->getContactProcessingThreshold();
+			//} else
+			//{
+			input.m_maximumDistanceSquared = min0->getMargin() + min1->getMargin() + m_manifoldPtr->getContactBreakingThreshold() + resultOut->m_closestPointDistanceThreshold;
+			//		}
 
-			btDummyResult()
-				: m_hasContact(false)
-			{
-			}
-			
-			
-			virtual void setShapeIdentifiersA(int partId0,int index0){}
-			virtual void setShapeIdentifiersB(int partId1,int index1){}
-			virtual void addContactPoint(const btVector3& normalOnBInWorld,const btVector3& pointInWorld,btScalar depth) 
-			{
-				m_hasContact = true;
-				m_normalOnBInWorld = normalOnBInWorld;
-				m_pointInWorld = pointInWorld;
-				m_depth = depth;
-			}
-		};
+			input.m_maximumDistanceSquared *= input.m_maximumDistanceSquared;
+		}
 
-		
-		struct btWithoutMarginResult : public btDiscreteCollisionDetectorInterface::Result
+		input.m_transformA = body0Wrap->getWorldTransform();
+		input.m_transformB = body1Wrap->getWorldTransform();
+
+#ifdef USE_SEPDISTANCE_UTIL2
+		btScalar sepDist = 0.f;
+		if (dispatchInfo.m_useConvexConservativeDistanceUtil)
 		{
-			btDiscreteCollisionDetectorInterface::Result* m_originalResult;
-			btVector3	m_reportedNormalOnWorld;
-			btScalar m_marginOnA;
-			btScalar m_marginOnB;
-			btScalar	m_reportedDistance;
-			
-			bool		m_foundResult;
-			btWithoutMarginResult(btDiscreteCollisionDetectorInterface::Result* result, btScalar marginOnA, btScalar marginOnB)
-			:m_originalResult(result),
-			m_marginOnA(marginOnA),
-			m_marginOnB(marginOnB),
-			m_foundResult(false)
+			sepDist = gjkPairDetector.getCachedSeparatingDistance();
+			if (sepDist > SIMD_EPSILON)
 			{
+				sepDist += dispatchInfo.m_convexConservativeDistanceThreshold;
+				//now perturbe directions to get multiple contact points
 			}
-			
-			virtual void setShapeIdentifiersA(int partId0,int index0){}
-			virtual void setShapeIdentifiersB(int partId1,int index1){}
-			virtual void addContactPoint(const btVector3& normalOnBInWorld,const btVector3& pointInWorldOrg,btScalar depthOrg) 
+		}
+#endif  //USE_SEPDISTANCE_UTIL2
+
+		if (min0->isPolyhedral() && min1->isPolyhedral())
+		{
+			struct btDummyResult : public btDiscreteCollisionDetectorInterface::Result
 			{
-				m_reportedDistance = depthOrg;
-				m_reportedNormalOnWorld = normalOnBInWorld;
-				
-				btVector3 adjustedPointB = pointInWorldOrg - normalOnBInWorld*m_marginOnB;
-				m_reportedDistance = depthOrg+(m_marginOnA+m_marginOnB);
-				if (m_reportedDistance<0.f)
+				btVector3 m_normalOnBInWorld;
+				btVector3 m_pointInWorld;
+				btScalar m_depth;
+				bool m_hasContact;
+
+				btDummyResult()
+					: m_hasContact(false)
 				{
-					m_foundResult = true;					
 				}
-				m_originalResult->addContactPoint(normalOnBInWorld,adjustedPointB,m_reportedDistance);
-			}
-		};
 
-		
-		btDummyResult dummy;
-
-///btBoxShape is an exception: its vertices are created WITH margin so don't subtract it
-
-		btScalar min0Margin = min0->getShapeType()==BOX_SHAPE_PROXYTYPE? 0.f : min0->getMargin();
-		btScalar min1Margin = min1->getShapeType()==BOX_SHAPE_PROXYTYPE? 0.f : min1->getMargin();
-
-		btWithoutMarginResult	withoutMargin(resultOut, min0Margin,min1Margin);
-
-		btPolyhedralConvexShape* polyhedronA = (btPolyhedralConvexShape*) min0;
-		btPolyhedralConvexShape* polyhedronB = (btPolyhedralConvexShape*) min1;
-		if (polyhedronA->getConvexPolyhedron() && polyhedronB->getConvexPolyhedron())
-		{
-
-
-			
-
-			btScalar threshold = m_manifoldPtr->getContactBreakingThreshold();
-
-			btScalar minDist = -1e30f;
-			btVector3 sepNormalWorldSpace;
-			bool foundSepAxis  = true;
-
-			if (dispatchInfo.m_enableSatConvex)
-			{
-				foundSepAxis = btPolyhedralContactClipping::findSeparatingAxis(
-					*polyhedronA->getConvexPolyhedron(), *polyhedronB->getConvexPolyhedron(),
-					body0Wrap->getWorldTransform(), 
-					body1Wrap->getWorldTransform(),
-					sepNormalWorldSpace,*resultOut);
-			} else
-			{
-#ifdef ZERO_MARGIN
-				gjkPairDetector.setIgnoreMargin(true);
-				gjkPairDetector.getClosestPoints(input,*resultOut,dispatchInfo.m_debugDraw);
-#else
-
-
-				gjkPairDetector.getClosestPoints(input,withoutMargin,dispatchInfo.m_debugDraw);
-				//gjkPairDetector.getClosestPoints(input,dummy,dispatchInfo.m_debugDraw);
-#endif //ZERO_MARGIN
-				//btScalar l2 = gjkPairDetector.getCachedSeparatingAxis().length2();
-				//if (l2>SIMD_EPSILON)
+				virtual void setShapeIdentifiersA(int partId0, int index0) {}
+				virtual void setShapeIdentifiersB(int partId1, int index1) {}
+				virtual void addContactPoint(const btVector3& normalOnBInWorld, const btVector3& pointInWorld, btScalar depth)
 				{
-					sepNormalWorldSpace = withoutMargin.m_reportedNormalOnWorld;//gjkPairDetector.getCachedSeparatingAxis()*(1.f/l2);
-					//minDist = -1e30f;//gjkPairDetector.getCachedSeparatingDistance();
-					minDist = withoutMargin.m_reportedDistance;//gjkPairDetector.getCachedSeparatingDistance()+min0->getMargin()+min1->getMargin();
-	
-#ifdef ZERO_MARGIN
-					foundSepAxis = true;//gjkPairDetector.getCachedSeparatingDistance()<0.f;
-#else
-					foundSepAxis = withoutMargin.m_foundResult && minDist<0;//-(min0->getMargin()+min1->getMargin());
-#endif
+					m_hasContact = true;
+					m_normalOnBInWorld = normalOnBInWorld;
+					m_pointInWorld = pointInWorld;
+					m_depth = depth;
 				}
-			}
-			if (foundSepAxis)
+			};
+
+			struct btWithoutMarginResult : public btDiscreteCollisionDetectorInterface::Result
 			{
-				
-//				printf("sepNormalWorldSpace=%f,%f,%f\n",sepNormalWorldSpace.getX(),sepNormalWorldSpace.getY(),sepNormalWorldSpace.getZ());
+				btDiscreteCollisionDetectorInterface::Result* m_originalResult;
+				btVector3 m_reportedNormalOnWorld;
+				btScalar m_marginOnA;
+				btScalar m_marginOnB;
+				btScalar m_reportedDistance;
 
-				worldVertsB1.resize(0);
-				btPolyhedralContactClipping::clipHullAgainstHull(sepNormalWorldSpace, *polyhedronA->getConvexPolyhedron(), *polyhedronB->getConvexPolyhedron(),
-					body0Wrap->getWorldTransform(), 
-																 body1Wrap->getWorldTransform(), minDist-threshold, threshold, worldVertsB1,worldVertsB2,
-																 *resultOut);
- 				
-			}
-			if (m_ownManifold)
+				bool m_foundResult;
+				btWithoutMarginResult(btDiscreteCollisionDetectorInterface::Result* result, btScalar marginOnA, btScalar marginOnB)
+					: m_originalResult(result),
+					  m_marginOnA(marginOnA),
+					  m_marginOnB(marginOnB),
+					  m_foundResult(false)
+				{
+				}
+
+				virtual void setShapeIdentifiersA(int partId0, int index0) {}
+				virtual void setShapeIdentifiersB(int partId1, int index1) {}
+				virtual void addContactPoint(const btVector3& normalOnBInWorld, const btVector3& pointInWorldOrg, btScalar depthOrg)
+				{
+					m_reportedDistance = depthOrg;
+					m_reportedNormalOnWorld = normalOnBInWorld;
+
+					btVector3 adjustedPointB = pointInWorldOrg - normalOnBInWorld * m_marginOnB;
+					m_reportedDistance = depthOrg + (m_marginOnA + m_marginOnB);
+					if (m_reportedDistance < 0.f)
+					{
+						m_foundResult = true;
+					}
+					m_originalResult->addContactPoint(normalOnBInWorld, adjustedPointB, m_reportedDistance);
+				}
+			};
+
+			btDummyResult dummy;
+
+			///btBoxShape is an exception: its vertices are created WITH margin so don't subtract it
+
+			btScalar min0Margin = min0->getShapeType() == BOX_SHAPE_PROXYTYPE ? 0.f : min0->getMargin();
+			btScalar min1Margin = min1->getShapeType() == BOX_SHAPE_PROXYTYPE ? 0.f : min1->getMargin();
+
+			btWithoutMarginResult withoutMargin(resultOut, min0Margin, min1Margin);
+
+			btPolyhedralConvexShape* polyhedronA = (btPolyhedralConvexShape*)min0;
+			btPolyhedralConvexShape* polyhedronB = (btPolyhedralConvexShape*)min1;
+			if (polyhedronA->getConvexPolyhedron() && polyhedronB->getConvexPolyhedron())
 			{
-				resultOut->refreshContactPoints();
-			}
-			return;
-
-		} else
-		{
-
-
-			//we can also deal with convex versus triangle (without connectivity data)
-			if (dispatchInfo.m_enableSatConvex && polyhedronA->getConvexPolyhedron() && polyhedronB->getShapeType()==TRIANGLE_SHAPE_PROXYTYPE)
-			{
-
-
-				btVertexArray worldSpaceVertices;
-				btTriangleShape* tri = (btTriangleShape*)polyhedronB;
-				worldSpaceVertices.push_back(	body1Wrap->getWorldTransform()*tri->m_vertices1[0]);
-				worldSpaceVertices.push_back(	body1Wrap->getWorldTransform()*tri->m_vertices1[1]);
-				worldSpaceVertices.push_back(	body1Wrap->getWorldTransform()*tri->m_vertices1[2]);
-				
-				//tri->initializePolyhedralFeatures();
-
-				btScalar threshold = m_manifoldPtr->getContactBreakingThreshold();
+				btScalar threshold = m_manifoldPtr->getContactBreakingThreshold()+ resultOut->m_closestPointDistanceThreshold;
 
+				btScalar minDist = -1e30f;
 				btVector3 sepNormalWorldSpace;
-				btScalar minDist =-1e30f;
-				btScalar maxDist = threshold;
-				
-				bool foundSepAxis = false;
-				bool useSatSepNormal = true;
+				bool foundSepAxis = true;
 
-				if (useSatSepNormal)
+				if (dispatchInfo.m_enableSatConvex)
 				{
+					foundSepAxis = btPolyhedralContactClipping::findSeparatingAxis(
+						*polyhedronA->getConvexPolyhedron(), *polyhedronB->getConvexPolyhedron(),
+						body0Wrap->getWorldTransform(),
+						body1Wrap->getWorldTransform(),
+						sepNormalWorldSpace, *resultOut);
+				}
+				else
+				{
+#ifdef ZERO_MARGIN
+					gjkPairDetector.setIgnoreMargin(true);
+					gjkPairDetector.getClosestPoints(input, *resultOut, dispatchInfo.m_debugDraw);
+#else
+
+					gjkPairDetector.getClosestPoints(input, withoutMargin, dispatchInfo.m_debugDraw);
+					//gjkPairDetector.getClosestPoints(input,dummy,dispatchInfo.m_debugDraw);
+#endif  //ZERO_MARGIN                                                    \
+	//btScalar l2 = gjkPairDetector.getCachedSeparatingAxis().length2(); \
+	//if (l2>SIMD_EPSILON)
+					{
+						sepNormalWorldSpace = withoutMargin.m_reportedNormalOnWorld;  //gjkPairDetector.getCachedSeparatingAxis()*(1.f/l2);
+						//minDist = -1e30f;//gjkPairDetector.getCachedSeparatingDistance();
+						minDist = withoutMargin.m_reportedDistance;  //gjkPairDetector.getCachedSeparatingDistance()+min0->getMargin()+min1->getMargin();
+
+#ifdef ZERO_MARGIN
+						foundSepAxis = true;  //gjkPairDetector.getCachedSeparatingDistance()<0.f;
+#else
+						foundSepAxis = withoutMargin.m_foundResult && minDist < 0;  //-(min0->getMargin()+min1->getMargin());
+#endif
+					}
+				}
+				if (foundSepAxis)
+				{
+					//				printf("sepNormalWorldSpace=%f,%f,%f\n",sepNormalWorldSpace.getX(),sepNormalWorldSpace.getY(),sepNormalWorldSpace.getZ());
+
+					worldVertsB1.resize(0);
+					btPolyhedralContactClipping::clipHullAgainstHull(sepNormalWorldSpace, *polyhedronA->getConvexPolyhedron(), *polyhedronB->getConvexPolyhedron(),
+																	 body0Wrap->getWorldTransform(),
+																	 body1Wrap->getWorldTransform(), minDist - threshold, threshold, worldVertsB1, worldVertsB2,
+																	 *resultOut);
+				}
+				if (m_ownManifold)
+				{
+					resultOut->refreshContactPoints();
+				}
+				return;
+			}
+			else
+			{
+				//we can also deal with convex versus triangle (without connectivity data)
+				if (dispatchInfo.m_enableSatConvex && polyhedronA->getConvexPolyhedron() && polyhedronB->getShapeType() == TRIANGLE_SHAPE_PROXYTYPE)
+				{
+					btVertexArray worldSpaceVertices;
+					btTriangleShape* tri = (btTriangleShape*)polyhedronB;
+					worldSpaceVertices.push_back(body1Wrap->getWorldTransform() * tri->m_vertices1[0]);
+					worldSpaceVertices.push_back(body1Wrap->getWorldTransform() * tri->m_vertices1[1]);
+					worldSpaceVertices.push_back(body1Wrap->getWorldTransform() * tri->m_vertices1[2]);
+
+					//tri->initializePolyhedralFeatures();
+
+					btScalar threshold = m_manifoldPtr->getContactBreakingThreshold()+ resultOut->m_closestPointDistanceThreshold;
+
+					btVector3 sepNormalWorldSpace;
+					btScalar minDist = -1e30f;
+					btScalar maxDist = threshold;
+
+					bool foundSepAxis = false;
+					bool useSatSepNormal = true;
+
+					if (useSatSepNormal)
+					{
 #if 0
 					if (0)
 					{
@@ -610,113 +565,107 @@ void btConvexConvexAlgorithm ::processCollision (const btCollisionObjectWrapper*
 						polyhedronB->initializePolyhedralFeatures();
 					} else
 #endif
-					{
-
-						btVector3 uniqueEdges[3] = {tri->m_vertices1[1]-tri->m_vertices1[0],
-							tri->m_vertices1[2]-tri->m_vertices1[1],
-							tri->m_vertices1[0]-tri->m_vertices1[2]};
-
-						uniqueEdges[0].normalize();
-						uniqueEdges[1].normalize();
-						uniqueEdges[2].normalize();
-
-						btConvexPolyhedron polyhedron;
-						polyhedron.m_vertices.push_back(tri->m_vertices1[2]);
-						polyhedron.m_vertices.push_back(tri->m_vertices1[0]);
-						polyhedron.m_vertices.push_back(tri->m_vertices1[1]);
-						
-						
 						{
-							btFace combinedFaceA;
-							combinedFaceA.m_indices.push_back(0);
-							combinedFaceA.m_indices.push_back(1);
-							combinedFaceA.m_indices.push_back(2);
-							btVector3 faceNormal = uniqueEdges[0].cross(uniqueEdges[1]);
-							faceNormal.normalize();
-							btScalar planeEq=1e30f;
-							for (int v=0;v<combinedFaceA.m_indices.size();v++)
+							btVector3 uniqueEdges[3] = {tri->m_vertices1[1] - tri->m_vertices1[0],
+														tri->m_vertices1[2] - tri->m_vertices1[1],
+														tri->m_vertices1[0] - tri->m_vertices1[2]};
+
+							uniqueEdges[0].normalize();
+							uniqueEdges[1].normalize();
+							uniqueEdges[2].normalize();
+
+							btConvexPolyhedron polyhedron;
+							polyhedron.m_vertices.push_back(tri->m_vertices1[2]);
+							polyhedron.m_vertices.push_back(tri->m_vertices1[0]);
+							polyhedron.m_vertices.push_back(tri->m_vertices1[1]);
+
 							{
-								btScalar eq = tri->m_vertices1[combinedFaceA.m_indices[v]].dot(faceNormal);
-								if (planeEq>eq)
+								btFace combinedFaceA;
+								combinedFaceA.m_indices.push_back(0);
+								combinedFaceA.m_indices.push_back(1);
+								combinedFaceA.m_indices.push_back(2);
+								btVector3 faceNormal = uniqueEdges[0].cross(uniqueEdges[1]);
+								faceNormal.normalize();
+								btScalar planeEq = 1e30f;
+								for (int v = 0; v < combinedFaceA.m_indices.size(); v++)
 								{
-									planeEq=eq;
+									btScalar eq = tri->m_vertices1[combinedFaceA.m_indices[v]].dot(faceNormal);
+									if (planeEq > eq)
+									{
+										planeEq = eq;
+									}
 								}
+								combinedFaceA.m_plane[0] = faceNormal[0];
+								combinedFaceA.m_plane[1] = faceNormal[1];
+								combinedFaceA.m_plane[2] = faceNormal[2];
+								combinedFaceA.m_plane[3] = -planeEq;
+								polyhedron.m_faces.push_back(combinedFaceA);
 							}
-							combinedFaceA.m_plane[0] = faceNormal[0];
-							combinedFaceA.m_plane[1] = faceNormal[1];
-							combinedFaceA.m_plane[2] = faceNormal[2];
-							combinedFaceA.m_plane[3] = -planeEq;
-							polyhedron.m_faces.push_back(combinedFaceA);
-						}
-						{
-							btFace combinedFaceB;
-							combinedFaceB.m_indices.push_back(0);
-							combinedFaceB.m_indices.push_back(2);
-							combinedFaceB.m_indices.push_back(1);
-							btVector3 faceNormal = -uniqueEdges[0].cross(uniqueEdges[1]);
-							faceNormal.normalize();
-							btScalar planeEq=1e30f;
-							for (int v=0;v<combinedFaceB.m_indices.size();v++)
 							{
-								btScalar eq = tri->m_vertices1[combinedFaceB.m_indices[v]].dot(faceNormal);
-								if (planeEq>eq)
+								btFace combinedFaceB;
+								combinedFaceB.m_indices.push_back(0);
+								combinedFaceB.m_indices.push_back(2);
+								combinedFaceB.m_indices.push_back(1);
+								btVector3 faceNormal = -uniqueEdges[0].cross(uniqueEdges[1]);
+								faceNormal.normalize();
+								btScalar planeEq = 1e30f;
+								for (int v = 0; v < combinedFaceB.m_indices.size(); v++)
 								{
-									planeEq=eq;
+									btScalar eq = tri->m_vertices1[combinedFaceB.m_indices[v]].dot(faceNormal);
+									if (planeEq > eq)
+									{
+										planeEq = eq;
+									}
 								}
+
+								combinedFaceB.m_plane[0] = faceNormal[0];
+								combinedFaceB.m_plane[1] = faceNormal[1];
+								combinedFaceB.m_plane[2] = faceNormal[2];
+								combinedFaceB.m_plane[3] = -planeEq;
+								polyhedron.m_faces.push_back(combinedFaceB);
 							}
 
-							combinedFaceB.m_plane[0] = faceNormal[0];
-							combinedFaceB.m_plane[1] = faceNormal[1];
-							combinedFaceB.m_plane[2] = faceNormal[2];
-							combinedFaceB.m_plane[3] = -planeEq;
-							polyhedron.m_faces.push_back(combinedFaceB);
+							polyhedron.m_uniqueEdges.push_back(uniqueEdges[0]);
+							polyhedron.m_uniqueEdges.push_back(uniqueEdges[1]);
+							polyhedron.m_uniqueEdges.push_back(uniqueEdges[2]);
+							polyhedron.initialize2();
+
+							polyhedronB->setPolyhedralFeatures(polyhedron);
 						}
 
-						
-						polyhedron.m_uniqueEdges.push_back(uniqueEdges[0]);
-						polyhedron.m_uniqueEdges.push_back(uniqueEdges[1]);
-						polyhedron.m_uniqueEdges.push_back(uniqueEdges[2]);
-						polyhedron.initialize2();
-
-						polyhedronB->setPolyhedralFeatures(polyhedron);
+						foundSepAxis = btPolyhedralContactClipping::findSeparatingAxis(
+							*polyhedronA->getConvexPolyhedron(), *polyhedronB->getConvexPolyhedron(),
+							body0Wrap->getWorldTransform(),
+							body1Wrap->getWorldTransform(),
+							sepNormalWorldSpace, *resultOut);
+						//	 printf("sepNormalWorldSpace=%f,%f,%f\n",sepNormalWorldSpace.getX(),sepNormalWorldSpace.getY(),sepNormalWorldSpace.getZ());
 					}
-
-					
-
-					 foundSepAxis = btPolyhedralContactClipping::findSeparatingAxis(
-						*polyhedronA->getConvexPolyhedron(), *polyhedronB->getConvexPolyhedron(),
-						body0Wrap->getWorldTransform(), 
-						body1Wrap->getWorldTransform(),
-						sepNormalWorldSpace,*resultOut);
-				//	 printf("sepNormalWorldSpace=%f,%f,%f\n",sepNormalWorldSpace.getX(),sepNormalWorldSpace.getY(),sepNormalWorldSpace.getZ());
-
-				} 
-				else
-				{
+					else
+					{
 #ifdef ZERO_MARGIN
-					gjkPairDetector.setIgnoreMargin(true);
-					gjkPairDetector.getClosestPoints(input,*resultOut,dispatchInfo.m_debugDraw);
+						gjkPairDetector.setIgnoreMargin(true);
+						gjkPairDetector.getClosestPoints(input, *resultOut, dispatchInfo.m_debugDraw);
 #else
-					gjkPairDetector.getClosestPoints(input,dummy,dispatchInfo.m_debugDraw);
-#endif//ZERO_MARGIN
-					
-					if (dummy.m_hasContact && dummy.m_depth<0)
-					{
-						
-						if (foundSepAxis)
+						gjkPairDetector.getClosestPoints(input, dummy, dispatchInfo.m_debugDraw);
+#endif  //ZERO_MARGIN
+
+						if (dummy.m_hasContact && dummy.m_depth < 0)
 						{
-							if (dummy.m_normalOnBInWorld.dot(sepNormalWorldSpace)<0.99)
+							if (foundSepAxis)
 							{
-								printf("?\n");
+								if (dummy.m_normalOnBInWorld.dot(sepNormalWorldSpace) < 0.99)
+								{
+									printf("?\n");
+								}
 							}
-						} else
-						{
-							printf("!\n");
+							else
+							{
+								printf("!\n");
+							}
+							sepNormalWorldSpace.setValue(0, 0, 1);  // = dummy.m_normalOnBInWorld;
+							//minDist = dummy.m_depth;
+							foundSepAxis = true;
 						}
-						sepNormalWorldSpace.setValue(0,0,1);// = dummy.m_normalOnBInWorld;
-						//minDist = dummy.m_depth;
-						foundSepAxis = true;
-					}
 #if 0
 					btScalar l2 = gjkPairDetector.getCachedSeparatingAxis().length2();
 					if (l2>SIMD_EPSILON)
@@ -728,145 +677,131 @@ void btConvexConvexAlgorithm ::processCollision (const btCollisionObjectWrapper*
 						foundSepAxis = true;
 					}
 #endif
-				}
+					}
 
-				
-			if (foundSepAxis)
-			{
-				worldVertsB2.resize(0);
-				btPolyhedralContactClipping::clipFaceAgainstHull(sepNormalWorldSpace, *polyhedronA->getConvexPolyhedron(), 
-					body0Wrap->getWorldTransform(), worldSpaceVertices, worldVertsB2,minDist-threshold, maxDist, *resultOut);
-			}
-				
-				
-				if (m_ownManifold)
-				{
-					resultOut->refreshContactPoints();
-				}
-				
-				return;
-			}
+					if (foundSepAxis)
+					{
+						worldVertsB2.resize(0);
+						btPolyhedralContactClipping::clipFaceAgainstHull(sepNormalWorldSpace, *polyhedronA->getConvexPolyhedron(),
+																		 body0Wrap->getWorldTransform(), worldSpaceVertices, worldVertsB2, minDist - threshold, maxDist, *resultOut);
+					}
 
-			
+					if (m_ownManifold)
+					{
+						resultOut->refreshContactPoints();
+					}
+
+					return;
+				}
+			}
 		}
 
+		gjkPairDetector.getClosestPoints(input, *resultOut, dispatchInfo.m_debugDraw);
 
-	}
-	
-	gjkPairDetector.getClosestPoints(input,*resultOut,dispatchInfo.m_debugDraw);
+		//now perform 'm_numPerturbationIterations' collision queries with the perturbated collision objects
 
-	//now perform 'm_numPerturbationIterations' collision queries with the perturbated collision objects
-	
-	//perform perturbation when more then 'm_minimumPointsPerturbationThreshold' points
-	if (m_numPerturbationIterations && resultOut->getPersistentManifold()->getNumContacts() < m_minimumPointsPerturbationThreshold)
-	{
-		
-		int i;
-		btVector3 v0,v1;
-		btVector3 sepNormalWorldSpace;
-		btScalar l2 = gjkPairDetector.getCachedSeparatingAxis().length2();
-	
-		if (l2>SIMD_EPSILON)
+		//perform perturbation when more then 'm_minimumPointsPerturbationThreshold' points
+		if (m_numPerturbationIterations && resultOut->getPersistentManifold()->getNumContacts() < m_minimumPointsPerturbationThreshold)
 		{
-			sepNormalWorldSpace = gjkPairDetector.getCachedSeparatingAxis()*(1.f/l2);
-			
-			btPlaneSpace1(sepNormalWorldSpace,v0,v1);
+			int i;
+			btVector3 v0, v1;
+			btVector3 sepNormalWorldSpace;
+			btScalar l2 = gjkPairDetector.getCachedSeparatingAxis().length2();
 
+			if (l2 > SIMD_EPSILON)
+			{
+				sepNormalWorldSpace = gjkPairDetector.getCachedSeparatingAxis() * (1.f / l2);
 
-			bool perturbeA = true;
-			const btScalar angleLimit = 0.125f * SIMD_PI;
-			btScalar perturbeAngle;
-			btScalar radiusA = min0->getAngularMotionDisc();
-			btScalar radiusB = min1->getAngularMotionDisc();
-			if (radiusA < radiusB)
-			{
-				perturbeAngle = gContactBreakingThreshold /radiusA;
-				perturbeA = true;
-			} else
-			{
-				perturbeAngle = gContactBreakingThreshold / radiusB;
-				perturbeA = false;
-			}
-			if ( perturbeAngle > angleLimit ) 
+				btPlaneSpace1(sepNormalWorldSpace, v0, v1);
+
+				bool perturbeA = true;
+				const btScalar angleLimit = 0.125f * SIMD_PI;
+				btScalar perturbeAngle;
+				btScalar radiusA = min0->getAngularMotionDisc();
+				btScalar radiusB = min1->getAngularMotionDisc();
+				if (radiusA < radiusB)
+				{
+					perturbeAngle = gContactBreakingThreshold / radiusA;
+					perturbeA = true;
+				}
+				else
+				{
+					perturbeAngle = gContactBreakingThreshold / radiusB;
+					perturbeA = false;
+				}
+				if (perturbeAngle > angleLimit)
 					perturbeAngle = angleLimit;
 
-			btTransform unPerturbedTransform;
-			if (perturbeA)
-			{
-				unPerturbedTransform = input.m_transformA;
-			} else
-			{
-				unPerturbedTransform = input.m_transformB;
-			}
-			
-			for ( i=0;i<m_numPerturbationIterations;i++)
-			{
-				if (v0.length2()>SIMD_EPSILON)
-				{
-				btQuaternion perturbeRot(v0,perturbeAngle);
-				btScalar iterationAngle = i*(SIMD_2_PI/btScalar(m_numPerturbationIterations));
-				btQuaternion rotq(sepNormalWorldSpace,iterationAngle);
-				
-				
+				btTransform unPerturbedTransform;
 				if (perturbeA)
 				{
-					input.m_transformA.setBasis(  btMatrix3x3(rotq.inverse()*perturbeRot*rotq)*body0Wrap->getWorldTransform().getBasis());
-					input.m_transformB = body1Wrap->getWorldTransform();
-	#ifdef DEBUG_CONTACTS
-					dispatchInfo.m_debugDraw->drawTransform(input.m_transformA,10.0);
-	#endif //DEBUG_CONTACTS
-				} else
-				{
-					input.m_transformA = body0Wrap->getWorldTransform();
-					input.m_transformB.setBasis( btMatrix3x3(rotq.inverse()*perturbeRot*rotq)*body1Wrap->getWorldTransform().getBasis());
-	#ifdef DEBUG_CONTACTS
-					dispatchInfo.m_debugDraw->drawTransform(input.m_transformB,10.0);
-	#endif
+					unPerturbedTransform = input.m_transformA;
 				}
-				
-				btPerturbedContactResult perturbedResultOut(resultOut,input.m_transformA,input.m_transformB,unPerturbedTransform,perturbeA,dispatchInfo.m_debugDraw);
-				gjkPairDetector.getClosestPoints(input,perturbedResultOut,dispatchInfo.m_debugDraw);
+				else
+				{
+					unPerturbedTransform = input.m_transformB;
+				}
+
+				for (i = 0; i < m_numPerturbationIterations; i++)
+				{
+					if (v0.length2() > SIMD_EPSILON)
+					{
+						btQuaternion perturbeRot(v0, perturbeAngle);
+						btScalar iterationAngle = i * (SIMD_2_PI / btScalar(m_numPerturbationIterations));
+						btQuaternion rotq(sepNormalWorldSpace, iterationAngle);
+
+						if (perturbeA)
+						{
+							input.m_transformA.setBasis(btMatrix3x3(rotq.inverse() * perturbeRot * rotq) * body0Wrap->getWorldTransform().getBasis());
+							input.m_transformB = body1Wrap->getWorldTransform();
+#ifdef DEBUG_CONTACTS
+							dispatchInfo.m_debugDraw->drawTransform(input.m_transformA, 10.0);
+#endif  //DEBUG_CONTACTS
+						}
+						else
+						{
+							input.m_transformA = body0Wrap->getWorldTransform();
+							input.m_transformB.setBasis(btMatrix3x3(rotq.inverse() * perturbeRot * rotq) * body1Wrap->getWorldTransform().getBasis());
+#ifdef DEBUG_CONTACTS
+							dispatchInfo.m_debugDraw->drawTransform(input.m_transformB, 10.0);
+#endif
+						}
+
+						btPerturbedContactResult perturbedResultOut(resultOut, input.m_transformA, input.m_transformB, unPerturbedTransform, perturbeA, dispatchInfo.m_debugDraw);
+						gjkPairDetector.getClosestPoints(input, perturbedResultOut, dispatchInfo.m_debugDraw);
+					}
 				}
 			}
 		}
-	}
-
-	
 
 #ifdef USE_SEPDISTANCE_UTIL2
-	if (dispatchInfo.m_useConvexConservativeDistanceUtil && (sepDist>SIMD_EPSILON))
-	{
-		m_sepDistance.initSeparatingDistance(gjkPairDetector.getCachedSeparatingAxis(),sepDist,body0->getWorldTransform(),body1->getWorldTransform());
-	}
-#endif //USE_SEPDISTANCE_UTIL2
-
-
+		if (dispatchInfo.m_useConvexConservativeDistanceUtil && (sepDist > SIMD_EPSILON))
+		{
+			m_sepDistance.initSeparatingDistance(gjkPairDetector.getCachedSeparatingAxis(), sepDist, body0->getWorldTransform(), body1->getWorldTransform());
+		}
+#endif  //USE_SEPDISTANCE_UTIL2
 	}
 
 	if (m_ownManifold)
 	{
 		resultOut->refreshContactPoints();
 	}
-
 }
 
-
-
 bool disableCcd = false;
-btScalar	btConvexConvexAlgorithm::calculateTimeOfImpact(btCollisionObject* col0,btCollisionObject* col1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+btScalar btConvexConvexAlgorithm::calculateTimeOfImpact(btCollisionObject* col0, btCollisionObject* col1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	(void)resultOut;
 	(void)dispatchInfo;
 	///Rather then checking ALL pairs, only calculate TOI when motion exceeds threshold
-    
+
 	///Linear motion for one of objects needs to exceed m_ccdSquareMotionThreshold
 	///col0->m_worldTransform,
 	btScalar resultFraction = btScalar(1.);
 
-
 	btScalar squareMot0 = (col0->getInterpolationWorldTransform().getOrigin() - col0->getWorldTransform().getOrigin()).length2();
 	btScalar squareMot1 = (col1->getInterpolationWorldTransform().getOrigin() - col1->getWorldTransform().getOrigin()).length2();
-    
+
 	if (squareMot0 < col0->getCcdSquareMotionThreshold() &&
 		squareMot1 < col1->getCcdSquareMotionThreshold())
 		return resultFraction;
@@ -874,77 +809,65 @@ btScalar	btConvexConvexAlgorithm::calculateTimeOfImpact(btCollisionObject* col0,
 	if (disableCcd)
 		return btScalar(1.);
 
-
 	//An adhoc way of testing the Continuous Collision Detection algorithms
 	//One object is approximated as a sphere, to simplify things
 	//Starting in penetration should report no time of impact
 	//For proper CCD, better accuracy and handling of 'allowed' penetration should be added
 	//also the mainloop of the physics should have a kind of toi queue (something like Brian Mirtich's application of Timewarp for Rigidbodies)
 
-		
 	/// Convex0 against sphere for Convex1
 	{
 		btConvexShape* convex0 = static_cast<btConvexShape*>(col0->getCollisionShape());
 
-		btSphereShape	sphere1(col1->getCcdSweptSphereRadius()); //todo: allow non-zero sphere sizes, for better approximation
+		btSphereShape sphere1(col1->getCcdSweptSphereRadius());  //todo: allow non-zero sphere sizes, for better approximation
 		btConvexCast::CastResult result;
 		btVoronoiSimplexSolver voronoiSimplex;
 		//SubsimplexConvexCast ccd0(&sphere,min0,&voronoiSimplex);
 		///Simplification, one object is simplified as a sphere
-		btGjkConvexCast ccd1( convex0 ,&sphere1,&voronoiSimplex);
+		btGjkConvexCast ccd1(convex0, &sphere1, &voronoiSimplex);
 		//ContinuousConvexCollision ccd(min0,min1,&voronoiSimplex,0);
-		if (ccd1.calcTimeOfImpact(col0->getWorldTransform(),col0->getInterpolationWorldTransform(),
-			col1->getWorldTransform(),col1->getInterpolationWorldTransform(),result))
+		if (ccd1.calcTimeOfImpact(col0->getWorldTransform(), col0->getInterpolationWorldTransform(),
+								  col1->getWorldTransform(), col1->getInterpolationWorldTransform(), result))
 		{
-		
 			//store result.m_fraction in both bodies
-		
-			if (col0->getHitFraction()> result.m_fraction)
-				col0->setHitFraction( result.m_fraction );
+
+			if (col0->getHitFraction() > result.m_fraction)
+				col0->setHitFraction(result.m_fraction);
 
 			if (col1->getHitFraction() > result.m_fraction)
-				col1->setHitFraction( result.m_fraction);
+				col1->setHitFraction(result.m_fraction);
 
 			if (resultFraction > result.m_fraction)
 				resultFraction = result.m_fraction;
-
 		}
-		
-		
-
-
 	}
 
 	/// Sphere (for convex0) against Convex1
 	{
 		btConvexShape* convex1 = static_cast<btConvexShape*>(col1->getCollisionShape());
 
-		btSphereShape	sphere0(col0->getCcdSweptSphereRadius()); //todo: allow non-zero sphere sizes, for better approximation
+		btSphereShape sphere0(col0->getCcdSweptSphereRadius());  //todo: allow non-zero sphere sizes, for better approximation
 		btConvexCast::CastResult result;
 		btVoronoiSimplexSolver voronoiSimplex;
 		//SubsimplexConvexCast ccd0(&sphere,min0,&voronoiSimplex);
 		///Simplification, one object is simplified as a sphere
-		btGjkConvexCast ccd1(&sphere0,convex1,&voronoiSimplex);
+		btGjkConvexCast ccd1(&sphere0, convex1, &voronoiSimplex);
 		//ContinuousConvexCollision ccd(min0,min1,&voronoiSimplex,0);
-		if (ccd1.calcTimeOfImpact(col0->getWorldTransform(),col0->getInterpolationWorldTransform(),
-			col1->getWorldTransform(),col1->getInterpolationWorldTransform(),result))
+		if (ccd1.calcTimeOfImpact(col0->getWorldTransform(), col0->getInterpolationWorldTransform(),
+								  col1->getWorldTransform(), col1->getInterpolationWorldTransform(), result))
 		{
-		
 			//store result.m_fraction in both bodies
-		
-			if (col0->getHitFraction()	> result.m_fraction)
-				col0->setHitFraction( result.m_fraction);
+
+			if (col0->getHitFraction() > result.m_fraction)
+				col0->setHitFraction(result.m_fraction);
 
 			if (col1->getHitFraction() > result.m_fraction)
-				col1->setHitFraction( result.m_fraction);
+				col1->setHitFraction(result.m_fraction);
 
 			if (resultFraction > result.m_fraction)
 				resultFraction = result.m_fraction;
-
 		}
 	}
-	
+
 	return resultFraction;
-
 }
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexConvexAlgorithm.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexConvexAlgorithm.h
index cd75ba12d7f..eac5b4d8248 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexConvexAlgorithm.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexConvexAlgorithm.h
@@ -23,7 +23,7 @@ subject to the following restrictions:
 #include "BulletCollision/NarrowPhaseCollision/btVoronoiSimplexSolver.h"
 #include "btCollisionCreateFunc.h"
 #include "btCollisionDispatcher.h"
-#include "LinearMath/btTransformUtil.h" //for btConvexSeparatingDistanceUtil
+#include "LinearMath/btTransformUtil.h"  //for btConvexSeparatingDistanceUtil
 #include "BulletCollision/NarrowPhaseCollision/btPolyhedralContactClipping.h"
 
 class btConvexPenetrationDepthSolver;
@@ -41,69 +41,61 @@ class btConvexPenetrationDepthSolver;
 class btConvexConvexAlgorithm : public btActivatingCollisionAlgorithm
 {
 #ifdef USE_SEPDISTANCE_UTIL2
-	btConvexSeparatingDistanceUtil	m_sepDistance;
+	btConvexSeparatingDistanceUtil m_sepDistance;
 #endif
 	btConvexPenetrationDepthSolver* m_pdSolver;
 
 	btVertexArray worldVertsB1;
 	btVertexArray worldVertsB2;
-	
-	bool	m_ownManifold;
-	btPersistentManifold*	m_manifoldPtr;
-	bool			m_lowLevelOfDetail;
-	
+
+	bool m_ownManifold;
+	btPersistentManifold* m_manifoldPtr;
+	bool m_lowLevelOfDetail;
+
 	int m_numPerturbationIterations;
 	int m_minimumPointsPerturbationThreshold;
 
-
 	///cache separating vector to speedup collision detection
-	
 
 public:
-
-	btConvexConvexAlgorithm(btPersistentManifold* mf,const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap, btConvexPenetrationDepthSolver* pdSolver, int numPerturbationIterations, int minimumPointsPerturbationThreshold);
+	btConvexConvexAlgorithm(btPersistentManifold* mf, const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, btConvexPenetrationDepthSolver* pdSolver, int numPerturbationIterations, int minimumPointsPerturbationThreshold);
 
 	virtual ~btConvexConvexAlgorithm();
 
-	virtual void processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual void processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual	void	getAllContactManifolds(btManifoldArray&	manifoldArray)
+	virtual void getAllContactManifolds(btManifoldArray& manifoldArray)
 	{
 		///should we use m_ownManifold to avoid adding duplicates?
 		if (m_manifoldPtr && m_ownManifold)
 			manifoldArray.push_back(m_manifoldPtr);
 	}
 
+	void setLowLevelOfDetail(bool useLowLevel);
 
-	void	setLowLevelOfDetail(bool useLowLevel);
-
-
-	const btPersistentManifold*	getManifold()
+	const btPersistentManifold* getManifold()
 	{
 		return m_manifoldPtr;
 	}
 
-	struct CreateFunc :public 	btCollisionAlgorithmCreateFunc
+	struct CreateFunc : public btCollisionAlgorithmCreateFunc
 	{
-
-		btConvexPenetrationDepthSolver*		m_pdSolver;
+		btConvexPenetrationDepthSolver* m_pdSolver;
 		int m_numPerturbationIterations;
 		int m_minimumPointsPerturbationThreshold;
 
 		CreateFunc(btConvexPenetrationDepthSolver* pdSolver);
-		
+
 		virtual ~CreateFunc();
 
-		virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+		virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 		{
 			void* mem = ci.m_dispatcher1->allocateCollisionAlgorithm(sizeof(btConvexConvexAlgorithm));
-			return new(mem) btConvexConvexAlgorithm(ci.m_manifold,ci,body0Wrap,body1Wrap,m_pdSolver,m_numPerturbationIterations,m_minimumPointsPerturbationThreshold);
+			return new (mem) btConvexConvexAlgorithm(ci.m_manifold, ci, body0Wrap, body1Wrap, m_pdSolver, m_numPerturbationIterations, m_minimumPointsPerturbationThreshold);
 		}
 	};
-
-
 };
 
-#endif //BT_CONVEX_CONVEX_ALGORITHM_H
+#endif  //BT_CONVEX_CONVEX_ALGORITHM_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexPlaneCollisionAlgorithm.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexPlaneCollisionAlgorithm.cpp
index cce2d95bcf9..ba1bc06b69a 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexPlaneCollisionAlgorithm.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexPlaneCollisionAlgorithm.cpp
@@ -23,25 +23,24 @@ subject to the following restrictions:
 
 //#include <stdio.h>
 
-btConvexPlaneCollisionAlgorithm::btConvexPlaneCollisionAlgorithm(btPersistentManifold* mf,const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* col0Wrap,const btCollisionObjectWrapper* col1Wrap, bool isSwapped, int numPerturbationIterations,int minimumPointsPerturbationThreshold)
-: btCollisionAlgorithm(ci),
-m_ownManifold(false),
-m_manifoldPtr(mf),
-m_isSwapped(isSwapped),
-m_numPerturbationIterations(numPerturbationIterations),
-m_minimumPointsPerturbationThreshold(minimumPointsPerturbationThreshold)
+btConvexPlaneCollisionAlgorithm::btConvexPlaneCollisionAlgorithm(btPersistentManifold* mf, const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* col0Wrap, const btCollisionObjectWrapper* col1Wrap, bool isSwapped, int numPerturbationIterations, int minimumPointsPerturbationThreshold)
+	: btCollisionAlgorithm(ci),
+	  m_ownManifold(false),
+	  m_manifoldPtr(mf),
+	  m_isSwapped(isSwapped),
+	  m_numPerturbationIterations(numPerturbationIterations),
+	  m_minimumPointsPerturbationThreshold(minimumPointsPerturbationThreshold)
 {
-	const btCollisionObjectWrapper* convexObjWrap = m_isSwapped? col1Wrap : col0Wrap;
-	const btCollisionObjectWrapper* planeObjWrap = m_isSwapped? col0Wrap : col1Wrap;
+	const btCollisionObjectWrapper* convexObjWrap = m_isSwapped ? col1Wrap : col0Wrap;
+	const btCollisionObjectWrapper* planeObjWrap = m_isSwapped ? col0Wrap : col1Wrap;
 
-	if (!m_manifoldPtr && m_dispatcher->needsCollision(convexObjWrap->getCollisionObject(),planeObjWrap->getCollisionObject()))
+	if (!m_manifoldPtr && m_dispatcher->needsCollision(convexObjWrap->getCollisionObject(), planeObjWrap->getCollisionObject()))
 	{
-		m_manifoldPtr = m_dispatcher->getNewManifold(convexObjWrap->getCollisionObject(),planeObjWrap->getCollisionObject());
+		m_manifoldPtr = m_dispatcher->getNewManifold(convexObjWrap->getCollisionObject(), planeObjWrap->getCollisionObject());
 		m_ownManifold = true;
 	}
 }
 
-
 btConvexPlaneCollisionAlgorithm::~btConvexPlaneCollisionAlgorithm()
 {
 	if (m_ownManifold)
@@ -51,32 +50,32 @@ btConvexPlaneCollisionAlgorithm::~btConvexPlaneCollisionAlgorithm()
 	}
 }
 
-void btConvexPlaneCollisionAlgorithm::collideSingleContact (const btQuaternion& perturbeRot, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+void btConvexPlaneCollisionAlgorithm::collideSingleContact(const btQuaternion& perturbeRot, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
-    const btCollisionObjectWrapper* convexObjWrap = m_isSwapped? body1Wrap : body0Wrap;
-	const btCollisionObjectWrapper* planeObjWrap = m_isSwapped? body0Wrap: body1Wrap;
+	const btCollisionObjectWrapper* convexObjWrap = m_isSwapped ? body1Wrap : body0Wrap;
+	const btCollisionObjectWrapper* planeObjWrap = m_isSwapped ? body0Wrap : body1Wrap;
 
-	btConvexShape* convexShape = (btConvexShape*) convexObjWrap->getCollisionShape();
-	btStaticPlaneShape* planeShape = (btStaticPlaneShape*) planeObjWrap->getCollisionShape();
+	btConvexShape* convexShape = (btConvexShape*)convexObjWrap->getCollisionShape();
+	btStaticPlaneShape* planeShape = (btStaticPlaneShape*)planeObjWrap->getCollisionShape();
 
-    bool hasCollision = false;
+	bool hasCollision = false;
 	const btVector3& planeNormal = planeShape->getPlaneNormal();
 	const btScalar& planeConstant = planeShape->getPlaneConstant();
-	
+
 	btTransform convexWorldTransform = convexObjWrap->getWorldTransform();
 	btTransform convexInPlaneTrans;
-	convexInPlaneTrans= planeObjWrap->getWorldTransform().inverse() * convexWorldTransform;
+	convexInPlaneTrans = planeObjWrap->getWorldTransform().inverse() * convexWorldTransform;
 	//now perturbe the convex-world transform
-	convexWorldTransform.getBasis()*=btMatrix3x3(perturbeRot);
+	convexWorldTransform.getBasis() *= btMatrix3x3(perturbeRot);
 	btTransform planeInConvex;
-	planeInConvex= convexWorldTransform.inverse() * planeObjWrap->getWorldTransform();
-	
-	btVector3 vtx = convexShape->localGetSupportingVertex(planeInConvex.getBasis()*-planeNormal);
+	planeInConvex = convexWorldTransform.inverse() * planeObjWrap->getWorldTransform();
+
+	btVector3 vtx = convexShape->localGetSupportingVertex(planeInConvex.getBasis() * -planeNormal);
 
 	btVector3 vtxInPlane = convexInPlaneTrans(vtx);
 	btScalar distance = (planeNormal.dot(vtxInPlane) - planeConstant);
 
-	btVector3 vtxInPlaneProjected = vtxInPlane - distance*planeNormal;
+	btVector3 vtxInPlaneProjected = vtxInPlane - distance * planeNormal;
 	btVector3 vtxInPlaneWorld = planeObjWrap->getWorldTransform() * vtxInPlaneProjected;
 
 	hasCollision = distance < m_manifoldPtr->getContactBreakingThreshold();
@@ -86,70 +85,69 @@ void btConvexPlaneCollisionAlgorithm::collideSingleContact (const btQuaternion&
 		/// report a contact. internally this will be kept persistent, and contact reduction is done
 		btVector3 normalOnSurfaceB = planeObjWrap->getWorldTransform().getBasis() * planeNormal;
 		btVector3 pOnB = vtxInPlaneWorld;
-		resultOut->addContactPoint(normalOnSurfaceB,pOnB,distance);
+		resultOut->addContactPoint(normalOnSurfaceB, pOnB, distance);
 	}
 }
 
-
-void btConvexPlaneCollisionAlgorithm::processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+void btConvexPlaneCollisionAlgorithm::processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	(void)dispatchInfo;
 	if (!m_manifoldPtr)
 		return;
 
-	const btCollisionObjectWrapper* convexObjWrap = m_isSwapped? body1Wrap : body0Wrap;
-	const btCollisionObjectWrapper* planeObjWrap = m_isSwapped? body0Wrap: body1Wrap;
+	const btCollisionObjectWrapper* convexObjWrap = m_isSwapped ? body1Wrap : body0Wrap;
+	const btCollisionObjectWrapper* planeObjWrap = m_isSwapped ? body0Wrap : body1Wrap;
 
-	btConvexShape* convexShape = (btConvexShape*) convexObjWrap->getCollisionShape();
-	btStaticPlaneShape* planeShape = (btStaticPlaneShape*) planeObjWrap->getCollisionShape();
+	btConvexShape* convexShape = (btConvexShape*)convexObjWrap->getCollisionShape();
+	btStaticPlaneShape* planeShape = (btStaticPlaneShape*)planeObjWrap->getCollisionShape();
 
 	bool hasCollision = false;
 	const btVector3& planeNormal = planeShape->getPlaneNormal();
 	const btScalar& planeConstant = planeShape->getPlaneConstant();
 	btTransform planeInConvex;
-	planeInConvex= convexObjWrap->getWorldTransform().inverse() * planeObjWrap->getWorldTransform();
+	planeInConvex = convexObjWrap->getWorldTransform().inverse() * planeObjWrap->getWorldTransform();
 	btTransform convexInPlaneTrans;
-	convexInPlaneTrans= planeObjWrap->getWorldTransform().inverse() * convexObjWrap->getWorldTransform();
+	convexInPlaneTrans = planeObjWrap->getWorldTransform().inverse() * convexObjWrap->getWorldTransform();
 
-	btVector3 vtx = convexShape->localGetSupportingVertex(planeInConvex.getBasis()*-planeNormal);
+	btVector3 vtx = convexShape->localGetSupportingVertex(planeInConvex.getBasis() * -planeNormal);
 	btVector3 vtxInPlane = convexInPlaneTrans(vtx);
 	btScalar distance = (planeNormal.dot(vtxInPlane) - planeConstant);
 
-	btVector3 vtxInPlaneProjected = vtxInPlane - distance*planeNormal;
+	btVector3 vtxInPlaneProjected = vtxInPlane - distance * planeNormal;
 	btVector3 vtxInPlaneWorld = planeObjWrap->getWorldTransform() * vtxInPlaneProjected;
 
-	hasCollision = distance < m_manifoldPtr->getContactBreakingThreshold();
+	hasCollision = distance < m_manifoldPtr->getContactBreakingThreshold()+ resultOut->m_closestPointDistanceThreshold;
 	resultOut->setPersistentManifold(m_manifoldPtr);
 	if (hasCollision)
 	{
 		/// report a contact. internally this will be kept persistent, and contact reduction is done
 		btVector3 normalOnSurfaceB = planeObjWrap->getWorldTransform().getBasis() * planeNormal;
 		btVector3 pOnB = vtxInPlaneWorld;
-		resultOut->addContactPoint(normalOnSurfaceB,pOnB,distance);
+		resultOut->addContactPoint(normalOnSurfaceB, pOnB, distance);
 	}
 
 	//the perturbation algorithm doesn't work well with implicit surfaces such as spheres, cylinder and cones:
 	//they keep on rolling forever because of the additional off-center contact points
 	//so only enable the feature for polyhedral shapes (btBoxShape, btConvexHullShape etc)
-	if (convexShape->isPolyhedral() && resultOut->getPersistentManifold()->getNumContacts()<m_minimumPointsPerturbationThreshold)
+	if (convexShape->isPolyhedral() && resultOut->getPersistentManifold()->getNumContacts() < m_minimumPointsPerturbationThreshold)
 	{
-		btVector3 v0,v1;
-		btPlaneSpace1(planeNormal,v0,v1);
+		btVector3 v0, v1;
+		btPlaneSpace1(planeNormal, v0, v1);
 		//now perform 'm_numPerturbationIterations' collision queries with the perturbated collision objects
 
 		const btScalar angleLimit = 0.125f * SIMD_PI;
 		btScalar perturbeAngle;
 		btScalar radius = convexShape->getAngularMotionDisc();
 		perturbeAngle = gContactBreakingThreshold / radius;
-		if ( perturbeAngle > angleLimit ) 
-				perturbeAngle = angleLimit;
+		if (perturbeAngle > angleLimit)
+			perturbeAngle = angleLimit;
 
-		btQuaternion perturbeRot(v0,perturbeAngle);
-		for (int i=0;i<m_numPerturbationIterations;i++)
+		btQuaternion perturbeRot(v0, perturbeAngle);
+		for (int i = 0; i < m_numPerturbationIterations; i++)
 		{
-			btScalar iterationAngle = i*(SIMD_2_PI/btScalar(m_numPerturbationIterations));
-			btQuaternion rotq(planeNormal,iterationAngle);
-			collideSingleContact(rotq.inverse()*perturbeRot*rotq,body0Wrap,body1Wrap,dispatchInfo,resultOut);
+			btScalar iterationAngle = i * (SIMD_2_PI / btScalar(m_numPerturbationIterations));
+			btQuaternion rotq(planeNormal, iterationAngle);
+			collideSingleContact(rotq.inverse() * perturbeRot * rotq, body0Wrap, body1Wrap, dispatchInfo, resultOut);
 		}
 	}
 
@@ -162,7 +160,7 @@ void btConvexPlaneCollisionAlgorithm::processCollision (const btCollisionObjectW
 	}
 }
 
-btScalar btConvexPlaneCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* col0,btCollisionObject* col1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+btScalar btConvexPlaneCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* col0, btCollisionObject* col1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	(void)resultOut;
 	(void)dispatchInfo;
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexPlaneCollisionAlgorithm.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexPlaneCollisionAlgorithm.h
index d28c430c4c1..b693da118fe 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexPlaneCollisionAlgorithm.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btConvexPlaneCollisionAlgorithm.h
@@ -28,25 +28,24 @@ class btPersistentManifold;
 /// Other features are frame-coherency (persistent data) and collision response.
 class btConvexPlaneCollisionAlgorithm : public btCollisionAlgorithm
 {
-	bool		m_ownManifold;
-	btPersistentManifold*	m_manifoldPtr;
-	bool		m_isSwapped;
-	int			m_numPerturbationIterations;
-	int			m_minimumPointsPerturbationThreshold;
+	bool m_ownManifold;
+	btPersistentManifold* m_manifoldPtr;
+	bool m_isSwapped;
+	int m_numPerturbationIterations;
+	int m_minimumPointsPerturbationThreshold;
 
 public:
-
-	btConvexPlaneCollisionAlgorithm(btPersistentManifold* mf,const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap, bool isSwapped, int numPerturbationIterations,int minimumPointsPerturbationThreshold);
+	btConvexPlaneCollisionAlgorithm(btPersistentManifold* mf, const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, bool isSwapped, int numPerturbationIterations, int minimumPointsPerturbationThreshold);
 
 	virtual ~btConvexPlaneCollisionAlgorithm();
 
-	virtual void processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual void processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	void collideSingleContact (const btQuaternion& perturbeRot, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	void collideSingleContact(const btQuaternion& perturbeRot, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual	void	getAllContactManifolds(btManifoldArray&	manifoldArray)
+	virtual void getAllContactManifolds(btManifoldArray& manifoldArray)
 	{
 		if (m_manifoldPtr && m_ownManifold)
 		{
@@ -54,31 +53,30 @@ public:
 		}
 	}
 
-	struct CreateFunc :public 	btCollisionAlgorithmCreateFunc
+	struct CreateFunc : public btCollisionAlgorithmCreateFunc
 	{
-		int	m_numPerturbationIterations;
+		int m_numPerturbationIterations;
 		int m_minimumPointsPerturbationThreshold;
-			
-		CreateFunc() 
+
+		CreateFunc()
 			: m_numPerturbationIterations(1),
-			m_minimumPointsPerturbationThreshold(0)
+			  m_minimumPointsPerturbationThreshold(0)
 		{
 		}
-		
-		virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+
+		virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 		{
 			void* mem = ci.m_dispatcher1->allocateCollisionAlgorithm(sizeof(btConvexPlaneCollisionAlgorithm));
 			if (!m_swapped)
 			{
-				return new(mem) btConvexPlaneCollisionAlgorithm(0,ci,body0Wrap,body1Wrap,false,m_numPerturbationIterations,m_minimumPointsPerturbationThreshold);
-			} else
+				return new (mem) btConvexPlaneCollisionAlgorithm(0, ci, body0Wrap, body1Wrap, false, m_numPerturbationIterations, m_minimumPointsPerturbationThreshold);
+			}
+			else
 			{
-				return new(mem) btConvexPlaneCollisionAlgorithm(0,ci,body0Wrap,body1Wrap,true,m_numPerturbationIterations,m_minimumPointsPerturbationThreshold);
+				return new (mem) btConvexPlaneCollisionAlgorithm(0, ci, body0Wrap, body1Wrap, true, m_numPerturbationIterations, m_minimumPointsPerturbationThreshold);
 			}
 		}
 	};
-
 };
 
-#endif //BT_CONVEX_PLANE_COLLISION_ALGORITHM_H
-
+#endif  //BT_CONVEX_PLANE_COLLISION_ALGORITHM_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btDefaultCollisionConfiguration.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btDefaultCollisionConfiguration.cpp
index f6e4e57b0a3..ef3ea9e394c 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btDefaultCollisionConfiguration.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btDefaultCollisionConfiguration.cpp
@@ -26,114 +26,108 @@ subject to the following restrictions:
 #include "BulletCollision/CollisionDispatch/btSphereSphereCollisionAlgorithm.h"
 #ifdef USE_BUGGY_SPHERE_BOX_ALGORITHM
 #include "BulletCollision/CollisionDispatch/btSphereBoxCollisionAlgorithm.h"
-#endif //USE_BUGGY_SPHERE_BOX_ALGORITHM
+#endif  //USE_BUGGY_SPHERE_BOX_ALGORITHM
 #include "BulletCollision/CollisionDispatch/btSphereTriangleCollisionAlgorithm.h"
 #include "BulletCollision/NarrowPhaseCollision/btGjkEpaPenetrationDepthSolver.h"
 #include "BulletCollision/NarrowPhaseCollision/btMinkowskiPenetrationDepthSolver.h"
 #include "BulletCollision/NarrowPhaseCollision/btVoronoiSimplexSolver.h"
 
-
-
 #include "LinearMath/btPoolAllocator.h"
 
-
-
-
-
 btDefaultCollisionConfiguration::btDefaultCollisionConfiguration(const btDefaultCollisionConstructionInfo& constructionInfo)
 //btDefaultCollisionConfiguration::btDefaultCollisionConfiguration(btStackAlloc*	stackAlloc,btPoolAllocator*	persistentManifoldPool,btPoolAllocator*	collisionAlgorithmPool)
 {
-
-    void* mem = NULL;
+	void* mem = NULL;
 	if (constructionInfo.m_useEpaPenetrationAlgorithm)
 	{
-		mem = btAlignedAlloc(sizeof(btGjkEpaPenetrationDepthSolver),16);
-		m_pdSolver = new (mem)btGjkEpaPenetrationDepthSolver;
-	}else
-	{
-		mem = btAlignedAlloc(sizeof(btMinkowskiPenetrationDepthSolver),16);
-		m_pdSolver = new (mem)btMinkowskiPenetrationDepthSolver;
+		mem = btAlignedAlloc(sizeof(btGjkEpaPenetrationDepthSolver), 16);
+		m_pdSolver = new (mem) btGjkEpaPenetrationDepthSolver;
 	}
-	
+	else
+	{
+		mem = btAlignedAlloc(sizeof(btMinkowskiPenetrationDepthSolver), 16);
+		m_pdSolver = new (mem) btMinkowskiPenetrationDepthSolver;
+	}
+
 	//default CreationFunctions, filling the m_doubleDispatch table
-	mem = btAlignedAlloc(sizeof(btConvexConvexAlgorithm::CreateFunc),16);
-	m_convexConvexCreateFunc = new(mem) btConvexConvexAlgorithm::CreateFunc(m_pdSolver);
-	mem = btAlignedAlloc(sizeof(btConvexConcaveCollisionAlgorithm::CreateFunc),16);
-	m_convexConcaveCreateFunc = new (mem)btConvexConcaveCollisionAlgorithm::CreateFunc;
-	mem = btAlignedAlloc(sizeof(btConvexConcaveCollisionAlgorithm::CreateFunc),16);
-	m_swappedConvexConcaveCreateFunc = new (mem)btConvexConcaveCollisionAlgorithm::SwappedCreateFunc;
-	mem = btAlignedAlloc(sizeof(btCompoundCollisionAlgorithm::CreateFunc),16);
-	m_compoundCreateFunc = new (mem)btCompoundCollisionAlgorithm::CreateFunc;
+	mem = btAlignedAlloc(sizeof(btConvexConvexAlgorithm::CreateFunc), 16);
+	m_convexConvexCreateFunc = new (mem) btConvexConvexAlgorithm::CreateFunc(m_pdSolver);
+	mem = btAlignedAlloc(sizeof(btConvexConcaveCollisionAlgorithm::CreateFunc), 16);
+	m_convexConcaveCreateFunc = new (mem) btConvexConcaveCollisionAlgorithm::CreateFunc;
+	mem = btAlignedAlloc(sizeof(btConvexConcaveCollisionAlgorithm::CreateFunc), 16);
+	m_swappedConvexConcaveCreateFunc = new (mem) btConvexConcaveCollisionAlgorithm::SwappedCreateFunc;
+	mem = btAlignedAlloc(sizeof(btCompoundCollisionAlgorithm::CreateFunc), 16);
+	m_compoundCreateFunc = new (mem) btCompoundCollisionAlgorithm::CreateFunc;
 
-	mem = btAlignedAlloc(sizeof(btCompoundCompoundCollisionAlgorithm::CreateFunc),16);
-	m_compoundCompoundCreateFunc = new (mem)btCompoundCompoundCollisionAlgorithm::CreateFunc;
+	mem = btAlignedAlloc(sizeof(btCompoundCompoundCollisionAlgorithm::CreateFunc), 16);
+	m_compoundCompoundCreateFunc = new (mem) btCompoundCompoundCollisionAlgorithm::CreateFunc;
 
-	mem = btAlignedAlloc(sizeof(btCompoundCollisionAlgorithm::SwappedCreateFunc),16);
-	m_swappedCompoundCreateFunc = new (mem)btCompoundCollisionAlgorithm::SwappedCreateFunc;
-	mem = btAlignedAlloc(sizeof(btEmptyAlgorithm::CreateFunc),16);
-	m_emptyCreateFunc = new(mem) btEmptyAlgorithm::CreateFunc;
-	
-	mem = btAlignedAlloc(sizeof(btSphereSphereCollisionAlgorithm::CreateFunc),16);
-	m_sphereSphereCF = new(mem) btSphereSphereCollisionAlgorithm::CreateFunc;
+	mem = btAlignedAlloc(sizeof(btCompoundCollisionAlgorithm::SwappedCreateFunc), 16);
+	m_swappedCompoundCreateFunc = new (mem) btCompoundCollisionAlgorithm::SwappedCreateFunc;
+	mem = btAlignedAlloc(sizeof(btEmptyAlgorithm::CreateFunc), 16);
+	m_emptyCreateFunc = new (mem) btEmptyAlgorithm::CreateFunc;
+
+	mem = btAlignedAlloc(sizeof(btSphereSphereCollisionAlgorithm::CreateFunc), 16);
+	m_sphereSphereCF = new (mem) btSphereSphereCollisionAlgorithm::CreateFunc;
 #ifdef USE_BUGGY_SPHERE_BOX_ALGORITHM
-	mem = btAlignedAlloc(sizeof(btSphereBoxCollisionAlgorithm::CreateFunc),16);
-	m_sphereBoxCF = new(mem) btSphereBoxCollisionAlgorithm::CreateFunc;
-	mem = btAlignedAlloc(sizeof(btSphereBoxCollisionAlgorithm::CreateFunc),16);
-	m_boxSphereCF = new (mem)btSphereBoxCollisionAlgorithm::CreateFunc;
+	mem = btAlignedAlloc(sizeof(btSphereBoxCollisionAlgorithm::CreateFunc), 16);
+	m_sphereBoxCF = new (mem) btSphereBoxCollisionAlgorithm::CreateFunc;
+	mem = btAlignedAlloc(sizeof(btSphereBoxCollisionAlgorithm::CreateFunc), 16);
+	m_boxSphereCF = new (mem) btSphereBoxCollisionAlgorithm::CreateFunc;
 	m_boxSphereCF->m_swapped = true;
-#endif //USE_BUGGY_SPHERE_BOX_ALGORITHM
+#endif  //USE_BUGGY_SPHERE_BOX_ALGORITHM
 
-	mem = btAlignedAlloc(sizeof(btSphereTriangleCollisionAlgorithm::CreateFunc),16);
-	m_sphereTriangleCF = new (mem)btSphereTriangleCollisionAlgorithm::CreateFunc;
-	mem = btAlignedAlloc(sizeof(btSphereTriangleCollisionAlgorithm::CreateFunc),16);
-	m_triangleSphereCF = new (mem)btSphereTriangleCollisionAlgorithm::CreateFunc;
+	mem = btAlignedAlloc(sizeof(btSphereTriangleCollisionAlgorithm::CreateFunc), 16);
+	m_sphereTriangleCF = new (mem) btSphereTriangleCollisionAlgorithm::CreateFunc;
+	mem = btAlignedAlloc(sizeof(btSphereTriangleCollisionAlgorithm::CreateFunc), 16);
+	m_triangleSphereCF = new (mem) btSphereTriangleCollisionAlgorithm::CreateFunc;
 	m_triangleSphereCF->m_swapped = true;
-	
-	mem = btAlignedAlloc(sizeof(btBoxBoxCollisionAlgorithm::CreateFunc),16);
-	m_boxBoxCF = new(mem)btBoxBoxCollisionAlgorithm::CreateFunc;
+
+	mem = btAlignedAlloc(sizeof(btBoxBoxCollisionAlgorithm::CreateFunc), 16);
+	m_boxBoxCF = new (mem) btBoxBoxCollisionAlgorithm::CreateFunc;
 
 	//convex versus plane
-	mem = btAlignedAlloc (sizeof(btConvexPlaneCollisionAlgorithm::CreateFunc),16);
+	mem = btAlignedAlloc(sizeof(btConvexPlaneCollisionAlgorithm::CreateFunc), 16);
 	m_convexPlaneCF = new (mem) btConvexPlaneCollisionAlgorithm::CreateFunc;
-	mem = btAlignedAlloc (sizeof(btConvexPlaneCollisionAlgorithm::CreateFunc),16);
+	mem = btAlignedAlloc(sizeof(btConvexPlaneCollisionAlgorithm::CreateFunc), 16);
 	m_planeConvexCF = new (mem) btConvexPlaneCollisionAlgorithm::CreateFunc;
 	m_planeConvexCF->m_swapped = true;
-	
+
 	///calculate maximum element size, big enough to fit any collision algorithm in the memory pool
 	int maxSize = sizeof(btConvexConvexAlgorithm);
 	int maxSize2 = sizeof(btConvexConcaveCollisionAlgorithm);
 	int maxSize3 = sizeof(btCompoundCollisionAlgorithm);
 	int maxSize4 = sizeof(btCompoundCompoundCollisionAlgorithm);
 
-	int	collisionAlgorithmMaxElementSize = btMax(maxSize,constructionInfo.m_customCollisionAlgorithmMaxElementSize);
-	collisionAlgorithmMaxElementSize = btMax(collisionAlgorithmMaxElementSize,maxSize2);
-	collisionAlgorithmMaxElementSize = btMax(collisionAlgorithmMaxElementSize,maxSize3);
-	collisionAlgorithmMaxElementSize = btMax(collisionAlgorithmMaxElementSize,maxSize4);
-		
+	int collisionAlgorithmMaxElementSize = btMax(maxSize, constructionInfo.m_customCollisionAlgorithmMaxElementSize);
+	collisionAlgorithmMaxElementSize = btMax(collisionAlgorithmMaxElementSize, maxSize2);
+	collisionAlgorithmMaxElementSize = btMax(collisionAlgorithmMaxElementSize, maxSize3);
+	collisionAlgorithmMaxElementSize = btMax(collisionAlgorithmMaxElementSize, maxSize4);
+
 	if (constructionInfo.m_persistentManifoldPool)
 	{
 		m_ownsPersistentManifoldPool = false;
 		m_persistentManifoldPool = constructionInfo.m_persistentManifoldPool;
-	} else
+	}
+	else
 	{
 		m_ownsPersistentManifoldPool = true;
-		void* mem = btAlignedAlloc(sizeof(btPoolAllocator),16);
-		m_persistentManifoldPool = new (mem) btPoolAllocator(sizeof(btPersistentManifold),constructionInfo.m_defaultMaxPersistentManifoldPoolSize);
+		void* mem = btAlignedAlloc(sizeof(btPoolAllocator), 16);
+		m_persistentManifoldPool = new (mem) btPoolAllocator(sizeof(btPersistentManifold), constructionInfo.m_defaultMaxPersistentManifoldPoolSize);
 	}
-	
-	collisionAlgorithmMaxElementSize = (collisionAlgorithmMaxElementSize+16)&0xffffffffffff0;
+
+	collisionAlgorithmMaxElementSize = (collisionAlgorithmMaxElementSize + 16) & 0xffffffffffff0;
 	if (constructionInfo.m_collisionAlgorithmPool)
 	{
 		m_ownsCollisionAlgorithmPool = false;
 		m_collisionAlgorithmPool = constructionInfo.m_collisionAlgorithmPool;
-	} else
+	}
+	else
 	{
 		m_ownsCollisionAlgorithmPool = true;
-		void* mem = btAlignedAlloc(sizeof(btPoolAllocator),16);
-		m_collisionAlgorithmPool = new(mem) btPoolAllocator(collisionAlgorithmMaxElementSize,constructionInfo.m_defaultMaxCollisionAlgorithmPoolSize);
+		void* mem = btAlignedAlloc(sizeof(btPoolAllocator), 16);
+		m_collisionAlgorithmPool = new (mem) btPoolAllocator(collisionAlgorithmMaxElementSize, constructionInfo.m_defaultMaxCollisionAlgorithmPoolSize);
 	}
-
-
 }
 
 btDefaultCollisionConfiguration::~btDefaultCollisionConfiguration()
@@ -150,83 +144,78 @@ btDefaultCollisionConfiguration::~btDefaultCollisionConfiguration()
 	}
 
 	m_convexConvexCreateFunc->~btCollisionAlgorithmCreateFunc();
-	btAlignedFree(	m_convexConvexCreateFunc);
+	btAlignedFree(m_convexConvexCreateFunc);
 
 	m_convexConcaveCreateFunc->~btCollisionAlgorithmCreateFunc();
-	btAlignedFree( m_convexConcaveCreateFunc);
+	btAlignedFree(m_convexConcaveCreateFunc);
 	m_swappedConvexConcaveCreateFunc->~btCollisionAlgorithmCreateFunc();
-	btAlignedFree( m_swappedConvexConcaveCreateFunc);
+	btAlignedFree(m_swappedConvexConcaveCreateFunc);
 
 	m_compoundCreateFunc->~btCollisionAlgorithmCreateFunc();
-	btAlignedFree( m_compoundCreateFunc);
+	btAlignedFree(m_compoundCreateFunc);
 
 	m_compoundCompoundCreateFunc->~btCollisionAlgorithmCreateFunc();
 	btAlignedFree(m_compoundCompoundCreateFunc);
 
 	m_swappedCompoundCreateFunc->~btCollisionAlgorithmCreateFunc();
-	btAlignedFree( m_swappedCompoundCreateFunc);
+	btAlignedFree(m_swappedCompoundCreateFunc);
 
 	m_emptyCreateFunc->~btCollisionAlgorithmCreateFunc();
-	btAlignedFree( m_emptyCreateFunc);
+	btAlignedFree(m_emptyCreateFunc);
 
 	m_sphereSphereCF->~btCollisionAlgorithmCreateFunc();
-	btAlignedFree( m_sphereSphereCF);
+	btAlignedFree(m_sphereSphereCF);
 
 #ifdef USE_BUGGY_SPHERE_BOX_ALGORITHM
 	m_sphereBoxCF->~btCollisionAlgorithmCreateFunc();
-	btAlignedFree( m_sphereBoxCF);
+	btAlignedFree(m_sphereBoxCF);
 	m_boxSphereCF->~btCollisionAlgorithmCreateFunc();
-	btAlignedFree( m_boxSphereCF);
-#endif //USE_BUGGY_SPHERE_BOX_ALGORITHM
+	btAlignedFree(m_boxSphereCF);
+#endif  //USE_BUGGY_SPHERE_BOX_ALGORITHM
 
 	m_sphereTriangleCF->~btCollisionAlgorithmCreateFunc();
-	btAlignedFree( m_sphereTriangleCF);
+	btAlignedFree(m_sphereTriangleCF);
 	m_triangleSphereCF->~btCollisionAlgorithmCreateFunc();
-	btAlignedFree( m_triangleSphereCF);
+	btAlignedFree(m_triangleSphereCF);
 	m_boxBoxCF->~btCollisionAlgorithmCreateFunc();
-	btAlignedFree( m_boxBoxCF);
+	btAlignedFree(m_boxBoxCF);
 
 	m_convexPlaneCF->~btCollisionAlgorithmCreateFunc();
-	btAlignedFree( m_convexPlaneCF);
+	btAlignedFree(m_convexPlaneCF);
 	m_planeConvexCF->~btCollisionAlgorithmCreateFunc();
-	btAlignedFree( m_planeConvexCF);
+	btAlignedFree(m_planeConvexCF);
 
 	m_pdSolver->~btConvexPenetrationDepthSolver();
-	
+
 	btAlignedFree(m_pdSolver);
-
-
 }
 
 btCollisionAlgorithmCreateFunc* btDefaultCollisionConfiguration::getClosestPointsAlgorithmCreateFunc(int proxyType0, int proxyType1)
 {
-
-
 	if ((proxyType0 == SPHERE_SHAPE_PROXYTYPE) && (proxyType1 == SPHERE_SHAPE_PROXYTYPE))
 	{
-		return	m_sphereSphereCF;
+		return m_sphereSphereCF;
 	}
 #ifdef USE_BUGGY_SPHERE_BOX_ALGORITHM
 	if ((proxyType0 == SPHERE_SHAPE_PROXYTYPE) && (proxyType1 == BOX_SHAPE_PROXYTYPE))
 	{
-		return	m_sphereBoxCF;
+		return m_sphereBoxCF;
 	}
 
 	if ((proxyType0 == BOX_SHAPE_PROXYTYPE) && (proxyType1 == SPHERE_SHAPE_PROXYTYPE))
 	{
-		return	m_boxSphereCF;
+		return m_boxSphereCF;
 	}
-#endif //USE_BUGGY_SPHERE_BOX_ALGORITHM
-
+#endif  //USE_BUGGY_SPHERE_BOX_ALGORITHM
 
 	if ((proxyType0 == SPHERE_SHAPE_PROXYTYPE) && (proxyType1 == TRIANGLE_SHAPE_PROXYTYPE))
 	{
-		return	m_sphereTriangleCF;
+		return m_sphereTriangleCF;
 	}
 
 	if ((proxyType0 == TRIANGLE_SHAPE_PROXYTYPE) && (proxyType1 == SPHERE_SHAPE_PROXYTYPE))
 	{
-		return	m_triangleSphereCF;
+		return m_triangleSphereCF;
 	}
 
 	if (btBroadphaseProxy::isConvex(proxyType0) && (proxyType1 == STATIC_PLANE_PROXYTYPE))
@@ -239,8 +228,6 @@ btCollisionAlgorithmCreateFunc* btDefaultCollisionConfiguration::getClosestPoint
 		return m_planeConvexCF;
 	}
 
-
-
 	if (btBroadphaseProxy::isConvex(proxyType0) && btBroadphaseProxy::isConvex(proxyType1))
 	{
 		return m_convexConvexCreateFunc;
@@ -256,7 +243,6 @@ btCollisionAlgorithmCreateFunc* btDefaultCollisionConfiguration::getClosestPoint
 		return m_swappedConvexConcaveCreateFunc;
 	}
 
-
 	if (btBroadphaseProxy::isCompound(proxyType0) && btBroadphaseProxy::isCompound(proxyType1))
 	{
 		return m_compoundCompoundCreateFunc;
@@ -276,46 +262,41 @@ btCollisionAlgorithmCreateFunc* btDefaultCollisionConfiguration::getClosestPoint
 
 	//failed to find an algorithm
 	return m_emptyCreateFunc;
-
 }
 
-btCollisionAlgorithmCreateFunc* btDefaultCollisionConfiguration::getCollisionAlgorithmCreateFunc(int proxyType0,int proxyType1)
+btCollisionAlgorithmCreateFunc* btDefaultCollisionConfiguration::getCollisionAlgorithmCreateFunc(int proxyType0, int proxyType1)
 {
-
-
-
-	if ((proxyType0 == SPHERE_SHAPE_PROXYTYPE) && (proxyType1==SPHERE_SHAPE_PROXYTYPE))
+	if ((proxyType0 == SPHERE_SHAPE_PROXYTYPE) && (proxyType1 == SPHERE_SHAPE_PROXYTYPE))
 	{
-		return	m_sphereSphereCF;
+		return m_sphereSphereCF;
 	}
 #ifdef USE_BUGGY_SPHERE_BOX_ALGORITHM
-	if ((proxyType0 == SPHERE_SHAPE_PROXYTYPE) && (proxyType1==BOX_SHAPE_PROXYTYPE))
+	if ((proxyType0 == SPHERE_SHAPE_PROXYTYPE) && (proxyType1 == BOX_SHAPE_PROXYTYPE))
 	{
-		return	m_sphereBoxCF;
+		return m_sphereBoxCF;
 	}
 
-	if ((proxyType0 == BOX_SHAPE_PROXYTYPE ) && (proxyType1==SPHERE_SHAPE_PROXYTYPE))
+	if ((proxyType0 == BOX_SHAPE_PROXYTYPE) && (proxyType1 == SPHERE_SHAPE_PROXYTYPE))
 	{
-		return	m_boxSphereCF;
+		return m_boxSphereCF;
 	}
-#endif //USE_BUGGY_SPHERE_BOX_ALGORITHM
+#endif  //USE_BUGGY_SPHERE_BOX_ALGORITHM
 
-
-	if ((proxyType0 == SPHERE_SHAPE_PROXYTYPE ) && (proxyType1==TRIANGLE_SHAPE_PROXYTYPE))
+	if ((proxyType0 == SPHERE_SHAPE_PROXYTYPE) && (proxyType1 == TRIANGLE_SHAPE_PROXYTYPE))
 	{
-		return	m_sphereTriangleCF;
+		return m_sphereTriangleCF;
 	}
 
-	if ((proxyType0 == TRIANGLE_SHAPE_PROXYTYPE  ) && (proxyType1==SPHERE_SHAPE_PROXYTYPE))
+	if ((proxyType0 == TRIANGLE_SHAPE_PROXYTYPE) && (proxyType1 == SPHERE_SHAPE_PROXYTYPE))
 	{
-		return	m_triangleSphereCF;
-	} 
+		return m_triangleSphereCF;
+	}
 
 	if ((proxyType0 == BOX_SHAPE_PROXYTYPE) && (proxyType1 == BOX_SHAPE_PROXYTYPE))
 	{
 		return m_boxBoxCF;
 	}
-	
+
 	if (btBroadphaseProxy::isConvex(proxyType0) && (proxyType1 == STATIC_PLANE_PROXYTYPE))
 	{
 		return m_convexPlaneCF;
@@ -325,8 +306,6 @@ btCollisionAlgorithmCreateFunc* btDefaultCollisionConfiguration::getCollisionAlg
 	{
 		return m_planeConvexCF;
 	}
-	
-
 
 	if (btBroadphaseProxy::isConvex(proxyType0) && btBroadphaseProxy::isConvex(proxyType1))
 	{
@@ -343,7 +322,6 @@ btCollisionAlgorithmCreateFunc* btDefaultCollisionConfiguration::getCollisionAlg
 		return m_swappedConvexConcaveCreateFunc;
 	}
 
-
 	if (btBroadphaseProxy::isCompound(proxyType0) && btBroadphaseProxy::isCompound(proxyType1))
 	{
 		return m_compoundCompoundCreateFunc;
@@ -352,7 +330,8 @@ btCollisionAlgorithmCreateFunc* btDefaultCollisionConfiguration::getCollisionAlg
 	if (btBroadphaseProxy::isCompound(proxyType0))
 	{
 		return m_compoundCreateFunc;
-	} else
+	}
+	else
 	{
 		if (btBroadphaseProxy::isCompound(proxyType1))
 		{
@@ -366,17 +345,17 @@ btCollisionAlgorithmCreateFunc* btDefaultCollisionConfiguration::getCollisionAlg
 
 void btDefaultCollisionConfiguration::setConvexConvexMultipointIterations(int numPerturbationIterations, int minimumPointsPerturbationThreshold)
 {
-	btConvexConvexAlgorithm::CreateFunc* convexConvex = (btConvexConvexAlgorithm::CreateFunc*) m_convexConvexCreateFunc;
+	btConvexConvexAlgorithm::CreateFunc* convexConvex = (btConvexConvexAlgorithm::CreateFunc*)m_convexConvexCreateFunc;
 	convexConvex->m_numPerturbationIterations = numPerturbationIterations;
 	convexConvex->m_minimumPointsPerturbationThreshold = minimumPointsPerturbationThreshold;
 }
 
-void	btDefaultCollisionConfiguration::setPlaneConvexMultipointIterations(int numPerturbationIterations, int minimumPointsPerturbationThreshold)
+void btDefaultCollisionConfiguration::setPlaneConvexMultipointIterations(int numPerturbationIterations, int minimumPointsPerturbationThreshold)
 {
 	btConvexPlaneCollisionAlgorithm::CreateFunc* cpCF = (btConvexPlaneCollisionAlgorithm::CreateFunc*)m_convexPlaneCF;
 	cpCF->m_numPerturbationIterations = numPerturbationIterations;
 	cpCF->m_minimumPointsPerturbationThreshold = minimumPointsPerturbationThreshold;
-	
+
 	btConvexPlaneCollisionAlgorithm::CreateFunc* pcCF = (btConvexPlaneCollisionAlgorithm::CreateFunc*)m_planeConvexCF;
 	pcCF->m_numPerturbationIterations = numPerturbationIterations;
 	pcCF->m_minimumPointsPerturbationThreshold = minimumPointsPerturbationThreshold;
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btDefaultCollisionConfiguration.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btDefaultCollisionConfiguration.h
index 17c7596cffb..b39a3f41de1 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btDefaultCollisionConfiguration.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btDefaultCollisionConfiguration.h
@@ -20,76 +20,68 @@ subject to the following restrictions:
 class btVoronoiSimplexSolver;
 class btConvexPenetrationDepthSolver;
 
-struct	btDefaultCollisionConstructionInfo
+struct btDefaultCollisionConstructionInfo
 {
-	btPoolAllocator*	m_persistentManifoldPool;
-	btPoolAllocator*	m_collisionAlgorithmPool;
-	int					m_defaultMaxPersistentManifoldPoolSize;
-	int					m_defaultMaxCollisionAlgorithmPoolSize;
-	int					m_customCollisionAlgorithmMaxElementSize;
-	int					m_useEpaPenetrationAlgorithm;
+	btPoolAllocator* m_persistentManifoldPool;
+	btPoolAllocator* m_collisionAlgorithmPool;
+	int m_defaultMaxPersistentManifoldPoolSize;
+	int m_defaultMaxCollisionAlgorithmPoolSize;
+	int m_customCollisionAlgorithmMaxElementSize;
+	int m_useEpaPenetrationAlgorithm;
 
 	btDefaultCollisionConstructionInfo()
-		:m_persistentManifoldPool(0),
-		m_collisionAlgorithmPool(0),
-		m_defaultMaxPersistentManifoldPoolSize(4096),
-		m_defaultMaxCollisionAlgorithmPoolSize(4096),
-		m_customCollisionAlgorithmMaxElementSize(0),
-		m_useEpaPenetrationAlgorithm(true)
+		: m_persistentManifoldPool(0),
+		  m_collisionAlgorithmPool(0),
+		  m_defaultMaxPersistentManifoldPoolSize(4096),
+		  m_defaultMaxCollisionAlgorithmPoolSize(4096),
+		  m_customCollisionAlgorithmMaxElementSize(0),
+		  m_useEpaPenetrationAlgorithm(true)
 	{
 	}
 };
 
-
-
 ///btCollisionConfiguration allows to configure Bullet collision detection
 ///stack allocator, pool memory allocators
 ///@todo: describe the meaning
-class	btDefaultCollisionConfiguration : public btCollisionConfiguration
+class btDefaultCollisionConfiguration : public btCollisionConfiguration
 {
-
 protected:
+	int m_persistentManifoldPoolSize;
 
-	int	m_persistentManifoldPoolSize;
-	
+	btPoolAllocator* m_persistentManifoldPool;
+	bool m_ownsPersistentManifoldPool;
 
-	btPoolAllocator*	m_persistentManifoldPool;
-	bool	m_ownsPersistentManifoldPool;
-
-
-	btPoolAllocator*	m_collisionAlgorithmPool;
-	bool	m_ownsCollisionAlgorithmPool;
+	btPoolAllocator* m_collisionAlgorithmPool;
+	bool m_ownsCollisionAlgorithmPool;
 
 	//default penetration depth solver
-	btConvexPenetrationDepthSolver*	m_pdSolver;
-	
+	btConvexPenetrationDepthSolver* m_pdSolver;
+
 	//default CreationFunctions, filling the m_doubleDispatch table
-	btCollisionAlgorithmCreateFunc*	m_convexConvexCreateFunc;
-	btCollisionAlgorithmCreateFunc*	m_convexConcaveCreateFunc;
-	btCollisionAlgorithmCreateFunc*	m_swappedConvexConcaveCreateFunc;
-	btCollisionAlgorithmCreateFunc*	m_compoundCreateFunc;
-	btCollisionAlgorithmCreateFunc*	m_compoundCompoundCreateFunc;
-	
-	btCollisionAlgorithmCreateFunc*	m_swappedCompoundCreateFunc;
+	btCollisionAlgorithmCreateFunc* m_convexConvexCreateFunc;
+	btCollisionAlgorithmCreateFunc* m_convexConcaveCreateFunc;
+	btCollisionAlgorithmCreateFunc* m_swappedConvexConcaveCreateFunc;
+	btCollisionAlgorithmCreateFunc* m_compoundCreateFunc;
+	btCollisionAlgorithmCreateFunc* m_compoundCompoundCreateFunc;
+
+	btCollisionAlgorithmCreateFunc* m_swappedCompoundCreateFunc;
 	btCollisionAlgorithmCreateFunc* m_emptyCreateFunc;
 	btCollisionAlgorithmCreateFunc* m_sphereSphereCF;
 	btCollisionAlgorithmCreateFunc* m_sphereBoxCF;
 	btCollisionAlgorithmCreateFunc* m_boxSphereCF;
 
 	btCollisionAlgorithmCreateFunc* m_boxBoxCF;
-	btCollisionAlgorithmCreateFunc*	m_sphereTriangleCF;
-	btCollisionAlgorithmCreateFunc*	m_triangleSphereCF;
-	btCollisionAlgorithmCreateFunc*	m_planeConvexCF;
-	btCollisionAlgorithmCreateFunc*	m_convexPlaneCF;
-	
+	btCollisionAlgorithmCreateFunc* m_sphereTriangleCF;
+	btCollisionAlgorithmCreateFunc* m_triangleSphereCF;
+	btCollisionAlgorithmCreateFunc* m_planeConvexCF;
+	btCollisionAlgorithmCreateFunc* m_convexPlaneCF;
+
 public:
-
-
 	btDefaultCollisionConfiguration(const btDefaultCollisionConstructionInfo& constructionInfo = btDefaultCollisionConstructionInfo());
 
 	virtual ~btDefaultCollisionConfiguration();
 
-		///memory pools
+	///memory pools
 	virtual btPoolAllocator* getPersistentManifoldPool()
 	{
 		return m_persistentManifoldPool;
@@ -100,8 +92,7 @@ public:
 		return m_collisionAlgorithmPool;
 	}
 
-
-	virtual btCollisionAlgorithmCreateFunc* getCollisionAlgorithmCreateFunc(int proxyType0,int proxyType1);
+	virtual btCollisionAlgorithmCreateFunc* getCollisionAlgorithmCreateFunc(int proxyType0, int proxyType1);
 
 	virtual btCollisionAlgorithmCreateFunc* getClosestPointsAlgorithmCreateFunc(int proxyType0, int proxyType1);
 
@@ -112,11 +103,9 @@ public:
 	///3 is a good value for both params, if you want to enable the feature. This is because the default contact cache contains a maximum of 4 points, and one collision query at the unperturbed orientation is performed first.
 	///See Bullet/Demos/CollisionDemo for an example how this feature gathers multiple points.
 	///@todo we could add a per-object setting of those parameters, for level-of-detail collision detection.
-	void	setConvexConvexMultipointIterations(int numPerturbationIterations=3, int minimumPointsPerturbationThreshold = 3);
-
-	void	setPlaneConvexMultipointIterations(int numPerturbationIterations=3, int minimumPointsPerturbationThreshold = 3);
+	void setConvexConvexMultipointIterations(int numPerturbationIterations = 3, int minimumPointsPerturbationThreshold = 3);
 
+	void setPlaneConvexMultipointIterations(int numPerturbationIterations = 3, int minimumPointsPerturbationThreshold = 3);
 };
 
-#endif //BT_DEFAULT_COLLISION_CONFIGURATION
-
+#endif  //BT_DEFAULT_COLLISION_CONFIGURATION
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btEmptyCollisionAlgorithm.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btEmptyCollisionAlgorithm.cpp
index 5fa1c8be5e4..7cd41bdb33e 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btEmptyCollisionAlgorithm.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btEmptyCollisionAlgorithm.cpp
@@ -15,20 +15,16 @@ subject to the following restrictions:
 
 #include "btEmptyCollisionAlgorithm.h"
 
-
-
 btEmptyAlgorithm::btEmptyAlgorithm(const btCollisionAlgorithmConstructionInfo& ci)
 	: btCollisionAlgorithm(ci)
 {
 }
 
-void btEmptyAlgorithm::processCollision (const btCollisionObjectWrapper* ,const btCollisionObjectWrapper* ,const btDispatcherInfo& ,btManifoldResult* )
+void btEmptyAlgorithm::processCollision(const btCollisionObjectWrapper*, const btCollisionObjectWrapper*, const btDispatcherInfo&, btManifoldResult*)
 {
 }
 
-btScalar btEmptyAlgorithm::calculateTimeOfImpact(btCollisionObject* ,btCollisionObject* ,const btDispatcherInfo& ,btManifoldResult* )
+btScalar btEmptyAlgorithm::calculateTimeOfImpact(btCollisionObject*, btCollisionObject*, const btDispatcherInfo&, btManifoldResult*)
 {
 	return btScalar(1.);
 }
-
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btEmptyCollisionAlgorithm.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btEmptyCollisionAlgorithm.h
index cb0f152183c..65ef83e0946 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btEmptyCollisionAlgorithm.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btEmptyCollisionAlgorithm.h
@@ -25,30 +25,28 @@ subject to the following restrictions:
 ///The dispatcher can dispatch a persistent btEmptyAlgorithm to avoid a search every frame.
 class btEmptyAlgorithm : public btCollisionAlgorithm
 {
-
 public:
-	
 	btEmptyAlgorithm(const btCollisionAlgorithmConstructionInfo& ci);
 
-	virtual void processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual void processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual	void	getAllContactManifolds(btManifoldArray&	manifoldArray)
+	virtual void getAllContactManifolds(btManifoldArray& manifoldArray)
 	{
 	}
 
-	struct CreateFunc :public 	btCollisionAlgorithmCreateFunc
+	struct CreateFunc : public btCollisionAlgorithmCreateFunc
 	{
-        virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
-        {
+		virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
+		{
 			(void)body0Wrap;
 			(void)body1Wrap;
 			void* mem = ci.m_dispatcher1->allocateCollisionAlgorithm(sizeof(btEmptyAlgorithm));
-			return new(mem) btEmptyAlgorithm(ci);
+			return new (mem) btEmptyAlgorithm(ci);
 		}
 	};
 
 } ATTRIBUTE_ALIGNED(16);
 
-#endif //BT_EMPTY_ALGORITH
+#endif  //BT_EMPTY_ALGORITH
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btGhostObject.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btGhostObject.cpp
index 86141fa6899..00f16fd0a8a 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btGhostObject.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btGhostObject.cpp
@@ -29,60 +29,58 @@ btGhostObject::~btGhostObject()
 	btAssert(!m_overlappingObjects.size());
 }
 
-
-void btGhostObject::addOverlappingObjectInternal(btBroadphaseProxy* otherProxy,btBroadphaseProxy* thisProxy)
+void btGhostObject::addOverlappingObjectInternal(btBroadphaseProxy* otherProxy, btBroadphaseProxy* thisProxy)
 {
 	btCollisionObject* otherObject = (btCollisionObject*)otherProxy->m_clientObject;
 	btAssert(otherObject);
 	///if this linearSearch becomes too slow (too many overlapping objects) we should add a more appropriate data structure
 	int index = m_overlappingObjects.findLinearSearch(otherObject);
-	if (index==m_overlappingObjects.size())
+	if (index == m_overlappingObjects.size())
 	{
 		//not found
 		m_overlappingObjects.push_back(otherObject);
 	}
 }
 
-void btGhostObject::removeOverlappingObjectInternal(btBroadphaseProxy* otherProxy,btDispatcher* dispatcher,btBroadphaseProxy* thisProxy)
+void btGhostObject::removeOverlappingObjectInternal(btBroadphaseProxy* otherProxy, btDispatcher* dispatcher, btBroadphaseProxy* thisProxy)
 {
 	btCollisionObject* otherObject = (btCollisionObject*)otherProxy->m_clientObject;
 	btAssert(otherObject);
 	int index = m_overlappingObjects.findLinearSearch(otherObject);
-	if (index<m_overlappingObjects.size())
+	if (index < m_overlappingObjects.size())
 	{
-		m_overlappingObjects[index] = m_overlappingObjects[m_overlappingObjects.size()-1];
+		m_overlappingObjects[index] = m_overlappingObjects[m_overlappingObjects.size() - 1];
 		m_overlappingObjects.pop_back();
 	}
 }
 
-
 btPairCachingGhostObject::btPairCachingGhostObject()
 {
-	m_hashPairCache = new (btAlignedAlloc(sizeof(btHashedOverlappingPairCache),16)) btHashedOverlappingPairCache();
+	m_hashPairCache = new (btAlignedAlloc(sizeof(btHashedOverlappingPairCache), 16)) btHashedOverlappingPairCache();
 }
 
 btPairCachingGhostObject::~btPairCachingGhostObject()
 {
 	m_hashPairCache->~btHashedOverlappingPairCache();
-	btAlignedFree( m_hashPairCache );
+	btAlignedFree(m_hashPairCache);
 }
 
-void btPairCachingGhostObject::addOverlappingObjectInternal(btBroadphaseProxy* otherProxy,btBroadphaseProxy* thisProxy)
+void btPairCachingGhostObject::addOverlappingObjectInternal(btBroadphaseProxy* otherProxy, btBroadphaseProxy* thisProxy)
 {
-	btBroadphaseProxy*actualThisProxy = thisProxy ? thisProxy : getBroadphaseHandle();
+	btBroadphaseProxy* actualThisProxy = thisProxy ? thisProxy : getBroadphaseHandle();
 	btAssert(actualThisProxy);
 
 	btCollisionObject* otherObject = (btCollisionObject*)otherProxy->m_clientObject;
 	btAssert(otherObject);
 	int index = m_overlappingObjects.findLinearSearch(otherObject);
-	if (index==m_overlappingObjects.size())
+	if (index == m_overlappingObjects.size())
 	{
 		m_overlappingObjects.push_back(otherObject);
-		m_hashPairCache->addOverlappingPair(actualThisProxy,otherProxy);
+		m_hashPairCache->addOverlappingPair(actualThisProxy, otherProxy);
 	}
 }
 
-void btPairCachingGhostObject::removeOverlappingObjectInternal(btBroadphaseProxy* otherProxy,btDispatcher* dispatcher,btBroadphaseProxy* thisProxy1)
+void btPairCachingGhostObject::removeOverlappingObjectInternal(btBroadphaseProxy* otherProxy, btDispatcher* dispatcher, btBroadphaseProxy* thisProxy1)
 {
 	btCollisionObject* otherObject = (btCollisionObject*)otherProxy->m_clientObject;
 	btBroadphaseProxy* actualThisProxy = thisProxy1 ? thisProxy1 : getBroadphaseHandle();
@@ -90,82 +88,79 @@ void btPairCachingGhostObject::removeOverlappingObjectInternal(btBroadphaseProxy
 
 	btAssert(otherObject);
 	int index = m_overlappingObjects.findLinearSearch(otherObject);
-	if (index<m_overlappingObjects.size())
+	if (index < m_overlappingObjects.size())
 	{
-		m_overlappingObjects[index] = m_overlappingObjects[m_overlappingObjects.size()-1];
+		m_overlappingObjects[index] = m_overlappingObjects[m_overlappingObjects.size() - 1];
 		m_overlappingObjects.pop_back();
-		m_hashPairCache->removeOverlappingPair(actualThisProxy,otherProxy,dispatcher);
+		m_hashPairCache->removeOverlappingPair(actualThisProxy, otherProxy, dispatcher);
 	}
 }
 
-
-void	btGhostObject::convexSweepTest(const btConvexShape* castShape, const btTransform& convexFromWorld, const btTransform& convexToWorld, btCollisionWorld::ConvexResultCallback& resultCallback, btScalar allowedCcdPenetration) const
+void btGhostObject::convexSweepTest(const btConvexShape* castShape, const btTransform& convexFromWorld, const btTransform& convexToWorld, btCollisionWorld::ConvexResultCallback& resultCallback, btScalar allowedCcdPenetration) const
 {
-	btTransform	convexFromTrans,convexToTrans;
+	btTransform convexFromTrans, convexToTrans;
 	convexFromTrans = convexFromWorld;
 	convexToTrans = convexToWorld;
 	btVector3 castShapeAabbMin, castShapeAabbMax;
 	/* Compute AABB that encompasses angular movement */
 	{
 		btVector3 linVel, angVel;
-		btTransformUtil::calculateVelocity (convexFromTrans, convexToTrans, 1.0, linVel, angVel);
+		btTransformUtil::calculateVelocity(convexFromTrans, convexToTrans, 1.0, linVel, angVel);
 		btTransform R;
-		R.setIdentity ();
-		R.setRotation (convexFromTrans.getRotation());
-		castShape->calculateTemporalAabb (R, linVel, angVel, 1.0, castShapeAabbMin, castShapeAabbMax);
+		R.setIdentity();
+		R.setRotation(convexFromTrans.getRotation());
+		castShape->calculateTemporalAabb(R, linVel, angVel, 1.0, castShapeAabbMin, castShapeAabbMax);
 	}
 
 	/// go over all objects, and if the ray intersects their aabb + cast shape aabb,
 	// do a ray-shape query using convexCaster (CCD)
 	int i;
-	for (i=0;i<m_overlappingObjects.size();i++)
+	for (i = 0; i < m_overlappingObjects.size(); i++)
 	{
-		btCollisionObject*	collisionObject= m_overlappingObjects[i];
+		btCollisionObject* collisionObject = m_overlappingObjects[i];
 		//only perform raycast if filterMask matches
-		if(resultCallback.needsCollision(collisionObject->getBroadphaseHandle())) {
+		if (resultCallback.needsCollision(collisionObject->getBroadphaseHandle()))
+		{
 			//RigidcollisionObject* collisionObject = ctrl->GetRigidcollisionObject();
-			btVector3 collisionObjectAabbMin,collisionObjectAabbMax;
-			collisionObject->getCollisionShape()->getAabb(collisionObject->getWorldTransform(),collisionObjectAabbMin,collisionObjectAabbMax);
-			AabbExpand (collisionObjectAabbMin, collisionObjectAabbMax, castShapeAabbMin, castShapeAabbMax);
-			btScalar hitLambda = btScalar(1.); //could use resultCallback.m_closestHitFraction, but needs testing
+			btVector3 collisionObjectAabbMin, collisionObjectAabbMax;
+			collisionObject->getCollisionShape()->getAabb(collisionObject->getWorldTransform(), collisionObjectAabbMin, collisionObjectAabbMax);
+			AabbExpand(collisionObjectAabbMin, collisionObjectAabbMax, castShapeAabbMin, castShapeAabbMax);
+			btScalar hitLambda = btScalar(1.);  //could use resultCallback.m_closestHitFraction, but needs testing
 			btVector3 hitNormal;
-			if (btRayAabb(convexFromWorld.getOrigin(),convexToWorld.getOrigin(),collisionObjectAabbMin,collisionObjectAabbMax,hitLambda,hitNormal))
+			if (btRayAabb(convexFromWorld.getOrigin(), convexToWorld.getOrigin(), collisionObjectAabbMin, collisionObjectAabbMax, hitLambda, hitNormal))
 			{
-				btCollisionWorld::objectQuerySingle(castShape, convexFromTrans,convexToTrans,
-					collisionObject,
-						collisionObject->getCollisionShape(),
-						collisionObject->getWorldTransform(),
-						resultCallback,
-						allowedCcdPenetration);
+				btCollisionWorld::objectQuerySingle(castShape, convexFromTrans, convexToTrans,
+													collisionObject,
+													collisionObject->getCollisionShape(),
+													collisionObject->getWorldTransform(),
+													resultCallback,
+													allowedCcdPenetration);
 			}
 		}
 	}
-
 }
 
-void	btGhostObject::rayTest(const btVector3& rayFromWorld, const btVector3& rayToWorld, btCollisionWorld::RayResultCallback& resultCallback) const
+void btGhostObject::rayTest(const btVector3& rayFromWorld, const btVector3& rayToWorld, btCollisionWorld::RayResultCallback& resultCallback) const
 {
 	btTransform rayFromTrans;
 	rayFromTrans.setIdentity();
 	rayFromTrans.setOrigin(rayFromWorld);
-	btTransform  rayToTrans;
+	btTransform rayToTrans;
 	rayToTrans.setIdentity();
 	rayToTrans.setOrigin(rayToWorld);
 
-
 	int i;
-	for (i=0;i<m_overlappingObjects.size();i++)
+	for (i = 0; i < m_overlappingObjects.size(); i++)
 	{
-		btCollisionObject*	collisionObject= m_overlappingObjects[i];
+		btCollisionObject* collisionObject = m_overlappingObjects[i];
 		//only perform raycast if filterMask matches
-		if(resultCallback.needsCollision(collisionObject->getBroadphaseHandle())) 
+		if (resultCallback.needsCollision(collisionObject->getBroadphaseHandle()))
 		{
-			btCollisionWorld::rayTestSingle(rayFromTrans,rayToTrans,
-							collisionObject,
-								collisionObject->getCollisionShape(),
-								collisionObject->getWorldTransform(),
-								resultCallback);
+			btCollisionWorld::rayTestSingle(rayFromTrans, rayToTrans,
+											collisionObject,
+											collisionObject->getCollisionShape(),
+											collisionObject->getWorldTransform(),
+											resultCallback);
 		}
 	}
 }
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btGhostObject.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btGhostObject.h
index 8ec86138575..aa7f48d5cb4 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btGhostObject.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btGhostObject.h
@@ -16,7 +16,6 @@ subject to the following restrictions:
 #ifndef BT_GHOST_OBJECT_H
 #define BT_GHOST_OBJECT_H
 
-
 #include "btCollisionObject.h"
 #include "BulletCollision/BroadphaseCollision/btOverlappingPairCallback.h"
 #include "LinearMath/btAlignedAllocator.h"
@@ -31,48 +30,47 @@ class btDispatcher;
 ///By default, this overlap is based on the AABB
 ///This is useful for creating a character controller, collision sensors/triggers, explosions etc.
 ///We plan on adding rayTest and other queries for the btGhostObject
-ATTRIBUTE_ALIGNED16(class) btGhostObject : public btCollisionObject
+ATTRIBUTE_ALIGNED16(class)
+btGhostObject : public btCollisionObject
 {
 protected:
-
 	btAlignedObjectArray<btCollisionObject*> m_overlappingObjects;
 
 public:
-
 	btGhostObject();
 
 	virtual ~btGhostObject();
 
-	void	convexSweepTest(const class btConvexShape* castShape, const btTransform& convexFromWorld, const btTransform& convexToWorld, btCollisionWorld::ConvexResultCallback& resultCallback, btScalar allowedCcdPenetration = 0.f) const;
+	void convexSweepTest(const class btConvexShape* castShape, const btTransform& convexFromWorld, const btTransform& convexToWorld, btCollisionWorld::ConvexResultCallback& resultCallback, btScalar allowedCcdPenetration = 0.f) const;
 
-	void	rayTest(const btVector3& rayFromWorld, const btVector3& rayToWorld, btCollisionWorld::RayResultCallback& resultCallback) const; 
+	void rayTest(const btVector3& rayFromWorld, const btVector3& rayToWorld, btCollisionWorld::RayResultCallback& resultCallback) const;
 
 	///this method is mainly for expert/internal use only.
-	virtual void	addOverlappingObjectInternal(btBroadphaseProxy* otherProxy, btBroadphaseProxy* thisProxy=0);
+	virtual void addOverlappingObjectInternal(btBroadphaseProxy * otherProxy, btBroadphaseProxy* thisProxy = 0);
 	///this method is mainly for expert/internal use only.
-	virtual void	removeOverlappingObjectInternal(btBroadphaseProxy* otherProxy,btDispatcher* dispatcher,btBroadphaseProxy* thisProxy=0);
+	virtual void removeOverlappingObjectInternal(btBroadphaseProxy * otherProxy, btDispatcher * dispatcher, btBroadphaseProxy* thisProxy = 0);
 
-	int	getNumOverlappingObjects() const
+	int getNumOverlappingObjects() const
 	{
 		return m_overlappingObjects.size();
 	}
 
-	btCollisionObject*	getOverlappingObject(int index)
+	btCollisionObject* getOverlappingObject(int index)
 	{
 		return m_overlappingObjects[index];
 	}
 
-	const btCollisionObject*	getOverlappingObject(int index) const
+	const btCollisionObject* getOverlappingObject(int index) const
 	{
 		return m_overlappingObjects[index];
 	}
 
-	btAlignedObjectArray<btCollisionObject*>&	getOverlappingPairs()
+	btAlignedObjectArray<btCollisionObject*>& getOverlappingPairs()
 	{
 		return m_overlappingObjects;
 	}
 
-	const btAlignedObjectArray<btCollisionObject*>	getOverlappingPairs() const
+	const btAlignedObjectArray<btCollisionObject*> getOverlappingPairs() const
 	{
 		return m_overlappingObjects;
 	}
@@ -81,49 +79,43 @@ public:
 	// internal cast
 	//
 
-	static const btGhostObject*	upcast(const btCollisionObject* colObj)
+	static const btGhostObject* upcast(const btCollisionObject* colObj)
 	{
-		if (colObj->getInternalType()==CO_GHOST_OBJECT)
+		if (colObj->getInternalType() == CO_GHOST_OBJECT)
 			return (const btGhostObject*)colObj;
 		return 0;
 	}
-	static btGhostObject*			upcast(btCollisionObject* colObj)
+	static btGhostObject* upcast(btCollisionObject * colObj)
 	{
-		if (colObj->getInternalType()==CO_GHOST_OBJECT)
+		if (colObj->getInternalType() == CO_GHOST_OBJECT)
 			return (btGhostObject*)colObj;
 		return 0;
 	}
-
 };
 
-class	btPairCachingGhostObject : public btGhostObject
+class btPairCachingGhostObject : public btGhostObject
 {
-	btHashedOverlappingPairCache*	m_hashPairCache;
+	btHashedOverlappingPairCache* m_hashPairCache;
 
 public:
-
 	btPairCachingGhostObject();
 
 	virtual ~btPairCachingGhostObject();
 
 	///this method is mainly for expert/internal use only.
-	virtual void	addOverlappingObjectInternal(btBroadphaseProxy* otherProxy, btBroadphaseProxy* thisProxy=0);
+	virtual void addOverlappingObjectInternal(btBroadphaseProxy* otherProxy, btBroadphaseProxy* thisProxy = 0);
 
-	virtual void	removeOverlappingObjectInternal(btBroadphaseProxy* otherProxy,btDispatcher* dispatcher,btBroadphaseProxy* thisProxy=0);
+	virtual void removeOverlappingObjectInternal(btBroadphaseProxy* otherProxy, btDispatcher* dispatcher, btBroadphaseProxy* thisProxy = 0);
 
-	btHashedOverlappingPairCache*	getOverlappingPairCache()
+	btHashedOverlappingPairCache* getOverlappingPairCache()
 	{
 		return m_hashPairCache;
 	}
-
 };
 
-
-
 ///The btGhostPairCallback interfaces and forwards adding and removal of overlapping pairs from the btBroadphaseInterface to btGhostObject.
 class btGhostPairCallback : public btOverlappingPairCallback
 {
-	
 public:
 	btGhostPairCallback()
 	{
@@ -131,15 +123,14 @@ public:
 
 	virtual ~btGhostPairCallback()
 	{
-		
 	}
 
-	virtual btBroadphasePair*	addOverlappingPair(btBroadphaseProxy* proxy0,btBroadphaseProxy* proxy1)
+	virtual btBroadphasePair* addOverlappingPair(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1)
 	{
-		btCollisionObject* colObj0 = (btCollisionObject*) proxy0->m_clientObject;
-		btCollisionObject* colObj1 = (btCollisionObject*) proxy1->m_clientObject;
-		btGhostObject* ghost0 = 		btGhostObject::upcast(colObj0);
-		btGhostObject* ghost1 = 		btGhostObject::upcast(colObj1);
+		btCollisionObject* colObj0 = (btCollisionObject*)proxy0->m_clientObject;
+		btCollisionObject* colObj1 = (btCollisionObject*)proxy1->m_clientObject;
+		btGhostObject* ghost0 = btGhostObject::upcast(colObj0);
+		btGhostObject* ghost1 = btGhostObject::upcast(colObj1);
 		if (ghost0)
 			ghost0->addOverlappingObjectInternal(proxy1, proxy0);
 		if (ghost1)
@@ -147,29 +138,25 @@ public:
 		return 0;
 	}
 
-	virtual void*	removeOverlappingPair(btBroadphaseProxy* proxy0,btBroadphaseProxy* proxy1,btDispatcher* dispatcher)
+	virtual void* removeOverlappingPair(btBroadphaseProxy* proxy0, btBroadphaseProxy* proxy1, btDispatcher* dispatcher)
 	{
-		btCollisionObject* colObj0 = (btCollisionObject*) proxy0->m_clientObject;
-		btCollisionObject* colObj1 = (btCollisionObject*) proxy1->m_clientObject;
-		btGhostObject* ghost0 = 		btGhostObject::upcast(colObj0);
-		btGhostObject* ghost1 = 		btGhostObject::upcast(colObj1);
+		btCollisionObject* colObj0 = (btCollisionObject*)proxy0->m_clientObject;
+		btCollisionObject* colObj1 = (btCollisionObject*)proxy1->m_clientObject;
+		btGhostObject* ghost0 = btGhostObject::upcast(colObj0);
+		btGhostObject* ghost1 = btGhostObject::upcast(colObj1);
 		if (ghost0)
-			ghost0->removeOverlappingObjectInternal(proxy1,dispatcher,proxy0);
+			ghost0->removeOverlappingObjectInternal(proxy1, dispatcher, proxy0);
 		if (ghost1)
-			ghost1->removeOverlappingObjectInternal(proxy0,dispatcher,proxy1);
+			ghost1->removeOverlappingObjectInternal(proxy0, dispatcher, proxy1);
 		return 0;
 	}
 
-	virtual void	removeOverlappingPairsContainingProxy(btBroadphaseProxy* /*proxy0*/,btDispatcher* /*dispatcher*/)
+	virtual void removeOverlappingPairsContainingProxy(btBroadphaseProxy* /*proxy0*/, btDispatcher* /*dispatcher*/)
 	{
 		btAssert(0);
 		//need to keep track of all ghost objects and call them here
 		//m_hashPairCache->removeOverlappingPairsContainingProxy(proxy0,dispatcher);
 	}
-
-	
-
 };
 
 #endif
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btHashedSimplePairCache.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btHashedSimplePairCache.cpp
index 8271981b29d..b686d98d1e7 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btHashedSimplePairCache.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btHashedSimplePairCache.cpp
@@ -13,64 +13,49 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #include "btHashedSimplePairCache.h"
 
-
 #include <stdio.h>
 
 #ifdef BT_DEBUG_COLLISION_PAIRS
-int	gOverlappingSimplePairs = 0;
-int gRemoveSimplePairs =0;
-int gAddedSimplePairs =0;
-int gFindSimplePairs =0;
-#endif //BT_DEBUG_COLLISION_PAIRS
+int gOverlappingSimplePairs = 0;
+int gRemoveSimplePairs = 0;
+int gAddedSimplePairs = 0;
+int gFindSimplePairs = 0;
+#endif  //BT_DEBUG_COLLISION_PAIRS
 
-
-
-btHashedSimplePairCache::btHashedSimplePairCache() {
-	int initialAllocatedSize= 2;
+btHashedSimplePairCache::btHashedSimplePairCache()
+{
+	int initialAllocatedSize = 2;
 	m_overlappingPairArray.reserve(initialAllocatedSize);
 	growTables();
 }
 
-
-
-
 btHashedSimplePairCache::~btHashedSimplePairCache()
 {
 }
 
-
-
-
-
-
 void btHashedSimplePairCache::removeAllPairs()
 {
 	m_overlappingPairArray.clear();
 	m_hashTable.clear();
 	m_next.clear();
 
-	int initialAllocatedSize= 2;
+	int initialAllocatedSize = 2;
 	m_overlappingPairArray.reserve(initialAllocatedSize);
 	growTables();
 }
 
-
-
 btSimplePair* btHashedSimplePairCache::findPair(int indexA, int indexB)
 {
 #ifdef BT_DEBUG_COLLISION_PAIRS
 	gFindSimplePairs++;
 #endif
-	
-	
+
 	/*if (indexA > indexB) 
 		btSwap(indexA, indexB);*/
 
-	int hash = static_cast<int>(getHash(static_cast<unsigned int>(indexA), static_cast<unsigned int>(indexB)) & (m_overlappingPairArray.capacity()-1));
+	int hash = static_cast<int>(getHash(static_cast<unsigned int>(indexA), static_cast<unsigned int>(indexB)) & (m_overlappingPairArray.capacity() - 1));
 
 	if (hash >= m_hashTable.size())
 	{
@@ -95,9 +80,8 @@ btSimplePair* btHashedSimplePairCache::findPair(int indexA, int indexB)
 
 //#include <stdio.h>
 
-void	btHashedSimplePairCache::growTables()
+void btHashedSimplePairCache::growTables()
 {
-
 	int newCapacity = m_overlappingPairArray.capacity();
 
 	if (m_hashTable.size() < newCapacity)
@@ -108,10 +92,9 @@ void	btHashedSimplePairCache::growTables()
 		m_hashTable.resize(newCapacity);
 		m_next.resize(newCapacity);
 
-
 		int i;
 
-		for (i= 0; i < newCapacity; ++i)
+		for (i = 0; i < newCapacity; ++i)
 		{
 			m_hashTable[i] = BT_SIMPLE_NULL_PAIR;
 		}
@@ -120,27 +103,22 @@ void	btHashedSimplePairCache::growTables()
 			m_next[i] = BT_SIMPLE_NULL_PAIR;
 		}
 
-		for(i=0;i<curHashtableSize;i++)
+		for (i = 0; i < curHashtableSize; i++)
 		{
-	
 			const btSimplePair& pair = m_overlappingPairArray[i];
 			int indexA = pair.m_indexA;
 			int indexB = pair.m_indexB;
-			
-			int	hashValue = static_cast<int>(getHash(static_cast<unsigned int>(indexA),static_cast<unsigned int>(indexB)) & (m_overlappingPairArray.capacity()-1));	// New hash value with new mask
+
+			int hashValue = static_cast<int>(getHash(static_cast<unsigned int>(indexA), static_cast<unsigned int>(indexB)) & (m_overlappingPairArray.capacity() - 1));  // New hash value with new mask
 			m_next[i] = m_hashTable[hashValue];
 			m_hashTable[hashValue] = i;
 		}
-
-
 	}
 }
 
 btSimplePair* btHashedSimplePairCache::internalAddPair(int indexA, int indexB)
 {
-
-	int	hash = static_cast<int>(getHash(static_cast<unsigned int>(indexA),static_cast<unsigned int>(indexB)) & (m_overlappingPairArray.capacity()-1));	// New hash value with new mask
-
+	int hash = static_cast<int>(getHash(static_cast<unsigned int>(indexA), static_cast<unsigned int>(indexB)) & (m_overlappingPairArray.capacity() - 1));  // New hash value with new mask
 
 	btSimplePair* pair = internalFindPair(indexA, indexB, hash);
 	if (pair != NULL)
@@ -158,32 +136,29 @@ btSimplePair* btHashedSimplePairCache::internalAddPair(int indexA, int indexB)
 	{
 		growTables();
 		//hash with new capacity
-		hash = static_cast<int>(getHash(static_cast<unsigned int>(indexA),static_cast<unsigned int>(indexB)) & (m_overlappingPairArray.capacity()-1));
+		hash = static_cast<int>(getHash(static_cast<unsigned int>(indexA), static_cast<unsigned int>(indexB)) & (m_overlappingPairArray.capacity() - 1));
 	}
-	
-	pair = new (mem) btSimplePair(indexA,indexB);
+
+	pair = new (mem) btSimplePair(indexA, indexB);
 
 	pair->m_userPointer = 0;
-	
+
 	m_next[count] = m_hashTable[hash];
 	m_hashTable[hash] = count;
 
 	return pair;
 }
 
-
-
 void* btHashedSimplePairCache::removeOverlappingPair(int indexA, int indexB)
 {
 #ifdef BT_DEBUG_COLLISION_PAIRS
 	gRemoveSimplePairs++;
 #endif
-	
 
 	/*if (indexA > indexB) 
 		btSwap(indexA, indexB);*/
 
-	int	hash = static_cast<int>(getHash(static_cast<unsigned int>(indexA),static_cast<unsigned int>(indexB)) & (m_overlappingPairArray.capacity()-1));
+	int hash = static_cast<int>(getHash(static_cast<unsigned int>(indexA), static_cast<unsigned int>(indexB)) & (m_overlappingPairArray.capacity() - 1));
 
 	btSimplePair* pair = internalFindPair(indexA, indexB, hash);
 	if (pair == NULL)
@@ -191,10 +166,8 @@ void* btHashedSimplePairCache::removeOverlappingPair(int indexA, int indexB)
 		return 0;
 	}
 
-	
 	void* userData = pair->m_userPointer;
 
-
 	int pairIndex = int(pair - &m_overlappingPairArray[0]);
 	btAssert(pairIndex < m_overlappingPairArray.size());
 
@@ -234,8 +207,8 @@ void* btHashedSimplePairCache::removeOverlappingPair(int indexA, int indexB)
 
 	// Remove the last pair from the hash table.
 	const btSimplePair* last = &m_overlappingPairArray[lastPairIndex];
-		/* missing swap here too, Nat. */ 
-	int lastHash = static_cast<int>(getHash(static_cast<unsigned int>(last->m_indexA), static_cast<unsigned int>(last->m_indexB)) & (m_overlappingPairArray.capacity()-1));
+	/* missing swap here too, Nat. */
+	int lastHash = static_cast<int>(getHash(static_cast<unsigned int>(last->m_indexA), static_cast<unsigned int>(last->m_indexB)) & (m_overlappingPairArray.capacity() - 1));
 
 	index = m_hashTable[lastHash];
 	btAssert(index != BT_SIMPLE_NULL_PAIR);
@@ -269,13 +242,3 @@ void* btHashedSimplePairCache::removeOverlappingPair(int indexA, int indexB)
 	return userData;
 }
 //#include <stdio.h>
-
-
-
-
-
-
-
-
-
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btHashedSimplePairCache.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btHashedSimplePairCache.h
index 318981cda1c..fd38a4f0e19 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btHashedSimplePairCache.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btHashedSimplePairCache.h
@@ -16,144 +16,126 @@ subject to the following restrictions:
 #ifndef BT_HASHED_SIMPLE_PAIR_CACHE_H
 #define BT_HASHED_SIMPLE_PAIR_CACHE_H
 
-
-
 #include "LinearMath/btAlignedObjectArray.h"
 
-const int BT_SIMPLE_NULL_PAIR=0xffffffff;
+const int BT_SIMPLE_NULL_PAIR = 0xffffffff;
 
 struct btSimplePair
 {
-	btSimplePair(int indexA,int indexB)
-		:m_indexA(indexA),
-		m_indexB(indexB),
-		m_userPointer(0)
+	btSimplePair(int indexA, int indexB)
+		: m_indexA(indexA),
+		  m_indexB(indexB),
+		  m_userPointer(0)
 	{
 	}
 
 	int m_indexA;
 	int m_indexB;
-	union
-	{
-		void*	m_userPointer;
-		int		m_userValue;
+	union {
+		void* m_userPointer;
+		int m_userValue;
 	};
 };
 
-typedef btAlignedObjectArray<btSimplePair>	btSimplePairArray;
-
+typedef btAlignedObjectArray<btSimplePair> btSimplePairArray;
 
 #ifdef BT_DEBUG_COLLISION_PAIRS
 extern int gOverlappingSimplePairs;
 extern int gRemoveSimplePairs;
 extern int gAddedSimplePairs;
 extern int gFindSimplePairs;
-#endif //BT_DEBUG_COLLISION_PAIRS
-
-
+#endif  //BT_DEBUG_COLLISION_PAIRS
 
 class btHashedSimplePairCache
 {
-	btSimplePairArray	m_overlappingPairArray;
-		
+	btSimplePairArray m_overlappingPairArray;
 
 protected:
-	
-	btAlignedObjectArray<int>	m_hashTable;
-	btAlignedObjectArray<int>	m_next;
-	
+	btAlignedObjectArray<int> m_hashTable;
+	btAlignedObjectArray<int> m_next;
 
 public:
 	btHashedSimplePairCache();
 	virtual ~btHashedSimplePairCache();
-	
+
 	void removeAllPairs();
 
-	virtual void*	removeOverlappingPair(int indexA,int indexB);
-	
+	virtual void* removeOverlappingPair(int indexA, int indexB);
+
 	// Add a pair and return the new pair. If the pair already exists,
 	// no new pair is created and the old one is returned.
-	virtual btSimplePair* 	addOverlappingPair(int indexA,int indexB)
+	virtual btSimplePair* addOverlappingPair(int indexA, int indexB)
 	{
 #ifdef BT_DEBUG_COLLISION_PAIRS
 		gAddedSimplePairs++;
 #endif
 
-		return internalAddPair(indexA,indexB);
+		return internalAddPair(indexA, indexB);
 	}
 
-	
-	virtual btSimplePair*	getOverlappingPairArrayPtr()
+	virtual btSimplePair* getOverlappingPairArrayPtr()
 	{
 		return &m_overlappingPairArray[0];
 	}
 
-	const btSimplePair*	getOverlappingPairArrayPtr() const
+	const btSimplePair* getOverlappingPairArrayPtr() const
 	{
 		return &m_overlappingPairArray[0];
 	}
 
-	btSimplePairArray&	getOverlappingPairArray()
+	btSimplePairArray& getOverlappingPairArray()
 	{
 		return m_overlappingPairArray;
 	}
 
-	const btSimplePairArray&	getOverlappingPairArray() const
+	const btSimplePairArray& getOverlappingPairArray() const
 	{
 		return m_overlappingPairArray;
 	}
 
-	
-	btSimplePair* findPair(int indexA,int indexB);
+	btSimplePair* findPair(int indexA, int indexB);
 
 	int GetCount() const { return m_overlappingPairArray.size(); }
 
-	int	getNumOverlappingPairs() const
+	int getNumOverlappingPairs() const
 	{
 		return m_overlappingPairArray.size();
 	}
-private:
-	
-	btSimplePair* 	internalAddPair(int indexA, int indexB);
 
-	void	growTables();
+private:
+	btSimplePair* internalAddPair(int indexA, int indexB);
+
+	void growTables();
 
 	SIMD_FORCE_INLINE bool equalsPair(const btSimplePair& pair, int indexA, int indexB)
-	{	
+	{
 		return pair.m_indexA == indexA && pair.m_indexB == indexB;
 	}
 
-	
-	
 	SIMD_FORCE_INLINE unsigned int getHash(unsigned int indexA, unsigned int indexB)
 	{
 		unsigned int key = indexA | (indexB << 16);
 		// Thomas Wang's hash
 
 		key += ~(key << 15);
-		key ^=  (key >> 10);
-		key +=  (key << 3);
-		key ^=  (key >> 6);
+		key ^= (key >> 10);
+		key += (key << 3);
+		key ^= (key >> 6);
 		key += ~(key << 11);
-		key ^=  (key >> 16);
+		key ^= (key >> 16);
 		return key;
 	}
-	
 
-
-
-
-	SIMD_FORCE_INLINE btSimplePair* internalFindPair(int proxyIdA , int proxyIdB, int hash)
+	SIMD_FORCE_INLINE btSimplePair* internalFindPair(int proxyIdA, int proxyIdB, int hash)
 	{
-		
 		int index = m_hashTable[hash];
-		
-		while( index != BT_SIMPLE_NULL_PAIR && equalsPair(m_overlappingPairArray[index], proxyIdA, proxyIdB) == false)
+
+		while (index != BT_SIMPLE_NULL_PAIR && equalsPair(m_overlappingPairArray[index], proxyIdA, proxyIdB) == false)
 		{
 			index = m_next[index];
 		}
 
-		if ( index == BT_SIMPLE_NULL_PAIR )
+		if (index == BT_SIMPLE_NULL_PAIR)
 		{
 			return NULL;
 		}
@@ -162,13 +144,6 @@ private:
 
 		return &m_overlappingPairArray[index];
 	}
-
-	
 };
 
-
-
-
-#endif //BT_HASHED_SIMPLE_PAIR_CACHE_H
-
-
+#endif  //BT_HASHED_SIMPLE_PAIR_CACHE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btInternalEdgeUtility.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btInternalEdgeUtility.cpp
index 898320ee1ae..e74c83f9f17 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btInternalEdgeUtility.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btInternalEdgeUtility.cpp
@@ -12,50 +12,44 @@
 
 #ifdef DEBUG_INTERNAL_EDGE
 #include <stdio.h>
-#endif //DEBUG_INTERNAL_EDGE
-
+#endif  //DEBUG_INTERNAL_EDGE
 
 #ifdef BT_INTERNAL_EDGE_DEBUG_DRAW
 static btIDebugDraw* gDebugDrawer = 0;
 
-void	btSetDebugDrawer(btIDebugDraw* debugDrawer)
+void btSetDebugDrawer(btIDebugDraw* debugDrawer)
 {
 	gDebugDrawer = debugDrawer;
 }
 
-static void    btDebugDrawLine(const btVector3& from,const btVector3& to, const btVector3& color)
+static void btDebugDrawLine(const btVector3& from, const btVector3& to, const btVector3& color)
 {
 	if (gDebugDrawer)
-		gDebugDrawer->drawLine(from,to,color);
+		gDebugDrawer->drawLine(from, to, color);
 }
-#endif //BT_INTERNAL_EDGE_DEBUG_DRAW
+#endif  //BT_INTERNAL_EDGE_DEBUG_DRAW
 
-
-static int	btGetHash(int partId, int triangleIndex)
+static int btGetHash(int partId, int triangleIndex)
 {
-	int hash = (partId<<(31-MAX_NUM_PARTS_IN_BITS)) | triangleIndex;
+	int hash = (partId << (31 - MAX_NUM_PARTS_IN_BITS)) | triangleIndex;
 	return hash;
 }
 
-
-
-static btScalar btGetAngle(const btVector3& edgeA, const btVector3& normalA,const btVector3& normalB)
+static btScalar btGetAngle(const btVector3& edgeA, const btVector3& normalA, const btVector3& normalB)
 {
-	const btVector3 refAxis0  = edgeA;
-	const btVector3 refAxis1  = normalA;
+	const btVector3 refAxis0 = edgeA;
+	const btVector3 refAxis1 = normalA;
 	const btVector3 swingAxis = normalB;
 	btScalar angle = btAtan2(swingAxis.dot(refAxis0), swingAxis.dot(refAxis1));
-	return  angle;
+	return angle;
 }
 
-
 struct btConnectivityProcessor : public btTriangleCallback
 {
-	int				m_partIdA;
-	int				m_triangleIndexA;
-	btVector3*		m_triangleVerticesA;
-	btTriangleInfoMap*	m_triangleInfoMap;
-
+	int m_partIdA;
+	int m_triangleIndexA;
+	btVector3* m_triangleVerticesA;
+	btTriangleInfoMap* m_triangleInfoMap;
 
 	virtual void processTriangle(btVector3* triangle, int partId, int triangleIndex)
 	{
@@ -69,18 +63,17 @@ struct btConnectivityProcessor : public btTriangleCallback
 
 		//search for shared vertices and edges
 		int numshared = 0;
-		int sharedVertsA[3]={-1,-1,-1};
-		int sharedVertsB[3]={-1,-1,-1};
+		int sharedVertsA[3] = {-1, -1, -1};
+		int sharedVertsB[3] = {-1, -1, -1};
 
 		///skip degenerate triangles
-		btScalar crossBSqr = ((triangle[1]-triangle[0]).cross(triangle[2]-triangle[0])).length2();
+		btScalar crossBSqr = ((triangle[1] - triangle[0]).cross(triangle[2] - triangle[0])).length2();
 		if (crossBSqr < m_triangleInfoMap->m_equalVertexThreshold)
 			return;
 
-
-		btScalar crossASqr = ((m_triangleVerticesA[1]-m_triangleVerticesA[0]).cross(m_triangleVerticesA[2]-m_triangleVerticesA[0])).length2();
+		btScalar crossASqr = ((m_triangleVerticesA[1] - m_triangleVerticesA[0]).cross(m_triangleVerticesA[2] - m_triangleVerticesA[0])).length2();
 		///skip degenerate triangles
-		if (crossASqr< m_triangleInfoMap->m_equalVertexThreshold)
+		if (crossASqr < m_triangleInfoMap->m_equalVertexThreshold)
 			return;
 
 #if 0
@@ -96,36 +89,36 @@ struct btConnectivityProcessor : public btTriangleCallback
 			triangle[2].getX(),triangle[2].getY(),triangle[2].getZ());
 #endif
 
-		for (int i=0;i<3;i++)
+		for (int i = 0; i < 3; i++)
 		{
-			for (int j=0;j<3;j++)
+			for (int j = 0; j < 3; j++)
 			{
-				if ( (m_triangleVerticesA[i]-triangle[j]).length2() < m_triangleInfoMap->m_equalVertexThreshold)
+				if ((m_triangleVerticesA[i] - triangle[j]).length2() < m_triangleInfoMap->m_equalVertexThreshold)
 				{
 					sharedVertsA[numshared] = i;
 					sharedVertsB[numshared] = j;
 					numshared++;
 					///degenerate case
-					if(numshared >= 3)
+					if (numshared >= 3)
 						return;
 				}
 			}
 			///degenerate case
-			if(numshared >= 3)
+			if (numshared >= 3)
 				return;
 		}
 		switch (numshared)
 		{
-		case 0:
+			case 0:
 			{
 				break;
 			}
-		case 1:
+			case 1:
 			{
 				//shared vertex
 				break;
 			}
-		case 2:
+			case 2:
 			{
 				//shared edge
 				//we need to make sure the edge is in the order V2V0 and not V0V2 so that the signs are correct
@@ -138,26 +131,25 @@ struct btConnectivityProcessor : public btTriangleCallback
 					sharedVertsB[0] = tmp;
 				}
 
-				int hash = btGetHash(m_partIdA,m_triangleIndexA);
+				int hash = btGetHash(m_partIdA, m_triangleIndexA);
 
 				btTriangleInfo* info = m_triangleInfoMap->find(hash);
 				if (!info)
 				{
 					btTriangleInfo tmp;
-					m_triangleInfoMap->insert(hash,tmp);
+					m_triangleInfoMap->insert(hash, tmp);
 					info = m_triangleInfoMap->find(hash);
 				}
 
-				int sumvertsA = sharedVertsA[0]+sharedVertsA[1];
-				int otherIndexA = 3-sumvertsA;
+				int sumvertsA = sharedVertsA[0] + sharedVertsA[1];
+				int otherIndexA = 3 - sumvertsA;
 
-				
-				btVector3 edge(m_triangleVerticesA[sharedVertsA[1]]-m_triangleVerticesA[sharedVertsA[0]]);
+				btVector3 edge(m_triangleVerticesA[sharedVertsA[1]] - m_triangleVerticesA[sharedVertsA[0]]);
 
-				btTriangleShape tA(m_triangleVerticesA[0],m_triangleVerticesA[1],m_triangleVerticesA[2]);
-				int otherIndexB = 3-(sharedVertsB[0]+sharedVertsB[1]);
+				btTriangleShape tA(m_triangleVerticesA[0], m_triangleVerticesA[1], m_triangleVerticesA[2]);
+				int otherIndexB = 3 - (sharedVertsB[0] + sharedVertsB[1]);
 
-				btTriangleShape tB(triangle[sharedVertsB[1]],triangle[sharedVertsB[0]],triangle[otherIndexB]);
+				btTriangleShape tB(triangle[sharedVertsB[1]], triangle[sharedVertsB[0]], triangle[otherIndexB]);
 				//btTriangleShape tB(triangle[0],triangle[1],triangle[2]);
 
 				btVector3 normalA;
@@ -168,26 +160,25 @@ struct btConnectivityProcessor : public btTriangleCallback
 				btVector3 edgeCrossA = edge.cross(normalA).normalize();
 
 				{
-					btVector3 tmp = m_triangleVerticesA[otherIndexA]-m_triangleVerticesA[sharedVertsA[0]];
+					btVector3 tmp = m_triangleVerticesA[otherIndexA] - m_triangleVerticesA[sharedVertsA[0]];
 					if (edgeCrossA.dot(tmp) < 0)
 					{
-						edgeCrossA*=-1;
+						edgeCrossA *= -1;
 					}
 				}
 
 				btVector3 edgeCrossB = edge.cross(normalB).normalize();
 
 				{
-					btVector3 tmp = triangle[otherIndexB]-triangle[sharedVertsB[0]];
+					btVector3 tmp = triangle[otherIndexB] - triangle[sharedVertsB[0]];
 					if (edgeCrossB.dot(tmp) < 0)
 					{
-						edgeCrossB*=-1;
+						edgeCrossB *= -1;
 					}
 				}
 
-				btScalar	angle2 = 0;
-				btScalar	ang4 = 0.f;
-
+				btScalar angle2 = 0;
+				btScalar ang4 = 0.f;
 
 				btVector3 calculatedEdge = edgeCrossA.cross(edgeCrossB);
 				btScalar len2 = calculatedEdge.length2();
@@ -196,52 +187,47 @@ struct btConnectivityProcessor : public btTriangleCallback
 				//btVector3 calculatedNormalB = normalA;
 				bool isConvex = false;
 
-				if (len2<m_triangleInfoMap->m_planarEpsilon)
+				if (len2 < m_triangleInfoMap->m_planarEpsilon)
 				{
 					angle2 = 0.f;
 					ang4 = 0.f;
-				} else
+				}
+				else
 				{
-
 					calculatedEdge.normalize();
 					btVector3 calculatedNormalA = calculatedEdge.cross(edgeCrossA);
 					calculatedNormalA.normalize();
-					angle2 = btGetAngle(calculatedNormalA,edgeCrossA,edgeCrossB);
-					ang4 = SIMD_PI-angle2;
+					angle2 = btGetAngle(calculatedNormalA, edgeCrossA, edgeCrossB);
+					ang4 = SIMD_PI - angle2;
 					btScalar dotA = normalA.dot(edgeCrossB);
 					///@todo: check if we need some epsilon, due to floating point imprecision
-					isConvex = (dotA<0.);
+					isConvex = (dotA < 0.);
 
 					correctedAngle = isConvex ? ang4 : -ang4;
 				}
 
-				
-
-				
-							
-				//alternatively use 
+				//alternatively use
 				//btVector3 calculatedNormalB2 = quatRotate(orn,normalA);
 
-
 				switch (sumvertsA)
 				{
-				case 1:
+					case 1:
 					{
-						btVector3 edge = m_triangleVerticesA[0]-m_triangleVerticesA[1];
-						btQuaternion orn(edge,-correctedAngle);
-						btVector3 computedNormalB = quatRotate(orn,normalA);
+						btVector3 edge = m_triangleVerticesA[0] - m_triangleVerticesA[1];
+						btQuaternion orn(edge, -correctedAngle);
+						btVector3 computedNormalB = quatRotate(orn, normalA);
 						btScalar bla = computedNormalB.dot(normalB);
-						if (bla<0)
+						if (bla < 0)
 						{
-							computedNormalB*=-1;
+							computedNormalB *= -1;
 							info->m_flags |= TRI_INFO_V0V1_SWAP_NORMALB;
 						}
 #ifdef DEBUG_INTERNAL_EDGE
-						if ((computedNormalB-normalB).length()>0.0001)
+						if ((computedNormalB - normalB).length() > 0.0001)
 						{
 							printf("warning: normals not identical\n");
 						}
-#endif//DEBUG_INTERNAL_EDGE
+#endif  //DEBUG_INTERNAL_EDGE
 
 						info->m_edgeV0V1Angle = -correctedAngle;
 
@@ -249,44 +235,44 @@ struct btConnectivityProcessor : public btTriangleCallback
 							info->m_flags |= TRI_INFO_V0V1_CONVEX;
 						break;
 					}
-				case 2:
+					case 2:
 					{
-						btVector3 edge = m_triangleVerticesA[2]-m_triangleVerticesA[0];
-						btQuaternion orn(edge,-correctedAngle);
-						btVector3 computedNormalB = quatRotate(orn,normalA);
-						if (computedNormalB.dot(normalB)<0)
+						btVector3 edge = m_triangleVerticesA[2] - m_triangleVerticesA[0];
+						btQuaternion orn(edge, -correctedAngle);
+						btVector3 computedNormalB = quatRotate(orn, normalA);
+						if (computedNormalB.dot(normalB) < 0)
 						{
-							computedNormalB*=-1;
+							computedNormalB *= -1;
 							info->m_flags |= TRI_INFO_V2V0_SWAP_NORMALB;
 						}
 
 #ifdef DEBUG_INTERNAL_EDGE
-						if ((computedNormalB-normalB).length()>0.0001)
+						if ((computedNormalB - normalB).length() > 0.0001)
 						{
 							printf("warning: normals not identical\n");
 						}
-#endif //DEBUG_INTERNAL_EDGE
+#endif  //DEBUG_INTERNAL_EDGE
 						info->m_edgeV2V0Angle = -correctedAngle;
 						if (isConvex)
 							info->m_flags |= TRI_INFO_V2V0_CONVEX;
-						break;	
+						break;
 					}
-				case 3:
+					case 3:
 					{
-						btVector3 edge = m_triangleVerticesA[1]-m_triangleVerticesA[2];
-						btQuaternion orn(edge,-correctedAngle);
-						btVector3 computedNormalB = quatRotate(orn,normalA);
-						if (computedNormalB.dot(normalB)<0)
+						btVector3 edge = m_triangleVerticesA[1] - m_triangleVerticesA[2];
+						btQuaternion orn(edge, -correctedAngle);
+						btVector3 computedNormalB = quatRotate(orn, normalA);
+						if (computedNormalB.dot(normalB) < 0)
 						{
 							info->m_flags |= TRI_INFO_V1V2_SWAP_NORMALB;
-							computedNormalB*=-1;
+							computedNormalB *= -1;
 						}
 #ifdef DEBUG_INTERNAL_EDGE
-						if ((computedNormalB-normalB).length()>0.0001)
+						if ((computedNormalB - normalB).length() > 0.0001)
 						{
 							printf("warning: normals not identical\n");
 						}
-#endif //DEBUG_INTERNAL_EDGE
+#endif  //DEBUG_INTERNAL_EDGE
 						info->m_edgeV1V2Angle = -correctedAngle;
 
 						if (isConvex)
@@ -297,18 +283,17 @@ struct btConnectivityProcessor : public btTriangleCallback
 
 				break;
 			}
-		default:
+			default:
 			{
 				//				printf("warning: duplicate triangle\n");
 			}
-
 		}
 	}
 };
 /////////////////////////////////////////////////////////
 /////////////////////////////////////////////////////////
 
-void btGenerateInternalEdgeInfo (btBvhTriangleMeshShape*trimeshShape, btTriangleInfoMap* triangleInfoMap)
+void btGenerateInternalEdgeInfo(btBvhTriangleMeshShape* trimeshShape, btTriangleInfoMap* triangleInfoMap)
 {
 	//the user pointer shouldn't already be used for other purposes, we intend to store connectivity info there!
 	if (trimeshShape->getTriangleInfoMap())
@@ -319,46 +304,45 @@ void btGenerateInternalEdgeInfo (btBvhTriangleMeshShape*trimeshShape, btTriangle
 	btStridingMeshInterface* meshInterface = trimeshShape->getMeshInterface();
 	const btVector3& meshScaling = meshInterface->getScaling();
 
-	for (int partId = 0; partId< meshInterface->getNumSubParts();partId++)
+	for (int partId = 0; partId < meshInterface->getNumSubParts(); partId++)
 	{
-		const unsigned char *vertexbase = 0;
+		const unsigned char* vertexbase = 0;
 		int numverts = 0;
 		PHY_ScalarType type = PHY_INTEGER;
 		int stride = 0;
-		const unsigned char *indexbase = 0;
+		const unsigned char* indexbase = 0;
 		int indexstride = 0;
 		int numfaces = 0;
 		PHY_ScalarType indicestype = PHY_INTEGER;
 		//PHY_ScalarType indexType=0;
 
 		btVector3 triangleVerts[3];
-		meshInterface->getLockedReadOnlyVertexIndexBase(&vertexbase,numverts,	type,stride,&indexbase,indexstride,numfaces,indicestype,partId);
-		btVector3 aabbMin,aabbMax;
+		meshInterface->getLockedReadOnlyVertexIndexBase(&vertexbase, numverts, type, stride, &indexbase, indexstride, numfaces, indicestype, partId);
+		btVector3 aabbMin, aabbMax;
 
-		for (int triangleIndex = 0 ; triangleIndex < numfaces;triangleIndex++)
+		for (int triangleIndex = 0; triangleIndex < numfaces; triangleIndex++)
 		{
-			unsigned int* gfxbase = (unsigned int*)(indexbase+triangleIndex*indexstride);
+			unsigned int* gfxbase = (unsigned int*)(indexbase + triangleIndex * indexstride);
 
-			for (int j=2;j>=0;j--)
+			for (int j = 2; j >= 0; j--)
 			{
-
-				int graphicsindex = indicestype==PHY_SHORT?((unsigned short*)gfxbase)[j]:gfxbase[j];
+				int graphicsindex = indicestype == PHY_SHORT ? ((unsigned short*)gfxbase)[j] : gfxbase[j];
 				if (type == PHY_FLOAT)
 				{
-					float* graphicsbase = (float*)(vertexbase+graphicsindex*stride);
+					float* graphicsbase = (float*)(vertexbase + graphicsindex * stride);
 					triangleVerts[j] = btVector3(
-						graphicsbase[0]*meshScaling.getX(),
-						graphicsbase[1]*meshScaling.getY(),
-						graphicsbase[2]*meshScaling.getZ());
+						graphicsbase[0] * meshScaling.getX(),
+						graphicsbase[1] * meshScaling.getY(),
+						graphicsbase[2] * meshScaling.getZ());
 				}
 				else
 				{
-					double* graphicsbase = (double*)(vertexbase+graphicsindex*stride);
-					triangleVerts[j] = btVector3( btScalar(graphicsbase[0]*meshScaling.getX()), btScalar(graphicsbase[1]*meshScaling.getY()), btScalar(graphicsbase[2]*meshScaling.getZ()));
+					double* graphicsbase = (double*)(vertexbase + graphicsindex * stride);
+					triangleVerts[j] = btVector3(btScalar(graphicsbase[0] * meshScaling.getX()), btScalar(graphicsbase[1] * meshScaling.getY()), btScalar(graphicsbase[2] * meshScaling.getZ()));
 				}
 			}
-			aabbMin.setValue(btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT));
-			aabbMax.setValue(btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT)); 
+			aabbMin.setValue(btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT));
+			aabbMax.setValue(btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT));
 			aabbMin.setMin(triangleVerts[0]);
 			aabbMax.setMax(triangleVerts[0]);
 			aabbMin.setMin(triangleVerts[1]);
@@ -370,140 +354,127 @@ void btGenerateInternalEdgeInfo (btBvhTriangleMeshShape*trimeshShape, btTriangle
 			connectivityProcessor.m_partIdA = partId;
 			connectivityProcessor.m_triangleIndexA = triangleIndex;
 			connectivityProcessor.m_triangleVerticesA = &triangleVerts[0];
-			connectivityProcessor.m_triangleInfoMap  = triangleInfoMap;
+			connectivityProcessor.m_triangleInfoMap = triangleInfoMap;
 
-			trimeshShape->processAllTriangles(&connectivityProcessor,aabbMin,aabbMax);
+			trimeshShape->processAllTriangles(&connectivityProcessor, aabbMin, aabbMax);
 		}
-
 	}
-
 }
 
-
-
-
 // Given a point and a line segment (defined by two points), compute the closest point
 // in the line.  Cap the point at the endpoints of the line segment.
-void btNearestPointInLineSegment(const btVector3 &point, const btVector3& line0, const btVector3& line1, btVector3& nearestPoint)
+void btNearestPointInLineSegment(const btVector3& point, const btVector3& line0, const btVector3& line1, btVector3& nearestPoint)
 {
-	btVector3 lineDelta     = line1 - line0;
+	btVector3 lineDelta = line1 - line0;
 
 	// Handle degenerate lines
-	if ( lineDelta.fuzzyZero())
+	if (lineDelta.fuzzyZero())
 	{
 		nearestPoint = line0;
 	}
 	else
 	{
-		btScalar delta = (point-line0).dot(lineDelta) / (lineDelta).dot(lineDelta);
+		btScalar delta = (point - line0).dot(lineDelta) / (lineDelta).dot(lineDelta);
 
 		// Clamp the point to conform to the segment's endpoints
-		if ( delta < 0 )
+		if (delta < 0)
 			delta = 0;
-		else if ( delta > 1 )
+		else if (delta > 1)
 			delta = 1;
 
-		nearestPoint = line0 + lineDelta*delta;
+		nearestPoint = line0 + lineDelta * delta;
 	}
 }
 
-
-
-
-bool	btClampNormal(const btVector3& edge,const btVector3& tri_normal_org,const btVector3& localContactNormalOnB, btScalar correctedEdgeAngle, btVector3 & clampedLocalNormal)
+bool btClampNormal(const btVector3& edge, const btVector3& tri_normal_org, const btVector3& localContactNormalOnB, btScalar correctedEdgeAngle, btVector3& clampedLocalNormal)
 {
 	btVector3 tri_normal = tri_normal_org;
 	//we only have a local triangle normal, not a local contact normal -> only normal in world space...
 	//either compute the current angle all in local space, or all in world space
 
 	btVector3 edgeCross = edge.cross(tri_normal).normalize();
-	btScalar curAngle = btGetAngle(edgeCross,tri_normal,localContactNormalOnB);
+	btScalar curAngle = btGetAngle(edgeCross, tri_normal, localContactNormalOnB);
 
-	if (correctedEdgeAngle<0)
+	if (correctedEdgeAngle < 0)
 	{
 		if (curAngle < correctedEdgeAngle)
 		{
-			btScalar diffAngle = correctedEdgeAngle-curAngle;
-			btQuaternion rotation(edge,diffAngle );
-			clampedLocalNormal = btMatrix3x3(rotation)*localContactNormalOnB;
+			btScalar diffAngle = correctedEdgeAngle - curAngle;
+			btQuaternion rotation(edge, diffAngle);
+			clampedLocalNormal = btMatrix3x3(rotation) * localContactNormalOnB;
 			return true;
 		}
 	}
 
-	if (correctedEdgeAngle>=0)
+	if (correctedEdgeAngle >= 0)
 	{
 		if (curAngle > correctedEdgeAngle)
 		{
-			btScalar diffAngle = correctedEdgeAngle-curAngle;
-			btQuaternion rotation(edge,diffAngle );
-			clampedLocalNormal = btMatrix3x3(rotation)*localContactNormalOnB;
+			btScalar diffAngle = correctedEdgeAngle - curAngle;
+			btQuaternion rotation(edge, diffAngle);
+			clampedLocalNormal = btMatrix3x3(rotation) * localContactNormalOnB;
 			return true;
 		}
 	}
 	return false;
 }
 
-
-
 /// Changes a btManifoldPoint collision normal to the normal from the mesh.
-void btAdjustInternalEdgeContacts(btManifoldPoint& cp, const btCollisionObjectWrapper* colObj0Wrap,const btCollisionObjectWrapper* colObj1Wrap, int partId0, int index0, int normalAdjustFlags)
+void btAdjustInternalEdgeContacts(btManifoldPoint& cp, const btCollisionObjectWrapper* colObj0Wrap, const btCollisionObjectWrapper* colObj1Wrap, int partId0, int index0, int normalAdjustFlags)
 {
 	//btAssert(colObj0->getCollisionShape()->getShapeType() == TRIANGLE_SHAPE_PROXYTYPE);
 	if (colObj0Wrap->getCollisionShape()->getShapeType() != TRIANGLE_SHAPE_PROXYTYPE)
 		return;
 
 	btBvhTriangleMeshShape* trimesh = 0;
-	
-	if( colObj0Wrap->getCollisionObject()->getCollisionShape()->getShapeType() == SCALED_TRIANGLE_MESH_SHAPE_PROXYTYPE )
+
+	if (colObj0Wrap->getCollisionObject()->getCollisionShape()->getShapeType() == SCALED_TRIANGLE_MESH_SHAPE_PROXYTYPE)
 	{
 		trimesh = ((btScaledBvhTriangleMeshShape*)colObj0Wrap->getCollisionObject()->getCollisionShape())->getChildShape();
 	}
 	else
 	{
-		if (colObj0Wrap->getCollisionObject()->getCollisionShape()->getShapeType()==TRIANGLE_MESH_SHAPE_PROXYTYPE)
+		if (colObj0Wrap->getCollisionObject()->getCollisionShape()->getShapeType() == TRIANGLE_MESH_SHAPE_PROXYTYPE)
 		{
 			trimesh = (btBvhTriangleMeshShape*)colObj0Wrap->getCollisionObject()->getCollisionShape();
 		}
 	}
-	if (trimesh==0)
+	if (trimesh == 0)
 		return;
 
-	btTriangleInfoMap* triangleInfoMapPtr = (btTriangleInfoMap*) trimesh->getTriangleInfoMap();
+	btTriangleInfoMap* triangleInfoMapPtr = (btTriangleInfoMap*)trimesh->getTriangleInfoMap();
 	if (!triangleInfoMapPtr)
 		return;
 
-	int hash = btGetHash(partId0,index0);
-
+	int hash = btGetHash(partId0, index0);
 
 	btTriangleInfo* info = triangleInfoMapPtr->find(hash);
 	if (!info)
 		return;
 
-	btScalar frontFacing = (normalAdjustFlags & BT_TRIANGLE_CONVEX_BACKFACE_MODE)==0? 1.f : -1.f;
-	
+	btScalar frontFacing = (normalAdjustFlags & BT_TRIANGLE_CONVEX_BACKFACE_MODE) == 0 ? 1.f : -1.f;
+
 	const btTriangleShape* tri_shape = static_cast<const btTriangleShape*>(colObj0Wrap->getCollisionShape());
-	btVector3 v0,v1,v2;
-	tri_shape->getVertex(0,v0);
-	tri_shape->getVertex(1,v1);
-	tri_shape->getVertex(2,v2);
+	btVector3 v0, v1, v2;
+	tri_shape->getVertex(0, v0);
+	tri_shape->getVertex(1, v1);
+	tri_shape->getVertex(2, v2);
 
 	//btVector3 center = (v0+v1+v2)*btScalar(1./3.);
 
-	btVector3 red(1,0,0), green(0,1,0),blue(0,0,1),white(1,1,1),black(0,0,0);
+	btVector3 red(1, 0, 0), green(0, 1, 0), blue(0, 0, 1), white(1, 1, 1), black(0, 0, 0);
 	btVector3 tri_normal;
 	tri_shape->calcNormal(tri_normal);
 
 	//btScalar dot = tri_normal.dot(cp.m_normalWorldOnB);
 	btVector3 nearest;
-	btNearestPointInLineSegment(cp.m_localPointB,v0,v1,nearest);
+	btNearestPointInLineSegment(cp.m_localPointB, v0, v1, nearest);
 
 	btVector3 contact = cp.m_localPointB;
 #ifdef BT_INTERNAL_EDGE_DEBUG_DRAW
 	const btTransform& tr = colObj0->getWorldTransform();
-	btDebugDrawLine(tr*nearest,tr*cp.m_localPointB,red);
-#endif //BT_INTERNAL_EDGE_DEBUG_DRAW
-
-
+	btDebugDrawLine(tr * nearest, tr * cp.m_localPointB, red);
+#endif  //BT_INTERNAL_EDGE_DEBUG_DRAW
 
 	bool isNearEdge = false;
 
@@ -511,334 +482,325 @@ void btAdjustInternalEdgeContacts(btManifoldPoint& cp, const btCollisionObjectWr
 	int numConvexEdgeHits = 0;
 
 	btVector3 localContactNormalOnB = colObj0Wrap->getWorldTransform().getBasis().transpose() * cp.m_normalWorldOnB;
-	localContactNormalOnB.normalize();//is this necessary?
-	
+	localContactNormalOnB.normalize();  //is this necessary?
+
 	// Get closest edge
-	int      bestedge=-1;
-	btScalar    disttobestedge=BT_LARGE_FLOAT;
+	int bestedge = -1;
+	btScalar disttobestedge = BT_LARGE_FLOAT;
 	//
 	// Edge 0 -> 1
-	if (btFabs(info->m_edgeV0V1Angle)< triangleInfoMapPtr->m_maxEdgeAngleThreshold)
-	{	
-	   btVector3 nearest;
-	   btNearestPointInLineSegment( cp.m_localPointB, v0, v1, nearest );
-	   btScalar     len=(contact-nearest).length();
-	   //
-	   if( len < disttobestedge )
-	   {
-	      bestedge=0;
-	      disttobestedge=len;
-      }	      
-   }	   
+	if (btFabs(info->m_edgeV0V1Angle) < triangleInfoMapPtr->m_maxEdgeAngleThreshold)
+	{
+		btVector3 nearest;
+		btNearestPointInLineSegment(cp.m_localPointB, v0, v1, nearest);
+		btScalar len = (contact - nearest).length();
+		//
+		if (len < disttobestedge)
+		{
+			bestedge = 0;
+			disttobestedge = len;
+		}
+	}
 	// Edge 1 -> 2
-	if (btFabs(info->m_edgeV1V2Angle)< triangleInfoMapPtr->m_maxEdgeAngleThreshold)
-	{	
-	   btVector3 nearest;
-	   btNearestPointInLineSegment( cp.m_localPointB, v1, v2, nearest );
-	   btScalar     len=(contact-nearest).length();
-	   //
-	   if( len < disttobestedge )
-	   {
-	      bestedge=1;
-	      disttobestedge=len;
-      }	      
-   }	   
+	if (btFabs(info->m_edgeV1V2Angle) < triangleInfoMapPtr->m_maxEdgeAngleThreshold)
+	{
+		btVector3 nearest;
+		btNearestPointInLineSegment(cp.m_localPointB, v1, v2, nearest);
+		btScalar len = (contact - nearest).length();
+		//
+		if (len < disttobestedge)
+		{
+			bestedge = 1;
+			disttobestedge = len;
+		}
+	}
 	// Edge 2 -> 0
-	if (btFabs(info->m_edgeV2V0Angle)< triangleInfoMapPtr->m_maxEdgeAngleThreshold)
-	{	
-	   btVector3 nearest;
-	   btNearestPointInLineSegment( cp.m_localPointB, v2, v0, nearest );
-	   btScalar     len=(contact-nearest).length();
-	   //
-	   if( len < disttobestedge )
-	   {
-	      bestedge=2;
-	      disttobestedge=len;
-      }	      
-   }   	      	
-	
-#ifdef BT_INTERNAL_EDGE_DEBUG_DRAW
-   btVector3 upfix=tri_normal * btVector3(0.1f,0.1f,0.1f);
-   btDebugDrawLine(tr * v0 + upfix, tr * v1 + upfix, red );
-#endif   
-	if (btFabs(info->m_edgeV0V1Angle)< triangleInfoMapPtr->m_maxEdgeAngleThreshold)
+	if (btFabs(info->m_edgeV2V0Angle) < triangleInfoMapPtr->m_maxEdgeAngleThreshold)
 	{
+		btVector3 nearest;
+		btNearestPointInLineSegment(cp.m_localPointB, v2, v0, nearest);
+		btScalar len = (contact - nearest).length();
+		//
+		if (len < disttobestedge)
+		{
+			bestedge = 2;
+			disttobestedge = len;
+		}
+	}
+
 #ifdef BT_INTERNAL_EDGE_DEBUG_DRAW
-		btDebugDrawLine(tr*contact,tr*(contact+cp.m_normalWorldOnB*10),black);
+	btVector3 upfix = tri_normal * btVector3(0.1f, 0.1f, 0.1f);
+	btDebugDrawLine(tr * v0 + upfix, tr * v1 + upfix, red);
 #endif
-		btScalar len = (contact-nearest).length();
-		if(len<triangleInfoMapPtr->m_edgeDistanceThreshold)
-		if( bestedge==0 )
-		{
-			btVector3 edge(v0-v1);
-			isNearEdge = true;
-
-			if (info->m_edgeV0V1Angle==btScalar(0))
-			{
-				numConcaveEdgeHits++;
-			} else
+	if (btFabs(info->m_edgeV0V1Angle) < triangleInfoMapPtr->m_maxEdgeAngleThreshold)
+	{
+#ifdef BT_INTERNAL_EDGE_DEBUG_DRAW
+		btDebugDrawLine(tr * contact, tr * (contact + cp.m_normalWorldOnB * 10), black);
+#endif
+		btScalar len = (contact - nearest).length();
+		if (len < triangleInfoMapPtr->m_edgeDistanceThreshold)
+			if (bestedge == 0)
 			{
+				btVector3 edge(v0 - v1);
+				isNearEdge = true;
 
-				bool isEdgeConvex = (info->m_flags & TRI_INFO_V0V1_CONVEX);
-				btScalar swapFactor = isEdgeConvex ? btScalar(1) : btScalar(-1);
-	#ifdef BT_INTERNAL_EDGE_DEBUG_DRAW
-				btDebugDrawLine(tr*nearest,tr*(nearest+swapFactor*tri_normal*10),white);
-	#endif //BT_INTERNAL_EDGE_DEBUG_DRAW
-
-				btVector3 nA = swapFactor * tri_normal;
-
-				btQuaternion orn(edge,info->m_edgeV0V1Angle);
-				btVector3 computedNormalB = quatRotate(orn,tri_normal);
-				if (info->m_flags & TRI_INFO_V0V1_SWAP_NORMALB)
-					computedNormalB*=-1;
-				btVector3 nB = swapFactor*computedNormalB;
-
-				btScalar	NdotA = localContactNormalOnB.dot(nA);
-				btScalar	NdotB = localContactNormalOnB.dot(nB);
-				bool backFacingNormal = (NdotA< triangleInfoMapPtr->m_convexEpsilon) && (NdotB<triangleInfoMapPtr->m_convexEpsilon);
-
-#ifdef DEBUG_INTERNAL_EDGE
-				{
-					
-					btDebugDrawLine(cp.getPositionWorldOnB(),cp.getPositionWorldOnB()+tr.getBasis()*(nB*20),red);
-				}
-#endif //DEBUG_INTERNAL_EDGE
-
-
-				if (backFacingNormal)
+				if (info->m_edgeV0V1Angle == btScalar(0))
 				{
 					numConcaveEdgeHits++;
 				}
 				else
 				{
-					numConvexEdgeHits++;
-					btVector3 clampedLocalNormal;
-					bool isClamped = btClampNormal(edge,swapFactor*tri_normal,localContactNormalOnB, info->m_edgeV0V1Angle,clampedLocalNormal);
-					if (isClamped)
+					bool isEdgeConvex = (info->m_flags & TRI_INFO_V0V1_CONVEX);
+					btScalar swapFactor = isEdgeConvex ? btScalar(1) : btScalar(-1);
+#ifdef BT_INTERNAL_EDGE_DEBUG_DRAW
+					btDebugDrawLine(tr * nearest, tr * (nearest + swapFactor * tri_normal * 10), white);
+#endif  //BT_INTERNAL_EDGE_DEBUG_DRAW
+
+					btVector3 nA = swapFactor * tri_normal;
+
+					btQuaternion orn(edge, info->m_edgeV0V1Angle);
+					btVector3 computedNormalB = quatRotate(orn, tri_normal);
+					if (info->m_flags & TRI_INFO_V0V1_SWAP_NORMALB)
+						computedNormalB *= -1;
+					btVector3 nB = swapFactor * computedNormalB;
+
+					btScalar NdotA = localContactNormalOnB.dot(nA);
+					btScalar NdotB = localContactNormalOnB.dot(nB);
+					bool backFacingNormal = (NdotA < triangleInfoMapPtr->m_convexEpsilon) && (NdotB < triangleInfoMapPtr->m_convexEpsilon);
+
+#ifdef DEBUG_INTERNAL_EDGE
 					{
-						if (((normalAdjustFlags & BT_TRIANGLE_CONVEX_DOUBLE_SIDED)!=0) || (clampedLocalNormal.dot(frontFacing*tri_normal)>0))
+						btDebugDrawLine(cp.getPositionWorldOnB(), cp.getPositionWorldOnB() + tr.getBasis() * (nB * 20), red);
+					}
+#endif  //DEBUG_INTERNAL_EDGE
+
+					if (backFacingNormal)
+					{
+						numConcaveEdgeHits++;
+					}
+					else
+					{
+						numConvexEdgeHits++;
+						btVector3 clampedLocalNormal;
+						bool isClamped = btClampNormal(edge, swapFactor * tri_normal, localContactNormalOnB, info->m_edgeV0V1Angle, clampedLocalNormal);
+						if (isClamped)
 						{
-							btVector3 newNormal = colObj0Wrap->getWorldTransform().getBasis() * clampedLocalNormal;
-							//					cp.m_distance1 = cp.m_distance1 * newNormal.dot(cp.m_normalWorldOnB);
-							cp.m_normalWorldOnB = newNormal;
-							// Reproject collision point along normal. (what about cp.m_distance1?)
-							cp.m_positionWorldOnB = cp.m_positionWorldOnA - cp.m_normalWorldOnB * cp.m_distance1;
-							cp.m_localPointB = colObj0Wrap->getWorldTransform().invXform(cp.m_positionWorldOnB);
-							
+							if (((normalAdjustFlags & BT_TRIANGLE_CONVEX_DOUBLE_SIDED) != 0) || (clampedLocalNormal.dot(frontFacing * tri_normal) > 0))
+							{
+								btVector3 newNormal = colObj0Wrap->getWorldTransform().getBasis() * clampedLocalNormal;
+								//					cp.m_distance1 = cp.m_distance1 * newNormal.dot(cp.m_normalWorldOnB);
+								cp.m_normalWorldOnB = newNormal;
+								// Reproject collision point along normal. (what about cp.m_distance1?)
+								cp.m_positionWorldOnB = cp.m_positionWorldOnA - cp.m_normalWorldOnB * cp.m_distance1;
+								cp.m_localPointB = colObj0Wrap->getWorldTransform().invXform(cp.m_positionWorldOnB);
+							}
 						}
 					}
 				}
 			}
-		}
 	}
 
-	btNearestPointInLineSegment(contact,v1,v2,nearest);
+	btNearestPointInLineSegment(contact, v1, v2, nearest);
 #ifdef BT_INTERNAL_EDGE_DEBUG_DRAW
-	btDebugDrawLine(tr*nearest,tr*cp.m_localPointB,green);
-#endif //BT_INTERNAL_EDGE_DEBUG_DRAW
+	btDebugDrawLine(tr * nearest, tr * cp.m_localPointB, green);
+#endif  //BT_INTERNAL_EDGE_DEBUG_DRAW
 
 #ifdef BT_INTERNAL_EDGE_DEBUG_DRAW
-   btDebugDrawLine(tr * v1 + upfix, tr * v2 + upfix , green );
-#endif   
+	btDebugDrawLine(tr * v1 + upfix, tr * v2 + upfix, green);
+#endif
 
-	if (btFabs(info->m_edgeV1V2Angle)< triangleInfoMapPtr->m_maxEdgeAngleThreshold)
+	if (btFabs(info->m_edgeV1V2Angle) < triangleInfoMapPtr->m_maxEdgeAngleThreshold)
 	{
 #ifdef BT_INTERNAL_EDGE_DEBUG_DRAW
-		btDebugDrawLine(tr*contact,tr*(contact+cp.m_normalWorldOnB*10),black);
-#endif //BT_INTERNAL_EDGE_DEBUG_DRAW
+		btDebugDrawLine(tr * contact, tr * (contact + cp.m_normalWorldOnB * 10), black);
+#endif  //BT_INTERNAL_EDGE_DEBUG_DRAW
 
-
-
-		btScalar len = (contact-nearest).length();
-		if(len<triangleInfoMapPtr->m_edgeDistanceThreshold)
-		if( bestedge==1 )
-		{
-			isNearEdge = true;
+		btScalar len = (contact - nearest).length();
+		if (len < triangleInfoMapPtr->m_edgeDistanceThreshold)
+			if (bestedge == 1)
+			{
+				isNearEdge = true;
 #ifdef BT_INTERNAL_EDGE_DEBUG_DRAW
-			btDebugDrawLine(tr*nearest,tr*(nearest+tri_normal*10),white);
-#endif //BT_INTERNAL_EDGE_DEBUG_DRAW
+				btDebugDrawLine(tr * nearest, tr * (nearest + tri_normal * 10), white);
+#endif  //BT_INTERNAL_EDGE_DEBUG_DRAW
 
-			btVector3 edge(v1-v2);
+				btVector3 edge(v1 - v2);
 
-			isNearEdge = true;
+				isNearEdge = true;
 
-			if (info->m_edgeV1V2Angle == btScalar(0))
-			{
-				numConcaveEdgeHits++;
-			} else
-			{
-				bool isEdgeConvex = (info->m_flags & TRI_INFO_V1V2_CONVEX)!=0;
-				btScalar swapFactor = isEdgeConvex ? btScalar(1) : btScalar(-1);
-	#ifdef BT_INTERNAL_EDGE_DEBUG_DRAW
-				btDebugDrawLine(tr*nearest,tr*(nearest+swapFactor*tri_normal*10),white);
-	#endif //BT_INTERNAL_EDGE_DEBUG_DRAW
-
-				btVector3 nA = swapFactor * tri_normal;
-				
-				btQuaternion orn(edge,info->m_edgeV1V2Angle);
-				btVector3 computedNormalB = quatRotate(orn,tri_normal);
-				if (info->m_flags & TRI_INFO_V1V2_SWAP_NORMALB)
-					computedNormalB*=-1;
-				btVector3 nB = swapFactor*computedNormalB;
-
-#ifdef DEBUG_INTERNAL_EDGE
-				{
-					btDebugDrawLine(cp.getPositionWorldOnB(),cp.getPositionWorldOnB()+tr.getBasis()*(nB*20),red);
-				}
-#endif //DEBUG_INTERNAL_EDGE
-
-
-				btScalar	NdotA = localContactNormalOnB.dot(nA);
-				btScalar	NdotB = localContactNormalOnB.dot(nB);
-				bool backFacingNormal = (NdotA< triangleInfoMapPtr->m_convexEpsilon) && (NdotB<triangleInfoMapPtr->m_convexEpsilon);
-
-				if (backFacingNormal)
+				if (info->m_edgeV1V2Angle == btScalar(0))
 				{
 					numConcaveEdgeHits++;
 				}
 				else
 				{
-					numConvexEdgeHits++;
-					btVector3 localContactNormalOnB = colObj0Wrap->getWorldTransform().getBasis().transpose() * cp.m_normalWorldOnB;
-					btVector3 clampedLocalNormal;
-					bool isClamped = btClampNormal(edge,swapFactor*tri_normal,localContactNormalOnB, info->m_edgeV1V2Angle,clampedLocalNormal);
-					if (isClamped)
+					bool isEdgeConvex = (info->m_flags & TRI_INFO_V1V2_CONVEX) != 0;
+					btScalar swapFactor = isEdgeConvex ? btScalar(1) : btScalar(-1);
+#ifdef BT_INTERNAL_EDGE_DEBUG_DRAW
+					btDebugDrawLine(tr * nearest, tr * (nearest + swapFactor * tri_normal * 10), white);
+#endif  //BT_INTERNAL_EDGE_DEBUG_DRAW
+
+					btVector3 nA = swapFactor * tri_normal;
+
+					btQuaternion orn(edge, info->m_edgeV1V2Angle);
+					btVector3 computedNormalB = quatRotate(orn, tri_normal);
+					if (info->m_flags & TRI_INFO_V1V2_SWAP_NORMALB)
+						computedNormalB *= -1;
+					btVector3 nB = swapFactor * computedNormalB;
+
+#ifdef DEBUG_INTERNAL_EDGE
 					{
-						if (((normalAdjustFlags & BT_TRIANGLE_CONVEX_DOUBLE_SIDED)!=0) || (clampedLocalNormal.dot(frontFacing*tri_normal)>0))
+						btDebugDrawLine(cp.getPositionWorldOnB(), cp.getPositionWorldOnB() + tr.getBasis() * (nB * 20), red);
+					}
+#endif  //DEBUG_INTERNAL_EDGE
+
+					btScalar NdotA = localContactNormalOnB.dot(nA);
+					btScalar NdotB = localContactNormalOnB.dot(nB);
+					bool backFacingNormal = (NdotA < triangleInfoMapPtr->m_convexEpsilon) && (NdotB < triangleInfoMapPtr->m_convexEpsilon);
+
+					if (backFacingNormal)
+					{
+						numConcaveEdgeHits++;
+					}
+					else
+					{
+						numConvexEdgeHits++;
+						btVector3 localContactNormalOnB = colObj0Wrap->getWorldTransform().getBasis().transpose() * cp.m_normalWorldOnB;
+						btVector3 clampedLocalNormal;
+						bool isClamped = btClampNormal(edge, swapFactor * tri_normal, localContactNormalOnB, info->m_edgeV1V2Angle, clampedLocalNormal);
+						if (isClamped)
 						{
-							btVector3 newNormal = colObj0Wrap->getWorldTransform().getBasis() * clampedLocalNormal;
-							//					cp.m_distance1 = cp.m_distance1 * newNormal.dot(cp.m_normalWorldOnB);
-							cp.m_normalWorldOnB = newNormal;
-							// Reproject collision point along normal.
-							cp.m_positionWorldOnB = cp.m_positionWorldOnA - cp.m_normalWorldOnB * cp.m_distance1;
-							cp.m_localPointB = colObj0Wrap->getWorldTransform().invXform(cp.m_positionWorldOnB);
+							if (((normalAdjustFlags & BT_TRIANGLE_CONVEX_DOUBLE_SIDED) != 0) || (clampedLocalNormal.dot(frontFacing * tri_normal) > 0))
+							{
+								btVector3 newNormal = colObj0Wrap->getWorldTransform().getBasis() * clampedLocalNormal;
+								//					cp.m_distance1 = cp.m_distance1 * newNormal.dot(cp.m_normalWorldOnB);
+								cp.m_normalWorldOnB = newNormal;
+								// Reproject collision point along normal.
+								cp.m_positionWorldOnB = cp.m_positionWorldOnA - cp.m_normalWorldOnB * cp.m_distance1;
+								cp.m_localPointB = colObj0Wrap->getWorldTransform().invXform(cp.m_positionWorldOnB);
+							}
 						}
 					}
 				}
 			}
-		}
 	}
 
-	btNearestPointInLineSegment(contact,v2,v0,nearest);
+	btNearestPointInLineSegment(contact, v2, v0, nearest);
 #ifdef BT_INTERNAL_EDGE_DEBUG_DRAW
-	btDebugDrawLine(tr*nearest,tr*cp.m_localPointB,blue);
-#endif //BT_INTERNAL_EDGE_DEBUG_DRAW
+	btDebugDrawLine(tr * nearest, tr * cp.m_localPointB, blue);
+#endif  //BT_INTERNAL_EDGE_DEBUG_DRAW
 #ifdef BT_INTERNAL_EDGE_DEBUG_DRAW
-   btDebugDrawLine(tr * v2 + upfix, tr * v0 + upfix , blue );
-#endif   
+	btDebugDrawLine(tr * v2 + upfix, tr * v0 + upfix, blue);
+#endif
 
-	if (btFabs(info->m_edgeV2V0Angle)< triangleInfoMapPtr->m_maxEdgeAngleThreshold)
+	if (btFabs(info->m_edgeV2V0Angle) < triangleInfoMapPtr->m_maxEdgeAngleThreshold)
 	{
-
 #ifdef BT_INTERNAL_EDGE_DEBUG_DRAW
-		btDebugDrawLine(tr*contact,tr*(contact+cp.m_normalWorldOnB*10),black);
-#endif //BT_INTERNAL_EDGE_DEBUG_DRAW
+		btDebugDrawLine(tr * contact, tr * (contact + cp.m_normalWorldOnB * 10), black);
+#endif  //BT_INTERNAL_EDGE_DEBUG_DRAW
 
-		btScalar len = (contact-nearest).length();
-		if(len<triangleInfoMapPtr->m_edgeDistanceThreshold)
-		if( bestedge==2 )
-		{
-			isNearEdge = true;
+		btScalar len = (contact - nearest).length();
+		if (len < triangleInfoMapPtr->m_edgeDistanceThreshold)
+			if (bestedge == 2)
+			{
+				isNearEdge = true;
 #ifdef BT_INTERNAL_EDGE_DEBUG_DRAW
-			btDebugDrawLine(tr*nearest,tr*(nearest+tri_normal*10),white);
-#endif //BT_INTERNAL_EDGE_DEBUG_DRAW
+				btDebugDrawLine(tr * nearest, tr * (nearest + tri_normal * 10), white);
+#endif  //BT_INTERNAL_EDGE_DEBUG_DRAW
 
-			btVector3 edge(v2-v0);
+				btVector3 edge(v2 - v0);
 
-			if (info->m_edgeV2V0Angle==btScalar(0))
-			{
-				numConcaveEdgeHits++;
-			} else
-			{
-
-				bool isEdgeConvex = (info->m_flags & TRI_INFO_V2V0_CONVEX)!=0;
-				btScalar swapFactor = isEdgeConvex ? btScalar(1) : btScalar(-1);
-	#ifdef BT_INTERNAL_EDGE_DEBUG_DRAW
-				btDebugDrawLine(tr*nearest,tr*(nearest+swapFactor*tri_normal*10),white);
-	#endif //BT_INTERNAL_EDGE_DEBUG_DRAW
-
-				btVector3 nA = swapFactor * tri_normal;
-				btQuaternion orn(edge,info->m_edgeV2V0Angle);
-				btVector3 computedNormalB = quatRotate(orn,tri_normal);
-				if (info->m_flags & TRI_INFO_V2V0_SWAP_NORMALB)
-					computedNormalB*=-1;
-				btVector3 nB = swapFactor*computedNormalB;
-
-#ifdef DEBUG_INTERNAL_EDGE
-				{
-					btDebugDrawLine(cp.getPositionWorldOnB(),cp.getPositionWorldOnB()+tr.getBasis()*(nB*20),red);
-				}
-#endif //DEBUG_INTERNAL_EDGE
-
-				btScalar	NdotA = localContactNormalOnB.dot(nA);
-				btScalar	NdotB = localContactNormalOnB.dot(nB);
-				bool backFacingNormal = (NdotA< triangleInfoMapPtr->m_convexEpsilon) && (NdotB<triangleInfoMapPtr->m_convexEpsilon);
-
-				if (backFacingNormal)
+				if (info->m_edgeV2V0Angle == btScalar(0))
 				{
 					numConcaveEdgeHits++;
 				}
 				else
 				{
-					numConvexEdgeHits++;
-					//				printf("hitting convex edge\n");
+					bool isEdgeConvex = (info->m_flags & TRI_INFO_V2V0_CONVEX) != 0;
+					btScalar swapFactor = isEdgeConvex ? btScalar(1) : btScalar(-1);
+#ifdef BT_INTERNAL_EDGE_DEBUG_DRAW
+					btDebugDrawLine(tr * nearest, tr * (nearest + swapFactor * tri_normal * 10), white);
+#endif  //BT_INTERNAL_EDGE_DEBUG_DRAW
 
+					btVector3 nA = swapFactor * tri_normal;
+					btQuaternion orn(edge, info->m_edgeV2V0Angle);
+					btVector3 computedNormalB = quatRotate(orn, tri_normal);
+					if (info->m_flags & TRI_INFO_V2V0_SWAP_NORMALB)
+						computedNormalB *= -1;
+					btVector3 nB = swapFactor * computedNormalB;
 
-					btVector3 localContactNormalOnB = colObj0Wrap->getWorldTransform().getBasis().transpose() * cp.m_normalWorldOnB;
-					btVector3 clampedLocalNormal;
-					bool isClamped = btClampNormal(edge,swapFactor*tri_normal,localContactNormalOnB,info->m_edgeV2V0Angle,clampedLocalNormal);
-					if (isClamped)
+#ifdef DEBUG_INTERNAL_EDGE
 					{
-						if (((normalAdjustFlags & BT_TRIANGLE_CONVEX_DOUBLE_SIDED)!=0) || (clampedLocalNormal.dot(frontFacing*tri_normal)>0))
+						btDebugDrawLine(cp.getPositionWorldOnB(), cp.getPositionWorldOnB() + tr.getBasis() * (nB * 20), red);
+					}
+#endif  //DEBUG_INTERNAL_EDGE
+
+					btScalar NdotA = localContactNormalOnB.dot(nA);
+					btScalar NdotB = localContactNormalOnB.dot(nB);
+					bool backFacingNormal = (NdotA < triangleInfoMapPtr->m_convexEpsilon) && (NdotB < triangleInfoMapPtr->m_convexEpsilon);
+
+					if (backFacingNormal)
+					{
+						numConcaveEdgeHits++;
+					}
+					else
+					{
+						numConvexEdgeHits++;
+						//				printf("hitting convex edge\n");
+
+						btVector3 localContactNormalOnB = colObj0Wrap->getWorldTransform().getBasis().transpose() * cp.m_normalWorldOnB;
+						btVector3 clampedLocalNormal;
+						bool isClamped = btClampNormal(edge, swapFactor * tri_normal, localContactNormalOnB, info->m_edgeV2V0Angle, clampedLocalNormal);
+						if (isClamped)
 						{
-							btVector3 newNormal = colObj0Wrap->getWorldTransform().getBasis() * clampedLocalNormal;
-							//					cp.m_distance1 = cp.m_distance1 * newNormal.dot(cp.m_normalWorldOnB);
-							cp.m_normalWorldOnB = newNormal;
-							// Reproject collision point along normal.
-							cp.m_positionWorldOnB = cp.m_positionWorldOnA - cp.m_normalWorldOnB * cp.m_distance1;
-							cp.m_localPointB = colObj0Wrap->getWorldTransform().invXform(cp.m_positionWorldOnB);
+							if (((normalAdjustFlags & BT_TRIANGLE_CONVEX_DOUBLE_SIDED) != 0) || (clampedLocalNormal.dot(frontFacing * tri_normal) > 0))
+							{
+								btVector3 newNormal = colObj0Wrap->getWorldTransform().getBasis() * clampedLocalNormal;
+								//					cp.m_distance1 = cp.m_distance1 * newNormal.dot(cp.m_normalWorldOnB);
+								cp.m_normalWorldOnB = newNormal;
+								// Reproject collision point along normal.
+								cp.m_positionWorldOnB = cp.m_positionWorldOnA - cp.m_normalWorldOnB * cp.m_distance1;
+								cp.m_localPointB = colObj0Wrap->getWorldTransform().invXform(cp.m_positionWorldOnB);
+							}
 						}
 					}
-				} 
+				}
 			}
-			
-
-		}
 	}
 
 #ifdef DEBUG_INTERNAL_EDGE
 	{
-		btVector3 color(0,1,1);
-		btDebugDrawLine(cp.getPositionWorldOnB(),cp.getPositionWorldOnB()+cp.m_normalWorldOnB*10,color);
+		btVector3 color(0, 1, 1);
+		btDebugDrawLine(cp.getPositionWorldOnB(), cp.getPositionWorldOnB() + cp.m_normalWorldOnB * 10, color);
 	}
-#endif //DEBUG_INTERNAL_EDGE
+#endif  //DEBUG_INTERNAL_EDGE
 
 	if (isNearEdge)
 	{
-
-		if (numConcaveEdgeHits>0)
+		if (numConcaveEdgeHits > 0)
 		{
-			if ((normalAdjustFlags & BT_TRIANGLE_CONCAVE_DOUBLE_SIDED)!=0)
+			if ((normalAdjustFlags & BT_TRIANGLE_CONCAVE_DOUBLE_SIDED) != 0)
 			{
 				//fix tri_normal so it pointing the same direction as the current local contact normal
 				if (tri_normal.dot(localContactNormalOnB) < 0)
 				{
 					tri_normal *= -1;
 				}
-				cp.m_normalWorldOnB = colObj0Wrap->getWorldTransform().getBasis()*tri_normal;
-			} else
+				cp.m_normalWorldOnB = colObj0Wrap->getWorldTransform().getBasis() * tri_normal;
+			}
+			else
 			{
-				btVector3 newNormal = tri_normal *frontFacing;
+				btVector3 newNormal = tri_normal * frontFacing;
 				//if the tri_normal is pointing opposite direction as the current local contact normal, skip it
-				btScalar d = newNormal.dot(localContactNormalOnB) ;
-				if (d< 0)
+				btScalar d = newNormal.dot(localContactNormalOnB);
+				if (d < 0)
 				{
 					return;
 				}
 				//modify the normal to be the triangle normal (or backfacing normal)
-				cp.m_normalWorldOnB = colObj0Wrap->getWorldTransform().getBasis() *newNormal;
+				cp.m_normalWorldOnB = colObj0Wrap->getWorldTransform().getBasis() * newNormal;
 			}
-						
+
 			// Reproject collision point along normal.
 			cp.m_positionWorldOnB = cp.m_positionWorldOnA - cp.m_normalWorldOnB * cp.m_distance1;
 			cp.m_localPointB = colObj0Wrap->getWorldTransform().invXform(cp.m_positionWorldOnB);
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btInternalEdgeUtility.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btInternalEdgeUtility.h
index 7d9aafeee69..9d9cff040fa 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btInternalEdgeUtility.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btInternalEdgeUtility.h
@@ -16,32 +16,26 @@ struct btCollisionObjectWrapper;
 class btManifoldPoint;
 class btIDebugDraw;
 
-
-
 enum btInternalEdgeAdjustFlags
 {
 	BT_TRIANGLE_CONVEX_BACKFACE_MODE = 1,
-	BT_TRIANGLE_CONCAVE_DOUBLE_SIDED = 2, //double sided options are experimental, single sided is recommended
+	BT_TRIANGLE_CONCAVE_DOUBLE_SIDED = 2,  //double sided options are experimental, single sided is recommended
 	BT_TRIANGLE_CONVEX_DOUBLE_SIDED = 4
 };
 
-
 ///Call btGenerateInternalEdgeInfo to create triangle info, store in the shape 'userInfo'
-void	btGenerateInternalEdgeInfo (btBvhTriangleMeshShape*trimeshShape, btTriangleInfoMap* triangleInfoMap);
-
+void btGenerateInternalEdgeInfo(btBvhTriangleMeshShape* trimeshShape, btTriangleInfoMap* triangleInfoMap);
 
 ///Call the btFixMeshNormal to adjust the collision normal, using the triangle info map (generated using btGenerateInternalEdgeInfo)
 ///If this info map is missing, or the triangle is not store in this map, nothing will be done
-void	btAdjustInternalEdgeContacts(btManifoldPoint& cp, const btCollisionObjectWrapper* trimeshColObj0Wrap,const btCollisionObjectWrapper* otherColObj1Wrap, int partId0, int index0, int normalAdjustFlags = 0);
+void btAdjustInternalEdgeContacts(btManifoldPoint& cp, const btCollisionObjectWrapper* trimeshColObj0Wrap, const btCollisionObjectWrapper* otherColObj1Wrap, int partId0, int index0, int normalAdjustFlags = 0);
 
 ///Enable the BT_INTERNAL_EDGE_DEBUG_DRAW define and call btSetDebugDrawer, to get visual info to see if the internal edge utility works properly.
 ///If the utility doesn't work properly, you might have to adjust the threshold values in btTriangleInfoMap
 //#define BT_INTERNAL_EDGE_DEBUG_DRAW
 
 #ifdef BT_INTERNAL_EDGE_DEBUG_DRAW
-void	btSetDebugDrawer(btIDebugDraw* debugDrawer);
-#endif //BT_INTERNAL_EDGE_DEBUG_DRAW
-
-
-#endif //BT_INTERNAL_EDGE_UTILITY_H
+void btSetDebugDrawer(btIDebugDraw* debugDrawer);
+#endif  //BT_INTERNAL_EDGE_DEBUG_DRAW
 
+#endif  //BT_INTERNAL_EDGE_UTILITY_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btManifoldResult.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btManifoldResult.cpp
index 23c73c88254..770eb243691 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btManifoldResult.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btManifoldResult.cpp
@@ -13,106 +13,102 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btManifoldResult.h"
 #include "BulletCollision/NarrowPhaseCollision/btPersistentManifold.h"
 #include "BulletCollision/CollisionDispatch/btCollisionObject.h"
 #include "BulletCollision/CollisionDispatch/btCollisionObjectWrapper.h"
 
 ///This is to allow MaterialCombiner/Custom Friction/Restitution values
-ContactAddedCallback		gContactAddedCallback=0;
+ContactAddedCallback gContactAddedCallback = 0;
 
-CalculateCombinedCallback		gCalculateCombinedRestitutionCallback = &btManifoldResult::calculateCombinedRestitution;
-CalculateCombinedCallback		gCalculateCombinedFrictionCallback = &btManifoldResult::calculateCombinedFriction;
-CalculateCombinedCallback		gCalculateCombinedRollingFrictionCallback = &btManifoldResult::calculateCombinedRollingFriction;
-CalculateCombinedCallback		gCalculateCombinedSpinningFrictionCallback = &btManifoldResult::calculateCombinedSpinningFriction;
-CalculateCombinedCallback		gCalculateCombinedContactDampingCallback = &btManifoldResult::calculateCombinedContactDamping;
-CalculateCombinedCallback		gCalculateCombinedContactStiffnessCallback = &btManifoldResult::calculateCombinedContactStiffness;
+CalculateCombinedCallback gCalculateCombinedRestitutionCallback = &btManifoldResult::calculateCombinedRestitution;
+CalculateCombinedCallback gCalculateCombinedFrictionCallback = &btManifoldResult::calculateCombinedFriction;
+CalculateCombinedCallback gCalculateCombinedRollingFrictionCallback = &btManifoldResult::calculateCombinedRollingFriction;
+CalculateCombinedCallback gCalculateCombinedSpinningFrictionCallback = &btManifoldResult::calculateCombinedSpinningFriction;
+CalculateCombinedCallback gCalculateCombinedContactDampingCallback = &btManifoldResult::calculateCombinedContactDamping;
+CalculateCombinedCallback gCalculateCombinedContactStiffnessCallback = &btManifoldResult::calculateCombinedContactStiffness;
 
-btScalar	btManifoldResult::calculateCombinedRollingFriction(const btCollisionObject* body0,const btCollisionObject* body1)
+btScalar btManifoldResult::calculateCombinedRollingFriction(const btCollisionObject* body0, const btCollisionObject* body1)
 {
 	btScalar friction = body0->getRollingFriction() * body1->getFriction() + body1->getRollingFriction() * body0->getFriction();
 
-	const btScalar MAX_FRICTION  = btScalar(10.);
+	const btScalar MAX_FRICTION = btScalar(10.);
 	if (friction < -MAX_FRICTION)
 		friction = -MAX_FRICTION;
 	if (friction > MAX_FRICTION)
 		friction = MAX_FRICTION;
 	return friction;
-
 }
 
-btScalar	btManifoldResult::calculateCombinedSpinningFriction(const btCollisionObject* body0,const btCollisionObject* body1)
+btScalar btManifoldResult::calculateCombinedSpinningFriction(const btCollisionObject* body0, const btCollisionObject* body1)
 {
-    btScalar friction = body0->getSpinningFriction() * body1->getFriction() + body1->getSpinningFriction() * body0->getFriction();
-    
-    const btScalar MAX_FRICTION  = btScalar(10.);
-    if (friction < -MAX_FRICTION)
-        friction = -MAX_FRICTION;
-    if (friction > MAX_FRICTION)
-        friction = MAX_FRICTION;
-    return friction;
+	btScalar friction = body0->getSpinningFriction() * body1->getFriction() + body1->getSpinningFriction() * body0->getFriction();
+
+	const btScalar MAX_FRICTION = btScalar(10.);
+	if (friction < -MAX_FRICTION)
+		friction = -MAX_FRICTION;
+	if (friction > MAX_FRICTION)
+		friction = MAX_FRICTION;
+	return friction;
 }
 
 ///User can override this material combiner by implementing gContactAddedCallback and setting body0->m_collisionFlags |= btCollisionObject::customMaterialCallback;
-btScalar	btManifoldResult::calculateCombinedFriction(const btCollisionObject* body0,const btCollisionObject* body1)
+btScalar btManifoldResult::calculateCombinedFriction(const btCollisionObject* body0, const btCollisionObject* body1)
 {
 	btScalar friction = body0->getFriction() * body1->getFriction();
 
-	const btScalar MAX_FRICTION  = btScalar(10.);
+	const btScalar MAX_FRICTION = btScalar(10.);
 	if (friction < -MAX_FRICTION)
 		friction = -MAX_FRICTION;
 	if (friction > MAX_FRICTION)
 		friction = MAX_FRICTION;
 	return friction;
-
 }
 
-btScalar	btManifoldResult::calculateCombinedRestitution(const btCollisionObject* body0,const btCollisionObject* body1)
+btScalar btManifoldResult::calculateCombinedRestitution(const btCollisionObject* body0, const btCollisionObject* body1)
 {
 	return body0->getRestitution() * body1->getRestitution();
 }
 
-btScalar	btManifoldResult::calculateCombinedContactDamping(const btCollisionObject* body0,const btCollisionObject* body1)
+btScalar btManifoldResult::calculateCombinedContactDamping(const btCollisionObject* body0, const btCollisionObject* body1)
 {
-    return body0->getContactDamping() + body1->getContactDamping();
+	return body0->getContactDamping() + body1->getContactDamping();
 }
 
-btScalar	btManifoldResult::calculateCombinedContactStiffness(const btCollisionObject* body0,const btCollisionObject* body1)
+btScalar btManifoldResult::calculateCombinedContactStiffness(const btCollisionObject* body0, const btCollisionObject* body1)
 {
-    
-    btScalar s0 = body0->getContactStiffness();
-    btScalar s1 = body1->getContactStiffness();
-    
-    btScalar tmp0 = btScalar(1)/s0;
-    btScalar tmp1 = btScalar(1)/s1;
-    btScalar combinedStiffness = btScalar(1) / (tmp0+tmp1);
-    return combinedStiffness;
+	btScalar s0 = body0->getContactStiffness();
+	btScalar s1 = body1->getContactStiffness();
+
+	btScalar tmp0 = btScalar(1) / s0;
+	btScalar tmp1 = btScalar(1) / s1;
+	btScalar combinedStiffness = btScalar(1) / (tmp0 + tmp1);
+	return combinedStiffness;
 }
 
-
-btManifoldResult::btManifoldResult(const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
-		:m_manifoldPtr(0),
-		m_body0Wrap(body0Wrap),
-		m_body1Wrap(body1Wrap)
+btManifoldResult::btManifoldResult(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
+	: m_manifoldPtr(0),
+	  m_body0Wrap(body0Wrap),
+	  m_body1Wrap(body1Wrap)
 #ifdef DEBUG_PART_INDEX
-		,m_partId0(-1),
-	m_partId1(-1),
-	m_index0(-1),
-	m_index1(-1)
-#endif //DEBUG_PART_INDEX
-	, m_closestPointDistanceThreshold(0)
+	  ,
+	  m_partId0(-1),
+	  m_partId1(-1),
+	  m_index0(-1),
+	  m_index1(-1)
+#endif  //DEBUG_PART_INDEX
+	  ,
+	  m_closestPointDistanceThreshold(0)
 {
 }
 
-
-void btManifoldResult::addContactPoint(const btVector3& normalOnBInWorld,const btVector3& pointInWorld,btScalar depth)
+void btManifoldResult::addContactPoint(const btVector3& normalOnBInWorld, const btVector3& pointInWorld, btScalar depth)
 {
 	btAssert(m_manifoldPtr);
 	//order in manifold needs to match
 
 	if (depth > m_manifoldPtr->getContactBreakingThreshold())
-//	if (depth > m_manifoldPtr->getContactProcessingThreshold())
+		//	if (depth > m_manifoldPtr->getContactProcessingThreshold())
 		return;
 
 	bool isSwapped = m_manifoldPtr->getBody0() != m_body0Wrap->getCollisionObject();
@@ -122,81 +118,82 @@ void btManifoldResult::addContactPoint(const btVector3& normalOnBInWorld,const b
 
 	btVector3 localA;
 	btVector3 localB;
-	
+
 	if (isSwapped)
 	{
-		localA = m_body1Wrap->getCollisionObject()->getWorldTransform().invXform(pointA );
+		localA = m_body1Wrap->getCollisionObject()->getWorldTransform().invXform(pointA);
 		localB = m_body0Wrap->getCollisionObject()->getWorldTransform().invXform(pointInWorld);
-	} else
+	}
+	else
 	{
-		localA = m_body0Wrap->getCollisionObject()->getWorldTransform().invXform(pointA );
+		localA = m_body0Wrap->getCollisionObject()->getWorldTransform().invXform(pointA);
 		localB = m_body1Wrap->getCollisionObject()->getWorldTransform().invXform(pointInWorld);
 	}
 
-	btManifoldPoint newPt(localA,localB,normalOnBInWorld,depth);
+	btManifoldPoint newPt(localA, localB, normalOnBInWorld, depth);
 	newPt.m_positionWorldOnA = pointA;
 	newPt.m_positionWorldOnB = pointInWorld;
-	
+
 	int insertIndex = m_manifoldPtr->getCacheEntry(newPt);
 
-	newPt.m_combinedFriction = gCalculateCombinedFrictionCallback(m_body0Wrap->getCollisionObject(),m_body1Wrap->getCollisionObject());
-	newPt.m_combinedRestitution = gCalculateCombinedRestitutionCallback(m_body0Wrap->getCollisionObject(),m_body1Wrap->getCollisionObject());
-	newPt.m_combinedRollingFriction = gCalculateCombinedRollingFrictionCallback(m_body0Wrap->getCollisionObject(),m_body1Wrap->getCollisionObject());
-	newPt.m_combinedSpinningFriction = gCalculateCombinedSpinningFrictionCallback(m_body0Wrap->getCollisionObject(),m_body1Wrap->getCollisionObject());
-	
-	if (    (m_body0Wrap->getCollisionObject()->getCollisionFlags()& btCollisionObject::CF_HAS_CONTACT_STIFFNESS_DAMPING) ||
-            (m_body1Wrap->getCollisionObject()->getCollisionFlags()& btCollisionObject::CF_HAS_CONTACT_STIFFNESS_DAMPING))
-    {
-		newPt.m_combinedContactDamping1 = gCalculateCombinedContactDampingCallback(m_body0Wrap->getCollisionObject(),m_body1Wrap->getCollisionObject());
-		newPt.m_combinedContactStiffness1 = gCalculateCombinedContactStiffnessCallback(m_body0Wrap->getCollisionObject(),m_body1Wrap->getCollisionObject());
-        newPt.m_contactPointFlags |= BT_CONTACT_FLAG_CONTACT_STIFFNESS_DAMPING;
-    }
+	newPt.m_combinedFriction = gCalculateCombinedFrictionCallback(m_body0Wrap->getCollisionObject(), m_body1Wrap->getCollisionObject());
+	newPt.m_combinedRestitution = gCalculateCombinedRestitutionCallback(m_body0Wrap->getCollisionObject(), m_body1Wrap->getCollisionObject());
+	newPt.m_combinedRollingFriction = gCalculateCombinedRollingFrictionCallback(m_body0Wrap->getCollisionObject(), m_body1Wrap->getCollisionObject());
+	newPt.m_combinedSpinningFriction = gCalculateCombinedSpinningFrictionCallback(m_body0Wrap->getCollisionObject(), m_body1Wrap->getCollisionObject());
 
-	if (    (m_body0Wrap->getCollisionObject()->getCollisionFlags()& btCollisionObject::CF_HAS_FRICTION_ANCHOR) ||
-            (m_body1Wrap->getCollisionObject()->getCollisionFlags()& btCollisionObject::CF_HAS_FRICTION_ANCHOR))
-    {
-        newPt.m_contactPointFlags |= BT_CONTACT_FLAG_FRICTION_ANCHOR;
-    }
+	if ((m_body0Wrap->getCollisionObject()->getCollisionFlags() & btCollisionObject::CF_HAS_CONTACT_STIFFNESS_DAMPING) ||
+		(m_body1Wrap->getCollisionObject()->getCollisionFlags() & btCollisionObject::CF_HAS_CONTACT_STIFFNESS_DAMPING))
+	{
+		newPt.m_combinedContactDamping1 = gCalculateCombinedContactDampingCallback(m_body0Wrap->getCollisionObject(), m_body1Wrap->getCollisionObject());
+		newPt.m_combinedContactStiffness1 = gCalculateCombinedContactStiffnessCallback(m_body0Wrap->getCollisionObject(), m_body1Wrap->getCollisionObject());
+		newPt.m_contactPointFlags |= BT_CONTACT_FLAG_CONTACT_STIFFNESS_DAMPING;
+	}
 
-	btPlaneSpace1(newPt.m_normalWorldOnB,newPt.m_lateralFrictionDir1,newPt.m_lateralFrictionDir2);
-	
+	if ((m_body0Wrap->getCollisionObject()->getCollisionFlags() & btCollisionObject::CF_HAS_FRICTION_ANCHOR) ||
+		(m_body1Wrap->getCollisionObject()->getCollisionFlags() & btCollisionObject::CF_HAS_FRICTION_ANCHOR))
+	{
+		newPt.m_contactPointFlags |= BT_CONTACT_FLAG_FRICTION_ANCHOR;
+	}
 
-	
-   //BP mod, store contact triangles.
+	btPlaneSpace1(newPt.m_normalWorldOnB, newPt.m_lateralFrictionDir1, newPt.m_lateralFrictionDir2);
+
+	//BP mod, store contact triangles.
 	if (isSwapped)
 	{
 		newPt.m_partId0 = m_partId1;
 		newPt.m_partId1 = m_partId0;
-		newPt.m_index0  = m_index1;
-		newPt.m_index1  = m_index0;
-	} else
+		newPt.m_index0 = m_index1;
+		newPt.m_index1 = m_index0;
+	}
+	else
 	{
 		newPt.m_partId0 = m_partId0;
 		newPt.m_partId1 = m_partId1;
-		newPt.m_index0  = m_index0;
-		newPt.m_index1  = m_index1;
+		newPt.m_index0 = m_index0;
+		newPt.m_index1 = m_index1;
 	}
 	//printf("depth=%f\n",depth);
 	///@todo, check this for any side effects
 	if (insertIndex >= 0)
 	{
 		//const btManifoldPoint& oldPoint = m_manifoldPtr->getContactPoint(insertIndex);
-		m_manifoldPtr->replaceContactPoint(newPt,insertIndex);
-	} else
+		m_manifoldPtr->replaceContactPoint(newPt, insertIndex);
+	}
+	else
 	{
 		insertIndex = m_manifoldPtr->addManifoldPoint(newPt);
 	}
-	
+
 	//User can override friction and/or restitution
 	if (gContactAddedCallback &&
 		//and if either of the two bodies requires custom material
-		 ((m_body0Wrap->getCollisionObject()->getCollisionFlags() & btCollisionObject::CF_CUSTOM_MATERIAL_CALLBACK) ||
-		   (m_body1Wrap->getCollisionObject()->getCollisionFlags() & btCollisionObject::CF_CUSTOM_MATERIAL_CALLBACK)))
+		((m_body0Wrap->getCollisionObject()->getCollisionFlags() & btCollisionObject::CF_CUSTOM_MATERIAL_CALLBACK) ||
+		 (m_body1Wrap->getCollisionObject()->getCollisionFlags() & btCollisionObject::CF_CUSTOM_MATERIAL_CALLBACK)))
 	{
 		//experimental feature info, for per-triangle material etc.
-		const btCollisionObjectWrapper* obj0Wrap = isSwapped? m_body1Wrap : m_body0Wrap;
-		const btCollisionObjectWrapper* obj1Wrap = isSwapped? m_body0Wrap : m_body1Wrap;
-		(*gContactAddedCallback)(m_manifoldPtr->getContactPoint(insertIndex),obj0Wrap,newPt.m_partId0,newPt.m_index0,obj1Wrap,newPt.m_partId1,newPt.m_index1);
+		const btCollisionObjectWrapper* obj0Wrap = isSwapped ? m_body1Wrap : m_body0Wrap;
+		const btCollisionObjectWrapper* obj1Wrap = isSwapped ? m_body0Wrap : m_body1Wrap;
+		(*gContactAddedCallback)(m_manifoldPtr->getContactPoint(insertIndex), obj0Wrap, newPt.m_partId0, newPt.m_index0, obj1Wrap, newPt.m_partId1, newPt.m_index1);
 	}
 
 	if (gContactStartedCallback && isNewCollision)
@@ -204,4 +201,3 @@ void btManifoldResult::addContactPoint(const btVector3& normalOnBInWorld,const b
 		gContactStartedCallback(m_manifoldPtr);
 	}
 }
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btManifoldResult.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btManifoldResult.h
index 12cdafd1b6a..6c0a2d9a43f 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btManifoldResult.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btManifoldResult.h
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef BT_MANIFOLD_RESULT_H
 #define BT_MANIFOLD_RESULT_H
 
@@ -29,85 +28,81 @@ class btManifoldPoint;
 #include "BulletCollision/CollisionDispatch/btCollisionObjectWrapper.h"
 #include "BulletCollision/CollisionDispatch/btCollisionObject.h"
 
-typedef bool (*ContactAddedCallback)(btManifoldPoint& cp,	const btCollisionObjectWrapper* colObj0Wrap,int partId0,int index0,const btCollisionObjectWrapper* colObj1Wrap,int partId1,int index1);
-extern ContactAddedCallback		gContactAddedCallback;
+typedef bool (*ContactAddedCallback)(btManifoldPoint& cp, const btCollisionObjectWrapper* colObj0Wrap, int partId0, int index0, const btCollisionObjectWrapper* colObj1Wrap, int partId1, int index1);
+extern ContactAddedCallback gContactAddedCallback;
 
 //#define DEBUG_PART_INDEX 1
 
 /// These callbacks are used to customize the algorith that combine restitution, friction, damping, Stiffness
-typedef btScalar (*CalculateCombinedCallback)(const btCollisionObject* body0,const btCollisionObject* body1);
+typedef btScalar (*CalculateCombinedCallback)(const btCollisionObject* body0, const btCollisionObject* body1);
 
-extern CalculateCombinedCallback		gCalculateCombinedRestitutionCallback;
-extern CalculateCombinedCallback		gCalculateCombinedFrictionCallback;
-extern CalculateCombinedCallback		gCalculateCombinedRollingFrictionCallback;
-extern CalculateCombinedCallback		gCalculateCombinedSpinningFrictionCallback;
-extern CalculateCombinedCallback		gCalculateCombinedContactDampingCallback;
-extern CalculateCombinedCallback		gCalculateCombinedContactStiffnessCallback;
+extern CalculateCombinedCallback gCalculateCombinedRestitutionCallback;
+extern CalculateCombinedCallback gCalculateCombinedFrictionCallback;
+extern CalculateCombinedCallback gCalculateCombinedRollingFrictionCallback;
+extern CalculateCombinedCallback gCalculateCombinedSpinningFrictionCallback;
+extern CalculateCombinedCallback gCalculateCombinedContactDampingCallback;
+extern CalculateCombinedCallback gCalculateCombinedContactStiffnessCallback;
 
 ///btManifoldResult is a helper class to manage  contact results.
 class btManifoldResult : public btDiscreteCollisionDetectorInterface::Result
 {
 protected:
-
 	btPersistentManifold* m_manifoldPtr;
 
 	const btCollisionObjectWrapper* m_body0Wrap;
 	const btCollisionObjectWrapper* m_body1Wrap;
-	int	m_partId0;
+	int m_partId0;
 	int m_partId1;
 	int m_index0;
 	int m_index1;
-	
-	
-public:
 
+public:
 	btManifoldResult()
 		:
 #ifdef DEBUG_PART_INDEX
-		
-	m_partId0(-1),
-	m_partId1(-1),
-	m_index0(-1),
-	m_index1(-1)
-#endif //DEBUG_PART_INDEX
-		m_closestPointDistanceThreshold(0)
+
+		  m_partId0(-1),
+		  m_partId1(-1),
+		  m_index0(-1),
+		  m_index1(-1)
+#endif  //DEBUG_PART_INDEX
+			  m_closestPointDistanceThreshold(0)
 	{
 	}
 
-	btManifoldResult(const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap);
+	btManifoldResult(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap);
 
-	virtual ~btManifoldResult() {};
+	virtual ~btManifoldResult(){};
 
-	void	setPersistentManifold(btPersistentManifold* manifoldPtr)
+	void setPersistentManifold(btPersistentManifold* manifoldPtr)
 	{
 		m_manifoldPtr = manifoldPtr;
 	}
 
-	const btPersistentManifold*	getPersistentManifold() const
+	const btPersistentManifold* getPersistentManifold() const
 	{
 		return m_manifoldPtr;
 	}
-	btPersistentManifold*	getPersistentManifold()
+	btPersistentManifold* getPersistentManifold()
 	{
 		return m_manifoldPtr;
 	}
 
-	virtual void setShapeIdentifiersA(int partId0,int index0)
+	virtual void setShapeIdentifiersA(int partId0, int index0)
 	{
-		m_partId0=partId0;
-		m_index0=index0;
+		m_partId0 = partId0;
+		m_index0 = index0;
 	}
 
-	virtual void setShapeIdentifiersB(	int partId1,int index1)
+	virtual void setShapeIdentifiersB(int partId1, int index1)
 	{
-		m_partId1=partId1;
-		m_index1=index1;
+		m_partId1 = partId1;
+		m_index1 = index1;
 	}
 
+	virtual void addContactPoint(const btVector3& normalOnBInWorld, const btVector3& pointInWorld, btScalar depth);
 
-	virtual void addContactPoint(const btVector3& normalOnBInWorld,const btVector3& pointInWorld,btScalar depth);
-
-	SIMD_FORCE_INLINE	void refreshContactPoints()
+	SIMD_FORCE_INLINE void refreshContactPoints()
 	{
 		btAssert(m_manifoldPtr);
 		if (!m_manifoldPtr->getNumContacts())
@@ -117,10 +112,11 @@ public:
 
 		if (isSwapped)
 		{
-			m_manifoldPtr->refreshContactPoints(m_body1Wrap->getCollisionObject()->getWorldTransform(),m_body0Wrap->getCollisionObject()->getWorldTransform());
-		} else
+			m_manifoldPtr->refreshContactPoints(m_body1Wrap->getCollisionObject()->getWorldTransform(), m_body0Wrap->getCollisionObject()->getWorldTransform());
+		}
+		else
 		{
-			m_manifoldPtr->refreshContactPoints(m_body0Wrap->getCollisionObject()->getWorldTransform(),m_body1Wrap->getCollisionObject()->getWorldTransform());
+			m_manifoldPtr->refreshContactPoints(m_body0Wrap->getCollisionObject()->getWorldTransform(), m_body1Wrap->getCollisionObject()->getWorldTransform());
 		}
 	}
 
@@ -153,15 +149,15 @@ public:
 		return m_body1Wrap->getCollisionObject();
 	}
 
-	btScalar	m_closestPointDistanceThreshold;
+	btScalar m_closestPointDistanceThreshold;
 
 	/// in the future we can let the user override the methods to combine restitution and friction
-	static btScalar	calculateCombinedRestitution(const btCollisionObject* body0,const btCollisionObject* body1);
-	static btScalar	calculateCombinedFriction(const btCollisionObject* body0,const btCollisionObject* body1);
-	static btScalar calculateCombinedRollingFriction(const btCollisionObject* body0,const btCollisionObject* body1);
-    static btScalar calculateCombinedSpinningFriction(const btCollisionObject* body0,const btCollisionObject* body1);
-    static btScalar calculateCombinedContactDamping(const btCollisionObject* body0,const btCollisionObject* body1);
-	static btScalar calculateCombinedContactStiffness(const btCollisionObject* body0,const btCollisionObject* body1);
+	static btScalar calculateCombinedRestitution(const btCollisionObject* body0, const btCollisionObject* body1);
+	static btScalar calculateCombinedFriction(const btCollisionObject* body0, const btCollisionObject* body1);
+	static btScalar calculateCombinedRollingFriction(const btCollisionObject* body0, const btCollisionObject* body1);
+	static btScalar calculateCombinedSpinningFriction(const btCollisionObject* body0, const btCollisionObject* body1);
+	static btScalar calculateCombinedContactDamping(const btCollisionObject* body0, const btCollisionObject* body1);
+	static btScalar calculateCombinedContactStiffness(const btCollisionObject* body0, const btCollisionObject* body1);
 };
 
-#endif //BT_MANIFOLD_RESULT_H
+#endif  //BT_MANIFOLD_RESULT_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btSimulationIslandManager.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btSimulationIslandManager.cpp
index 91c76a8dac4..e5097ccbbf5 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btSimulationIslandManager.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btSimulationIslandManager.cpp
@@ -14,7 +14,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "LinearMath/btScalar.h"
 #include "btSimulationIslandManager.h"
 #include "BulletCollision/BroadphaseCollision/btDispatcher.h"
@@ -25,8 +24,7 @@ subject to the following restrictions:
 //#include <stdio.h>
 #include "LinearMath/btQuickprof.h"
 
-btSimulationIslandManager::btSimulationIslandManager():
-m_splitIslands(true)
+btSimulationIslandManager::btSimulationIslandManager() : m_splitIslands(true)
 {
 }
 
@@ -34,53 +32,47 @@ btSimulationIslandManager::~btSimulationIslandManager()
 {
 }
 
-
 void btSimulationIslandManager::initUnionFind(int n)
 {
-		m_unionFind.reset(n);
+	m_unionFind.reset(n);
 }
-		
 
-void btSimulationIslandManager::findUnions(btDispatcher* /* dispatcher */,btCollisionWorld* colWorld)
+void btSimulationIslandManager::findUnions(btDispatcher* /* dispatcher */, btCollisionWorld* colWorld)
 {
-	
 	{
 		btOverlappingPairCache* pairCachePtr = colWorld->getPairCache();
 		const int numOverlappingPairs = pairCachePtr->getNumOverlappingPairs();
 		if (numOverlappingPairs)
 		{
-		btBroadphasePair* pairPtr = pairCachePtr->getOverlappingPairArrayPtr();
-		
-		for (int i=0;i<numOverlappingPairs;i++)
-		{
-			const btBroadphasePair& collisionPair = pairPtr[i];
-			btCollisionObject* colObj0 = (btCollisionObject*)collisionPair.m_pProxy0->m_clientObject;
-			btCollisionObject* colObj1 = (btCollisionObject*)collisionPair.m_pProxy1->m_clientObject;
+			btBroadphasePair* pairPtr = pairCachePtr->getOverlappingPairArrayPtr();
 
-			if (((colObj0) && ((colObj0)->mergesSimulationIslands())) &&
-				((colObj1) && ((colObj1)->mergesSimulationIslands())))
+			for (int i = 0; i < numOverlappingPairs; i++)
 			{
+				const btBroadphasePair& collisionPair = pairPtr[i];
+				btCollisionObject* colObj0 = (btCollisionObject*)collisionPair.m_pProxy0->m_clientObject;
+				btCollisionObject* colObj1 = (btCollisionObject*)collisionPair.m_pProxy1->m_clientObject;
 
-				m_unionFind.unite((colObj0)->getIslandTag(),
-					(colObj1)->getIslandTag());
+				if (((colObj0) && ((colObj0)->mergesSimulationIslands())) &&
+					((colObj1) && ((colObj1)->mergesSimulationIslands())))
+				{
+					m_unionFind.unite((colObj0)->getIslandTag(),
+									  (colObj1)->getIslandTag());
+				}
 			}
 		}
-		}
 	}
 }
 
 #ifdef STATIC_SIMULATION_ISLAND_OPTIMIZATION
-void   btSimulationIslandManager::updateActivationState(btCollisionWorld* colWorld,btDispatcher* dispatcher)
+void btSimulationIslandManager::updateActivationState(btCollisionWorld* colWorld, btDispatcher* dispatcher)
 {
-
-	// put the index into m_controllers into m_tag   
+	// put the index into m_controllers into m_tag
 	int index = 0;
 	{
-
 		int i;
-		for (i=0;i<colWorld->getCollisionObjectArray().size(); i++)
+		for (i = 0; i < colWorld->getCollisionObjectArray().size(); i++)
 		{
-			btCollisionObject*   collisionObject= colWorld->getCollisionObjectArray()[i];
+			btCollisionObject* collisionObject = colWorld->getCollisionObjectArray()[i];
 			//Adding filtering here
 			if (!collisionObject->isStaticOrKinematicObject())
 			{
@@ -92,28 +84,29 @@ void   btSimulationIslandManager::updateActivationState(btCollisionWorld* colWor
 	}
 	// do the union find
 
-	initUnionFind( index );
+	initUnionFind(index);
 
-	findUnions(dispatcher,colWorld);
+	findUnions(dispatcher, colWorld);
 }
 
-void   btSimulationIslandManager::storeIslandActivationState(btCollisionWorld* colWorld)
+void btSimulationIslandManager::storeIslandActivationState(btCollisionWorld* colWorld)
 {
-	// put the islandId ('find' value) into m_tag   
+	// put the islandId ('find' value) into m_tag
 	{
 		int index = 0;
 		int i;
-		for (i=0;i<colWorld->getCollisionObjectArray().size();i++)
+		for (i = 0; i < colWorld->getCollisionObjectArray().size(); i++)
 		{
-			btCollisionObject* collisionObject= colWorld->getCollisionObjectArray()[i];
+			btCollisionObject* collisionObject = colWorld->getCollisionObjectArray()[i];
 			if (!collisionObject->isStaticOrKinematicObject())
 			{
-				collisionObject->setIslandTag( m_unionFind.find(index) );
+				collisionObject->setIslandTag(m_unionFind.find(index));
 				//Set the correct object offset in Collision Object Array
 				m_unionFind.getElement(index).m_sz = i;
 				collisionObject->setCompanionId(-1);
 				index++;
-			} else
+			}
+			else
 			{
 				collisionObject->setIslandTag(-1);
 				collisionObject->setCompanionId(-2);
@@ -122,49 +115,44 @@ void   btSimulationIslandManager::storeIslandActivationState(btCollisionWorld* c
 	}
 }
 
-
-#else //STATIC_SIMULATION_ISLAND_OPTIMIZATION
-void	btSimulationIslandManager::updateActivationState(btCollisionWorld* colWorld,btDispatcher* dispatcher)
+#else  //STATIC_SIMULATION_ISLAND_OPTIMIZATION
+void btSimulationIslandManager::updateActivationState(btCollisionWorld* colWorld, btDispatcher* dispatcher)
 {
+	initUnionFind(int(colWorld->getCollisionObjectArray().size()));
 
-	initUnionFind( int (colWorld->getCollisionObjectArray().size()));
-
-	// put the index into m_controllers into m_tag	
+	// put the index into m_controllers into m_tag
 	{
-
 		int index = 0;
 		int i;
-		for (i=0;i<colWorld->getCollisionObjectArray().size(); i++)
+		for (i = 0; i < colWorld->getCollisionObjectArray().size(); i++)
 		{
-			btCollisionObject*	collisionObject= colWorld->getCollisionObjectArray()[i];
+			btCollisionObject* collisionObject = colWorld->getCollisionObjectArray()[i];
 			collisionObject->setIslandTag(index);
 			collisionObject->setCompanionId(-1);
 			collisionObject->setHitFraction(btScalar(1.));
 			index++;
-
 		}
 	}
 	// do the union find
 
-	findUnions(dispatcher,colWorld);
+	findUnions(dispatcher, colWorld);
 }
 
-void	btSimulationIslandManager::storeIslandActivationState(btCollisionWorld* colWorld)
+void btSimulationIslandManager::storeIslandActivationState(btCollisionWorld* colWorld)
 {
-	// put the islandId ('find' value) into m_tag	
+	// put the islandId ('find' value) into m_tag
 	{
-
-
 		int index = 0;
 		int i;
-		for (i=0;i<colWorld->getCollisionObjectArray().size();i++)
+		for (i = 0; i < colWorld->getCollisionObjectArray().size(); i++)
 		{
-			btCollisionObject* collisionObject= colWorld->getCollisionObjectArray()[i];
+			btCollisionObject* collisionObject = colWorld->getCollisionObjectArray()[i];
 			if (!collisionObject->isStaticOrKinematicObject())
 			{
-				collisionObject->setIslandTag( m_unionFind.find(index) );
+				collisionObject->setIslandTag(m_unionFind.find(index));
 				collisionObject->setCompanionId(-1);
-			} else
+			}
+			else
 			{
 				collisionObject->setIslandTag(-1);
 				collisionObject->setCompanionId(-2);
@@ -174,72 +162,59 @@ void	btSimulationIslandManager::storeIslandActivationState(btCollisionWorld* col
 	}
 }
 
-#endif //STATIC_SIMULATION_ISLAND_OPTIMIZATION
+#endif  //STATIC_SIMULATION_ISLAND_OPTIMIZATION
 
-inline	int	getIslandId(const btPersistentManifold* lhs)
+inline int getIslandId(const btPersistentManifold* lhs)
 {
 	int islandId;
 	const btCollisionObject* rcolObj0 = static_cast<const btCollisionObject*>(lhs->getBody0());
 	const btCollisionObject* rcolObj1 = static_cast<const btCollisionObject*>(lhs->getBody1());
-	islandId= rcolObj0->getIslandTag()>=0?rcolObj0->getIslandTag():rcolObj1->getIslandTag();
+	islandId = rcolObj0->getIslandTag() >= 0 ? rcolObj0->getIslandTag() : rcolObj1->getIslandTag();
 	return islandId;
-
 }
 
-
-
 /// function object that routes calls to operator<
 class btPersistentManifoldSortPredicate
 {
-	public:
-
-		SIMD_FORCE_INLINE bool operator() ( const btPersistentManifold* lhs, const btPersistentManifold* rhs ) const
-		{
-			return getIslandId(lhs) < getIslandId(rhs);
-		}
+public:
+	SIMD_FORCE_INLINE bool operator()(const btPersistentManifold* lhs, const btPersistentManifold* rhs) const
+	{
+		return getIslandId(lhs) < getIslandId(rhs);
+	}
 };
 
 class btPersistentManifoldSortPredicateDeterministic
 {
 public:
-
-	SIMD_FORCE_INLINE bool operator() (const btPersistentManifold* lhs, const btPersistentManifold* rhs) const
+	SIMD_FORCE_INLINE bool operator()(const btPersistentManifold* lhs, const btPersistentManifold* rhs) const
 	{
 		return (
-			(getIslandId(lhs) < getIslandId(rhs))
-							|| ((getIslandId(lhs) == getIslandId(rhs)) && lhs->getBody0()->getBroadphaseHandle()->m_uniqueId < rhs->getBody0()->getBroadphaseHandle()->m_uniqueId) 
-						||((getIslandId(lhs) == getIslandId(rhs)) && (lhs->getBody0()->getBroadphaseHandle()->m_uniqueId == rhs->getBody0()->getBroadphaseHandle()->m_uniqueId) &&
-					(lhs->getBody1()->getBroadphaseHandle()->m_uniqueId < rhs->getBody1()->getBroadphaseHandle()->m_uniqueId))
-			);
-
+			(getIslandId(lhs) < getIslandId(rhs)) || ((getIslandId(lhs) == getIslandId(rhs)) && lhs->getBody0()->getBroadphaseHandle()->m_uniqueId < rhs->getBody0()->getBroadphaseHandle()->m_uniqueId) || ((getIslandId(lhs) == getIslandId(rhs)) && (lhs->getBody0()->getBroadphaseHandle()->m_uniqueId == rhs->getBody0()->getBroadphaseHandle()->m_uniqueId) && (lhs->getBody1()->getBroadphaseHandle()->m_uniqueId < rhs->getBody1()->getBroadphaseHandle()->m_uniqueId)));
 	}
 };
 
-
-void btSimulationIslandManager::buildIslands(btDispatcher* dispatcher,btCollisionWorld* collisionWorld)
+void btSimulationIslandManager::buildIslands(btDispatcher* dispatcher, btCollisionWorld* collisionWorld)
 {
-
 	BT_PROFILE("islandUnionFindAndQuickSort");
-	
+
 	btCollisionObjectArray& collisionObjects = collisionWorld->getCollisionObjectArray();
 
 	m_islandmanifold.resize(0);
 
 	//we are going to sort the unionfind array, and store the element id in the size
 	//afterwards, we clean unionfind, to make sure no-one uses it anymore
-	
+
 	getUnionFind().sortIslands();
 	int numElem = getUnionFind().getNumElements();
 
-	int endIslandIndex=1;
+	int endIslandIndex = 1;
 	int startIslandIndex;
 
-
 	//update the sleeping state for bodies, if all are sleeping
-	for ( startIslandIndex=0;startIslandIndex<numElem;startIslandIndex = endIslandIndex)
+	for (startIslandIndex = 0; startIslandIndex < numElem; startIslandIndex = endIslandIndex)
 	{
 		int islandId = getUnionFind().getElement(startIslandIndex).m_id;
-		for (endIslandIndex = startIslandIndex+1;(endIslandIndex<numElem) && (getUnionFind().getElement(endIslandIndex).m_id == islandId);endIslandIndex++)
+		for (endIslandIndex = startIslandIndex + 1; (endIslandIndex < numElem) && (getUnionFind().getElement(endIslandIndex).m_id == islandId); endIslandIndex++)
 		{
 		}
 
@@ -248,69 +223,68 @@ void btSimulationIslandManager::buildIslands(btDispatcher* dispatcher,btCollisio
 		bool allSleeping = true;
 
 		int idx;
-		for (idx=startIslandIndex;idx<endIslandIndex;idx++)
+		for (idx = startIslandIndex; idx < endIslandIndex; idx++)
 		{
 			int i = getUnionFind().getElement(idx).m_sz;
 
 			btCollisionObject* colObj0 = collisionObjects[i];
 			if ((colObj0->getIslandTag() != islandId) && (colObj0->getIslandTag() != -1))
 			{
-//				printf("error in island management\n");
+				//				printf("error in island management\n");
 			}
 
 			btAssert((colObj0->getIslandTag() == islandId) || (colObj0->getIslandTag() == -1));
 			if (colObj0->getIslandTag() == islandId)
 			{
-				if (colObj0->getActivationState()== ACTIVE_TAG ||
-				   colObj0->getActivationState()== DISABLE_DEACTIVATION)
+				if (colObj0->getActivationState() == ACTIVE_TAG ||
+					colObj0->getActivationState() == DISABLE_DEACTIVATION)
 				{
 					allSleeping = false;
 					break;
 				}
 			}
 		}
-			
 
 		if (allSleeping)
 		{
 			int idx;
-			for (idx=startIslandIndex;idx<endIslandIndex;idx++)
+			for (idx = startIslandIndex; idx < endIslandIndex; idx++)
 			{
 				int i = getUnionFind().getElement(idx).m_sz;
 				btCollisionObject* colObj0 = collisionObjects[i];
 				if ((colObj0->getIslandTag() != islandId) && (colObj0->getIslandTag() != -1))
 				{
-//					printf("error in island management\n");
+					//					printf("error in island management\n");
 				}
 
 				btAssert((colObj0->getIslandTag() == islandId) || (colObj0->getIslandTag() == -1));
 
 				if (colObj0->getIslandTag() == islandId)
 				{
-					colObj0->setActivationState( ISLAND_SLEEPING );
+					colObj0->setActivationState(ISLAND_SLEEPING);
 				}
 			}
-		} else
+		}
+		else
 		{
-
 			int idx;
-			for (idx=startIslandIndex;idx<endIslandIndex;idx++)
+			for (idx = startIslandIndex; idx < endIslandIndex; idx++)
 			{
 				int i = getUnionFind().getElement(idx).m_sz;
 
 				btCollisionObject* colObj0 = collisionObjects[i];
 				if ((colObj0->getIslandTag() != islandId) && (colObj0->getIslandTag() != -1))
 				{
-//					printf("error in island management\n");
+					//					printf("error in island management\n");
 				}
 
 				btAssert((colObj0->getIslandTag() == islandId) || (colObj0->getIslandTag() == -1));
 
 				if (colObj0->getIslandTag() == islandId)
 				{
-					if ( colObj0->getActivationState() == ISLAND_SLEEPING)
+					if (colObj0->getActivationState() == ISLAND_SLEEPING)
 					{
-						colObj0->setActivationState( WANTS_DEACTIVATION);
+						colObj0->setActivationState(WANTS_DEACTIVATION);
 						colObj0->setDeactivationTime(0.f);
 					}
 				}
@@ -318,34 +292,30 @@ void btSimulationIslandManager::buildIslands(btDispatcher* dispatcher,btCollisio
 		}
 	}
 
-	
 	int i;
 	int maxNumManifolds = dispatcher->getNumManifolds();
 
-//#define SPLIT_ISLANDS 1
-//#ifdef SPLIT_ISLANDS
+	//#define SPLIT_ISLANDS 1
+	//#ifdef SPLIT_ISLANDS
 
-	
-//#endif //SPLIT_ISLANDS
+	//#endif //SPLIT_ISLANDS
 
-	
-	for (i=0;i<maxNumManifolds ;i++)
+	for (i = 0; i < maxNumManifolds; i++)
 	{
-		 btPersistentManifold* manifold = dispatcher->getManifoldByIndexInternal(i);
-		 if (collisionWorld->getDispatchInfo().m_deterministicOverlappingPairs)
-		 {
+		btPersistentManifold* manifold = dispatcher->getManifoldByIndexInternal(i);
+		if (collisionWorld->getDispatchInfo().m_deterministicOverlappingPairs)
+		{
 			if (manifold->getNumContacts() == 0)
 				continue;
-		 }
+		}
 
-		 const btCollisionObject* colObj0 = static_cast<const btCollisionObject*>(manifold->getBody0());
-		 const btCollisionObject* colObj1 = static_cast<const btCollisionObject*>(manifold->getBody1());
-		
-		 ///@todo: check sleeping conditions!
-		 if (((colObj0) && colObj0->getActivationState() != ISLAND_SLEEPING) ||
+		const btCollisionObject* colObj0 = static_cast<const btCollisionObject*>(manifold->getBody0());
+		const btCollisionObject* colObj1 = static_cast<const btCollisionObject*>(manifold->getBody1());
+
+		///@todo: check sleeping conditions!
+		if (((colObj0) && colObj0->getActivationState() != ISLAND_SLEEPING) ||
 			((colObj1) && colObj1->getActivationState() != ISLAND_SLEEPING))
 		{
-		
 			//kinematic objects don't merge islands, but wake up all connected objects
 			if (colObj0->isKinematicObject() && colObj0->getActivationState() != ISLAND_SLEEPING)
 			{
@@ -357,36 +327,34 @@ void btSimulationIslandManager::buildIslands(btDispatcher* dispatcher,btCollisio
 				if (colObj1->hasContactResponse())
 					colObj0->activate();
 			}
-			if(m_splitIslands)
-			{ 
+			if (m_splitIslands)
+			{
 				//filtering for response
-				if (dispatcher->needsResponse(colObj0,colObj1))
+				if (dispatcher->needsResponse(colObj0, colObj1))
 					m_islandmanifold.push_back(manifold);
 			}
 		}
 	}
 }
 
-
-
 ///@todo: this is random access, it can be walked 'cache friendly'!
-void btSimulationIslandManager::buildAndProcessIslands(btDispatcher* dispatcher,btCollisionWorld* collisionWorld, IslandCallback* callback)
+void btSimulationIslandManager::buildAndProcessIslands(btDispatcher* dispatcher, btCollisionWorld* collisionWorld, IslandCallback* callback)
 {
 	btCollisionObjectArray& collisionObjects = collisionWorld->getCollisionObjectArray();
 
-	buildIslands(dispatcher,collisionWorld);
+	buildIslands(dispatcher, collisionWorld);
 
-	int endIslandIndex=1;
+	int endIslandIndex = 1;
 	int startIslandIndex;
 	int numElem = getUnionFind().getNumElements();
 
 	BT_PROFILE("processIslands");
 
-	if(!m_splitIslands)
+	if (!m_splitIslands)
 	{
 		btPersistentManifold** manifold = dispatcher->getInternalManifoldPointer();
 		int maxNumManifolds = dispatcher->getNumManifolds();
-		callback->processIsland(&collisionObjects[0],collisionObjects.size(),manifold,maxNumManifolds, -1);
+		callback->processIsland(&collisionObjects[0], collisionObjects.size(), manifold, maxNumManifolds, -1);
 	}
 	else
 	{
@@ -394,7 +362,7 @@ void btSimulationIslandManager::buildAndProcessIslands(btDispatcher* dispatcher,
 		// Sort the vector using predicate and std::sort
 		//std::sort(islandmanifold.begin(), islandmanifold.end(), btPersistentManifoldSortPredicate);
 
-		int numManifolds = int (m_islandmanifold.size());
+		int numManifolds = int(m_islandmanifold.size());
 
 		//tried a radix sort, but quicksort/heapsort seems still faster
 		//@todo rewrite island management
@@ -403,7 +371,8 @@ void btSimulationIslandManager::buildAndProcessIslands(btDispatcher* dispatcher,
 		if (collisionWorld->getDispatchInfo().m_deterministicOverlappingPairs)
 		{
 			m_islandmanifold.quickSort(btPersistentManifoldSortPredicateDeterministic());
-		} else
+		}
+		else
 		{
 			m_islandmanifold.quickSort(btPersistentManifoldSortPredicate());
 		}
@@ -417,55 +386,49 @@ void btSimulationIslandManager::buildAndProcessIslands(btDispatcher* dispatcher,
 
 		//int islandId;
 
-		
-
-	//	printf("Start Islands\n");
+		//	printf("Start Islands\n");
 
 		//traverse the simulation islands, and call the solver, unless all objects are sleeping/deactivated
-		for ( startIslandIndex=0;startIslandIndex<numElem;startIslandIndex = endIslandIndex)
+		for (startIslandIndex = 0; startIslandIndex < numElem; startIslandIndex = endIslandIndex)
 		{
 			int islandId = getUnionFind().getElement(startIslandIndex).m_id;
 
+			bool islandSleeping = true;
 
-			   bool islandSleeping = true;
-	                
-					for (endIslandIndex = startIslandIndex;(endIslandIndex<numElem) && (getUnionFind().getElement(endIslandIndex).m_id == islandId);endIslandIndex++)
-					{
-							int i = getUnionFind().getElement(endIslandIndex).m_sz;
-							btCollisionObject* colObj0 = collisionObjects[i];
-							m_islandBodies.push_back(colObj0);
-							if (colObj0->isActive())
-									islandSleeping = false;
-					}
-	                
+			for (endIslandIndex = startIslandIndex; (endIslandIndex < numElem) && (getUnionFind().getElement(endIslandIndex).m_id == islandId); endIslandIndex++)
+			{
+				int i = getUnionFind().getElement(endIslandIndex).m_sz;
+				btCollisionObject* colObj0 = collisionObjects[i];
+				m_islandBodies.push_back(colObj0);
+				if (colObj0->isActive())
+					islandSleeping = false;
+			}
 
 			//find the accompanying contact manifold for this islandId
 			int numIslandManifolds = 0;
 			btPersistentManifold** startManifold = 0;
 
-			if (startManifoldIndex<numManifolds)
+			if (startManifoldIndex < numManifolds)
 			{
 				int curIslandId = getIslandId(m_islandmanifold[startManifoldIndex]);
 				if (curIslandId == islandId)
 				{
 					startManifold = &m_islandmanifold[startManifoldIndex];
-				
-					for (endManifoldIndex = startManifoldIndex+1;(endManifoldIndex<numManifolds) && (islandId == getIslandId(m_islandmanifold[endManifoldIndex]));endManifoldIndex++)
-					{
 
+					for (endManifoldIndex = startManifoldIndex + 1; (endManifoldIndex < numManifolds) && (islandId == getIslandId(m_islandmanifold[endManifoldIndex])); endManifoldIndex++)
+					{
 					}
 					/// Process the actual simulation, only if not sleeping/deactivated
-					numIslandManifolds = endManifoldIndex-startManifoldIndex;
+					numIslandManifolds = endManifoldIndex - startManifoldIndex;
 				}
-
 			}
 
 			if (!islandSleeping)
 			{
-				callback->processIsland(&m_islandBodies[0],m_islandBodies.size(),startManifold,numIslandManifolds, islandId);
-	//			printf("Island callback of size:%d bodies, %d manifolds\n",islandBodies.size(),numIslandManifolds);
+				callback->processIsland(&m_islandBodies[0], m_islandBodies.size(), startManifold, numIslandManifolds, islandId);
+				//			printf("Island callback of size:%d bodies, %d manifolds\n",islandBodies.size(),numIslandManifolds);
 			}
-			
+
 			if (numIslandManifolds)
 			{
 				startManifoldIndex = endManifoldIndex;
@@ -473,6 +436,5 @@ void btSimulationIslandManager::buildAndProcessIslands(btDispatcher* dispatcher,
 
 			m_islandBodies.resize(0);
 		}
-	} // else if(!splitIslands) 
-
+	}  // else if(!splitIslands)
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btSimulationIslandManager.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btSimulationIslandManager.h
index e24c6afeca1..6c2802141c6 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btSimulationIslandManager.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btSimulationIslandManager.h
@@ -26,45 +26,39 @@ class btCollisionWorld;
 class btDispatcher;
 class btPersistentManifold;
 
-
 ///SimulationIslandManager creates and handles simulation islands, using btUnionFind
 class btSimulationIslandManager
 {
 	btUnionFind m_unionFind;
 
-	btAlignedObjectArray<btPersistentManifold*>  m_islandmanifold;
-	btAlignedObjectArray<btCollisionObject* >  m_islandBodies;
-	
+	btAlignedObjectArray<btPersistentManifold*> m_islandmanifold;
+	btAlignedObjectArray<btCollisionObject*> m_islandBodies;
+
 	bool m_splitIslands;
-	
+
 public:
 	btSimulationIslandManager();
 	virtual ~btSimulationIslandManager();
 
+	void initUnionFind(int n);
 
-	void initUnionFind(int n);	
-	
-		
-	btUnionFind& getUnionFind() { return m_unionFind;}
+	btUnionFind& getUnionFind() { return m_unionFind; }
 
-	virtual	void	updateActivationState(btCollisionWorld* colWorld,btDispatcher* dispatcher);
-	virtual	void	storeIslandActivationState(btCollisionWorld* world);
+	virtual void updateActivationState(btCollisionWorld* colWorld, btDispatcher* dispatcher);
+	virtual void storeIslandActivationState(btCollisionWorld* world);
 
+	void findUnions(btDispatcher* dispatcher, btCollisionWorld* colWorld);
 
-	void	findUnions(btDispatcher* dispatcher,btCollisionWorld* colWorld);
-
-	
-
-	struct	IslandCallback
+	struct IslandCallback
 	{
-		virtual ~IslandCallback() {};
+		virtual ~IslandCallback(){};
 
-		virtual	void	processIsland(btCollisionObject** bodies,int numBodies,class btPersistentManifold**	manifolds,int numManifolds, int islandId) = 0;
+		virtual void processIsland(btCollisionObject** bodies, int numBodies, class btPersistentManifold** manifolds, int numManifolds, int islandId) = 0;
 	};
 
-	void	buildAndProcessIslands(btDispatcher* dispatcher,btCollisionWorld* collisionWorld, IslandCallback* callback);
+	void buildAndProcessIslands(btDispatcher* dispatcher, btCollisionWorld* collisionWorld, IslandCallback* callback);
 
-	void buildIslands(btDispatcher* dispatcher,btCollisionWorld* colWorld);
+	void buildIslands(btDispatcher* dispatcher, btCollisionWorld* colWorld);
 
 	bool getSplitIslands()
 	{
@@ -74,8 +68,6 @@ public:
 	{
 		m_splitIslands = doSplitIslands;
 	}
-
 };
 
-#endif //BT_SIMULATION_ISLAND_MANAGER_H
-
+#endif  //BT_SIMULATION_ISLAND_MANAGER_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereBoxCollisionAlgorithm.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereBoxCollisionAlgorithm.cpp
index e8b567e0efc..bc68b285b8d 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereBoxCollisionAlgorithm.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereBoxCollisionAlgorithm.cpp
@@ -21,23 +21,22 @@ subject to the following restrictions:
 #include "BulletCollision/CollisionDispatch/btCollisionObjectWrapper.h"
 //#include <stdio.h>
 
-btSphereBoxCollisionAlgorithm::btSphereBoxCollisionAlgorithm(btPersistentManifold* mf,const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* col0Wrap,const btCollisionObjectWrapper* col1Wrap, bool isSwapped)
-: btActivatingCollisionAlgorithm(ci,col0Wrap,col1Wrap),
-m_ownManifold(false),
-m_manifoldPtr(mf),
-m_isSwapped(isSwapped)
+btSphereBoxCollisionAlgorithm::btSphereBoxCollisionAlgorithm(btPersistentManifold* mf, const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* col0Wrap, const btCollisionObjectWrapper* col1Wrap, bool isSwapped)
+	: btActivatingCollisionAlgorithm(ci, col0Wrap, col1Wrap),
+	  m_ownManifold(false),
+	  m_manifoldPtr(mf),
+	  m_isSwapped(isSwapped)
 {
-	const btCollisionObjectWrapper* sphereObjWrap = m_isSwapped? col1Wrap : col0Wrap;
-	const btCollisionObjectWrapper* boxObjWrap = m_isSwapped? col0Wrap : col1Wrap;
-	
-	if (!m_manifoldPtr && m_dispatcher->needsCollision(sphereObjWrap->getCollisionObject(),boxObjWrap->getCollisionObject()))
+	const btCollisionObjectWrapper* sphereObjWrap = m_isSwapped ? col1Wrap : col0Wrap;
+	const btCollisionObjectWrapper* boxObjWrap = m_isSwapped ? col0Wrap : col1Wrap;
+
+	if (!m_manifoldPtr && m_dispatcher->needsCollision(sphereObjWrap->getCollisionObject(), boxObjWrap->getCollisionObject()))
 	{
-		m_manifoldPtr = m_dispatcher->getNewManifold(sphereObjWrap->getCollisionObject(),boxObjWrap->getCollisionObject());
+		m_manifoldPtr = m_dispatcher->getNewManifold(sphereObjWrap->getCollisionObject(), boxObjWrap->getCollisionObject());
 		m_ownManifold = true;
 	}
 }
 
-
 btSphereBoxCollisionAlgorithm::~btSphereBoxCollisionAlgorithm()
 {
 	if (m_ownManifold)
@@ -47,17 +46,15 @@ btSphereBoxCollisionAlgorithm::~btSphereBoxCollisionAlgorithm()
 	}
 }
 
-
-
-void btSphereBoxCollisionAlgorithm::processCollision (const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
+void btSphereBoxCollisionAlgorithm::processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	(void)dispatchInfo;
 	(void)resultOut;
 	if (!m_manifoldPtr)
 		return;
 
-	const btCollisionObjectWrapper* sphereObjWrap = m_isSwapped? body1Wrap : body0Wrap;
-	const btCollisionObjectWrapper* boxObjWrap = m_isSwapped? body0Wrap : body1Wrap;
+	const btCollisionObjectWrapper* sphereObjWrap = m_isSwapped ? body1Wrap : body0Wrap;
+	const btCollisionObjectWrapper* boxObjWrap = m_isSwapped ? body0Wrap : body1Wrap;
 
 	btVector3 pOnBox;
 
@@ -83,10 +80,9 @@ void btSphereBoxCollisionAlgorithm::processCollision (const btCollisionObjectWra
 			resultOut->refreshContactPoints();
 		}
 	}
-
 }
 
-btScalar btSphereBoxCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* col0,btCollisionObject* col1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+btScalar btSphereBoxCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* col0, btCollisionObject* col1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	(void)resultOut;
 	(void)dispatchInfo;
@@ -97,27 +93,26 @@ btScalar btSphereBoxCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject*
 	return btScalar(1.);
 }
 
-
-bool btSphereBoxCollisionAlgorithm::getSphereDistance(const btCollisionObjectWrapper* boxObjWrap, btVector3& pointOnBox, btVector3& normal, btScalar& penetrationDepth, const btVector3& sphereCenter, btScalar fRadius, btScalar maxContactDistance ) 
+bool btSphereBoxCollisionAlgorithm::getSphereDistance(const btCollisionObjectWrapper* boxObjWrap, btVector3& pointOnBox, btVector3& normal, btScalar& penetrationDepth, const btVector3& sphereCenter, btScalar fRadius, btScalar maxContactDistance)
 {
-	const btBoxShape* boxShape= (const btBoxShape*)boxObjWrap->getCollisionShape();
-	btVector3 const &boxHalfExtent = boxShape->getHalfExtentsWithoutMargin();
+	const btBoxShape* boxShape = (const btBoxShape*)boxObjWrap->getCollisionShape();
+	btVector3 const& boxHalfExtent = boxShape->getHalfExtentsWithoutMargin();
 	btScalar boxMargin = boxShape->getMargin();
 	penetrationDepth = 1.0f;
 
 	// convert the sphere position to the box's local space
-	btTransform const &m44T = boxObjWrap->getWorldTransform();
+	btTransform const& m44T = boxObjWrap->getWorldTransform();
 	btVector3 sphereRelPos = m44T.invXform(sphereCenter);
 
 	// Determine the closest point to the sphere center in the box
 	btVector3 closestPoint = sphereRelPos;
-	closestPoint.setX( btMin(boxHalfExtent.getX(), closestPoint.getX()) );
-	closestPoint.setX( btMax(-boxHalfExtent.getX(), closestPoint.getX()) );
-	closestPoint.setY( btMin(boxHalfExtent.getY(), closestPoint.getY()) );
-	closestPoint.setY( btMax(-boxHalfExtent.getY(), closestPoint.getY()) );
-	closestPoint.setZ( btMin(boxHalfExtent.getZ(), closestPoint.getZ()) );
-	closestPoint.setZ( btMax(-boxHalfExtent.getZ(), closestPoint.getZ()) );
-	
+	closestPoint.setX(btMin(boxHalfExtent.getX(), closestPoint.getX()));
+	closestPoint.setX(btMax(-boxHalfExtent.getX(), closestPoint.getX()));
+	closestPoint.setY(btMin(boxHalfExtent.getY(), closestPoint.getY()));
+	closestPoint.setY(btMax(-boxHalfExtent.getY(), closestPoint.getY()));
+	closestPoint.setZ(btMin(boxHalfExtent.getZ(), closestPoint.getZ()));
+	closestPoint.setZ(btMax(-boxHalfExtent.getZ(), closestPoint.getZ()));
+
 	btScalar intersectionDist = fRadius + boxMargin;
 	btScalar contactDist = intersectionDist + maxContactDistance;
 	normal = sphereRelPos - closestPoint;
@@ -136,42 +131,42 @@ bool btSphereBoxCollisionAlgorithm::getSphereDistance(const btCollisionObjectWra
 	{
 		distance = -getSpherePenetration(boxHalfExtent, sphereRelPos, closestPoint, normal);
 	}
-	else //compute the penetration details
+	else  //compute the penetration details
 	{
 		distance = normal.length();
 		normal /= distance;
 	}
 
 	pointOnBox = closestPoint + normal * boxMargin;
-//	v3PointOnSphere = sphereRelPos - (normal * fRadius);	
+	//	v3PointOnSphere = sphereRelPos - (normal * fRadius);
 	penetrationDepth = distance - intersectionDist;
 
 	// transform back in world space
 	btVector3 tmp = m44T(pointOnBox);
 	pointOnBox = tmp;
-//	tmp = m44T(v3PointOnSphere);
-//	v3PointOnSphere = tmp;
+	//	tmp = m44T(v3PointOnSphere);
+	//	v3PointOnSphere = tmp;
 	tmp = m44T.getBasis() * normal;
 	normal = tmp;
 
 	return true;
 }
 
-btScalar btSphereBoxCollisionAlgorithm::getSpherePenetration( btVector3 const &boxHalfExtent, btVector3 const &sphereRelPos, btVector3 &closestPoint, btVector3& normal ) 
+btScalar btSphereBoxCollisionAlgorithm::getSpherePenetration(btVector3 const& boxHalfExtent, btVector3 const& sphereRelPos, btVector3& closestPoint, btVector3& normal)
 {
 	//project the center of the sphere on the closest face of the box
 	btScalar faceDist = boxHalfExtent.getX() - sphereRelPos.getX();
 	btScalar minDist = faceDist;
-	closestPoint.setX( boxHalfExtent.getX() );
-	normal.setValue(btScalar(1.0f),  btScalar(0.0f),  btScalar(0.0f));
+	closestPoint.setX(boxHalfExtent.getX());
+	normal.setValue(btScalar(1.0f), btScalar(0.0f), btScalar(0.0f));
 
 	faceDist = boxHalfExtent.getX() + sphereRelPos.getX();
 	if (faceDist < minDist)
 	{
 		minDist = faceDist;
 		closestPoint = sphereRelPos;
-		closestPoint.setX( -boxHalfExtent.getX() );
-		normal.setValue(btScalar(-1.0f),  btScalar(0.0f),  btScalar(0.0f));
+		closestPoint.setX(-boxHalfExtent.getX());
+		normal.setValue(btScalar(-1.0f), btScalar(0.0f), btScalar(0.0f));
 	}
 
 	faceDist = boxHalfExtent.getY() - sphereRelPos.getY();
@@ -179,8 +174,8 @@ btScalar btSphereBoxCollisionAlgorithm::getSpherePenetration( btVector3 const &b
 	{
 		minDist = faceDist;
 		closestPoint = sphereRelPos;
-		closestPoint.setY( boxHalfExtent.getY() );
-		normal.setValue(btScalar(0.0f),  btScalar(1.0f),  btScalar(0.0f));
+		closestPoint.setY(boxHalfExtent.getY());
+		normal.setValue(btScalar(0.0f), btScalar(1.0f), btScalar(0.0f));
 	}
 
 	faceDist = boxHalfExtent.getY() + sphereRelPos.getY();
@@ -188,8 +183,8 @@ btScalar btSphereBoxCollisionAlgorithm::getSpherePenetration( btVector3 const &b
 	{
 		minDist = faceDist;
 		closestPoint = sphereRelPos;
-		closestPoint.setY( -boxHalfExtent.getY() );
-		normal.setValue(btScalar(0.0f),  btScalar(-1.0f),  btScalar(0.0f));
+		closestPoint.setY(-boxHalfExtent.getY());
+		normal.setValue(btScalar(0.0f), btScalar(-1.0f), btScalar(0.0f));
 	}
 
 	faceDist = boxHalfExtent.getZ() - sphereRelPos.getZ();
@@ -197,8 +192,8 @@ btScalar btSphereBoxCollisionAlgorithm::getSpherePenetration( btVector3 const &b
 	{
 		minDist = faceDist;
 		closestPoint = sphereRelPos;
-		closestPoint.setZ( boxHalfExtent.getZ() );
-		normal.setValue(btScalar(0.0f),  btScalar(0.0f),  btScalar(1.0f));
+		closestPoint.setZ(boxHalfExtent.getZ());
+		normal.setValue(btScalar(0.0f), btScalar(0.0f), btScalar(1.0f));
 	}
 
 	faceDist = boxHalfExtent.getZ() + sphereRelPos.getZ();
@@ -206,8 +201,8 @@ btScalar btSphereBoxCollisionAlgorithm::getSpherePenetration( btVector3 const &b
 	{
 		minDist = faceDist;
 		closestPoint = sphereRelPos;
-		closestPoint.setZ( -boxHalfExtent.getZ() );
-		normal.setValue(btScalar(0.0f),  btScalar(0.0f),  btScalar(-1.0f));
+		closestPoint.setZ(-boxHalfExtent.getZ());
+		normal.setValue(btScalar(0.0f), btScalar(0.0f), btScalar(-1.0f));
 	}
 
 	return minDist;
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereBoxCollisionAlgorithm.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereBoxCollisionAlgorithm.h
index eefaedc9e7e..3348bc89af5 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereBoxCollisionAlgorithm.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereBoxCollisionAlgorithm.h
@@ -28,21 +28,20 @@ class btPersistentManifold;
 /// Other features are frame-coherency (persistent data) and collision response.
 class btSphereBoxCollisionAlgorithm : public btActivatingCollisionAlgorithm
 {
-	bool	m_ownManifold;
-	btPersistentManifold*	m_manifoldPtr;
-	bool	m_isSwapped;
-	
-public:
+	bool m_ownManifold;
+	btPersistentManifold* m_manifoldPtr;
+	bool m_isSwapped;
 
-	btSphereBoxCollisionAlgorithm(btPersistentManifold* mf,const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap, bool isSwapped);
+public:
+	btSphereBoxCollisionAlgorithm(btPersistentManifold* mf, const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, bool isSwapped);
 
 	virtual ~btSphereBoxCollisionAlgorithm();
 
-	virtual void processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual void processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual	void	getAllContactManifolds(btManifoldArray&	manifoldArray)
+	virtual void getAllContactManifolds(btManifoldArray& manifoldArray)
 	{
 		if (m_manifoldPtr && m_ownManifold)
 		{
@@ -50,26 +49,25 @@ public:
 		}
 	}
 
-	bool getSphereDistance( const btCollisionObjectWrapper* boxObjWrap, btVector3& v3PointOnBox, btVector3& normal, btScalar& penetrationDepth, const btVector3& v3SphereCenter, btScalar fRadius, btScalar maxContactDistance );
+	bool getSphereDistance(const btCollisionObjectWrapper* boxObjWrap, btVector3& v3PointOnBox, btVector3& normal, btScalar& penetrationDepth, const btVector3& v3SphereCenter, btScalar fRadius, btScalar maxContactDistance);
 
-	btScalar getSpherePenetration( btVector3 const &boxHalfExtent, btVector3 const &sphereRelPos, btVector3 &closestPoint, btVector3& normal );
-	
-	struct CreateFunc :public 	btCollisionAlgorithmCreateFunc
+	btScalar getSpherePenetration(btVector3 const& boxHalfExtent, btVector3 const& sphereRelPos, btVector3& closestPoint, btVector3& normal);
+
+	struct CreateFunc : public btCollisionAlgorithmCreateFunc
 	{
-		virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+		virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 		{
 			void* mem = ci.m_dispatcher1->allocateCollisionAlgorithm(sizeof(btSphereBoxCollisionAlgorithm));
 			if (!m_swapped)
 			{
-				return new(mem) btSphereBoxCollisionAlgorithm(0,ci,body0Wrap,body1Wrap,false);
-			} else
+				return new (mem) btSphereBoxCollisionAlgorithm(0, ci, body0Wrap, body1Wrap, false);
+			}
+			else
 			{
-				return new(mem) btSphereBoxCollisionAlgorithm(0,ci,body0Wrap,body1Wrap,true);
+				return new (mem) btSphereBoxCollisionAlgorithm(0, ci, body0Wrap, body1Wrap, true);
 			}
 		}
 	};
-
 };
 
-#endif //BT_SPHERE_BOX_COLLISION_ALGORITHM_H
-
+#endif  //BT_SPHERE_BOX_COLLISION_ALGORITHM_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereSphereCollisionAlgorithm.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereSphereCollisionAlgorithm.cpp
index 27eaec3059f..7fa0559f975 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereSphereCollisionAlgorithm.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereSphereCollisionAlgorithm.cpp
@@ -20,14 +20,14 @@ subject to the following restrictions:
 #include "BulletCollision/CollisionDispatch/btCollisionObject.h"
 #include "BulletCollision/CollisionDispatch/btCollisionObjectWrapper.h"
 
-btSphereSphereCollisionAlgorithm::btSphereSphereCollisionAlgorithm(btPersistentManifold* mf,const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* col0Wrap,const btCollisionObjectWrapper* col1Wrap)
-: btActivatingCollisionAlgorithm(ci,col0Wrap,col1Wrap),
-m_ownManifold(false),
-m_manifoldPtr(mf)
+btSphereSphereCollisionAlgorithm::btSphereSphereCollisionAlgorithm(btPersistentManifold* mf, const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* col0Wrap, const btCollisionObjectWrapper* col1Wrap)
+	: btActivatingCollisionAlgorithm(ci, col0Wrap, col1Wrap),
+	  m_ownManifold(false),
+	  m_manifoldPtr(mf)
 {
 	if (!m_manifoldPtr)
 	{
-		m_manifoldPtr = m_dispatcher->getNewManifold(col0Wrap->getCollisionObject(),col1Wrap->getCollisionObject());
+		m_manifoldPtr = m_dispatcher->getNewManifold(col0Wrap->getCollisionObject(), col1Wrap->getCollisionObject());
 		m_ownManifold = true;
 	}
 }
@@ -41,7 +41,7 @@ btSphereSphereCollisionAlgorithm::~btSphereSphereCollisionAlgorithm()
 	}
 }
 
-void btSphereSphereCollisionAlgorithm::processCollision (const btCollisionObjectWrapper* col0Wrap,const btCollisionObjectWrapper* col1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+void btSphereSphereCollisionAlgorithm::processCollision(const btCollisionObjectWrapper* col0Wrap, const btCollisionObjectWrapper* col1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	(void)dispatchInfo;
 
@@ -53,27 +53,27 @@ void btSphereSphereCollisionAlgorithm::processCollision (const btCollisionObject
 	btSphereShape* sphere0 = (btSphereShape*)col0Wrap->getCollisionShape();
 	btSphereShape* sphere1 = (btSphereShape*)col1Wrap->getCollisionShape();
 
-	btVector3 diff = col0Wrap->getWorldTransform().getOrigin()-  col1Wrap->getWorldTransform().getOrigin();
+	btVector3 diff = col0Wrap->getWorldTransform().getOrigin() - col1Wrap->getWorldTransform().getOrigin();
 	btScalar len = diff.length();
 	btScalar radius0 = sphere0->getRadius();
 	btScalar radius1 = sphere1->getRadius();
 
 #ifdef CLEAR_MANIFOLD
-	m_manifoldPtr->clearManifold(); //don't do this, it disables warmstarting
+	m_manifoldPtr->clearManifold();  //don't do this, it disables warmstarting
 #endif
 
 	///iff distance positive, don't generate a new contact
-	if ( len > (radius0+radius1+resultOut->m_closestPointDistanceThreshold))
+	if (len > (radius0 + radius1 + resultOut->m_closestPointDistanceThreshold))
 	{
 #ifndef CLEAR_MANIFOLD
 		resultOut->refreshContactPoints();
-#endif //CLEAR_MANIFOLD
+#endif  //CLEAR_MANIFOLD
 		return;
 	}
 	///distance (negative means penetration)
-	btScalar dist = len - (radius0+radius1);
+	btScalar dist = len - (radius0 + radius1);
 
-	btVector3 normalOnSurfaceB(1,0,0);
+	btVector3 normalOnSurfaceB(1, 0, 0);
 	if (len > SIMD_EPSILON)
 	{
 		normalOnSurfaceB = diff / len;
@@ -82,20 +82,18 @@ void btSphereSphereCollisionAlgorithm::processCollision (const btCollisionObject
 	///point on A (worldspace)
 	///btVector3 pos0 = col0->getWorldTransform().getOrigin() - radius0 * normalOnSurfaceB;
 	///point on B (worldspace)
-	btVector3 pos1 = col1Wrap->getWorldTransform().getOrigin() + radius1* normalOnSurfaceB;
+	btVector3 pos1 = col1Wrap->getWorldTransform().getOrigin() + radius1 * normalOnSurfaceB;
 
 	/// report a contact. internally this will be kept persistent, and contact reduction is done
-	
-	
-	resultOut->addContactPoint(normalOnSurfaceB,pos1,dist);
+
+	resultOut->addContactPoint(normalOnSurfaceB, pos1, dist);
 
 #ifndef CLEAR_MANIFOLD
 	resultOut->refreshContactPoints();
-#endif //CLEAR_MANIFOLD
-
+#endif  //CLEAR_MANIFOLD
 }
 
-btScalar btSphereSphereCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* col0,btCollisionObject* col1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+btScalar btSphereSphereCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* col0, btCollisionObject* col1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	(void)col0;
 	(void)col1;
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereSphereCollisionAlgorithm.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereSphereCollisionAlgorithm.h
index 3517a568a99..b08d0df76d8 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereSphereCollisionAlgorithm.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereSphereCollisionAlgorithm.h
@@ -28,39 +28,37 @@ class btPersistentManifold;
 /// Also provides the most basic sample for custom/user btCollisionAlgorithm
 class btSphereSphereCollisionAlgorithm : public btActivatingCollisionAlgorithm
 {
-	bool	m_ownManifold;
-	btPersistentManifold*	m_manifoldPtr;
-	
+	bool m_ownManifold;
+	btPersistentManifold* m_manifoldPtr;
+
 public:
-	btSphereSphereCollisionAlgorithm(btPersistentManifold* mf,const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* col0Wrap,const btCollisionObjectWrapper* col1Wrap);
+	btSphereSphereCollisionAlgorithm(btPersistentManifold* mf, const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* col0Wrap, const btCollisionObjectWrapper* col1Wrap);
 
 	btSphereSphereCollisionAlgorithm(const btCollisionAlgorithmConstructionInfo& ci)
 		: btActivatingCollisionAlgorithm(ci) {}
 
-	virtual void processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual void processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual	void	getAllContactManifolds(btManifoldArray&	manifoldArray)
+	virtual void getAllContactManifolds(btManifoldArray& manifoldArray)
 	{
 		if (m_manifoldPtr && m_ownManifold)
 		{
 			manifoldArray.push_back(m_manifoldPtr);
 		}
 	}
-	
+
 	virtual ~btSphereSphereCollisionAlgorithm();
 
-	struct CreateFunc :public 	btCollisionAlgorithmCreateFunc
+	struct CreateFunc : public btCollisionAlgorithmCreateFunc
 	{
-		virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* col0Wrap,const btCollisionObjectWrapper* col1Wrap)
+		virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* col0Wrap, const btCollisionObjectWrapper* col1Wrap)
 		{
 			void* mem = ci.m_dispatcher1->allocateCollisionAlgorithm(sizeof(btSphereSphereCollisionAlgorithm));
-			return new(mem) btSphereSphereCollisionAlgorithm(0,ci,col0Wrap,col1Wrap);
+			return new (mem) btSphereSphereCollisionAlgorithm(0, ci, col0Wrap, col1Wrap);
 		}
 	};
-
 };
 
-#endif //BT_SPHERE_SPHERE_COLLISION_ALGORITHM_H
-
+#endif  //BT_SPHERE_SPHERE_COLLISION_ALGORITHM_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereTriangleCollisionAlgorithm.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereTriangleCollisionAlgorithm.cpp
index 86d4e744009..1bc3056c013 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereTriangleCollisionAlgorithm.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereTriangleCollisionAlgorithm.cpp
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btSphereTriangleCollisionAlgorithm.h"
 #include "BulletCollision/CollisionDispatch/btCollisionDispatcher.h"
 #include "BulletCollision/CollisionShapes/btSphereShape.h"
@@ -21,15 +20,15 @@ subject to the following restrictions:
 #include "SphereTriangleDetector.h"
 #include "BulletCollision/CollisionDispatch/btCollisionObjectWrapper.h"
 
-btSphereTriangleCollisionAlgorithm::btSphereTriangleCollisionAlgorithm(btPersistentManifold* mf,const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,bool swapped)
-: btActivatingCollisionAlgorithm(ci,body0Wrap,body1Wrap),
-m_ownManifold(false),
-m_manifoldPtr(mf),
-m_swapped(swapped)
+btSphereTriangleCollisionAlgorithm::btSphereTriangleCollisionAlgorithm(btPersistentManifold* mf, const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, bool swapped)
+	: btActivatingCollisionAlgorithm(ci, body0Wrap, body1Wrap),
+	  m_ownManifold(false),
+	  m_manifoldPtr(mf),
+	  m_swapped(swapped)
 {
 	if (!m_manifoldPtr)
 	{
-		m_manifoldPtr = m_dispatcher->getNewManifold(body0Wrap->getCollisionObject(),body1Wrap->getCollisionObject());
+		m_manifoldPtr = m_dispatcher->getNewManifold(body0Wrap->getCollisionObject(), body1Wrap->getCollisionObject());
 		m_ownManifold = true;
 	}
 }
@@ -43,36 +42,35 @@ btSphereTriangleCollisionAlgorithm::~btSphereTriangleCollisionAlgorithm()
 	}
 }
 
-void btSphereTriangleCollisionAlgorithm::processCollision (const btCollisionObjectWrapper* col0Wrap,const btCollisionObjectWrapper* col1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+void btSphereTriangleCollisionAlgorithm::processCollision(const btCollisionObjectWrapper* col0Wrap, const btCollisionObjectWrapper* col1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	if (!m_manifoldPtr)
 		return;
 
-	const btCollisionObjectWrapper* sphereObjWrap = m_swapped? col1Wrap : col0Wrap;
-	const btCollisionObjectWrapper* triObjWrap = m_swapped? col0Wrap : col1Wrap;
+	const btCollisionObjectWrapper* sphereObjWrap = m_swapped ? col1Wrap : col0Wrap;
+	const btCollisionObjectWrapper* triObjWrap = m_swapped ? col0Wrap : col1Wrap;
 
 	btSphereShape* sphere = (btSphereShape*)sphereObjWrap->getCollisionShape();
 	btTriangleShape* triangle = (btTriangleShape*)triObjWrap->getCollisionShape();
-	
+
 	/// report a contact. internally this will be kept persistent, and contact reduction is done
 	resultOut->setPersistentManifold(m_manifoldPtr);
-	SphereTriangleDetector detector(sphere,triangle, m_manifoldPtr->getContactBreakingThreshold()+ resultOut->m_closestPointDistanceThreshold);
-	
+	SphereTriangleDetector detector(sphere, triangle, m_manifoldPtr->getContactBreakingThreshold() + resultOut->m_closestPointDistanceThreshold);
+
 	btDiscreteCollisionDetectorInterface::ClosestPointInput input;
-	input.m_maximumDistanceSquared = btScalar(BT_LARGE_FLOAT);///@todo: tighter bounds
+	input.m_maximumDistanceSquared = btScalar(BT_LARGE_FLOAT);  ///@todo: tighter bounds
 	input.m_transformA = sphereObjWrap->getWorldTransform();
 	input.m_transformB = triObjWrap->getWorldTransform();
 
 	bool swapResults = m_swapped;
 
-	detector.getClosestPoints(input,*resultOut,dispatchInfo.m_debugDraw,swapResults);
+	detector.getClosestPoints(input, *resultOut, dispatchInfo.m_debugDraw, swapResults);
 
 	if (m_ownManifold)
 		resultOut->refreshContactPoints();
-	
 }
 
-btScalar btSphereTriangleCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* col0,btCollisionObject* col1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+btScalar btSphereTriangleCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* col0, btCollisionObject* col1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	(void)resultOut;
 	(void)dispatchInfo;
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereTriangleCollisionAlgorithm.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereTriangleCollisionAlgorithm.h
index 6b6e39a72b6..d660222f164 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereTriangleCollisionAlgorithm.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btSphereTriangleCollisionAlgorithm.h
@@ -27,43 +27,39 @@ class btPersistentManifold;
 /// Also provides the most basic sample for custom/user btCollisionAlgorithm
 class btSphereTriangleCollisionAlgorithm : public btActivatingCollisionAlgorithm
 {
-	bool	m_ownManifold;
-	btPersistentManifold*	m_manifoldPtr;
-	bool	m_swapped;
-	
+	bool m_ownManifold;
+	btPersistentManifold* m_manifoldPtr;
+	bool m_swapped;
+
 public:
-	btSphereTriangleCollisionAlgorithm(btPersistentManifold* mf,const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,bool swapped);
+	btSphereTriangleCollisionAlgorithm(btPersistentManifold* mf, const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, bool swapped);
 
 	btSphereTriangleCollisionAlgorithm(const btCollisionAlgorithmConstructionInfo& ci)
 		: btActivatingCollisionAlgorithm(ci) {}
 
-	virtual void processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual void processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual	void	getAllContactManifolds(btManifoldArray&	manifoldArray)
+	virtual void getAllContactManifolds(btManifoldArray& manifoldArray)
 	{
 		if (m_manifoldPtr && m_ownManifold)
 		{
 			manifoldArray.push_back(m_manifoldPtr);
 		}
 	}
-	
+
 	virtual ~btSphereTriangleCollisionAlgorithm();
 
-	struct CreateFunc :public 	btCollisionAlgorithmCreateFunc
+	struct CreateFunc : public btCollisionAlgorithmCreateFunc
 	{
-		
-		virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+		virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 		{
-			
 			void* mem = ci.m_dispatcher1->allocateCollisionAlgorithm(sizeof(btSphereTriangleCollisionAlgorithm));
 
-			return new(mem) btSphereTriangleCollisionAlgorithm(ci.m_manifold,ci,body0Wrap,body1Wrap,m_swapped);
+			return new (mem) btSphereTriangleCollisionAlgorithm(ci.m_manifold, ci, body0Wrap, body1Wrap, m_swapped);
 		}
 	};
-
 };
 
-#endif //BT_SPHERE_TRIANGLE_COLLISION_ALGORITHM_H
-
+#endif  //BT_SPHERE_TRIANGLE_COLLISION_ALGORITHM_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btUnionFind.cpp b/thirdparty/bullet/BulletCollision/CollisionDispatch/btUnionFind.cpp
index 5222933595d..816bf1e6adc 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btUnionFind.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btUnionFind.cpp
@@ -15,68 +15,60 @@ subject to the following restrictions:
 
 #include "btUnionFind.h"
 
-
-
 btUnionFind::~btUnionFind()
 {
 	Free();
-
 }
 
 btUnionFind::btUnionFind()
-{ 
-
+{
 }
 
-void	btUnionFind::allocate(int N)
+void btUnionFind::allocate(int N)
 {
 	m_elements.resize(N);
 }
-void	btUnionFind::Free()
+void btUnionFind::Free()
 {
 	m_elements.clear();
 }
 
-
-void	btUnionFind::reset(int N)
+void btUnionFind::reset(int N)
 {
 	allocate(N);
 
-	for (int i = 0; i < N; i++) 
-	{ 
-		m_elements[i].m_id = i; m_elements[i].m_sz = 1; 
-	} 
+	for (int i = 0; i < N; i++)
+	{
+		m_elements[i].m_id = i;
+		m_elements[i].m_sz = 1;
+	}
 }
 
-
 class btUnionFindElementSortPredicate
 {
-	public:
-
-		bool operator() ( const btElement& lhs, const btElement& rhs ) const
-		{
-			return lhs.m_id < rhs.m_id;
-		}
+public:
+	bool operator()(const btElement& lhs, const btElement& rhs) const
+	{
+		return lhs.m_id < rhs.m_id;
+	}
 };
 
 ///this is a special operation, destroying the content of btUnionFind.
 ///it sorts the elements, based on island id, in order to make it easy to iterate over islands
-void	btUnionFind::sortIslands()
+void btUnionFind::sortIslands()
 {
-
 	//first store the original body index, and islandId
 	int numElements = m_elements.size();
-	
-	for (int i=0;i<numElements;i++)
+
+	for (int i = 0; i < numElements; i++)
 	{
 		m_elements[i].m_id = find(i);
 #ifndef STATIC_SIMULATION_ISLAND_OPTIMIZATION
 		m_elements[i].m_sz = i;
-#endif //STATIC_SIMULATION_ISLAND_OPTIMIZATION
+#endif  //STATIC_SIMULATION_ISLAND_OPTIMIZATION
 	}
-	
-	 // Sort the vector using predicate and std::sort
-	  //std::sort(m_elements.begin(), m_elements.end(), btUnionFindElementSortPredicate);
-	  m_elements.quickSort(btUnionFindElementSortPredicate());
 
+	// Sort the vector using predicate and std::sort
+	//std::sort(m_elements.begin(), m_elements.end(), btUnionFindElementSortPredicate);
+	m_elements.quickSort(btUnionFindElementSortPredicate());
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionDispatch/btUnionFind.h b/thirdparty/bullet/BulletCollision/CollisionDispatch/btUnionFind.h
index ef2a29202f7..d422ef55eb3 100644
--- a/thirdparty/bullet/BulletCollision/CollisionDispatch/btUnionFind.h
+++ b/thirdparty/bullet/BulletCollision/CollisionDispatch/btUnionFind.h
@@ -23,107 +23,101 @@ subject to the following restrictions:
 ///see for discussion of static island optimizations by Vroonsh here: http://code.google.com/p/bullet/issues/detail?id=406
 #define STATIC_SIMULATION_ISLAND_OPTIMIZATION 1
 
-struct	btElement
+struct btElement
 {
-	int	m_id;
-	int	m_sz;
+	int m_id;
+	int m_sz;
 };
 
 ///UnionFind calculates connected subsets
 // Implements weighted Quick Union with path compression
 // optimization: could use short ints instead of ints (halving memory, would limit the number of rigid bodies to 64k, sounds reasonable)
 class btUnionFind
-  {
-    private:
-		btAlignedObjectArray<btElement>	m_elements;
+{
+private:
+	btAlignedObjectArray<btElement> m_elements;
 
-    public:
-	  
-		btUnionFind();
-		~btUnionFind();
+public:
+	btUnionFind();
+	~btUnionFind();
 
-	
-		//this is a special operation, destroying the content of btUnionFind.
-		//it sorts the elements, based on island id, in order to make it easy to iterate over islands
-		void	sortIslands();
+	//this is a special operation, destroying the content of btUnionFind.
+	//it sorts the elements, based on island id, in order to make it easy to iterate over islands
+	void sortIslands();
 
-	  void	reset(int N);
+	void reset(int N);
 
-	  SIMD_FORCE_INLINE int	getNumElements() const
-	  {
-		  return int(m_elements.size());
-	  }
-	  SIMD_FORCE_INLINE bool  isRoot(int x) const
-	  {
-		  return (x == m_elements[x].m_id);
-	  }
+	SIMD_FORCE_INLINE int getNumElements() const
+	{
+		return int(m_elements.size());
+	}
+	SIMD_FORCE_INLINE bool isRoot(int x) const
+	{
+		return (x == m_elements[x].m_id);
+	}
 
-	  btElement&	getElement(int index)
-	  {
-		  return m_elements[index];
-	  }
-	  const btElement& getElement(int index) const
-	  {
-		  return m_elements[index];
-	  }
-   
-	  void	allocate(int N);
-	  void	Free();
+	btElement& getElement(int index)
+	{
+		return m_elements[index];
+	}
+	const btElement& getElement(int index) const
+	{
+		return m_elements[index];
+	}
 
+	void allocate(int N);
+	void Free();
 
+	int find(int p, int q)
+	{
+		return (find(p) == find(q));
+	}
 
-
-	  int find(int p, int q)
-		{ 
-			return (find(p) == find(q)); 
-		}
-
-		void unite(int p, int q)
-		{
-			int i = find(p), j = find(q);
-			if (i == j) 
-				return;
+	void unite(int p, int q)
+	{
+		int i = find(p), j = find(q);
+		if (i == j)
+			return;
 
 #ifndef USE_PATH_COMPRESSION
-			//weighted quick union, this keeps the 'trees' balanced, and keeps performance of unite O( log(n) )
-			if (m_elements[i].m_sz < m_elements[j].m_sz)
-			{ 
-				m_elements[i].m_id = j; m_elements[j].m_sz += m_elements[i].m_sz; 
-			}
-			else 
-			{ 
-				m_elements[j].m_id = i; m_elements[i].m_sz += m_elements[j].m_sz; 
-			}
-#else
-			m_elements[i].m_id = j; m_elements[j].m_sz += m_elements[i].m_sz; 
-#endif //USE_PATH_COMPRESSION
+		//weighted quick union, this keeps the 'trees' balanced, and keeps performance of unite O( log(n) )
+		if (m_elements[i].m_sz < m_elements[j].m_sz)
+		{
+			m_elements[i].m_id = j;
+			m_elements[j].m_sz += m_elements[i].m_sz;
 		}
+		else
+		{
+			m_elements[j].m_id = i;
+			m_elements[i].m_sz += m_elements[j].m_sz;
+		}
+#else
+		m_elements[i].m_id = j;
+		m_elements[j].m_sz += m_elements[i].m_sz;
+#endif  //USE_PATH_COMPRESSION
+	}
 
-		int find(int x)
-		{ 
+	int find(int x)
+	{
+		//btAssert(x < m_N);
+		//btAssert(x >= 0);
+
+		while (x != m_elements[x].m_id)
+		{
+			//not really a reason not to use path compression, and it flattens the trees/improves find performance dramatically
+
+#ifdef USE_PATH_COMPRESSION
+			const btElement* elementPtr = &m_elements[m_elements[x].m_id];
+			m_elements[x].m_id = elementPtr->m_id;
+			x = elementPtr->m_id;
+#else  //
+			x = m_elements[x].m_id;
+#endif
 			//btAssert(x < m_N);
 			//btAssert(x >= 0);
-
-			while (x != m_elements[x].m_id) 
-			{
-		//not really a reason not to use path compression, and it flattens the trees/improves find performance dramatically
-	
-		#ifdef USE_PATH_COMPRESSION
-				const btElement* elementPtr = &m_elements[m_elements[x].m_id];
-				m_elements[x].m_id = elementPtr->m_id;
-				x = elementPtr->m_id;			
-		#else//
-				x = m_elements[x].m_id;
-		#endif		
-				//btAssert(x < m_N);
-				//btAssert(x >= 0);
-
-			}
-			return x; 
 		}
+		return x;
+	}
+};
 
-
-  };
-
-
-#endif //BT_UNION_FIND_H
+#endif  //BT_UNION_FIND_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btBox2dShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btBox2dShape.cpp
index ecce028c2e6..a3d8075dafa 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btBox2dShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btBox2dShape.cpp
@@ -15,28 +15,23 @@ subject to the following restrictions:
 
 #include "btBox2dShape.h"
 
+//{
 
-//{ 
-
-
-void btBox2dShape::getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const
+void btBox2dShape::getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
 {
-	btTransformAabb(getHalfExtentsWithoutMargin(),getMargin(),t,aabbMin,aabbMax);
+	btTransformAabb(getHalfExtentsWithoutMargin(), getMargin(), t, aabbMin, aabbMax);
 }
 
-
-void	btBox2dShape::calculateLocalInertia(btScalar mass,btVector3& inertia) const
+void btBox2dShape::calculateLocalInertia(btScalar mass, btVector3& inertia) const
 {
 	//btScalar margin = btScalar(0.);
 	btVector3 halfExtents = getHalfExtentsWithMargin();
 
-	btScalar lx=btScalar(2.)*(halfExtents.x());
-	btScalar ly=btScalar(2.)*(halfExtents.y());
-	btScalar lz=btScalar(2.)*(halfExtents.z());
-
-	inertia.setValue(mass/(btScalar(12.0)) * (ly*ly + lz*lz),
-					mass/(btScalar(12.0)) * (lx*lx + lz*lz),
-					mass/(btScalar(12.0)) * (lx*lx + ly*ly));
+	btScalar lx = btScalar(2.) * (halfExtents.x());
+	btScalar ly = btScalar(2.) * (halfExtents.y());
+	btScalar lz = btScalar(2.) * (halfExtents.z());
 
+	inertia.setValue(mass / (btScalar(12.0)) * (ly * ly + lz * lz),
+					 mass / (btScalar(12.0)) * (lx * lx + lz * lz),
+					 mass / (btScalar(12.0)) * (lx * lx + ly * ly));
 }
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btBox2dShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btBox2dShape.h
index 22bee4f2c83..7e085f9e2ea 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btBox2dShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btBox2dShape.h
@@ -23,9 +23,9 @@ subject to the following restrictions:
 #include "LinearMath/btMinMax.h"
 
 ///The btBox2dShape is a box primitive around the origin, its sides axis aligned with length specified by half extents, in local shape coordinates. When used as part of a btCollisionObject or btRigidBody it will be an oriented box in world space.
-ATTRIBUTE_ALIGNED16(class) btBox2dShape: public btPolyhedralConvexShape
+ATTRIBUTE_ALIGNED16(class)
+btBox2dShape : public btPolyhedralConvexShape
 {
-
 	//btVector3	m_boxHalfExtents1; //use m_implicitShapeDimensions instead
 
 	btVector3 m_centroid;
@@ -33,79 +33,75 @@ ATTRIBUTE_ALIGNED16(class) btBox2dShape: public btPolyhedralConvexShape
 	btVector3 m_normals[4];
 
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
 	btVector3 getHalfExtentsWithMargin() const
 	{
 		btVector3 halfExtents = getHalfExtentsWithoutMargin();
-		btVector3 margin(getMargin(),getMargin(),getMargin());
+		btVector3 margin(getMargin(), getMargin(), getMargin());
 		halfExtents += margin;
 		return halfExtents;
 	}
-	
+
 	const btVector3& getHalfExtentsWithoutMargin() const
 	{
-		return m_implicitShapeDimensions;//changed in Bullet 2.63: assume the scaling and margin are included
+		return m_implicitShapeDimensions;  //changed in Bullet 2.63: assume the scaling and margin are included
 	}
-	
 
-	virtual btVector3	localGetSupportingVertex(const btVector3& vec) const
+	virtual btVector3 localGetSupportingVertex(const btVector3& vec) const
 	{
 		btVector3 halfExtents = getHalfExtentsWithoutMargin();
-		btVector3 margin(getMargin(),getMargin(),getMargin());
+		btVector3 margin(getMargin(), getMargin(), getMargin());
 		halfExtents += margin;
-		
+
 		return btVector3(btFsels(vec.x(), halfExtents.x(), -halfExtents.x()),
-			btFsels(vec.y(), halfExtents.y(), -halfExtents.y()),
-			btFsels(vec.z(), halfExtents.z(), -halfExtents.z()));
+						 btFsels(vec.y(), halfExtents.y(), -halfExtents.y()),
+						 btFsels(vec.z(), halfExtents.z(), -halfExtents.z()));
 	}
 
-	SIMD_FORCE_INLINE  btVector3	localGetSupportingVertexWithoutMargin(const btVector3& vec)const
+	SIMD_FORCE_INLINE btVector3 localGetSupportingVertexWithoutMargin(const btVector3& vec) const
 	{
 		const btVector3& halfExtents = getHalfExtentsWithoutMargin();
-		
+
 		return btVector3(btFsels(vec.x(), halfExtents.x(), -halfExtents.x()),
-			btFsels(vec.y(), halfExtents.y(), -halfExtents.y()),
-			btFsels(vec.z(), halfExtents.z(), -halfExtents.z()));
+						 btFsels(vec.y(), halfExtents.y(), -halfExtents.y()),
+						 btFsels(vec.z(), halfExtents.z(), -halfExtents.z()));
 	}
 
-	virtual void	batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const
+	virtual void batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const
 	{
 		const btVector3& halfExtents = getHalfExtentsWithoutMargin();
-	
-		for (int i=0;i<numVectors;i++)
+
+		for (int i = 0; i < numVectors; i++)
 		{
 			const btVector3& vec = vectors[i];
 			supportVerticesOut[i].setValue(btFsels(vec.x(), halfExtents.x(), -halfExtents.x()),
-				btFsels(vec.y(), halfExtents.y(), -halfExtents.y()),
-				btFsels(vec.z(), halfExtents.z(), -halfExtents.z())); 
+										   btFsels(vec.y(), halfExtents.y(), -halfExtents.y()),
+										   btFsels(vec.z(), halfExtents.z(), -halfExtents.z()));
 		}
-
 	}
 
-
 	///a btBox2dShape is a flat 2D box in the X-Y plane (Z extents are zero)
-	btBox2dShape( const btVector3& boxHalfExtents) 
+	btBox2dShape(const btVector3& boxHalfExtents)
 		: btPolyhedralConvexShape(),
-		m_centroid(0,0,0)
+		  m_centroid(0, 0, 0)
 	{
-		m_vertices[0].setValue(-boxHalfExtents.getX(),-boxHalfExtents.getY(),0);
-		m_vertices[1].setValue(boxHalfExtents.getX(),-boxHalfExtents.getY(),0);
-		m_vertices[2].setValue(boxHalfExtents.getX(),boxHalfExtents.getY(),0);
-		m_vertices[3].setValue(-boxHalfExtents.getX(),boxHalfExtents.getY(),0);
+		m_vertices[0].setValue(-boxHalfExtents.getX(), -boxHalfExtents.getY(), 0);
+		m_vertices[1].setValue(boxHalfExtents.getX(), -boxHalfExtents.getY(), 0);
+		m_vertices[2].setValue(boxHalfExtents.getX(), boxHalfExtents.getY(), 0);
+		m_vertices[3].setValue(-boxHalfExtents.getX(), boxHalfExtents.getY(), 0);
 
-		m_normals[0].setValue(0,-1,0);
-		m_normals[1].setValue(1,0,0);
-		m_normals[2].setValue(0,1,0);
-		m_normals[3].setValue(-1,0,0);
+		m_normals[0].setValue(0, -1, 0);
+		m_normals[1].setValue(1, 0, 0);
+		m_normals[2].setValue(0, 1, 0);
+		m_normals[3].setValue(-1, 0, 0);
 
 		btScalar minDimension = boxHalfExtents.getX();
-		if (minDimension>boxHalfExtents.getY())
+		if (minDimension > boxHalfExtents.getY())
 			minDimension = boxHalfExtents.getY();
 
 		m_shapeType = BOX_2D_SHAPE_PROXYTYPE;
-		btVector3 margin(getMargin(),getMargin(),getMargin());
+		btVector3 margin(getMargin(), getMargin(), getMargin());
 		m_implicitShapeDimensions = (boxHalfExtents * m_localScaling) - margin;
 
 		setSafeMargin(minDimension);
@@ -114,42 +110,34 @@ public:
 	virtual void setMargin(btScalar collisionMargin)
 	{
 		//correct the m_implicitShapeDimensions for the margin
-		btVector3 oldMargin(getMargin(),getMargin(),getMargin());
-		btVector3 implicitShapeDimensionsWithMargin = m_implicitShapeDimensions+oldMargin;
-		
-		btConvexInternalShape::setMargin(collisionMargin);
-		btVector3 newMargin(getMargin(),getMargin(),getMargin());
-		m_implicitShapeDimensions = implicitShapeDimensionsWithMargin - newMargin;
+		btVector3 oldMargin(getMargin(), getMargin(), getMargin());
+		btVector3 implicitShapeDimensionsWithMargin = m_implicitShapeDimensions + oldMargin;
 
+		btConvexInternalShape::setMargin(collisionMargin);
+		btVector3 newMargin(getMargin(), getMargin(), getMargin());
+		m_implicitShapeDimensions = implicitShapeDimensionsWithMargin - newMargin;
 	}
-	virtual void	setLocalScaling(const btVector3& scaling)
+	virtual void setLocalScaling(const btVector3& scaling)
 	{
-		btVector3 oldMargin(getMargin(),getMargin(),getMargin());
-		btVector3 implicitShapeDimensionsWithMargin = m_implicitShapeDimensions+oldMargin;
+		btVector3 oldMargin(getMargin(), getMargin(), getMargin());
+		btVector3 implicitShapeDimensionsWithMargin = m_implicitShapeDimensions + oldMargin;
 		btVector3 unScaledImplicitShapeDimensionsWithMargin = implicitShapeDimensionsWithMargin / m_localScaling;
 
 		btConvexInternalShape::setLocalScaling(scaling);
 
 		m_implicitShapeDimensions = (unScaledImplicitShapeDimensionsWithMargin * m_localScaling) - oldMargin;
-
 	}
 
-	virtual void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const;
+	virtual void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
 
-	
+	virtual void calculateLocalInertia(btScalar mass, btVector3 & inertia) const;
 
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
-
-
-
-
-
-	int	getVertexCount() const
+	int getVertexCount() const
 	{
 		return 4;
 	}
 
-	virtual int getNumVertices()const
+	virtual int getNumVertices() const
 	{
 		return 4;
 	}
@@ -164,82 +152,70 @@ public:
 		return &m_normals[0];
 	}
 
-
-
-
-
-
-
-	virtual void getPlane(btVector3& planeNormal,btVector3& planeSupport,int i ) const
+	virtual void getPlane(btVector3 & planeNormal, btVector3 & planeSupport, int i) const
 	{
 		//this plane might not be aligned...
-		btVector4 plane ;
-		getPlaneEquation(plane,i);
-		planeNormal = btVector3(plane.getX(),plane.getY(),plane.getZ());
+		btVector4 plane;
+		getPlaneEquation(plane, i);
+		planeNormal = btVector3(plane.getX(), plane.getY(), plane.getZ());
 		planeSupport = localGetSupportingVertex(-planeNormal);
 	}
 
-
 	const btVector3& getCentroid() const
 	{
 		return m_centroid;
 	}
-	
+
 	virtual int getNumPlanes() const
 	{
 		return 6;
-	}	
-	
-	
+	}
 
 	virtual int getNumEdges() const
 	{
 		return 12;
 	}
 
-
-	virtual void getVertex(int i,btVector3& vtx) const
+	virtual void getVertex(int i, btVector3& vtx) const
 	{
 		btVector3 halfExtents = getHalfExtentsWithoutMargin();
 
 		vtx = btVector3(
-				halfExtents.x() * (1-(i&1)) - halfExtents.x() * (i&1),
-				halfExtents.y() * (1-((i&2)>>1)) - halfExtents.y() * ((i&2)>>1),
-				halfExtents.z() * (1-((i&4)>>2)) - halfExtents.z() * ((i&4)>>2));
+			halfExtents.x() * (1 - (i & 1)) - halfExtents.x() * (i & 1),
+			halfExtents.y() * (1 - ((i & 2) >> 1)) - halfExtents.y() * ((i & 2) >> 1),
+			halfExtents.z() * (1 - ((i & 4) >> 2)) - halfExtents.z() * ((i & 4) >> 2));
 	}
-	
 
-	virtual void	getPlaneEquation(btVector4& plane,int i) const
+	virtual void getPlaneEquation(btVector4 & plane, int i) const
 	{
 		btVector3 halfExtents = getHalfExtentsWithoutMargin();
 
 		switch (i)
 		{
-		case 0:
-			plane.setValue(btScalar(1.),btScalar(0.),btScalar(0.),-halfExtents.x());
-			break;
-		case 1:
-			plane.setValue(btScalar(-1.),btScalar(0.),btScalar(0.),-halfExtents.x());
-			break;
-		case 2:
-			plane.setValue(btScalar(0.),btScalar(1.),btScalar(0.),-halfExtents.y());
-			break;
-		case 3:
-			plane.setValue(btScalar(0.),btScalar(-1.),btScalar(0.),-halfExtents.y());
-			break;
-		case 4:
-			plane.setValue(btScalar(0.),btScalar(0.),btScalar(1.),-halfExtents.z());
-			break;
-		case 5:
-			plane.setValue(btScalar(0.),btScalar(0.),btScalar(-1.),-halfExtents.z());
-			break;
-		default:
-			btAssert(0);
+			case 0:
+				plane.setValue(btScalar(1.), btScalar(0.), btScalar(0.), -halfExtents.x());
+				break;
+			case 1:
+				plane.setValue(btScalar(-1.), btScalar(0.), btScalar(0.), -halfExtents.x());
+				break;
+			case 2:
+				plane.setValue(btScalar(0.), btScalar(1.), btScalar(0.), -halfExtents.y());
+				break;
+			case 3:
+				plane.setValue(btScalar(0.), btScalar(-1.), btScalar(0.), -halfExtents.y());
+				break;
+			case 4:
+				plane.setValue(btScalar(0.), btScalar(0.), btScalar(1.), -halfExtents.z());
+				break;
+			case 5:
+				plane.setValue(btScalar(0.), btScalar(0.), btScalar(-1.), -halfExtents.z());
+				break;
+			default:
+				btAssert(0);
 		}
 	}
 
-	
-	virtual void getEdge(int i,btVector3& pa,btVector3& pb) const
+	virtual void getEdge(int i, btVector3& pa, btVector3& pb) const
 	//virtual void getEdge(int i,Edge& edge) const
 	{
 		int edgeVert0 = 0;
@@ -247,126 +223,117 @@ public:
 
 		switch (i)
 		{
-		case 0:
+			case 0:
 				edgeVert0 = 0;
 				edgeVert1 = 1;
-			break;
-		case 1:
+				break;
+			case 1:
 				edgeVert0 = 0;
 				edgeVert1 = 2;
-			break;
-		case 2:
-			edgeVert0 = 1;
-			edgeVert1 = 3;
+				break;
+			case 2:
+				edgeVert0 = 1;
+				edgeVert1 = 3;
 
-			break;
-		case 3:
-			edgeVert0 = 2;
-			edgeVert1 = 3;
-			break;
-		case 4:
-			edgeVert0 = 0;
-			edgeVert1 = 4;
-			break;
-		case 5:
-			edgeVert0 = 1;
-			edgeVert1 = 5;
-
-			break;
-		case 6:
-			edgeVert0 = 2;
-			edgeVert1 = 6;
-			break;
-		case 7:
-			edgeVert0 = 3;
-			edgeVert1 = 7;
-			break;
-		case 8:
-			edgeVert0 = 4;
-			edgeVert1 = 5;
-			break;
-		case 9:
-			edgeVert0 = 4;
-			edgeVert1 = 6;
-			break;
-		case 10:
-			edgeVert0 = 5;
-			edgeVert1 = 7;
-			break;
-		case 11:
-			edgeVert0 = 6;
-			edgeVert1 = 7;
-			break;
-		default:
-			btAssert(0);
+				break;
+			case 3:
+				edgeVert0 = 2;
+				edgeVert1 = 3;
+				break;
+			case 4:
+				edgeVert0 = 0;
+				edgeVert1 = 4;
+				break;
+			case 5:
+				edgeVert0 = 1;
+				edgeVert1 = 5;
 
+				break;
+			case 6:
+				edgeVert0 = 2;
+				edgeVert1 = 6;
+				break;
+			case 7:
+				edgeVert0 = 3;
+				edgeVert1 = 7;
+				break;
+			case 8:
+				edgeVert0 = 4;
+				edgeVert1 = 5;
+				break;
+			case 9:
+				edgeVert0 = 4;
+				edgeVert1 = 6;
+				break;
+			case 10:
+				edgeVert0 = 5;
+				edgeVert1 = 7;
+				break;
+			case 11:
+				edgeVert0 = 6;
+				edgeVert1 = 7;
+				break;
+			default:
+				btAssert(0);
 		}
 
-		getVertex(edgeVert0,pa );
-		getVertex(edgeVert1,pb );
+		getVertex(edgeVert0, pa);
+		getVertex(edgeVert1, pb);
 	}
 
-
-
-
-	
-	virtual	bool isInside(const btVector3& pt,btScalar tolerance) const
+	virtual bool isInside(const btVector3& pt, btScalar tolerance) const
 	{
 		btVector3 halfExtents = getHalfExtentsWithoutMargin();
 
 		//btScalar minDist = 2*tolerance;
-		
-		bool result =	(pt.x() <= (halfExtents.x()+tolerance)) &&
-						(pt.x() >= (-halfExtents.x()-tolerance)) &&
-						(pt.y() <= (halfExtents.y()+tolerance)) &&
-						(pt.y() >= (-halfExtents.y()-tolerance)) &&
-						(pt.z() <= (halfExtents.z()+tolerance)) &&
-						(pt.z() >= (-halfExtents.z()-tolerance));
-		
+
+		bool result = (pt.x() <= (halfExtents.x() + tolerance)) &&
+					  (pt.x() >= (-halfExtents.x() - tolerance)) &&
+					  (pt.y() <= (halfExtents.y() + tolerance)) &&
+					  (pt.y() >= (-halfExtents.y() - tolerance)) &&
+					  (pt.z() <= (halfExtents.z() + tolerance)) &&
+					  (pt.z() >= (-halfExtents.z() - tolerance));
+
 		return result;
 	}
 
-
 	//debugging
-	virtual const char*	getName()const
+	virtual const char* getName() const
 	{
 		return "Box2d";
 	}
 
-	virtual int		getNumPreferredPenetrationDirections() const
+	virtual int getNumPreferredPenetrationDirections() const
 	{
 		return 6;
 	}
-	
-	virtual void	getPreferredPenetrationDirection(int index, btVector3& penetrationVector) const
+
+	virtual void getPreferredPenetrationDirection(int index, btVector3& penetrationVector) const
 	{
 		switch (index)
 		{
-		case 0:
-			penetrationVector.setValue(btScalar(1.),btScalar(0.),btScalar(0.));
-			break;
-		case 1:
-			penetrationVector.setValue(btScalar(-1.),btScalar(0.),btScalar(0.));
-			break;
-		case 2:
-			penetrationVector.setValue(btScalar(0.),btScalar(1.),btScalar(0.));
-			break;
-		case 3:
-			penetrationVector.setValue(btScalar(0.),btScalar(-1.),btScalar(0.));
-			break;
-		case 4:
-			penetrationVector.setValue(btScalar(0.),btScalar(0.),btScalar(1.));
-			break;
-		case 5:
-			penetrationVector.setValue(btScalar(0.),btScalar(0.),btScalar(-1.));
-			break;
-		default:
-			btAssert(0);
+			case 0:
+				penetrationVector.setValue(btScalar(1.), btScalar(0.), btScalar(0.));
+				break;
+			case 1:
+				penetrationVector.setValue(btScalar(-1.), btScalar(0.), btScalar(0.));
+				break;
+			case 2:
+				penetrationVector.setValue(btScalar(0.), btScalar(1.), btScalar(0.));
+				break;
+			case 3:
+				penetrationVector.setValue(btScalar(0.), btScalar(-1.), btScalar(0.));
+				break;
+			case 4:
+				penetrationVector.setValue(btScalar(0.), btScalar(0.), btScalar(1.));
+				break;
+			case 5:
+				penetrationVector.setValue(btScalar(0.), btScalar(0.), btScalar(-1.));
+				break;
+			default:
+				btAssert(0);
 		}
 	}
-
 };
 
-#endif //BT_OBB_BOX_2D_SHAPE_H
-
-
+#endif  //BT_OBB_BOX_2D_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btBoxShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btBoxShape.cpp
index 72eeb38911f..cb91d023e40 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btBoxShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btBoxShape.cpp
@@ -14,38 +14,32 @@ subject to the following restrictions:
 */
 #include "btBoxShape.h"
 
-btBoxShape::btBoxShape( const btVector3& boxHalfExtents) 
-: btPolyhedralConvexShape()
+btBoxShape::btBoxShape(const btVector3& boxHalfExtents)
+	: btPolyhedralConvexShape()
 {
 	m_shapeType = BOX_SHAPE_PROXYTYPE;
 
-	btVector3 margin(getMargin(),getMargin(),getMargin());
+	btVector3 margin(getMargin(), getMargin(), getMargin());
 	m_implicitShapeDimensions = (boxHalfExtents * m_localScaling) - margin;
 
 	setSafeMargin(boxHalfExtents);
 };
 
-
-
-
-void btBoxShape::getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const
+void btBoxShape::getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
 {
-	btTransformAabb(getHalfExtentsWithoutMargin(),getMargin(),t,aabbMin,aabbMax);
+	btTransformAabb(getHalfExtentsWithoutMargin(), getMargin(), t, aabbMin, aabbMax);
 }
 
-
-void	btBoxShape::calculateLocalInertia(btScalar mass,btVector3& inertia) const
+void btBoxShape::calculateLocalInertia(btScalar mass, btVector3& inertia) const
 {
 	//btScalar margin = btScalar(0.);
 	btVector3 halfExtents = getHalfExtentsWithMargin();
 
-	btScalar lx=btScalar(2.)*(halfExtents.x());
-	btScalar ly=btScalar(2.)*(halfExtents.y());
-	btScalar lz=btScalar(2.)*(halfExtents.z());
-
-	inertia.setValue(mass/(btScalar(12.0)) * (ly*ly + lz*lz),
-					mass/(btScalar(12.0)) * (lx*lx + lz*lz),
-					mass/(btScalar(12.0)) * (lx*lx + ly*ly));
+	btScalar lx = btScalar(2.) * (halfExtents.x());
+	btScalar ly = btScalar(2.) * (halfExtents.y());
+	btScalar lz = btScalar(2.) * (halfExtents.z());
 
+	inertia.setValue(mass / (btScalar(12.0)) * (ly * ly + lz * lz),
+					 mass / (btScalar(12.0)) * (lx * lx + lz * lz),
+					 mass / (btScalar(12.0)) * (lx * lx + ly * ly));
 }
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btBoxShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btBoxShape.h
index 715e3f2ab98..3c65505d5bb 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btBoxShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btBoxShape.h
@@ -23,112 +23,102 @@ subject to the following restrictions:
 #include "LinearMath/btMinMax.h"
 
 ///The btBoxShape is a box primitive around the origin, its sides axis aligned with length specified by half extents, in local shape coordinates. When used as part of a btCollisionObject or btRigidBody it will be an oriented box in world space.
-ATTRIBUTE_ALIGNED16(class) btBoxShape: public btPolyhedralConvexShape
+ATTRIBUTE_ALIGNED16(class)
+btBoxShape : public btPolyhedralConvexShape
 {
-
 	//btVector3	m_boxHalfExtents1; //use m_implicitShapeDimensions instead
 
-
 public:
-
-BT_DECLARE_ALIGNED_ALLOCATOR();
+	BT_DECLARE_ALIGNED_ALLOCATOR();
 
 	btVector3 getHalfExtentsWithMargin() const
 	{
 		btVector3 halfExtents = getHalfExtentsWithoutMargin();
-		btVector3 margin(getMargin(),getMargin(),getMargin());
+		btVector3 margin(getMargin(), getMargin(), getMargin());
 		halfExtents += margin;
 		return halfExtents;
 	}
-	
+
 	const btVector3& getHalfExtentsWithoutMargin() const
 	{
-		return m_implicitShapeDimensions;//scaling is included, margin is not
+		return m_implicitShapeDimensions;  //scaling is included, margin is not
 	}
-	
 
-	virtual btVector3	localGetSupportingVertex(const btVector3& vec) const
+	virtual btVector3 localGetSupportingVertex(const btVector3& vec) const
 	{
 		btVector3 halfExtents = getHalfExtentsWithoutMargin();
-		btVector3 margin(getMargin(),getMargin(),getMargin());
+		btVector3 margin(getMargin(), getMargin(), getMargin());
 		halfExtents += margin;
-		
+
 		return btVector3(btFsels(vec.x(), halfExtents.x(), -halfExtents.x()),
-			btFsels(vec.y(), halfExtents.y(), -halfExtents.y()),
-			btFsels(vec.z(), halfExtents.z(), -halfExtents.z()));
+						 btFsels(vec.y(), halfExtents.y(), -halfExtents.y()),
+						 btFsels(vec.z(), halfExtents.z(), -halfExtents.z()));
 	}
 
-	SIMD_FORCE_INLINE  btVector3	localGetSupportingVertexWithoutMargin(const btVector3& vec)const
+	SIMD_FORCE_INLINE btVector3 localGetSupportingVertexWithoutMargin(const btVector3& vec) const
 	{
 		const btVector3& halfExtents = getHalfExtentsWithoutMargin();
-		
+
 		return btVector3(btFsels(vec.x(), halfExtents.x(), -halfExtents.x()),
-			btFsels(vec.y(), halfExtents.y(), -halfExtents.y()),
-			btFsels(vec.z(), halfExtents.z(), -halfExtents.z()));
+						 btFsels(vec.y(), halfExtents.y(), -halfExtents.y()),
+						 btFsels(vec.z(), halfExtents.z(), -halfExtents.z()));
 	}
 
-	virtual void	batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const
+	virtual void batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const
 	{
 		const btVector3& halfExtents = getHalfExtentsWithoutMargin();
-	
-		for (int i=0;i<numVectors;i++)
+
+		for (int i = 0; i < numVectors; i++)
 		{
 			const btVector3& vec = vectors[i];
 			supportVerticesOut[i].setValue(btFsels(vec.x(), halfExtents.x(), -halfExtents.x()),
-				btFsels(vec.y(), halfExtents.y(), -halfExtents.y()),
-				btFsels(vec.z(), halfExtents.z(), -halfExtents.z())); 
+										   btFsels(vec.y(), halfExtents.y(), -halfExtents.y()),
+										   btFsels(vec.z(), halfExtents.z(), -halfExtents.z()));
 		}
-
 	}
 
-
-	btBoxShape( const btVector3& boxHalfExtents);
+	btBoxShape(const btVector3& boxHalfExtents);
 
 	virtual void setMargin(btScalar collisionMargin)
 	{
 		//correct the m_implicitShapeDimensions for the margin
-		btVector3 oldMargin(getMargin(),getMargin(),getMargin());
-		btVector3 implicitShapeDimensionsWithMargin = m_implicitShapeDimensions+oldMargin;
-		
-		btConvexInternalShape::setMargin(collisionMargin);
-		btVector3 newMargin(getMargin(),getMargin(),getMargin());
-		m_implicitShapeDimensions = implicitShapeDimensionsWithMargin - newMargin;
+		btVector3 oldMargin(getMargin(), getMargin(), getMargin());
+		btVector3 implicitShapeDimensionsWithMargin = m_implicitShapeDimensions + oldMargin;
 
+		btConvexInternalShape::setMargin(collisionMargin);
+		btVector3 newMargin(getMargin(), getMargin(), getMargin());
+		m_implicitShapeDimensions = implicitShapeDimensionsWithMargin - newMargin;
 	}
-	virtual void	setLocalScaling(const btVector3& scaling)
+	virtual void setLocalScaling(const btVector3& scaling)
 	{
-		btVector3 oldMargin(getMargin(),getMargin(),getMargin());
-		btVector3 implicitShapeDimensionsWithMargin = m_implicitShapeDimensions+oldMargin;
+		btVector3 oldMargin(getMargin(), getMargin(), getMargin());
+		btVector3 implicitShapeDimensionsWithMargin = m_implicitShapeDimensions + oldMargin;
 		btVector3 unScaledImplicitShapeDimensionsWithMargin = implicitShapeDimensionsWithMargin / m_localScaling;
 
 		btConvexInternalShape::setLocalScaling(scaling);
 
 		m_implicitShapeDimensions = (unScaledImplicitShapeDimensionsWithMargin * m_localScaling) - oldMargin;
-
 	}
 
-	virtual void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const;
+	virtual void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
 
-	
+	virtual void calculateLocalInertia(btScalar mass, btVector3 & inertia) const;
 
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
-
-	virtual void getPlane(btVector3& planeNormal,btVector3& planeSupport,int i ) const
+	virtual void getPlane(btVector3 & planeNormal, btVector3 & planeSupport, int i) const
 	{
 		//this plane might not be aligned...
-		btVector4 plane ;
-		getPlaneEquation(plane,i);
-		planeNormal = btVector3(plane.getX(),plane.getY(),plane.getZ());
+		btVector4 plane;
+		getPlaneEquation(plane, i);
+		planeNormal = btVector3(plane.getX(), plane.getY(), plane.getZ());
 		planeSupport = localGetSupportingVertex(-planeNormal);
 	}
 
-	
 	virtual int getNumPlanes() const
 	{
 		return 6;
-	}	
-	
-	virtual int	getNumVertices() const 
+	}
+
+	virtual int getNumVertices() const
 	{
 		return 8;
 	}
@@ -138,49 +128,46 @@ BT_DECLARE_ALIGNED_ALLOCATOR();
 		return 12;
 	}
 
-
-	virtual void getVertex(int i,btVector3& vtx) const
+	virtual void getVertex(int i, btVector3& vtx) const
 	{
 		btVector3 halfExtents = getHalfExtentsWithMargin();
 
 		vtx = btVector3(
-				halfExtents.x() * (1-(i&1)) - halfExtents.x() * (i&1),
-				halfExtents.y() * (1-((i&2)>>1)) - halfExtents.y() * ((i&2)>>1),
-				halfExtents.z() * (1-((i&4)>>2)) - halfExtents.z() * ((i&4)>>2));
+			halfExtents.x() * (1 - (i & 1)) - halfExtents.x() * (i & 1),
+			halfExtents.y() * (1 - ((i & 2) >> 1)) - halfExtents.y() * ((i & 2) >> 1),
+			halfExtents.z() * (1 - ((i & 4) >> 2)) - halfExtents.z() * ((i & 4) >> 2));
 	}
-	
 
-	virtual void	getPlaneEquation(btVector4& plane,int i) const
+	virtual void getPlaneEquation(btVector4 & plane, int i) const
 	{
 		btVector3 halfExtents = getHalfExtentsWithoutMargin();
 
 		switch (i)
 		{
-		case 0:
-			plane.setValue(btScalar(1.),btScalar(0.),btScalar(0.),-halfExtents.x());
-			break;
-		case 1:
-			plane.setValue(btScalar(-1.),btScalar(0.),btScalar(0.),-halfExtents.x());
-			break;
-		case 2:
-			plane.setValue(btScalar(0.),btScalar(1.),btScalar(0.),-halfExtents.y());
-			break;
-		case 3:
-			plane.setValue(btScalar(0.),btScalar(-1.),btScalar(0.),-halfExtents.y());
-			break;
-		case 4:
-			plane.setValue(btScalar(0.),btScalar(0.),btScalar(1.),-halfExtents.z());
-			break;
-		case 5:
-			plane.setValue(btScalar(0.),btScalar(0.),btScalar(-1.),-halfExtents.z());
-			break;
-		default:
-			btAssert(0);
+			case 0:
+				plane.setValue(btScalar(1.), btScalar(0.), btScalar(0.), -halfExtents.x());
+				break;
+			case 1:
+				plane.setValue(btScalar(-1.), btScalar(0.), btScalar(0.), -halfExtents.x());
+				break;
+			case 2:
+				plane.setValue(btScalar(0.), btScalar(1.), btScalar(0.), -halfExtents.y());
+				break;
+			case 3:
+				plane.setValue(btScalar(0.), btScalar(-1.), btScalar(0.), -halfExtents.y());
+				break;
+			case 4:
+				plane.setValue(btScalar(0.), btScalar(0.), btScalar(1.), -halfExtents.z());
+				break;
+			case 5:
+				plane.setValue(btScalar(0.), btScalar(0.), btScalar(-1.), -halfExtents.z());
+				break;
+			default:
+				btAssert(0);
 		}
 	}
 
-	
-	virtual void getEdge(int i,btVector3& pa,btVector3& pb) const
+	virtual void getEdge(int i, btVector3& pa, btVector3& pb) const
 	//virtual void getEdge(int i,Edge& edge) const
 	{
 		int edgeVert0 = 0;
@@ -188,127 +175,117 @@ BT_DECLARE_ALIGNED_ALLOCATOR();
 
 		switch (i)
 		{
-		case 0:
+			case 0:
 				edgeVert0 = 0;
 				edgeVert1 = 1;
-			break;
-		case 1:
+				break;
+			case 1:
 				edgeVert0 = 0;
 				edgeVert1 = 2;
-			break;
-		case 2:
-			edgeVert0 = 1;
-			edgeVert1 = 3;
+				break;
+			case 2:
+				edgeVert0 = 1;
+				edgeVert1 = 3;
 
-			break;
-		case 3:
-			edgeVert0 = 2;
-			edgeVert1 = 3;
-			break;
-		case 4:
-			edgeVert0 = 0;
-			edgeVert1 = 4;
-			break;
-		case 5:
-			edgeVert0 = 1;
-			edgeVert1 = 5;
-
-			break;
-		case 6:
-			edgeVert0 = 2;
-			edgeVert1 = 6;
-			break;
-		case 7:
-			edgeVert0 = 3;
-			edgeVert1 = 7;
-			break;
-		case 8:
-			edgeVert0 = 4;
-			edgeVert1 = 5;
-			break;
-		case 9:
-			edgeVert0 = 4;
-			edgeVert1 = 6;
-			break;
-		case 10:
-			edgeVert0 = 5;
-			edgeVert1 = 7;
-			break;
-		case 11:
-			edgeVert0 = 6;
-			edgeVert1 = 7;
-			break;
-		default:
-			btAssert(0);
+				break;
+			case 3:
+				edgeVert0 = 2;
+				edgeVert1 = 3;
+				break;
+			case 4:
+				edgeVert0 = 0;
+				edgeVert1 = 4;
+				break;
+			case 5:
+				edgeVert0 = 1;
+				edgeVert1 = 5;
 
+				break;
+			case 6:
+				edgeVert0 = 2;
+				edgeVert1 = 6;
+				break;
+			case 7:
+				edgeVert0 = 3;
+				edgeVert1 = 7;
+				break;
+			case 8:
+				edgeVert0 = 4;
+				edgeVert1 = 5;
+				break;
+			case 9:
+				edgeVert0 = 4;
+				edgeVert1 = 6;
+				break;
+			case 10:
+				edgeVert0 = 5;
+				edgeVert1 = 7;
+				break;
+			case 11:
+				edgeVert0 = 6;
+				edgeVert1 = 7;
+				break;
+			default:
+				btAssert(0);
 		}
 
-		getVertex(edgeVert0,pa );
-		getVertex(edgeVert1,pb );
+		getVertex(edgeVert0, pa);
+		getVertex(edgeVert1, pb);
 	}
 
-
-
-
-	
-	virtual	bool isInside(const btVector3& pt,btScalar tolerance) const
+	virtual bool isInside(const btVector3& pt, btScalar tolerance) const
 	{
 		btVector3 halfExtents = getHalfExtentsWithoutMargin();
 
 		//btScalar minDist = 2*tolerance;
-		
-		bool result =	(pt.x() <= (halfExtents.x()+tolerance)) &&
-						(pt.x() >= (-halfExtents.x()-tolerance)) &&
-						(pt.y() <= (halfExtents.y()+tolerance)) &&
-						(pt.y() >= (-halfExtents.y()-tolerance)) &&
-						(pt.z() <= (halfExtents.z()+tolerance)) &&
-						(pt.z() >= (-halfExtents.z()-tolerance));
-		
+
+		bool result = (pt.x() <= (halfExtents.x() + tolerance)) &&
+					  (pt.x() >= (-halfExtents.x() - tolerance)) &&
+					  (pt.y() <= (halfExtents.y() + tolerance)) &&
+					  (pt.y() >= (-halfExtents.y() - tolerance)) &&
+					  (pt.z() <= (halfExtents.z() + tolerance)) &&
+					  (pt.z() >= (-halfExtents.z() - tolerance));
+
 		return result;
 	}
 
-
 	//debugging
-	virtual const char*	getName()const
+	virtual const char* getName() const
 	{
 		return "Box";
 	}
 
-	virtual int		getNumPreferredPenetrationDirections() const
+	virtual int getNumPreferredPenetrationDirections() const
 	{
 		return 6;
 	}
-	
-	virtual void	getPreferredPenetrationDirection(int index, btVector3& penetrationVector) const
+
+	virtual void getPreferredPenetrationDirection(int index, btVector3& penetrationVector) const
 	{
 		switch (index)
 		{
-		case 0:
-			penetrationVector.setValue(btScalar(1.),btScalar(0.),btScalar(0.));
-			break;
-		case 1:
-			penetrationVector.setValue(btScalar(-1.),btScalar(0.),btScalar(0.));
-			break;
-		case 2:
-			penetrationVector.setValue(btScalar(0.),btScalar(1.),btScalar(0.));
-			break;
-		case 3:
-			penetrationVector.setValue(btScalar(0.),btScalar(-1.),btScalar(0.));
-			break;
-		case 4:
-			penetrationVector.setValue(btScalar(0.),btScalar(0.),btScalar(1.));
-			break;
-		case 5:
-			penetrationVector.setValue(btScalar(0.),btScalar(0.),btScalar(-1.));
-			break;
-		default:
-			btAssert(0);
+			case 0:
+				penetrationVector.setValue(btScalar(1.), btScalar(0.), btScalar(0.));
+				break;
+			case 1:
+				penetrationVector.setValue(btScalar(-1.), btScalar(0.), btScalar(0.));
+				break;
+			case 2:
+				penetrationVector.setValue(btScalar(0.), btScalar(1.), btScalar(0.));
+				break;
+			case 3:
+				penetrationVector.setValue(btScalar(0.), btScalar(-1.), btScalar(0.));
+				break;
+			case 4:
+				penetrationVector.setValue(btScalar(0.), btScalar(0.), btScalar(1.));
+				break;
+			case 5:
+				penetrationVector.setValue(btScalar(0.), btScalar(0.), btScalar(-1.));
+				break;
+			default:
+				btAssert(0);
 		}
 	}
-
 };
 
-
-#endif //BT_OBB_BOX_MINKOWSKI_H
-
-
+#endif  //BT_OBB_BOX_MINKOWSKI_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btBvhTriangleMeshShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btBvhTriangleMeshShape.cpp
index 61f465cb729..d663b3d6d6b 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btBvhTriangleMeshShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btBvhTriangleMeshShape.cpp
@@ -22,11 +22,11 @@ subject to the following restrictions:
 ///Bvh Concave triangle mesh is a static-triangle mesh shape with Bounding Volume Hierarchy optimization.
 ///Uses an interface to access the triangles to allow for sharing graphics/physics triangles.
 btBvhTriangleMeshShape::btBvhTriangleMeshShape(btStridingMeshInterface* meshInterface, bool useQuantizedAabbCompression, bool buildBvh)
-:btTriangleMeshShape(meshInterface),
-m_bvh(0),
-m_triangleInfoMap(0),
-m_useQuantizedAabbCompression(useQuantizedAabbCompression),
-m_ownsBvh(false)
+	: btTriangleMeshShape(meshInterface),
+	  m_bvh(0),
+	  m_triangleInfoMap(0),
+	  m_useQuantizedAabbCompression(useQuantizedAabbCompression),
+	  m_ownsBvh(false)
 {
 	m_shapeType = TRIANGLE_MESH_SHAPE_PROXYTYPE;
 	//construct bvh from meshInterface
@@ -37,16 +37,15 @@ m_ownsBvh(false)
 		buildOptimizedBvh();
 	}
 
-#endif //DISABLE_BVH
-
+#endif  //DISABLE_BVH
 }
 
-btBvhTriangleMeshShape::btBvhTriangleMeshShape(btStridingMeshInterface* meshInterface, bool useQuantizedAabbCompression,const btVector3& bvhAabbMin,const btVector3& bvhAabbMax,bool buildBvh)
-:btTriangleMeshShape(meshInterface),
-m_bvh(0),
-m_triangleInfoMap(0),
-m_useQuantizedAabbCompression(useQuantizedAabbCompression),
-m_ownsBvh(false)
+btBvhTriangleMeshShape::btBvhTriangleMeshShape(btStridingMeshInterface* meshInterface, bool useQuantizedAabbCompression, const btVector3& bvhAabbMin, const btVector3& bvhAabbMax, bool buildBvh)
+	: btTriangleMeshShape(meshInterface),
+	  m_bvh(0),
+	  m_triangleInfoMap(0),
+	  m_useQuantizedAabbCompression(useQuantizedAabbCompression),
+	  m_ownsBvh(false)
 {
 	m_shapeType = TRIANGLE_MESH_SHAPE_PROXYTYPE;
 	//construct bvh from meshInterface
@@ -54,30 +53,28 @@ m_ownsBvh(false)
 
 	if (buildBvh)
 	{
-		void* mem = btAlignedAlloc(sizeof(btOptimizedBvh),16);
+		void* mem = btAlignedAlloc(sizeof(btOptimizedBvh), 16);
 		m_bvh = new (mem) btOptimizedBvh();
-		
-		m_bvh->build(meshInterface,m_useQuantizedAabbCompression,bvhAabbMin,bvhAabbMax);
+
+		m_bvh->build(meshInterface, m_useQuantizedAabbCompression, bvhAabbMin, bvhAabbMax);
 		m_ownsBvh = true;
 	}
 
-#endif //DISABLE_BVH
-
+#endif  //DISABLE_BVH
 }
 
-void	btBvhTriangleMeshShape::partialRefitTree(const btVector3& aabbMin,const btVector3& aabbMax)
+void btBvhTriangleMeshShape::partialRefitTree(const btVector3& aabbMin, const btVector3& aabbMax)
 {
-	m_bvh->refitPartial( m_meshInterface,aabbMin,aabbMax );
-	
+	m_bvh->refitPartial(m_meshInterface, aabbMin, aabbMax);
+
 	m_localAabbMin.setMin(aabbMin);
 	m_localAabbMax.setMax(aabbMax);
 }
 
-
-void	btBvhTriangleMeshShape::refitTree(const btVector3& aabbMin,const btVector3& aabbMax)
+void btBvhTriangleMeshShape::refitTree(const btVector3& aabbMin, const btVector3& aabbMax)
 {
-	m_bvh->refit( m_meshInterface, aabbMin,aabbMax );
-	
+	m_bvh->refit(m_meshInterface, aabbMin, aabbMax);
+
 	recalcLocalAabb();
 }
 
@@ -90,27 +87,27 @@ btBvhTriangleMeshShape::~btBvhTriangleMeshShape()
 	}
 }
 
-void	btBvhTriangleMeshShape::performRaycast (btTriangleCallback* callback, const btVector3& raySource, const btVector3& rayTarget)
+void btBvhTriangleMeshShape::performRaycast(btTriangleCallback* callback, const btVector3& raySource, const btVector3& rayTarget)
 {
-	struct	MyNodeOverlapCallback : public btNodeOverlapCallback
+	struct MyNodeOverlapCallback : public btNodeOverlapCallback
 	{
-		btStridingMeshInterface*	m_meshInterface;
+		btStridingMeshInterface* m_meshInterface;
 		btTriangleCallback* m_callback;
 
-		MyNodeOverlapCallback(btTriangleCallback* callback,btStridingMeshInterface* meshInterface)
-			:m_meshInterface(meshInterface),
-			m_callback(callback)
+		MyNodeOverlapCallback(btTriangleCallback* callback, btStridingMeshInterface* meshInterface)
+			: m_meshInterface(meshInterface),
+			  m_callback(callback)
 		{
 		}
-				
+
 		virtual void processNode(int nodeSubPart, int nodeTriangleIndex)
 		{
 			btVector3 m_triangle[3];
-			const unsigned char *vertexbase;
+			const unsigned char* vertexbase;
 			int numverts;
 			PHY_ScalarType type;
 			int stride;
-			const unsigned char *indexbase;
+			const unsigned char* indexbase;
 			int indexstride;
 			int numfaces;
 			PHY_ScalarType indicestype;
@@ -126,60 +123,60 @@ void	btBvhTriangleMeshShape::performRaycast (btTriangleCallback* callback, const
 				indicestype,
 				nodeSubPart);
 
-			unsigned int* gfxbase = (unsigned int*)(indexbase+nodeTriangleIndex*indexstride);
-			btAssert(indicestype==PHY_INTEGER||indicestype==PHY_SHORT);
-	
+			unsigned int* gfxbase = (unsigned int*)(indexbase + nodeTriangleIndex * indexstride);
+			btAssert(indicestype == PHY_INTEGER || indicestype == PHY_SHORT);
+
 			const btVector3& meshScaling = m_meshInterface->getScaling();
-			for (int j=2;j>=0;j--)
+			for (int j = 2; j >= 0; j--)
 			{
-				int graphicsindex = indicestype==PHY_SHORT?((unsigned short*)gfxbase)[j]:gfxbase[j];
-				
+				int graphicsindex = indicestype == PHY_SHORT ? ((unsigned short*)gfxbase)[j] : gfxbase[j];
+
 				if (type == PHY_FLOAT)
 				{
-					float* graphicsbase = (float*)(vertexbase+graphicsindex*stride);
-					
-					m_triangle[j] = btVector3(graphicsbase[0]*meshScaling.getX(),graphicsbase[1]*meshScaling.getY(),graphicsbase[2]*meshScaling.getZ());		
+					float* graphicsbase = (float*)(vertexbase + graphicsindex * stride);
+
+					m_triangle[j] = btVector3(graphicsbase[0] * meshScaling.getX(), graphicsbase[1] * meshScaling.getY(), graphicsbase[2] * meshScaling.getZ());
 				}
 				else
 				{
-					double* graphicsbase = (double*)(vertexbase+graphicsindex*stride);
-					
-					m_triangle[j] = btVector3(btScalar(graphicsbase[0])*meshScaling.getX(),btScalar(graphicsbase[1])*meshScaling.getY(),btScalar(graphicsbase[2])*meshScaling.getZ());		
+					double* graphicsbase = (double*)(vertexbase + graphicsindex * stride);
+
+					m_triangle[j] = btVector3(btScalar(graphicsbase[0]) * meshScaling.getX(), btScalar(graphicsbase[1]) * meshScaling.getY(), btScalar(graphicsbase[2]) * meshScaling.getZ());
 				}
 			}
 
 			/* Perform ray vs. triangle collision here */
-			m_callback->processTriangle(m_triangle,nodeSubPart,nodeTriangleIndex);
+			m_callback->processTriangle(m_triangle, nodeSubPart, nodeTriangleIndex);
 			m_meshInterface->unLockReadOnlyVertexBase(nodeSubPart);
 		}
 	};
 
-	MyNodeOverlapCallback	myNodeCallback(callback,m_meshInterface);
+	MyNodeOverlapCallback myNodeCallback(callback, m_meshInterface);
 
-	m_bvh->reportRayOverlappingNodex(&myNodeCallback,raySource,rayTarget);
+	m_bvh->reportRayOverlappingNodex(&myNodeCallback, raySource, rayTarget);
 }
 
-void	btBvhTriangleMeshShape::performConvexcast (btTriangleCallback* callback, const btVector3& raySource, const btVector3& rayTarget, const btVector3& aabbMin, const btVector3& aabbMax)
+void btBvhTriangleMeshShape::performConvexcast(btTriangleCallback* callback, const btVector3& raySource, const btVector3& rayTarget, const btVector3& aabbMin, const btVector3& aabbMax)
 {
-	struct	MyNodeOverlapCallback : public btNodeOverlapCallback
+	struct MyNodeOverlapCallback : public btNodeOverlapCallback
 	{
-		btStridingMeshInterface*	m_meshInterface;
+		btStridingMeshInterface* m_meshInterface;
 		btTriangleCallback* m_callback;
 
-		MyNodeOverlapCallback(btTriangleCallback* callback,btStridingMeshInterface* meshInterface)
-			:m_meshInterface(meshInterface),
-			m_callback(callback)
+		MyNodeOverlapCallback(btTriangleCallback* callback, btStridingMeshInterface* meshInterface)
+			: m_meshInterface(meshInterface),
+			  m_callback(callback)
 		{
 		}
-				
+
 		virtual void processNode(int nodeSubPart, int nodeTriangleIndex)
 		{
 			btVector3 m_triangle[3];
-			const unsigned char *vertexbase;
+			const unsigned char* vertexbase;
 			int numverts;
 			PHY_ScalarType type;
 			int stride;
-			const unsigned char *indexbase;
+			const unsigned char* indexbase;
 			int indexstride;
 			int numfaces;
 			PHY_ScalarType indicestype;
@@ -195,77 +192,74 @@ void	btBvhTriangleMeshShape::performConvexcast (btTriangleCallback* callback, co
 				indicestype,
 				nodeSubPart);
 
-			unsigned int* gfxbase = (unsigned int*)(indexbase+nodeTriangleIndex*indexstride);
-			btAssert(indicestype==PHY_INTEGER||indicestype==PHY_SHORT);
-	
+			unsigned int* gfxbase = (unsigned int*)(indexbase + nodeTriangleIndex * indexstride);
+			btAssert(indicestype == PHY_INTEGER || indicestype == PHY_SHORT);
+
 			const btVector3& meshScaling = m_meshInterface->getScaling();
-			for (int j=2;j>=0;j--)
+			for (int j = 2; j >= 0; j--)
 			{
-				int graphicsindex = indicestype==PHY_SHORT?((unsigned short*)gfxbase)[j]:gfxbase[j];
+				int graphicsindex = indicestype == PHY_SHORT ? ((unsigned short*)gfxbase)[j] : gfxbase[j];
 
 				if (type == PHY_FLOAT)
 				{
-					float* graphicsbase = (float*)(vertexbase+graphicsindex*stride);
+					float* graphicsbase = (float*)(vertexbase + graphicsindex * stride);
 
-					m_triangle[j] = btVector3(graphicsbase[0]*meshScaling.getX(),graphicsbase[1]*meshScaling.getY(),graphicsbase[2]*meshScaling.getZ());		
+					m_triangle[j] = btVector3(graphicsbase[0] * meshScaling.getX(), graphicsbase[1] * meshScaling.getY(), graphicsbase[2] * meshScaling.getZ());
 				}
 				else
 				{
-					double* graphicsbase = (double*)(vertexbase+graphicsindex*stride);
-					
-					m_triangle[j] = btVector3(btScalar(graphicsbase[0])*meshScaling.getX(),btScalar(graphicsbase[1])*meshScaling.getY(),btScalar(graphicsbase[2])*meshScaling.getZ());		
+					double* graphicsbase = (double*)(vertexbase + graphicsindex * stride);
+
+					m_triangle[j] = btVector3(btScalar(graphicsbase[0]) * meshScaling.getX(), btScalar(graphicsbase[1]) * meshScaling.getY(), btScalar(graphicsbase[2]) * meshScaling.getZ());
 				}
 			}
 
 			/* Perform ray vs. triangle collision here */
-			m_callback->processTriangle(m_triangle,nodeSubPart,nodeTriangleIndex);
+			m_callback->processTriangle(m_triangle, nodeSubPart, nodeTriangleIndex);
 			m_meshInterface->unLockReadOnlyVertexBase(nodeSubPart);
 		}
 	};
 
-	MyNodeOverlapCallback	myNodeCallback(callback,m_meshInterface);
+	MyNodeOverlapCallback myNodeCallback(callback, m_meshInterface);
 
-	m_bvh->reportBoxCastOverlappingNodex (&myNodeCallback, raySource, rayTarget, aabbMin, aabbMax);
+	m_bvh->reportBoxCastOverlappingNodex(&myNodeCallback, raySource, rayTarget, aabbMin, aabbMax);
 }
 
 //perform bvh tree traversal and report overlapping triangles to 'callback'
-void	btBvhTriangleMeshShape::processAllTriangles(btTriangleCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax) const
+void btBvhTriangleMeshShape::processAllTriangles(btTriangleCallback* callback, const btVector3& aabbMin, const btVector3& aabbMax) const
 {
-
 #ifdef DISABLE_BVH
 	//brute force traverse all triangles
-	btTriangleMeshShape::processAllTriangles(callback,aabbMin,aabbMax);
+	btTriangleMeshShape::processAllTriangles(callback, aabbMin, aabbMax);
 #else
 
 	//first get all the nodes
 
-	
-	struct	MyNodeOverlapCallback : public btNodeOverlapCallback
+	struct MyNodeOverlapCallback : public btNodeOverlapCallback
 	{
-		btStridingMeshInterface*	m_meshInterface;
-		btTriangleCallback*		m_callback;
-		btVector3				m_triangle[3];
+		btStridingMeshInterface* m_meshInterface;
+		btTriangleCallback* m_callback;
+		btVector3 m_triangle[3];
 		int m_numOverlap;
 
-		MyNodeOverlapCallback(btTriangleCallback* callback,btStridingMeshInterface* meshInterface)
-			:m_meshInterface(meshInterface),
-			m_callback(callback),
-			m_numOverlap(0)
+		MyNodeOverlapCallback(btTriangleCallback* callback, btStridingMeshInterface* meshInterface)
+			: m_meshInterface(meshInterface),
+			  m_callback(callback),
+			  m_numOverlap(0)
 		{
 		}
-				
+
 		virtual void processNode(int nodeSubPart, int nodeTriangleIndex)
 		{
 			m_numOverlap++;
-			const unsigned char *vertexbase;
+			const unsigned char* vertexbase;
 			int numverts;
 			PHY_ScalarType type;
 			int stride;
-			const unsigned char *indexbase;
+			const unsigned char* indexbase;
 			int indexstride;
 			int numfaces;
 			PHY_ScalarType indicestype;
-			
 
 			m_meshInterface->getLockedReadOnlyVertexIndexBase(
 				&vertexbase,
@@ -278,67 +272,62 @@ void	btBvhTriangleMeshShape::processAllTriangles(btTriangleCallback* callback,co
 				indicestype,
 				nodeSubPart);
 
-			unsigned int* gfxbase = (unsigned int*)(indexbase+nodeTriangleIndex*indexstride);
-			btAssert(indicestype==PHY_INTEGER||indicestype==PHY_SHORT||indicestype==PHY_UCHAR);
-	
-			const btVector3& meshScaling = m_meshInterface->getScaling();
-			for (int j=2;j>=0;j--)
-			{
-				
-				int graphicsindex = indicestype==PHY_SHORT?((unsigned short*)gfxbase)[j]:indicestype==PHY_INTEGER?gfxbase[j]:((unsigned char*)gfxbase)[j];
+			unsigned int* gfxbase = (unsigned int*)(indexbase + nodeTriangleIndex * indexstride);
+			btAssert(indicestype == PHY_INTEGER || indicestype == PHY_SHORT || indicestype == PHY_UCHAR);
 
+			const btVector3& meshScaling = m_meshInterface->getScaling();
+			for (int j = 2; j >= 0; j--)
+			{
+				int graphicsindex = indicestype == PHY_SHORT ? ((unsigned short*)gfxbase)[j] : indicestype == PHY_INTEGER ? gfxbase[j] : ((unsigned char*)gfxbase)[j];
 
 #ifdef DEBUG_TRIANGLE_MESH
-				printf("%d ,",graphicsindex);
-#endif //DEBUG_TRIANGLE_MESH
+				printf("%d ,", graphicsindex);
+#endif  //DEBUG_TRIANGLE_MESH
 				if (type == PHY_FLOAT)
 				{
-					float* graphicsbase = (float*)(vertexbase+graphicsindex*stride);
-					
+					float* graphicsbase = (float*)(vertexbase + graphicsindex * stride);
+
 					m_triangle[j] = btVector3(
-																		graphicsbase[0]*meshScaling.getX(),
-																		graphicsbase[1]*meshScaling.getY(),
-																		graphicsbase[2]*meshScaling.getZ());
+						graphicsbase[0] * meshScaling.getX(),
+						graphicsbase[1] * meshScaling.getY(),
+						graphicsbase[2] * meshScaling.getZ());
 				}
 				else
 				{
-					double* graphicsbase = (double*)(vertexbase+graphicsindex*stride);
+					double* graphicsbase = (double*)(vertexbase + graphicsindex * stride);
 
 					m_triangle[j] = btVector3(
-						btScalar(graphicsbase[0])*meshScaling.getX(),
-						btScalar(graphicsbase[1])*meshScaling.getY(),
-						btScalar(graphicsbase[2])*meshScaling.getZ());
+						btScalar(graphicsbase[0]) * meshScaling.getX(),
+						btScalar(graphicsbase[1]) * meshScaling.getY(),
+						btScalar(graphicsbase[2]) * meshScaling.getZ());
 				}
 #ifdef DEBUG_TRIANGLE_MESH
-				printf("triangle vertices:%f,%f,%f\n",triangle[j].x(),triangle[j].y(),triangle[j].z());
-#endif //DEBUG_TRIANGLE_MESH
+				printf("triangle vertices:%f,%f,%f\n", triangle[j].x(), triangle[j].y(), triangle[j].z());
+#endif  //DEBUG_TRIANGLE_MESH
 			}
 
-			m_callback->processTriangle(m_triangle,nodeSubPart,nodeTriangleIndex);
+			m_callback->processTriangle(m_triangle, nodeSubPart, nodeTriangleIndex);
 			m_meshInterface->unLockReadOnlyVertexBase(nodeSubPart);
 		}
-
 	};
 
-	MyNodeOverlapCallback	myNodeCallback(callback,m_meshInterface);
-
-	m_bvh->reportAabbOverlappingNodex(&myNodeCallback,aabbMin,aabbMax);
-	
-#endif//DISABLE_BVH
+	MyNodeOverlapCallback myNodeCallback(callback, m_meshInterface);
 
+	m_bvh->reportAabbOverlappingNodex(&myNodeCallback, aabbMin, aabbMax);
 
+#endif  //DISABLE_BVH
 }
 
-void   btBvhTriangleMeshShape::setLocalScaling(const btVector3& scaling)
+void btBvhTriangleMeshShape::setLocalScaling(const btVector3& scaling)
 {
-   if ((getLocalScaling() -scaling).length2() > SIMD_EPSILON)
-   {
-      btTriangleMeshShape::setLocalScaling(scaling);
-	  buildOptimizedBvh();
-   }
+	if ((getLocalScaling() - scaling).length2() > SIMD_EPSILON)
+	{
+		btTriangleMeshShape::setLocalScaling(scaling);
+		buildOptimizedBvh();
+	}
 }
 
-void   btBvhTriangleMeshShape::buildOptimizedBvh()
+void btBvhTriangleMeshShape::buildOptimizedBvh()
 {
 	if (m_ownsBvh)
 	{
@@ -346,43 +335,39 @@ void   btBvhTriangleMeshShape::buildOptimizedBvh()
 		btAlignedFree(m_bvh);
 	}
 	///m_localAabbMin/m_localAabbMax is already re-calculated in btTriangleMeshShape. We could just scale aabb, but this needs some more work
-	void* mem = btAlignedAlloc(sizeof(btOptimizedBvh),16);
-	m_bvh = new(mem) btOptimizedBvh();
+	void* mem = btAlignedAlloc(sizeof(btOptimizedBvh), 16);
+	m_bvh = new (mem) btOptimizedBvh();
 	//rebuild the bvh...
-	m_bvh->build(m_meshInterface,m_useQuantizedAabbCompression,m_localAabbMin,m_localAabbMax);
+	m_bvh->build(m_meshInterface, m_useQuantizedAabbCompression, m_localAabbMin, m_localAabbMax);
 	m_ownsBvh = true;
 }
 
-void   btBvhTriangleMeshShape::setOptimizedBvh(btOptimizedBvh* bvh, const btVector3& scaling)
+void btBvhTriangleMeshShape::setOptimizedBvh(btOptimizedBvh* bvh, const btVector3& scaling)
 {
-   btAssert(!m_bvh);
-   btAssert(!m_ownsBvh);
+	btAssert(!m_bvh);
+	btAssert(!m_ownsBvh);
 
-   m_bvh = bvh;
-   m_ownsBvh = false;
-   // update the scaling without rebuilding the bvh
-   if ((getLocalScaling() -scaling).length2() > SIMD_EPSILON)
-   {
-      btTriangleMeshShape::setLocalScaling(scaling);
-   }
+	m_bvh = bvh;
+	m_ownsBvh = false;
+	// update the scaling without rebuilding the bvh
+	if ((getLocalScaling() - scaling).length2() > SIMD_EPSILON)
+	{
+		btTriangleMeshShape::setLocalScaling(scaling);
+	}
 }
 
-
-
 ///fills the dataBuffer and returns the struct name (and 0 on failure)
-const char*	btBvhTriangleMeshShape::serialize(void* dataBuffer, btSerializer* serializer) const
+const char* btBvhTriangleMeshShape::serialize(void* dataBuffer, btSerializer* serializer) const
 {
-	btTriangleMeshShapeData* trimeshData = (btTriangleMeshShapeData*) dataBuffer;
+	btTriangleMeshShapeData* trimeshData = (btTriangleMeshShapeData*)dataBuffer;
 
-	btCollisionShape::serialize(&trimeshData->m_collisionShapeData,serializer);
+	btCollisionShape::serialize(&trimeshData->m_collisionShapeData, serializer);
 
 	m_meshInterface->serialize(&trimeshData->m_meshInterface, serializer);
 
 	trimeshData->m_collisionMargin = float(m_collisionMargin);
 
-	
-
-	if (m_bvh && !(serializer->getSerializationFlags()&BT_SERIALIZE_NO_BVH))
+	if (m_bvh && !(serializer->getSerializationFlags() & BT_SERIALIZE_NO_BVH))
 	{
 		void* chunk = serializer->findPointer(m_bvh);
 		if (chunk)
@@ -391,48 +376,49 @@ const char*	btBvhTriangleMeshShape::serialize(void* dataBuffer, btSerializer* se
 			trimeshData->m_quantizedDoubleBvh = (btQuantizedBvhData*)chunk;
 			trimeshData->m_quantizedFloatBvh = 0;
 #else
-			trimeshData->m_quantizedFloatBvh  = (btQuantizedBvhData*)chunk;
-			trimeshData->m_quantizedDoubleBvh= 0;
-#endif //BT_USE_DOUBLE_PRECISION
-		} else
+			trimeshData->m_quantizedFloatBvh = (btQuantizedBvhData*)chunk;
+			trimeshData->m_quantizedDoubleBvh = 0;
+#endif  //BT_USE_DOUBLE_PRECISION
+		}
+		else
 		{
-
 #ifdef BT_USE_DOUBLE_PRECISION
 			trimeshData->m_quantizedDoubleBvh = (btQuantizedBvhData*)serializer->getUniquePointer(m_bvh);
 			trimeshData->m_quantizedFloatBvh = 0;
 #else
-			trimeshData->m_quantizedFloatBvh  = (btQuantizedBvhData*)serializer->getUniquePointer(m_bvh);
-			trimeshData->m_quantizedDoubleBvh= 0;
-#endif //BT_USE_DOUBLE_PRECISION
-	
+			trimeshData->m_quantizedFloatBvh = (btQuantizedBvhData*)serializer->getUniquePointer(m_bvh);
+			trimeshData->m_quantizedDoubleBvh = 0;
+#endif  //BT_USE_DOUBLE_PRECISION
+
 			int sz = m_bvh->calculateSerializeBufferSizeNew();
-			btChunk* chunk = serializer->allocate(sz,1);
+			btChunk* chunk = serializer->allocate(sz, 1);
 			const char* structType = m_bvh->serialize(chunk->m_oldPtr, serializer);
-			serializer->finalizeChunk(chunk,structType,BT_QUANTIZED_BVH_CODE,m_bvh);
+			serializer->finalizeChunk(chunk, structType, BT_QUANTIZED_BVH_CODE, m_bvh);
 		}
-	} else
+	}
+	else
 	{
 		trimeshData->m_quantizedFloatBvh = 0;
 		trimeshData->m_quantizedDoubleBvh = 0;
 	}
 
-	
-
-	if (m_triangleInfoMap && !(serializer->getSerializationFlags()&BT_SERIALIZE_NO_TRIANGLEINFOMAP))
+	if (m_triangleInfoMap && !(serializer->getSerializationFlags() & BT_SERIALIZE_NO_TRIANGLEINFOMAP))
 	{
 		void* chunk = serializer->findPointer(m_triangleInfoMap);
 		if (chunk)
 		{
 			trimeshData->m_triangleInfoMap = (btTriangleInfoMapData*)chunk;
-		} else
+		}
+		else
 		{
 			trimeshData->m_triangleInfoMap = (btTriangleInfoMapData*)serializer->getUniquePointer(m_triangleInfoMap);
 			int sz = m_triangleInfoMap->calculateSerializeBufferSize();
-			btChunk* chunk = serializer->allocate(sz,1);
+			btChunk* chunk = serializer->allocate(sz, 1);
 			const char* structType = m_triangleInfoMap->serialize(chunk->m_oldPtr, serializer);
-			serializer->finalizeChunk(chunk,structType,BT_TRIANLGE_INFO_MAP,m_triangleInfoMap);
+			serializer->finalizeChunk(chunk, structType, BT_TRIANLGE_INFO_MAP, m_triangleInfoMap);
 		}
-	} else
+	}
+	else
 	{
 		trimeshData->m_triangleInfoMap = 0;
 	}
@@ -443,28 +429,24 @@ const char*	btBvhTriangleMeshShape::serialize(void* dataBuffer, btSerializer* se
 	return "btTriangleMeshShapeData";
 }
 
-void	btBvhTriangleMeshShape::serializeSingleBvh(btSerializer* serializer) const
+void btBvhTriangleMeshShape::serializeSingleBvh(btSerializer* serializer) const
 {
 	if (m_bvh)
 	{
-		int len = m_bvh->calculateSerializeBufferSizeNew(); //make sure not to use calculateSerializeBufferSize because it is used for in-place
-		btChunk* chunk = serializer->allocate(len,1);
+		int len = m_bvh->calculateSerializeBufferSizeNew();  //make sure not to use calculateSerializeBufferSize because it is used for in-place
+		btChunk* chunk = serializer->allocate(len, 1);
 		const char* structType = m_bvh->serialize(chunk->m_oldPtr, serializer);
-		serializer->finalizeChunk(chunk,structType,BT_QUANTIZED_BVH_CODE,(void*)m_bvh);
+		serializer->finalizeChunk(chunk, structType, BT_QUANTIZED_BVH_CODE, (void*)m_bvh);
 	}
 }
 
-void	btBvhTriangleMeshShape::serializeSingleTriangleInfoMap(btSerializer* serializer) const
+void btBvhTriangleMeshShape::serializeSingleTriangleInfoMap(btSerializer* serializer) const
 {
 	if (m_triangleInfoMap)
 	{
 		int len = m_triangleInfoMap->calculateSerializeBufferSize();
-		btChunk* chunk = serializer->allocate(len,1);
+		btChunk* chunk = serializer->allocate(len, 1);
 		const char* structType = m_triangleInfoMap->serialize(chunk->m_oldPtr, serializer);
-		serializer->finalizeChunk(chunk,structType,BT_TRIANLGE_INFO_MAP,(void*)m_triangleInfoMap);
+		serializer->finalizeChunk(chunk, structType, BT_TRIANLGE_INFO_MAP, (void*)m_triangleInfoMap);
 	}
 }
-
-
-
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btBvhTriangleMeshShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btBvhTriangleMeshShape.h
index 1fa4995d160..8b2f2ee85e8 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btBvhTriangleMeshShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btBvhTriangleMeshShape.h
@@ -23,103 +23,99 @@ subject to the following restrictions:
 
 ///The btBvhTriangleMeshShape is a static-triangle mesh shape, it can only be used for fixed/non-moving objects.
 ///If you required moving concave triangle meshes, it is recommended to perform convex decomposition
-///using HACD, see Bullet/Demos/ConvexDecompositionDemo. 
+///using HACD, see Bullet/Demos/ConvexDecompositionDemo.
 ///Alternatively, you can use btGimpactMeshShape for moving concave triangle meshes.
-///btBvhTriangleMeshShape has several optimizations, such as bounding volume hierarchy and 
-///cache friendly traversal for PlayStation 3 Cell SPU. 
+///btBvhTriangleMeshShape has several optimizations, such as bounding volume hierarchy and
+///cache friendly traversal for PlayStation 3 Cell SPU.
 ///It is recommended to enable useQuantizedAabbCompression for better memory usage.
 ///It takes a triangle mesh as input, for example a btTriangleMesh or btTriangleIndexVertexArray. The btBvhTriangleMeshShape class allows for triangle mesh deformations by a refit or partialRefit method.
 ///Instead of building the bounding volume hierarchy acceleration structure, it is also possible to serialize (save) and deserialize (load) the structure from disk.
 ///See Demos\ConcaveDemo\ConcavePhysicsDemo.cpp for an example.
-ATTRIBUTE_ALIGNED16(class) btBvhTriangleMeshShape : public btTriangleMeshShape
+ATTRIBUTE_ALIGNED16(class)
+btBvhTriangleMeshShape : public btTriangleMeshShape
 {
-
-	btOptimizedBvh*	m_bvh;
-	btTriangleInfoMap*	m_triangleInfoMap;
+	btOptimizedBvh* m_bvh;
+	btTriangleInfoMap* m_triangleInfoMap;
 
 	bool m_useQuantizedAabbCompression;
 	bool m_ownsBvh;
 #ifdef __clang__
-	bool m_pad[11] __attribute__((unused));////need padding due to alignment
+	bool m_pad[11] __attribute__((unused));  ////need padding due to alignment
 #else
-	bool m_pad[11];////need padding due to alignment
+	bool m_pad[11];  ////need padding due to alignment
 #endif
 
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-	
-	btBvhTriangleMeshShape(btStridingMeshInterface* meshInterface, bool useQuantizedAabbCompression, bool buildBvh = true);
+	btBvhTriangleMeshShape(btStridingMeshInterface * meshInterface, bool useQuantizedAabbCompression, bool buildBvh = true);
 
 	///optionally pass in a larger bvh aabb, used for quantization. This allows for deformations within this aabb
-	btBvhTriangleMeshShape(btStridingMeshInterface* meshInterface, bool useQuantizedAabbCompression,const btVector3& bvhAabbMin,const btVector3& bvhAabbMax, bool buildBvh = true);
-	
+	btBvhTriangleMeshShape(btStridingMeshInterface * meshInterface, bool useQuantizedAabbCompression, const btVector3& bvhAabbMin, const btVector3& bvhAabbMax, bool buildBvh = true);
+
 	virtual ~btBvhTriangleMeshShape();
 
-	bool getOwnsBvh () const
+	bool getOwnsBvh() const
 	{
 		return m_ownsBvh;
 	}
 
+	void performRaycast(btTriangleCallback * callback, const btVector3& raySource, const btVector3& rayTarget);
+	void performConvexcast(btTriangleCallback * callback, const btVector3& boxSource, const btVector3& boxTarget, const btVector3& boxMin, const btVector3& boxMax);
 
-	
-	void performRaycast (btTriangleCallback* callback, const btVector3& raySource, const btVector3& rayTarget);
-	void performConvexcast (btTriangleCallback* callback, const btVector3& boxSource, const btVector3& boxTarget, const btVector3& boxMin, const btVector3& boxMax);
+	virtual void processAllTriangles(btTriangleCallback * callback, const btVector3& aabbMin, const btVector3& aabbMax) const;
 
-	virtual void	processAllTriangles(btTriangleCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax) const;
-
-	void	refitTree(const btVector3& aabbMin,const btVector3& aabbMax);
+	void refitTree(const btVector3& aabbMin, const btVector3& aabbMax);
 
 	///for a fast incremental refit of parts of the tree. Note: the entire AABB of the tree will become more conservative, it never shrinks
-	void	partialRefitTree(const btVector3& aabbMin,const btVector3& aabbMax);
+	void partialRefitTree(const btVector3& aabbMin, const btVector3& aabbMax);
 
 	//debugging
-	virtual const char*	getName()const {return "BVHTRIANGLEMESH";}
+	virtual const char* getName() const { return "BVHTRIANGLEMESH"; }
 
+	virtual void setLocalScaling(const btVector3& scaling);
 
-	virtual void	setLocalScaling(const btVector3& scaling);
-	
-	btOptimizedBvh*	getOptimizedBvh()
+	btOptimizedBvh* getOptimizedBvh()
 	{
 		return m_bvh;
 	}
 
-	void	setOptimizedBvh(btOptimizedBvh* bvh, const btVector3& localScaling=btVector3(1,1,1));
+	void setOptimizedBvh(btOptimizedBvh * bvh, const btVector3& localScaling = btVector3(1, 1, 1));
 
-	void    buildOptimizedBvh();
+	void buildOptimizedBvh();
 
-	bool	usesQuantizedAabbCompression() const
+	bool usesQuantizedAabbCompression() const
 	{
-		return	m_useQuantizedAabbCompression;
+		return m_useQuantizedAabbCompression;
 	}
 
-	void	setTriangleInfoMap(btTriangleInfoMap* triangleInfoMap)
+	void setTriangleInfoMap(btTriangleInfoMap * triangleInfoMap)
 	{
 		m_triangleInfoMap = triangleInfoMap;
 	}
 
-	const btTriangleInfoMap*	getTriangleInfoMap() const
-	{
-		return m_triangleInfoMap;
-	}
-	
-	btTriangleInfoMap*	getTriangleInfoMap()
+	const btTriangleInfoMap* getTriangleInfoMap() const
 	{
 		return m_triangleInfoMap;
 	}
 
-	virtual	int	calculateSerializeBufferSize() const;
+	btTriangleInfoMap* getTriangleInfoMap()
+	{
+		return m_triangleInfoMap;
+	}
+
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 
-	virtual void	serializeSingleBvh(btSerializer* serializer) const;
-
-	virtual void	serializeSingleTriangleInfoMap(btSerializer* serializer) const;
+	virtual void serializeSingleBvh(btSerializer * serializer) const;
 
+	virtual void serializeSingleTriangleInfoMap(btSerializer * serializer) const;
 };
 
+// clang-format off
+
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
 struct	btTriangleMeshShapeData
 {
@@ -138,12 +134,11 @@ struct	btTriangleMeshShapeData
 	
 };
 
+// clang-format on
 
-SIMD_FORCE_INLINE	int	btBvhTriangleMeshShape::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btBvhTriangleMeshShape::calculateSerializeBufferSize() const
 {
 	return sizeof(btTriangleMeshShapeData);
 }
 
-
-
-#endif //BT_BVH_TRIANGLE_MESH_SHAPE_H
+#endif  //BT_BVH_TRIANGLE_MESH_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btCapsuleShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btCapsuleShape.cpp
index 0345501ce24..7c337742841 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btCapsuleShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btCapsuleShape.cpp
@@ -13,24 +13,21 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btCapsuleShape.h"
 
 #include "LinearMath/btQuaternion.h"
 
-btCapsuleShape::btCapsuleShape(btScalar radius, btScalar height) : btConvexInternalShape ()
+btCapsuleShape::btCapsuleShape(btScalar radius, btScalar height) : btConvexInternalShape()
 {
 	m_collisionMargin = radius;
 	m_shapeType = CAPSULE_SHAPE_PROXYTYPE;
 	m_upAxis = 1;
-	m_implicitShapeDimensions.setValue(radius,0.5f*height,radius);
+	m_implicitShapeDimensions.setValue(radius, 0.5f * height, radius);
 }
 
- 
- btVector3	btCapsuleShape::localGetSupportingVertexWithoutMargin(const btVector3& vec0)const
+btVector3 btCapsuleShape::localGetSupportingVertexWithoutMargin(const btVector3& vec0) const
 {
-
-	btVector3 supVec(0,0,0);
+	btVector3 supVec(0, 0, 0);
 
 	btScalar maxDot(btScalar(-BT_LARGE_FLOAT));
 
@@ -38,20 +35,19 @@ btCapsuleShape::btCapsuleShape(btScalar radius, btScalar height) : btConvexInter
 	btScalar lenSqr = vec.length2();
 	if (lenSqr < btScalar(0.0001))
 	{
-		vec.setValue(1,0,0);
-	} else
+		vec.setValue(1, 0, 0);
+	}
+	else
 	{
-		btScalar rlen = btScalar(1.) / btSqrt(lenSqr );
+		btScalar rlen = btScalar(1.) / btSqrt(lenSqr);
 		vec *= rlen;
 	}
 
 	btVector3 vtx;
 	btScalar newDot;
-	
-	
 
 	{
-		btVector3 pos(0,0,0);
+		btVector3 pos(0, 0, 0);
 		pos[getUpAxis()] = getHalfHeight();
 
 		vtx = pos;
@@ -63,7 +59,7 @@ btCapsuleShape::btCapsuleShape(btScalar radius, btScalar height) : btConvexInter
 		}
 	}
 	{
-		btVector3 pos(0,0,0);
+		btVector3 pos(0, 0, 0);
 		pos[getUpAxis()] = -getHalfHeight();
 
 		vtx = pos;
@@ -76,15 +72,11 @@ btCapsuleShape::btCapsuleShape(btScalar radius, btScalar height) : btConvexInter
 	}
 
 	return supVec;
-
 }
 
- void	btCapsuleShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const
+void btCapsuleShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const
 {
-
-	
-	
-	for (int j=0;j<numVectors;j++)
+	for (int j = 0; j < numVectors; j++)
 	{
 		btScalar maxDot(btScalar(-BT_LARGE_FLOAT));
 		const btVector3& vec = vectors[j];
@@ -92,7 +84,7 @@ btCapsuleShape::btCapsuleShape(btScalar radius, btScalar height) : btConvexInter
 		btVector3 vtx;
 		btScalar newDot;
 		{
-			btVector3 pos(0,0,0);
+			btVector3 pos(0, 0, 0);
 			pos[getUpAxis()] = getHalfHeight();
 			vtx = pos;
 			newDot = vec.dot(vtx);
@@ -103,7 +95,7 @@ btCapsuleShape::btCapsuleShape(btScalar radius, btScalar height) : btConvexInter
 			}
 		}
 		{
-			btVector3 pos(0,0,0);
+			btVector3 pos(0, 0, 0);
 			pos[getUpAxis()] = -getHalfHeight();
 			vtx = pos;
 			newDot = vec.dot(vtx);
@@ -113,57 +105,44 @@ btCapsuleShape::btCapsuleShape(btScalar radius, btScalar height) : btConvexInter
 				supportVerticesOut[j] = vtx;
 			}
 		}
-		
 	}
 }
 
-
-void	btCapsuleShape::calculateLocalInertia(btScalar mass,btVector3& inertia) const
+void btCapsuleShape::calculateLocalInertia(btScalar mass, btVector3& inertia) const
 {
 	//as an approximation, take the inertia of the box that bounds the spheres
 
 	btTransform ident;
 	ident.setIdentity();
 
-	
 	btScalar radius = getRadius();
 
-	btVector3 halfExtents(radius,radius,radius);
-	halfExtents[getUpAxis()]+=getHalfHeight();
+	btVector3 halfExtents(radius, radius, radius);
+	halfExtents[getUpAxis()] += getHalfHeight();
 
-	btScalar lx=btScalar(2.)*(halfExtents[0]);
-	btScalar ly=btScalar(2.)*(halfExtents[1]);
-	btScalar lz=btScalar(2.)*(halfExtents[2]);
-	const btScalar x2 = lx*lx;
-	const btScalar y2 = ly*ly;
-	const btScalar z2 = lz*lz;
+	btScalar lx = btScalar(2.) * (halfExtents[0]);
+	btScalar ly = btScalar(2.) * (halfExtents[1]);
+	btScalar lz = btScalar(2.) * (halfExtents[2]);
+	const btScalar x2 = lx * lx;
+	const btScalar y2 = ly * ly;
+	const btScalar z2 = lz * lz;
 	const btScalar scaledmass = mass * btScalar(.08333333);
 
-	inertia[0] = scaledmass * (y2+z2);
-	inertia[1] = scaledmass * (x2+z2);
-	inertia[2] = scaledmass * (x2+y2);
-
+	inertia[0] = scaledmass * (y2 + z2);
+	inertia[1] = scaledmass * (x2 + z2);
+	inertia[2] = scaledmass * (x2 + y2);
 }
 
-btCapsuleShapeX::btCapsuleShapeX(btScalar radius,btScalar height)
+btCapsuleShapeX::btCapsuleShapeX(btScalar radius, btScalar height)
 {
 	m_collisionMargin = radius;
 	m_upAxis = 0;
-	m_implicitShapeDimensions.setValue(0.5f*height, radius,radius);
+	m_implicitShapeDimensions.setValue(0.5f * height, radius, radius);
 }
 
-
-
-
-
-
-btCapsuleShapeZ::btCapsuleShapeZ(btScalar radius,btScalar height)
+btCapsuleShapeZ::btCapsuleShapeZ(btScalar radius, btScalar height)
 {
 	m_collisionMargin = radius;
 	m_upAxis = 2;
-	m_implicitShapeDimensions.setValue(radius,radius,0.5f*height);
+	m_implicitShapeDimensions.setValue(radius, radius, 0.5f * height);
 }
-
-
-
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btCapsuleShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btCapsuleShape.h
index 5a3362834a5..138d0c0f7c1 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btCapsuleShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btCapsuleShape.h
@@ -17,99 +17,96 @@ subject to the following restrictions:
 #define BT_CAPSULE_SHAPE_H
 
 #include "btConvexInternalShape.h"
-#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h" // for the types
-
+#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h"  // for the types
 
 ///The btCapsuleShape represents a capsule around the Y axis, there is also the btCapsuleShapeX aligned around the X axis and btCapsuleShapeZ around the Z axis.
 ///The total height is height+2*radius, so the height is just the height between the center of each 'sphere' of the capsule caps.
 ///The btCapsuleShape is a convex hull of two spheres. The btMultiSphereShape is a more general collision shape that takes the convex hull of multiple sphere, so it can also represent a capsule when just using two spheres.
-ATTRIBUTE_ALIGNED16(class) btCapsuleShape : public btConvexInternalShape
+ATTRIBUTE_ALIGNED16(class)
+btCapsuleShape : public btConvexInternalShape
 {
 protected:
-	int	m_upAxis;
+	int m_upAxis;
 
 protected:
 	///only used for btCapsuleShapeZ and btCapsuleShapeX subclasses.
-	btCapsuleShape() : btConvexInternalShape() {m_shapeType = CAPSULE_SHAPE_PROXYTYPE;};
+	btCapsuleShape() : btConvexInternalShape() { m_shapeType = CAPSULE_SHAPE_PROXYTYPE; };
 
 public:
-	
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
-	btCapsuleShape(btScalar radius,btScalar height);
+
+	btCapsuleShape(btScalar radius, btScalar height);
 
 	///CollisionShape Interface
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
+	virtual void calculateLocalInertia(btScalar mass, btVector3 & inertia) const;
 
 	/// btConvexShape Interface
-	virtual btVector3	localGetSupportingVertexWithoutMargin(const btVector3& vec)const;
+	virtual btVector3 localGetSupportingVertexWithoutMargin(const btVector3& vec) const;
+
+	virtual void batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const;
 
-	virtual void	batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const;
-	
 	virtual void setMargin(btScalar collisionMargin)
 	{
 		//don't override the margin for capsules, their entire radius == margin
 		(void)collisionMargin;
 	}
 
-	virtual void getAabb (const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
+	virtual void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
 	{
-			btVector3 halfExtents(getRadius(),getRadius(),getRadius());
-			halfExtents[m_upAxis] = getRadius() + getHalfHeight();
-			btMatrix3x3 abs_b = t.getBasis().absolute();  
-			btVector3 center = t.getOrigin();
-            btVector3 extent = halfExtents.dot3(abs_b[0], abs_b[1], abs_b[2]);
-        
-			aabbMin = center - extent;
-			aabbMax = center + extent;
+		btVector3 halfExtents(getRadius(), getRadius(), getRadius());
+		halfExtents[m_upAxis] = getRadius() + getHalfHeight();
+		btMatrix3x3 abs_b = t.getBasis().absolute();
+		btVector3 center = t.getOrigin();
+		btVector3 extent = halfExtents.dot3(abs_b[0], abs_b[1], abs_b[2]);
+
+		aabbMin = center - extent;
+		aabbMax = center + extent;
 	}
 
-	virtual const char*	getName()const 
+	virtual const char* getName() const
 	{
 		return "CapsuleShape";
 	}
 
-	int	getUpAxis() const
+	int getUpAxis() const
 	{
 		return m_upAxis;
 	}
 
-	btScalar	getRadius() const
+	btScalar getRadius() const
 	{
-		int radiusAxis = (m_upAxis+2)%3;
+		int radiusAxis = (m_upAxis + 2) % 3;
 		return m_implicitShapeDimensions[radiusAxis];
 	}
 
-	btScalar	getHalfHeight() const
+	btScalar getHalfHeight() const
 	{
 		return m_implicitShapeDimensions[m_upAxis];
 	}
 
-	virtual void	setLocalScaling(const btVector3& scaling)
+	virtual void setLocalScaling(const btVector3& scaling)
 	{
 		btVector3 unScaledImplicitShapeDimensions = m_implicitShapeDimensions / m_localScaling;
-                btConvexInternalShape::setLocalScaling(scaling);
+		btConvexInternalShape::setLocalScaling(scaling);
 		m_implicitShapeDimensions = (unScaledImplicitShapeDimensions * scaling);
 		//update m_collisionMargin, since entire radius==margin
-		int radiusAxis = (m_upAxis+2)%3;
+		int radiusAxis = (m_upAxis + 2) % 3;
 		m_collisionMargin = m_implicitShapeDimensions[radiusAxis];
 	}
 
-	virtual btVector3	getAnisotropicRollingFrictionDirection() const
+	virtual btVector3 getAnisotropicRollingFrictionDirection() const
 	{
-		btVector3 aniDir(0,0,0);
-		aniDir[getUpAxis()]=1;
+		btVector3 aniDir(0, 0, 0);
+		aniDir[getUpAxis()] = 1;
 		return aniDir;
 	}
 
-
-	virtual	int	calculateSerializeBufferSize() const;
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
-
-	SIMD_FORCE_INLINE	void	deSerializeFloat(struct btCapsuleShapeData* dataBuffer);
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 
+	SIMD_FORCE_INLINE void deSerializeFloat(struct btCapsuleShapeData * dataBuffer);
 };
 
 ///btCapsuleShapeX represents a capsule around the Z axis
@@ -117,17 +114,13 @@ public:
 class btCapsuleShapeX : public btCapsuleShape
 {
 public:
+	btCapsuleShapeX(btScalar radius, btScalar height);
 
-	btCapsuleShapeX(btScalar radius,btScalar height);
-		
 	//debugging
-	virtual const char*	getName()const
+	virtual const char* getName() const
 	{
 		return "CapsuleX";
 	}
-
-	
-
 };
 
 ///btCapsuleShapeZ represents a capsule around the Z axis
@@ -135,38 +128,36 @@ public:
 class btCapsuleShapeZ : public btCapsuleShape
 {
 public:
-	btCapsuleShapeZ(btScalar radius,btScalar height);
+	btCapsuleShapeZ(btScalar radius, btScalar height);
 
-		//debugging
-	virtual const char*	getName()const
+	//debugging
+	virtual const char* getName() const
 	{
 		return "CapsuleZ";
 	}
-
-	
 };
 
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
-struct	btCapsuleShapeData
+struct btCapsuleShapeData
 {
-	btConvexInternalShapeData	m_convexInternalShapeData;
+	btConvexInternalShapeData m_convexInternalShapeData;
 
-	int	m_upAxis;
+	int m_upAxis;
 
-	char	m_padding[4];
+	char m_padding[4];
 };
 
-SIMD_FORCE_INLINE	int	btCapsuleShape::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btCapsuleShape::calculateSerializeBufferSize() const
 {
 	return sizeof(btCapsuleShapeData);
 }
 
-	///fills the dataBuffer and returns the struct name (and 0 on failure)
-SIMD_FORCE_INLINE	const char*	btCapsuleShape::serialize(void* dataBuffer, btSerializer* serializer) const
+///fills the dataBuffer and returns the struct name (and 0 on failure)
+SIMD_FORCE_INLINE const char* btCapsuleShape::serialize(void* dataBuffer, btSerializer* serializer) const
 {
-	btCapsuleShapeData* shapeData = (btCapsuleShapeData*) dataBuffer;
+	btCapsuleShapeData* shapeData = (btCapsuleShapeData*)dataBuffer;
 
-	btConvexInternalShape::serialize(&shapeData->m_convexInternalShapeData,serializer);
+	btConvexInternalShape::serialize(&shapeData->m_convexInternalShapeData, serializer);
 
 	shapeData->m_upAxis = m_upAxis;
 
@@ -179,7 +170,7 @@ SIMD_FORCE_INLINE	const char*	btCapsuleShape::serialize(void* dataBuffer, btSeri
 	return "btCapsuleShapeData";
 }
 
-SIMD_FORCE_INLINE	void	btCapsuleShape::deSerializeFloat(btCapsuleShapeData* dataBuffer)
+SIMD_FORCE_INLINE void btCapsuleShape::deSerializeFloat(btCapsuleShapeData* dataBuffer)
 {
 	m_implicitShapeDimensions.deSerializeFloat(dataBuffer->m_convexInternalShapeData.m_implicitShapeDimensions);
 	m_collisionMargin = dataBuffer->m_convexInternalShapeData.m_collisionMargin;
@@ -188,4 +179,4 @@ SIMD_FORCE_INLINE	void	btCapsuleShape::deSerializeFloat(btCapsuleShapeData* data
 	m_upAxis = dataBuffer->m_upAxis;
 }
 
-#endif //BT_CAPSULE_SHAPE_H
+#endif  //BT_CAPSULE_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btCollisionMargin.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btCollisionMargin.h
index 474bf1fb499..abd8ab3eb5f 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btCollisionMargin.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btCollisionMargin.h
@@ -19,9 +19,6 @@ subject to the following restrictions:
 ///The CONVEX_DISTANCE_MARGIN is a default collision margin for convex collision shapes derived from btConvexInternalShape.
 ///This collision margin is used by Gjk and some other algorithms
 ///Note that when creating small objects, you need to make sure to set a smaller collision margin, using the 'setMargin' API
-#define CONVEX_DISTANCE_MARGIN btScalar(0.04)// btScalar(0.1)//;//btScalar(0.01)
-
-
-
-#endif //BT_COLLISION_MARGIN_H
+#define CONVEX_DISTANCE_MARGIN btScalar(0.04)  // btScalar(0.1)//;//btScalar(0.01)
 
+#endif  //BT_COLLISION_MARGIN_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btCollisionShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btCollisionShape.cpp
index 823e2788f2e..0b3640a65be 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btCollisionShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btCollisionShape.cpp
@@ -20,47 +20,44 @@ subject to the following restrictions:
   can be used by probes that are checking whether the
   library is actually installed.
 */
-extern "C" 
+extern "C"
 {
-void btBulletCollisionProbe ();
+	void btBulletCollisionProbe();
 
-void btBulletCollisionProbe () {}
+	void btBulletCollisionProbe() {}
 }
 
-
-
-void	btCollisionShape::getBoundingSphere(btVector3& center,btScalar& radius) const
+void btCollisionShape::getBoundingSphere(btVector3& center, btScalar& radius) const
 {
 	btTransform tr;
 	tr.setIdentity();
-	btVector3 aabbMin,aabbMax;
+	btVector3 aabbMin, aabbMax;
 
-	getAabb(tr,aabbMin,aabbMax);
+	getAabb(tr, aabbMin, aabbMax);
 
-	radius = (aabbMax-aabbMin).length()*btScalar(0.5);
-	center = (aabbMin+aabbMax)*btScalar(0.5);
+	radius = (aabbMax - aabbMin).length() * btScalar(0.5);
+	center = (aabbMin + aabbMax) * btScalar(0.5);
 }
 
-
-btScalar	btCollisionShape::getContactBreakingThreshold(btScalar defaultContactThreshold) const
+btScalar btCollisionShape::getContactBreakingThreshold(btScalar defaultContactThreshold) const
 {
 	return getAngularMotionDisc() * defaultContactThreshold;
 }
 
-btScalar	btCollisionShape::getAngularMotionDisc() const
+btScalar btCollisionShape::getAngularMotionDisc() const
 {
 	///@todo cache this value, to improve performance
-	btVector3	center;
+	btVector3 center;
 	btScalar disc;
-	getBoundingSphere(center,disc);
+	getBoundingSphere(center, disc);
 	disc += (center).length();
 	return disc;
 }
 
-void btCollisionShape::calculateTemporalAabb(const btTransform& curTrans,const btVector3& linvel,const btVector3& angvel,btScalar timeStep, btVector3& temporalAabbMin,btVector3& temporalAabbMax) const
+void btCollisionShape::calculateTemporalAabb(const btTransform& curTrans, const btVector3& linvel, const btVector3& angvel, btScalar timeStep, btVector3& temporalAabbMin, btVector3& temporalAabbMax) const
 {
 	//start with static aabb
-	getAabb(curTrans,temporalAabbMin,temporalAabbMax);
+	getAabb(curTrans, temporalAabbMin, temporalAabbMax);
 
 	btScalar temporalAabbMaxx = temporalAabbMax.getX();
 	btScalar temporalAabbMaxy = temporalAabbMax.getY();
@@ -70,36 +67,36 @@ void btCollisionShape::calculateTemporalAabb(const btTransform& curTrans,const b
 	btScalar temporalAabbMinz = temporalAabbMin.getZ();
 
 	// add linear motion
-	btVector3 linMotion = linvel*timeStep;
+	btVector3 linMotion = linvel * timeStep;
 	///@todo: simd would have a vector max/min operation, instead of per-element access
 	if (linMotion.x() > btScalar(0.))
-		temporalAabbMaxx += linMotion.x(); 
+		temporalAabbMaxx += linMotion.x();
 	else
 		temporalAabbMinx += linMotion.x();
 	if (linMotion.y() > btScalar(0.))
-		temporalAabbMaxy += linMotion.y(); 
+		temporalAabbMaxy += linMotion.y();
 	else
 		temporalAabbMiny += linMotion.y();
 	if (linMotion.z() > btScalar(0.))
-		temporalAabbMaxz += linMotion.z(); 
+		temporalAabbMaxz += linMotion.z();
 	else
 		temporalAabbMinz += linMotion.z();
 
 	//add conservative angular motion
 	btScalar angularMotion = angvel.length() * getAngularMotionDisc() * timeStep;
-	btVector3 angularMotion3d(angularMotion,angularMotion,angularMotion);
-	temporalAabbMin = btVector3(temporalAabbMinx,temporalAabbMiny,temporalAabbMinz);
-	temporalAabbMax = btVector3(temporalAabbMaxx,temporalAabbMaxy,temporalAabbMaxz);
+	btVector3 angularMotion3d(angularMotion, angularMotion, angularMotion);
+	temporalAabbMin = btVector3(temporalAabbMinx, temporalAabbMiny, temporalAabbMinz);
+	temporalAabbMax = btVector3(temporalAabbMaxx, temporalAabbMaxy, temporalAabbMaxz);
 
 	temporalAabbMin -= angularMotion3d;
 	temporalAabbMax += angularMotion3d;
 }
 
 ///fills the dataBuffer and returns the struct name (and 0 on failure)
-const char*	btCollisionShape::serialize(void* dataBuffer, btSerializer* serializer) const
+const char* btCollisionShape::serialize(void* dataBuffer, btSerializer* serializer) const
 {
-	btCollisionShapeData* shapeData = (btCollisionShapeData*) dataBuffer;
-	char* name = (char*) serializer->findNameForPointer(this);
+	btCollisionShapeData* shapeData = (btCollisionShapeData*)dataBuffer;
+	char* name = (char*)serializer->findNameForPointer(this);
 	shapeData->m_name = (char*)serializer->getUniquePointer(name);
 	if (shapeData->m_name)
 	{
@@ -113,10 +110,10 @@ const char*	btCollisionShape::serialize(void* dataBuffer, btSerializer* serializ
 	return "btCollisionShapeData";
 }
 
-void	btCollisionShape::serializeSingleShape(btSerializer* serializer) const
+void btCollisionShape::serializeSingleShape(btSerializer* serializer) const
 {
 	int len = calculateSerializeBufferSize();
-	btChunk* chunk = serializer->allocate(len,1);
+	btChunk* chunk = serializer->allocate(len, 1);
 	const char* structType = serialize(chunk->m_oldPtr, serializer);
-	serializer->finalizeChunk(chunk,structType,BT_SHAPE_CODE,(void*)this);
+	serializer->finalizeChunk(chunk, structType, BT_SHAPE_CODE, (void*)this);
 }
\ No newline at end of file
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btCollisionShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btCollisionShape.h
index 6c4916fbd4b..c80e105a4d8 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btCollisionShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btCollisionShape.h
@@ -19,12 +19,12 @@ subject to the following restrictions:
 #include "LinearMath/btTransform.h"
 #include "LinearMath/btVector3.h"
 #include "LinearMath/btMatrix3x3.h"
-#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h" //for the shape types
+#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h"  //for the shape types
 class btSerializer;
 
-
 ///The btCollisionShape class provides an interface for collision shapes that can be shared among btCollisionObjects.
-ATTRIBUTE_ALIGNED16(class) btCollisionShape
+ATTRIBUTE_ALIGNED16(class)
+btCollisionShape
 {
 protected:
 	int m_shapeType;
@@ -32,10 +32,9 @@ protected:
 	int m_userIndex;
 
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-	btCollisionShape() : m_shapeType (INVALID_SHAPE_PROXYTYPE), m_userPointer(0), m_userIndex(-1)
+	btCollisionShape() : m_shapeType(INVALID_SHAPE_PROXYTYPE), m_userPointer(0), m_userIndex(-1)
 	{
 	}
 
@@ -44,50 +43,47 @@ public:
 	}
 
 	///getAabb returns the axis aligned bounding box in the coordinate frame of the given transform t.
-	virtual void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const =0;
+	virtual void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const = 0;
 
-	virtual void	getBoundingSphere(btVector3& center,btScalar& radius) const;
+	virtual void getBoundingSphere(btVector3 & center, btScalar & radius) const;
 
 	///getAngularMotionDisc returns the maximum radius needed for Conservative Advancement to handle time-of-impact with rotations.
-	virtual btScalar	getAngularMotionDisc() const;
-
-	virtual btScalar	getContactBreakingThreshold(btScalar defaultContactThresholdFactor) const;
+	virtual btScalar getAngularMotionDisc() const;
 
+	virtual btScalar getContactBreakingThreshold(btScalar defaultContactThresholdFactor) const;
 
 	///calculateTemporalAabb calculates the enclosing aabb for the moving object over interval [0..timeStep)
 	///result is conservative
-	void calculateTemporalAabb(const btTransform& curTrans,const btVector3& linvel,const btVector3& angvel,btScalar timeStep, btVector3& temporalAabbMin,btVector3& temporalAabbMax) const;
+	void calculateTemporalAabb(const btTransform& curTrans, const btVector3& linvel, const btVector3& angvel, btScalar timeStep, btVector3& temporalAabbMin, btVector3& temporalAabbMax) const;
 
-
-
-	SIMD_FORCE_INLINE bool	isPolyhedral() const
+	SIMD_FORCE_INLINE bool isPolyhedral() const
 	{
 		return btBroadphaseProxy::isPolyhedral(getShapeType());
 	}
 
-	SIMD_FORCE_INLINE bool	isConvex2d() const
+	SIMD_FORCE_INLINE bool isConvex2d() const
 	{
 		return btBroadphaseProxy::isConvex2d(getShapeType());
 	}
 
-	SIMD_FORCE_INLINE bool	isConvex() const
+	SIMD_FORCE_INLINE bool isConvex() const
 	{
 		return btBroadphaseProxy::isConvex(getShapeType());
 	}
-	SIMD_FORCE_INLINE bool	isNonMoving() const
+	SIMD_FORCE_INLINE bool isNonMoving() const
 	{
 		return btBroadphaseProxy::isNonMoving(getShapeType());
 	}
-	SIMD_FORCE_INLINE bool	isConcave() const
+	SIMD_FORCE_INLINE bool isConcave() const
 	{
 		return btBroadphaseProxy::isConcave(getShapeType());
 	}
-	SIMD_FORCE_INLINE bool	isCompound() const
+	SIMD_FORCE_INLINE bool isCompound() const
 	{
 		return btBroadphaseProxy::isCompound(getShapeType());
 	}
 
-	SIMD_FORCE_INLINE bool	isSoftBody() const
+	SIMD_FORCE_INLINE bool isSoftBody() const
 	{
 		return btBroadphaseProxy::isSoftBody(getShapeType());
 	}
@@ -99,35 +95,35 @@ public:
 	}
 
 #ifndef __SPU__
-	virtual void	setLocalScaling(const btVector3& scaling) =0;
-	virtual const btVector3& getLocalScaling() const =0;
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const = 0;
+	virtual void setLocalScaling(const btVector3& scaling) = 0;
+	virtual const btVector3& getLocalScaling() const = 0;
+	virtual void calculateLocalInertia(btScalar mass, btVector3 & inertia) const = 0;
 
+	//debugging support
+	virtual const char* getName() const = 0;
+#endif  //__SPU__
 
-//debugging support
-	virtual const char*	getName()const =0 ;
-#endif //__SPU__
-
-	
-	int		getShapeType() const { return m_shapeType; }
+	int getShapeType() const
+	{
+		return m_shapeType;
+	}
 
 	///the getAnisotropicRollingFrictionDirection can be used in combination with setAnisotropicFriction
 	///See Bullet/Demos/RollingFrictionDemo for an example
-	virtual btVector3	getAnisotropicRollingFrictionDirection() const
+	virtual btVector3 getAnisotropicRollingFrictionDirection() const
 	{
-		return btVector3(1,1,1);
+		return btVector3(1, 1, 1);
 	}
-	virtual void	setMargin(btScalar margin) = 0;
-	virtual btScalar	getMargin() const = 0;
+	virtual void setMargin(btScalar margin) = 0;
+	virtual btScalar getMargin() const = 0;
 
-	
 	///optional user data pointer
-	void	setUserPointer(void*  userPtr)
+	void setUserPointer(void* userPtr)
 	{
 		m_userPointer = userPtr;
 	}
 
-	void*	getUserPointer() const
+	void* getUserPointer() const
 	{
 		return m_userPointer;
 	}
@@ -141,16 +137,16 @@ public:
 		return m_userIndex;
 	}
 
-
-	virtual	int	calculateSerializeBufferSize() const;
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 
-	virtual void	serializeSingleShape(btSerializer* serializer) const;
-
-};	
+	virtual void serializeSingleShape(btSerializer * serializer) const;
+};
 
+// clang-format off
+// parser needs * with the name
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
 struct	btCollisionShapeData
 {
@@ -158,13 +154,10 @@ struct	btCollisionShapeData
 	int		m_shapeType;
 	char	m_padding[4];
 };
-
-SIMD_FORCE_INLINE	int	btCollisionShape::calculateSerializeBufferSize() const
+// clang-format on
+SIMD_FORCE_INLINE int btCollisionShape::calculateSerializeBufferSize() const
 {
 	return sizeof(btCollisionShapeData);
 }
 
-
-
-#endif //BT_COLLISION_SHAPE_H
-
+#endif  //BT_COLLISION_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btCompoundShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btCompoundShape.cpp
index 85572da307d..fd7828b1047 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btCompoundShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btCompoundShape.cpp
@@ -19,26 +19,25 @@ subject to the following restrictions:
 #include "LinearMath/btSerializer.h"
 
 btCompoundShape::btCompoundShape(bool enableDynamicAabbTree, const int initialChildCapacity)
-: m_localAabbMin(btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT)),
-m_localAabbMax(btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT)),
-m_dynamicAabbTree(0),
-m_updateRevision(1),
-m_collisionMargin(btScalar(0.)),
-m_localScaling(btScalar(1.),btScalar(1.),btScalar(1.))
+	: m_localAabbMin(btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT)),
+	  m_localAabbMax(btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT)),
+	  m_dynamicAabbTree(0),
+	  m_updateRevision(1),
+	  m_collisionMargin(btScalar(0.)),
+	  m_localScaling(btScalar(1.), btScalar(1.), btScalar(1.))
 {
 	m_shapeType = COMPOUND_SHAPE_PROXYTYPE;
 
 	if (enableDynamicAabbTree)
 	{
-		void* mem = btAlignedAlloc(sizeof(btDbvt),16);
-		m_dynamicAabbTree = new(mem) btDbvt();
-		btAssert(mem==m_dynamicAabbTree);
+		void* mem = btAlignedAlloc(sizeof(btDbvt), 16);
+		m_dynamicAabbTree = new (mem) btDbvt();
+		btAssert(mem == m_dynamicAabbTree);
 	}
 
 	m_children.reserve(initialChildCapacity);
 }
 
-
 btCompoundShape::~btCompoundShape()
 {
 	if (m_dynamicAabbTree)
@@ -48,7 +47,7 @@ btCompoundShape::~btCompoundShape()
 	}
 }
 
-void	btCompoundShape::addChildShape(const btTransform& localTransform,btCollisionShape* shape)
+void btCompoundShape::addChildShape(const btTransform& localTransform, btCollisionShape* shape)
 {
 	m_updateRevision++;
 	//m_childTransforms.push_back(localTransform);
@@ -60,11 +59,10 @@ void	btCompoundShape::addChildShape(const btTransform& localTransform,btCollisio
 	child.m_childShapeType = shape->getShapeType();
 	child.m_childMargin = shape->getMargin();
 
-	
 	//extend the local aabbMin/aabbMax
-	btVector3 localAabbMin,localAabbMax;
-	shape->getAabb(localTransform,localAabbMin,localAabbMax);
-	for (int i=0;i<3;i++)
+	btVector3 localAabbMin, localAabbMax;
+	shape->getAabb(localTransform, localAabbMin, localAabbMax);
+	for (int i = 0; i < 3; i++)
 	{
 		if (m_localAabbMin[i] > localAabbMin[i])
 		{
@@ -74,31 +72,30 @@ void	btCompoundShape::addChildShape(const btTransform& localTransform,btCollisio
 		{
 			m_localAabbMax[i] = localAabbMax[i];
 		}
-
 	}
 	if (m_dynamicAabbTree)
 	{
-		const btDbvtVolume	bounds=btDbvtVolume::FromMM(localAabbMin,localAabbMax);
+		const btDbvtVolume bounds = btDbvtVolume::FromMM(localAabbMin, localAabbMax);
 		size_t index = m_children.size();
-		child.m_node = m_dynamicAabbTree->insert(bounds,reinterpret_cast<void*>(index) );
+		child.m_node = m_dynamicAabbTree->insert(bounds, reinterpret_cast<void*>(index));
 	}
 
 	m_children.push_back(child);
-
 }
 
-void	btCompoundShape::updateChildTransform(int childIndex, const btTransform& newChildTransform,bool shouldRecalculateLocalAabb)
+void btCompoundShape::updateChildTransform(int childIndex, const btTransform& newChildTransform, bool shouldRecalculateLocalAabb)
 {
 	m_children[childIndex].m_transform = newChildTransform;
 
 	if (m_dynamicAabbTree)
 	{
 		///update the dynamic aabb tree
-		btVector3 localAabbMin,localAabbMax;
-		m_children[childIndex].m_childShape->getAabb(newChildTransform,localAabbMin,localAabbMax);
-		ATTRIBUTE_ALIGNED16(btDbvtVolume)	bounds=btDbvtVolume::FromMM(localAabbMin,localAabbMax);
+		btVector3 localAabbMin, localAabbMax;
+		m_children[childIndex].m_childShape->getAabb(newChildTransform, localAabbMin, localAabbMax);
+		ATTRIBUTE_ALIGNED16(btDbvtVolume)
+		bounds = btDbvtVolume::FromMM(localAabbMin, localAabbMax);
 		//int index = m_children.size()-1;
-		m_dynamicAabbTree->update(m_children[childIndex].m_node,bounds);
+		m_dynamicAabbTree->update(m_children[childIndex].m_node, bounds);
 	}
 
 	if (shouldRecalculateLocalAabb)
@@ -110,35 +107,30 @@ void	btCompoundShape::updateChildTransform(int childIndex, const btTransform& ne
 void btCompoundShape::removeChildShapeByIndex(int childShapeIndex)
 {
 	m_updateRevision++;
-	btAssert(childShapeIndex >=0 && childShapeIndex < m_children.size());
+	btAssert(childShapeIndex >= 0 && childShapeIndex < m_children.size());
 	if (m_dynamicAabbTree)
 	{
 		m_dynamicAabbTree->remove(m_children[childShapeIndex].m_node);
 	}
-	m_children.swap(childShapeIndex,m_children.size()-1);
-    if (m_dynamicAabbTree) 
+	m_children.swap(childShapeIndex, m_children.size() - 1);
+	if (m_dynamicAabbTree)
 		m_children[childShapeIndex].m_node->dataAsInt = childShapeIndex;
 	m_children.pop_back();
-
 }
 
-
-
 void btCompoundShape::removeChildShape(btCollisionShape* shape)
 {
 	m_updateRevision++;
 	// Find the children containing the shape specified, and remove those children.
 	//note: there might be multiple children using the same shape!
-	for(int i = m_children.size()-1; i >= 0 ; i--)
+	for (int i = m_children.size() - 1; i >= 0; i--)
 	{
-		if(m_children[i].m_childShape == shape)
+		if (m_children[i].m_childShape == shape)
 		{
 			removeChildShapeByIndex(i);
 		}
 	}
 
-
-
 	recalculateLocalAabb();
 }
 
@@ -147,15 +139,15 @@ void btCompoundShape::recalculateLocalAabb()
 	// Recalculate the local aabb
 	// Brute force, it iterates over all the shapes left.
 
-	m_localAabbMin = btVector3(btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT));
-	m_localAabbMax = btVector3(btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT));
+	m_localAabbMin = btVector3(btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT));
+	m_localAabbMax = btVector3(btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT));
 
 	//extend the local aabbMin/aabbMax
 	for (int j = 0; j < m_children.size(); j++)
 	{
-		btVector3 localAabbMin,localAabbMax;
+		btVector3 localAabbMin, localAabbMax;
 		m_children[j].m_childShape->getAabb(m_children[j].m_transform, localAabbMin, localAabbMax);
-		for (int i=0;i<3;i++)
+		for (int i = 0; i < 3; i++)
 		{
 			if (m_localAabbMin[i] > localAabbMin[i])
 				m_localAabbMin[i] = localAabbMin[i];
@@ -166,53 +158,47 @@ void btCompoundShape::recalculateLocalAabb()
 }
 
 ///getAabb's default implementation is brute force, expected derived classes to implement a fast dedicated version
-void btCompoundShape::getAabb(const btTransform& trans,btVector3& aabbMin,btVector3& aabbMax) const
+void btCompoundShape::getAabb(const btTransform& trans, btVector3& aabbMin, btVector3& aabbMax) const
 {
-	btVector3 localHalfExtents = btScalar(0.5)*(m_localAabbMax-m_localAabbMin);
-	btVector3 localCenter = btScalar(0.5)*(m_localAabbMax+m_localAabbMin);
-	
+	btVector3 localHalfExtents = btScalar(0.5) * (m_localAabbMax - m_localAabbMin);
+	btVector3 localCenter = btScalar(0.5) * (m_localAabbMax + m_localAabbMin);
+
 	//avoid an illegal AABB when there are no children
 	if (!m_children.size())
 	{
-		localHalfExtents.setValue(0,0,0);
-		localCenter.setValue(0,0,0);
+		localHalfExtents.setValue(0, 0, 0);
+		localCenter.setValue(0, 0, 0);
 	}
-	localHalfExtents += btVector3(getMargin(),getMargin(),getMargin());
-		
+	localHalfExtents += btVector3(getMargin(), getMargin(), getMargin());
 
-	btMatrix3x3 abs_b = trans.getBasis().absolute();  
+	btMatrix3x3 abs_b = trans.getBasis().absolute();
 
 	btVector3 center = trans(localCenter);
 
-    btVector3 extent = localHalfExtents.dot3(abs_b[0], abs_b[1], abs_b[2]);
-	aabbMin = center-extent;
-	aabbMax = center+extent;
-	
+	btVector3 extent = localHalfExtents.dot3(abs_b[0], abs_b[1], abs_b[2]);
+	aabbMin = center - extent;
+	aabbMax = center + extent;
 }
 
-void	btCompoundShape::calculateLocalInertia(btScalar mass,btVector3& inertia) const
+void btCompoundShape::calculateLocalInertia(btScalar mass, btVector3& inertia) const
 {
 	//approximation: take the inertia from the aabb for now
 	btTransform ident;
 	ident.setIdentity();
-	btVector3 aabbMin,aabbMax;
-	getAabb(ident,aabbMin,aabbMax);
+	btVector3 aabbMin, aabbMax;
+	getAabb(ident, aabbMin, aabbMax);
 
-	btVector3 halfExtents = (aabbMax-aabbMin)*btScalar(0.5);
+	btVector3 halfExtents = (aabbMax - aabbMin) * btScalar(0.5);
 
-	btScalar lx=btScalar(2.)*(halfExtents.x());
-	btScalar ly=btScalar(2.)*(halfExtents.y());
-	btScalar lz=btScalar(2.)*(halfExtents.z());
-
-	inertia[0] = mass/(btScalar(12.0)) * (ly*ly + lz*lz);
-	inertia[1] = mass/(btScalar(12.0)) * (lx*lx + lz*lz);
-	inertia[2] = mass/(btScalar(12.0)) * (lx*lx + ly*ly);
+	btScalar lx = btScalar(2.) * (halfExtents.x());
+	btScalar ly = btScalar(2.) * (halfExtents.y());
+	btScalar lz = btScalar(2.) * (halfExtents.z());
 
+	inertia[0] = mass / (btScalar(12.0)) * (ly * ly + lz * lz);
+	inertia[1] = mass / (btScalar(12.0)) * (lx * lx + lz * lz);
+	inertia[2] = mass / (btScalar(12.0)) * (lx * lx + ly * ly);
 }
 
-
-
-
 void btCompoundShape::calculatePrincipalAxisTransform(const btScalar* masses, btTransform& principal, btVector3& inertia) const
 {
 	int n = m_children.size();
@@ -223,18 +209,18 @@ void btCompoundShape::calculatePrincipalAxisTransform(const btScalar* masses, bt
 
 	for (k = 0; k < n; k++)
 	{
-		btAssert(masses[k]>0);
+		btAssert(masses[k] > 0);
 		center += m_children[k].m_transform.getOrigin() * masses[k];
 		totalMass += masses[k];
 	}
 
-	btAssert(totalMass>0);
+	btAssert(totalMass > 0);
 
 	center /= totalMass;
 	principal.setOrigin(center);
 
 	btMatrix3x3 tensor(0, 0, 0, 0, 0, 0, 0, 0, 0);
-	for ( k = 0; k < n; k++)
+	for (k = 0; k < n; k++)
 	{
 		btVector3 i;
 		m_children[k].m_childShape->calculateLocalInertia(masses[k], i);
@@ -259,8 +245,8 @@ void btCompoundShape::calculatePrincipalAxisTransform(const btScalar* masses, bt
 		j[0].setValue(o2, 0, 0);
 		j[1].setValue(0, o2, 0);
 		j[2].setValue(0, 0, o2);
-		j[0] += o * -o.x(); 
-		j[1] += o * -o.y(); 
+		j[0] += o * -o.x();
+		j[1] += o * -o.y();
 		j[2] += o * -o.z();
 
 		//add inertia tensor of pointmass
@@ -273,59 +259,50 @@ void btCompoundShape::calculatePrincipalAxisTransform(const btScalar* masses, bt
 	inertia.setValue(tensor[0][0], tensor[1][1], tensor[2][2]);
 }
 
-
-
-
-
 void btCompoundShape::setLocalScaling(const btVector3& scaling)
 {
-
-	for(int i = 0; i < m_children.size(); i++)
+	for (int i = 0; i < m_children.size(); i++)
 	{
 		btTransform childTrans = getChildTransform(i);
 		btVector3 childScale = m_children[i].m_childShape->getLocalScaling();
-//		childScale = childScale * (childTrans.getBasis() * scaling);
+		//		childScale = childScale * (childTrans.getBasis() * scaling);
 		childScale = childScale * scaling / m_localScaling;
 		m_children[i].m_childShape->setLocalScaling(childScale);
 		childTrans.setOrigin((childTrans.getOrigin()) * scaling / m_localScaling);
-		updateChildTransform(i, childTrans,false);
+		updateChildTransform(i, childTrans, false);
 	}
-	
+
 	m_localScaling = scaling;
 	recalculateLocalAabb();
-
 }
 
-
 void btCompoundShape::createAabbTreeFromChildren()
 {
-    if ( !m_dynamicAabbTree )
-    {
-        void* mem = btAlignedAlloc(sizeof(btDbvt),16);
-        m_dynamicAabbTree = new(mem) btDbvt();
-        btAssert(mem==m_dynamicAabbTree);
+	if (!m_dynamicAabbTree)
+	{
+		void* mem = btAlignedAlloc(sizeof(btDbvt), 16);
+		m_dynamicAabbTree = new (mem) btDbvt();
+		btAssert(mem == m_dynamicAabbTree);
 
-        for ( int index = 0; index < m_children.size(); index++ )
-        {
-            btCompoundShapeChild &child = m_children[index];
+		for (int index = 0; index < m_children.size(); index++)
+		{
+			btCompoundShapeChild& child = m_children[index];
 
-            //extend the local aabbMin/aabbMax
-            btVector3 localAabbMin,localAabbMax;
-            child.m_childShape->getAabb(child.m_transform,localAabbMin,localAabbMax);
+			//extend the local aabbMin/aabbMax
+			btVector3 localAabbMin, localAabbMax;
+			child.m_childShape->getAabb(child.m_transform, localAabbMin, localAabbMax);
 
-            const btDbvtVolume  bounds=btDbvtVolume::FromMM(localAabbMin,localAabbMax);
+			const btDbvtVolume bounds = btDbvtVolume::FromMM(localAabbMin, localAabbMax);
 			size_t index2 = index;
-            child.m_node = m_dynamicAabbTree->insert(bounds, reinterpret_cast<void*>(index2) );
-        }
-    }
+			child.m_node = m_dynamicAabbTree->insert(bounds, reinterpret_cast<void*>(index2));
+		}
+	}
 }
 
-
 ///fills the dataBuffer and returns the struct name (and 0 on failure)
-const char*	btCompoundShape::serialize(void* dataBuffer, btSerializer* serializer) const
+const char* btCompoundShape::serialize(void* dataBuffer, btSerializer* serializer) const
 {
-
-	btCompoundShapeData* shapeData = (btCompoundShapeData*) dataBuffer;
+	btCompoundShapeData* shapeData = (btCompoundShapeData*)dataBuffer;
 	btCollisionShape::serialize(&shapeData->m_collisionShapeData, serializer);
 
 	shapeData->m_collisionMargin = float(m_collisionMargin);
@@ -333,27 +310,26 @@ const char*	btCompoundShape::serialize(void* dataBuffer, btSerializer* serialize
 	shapeData->m_childShapePtr = 0;
 	if (shapeData->m_numChildShapes)
 	{
-		btChunk* chunk = serializer->allocate(sizeof(btCompoundShapeChildData),shapeData->m_numChildShapes);
+		btChunk* chunk = serializer->allocate(sizeof(btCompoundShapeChildData), shapeData->m_numChildShapes);
 		btCompoundShapeChildData* memPtr = (btCompoundShapeChildData*)chunk->m_oldPtr;
 		shapeData->m_childShapePtr = (btCompoundShapeChildData*)serializer->getUniquePointer(memPtr);
 
-		for (int i=0;i<shapeData->m_numChildShapes;i++,memPtr++)
+		for (int i = 0; i < shapeData->m_numChildShapes; i++, memPtr++)
 		{
 			memPtr->m_childMargin = float(m_children[i].m_childMargin);
 			memPtr->m_childShape = (btCollisionShapeData*)serializer->getUniquePointer(m_children[i].m_childShape);
 			//don't serialize shapes that already have been serialized
 			if (!serializer->findPointer(m_children[i].m_childShape))
 			{
-				btChunk* chunk = serializer->allocate(m_children[i].m_childShape->calculateSerializeBufferSize(),1);
-				const char* structType = m_children[i].m_childShape->serialize(chunk->m_oldPtr,serializer);
-				serializer->finalizeChunk(chunk,structType,BT_SHAPE_CODE,m_children[i].m_childShape);
-			} 
+				btChunk* chunk = serializer->allocate(m_children[i].m_childShape->calculateSerializeBufferSize(), 1);
+				const char* structType = m_children[i].m_childShape->serialize(chunk->m_oldPtr, serializer);
+				serializer->finalizeChunk(chunk, structType, BT_SHAPE_CODE, m_children[i].m_childShape);
+			}
 
 			memPtr->m_childShapeType = m_children[i].m_childShapeType;
 			m_children[i].m_transform.serializeFloat(memPtr->m_transform);
 		}
-		serializer->finalizeChunk(chunk,"btCompoundShapeChildData",BT_ARRAY_CODE,chunk->m_oldPtr);
+		serializer->finalizeChunk(chunk, "btCompoundShapeChildData", BT_ARRAY_CODE, chunk->m_oldPtr);
 	}
 	return "btCompoundShapeData";
 }
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btCompoundShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btCompoundShape.h
index 2cbcd1bfca2..7e2d0eb8170 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btCompoundShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btCompoundShape.h
@@ -27,45 +27,47 @@ subject to the following restrictions:
 //class btOptimizedBvh;
 struct btDbvt;
 
-ATTRIBUTE_ALIGNED16(struct) btCompoundShapeChild
+ATTRIBUTE_ALIGNED16(struct)
+btCompoundShapeChild
 {
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-	btTransform			m_transform;
-	btCollisionShape*	m_childShape;
-	int					m_childShapeType;
-	btScalar			m_childMargin;
-	struct btDbvtNode*	m_node;
+	btTransform m_transform;
+	btCollisionShape* m_childShape;
+	int m_childShapeType;
+	btScalar m_childMargin;
+	struct btDbvtNode* m_node;
 };
 
 SIMD_FORCE_INLINE bool operator==(const btCompoundShapeChild& c1, const btCompoundShapeChild& c2)
 {
-	return  ( c1.m_transform      == c2.m_transform &&
-		c1.m_childShape     == c2.m_childShape &&
-		c1.m_childShapeType == c2.m_childShapeType &&
-		c1.m_childMargin    == c2.m_childMargin );
+	return (c1.m_transform == c2.m_transform &&
+			c1.m_childShape == c2.m_childShape &&
+			c1.m_childShapeType == c2.m_childShapeType &&
+			c1.m_childMargin == c2.m_childMargin);
 }
 
 /// The btCompoundShape allows to store multiple other btCollisionShapes
 /// This allows for moving concave collision objects. This is more general then the static concave btBvhTriangleMeshShape.
-/// It has an (optional) dynamic aabb tree to accelerate early rejection tests. 
+/// It has an (optional) dynamic aabb tree to accelerate early rejection tests.
 /// @todo: This aabb tree can also be use to speed up ray tests on btCompoundShape, see http://code.google.com/p/bullet/issues/detail?id=25
 /// Currently, removal of child shapes is only supported when disabling the aabb tree (pass 'false' in the constructor of btCompoundShape)
-ATTRIBUTE_ALIGNED16(class) btCompoundShape	: public btCollisionShape
+ATTRIBUTE_ALIGNED16(class)
+btCompoundShape : public btCollisionShape
 {
 protected:
 	btAlignedObjectArray<btCompoundShapeChild> m_children;
-	btVector3						m_localAabbMin;
-	btVector3						m_localAabbMax;
+	btVector3 m_localAabbMin;
+	btVector3 m_localAabbMax;
 
-	btDbvt*							m_dynamicAabbTree;
+	btDbvt* m_dynamicAabbTree;
 
 	///increment m_updateRevision when adding/removing/replacing child shapes, so that some caches can be updated
-	int								m_updateRevision;
+	int m_updateRevision;
 
-	btScalar	m_collisionMargin;
+	btScalar m_collisionMargin;
 
-	btVector3	m_localScaling;
+	btVector3 m_localScaling;
 
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
@@ -74,17 +76,16 @@ public:
 
 	virtual ~btCompoundShape();
 
-	void	addChildShape(const btTransform& localTransform,btCollisionShape* shape);
+	void addChildShape(const btTransform& localTransform, btCollisionShape* shape);
 
 	/// Remove all children shapes that contain the specified shape
-	virtual void removeChildShape(btCollisionShape* shape);
+	virtual void removeChildShape(btCollisionShape * shape);
 
 	void removeChildShapeByIndex(int childShapeindex);
 
-
-	int		getNumChildShapes() const
+	int getNumChildShapes() const
 	{
-		return int (m_children.size());
+		return int(m_children.size());
 	}
 
 	btCollisionShape* getChildShape(int index)
@@ -96,18 +97,17 @@ public:
 		return m_children[index].m_childShape;
 	}
 
-	btTransform&	getChildTransform(int index)
+	btTransform& getChildTransform(int index)
 	{
 		return m_children[index].m_transform;
 	}
-	const btTransform&	getChildTransform(int index) const
+	const btTransform& getChildTransform(int index) const
 	{
 		return m_children[index].m_transform;
 	}
 
 	///set a new transform for a child, and update internal data structures (local aabb and dynamic tree)
-	void	updateChildTransform(int childIndex, const btTransform& newChildTransform, bool shouldRecalculateLocalAabb = true);
-
+	void updateChildTransform(int childIndex, const btTransform& newChildTransform, bool shouldRecalculateLocalAabb = true);
 
 	btCompoundShapeChild* getChildList()
 	{
@@ -115,40 +115,40 @@ public:
 	}
 
 	///getAabb's default implementation is brute force, expected derived classes to implement a fast dedicated version
-	virtual	void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const;
+	virtual void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
 
 	/** Re-calculate the local Aabb. Is called at the end of removeChildShapes. 
 	Use this yourself if you modify the children or their transforms. */
-	virtual void recalculateLocalAabb(); 
+	virtual void recalculateLocalAabb();
 
-	virtual void	setLocalScaling(const btVector3& scaling);
+	virtual void setLocalScaling(const btVector3& scaling);
 
-	virtual const btVector3& getLocalScaling() const 
+	virtual const btVector3& getLocalScaling() const
 	{
 		return m_localScaling;
 	}
 
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
+	virtual void calculateLocalInertia(btScalar mass, btVector3 & inertia) const;
 
-	virtual void	setMargin(btScalar margin)
+	virtual void setMargin(btScalar margin)
 	{
 		m_collisionMargin = margin;
 	}
-	virtual btScalar	getMargin() const
+	virtual btScalar getMargin() const
 	{
 		return m_collisionMargin;
 	}
-	virtual const char*	getName()const
+	virtual const char* getName() const
 	{
 		return "Compound";
 	}
 
-	const btDbvt*	getDynamicAabbTree() const
+	const btDbvt* getDynamicAabbTree() const
 	{
 		return m_dynamicAabbTree;
 	}
-	
-	btDbvt*	getDynamicAabbTree()
+
+	btDbvt* getDynamicAabbTree()
 	{
 		return m_dynamicAabbTree;
 	}
@@ -162,19 +162,19 @@ public:
 	///of the collision object by the principal transform.
 	void calculatePrincipalAxisTransform(const btScalar* masses, btTransform& principal, btVector3& inertia) const;
 
-	int	getUpdateRevision() const
+	int getUpdateRevision() const
 	{
 		return m_updateRevision;
 	}
 
-	virtual	int	calculateSerializeBufferSize() const;
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
-
-
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 };
 
+// clang-format off
+
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
 struct btCompoundShapeChildData
 {
@@ -197,16 +197,11 @@ struct	btCompoundShapeData
 
 };
 
+// clang-format on
 
-SIMD_FORCE_INLINE	int	btCompoundShape::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btCompoundShape::calculateSerializeBufferSize() const
 {
 	return sizeof(btCompoundShapeData);
 }
 
-
-
-
-
-
-
-#endif //BT_COMPOUND_SHAPE_H
+#endif  //BT_COMPOUND_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btConcaveShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btConcaveShape.cpp
index 58ff84a5b02..5d396844dd8 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btConcaveShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btConcaveShape.cpp
@@ -13,15 +13,12 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btConcaveShape.h"
 
 btConcaveShape::btConcaveShape() : m_collisionMargin(btScalar(0.))
 {
-
 }
 
 btConcaveShape::~btConcaveShape()
 {
-
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btConcaveShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btConcaveShape.h
index 2917cc5b60f..716624e182b 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btConcaveShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btConcaveShape.h
@@ -17,12 +17,13 @@ subject to the following restrictions:
 #define BT_CONCAVE_SHAPE_H
 
 #include "btCollisionShape.h"
-#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h" // for the types
+#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h"  // for the types
 #include "btTriangleCallback.h"
 
 /// PHY_ScalarType enumerates possible scalar types.
 /// See the btStridingMeshInterface or btHeightfieldTerrainShape for its use
-typedef enum PHY_ScalarType {
+typedef enum PHY_ScalarType
+{
 	PHY_FLOAT,
 	PHY_DOUBLE,
 	PHY_INTEGER,
@@ -33,30 +34,29 @@ typedef enum PHY_ScalarType {
 
 ///The btConcaveShape class provides an interface for non-moving (static) concave shapes.
 ///It has been implemented by the btStaticPlaneShape, btBvhTriangleMeshShape and btHeightfieldTerrainShape.
-ATTRIBUTE_ALIGNED16(class) btConcaveShape : public btCollisionShape
+ATTRIBUTE_ALIGNED16(class)
+btConcaveShape : public btCollisionShape
 {
 protected:
 	btScalar m_collisionMargin;
 
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
+
 	btConcaveShape();
 
 	virtual ~btConcaveShape();
 
-	virtual void	processAllTriangles(btTriangleCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax) const = 0;
+	virtual void processAllTriangles(btTriangleCallback * callback, const btVector3& aabbMin, const btVector3& aabbMax) const = 0;
 
-	virtual btScalar getMargin() const {
+	virtual btScalar getMargin() const
+	{
 		return m_collisionMargin;
 	}
 	virtual void setMargin(btScalar collisionMargin)
 	{
 		m_collisionMargin = collisionMargin;
 	}
-
-
-
 };
 
-#endif //BT_CONCAVE_SHAPE_H
+#endif  //BT_CONCAVE_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btConeShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btConeShape.cpp
index 2d83c8bfbac..64a6f272ca5 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btConeShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btConeShape.cpp
@@ -15,11 +15,9 @@ subject to the following restrictions:
 
 #include "btConeShape.h"
 
-
-
-btConeShape::btConeShape (btScalar radius,btScalar height): btConvexInternalShape (),
-m_radius (radius),
-m_height(height)
+btConeShape::btConeShape(btScalar radius, btScalar height) : btConvexInternalShape(),
+															 m_radius(radius),
+															 m_height(height)
 {
 	m_shapeType = CONE_SHAPE_PROXYTYPE;
 	setConeUpIndex(1);
@@ -27,42 +25,40 @@ m_height(height)
 	m_sinAngle = (m_radius / btSqrt(m_radius * m_radius + m_height * m_height));
 }
 
-btConeShapeZ::btConeShapeZ (btScalar radius,btScalar height):
-btConeShape(radius,height)
+btConeShapeZ::btConeShapeZ(btScalar radius, btScalar height) : btConeShape(radius, height)
 {
 	setConeUpIndex(2);
 }
 
-btConeShapeX::btConeShapeX (btScalar radius,btScalar height):
-btConeShape(radius,height)
+btConeShapeX::btConeShapeX(btScalar radius, btScalar height) : btConeShape(radius, height)
 {
 	setConeUpIndex(0);
 }
 
 ///choose upAxis index
-void	btConeShape::setConeUpIndex(int upIndex)
+void btConeShape::setConeUpIndex(int upIndex)
 {
 	switch (upIndex)
 	{
-	case 0:
+		case 0:
 			m_coneIndices[0] = 1;
 			m_coneIndices[1] = 0;
 			m_coneIndices[2] = 2;
-		break;
-	case 1:
+			break;
+		case 1:
 			m_coneIndices[0] = 0;
 			m_coneIndices[1] = 1;
 			m_coneIndices[2] = 2;
-		break;
-	case 2:
+			break;
+		case 2:
 			m_coneIndices[0] = 0;
 			m_coneIndices[1] = 2;
 			m_coneIndices[2] = 1;
-		break;
-	default:
-		btAssert(0);
+			break;
+		default:
+			btAssert(0);
 	};
-	
+
 	m_implicitShapeDimensions[m_coneIndices[0]] = m_radius;
 	m_implicitShapeDimensions[m_coneIndices[1]] = m_height;
 	m_implicitShapeDimensions[m_coneIndices[2]] = m_radius;
@@ -70,72 +66,71 @@ void	btConeShape::setConeUpIndex(int upIndex)
 
 btVector3 btConeShape::coneLocalSupport(const btVector3& v) const
 {
-	
 	btScalar halfHeight = m_height * btScalar(0.5);
 
- if (v[m_coneIndices[1]] > v.length() * m_sinAngle)
- {
-	btVector3 tmp;
-
-	tmp[m_coneIndices[0]] = btScalar(0.);
-	tmp[m_coneIndices[1]] = halfHeight;
-	tmp[m_coneIndices[2]] = btScalar(0.);
-	return tmp;
- }
-  else {
-    btScalar s = btSqrt(v[m_coneIndices[0]] * v[m_coneIndices[0]] + v[m_coneIndices[2]] * v[m_coneIndices[2]]);
-    if (s > SIMD_EPSILON) {
-      btScalar d = m_radius / s;
-	  btVector3 tmp;
-	  tmp[m_coneIndices[0]] = v[m_coneIndices[0]] * d;
-	  tmp[m_coneIndices[1]] = -halfHeight;
-	  tmp[m_coneIndices[2]] = v[m_coneIndices[2]] * d;
-	  return tmp;
-    }
-    else  {
+	if (v[m_coneIndices[1]] > v.length() * m_sinAngle)
+	{
 		btVector3 tmp;
+
 		tmp[m_coneIndices[0]] = btScalar(0.);
-		tmp[m_coneIndices[1]] = -halfHeight;
+		tmp[m_coneIndices[1]] = halfHeight;
 		tmp[m_coneIndices[2]] = btScalar(0.);
 		return tmp;
 	}
-  }
-
+	else
+	{
+		btScalar s = btSqrt(v[m_coneIndices[0]] * v[m_coneIndices[0]] + v[m_coneIndices[2]] * v[m_coneIndices[2]]);
+		if (s > SIMD_EPSILON)
+		{
+			btScalar d = m_radius / s;
+			btVector3 tmp;
+			tmp[m_coneIndices[0]] = v[m_coneIndices[0]] * d;
+			tmp[m_coneIndices[1]] = -halfHeight;
+			tmp[m_coneIndices[2]] = v[m_coneIndices[2]] * d;
+			return tmp;
+		}
+		else
+		{
+			btVector3 tmp;
+			tmp[m_coneIndices[0]] = btScalar(0.);
+			tmp[m_coneIndices[1]] = -halfHeight;
+			tmp[m_coneIndices[2]] = btScalar(0.);
+			return tmp;
+		}
+	}
 }
 
-btVector3	btConeShape::localGetSupportingVertexWithoutMargin(const btVector3& vec) const
+btVector3 btConeShape::localGetSupportingVertexWithoutMargin(const btVector3& vec) const
 {
-		return coneLocalSupport(vec);
+	return coneLocalSupport(vec);
 }
 
-void	btConeShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const
+void btConeShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const
 {
-	for (int i=0;i<numVectors;i++)
+	for (int i = 0; i < numVectors; i++)
 	{
 		const btVector3& vec = vectors[i];
 		supportVerticesOut[i] = coneLocalSupport(vec);
 	}
 }
 
-
-btVector3	btConeShape::localGetSupportingVertex(const btVector3& vec)  const
+btVector3 btConeShape::localGetSupportingVertex(const btVector3& vec) const
 {
 	btVector3 supVertex = coneLocalSupport(vec);
-	if ( getMargin()!=btScalar(0.) )
+	if (getMargin() != btScalar(0.))
 	{
 		btVector3 vecnorm = vec;
-		if (vecnorm .length2() < (SIMD_EPSILON*SIMD_EPSILON))
+		if (vecnorm.length2() < (SIMD_EPSILON * SIMD_EPSILON))
 		{
-			vecnorm.setValue(btScalar(-1.),btScalar(-1.),btScalar(-1.));
-		} 
+			vecnorm.setValue(btScalar(-1.), btScalar(-1.), btScalar(-1.));
+		}
 		vecnorm.normalize();
-		supVertex+= getMargin() * vecnorm;
+		supVertex += getMargin() * vecnorm;
 	}
 	return supVertex;
 }
 
-
-void	btConeShape::setLocalScaling(const btVector3& scaling)
+void btConeShape::setLocalScaling(const btVector3& scaling)
 {
 	int axis = m_coneIndices[1];
 	int r1 = m_coneIndices[0];
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btConeShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btConeShape.h
index 3b44e3f272c..49f26bc4e58 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btConeShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btConeShape.h
@@ -17,31 +17,30 @@ subject to the following restrictions:
 #define BT_CONE_MINKOWSKI_H
 
 #include "btConvexInternalShape.h"
-#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h" // for the types
+#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h"  // for the types
 
 ///The btConeShape implements a cone shape primitive, centered around the origin and aligned with the Y axis. The btConeShapeX is aligned around the X axis and btConeShapeZ around the Z axis.
-ATTRIBUTE_ALIGNED16(class) btConeShape : public btConvexInternalShape
+ATTRIBUTE_ALIGNED16(class)
+btConeShape : public btConvexInternalShape
 
 {
-
 	btScalar m_sinAngle;
 	btScalar m_radius;
 	btScalar m_height;
-	int		m_coneIndices[3];
+	int m_coneIndices[3];
 	btVector3 coneLocalSupport(const btVector3& v) const;
 
-
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
-	btConeShape (btScalar radius,btScalar height);
-	
-	virtual btVector3	localGetSupportingVertex(const btVector3& vec) const;
-	virtual btVector3	localGetSupportingVertexWithoutMargin(const btVector3& vec) const;
-	virtual void	batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const;
 
-	btScalar getRadius() const { return m_radius;}
-	btScalar getHeight() const { return m_height;}
+	btConeShape(btScalar radius, btScalar height);
+
+	virtual btVector3 localGetSupportingVertex(const btVector3& vec) const;
+	virtual btVector3 localGetSupportingVertexWithoutMargin(const btVector3& vec) const;
+	virtual void batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const;
+
+	btScalar getRadius() const { return m_radius; }
+	btScalar getHeight() const { return m_height; }
 
 	void setRadius(const btScalar radius)
 	{
@@ -52,124 +51,115 @@ public:
 		m_height = height;
 	}
 
-
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const
+	virtual void calculateLocalInertia(btScalar mass, btVector3 & inertia) const
 	{
 		btTransform identity;
 		identity.setIdentity();
-		btVector3 aabbMin,aabbMax;
-		getAabb(identity,aabbMin,aabbMax);
+		btVector3 aabbMin, aabbMax;
+		getAabb(identity, aabbMin, aabbMax);
 
-		btVector3 halfExtents = (aabbMax-aabbMin)*btScalar(0.5);
+		btVector3 halfExtents = (aabbMax - aabbMin) * btScalar(0.5);
 
 		btScalar margin = getMargin();
 
-		btScalar lx=btScalar(2.)*(halfExtents.x()+margin);
-		btScalar ly=btScalar(2.)*(halfExtents.y()+margin);
-		btScalar lz=btScalar(2.)*(halfExtents.z()+margin);
-		const btScalar x2 = lx*lx;
-		const btScalar y2 = ly*ly;
-		const btScalar z2 = lz*lz;
+		btScalar lx = btScalar(2.) * (halfExtents.x() + margin);
+		btScalar ly = btScalar(2.) * (halfExtents.y() + margin);
+		btScalar lz = btScalar(2.) * (halfExtents.z() + margin);
+		const btScalar x2 = lx * lx;
+		const btScalar y2 = ly * ly;
+		const btScalar z2 = lz * lz;
 		const btScalar scaledmass = mass * btScalar(0.08333333);
 
-		inertia = scaledmass * (btVector3(y2+z2,x2+z2,x2+y2));
+		inertia = scaledmass * (btVector3(y2 + z2, x2 + z2, x2 + y2));
 
-//		inertia.x() = scaledmass * (y2+z2);
-//		inertia.y() = scaledmass * (x2+z2);
-//		inertia.z() = scaledmass * (x2+y2);
+		//		inertia.x() = scaledmass * (y2+z2);
+		//		inertia.y() = scaledmass * (x2+z2);
+		//		inertia.z() = scaledmass * (x2+y2);
 	}
 
-
-		virtual const char*	getName()const 
-		{
-			return "Cone";
-		}
-		
-		///choose upAxis index
-		void	setConeUpIndex(int upIndex);
-		
-		int	getConeUpIndex() const
-		{
-			return m_coneIndices[1];
-		}
-
-	virtual btVector3	getAnisotropicRollingFrictionDirection() const
+	virtual const char* getName() const
 	{
-		return btVector3 (0,1,0);
+		return "Cone";
 	}
 
-	virtual void	setLocalScaling(const btVector3& scaling);
-	
-	
-	virtual	int	calculateSerializeBufferSize() const;
-	
-	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
-	
+	///choose upAxis index
+	void setConeUpIndex(int upIndex);
 
+	int getConeUpIndex() const
+	{
+		return m_coneIndices[1];
+	}
+
+	virtual btVector3 getAnisotropicRollingFrictionDirection() const
+	{
+		return btVector3(0, 1, 0);
+	}
+
+	virtual void setLocalScaling(const btVector3& scaling);
+
+	virtual int calculateSerializeBufferSize() const;
+
+	///fills the dataBuffer and returns the struct name (and 0 on failure)
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 };
 
 ///btConeShape implements a Cone shape, around the X axis
 class btConeShapeX : public btConeShape
 {
-	public:
-		btConeShapeX(btScalar radius,btScalar height);
+public:
+	btConeShapeX(btScalar radius, btScalar height);
 
-	virtual btVector3	getAnisotropicRollingFrictionDirection() const
+	virtual btVector3 getAnisotropicRollingFrictionDirection() const
 	{
-		return btVector3 (1,0,0);
+		return btVector3(1, 0, 0);
 	}
 
 	//debugging
-	virtual const char*	getName()const
+	virtual const char* getName() const
 	{
 		return "ConeX";
 	}
-	
-	
 };
 
 ///btConeShapeZ implements a Cone shape, around the Z axis
 class btConeShapeZ : public btConeShape
 {
 public:
-	btConeShapeZ(btScalar radius,btScalar height);
+	btConeShapeZ(btScalar radius, btScalar height);
 
-	virtual btVector3	getAnisotropicRollingFrictionDirection() const
+	virtual btVector3 getAnisotropicRollingFrictionDirection() const
 	{
-		return btVector3 (0,0,1);
+		return btVector3(0, 0, 1);
 	}
 
 	//debugging
-	virtual const char*	getName()const
+	virtual const char* getName() const
 	{
 		return "ConeZ";
 	}
-	
-	
 };
 
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
-struct	btConeShapeData
+struct btConeShapeData
 {
-	btConvexInternalShapeData	m_convexInternalShapeData;
-	
-	int	m_upIndex;
-	
-	char	m_padding[4];
+	btConvexInternalShapeData m_convexInternalShapeData;
+
+	int m_upIndex;
+
+	char m_padding[4];
 };
 
-SIMD_FORCE_INLINE	int	btConeShape::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btConeShape::calculateSerializeBufferSize() const
 {
 	return sizeof(btConeShapeData);
 }
 
 ///fills the dataBuffer and returns the struct name (and 0 on failure)
-SIMD_FORCE_INLINE	const char*	btConeShape::serialize(void* dataBuffer, btSerializer* serializer) const
+SIMD_FORCE_INLINE const char* btConeShape::serialize(void* dataBuffer, btSerializer* serializer) const
 {
-	btConeShapeData* shapeData = (btConeShapeData*) dataBuffer;
+	btConeShapeData* shapeData = (btConeShapeData*)dataBuffer;
 
-	btConvexInternalShape::serialize(&shapeData->m_convexInternalShapeData,serializer);
+	btConvexInternalShape::serialize(&shapeData->m_convexInternalShapeData, serializer);
 
 	shapeData->m_upIndex = m_coneIndices[1];
 
@@ -182,5 +172,4 @@ SIMD_FORCE_INLINE	const char*	btConeShape::serialize(void* dataBuffer, btSeriali
 	return "btConeShapeData";
 }
 
-#endif //BT_CONE_MINKOWSKI_H
-
+#endif  //BT_CONE_MINKOWSKI_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvex2dShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvex2dShape.cpp
index 10ea3e981af..7d3d1d362fb 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvex2dShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvex2dShape.cpp
@@ -15,54 +15,48 @@ subject to the following restrictions:
 
 #include "btConvex2dShape.h"
 
-btConvex2dShape::btConvex2dShape(	btConvexShape* convexChildShape):
-btConvexShape (), m_childConvexShape(convexChildShape)
+btConvex2dShape::btConvex2dShape(btConvexShape* convexChildShape) : btConvexShape(), m_childConvexShape(convexChildShape)
 {
 	m_shapeType = CONVEX_2D_SHAPE_PROXYTYPE;
 }
-	
+
 btConvex2dShape::~btConvex2dShape()
 {
 }
 
-	
-
-btVector3	btConvex2dShape::localGetSupportingVertexWithoutMargin(const btVector3& vec)const
+btVector3 btConvex2dShape::localGetSupportingVertexWithoutMargin(const btVector3& vec) const
 {
 	return m_childConvexShape->localGetSupportingVertexWithoutMargin(vec);
 }
 
-void	btConvex2dShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const
+void btConvex2dShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const
 {
-	m_childConvexShape->batchedUnitVectorGetSupportingVertexWithoutMargin(vectors,supportVerticesOut,numVectors);
+	m_childConvexShape->batchedUnitVectorGetSupportingVertexWithoutMargin(vectors, supportVerticesOut, numVectors);
 }
 
-
-btVector3	btConvex2dShape::localGetSupportingVertex(const btVector3& vec)const
+btVector3 btConvex2dShape::localGetSupportingVertex(const btVector3& vec) const
 {
 	return m_childConvexShape->localGetSupportingVertex(vec);
 }
 
-
-void	btConvex2dShape::calculateLocalInertia(btScalar mass,btVector3& inertia) const
+void btConvex2dShape::calculateLocalInertia(btScalar mass, btVector3& inertia) const
 {
 	///this linear upscaling is not realistic, but we don't deal with large mass ratios...
-	m_childConvexShape->calculateLocalInertia(mass,inertia);
+	m_childConvexShape->calculateLocalInertia(mass, inertia);
 }
 
-
-	///getAabb's default implementation is brute force, expected derived classes to implement a fast dedicated version
-void btConvex2dShape::getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const
+///getAabb's default implementation is brute force, expected derived classes to implement a fast dedicated version
+void btConvex2dShape::getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
 {
-	m_childConvexShape->getAabb(t,aabbMin,aabbMax);
+	m_childConvexShape->getAabb(t, aabbMin, aabbMax);
 }
 
-void btConvex2dShape::getAabbSlow(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const
+void btConvex2dShape::getAabbSlow(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
 {
-	m_childConvexShape->getAabbSlow(t,aabbMin,aabbMax);
+	m_childConvexShape->getAabbSlow(t, aabbMin, aabbMax);
 }
 
-void	btConvex2dShape::setLocalScaling(const btVector3& scaling) 
+void btConvex2dShape::setLocalScaling(const btVector3& scaling)
 {
 	m_childConvexShape->setLocalScaling(scaling);
 }
@@ -72,21 +66,21 @@ const btVector3& btConvex2dShape::getLocalScaling() const
 	return m_childConvexShape->getLocalScaling();
 }
 
-void	btConvex2dShape::setMargin(btScalar margin)
+void btConvex2dShape::setMargin(btScalar margin)
 {
 	m_childConvexShape->setMargin(margin);
 }
-btScalar	btConvex2dShape::getMargin() const
+btScalar btConvex2dShape::getMargin() const
 {
 	return m_childConvexShape->getMargin();
 }
 
-int		btConvex2dShape::getNumPreferredPenetrationDirections() const
+int btConvex2dShape::getNumPreferredPenetrationDirections() const
 {
 	return m_childConvexShape->getNumPreferredPenetrationDirections();
 }
-	
-void	btConvex2dShape::getPreferredPenetrationDirection(int index, btVector3& penetrationVector) const
+
+void btConvex2dShape::getPreferredPenetrationDirection(int index, btVector3& penetrationVector) const
 {
-	m_childConvexShape->getPreferredPenetrationDirection(index,penetrationVector);
+	m_childConvexShape->getPreferredPenetrationDirection(index, penetrationVector);
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvex2dShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvex2dShape.h
index bbd1caf42fb..cd4f1ef7b89 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvex2dShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvex2dShape.h
@@ -17,66 +17,61 @@ subject to the following restrictions:
 #define BT_CONVEX_2D_SHAPE_H
 
 #include "BulletCollision/CollisionShapes/btConvexShape.h"
-#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h" // for the types
+#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h"  // for the types
 
 ///The btConvex2dShape allows to use arbitrary convex shapes as 2d convex shapes, with the Z component assumed to be 0.
 ///For 2d boxes, the btBox2dShape is recommended.
-ATTRIBUTE_ALIGNED16(class) btConvex2dShape : public btConvexShape
+ATTRIBUTE_ALIGNED16(class)
+btConvex2dShape : public btConvexShape
 {
-	btConvexShape*	m_childConvexShape;
+	btConvexShape* m_childConvexShape;
 
-	public:
-	
+public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
-	btConvex2dShape(	btConvexShape* convexChildShape);
-	
+
+	btConvex2dShape(btConvexShape * convexChildShape);
+
 	virtual ~btConvex2dShape();
-	
-	virtual btVector3	localGetSupportingVertexWithoutMargin(const btVector3& vec)const;
 
-	virtual btVector3	localGetSupportingVertex(const btVector3& vec)const;
+	virtual btVector3 localGetSupportingVertexWithoutMargin(const btVector3& vec) const;
 
-	virtual void	batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const;
+	virtual btVector3 localGetSupportingVertex(const btVector3& vec) const;
 
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
+	virtual void batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const;
 
-	btConvexShape*	getChildShape() 
+	virtual void calculateLocalInertia(btScalar mass, btVector3 & inertia) const;
+
+	btConvexShape* getChildShape()
 	{
 		return m_childConvexShape;
 	}
 
-	const btConvexShape*	getChildShape() const
+	const btConvexShape* getChildShape() const
 	{
 		return m_childConvexShape;
 	}
 
-	virtual const char*	getName()const 
+	virtual const char* getName() const
 	{
 		return "Convex2dShape";
 	}
-	
-
 
 	///////////////////////////
 
-
 	///getAabb's default implementation is brute force, expected derived classes to implement a fast dedicated version
-	void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const;
+	void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
 
-	virtual void getAabbSlow(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const;
+	virtual void getAabbSlow(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
 
-	virtual void	setLocalScaling(const btVector3& scaling) ;
-	virtual const btVector3& getLocalScaling() const ;
+	virtual void setLocalScaling(const btVector3& scaling);
+	virtual const btVector3& getLocalScaling() const;
 
-	virtual void	setMargin(btScalar margin);
-	virtual btScalar	getMargin() const;
-
-	virtual int		getNumPreferredPenetrationDirections() const;
-	
-	virtual void	getPreferredPenetrationDirection(int index, btVector3& penetrationVector) const;
+	virtual void setMargin(btScalar margin);
+	virtual btScalar getMargin() const;
 
+	virtual int getNumPreferredPenetrationDirections() const;
 
+	virtual void getPreferredPenetrationDirection(int index, btVector3& penetrationVector) const;
 };
 
-#endif //BT_CONVEX_2D_SHAPE_H
+#endif  //BT_CONVEX_2D_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexHullShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexHullShape.cpp
index a7a9598406a..703de45922e 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexHullShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexHullShape.cpp
@@ -13,7 +13,7 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-#if defined (_WIN32) || defined (__i386__)
+#if defined(_WIN32) || defined(__i386__)
 #define BT_USE_SSE_IN_API
 #endif
 
@@ -25,14 +25,14 @@ subject to the following restrictions:
 #include "btConvexPolyhedron.h"
 #include "LinearMath/btConvexHullComputer.h"
 
-btConvexHullShape ::btConvexHullShape (const btScalar* points,int numPoints,int stride) : btPolyhedralConvexAabbCachingShape ()
+btConvexHullShape ::btConvexHullShape(const btScalar* points, int numPoints, int stride) : btPolyhedralConvexAabbCachingShape()
 {
 	m_shapeType = CONVEX_HULL_SHAPE_PROXYTYPE;
 	m_unscaledPoints.resize(numPoints);
 
 	unsigned char* pointsAddress = (unsigned char*)points;
 
-	for (int i=0;i<numPoints;i++)
+	for (int i = 0; i < numPoints; i++)
 	{
 		btScalar* point = (btScalar*)pointsAddress;
 		m_unscaledPoints[i] = btVector3(point[0], point[1], point[2]);
@@ -40,11 +40,8 @@ btConvexHullShape ::btConvexHullShape (const btScalar* points,int numPoints,int
 	}
 
 	recalcLocalAabb();
-
 }
 
-
-
 void btConvexHullShape::setLocalScaling(const btVector3& scaling)
 {
 	m_localScaling = scaling;
@@ -56,90 +53,81 @@ void btConvexHullShape::addPoint(const btVector3& point, bool recalculateLocalAa
 	m_unscaledPoints.push_back(point);
 	if (recalculateLocalAabb)
 		recalcLocalAabb();
-
 }
 
-btVector3	btConvexHullShape::localGetSupportingVertexWithoutMargin(const btVector3& vec)const
+btVector3 btConvexHullShape::localGetSupportingVertexWithoutMargin(const btVector3& vec) const
 {
-	btVector3 supVec(btScalar(0.),btScalar(0.),btScalar(0.));
+	btVector3 supVec(btScalar(0.), btScalar(0.), btScalar(0.));
 	btScalar maxDot = btScalar(-BT_LARGE_FLOAT);
 
-    // Here we take advantage of dot(a, b*c) = dot(a*b, c).  Note: This is true mathematically, but not numerically. 
-    if( 0 < m_unscaledPoints.size() )
-    {
-        btVector3 scaled = vec * m_localScaling;
-        int index = (int) scaled.maxDot( &m_unscaledPoints[0], m_unscaledPoints.size(), maxDot); // FIXME: may violate encapsulation of m_unscaledPoints
-        return m_unscaledPoints[index] * m_localScaling;
-    }
+	// Here we take advantage of dot(a, b*c) = dot(a*b, c).  Note: This is true mathematically, but not numerically.
+	if (0 < m_unscaledPoints.size())
+	{
+		btVector3 scaled = vec * m_localScaling;
+		int index = (int)scaled.maxDot(&m_unscaledPoints[0], m_unscaledPoints.size(), maxDot);  // FIXME: may violate encapsulation of m_unscaledPoints
+		return m_unscaledPoints[index] * m_localScaling;
+	}
 
-    return supVec;
+	return supVec;
 }
 
-void	btConvexHullShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const
+void btConvexHullShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const
 {
 	btScalar newDot;
 	//use 'w' component of supportVerticesOut?
 	{
-		for (int i=0;i<numVectors;i++)
+		for (int i = 0; i < numVectors; i++)
 		{
 			supportVerticesOut[i][3] = btScalar(-BT_LARGE_FLOAT);
 		}
 	}
 
-    for (int j=0;j<numVectors;j++)
-    {
-        btVector3 vec = vectors[j] * m_localScaling;        // dot(a*b,c) = dot(a,b*c)
-        if( 0 <  m_unscaledPoints.size() )
-        {
-            int i = (int) vec.maxDot( &m_unscaledPoints[0], m_unscaledPoints.size(), newDot);
-            supportVerticesOut[j] = getScaledPoint(i);
-            supportVerticesOut[j][3] = newDot;        
-        }
-        else
-            supportVerticesOut[j][3] = -BT_LARGE_FLOAT;
-    }
-
-
-
+	for (int j = 0; j < numVectors; j++)
+	{
+		btVector3 vec = vectors[j] * m_localScaling;  // dot(a*b,c) = dot(a,b*c)
+		if (0 < m_unscaledPoints.size())
+		{
+			int i = (int)vec.maxDot(&m_unscaledPoints[0], m_unscaledPoints.size(), newDot);
+			supportVerticesOut[j] = getScaledPoint(i);
+			supportVerticesOut[j][3] = newDot;
+		}
+		else
+			supportVerticesOut[j][3] = -BT_LARGE_FLOAT;
+	}
 }
-	
 
-
-btVector3	btConvexHullShape::localGetSupportingVertex(const btVector3& vec)const
+btVector3 btConvexHullShape::localGetSupportingVertex(const btVector3& vec) const
 {
 	btVector3 supVertex = localGetSupportingVertexWithoutMargin(vec);
 
-	if ( getMargin()!=btScalar(0.) )
+	if (getMargin() != btScalar(0.))
 	{
 		btVector3 vecnorm = vec;
-		if (vecnorm .length2() < (SIMD_EPSILON*SIMD_EPSILON))
+		if (vecnorm.length2() < (SIMD_EPSILON * SIMD_EPSILON))
 		{
-			vecnorm.setValue(btScalar(-1.),btScalar(-1.),btScalar(-1.));
-		} 
+			vecnorm.setValue(btScalar(-1.), btScalar(-1.), btScalar(-1.));
+		}
 		vecnorm.normalize();
-		supVertex+= getMargin() * vecnorm;
+		supVertex += getMargin() * vecnorm;
 	}
 	return supVertex;
 }
 
-
 void btConvexHullShape::optimizeConvexHull()
 {
 	btConvexHullComputer conv;
-	conv.compute(&m_unscaledPoints[0].getX(), sizeof(btVector3),m_unscaledPoints.size(),0.f,0.f);
+	conv.compute(&m_unscaledPoints[0].getX(), sizeof(btVector3), m_unscaledPoints.size(), 0.f, 0.f);
 	int numVerts = conv.vertices.size();
 	m_unscaledPoints.resize(0);
-	for (int i=0;i<numVerts;i++)
-    {
-        m_unscaledPoints.push_back(conv.vertices[i]);
-    }
+	for (int i = 0; i < numVerts; i++)
+	{
+		m_unscaledPoints.push_back(conv.vertices[i]);
+	}
 }
 
-
-
 //currently just for debugging (drawing), perhaps future support for algebraic continuous collision detection
 //Please note that you can debug-draw btConvexHullShape with the Raytracer Demo
-int	btConvexHullShape::getNumVertices() const
+int btConvexHullShape::getNumVertices() const
 {
 	return m_unscaledPoints.size();
 }
@@ -149,67 +137,65 @@ int btConvexHullShape::getNumEdges() const
 	return m_unscaledPoints.size();
 }
 
-void btConvexHullShape::getEdge(int i,btVector3& pa,btVector3& pb) const
+void btConvexHullShape::getEdge(int i, btVector3& pa, btVector3& pb) const
 {
-
-	int index0 = i%m_unscaledPoints.size();
-	int index1 = (i+1)%m_unscaledPoints.size();
+	int index0 = i % m_unscaledPoints.size();
+	int index1 = (i + 1) % m_unscaledPoints.size();
 	pa = getScaledPoint(index0);
 	pb = getScaledPoint(index1);
 }
 
-void btConvexHullShape::getVertex(int i,btVector3& vtx) const
+void btConvexHullShape::getVertex(int i, btVector3& vtx) const
 {
 	vtx = getScaledPoint(i);
 }
 
-int	btConvexHullShape::getNumPlanes() const
+int btConvexHullShape::getNumPlanes() const
 {
 	return 0;
 }
 
-void btConvexHullShape::getPlane(btVector3& ,btVector3& ,int ) const
+void btConvexHullShape::getPlane(btVector3&, btVector3&, int) const
 {
-
 	btAssert(0);
 }
 
 //not yet
-bool btConvexHullShape::isInside(const btVector3& ,btScalar ) const
+bool btConvexHullShape::isInside(const btVector3&, btScalar) const
 {
 	btAssert(0);
 	return false;
 }
 
 ///fills the dataBuffer and returns the struct name (and 0 on failure)
-const char*	btConvexHullShape::serialize(void* dataBuffer, btSerializer* serializer) const
+const char* btConvexHullShape::serialize(void* dataBuffer, btSerializer* serializer) const
 {
 	//int szc = sizeof(btConvexHullShapeData);
-	btConvexHullShapeData* shapeData = (btConvexHullShapeData*) dataBuffer;
+	btConvexHullShapeData* shapeData = (btConvexHullShapeData*)dataBuffer;
 	btConvexInternalShape::serialize(&shapeData->m_convexInternalShapeData, serializer);
 
 	int numElem = m_unscaledPoints.size();
 	shapeData->m_numUnscaledPoints = numElem;
 #ifdef BT_USE_DOUBLE_PRECISION
 	shapeData->m_unscaledPointsFloatPtr = 0;
-	shapeData->m_unscaledPointsDoublePtr = numElem ? (btVector3Data*)serializer->getUniquePointer((void*)&m_unscaledPoints[0]):  0;
+	shapeData->m_unscaledPointsDoublePtr = numElem ? (btVector3Data*)serializer->getUniquePointer((void*)&m_unscaledPoints[0]) : 0;
 #else
-	shapeData->m_unscaledPointsFloatPtr = numElem ? (btVector3Data*)serializer->getUniquePointer((void*)&m_unscaledPoints[0]):  0;
+	shapeData->m_unscaledPointsFloatPtr = numElem ? (btVector3Data*)serializer->getUniquePointer((void*)&m_unscaledPoints[0]) : 0;
 	shapeData->m_unscaledPointsDoublePtr = 0;
 #endif
-	
+
 	if (numElem)
 	{
 		int sz = sizeof(btVector3Data);
-	//	int sz2 = sizeof(btVector3DoubleData);
-	//	int sz3 = sizeof(btVector3FloatData);
-		btChunk* chunk = serializer->allocate(sz,numElem);
+		//	int sz2 = sizeof(btVector3DoubleData);
+		//	int sz3 = sizeof(btVector3FloatData);
+		btChunk* chunk = serializer->allocate(sz, numElem);
 		btVector3Data* memPtr = (btVector3Data*)chunk->m_oldPtr;
-		for (int i=0;i<numElem;i++,memPtr++)
+		for (int i = 0; i < numElem; i++, memPtr++)
 		{
 			m_unscaledPoints[i].serialize(*memPtr);
 		}
-		serializer->finalizeChunk(chunk,btVector3DataName,BT_ARRAY_CODE,(void*)&m_unscaledPoints[0]);
+		serializer->finalizeChunk(chunk, btVector3DataName, BT_ARRAY_CODE, (void*)&m_unscaledPoints[0]);
 	}
 
 	// Fill padding with zeros to appease msan.
@@ -218,45 +204,41 @@ const char*	btConvexHullShape::serialize(void* dataBuffer, btSerializer* seriali
 	return "btConvexHullShapeData";
 }
 
-void btConvexHullShape::project(const btTransform& trans, const btVector3& dir, btScalar& minProj, btScalar& maxProj, btVector3& witnesPtMin,btVector3& witnesPtMax) const
+void btConvexHullShape::project(const btTransform& trans, const btVector3& dir, btScalar& minProj, btScalar& maxProj, btVector3& witnesPtMin, btVector3& witnesPtMax) const
 {
 #if 1
 	minProj = FLT_MAX;
 	maxProj = -FLT_MAX;
 
 	int numVerts = m_unscaledPoints.size();
-	for(int i=0;i<numVerts;i++)
+	for (int i = 0; i < numVerts; i++)
 	{
 		btVector3 vtx = m_unscaledPoints[i] * m_localScaling;
 		btVector3 pt = trans * vtx;
 		btScalar dp = pt.dot(dir);
-		if(dp < minProj)	
+		if (dp < minProj)
 		{
 			minProj = dp;
 			witnesPtMin = pt;
 		}
-		if(dp > maxProj)	
+		if (dp > maxProj)
 		{
 			maxProj = dp;
-			witnesPtMax=pt;
+			witnesPtMax = pt;
 		}
 	}
 #else
-	btVector3 localAxis = dir*trans.getBasis();
-	witnesPtMin  = trans(localGetSupportingVertex(localAxis));
+	btVector3 localAxis = dir * trans.getBasis();
+	witnesPtMin = trans(localGetSupportingVertex(localAxis));
 	witnesPtMax = trans(localGetSupportingVertex(-localAxis));
 
 	minProj = witnesPtMin.dot(dir);
 	maxProj = witnesPtMax.dot(dir);
 #endif
 
-	if(minProj>maxProj)
+	if (minProj > maxProj)
 	{
-		btSwap(minProj,maxProj);
-		btSwap(witnesPtMin,witnesPtMax);
+		btSwap(minProj, maxProj);
+		btSwap(witnesPtMin, witnesPtMax);
 	}
-
-
 }
-
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexHullShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexHullShape.h
index 0c12aeef156..96136d7dd72 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexHullShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexHullShape.h
@@ -17,28 +17,26 @@ subject to the following restrictions:
 #define BT_CONVEX_HULL_SHAPE_H
 
 #include "btPolyhedralConvexShape.h"
-#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h" // for the types
+#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h"  // for the types
 #include "LinearMath/btAlignedObjectArray.h"
 
-
 ///The btConvexHullShape implements an implicit convex hull of an array of vertices.
 ///Bullet provides a general and fast collision detector for convex shapes based on GJK and EPA using localGetSupportingVertex.
-ATTRIBUTE_ALIGNED16(class) btConvexHullShape : public btPolyhedralConvexAabbCachingShape
+ATTRIBUTE_ALIGNED16(class)
+btConvexHullShape : public btPolyhedralConvexAabbCachingShape
 {
-	btAlignedObjectArray<btVector3>	m_unscaledPoints;
+	btAlignedObjectArray<btVector3> m_unscaledPoints;
 
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-	
 	///this constructor optionally takes in a pointer to points. Each point is assumed to be 3 consecutive btScalar (x,y,z), the striding defines the number of bytes between each point, in memory.
 	///It is easier to not pass any points in the constructor, and just add one point at a time, using addPoint.
 	///btConvexHullShape make an internal copy of the points.
-	btConvexHullShape(const btScalar* points=0,int numPoints=0, int stride=sizeof(btVector3));
+	btConvexHullShape(const btScalar* points = 0, int numPoints = 0, int stride = sizeof(btVector3));
 
 	void addPoint(const btVector3& point, bool recalculateLocalAabb = true);
 
-	
 	btVector3* getUnscaledPoints()
 	{
 		return &m_unscaledPoints[0];
@@ -55,48 +53,46 @@ public:
 		return getUnscaledPoints();
 	}
 
-    void optimizeConvexHull();
-    
-	SIMD_FORCE_INLINE	btVector3 getScaledPoint(int i) const
+	void optimizeConvexHull();
+
+	SIMD_FORCE_INLINE btVector3 getScaledPoint(int i) const
 	{
 		return m_unscaledPoints[i] * m_localScaling;
 	}
 
-	SIMD_FORCE_INLINE	int getNumPoints() const 
+	SIMD_FORCE_INLINE int getNumPoints() const
 	{
 		return m_unscaledPoints.size();
 	}
 
-	virtual btVector3	localGetSupportingVertex(const btVector3& vec)const;
-	virtual btVector3	localGetSupportingVertexWithoutMargin(const btVector3& vec)const;
-	virtual void	batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const;
-	
-
-	virtual void project(const btTransform& trans, const btVector3& dir, btScalar& minProj, btScalar& maxProj, btVector3& witnesPtMin,btVector3& witnesPtMax) const;
+	virtual btVector3 localGetSupportingVertex(const btVector3& vec) const;
+	virtual btVector3 localGetSupportingVertexWithoutMargin(const btVector3& vec) const;
+	virtual void batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const;
 
+	virtual void project(const btTransform& trans, const btVector3& dir, btScalar& minProj, btScalar& maxProj, btVector3& witnesPtMin, btVector3& witnesPtMax) const;
 
 	//debugging
-	virtual const char*	getName()const {return "Convex";}
+	virtual const char* getName() const { return "Convex"; }
 
-	
-	virtual int	getNumVertices() const;
+	virtual int getNumVertices() const;
 	virtual int getNumEdges() const;
-	virtual void getEdge(int i,btVector3& pa,btVector3& pb) const;
-	virtual void getVertex(int i,btVector3& vtx) const;
-	virtual int	getNumPlanes() const;
-	virtual void getPlane(btVector3& planeNormal,btVector3& planeSupport,int i ) const;
-	virtual	bool isInside(const btVector3& pt,btScalar tolerance) const;
+	virtual void getEdge(int i, btVector3& pa, btVector3& pb) const;
+	virtual void getVertex(int i, btVector3& vtx) const;
+	virtual int getNumPlanes() const;
+	virtual void getPlane(btVector3 & planeNormal, btVector3 & planeSupport, int i) const;
+	virtual bool isInside(const btVector3& pt, btScalar tolerance) const;
 
 	///in case we receive negative scaling
-	virtual void	setLocalScaling(const btVector3& scaling);
+	virtual void setLocalScaling(const btVector3& scaling);
 
-	virtual	int	calculateSerializeBufferSize() const;
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
-
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 };
 
+// clang-format off
+
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
 struct	btConvexHullShapeData
 {
@@ -110,12 +106,11 @@ struct	btConvexHullShapeData
 
 };
 
+// clang-format on
 
-SIMD_FORCE_INLINE	int	btConvexHullShape::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btConvexHullShape::calculateSerializeBufferSize() const
 {
 	return sizeof(btConvexHullShapeData);
 }
 
-
-#endif //BT_CONVEX_HULL_SHAPE_H
-
+#endif  //BT_CONVEX_HULL_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexInternalShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexInternalShape.cpp
index 083d60b1b1e..4d598b1aa20 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexInternalShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexInternalShape.cpp
@@ -13,139 +13,125 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btConvexInternalShape.h"
 
-
-
 btConvexInternalShape::btConvexInternalShape()
-: m_localScaling(btScalar(1.),btScalar(1.),btScalar(1.)),
-m_collisionMargin(CONVEX_DISTANCE_MARGIN)
+	: m_localScaling(btScalar(1.), btScalar(1.), btScalar(1.)),
+	  m_collisionMargin(CONVEX_DISTANCE_MARGIN)
 {
 }
 
-
-void	btConvexInternalShape::setLocalScaling(const btVector3& scaling)
+void btConvexInternalShape::setLocalScaling(const btVector3& scaling)
 {
 	m_localScaling = scaling.absolute();
 }
 
-
-
-void	btConvexInternalShape::getAabbSlow(const btTransform& trans,btVector3&minAabb,btVector3&maxAabb) const
+void btConvexInternalShape::getAabbSlow(const btTransform& trans, btVector3& minAabb, btVector3& maxAabb) const
 {
 #ifndef __SPU__
 	//use localGetSupportingVertexWithoutMargin?
 	btScalar margin = getMargin();
-	for (int i=0;i<3;i++)
+	for (int i = 0; i < 3; i++)
 	{
-		btVector3 vec(btScalar(0.),btScalar(0.),btScalar(0.));
+		btVector3 vec(btScalar(0.), btScalar(0.), btScalar(0.));
 		vec[i] = btScalar(1.);
 
-		btVector3 sv = localGetSupportingVertex(vec*trans.getBasis());
+		btVector3 sv = localGetSupportingVertex(vec * trans.getBasis());
 
 		btVector3 tmp = trans(sv);
-		maxAabb[i] = tmp[i]+margin;
+		maxAabb[i] = tmp[i] + margin;
 		vec[i] = btScalar(-1.);
-		tmp = trans(localGetSupportingVertex(vec*trans.getBasis()));
-		minAabb[i] = tmp[i]-margin;
+		tmp = trans(localGetSupportingVertex(vec * trans.getBasis()));
+		minAabb[i] = tmp[i] - margin;
 	}
 #endif
 }
 
-
-
-btVector3	btConvexInternalShape::localGetSupportingVertex(const btVector3& vec)const
+btVector3 btConvexInternalShape::localGetSupportingVertex(const btVector3& vec) const
 {
 #ifndef __SPU__
 
-	 btVector3	supVertex = localGetSupportingVertexWithoutMargin(vec);
+	btVector3 supVertex = localGetSupportingVertexWithoutMargin(vec);
 
-	if ( getMargin()!=btScalar(0.) )
+	if (getMargin() != btScalar(0.))
 	{
 		btVector3 vecnorm = vec;
-		if (vecnorm .length2() < (SIMD_EPSILON*SIMD_EPSILON))
+		if (vecnorm.length2() < (SIMD_EPSILON * SIMD_EPSILON))
 		{
-			vecnorm.setValue(btScalar(-1.),btScalar(-1.),btScalar(-1.));
-		} 
+			vecnorm.setValue(btScalar(-1.), btScalar(-1.), btScalar(-1.));
+		}
 		vecnorm.normalize();
-		supVertex+= getMargin() * vecnorm;
+		supVertex += getMargin() * vecnorm;
 	}
 	return supVertex;
 
 #else
 	btAssert(0);
-	return btVector3(0,0,0);
-#endif //__SPU__
-
- }
-
+	return btVector3(0, 0, 0);
+#endif  //__SPU__
+}
 
 btConvexInternalAabbCachingShape::btConvexInternalAabbCachingShape()
-	:	btConvexInternalShape(),
-m_localAabbMin(1,1,1),
-m_localAabbMax(-1,-1,-1),
-m_isLocalAabbValid(false)
+	: btConvexInternalShape(),
+	  m_localAabbMin(1, 1, 1),
+	  m_localAabbMax(-1, -1, -1),
+	  m_isLocalAabbValid(false)
 {
 }
 
-
-void btConvexInternalAabbCachingShape::getAabb(const btTransform& trans,btVector3& aabbMin,btVector3& aabbMax) const
+void btConvexInternalAabbCachingShape::getAabb(const btTransform& trans, btVector3& aabbMin, btVector3& aabbMax) const
 {
-	getNonvirtualAabb(trans,aabbMin,aabbMax,getMargin());
+	getNonvirtualAabb(trans, aabbMin, aabbMax, getMargin());
 }
 
-void	btConvexInternalAabbCachingShape::setLocalScaling(const btVector3& scaling)
+void btConvexInternalAabbCachingShape::setLocalScaling(const btVector3& scaling)
 {
 	btConvexInternalShape::setLocalScaling(scaling);
 	recalcLocalAabb();
 }
 
-
-void	btConvexInternalAabbCachingShape::recalcLocalAabb()
+void btConvexInternalAabbCachingShape::recalcLocalAabb()
 {
 	m_isLocalAabbValid = true;
-	
-	#if 1
+
+#if 1
 	static const btVector3 _directions[] =
-	{
-		btVector3( 1.,  0.,  0.),
-		btVector3( 0.,  1.,  0.),
-		btVector3( 0.,  0.,  1.),
-		btVector3( -1., 0.,  0.),
-		btVector3( 0., -1.,  0.),
-		btVector3( 0.,  0., -1.)
-	};
-	
+		{
+			btVector3(1., 0., 0.),
+			btVector3(0., 1., 0.),
+			btVector3(0., 0., 1.),
+			btVector3(-1., 0., 0.),
+			btVector3(0., -1., 0.),
+			btVector3(0., 0., -1.)};
+
 	btVector3 _supporting[] =
-	{
-		btVector3( 0., 0., 0.),
-		btVector3( 0., 0., 0.),
-		btVector3( 0., 0., 0.),
-		btVector3( 0., 0., 0.),
-		btVector3( 0., 0., 0.),
-		btVector3( 0., 0., 0.)
-	};
-	
+		{
+			btVector3(0., 0., 0.),
+			btVector3(0., 0., 0.),
+			btVector3(0., 0., 0.),
+			btVector3(0., 0., 0.),
+			btVector3(0., 0., 0.),
+			btVector3(0., 0., 0.)};
+
 	batchedUnitVectorGetSupportingVertexWithoutMargin(_directions, _supporting, 6);
-	
-	for ( int i = 0; i < 3; ++i )
+
+	for (int i = 0; i < 3; ++i)
 	{
 		m_localAabbMax[i] = _supporting[i][i] + m_collisionMargin;
 		m_localAabbMin[i] = _supporting[i + 3][i] - m_collisionMargin;
 	}
-	
-	#else
 
-	for (int i=0;i<3;i++)
+#else
+
+	for (int i = 0; i < 3; i++)
 	{
-		btVector3 vec(btScalar(0.),btScalar(0.),btScalar(0.));
+		btVector3 vec(btScalar(0.), btScalar(0.), btScalar(0.));
 		vec[i] = btScalar(1.);
 		btVector3 tmp = localGetSupportingVertex(vec);
-		m_localAabbMax[i] = tmp[i]+m_collisionMargin;
+		m_localAabbMax[i] = tmp[i] + m_collisionMargin;
 		vec[i] = btScalar(-1.);
 		tmp = localGetSupportingVertex(vec);
-		m_localAabbMin[i] = tmp[i]-m_collisionMargin;
+		m_localAabbMin[i] = tmp[i] - m_collisionMargin;
 	}
-	#endif
+#endif
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexInternalShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexInternalShape.h
index 1213b82fbe3..a28c57de4b8 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexInternalShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexInternalShape.h
@@ -19,39 +19,35 @@ subject to the following restrictions:
 #include "btConvexShape.h"
 #include "LinearMath/btAabbUtil2.h"
 
-
 ///The btConvexInternalShape is an internal base class, shared by most convex shape implementations.
 ///The btConvexInternalShape uses a default collision margin set to CONVEX_DISTANCE_MARGIN.
 ///This collision margin used by Gjk and some other algorithms, see also btCollisionMargin.h
-///Note that when creating small shapes (derived from btConvexInternalShape), 
+///Note that when creating small shapes (derived from btConvexInternalShape),
 ///you need to make sure to set a smaller collision margin, using the 'setMargin' API
 ///There is a automatic mechanism 'setSafeMargin' used by btBoxShape and btCylinderShape
-ATTRIBUTE_ALIGNED16(class) btConvexInternalShape : public btConvexShape
+ATTRIBUTE_ALIGNED16(class)
+btConvexInternalShape : public btConvexShape
 {
-
-	protected:
-
+protected:
 	//local scaling. collisionMargin is not scaled !
-	btVector3	m_localScaling;
+	btVector3 m_localScaling;
 
-	btVector3	m_implicitShapeDimensions;
-	
-	btScalar	m_collisionMargin;
+	btVector3 m_implicitShapeDimensions;
 
-	btScalar	m_padding;
+	btScalar m_collisionMargin;
+
+	btScalar m_padding;
 
 	btConvexInternalShape();
 
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
 	virtual ~btConvexInternalShape()
 	{
-
 	}
 
-	virtual btVector3	localGetSupportingVertex(const btVector3& vec)const;
+	virtual btVector3 localGetSupportingVertex(const btVector3& vec) const;
 
 	const btVector3& getImplicitShapeDimensions() const
 	{
@@ -62,110 +58,102 @@ public:
 	///changing a collision shape while the body is in the world is not recommended,
 	///it is best to remove the body from the world, then make the change, and re-add it
 	///alternatively flush the contact points, see documentation for 'cleanProxyFromPairs'
-	void	setImplicitShapeDimensions(const btVector3& dimensions)
+	void setImplicitShapeDimensions(const btVector3& dimensions)
 	{
 		m_implicitShapeDimensions = dimensions;
 	}
 
-	void	setSafeMargin(btScalar minDimension, btScalar defaultMarginMultiplier = 0.1f)
+	void setSafeMargin(btScalar minDimension, btScalar defaultMarginMultiplier = 0.1f)
 	{
-		btScalar safeMargin = defaultMarginMultiplier*minDimension;
+		btScalar safeMargin = defaultMarginMultiplier * minDimension;
 		if (safeMargin < getMargin())
 		{
 			setMargin(safeMargin);
 		}
 	}
-	void	setSafeMargin(const btVector3& halfExtents, btScalar defaultMarginMultiplier = 0.1f)
+	void setSafeMargin(const btVector3& halfExtents, btScalar defaultMarginMultiplier = 0.1f)
 	{
 		//see http://code.google.com/p/bullet/issues/detail?id=349
 		//this margin check could could be added to other collision shapes too,
 		//or add some assert/warning somewhere
-		btScalar minDimension=halfExtents[halfExtents.minAxis()]; 		
+		btScalar minDimension = halfExtents[halfExtents.minAxis()];
 		setSafeMargin(minDimension, defaultMarginMultiplier);
 	}
 
 	///getAabb's default implementation is brute force, expected derived classes to implement a fast dedicated version
-	void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const
+	void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
 	{
-		getAabbSlow(t,aabbMin,aabbMax);
+		getAabbSlow(t, aabbMin, aabbMax);
 	}
 
+	virtual void getAabbSlow(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
 
-	
-	virtual void getAabbSlow(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const;
-
-
-	virtual void	setLocalScaling(const btVector3& scaling);
-	virtual const btVector3& getLocalScaling() const 
+	virtual void setLocalScaling(const btVector3& scaling);
+	virtual const btVector3& getLocalScaling() const
 	{
 		return m_localScaling;
 	}
 
-	const btVector3& getLocalScalingNV() const 
+	const btVector3& getLocalScalingNV() const
 	{
 		return m_localScaling;
 	}
 
-	virtual void	setMargin(btScalar margin)
+	virtual void setMargin(btScalar margin)
 	{
 		m_collisionMargin = margin;
 	}
-	virtual btScalar	getMargin() const
+	virtual btScalar getMargin() const
 	{
 		return m_collisionMargin;
 	}
 
-	btScalar	getMarginNV() const
+	btScalar getMarginNV() const
 	{
 		return m_collisionMargin;
 	}
 
-	virtual int		getNumPreferredPenetrationDirections() const
+	virtual int getNumPreferredPenetrationDirections() const
 	{
 		return 0;
 	}
-	
-	virtual void	getPreferredPenetrationDirection(int index, btVector3& penetrationVector) const
+
+	virtual void getPreferredPenetrationDirection(int index, btVector3& penetrationVector) const
 	{
 		(void)penetrationVector;
 		(void)index;
 		btAssert(0);
 	}
 
-	virtual	int	calculateSerializeBufferSize() const;
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
-
-	
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 };
 
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
-struct	btConvexInternalShapeData
+struct btConvexInternalShapeData
 {
-	btCollisionShapeData	m_collisionShapeData;
+	btCollisionShapeData m_collisionShapeData;
 
-	btVector3FloatData	m_localScaling;
+	btVector3FloatData m_localScaling;
 
-	btVector3FloatData	m_implicitShapeDimensions;
-	
-	float			m_collisionMargin;
+	btVector3FloatData m_implicitShapeDimensions;
 
-	int	m_padding;
+	float m_collisionMargin;
 
+	int m_padding;
 };
 
-
-
-SIMD_FORCE_INLINE	int	btConvexInternalShape::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btConvexInternalShape::calculateSerializeBufferSize() const
 {
 	return sizeof(btConvexInternalShapeData);
 }
 
 ///fills the dataBuffer and returns the struct name (and 0 on failure)
-SIMD_FORCE_INLINE	const char*	btConvexInternalShape::serialize(void* dataBuffer, btSerializer* serializer) const
+SIMD_FORCE_INLINE const char* btConvexInternalShape::serialize(void* dataBuffer, btSerializer* serializer) const
 {
-	btConvexInternalShapeData* shapeData = (btConvexInternalShapeData*) dataBuffer;
+	btConvexInternalShapeData* shapeData = (btConvexInternalShapeData*)dataBuffer;
 	btCollisionShape::serialize(&shapeData->m_collisionShapeData, serializer);
 
 	m_implicitShapeDimensions.serializeFloat(shapeData->m_implicitShapeDimensions);
@@ -178,50 +166,43 @@ SIMD_FORCE_INLINE	const char*	btConvexInternalShape::serialize(void* dataBuffer,
 	return "btConvexInternalShapeData";
 }
 
-
-
-
 ///btConvexInternalAabbCachingShape adds local aabb caching for convex shapes, to avoid expensive bounding box calculations
 class btConvexInternalAabbCachingShape : public btConvexInternalShape
 {
-	btVector3	m_localAabbMin;
-	btVector3	m_localAabbMax;
-	bool		m_isLocalAabbValid;
-	
+	btVector3 m_localAabbMin;
+	btVector3 m_localAabbMax;
+	bool m_isLocalAabbValid;
+
 protected:
-					
 	btConvexInternalAabbCachingShape();
-	
-	void setCachedLocalAabb (const btVector3& aabbMin, const btVector3& aabbMax)
+
+	void setCachedLocalAabb(const btVector3& aabbMin, const btVector3& aabbMax)
 	{
 		m_isLocalAabbValid = true;
 		m_localAabbMin = aabbMin;
 		m_localAabbMax = aabbMax;
 	}
 
-	inline void getCachedLocalAabb (btVector3& aabbMin, btVector3& aabbMax) const
+	inline void getCachedLocalAabb(btVector3& aabbMin, btVector3& aabbMax) const
 	{
 		btAssert(m_isLocalAabbValid);
 		aabbMin = m_localAabbMin;
 		aabbMax = m_localAabbMax;
 	}
 
-	inline void getNonvirtualAabb(const btTransform& trans,btVector3& aabbMin,btVector3& aabbMax, btScalar margin) const
+	inline void getNonvirtualAabb(const btTransform& trans, btVector3& aabbMin, btVector3& aabbMax, btScalar margin) const
 	{
-
 		//lazy evaluation of local aabb
 		btAssert(m_isLocalAabbValid);
-		btTransformAabb(m_localAabbMin,m_localAabbMax,margin,trans,aabbMin,aabbMax);
+		btTransformAabb(m_localAabbMin, m_localAabbMax, margin, trans, aabbMin, aabbMax);
 	}
-		
+
 public:
-		
-	virtual void	setLocalScaling(const btVector3& scaling);
+	virtual void setLocalScaling(const btVector3& scaling);
 
-	virtual void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const;
-
-	void	recalcLocalAabb();
+	virtual void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
 
+	void recalcLocalAabb();
 };
 
-#endif //BT_CONVEX_INTERNAL_SHAPE_H
+#endif  //BT_CONVEX_INTERNAL_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexPointCloudShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexPointCloudShape.cpp
index ad1d1bf78f8..f00a440fa35 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexPointCloudShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexPointCloudShape.cpp
@@ -25,81 +25,73 @@ void btConvexPointCloudShape::setLocalScaling(const btVector3& scaling)
 }
 
 #ifndef __SPU__
-btVector3	btConvexPointCloudShape::localGetSupportingVertexWithoutMargin(const btVector3& vec0)const
+btVector3 btConvexPointCloudShape::localGetSupportingVertexWithoutMargin(const btVector3& vec0) const
 {
-	btVector3 supVec(btScalar(0.),btScalar(0.),btScalar(0.));
+	btVector3 supVec(btScalar(0.), btScalar(0.), btScalar(0.));
 	btScalar maxDot = btScalar(-BT_LARGE_FLOAT);
 
 	btVector3 vec = vec0;
 	btScalar lenSqr = vec.length2();
 	if (lenSqr < btScalar(0.0001))
 	{
-		vec.setValue(1,0,0);
-	} else
+		vec.setValue(1, 0, 0);
+	}
+	else
 	{
-		btScalar rlen = btScalar(1.) / btSqrt(lenSqr );
+		btScalar rlen = btScalar(1.) / btSqrt(lenSqr);
 		vec *= rlen;
 	}
-    
-    if( m_numPoints > 0 )
-    {
-        // Here we take advantage of dot(a*b, c) = dot( a, b*c) to do less work. Note this transformation is true mathematically, not numerically.
-    //    btVector3 scaled = vec * m_localScaling;
-        int index = (int) vec.maxDot( &m_unscaledPoints[0], m_numPoints, maxDot);   //FIXME: may violate encapsulation of m_unscaledPoints
-        return getScaledPoint(index);
-    }
+
+	if (m_numPoints > 0)
+	{
+		// Here we take advantage of dot(a*b, c) = dot( a, b*c) to do less work. Note this transformation is true mathematically, not numerically.
+		//    btVector3 scaled = vec * m_localScaling;
+		int index = (int)vec.maxDot(&m_unscaledPoints[0], m_numPoints, maxDot);  //FIXME: may violate encapsulation of m_unscaledPoints
+		return getScaledPoint(index);
+	}
 
 	return supVec;
 }
 
-void	btConvexPointCloudShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const
+void btConvexPointCloudShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const
 {
-    for( int j = 0; j < numVectors; j++ )
-    {
-        const btVector3& vec = vectors[j] * m_localScaling;  // dot( a*c, b) = dot(a, b*c)
-        btScalar maxDot;
-        int index = (int) vec.maxDot( &m_unscaledPoints[0], m_numPoints, maxDot);
-        supportVerticesOut[j][3] = btScalar(-BT_LARGE_FLOAT);
-        if( 0 <= index )
-        {
-            //WARNING: don't swap next lines, the w component would get overwritten!
-            supportVerticesOut[j] = getScaledPoint(index);
-            supportVerticesOut[j][3] = maxDot;
-        }
-    }
-
+	for (int j = 0; j < numVectors; j++)
+	{
+		const btVector3& vec = vectors[j] * m_localScaling;  // dot( a*c, b) = dot(a, b*c)
+		btScalar maxDot;
+		int index = (int)vec.maxDot(&m_unscaledPoints[0], m_numPoints, maxDot);
+		supportVerticesOut[j][3] = btScalar(-BT_LARGE_FLOAT);
+		if (0 <= index)
+		{
+			//WARNING: don't swap next lines, the w component would get overwritten!
+			supportVerticesOut[j] = getScaledPoint(index);
+			supportVerticesOut[j][3] = maxDot;
+		}
+	}
 }
-	
 
-
-btVector3	btConvexPointCloudShape::localGetSupportingVertex(const btVector3& vec)const
+btVector3 btConvexPointCloudShape::localGetSupportingVertex(const btVector3& vec) const
 {
 	btVector3 supVertex = localGetSupportingVertexWithoutMargin(vec);
 
-	if ( getMargin()!=btScalar(0.) )
+	if (getMargin() != btScalar(0.))
 	{
 		btVector3 vecnorm = vec;
-		if (vecnorm .length2() < (SIMD_EPSILON*SIMD_EPSILON))
+		if (vecnorm.length2() < (SIMD_EPSILON * SIMD_EPSILON))
 		{
-			vecnorm.setValue(btScalar(-1.),btScalar(-1.),btScalar(-1.));
-		} 
+			vecnorm.setValue(btScalar(-1.), btScalar(-1.), btScalar(-1.));
+		}
 		vecnorm.normalize();
-		supVertex+= getMargin() * vecnorm;
+		supVertex += getMargin() * vecnorm;
 	}
 	return supVertex;
 }
 
-
 #endif
 
-
-
-
-
-
 //currently just for debugging (drawing), perhaps future support for algebraic continuous collision detection
 //Please note that you can debug-draw btConvexHullShape with the Raytracer Demo
-int	btConvexPointCloudShape::getNumVertices() const
+int btConvexPointCloudShape::getNumVertices() const
 {
 	return m_numPoints;
 }
@@ -109,31 +101,29 @@ int btConvexPointCloudShape::getNumEdges() const
 	return 0;
 }
 
-void btConvexPointCloudShape::getEdge(int i,btVector3& pa,btVector3& pb) const
+void btConvexPointCloudShape::getEdge(int i, btVector3& pa, btVector3& pb) const
 {
-	btAssert (0);
+	btAssert(0);
 }
 
-void btConvexPointCloudShape::getVertex(int i,btVector3& vtx) const
+void btConvexPointCloudShape::getVertex(int i, btVector3& vtx) const
 {
-	vtx = m_unscaledPoints[i]*m_localScaling;
+	vtx = m_unscaledPoints[i] * m_localScaling;
 }
 
-int	btConvexPointCloudShape::getNumPlanes() const
+int btConvexPointCloudShape::getNumPlanes() const
 {
 	return 0;
 }
 
-void btConvexPointCloudShape::getPlane(btVector3& ,btVector3& ,int ) const
+void btConvexPointCloudShape::getPlane(btVector3&, btVector3&, int) const
 {
-
 	btAssert(0);
 }
 
 //not yet
-bool btConvexPointCloudShape::isInside(const btVector3& ,btScalar ) const
+bool btConvexPointCloudShape::isInside(const btVector3&, btScalar) const
 {
 	btAssert(0);
 	return false;
 }
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexPointCloudShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexPointCloudShape.h
index 54b5afac3ec..c7d554a4d3d 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexPointCloudShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexPointCloudShape.h
@@ -17,11 +17,12 @@ subject to the following restrictions:
 #define BT_CONVEX_POINT_CLOUD_SHAPE_H
 
 #include "btPolyhedralConvexShape.h"
-#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h" // for the types
+#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h"  // for the types
 #include "LinearMath/btAlignedObjectArray.h"
 
 ///The btConvexPointCloudShape implements an implicit convex hull of an array of vertices.
-ATTRIBUTE_ALIGNED16(class) btConvexPointCloudShape : public btPolyhedralConvexAabbCachingShape
+ATTRIBUTE_ALIGNED16(class)
+btConvexPointCloudShape : public btPolyhedralConvexAabbCachingShape
 {
 	btVector3* m_unscaledPoints;
 	int m_numPoints;
@@ -31,13 +32,13 @@ public:
 
 	btConvexPointCloudShape()
 	{
-		m_localScaling.setValue(1.f,1.f,1.f);
+		m_localScaling.setValue(1.f, 1.f, 1.f);
 		m_shapeType = CONVEX_POINT_CLOUD_SHAPE_PROXYTYPE;
 		m_unscaledPoints = 0;
 		m_numPoints = 0;
 	}
 
-	btConvexPointCloudShape(btVector3* points,int numPoints, const btVector3& localScaling,bool computeAabb = true)
+	btConvexPointCloudShape(btVector3 * points, int numPoints, const btVector3& localScaling, bool computeAabb = true)
 	{
 		m_localScaling = localScaling;
 		m_shapeType = CONVEX_POINT_CLOUD_SHAPE_PROXYTYPE;
@@ -48,7 +49,7 @@ public:
 			recalcLocalAabb();
 	}
 
-	void setPoints (btVector3* points, int numPoints, bool computeAabb = true,const btVector3& localScaling=btVector3(1.f,1.f,1.f))
+	void setPoints(btVector3 * points, int numPoints, bool computeAabb = true, const btVector3& localScaling = btVector3(1.f, 1.f, 1.f))
 	{
 		m_unscaledPoints = points;
 		m_numPoints = numPoints;
@@ -58,48 +59,45 @@ public:
 			recalcLocalAabb();
 	}
 
-	SIMD_FORCE_INLINE	btVector3* getUnscaledPoints()
+	SIMD_FORCE_INLINE btVector3* getUnscaledPoints()
 	{
 		return m_unscaledPoints;
 	}
 
-	SIMD_FORCE_INLINE	const btVector3* getUnscaledPoints() const
+	SIMD_FORCE_INLINE const btVector3* getUnscaledPoints() const
 	{
 		return m_unscaledPoints;
 	}
 
-	SIMD_FORCE_INLINE	int getNumPoints() const 
+	SIMD_FORCE_INLINE int getNumPoints() const
 	{
 		return m_numPoints;
 	}
 
-	SIMD_FORCE_INLINE	btVector3	getScaledPoint( int index) const
+	SIMD_FORCE_INLINE btVector3 getScaledPoint(int index) const
 	{
 		return m_unscaledPoints[index] * m_localScaling;
 	}
 
 #ifndef __SPU__
-	virtual btVector3	localGetSupportingVertex(const btVector3& vec)const;
-	virtual btVector3	localGetSupportingVertexWithoutMargin(const btVector3& vec)const;
-	virtual void	batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const;
+	virtual btVector3 localGetSupportingVertex(const btVector3& vec) const;
+	virtual btVector3 localGetSupportingVertexWithoutMargin(const btVector3& vec) const;
+	virtual void batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const;
 #endif
 
-
 	//debugging
-	virtual const char*	getName()const {return "ConvexPointCloud";}
+	virtual const char* getName() const { return "ConvexPointCloud"; }
 
-	virtual int	getNumVertices() const;
+	virtual int getNumVertices() const;
 	virtual int getNumEdges() const;
-	virtual void getEdge(int i,btVector3& pa,btVector3& pb) const;
-	virtual void getVertex(int i,btVector3& vtx) const;
-	virtual int	getNumPlanes() const;
-	virtual void getPlane(btVector3& planeNormal,btVector3& planeSupport,int i ) const;
-	virtual	bool isInside(const btVector3& pt,btScalar tolerance) const;
+	virtual void getEdge(int i, btVector3& pa, btVector3& pb) const;
+	virtual void getVertex(int i, btVector3& vtx) const;
+	virtual int getNumPlanes() const;
+	virtual void getPlane(btVector3 & planeNormal, btVector3 & planeSupport, int i) const;
+	virtual bool isInside(const btVector3& pt, btScalar tolerance) const;
 
 	///in case we receive negative scaling
-	virtual void	setLocalScaling(const btVector3& scaling);
+	virtual void setLocalScaling(const btVector3& scaling);
 };
 
-
-#endif //BT_CONVEX_POINT_CLOUD_SHAPE_H
-
+#endif  //BT_CONVEX_POINT_CLOUD_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexPolyhedron.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexPolyhedron.cpp
index 0fea00df5c1..65b669e1c05 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexPolyhedron.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexPolyhedron.cpp
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 ///This file was written by Erwin Coumans
 ///Separating axis rest based on work from Pierre Terdiman, see
 ///And contact clipping based on work from Simon Hobbs
@@ -21,49 +20,45 @@ subject to the following restrictions:
 #include "btConvexPolyhedron.h"
 #include "LinearMath/btHashMap.h"
 
-
 btConvexPolyhedron::btConvexPolyhedron()
 {
-
 }
 btConvexPolyhedron::~btConvexPolyhedron()
 {
-
 }
 
-
 inline bool IsAlmostZero(const btVector3& v)
 {
-	if(btFabs(v.x())>1e-6 || btFabs(v.y())>1e-6 || btFabs(v.z())>1e-6)	return false;
+	if (btFabs(v.x()) > 1e-6 || btFabs(v.y()) > 1e-6 || btFabs(v.z()) > 1e-6) return false;
 	return true;
 }
 
 struct btInternalVertexPair
 {
-	btInternalVertexPair(short int v0,short int v1)
-		:m_v0(v0),
-		m_v1(v1)
+	btInternalVertexPair(short int v0, short int v1)
+		: m_v0(v0),
+		  m_v1(v1)
 	{
-		if (m_v1>m_v0)
-			btSwap(m_v0,m_v1);
+		if (m_v1 > m_v0)
+			btSwap(m_v0, m_v1);
 	}
 	short int m_v0;
 	short int m_v1;
 	int getHash() const
 	{
-		return m_v0+(m_v1<<16);
+		return m_v0 + (m_v1 << 16);
 	}
 	bool equals(const btInternalVertexPair& other) const
 	{
-		return m_v0==other.m_v0 && m_v1==other.m_v1;
+		return m_v0 == other.m_v0 && m_v1 == other.m_v1;
 	}
 };
 
 struct btInternalEdge
 {
 	btInternalEdge()
-		:m_face0(-1),
-		m_face1(-1)
+		: m_face0(-1),
+		  m_face1(-1)
 	{
 	}
 	short int m_face0;
@@ -75,23 +70,31 @@ struct btInternalEdge
 #ifdef TEST_INTERNAL_OBJECTS
 bool btConvexPolyhedron::testContainment() const
 {
-	for(int p=0;p<8;p++)
+	for (int p = 0; p < 8; p++)
 	{
 		btVector3 LocalPt;
-		if(p==0)		LocalPt = m_localCenter + btVector3(m_extents[0], m_extents[1], m_extents[2]);
-		else if(p==1)	LocalPt = m_localCenter + btVector3(m_extents[0], m_extents[1], -m_extents[2]);
-		else if(p==2)	LocalPt = m_localCenter + btVector3(m_extents[0], -m_extents[1], m_extents[2]);
-		else if(p==3)	LocalPt = m_localCenter + btVector3(m_extents[0], -m_extents[1], -m_extents[2]);
-		else if(p==4)	LocalPt = m_localCenter + btVector3(-m_extents[0], m_extents[1], m_extents[2]);
-		else if(p==5)	LocalPt = m_localCenter + btVector3(-m_extents[0], m_extents[1], -m_extents[2]);
-		else if(p==6)	LocalPt = m_localCenter + btVector3(-m_extents[0], -m_extents[1], m_extents[2]);
-		else if(p==7)	LocalPt = m_localCenter + btVector3(-m_extents[0], -m_extents[1], -m_extents[2]);
+		if (p == 0)
+			LocalPt = m_localCenter + btVector3(m_extents[0], m_extents[1], m_extents[2]);
+		else if (p == 1)
+			LocalPt = m_localCenter + btVector3(m_extents[0], m_extents[1], -m_extents[2]);
+		else if (p == 2)
+			LocalPt = m_localCenter + btVector3(m_extents[0], -m_extents[1], m_extents[2]);
+		else if (p == 3)
+			LocalPt = m_localCenter + btVector3(m_extents[0], -m_extents[1], -m_extents[2]);
+		else if (p == 4)
+			LocalPt = m_localCenter + btVector3(-m_extents[0], m_extents[1], m_extents[2]);
+		else if (p == 5)
+			LocalPt = m_localCenter + btVector3(-m_extents[0], m_extents[1], -m_extents[2]);
+		else if (p == 6)
+			LocalPt = m_localCenter + btVector3(-m_extents[0], -m_extents[1], m_extents[2]);
+		else if (p == 7)
+			LocalPt = m_localCenter + btVector3(-m_extents[0], -m_extents[1], -m_extents[2]);
 
-		for(int i=0;i<m_faces.size();i++)
+		for (int i = 0; i < m_faces.size(); i++)
 		{
 			const btVector3 Normal(m_faces[i].m_plane[0], m_faces[i].m_plane[1], m_faces[i].m_plane[2]);
 			const btScalar d = LocalPt.dot(Normal) + m_faces[i].m_plane[3];
-			if(d>0.0f)
+			if (d > 0.0f)
 				return false;
 		}
 	}
@@ -99,33 +102,28 @@ bool btConvexPolyhedron::testContainment() const
 }
 #endif
 
-void	btConvexPolyhedron::initialize()
+void btConvexPolyhedron::initialize()
 {
+	btHashMap<btInternalVertexPair, btInternalEdge> edges;
 
-	btHashMap<btInternalVertexPair,btInternalEdge> edges;
-
-	
-	
-	
-	for(int i=0;i<m_faces.size();i++)
+	for (int i = 0; i < m_faces.size(); i++)
 	{
 		int numVertices = m_faces[i].m_indices.size();
 		int NbTris = numVertices;
-		for(int j=0;j<NbTris;j++)
+		for (int j = 0; j < NbTris; j++)
 		{
-			int k = (j+1)%numVertices;
-			btInternalVertexPair vp(m_faces[i].m_indices[j],m_faces[i].m_indices[k]);
+			int k = (j + 1) % numVertices;
+			btInternalVertexPair vp(m_faces[i].m_indices[j], m_faces[i].m_indices[k]);
 			btInternalEdge* edptr = edges.find(vp);
-			btVector3 edge = m_vertices[vp.m_v1]-m_vertices[vp.m_v0];
+			btVector3 edge = m_vertices[vp.m_v1] - m_vertices[vp.m_v0];
 			edge.normalize();
 
 			bool found = false;
 
-			for (int p=0;p<m_uniqueEdges.size();p++)
+			for (int p = 0; p < m_uniqueEdges.size(); p++)
 			{
-				
-				if (IsAlmostZero(m_uniqueEdges[p]-edge) || 
-					IsAlmostZero(m_uniqueEdges[p]+edge))
+				if (IsAlmostZero(m_uniqueEdges[p] - edge) ||
+					IsAlmostZero(m_uniqueEdges[p] + edge))
 				{
 					found = true;
 					break;
@@ -139,112 +137,107 @@ void	btConvexPolyhedron::initialize()
 
 			if (edptr)
 			{
-				btAssert(edptr->m_face0>=0);
-				btAssert(edptr->m_face1<0);
+				btAssert(edptr->m_face0 >= 0);
+				btAssert(edptr->m_face1 < 0);
 				edptr->m_face1 = i;
-			} else
+			}
+			else
 			{
 				btInternalEdge ed;
 				ed.m_face0 = i;
-				edges.insert(vp,ed);
+				edges.insert(vp, ed);
 			}
 		}
 	}
 
 #ifdef USE_CONNECTED_FACES
-	for(int i=0;i<m_faces.size();i++)
+	for (int i = 0; i < m_faces.size(); i++)
 	{
 		int numVertices = m_faces[i].m_indices.size();
 		m_faces[i].m_connectedFaces.resize(numVertices);
 
-		for(int j=0;j<numVertices;j++)
+		for (int j = 0; j < numVertices; j++)
 		{
-			int k = (j+1)%numVertices;
-			btInternalVertexPair vp(m_faces[i].m_indices[j],m_faces[i].m_indices[k]);
+			int k = (j + 1) % numVertices;
+			btInternalVertexPair vp(m_faces[i].m_indices[j], m_faces[i].m_indices[k]);
 			btInternalEdge* edptr = edges.find(vp);
 			btAssert(edptr);
-			btAssert(edptr->m_face0>=0);
-			btAssert(edptr->m_face1>=0);
+			btAssert(edptr->m_face0 >= 0);
+			btAssert(edptr->m_face1 >= 0);
 
-			int connectedFace = (edptr->m_face0==i)?edptr->m_face1:edptr->m_face0;
+			int connectedFace = (edptr->m_face0 == i) ? edptr->m_face1 : edptr->m_face0;
 			m_faces[i].m_connectedFaces[j] = connectedFace;
 		}
 	}
-#endif//USE_CONNECTED_FACES
+#endif  //USE_CONNECTED_FACES
 
 	initialize2();
 }
 
-void	btConvexPolyhedron::initialize2()
+void btConvexPolyhedron::initialize2()
 {
 	m_localCenter.setValue(0, 0, 0);
 	btScalar TotalArea = 0.0f;
-	for(int i=0;i<m_faces.size();i++)
+	for (int i = 0; i < m_faces.size(); i++)
 	{
 		int numVertices = m_faces[i].m_indices.size();
-		int NbTris = numVertices-2;
-		
+		int NbTris = numVertices - 2;
+
 		const btVector3& p0 = m_vertices[m_faces[i].m_indices[0]];
-		for(int j=1;j<=NbTris;j++)
+		for (int j = 1; j <= NbTris; j++)
 		{
-			int k = (j+1)%numVertices;
+			int k = (j + 1) % numVertices;
 			const btVector3& p1 = m_vertices[m_faces[i].m_indices[j]];
 			const btVector3& p2 = m_vertices[m_faces[i].m_indices[k]];
 			btScalar Area = ((p0 - p1).cross(p0 - p2)).length() * 0.5f;
-			btVector3 Center = (p0+p1+p2)/3.0f;
+			btVector3 Center = (p0 + p1 + p2) / 3.0f;
 			m_localCenter += Area * Center;
 			TotalArea += Area;
 		}
 	}
 	m_localCenter /= TotalArea;
 
-
-
-
 #ifdef TEST_INTERNAL_OBJECTS
-	if(1)
+	if (1)
 	{
 		m_radius = FLT_MAX;
-		for(int i=0;i<m_faces.size();i++)
+		for (int i = 0; i < m_faces.size(); i++)
 		{
 			const btVector3 Normal(m_faces[i].m_plane[0], m_faces[i].m_plane[1], m_faces[i].m_plane[2]);
 			const btScalar dist = btFabs(m_localCenter.dot(Normal) + m_faces[i].m_plane[3]);
-			if(dist<m_radius)
+			if (dist < m_radius)
 				m_radius = dist;
 		}
 
-	
 		btScalar MinX = FLT_MAX;
 		btScalar MinY = FLT_MAX;
 		btScalar MinZ = FLT_MAX;
 		btScalar MaxX = -FLT_MAX;
 		btScalar MaxY = -FLT_MAX;
 		btScalar MaxZ = -FLT_MAX;
-		for(int i=0; i<m_vertices.size(); i++)
+		for (int i = 0; i < m_vertices.size(); i++)
 		{
 			const btVector3& pt = m_vertices[i];
-			if(pt.x()<MinX)	MinX = pt.x();
-			if(pt.x()>MaxX)	MaxX = pt.x();
-			if(pt.y()<MinY)	MinY = pt.y();
-			if(pt.y()>MaxY)	MaxY = pt.y();
-			if(pt.z()<MinZ)	MinZ = pt.z();
-			if(pt.z()>MaxZ)	MaxZ = pt.z();
+			if (pt.x() < MinX) MinX = pt.x();
+			if (pt.x() > MaxX) MaxX = pt.x();
+			if (pt.y() < MinY) MinY = pt.y();
+			if (pt.y() > MaxY) MaxY = pt.y();
+			if (pt.z() < MinZ) MinZ = pt.z();
+			if (pt.z() > MaxZ) MaxZ = pt.z();
 		}
-		mC.setValue(MaxX+MinX, MaxY+MinY, MaxZ+MinZ);
-		mE.setValue(MaxX-MinX, MaxY-MinY, MaxZ-MinZ);
+		mC.setValue(MaxX + MinX, MaxY + MinY, MaxZ + MinZ);
+		mE.setValue(MaxX - MinX, MaxY - MinY, MaxZ - MinZ);
 
-
-
-//		const btScalar r = m_radius / sqrtf(2.0f);
+		//		const btScalar r = m_radius / sqrtf(2.0f);
 		const btScalar r = m_radius / sqrtf(3.0f);
 		const int LargestExtent = mE.maxAxis();
-		const btScalar Step = (mE[LargestExtent]*0.5f - r)/1024.0f;
+		const btScalar Step = (mE[LargestExtent] * 0.5f - r) / 1024.0f;
 		m_extents[0] = m_extents[1] = m_extents[2] = r;
-		m_extents[LargestExtent] = mE[LargestExtent]*0.5f;
+		m_extents[LargestExtent] = mE[LargestExtent] * 0.5f;
 		bool FoundBox = false;
-		for(int j=0;j<1024;j++)
+		for (int j = 0; j < 1024; j++)
 		{
-			if(testContainment())
+			if (testContainment())
 			{
 				FoundBox = true;
 				break;
@@ -252,25 +245,25 @@ void	btConvexPolyhedron::initialize2()
 
 			m_extents[LargestExtent] -= Step;
 		}
-		if(!FoundBox)
+		if (!FoundBox)
 		{
 			m_extents[0] = m_extents[1] = m_extents[2] = r;
 		}
 		else
 		{
 			// Refine the box
-			const btScalar Step = (m_radius - r)/1024.0f;
-			const int e0 = (1<<LargestExtent) & 3;
-			const int e1 = (1<<e0) & 3;
+			const btScalar Step = (m_radius - r) / 1024.0f;
+			const int e0 = (1 << LargestExtent) & 3;
+			const int e1 = (1 << e0) & 3;
 
-			for(int j=0;j<1024;j++)
+			for (int j = 0; j < 1024; j++)
 			{
 				const btScalar Saved0 = m_extents[e0];
 				const btScalar Saved1 = m_extents[e1];
 				m_extents[e0] += Step;
 				m_extents[e1] += Step;
 
-				if(!testContainment())
+				if (!testContainment())
 				{
 					m_extents[e0] = Saved0;
 					m_extents[e1] = Saved1;
@@ -281,29 +274,29 @@ void	btConvexPolyhedron::initialize2()
 	}
 #endif
 }
-void btConvexPolyhedron::project(const btTransform& trans, const btVector3& dir, btScalar& minProj, btScalar& maxProj, btVector3& witnesPtMin,btVector3& witnesPtMax) const
+void btConvexPolyhedron::project(const btTransform& trans, const btVector3& dir, btScalar& minProj, btScalar& maxProj, btVector3& witnesPtMin, btVector3& witnesPtMax) const
 {
 	minProj = FLT_MAX;
 	maxProj = -FLT_MAX;
 	int numVerts = m_vertices.size();
-	for(int i=0;i<numVerts;i++)
+	for (int i = 0; i < numVerts; i++)
 	{
 		btVector3 pt = trans * m_vertices[i];
 		btScalar dp = pt.dot(dir);
-		if(dp < minProj)
+		if (dp < minProj)
 		{
 			minProj = dp;
 			witnesPtMin = pt;
 		}
-		if(dp > maxProj)
+		if (dp > maxProj)
 		{
 			maxProj = dp;
 			witnesPtMax = pt;
 		}
 	}
-	if(minProj>maxProj)
+	if (minProj > maxProj)
 	{
-		btSwap(minProj,maxProj);
-		btSwap(witnesPtMin,witnesPtMax);
+		btSwap(minProj, maxProj);
+		btSwap(witnesPtMin, witnesPtMax);
 	}
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexPolyhedron.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexPolyhedron.h
index c5aa20f4532..638aa9b3df0 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexPolyhedron.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexPolyhedron.h
@@ -13,10 +13,8 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 ///This file was written by Erwin Coumans
 
-
 #ifndef _BT_POLYHEDRAL_FEATURES_H
 #define _BT_POLYHEDRAL_FEATURES_H
 
@@ -25,42 +23,37 @@ subject to the following restrictions:
 
 #define TEST_INTERNAL_OBJECTS 1
 
-
 struct btFace
 {
-	btAlignedObjectArray<int>	m_indices;
-//	btAlignedObjectArray<int>	m_connectedFaces;
-	btScalar	m_plane[4];
+	btAlignedObjectArray<int> m_indices;
+	//	btAlignedObjectArray<int>	m_connectedFaces;
+	btScalar m_plane[4];
 };
 
-
-ATTRIBUTE_ALIGNED16(class) btConvexPolyhedron
+ATTRIBUTE_ALIGNED16(class)
+btConvexPolyhedron
 {
-	public:
-		
+public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-		
-	btConvexPolyhedron();
-	virtual	~btConvexPolyhedron();
 
-	btAlignedObjectArray<btVector3>	m_vertices;
-	btAlignedObjectArray<btFace>	m_faces;
+	btConvexPolyhedron();
+	virtual ~btConvexPolyhedron();
+
+	btAlignedObjectArray<btVector3> m_vertices;
+	btAlignedObjectArray<btFace> m_faces;
 	btAlignedObjectArray<btVector3> m_uniqueEdges;
 
-	btVector3		m_localCenter;
-	btVector3		m_extents;
-	btScalar		m_radius;
-	btVector3		mC;
-	btVector3		mE;
+	btVector3 m_localCenter;
+	btVector3 m_extents;
+	btScalar m_radius;
+	btVector3 mC;
+	btVector3 mE;
 
-	void	initialize();
-	void	initialize2();
+	void initialize();
+	void initialize2();
 	bool testContainment() const;
 
-	void project(const btTransform& trans, const btVector3& dir, btScalar& minProj, btScalar& maxProj, btVector3& witnesPtMin,btVector3& witnesPtMax) const;
+	void project(const btTransform& trans, const btVector3& dir, btScalar& minProj, btScalar& maxProj, btVector3& witnesPtMin, btVector3& witnesPtMax) const;
 };
 
-	
-#endif //_BT_POLYHEDRAL_FEATURES_H
-
-
+#endif  //_BT_POLYHEDRAL_FEATURES_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexShape.cpp
index 2f84858598a..f8fb0aa9fda 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexShape.cpp
@@ -13,7 +13,7 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-#if defined (_WIN32) || defined (__i386__)
+#if defined(_WIN32) || defined(__i386__)
 #define BT_USE_SSE_IN_API
 #endif
 
@@ -27,30 +27,28 @@ subject to the following restrictions:
 #include "btConvexPointCloudShape.h"
 
 ///not supported on IBM SDK, until we fix the alignment of btVector3
-#if defined (__CELLOS_LV2__) && defined (__SPU__)
+#if defined(__CELLOS_LV2__) && defined(__SPU__)
 #include <spu_intrinsics.h>
-static inline vec_float4 vec_dot3( vec_float4 vec0, vec_float4 vec1 )
+static inline vec_float4 vec_dot3(vec_float4 vec0, vec_float4 vec1)
 {
-    vec_float4 result;
-    result = spu_mul( vec0, vec1 );
-    result = spu_madd( spu_rlqwbyte( vec0, 4 ), spu_rlqwbyte( vec1, 4 ), result );
-    return spu_madd( spu_rlqwbyte( vec0, 8 ), spu_rlqwbyte( vec1, 8 ), result );
+	vec_float4 result;
+	result = spu_mul(vec0, vec1);
+	result = spu_madd(spu_rlqwbyte(vec0, 4), spu_rlqwbyte(vec1, 4), result);
+	return spu_madd(spu_rlqwbyte(vec0, 8), spu_rlqwbyte(vec1, 8), result);
 }
-#endif //__SPU__
+#endif  //__SPU__
 
-btConvexShape::btConvexShape ()
+btConvexShape::btConvexShape()
 {
 }
 
 btConvexShape::~btConvexShape()
 {
-
 }
 
-
-void btConvexShape::project(const btTransform& trans, const btVector3& dir, btScalar& min, btScalar& max, btVector3& witnesPtMin,btVector3& witnesPtMax) const
+void btConvexShape::project(const btTransform& trans, const btVector3& dir, btScalar& min, btScalar& max, btVector3& witnesPtMin, btVector3& witnesPtMax) const
 {
-	btVector3 localAxis = dir*trans.getBasis();
+	btVector3 localAxis = dir * trans.getBasis();
 	btVector3 vtx1 = trans(localGetSupportingVertex(localAxis));
 	btVector3 vtx2 = trans(localGetSupportingVertex(-localAxis));
 
@@ -58,8 +56,8 @@ void btConvexShape::project(const btTransform& trans, const btVector3& dir, btSc
 	max = vtx2.dot(dir);
 	witnesPtMax = vtx2;
 	witnesPtMin = vtx1;
-	
-	if(min>max)
+
+	if (min > max)
 	{
 		btScalar tmp = min;
 		min = max;
@@ -69,391 +67,392 @@ void btConvexShape::project(const btTransform& trans, const btVector3& dir, btSc
 	}
 }
 
-
-static btVector3 convexHullSupport (const btVector3& localDirOrg, const btVector3* points, int numPoints, const btVector3& localScaling)
-{	
-
+static btVector3 convexHullSupport(const btVector3& localDirOrg, const btVector3* points, int numPoints, const btVector3& localScaling)
+{
 	btVector3 vec = localDirOrg * localScaling;
 
-#if defined (__CELLOS_LV2__) && defined (__SPU__)
+#if defined(__CELLOS_LV2__) && defined(__SPU__)
 
 	btVector3 localDir = vec;
 
-	vec_float4 v_distMax = {-FLT_MAX,0,0,0};
-	vec_int4 v_idxMax = {-999,0,0,0};
-	int v=0;
+	vec_float4 v_distMax = {-FLT_MAX, 0, 0, 0};
+	vec_int4 v_idxMax = {-999, 0, 0, 0};
+	int v = 0;
 	int numverts = numPoints;
 
-	for(;v<(int)numverts-4;v+=4) {
-		vec_float4 p0 = vec_dot3(points[v  ].get128(),localDir.get128());
-		vec_float4 p1 = vec_dot3(points[v+1].get128(),localDir.get128());
-		vec_float4 p2 = vec_dot3(points[v+2].get128(),localDir.get128());
-		vec_float4 p3 = vec_dot3(points[v+3].get128(),localDir.get128());
-		const vec_int4 i0 = {v  ,0,0,0};
-		const vec_int4 i1 = {v+1,0,0,0};
-		const vec_int4 i2 = {v+2,0,0,0};
-		const vec_int4 i3 = {v+3,0,0,0};
-		vec_uint4  retGt01 = spu_cmpgt(p0,p1);
-		vec_float4 pmax01 = spu_sel(p1,p0,retGt01);
-		vec_int4   imax01 = spu_sel(i1,i0,retGt01);
-		vec_uint4  retGt23 = spu_cmpgt(p2,p3);
-		vec_float4 pmax23 = spu_sel(p3,p2,retGt23);
-		vec_int4   imax23 = spu_sel(i3,i2,retGt23);
-		vec_uint4  retGt0123 = spu_cmpgt(pmax01,pmax23);
-		vec_float4 pmax0123 = spu_sel(pmax23,pmax01,retGt0123);
-		vec_int4   imax0123 = spu_sel(imax23,imax01,retGt0123);
-		vec_uint4  retGtMax = spu_cmpgt(v_distMax,pmax0123);
-		v_distMax = spu_sel(pmax0123,v_distMax,retGtMax);
-		v_idxMax = spu_sel(imax0123,v_idxMax,retGtMax);
+	for (; v < (int)numverts - 4; v += 4)
+	{
+		vec_float4 p0 = vec_dot3(points[v].get128(), localDir.get128());
+		vec_float4 p1 = vec_dot3(points[v + 1].get128(), localDir.get128());
+		vec_float4 p2 = vec_dot3(points[v + 2].get128(), localDir.get128());
+		vec_float4 p3 = vec_dot3(points[v + 3].get128(), localDir.get128());
+		const vec_int4 i0 = {v, 0, 0, 0};
+		const vec_int4 i1 = {v + 1, 0, 0, 0};
+		const vec_int4 i2 = {v + 2, 0, 0, 0};
+		const vec_int4 i3 = {v + 3, 0, 0, 0};
+		vec_uint4 retGt01 = spu_cmpgt(p0, p1);
+		vec_float4 pmax01 = spu_sel(p1, p0, retGt01);
+		vec_int4 imax01 = spu_sel(i1, i0, retGt01);
+		vec_uint4 retGt23 = spu_cmpgt(p2, p3);
+		vec_float4 pmax23 = spu_sel(p3, p2, retGt23);
+		vec_int4 imax23 = spu_sel(i3, i2, retGt23);
+		vec_uint4 retGt0123 = spu_cmpgt(pmax01, pmax23);
+		vec_float4 pmax0123 = spu_sel(pmax23, pmax01, retGt0123);
+		vec_int4 imax0123 = spu_sel(imax23, imax01, retGt0123);
+		vec_uint4 retGtMax = spu_cmpgt(v_distMax, pmax0123);
+		v_distMax = spu_sel(pmax0123, v_distMax, retGtMax);
+		v_idxMax = spu_sel(imax0123, v_idxMax, retGtMax);
 	}
-	for(;v<(int)numverts;v++) {
-		vec_float4 p = vec_dot3(points[v].get128(),localDir.get128());
-		const vec_int4 i = {v,0,0,0};
-		vec_uint4  retGtMax = spu_cmpgt(v_distMax,p);
-		v_distMax = spu_sel(p,v_distMax,retGtMax);
-		v_idxMax = spu_sel(i,v_idxMax,retGtMax);
+	for (; v < (int)numverts; v++)
+	{
+		vec_float4 p = vec_dot3(points[v].get128(), localDir.get128());
+		const vec_int4 i = {v, 0, 0, 0};
+		vec_uint4 retGtMax = spu_cmpgt(v_distMax, p);
+		v_distMax = spu_sel(p, v_distMax, retGtMax);
+		v_idxMax = spu_sel(i, v_idxMax, retGtMax);
 	}
-	int ptIndex = spu_extract(v_idxMax,0);
-	const btVector3& supVec= points[ptIndex] * localScaling;
+	int ptIndex = spu_extract(v_idxMax, 0);
+	const btVector3& supVec = points[ptIndex] * localScaling;
 	return supVec;
 #else
 
 	btScalar maxDot;
-	long ptIndex = vec.maxDot( points, numPoints, maxDot);
+	long ptIndex = vec.maxDot(points, numPoints, maxDot);
 	btAssert(ptIndex >= 0);
-	if (ptIndex<0)
+	if (ptIndex < 0)
 	{
 		ptIndex = 0;
 	}
 	btVector3 supVec = points[ptIndex] * localScaling;
 	return supVec;
-#endif //__SPU__
+#endif  //__SPU__
 }
 
-btVector3 btConvexShape::localGetSupportVertexWithoutMarginNonVirtual (const btVector3& localDir) const
+btVector3 btConvexShape::localGetSupportVertexWithoutMarginNonVirtual(const btVector3& localDir) const
 {
 	switch (m_shapeType)
 	{
-    case SPHERE_SHAPE_PROXYTYPE:
-	{
-		return btVector3(0,0,0);
-    }
-	case BOX_SHAPE_PROXYTYPE:
-	{
-		btBoxShape* convexShape = (btBoxShape*)this;
-		const btVector3& halfExtents = convexShape->getImplicitShapeDimensions();
+		case SPHERE_SHAPE_PROXYTYPE:
+		{
+			return btVector3(0, 0, 0);
+		}
+		case BOX_SHAPE_PROXYTYPE:
+		{
+			btBoxShape* convexShape = (btBoxShape*)this;
+			const btVector3& halfExtents = convexShape->getImplicitShapeDimensions();
 
-#if defined( __APPLE__ ) && (defined( BT_USE_SSE )||defined( BT_USE_NEON ))
-    #if defined( BT_USE_SSE )
-            return btVector3( _mm_xor_ps( _mm_and_ps( localDir.mVec128, (__m128){-0.0f, -0.0f, -0.0f, -0.0f }), halfExtents.mVec128 ));
-    #elif defined( BT_USE_NEON )
-            return btVector3( (float32x4_t) (((uint32x4_t) localDir.mVec128 & (uint32x4_t){ 0x80000000, 0x80000000, 0x80000000, 0x80000000}) ^ (uint32x4_t) halfExtents.mVec128 ));
-    #else
-        #error unknown vector arch
-    #endif
+#if defined(__APPLE__) && (defined(BT_USE_SSE) || defined(BT_USE_NEON))
+#if defined(BT_USE_SSE)
+			return btVector3(_mm_xor_ps(_mm_and_ps(localDir.mVec128, (__m128){-0.0f, -0.0f, -0.0f, -0.0f}), halfExtents.mVec128));
+#elif defined(BT_USE_NEON)
+			return btVector3((float32x4_t)(((uint32x4_t)localDir.mVec128 & (uint32x4_t){0x80000000, 0x80000000, 0x80000000, 0x80000000}) ^ (uint32x4_t)halfExtents.mVec128));
 #else
-		return btVector3(btFsels(localDir.x(), halfExtents.x(), -halfExtents.x()),
-			btFsels(localDir.y(), halfExtents.y(), -halfExtents.y()),
-			btFsels(localDir.z(), halfExtents.z(), -halfExtents.z()));
+#error unknown vector arch
 #endif
-	}
-	case TRIANGLE_SHAPE_PROXYTYPE:
-	{
-		btTriangleShape* triangleShape = (btTriangleShape*)this;
-		btVector3 dir(localDir.getX(),localDir.getY(),localDir.getZ());
-		btVector3* vertices = &triangleShape->m_vertices1[0];
-        btVector3 dots = dir.dot3(vertices[0], vertices[1], vertices[2]);
-		btVector3 sup = vertices[dots.maxAxis()];
-		return btVector3(sup.getX(),sup.getY(),sup.getZ());
-	}
-	case CYLINDER_SHAPE_PROXYTYPE:
-	{
-		btCylinderShape* cylShape = (btCylinderShape*)this;
-		//mapping of halfextents/dimension onto radius/height depends on how cylinder local orientation is (upAxis)
-
-		btVector3 halfExtents = cylShape->getImplicitShapeDimensions();
-		btVector3 v(localDir.getX(),localDir.getY(),localDir.getZ());
-		int cylinderUpAxis = cylShape->getUpAxis();
-		int XX(1),YY(0),ZZ(2);
-
-		switch (cylinderUpAxis)
-		{
-		case 0:
-		{
-			XX = 1;
-			YY = 0;
-			ZZ = 2;
-		}
-		break;
-		case 1:
-		{
-			XX = 0;
-			YY = 1;
-			ZZ = 2;	
-		}
-		break;
-		case 2:
-		{
-			XX = 0;
-			YY = 2;
-			ZZ = 1;
-			
-		}
-		break;
-		default:
-			btAssert(0);
-		break;
-		};
-
-		btScalar radius = halfExtents[XX];
-		btScalar halfHeight = halfExtents[cylinderUpAxis];
-
-		btVector3 tmp;
-		btScalar d ;
-
-		btScalar s = btSqrt(v[XX] * v[XX] + v[ZZ] * v[ZZ]);
-		if (s != btScalar(0.0))
-		{
-			d = radius / s;  
-			tmp[XX] = v[XX] * d;
-			tmp[YY] = v[YY] < 0.0 ? -halfHeight : halfHeight;
-			tmp[ZZ] = v[ZZ] * d;
-			return btVector3(tmp.getX(),tmp.getY(),tmp.getZ());
-		} else {
-			tmp[XX] = radius;
-			tmp[YY] = v[YY] < 0.0 ? -halfHeight : halfHeight;
-			tmp[ZZ] = btScalar(0.0);
-			return btVector3(tmp.getX(),tmp.getY(),tmp.getZ());
-		}
-	}
-	case CAPSULE_SHAPE_PROXYTYPE:
-	{
-		btVector3 vec0(localDir.getX(),localDir.getY(),localDir.getZ());
-
-		btCapsuleShape* capsuleShape = (btCapsuleShape*)this;
-		btScalar halfHeight = capsuleShape->getHalfHeight();
-		int capsuleUpAxis = capsuleShape->getUpAxis();
-
-		btVector3 supVec(0,0,0);
-
-		btScalar maxDot(btScalar(-BT_LARGE_FLOAT));
-
-		btVector3 vec = vec0;
-		btScalar lenSqr = vec.length2();
-		if (lenSqr < SIMD_EPSILON*SIMD_EPSILON)
-		{
-			vec.setValue(1,0,0);
-		} else
-		{
-			btScalar rlen = btScalar(1.) / btSqrt(lenSqr );
-			vec *= rlen;
-		}
-		btVector3 vtx;
-		btScalar newDot;
-		{
-			btVector3 pos(0,0,0);
-			pos[capsuleUpAxis] = halfHeight;
-
-			vtx = pos;
-			newDot = vec.dot(vtx);
-			
-
-			if (newDot > maxDot)
-			{
-				maxDot = newDot;
-				supVec = vtx;
-			}
-		}
-		{
-			btVector3 pos(0,0,0);
-			pos[capsuleUpAxis] = -halfHeight;
-
-			vtx = pos;
-			newDot = vec.dot(vtx);
-			if (newDot > maxDot)
-			{
-				maxDot = newDot;
-				supVec = vtx;
-			}
-		}
-		return btVector3(supVec.getX(),supVec.getY(),supVec.getZ());	
-	}
-	case CONVEX_POINT_CLOUD_SHAPE_PROXYTYPE:
-	{
-		btConvexPointCloudShape* convexPointCloudShape = (btConvexPointCloudShape*)this;
-		btVector3* points = convexPointCloudShape->getUnscaledPoints ();
-		int numPoints = convexPointCloudShape->getNumPoints ();
-		return convexHullSupport (localDir, points, numPoints,convexPointCloudShape->getLocalScalingNV());
-	}
-	case CONVEX_HULL_SHAPE_PROXYTYPE:
-	{
-		btConvexHullShape* convexHullShape = (btConvexHullShape*)this;
-		btVector3* points = convexHullShape->getUnscaledPoints();
-		int numPoints = convexHullShape->getNumPoints ();
-		return convexHullSupport (localDir, points, numPoints,convexHullShape->getLocalScalingNV());
-	}
-    default:
-#ifndef __SPU__
-		return this->localGetSupportingVertexWithoutMargin (localDir);
 #else
-		btAssert (0);
+			return btVector3(btFsels(localDir.x(), halfExtents.x(), -halfExtents.x()),
+							 btFsels(localDir.y(), halfExtents.y(), -halfExtents.y()),
+							 btFsels(localDir.z(), halfExtents.z(), -halfExtents.z()));
+#endif
+		}
+		case TRIANGLE_SHAPE_PROXYTYPE:
+		{
+			btTriangleShape* triangleShape = (btTriangleShape*)this;
+			btVector3 dir(localDir.getX(), localDir.getY(), localDir.getZ());
+			btVector3* vertices = &triangleShape->m_vertices1[0];
+			btVector3 dots = dir.dot3(vertices[0], vertices[1], vertices[2]);
+			btVector3 sup = vertices[dots.maxAxis()];
+			return btVector3(sup.getX(), sup.getY(), sup.getZ());
+		}
+		case CYLINDER_SHAPE_PROXYTYPE:
+		{
+			btCylinderShape* cylShape = (btCylinderShape*)this;
+			//mapping of halfextents/dimension onto radius/height depends on how cylinder local orientation is (upAxis)
+
+			btVector3 halfExtents = cylShape->getImplicitShapeDimensions();
+			btVector3 v(localDir.getX(), localDir.getY(), localDir.getZ());
+			int cylinderUpAxis = cylShape->getUpAxis();
+			int XX(1), YY(0), ZZ(2);
+
+			switch (cylinderUpAxis)
+			{
+				case 0:
+				{
+					XX = 1;
+					YY = 0;
+					ZZ = 2;
+				}
+				break;
+				case 1:
+				{
+					XX = 0;
+					YY = 1;
+					ZZ = 2;
+				}
+				break;
+				case 2:
+				{
+					XX = 0;
+					YY = 2;
+					ZZ = 1;
+				}
+				break;
+				default:
+					btAssert(0);
+					break;
+			};
+
+			btScalar radius = halfExtents[XX];
+			btScalar halfHeight = halfExtents[cylinderUpAxis];
+
+			btVector3 tmp;
+			btScalar d;
+
+			btScalar s = btSqrt(v[XX] * v[XX] + v[ZZ] * v[ZZ]);
+			if (s != btScalar(0.0))
+			{
+				d = radius / s;
+				tmp[XX] = v[XX] * d;
+				tmp[YY] = v[YY] < 0.0 ? -halfHeight : halfHeight;
+				tmp[ZZ] = v[ZZ] * d;
+				return btVector3(tmp.getX(), tmp.getY(), tmp.getZ());
+			}
+			else
+			{
+				tmp[XX] = radius;
+				tmp[YY] = v[YY] < 0.0 ? -halfHeight : halfHeight;
+				tmp[ZZ] = btScalar(0.0);
+				return btVector3(tmp.getX(), tmp.getY(), tmp.getZ());
+			}
+		}
+		case CAPSULE_SHAPE_PROXYTYPE:
+		{
+			btVector3 vec0(localDir.getX(), localDir.getY(), localDir.getZ());
+
+			btCapsuleShape* capsuleShape = (btCapsuleShape*)this;
+			btScalar halfHeight = capsuleShape->getHalfHeight();
+			int capsuleUpAxis = capsuleShape->getUpAxis();
+
+			btVector3 supVec(0, 0, 0);
+
+			btScalar maxDot(btScalar(-BT_LARGE_FLOAT));
+
+			btVector3 vec = vec0;
+			btScalar lenSqr = vec.length2();
+			if (lenSqr < SIMD_EPSILON * SIMD_EPSILON)
+			{
+				vec.setValue(1, 0, 0);
+			}
+			else
+			{
+				btScalar rlen = btScalar(1.) / btSqrt(lenSqr);
+				vec *= rlen;
+			}
+			btVector3 vtx;
+			btScalar newDot;
+			{
+				btVector3 pos(0, 0, 0);
+				pos[capsuleUpAxis] = halfHeight;
+
+				vtx = pos;
+				newDot = vec.dot(vtx);
+
+				if (newDot > maxDot)
+				{
+					maxDot = newDot;
+					supVec = vtx;
+				}
+			}
+			{
+				btVector3 pos(0, 0, 0);
+				pos[capsuleUpAxis] = -halfHeight;
+
+				vtx = pos;
+				newDot = vec.dot(vtx);
+				if (newDot > maxDot)
+				{
+					maxDot = newDot;
+					supVec = vtx;
+				}
+			}
+			return btVector3(supVec.getX(), supVec.getY(), supVec.getZ());
+		}
+		case CONVEX_POINT_CLOUD_SHAPE_PROXYTYPE:
+		{
+			btConvexPointCloudShape* convexPointCloudShape = (btConvexPointCloudShape*)this;
+			btVector3* points = convexPointCloudShape->getUnscaledPoints();
+			int numPoints = convexPointCloudShape->getNumPoints();
+			return convexHullSupport(localDir, points, numPoints, convexPointCloudShape->getLocalScalingNV());
+		}
+		case CONVEX_HULL_SHAPE_PROXYTYPE:
+		{
+			btConvexHullShape* convexHullShape = (btConvexHullShape*)this;
+			btVector3* points = convexHullShape->getUnscaledPoints();
+			int numPoints = convexHullShape->getNumPoints();
+			return convexHullSupport(localDir, points, numPoints, convexHullShape->getLocalScalingNV());
+		}
+		default:
+#ifndef __SPU__
+			return this->localGetSupportingVertexWithoutMargin(localDir);
+#else
+			btAssert(0);
 #endif
 	}
 
 	// should never reach here
-	btAssert (0);
-	return btVector3 (btScalar(0.0f), btScalar(0.0f), btScalar(0.0f));
+	btAssert(0);
+	return btVector3(btScalar(0.0f), btScalar(0.0f), btScalar(0.0f));
 }
 
-btVector3 btConvexShape::localGetSupportVertexNonVirtual (const btVector3& localDir) const
+btVector3 btConvexShape::localGetSupportVertexNonVirtual(const btVector3& localDir) const
 {
 	btVector3 localDirNorm = localDir;
-	if (localDirNorm .length2() < (SIMD_EPSILON*SIMD_EPSILON))
+	if (localDirNorm.length2() < (SIMD_EPSILON * SIMD_EPSILON))
 	{
-		localDirNorm.setValue(btScalar(-1.),btScalar(-1.),btScalar(-1.));
+		localDirNorm.setValue(btScalar(-1.), btScalar(-1.), btScalar(-1.));
 	}
-	localDirNorm.normalize ();
+	localDirNorm.normalize();
 
-	return localGetSupportVertexWithoutMarginNonVirtual(localDirNorm)+ getMarginNonVirtual() * localDirNorm;
+	return localGetSupportVertexWithoutMarginNonVirtual(localDirNorm) + getMarginNonVirtual() * localDirNorm;
 }
 
 /* TODO: This should be bumped up to btCollisionShape () */
-btScalar btConvexShape::getMarginNonVirtual () const
+btScalar btConvexShape::getMarginNonVirtual() const
 {
 	switch (m_shapeType)
 	{
-    case SPHERE_SHAPE_PROXYTYPE:
-	{
-		btSphereShape* sphereShape = (btSphereShape*)this;
-		return sphereShape->getRadius ();
-	}
-	case BOX_SHAPE_PROXYTYPE:
-	{
-		btBoxShape* convexShape = (btBoxShape*)this;
-		return convexShape->getMarginNV ();
-	}
-	case TRIANGLE_SHAPE_PROXYTYPE:
-	{
-		btTriangleShape* triangleShape = (btTriangleShape*)this;
-		return triangleShape->getMarginNV ();
-	}
-	case CYLINDER_SHAPE_PROXYTYPE:
-	{
-		btCylinderShape* cylShape = (btCylinderShape*)this;
-		return cylShape->getMarginNV();
-	}
-	case CONE_SHAPE_PROXYTYPE:
-	{
-		btConeShape* conShape = (btConeShape*)this;
-		return conShape->getMarginNV();
-	}
-	case CAPSULE_SHAPE_PROXYTYPE:
-	{
-		btCapsuleShape* capsuleShape = (btCapsuleShape*)this;
-		return capsuleShape->getMarginNV();
-	}
-	case CONVEX_POINT_CLOUD_SHAPE_PROXYTYPE:
-	/* fall through */
-	case CONVEX_HULL_SHAPE_PROXYTYPE:
-	{
-		btPolyhedralConvexShape* convexHullShape = (btPolyhedralConvexShape*)this;
-		return convexHullShape->getMarginNV();
-	}
-    default:
+		case SPHERE_SHAPE_PROXYTYPE:
+		{
+			btSphereShape* sphereShape = (btSphereShape*)this;
+			return sphereShape->getRadius();
+		}
+		case BOX_SHAPE_PROXYTYPE:
+		{
+			btBoxShape* convexShape = (btBoxShape*)this;
+			return convexShape->getMarginNV();
+		}
+		case TRIANGLE_SHAPE_PROXYTYPE:
+		{
+			btTriangleShape* triangleShape = (btTriangleShape*)this;
+			return triangleShape->getMarginNV();
+		}
+		case CYLINDER_SHAPE_PROXYTYPE:
+		{
+			btCylinderShape* cylShape = (btCylinderShape*)this;
+			return cylShape->getMarginNV();
+		}
+		case CONE_SHAPE_PROXYTYPE:
+		{
+			btConeShape* conShape = (btConeShape*)this;
+			return conShape->getMarginNV();
+		}
+		case CAPSULE_SHAPE_PROXYTYPE:
+		{
+			btCapsuleShape* capsuleShape = (btCapsuleShape*)this;
+			return capsuleShape->getMarginNV();
+		}
+		case CONVEX_POINT_CLOUD_SHAPE_PROXYTYPE:
+		/* fall through */
+		case CONVEX_HULL_SHAPE_PROXYTYPE:
+		{
+			btPolyhedralConvexShape* convexHullShape = (btPolyhedralConvexShape*)this;
+			return convexHullShape->getMarginNV();
+		}
+		default:
 #ifndef __SPU__
-		return this->getMargin ();
+			return this->getMargin();
 #else
-		btAssert (0);
+			btAssert(0);
 #endif
 	}
 
 	// should never reach here
-	btAssert (0);
+	btAssert(0);
 	return btScalar(0.0f);
 }
 #ifndef __SPU__
-void btConvexShape::getAabbNonVirtual (const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
+void btConvexShape::getAabbNonVirtual(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
 {
 	switch (m_shapeType)
 	{
-    case SPHERE_SHAPE_PROXYTYPE:
-	{
-		btSphereShape* sphereShape = (btSphereShape*)this;
-		btScalar radius = sphereShape->getImplicitShapeDimensions().getX();// * convexShape->getLocalScaling().getX();
-		btScalar margin = radius + sphereShape->getMarginNonVirtual();
-		const btVector3& center = t.getOrigin();
-		btVector3 extent(margin,margin,margin);
-		aabbMin = center - extent;
-		aabbMax = center + extent;
-    }
-	break;
-	case CYLINDER_SHAPE_PROXYTYPE:
-	/* fall through */
-	case BOX_SHAPE_PROXYTYPE:
-	{
-		btBoxShape* convexShape = (btBoxShape*)this;
-		btScalar margin=convexShape->getMarginNonVirtual();
-		btVector3 halfExtents = convexShape->getImplicitShapeDimensions();
-		halfExtents += btVector3(margin,margin,margin);
-		btMatrix3x3 abs_b = t.getBasis().absolute();  
-		btVector3 center = t.getOrigin();
-        btVector3 extent = halfExtents.dot3(abs_b[0], abs_b[1], abs_b[2]);    
-        
-		aabbMin = center - extent;
-		aabbMax = center + extent;
-		break;
-	}
-	case TRIANGLE_SHAPE_PROXYTYPE:
-	{
-		btTriangleShape* triangleShape = (btTriangleShape*)this;
-		btScalar margin = triangleShape->getMarginNonVirtual();
-		for (int i=0;i<3;i++)
+		case SPHERE_SHAPE_PROXYTYPE:
 		{
-			btVector3 vec(btScalar(0.),btScalar(0.),btScalar(0.));
-			vec[i] = btScalar(1.);
+			btSphereShape* sphereShape = (btSphereShape*)this;
+			btScalar radius = sphereShape->getImplicitShapeDimensions().getX();  // * convexShape->getLocalScaling().getX();
+			btScalar margin = radius + sphereShape->getMarginNonVirtual();
+			const btVector3& center = t.getOrigin();
+			btVector3 extent(margin, margin, margin);
+			aabbMin = center - extent;
+			aabbMax = center + extent;
+		}
+		break;
+		case CYLINDER_SHAPE_PROXYTYPE:
+		/* fall through */
+		case BOX_SHAPE_PROXYTYPE:
+		{
+			btBoxShape* convexShape = (btBoxShape*)this;
+			btScalar margin = convexShape->getMarginNonVirtual();
+			btVector3 halfExtents = convexShape->getImplicitShapeDimensions();
+			halfExtents += btVector3(margin, margin, margin);
+			btMatrix3x3 abs_b = t.getBasis().absolute();
+			btVector3 center = t.getOrigin();
+			btVector3 extent = halfExtents.dot3(abs_b[0], abs_b[1], abs_b[2]);
 
-			btVector3 sv = localGetSupportVertexWithoutMarginNonVirtual(vec*t.getBasis());
+			aabbMin = center - extent;
+			aabbMax = center + extent;
+			break;
+		}
+		case TRIANGLE_SHAPE_PROXYTYPE:
+		{
+			btTriangleShape* triangleShape = (btTriangleShape*)this;
+			btScalar margin = triangleShape->getMarginNonVirtual();
+			for (int i = 0; i < 3; i++)
+			{
+				btVector3 vec(btScalar(0.), btScalar(0.), btScalar(0.));
+				vec[i] = btScalar(1.);
 
-			btVector3 tmp = t(sv);
-			aabbMax[i] = tmp[i]+margin;
-			vec[i] = btScalar(-1.);
-			tmp = t(localGetSupportVertexWithoutMarginNonVirtual(vec*t.getBasis()));
-			aabbMin[i] = tmp[i]-margin;
-		}	
-	}
-	break;
-	case CAPSULE_SHAPE_PROXYTYPE:
-	{
-		btCapsuleShape* capsuleShape = (btCapsuleShape*)this;
-		btVector3 halfExtents(capsuleShape->getRadius(),capsuleShape->getRadius(),capsuleShape->getRadius());
-		int m_upAxis = capsuleShape->getUpAxis();
-		halfExtents[m_upAxis] = capsuleShape->getRadius() + capsuleShape->getHalfHeight();
-		btMatrix3x3 abs_b = t.getBasis().absolute();  
-		btVector3 center = t.getOrigin();
-        btVector3 extent = halfExtents.dot3(abs_b[0], abs_b[1], abs_b[2]);    
-		aabbMin = center - extent;
-		aabbMax = center + extent;
-	}
-	break;
-	case CONVEX_POINT_CLOUD_SHAPE_PROXYTYPE:
-	case CONVEX_HULL_SHAPE_PROXYTYPE:
-	{
-		btPolyhedralConvexAabbCachingShape* convexHullShape = (btPolyhedralConvexAabbCachingShape*)this;
-		btScalar margin = convexHullShape->getMarginNonVirtual();
-		convexHullShape->getNonvirtualAabb (t, aabbMin, aabbMax, margin);
-	}
-	break;
-    default:
+				btVector3 sv = localGetSupportVertexWithoutMarginNonVirtual(vec * t.getBasis());
+
+				btVector3 tmp = t(sv);
+				aabbMax[i] = tmp[i] + margin;
+				vec[i] = btScalar(-1.);
+				tmp = t(localGetSupportVertexWithoutMarginNonVirtual(vec * t.getBasis()));
+				aabbMin[i] = tmp[i] - margin;
+			}
+		}
+		break;
+		case CAPSULE_SHAPE_PROXYTYPE:
+		{
+			btCapsuleShape* capsuleShape = (btCapsuleShape*)this;
+			btVector3 halfExtents(capsuleShape->getRadius(), capsuleShape->getRadius(), capsuleShape->getRadius());
+			int m_upAxis = capsuleShape->getUpAxis();
+			halfExtents[m_upAxis] = capsuleShape->getRadius() + capsuleShape->getHalfHeight();
+			btMatrix3x3 abs_b = t.getBasis().absolute();
+			btVector3 center = t.getOrigin();
+			btVector3 extent = halfExtents.dot3(abs_b[0], abs_b[1], abs_b[2]);
+			aabbMin = center - extent;
+			aabbMax = center + extent;
+		}
+		break;
+		case CONVEX_POINT_CLOUD_SHAPE_PROXYTYPE:
+		case CONVEX_HULL_SHAPE_PROXYTYPE:
+		{
+			btPolyhedralConvexAabbCachingShape* convexHullShape = (btPolyhedralConvexAabbCachingShape*)this;
+			btScalar margin = convexHullShape->getMarginNonVirtual();
+			convexHullShape->getNonvirtualAabb(t, aabbMin, aabbMax, margin);
+		}
+		break;
+		default:
 #ifndef __SPU__
-		this->getAabb (t, aabbMin, aabbMax);
+			this->getAabb(t, aabbMin, aabbMax);
 #else
-		btAssert (0);
+			btAssert(0);
 #endif
-	break;
+			break;
 	}
 
 	// should never reach here
-	btAssert (0);
+	btAssert(0);
 }
 
-#endif //__SPU__
+#endif  //__SPU__
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexShape.h
index 875f2ac195a..d3b3ed816ec 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexShape.h
@@ -28,58 +28,48 @@ subject to the following restrictions:
 
 /// The btConvexShape is an abstract shape interface, implemented by all convex shapes such as btBoxShape, btConvexHullShape etc.
 /// It describes general convex shapes using the localGetSupportingVertex interface, used by collision detectors such as btGjkPairDetector.
-ATTRIBUTE_ALIGNED16(class) btConvexShape : public btCollisionShape
+ATTRIBUTE_ALIGNED16(class)
+btConvexShape : public btCollisionShape
 {
-
-
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-	btConvexShape ();
+	btConvexShape();
 
 	virtual ~btConvexShape();
 
-	virtual btVector3	localGetSupportingVertex(const btVector3& vec)const = 0;
+	virtual btVector3 localGetSupportingVertex(const btVector3& vec) const = 0;
 
-	////////
-	#ifndef __SPU__
-	virtual btVector3	localGetSupportingVertexWithoutMargin(const btVector3& vec) const=0;
-	#endif //#ifndef __SPU__
+////////
+#ifndef __SPU__
+	virtual btVector3 localGetSupportingVertexWithoutMargin(const btVector3& vec) const = 0;
+#endif  //#ifndef __SPU__
 
-	btVector3 localGetSupportVertexWithoutMarginNonVirtual (const btVector3& vec) const;
-	btVector3 localGetSupportVertexNonVirtual (const btVector3& vec) const;
-	btScalar getMarginNonVirtual () const;
-	void getAabbNonVirtual (const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
+	btVector3 localGetSupportVertexWithoutMarginNonVirtual(const btVector3& vec) const;
+	btVector3 localGetSupportVertexNonVirtual(const btVector3& vec) const;
+	btScalar getMarginNonVirtual() const;
+	void getAabbNonVirtual(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
 
+	virtual void project(const btTransform& trans, const btVector3& dir, btScalar& minProj, btScalar& maxProj, btVector3& witnesPtMin, btVector3& witnesPtMax) const;
 
-	virtual void project(const btTransform& trans, const btVector3& dir, btScalar& minProj, btScalar& maxProj, btVector3& witnesPtMin,btVector3& witnesPtMax) const;
-
-	
 	//notice that the vectors should be unit length
-	virtual void	batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const= 0;
+	virtual void batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const = 0;
 
 	///getAabb's default implementation is brute force, expected derived classes to implement a fast dedicated version
-	void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const =0;
+	void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const = 0;
 
-	virtual void getAabbSlow(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const =0;
+	virtual void getAabbSlow(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const = 0;
 
-	virtual void	setLocalScaling(const btVector3& scaling) =0;
-	virtual const btVector3& getLocalScaling() const =0;
+	virtual void setLocalScaling(const btVector3& scaling) = 0;
+	virtual const btVector3& getLocalScaling() const = 0;
 
-	virtual void	setMargin(btScalar margin)=0;
+	virtual void setMargin(btScalar margin) = 0;
 
-	virtual btScalar	getMargin() const=0;
+	virtual btScalar getMargin() const = 0;
 
-	virtual int		getNumPreferredPenetrationDirections() const=0;
-	
-	virtual void	getPreferredPenetrationDirection(int index, btVector3& penetrationVector) const=0;
+	virtual int getNumPreferredPenetrationDirections() const = 0;
 
-
-	
-	
+	virtual void getPreferredPenetrationDirection(int index, btVector3& penetrationVector) const = 0;
 };
 
-
-
-#endif //BT_CONVEX_SHAPE_INTERFACE1
+#endif  //BT_CONVEX_SHAPE_INTERFACE1
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexTriangleMeshShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexTriangleMeshShape.cpp
index 0f9ced554b8..f6987cc760c 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexTriangleMeshShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexTriangleMeshShape.cpp
@@ -19,42 +19,37 @@ subject to the following restrictions:
 #include "LinearMath/btQuaternion.h"
 #include "BulletCollision/CollisionShapes/btStridingMeshInterface.h"
 
-
-btConvexTriangleMeshShape ::btConvexTriangleMeshShape (btStridingMeshInterface* meshInterface, bool calcAabb)
-: btPolyhedralConvexAabbCachingShape(), m_stridingMesh(meshInterface)
+btConvexTriangleMeshShape ::btConvexTriangleMeshShape(btStridingMeshInterface* meshInterface, bool calcAabb)
+	: btPolyhedralConvexAabbCachingShape(), m_stridingMesh(meshInterface)
 {
 	m_shapeType = CONVEX_TRIANGLEMESH_SHAPE_PROXYTYPE;
-	if ( calcAabb )
+	if (calcAabb)
 		recalcLocalAabb();
 }
 
-
-
-
 ///It's not nice to have all this virtual function overhead, so perhaps we can also gather the points once
 ///but then we are duplicating
-class LocalSupportVertexCallback: public btInternalTriangleIndexCallback
+class LocalSupportVertexCallback : public btInternalTriangleIndexCallback
 {
-
 	btVector3 m_supportVertexLocal;
-public:
 
+public:
 	btScalar m_maxDot;
 	btVector3 m_supportVecLocal;
 
 	LocalSupportVertexCallback(const btVector3& supportVecLocal)
-		: m_supportVertexLocal(btScalar(0.),btScalar(0.),btScalar(0.)),
-		m_maxDot(btScalar(-BT_LARGE_FLOAT)),
-                m_supportVecLocal(supportVecLocal)
+		: m_supportVertexLocal(btScalar(0.), btScalar(0.), btScalar(0.)),
+		  m_maxDot(btScalar(-BT_LARGE_FLOAT)),
+		  m_supportVecLocal(supportVecLocal)
 	{
 	}
 
-	virtual void internalProcessTriangleIndex(btVector3* triangle,int partId,int  triangleIndex)
+	virtual void internalProcessTriangleIndex(btVector3* triangle, int partId, int triangleIndex)
 	{
 		(void)triangleIndex;
 		(void)partId;
 
-		for (int i=0;i<3;i++)
+		for (int i = 0; i < 3; i++)
 		{
 			btScalar dot = m_supportVecLocal.dot(triangle[i]);
 			if (dot > m_maxDot)
@@ -64,99 +59,82 @@ public:
 			}
 		}
 	}
-	
-	btVector3	GetSupportVertexLocal()
+
+	btVector3 GetSupportVertexLocal()
 	{
 		return m_supportVertexLocal;
 	}
-
 };
 
-
-
-
-
-btVector3	btConvexTriangleMeshShape::localGetSupportingVertexWithoutMargin(const btVector3& vec0)const
+btVector3 btConvexTriangleMeshShape::localGetSupportingVertexWithoutMargin(const btVector3& vec0) const
 {
-	btVector3 supVec(btScalar(0.),btScalar(0.),btScalar(0.));
+	btVector3 supVec(btScalar(0.), btScalar(0.), btScalar(0.));
 
 	btVector3 vec = vec0;
 	btScalar lenSqr = vec.length2();
 	if (lenSqr < btScalar(0.0001))
 	{
-		vec.setValue(1,0,0);
-	} else
+		vec.setValue(1, 0, 0);
+	}
+	else
 	{
-		btScalar rlen = btScalar(1.) / btSqrt(lenSqr );
+		btScalar rlen = btScalar(1.) / btSqrt(lenSqr);
 		vec *= rlen;
 	}
 
-	LocalSupportVertexCallback	supportCallback(vec);
-	btVector3 aabbMax(btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT));
-	m_stridingMesh->InternalProcessAllTriangles(&supportCallback,-aabbMax,aabbMax);
+	LocalSupportVertexCallback supportCallback(vec);
+	btVector3 aabbMax(btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT));
+	m_stridingMesh->InternalProcessAllTriangles(&supportCallback, -aabbMax, aabbMax);
 	supVec = supportCallback.GetSupportVertexLocal();
 
 	return supVec;
 }
 
-void	btConvexTriangleMeshShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const
+void btConvexTriangleMeshShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const
 {
 	//use 'w' component of supportVerticesOut?
 	{
-		for (int i=0;i<numVectors;i++)
+		for (int i = 0; i < numVectors; i++)
 		{
 			supportVerticesOut[i][3] = btScalar(-BT_LARGE_FLOAT);
 		}
 	}
-	
+
 	///@todo: could do the batch inside the callback!
 
-
-	for (int j=0;j<numVectors;j++)
+	for (int j = 0; j < numVectors; j++)
 	{
 		const btVector3& vec = vectors[j];
-		LocalSupportVertexCallback	supportCallback(vec);
-		btVector3 aabbMax(btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT));
-		m_stridingMesh->InternalProcessAllTriangles(&supportCallback,-aabbMax,aabbMax);
+		LocalSupportVertexCallback supportCallback(vec);
+		btVector3 aabbMax(btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT));
+		m_stridingMesh->InternalProcessAllTriangles(&supportCallback, -aabbMax, aabbMax);
 		supportVerticesOut[j] = supportCallback.GetSupportVertexLocal();
 	}
-	
 }
-	
 
-
-btVector3	btConvexTriangleMeshShape::localGetSupportingVertex(const btVector3& vec)const
+btVector3 btConvexTriangleMeshShape::localGetSupportingVertex(const btVector3& vec) const
 {
 	btVector3 supVertex = localGetSupportingVertexWithoutMargin(vec);
 
-	if ( getMargin()!=btScalar(0.) )
+	if (getMargin() != btScalar(0.))
 	{
 		btVector3 vecnorm = vec;
-		if (vecnorm .length2() < (SIMD_EPSILON*SIMD_EPSILON))
+		if (vecnorm.length2() < (SIMD_EPSILON * SIMD_EPSILON))
 		{
-			vecnorm.setValue(btScalar(-1.),btScalar(-1.),btScalar(-1.));
-		} 
+			vecnorm.setValue(btScalar(-1.), btScalar(-1.), btScalar(-1.));
+		}
 		vecnorm.normalize();
-		supVertex+= getMargin() * vecnorm;
+		supVertex += getMargin() * vecnorm;
 	}
 	return supVertex;
 }
 
-
-
-
-
-
-
-
-
 //currently just for debugging (drawing), perhaps future support for algebraic continuous collision detection
 //Please note that you can debug-draw btConvexTriangleMeshShape with the Raytracer Demo
-int	btConvexTriangleMeshShape::getNumVertices() const
+int btConvexTriangleMeshShape::getNumVertices() const
 {
 	//cache this?
 	return 0;
-	
 }
 
 int btConvexTriangleMeshShape::getNumEdges() const
@@ -164,43 +142,39 @@ int btConvexTriangleMeshShape::getNumEdges() const
 	return 0;
 }
 
-void btConvexTriangleMeshShape::getEdge(int ,btVector3& ,btVector3& ) const
-{
-	btAssert(0);	
-}
-
-void btConvexTriangleMeshShape::getVertex(int ,btVector3& ) const
+void btConvexTriangleMeshShape::getEdge(int, btVector3&, btVector3&) const
 {
 	btAssert(0);
 }
 
-int	btConvexTriangleMeshShape::getNumPlanes() const
+void btConvexTriangleMeshShape::getVertex(int, btVector3&) const
+{
+	btAssert(0);
+}
+
+int btConvexTriangleMeshShape::getNumPlanes() const
 {
 	return 0;
 }
 
-void btConvexTriangleMeshShape::getPlane(btVector3& ,btVector3& ,int  ) const
+void btConvexTriangleMeshShape::getPlane(btVector3&, btVector3&, int) const
 {
 	btAssert(0);
 }
 
 //not yet
-bool btConvexTriangleMeshShape::isInside(const btVector3& ,btScalar ) const
+bool btConvexTriangleMeshShape::isInside(const btVector3&, btScalar) const
 {
 	btAssert(0);
 	return false;
 }
 
-
-
-void	btConvexTriangleMeshShape::setLocalScaling(const btVector3& scaling)
+void btConvexTriangleMeshShape::setLocalScaling(const btVector3& scaling)
 {
 	m_stridingMesh->setScaling(scaling);
-	
-	recalcLocalAabb();
-	
-}
 
+	recalcLocalAabb();
+}
 
 const btVector3& btConvexTriangleMeshShape::getLocalScaling() const
 {
@@ -209,107 +183,101 @@ const btVector3& btConvexTriangleMeshShape::getLocalScaling() const
 
 void btConvexTriangleMeshShape::calculatePrincipalAxisTransform(btTransform& principal, btVector3& inertia, btScalar& volume) const
 {
-   class CenterCallback: public btInternalTriangleIndexCallback
-   {
-      bool first;
-      btVector3 ref;
-      btVector3 sum;
-      btScalar volume;
+	class CenterCallback : public btInternalTriangleIndexCallback
+	{
+		bool first;
+		btVector3 ref;
+		btVector3 sum;
+		btScalar volume;
 
-   public:
+	public:
+		CenterCallback() : first(true), ref(0, 0, 0), sum(0, 0, 0), volume(0)
+		{
+		}
 
-      CenterCallback() : first(true), ref(0, 0, 0), sum(0, 0, 0), volume(0)
-      {
-      }
+		virtual void internalProcessTriangleIndex(btVector3* triangle, int partId, int triangleIndex)
+		{
+			(void)triangleIndex;
+			(void)partId;
+			if (first)
+			{
+				ref = triangle[0];
+				first = false;
+			}
+			else
+			{
+				btScalar vol = btFabs((triangle[0] - ref).triple(triangle[1] - ref, triangle[2] - ref));
+				sum += (btScalar(0.25) * vol) * ((triangle[0] + triangle[1] + triangle[2] + ref));
+				volume += vol;
+			}
+		}
 
-      virtual void internalProcessTriangleIndex(btVector3* triangle, int partId, int triangleIndex)
-      {
-         (void) triangleIndex;
-         (void) partId;
-         if (first)
-         {
-            ref = triangle[0];
-            first = false;
-         }
-         else
-         {
-            btScalar vol = btFabs((triangle[0] - ref).triple(triangle[1] - ref, triangle[2] - ref));
-            sum += (btScalar(0.25) * vol) * ((triangle[0] + triangle[1] + triangle[2] + ref));
-            volume += vol;
-         }
-      }
-      
-      btVector3 getCenter()
-      {
-         return (volume > 0) ? sum / volume : ref;
-      }
+		btVector3 getCenter()
+		{
+			return (volume > 0) ? sum / volume : ref;
+		}
 
-      btScalar getVolume()
-      {
-         return volume * btScalar(1. / 6);
-      }
+		btScalar getVolume()
+		{
+			return volume * btScalar(1. / 6);
+		}
+	};
 
-   };
+	class InertiaCallback : public btInternalTriangleIndexCallback
+	{
+		btMatrix3x3 sum;
+		btVector3 center;
 
-   class InertiaCallback: public btInternalTriangleIndexCallback
-   {
-      btMatrix3x3 sum;
-      btVector3 center;
+	public:
+		InertiaCallback(btVector3& center) : sum(0, 0, 0, 0, 0, 0, 0, 0, 0), center(center)
+		{
+		}
 
-   public:
+		virtual void internalProcessTriangleIndex(btVector3* triangle, int partId, int triangleIndex)
+		{
+			(void)triangleIndex;
+			(void)partId;
+			btMatrix3x3 i;
+			btVector3 a = triangle[0] - center;
+			btVector3 b = triangle[1] - center;
+			btVector3 c = triangle[2] - center;
+			btScalar volNeg = -btFabs(a.triple(b, c)) * btScalar(1. / 6);
+			for (int j = 0; j < 3; j++)
+			{
+				for (int k = 0; k <= j; k++)
+				{
+					i[j][k] = i[k][j] = volNeg * (btScalar(0.1) * (a[j] * a[k] + b[j] * b[k] + c[j] * c[k]) + btScalar(0.05) * (a[j] * b[k] + a[k] * b[j] + a[j] * c[k] + a[k] * c[j] + b[j] * c[k] + b[k] * c[j]));
+				}
+			}
+			btScalar i00 = -i[0][0];
+			btScalar i11 = -i[1][1];
+			btScalar i22 = -i[2][2];
+			i[0][0] = i11 + i22;
+			i[1][1] = i22 + i00;
+			i[2][2] = i00 + i11;
+			sum[0] += i[0];
+			sum[1] += i[1];
+			sum[2] += i[2];
+		}
 
-      InertiaCallback(btVector3& center) : sum(0, 0, 0, 0, 0, 0, 0, 0, 0), center(center)
-      {
-      }
+		btMatrix3x3& getInertia()
+		{
+			return sum;
+		}
+	};
 
-      virtual void internalProcessTriangleIndex(btVector3* triangle, int partId, int triangleIndex)
-      {
-         (void) triangleIndex;
-         (void) partId;
-         btMatrix3x3 i;
-         btVector3 a = triangle[0] - center;
-         btVector3 b = triangle[1] - center;
-         btVector3 c = triangle[2] - center;
-         btScalar volNeg = -btFabs(a.triple(b, c)) * btScalar(1. / 6);
-         for (int j = 0; j < 3; j++)
-         {
-            for (int k = 0; k <= j; k++)
-            {
-               i[j][k] = i[k][j] = volNeg * (btScalar(0.1) * (a[j] * a[k] + b[j] * b[k] + c[j] * c[k])
-                  + btScalar(0.05) * (a[j] * b[k] + a[k] * b[j] + a[j] * c[k] + a[k] * c[j] + b[j] * c[k] + b[k] * c[j]));
-            }
-         }
-         btScalar i00 = -i[0][0];
-         btScalar i11 = -i[1][1];
-         btScalar i22 = -i[2][2];
-         i[0][0] = i11 + i22; 
-         i[1][1] = i22 + i00; 
-         i[2][2] = i00 + i11;
-         sum[0] += i[0];
-         sum[1] += i[1];
-         sum[2] += i[2];
-      }
-      
-      btMatrix3x3& getInertia()
-      {
-         return sum;
-      }
+	CenterCallback centerCallback;
+	btVector3 aabbMax(btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT));
+	m_stridingMesh->InternalProcessAllTriangles(&centerCallback, -aabbMax, aabbMax);
+	btVector3 center = centerCallback.getCenter();
+	principal.setOrigin(center);
+	volume = centerCallback.getVolume();
 
-   };
+	InertiaCallback inertiaCallback(center);
+	m_stridingMesh->InternalProcessAllTriangles(&inertiaCallback, -aabbMax, aabbMax);
 
-   CenterCallback centerCallback;
-   btVector3 aabbMax(btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT));
-   m_stridingMesh->InternalProcessAllTriangles(&centerCallback, -aabbMax, aabbMax);
-   btVector3 center = centerCallback.getCenter();
-   principal.setOrigin(center);
-   volume = centerCallback.getVolume();
-
-   InertiaCallback inertiaCallback(center);
-   m_stridingMesh->InternalProcessAllTriangles(&inertiaCallback, -aabbMax, aabbMax);
-
-   btMatrix3x3& i = inertiaCallback.getInertia();
-   i.diagonalize(principal.getBasis(), btScalar(0.00001), 20);
-   inertia.setValue(i[0][0], i[1][1], i[2][2]);
-   inertia /= volume;
+	btMatrix3x3& i = inertiaCallback.getInertia();
+	i.diagonalize(principal.getBasis(), btScalar(0.00001), 20);
+	inertia.setValue(i[0][0], i[1][1], i[2][2]);
+	inertia /= volume;
 }
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexTriangleMeshShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexTriangleMeshShape.h
index f338865ca13..6dac9fff040 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexTriangleMeshShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btConvexTriangleMeshShape.h
@@ -15,24 +15,22 @@ subject to the following restrictions:
 #ifndef BT_CONVEX_TRIANGLEMESH_SHAPE_H
 #define BT_CONVEX_TRIANGLEMESH_SHAPE_H
 
-
 #include "btPolyhedralConvexShape.h"
-#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h" // for the types
-
+#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h"  // for the types
 
 /// The btConvexTriangleMeshShape is a convex hull of a triangle mesh, but the performance is not as good as btConvexHullShape.
 /// A small benefit of this class is that it uses the btStridingMeshInterface, so you can avoid the duplication of the triangle mesh data. Nevertheless, most users should use the much better performing btConvexHullShape instead.
-ATTRIBUTE_ALIGNED16(class) btConvexTriangleMeshShape : public btPolyhedralConvexAabbCachingShape
+ATTRIBUTE_ALIGNED16(class)
+btConvexTriangleMeshShape : public btPolyhedralConvexAabbCachingShape
 {
-
-	class btStridingMeshInterface*	m_stridingMesh;
+	class btStridingMeshInterface* m_stridingMesh;
 
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
-	btConvexTriangleMeshShape(btStridingMeshInterface* meshInterface, bool calcAabb = true);
 
-	class btStridingMeshInterface*	getMeshInterface()
+	btConvexTriangleMeshShape(btStridingMeshInterface * meshInterface, bool calcAabb = true);
+
+	class btStridingMeshInterface* getMeshInterface()
 	{
 		return m_stridingMesh;
 	}
@@ -40,24 +38,23 @@ public:
 	{
 		return m_stridingMesh;
 	}
-	
-	virtual btVector3	localGetSupportingVertex(const btVector3& vec)const;
-	virtual btVector3	localGetSupportingVertexWithoutMargin(const btVector3& vec)const;
-	virtual void	batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const;
-	
-	//debugging
-	virtual const char*	getName()const {return "ConvexTrimesh";}
-	
-	virtual int	getNumVertices() const;
-	virtual int getNumEdges() const;
-	virtual void getEdge(int i,btVector3& pa,btVector3& pb) const;
-	virtual void getVertex(int i,btVector3& vtx) const;
-	virtual int	getNumPlanes() const;
-	virtual void getPlane(btVector3& planeNormal,btVector3& planeSupport,int i ) const;
-	virtual	bool isInside(const btVector3& pt,btScalar tolerance) const;
 
-	
-	virtual void	setLocalScaling(const btVector3& scaling);
+	virtual btVector3 localGetSupportingVertex(const btVector3& vec) const;
+	virtual btVector3 localGetSupportingVertexWithoutMargin(const btVector3& vec) const;
+	virtual void batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const;
+
+	//debugging
+	virtual const char* getName() const { return "ConvexTrimesh"; }
+
+	virtual int getNumVertices() const;
+	virtual int getNumEdges() const;
+	virtual void getEdge(int i, btVector3& pa, btVector3& pb) const;
+	virtual void getVertex(int i, btVector3& vtx) const;
+	virtual int getNumPlanes() const;
+	virtual void getPlane(btVector3 & planeNormal, btVector3 & planeSupport, int i) const;
+	virtual bool isInside(const btVector3& pt, btScalar tolerance) const;
+
+	virtual void setLocalScaling(const btVector3& scaling);
 	virtual const btVector3& getLocalScaling() const;
 
 	///computes the exact moment of inertia and the transform from the coordinate system defined by the principal axes of the moment of inertia
@@ -65,13 +62,7 @@ public:
 	///by the mass. The resulting transform "principal" has to be applied inversely to the mesh in order for the local coordinate system of the
 	///shape to be centered at the center of mass and to coincide with the principal axes. This also necessitates a correction of the world transform
 	///of the collision object by the principal transform. This method also computes the volume of the convex mesh.
-	void calculatePrincipalAxisTransform(btTransform& principal, btVector3& inertia, btScalar& volume) const;
-
+	void calculatePrincipalAxisTransform(btTransform & principal, btVector3 & inertia, btScalar & volume) const;
 };
 
-
-
-#endif //BT_CONVEX_TRIANGLEMESH_SHAPE_H
-
-
-
+#endif  //BT_CONVEX_TRIANGLEMESH_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btCylinderShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btCylinderShape.cpp
index 604b3fc7708..66dbb8e53de 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btCylinderShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btCylinderShape.cpp
@@ -15,11 +15,11 @@ subject to the following restrictions:
 
 #include "btCylinderShape.h"
 
-btCylinderShape::btCylinderShape (const btVector3& halfExtents)
-:btConvexInternalShape(),
-m_upAxis(1)
+btCylinderShape::btCylinderShape(const btVector3& halfExtents)
+	: btConvexInternalShape(),
+	  m_upAxis(1)
 {
-	btVector3 margin(getMargin(),getMargin(),getMargin());
+	btVector3 margin(getMargin(), getMargin(), getMargin());
 	m_implicitShapeDimensions = (halfExtents * m_localScaling) - margin;
 
 	setSafeMargin(halfExtents);
@@ -27,30 +27,25 @@ m_upAxis(1)
 	m_shapeType = CYLINDER_SHAPE_PROXYTYPE;
 }
 
-
-btCylinderShapeX::btCylinderShapeX (const btVector3& halfExtents)
-:btCylinderShape(halfExtents)
+btCylinderShapeX::btCylinderShapeX(const btVector3& halfExtents)
+	: btCylinderShape(halfExtents)
 {
 	m_upAxis = 0;
-
 }
 
-
-btCylinderShapeZ::btCylinderShapeZ (const btVector3& halfExtents)
-:btCylinderShape(halfExtents)
+btCylinderShapeZ::btCylinderShapeZ(const btVector3& halfExtents)
+	: btCylinderShape(halfExtents)
 {
 	m_upAxis = 2;
-
 }
 
-void btCylinderShape::getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const
+void btCylinderShape::getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
 {
-	btTransformAabb(getHalfExtentsWithoutMargin(),getMargin(),t,aabbMin,aabbMax);
+	btTransformAabb(getHalfExtentsWithoutMargin(), getMargin(), t, aabbMin, aabbMax);
 }
 
-void	btCylinderShape::calculateLocalInertia(btScalar mass,btVector3& inertia) const
+void btCylinderShape::calculateLocalInertia(btScalar mass, btVector3& inertia) const
 {
-
 //Until Bullet 2.77 a box approximation was used, so uncomment this if you need backwards compatibility
 //#define USE_BOX_INERTIA_APPROXIMATION 1
 #ifndef USE_BOX_INERTIA_APPROXIMATION
@@ -64,25 +59,25 @@ void	btCylinderShape::calculateLocalInertia(btScalar mass,btVector3& inertia) co
 	*
 	*/
 
-	btScalar radius2;	// square of cylinder radius
-	btScalar height2;	// square of cylinder height
-	btVector3 halfExtents = getHalfExtentsWithMargin();	// get cylinder dimension
+	btScalar radius2;                                    // square of cylinder radius
+	btScalar height2;                                    // square of cylinder height
+	btVector3 halfExtents = getHalfExtentsWithMargin();  // get cylinder dimension
 	btScalar div12 = mass / 12.f;
 	btScalar div4 = mass / 4.f;
 	btScalar div2 = mass / 2.f;
 	int idxRadius, idxHeight;
 
-	switch (m_upAxis)	// get indices of radius and height of cylinder
+	switch (m_upAxis)  // get indices of radius and height of cylinder
 	{
-		case 0:		// cylinder is aligned along x
+		case 0:  // cylinder is aligned along x
 			idxRadius = 1;
 			idxHeight = 0;
 			break;
-		case 2:		// cylinder is aligned along z
+		case 2:  // cylinder is aligned along z
 			idxRadius = 0;
 			idxHeight = 2;
 			break;
-		default:	// cylinder is aligned along y
+		default:  // cylinder is aligned along y
 			idxRadius = 0;
 			idxHeight = 1;
 	}
@@ -95,188 +90,164 @@ void	btCylinderShape::calculateLocalInertia(btScalar mass,btVector3& inertia) co
 	btScalar t1 = div12 * height2 + div4 * radius2;
 	btScalar t2 = div2 * radius2;
 
-	switch (m_upAxis)	// set diagonal elements of inertia tensor
+	switch (m_upAxis)  // set diagonal elements of inertia tensor
 	{
-		case 0:		// cylinder is aligned along x
-			inertia.setValue(t2,t1,t1);
+		case 0:  // cylinder is aligned along x
+			inertia.setValue(t2, t1, t1);
 			break;
-		case 2:		// cylinder is aligned along z
-			inertia.setValue(t1,t1,t2);
+		case 2:  // cylinder is aligned along z
+			inertia.setValue(t1, t1, t2);
 			break;
-		default:	// cylinder is aligned along y
-			inertia.setValue(t1,t2,t1);
+		default:  // cylinder is aligned along y
+			inertia.setValue(t1, t2, t1);
 	}
-#else //USE_BOX_INERTIA_APPROXIMATION
+#else   //USE_BOX_INERTIA_APPROXIMATION
 	//approximation of box shape
 	btVector3 halfExtents = getHalfExtentsWithMargin();
 
-	btScalar lx=btScalar(2.)*(halfExtents.x());
-	btScalar ly=btScalar(2.)*(halfExtents.y());
-	btScalar lz=btScalar(2.)*(halfExtents.z());
+	btScalar lx = btScalar(2.) * (halfExtents.x());
+	btScalar ly = btScalar(2.) * (halfExtents.y());
+	btScalar lz = btScalar(2.) * (halfExtents.z());
 
-	inertia.setValue(mass/(btScalar(12.0)) * (ly*ly + lz*lz),
-					mass/(btScalar(12.0)) * (lx*lx + lz*lz),
-					mass/(btScalar(12.0)) * (lx*lx + ly*ly));
-#endif //USE_BOX_INERTIA_APPROXIMATION
+	inertia.setValue(mass / (btScalar(12.0)) * (ly * ly + lz * lz),
+					 mass / (btScalar(12.0)) * (lx * lx + lz * lz),
+					 mass / (btScalar(12.0)) * (lx * lx + ly * ly));
+#endif  //USE_BOX_INERTIA_APPROXIMATION
 }
 
-
-SIMD_FORCE_INLINE btVector3 CylinderLocalSupportX(const btVector3& halfExtents,const btVector3& v) 
+SIMD_FORCE_INLINE btVector3 CylinderLocalSupportX(const btVector3& halfExtents, const btVector3& v)
 {
-const int cylinderUpAxis = 0;
-const int XX = 1;
-const int YY = 0;
-const int ZZ = 2;
+	const int cylinderUpAxis = 0;
+	const int XX = 1;
+	const int YY = 0;
+	const int ZZ = 2;
 
 	//mapping depends on how cylinder local orientation is
 	// extents of the cylinder is: X,Y is for radius, and Z for height
 
-
 	btScalar radius = halfExtents[XX];
 	btScalar halfHeight = halfExtents[cylinderUpAxis];
 
+	btVector3 tmp;
+	btScalar d;
 
-    btVector3 tmp;
-	btScalar d ;
-
-    btScalar s = btSqrt(v[XX] * v[XX] + v[ZZ] * v[ZZ]);
-    if (s != btScalar(0.0))
+	btScalar s = btSqrt(v[XX] * v[XX] + v[ZZ] * v[ZZ]);
+	if (s != btScalar(0.0))
 	{
-        d = radius / s;  
+		d = radius / s;
 		tmp[XX] = v[XX] * d;
 		tmp[YY] = v[YY] < 0.0 ? -halfHeight : halfHeight;
 		tmp[ZZ] = v[ZZ] * d;
 		return tmp;
 	}
-    else
+	else
 	{
-	    tmp[XX] = radius;
+		tmp[XX] = radius;
 		tmp[YY] = v[YY] < 0.0 ? -halfHeight : halfHeight;
 		tmp[ZZ] = btScalar(0.0);
 		return tmp;
-    }
-
-
+	}
 }
 
-
-
-
-
-
-inline  btVector3 CylinderLocalSupportY(const btVector3& halfExtents,const btVector3& v) 
+inline btVector3 CylinderLocalSupportY(const btVector3& halfExtents, const btVector3& v)
 {
-
-const int cylinderUpAxis = 1;
-const int XX = 0;
-const int YY = 1;
-const int ZZ = 2;
-
+	const int cylinderUpAxis = 1;
+	const int XX = 0;
+	const int YY = 1;
+	const int ZZ = 2;
 
 	btScalar radius = halfExtents[XX];
 	btScalar halfHeight = halfExtents[cylinderUpAxis];
 
+	btVector3 tmp;
+	btScalar d;
 
-    btVector3 tmp;
-	btScalar d ;
-
-    btScalar s = btSqrt(v[XX] * v[XX] + v[ZZ] * v[ZZ]);
-    if (s != btScalar(0.0))
+	btScalar s = btSqrt(v[XX] * v[XX] + v[ZZ] * v[ZZ]);
+	if (s != btScalar(0.0))
 	{
-        d = radius / s;  
+		d = radius / s;
 		tmp[XX] = v[XX] * d;
 		tmp[YY] = v[YY] < 0.0 ? -halfHeight : halfHeight;
 		tmp[ZZ] = v[ZZ] * d;
 		return tmp;
 	}
-    else
+	else
 	{
-	    tmp[XX] = radius;
+		tmp[XX] = radius;
 		tmp[YY] = v[YY] < 0.0 ? -halfHeight : halfHeight;
 		tmp[ZZ] = btScalar(0.0);
 		return tmp;
-    }
-
+	}
 }
 
-inline btVector3 CylinderLocalSupportZ(const btVector3& halfExtents,const btVector3& v) 
+inline btVector3 CylinderLocalSupportZ(const btVector3& halfExtents, const btVector3& v)
 {
-const int cylinderUpAxis = 2;
-const int XX = 0;
-const int YY = 2;
-const int ZZ = 1;
+	const int cylinderUpAxis = 2;
+	const int XX = 0;
+	const int YY = 2;
+	const int ZZ = 1;
 
 	//mapping depends on how cylinder local orientation is
 	// extents of the cylinder is: X,Y is for radius, and Z for height
 
-
 	btScalar radius = halfExtents[XX];
 	btScalar halfHeight = halfExtents[cylinderUpAxis];
 
+	btVector3 tmp;
+	btScalar d;
 
-    btVector3 tmp;
-	btScalar d ;
-
-    btScalar s = btSqrt(v[XX] * v[XX] + v[ZZ] * v[ZZ]);
-    if (s != btScalar(0.0))
+	btScalar s = btSqrt(v[XX] * v[XX] + v[ZZ] * v[ZZ]);
+	if (s != btScalar(0.0))
 	{
-        d = radius / s;  
+		d = radius / s;
 		tmp[XX] = v[XX] * d;
 		tmp[YY] = v[YY] < 0.0 ? -halfHeight : halfHeight;
 		tmp[ZZ] = v[ZZ] * d;
 		return tmp;
 	}
-    else
+	else
 	{
-	    tmp[XX] = radius;
+		tmp[XX] = radius;
 		tmp[YY] = v[YY] < 0.0 ? -halfHeight : halfHeight;
 		tmp[ZZ] = btScalar(0.0);
 		return tmp;
-    }
-
-
-}
-
-btVector3	btCylinderShapeX::localGetSupportingVertexWithoutMargin(const btVector3& vec)const
-{
-	return CylinderLocalSupportX(getHalfExtentsWithoutMargin(),vec);
-}
-
-
-btVector3	btCylinderShapeZ::localGetSupportingVertexWithoutMargin(const btVector3& vec)const
-{
-	return CylinderLocalSupportZ(getHalfExtentsWithoutMargin(),vec);
-}
-btVector3	btCylinderShape::localGetSupportingVertexWithoutMargin(const btVector3& vec)const
-{
-	return CylinderLocalSupportY(getHalfExtentsWithoutMargin(),vec);
-}
-
-void	btCylinderShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const
-{
-	for (int i=0;i<numVectors;i++)
-	{
-		supportVerticesOut[i] = CylinderLocalSupportY(getHalfExtentsWithoutMargin(),vectors[i]);
 	}
 }
 
-void	btCylinderShapeZ::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const
+btVector3 btCylinderShapeX::localGetSupportingVertexWithoutMargin(const btVector3& vec) const
 {
-	for (int i=0;i<numVectors;i++)
+	return CylinderLocalSupportX(getHalfExtentsWithoutMargin(), vec);
+}
+
+btVector3 btCylinderShapeZ::localGetSupportingVertexWithoutMargin(const btVector3& vec) const
+{
+	return CylinderLocalSupportZ(getHalfExtentsWithoutMargin(), vec);
+}
+btVector3 btCylinderShape::localGetSupportingVertexWithoutMargin(const btVector3& vec) const
+{
+	return CylinderLocalSupportY(getHalfExtentsWithoutMargin(), vec);
+}
+
+void btCylinderShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const
+{
+	for (int i = 0; i < numVectors; i++)
 	{
-		supportVerticesOut[i] = CylinderLocalSupportZ(getHalfExtentsWithoutMargin(),vectors[i]);
+		supportVerticesOut[i] = CylinderLocalSupportY(getHalfExtentsWithoutMargin(), vectors[i]);
 	}
 }
 
-
-
-
-void	btCylinderShapeX::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const
+void btCylinderShapeZ::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const
 {
-	for (int i=0;i<numVectors;i++)
+	for (int i = 0; i < numVectors; i++)
 	{
-		supportVerticesOut[i] = CylinderLocalSupportX(getHalfExtentsWithoutMargin(),vectors[i]);
+		supportVerticesOut[i] = CylinderLocalSupportZ(getHalfExtentsWithoutMargin(), vectors[i]);
 	}
 }
 
-
+void btCylinderShapeX::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const
+{
+	for (int i = 0; i < numVectors; i++)
+	{
+		supportVerticesOut[i] = CylinderLocalSupportX(getHalfExtentsWithoutMargin(), vectors[i]);
+	}
+}
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btCylinderShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btCylinderShape.h
index a214a827c9d..d3f64508bae 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btCylinderShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btCylinderShape.h
@@ -17,90 +17,84 @@ subject to the following restrictions:
 #define BT_CYLINDER_MINKOWSKI_H
 
 #include "btBoxShape.h"
-#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h" // for the types
+#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h"  // for the types
 #include "LinearMath/btVector3.h"
 
 /// The btCylinderShape class implements a cylinder shape primitive, centered around the origin. Its central axis aligned with the Y axis. btCylinderShapeX is aligned with the X axis and btCylinderShapeZ around the Z axis.
-ATTRIBUTE_ALIGNED16(class) btCylinderShape : public btConvexInternalShape
+ATTRIBUTE_ALIGNED16(class)
+btCylinderShape : public btConvexInternalShape
 
 {
-
 protected:
-
-	int	m_upAxis;
+	int m_upAxis;
 
 public:
-
-BT_DECLARE_ALIGNED_ALLOCATOR();
+	BT_DECLARE_ALIGNED_ALLOCATOR();
 
 	btVector3 getHalfExtentsWithMargin() const
 	{
 		btVector3 halfExtents = getHalfExtentsWithoutMargin();
-		btVector3 margin(getMargin(),getMargin(),getMargin());
+		btVector3 margin(getMargin(), getMargin(), getMargin());
 		halfExtents += margin;
 		return halfExtents;
 	}
-	
+
 	const btVector3& getHalfExtentsWithoutMargin() const
 	{
-		return m_implicitShapeDimensions;//changed in Bullet 2.63: assume the scaling and margin are included
+		return m_implicitShapeDimensions;  //changed in Bullet 2.63: assume the scaling and margin are included
 	}
 
-	btCylinderShape (const btVector3& halfExtents);
-	
-	void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const;
+	btCylinderShape(const btVector3& halfExtents);
 
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
+	void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
 
-	virtual btVector3	localGetSupportingVertexWithoutMargin(const btVector3& vec)const;
+	virtual void calculateLocalInertia(btScalar mass, btVector3 & inertia) const;
 
-	virtual void	batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const;
+	virtual btVector3 localGetSupportingVertexWithoutMargin(const btVector3& vec) const;
+
+	virtual void batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const;
 
 	virtual void setMargin(btScalar collisionMargin)
 	{
 		//correct the m_implicitShapeDimensions for the margin
-		btVector3 oldMargin(getMargin(),getMargin(),getMargin());
-		btVector3 implicitShapeDimensionsWithMargin = m_implicitShapeDimensions+oldMargin;
-		
-		btConvexInternalShape::setMargin(collisionMargin);
-		btVector3 newMargin(getMargin(),getMargin(),getMargin());
-		m_implicitShapeDimensions = implicitShapeDimensionsWithMargin - newMargin;
+		btVector3 oldMargin(getMargin(), getMargin(), getMargin());
+		btVector3 implicitShapeDimensionsWithMargin = m_implicitShapeDimensions + oldMargin;
 
+		btConvexInternalShape::setMargin(collisionMargin);
+		btVector3 newMargin(getMargin(), getMargin(), getMargin());
+		m_implicitShapeDimensions = implicitShapeDimensionsWithMargin - newMargin;
 	}
 
-	virtual btVector3	localGetSupportingVertex(const btVector3& vec) const
+	virtual btVector3 localGetSupportingVertex(const btVector3& vec) const
 	{
-
 		btVector3 supVertex;
 		supVertex = localGetSupportingVertexWithoutMargin(vec);
-		
-		if ( getMargin()!=btScalar(0.) )
+
+		if (getMargin() != btScalar(0.))
 		{
 			btVector3 vecnorm = vec;
-			if (vecnorm .length2() < (SIMD_EPSILON*SIMD_EPSILON))
+			if (vecnorm.length2() < (SIMD_EPSILON * SIMD_EPSILON))
 			{
-				vecnorm.setValue(btScalar(-1.),btScalar(-1.),btScalar(-1.));
-			} 
+				vecnorm.setValue(btScalar(-1.), btScalar(-1.), btScalar(-1.));
+			}
 			vecnorm.normalize();
-			supVertex+= getMargin() * vecnorm;
+			supVertex += getMargin() * vecnorm;
 		}
 		return supVertex;
 	}
 
-
 	//use box inertia
 	//	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
 
-
-	int	getUpAxis() const
+	int getUpAxis() const
 	{
 		return m_upAxis;
 	}
 
-	virtual btVector3	getAnisotropicRollingFrictionDirection() const
+	virtual btVector3 getAnisotropicRollingFrictionDirection() const
 	{
-		btVector3 aniDir(0,0,0);
-		aniDir[getUpAxis()]=1;
+		btVector3 aniDir(0, 0, 0);
+		aniDir[getUpAxis()] = 1;
 		return aniDir;
 	}
 
@@ -109,43 +103,41 @@ BT_DECLARE_ALIGNED_ALLOCATOR();
 		return getHalfExtentsWithMargin().getX();
 	}
 
-	virtual void	setLocalScaling(const btVector3& scaling)
+	virtual void setLocalScaling(const btVector3& scaling)
 	{
-		btVector3 oldMargin(getMargin(),getMargin(),getMargin());
-		btVector3 implicitShapeDimensionsWithMargin = m_implicitShapeDimensions+oldMargin;
+		btVector3 oldMargin(getMargin(), getMargin(), getMargin());
+		btVector3 implicitShapeDimensionsWithMargin = m_implicitShapeDimensions + oldMargin;
 		btVector3 unScaledImplicitShapeDimensionsWithMargin = implicitShapeDimensionsWithMargin / m_localScaling;
 
 		btConvexInternalShape::setLocalScaling(scaling);
 
 		m_implicitShapeDimensions = (unScaledImplicitShapeDimensionsWithMargin * m_localScaling) - oldMargin;
-
 	}
 
 	//debugging
-	virtual const char*	getName()const
+	virtual const char* getName() const
 	{
 		return "CylinderY";
 	}
 
-	virtual	int	calculateSerializeBufferSize() const;
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
-
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 };
 
 class btCylinderShapeX : public btCylinderShape
 {
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
-	btCylinderShapeX (const btVector3& halfExtents);
 
-	virtual btVector3	localGetSupportingVertexWithoutMargin(const btVector3& vec)const;
-	virtual void	batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const;
-	
-		//debugging
-	virtual const char*	getName()const
+	btCylinderShapeX(const btVector3& halfExtents);
+
+	virtual btVector3 localGetSupportingVertexWithoutMargin(const btVector3& vec) const;
+	virtual void batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const;
+
+	//debugging
+	virtual const char* getName() const
 	{
 		return "CylinderX";
 	}
@@ -154,21 +146,20 @@ public:
 	{
 		return getHalfExtentsWithMargin().getY();
 	}
-
 };
 
 class btCylinderShapeZ : public btCylinderShape
 {
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
-	btCylinderShapeZ (const btVector3& halfExtents);
 
-	virtual btVector3	localGetSupportingVertexWithoutMargin(const btVector3& vec)const;
-	virtual void	batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const;
+	btCylinderShapeZ(const btVector3& halfExtents);
 
-		//debugging
-	virtual const char*	getName()const
+	virtual btVector3 localGetSupportingVertexWithoutMargin(const btVector3& vec) const;
+	virtual void batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const;
+
+	//debugging
+	virtual const char* getName() const
 	{
 		return "CylinderZ";
 	}
@@ -177,30 +168,29 @@ public:
 	{
 		return getHalfExtentsWithMargin().getX();
 	}
-
 };
 
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
-struct	btCylinderShapeData
+struct btCylinderShapeData
 {
-	btConvexInternalShapeData	m_convexInternalShapeData;
+	btConvexInternalShapeData m_convexInternalShapeData;
 
-	int	m_upAxis;
+	int m_upAxis;
 
-	char	m_padding[4];
+	char m_padding[4];
 };
 
-SIMD_FORCE_INLINE	int	btCylinderShape::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btCylinderShape::calculateSerializeBufferSize() const
 {
 	return sizeof(btCylinderShapeData);
 }
 
-	///fills the dataBuffer and returns the struct name (and 0 on failure)
-SIMD_FORCE_INLINE	const char*	btCylinderShape::serialize(void* dataBuffer, btSerializer* serializer) const
+///fills the dataBuffer and returns the struct name (and 0 on failure)
+SIMD_FORCE_INLINE const char* btCylinderShape::serialize(void* dataBuffer, btSerializer* serializer) const
 {
-	btCylinderShapeData* shapeData = (btCylinderShapeData*) dataBuffer;
+	btCylinderShapeData* shapeData = (btCylinderShapeData*)dataBuffer;
 
-	btConvexInternalShape::serialize(&shapeData->m_convexInternalShapeData,serializer);
+	btConvexInternalShape::serialize(&shapeData->m_convexInternalShapeData, serializer);
 
 	shapeData->m_upAxis = m_upAxis;
 
@@ -213,7 +203,4 @@ SIMD_FORCE_INLINE	const char*	btCylinderShape::serialize(void* dataBuffer, btSer
 	return "btCylinderShapeData";
 }
 
-
-
-#endif //BT_CYLINDER_MINKOWSKI_H
-
+#endif  //BT_CYLINDER_MINKOWSKI_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btEmptyShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btEmptyShape.cpp
index a9e6df5c58e..4699555bd87 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btEmptyShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btEmptyShape.cpp
@@ -15,36 +15,28 @@ subject to the following restrictions:
 
 #include "btEmptyShape.h"
 
-
 #include "btCollisionShape.h"
 
-
-btEmptyShape::btEmptyShape() : btConcaveShape ()
+btEmptyShape::btEmptyShape() : btConcaveShape()
 {
 	m_shapeType = EMPTY_SHAPE_PROXYTYPE;
 }
 
-
 btEmptyShape::~btEmptyShape()
 {
 }
 
-
-	///getAabb's default implementation is brute force, expected derived classes to implement a fast dedicated version
-void btEmptyShape::getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const
+///getAabb's default implementation is brute force, expected derived classes to implement a fast dedicated version
+void btEmptyShape::getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
 {
-	btVector3 margin(getMargin(),getMargin(),getMargin());
+	btVector3 margin(getMargin(), getMargin(), getMargin());
 
 	aabbMin = t.getOrigin() - margin;
 
 	aabbMax = t.getOrigin() + margin;
-
 }
 
-void	btEmptyShape::calculateLocalInertia(btScalar ,btVector3& ) const
+void btEmptyShape::calculateLocalInertia(btScalar, btVector3&) const
 {
 	btAssert(0);
 }
-
-	
-	
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btEmptyShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btEmptyShape.h
index 069a79402bf..d2e21173b21 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btEmptyShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btEmptyShape.h
@@ -23,50 +23,43 @@ subject to the following restrictions:
 #include "LinearMath/btMatrix3x3.h"
 #include "btCollisionMargin.h"
 
-
-
-
 /// The btEmptyShape is a collision shape without actual collision detection shape, so most users should ignore this class.
 /// It can be replaced by another shape during runtime, but the inertia tensor should be recomputed.
-ATTRIBUTE_ALIGNED16(class) btEmptyShape	: public btConcaveShape
+ATTRIBUTE_ALIGNED16(class)
+btEmptyShape : public btConcaveShape
 {
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
+
 	btEmptyShape();
 
 	virtual ~btEmptyShape();
 
-
 	///getAabb's default implementation is brute force, expected derived classes to implement a fast dedicated version
-	void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const;
+	void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
 
-
-	virtual void	setLocalScaling(const btVector3& scaling)
+	virtual void setLocalScaling(const btVector3& scaling)
 	{
 		m_localScaling = scaling;
 	}
-	virtual const btVector3& getLocalScaling() const 
+	virtual const btVector3& getLocalScaling() const
 	{
 		return m_localScaling;
 	}
 
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
-	
-	virtual const char*	getName()const
+	virtual void calculateLocalInertia(btScalar mass, btVector3 & inertia) const;
+
+	virtual const char* getName() const
 	{
 		return "Empty";
 	}
 
-	virtual void processAllTriangles(btTriangleCallback* ,const btVector3& ,const btVector3& ) const
+	virtual void processAllTriangles(btTriangleCallback*, const btVector3&, const btVector3&) const
 	{
 	}
 
 protected:
-	btVector3	m_localScaling;
-
+	btVector3 m_localScaling;
 };
 
-
-
-#endif //BT_EMPTY_SHAPE_H
+#endif  //BT_EMPTY_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btHeightfieldTerrainShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btHeightfieldTerrainShape.cpp
index 441a89c6bb6..c85ce2498e9 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btHeightfieldTerrainShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btHeightfieldTerrainShape.cpp
@@ -17,23 +17,17 @@ subject to the following restrictions:
 
 #include "LinearMath/btTransformUtil.h"
 
-
-
-btHeightfieldTerrainShape::btHeightfieldTerrainShape
-(
-int heightStickWidth, int heightStickLength, const void* heightfieldData,
-btScalar heightScale, btScalar minHeight, btScalar maxHeight,int upAxis,
-PHY_ScalarType hdt, bool flipQuadEdges
-)
+btHeightfieldTerrainShape::btHeightfieldTerrainShape(
+	int heightStickWidth, int heightStickLength, const void* heightfieldData,
+	btScalar heightScale, btScalar minHeight, btScalar maxHeight, int upAxis,
+	PHY_ScalarType hdt, bool flipQuadEdges)
 {
 	initialize(heightStickWidth, heightStickLength, heightfieldData,
-	           heightScale, minHeight, maxHeight, upAxis, hdt,
-	           flipQuadEdges);
+			   heightScale, minHeight, maxHeight, upAxis, hdt,
+			   flipQuadEdges);
 }
 
-
-
-btHeightfieldTerrainShape::btHeightfieldTerrainShape(int heightStickWidth, int heightStickLength,const void* heightfieldData,btScalar maxHeight,int upAxis,bool useFloatData,bool flipQuadEdges)
+btHeightfieldTerrainShape::btHeightfieldTerrainShape(int heightStickWidth, int heightStickLength, const void* heightfieldData, btScalar maxHeight, int upAxis, bool useFloatData, bool flipQuadEdges)
 {
 	// legacy constructor: support only float or unsigned char,
 	// 	and min height is zero
@@ -45,27 +39,23 @@ btHeightfieldTerrainShape::btHeightfieldTerrainShape(int heightStickWidth, int h
 	btScalar heightScale = maxHeight / 65535;
 
 	initialize(heightStickWidth, heightStickLength, heightfieldData,
-	           heightScale, minHeight, maxHeight, upAxis, hdt,
-	           flipQuadEdges);
+			   heightScale, minHeight, maxHeight, upAxis, hdt,
+			   flipQuadEdges);
 }
 
-
-
-void btHeightfieldTerrainShape::initialize
-(
-int heightStickWidth, int heightStickLength, const void* heightfieldData,
-btScalar heightScale, btScalar minHeight, btScalar maxHeight, int upAxis,
-PHY_ScalarType hdt, bool flipQuadEdges
-)
+void btHeightfieldTerrainShape::initialize(
+	int heightStickWidth, int heightStickLength, const void* heightfieldData,
+	btScalar heightScale, btScalar minHeight, btScalar maxHeight, int upAxis,
+	PHY_ScalarType hdt, bool flipQuadEdges)
 {
 	// validation
-	btAssert(heightStickWidth > 1);// && "bad width");
-	btAssert(heightStickLength > 1);// && "bad length");
-	btAssert(heightfieldData);// && "null heightfield data");
+	btAssert(heightStickWidth > 1);   // && "bad width");
+	btAssert(heightStickLength > 1);  // && "bad length");
+	btAssert(heightfieldData);        // && "null heightfield data");
 	// btAssert(heightScale) -- do we care?  Trust caller here
-	btAssert(minHeight <= maxHeight);// && "bad min/max height");
-	btAssert(upAxis >= 0 && upAxis < 3);// && "bad upAxis--should be in range [0,2]");
-	btAssert(hdt != PHY_UCHAR || hdt != PHY_FLOAT || hdt != PHY_SHORT);// && "Bad height data type enum");
+	btAssert(minHeight <= maxHeight);                                    // && "bad min/max height");
+	btAssert(upAxis >= 0 && upAxis < 3);                                 // && "bad upAxis--should be in range [0,2]");
+	btAssert(hdt != PHY_UCHAR || hdt != PHY_FLOAT || hdt != PHY_SHORT);  // && "Bad height data type enum");
 
 	// initialize member variables
 	m_shapeType = TERRAIN_SHAPE_PROXYTYPE;
@@ -73,8 +63,8 @@ PHY_ScalarType hdt, bool flipQuadEdges
 	m_heightStickLength = heightStickLength;
 	m_minHeight = minHeight;
 	m_maxHeight = maxHeight;
-	m_width = (btScalar) (heightStickWidth - 1);
-	m_length = (btScalar) (heightStickLength - 1);
+	m_width = (btScalar)(heightStickWidth - 1);
+	m_length = (btScalar)(heightStickLength - 1);
 	m_heightScale = heightScale;
 	m_heightfieldDataUnknown = heightfieldData;
 	m_heightDataType = hdt;
@@ -87,28 +77,28 @@ PHY_ScalarType hdt, bool flipQuadEdges
 	// determine min/max axis-aligned bounding box (aabb) values
 	switch (m_upAxis)
 	{
-	case 0:
+		case 0:
 		{
 			m_localAabbMin.setValue(m_minHeight, 0, 0);
 			m_localAabbMax.setValue(m_maxHeight, m_width, m_length);
 			break;
 		}
-	case 1:
+		case 1:
 		{
 			m_localAabbMin.setValue(0, m_minHeight, 0);
 			m_localAabbMax.setValue(m_width, m_maxHeight, m_length);
 			break;
 		};
-	case 2:
+		case 2:
 		{
 			m_localAabbMin.setValue(0, 0, m_minHeight);
 			m_localAabbMax.setValue(m_width, m_length, m_maxHeight);
 			break;
 		}
-	default:
+		default:
 		{
 			//need to get valid m_upAxis
-			btAssert(0);// && "Bad m_upAxis");
+			btAssert(0);  // && "Bad m_upAxis");
 		}
 	}
 
@@ -116,62 +106,57 @@ PHY_ScalarType hdt, bool flipQuadEdges
 	m_localOrigin = btScalar(0.5) * (m_localAabbMin + m_localAabbMax);
 }
 
-
-
 btHeightfieldTerrainShape::~btHeightfieldTerrainShape()
 {
 }
 
-
-
-void btHeightfieldTerrainShape::getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const
+void btHeightfieldTerrainShape::getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
 {
-	btVector3 halfExtents = (m_localAabbMax-m_localAabbMin)* m_localScaling * btScalar(0.5);
+	btVector3 halfExtents = (m_localAabbMax - m_localAabbMin) * m_localScaling * btScalar(0.5);
 
 	btVector3 localOrigin(0, 0, 0);
 	localOrigin[m_upAxis] = (m_minHeight + m_maxHeight) * btScalar(0.5);
 	localOrigin *= m_localScaling;
 
-	btMatrix3x3 abs_b = t.getBasis().absolute();  
+	btMatrix3x3 abs_b = t.getBasis().absolute();
 	btVector3 center = t.getOrigin();
-    btVector3 extent = halfExtents.dot3(abs_b[0], abs_b[1], abs_b[2]);
-	extent += btVector3(getMargin(),getMargin(),getMargin());
+	btVector3 extent = halfExtents.dot3(abs_b[0], abs_b[1], abs_b[2]);
+	extent += btVector3(getMargin(), getMargin(), getMargin());
 
 	aabbMin = center - extent;
 	aabbMax = center + extent;
 }
 
-
 /// This returns the "raw" (user's initial) height, not the actual height.
 /// The actual height needs to be adjusted to be relative to the center
 ///   of the heightfield's AABB.
 btScalar
-btHeightfieldTerrainShape::getRawHeightFieldValue(int x,int y) const
+btHeightfieldTerrainShape::getRawHeightFieldValue(int x, int y) const
 {
 	btScalar val = 0.f;
 	switch (m_heightDataType)
 	{
-	case PHY_FLOAT:
+		case PHY_FLOAT:
 		{
-			val = m_heightfieldDataFloat[(y*m_heightStickWidth)+x];
+			val = m_heightfieldDataFloat[(y * m_heightStickWidth) + x];
 			break;
 		}
 
-	case PHY_UCHAR:
+		case PHY_UCHAR:
 		{
-			unsigned char heightFieldValue = m_heightfieldDataUnsignedChar[(y*m_heightStickWidth)+x];
+			unsigned char heightFieldValue = m_heightfieldDataUnsignedChar[(y * m_heightStickWidth) + x];
 			val = heightFieldValue * m_heightScale;
 			break;
 		}
 
-	case PHY_SHORT:
+		case PHY_SHORT:
 		{
 			short hfValue = m_heightfieldDataShort[(y * m_heightStickWidth) + x];
 			val = hfValue * m_heightScale;
 			break;
 		}
 
-	default:
+		default:
 		{
 			btAssert(!"Bad m_heightDataType");
 		}
@@ -180,74 +165,63 @@ btHeightfieldTerrainShape::getRawHeightFieldValue(int x,int y) const
 	return val;
 }
 
-
-
-
 /// this returns the vertex in bullet-local coordinates
-void	btHeightfieldTerrainShape::getVertex(int x,int y,btVector3& vertex) const
+void btHeightfieldTerrainShape::getVertex(int x, int y, btVector3& vertex) const
 {
-	btAssert(x>=0);
-	btAssert(y>=0);
-	btAssert(x<m_heightStickWidth);
-	btAssert(y<m_heightStickLength);
+	btAssert(x >= 0);
+	btAssert(y >= 0);
+	btAssert(x < m_heightStickWidth);
+	btAssert(y < m_heightStickLength);
 
-	btScalar	height = getRawHeightFieldValue(x,y);
+	btScalar height = getRawHeightFieldValue(x, y);
 
 	switch (m_upAxis)
 	{
-	case 0:
-		{
-		vertex.setValue(
-			height - m_localOrigin.getX(),
-			(-m_width/btScalar(2.0)) + x,
-			(-m_length/btScalar(2.0) ) + y
-			);
-			break;
-		}
-	case 1:
+		case 0:
 		{
 			vertex.setValue(
-			(-m_width/btScalar(2.0)) + x,
-			height - m_localOrigin.getY(),
-			(-m_length/btScalar(2.0)) + y
-			);
+				height - m_localOrigin.getX(),
+				(-m_width / btScalar(2.0)) + x,
+				(-m_length / btScalar(2.0)) + y);
+			break;
+		}
+		case 1:
+		{
+			vertex.setValue(
+				(-m_width / btScalar(2.0)) + x,
+				height - m_localOrigin.getY(),
+				(-m_length / btScalar(2.0)) + y);
 			break;
 		};
-	case 2:
+		case 2:
 		{
 			vertex.setValue(
-			(-m_width/btScalar(2.0)) + x,
-			(-m_length/btScalar(2.0)) + y,
-			height - m_localOrigin.getZ()
-			);
+				(-m_width / btScalar(2.0)) + x,
+				(-m_length / btScalar(2.0)) + y,
+				height - m_localOrigin.getZ());
 			break;
 		}
-	default:
+		default:
 		{
 			//need to get valid m_upAxis
 			btAssert(0);
 		}
 	}
 
-	vertex*=m_localScaling;
+	vertex *= m_localScaling;
 }
 
-
-
 static inline int
-getQuantized
-(
-btScalar x
-)
+getQuantized(
+	btScalar x)
 {
-	if (x < 0.0) {
-		return (int) (x - 0.5);
+	if (x < 0.0)
+	{
+		return (int)(x - 0.5);
 	}
-	return (int) (x + 0.5);
+	return (int)(x + 0.5);
 }
 
-
-
 /// given input vector, return quantized version
 /**
   This routine is basically determining the gridpoint indices for a given
@@ -257,7 +231,7 @@ btScalar x
   "with clamp" means that we restrict the point to be in the heightfield's
   axis-aligned bounding box.
  */
-void btHeightfieldTerrainShape::quantizeWithClamp(int* out, const btVector3& point,int /*isMax*/) const
+void btHeightfieldTerrainShape::quantizeWithClamp(int* out, const btVector3& point, int /*isMax*/) const
 {
 	btVector3 clampedPoint(point);
 	clampedPoint.setMax(m_localAabbMin);
@@ -266,11 +240,8 @@ void btHeightfieldTerrainShape::quantizeWithClamp(int* out, const btVector3& poi
 	out[0] = getQuantized(clampedPoint.getX());
 	out[1] = getQuantized(clampedPoint.getY());
 	out[2] = getQuantized(clampedPoint.getZ());
-		
 }
 
-
-
 /// process all triangles within the provided axis-aligned bounding box
 /**
   basic algorithm:
@@ -278,128 +249,123 @@ void btHeightfieldTerrainShape::quantizeWithClamp(int* out, const btVector3& poi
     - convert input aabb to a range of heightfield grid points (quantize)
     - iterate over all triangles in that subset of the grid
  */
-void	btHeightfieldTerrainShape::processAllTriangles(btTriangleCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax) const
+void btHeightfieldTerrainShape::processAllTriangles(btTriangleCallback* callback, const btVector3& aabbMin, const btVector3& aabbMax) const
 {
 	// scale down the input aabb's so they are in local (non-scaled) coordinates
-	btVector3	localAabbMin = aabbMin*btVector3(1.f/m_localScaling[0],1.f/m_localScaling[1],1.f/m_localScaling[2]);
-	btVector3	localAabbMax = aabbMax*btVector3(1.f/m_localScaling[0],1.f/m_localScaling[1],1.f/m_localScaling[2]);
+	btVector3 localAabbMin = aabbMin * btVector3(1.f / m_localScaling[0], 1.f / m_localScaling[1], 1.f / m_localScaling[2]);
+	btVector3 localAabbMax = aabbMax * btVector3(1.f / m_localScaling[0], 1.f / m_localScaling[1], 1.f / m_localScaling[2]);
 
 	// account for local origin
 	localAabbMin += m_localOrigin;
 	localAabbMax += m_localOrigin;
 
 	//quantize the aabbMin and aabbMax, and adjust the start/end ranges
-	int	quantizedAabbMin[3];
-	int	quantizedAabbMax[3];
-	quantizeWithClamp(quantizedAabbMin, localAabbMin,0);
-	quantizeWithClamp(quantizedAabbMax, localAabbMax,1);
-	
+	int quantizedAabbMin[3];
+	int quantizedAabbMax[3];
+	quantizeWithClamp(quantizedAabbMin, localAabbMin, 0);
+	quantizeWithClamp(quantizedAabbMax, localAabbMax, 1);
+
 	// expand the min/max quantized values
 	// this is to catch the case where the input aabb falls between grid points!
-	for (int i = 0; i < 3; ++i) {
+	for (int i = 0; i < 3; ++i)
+	{
 		quantizedAabbMin[i]--;
 		quantizedAabbMax[i]++;
-	}	
+	}
 
-	int startX=0;
-	int endX=m_heightStickWidth-1;
-	int startJ=0;
-	int endJ=m_heightStickLength-1;
+	int startX = 0;
+	int endX = m_heightStickWidth - 1;
+	int startJ = 0;
+	int endJ = m_heightStickLength - 1;
 
 	switch (m_upAxis)
 	{
-	case 0:
+		case 0:
 		{
-			if (quantizedAabbMin[1]>startX)
+			if (quantizedAabbMin[1] > startX)
 				startX = quantizedAabbMin[1];
-			if (quantizedAabbMax[1]<endX)
+			if (quantizedAabbMax[1] < endX)
 				endX = quantizedAabbMax[1];
-			if (quantizedAabbMin[2]>startJ)
+			if (quantizedAabbMin[2] > startJ)
 				startJ = quantizedAabbMin[2];
-			if (quantizedAabbMax[2]<endJ)
+			if (quantizedAabbMax[2] < endJ)
 				endJ = quantizedAabbMax[2];
 			break;
 		}
-	case 1:
+		case 1:
 		{
-			if (quantizedAabbMin[0]>startX)
+			if (quantizedAabbMin[0] > startX)
 				startX = quantizedAabbMin[0];
-			if (quantizedAabbMax[0]<endX)
+			if (quantizedAabbMax[0] < endX)
 				endX = quantizedAabbMax[0];
-			if (quantizedAabbMin[2]>startJ)
+			if (quantizedAabbMin[2] > startJ)
 				startJ = quantizedAabbMin[2];
-			if (quantizedAabbMax[2]<endJ)
+			if (quantizedAabbMax[2] < endJ)
 				endJ = quantizedAabbMax[2];
 			break;
 		};
-	case 2:
+		case 2:
 		{
-			if (quantizedAabbMin[0]>startX)
+			if (quantizedAabbMin[0] > startX)
 				startX = quantizedAabbMin[0];
-			if (quantizedAabbMax[0]<endX)
+			if (quantizedAabbMax[0] < endX)
 				endX = quantizedAabbMax[0];
-			if (quantizedAabbMin[1]>startJ)
+			if (quantizedAabbMin[1] > startJ)
 				startJ = quantizedAabbMin[1];
-			if (quantizedAabbMax[1]<endJ)
+			if (quantizedAabbMax[1] < endJ)
 				endJ = quantizedAabbMax[1];
 			break;
 		}
-	default:
+		default:
 		{
 			//need to get valid m_upAxis
 			btAssert(0);
 		}
 	}
 
-	
-  
-
-	for(int j=startJ; j<endJ; j++)
+	for (int j = startJ; j < endJ; j++)
 	{
-		for(int x=startX; x<endX; x++)
+		for (int x = startX; x < endX; x++)
 		{
 			btVector3 vertices[3];
-			if (m_flipQuadEdges || (m_useDiamondSubdivision && !((j+x) & 1))|| (m_useZigzagSubdivision  && !(j & 1)))
+			if (m_flipQuadEdges || (m_useDiamondSubdivision && !((j + x) & 1)) || (m_useZigzagSubdivision && !(j & 1)))
 			{
-        //first triangle
-        getVertex(x,j,vertices[0]);
-		getVertex(x, j + 1, vertices[1]);
-		getVertex(x + 1, j + 1, vertices[2]);
-        callback->processTriangle(vertices,x,j);
-        //second triangle
-      //  getVertex(x,j,vertices[0]);//already got this vertex before, thanks to Danny Chapman
-        getVertex(x+1,j+1,vertices[1]);
-		getVertex(x + 1, j, vertices[2]);
-		callback->processTriangle(vertices, x, j);
-
-			} else
+				//first triangle
+				getVertex(x, j, vertices[0]);
+				getVertex(x, j + 1, vertices[1]);
+				getVertex(x + 1, j + 1, vertices[2]);
+				callback->processTriangle(vertices, x, j);
+				//second triangle
+				//  getVertex(x,j,vertices[0]);//already got this vertex before, thanks to Danny Chapman
+				getVertex(x + 1, j + 1, vertices[1]);
+				getVertex(x + 1, j, vertices[2]);
+				callback->processTriangle(vertices, x, j);
+			}
+			else
 			{
-        //first triangle
-        getVertex(x,j,vertices[0]);
-        getVertex(x,j+1,vertices[1]);
-        getVertex(x+1,j,vertices[2]);
-        callback->processTriangle(vertices,x,j);
-        //second triangle
-        getVertex(x+1,j,vertices[0]);
-        //getVertex(x,j+1,vertices[1]);
-        getVertex(x+1,j+1,vertices[2]);
-        callback->processTriangle(vertices,x,j);
+				//first triangle
+				getVertex(x, j, vertices[0]);
+				getVertex(x, j + 1, vertices[1]);
+				getVertex(x + 1, j, vertices[2]);
+				callback->processTriangle(vertices, x, j);
+				//second triangle
+				getVertex(x + 1, j, vertices[0]);
+				//getVertex(x,j+1,vertices[1]);
+				getVertex(x + 1, j + 1, vertices[2]);
+				callback->processTriangle(vertices, x, j);
 			}
 		}
 	}
-
-	
-
 }
 
-void	btHeightfieldTerrainShape::calculateLocalInertia(btScalar ,btVector3& inertia) const
+void btHeightfieldTerrainShape::calculateLocalInertia(btScalar, btVector3& inertia) const
 {
 	//moving concave objects not supported
-	
-	inertia.setValue(btScalar(0.),btScalar(0.),btScalar(0.));
+
+	inertia.setValue(btScalar(0.), btScalar(0.), btScalar(0.));
 }
 
-void	btHeightfieldTerrainShape::setLocalScaling(const btVector3& scaling)
+void btHeightfieldTerrainShape::setLocalScaling(const btVector3& scaling)
 {
 	m_localScaling = scaling;
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btHeightfieldTerrainShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btHeightfieldTerrainShape.h
index 4a7a4a4bda4..8a50a57e316 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btHeightfieldTerrainShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btHeightfieldTerrainShape.h
@@ -68,43 +68,41 @@ subject to the following restrictions:
 
   For usage and testing see the TerrainDemo.
  */
-ATTRIBUTE_ALIGNED16(class) btHeightfieldTerrainShape : public btConcaveShape
+ATTRIBUTE_ALIGNED16(class)
+btHeightfieldTerrainShape : public btConcaveShape
 {
 protected:
-	btVector3	m_localAabbMin;
-	btVector3	m_localAabbMax;
-	btVector3	m_localOrigin;
+	btVector3 m_localAabbMin;
+	btVector3 m_localAabbMax;
+	btVector3 m_localOrigin;
 
 	///terrain data
-	int	m_heightStickWidth;
+	int m_heightStickWidth;
 	int m_heightStickLength;
-	btScalar	m_minHeight;
-	btScalar	m_maxHeight;
+	btScalar m_minHeight;
+	btScalar m_maxHeight;
 	btScalar m_width;
 	btScalar m_length;
 	btScalar m_heightScale;
-	union
-	{
-		const unsigned char*	m_heightfieldDataUnsignedChar;
-		const short*		m_heightfieldDataShort;
-		const btScalar*			m_heightfieldDataFloat;
-		const void*	m_heightfieldDataUnknown;
+	union {
+		const unsigned char* m_heightfieldDataUnsignedChar;
+		const short* m_heightfieldDataShort;
+		const btScalar* m_heightfieldDataFloat;
+		const void* m_heightfieldDataUnknown;
 	};
 
-	PHY_ScalarType	m_heightDataType;	
-	bool	m_flipQuadEdges;
-  	bool  m_useDiamondSubdivision;
+	PHY_ScalarType m_heightDataType;
+	bool m_flipQuadEdges;
+	bool m_useDiamondSubdivision;
 	bool m_useZigzagSubdivision;
 
-	int	m_upAxis;
-	
-	btVector3	m_localScaling;
-
-	virtual btScalar	getRawHeightFieldValue(int x,int y) const;
-	void		quantizeWithClamp(int* out, const btVector3& point,int isMax) const;
-	void		getVertex(int x,int y,btVector3& vertex) const;
+	int m_upAxis;
 
+	btVector3 m_localScaling;
 
+	virtual btScalar getRawHeightFieldValue(int x, int y) const;
+	void quantizeWithClamp(int* out, const btVector3& point, int isMax) const;
+	void getVertex(int x, int y, btVector3& vertex) const;
 
 	/// protected initialization
 	/**
@@ -112,25 +110,24 @@ protected:
 	  backwards-compatible without a lot of copy/paste.
 	 */
 	void initialize(int heightStickWidth, int heightStickLength,
-	                const void* heightfieldData, btScalar heightScale,
-	                btScalar minHeight, btScalar maxHeight, int upAxis,
-	                PHY_ScalarType heightDataType, bool flipQuadEdges);
+					const void* heightfieldData, btScalar heightScale,
+					btScalar minHeight, btScalar maxHeight, int upAxis,
+					PHY_ScalarType heightDataType, bool flipQuadEdges);
 
 public:
-	
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
+
 	/// preferred constructor
 	/**
 	  This constructor supports a range of heightfield
 	  data types, and allows for a non-zero minimum height value.
 	  heightScale is needed for any integer-based heightfield data types.
 	 */
-	btHeightfieldTerrainShape(int heightStickWidth,int heightStickLength,
-	                          const void* heightfieldData, btScalar heightScale,
-	                          btScalar minHeight, btScalar maxHeight,
-	                          int upAxis, PHY_ScalarType heightDataType,
-	                          bool flipQuadEdges);
+	btHeightfieldTerrainShape(int heightStickWidth, int heightStickLength,
+							  const void* heightfieldData, btScalar heightScale,
+							  btScalar minHeight, btScalar maxHeight,
+							  int upAxis, PHY_ScalarType heightDataType,
+							  bool flipQuadEdges);
 
 	/// legacy constructor
 	/**
@@ -139,29 +136,27 @@ public:
 	  compatibility reasons, heightScale is calculated as maxHeight / 65535 
 	  (and is only used when useFloatData = false).
  	 */
-	btHeightfieldTerrainShape(int heightStickWidth,int heightStickLength,const void* heightfieldData, btScalar maxHeight,int upAxis,bool useFloatData,bool flipQuadEdges);
+	btHeightfieldTerrainShape(int heightStickWidth, int heightStickLength, const void* heightfieldData, btScalar maxHeight, int upAxis, bool useFloatData, bool flipQuadEdges);
 
 	virtual ~btHeightfieldTerrainShape();
 
+	void setUseDiamondSubdivision(bool useDiamondSubdivision = true) { m_useDiamondSubdivision = useDiamondSubdivision; }
 
-	void setUseDiamondSubdivision(bool useDiamondSubdivision=true) { m_useDiamondSubdivision = useDiamondSubdivision;}
+	///could help compatibility with Ogre heightfields. See https://code.google.com/p/bullet/issues/detail?id=625
+	void setUseZigzagSubdivision(bool useZigzagSubdivision = true) { m_useZigzagSubdivision = useZigzagSubdivision; }
 
-	///could help compatibility with Ogre heightfields. See https://code.google.com/p/bullet/issues/detail?id=625	
-	void setUseZigzagSubdivision(bool useZigzagSubdivision=true) { m_useZigzagSubdivision = useZigzagSubdivision;}
+	virtual void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
 
-	virtual void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const;
+	virtual void processAllTriangles(btTriangleCallback * callback, const btVector3& aabbMin, const btVector3& aabbMax) const;
 
-	virtual void	processAllTriangles(btTriangleCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax) const;
+	virtual void calculateLocalInertia(btScalar mass, btVector3 & inertia) const;
 
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
+	virtual void setLocalScaling(const btVector3& scaling);
 
-	virtual void	setLocalScaling(const btVector3& scaling);
-	
 	virtual const btVector3& getLocalScaling() const;
-	
-	//debugging
-	virtual const char*	getName()const {return "HEIGHTFIELD";}
 
+	//debugging
+	virtual const char* getName() const { return "HEIGHTFIELD"; }
 };
 
-#endif //BT_HEIGHTFIELD_TERRAIN_SHAPE_H
+#endif  //BT_HEIGHTFIELD_TERRAIN_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btMaterial.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btMaterial.h
index 866f9b4da4d..c9a436bf26a 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btMaterial.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btMaterial.h
@@ -21,15 +21,18 @@ subject to the following restrictions:
 // Material class to be used by btMultimaterialTriangleMeshShape to store triangle properties
 class btMaterial
 {
-    // public members so that materials can change due to world events
+	// public members so that materials can change due to world events
 public:
-    btScalar m_friction;
-    btScalar m_restitution;
-    int pad[2];
+	btScalar m_friction;
+	btScalar m_restitution;
+	int pad[2];
 
-    btMaterial(){}
-    btMaterial(btScalar fric, btScalar rest) { m_friction = fric; m_restitution = rest; }
+	btMaterial() {}
+	btMaterial(btScalar fric, btScalar rest)
+	{
+		m_friction = fric;
+		m_restitution = rest;
+	}
 };
 
-#endif // BT_MATERIAL_H
-
+#endif  // BT_MATERIAL_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btMiniSDF.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btMiniSDF.cpp
index afe45e1d2d0..13c0a343f17 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btMiniSDF.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btMiniSDF.cpp
@@ -10,31 +10,31 @@
 //
 
 #include <limits.h>
-#include <string.h> //memcpy
+#include <string.h>  //memcpy
 
 struct btSdfDataStream
 {
 	const char* m_data;
 	int m_size;
-	
+
 	int m_currentOffset;
-	
+
 	btSdfDataStream(const char* data, int size)
-		:m_data(data),
-		m_size(size),
-		m_currentOffset(0)
+		: m_data(data),
+		  m_size(size),
+		  m_currentOffset(0)
 	{
-		
 	}
 
-	template<class T> bool read(T& val)
+	template <class T>
+	bool read(T& val)
 	{
 		int bytes = sizeof(T);
-		if (m_currentOffset+bytes<=m_size)
+		if (m_currentOffset + bytes <= m_size)
 		{
 			char* dest = (char*)&val;
-			memcpy(dest,&m_data[m_currentOffset],bytes);
-			m_currentOffset+=bytes;
+			memcpy(dest, &m_data[m_currentOffset], bytes);
+			m_currentOffset += bytes;
 			return true;
 		}
 		btAssert(0);
@@ -42,137 +42,133 @@ struct btSdfDataStream
 	}
 };
 
-
 bool btMiniSDF::load(const char* data, int size)
 {
-		int fileSize = -1;
+	int fileSize = -1;
 
-		btSdfDataStream ds(data,size);
-		{
-			double buf[6];
-			ds.read(buf);
-			m_domain.m_min[0] = buf[0];
-			m_domain.m_min[1] = buf[1];
-			m_domain.m_min[2] = buf[2];
-			m_domain.m_min[3] = 0;
-			m_domain.m_max[0] = buf[3];
-			m_domain.m_max[1] = buf[4];
-			m_domain.m_max[2] = buf[5];
-			m_domain.m_max[3] = 0;
-		}
-		{
-			unsigned int buf2[3];
-			ds.read(buf2);
-			m_resolution[0] = buf2[0];
-			m_resolution[1] = buf2[1];
-			m_resolution[2] = buf2[2];
-		}
-		{
-			double buf[3];
-			ds.read(buf);
-			m_cell_size[0] = buf[0];
-			m_cell_size[1] = buf[1];
-			m_cell_size[2] = buf[2];
-		}
-		{
-			double buf[3];
-			ds.read(buf);
-			m_inv_cell_size[0] = buf[0];
-			m_inv_cell_size[1] = buf[1];
-			m_inv_cell_size[2] = buf[2];
-		}
-		{
-			unsigned long long int cells;
-			ds.read(cells);
-			m_n_cells = cells;
-		}
-		{
-			unsigned long long int fields;
-			ds.read(fields);
-			m_n_fields = fields;
-		}
+	btSdfDataStream ds(data, size);
+	{
+		double buf[6];
+		ds.read(buf);
+		m_domain.m_min[0] = buf[0];
+		m_domain.m_min[1] = buf[1];
+		m_domain.m_min[2] = buf[2];
+		m_domain.m_min[3] = 0;
+		m_domain.m_max[0] = buf[3];
+		m_domain.m_max[1] = buf[4];
+		m_domain.m_max[2] = buf[5];
+		m_domain.m_max[3] = 0;
+	}
+	{
+		unsigned int buf2[3];
+		ds.read(buf2);
+		m_resolution[0] = buf2[0];
+		m_resolution[1] = buf2[1];
+		m_resolution[2] = buf2[2];
+	}
+	{
+		double buf[3];
+		ds.read(buf);
+		m_cell_size[0] = buf[0];
+		m_cell_size[1] = buf[1];
+		m_cell_size[2] = buf[2];
+	}
+	{
+		double buf[3];
+		ds.read(buf);
+		m_inv_cell_size[0] = buf[0];
+		m_inv_cell_size[1] = buf[1];
+		m_inv_cell_size[2] = buf[2];
+	}
+	{
+		unsigned long long int cells;
+		ds.read(cells);
+		m_n_cells = cells;
+	}
+	{
+		unsigned long long int fields;
+		ds.read(fields);
+		m_n_fields = fields;
+	}
 
-		
-		unsigned long long int nodes0;
-		std::size_t n_nodes0;
-		ds.read(nodes0);
-		n_nodes0 = nodes0;
-		if (n_nodes0 > 1024 * 1024 * 1024)
-		{
-			return m_isValid;
-		}
-		m_nodes.resize(n_nodes0);
-		for (unsigned int i=0;i<n_nodes0;i++)
-		{
-			unsigned long long int n_nodes1;
-			ds.read(n_nodes1);
-			btAlignedObjectArray<double>& nodes = m_nodes[i];
-			nodes.resize(n_nodes1);
-			for ( int j=0;j<nodes.size();j++)
-			{
-				double& node = nodes[j];
-				ds.read(node);
-			}
-		}
-
-		unsigned long long int n_cells0;
-		ds.read(n_cells0);
-		m_cells.resize(n_cells0);
-		for (int i=0;i<n_cells0;i++)
-		{
-			unsigned long long int n_cells1;
-			btAlignedObjectArray<btCell32 >& cells = m_cells[i];
-			ds.read(n_cells1);
-			cells.resize(n_cells1);
-			for (int j=0;j<n_cells1;j++)
-			{
-				btCell32& cell = cells[j];
-				ds.read(cell);
-			}
-		}
-
-		{
-			unsigned long long int n_cell_maps0;
-			ds.read(n_cell_maps0);
-		
-			m_cell_map.resize(n_cell_maps0);
-			for (int i=0;i<n_cell_maps0;i++)
-			{
-				unsigned long long int n_cell_maps1;
-				btAlignedObjectArray<unsigned int>& cell_maps = m_cell_map[i];
-				ds.read(n_cell_maps1);
-				cell_maps.resize(n_cell_maps1);
-				for (int j=0;j<n_cell_maps1;j++)
-				{
-					unsigned int& cell_map = cell_maps[j];
-					ds.read(cell_map);
-				}
-			}
-		}
-
-		m_isValid = (ds.m_currentOffset == ds.m_size);
+	unsigned long long int nodes0;
+	std::size_t n_nodes0;
+	ds.read(nodes0);
+	n_nodes0 = nodes0;
+	if (n_nodes0 > 1024 * 1024 * 1024)
+	{
 		return m_isValid;
+	}
+	m_nodes.resize(n_nodes0);
+	for (unsigned int i = 0; i < n_nodes0; i++)
+	{
+		unsigned long long int n_nodes1;
+		ds.read(n_nodes1);
+		btAlignedObjectArray<double>& nodes = m_nodes[i];
+		nodes.resize(n_nodes1);
+		for (int j = 0; j < nodes.size(); j++)
+		{
+			double& node = nodes[j];
+			ds.read(node);
+		}
+	}
+
+	unsigned long long int n_cells0;
+	ds.read(n_cells0);
+	m_cells.resize(n_cells0);
+	for (int i = 0; i < n_cells0; i++)
+	{
+		unsigned long long int n_cells1;
+		btAlignedObjectArray<btCell32>& cells = m_cells[i];
+		ds.read(n_cells1);
+		cells.resize(n_cells1);
+		for (int j = 0; j < n_cells1; j++)
+		{
+			btCell32& cell = cells[j];
+			ds.read(cell);
+		}
+	}
+
+	{
+		unsigned long long int n_cell_maps0;
+		ds.read(n_cell_maps0);
+
+		m_cell_map.resize(n_cell_maps0);
+		for (int i = 0; i < n_cell_maps0; i++)
+		{
+			unsigned long long int n_cell_maps1;
+			btAlignedObjectArray<unsigned int>& cell_maps = m_cell_map[i];
+			ds.read(n_cell_maps1);
+			cell_maps.resize(n_cell_maps1);
+			for (int j = 0; j < n_cell_maps1; j++)
+			{
+				unsigned int& cell_map = cell_maps[j];
+				ds.read(cell_map);
+			}
+		}
+	}
+
+	m_isValid = (ds.m_currentOffset == ds.m_size);
+	return m_isValid;
 }
 
-
-unsigned int btMiniSDF::multiToSingleIndex(btMultiIndex const & ijk) const
+unsigned int btMiniSDF::multiToSingleIndex(btMultiIndex const& ijk) const
 {
 	return m_resolution[1] * m_resolution[0] * ijk.ijk[2] + m_resolution[0] * ijk.ijk[1] + ijk.ijk[0];
 }
-	
+
 btAlignedBox3d
 btMiniSDF::subdomain(btMultiIndex const& ijk) const
 {
 	btAssert(m_isValid);
 	btVector3 tmp;
-	tmp.m_floats[0] = m_cell_size[0]*(double)ijk.ijk[0];
-	tmp.m_floats[1] = m_cell_size[1]*(double)ijk.ijk[1];
-	tmp.m_floats[2] = m_cell_size[2]*(double)ijk.ijk[2];
-	
+	tmp.m_floats[0] = m_cell_size[0] * (double)ijk.ijk[0];
+	tmp.m_floats[1] = m_cell_size[1] * (double)ijk.ijk[1];
+	tmp.m_floats[2] = m_cell_size[2] * (double)ijk.ijk[2];
 
 	btVector3 origin = m_domain.min() + tmp;
 
-	btAlignedBox3d box = btAlignedBox3d (origin, origin + m_cell_size);
+	btAlignedBox3d box = btAlignedBox3d(origin, origin + m_cell_size);
 	return box;
 }
 
@@ -181,8 +177,8 @@ btMiniSDF::singleToMultiIndex(unsigned int l) const
 {
 	btAssert(m_isValid);
 	unsigned int n01 = m_resolution[0] * m_resolution[1];
-	unsigned int  k = l / n01;
-	unsigned int  temp = l % n01;
+	unsigned int k = l / n01;
+	unsigned int temp = l % n01;
 	unsigned int j = temp / m_resolution[0];
 	unsigned int i = temp % m_resolution[0];
 	btMultiIndex mi;
@@ -199,59 +195,57 @@ btMiniSDF::subdomain(unsigned int l) const
 	return subdomain(singleToMultiIndex(l));
 }
 
-
 btShapeMatrix
 btMiniSDF::shape_function_(btVector3 const& xi, btShapeGradients* gradient) const
 {
 	btAssert(m_isValid);
 	btShapeMatrix res;
-	
+
 	btScalar x = xi[0];
 	btScalar y = xi[1];
 	btScalar z = xi[2];
 
-	btScalar  x2 = x*x;
-	btScalar  y2 = y*y;
-	btScalar  z2 = z*z;
+	btScalar x2 = x * x;
+	btScalar y2 = y * y;
+	btScalar z2 = z * z;
 
-	btScalar  _1mx = 1.0 - x;
-	btScalar  _1my = 1.0 - y;
-	btScalar  _1mz = 1.0 - z;
+	btScalar _1mx = 1.0 - x;
+	btScalar _1my = 1.0 - y;
+	btScalar _1mz = 1.0 - z;
 
-	btScalar  _1px = 1.0 + x;
-	btScalar  _1py = 1.0 + y;
-	btScalar  _1pz = 1.0 + z;
+	btScalar _1px = 1.0 + x;
+	btScalar _1py = 1.0 + y;
+	btScalar _1pz = 1.0 + z;
 
-	btScalar  _1m3x = 1.0 - 3.0 * x;
-	btScalar  _1m3y = 1.0 - 3.0 * y;
-	btScalar  _1m3z = 1.0 - 3.0 * z;
+	btScalar _1m3x = 1.0 - 3.0 * x;
+	btScalar _1m3y = 1.0 - 3.0 * y;
+	btScalar _1m3z = 1.0 - 3.0 * z;
 
-	btScalar  _1p3x = 1.0 + 3.0 * x;
-	btScalar  _1p3y = 1.0 + 3.0 * y;
-	btScalar  _1p3z = 1.0 + 3.0 * z;
+	btScalar _1p3x = 1.0 + 3.0 * x;
+	btScalar _1p3y = 1.0 + 3.0 * y;
+	btScalar _1p3z = 1.0 + 3.0 * z;
 
-	btScalar  _1mxt1my = _1mx * _1my;
-	btScalar  _1mxt1py = _1mx * _1py;
-	btScalar  _1pxt1my = _1px * _1my;
-	btScalar  _1pxt1py = _1px * _1py;
+	btScalar _1mxt1my = _1mx * _1my;
+	btScalar _1mxt1py = _1mx * _1py;
+	btScalar _1pxt1my = _1px * _1my;
+	btScalar _1pxt1py = _1px * _1py;
 
-	btScalar  _1mxt1mz = _1mx * _1mz;
-	btScalar  _1mxt1pz = _1mx * _1pz;
-	btScalar  _1pxt1mz = _1px * _1mz;
-	btScalar  _1pxt1pz = _1px * _1pz;
+	btScalar _1mxt1mz = _1mx * _1mz;
+	btScalar _1mxt1pz = _1mx * _1pz;
+	btScalar _1pxt1mz = _1px * _1mz;
+	btScalar _1pxt1pz = _1px * _1pz;
 
-	btScalar  _1myt1mz = _1my * _1mz;
-	btScalar  _1myt1pz = _1my * _1pz;
-	btScalar  _1pyt1mz = _1py * _1mz;
-	btScalar  _1pyt1pz = _1py * _1pz;
-
-	btScalar  _1mx2 = 1.0 - x2;
-	btScalar  _1my2 = 1.0 - y2;
-	btScalar  _1mz2 = 1.0 - z2;
+	btScalar _1myt1mz = _1my * _1mz;
+	btScalar _1myt1pz = _1my * _1pz;
+	btScalar _1pyt1mz = _1py * _1mz;
+	btScalar _1pyt1pz = _1py * _1pz;
 
+	btScalar _1mx2 = 1.0 - x2;
+	btScalar _1my2 = 1.0 - y2;
+	btScalar _1mz2 = 1.0 - z2;
 
 	// Corner nodes.
-	btScalar  fac = 1.0 / 64.0 * (9.0 * (x2 + y2 + z2) - 19.0);
+	btScalar fac = 1.0 / 64.0 * (9.0 * (x2 + y2 + z2) - 19.0);
 	res[0] = fac * _1mxt1my * _1mz;
 	res[1] = fac * _1pxt1my * _1mz;
 	res[2] = fac * _1mxt1py * _1mz;
@@ -264,10 +258,10 @@ btMiniSDF::shape_function_(btVector3 const& xi, btShapeGradients* gradient) cons
 	// Edge nodes.
 
 	fac = 9.0 / 64.0 * _1mx2;
-	btScalar  fact1m3x = fac * _1m3x;
-	btScalar  fact1p3x = fac * _1p3x;
-	res[ 8] = fact1m3x * _1myt1mz;
-	res[ 9] = fact1p3x * _1myt1mz;
+	btScalar fact1m3x = fac * _1m3x;
+	btScalar fact1p3x = fac * _1p3x;
+	res[8] = fact1m3x * _1myt1mz;
+	res[9] = fact1p3x * _1myt1mz;
 	res[10] = fact1m3x * _1myt1pz;
 	res[11] = fact1p3x * _1myt1pz;
 	res[12] = fact1m3x * _1pyt1mz;
@@ -276,8 +270,8 @@ btMiniSDF::shape_function_(btVector3 const& xi, btShapeGradients* gradient) cons
 	res[15] = fact1p3x * _1pyt1pz;
 
 	fac = 9.0 / 64.0 * _1my2;
-	btScalar  fact1m3y = fac * _1m3y;
-	btScalar  fact1p3y = fac * _1p3y;
+	btScalar fact1m3y = fac * _1m3y;
+	btScalar fact1p3y = fac * _1p3y;
 	res[16] = fact1m3y * _1mxt1mz;
 	res[17] = fact1p3y * _1mxt1mz;
 	res[18] = fact1m3y * _1pxt1mz;
@@ -288,8 +282,8 @@ btMiniSDF::shape_function_(btVector3 const& xi, btShapeGradients* gradient) cons
 	res[23] = fact1p3y * _1pxt1pz;
 
 	fac = 9.0 / 64.0 * _1mz2;
-	btScalar  fact1m3z = fac * _1m3z;
-	btScalar  fact1p3z = fac * _1p3z;
+	btScalar fact1m3z = fac * _1m3z;
+	btScalar fact1p3z = fac * _1p3z;
 	res[24] = fact1m3z * _1mxt1my;
 	res[25] = fact1p3z * _1mxt1my;
 	res[26] = fact1m3z * _1mxt1py;
@@ -309,7 +303,7 @@ btMiniSDF::shape_function_(btVector3 const& xi, btShapeGradients* gradient) cons
 		btScalar _18x = 18.0 * x;
 		btScalar _18y = 18.0 * y;
 		btScalar _18z = 18.0 * z;
-		
+
 		btScalar _3m9x2 = 3.0 - 9.0 * x2;
 		btScalar _3m9y2 = 3.0 - 9.0 * y2;
 		btScalar _3m9z2 = 3.0 - 9.0 * z2;
@@ -325,132 +319,128 @@ btMiniSDF::shape_function_(btVector3 const& xi, btShapeGradients* gradient) cons
 		btScalar _18zm9tx2py2p3z2m19 = _18z - _9tx2py2p3z2m19;
 		btScalar _18zp9tx2py2p3z2m19 = _18z + _9tx2py2p3z2m19;
 
-		dN(0,0) =_18xm9t3x2py2pz2m19 * _1myt1mz;
-		dN(0,1) =_1mxt1mz * _18ym9tx2p3y2pz2m19;
-		dN(0,2) =_1mxt1my * _18zm9tx2py2p3z2m19;
-		dN(1,0) =_18xp9t3x2py2pz2m19 * _1myt1mz;
-		dN(1,1) =_1pxt1mz * _18ym9tx2p3y2pz2m19;
-		dN(1,2) =_1pxt1my * _18zm9tx2py2p3z2m19;
-		dN(2,0) =_18xm9t3x2py2pz2m19 * _1pyt1mz;
-		dN(2,1) =_1mxt1mz * _18yp9tx2p3y2pz2m19;
-		dN(2,2) =_1mxt1py * _18zm9tx2py2p3z2m19;
-		dN(3,0) =_18xp9t3x2py2pz2m19 * _1pyt1mz;
-		dN(3,1) =_1pxt1mz * _18yp9tx2p3y2pz2m19;
-		dN(3,2) =_1pxt1py * _18zm9tx2py2p3z2m19;
-		dN(4,0) =_18xm9t3x2py2pz2m19 * _1myt1pz;
-		dN(4,1) =_1mxt1pz * _18ym9tx2p3y2pz2m19;
-		dN(4,2) =_1mxt1my * _18zp9tx2py2p3z2m19;
-		dN(5,0) =_18xp9t3x2py2pz2m19 * _1myt1pz;
-		dN(5,1) =_1pxt1pz * _18ym9tx2p3y2pz2m19;
-		dN(5,2) =_1pxt1my * _18zp9tx2py2p3z2m19;
-		dN(6,0) =_18xm9t3x2py2pz2m19 * _1pyt1pz;
-		dN(6,1) =_1mxt1pz * _18yp9tx2p3y2pz2m19;
-		dN(6,2) =_1mxt1py * _18zp9tx2py2p3z2m19;
-		dN(7,0) =_18xp9t3x2py2pz2m19 * _1pyt1pz;
-		dN(7,1) =_1pxt1pz * _18yp9tx2p3y2pz2m19;
-		dN(7,2) =_1pxt1py * _18zp9tx2py2p3z2m19;
+		dN(0, 0) = _18xm9t3x2py2pz2m19 * _1myt1mz;
+		dN(0, 1) = _1mxt1mz * _18ym9tx2p3y2pz2m19;
+		dN(0, 2) = _1mxt1my * _18zm9tx2py2p3z2m19;
+		dN(1, 0) = _18xp9t3x2py2pz2m19 * _1myt1mz;
+		dN(1, 1) = _1pxt1mz * _18ym9tx2p3y2pz2m19;
+		dN(1, 2) = _1pxt1my * _18zm9tx2py2p3z2m19;
+		dN(2, 0) = _18xm9t3x2py2pz2m19 * _1pyt1mz;
+		dN(2, 1) = _1mxt1mz * _18yp9tx2p3y2pz2m19;
+		dN(2, 2) = _1mxt1py * _18zm9tx2py2p3z2m19;
+		dN(3, 0) = _18xp9t3x2py2pz2m19 * _1pyt1mz;
+		dN(3, 1) = _1pxt1mz * _18yp9tx2p3y2pz2m19;
+		dN(3, 2) = _1pxt1py * _18zm9tx2py2p3z2m19;
+		dN(4, 0) = _18xm9t3x2py2pz2m19 * _1myt1pz;
+		dN(4, 1) = _1mxt1pz * _18ym9tx2p3y2pz2m19;
+		dN(4, 2) = _1mxt1my * _18zp9tx2py2p3z2m19;
+		dN(5, 0) = _18xp9t3x2py2pz2m19 * _1myt1pz;
+		dN(5, 1) = _1pxt1pz * _18ym9tx2p3y2pz2m19;
+		dN(5, 2) = _1pxt1my * _18zp9tx2py2p3z2m19;
+		dN(6, 0) = _18xm9t3x2py2pz2m19 * _1pyt1pz;
+		dN(6, 1) = _1mxt1pz * _18yp9tx2p3y2pz2m19;
+		dN(6, 2) = _1mxt1py * _18zp9tx2py2p3z2m19;
+		dN(7, 0) = _18xp9t3x2py2pz2m19 * _1pyt1pz;
+		dN(7, 1) = _1pxt1pz * _18yp9tx2p3y2pz2m19;
+		dN(7, 2) = _1pxt1py * _18zp9tx2py2p3z2m19;
 
 		dN.topRowsDivide(8, 64.0);
 
 		btScalar _m3m9x2m2x = -_3m9x2 - _2x;
-		btScalar _p3m9x2m2x =  _3m9x2 - _2x;
+		btScalar _p3m9x2m2x = _3m9x2 - _2x;
 		btScalar _1mx2t1m3x = _1mx2 * _1m3x;
 		btScalar _1mx2t1p3x = _1mx2 * _1p3x;
-		dN( 8,0) =    _m3m9x2m2x * _1myt1mz,
-		dN( 8,1) =   -_1mx2t1m3x * _1mz,
-		dN( 8,2) =   -_1mx2t1m3x * _1my;
-		dN( 9,0) =    _p3m9x2m2x * _1myt1mz,
-		dN( 9,1) =   -_1mx2t1p3x * _1mz,
-		dN( 9,2) =   -_1mx2t1p3x * _1my;
-		dN(10,0) =    _m3m9x2m2x * _1myt1pz,
-		dN(10,1) =   -_1mx2t1m3x * _1pz,
-		dN(10,2) =   _1mx2t1m3x * _1my;
-		dN(11,0) =    _p3m9x2m2x * _1myt1pz,
-		dN(11,1) =   -_1mx2t1p3x * _1pz,
-		dN(11,2) =   _1mx2t1p3x * _1my;
-		dN(12,0) =    _m3m9x2m2x * _1pyt1mz,
-		dN(12,1) =   _1mx2t1m3x * _1mz,
-		dN(12,2) =   -_1mx2t1m3x * _1py;
-		dN(13,0) =    _p3m9x2m2x * _1pyt1mz,
-		dN(13,1) =   _1mx2t1p3x * _1mz,
-		dN(13,2) =   -_1mx2t1p3x * _1py;
-		dN(14,0) =    _m3m9x2m2x * _1pyt1pz,
-		dN(14,1) =   _1mx2t1m3x * _1pz,
-		dN(14,2) =   _1mx2t1m3x * _1py;
-		dN(15,0) =    _p3m9x2m2x * _1pyt1pz,
-		dN(15,1) =   _1mx2t1p3x * _1pz,
-		dN(15,2) =   _1mx2t1p3x * _1py;
+		dN(8, 0) = _m3m9x2m2x * _1myt1mz,
+			  dN(8, 1) = -_1mx2t1m3x * _1mz,
+			  dN(8, 2) = -_1mx2t1m3x * _1my;
+		dN(9, 0) = _p3m9x2m2x * _1myt1mz,
+			  dN(9, 1) = -_1mx2t1p3x * _1mz,
+			  dN(9, 2) = -_1mx2t1p3x * _1my;
+		dN(10, 0) = _m3m9x2m2x * _1myt1pz,
+			   dN(10, 1) = -_1mx2t1m3x * _1pz,
+			   dN(10, 2) = _1mx2t1m3x * _1my;
+		dN(11, 0) = _p3m9x2m2x * _1myt1pz,
+			   dN(11, 1) = -_1mx2t1p3x * _1pz,
+			   dN(11, 2) = _1mx2t1p3x * _1my;
+		dN(12, 0) = _m3m9x2m2x * _1pyt1mz,
+			   dN(12, 1) = _1mx2t1m3x * _1mz,
+			   dN(12, 2) = -_1mx2t1m3x * _1py;
+		dN(13, 0) = _p3m9x2m2x * _1pyt1mz,
+			   dN(13, 1) = _1mx2t1p3x * _1mz,
+			   dN(13, 2) = -_1mx2t1p3x * _1py;
+		dN(14, 0) = _m3m9x2m2x * _1pyt1pz,
+			   dN(14, 1) = _1mx2t1m3x * _1pz,
+			   dN(14, 2) = _1mx2t1m3x * _1py;
+		dN(15, 0) = _p3m9x2m2x * _1pyt1pz,
+			   dN(15, 1) = _1mx2t1p3x * _1pz,
+			   dN(15, 2) = _1mx2t1p3x * _1py;
 
 		btScalar _m3m9y2m2y = -_3m9y2 - _2y;
-		btScalar _p3m9y2m2y =  _3m9y2 - _2y;
+		btScalar _p3m9y2m2y = _3m9y2 - _2y;
 		btScalar _1my2t1m3y = _1my2 * _1m3y;
 		btScalar _1my2t1p3y = _1my2 * _1p3y;
-		dN(16,0) =    -_1my2t1m3y * _1mz,
-		dN(16,1) =   _m3m9y2m2y * _1mxt1mz,
-		dN(16,2) =   -_1my2t1m3y * _1mx;
-		dN(17,0) =    -_1my2t1p3y * _1mz,
-		dN(17,1) =   _p3m9y2m2y * _1mxt1mz,
-		dN(17,2) =   -_1my2t1p3y * _1mx;
-		dN(18,0) =    _1my2t1m3y * _1mz,
-		dN(18,1) =   _m3m9y2m2y * _1pxt1mz,
-		dN(18,2) =   -_1my2t1m3y * _1px;
-		dN(19,0) =    _1my2t1p3y * _1mz,
-		dN(19,1) =   _p3m9y2m2y * _1pxt1mz,
-		dN(19,2) =   -_1my2t1p3y * _1px;
-		dN(20,0) =    -_1my2t1m3y * _1pz,
-		dN(20,1) =   _m3m9y2m2y * _1mxt1pz,
-		dN(20,2) =   _1my2t1m3y * _1mx;
-		dN(21,0) =    -_1my2t1p3y * _1pz,
-		dN(21,1) =   _p3m9y2m2y * _1mxt1pz,
-		dN(21,2) =   _1my2t1p3y * _1mx;
-		dN(22,0) =    _1my2t1m3y * _1pz,
-		dN(22,1) =   _m3m9y2m2y * _1pxt1pz,
-		dN(22,2) =   _1my2t1m3y * _1px;
-		dN(23,0) =    _1my2t1p3y * _1pz,
-		dN(23,1) =   _p3m9y2m2y * _1pxt1pz,
-		dN(23,2) =   _1my2t1p3y * _1px;
-
+		dN(16, 0) = -_1my2t1m3y * _1mz,
+			   dN(16, 1) = _m3m9y2m2y * _1mxt1mz,
+			   dN(16, 2) = -_1my2t1m3y * _1mx;
+		dN(17, 0) = -_1my2t1p3y * _1mz,
+			   dN(17, 1) = _p3m9y2m2y * _1mxt1mz,
+			   dN(17, 2) = -_1my2t1p3y * _1mx;
+		dN(18, 0) = _1my2t1m3y * _1mz,
+			   dN(18, 1) = _m3m9y2m2y * _1pxt1mz,
+			   dN(18, 2) = -_1my2t1m3y * _1px;
+		dN(19, 0) = _1my2t1p3y * _1mz,
+			   dN(19, 1) = _p3m9y2m2y * _1pxt1mz,
+			   dN(19, 2) = -_1my2t1p3y * _1px;
+		dN(20, 0) = -_1my2t1m3y * _1pz,
+			   dN(20, 1) = _m3m9y2m2y * _1mxt1pz,
+			   dN(20, 2) = _1my2t1m3y * _1mx;
+		dN(21, 0) = -_1my2t1p3y * _1pz,
+			   dN(21, 1) = _p3m9y2m2y * _1mxt1pz,
+			   dN(21, 2) = _1my2t1p3y * _1mx;
+		dN(22, 0) = _1my2t1m3y * _1pz,
+			   dN(22, 1) = _m3m9y2m2y * _1pxt1pz,
+			   dN(22, 2) = _1my2t1m3y * _1px;
+		dN(23, 0) = _1my2t1p3y * _1pz,
+			   dN(23, 1) = _p3m9y2m2y * _1pxt1pz,
+			   dN(23, 2) = _1my2t1p3y * _1px;
 
 		btScalar _m3m9z2m2z = -_3m9z2 - _2z;
-		btScalar _p3m9z2m2z =  _3m9z2 - _2z;
+		btScalar _p3m9z2m2z = _3m9z2 - _2z;
 		btScalar _1mz2t1m3z = _1mz2 * _1m3z;
 		btScalar _1mz2t1p3z = _1mz2 * _1p3z;
-		dN(24,0) =    -_1mz2t1m3z * _1my,
-		dN(24,1) =  -_1mz2t1m3z * _1mx,
-		dN(24,2) =   _m3m9z2m2z * _1mxt1my;
-		dN(25,0) =    -_1mz2t1p3z * _1my,
-		dN(25,1) =   -_1mz2t1p3z * _1mx,
-		dN(25,2) =   _p3m9z2m2z * _1mxt1my;
-		dN(26,0) =    -_1mz2t1m3z * _1py,
-		dN(26,1) =   _1mz2t1m3z * _1mx,
-		dN(26,2) =   _m3m9z2m2z * _1mxt1py;
-		dN(27,0) =    -_1mz2t1p3z * _1py,
-		dN(27,1) =   _1mz2t1p3z * _1mx,
-		dN(27,2) =   _p3m9z2m2z * _1mxt1py;
-		dN(28,0) =    _1mz2t1m3z * _1my,
-		dN(28,1) =   -_1mz2t1m3z * _1px,
-		dN(28,2) =   _m3m9z2m2z * _1pxt1my;
-		dN(29,0) =    _1mz2t1p3z * _1my,
-		dN(29,1) =   -_1mz2t1p3z * _1px,
-		dN(29,2) =   _p3m9z2m2z * _1pxt1my;
-		dN(30,0) =    _1mz2t1m3z * _1py,
-		dN(30,1) =   _1mz2t1m3z * _1px,
-		dN(30,2) =   _m3m9z2m2z * _1pxt1py;
-		dN(31,0) =    _1mz2t1p3z * _1py,
-		dN(31,1) =   _1mz2t1p3z * _1px,
-		dN(31,2) =   _p3m9z2m2z * _1pxt1py;
+		dN(24, 0) = -_1mz2t1m3z * _1my,
+			   dN(24, 1) = -_1mz2t1m3z * _1mx,
+			   dN(24, 2) = _m3m9z2m2z * _1mxt1my;
+		dN(25, 0) = -_1mz2t1p3z * _1my,
+			   dN(25, 1) = -_1mz2t1p3z * _1mx,
+			   dN(25, 2) = _p3m9z2m2z * _1mxt1my;
+		dN(26, 0) = -_1mz2t1m3z * _1py,
+			   dN(26, 1) = _1mz2t1m3z * _1mx,
+			   dN(26, 2) = _m3m9z2m2z * _1mxt1py;
+		dN(27, 0) = -_1mz2t1p3z * _1py,
+			   dN(27, 1) = _1mz2t1p3z * _1mx,
+			   dN(27, 2) = _p3m9z2m2z * _1mxt1py;
+		dN(28, 0) = _1mz2t1m3z * _1my,
+			   dN(28, 1) = -_1mz2t1m3z * _1px,
+			   dN(28, 2) = _m3m9z2m2z * _1pxt1my;
+		dN(29, 0) = _1mz2t1p3z * _1my,
+			   dN(29, 1) = -_1mz2t1p3z * _1px,
+			   dN(29, 2) = _p3m9z2m2z * _1pxt1my;
+		dN(30, 0) = _1mz2t1m3z * _1py,
+			   dN(30, 1) = _1mz2t1m3z * _1px,
+			   dN(30, 2) = _m3m9z2m2z * _1pxt1py;
+		dN(31, 0) = _1mz2t1p3z * _1py,
+			   dN(31, 1) = _1mz2t1p3z * _1px,
+			   dN(31, 2) = _p3m9z2m2z * _1pxt1py;
 
 		dN.bottomRowsMul(32u - 8u, 9.0 / 64.0);
-
 	}
 
 	return res;
 }
 
-
-
 bool btMiniSDF::interpolate(unsigned int field_id, double& dist, btVector3 const& x,
-	btVector3* gradient) const
+							btVector3* gradient) const
 {
 	btAssert(m_isValid);
 	if (!m_isValid)
@@ -459,15 +449,15 @@ bool btMiniSDF::interpolate(unsigned int field_id, double& dist, btVector3 const
 	if (!m_domain.contains(x))
 		return false;
 
-	btVector3 tmpmi = ((x - m_domain.min())*(m_inv_cell_size));//.cast<unsigned int>().eval();
-	unsigned int mi[3] = {(unsigned int )tmpmi[0],(unsigned int )tmpmi[1],(unsigned int )tmpmi[2]};
+	btVector3 tmpmi = ((x - m_domain.min()) * (m_inv_cell_size));  //.cast<unsigned int>().eval();
+	unsigned int mi[3] = {(unsigned int)tmpmi[0], (unsigned int)tmpmi[1], (unsigned int)tmpmi[2]};
 	if (mi[0] >= m_resolution[0])
-		mi[0] = m_resolution[0]-1;
+		mi[0] = m_resolution[0] - 1;
 	if (mi[1] >= m_resolution[1])
-		mi[1] = m_resolution[1]-1;
+		mi[1] = m_resolution[1] - 1;
 	if (mi[2] >= m_resolution[2])
-		mi[2] = m_resolution[2]-1;
-	btMultiIndex mui; 
+		mi[2] = m_resolution[2] - 1;
+	btMultiIndex mui;
 	mui.ijk[0] = mi[0];
 	mui.ijk[1] = mi[1];
 	mui.ijk[2] = mi[2];
@@ -478,12 +468,12 @@ bool btMiniSDF::interpolate(unsigned int field_id, double& dist, btVector3 const
 
 	btAlignedBox3d sd = subdomain(i);
 	i = i_;
-	btVector3 d = sd.m_max-sd.m_min;//.diagonal().eval();
+	btVector3 d = sd.m_max - sd.m_min;  //.diagonal().eval();
 
 	btVector3 denom = (sd.max() - sd.min());
-	btVector3 c0 = btVector3(2.0,2.0,2.0)/denom;
-	btVector3 c1 = (sd.max() + sd.min())/denom;
-	btVector3 xi = (c0*x - c1);
+	btVector3 c0 = btVector3(2.0, 2.0, 2.0) / denom;
+	btVector3 c1 = (sd.max() + sd.min()) / denom;
+	btVector3 xi = (c0 * x - c1);
 
 	btCell32 const& cell = m_cells[field_id][i];
 	if (!gradient)
@@ -497,7 +487,8 @@ bool btMiniSDF::interpolate(unsigned int field_id, double& dist, btVector3 const
 			double c = m_nodes[field_id][v];
 			if (c == DBL_MAX)
 			{
-				return false;;
+				return false;
+				;
 			}
 			phi += c * N[j];
 		}
@@ -529,4 +520,3 @@ bool btMiniSDF::interpolate(unsigned int field_id, double& dist, btVector3 const
 	dist = phi;
 	return true;
 }
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btMiniSDF.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btMiniSDF.h
index 3de90e4f8ad..b60fd102fdc 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btMiniSDF.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btMiniSDF.h
@@ -5,7 +5,6 @@
 #include "LinearMath/btAabbUtil2.h"
 #include "LinearMath/btAlignedObjectArray.h"
 
-
 struct btMultiIndex
 {
 	unsigned int ijk[3];
@@ -25,7 +24,6 @@ struct btAlignedBox3d
 	{
 		return m_max;
 	}
-	
 
 	bool contains(const btVector3& x) const
 	{
@@ -33,8 +31,8 @@ struct btAlignedBox3d
 	}
 
 	btAlignedBox3d(const btVector3& mn, const btVector3& mx)
-		:m_min(mn),
-		m_max(mx)
+		: m_min(mn),
+		  m_max(mx)
 	{
 	}
 
@@ -47,39 +45,38 @@ struct btShapeMatrix
 {
 	double m_vec[32];
 
-	inline double&       operator[](int i)
+	inline double& operator[](int i)
 	{
 		return m_vec[i];
 	}
 
-	inline const double&       operator[](int i) const
+	inline const double& operator[](int i) const
 	{
 		return m_vec[i];
 	}
-
 };
 
 struct btShapeGradients
 {
-	btVector3	m_vec[32];
+	btVector3 m_vec[32];
 
 	void topRowsDivide(int row, double denom)
 	{
-		for (int i=0;i<row;i++)
+		for (int i = 0; i < row; i++)
 		{
 			m_vec[i] /= denom;
 		}
 	}
-	
+
 	void bottomRowsMul(int row, double val)
 	{
-		for (int i=32-row;i<32;i++)
+		for (int i = 32 - row; i < 32; i++)
 		{
 			m_vec[i] *= val;
 		}
 	}
-	
-	inline btScalar&       operator()(int i, int j)
+
+	inline btScalar& operator()(int i, int j)
 	{
 		return m_vec[i][j];
 	}
@@ -92,7 +89,6 @@ struct btCell32
 
 struct btMiniSDF
 {
-
 	btAlignedBox3d m_domain;
 	unsigned int m_resolution[3];
 	btVector3 m_cell_size;
@@ -101,13 +97,12 @@ struct btMiniSDF
 	std::size_t m_n_fields;
 	bool m_isValid;
 
-
 	btAlignedObjectArray<btAlignedObjectArray<double> > m_nodes;
-	btAlignedObjectArray<btAlignedObjectArray<btCell32 > > m_cells;
+	btAlignedObjectArray<btAlignedObjectArray<btCell32> > m_cells;
 	btAlignedObjectArray<btAlignedObjectArray<unsigned int> > m_cell_map;
 
 	btMiniSDF()
-	:m_isValid(false)
+		: m_isValid(false)
 	{
 	}
 	bool load(const char* data, int size);
@@ -115,20 +110,18 @@ struct btMiniSDF
 	{
 		return m_isValid;
 	}
-	unsigned int multiToSingleIndex(btMultiIndex const & ijk) const;
-	
+	unsigned int multiToSingleIndex(btMultiIndex const& ijk) const;
+
 	btAlignedBox3d subdomain(btMultiIndex const& ijk) const;
 
 	btMultiIndex singleToMultiIndex(unsigned int l) const;
 
 	btAlignedBox3d subdomain(unsigned int l) const;
 
-
 	btShapeMatrix
 	shape_function_(btVector3 const& xi, btShapeGradients* gradient = 0) const;
 
 	bool interpolate(unsigned int field_id, double& dist, btVector3 const& x, btVector3* gradient) const;
 };
 
-
-#endif //MINISDF_H
+#endif  //MINISDF_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btMinkowskiSumShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btMinkowskiSumShape.cpp
index 899ef500564..d4b6a651de0 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btMinkowskiSumShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btMinkowskiSumShape.cpp
@@ -13,65 +13,59 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btMinkowskiSumShape.h"
 
-
-btMinkowskiSumShape::btMinkowskiSumShape(const btConvexShape* shapeA,const btConvexShape* shapeB)
-: btConvexInternalShape (),
-m_shapeA(shapeA),
-m_shapeB(shapeB)
+btMinkowskiSumShape::btMinkowskiSumShape(const btConvexShape* shapeA, const btConvexShape* shapeB)
+	: btConvexInternalShape(),
+	  m_shapeA(shapeA),
+	  m_shapeB(shapeB)
 {
 	m_shapeType = MINKOWSKI_DIFFERENCE_SHAPE_PROXYTYPE;
 	m_transA.setIdentity();
 	m_transB.setIdentity();
 }
 
-btVector3 btMinkowskiSumShape::localGetSupportingVertexWithoutMargin(const btVector3& vec)const
+btVector3 btMinkowskiSumShape::localGetSupportingVertexWithoutMargin(const btVector3& vec) const
 {
-	btVector3 supVertexA = m_transA(m_shapeA->localGetSupportingVertexWithoutMargin(vec*m_transA.getBasis()));
-	btVector3 supVertexB = m_transB(m_shapeB->localGetSupportingVertexWithoutMargin(-vec*m_transB.getBasis()));
-	return  supVertexA - supVertexB;
+	btVector3 supVertexA = m_transA(m_shapeA->localGetSupportingVertexWithoutMargin(vec * m_transA.getBasis()));
+	btVector3 supVertexB = m_transB(m_shapeB->localGetSupportingVertexWithoutMargin(-vec * m_transB.getBasis()));
+	return supVertexA - supVertexB;
 }
 
-void	btMinkowskiSumShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const
+void btMinkowskiSumShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const
 {
 	///@todo: could make recursive use of batching. probably this shape is not used frequently.
-	for (int i=0;i<numVectors;i++)
+	for (int i = 0; i < numVectors; i++)
 	{
 		supportVerticesOut[i] = localGetSupportingVertexWithoutMargin(vectors[i]);
 	}
-
 }
 
-
-
-btScalar	btMinkowskiSumShape::getMargin() const
+btScalar btMinkowskiSumShape::getMargin() const
 {
 	return m_shapeA->getMargin() + m_shapeB->getMargin();
 }
 
-
-void	btMinkowskiSumShape::calculateLocalInertia(btScalar mass,btVector3& inertia) const
+void btMinkowskiSumShape::calculateLocalInertia(btScalar mass, btVector3& inertia) const
 {
 	(void)mass;
 	//inertia of the AABB of the Minkowski sum
 	btTransform identity;
 	identity.setIdentity();
-	btVector3 aabbMin,aabbMax;
-	getAabb(identity,aabbMin,aabbMax);
+	btVector3 aabbMin, aabbMax;
+	getAabb(identity, aabbMin, aabbMax);
 
-	btVector3 halfExtents = (aabbMax-aabbMin)*btScalar(0.5);
+	btVector3 halfExtents = (aabbMax - aabbMin) * btScalar(0.5);
 
 	btScalar margin = getMargin();
 
-	btScalar lx=btScalar(2.)*(halfExtents.x()+margin);
-	btScalar ly=btScalar(2.)*(halfExtents.y()+margin);
-	btScalar lz=btScalar(2.)*(halfExtents.z()+margin);
-	const btScalar x2 = lx*lx;
-	const btScalar y2 = ly*ly;
-	const btScalar z2 = lz*lz;
+	btScalar lx = btScalar(2.) * (halfExtents.x() + margin);
+	btScalar ly = btScalar(2.) * (halfExtents.y() + margin);
+	btScalar lz = btScalar(2.) * (halfExtents.z() + margin);
+	const btScalar x2 = lx * lx;
+	const btScalar y2 = ly * ly;
+	const btScalar z2 = lz * lz;
 	const btScalar scaledmass = mass * btScalar(0.08333333);
 
-	inertia = scaledmass * (btVector3(y2+z2,x2+z2,x2+y2));
+	inertia = scaledmass * (btVector3(y2 + z2, x2 + z2, x2 + y2));
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btMinkowskiSumShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btMinkowskiSumShape.h
index a3f9a472394..3b5150f6d50 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btMinkowskiSumShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btMinkowskiSumShape.h
@@ -17,46 +17,43 @@ subject to the following restrictions:
 #define BT_MINKOWSKI_SUM_SHAPE_H
 
 #include "btConvexInternalShape.h"
-#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h" // for the types
+#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h"  // for the types
 
 /// The btMinkowskiSumShape is only for advanced users. This shape represents implicit based minkowski sum of two convex implicit shapes.
-ATTRIBUTE_ALIGNED16(class) btMinkowskiSumShape : public btConvexInternalShape
+ATTRIBUTE_ALIGNED16(class)
+btMinkowskiSumShape : public btConvexInternalShape
 {
-
-	btTransform	m_transA;
-	btTransform	m_transB;
-	const btConvexShape*	m_shapeA;
-	const btConvexShape*	m_shapeB;
+	btTransform m_transA;
+	btTransform m_transB;
+	const btConvexShape* m_shapeA;
+	const btConvexShape* m_shapeB;
 
 public:
+	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-BT_DECLARE_ALIGNED_ALLOCATOR();
+	btMinkowskiSumShape(const btConvexShape* shapeA, const btConvexShape* shapeB);
 
-	btMinkowskiSumShape(const btConvexShape* shapeA,const btConvexShape* shapeB);
+	virtual btVector3 localGetSupportingVertexWithoutMargin(const btVector3& vec) const;
 
-	virtual btVector3	localGetSupportingVertexWithoutMargin(const btVector3& vec)const;
+	virtual void batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const;
 
-	virtual void	batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const;
+	virtual void calculateLocalInertia(btScalar mass, btVector3 & inertia) const;
 
+	void setTransformA(const btTransform& transA) { m_transA = transA; }
+	void setTransformB(const btTransform& transB) { m_transB = transB; }
 
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
+	const btTransform& getTransformA() const { return m_transA; }
+	const btTransform& GetTransformB() const { return m_transB; }
 
-	void	setTransformA(const btTransform&	transA) { m_transA = transA;}
-	void	setTransformB(const btTransform&	transB) { m_transB = transB;}
+	virtual btScalar getMargin() const;
 
-	const btTransform& getTransformA()const  { return m_transA;}
-	const btTransform& GetTransformB()const  { return m_transB;}
+	const btConvexShape* getShapeA() const { return m_shapeA; }
+	const btConvexShape* getShapeB() const { return m_shapeB; }
 
-
-	virtual btScalar	getMargin() const;
-
-	const btConvexShape*	getShapeA() const { return m_shapeA;}
-	const btConvexShape*	getShapeB() const { return m_shapeB;}
-
-	virtual const char*	getName()const 
+	virtual const char* getName() const
 	{
 		return "MinkowskiSum";
 	}
 };
 
-#endif //BT_MINKOWSKI_SUM_SHAPE_H
+#endif  //BT_MINKOWSKI_SUM_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btMultiSphereShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btMultiSphereShape.cpp
index 4195fa31388..c0cc55dfb08 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btMultiSphereShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btMultiSphereShape.cpp
@@ -13,7 +13,7 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-#if defined (_WIN32) || defined (__i386__)
+#if defined(_WIN32) || defined(__i386__)
 #define BT_USE_SSE_IN_API
 #endif
 
@@ -22,43 +22,41 @@ subject to the following restrictions:
 #include "LinearMath/btQuaternion.h"
 #include "LinearMath/btSerializer.h"
 
-btMultiSphereShape::btMultiSphereShape (const btVector3* positions,const btScalar* radi,int numSpheres)
-:btConvexInternalAabbCachingShape ()
+btMultiSphereShape::btMultiSphereShape(const btVector3* positions, const btScalar* radi, int numSpheres)
+	: btConvexInternalAabbCachingShape()
 {
 	m_shapeType = MULTI_SPHERE_SHAPE_PROXYTYPE;
 	//btScalar startMargin = btScalar(BT_LARGE_FLOAT);
 
 	m_localPositionArray.resize(numSpheres);
 	m_radiArray.resize(numSpheres);
-	for (int i=0;i<numSpheres;i++)
+	for (int i = 0; i < numSpheres; i++)
 	{
 		m_localPositionArray[i] = positions[i];
 		m_radiArray[i] = radi[i];
-		
 	}
 
 	recalcLocalAabb();
-
 }
 
 #ifndef MIN
-	#define MIN( _a, _b)    ((_a) < (_b) ? (_a) : (_b))
+#define MIN(_a, _b) ((_a) < (_b) ? (_a) : (_b))
 #endif
- btVector3	btMultiSphereShape::localGetSupportingVertexWithoutMargin(const btVector3& vec0)const
+btVector3 btMultiSphereShape::localGetSupportingVertexWithoutMargin(const btVector3& vec0) const
 {
-	btVector3 supVec(0,0,0);
+	btVector3 supVec(0, 0, 0);
 
 	btScalar maxDot(btScalar(-BT_LARGE_FLOAT));
 
-
 	btVector3 vec = vec0;
 	btScalar lenSqr = vec.length2();
-	if (lenSqr < (SIMD_EPSILON*SIMD_EPSILON))
+	if (lenSqr < (SIMD_EPSILON * SIMD_EPSILON))
 	{
-		vec.setValue(1,0,0);
-	} else
+		vec.setValue(1, 0, 0);
+	}
+	else
 	{
-		btScalar rlen = btScalar(1.) / btSqrt(lenSqr );
+		btScalar rlen = btScalar(1.) / btSqrt(lenSqr);
 		vec *= rlen;
 	}
 
@@ -69,32 +67,30 @@ btMultiSphereShape::btMultiSphereShape (const btVector3* positions,const btScala
 	const btScalar* rad = &m_radiArray[0];
 	int numSpheres = m_localPositionArray.size();
 
-	for( int k = 0; k < numSpheres; k+= 128 )
+	for (int k = 0; k < numSpheres; k += 128)
 	{
 		btVector3 temp[128];
-		int inner_count = MIN( numSpheres - k, 128 );
-        for( long i = 0; i < inner_count; i++ )
-        {
-            temp[i] = (*pos)*m_localScaling +vec*m_localScaling*(*rad) - vec * getMargin();
-            pos++;
-            rad++;
-        }
-        long i = vec.maxDot( temp, inner_count, newDot);
-        if( newDot > maxDot )
+		int inner_count = MIN(numSpheres - k, 128);
+		for (long i = 0; i < inner_count; i++)
+		{
+			temp[i] = (*pos) * m_localScaling + vec * m_localScaling * (*rad) - vec * getMargin();
+			pos++;
+			rad++;
+		}
+		long i = vec.maxDot(temp, inner_count, newDot);
+		if (newDot > maxDot)
 		{
 			maxDot = newDot;
 			supVec = temp[i];
 		}
-    }
+	}
 
 	return supVec;
-
 }
 
- void	btMultiSphereShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const
+void btMultiSphereShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const
 {
-
-	for (int j=0;j<numVectors;j++)
+	for (int j = 0; j < numVectors; j++)
 	{
 		btScalar maxDot(btScalar(-BT_LARGE_FLOAT));
 
@@ -107,73 +103,63 @@ btMultiSphereShape::btMultiSphereShape (const btVector3* positions,const btScala
 		const btScalar* rad = &m_radiArray[0];
 		int numSpheres = m_localPositionArray.size();
 
-        for( int k = 0; k < numSpheres; k+= 128 )
-        {
-            btVector3 temp[128];
-            int inner_count = MIN( numSpheres - k, 128 );
-            for( long i = 0; i < inner_count; i++ )
-            {
-                temp[i] = (*pos)*m_localScaling +vec*m_localScaling*(*rad) - vec * getMargin();
-                pos++;
-                rad++;
-            }
-            long i = vec.maxDot( temp, inner_count, newDot);
-            if( newDot > maxDot )
-            {
-                maxDot = newDot;
-                supportVerticesOut[j] = temp[i];
-            }
-        }
-        
+		for (int k = 0; k < numSpheres; k += 128)
+		{
+			btVector3 temp[128];
+			int inner_count = MIN(numSpheres - k, 128);
+			for (long i = 0; i < inner_count; i++)
+			{
+				temp[i] = (*pos) * m_localScaling + vec * m_localScaling * (*rad) - vec * getMargin();
+				pos++;
+				rad++;
+			}
+			long i = vec.maxDot(temp, inner_count, newDot);
+			if (newDot > maxDot)
+			{
+				maxDot = newDot;
+				supportVerticesOut[j] = temp[i];
+			}
+		}
 	}
 }
 
-
-
-
-
-
-
-
-void	btMultiSphereShape::calculateLocalInertia(btScalar mass,btVector3& inertia) const
+void btMultiSphereShape::calculateLocalInertia(btScalar mass, btVector3& inertia) const
 {
 	//as an approximation, take the inertia of the box that bounds the spheres
 
-	btVector3 localAabbMin,localAabbMax;
-	getCachedLocalAabb(localAabbMin,localAabbMax);
-	btVector3 halfExtents = (localAabbMax-localAabbMin)*btScalar(0.5);
+	btVector3 localAabbMin, localAabbMax;
+	getCachedLocalAabb(localAabbMin, localAabbMax);
+	btVector3 halfExtents = (localAabbMax - localAabbMin) * btScalar(0.5);
 
-	btScalar lx=btScalar(2.)*(halfExtents.x());
-	btScalar ly=btScalar(2.)*(halfExtents.y());
-	btScalar lz=btScalar(2.)*(halfExtents.z());
-
-	inertia.setValue(mass/(btScalar(12.0)) * (ly*ly + lz*lz),
-					mass/(btScalar(12.0)) * (lx*lx + lz*lz),
-					mass/(btScalar(12.0)) * (lx*lx + ly*ly));
+	btScalar lx = btScalar(2.) * (halfExtents.x());
+	btScalar ly = btScalar(2.) * (halfExtents.y());
+	btScalar lz = btScalar(2.) * (halfExtents.z());
 
+	inertia.setValue(mass / (btScalar(12.0)) * (ly * ly + lz * lz),
+					 mass / (btScalar(12.0)) * (lx * lx + lz * lz),
+					 mass / (btScalar(12.0)) * (lx * lx + ly * ly));
 }
 
-
 ///fills the dataBuffer and returns the struct name (and 0 on failure)
-const char*	btMultiSphereShape::serialize(void* dataBuffer, btSerializer* serializer) const
+const char* btMultiSphereShape::serialize(void* dataBuffer, btSerializer* serializer) const
 {
-	btMultiSphereShapeData* shapeData = (btMultiSphereShapeData*) dataBuffer;
+	btMultiSphereShapeData* shapeData = (btMultiSphereShapeData*)dataBuffer;
 	btConvexInternalShape::serialize(&shapeData->m_convexInternalShapeData, serializer);
 
 	int numElem = m_localPositionArray.size();
-	shapeData->m_localPositionArrayPtr = numElem ? (btPositionAndRadius*)serializer->getUniquePointer((void*)&m_localPositionArray[0]):  0;
-	
+	shapeData->m_localPositionArrayPtr = numElem ? (btPositionAndRadius*)serializer->getUniquePointer((void*)&m_localPositionArray[0]) : 0;
+
 	shapeData->m_localPositionArraySize = numElem;
 	if (numElem)
 	{
-		btChunk* chunk = serializer->allocate(sizeof(btPositionAndRadius),numElem);
+		btChunk* chunk = serializer->allocate(sizeof(btPositionAndRadius), numElem);
 		btPositionAndRadius* memPtr = (btPositionAndRadius*)chunk->m_oldPtr;
-		for (int i=0;i<numElem;i++,memPtr++)
+		for (int i = 0; i < numElem; i++, memPtr++)
 		{
 			m_localPositionArray[i].serializeFloat(memPtr->m_pos);
 			memPtr->m_radius = float(m_radiArray[i]);
 		}
-		serializer->finalizeChunk(chunk,"btPositionAndRadius",BT_ARRAY_CODE,(void*)&m_localPositionArray[0]);
+		serializer->finalizeChunk(chunk, "btPositionAndRadius", BT_ARRAY_CODE, (void*)&m_localPositionArray[0]);
 	}
 
 	// Fill padding with zeros to appease msan.
@@ -181,5 +167,3 @@ const char*	btMultiSphereShape::serialize(void* dataBuffer, btSerializer* serial
 
 	return "btMultiSphereShapeData";
 }
-
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btMultiSphereShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btMultiSphereShape.h
index 5d3b4026848..2d79c07ca42 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btMultiSphereShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btMultiSphereShape.h
@@ -17,69 +17,65 @@ subject to the following restrictions:
 #define BT_MULTI_SPHERE_MINKOWSKI_H
 
 #include "btConvexInternalShape.h"
-#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h" // for the types
+#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h"  // for the types
 #include "LinearMath/btAlignedObjectArray.h"
 #include "LinearMath/btAabbUtil2.h"
 
-
-
 ///The btMultiSphereShape represents the convex hull of a collection of spheres. You can create special capsules or other smooth volumes.
 ///It is possible to animate the spheres for deformation, but call 'recalcLocalAabb' after changing any sphere position/radius
-ATTRIBUTE_ALIGNED16(class) btMultiSphereShape : public btConvexInternalAabbCachingShape
+ATTRIBUTE_ALIGNED16(class)
+btMultiSphereShape : public btConvexInternalAabbCachingShape
 {
-	
 	btAlignedObjectArray<btVector3> m_localPositionArray;
-	btAlignedObjectArray<btScalar>  m_radiArray;
-	
+	btAlignedObjectArray<btScalar> m_radiArray;
+
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
-	btMultiSphereShape (const btVector3* positions,const btScalar* radi,int numSpheres);
+
+	btMultiSphereShape(const btVector3* positions, const btScalar* radi, int numSpheres);
 
 	///CollisionShape Interface
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
+	virtual void calculateLocalInertia(btScalar mass, btVector3 & inertia) const;
 
 	/// btConvexShape Interface
-	virtual btVector3	localGetSupportingVertexWithoutMargin(const btVector3& vec)const;
+	virtual btVector3 localGetSupportingVertexWithoutMargin(const btVector3& vec) const;
 
-	virtual void	batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const;
-	
-	int	getSphereCount() const
+	virtual void batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const;
+
+	int getSphereCount() const
 	{
 		return m_localPositionArray.size();
 	}
 
-	const btVector3&	getSpherePosition(int index) const
+	const btVector3& getSpherePosition(int index) const
 	{
 		return m_localPositionArray[index];
 	}
 
-	btScalar	getSphereRadius(int index) const
+	btScalar getSphereRadius(int index) const
 	{
 		return m_radiArray[index];
 	}
 
-
-	virtual const char*	getName()const 
+	virtual const char* getName() const
 	{
 		return "MultiSphere";
 	}
 
-	virtual	int	calculateSerializeBufferSize() const;
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
-
-
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 };
 
-
-struct	btPositionAndRadius
+struct btPositionAndRadius
 {
-	btVector3FloatData	m_pos;
-	float		m_radius;
+	btVector3FloatData m_pos;
+	float m_radius;
 };
 
+// clang-format off
+
 struct	btMultiSphereShapeData
 {
 	btConvexInternalShapeData	m_convexInternalShapeData;
@@ -89,13 +85,11 @@ struct	btMultiSphereShapeData
 	char	m_padding[4];
 };
 
+// clang-format on
 
-
-SIMD_FORCE_INLINE	int	btMultiSphereShape::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btMultiSphereShape::calculateSerializeBufferSize() const
 {
 	return sizeof(btMultiSphereShapeData);
 }
 
-
-
-#endif //BT_MULTI_SPHERE_MINKOWSKI_H
+#endif  //BT_MULTI_SPHERE_MINKOWSKI_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btMultimaterialTriangleMeshShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btMultimaterialTriangleMeshShape.cpp
index 58799ac96ad..30108c9e7b4 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btMultimaterialTriangleMeshShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btMultimaterialTriangleMeshShape.cpp
@@ -19,27 +19,25 @@ subject to the following restrictions:
 #include "BulletCollision/CollisionShapes/btTriangleIndexVertexMaterialArray.h"
 //#include "BulletCollision/CollisionShapes/btOptimizedBvh.h"
 
-
 ///Obtains the material for a specific triangle
-const btMaterial * btMultimaterialTriangleMeshShape::getMaterialProperties(int partID, int triIndex)
+const btMaterial *btMultimaterialTriangleMeshShape::getMaterialProperties(int partID, int triIndex)
 {
-    const unsigned char * materialBase = 0;
-    int numMaterials;
-    PHY_ScalarType materialType;
-    int materialStride;
-    const unsigned char * triangleMaterialBase = 0;
-    int numTriangles;
-    int triangleMaterialStride;
-    PHY_ScalarType triangleType;
+	const unsigned char *materialBase = 0;
+	int numMaterials;
+	PHY_ScalarType materialType;
+	int materialStride;
+	const unsigned char *triangleMaterialBase = 0;
+	int numTriangles;
+	int triangleMaterialStride;
+	PHY_ScalarType triangleType;
 
-    ((btTriangleIndexVertexMaterialArray*)m_meshInterface)->getLockedReadOnlyMaterialBase(&materialBase, numMaterials, materialType, materialStride,
-        &triangleMaterialBase, numTriangles, triangleMaterialStride, triangleType, partID);
+	((btTriangleIndexVertexMaterialArray *)m_meshInterface)->getLockedReadOnlyMaterialBase(&materialBase, numMaterials, materialType, materialStride, &triangleMaterialBase, numTriangles, triangleMaterialStride, triangleType, partID);
 
-    // return the pointer to the place with the friction for the triangle
-    // TODO: This depends on whether it's a moving mesh or not
-    // BUG IN GIMPACT
-    //return (btScalar*)(&materialBase[triangleMaterialBase[(triIndex-1) * triangleMaterialStride] * materialStride]);
-    int * matInd = (int *)(&(triangleMaterialBase[(triIndex * triangleMaterialStride)]));
-    btMaterial *matVal = (btMaterial *)(&(materialBase[*matInd * materialStride]));
-    return (matVal);
+	// return the pointer to the place with the friction for the triangle
+	// TODO: This depends on whether it's a moving mesh or not
+	// BUG IN GIMPACT
+	//return (btScalar*)(&materialBase[triangleMaterialBase[(triIndex-1) * triangleMaterialStride] * materialStride]);
+	int *matInd = (int *)(&(triangleMaterialBase[(triIndex * triangleMaterialStride)]));
+	btMaterial *matVal = (btMaterial *)(&(materialBase[*matInd * materialStride]));
+	return (matVal);
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btMultimaterialTriangleMeshShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btMultimaterialTriangleMeshShape.h
index 5ebaede4a88..d1d42f8e049 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btMultimaterialTriangleMeshShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btMultimaterialTriangleMeshShape.h
@@ -22,82 +22,80 @@ subject to the following restrictions:
 #include "btMaterial.h"
 
 ///The BvhTriangleMaterialMeshShape extends the btBvhTriangleMeshShape. Its main contribution is the interface into a material array, which allows per-triangle friction and restitution.
-ATTRIBUTE_ALIGNED16(class) btMultimaterialTriangleMeshShape : public btBvhTriangleMeshShape
+ATTRIBUTE_ALIGNED16(class)
+btMultimaterialTriangleMeshShape : public btBvhTriangleMeshShape
 {
-    btAlignedObjectArray <btMaterial*> m_materialList;
+	btAlignedObjectArray<btMaterial *> m_materialList;
 
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-    btMultimaterialTriangleMeshShape(btStridingMeshInterface* meshInterface, bool useQuantizedAabbCompression, bool buildBvh = true):
-        btBvhTriangleMeshShape(meshInterface, useQuantizedAabbCompression, buildBvh)
-        {
-            m_shapeType = MULTIMATERIAL_TRIANGLE_MESH_PROXYTYPE;
+	btMultimaterialTriangleMeshShape(btStridingMeshInterface * meshInterface, bool useQuantizedAabbCompression, bool buildBvh = true) : btBvhTriangleMeshShape(meshInterface, useQuantizedAabbCompression, buildBvh)
+	{
+		m_shapeType = MULTIMATERIAL_TRIANGLE_MESH_PROXYTYPE;
 
-            const unsigned char *vertexbase;
-            int numverts;
-            PHY_ScalarType type;
-            int stride;
-            const unsigned char *indexbase;
-            int indexstride;
-            int numfaces;
-            PHY_ScalarType indicestype;
+		const unsigned char *vertexbase;
+		int numverts;
+		PHY_ScalarType type;
+		int stride;
+		const unsigned char *indexbase;
+		int indexstride;
+		int numfaces;
+		PHY_ScalarType indicestype;
 
-            //m_materialLookup = (int**)(btAlignedAlloc(sizeof(int*) * meshInterface->getNumSubParts(), 16));
+		//m_materialLookup = (int**)(btAlignedAlloc(sizeof(int*) * meshInterface->getNumSubParts(), 16));
 
-            for(int i = 0; i < meshInterface->getNumSubParts(); i++)
-            {
-                m_meshInterface->getLockedReadOnlyVertexIndexBase(
-                    &vertexbase,
-                    numverts,
-                    type,
-                    stride,
-                    &indexbase,
-                    indexstride,
-                    numfaces,
-                    indicestype,
-                    i);
-                //m_materialLookup[i] = (int*)(btAlignedAlloc(sizeof(int) * numfaces, 16));
-            }
-        }
+		for (int i = 0; i < meshInterface->getNumSubParts(); i++)
+		{
+			m_meshInterface->getLockedReadOnlyVertexIndexBase(
+				&vertexbase,
+				numverts,
+				type,
+				stride,
+				&indexbase,
+				indexstride,
+				numfaces,
+				indicestype,
+				i);
+			//m_materialLookup[i] = (int*)(btAlignedAlloc(sizeof(int) * numfaces, 16));
+		}
+	}
 
 	///optionally pass in a larger bvh aabb, used for quantization. This allows for deformations within this aabb
-	btMultimaterialTriangleMeshShape(btStridingMeshInterface* meshInterface, bool useQuantizedAabbCompression,const btVector3& bvhAabbMin,const btVector3& bvhAabbMax, bool buildBvh = true):
-        btBvhTriangleMeshShape(meshInterface, useQuantizedAabbCompression, bvhAabbMin, bvhAabbMax, buildBvh)
-        {
-            m_shapeType = MULTIMATERIAL_TRIANGLE_MESH_PROXYTYPE;
+	btMultimaterialTriangleMeshShape(btStridingMeshInterface * meshInterface, bool useQuantizedAabbCompression, const btVector3 &bvhAabbMin, const btVector3 &bvhAabbMax, bool buildBvh = true) : btBvhTriangleMeshShape(meshInterface, useQuantizedAabbCompression, bvhAabbMin, bvhAabbMax, buildBvh)
+	{
+		m_shapeType = MULTIMATERIAL_TRIANGLE_MESH_PROXYTYPE;
 
-            const unsigned char *vertexbase;
-            int numverts;
-            PHY_ScalarType type;
-            int stride;
-            const unsigned char *indexbase;
-            int indexstride;
-            int numfaces;
-            PHY_ScalarType indicestype;
+		const unsigned char *vertexbase;
+		int numverts;
+		PHY_ScalarType type;
+		int stride;
+		const unsigned char *indexbase;
+		int indexstride;
+		int numfaces;
+		PHY_ScalarType indicestype;
 
-            //m_materialLookup = (int**)(btAlignedAlloc(sizeof(int*) * meshInterface->getNumSubParts(), 16));
+		//m_materialLookup = (int**)(btAlignedAlloc(sizeof(int*) * meshInterface->getNumSubParts(), 16));
 
-            for(int i = 0; i < meshInterface->getNumSubParts(); i++)
-            {
-                m_meshInterface->getLockedReadOnlyVertexIndexBase(
-                    &vertexbase,
-                    numverts,
-                    type,
-                    stride,
-                    &indexbase,
-                    indexstride,
-                    numfaces,
-                    indicestype,
-                    i);
-                //m_materialLookup[i] = (int*)(btAlignedAlloc(sizeof(int) * numfaces * 2, 16));
-            }
-        }
-	
-    virtual ~btMultimaterialTriangleMeshShape()
-    {
-/*
+		for (int i = 0; i < meshInterface->getNumSubParts(); i++)
+		{
+			m_meshInterface->getLockedReadOnlyVertexIndexBase(
+				&vertexbase,
+				numverts,
+				type,
+				stride,
+				&indexbase,
+				indexstride,
+				numfaces,
+				indicestype,
+				i);
+			//m_materialLookup[i] = (int*)(btAlignedAlloc(sizeof(int) * numfaces * 2, 16));
+		}
+	}
+
+	virtual ~btMultimaterialTriangleMeshShape()
+	{
+		/*
         for(int i = 0; i < m_meshInterface->getNumSubParts(); i++)
         {
             btAlignedFree(m_materialValues[i]);
@@ -106,14 +104,12 @@ public:
         btAlignedFree(m_materialValues);
         m_materialLookup = NULL;
 */
-    }
+	}
 	//debugging
-	virtual const char*	getName()const {return "MULTIMATERIALTRIANGLEMESH";}
+	virtual const char *getName() const { return "MULTIMATERIALTRIANGLEMESH"; }
 
-    ///Obtains the material for a specific triangle
-    const btMaterial * getMaterialProperties(int partID, int triIndex);
+	///Obtains the material for a specific triangle
+	const btMaterial *getMaterialProperties(int partID, int triIndex);
+};
 
-}
-;
-
-#endif //BT_BVH_TRIANGLE_MATERIAL_MESH_SHAPE_H
+#endif  //BT_BVH_TRIANGLE_MATERIAL_MESH_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btOptimizedBvh.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btOptimizedBvh.cpp
index 6f36775f7c9..687399e0a97 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btOptimizedBvh.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btOptimizedBvh.cpp
@@ -13,51 +13,46 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btOptimizedBvh.h"
 #include "btStridingMeshInterface.h"
 #include "LinearMath/btAabbUtil2.h"
 #include "LinearMath/btIDebugDraw.h"
 
-
 btOptimizedBvh::btOptimizedBvh()
-{ 
+{
 }
 
 btOptimizedBvh::~btOptimizedBvh()
 {
 }
 
-
 void btOptimizedBvh::build(btStridingMeshInterface* triangles, bool useQuantizedAabbCompression, const btVector3& bvhAabbMin, const btVector3& bvhAabbMax)
 {
 	m_useQuantization = useQuantizedAabbCompression;
 
-
 	// NodeArray	triangleNodes;
 
-	struct	NodeTriangleCallback : public btInternalTriangleIndexCallback
+	struct NodeTriangleCallback : public btInternalTriangleIndexCallback
 	{
-
-		NodeArray&	m_triangleNodes;
+		NodeArray& m_triangleNodes;
 
 		NodeTriangleCallback& operator=(NodeTriangleCallback& other)
 		{
 			m_triangleNodes.copyFromArray(other.m_triangleNodes);
 			return *this;
 		}
-		
-		NodeTriangleCallback(NodeArray&	triangleNodes)
-			:m_triangleNodes(triangleNodes)
+
+		NodeTriangleCallback(NodeArray& triangleNodes)
+			: m_triangleNodes(triangleNodes)
 		{
 		}
 
-		virtual void internalProcessTriangleIndex(btVector3* triangle,int partId,int  triangleIndex)
+		virtual void internalProcessTriangleIndex(btVector3* triangle, int partId, int triangleIndex)
 		{
 			btOptimizedBvhNode node;
-			btVector3	aabbMin,aabbMax;
-			aabbMin.setValue(btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT));
-			aabbMax.setValue(btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT)); 
+			btVector3 aabbMin, aabbMax;
+			aabbMin.setValue(btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT));
+			aabbMax.setValue(btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT));
 			aabbMin.setMin(triangle[0]);
 			aabbMax.setMax(triangle[0]);
 			aabbMin.setMin(triangle[1]);
@@ -70,17 +65,17 @@ void btOptimizedBvh::build(btStridingMeshInterface* triangles, bool useQuantized
 			node.m_aabbMaxOrg = aabbMax;
 
 			node.m_escapeIndex = -1;
-	
+
 			//for child nodes
 			node.m_subPart = partId;
 			node.m_triangleIndex = triangleIndex;
 			m_triangleNodes.push_back(node);
 		}
 	};
-	struct	QuantizedNodeTriangleCallback : public btInternalTriangleIndexCallback
+	struct QuantizedNodeTriangleCallback : public btInternalTriangleIndexCallback
 	{
-		QuantizedNodeArray&	m_triangleNodes;
-		const btQuantizedBvh* m_optimizedTree; // for quantization
+		QuantizedNodeArray& m_triangleNodes;
+		const btQuantizedBvh* m_optimizedTree;  // for quantization
 
 		QuantizedNodeTriangleCallback& operator=(QuantizedNodeTriangleCallback& other)
 		{
@@ -89,23 +84,23 @@ void btOptimizedBvh::build(btStridingMeshInterface* triangles, bool useQuantized
 			return *this;
 		}
 
-		QuantizedNodeTriangleCallback(QuantizedNodeArray&	triangleNodes,const btQuantizedBvh* tree)
-			:m_triangleNodes(triangleNodes),m_optimizedTree(tree)
+		QuantizedNodeTriangleCallback(QuantizedNodeArray& triangleNodes, const btQuantizedBvh* tree)
+			: m_triangleNodes(triangleNodes), m_optimizedTree(tree)
 		{
 		}
 
-		virtual void internalProcessTriangleIndex(btVector3* triangle,int partId,int  triangleIndex)
+		virtual void internalProcessTriangleIndex(btVector3* triangle, int partId, int triangleIndex)
 		{
 			// The partId and triangle index must fit in the same (positive) integer
-			btAssert(partId < (1<<MAX_NUM_PARTS_IN_BITS));
-			btAssert(triangleIndex < (1<<(31-MAX_NUM_PARTS_IN_BITS)));
+			btAssert(partId < (1 << MAX_NUM_PARTS_IN_BITS));
+			btAssert(triangleIndex < (1 << (31 - MAX_NUM_PARTS_IN_BITS)));
 			//negative indices are reserved for escapeIndex
-			btAssert(triangleIndex>=0);
+			btAssert(triangleIndex >= 0);
 
 			btQuantizedBvhNode node;
-			btVector3	aabbMin,aabbMax;
-			aabbMin.setValue(btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT));
-			aabbMax.setValue(btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT)); 
+			btVector3 aabbMin, aabbMax;
+			aabbMin.setValue(btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT));
+			aabbMax.setValue(btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT));
 			aabbMin.setMin(triangle[0]);
 			aabbMax.setMax(triangle[0]);
 			aabbMin.setMin(triangle[1]);
@@ -132,59 +127,52 @@ void btOptimizedBvh::build(btStridingMeshInterface* triangles, bool useQuantized
 				aabbMin.setZ(aabbMin.z() - MIN_AABB_HALF_DIMENSION);
 			}
 
-			m_optimizedTree->quantize(&node.m_quantizedAabbMin[0],aabbMin,0);
-			m_optimizedTree->quantize(&node.m_quantizedAabbMax[0],aabbMax,1);
+			m_optimizedTree->quantize(&node.m_quantizedAabbMin[0], aabbMin, 0);
+			m_optimizedTree->quantize(&node.m_quantizedAabbMax[0], aabbMax, 1);
 
-			node.m_escapeIndexOrTriangleIndex = (partId<<(31-MAX_NUM_PARTS_IN_BITS)) | triangleIndex;
+			node.m_escapeIndexOrTriangleIndex = (partId << (31 - MAX_NUM_PARTS_IN_BITS)) | triangleIndex;
 
 			m_triangleNodes.push_back(node);
 		}
 	};
-	
-
 
 	int numLeafNodes = 0;
 
-	
 	if (m_useQuantization)
 	{
-
 		//initialize quantization values
-		setQuantizationValues(bvhAabbMin,bvhAabbMax);
+		setQuantizationValues(bvhAabbMin, bvhAabbMax);
 
-		QuantizedNodeTriangleCallback	callback(m_quantizedLeafNodes,this);
+		QuantizedNodeTriangleCallback callback(m_quantizedLeafNodes, this);
 
-	
-		triangles->InternalProcessAllTriangles(&callback,m_bvhAabbMin,m_bvhAabbMax);
+		triangles->InternalProcessAllTriangles(&callback, m_bvhAabbMin, m_bvhAabbMax);
 
 		//now we have an array of leafnodes in m_leafNodes
 		numLeafNodes = m_quantizedLeafNodes.size();
 
-
-		m_quantizedContiguousNodes.resize(2*numLeafNodes);
-
-
-	} else
+		m_quantizedContiguousNodes.resize(2 * numLeafNodes);
+	}
+	else
 	{
-		NodeTriangleCallback	callback(m_leafNodes);
+		NodeTriangleCallback callback(m_leafNodes);
 
-		btVector3 aabbMin(btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT));
-		btVector3 aabbMax(btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT));
+		btVector3 aabbMin(btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT));
+		btVector3 aabbMax(btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT));
 
-		triangles->InternalProcessAllTriangles(&callback,aabbMin,aabbMax);
+		triangles->InternalProcessAllTriangles(&callback, aabbMin, aabbMax);
 
 		//now we have an array of leafnodes in m_leafNodes
 		numLeafNodes = m_leafNodes.size();
 
-		m_contiguousNodes.resize(2*numLeafNodes);
+		m_contiguousNodes.resize(2 * numLeafNodes);
 	}
 
 	m_curNodeIndex = 0;
 
-	buildTree(0,numLeafNodes);
+	buildTree(0, numLeafNodes);
 
 	///if the entire tree is small then subtree size, we need to create a header info for the tree
-	if(m_useQuantization && !m_SubtreeHeaders.size())
+	if (m_useQuantization && !m_SubtreeHeaders.size())
 	{
 		btBvhSubtreeInfo& subtree = m_SubtreeHeaders.expand();
 		subtree.setAabbFromQuantizeNode(m_quantizedContiguousNodes[0]);
@@ -200,37 +188,29 @@ void btOptimizedBvh::build(btStridingMeshInterface* triangles, bool useQuantized
 	m_leafNodes.clear();
 }
 
-
-
-
-void	btOptimizedBvh::refit(btStridingMeshInterface* meshInterface,const btVector3& aabbMin,const btVector3& aabbMax)
+void btOptimizedBvh::refit(btStridingMeshInterface* meshInterface, const btVector3& aabbMin, const btVector3& aabbMax)
 {
 	if (m_useQuantization)
 	{
+		setQuantizationValues(aabbMin, aabbMax);
 
-		setQuantizationValues(aabbMin,aabbMax);
-
-		updateBvhNodes(meshInterface,0,m_curNodeIndex,0);
+		updateBvhNodes(meshInterface, 0, m_curNodeIndex, 0);
 
 		///now update all subtree headers
 
 		int i;
-		for (i=0;i<m_SubtreeHeaders.size();i++)
+		for (i = 0; i < m_SubtreeHeaders.size(); i++)
 		{
 			btBvhSubtreeInfo& subtree = m_SubtreeHeaders[i];
 			subtree.setAabbFromQuantizeNode(m_quantizedContiguousNodes[subtree.m_rootNodeIndex]);
 		}
-
-	} else
+	}
+	else
 	{
-
 	}
 }
 
-
-
-
-void	btOptimizedBvh::refitPartial(btStridingMeshInterface* meshInterface,const btVector3& aabbMin,const btVector3& aabbMax)
+void btOptimizedBvh::refitPartial(btStridingMeshInterface* meshInterface, const btVector3& aabbMin, const btVector3& aabbMax)
 {
 	//incrementally initialize quantization values
 	btAssert(m_useQuantization);
@@ -245,147 +225,135 @@ void	btOptimizedBvh::refitPartial(btStridingMeshInterface* meshInterface,const b
 
 	///we should update all quantization values, using updateBvhNodes(meshInterface);
 	///but we only update chunks that overlap the given aabb
-	
-	unsigned short	quantizedQueryAabbMin[3];
-	unsigned short	quantizedQueryAabbMax[3];
 
-	quantize(&quantizedQueryAabbMin[0],aabbMin,0);
-	quantize(&quantizedQueryAabbMax[0],aabbMax,1);
+	unsigned short quantizedQueryAabbMin[3];
+	unsigned short quantizedQueryAabbMax[3];
+
+	quantize(&quantizedQueryAabbMin[0], aabbMin, 0);
+	quantize(&quantizedQueryAabbMax[0], aabbMax, 1);
 
 	int i;
-	for (i=0;i<this->m_SubtreeHeaders.size();i++)
+	for (i = 0; i < this->m_SubtreeHeaders.size(); i++)
 	{
 		btBvhSubtreeInfo& subtree = m_SubtreeHeaders[i];
 
 		//PCK: unsigned instead of bool
-		unsigned overlap = testQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin,quantizedQueryAabbMax,subtree.m_quantizedAabbMin,subtree.m_quantizedAabbMax);
+		unsigned overlap = testQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin, quantizedQueryAabbMax, subtree.m_quantizedAabbMin, subtree.m_quantizedAabbMax);
 		if (overlap != 0)
 		{
-			updateBvhNodes(meshInterface,subtree.m_rootNodeIndex,subtree.m_rootNodeIndex+subtree.m_subtreeSize,i);
+			updateBvhNodes(meshInterface, subtree.m_rootNodeIndex, subtree.m_rootNodeIndex + subtree.m_subtreeSize, i);
 
 			subtree.setAabbFromQuantizeNode(m_quantizedContiguousNodes[subtree.m_rootNodeIndex]);
 		}
 	}
-	
 }
 
-void	btOptimizedBvh::updateBvhNodes(btStridingMeshInterface* meshInterface,int firstNode,int endNode,int index)
+void btOptimizedBvh::updateBvhNodes(btStridingMeshInterface* meshInterface, int firstNode, int endNode, int index)
 {
 	(void)index;
 
 	btAssert(m_useQuantization);
 
-	int curNodeSubPart=-1;
+	int curNodeSubPart = -1;
 
 	//get access info to trianglemesh data
-		const unsigned char *vertexbase = 0;
-		int numverts = 0;
-		PHY_ScalarType type = PHY_INTEGER;
-		int stride = 0;
-		const unsigned char *indexbase = 0;
-		int indexstride = 0;
-		int numfaces = 0;
-		PHY_ScalarType indicestype = PHY_INTEGER;
+	const unsigned char* vertexbase = 0;
+	int numverts = 0;
+	PHY_ScalarType type = PHY_INTEGER;
+	int stride = 0;
+	const unsigned char* indexbase = 0;
+	int indexstride = 0;
+	int numfaces = 0;
+	PHY_ScalarType indicestype = PHY_INTEGER;
 
-		btVector3	triangleVerts[3];
-		btVector3	aabbMin,aabbMax;
-		const btVector3& meshScaling = meshInterface->getScaling();
-		
-		int i;
-		for (i=endNode-1;i>=firstNode;i--)
+	btVector3 triangleVerts[3];
+	btVector3 aabbMin, aabbMax;
+	const btVector3& meshScaling = meshInterface->getScaling();
+
+	int i;
+	for (i = endNode - 1; i >= firstNode; i--)
+	{
+		btQuantizedBvhNode& curNode = m_quantizedContiguousNodes[i];
+		if (curNode.isLeafNode())
 		{
-
-
-			btQuantizedBvhNode& curNode = m_quantizedContiguousNodes[i];
-			if (curNode.isLeafNode())
+			//recalc aabb from triangle data
+			int nodeSubPart = curNode.getPartId();
+			int nodeTriangleIndex = curNode.getTriangleIndex();
+			if (nodeSubPart != curNodeSubPart)
 			{
-				//recalc aabb from triangle data
-				int nodeSubPart = curNode.getPartId();
-				int nodeTriangleIndex = curNode.getTriangleIndex();
-				if (nodeSubPart != curNodeSubPart)
-				{
-					if (curNodeSubPart >= 0)
-						meshInterface->unLockReadOnlyVertexBase(curNodeSubPart);
-					meshInterface->getLockedReadOnlyVertexIndexBase(&vertexbase,numverts,	type,stride,&indexbase,indexstride,numfaces,indicestype,nodeSubPart);
+				if (curNodeSubPart >= 0)
+					meshInterface->unLockReadOnlyVertexBase(curNodeSubPart);
+				meshInterface->getLockedReadOnlyVertexIndexBase(&vertexbase, numverts, type, stride, &indexbase, indexstride, numfaces, indicestype, nodeSubPart);
 
-					curNodeSubPart = nodeSubPart;
-					btAssert(indicestype==PHY_INTEGER||indicestype==PHY_SHORT);
-				}
-				//triangles->getLockedReadOnlyVertexIndexBase(vertexBase,numVerts,
+				curNodeSubPart = nodeSubPart;
+				btAssert(indicestype == PHY_INTEGER || indicestype == PHY_SHORT);
+			}
+			//triangles->getLockedReadOnlyVertexIndexBase(vertexBase,numVerts,
 
-				unsigned int* gfxbase = (unsigned int*)(indexbase+nodeTriangleIndex*indexstride);
-				
-				
-				for (int j=2;j>=0;j--)
-				{
-					
-					int graphicsindex = indicestype==PHY_SHORT?((unsigned short*)gfxbase)[j]:gfxbase[j];
-					if (type == PHY_FLOAT)
-					{
-						float* graphicsbase = (float*)(vertexbase+graphicsindex*stride);
-						triangleVerts[j] = btVector3(
-							graphicsbase[0]*meshScaling.getX(),
-							graphicsbase[1]*meshScaling.getY(),
-							graphicsbase[2]*meshScaling.getZ());
-					}
-					else
-					{
-						double* graphicsbase = (double*)(vertexbase+graphicsindex*stride);
-						triangleVerts[j] = btVector3( btScalar(graphicsbase[0]*meshScaling.getX()), btScalar(graphicsbase[1]*meshScaling.getY()), btScalar(graphicsbase[2]*meshScaling.getZ()));
-					}
-				}
+			unsigned int* gfxbase = (unsigned int*)(indexbase + nodeTriangleIndex * indexstride);
 
-
-				
-				aabbMin.setValue(btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT));
-				aabbMax.setValue(btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT)); 
-				aabbMin.setMin(triangleVerts[0]);
-				aabbMax.setMax(triangleVerts[0]);
-				aabbMin.setMin(triangleVerts[1]);
-				aabbMax.setMax(triangleVerts[1]);
-				aabbMin.setMin(triangleVerts[2]);
-				aabbMax.setMax(triangleVerts[2]);
-
-				quantize(&curNode.m_quantizedAabbMin[0],aabbMin,0);
-				quantize(&curNode.m_quantizedAabbMax[0],aabbMax,1);
-				
-			} else
+			for (int j = 2; j >= 0; j--)
 			{
-				//combine aabb from both children
-
-				btQuantizedBvhNode* leftChildNode = &m_quantizedContiguousNodes[i+1];
-				
-				btQuantizedBvhNode* rightChildNode = leftChildNode->isLeafNode() ? &m_quantizedContiguousNodes[i+2] :
-					&m_quantizedContiguousNodes[i+1+leftChildNode->getEscapeIndex()];
-				
-
+				int graphicsindex = indicestype == PHY_SHORT ? ((unsigned short*)gfxbase)[j] : gfxbase[j];
+				if (type == PHY_FLOAT)
 				{
-					for (int i=0;i<3;i++)
-					{
-						curNode.m_quantizedAabbMin[i] = leftChildNode->m_quantizedAabbMin[i];
-						if (curNode.m_quantizedAabbMin[i]>rightChildNode->m_quantizedAabbMin[i])
-							curNode.m_quantizedAabbMin[i]=rightChildNode->m_quantizedAabbMin[i];
-
-						curNode.m_quantizedAabbMax[i] = leftChildNode->m_quantizedAabbMax[i];
-						if (curNode.m_quantizedAabbMax[i] < rightChildNode->m_quantizedAabbMax[i])
-							curNode.m_quantizedAabbMax[i] = rightChildNode->m_quantizedAabbMax[i];
-					}
+					float* graphicsbase = (float*)(vertexbase + graphicsindex * stride);
+					triangleVerts[j] = btVector3(
+						graphicsbase[0] * meshScaling.getX(),
+						graphicsbase[1] * meshScaling.getY(),
+						graphicsbase[2] * meshScaling.getZ());
+				}
+				else
+				{
+					double* graphicsbase = (double*)(vertexbase + graphicsindex * stride);
+					triangleVerts[j] = btVector3(btScalar(graphicsbase[0] * meshScaling.getX()), btScalar(graphicsbase[1] * meshScaling.getY()), btScalar(graphicsbase[2] * meshScaling.getZ()));
 				}
 			}
 
+			aabbMin.setValue(btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT));
+			aabbMax.setValue(btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT));
+			aabbMin.setMin(triangleVerts[0]);
+			aabbMax.setMax(triangleVerts[0]);
+			aabbMin.setMin(triangleVerts[1]);
+			aabbMax.setMax(triangleVerts[1]);
+			aabbMin.setMin(triangleVerts[2]);
+			aabbMax.setMax(triangleVerts[2]);
+
+			quantize(&curNode.m_quantizedAabbMin[0], aabbMin, 0);
+			quantize(&curNode.m_quantizedAabbMax[0], aabbMax, 1);
 		}
+		else
+		{
+			//combine aabb from both children
 
-		if (curNodeSubPart >= 0)
-			meshInterface->unLockReadOnlyVertexBase(curNodeSubPart);
+			btQuantizedBvhNode* leftChildNode = &m_quantizedContiguousNodes[i + 1];
 
-		
+			btQuantizedBvhNode* rightChildNode = leftChildNode->isLeafNode() ? &m_quantizedContiguousNodes[i + 2] : &m_quantizedContiguousNodes[i + 1 + leftChildNode->getEscapeIndex()];
+
+			{
+				for (int i = 0; i < 3; i++)
+				{
+					curNode.m_quantizedAabbMin[i] = leftChildNode->m_quantizedAabbMin[i];
+					if (curNode.m_quantizedAabbMin[i] > rightChildNode->m_quantizedAabbMin[i])
+						curNode.m_quantizedAabbMin[i] = rightChildNode->m_quantizedAabbMin[i];
+
+					curNode.m_quantizedAabbMax[i] = leftChildNode->m_quantizedAabbMax[i];
+					if (curNode.m_quantizedAabbMax[i] < rightChildNode->m_quantizedAabbMax[i])
+						curNode.m_quantizedAabbMax[i] = rightChildNode->m_quantizedAabbMax[i];
+				}
+			}
+		}
+	}
+
+	if (curNodeSubPart >= 0)
+		meshInterface->unLockReadOnlyVertexBase(curNodeSubPart);
 }
 
 ///deSerializeInPlace loads and initializes a BVH from a buffer in memory 'in place'
-btOptimizedBvh* btOptimizedBvh::deSerializeInPlace(void *i_alignedDataBuffer, unsigned int i_dataBufferSize, bool i_swapEndian)
+btOptimizedBvh* btOptimizedBvh::deSerializeInPlace(void* i_alignedDataBuffer, unsigned int i_dataBufferSize, bool i_swapEndian)
 {
-	btQuantizedBvh* bvh = btQuantizedBvh::deSerializeInPlace(i_alignedDataBuffer,i_dataBufferSize,i_swapEndian);
-	
+	btQuantizedBvh* bvh = btQuantizedBvh::deSerializeInPlace(i_alignedDataBuffer, i_dataBufferSize, i_swapEndian);
+
 	//we don't add additional data so just do a static upcast
 	return static_cast<btOptimizedBvh*>(bvh);
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btOptimizedBvh.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btOptimizedBvh.h
index 715961f5528..22f131c8b25 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btOptimizedBvh.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btOptimizedBvh.h
@@ -22,44 +22,35 @@ subject to the following restrictions:
 
 class btStridingMeshInterface;
 
-
 ///The btOptimizedBvh extends the btQuantizedBvh to create AABB tree for triangle meshes, through the btStridingMeshInterface.
-ATTRIBUTE_ALIGNED16(class) btOptimizedBvh : public btQuantizedBvh
+ATTRIBUTE_ALIGNED16(class)
+btOptimizedBvh : public btQuantizedBvh
 {
-	
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
 protected:
-
 public:
-
 	btOptimizedBvh();
 
 	virtual ~btOptimizedBvh();
 
-	void	build(btStridingMeshInterface* triangles,bool useQuantizedAabbCompression, const btVector3& bvhAabbMin, const btVector3& bvhAabbMax);
+	void build(btStridingMeshInterface * triangles, bool useQuantizedAabbCompression, const btVector3& bvhAabbMin, const btVector3& bvhAabbMax);
 
-	void	refit(btStridingMeshInterface* triangles,const btVector3& aabbMin,const btVector3& aabbMax);
+	void refit(btStridingMeshInterface * triangles, const btVector3& aabbMin, const btVector3& aabbMax);
 
-	void	refitPartial(btStridingMeshInterface* triangles,const btVector3& aabbMin, const btVector3& aabbMax);
+	void refitPartial(btStridingMeshInterface * triangles, const btVector3& aabbMin, const btVector3& aabbMax);
 
-	void	updateBvhNodes(btStridingMeshInterface* meshInterface,int firstNode,int endNode,int index);
+	void updateBvhNodes(btStridingMeshInterface * meshInterface, int firstNode, int endNode, int index);
 
 	/// Data buffer MUST be 16 byte aligned
-	virtual bool serializeInPlace(void *o_alignedDataBuffer, unsigned i_dataBufferSize, bool i_swapEndian) const
+	virtual bool serializeInPlace(void* o_alignedDataBuffer, unsigned i_dataBufferSize, bool i_swapEndian) const
 	{
-		return btQuantizedBvh::serialize(o_alignedDataBuffer,i_dataBufferSize,i_swapEndian);
-
+		return btQuantizedBvh::serialize(o_alignedDataBuffer, i_dataBufferSize, i_swapEndian);
 	}
 
 	///deSerializeInPlace loads and initializes a BVH from a buffer in memory 'in place'
-	static btOptimizedBvh *deSerializeInPlace(void *i_alignedDataBuffer, unsigned int i_dataBufferSize, bool i_swapEndian);
-
-
+	static btOptimizedBvh* deSerializeInPlace(void* i_alignedDataBuffer, unsigned int i_dataBufferSize, bool i_swapEndian);
 };
 
-
-#endif //BT_OPTIMIZED_BVH_H
-
-
+#endif  //BT_OPTIMIZED_BVH_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btPolyhedralConvexShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btPolyhedralConvexShape.cpp
index d51b6760fc1..521ecfc7606 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btPolyhedralConvexShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btPolyhedralConvexShape.cpp
@@ -12,7 +12,7 @@ subject to the following restrictions:
 2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.
 3. This notice may not be removed or altered from any source distribution.
 */
-#if defined (_WIN32) || defined (__i386__)
+#if defined(_WIN32) || defined(__i386__)
 #define BT_USE_SSE_IN_API
 #endif
 
@@ -23,11 +23,9 @@ subject to the following restrictions:
 #include "LinearMath/btGeometryUtil.h"
 #include "LinearMath/btGrahamScan2dConvexHull.h"
 
-
-btPolyhedralConvexShape::btPolyhedralConvexShape() :btConvexInternalShape(),
-m_polyhedron(0)
+btPolyhedralConvexShape::btPolyhedralConvexShape() : btConvexInternalShape(),
+													 m_polyhedron(0)
 {
-
 }
 
 btPolyhedralConvexShape::~btPolyhedralConvexShape()
@@ -44,65 +42,65 @@ void btPolyhedralConvexShape::setPolyhedralFeatures(btConvexPolyhedron& polyhedr
 	if (m_polyhedron)
 	{
 		*m_polyhedron = polyhedron;
-	} else
+	}
+	else
 	{
-		void* mem = btAlignedAlloc(sizeof(btConvexPolyhedron),16);
+		void* mem = btAlignedAlloc(sizeof(btConvexPolyhedron), 16);
 		m_polyhedron = new (mem) btConvexPolyhedron(polyhedron);
 	}
 }
 
-bool	btPolyhedralConvexShape::initializePolyhedralFeatures(int shiftVerticesByMargin)
+bool btPolyhedralConvexShape::initializePolyhedralFeatures(int shiftVerticesByMargin)
 {
-
 	if (m_polyhedron)
 	{
 		m_polyhedron->~btConvexPolyhedron();
 		btAlignedFree(m_polyhedron);
 	}
-	
-	void* mem = btAlignedAlloc(sizeof(btConvexPolyhedron),16);
+
+	void* mem = btAlignedAlloc(sizeof(btConvexPolyhedron), 16);
 	m_polyhedron = new (mem) btConvexPolyhedron;
 
 	btAlignedObjectArray<btVector3> orgVertices;
 
-	for (int i=0;i<getNumVertices();i++)
+	for (int i = 0; i < getNumVertices(); i++)
 	{
 		btVector3& newVertex = orgVertices.expand();
-		getVertex(i,newVertex);
+		getVertex(i, newVertex);
 	}
-	
+
 	btConvexHullComputer conv;
-	
+
 	if (shiftVerticesByMargin)
 	{
 		btAlignedObjectArray<btVector3> planeEquations;
-		btGeometryUtil::getPlaneEquationsFromVertices(orgVertices,planeEquations);
+		btGeometryUtil::getPlaneEquationsFromVertices(orgVertices, planeEquations);
 
 		btAlignedObjectArray<btVector3> shiftedPlaneEquations;
-		for (int p=0;p<planeEquations.size();p++)
+		for (int p = 0; p < planeEquations.size(); p++)
 		{
-			   btVector3 plane = planeEquations[p];
-		//	   btScalar margin = getMargin();
-			   plane[3] -= getMargin();
-			   shiftedPlaneEquations.push_back(plane);
+			btVector3 plane = planeEquations[p];
+			//	   btScalar margin = getMargin();
+			plane[3] -= getMargin();
+			shiftedPlaneEquations.push_back(plane);
 		}
 
 		btAlignedObjectArray<btVector3> tmpVertices;
 
-		btGeometryUtil::getVerticesFromPlaneEquations(shiftedPlaneEquations,tmpVertices);
-	
-		conv.compute(&tmpVertices[0].getX(), sizeof(btVector3),tmpVertices.size(),0.f,0.f);
-	} else
+		btGeometryUtil::getVerticesFromPlaneEquations(shiftedPlaneEquations, tmpVertices);
+
+		conv.compute(&tmpVertices[0].getX(), sizeof(btVector3), tmpVertices.size(), 0.f, 0.f);
+	}
+	else
 	{
-		
-		conv.compute(&orgVertices[0].getX(), sizeof(btVector3),orgVertices.size(),0.f,0.f);
+		conv.compute(&orgVertices[0].getX(), sizeof(btVector3), orgVertices.size(), 0.f, 0.f);
 	}
 
 #ifndef BT_RECONSTRUCT_FACES
-	
+
 	int numVertices = conv.vertices.size();
 	m_polyhedron->m_vertices.resize(numVertices);
-	for (int p=0;p<numVertices;p++)
+	for (int p = 0; p < numVertices; p++)
 	{
 		m_polyhedron->m_vertices[p] = conv.vertices[p];
 	}
@@ -115,78 +113,72 @@ bool	btPolyhedralConvexShape::initializePolyhedralFeatures(int shiftVerticesByMa
 		btFace combinedFace;
 		const btConvexHullComputer::Edge* edge = &conv.edges[conv.faces[j]];
 		v0 = edge->getSourceVertex();
-		int prevVertex=v0;
+		int prevVertex = v0;
 		combinedFace.m_indices.push_back(v0);
 		v1 = edge->getTargetVertex();
 		while (v1 != v0)
 		{
-		
 			btVector3 wa = conv.vertices[prevVertex];
 			btVector3 wb = conv.vertices[v1];
-			btVector3 newEdge = wb-wa;
+			btVector3 newEdge = wb - wa;
 			newEdge.normalize();
-			if (numEdges<2)
+			if (numEdges < 2)
 				edges[numEdges++] = newEdge;
 
-
 			//face->addIndex(v1);
 			combinedFace.m_indices.push_back(v1);
 			edge = edge->getNextEdgeOfFace();
 			prevVertex = v1;
 			int v01 = edge->getSourceVertex();
 			v1 = edge->getTargetVertex();
-
 		}
-		
+
 		btAssert(combinedFace.m_indices.size() > 2);
 
 		btVector3 faceNormal = edges[0].cross(edges[1]);
 		faceNormal.normalize();
 
-		btScalar planeEq=1e30f;
+		btScalar planeEq = 1e30f;
 
-		for (int v=0;v<combinedFace.m_indices.size();v++)
+		for (int v = 0; v < combinedFace.m_indices.size(); v++)
 		{
 			btScalar eq = m_polyhedron->m_vertices[combinedFace.m_indices[v]].dot(faceNormal);
-			if (planeEq>eq)
+			if (planeEq > eq)
 			{
-				planeEq=eq;
+				planeEq = eq;
 			}
 		}
 		combinedFace.m_plane[0] = faceNormal.getX();
 		combinedFace.m_plane[1] = faceNormal.getY();
 		combinedFace.m_plane[2] = faceNormal.getZ();
 		combinedFace.m_plane[3] = -planeEq;
-		
+
 		m_polyhedron->m_faces.push_back(combinedFace);
 	}
 
-
-#else//BT_RECONSTRUCT_FACES
+#else  //BT_RECONSTRUCT_FACES
 
 	btAlignedObjectArray<btVector3> faceNormals;
 	int numFaces = conv.faces.size();
 	faceNormals.resize(numFaces);
 	btConvexHullComputer* convexUtil = &conv;
 
-	
-	btAlignedObjectArray<btFace>	tmpFaces;
+	btAlignedObjectArray<btFace> tmpFaces;
 	tmpFaces.resize(numFaces);
 
 	int numVertices = convexUtil->vertices.size();
 	m_polyhedron->m_vertices.resize(numVertices);
-	for (int p=0;p<numVertices;p++)
+	for (int p = 0; p < numVertices; p++)
 	{
 		m_polyhedron->m_vertices[p] = convexUtil->vertices[p];
 	}
 
-
-	for (int i=0;i<numFaces;i++)
+	for (int i = 0; i < numFaces; i++)
 	{
 		int face = convexUtil->faces[i];
 		//printf("face=%d\n",face);
-		const btConvexHullComputer::Edge*  firstEdge = &convexUtil->edges[face];
-		const btConvexHullComputer::Edge*  edge = firstEdge;
+		const btConvexHullComputer::Edge* firstEdge = &convexUtil->edges[face];
+		const btConvexHullComputer::Edge* edge = firstEdge;
 
 		btVector3 edges[3];
 		int numEdges = 0;
@@ -194,25 +186,23 @@ bool	btPolyhedralConvexShape::initializePolyhedralFeatures(int shiftVerticesByMa
 
 		do
 		{
-			
 			int src = edge->getSourceVertex();
 			tmpFaces[i].m_indices.push_back(src);
 			int targ = edge->getTargetVertex();
 			btVector3 wa = convexUtil->vertices[src];
 
 			btVector3 wb = convexUtil->vertices[targ];
-			btVector3 newEdge = wb-wa;
+			btVector3 newEdge = wb - wa;
 			newEdge.normalize();
-			if (numEdges<2)
+			if (numEdges < 2)
 				edges[numEdges++] = newEdge;
 
 			edge = edge->getNextEdgeOfFace();
-		} while (edge!=firstEdge);
+		} while (edge != firstEdge);
 
 		btScalar planeEq = 1e30f;
 
-		
-		if (numEdges==2)
+		if (numEdges == 2)
 		{
 			faceNormals[i] = edges[0].cross(edges[1]);
 			faceNormals[i].normalize();
@@ -220,20 +210,19 @@ bool	btPolyhedralConvexShape::initializePolyhedralFeatures(int shiftVerticesByMa
 			tmpFaces[i].m_plane[1] = faceNormals[i].getY();
 			tmpFaces[i].m_plane[2] = faceNormals[i].getZ();
 			tmpFaces[i].m_plane[3] = planeEq;
-
 		}
 		else
 		{
-			btAssert(0);//degenerate?
+			btAssert(0);  //degenerate?
 			faceNormals[i].setZero();
 		}
 
-		for (int v=0;v<tmpFaces[i].m_indices.size();v++)
+		for (int v = 0; v < tmpFaces[i].m_indices.size(); v++)
 		{
 			btScalar eq = m_polyhedron->m_vertices[tmpFaces[i].m_indices[v]].dot(faceNormals[i]);
-			if (planeEq>eq)
+			if (planeEq > eq)
 			{
-				planeEq=eq;
+				planeEq = eq;
 			}
 		}
 		tmpFaces[i].m_plane[3] = -planeEq;
@@ -241,89 +230,86 @@ bool	btPolyhedralConvexShape::initializePolyhedralFeatures(int shiftVerticesByMa
 
 	//merge coplanar faces and copy them to m_polyhedron
 
-	btScalar faceWeldThreshold= 0.999f;
+	btScalar faceWeldThreshold = 0.999f;
 	btAlignedObjectArray<int> todoFaces;
-	for (int i=0;i<tmpFaces.size();i++)
+	for (int i = 0; i < tmpFaces.size(); i++)
 		todoFaces.push_back(i);
 
 	while (todoFaces.size())
 	{
 		btAlignedObjectArray<int> coplanarFaceGroup;
-		int refFace = todoFaces[todoFaces.size()-1];
+		int refFace = todoFaces[todoFaces.size() - 1];
 
 		coplanarFaceGroup.push_back(refFace);
 		btFace& faceA = tmpFaces[refFace];
 		todoFaces.pop_back();
 
-		btVector3 faceNormalA(faceA.m_plane[0],faceA.m_plane[1],faceA.m_plane[2]);
-		for (int j=todoFaces.size()-1;j>=0;j--)
+		btVector3 faceNormalA(faceA.m_plane[0], faceA.m_plane[1], faceA.m_plane[2]);
+		for (int j = todoFaces.size() - 1; j >= 0; j--)
 		{
 			int i = todoFaces[j];
 			btFace& faceB = tmpFaces[i];
-			btVector3 faceNormalB(faceB.m_plane[0],faceB.m_plane[1],faceB.m_plane[2]);
-			if (faceNormalA.dot(faceNormalB)>faceWeldThreshold)
+			btVector3 faceNormalB(faceB.m_plane[0], faceB.m_plane[1], faceB.m_plane[2]);
+			if (faceNormalA.dot(faceNormalB) > faceWeldThreshold)
 			{
 				coplanarFaceGroup.push_back(i);
 				todoFaces.remove(i);
 			}
 		}
 
-
 		bool did_merge = false;
-		if (coplanarFaceGroup.size()>1)
+		if (coplanarFaceGroup.size() > 1)
 		{
 			//do the merge: use Graham Scan 2d convex hull
 
 			btAlignedObjectArray<GrahamVector3> orgpoints;
-			btVector3 averageFaceNormal(0,0,0);
+			btVector3 averageFaceNormal(0, 0, 0);
 
-			for (int i=0;i<coplanarFaceGroup.size();i++)
+			for (int i = 0; i < coplanarFaceGroup.size(); i++)
 			{
-//				m_polyhedron->m_faces.push_back(tmpFaces[coplanarFaceGroup[i]]);
+				//				m_polyhedron->m_faces.push_back(tmpFaces[coplanarFaceGroup[i]]);
 
 				btFace& face = tmpFaces[coplanarFaceGroup[i]];
-				btVector3 faceNormal(face.m_plane[0],face.m_plane[1],face.m_plane[2]);
-				averageFaceNormal+=faceNormal;
-				for (int f=0;f<face.m_indices.size();f++)
+				btVector3 faceNormal(face.m_plane[0], face.m_plane[1], face.m_plane[2]);
+				averageFaceNormal += faceNormal;
+				for (int f = 0; f < face.m_indices.size(); f++)
 				{
 					int orgIndex = face.m_indices[f];
 					btVector3 pt = m_polyhedron->m_vertices[orgIndex];
-					
+
 					bool found = false;
 
-					for (int i=0;i<orgpoints.size();i++)
+					for (int i = 0; i < orgpoints.size(); i++)
 					{
 						//if ((orgpoints[i].m_orgIndex == orgIndex) || ((rotatedPt-orgpoints[i]).length2()<0.0001))
 						if (orgpoints[i].m_orgIndex == orgIndex)
 						{
-							found=true;
+							found = true;
 							break;
 						}
 					}
 					if (!found)
-						orgpoints.push_back(GrahamVector3(pt,orgIndex));
+						orgpoints.push_back(GrahamVector3(pt, orgIndex));
 				}
 			}
 
-			
-
 			btFace combinedFace;
-			for (int i=0;i<4;i++)
+			for (int i = 0; i < 4; i++)
 				combinedFace.m_plane[i] = tmpFaces[coplanarFaceGroup[0]].m_plane[i];
 
 			btAlignedObjectArray<GrahamVector3> hull;
 
 			averageFaceNormal.normalize();
-			GrahamScanConvexHull2D(orgpoints,hull,averageFaceNormal);
+			GrahamScanConvexHull2D(orgpoints, hull, averageFaceNormal);
 
-			for (int i=0;i<hull.size();i++)
+			for (int i = 0; i < hull.size(); i++)
 			{
 				combinedFace.m_indices.push_back(hull[i].m_orgIndex);
-				for(int k = 0; k < orgpoints.size(); k++) 
+				for (int k = 0; k < orgpoints.size(); k++)
 				{
-					if(orgpoints[k].m_orgIndex == hull[i].m_orgIndex) 
+					if (orgpoints[k].m_orgIndex == hull[i].m_orgIndex)
 					{
-						orgpoints[k].m_orgIndex = -1; // invalidate...
+						orgpoints[k].m_orgIndex = -1;  // invalidate...
 						break;
 					}
 				}
@@ -331,38 +317,41 @@ bool	btPolyhedralConvexShape::initializePolyhedralFeatures(int shiftVerticesByMa
 
 			// are there rejected vertices?
 			bool reject_merge = false;
-			
 
-
-			for(int i = 0; i < orgpoints.size(); i++) {
-				if(orgpoints[i].m_orgIndex == -1)
-					continue; // this is in the hull...
+			for (int i = 0; i < orgpoints.size(); i++)
+			{
+				if (orgpoints[i].m_orgIndex == -1)
+					continue;  // this is in the hull...
 				// this vertex is rejected -- is anybody else using this vertex?
-				for(int j = 0; j < tmpFaces.size(); j++) {
-					
+				for (int j = 0; j < tmpFaces.size(); j++)
+				{
 					btFace& face = tmpFaces[j];
 					// is this a face of the current coplanar group?
 					bool is_in_current_group = false;
-					for(int k = 0; k < coplanarFaceGroup.size(); k++) {
-						if(coplanarFaceGroup[k] == j) {
+					for (int k = 0; k < coplanarFaceGroup.size(); k++)
+					{
+						if (coplanarFaceGroup[k] == j)
+						{
 							is_in_current_group = true;
 							break;
 						}
 					}
-					if(is_in_current_group) // ignore this face...
+					if (is_in_current_group)  // ignore this face...
 						continue;
 					// does this face use this rejected vertex?
-					for(int v = 0; v < face.m_indices.size(); v++) {
-						if(face.m_indices[v] == orgpoints[i].m_orgIndex) {
+					for (int v = 0; v < face.m_indices.size(); v++)
+					{
+						if (face.m_indices[v] == orgpoints[i].m_orgIndex)
+						{
 							// this rejected vertex is used in another face -- reject merge
 							reject_merge = true;
 							break;
 						}
 					}
-					if(reject_merge)
+					if (reject_merge)
 						break;
 				}
-				if(reject_merge)
+				if (reject_merge)
 					break;
 			}
 
@@ -373,21 +362,17 @@ bool	btPolyhedralConvexShape::initializePolyhedralFeatures(int shiftVerticesByMa
 				m_polyhedron->m_faces.push_back(combinedFace);
 			}
 		}
-		if(!did_merge)
+		if (!did_merge)
 		{
-			for (int i=0;i<coplanarFaceGroup.size();i++)
+			for (int i = 0; i < coplanarFaceGroup.size(); i++)
 			{
 				btFace face = tmpFaces[coplanarFaceGroup[i]];
 				m_polyhedron->m_faces.push_back(face);
 			}
-
-		} 
-
-
-
+		}
 	}
 
-#endif //BT_RECONSTRUCT_FACES
+#endif  //BT_RECONSTRUCT_FACES
 
 	m_polyhedron->initialize();
 
@@ -395,14 +380,12 @@ bool	btPolyhedralConvexShape::initializePolyhedralFeatures(int shiftVerticesByMa
 }
 
 #ifndef MIN
-    #define MIN(_a, _b)     ((_a) < (_b) ? (_a) : (_b))
+#define MIN(_a, _b) ((_a) < (_b) ? (_a) : (_b))
 #endif
 
-btVector3	btPolyhedralConvexShape::localGetSupportingVertexWithoutMargin(const btVector3& vec0)const
+btVector3 btPolyhedralConvexShape::localGetSupportingVertexWithoutMargin(const btVector3& vec0) const
 {
-
-
-	btVector3 supVec(0,0,0);
+	btVector3 supVec(0, 0, 0);
 #ifndef __SPU__
 	int i;
 	btScalar maxDot(btScalar(-BT_LARGE_FLOAT));
@@ -411,37 +394,36 @@ btVector3	btPolyhedralConvexShape::localGetSupportingVertexWithoutMargin(const b
 	btScalar lenSqr = vec.length2();
 	if (lenSqr < btScalar(0.0001))
 	{
-		vec.setValue(1,0,0);
-	} else
+		vec.setValue(1, 0, 0);
+	}
+	else
 	{
-		btScalar rlen = btScalar(1.) / btSqrt(lenSqr );
+		btScalar rlen = btScalar(1.) / btSqrt(lenSqr);
 		vec *= rlen;
 	}
 
 	btVector3 vtx;
 	btScalar newDot;
 
-    for( int k = 0; k < getNumVertices(); k += 128 )
-    {
-        btVector3 temp[128];
-        int inner_count = MIN(getNumVertices() - k, 128);
-        for( i = 0; i < inner_count; i++ )
-            getVertex(i,temp[i]); 
-        i = (int) vec.maxDot( temp, inner_count, newDot);
+	for (int k = 0; k < getNumVertices(); k += 128)
+	{
+		btVector3 temp[128];
+		int inner_count = MIN(getNumVertices() - k, 128);
+		for (i = 0; i < inner_count; i++)
+			getVertex(i, temp[i]);
+		i = (int)vec.maxDot(temp, inner_count, newDot);
 		if (newDot > maxDot)
 		{
 			maxDot = newDot;
 			supVec = temp[i];
-		}        
-    }
-	
-#endif //__SPU__
+		}
+	}
+
+#endif  //__SPU__
 	return supVec;
 }
 
-
-
-void	btPolyhedralConvexShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const
+void btPolyhedralConvexShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const
 {
 #ifndef __SPU__
 	int i;
@@ -449,36 +431,34 @@ void	btPolyhedralConvexShape::batchedUnitVectorGetSupportingVertexWithoutMargin(
 	btVector3 vtx;
 	btScalar newDot;
 
-	for (i=0;i<numVectors;i++)
+	for (i = 0; i < numVectors; i++)
 	{
 		supportVerticesOut[i][3] = btScalar(-BT_LARGE_FLOAT);
 	}
 
-	for (int j=0;j<numVectors;j++)
+	for (int j = 0; j < numVectors; j++)
 	{
-        const btVector3& vec = vectors[j];
-        
-        for( int k = 0; k < getNumVertices(); k += 128 )
-        {
-            btVector3 temp[128];
-            int inner_count = MIN(getNumVertices() - k, 128);
-            for( i = 0; i < inner_count; i++ )
-                getVertex(i,temp[i]); 
-            i = (int) vec.maxDot( temp, inner_count, newDot);
-            if (newDot > supportVerticesOut[j][3])
-            {
+		const btVector3& vec = vectors[j];
+
+		for (int k = 0; k < getNumVertices(); k += 128)
+		{
+			btVector3 temp[128];
+			int inner_count = MIN(getNumVertices() - k, 128);
+			for (i = 0; i < inner_count; i++)
+				getVertex(i, temp[i]);
+			i = (int)vec.maxDot(temp, inner_count, newDot);
+			if (newDot > supportVerticesOut[j][3])
+			{
 				supportVerticesOut[j] = temp[i];
 				supportVerticesOut[j][3] = newDot;
-            }        
-        }
-    }
+			}
+		}
+	}
 
-#endif //__SPU__
+#endif  //__SPU__
 }
 
-
-
-void	btPolyhedralConvexShape::calculateLocalInertia(btScalar mass,btVector3& inertia) const
+void btPolyhedralConvexShape::calculateLocalInertia(btScalar mass, btVector3& inertia) const
 {
 #ifndef __SPU__
 	//not yet, return box inertia
@@ -487,81 +467,77 @@ void	btPolyhedralConvexShape::calculateLocalInertia(btScalar mass,btVector3& ine
 
 	btTransform ident;
 	ident.setIdentity();
-	btVector3 aabbMin,aabbMax;
-	getAabb(ident,aabbMin,aabbMax);
-	btVector3 halfExtents = (aabbMax-aabbMin)*btScalar(0.5);
+	btVector3 aabbMin, aabbMax;
+	getAabb(ident, aabbMin, aabbMax);
+	btVector3 halfExtents = (aabbMax - aabbMin) * btScalar(0.5);
 
-	btScalar lx=btScalar(2.)*(halfExtents.x()+margin);
-	btScalar ly=btScalar(2.)*(halfExtents.y()+margin);
-	btScalar lz=btScalar(2.)*(halfExtents.z()+margin);
-	const btScalar x2 = lx*lx;
-	const btScalar y2 = ly*ly;
-	const btScalar z2 = lz*lz;
+	btScalar lx = btScalar(2.) * (halfExtents.x() + margin);
+	btScalar ly = btScalar(2.) * (halfExtents.y() + margin);
+	btScalar lz = btScalar(2.) * (halfExtents.z() + margin);
+	const btScalar x2 = lx * lx;
+	const btScalar y2 = ly * ly;
+	const btScalar z2 = lz * lz;
 	const btScalar scaledmass = mass * btScalar(0.08333333);
 
-	inertia = scaledmass * (btVector3(y2+z2,x2+z2,x2+y2));
-#endif //__SPU__
+	inertia = scaledmass * (btVector3(y2 + z2, x2 + z2, x2 + y2));
+#endif  //__SPU__
 }
 
-
-
-void	btPolyhedralConvexAabbCachingShape::setLocalScaling(const btVector3& scaling)
+void btPolyhedralConvexAabbCachingShape::setLocalScaling(const btVector3& scaling)
 {
 	btConvexInternalShape::setLocalScaling(scaling);
 	recalcLocalAabb();
 }
 
 btPolyhedralConvexAabbCachingShape::btPolyhedralConvexAabbCachingShape()
-:btPolyhedralConvexShape(),
-m_localAabbMin(1,1,1),
-m_localAabbMax(-1,-1,-1),
-m_isLocalAabbValid(false)
+	: btPolyhedralConvexShape(),
+	  m_localAabbMin(1, 1, 1),
+	  m_localAabbMax(-1, -1, -1),
+	  m_isLocalAabbValid(false)
 {
 }
 
-void btPolyhedralConvexAabbCachingShape::getAabb(const btTransform& trans,btVector3& aabbMin,btVector3& aabbMax) const
+void btPolyhedralConvexAabbCachingShape::getAabb(const btTransform& trans, btVector3& aabbMin, btVector3& aabbMax) const
 {
-	getNonvirtualAabb(trans,aabbMin,aabbMax,getMargin());
+	getNonvirtualAabb(trans, aabbMin, aabbMax, getMargin());
 }
 
-void	btPolyhedralConvexAabbCachingShape::recalcLocalAabb()
+void btPolyhedralConvexAabbCachingShape::recalcLocalAabb()
 {
 	m_isLocalAabbValid = true;
-	
-	#if 1
+
+#if 1
 	static const btVector3 _directions[] =
-	{
-		btVector3( 1.,  0.,  0.),
-		btVector3( 0.,  1.,  0.),
-		btVector3( 0.,  0.,  1.),
-		btVector3( -1., 0.,  0.),
-		btVector3( 0., -1.,  0.),
-		btVector3( 0.,  0., -1.)
-	};
-	
+		{
+			btVector3(1., 0., 0.),
+			btVector3(0., 1., 0.),
+			btVector3(0., 0., 1.),
+			btVector3(-1., 0., 0.),
+			btVector3(0., -1., 0.),
+			btVector3(0., 0., -1.)};
+
 	btVector3 _supporting[] =
-	{
-		btVector3( 0., 0., 0.),
-		btVector3( 0., 0., 0.),
-		btVector3( 0., 0., 0.),
-		btVector3( 0., 0., 0.),
-		btVector3( 0., 0., 0.),
-		btVector3( 0., 0., 0.)
-	};
-	
+		{
+			btVector3(0., 0., 0.),
+			btVector3(0., 0., 0.),
+			btVector3(0., 0., 0.),
+			btVector3(0., 0., 0.),
+			btVector3(0., 0., 0.),
+			btVector3(0., 0., 0.)};
+
 	batchedUnitVectorGetSupportingVertexWithoutMargin(_directions, _supporting, 6);
-	
-	for ( int i = 0; i < 3; ++i )
+
+	for (int i = 0; i < 3; ++i)
 	{
 		m_localAabbMax[i] = _supporting[i][i] + m_collisionMargin;
 		m_localAabbMin[i] = _supporting[i + 3][i] - m_collisionMargin;
 	}
-	
-	#else
 
-	for (int i=0;i<3;i++)
+#else
+
+	for (int i = 0; i < 3; i++)
 	{
-		btVector3 vec(btScalar(0.),btScalar(0.),btScalar(0.));
+		btVector3 vec(btScalar(0.), btScalar(0.), btScalar(0.));
 		vec[i] = btScalar(1.);
 		btVector3 tmp = localGetSupportingVertex(vec);
 		m_localAabbMax[i] = tmp[i];
@@ -569,9 +545,5 @@ void	btPolyhedralConvexAabbCachingShape::recalcLocalAabb()
 		tmp = localGetSupportingVertex(vec);
 		m_localAabbMin[i] = tmp[i];
 	}
-	#endif
+#endif
 }
-
-
-
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btPolyhedralConvexShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btPolyhedralConvexShape.h
index b7ddb6e0608..b3ffab7a23c 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btPolyhedralConvexShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btPolyhedralConvexShape.h
@@ -20,20 +20,15 @@ subject to the following restrictions:
 #include "btConvexInternalShape.h"
 class btConvexPolyhedron;
 
-
 ///The btPolyhedralConvexShape is an internal interface class for polyhedral convex shapes.
-ATTRIBUTE_ALIGNED16(class) btPolyhedralConvexShape : public btConvexInternalShape
+ATTRIBUTE_ALIGNED16(class)
+btPolyhedralConvexShape : public btConvexInternalShape
 {
-	
-
 protected:
-	
 	btConvexPolyhedron* m_polyhedron;
 
 public:
-	
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
 
 	btPolyhedralConvexShape();
 
@@ -41,55 +36,49 @@ public:
 
 	///optional method mainly used to generate multiple contact points by clipping polyhedral features (faces/edges)
 	///experimental/work-in-progress
-	virtual bool	initializePolyhedralFeatures(int shiftVerticesByMargin=0);
+	virtual bool initializePolyhedralFeatures(int shiftVerticesByMargin = 0);
 
-	virtual void setPolyhedralFeatures(btConvexPolyhedron& polyhedron);
-	
+	virtual void setPolyhedralFeatures(btConvexPolyhedron & polyhedron);
 
-	const btConvexPolyhedron*	getConvexPolyhedron() const
+	const btConvexPolyhedron* getConvexPolyhedron() const
 	{
 		return m_polyhedron;
 	}
 
 	//brute force implementations
 
-	virtual btVector3	localGetSupportingVertexWithoutMargin(const btVector3& vec)const;
-	virtual void	batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const;
-	
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
-	
-	
-	virtual int	getNumVertices() const = 0 ;
+	virtual btVector3 localGetSupportingVertexWithoutMargin(const btVector3& vec) const;
+	virtual void batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const;
+
+	virtual void calculateLocalInertia(btScalar mass, btVector3 & inertia) const;
+
+	virtual int getNumVertices() const = 0;
 	virtual int getNumEdges() const = 0;
-	virtual void getEdge(int i,btVector3& pa,btVector3& pb) const = 0;
-	virtual void getVertex(int i,btVector3& vtx) const = 0;
-	virtual int	getNumPlanes() const = 0;
-	virtual void getPlane(btVector3& planeNormal,btVector3& planeSupport,int i ) const = 0;
-//	virtual int getIndex(int i) const = 0 ; 
+	virtual void getEdge(int i, btVector3& pa, btVector3& pb) const = 0;
+	virtual void getVertex(int i, btVector3& vtx) const = 0;
+	virtual int getNumPlanes() const = 0;
+	virtual void getPlane(btVector3 & planeNormal, btVector3 & planeSupport, int i) const = 0;
+	//	virtual int getIndex(int i) const = 0 ;
 
-	virtual	bool isInside(const btVector3& pt,btScalar tolerance) const = 0;
-	
+	virtual bool isInside(const btVector3& pt, btScalar tolerance) const = 0;
 };
 
-
 ///The btPolyhedralConvexAabbCachingShape adds aabb caching to the btPolyhedralConvexShape
 class btPolyhedralConvexAabbCachingShape : public btPolyhedralConvexShape
 {
+	btVector3 m_localAabbMin;
+	btVector3 m_localAabbMax;
+	bool m_isLocalAabbValid;
 
-	btVector3	m_localAabbMin;
-	btVector3	m_localAabbMax;
-	bool		m_isLocalAabbValid;
-		
 protected:
-
-	void setCachedLocalAabb (const btVector3& aabbMin, const btVector3& aabbMax)
+	void setCachedLocalAabb(const btVector3& aabbMin, const btVector3& aabbMax)
 	{
 		m_isLocalAabbValid = true;
 		m_localAabbMin = aabbMin;
 		m_localAabbMax = aabbMax;
 	}
 
-	inline void getCachedLocalAabb (btVector3& aabbMin, btVector3& aabbMax) const
+	inline void getCachedLocalAabb(btVector3& aabbMin, btVector3& aabbMax) const
 	{
 		btAssert(m_isLocalAabbValid);
 		aabbMin = m_localAabbMin;
@@ -97,25 +86,21 @@ protected:
 	}
 
 protected:
-
 	btPolyhedralConvexAabbCachingShape();
-	
-public:
-	
-	inline void getNonvirtualAabb(const btTransform& trans,btVector3& aabbMin,btVector3& aabbMax, btScalar margin) const
-	{
 
+public:
+	inline void getNonvirtualAabb(const btTransform& trans, btVector3& aabbMin, btVector3& aabbMax, btScalar margin) const
+	{
 		//lazy evaluation of local aabb
 		btAssert(m_isLocalAabbValid);
-		btTransformAabb(m_localAabbMin,m_localAabbMax,margin,trans,aabbMin,aabbMax);
+		btTransformAabb(m_localAabbMin, m_localAabbMax, margin, trans, aabbMin, aabbMax);
 	}
 
-	virtual void	setLocalScaling(const btVector3& scaling);
+	virtual void setLocalScaling(const btVector3& scaling);
 
-	virtual void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const;
-
-	void	recalcLocalAabb();
+	virtual void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
 
+	void recalcLocalAabb();
 };
 
-#endif //BT_POLYHEDRAL_CONVEX_SHAPE_H
+#endif  //BT_POLYHEDRAL_CONVEX_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btScaledBvhTriangleMeshShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btScaledBvhTriangleMeshShape.cpp
index 6a337c786c4..f4273199749 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btScaledBvhTriangleMeshShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btScaledBvhTriangleMeshShape.cpp
@@ -13,11 +13,10 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btScaledBvhTriangleMeshShape.h"
 
-btScaledBvhTriangleMeshShape::btScaledBvhTriangleMeshShape(btBvhTriangleMeshShape* childShape,const btVector3& localScaling)
-:m_localScaling(localScaling),m_bvhTriMeshShape(childShape)
+btScaledBvhTriangleMeshShape::btScaledBvhTriangleMeshShape(btBvhTriangleMeshShape* childShape, const btVector3& localScaling)
+	: m_localScaling(localScaling), m_bvhTriMeshShape(childShape)
 {
 	m_shapeType = SCALED_TRIANGLE_MESH_SHAPE_PROXYTYPE;
 }
@@ -26,55 +25,51 @@ btScaledBvhTriangleMeshShape::~btScaledBvhTriangleMeshShape()
 {
 }
 
-
 class btScaledTriangleCallback : public btTriangleCallback
 {
 	btTriangleCallback* m_originalCallback;
 
-	btVector3	m_localScaling;
+	btVector3 m_localScaling;
 
 public:
-
-	btScaledTriangleCallback(btTriangleCallback* originalCallback,const btVector3& localScaling)
-		:m_originalCallback(originalCallback),
-		m_localScaling(localScaling)
+	btScaledTriangleCallback(btTriangleCallback* originalCallback, const btVector3& localScaling)
+		: m_originalCallback(originalCallback),
+		  m_localScaling(localScaling)
 	{
 	}
 
 	virtual void processTriangle(btVector3* triangle, int partId, int triangleIndex)
 	{
 		btVector3 newTriangle[3];
-		newTriangle[0] = triangle[0]*m_localScaling;
-		newTriangle[1] = triangle[1]*m_localScaling;
-		newTriangle[2] = triangle[2]*m_localScaling;
-		m_originalCallback->processTriangle(&newTriangle[0],partId,triangleIndex);
+		newTriangle[0] = triangle[0] * m_localScaling;
+		newTriangle[1] = triangle[1] * m_localScaling;
+		newTriangle[2] = triangle[2] * m_localScaling;
+		m_originalCallback->processTriangle(&newTriangle[0], partId, triangleIndex);
 	}
 };
 
-void	btScaledBvhTriangleMeshShape::processAllTriangles(btTriangleCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax) const
+void btScaledBvhTriangleMeshShape::processAllTriangles(btTriangleCallback* callback, const btVector3& aabbMin, const btVector3& aabbMax) const
 {
-	btScaledTriangleCallback scaledCallback(callback,m_localScaling);
-	
-	btVector3 invLocalScaling(1.f/m_localScaling.getX(),1.f/m_localScaling.getY(),1.f/m_localScaling.getZ());
-	btVector3 scaledAabbMin,scaledAabbMax;
+	btScaledTriangleCallback scaledCallback(callback, m_localScaling);
+
+	btVector3 invLocalScaling(1.f / m_localScaling.getX(), 1.f / m_localScaling.getY(), 1.f / m_localScaling.getZ());
+	btVector3 scaledAabbMin, scaledAabbMax;
 
 	///support negative scaling
 	scaledAabbMin[0] = m_localScaling.getX() >= 0. ? aabbMin[0] * invLocalScaling[0] : aabbMax[0] * invLocalScaling[0];
 	scaledAabbMin[1] = m_localScaling.getY() >= 0. ? aabbMin[1] * invLocalScaling[1] : aabbMax[1] * invLocalScaling[1];
 	scaledAabbMin[2] = m_localScaling.getZ() >= 0. ? aabbMin[2] * invLocalScaling[2] : aabbMax[2] * invLocalScaling[2];
 	scaledAabbMin[3] = 0.f;
-	
+
 	scaledAabbMax[0] = m_localScaling.getX() <= 0. ? aabbMin[0] * invLocalScaling[0] : aabbMax[0] * invLocalScaling[0];
 	scaledAabbMax[1] = m_localScaling.getY() <= 0. ? aabbMin[1] * invLocalScaling[1] : aabbMax[1] * invLocalScaling[1];
 	scaledAabbMax[2] = m_localScaling.getZ() <= 0. ? aabbMin[2] * invLocalScaling[2] : aabbMax[2] * invLocalScaling[2];
 	scaledAabbMax[3] = 0.f;
-	
-	
-	m_bvhTriMeshShape->processAllTriangles(&scaledCallback,scaledAabbMin,scaledAabbMax);
+
+	m_bvhTriMeshShape->processAllTriangles(&scaledCallback, scaledAabbMin, scaledAabbMax);
 }
 
-
-void	btScaledBvhTriangleMeshShape::getAabb(const btTransform& trans,btVector3& aabbMin,btVector3& aabbMax) const
+void btScaledBvhTriangleMeshShape::getAabb(const btTransform& trans, btVector3& aabbMin, btVector3& aabbMax) const
 {
 	btVector3 localAabbMin = m_bvhTriMeshShape->getLocalAabbMin();
 	btVector3 localAabbMax = m_bvhTriMeshShape->getLocalAabbMax();
@@ -89,22 +84,21 @@ void	btScaledBvhTriangleMeshShape::getAabb(const btTransform& trans,btVector3& a
 	localAabbMax[1] = (m_localScaling.getY() <= 0.) ? tmpLocalAabbMin[1] : tmpLocalAabbMax[1];
 	localAabbMax[2] = (m_localScaling.getZ() <= 0.) ? tmpLocalAabbMin[2] : tmpLocalAabbMax[2];
 
-	btVector3 localHalfExtents = btScalar(0.5)*(localAabbMax-localAabbMin);
+	btVector3 localHalfExtents = btScalar(0.5) * (localAabbMax - localAabbMin);
 	btScalar margin = m_bvhTriMeshShape->getMargin();
-	localHalfExtents += btVector3(margin,margin,margin);
-	btVector3 localCenter = btScalar(0.5)*(localAabbMax+localAabbMin);
-	
-	btMatrix3x3 abs_b = trans.getBasis().absolute();  
+	localHalfExtents += btVector3(margin, margin, margin);
+	btVector3 localCenter = btScalar(0.5) * (localAabbMax + localAabbMin);
+
+	btMatrix3x3 abs_b = trans.getBasis().absolute();
 
 	btVector3 center = trans(localCenter);
 
-    btVector3 extent = localHalfExtents.dot3(abs_b[0], abs_b[1], abs_b[2]);
+	btVector3 extent = localHalfExtents.dot3(abs_b[0], abs_b[1], abs_b[2]);
 	aabbMin = center - extent;
 	aabbMax = center + extent;
-
 }
 
-void	btScaledBvhTriangleMeshShape::setLocalScaling(const btVector3& scaling)
+void btScaledBvhTriangleMeshShape::setLocalScaling(const btVector3& scaling)
 {
 	m_localScaling = scaling;
 }
@@ -114,8 +108,8 @@ const btVector3& btScaledBvhTriangleMeshShape::getLocalScaling() const
 	return m_localScaling;
 }
 
-void	btScaledBvhTriangleMeshShape::calculateLocalInertia(btScalar mass,btVector3& inertia) const
+void btScaledBvhTriangleMeshShape::calculateLocalInertia(btScalar mass, btVector3& inertia) const
 {
 	///don't make this a movable object!
-//	btAssert(0);
+	//	btAssert(0);
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btScaledBvhTriangleMeshShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btScaledBvhTriangleMeshShape.h
index 39049eaf083..4d6feb61af5 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btScaledBvhTriangleMeshShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btScaledBvhTriangleMeshShape.h
@@ -18,78 +18,69 @@ subject to the following restrictions:
 
 #include "BulletCollision/CollisionShapes/btBvhTriangleMeshShape.h"
 
-
 ///The btScaledBvhTriangleMeshShape allows to instance a scaled version of an existing btBvhTriangleMeshShape.
 ///Note that each btBvhTriangleMeshShape still can have its own local scaling, independent from this btScaledBvhTriangleMeshShape 'localScaling'
-ATTRIBUTE_ALIGNED16(class) btScaledBvhTriangleMeshShape : public btConcaveShape
+ATTRIBUTE_ALIGNED16(class)
+btScaledBvhTriangleMeshShape : public btConcaveShape
 {
-	
-	
-	btVector3	m_localScaling;
+	btVector3 m_localScaling;
 
-	btBvhTriangleMeshShape*	m_bvhTriMeshShape;
+	btBvhTriangleMeshShape* m_bvhTriMeshShape;
 
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-
-	btScaledBvhTriangleMeshShape(btBvhTriangleMeshShape* childShape,const btVector3& localScaling);
+	btScaledBvhTriangleMeshShape(btBvhTriangleMeshShape * childShape, const btVector3& localScaling);
 
 	virtual ~btScaledBvhTriangleMeshShape();
 
-
-	virtual void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const;
-	virtual void	setLocalScaling(const btVector3& scaling);
+	virtual void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
+	virtual void setLocalScaling(const btVector3& scaling);
 	virtual const btVector3& getLocalScaling() const;
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
+	virtual void calculateLocalInertia(btScalar mass, btVector3 & inertia) const;
 
-	virtual void	processAllTriangles(btTriangleCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax) const;
+	virtual void processAllTriangles(btTriangleCallback * callback, const btVector3& aabbMin, const btVector3& aabbMax) const;
 
-	btBvhTriangleMeshShape*	getChildShape()
+	btBvhTriangleMeshShape* getChildShape()
 	{
 		return m_bvhTriMeshShape;
 	}
 
-	const btBvhTriangleMeshShape*	getChildShape() const
+	const btBvhTriangleMeshShape* getChildShape() const
 	{
 		return m_bvhTriMeshShape;
 	}
 
 	//debugging
-	virtual const char*	getName()const {return "SCALEDBVHTRIANGLEMESH";}
+	virtual const char* getName() const { return "SCALEDBVHTRIANGLEMESH"; }
 
-	virtual	int	calculateSerializeBufferSize() const;
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
-
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 };
 
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
-struct	btScaledTriangleMeshShapeData
+struct btScaledTriangleMeshShapeData
 {
-	btTriangleMeshShapeData	m_trimeshShapeData;
+	btTriangleMeshShapeData m_trimeshShapeData;
 
-	btVector3FloatData	m_localScaling;
+	btVector3FloatData m_localScaling;
 };
 
-
-SIMD_FORCE_INLINE	int	btScaledBvhTriangleMeshShape::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btScaledBvhTriangleMeshShape::calculateSerializeBufferSize() const
 {
 	return sizeof(btScaledTriangleMeshShapeData);
 }
 
-
 ///fills the dataBuffer and returns the struct name (and 0 on failure)
-SIMD_FORCE_INLINE	const char*	btScaledBvhTriangleMeshShape::serialize(void* dataBuffer, btSerializer* serializer) const
+SIMD_FORCE_INLINE const char* btScaledBvhTriangleMeshShape::serialize(void* dataBuffer, btSerializer* serializer) const
 {
-	btScaledTriangleMeshShapeData* scaledMeshData = (btScaledTriangleMeshShapeData*) dataBuffer;
-	m_bvhTriMeshShape->serialize(&scaledMeshData->m_trimeshShapeData,serializer);
+	btScaledTriangleMeshShapeData* scaledMeshData = (btScaledTriangleMeshShapeData*)dataBuffer;
+	m_bvhTriMeshShape->serialize(&scaledMeshData->m_trimeshShapeData, serializer);
 	scaledMeshData->m_trimeshShapeData.m_collisionShapeData.m_shapeType = SCALED_TRIANGLE_MESH_SHAPE_PROXYTYPE;
 	m_localScaling.serializeFloat(scaledMeshData->m_localScaling);
 	return "btScaledTriangleMeshShapeData";
 }
 
-
-#endif //BT_SCALED_BVH_TRIANGLE_MESH_SHAPE_H
+#endif  //BT_SCALED_BVH_TRIANGLE_MESH_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btSdfCollisionShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btSdfCollisionShape.cpp
index 828acda4709..4a95dbea4f2 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btSdfCollisionShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btSdfCollisionShape.cpp
@@ -9,10 +9,9 @@ struct btSdfCollisionShapeInternalData
 	btMiniSDF m_sdf;
 
 	btSdfCollisionShapeInternalData()
-		:m_localScaling(1,1,1),
-		m_margin(0)
+		: m_localScaling(1, 1, 1),
+		  m_margin(0)
 	{
-
 	}
 };
 
@@ -26,8 +25,6 @@ btSdfCollisionShape::btSdfCollisionShape()
 	m_shapeType = SDF_SHAPE_PROXYTYPE;
 	m_data = new btSdfCollisionShapeInternalData();
 
-
-	
 	//"E:/develop/bullet3/data/toys/ground_hole64_64_8.cdf");//ground_cube.cdf");
 	/*unsigned int field_id=0;
 	Eigen::Vector3d x (1,10,1);
@@ -35,25 +32,22 @@ btSdfCollisionShape::btSdfCollisionShape()
 	double dist = m_data->m_sdf.interpolate(field_id, x, &gradient);
 	printf("dist=%g\n", dist);
 	*/
-
 }
 btSdfCollisionShape::~btSdfCollisionShape()
 {
 	delete m_data;
 }
 
-void btSdfCollisionShape::getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const
+void btSdfCollisionShape::getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
 {
 	btAssert(m_data->m_sdf.isValid());
 	btVector3 localAabbMin = m_data->m_sdf.m_domain.m_min;
 	btVector3 localAabbMax = m_data->m_sdf.m_domain.m_max;
 	btScalar margin(0);
-	btTransformAabb(localAabbMin,localAabbMax,margin,t,aabbMin,aabbMax);
-
+	btTransformAabb(localAabbMin, localAabbMax, margin, t, aabbMin, aabbMax);
 }
 
-	
-void	btSdfCollisionShape::setLocalScaling(const btVector3& scaling)
+void btSdfCollisionShape::setLocalScaling(const btVector3& scaling)
 {
 	m_data->m_localScaling = scaling;
 }
@@ -61,39 +55,38 @@ const btVector3& btSdfCollisionShape::getLocalScaling() const
 {
 	return m_data->m_localScaling;
 }
-void	btSdfCollisionShape::calculateLocalInertia(btScalar mass,btVector3& inertia) const
+void btSdfCollisionShape::calculateLocalInertia(btScalar mass, btVector3& inertia) const
 {
-	inertia.setValue(0,0,0);
+	inertia.setValue(0, 0, 0);
 }
-const char*	btSdfCollisionShape::getName()const
+const char* btSdfCollisionShape::getName() const
 {
 	return "btSdfCollisionShape";
 }
-void	btSdfCollisionShape::setMargin(btScalar margin)
+void btSdfCollisionShape::setMargin(btScalar margin)
 {
 	m_data->m_margin = margin;
 }
-btScalar	btSdfCollisionShape::getMargin() const
+btScalar btSdfCollisionShape::getMargin() const
 {
 	return m_data->m_margin;
 }
 
-void	btSdfCollisionShape::processAllTriangles(btTriangleCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax) const
+void btSdfCollisionShape::processAllTriangles(btTriangleCallback* callback, const btVector3& aabbMin, const btVector3& aabbMax) const
 {
 	//not yet
 }
 
-
 bool btSdfCollisionShape::queryPoint(const btVector3& ptInSDF, btScalar& distOut, btVector3& normal)
 {
 	int field = 0;
 	btVector3 grad;
 	double dist;
-	bool hasResult = m_data->m_sdf.interpolate(field,dist, ptInSDF,&grad);
+	bool hasResult = m_data->m_sdf.interpolate(field, dist, ptInSDF, &grad);
 	if (hasResult)
 	{
-		normal.setValue(grad[0],grad[1],grad[2]);
-		distOut= dist;
+		normal.setValue(grad[0], grad[1], grad[2]);
+		distOut = dist;
 	}
 	return hasResult;
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btSdfCollisionShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btSdfCollisionShape.h
index 6e32db9cd86..3989d6245ed 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btSdfCollisionShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btSdfCollisionShape.h
@@ -8,23 +8,22 @@ class btSdfCollisionShape : public btConcaveShape
 	struct btSdfCollisionShapeInternalData* m_data;
 
 public:
-		
 	btSdfCollisionShape();
 	virtual ~btSdfCollisionShape();
-	
+
 	bool initializeSDF(const char* sdfData, int sizeInBytes);
 
-	virtual void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const;
-	virtual void	setLocalScaling(const btVector3& scaling);
+	virtual void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
+	virtual void setLocalScaling(const btVector3& scaling);
 	virtual const btVector3& getLocalScaling() const;
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
-	virtual const char*	getName()const;
-	virtual void	setMargin(btScalar margin);
-	virtual btScalar	getMargin() const;
+	virtual void calculateLocalInertia(btScalar mass, btVector3& inertia) const;
+	virtual const char* getName() const;
+	virtual void setMargin(btScalar margin);
+	virtual btScalar getMargin() const;
 
-	virtual void	processAllTriangles(btTriangleCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax) const;
+	virtual void processAllTriangles(btTriangleCallback* callback, const btVector3& aabbMin, const btVector3& aabbMax) const;
 
 	bool queryPoint(const btVector3& ptInSDF, btScalar& distOut, btVector3& normal);
 };
 
-#endif //BT_SDF_COLLISION_SHAPE_H
+#endif  //BT_SDF_COLLISION_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btShapeHull.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btShapeHull.cpp
index 9f712fe555a..a2c490faf9d 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btShapeHull.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btShapeHull.cpp
@@ -15,53 +15,48 @@ subject to the following restrictions:
 
 //btShapeHull was implemented by John McCutchan.
 
-
 #include "btShapeHull.h"
 #include "LinearMath/btConvexHull.h"
 
 #define NUM_UNITSPHERE_POINTS 42
 #define NUM_UNITSPHERE_POINTS_HIGHRES 256
 
-
-btShapeHull::btShapeHull (const btConvexShape* shape)
+btShapeHull::btShapeHull(const btConvexShape* shape)
 {
 	m_shape = shape;
-	m_vertices.clear ();
+	m_vertices.clear();
 	m_indices.clear();
 	m_numIndices = 0;
 }
 
-btShapeHull::~btShapeHull ()
+btShapeHull::~btShapeHull()
 {
-	m_indices.clear();	
-	m_vertices.clear ();
+	m_indices.clear();
+	m_vertices.clear();
 }
 
-bool
-btShapeHull::buildHull (btScalar /*margin*/, int highres)
+bool btShapeHull::buildHull(btScalar /*margin*/, int highres)
 {
-	int numSampleDirections = highres? NUM_UNITSPHERE_POINTS_HIGHRES:NUM_UNITSPHERE_POINTS;
-	{
-		int numPDA = m_shape->getNumPreferredPenetrationDirections();
-		if (numPDA)
-		{
-			for (int i=0;i<numPDA;i++)
-			{
-				btVector3 norm;
-				m_shape->getPreferredPenetrationDirection(i,norm);
-				getUnitSpherePoints(highres)[numSampleDirections] = norm;
-				numSampleDirections++;
-			}
-		}
-	}
-
-	btVector3 supportPoints[NUM_UNITSPHERE_POINTS_HIGHRES+MAX_PREFERRED_PENETRATION_DIRECTIONS*2];
+	
+	int numSampleDirections = highres ? NUM_UNITSPHERE_POINTS_HIGHRES : NUM_UNITSPHERE_POINTS;
+	btVector3 supportPoints[NUM_UNITSPHERE_POINTS_HIGHRES + MAX_PREFERRED_PENETRATION_DIRECTIONS * 2];
 	int i;
 	for (i = 0; i < numSampleDirections; i++)
 	{
 		supportPoints[i] = m_shape->localGetSupportingVertex(getUnitSpherePoints(highres)[i]);
 	}
 
+	int numPDA = m_shape->getNumPreferredPenetrationDirections();
+	if (numPDA)
+	{
+		for (int s = 0; s < numPDA; s++)
+		{
+			btVector3 norm;
+			m_shape->getPreferredPenetrationDirection(s, norm);
+			supportPoints[i++] = m_shape->localGetSupportingVertex(norm);
+			numSampleDirections++;
+		}
+	}
 	HullDesc hd;
 	hd.mFlags = QF_TRIANGLES;
 	hd.mVcount = static_cast<unsigned int>(numSampleDirections);
@@ -71,18 +66,17 @@ btShapeHull::buildHull (btScalar /*margin*/, int highres)
 	hd.mVertexStride = sizeof(btVector3);
 #else
 	hd.mVertices = &supportPoints[0];
-	hd.mVertexStride = sizeof (btVector3);
+	hd.mVertexStride = sizeof(btVector3);
 #endif
 
 	HullLibrary hl;
 	HullResult hr;
-	if (hl.CreateConvexHull (hd, hr) == QE_FAIL)
+	if (hl.CreateConvexHull(hd, hr) == QE_FAIL)
 	{
 		return false;
 	}
 
-	m_vertices.resize (static_cast<int>(hr.mNumOutputVertices));
-
+	m_vertices.resize(static_cast<int>(hr.mNumOutputVertices));
 
 	for (i = 0; i < static_cast<int>(hr.mNumOutputVertices); i++)
 	{
@@ -96,338 +90,332 @@ btShapeHull::buildHull (btScalar /*margin*/, int highres)
 	}
 
 	// free temporary hull result that we just copied
-	hl.ReleaseResult (hr);
+	hl.ReleaseResult(hr);
 
 	return true;
 }
 
-int
-btShapeHull::numTriangles () const
+int btShapeHull::numTriangles() const
 {
 	return static_cast<int>(m_numIndices / 3);
 }
 
-int
-btShapeHull::numVertices () const
+int btShapeHull::numVertices() const
 {
-	return m_vertices.size ();
+	return m_vertices.size();
 }
 
-int
-btShapeHull::numIndices () const
+int btShapeHull::numIndices() const
 {
 	return static_cast<int>(m_numIndices);
 }
 
-
 btVector3* btShapeHull::getUnitSpherePoints(int highres)
 {
 	static btVector3 sUnitSpherePointsHighres[NUM_UNITSPHERE_POINTS_HIGHRES + MAX_PREFERRED_PENETRATION_DIRECTIONS * 2] =
-	{
-		btVector3(btScalar(0.997604), btScalar(0.067004), btScalar(0.017144)),
-		btVector3(btScalar(0.984139), btScalar(-0.086784), btScalar(-0.154427)),
-		btVector3(btScalar(0.971065), btScalar(0.124164), btScalar(-0.203224)),
-		btVector3(btScalar(0.955844), btScalar(0.291173), btScalar(-0.037704)),
-		btVector3(btScalar(0.957405), btScalar(0.212238), btScalar(0.195157)),
-		btVector3(btScalar(0.971650), btScalar(-0.012709), btScalar(0.235561)),
-		btVector3(btScalar(0.984920), btScalar(-0.161831), btScalar(0.059695)),
-		btVector3(btScalar(0.946673), btScalar(-0.299288), btScalar(-0.117536)),
-		btVector3(btScalar(0.922670), btScalar(-0.219186), btScalar(-0.317019)),
-		btVector3(btScalar(0.928134), btScalar(-0.007265), btScalar(-0.371867)),
-		btVector3(btScalar(0.875642), btScalar(0.198434), btScalar(-0.439988)),
-		btVector3(btScalar(0.908035), btScalar(0.325975), btScalar(-0.262562)),
-		btVector3(btScalar(0.864519), btScalar(0.488706), btScalar(-0.116755)),
-		btVector3(btScalar(0.893009), btScalar(0.428046), btScalar(0.137185)),
-		btVector3(btScalar(0.857494), btScalar(0.362137), btScalar(0.364776)),
-		btVector3(btScalar(0.900815), btScalar(0.132524), btScalar(0.412987)),
-		btVector3(btScalar(0.934964), btScalar(-0.241739), btScalar(0.259179)),
-		btVector3(btScalar(0.894570), btScalar(-0.103504), btScalar(0.434263)),
-		btVector3(btScalar(0.922085), btScalar(-0.376668), btScalar(0.086241)),
-		btVector3(btScalar(0.862177), btScalar(-0.499154), btScalar(-0.085330)),
-		btVector3(btScalar(0.861982), btScalar(-0.420218), btScalar(-0.282861)),
-		btVector3(btScalar(0.818076), btScalar(-0.328256), btScalar(-0.471804)),
-		btVector3(btScalar(0.762657), btScalar(-0.179329), btScalar(-0.621124)),
-		btVector3(btScalar(0.826857), btScalar(0.019760), btScalar(-0.561786)),
-		btVector3(btScalar(0.731434), btScalar(0.206599), btScalar(-0.649817)),
-		btVector3(btScalar(0.769486), btScalar(0.379052), btScalar(-0.513770)),
-		btVector3(btScalar(0.796806), btScalar(0.507176), btScalar(-0.328145)),
-		btVector3(btScalar(0.679722), btScalar(0.684101), btScalar(-0.264123)),
-		btVector3(btScalar(0.786854), btScalar(0.614886), btScalar(0.050912)),
-		btVector3(btScalar(0.769486), btScalar(0.571141), btScalar(0.285139)),
-		btVector3(btScalar(0.707432), btScalar(0.492789), btScalar(0.506288)),
-		btVector3(btScalar(0.774560), btScalar(0.268037), btScalar(0.572652)),
-		btVector3(btScalar(0.796220), btScalar(0.031230), btScalar(0.604077)),
-		btVector3(btScalar(0.837395), btScalar(-0.320285), btScalar(0.442461)),
-		btVector3(btScalar(0.848127), btScalar(-0.450548), btScalar(0.278307)),
-		btVector3(btScalar(0.775536), btScalar(-0.206354), btScalar(0.596465)),
-		btVector3(btScalar(0.816320), btScalar(-0.567007), btScalar(0.109469)),
-		btVector3(btScalar(0.741191), btScalar(-0.668690), btScalar(-0.056832)),
-		btVector3(btScalar(0.755632), btScalar(-0.602975), btScalar(-0.254949)),
-		btVector3(btScalar(0.720311), btScalar(-0.521318), btScalar(-0.457165)),
-		btVector3(btScalar(0.670746), btScalar(-0.386583), btScalar(-0.632835)),
-		btVector3(btScalar(0.587031), btScalar(-0.219769), btScalar(-0.778836)),
-		btVector3(btScalar(0.676015), btScalar(-0.003182), btScalar(-0.736676)),
-		btVector3(btScalar(0.566932), btScalar(0.186963), btScalar(-0.802064)),
-		btVector3(btScalar(0.618254), btScalar(0.398105), btScalar(-0.677533)),
-		btVector3(btScalar(0.653964), btScalar(0.575224), btScalar(-0.490933)),
-		btVector3(btScalar(0.525367), btScalar(0.743205), btScalar(-0.414028)),
-		btVector3(btScalar(0.506439), btScalar(0.836528), btScalar(-0.208885)),
-		btVector3(btScalar(0.651427), btScalar(0.756426), btScalar(-0.056247)),
-		btVector3(btScalar(0.641670), btScalar(0.745149), btScalar(0.180908)),
-		btVector3(btScalar(0.602643), btScalar(0.687211), btScalar(0.405180)),
-		btVector3(btScalar(0.516586), btScalar(0.596999), btScalar(0.613447)),
-		btVector3(btScalar(0.602252), btScalar(0.387801), btScalar(0.697573)),
-		btVector3(btScalar(0.646549), btScalar(0.153911), btScalar(0.746956)),
-		btVector3(btScalar(0.650842), btScalar(-0.087756), btScalar(0.753983)),
-		btVector3(btScalar(0.740411), btScalar(-0.497404), btScalar(0.451830)),
-		btVector3(btScalar(0.726946), btScalar(-0.619890), btScalar(0.295093)),
-		btVector3(btScalar(0.637768), btScalar(-0.313092), btScalar(0.703624)),
-		btVector3(btScalar(0.678942), btScalar(-0.722934), btScalar(0.126645)),
-		btVector3(btScalar(0.489072), btScalar(-0.867195), btScalar(-0.092942)),
-		btVector3(btScalar(0.622742), btScalar(-0.757541), btScalar(-0.194636)),
-		btVector3(btScalar(0.596788), btScalar(-0.693576), btScalar(-0.403098)),
-		btVector3(btScalar(0.550150), btScalar(-0.582172), btScalar(-0.598287)),
-		btVector3(btScalar(0.474436), btScalar(-0.429745), btScalar(-0.768101)),
-		btVector3(btScalar(0.372574), btScalar(-0.246016), btScalar(-0.894583)),
-		btVector3(btScalar(0.480095), btScalar(-0.026513), btScalar(-0.876626)),
-		btVector3(btScalar(0.352474), btScalar(0.177242), btScalar(-0.918787)),
-		btVector3(btScalar(0.441848), btScalar(0.374386), btScalar(-0.814946)),
-		btVector3(btScalar(0.492389), btScalar(0.582223), btScalar(-0.646693)),
-		btVector3(btScalar(0.343498), btScalar(0.866080), btScalar(-0.362693)),
-		btVector3(btScalar(0.362036), btScalar(0.745149), btScalar(-0.559639)),
-		btVector3(btScalar(0.334131), btScalar(0.937044), btScalar(-0.099774)),
-		btVector3(btScalar(0.486925), btScalar(0.871718), btScalar(0.052473)),
-		btVector3(btScalar(0.452776), btScalar(0.845665), btScalar(0.281820)),
-		btVector3(btScalar(0.399503), btScalar(0.771785), btScalar(0.494576)),
-		btVector3(btScalar(0.296469), btScalar(0.673018), btScalar(0.677469)),
-		btVector3(btScalar(0.392088), btScalar(0.479179), btScalar(0.785213)),
-		btVector3(btScalar(0.452190), btScalar(0.252094), btScalar(0.855286)),
-		btVector3(btScalar(0.478339), btScalar(0.013149), btScalar(0.877928)),
-		btVector3(btScalar(0.481656), btScalar(-0.219380), btScalar(0.848259)),
-		btVector3(btScalar(0.615327), btScalar(-0.494293), btScalar(0.613837)),
-		btVector3(btScalar(0.594642), btScalar(-0.650414), btScalar(0.472325)),
-		btVector3(btScalar(0.562249), btScalar(-0.771345), btScalar(0.297631)),
-		btVector3(btScalar(0.467411), btScalar(-0.437133), btScalar(0.768231)),
-		btVector3(btScalar(0.519513), btScalar(-0.847947), btScalar(0.103808)),
-		btVector3(btScalar(0.297640), btScalar(-0.938159), btScalar(-0.176288)),
-		btVector3(btScalar(0.446727), btScalar(-0.838615), btScalar(-0.311359)),
-		btVector3(btScalar(0.331790), btScalar(-0.942437), btScalar(0.040762)),
-		btVector3(btScalar(0.413358), btScalar(-0.748403), btScalar(-0.518259)),
-		btVector3(btScalar(0.347596), btScalar(-0.621640), btScalar(-0.701737)),
-		btVector3(btScalar(0.249831), btScalar(-0.456186), btScalar(-0.853984)),
-		btVector3(btScalar(0.131772), btScalar(-0.262931), btScalar(-0.955678)),
-		btVector3(btScalar(0.247099), btScalar(-0.042261), btScalar(-0.967975)),
-		btVector3(btScalar(0.113624), btScalar(0.165965), btScalar(-0.979491)),
-		btVector3(btScalar(0.217438), btScalar(0.374580), btScalar(-0.901220)),
-		btVector3(btScalar(0.307983), btScalar(0.554615), btScalar(-0.772786)),
-		btVector3(btScalar(0.166702), btScalar(0.953181), btScalar(-0.252021)),
-		btVector3(btScalar(0.172751), btScalar(0.844499), btScalar(-0.506743)),
-		btVector3(btScalar(0.177630), btScalar(0.711125), btScalar(-0.679876)),
-		btVector3(btScalar(0.120064), btScalar(0.992260), btScalar(-0.030482)),
-		btVector3(btScalar(0.289640), btScalar(0.949098), btScalar(0.122546)),
-		btVector3(btScalar(0.239879), btScalar(0.909047), btScalar(0.340377)),
-		btVector3(btScalar(0.181142), btScalar(0.821363), btScalar(0.540641)),
-		btVector3(btScalar(0.066986), btScalar(0.719097), btScalar(0.691327)),
-		btVector3(btScalar(0.156750), btScalar(0.545478), btScalar(0.823079)),
-		btVector3(btScalar(0.236172), btScalar(0.342306), btScalar(0.909353)),
-		btVector3(btScalar(0.277541), btScalar(0.112693), btScalar(0.953856)),
-		btVector3(btScalar(0.295299), btScalar(-0.121974), btScalar(0.947415)),
-		btVector3(btScalar(0.287883), btScalar(-0.349254), btScalar(0.891591)),
-		btVector3(btScalar(0.437165), btScalar(-0.634666), btScalar(0.636869)),
-		btVector3(btScalar(0.407113), btScalar(-0.784954), btScalar(0.466664)),
-		btVector3(btScalar(0.375111), btScalar(-0.888193), btScalar(0.264839)),
-		btVector3(btScalar(0.275394), btScalar(-0.560591), btScalar(0.780723)),
-		btVector3(btScalar(0.122015), btScalar(-0.992209), btScalar(-0.024821)),
-		btVector3(btScalar(0.087866), btScalar(-0.966156), btScalar(-0.241676)),
-		btVector3(btScalar(0.239489), btScalar(-0.885665), btScalar(-0.397437)),
-		btVector3(btScalar(0.167287), btScalar(-0.965184), btScalar(0.200817)),
-		btVector3(btScalar(0.201632), btScalar(-0.776789), btScalar(-0.596335)),
-		btVector3(btScalar(0.122015), btScalar(-0.637971), btScalar(-0.760098)),
-		btVector3(btScalar(0.008054), btScalar(-0.464741), btScalar(-0.885214)),
-		btVector3(btScalar(-0.116054), btScalar(-0.271096), btScalar(-0.955482)),
-		btVector3(btScalar(-0.000727), btScalar(-0.056065), btScalar(-0.998424)),
-		btVector3(btScalar(-0.134007), btScalar(0.152939), btScalar(-0.978905)),
-		btVector3(btScalar(-0.025900), btScalar(0.366026), btScalar(-0.930108)),
-		btVector3(btScalar(0.081231), btScalar(0.557337), btScalar(-0.826072)),
-		btVector3(btScalar(-0.002874), btScalar(0.917213), btScalar(-0.398023)),
-		btVector3(btScalar(-0.050683), btScalar(0.981761), btScalar(-0.182534)),
-		btVector3(btScalar(-0.040536), btScalar(0.710153), btScalar(-0.702713)),
-		btVector3(btScalar(-0.139081), btScalar(0.827973), btScalar(-0.543048)),
-		btVector3(btScalar(-0.101029), btScalar(0.994010), btScalar(0.041152)),
-		btVector3(btScalar(0.069328), btScalar(0.978067), btScalar(0.196133)),
-		btVector3(btScalar(0.023860), btScalar(0.911380), btScalar(0.410645)),
-		btVector3(btScalar(-0.153521), btScalar(0.736789), btScalar(0.658145)),
-		btVector3(btScalar(-0.070002), btScalar(0.591750), btScalar(0.802780)),
-		btVector3(btScalar(0.002590), btScalar(0.312948), btScalar(0.949562)),
-		btVector3(btScalar(0.090988), btScalar(-0.020680), btScalar(0.995627)),
-		btVector3(btScalar(0.088842), btScalar(-0.250099), btScalar(0.964006)),
-		btVector3(btScalar(0.083378), btScalar(-0.470185), btScalar(0.878318)),
-		btVector3(btScalar(0.240074), btScalar(-0.749764), btScalar(0.616374)),
-		btVector3(btScalar(0.210803), btScalar(-0.885860), btScalar(0.412987)),
-		btVector3(btScalar(0.077524), btScalar(-0.660524), btScalar(0.746565)),
-		btVector3(btScalar(-0.096736), btScalar(-0.990070), btScalar(-0.100945)),
-		btVector3(btScalar(-0.052634), btScalar(-0.990264), btScalar(0.127426)),
-		btVector3(btScalar(-0.106102), btScalar(-0.938354), btScalar(-0.328340)),
-		btVector3(btScalar(0.013323), btScalar(-0.863112), btScalar(-0.504596)),
-		btVector3(btScalar(-0.002093), btScalar(-0.936993), btScalar(0.349161)),
-		btVector3(btScalar(-0.106297), btScalar(-0.636610), btScalar(-0.763612)),
-		btVector3(btScalar(-0.229430), btScalar(-0.463769), btScalar(-0.855546)),
-		btVector3(btScalar(-0.245236), btScalar(-0.066175), btScalar(-0.966999)),
-		btVector3(btScalar(-0.351587), btScalar(-0.270513), btScalar(-0.896145)),
-		btVector3(btScalar(-0.370906), btScalar(0.133108), btScalar(-0.918982)),
-		btVector3(btScalar(-0.264360), btScalar(0.346000), btScalar(-0.900049)),
-		btVector3(btScalar(-0.151375), btScalar(0.543728), btScalar(-0.825291)),
-		btVector3(btScalar(-0.218697), btScalar(0.912741), btScalar(-0.344346)),
-		btVector3(btScalar(-0.274507), btScalar(0.953764), btScalar(-0.121635)),
-		btVector3(btScalar(-0.259677), btScalar(0.692266), btScalar(-0.673044)),
-		btVector3(btScalar(-0.350416), btScalar(0.798810), btScalar(-0.488786)),
-		btVector3(btScalar(-0.320170), btScalar(0.941127), btScalar(0.108297)),
-		btVector3(btScalar(-0.147667), btScalar(0.952792), btScalar(0.265034)),
-		btVector3(btScalar(-0.188061), btScalar(0.860636), btScalar(0.472910)),
-		btVector3(btScalar(-0.370906), btScalar(0.739900), btScalar(0.560941)),
-		btVector3(btScalar(-0.297143), btScalar(0.585334), btScalar(0.754178)),
-		btVector3(btScalar(-0.189622), btScalar(0.428241), btScalar(0.883393)),
-		btVector3(btScalar(-0.091272), btScalar(0.098695), btScalar(0.990747)),
-		btVector3(btScalar(-0.256945), btScalar(0.228375), btScalar(0.938827)),
-		btVector3(btScalar(-0.111761), btScalar(-0.133251), btScalar(0.984696)),
-		btVector3(btScalar(-0.118006), btScalar(-0.356253), btScalar(0.926725)),
-		btVector3(btScalar(-0.119372), btScalar(-0.563896), btScalar(0.817029)),
-		btVector3(btScalar(0.041228), btScalar(-0.833949), btScalar(0.550010)),
-		btVector3(btScalar(-0.121909), btScalar(-0.736543), btScalar(0.665172)),
-		btVector3(btScalar(-0.307681), btScalar(-0.931160), btScalar(-0.195026)),
-		btVector3(btScalar(-0.283679), btScalar(-0.957990), btScalar(0.041348)),
-		btVector3(btScalar(-0.227284), btScalar(-0.935243), btScalar(0.270890)),
-		btVector3(btScalar(-0.293436), btScalar(-0.858252), btScalar(-0.420860)),
-		btVector3(btScalar(-0.175767), btScalar(-0.780677), btScalar(-0.599262)),
-		btVector3(btScalar(-0.170108), btScalar(-0.858835), btScalar(0.482865)),
-		btVector3(btScalar(-0.332854), btScalar(-0.635055), btScalar(-0.696857)),
-		btVector3(btScalar(-0.447791), btScalar(-0.445299), btScalar(-0.775128)),
-		btVector3(btScalar(-0.470622), btScalar(-0.074146), btScalar(-0.879164)),
-		btVector3(btScalar(-0.639417), btScalar(-0.340505), btScalar(-0.689049)),
-		btVector3(btScalar(-0.598438), btScalar(0.104722), btScalar(-0.794256)),
-		btVector3(btScalar(-0.488575), btScalar(0.307699), btScalar(-0.816313)),
-		btVector3(btScalar(-0.379882), btScalar(0.513592), btScalar(-0.769077)),
-		btVector3(btScalar(-0.425740), btScalar(0.862775), btScalar(-0.272516)),
-		btVector3(btScalar(-0.480769), btScalar(0.875412), btScalar(-0.048439)),
-		btVector3(btScalar(-0.467890), btScalar(0.648716), btScalar(-0.600043)),
-		btVector3(btScalar(-0.543799), btScalar(0.730956), btScalar(-0.411881)),
-		btVector3(btScalar(-0.516284), btScalar(0.838277), btScalar(0.174076)),
-		btVector3(btScalar(-0.353343), btScalar(0.876384), btScalar(0.326519)),
-		btVector3(btScalar(-0.572875), btScalar(0.614497), btScalar(0.542007)),
-		btVector3(btScalar(-0.503600), btScalar(0.497261), btScalar(0.706161)),
-		btVector3(btScalar(-0.530920), btScalar(0.754870), btScalar(0.384685)),
-		btVector3(btScalar(-0.395884), btScalar(0.366414), btScalar(0.841818)),
-		btVector3(btScalar(-0.300656), btScalar(0.001678), btScalar(0.953661)),
-		btVector3(btScalar(-0.461060), btScalar(0.146912), btScalar(0.875000)),
-		btVector3(btScalar(-0.315486), btScalar(-0.232212), btScalar(0.919893)),
-		btVector3(btScalar(-0.323682), btScalar(-0.449187), btScalar(0.832644)),
-		btVector3(btScalar(-0.318999), btScalar(-0.639527), btScalar(0.699134)),
-		btVector3(btScalar(-0.496771), btScalar(-0.866029), btScalar(-0.055271)),
-		btVector3(btScalar(-0.496771), btScalar(-0.816257), btScalar(-0.294377)),
-		btVector3(btScalar(-0.456377), btScalar(-0.869528), btScalar(0.188130)),
-		btVector3(btScalar(-0.380858), btScalar(-0.827144), btScalar(0.412792)),
-		btVector3(btScalar(-0.449352), btScalar(-0.727405), btScalar(-0.518259)),
-		btVector3(btScalar(-0.570533), btScalar(-0.551064), btScalar(-0.608632)),
-		btVector3(btScalar(-0.656394), btScalar(-0.118280), btScalar(-0.744874)),
-		btVector3(btScalar(-0.756696), btScalar(-0.438105), btScalar(-0.484882)),
-		btVector3(btScalar(-0.801773), btScalar(-0.204798), btScalar(-0.561005)),
-		btVector3(btScalar(-0.785186), btScalar(0.038618), btScalar(-0.617805)),
-		btVector3(btScalar(-0.709082), btScalar(0.262399), btScalar(-0.654306)),
-		btVector3(btScalar(-0.583412), btScalar(0.462265), btScalar(-0.667383)),
-		btVector3(btScalar(-0.616001), btScalar(0.761286), btScalar(-0.201272)),
-		btVector3(btScalar(-0.660687), btScalar(0.750204), btScalar(0.020072)),
-		btVector3(btScalar(-0.744987), btScalar(0.435823), btScalar(-0.504791)),
-		btVector3(btScalar(-0.713765), btScalar(0.605554), btScalar(-0.351373)),
-		btVector3(btScalar(-0.686251), btScalar(0.687600), btScalar(0.236927)),
-		btVector3(btScalar(-0.680201), btScalar(0.429407), btScalar(0.593732)),
-		btVector3(btScalar(-0.733474), btScalar(0.546450), btScalar(0.403814)),
-		btVector3(btScalar(-0.591023), btScalar(0.292923), btScalar(0.751445)),
-		btVector3(btScalar(-0.500283), btScalar(-0.080757), btScalar(0.861922)),
-		btVector3(btScalar(-0.643710), btScalar(0.070115), btScalar(0.761985)),
-		btVector3(btScalar(-0.506332), btScalar(-0.308425), btScalar(0.805122)),
-		btVector3(btScalar(-0.503015), btScalar(-0.509847), btScalar(0.697573)),
-		btVector3(btScalar(-0.482525), btScalar(-0.682105), btScalar(0.549229)),
-		btVector3(btScalar(-0.680396), btScalar(-0.716323), btScalar(-0.153451)),
-		btVector3(btScalar(-0.658346), btScalar(-0.746264), btScalar(0.097562)),
-		btVector3(btScalar(-0.653272), btScalar(-0.646915), btScalar(-0.392948)),
-		btVector3(btScalar(-0.590828), btScalar(-0.732655), btScalar(0.337645)),
-		btVector3(btScalar(-0.819140), btScalar(-0.518013), btScalar(-0.246166)),
-		btVector3(btScalar(-0.900513), btScalar(-0.282178), btScalar(-0.330487)),
-		btVector3(btScalar(-0.914953), btScalar(-0.028652), btScalar(-0.402122)),
-		btVector3(btScalar(-0.859924), btScalar(0.220209), btScalar(-0.459898)),
-		btVector3(btScalar(-0.777185), btScalar(0.613720), btScalar(-0.137836)),
-		btVector3(btScalar(-0.805285), btScalar(0.586889), btScalar(0.082728)),
-		btVector3(btScalar(-0.872413), btScalar(0.406077), btScalar(-0.271735)),
-		btVector3(btScalar(-0.859339), btScalar(0.448072), btScalar(0.246101)),
-		btVector3(btScalar(-0.757671), btScalar(0.216320), btScalar(0.615594)),
-		btVector3(btScalar(-0.826165), btScalar(0.348139), btScalar(0.442851)),
-		btVector3(btScalar(-0.671810), btScalar(-0.162803), btScalar(0.722557)),
-		btVector3(btScalar(-0.796504), btScalar(-0.004543), btScalar(0.604468)),
-		btVector3(btScalar(-0.676298), btScalar(-0.378223), btScalar(0.631794)),
-		btVector3(btScalar(-0.668883), btScalar(-0.558258), btScalar(0.490673)),
-		btVector3(btScalar(-0.821287), btScalar(-0.570118), btScalar(0.006994)),
-		btVector3(btScalar(-0.767428), btScalar(-0.587810), btScalar(0.255470)),
-		btVector3(btScalar(-0.933296), btScalar(-0.349837), btScalar(-0.079865)),
-		btVector3(btScalar(-0.982667), btScalar(-0.100393), btScalar(-0.155208)),
-		btVector3(btScalar(-0.961396), btScalar(0.160910), btScalar(-0.222938)),
-		btVector3(btScalar(-0.934858), btScalar(0.354555), btScalar(-0.006864)),
-		btVector3(btScalar(-0.941687), btScalar(0.229736), btScalar(0.245711)),
-		btVector3(btScalar(-0.884317), btScalar(0.131552), btScalar(0.447536)),
-		btVector3(btScalar(-0.810359), btScalar(-0.219769), btScalar(0.542788)),
-		btVector3(btScalar(-0.915929), btScalar(-0.210048), btScalar(0.341743)),
-		btVector3(btScalar(-0.816799), btScalar(-0.407192), btScalar(0.408303)),
-		btVector3(btScalar(-0.903050), btScalar(-0.392416), btScalar(0.174076)),
-		btVector3(btScalar(-0.980325), btScalar(-0.170969), btScalar(0.096586)),
-		btVector3(btScalar(-0.995936), btScalar(0.084891), btScalar(0.029441)),
-		btVector3(btScalar(-0.960031), btScalar(0.002650), btScalar(0.279283)),
-	};
+		{
+			btVector3(btScalar(0.997604), btScalar(0.067004), btScalar(0.017144)),
+			btVector3(btScalar(0.984139), btScalar(-0.086784), btScalar(-0.154427)),
+			btVector3(btScalar(0.971065), btScalar(0.124164), btScalar(-0.203224)),
+			btVector3(btScalar(0.955844), btScalar(0.291173), btScalar(-0.037704)),
+			btVector3(btScalar(0.957405), btScalar(0.212238), btScalar(0.195157)),
+			btVector3(btScalar(0.971650), btScalar(-0.012709), btScalar(0.235561)),
+			btVector3(btScalar(0.984920), btScalar(-0.161831), btScalar(0.059695)),
+			btVector3(btScalar(0.946673), btScalar(-0.299288), btScalar(-0.117536)),
+			btVector3(btScalar(0.922670), btScalar(-0.219186), btScalar(-0.317019)),
+			btVector3(btScalar(0.928134), btScalar(-0.007265), btScalar(-0.371867)),
+			btVector3(btScalar(0.875642), btScalar(0.198434), btScalar(-0.439988)),
+			btVector3(btScalar(0.908035), btScalar(0.325975), btScalar(-0.262562)),
+			btVector3(btScalar(0.864519), btScalar(0.488706), btScalar(-0.116755)),
+			btVector3(btScalar(0.893009), btScalar(0.428046), btScalar(0.137185)),
+			btVector3(btScalar(0.857494), btScalar(0.362137), btScalar(0.364776)),
+			btVector3(btScalar(0.900815), btScalar(0.132524), btScalar(0.412987)),
+			btVector3(btScalar(0.934964), btScalar(-0.241739), btScalar(0.259179)),
+			btVector3(btScalar(0.894570), btScalar(-0.103504), btScalar(0.434263)),
+			btVector3(btScalar(0.922085), btScalar(-0.376668), btScalar(0.086241)),
+			btVector3(btScalar(0.862177), btScalar(-0.499154), btScalar(-0.085330)),
+			btVector3(btScalar(0.861982), btScalar(-0.420218), btScalar(-0.282861)),
+			btVector3(btScalar(0.818076), btScalar(-0.328256), btScalar(-0.471804)),
+			btVector3(btScalar(0.762657), btScalar(-0.179329), btScalar(-0.621124)),
+			btVector3(btScalar(0.826857), btScalar(0.019760), btScalar(-0.561786)),
+			btVector3(btScalar(0.731434), btScalar(0.206599), btScalar(-0.649817)),
+			btVector3(btScalar(0.769486), btScalar(0.379052), btScalar(-0.513770)),
+			btVector3(btScalar(0.796806), btScalar(0.507176), btScalar(-0.328145)),
+			btVector3(btScalar(0.679722), btScalar(0.684101), btScalar(-0.264123)),
+			btVector3(btScalar(0.786854), btScalar(0.614886), btScalar(0.050912)),
+			btVector3(btScalar(0.769486), btScalar(0.571141), btScalar(0.285139)),
+			btVector3(btScalar(0.707432), btScalar(0.492789), btScalar(0.506288)),
+			btVector3(btScalar(0.774560), btScalar(0.268037), btScalar(0.572652)),
+			btVector3(btScalar(0.796220), btScalar(0.031230), btScalar(0.604077)),
+			btVector3(btScalar(0.837395), btScalar(-0.320285), btScalar(0.442461)),
+			btVector3(btScalar(0.848127), btScalar(-0.450548), btScalar(0.278307)),
+			btVector3(btScalar(0.775536), btScalar(-0.206354), btScalar(0.596465)),
+			btVector3(btScalar(0.816320), btScalar(-0.567007), btScalar(0.109469)),
+			btVector3(btScalar(0.741191), btScalar(-0.668690), btScalar(-0.056832)),
+			btVector3(btScalar(0.755632), btScalar(-0.602975), btScalar(-0.254949)),
+			btVector3(btScalar(0.720311), btScalar(-0.521318), btScalar(-0.457165)),
+			btVector3(btScalar(0.670746), btScalar(-0.386583), btScalar(-0.632835)),
+			btVector3(btScalar(0.587031), btScalar(-0.219769), btScalar(-0.778836)),
+			btVector3(btScalar(0.676015), btScalar(-0.003182), btScalar(-0.736676)),
+			btVector3(btScalar(0.566932), btScalar(0.186963), btScalar(-0.802064)),
+			btVector3(btScalar(0.618254), btScalar(0.398105), btScalar(-0.677533)),
+			btVector3(btScalar(0.653964), btScalar(0.575224), btScalar(-0.490933)),
+			btVector3(btScalar(0.525367), btScalar(0.743205), btScalar(-0.414028)),
+			btVector3(btScalar(0.506439), btScalar(0.836528), btScalar(-0.208885)),
+			btVector3(btScalar(0.651427), btScalar(0.756426), btScalar(-0.056247)),
+			btVector3(btScalar(0.641670), btScalar(0.745149), btScalar(0.180908)),
+			btVector3(btScalar(0.602643), btScalar(0.687211), btScalar(0.405180)),
+			btVector3(btScalar(0.516586), btScalar(0.596999), btScalar(0.613447)),
+			btVector3(btScalar(0.602252), btScalar(0.387801), btScalar(0.697573)),
+			btVector3(btScalar(0.646549), btScalar(0.153911), btScalar(0.746956)),
+			btVector3(btScalar(0.650842), btScalar(-0.087756), btScalar(0.753983)),
+			btVector3(btScalar(0.740411), btScalar(-0.497404), btScalar(0.451830)),
+			btVector3(btScalar(0.726946), btScalar(-0.619890), btScalar(0.295093)),
+			btVector3(btScalar(0.637768), btScalar(-0.313092), btScalar(0.703624)),
+			btVector3(btScalar(0.678942), btScalar(-0.722934), btScalar(0.126645)),
+			btVector3(btScalar(0.489072), btScalar(-0.867195), btScalar(-0.092942)),
+			btVector3(btScalar(0.622742), btScalar(-0.757541), btScalar(-0.194636)),
+			btVector3(btScalar(0.596788), btScalar(-0.693576), btScalar(-0.403098)),
+			btVector3(btScalar(0.550150), btScalar(-0.582172), btScalar(-0.598287)),
+			btVector3(btScalar(0.474436), btScalar(-0.429745), btScalar(-0.768101)),
+			btVector3(btScalar(0.372574), btScalar(-0.246016), btScalar(-0.894583)),
+			btVector3(btScalar(0.480095), btScalar(-0.026513), btScalar(-0.876626)),
+			btVector3(btScalar(0.352474), btScalar(0.177242), btScalar(-0.918787)),
+			btVector3(btScalar(0.441848), btScalar(0.374386), btScalar(-0.814946)),
+			btVector3(btScalar(0.492389), btScalar(0.582223), btScalar(-0.646693)),
+			btVector3(btScalar(0.343498), btScalar(0.866080), btScalar(-0.362693)),
+			btVector3(btScalar(0.362036), btScalar(0.745149), btScalar(-0.559639)),
+			btVector3(btScalar(0.334131), btScalar(0.937044), btScalar(-0.099774)),
+			btVector3(btScalar(0.486925), btScalar(0.871718), btScalar(0.052473)),
+			btVector3(btScalar(0.452776), btScalar(0.845665), btScalar(0.281820)),
+			btVector3(btScalar(0.399503), btScalar(0.771785), btScalar(0.494576)),
+			btVector3(btScalar(0.296469), btScalar(0.673018), btScalar(0.677469)),
+			btVector3(btScalar(0.392088), btScalar(0.479179), btScalar(0.785213)),
+			btVector3(btScalar(0.452190), btScalar(0.252094), btScalar(0.855286)),
+			btVector3(btScalar(0.478339), btScalar(0.013149), btScalar(0.877928)),
+			btVector3(btScalar(0.481656), btScalar(-0.219380), btScalar(0.848259)),
+			btVector3(btScalar(0.615327), btScalar(-0.494293), btScalar(0.613837)),
+			btVector3(btScalar(0.594642), btScalar(-0.650414), btScalar(0.472325)),
+			btVector3(btScalar(0.562249), btScalar(-0.771345), btScalar(0.297631)),
+			btVector3(btScalar(0.467411), btScalar(-0.437133), btScalar(0.768231)),
+			btVector3(btScalar(0.519513), btScalar(-0.847947), btScalar(0.103808)),
+			btVector3(btScalar(0.297640), btScalar(-0.938159), btScalar(-0.176288)),
+			btVector3(btScalar(0.446727), btScalar(-0.838615), btScalar(-0.311359)),
+			btVector3(btScalar(0.331790), btScalar(-0.942437), btScalar(0.040762)),
+			btVector3(btScalar(0.413358), btScalar(-0.748403), btScalar(-0.518259)),
+			btVector3(btScalar(0.347596), btScalar(-0.621640), btScalar(-0.701737)),
+			btVector3(btScalar(0.249831), btScalar(-0.456186), btScalar(-0.853984)),
+			btVector3(btScalar(0.131772), btScalar(-0.262931), btScalar(-0.955678)),
+			btVector3(btScalar(0.247099), btScalar(-0.042261), btScalar(-0.967975)),
+			btVector3(btScalar(0.113624), btScalar(0.165965), btScalar(-0.979491)),
+			btVector3(btScalar(0.217438), btScalar(0.374580), btScalar(-0.901220)),
+			btVector3(btScalar(0.307983), btScalar(0.554615), btScalar(-0.772786)),
+			btVector3(btScalar(0.166702), btScalar(0.953181), btScalar(-0.252021)),
+			btVector3(btScalar(0.172751), btScalar(0.844499), btScalar(-0.506743)),
+			btVector3(btScalar(0.177630), btScalar(0.711125), btScalar(-0.679876)),
+			btVector3(btScalar(0.120064), btScalar(0.992260), btScalar(-0.030482)),
+			btVector3(btScalar(0.289640), btScalar(0.949098), btScalar(0.122546)),
+			btVector3(btScalar(0.239879), btScalar(0.909047), btScalar(0.340377)),
+			btVector3(btScalar(0.181142), btScalar(0.821363), btScalar(0.540641)),
+			btVector3(btScalar(0.066986), btScalar(0.719097), btScalar(0.691327)),
+			btVector3(btScalar(0.156750), btScalar(0.545478), btScalar(0.823079)),
+			btVector3(btScalar(0.236172), btScalar(0.342306), btScalar(0.909353)),
+			btVector3(btScalar(0.277541), btScalar(0.112693), btScalar(0.953856)),
+			btVector3(btScalar(0.295299), btScalar(-0.121974), btScalar(0.947415)),
+			btVector3(btScalar(0.287883), btScalar(-0.349254), btScalar(0.891591)),
+			btVector3(btScalar(0.437165), btScalar(-0.634666), btScalar(0.636869)),
+			btVector3(btScalar(0.407113), btScalar(-0.784954), btScalar(0.466664)),
+			btVector3(btScalar(0.375111), btScalar(-0.888193), btScalar(0.264839)),
+			btVector3(btScalar(0.275394), btScalar(-0.560591), btScalar(0.780723)),
+			btVector3(btScalar(0.122015), btScalar(-0.992209), btScalar(-0.024821)),
+			btVector3(btScalar(0.087866), btScalar(-0.966156), btScalar(-0.241676)),
+			btVector3(btScalar(0.239489), btScalar(-0.885665), btScalar(-0.397437)),
+			btVector3(btScalar(0.167287), btScalar(-0.965184), btScalar(0.200817)),
+			btVector3(btScalar(0.201632), btScalar(-0.776789), btScalar(-0.596335)),
+			btVector3(btScalar(0.122015), btScalar(-0.637971), btScalar(-0.760098)),
+			btVector3(btScalar(0.008054), btScalar(-0.464741), btScalar(-0.885214)),
+			btVector3(btScalar(-0.116054), btScalar(-0.271096), btScalar(-0.955482)),
+			btVector3(btScalar(-0.000727), btScalar(-0.056065), btScalar(-0.998424)),
+			btVector3(btScalar(-0.134007), btScalar(0.152939), btScalar(-0.978905)),
+			btVector3(btScalar(-0.025900), btScalar(0.366026), btScalar(-0.930108)),
+			btVector3(btScalar(0.081231), btScalar(0.557337), btScalar(-0.826072)),
+			btVector3(btScalar(-0.002874), btScalar(0.917213), btScalar(-0.398023)),
+			btVector3(btScalar(-0.050683), btScalar(0.981761), btScalar(-0.182534)),
+			btVector3(btScalar(-0.040536), btScalar(0.710153), btScalar(-0.702713)),
+			btVector3(btScalar(-0.139081), btScalar(0.827973), btScalar(-0.543048)),
+			btVector3(btScalar(-0.101029), btScalar(0.994010), btScalar(0.041152)),
+			btVector3(btScalar(0.069328), btScalar(0.978067), btScalar(0.196133)),
+			btVector3(btScalar(0.023860), btScalar(0.911380), btScalar(0.410645)),
+			btVector3(btScalar(-0.153521), btScalar(0.736789), btScalar(0.658145)),
+			btVector3(btScalar(-0.070002), btScalar(0.591750), btScalar(0.802780)),
+			btVector3(btScalar(0.002590), btScalar(0.312948), btScalar(0.949562)),
+			btVector3(btScalar(0.090988), btScalar(-0.020680), btScalar(0.995627)),
+			btVector3(btScalar(0.088842), btScalar(-0.250099), btScalar(0.964006)),
+			btVector3(btScalar(0.083378), btScalar(-0.470185), btScalar(0.878318)),
+			btVector3(btScalar(0.240074), btScalar(-0.749764), btScalar(0.616374)),
+			btVector3(btScalar(0.210803), btScalar(-0.885860), btScalar(0.412987)),
+			btVector3(btScalar(0.077524), btScalar(-0.660524), btScalar(0.746565)),
+			btVector3(btScalar(-0.096736), btScalar(-0.990070), btScalar(-0.100945)),
+			btVector3(btScalar(-0.052634), btScalar(-0.990264), btScalar(0.127426)),
+			btVector3(btScalar(-0.106102), btScalar(-0.938354), btScalar(-0.328340)),
+			btVector3(btScalar(0.013323), btScalar(-0.863112), btScalar(-0.504596)),
+			btVector3(btScalar(-0.002093), btScalar(-0.936993), btScalar(0.349161)),
+			btVector3(btScalar(-0.106297), btScalar(-0.636610), btScalar(-0.763612)),
+			btVector3(btScalar(-0.229430), btScalar(-0.463769), btScalar(-0.855546)),
+			btVector3(btScalar(-0.245236), btScalar(-0.066175), btScalar(-0.966999)),
+			btVector3(btScalar(-0.351587), btScalar(-0.270513), btScalar(-0.896145)),
+			btVector3(btScalar(-0.370906), btScalar(0.133108), btScalar(-0.918982)),
+			btVector3(btScalar(-0.264360), btScalar(0.346000), btScalar(-0.900049)),
+			btVector3(btScalar(-0.151375), btScalar(0.543728), btScalar(-0.825291)),
+			btVector3(btScalar(-0.218697), btScalar(0.912741), btScalar(-0.344346)),
+			btVector3(btScalar(-0.274507), btScalar(0.953764), btScalar(-0.121635)),
+			btVector3(btScalar(-0.259677), btScalar(0.692266), btScalar(-0.673044)),
+			btVector3(btScalar(-0.350416), btScalar(0.798810), btScalar(-0.488786)),
+			btVector3(btScalar(-0.320170), btScalar(0.941127), btScalar(0.108297)),
+			btVector3(btScalar(-0.147667), btScalar(0.952792), btScalar(0.265034)),
+			btVector3(btScalar(-0.188061), btScalar(0.860636), btScalar(0.472910)),
+			btVector3(btScalar(-0.370906), btScalar(0.739900), btScalar(0.560941)),
+			btVector3(btScalar(-0.297143), btScalar(0.585334), btScalar(0.754178)),
+			btVector3(btScalar(-0.189622), btScalar(0.428241), btScalar(0.883393)),
+			btVector3(btScalar(-0.091272), btScalar(0.098695), btScalar(0.990747)),
+			btVector3(btScalar(-0.256945), btScalar(0.228375), btScalar(0.938827)),
+			btVector3(btScalar(-0.111761), btScalar(-0.133251), btScalar(0.984696)),
+			btVector3(btScalar(-0.118006), btScalar(-0.356253), btScalar(0.926725)),
+			btVector3(btScalar(-0.119372), btScalar(-0.563896), btScalar(0.817029)),
+			btVector3(btScalar(0.041228), btScalar(-0.833949), btScalar(0.550010)),
+			btVector3(btScalar(-0.121909), btScalar(-0.736543), btScalar(0.665172)),
+			btVector3(btScalar(-0.307681), btScalar(-0.931160), btScalar(-0.195026)),
+			btVector3(btScalar(-0.283679), btScalar(-0.957990), btScalar(0.041348)),
+			btVector3(btScalar(-0.227284), btScalar(-0.935243), btScalar(0.270890)),
+			btVector3(btScalar(-0.293436), btScalar(-0.858252), btScalar(-0.420860)),
+			btVector3(btScalar(-0.175767), btScalar(-0.780677), btScalar(-0.599262)),
+			btVector3(btScalar(-0.170108), btScalar(-0.858835), btScalar(0.482865)),
+			btVector3(btScalar(-0.332854), btScalar(-0.635055), btScalar(-0.696857)),
+			btVector3(btScalar(-0.447791), btScalar(-0.445299), btScalar(-0.775128)),
+			btVector3(btScalar(-0.470622), btScalar(-0.074146), btScalar(-0.879164)),
+			btVector3(btScalar(-0.639417), btScalar(-0.340505), btScalar(-0.689049)),
+			btVector3(btScalar(-0.598438), btScalar(0.104722), btScalar(-0.794256)),
+			btVector3(btScalar(-0.488575), btScalar(0.307699), btScalar(-0.816313)),
+			btVector3(btScalar(-0.379882), btScalar(0.513592), btScalar(-0.769077)),
+			btVector3(btScalar(-0.425740), btScalar(0.862775), btScalar(-0.272516)),
+			btVector3(btScalar(-0.480769), btScalar(0.875412), btScalar(-0.048439)),
+			btVector3(btScalar(-0.467890), btScalar(0.648716), btScalar(-0.600043)),
+			btVector3(btScalar(-0.543799), btScalar(0.730956), btScalar(-0.411881)),
+			btVector3(btScalar(-0.516284), btScalar(0.838277), btScalar(0.174076)),
+			btVector3(btScalar(-0.353343), btScalar(0.876384), btScalar(0.326519)),
+			btVector3(btScalar(-0.572875), btScalar(0.614497), btScalar(0.542007)),
+			btVector3(btScalar(-0.503600), btScalar(0.497261), btScalar(0.706161)),
+			btVector3(btScalar(-0.530920), btScalar(0.754870), btScalar(0.384685)),
+			btVector3(btScalar(-0.395884), btScalar(0.366414), btScalar(0.841818)),
+			btVector3(btScalar(-0.300656), btScalar(0.001678), btScalar(0.953661)),
+			btVector3(btScalar(-0.461060), btScalar(0.146912), btScalar(0.875000)),
+			btVector3(btScalar(-0.315486), btScalar(-0.232212), btScalar(0.919893)),
+			btVector3(btScalar(-0.323682), btScalar(-0.449187), btScalar(0.832644)),
+			btVector3(btScalar(-0.318999), btScalar(-0.639527), btScalar(0.699134)),
+			btVector3(btScalar(-0.496771), btScalar(-0.866029), btScalar(-0.055271)),
+			btVector3(btScalar(-0.496771), btScalar(-0.816257), btScalar(-0.294377)),
+			btVector3(btScalar(-0.456377), btScalar(-0.869528), btScalar(0.188130)),
+			btVector3(btScalar(-0.380858), btScalar(-0.827144), btScalar(0.412792)),
+			btVector3(btScalar(-0.449352), btScalar(-0.727405), btScalar(-0.518259)),
+			btVector3(btScalar(-0.570533), btScalar(-0.551064), btScalar(-0.608632)),
+			btVector3(btScalar(-0.656394), btScalar(-0.118280), btScalar(-0.744874)),
+			btVector3(btScalar(-0.756696), btScalar(-0.438105), btScalar(-0.484882)),
+			btVector3(btScalar(-0.801773), btScalar(-0.204798), btScalar(-0.561005)),
+			btVector3(btScalar(-0.785186), btScalar(0.038618), btScalar(-0.617805)),
+			btVector3(btScalar(-0.709082), btScalar(0.262399), btScalar(-0.654306)),
+			btVector3(btScalar(-0.583412), btScalar(0.462265), btScalar(-0.667383)),
+			btVector3(btScalar(-0.616001), btScalar(0.761286), btScalar(-0.201272)),
+			btVector3(btScalar(-0.660687), btScalar(0.750204), btScalar(0.020072)),
+			btVector3(btScalar(-0.744987), btScalar(0.435823), btScalar(-0.504791)),
+			btVector3(btScalar(-0.713765), btScalar(0.605554), btScalar(-0.351373)),
+			btVector3(btScalar(-0.686251), btScalar(0.687600), btScalar(0.236927)),
+			btVector3(btScalar(-0.680201), btScalar(0.429407), btScalar(0.593732)),
+			btVector3(btScalar(-0.733474), btScalar(0.546450), btScalar(0.403814)),
+			btVector3(btScalar(-0.591023), btScalar(0.292923), btScalar(0.751445)),
+			btVector3(btScalar(-0.500283), btScalar(-0.080757), btScalar(0.861922)),
+			btVector3(btScalar(-0.643710), btScalar(0.070115), btScalar(0.761985)),
+			btVector3(btScalar(-0.506332), btScalar(-0.308425), btScalar(0.805122)),
+			btVector3(btScalar(-0.503015), btScalar(-0.509847), btScalar(0.697573)),
+			btVector3(btScalar(-0.482525), btScalar(-0.682105), btScalar(0.549229)),
+			btVector3(btScalar(-0.680396), btScalar(-0.716323), btScalar(-0.153451)),
+			btVector3(btScalar(-0.658346), btScalar(-0.746264), btScalar(0.097562)),
+			btVector3(btScalar(-0.653272), btScalar(-0.646915), btScalar(-0.392948)),
+			btVector3(btScalar(-0.590828), btScalar(-0.732655), btScalar(0.337645)),
+			btVector3(btScalar(-0.819140), btScalar(-0.518013), btScalar(-0.246166)),
+			btVector3(btScalar(-0.900513), btScalar(-0.282178), btScalar(-0.330487)),
+			btVector3(btScalar(-0.914953), btScalar(-0.028652), btScalar(-0.402122)),
+			btVector3(btScalar(-0.859924), btScalar(0.220209), btScalar(-0.459898)),
+			btVector3(btScalar(-0.777185), btScalar(0.613720), btScalar(-0.137836)),
+			btVector3(btScalar(-0.805285), btScalar(0.586889), btScalar(0.082728)),
+			btVector3(btScalar(-0.872413), btScalar(0.406077), btScalar(-0.271735)),
+			btVector3(btScalar(-0.859339), btScalar(0.448072), btScalar(0.246101)),
+			btVector3(btScalar(-0.757671), btScalar(0.216320), btScalar(0.615594)),
+			btVector3(btScalar(-0.826165), btScalar(0.348139), btScalar(0.442851)),
+			btVector3(btScalar(-0.671810), btScalar(-0.162803), btScalar(0.722557)),
+			btVector3(btScalar(-0.796504), btScalar(-0.004543), btScalar(0.604468)),
+			btVector3(btScalar(-0.676298), btScalar(-0.378223), btScalar(0.631794)),
+			btVector3(btScalar(-0.668883), btScalar(-0.558258), btScalar(0.490673)),
+			btVector3(btScalar(-0.821287), btScalar(-0.570118), btScalar(0.006994)),
+			btVector3(btScalar(-0.767428), btScalar(-0.587810), btScalar(0.255470)),
+			btVector3(btScalar(-0.933296), btScalar(-0.349837), btScalar(-0.079865)),
+			btVector3(btScalar(-0.982667), btScalar(-0.100393), btScalar(-0.155208)),
+			btVector3(btScalar(-0.961396), btScalar(0.160910), btScalar(-0.222938)),
+			btVector3(btScalar(-0.934858), btScalar(0.354555), btScalar(-0.006864)),
+			btVector3(btScalar(-0.941687), btScalar(0.229736), btScalar(0.245711)),
+			btVector3(btScalar(-0.884317), btScalar(0.131552), btScalar(0.447536)),
+			btVector3(btScalar(-0.810359), btScalar(-0.219769), btScalar(0.542788)),
+			btVector3(btScalar(-0.915929), btScalar(-0.210048), btScalar(0.341743)),
+			btVector3(btScalar(-0.816799), btScalar(-0.407192), btScalar(0.408303)),
+			btVector3(btScalar(-0.903050), btScalar(-0.392416), btScalar(0.174076)),
+			btVector3(btScalar(-0.980325), btScalar(-0.170969), btScalar(0.096586)),
+			btVector3(btScalar(-0.995936), btScalar(0.084891), btScalar(0.029441)),
+			btVector3(btScalar(-0.960031), btScalar(0.002650), btScalar(0.279283)),
+		};
 	static btVector3 sUnitSpherePoints[NUM_UNITSPHERE_POINTS + MAX_PREFERRED_PENETRATION_DIRECTIONS * 2] =
-	{
-		btVector3(btScalar(0.000000) , btScalar(-0.000000),btScalar(-1.000000)),
-		btVector3(btScalar(0.723608) , btScalar(-0.525725),btScalar(-0.447219)),
-		btVector3(btScalar(-0.276388) , btScalar(-0.850649),btScalar(-0.447219)),
-		btVector3(btScalar(-0.894426) , btScalar(-0.000000),btScalar(-0.447216)),
-		btVector3(btScalar(-0.276388) , btScalar(0.850649),btScalar(-0.447220)),
-		btVector3(btScalar(0.723608) , btScalar(0.525725),btScalar(-0.447219)),
-		btVector3(btScalar(0.276388) , btScalar(-0.850649),btScalar(0.447220)),
-		btVector3(btScalar(-0.723608) , btScalar(-0.525725),btScalar(0.447219)),
-		btVector3(btScalar(-0.723608) , btScalar(0.525725),btScalar(0.447219)),
-		btVector3(btScalar(0.276388) , btScalar(0.850649),btScalar(0.447219)),
-		btVector3(btScalar(0.894426) , btScalar(0.000000),btScalar(0.447216)),
-		btVector3(btScalar(-0.000000) , btScalar(0.000000),btScalar(1.000000)),
-		btVector3(btScalar(0.425323) , btScalar(-0.309011),btScalar(-0.850654)),
-		btVector3(btScalar(-0.162456) , btScalar(-0.499995),btScalar(-0.850654)),
-		btVector3(btScalar(0.262869) , btScalar(-0.809012),btScalar(-0.525738)),
-		btVector3(btScalar(0.425323) , btScalar(0.309011),btScalar(-0.850654)),
-		btVector3(btScalar(0.850648) , btScalar(-0.000000),btScalar(-0.525736)),
-		btVector3(btScalar(-0.525730) , btScalar(-0.000000),btScalar(-0.850652)),
-		btVector3(btScalar(-0.688190) , btScalar(-0.499997),btScalar(-0.525736)),
-		btVector3(btScalar(-0.162456) , btScalar(0.499995),btScalar(-0.850654)),
-		btVector3(btScalar(-0.688190) , btScalar(0.499997),btScalar(-0.525736)),
-		btVector3(btScalar(0.262869) , btScalar(0.809012),btScalar(-0.525738)),
-		btVector3(btScalar(0.951058) , btScalar(0.309013),btScalar(0.000000)),
-		btVector3(btScalar(0.951058) , btScalar(-0.309013),btScalar(0.000000)),
-		btVector3(btScalar(0.587786) , btScalar(-0.809017),btScalar(0.000000)),
-		btVector3(btScalar(0.000000) , btScalar(-1.000000),btScalar(0.000000)),
-		btVector3(btScalar(-0.587786) , btScalar(-0.809017),btScalar(0.000000)),
-		btVector3(btScalar(-0.951058) , btScalar(-0.309013),btScalar(-0.000000)),
-		btVector3(btScalar(-0.951058) , btScalar(0.309013),btScalar(-0.000000)),
-		btVector3(btScalar(-0.587786) , btScalar(0.809017),btScalar(-0.000000)),
-		btVector3(btScalar(-0.000000) , btScalar(1.000000),btScalar(-0.000000)),
-		btVector3(btScalar(0.587786) , btScalar(0.809017),btScalar(-0.000000)),
-		btVector3(btScalar(0.688190) , btScalar(-0.499997),btScalar(0.525736)),
-		btVector3(btScalar(-0.262869) , btScalar(-0.809012),btScalar(0.525738)),
-		btVector3(btScalar(-0.850648) , btScalar(0.000000),btScalar(0.525736)),
-		btVector3(btScalar(-0.262869) , btScalar(0.809012),btScalar(0.525738)),
-		btVector3(btScalar(0.688190) , btScalar(0.499997),btScalar(0.525736)),
-		btVector3(btScalar(0.525730) , btScalar(0.000000),btScalar(0.850652)),
-		btVector3(btScalar(0.162456) , btScalar(-0.499995),btScalar(0.850654)),
-		btVector3(btScalar(-0.425323) , btScalar(-0.309011),btScalar(0.850654)),
-		btVector3(btScalar(-0.425323) , btScalar(0.309011),btScalar(0.850654)),
-		btVector3(btScalar(0.162456) , btScalar(0.499995),btScalar(0.850654))
-	};
+		{
+			btVector3(btScalar(0.000000), btScalar(-0.000000), btScalar(-1.000000)),
+			btVector3(btScalar(0.723608), btScalar(-0.525725), btScalar(-0.447219)),
+			btVector3(btScalar(-0.276388), btScalar(-0.850649), btScalar(-0.447219)),
+			btVector3(btScalar(-0.894426), btScalar(-0.000000), btScalar(-0.447216)),
+			btVector3(btScalar(-0.276388), btScalar(0.850649), btScalar(-0.447220)),
+			btVector3(btScalar(0.723608), btScalar(0.525725), btScalar(-0.447219)),
+			btVector3(btScalar(0.276388), btScalar(-0.850649), btScalar(0.447220)),
+			btVector3(btScalar(-0.723608), btScalar(-0.525725), btScalar(0.447219)),
+			btVector3(btScalar(-0.723608), btScalar(0.525725), btScalar(0.447219)),
+			btVector3(btScalar(0.276388), btScalar(0.850649), btScalar(0.447219)),
+			btVector3(btScalar(0.894426), btScalar(0.000000), btScalar(0.447216)),
+			btVector3(btScalar(-0.000000), btScalar(0.000000), btScalar(1.000000)),
+			btVector3(btScalar(0.425323), btScalar(-0.309011), btScalar(-0.850654)),
+			btVector3(btScalar(-0.162456), btScalar(-0.499995), btScalar(-0.850654)),
+			btVector3(btScalar(0.262869), btScalar(-0.809012), btScalar(-0.525738)),
+			btVector3(btScalar(0.425323), btScalar(0.309011), btScalar(-0.850654)),
+			btVector3(btScalar(0.850648), btScalar(-0.000000), btScalar(-0.525736)),
+			btVector3(btScalar(-0.525730), btScalar(-0.000000), btScalar(-0.850652)),
+			btVector3(btScalar(-0.688190), btScalar(-0.499997), btScalar(-0.525736)),
+			btVector3(btScalar(-0.162456), btScalar(0.499995), btScalar(-0.850654)),
+			btVector3(btScalar(-0.688190), btScalar(0.499997), btScalar(-0.525736)),
+			btVector3(btScalar(0.262869), btScalar(0.809012), btScalar(-0.525738)),
+			btVector3(btScalar(0.951058), btScalar(0.309013), btScalar(0.000000)),
+			btVector3(btScalar(0.951058), btScalar(-0.309013), btScalar(0.000000)),
+			btVector3(btScalar(0.587786), btScalar(-0.809017), btScalar(0.000000)),
+			btVector3(btScalar(0.000000), btScalar(-1.000000), btScalar(0.000000)),
+			btVector3(btScalar(-0.587786), btScalar(-0.809017), btScalar(0.000000)),
+			btVector3(btScalar(-0.951058), btScalar(-0.309013), btScalar(-0.000000)),
+			btVector3(btScalar(-0.951058), btScalar(0.309013), btScalar(-0.000000)),
+			btVector3(btScalar(-0.587786), btScalar(0.809017), btScalar(-0.000000)),
+			btVector3(btScalar(-0.000000), btScalar(1.000000), btScalar(-0.000000)),
+			btVector3(btScalar(0.587786), btScalar(0.809017), btScalar(-0.000000)),
+			btVector3(btScalar(0.688190), btScalar(-0.499997), btScalar(0.525736)),
+			btVector3(btScalar(-0.262869), btScalar(-0.809012), btScalar(0.525738)),
+			btVector3(btScalar(-0.850648), btScalar(0.000000), btScalar(0.525736)),
+			btVector3(btScalar(-0.262869), btScalar(0.809012), btScalar(0.525738)),
+			btVector3(btScalar(0.688190), btScalar(0.499997), btScalar(0.525736)),
+			btVector3(btScalar(0.525730), btScalar(0.000000), btScalar(0.850652)),
+			btVector3(btScalar(0.162456), btScalar(-0.499995), btScalar(0.850654)),
+			btVector3(btScalar(-0.425323), btScalar(-0.309011), btScalar(0.850654)),
+			btVector3(btScalar(-0.425323), btScalar(0.309011), btScalar(0.850654)),
+			btVector3(btScalar(0.162456), btScalar(0.499995), btScalar(0.850654))};
 	if (highres)
 		return sUnitSpherePointsHighres;
 	return sUnitSpherePoints;
 }
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btShapeHull.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btShapeHull.h
index 78ea4b65011..54439f9ca25 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btShapeHull.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btShapeHull.h
@@ -21,32 +21,31 @@ subject to the following restrictions:
 #include "LinearMath/btAlignedObjectArray.h"
 #include "BulletCollision/CollisionShapes/btConvexShape.h"
 
-
 ///The btShapeHull class takes a btConvexShape, builds a simplified convex hull using btConvexHull and provides triangle indices and vertices.
 ///It can be useful for to simplify a complex convex object and for visualization of a non-polyhedral convex object.
 ///It approximates the convex hull using the supporting vertex of 42 directions.
-ATTRIBUTE_ALIGNED16(class) btShapeHull
+ATTRIBUTE_ALIGNED16(class)
+btShapeHull
 {
 protected:
-
 	btAlignedObjectArray<btVector3> m_vertices;
 	btAlignedObjectArray<unsigned int> m_indices;
 	unsigned int m_numIndices;
 	const btConvexShape* m_shape;
 
-	static btVector3* getUnitSpherePoints(int highres=0);
+	static btVector3* getUnitSpherePoints(int highres = 0);
 
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
-	btShapeHull (const btConvexShape* shape);
-	~btShapeHull ();
 
-	bool buildHull (btScalar margin, int highres=0);
+	btShapeHull(const btConvexShape* shape);
+	~btShapeHull();
 
-	int numTriangles () const;
-	int numVertices () const;
-	int numIndices () const;
+	bool buildHull(btScalar margin, int highres = 0);
+
+	int numTriangles() const;
+	int numVertices() const;
+	int numIndices() const;
 
 	const btVector3* getVertexPointer() const
 	{
@@ -58,4 +57,4 @@ public:
 	}
 };
 
-#endif //BT_SHAPE_HULL_H
+#endif  //BT_SHAPE_HULL_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btSphereShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btSphereShape.cpp
index b9a736c0fdd..027db2e1046 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btSphereShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btSphereShape.cpp
@@ -18,54 +18,48 @@ subject to the following restrictions:
 
 #include "LinearMath/btQuaternion.h"
 
-btVector3	btSphereShape::localGetSupportingVertexWithoutMargin(const btVector3& vec)const
+btVector3 btSphereShape::localGetSupportingVertexWithoutMargin(const btVector3& vec) const
 {
 	(void)vec;
-	return btVector3(btScalar(0.),btScalar(0.),btScalar(0.));
+	return btVector3(btScalar(0.), btScalar(0.), btScalar(0.));
 }
 
-void	btSphereShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const
+void btSphereShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const
 {
 	(void)vectors;
 
-	for (int i=0;i<numVectors;i++)
+	for (int i = 0; i < numVectors; i++)
 	{
-		supportVerticesOut[i].setValue(btScalar(0.),btScalar(0.),btScalar(0.));
+		supportVerticesOut[i].setValue(btScalar(0.), btScalar(0.), btScalar(0.));
 	}
 }
 
-
-btVector3	btSphereShape::localGetSupportingVertex(const btVector3& vec)const
+btVector3 btSphereShape::localGetSupportingVertex(const btVector3& vec) const
 {
 	btVector3 supVertex;
 	supVertex = localGetSupportingVertexWithoutMargin(vec);
 
 	btVector3 vecnorm = vec;
-	if (vecnorm .length2() < (SIMD_EPSILON*SIMD_EPSILON))
+	if (vecnorm.length2() < (SIMD_EPSILON * SIMD_EPSILON))
 	{
-		vecnorm.setValue(btScalar(-1.),btScalar(-1.),btScalar(-1.));
-	} 
+		vecnorm.setValue(btScalar(-1.), btScalar(-1.), btScalar(-1.));
+	}
 	vecnorm.normalize();
-	supVertex+= getMargin() * vecnorm;
+	supVertex += getMargin() * vecnorm;
 	return supVertex;
 }
 
-
 //broken due to scaling
-void btSphereShape::getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const
+void btSphereShape::getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
 {
 	const btVector3& center = t.getOrigin();
-	btVector3 extent(getMargin(),getMargin(),getMargin());
+	btVector3 extent(getMargin(), getMargin(), getMargin());
 	aabbMin = center - extent;
 	aabbMax = center + extent;
 }
 
-
-
-void	btSphereShape::calculateLocalInertia(btScalar mass,btVector3& inertia) const
+void btSphereShape::calculateLocalInertia(btScalar mass, btVector3& inertia) const
 {
-	btScalar elem = btScalar(0.4) * mass * getMargin()*getMargin();
-	inertia.setValue(elem,elem,elem);
-
+	btScalar elem = btScalar(0.4) * mass * getMargin() * getMargin();
+	inertia.setValue(elem, elem, elem);
 }
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btSphereShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btSphereShape.h
index 50561f7f54f..75e4fd8e189 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btSphereShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btSphereShape.h
@@ -16,17 +16,17 @@ subject to the following restrictions:
 #define BT_SPHERE_MINKOWSKI_H
 
 #include "btConvexInternalShape.h"
-#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h" // for the types
+#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h"  // for the types
 
 ///The btSphereShape implements an implicit sphere, centered around a local origin with radius.
-ATTRIBUTE_ALIGNED16(class) btSphereShape : public btConvexInternalShape
+ATTRIBUTE_ALIGNED16(class)
+btSphereShape : public btConvexInternalShape
 
 {
-	
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-	btSphereShape (btScalar radius) : btConvexInternalShape ()
+	btSphereShape(btScalar radius) : btConvexInternalShape()
 	{
 		m_shapeType = SPHERE_SHAPE_PROXYTYPE;
 		m_localScaling.setValue(1.0, 1.0, 1.0);
@@ -35,42 +35,37 @@ public:
 		m_collisionMargin = radius;
 		m_padding = 0;
 	}
-	
-	virtual btVector3	localGetSupportingVertex(const btVector3& vec)const;
-	virtual btVector3	localGetSupportingVertexWithoutMargin(const btVector3& vec)const;
+
+	virtual btVector3 localGetSupportingVertex(const btVector3& vec) const;
+	virtual btVector3 localGetSupportingVertexWithoutMargin(const btVector3& vec) const;
 	//notice that the vectors should be unit length
-	virtual void	batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const;
+	virtual void batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const;
 
+	virtual void calculateLocalInertia(btScalar mass, btVector3 & inertia) const;
 
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
+	virtual void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
 
-	virtual void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const;
+	btScalar getRadius() const { return m_implicitShapeDimensions.getX() * m_localScaling.getX(); }
 
-
-	btScalar	getRadius() const { return m_implicitShapeDimensions.getX() * m_localScaling.getX();}
-
-	void	setUnscaledRadius(btScalar	radius)
+	void setUnscaledRadius(btScalar radius)
 	{
 		m_implicitShapeDimensions.setX(radius);
 		btConvexInternalShape::setMargin(radius);
 	}
 
 	//debugging
-	virtual const char*	getName()const {return "SPHERE";}
+	virtual const char* getName() const { return "SPHERE"; }
 
-	virtual void	setMargin(btScalar margin)
+	virtual void setMargin(btScalar margin)
 	{
 		btConvexInternalShape::setMargin(margin);
 	}
-	virtual btScalar	getMargin() const
+	virtual btScalar getMargin() const
 	{
 		//to improve gjk behaviour, use radius+margin as the full margin, so never get into the penetration case
 		//this means, non-uniform scaling is not supported anymore
 		return getRadius();
 	}
-
-
 };
 
-
-#endif //BT_SPHERE_MINKOWSKI_H
+#endif  //BT_SPHERE_MINKOWSKI_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btStaticPlaneShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btStaticPlaneShape.cpp
index b5e0e716d48..9238c919d51 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btStaticPlaneShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btStaticPlaneShape.cpp
@@ -17,24 +17,18 @@ subject to the following restrictions:
 
 #include "LinearMath/btTransformUtil.h"
 
-
-btStaticPlaneShape::btStaticPlaneShape(const btVector3& planeNormal,btScalar planeConstant)
-: btConcaveShape (), m_planeNormal(planeNormal.normalized()),
-m_planeConstant(planeConstant),
-m_localScaling(btScalar(1.),btScalar(1.),btScalar(1.))
+btStaticPlaneShape::btStaticPlaneShape(const btVector3& planeNormal, btScalar planeConstant)
+	: btConcaveShape(), m_planeNormal(planeNormal.normalized()), m_planeConstant(planeConstant), m_localScaling(btScalar(1.), btScalar(1.), btScalar(1.))
 {
 	m_shapeType = STATIC_PLANE_PROXYTYPE;
 	//	btAssert( btFuzzyZero(m_planeNormal.length() - btScalar(1.)) );
 }
 
-
 btStaticPlaneShape::~btStaticPlaneShape()
 {
 }
 
-
-
-void btStaticPlaneShape::getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const
+void btStaticPlaneShape::getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
 {
 	(void)t;
 	/*
@@ -47,55 +41,49 @@ void btStaticPlaneShape::getAabb(const btTransform& t,btVector3& aabbMin,btVecto
 	aabbMax.setMax(center - infvec*m_planeNormal); 
 	*/
 
-	aabbMin.setValue(btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT));
-	aabbMax.setValue(btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT));
-
+	aabbMin.setValue(btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT));
+	aabbMax.setValue(btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT));
 }
 
-
-
-
-void	btStaticPlaneShape::processAllTriangles(btTriangleCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax) const
+void btStaticPlaneShape::processAllTriangles(btTriangleCallback* callback, const btVector3& aabbMin, const btVector3& aabbMax) const
 {
-
 	btVector3 halfExtents = (aabbMax - aabbMin) * btScalar(0.5);
 	btScalar radius = halfExtents.length();
 	btVector3 center = (aabbMax + aabbMin) * btScalar(0.5);
-	
+
 	//this is where the triangles are generated, given AABB and plane equation (normal/constant)
 
-	btVector3 tangentDir0,tangentDir1;
+	btVector3 tangentDir0, tangentDir1;
 
 	//tangentDir0/tangentDir1 can be precalculated
-	btPlaneSpace1(m_planeNormal,tangentDir0,tangentDir1);
+	btPlaneSpace1(m_planeNormal, tangentDir0, tangentDir1);
+
+	btVector3 projectedCenter = center - (m_planeNormal.dot(center) - m_planeConstant) * m_planeNormal;
 
-	btVector3 projectedCenter = center - (m_planeNormal.dot(center) - m_planeConstant)*m_planeNormal;
-	
 	btVector3 triangle[3];
-	triangle[0] = projectedCenter + tangentDir0*radius + tangentDir1*radius;
-	triangle[1] = projectedCenter + tangentDir0*radius - tangentDir1*radius;
-	triangle[2] = projectedCenter - tangentDir0*radius - tangentDir1*radius;
+	triangle[0] = projectedCenter + tangentDir0 * radius + tangentDir1 * radius;
+	triangle[1] = projectedCenter + tangentDir0 * radius - tangentDir1 * radius;
+	triangle[2] = projectedCenter - tangentDir0 * radius - tangentDir1 * radius;
 
-	callback->processTriangle(triangle,0,0);
+	callback->processTriangle(triangle, 0, 0);
 
-	triangle[0] = projectedCenter - tangentDir0*radius - tangentDir1*radius;
-	triangle[1] = projectedCenter - tangentDir0*radius + tangentDir1*radius;
-	triangle[2] = projectedCenter + tangentDir0*radius + tangentDir1*radius;
-
-	callback->processTriangle(triangle,0,1);
+	triangle[0] = projectedCenter - tangentDir0 * radius - tangentDir1 * radius;
+	triangle[1] = projectedCenter - tangentDir0 * radius + tangentDir1 * radius;
+	triangle[2] = projectedCenter + tangentDir0 * radius + tangentDir1 * radius;
 
+	callback->processTriangle(triangle, 0, 1);
 }
 
-void	btStaticPlaneShape::calculateLocalInertia(btScalar mass,btVector3& inertia) const
+void btStaticPlaneShape::calculateLocalInertia(btScalar mass, btVector3& inertia) const
 {
 	(void)mass;
 
 	//moving concave objects not supported
-	
-	inertia.setValue(btScalar(0.),btScalar(0.),btScalar(0.));
+
+	inertia.setValue(btScalar(0.), btScalar(0.), btScalar(0.));
 }
 
-void	btStaticPlaneShape::setLocalScaling(const btVector3& scaling)
+void btStaticPlaneShape::setLocalScaling(const btVector3& scaling)
 {
 	m_localScaling = scaling;
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btStaticPlaneShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btStaticPlaneShape.h
index 5e9eccc77d9..1cda8bbc75b 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btStaticPlaneShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btStaticPlaneShape.h
@@ -18,78 +18,74 @@ subject to the following restrictions:
 
 #include "btConcaveShape.h"
 
-
 ///The btStaticPlaneShape simulates an infinite non-moving (static) collision plane.
-ATTRIBUTE_ALIGNED16(class) btStaticPlaneShape : public btConcaveShape
+ATTRIBUTE_ALIGNED16(class)
+btStaticPlaneShape : public btConcaveShape
 {
 protected:
-	btVector3	m_localAabbMin;
-	btVector3	m_localAabbMax;
-	
-	btVector3	m_planeNormal;
-	btScalar      m_planeConstant;
-	btVector3	m_localScaling;
+	btVector3 m_localAabbMin;
+	btVector3 m_localAabbMax;
+
+	btVector3 m_planeNormal;
+	btScalar m_planeConstant;
+	btVector3 m_localScaling;
 
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-	btStaticPlaneShape(const btVector3& planeNormal,btScalar planeConstant);
+	btStaticPlaneShape(const btVector3& planeNormal, btScalar planeConstant);
 
 	virtual ~btStaticPlaneShape();
 
+	virtual void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
 
-	virtual void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const;
+	virtual void processAllTriangles(btTriangleCallback * callback, const btVector3& aabbMin, const btVector3& aabbMax) const;
 
-	virtual void	processAllTriangles(btTriangleCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax) const;
+	virtual void calculateLocalInertia(btScalar mass, btVector3 & inertia) const;
 
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
-
-	virtual void	setLocalScaling(const btVector3& scaling);
+	virtual void setLocalScaling(const btVector3& scaling);
 	virtual const btVector3& getLocalScaling() const;
-	
-	const btVector3&	getPlaneNormal() const
+
+	const btVector3& getPlaneNormal() const
 	{
-		return	m_planeNormal;
+		return m_planeNormal;
 	}
 
-	const btScalar&	getPlaneConstant() const
+	const btScalar& getPlaneConstant() const
 	{
-		return	m_planeConstant;
+		return m_planeConstant;
 	}
 
 	//debugging
-	virtual const char*	getName()const {return "STATICPLANE";}
+	virtual const char* getName() const { return "STATICPLANE"; }
 
-	virtual	int	calculateSerializeBufferSize() const;
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
-
-
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 };
 
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
-struct	btStaticPlaneShapeData
+struct btStaticPlaneShapeData
 {
-	btCollisionShapeData	m_collisionShapeData;
+	btCollisionShapeData m_collisionShapeData;
 
-	btVector3FloatData	m_localScaling;
-	btVector3FloatData	m_planeNormal;
-	float			m_planeConstant;
-	char	m_pad[4];
+	btVector3FloatData m_localScaling;
+	btVector3FloatData m_planeNormal;
+	float m_planeConstant;
+	char m_pad[4];
 };
 
-
-SIMD_FORCE_INLINE	int	btStaticPlaneShape::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btStaticPlaneShape::calculateSerializeBufferSize() const
 {
 	return sizeof(btStaticPlaneShapeData);
 }
 
 ///fills the dataBuffer and returns the struct name (and 0 on failure)
-SIMD_FORCE_INLINE	const char*	btStaticPlaneShape::serialize(void* dataBuffer, btSerializer* serializer) const
+SIMD_FORCE_INLINE const char* btStaticPlaneShape::serialize(void* dataBuffer, btSerializer* serializer) const
 {
-	btStaticPlaneShapeData* planeData = (btStaticPlaneShapeData*) dataBuffer;
-	btCollisionShape::serialize(&planeData->m_collisionShapeData,serializer);
+	btStaticPlaneShapeData* planeData = (btStaticPlaneShapeData*)dataBuffer;
+	btCollisionShape::serialize(&planeData->m_collisionShapeData, serializer);
 
 	m_localScaling.serializeFloat(planeData->m_localScaling);
 	m_planeNormal.serializeFloat(planeData->m_planeNormal);
@@ -104,8 +100,4 @@ SIMD_FORCE_INLINE	const char*	btStaticPlaneShape::serialize(void* dataBuffer, bt
 	return "btStaticPlaneShapeData";
 }
 
-
-#endif //BT_STATIC_PLANE_SHAPE_H
-
-
-
+#endif  //BT_STATIC_PLANE_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btStridingMeshInterface.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btStridingMeshInterface.cpp
index 78ddeb37047..eb288e99c91 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btStridingMeshInterface.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btStridingMeshInterface.cpp
@@ -18,32 +18,30 @@ subject to the following restrictions:
 
 btStridingMeshInterface::~btStridingMeshInterface()
 {
-
 }
 
-
-void	btStridingMeshInterface::InternalProcessAllTriangles(btInternalTriangleIndexCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax) const
+void btStridingMeshInterface::InternalProcessAllTriangles(btInternalTriangleIndexCallback* callback, const btVector3& aabbMin, const btVector3& aabbMax) const
 {
 	(void)aabbMin;
 	(void)aabbMax;
 	int numtotalphysicsverts = 0;
-	int part,graphicssubparts = getNumSubParts();
-	const unsigned char * vertexbase;
-	const unsigned char * indexbase;
+	int part, graphicssubparts = getNumSubParts();
+	const unsigned char* vertexbase;
+	const unsigned char* indexbase;
 	int indexstride;
 	PHY_ScalarType type;
 	PHY_ScalarType gfxindextype;
-	int stride,numverts,numtriangles;
+	int stride, numverts, numtriangles;
 	int gfxindex;
 	btVector3 triangle[3];
 
 	btVector3 meshScaling = getScaling();
 
 	///if the number of parts is big, the performance might drop due to the innerloop switch on indextype
-	for (part=0;part<graphicssubparts ;part++)
+	for (part = 0; part < graphicssubparts; part++)
 	{
-		getLockedReadOnlyVertexIndexBase(&vertexbase,numverts,type,stride,&indexbase,indexstride,numtriangles,gfxindextype,part);
-		numtotalphysicsverts+=numtriangles*3; //upper bound
+		getLockedReadOnlyVertexIndexBase(&vertexbase, numverts, type, stride, &indexbase, indexstride, numtriangles, gfxindextype, part);
+		numtotalphysicsverts += numtriangles * 3;  //upper bound
 
 		///unlike that developers want to pass in double-precision meshes in single-precision Bullet build
 		///so disable this feature by default
@@ -51,143 +49,141 @@ void	btStridingMeshInterface::InternalProcessAllTriangles(btInternalTriangleInde
 
 		switch (type)
 		{
-		case PHY_FLOAT:
-		 {
+			case PHY_FLOAT:
+			{
+				float* graphicsbase;
 
-			 float* graphicsbase;
+				switch (gfxindextype)
+				{
+					case PHY_INTEGER:
+					{
+						for (gfxindex = 0; gfxindex < numtriangles; gfxindex++)
+						{
+							unsigned int* tri_indices = (unsigned int*)(indexbase + gfxindex * indexstride);
+							graphicsbase = (float*)(vertexbase + tri_indices[0] * stride);
+							triangle[0].setValue(graphicsbase[0] * meshScaling.getX(), graphicsbase[1] * meshScaling.getY(), graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (float*)(vertexbase + tri_indices[1] * stride);
+							triangle[1].setValue(graphicsbase[0] * meshScaling.getX(), graphicsbase[1] * meshScaling.getY(), graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (float*)(vertexbase + tri_indices[2] * stride);
+							triangle[2].setValue(graphicsbase[0] * meshScaling.getX(), graphicsbase[1] * meshScaling.getY(), graphicsbase[2] * meshScaling.getZ());
+							callback->internalProcessTriangleIndex(triangle, part, gfxindex);
+						}
+						break;
+					}
+					case PHY_SHORT:
+					{
+						for (gfxindex = 0; gfxindex < numtriangles; gfxindex++)
+						{
+							unsigned short int* tri_indices = (unsigned short int*)(indexbase + gfxindex * indexstride);
+							graphicsbase = (float*)(vertexbase + tri_indices[0] * stride);
+							triangle[0].setValue(graphicsbase[0] * meshScaling.getX(), graphicsbase[1] * meshScaling.getY(), graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (float*)(vertexbase + tri_indices[1] * stride);
+							triangle[1].setValue(graphicsbase[0] * meshScaling.getX(), graphicsbase[1] * meshScaling.getY(), graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (float*)(vertexbase + tri_indices[2] * stride);
+							triangle[2].setValue(graphicsbase[0] * meshScaling.getX(), graphicsbase[1] * meshScaling.getY(), graphicsbase[2] * meshScaling.getZ());
+							callback->internalProcessTriangleIndex(triangle, part, gfxindex);
+						}
+						break;
+					}
+					case PHY_UCHAR:
+					{
+						for (gfxindex = 0; gfxindex < numtriangles; gfxindex++)
+						{
+							unsigned char* tri_indices = (unsigned char*)(indexbase + gfxindex * indexstride);
+							graphicsbase = (float*)(vertexbase + tri_indices[0] * stride);
+							triangle[0].setValue(graphicsbase[0] * meshScaling.getX(), graphicsbase[1] * meshScaling.getY(), graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (float*)(vertexbase + tri_indices[1] * stride);
+							triangle[1].setValue(graphicsbase[0] * meshScaling.getX(), graphicsbase[1] * meshScaling.getY(), graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (float*)(vertexbase + tri_indices[2] * stride);
+							triangle[2].setValue(graphicsbase[0] * meshScaling.getX(), graphicsbase[1] * meshScaling.getY(), graphicsbase[2] * meshScaling.getZ());
+							callback->internalProcessTriangleIndex(triangle, part, gfxindex);
+						}
+						break;
+					}
+					default:
+						btAssert((gfxindextype == PHY_INTEGER) || (gfxindextype == PHY_SHORT));
+				}
+				break;
+			}
 
-			 switch (gfxindextype)
-			 {
-			 case PHY_INTEGER:
-				 {
-					 for (gfxindex=0;gfxindex<numtriangles;gfxindex++)
-					 {
-						 unsigned int* tri_indices= (unsigned int*)(indexbase+gfxindex*indexstride);
-						 graphicsbase = (float*)(vertexbase+tri_indices[0]*stride);
-						 triangle[0].setValue(graphicsbase[0]*meshScaling.getX(),graphicsbase[1]*meshScaling.getY(),graphicsbase[2]*meshScaling.getZ());
-						 graphicsbase = (float*)(vertexbase+tri_indices[1]*stride);
-						 triangle[1].setValue(graphicsbase[0]*meshScaling.getX(),graphicsbase[1]*meshScaling.getY(),	graphicsbase[2]*meshScaling.getZ());
-						 graphicsbase = (float*)(vertexbase+tri_indices[2]*stride);
-						 triangle[2].setValue(graphicsbase[0]*meshScaling.getX(),graphicsbase[1]*meshScaling.getY(),	graphicsbase[2]*meshScaling.getZ());
-						 callback->internalProcessTriangleIndex(triangle,part,gfxindex);
-					 }
-					 break;
-				 }
-			 case PHY_SHORT:
-				 {
-					 for (gfxindex=0;gfxindex<numtriangles;gfxindex++)
-					 {
-						 unsigned short int* tri_indices= (unsigned short int*)(indexbase+gfxindex*indexstride);
-						 graphicsbase = (float*)(vertexbase+tri_indices[0]*stride);
-						 triangle[0].setValue(graphicsbase[0]*meshScaling.getX(),graphicsbase[1]*meshScaling.getY(),graphicsbase[2]*meshScaling.getZ());
-						 graphicsbase = (float*)(vertexbase+tri_indices[1]*stride);
-						 triangle[1].setValue(graphicsbase[0]*meshScaling.getX(),graphicsbase[1]*meshScaling.getY(),	graphicsbase[2]*meshScaling.getZ());
-						 graphicsbase = (float*)(vertexbase+tri_indices[2]*stride);
-						 triangle[2].setValue(graphicsbase[0]*meshScaling.getX(),graphicsbase[1]*meshScaling.getY(),	graphicsbase[2]*meshScaling.getZ());
-						 callback->internalProcessTriangleIndex(triangle,part,gfxindex);
-					 }
-					 break;
-				 }
-			case PHY_UCHAR:
-				 {
-					 for (gfxindex=0;gfxindex<numtriangles;gfxindex++)
-					 {
-						 unsigned char* tri_indices= (unsigned char*)(indexbase+gfxindex*indexstride);
-						 graphicsbase = (float*)(vertexbase+tri_indices[0]*stride);
-						 triangle[0].setValue(graphicsbase[0]*meshScaling.getX(),graphicsbase[1]*meshScaling.getY(),graphicsbase[2]*meshScaling.getZ());
-						 graphicsbase = (float*)(vertexbase+tri_indices[1]*stride);
-						 triangle[1].setValue(graphicsbase[0]*meshScaling.getX(),graphicsbase[1]*meshScaling.getY(),	graphicsbase[2]*meshScaling.getZ());
-						 graphicsbase = (float*)(vertexbase+tri_indices[2]*stride);
-						 triangle[2].setValue(graphicsbase[0]*meshScaling.getX(),graphicsbase[1]*meshScaling.getY(),	graphicsbase[2]*meshScaling.getZ());
-						 callback->internalProcessTriangleIndex(triangle,part,gfxindex);
-					 }
-					 break;
-				 }
-			 default:
-				 btAssert((gfxindextype == PHY_INTEGER) || (gfxindextype == PHY_SHORT));
-			 }
-			 break;
-		 }
-
-		case PHY_DOUBLE:
+			case PHY_DOUBLE:
 			{
 				double* graphicsbase;
 
 				switch (gfxindextype)
 				{
-				case PHY_INTEGER:
+					case PHY_INTEGER:
 					{
-						for (gfxindex=0;gfxindex<numtriangles;gfxindex++)
+						for (gfxindex = 0; gfxindex < numtriangles; gfxindex++)
 						{
-							unsigned int* tri_indices= (unsigned int*)(indexbase+gfxindex*indexstride);
-							graphicsbase = (double*)(vertexbase+tri_indices[0]*stride);
-							triangle[0].setValue((btScalar)graphicsbase[0]*meshScaling.getX(),(btScalar)graphicsbase[1]*meshScaling.getY(),(btScalar)graphicsbase[2]*meshScaling.getZ());
-							graphicsbase = (double*)(vertexbase+tri_indices[1]*stride);
-							triangle[1].setValue((btScalar)graphicsbase[0]*meshScaling.getX(),(btScalar)graphicsbase[1]*meshScaling.getY(),  (btScalar)graphicsbase[2]*meshScaling.getZ());
-							graphicsbase = (double*)(vertexbase+tri_indices[2]*stride);
-							triangle[2].setValue((btScalar)graphicsbase[0]*meshScaling.getX(),(btScalar)graphicsbase[1]*meshScaling.getY(),  (btScalar)graphicsbase[2]*meshScaling.getZ());
-							callback->internalProcessTriangleIndex(triangle,part,gfxindex);
+							unsigned int* tri_indices = (unsigned int*)(indexbase + gfxindex * indexstride);
+							graphicsbase = (double*)(vertexbase + tri_indices[0] * stride);
+							triangle[0].setValue((btScalar)graphicsbase[0] * meshScaling.getX(), (btScalar)graphicsbase[1] * meshScaling.getY(), (btScalar)graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (double*)(vertexbase + tri_indices[1] * stride);
+							triangle[1].setValue((btScalar)graphicsbase[0] * meshScaling.getX(), (btScalar)graphicsbase[1] * meshScaling.getY(), (btScalar)graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (double*)(vertexbase + tri_indices[2] * stride);
+							triangle[2].setValue((btScalar)graphicsbase[0] * meshScaling.getX(), (btScalar)graphicsbase[1] * meshScaling.getY(), (btScalar)graphicsbase[2] * meshScaling.getZ());
+							callback->internalProcessTriangleIndex(triangle, part, gfxindex);
 						}
 						break;
 					}
-				case PHY_SHORT:
+					case PHY_SHORT:
 					{
-						for (gfxindex=0;gfxindex<numtriangles;gfxindex++)
+						for (gfxindex = 0; gfxindex < numtriangles; gfxindex++)
 						{
-							unsigned short int* tri_indices= (unsigned short int*)(indexbase+gfxindex*indexstride);
-							graphicsbase = (double*)(vertexbase+tri_indices[0]*stride);
-							triangle[0].setValue((btScalar)graphicsbase[0]*meshScaling.getX(),(btScalar)graphicsbase[1]*meshScaling.getY(),(btScalar)graphicsbase[2]*meshScaling.getZ());
-							graphicsbase = (double*)(vertexbase+tri_indices[1]*stride);
-							triangle[1].setValue((btScalar)graphicsbase[0]*meshScaling.getX(),(btScalar)graphicsbase[1]*meshScaling.getY(),  (btScalar)graphicsbase[2]*meshScaling.getZ());
-							graphicsbase = (double*)(vertexbase+tri_indices[2]*stride);
-							triangle[2].setValue((btScalar)graphicsbase[0]*meshScaling.getX(),(btScalar)graphicsbase[1]*meshScaling.getY(),  (btScalar)graphicsbase[2]*meshScaling.getZ());
-							callback->internalProcessTriangleIndex(triangle,part,gfxindex);
+							unsigned short int* tri_indices = (unsigned short int*)(indexbase + gfxindex * indexstride);
+							graphicsbase = (double*)(vertexbase + tri_indices[0] * stride);
+							triangle[0].setValue((btScalar)graphicsbase[0] * meshScaling.getX(), (btScalar)graphicsbase[1] * meshScaling.getY(), (btScalar)graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (double*)(vertexbase + tri_indices[1] * stride);
+							triangle[1].setValue((btScalar)graphicsbase[0] * meshScaling.getX(), (btScalar)graphicsbase[1] * meshScaling.getY(), (btScalar)graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (double*)(vertexbase + tri_indices[2] * stride);
+							triangle[2].setValue((btScalar)graphicsbase[0] * meshScaling.getX(), (btScalar)graphicsbase[1] * meshScaling.getY(), (btScalar)graphicsbase[2] * meshScaling.getZ());
+							callback->internalProcessTriangleIndex(triangle, part, gfxindex);
 						}
 						break;
 					}
-				case PHY_UCHAR:
+					case PHY_UCHAR:
 					{
-						for (gfxindex=0;gfxindex<numtriangles;gfxindex++)
+						for (gfxindex = 0; gfxindex < numtriangles; gfxindex++)
 						{
-							unsigned char* tri_indices= (unsigned char*)(indexbase+gfxindex*indexstride);
-							graphicsbase = (double*)(vertexbase+tri_indices[0]*stride);
-							triangle[0].setValue((btScalar)graphicsbase[0]*meshScaling.getX(),(btScalar)graphicsbase[1]*meshScaling.getY(),(btScalar)graphicsbase[2]*meshScaling.getZ());
-							graphicsbase = (double*)(vertexbase+tri_indices[1]*stride);
-							triangle[1].setValue((btScalar)graphicsbase[0]*meshScaling.getX(),(btScalar)graphicsbase[1]*meshScaling.getY(),  (btScalar)graphicsbase[2]*meshScaling.getZ());
-							graphicsbase = (double*)(vertexbase+tri_indices[2]*stride);
-							triangle[2].setValue((btScalar)graphicsbase[0]*meshScaling.getX(),(btScalar)graphicsbase[1]*meshScaling.getY(),  (btScalar)graphicsbase[2]*meshScaling.getZ());
-							callback->internalProcessTriangleIndex(triangle,part,gfxindex);
+							unsigned char* tri_indices = (unsigned char*)(indexbase + gfxindex * indexstride);
+							graphicsbase = (double*)(vertexbase + tri_indices[0] * stride);
+							triangle[0].setValue((btScalar)graphicsbase[0] * meshScaling.getX(), (btScalar)graphicsbase[1] * meshScaling.getY(), (btScalar)graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (double*)(vertexbase + tri_indices[1] * stride);
+							triangle[1].setValue((btScalar)graphicsbase[0] * meshScaling.getX(), (btScalar)graphicsbase[1] * meshScaling.getY(), (btScalar)graphicsbase[2] * meshScaling.getZ());
+							graphicsbase = (double*)(vertexbase + tri_indices[2] * stride);
+							triangle[2].setValue((btScalar)graphicsbase[0] * meshScaling.getX(), (btScalar)graphicsbase[1] * meshScaling.getY(), (btScalar)graphicsbase[2] * meshScaling.getZ());
+							callback->internalProcessTriangleIndex(triangle, part, gfxindex);
 						}
 						break;
 					}
-				default:
-					btAssert((gfxindextype == PHY_INTEGER) || (gfxindextype == PHY_SHORT));
+					default:
+						btAssert((gfxindextype == PHY_INTEGER) || (gfxindextype == PHY_SHORT));
 				}
 				break;
 			}
-		default:
-			btAssert((type == PHY_FLOAT) || (type == PHY_DOUBLE));
+			default:
+				btAssert((type == PHY_FLOAT) || (type == PHY_DOUBLE));
 		}
 
 		unLockReadOnlyVertexBase(part);
 	}
 }
 
-void	btStridingMeshInterface::calculateAabbBruteForce(btVector3& aabbMin,btVector3& aabbMax)
+void btStridingMeshInterface::calculateAabbBruteForce(btVector3& aabbMin, btVector3& aabbMax)
 {
-
-	struct	AabbCalculationCallback : public btInternalTriangleIndexCallback
+	struct AabbCalculationCallback : public btInternalTriangleIndexCallback
 	{
-		btVector3	m_aabbMin;
-		btVector3	m_aabbMax;
+		btVector3 m_aabbMin;
+		btVector3 m_aabbMax;
 
 		AabbCalculationCallback()
 		{
-			m_aabbMin.setValue(btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT));
-			m_aabbMax.setValue(btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT));
+			m_aabbMin.setValue(btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT));
+			m_aabbMax.setValue(btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT));
 		}
 
-		virtual void internalProcessTriangleIndex(btVector3* triangle,int partId,int  triangleIndex)
+		virtual void internalProcessTriangleIndex(btVector3* triangle, int partId, int triangleIndex)
 		{
 			(void)partId;
 			(void)triangleIndex;
@@ -202,21 +198,19 @@ void	btStridingMeshInterface::calculateAabbBruteForce(btVector3& aabbMin,btVecto
 	};
 
 	//first calculate the total aabb for all triangles
-	AabbCalculationCallback	aabbCallback;
-	aabbMin.setValue(btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT),btScalar(-BT_LARGE_FLOAT));
-	aabbMax.setValue(btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT));
-	InternalProcessAllTriangles(&aabbCallback,aabbMin,aabbMax);
+	AabbCalculationCallback aabbCallback;
+	aabbMin.setValue(btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT), btScalar(-BT_LARGE_FLOAT));
+	aabbMax.setValue(btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT));
+	InternalProcessAllTriangles(&aabbCallback, aabbMin, aabbMax);
 
 	aabbMin = aabbCallback.m_aabbMin;
 	aabbMax = aabbCallback.m_aabbMax;
 }
 
-
-
 ///fills the dataBuffer and returns the struct name (and 0 on failure)
-const char*	btStridingMeshInterface::serialize(void* dataBuffer, btSerializer* serializer) const
+const char* btStridingMeshInterface::serialize(void* dataBuffer, btSerializer* serializer) const
 {
-	btStridingMeshInterfaceData* trimeshData = (btStridingMeshInterfaceData*) dataBuffer;
+	btStridingMeshInterfaceData* trimeshData = (btStridingMeshInterfaceData*)dataBuffer;
 
 	trimeshData->m_numMeshParts = getNumSubParts();
 
@@ -226,29 +220,28 @@ const char*	btStridingMeshInterface::serialize(void* dataBuffer, btSerializer* s
 
 	if (trimeshData->m_numMeshParts)
 	{
-		btChunk* chunk = serializer->allocate(sizeof(btMeshPartData),trimeshData->m_numMeshParts);
+		btChunk* chunk = serializer->allocate(sizeof(btMeshPartData), trimeshData->m_numMeshParts);
 		btMeshPartData* memPtr = (btMeshPartData*)chunk->m_oldPtr;
-		trimeshData->m_meshPartsPtr = (btMeshPartData *)serializer->getUniquePointer(memPtr);
+		trimeshData->m_meshPartsPtr = (btMeshPartData*)serializer->getUniquePointer(memPtr);
 
-
-	//	int numtotalphysicsverts = 0;
-		int part,graphicssubparts = getNumSubParts();
-		const unsigned char * vertexbase;
-		const unsigned char * indexbase;
+		//	int numtotalphysicsverts = 0;
+		int part, graphicssubparts = getNumSubParts();
+		const unsigned char* vertexbase;
+		const unsigned char* indexbase;
 		int indexstride;
 		PHY_ScalarType type;
 		PHY_ScalarType gfxindextype;
-		int stride,numverts,numtriangles;
+		int stride, numverts, numtriangles;
 		int gfxindex;
-	//	btVector3 triangle[3];
+		//	btVector3 triangle[3];
 
-	//	btVector3 meshScaling = getScaling();
+		//	btVector3 meshScaling = getScaling();
 
 		///if the number of parts is big, the performance might drop due to the innerloop switch on indextype
-		for (part=0;part<graphicssubparts ;part++,memPtr++)
+		for (part = 0; part < graphicssubparts; part++, memPtr++)
 		{
-			getLockedReadOnlyVertexIndexBase(&vertexbase,numverts,type,stride,&indexbase,indexstride,numtriangles,gfxindextype,part);
-			memPtr->m_numTriangles = numtriangles;//indices = 3*numtriangles
+			getLockedReadOnlyVertexIndexBase(&vertexbase, numverts, type, stride, &indexbase, indexstride, numtriangles, gfxindextype, part);
+			memPtr->m_numTriangles = numtriangles;  //indices = 3*numtriangles
 			memPtr->m_numVertices = numverts;
 			memPtr->m_indices16 = 0;
 			memPtr->m_indices32 = 0;
@@ -257,39 +250,38 @@ const char*	btStridingMeshInterface::serialize(void* dataBuffer, btSerializer* s
 			memPtr->m_vertices3f = 0;
 			memPtr->m_vertices3d = 0;
 
-
 			switch (gfxindextype)
 			{
-			case PHY_INTEGER:
+				case PHY_INTEGER:
 				{
-					int numindices = numtriangles*3;
-				
+					int numindices = numtriangles * 3;
+
 					if (numindices)
 					{
-						btChunk* chunk = serializer->allocate(sizeof(btIntIndexData),numindices);
+						btChunk* chunk = serializer->allocate(sizeof(btIntIndexData), numindices);
 						btIntIndexData* tmpIndices = (btIntIndexData*)chunk->m_oldPtr;
 						memPtr->m_indices32 = (btIntIndexData*)serializer->getUniquePointer(tmpIndices);
-						for (gfxindex=0;gfxindex<numtriangles;gfxindex++)
+						for (gfxindex = 0; gfxindex < numtriangles; gfxindex++)
 						{
-							unsigned int* tri_indices= (unsigned int*)(indexbase+gfxindex*indexstride);
-							tmpIndices[gfxindex*3].m_value = tri_indices[0];
-							tmpIndices[gfxindex*3+1].m_value = tri_indices[1];
-							tmpIndices[gfxindex*3+2].m_value = tri_indices[2];
+							unsigned int* tri_indices = (unsigned int*)(indexbase + gfxindex * indexstride);
+							tmpIndices[gfxindex * 3].m_value = tri_indices[0];
+							tmpIndices[gfxindex * 3 + 1].m_value = tri_indices[1];
+							tmpIndices[gfxindex * 3 + 2].m_value = tri_indices[2];
 						}
-						serializer->finalizeChunk(chunk,"btIntIndexData",BT_ARRAY_CODE,(void*)chunk->m_oldPtr);
+						serializer->finalizeChunk(chunk, "btIntIndexData", BT_ARRAY_CODE, (void*)chunk->m_oldPtr);
 					}
 					break;
 				}
-			case PHY_SHORT:
+				case PHY_SHORT:
 				{
 					if (numtriangles)
 					{
-						btChunk* chunk = serializer->allocate(sizeof(btShortIntIndexTripletData),numtriangles);
+						btChunk* chunk = serializer->allocate(sizeof(btShortIntIndexTripletData), numtriangles);
 						btShortIntIndexTripletData* tmpIndices = (btShortIntIndexTripletData*)chunk->m_oldPtr;
-						memPtr->m_3indices16 = (btShortIntIndexTripletData*) serializer->getUniquePointer(tmpIndices);
-						for (gfxindex=0;gfxindex<numtriangles;gfxindex++)
+						memPtr->m_3indices16 = (btShortIntIndexTripletData*)serializer->getUniquePointer(tmpIndices);
+						for (gfxindex = 0; gfxindex < numtriangles; gfxindex++)
 						{
-							unsigned short int* tri_indices= (unsigned short int*)(indexbase+gfxindex*indexstride);
+							unsigned short int* tri_indices = (unsigned short int*)(indexbase + gfxindex * indexstride);
 							tmpIndices[gfxindex].m_values[0] = tri_indices[0];
 							tmpIndices[gfxindex].m_values[1] = tri_indices[1];
 							tmpIndices[gfxindex].m_values[2] = tri_indices[2];
@@ -297,7 +289,7 @@ const char*	btStridingMeshInterface::serialize(void* dataBuffer, btSerializer* s
 							tmpIndices[gfxindex].m_pad[0] = 0;
 							tmpIndices[gfxindex].m_pad[1] = 0;
 						}
-						serializer->finalizeChunk(chunk,"btShortIntIndexTripletData",BT_ARRAY_CODE,(void*)chunk->m_oldPtr);
+						serializer->finalizeChunk(chunk, "btShortIntIndexTripletData", BT_ARRAY_CODE, (void*)chunk->m_oldPtr);
 					}
 					break;
 				}
@@ -305,23 +297,23 @@ const char*	btStridingMeshInterface::serialize(void* dataBuffer, btSerializer* s
 				{
 					if (numtriangles)
 					{
-						btChunk* chunk = serializer->allocate(sizeof(btCharIndexTripletData),numtriangles);
+						btChunk* chunk = serializer->allocate(sizeof(btCharIndexTripletData), numtriangles);
 						btCharIndexTripletData* tmpIndices = (btCharIndexTripletData*)chunk->m_oldPtr;
-						memPtr->m_3indices8 = (btCharIndexTripletData*) serializer->getUniquePointer(tmpIndices);
-						for (gfxindex=0;gfxindex<numtriangles;gfxindex++)
+						memPtr->m_3indices8 = (btCharIndexTripletData*)serializer->getUniquePointer(tmpIndices);
+						for (gfxindex = 0; gfxindex < numtriangles; gfxindex++)
 						{
-							unsigned char* tri_indices= (unsigned char*)(indexbase+gfxindex*indexstride);
+							unsigned char* tri_indices = (unsigned char*)(indexbase + gfxindex * indexstride);
 							tmpIndices[gfxindex].m_values[0] = tri_indices[0];
 							tmpIndices[gfxindex].m_values[1] = tri_indices[1];
 							tmpIndices[gfxindex].m_values[2] = tri_indices[2];
 							// Fill padding with zeros to appease msan.
 							tmpIndices[gfxindex].m_pad = 0;
 						}
-						serializer->finalizeChunk(chunk,"btCharIndexTripletData",BT_ARRAY_CODE,(void*)chunk->m_oldPtr);
+						serializer->finalizeChunk(chunk, "btCharIndexTripletData", BT_ARRAY_CODE, (void*)chunk->m_oldPtr);
 					}
 					break;
 				}
-			default:
+				default:
 				{
 					btAssert(0);
 					//unknown index type
@@ -330,54 +322,54 @@ const char*	btStridingMeshInterface::serialize(void* dataBuffer, btSerializer* s
 
 			switch (type)
 			{
-			case PHY_FLOAT:
-			 {
-				 float* graphicsbase;
-
-				 if (numverts)
-				 {
-					 btChunk* chunk = serializer->allocate(sizeof(btVector3FloatData),numverts);
-					 btVector3FloatData* tmpVertices = (btVector3FloatData*) chunk->m_oldPtr;
-					 memPtr->m_vertices3f = (btVector3FloatData *)serializer->getUniquePointer(tmpVertices);
-					 for (int i=0;i<numverts;i++)
-					 {
-						 graphicsbase = (float*)(vertexbase+i*stride);
-						 tmpVertices[i].m_floats[0] = graphicsbase[0];
-						 tmpVertices[i].m_floats[1] = graphicsbase[1];
-						 tmpVertices[i].m_floats[2] = graphicsbase[2];
-					 }
-					 serializer->finalizeChunk(chunk,"btVector3FloatData",BT_ARRAY_CODE,(void*)chunk->m_oldPtr);
-				 }
-				 break;
-				}
-
-			case PHY_DOUBLE:
+				case PHY_FLOAT:
 				{
+					float* graphicsbase;
+
 					if (numverts)
 					{
-						btChunk* chunk = serializer->allocate(sizeof(btVector3DoubleData),numverts);
-						btVector3DoubleData* tmpVertices = (btVector3DoubleData*) chunk->m_oldPtr;
-						memPtr->m_vertices3d = (btVector3DoubleData *) serializer->getUniquePointer(tmpVertices);
-						for (int i=0;i<numverts;i++)
-					 {
-						 double* graphicsbase = (double*)(vertexbase+i*stride);//for now convert to float, might leave it at double
-						 tmpVertices[i].m_floats[0] = graphicsbase[0];
-						 tmpVertices[i].m_floats[1] = graphicsbase[1];
-						 tmpVertices[i].m_floats[2] = graphicsbase[2];
-					 }
-						serializer->finalizeChunk(chunk,"btVector3DoubleData",BT_ARRAY_CODE,(void*)chunk->m_oldPtr);
+						btChunk* chunk = serializer->allocate(sizeof(btVector3FloatData), numverts);
+						btVector3FloatData* tmpVertices = (btVector3FloatData*)chunk->m_oldPtr;
+						memPtr->m_vertices3f = (btVector3FloatData*)serializer->getUniquePointer(tmpVertices);
+						for (int i = 0; i < numverts; i++)
+						{
+							graphicsbase = (float*)(vertexbase + i * stride);
+							tmpVertices[i].m_floats[0] = graphicsbase[0];
+							tmpVertices[i].m_floats[1] = graphicsbase[1];
+							tmpVertices[i].m_floats[2] = graphicsbase[2];
+						}
+						serializer->finalizeChunk(chunk, "btVector3FloatData", BT_ARRAY_CODE, (void*)chunk->m_oldPtr);
 					}
 					break;
 				}
 
-			default:
-				btAssert((type == PHY_FLOAT) || (type == PHY_DOUBLE));
+				case PHY_DOUBLE:
+				{
+					if (numverts)
+					{
+						btChunk* chunk = serializer->allocate(sizeof(btVector3DoubleData), numverts);
+						btVector3DoubleData* tmpVertices = (btVector3DoubleData*)chunk->m_oldPtr;
+						memPtr->m_vertices3d = (btVector3DoubleData*)serializer->getUniquePointer(tmpVertices);
+						for (int i = 0; i < numverts; i++)
+						{
+							double* graphicsbase = (double*)(vertexbase + i * stride);  //for now convert to float, might leave it at double
+							tmpVertices[i].m_floats[0] = graphicsbase[0];
+							tmpVertices[i].m_floats[1] = graphicsbase[1];
+							tmpVertices[i].m_floats[2] = graphicsbase[2];
+						}
+						serializer->finalizeChunk(chunk, "btVector3DoubleData", BT_ARRAY_CODE, (void*)chunk->m_oldPtr);
+					}
+					break;
+				}
+
+				default:
+					btAssert((type == PHY_FLOAT) || (type == PHY_DOUBLE));
 			}
 
 			unLockReadOnlyVertexBase(part);
 		}
 
-		serializer->finalizeChunk(chunk,"btMeshPartData",BT_ARRAY_CODE,chunk->m_oldPtr);
+		serializer->finalizeChunk(chunk, "btMeshPartData", BT_ARRAY_CODE, chunk->m_oldPtr);
 	}
 
 	// Fill padding with zeros to appease msan.
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btStridingMeshInterface.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btStridingMeshInterface.h
index 9fbe139768d..7d729ee0d73 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btStridingMeshInterface.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btStridingMeshInterface.h
@@ -20,110 +20,102 @@ subject to the following restrictions:
 #include "btTriangleCallback.h"
 #include "btConcaveShape.h"
 
-
-
-
-
 ///	The btStridingMeshInterface is the interface class for high performance generic access to triangle meshes, used in combination with btBvhTriangleMeshShape and some other collision shapes.
 /// Using index striding of 3*sizeof(integer) it can use triangle arrays, using index striding of 1*sizeof(integer) it can handle triangle strips.
 /// It allows for sharing graphics and collision meshes. Also it provides locking/unlocking of graphics meshes that are in gpu memory.
-ATTRIBUTE_ALIGNED16(class ) btStridingMeshInterface
+ATTRIBUTE_ALIGNED16(class)
+btStridingMeshInterface
 {
-	protected:
-	
-		btVector3 m_scaling;
+protected:
+	btVector3 m_scaling;
 
-	public:
-		BT_DECLARE_ALIGNED_ALLOCATOR();
-		
-		btStridingMeshInterface() :m_scaling(btScalar(1.),btScalar(1.),btScalar(1.))
-		{
+public:
+	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-		}
+	btStridingMeshInterface() : m_scaling(btScalar(1.), btScalar(1.), btScalar(1.))
+	{
+	}
 
-		virtual ~btStridingMeshInterface();
+	virtual ~btStridingMeshInterface();
 
+	virtual void InternalProcessAllTriangles(btInternalTriangleIndexCallback * callback, const btVector3& aabbMin, const btVector3& aabbMax) const;
 
+	///brute force method to calculate aabb
+	void calculateAabbBruteForce(btVector3 & aabbMin, btVector3 & aabbMax);
 
-		virtual void	InternalProcessAllTriangles(btInternalTriangleIndexCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax) const;
+	/// get read and write access to a subpart of a triangle mesh
+	/// this subpart has a continuous array of vertices and indices
+	/// in this way the mesh can be handled as chunks of memory with striding
+	/// very similar to OpenGL vertexarray support
+	/// make a call to unLockVertexBase when the read and write access is finished
+	virtual void getLockedVertexIndexBase(unsigned char** vertexbase, int& numverts, PHY_ScalarType& type, int& stride, unsigned char** indexbase, int& indexstride, int& numfaces, PHY_ScalarType& indicestype, int subpart = 0) = 0;
 
-		///brute force method to calculate aabb
-		void	calculateAabbBruteForce(btVector3& aabbMin,btVector3& aabbMax);
+	virtual void getLockedReadOnlyVertexIndexBase(const unsigned char** vertexbase, int& numverts, PHY_ScalarType& type, int& stride, const unsigned char** indexbase, int& indexstride, int& numfaces, PHY_ScalarType& indicestype, int subpart = 0) const = 0;
 
-		/// get read and write access to a subpart of a triangle mesh
-		/// this subpart has a continuous array of vertices and indices
-		/// in this way the mesh can be handled as chunks of memory with striding
-		/// very similar to OpenGL vertexarray support
-		/// make a call to unLockVertexBase when the read and write access is finished	
-		virtual void	getLockedVertexIndexBase(unsigned char **vertexbase, int& numverts,PHY_ScalarType& type, int& stride,unsigned char **indexbase,int & indexstride,int& numfaces,PHY_ScalarType& indicestype,int subpart=0)=0;
-		
-		virtual void	getLockedReadOnlyVertexIndexBase(const unsigned char **vertexbase, int& numverts,PHY_ScalarType& type, int& stride,const unsigned char **indexbase,int & indexstride,int& numfaces,PHY_ScalarType& indicestype,int subpart=0) const=0;
-	
-		/// unLockVertexBase finishes the access to a subpart of the triangle mesh
-		/// make a call to unLockVertexBase when the read and write access (using getLockedVertexIndexBase) is finished
-		virtual void	unLockVertexBase(int subpart)=0;
+	/// unLockVertexBase finishes the access to a subpart of the triangle mesh
+	/// make a call to unLockVertexBase when the read and write access (using getLockedVertexIndexBase) is finished
+	virtual void unLockVertexBase(int subpart) = 0;
 
-		virtual void	unLockReadOnlyVertexBase(int subpart) const=0;
+	virtual void unLockReadOnlyVertexBase(int subpart) const = 0;
 
+	/// getNumSubParts returns the number of seperate subparts
+	/// each subpart has a continuous array of vertices and indices
+	virtual int getNumSubParts() const = 0;
 
-		/// getNumSubParts returns the number of seperate subparts
-		/// each subpart has a continuous array of vertices and indices
-		virtual int		getNumSubParts() const=0;
+	virtual void preallocateVertices(int numverts) = 0;
+	virtual void preallocateIndices(int numindices) = 0;
 
-		virtual void	preallocateVertices(int numverts)=0;
-		virtual void	preallocateIndices(int numindices)=0;
+	virtual bool hasPremadeAabb() const { return false; }
+	virtual void setPremadeAabb(const btVector3& aabbMin, const btVector3& aabbMax) const
+	{
+		(void)aabbMin;
+		(void)aabbMax;
+	}
+	virtual void getPremadeAabb(btVector3 * aabbMin, btVector3 * aabbMax) const
+	{
+		(void)aabbMin;
+		(void)aabbMax;
+	}
 
-		virtual bool	hasPremadeAabb() const { return false; }
-		virtual void	setPremadeAabb(const btVector3& aabbMin, const btVector3& aabbMax ) const
-                {
-                        (void) aabbMin;
-                        (void) aabbMax;
-                }
-		virtual void	getPremadeAabb(btVector3* aabbMin, btVector3* aabbMax ) const
-        {
-            (void) aabbMin;
-            (void) aabbMax;
-        }
-
-		const btVector3&	getScaling() const {
-			return m_scaling;
-		}
-		void	setScaling(const btVector3& scaling)
-		{
-			m_scaling = scaling;
-		}
-
-		virtual	int	calculateSerializeBufferSize() const;
-
-		///fills the dataBuffer and returns the struct name (and 0 on failure)
-		virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
+	const btVector3& getScaling() const
+	{
+		return m_scaling;
+	}
+	void setScaling(const btVector3& scaling)
+	{
+		m_scaling = scaling;
+	}
 
+	virtual int calculateSerializeBufferSize() const;
 
+	///fills the dataBuffer and returns the struct name (and 0 on failure)
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 };
 
-struct	btIntIndexData
+struct btIntIndexData
 {
-	int	m_value;
+	int m_value;
 };
 
-struct	btShortIntIndexData
+struct btShortIntIndexData
 {
 	short m_value;
 	char m_pad[2];
 };
 
-struct	btShortIntIndexTripletData
+struct btShortIntIndexTripletData
 {
-	short	m_values[3];
-	char	m_pad[2];
+	short m_values[3];
+	char m_pad[2];
 };
 
-struct	btCharIndexTripletData
+struct btCharIndexTripletData
 {
 	unsigned char m_values[3];
-	char	m_pad;
+	char m_pad;
 };
 
+// clang-format off
 
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
 struct	btMeshPartData
@@ -151,14 +143,11 @@ struct	btStridingMeshInterfaceData
 	char m_padding[4];
 };
 
+// clang-format on
 
-
-
-SIMD_FORCE_INLINE	int	btStridingMeshInterface::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btStridingMeshInterface::calculateSerializeBufferSize() const
 {
 	return sizeof(btStridingMeshInterfaceData);
 }
 
-
-
-#endif //BT_STRIDING_MESHINTERFACE_H
+#endif  //BT_STRIDING_MESHINTERFACE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btTetrahedronShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btTetrahedronShape.cpp
index 52f346bf726..c4d33c429f0 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btTetrahedronShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btTetrahedronShape.cpp
@@ -16,29 +16,29 @@ subject to the following restrictions:
 #include "btTetrahedronShape.h"
 #include "LinearMath/btMatrix3x3.h"
 
-btBU_Simplex1to4::btBU_Simplex1to4() : btPolyhedralConvexAabbCachingShape (),
-m_numVertices(0)
+btBU_Simplex1to4::btBU_Simplex1to4() : btPolyhedralConvexAabbCachingShape(),
+									   m_numVertices(0)
 {
 	m_shapeType = TETRAHEDRAL_SHAPE_PROXYTYPE;
 }
 
-btBU_Simplex1to4::btBU_Simplex1to4(const btVector3& pt0) : btPolyhedralConvexAabbCachingShape (),
-m_numVertices(0)
+btBU_Simplex1to4::btBU_Simplex1to4(const btVector3& pt0) : btPolyhedralConvexAabbCachingShape(),
+														   m_numVertices(0)
 {
 	m_shapeType = TETRAHEDRAL_SHAPE_PROXYTYPE;
 	addVertex(pt0);
 }
 
-btBU_Simplex1to4::btBU_Simplex1to4(const btVector3& pt0,const btVector3& pt1) : btPolyhedralConvexAabbCachingShape (),
-m_numVertices(0)
+btBU_Simplex1to4::btBU_Simplex1to4(const btVector3& pt0, const btVector3& pt1) : btPolyhedralConvexAabbCachingShape(),
+																				 m_numVertices(0)
 {
 	m_shapeType = TETRAHEDRAL_SHAPE_PROXYTYPE;
 	addVertex(pt0);
 	addVertex(pt1);
 }
 
-btBU_Simplex1to4::btBU_Simplex1to4(const btVector3& pt0,const btVector3& pt1,const btVector3& pt2) : btPolyhedralConvexAabbCachingShape (),
-m_numVertices(0)
+btBU_Simplex1to4::btBU_Simplex1to4(const btVector3& pt0, const btVector3& pt1, const btVector3& pt2) : btPolyhedralConvexAabbCachingShape(),
+																									   m_numVertices(0)
 {
 	m_shapeType = TETRAHEDRAL_SHAPE_PROXYTYPE;
 	addVertex(pt0);
@@ -46,8 +46,8 @@ m_numVertices(0)
 	addVertex(pt2);
 }
 
-btBU_Simplex1to4::btBU_Simplex1to4(const btVector3& pt0,const btVector3& pt1,const btVector3& pt2,const btVector3& pt3) : btPolyhedralConvexAabbCachingShape (),
-m_numVertices(0)
+btBU_Simplex1to4::btBU_Simplex1to4(const btVector3& pt0, const btVector3& pt1, const btVector3& pt2, const btVector3& pt3) : btPolyhedralConvexAabbCachingShape(),
+																															 m_numVertices(0)
 {
 	m_shapeType = TETRAHEDRAL_SHAPE_PROXYTYPE;
 	addVertex(pt0);
@@ -56,17 +56,16 @@ m_numVertices(0)
 	addVertex(pt3);
 }
 
-
-void btBU_Simplex1to4::getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const
+void btBU_Simplex1to4::getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
 {
 #if 1
-	btPolyhedralConvexAabbCachingShape::getAabb(t,aabbMin,aabbMax);
+	btPolyhedralConvexAabbCachingShape::getAabb(t, aabbMin, aabbMax);
 #else
-	aabbMin.setValue(BT_LARGE_FLOAT,BT_LARGE_FLOAT,BT_LARGE_FLOAT);
-	aabbMax.setValue(-BT_LARGE_FLOAT,-BT_LARGE_FLOAT,-BT_LARGE_FLOAT);
+	aabbMin.setValue(BT_LARGE_FLOAT, BT_LARGE_FLOAT, BT_LARGE_FLOAT);
+	aabbMax.setValue(-BT_LARGE_FLOAT, -BT_LARGE_FLOAT, -BT_LARGE_FLOAT);
 
 	//just transform the vertices in worldspace, and take their AABB
-	for (int i=0;i<m_numVertices;i++)
+	for (int i = 0; i < m_numVertices; i++)
 	{
 		btVector3 worldVertex = t(m_vertices[i]);
 		aabbMin.setMin(worldVertex);
@@ -75,18 +74,13 @@ void btBU_Simplex1to4::getAabb(const btTransform& t,btVector3& aabbMin,btVector3
 #endif
 }
 
-
-
-
-
 void btBU_Simplex1to4::addVertex(const btVector3& pt)
 {
 	m_vertices[m_numVertices++] = pt;
 	recalcLocalAabb();
 }
 
-
-int	btBU_Simplex1to4::getNumVertices() const
+int btBU_Simplex1to4::getNumVertices() const
 {
 	return m_numVertices;
 }
@@ -97,122 +91,113 @@ int btBU_Simplex1to4::getNumEdges() const
 
 	switch (m_numVertices)
 	{
-	case 0:
-		return 0;
-	case 1: return 0;
-	case 2: return 1;
-	case 3: return 3;
-	case 4: return 6;
-
-
+		case 0:
+			return 0;
+		case 1:
+			return 0;
+		case 2:
+			return 1;
+		case 3:
+			return 3;
+		case 4:
+			return 6;
 	}
 
 	return 0;
 }
 
-void btBU_Simplex1to4::getEdge(int i,btVector3& pa,btVector3& pb) const
+void btBU_Simplex1to4::getEdge(int i, btVector3& pa, btVector3& pb) const
 {
-	
-    switch (m_numVertices)
+	switch (m_numVertices)
 	{
-
-	case 2: 
-		pa = m_vertices[0];
-		pb = m_vertices[1];
-		break;
-	case 3:  
-		switch (i)
-		{
-		case 0:
+		case 2:
 			pa = m_vertices[0];
 			pb = m_vertices[1];
 			break;
-		case 1:
-			pa = m_vertices[1];
-			pb = m_vertices[2];
-			break;
-		case 2:
-			pa = m_vertices[2];
-			pb = m_vertices[0];
-			break;
-
-		}
-		break;
-	case 4: 
-		switch (i)
-		{
-		case 0:
-			pa = m_vertices[0];
-			pb = m_vertices[1];
-			break;
-		case 1:
-			pa = m_vertices[1];
-			pb = m_vertices[2];
-			break;
-		case 2:
-			pa = m_vertices[2];
-			pb = m_vertices[0];
-			break;
 		case 3:
-			pa = m_vertices[0];
-			pb = m_vertices[3];
+			switch (i)
+			{
+				case 0:
+					pa = m_vertices[0];
+					pb = m_vertices[1];
+					break;
+				case 1:
+					pa = m_vertices[1];
+					pb = m_vertices[2];
+					break;
+				case 2:
+					pa = m_vertices[2];
+					pb = m_vertices[0];
+					break;
+			}
 			break;
 		case 4:
-			pa = m_vertices[1];
-			pb = m_vertices[3];
-			break;
-		case 5:
-			pa = m_vertices[2];
-			pb = m_vertices[3];
-			break;
-		}
-
+			switch (i)
+			{
+				case 0:
+					pa = m_vertices[0];
+					pb = m_vertices[1];
+					break;
+				case 1:
+					pa = m_vertices[1];
+					pb = m_vertices[2];
+					break;
+				case 2:
+					pa = m_vertices[2];
+					pb = m_vertices[0];
+					break;
+				case 3:
+					pa = m_vertices[0];
+					pb = m_vertices[3];
+					break;
+				case 4:
+					pa = m_vertices[1];
+					pb = m_vertices[3];
+					break;
+				case 5:
+					pa = m_vertices[2];
+					pb = m_vertices[3];
+					break;
+			}
 	}
-
-
-
-
 }
 
-void btBU_Simplex1to4::getVertex(int i,btVector3& vtx) const
+void btBU_Simplex1to4::getVertex(int i, btVector3& vtx) const
 {
 	vtx = m_vertices[i];
 }
 
-int	btBU_Simplex1to4::getNumPlanes() const
+int btBU_Simplex1to4::getNumPlanes() const
 {
 	switch (m_numVertices)
 	{
-	case 0:
+		case 0:
 			return 0;
-	case 1:
+		case 1:
 			return 0;
-	case 2:
+		case 2:
 			return 0;
-	case 3:
+		case 3:
 			return 2;
-	case 4:
+		case 4:
 			return 4;
-	default:
+		default:
 		{
 		}
 	}
 	return 0;
 }
 
-
-void btBU_Simplex1to4::getPlane(btVector3&, btVector3& ,int ) const
+void btBU_Simplex1to4::getPlane(btVector3&, btVector3&, int) const
 {
-	
 }
 
-int btBU_Simplex1to4::getIndex(int ) const
+int btBU_Simplex1to4::getIndex(int) const
 {
 	return 0;
 }
 
-bool btBU_Simplex1to4::isInside(const btVector3& ,btScalar ) const
+bool btBU_Simplex1to4::isInside(const btVector3&, btScalar) const
 {
 	return false;
 }
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btTetrahedronShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btTetrahedronShape.h
index b69209835ef..f5e2209ed09 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btTetrahedronShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btTetrahedronShape.h
@@ -16,61 +16,56 @@ subject to the following restrictions:
 #ifndef BT_SIMPLEX_1TO4_SHAPE
 #define BT_SIMPLEX_1TO4_SHAPE
 
-
 #include "btPolyhedralConvexShape.h"
 #include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h"
 
-
 ///The btBU_Simplex1to4 implements tetrahedron, triangle, line, vertex collision shapes. In most cases it is better to use btConvexHullShape instead.
-ATTRIBUTE_ALIGNED16(class) btBU_Simplex1to4 : public btPolyhedralConvexAabbCachingShape
+ATTRIBUTE_ALIGNED16(class)
+btBU_Simplex1to4 : public btPolyhedralConvexAabbCachingShape
 {
 protected:
-
-	int	m_numVertices;
-	btVector3	m_vertices[4];
+	int m_numVertices;
+	btVector3 m_vertices[4];
 
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
+
 	btBU_Simplex1to4();
 
 	btBU_Simplex1to4(const btVector3& pt0);
-	btBU_Simplex1to4(const btVector3& pt0,const btVector3& pt1);
-	btBU_Simplex1to4(const btVector3& pt0,const btVector3& pt1,const btVector3& pt2);
-	btBU_Simplex1to4(const btVector3& pt0,const btVector3& pt1,const btVector3& pt2,const btVector3& pt3);
+	btBU_Simplex1to4(const btVector3& pt0, const btVector3& pt1);
+	btBU_Simplex1to4(const btVector3& pt0, const btVector3& pt1, const btVector3& pt2);
+	btBU_Simplex1to4(const btVector3& pt0, const btVector3& pt1, const btVector3& pt2, const btVector3& pt3);
 
-    
-	void	reset()
+	void reset()
 	{
 		m_numVertices = 0;
 	}
-	
-	virtual void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const;
+
+	virtual void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
 
 	void addVertex(const btVector3& pt);
 
 	//PolyhedralConvexShape interface
 
-	virtual int	getNumVertices() const;
+	virtual int getNumVertices() const;
 
 	virtual int getNumEdges() const;
 
-	virtual void getEdge(int i,btVector3& pa,btVector3& pb) const;
-	
-	virtual void getVertex(int i,btVector3& vtx) const;
+	virtual void getEdge(int i, btVector3& pa, btVector3& pb) const;
 
-	virtual int	getNumPlanes() const;
+	virtual void getVertex(int i, btVector3& vtx) const;
 
-	virtual void getPlane(btVector3& planeNormal,btVector3& planeSupport,int i) const;
+	virtual int getNumPlanes() const;
+
+	virtual void getPlane(btVector3 & planeNormal, btVector3 & planeSupport, int i) const;
 
 	virtual int getIndex(int i) const;
 
-	virtual	bool isInside(const btVector3& pt,btScalar tolerance) const;
-
+	virtual bool isInside(const btVector3& pt, btScalar tolerance) const;
 
 	///getName is for debugging
-	virtual const char*	getName()const { return "btBU_Simplex1to4";}
-
+	virtual const char* getName() const { return "btBU_Simplex1to4"; }
 };
 
-#endif //BT_SIMPLEX_1TO4_SHAPE
+#endif  //BT_SIMPLEX_1TO4_SHAPE
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleBuffer.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleBuffer.cpp
index 3027e65b256..3b6db2b39fa 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleBuffer.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleBuffer.cpp
@@ -15,21 +15,14 @@ subject to the following restrictions:
 
 #include "btTriangleBuffer.h"
 
-
-
-
-
-
-
-void btTriangleBuffer::processTriangle(btVector3* triangle,int partId,int  triangleIndex)
+void btTriangleBuffer::processTriangle(btVector3* triangle, int partId, int triangleIndex)
 {
-		btTriangle	tri;
-		tri.m_vertex0 = triangle[0];
-		tri.m_vertex1 = triangle[1];
-		tri.m_vertex2 = triangle[2];
-		tri.m_partId = partId;
-		tri.m_triangleIndex = triangleIndex;
-			
-		m_triangleBuffer.push_back(tri);
-}
+	btTriangle tri;
+	tri.m_vertex0 = triangle[0];
+	tri.m_vertex1 = triangle[1];
+	tri.m_vertex2 = triangle[2];
+	tri.m_partId = partId;
+	tri.m_triangleIndex = triangleIndex;
 
+	m_triangleBuffer.push_back(tri);
+}
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleBuffer.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleBuffer.h
index b71fc8b3746..a89b9cd8a48 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleBuffer.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleBuffer.h
@@ -19,13 +19,13 @@ subject to the following restrictions:
 #include "btTriangleCallback.h"
 #include "LinearMath/btAlignedObjectArray.h"
 
-struct	btTriangle
+struct btTriangle
 {
-	btVector3	m_vertex0;
-	btVector3	m_vertex1;
-	btVector3	m_vertex2;
-	int	m_partId;
-	int	m_triangleIndex;
+	btVector3 m_vertex0;
+	btVector3 m_vertex1;
+	btVector3 m_vertex2;
+	int m_partId;
+	int m_triangleIndex;
 };
 
 ///The btTriangleBuffer callback can be useful to collect and store overlapping triangles between AABB and concave objects that support 'processAllTriangles'
@@ -39,31 +39,25 @@ struct	btTriangle
 ///			}
 class btTriangleBuffer : public btTriangleCallback
 {
+	btAlignedObjectArray<btTriangle> m_triangleBuffer;
 
-	btAlignedObjectArray<btTriangle>	m_triangleBuffer;
-	
 public:
-
-
 	virtual void processTriangle(btVector3* triangle, int partId, int triangleIndex);
-	
-	int	getNumTriangles() const
+
+	int getNumTriangles() const
 	{
 		return int(m_triangleBuffer.size());
 	}
-	
-	const btTriangle&	getTriangle(int index) const
+
+	const btTriangle& getTriangle(int index) const
 	{
 		return m_triangleBuffer[index];
 	}
 
-	void	clearBuffer()
+	void clearBuffer()
 	{
 		m_triangleBuffer.clear();
 	}
-	
 };
 
-
-#endif //BT_TRIANGLE_BUFFER_H
-
+#endif  //BT_TRIANGLE_BUFFER_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleCallback.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleCallback.cpp
index f558bf6d241..5bd2c595fed 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleCallback.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleCallback.cpp
@@ -17,12 +17,8 @@ subject to the following restrictions:
 
 btTriangleCallback::~btTriangleCallback()
 {
-
 }
 
-
 btInternalTriangleIndexCallback::~btInternalTriangleIndexCallback()
 {
-
 }
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleCallback.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleCallback.h
index 461c57f8773..d3644891ee6 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleCallback.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleCallback.h
@@ -18,13 +18,11 @@ subject to the following restrictions:
 
 #include "LinearMath/btVector3.h"
 
-
 ///The btTriangleCallback provides a callback for each overlapping triangle when calling processAllTriangles.
 ///This callback is called by processAllTriangles for all btConcaveShape derived class, such as  btBvhTriangleMeshShape, btStaticPlaneShape and btHeightfieldTerrainShape.
 class btTriangleCallback
 {
 public:
-
 	virtual ~btTriangleCallback();
 	virtual void processTriangle(btVector3* triangle, int partId, int triangleIndex) = 0;
 };
@@ -32,11 +30,8 @@ public:
 class btInternalTriangleIndexCallback
 {
 public:
-
 	virtual ~btInternalTriangleIndexCallback();
-	virtual void internalProcessTriangleIndex(btVector3* triangle,int partId,int  triangleIndex) = 0;
+	virtual void internalProcessTriangleIndex(btVector3* triangle, int partId, int triangleIndex) = 0;
 };
 
-
-
-#endif //BT_TRIANGLE_CALLBACK_H
+#endif  //BT_TRIANGLE_CALLBACK_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleIndexVertexArray.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleIndexVertexArray.cpp
index a665024cb61..dae42551941 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleIndexVertexArray.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleIndexVertexArray.cpp
@@ -15,81 +15,76 @@ subject to the following restrictions:
 
 #include "btTriangleIndexVertexArray.h"
 
-btTriangleIndexVertexArray::btTriangleIndexVertexArray(int numTriangles,int* triangleIndexBase,int triangleIndexStride,int numVertices,btScalar* vertexBase,int vertexStride)
-: m_hasAabb(0)
+btTriangleIndexVertexArray::btTriangleIndexVertexArray(int numTriangles, int* triangleIndexBase, int triangleIndexStride, int numVertices, btScalar* vertexBase, int vertexStride)
+	: m_hasAabb(0)
 {
 	btIndexedMesh mesh;
 
 	mesh.m_numTriangles = numTriangles;
-	mesh.m_triangleIndexBase = (const unsigned char *)triangleIndexBase;
+	mesh.m_triangleIndexBase = (const unsigned char*)triangleIndexBase;
 	mesh.m_triangleIndexStride = triangleIndexStride;
 	mesh.m_numVertices = numVertices;
-	mesh.m_vertexBase = (const unsigned char *)vertexBase;
+	mesh.m_vertexBase = (const unsigned char*)vertexBase;
 	mesh.m_vertexStride = vertexStride;
 
 	addIndexedMesh(mesh);
-
 }
 
 btTriangleIndexVertexArray::~btTriangleIndexVertexArray()
 {
-
 }
 
-void	btTriangleIndexVertexArray::getLockedVertexIndexBase(unsigned char **vertexbase, int& numverts,PHY_ScalarType& type, int& vertexStride,unsigned char **indexbase,int & indexstride,int& numfaces,PHY_ScalarType& indicestype,int subpart)
+void btTriangleIndexVertexArray::getLockedVertexIndexBase(unsigned char** vertexbase, int& numverts, PHY_ScalarType& type, int& vertexStride, unsigned char** indexbase, int& indexstride, int& numfaces, PHY_ScalarType& indicestype, int subpart)
 {
-	btAssert(subpart< getNumSubParts() );
+	btAssert(subpart < getNumSubParts());
 
 	btIndexedMesh& mesh = m_indexedMeshes[subpart];
 
 	numverts = mesh.m_numVertices;
-	(*vertexbase) = (unsigned char *) mesh.m_vertexBase;
+	(*vertexbase) = (unsigned char*)mesh.m_vertexBase;
 
-   type = mesh.m_vertexType;
+	type = mesh.m_vertexType;
 
 	vertexStride = mesh.m_vertexStride;
 
 	numfaces = mesh.m_numTriangles;
 
-	(*indexbase) = (unsigned char *)mesh.m_triangleIndexBase;
+	(*indexbase) = (unsigned char*)mesh.m_triangleIndexBase;
 	indexstride = mesh.m_triangleIndexStride;
 	indicestype = mesh.m_indexType;
 }
 
-void	btTriangleIndexVertexArray::getLockedReadOnlyVertexIndexBase(const unsigned char **vertexbase, int& numverts,PHY_ScalarType& type, int& vertexStride,const unsigned char **indexbase,int & indexstride,int& numfaces,PHY_ScalarType& indicestype,int subpart) const
+void btTriangleIndexVertexArray::getLockedReadOnlyVertexIndexBase(const unsigned char** vertexbase, int& numverts, PHY_ScalarType& type, int& vertexStride, const unsigned char** indexbase, int& indexstride, int& numfaces, PHY_ScalarType& indicestype, int subpart) const
 {
 	const btIndexedMesh& mesh = m_indexedMeshes[subpart];
 
 	numverts = mesh.m_numVertices;
-	(*vertexbase) = (const unsigned char *)mesh.m_vertexBase;
+	(*vertexbase) = (const unsigned char*)mesh.m_vertexBase;
+
+	type = mesh.m_vertexType;
 
-   type = mesh.m_vertexType;
-   
 	vertexStride = mesh.m_vertexStride;
 
 	numfaces = mesh.m_numTriangles;
-	(*indexbase) = (const unsigned char *)mesh.m_triangleIndexBase;
+	(*indexbase) = (const unsigned char*)mesh.m_triangleIndexBase;
 	indexstride = mesh.m_triangleIndexStride;
 	indicestype = mesh.m_indexType;
 }
 
-bool	btTriangleIndexVertexArray::hasPremadeAabb() const
+bool btTriangleIndexVertexArray::hasPremadeAabb() const
 {
 	return (m_hasAabb == 1);
 }
 
-
-void	btTriangleIndexVertexArray::setPremadeAabb(const btVector3& aabbMin, const btVector3& aabbMax ) const
+void btTriangleIndexVertexArray::setPremadeAabb(const btVector3& aabbMin, const btVector3& aabbMax) const
 {
 	m_aabbMin = aabbMin;
 	m_aabbMax = aabbMax;
-	m_hasAabb = 1; // this is intentionally an int see notes in header
+	m_hasAabb = 1;  // this is intentionally an int see notes in header
 }
 
-void	btTriangleIndexVertexArray::getPremadeAabb(btVector3* aabbMin, btVector3* aabbMax ) const
+void btTriangleIndexVertexArray::getPremadeAabb(btVector3* aabbMin, btVector3* aabbMax) const
 {
 	*aabbMin = m_aabbMin;
 	*aabbMax = m_aabbMax;
 }
-
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleIndexVertexArray.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleIndexVertexArray.h
index b7a6f74361f..8ebb22baae7 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleIndexVertexArray.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleIndexVertexArray.h
@@ -20,62 +20,59 @@ subject to the following restrictions:
 #include "LinearMath/btAlignedObjectArray.h"
 #include "LinearMath/btScalar.h"
 
-
 ///The btIndexedMesh indexes a single vertex and index array. Multiple btIndexedMesh objects can be passed into a btTriangleIndexVertexArray using addIndexedMesh.
 ///Instead of the number of indices, we pass the number of triangles.
-ATTRIBUTE_ALIGNED16( struct)	btIndexedMesh
+ATTRIBUTE_ALIGNED16(struct)
+btIndexedMesh
 {
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-   int                     m_numTriangles;
-   const unsigned char *   m_triangleIndexBase;
-   // Size in byte of the indices for one triangle (3*sizeof(index_type) if the indices are tightly packed)
-   int                     m_triangleIndexStride;
-   int                     m_numVertices;
-   const unsigned char *   m_vertexBase;
-   // Size of a vertex, in bytes
-   int                     m_vertexStride;
+	int m_numTriangles;
+	const unsigned char* m_triangleIndexBase;
+	// Size in byte of the indices for one triangle (3*sizeof(index_type) if the indices are tightly packed)
+	int m_triangleIndexStride;
+	int m_numVertices;
+	const unsigned char* m_vertexBase;
+	// Size of a vertex, in bytes
+	int m_vertexStride;
 
-   // The index type is set when adding an indexed mesh to the
-   // btTriangleIndexVertexArray, do not set it manually
-   PHY_ScalarType m_indexType;
+	// The index type is set when adding an indexed mesh to the
+	// btTriangleIndexVertexArray, do not set it manually
+	PHY_ScalarType m_indexType;
 
-   // The vertex type has a default type similar to Bullet's precision mode (float or double)
-   // but can be set manually if you for example run Bullet with double precision but have
-   // mesh data in single precision..
-   PHY_ScalarType m_vertexType;
+	// The vertex type has a default type similar to Bullet's precision mode (float or double)
+	// but can be set manually if you for example run Bullet with double precision but have
+	// mesh data in single precision..
+	PHY_ScalarType m_vertexType;
 
-
-   btIndexedMesh()
-	   :m_indexType(PHY_INTEGER),
+	btIndexedMesh()
+		: m_indexType(PHY_INTEGER),
 #ifdef BT_USE_DOUBLE_PRECISION
-      m_vertexType(PHY_DOUBLE)
-#else // BT_USE_DOUBLE_PRECISION
-      m_vertexType(PHY_FLOAT)
-#endif // BT_USE_DOUBLE_PRECISION
-      {
-      }
-}
-;
+		  m_vertexType(PHY_DOUBLE)
+#else   // BT_USE_DOUBLE_PRECISION
+		  m_vertexType(PHY_FLOAT)
+#endif  // BT_USE_DOUBLE_PRECISION
+	{
+	}
+};
 
-
-typedef btAlignedObjectArray<btIndexedMesh>	IndexedMeshArray;
+typedef btAlignedObjectArray<btIndexedMesh> IndexedMeshArray;
 
 ///The btTriangleIndexVertexArray allows to access multiple triangle meshes, by indexing into existing triangle/index arrays.
 ///Additional meshes can be added using addIndexedMesh
 ///No duplicate is made of the vertex/index data, it only indexes into external vertex/index arrays.
 ///So keep those arrays around during the lifetime of this btTriangleIndexVertexArray.
-ATTRIBUTE_ALIGNED16( class) btTriangleIndexVertexArray : public btStridingMeshInterface
+ATTRIBUTE_ALIGNED16(class)
+btTriangleIndexVertexArray : public btStridingMeshInterface
 {
 protected:
-	IndexedMeshArray	m_indexedMeshes;
+	IndexedMeshArray m_indexedMeshes;
 	int m_pad[2];
-	mutable int m_hasAabb; // using int instead of bool to maintain alignment
+	mutable int m_hasAabb;  // using int instead of bool to maintain alignment
 	mutable btVector3 m_aabbMin;
 	mutable btVector3 m_aabbMax;
 
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
 	btTriangleIndexVertexArray() : m_hasAabb(0)
@@ -85,49 +82,47 @@ public:
 	virtual ~btTriangleIndexVertexArray();
 
 	//just to be backwards compatible
-	btTriangleIndexVertexArray(int numTriangles,int* triangleIndexBase,int triangleIndexStride,int numVertices,btScalar* vertexBase,int vertexStride);
-	
-	void	addIndexedMesh(const btIndexedMesh& mesh, PHY_ScalarType indexType = PHY_INTEGER)
+	btTriangleIndexVertexArray(int numTriangles, int* triangleIndexBase, int triangleIndexStride, int numVertices, btScalar* vertexBase, int vertexStride);
+
+	void addIndexedMesh(const btIndexedMesh& mesh, PHY_ScalarType indexType = PHY_INTEGER)
 	{
 		m_indexedMeshes.push_back(mesh);
-		m_indexedMeshes[m_indexedMeshes.size()-1].m_indexType = indexType;
+		m_indexedMeshes[m_indexedMeshes.size() - 1].m_indexType = indexType;
 	}
-	
-	
-	virtual void	getLockedVertexIndexBase(unsigned char **vertexbase, int& numverts,PHY_ScalarType& type, int& vertexStride,unsigned char **indexbase,int & indexstride,int& numfaces,PHY_ScalarType& indicestype,int subpart=0);
 
-	virtual void	getLockedReadOnlyVertexIndexBase(const unsigned char **vertexbase, int& numverts,PHY_ScalarType& type, int& vertexStride,const unsigned char **indexbase,int & indexstride,int& numfaces,PHY_ScalarType& indicestype,int subpart=0) const;
+	virtual void getLockedVertexIndexBase(unsigned char** vertexbase, int& numverts, PHY_ScalarType& type, int& vertexStride, unsigned char** indexbase, int& indexstride, int& numfaces, PHY_ScalarType& indicestype, int subpart = 0);
+
+	virtual void getLockedReadOnlyVertexIndexBase(const unsigned char** vertexbase, int& numverts, PHY_ScalarType& type, int& vertexStride, const unsigned char** indexbase, int& indexstride, int& numfaces, PHY_ScalarType& indicestype, int subpart = 0) const;
 
 	/// unLockVertexBase finishes the access to a subpart of the triangle mesh
 	/// make a call to unLockVertexBase when the read and write access (using getLockedVertexIndexBase) is finished
-	virtual void	unLockVertexBase(int subpart) {(void)subpart;}
+	virtual void unLockVertexBase(int subpart) { (void)subpart; }
 
-	virtual void	unLockReadOnlyVertexBase(int subpart) const {(void)subpart;}
+	virtual void unLockReadOnlyVertexBase(int subpart) const { (void)subpart; }
 
 	/// getNumSubParts returns the number of seperate subparts
 	/// each subpart has a continuous array of vertices and indices
-	virtual int		getNumSubParts() const { 
+	virtual int getNumSubParts() const
+	{
 		return (int)m_indexedMeshes.size();
 	}
 
-	IndexedMeshArray&	getIndexedMeshArray()
+	IndexedMeshArray& getIndexedMeshArray()
 	{
 		return m_indexedMeshes;
 	}
 
-	const IndexedMeshArray&	getIndexedMeshArray() const
+	const IndexedMeshArray& getIndexedMeshArray() const
 	{
 		return m_indexedMeshes;
 	}
 
-	virtual void	preallocateVertices(int numverts){(void) numverts;}
-	virtual void	preallocateIndices(int numindices){(void) numindices;}
+	virtual void preallocateVertices(int numverts) { (void)numverts; }
+	virtual void preallocateIndices(int numindices) { (void)numindices; }
 
-	virtual bool	hasPremadeAabb() const;
-	virtual void	setPremadeAabb(const btVector3& aabbMin, const btVector3& aabbMax ) const;
-	virtual void	getPremadeAabb(btVector3* aabbMin, btVector3* aabbMax ) const;
+	virtual bool hasPremadeAabb() const;
+	virtual void setPremadeAabb(const btVector3& aabbMin, const btVector3& aabbMax) const;
+	virtual void getPremadeAabb(btVector3 * aabbMin, btVector3 * aabbMax) const;
+};
 
-}
-;
-
-#endif //BT_TRIANGLE_INDEX_VERTEX_ARRAY_H
+#endif  //BT_TRIANGLE_INDEX_VERTEX_ARRAY_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleIndexVertexMaterialArray.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleIndexVertexMaterialArray.cpp
index dc562941ad6..4bf133d7ac7 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleIndexVertexMaterialArray.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleIndexVertexMaterialArray.cpp
@@ -17,70 +17,68 @@ subject to the following restrictions:
 
 #include "btTriangleIndexVertexMaterialArray.h"
 
-btTriangleIndexVertexMaterialArray::btTriangleIndexVertexMaterialArray(int numTriangles,int* triangleIndexBase,int triangleIndexStride,
-                                   int numVertices,btScalar* vertexBase,int vertexStride,
-                                   int numMaterials, unsigned char* materialBase, int materialStride,
-                                   int* triangleMaterialsBase, int materialIndexStride) :
-btTriangleIndexVertexArray(numTriangles, triangleIndexBase, triangleIndexStride, numVertices, vertexBase, vertexStride)
+btTriangleIndexVertexMaterialArray::btTriangleIndexVertexMaterialArray(int numTriangles, int* triangleIndexBase, int triangleIndexStride,
+																	   int numVertices, btScalar* vertexBase, int vertexStride,
+																	   int numMaterials, unsigned char* materialBase, int materialStride,
+																	   int* triangleMaterialsBase, int materialIndexStride) : btTriangleIndexVertexArray(numTriangles, triangleIndexBase, triangleIndexStride, numVertices, vertexBase, vertexStride)
 {
-    btMaterialProperties mat;
+	btMaterialProperties mat;
 
-    mat.m_numMaterials = numMaterials;
-    mat.m_materialBase = materialBase;
-    mat.m_materialStride = materialStride;
+	mat.m_numMaterials = numMaterials;
+	mat.m_materialBase = materialBase;
+	mat.m_materialStride = materialStride;
 #ifdef BT_USE_DOUBLE_PRECISION
-    mat.m_materialType = PHY_DOUBLE;
+	mat.m_materialType = PHY_DOUBLE;
 #else
-    mat.m_materialType = PHY_FLOAT;
+	mat.m_materialType = PHY_FLOAT;
 #endif
 
-    mat.m_numTriangles = numTriangles;
-    mat.m_triangleMaterialsBase = (unsigned char *)triangleMaterialsBase;
-    mat.m_triangleMaterialStride = materialIndexStride;
-    mat.m_triangleType = PHY_INTEGER;
+	mat.m_numTriangles = numTriangles;
+	mat.m_triangleMaterialsBase = (unsigned char*)triangleMaterialsBase;
+	mat.m_triangleMaterialStride = materialIndexStride;
+	mat.m_triangleType = PHY_INTEGER;
 
-    addMaterialProperties(mat);
+	addMaterialProperties(mat);
 }
 
-
-void btTriangleIndexVertexMaterialArray::getLockedMaterialBase(unsigned char **materialBase, int& numMaterials, PHY_ScalarType& materialType, int& materialStride,
-                                   unsigned char ** triangleMaterialBase, int& numTriangles, int& triangleMaterialStride, PHY_ScalarType& triangleType, int subpart)
+void btTriangleIndexVertexMaterialArray::getLockedMaterialBase(unsigned char** materialBase, int& numMaterials, PHY_ScalarType& materialType, int& materialStride,
+															   unsigned char** triangleMaterialBase, int& numTriangles, int& triangleMaterialStride, PHY_ScalarType& triangleType, int subpart)
 {
-    btAssert(subpart< getNumSubParts() );
+	btAssert(subpart < getNumSubParts());
 
-    btMaterialProperties& mats = m_materials[subpart];
+	btMaterialProperties& mats = m_materials[subpart];
 
-    numMaterials = mats.m_numMaterials;
-    (*materialBase) = (unsigned char *) mats.m_materialBase;
+	numMaterials = mats.m_numMaterials;
+	(*materialBase) = (unsigned char*)mats.m_materialBase;
 #ifdef BT_USE_DOUBLE_PRECISION
-    materialType = PHY_DOUBLE;
+	materialType = PHY_DOUBLE;
 #else
-    materialType = PHY_FLOAT;
+	materialType = PHY_FLOAT;
 #endif
-    materialStride = mats.m_materialStride;
+	materialStride = mats.m_materialStride;
 
-    numTriangles = mats.m_numTriangles;
-    (*triangleMaterialBase) = (unsigned char *)mats.m_triangleMaterialsBase;
-    triangleMaterialStride = mats.m_triangleMaterialStride;
-    triangleType = mats.m_triangleType;
+	numTriangles = mats.m_numTriangles;
+	(*triangleMaterialBase) = (unsigned char*)mats.m_triangleMaterialsBase;
+	triangleMaterialStride = mats.m_triangleMaterialStride;
+	triangleType = mats.m_triangleType;
 }
 
-void btTriangleIndexVertexMaterialArray::getLockedReadOnlyMaterialBase(const unsigned char **materialBase, int& numMaterials, PHY_ScalarType& materialType, int& materialStride,
-                                           const unsigned char ** triangleMaterialBase, int& numTriangles, int& triangleMaterialStride, PHY_ScalarType& triangleType, int subpart)
+void btTriangleIndexVertexMaterialArray::getLockedReadOnlyMaterialBase(const unsigned char** materialBase, int& numMaterials, PHY_ScalarType& materialType, int& materialStride,
+																	   const unsigned char** triangleMaterialBase, int& numTriangles, int& triangleMaterialStride, PHY_ScalarType& triangleType, int subpart)
 {
-    btMaterialProperties& mats = m_materials[subpart];
+	btMaterialProperties& mats = m_materials[subpart];
 
-    numMaterials = mats.m_numMaterials;
-    (*materialBase) = (const unsigned char *) mats.m_materialBase;
+	numMaterials = mats.m_numMaterials;
+	(*materialBase) = (const unsigned char*)mats.m_materialBase;
 #ifdef BT_USE_DOUBLE_PRECISION
-    materialType = PHY_DOUBLE;
+	materialType = PHY_DOUBLE;
 #else
-    materialType = PHY_FLOAT;
+	materialType = PHY_FLOAT;
 #endif
-    materialStride = mats.m_materialStride;
+	materialStride = mats.m_materialStride;
 
-    numTriangles = mats.m_numTriangles;
-    (*triangleMaterialBase) = (const unsigned char *)mats.m_triangleMaterialsBase;
-    triangleMaterialStride = mats.m_triangleMaterialStride;
-    triangleType = mats.m_triangleType;
+	numTriangles = mats.m_numTriangles;
+	(*triangleMaterialBase) = (const unsigned char*)mats.m_triangleMaterialsBase;
+	triangleMaterialStride = mats.m_triangleMaterialStride;
+	triangleType = mats.m_triangleType;
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleIndexVertexMaterialArray.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleIndexVertexMaterialArray.h
index ba4f7b46076..315b1e21f35 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleIndexVertexMaterialArray.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleIndexVertexMaterialArray.h
@@ -20,26 +20,26 @@ subject to the following restrictions:
 
 #include "btTriangleIndexVertexArray.h"
 
-
-ATTRIBUTE_ALIGNED16( struct)	btMaterialProperties
+ATTRIBUTE_ALIGNED16(struct)
+btMaterialProperties
 {
-    ///m_materialBase ==========> 2 btScalar values make up one material, friction then restitution
-    int m_numMaterials;
-    const unsigned char * m_materialBase;
-    int m_materialStride;
-    PHY_ScalarType m_materialType;
-    ///m_numTriangles <=========== This exists in the btIndexedMesh object for the same subpart, but since we're
-    ///                           padding the structure, it can be reproduced at no real cost
-    ///m_triangleMaterials =====> 1 integer value makes up one entry
-    ///                           eg: m_triangleMaterials[1] = 5; // This will set triangle 2 to use material 5
-    int m_numTriangles; 
-    const unsigned char * m_triangleMaterialsBase;
-    int m_triangleMaterialStride;
-    ///m_triangleType <========== Automatically set in addMaterialProperties
-    PHY_ScalarType m_triangleType;
+	///m_materialBase ==========> 2 btScalar values make up one material, friction then restitution
+	int m_numMaterials;
+	const unsigned char* m_materialBase;
+	int m_materialStride;
+	PHY_ScalarType m_materialType;
+	///m_numTriangles <=========== This exists in the btIndexedMesh object for the same subpart, but since we're
+	///                           padding the structure, it can be reproduced at no real cost
+	///m_triangleMaterials =====> 1 integer value makes up one entry
+	///                           eg: m_triangleMaterials[1] = 5; // This will set triangle 2 to use material 5
+	int m_numTriangles;
+	const unsigned char* m_triangleMaterialsBase;
+	int m_triangleMaterialStride;
+	///m_triangleType <========== Automatically set in addMaterialProperties
+	PHY_ScalarType m_triangleType;
 };
 
-typedef btAlignedObjectArray<btMaterialProperties>	MaterialArray;
+typedef btAlignedObjectArray<btMaterialProperties> MaterialArray;
 
 ///Teh btTriangleIndexVertexMaterialArray is built on TriangleIndexVertexArray
 ///The addition of a material array allows for the utilization of the partID and
@@ -47,38 +47,37 @@ typedef btAlignedObjectArray<btMaterialProperties>	MaterialArray;
 ///TriangleIndexVertexArray, no duplicate is made of the material data, so it
 ///is the users responsibility to maintain the array during the lifetime of the
 ///TriangleIndexVertexMaterialArray.
-ATTRIBUTE_ALIGNED16(class) btTriangleIndexVertexMaterialArray : public btTriangleIndexVertexArray
+ATTRIBUTE_ALIGNED16(class)
+btTriangleIndexVertexMaterialArray : public btTriangleIndexVertexArray
 {
 protected:
-    MaterialArray       m_materials;
-		
+	MaterialArray m_materials;
+
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-    btTriangleIndexVertexMaterialArray()
+	btTriangleIndexVertexMaterialArray()
 	{
 	}
 
-    btTriangleIndexVertexMaterialArray(int numTriangles,int* triangleIndexBase,int triangleIndexStride,
-        int numVertices,btScalar* vertexBase,int vertexStride,
-        int numMaterials, unsigned char* materialBase, int materialStride,
-        int* triangleMaterialsBase, int materialIndexStride);
+	btTriangleIndexVertexMaterialArray(int numTriangles, int* triangleIndexBase, int triangleIndexStride,
+									   int numVertices, btScalar* vertexBase, int vertexStride,
+									   int numMaterials, unsigned char* materialBase, int materialStride,
+									   int* triangleMaterialsBase, int materialIndexStride);
 
-    virtual ~btTriangleIndexVertexMaterialArray() {}
+	virtual ~btTriangleIndexVertexMaterialArray() {}
 
-    void	addMaterialProperties(const btMaterialProperties& mat, PHY_ScalarType triangleType = PHY_INTEGER)
-    {
-        m_materials.push_back(mat);
-        m_materials[m_materials.size()-1].m_triangleType = triangleType;
-    }
+	void addMaterialProperties(const btMaterialProperties& mat, PHY_ScalarType triangleType = PHY_INTEGER)
+	{
+		m_materials.push_back(mat);
+		m_materials[m_materials.size() - 1].m_triangleType = triangleType;
+	}
 
-    virtual void getLockedMaterialBase(unsigned char **materialBase, int& numMaterials, PHY_ScalarType& materialType, int& materialStride,
-        unsigned char ** triangleMaterialBase, int& numTriangles, int& triangleMaterialStride, PHY_ScalarType& triangleType ,int subpart = 0);
+	virtual void getLockedMaterialBase(unsigned char** materialBase, int& numMaterials, PHY_ScalarType& materialType, int& materialStride,
+									   unsigned char** triangleMaterialBase, int& numTriangles, int& triangleMaterialStride, PHY_ScalarType& triangleType, int subpart = 0);
 
-    virtual void getLockedReadOnlyMaterialBase(const unsigned char **materialBase, int& numMaterials, PHY_ScalarType& materialType, int& materialStride,
-        const unsigned char ** triangleMaterialBase, int& numTriangles, int& triangleMaterialStride, PHY_ScalarType& triangleType, int subpart = 0);
+	virtual void getLockedReadOnlyMaterialBase(const unsigned char** materialBase, int& numMaterials, PHY_ScalarType& materialType, int& materialStride,
+											   const unsigned char** triangleMaterialBase, int& numTriangles, int& triangleMaterialStride, PHY_ScalarType& triangleType, int subpart = 0);
+};
 
-}
-;
-
-#endif //BT_MULTIMATERIAL_TRIANGLE_INDEX_VERTEX_ARRAY_H
+#endif  //BT_MULTIMATERIAL_TRIANGLE_INDEX_VERTEX_ARRAY_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleInfoMap.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleInfoMap.h
index 64275895901..8ee35ef5faf 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleInfoMap.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleInfoMap.h
@@ -16,11 +16,9 @@ subject to the following restrictions:
 #ifndef _BT_TRIANGLE_INFO_MAP_H
 #define _BT_TRIANGLE_INFO_MAP_H
 
-
 #include "LinearMath/btHashMap.h"
 #include "LinearMath/btSerializer.h"
 
-
 ///for btTriangleInfo m_flags
 #define TRI_INFO_V0V1_CONVEX 1
 #define TRI_INFO_V1V2_CONVEX 2
@@ -30,61 +28,58 @@ subject to the following restrictions:
 #define TRI_INFO_V1V2_SWAP_NORMALB 16
 #define TRI_INFO_V2V0_SWAP_NORMALB 32
 
-
 ///The btTriangleInfo structure stores information to adjust collision normals to avoid collisions against internal edges
-///it can be generated using 
-struct	btTriangleInfo
+///it can be generated using
+struct btTriangleInfo
 {
 	btTriangleInfo()
 	{
 		m_edgeV0V1Angle = SIMD_2_PI;
 		m_edgeV1V2Angle = SIMD_2_PI;
 		m_edgeV2V0Angle = SIMD_2_PI;
-		m_flags=0;
+		m_flags = 0;
 	}
 
-	int			m_flags;
-
-	btScalar	m_edgeV0V1Angle;
-	btScalar	m_edgeV1V2Angle;
-	btScalar	m_edgeV2V0Angle;
+	int m_flags;
 
+	btScalar m_edgeV0V1Angle;
+	btScalar m_edgeV1V2Angle;
+	btScalar m_edgeV2V0Angle;
 };
 
-typedef btHashMap<btHashInt,btTriangleInfo> btInternalTriangleInfoMap;
-
+typedef btHashMap<btHashInt, btTriangleInfo> btInternalTriangleInfoMap;
 
 ///The btTriangleInfoMap stores edge angle information for some triangles. You can compute this information yourself or using btGenerateInternalEdgeInfo.
-struct	btTriangleInfoMap : public btInternalTriangleInfoMap
+struct btTriangleInfoMap : public btInternalTriangleInfoMap
 {
-	btScalar	m_convexEpsilon;///used to determine if an edge or contact normal is convex, using the dot product
-	btScalar	m_planarEpsilon; ///used to determine if a triangle edge is planar with zero angle
-	btScalar	m_equalVertexThreshold; ///used to compute connectivity: if the distance between two vertices is smaller than m_equalVertexThreshold, they are considered to be 'shared'
-	btScalar	m_edgeDistanceThreshold; ///used to determine edge contacts: if the closest distance between a contact point and an edge is smaller than this distance threshold it is considered to "hit the edge"
-	btScalar	m_maxEdgeAngleThreshold; //ignore edges that connect triangles at an angle larger than this m_maxEdgeAngleThreshold
-	btScalar	m_zeroAreaThreshold; ///used to determine if a triangle is degenerate (length squared of cross product of 2 triangle edges < threshold)
-	
-	
+	btScalar m_convexEpsilon;          ///used to determine if an edge or contact normal is convex, using the dot product
+	btScalar m_planarEpsilon;          ///used to determine if a triangle edge is planar with zero angle
+	btScalar m_equalVertexThreshold;   ///used to compute connectivity: if the distance between two vertices is smaller than m_equalVertexThreshold, they are considered to be 'shared'
+	btScalar m_edgeDistanceThreshold;  ///used to determine edge contacts: if the closest distance between a contact point and an edge is smaller than this distance threshold it is considered to "hit the edge"
+	btScalar m_maxEdgeAngleThreshold;  //ignore edges that connect triangles at an angle larger than this m_maxEdgeAngleThreshold
+	btScalar m_zeroAreaThreshold;      ///used to determine if a triangle is degenerate (length squared of cross product of 2 triangle edges < threshold)
+
 	btTriangleInfoMap()
 	{
 		m_convexEpsilon = 0.00f;
 		m_planarEpsilon = 0.0001f;
-		m_equalVertexThreshold = btScalar(0.0001)*btScalar(0.0001);
+		m_equalVertexThreshold = btScalar(0.0001) * btScalar(0.0001);
 		m_edgeDistanceThreshold = btScalar(0.1);
-		m_zeroAreaThreshold = btScalar(0.0001)*btScalar(0.0001);
+		m_zeroAreaThreshold = btScalar(0.0001) * btScalar(0.0001);
 		m_maxEdgeAngleThreshold = SIMD_2_PI;
 	}
 	virtual ~btTriangleInfoMap() {}
 
-	virtual	int	calculateSerializeBufferSize() const;
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
-
-	void	deSerialize(struct btTriangleInfoMapData& data);
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 
+	void deSerialize(struct btTriangleInfoMapData& data);
 };
 
+// clang-format off
+
 ///those fields have to be float and not btScalar for the serialization to work properly
 struct	btTriangleInfoData
 {
@@ -114,86 +109,86 @@ struct	btTriangleInfoMapData
 	char	m_padding[4];
 };
 
-SIMD_FORCE_INLINE	int	btTriangleInfoMap::calculateSerializeBufferSize() const
+// clang-format on
+
+SIMD_FORCE_INLINE int btTriangleInfoMap::calculateSerializeBufferSize() const
 {
 	return sizeof(btTriangleInfoMapData);
 }
 
 ///fills the dataBuffer and returns the struct name (and 0 on failure)
-SIMD_FORCE_INLINE	const char*	btTriangleInfoMap::serialize(void* dataBuffer, btSerializer* serializer) const
+SIMD_FORCE_INLINE const char* btTriangleInfoMap::serialize(void* dataBuffer, btSerializer* serializer) const
 {
-	btTriangleInfoMapData* tmapData = (btTriangleInfoMapData*) dataBuffer;
+	btTriangleInfoMapData* tmapData = (btTriangleInfoMapData*)dataBuffer;
 	tmapData->m_convexEpsilon = (float)m_convexEpsilon;
 	tmapData->m_planarEpsilon = (float)m_planarEpsilon;
-	tmapData->m_equalVertexThreshold =(float) m_equalVertexThreshold;
+	tmapData->m_equalVertexThreshold = (float)m_equalVertexThreshold;
 	tmapData->m_edgeDistanceThreshold = (float)m_edgeDistanceThreshold;
 	tmapData->m_zeroAreaThreshold = (float)m_zeroAreaThreshold;
-	
+
 	tmapData->m_hashTableSize = m_hashTable.size();
 
 	tmapData->m_hashTablePtr = tmapData->m_hashTableSize ? (int*)serializer->getUniquePointer((void*)&m_hashTable[0]) : 0;
 	if (tmapData->m_hashTablePtr)
-	{ 
+	{
 		//serialize an int buffer
 		int sz = sizeof(int);
 		int numElem = tmapData->m_hashTableSize;
-		btChunk* chunk = serializer->allocate(sz,numElem);
+		btChunk* chunk = serializer->allocate(sz, numElem);
 		int* memPtr = (int*)chunk->m_oldPtr;
-		for (int i=0;i<numElem;i++,memPtr++)
+		for (int i = 0; i < numElem; i++, memPtr++)
 		{
 			*memPtr = m_hashTable[i];
 		}
-		serializer->finalizeChunk(chunk,"int",BT_ARRAY_CODE,(void*)&m_hashTable[0]);
-
+		serializer->finalizeChunk(chunk, "int", BT_ARRAY_CODE, (void*)&m_hashTable[0]);
 	}
 
 	tmapData->m_nextSize = m_next.size();
-	tmapData->m_nextPtr = tmapData->m_nextSize? (int*)serializer->getUniquePointer((void*)&m_next[0]): 0;
+	tmapData->m_nextPtr = tmapData->m_nextSize ? (int*)serializer->getUniquePointer((void*)&m_next[0]) : 0;
 	if (tmapData->m_nextPtr)
 	{
 		int sz = sizeof(int);
 		int numElem = tmapData->m_nextSize;
-		btChunk* chunk = serializer->allocate(sz,numElem);
+		btChunk* chunk = serializer->allocate(sz, numElem);
 		int* memPtr = (int*)chunk->m_oldPtr;
-		for (int i=0;i<numElem;i++,memPtr++)
+		for (int i = 0; i < numElem; i++, memPtr++)
 		{
 			*memPtr = m_next[i];
 		}
-		serializer->finalizeChunk(chunk,"int",BT_ARRAY_CODE,(void*)&m_next[0]);
+		serializer->finalizeChunk(chunk, "int", BT_ARRAY_CODE, (void*)&m_next[0]);
 	}
-	
+
 	tmapData->m_numValues = m_valueArray.size();
-	tmapData->m_valueArrayPtr = tmapData->m_numValues ? (btTriangleInfoData*)serializer->getUniquePointer((void*)&m_valueArray[0]): 0;
+	tmapData->m_valueArrayPtr = tmapData->m_numValues ? (btTriangleInfoData*)serializer->getUniquePointer((void*)&m_valueArray[0]) : 0;
 	if (tmapData->m_valueArrayPtr)
 	{
 		int sz = sizeof(btTriangleInfoData);
 		int numElem = tmapData->m_numValues;
-		btChunk* chunk = serializer->allocate(sz,numElem);
+		btChunk* chunk = serializer->allocate(sz, numElem);
 		btTriangleInfoData* memPtr = (btTriangleInfoData*)chunk->m_oldPtr;
-		for (int i=0;i<numElem;i++,memPtr++)
+		for (int i = 0; i < numElem; i++, memPtr++)
 		{
 			memPtr->m_edgeV0V1Angle = (float)m_valueArray[i].m_edgeV0V1Angle;
 			memPtr->m_edgeV1V2Angle = (float)m_valueArray[i].m_edgeV1V2Angle;
 			memPtr->m_edgeV2V0Angle = (float)m_valueArray[i].m_edgeV2V0Angle;
 			memPtr->m_flags = m_valueArray[i].m_flags;
 		}
-		serializer->finalizeChunk(chunk,"btTriangleInfoData",BT_ARRAY_CODE,(void*) &m_valueArray[0]);
+		serializer->finalizeChunk(chunk, "btTriangleInfoData", BT_ARRAY_CODE, (void*)&m_valueArray[0]);
 	}
-	
+
 	tmapData->m_numKeys = m_keyArray.size();
 	tmapData->m_keyArrayPtr = tmapData->m_numKeys ? (int*)serializer->getUniquePointer((void*)&m_keyArray[0]) : 0;
 	if (tmapData->m_keyArrayPtr)
 	{
 		int sz = sizeof(int);
 		int numElem = tmapData->m_numValues;
-		btChunk* chunk = serializer->allocate(sz,numElem);
+		btChunk* chunk = serializer->allocate(sz, numElem);
 		int* memPtr = (int*)chunk->m_oldPtr;
-		for (int i=0;i<numElem;i++,memPtr++)
+		for (int i = 0; i < numElem; i++, memPtr++)
 		{
 			*memPtr = m_keyArray[i].getUid1();
 		}
-		serializer->finalizeChunk(chunk,"int",BT_ARRAY_CODE,(void*) &m_keyArray[0]);
-
+		serializer->finalizeChunk(chunk, "int", BT_ARRAY_CODE, (void*)&m_keyArray[0]);
 	}
 
 	// Fill padding with zeros to appease msan.
@@ -205,44 +200,39 @@ SIMD_FORCE_INLINE	const char*	btTriangleInfoMap::serialize(void* dataBuffer, btS
 	return "btTriangleInfoMapData";
 }
 
-
-
 ///fills the dataBuffer and returns the struct name (and 0 on failure)
-SIMD_FORCE_INLINE	void	btTriangleInfoMap::deSerialize(btTriangleInfoMapData& tmapData )
+SIMD_FORCE_INLINE void btTriangleInfoMap::deSerialize(btTriangleInfoMapData& tmapData)
 {
-
-
 	m_convexEpsilon = tmapData.m_convexEpsilon;
 	m_planarEpsilon = tmapData.m_planarEpsilon;
 	m_equalVertexThreshold = tmapData.m_equalVertexThreshold;
 	m_edgeDistanceThreshold = tmapData.m_edgeDistanceThreshold;
 	m_zeroAreaThreshold = tmapData.m_zeroAreaThreshold;
 	m_hashTable.resize(tmapData.m_hashTableSize);
-	int i =0;
-	for (i=0;i<tmapData.m_hashTableSize;i++)
+	int i = 0;
+	for (i = 0; i < tmapData.m_hashTableSize; i++)
 	{
 		m_hashTable[i] = tmapData.m_hashTablePtr[i];
 	}
 	m_next.resize(tmapData.m_nextSize);
-	for (i=0;i<tmapData.m_nextSize;i++)
+	for (i = 0; i < tmapData.m_nextSize; i++)
 	{
 		m_next[i] = tmapData.m_nextPtr[i];
 	}
 	m_valueArray.resize(tmapData.m_numValues);
-	for (i=0;i<tmapData.m_numValues;i++)
+	for (i = 0; i < tmapData.m_numValues; i++)
 	{
 		m_valueArray[i].m_edgeV0V1Angle = tmapData.m_valueArrayPtr[i].m_edgeV0V1Angle;
 		m_valueArray[i].m_edgeV1V2Angle = tmapData.m_valueArrayPtr[i].m_edgeV1V2Angle;
 		m_valueArray[i].m_edgeV2V0Angle = tmapData.m_valueArrayPtr[i].m_edgeV2V0Angle;
 		m_valueArray[i].m_flags = tmapData.m_valueArrayPtr[i].m_flags;
 	}
-	
-	m_keyArray.resize(tmapData.m_numKeys,btHashInt(0));
-	for (i=0;i<tmapData.m_numKeys;i++)
+
+	m_keyArray.resize(tmapData.m_numKeys, btHashInt(0));
+	for (i = 0; i < tmapData.m_numKeys; i++)
 	{
 		m_keyArray[i].setUid1(tmapData.m_keyArrayPtr[i]);
 	}
 }
 
-
-#endif //_BT_TRIANGLE_INFO_MAP_H
+#endif  //_BT_TRIANGLE_INFO_MAP_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleMesh.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleMesh.cpp
index e4de7320930..abd8c227866 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleMesh.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleMesh.cpp
@@ -13,38 +13,36 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btTriangleMesh.h"
 
-
-
-btTriangleMesh::btTriangleMesh (bool use32bitIndices,bool use4componentVertices)
-:m_use32bitIndices(use32bitIndices),
-m_use4componentVertices(use4componentVertices),
-m_weldingThreshold(0.0)
+btTriangleMesh::btTriangleMesh(bool use32bitIndices, bool use4componentVertices)
+	: m_use32bitIndices(use32bitIndices),
+	  m_use4componentVertices(use4componentVertices),
+	  m_weldingThreshold(0.0)
 {
 	btIndexedMesh meshIndex;
 	meshIndex.m_numTriangles = 0;
 	meshIndex.m_numVertices = 0;
 	meshIndex.m_indexType = PHY_INTEGER;
 	meshIndex.m_triangleIndexBase = 0;
-	meshIndex.m_triangleIndexStride = 3*sizeof(int);
+	meshIndex.m_triangleIndexStride = 3 * sizeof(int);
 	meshIndex.m_vertexBase = 0;
 	meshIndex.m_vertexStride = sizeof(btVector3);
 	m_indexedMeshes.push_back(meshIndex);
 
 	if (m_use32bitIndices)
 	{
-		m_indexedMeshes[0].m_numTriangles = m_32bitIndices.size()/3;
+		m_indexedMeshes[0].m_numTriangles = m_32bitIndices.size() / 3;
 		m_indexedMeshes[0].m_triangleIndexBase = 0;
 		m_indexedMeshes[0].m_indexType = PHY_INTEGER;
-		m_indexedMeshes[0].m_triangleIndexStride = 3*sizeof(int);
-	} else
+		m_indexedMeshes[0].m_triangleIndexStride = 3 * sizeof(int);
+	}
+	else
 	{
-		m_indexedMeshes[0].m_numTriangles = m_16bitIndices.size()/3;
+		m_indexedMeshes[0].m_numTriangles = m_16bitIndices.size() / 3;
 		m_indexedMeshes[0].m_triangleIndexBase = 0;
 		m_indexedMeshes[0].m_indexType = PHY_SHORT;
-		m_indexedMeshes[0].m_triangleIndexStride = 3*sizeof(short int);
+		m_indexedMeshes[0].m_triangleIndexStride = 3 * sizeof(short int);
 	}
 
 	if (m_use4componentVertices)
@@ -52,48 +50,48 @@ m_weldingThreshold(0.0)
 		m_indexedMeshes[0].m_numVertices = m_4componentVertices.size();
 		m_indexedMeshes[0].m_vertexBase = 0;
 		m_indexedMeshes[0].m_vertexStride = sizeof(btVector3);
-	} else
-	{
-		m_indexedMeshes[0].m_numVertices = m_3componentVertices.size()/3;
-		m_indexedMeshes[0].m_vertexBase = 0;
-		m_indexedMeshes[0].m_vertexStride = 3*sizeof(btScalar);
 	}
-
-
+	else
+	{
+		m_indexedMeshes[0].m_numVertices = m_3componentVertices.size() / 3;
+		m_indexedMeshes[0].m_vertexBase = 0;
+		m_indexedMeshes[0].m_vertexStride = 3 * sizeof(btScalar);
+	}
 }
 
-void	btTriangleMesh::addIndex(int index)
+void btTriangleMesh::addIndex(int index)
 {
 	if (m_use32bitIndices)
 	{
 		m_32bitIndices.push_back(index);
-		m_indexedMeshes[0].m_triangleIndexBase = (unsigned char*) &m_32bitIndices[0];
-	} else
+		m_indexedMeshes[0].m_triangleIndexBase = (unsigned char*)&m_32bitIndices[0];
+	}
+	else
 	{
 		m_16bitIndices.push_back(index);
-		m_indexedMeshes[0].m_triangleIndexBase = (unsigned char*) &m_16bitIndices[0];
+		m_indexedMeshes[0].m_triangleIndexBase = (unsigned char*)&m_16bitIndices[0];
 	}
 }
 
-void	btTriangleMesh::addTriangleIndices(int index1, int index2, int index3 )
+void btTriangleMesh::addTriangleIndices(int index1, int index2, int index3)
 {
 	m_indexedMeshes[0].m_numTriangles++;
-	addIndex( index1 );
-	addIndex( index2 );
-	addIndex( index3 );
+	addIndex(index1);
+	addIndex(index2);
+	addIndex(index3);
 }
 
-int	btTriangleMesh::findOrAddVertex(const btVector3& vertex, bool removeDuplicateVertices)
+int btTriangleMesh::findOrAddVertex(const btVector3& vertex, bool removeDuplicateVertices)
 {
 	//return index of new/existing vertex
 	///@todo: could use acceleration structure for this
 	if (m_use4componentVertices)
 	{
 		if (removeDuplicateVertices)
+		{
+			for (int i = 0; i < m_4componentVertices.size(); i++)
 			{
-			for (int i=0;i< m_4componentVertices.size();i++)
-			{
-				if ((m_4componentVertices[i]-vertex).length2() <= m_weldingThreshold)
+				if ((m_4componentVertices[i] - vertex).length2() <= m_weldingThreshold)
 				{
 					return i;
 				}
@@ -103,19 +101,18 @@ int	btTriangleMesh::findOrAddVertex(const btVector3& vertex, bool removeDuplicat
 		m_4componentVertices.push_back(vertex);
 		m_indexedMeshes[0].m_vertexBase = (unsigned char*)&m_4componentVertices[0];
 
-		return m_4componentVertices.size()-1;
-		
-	} else
+		return m_4componentVertices.size() - 1;
+	}
+	else
 	{
-		
 		if (removeDuplicateVertices)
 		{
-			for (int i=0;i< m_3componentVertices.size();i+=3)
+			for (int i = 0; i < m_3componentVertices.size(); i += 3)
 			{
-				btVector3 vtx(m_3componentVertices[i],m_3componentVertices[i+1],m_3componentVertices[i+2]);
-				if ((vtx-vertex).length2() <= m_weldingThreshold)
+				btVector3 vtx(m_3componentVertices[i], m_3componentVertices[i + 1], m_3componentVertices[i + 2]);
+				if ((vtx - vertex).length2() <= m_weldingThreshold)
 				{
-					return i/3;
+					return i / 3;
 				}
 			}
 		}
@@ -124,17 +121,16 @@ int	btTriangleMesh::findOrAddVertex(const btVector3& vertex, bool removeDuplicat
 		m_3componentVertices.push_back(vertex.getZ());
 		m_indexedMeshes[0].m_numVertices++;
 		m_indexedMeshes[0].m_vertexBase = (unsigned char*)&m_3componentVertices[0];
-		return (m_3componentVertices.size()/3)-1;
+		return (m_3componentVertices.size() / 3) - 1;
 	}
-
 }
-		
-void	btTriangleMesh::addTriangle(const btVector3& vertex0,const btVector3& vertex1,const btVector3& vertex2,bool removeDuplicateVertices)
+
+void btTriangleMesh::addTriangle(const btVector3& vertex0, const btVector3& vertex1, const btVector3& vertex2, bool removeDuplicateVertices)
 {
 	m_indexedMeshes[0].m_numTriangles++;
-	addIndex(findOrAddVertex(vertex0,removeDuplicateVertices));
-	addIndex(findOrAddVertex(vertex1,removeDuplicateVertices));
-	addIndex(findOrAddVertex(vertex2,removeDuplicateVertices));
+	addIndex(findOrAddVertex(vertex0, removeDuplicateVertices));
+	addIndex(findOrAddVertex(vertex1, removeDuplicateVertices));
+	addIndex(findOrAddVertex(vertex2, removeDuplicateVertices));
 }
 
 int btTriangleMesh::getNumTriangles() const
@@ -151,7 +147,8 @@ void btTriangleMesh::preallocateVertices(int numverts)
 	if (m_use4componentVertices)
 	{
 		m_4componentVertices.reserve(numverts);
-	} else
+	}
+	else
 	{
 		m_3componentVertices.reserve(numverts);
 	}
@@ -162,7 +159,8 @@ void btTriangleMesh::preallocateIndices(int numindices)
 	if (m_use32bitIndices)
 	{
 		m_32bitIndices.reserve(numindices);
-	} else
+	}
+	else
 	{
 		m_16bitIndices.reserve(numindices);
 	}
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleMesh.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleMesh.h
index ac4afa7f6b2..a8a362355cc 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleMesh.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleMesh.h
@@ -26,47 +26,44 @@ subject to the following restrictions:
 ///Performance of btTriangleMesh and btTriangleIndexVertexArray used in a btBvhTriangleMeshShape is the same.
 class btTriangleMesh : public btTriangleIndexVertexArray
 {
-	btAlignedObjectArray<btVector3>	m_4componentVertices;
-	btAlignedObjectArray<btScalar>	m_3componentVertices;
+	btAlignedObjectArray<btVector3> m_4componentVertices;
+	btAlignedObjectArray<btScalar> m_3componentVertices;
 
-	btAlignedObjectArray<unsigned int>		m_32bitIndices;
-	btAlignedObjectArray<unsigned short int>		m_16bitIndices;
-	bool	m_use32bitIndices;
-	bool	m_use4componentVertices;
-	
+	btAlignedObjectArray<unsigned int> m_32bitIndices;
+	btAlignedObjectArray<unsigned short int> m_16bitIndices;
+	bool m_use32bitIndices;
+	bool m_use4componentVertices;
 
-	public:
-		btScalar	m_weldingThreshold;
+public:
+	btScalar m_weldingThreshold;
 
-		btTriangleMesh (bool use32bitIndices=true,bool use4componentVertices=true);
+	btTriangleMesh(bool use32bitIndices = true, bool use4componentVertices = true);
 
-		bool	getUse32bitIndices() const
-		{
-			return m_use32bitIndices;
-		}
+	bool getUse32bitIndices() const
+	{
+		return m_use32bitIndices;
+	}
 
-		bool	getUse4componentVertices() const
-		{
-			return m_use4componentVertices;
-		}
-		///By default addTriangle won't search for duplicate vertices, because the search is very slow for large triangle meshes.
-		///In general it is better to directly use btTriangleIndexVertexArray instead.
-		void	addTriangle(const btVector3& vertex0,const btVector3& vertex1,const btVector3& vertex2, bool removeDuplicateVertices=false);
+	bool getUse4componentVertices() const
+	{
+		return m_use4componentVertices;
+	}
+	///By default addTriangle won't search for duplicate vertices, because the search is very slow for large triangle meshes.
+	///In general it is better to directly use btTriangleIndexVertexArray instead.
+	void addTriangle(const btVector3& vertex0, const btVector3& vertex1, const btVector3& vertex2, bool removeDuplicateVertices = false);
 
-		///Add a triangle using its indices. Make sure the indices are pointing within the vertices array, so add the vertices first (and to be sure, avoid removal of duplicate vertices)	
-		void	addTriangleIndices(int index1, int index2, int index3 );
-	
-		int getNumTriangles() const;
+	///Add a triangle using its indices. Make sure the indices are pointing within the vertices array, so add the vertices first (and to be sure, avoid removal of duplicate vertices)
+	void addTriangleIndices(int index1, int index2, int index3);
 
-		virtual void	preallocateVertices(int numverts);
-		virtual void	preallocateIndices(int numindices);
+	int getNumTriangles() const;
 
-		///findOrAddVertex is an internal method, use addTriangle instead
-		int		findOrAddVertex(const btVector3& vertex, bool removeDuplicateVertices);
-		///addIndex is an internal method, use addTriangle instead
-		void	addIndex(int index);
-		
+	virtual void preallocateVertices(int numverts);
+	virtual void preallocateIndices(int numindices);
+
+	///findOrAddVertex is an internal method, use addTriangle instead
+	int findOrAddVertex(const btVector3& vertex, bool removeDuplicateVertices);
+	///addIndex is an internal method, use addTriangle instead
+	void addIndex(int index);
 };
 
-#endif //BT_TRIANGLE_MESH_H
-
+#endif  //BT_TRIANGLE_MESH_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleMeshShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleMeshShape.cpp
index 0e17951405c..aec239063cd 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleMeshShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleMeshShape.cpp
@@ -20,12 +20,11 @@ subject to the following restrictions:
 #include "LinearMath/btAabbUtil2.h"
 #include "BulletCollision/CollisionShapes/btCollisionMargin.h"
 
-
 btTriangleMeshShape::btTriangleMeshShape(btStridingMeshInterface* meshInterface)
-: btConcaveShape (), m_meshInterface(meshInterface)
+	: btConcaveShape(), m_meshInterface(meshInterface)
 {
 	m_shapeType = TRIANGLE_MESH_SHAPE_PROXYTYPE;
-	if(meshInterface->hasPremadeAabb())
+	if (meshInterface->hasPremadeAabb())
 	{
 		meshInterface->getPremadeAabb(&m_localAabbMin, &m_localAabbMax);
 	}
@@ -35,69 +34,60 @@ btTriangleMeshShape::btTriangleMeshShape(btStridingMeshInterface* meshInterface)
 	}
 }
 
-
 btTriangleMeshShape::~btTriangleMeshShape()
 {
-		
 }
 
-
-
-
-void btTriangleMeshShape::getAabb(const btTransform& trans,btVector3& aabbMin,btVector3& aabbMax) const
+void btTriangleMeshShape::getAabb(const btTransform& trans, btVector3& aabbMin, btVector3& aabbMax) const
 {
+	btVector3 localHalfExtents = btScalar(0.5) * (m_localAabbMax - m_localAabbMin);
+	localHalfExtents += btVector3(getMargin(), getMargin(), getMargin());
+	btVector3 localCenter = btScalar(0.5) * (m_localAabbMax + m_localAabbMin);
 
-	btVector3 localHalfExtents = btScalar(0.5)*(m_localAabbMax-m_localAabbMin);
-	localHalfExtents += btVector3(getMargin(),getMargin(),getMargin());
-	btVector3 localCenter = btScalar(0.5)*(m_localAabbMax+m_localAabbMin);
-	
-	btMatrix3x3 abs_b = trans.getBasis().absolute();  
+	btMatrix3x3 abs_b = trans.getBasis().absolute();
 
 	btVector3 center = trans(localCenter);
 
-    btVector3 extent = localHalfExtents.dot3(abs_b[0], abs_b[1], abs_b[2]);
+	btVector3 extent = localHalfExtents.dot3(abs_b[0], abs_b[1], abs_b[2]);
 	aabbMin = center - extent;
 	aabbMax = center + extent;
 }
 
-void	btTriangleMeshShape::recalcLocalAabb()
+void btTriangleMeshShape::recalcLocalAabb()
 {
-	for (int i=0;i<3;i++)
+	for (int i = 0; i < 3; i++)
 	{
-		btVector3 vec(btScalar(0.),btScalar(0.),btScalar(0.));
+		btVector3 vec(btScalar(0.), btScalar(0.), btScalar(0.));
 		vec[i] = btScalar(1.);
 		btVector3 tmp = localGetSupportingVertex(vec);
-		m_localAabbMax[i] = tmp[i]+m_collisionMargin;
+		m_localAabbMax[i] = tmp[i] + m_collisionMargin;
 		vec[i] = btScalar(-1.);
 		tmp = localGetSupportingVertex(vec);
-		m_localAabbMin[i] = tmp[i]-m_collisionMargin;
+		m_localAabbMin[i] = tmp[i] - m_collisionMargin;
 	}
 }
 
-
-
 class SupportVertexCallback : public btTriangleCallback
 {
-
 	btVector3 m_supportVertexLocal;
-public:
 
-	btTransform	m_worldTrans;
+public:
+	btTransform m_worldTrans;
 	btScalar m_maxDot;
 	btVector3 m_supportVecLocal;
 
-	SupportVertexCallback(const btVector3& supportVecWorld,const btTransform& trans)
-		: m_supportVertexLocal(btScalar(0.),btScalar(0.),btScalar(0.)), m_worldTrans(trans) ,m_maxDot(btScalar(-BT_LARGE_FLOAT))
-		
+	SupportVertexCallback(const btVector3& supportVecWorld, const btTransform& trans)
+		: m_supportVertexLocal(btScalar(0.), btScalar(0.), btScalar(0.)), m_worldTrans(trans), m_maxDot(btScalar(-BT_LARGE_FLOAT))
+
 	{
 		m_supportVecLocal = supportVecWorld * m_worldTrans.getBasis();
 	}
 
-	virtual void processTriangle( btVector3* triangle,int partId, int triangleIndex)
+	virtual void processTriangle(btVector3* triangle, int partId, int triangleIndex)
 	{
 		(void)partId;
 		(void)triangleIndex;
-		for (int i=0;i<3;i++)
+		for (int i = 0; i < 3; i++)
 		{
 			btScalar dot = m_supportVecLocal.dot(triangle[i]);
 			if (dot > m_maxDot)
@@ -113,14 +103,12 @@ public:
 		return m_worldTrans(m_supportVertexLocal);
 	}
 
-	btVector3	GetSupportVertexLocal()
+	btVector3 GetSupportVertexLocal()
 	{
 		return m_supportVertexLocal;
 	}
-
 };
 
-	
 void btTriangleMeshShape::setLocalScaling(const btVector3& scaling)
 {
 	m_meshInterface->setScaling(scaling);
@@ -132,60 +120,46 @@ const btVector3& btTriangleMeshShape::getLocalScaling() const
 	return m_meshInterface->getScaling();
 }
 
-
-
-
-
-
 //#define DEBUG_TRIANGLE_MESH
 
-
-
-void	btTriangleMeshShape::processAllTriangles(btTriangleCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax) const
+void btTriangleMeshShape::processAllTriangles(btTriangleCallback* callback, const btVector3& aabbMin, const btVector3& aabbMax) const
 {
-		struct FilteredCallback : public btInternalTriangleIndexCallback
+	struct FilteredCallback : public btInternalTriangleIndexCallback
 	{
 		btTriangleCallback* m_callback;
 		btVector3 m_aabbMin;
 		btVector3 m_aabbMax;
 
-		FilteredCallback(btTriangleCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax)
-			:m_callback(callback),
-			m_aabbMin(aabbMin),
-			m_aabbMax(aabbMax)
+		FilteredCallback(btTriangleCallback* callback, const btVector3& aabbMin, const btVector3& aabbMax)
+			: m_callback(callback),
+			  m_aabbMin(aabbMin),
+			  m_aabbMax(aabbMax)
 		{
 		}
 
-		virtual void internalProcessTriangleIndex(btVector3* triangle,int partId,int triangleIndex)
+		virtual void internalProcessTriangleIndex(btVector3* triangle, int partId, int triangleIndex)
 		{
-			if (TestTriangleAgainstAabb2(&triangle[0],m_aabbMin,m_aabbMax))
+			if (TestTriangleAgainstAabb2(&triangle[0], m_aabbMin, m_aabbMax))
 			{
 				//check aabb in triangle-space, before doing this
-				m_callback->processTriangle(triangle,partId,triangleIndex);
+				m_callback->processTriangle(triangle, partId, triangleIndex);
 			}
-			
 		}
-
 	};
 
-	FilteredCallback filterCallback(callback,aabbMin,aabbMax);
+	FilteredCallback filterCallback(callback, aabbMin, aabbMax);
 
-	m_meshInterface->InternalProcessAllTriangles(&filterCallback,aabbMin,aabbMax);
+	m_meshInterface->InternalProcessAllTriangles(&filterCallback, aabbMin, aabbMax);
 }
 
-
-
-
-
-void	btTriangleMeshShape::calculateLocalInertia(btScalar mass,btVector3& inertia) const
+void btTriangleMeshShape::calculateLocalInertia(btScalar mass, btVector3& inertia) const
 {
 	(void)mass;
 	//moving concave objects not supported
 	btAssert(0);
-	inertia.setValue(btScalar(0.),btScalar(0.),btScalar(0.));
+	inertia.setValue(btScalar(0.), btScalar(0.), btScalar(0.));
 }
 
-
 btVector3 btTriangleMeshShape::localGetSupportingVertex(const btVector3& vec) const
 {
 	btVector3 supportVertex;
@@ -193,15 +167,13 @@ btVector3 btTriangleMeshShape::localGetSupportingVertex(const btVector3& vec) co
 	btTransform ident;
 	ident.setIdentity();
 
-	SupportVertexCallback supportCallback(vec,ident);
+	SupportVertexCallback supportCallback(vec, ident);
+
+	btVector3 aabbMax(btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT));
+
+	processAllTriangles(&supportCallback, -aabbMax, aabbMax);
 
-	btVector3 aabbMax(btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT));
-	
-	processAllTriangles(&supportCallback,-aabbMax,aabbMax);
-		
 	supportVertex = supportCallback.GetSupportVertexLocal();
 
 	return supportVertex;
 }
-
-
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleMeshShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleMeshShape.h
index 453e58005a4..4a70e283fae 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleMeshShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleMeshShape.h
@@ -19,18 +19,18 @@ subject to the following restrictions:
 #include "btConcaveShape.h"
 #include "btStridingMeshInterface.h"
 
-
 ///The btTriangleMeshShape is an internal concave triangle mesh interface. Don't use this class directly, use btBvhTriangleMeshShape instead.
-ATTRIBUTE_ALIGNED16(class) btTriangleMeshShape : public btConcaveShape
+ATTRIBUTE_ALIGNED16(class)
+btTriangleMeshShape : public btConcaveShape
 {
 protected:
-	btVector3	m_localAabbMin;
-	btVector3	m_localAabbMax;
+	btVector3 m_localAabbMin;
+	btVector3 m_localAabbMax;
 	btStridingMeshInterface* m_meshInterface;
 
 	///btTriangleMeshShape constructor has been disabled/protected, so that users will not mistakenly use this class.
 	///Don't use btTriangleMeshShape but use btBvhTriangleMeshShape instead!
-	btTriangleMeshShape(btStridingMeshInterface* meshInterface);
+	btTriangleMeshShape(btStridingMeshInterface * meshInterface);
 
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
@@ -39,23 +39,23 @@ public:
 
 	virtual btVector3 localGetSupportingVertex(const btVector3& vec) const;
 
-	virtual btVector3	localGetSupportingVertexWithoutMargin(const btVector3& vec)const
+	virtual btVector3 localGetSupportingVertexWithoutMargin(const btVector3& vec) const
 	{
 		btAssert(0);
 		return localGetSupportingVertex(vec);
 	}
 
-	void	recalcLocalAabb();
+	void recalcLocalAabb();
 
-	virtual void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const;
+	virtual void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
 
-	virtual void	processAllTriangles(btTriangleCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax) const;
+	virtual void processAllTriangles(btTriangleCallback * callback, const btVector3& aabbMin, const btVector3& aabbMax) const;
 
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
+	virtual void calculateLocalInertia(btScalar mass, btVector3 & inertia) const;
 
-	virtual void	setLocalScaling(const btVector3& scaling);
+	virtual void setLocalScaling(const btVector3& scaling);
 	virtual const btVector3& getLocalScaling() const;
-	
+
 	btStridingMeshInterface* getMeshInterface()
 	{
 		return m_meshInterface;
@@ -75,16 +75,8 @@ public:
 		return m_localAabbMax;
 	}
 
-
-
 	//debugging
-	virtual const char*	getName()const {return "TRIANGLEMESH";}
-
-	
-
+	virtual const char* getName() const { return "TRIANGLEMESH"; }
 };
 
-
-
-
-#endif //BT_TRIANGLE_MESH_SHAPE_H
+#endif  //BT_TRIANGLE_MESH_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleShape.h
index a8a80f82fe2..190cbdae69e 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btTriangleShape.h
@@ -19,15 +19,13 @@ subject to the following restrictions:
 #include "btConvexShape.h"
 #include "btBoxShape.h"
 
-ATTRIBUTE_ALIGNED16(class) btTriangleShape : public btPolyhedralConvexShape
+ATTRIBUTE_ALIGNED16(class)
+btTriangleShape : public btPolyhedralConvexShape
 {
-
-
 public:
+	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-BT_DECLARE_ALIGNED_ALLOCATOR();
-
-	btVector3	m_vertices1[3];
+	btVector3 m_vertices1[3];
 
 	virtual int getNumVertices() const
 	{
@@ -43,7 +41,7 @@ BT_DECLARE_ALIGNED_ALLOCATOR();
 	{
 		return m_vertices1[index];
 	}
-	virtual void getVertex(int index,btVector3& vert) const
+	virtual void getVertex(int index, btVector3& vert) const
 	{
 		vert = m_vertices1[index];
 	}
@@ -52,83 +50,79 @@ BT_DECLARE_ALIGNED_ALLOCATOR();
 	{
 		return 3;
 	}
-	
-	virtual void getEdge(int i,btVector3& pa,btVector3& pb) const
+
+	virtual void getEdge(int i, btVector3& pa, btVector3& pb) const
 	{
-		getVertex(i,pa);
-		getVertex((i+1)%3,pb);
+		getVertex(i, pa);
+		getVertex((i + 1) % 3, pb);
 	}
 
-
-	virtual void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax)const 
+	virtual void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
 	{
-//		btAssert(0);
-		getAabbSlow(t,aabbMin,aabbMax);
+		//		btAssert(0);
+		getAabbSlow(t, aabbMin, aabbMax);
 	}
 
-	btVector3 localGetSupportingVertexWithoutMargin(const btVector3& dir)const 
+	btVector3 localGetSupportingVertexWithoutMargin(const btVector3& dir) const
 	{
-        btVector3 dots = dir.dot3(m_vertices1[0], m_vertices1[1], m_vertices1[2]);
-	  	return m_vertices1[dots.maxAxis()];
-
+		btVector3 dots = dir.dot3(m_vertices1[0], m_vertices1[1], m_vertices1[2]);
+		return m_vertices1[dots.maxAxis()];
 	}
 
-	virtual void	batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const
+	virtual void batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const
 	{
-		for (int i=0;i<numVectors;i++)
+		for (int i = 0; i < numVectors; i++)
 		{
 			const btVector3& dir = vectors[i];
-            btVector3 dots = dir.dot3(m_vertices1[0], m_vertices1[1], m_vertices1[2]);
-  			supportVerticesOut[i] = m_vertices1[dots.maxAxis()];
+			btVector3 dots = dir.dot3(m_vertices1[0], m_vertices1[1], m_vertices1[2]);
+			supportVerticesOut[i] = m_vertices1[dots.maxAxis()];
 		}
-
 	}
 
-	btTriangleShape() : btPolyhedralConvexShape ()
-    {
-		m_shapeType = TRIANGLE_SHAPE_PROXYTYPE;
-	}
-
-	btTriangleShape(const btVector3& p0,const btVector3& p1,const btVector3& p2) : btPolyhedralConvexShape ()
-    {
-		m_shapeType = TRIANGLE_SHAPE_PROXYTYPE;
-        m_vertices1[0] = p0;
-        m_vertices1[1] = p1;
-        m_vertices1[2] = p2;
-    }
-
-
-	virtual void getPlane(btVector3& planeNormal,btVector3& planeSupport,int i) const
+	btTriangleShape() : btPolyhedralConvexShape()
 	{
-		getPlaneEquation(i,planeNormal,planeSupport);
+		m_shapeType = TRIANGLE_SHAPE_PROXYTYPE;
 	}
 
-	virtual int	getNumPlanes() const
+	btTriangleShape(const btVector3& p0, const btVector3& p1, const btVector3& p2) : btPolyhedralConvexShape()
+	{
+		m_shapeType = TRIANGLE_SHAPE_PROXYTYPE;
+		m_vertices1[0] = p0;
+		m_vertices1[1] = p1;
+		m_vertices1[2] = p2;
+	}
+
+	virtual void getPlane(btVector3 & planeNormal, btVector3 & planeSupport, int i) const
+	{
+		getPlaneEquation(i, planeNormal, planeSupport);
+	}
+
+	virtual int getNumPlanes() const
 	{
 		return 1;
 	}
 
-	void calcNormal(btVector3& normal) const
+	void calcNormal(btVector3 & normal) const
 	{
-		normal = (m_vertices1[1]-m_vertices1[0]).cross(m_vertices1[2]-m_vertices1[0]);
+		normal = (m_vertices1[1] - m_vertices1[0]).cross(m_vertices1[2] - m_vertices1[0]);
 		normal.normalize();
 	}
 
-	virtual void getPlaneEquation(int i, btVector3& planeNormal,btVector3& planeSupport) const
+	virtual void getPlaneEquation(int i, btVector3& planeNormal, btVector3& planeSupport) const
 	{
 		(void)i;
 		calcNormal(planeNormal);
 		planeSupport = m_vertices1[0];
 	}
 
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const
+	virtual void calculateLocalInertia(btScalar mass, btVector3 & inertia) const
 	{
 		(void)mass;
 		btAssert(0);
-		inertia.setValue(btScalar(0.),btScalar(0.),btScalar(0.));
+		inertia.setValue(btScalar(0.), btScalar(0.), btScalar(0.));
 	}
 
-		virtual	bool isInside(const btVector3& pt,btScalar tolerance) const
+	virtual bool isInside(const btVector3& pt, btScalar tolerance) const
 	{
 		btVector3 normal;
 		calcNormal(normal);
@@ -140,45 +134,42 @@ BT_DECLARE_ALIGNED_ALLOCATOR();
 		{
 			//inside check on edge-planes
 			int i;
-			for (i=0;i<3;i++)
+			for (i = 0; i < 3; i++)
 			{
-				btVector3 pa,pb;
-				getEdge(i,pa,pb);
-				btVector3 edge = pb-pa;
+				btVector3 pa, pb;
+				getEdge(i, pa, pb);
+				btVector3 edge = pb - pa;
 				btVector3 edgeNormal = edge.cross(normal);
 				edgeNormal.normalize();
-				btScalar dist = pt.dot( edgeNormal);
+				btScalar dist = pt.dot(edgeNormal);
 				btScalar edgeConst = pa.dot(edgeNormal);
 				dist -= edgeConst;
 				if (dist < -tolerance)
 					return false;
 			}
-			
+
 			return true;
 		}
 
 		return false;
 	}
-		//debugging
-		virtual const char*	getName()const
-		{
-			return "Triangle";
-		}
-
-		virtual int		getNumPreferredPenetrationDirections() const
-		{
-			return 2;
-		}
-		
-		virtual void	getPreferredPenetrationDirection(int index, btVector3& penetrationVector) const
-		{
-			calcNormal(penetrationVector);
-			if (index)
-				penetrationVector *= btScalar(-1.);
-		}
+	//debugging
+	virtual const char* getName() const
+	{
+		return "Triangle";
+	}
 
+	virtual int getNumPreferredPenetrationDirections() const
+	{
+		return 2;
+	}
 
+	virtual void getPreferredPenetrationDirection(int index, btVector3& penetrationVector) const
+	{
+		calcNormal(penetrationVector);
+		if (index)
+			penetrationVector *= btScalar(-1.);
+	}
 };
 
-#endif //BT_OBB_TRIANGLE_MINKOWSKI_H
-
+#endif  //BT_OBB_TRIANGLE_MINKOWSKI_H
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btUniformScalingShape.cpp b/thirdparty/bullet/BulletCollision/CollisionShapes/btUniformScalingShape.cpp
index b148bbd99a5..ed3cd2d2597 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btUniformScalingShape.cpp
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btUniformScalingShape.cpp
@@ -15,122 +15,112 @@ subject to the following restrictions:
 
 #include "btUniformScalingShape.h"
 
-btUniformScalingShape::btUniformScalingShape(	btConvexShape* convexChildShape,btScalar uniformScalingFactor):
-btConvexShape (), m_childConvexShape(convexChildShape),
-m_uniformScalingFactor(uniformScalingFactor)
+btUniformScalingShape::btUniformScalingShape(btConvexShape* convexChildShape, btScalar uniformScalingFactor) : btConvexShape(), m_childConvexShape(convexChildShape), m_uniformScalingFactor(uniformScalingFactor)
 {
 	m_shapeType = UNIFORM_SCALING_SHAPE_PROXYTYPE;
 }
-	
+
 btUniformScalingShape::~btUniformScalingShape()
 {
 }
-	
 
-btVector3	btUniformScalingShape::localGetSupportingVertexWithoutMargin(const btVector3& vec)const
+btVector3 btUniformScalingShape::localGetSupportingVertexWithoutMargin(const btVector3& vec) const
 {
 	btVector3 tmpVertex;
 	tmpVertex = m_childConvexShape->localGetSupportingVertexWithoutMargin(vec);
-	return tmpVertex*m_uniformScalingFactor;
+	return tmpVertex * m_uniformScalingFactor;
 }
 
-void	btUniformScalingShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const
+void btUniformScalingShape::batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const
 {
-	m_childConvexShape->batchedUnitVectorGetSupportingVertexWithoutMargin(vectors,supportVerticesOut,numVectors);
+	m_childConvexShape->batchedUnitVectorGetSupportingVertexWithoutMargin(vectors, supportVerticesOut, numVectors);
 	int i;
-	for (i=0;i<numVectors;i++)
+	for (i = 0; i < numVectors; i++)
 	{
 		supportVerticesOut[i] = supportVerticesOut[i] * m_uniformScalingFactor;
 	}
 }
 
-
-btVector3	btUniformScalingShape::localGetSupportingVertex(const btVector3& vec)const
+btVector3 btUniformScalingShape::localGetSupportingVertex(const btVector3& vec) const
 {
 	btVector3 tmpVertex;
 	tmpVertex = m_childConvexShape->localGetSupportingVertex(vec);
-	return tmpVertex*m_uniformScalingFactor;
+	return tmpVertex * m_uniformScalingFactor;
 }
 
-
-void	btUniformScalingShape::calculateLocalInertia(btScalar mass,btVector3& inertia) const
+void btUniformScalingShape::calculateLocalInertia(btScalar mass, btVector3& inertia) const
 {
-
 	///this linear upscaling is not realistic, but we don't deal with large mass ratios...
 	btVector3 tmpInertia;
-	m_childConvexShape->calculateLocalInertia(mass,tmpInertia);
+	m_childConvexShape->calculateLocalInertia(mass, tmpInertia);
 	inertia = tmpInertia * m_uniformScalingFactor;
 }
 
-
-	///getAabb's default implementation is brute force, expected derived classes to implement a fast dedicated version
-void btUniformScalingShape::getAabb(const btTransform& trans,btVector3& aabbMin,btVector3& aabbMax) const
+///getAabb's default implementation is brute force, expected derived classes to implement a fast dedicated version
+void btUniformScalingShape::getAabb(const btTransform& trans, btVector3& aabbMin, btVector3& aabbMax) const
 {
-	getAabbSlow(trans,aabbMin,aabbMax);
-
+	getAabbSlow(trans, aabbMin, aabbMax);
 }
 
-void btUniformScalingShape::getAabbSlow(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const
+void btUniformScalingShape::getAabbSlow(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
 {
 #if 1
 	btVector3 _directions[] =
-	{
-		btVector3( 1.,  0.,  0.),
-		btVector3( 0.,  1.,  0.),
-		btVector3( 0.,  0.,  1.),
-		btVector3( -1., 0.,  0.),
-		btVector3( 0., -1.,  0.),
-		btVector3( 0.,  0., -1.)
-	};
-	
+		{
+			btVector3(1., 0., 0.),
+			btVector3(0., 1., 0.),
+			btVector3(0., 0., 1.),
+			btVector3(-1., 0., 0.),
+			btVector3(0., -1., 0.),
+			btVector3(0., 0., -1.)};
+
 	btVector3 _supporting[] =
-	{
-		btVector3( 0., 0., 0.),
-		btVector3( 0., 0., 0.),
-		btVector3( 0., 0., 0.),
-		btVector3( 0., 0., 0.),
-		btVector3( 0., 0., 0.),
-		btVector3( 0., 0., 0.)
-	};
+		{
+			btVector3(0., 0., 0.),
+			btVector3(0., 0., 0.),
+			btVector3(0., 0., 0.),
+			btVector3(0., 0., 0.),
+			btVector3(0., 0., 0.),
+			btVector3(0., 0., 0.)};
 
-	for (int i=0;i<6;i++)
+	for (int i = 0; i < 6; i++)
 	{
-		_directions[i] = _directions[i]*t.getBasis();
+		_directions[i] = _directions[i] * t.getBasis();
 	}
-	
-	batchedUnitVectorGetSupportingVertexWithoutMargin(_directions, _supporting, 6);
-	
-	btVector3 aabbMin1(0,0,0),aabbMax1(0,0,0);
 
-	for ( int i = 0; i < 3; ++i )
+	batchedUnitVectorGetSupportingVertexWithoutMargin(_directions, _supporting, 6);
+
+	btVector3 aabbMin1(0, 0, 0), aabbMax1(0, 0, 0);
+
+	for (int i = 0; i < 3; ++i)
 	{
 		aabbMax1[i] = t(_supporting[i])[i];
 		aabbMin1[i] = t(_supporting[i + 3])[i];
 	}
-	btVector3 marginVec(getMargin(),getMargin(),getMargin());
-	aabbMin = aabbMin1-marginVec;
-	aabbMax = aabbMax1+marginVec;
-	
+	btVector3 marginVec(getMargin(), getMargin(), getMargin());
+	aabbMin = aabbMin1 - marginVec;
+	aabbMax = aabbMax1 + marginVec;
+
 #else
 
 	btScalar margin = getMargin();
-	for (int i=0;i<3;i++)
+	for (int i = 0; i < 3; i++)
 	{
-		btVector3 vec(btScalar(0.),btScalar(0.),btScalar(0.));
+		btVector3 vec(btScalar(0.), btScalar(0.), btScalar(0.));
 		vec[i] = btScalar(1.);
-		btVector3 sv = localGetSupportingVertex(vec*t.getBasis());
+		btVector3 sv = localGetSupportingVertex(vec * t.getBasis());
 		btVector3 tmp = t(sv);
-		aabbMax[i] = tmp[i]+margin;
+		aabbMax[i] = tmp[i] + margin;
 		vec[i] = btScalar(-1.);
-		sv = localGetSupportingVertex(vec*t.getBasis());
+		sv = localGetSupportingVertex(vec * t.getBasis());
 		tmp = t(sv);
-		aabbMin[i] = tmp[i]-margin;
+		aabbMin[i] = tmp[i] - margin;
 	}
 
 #endif
 }
 
-void	btUniformScalingShape::setLocalScaling(const btVector3& scaling) 
+void btUniformScalingShape::setLocalScaling(const btVector3& scaling)
 {
 	m_childConvexShape->setLocalScaling(scaling);
 }
@@ -140,21 +130,21 @@ const btVector3& btUniformScalingShape::getLocalScaling() const
 	return m_childConvexShape->getLocalScaling();
 }
 
-void	btUniformScalingShape::setMargin(btScalar margin)
+void btUniformScalingShape::setMargin(btScalar margin)
 {
 	m_childConvexShape->setMargin(margin);
 }
-btScalar	btUniformScalingShape::getMargin() const
+btScalar btUniformScalingShape::getMargin() const
 {
 	return m_childConvexShape->getMargin() * m_uniformScalingFactor;
 }
 
-int		btUniformScalingShape::getNumPreferredPenetrationDirections() const
+int btUniformScalingShape::getNumPreferredPenetrationDirections() const
 {
 	return m_childConvexShape->getNumPreferredPenetrationDirections();
 }
-	
-void	btUniformScalingShape::getPreferredPenetrationDirection(int index, btVector3& penetrationVector) const
+
+void btUniformScalingShape::getPreferredPenetrationDirection(int index, btVector3& penetrationVector) const
 {
-	m_childConvexShape->getPreferredPenetrationDirection(index,penetrationVector);
+	m_childConvexShape->getPreferredPenetrationDirection(index, penetrationVector);
 }
diff --git a/thirdparty/bullet/BulletCollision/CollisionShapes/btUniformScalingShape.h b/thirdparty/bullet/BulletCollision/CollisionShapes/btUniformScalingShape.h
index a10f58d2429..4dfe34efbdc 100644
--- a/thirdparty/bullet/BulletCollision/CollisionShapes/btUniformScalingShape.h
+++ b/thirdparty/bullet/BulletCollision/CollisionShapes/btUniformScalingShape.h
@@ -17,73 +17,68 @@ subject to the following restrictions:
 #define BT_UNIFORM_SCALING_SHAPE_H
 
 #include "btConvexShape.h"
-#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h" // for the types
+#include "BulletCollision/BroadphaseCollision/btBroadphaseProxy.h"  // for the types
 
 ///The btUniformScalingShape allows to re-use uniform scaled instances of btConvexShape in a memory efficient way.
 ///Istead of using btUniformScalingShape, it is better to use the non-uniform setLocalScaling method on convex shapes that implement it.
-ATTRIBUTE_ALIGNED16(class) btUniformScalingShape : public btConvexShape
+ATTRIBUTE_ALIGNED16(class)
+btUniformScalingShape : public btConvexShape
 {
-	btConvexShape*	m_childConvexShape;
+	btConvexShape* m_childConvexShape;
 
-	btScalar	m_uniformScalingFactor;
-	
-	public:
-	
+	btScalar m_uniformScalingFactor;
+
+public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
-	btUniformScalingShape(	btConvexShape* convexChildShape, btScalar uniformScalingFactor);
-	
+
+	btUniformScalingShape(btConvexShape * convexChildShape, btScalar uniformScalingFactor);
+
 	virtual ~btUniformScalingShape();
-	
-	virtual btVector3	localGetSupportingVertexWithoutMargin(const btVector3& vec)const;
 
-	virtual btVector3	localGetSupportingVertex(const btVector3& vec)const;
+	virtual btVector3 localGetSupportingVertexWithoutMargin(const btVector3& vec) const;
 
-	virtual void	batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const;
+	virtual btVector3 localGetSupportingVertex(const btVector3& vec) const;
 
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
+	virtual void batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const;
 
-	btScalar	getUniformScalingFactor() const
+	virtual void calculateLocalInertia(btScalar mass, btVector3 & inertia) const;
+
+	btScalar getUniformScalingFactor() const
 	{
 		return m_uniformScalingFactor;
 	}
 
-	btConvexShape*	getChildShape() 
+	btConvexShape* getChildShape()
 	{
 		return m_childConvexShape;
 	}
 
-	const btConvexShape*	getChildShape() const
+	const btConvexShape* getChildShape() const
 	{
 		return m_childConvexShape;
 	}
 
-	virtual const char*	getName()const 
+	virtual const char* getName() const
 	{
 		return "UniformScalingShape";
 	}
-	
-
 
 	///////////////////////////
 
-
 	///getAabb's default implementation is brute force, expected derived classes to implement a fast dedicated version
-	void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const;
+	void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
 
-	virtual void getAabbSlow(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const;
+	virtual void getAabbSlow(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const;
 
-	virtual void	setLocalScaling(const btVector3& scaling) ;
-	virtual const btVector3& getLocalScaling() const ;
+	virtual void setLocalScaling(const btVector3& scaling);
+	virtual const btVector3& getLocalScaling() const;
 
-	virtual void	setMargin(btScalar margin);
-	virtual btScalar	getMargin() const;
-
-	virtual int		getNumPreferredPenetrationDirections() const;
-	
-	virtual void	getPreferredPenetrationDirection(int index, btVector3& penetrationVector) const;
+	virtual void setMargin(btScalar margin);
+	virtual btScalar getMargin() const;
 
+	virtual int getNumPreferredPenetrationDirections() const;
 
+	virtual void getPreferredPenetrationDirection(int index, btVector3& penetrationVector) const;
 };
 
-#endif //BT_UNIFORM_SCALING_SHAPE_H
+#endif  //BT_UNIFORM_SCALING_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btBoxCollision.h b/thirdparty/bullet/BulletCollision/Gimpact/btBoxCollision.h
index 0a0357e5a81..182835c3b44 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btBoxCollision.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btBoxCollision.h
@@ -26,27 +26,21 @@ subject to the following restrictions:
 
 #include "LinearMath/btTransform.h"
 
-
 ///Swap numbers
-#define BT_SWAP_NUMBERS(a,b){ \
-    a = a+b; \
-    b = a-b; \
-    a = a-b; \
-}\
-
-
-#define BT_MAX(a,b) (a<b?b:a)
-#define BT_MIN(a,b) (a>b?b:a)
-
-#define BT_GREATER(x, y)	btFabs(x) > (y)
-
-#define BT_MAX3(a,b,c) BT_MAX(a,BT_MAX(b,c))
-#define BT_MIN3(a,b,c) BT_MIN(a,BT_MIN(b,c))
-
-
+#define BT_SWAP_NUMBERS(a, b) \
+	{                         \
+		a = a + b;            \
+		b = a - b;            \
+		a = a - b;            \
+	}
 
+#define BT_MAX(a, b) (a < b ? b : a)
+#define BT_MIN(a, b) (a > b ? b : a)
 
+#define BT_GREATER(x, y) btFabs(x) > (y)
 
+#define BT_MAX3(a, b, c) BT_MAX(a, BT_MAX(b, c))
+#define BT_MIN3(a, b, c) BT_MIN(a, BT_MIN(b, c))
 
 enum eBT_PLANE_INTERSECTION_TYPE
 {
@@ -115,152 +109,144 @@ enum eBT_PLANE_INTERSECTION_TYPE
 //	return test_cross_edge_box(edge,absolute_edge,pointa,pointb,extend,1,0,0,1);
 //}
 
+#define TEST_CROSS_EDGE_BOX_MCR(edge, absolute_edge, pointa, pointb, _extend, i_dir_0, i_dir_1, i_comp_0, i_comp_1) \
+	{                                                                                                               \
+		const btScalar dir0 = -edge[i_dir_0];                                                                       \
+		const btScalar dir1 = edge[i_dir_1];                                                                        \
+		btScalar pmin = pointa[i_comp_0] * dir0 + pointa[i_comp_1] * dir1;                                          \
+		btScalar pmax = pointb[i_comp_0] * dir0 + pointb[i_comp_1] * dir1;                                          \
+		if (pmin > pmax)                                                                                            \
+		{                                                                                                           \
+			BT_SWAP_NUMBERS(pmin, pmax);                                                                            \
+		}                                                                                                           \
+		const btScalar abs_dir0 = absolute_edge[i_dir_0];                                                           \
+		const btScalar abs_dir1 = absolute_edge[i_dir_1];                                                           \
+		const btScalar rad = _extend[i_comp_0] * abs_dir0 + _extend[i_comp_1] * abs_dir1;                           \
+		if (pmin > rad || -rad > pmax) return false;                                                                \
+	}
 
-#define TEST_CROSS_EDGE_BOX_MCR(edge,absolute_edge,pointa,pointb,_extend,i_dir_0,i_dir_1,i_comp_0,i_comp_1)\
-{\
-	const btScalar dir0 = -edge[i_dir_0];\
-	const btScalar dir1 = edge[i_dir_1];\
-	btScalar pmin = pointa[i_comp_0]*dir0 + pointa[i_comp_1]*dir1;\
-	btScalar pmax = pointb[i_comp_0]*dir0 + pointb[i_comp_1]*dir1;\
-	if(pmin>pmax)\
-	{\
-		BT_SWAP_NUMBERS(pmin,pmax); \
-	}\
-	const btScalar abs_dir0 = absolute_edge[i_dir_0];\
-	const btScalar abs_dir1 = absolute_edge[i_dir_1];\
-	const btScalar rad = _extend[i_comp_0] * abs_dir0 + _extend[i_comp_1] * abs_dir1;\
-	if(pmin>rad || -rad>pmax) return false;\
-}\
+#define TEST_CROSS_EDGE_BOX_X_AXIS_MCR(edge, absolute_edge, pointa, pointb, _extend)       \
+	{                                                                                      \
+		TEST_CROSS_EDGE_BOX_MCR(edge, absolute_edge, pointa, pointb, _extend, 2, 1, 1, 2); \
+	}
 
+#define TEST_CROSS_EDGE_BOX_Y_AXIS_MCR(edge, absolute_edge, pointa, pointb, _extend)       \
+	{                                                                                      \
+		TEST_CROSS_EDGE_BOX_MCR(edge, absolute_edge, pointa, pointb, _extend, 0, 2, 2, 0); \
+	}
 
-#define TEST_CROSS_EDGE_BOX_X_AXIS_MCR(edge,absolute_edge,pointa,pointb,_extend)\
-{\
-	TEST_CROSS_EDGE_BOX_MCR(edge,absolute_edge,pointa,pointb,_extend,2,1,1,2);\
-}\
-
-#define TEST_CROSS_EDGE_BOX_Y_AXIS_MCR(edge,absolute_edge,pointa,pointb,_extend)\
-{\
-	TEST_CROSS_EDGE_BOX_MCR(edge,absolute_edge,pointa,pointb,_extend,0,2,2,0);\
-}\
-
-#define TEST_CROSS_EDGE_BOX_Z_AXIS_MCR(edge,absolute_edge,pointa,pointb,_extend)\
-{\
-	TEST_CROSS_EDGE_BOX_MCR(edge,absolute_edge,pointa,pointb,_extend,1,0,0,1);\
-}\
-
+#define TEST_CROSS_EDGE_BOX_Z_AXIS_MCR(edge, absolute_edge, pointa, pointb, _extend)       \
+	{                                                                                      \
+		TEST_CROSS_EDGE_BOX_MCR(edge, absolute_edge, pointa, pointb, _extend, 1, 0, 0, 1); \
+	}
 
 //! Returns the dot product between a vec3f and the col of a matrix
 SIMD_FORCE_INLINE btScalar bt_mat3_dot_col(
-const btMatrix3x3 & mat, const btVector3 & vec3, int colindex)
+	const btMatrix3x3 &mat, const btVector3 &vec3, int colindex)
 {
-	return vec3[0]*mat[0][colindex] + vec3[1]*mat[1][colindex] + vec3[2]*mat[2][colindex];
+	return vec3[0] * mat[0][colindex] + vec3[1] * mat[1][colindex] + vec3[2] * mat[2][colindex];
 }
 
-
 //!  Class for transforming a model1 to the space of model0
-ATTRIBUTE_ALIGNED16	(class) BT_BOX_BOX_TRANSFORM_CACHE
+ATTRIBUTE_ALIGNED16(class)
+BT_BOX_BOX_TRANSFORM_CACHE
 {
 public:
-    btVector3  m_T1to0;//!< Transforms translation of model1 to model 0
-	btMatrix3x3 m_R1to0;//!< Transforms Rotation of model1 to model 0, equal  to R0' * R1
-	btMatrix3x3 m_AR;//!< Absolute value of m_R1to0
+	btVector3 m_T1to0;    //!< Transforms translation of model1 to model 0
+	btMatrix3x3 m_R1to0;  //!< Transforms Rotation of model1 to model 0, equal  to R0' * R1
+	btMatrix3x3 m_AR;     //!< Absolute value of m_R1to0
 
 	SIMD_FORCE_INLINE void calc_absolute_matrix()
 	{
-//		static const btVector3 vepsi(1e-6f,1e-6f,1e-6f);
-//		m_AR[0] = vepsi + m_R1to0[0].absolute();
-//		m_AR[1] = vepsi + m_R1to0[1].absolute();
-//		m_AR[2] = vepsi + m_R1to0[2].absolute();
+		//		static const btVector3 vepsi(1e-6f,1e-6f,1e-6f);
+		//		m_AR[0] = vepsi + m_R1to0[0].absolute();
+		//		m_AR[1] = vepsi + m_R1to0[1].absolute();
+		//		m_AR[2] = vepsi + m_R1to0[2].absolute();
 
-		int i,j;
-
-        for(i=0;i<3;i++)
-        {
-            for(j=0;j<3;j++ )
-            {
-            	m_AR[i][j] = 1e-6f + btFabs(m_R1to0[i][j]);
-            }
-        }
+		int i, j;
 
+		for (i = 0; i < 3; i++)
+		{
+			for (j = 0; j < 3; j++)
+			{
+				m_AR[i][j] = 1e-6f + btFabs(m_R1to0[i][j]);
+			}
+		}
 	}
 
 	BT_BOX_BOX_TRANSFORM_CACHE()
 	{
 	}
 
-
-
 	//! Calc the transformation relative  1 to 0. Inverts matrics by transposing
-	SIMD_FORCE_INLINE void calc_from_homogenic(const btTransform & trans0,const btTransform & trans1)
+	SIMD_FORCE_INLINE void calc_from_homogenic(const btTransform &trans0, const btTransform &trans1)
 	{
-
 		btTransform temp_trans = trans0.inverse();
 		temp_trans = temp_trans * trans1;
 
 		m_T1to0 = temp_trans.getOrigin();
 		m_R1to0 = temp_trans.getBasis();
 
-
 		calc_absolute_matrix();
 	}
 
 	//! Calcs the full invertion of the matrices. Useful for scaling matrices
-	SIMD_FORCE_INLINE void calc_from_full_invert(const btTransform & trans0,const btTransform & trans1)
+	SIMD_FORCE_INLINE void calc_from_full_invert(const btTransform &trans0, const btTransform &trans1)
 	{
 		m_R1to0 = trans0.getBasis().inverse();
 		m_T1to0 = m_R1to0 * (-trans0.getOrigin());
 
-		m_T1to0 += m_R1to0*trans1.getOrigin();
+		m_T1to0 += m_R1to0 * trans1.getOrigin();
 		m_R1to0 *= trans1.getBasis();
 
 		calc_absolute_matrix();
 	}
 
-	SIMD_FORCE_INLINE btVector3 transform(const btVector3 & point) const
+	SIMD_FORCE_INLINE btVector3 transform(const btVector3 &point) const
 	{
-        return point.dot3( m_R1to0[0], m_R1to0[1], m_R1to0[2] ) + m_T1to0;
+		return point.dot3(m_R1to0[0], m_R1to0[1], m_R1to0[2]) + m_T1to0;
 	}
 };
 
-
 #define BOX_PLANE_EPSILON 0.000001f
 
 //! Axis aligned box
-ATTRIBUTE_ALIGNED16	(class) btAABB
+ATTRIBUTE_ALIGNED16(class)
+btAABB
 {
 public:
 	btVector3 m_min;
 	btVector3 m_max;
 
 	btAABB()
-	{}
-
-
-	btAABB(const btVector3 & V1,
-			 const btVector3 & V2,
-			 const btVector3 & V3)
 	{
-		m_min[0] = BT_MIN3(V1[0],V2[0],V3[0]);
-		m_min[1] = BT_MIN3(V1[1],V2[1],V3[1]);
-		m_min[2] = BT_MIN3(V1[2],V2[2],V3[2]);
-
-		m_max[0] = BT_MAX3(V1[0],V2[0],V3[0]);
-		m_max[1] = BT_MAX3(V1[1],V2[1],V3[1]);
-		m_max[2] = BT_MAX3(V1[2],V2[2],V3[2]);
 	}
 
-	btAABB(const btVector3 & V1,
-			 const btVector3 & V2,
-			 const btVector3 & V3,
-			 btScalar margin)
+	btAABB(const btVector3 &V1,
+		   const btVector3 &V2,
+		   const btVector3 &V3)
 	{
-		m_min[0] = BT_MIN3(V1[0],V2[0],V3[0]);
-		m_min[1] = BT_MIN3(V1[1],V2[1],V3[1]);
-		m_min[2] = BT_MIN3(V1[2],V2[2],V3[2]);
+		m_min[0] = BT_MIN3(V1[0], V2[0], V3[0]);
+		m_min[1] = BT_MIN3(V1[1], V2[1], V3[1]);
+		m_min[2] = BT_MIN3(V1[2], V2[2], V3[2]);
 
-		m_max[0] = BT_MAX3(V1[0],V2[0],V3[0]);
-		m_max[1] = BT_MAX3(V1[1],V2[1],V3[1]);
-		m_max[2] = BT_MAX3(V1[2],V2[2],V3[2]);
+		m_max[0] = BT_MAX3(V1[0], V2[0], V3[0]);
+		m_max[1] = BT_MAX3(V1[1], V2[1], V3[1]);
+		m_max[2] = BT_MAX3(V1[2], V2[2], V3[2]);
+	}
+
+	btAABB(const btVector3 &V1,
+		   const btVector3 &V2,
+		   const btVector3 &V3,
+		   btScalar margin)
+	{
+		m_min[0] = BT_MIN3(V1[0], V2[0], V3[0]);
+		m_min[1] = BT_MIN3(V1[1], V2[1], V3[1]);
+		m_min[2] = BT_MIN3(V1[2], V2[2], V3[2]);
+
+		m_max[0] = BT_MAX3(V1[0], V2[0], V3[0]);
+		m_max[1] = BT_MAX3(V1[1], V2[1], V3[1]);
+		m_max[2] = BT_MAX3(V1[2], V2[2], V3[2]);
 
 		m_min[0] -= margin;
 		m_min[1] -= margin;
@@ -270,13 +256,11 @@ public:
 		m_max[2] += margin;
 	}
 
-	btAABB(const btAABB &other):
-		m_min(other.m_min),m_max(other.m_max)
+	btAABB(const btAABB &other) : m_min(other.m_min), m_max(other.m_max)
 	{
 	}
 
-	btAABB(const btAABB &other,btScalar margin ):
-		m_min(other.m_min),m_max(other.m_max)
+	btAABB(const btAABB &other, btScalar margin) : m_min(other.m_min), m_max(other.m_max)
 	{
 		m_min[0] -= margin;
 		m_min[1] -= margin;
@@ -317,34 +301,34 @@ public:
 		m_max[2] = other.m_max[2] + margin;
 	}
 
-	template<typename CLASS_POINT>
+	template <typename CLASS_POINT>
 	SIMD_FORCE_INLINE void calc_from_triangle(
-							const CLASS_POINT & V1,
-							const CLASS_POINT & V2,
-							const CLASS_POINT & V3)
+		const CLASS_POINT &V1,
+		const CLASS_POINT &V2,
+		const CLASS_POINT &V3)
 	{
-		m_min[0] = BT_MIN3(V1[0],V2[0],V3[0]);
-		m_min[1] = BT_MIN3(V1[1],V2[1],V3[1]);
-		m_min[2] = BT_MIN3(V1[2],V2[2],V3[2]);
+		m_min[0] = BT_MIN3(V1[0], V2[0], V3[0]);
+		m_min[1] = BT_MIN3(V1[1], V2[1], V3[1]);
+		m_min[2] = BT_MIN3(V1[2], V2[2], V3[2]);
 
-		m_max[0] = BT_MAX3(V1[0],V2[0],V3[0]);
-		m_max[1] = BT_MAX3(V1[1],V2[1],V3[1]);
-		m_max[2] = BT_MAX3(V1[2],V2[2],V3[2]);
+		m_max[0] = BT_MAX3(V1[0], V2[0], V3[0]);
+		m_max[1] = BT_MAX3(V1[1], V2[1], V3[1]);
+		m_max[2] = BT_MAX3(V1[2], V2[2], V3[2]);
 	}
 
-	template<typename CLASS_POINT>
+	template <typename CLASS_POINT>
 	SIMD_FORCE_INLINE void calc_from_triangle_margin(
-							const CLASS_POINT & V1,
-							const CLASS_POINT & V2,
-							const CLASS_POINT & V3, btScalar margin)
+		const CLASS_POINT &V1,
+		const CLASS_POINT &V2,
+		const CLASS_POINT &V3, btScalar margin)
 	{
-		m_min[0] = BT_MIN3(V1[0],V2[0],V3[0]);
-		m_min[1] = BT_MIN3(V1[1],V2[1],V3[1]);
-		m_min[2] = BT_MIN3(V1[2],V2[2],V3[2]);
+		m_min[0] = BT_MIN3(V1[0], V2[0], V3[0]);
+		m_min[1] = BT_MIN3(V1[1], V2[1], V3[1]);
+		m_min[2] = BT_MIN3(V1[2], V2[2], V3[2]);
 
-		m_max[0] = BT_MAX3(V1[0],V2[0],V3[0]);
-		m_max[1] = BT_MAX3(V1[1],V2[1],V3[1]);
-		m_max[2] = BT_MAX3(V1[2],V2[2],V3[2]);
+		m_max[0] = BT_MAX3(V1[0], V2[0], V3[0]);
+		m_max[1] = BT_MAX3(V1[1], V2[1], V3[1]);
+		m_max[2] = BT_MAX3(V1[2], V2[2], V3[2]);
 
 		m_min[0] -= margin;
 		m_min[1] -= margin;
@@ -355,91 +339,89 @@ public:
 	}
 
 	//! Apply a transform to an AABB
-	SIMD_FORCE_INLINE void appy_transform(const btTransform & trans)
+	SIMD_FORCE_INLINE void appy_transform(const btTransform &trans)
 	{
-		btVector3 center = (m_max+m_min)*0.5f;
+		btVector3 center = (m_max + m_min) * 0.5f;
 		btVector3 extends = m_max - center;
 		// Compute new center
 		center = trans(center);
 
-        btVector3 textends = extends.dot3(trans.getBasis().getRow(0).absolute(), 
-                                          trans.getBasis().getRow(1).absolute(), 
-                                          trans.getBasis().getRow(2).absolute());
+		btVector3 textends = extends.dot3(trans.getBasis().getRow(0).absolute(),
+										  trans.getBasis().getRow(1).absolute(),
+										  trans.getBasis().getRow(2).absolute());
 
 		m_min = center - textends;
 		m_max = center + textends;
 	}
 
-
 	//! Apply a transform to an AABB
-	SIMD_FORCE_INLINE void appy_transform_trans_cache(const BT_BOX_BOX_TRANSFORM_CACHE & trans)
+	SIMD_FORCE_INLINE void appy_transform_trans_cache(const BT_BOX_BOX_TRANSFORM_CACHE &trans)
 	{
-		btVector3 center = (m_max+m_min)*0.5f;
+		btVector3 center = (m_max + m_min) * 0.5f;
 		btVector3 extends = m_max - center;
 		// Compute new center
 		center = trans.transform(center);
 
-        btVector3 textends = extends.dot3(trans.m_R1to0.getRow(0).absolute(), 
-                                          trans.m_R1to0.getRow(1).absolute(), 
-                                          trans.m_R1to0.getRow(2).absolute());
-        
+		btVector3 textends = extends.dot3(trans.m_R1to0.getRow(0).absolute(),
+										  trans.m_R1to0.getRow(1).absolute(),
+										  trans.m_R1to0.getRow(2).absolute());
+
 		m_min = center - textends;
 		m_max = center + textends;
 	}
 
 	//! Merges a Box
-	SIMD_FORCE_INLINE void merge(const btAABB & box)
+	SIMD_FORCE_INLINE void merge(const btAABB &box)
 	{
-		m_min[0] = BT_MIN(m_min[0],box.m_min[0]);
-		m_min[1] = BT_MIN(m_min[1],box.m_min[1]);
-		m_min[2] = BT_MIN(m_min[2],box.m_min[2]);
+		m_min[0] = BT_MIN(m_min[0], box.m_min[0]);
+		m_min[1] = BT_MIN(m_min[1], box.m_min[1]);
+		m_min[2] = BT_MIN(m_min[2], box.m_min[2]);
 
-		m_max[0] = BT_MAX(m_max[0],box.m_max[0]);
-		m_max[1] = BT_MAX(m_max[1],box.m_max[1]);
-		m_max[2] = BT_MAX(m_max[2],box.m_max[2]);
+		m_max[0] = BT_MAX(m_max[0], box.m_max[0]);
+		m_max[1] = BT_MAX(m_max[1], box.m_max[1]);
+		m_max[2] = BT_MAX(m_max[2], box.m_max[2]);
 	}
 
 	//! Merges a point
-	template<typename CLASS_POINT>
-	SIMD_FORCE_INLINE void merge_point(const CLASS_POINT & point)
+	template <typename CLASS_POINT>
+	SIMD_FORCE_INLINE void merge_point(const CLASS_POINT &point)
 	{
-		m_min[0] = BT_MIN(m_min[0],point[0]);
-		m_min[1] = BT_MIN(m_min[1],point[1]);
-		m_min[2] = BT_MIN(m_min[2],point[2]);
+		m_min[0] = BT_MIN(m_min[0], point[0]);
+		m_min[1] = BT_MIN(m_min[1], point[1]);
+		m_min[2] = BT_MIN(m_min[2], point[2]);
 
-		m_max[0] = BT_MAX(m_max[0],point[0]);
-		m_max[1] = BT_MAX(m_max[1],point[1]);
-		m_max[2] = BT_MAX(m_max[2],point[2]);
+		m_max[0] = BT_MAX(m_max[0], point[0]);
+		m_max[1] = BT_MAX(m_max[1], point[1]);
+		m_max[2] = BT_MAX(m_max[2], point[2]);
 	}
 
 	//! Gets the extend and center
-	SIMD_FORCE_INLINE void get_center_extend(btVector3 & center,btVector3 & extend)  const
+	SIMD_FORCE_INLINE void get_center_extend(btVector3 & center, btVector3 & extend) const
 	{
-		center = (m_max+m_min)*0.5f;
+		center = (m_max + m_min) * 0.5f;
 		extend = m_max - center;
 	}
 
 	//! Finds the intersecting box between this box and the other.
-	SIMD_FORCE_INLINE void find_intersection(const btAABB & other, btAABB & intersection)  const
+	SIMD_FORCE_INLINE void find_intersection(const btAABB &other, btAABB &intersection) const
 	{
-		intersection.m_min[0] = BT_MAX(other.m_min[0],m_min[0]);
-		intersection.m_min[1] = BT_MAX(other.m_min[1],m_min[1]);
-		intersection.m_min[2] = BT_MAX(other.m_min[2],m_min[2]);
+		intersection.m_min[0] = BT_MAX(other.m_min[0], m_min[0]);
+		intersection.m_min[1] = BT_MAX(other.m_min[1], m_min[1]);
+		intersection.m_min[2] = BT_MAX(other.m_min[2], m_min[2]);
 
-		intersection.m_max[0] = BT_MIN(other.m_max[0],m_max[0]);
-		intersection.m_max[1] = BT_MIN(other.m_max[1],m_max[1]);
-		intersection.m_max[2] = BT_MIN(other.m_max[2],m_max[2]);
+		intersection.m_max[0] = BT_MIN(other.m_max[0], m_max[0]);
+		intersection.m_max[1] = BT_MIN(other.m_max[1], m_max[1]);
+		intersection.m_max[2] = BT_MIN(other.m_max[2], m_max[2]);
 	}
 
-
-	SIMD_FORCE_INLINE bool has_collision(const btAABB & other) const
+	SIMD_FORCE_INLINE bool has_collision(const btAABB &other) const
 	{
-		if(m_min[0] > other.m_max[0] ||
-		   m_max[0] < other.m_min[0] ||
-		   m_min[1] > other.m_max[1] ||
-		   m_max[1] < other.m_min[1] ||
-		   m_min[2] > other.m_max[2] ||
-		   m_max[2] < other.m_min[2])
+		if (m_min[0] > other.m_max[0] ||
+			m_max[0] < other.m_min[0] ||
+			m_min[1] > other.m_max[1] ||
+			m_max[1] < other.m_min[1] ||
+			m_min[2] > other.m_max[2] ||
+			m_max[2] < other.m_min[2])
 		{
 			return false;
 		}
@@ -451,35 +433,34 @@ public:
 	\param vorigin A vec3f with the origin of the ray
 	\param vdir A vec3f with the direction of the ray
 	*/
-	SIMD_FORCE_INLINE bool collide_ray(const btVector3 & vorigin,const btVector3 & vdir)  const
+	SIMD_FORCE_INLINE bool collide_ray(const btVector3 &vorigin, const btVector3 &vdir) const
 	{
-		btVector3 extents,center;
-		this->get_center_extend(center,extents);;
+		btVector3 extents, center;
+		this->get_center_extend(center, extents);
+		;
 
 		btScalar Dx = vorigin[0] - center[0];
-		if(BT_GREATER(Dx, extents[0]) && Dx*vdir[0]>=0.0f)	return false;
+		if (BT_GREATER(Dx, extents[0]) && Dx * vdir[0] >= 0.0f) return false;
 		btScalar Dy = vorigin[1] - center[1];
-		if(BT_GREATER(Dy, extents[1]) && Dy*vdir[1]>=0.0f)	return false;
+		if (BT_GREATER(Dy, extents[1]) && Dy * vdir[1] >= 0.0f) return false;
 		btScalar Dz = vorigin[2] - center[2];
-		if(BT_GREATER(Dz, extents[2]) && Dz*vdir[2]>=0.0f)	return false;
-
+		if (BT_GREATER(Dz, extents[2]) && Dz * vdir[2] >= 0.0f) return false;
 
 		btScalar f = vdir[1] * Dz - vdir[2] * Dy;
-		if(btFabs(f) > extents[1]*btFabs(vdir[2]) + extents[2]*btFabs(vdir[1])) return false;
+		if (btFabs(f) > extents[1] * btFabs(vdir[2]) + extents[2] * btFabs(vdir[1])) return false;
 		f = vdir[2] * Dx - vdir[0] * Dz;
-		if(btFabs(f) > extents[0]*btFabs(vdir[2]) + extents[2]*btFabs(vdir[0]))return false;
+		if (btFabs(f) > extents[0] * btFabs(vdir[2]) + extents[2] * btFabs(vdir[0])) return false;
 		f = vdir[0] * Dy - vdir[1] * Dx;
-		if(btFabs(f) > extents[0]*btFabs(vdir[1]) + extents[1]*btFabs(vdir[0]))return false;
+		if (btFabs(f) > extents[0] * btFabs(vdir[1]) + extents[1] * btFabs(vdir[0])) return false;
 		return true;
 	}
 
-
-	SIMD_FORCE_INLINE void projection_interval(const btVector3 & direction, btScalar &vmin, btScalar &vmax) const
+	SIMD_FORCE_INLINE void projection_interval(const btVector3 &direction, btScalar &vmin, btScalar &vmax) const
 	{
-		btVector3 center = (m_max+m_min)*0.5f;
-		btVector3 extend = m_max-center;
+		btVector3 center = (m_max + m_min) * 0.5f;
+		btVector3 extend = m_max - center;
 
-		btScalar _fOrigin =  direction.dot(center);
+		btScalar _fOrigin = direction.dot(center);
 		btScalar _fMaximumExtent = extend.dot(direction.absolute());
 		vmin = _fOrigin - _fMaximumExtent;
 		vmax = _fOrigin + _fMaximumExtent;
@@ -487,30 +468,30 @@ public:
 
 	SIMD_FORCE_INLINE eBT_PLANE_INTERSECTION_TYPE plane_classify(const btVector4 &plane) const
 	{
-		btScalar _fmin,_fmax;
-		this->projection_interval(plane,_fmin,_fmax);
+		btScalar _fmin, _fmax;
+		this->projection_interval(plane, _fmin, _fmax);
 
-		if(plane[3] > _fmax + BOX_PLANE_EPSILON)
+		if (plane[3] > _fmax + BOX_PLANE_EPSILON)
 		{
-			return BT_CONST_BACK_PLANE; // 0
+			return BT_CONST_BACK_PLANE;  // 0
 		}
 
-		if(plane[3]+BOX_PLANE_EPSILON >=_fmin)
+		if (plane[3] + BOX_PLANE_EPSILON >= _fmin)
 		{
-			return BT_CONST_COLLIDE_PLANE; //1
+			return BT_CONST_COLLIDE_PLANE;  //1
 		}
-		return BT_CONST_FRONT_PLANE;//2
+		return BT_CONST_FRONT_PLANE;  //2
 	}
 
-	SIMD_FORCE_INLINE bool overlapping_trans_conservative(const btAABB & box, btTransform & trans1_to_0) const
+	SIMD_FORCE_INLINE bool overlapping_trans_conservative(const btAABB &box, btTransform &trans1_to_0) const
 	{
 		btAABB tbox = box;
 		tbox.appy_transform(trans1_to_0);
 		return has_collision(tbox);
 	}
 
-	SIMD_FORCE_INLINE bool overlapping_trans_conservative2(const btAABB & box,
-		const BT_BOX_BOX_TRANSFORM_CACHE & trans1_to_0) const
+	SIMD_FORCE_INLINE bool overlapping_trans_conservative2(const btAABB &box,
+														   const BT_BOX_BOX_TRANSFORM_CACHE &trans1_to_0) const
 	{
 		btAABB tbox = box;
 		tbox.appy_transform_trans_cache(trans1_to_0);
@@ -519,52 +500,50 @@ public:
 
 	//! transcache is the transformation cache from box to this AABB
 	SIMD_FORCE_INLINE bool overlapping_trans_cache(
-		const btAABB & box,const BT_BOX_BOX_TRANSFORM_CACHE & transcache, bool fulltest) const
+		const btAABB &box, const BT_BOX_BOX_TRANSFORM_CACHE &transcache, bool fulltest) const
 	{
-
 		//Taken from OPCODE
-		btVector3 ea,eb;//extends
-		btVector3 ca,cb;//extends
-		get_center_extend(ca,ea);
-		box.get_center_extend(cb,eb);
-
+		btVector3 ea, eb;  //extends
+		btVector3 ca, cb;  //extends
+		get_center_extend(ca, ea);
+		box.get_center_extend(cb, eb);
 
 		btVector3 T;
-		btScalar t,t2;
+		btScalar t, t2;
 		int i;
 
 		// Class I : A's basis vectors
-		for(i=0;i<3;i++)
+		for (i = 0; i < 3; i++)
 		{
-			T[i] =  transcache.m_R1to0[i].dot(cb) + transcache.m_T1to0[i] - ca[i];
+			T[i] = transcache.m_R1to0[i].dot(cb) + transcache.m_T1to0[i] - ca[i];
 			t = transcache.m_AR[i].dot(eb) + ea[i];
-			if(BT_GREATER(T[i], t))	return false;
+			if (BT_GREATER(T[i], t)) return false;
 		}
 		// Class II : B's basis vectors
-		for(i=0;i<3;i++)
+		for (i = 0; i < 3; i++)
 		{
-			t = bt_mat3_dot_col(transcache.m_R1to0,T,i);
-			t2 = bt_mat3_dot_col(transcache.m_AR,ea,i) + eb[i];
-			if(BT_GREATER(t,t2))	return false;
+			t = bt_mat3_dot_col(transcache.m_R1to0, T, i);
+			t2 = bt_mat3_dot_col(transcache.m_AR, ea, i) + eb[i];
+			if (BT_GREATER(t, t2)) return false;
 		}
 		// Class III : 9 cross products
-		if(fulltest)
+		if (fulltest)
 		{
-			int j,m,n,o,p,q,r;
-			for(i=0;i<3;i++)
+			int j, m, n, o, p, q, r;
+			for (i = 0; i < 3; i++)
 			{
-				m = (i+1)%3;
-				n = (i+2)%3;
-				o = i==0?1:0;
-				p = i==2?1:2;
-				for(j=0;j<3;j++)
+				m = (i + 1) % 3;
+				n = (i + 2) % 3;
+				o = i == 0 ? 1 : 0;
+				p = i == 2 ? 1 : 2;
+				for (j = 0; j < 3; j++)
 				{
-					q = j==2?1:2;
-					r = j==0?1:0;
-					t = T[n]*transcache.m_R1to0[m][j] - T[m]*transcache.m_R1to0[n][j];
-					t2 = ea[o]*transcache.m_AR[p][j] + ea[p]*transcache.m_AR[o][j] +
-						eb[r]*transcache.m_AR[i][q] + eb[q]*transcache.m_AR[i][r];
-					if(BT_GREATER(t,t2))	return false;
+					q = j == 2 ? 1 : 2;
+					r = j == 0 ? 1 : 0;
+					t = T[n] * transcache.m_R1to0[m][j] - T[m] * transcache.m_R1to0[n][j];
+					t2 = ea[o] * transcache.m_AR[p][j] + ea[p] * transcache.m_AR[o][j] +
+						 eb[r] * transcache.m_AR[i][q] + eb[q] * transcache.m_AR[i][r];
+					if (BT_GREATER(t, t2)) return false;
 				}
 			}
 		}
@@ -573,7 +552,7 @@ public:
 
 	//! Simple test for planes.
 	SIMD_FORCE_INLINE bool collide_plane(
-		const btVector4 & plane) const
+		const btVector4 &plane) const
 	{
 		eBT_PLANE_INTERSECTION_TYPE classify = plane_classify(plane);
 		return (classify == BT_CONST_COLLIDE_PLANE);
@@ -581,15 +560,15 @@ public:
 
 	//! test for a triangle, with edges
 	SIMD_FORCE_INLINE bool collide_triangle_exact(
-		const btVector3 & p1,
-		const btVector3 & p2,
-		const btVector3 & p3,
-		const btVector4 & triangle_plane) const
+		const btVector3 &p1,
+		const btVector3 &p2,
+		const btVector3 &p3,
+		const btVector4 &triangle_plane) const
 	{
-		if(!collide_plane(triangle_plane)) return false;
+		if (!collide_plane(triangle_plane)) return false;
 
-		btVector3 center,extends;
-		this->get_center_extend(center,extends);
+		btVector3 center, extends;
+		this->get_center_extend(center, extends);
 
 		const btVector3 v1(p1 - center);
 		const btVector3 v2(p2 - center);
@@ -599,47 +578,43 @@ public:
 		btVector3 diff(v2 - v1);
 		btVector3 abs_diff = diff.absolute();
 		//Test With X axis
-		TEST_CROSS_EDGE_BOX_X_AXIS_MCR(diff,abs_diff,v1,v3,extends);
+		TEST_CROSS_EDGE_BOX_X_AXIS_MCR(diff, abs_diff, v1, v3, extends);
 		//Test With Y axis
-		TEST_CROSS_EDGE_BOX_Y_AXIS_MCR(diff,abs_diff,v1,v3,extends);
+		TEST_CROSS_EDGE_BOX_Y_AXIS_MCR(diff, abs_diff, v1, v3, extends);
 		//Test With Z axis
-		TEST_CROSS_EDGE_BOX_Z_AXIS_MCR(diff,abs_diff,v1,v3,extends);
-
+		TEST_CROSS_EDGE_BOX_Z_AXIS_MCR(diff, abs_diff, v1, v3, extends);
 
 		diff = v3 - v2;
 		abs_diff = diff.absolute();
 		//Test With X axis
-		TEST_CROSS_EDGE_BOX_X_AXIS_MCR(diff,abs_diff,v2,v1,extends);
+		TEST_CROSS_EDGE_BOX_X_AXIS_MCR(diff, abs_diff, v2, v1, extends);
 		//Test With Y axis
-		TEST_CROSS_EDGE_BOX_Y_AXIS_MCR(diff,abs_diff,v2,v1,extends);
+		TEST_CROSS_EDGE_BOX_Y_AXIS_MCR(diff, abs_diff, v2, v1, extends);
 		//Test With Z axis
-		TEST_CROSS_EDGE_BOX_Z_AXIS_MCR(diff,abs_diff,v2,v1,extends);
+		TEST_CROSS_EDGE_BOX_Z_AXIS_MCR(diff, abs_diff, v2, v1, extends);
 
 		diff = v1 - v3;
 		abs_diff = diff.absolute();
 		//Test With X axis
-		TEST_CROSS_EDGE_BOX_X_AXIS_MCR(diff,abs_diff,v3,v2,extends);
+		TEST_CROSS_EDGE_BOX_X_AXIS_MCR(diff, abs_diff, v3, v2, extends);
 		//Test With Y axis
-		TEST_CROSS_EDGE_BOX_Y_AXIS_MCR(diff,abs_diff,v3,v2,extends);
+		TEST_CROSS_EDGE_BOX_Y_AXIS_MCR(diff, abs_diff, v3, v2, extends);
 		//Test With Z axis
-		TEST_CROSS_EDGE_BOX_Z_AXIS_MCR(diff,abs_diff,v3,v2,extends);
+		TEST_CROSS_EDGE_BOX_Z_AXIS_MCR(diff, abs_diff, v3, v2, extends);
 
 		return true;
 	}
 };
 
-
 //! Compairison of transformation objects
-SIMD_FORCE_INLINE bool btCompareTransformsEqual(const btTransform & t1,const btTransform & t2)
+SIMD_FORCE_INLINE bool btCompareTransformsEqual(const btTransform &t1, const btTransform &t2)
 {
-	if(!(t1.getOrigin() == t2.getOrigin()) ) return false;
+	if (!(t1.getOrigin() == t2.getOrigin())) return false;
 
-	if(!(t1.getBasis().getRow(0) == t2.getBasis().getRow(0)) ) return false;
-	if(!(t1.getBasis().getRow(1) == t2.getBasis().getRow(1)) ) return false;
-	if(!(t1.getBasis().getRow(2) == t2.getBasis().getRow(2)) ) return false;
+	if (!(t1.getBasis().getRow(0) == t2.getBasis().getRow(0))) return false;
+	if (!(t1.getBasis().getRow(1) == t2.getBasis().getRow(1))) return false;
+	if (!(t1.getBasis().getRow(2) == t2.getBasis().getRow(2))) return false;
 	return true;
 }
 
-
-
-#endif // GIM_BOX_COLLISION_H_INCLUDED
+#endif  // GIM_BOX_COLLISION_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btClipPolygon.h b/thirdparty/bullet/BulletCollision/Gimpact/btClipPolygon.h
index de0a5231baf..38c23e222de 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btClipPolygon.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btClipPolygon.h
@@ -27,77 +27,74 @@ subject to the following restrictions:
 #include "LinearMath/btTransform.h"
 #include "LinearMath/btGeometryUtil.h"
 
-
-SIMD_FORCE_INLINE btScalar bt_distance_point_plane(const btVector4 & plane,const btVector3 &point)
+SIMD_FORCE_INLINE btScalar bt_distance_point_plane(const btVector4 &plane, const btVector3 &point)
 {
 	return point.dot(plane) - plane[3];
 }
 
 /*! Vector blending
 Takes two vectors a, b, blends them together*/
-SIMD_FORCE_INLINE void bt_vec_blend(btVector3 &vr, const btVector3 &va,const btVector3 &vb, btScalar blend_factor)
+SIMD_FORCE_INLINE void bt_vec_blend(btVector3 &vr, const btVector3 &va, const btVector3 &vb, btScalar blend_factor)
 {
-	vr = (1-blend_factor)*va + blend_factor*vb;
+	vr = (1 - blend_factor) * va + blend_factor * vb;
 }
 
 //! This function calcs the distance from a 3D plane
 SIMD_FORCE_INLINE void bt_plane_clip_polygon_collect(
-						const btVector3 & point0,
-						const btVector3 & point1,
-						btScalar dist0,
-						btScalar dist1,
-						btVector3 * clipped,
-						int & clipped_count)
+	const btVector3 &point0,
+	const btVector3 &point1,
+	btScalar dist0,
+	btScalar dist1,
+	btVector3 *clipped,
+	int &clipped_count)
 {
-	bool _prevclassif = (dist0>SIMD_EPSILON);
-	bool _classif = (dist1>SIMD_EPSILON);
-	if(_classif!=_prevclassif)
+	bool _prevclassif = (dist0 > SIMD_EPSILON);
+	bool _classif = (dist1 > SIMD_EPSILON);
+	if (_classif != _prevclassif)
 	{
-		btScalar blendfactor = -dist0/(dist1-dist0);
-		bt_vec_blend(clipped[clipped_count],point0,point1,blendfactor);
+		btScalar blendfactor = -dist0 / (dist1 - dist0);
+		bt_vec_blend(clipped[clipped_count], point0, point1, blendfactor);
 		clipped_count++;
 	}
-	if(!_classif)
+	if (!_classif)
 	{
 		clipped[clipped_count] = point1;
 		clipped_count++;
 	}
 }
 
-
 //! Clips a polygon by a plane
 /*!
 *\return The count of the clipped counts
 */
 SIMD_FORCE_INLINE int bt_plane_clip_polygon(
-						const btVector4 & plane,
-						const btVector3 * polygon_points,
-						int polygon_point_count,
-						btVector3 * clipped)
+	const btVector4 &plane,
+	const btVector3 *polygon_points,
+	int polygon_point_count,
+	btVector3 *clipped)
 {
-    int clipped_count = 0;
+	int clipped_count = 0;
 
-
-    //clip first point
-	btScalar firstdist = bt_distance_point_plane(plane,polygon_points[0]);;
-	if(!(firstdist>SIMD_EPSILON))
+	//clip first point
+	btScalar firstdist = bt_distance_point_plane(plane, polygon_points[0]);
+	;
+	if (!(firstdist > SIMD_EPSILON))
 	{
 		clipped[clipped_count] = polygon_points[0];
 		clipped_count++;
 	}
 
 	btScalar olddist = firstdist;
-	for(int i=1;i<polygon_point_count;i++)
+	for (int i = 1; i < polygon_point_count; i++)
 	{
-		btScalar dist = bt_distance_point_plane(plane,polygon_points[i]);
+		btScalar dist = bt_distance_point_plane(plane, polygon_points[i]);
 
 		bt_plane_clip_polygon_collect(
-						polygon_points[i-1],polygon_points[i],
-						olddist,
-						dist,
-						clipped,
-						clipped_count);
-
+			polygon_points[i - 1], polygon_points[i],
+			olddist,
+			dist,
+			clipped,
+			clipped_count);
 
 		olddist = dist;
 	}
@@ -105,11 +102,11 @@ SIMD_FORCE_INLINE int bt_plane_clip_polygon(
 	//RETURN TO FIRST  point
 
 	bt_plane_clip_polygon_collect(
-					polygon_points[polygon_point_count-1],polygon_points[0],
-					olddist,
-					firstdist,
-					clipped,
-					clipped_count);
+		polygon_points[polygon_point_count - 1], polygon_points[0],
+		olddist,
+		firstdist,
+		clipped,
+		clipped_count);
 
 	return clipped_count;
 }
@@ -120,18 +117,19 @@ SIMD_FORCE_INLINE int bt_plane_clip_polygon(
 *\return The count of the clipped counts
 */
 SIMD_FORCE_INLINE int bt_plane_clip_triangle(
-						const btVector4 & plane,
-						const btVector3 & point0,
-						const btVector3 & point1,
-						const btVector3& point2,
-						btVector3 * clipped // an allocated array of 16 points at least
-						)
+	const btVector4 &plane,
+	const btVector3 &point0,
+	const btVector3 &point1,
+	const btVector3 &point2,
+	btVector3 *clipped  // an allocated array of 16 points at least
+)
 {
-    int clipped_count = 0;
+	int clipped_count = 0;
 
-    //clip first point0
-	btScalar firstdist = bt_distance_point_plane(plane,point0);;
-	if(!(firstdist>SIMD_EPSILON))
+	//clip first point0
+	btScalar firstdist = bt_distance_point_plane(plane, point0);
+	;
+	if (!(firstdist > SIMD_EPSILON))
 	{
 		clipped[clipped_count] = point0;
 		clipped_count++;
@@ -139,44 +137,37 @@ SIMD_FORCE_INLINE int bt_plane_clip_triangle(
 
 	// point 1
 	btScalar olddist = firstdist;
-	btScalar dist = bt_distance_point_plane(plane,point1);
+	btScalar dist = bt_distance_point_plane(plane, point1);
 
 	bt_plane_clip_polygon_collect(
-					point0,point1,
-					olddist,
-					dist,
-					clipped,
-					clipped_count);
+		point0, point1,
+		olddist,
+		dist,
+		clipped,
+		clipped_count);
 
 	olddist = dist;
 
-
 	// point 2
-	dist = bt_distance_point_plane(plane,point2);
+	dist = bt_distance_point_plane(plane, point2);
 
 	bt_plane_clip_polygon_collect(
-					point1,point2,
-					olddist,
-					dist,
-					clipped,
-					clipped_count);
+		point1, point2,
+		olddist,
+		dist,
+		clipped,
+		clipped_count);
 	olddist = dist;
 
-
-
 	//RETURN TO FIRST  point0
 	bt_plane_clip_polygon_collect(
-					point2,point0,
-					olddist,
-					firstdist,
-					clipped,
-					clipped_count);
+		point2, point0,
+		olddist,
+		firstdist,
+		clipped,
+		clipped_count);
 
 	return clipped_count;
 }
 
-
-
-
-
-#endif // GIM_TRI_COLLISION_H_INCLUDED
+#endif  // GIM_TRI_COLLISION_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btCompoundFromGimpact.h b/thirdparty/bullet/BulletCollision/Gimpact/btCompoundFromGimpact.h
index 19f7ecddd0e..ede59e8a57f 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btCompoundFromGimpact.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btCompoundFromGimpact.h
@@ -5,7 +5,8 @@
 #include "btGImpactShape.h"
 #include "BulletCollision/NarrowPhaseCollision/btRaycastCallback.h"
 
-ATTRIBUTE_ALIGNED16(class) btCompoundFromGimpactShape	: public btCompoundShape
+ATTRIBUTE_ALIGNED16(class)
+btCompoundFromGimpactShape : public btCompoundShape
 {
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
@@ -18,92 +19,87 @@ public:
 			delete m_children[i].m_childShape;
 		}
 	}
-
 };
 
 struct MyCallback : public btTriangleRaycastCallback
+{
+	int m_ignorePart;
+	int m_ignoreTriangleIndex;
+
+	MyCallback(const btVector3& from, const btVector3& to, int ignorePart, int ignoreTriangleIndex)
+		: btTriangleRaycastCallback(from, to),
+		  m_ignorePart(ignorePart),
+		  m_ignoreTriangleIndex(ignoreTriangleIndex)
+	{
+	}
+	virtual btScalar reportHit(const btVector3& hitNormalLocal, btScalar hitFraction, int partId, int triangleIndex)
+	{
+		if (partId != m_ignorePart || triangleIndex != m_ignoreTriangleIndex)
 		{
-			int	m_ignorePart;
-			int	m_ignoreTriangleIndex;
-			
+			if (hitFraction < m_hitFraction)
+				return hitFraction;
+		}
 
-			MyCallback(const btVector3& from, const btVector3& to, int ignorePart, int ignoreTriangleIndex)
-			:btTriangleRaycastCallback(from,to),
-			m_ignorePart(ignorePart),
-			m_ignoreTriangleIndex(ignoreTriangleIndex)
-			{
-				
-			}
-			virtual btScalar reportHit(const btVector3& hitNormalLocal, btScalar hitFraction, int partId, int triangleIndex)
-			{
-				if (partId!=m_ignorePart || triangleIndex!=m_ignoreTriangleIndex)
-				{
-					if (hitFraction < m_hitFraction)
-						return hitFraction;
-				}
+		return m_hitFraction;
+	}
+};
+struct MyInternalTriangleIndexCallback : public btInternalTriangleIndexCallback
+{
+	const btGImpactMeshShape* m_gimpactShape;
+	btCompoundShape* m_colShape;
+	btScalar m_depth;
 
-				return m_hitFraction;
-			}
-		};
-		struct MyInternalTriangleIndexCallback :public btInternalTriangleIndexCallback
+	MyInternalTriangleIndexCallback(btCompoundShape* colShape, const btGImpactMeshShape* meshShape, btScalar depth)
+		: m_colShape(colShape),
+		  m_gimpactShape(meshShape),
+		  m_depth(depth)
+	{
+	}
+
+	virtual void internalProcessTriangleIndex(btVector3* triangle, int partId, int triangleIndex)
+	{
+		btVector3 scale = m_gimpactShape->getLocalScaling();
+		btVector3 v0 = triangle[0] * scale;
+		btVector3 v1 = triangle[1] * scale;
+		btVector3 v2 = triangle[2] * scale;
+
+		btVector3 centroid = (v0 + v1 + v2) / 3;
+		btVector3 normal = (v1 - v0).cross(v2 - v0);
+		normal.normalize();
+		btVector3 rayFrom = centroid;
+		btVector3 rayTo = centroid - normal * m_depth;
+
+		MyCallback cb(rayFrom, rayTo, partId, triangleIndex);
+
+		m_gimpactShape->processAllTrianglesRay(&cb, rayFrom, rayTo);
+		if (cb.m_hitFraction < 1)
 		{
-			const btGImpactMeshShape*		m_gimpactShape;
-			btCompoundShape*			m_colShape;
-			btScalar	m_depth;
+			rayTo.setInterpolate3(cb.m_from, cb.m_to, cb.m_hitFraction);
+			//rayTo = cb.m_from;
+			//rayTo = rayTo.lerp(cb.m_to,cb.m_hitFraction);
+			//gDebugDraw.drawLine(tr(centroid),tr(centroid+normal),btVector3(1,0,0));
+		}
 
-			MyInternalTriangleIndexCallback (btCompoundShape* colShape, const btGImpactMeshShape* meshShape, btScalar depth)
-			:m_colShape(colShape),
-			m_gimpactShape(meshShape),
-			m_depth(depth)
-			{
-			}
-			
-			virtual void internalProcessTriangleIndex(btVector3* triangle,int partId,int  triangleIndex)
-			{
-				btVector3 scale = m_gimpactShape->getLocalScaling();
-				btVector3 v0=triangle[0]*scale;
-				btVector3 v1=triangle[1]*scale;
-				btVector3 v2=triangle[2]*scale;
-				
-				btVector3 centroid = (v0+v1+v2)/3;
-				btVector3 normal = (v1-v0).cross(v2-v0);
-				normal.normalize();
-				btVector3 rayFrom = centroid;
-				btVector3 rayTo = centroid-normal*m_depth;
-				
-				MyCallback cb(rayFrom,rayTo,partId,triangleIndex);
-				
-				m_gimpactShape->processAllTrianglesRay(&cb,rayFrom, rayTo);
-				if (cb.m_hitFraction<1)
-				{
-					rayTo.setInterpolate3(cb.m_from,cb.m_to,cb.m_hitFraction);
-					//rayTo = cb.m_from;
-					//rayTo = rayTo.lerp(cb.m_to,cb.m_hitFraction);
-					//gDebugDraw.drawLine(tr(centroid),tr(centroid+normal),btVector3(1,0,0));
-				}
-				
+		btBU_Simplex1to4* tet = new btBU_Simplex1to4(v0, v1, v2, rayTo);
+		btTransform ident;
+		ident.setIdentity();
+		m_colShape->addChildShape(ident, tet);
+	}
+};
 
-				
-				btBU_Simplex1to4* tet = new btBU_Simplex1to4(v0,v1,v2,rayTo);
-				btTransform ident;
-				ident.setIdentity();
-				m_colShape->addChildShape(ident,tet);
-			}
-		};
-		
-btCompoundShape*	btCreateCompoundFromGimpactShape(const btGImpactMeshShape* gimpactMesh, btScalar depth)
+btCompoundShape* btCreateCompoundFromGimpactShape(const btGImpactMeshShape* gimpactMesh, btScalar depth)
 {
 	btCompoundShape* colShape = new btCompoundFromGimpactShape();
-		
-		btTransform tr;
-		tr.setIdentity();
-		
-		MyInternalTriangleIndexCallback cb(colShape,gimpactMesh, depth);
-		btVector3 aabbMin,aabbMax;
-		gimpactMesh->getAabb(tr,aabbMin,aabbMax);
-		gimpactMesh->getMeshInterface()->InternalProcessAllTriangles(&cb,aabbMin,aabbMax);
 
-	return colShape;	
-}	
+	btTransform tr;
+	tr.setIdentity();
 
-#endif //BT_COMPOUND_FROM_GIMPACT
+	MyInternalTriangleIndexCallback cb(colShape, gimpactMesh, depth);
+	btVector3 aabbMin, aabbMax;
+	gimpactMesh->getAabb(tr, aabbMin, aabbMax);
+	gimpactMesh->getMeshInterface()->InternalProcessAllTriangles(&cb, aabbMin, aabbMax);
+
+	return colShape;
+}
+
+#endif  //BT_COMPOUND_FROM_GIMPACT
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btContactProcessing.cpp b/thirdparty/bullet/BulletCollision/Gimpact/btContactProcessing.cpp
index eed31d839f8..f2e3e18d610 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btContactProcessing.cpp
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btContactProcessing.cpp
@@ -27,54 +27,50 @@ struct CONTACT_KEY_TOKEN
 	unsigned int m_key;
 	int m_value;
 	CONTACT_KEY_TOKEN()
-    {
-    }
+	{
+	}
 
-    CONTACT_KEY_TOKEN(unsigned int key,int token)
-    {
-    	m_key = key;
-    	m_value =  token;
-    }
+	CONTACT_KEY_TOKEN(unsigned int key, int token)
+	{
+		m_key = key;
+		m_value = token;
+	}
 
-    CONTACT_KEY_TOKEN(const CONTACT_KEY_TOKEN& rtoken)
-    {
-    	m_key = rtoken.m_key;
-    	m_value = rtoken.m_value;
-    }
+	CONTACT_KEY_TOKEN(const CONTACT_KEY_TOKEN& rtoken)
+	{
+		m_key = rtoken.m_key;
+		m_value = rtoken.m_value;
+	}
 
-    inline bool operator <(const CONTACT_KEY_TOKEN& other) const
+	inline bool operator<(const CONTACT_KEY_TOKEN& other) const
 	{
 		return (m_key < other.m_key);
 	}
 
-	inline bool operator >(const CONTACT_KEY_TOKEN& other) const
+	inline bool operator>(const CONTACT_KEY_TOKEN& other) const
 	{
 		return (m_key > other.m_key);
 	}
-
 };
 
 class CONTACT_KEY_TOKEN_COMP
 {
-	public:
-
-		bool operator() ( const CONTACT_KEY_TOKEN& a, const CONTACT_KEY_TOKEN& b ) const
-		{
-			return ( a < b );
-		}
+public:
+	bool operator()(const CONTACT_KEY_TOKEN& a, const CONTACT_KEY_TOKEN& b) const
+	{
+		return (a < b);
+	}
 };
 
-
 void btContactArray::merge_contacts(
-	const btContactArray & contacts, bool normal_contact_average)
+	const btContactArray& contacts, bool normal_contact_average)
 {
 	clear();
 
 	int i;
-	if(contacts.size()==0) return;
+	if (contacts.size() == 0) return;
 
-
-	if(contacts.size()==1)
+	if (contacts.size() == 1)
 	{
 		push_back(contacts[0]);
 		return;
@@ -86,16 +82,16 @@ void btContactArray::merge_contacts(
 
 	//fill key contacts
 
-	for ( i = 0;i<contacts.size() ;i++ )
+	for (i = 0; i < contacts.size(); i++)
 	{
-		keycontacts.push_back(CONTACT_KEY_TOKEN(contacts[i].calc_key_contact(),i));
+		keycontacts.push_back(CONTACT_KEY_TOKEN(contacts[i].calc_key_contact(), i));
 	}
 
 	//sort keys
 	keycontacts.quickSort(CONTACT_KEY_TOKEN_COMP());
 
 	// Merge contacts
-	int coincident_count=0;
+	int coincident_count = 0;
 	btVector3 coincident_normals[MAX_COINCIDENT];
 
 	unsigned int last_key = keycontacts[0].m_key;
@@ -103,56 +99,56 @@ void btContactArray::merge_contacts(
 
 	push_back(contacts[keycontacts[0].m_value]);
 
-	GIM_CONTACT * pcontact = &(*this)[0];
+	GIM_CONTACT* pcontact = &(*this)[0];
 
-	for( i=1;i<keycontacts.size();i++)
+	for (i = 1; i < keycontacts.size(); i++)
 	{
-	    key = keycontacts[i].m_key;
-		const GIM_CONTACT * scontact = &contacts[keycontacts[i].m_value];
+		key = keycontacts[i].m_key;
+		const GIM_CONTACT* scontact = &contacts[keycontacts[i].m_value];
 
-		if(last_key ==  key)//same points
+		if (last_key == key)  //same points
 		{
 			//merge contact
-			if(pcontact->m_depth - CONTACT_DIFF_EPSILON > scontact->m_depth)//)
+			if (pcontact->m_depth - CONTACT_DIFF_EPSILON > scontact->m_depth)  //)
 			{
 				*pcontact = *scontact;
-                coincident_count = 0;
+				coincident_count = 0;
 			}
-			else if(normal_contact_average)
+			else if (normal_contact_average)
 			{
-				if(btFabs(pcontact->m_depth - scontact->m_depth)<CONTACT_DIFF_EPSILON)
-                {
-                    if(coincident_count<MAX_COINCIDENT)
-                    {
-                    	coincident_normals[coincident_count] = scontact->m_normal;
-                        coincident_count++;
-                    }
-                }
+				if (btFabs(pcontact->m_depth - scontact->m_depth) < CONTACT_DIFF_EPSILON)
+				{
+					if (coincident_count < MAX_COINCIDENT)
+					{
+						coincident_normals[coincident_count] = scontact->m_normal;
+						coincident_count++;
+					}
+				}
 			}
 		}
 		else
-		{//add new contact
+		{  //add new contact
 
-		    if(normal_contact_average && coincident_count>0)
-		    {
-		    	pcontact->interpolate_normals(coincident_normals,coincident_count);
-		        coincident_count = 0;
-		    }
+			if (normal_contact_average && coincident_count > 0)
+			{
+				pcontact->interpolate_normals(coincident_normals, coincident_count);
+				coincident_count = 0;
+			}
 
-		    push_back(*scontact);
-		    pcontact = &(*this)[this->size()-1];
-        }
+			push_back(*scontact);
+			pcontact = &(*this)[this->size() - 1];
+		}
 		last_key = key;
 	}
 }
 
-void btContactArray::merge_contacts_unique(const btContactArray & contacts)
+void btContactArray::merge_contacts_unique(const btContactArray& contacts)
 {
 	clear();
 
-	if(contacts.size()==0) return;
+	if (contacts.size() == 0) return;
 
-	if(contacts.size()==1)
+	if (contacts.size() == 1)
 	{
 		push_back(contacts[0]);
 		return;
@@ -160,14 +156,14 @@ void btContactArray::merge_contacts_unique(const btContactArray & contacts)
 
 	GIM_CONTACT average_contact = contacts[0];
 
-	for (int i=1;i<contacts.size() ;i++ )
+	for (int i = 1; i < contacts.size(); i++)
 	{
 		average_contact.m_point += contacts[i].m_point;
 		average_contact.m_normal += contacts[i].m_normal * contacts[i].m_depth;
 	}
 
 	//divide
-	btScalar divide_average = 1.0f/((btScalar)contacts.size());
+	btScalar divide_average = 1.0f / ((btScalar)contacts.size());
 
 	average_contact.m_point *= divide_average;
 
@@ -176,6 +172,4 @@ void btContactArray::merge_contacts_unique(const btContactArray & contacts)
 	average_contact.m_depth = average_contact.m_normal.length();
 
 	average_contact.m_normal /= average_contact.m_depth;
-
 }
-
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btContactProcessing.h b/thirdparty/bullet/BulletCollision/Gimpact/btContactProcessing.h
index d1027dbe67e..4ff09d7cddf 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btContactProcessing.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btContactProcessing.h
@@ -29,7 +29,7 @@ subject to the following restrictions:
 #include "btTriangleShapeEx.h"
 #include "btContactProcessingStructs.h"
 
-class btContactArray:public btAlignedObjectArray<GIM_CONTACT>
+class btContactArray : public btAlignedObjectArray<GIM_CONTACT>
 {
 public:
 	btContactArray()
@@ -38,28 +38,28 @@ public:
 	}
 
 	SIMD_FORCE_INLINE void push_contact(
-		const btVector3 &point,const btVector3 & normal,
+		const btVector3 &point, const btVector3 &normal,
 		btScalar depth, int feature1, int feature2)
 	{
-		push_back( GIM_CONTACT(point,normal,depth,feature1,feature2) );
+		push_back(GIM_CONTACT(point, normal, depth, feature1, feature2));
 	}
 
 	SIMD_FORCE_INLINE void push_triangle_contacts(
-		const GIM_TRIANGLE_CONTACT & tricontact,
-		int feature1,int feature2)
+		const GIM_TRIANGLE_CONTACT &tricontact,
+		int feature1, int feature2)
 	{
-		for(int i = 0;i<tricontact.m_point_count ;i++ )
+		for (int i = 0; i < tricontact.m_point_count; i++)
 		{
 			push_contact(
 				tricontact.m_points[i],
 				tricontact.m_separating_normal,
-				tricontact.m_penetration_depth,feature1,feature2);
+				tricontact.m_penetration_depth, feature1, feature2);
 		}
 	}
 
-	void merge_contacts(const btContactArray & contacts, bool normal_contact_average = true);
+	void merge_contacts(const btContactArray &contacts, bool normal_contact_average = true);
 
-	void merge_contacts_unique(const btContactArray & contacts);
+	void merge_contacts_unique(const btContactArray &contacts);
 };
 
-#endif // GIM_CONTACT_H_INCLUDED
+#endif  // GIM_CONTACT_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btContactProcessingStructs.h b/thirdparty/bullet/BulletCollision/Gimpact/btContactProcessingStructs.h
index efbc4a567ae..bc8a709246f 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btContactProcessingStructs.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btContactProcessingStructs.h
@@ -28,7 +28,6 @@ subject to the following restrictions:
 #include "LinearMath/btAlignedObjectArray.h"
 #include "btTriangleShapeEx.h"
 
-
 /**
 Configuration var for applying interpolation of  contact normals
 */
@@ -41,69 +40,66 @@ Configuration var for applying interpolation of  contact normals
 class GIM_CONTACT
 {
 public:
-    btVector3 m_point;
-    btVector3 m_normal;
-    btScalar m_depth;//Positive value indicates interpenetration
-    btScalar m_distance;//Padding not for use
-    int m_feature1;//Face number
-    int m_feature2;//Face number
+	btVector3 m_point;
+	btVector3 m_normal;
+	btScalar m_depth;     //Positive value indicates interpenetration
+	btScalar m_distance;  //Padding not for use
+	int m_feature1;       //Face number
+	int m_feature2;       //Face number
 public:
-    GIM_CONTACT()
-    {
-    }
+	GIM_CONTACT()
+	{
+	}
 
-    GIM_CONTACT(const GIM_CONTACT & contact):
-				m_point(contact.m_point),
-				m_normal(contact.m_normal),
-				m_depth(contact.m_depth),
-				m_feature1(contact.m_feature1),
-				m_feature2(contact.m_feature2)
-    {
-    }
+	GIM_CONTACT(const GIM_CONTACT &contact) : m_point(contact.m_point),
+											  m_normal(contact.m_normal),
+											  m_depth(contact.m_depth),
+											  m_feature1(contact.m_feature1),
+											  m_feature2(contact.m_feature2)
+	{
+	}
 
-    GIM_CONTACT(const btVector3 &point,const btVector3 & normal,
-    	 			btScalar depth, int feature1, int feature2):
-				m_point(point),
-				m_normal(normal),
-				m_depth(depth),
-				m_feature1(feature1),
-				m_feature2(feature2)
-    {
-    }
+	GIM_CONTACT(const btVector3 &point, const btVector3 &normal,
+				btScalar depth, int feature1, int feature2) : m_point(point),
+															  m_normal(normal),
+															  m_depth(depth),
+															  m_feature1(feature1),
+															  m_feature2(feature2)
+	{
+	}
 
 	//! Calcs key for coord classification
-    SIMD_FORCE_INLINE unsigned int calc_key_contact() const
-    {
-    	int _coords[] = {
-    		(int)(m_point[0]*1000.0f+1.0f),
-    		(int)(m_point[1]*1333.0f),
-    		(int)(m_point[2]*2133.0f+3.0f)};
-		unsigned int _hash=0;
+	SIMD_FORCE_INLINE unsigned int calc_key_contact() const
+	{
+		int _coords[] = {
+			(int)(m_point[0] * 1000.0f + 1.0f),
+			(int)(m_point[1] * 1333.0f),
+			(int)(m_point[2] * 2133.0f + 3.0f)};
+		unsigned int _hash = 0;
 		unsigned int *_uitmp = (unsigned int *)(&_coords[0]);
 		_hash = *_uitmp;
 		_uitmp++;
-		_hash += (*_uitmp)<<4;
+		_hash += (*_uitmp) << 4;
 		_uitmp++;
-		_hash += (*_uitmp)<<8;
+		_hash += (*_uitmp) << 8;
 		return _hash;
-    }
+	}
 
-    SIMD_FORCE_INLINE void interpolate_normals( btVector3 * normals,int normal_count)
-    {
-    	btVector3 vec_sum(m_normal);
-		for(int i=0;i<normal_count;i++)
+	SIMD_FORCE_INLINE void interpolate_normals(btVector3 *normals, int normal_count)
+	{
+		btVector3 vec_sum(m_normal);
+		for (int i = 0; i < normal_count; i++)
 		{
 			vec_sum += normals[i];
 		}
 
 		btScalar vec_sum_len = vec_sum.length2();
-		if(vec_sum_len <CONTACT_DIFF_EPSILON) return;
+		if (vec_sum_len < CONTACT_DIFF_EPSILON) return;
 
 		//GIM_INV_SQRT(vec_sum_len,vec_sum_len); // 1/sqrt(vec_sum_len)
 
-		m_normal = vec_sum/btSqrt(vec_sum_len);
-    }
-
+		m_normal = vec_sum / btSqrt(vec_sum_len);
+	}
 };
 
-#endif // BT_CONTACT_H_STRUCTS_INCLUDED
+#endif  // BT_CONTACT_H_STRUCTS_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactBvh.cpp b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactBvh.cpp
index 863233163a6..bb520e061d0 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactBvh.cpp
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactBvh.cpp
@@ -30,7 +30,6 @@ btClock g_tree_clock;
 float g_accum_tree_collision_time = 0;
 int g_count_traversing = 0;
 
-
 void bt_begin_gim02_tree_time()
 {
 	g_tree_clock.reset();
@@ -45,7 +44,7 @@ void bt_end_gim02_tree_time()
 //! Gets the average time in miliseconds of tree collisions
 float btGImpactBvh::getAverageTreeCollisionTime()
 {
-	if(g_count_traversing == 0) return 0;
+	if (g_count_traversing == 0) return 0;
 
 	float avgtime = g_accum_tree_collision_time;
 	avgtime /= (float)g_count_traversing;
@@ -54,80 +53,76 @@ float btGImpactBvh::getAverageTreeCollisionTime()
 	g_count_traversing = 0;
 	return avgtime;
 
-//	float avgtime = g_count_traversing;
-//	g_count_traversing = 0;
-//	return avgtime;
-
+	//	float avgtime = g_count_traversing;
+	//	g_count_traversing = 0;
+	//	return avgtime;
 }
 
-#endif //TRI_COLLISION_PROFILING
+#endif  //TRI_COLLISION_PROFILING
 
 /////////////////////// btBvhTree /////////////////////////////////
 
 int btBvhTree::_calc_splitting_axis(
-	GIM_BVH_DATA_ARRAY & primitive_boxes, int startIndex,  int endIndex)
+	GIM_BVH_DATA_ARRAY& primitive_boxes, int startIndex, int endIndex)
 {
-
 	int i;
 
-	btVector3 means(btScalar(0.),btScalar(0.),btScalar(0.));
-	btVector3 variance(btScalar(0.),btScalar(0.),btScalar(0.));
-	int numIndices = endIndex-startIndex;
+	btVector3 means(btScalar(0.), btScalar(0.), btScalar(0.));
+	btVector3 variance(btScalar(0.), btScalar(0.), btScalar(0.));
+	int numIndices = endIndex - startIndex;
 
-	for (i=startIndex;i<endIndex;i++)
+	for (i = startIndex; i < endIndex; i++)
 	{
-		btVector3 center = btScalar(0.5)*(primitive_boxes[i].m_bound.m_max +
-					 primitive_boxes[i].m_bound.m_min);
-		means+=center;
+		btVector3 center = btScalar(0.5) * (primitive_boxes[i].m_bound.m_max +
+											primitive_boxes[i].m_bound.m_min);
+		means += center;
 	}
-	means *= (btScalar(1.)/(btScalar)numIndices);
+	means *= (btScalar(1.) / (btScalar)numIndices);
 
-	for (i=startIndex;i<endIndex;i++)
+	for (i = startIndex; i < endIndex; i++)
 	{
-		btVector3 center = btScalar(0.5)*(primitive_boxes[i].m_bound.m_max +
-					 primitive_boxes[i].m_bound.m_min);
-		btVector3 diff2 = center-means;
+		btVector3 center = btScalar(0.5) * (primitive_boxes[i].m_bound.m_max +
+											primitive_boxes[i].m_bound.m_min);
+		btVector3 diff2 = center - means;
 		diff2 = diff2 * diff2;
 		variance += diff2;
 	}
-	variance *= (btScalar(1.)/	((btScalar)numIndices-1)	);
+	variance *= (btScalar(1.) / ((btScalar)numIndices - 1));
 
 	return variance.maxAxis();
 }
 
-
 int btBvhTree::_sort_and_calc_splitting_index(
-	GIM_BVH_DATA_ARRAY & primitive_boxes, int startIndex,
+	GIM_BVH_DATA_ARRAY& primitive_boxes, int startIndex,
 	int endIndex, int splitAxis)
 {
 	int i;
-	int splitIndex =startIndex;
+	int splitIndex = startIndex;
 	int numIndices = endIndex - startIndex;
 
 	// average of centers
 	btScalar splitValue = 0.0f;
 
-	btVector3 means(btScalar(0.),btScalar(0.),btScalar(0.));
-	for (i=startIndex;i<endIndex;i++)
+	btVector3 means(btScalar(0.), btScalar(0.), btScalar(0.));
+	for (i = startIndex; i < endIndex; i++)
 	{
-		btVector3 center = btScalar(0.5)*(primitive_boxes[i].m_bound.m_max +
-					 primitive_boxes[i].m_bound.m_min);
-		means+=center;
+		btVector3 center = btScalar(0.5) * (primitive_boxes[i].m_bound.m_max +
+											primitive_boxes[i].m_bound.m_min);
+		means += center;
 	}
-	means *= (btScalar(1.)/(btScalar)numIndices);
+	means *= (btScalar(1.) / (btScalar)numIndices);
 
 	splitValue = means[splitAxis];
 
-
 	//sort leafNodes so all values larger then splitValue comes first, and smaller values start from 'splitIndex'.
-	for (i=startIndex;i<endIndex;i++)
+	for (i = startIndex; i < endIndex; i++)
 	{
-		btVector3 center = btScalar(0.5)*(primitive_boxes[i].m_bound.m_max +
-					 primitive_boxes[i].m_bound.m_min);
+		btVector3 center = btScalar(0.5) * (primitive_boxes[i].m_bound.m_max +
+											primitive_boxes[i].m_bound.m_min);
 		if (center[splitAxis] > splitValue)
 		{
 			//swap
-			primitive_boxes.swap(i,splitIndex);
+			primitive_boxes.swap(i, splitIndex);
 			//swapLeafNodes(i,splitIndex);
 			splitIndex++;
 		}
@@ -142,32 +137,30 @@ int btBvhTree::_sort_and_calc_splitting_index(
 	//bool unbalanced2 = true;
 
 	//this should be safe too:
-	int rangeBalancedIndices = numIndices/3;
-	bool unbalanced = ((splitIndex<=(startIndex+rangeBalancedIndices)) || (splitIndex >=(endIndex-1-rangeBalancedIndices)));
+	int rangeBalancedIndices = numIndices / 3;
+	bool unbalanced = ((splitIndex <= (startIndex + rangeBalancedIndices)) || (splitIndex >= (endIndex - 1 - rangeBalancedIndices)));
 
 	if (unbalanced)
 	{
-		splitIndex = startIndex+ (numIndices>>1);
+		splitIndex = startIndex + (numIndices >> 1);
 	}
 
-	btAssert(!((splitIndex==startIndex) || (splitIndex == (endIndex))));
+	btAssert(!((splitIndex == startIndex) || (splitIndex == (endIndex))));
 
 	return splitIndex;
-
 }
 
-
-void btBvhTree::_build_sub_tree(GIM_BVH_DATA_ARRAY & primitive_boxes, int startIndex,  int endIndex)
+void btBvhTree::_build_sub_tree(GIM_BVH_DATA_ARRAY& primitive_boxes, int startIndex, int endIndex)
 {
 	int curIndex = m_num_nodes;
 	m_num_nodes++;
 
-	btAssert((endIndex-startIndex)>0);
+	btAssert((endIndex - startIndex) > 0);
 
-	if ((endIndex-startIndex)==1)
+	if ((endIndex - startIndex) == 1)
 	{
-	    //We have a leaf node
-	    setNodeBound(curIndex,primitive_boxes[startIndex].m_bound);
+		//We have a leaf node
+		setNodeBound(curIndex, primitive_boxes[startIndex].m_bound);
 		m_node_array[curIndex].setDataIndex(primitive_boxes[startIndex].m_data);
 
 		return;
@@ -175,47 +168,42 @@ void btBvhTree::_build_sub_tree(GIM_BVH_DATA_ARRAY & primitive_boxes, int startI
 	//calculate Best Splitting Axis and where to split it. Sort the incoming 'leafNodes' array within range 'startIndex/endIndex'.
 
 	//split axis
-	int splitIndex = _calc_splitting_axis(primitive_boxes,startIndex,endIndex);
+	int splitIndex = _calc_splitting_axis(primitive_boxes, startIndex, endIndex);
 
 	splitIndex = _sort_and_calc_splitting_index(
-			primitive_boxes,startIndex,endIndex,
-			splitIndex//split axis
-			);
-
+		primitive_boxes, startIndex, endIndex,
+		splitIndex  //split axis
+	);
 
 	//calc this node bounding box
 
 	btAABB node_bound;
 	node_bound.invalidate();
 
-	for (int i=startIndex;i<endIndex;i++)
+	for (int i = startIndex; i < endIndex; i++)
 	{
 		node_bound.merge(primitive_boxes[i].m_bound);
 	}
 
-	setNodeBound(curIndex,node_bound);
-
+	setNodeBound(curIndex, node_bound);
 
 	//build left branch
-	_build_sub_tree(primitive_boxes, startIndex, splitIndex );
-
+	_build_sub_tree(primitive_boxes, startIndex, splitIndex);
 
 	//build right branch
-	 _build_sub_tree(primitive_boxes, splitIndex ,endIndex);
+	_build_sub_tree(primitive_boxes, splitIndex, endIndex);
 
 	m_node_array[curIndex].setEscapeIndex(m_num_nodes - curIndex);
-
-
 }
 
 //! stackless build tree
 void btBvhTree::build_tree(
-	GIM_BVH_DATA_ARRAY & primitive_boxes)
+	GIM_BVH_DATA_ARRAY& primitive_boxes)
 {
 	// initialize node count to 0
 	m_num_nodes = 0;
 	// allocate nodes
-	m_node_array.resize(primitive_boxes.size()*2);
+	m_node_array.resize(primitive_boxes.size() * 2);
 
 	_build_sub_tree(primitive_boxes, 0, primitive_boxes.size());
 }
@@ -225,13 +213,13 @@ void btBvhTree::build_tree(
 void btGImpactBvh::refit()
 {
 	int nodecount = getNodeCount();
-	while(nodecount--)
+	while (nodecount--)
 	{
-		if(isLeafNode(nodecount))
+		if (isLeafNode(nodecount))
 		{
 			btAABB leafbox;
-			m_primitive_manager->get_primitive_box(getNodeData(nodecount),leafbox);
-			setNodeBound(nodecount,leafbox);
+			m_primitive_manager->get_primitive_box(getNodeData(nodecount), leafbox);
+			setNodeBound(nodecount, leafbox);
 		}
 		else
 		{
@@ -243,20 +231,20 @@ void btGImpactBvh::refit()
 			btAABB temp_box;
 
 			int child_node = getLeftNode(nodecount);
-			if(child_node)
+			if (child_node)
 			{
-				getNodeBound(child_node,temp_box);
+				getNodeBound(child_node, temp_box);
 				bound.merge(temp_box);
 			}
 
 			child_node = getRightNode(nodecount);
-			if(child_node)
+			if (child_node)
 			{
-				getNodeBound(child_node,temp_box);
+				getNodeBound(child_node, temp_box);
 				bound.merge(temp_box);
 			}
 
-			setNodeBound(nodecount,bound);
+			setNodeBound(nodecount, bound);
 		}
 	}
 }
@@ -268,17 +256,17 @@ void btGImpactBvh::buildSet()
 	GIM_BVH_DATA_ARRAY primitive_boxes;
 	primitive_boxes.resize(m_primitive_manager->get_primitive_count());
 
-	for (int i = 0;i<primitive_boxes.size() ;i++ )
+	for (int i = 0; i < primitive_boxes.size(); i++)
 	{
-		 m_primitive_manager->get_primitive_box(i,primitive_boxes[i].m_bound);
-		 primitive_boxes[i].m_data = i;
+		m_primitive_manager->get_primitive_box(i, primitive_boxes[i].m_bound);
+		primitive_boxes[i].m_data = i;
 	}
 
 	m_box_tree.build_tree(primitive_boxes);
 }
 
 //! returns the indices of the primitives in the m_primitive_manager
-bool btGImpactBvh::boxQuery(const btAABB & box, btAlignedObjectArray<int> & collided_results) const
+bool btGImpactBvh::boxQuery(const btAABB& box, btAlignedObjectArray<int>& collided_results) const
 {
 	int curIndex = 0;
 	int numNodes = getNodeCount();
@@ -286,7 +274,7 @@ bool btGImpactBvh::boxQuery(const btAABB & box, btAlignedObjectArray<int> & coll
 	while (curIndex < numNodes)
 	{
 		btAABB bound;
-		getNodeBound(curIndex,bound);
+		getNodeBound(curIndex, bound);
 
 		//catch bugs in tree data
 
@@ -306,19 +294,17 @@ bool btGImpactBvh::boxQuery(const btAABB & box, btAlignedObjectArray<int> & coll
 		else
 		{
 			//skip node
-			curIndex+= getEscapeNodeIndex(curIndex);
+			curIndex += getEscapeNodeIndex(curIndex);
 		}
 	}
-	if(collided_results.size()>0) return true;
+	if (collided_results.size() > 0) return true;
 	return false;
 }
 
-
-
 //! returns the indices of the primitives in the m_primitive_manager
 bool btGImpactBvh::rayQuery(
-	const btVector3 & ray_dir,const btVector3 & ray_origin ,
-	btAlignedObjectArray<int> & collided_results) const
+	const btVector3& ray_dir, const btVector3& ray_origin,
+	btAlignedObjectArray<int>& collided_results) const
 {
 	int curIndex = 0;
 	int numNodes = getNodeCount();
@@ -326,16 +312,16 @@ bool btGImpactBvh::rayQuery(
 	while (curIndex < numNodes)
 	{
 		btAABB bound;
-		getNodeBound(curIndex,bound);
+		getNodeBound(curIndex, bound);
 
 		//catch bugs in tree data
 
-		bool aabbOverlap = bound.collide_ray(ray_origin,ray_dir);
+		bool aabbOverlap = bound.collide_ray(ray_origin, ray_dir);
 		bool isleafnode = isLeafNode(curIndex);
 
 		if (isleafnode && aabbOverlap)
 		{
-			collided_results.push_back(getNodeData( curIndex));
+			collided_results.push_back(getNodeData(curIndex));
 		}
 
 		if (aabbOverlap || isleafnode)
@@ -346,153 +332,133 @@ bool btGImpactBvh::rayQuery(
 		else
 		{
 			//skip node
-			curIndex+= getEscapeNodeIndex(curIndex);
+			curIndex += getEscapeNodeIndex(curIndex);
 		}
 	}
-	if(collided_results.size()>0) return true;
+	if (collided_results.size() > 0) return true;
 	return false;
 }
 
-
 SIMD_FORCE_INLINE bool _node_collision(
-	btGImpactBvh * boxset0, btGImpactBvh * boxset1,
-	const BT_BOX_BOX_TRANSFORM_CACHE & trans_cache_1to0,
-	int node0 ,int node1, bool complete_primitive_tests)
+	btGImpactBvh* boxset0, btGImpactBvh* boxset1,
+	const BT_BOX_BOX_TRANSFORM_CACHE& trans_cache_1to0,
+	int node0, int node1, bool complete_primitive_tests)
 {
 	btAABB box0;
-	boxset0->getNodeBound(node0,box0);
+	boxset0->getNodeBound(node0, box0);
 	btAABB box1;
-	boxset1->getNodeBound(node1,box1);
-
-	return box0.overlapping_trans_cache(box1,trans_cache_1to0,complete_primitive_tests );
-//	box1.appy_transform_trans_cache(trans_cache_1to0);
-//	return box0.has_collision(box1);
+	boxset1->getNodeBound(node1, box1);
 
+	return box0.overlapping_trans_cache(box1, trans_cache_1to0, complete_primitive_tests);
+	//	box1.appy_transform_trans_cache(trans_cache_1to0);
+	//	return box0.has_collision(box1);
 }
 
-
 //stackless recursive collision routine
 static void _find_collision_pairs_recursive(
-	btGImpactBvh * boxset0, btGImpactBvh * boxset1,
-	btPairSet * collision_pairs,
-	const BT_BOX_BOX_TRANSFORM_CACHE & trans_cache_1to0,
+	btGImpactBvh* boxset0, btGImpactBvh* boxset1,
+	btPairSet* collision_pairs,
+	const BT_BOX_BOX_TRANSFORM_CACHE& trans_cache_1to0,
 	int node0, int node1, bool complete_primitive_tests)
 {
+	if (_node_collision(
+			boxset0, boxset1, trans_cache_1to0,
+			node0, node1, complete_primitive_tests) == false) return;  //avoid colliding internal nodes
 
-
-
-	if( _node_collision(
-		boxset0,boxset1,trans_cache_1to0,
-		node0,node1,complete_primitive_tests) ==false) return;//avoid colliding internal nodes
-
-	if(boxset0->isLeafNode(node0))
+	if (boxset0->isLeafNode(node0))
 	{
-		if(boxset1->isLeafNode(node1))
+		if (boxset1->isLeafNode(node1))
 		{
 			// collision result
 			collision_pairs->push_pair(
-				boxset0->getNodeData(node0),boxset1->getNodeData(node1));
+				boxset0->getNodeData(node0), boxset1->getNodeData(node1));
 			return;
 		}
 		else
 		{
-
 			//collide left recursive
 
 			_find_collision_pairs_recursive(
-								boxset0,boxset1,
-								collision_pairs,trans_cache_1to0,
-								node0,boxset1->getLeftNode(node1),false);
+				boxset0, boxset1,
+				collision_pairs, trans_cache_1to0,
+				node0, boxset1->getLeftNode(node1), false);
 
 			//collide right recursive
 			_find_collision_pairs_recursive(
-								boxset0,boxset1,
-								collision_pairs,trans_cache_1to0,
-								node0,boxset1->getRightNode(node1),false);
-
-
+				boxset0, boxset1,
+				collision_pairs, trans_cache_1to0,
+				node0, boxset1->getRightNode(node1), false);
 		}
 	}
 	else
 	{
-		if(boxset1->isLeafNode(node1))
+		if (boxset1->isLeafNode(node1))
 		{
-
 			//collide left recursive
 			_find_collision_pairs_recursive(
-								boxset0,boxset1,
-								collision_pairs,trans_cache_1to0,
-								boxset0->getLeftNode(node0),node1,false);
-
+				boxset0, boxset1,
+				collision_pairs, trans_cache_1to0,
+				boxset0->getLeftNode(node0), node1, false);
 
 			//collide right recursive
 
 			_find_collision_pairs_recursive(
-								boxset0,boxset1,
-								collision_pairs,trans_cache_1to0,
-								boxset0->getRightNode(node0),node1,false);
-
-
+				boxset0, boxset1,
+				collision_pairs, trans_cache_1to0,
+				boxset0->getRightNode(node0), node1, false);
 		}
 		else
 		{
 			//collide left0 left1
 
-
-
 			_find_collision_pairs_recursive(
-				boxset0,boxset1,
-				collision_pairs,trans_cache_1to0,
-				boxset0->getLeftNode(node0),boxset1->getLeftNode(node1),false);
+				boxset0, boxset1,
+				collision_pairs, trans_cache_1to0,
+				boxset0->getLeftNode(node0), boxset1->getLeftNode(node1), false);
 
 			//collide left0 right1
 
 			_find_collision_pairs_recursive(
-				boxset0,boxset1,
-				collision_pairs,trans_cache_1to0,
-				boxset0->getLeftNode(node0),boxset1->getRightNode(node1),false);
-
+				boxset0, boxset1,
+				collision_pairs, trans_cache_1to0,
+				boxset0->getLeftNode(node0), boxset1->getRightNode(node1), false);
 
 			//collide right0 left1
 
 			_find_collision_pairs_recursive(
-				boxset0,boxset1,
-				collision_pairs,trans_cache_1to0,
-				boxset0->getRightNode(node0),boxset1->getLeftNode(node1),false);
+				boxset0, boxset1,
+				collision_pairs, trans_cache_1to0,
+				boxset0->getRightNode(node0), boxset1->getLeftNode(node1), false);
 
 			//collide right0 right1
 
 			_find_collision_pairs_recursive(
-				boxset0,boxset1,
-				collision_pairs,trans_cache_1to0,
-				boxset0->getRightNode(node0),boxset1->getRightNode(node1),false);
+				boxset0, boxset1,
+				collision_pairs, trans_cache_1to0,
+				boxset0->getRightNode(node0), boxset1->getRightNode(node1), false);
 
-		}// else if node1 is not a leaf
-	}// else if node0 is not a leaf
+		}  // else if node1 is not a leaf
+	}      // else if node0 is not a leaf
 }
 
-
-void btGImpactBvh::find_collision(btGImpactBvh * boxset0, const btTransform & trans0,
-		btGImpactBvh * boxset1, const btTransform & trans1,
-		btPairSet & collision_pairs)
+void btGImpactBvh::find_collision(btGImpactBvh* boxset0, const btTransform& trans0,
+								  btGImpactBvh* boxset1, const btTransform& trans1,
+								  btPairSet& collision_pairs)
 {
-
-	if(boxset0->getNodeCount()==0 || boxset1->getNodeCount()==0 ) return;
+	if (boxset0->getNodeCount() == 0 || boxset1->getNodeCount() == 0) return;
 
 	BT_BOX_BOX_TRANSFORM_CACHE trans_cache_1to0;
 
-	trans_cache_1to0.calc_from_homogenic(trans0,trans1);
+	trans_cache_1to0.calc_from_homogenic(trans0, trans1);
 
 #ifdef TRI_COLLISION_PROFILING
 	bt_begin_gim02_tree_time();
-#endif //TRI_COLLISION_PROFILING
+#endif  //TRI_COLLISION_PROFILING
 
 	_find_collision_pairs_recursive(
-		boxset0,boxset1,
-		&collision_pairs,trans_cache_1to0,0,0,true);
+		boxset0, boxset1,
+		&collision_pairs, trans_cache_1to0, 0, 0, true);
 #ifdef TRI_COLLISION_PROFILING
 	bt_end_gim02_tree_time();
-#endif //TRI_COLLISION_PROFILING
-
+#endif  //TRI_COLLISION_PROFILING
 }
-
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactBvh.h b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactBvh.h
index e20e03cc1dc..3cd8fa24e77 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactBvh.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactBvh.h
@@ -1,5 +1,5 @@
-#ifndef GIM_BOX_SET_H_INCLUDED
-#define GIM_BOX_SET_H_INCLUDED
+#ifndef BT_GIMPACT_BVH_H_INCLUDED
+#define BT_GIMPACT_BVH_H_INCLUDED
 
 /*! \file gim_box_set.h
 \author Francisco Leon Najera
@@ -24,7 +24,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "LinearMath/btAlignedObjectArray.h"
 
 #include "btBoxCollision.h"
@@ -32,50 +31,48 @@ subject to the following restrictions:
 #include "btGImpactBvhStructs.h"
 
 //! A pairset array
-class btPairSet: public btAlignedObjectArray<GIM_PAIR>
+class btPairSet : public btAlignedObjectArray<GIM_PAIR>
 {
 public:
 	btPairSet()
 	{
 		reserve(32);
 	}
-	inline void push_pair(int index1,int index2)
+	inline void push_pair(int index1, int index2)
 	{
-		push_back(GIM_PAIR(index1,index2));
+		push_back(GIM_PAIR(index1, index2));
 	}
 
-	inline void push_pair_inv(int index1,int index2)
+	inline void push_pair_inv(int index1, int index2)
 	{
-		push_back(GIM_PAIR(index2,index1));
+		push_back(GIM_PAIR(index2, index1));
 	}
 };
 
-class GIM_BVH_DATA_ARRAY:public btAlignedObjectArray<GIM_BVH_DATA>
+class GIM_BVH_DATA_ARRAY : public btAlignedObjectArray<GIM_BVH_DATA>
 {
 };
 
-
-class GIM_BVH_TREE_NODE_ARRAY:public btAlignedObjectArray<GIM_BVH_TREE_NODE>
+class GIM_BVH_TREE_NODE_ARRAY : public btAlignedObjectArray<GIM_BVH_TREE_NODE>
 {
 };
 
-
-
-
 //! Basic Box tree structure
 class btBvhTree
 {
 protected:
 	int m_num_nodes;
 	GIM_BVH_TREE_NODE_ARRAY m_node_array;
+
 protected:
 	int _sort_and_calc_splitting_index(
-		GIM_BVH_DATA_ARRAY & primitive_boxes,
-		 int startIndex,  int endIndex, int splitAxis);
+		GIM_BVH_DATA_ARRAY& primitive_boxes,
+		int startIndex, int endIndex, int splitAxis);
 
-	int _calc_splitting_axis(GIM_BVH_DATA_ARRAY & primitive_boxes, int startIndex,  int endIndex);
+	int _calc_splitting_axis(GIM_BVH_DATA_ARRAY& primitive_boxes, int startIndex, int endIndex);
+
+	void _build_sub_tree(GIM_BVH_DATA_ARRAY& primitive_boxes, int startIndex, int endIndex);
 
-	void _build_sub_tree(GIM_BVH_DATA_ARRAY & primitive_boxes, int startIndex,  int endIndex);
 public:
 	btBvhTree()
 	{
@@ -84,7 +81,7 @@ public:
 
 	//! prototype functions for box tree management
 	//!@{
-	void build_tree(GIM_BVH_DATA_ARRAY & primitive_boxes);
+	void build_tree(GIM_BVH_DATA_ARRAY& primitive_boxes);
 
 	SIMD_FORCE_INLINE void clearNodes()
 	{
@@ -109,25 +106,25 @@ public:
 		return m_node_array[nodeindex].getDataIndex();
 	}
 
-	SIMD_FORCE_INLINE void getNodeBound(int nodeindex, btAABB & bound) const
+	SIMD_FORCE_INLINE void getNodeBound(int nodeindex, btAABB& bound) const
 	{
 		bound = m_node_array[nodeindex].m_bound;
 	}
 
-	SIMD_FORCE_INLINE void setNodeBound(int nodeindex, const btAABB & bound)
+	SIMD_FORCE_INLINE void setNodeBound(int nodeindex, const btAABB& bound)
 	{
 		m_node_array[nodeindex].m_bound = bound;
 	}
 
 	SIMD_FORCE_INLINE int getLeftNode(int nodeindex) const
 	{
-		return nodeindex+1;
+		return nodeindex + 1;
 	}
 
 	SIMD_FORCE_INLINE int getRightNode(int nodeindex) const
 	{
-		if(m_node_array[nodeindex+1].isLeafNode()) return nodeindex+2;
-		return nodeindex+1 + m_node_array[nodeindex+1].getEscapeIndex();
+		if (m_node_array[nodeindex + 1].isLeafNode()) return nodeindex + 2;
+		return nodeindex + 1 + m_node_array[nodeindex + 1].getEscapeIndex();
 	}
 
 	SIMD_FORCE_INLINE int getEscapeNodeIndex(int nodeindex) const
@@ -135,7 +132,7 @@ public:
 		return m_node_array[nodeindex].getEscapeIndex();
 	}
 
-	SIMD_FORCE_INLINE const GIM_BVH_TREE_NODE * get_node_pointer(int index = 0) const
+	SIMD_FORCE_INLINE const GIM_BVH_TREE_NODE* get_node_pointer(int index = 0) const
 	{
 		return &m_node_array[index];
 	}
@@ -143,7 +140,6 @@ public:
 	//!@}
 };
 
-
 //! Prototype Base class for primitive classification
 /*!
 This class is a wrapper for primitive collections.
@@ -153,18 +149,16 @@ This class can manage Compound shapes and trimeshes, and if it is managing trime
 class btPrimitiveManagerBase
 {
 public:
-
 	virtual ~btPrimitiveManagerBase() {}
 
 	//! determines if this manager consist on only triangles, which special case will be optimized
 	virtual bool is_trimesh() const = 0;
 	virtual int get_primitive_count() const = 0;
-	virtual void get_primitive_box(int prim_index ,btAABB & primbox) const = 0;
+	virtual void get_primitive_box(int prim_index, btAABB& primbox) const = 0;
 	//! retrieves only the points of the triangle, and the collision margin
-	virtual void get_primitive_triangle(int prim_index,btPrimitiveTriangle & triangle) const= 0;
+	virtual void get_primitive_triangle(int prim_index, btPrimitiveTriangle& triangle) const = 0;
 };
 
-
 //! Structure for containing Boxes
 /*!
 This class offers an structure for managing a box tree of primitives.
@@ -174,13 +168,13 @@ class btGImpactBvh
 {
 protected:
 	btBvhTree m_box_tree;
-	btPrimitiveManagerBase * m_primitive_manager;
+	btPrimitiveManagerBase* m_primitive_manager;
 
 protected:
 	//stackless refit
 	void refit();
-public:
 
+public:
 	//! this constructor doesn't build the tree. you must call	buildSet
 	btGImpactBvh()
 	{
@@ -188,31 +182,30 @@ public:
 	}
 
 	//! this constructor doesn't build the tree. you must call	buildSet
-	btGImpactBvh(btPrimitiveManagerBase * primitive_manager)
+	btGImpactBvh(btPrimitiveManagerBase* primitive_manager)
 	{
 		m_primitive_manager = primitive_manager;
 	}
 
-	SIMD_FORCE_INLINE btAABB getGlobalBox()  const
+	SIMD_FORCE_INLINE btAABB getGlobalBox() const
 	{
 		btAABB totalbox;
 		getNodeBound(0, totalbox);
 		return totalbox;
 	}
 
-	SIMD_FORCE_INLINE void setPrimitiveManager(btPrimitiveManagerBase * primitive_manager)
+	SIMD_FORCE_INLINE void setPrimitiveManager(btPrimitiveManagerBase* primitive_manager)
 	{
 		m_primitive_manager = primitive_manager;
 	}
 
-	SIMD_FORCE_INLINE btPrimitiveManagerBase * getPrimitiveManager() const
+	SIMD_FORCE_INLINE btPrimitiveManagerBase* getPrimitiveManager() const
 	{
 		return m_primitive_manager;
 	}
 
-
-//! node manager prototype functions
-///@{
+	//! node manager prototype functions
+	///@{
 
 	//! this attemps to refit the box set.
 	SIMD_FORCE_INLINE void update()
@@ -224,21 +217,21 @@ public:
 	void buildSet();
 
 	//! returns the indices of the primitives in the m_primitive_manager
-	bool boxQuery(const btAABB & box, btAlignedObjectArray<int> & collided_results) const;
+	bool boxQuery(const btAABB& box, btAlignedObjectArray<int>& collided_results) const;
 
 	//! returns the indices of the primitives in the m_primitive_manager
-	SIMD_FORCE_INLINE bool boxQueryTrans(const btAABB & box,
-		 const btTransform & transform, btAlignedObjectArray<int> & collided_results) const
+	SIMD_FORCE_INLINE bool boxQueryTrans(const btAABB& box,
+										 const btTransform& transform, btAlignedObjectArray<int>& collided_results) const
 	{
-		btAABB transbox=box;
+		btAABB transbox = box;
 		transbox.appy_transform(transform);
-		return boxQuery(transbox,collided_results);
+		return boxQuery(transbox, collided_results);
 	}
 
 	//! returns the indices of the primitives in the m_primitive_manager
 	bool rayQuery(
-		const btVector3 & ray_dir,const btVector3 & ray_origin ,
-		btAlignedObjectArray<int> & collided_results) const;
+		const btVector3& ray_dir, const btVector3& ray_origin,
+		btAlignedObjectArray<int>& collided_results) const;
 
 	//! tells if this set has hierarcht
 	SIMD_FORCE_INLINE bool hasHierarchy() const
@@ -247,7 +240,7 @@ public:
 	}
 
 	//! tells if this set is a trimesh
-	SIMD_FORCE_INLINE bool isTrimesh()  const
+	SIMD_FORCE_INLINE bool isTrimesh() const
 	{
 		return m_primitive_manager->is_trimesh();
 	}
@@ -269,17 +262,16 @@ public:
 		return m_box_tree.getNodeData(nodeindex);
 	}
 
-	SIMD_FORCE_INLINE void getNodeBound(int nodeindex, btAABB & bound)  const
+	SIMD_FORCE_INLINE void getNodeBound(int nodeindex, btAABB& bound) const
 	{
 		m_box_tree.getNodeBound(nodeindex, bound);
 	}
 
-	SIMD_FORCE_INLINE void setNodeBound(int nodeindex, const btAABB & bound)
+	SIMD_FORCE_INLINE void setNodeBound(int nodeindex, const btAABB& bound)
 	{
 		m_box_tree.setNodeBound(nodeindex, bound);
 	}
 
-
 	SIMD_FORCE_INLINE int getLeftNode(int nodeindex) const
 	{
 		return m_box_tree.getLeftNode(nodeindex);
@@ -295,24 +287,23 @@ public:
 		return m_box_tree.getEscapeNodeIndex(nodeindex);
 	}
 
-	SIMD_FORCE_INLINE void getNodeTriangle(int nodeindex,btPrimitiveTriangle & triangle) const
+	SIMD_FORCE_INLINE void getNodeTriangle(int nodeindex, btPrimitiveTriangle& triangle) const
 	{
-		m_primitive_manager->get_primitive_triangle(getNodeData(nodeindex),triangle);
+		m_primitive_manager->get_primitive_triangle(getNodeData(nodeindex), triangle);
 	}
 
-
-	SIMD_FORCE_INLINE const GIM_BVH_TREE_NODE * get_node_pointer(int index = 0) const
+	SIMD_FORCE_INLINE const GIM_BVH_TREE_NODE* get_node_pointer(int index = 0) const
 	{
 		return m_box_tree.get_node_pointer(index);
 	}
 
 #ifdef TRI_COLLISION_PROFILING
 	static float getAverageTreeCollisionTime();
-#endif //TRI_COLLISION_PROFILING
+#endif  //TRI_COLLISION_PROFILING
 
-	static void find_collision(btGImpactBvh * boxset1, const btTransform & trans1,
-		btGImpactBvh * boxset2, const btTransform & trans2,
-		btPairSet & collision_pairs);
+	static void find_collision(btGImpactBvh* boxset1, const btTransform& trans1,
+							   btGImpactBvh* boxset2, const btTransform& trans2,
+							   btPairSet& collision_pairs);
 };
 
-#endif // GIM_BOXPRUNING_H_INCLUDED
+#endif  // BT_GIMPACT_BVH_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactBvhStructs.h b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactBvhStructs.h
index 9342a572d09..54888c6757b 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactBvhStructs.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactBvhStructs.h
@@ -24,7 +24,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "LinearMath/btAlignedObjectArray.h"
 
 #include "btBoxCollision.h"
@@ -33,21 +32,22 @@ subject to the following restrictions:
 //! Overlapping pair
 struct GIM_PAIR
 {
-    int m_index1;
-    int m_index2;
-    GIM_PAIR()
-    {}
+	int m_index1;
+	int m_index2;
+	GIM_PAIR()
+	{
+	}
 
-    GIM_PAIR(const GIM_PAIR & p)
-    {
-    	m_index1 = p.m_index1;
-    	m_index2 = p.m_index2;
+	GIM_PAIR(const GIM_PAIR& p)
+	{
+		m_index1 = p.m_index1;
+		m_index2 = p.m_index2;
 	}
 
 	GIM_PAIR(int index1, int index2)
-    {
-    	m_index1 = index1;
-    	m_index2 = index2;
+	{
+		m_index1 = index1;
+		m_index2 = index2;
 	}
 };
 
@@ -63,8 +63,10 @@ class GIM_BVH_TREE_NODE
 {
 public:
 	btAABB m_bound;
+
 protected:
-	int	m_escapeIndexOrDataIndex;
+	int m_escapeIndexOrDataIndex;
+
 public:
 	GIM_BVH_TREE_NODE()
 	{
@@ -74,7 +76,7 @@ public:
 	SIMD_FORCE_INLINE bool isLeafNode() const
 	{
 		//skipindex is negative (internal node), triangleindex >=0 (leafnode)
-		return (m_escapeIndexOrDataIndex>=0);
+		return (m_escapeIndexOrDataIndex >= 0);
 	}
 
 	SIMD_FORCE_INLINE int getEscapeIndex() const
@@ -99,7 +101,6 @@ public:
 	{
 		m_escapeIndexOrDataIndex = index;
 	}
-
 };
 
-#endif // GIM_BOXPRUNING_H_INCLUDED
+#endif  // GIM_BOXPRUNING_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactCollisionAlgorithm.cpp b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactCollisionAlgorithm.cpp
index 2e87475e393..3d8ab9f5206 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactCollisionAlgorithm.cpp
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactCollisionAlgorithm.cpp
@@ -31,18 +31,16 @@ Concave-Concave Collision
 #include "btContactProcessing.h"
 #include "LinearMath/btQuickprof.h"
 
-
 //! Class for accessing the plane equation
 class btPlaneShape : public btStaticPlaneShape
 {
 public:
-
 	btPlaneShape(const btVector3& v, float f)
-		:btStaticPlaneShape(v,f)
+		: btStaticPlaneShape(v, f)
 	{
 	}
 
-	void get_plane_equation(btVector4 &equation)
+	void get_plane_equation(btVector4& equation)
 	{
 		equation[0] = m_planeNormal[0];
 		equation[1] = m_planeNormal[1];
@@ -50,18 +48,16 @@ public:
 		equation[3] = m_planeConstant;
 	}
 
-
-	void get_plane_equation_transformed(const btTransform & trans,btVector4 &equation) const 
+	void get_plane_equation_transformed(const btTransform& trans, btVector4& equation) const
 	{
-		equation[0] = trans.getBasis().getRow(0).dot(m_planeNormal);
-		equation[1] = trans.getBasis().getRow(1).dot(m_planeNormal);
-		equation[2] = trans.getBasis().getRow(2).dot(m_planeNormal);
-		equation[3] = trans.getOrigin().dot(m_planeNormal) + m_planeConstant;
+		const btVector3 normal = trans.getBasis() * m_planeNormal;
+		equation[0] = normal[0];
+		equation[1] = normal[1];
+		equation[2] = normal[2];
+		equation[3] = normal.dot(trans * (m_planeConstant * m_planeNormal));
 	}
 };
 
-
-
 //////////////////////////////////////////////////////////////////////////////////////////////
 #ifdef TRI_COLLISION_PROFILING
 
@@ -80,7 +76,7 @@ void bt_end_gim02_tri_time()
 	g_accum_triangle_collision_time += g_triangle_clock.getTimeMicroseconds();
 	g_count_triangle_collision++;
 }
-#endif //TRI_COLLISION_PROFILING
+#endif  //TRI_COLLISION_PROFILING
 //! Retrieving shapes shapes
 /*!
 Declared here due of insuficent space on Pool allocators
@@ -89,7 +85,7 @@ Declared here due of insuficent space on Pool allocators
 class GIM_ShapeRetriever
 {
 public:
-	const btGImpactShapeInterface * m_gim_shape;
+	const btGImpactShapeInterface* m_gim_shape;
 	btTriangleShapeEx m_trishape;
 	btTetrahedronShapeEx m_tetrashape;
 
@@ -97,51 +93,50 @@ public:
 	class ChildShapeRetriever
 	{
 	public:
-		GIM_ShapeRetriever * m_parent;
-		virtual const btCollisionShape * getChildShape(int index)
+		GIM_ShapeRetriever* m_parent;
+		virtual const btCollisionShape* getChildShape(int index)
 		{
 			return m_parent->m_gim_shape->getChildShape(index);
 		}
 		virtual ~ChildShapeRetriever() {}
 	};
 
-	class TriangleShapeRetriever:public ChildShapeRetriever
+	class TriangleShapeRetriever : public ChildShapeRetriever
 	{
 	public:
-
-		virtual btCollisionShape * getChildShape(int index)
+		virtual btCollisionShape* getChildShape(int index)
 		{
-			m_parent->m_gim_shape->getBulletTriangle(index,m_parent->m_trishape);
+			m_parent->m_gim_shape->getBulletTriangle(index, m_parent->m_trishape);
 			return &m_parent->m_trishape;
 		}
 		virtual ~TriangleShapeRetriever() {}
 	};
 
-	class TetraShapeRetriever:public ChildShapeRetriever
+	class TetraShapeRetriever : public ChildShapeRetriever
 	{
 	public:
-
-		virtual btCollisionShape * getChildShape(int index)
+		virtual btCollisionShape* getChildShape(int index)
 		{
-			m_parent->m_gim_shape->getBulletTetrahedron(index,m_parent->m_tetrashape);
+			m_parent->m_gim_shape->getBulletTetrahedron(index, m_parent->m_tetrashape);
 			return &m_parent->m_tetrashape;
 		}
 	};
+
 public:
 	ChildShapeRetriever m_child_retriever;
 	TriangleShapeRetriever m_tri_retriever;
-	TetraShapeRetriever  m_tetra_retriever;
-	ChildShapeRetriever * m_current_retriever;
+	TetraShapeRetriever m_tetra_retriever;
+	ChildShapeRetriever* m_current_retriever;
 
-	GIM_ShapeRetriever(const btGImpactShapeInterface * gim_shape)
+	GIM_ShapeRetriever(const btGImpactShapeInterface* gim_shape)
 	{
 		m_gim_shape = gim_shape;
 		//select retriever
-		if(m_gim_shape->needsRetrieveTriangles())
+		if (m_gim_shape->needsRetrieveTriangles())
 		{
 			m_current_retriever = &m_tri_retriever;
 		}
-		else if(m_gim_shape->needsRetrieveTetrahedrons())
+		else if (m_gim_shape->needsRetrieveTetrahedrons())
 		{
 			m_current_retriever = &m_tetra_retriever;
 		}
@@ -153,32 +148,26 @@ public:
 		m_current_retriever->m_parent = this;
 	}
 
-	const btCollisionShape * getChildShape(int index)
+	const btCollisionShape* getChildShape(int index)
 	{
 		return m_current_retriever->getChildShape(index);
 	}
-
-
 };
 
-
-
 //!@}
 
-
 #ifdef TRI_COLLISION_PROFILING
 
 //! Gets the average time in miliseconds of tree collisions
 float btGImpactCollisionAlgorithm::getAverageTreeCollisionTime()
 {
 	return btGImpactBoxSet::getAverageTreeCollisionTime();
-
 }
 
 //! Gets the average time in miliseconds of triangle collisions
 float btGImpactCollisionAlgorithm::getAverageTriangleCollisionTime()
 {
-	if(g_count_triangle_collision == 0) return 0;
+	if (g_count_triangle_collision == 0) return 0;
 
 	float avgtime = g_accum_triangle_collision_time;
 	avgtime /= (float)g_count_triangle_collision;
@@ -189,12 +178,10 @@ float btGImpactCollisionAlgorithm::getAverageTriangleCollisionTime()
 	return avgtime;
 }
 
-#endif //TRI_COLLISION_PROFILING
+#endif  //TRI_COLLISION_PROFILING
 
-
-
-btGImpactCollisionAlgorithm::btGImpactCollisionAlgorithm( const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
-: btActivatingCollisionAlgorithm(ci,body0Wrap,body1Wrap)
+btGImpactCollisionAlgorithm::btGImpactCollisionAlgorithm(const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
+	: btActivatingCollisionAlgorithm(ci, body0Wrap, body1Wrap)
 {
 	m_manifoldPtr = NULL;
 	m_convex_algorithm = NULL;
@@ -205,77 +192,62 @@ btGImpactCollisionAlgorithm::~btGImpactCollisionAlgorithm()
 	clearCache();
 }
 
-
-
-
-
-void btGImpactCollisionAlgorithm::addContactPoint(const btCollisionObjectWrapper * body0Wrap,
-				const btCollisionObjectWrapper * body1Wrap,
-				const btVector3 & point,
-				const btVector3 & normal,
-				btScalar distance)
+void btGImpactCollisionAlgorithm::addContactPoint(const btCollisionObjectWrapper* body0Wrap,
+												  const btCollisionObjectWrapper* body1Wrap,
+												  const btVector3& point,
+												  const btVector3& normal,
+												  btScalar distance)
 {
-	m_resultOut->setShapeIdentifiersA(m_part0,m_triface0);
-	m_resultOut->setShapeIdentifiersB(m_part1,m_triface1);
-	checkManifold(body0Wrap,body1Wrap);
-	m_resultOut->addContactPoint(normal,point,distance);
+	m_resultOut->setShapeIdentifiersA(m_part0, m_triface0);
+	m_resultOut->setShapeIdentifiersB(m_part1, m_triface1);
+	checkManifold(body0Wrap, body1Wrap);
+	m_resultOut->addContactPoint(normal, point, distance);
 }
 
-
 void btGImpactCollisionAlgorithm::shape_vs_shape_collision(
-					  const btCollisionObjectWrapper * body0Wrap,
-					  const btCollisionObjectWrapper* body1Wrap,
-					  const btCollisionShape * shape0,
-					  const btCollisionShape * shape1)
+	const btCollisionObjectWrapper* body0Wrap,
+	const btCollisionObjectWrapper* body1Wrap,
+	const btCollisionShape* shape0,
+	const btCollisionShape* shape1)
 {
-
-
 	{
-		
-		btCollisionAlgorithm* algor = newAlgorithm(body0Wrap,body1Wrap);
+		btCollisionAlgorithm* algor = newAlgorithm(body0Wrap, body1Wrap);
 		// post :	checkManifold is called
 
-		m_resultOut->setShapeIdentifiersA(m_part0,m_triface0);
-		m_resultOut->setShapeIdentifiersB(m_part1,m_triface1);
-		
-		algor->processCollision(body0Wrap,body1Wrap,*m_dispatchInfo,m_resultOut);
-		
+		m_resultOut->setShapeIdentifiersA(m_part0, m_triface0);
+		m_resultOut->setShapeIdentifiersB(m_part1, m_triface1);
+
+		algor->processCollision(body0Wrap, body1Wrap, *m_dispatchInfo, m_resultOut);
+
 		algor->~btCollisionAlgorithm();
 		m_dispatcher->freeCollisionAlgorithm(algor);
 	}
-
 }
 
 void btGImpactCollisionAlgorithm::convex_vs_convex_collision(
-					  const btCollisionObjectWrapper* body0Wrap,
-					  const btCollisionObjectWrapper* body1Wrap,
-					  const btCollisionShape* shape0,
-					  const btCollisionShape* shape1)
+	const btCollisionObjectWrapper* body0Wrap,
+	const btCollisionObjectWrapper* body1Wrap,
+	const btCollisionShape* shape0,
+	const btCollisionShape* shape1)
 {
+	m_resultOut->setShapeIdentifiersA(m_part0, m_triface0);
+	m_resultOut->setShapeIdentifiersB(m_part1, m_triface1);
 
-	m_resultOut->setShapeIdentifiersA(m_part0,m_triface0);
-	m_resultOut->setShapeIdentifiersB(m_part1,m_triface1);
-
-	btCollisionObjectWrapper ob0(body0Wrap,shape0,body0Wrap->getCollisionObject(),body0Wrap->getWorldTransform(),m_part0,m_triface0);
-	btCollisionObjectWrapper ob1(body1Wrap,shape1,body1Wrap->getCollisionObject(),body1Wrap->getWorldTransform(),m_part1,m_triface1);
-	checkConvexAlgorithm(&ob0,&ob1);
-	m_convex_algorithm->processCollision(&ob0,&ob1,*m_dispatchInfo,m_resultOut);
-
-
+	btCollisionObjectWrapper ob0(body0Wrap, shape0, body0Wrap->getCollisionObject(), body0Wrap->getWorldTransform(), m_part0, m_triface0);
+	btCollisionObjectWrapper ob1(body1Wrap, shape1, body1Wrap->getCollisionObject(), body1Wrap->getWorldTransform(), m_part1, m_triface1);
+	checkConvexAlgorithm(&ob0, &ob1);
+	m_convex_algorithm->processCollision(&ob0, &ob1, *m_dispatchInfo, m_resultOut);
 }
 
-
-
-
 void btGImpactCollisionAlgorithm::gimpact_vs_gimpact_find_pairs(
-					  const btTransform & trans0,
-					  const btTransform & trans1,
-					  const btGImpactShapeInterface * shape0,
-					  const btGImpactShapeInterface * shape1,btPairSet & pairset)
+	const btTransform& trans0,
+	const btTransform& trans1,
+	const btGImpactShapeInterface* shape0,
+	const btGImpactShapeInterface* shape1, btPairSet& pairset)
 {
-	if(shape0->hasBoxSet() && shape1->hasBoxSet())
+	if (shape0->hasBoxSet() && shape1->hasBoxSet())
 	{
-		btGImpactBoxSet::find_collision(shape0->getBoxSet(),trans0,shape1->getBoxSet(),trans1,pairset);
+		btGImpactBoxSet::find_collision(shape0->getBoxSet(), trans0, shape1->getBoxSet(), trans1, pairset);
 	}
 	else
 	{
@@ -283,74 +255,66 @@ void btGImpactCollisionAlgorithm::gimpact_vs_gimpact_find_pairs(
 		btAABB boxshape1;
 		int i = shape0->getNumChildShapes();
 
-		while(i--)
+		while (i--)
 		{
-			shape0->getChildAabb(i,trans0,boxshape0.m_min,boxshape0.m_max);
+			shape0->getChildAabb(i, trans0, boxshape0.m_min, boxshape0.m_max);
 
 			int j = shape1->getNumChildShapes();
-			while(j--)
+			while (j--)
 			{
-				shape1->getChildAabb(i,trans1,boxshape1.m_min,boxshape1.m_max);
+				shape1->getChildAabb(i, trans1, boxshape1.m_min, boxshape1.m_max);
 
-				if(boxshape1.has_collision(boxshape0))
+				if (boxshape1.has_collision(boxshape0))
 				{
-					pairset.push_pair(i,j);
+					pairset.push_pair(i, j);
 				}
 			}
 		}
 	}
-
-
 }
 
-
 void btGImpactCollisionAlgorithm::gimpact_vs_shape_find_pairs(
-					  const btTransform & trans0,
-					  const btTransform & trans1,
-					  const btGImpactShapeInterface * shape0,
-					  const btCollisionShape * shape1,
-					  btAlignedObjectArray<int> & collided_primitives)
+	const btTransform& trans0,
+	const btTransform& trans1,
+	const btGImpactShapeInterface* shape0,
+	const btCollisionShape* shape1,
+	btAlignedObjectArray<int>& collided_primitives)
 {
-
 	btAABB boxshape;
 
-
-	if(shape0->hasBoxSet())
+	if (shape0->hasBoxSet())
 	{
 		btTransform trans1to0 = trans0.inverse();
 		trans1to0 *= trans1;
 
-		shape1->getAabb(trans1to0,boxshape.m_min,boxshape.m_max);
+		shape1->getAabb(trans1to0, boxshape.m_min, boxshape.m_max);
 
 		shape0->getBoxSet()->boxQuery(boxshape, collided_primitives);
 	}
 	else
 	{
-		shape1->getAabb(trans1,boxshape.m_min,boxshape.m_max);
+		shape1->getAabb(trans1, boxshape.m_min, boxshape.m_max);
 
 		btAABB boxshape0;
 		int i = shape0->getNumChildShapes();
 
-		while(i--)
+		while (i--)
 		{
-			shape0->getChildAabb(i,trans0,boxshape0.m_min,boxshape0.m_max);
+			shape0->getChildAabb(i, trans0, boxshape0.m_min, boxshape0.m_max);
 
-			if(boxshape.has_collision(boxshape0))
+			if (boxshape.has_collision(boxshape0))
 			{
 				collided_primitives.push_back(i);
 			}
 		}
-
 	}
-
 }
 
-
-void btGImpactCollisionAlgorithm::collide_gjk_triangles(const btCollisionObjectWrapper * body0Wrap,
-				  const btCollisionObjectWrapper * body1Wrap,
-				  const btGImpactMeshShapePart * shape0,
-				  const btGImpactMeshShapePart * shape1,
-				  const int * pairs, int pair_count)
+void btGImpactCollisionAlgorithm::collide_gjk_triangles(const btCollisionObjectWrapper* body0Wrap,
+														const btCollisionObjectWrapper* body1Wrap,
+														const btGImpactMeshShapePart* shape0,
+														const btGImpactMeshShapePart* shape1,
+														const int* pairs, int pair_count)
 {
 	btTriangleShapeEx tri0;
 	btTriangleShapeEx tri1;
@@ -358,27 +322,22 @@ void btGImpactCollisionAlgorithm::collide_gjk_triangles(const btCollisionObjectW
 	shape0->lockChildShapes();
 	shape1->lockChildShapes();
 
-	const int * pair_pointer = pairs;
+	const int* pair_pointer = pairs;
 
-	while(pair_count--)
+	while (pair_count--)
 	{
-
 		m_triface0 = *(pair_pointer);
-		m_triface1 = *(pair_pointer+1);
-		pair_pointer+=2;
-
-
-
-		shape0->getBulletTriangle(m_triface0,tri0);
-		shape1->getBulletTriangle(m_triface1,tri1);
+		m_triface1 = *(pair_pointer + 1);
+		pair_pointer += 2;
 
+		shape0->getBulletTriangle(m_triface0, tri0);
+		shape1->getBulletTriangle(m_triface1, tri1);
 
 		//collide two convex shapes
-		if(tri0.overlap_test_conservative(tri1))
+		if (tri0.overlap_test_conservative(tri1))
 		{
-			convex_vs_convex_collision(body0Wrap,body1Wrap,&tri0,&tri1);
+			convex_vs_convex_collision(body0Wrap, body1Wrap, &tri0, &tri1);
 		}
-
 	}
 
 	shape0->unlockChildShapes();
@@ -386,10 +345,10 @@ void btGImpactCollisionAlgorithm::collide_gjk_triangles(const btCollisionObjectW
 }
 
 void btGImpactCollisionAlgorithm::collide_sat_triangles(const btCollisionObjectWrapper* body0Wrap,
-					  const btCollisionObjectWrapper* body1Wrap,
-					  const btGImpactMeshShapePart * shape0,
-					  const btGImpactMeshShapePart * shape1,
-					  const int * pairs, int pair_count)
+														const btCollisionObjectWrapper* body1Wrap,
+														const btGImpactMeshShapePart* shape0,
+														const btGImpactMeshShapePart* shape1,
+														const int* pairs, int pair_count)
 {
 	btTransform orgtrans0 = body0Wrap->getWorldTransform();
 	btTransform orgtrans1 = body1Wrap->getWorldTransform();
@@ -401,119 +360,105 @@ void btGImpactCollisionAlgorithm::collide_sat_triangles(const btCollisionObjectW
 	shape0->lockChildShapes();
 	shape1->lockChildShapes();
 
-	const int * pair_pointer = pairs;
+	const int* pair_pointer = pairs;
 
-	while(pair_count--)
+	while (pair_count--)
 	{
-
 		m_triface0 = *(pair_pointer);
-		m_triface1 = *(pair_pointer+1);
-		pair_pointer+=2;
+		m_triface1 = *(pair_pointer + 1);
+		pair_pointer += 2;
 
+		shape0->getPrimitiveTriangle(m_triface0, ptri0);
+		shape1->getPrimitiveTriangle(m_triface1, ptri1);
 
-		shape0->getPrimitiveTriangle(m_triface0,ptri0);
-		shape1->getPrimitiveTriangle(m_triface1,ptri1);
-
-		#ifdef TRI_COLLISION_PROFILING
+#ifdef TRI_COLLISION_PROFILING
 		bt_begin_gim02_tri_time();
-		#endif
+#endif
 
 		ptri0.applyTransform(orgtrans0);
 		ptri1.applyTransform(orgtrans1);
 
-
 		//build planes
 		ptri0.buildTriPlane();
 		ptri1.buildTriPlane();
 		// test conservative
 
-
-
-		if(ptri0.overlap_test_conservative(ptri1))
+		if (ptri0.overlap_test_conservative(ptri1))
 		{
-			if(ptri0.find_triangle_collision_clip_method(ptri1,contact_data))
+			if (ptri0.find_triangle_collision_clip_method(ptri1, contact_data))
 			{
-
 				int j = contact_data.m_point_count;
-				while(j--)
+				while (j--)
 				{
-
 					addContactPoint(body0Wrap, body1Wrap,
-								contact_data.m_points[j],
-								contact_data.m_separating_normal,
-								-contact_data.m_penetration_depth);
+									contact_data.m_points[j],
+									contact_data.m_separating_normal,
+									-contact_data.m_penetration_depth);
 				}
 			}
 		}
 
-		#ifdef TRI_COLLISION_PROFILING
+#ifdef TRI_COLLISION_PROFILING
 		bt_end_gim02_tri_time();
-		#endif
-
+#endif
 	}
 
 	shape0->unlockChildShapes();
 	shape1->unlockChildShapes();
-
 }
 
-
 void btGImpactCollisionAlgorithm::gimpact_vs_gimpact(
-						const btCollisionObjectWrapper* body0Wrap,
-					   	const btCollisionObjectWrapper * body1Wrap,
-					  	const btGImpactShapeInterface * shape0,
-					  	const btGImpactShapeInterface * shape1)
+	const btCollisionObjectWrapper* body0Wrap,
+	const btCollisionObjectWrapper* body1Wrap,
+	const btGImpactShapeInterface* shape0,
+	const btGImpactShapeInterface* shape1)
 {
-
-	if(shape0->getGImpactShapeType()==CONST_GIMPACT_TRIMESH_SHAPE)
+	if (shape0->getGImpactShapeType() == CONST_GIMPACT_TRIMESH_SHAPE)
 	{
-		const btGImpactMeshShape * meshshape0 = static_cast<const btGImpactMeshShape *>(shape0);
+		const btGImpactMeshShape* meshshape0 = static_cast<const btGImpactMeshShape*>(shape0);
 		m_part0 = meshshape0->getMeshPartCount();
 
-		while(m_part0--)
+		while (m_part0--)
 		{
-			gimpact_vs_gimpact(body0Wrap,body1Wrap,meshshape0->getMeshPart(m_part0),shape1);
+			gimpact_vs_gimpact(body0Wrap, body1Wrap, meshshape0->getMeshPart(m_part0), shape1);
 		}
 
 		return;
 	}
 
-	if(shape1->getGImpactShapeType()==CONST_GIMPACT_TRIMESH_SHAPE)
+	if (shape1->getGImpactShapeType() == CONST_GIMPACT_TRIMESH_SHAPE)
 	{
-		const btGImpactMeshShape * meshshape1 = static_cast<const btGImpactMeshShape *>(shape1);
+		const btGImpactMeshShape* meshshape1 = static_cast<const btGImpactMeshShape*>(shape1);
 		m_part1 = meshshape1->getMeshPartCount();
 
-		while(m_part1--)
+		while (m_part1--)
 		{
-
-			gimpact_vs_gimpact(body0Wrap,body1Wrap,shape0,meshshape1->getMeshPart(m_part1));
-
+			gimpact_vs_gimpact(body0Wrap, body1Wrap, shape0, meshshape1->getMeshPart(m_part1));
 		}
 
 		return;
 	}
 
-
 	btTransform orgtrans0 = body0Wrap->getWorldTransform();
 	btTransform orgtrans1 = body1Wrap->getWorldTransform();
 
 	btPairSet pairset;
 
-	gimpact_vs_gimpact_find_pairs(orgtrans0,orgtrans1,shape0,shape1,pairset);
+	gimpact_vs_gimpact_find_pairs(orgtrans0, orgtrans1, shape0, shape1, pairset);
 
-	if(pairset.size()== 0) return;
+	if (pairset.size() == 0) return;
 
-	if(shape0->getGImpactShapeType() == CONST_GIMPACT_TRIMESH_SHAPE_PART &&
+	if (shape0->getGImpactShapeType() == CONST_GIMPACT_TRIMESH_SHAPE_PART &&
 		shape1->getGImpactShapeType() == CONST_GIMPACT_TRIMESH_SHAPE_PART)
 	{
-		const btGImpactMeshShapePart * shapepart0 = static_cast<const btGImpactMeshShapePart * >(shape0);
-		const btGImpactMeshShapePart * shapepart1 = static_cast<const btGImpactMeshShapePart * >(shape1);
-		//specialized function
-		#ifdef BULLET_TRIANGLE_COLLISION
-		collide_gjk_triangles(body0Wrap,body1Wrap,shapepart0,shapepart1,&pairset[0].m_index1,pairset.size());
-		#else
-		collide_sat_triangles(body0Wrap,body1Wrap,shapepart0,shapepart1,&pairset[0].m_index1,pairset.size());
-		#endif
+		const btGImpactMeshShapePart* shapepart0 = static_cast<const btGImpactMeshShapePart*>(shape0);
+		const btGImpactMeshShapePart* shapepart1 = static_cast<const btGImpactMeshShapePart*>(shape1);
+//specialized function
+#ifdef BULLET_TRIANGLE_COLLISION
+		collide_gjk_triangles(body0Wrap, body1Wrap, shapepart0, shapepart1, &pairset[0].m_index1, pairset.size());
+#else
+		collide_sat_triangles(body0Wrap, body1Wrap, shapepart0, shapepart1, &pairset[0].m_index1, pairset.size());
+#endif
 
 		return;
 	}
@@ -530,32 +475,32 @@ void btGImpactCollisionAlgorithm::gimpact_vs_gimpact(
 	bool child_has_transform1 = shape1->childrenHasTransform();
 
 	int i = pairset.size();
-	while(i--)
+	while (i--)
 	{
-		GIM_PAIR * pair = &pairset[i];
+		GIM_PAIR* pair = &pairset[i];
 		m_triface0 = pair->m_index1;
 		m_triface1 = pair->m_index2;
-		const btCollisionShape * colshape0 = retriever0.getChildShape(m_triface0);
-		const btCollisionShape * colshape1 = retriever1.getChildShape(m_triface1);
+		const btCollisionShape* colshape0 = retriever0.getChildShape(m_triface0);
+		const btCollisionShape* colshape1 = retriever1.getChildShape(m_triface1);
 
 		btTransform tr0 = body0Wrap->getWorldTransform();
 		btTransform tr1 = body1Wrap->getWorldTransform();
 
-		if(child_has_transform0)
+		if (child_has_transform0)
 		{
-			tr0 = orgtrans0*shape0->getChildTransform(m_triface0);
+			tr0 = orgtrans0 * shape0->getChildTransform(m_triface0);
 		}
 
-		if(child_has_transform1)
+		if (child_has_transform1)
 		{
-			tr1 = orgtrans1*shape1->getChildTransform(m_triface1);
+			tr1 = orgtrans1 * shape1->getChildTransform(m_triface1);
 		}
 
-		btCollisionObjectWrapper ob0(body0Wrap,colshape0,body0Wrap->getCollisionObject(),tr0,m_part0,m_triface0);
-		btCollisionObjectWrapper ob1(body1Wrap,colshape1,body1Wrap->getCollisionObject(),tr1,m_part1,m_triface1);
+		btCollisionObjectWrapper ob0(body0Wrap, colshape0, body0Wrap->getCollisionObject(), tr0, m_part0, m_triface0);
+		btCollisionObjectWrapper ob1(body1Wrap, colshape1, body1Wrap->getCollisionObject(), tr1, m_part1, m_triface1);
 
 		//collide two convex shapes
-		convex_vs_convex_collision(&ob0,&ob1,colshape0,colshape1);
+		convex_vs_convex_collision(&ob0, &ob1, colshape0, colshape1);
 	}
 
 	shape0->unlockChildShapes();
@@ -563,159 +508,149 @@ void btGImpactCollisionAlgorithm::gimpact_vs_gimpact(
 }
 
 void btGImpactCollisionAlgorithm::gimpact_vs_shape(const btCollisionObjectWrapper* body0Wrap,
-				  const btCollisionObjectWrapper * body1Wrap,
-				  const btGImpactShapeInterface * shape0,
-				  const btCollisionShape * shape1,bool swapped)
+												   const btCollisionObjectWrapper* body1Wrap,
+												   const btGImpactShapeInterface* shape0,
+												   const btCollisionShape* shape1, bool swapped)
 {
-	if(shape0->getGImpactShapeType()==CONST_GIMPACT_TRIMESH_SHAPE)
+	if (shape0->getGImpactShapeType() == CONST_GIMPACT_TRIMESH_SHAPE)
 	{
-		const btGImpactMeshShape * meshshape0 = static_cast<const btGImpactMeshShape *>(shape0);
+		const btGImpactMeshShape* meshshape0 = static_cast<const btGImpactMeshShape*>(shape0);
 		int& part = swapped ? m_part1 : m_part0;
 		part = meshshape0->getMeshPartCount();
 
-		while(part--)
+		while (part--)
 		{
-
 			gimpact_vs_shape(body0Wrap,
-				  body1Wrap,
-				  meshshape0->getMeshPart(part),
-				  shape1,swapped);
-
+							 body1Wrap,
+							 meshshape0->getMeshPart(part),
+							 shape1, swapped);
 		}
 
 		return;
 	}
 
-	#ifdef GIMPACT_VS_PLANE_COLLISION
-	if(shape0->getGImpactShapeType() == CONST_GIMPACT_TRIMESH_SHAPE_PART &&
+#ifdef GIMPACT_VS_PLANE_COLLISION
+	if (shape0->getGImpactShapeType() == CONST_GIMPACT_TRIMESH_SHAPE_PART &&
 		shape1->getShapeType() == STATIC_PLANE_PROXYTYPE)
 	{
-		const btGImpactMeshShapePart * shapepart = static_cast<const btGImpactMeshShapePart *>(shape0);
-		const btStaticPlaneShape * planeshape = static_cast<const btStaticPlaneShape * >(shape1);
-		gimpacttrimeshpart_vs_plane_collision(body0Wrap,body1Wrap,shapepart,planeshape,swapped);
+		const btGImpactMeshShapePart* shapepart = static_cast<const btGImpactMeshShapePart*>(shape0);
+		const btStaticPlaneShape* planeshape = static_cast<const btStaticPlaneShape*>(shape1);
+		gimpacttrimeshpart_vs_plane_collision(body0Wrap, body1Wrap, shapepart, planeshape, swapped);
 		return;
 	}
 
-	#endif
+#endif
 
-
-
-	if(shape1->isCompound())
+	if (shape1->isCompound())
 	{
-		const btCompoundShape * compoundshape = static_cast<const btCompoundShape *>(shape1);
-		gimpact_vs_compoundshape(body0Wrap,body1Wrap,shape0,compoundshape,swapped);
+		const btCompoundShape* compoundshape = static_cast<const btCompoundShape*>(shape1);
+		gimpact_vs_compoundshape(body0Wrap, body1Wrap, shape0, compoundshape, swapped);
 		return;
 	}
-	else if(shape1->isConcave())
+	else if (shape1->isConcave())
 	{
-		const btConcaveShape * concaveshape = static_cast<const btConcaveShape *>(shape1);
-		gimpact_vs_concave(body0Wrap,body1Wrap,shape0,concaveshape,swapped);
+		const btConcaveShape* concaveshape = static_cast<const btConcaveShape*>(shape1);
+		gimpact_vs_concave(body0Wrap, body1Wrap, shape0, concaveshape, swapped);
 		return;
 	}
 
-
 	btTransform orgtrans0 = body0Wrap->getWorldTransform();
 
 	btTransform orgtrans1 = body1Wrap->getWorldTransform();
 
 	btAlignedObjectArray<int> collided_results;
 
-	gimpact_vs_shape_find_pairs(orgtrans0,orgtrans1,shape0,shape1,collided_results);
-
-	if(collided_results.size() == 0) return;
+	gimpact_vs_shape_find_pairs(orgtrans0, orgtrans1, shape0, shape1, collided_results);
 
+	if (collided_results.size() == 0) return;
 
 	shape0->lockChildShapes();
 
 	GIM_ShapeRetriever retriever0(shape0);
 
-
 	bool child_has_transform0 = shape0->childrenHasTransform();
 
-
 	int i = collided_results.size();
 
-	while(i--)
+	while (i--)
 	{
 		int child_index = collided_results[i];
-        if(swapped)
-    		m_triface1 = child_index;
-        else
-            m_triface0 = child_index;
+		if (swapped)
+			m_triface1 = child_index;
+		else
+			m_triface0 = child_index;
 
-		const btCollisionShape * colshape0 = retriever0.getChildShape(child_index);
+		const btCollisionShape* colshape0 = retriever0.getChildShape(child_index);
 
 		btTransform tr0 = body0Wrap->getWorldTransform();
 
-		if(child_has_transform0)
+		if (child_has_transform0)
 		{
-			tr0 = orgtrans0*shape0->getChildTransform(child_index);
+			tr0 = orgtrans0 * shape0->getChildTransform(child_index);
 		}
 
-		btCollisionObjectWrapper ob0(body0Wrap,colshape0,body0Wrap->getCollisionObject(),body0Wrap->getWorldTransform(),m_part0,m_triface0);
+		btCollisionObjectWrapper ob0(body0Wrap, colshape0, body0Wrap->getCollisionObject(), body0Wrap->getWorldTransform(), m_part0, m_triface0);
 		const btCollisionObjectWrapper* prevObj0 = m_resultOut->getBody0Wrap();
-		
-		if (m_resultOut->getBody0Wrap()->getCollisionObject()==ob0.getCollisionObject())
+
+		if (m_resultOut->getBody0Wrap()->getCollisionObject() == ob0.getCollisionObject())
 		{
 			m_resultOut->setBody0Wrap(&ob0);
-		} else
+		}
+		else
 		{
 			m_resultOut->setBody1Wrap(&ob0);
 		}
 
 		//collide two shapes
-		if(swapped)
+		if (swapped)
 		{
-			
-			shape_vs_shape_collision(body1Wrap,&ob0,shape1,colshape0);
+			shape_vs_shape_collision(body1Wrap, &ob0, shape1, colshape0);
 		}
 		else
 		{
-			
-			shape_vs_shape_collision(&ob0,body1Wrap,colshape0,shape1);
+			shape_vs_shape_collision(&ob0, body1Wrap, colshape0, shape1);
 		}
 		m_resultOut->setBody0Wrap(prevObj0);
-
 	}
 
 	shape0->unlockChildShapes();
-
 }
 
 void btGImpactCollisionAlgorithm::gimpact_vs_compoundshape(const btCollisionObjectWrapper* body0Wrap,
-				  const btCollisionObjectWrapper* body1Wrap,
-				  const btGImpactShapeInterface * shape0,
-				  const btCompoundShape * shape1,bool swapped)
+														   const btCollisionObjectWrapper* body1Wrap,
+														   const btGImpactShapeInterface* shape0,
+														   const btCompoundShape* shape1, bool swapped)
 {
 	btTransform orgtrans1 = body1Wrap->getWorldTransform();
 
 	int i = shape1->getNumChildShapes();
-	while(i--)
+	while (i--)
 	{
+		const btCollisionShape* colshape1 = shape1->getChildShape(i);
+		btTransform childtrans1 = orgtrans1 * shape1->getChildTransform(i);
 
-		const btCollisionShape * colshape1 = shape1->getChildShape(i);
-		btTransform childtrans1 = orgtrans1*shape1->getChildTransform(i);
+		btCollisionObjectWrapper ob1(body1Wrap, colshape1, body1Wrap->getCollisionObject(), childtrans1, -1, i);
 
-		btCollisionObjectWrapper ob1(body1Wrap,colshape1,body1Wrap->getCollisionObject(),childtrans1,-1,i);
-		
 		const btCollisionObjectWrapper* tmp = 0;
-		if (m_resultOut->getBody0Wrap()->getCollisionObject()==ob1.getCollisionObject())
+		if (m_resultOut->getBody0Wrap()->getCollisionObject() == ob1.getCollisionObject())
 		{
 			tmp = m_resultOut->getBody0Wrap();
 			m_resultOut->setBody0Wrap(&ob1);
-		} else
+		}
+		else
 		{
 			tmp = m_resultOut->getBody1Wrap();
 			m_resultOut->setBody1Wrap(&ob1);
 		}
 		//collide child shape
 		gimpact_vs_shape(body0Wrap, &ob1,
-					  shape0,colshape1,swapped);
+						 shape0, colshape1, swapped);
 
-		if (m_resultOut->getBody0Wrap()->getCollisionObject()==ob1.getCollisionObject())
+		if (m_resultOut->getBody0Wrap()->getCollisionObject() == ob1.getCollisionObject())
 		{
 			m_resultOut->setBody0Wrap(tmp);
-		} else
+		}
+		else
 		{
 			m_resultOut->setBody1Wrap(tmp);
 		}
@@ -723,27 +658,25 @@ void btGImpactCollisionAlgorithm::gimpact_vs_compoundshape(const btCollisionObje
 }
 
 void btGImpactCollisionAlgorithm::gimpacttrimeshpart_vs_plane_collision(
-					  const btCollisionObjectWrapper * body0Wrap,
-					  const btCollisionObjectWrapper * body1Wrap,
-					  const btGImpactMeshShapePart * shape0,
-					  const btStaticPlaneShape * shape1,bool swapped)
+	const btCollisionObjectWrapper* body0Wrap,
+	const btCollisionObjectWrapper* body1Wrap,
+	const btGImpactMeshShapePart* shape0,
+	const btStaticPlaneShape* shape1, bool swapped)
 {
-
-
 	btTransform orgtrans0 = body0Wrap->getWorldTransform();
 	btTransform orgtrans1 = body1Wrap->getWorldTransform();
 
-	const btPlaneShape * planeshape = static_cast<const btPlaneShape *>(shape1);
+	const btPlaneShape* planeshape = static_cast<const btPlaneShape*>(shape1);
 	btVector4 plane;
-	planeshape->get_plane_equation_transformed(orgtrans1,plane);
+	planeshape->get_plane_equation_transformed(orgtrans1, plane);
 
 	//test box against plane
 
 	btAABB tribox;
-	shape0->getAabb(orgtrans0,tribox.m_min,tribox.m_max);
+	shape0->getAabb(orgtrans0, tribox.m_min, tribox.m_max);
 	tribox.increment_margin(planeshape->getMargin());
 
-	if( tribox.plane_classify(plane)!= BT_CONST_COLLIDE_PLANE) return;
+	if (tribox.plane_classify(plane) != BT_CONST_COLLIDE_PLANE) return;
 
 	shape0->lockChildShapes();
 
@@ -751,28 +684,28 @@ void btGImpactCollisionAlgorithm::gimpacttrimeshpart_vs_plane_collision(
 
 	btVector3 vertex;
 	int vi = shape0->getVertexCount();
-	while(vi--)
+	while (vi--)
 	{
-		shape0->getVertex(vi,vertex);
+		shape0->getVertex(vi, vertex);
 		vertex = orgtrans0(vertex);
 
 		btScalar distance = vertex.dot(plane) - plane[3] - margin;
 
-		if(distance<0.0)//add contact
+		if (distance < 0.0)  //add contact
 		{
-			if(swapped)
+			if (swapped)
 			{
 				addContactPoint(body1Wrap, body0Wrap,
-					vertex,
-					-plane,
-					distance);
+								vertex,
+								-plane,
+								distance);
 			}
 			else
 			{
 				addContactPoint(body0Wrap, body1Wrap,
-					vertex,
-					plane,
-					distance);
+								vertex,
+								plane,
+								distance);
 			}
 		}
 	}
@@ -780,69 +713,64 @@ void btGImpactCollisionAlgorithm::gimpacttrimeshpart_vs_plane_collision(
 	shape0->unlockChildShapes();
 }
 
-
-
-
-class btGImpactTriangleCallback: public btTriangleCallback
+class btGImpactTriangleCallback : public btTriangleCallback
 {
 public:
-	btGImpactCollisionAlgorithm * algorithm;
-	const btCollisionObjectWrapper * body0Wrap;
-	const btCollisionObjectWrapper * body1Wrap;
-	const btGImpactShapeInterface * gimpactshape0;
+	btGImpactCollisionAlgorithm* algorithm;
+	const btCollisionObjectWrapper* body0Wrap;
+	const btCollisionObjectWrapper* body1Wrap;
+	const btGImpactShapeInterface* gimpactshape0;
 	bool swapped;
 	btScalar margin;
 
 	virtual void processTriangle(btVector3* triangle, int partId, int triangleIndex)
 	{
-		btTriangleShapeEx tri1(triangle[0],triangle[1],triangle[2]);
+		btTriangleShapeEx tri1(triangle[0], triangle[1], triangle[2]);
 		tri1.setMargin(margin);
-        if(swapped)
-        {
-            algorithm->setPart0(partId);
-            algorithm->setFace0(triangleIndex);
-        }
-        else
-        {
-            algorithm->setPart1(partId);
-            algorithm->setFace1(triangleIndex);
-        }
+		if (swapped)
+		{
+			algorithm->setPart0(partId);
+			algorithm->setFace0(triangleIndex);
+		}
+		else
+		{
+			algorithm->setPart1(partId);
+			algorithm->setFace1(triangleIndex);
+		}
 
-		btCollisionObjectWrapper ob1Wrap(body1Wrap,&tri1,body1Wrap->getCollisionObject(),body1Wrap->getWorldTransform(),partId,triangleIndex);
-		const btCollisionObjectWrapper * tmp = 0;
+		btCollisionObjectWrapper ob1Wrap(body1Wrap, &tri1, body1Wrap->getCollisionObject(), body1Wrap->getWorldTransform(), partId, triangleIndex);
+		const btCollisionObjectWrapper* tmp = 0;
 
-		if (algorithm->internalGetResultOut()->getBody0Wrap()->getCollisionObject()==ob1Wrap.getCollisionObject())
+		if (algorithm->internalGetResultOut()->getBody0Wrap()->getCollisionObject() == ob1Wrap.getCollisionObject())
 		{
 			tmp = algorithm->internalGetResultOut()->getBody0Wrap();
 			algorithm->internalGetResultOut()->setBody0Wrap(&ob1Wrap);
-		} else
+		}
+		else
 		{
 			tmp = algorithm->internalGetResultOut()->getBody1Wrap();
 			algorithm->internalGetResultOut()->setBody1Wrap(&ob1Wrap);
 		}
-		
-		algorithm->gimpact_vs_shape(
-							body0Wrap,&ob1Wrap,gimpactshape0,&tri1,swapped);
 
-		if (algorithm->internalGetResultOut()->getBody0Wrap()->getCollisionObject()==ob1Wrap.getCollisionObject())
+		algorithm->gimpact_vs_shape(
+			body0Wrap, &ob1Wrap, gimpactshape0, &tri1, swapped);
+
+		if (algorithm->internalGetResultOut()->getBody0Wrap()->getCollisionObject() == ob1Wrap.getCollisionObject())
 		{
 			algorithm->internalGetResultOut()->setBody0Wrap(tmp);
-		} else
+		}
+		else
 		{
 			algorithm->internalGetResultOut()->setBody1Wrap(tmp);
 		}
-
 	}
 };
 
-
-
-
 void btGImpactCollisionAlgorithm::gimpact_vs_concave(
-				  const btCollisionObjectWrapper* body0Wrap,
-				  const btCollisionObjectWrapper * body1Wrap,
-				  const btGImpactShapeInterface * shape0,
-				  const btConcaveShape * shape1,bool swapped)
+	const btCollisionObjectWrapper* body0Wrap,
+	const btCollisionObjectWrapper* body1Wrap,
+	const btGImpactShapeInterface* shape0,
+	const btConcaveShape* shape1, bool swapped)
 {
 	//create the callback
 	btGImpactTriangleCallback tricallback;
@@ -858,75 +786,71 @@ void btGImpactCollisionAlgorithm::gimpact_vs_concave(
 
 	gimpactInConcaveSpace = body1Wrap->getWorldTransform().inverse() * body0Wrap->getWorldTransform();
 
-	btVector3 minAABB,maxAABB;
-	shape0->getAabb(gimpactInConcaveSpace,minAABB,maxAABB);
-
-	shape1->processAllTriangles(&tricallback,minAABB,maxAABB);
+	btVector3 minAABB, maxAABB;
+	shape0->getAabb(gimpactInConcaveSpace, minAABB, maxAABB);
 
+	shape1->processAllTriangles(&tricallback, minAABB, maxAABB);
 }
 
-
-
-void btGImpactCollisionAlgorithm::processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+void btGImpactCollisionAlgorithm::processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
-    clearCache();
+	clearCache();
 
-    m_resultOut = resultOut;
+	m_resultOut = resultOut;
 	m_dispatchInfo = &dispatchInfo;
-    const btGImpactShapeInterface * gimpactshape0;
-    const btGImpactShapeInterface * gimpactshape1;
+	const btGImpactShapeInterface* gimpactshape0;
+	const btGImpactShapeInterface* gimpactshape1;
 
-	if (body0Wrap->getCollisionShape()->getShapeType()==GIMPACT_SHAPE_PROXYTYPE)
+	if (body0Wrap->getCollisionShape()->getShapeType() == GIMPACT_SHAPE_PROXYTYPE)
 	{
-		gimpactshape0 = static_cast<const btGImpactShapeInterface *>(body0Wrap->getCollisionShape());
+		gimpactshape0 = static_cast<const btGImpactShapeInterface*>(body0Wrap->getCollisionShape());
 
-		if( body1Wrap->getCollisionShape()->getShapeType()==GIMPACT_SHAPE_PROXYTYPE )
+		if (body1Wrap->getCollisionShape()->getShapeType() == GIMPACT_SHAPE_PROXYTYPE)
 		{
-			gimpactshape1 = static_cast<const btGImpactShapeInterface *>(body1Wrap->getCollisionShape());
+			gimpactshape1 = static_cast<const btGImpactShapeInterface*>(body1Wrap->getCollisionShape());
 
-			gimpact_vs_gimpact(body0Wrap,body1Wrap,gimpactshape0,gimpactshape1);
+			gimpact_vs_gimpact(body0Wrap, body1Wrap, gimpactshape0, gimpactshape1);
 		}
 		else
 		{
-			gimpact_vs_shape(body0Wrap,body1Wrap,gimpactshape0,body1Wrap->getCollisionShape(),false);
+			gimpact_vs_shape(body0Wrap, body1Wrap, gimpactshape0, body1Wrap->getCollisionShape(), false);
 		}
-
 	}
-	else if (body1Wrap->getCollisionShape()->getShapeType()==GIMPACT_SHAPE_PROXYTYPE )
+	else if (body1Wrap->getCollisionShape()->getShapeType() == GIMPACT_SHAPE_PROXYTYPE)
 	{
-		gimpactshape1 = static_cast<const btGImpactShapeInterface *>(body1Wrap->getCollisionShape());
+		gimpactshape1 = static_cast<const btGImpactShapeInterface*>(body1Wrap->getCollisionShape());
 
-		gimpact_vs_shape(body1Wrap,body0Wrap,gimpactshape1,body0Wrap->getCollisionShape(),true);
+		gimpact_vs_shape(body1Wrap, body0Wrap, gimpactshape1, body0Wrap->getCollisionShape(), true);
+	}
+
+	// Ensure that gContactProcessedCallback is called for concave shapes.
+	if (getLastManifold())
+	{
+		m_resultOut->refreshContactPoints();
 	}
 }
 
-
-btScalar btGImpactCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+btScalar btGImpactCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	return 1.f;
-
 }
 
 ///////////////////////////////////// REGISTERING ALGORITHM //////////////////////////////////////////////
 
-
-
 //! Use this function for register the algorithm externally
-void btGImpactCollisionAlgorithm::registerAlgorithm(btCollisionDispatcher * dispatcher)
+void btGImpactCollisionAlgorithm::registerAlgorithm(btCollisionDispatcher* dispatcher)
 {
-
 	static btGImpactCollisionAlgorithm::CreateFunc s_gimpact_cf;
 
 	int i;
 
-	for ( i = 0;i < MAX_BROADPHASE_COLLISION_TYPES ;i++ )
+	for (i = 0; i < MAX_BROADPHASE_COLLISION_TYPES; i++)
 	{
-		dispatcher->registerCollisionCreateFunc(GIMPACT_SHAPE_PROXYTYPE,i ,&s_gimpact_cf);
+		dispatcher->registerCollisionCreateFunc(GIMPACT_SHAPE_PROXYTYPE, i, &s_gimpact_cf);
 	}
 
-	for ( i = 0;i < MAX_BROADPHASE_COLLISION_TYPES ;i++ )
+	for (i = 0; i < MAX_BROADPHASE_COLLISION_TYPES; i++)
 	{
-		dispatcher->registerCollisionCreateFunc(i,GIMPACT_SHAPE_PROXYTYPE ,&s_gimpact_cf);
+		dispatcher->registerCollisionCreateFunc(i, GIMPACT_SHAPE_PROXYTYPE, &s_gimpact_cf);
 	}
-
 }
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactCollisionAlgorithm.h b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactCollisionAlgorithm.h
index 3e5675f729d..a368c8a0c0b 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactCollisionAlgorithm.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactCollisionAlgorithm.h
@@ -42,7 +42,6 @@ class btDispatcher;
 #include "LinearMath/btIDebugDraw.h"
 #include "BulletCollision/CollisionDispatch/btCollisionObjectWrapper.h"
 
-
 //! Collision Algorithm for GImpact Shapes
 /*!
 For register this algorithm in Bullet, proceed as following:
@@ -54,36 +53,35 @@ btGImpactCollisionAlgorithm::registerAlgorithm(dispatcher);
 class btGImpactCollisionAlgorithm : public btActivatingCollisionAlgorithm
 {
 protected:
-	btCollisionAlgorithm * m_convex_algorithm;
-    btPersistentManifold * m_manifoldPtr;
+	btCollisionAlgorithm* m_convex_algorithm;
+	btPersistentManifold* m_manifoldPtr;
 	btManifoldResult* m_resultOut;
-	const btDispatcherInfo * m_dispatchInfo;
+	const btDispatcherInfo* m_dispatchInfo;
 	int m_triface0;
 	int m_part0;
 	int m_triface1;
 	int m_part1;
 
-
 	//! Creates a new contact point
-	SIMD_FORCE_INLINE btPersistentManifold* newContactManifold(const btCollisionObject* body0,const btCollisionObject* body1)
+	SIMD_FORCE_INLINE btPersistentManifold* newContactManifold(const btCollisionObject* body0, const btCollisionObject* body1)
 	{
-		m_manifoldPtr = m_dispatcher->getNewManifold(body0,body1);
+		m_manifoldPtr = m_dispatcher->getNewManifold(body0, body1);
 		return m_manifoldPtr;
 	}
 
 	SIMD_FORCE_INLINE void destroyConvexAlgorithm()
 	{
-		if(m_convex_algorithm)
+		if (m_convex_algorithm)
 		{
 			m_convex_algorithm->~btCollisionAlgorithm();
-			m_dispatcher->freeCollisionAlgorithm( m_convex_algorithm);
+			m_dispatcher->freeCollisionAlgorithm(m_convex_algorithm);
 			m_convex_algorithm = NULL;
 		}
 	}
 
 	SIMD_FORCE_INLINE void destroyContactManifolds()
 	{
-		if(m_manifoldPtr == NULL) return;
+		if (m_manifoldPtr == NULL) return;
 		m_dispatcher->releaseManifold(m_manifoldPtr);
 		m_manifoldPtr = NULL;
 	}
@@ -104,207 +102,187 @@ protected:
 		return m_manifoldPtr;
 	}
 
-
 	// Call before process collision
-	SIMD_FORCE_INLINE void checkManifold(const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+	SIMD_FORCE_INLINE void checkManifold(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 	{
-		if(getLastManifold() == 0)
+		if (getLastManifold() == 0)
 		{
-			newContactManifold(body0Wrap->getCollisionObject(),body1Wrap->getCollisionObject());
+			newContactManifold(body0Wrap->getCollisionObject(), body1Wrap->getCollisionObject());
 		}
 
 		m_resultOut->setPersistentManifold(getLastManifold());
 	}
 
 	// Call before process collision
-	SIMD_FORCE_INLINE btCollisionAlgorithm * newAlgorithm(const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+	SIMD_FORCE_INLINE btCollisionAlgorithm* newAlgorithm(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 	{
-		checkManifold(body0Wrap,body1Wrap);
+		checkManifold(body0Wrap, body1Wrap);
 
-		btCollisionAlgorithm * convex_algorithm = m_dispatcher->findAlgorithm(
-				body0Wrap,body1Wrap,getLastManifold(), BT_CONTACT_POINT_ALGORITHMS);
-		return convex_algorithm ;
+		btCollisionAlgorithm* convex_algorithm = m_dispatcher->findAlgorithm(
+			body0Wrap, body1Wrap, getLastManifold(), BT_CONTACT_POINT_ALGORITHMS);
+		return convex_algorithm;
 	}
 
 	// Call before process collision
-	SIMD_FORCE_INLINE void checkConvexAlgorithm(const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+	SIMD_FORCE_INLINE void checkConvexAlgorithm(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 	{
-		if(m_convex_algorithm) return;
-		m_convex_algorithm = newAlgorithm(body0Wrap,body1Wrap);
+		if (m_convex_algorithm) return;
+		m_convex_algorithm = newAlgorithm(body0Wrap, body1Wrap);
 	}
 
+	void addContactPoint(const btCollisionObjectWrapper* body0Wrap,
+						 const btCollisionObjectWrapper* body1Wrap,
+						 const btVector3& point,
+						 const btVector3& normal,
+						 btScalar distance);
 
-
-
-	void addContactPoint(const btCollisionObjectWrapper * body0Wrap,
-					const btCollisionObjectWrapper * body1Wrap,
-					const btVector3 & point,
-					const btVector3 & normal,
-					btScalar distance);
-
-//! Collision routines
-//!@{
+	//! Collision routines
+	//!@{
 
 	void collide_gjk_triangles(const btCollisionObjectWrapper* body0Wrap,
-				  const btCollisionObjectWrapper* body1Wrap,
-				  const btGImpactMeshShapePart * shape0,
-				  const btGImpactMeshShapePart * shape1,
-				  const int * pairs, int pair_count);
+							   const btCollisionObjectWrapper* body1Wrap,
+							   const btGImpactMeshShapePart* shape0,
+							   const btGImpactMeshShapePart* shape1,
+							   const int* pairs, int pair_count);
 
 	void collide_sat_triangles(const btCollisionObjectWrapper* body0Wrap,
-					  const btCollisionObjectWrapper* body1Wrap,
-					  const btGImpactMeshShapePart * shape0,
-					  const btGImpactMeshShapePart * shape1,
-					  const int * pairs, int pair_count);
-
-
-
+							   const btCollisionObjectWrapper* body1Wrap,
+							   const btGImpactMeshShapePart* shape0,
+							   const btGImpactMeshShapePart* shape1,
+							   const int* pairs, int pair_count);
 
 	void shape_vs_shape_collision(
-					  const btCollisionObjectWrapper* body0,
-					  const btCollisionObjectWrapper* body1,
-					  const btCollisionShape * shape0,
-					  const btCollisionShape * shape1);
+		const btCollisionObjectWrapper* body0,
+		const btCollisionObjectWrapper* body1,
+		const btCollisionShape* shape0,
+		const btCollisionShape* shape1);
 
 	void convex_vs_convex_collision(const btCollisionObjectWrapper* body0Wrap,
-					  const btCollisionObjectWrapper* body1Wrap,
-					  const btCollisionShape* shape0,
-					  const btCollisionShape* shape1);
-
-
+									const btCollisionObjectWrapper* body1Wrap,
+									const btCollisionShape* shape0,
+									const btCollisionShape* shape1);
 
 	void gimpact_vs_gimpact_find_pairs(
-					  const btTransform & trans0,
-					  const btTransform & trans1,
-					  const btGImpactShapeInterface * shape0,
-					  const btGImpactShapeInterface * shape1,btPairSet & pairset);
+		const btTransform& trans0,
+		const btTransform& trans1,
+		const btGImpactShapeInterface* shape0,
+		const btGImpactShapeInterface* shape1, btPairSet& pairset);
 
 	void gimpact_vs_shape_find_pairs(
-					  const btTransform & trans0,
-					  const btTransform & trans1,
-					  const btGImpactShapeInterface * shape0,
-					  const btCollisionShape * shape1,
-					  btAlignedObjectArray<int> & collided_primitives);
-
+		const btTransform& trans0,
+		const btTransform& trans1,
+		const btGImpactShapeInterface* shape0,
+		const btCollisionShape* shape1,
+		btAlignedObjectArray<int>& collided_primitives);
 
 	void gimpacttrimeshpart_vs_plane_collision(
-					  const btCollisionObjectWrapper * body0Wrap,
-					  const btCollisionObjectWrapper * body1Wrap,
-					  const btGImpactMeshShapePart * shape0,
-					  const btStaticPlaneShape * shape1,bool swapped);
-
+		const btCollisionObjectWrapper* body0Wrap,
+		const btCollisionObjectWrapper* body1Wrap,
+		const btGImpactMeshShapePart* shape0,
+		const btStaticPlaneShape* shape1, bool swapped);
 
 public:
-
-	btGImpactCollisionAlgorithm( const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap);
+	btGImpactCollisionAlgorithm(const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap);
 
 	virtual ~btGImpactCollisionAlgorithm();
 
-	virtual void processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual void processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	btScalar	calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	btScalar calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual	void	getAllContactManifolds(btManifoldArray&	manifoldArray)
+	virtual void getAllContactManifolds(btManifoldArray& manifoldArray)
 	{
 		if (m_manifoldPtr)
 			manifoldArray.push_back(m_manifoldPtr);
 	}
 
-	btManifoldResult*	internalGetResultOut()
+	btManifoldResult* internalGetResultOut()
 	{
 		return m_resultOut;
 	}
 
-	struct CreateFunc :public 	btCollisionAlgorithmCreateFunc
+	struct CreateFunc : public btCollisionAlgorithmCreateFunc
 	{
-		virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+		virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 		{
 			void* mem = ci.m_dispatcher1->allocateCollisionAlgorithm(sizeof(btGImpactCollisionAlgorithm));
-			return new(mem) btGImpactCollisionAlgorithm(ci,body0Wrap,body1Wrap);
+			return new (mem) btGImpactCollisionAlgorithm(ci, body0Wrap, body1Wrap);
 		}
 	};
 
 	//! Use this function for register the algorithm externally
-	static void registerAlgorithm(btCollisionDispatcher * dispatcher);
+	static void registerAlgorithm(btCollisionDispatcher* dispatcher);
 #ifdef TRI_COLLISION_PROFILING
 	//! Gets the average time in miliseconds of tree collisions
 	static float getAverageTreeCollisionTime();
 
 	//! Gets the average time in miliseconds of triangle collisions
 	static float getAverageTriangleCollisionTime();
-#endif //TRI_COLLISION_PROFILING
+#endif  //TRI_COLLISION_PROFILING
 
 	//! Collides two gimpact shapes
 	/*!
 	\pre shape0 and shape1 couldn't be btGImpactMeshShape objects
 	*/
 
-
 	void gimpact_vs_gimpact(const btCollisionObjectWrapper* body0Wrap,
-					  const btCollisionObjectWrapper * body1Wrap,
-					  const btGImpactShapeInterface * shape0,
-					  const btGImpactShapeInterface * shape1);
+							const btCollisionObjectWrapper* body1Wrap,
+							const btGImpactShapeInterface* shape0,
+							const btGImpactShapeInterface* shape1);
 
 	void gimpact_vs_shape(const btCollisionObjectWrapper* body0Wrap,
-					  const btCollisionObjectWrapper* body1Wrap,
-					  const btGImpactShapeInterface * shape0,
-					  const btCollisionShape * shape1,bool swapped);
+						  const btCollisionObjectWrapper* body1Wrap,
+						  const btGImpactShapeInterface* shape0,
+						  const btCollisionShape* shape1, bool swapped);
 
-	void gimpact_vs_compoundshape(const btCollisionObjectWrapper * body0Wrap,
-					  const btCollisionObjectWrapper * body1Wrap,
-					  const btGImpactShapeInterface * shape0,
-					  const btCompoundShape * shape1,bool swapped);
+	void gimpact_vs_compoundshape(const btCollisionObjectWrapper* body0Wrap,
+								  const btCollisionObjectWrapper* body1Wrap,
+								  const btGImpactShapeInterface* shape0,
+								  const btCompoundShape* shape1, bool swapped);
 
 	void gimpact_vs_concave(
-					  const btCollisionObjectWrapper * body0Wrap,
-					  const btCollisionObjectWrapper * body1Wrap,
-					  const btGImpactShapeInterface * shape0,
-					  const btConcaveShape * shape1,bool swapped);
-
-
-
-
-		/// Accessor/Mutator pairs for Part and triangleID
-    void 	setFace0(int value) 
-    { 
-    	m_triface0 = value; 
-    }
-    int getFace0() 
-    { 
-    	return m_triface0; 
-    }
-    void setFace1(int value) 
-    { 
-    	m_triface1 = value; 
-    }
-    int getFace1() 
-    { 
-    	return m_triface1; 
-    }
-    void setPart0(int value) 
-    { 
-    	m_part0 = value; 
-    }
-    int getPart0() 
-    { 
-    	return m_part0; 
-    }
-    void setPart1(int value) 
-    { 
-    	m_part1 = value; 
-		}
-    int getPart1() 
-    { 
-    	return m_part1; 
-    }
+		const btCollisionObjectWrapper* body0Wrap,
+		const btCollisionObjectWrapper* body1Wrap,
+		const btGImpactShapeInterface* shape0,
+		const btConcaveShape* shape1, bool swapped);
 
+	/// Accessor/Mutator pairs for Part and triangleID
+	void setFace0(int value)
+	{
+		m_triface0 = value;
+	}
+	int getFace0()
+	{
+		return m_triface0;
+	}
+	void setFace1(int value)
+	{
+		m_triface1 = value;
+	}
+	int getFace1()
+	{
+		return m_triface1;
+	}
+	void setPart0(int value)
+	{
+		m_part0 = value;
+	}
+	int getPart0()
+	{
+		return m_part0;
+	}
+	void setPart1(int value)
+	{
+		m_part1 = value;
+	}
+	int getPart1()
+	{
+		return m_part1;
+	}
 };
 
-
 //algorithm details
 //#define BULLET_TRIANGLE_COLLISION 1
 #define GIMPACT_VS_PLANE_COLLISION 1
 
-
-
-#endif //BT_GIMPACT_BVH_CONCAVE_COLLISION_ALGORITHM_H
+#endif  //BT_GIMPACT_BVH_CONCAVE_COLLISION_ALGORITHM_H
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactMassUtil.h b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactMassUtil.h
index 2543aefcfc6..1cde46ed8ba 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactMassUtil.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactMassUtil.h
@@ -21,40 +21,36 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef GIMPACT_MASS_UTIL_H
 #define GIMPACT_MASS_UTIL_H
 
 #include "LinearMath/btTransform.h"
 
-
-
 SIMD_FORCE_INLINE btVector3 gim_inertia_add_transformed(
-	const btVector3 & source_inertia, const btVector3 & added_inertia, const btTransform & transform)
+	const btVector3& source_inertia, const btVector3& added_inertia, const btTransform& transform)
 {
-	btMatrix3x3  rotatedTensor = transform.getBasis().scaled(added_inertia) * transform.getBasis().transpose();
+	btMatrix3x3 rotatedTensor = transform.getBasis().scaled(added_inertia) * transform.getBasis().transpose();
 
 	btScalar x2 = transform.getOrigin()[0];
-	x2*= x2;
+	x2 *= x2;
 	btScalar y2 = transform.getOrigin()[1];
-	y2*= y2;
+	y2 *= y2;
 	btScalar z2 = transform.getOrigin()[2];
-	z2*= z2;
+	z2 *= z2;
 
-	btScalar ix = rotatedTensor[0][0]*(y2+z2);
-	btScalar iy = rotatedTensor[1][1]*(x2+z2);
-	btScalar iz = rotatedTensor[2][2]*(x2+y2);
+	btScalar ix = rotatedTensor[0][0] * (y2 + z2);
+	btScalar iy = rotatedTensor[1][1] * (x2 + z2);
+	btScalar iz = rotatedTensor[2][2] * (x2 + y2);
 
-	return btVector3(source_inertia[0]+ix,source_inertia[1]+iy,source_inertia[2] + iz);
+	return btVector3(source_inertia[0] + ix, source_inertia[1] + iy, source_inertia[2] + iz);
 }
 
-SIMD_FORCE_INLINE btVector3 gim_get_point_inertia(const btVector3 & point, btScalar mass)
+SIMD_FORCE_INLINE btVector3 gim_get_point_inertia(const btVector3& point, btScalar mass)
 {
-	btScalar x2 = point[0]*point[0];
-	btScalar y2 = point[1]*point[1];
-	btScalar z2 = point[2]*point[2];
-	return btVector3(mass*(y2+z2),mass*(x2+z2),mass*(x2+y2));
+	btScalar x2 = point[0] * point[0];
+	btScalar y2 = point[1] * point[1];
+	btScalar z2 = point[2] * point[2];
+	return btVector3(mass * (y2 + z2), mass * (x2 + z2), mass * (x2 + y2));
 }
 
-
-#endif //GIMPACT_MESH_SHAPE_H
+#endif  //GIMPACT_MESH_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactQuantizedBvh.cpp b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactQuantizedBvh.cpp
index 4528758c370..b81fc970449 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactQuantizedBvh.cpp
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactQuantizedBvh.cpp
@@ -27,11 +27,9 @@ subject to the following restrictions:
 #ifdef TRI_COLLISION_PROFILING
 btClock g_q_tree_clock;
 
-
 float g_q_accum_tree_collision_time = 0;
 int g_q_count_traversing = 0;
 
-
 void bt_begin_gim02_q_tree_time()
 {
 	g_q_tree_clock.reset();
@@ -43,11 +41,10 @@ void bt_end_gim02_q_tree_time()
 	g_q_count_traversing++;
 }
 
-
 //! Gets the average time in miliseconds of tree collisions
 float btGImpactQuantizedBvh::getAverageTreeCollisionTime()
 {
-	if(g_q_count_traversing == 0) return 0;
+	if (g_q_count_traversing == 0) return 0;
 
 	float avgtime = g_q_accum_tree_collision_time;
 	avgtime /= (float)g_q_count_traversing;
@@ -56,99 +53,92 @@ float btGImpactQuantizedBvh::getAverageTreeCollisionTime()
 	g_q_count_traversing = 0;
 	return avgtime;
 
-//	float avgtime = g_q_count_traversing;
-//	g_q_count_traversing = 0;
-//	return avgtime;
-
+	//	float avgtime = g_q_count_traversing;
+	//	g_q_count_traversing = 0;
+	//	return avgtime;
 }
 
-#endif //TRI_COLLISION_PROFILING
+#endif  //TRI_COLLISION_PROFILING
 
 /////////////////////// btQuantizedBvhTree /////////////////////////////////
 
 void btQuantizedBvhTree::calc_quantization(
-	GIM_BVH_DATA_ARRAY & primitive_boxes, btScalar boundMargin)
+	GIM_BVH_DATA_ARRAY& primitive_boxes, btScalar boundMargin)
 {
 	//calc globa box
 	btAABB global_bound;
 	global_bound.invalidate();
 
-	for (int i=0;i<primitive_boxes.size() ;i++ )
+	for (int i = 0; i < primitive_boxes.size(); i++)
 	{
 		global_bound.merge(primitive_boxes[i].m_bound);
 	}
 
 	bt_calc_quantization_parameters(
-		m_global_bound.m_min,m_global_bound.m_max,m_bvhQuantization,global_bound.m_min,global_bound.m_max,boundMargin);
-
+		m_global_bound.m_min, m_global_bound.m_max, m_bvhQuantization, global_bound.m_min, global_bound.m_max, boundMargin);
 }
 
-
-
 int btQuantizedBvhTree::_calc_splitting_axis(
-	GIM_BVH_DATA_ARRAY & primitive_boxes, int startIndex,  int endIndex)
+	GIM_BVH_DATA_ARRAY& primitive_boxes, int startIndex, int endIndex)
 {
-
 	int i;
 
-	btVector3 means(btScalar(0.),btScalar(0.),btScalar(0.));
-	btVector3 variance(btScalar(0.),btScalar(0.),btScalar(0.));
-	int numIndices = endIndex-startIndex;
+	btVector3 means(btScalar(0.), btScalar(0.), btScalar(0.));
+	btVector3 variance(btScalar(0.), btScalar(0.), btScalar(0.));
+	int numIndices = endIndex - startIndex;
 
-	for (i=startIndex;i<endIndex;i++)
+	for (i = startIndex; i < endIndex; i++)
 	{
-		btVector3 center = btScalar(0.5)*(primitive_boxes[i].m_bound.m_max +
-					 primitive_boxes[i].m_bound.m_min);
-		means+=center;
+		btVector3 center = btScalar(0.5) * (primitive_boxes[i].m_bound.m_max +
+											primitive_boxes[i].m_bound.m_min);
+		means += center;
 	}
-	means *= (btScalar(1.)/(btScalar)numIndices);
+	means *= (btScalar(1.) / (btScalar)numIndices);
 
-	for (i=startIndex;i<endIndex;i++)
+	for (i = startIndex; i < endIndex; i++)
 	{
-		btVector3 center = btScalar(0.5)*(primitive_boxes[i].m_bound.m_max +
-					 primitive_boxes[i].m_bound.m_min);
-		btVector3 diff2 = center-means;
+		btVector3 center = btScalar(0.5) * (primitive_boxes[i].m_bound.m_max +
+											primitive_boxes[i].m_bound.m_min);
+		btVector3 diff2 = center - means;
 		diff2 = diff2 * diff2;
 		variance += diff2;
 	}
-	variance *= (btScalar(1.)/	((btScalar)numIndices-1)	);
+	variance *= (btScalar(1.) / ((btScalar)numIndices - 1));
 
 	return variance.maxAxis();
 }
 
-
 int btQuantizedBvhTree::_sort_and_calc_splitting_index(
-	GIM_BVH_DATA_ARRAY & primitive_boxes, int startIndex,
+	GIM_BVH_DATA_ARRAY& primitive_boxes, int startIndex,
 	int endIndex, int splitAxis)
 {
 	int i;
-	int splitIndex =startIndex;
+	int splitIndex = startIndex;
 	int numIndices = endIndex - startIndex;
 
 	// average of centers
 	btScalar splitValue = 0.0f;
 
-	btVector3 means(btScalar(0.),btScalar(0.),btScalar(0.));
-	for (i=startIndex;i<endIndex;i++)
+	btVector3 means(btScalar(0.), btScalar(0.), btScalar(0.));
+	for (i = startIndex; i < endIndex; i++)
 	{
-		btVector3 center = btScalar(0.5)*(primitive_boxes[i].m_bound.m_max +
-					 primitive_boxes[i].m_bound.m_min);
-		means+=center;
+		btVector3 center = btScalar(0.5) * (primitive_boxes[i].m_bound.m_max +
+											primitive_boxes[i].m_bound.m_min);
+		means += center;
 	}
-	means *= (btScalar(1.)/(btScalar)numIndices);
+	means *= (btScalar(1.) / (btScalar)numIndices);
 
 	splitValue = means[splitAxis];
 
-
 	//sort leafNodes so all values larger then splitValue comes first, and smaller values start from 'splitIndex'.
-	for (i=startIndex;i<endIndex;i++)
+	for (i = startIndex; i < endIndex; i++)
 	{
-		btVector3 center = btScalar(0.5)*(primitive_boxes[i].m_bound.m_max +
-					 primitive_boxes[i].m_bound.m_min);
+		btVector3 center = btScalar(0.5) * (primitive_boxes[i].m_bound.m_max +
+											primitive_boxes[i].m_bound.m_min);
 		if (center[splitAxis] > splitValue)
 		{
 			//swap
-			primitive_boxes.swap(i,splitIndex);
+			primitive_boxes.swap(i, splitIndex);
 			//swapLeafNodes(i,splitIndex);
 			splitIndex++;
 		}
@@ -163,32 +153,30 @@ int btQuantizedBvhTree::_sort_and_calc_splitting_index(
 	//bool unbalanced2 = true;
 
 	//this should be safe too:
-	int rangeBalancedIndices = numIndices/3;
-	bool unbalanced = ((splitIndex<=(startIndex+rangeBalancedIndices)) || (splitIndex >=(endIndex-1-rangeBalancedIndices)));
+	int rangeBalancedIndices = numIndices / 3;
+	bool unbalanced = ((splitIndex <= (startIndex + rangeBalancedIndices)) || (splitIndex >= (endIndex - 1 - rangeBalancedIndices)));
 
 	if (unbalanced)
 	{
-		splitIndex = startIndex+ (numIndices>>1);
+		splitIndex = startIndex + (numIndices >> 1);
 	}
 
-	btAssert(!((splitIndex==startIndex) || (splitIndex == (endIndex))));
+	btAssert(!((splitIndex == startIndex) || (splitIndex == (endIndex))));
 
 	return splitIndex;
-
 }
 
-
-void btQuantizedBvhTree::_build_sub_tree(GIM_BVH_DATA_ARRAY & primitive_boxes, int startIndex,  int endIndex)
+void btQuantizedBvhTree::_build_sub_tree(GIM_BVH_DATA_ARRAY& primitive_boxes, int startIndex, int endIndex)
 {
 	int curIndex = m_num_nodes;
 	m_num_nodes++;
 
-	btAssert((endIndex-startIndex)>0);
+	btAssert((endIndex - startIndex) > 0);
 
-	if ((endIndex-startIndex)==1)
+	if ((endIndex - startIndex) == 1)
 	{
-	    //We have a leaf node
-	    setNodeBound(curIndex,primitive_boxes[startIndex].m_bound);
+		//We have a leaf node
+		setNodeBound(curIndex, primitive_boxes[startIndex].m_bound);
 		m_node_array[curIndex].setDataIndex(primitive_boxes[startIndex].m_data);
 
 		return;
@@ -196,48 +184,43 @@ void btQuantizedBvhTree::_build_sub_tree(GIM_BVH_DATA_ARRAY & primitive_boxes, i
 	//calculate Best Splitting Axis and where to split it. Sort the incoming 'leafNodes' array within range 'startIndex/endIndex'.
 
 	//split axis
-	int splitIndex = _calc_splitting_axis(primitive_boxes,startIndex,endIndex);
+	int splitIndex = _calc_splitting_axis(primitive_boxes, startIndex, endIndex);
 
 	splitIndex = _sort_and_calc_splitting_index(
-			primitive_boxes,startIndex,endIndex,
-			splitIndex//split axis
-			);
-
+		primitive_boxes, startIndex, endIndex,
+		splitIndex  //split axis
+	);
 
 	//calc this node bounding box
 
 	btAABB node_bound;
 	node_bound.invalidate();
 
-	for (int i=startIndex;i<endIndex;i++)
+	for (int i = startIndex; i < endIndex; i++)
 	{
 		node_bound.merge(primitive_boxes[i].m_bound);
 	}
 
-	setNodeBound(curIndex,node_bound);
-
+	setNodeBound(curIndex, node_bound);
 
 	//build left branch
-	_build_sub_tree(primitive_boxes, startIndex, splitIndex );
-
+	_build_sub_tree(primitive_boxes, startIndex, splitIndex);
 
 	//build right branch
-	 _build_sub_tree(primitive_boxes, splitIndex ,endIndex);
+	_build_sub_tree(primitive_boxes, splitIndex, endIndex);
 
 	m_node_array[curIndex].setEscapeIndex(m_num_nodes - curIndex);
-
-
 }
 
 //! stackless build tree
 void btQuantizedBvhTree::build_tree(
-	GIM_BVH_DATA_ARRAY & primitive_boxes)
+	GIM_BVH_DATA_ARRAY& primitive_boxes)
 {
 	calc_quantization(primitive_boxes);
 	// initialize node count to 0
 	m_num_nodes = 0;
 	// allocate nodes
-	m_node_array.resize(primitive_boxes.size()*2);
+	m_node_array.resize(primitive_boxes.size() * 2);
 
 	_build_sub_tree(primitive_boxes, 0, primitive_boxes.size());
 }
@@ -247,13 +230,13 @@ void btQuantizedBvhTree::build_tree(
 void btGImpactQuantizedBvh::refit()
 {
 	int nodecount = getNodeCount();
-	while(nodecount--)
+	while (nodecount--)
 	{
-		if(isLeafNode(nodecount))
+		if (isLeafNode(nodecount))
 		{
 			btAABB leafbox;
-			m_primitive_manager->get_primitive_box(getNodeData(nodecount),leafbox);
-			setNodeBound(nodecount,leafbox);
+			m_primitive_manager->get_primitive_box(getNodeData(nodecount), leafbox);
+			setNodeBound(nodecount, leafbox);
 		}
 		else
 		{
@@ -265,20 +248,20 @@ void btGImpactQuantizedBvh::refit()
 			btAABB temp_box;
 
 			int child_node = getLeftNode(nodecount);
-			if(child_node)
+			if (child_node)
 			{
-				getNodeBound(child_node,temp_box);
+				getNodeBound(child_node, temp_box);
 				bound.merge(temp_box);
 			}
 
 			child_node = getRightNode(nodecount);
-			if(child_node)
+			if (child_node)
 			{
-				getNodeBound(child_node,temp_box);
+				getNodeBound(child_node, temp_box);
 				bound.merge(temp_box);
 			}
 
-			setNodeBound(nodecount,bound);
+			setNodeBound(nodecount, bound);
 		}
 	}
 }
@@ -290,17 +273,17 @@ void btGImpactQuantizedBvh::buildSet()
 	GIM_BVH_DATA_ARRAY primitive_boxes;
 	primitive_boxes.resize(m_primitive_manager->get_primitive_count());
 
-	for (int i = 0;i<primitive_boxes.size() ;i++ )
+	for (int i = 0; i < primitive_boxes.size(); i++)
 	{
-		 m_primitive_manager->get_primitive_box(i,primitive_boxes[i].m_bound);
-		 primitive_boxes[i].m_data = i;
+		m_primitive_manager->get_primitive_box(i, primitive_boxes[i].m_bound);
+		primitive_boxes[i].m_data = i;
 	}
 
 	m_box_tree.build_tree(primitive_boxes);
 }
 
 //! returns the indices of the primitives in the m_primitive_manager
-bool btGImpactQuantizedBvh::boxQuery(const btAABB & box, btAlignedObjectArray<int> & collided_results) const
+bool btGImpactQuantizedBvh::boxQuery(const btAABB& box, btAlignedObjectArray<int>& collided_results) const
 {
 	int curIndex = 0;
 	int numNodes = getNodeCount();
@@ -310,16 +293,14 @@ bool btGImpactQuantizedBvh::boxQuery(const btAABB & box, btAlignedObjectArray<in
 	unsigned short quantizedMin[3];
 	unsigned short quantizedMax[3];
 
-	m_box_tree.quantizePoint(quantizedMin,box.m_min);
-	m_box_tree.quantizePoint(quantizedMax,box.m_max);
-
+	m_box_tree.quantizePoint(quantizedMin, box.m_min);
+	m_box_tree.quantizePoint(quantizedMax, box.m_max);
 
 	while (curIndex < numNodes)
 	{
-
 		//catch bugs in tree data
 
-		bool aabbOverlap = m_box_tree.testQuantizedBoxOverlapp(curIndex, quantizedMin,quantizedMax);
+		bool aabbOverlap = m_box_tree.testQuantizedBoxOverlapp(curIndex, quantizedMin, quantizedMax);
 		bool isleafnode = isLeafNode(curIndex);
 
 		if (isleafnode && aabbOverlap)
@@ -335,19 +316,17 @@ bool btGImpactQuantizedBvh::boxQuery(const btAABB & box, btAlignedObjectArray<in
 		else
 		{
 			//skip node
-			curIndex+= getEscapeNodeIndex(curIndex);
+			curIndex += getEscapeNodeIndex(curIndex);
 		}
 	}
-	if(collided_results.size()>0) return true;
+	if (collided_results.size() > 0) return true;
 	return false;
 }
 
-
-
 //! returns the indices of the primitives in the m_primitive_manager
 bool btGImpactQuantizedBvh::rayQuery(
-	const btVector3 & ray_dir,const btVector3 & ray_origin ,
-	btAlignedObjectArray<int> & collided_results) const
+	const btVector3& ray_dir, const btVector3& ray_origin,
+	btAlignedObjectArray<int>& collided_results) const
 {
 	int curIndex = 0;
 	int numNodes = getNodeCount();
@@ -355,16 +334,16 @@ bool btGImpactQuantizedBvh::rayQuery(
 	while (curIndex < numNodes)
 	{
 		btAABB bound;
-		getNodeBound(curIndex,bound);
+		getNodeBound(curIndex, bound);
 
 		//catch bugs in tree data
 
-		bool aabbOverlap = bound.collide_ray(ray_origin,ray_dir);
+		bool aabbOverlap = bound.collide_ray(ray_origin, ray_dir);
 		bool isleafnode = isLeafNode(curIndex);
 
 		if (isleafnode && aabbOverlap)
 		{
-			collided_results.push_back(getNodeData( curIndex));
+			collided_results.push_back(getNodeData(curIndex));
 		}
 
 		if (aabbOverlap || isleafnode)
@@ -375,154 +354,133 @@ bool btGImpactQuantizedBvh::rayQuery(
 		else
 		{
 			//skip node
-			curIndex+= getEscapeNodeIndex(curIndex);
+			curIndex += getEscapeNodeIndex(curIndex);
 		}
 	}
-	if(collided_results.size()>0) return true;
+	if (collided_results.size() > 0) return true;
 	return false;
 }
 
-
 SIMD_FORCE_INLINE bool _quantized_node_collision(
-	const btGImpactQuantizedBvh * boxset0, const btGImpactQuantizedBvh * boxset1,
-	const BT_BOX_BOX_TRANSFORM_CACHE & trans_cache_1to0,
-	int node0 ,int node1, bool complete_primitive_tests)
+	const btGImpactQuantizedBvh* boxset0, const btGImpactQuantizedBvh* boxset1,
+	const BT_BOX_BOX_TRANSFORM_CACHE& trans_cache_1to0,
+	int node0, int node1, bool complete_primitive_tests)
 {
 	btAABB box0;
-	boxset0->getNodeBound(node0,box0);
+	boxset0->getNodeBound(node0, box0);
 	btAABB box1;
-	boxset1->getNodeBound(node1,box1);
-
-	return box0.overlapping_trans_cache(box1,trans_cache_1to0,complete_primitive_tests );
-//	box1.appy_transform_trans_cache(trans_cache_1to0);
-//	return box0.has_collision(box1);
+	boxset1->getNodeBound(node1, box1);
 
+	return box0.overlapping_trans_cache(box1, trans_cache_1to0, complete_primitive_tests);
+	//	box1.appy_transform_trans_cache(trans_cache_1to0);
+	//	return box0.has_collision(box1);
 }
 
-
 //stackless recursive collision routine
 static void _find_quantized_collision_pairs_recursive(
-	const btGImpactQuantizedBvh * boxset0, const btGImpactQuantizedBvh * boxset1,
-	btPairSet * collision_pairs,
-	const BT_BOX_BOX_TRANSFORM_CACHE & trans_cache_1to0,
+	const btGImpactQuantizedBvh* boxset0, const btGImpactQuantizedBvh* boxset1,
+	btPairSet* collision_pairs,
+	const BT_BOX_BOX_TRANSFORM_CACHE& trans_cache_1to0,
 	int node0, int node1, bool complete_primitive_tests)
 {
+	if (_quantized_node_collision(
+			boxset0, boxset1, trans_cache_1to0,
+			node0, node1, complete_primitive_tests) == false) return;  //avoid colliding internal nodes
 
-
-
-	if( _quantized_node_collision(
-		boxset0,boxset1,trans_cache_1to0,
-		node0,node1,complete_primitive_tests) ==false) return;//avoid colliding internal nodes
-
-	if(boxset0->isLeafNode(node0))
+	if (boxset0->isLeafNode(node0))
 	{
-		if(boxset1->isLeafNode(node1))
+		if (boxset1->isLeafNode(node1))
 		{
 			// collision result
 			collision_pairs->push_pair(
-				boxset0->getNodeData(node0),boxset1->getNodeData(node1));
+				boxset0->getNodeData(node0), boxset1->getNodeData(node1));
 			return;
 		}
 		else
 		{
-
 			//collide left recursive
 
 			_find_quantized_collision_pairs_recursive(
-								boxset0,boxset1,
-								collision_pairs,trans_cache_1to0,
-								node0,boxset1->getLeftNode(node1),false);
+				boxset0, boxset1,
+				collision_pairs, trans_cache_1to0,
+				node0, boxset1->getLeftNode(node1), false);
 
 			//collide right recursive
 			_find_quantized_collision_pairs_recursive(
-								boxset0,boxset1,
-								collision_pairs,trans_cache_1to0,
-								node0,boxset1->getRightNode(node1),false);
-
-
+				boxset0, boxset1,
+				collision_pairs, trans_cache_1to0,
+				node0, boxset1->getRightNode(node1), false);
 		}
 	}
 	else
 	{
-		if(boxset1->isLeafNode(node1))
+		if (boxset1->isLeafNode(node1))
 		{
-
 			//collide left recursive
 			_find_quantized_collision_pairs_recursive(
-								boxset0,boxset1,
-								collision_pairs,trans_cache_1to0,
-								boxset0->getLeftNode(node0),node1,false);
-
+				boxset0, boxset1,
+				collision_pairs, trans_cache_1to0,
+				boxset0->getLeftNode(node0), node1, false);
 
 			//collide right recursive
 
 			_find_quantized_collision_pairs_recursive(
-								boxset0,boxset1,
-								collision_pairs,trans_cache_1to0,
-								boxset0->getRightNode(node0),node1,false);
-
-
+				boxset0, boxset1,
+				collision_pairs, trans_cache_1to0,
+				boxset0->getRightNode(node0), node1, false);
 		}
 		else
 		{
 			//collide left0 left1
 
-
-
 			_find_quantized_collision_pairs_recursive(
-				boxset0,boxset1,
-				collision_pairs,trans_cache_1to0,
-				boxset0->getLeftNode(node0),boxset1->getLeftNode(node1),false);
+				boxset0, boxset1,
+				collision_pairs, trans_cache_1to0,
+				boxset0->getLeftNode(node0), boxset1->getLeftNode(node1), false);
 
 			//collide left0 right1
 
 			_find_quantized_collision_pairs_recursive(
-				boxset0,boxset1,
-				collision_pairs,trans_cache_1to0,
-				boxset0->getLeftNode(node0),boxset1->getRightNode(node1),false);
-
+				boxset0, boxset1,
+				collision_pairs, trans_cache_1to0,
+				boxset0->getLeftNode(node0), boxset1->getRightNode(node1), false);
 
 			//collide right0 left1
 
 			_find_quantized_collision_pairs_recursive(
-				boxset0,boxset1,
-				collision_pairs,trans_cache_1to0,
-				boxset0->getRightNode(node0),boxset1->getLeftNode(node1),false);
+				boxset0, boxset1,
+				collision_pairs, trans_cache_1to0,
+				boxset0->getRightNode(node0), boxset1->getLeftNode(node1), false);
 
 			//collide right0 right1
 
 			_find_quantized_collision_pairs_recursive(
-				boxset0,boxset1,
-				collision_pairs,trans_cache_1to0,
-				boxset0->getRightNode(node0),boxset1->getRightNode(node1),false);
+				boxset0, boxset1,
+				collision_pairs, trans_cache_1to0,
+				boxset0->getRightNode(node0), boxset1->getRightNode(node1), false);
 
-		}// else if node1 is not a leaf
-	}// else if node0 is not a leaf
+		}  // else if node1 is not a leaf
+	}      // else if node0 is not a leaf
 }
 
-
-void btGImpactQuantizedBvh::find_collision(const btGImpactQuantizedBvh * boxset0, const btTransform & trans0,
-		const btGImpactQuantizedBvh * boxset1, const btTransform & trans1,
-		btPairSet & collision_pairs)
+void btGImpactQuantizedBvh::find_collision(const btGImpactQuantizedBvh* boxset0, const btTransform& trans0,
+										   const btGImpactQuantizedBvh* boxset1, const btTransform& trans1,
+										   btPairSet& collision_pairs)
 {
-
-	if(boxset0->getNodeCount()==0 || boxset1->getNodeCount()==0 ) return;
+	if (boxset0->getNodeCount() == 0 || boxset1->getNodeCount() == 0) return;
 
 	BT_BOX_BOX_TRANSFORM_CACHE trans_cache_1to0;
 
-	trans_cache_1to0.calc_from_homogenic(trans0,trans1);
+	trans_cache_1to0.calc_from_homogenic(trans0, trans1);
 
 #ifdef TRI_COLLISION_PROFILING
 	bt_begin_gim02_q_tree_time();
-#endif //TRI_COLLISION_PROFILING
+#endif  //TRI_COLLISION_PROFILING
 
 	_find_quantized_collision_pairs_recursive(
-		boxset0,boxset1,
-		&collision_pairs,trans_cache_1to0,0,0,true);
+		boxset0, boxset1,
+		&collision_pairs, trans_cache_1to0, 0, 0, true);
 #ifdef TRI_COLLISION_PROFILING
 	bt_end_gim02_q_tree_time();
-#endif //TRI_COLLISION_PROFILING
-
+#endif  //TRI_COLLISION_PROFILING
 }
-
-
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactQuantizedBvh.h b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactQuantizedBvh.h
index 42e5520fc03..b231c1e8327 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactQuantizedBvh.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactQuantizedBvh.h
@@ -28,13 +28,10 @@ subject to the following restrictions:
 #include "btQuantization.h"
 #include "btGImpactQuantizedBvhStructs.h"
 
-class GIM_QUANTIZED_BVH_NODE_ARRAY:public btAlignedObjectArray<BT_QUANTIZED_BVH_NODE>
+class GIM_QUANTIZED_BVH_NODE_ARRAY : public btAlignedObjectArray<BT_QUANTIZED_BVH_NODE>
 {
 };
 
-
-
-
 //! Basic Box tree structure
 class btQuantizedBvhTree
 {
@@ -43,16 +40,18 @@ protected:
 	GIM_QUANTIZED_BVH_NODE_ARRAY m_node_array;
 	btAABB m_global_bound;
 	btVector3 m_bvhQuantization;
+
 protected:
-	void calc_quantization(GIM_BVH_DATA_ARRAY & primitive_boxes, btScalar boundMargin = btScalar(1.0) );
+	void calc_quantization(GIM_BVH_DATA_ARRAY& primitive_boxes, btScalar boundMargin = btScalar(1.0));
 
 	int _sort_and_calc_splitting_index(
-		GIM_BVH_DATA_ARRAY & primitive_boxes,
-		 int startIndex,  int endIndex, int splitAxis);
+		GIM_BVH_DATA_ARRAY& primitive_boxes,
+		int startIndex, int endIndex, int splitAxis);
 
-	int _calc_splitting_axis(GIM_BVH_DATA_ARRAY & primitive_boxes, int startIndex,  int endIndex);
+	int _calc_splitting_axis(GIM_BVH_DATA_ARRAY& primitive_boxes, int startIndex, int endIndex);
+
+	void _build_sub_tree(GIM_BVH_DATA_ARRAY& primitive_boxes, int startIndex, int endIndex);
 
-	void _build_sub_tree(GIM_BVH_DATA_ARRAY & primitive_boxes, int startIndex,  int endIndex);
 public:
 	btQuantizedBvhTree()
 	{
@@ -61,20 +60,19 @@ public:
 
 	//! prototype functions for box tree management
 	//!@{
-	void build_tree(GIM_BVH_DATA_ARRAY & primitive_boxes);
+	void build_tree(GIM_BVH_DATA_ARRAY& primitive_boxes);
 
 	SIMD_FORCE_INLINE void quantizePoint(
-		unsigned short * quantizedpoint, const btVector3 & point) const
+		unsigned short* quantizedpoint, const btVector3& point) const
 	{
-		bt_quantize_clamp(quantizedpoint,point,m_global_bound.m_min,m_global_bound.m_max,m_bvhQuantization);
+		bt_quantize_clamp(quantizedpoint, point, m_global_bound.m_min, m_global_bound.m_max, m_bvhQuantization);
 	}
 
-
 	SIMD_FORCE_INLINE bool testQuantizedBoxOverlapp(
 		int node_index,
-		unsigned short * quantizedMin,unsigned short * quantizedMax) const
+		unsigned short* quantizedMin, unsigned short* quantizedMax) const
 	{
-		return m_node_array[node_index].testQuantizedBoxOverlapp(quantizedMin,quantizedMax);
+		return m_node_array[node_index].testQuantizedBoxOverlapp(quantizedMin, quantizedMax);
 	}
 
 	SIMD_FORCE_INLINE void clearNodes()
@@ -100,41 +98,41 @@ public:
 		return m_node_array[nodeindex].getDataIndex();
 	}
 
-	SIMD_FORCE_INLINE void getNodeBound(int nodeindex, btAABB & bound) const
+	SIMD_FORCE_INLINE void getNodeBound(int nodeindex, btAABB& bound) const
 	{
 		bound.m_min = bt_unquantize(
 			m_node_array[nodeindex].m_quantizedAabbMin,
-			m_global_bound.m_min,m_bvhQuantization);
+			m_global_bound.m_min, m_bvhQuantization);
 
 		bound.m_max = bt_unquantize(
 			m_node_array[nodeindex].m_quantizedAabbMax,
-			m_global_bound.m_min,m_bvhQuantization);
+			m_global_bound.m_min, m_bvhQuantization);
 	}
 
-	SIMD_FORCE_INLINE void setNodeBound(int nodeindex, const btAABB & bound)
+	SIMD_FORCE_INLINE void setNodeBound(int nodeindex, const btAABB& bound)
 	{
-		bt_quantize_clamp(	m_node_array[nodeindex].m_quantizedAabbMin,
-							bound.m_min,
-							m_global_bound.m_min,
-							m_global_bound.m_max,
-							m_bvhQuantization);
+		bt_quantize_clamp(m_node_array[nodeindex].m_quantizedAabbMin,
+						  bound.m_min,
+						  m_global_bound.m_min,
+						  m_global_bound.m_max,
+						  m_bvhQuantization);
 
-		bt_quantize_clamp(	m_node_array[nodeindex].m_quantizedAabbMax,
-							bound.m_max,
-							m_global_bound.m_min,
-							m_global_bound.m_max,
-							m_bvhQuantization);
+		bt_quantize_clamp(m_node_array[nodeindex].m_quantizedAabbMax,
+						  bound.m_max,
+						  m_global_bound.m_min,
+						  m_global_bound.m_max,
+						  m_bvhQuantization);
 	}
 
 	SIMD_FORCE_INLINE int getLeftNode(int nodeindex) const
 	{
-		return nodeindex+1;
+		return nodeindex + 1;
 	}
 
 	SIMD_FORCE_INLINE int getRightNode(int nodeindex) const
 	{
-		if(m_node_array[nodeindex+1].isLeafNode()) return nodeindex+2;
-		return nodeindex+1 + m_node_array[nodeindex+1].getEscapeIndex();
+		if (m_node_array[nodeindex + 1].isLeafNode()) return nodeindex + 2;
+		return nodeindex + 1 + m_node_array[nodeindex + 1].getEscapeIndex();
 	}
 
 	SIMD_FORCE_INLINE int getEscapeNodeIndex(int nodeindex) const
@@ -142,7 +140,7 @@ public:
 		return m_node_array[nodeindex].getEscapeIndex();
 	}
 
-	SIMD_FORCE_INLINE const BT_QUANTIZED_BVH_NODE * get_node_pointer(int index = 0) const
+	SIMD_FORCE_INLINE const BT_QUANTIZED_BVH_NODE* get_node_pointer(int index = 0) const
 	{
 		return &m_node_array[index];
 	}
@@ -150,8 +148,6 @@ public:
 	//!@}
 };
 
-
-
 //! Structure for containing Boxes
 /*!
 This class offers an structure for managing a box tree of primitives.
@@ -161,13 +157,13 @@ class btGImpactQuantizedBvh
 {
 protected:
 	btQuantizedBvhTree m_box_tree;
-	btPrimitiveManagerBase * m_primitive_manager;
+	btPrimitiveManagerBase* m_primitive_manager;
 
 protected:
 	//stackless refit
 	void refit();
-public:
 
+public:
 	//! this constructor doesn't build the tree. you must call	buildSet
 	btGImpactQuantizedBvh()
 	{
@@ -175,31 +171,30 @@ public:
 	}
 
 	//! this constructor doesn't build the tree. you must call	buildSet
-	btGImpactQuantizedBvh(btPrimitiveManagerBase * primitive_manager)
+	btGImpactQuantizedBvh(btPrimitiveManagerBase* primitive_manager)
 	{
 		m_primitive_manager = primitive_manager;
 	}
 
-	SIMD_FORCE_INLINE btAABB getGlobalBox()  const
+	SIMD_FORCE_INLINE btAABB getGlobalBox() const
 	{
 		btAABB totalbox;
 		getNodeBound(0, totalbox);
 		return totalbox;
 	}
 
-	SIMD_FORCE_INLINE void setPrimitiveManager(btPrimitiveManagerBase * primitive_manager)
+	SIMD_FORCE_INLINE void setPrimitiveManager(btPrimitiveManagerBase* primitive_manager)
 	{
 		m_primitive_manager = primitive_manager;
 	}
 
-	SIMD_FORCE_INLINE btPrimitiveManagerBase * getPrimitiveManager() const
+	SIMD_FORCE_INLINE btPrimitiveManagerBase* getPrimitiveManager() const
 	{
 		return m_primitive_manager;
 	}
 
-
-//! node manager prototype functions
-///@{
+	//! node manager prototype functions
+	///@{
 
 	//! this attemps to refit the box set.
 	SIMD_FORCE_INLINE void update()
@@ -211,21 +206,21 @@ public:
 	void buildSet();
 
 	//! returns the indices of the primitives in the m_primitive_manager
-	bool boxQuery(const btAABB & box, btAlignedObjectArray<int> & collided_results) const;
+	bool boxQuery(const btAABB& box, btAlignedObjectArray<int>& collided_results) const;
 
 	//! returns the indices of the primitives in the m_primitive_manager
-	SIMD_FORCE_INLINE bool boxQueryTrans(const btAABB & box,
-		 const btTransform & transform, btAlignedObjectArray<int> & collided_results) const
+	SIMD_FORCE_INLINE bool boxQueryTrans(const btAABB& box,
+										 const btTransform& transform, btAlignedObjectArray<int>& collided_results) const
 	{
-		btAABB transbox=box;
+		btAABB transbox = box;
 		transbox.appy_transform(transform);
-		return boxQuery(transbox,collided_results);
+		return boxQuery(transbox, collided_results);
 	}
 
 	//! returns the indices of the primitives in the m_primitive_manager
 	bool rayQuery(
-		const btVector3 & ray_dir,const btVector3 & ray_origin ,
-		btAlignedObjectArray<int> & collided_results) const;
+		const btVector3& ray_dir, const btVector3& ray_origin,
+		btAlignedObjectArray<int>& collided_results) const;
 
 	//! tells if this set has hierarcht
 	SIMD_FORCE_INLINE bool hasHierarchy() const
@@ -234,7 +229,7 @@ public:
 	}
 
 	//! tells if this set is a trimesh
-	SIMD_FORCE_INLINE bool isTrimesh()  const
+	SIMD_FORCE_INLINE bool isTrimesh() const
 	{
 		return m_primitive_manager->is_trimesh();
 	}
@@ -256,17 +251,16 @@ public:
 		return m_box_tree.getNodeData(nodeindex);
 	}
 
-	SIMD_FORCE_INLINE void getNodeBound(int nodeindex, btAABB & bound)  const
+	SIMD_FORCE_INLINE void getNodeBound(int nodeindex, btAABB& bound) const
 	{
 		m_box_tree.getNodeBound(nodeindex, bound);
 	}
 
-	SIMD_FORCE_INLINE void setNodeBound(int nodeindex, const btAABB & bound)
+	SIMD_FORCE_INLINE void setNodeBound(int nodeindex, const btAABB& bound)
 	{
 		m_box_tree.setNodeBound(nodeindex, bound);
 	}
 
-
 	SIMD_FORCE_INLINE int getLeftNode(int nodeindex) const
 	{
 		return m_box_tree.getLeftNode(nodeindex);
@@ -282,24 +276,23 @@ public:
 		return m_box_tree.getEscapeNodeIndex(nodeindex);
 	}
 
-	SIMD_FORCE_INLINE void getNodeTriangle(int nodeindex,btPrimitiveTriangle & triangle) const
+	SIMD_FORCE_INLINE void getNodeTriangle(int nodeindex, btPrimitiveTriangle& triangle) const
 	{
-		m_primitive_manager->get_primitive_triangle(getNodeData(nodeindex),triangle);
+		m_primitive_manager->get_primitive_triangle(getNodeData(nodeindex), triangle);
 	}
 
-
-	SIMD_FORCE_INLINE const BT_QUANTIZED_BVH_NODE * get_node_pointer(int index = 0) const
+	SIMD_FORCE_INLINE const BT_QUANTIZED_BVH_NODE* get_node_pointer(int index = 0) const
 	{
 		return m_box_tree.get_node_pointer(index);
 	}
 
 #ifdef TRI_COLLISION_PROFILING
 	static float getAverageTreeCollisionTime();
-#endif //TRI_COLLISION_PROFILING
+#endif  //TRI_COLLISION_PROFILING
 
-	static void find_collision(const btGImpactQuantizedBvh * boxset1, const btTransform & trans1,
-		const btGImpactQuantizedBvh * boxset2, const btTransform & trans2,
-		btPairSet & collision_pairs);
+	static void find_collision(const btGImpactQuantizedBvh* boxset1, const btTransform& trans1,
+							   const btGImpactQuantizedBvh* boxset2, const btTransform& trans2,
+							   btPairSet& collision_pairs);
 };
 
-#endif // GIM_BOXPRUNING_H_INCLUDED
+#endif  // GIM_BOXPRUNING_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactQuantizedBvhStructs.h b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactQuantizedBvhStructs.h
index 7dd5a1b9d01..bd50cb5b87e 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactQuantizedBvhStructs.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactQuantizedBvhStructs.h
@@ -29,13 +29,14 @@ subject to the following restrictions:
 
 ///btQuantizedBvhNode is a compressed aabb node, 16 bytes.
 ///Node can be used for leafnode or internal node. Leafnodes can point to 32-bit triangle index (non-negative range).
-ATTRIBUTE_ALIGNED16	(struct) BT_QUANTIZED_BVH_NODE
+ATTRIBUTE_ALIGNED16(struct)
+BT_QUANTIZED_BVH_NODE
 {
 	//12 bytes
-	unsigned short int	m_quantizedAabbMin[3];
-	unsigned short int	m_quantizedAabbMax[3];
+	unsigned short int m_quantizedAabbMin[3];
+	unsigned short int m_quantizedAabbMax[3];
 	//4 bytes
-	int	m_escapeIndexOrDataIndex;
+	int m_escapeIndexOrDataIndex;
 
 	BT_QUANTIZED_BVH_NODE()
 	{
@@ -45,7 +46,7 @@ ATTRIBUTE_ALIGNED16	(struct) BT_QUANTIZED_BVH_NODE
 	SIMD_FORCE_INLINE bool isLeafNode() const
 	{
 		//skipindex is negative (internal node), triangleindex >=0 (leafnode)
-		return (m_escapeIndexOrDataIndex>=0);
+		return (m_escapeIndexOrDataIndex >= 0);
 	}
 
 	SIMD_FORCE_INLINE int getEscapeIndex() const
@@ -72,20 +73,19 @@ ATTRIBUTE_ALIGNED16	(struct) BT_QUANTIZED_BVH_NODE
 	}
 
 	SIMD_FORCE_INLINE bool testQuantizedBoxOverlapp(
-		unsigned short * quantizedMin,unsigned short * quantizedMax) const
+		unsigned short* quantizedMin, unsigned short* quantizedMax) const
 	{
-		if(m_quantizedAabbMin[0] > quantizedMax[0] ||
-		   m_quantizedAabbMax[0] < quantizedMin[0] ||
-		   m_quantizedAabbMin[1] > quantizedMax[1] ||
-		   m_quantizedAabbMax[1] < quantizedMin[1] ||
-		   m_quantizedAabbMin[2] > quantizedMax[2] ||
-		   m_quantizedAabbMax[2] < quantizedMin[2])
+		if (m_quantizedAabbMin[0] > quantizedMax[0] ||
+			m_quantizedAabbMax[0] < quantizedMin[0] ||
+			m_quantizedAabbMin[1] > quantizedMax[1] ||
+			m_quantizedAabbMax[1] < quantizedMin[1] ||
+			m_quantizedAabbMin[2] > quantizedMax[2] ||
+			m_quantizedAabbMax[2] < quantizedMin[2])
 		{
 			return false;
 		}
 		return true;
 	}
-
 };
 
-#endif // GIM_QUANTIZED_SET_STRUCTS_H_INCLUDED
+#endif  // GIM_QUANTIZED_SET_STRUCTS_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactShape.cpp b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactShape.cpp
index 30c85e3fffd..34c229a3aba 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactShape.cpp
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactShape.cpp
@@ -18,178 +18,169 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btGImpactShape.h"
 #include "btGImpactMassUtil.h"
 
-
-btGImpactMeshShapePart::btGImpactMeshShapePart( btStridingMeshInterface * meshInterface, int part )
+btGImpactMeshShapePart::btGImpactMeshShapePart(btStridingMeshInterface* meshInterface, int part)
 {
-    // moved from .h to .cpp because of conditional compilation
-    // (The setting of BT_THREADSAFE may differ between various cpp files, so it is best to
-    // avoid using it in h files)
-    m_primitive_manager.m_meshInterface = meshInterface;
-    m_primitive_manager.m_part = part;
-    m_box_set.setPrimitiveManager( &m_primitive_manager );
+	// moved from .h to .cpp because of conditional compilation
+	// (The setting of BT_THREADSAFE may differ between various cpp files, so it is best to
+	// avoid using it in h files)
+	m_primitive_manager.m_meshInterface = meshInterface;
+	m_primitive_manager.m_part = part;
+	m_box_set.setPrimitiveManager(&m_primitive_manager);
 #if BT_THREADSAFE
-    // If threadsafe is requested, this object uses a different lock/unlock
-    //  model with the btStridingMeshInterface -- lock once when the object is constructed
-    //  and unlock once in the destructor.
-    // The other way of locking and unlocking for each collision check in the narrowphase
-    // is not threadsafe.  Note these are not thread-locks, they are calls to the meshInterface's
-    // getLockedReadOnlyVertexIndexBase virtual function, which by default just returns a couple of
-    // pointers.  In theory a client could override the lock function to do all sorts of
-    // things like reading data from GPU memory, or decompressing data on the fly, but such things
-    // do not seem all that likely or useful, given the performance cost.
-    m_primitive_manager.lock();
+	// If threadsafe is requested, this object uses a different lock/unlock
+	//  model with the btStridingMeshInterface -- lock once when the object is constructed
+	//  and unlock once in the destructor.
+	// The other way of locking and unlocking for each collision check in the narrowphase
+	// is not threadsafe.  Note these are not thread-locks, they are calls to the meshInterface's
+	// getLockedReadOnlyVertexIndexBase virtual function, which by default just returns a couple of
+	// pointers.  In theory a client could override the lock function to do all sorts of
+	// things like reading data from GPU memory, or decompressing data on the fly, but such things
+	// do not seem all that likely or useful, given the performance cost.
+	m_primitive_manager.lock();
 #endif
 }
 
 btGImpactMeshShapePart::~btGImpactMeshShapePart()
 {
-    // moved from .h to .cpp because of conditional compilation
+	// moved from .h to .cpp because of conditional compilation
 #if BT_THREADSAFE
-    m_primitive_manager.unlock();
+	m_primitive_manager.unlock();
 #endif
 }
 
 void btGImpactMeshShapePart::lockChildShapes() const
 {
-    // moved from .h to .cpp because of conditional compilation
-#if ! BT_THREADSAFE
-    // called in the narrowphase -- not threadsafe!
-    void * dummy = (void*) ( m_box_set.getPrimitiveManager() );
-    TrimeshPrimitiveManager * dummymanager = static_cast<TrimeshPrimitiveManager *>( dummy );
-    dummymanager->lock();
+	// moved from .h to .cpp because of conditional compilation
+#if !BT_THREADSAFE
+	// called in the narrowphase -- not threadsafe!
+	void* dummy = (void*)(m_box_set.getPrimitiveManager());
+	TrimeshPrimitiveManager* dummymanager = static_cast<TrimeshPrimitiveManager*>(dummy);
+	dummymanager->lock();
 #endif
 }
 
-void btGImpactMeshShapePart::unlockChildShapes()  const
+void btGImpactMeshShapePart::unlockChildShapes() const
 {
-    // moved from .h to .cpp because of conditional compilation
-#if ! BT_THREADSAFE
-    // called in the narrowphase -- not threadsafe!
-    void * dummy = (void*) ( m_box_set.getPrimitiveManager() );
-    TrimeshPrimitiveManager * dummymanager = static_cast<TrimeshPrimitiveManager *>( dummy );
-    dummymanager->unlock();
+	// moved from .h to .cpp because of conditional compilation
+#if !BT_THREADSAFE
+	// called in the narrowphase -- not threadsafe!
+	void* dummy = (void*)(m_box_set.getPrimitiveManager());
+	TrimeshPrimitiveManager* dummymanager = static_cast<TrimeshPrimitiveManager*>(dummy);
+	dummymanager->unlock();
 #endif
 }
 
-
 #define CALC_EXACT_INERTIA 1
 
-
-void btGImpactCompoundShape::calculateLocalInertia(btScalar mass,btVector3& inertia) const
+void btGImpactCompoundShape::calculateLocalInertia(btScalar mass, btVector3& inertia) const
 {
 	lockChildShapes();
 #ifdef CALC_EXACT_INERTIA
-	inertia.setValue(0.f,0.f,0.f);
+	inertia.setValue(0.f, 0.f, 0.f);
 
 	int i = this->getNumChildShapes();
-	btScalar shapemass = mass/btScalar(i);
+	btScalar shapemass = mass / btScalar(i);
 
-	while(i--)
+	while (i--)
 	{
 		btVector3 temp_inertia;
-		m_childShapes[i]->calculateLocalInertia(shapemass,temp_inertia);
-		if(childrenHasTransform())
+		m_childShapes[i]->calculateLocalInertia(shapemass, temp_inertia);
+		if (childrenHasTransform())
 		{
-			inertia = gim_inertia_add_transformed( inertia,temp_inertia,m_childTransforms[i]);
+			inertia = gim_inertia_add_transformed(inertia, temp_inertia, m_childTransforms[i]);
 		}
 		else
 		{
-			inertia = gim_inertia_add_transformed( inertia,temp_inertia,btTransform::getIdentity());
+			inertia = gim_inertia_add_transformed(inertia, temp_inertia, btTransform::getIdentity());
 		}
-
 	}
 
 #else
 
 	// Calc box inertia
 
-	btScalar lx= m_localAABB.m_max[0] - m_localAABB.m_min[0];
-	btScalar ly= m_localAABB.m_max[1] - m_localAABB.m_min[1];
-	btScalar lz= m_localAABB.m_max[2] - m_localAABB.m_min[2];
-	const btScalar x2 = lx*lx;
-	const btScalar y2 = ly*ly;
-	const btScalar z2 = lz*lz;
+	btScalar lx = m_localAABB.m_max[0] - m_localAABB.m_min[0];
+	btScalar ly = m_localAABB.m_max[1] - m_localAABB.m_min[1];
+	btScalar lz = m_localAABB.m_max[2] - m_localAABB.m_min[2];
+	const btScalar x2 = lx * lx;
+	const btScalar y2 = ly * ly;
+	const btScalar z2 = lz * lz;
 	const btScalar scaledmass = mass * btScalar(0.08333333);
 
-	inertia = scaledmass * (btVector3(y2+z2,x2+z2,x2+y2));
+	inertia = scaledmass * (btVector3(y2 + z2, x2 + z2, x2 + y2));
 
 #endif
 	unlockChildShapes();
 }
 
-
-
-void btGImpactMeshShapePart::calculateLocalInertia(btScalar mass,btVector3& inertia) const
+void btGImpactMeshShapePart::calculateLocalInertia(btScalar mass, btVector3& inertia) const
 {
 	lockChildShapes();
 
-
 #ifdef CALC_EXACT_INERTIA
-	inertia.setValue(0.f,0.f,0.f);
+	inertia.setValue(0.f, 0.f, 0.f);
 
 	int i = this->getVertexCount();
-	btScalar pointmass = mass/btScalar(i);
+	btScalar pointmass = mass / btScalar(i);
 
-	while(i--)
+	while (i--)
 	{
 		btVector3 pointintertia;
-		this->getVertex(i,pointintertia);
-		pointintertia = gim_get_point_inertia(pointintertia,pointmass);
-		inertia+=pointintertia;
+		this->getVertex(i, pointintertia);
+		pointintertia = gim_get_point_inertia(pointintertia, pointmass);
+		inertia += pointintertia;
 	}
 
 #else
 
 	// Calc box inertia
 
-	btScalar lx= m_localAABB.m_max[0] - m_localAABB.m_min[0];
-	btScalar ly= m_localAABB.m_max[1] - m_localAABB.m_min[1];
-	btScalar lz= m_localAABB.m_max[2] - m_localAABB.m_min[2];
-	const btScalar x2 = lx*lx;
-	const btScalar y2 = ly*ly;
-	const btScalar z2 = lz*lz;
+	btScalar lx = m_localAABB.m_max[0] - m_localAABB.m_min[0];
+	btScalar ly = m_localAABB.m_max[1] - m_localAABB.m_min[1];
+	btScalar lz = m_localAABB.m_max[2] - m_localAABB.m_min[2];
+	const btScalar x2 = lx * lx;
+	const btScalar y2 = ly * ly;
+	const btScalar z2 = lz * lz;
 	const btScalar scaledmass = mass * btScalar(0.08333333);
 
-	inertia = scaledmass * (btVector3(y2+z2,x2+z2,x2+y2));
+	inertia = scaledmass * (btVector3(y2 + z2, x2 + z2, x2 + y2));
 
 #endif
 
 	unlockChildShapes();
 }
 
-void btGImpactMeshShape::calculateLocalInertia(btScalar mass,btVector3& inertia) const
+void btGImpactMeshShape::calculateLocalInertia(btScalar mass, btVector3& inertia) const
 {
-
 #ifdef CALC_EXACT_INERTIA
-	inertia.setValue(0.f,0.f,0.f);
+	inertia.setValue(0.f, 0.f, 0.f);
 
 	int i = this->getMeshPartCount();
-	btScalar partmass = mass/btScalar(i);
+	btScalar partmass = mass / btScalar(i);
 
-	while(i--)
+	while (i--)
 	{
 		btVector3 partinertia;
-		getMeshPart(i)->calculateLocalInertia(partmass,partinertia);
-		inertia+=partinertia;
+		getMeshPart(i)->calculateLocalInertia(partmass, partinertia);
+		inertia += partinertia;
 	}
 
 #else
 
 	// Calc box inertia
 
-	btScalar lx= m_localAABB.m_max[0] - m_localAABB.m_min[0];
-	btScalar ly= m_localAABB.m_max[1] - m_localAABB.m_min[1];
-	btScalar lz= m_localAABB.m_max[2] - m_localAABB.m_min[2];
-	const btScalar x2 = lx*lx;
-	const btScalar y2 = ly*ly;
-	const btScalar z2 = lz*lz;
+	btScalar lx = m_localAABB.m_max[0] - m_localAABB.m_min[0];
+	btScalar ly = m_localAABB.m_max[1] - m_localAABB.m_min[1];
+	btScalar lz = m_localAABB.m_max[2] - m_localAABB.m_min[2];
+	const btScalar x2 = lx * lx;
+	const btScalar y2 = ly * ly;
+	const btScalar z2 = lz * lz;
 	const btScalar scaledmass = mass * btScalar(0.08333333);
 
-	inertia = scaledmass * (btVector3(y2+z2,x2+z2,x2+y2));
+	inertia = scaledmass * (btVector3(y2 + z2, x2 + z2, x2 + y2));
 
 #endif
 }
@@ -198,7 +189,7 @@ void btGImpactMeshShape::rayTest(const btVector3& rayFrom, const btVector3& rayT
 {
 }
 
-void btGImpactMeshShapePart::processAllTrianglesRay(btTriangleCallback* callback,const btVector3& rayFrom, const btVector3& rayTo) const
+void btGImpactMeshShapePart::processAllTrianglesRay(btTriangleCallback* callback, const btVector3& rayFrom, const btVector3& rayTo) const
 {
 	lockChildShapes();
 
@@ -207,7 +198,7 @@ void btGImpactMeshShapePart::processAllTrianglesRay(btTriangleCallback* callback
 	rayDir.normalize();
 	m_box_set.rayQuery(rayDir, rayFrom, collided);
 
-	if(collided.size()==0)
+	if (collided.size() == 0)
 	{
 		unlockChildShapes();
 		return;
@@ -216,15 +207,15 @@ void btGImpactMeshShapePart::processAllTrianglesRay(btTriangleCallback* callback
 	int part = (int)getPart();
 	btPrimitiveTriangle triangle;
 	int i = collided.size();
-	while(i--)
+	while (i--)
 	{
-		getPrimitiveTriangle(collided[i],triangle);
-		callback->processTriangle(triangle.m_vertices,part,collided[i]);
+		getPrimitiveTriangle(collided[i], triangle);
+		callback->processTriangle(triangle.m_vertices, part, collided[i]);
 	}
 	unlockChildShapes();
 }
 
-void btGImpactMeshShapePart::processAllTriangles(btTriangleCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax) const
+void btGImpactMeshShapePart::processAllTriangles(btTriangleCallback* callback, const btVector3& aabbMin, const btVector3& aabbMax) const
 {
 	lockChildShapes();
 	btAABB box;
@@ -232,9 +223,9 @@ void btGImpactMeshShapePart::processAllTriangles(btTriangleCallback* callback,co
 	box.m_max = aabbMax;
 
 	btAlignedObjectArray<int> collided;
-	m_box_set.boxQuery(box,collided);
+	m_box_set.boxQuery(box, collided);
 
-	if(collided.size()==0)
+	if (collided.size() == 0)
 	{
 		unlockChildShapes();
 		return;
@@ -243,40 +234,38 @@ void btGImpactMeshShapePart::processAllTriangles(btTriangleCallback* callback,co
 	int part = (int)getPart();
 	btPrimitiveTriangle triangle;
 	int i = collided.size();
-	while(i--)
+	while (i--)
 	{
-		this->getPrimitiveTriangle(collided[i],triangle);
-		callback->processTriangle(triangle.m_vertices,part,collided[i]);
+		this->getPrimitiveTriangle(collided[i], triangle);
+		callback->processTriangle(triangle.m_vertices, part, collided[i]);
 	}
 	unlockChildShapes();
-
 }
 
-void btGImpactMeshShape::processAllTriangles(btTriangleCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax) const
+void btGImpactMeshShape::processAllTriangles(btTriangleCallback* callback, const btVector3& aabbMin, const btVector3& aabbMax) const
 {
 	int i = m_mesh_parts.size();
-	while(i--)
+	while (i--)
 	{
-		m_mesh_parts[i]->processAllTriangles(callback,aabbMin,aabbMax);
+		m_mesh_parts[i]->processAllTriangles(callback, aabbMin, aabbMax);
 	}
 }
 
-void btGImpactMeshShape::processAllTrianglesRay(btTriangleCallback* callback,const btVector3& rayFrom, const btVector3& rayTo) const
+void btGImpactMeshShape::processAllTrianglesRay(btTriangleCallback* callback, const btVector3& rayFrom, const btVector3& rayTo) const
 {
 	int i = m_mesh_parts.size();
-	while(i--)
+	while (i--)
 	{
 		m_mesh_parts[i]->processAllTrianglesRay(callback, rayFrom, rayTo);
 	}
 }
 
-
 ///fills the dataBuffer and returns the struct name (and 0 on failure)
-const char*	btGImpactMeshShape::serialize(void* dataBuffer, btSerializer* serializer) const
+const char* btGImpactMeshShape::serialize(void* dataBuffer, btSerializer* serializer) const
 {
-	btGImpactMeshShapeData* trimeshData = (btGImpactMeshShapeData*) dataBuffer;
+	btGImpactMeshShapeData* trimeshData = (btGImpactMeshShapeData*)dataBuffer;
 
-	btCollisionShape::serialize(&trimeshData->m_collisionShapeData,serializer);
+	btCollisionShape::serialize(&trimeshData->m_collisionShapeData, serializer);
 
 	m_meshInterface->serialize(&trimeshData->m_meshInterface, serializer);
 
@@ -288,4 +277,3 @@ const char*	btGImpactMeshShape::serialize(void* dataBuffer, btSerializer* serial
 
 	return "btGImpactMeshShapeData";
 }
-
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactShape.h b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactShape.h
index 9d7e40562c6..5b85e87041b 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btGImpactShape.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btGImpactShape.h
@@ -21,7 +21,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef GIMPACT_SHAPE_H
 #define GIMPACT_SHAPE_H
 
@@ -37,8 +36,7 @@ subject to the following restrictions:
 #include "LinearMath/btMatrix3x3.h"
 #include "LinearMath/btAlignedObjectArray.h"
 
-#include "btGImpactQuantizedBvh.h" // box tree class
-
+#include "btGImpactQuantizedBvh.h"  // box tree class
 
 //! declare Quantized trees, (you can change to float based trees)
 typedef btGImpactQuantizedBvh btGImpactBoxSet;
@@ -50,10 +48,8 @@ enum eGIMPACT_SHAPE_TYPE
 	CONST_GIMPACT_TRIMESH_SHAPE
 };
 
-
-
 //! Helper class for tetrahedrons
-class btTetrahedronShapeEx:public btBU_Simplex1to4
+class btTetrahedronShapeEx : public btBU_Simplex1to4
 {
 public:
 	btTetrahedronShapeEx()
@@ -61,10 +57,9 @@ public:
 		m_numVertices = 4;
 	}
 
-
 	SIMD_FORCE_INLINE void setVertices(
-		const btVector3 & v0,const btVector3 & v1,
-		const btVector3 & v2,const btVector3 & v3)
+		const btVector3& v0, const btVector3& v1,
+		const btVector3& v2, const btVector3& v3)
 	{
 		m_vertices[0] = v0;
 		m_vertices[1] = v1;
@@ -74,45 +69,42 @@ public:
 	}
 };
 
-
 //! Base class for gimpact shapes
 class btGImpactShapeInterface : public btConcaveShape
 {
 protected:
-    btAABB m_localAABB;
-    bool m_needs_update;
-    btVector3  localScaling;
-    btGImpactBoxSet m_box_set;// optionally boxset
+	btAABB m_localAABB;
+	bool m_needs_update;
+	btVector3 localScaling;
+	btGImpactBoxSet m_box_set;  // optionally boxset
 
 	//! use this function for perfofm refit in bounding boxes
-    //! use this function for perfofm refit in bounding boxes
-    virtual void calcLocalAABB()
-    {
+	//! use this function for perfofm refit in bounding boxes
+	virtual void calcLocalAABB()
+	{
 		lockChildShapes();
-    	if(m_box_set.getNodeCount() == 0)
-    	{
-    		m_box_set.buildSet();
-    	}
-    	else
-    	{
-    		m_box_set.update();
-    	}
-    	unlockChildShapes();
-
-    	m_localAABB = m_box_set.getGlobalBox();
-    }
+		if (m_box_set.getNodeCount() == 0)
+		{
+			m_box_set.buildSet();
+		}
+		else
+		{
+			m_box_set.update();
+		}
+		unlockChildShapes();
 
+		m_localAABB = m_box_set.getGlobalBox();
+	}
 
 public:
 	btGImpactShapeInterface()
 	{
-		m_shapeType=GIMPACT_SHAPE_PROXYTYPE;
+		m_shapeType = GIMPACT_SHAPE_PROXYTYPE;
 		m_localAABB.invalidate();
 		m_needs_update = true;
-		localScaling.setValue(1.f,1.f,1.f);
+		localScaling.setValue(1.f, 1.f, 1.f);
 	}
 
-
 	//! performs refit operation
 	/*!
 	Updates the entire Box set of this shape.
@@ -120,47 +112,46 @@ public:
 		will does nothing.
 	\post if m_needs_update == true, then it calls calcLocalAABB();
 	*/
-    SIMD_FORCE_INLINE void updateBound()
-    {
-    	if(!m_needs_update) return;
-    	calcLocalAABB();
-    	m_needs_update  = false;
-    }
+	SIMD_FORCE_INLINE void updateBound()
+	{
+		if (!m_needs_update) return;
+		calcLocalAABB();
+		m_needs_update = false;
+	}
 
-    //! If the Bounding box is not updated, then this class attemps to calculate it.
-    /*!
+	//! If the Bounding box is not updated, then this class attemps to calculate it.
+	/*!
     \post Calls updateBound() for update the box set.
     */
-    void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const
-    {
-        btAABB transformedbox = m_localAABB;
-        transformedbox.appy_transform(t);
-        aabbMin = transformedbox.m_min;
-        aabbMax = transformedbox.m_max;
-    }
+	void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
+	{
+		btAABB transformedbox = m_localAABB;
+		transformedbox.appy_transform(t);
+		aabbMin = transformedbox.m_min;
+		aabbMax = transformedbox.m_max;
+	}
 
-    //! Tells to this object that is needed to refit the box set
-    virtual void postUpdate()
-    {
-    	m_needs_update = true;
-    }
+	//! Tells to this object that is needed to refit the box set
+	virtual void postUpdate()
+	{
+		m_needs_update = true;
+	}
 
 	//! Obtains the local box, which is the global calculated box of the total of subshapes
-	SIMD_FORCE_INLINE const btAABB & getLocalBox()
+	SIMD_FORCE_INLINE const btAABB& getLocalBox()
 	{
 		return m_localAABB;
 	}
 
+	virtual int getShapeType() const
+	{
+		return GIMPACT_SHAPE_PROXYTYPE;
+	}
 
-    virtual int	getShapeType() const
-    {
-        return GIMPACT_SHAPE_PROXYTYPE;
-    }
-
-    /*!
+	/*!
 	\post You must call updateBound() for update the box set.
 	*/
-	virtual void	setLocalScaling(const btVector3& scaling)
+	virtual void setLocalScaling(const btVector3& scaling)
 	{
 		localScaling = scaling;
 		postUpdate();
@@ -171,46 +162,43 @@ public:
 		return localScaling;
 	}
 
-
 	virtual void setMargin(btScalar margin)
-    {
-    	m_collisionMargin = margin;
-    	int i = getNumChildShapes();
-    	while(i--)
-    	{
+	{
+		m_collisionMargin = margin;
+		int i = getNumChildShapes();
+		while (i--)
+		{
 			btCollisionShape* child = getChildShape(i);
 			child->setMargin(margin);
-    	}
+		}
 
 		m_needs_update = true;
-    }
-
+	}
 
 	//! Subshape member functions
 	//!@{
 
 	//! Base method for determinig which kind of GIMPACT shape we get
-	virtual eGIMPACT_SHAPE_TYPE getGImpactShapeType() const = 0 ;
+	virtual eGIMPACT_SHAPE_TYPE getGImpactShapeType() const = 0;
 
 	//! gets boxset
-	SIMD_FORCE_INLINE const btGImpactBoxSet * getBoxSet() const
+	SIMD_FORCE_INLINE const btGImpactBoxSet* getBoxSet() const
 	{
 		return &m_box_set;
 	}
 
 	//! Determines if this class has a hierarchy structure for sorting its primitives
-	SIMD_FORCE_INLINE bool hasBoxSet()  const
+	SIMD_FORCE_INLINE bool hasBoxSet() const
 	{
-		if(m_box_set.getNodeCount() == 0) return false;
+		if (m_box_set.getNodeCount() == 0) return false;
 		return true;
 	}
 
 	//! Obtains the primitive manager
-	virtual const btPrimitiveManagerBase * getPrimitiveManager()  const = 0;
-
+	virtual const btPrimitiveManagerBase* getPrimitiveManager() const = 0;
 
 	//! Gets the number of children
-	virtual int	getNumChildShapes() const  = 0;
+	virtual int getNumChildShapes() const = 0;
 
 	//! if true, then its children must get transforms.
 	virtual bool childrenHasTransform() const = 0;
@@ -221,11 +209,9 @@ public:
 	//! Determines if this shape has tetrahedrons
 	virtual bool needsRetrieveTetrahedrons() const = 0;
 
-	virtual void getBulletTriangle(int prim_index,btTriangleShapeEx & triangle) const = 0;
-
-	virtual void getBulletTetrahedron(int prim_index,btTetrahedronShapeEx & tetrahedron) const = 0;
-
+	virtual void getBulletTriangle(int prim_index, btTriangleShapeEx& triangle) const = 0;
 
+	virtual void getBulletTetrahedron(int prim_index, btTetrahedronShapeEx& tetrahedron) const = 0;
 
 	//! call when reading child shapes
 	virtual void lockChildShapes() const
@@ -237,94 +223,91 @@ public:
 	}
 
 	//! if this trimesh
-	SIMD_FORCE_INLINE void getPrimitiveTriangle(int index,btPrimitiveTriangle & triangle) const
+	SIMD_FORCE_INLINE void getPrimitiveTriangle(int index, btPrimitiveTriangle& triangle) const
 	{
-		getPrimitiveManager()->get_primitive_triangle(index,triangle);
+		getPrimitiveManager()->get_primitive_triangle(index, triangle);
 	}
 
-
 	//! Retrieves the bound from a child
-    /*!
+	/*!
     */
-    virtual void getChildAabb(int child_index,const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const
-    {
-        btAABB child_aabb;
-        getPrimitiveManager()->get_primitive_box(child_index,child_aabb);
-        child_aabb.appy_transform(t);
-        aabbMin = child_aabb.m_min;
-        aabbMax = child_aabb.m_max;
-    }
+	virtual void getChildAabb(int child_index, const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
+	{
+		btAABB child_aabb;
+		getPrimitiveManager()->get_primitive_box(child_index, child_aabb);
+		child_aabb.appy_transform(t);
+		aabbMin = child_aabb.m_min;
+		aabbMax = child_aabb.m_max;
+	}
 
 	//! Gets the children
 	virtual btCollisionShape* getChildShape(int index) = 0;
 
-
 	//! Gets the child
 	virtual const btCollisionShape* getChildShape(int index) const = 0;
 
 	//! Gets the children transform
-	virtual btTransform	getChildTransform(int index) const = 0;
+	virtual btTransform getChildTransform(int index) const = 0;
 
 	//! Sets the children transform
 	/*!
 	\post You must call updateBound() for update the box set.
 	*/
-	virtual void setChildTransform(int index, const btTransform & transform) = 0;
+	virtual void setChildTransform(int index, const btTransform& transform) = 0;
 
 	//!@}
 
-
 	//! virtual method for ray collision
-	virtual void rayTest(const btVector3& rayFrom, const btVector3& rayTo, btCollisionWorld::RayResultCallback& resultCallback)  const
+	virtual void rayTest(const btVector3& rayFrom, const btVector3& rayTo, btCollisionWorld::RayResultCallback& resultCallback) const
 	{
-        (void) rayFrom; (void) rayTo; (void) resultCallback;
+		(void)rayFrom;
+		(void)rayTo;
+		(void)resultCallback;
 	}
 
 	//! Function for retrieve triangles.
 	/*!
 	It gives the triangles in local space
 	*/
-	virtual void	processAllTriangles(btTriangleCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax) const
+	virtual void processAllTriangles(btTriangleCallback* callback, const btVector3& aabbMin, const btVector3& aabbMax) const
 	{
-        (void) callback; (void) aabbMin; (void) aabbMax;
+		(void)callback;
+		(void)aabbMin;
+		(void)aabbMax;
 	}
 
 	//! Function for retrieve triangles.
 	/*!
 	It gives the triangles in local space
 	*/
-	virtual void processAllTrianglesRay(btTriangleCallback* /*callback*/,const btVector3& /*rayFrom*/, const btVector3& /*rayTo*/) const
+	virtual void processAllTrianglesRay(btTriangleCallback* /*callback*/, const btVector3& /*rayFrom*/, const btVector3& /*rayTo*/) const
 	{
-		
 	}
 
 	//!@}
-
 };
 
-
 //! btGImpactCompoundShape allows to handle multiple btCollisionShape objects at once
 /*!
 This class only can manage Convex subshapes
 */
-class btGImpactCompoundShape	: public btGImpactShapeInterface
+class btGImpactCompoundShape : public btGImpactShapeInterface
 {
 public:
 	//! compound primitive manager
-	class CompoundPrimitiveManager:public btPrimitiveManagerBase
+	class CompoundPrimitiveManager : public btPrimitiveManagerBase
 	{
 	public:
 		virtual ~CompoundPrimitiveManager() {}
-		btGImpactCompoundShape * m_compoundShape;
-
+		btGImpactCompoundShape* m_compoundShape;
 
 		CompoundPrimitiveManager(const CompoundPrimitiveManager& compound)
-            : btPrimitiveManagerBase()
+			: btPrimitiveManagerBase()
 		{
 			m_compoundShape = compound.m_compoundShape;
 		}
 
-		CompoundPrimitiveManager(btGImpactCompoundShape * compoundShape)
+		CompoundPrimitiveManager(btGImpactCompoundShape* compoundShape)
 		{
 			m_compoundShape = compoundShape;
 		}
@@ -341,13 +324,13 @@ public:
 
 		virtual int get_primitive_count() const
 		{
-			return (int )m_compoundShape->getNumChildShapes();
+			return (int)m_compoundShape->getNumChildShapes();
 		}
 
-		virtual void get_primitive_box(int prim_index ,btAABB & primbox) const
+		virtual void get_primitive_box(int prim_index, btAABB& primbox) const
 		{
 			btTransform prim_trans;
-			if(m_compoundShape->childrenHasTransform())
+			if (m_compoundShape->childrenHasTransform())
 			{
 				prim_trans = m_compoundShape->getChildTransform(prim_index);
 			}
@@ -356,30 +339,26 @@ public:
 				prim_trans.setIdentity();
 			}
 			const btCollisionShape* shape = m_compoundShape->getChildShape(prim_index);
-			shape->getAabb(prim_trans,primbox.m_min,primbox.m_max);
+			shape->getAabb(prim_trans, primbox.m_min, primbox.m_max);
 		}
 
-		virtual void get_primitive_triangle(int prim_index,btPrimitiveTriangle & triangle) const
+		virtual void get_primitive_triangle(int prim_index, btPrimitiveTriangle& triangle) const
 		{
 			btAssert(0);
-            (void) prim_index; (void) triangle;
+			(void)prim_index;
+			(void)triangle;
 		}
-
 	};
 
-
-
 protected:
 	CompoundPrimitiveManager m_primitive_manager;
-	btAlignedObjectArray<btTransform>		m_childTransforms;
-	btAlignedObjectArray<btCollisionShape*>	m_childShapes;
-
+	btAlignedObjectArray<btTransform> m_childTransforms;
+	btAlignedObjectArray<btCollisionShape*> m_childShapes;
 
 public:
-
 	btGImpactCompoundShape(bool children_has_transform = true)
 	{
-        (void) children_has_transform;
+		(void)children_has_transform;
 		m_primitive_manager.m_compoundShape = this;
 		m_box_set.setPrimitiveManager(&m_primitive_manager);
 	}
@@ -388,36 +367,33 @@ public:
 	{
 	}
 
-
 	//! if true, then its children must get transforms.
 	virtual bool childrenHasTransform() const
 	{
-		if(m_childTransforms.size()==0) return false;
+		if (m_childTransforms.size() == 0) return false;
 		return true;
 	}
 
-
 	//! Obtains the primitive manager
-	virtual const btPrimitiveManagerBase * getPrimitiveManager()  const
+	virtual const btPrimitiveManagerBase* getPrimitiveManager() const
 	{
 		return &m_primitive_manager;
 	}
 
 	//! Obtains the compopund primitive manager
-	SIMD_FORCE_INLINE CompoundPrimitiveManager * getCompoundPrimitiveManager()
+	SIMD_FORCE_INLINE CompoundPrimitiveManager* getCompoundPrimitiveManager()
 	{
 		return &m_primitive_manager;
 	}
 
 	//! Gets the number of children
-	virtual int	getNumChildShapes() const
+	virtual int getNumChildShapes() const
 	{
 		return m_childShapes.size();
 	}
 
-
 	//! Use this method for adding children. Only Convex shapes are allowed.
-	void addChildShape(const btTransform& localTransform,btCollisionShape* shape)
+	void addChildShape(const btTransform& localTransform, btCollisionShape* shape)
 	{
 		btAssert(shape->isConvex());
 		m_childTransforms.push_back(localTransform);
@@ -444,24 +420,22 @@ public:
 	}
 
 	//! Retrieves the bound from a child
-    /*!
+	/*!
     */
-    virtual void getChildAabb(int child_index,const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const
-    {
-
-    	if(childrenHasTransform())
-    	{
-    		m_childShapes[child_index]->getAabb(t*m_childTransforms[child_index],aabbMin,aabbMax);
-    	}
-    	else
-    	{
-    		m_childShapes[child_index]->getAabb(t,aabbMin,aabbMax);
-    	}
-    }
-
+	virtual void getChildAabb(int child_index, const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
+	{
+		if (childrenHasTransform())
+		{
+			m_childShapes[child_index]->getAabb(t * m_childTransforms[child_index], aabbMin, aabbMax);
+		}
+		else
+		{
+			m_childShapes[child_index]->getAabb(t, aabbMin, aabbMax);
+		}
+	}
 
 	//! Gets the children transform
-	virtual btTransform	getChildTransform(int index) const
+	virtual btTransform getChildTransform(int index) const
 	{
 		btAssert(m_childTransforms.size() == m_childShapes.size());
 		return m_childTransforms[index];
@@ -471,7 +445,7 @@ public:
 	/*!
 	\post You must call updateBound() for update the box set.
 	*/
-	virtual void setChildTransform(int index, const btTransform & transform)
+	virtual void setChildTransform(int index, const btTransform& transform)
 	{
 		btAssert(m_childTransforms.size() == m_childShapes.size());
 		m_childTransforms[index] = transform;
@@ -490,24 +464,24 @@ public:
 		return false;
 	}
 
-
-	virtual void getBulletTriangle(int prim_index,btTriangleShapeEx & triangle) const
+	virtual void getBulletTriangle(int prim_index, btTriangleShapeEx& triangle) const
 	{
-        (void) prim_index; (void) triangle;
+		(void)prim_index;
+		(void)triangle;
 		btAssert(0);
 	}
 
-	virtual void getBulletTetrahedron(int prim_index,btTetrahedronShapeEx & tetrahedron) const
+	virtual void getBulletTetrahedron(int prim_index, btTetrahedronShapeEx& tetrahedron) const
 	{
-        (void) prim_index; (void) tetrahedron;
+		(void)prim_index;
+		(void)tetrahedron;
 		btAssert(0);
 	}
 
-
 	//! Calculates the exact inertia tensor for this shape
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
+	virtual void calculateLocalInertia(btScalar mass, btVector3& inertia) const;
 
-	virtual const char*	getName()const
+	virtual const char* getName() const
 	{
 		return "GImpactCompound";
 	}
@@ -516,11 +490,8 @@ public:
 	{
 		return CONST_GIMPACT_COMPOUND_SHAPE;
 	}
-
 };
 
-
-
 //! This class manages a sub part of a mesh supplied by the btStridingMeshInterface interface.
 /*!
 - Simply create this shape by passing the btStridingMeshInterface to the constructor btGImpactMeshShapePart, then you must call updateBound() after creating the mesh
@@ -535,21 +506,21 @@ public:
 	/*!
 	Manages the info from btStridingMeshInterface object and controls the Lock/Unlock mechanism
 	*/
-	class TrimeshPrimitiveManager:public btPrimitiveManagerBase
+	class TrimeshPrimitiveManager : public btPrimitiveManagerBase
 	{
 	public:
 		btScalar m_margin;
-		btStridingMeshInterface * m_meshInterface;
+		btStridingMeshInterface* m_meshInterface;
 		btVector3 m_scale;
 		int m_part;
 		int m_lock_count;
-		const unsigned char *vertexbase;
+		const unsigned char* vertexbase;
 		int numverts;
 		PHY_ScalarType type;
 		int stride;
-		const unsigned char *indexbase;
+		const unsigned char* indexbase;
 		int indexstride;
-		int  numfaces;
+		int numfaces;
 		PHY_ScalarType indicestype;
 
 		TrimeshPrimitiveManager()
@@ -557,7 +528,7 @@ public:
 			m_meshInterface = NULL;
 			m_part = 0;
 			m_margin = 0.01f;
-			m_scale = btVector3(1.f,1.f,1.f);
+			m_scale = btVector3(1.f, 1.f, 1.f);
 			m_lock_count = 0;
 			vertexbase = 0;
 			numverts = 0;
@@ -567,8 +538,8 @@ public:
 			numfaces = 0;
 		}
 
- 		TrimeshPrimitiveManager(const TrimeshPrimitiveManager & manager)
-            : btPrimitiveManagerBase()
+		TrimeshPrimitiveManager(const TrimeshPrimitiveManager& manager)
+			: btPrimitiveManagerBase()
 		{
 			m_meshInterface = manager.m_meshInterface;
 			m_part = manager.m_part;
@@ -581,11 +552,10 @@ public:
 			indexbase = 0;
 			indexstride = 0;
 			numfaces = 0;
-
 		}
 
 		TrimeshPrimitiveManager(
-			btStridingMeshInterface * meshInterface,	int part)
+			btStridingMeshInterface* meshInterface, int part)
 		{
 			m_meshInterface = meshInterface;
 			m_part = part;
@@ -598,29 +568,28 @@ public:
 			indexbase = 0;
 			indexstride = 0;
 			numfaces = 0;
-
 		}
 
 		virtual ~TrimeshPrimitiveManager() {}
 
 		void lock()
 		{
-			if(m_lock_count>0)
+			if (m_lock_count > 0)
 			{
 				m_lock_count++;
 				return;
 			}
 			m_meshInterface->getLockedReadOnlyVertexIndexBase(
-				&vertexbase,numverts,
-				type, stride,&indexbase, indexstride, numfaces,indicestype,m_part);
+				&vertexbase, numverts,
+				type, stride, &indexbase, indexstride, numfaces, indicestype, m_part);
 
 			m_lock_count = 1;
 		}
 
 		void unlock()
 		{
-			if(m_lock_count == 0) return;
-			if(m_lock_count>1)
+			if (m_lock_count == 0) return;
+			if (m_lock_count > 1)
 			{
 				--m_lock_count;
 				return;
@@ -637,93 +606,91 @@ public:
 
 		virtual int get_primitive_count() const
 		{
-			return (int )numfaces;
+			return (int)numfaces;
 		}
 
 		SIMD_FORCE_INLINE int get_vertex_count() const
 		{
-			return (int )numverts;
+			return (int)numverts;
 		}
 
-		SIMD_FORCE_INLINE void get_indices(int face_index,unsigned int &i0,unsigned int &i1,unsigned int &i2) const
+		SIMD_FORCE_INLINE void get_indices(int face_index, unsigned int& i0, unsigned int& i1, unsigned int& i2) const
 		{
-			if(indicestype == PHY_SHORT)
+			if (indicestype == PHY_SHORT)
 			{
-				unsigned short* s_indices = (unsigned short *)(indexbase + face_index * indexstride);
+				unsigned short* s_indices = (unsigned short*)(indexbase + face_index * indexstride);
 				i0 = s_indices[0];
 				i1 = s_indices[1];
 				i2 = s_indices[2];
 			}
 			else
 			{
-				unsigned int * i_indices = (unsigned int *)(indexbase + face_index*indexstride);
+				unsigned int* i_indices = (unsigned int*)(indexbase + face_index * indexstride);
 				i0 = i_indices[0];
 				i1 = i_indices[1];
 				i2 = i_indices[2];
 			}
 		}
 
-		SIMD_FORCE_INLINE void get_vertex(unsigned int vertex_index, btVector3 & vertex) const
+		SIMD_FORCE_INLINE void get_vertex(unsigned int vertex_index, btVector3& vertex) const
 		{
-			if(type == PHY_DOUBLE)
+			if (type == PHY_DOUBLE)
 			{
-				double * dvertices = (double *)(vertexbase + vertex_index*stride);
-				vertex[0] = btScalar(dvertices[0]*m_scale[0]);
-				vertex[1] = btScalar(dvertices[1]*m_scale[1]);
-				vertex[2] = btScalar(dvertices[2]*m_scale[2]);
+				double* dvertices = (double*)(vertexbase + vertex_index * stride);
+				vertex[0] = btScalar(dvertices[0] * m_scale[0]);
+				vertex[1] = btScalar(dvertices[1] * m_scale[1]);
+				vertex[2] = btScalar(dvertices[2] * m_scale[2]);
 			}
 			else
 			{
-				float * svertices = (float *)(vertexbase + vertex_index*stride);
-				vertex[0] = svertices[0]*m_scale[0];
-				vertex[1] = svertices[1]*m_scale[1];
-				vertex[2] = svertices[2]*m_scale[2];
+				float* svertices = (float*)(vertexbase + vertex_index * stride);
+				vertex[0] = svertices[0] * m_scale[0];
+				vertex[1] = svertices[1] * m_scale[1];
+				vertex[2] = svertices[2] * m_scale[2];
 			}
 		}
 
-		virtual void get_primitive_box(int prim_index ,btAABB & primbox) const
+		virtual void get_primitive_box(int prim_index, btAABB& primbox) const
 		{
-			btPrimitiveTriangle  triangle;
-			get_primitive_triangle(prim_index,triangle);
+			btPrimitiveTriangle triangle;
+			get_primitive_triangle(prim_index, triangle);
 			primbox.calc_from_triangle_margin(
 				triangle.m_vertices[0],
-				triangle.m_vertices[1],triangle.m_vertices[2],triangle.m_margin);
+				triangle.m_vertices[1], triangle.m_vertices[2], triangle.m_margin);
 		}
 
-		virtual void get_primitive_triangle(int prim_index,btPrimitiveTriangle & triangle) const
+		virtual void get_primitive_triangle(int prim_index, btPrimitiveTriangle& triangle) const
 		{
 			unsigned int indices[3];
-			get_indices(prim_index,indices[0],indices[1],indices[2]);
-			get_vertex(indices[0],triangle.m_vertices[0]);
-			get_vertex(indices[1],triangle.m_vertices[1]);
-			get_vertex(indices[2],triangle.m_vertices[2]);
+			get_indices(prim_index, indices[0], indices[1], indices[2]);
+			get_vertex(indices[0], triangle.m_vertices[0]);
+			get_vertex(indices[1], triangle.m_vertices[1]);
+			get_vertex(indices[2], triangle.m_vertices[2]);
 			triangle.m_margin = m_margin;
 		}
 
-		SIMD_FORCE_INLINE void get_bullet_triangle(int prim_index,btTriangleShapeEx & triangle) const
+		SIMD_FORCE_INLINE void get_bullet_triangle(int prim_index, btTriangleShapeEx& triangle) const
 		{
 			unsigned int indices[3];
-			get_indices(prim_index,indices[0],indices[1],indices[2]);
-			get_vertex(indices[0],triangle.m_vertices1[0]);
-			get_vertex(indices[1],triangle.m_vertices1[1]);
-			get_vertex(indices[2],triangle.m_vertices1[2]);
+			get_indices(prim_index, indices[0], indices[1], indices[2]);
+			get_vertex(indices[0], triangle.m_vertices1[0]);
+			get_vertex(indices[1], triangle.m_vertices1[1]);
+			get_vertex(indices[2], triangle.m_vertices1[2]);
 			triangle.setMargin(m_margin);
 		}
-
 	};
 
-
 protected:
 	TrimeshPrimitiveManager m_primitive_manager;
-public:
 
+public:
 	btGImpactMeshShapePart()
 	{
 		m_box_set.setPrimitiveManager(&m_primitive_manager);
 	}
 
-    btGImpactMeshShapePart( btStridingMeshInterface * meshInterface, int part );
-    virtual ~btGImpactMeshShapePart();
+	btGImpactMeshShapePart(btStridingMeshInterface* meshInterface, int part);
+	virtual ~btGImpactMeshShapePart();
 
 	//! if true, then its children must get transforms.
 	virtual bool childrenHasTransform() const
@@ -731,40 +698,36 @@ public:
 		return false;
 	}
 
-
 	//! call when reading child shapes
-    virtual void lockChildShapes() const;
-    virtual void unlockChildShapes()  const;
+	virtual void lockChildShapes() const;
+	virtual void unlockChildShapes() const;
 
 	//! Gets the number of children
-	virtual int	getNumChildShapes() const
+	virtual int getNumChildShapes() const
 	{
 		return m_primitive_manager.get_primitive_count();
 	}
 
-
 	//! Gets the children
 	virtual btCollisionShape* getChildShape(int index)
 	{
-        (void) index;
+		(void)index;
 		btAssert(0);
 		return NULL;
 	}
 
-
-
 	//! Gets the child
 	virtual const btCollisionShape* getChildShape(int index) const
 	{
-        (void) index;
+		(void)index;
 		btAssert(0);
 		return NULL;
 	}
 
 	//! Gets the children transform
-	virtual btTransform	getChildTransform(int index) const
+	virtual btTransform getChildTransform(int index) const
 	{
-        (void) index;
+		(void)index;
 		btAssert(0);
 		return btTransform();
 	}
@@ -773,35 +736,27 @@ public:
 	/*!
 	\post You must call updateBound() for update the box set.
 	*/
-	virtual void setChildTransform(int index, const btTransform & transform)
+	virtual void setChildTransform(int index, const btTransform& transform)
 	{
-        (void) index;
-        (void) transform;
+		(void)index;
+		(void)transform;
 		btAssert(0);
 	}
 
-
 	//! Obtains the primitive manager
-	virtual const btPrimitiveManagerBase * getPrimitiveManager()  const
+	virtual const btPrimitiveManagerBase* getPrimitiveManager() const
 	{
 		return &m_primitive_manager;
 	}
 
-	SIMD_FORCE_INLINE TrimeshPrimitiveManager * getTrimeshPrimitiveManager()
+	SIMD_FORCE_INLINE TrimeshPrimitiveManager* getTrimeshPrimitiveManager()
 	{
 		return &m_primitive_manager;
 	}
 
+	virtual void calculateLocalInertia(btScalar mass, btVector3& inertia) const;
 
-
-
-
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
-
-
-
-
-	virtual const char*	getName()const
+	virtual const char* getName() const
 	{
 		return "GImpactMeshShapePart";
 	}
@@ -823,62 +778,59 @@ public:
 		return false;
 	}
 
-	virtual void getBulletTriangle(int prim_index,btTriangleShapeEx & triangle) const
+	virtual void getBulletTriangle(int prim_index, btTriangleShapeEx& triangle) const
 	{
-		m_primitive_manager.get_bullet_triangle(prim_index,triangle);
+		m_primitive_manager.get_bullet_triangle(prim_index, triangle);
 	}
 
-	virtual void getBulletTetrahedron(int prim_index,btTetrahedronShapeEx & tetrahedron) const
+	virtual void getBulletTetrahedron(int prim_index, btTetrahedronShapeEx& tetrahedron) const
 	{
-        (void) prim_index;
-        (void) tetrahedron;
+		(void)prim_index;
+		(void)tetrahedron;
 		btAssert(0);
 	}
 
-
-
 	SIMD_FORCE_INLINE int getVertexCount() const
 	{
 		return m_primitive_manager.get_vertex_count();
 	}
 
-	SIMD_FORCE_INLINE void getVertex(int vertex_index, btVector3 & vertex) const
+	SIMD_FORCE_INLINE void getVertex(int vertex_index, btVector3& vertex) const
 	{
-		m_primitive_manager.get_vertex(vertex_index,vertex);
+		m_primitive_manager.get_vertex(vertex_index, vertex);
 	}
 
 	SIMD_FORCE_INLINE void setMargin(btScalar margin)
-    {
-    	m_primitive_manager.m_margin = margin;
-    	postUpdate();
-    }
+	{
+		m_primitive_manager.m_margin = margin;
+		postUpdate();
+	}
 
-    SIMD_FORCE_INLINE btScalar getMargin() const
-    {
-    	return m_primitive_manager.m_margin;
-    }
+	SIMD_FORCE_INLINE btScalar getMargin() const
+	{
+		return m_primitive_manager.m_margin;
+	}
 
-    virtual void	setLocalScaling(const btVector3& scaling)
-    {
-    	m_primitive_manager.m_scale = scaling;
-    	postUpdate();
-    }
+	virtual void setLocalScaling(const btVector3& scaling)
+	{
+		m_primitive_manager.m_scale = scaling;
+		postUpdate();
+	}
 
-    virtual const btVector3& getLocalScaling() const
-    {
-    	return m_primitive_manager.m_scale;
-    }
+	virtual const btVector3& getLocalScaling() const
+	{
+		return m_primitive_manager.m_scale;
+	}
 
-    SIMD_FORCE_INLINE int getPart() const
-    {
-    	return (int)m_primitive_manager.m_part;
-    }
+	SIMD_FORCE_INLINE int getPart() const
+	{
+		return (int)m_primitive_manager.m_part;
+	}
 
-	virtual void	processAllTriangles(btTriangleCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax) const;
-	virtual void	processAllTrianglesRay(btTriangleCallback* callback,const btVector3& rayFrom,const btVector3& rayTo) const;
+	virtual void processAllTriangles(btTriangleCallback* callback, const btVector3& aabbMin, const btVector3& aabbMax) const;
+	virtual void processAllTrianglesRay(btTriangleCallback* callback, const btVector3& rayFrom, const btVector3& rayTo) const;
 };
 
-
 //! This class manages a mesh supplied by the btStridingMeshInterface interface.
 /*!
 Set of btGImpactMeshShapePart parts
@@ -893,29 +845,29 @@ class btGImpactMeshShape : public btGImpactShapeInterface
 
 protected:
 	btAlignedObjectArray<btGImpactMeshShapePart*> m_mesh_parts;
-	void buildMeshParts(btStridingMeshInterface * meshInterface)
+	void buildMeshParts(btStridingMeshInterface* meshInterface)
 	{
-		for (int i=0;i<meshInterface->getNumSubParts() ;++i )
+		for (int i = 0; i < meshInterface->getNumSubParts(); ++i)
 		{
-			btGImpactMeshShapePart * newpart = new btGImpactMeshShapePart(meshInterface,i);
+			btGImpactMeshShapePart* newpart = new btGImpactMeshShapePart(meshInterface, i);
 			m_mesh_parts.push_back(newpart);
 		}
 	}
 
 	//! use this function for perfofm refit in bounding boxes
-    virtual void calcLocalAABB()
-    {
-    	m_localAABB.invalidate();
-    	int i = m_mesh_parts.size();
-    	while(i--)
-    	{
-    		m_mesh_parts[i]->updateBound();
-    		m_localAABB.merge(m_mesh_parts[i]->getLocalBox());
-    	}
-    }
+	virtual void calcLocalAABB()
+	{
+		m_localAABB.invalidate();
+		int i = m_mesh_parts.size();
+		while (i--)
+		{
+			m_mesh_parts[i]->updateBound();
+			m_localAABB.merge(m_mesh_parts[i]->getLocalBox());
+		}
+	}
 
 public:
-	btGImpactMeshShape(btStridingMeshInterface * meshInterface)
+	btGImpactMeshShape(btStridingMeshInterface* meshInterface)
 	{
 		m_meshInterface = meshInterface;
 		buildMeshParts(meshInterface);
@@ -924,15 +876,14 @@ public:
 	virtual ~btGImpactMeshShape()
 	{
 		int i = m_mesh_parts.size();
-    	while(i--)
-    	{
-			btGImpactMeshShapePart * part = m_mesh_parts[i];
+		while (i--)
+		{
+			btGImpactMeshShapePart* part = m_mesh_parts[i];
 			delete part;
-    	}
+		}
 		m_mesh_parts.clear();
 	}
 
-
 	btStridingMeshInterface* getMeshInterface()
 	{
 		return m_meshInterface;
@@ -948,79 +899,73 @@ public:
 		return m_mesh_parts.size();
 	}
 
-	btGImpactMeshShapePart * getMeshPart(int index)
+	btGImpactMeshShapePart* getMeshPart(int index)
 	{
 		return m_mesh_parts[index];
 	}
 
-
-
-	const btGImpactMeshShapePart * getMeshPart(int index) const
+	const btGImpactMeshShapePart* getMeshPart(int index) const
 	{
 		return m_mesh_parts[index];
 	}
 
-
-	virtual void	setLocalScaling(const btVector3& scaling)
+	virtual void setLocalScaling(const btVector3& scaling)
 	{
 		localScaling = scaling;
 
 		int i = m_mesh_parts.size();
-    	while(i--)
-    	{
-			btGImpactMeshShapePart * part = m_mesh_parts[i];
+		while (i--)
+		{
+			btGImpactMeshShapePart* part = m_mesh_parts[i];
 			part->setLocalScaling(scaling);
-    	}
+		}
 
 		m_needs_update = true;
 	}
 
 	virtual void setMargin(btScalar margin)
-    {
-    	m_collisionMargin = margin;
+	{
+		m_collisionMargin = margin;
 
 		int i = m_mesh_parts.size();
-    	while(i--)
-    	{
-			btGImpactMeshShapePart * part = m_mesh_parts[i];
+		while (i--)
+		{
+			btGImpactMeshShapePart* part = m_mesh_parts[i];
 			part->setMargin(margin);
-    	}
+		}
 
 		m_needs_update = true;
-    }
+	}
 
 	//! Tells to this object that is needed to refit all the meshes
-    virtual void postUpdate()
-    {
+	virtual void postUpdate()
+	{
 		int i = m_mesh_parts.size();
-    	while(i--)
-    	{
-			btGImpactMeshShapePart * part = m_mesh_parts[i];
+		while (i--)
+		{
+			btGImpactMeshShapePart* part = m_mesh_parts[i];
 			part->postUpdate();
-    	}
+		}
 
-    	m_needs_update = true;
-    }
-
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const;
+		m_needs_update = true;
+	}
 
+	virtual void calculateLocalInertia(btScalar mass, btVector3& inertia) const;
 
 	//! Obtains the primitive manager
-	virtual const btPrimitiveManagerBase * getPrimitiveManager()  const
+	virtual const btPrimitiveManagerBase* getPrimitiveManager() const
 	{
 		btAssert(0);
 		return NULL;
 	}
 
-
 	//! Gets the number of children
-	virtual int	getNumChildShapes() const
+	virtual int getNumChildShapes() const
 	{
 		btAssert(0);
 		return 0;
 	}
 
-
 	//! if true, then its children must get transforms.
 	virtual bool childrenHasTransform() const
 	{
@@ -1042,15 +987,17 @@ public:
 		return false;
 	}
 
-	virtual void getBulletTriangle(int prim_index,btTriangleShapeEx & triangle) const
+	virtual void getBulletTriangle(int prim_index, btTriangleShapeEx& triangle) const
 	{
-        (void) prim_index; (void) triangle;
+		(void)prim_index;
+		(void)triangle;
 		btAssert(0);
 	}
 
-	virtual void getBulletTetrahedron(int prim_index,btTetrahedronShapeEx & tetrahedron) const
+	virtual void getBulletTetrahedron(int prim_index, btTetrahedronShapeEx& tetrahedron) const
 	{
-        (void) prim_index; (void) tetrahedron;
+		(void)prim_index;
+		(void)tetrahedron;
 		btAssert(0);
 	}
 
@@ -1065,39 +1012,38 @@ public:
 		btAssert(0);
 	}
 
-
-
-
 	//! Retrieves the bound from a child
-    /*!
+	/*!
     */
-    virtual void getChildAabb(int child_index,const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const
-    {
-        (void) child_index; (void) t; (void) aabbMin; (void) aabbMax;
-        btAssert(0);
-    }
+	virtual void getChildAabb(int child_index, const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
+	{
+		(void)child_index;
+		(void)t;
+		(void)aabbMin;
+		(void)aabbMax;
+		btAssert(0);
+	}
 
 	//! Gets the children
 	virtual btCollisionShape* getChildShape(int index)
 	{
-        (void) index;
+		(void)index;
 		btAssert(0);
 		return NULL;
 	}
 
-
 	//! Gets the child
 	virtual const btCollisionShape* getChildShape(int index) const
 	{
-        (void) index;
+		(void)index;
 		btAssert(0);
 		return NULL;
 	}
 
 	//! Gets the children transform
-	virtual btTransform	getChildTransform(int index) const
+	virtual btTransform getChildTransform(int index) const
 	{
-        (void) index;
+		(void)index;
 		btAssert(0);
 		return btTransform();
 	}
@@ -1106,59 +1052,56 @@ public:
 	/*!
 	\post You must call updateBound() for update the box set.
 	*/
-	virtual void setChildTransform(int index, const btTransform & transform)
+	virtual void setChildTransform(int index, const btTransform& transform)
 	{
-        (void) index; (void) transform;
+		(void)index;
+		(void)transform;
 		btAssert(0);
 	}
 
-
 	virtual eGIMPACT_SHAPE_TYPE getGImpactShapeType() const
 	{
 		return CONST_GIMPACT_TRIMESH_SHAPE;
 	}
 
-
-	virtual const char*	getName()const
+	virtual const char* getName() const
 	{
 		return "GImpactMesh";
 	}
 
-	virtual void rayTest(const btVector3& rayFrom, const btVector3& rayTo, btCollisionWorld::RayResultCallback& resultCallback)  const;
+	virtual void rayTest(const btVector3& rayFrom, const btVector3& rayTo, btCollisionWorld::RayResultCallback& resultCallback) const;
 
 	//! Function for retrieve triangles.
 	/*!
 	It gives the triangles in local space
 	*/
-	virtual void	processAllTriangles(btTriangleCallback* callback,const btVector3& aabbMin,const btVector3& aabbMax) const;
+	virtual void processAllTriangles(btTriangleCallback* callback, const btVector3& aabbMin, const btVector3& aabbMax) const;
 
-	virtual void	processAllTrianglesRay (btTriangleCallback* callback,const btVector3& rayFrom,const btVector3& rayTo) const;
+	virtual void processAllTrianglesRay(btTriangleCallback* callback, const btVector3& rayFrom, const btVector3& rayTo) const;
 
-	virtual	int	calculateSerializeBufferSize() const;
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
-
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 };
 
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
-struct	btGImpactMeshShapeData
+struct btGImpactMeshShapeData
 {
-	btCollisionShapeData	m_collisionShapeData;
+	btCollisionShapeData m_collisionShapeData;
 
 	btStridingMeshInterfaceData m_meshInterface;
 
-	btVector3FloatData	m_localScaling;
+	btVector3FloatData m_localScaling;
 
-	float	m_collisionMargin;
+	float m_collisionMargin;
 
-	int		m_gimpactSubType;
+	int m_gimpactSubType;
 };
 
-SIMD_FORCE_INLINE	int	btGImpactMeshShape::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btGImpactMeshShape::calculateSerializeBufferSize() const
 {
 	return sizeof(btGImpactMeshShapeData);
 }
 
-
-#endif //GIMPACT_MESH_SHAPE_H
+#endif  //GIMPACT_MESH_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btGenericPoolAllocator.cpp b/thirdparty/bullet/BulletCollision/Gimpact/btGenericPoolAllocator.cpp
index 5d07d1adb97..bfdb3db5d08 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btGenericPoolAllocator.cpp
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btGenericPoolAllocator.cpp
@@ -20,48 +20,45 @@ subject to the following restrictions:
 
 #include "btGenericPoolAllocator.h"
 
-
-
 /// *************** btGenericMemoryPool ******************///////////
 
 size_t btGenericMemoryPool::allocate_from_free_nodes(size_t num_elements)
 {
 	size_t ptr = BT_UINT_MAX;
 
-	if(m_free_nodes_count == 0) return BT_UINT_MAX;
+	if (m_free_nodes_count == 0) return BT_UINT_MAX;
 	// find an avaliable free node with the correct size
 	size_t revindex = m_free_nodes_count;
 
-	while(revindex-- && ptr == BT_UINT_MAX)
+	while (revindex-- && ptr == BT_UINT_MAX)
 	{
-		if(m_allocated_sizes[m_free_nodes[revindex]]>=num_elements)
+		if (m_allocated_sizes[m_free_nodes[revindex]] >= num_elements)
 		{
 			ptr = revindex;
 		}
 	}
-	if(ptr == BT_UINT_MAX) return BT_UINT_MAX; // not found
-
+	if (ptr == BT_UINT_MAX) return BT_UINT_MAX;  // not found
 
 	revindex = ptr;
 	ptr = m_free_nodes[revindex];
 	// post: ptr contains the node index, and revindex the index in m_free_nodes
 
-	size_t  finalsize = m_allocated_sizes[ptr];
+	size_t finalsize = m_allocated_sizes[ptr];
 	finalsize -= num_elements;
 
 	m_allocated_sizes[ptr] = num_elements;
 
 	// post: finalsize>=0, m_allocated_sizes[ptr] has the requested size
 
-	if(finalsize>0) // preserve free node, there are some free memory
+	if (finalsize > 0)  // preserve free node, there are some free memory
 	{
 		m_free_nodes[revindex] = ptr + num_elements;
 		m_allocated_sizes[ptr + num_elements] = finalsize;
 	}
-	else // delete free node
+	else  // delete free node
 	{
 		// swap with end
-		m_free_nodes[revindex] = m_free_nodes[m_free_nodes_count-1];
+		m_free_nodes[revindex] = m_free_nodes[m_free_nodes_count - 1];
 		m_free_nodes_count--;
 	}
 
@@ -70,17 +67,16 @@ size_t btGenericMemoryPool::allocate_from_free_nodes(size_t num_elements)
 
 size_t btGenericMemoryPool::allocate_from_pool(size_t num_elements)
 {
-	if(m_allocated_count+num_elements>m_max_element_count) return BT_UINT_MAX;
+	if (m_allocated_count + num_elements > m_max_element_count) return BT_UINT_MAX;
 
 	size_t ptr = m_allocated_count;
 
 	m_allocated_sizes[m_allocated_count] = num_elements;
-	m_allocated_count+=num_elements;
+	m_allocated_count += num_elements;
 
 	return ptr;
 }
 
-
 void btGenericMemoryPool::init_pool(size_t element_size, size_t element_count)
 {
 	m_allocated_count = 0;
@@ -89,14 +85,11 @@ void btGenericMemoryPool::init_pool(size_t element_size, size_t element_count)
 	m_element_size = element_size;
 	m_max_element_count = element_count;
 
+	m_pool = (unsigned char *)btAlignedAlloc(m_element_size * m_max_element_count, 16);
+	m_free_nodes = (size_t *)btAlignedAlloc(sizeof(size_t) * m_max_element_count, 16);
+	m_allocated_sizes = (size_t *)btAlignedAlloc(sizeof(size_t) * m_max_element_count, 16);
 
-
-
-	m_pool = (unsigned char *) btAlignedAlloc(m_element_size*m_max_element_count,16);
-	m_free_nodes = (size_t *) btAlignedAlloc(sizeof(size_t)*m_max_element_count,16);
-	m_allocated_sizes = (size_t *) btAlignedAlloc(sizeof(size_t)*m_max_element_count,16);
-
-	for (size_t i = 0;i< m_max_element_count;i++ )
+	for (size_t i = 0; i < m_max_element_count; i++)
 	{
 		m_allocated_sizes[i] = 0;
 	}
@@ -111,150 +104,141 @@ void btGenericMemoryPool::end_pool()
 	m_free_nodes_count = 0;
 }
 
-
 //! Allocates memory in pool
 /*!
 \param size_bytes size in bytes of the buffer
 */
-void * btGenericMemoryPool::allocate(size_t size_bytes)
+void *btGenericMemoryPool::allocate(size_t size_bytes)
 {
-
-	size_t module = size_bytes%m_element_size;
-	size_t element_count = size_bytes/m_element_size;
-	if(module>0) element_count++;
+	size_t module = size_bytes % m_element_size;
+	size_t element_count = size_bytes / m_element_size;
+	if (module > 0) element_count++;
 
 	size_t alloc_pos = allocate_from_free_nodes(element_count);
 	// a free node is found
-	if(alloc_pos != BT_UINT_MAX)
+	if (alloc_pos != BT_UINT_MAX)
 	{
 		return get_element_data(alloc_pos);
 	}
 	// allocate directly on pool
 	alloc_pos = allocate_from_pool(element_count);
 
-	if(alloc_pos == BT_UINT_MAX) return NULL; // not space
+	if (alloc_pos == BT_UINT_MAX) return NULL;  // not space
 	return get_element_data(alloc_pos);
 }
 
-bool btGenericMemoryPool::freeMemory(void * pointer)
+bool btGenericMemoryPool::freeMemory(void *pointer)
 {
-	unsigned char * pointer_pos = (unsigned char *)pointer;
-	unsigned char * pool_pos = (unsigned char *)m_pool;
+	unsigned char *pointer_pos = (unsigned char *)pointer;
+	unsigned char *pool_pos = (unsigned char *)m_pool;
 	// calc offset
-	if(pointer_pos<pool_pos) return false;//other pool
+	if (pointer_pos < pool_pos) return false;  //other pool
 	size_t offset = size_t(pointer_pos - pool_pos);
-	if(offset>=get_pool_capacity()) return false;// far away
+	if (offset >= get_pool_capacity()) return false;  // far away
 
 	// find free position
-	m_free_nodes[m_free_nodes_count] = offset/m_element_size;
+	m_free_nodes[m_free_nodes_count] = offset / m_element_size;
 	m_free_nodes_count++;
 	return true;
 }
 
-
 /// *******************! btGenericPoolAllocator *******************!///
 
-
 btGenericPoolAllocator::~btGenericPoolAllocator()
 {
 	// destroy pools
 	size_t i;
-	for (i=0;i<m_pool_count;i++)
+	for (i = 0; i < m_pool_count; i++)
 	{
 		m_pools[i]->end_pool();
 		btAlignedFree(m_pools[i]);
 	}
 }
 
-
 // creates a pool
-btGenericMemoryPool * btGenericPoolAllocator::push_new_pool()
+btGenericMemoryPool *btGenericPoolAllocator::push_new_pool()
 {
-	if(m_pool_count >= BT_DEFAULT_MAX_POOLS) return NULL;
+	if (m_pool_count >= BT_DEFAULT_MAX_POOLS) return NULL;
 
-	btGenericMemoryPool * newptr = (btGenericMemoryPool *)btAlignedAlloc(sizeof(btGenericMemoryPool),16);
+	btGenericMemoryPool *newptr = (btGenericMemoryPool *)btAlignedAlloc(sizeof(btGenericMemoryPool), 16);
 
 	m_pools[m_pool_count] = newptr;
 
-	m_pools[m_pool_count]->init_pool(m_pool_element_size,m_pool_element_count);
+	m_pools[m_pool_count]->init_pool(m_pool_element_size, m_pool_element_count);
 
 	m_pool_count++;
 	return newptr;
 }
 
-void * btGenericPoolAllocator::failback_alloc(size_t size_bytes)
+void *btGenericPoolAllocator::failback_alloc(size_t size_bytes)
 {
+	btGenericMemoryPool *pool = NULL;
 
-	btGenericMemoryPool * pool = NULL;
-
-
-	if(size_bytes<=get_pool_capacity())
+	if (size_bytes <= get_pool_capacity())
 	{
-		pool = 	push_new_pool();
+		pool = push_new_pool();
 	}
 
-	if(pool==NULL) // failback
+	if (pool == NULL)  // failback
 	{
-		return btAlignedAlloc(size_bytes,16);
+		return btAlignedAlloc(size_bytes, 16);
 	}
 
 	return pool->allocate(size_bytes);
 }
 
-bool btGenericPoolAllocator::failback_free(void * pointer)
+bool btGenericPoolAllocator::failback_free(void *pointer)
 {
 	btAlignedFree(pointer);
 	return true;
 }
 
-
 //! Allocates memory in pool
 /*!
 \param size_bytes size in bytes of the buffer
 */
-void * btGenericPoolAllocator::allocate(size_t size_bytes)
+void *btGenericPoolAllocator::allocate(size_t size_bytes)
 {
-	void * ptr = NULL;
+	void *ptr = NULL;
 
 	size_t i = 0;
-	while(i<m_pool_count && ptr == NULL)
+	while (i < m_pool_count && ptr == NULL)
 	{
 		ptr = m_pools[i]->allocate(size_bytes);
 		++i;
 	}
 
-	if(ptr) return ptr;
+	if (ptr) return ptr;
 
 	return failback_alloc(size_bytes);
 }
 
-bool btGenericPoolAllocator::freeMemory(void * pointer)
+bool btGenericPoolAllocator::freeMemory(void *pointer)
 {
 	bool result = false;
 
 	size_t i = 0;
-	while(i<m_pool_count && result == false)
+	while (i < m_pool_count && result == false)
 	{
 		result = m_pools[i]->freeMemory(pointer);
 		++i;
 	}
 
-	if(result) return true;
+	if (result) return true;
 
 	return failback_free(pointer);
 }
 
 /// ************** STANDARD ALLOCATOR ***************************///
 
-
 #define BT_DEFAULT_POOL_SIZE 32768
 #define BT_DEFAULT_POOL_ELEMENT_SIZE 8
 
 // main allocator
-class GIM_STANDARD_ALLOCATOR: public btGenericPoolAllocator
+class GIM_STANDARD_ALLOCATOR : public btGenericPoolAllocator
 {
 public:
-	GIM_STANDARD_ALLOCATOR():btGenericPoolAllocator(BT_DEFAULT_POOL_ELEMENT_SIZE,BT_DEFAULT_POOL_SIZE)
+	GIM_STANDARD_ALLOCATOR() : btGenericPoolAllocator(BT_DEFAULT_POOL_ELEMENT_SIZE, BT_DEFAULT_POOL_SIZE)
 	{
 	}
 };
@@ -262,19 +246,18 @@ public:
 // global allocator
 GIM_STANDARD_ALLOCATOR g_main_allocator;
 
-
-void * btPoolAlloc(size_t size)
+void *btPoolAlloc(size_t size)
 {
 	return g_main_allocator.allocate(size);
 }
 
-void * btPoolRealloc(void *ptr, size_t oldsize, size_t newsize)
+void *btPoolRealloc(void *ptr, size_t oldsize, size_t newsize)
 {
-	void * newptr = btPoolAlloc(newsize);
-    size_t copysize = oldsize<newsize?oldsize:newsize;
-    memcpy(newptr,ptr,copysize);
-    btPoolFree(ptr);
-    return newptr;
+	void *newptr = btPoolAlloc(newsize);
+	size_t copysize = oldsize < newsize ? oldsize : newsize;
+	memcpy(newptr, ptr, copysize);
+	btPoolFree(ptr);
+	return newptr;
 }
 
 void btPoolFree(void *ptr)
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btGenericPoolAllocator.h b/thirdparty/bullet/BulletCollision/Gimpact/btGenericPoolAllocator.h
index b46d8516347..a535088e484 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btGenericPoolAllocator.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btGenericPoolAllocator.h
@@ -29,16 +29,16 @@ subject to the following restrictions:
 #define BT_UINT_MAX UINT_MAX
 #define BT_DEFAULT_MAX_POOLS 16
 
-
 //! Generic Pool class
 class btGenericMemoryPool
 {
 public:
-	unsigned char * m_pool; //[m_element_size*m_max_element_count];
-	size_t * m_free_nodes; //[m_max_element_count];//! free nodes
-	size_t * m_allocated_sizes;//[m_max_element_count];//! Number of elements allocated per node
+	unsigned char *m_pool;      //[m_element_size*m_max_element_count];
+	size_t *m_free_nodes;       //[m_max_element_count];//! free nodes
+	size_t *m_allocated_sizes;  //[m_max_element_count];//! Number of elements allocated per node
 	size_t m_allocated_count;
 	size_t m_free_nodes_count;
+
 protected:
 	size_t m_element_size;
 	size_t m_max_element_count;
@@ -47,12 +47,10 @@ protected:
 	size_t allocate_from_pool(size_t num_elements);
 
 public:
-
 	void init_pool(size_t element_size, size_t element_count);
 
 	void end_pool();
 
-
 	btGenericMemoryPool(size_t element_size, size_t element_count)
 	{
 		init_pool(element_size, element_count);
@@ -63,10 +61,9 @@ public:
 		end_pool();
 	}
 
-
 	inline size_t get_pool_capacity()
 	{
-		return m_element_size*m_max_element_count;
+		return m_element_size * m_max_element_count;
 	}
 
 	inline size_t gem_element_size()
@@ -89,23 +86,20 @@ public:
 		return m_free_nodes_count;
 	}
 
-	inline void * get_element_data(size_t element_index)
+	inline void *get_element_data(size_t element_index)
 	{
-		return &m_pool[element_index*m_element_size];
+		return &m_pool[element_index * m_element_size];
 	}
 
 	//! Allocates memory in pool
 	/*!
 	\param size_bytes size in bytes of the buffer
 	*/
-	void * allocate(size_t size_bytes);
+	void *allocate(size_t size_bytes);
 
-	bool freeMemory(void * pointer);
+	bool freeMemory(void *pointer);
 };
 
-
-
-
 //! Generic Allocator with pools
 /*!
 General purpose Allocator which can create Memory Pools dynamiacally as needed.
@@ -115,26 +109,25 @@ class btGenericPoolAllocator
 protected:
 	size_t m_pool_element_size;
 	size_t m_pool_element_count;
-public:
-	btGenericMemoryPool * m_pools[BT_DEFAULT_MAX_POOLS];
-	size_t m_pool_count;
 
+public:
+	btGenericMemoryPool *m_pools[BT_DEFAULT_MAX_POOLS];
+	size_t m_pool_count;
 
 	inline size_t get_pool_capacity()
 	{
-		return m_pool_element_size*m_pool_element_count;
+		return m_pool_element_size * m_pool_element_count;
 	}
 
-
 protected:
 	// creates a pool
-	btGenericMemoryPool * push_new_pool();
+	btGenericMemoryPool *push_new_pool();
 
-	void * failback_alloc(size_t size_bytes);
+	void *failback_alloc(size_t size_bytes);
+
+	bool failback_free(void *pointer);
 
-	bool failback_free(void * pointer);
 public:
-
 	btGenericPoolAllocator(size_t pool_element_size, size_t pool_element_count)
 	{
 		m_pool_count = 0;
@@ -148,16 +141,13 @@ public:
 	/*!
 	\param size_bytes size in bytes of the buffer
 	*/
-	void * allocate(size_t size_bytes);
+	void *allocate(size_t size_bytes);
 
-	bool freeMemory(void * pointer);
+	bool freeMemory(void *pointer);
 };
 
-
-
-void * btPoolAlloc(size_t size);
-void * btPoolRealloc(void *ptr, size_t oldsize, size_t newsize);
+void *btPoolAlloc(size_t size);
+void *btPoolRealloc(void *ptr, size_t oldsize, size_t newsize);
 void btPoolFree(void *ptr);
 
-
 #endif
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btGeometryOperations.h b/thirdparty/bullet/BulletCollision/Gimpact/btGeometryOperations.h
index 60f06510ad7..6a1ee6dcf91 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btGeometryOperations.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btGeometryOperations.h
@@ -27,52 +27,44 @@ subject to the following restrictions:
 
 #include "btBoxCollision.h"
 
-
-
-
-
 #define PLANEDIREPSILON 0.0000001f
 #define PARALELENORMALS 0.000001f
 
-
-#define BT_CLAMP(number,minval,maxval) (number<minval?minval:(number>maxval?maxval:number))
+#define BT_CLAMP(number, minval, maxval) (number < minval ? minval : (number > maxval ? maxval : number))
 
 /// Calc a plane from a triangle edge an a normal. plane is a vec4f
-SIMD_FORCE_INLINE void bt_edge_plane(const btVector3 & e1,const btVector3 &  e2, const btVector3 & normal,btVector4 & plane)
+SIMD_FORCE_INLINE void bt_edge_plane(const btVector3 &e1, const btVector3 &e2, const btVector3 &normal, btVector4 &plane)
 {
-	btVector3 planenormal = (e2-e1).cross(normal);
+	btVector3 planenormal = (e2 - e1).cross(normal);
 	planenormal.normalize();
-	plane.setValue(planenormal[0],planenormal[1],planenormal[2],e2.dot(planenormal));
+	plane.setValue(planenormal[0], planenormal[1], planenormal[2], e2.dot(planenormal));
 }
 
-
-
 //***************** SEGMENT and LINE FUNCTIONS **********************************///
 
 /*! Finds the closest point(cp) to (v) on a segment (e1,e2)
  */
 SIMD_FORCE_INLINE void bt_closest_point_on_segment(
-	btVector3 & cp, const btVector3 & v,
-	const btVector3  &e1,const btVector3 &e2)
+	btVector3 &cp, const btVector3 &v,
+	const btVector3 &e1, const btVector3 &e2)
 {
-    btVector3 n = e2-e1;
-    cp = v - e1;
-	btScalar _scalar = cp.dot(n)/n.dot(n);
-	if(_scalar <0.0f)
+	btVector3 n = e2 - e1;
+	cp = v - e1;
+	btScalar _scalar = cp.dot(n) / n.dot(n);
+	if (_scalar < 0.0f)
 	{
-	    cp = e1;
+		cp = e1;
 	}
-	else if(_scalar >1.0f)
+	else if (_scalar > 1.0f)
 	{
-	    cp = e2;
+		cp = e2;
 	}
 	else
 	{
-		cp = _scalar*n + e1;
+		cp = _scalar * n + e1;
 	}
 }
 
-
 //! line plane collision
 /*!
 *\return
@@ -82,131 +74,125 @@ SIMD_FORCE_INLINE void bt_closest_point_on_segment(
 */
 
 SIMD_FORCE_INLINE int bt_line_plane_collision(
-	const btVector4 & plane,
-	const btVector3 & vDir,
-	const btVector3 & vPoint,
-	btVector3 & pout,
+	const btVector4 &plane,
+	const btVector3 &vDir,
+	const btVector3 &vPoint,
+	btVector3 &pout,
 	btScalar &tparam,
 	btScalar tmin, btScalar tmax)
 {
-
 	btScalar _dotdir = vDir.dot(plane);
 
-	if(btFabs(_dotdir)<PLANEDIREPSILON)
+	if (btFabs(_dotdir) < PLANEDIREPSILON)
 	{
 		tparam = tmax;
-	    return 0;
+		return 0;
 	}
 
-	btScalar _dis = bt_distance_point_plane(plane,vPoint);
-	char returnvalue = _dis<0.0f? 2:1;
-	tparam = -_dis/_dotdir;
+	btScalar _dis = bt_distance_point_plane(plane, vPoint);
+	char returnvalue = _dis < 0.0f ? 2 : 1;
+	tparam = -_dis / _dotdir;
 
-	if(tparam<tmin)
+	if (tparam < tmin)
 	{
 		returnvalue = 0;
 		tparam = tmin;
 	}
-	else if(tparam>tmax)
+	else if (tparam > tmax)
 	{
 		returnvalue = 0;
 		tparam = tmax;
 	}
-	pout = tparam*vDir + vPoint;
+	pout = tparam * vDir + vPoint;
 	return returnvalue;
 }
 
-
 //! Find closest points on segments
 SIMD_FORCE_INLINE void bt_segment_collision(
-	const btVector3 & vA1,
-	const btVector3 & vA2,
-	const btVector3 & vB1,
-	const btVector3 & vB2,
-	btVector3 & vPointA,
-	btVector3 & vPointB)
+	const btVector3 &vA1,
+	const btVector3 &vA2,
+	const btVector3 &vB1,
+	const btVector3 &vB2,
+	btVector3 &vPointA,
+	btVector3 &vPointB)
 {
-    btVector3 AD = vA2 - vA1;
-    btVector3 BD = vB2 - vB1;
-    btVector3 N = AD.cross(BD);
-    btScalar tp = N.length2();
+	btVector3 AD = vA2 - vA1;
+	btVector3 BD = vB2 - vB1;
+	btVector3 N = AD.cross(BD);
+	btScalar tp = N.length2();
 
-    btVector4 _M;//plane
+	btVector4 _M;  //plane
 
-    if(tp<SIMD_EPSILON)//ARE PARALELE
-    {
-    	//project B over A
-    	bool invert_b_order = false;
-    	_M[0] = vB1.dot(AD);
-    	_M[1] = vB2.dot(AD);
+	if (tp < SIMD_EPSILON)  //ARE PARALELE
+	{
+		//project B over A
+		bool invert_b_order = false;
+		_M[0] = vB1.dot(AD);
+		_M[1] = vB2.dot(AD);
 
-    	if(_M[0]>_M[1])
-    	{
-    		invert_b_order  = true;
-    		BT_SWAP_NUMBERS(_M[0],_M[1]);
-    	}
-    	_M[2] = vA1.dot(AD);
-    	_M[3] = vA2.dot(AD);
-    	//mid points
-    	N[0] = (_M[0]+_M[1])*0.5f;
-    	N[1] = (_M[2]+_M[3])*0.5f;
+		if (_M[0] > _M[1])
+		{
+			invert_b_order = true;
+			BT_SWAP_NUMBERS(_M[0], _M[1]);
+		}
+		_M[2] = vA1.dot(AD);
+		_M[3] = vA2.dot(AD);
+		//mid points
+		N[0] = (_M[0] + _M[1]) * 0.5f;
+		N[1] = (_M[2] + _M[3]) * 0.5f;
 
-    	if(N[0]<N[1])
-    	{
-    		if(_M[1]<_M[2])
-    		{
-    			vPointB = invert_b_order?vB1:vB2;
-    			vPointA = vA1;
-    		}
-    		else if(_M[1]<_M[3])
-    		{
-    			vPointB = invert_b_order?vB1:vB2;
-    			bt_closest_point_on_segment(vPointA,vPointB,vA1,vA2);
-    		}
-    		else
-    		{
-    			vPointA = vA2;
-    			bt_closest_point_on_segment(vPointB,vPointA,vB1,vB2);
-    		}
-    	}
-    	else
-    	{
-    		if(_M[3]<_M[0])
-    		{
-    			vPointB = invert_b_order?vB2:vB1;
-    			vPointA = vA2;
-    		}
-    		else if(_M[3]<_M[1])
-    		{
-    			vPointA = vA2;
-    			bt_closest_point_on_segment(vPointB,vPointA,vB1,vB2);
-    		}
-    		else
-    		{
-    			vPointB = invert_b_order?vB1:vB2;
-    			bt_closest_point_on_segment(vPointA,vPointB,vA1,vA2);
-    		}
-    	}
-    	return;
-    }
+		if (N[0] < N[1])
+		{
+			if (_M[1] < _M[2])
+			{
+				vPointB = invert_b_order ? vB1 : vB2;
+				vPointA = vA1;
+			}
+			else if (_M[1] < _M[3])
+			{
+				vPointB = invert_b_order ? vB1 : vB2;
+				bt_closest_point_on_segment(vPointA, vPointB, vA1, vA2);
+			}
+			else
+			{
+				vPointA = vA2;
+				bt_closest_point_on_segment(vPointB, vPointA, vB1, vB2);
+			}
+		}
+		else
+		{
+			if (_M[3] < _M[0])
+			{
+				vPointB = invert_b_order ? vB2 : vB1;
+				vPointA = vA2;
+			}
+			else if (_M[3] < _M[1])
+			{
+				vPointA = vA2;
+				bt_closest_point_on_segment(vPointB, vPointA, vB1, vB2);
+			}
+			else
+			{
+				vPointB = invert_b_order ? vB1 : vB2;
+				bt_closest_point_on_segment(vPointA, vPointB, vA1, vA2);
+			}
+		}
+		return;
+	}
 
-    N = N.cross(BD);
-    _M.setValue(N[0],N[1],N[2],vB1.dot(N));
+	N = N.cross(BD);
+	_M.setValue(N[0], N[1], N[2], vB1.dot(N));
 
 	// get point A as the plane collision point
-    bt_line_plane_collision(_M,AD,vA1,vPointA,tp,btScalar(0), btScalar(1));
+	bt_line_plane_collision(_M, AD, vA1, vPointA, tp, btScalar(0), btScalar(1));
 
-    /*Closest point on segment*/
-    vPointB = vPointA - vB1;
+	/*Closest point on segment*/
+	vPointB = vPointA - vB1;
 	tp = vPointB.dot(BD);
-	tp/= BD.dot(BD);
-	tp = BT_CLAMP(tp,0.0f,1.0f);
+	tp /= BD.dot(BD);
+	tp = BT_CLAMP(tp, 0.0f, 1.0f);
 
-	vPointB = tp*BD + vB1;
+	vPointB = tp * BD + vB1;
 }
 
-
-
-
-
-#endif // GIM_VECTOR_H_INCLUDED
+#endif  // GIM_VECTOR_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btQuantization.h b/thirdparty/bullet/BulletCollision/Gimpact/btQuantization.h
index bd2633cfc59..19a02a21772 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btQuantization.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btQuantization.h
@@ -27,54 +27,47 @@ subject to the following restrictions:
 
 #include "LinearMath/btTransform.h"
 
-
-
-
-
-
 SIMD_FORCE_INLINE void bt_calc_quantization_parameters(
-	btVector3 & outMinBound,
-	btVector3 & outMaxBound,
-	btVector3 & bvhQuantization,
-	const btVector3& srcMinBound,const btVector3& srcMaxBound,
+	btVector3& outMinBound,
+	btVector3& outMaxBound,
+	btVector3& bvhQuantization,
+	const btVector3& srcMinBound, const btVector3& srcMaxBound,
 	btScalar quantizationMargin)
 {
 	//enlarge the AABB to avoid division by zero when initializing the quantization values
-	btVector3 clampValue(quantizationMargin,quantizationMargin,quantizationMargin);
+	btVector3 clampValue(quantizationMargin, quantizationMargin, quantizationMargin);
 	outMinBound = srcMinBound - clampValue;
 	outMaxBound = srcMaxBound + clampValue;
 	btVector3 aabbSize = outMaxBound - outMinBound;
 	bvhQuantization = btVector3(btScalar(65535.0),
 								btScalar(65535.0),
-								btScalar(65535.0)) / aabbSize;
+								btScalar(65535.0)) /
+					  aabbSize;
 }
 
-
 SIMD_FORCE_INLINE void bt_quantize_clamp(
 	unsigned short* out,
 	const btVector3& point,
-	const btVector3 & min_bound,
-	const btVector3 & max_bound,
-	const btVector3 & bvhQuantization)
+	const btVector3& min_bound,
+	const btVector3& max_bound,
+	const btVector3& bvhQuantization)
 {
-
 	btVector3 clampedPoint(point);
 	clampedPoint.setMax(min_bound);
 	clampedPoint.setMin(max_bound);
 
 	btVector3 v = (clampedPoint - min_bound) * bvhQuantization;
-	out[0] = (unsigned short)(v.getX()+0.5f);
-	out[1] = (unsigned short)(v.getY()+0.5f);
-	out[2] = (unsigned short)(v.getZ()+0.5f);
+	out[0] = (unsigned short)(v.getX() + 0.5f);
+	out[1] = (unsigned short)(v.getY() + 0.5f);
+	out[2] = (unsigned short)(v.getZ() + 0.5f);
 }
 
-
 SIMD_FORCE_INLINE btVector3 bt_unquantize(
 	const unsigned short* vecIn,
-	const btVector3 & offset,
-	const btVector3 & bvhQuantization)
+	const btVector3& offset,
+	const btVector3& bvhQuantization)
 {
-	btVector3	vecOut;
+	btVector3 vecOut;
 	vecOut.setValue(
 		(btScalar)(vecIn[0]) / (bvhQuantization.getX()),
 		(btScalar)(vecIn[1]) / (bvhQuantization.getY()),
@@ -83,6 +76,4 @@ SIMD_FORCE_INLINE btVector3 bt_unquantize(
 	return vecOut;
 }
 
-
-
-#endif // BT_GIMPACT_QUANTIZATION_H_INCLUDED
+#endif  // BT_GIMPACT_QUANTIZATION_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btTriangleShapeEx.cpp b/thirdparty/bullet/BulletCollision/Gimpact/btTriangleShapeEx.cpp
index ca76cc54a12..292ef8c1ff3 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btTriangleShapeEx.cpp
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btTriangleShapeEx.cpp
@@ -23,196 +23,181 @@ subject to the following restrictions:
 
 #include "btTriangleShapeEx.h"
 
-
-
-void GIM_TRIANGLE_CONTACT::merge_points(const btVector4 & plane,
-                                       btScalar margin, const btVector3 * points, int point_count)
+void GIM_TRIANGLE_CONTACT::merge_points(const btVector4& plane,
+										btScalar margin, const btVector3* points, int point_count)
 {
-    m_point_count = 0;
-    m_penetration_depth= -1000.0f;
+	m_point_count = 0;
+	m_penetration_depth = -1000.0f;
 
-    int point_indices[MAX_TRI_CLIPPING];
+	int point_indices[MAX_TRI_CLIPPING];
 
 	int _k;
 
-    for ( _k=0;_k<point_count;_k++)
-    {
-        btScalar _dist = - bt_distance_point_plane(plane,points[_k]) + margin;
+	for (_k = 0; _k < point_count; _k++)
+	{
+		btScalar _dist = -bt_distance_point_plane(plane, points[_k]) + margin;
 
-        if (_dist>=0.0f)
-        {
-            if (_dist>m_penetration_depth)
-            {
-                m_penetration_depth = _dist;
-                point_indices[0] = _k;
-                m_point_count=1;
-            }
-            else if ((_dist+SIMD_EPSILON)>=m_penetration_depth)
-            {
-                point_indices[m_point_count] = _k;
-                m_point_count++;
-            }
-        }
-    }
+		if (_dist >= 0.0f)
+		{
+			if (_dist > m_penetration_depth)
+			{
+				m_penetration_depth = _dist;
+				point_indices[0] = _k;
+				m_point_count = 1;
+			}
+			else if ((_dist + SIMD_EPSILON) >= m_penetration_depth)
+			{
+				point_indices[m_point_count] = _k;
+				m_point_count++;
+			}
+		}
+	}
 
-    for ( _k=0;_k<m_point_count;_k++)
-    {
-        m_points[_k] = points[point_indices[_k]];
-    }
+	for (_k = 0; _k < m_point_count; _k++)
+	{
+		m_points[_k] = points[point_indices[_k]];
+	}
 }
 
 ///class btPrimitiveTriangle
 bool btPrimitiveTriangle::overlap_test_conservative(const btPrimitiveTriangle& other)
 {
-    btScalar total_margin = m_margin + other.m_margin;
-    // classify points on other triangle
-    btScalar dis0 = bt_distance_point_plane(m_plane,other.m_vertices[0]) - total_margin;
+	btScalar total_margin = m_margin + other.m_margin;
+	// classify points on other triangle
+	btScalar dis0 = bt_distance_point_plane(m_plane, other.m_vertices[0]) - total_margin;
 
-    btScalar dis1 = bt_distance_point_plane(m_plane,other.m_vertices[1]) - total_margin;
+	btScalar dis1 = bt_distance_point_plane(m_plane, other.m_vertices[1]) - total_margin;
 
-    btScalar dis2 = bt_distance_point_plane(m_plane,other.m_vertices[2]) - total_margin;
+	btScalar dis2 = bt_distance_point_plane(m_plane, other.m_vertices[2]) - total_margin;
 
-    if (dis0>0.0f&&dis1>0.0f&&dis2>0.0f) return false;
+	if (dis0 > 0.0f && dis1 > 0.0f && dis2 > 0.0f) return false;
 
-    // classify points on this triangle
-    dis0 = bt_distance_point_plane(other.m_plane,m_vertices[0]) - total_margin;
+	// classify points on this triangle
+	dis0 = bt_distance_point_plane(other.m_plane, m_vertices[0]) - total_margin;
 
-    dis1 = bt_distance_point_plane(other.m_plane,m_vertices[1]) - total_margin;
+	dis1 = bt_distance_point_plane(other.m_plane, m_vertices[1]) - total_margin;
 
-    dis2 = bt_distance_point_plane(other.m_plane,m_vertices[2]) - total_margin;
+	dis2 = bt_distance_point_plane(other.m_plane, m_vertices[2]) - total_margin;
 
-    if (dis0>0.0f&&dis1>0.0f&&dis2>0.0f) return false;
+	if (dis0 > 0.0f && dis1 > 0.0f && dis2 > 0.0f) return false;
 
-    return true;
+	return true;
 }
 
-int btPrimitiveTriangle::clip_triangle(btPrimitiveTriangle & other, btVector3 * clipped_points )
+int btPrimitiveTriangle::clip_triangle(btPrimitiveTriangle& other, btVector3* clipped_points)
 {
-    // edge 0
+	// edge 0
 
-    btVector3 temp_points[MAX_TRI_CLIPPING];
+	btVector3 temp_points[MAX_TRI_CLIPPING];
 
+	btVector4 edgeplane;
 
-    btVector4 edgeplane;
+	get_edge_plane(0, edgeplane);
 
-    get_edge_plane(0,edgeplane);
+	int clipped_count = bt_plane_clip_triangle(
+		edgeplane, other.m_vertices[0], other.m_vertices[1], other.m_vertices[2], temp_points);
 
+	if (clipped_count == 0) return 0;
 
-    int clipped_count = bt_plane_clip_triangle(
-                            edgeplane,other.m_vertices[0],other.m_vertices[1],other.m_vertices[2],temp_points);
+	btVector3 temp_points1[MAX_TRI_CLIPPING];
 
-    if (clipped_count == 0) return 0;
+	// edge 1
+	get_edge_plane(1, edgeplane);
 
-    btVector3 temp_points1[MAX_TRI_CLIPPING];
+	clipped_count = bt_plane_clip_polygon(edgeplane, temp_points, clipped_count, temp_points1);
 
+	if (clipped_count == 0) return 0;
 
-    // edge 1
-    get_edge_plane(1,edgeplane);
+	// edge 2
+	get_edge_plane(2, edgeplane);
 
+	clipped_count = bt_plane_clip_polygon(
+		edgeplane, temp_points1, clipped_count, clipped_points);
 
-    clipped_count = bt_plane_clip_polygon(edgeplane,temp_points,clipped_count,temp_points1);
-
-    if (clipped_count == 0) return 0;
-
-    // edge 2
-    get_edge_plane(2,edgeplane);
-
-    clipped_count = bt_plane_clip_polygon(
-                        edgeplane,temp_points1,clipped_count,clipped_points);
-
-    return clipped_count;
+	return clipped_count;
 }
 
-bool btPrimitiveTriangle::find_triangle_collision_clip_method(btPrimitiveTriangle & other, GIM_TRIANGLE_CONTACT & contacts)
+bool btPrimitiveTriangle::find_triangle_collision_clip_method(btPrimitiveTriangle& other, GIM_TRIANGLE_CONTACT& contacts)
 {
-    btScalar margin = m_margin + other.m_margin;
+	btScalar margin = m_margin + other.m_margin;
 
-    btVector3 clipped_points[MAX_TRI_CLIPPING];
-    int clipped_count;
-    //create planes
-    // plane v vs U points
+	btVector3 clipped_points[MAX_TRI_CLIPPING];
+	int clipped_count;
+	//create planes
+	// plane v vs U points
 
-    GIM_TRIANGLE_CONTACT contacts1;
+	GIM_TRIANGLE_CONTACT contacts1;
 
-    contacts1.m_separating_normal = m_plane;
+	contacts1.m_separating_normal = m_plane;
 
+	clipped_count = clip_triangle(other, clipped_points);
 
-    clipped_count = clip_triangle(other,clipped_points);
+	if (clipped_count == 0)
+	{
+		return false;  //Reject
+	}
 
-    if (clipped_count == 0 )
-    {
-        return false;//Reject
-    }
+	//find most deep interval face1
+	contacts1.merge_points(contacts1.m_separating_normal, margin, clipped_points, clipped_count);
+	if (contacts1.m_point_count == 0) return false;  // too far
+	//Normal pointing to this triangle
+	contacts1.m_separating_normal *= -1.f;
 
-    //find most deep interval face1
-    contacts1.merge_points(contacts1.m_separating_normal,margin,clipped_points,clipped_count);
-    if (contacts1.m_point_count == 0) return false; // too far
-    //Normal pointing to this triangle
-    contacts1.m_separating_normal *= -1.f;
+	//Clip tri1 by tri2 edges
+	GIM_TRIANGLE_CONTACT contacts2;
+	contacts2.m_separating_normal = other.m_plane;
 
+	clipped_count = other.clip_triangle(*this, clipped_points);
 
-    //Clip tri1 by tri2 edges
-    GIM_TRIANGLE_CONTACT contacts2;
-    contacts2.m_separating_normal = other.m_plane;
+	if (clipped_count == 0)
+	{
+		return false;  //Reject
+	}
 
-    clipped_count = other.clip_triangle(*this,clipped_points);
+	//find most deep interval face1
+	contacts2.merge_points(contacts2.m_separating_normal, margin, clipped_points, clipped_count);
+	if (contacts2.m_point_count == 0) return false;  // too far
 
-    if (clipped_count == 0 )
-    {
-        return false;//Reject
-    }
-
-    //find most deep interval face1
-    contacts2.merge_points(contacts2.m_separating_normal,margin,clipped_points,clipped_count);
-    if (contacts2.m_point_count == 0) return false; // too far
-
-
-
-
-    ////check most dir for contacts
-    if (contacts2.m_penetration_depth<contacts1.m_penetration_depth)
-    {
-        contacts.copy_from(contacts2);
-    }
-    else
-    {
-        contacts.copy_from(contacts1);
-    }
-    return true;
+	////check most dir for contacts
+	if (contacts2.m_penetration_depth < contacts1.m_penetration_depth)
+	{
+		contacts.copy_from(contacts2);
+	}
+	else
+	{
+		contacts.copy_from(contacts1);
+	}
+	return true;
 }
 
-
-
 ///class btTriangleShapeEx: public btTriangleShape
 
 bool btTriangleShapeEx::overlap_test_conservative(const btTriangleShapeEx& other)
 {
-    btScalar total_margin = getMargin() + other.getMargin();
+	btScalar total_margin = getMargin() + other.getMargin();
 
-    btVector4 plane0;
-    buildTriPlane(plane0);
-    btVector4 plane1;
-    other.buildTriPlane(plane1);
+	btVector4 plane0;
+	buildTriPlane(plane0);
+	btVector4 plane1;
+	other.buildTriPlane(plane1);
 
-    // classify points on other triangle
-    btScalar dis0 = bt_distance_point_plane(plane0,other.m_vertices1[0]) - total_margin;
+	// classify points on other triangle
+	btScalar dis0 = bt_distance_point_plane(plane0, other.m_vertices1[0]) - total_margin;
 
-    btScalar dis1 = bt_distance_point_plane(plane0,other.m_vertices1[1]) - total_margin;
+	btScalar dis1 = bt_distance_point_plane(plane0, other.m_vertices1[1]) - total_margin;
 
-    btScalar dis2 = bt_distance_point_plane(plane0,other.m_vertices1[2]) - total_margin;
+	btScalar dis2 = bt_distance_point_plane(plane0, other.m_vertices1[2]) - total_margin;
 
-    if (dis0>0.0f&&dis1>0.0f&&dis2>0.0f) return false;
+	if (dis0 > 0.0f && dis1 > 0.0f && dis2 > 0.0f) return false;
 
-    // classify points on this triangle
-    dis0 = bt_distance_point_plane(plane1,m_vertices1[0]) - total_margin;
+	// classify points on this triangle
+	dis0 = bt_distance_point_plane(plane1, m_vertices1[0]) - total_margin;
 
-    dis1 = bt_distance_point_plane(plane1,m_vertices1[1]) - total_margin;
+	dis1 = bt_distance_point_plane(plane1, m_vertices1[1]) - total_margin;
 
-    dis2 = bt_distance_point_plane(plane1,m_vertices1[2]) - total_margin;
+	dis2 = bt_distance_point_plane(plane1, m_vertices1[2]) - total_margin;
 
-    if (dis0>0.0f&&dis1>0.0f&&dis2>0.0f) return false;
+	if (dis0 > 0.0f && dis1 > 0.0f && dis2 > 0.0f) return false;
 
-    return true;
+	return true;
 }
-
-
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/btTriangleShapeEx.h b/thirdparty/bullet/BulletCollision/Gimpact/btTriangleShapeEx.h
index 973c2ed1277..568a1ce8116 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/btTriangleShapeEx.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/btTriangleShapeEx.h
@@ -21,7 +21,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef GIMPACT_TRIANGLE_SHAPE_EX_H
 #define GIMPACT_TRIANGLE_SHAPE_EX_H
 
@@ -31,16 +30,15 @@ subject to the following restrictions:
 #include "btClipPolygon.h"
 #include "btGeometryOperations.h"
 
-
 #define MAX_TRI_CLIPPING 16
 
 //! Structure for collision
 struct GIM_TRIANGLE_CONTACT
 {
-    btScalar m_penetration_depth;
-    int m_point_count;
-    btVector4 m_separating_normal;
-    btVector3 m_points[MAX_TRI_CLIPPING];
+	btScalar m_penetration_depth;
+	int m_point_count;
+	btVector4 m_separating_normal;
+	btVector3 m_points[MAX_TRI_CLIPPING];
 
 	SIMD_FORCE_INLINE void copy_from(const GIM_TRIANGLE_CONTACT& other)
 	{
@@ -48,7 +46,7 @@ struct GIM_TRIANGLE_CONTACT
 		m_separating_normal = other.m_separating_normal;
 		m_point_count = other.m_point_count;
 		int i = m_point_count;
-		while(i--)
+		while (i--)
 		{
 			m_points[i] = other.m_points[i];
 		}
@@ -63,14 +61,11 @@ struct GIM_TRIANGLE_CONTACT
 		copy_from(other);
 	}
 
-    //! classify points that are closer
-    void merge_points(const btVector4 & plane,
-    				btScalar margin, const btVector3 * points, int point_count);
-
+	//! classify points that are closer
+	void merge_points(const btVector4& plane,
+					  btScalar margin, const btVector3* points, int point_count);
 };
 
-
-
 class btPrimitiveTriangle
 {
 public:
@@ -78,17 +73,15 @@ public:
 	btVector4 m_plane;
 	btScalar m_margin;
 	btScalar m_dummy;
-	btPrimitiveTriangle():m_margin(0.01f)
+	btPrimitiveTriangle() : m_margin(0.01f)
 	{
-
 	}
 
-
 	SIMD_FORCE_INLINE void buildTriPlane()
 	{
-		btVector3 normal = (m_vertices[1]-m_vertices[0]).cross(m_vertices[2]-m_vertices[0]);
+		btVector3 normal = (m_vertices[1] - m_vertices[0]).cross(m_vertices[2] - m_vertices[0]);
 		normal.normalize();
-		m_plane.setValue(normal[0],normal[1],normal[2],m_vertices[0].dot(normal));
+		m_plane.setValue(normal[0], normal[1], normal[2], m_vertices[0].dot(normal));
 	}
 
 	//! Test if triangles could collide
@@ -98,14 +91,14 @@ public:
 	/*!
 	\pre this triangle must have its plane calculated.
 	*/
-	SIMD_FORCE_INLINE void get_edge_plane(int edge_index, btVector4 &plane)  const
-    {
-		const btVector3 & e0 = m_vertices[edge_index];
-		const btVector3 & e1 = m_vertices[(edge_index+1)%3];
-		bt_edge_plane(e0,e1,m_plane,plane);
-    }
+	SIMD_FORCE_INLINE void get_edge_plane(int edge_index, btVector4& plane) const
+	{
+		const btVector3& e0 = m_vertices[edge_index];
+		const btVector3& e1 = m_vertices[(edge_index + 1) % 3];
+		bt_edge_plane(e0, e1, m_plane, plane);
+	}
 
-    void applyTransform(const btTransform& t)
+	void applyTransform(const btTransform& t)
 	{
 		m_vertices[0] = t(m_vertices[0]);
 		m_vertices[1] = t(m_vertices[1]);
@@ -117,44 +110,41 @@ public:
 	\pre clipped_points must have MAX_TRI_CLIPPING size, and this triangle must have its plane calculated.
 	\return the number of clipped points
 	*/
-    int clip_triangle(btPrimitiveTriangle & other, btVector3 * clipped_points );
+	int clip_triangle(btPrimitiveTriangle& other, btVector3* clipped_points);
 
 	//! Find collision using the clipping method
 	/*!
 	\pre this triangle and other must have their triangles calculated
 	*/
-    bool find_triangle_collision_clip_method(btPrimitiveTriangle & other, GIM_TRIANGLE_CONTACT & contacts);
+	bool find_triangle_collision_clip_method(btPrimitiveTriangle& other, GIM_TRIANGLE_CONTACT& contacts);
 };
 
-
-
 //! Helper class for colliding Bullet Triangle Shapes
 /*!
 This class implements a better getAabb method than the previous btTriangleShape class
 */
-class btTriangleShapeEx: public btTriangleShape
+class btTriangleShapeEx : public btTriangleShape
 {
 public:
-
-	btTriangleShapeEx():btTriangleShape(btVector3(0,0,0),btVector3(0,0,0),btVector3(0,0,0))
+	btTriangleShapeEx() : btTriangleShape(btVector3(0, 0, 0), btVector3(0, 0, 0), btVector3(0, 0, 0))
 	{
 	}
 
-	btTriangleShapeEx(const btVector3& p0,const btVector3& p1,const btVector3& p2):	btTriangleShape(p0,p1,p2)
+	btTriangleShapeEx(const btVector3& p0, const btVector3& p1, const btVector3& p2) : btTriangleShape(p0, p1, p2)
 	{
 	}
 
-	btTriangleShapeEx(const btTriangleShapeEx & other):	btTriangleShape(other.m_vertices1[0],other.m_vertices1[1],other.m_vertices1[2])
+	btTriangleShapeEx(const btTriangleShapeEx& other) : btTriangleShape(other.m_vertices1[0], other.m_vertices1[1], other.m_vertices1[2])
 	{
 	}
 
-	virtual void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax)const
+	virtual void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
 	{
 		btVector3 tv0 = t(m_vertices1[0]);
 		btVector3 tv1 = t(m_vertices1[1]);
 		btVector3 tv2 = t(m_vertices1[2]);
 
-		btAABB trianglebox(tv0,tv1,tv2,m_collisionMargin);
+		btAABB trianglebox(tv0, tv1, tv2, m_collisionMargin);
 		aabbMin = trianglebox.m_min;
 		aabbMax = trianglebox.m_max;
 	}
@@ -166,15 +156,14 @@ public:
 		m_vertices1[2] = t(m_vertices1[2]);
 	}
 
-	SIMD_FORCE_INLINE void buildTriPlane(btVector4 & plane) const
+	SIMD_FORCE_INLINE void buildTriPlane(btVector4& plane) const
 	{
-		btVector3 normal = (m_vertices1[1]-m_vertices1[0]).cross(m_vertices1[2]-m_vertices1[0]);
+		btVector3 normal = (m_vertices1[1] - m_vertices1[0]).cross(m_vertices1[2] - m_vertices1[0]);
 		normal.normalize();
-		plane.setValue(normal[0],normal[1],normal[2],m_vertices1[0].dot(normal));
+		plane.setValue(normal[0], normal[1], normal[2], m_vertices1[0].dot(normal));
 	}
 
 	bool overlap_test_conservative(const btTriangleShapeEx& other);
 };
 
-
-#endif //GIMPACT_TRIANGLE_MESH_SHAPE_H
+#endif  //GIMPACT_TRIANGLE_MESH_SHAPE_H
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/gim_array.h b/thirdparty/bullet/BulletCollision/Gimpact/gim_array.h
index cda51a5fcec..fc2dc38a330 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/gim_array.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/gim_array.h
@@ -34,47 +34,46 @@ email: projectileman@yahoo.com
 
 #include "gim_memory.h"
 
-
 #define GIM_ARRAY_GROW_INCREMENT 2
 #define GIM_ARRAY_GROW_FACTOR 2
 
 //!	Very simple array container with fast access and simd memory
-template<typename T>
+template <typename T>
 class gim_array
 {
 public:
-//! properties
-//!@{
-    T *m_data;
-    GUINT m_size;
-    GUINT m_allocated_size;
-//!@}
-//! protected operations
-//!@{
+	//! properties
+	//!@{
+	T* m_data;
+	GUINT m_size;
+	GUINT m_allocated_size;
+	//!@}
+	//! protected operations
+	//!@{
 
-    inline void destroyData()
+	inline void destroyData()
 	{
-	    m_allocated_size = 0;
-		if(m_data==NULL) return;
+		m_allocated_size = 0;
+		if (m_data == NULL) return;
 		gim_free(m_data);
 		m_data = NULL;
 	}
 
 	inline bool resizeData(GUINT newsize)
 	{
-		if(newsize==0)
+		if (newsize == 0)
 		{
 			destroyData();
 			return true;
 		}
 
-		if(m_size>0)
+		if (m_size > 0)
 		{
-            m_data = (T*)gim_realloc(m_data,m_size*sizeof(T),newsize*sizeof(T));
+			m_data = (T*)gim_realloc(m_data, m_size * sizeof(T), newsize * sizeof(T));
 		}
 		else
 		{
-		    m_data = (T*)gim_alloc(newsize*sizeof(T));
+			m_data = (T*)gim_alloc(newsize * sizeof(T));
 		}
 		m_allocated_size = newsize;
 		return true;
@@ -82,243 +81,238 @@ public:
 
 	inline bool growingCheck()
 	{
-		if(m_allocated_size<=m_size)
+		if (m_allocated_size <= m_size)
 		{
-		    GUINT requestsize = m_size;
-		    m_size = m_allocated_size;
-			if(resizeData((requestsize+GIM_ARRAY_GROW_INCREMENT)*GIM_ARRAY_GROW_FACTOR)==false) return false;
+			GUINT requestsize = m_size;
+			m_size = m_allocated_size;
+			if (resizeData((requestsize + GIM_ARRAY_GROW_INCREMENT) * GIM_ARRAY_GROW_FACTOR) == false) return false;
 		}
 		return true;
 	}
 
-//!@}
-//! public operations
-//!@{
-    inline  bool reserve(GUINT size)
-    {
-        if(m_allocated_size>=size) return false;
-        return resizeData(size);
-    }
+	//!@}
+	//! public operations
+	//!@{
+	inline bool reserve(GUINT size)
+	{
+		if (m_allocated_size >= size) return false;
+		return resizeData(size);
+	}
 
-    inline void clear_range(GUINT start_range)
-    {
-        while(m_size>start_range)
-        {
-            m_data[--m_size].~T();
-        }
-    }
+	inline void clear_range(GUINT start_range)
+	{
+		while (m_size > start_range)
+		{
+			m_data[--m_size].~T();
+		}
+	}
 
-    inline void clear()
-    {
-        if(m_size==0)return;
-        clear_range(0);
-    }
+	inline void clear()
+	{
+		if (m_size == 0) return;
+		clear_range(0);
+	}
 
-    inline void clear_memory()
-    {
-        clear();
-        destroyData();
-    }
+	inline void clear_memory()
+	{
+		clear();
+		destroyData();
+	}
 
-    gim_array()
-    {
-        m_data = 0;
-        m_size = 0;
-        m_allocated_size = 0;
-    }
+	gim_array()
+	{
+		m_data = 0;
+		m_size = 0;
+		m_allocated_size = 0;
+	}
 
-    gim_array(GUINT reservesize)
-    {
-        m_data = 0;
-        m_size = 0;
+	gim_array(GUINT reservesize)
+	{
+		m_data = 0;
+		m_size = 0;
 
-        m_allocated_size = 0;
-        reserve(reservesize);
-    }
+		m_allocated_size = 0;
+		reserve(reservesize);
+	}
 
-    ~gim_array()
-    {
-        clear_memory();
-    }
+	~gim_array()
+	{
+		clear_memory();
+	}
 
-    inline GUINT size() const
-    {
-        return m_size;
-    }
+	inline GUINT size() const
+	{
+		return m_size;
+	}
 
-    inline GUINT max_size() const
-    {
-        return m_allocated_size;
-    }
+	inline GUINT max_size() const
+	{
+		return m_allocated_size;
+	}
 
-    inline T & operator[](size_t i)
+	inline T& operator[](size_t i)
 	{
 		return m_data[i];
 	}
-	inline  const T & operator[](size_t i) const
+	inline const T& operator[](size_t i) const
 	{
 		return m_data[i];
 	}
 
-    inline T * pointer(){ return m_data;}
-    inline const T * pointer() const
-    { return m_data;}
+	inline T* pointer() { return m_data; }
+	inline const T* pointer() const
+	{
+		return m_data;
+	}
 
-
-    inline T * get_pointer_at(GUINT i)
+	inline T* get_pointer_at(GUINT i)
 	{
 		return m_data + i;
 	}
 
-	inline const T * get_pointer_at(GUINT i) const
+	inline const T* get_pointer_at(GUINT i) const
 	{
 		return m_data + i;
 	}
 
-	inline T & at(GUINT i)
+	inline T& at(GUINT i)
 	{
 		return m_data[i];
 	}
 
-	inline const T & at(GUINT i) const
+	inline const T& at(GUINT i) const
 	{
 		return m_data[i];
 	}
 
-	inline T & front()
+	inline T& front()
 	{
 		return *m_data;
 	}
 
-	inline const T & front() const
+	inline const T& front() const
 	{
 		return *m_data;
 	}
 
-	inline T & back()
+	inline T& back()
 	{
-		return m_data[m_size-1];
+		return m_data[m_size - 1];
 	}
 
-	inline const T & back() const
+	inline const T& back() const
 	{
-		return m_data[m_size-1];
+		return m_data[m_size - 1];
 	}
 
-
 	inline void swap(GUINT i, GUINT j)
 	{
-	    gim_swap_elements(m_data,i,j);
+		gim_swap_elements(m_data, i, j);
 	}
 
-	inline void push_back(const T & obj)
+	inline void push_back(const T& obj)
 	{
-	    this->growingCheck();
-	    m_data[m_size] = obj;
-	    m_size++;
+		this->growingCheck();
+		m_data[m_size] = obj;
+		m_size++;
 	}
 
 	//!Simply increase the m_size, doesn't call the new element constructor
 	inline void push_back_mem()
 	{
-	    this->growingCheck();
-	    m_size++;
+		this->growingCheck();
+		m_size++;
 	}
 
-	inline void push_back_memcpy(const T & obj)
+	inline void push_back_memcpy(const T& obj)
 	{
-	    this->growingCheck();
-	    gim_simd_memcpy(&m_data[m_size],&obj,sizeof(T));
-	    m_size++;
+		this->growingCheck();
+		gim_simd_memcpy(&m_data[m_size], &obj, sizeof(T));
+		m_size++;
 	}
 
 	inline void pop_back()
 	{
-	    m_size--;
-        m_data[m_size].~T();
+		m_size--;
+		m_data[m_size].~T();
 	}
 
 	//!Simply decrease the m_size, doesn't call the deleted element destructor
 	inline void pop_back_mem()
 	{
-	    m_size--;
+		m_size--;
 	}
 
-    //! fast erase
+	//! fast erase
 	inline void erase(GUINT index)
 	{
-	    if(index<m_size-1)
-	    {
-	        swap(index,m_size-1);
-	    }
-	    pop_back();
+		if (index < m_size - 1)
+		{
+			swap(index, m_size - 1);
+		}
+		pop_back();
 	}
 
 	inline void erase_sorted_mem(GUINT index)
 	{
-	    m_size--;
-	    for(GUINT i = index;i<m_size;i++)
-	    {
-	        gim_simd_memcpy(m_data+i,m_data+i+1,sizeof(T));
-	    }
+		m_size--;
+		for (GUINT i = index; i < m_size; i++)
+		{
+			gim_simd_memcpy(m_data + i, m_data + i + 1, sizeof(T));
+		}
 	}
 
 	inline void erase_sorted(GUINT index)
 	{
-	    m_data[index].~T();
-	    erase_sorted_mem(index);
+		m_data[index].~T();
+		erase_sorted_mem(index);
 	}
 
 	inline void insert_mem(GUINT index)
 	{
-	    this->growingCheck();
-	    for(GUINT i = m_size;i>index;i--)
-	    {
-	        gim_simd_memcpy(m_data+i,m_data+i-1,sizeof(T));
-	    }
-	    m_size++;
+		this->growingCheck();
+		for (GUINT i = m_size; i > index; i--)
+		{
+			gim_simd_memcpy(m_data + i, m_data + i - 1, sizeof(T));
+		}
+		m_size++;
 	}
 
-	inline void insert(const T & obj,GUINT index)
+	inline void insert(const T& obj, GUINT index)
 	{
-	    insert_mem(index);
-	    m_data[index] = obj;
+		insert_mem(index);
+		m_data[index] = obj;
 	}
 
-	inline void resize(GUINT size, bool call_constructor = true, const T& fillData=T())
+	inline void resize(GUINT size, bool call_constructor = true, const T& fillData = T())
 	{
-	    if(size>m_size)
-	    {
-            reserve(size);
-            if(call_constructor)
-            {
-                while(m_size<size)
-                {
-                    m_data[m_size] = fillData;
-                    m_size++;
-                }
-            }
-            else
-            {
-            	m_size = size;
-            }
-	    }
-	    else if(size<m_size)
-	    {
-	        if(call_constructor) clear_range(size);
-	        m_size = size;
-	    }
+		if (size > m_size)
+		{
+			reserve(size);
+			if (call_constructor)
+			{
+				while (m_size < size)
+				{
+					m_data[m_size] = fillData;
+					m_size++;
+				}
+			}
+			else
+			{
+				m_size = size;
+			}
+		}
+		else if (size < m_size)
+		{
+			if (call_constructor) clear_range(size);
+			m_size = size;
+		}
 	}
 
 	inline void refit()
 	{
-	    resizeData(m_size);
+		resizeData(m_size);
 	}
-
 };
 
-
-
-
-
-#endif // GIM_CONTAINERS_H_INCLUDED
+#endif  // GIM_CONTAINERS_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/gim_basic_geometry_operations.h b/thirdparty/bullet/BulletCollision/Gimpact/gim_basic_geometry_operations.h
index 0c48cb60fc5..7ab783672d6 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/gim_basic_geometry_operations.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/gim_basic_geometry_operations.h
@@ -35,12 +35,8 @@ email: projectileman@yahoo.com
 -----------------------------------------------------------------------------
 */
 
-
 #include "gim_linear_math.h"
 
-
-
-
 #ifndef PLANEDIREPSILON
 #define PLANEDIREPSILON 0.0000001f
 #endif
@@ -49,77 +45,82 @@ email: projectileman@yahoo.com
 #define PARALELENORMALS 0.000001f
 #endif
 
-#define TRIANGLE_NORMAL(v1,v2,v3,n)\
-{\
-	vec3f _dif1,_dif2;\
-    VEC_DIFF(_dif1,v2,v1);\
-    VEC_DIFF(_dif2,v3,v1);\
-    VEC_CROSS(n,_dif1,_dif2);\
-    VEC_NORMALIZE(n);\
-}\
+#define TRIANGLE_NORMAL(v1, v2, v3, n) \
+	{                                  \
+		vec3f _dif1, _dif2;            \
+		VEC_DIFF(_dif1, v2, v1);       \
+		VEC_DIFF(_dif2, v3, v1);       \
+		VEC_CROSS(n, _dif1, _dif2);    \
+		VEC_NORMALIZE(n);              \
+	}
 
-#define TRIANGLE_NORMAL_FAST(v1,v2,v3,n){\
-    vec3f _dif1,_dif2; \
-    VEC_DIFF(_dif1,v2,v1); \
-    VEC_DIFF(_dif2,v3,v1); \
-    VEC_CROSS(n,_dif1,_dif2); \
-}\
+#define TRIANGLE_NORMAL_FAST(v1, v2, v3, n) \
+	{                                       \
+		vec3f _dif1, _dif2;                 \
+		VEC_DIFF(_dif1, v2, v1);            \
+		VEC_DIFF(_dif2, v3, v1);            \
+		VEC_CROSS(n, _dif1, _dif2);         \
+	}
 
 /// plane is a vec4f
-#define TRIANGLE_PLANE(v1,v2,v3,plane) {\
-    TRIANGLE_NORMAL(v1,v2,v3,plane);\
-    plane[3] = VEC_DOT(v1,plane);\
-}\
+#define TRIANGLE_PLANE(v1, v2, v3, plane)   \
+	{                                       \
+		TRIANGLE_NORMAL(v1, v2, v3, plane); \
+		plane[3] = VEC_DOT(v1, plane);      \
+	}
 
 /// plane is a vec4f
-#define TRIANGLE_PLANE_FAST(v1,v2,v3,plane) {\
-    TRIANGLE_NORMAL_FAST(v1,v2,v3,plane);\
-    plane[3] = VEC_DOT(v1,plane);\
-}\
+#define TRIANGLE_PLANE_FAST(v1, v2, v3, plane)   \
+	{                                            \
+		TRIANGLE_NORMAL_FAST(v1, v2, v3, plane); \
+		plane[3] = VEC_DOT(v1, plane);           \
+	}
 
 /// Calc a plane from an edge an a normal. plane is a vec4f
-#define EDGE_PLANE(e1,e2,n,plane) {\
-    vec3f _dif; \
-    VEC_DIFF(_dif,e2,e1); \
-    VEC_CROSS(plane,_dif,n); \
-    VEC_NORMALIZE(plane); \
-    plane[3] = VEC_DOT(e1,plane);\
-}\
+#define EDGE_PLANE(e1, e2, n, plane)   \
+	{                                  \
+		vec3f _dif;                    \
+		VEC_DIFF(_dif, e2, e1);        \
+		VEC_CROSS(plane, _dif, n);     \
+		VEC_NORMALIZE(plane);          \
+		plane[3] = VEC_DOT(e1, plane); \
+	}
 
-#define DISTANCE_PLANE_POINT(plane,point) (VEC_DOT(plane,point) - plane[3])
+#define DISTANCE_PLANE_POINT(plane, point) (VEC_DOT(plane, point) - plane[3])
 
-#define PROJECT_POINT_PLANE(point,plane,projected) {\
-	GREAL _dis;\
-	_dis = DISTANCE_PLANE_POINT(plane,point);\
-	VEC_SCALE(projected,-_dis,plane);\
-	VEC_SUM(projected,projected,point);	\
-}\
+#define PROJECT_POINT_PLANE(point, plane, projected) \
+	{                                                \
+		GREAL _dis;                                  \
+		_dis = DISTANCE_PLANE_POINT(plane, point);   \
+		VEC_SCALE(projected, -_dis, plane);          \
+		VEC_SUM(projected, projected, point);        \
+	}
 
 //! Verifies if a point is in the plane hull
-template<typename CLASS_POINT,typename CLASS_PLANE>
+template <typename CLASS_POINT, typename CLASS_PLANE>
 SIMD_FORCE_INLINE bool POINT_IN_HULL(
-	const CLASS_POINT& point,const CLASS_PLANE * planes,GUINT plane_count)
+	const CLASS_POINT &point, const CLASS_PLANE *planes, GUINT plane_count)
 {
 	GREAL _dis;
-	for (GUINT _i = 0;_i< plane_count;++_i)
+	for (GUINT _i = 0; _i < plane_count; ++_i)
 	{
-		_dis = DISTANCE_PLANE_POINT(planes[_i],point);
-	    if(_dis>0.0f) return false;
+		_dis = DISTANCE_PLANE_POINT(planes[_i], point);
+		if (_dis > 0.0f) return false;
 	}
 	return true;
 }
 
-template<typename CLASS_POINT,typename CLASS_PLANE>
+template <typename CLASS_POINT, typename CLASS_PLANE>
 SIMD_FORCE_INLINE void PLANE_CLIP_SEGMENT(
-	const CLASS_POINT& s1,
-	const CLASS_POINT &s2,const CLASS_PLANE &plane,CLASS_POINT &clipped)
+	const CLASS_POINT &s1,
+	const CLASS_POINT &s2, const CLASS_PLANE &plane, CLASS_POINT &clipped)
 {
-	GREAL _dis1,_dis2;
-	_dis1 = DISTANCE_PLANE_POINT(plane,s1);
-	VEC_DIFF(clipped,s2,s1);
-	_dis2 = VEC_DOT(clipped,plane);
-	VEC_SCALE(clipped,-_dis1/_dis2,clipped);
-	VEC_SUM(clipped,clipped,s1);
+	GREAL _dis1, _dis2;
+	_dis1 = DISTANCE_PLANE_POINT(plane, s1);
+	VEC_DIFF(clipped, s2, s1);
+	_dis2 = VEC_DOT(clipped, plane);
+	VEC_SCALE(clipped, -_dis1 / _dis2, clipped);
+	VEC_SUM(clipped, clipped, s1);
 }
 
 enum ePLANE_INTERSECTION_TYPE
@@ -152,30 +153,30 @@ intersection type must have the following values
 </ul>
 */
 
-template<typename CLASS_POINT,typename CLASS_PLANE>
+template <typename CLASS_POINT, typename CLASS_PLANE>
 SIMD_FORCE_INLINE eLINE_PLANE_INTERSECTION_TYPE PLANE_CLIP_SEGMENT2(
-	const CLASS_POINT& s1,
+	const CLASS_POINT &s1,
 	const CLASS_POINT &s2,
-	const CLASS_PLANE &plane,CLASS_POINT &clipped)
+	const CLASS_PLANE &plane, CLASS_POINT &clipped)
 {
-	GREAL _dis1 = DISTANCE_PLANE_POINT(plane,s1);
-	GREAL _dis2 = DISTANCE_PLANE_POINT(plane,s2);
-	if(_dis1 >-G_EPSILON && _dis2 >-G_EPSILON)
+	GREAL _dis1 = DISTANCE_PLANE_POINT(plane, s1);
+	GREAL _dis2 = DISTANCE_PLANE_POINT(plane, s2);
+	if (_dis1 > -G_EPSILON && _dis2 > -G_EPSILON)
 	{
-	    if(_dis1<_dis2) return G_FRONT_PLANE_S1;
-	    return G_FRONT_PLANE_S2;
+		if (_dis1 < _dis2) return G_FRONT_PLANE_S1;
+		return G_FRONT_PLANE_S2;
 	}
-	else if(_dis1 <G_EPSILON && _dis2 <G_EPSILON)
+	else if (_dis1 < G_EPSILON && _dis2 < G_EPSILON)
 	{
-	    if(_dis1>_dis2) return G_BACK_PLANE_S1;
-	    return G_BACK_PLANE_S2;
+		if (_dis1 > _dis2) return G_BACK_PLANE_S1;
+		return G_BACK_PLANE_S2;
 	}
 
-	VEC_DIFF(clipped,s2,s1);
-	_dis2 = VEC_DOT(clipped,plane);
-	VEC_SCALE(clipped,-_dis1/_dis2,clipped);
-	VEC_SUM(clipped,clipped,s1);
-	if(_dis1<_dis2) return G_COLLIDE_PLANE_S1;
+	VEC_DIFF(clipped, s2, s1);
+	_dis2 = VEC_DOT(clipped, plane);
+	VEC_SCALE(clipped, -_dis1 / _dis2, clipped);
+	VEC_SUM(clipped, clipped, s1);
+	if (_dis1 < _dis2) return G_COLLIDE_PLANE_S1;
 	return G_COLLIDE_PLANE_S2;
 }
 
@@ -194,43 +195,42 @@ intersection_type must have the following values
 <li> 5 : Segment collides plane, s2 in back
 </ul>
 */
-template<typename CLASS_POINT,typename CLASS_PLANE>
+template <typename CLASS_POINT, typename CLASS_PLANE>
 SIMD_FORCE_INLINE eLINE_PLANE_INTERSECTION_TYPE PLANE_CLIP_SEGMENT_CLOSEST(
-	const CLASS_POINT& s1,
+	const CLASS_POINT &s1,
 	const CLASS_POINT &s2,
 	const CLASS_PLANE &plane,
-	CLASS_POINT &clipped1,CLASS_POINT &clipped2)
+	CLASS_POINT &clipped1, CLASS_POINT &clipped2)
 {
-	eLINE_PLANE_INTERSECTION_TYPE intersection_type = PLANE_CLIP_SEGMENT2(s1,s2,plane,clipped1);
-	switch(intersection_type)
+	eLINE_PLANE_INTERSECTION_TYPE intersection_type = PLANE_CLIP_SEGMENT2(s1, s2, plane, clipped1);
+	switch (intersection_type)
 	{
-	case G_FRONT_PLANE_S1:
-		VEC_COPY(clipped1,s1);
-	    VEC_COPY(clipped2,s2);
-		break;
-	case G_FRONT_PLANE_S2:
-		VEC_COPY(clipped1,s2);
-	    VEC_COPY(clipped2,s1);
-		break;
-	case G_BACK_PLANE_S1:
-		VEC_COPY(clipped1,s1);
-	    VEC_COPY(clipped2,s2);
-		break;
-	case G_BACK_PLANE_S2:
-		VEC_COPY(clipped1,s2);
-	    VEC_COPY(clipped2,s1);
-		break;
-	case G_COLLIDE_PLANE_S1:
-		VEC_COPY(clipped2,s1);
-		break;
-	case G_COLLIDE_PLANE_S2:
-		VEC_COPY(clipped2,s2);
-		break;
+		case G_FRONT_PLANE_S1:
+			VEC_COPY(clipped1, s1);
+			VEC_COPY(clipped2, s2);
+			break;
+		case G_FRONT_PLANE_S2:
+			VEC_COPY(clipped1, s2);
+			VEC_COPY(clipped2, s1);
+			break;
+		case G_BACK_PLANE_S1:
+			VEC_COPY(clipped1, s1);
+			VEC_COPY(clipped2, s2);
+			break;
+		case G_BACK_PLANE_S2:
+			VEC_COPY(clipped1, s2);
+			VEC_COPY(clipped2, s1);
+			break;
+		case G_COLLIDE_PLANE_S1:
+			VEC_COPY(clipped2, s1);
+			break;
+		case G_COLLIDE_PLANE_S2:
+			VEC_COPY(clipped2, s2);
+			break;
 	}
 	return intersection_type;
 }
 
-
 //! Finds the 2 smallest cartesian coordinates of a plane normal
 #define PLANE_MINOR_AXES(plane, i0, i1) VEC_MINOR_AXES(plane, i0, i1)
 
@@ -239,23 +239,23 @@ SIMD_FORCE_INLINE eLINE_PLANE_INTERSECTION_TYPE PLANE_CLIP_SEGMENT_CLOSEST(
 Intersects plane in one way only. The ray must face the plane (normals must be in opossite directions).<br/>
 It uses the PLANEDIREPSILON constant.
 */
-template<typename T,typename CLASS_POINT,typename CLASS_PLANE>
+template <typename T, typename CLASS_POINT, typename CLASS_PLANE>
 SIMD_FORCE_INLINE bool RAY_PLANE_COLLISION(
-	const CLASS_PLANE & plane,
-	const CLASS_POINT & vDir,
-	const CLASS_POINT & vPoint,
-	CLASS_POINT & pout,T &tparam)
+	const CLASS_PLANE &plane,
+	const CLASS_POINT &vDir,
+	const CLASS_POINT &vPoint,
+	CLASS_POINT &pout, T &tparam)
 {
-	GREAL _dis,_dotdir;
-	_dotdir = VEC_DOT(plane,vDir);
-	if(_dotdir<PLANEDIREPSILON)
+	GREAL _dis, _dotdir;
+	_dotdir = VEC_DOT(plane, vDir);
+	if (_dotdir < PLANEDIREPSILON)
 	{
-	    return false;
+		return false;
 	}
-	_dis = DISTANCE_PLANE_POINT(plane,vPoint);
-	tparam = -_dis/_dotdir;
-	VEC_SCALE(pout,tparam,vDir);
-	VEC_SUM(pout,vPoint,pout);
+	_dis = DISTANCE_PLANE_POINT(plane, vPoint);
+	tparam = -_dis / _dotdir;
+	VEC_SCALE(pout, tparam, vDir);
+	VEC_SUM(pout, vPoint, pout);
 	return true;
 }
 
@@ -266,39 +266,39 @@ SIMD_FORCE_INLINE bool RAY_PLANE_COLLISION(
 	-1 if the ray collides in front
 	-2 if the ray collides in back
 */
-template<typename T,typename CLASS_POINT,typename CLASS_PLANE>
+template <typename T, typename CLASS_POINT, typename CLASS_PLANE>
 SIMD_FORCE_INLINE GUINT LINE_PLANE_COLLISION(
-	const CLASS_PLANE & plane,
-	const CLASS_POINT & vDir,
-	const CLASS_POINT & vPoint,
-	CLASS_POINT & pout,
+	const CLASS_PLANE &plane,
+	const CLASS_POINT &vDir,
+	const CLASS_POINT &vPoint,
+	CLASS_POINT &pout,
 	T &tparam,
 	T tmin, T tmax)
 {
-	GREAL _dis,_dotdir;
-	_dotdir = VEC_DOT(plane,vDir);
-	if(btFabs(_dotdir)<PLANEDIREPSILON)
+	GREAL _dis, _dotdir;
+	_dotdir = VEC_DOT(plane, vDir);
+	if (btFabs(_dotdir) < PLANEDIREPSILON)
 	{
 		tparam = tmax;
-	    return 0;
+		return 0;
 	}
-	_dis = DISTANCE_PLANE_POINT(plane,vPoint);
-	char returnvalue = _dis<0.0f?2:1;
-	tparam = -_dis/_dotdir;
+	_dis = DISTANCE_PLANE_POINT(plane, vPoint);
+	char returnvalue = _dis < 0.0f ? 2 : 1;
+	tparam = -_dis / _dotdir;
 
-	if(tparam<tmin)
+	if (tparam < tmin)
 	{
 		returnvalue = 0;
 		tparam = tmin;
 	}
-	else if(tparam>tmax)
+	else if (tparam > tmax)
 	{
 		returnvalue = 0;
 		tparam = tmax;
 	}
 
-	VEC_SCALE(pout,tparam,vDir);
-	VEC_SUM(pout,vPoint,pout);
+	VEC_SCALE(pout, tparam, vDir);
+	VEC_SUM(pout, vPoint, pout);
 	return returnvalue;
 }
 
@@ -312,24 +312,24 @@ SIMD_FORCE_INLINE GUINT LINE_PLANE_COLLISION(
   \return true if the planes intersect, 0 if paralell.
 
 */
-template<typename CLASS_POINT,typename CLASS_PLANE>
+template <typename CLASS_POINT, typename CLASS_PLANE>
 SIMD_FORCE_INLINE bool INTERSECT_PLANES(
-		const CLASS_PLANE &p1,
-		const CLASS_PLANE &p2,
-		CLASS_POINT &p,
-		CLASS_POINT &d)
+	const CLASS_PLANE &p1,
+	const CLASS_PLANE &p2,
+	CLASS_POINT &p,
+	CLASS_POINT &d)
 {
-	VEC_CROSS(d,p1,p2);
-  	GREAL denom = VEC_DOT(d, d);
-  	if(GIM_IS_ZERO(denom)) return false;
+	VEC_CROSS(d, p1, p2);
+	GREAL denom = VEC_DOT(d, d);
+	if (GIM_IS_ZERO(denom)) return false;
 	vec3f _n;
-	_n[0]=p1[3]*p2[0] - p2[3]*p1[0];
-	_n[1]=p1[3]*p2[1] - p2[3]*p1[1];
-	_n[2]=p1[3]*p2[2] - p2[3]*p1[2];
-	VEC_CROSS(p,_n,d);
-	p[0]/=denom;
-	p[1]/=denom;
-	p[2]/=denom;
+	_n[0] = p1[3] * p2[0] - p2[3] * p1[0];
+	_n[1] = p1[3] * p2[1] - p2[3] * p1[1];
+	_n[2] = p1[3] * p2[2] - p2[3] * p1[2];
+	VEC_CROSS(p, _n, d);
+	p[0] /= denom;
+	p[1] /= denom;
+	p[2] /= denom;
 	return true;
 }
 
@@ -337,32 +337,31 @@ SIMD_FORCE_INLINE bool INTERSECT_PLANES(
 
 /*! Finds the closest point(cp) to (v) on a segment (e1,e2)
  */
-template<typename CLASS_POINT>
+template <typename CLASS_POINT>
 SIMD_FORCE_INLINE void CLOSEST_POINT_ON_SEGMENT(
-	CLASS_POINT & cp, const CLASS_POINT & v,
-	const CLASS_POINT &e1,const CLASS_POINT &e2)
+	CLASS_POINT &cp, const CLASS_POINT &v,
+	const CLASS_POINT &e1, const CLASS_POINT &e2)
 {
-    vec3f _n;
-    VEC_DIFF(_n,e2,e1);
-    VEC_DIFF(cp,v,e1);
+	vec3f _n;
+	VEC_DIFF(_n, e2, e1);
+	VEC_DIFF(cp, v, e1);
 	GREAL _scalar = VEC_DOT(cp, _n);
-	_scalar/= VEC_DOT(_n, _n);
-	if(_scalar <0.0f)
+	_scalar /= VEC_DOT(_n, _n);
+	if (_scalar < 0.0f)
 	{
-	    VEC_COPY(cp,e1);
+		VEC_COPY(cp, e1);
 	}
-	else if(_scalar >1.0f)
+	else if (_scalar > 1.0f)
 	{
-	    VEC_COPY(cp,e2);
+		VEC_COPY(cp, e2);
 	}
 	else
 	{
-        VEC_SCALE(cp,_scalar,_n);
-        VEC_SUM(cp,cp,e1);
+		VEC_SCALE(cp, _scalar, _n);
+		VEC_SUM(cp, cp, e1);
 	}
 }
 
-
 /*! \brief Finds the line params where these lines intersect.
 
 \param dir1 Direction of line 1
@@ -374,118 +373,114 @@ SIMD_FORCE_INLINE void CLOSEST_POINT_ON_SEGMENT(
 \param dointersect  0  if the lines won't intersect, else 1
 
 */
-template<typename T,typename CLASS_POINT>
+template <typename T, typename CLASS_POINT>
 SIMD_FORCE_INLINE bool LINE_INTERSECTION_PARAMS(
-	const CLASS_POINT & dir1,
-	CLASS_POINT & point1,
-	const CLASS_POINT & dir2,
-	CLASS_POINT &  point2,
-	T& t1,T& t2)
+	const CLASS_POINT &dir1,
+	CLASS_POINT &point1,
+	const CLASS_POINT &dir2,
+	CLASS_POINT &point2,
+	T &t1, T &t2)
 {
-    GREAL det;
-	GREAL e1e1 = VEC_DOT(dir1,dir1);
-	GREAL e1e2 = VEC_DOT(dir1,dir2);
-	GREAL e2e2 = VEC_DOT(dir2,dir2);
+	GREAL det;
+	GREAL e1e1 = VEC_DOT(dir1, dir1);
+	GREAL e1e2 = VEC_DOT(dir1, dir2);
+	GREAL e2e2 = VEC_DOT(dir2, dir2);
 	vec3f p1p2;
-    VEC_DIFF(p1p2,point1,point2);
-    GREAL p1p2e1 = VEC_DOT(p1p2,dir1);
-	GREAL p1p2e2 = VEC_DOT(p1p2,dir2);
-	det = e1e2*e1e2 - e1e1*e2e2;
-	if(GIM_IS_ZERO(det)) return false;
-	t1 = (e1e2*p1p2e2 - e2e2*p1p2e1)/det;
-	t2 = (e1e1*p1p2e2 - e1e2*p1p2e1)/det;
+	VEC_DIFF(p1p2, point1, point2);
+	GREAL p1p2e1 = VEC_DOT(p1p2, dir1);
+	GREAL p1p2e2 = VEC_DOT(p1p2, dir2);
+	det = e1e2 * e1e2 - e1e1 * e2e2;
+	if (GIM_IS_ZERO(det)) return false;
+	t1 = (e1e2 * p1p2e2 - e2e2 * p1p2e1) / det;
+	t2 = (e1e1 * p1p2e2 - e1e2 * p1p2e1) / det;
 	return true;
 }
 
 //! Find closest points on segments
-template<typename CLASS_POINT>
+template <typename CLASS_POINT>
 SIMD_FORCE_INLINE void SEGMENT_COLLISION(
-	const CLASS_POINT & vA1,
-	const CLASS_POINT & vA2,
-	const CLASS_POINT & vB1,
-	const CLASS_POINT & vB2,
-	CLASS_POINT & vPointA,
-	CLASS_POINT & vPointB)
+	const CLASS_POINT &vA1,
+	const CLASS_POINT &vA2,
+	const CLASS_POINT &vB1,
+	const CLASS_POINT &vB2,
+	CLASS_POINT &vPointA,
+	CLASS_POINT &vPointB)
 {
-    CLASS_POINT _AD,_BD,n;
-    vec4f _M;//plane
-    VEC_DIFF(_AD,vA2,vA1);
-    VEC_DIFF(_BD,vB2,vB1);
-    VEC_CROSS(n,_AD,_BD);
-    GREAL _tp = VEC_DOT(n,n);
-    if(_tp<G_EPSILON)//ARE PARALELE
-    {
-    	//project B over A
-    	bool invert_b_order = false;
-    	_M[0] = VEC_DOT(vB1,_AD);
-    	_M[1] = VEC_DOT(vB2,_AD);
-    	if(_M[0]>_M[1])
-    	{
-    		invert_b_order  = true;
-    		GIM_SWAP_NUMBERS(_M[0],_M[1]);
-    	}
-    	_M[2] = VEC_DOT(vA1,_AD);
-    	_M[3] = VEC_DOT(vA2,_AD);
-    	//mid points
-    	n[0] = (_M[0]+_M[1])*0.5f;
-    	n[1] = (_M[2]+_M[3])*0.5f;
+	CLASS_POINT _AD, _BD, n;
+	vec4f _M;  //plane
+	VEC_DIFF(_AD, vA2, vA1);
+	VEC_DIFF(_BD, vB2, vB1);
+	VEC_CROSS(n, _AD, _BD);
+	GREAL _tp = VEC_DOT(n, n);
+	if (_tp < G_EPSILON)  //ARE PARALELE
+	{
+		//project B over A
+		bool invert_b_order = false;
+		_M[0] = VEC_DOT(vB1, _AD);
+		_M[1] = VEC_DOT(vB2, _AD);
+		if (_M[0] > _M[1])
+		{
+			invert_b_order = true;
+			GIM_SWAP_NUMBERS(_M[0], _M[1]);
+		}
+		_M[2] = VEC_DOT(vA1, _AD);
+		_M[3] = VEC_DOT(vA2, _AD);
+		//mid points
+		n[0] = (_M[0] + _M[1]) * 0.5f;
+		n[1] = (_M[2] + _M[3]) * 0.5f;
 
-    	if(n[0]<n[1])
-    	{
-    		if(_M[1]<_M[2])
-    		{
-    			vPointB = invert_b_order?vB1:vB2;
-    			vPointA = vA1;
-    		}
-    		else if(_M[1]<_M[3])
-    		{
-    			vPointB = invert_b_order?vB1:vB2;
-    			CLOSEST_POINT_ON_SEGMENT(vPointA,vPointB,vA1,vA2);
-    		}
-    		else
-    		{
-    			vPointA = vA2;
-    			CLOSEST_POINT_ON_SEGMENT(vPointB,vPointA,vB1,vB2);
-    		}
-    	}
-    	else
-    	{
-    		if(_M[3]<_M[0])
-    		{
-    			vPointB = invert_b_order?vB2:vB1;
-    			vPointA = vA2;
-    		}
-    		else if(_M[3]<_M[1])
-    		{
-    			vPointA = vA2;
-    			CLOSEST_POINT_ON_SEGMENT(vPointB,vPointA,vB1,vB2);
-    		}
-    		else
-    		{
-    			vPointB = invert_b_order?vB1:vB2;
-    			CLOSEST_POINT_ON_SEGMENT(vPointA,vPointB,vA1,vA2);
-    		}
-    	}
-    	return;
-    }
+		if (n[0] < n[1])
+		{
+			if (_M[1] < _M[2])
+			{
+				vPointB = invert_b_order ? vB1 : vB2;
+				vPointA = vA1;
+			}
+			else if (_M[1] < _M[3])
+			{
+				vPointB = invert_b_order ? vB1 : vB2;
+				CLOSEST_POINT_ON_SEGMENT(vPointA, vPointB, vA1, vA2);
+			}
+			else
+			{
+				vPointA = vA2;
+				CLOSEST_POINT_ON_SEGMENT(vPointB, vPointA, vB1, vB2);
+			}
+		}
+		else
+		{
+			if (_M[3] < _M[0])
+			{
+				vPointB = invert_b_order ? vB2 : vB1;
+				vPointA = vA2;
+			}
+			else if (_M[3] < _M[1])
+			{
+				vPointA = vA2;
+				CLOSEST_POINT_ON_SEGMENT(vPointB, vPointA, vB1, vB2);
+			}
+			else
+			{
+				vPointB = invert_b_order ? vB1 : vB2;
+				CLOSEST_POINT_ON_SEGMENT(vPointA, vPointB, vA1, vA2);
+			}
+		}
+		return;
+	}
 
+	VEC_CROSS(_M, n, _BD);
+	_M[3] = VEC_DOT(_M, vB1);
 
-    VEC_CROSS(_M,n,_BD);
-    _M[3] = VEC_DOT(_M,vB1);
-
-    LINE_PLANE_COLLISION(_M,_AD,vA1,vPointA,_tp,btScalar(0), btScalar(1));
-    /*Closest point on segment*/
-    VEC_DIFF(vPointB,vPointA,vB1);
+	LINE_PLANE_COLLISION(_M, _AD, vA1, vPointA, _tp, btScalar(0), btScalar(1));
+	/*Closest point on segment*/
+	VEC_DIFF(vPointB, vPointA, vB1);
 	_tp = VEC_DOT(vPointB, _BD);
-	_tp/= VEC_DOT(_BD, _BD);
-	_tp = GIM_CLAMP(_tp,0.0f,1.0f);
-    VEC_SCALE(vPointB,_tp,_BD);
-    VEC_SUM(vPointB,vPointB,vB1);
+	_tp /= VEC_DOT(_BD, _BD);
+	_tp = GIM_CLAMP(_tp, 0.0f, 1.0f);
+	VEC_SCALE(vPointB, _tp, _BD);
+	VEC_SUM(vPointB, vPointB, vB1);
 }
 
-
-
-
 //! Line box intersection in one dimension
 /*!
 
@@ -497,37 +492,36 @@ SIMD_FORCE_INLINE void SEGMENT_COLLISION(
 *\param tlast the maximum projection. Assign to INFINITY at first.
 *\return true if there is an intersection.
 */
-template<typename T>
-SIMD_FORCE_INLINE bool BOX_AXIS_INTERSECT(T pos, T dir,T bmin, T bmax, T & tfirst, T & tlast)
+template <typename T>
+SIMD_FORCE_INLINE bool BOX_AXIS_INTERSECT(T pos, T dir, T bmin, T bmax, T &tfirst, T &tlast)
 {
-	if(GIM_IS_ZERO(dir))
+	if (GIM_IS_ZERO(dir))
 	{
-        return !(pos < bmin || pos > bmax);
+		return !(pos < bmin || pos > bmax);
 	}
 	GREAL a0 = (bmin - pos) / dir;
 	GREAL a1 = (bmax - pos) / dir;
-	if(a0 > a1)   GIM_SWAP_NUMBERS(a0, a1);
+	if (a0 > a1) GIM_SWAP_NUMBERS(a0, a1);
 	tfirst = GIM_MAX(a0, tfirst);
 	tlast = GIM_MIN(a1, tlast);
 	if (tlast < tfirst) return false;
 	return true;
 }
 
-
 //! Sorts 3 componets
-template<typename T>
+template <typename T>
 SIMD_FORCE_INLINE void SORT_3_INDICES(
-		const T * values,
-		GUINT * order_indices)
+	const T *values,
+	GUINT *order_indices)
 {
 	//get minimum
 	order_indices[0] = values[0] < values[1] ? (values[0] < values[2] ? 0 : 2) : (values[1] < values[2] ? 1 : 2);
 
 	//get second and third
-	GUINT i0 = (order_indices[0] + 1)%3;
-	GUINT i1 = (i0 + 1)%3;
+	GUINT i0 = (order_indices[0] + 1) % 3;
+	GUINT i1 = (i0 + 1) % 3;
 
-	if(values[i0] < values[i1])
+	if (values[i0] < values[i1])
 	{
 		order_indices[1] = i0;
 		order_indices[2] = i1;
@@ -539,8 +533,4 @@ SIMD_FORCE_INLINE void SORT_3_INDICES(
 	}
 }
 
-
-
-
-
-#endif // GIM_VECTOR_H_INCLUDED
+#endif  // GIM_VECTOR_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/gim_bitset.h b/thirdparty/bullet/BulletCollision/Gimpact/gim_bitset.h
index 7dee48a4c76..c1fb41a5c0c 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/gim_bitset.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/gim_bitset.h
@@ -34,34 +34,32 @@ email: projectileman@yahoo.com
 
 #include "gim_array.h"
 
-
 #define GUINT_BIT_COUNT 32
 #define GUINT_EXPONENT 5
 
 class gim_bitset
 {
 public:
-    gim_array<GUINT> m_container;
+	gim_array<GUINT> m_container;
 
-    gim_bitset()
-    {
+	gim_bitset()
+	{
+	}
 
-    }
+	gim_bitset(GUINT bits_count)
+	{
+		resize(bits_count);
+	}
 
-    gim_bitset(GUINT bits_count)
-    {
-        resize(bits_count);
-    }
-
-    ~gim_bitset()
-    {
-    }
+	~gim_bitset()
+	{
+	}
 
 	inline bool resize(GUINT newsize)
 	{
 		GUINT oldsize = m_container.size();
-		m_container.resize(newsize/GUINT_BIT_COUNT + 1,false);
-		while(oldsize<m_container.size())
+		m_container.resize(newsize / GUINT_BIT_COUNT + 1, false);
+		while (oldsize < m_container.size())
 		{
 			m_container[oldsize] = 0;
 		}
@@ -70,12 +68,12 @@ public:
 
 	inline GUINT size()
 	{
-		return m_container.size()*GUINT_BIT_COUNT;
+		return m_container.size() * GUINT_BIT_COUNT;
 	}
 
 	inline void set_all()
 	{
-		for(GUINT i = 0;i<m_container.size();++i)
+		for (GUINT i = 0; i < m_container.size(); ++i)
 		{
 			m_container[i] = 0xffffffff;
 		}
@@ -83,7 +81,7 @@ public:
 
 	inline void clear_all()
 	{
-	    for(GUINT i = 0;i<m_container.size();++i)
+		for (GUINT i = 0; i < m_container.size(); ++i)
 		{
 			m_container[i] = 0;
 		}
@@ -91,33 +89,29 @@ public:
 
 	inline void set(GUINT bit_index)
 	{
-		if(bit_index>=size())
+		if (bit_index >= size())
 		{
 			resize(bit_index);
 		}
-		m_container[bit_index >> GUINT_EXPONENT] |= (1 << (bit_index & (GUINT_BIT_COUNT-1)));
+		m_container[bit_index >> GUINT_EXPONENT] |= (1 << (bit_index & (GUINT_BIT_COUNT - 1)));
 	}
 
 	///Return 0 or 1
 	inline char get(GUINT bit_index)
 	{
-		if(bit_index>=size())
+		if (bit_index >= size())
 		{
 			return 0;
 		}
 		char value = m_container[bit_index >> GUINT_EXPONENT] &
-					 (1 << (bit_index & (GUINT_BIT_COUNT-1)));
+					 (1 << (bit_index & (GUINT_BIT_COUNT - 1)));
 		return value;
 	}
 
 	inline void clear(GUINT bit_index)
 	{
-	    m_container[bit_index >> GUINT_EXPONENT] &= ~(1 << (bit_index & (GUINT_BIT_COUNT-1)));
+		m_container[bit_index >> GUINT_EXPONENT] &= ~(1 << (bit_index & (GUINT_BIT_COUNT - 1)));
 	}
 };
 
-
-
-
-
-#endif // GIM_CONTAINERS_H_INCLUDED
+#endif  // GIM_CONTAINERS_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/gim_box_collision.h b/thirdparty/bullet/BulletCollision/Gimpact/gim_box_collision.h
index a051b4fdbf2..9f7cbe732fa 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/gim_box_collision.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/gim_box_collision.h
@@ -35,8 +35,6 @@ email: projectileman@yahoo.com
 #include "gim_basic_geometry_operations.h"
 #include "LinearMath/btTransform.h"
 
-
-
 //SIMD_FORCE_INLINE bool test_cross_edge_box(
 //	const btVector3 & edge,
 //	const btVector3 & absolute_edge,
@@ -99,52 +97,50 @@ email: projectileman@yahoo.com
 
 #ifndef TEST_CROSS_EDGE_BOX_MCR
 
-#define TEST_CROSS_EDGE_BOX_MCR(edge,absolute_edge,pointa,pointb,_extend,i_dir_0,i_dir_1,i_comp_0,i_comp_1)\
-{\
-	const btScalar dir0 = -edge[i_dir_0];\
-	const btScalar dir1 = edge[i_dir_1];\
-	btScalar pmin = pointa[i_comp_0]*dir0 + pointa[i_comp_1]*dir1;\
-	btScalar pmax = pointb[i_comp_0]*dir0 + pointb[i_comp_1]*dir1;\
-	if(pmin>pmax)\
-	{\
-		GIM_SWAP_NUMBERS(pmin,pmax); \
-	}\
-	const btScalar abs_dir0 = absolute_edge[i_dir_0];\
-	const btScalar abs_dir1 = absolute_edge[i_dir_1];\
-	const btScalar rad = _extend[i_comp_0] * abs_dir0 + _extend[i_comp_1] * abs_dir1;\
-	if(pmin>rad || -rad>pmax) return false;\
-}\
+#define TEST_CROSS_EDGE_BOX_MCR(edge, absolute_edge, pointa, pointb, _extend, i_dir_0, i_dir_1, i_comp_0, i_comp_1) \
+	{                                                                                                               \
+		const btScalar dir0 = -edge[i_dir_0];                                                                       \
+		const btScalar dir1 = edge[i_dir_1];                                                                        \
+		btScalar pmin = pointa[i_comp_0] * dir0 + pointa[i_comp_1] * dir1;                                          \
+		btScalar pmax = pointb[i_comp_0] * dir0 + pointb[i_comp_1] * dir1;                                          \
+		if (pmin > pmax)                                                                                            \
+		{                                                                                                           \
+			GIM_SWAP_NUMBERS(pmin, pmax);                                                                           \
+		}                                                                                                           \
+		const btScalar abs_dir0 = absolute_edge[i_dir_0];                                                           \
+		const btScalar abs_dir1 = absolute_edge[i_dir_1];                                                           \
+		const btScalar rad = _extend[i_comp_0] * abs_dir0 + _extend[i_comp_1] * abs_dir1;                           \
+		if (pmin > rad || -rad > pmax) return false;                                                                \
+	}
 
 #endif
 
-#define TEST_CROSS_EDGE_BOX_X_AXIS_MCR(edge,absolute_edge,pointa,pointb,_extend)\
-{\
-	TEST_CROSS_EDGE_BOX_MCR(edge,absolute_edge,pointa,pointb,_extend,2,1,1,2);\
-}\
-
-#define TEST_CROSS_EDGE_BOX_Y_AXIS_MCR(edge,absolute_edge,pointa,pointb,_extend)\
-{\
-	TEST_CROSS_EDGE_BOX_MCR(edge,absolute_edge,pointa,pointb,_extend,0,2,2,0);\
-}\
-
-#define TEST_CROSS_EDGE_BOX_Z_AXIS_MCR(edge,absolute_edge,pointa,pointb,_extend)\
-{\
-	TEST_CROSS_EDGE_BOX_MCR(edge,absolute_edge,pointa,pointb,_extend,1,0,0,1);\
-}\
+#define TEST_CROSS_EDGE_BOX_X_AXIS_MCR(edge, absolute_edge, pointa, pointb, _extend)       \
+	{                                                                                      \
+		TEST_CROSS_EDGE_BOX_MCR(edge, absolute_edge, pointa, pointb, _extend, 2, 1, 1, 2); \
+	}
 
+#define TEST_CROSS_EDGE_BOX_Y_AXIS_MCR(edge, absolute_edge, pointa, pointb, _extend)       \
+	{                                                                                      \
+		TEST_CROSS_EDGE_BOX_MCR(edge, absolute_edge, pointa, pointb, _extend, 0, 2, 2, 0); \
+	}
 
+#define TEST_CROSS_EDGE_BOX_Z_AXIS_MCR(edge, absolute_edge, pointa, pointb, _extend)       \
+	{                                                                                      \
+		TEST_CROSS_EDGE_BOX_MCR(edge, absolute_edge, pointa, pointb, _extend, 1, 0, 0, 1); \
+	}
 
 //!  Class for transforming a model1 to the space of model0
 class GIM_BOX_BOX_TRANSFORM_CACHE
 {
 public:
-    btVector3  m_T1to0;//!< Transforms translation of model1 to model 0
-	btMatrix3x3 m_R1to0;//!< Transforms Rotation of model1 to model 0, equal  to R0' * R1
-	btMatrix3x3 m_AR;//!< Absolute value of m_R1to0
+	btVector3 m_T1to0;    //!< Transforms translation of model1 to model 0
+	btMatrix3x3 m_R1to0;  //!< Transforms Rotation of model1 to model 0, equal  to R0' * R1
+	btMatrix3x3 m_AR;     //!< Absolute value of m_R1to0
 
 	SIMD_FORCE_INLINE void calc_absolute_matrix()
 	{
-		static const btVector3 vepsi(1e-6f,1e-6f,1e-6f);
+		static const btVector3 vepsi(1e-6f, 1e-6f, 1e-6f);
 		m_AR[0] = vepsi + m_R1to0[0].absolute();
 		m_AR[1] = vepsi + m_R1to0[1].absolute();
 		m_AR[2] = vepsi + m_R1to0[2].absolute();
@@ -154,42 +150,40 @@ public:
 	{
 	}
 
-
-	GIM_BOX_BOX_TRANSFORM_CACHE(mat4f  trans1_to_0)
+	GIM_BOX_BOX_TRANSFORM_CACHE(mat4f trans1_to_0)
 	{
-		COPY_MATRIX_3X3(m_R1to0,trans1_to_0)
-        MAT_GET_TRANSLATION(trans1_to_0,m_T1to0)
+		COPY_MATRIX_3X3(m_R1to0, trans1_to_0)
+		MAT_GET_TRANSLATION(trans1_to_0, m_T1to0)
 		calc_absolute_matrix();
 	}
 
 	//! Calc the transformation relative  1 to 0. Inverts matrics by transposing
-	SIMD_FORCE_INLINE void calc_from_homogenic(const btTransform & trans0,const btTransform & trans1)
+	SIMD_FORCE_INLINE void calc_from_homogenic(const btTransform &trans0, const btTransform &trans1)
 	{
-
 		m_R1to0 = trans0.getBasis().transpose();
 		m_T1to0 = m_R1to0 * (-trans0.getOrigin());
 
-		m_T1to0 += m_R1to0*trans1.getOrigin();
+		m_T1to0 += m_R1to0 * trans1.getOrigin();
 		m_R1to0 *= trans1.getBasis();
 
 		calc_absolute_matrix();
 	}
 
 	//! Calcs the full invertion of the matrices. Useful for scaling matrices
-	SIMD_FORCE_INLINE void calc_from_full_invert(const btTransform & trans0,const btTransform & trans1)
+	SIMD_FORCE_INLINE void calc_from_full_invert(const btTransform &trans0, const btTransform &trans1)
 	{
 		m_R1to0 = trans0.getBasis().inverse();
 		m_T1to0 = m_R1to0 * (-trans0.getOrigin());
 
-		m_T1to0 += m_R1to0*trans1.getOrigin();
+		m_T1to0 += m_R1to0 * trans1.getOrigin();
 		m_R1to0 *= trans1.getBasis();
 
 		calc_absolute_matrix();
 	}
 
-	SIMD_FORCE_INLINE btVector3 transform(const btVector3 & point)
+	SIMD_FORCE_INLINE btVector3 transform(const btVector3 &point)
 	{
-        return point.dot3(m_R1to0[0], m_R1to0[1], m_R1to0[2]) + m_T1to0;
+		return point.dot3(m_R1to0[0], m_R1to0[1], m_R1to0[2]) + m_T1to0;
 	}
 };
 
@@ -205,34 +199,34 @@ public:
 	btVector3 m_max;
 
 	GIM_AABB()
-	{}
-
-
-	GIM_AABB(const btVector3 & V1,
-			 const btVector3 & V2,
-			 const btVector3 & V3)
 	{
-		m_min[0] = GIM_MIN3(V1[0],V2[0],V3[0]);
-		m_min[1] = GIM_MIN3(V1[1],V2[1],V3[1]);
-		m_min[2] = GIM_MIN3(V1[2],V2[2],V3[2]);
-
-		m_max[0] = GIM_MAX3(V1[0],V2[0],V3[0]);
-		m_max[1] = GIM_MAX3(V1[1],V2[1],V3[1]);
-		m_max[2] = GIM_MAX3(V1[2],V2[2],V3[2]);
 	}
 
-	GIM_AABB(const btVector3 & V1,
-			 const btVector3 & V2,
-			 const btVector3 & V3,
+	GIM_AABB(const btVector3 &V1,
+			 const btVector3 &V2,
+			 const btVector3 &V3)
+	{
+		m_min[0] = GIM_MIN3(V1[0], V2[0], V3[0]);
+		m_min[1] = GIM_MIN3(V1[1], V2[1], V3[1]);
+		m_min[2] = GIM_MIN3(V1[2], V2[2], V3[2]);
+
+		m_max[0] = GIM_MAX3(V1[0], V2[0], V3[0]);
+		m_max[1] = GIM_MAX3(V1[1], V2[1], V3[1]);
+		m_max[2] = GIM_MAX3(V1[2], V2[2], V3[2]);
+	}
+
+	GIM_AABB(const btVector3 &V1,
+			 const btVector3 &V2,
+			 const btVector3 &V3,
 			 GREAL margin)
 	{
-		m_min[0] = GIM_MIN3(V1[0],V2[0],V3[0]);
-		m_min[1] = GIM_MIN3(V1[1],V2[1],V3[1]);
-		m_min[2] = GIM_MIN3(V1[2],V2[2],V3[2]);
+		m_min[0] = GIM_MIN3(V1[0], V2[0], V3[0]);
+		m_min[1] = GIM_MIN3(V1[1], V2[1], V3[1]);
+		m_min[2] = GIM_MIN3(V1[2], V2[2], V3[2]);
 
-		m_max[0] = GIM_MAX3(V1[0],V2[0],V3[0]);
-		m_max[1] = GIM_MAX3(V1[1],V2[1],V3[1]);
-		m_max[2] = GIM_MAX3(V1[2],V2[2],V3[2]);
+		m_max[0] = GIM_MAX3(V1[0], V2[0], V3[0]);
+		m_max[1] = GIM_MAX3(V1[1], V2[1], V3[1]);
+		m_max[2] = GIM_MAX3(V1[2], V2[2], V3[2]);
 
 		m_min[0] -= margin;
 		m_min[1] -= margin;
@@ -242,13 +236,11 @@ public:
 		m_max[2] += margin;
 	}
 
-	GIM_AABB(const GIM_AABB &other):
-		m_min(other.m_min),m_max(other.m_max)
+	GIM_AABB(const GIM_AABB &other) : m_min(other.m_min), m_max(other.m_max)
 	{
 	}
 
-	GIM_AABB(const GIM_AABB &other,btScalar margin ):
-		m_min(other.m_min),m_max(other.m_max)
+	GIM_AABB(const GIM_AABB &other, btScalar margin) : m_min(other.m_min), m_max(other.m_max)
 	{
 		m_min[0] -= margin;
 		m_min[1] -= margin;
@@ -289,34 +281,34 @@ public:
 		m_max[2] = other.m_max[2] + margin;
 	}
 
-	template<typename CLASS_POINT>
+	template <typename CLASS_POINT>
 	SIMD_FORCE_INLINE void calc_from_triangle(
-							const CLASS_POINT & V1,
-							const CLASS_POINT & V2,
-							const CLASS_POINT & V3)
+		const CLASS_POINT &V1,
+		const CLASS_POINT &V2,
+		const CLASS_POINT &V3)
 	{
-		m_min[0] = GIM_MIN3(V1[0],V2[0],V3[0]);
-		m_min[1] = GIM_MIN3(V1[1],V2[1],V3[1]);
-		m_min[2] = GIM_MIN3(V1[2],V2[2],V3[2]);
+		m_min[0] = GIM_MIN3(V1[0], V2[0], V3[0]);
+		m_min[1] = GIM_MIN3(V1[1], V2[1], V3[1]);
+		m_min[2] = GIM_MIN3(V1[2], V2[2], V3[2]);
 
-		m_max[0] = GIM_MAX3(V1[0],V2[0],V3[0]);
-		m_max[1] = GIM_MAX3(V1[1],V2[1],V3[1]);
-		m_max[2] = GIM_MAX3(V1[2],V2[2],V3[2]);
+		m_max[0] = GIM_MAX3(V1[0], V2[0], V3[0]);
+		m_max[1] = GIM_MAX3(V1[1], V2[1], V3[1]);
+		m_max[2] = GIM_MAX3(V1[2], V2[2], V3[2]);
 	}
 
-	template<typename CLASS_POINT>
+	template <typename CLASS_POINT>
 	SIMD_FORCE_INLINE void calc_from_triangle_margin(
-							const CLASS_POINT & V1,
-							const CLASS_POINT & V2,
-							const CLASS_POINT & V3, btScalar margin)
+		const CLASS_POINT &V1,
+		const CLASS_POINT &V2,
+		const CLASS_POINT &V3, btScalar margin)
 	{
-		m_min[0] = GIM_MIN3(V1[0],V2[0],V3[0]);
-		m_min[1] = GIM_MIN3(V1[1],V2[1],V3[1]);
-		m_min[2] = GIM_MIN3(V1[2],V2[2],V3[2]);
+		m_min[0] = GIM_MIN3(V1[0], V2[0], V3[0]);
+		m_min[1] = GIM_MIN3(V1[1], V2[1], V3[1]);
+		m_min[2] = GIM_MIN3(V1[2], V2[2], V3[2]);
 
-		m_max[0] = GIM_MAX3(V1[0],V2[0],V3[0]);
-		m_max[1] = GIM_MAX3(V1[1],V2[1],V3[1]);
-		m_max[2] = GIM_MAX3(V1[2],V2[2],V3[2]);
+		m_max[0] = GIM_MAX3(V1[0], V2[0], V3[0]);
+		m_max[1] = GIM_MAX3(V1[1], V2[1], V3[1]);
+		m_max[2] = GIM_MAX3(V1[2], V2[2], V3[2]);
 
 		m_min[0] -= margin;
 		m_min[1] -= margin;
@@ -327,74 +319,73 @@ public:
 	}
 
 	//! Apply a transform to an AABB
-	SIMD_FORCE_INLINE void appy_transform(const btTransform & trans)
+	SIMD_FORCE_INLINE void appy_transform(const btTransform &trans)
 	{
-		btVector3 center = (m_max+m_min)*0.5f;
+		btVector3 center = (m_max + m_min) * 0.5f;
 		btVector3 extends = m_max - center;
 		// Compute new center
 		center = trans(center);
 
-        btVector3 textends = extends.dot3(trans.getBasis().getRow(0).absolute(), 
-                                          trans.getBasis().getRow(1).absolute(), 
-                                          trans.getBasis().getRow(2).absolute());
-        
+		btVector3 textends = extends.dot3(trans.getBasis().getRow(0).absolute(),
+										  trans.getBasis().getRow(1).absolute(),
+										  trans.getBasis().getRow(2).absolute());
+
 		m_min = center - textends;
 		m_max = center + textends;
 	}
 
 	//! Merges a Box
-	SIMD_FORCE_INLINE void merge(const GIM_AABB & box)
+	SIMD_FORCE_INLINE void merge(const GIM_AABB &box)
 	{
-		m_min[0] = GIM_MIN(m_min[0],box.m_min[0]);
-		m_min[1] = GIM_MIN(m_min[1],box.m_min[1]);
-		m_min[2] = GIM_MIN(m_min[2],box.m_min[2]);
+		m_min[0] = GIM_MIN(m_min[0], box.m_min[0]);
+		m_min[1] = GIM_MIN(m_min[1], box.m_min[1]);
+		m_min[2] = GIM_MIN(m_min[2], box.m_min[2]);
 
-		m_max[0] = GIM_MAX(m_max[0],box.m_max[0]);
-		m_max[1] = GIM_MAX(m_max[1],box.m_max[1]);
-		m_max[2] = GIM_MAX(m_max[2],box.m_max[2]);
+		m_max[0] = GIM_MAX(m_max[0], box.m_max[0]);
+		m_max[1] = GIM_MAX(m_max[1], box.m_max[1]);
+		m_max[2] = GIM_MAX(m_max[2], box.m_max[2]);
 	}
 
 	//! Merges a point
-	template<typename CLASS_POINT>
-	SIMD_FORCE_INLINE void merge_point(const CLASS_POINT & point)
+	template <typename CLASS_POINT>
+	SIMD_FORCE_INLINE void merge_point(const CLASS_POINT &point)
 	{
-		m_min[0] = GIM_MIN(m_min[0],point[0]);
-		m_min[1] = GIM_MIN(m_min[1],point[1]);
-		m_min[2] = GIM_MIN(m_min[2],point[2]);
+		m_min[0] = GIM_MIN(m_min[0], point[0]);
+		m_min[1] = GIM_MIN(m_min[1], point[1]);
+		m_min[2] = GIM_MIN(m_min[2], point[2]);
 
-		m_max[0] = GIM_MAX(m_max[0],point[0]);
-		m_max[1] = GIM_MAX(m_max[1],point[1]);
-		m_max[2] = GIM_MAX(m_max[2],point[2]);
+		m_max[0] = GIM_MAX(m_max[0], point[0]);
+		m_max[1] = GIM_MAX(m_max[1], point[1]);
+		m_max[2] = GIM_MAX(m_max[2], point[2]);
 	}
 
 	//! Gets the extend and center
-	SIMD_FORCE_INLINE void get_center_extend(btVector3 & center,btVector3 & extend)  const
+	SIMD_FORCE_INLINE void get_center_extend(btVector3 &center, btVector3 &extend) const
 	{
-		center = (m_max+m_min)*0.5f;
+		center = (m_max + m_min) * 0.5f;
 		extend = m_max - center;
 	}
 
 	//! Finds the intersecting box between this box and the other.
-	SIMD_FORCE_INLINE void find_intersection(const GIM_AABB & other, GIM_AABB & intersection)  const
+	SIMD_FORCE_INLINE void find_intersection(const GIM_AABB &other, GIM_AABB &intersection) const
 	{
-		intersection.m_min[0] = GIM_MAX(other.m_min[0],m_min[0]);
-		intersection.m_min[1] = GIM_MAX(other.m_min[1],m_min[1]);
-		intersection.m_min[2] = GIM_MAX(other.m_min[2],m_min[2]);
+		intersection.m_min[0] = GIM_MAX(other.m_min[0], m_min[0]);
+		intersection.m_min[1] = GIM_MAX(other.m_min[1], m_min[1]);
+		intersection.m_min[2] = GIM_MAX(other.m_min[2], m_min[2]);
 
-		intersection.m_max[0] = GIM_MIN(other.m_max[0],m_max[0]);
-		intersection.m_max[1] = GIM_MIN(other.m_max[1],m_max[1]);
-		intersection.m_max[2] = GIM_MIN(other.m_max[2],m_max[2]);
+		intersection.m_max[0] = GIM_MIN(other.m_max[0], m_max[0]);
+		intersection.m_max[1] = GIM_MIN(other.m_max[1], m_max[1]);
+		intersection.m_max[2] = GIM_MIN(other.m_max[2], m_max[2]);
 	}
 
-
-	SIMD_FORCE_INLINE bool has_collision(const GIM_AABB & other) const
+	SIMD_FORCE_INLINE bool has_collision(const GIM_AABB &other) const
 	{
-		if(m_min[0] > other.m_max[0] ||
-		   m_max[0] < other.m_min[0] ||
-		   m_min[1] > other.m_max[1] ||
-		   m_max[1] < other.m_min[1] ||
-		   m_min[2] > other.m_max[2] ||
-		   m_max[2] < other.m_min[2])
+		if (m_min[0] > other.m_max[0] ||
+			m_max[0] < other.m_min[0] ||
+			m_min[1] > other.m_max[1] ||
+			m_max[1] < other.m_min[1] ||
+			m_min[2] > other.m_max[2] ||
+			m_max[2] < other.m_min[2])
 		{
 			return false;
 		}
@@ -406,35 +397,34 @@ public:
 	\param vorigin A vec3f with the origin of the ray
 	\param vdir A vec3f with the direction of the ray
 	*/
-	SIMD_FORCE_INLINE bool collide_ray(const btVector3 & vorigin,const btVector3 & vdir)
+	SIMD_FORCE_INLINE bool collide_ray(const btVector3 &vorigin, const btVector3 &vdir)
 	{
-		btVector3 extents,center;
-		this->get_center_extend(center,extents);;
+		btVector3 extents, center;
+		this->get_center_extend(center, extents);
+		;
 
 		btScalar Dx = vorigin[0] - center[0];
-		if(GIM_GREATER(Dx, extents[0]) && Dx*vdir[0]>=0.0f)	return false;
+		if (GIM_GREATER(Dx, extents[0]) && Dx * vdir[0] >= 0.0f) return false;
 		btScalar Dy = vorigin[1] - center[1];
-		if(GIM_GREATER(Dy, extents[1]) && Dy*vdir[1]>=0.0f)	return false;
+		if (GIM_GREATER(Dy, extents[1]) && Dy * vdir[1] >= 0.0f) return false;
 		btScalar Dz = vorigin[2] - center[2];
-		if(GIM_GREATER(Dz, extents[2]) && Dz*vdir[2]>=0.0f)	return false;
-
+		if (GIM_GREATER(Dz, extents[2]) && Dz * vdir[2] >= 0.0f) return false;
 
 		btScalar f = vdir[1] * Dz - vdir[2] * Dy;
-		if(btFabs(f) > extents[1]*btFabs(vdir[2]) + extents[2]*btFabs(vdir[1])) return false;
+		if (btFabs(f) > extents[1] * btFabs(vdir[2]) + extents[2] * btFabs(vdir[1])) return false;
 		f = vdir[2] * Dx - vdir[0] * Dz;
-		if(btFabs(f) > extents[0]*btFabs(vdir[2]) + extents[2]*btFabs(vdir[0]))return false;
+		if (btFabs(f) > extents[0] * btFabs(vdir[2]) + extents[2] * btFabs(vdir[0])) return false;
 		f = vdir[0] * Dy - vdir[1] * Dx;
-		if(btFabs(f) > extents[0]*btFabs(vdir[1]) + extents[1]*btFabs(vdir[0]))return false;
+		if (btFabs(f) > extents[0] * btFabs(vdir[1]) + extents[1] * btFabs(vdir[0])) return false;
 		return true;
 	}
 
-
-	SIMD_FORCE_INLINE void projection_interval(const btVector3 & direction, btScalar &vmin, btScalar &vmax) const
+	SIMD_FORCE_INLINE void projection_interval(const btVector3 &direction, btScalar &vmin, btScalar &vmax) const
 	{
-		btVector3 center = (m_max+m_min)*0.5f;
-		btVector3 extend = m_max-center;
+		btVector3 center = (m_max + m_min) * 0.5f;
+		btVector3 extend = m_max - center;
 
-		btScalar _fOrigin =  direction.dot(center);
+		btScalar _fOrigin = direction.dot(center);
 		btScalar _fMaximumExtent = extend.dot(direction.absolute());
 		vmin = _fOrigin - _fMaximumExtent;
 		vmax = _fOrigin + _fMaximumExtent;
@@ -442,22 +432,22 @@ public:
 
 	SIMD_FORCE_INLINE ePLANE_INTERSECTION_TYPE plane_classify(const btVector4 &plane) const
 	{
-		btScalar _fmin,_fmax;
-		this->projection_interval(plane,_fmin,_fmax);
+		btScalar _fmin, _fmax;
+		this->projection_interval(plane, _fmin, _fmax);
 
-		if(plane[3] > _fmax + BOX_PLANE_EPSILON)
+		if (plane[3] > _fmax + BOX_PLANE_EPSILON)
 		{
-			return G_BACK_PLANE; // 0
+			return G_BACK_PLANE;  // 0
 		}
 
-		if(plane[3]+BOX_PLANE_EPSILON >=_fmin)
+		if (plane[3] + BOX_PLANE_EPSILON >= _fmin)
 		{
-			return G_COLLIDE_PLANE; //1
+			return G_COLLIDE_PLANE;  //1
 		}
-		return G_FRONT_PLANE;//2
+		return G_FRONT_PLANE;  //2
 	}
 
-	SIMD_FORCE_INLINE bool overlapping_trans_conservative(const GIM_AABB & box, btTransform & trans1_to_0)
+	SIMD_FORCE_INLINE bool overlapping_trans_conservative(const GIM_AABB &box, btTransform &trans1_to_0)
 	{
 		GIM_AABB tbox = box;
 		tbox.appy_transform(trans1_to_0);
@@ -466,52 +456,50 @@ public:
 
 	//! transcache is the transformation cache from box to this AABB
 	SIMD_FORCE_INLINE bool overlapping_trans_cache(
-		const GIM_AABB & box,const GIM_BOX_BOX_TRANSFORM_CACHE & transcache, bool fulltest)
+		const GIM_AABB &box, const GIM_BOX_BOX_TRANSFORM_CACHE &transcache, bool fulltest)
 	{
-
 		//Taken from OPCODE
-		btVector3 ea,eb;//extends
-		btVector3 ca,cb;//extends
-		get_center_extend(ca,ea);
-		box.get_center_extend(cb,eb);
-
+		btVector3 ea, eb;  //extends
+		btVector3 ca, cb;  //extends
+		get_center_extend(ca, ea);
+		box.get_center_extend(cb, eb);
 
 		btVector3 T;
-		btScalar t,t2;
+		btScalar t, t2;
 		int i;
 
 		// Class I : A's basis vectors
-		for(i=0;i<3;i++)
+		for (i = 0; i < 3; i++)
 		{
-			T[i] =  transcache.m_R1to0[i].dot(cb) + transcache.m_T1to0[i] - ca[i];
+			T[i] = transcache.m_R1to0[i].dot(cb) + transcache.m_T1to0[i] - ca[i];
 			t = transcache.m_AR[i].dot(eb) + ea[i];
-			if(GIM_GREATER(T[i], t))	return false;
+			if (GIM_GREATER(T[i], t)) return false;
 		}
 		// Class II : B's basis vectors
-		for(i=0;i<3;i++)
+		for (i = 0; i < 3; i++)
 		{
-			t = MAT_DOT_COL(transcache.m_R1to0,T,i);
-			t2 = MAT_DOT_COL(transcache.m_AR,ea,i) + eb[i];
-			if(GIM_GREATER(t,t2))	return false;
+			t = MAT_DOT_COL(transcache.m_R1to0, T, i);
+			t2 = MAT_DOT_COL(transcache.m_AR, ea, i) + eb[i];
+			if (GIM_GREATER(t, t2)) return false;
 		}
 		// Class III : 9 cross products
-		if(fulltest)
+		if (fulltest)
 		{
-			int j,m,n,o,p,q,r;
-			for(i=0;i<3;i++)
+			int j, m, n, o, p, q, r;
+			for (i = 0; i < 3; i++)
 			{
-				m = (i+1)%3;
-				n = (i+2)%3;
-				o = i==0?1:0;
-				p = i==2?1:2;
-				for(j=0;j<3;j++)
+				m = (i + 1) % 3;
+				n = (i + 2) % 3;
+				o = i == 0 ? 1 : 0;
+				p = i == 2 ? 1 : 2;
+				for (j = 0; j < 3; j++)
 				{
-					q = j==2?1:2;
-					r = j==0?1:0;
-					t = T[n]*transcache.m_R1to0[m][j] - T[m]*transcache.m_R1to0[n][j];
-					t2 = ea[o]*transcache.m_AR[p][j] + ea[p]*transcache.m_AR[o][j] +
-						eb[r]*transcache.m_AR[i][q] + eb[q]*transcache.m_AR[i][r];
-					if(GIM_GREATER(t,t2))	return false;
+					q = j == 2 ? 1 : 2;
+					r = j == 0 ? 1 : 0;
+					t = T[n] * transcache.m_R1to0[m][j] - T[m] * transcache.m_R1to0[n][j];
+					t2 = ea[o] * transcache.m_AR[p][j] + ea[p] * transcache.m_AR[o][j] +
+						 eb[r] * transcache.m_AR[i][q] + eb[q] * transcache.m_AR[i][r];
+					if (GIM_GREATER(t, t2)) return false;
 				}
 			}
 		}
@@ -520,7 +508,7 @@ public:
 
 	//! Simple test for planes.
 	SIMD_FORCE_INLINE bool collide_plane(
-		const btVector4 & plane)
+		const btVector4 &plane)
 	{
 		ePLANE_INTERSECTION_TYPE classify = plane_classify(plane);
 		return (classify == G_COLLIDE_PLANE);
@@ -528,15 +516,15 @@ public:
 
 	//! test for a triangle, with edges
 	SIMD_FORCE_INLINE bool collide_triangle_exact(
-		const btVector3 & p1,
-		const btVector3 & p2,
-		const btVector3 & p3,
-		const btVector4 & triangle_plane)
+		const btVector3 &p1,
+		const btVector3 &p2,
+		const btVector3 &p3,
+		const btVector4 &triangle_plane)
 	{
-		if(!collide_plane(triangle_plane)) return false;
+		if (!collide_plane(triangle_plane)) return false;
 
-		btVector3 center,extends;
-		this->get_center_extend(center,extends);
+		btVector3 center, extends;
+		this->get_center_extend(center, extends);
 
 		const btVector3 v1(p1 - center);
 		const btVector3 v2(p2 - center);
@@ -546,30 +534,29 @@ public:
 		btVector3 diff(v2 - v1);
 		btVector3 abs_diff = diff.absolute();
 		//Test With X axis
-		TEST_CROSS_EDGE_BOX_X_AXIS_MCR(diff,abs_diff,v1,v3,extends);
+		TEST_CROSS_EDGE_BOX_X_AXIS_MCR(diff, abs_diff, v1, v3, extends);
 		//Test With Y axis
-		TEST_CROSS_EDGE_BOX_Y_AXIS_MCR(diff,abs_diff,v1,v3,extends);
+		TEST_CROSS_EDGE_BOX_Y_AXIS_MCR(diff, abs_diff, v1, v3, extends);
 		//Test With Z axis
-		TEST_CROSS_EDGE_BOX_Z_AXIS_MCR(diff,abs_diff,v1,v3,extends);
-
+		TEST_CROSS_EDGE_BOX_Z_AXIS_MCR(diff, abs_diff, v1, v3, extends);
 
 		diff = v3 - v2;
 		abs_diff = diff.absolute();
 		//Test With X axis
-		TEST_CROSS_EDGE_BOX_X_AXIS_MCR(diff,abs_diff,v2,v1,extends);
+		TEST_CROSS_EDGE_BOX_X_AXIS_MCR(diff, abs_diff, v2, v1, extends);
 		//Test With Y axis
-		TEST_CROSS_EDGE_BOX_Y_AXIS_MCR(diff,abs_diff,v2,v1,extends);
+		TEST_CROSS_EDGE_BOX_Y_AXIS_MCR(diff, abs_diff, v2, v1, extends);
 		//Test With Z axis
-		TEST_CROSS_EDGE_BOX_Z_AXIS_MCR(diff,abs_diff,v2,v1,extends);
+		TEST_CROSS_EDGE_BOX_Z_AXIS_MCR(diff, abs_diff, v2, v1, extends);
 
 		diff = v1 - v3;
 		abs_diff = diff.absolute();
 		//Test With X axis
-		TEST_CROSS_EDGE_BOX_X_AXIS_MCR(diff,abs_diff,v3,v2,extends);
+		TEST_CROSS_EDGE_BOX_X_AXIS_MCR(diff, abs_diff, v3, v2, extends);
 		//Test With Y axis
-		TEST_CROSS_EDGE_BOX_Y_AXIS_MCR(diff,abs_diff,v3,v2,extends);
+		TEST_CROSS_EDGE_BOX_Y_AXIS_MCR(diff, abs_diff, v3, v2, extends);
 		//Test With Z axis
-		TEST_CROSS_EDGE_BOX_Z_AXIS_MCR(diff,abs_diff,v3,v2,extends);
+		TEST_CROSS_EDGE_BOX_Z_AXIS_MCR(diff, abs_diff, v3, v2, extends);
 
 		return true;
 	}
@@ -577,17 +564,15 @@ public:
 
 #ifndef BT_BOX_COLLISION_H_INCLUDED
 //! Compairison of transformation objects
-SIMD_FORCE_INLINE bool btCompareTransformsEqual(const btTransform & t1,const btTransform & t2)
+SIMD_FORCE_INLINE bool btCompareTransformsEqual(const btTransform &t1, const btTransform &t2)
 {
-	if(!(t1.getOrigin() == t2.getOrigin()) ) return false;
+	if (!(t1.getOrigin() == t2.getOrigin())) return false;
 
-	if(!(t1.getBasis().getRow(0) == t2.getBasis().getRow(0)) ) return false;
-	if(!(t1.getBasis().getRow(1) == t2.getBasis().getRow(1)) ) return false;
-	if(!(t1.getBasis().getRow(2) == t2.getBasis().getRow(2)) ) return false;
+	if (!(t1.getBasis().getRow(0) == t2.getBasis().getRow(0))) return false;
+	if (!(t1.getBasis().getRow(1) == t2.getBasis().getRow(1))) return false;
+	if (!(t1.getBasis().getRow(2) == t2.getBasis().getRow(2))) return false;
 	return true;
 }
 #endif
 
-
-
-#endif // GIM_BOX_COLLISION_H_INCLUDED
+#endif  // GIM_BOX_COLLISION_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/gim_box_set.cpp b/thirdparty/bullet/BulletCollision/Gimpact/gim_box_set.cpp
index 0c3d7ba8db0..0c7a6b7fc1b 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/gim_box_set.cpp
+++ b/thirdparty/bullet/BulletCollision/Gimpact/gim_box_set.cpp
@@ -28,67 +28,64 @@ email: projectileman@yahoo.com
 -----------------------------------------------------------------------------
 */
 
-
 #include "gim_box_set.h"
 
-
 GUINT GIM_BOX_TREE::_calc_splitting_axis(
-	gim_array<GIM_AABB_DATA> & primitive_boxes, GUINT startIndex,  GUINT endIndex)
+	gim_array<GIM_AABB_DATA>& primitive_boxes, GUINT startIndex, GUINT endIndex)
 {
 	GUINT i;
 
-	btVector3 means(btScalar(0.),btScalar(0.),btScalar(0.));
-	btVector3 variance(btScalar(0.),btScalar(0.),btScalar(0.));
-	GUINT numIndices = endIndex-startIndex;
+	btVector3 means(btScalar(0.), btScalar(0.), btScalar(0.));
+	btVector3 variance(btScalar(0.), btScalar(0.), btScalar(0.));
+	GUINT numIndices = endIndex - startIndex;
 
-	for (i=startIndex;i<endIndex;i++)
+	for (i = startIndex; i < endIndex; i++)
 	{
-		btVector3 center = btScalar(0.5)*(primitive_boxes[i].m_bound.m_max +
-					 primitive_boxes[i].m_bound.m_min);
-		means+=center;
+		btVector3 center = btScalar(0.5) * (primitive_boxes[i].m_bound.m_max +
+											primitive_boxes[i].m_bound.m_min);
+		means += center;
 	}
-	means *= (btScalar(1.)/(btScalar)numIndices);
+	means *= (btScalar(1.) / (btScalar)numIndices);
 
-	for (i=startIndex;i<endIndex;i++)
+	for (i = startIndex; i < endIndex; i++)
 	{
-		btVector3 center = btScalar(0.5)*(primitive_boxes[i].m_bound.m_max +
-					 primitive_boxes[i].m_bound.m_min);
-		btVector3 diff2 = center-means;
+		btVector3 center = btScalar(0.5) * (primitive_boxes[i].m_bound.m_max +
+											primitive_boxes[i].m_bound.m_min);
+		btVector3 diff2 = center - means;
 		diff2 = diff2 * diff2;
 		variance += diff2;
 	}
-	variance *= (btScalar(1.)/	((btScalar)numIndices-1)	);
+	variance *= (btScalar(1.) / ((btScalar)numIndices - 1));
 
 	return variance.maxAxis();
 }
 
-
 GUINT GIM_BOX_TREE::_sort_and_calc_splitting_index(
-	gim_array<GIM_AABB_DATA> & primitive_boxes, GUINT startIndex,
+	gim_array<GIM_AABB_DATA>& primitive_boxes, GUINT startIndex,
 	GUINT endIndex, GUINT splitAxis)
 {
 	GUINT i;
-	GUINT splitIndex =startIndex;
+	GUINT splitIndex = startIndex;
 	GUINT numIndices = endIndex - startIndex;
 
 	// average of centers
 	btScalar splitValue = 0.0f;
-	for (i=startIndex;i<endIndex;i++)
+	for (i = startIndex; i < endIndex; i++)
 	{
-		splitValue+= 0.5f*(primitive_boxes[i].m_bound.m_max[splitAxis] +
-					 primitive_boxes[i].m_bound.m_min[splitAxis]);
+		splitValue += 0.5f * (primitive_boxes[i].m_bound.m_max[splitAxis] +
+							  primitive_boxes[i].m_bound.m_min[splitAxis]);
 	}
 	splitValue /= (btScalar)numIndices;
 
 	//sort leafNodes so all values larger then splitValue comes first, and smaller values start from 'splitIndex'.
-	for (i=startIndex;i<endIndex;i++)
+	for (i = startIndex; i < endIndex; i++)
 	{
-		btScalar center = 0.5f*(primitive_boxes[i].m_bound.m_max[splitAxis] +
-					 primitive_boxes[i].m_bound.m_min[splitAxis]);
+		btScalar center = 0.5f * (primitive_boxes[i].m_bound.m_max[splitAxis] +
+								  primitive_boxes[i].m_bound.m_min[splitAxis]);
 		if (center > splitValue)
 		{
 			//swap
-			primitive_boxes.swap(i,splitIndex);
+			primitive_boxes.swap(i, splitIndex);
 			splitIndex++;
 		}
 	}
@@ -102,28 +99,27 @@ GUINT GIM_BOX_TREE::_sort_and_calc_splitting_index(
 	//bool unbalanced2 = true;
 
 	//this should be safe too:
-	GUINT rangeBalancedIndices = numIndices/3;
-	bool unbalanced = ((splitIndex<=(startIndex+rangeBalancedIndices)) || (splitIndex >=(endIndex-1-rangeBalancedIndices)));
+	GUINT rangeBalancedIndices = numIndices / 3;
+	bool unbalanced = ((splitIndex <= (startIndex + rangeBalancedIndices)) || (splitIndex >= (endIndex - 1 - rangeBalancedIndices)));
 
 	if (unbalanced)
 	{
-		splitIndex = startIndex+ (numIndices>>1);
+		splitIndex = startIndex + (numIndices >> 1);
 	}
 
-	btAssert(!((splitIndex==startIndex) || (splitIndex == (endIndex))));
+	btAssert(!((splitIndex == startIndex) || (splitIndex == (endIndex))));
 
 	return splitIndex;
 }
 
-
-void GIM_BOX_TREE::_build_sub_tree(gim_array<GIM_AABB_DATA> & primitive_boxes, GUINT startIndex,  GUINT endIndex)
+void GIM_BOX_TREE::_build_sub_tree(gim_array<GIM_AABB_DATA>& primitive_boxes, GUINT startIndex, GUINT endIndex)
 {
 	GUINT current_index = m_num_nodes++;
 
-	btAssert((endIndex-startIndex)>0);
+	btAssert((endIndex - startIndex) > 0);
 
-	if((endIndex-startIndex) == 1) //we got a leaf
-	{		
+	if ((endIndex - startIndex) == 1)  //we got a leaf
+	{
 		m_node_array[current_index].m_left = 0;
 		m_node_array[current_index].m_right = 0;
 		m_node_array[current_index].m_escapeIndex = 0;
@@ -138,8 +134,8 @@ void GIM_BOX_TREE::_build_sub_tree(gim_array<GIM_AABB_DATA> & primitive_boxes, G
 	GUINT splitIndex;
 
 	//calc this node bounding box
-	m_node_array[current_index].m_bound.invalidate();	
-	for (splitIndex=startIndex;splitIndex<endIndex;splitIndex++)
+	m_node_array[current_index].m_bound.invalidate();
+	for (splitIndex = startIndex; splitIndex < endIndex; splitIndex++)
 	{
 		m_node_array[current_index].m_bound.merge(primitive_boxes[splitIndex].m_bound);
 	}
@@ -147,36 +143,34 @@ void GIM_BOX_TREE::_build_sub_tree(gim_array<GIM_AABB_DATA> & primitive_boxes, G
 	//calculate Best Splitting Axis and where to split it. Sort the incoming 'leafNodes' array within range 'startIndex/endIndex'.
 
 	//split axis
-	splitIndex = _calc_splitting_axis(primitive_boxes,startIndex,endIndex);
+	splitIndex = _calc_splitting_axis(primitive_boxes, startIndex, endIndex);
 
 	splitIndex = _sort_and_calc_splitting_index(
-			primitive_boxes,startIndex,endIndex,splitIndex);
+		primitive_boxes, startIndex, endIndex, splitIndex);
 
 	//configure this inner node : the left node index
 	m_node_array[current_index].m_left = m_num_nodes;
 	//build left child tree
-	_build_sub_tree(primitive_boxes, startIndex, splitIndex );
+	_build_sub_tree(primitive_boxes, startIndex, splitIndex);
 
 	//configure this inner node : the right node index
 	m_node_array[current_index].m_right = m_num_nodes;
 
 	//build right child tree
-	_build_sub_tree(primitive_boxes, splitIndex ,endIndex);
+	_build_sub_tree(primitive_boxes, splitIndex, endIndex);
 
 	//configure this inner node : the escape index
-	m_node_array[current_index].m_escapeIndex  = m_num_nodes - current_index;
+	m_node_array[current_index].m_escapeIndex = m_num_nodes - current_index;
 }
 
 //! stackless build tree
 void GIM_BOX_TREE::build_tree(
-	gim_array<GIM_AABB_DATA> & primitive_boxes)
+	gim_array<GIM_AABB_DATA>& primitive_boxes)
 {
 	// initialize node count to 0
 	m_num_nodes = 0;
 	// allocate nodes
-	m_node_array.resize(primitive_boxes.size()*2);
-	
+	m_node_array.resize(primitive_boxes.size() * 2);
+
 	_build_sub_tree(primitive_boxes, 0, primitive_boxes.size());
 }
-
-
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/gim_box_set.h b/thirdparty/bullet/BulletCollision/Gimpact/gim_box_set.h
index 61d190a7dfe..0522007e4fa 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/gim_box_set.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/gim_box_set.h
@@ -33,54 +33,51 @@ email: projectileman@yahoo.com
 -----------------------------------------------------------------------------
 */
 
-
 #include "gim_array.h"
 #include "gim_radixsort.h"
 #include "gim_box_collision.h"
 #include "gim_tri_collision.h"
 
-
-
 //! Overlapping pair
 struct GIM_PAIR
 {
-    GUINT m_index1;
-    GUINT m_index2;
-    GIM_PAIR()
-    {}
+	GUINT m_index1;
+	GUINT m_index2;
+	GIM_PAIR()
+	{
+	}
 
-    GIM_PAIR(const GIM_PAIR & p)
-    {
-    	m_index1 = p.m_index1;
-    	m_index2 = p.m_index2;
+	GIM_PAIR(const GIM_PAIR& p)
+	{
+		m_index1 = p.m_index1;
+		m_index2 = p.m_index2;
 	}
 
 	GIM_PAIR(GUINT index1, GUINT index2)
-    {
-    	m_index1 = index1;
-    	m_index2 = index2;
+	{
+		m_index1 = index1;
+		m_index2 = index2;
 	}
 };
 
 //! A pairset array
-class gim_pair_set: public gim_array<GIM_PAIR>
+class gim_pair_set : public gim_array<GIM_PAIR>
 {
 public:
-	gim_pair_set():gim_array<GIM_PAIR>(32)
+	gim_pair_set() : gim_array<GIM_PAIR>(32)
 	{
 	}
-	inline void push_pair(GUINT index1,GUINT index2)
+	inline void push_pair(GUINT index1, GUINT index2)
 	{
-		push_back(GIM_PAIR(index1,index2));
+		push_back(GIM_PAIR(index1, index2));
 	}
 
-	inline void push_pair_inv(GUINT index1,GUINT index2)
+	inline void push_pair_inv(GUINT index1, GUINT index2)
 	{
-		push_back(GIM_PAIR(index2,index1));
+		push_back(GIM_PAIR(index2, index1));
 	}
 };
 
-
 //! Prototype Base class for primitive classification
 /*!
 This class is a wrapper for primitive collections.
@@ -90,16 +87,14 @@ This class can manage Compound shapes and trimeshes, and if it is managing trime
 class GIM_PRIMITIVE_MANAGER_PROTOTYPE
 {
 public:
-
 	virtual ~GIM_PRIMITIVE_MANAGER_PROTOTYPE() {}
 	//! determines if this manager consist on only triangles, which special case will be optimized
 	virtual bool is_trimesh() = 0;
 	virtual GUINT get_primitive_count() = 0;
-	virtual void get_primitive_box(GUINT prim_index ,GIM_AABB & primbox) = 0;
-	virtual void get_primitive_triangle(GUINT prim_index,GIM_TRIANGLE & triangle) = 0;
+	virtual void get_primitive_box(GUINT prim_index, GIM_AABB& primbox) = 0;
+	virtual void get_primitive_triangle(GUINT prim_index, GIM_TRIANGLE& triangle) = 0;
 };
 
-
 struct GIM_AABB_DATA
 {
 	GIM_AABB m_bound;
@@ -110,22 +105,22 @@ struct GIM_AABB_DATA
 struct GIM_BOX_TREE_NODE
 {
 	GIM_AABB m_bound;
-	GUINT m_left;//!< Left subtree
-	GUINT m_right;//!< Right subtree
-	GUINT m_escapeIndex;//!< Scape index for traversing
-	GUINT m_data;//!< primitive index if apply
+	GUINT m_left;         //!< Left subtree
+	GUINT m_right;        //!< Right subtree
+	GUINT m_escapeIndex;  //!< Scape index for traversing
+	GUINT m_data;         //!< primitive index if apply
 
 	GIM_BOX_TREE_NODE()
 	{
-	    m_left = 0;
-	    m_right = 0;
-	    m_escapeIndex = 0;
-	    m_data = 0;
+		m_left = 0;
+		m_right = 0;
+		m_escapeIndex = 0;
+		m_data = 0;
 	}
 
 	SIMD_FORCE_INLINE bool is_leaf_node() const
 	{
-	    return  (!m_left && !m_right);
+		return (!m_left && !m_right);
 	}
 };
 
@@ -135,14 +130,16 @@ class GIM_BOX_TREE
 protected:
 	GUINT m_num_nodes;
 	gim_array<GIM_BOX_TREE_NODE> m_node_array;
+
 protected:
 	GUINT _sort_and_calc_splitting_index(
-		gim_array<GIM_AABB_DATA> & primitive_boxes,
-		 GUINT startIndex,  GUINT endIndex, GUINT splitAxis);
+		gim_array<GIM_AABB_DATA>& primitive_boxes,
+		GUINT startIndex, GUINT endIndex, GUINT splitAxis);
 
-	GUINT _calc_splitting_axis(gim_array<GIM_AABB_DATA> & primitive_boxes, GUINT startIndex,  GUINT endIndex);
+	GUINT _calc_splitting_axis(gim_array<GIM_AABB_DATA>& primitive_boxes, GUINT startIndex, GUINT endIndex);
+
+	void _build_sub_tree(gim_array<GIM_AABB_DATA>& primitive_boxes, GUINT startIndex, GUINT endIndex);
 
-	void _build_sub_tree(gim_array<GIM_AABB_DATA> & primitive_boxes, GUINT startIndex,  GUINT endIndex);
 public:
 	GIM_BOX_TREE()
 	{
@@ -151,7 +148,7 @@ public:
 
 	//! prototype functions for box tree management
 	//!@{
-	void build_tree(gim_array<GIM_AABB_DATA> & primitive_boxes);
+	void build_tree(gim_array<GIM_AABB_DATA>& primitive_boxes);
 
 	SIMD_FORCE_INLINE void clearNodes()
 	{
@@ -176,22 +173,22 @@ public:
 		return m_node_array[nodeindex].m_data;
 	}
 
-	SIMD_FORCE_INLINE void getNodeBound(GUINT nodeindex, GIM_AABB & bound) const
+	SIMD_FORCE_INLINE void getNodeBound(GUINT nodeindex, GIM_AABB& bound) const
 	{
 		bound = m_node_array[nodeindex].m_bound;
 	}
 
-	SIMD_FORCE_INLINE void setNodeBound(GUINT nodeindex, const GIM_AABB & bound)
+	SIMD_FORCE_INLINE void setNodeBound(GUINT nodeindex, const GIM_AABB& bound)
 	{
 		m_node_array[nodeindex].m_bound = bound;
 	}
 
-	SIMD_FORCE_INLINE GUINT getLeftNodeIndex(GUINT nodeindex)  const
+	SIMD_FORCE_INLINE GUINT getLeftNodeIndex(GUINT nodeindex) const
 	{
 		return m_node_array[nodeindex].m_left;
 	}
 
-	SIMD_FORCE_INLINE GUINT getRightNodeIndex(GUINT nodeindex)  const
+	SIMD_FORCE_INLINE GUINT getRightNodeIndex(GUINT nodeindex) const
 	{
 		return m_node_array[nodeindex].m_right;
 	}
@@ -204,78 +201,78 @@ public:
 	//!@}
 };
 
-
 //! Generic Box Tree Template
 /*!
 This class offers an structure for managing a box tree of primitives.
 Requires a Primitive prototype (like GIM_PRIMITIVE_MANAGER_PROTOTYPE ) and
 a Box tree structure ( like GIM_BOX_TREE).
 */
-template<typename _GIM_PRIMITIVE_MANAGER_PROTOTYPE, typename _GIM_BOX_TREE_PROTOTYPE>
+template <typename _GIM_PRIMITIVE_MANAGER_PROTOTYPE, typename _GIM_BOX_TREE_PROTOTYPE>
 class GIM_BOX_TREE_TEMPLATE_SET
 {
 protected:
 	_GIM_PRIMITIVE_MANAGER_PROTOTYPE m_primitive_manager;
 	_GIM_BOX_TREE_PROTOTYPE m_box_tree;
+
 protected:
 	//stackless refit
 	SIMD_FORCE_INLINE void refit()
 	{
 		GUINT nodecount = getNodeCount();
-		while(nodecount--)
+		while (nodecount--)
 		{
-			if(isLeafNode(nodecount))
+			if (isLeafNode(nodecount))
 			{
 				GIM_AABB leafbox;
-				m_primitive_manager.get_primitive_box(getNodeData(nodecount),leafbox);
-				setNodeBound(nodecount,leafbox);
+				m_primitive_manager.get_primitive_box(getNodeData(nodecount), leafbox);
+				setNodeBound(nodecount, leafbox);
 			}
 			else
 			{
 				//get left bound
 				GUINT childindex = getLeftNodeIndex(nodecount);
 				GIM_AABB bound;
-				getNodeBound(childindex,bound);
+				getNodeBound(childindex, bound);
 				//get right bound
 				childindex = getRightNodeIndex(nodecount);
 				GIM_AABB bound2;
-				getNodeBound(childindex,bound2);
+				getNodeBound(childindex, bound2);
 				bound.merge(bound2);
 
-				setNodeBound(nodecount,bound);
+				setNodeBound(nodecount, bound);
 			}
 		}
 	}
-public:
 
+public:
 	GIM_BOX_TREE_TEMPLATE_SET()
 	{
 	}
 
-	SIMD_FORCE_INLINE GIM_AABB getGlobalBox()  const
+	SIMD_FORCE_INLINE GIM_AABB getGlobalBox() const
 	{
 		GIM_AABB totalbox;
 		getNodeBound(0, totalbox);
 		return totalbox;
 	}
 
-	SIMD_FORCE_INLINE void setPrimitiveManager(const _GIM_PRIMITIVE_MANAGER_PROTOTYPE & primitive_manager)
+	SIMD_FORCE_INLINE void setPrimitiveManager(const _GIM_PRIMITIVE_MANAGER_PROTOTYPE& primitive_manager)
 	{
 		m_primitive_manager = primitive_manager;
 	}
 
-	const _GIM_PRIMITIVE_MANAGER_PROTOTYPE & getPrimitiveManager() const
+	const _GIM_PRIMITIVE_MANAGER_PROTOTYPE& getPrimitiveManager() const
 	{
 		return m_primitive_manager;
 	}
 
-	_GIM_PRIMITIVE_MANAGER_PROTOTYPE & getPrimitiveManager()
+	_GIM_PRIMITIVE_MANAGER_PROTOTYPE& getPrimitiveManager()
 	{
 		return m_primitive_manager;
 	}
 
-//! node manager prototype functions
-///@{
+	//! node manager prototype functions
+	///@{
 
 	//! this attemps to refit the box set.
 	SIMD_FORCE_INLINE void update()
@@ -288,19 +285,19 @@ public:
 	{
 		//obtain primitive boxes
 		gim_array<GIM_AABB_DATA> primitive_boxes;
-		primitive_boxes.resize(m_primitive_manager.get_primitive_count(),false);
+		primitive_boxes.resize(m_primitive_manager.get_primitive_count(), false);
 
-		for (GUINT i = 0;i<primitive_boxes.size() ;i++ )
+		for (GUINT i = 0; i < primitive_boxes.size(); i++)
 		{
-			 m_primitive_manager.get_primitive_box(i,primitive_boxes[i].m_bound);
-			 primitive_boxes[i].m_data = i;
+			m_primitive_manager.get_primitive_box(i, primitive_boxes[i].m_bound);
+			primitive_boxes[i].m_data = i;
 		}
 
 		m_box_tree.build_tree(primitive_boxes);
 	}
 
 	//! returns the indices of the primitives in the m_primitive_manager
-	SIMD_FORCE_INLINE bool boxQuery(const GIM_AABB & box, gim_array<GUINT> & collided_results) const
+	SIMD_FORCE_INLINE bool boxQuery(const GIM_AABB& box, gim_array<GUINT>& collided_results) const
 	{
 		GUINT curIndex = 0;
 		GUINT numNodes = getNodeCount();
@@ -308,7 +305,7 @@ public:
 		while (curIndex < numNodes)
 		{
 			GIM_AABB bound;
-			getNodeBound(curIndex,bound);
+			getNodeBound(curIndex, bound);
 
 			//catch bugs in tree data
 
@@ -328,26 +325,26 @@ public:
 			else
 			{
 				//skip node
-				curIndex+= getScapeNodeIndex(curIndex);
+				curIndex += getScapeNodeIndex(curIndex);
 			}
 		}
-		if(collided_results.size()>0) return true;
+		if (collided_results.size() > 0) return true;
 		return false;
 	}
 
 	//! returns the indices of the primitives in the m_primitive_manager
-	SIMD_FORCE_INLINE bool boxQueryTrans(const GIM_AABB & box,
-		 const btTransform & transform, gim_array<GUINT> & collided_results) const
+	SIMD_FORCE_INLINE bool boxQueryTrans(const GIM_AABB& box,
+										 const btTransform& transform, gim_array<GUINT>& collided_results) const
 	{
-		GIM_AABB transbox=box;
+		GIM_AABB transbox = box;
 		transbox.appy_transform(transform);
-		return boxQuery(transbox,collided_results);
+		return boxQuery(transbox, collided_results);
 	}
 
 	//! returns the indices of the primitives in the m_primitive_manager
 	SIMD_FORCE_INLINE bool rayQuery(
-		const btVector3 & ray_dir,const btVector3 & ray_origin ,
-		gim_array<GUINT> & collided_results) const
+		const btVector3& ray_dir, const btVector3& ray_origin,
+		gim_array<GUINT>& collided_results) const
 	{
 		GUINT curIndex = 0;
 		GUINT numNodes = getNodeCount();
@@ -355,16 +352,16 @@ public:
 		while (curIndex < numNodes)
 		{
 			GIM_AABB bound;
-			getNodeBound(curIndex,bound);
+			getNodeBound(curIndex, bound);
 
 			//catch bugs in tree data
 
-			bool aabbOverlap = bound.collide_ray(ray_origin,ray_dir);
+			bool aabbOverlap = bound.collide_ray(ray_origin, ray_dir);
 			bool isleafnode = isLeafNode(curIndex);
 
 			if (isleafnode && aabbOverlap)
 			{
-				collided_results.push_back(getNodeData( curIndex));
+				collided_results.push_back(getNodeData(curIndex));
 			}
 
 			if (aabbOverlap || isleafnode)
@@ -375,10 +372,10 @@ public:
 			else
 			{
 				//skip node
-				curIndex+= getScapeNodeIndex(curIndex);
+				curIndex += getScapeNodeIndex(curIndex);
 			}
 		}
-		if(collided_results.size()>0) return true;
+		if (collided_results.size() > 0) return true;
 		return false;
 	}
 
@@ -389,7 +386,7 @@ public:
 	}
 
 	//! tells if this set is a trimesh
-	SIMD_FORCE_INLINE bool isTrimesh()  const
+	SIMD_FORCE_INLINE bool isTrimesh() const
 	{
 		return m_primitive_manager.is_trimesh();
 	}
@@ -411,12 +408,12 @@ public:
 		return m_box_tree.getNodeData(nodeindex);
 	}
 
-	SIMD_FORCE_INLINE void getNodeBound(GUINT nodeindex, GIM_AABB & bound)  const
+	SIMD_FORCE_INLINE void getNodeBound(GUINT nodeindex, GIM_AABB& bound) const
 	{
 		m_box_tree.getNodeBound(nodeindex, bound);
 	}
 
-	SIMD_FORCE_INLINE void setNodeBound(GUINT nodeindex, const GIM_AABB & bound)
+	SIMD_FORCE_INLINE void setNodeBound(GUINT nodeindex, const GIM_AABB& bound)
 	{
 		m_box_tree.setNodeBound(nodeindex, bound);
 	}
@@ -436,36 +433,30 @@ public:
 		return m_box_tree.getScapeNodeIndex(nodeindex);
 	}
 
-	SIMD_FORCE_INLINE void getNodeTriangle(GUINT nodeindex,GIM_TRIANGLE & triangle) const
+	SIMD_FORCE_INLINE void getNodeTriangle(GUINT nodeindex, GIM_TRIANGLE& triangle) const
 	{
-		m_primitive_manager.get_primitive_triangle(getNodeData(nodeindex),triangle);
+		m_primitive_manager.get_primitive_triangle(getNodeData(nodeindex), triangle);
 	}
-
 };
 
 //! Class for Box Tree Sets
 /*!
 this has the GIM_BOX_TREE implementation for bounding boxes.
 */
-template<typename _GIM_PRIMITIVE_MANAGER_PROTOTYPE>
-class GIM_BOX_TREE_SET: public GIM_BOX_TREE_TEMPLATE_SET< _GIM_PRIMITIVE_MANAGER_PROTOTYPE, GIM_BOX_TREE>
+template <typename _GIM_PRIMITIVE_MANAGER_PROTOTYPE>
+class GIM_BOX_TREE_SET : public GIM_BOX_TREE_TEMPLATE_SET<_GIM_PRIMITIVE_MANAGER_PROTOTYPE, GIM_BOX_TREE>
 {
 public:
-
 };
 
-
-
-
-
 /// GIM_BOX_SET collision methods
-template<typename BOX_SET_CLASS0,typename BOX_SET_CLASS1>
+template <typename BOX_SET_CLASS0, typename BOX_SET_CLASS1>
 class GIM_TREE_TREE_COLLIDER
 {
 public:
-	gim_pair_set * m_collision_pairs;
-	BOX_SET_CLASS0 * m_boxset0;
-	BOX_SET_CLASS1 * m_boxset1;
+	gim_pair_set* m_collision_pairs;
+	BOX_SET_CLASS0* m_boxset0;
+	BOX_SET_CLASS1* m_boxset1;
 	GUINT current_node0;
 	GUINT current_node1;
 	bool node0_is_leaf;
@@ -483,18 +474,18 @@ public:
 	GIM_TRIANGLE m_tri1;
 	btVector4 m_tri1_plane;
 
-
 public:
 	GIM_TREE_TREE_COLLIDER()
 	{
 		current_node0 = G_UINT_INFINITY;
 		current_node1 = G_UINT_INFINITY;
 	}
+
 protected:
 	SIMD_FORCE_INLINE void retrieve_node0_triangle(GUINT node0)
 	{
-		if(node0_has_triangle) return;
-		m_boxset0->getNodeTriangle(node0,m_tri0);
+		if (node0_has_triangle) return;
+		m_boxset0->getNodeTriangle(node0, m_tri0);
 		//transform triangle
 		m_tri0.m_vertices[0] = trans_cache_0to1(m_tri0.m_vertices[0]);
 		m_tri0.m_vertices[1] = trans_cache_0to1(m_tri0.m_vertices[1]);
@@ -506,8 +497,8 @@ protected:
 
 	SIMD_FORCE_INLINE void retrieve_node1_triangle(GUINT node1)
 	{
-		if(node1_has_triangle) return;
-		m_boxset1->getNodeTriangle(node1,m_tri1);
+		if (node1_has_triangle) return;
+		m_boxset1->getNodeTriangle(node1, m_tri1);
 		//transform triangle
 		m_tri1.m_vertices[0] = trans_cache_1to0.transform(m_tri1.m_vertices[0]);
 		m_tri1.m_vertices[1] = trans_cache_1to0.transform(m_tri1.m_vertices[1]);
@@ -519,8 +510,8 @@ protected:
 
 	SIMD_FORCE_INLINE void retrieve_node0_info(GUINT node0)
 	{
-		if(node0 == current_node0) return;
-		m_boxset0->getNodeBound(node0,m_box0);
+		if (node0 == current_node0) return;
+		m_boxset0->getNodeBound(node0, m_box0);
 		node0_is_leaf = m_boxset0->isLeafNode(node0);
 		node0_has_triangle = false;
 		current_node0 = node0;
@@ -528,21 +519,21 @@ protected:
 
 	SIMD_FORCE_INLINE void retrieve_node1_info(GUINT node1)
 	{
-		if(node1 == current_node1) return;
-		m_boxset1->getNodeBound(node1,m_box1);
+		if (node1 == current_node1) return;
+		m_boxset1->getNodeBound(node1, m_box1);
 		node1_is_leaf = m_boxset1->isLeafNode(node1);
 		node1_has_triangle = false;
 		current_node1 = node1;
 	}
 
-	SIMD_FORCE_INLINE bool node_collision(GUINT node0 ,GUINT node1)
+	SIMD_FORCE_INLINE bool node_collision(GUINT node0, GUINT node1)
 	{
 		retrieve_node0_info(node0);
 		retrieve_node1_info(node1);
-		bool result = m_box0.overlapping_trans_cache(m_box1,trans_cache_1to0,true);
-		if(!result) return false;
+		bool result = m_box0.overlapping_trans_cache(m_box1, trans_cache_1to0, true);
+		if (!result) return false;
 
-		if(t0_is_trimesh && node0_is_leaf)
+		if (t0_is_trimesh && node0_is_leaf)
 		{
 			//perform primitive vs box collision
 			retrieve_node0_triangle(node0);
@@ -550,14 +541,14 @@ protected:
 			m_box1.increment_margin(m_tri0.m_margin);
 
 			result = m_box1.collide_triangle_exact(
-				m_tri0.m_vertices[0],m_tri0.m_vertices[1],m_tri0.m_vertices[2],m_tri0_plane);
+				m_tri0.m_vertices[0], m_tri0.m_vertices[1], m_tri0.m_vertices[2], m_tri0_plane);
 
 			m_box1.increment_margin(-m_tri0.m_margin);
 
-			if(!result) return false;
+			if (!result) return false;
 			return true;
 		}
-		else if(t1_is_trimesh && node1_is_leaf)
+		else if (t1_is_trimesh && node1_is_leaf)
 		{
 			//perform primitive vs box collision
 			retrieve_node1_triangle(node1);
@@ -565,11 +556,11 @@ protected:
 			m_box0.increment_margin(m_tri1.m_margin);
 
 			result = m_box0.collide_triangle_exact(
-				m_tri1.m_vertices[0],m_tri1.m_vertices[1],m_tri1.m_vertices[2],m_tri1_plane);
+				m_tri1.m_vertices[0], m_tri1.m_vertices[1], m_tri1.m_vertices[2], m_tri1_plane);
 
 			m_box0.increment_margin(-m_tri1.m_margin);
 
-			if(!result) return false;
+			if (!result) return false;
 			return true;
 		}
 		return true;
@@ -582,40 +573,39 @@ protected:
 		stack_collisions.reserve(32);
 
 		//add the first pair
-		stack_collisions.push_pair(0,0);
+		stack_collisions.push_pair(0, 0);
 
-
-		while(stack_collisions.size())
+		while (stack_collisions.size())
 		{
 			//retrieve the last pair and pop
 			GUINT node0 = stack_collisions.back().m_index1;
 			GUINT node1 = stack_collisions.back().m_index2;
 			stack_collisions.pop_back();
-			if(node_collision(node0,node1)) // a collision is found
+			if (node_collision(node0, node1))  // a collision is found
 			{
-				if(node0_is_leaf)
+				if (node0_is_leaf)
 				{
-					if(node1_is_leaf)
+					if (node1_is_leaf)
 					{
-						m_collision_pairs->push_pair(m_boxset0->getNodeData(node0),m_boxset1->getNodeData(node1));
+						m_collision_pairs->push_pair(m_boxset0->getNodeData(node0), m_boxset1->getNodeData(node1));
 					}
 					else
 					{
 						//collide left
-						stack_collisions.push_pair(node0,m_boxset1->getLeftNodeIndex(node1));
+						stack_collisions.push_pair(node0, m_boxset1->getLeftNodeIndex(node1));
 
 						//collide right
-						stack_collisions.push_pair(node0,m_boxset1->getRightNodeIndex(node1));
+						stack_collisions.push_pair(node0, m_boxset1->getRightNodeIndex(node1));
 					}
 				}
 				else
 				{
-					if(node1_is_leaf)
+					if (node1_is_leaf)
 					{
 						//collide left
-						stack_collisions.push_pair(m_boxset0->getLeftNodeIndex(node0),node1);
+						stack_collisions.push_pair(m_boxset0->getLeftNodeIndex(node0), node1);
 						//collide right
-						stack_collisions.push_pair(m_boxset0->getRightNodeIndex(node0),node1);
+						stack_collisions.push_pair(m_boxset0->getRightNodeIndex(node0), node1);
 					}
 					else
 					{
@@ -624,36 +614,36 @@ protected:
 						GUINT left1 = m_boxset1->getLeftNodeIndex(node1);
 						GUINT right1 = m_boxset1->getRightNodeIndex(node1);
 						//collide left
-						stack_collisions.push_pair(left0,left1);
+						stack_collisions.push_pair(left0, left1);
 						//collide right
-						stack_collisions.push_pair(left0,right1);
+						stack_collisions.push_pair(left0, right1);
 						//collide left
-						stack_collisions.push_pair(right0,left1);
+						stack_collisions.push_pair(right0, left1);
 						//collide right
-						stack_collisions.push_pair(right0,right1);
+						stack_collisions.push_pair(right0, right1);
 
-					}// else if node1 is not a leaf
-				}// else if node0 is not a leaf
+					}  // else if node1 is not a leaf
+				}      // else if node0 is not a leaf
 
-			}// if(node_collision(node0,node1))
-		}//while(stack_collisions.size())
+			}  // if(node_collision(node0,node1))
+		}      //while(stack_collisions.size())
 	}
+
 public:
-	void find_collision(BOX_SET_CLASS0 * boxset1, const btTransform & trans1,
-		BOX_SET_CLASS1 * boxset2, const btTransform & trans2,
-		gim_pair_set & collision_pairs, bool complete_primitive_tests = true)
+	void find_collision(BOX_SET_CLASS0* boxset1, const btTransform& trans1,
+						BOX_SET_CLASS1* boxset2, const btTransform& trans2,
+						gim_pair_set& collision_pairs, bool complete_primitive_tests = true)
 	{
 		m_collision_pairs = &collision_pairs;
 		m_boxset0 = boxset1;
 		m_boxset1 = boxset2;
 
-		trans_cache_1to0.calc_from_homogenic(trans1,trans2);
+		trans_cache_1to0.calc_from_homogenic(trans1, trans2);
 
-		trans_cache_0to1 =  trans2.inverse();
+		trans_cache_0to1 = trans2.inverse();
 		trans_cache_0to1 *= trans1;
 
-
-		if(complete_primitive_tests)
+		if (complete_primitive_tests)
 		{
 			t0_is_trimesh = boxset1->getPrimitiveManager().is_trimesh();
 			t1_is_trimesh = boxset2->getPrimitiveManager().is_trimesh();
@@ -668,7 +658,4 @@ public:
 	}
 };
 
-
-#endif // GIM_BOXPRUNING_H_INCLUDED
-
-
+#endif  // GIM_BOXPRUNING_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/gim_clip_polygon.h b/thirdparty/bullet/BulletCollision/Gimpact/gim_clip_polygon.h
index e342459ce56..57b9c5c91ad 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/gim_clip_polygon.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/gim_clip_polygon.h
@@ -33,91 +33,86 @@ email: projectileman@yahoo.com
 -----------------------------------------------------------------------------
 */
 
-
 //! This function calcs the distance from a 3D plane
 class DISTANCE_PLANE_3D_FUNC
 {
 public:
-	template<typename CLASS_POINT,typename CLASS_PLANE>
-	inline GREAL operator()(const CLASS_PLANE & plane, const CLASS_POINT & point)
+	template <typename CLASS_POINT, typename CLASS_PLANE>
+	inline GREAL operator()(const CLASS_PLANE& plane, const CLASS_POINT& point)
 	{
 		return DISTANCE_PLANE_POINT(plane, point);
 	}
 };
 
-
-
-template<typename CLASS_POINT>
+template <typename CLASS_POINT>
 SIMD_FORCE_INLINE void PLANE_CLIP_POLYGON_COLLECT(
-						const CLASS_POINT & point0,
-						const CLASS_POINT & point1,
-						GREAL dist0,
-						GREAL dist1,
-						CLASS_POINT * clipped,
-						GUINT & clipped_count)
+	const CLASS_POINT& point0,
+	const CLASS_POINT& point1,
+	GREAL dist0,
+	GREAL dist1,
+	CLASS_POINT* clipped,
+	GUINT& clipped_count)
 {
-	GUINT _prevclassif = (dist0>G_EPSILON);
-	GUINT _classif = (dist1>G_EPSILON);
-	if(_classif!=_prevclassif)
+	GUINT _prevclassif = (dist0 > G_EPSILON);
+	GUINT _classif = (dist1 > G_EPSILON);
+	if (_classif != _prevclassif)
 	{
-		GREAL blendfactor = -dist0/(dist1-dist0);
-		VEC_BLEND(clipped[clipped_count],point0,point1,blendfactor);
+		GREAL blendfactor = -dist0 / (dist1 - dist0);
+		VEC_BLEND(clipped[clipped_count], point0, point1, blendfactor);
 		clipped_count++;
 	}
-	if(!_classif)
+	if (!_classif)
 	{
-		VEC_COPY(clipped[clipped_count],point1);
+		VEC_COPY(clipped[clipped_count], point1);
 		clipped_count++;
 	}
 }
 
-
 //! Clips a polygon by a plane
 /*!
 *\return The count of the clipped counts
 */
-template<typename CLASS_POINT,typename CLASS_PLANE, typename DISTANCE_PLANE_FUNC>
+template <typename CLASS_POINT, typename CLASS_PLANE, typename DISTANCE_PLANE_FUNC>
 SIMD_FORCE_INLINE GUINT PLANE_CLIP_POLYGON_GENERIC(
-						const CLASS_PLANE & plane,
-						const CLASS_POINT * polygon_points,
-						GUINT polygon_point_count,
-						CLASS_POINT * clipped,DISTANCE_PLANE_FUNC distance_func)
+	const CLASS_PLANE& plane,
+	const CLASS_POINT* polygon_points,
+	GUINT polygon_point_count,
+	CLASS_POINT* clipped, DISTANCE_PLANE_FUNC distance_func)
 {
-    GUINT clipped_count = 0;
+	GUINT clipped_count = 0;
 
-
-    //clip first point
-	GREAL firstdist = distance_func(plane,polygon_points[0]);;
-	if(!(firstdist>G_EPSILON))
+	//clip first point
+	GREAL firstdist = distance_func(plane, polygon_points[0]);
+	;
+	if (!(firstdist > G_EPSILON))
 	{
-		VEC_COPY(clipped[clipped_count],polygon_points[0]);
+		VEC_COPY(clipped[clipped_count], polygon_points[0]);
 		clipped_count++;
 	}
 
 	GREAL olddist = firstdist;
-	for(GUINT _i=1;_i<polygon_point_count;_i++)
-	{		
-		GREAL dist = distance_func(plane,polygon_points[_i]);
+	for (GUINT _i = 1; _i < polygon_point_count; _i++)
+	{
+		GREAL dist = distance_func(plane, polygon_points[_i]);
 
 		PLANE_CLIP_POLYGON_COLLECT(
-						polygon_points[_i-1],polygon_points[_i],
-						olddist,
-						dist,
-						clipped,
-						clipped_count);
+			polygon_points[_i - 1], polygon_points[_i],
+			olddist,
+			dist,
+			clipped,
+			clipped_count);
 
-
-		olddist = dist;		
+		olddist = dist;
 	}
 
-	//RETURN TO FIRST  point	
+	//RETURN TO FIRST  point
 
 	PLANE_CLIP_POLYGON_COLLECT(
-					polygon_points[polygon_point_count-1],polygon_points[0],
-					olddist,
-					firstdist,
-					clipped,
-					clipped_count);
+		polygon_points[polygon_point_count - 1], polygon_points[0],
+		olddist,
+		firstdist,
+		clipped,
+		clipped_count);
 
 	return clipped_count;
 }
@@ -126,85 +121,79 @@ SIMD_FORCE_INLINE GUINT PLANE_CLIP_POLYGON_GENERIC(
 /*!
 *\return The count of the clipped counts
 */
-template<typename CLASS_POINT,typename CLASS_PLANE, typename DISTANCE_PLANE_FUNC>
+template <typename CLASS_POINT, typename CLASS_PLANE, typename DISTANCE_PLANE_FUNC>
 SIMD_FORCE_INLINE GUINT PLANE_CLIP_TRIANGLE_GENERIC(
-						const CLASS_PLANE & plane,
-						const CLASS_POINT & point0,
-						const CLASS_POINT & point1,
-						const CLASS_POINT & point2,
-						CLASS_POINT * clipped,DISTANCE_PLANE_FUNC distance_func)
+	const CLASS_PLANE& plane,
+	const CLASS_POINT& point0,
+	const CLASS_POINT& point1,
+	const CLASS_POINT& point2,
+	CLASS_POINT* clipped, DISTANCE_PLANE_FUNC distance_func)
 {
-    GUINT clipped_count = 0;
+	GUINT clipped_count = 0;
 
-    //clip first point
-	GREAL firstdist = distance_func(plane,point0);;
-	if(!(firstdist>G_EPSILON))
+	//clip first point
+	GREAL firstdist = distance_func(plane, point0);
+	;
+	if (!(firstdist > G_EPSILON))
 	{
-		VEC_COPY(clipped[clipped_count],point0);
+		VEC_COPY(clipped[clipped_count], point0);
 		clipped_count++;
 	}
 
 	// point 1
 	GREAL olddist = firstdist;
-	GREAL dist = distance_func(plane,point1);
+	GREAL dist = distance_func(plane, point1);
 
 	PLANE_CLIP_POLYGON_COLLECT(
-					point0,point1,
-					olddist,
-					dist,
-					clipped,
-					clipped_count);
+		point0, point1,
+		olddist,
+		dist,
+		clipped,
+		clipped_count);
 
 	olddist = dist;
 
-
 	// point 2
-	dist = distance_func(plane,point2);
+	dist = distance_func(plane, point2);
 
 	PLANE_CLIP_POLYGON_COLLECT(
-					point1,point2,
-					olddist,
-					dist,
-					clipped,
-					clipped_count);
+		point1, point2,
+		olddist,
+		dist,
+		clipped,
+		clipped_count);
 	olddist = dist;
 
-
-
 	//RETURN TO FIRST  point
 	PLANE_CLIP_POLYGON_COLLECT(
-					point2,point0,
-					olddist,
-					firstdist,
-					clipped,
-					clipped_count);
+		point2, point0,
+		olddist,
+		firstdist,
+		clipped,
+		clipped_count);
 
 	return clipped_count;
 }
 
-
-template<typename CLASS_POINT,typename CLASS_PLANE>
+template <typename CLASS_POINT, typename CLASS_PLANE>
 SIMD_FORCE_INLINE GUINT PLANE_CLIP_POLYGON3D(
-						const CLASS_PLANE & plane,
-						const CLASS_POINT * polygon_points,
-						GUINT polygon_point_count,
-						CLASS_POINT * clipped)
+	const CLASS_PLANE& plane,
+	const CLASS_POINT* polygon_points,
+	GUINT polygon_point_count,
+	CLASS_POINT* clipped)
 {
-	return PLANE_CLIP_POLYGON_GENERIC<CLASS_POINT,CLASS_PLANE>(plane,polygon_points,polygon_point_count,clipped,DISTANCE_PLANE_3D_FUNC());
+	return PLANE_CLIP_POLYGON_GENERIC<CLASS_POINT, CLASS_PLANE>(plane, polygon_points, polygon_point_count, clipped, DISTANCE_PLANE_3D_FUNC());
 }
 
-
-template<typename CLASS_POINT,typename CLASS_PLANE>
+template <typename CLASS_POINT, typename CLASS_PLANE>
 SIMD_FORCE_INLINE GUINT PLANE_CLIP_TRIANGLE3D(
-						const CLASS_PLANE & plane,
-						const CLASS_POINT & point0,
-						const CLASS_POINT & point1,
-						const CLASS_POINT & point2,
-						CLASS_POINT * clipped)
+	const CLASS_PLANE& plane,
+	const CLASS_POINT& point0,
+	const CLASS_POINT& point1,
+	const CLASS_POINT& point2,
+	CLASS_POINT* clipped)
 {
-	return PLANE_CLIP_TRIANGLE_GENERIC<CLASS_POINT,CLASS_PLANE>(plane,point0,point1,point2,clipped,DISTANCE_PLANE_3D_FUNC());
+	return PLANE_CLIP_TRIANGLE_GENERIC<CLASS_POINT, CLASS_PLANE>(plane, point0, point1, point2, clipped, DISTANCE_PLANE_3D_FUNC());
 }
 
-
-
-#endif // GIM_TRI_COLLISION_H_INCLUDED
+#endif  // GIM_TRI_COLLISION_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/gim_contact.cpp b/thirdparty/bullet/BulletCollision/Gimpact/gim_contact.cpp
index 20e41de089f..390225709e1 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/gim_contact.cpp
+++ b/thirdparty/bullet/BulletCollision/Gimpact/gim_contact.cpp
@@ -33,91 +33,89 @@ email: projectileman@yahoo.com
 #define MAX_COINCIDENT 8
 
 void gim_contact_array::merge_contacts(
-	const gim_contact_array & contacts, bool normal_contact_average)
+	const gim_contact_array& contacts, bool normal_contact_average)
 {
 	clear();
 
-	if(contacts.size()==1)
+	if (contacts.size() == 1)
 	{
 		push_back(contacts.back());
 		return;
 	}
 
 	gim_array<GIM_RSORT_TOKEN> keycontacts(contacts.size());
-	keycontacts.resize(contacts.size(),false);
+	keycontacts.resize(contacts.size(), false);
 
 	//fill key contacts
 
 	GUINT i;
 
-	for (i = 0;i<contacts.size() ;i++ )
+	for (i = 0; i < contacts.size(); i++)
 	{
 		keycontacts[i].m_key = contacts[i].calc_key_contact();
 		keycontacts[i].m_value = i;
 	}
 
 	//sort keys
-	gim_heap_sort(keycontacts.pointer(),keycontacts.size(),GIM_RSORT_TOKEN_COMPARATOR());
+	gim_heap_sort(keycontacts.pointer(), keycontacts.size(), GIM_RSORT_TOKEN_COMPARATOR());
 
 	// Merge contacts
 
-	GUINT coincident_count=0;
+	GUINT coincident_count = 0;
 	btVector3 coincident_normals[MAX_COINCIDENT];
 
 	GUINT last_key = keycontacts[0].m_key;
 	GUINT key = 0;
 
 	push_back(contacts[keycontacts[0].m_value]);
-	GIM_CONTACT * pcontact = &back();
+	GIM_CONTACT* pcontact = &back();
 
-
-
-	for( i=1;i<keycontacts.size();i++)
+	for (i = 1; i < keycontacts.size(); i++)
 	{
-	    key = keycontacts[i].m_key;
-		const GIM_CONTACT * scontact = &contacts[keycontacts[i].m_value];
+		key = keycontacts[i].m_key;
+		const GIM_CONTACT* scontact = &contacts[keycontacts[i].m_value];
 
-		if(last_key ==  key)//same points
+		if (last_key == key)  //same points
 		{
 			//merge contact
-			if(pcontact->m_depth - CONTACT_DIFF_EPSILON > scontact->m_depth)//)
+			if (pcontact->m_depth - CONTACT_DIFF_EPSILON > scontact->m_depth)  //)
 			{
 				*pcontact = *scontact;
-                coincident_count = 0;
+				coincident_count = 0;
 			}
-			else if(normal_contact_average)
+			else if (normal_contact_average)
 			{
-				if(btFabs(pcontact->m_depth - scontact->m_depth)<CONTACT_DIFF_EPSILON)
-                {
-                    if(coincident_count<MAX_COINCIDENT)
-                    {
-                    	coincident_normals[coincident_count] = scontact->m_normal;
-                        coincident_count++;
-                    }
-                }
+				if (btFabs(pcontact->m_depth - scontact->m_depth) < CONTACT_DIFF_EPSILON)
+				{
+					if (coincident_count < MAX_COINCIDENT)
+					{
+						coincident_normals[coincident_count] = scontact->m_normal;
+						coincident_count++;
+					}
+				}
 			}
 		}
 		else
-		{//add new contact
+		{  //add new contact
 
-		    if(normal_contact_average && coincident_count>0)
-		    {
-		    	pcontact->interpolate_normals(coincident_normals,coincident_count);
-		        coincident_count = 0;
-		    }
+			if (normal_contact_average && coincident_count > 0)
+			{
+				pcontact->interpolate_normals(coincident_normals, coincident_count);
+				coincident_count = 0;
+			}
 
-		    push_back(*scontact);
-		    pcontact = &back();
-        }
+			push_back(*scontact);
+			pcontact = &back();
+		}
 		last_key = key;
 	}
 }
 
-void gim_contact_array::merge_contacts_unique(const gim_contact_array & contacts)
+void gim_contact_array::merge_contacts_unique(const gim_contact_array& contacts)
 {
 	clear();
 
-	if(contacts.size()==1)
+	if (contacts.size() == 1)
 	{
 		push_back(contacts.back());
 		return;
@@ -125,14 +123,14 @@ void gim_contact_array::merge_contacts_unique(const gim_contact_array & contacts
 
 	GIM_CONTACT average_contact = contacts.back();
 
-	for (GUINT i=1;i<contacts.size() ;i++ )
+	for (GUINT i = 1; i < contacts.size(); i++)
 	{
 		average_contact.m_point += contacts[i].m_point;
 		average_contact.m_normal += contacts[i].m_normal * contacts[i].m_depth;
 	}
 
 	//divide
-	GREAL divide_average = 1.0f/((GREAL)contacts.size());
+	GREAL divide_average = 1.0f / ((GREAL)contacts.size());
 
 	average_contact.m_point *= divide_average;
 
@@ -141,6 +139,4 @@ void gim_contact_array::merge_contacts_unique(const gim_contact_array & contacts
 	average_contact.m_depth = average_contact.m_normal.length();
 
 	average_contact.m_normal /= average_contact.m_depth;
-
 }
-
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/gim_contact.h b/thirdparty/bullet/BulletCollision/Gimpact/gim_contact.h
index b41c714b5f7..9deb28a26eb 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/gim_contact.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/gim_contact.h
@@ -36,7 +36,6 @@ email: projectileman@yahoo.com
 #include "gim_radixsort.h"
 #include "gim_array.h"
 
-
 /**
 Configuration var for applying interpolation of  contact normals
 */
@@ -58,90 +57,87 @@ Configuration var for applying interpolation of  contact normals
 class GIM_CONTACT
 {
 public:
-    btVector3 m_point;
-    btVector3 m_normal;
-    GREAL m_depth;//Positive value indicates interpenetration
-    GREAL m_distance;//Padding not for use
-    GUINT m_feature1;//Face number
-    GUINT m_feature2;//Face number
+	btVector3 m_point;
+	btVector3 m_normal;
+	GREAL m_depth;     //Positive value indicates interpenetration
+	GREAL m_distance;  //Padding not for use
+	GUINT m_feature1;  //Face number
+	GUINT m_feature2;  //Face number
 public:
-    GIM_CONTACT()
-    {
-    }
+	GIM_CONTACT()
+	{
+	}
 
-    GIM_CONTACT(const GIM_CONTACT & contact):
-				m_point(contact.m_point),
-				m_normal(contact.m_normal),
-				m_depth(contact.m_depth),
-				m_feature1(contact.m_feature1),
-				m_feature2(contact.m_feature2)
-    {
-    	m_point = contact.m_point;
-    	m_normal = contact.m_normal;
-    	m_depth = contact.m_depth;
-    	m_feature1 = contact.m_feature1;
-    	m_feature2 = contact.m_feature2;
-    }
+	GIM_CONTACT(const GIM_CONTACT &contact) : m_point(contact.m_point),
+											  m_normal(contact.m_normal),
+											  m_depth(contact.m_depth),
+											  m_feature1(contact.m_feature1),
+											  m_feature2(contact.m_feature2)
+	{
+		m_point = contact.m_point;
+		m_normal = contact.m_normal;
+		m_depth = contact.m_depth;
+		m_feature1 = contact.m_feature1;
+		m_feature2 = contact.m_feature2;
+	}
 
-    GIM_CONTACT(const btVector3 &point,const btVector3 & normal,
-    	 			GREAL depth, GUINT feature1, GUINT feature2):
-				m_point(point),
-				m_normal(normal),
-				m_depth(depth),
-				m_feature1(feature1),
-				m_feature2(feature2)
-    {
-    }
+	GIM_CONTACT(const btVector3 &point, const btVector3 &normal,
+				GREAL depth, GUINT feature1, GUINT feature2) : m_point(point),
+															   m_normal(normal),
+															   m_depth(depth),
+															   m_feature1(feature1),
+															   m_feature2(feature2)
+	{
+	}
 
 	//! Calcs key for coord classification
-    SIMD_FORCE_INLINE GUINT calc_key_contact() const
-    {
-    	GINT _coords[] = {
-    		(GINT)(m_point[0]*1000.0f+1.0f),
-    		(GINT)(m_point[1]*1333.0f),
-    		(GINT)(m_point[2]*2133.0f+3.0f)};
-		GUINT _hash=0;
+	SIMD_FORCE_INLINE GUINT calc_key_contact() const
+	{
+		GINT _coords[] = {
+			(GINT)(m_point[0] * 1000.0f + 1.0f),
+			(GINT)(m_point[1] * 1333.0f),
+			(GINT)(m_point[2] * 2133.0f + 3.0f)};
+		GUINT _hash = 0;
 		GUINT *_uitmp = (GUINT *)(&_coords[0]);
 		_hash = *_uitmp;
 		_uitmp++;
-		_hash += (*_uitmp)<<4;
+		_hash += (*_uitmp) << 4;
 		_uitmp++;
-		_hash += (*_uitmp)<<8;
+		_hash += (*_uitmp) << 8;
 		return _hash;
-    }
+	}
 
-    SIMD_FORCE_INLINE void interpolate_normals( btVector3 * normals,GUINT normal_count)
-    {
-    	btVector3 vec_sum(m_normal);
-		for(GUINT i=0;i<normal_count;i++)
+	SIMD_FORCE_INLINE void interpolate_normals(btVector3 *normals, GUINT normal_count)
+	{
+		btVector3 vec_sum(m_normal);
+		for (GUINT i = 0; i < normal_count; i++)
 		{
 			vec_sum += normals[i];
 		}
 
 		GREAL vec_sum_len = vec_sum.length2();
-		if(vec_sum_len <CONTACT_DIFF_EPSILON) return;
+		if (vec_sum_len < CONTACT_DIFF_EPSILON) return;
 
-		GIM_INV_SQRT(vec_sum_len,vec_sum_len); // 1/sqrt(vec_sum_len)
-
-		m_normal = vec_sum*vec_sum_len;
-    }
+		GIM_INV_SQRT(vec_sum_len, vec_sum_len);  // 1/sqrt(vec_sum_len)
 
+		m_normal = vec_sum * vec_sum_len;
+	}
 };
 
 #endif
 
-class gim_contact_array:public gim_array<GIM_CONTACT>
+class gim_contact_array : public gim_array<GIM_CONTACT>
 {
 public:
-	gim_contact_array():gim_array<GIM_CONTACT>(64)
+	gim_contact_array() : gim_array<GIM_CONTACT>(64)
 	{
 	}
 
-	SIMD_FORCE_INLINE void push_contact(const btVector3 &point,const btVector3 & normal,
-    	 			GREAL depth, GUINT feature1, GUINT feature2)
+	SIMD_FORCE_INLINE void push_contact(const btVector3 &point, const btVector3 &normal,
+										GREAL depth, GUINT feature1, GUINT feature2)
 	{
 		push_back_mem();
-		GIM_CONTACT & newele = back();
+		GIM_CONTACT &newele = back();
 		newele.m_point = point;
 		newele.m_normal = normal;
 		newele.m_depth = depth;
@@ -150,13 +146,13 @@ public:
 	}
 
 	SIMD_FORCE_INLINE void push_triangle_contacts(
-		const GIM_TRIANGLE_CONTACT_DATA & tricontact,
-		GUINT feature1,GUINT feature2)
+		const GIM_TRIANGLE_CONTACT_DATA &tricontact,
+		GUINT feature1, GUINT feature2)
 	{
-		for(GUINT i = 0;i<tricontact.m_point_count ;i++ )
+		for (GUINT i = 0; i < tricontact.m_point_count; i++)
 		{
 			push_back_mem();
-			GIM_CONTACT & newele = back();
+			GIM_CONTACT &newele = back();
 			newele.m_point = tricontact.m_points[i];
 			newele.m_normal = tricontact.m_separating_normal;
 			newele.m_depth = tricontact.m_penetration_depth;
@@ -165,8 +161,8 @@ public:
 		}
 	}
 
-	void merge_contacts(const gim_contact_array & contacts, bool normal_contact_average = true);
-	void merge_contacts_unique(const gim_contact_array & contacts);
+	void merge_contacts(const gim_contact_array &contacts, bool normal_contact_average = true);
+	void merge_contacts_unique(const gim_contact_array &contacts);
 };
 
-#endif // GIM_CONTACT_H_INCLUDED
+#endif  // GIM_CONTACT_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/gim_geom_types.h b/thirdparty/bullet/BulletCollision/Gimpact/gim_geom_types.h
index 6b8f9ea6c21..9dc48f354b9 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/gim_geom_types.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/gim_geom_types.h
@@ -33,11 +33,8 @@ email: projectileman@yahoo.com
 -----------------------------------------------------------------------------
 */
 
-
 #include "gim_math.h"
 
-
-
 //! Short Integer vector 2D
 typedef GSHORT vec2s[2];
 //! Integer vector 3D
@@ -92,6 +89,4 @@ typedef GREAL quatf[4];
 
 //typedef struct _aabb3f aabb3f;
 
-
-
-#endif // GIM_GEOM_TYPES_H_INCLUDED
+#endif  // GIM_GEOM_TYPES_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/gim_geometry.h b/thirdparty/bullet/BulletCollision/Gimpact/gim_geometry.h
index c67a6991c02..4a7ac3c4d80 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/gim_geometry.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/gim_geometry.h
@@ -39,4 +39,4 @@ email: projectileman@yahoo.com
 #include "gim_box_collision.h"
 #include "gim_tri_collision.h"
 
-#endif // GIM_VECTOR_H_INCLUDED
+#endif  // GIM_VECTOR_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/gim_hash_table.h b/thirdparty/bullet/BulletCollision/Gimpact/gim_hash_table.h
index e4237c2c570..abf88d3108d 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/gim_hash_table.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/gim_hash_table.h
@@ -34,53 +34,52 @@ email: projectileman@yahoo.com
 
 #include "gim_radixsort.h"
 
-
-#define GIM_INVALID_HASH 0xffffffff //!< A very very high value
+#define GIM_INVALID_HASH 0xffffffff  //!< A very very high value
 #define GIM_DEFAULT_HASH_TABLE_SIZE 380
 #define GIM_DEFAULT_HASH_TABLE_NODE_SIZE 4
 #define GIM_HASH_TABLE_GROW_FACTOR 2
 
-#define GIM_MIN_RADIX_SORT_SIZE 860 //!< calibrated on a PIII
+#define GIM_MIN_RADIX_SORT_SIZE 860  //!< calibrated on a PIII
 
-template<typename T>
+template <typename T>
 struct GIM_HASH_TABLE_NODE
 {
-    GUINT m_key;
-    T m_data;
-    GIM_HASH_TABLE_NODE()
-    {
-    }
+	GUINT m_key;
+	T m_data;
+	GIM_HASH_TABLE_NODE()
+	{
+	}
 
-    GIM_HASH_TABLE_NODE(const GIM_HASH_TABLE_NODE & value)
-    {
-        m_key = value.m_key;
-        m_data = value.m_data;
-    }
+	GIM_HASH_TABLE_NODE(const GIM_HASH_TABLE_NODE& value)
+	{
+		m_key = value.m_key;
+		m_data = value.m_data;
+	}
 
-    GIM_HASH_TABLE_NODE(GUINT key, const T & data)
-    {
-        m_key = key;
-        m_data = data;
-    }
+	GIM_HASH_TABLE_NODE(GUINT key, const T& data)
+	{
+		m_key = key;
+		m_data = data;
+	}
 
-    bool operator <(const GIM_HASH_TABLE_NODE<T> & other) const
+	bool operator<(const GIM_HASH_TABLE_NODE<T>& other) const
 	{
 		///inverse order, further objects are first
-		if(m_key <  other.m_key) return true;
+		if (m_key < other.m_key) return true;
 		return false;
 	}
 
-	bool operator >(const GIM_HASH_TABLE_NODE<T> & other) const
+	bool operator>(const GIM_HASH_TABLE_NODE<T>& other) const
 	{
 		///inverse order, further objects are first
-		if(m_key >  other.m_key) return true;
+		if (m_key > other.m_key) return true;
 		return false;
 	}
 
-	bool operator ==(const GIM_HASH_TABLE_NODE<T> & other) const
+	bool operator==(const GIM_HASH_TABLE_NODE<T>& other) const
 	{
 		///inverse order, further objects are first
-		if(m_key ==  other.m_key) return true;
+		if (m_key == other.m_key) return true;
 		return false;
 	}
 };
@@ -89,21 +88,19 @@ struct GIM_HASH_TABLE_NODE
 class GIM_HASH_NODE_GET_KEY
 {
 public:
-	template<class T>
-	inline GUINT operator()( const T& a)
+	template <class T>
+	inline GUINT operator()(const T& a)
 	{
 		return a.m_key;
 	}
 };
 
-
-
 ///Macro for comparing the key and the element
 class GIM_HASH_NODE_CMP_KEY_MACRO
 {
 public:
-	template<class T>
-	inline int operator() ( const T& a, GUINT key)
+	template <class T>
+	inline int operator()(const T& a, GUINT key)
 	{
 		return ((int)(a.m_key - key));
 	}
@@ -113,65 +110,53 @@ public:
 class GIM_HASH_NODE_CMP_MACRO
 {
 public:
-	template<class T>
-	inline int operator() ( const T& a, const T& b )
+	template <class T>
+	inline int operator()(const T& a, const T& b)
 	{
 		return ((int)(a.m_key - b.m_key));
 	}
 };
 
-
-
-
-
 //! Sorting for hash table
 /*!
 switch automatically between quicksort and radixsort
 */
-template<typename T>
-void gim_sort_hash_node_array(T * array, GUINT array_count)
+template <typename T>
+void gim_sort_hash_node_array(T* array, GUINT array_count)
 {
-    if(array_count<GIM_MIN_RADIX_SORT_SIZE)
-    {
-    	gim_heap_sort(array,array_count,GIM_HASH_NODE_CMP_MACRO());
-    }
-    else
-    {
-    	memcopy_elements_func cmpfunc;
-    	gim_radix_sort(array,array_count,GIM_HASH_NODE_GET_KEY(),cmpfunc);
-    }
+	if (array_count < GIM_MIN_RADIX_SORT_SIZE)
+	{
+		gim_heap_sort(array, array_count, GIM_HASH_NODE_CMP_MACRO());
+	}
+	else
+	{
+		memcopy_elements_func cmpfunc;
+		gim_radix_sort(array, array_count, GIM_HASH_NODE_GET_KEY(), cmpfunc);
+	}
 }
 
-
-
-
-
-
 // Note: assumes long is at least 32 bits.
 #define GIM_NUM_PRIME 28
 
 static const GUINT gim_prime_list[GIM_NUM_PRIME] =
-{
-  53ul,         97ul,         193ul,       389ul,       769ul,
-  1543ul,       3079ul,       6151ul,      12289ul,     24593ul,
-  49157ul,      98317ul,      196613ul,    393241ul,    786433ul,
-  1572869ul,    3145739ul,    6291469ul,   12582917ul,  25165843ul,
-  50331653ul,   100663319ul,  201326611ul, 402653189ul, 805306457ul,
-  1610612741ul, 3221225473ul, 4294967291ul
-};
+	{
+		53ul, 97ul, 193ul, 389ul, 769ul,
+		1543ul, 3079ul, 6151ul, 12289ul, 24593ul,
+		49157ul, 98317ul, 196613ul, 393241ul, 786433ul,
+		1572869ul, 3145739ul, 6291469ul, 12582917ul, 25165843ul,
+		50331653ul, 100663319ul, 201326611ul, 402653189ul, 805306457ul,
+		1610612741ul, 3221225473ul, 4294967291ul};
 
 inline GUINT gim_next_prime(GUINT number)
 {
-    //Find nearest upper prime
-    GUINT result_ind = 0;
-    gim_binary_search(gim_prime_list,0,(GIM_NUM_PRIME-2),number,result_ind);
+	//Find nearest upper prime
+	GUINT result_ind = 0;
+	gim_binary_search(gim_prime_list, 0, (GIM_NUM_PRIME - 2), number, result_ind);
 
-    // inv: result_ind < 28
-    return gim_prime_list[result_ind];
+	// inv: result_ind < 28
+	return gim_prime_list[result_ind];
 }
 
-
-
 //! A compact hash table implementation
 /*!
 A memory aligned compact hash table that coud be treated as an array.
@@ -187,129 +172,124 @@ When the array size reaches the size equivalent to 'min_hash_table_size', then i
 </ul>
 
 */
-template<class T>
+template <class T>
 class gim_hash_table
 {
 protected:
-    typedef GIM_HASH_TABLE_NODE<T> _node_type;
+	typedef GIM_HASH_TABLE_NODE<T> _node_type;
 
-    //!The nodes
-    //array< _node_type, SuperAllocator<_node_type> > m_nodes;
-    gim_array< _node_type > m_nodes;
-    //SuperBufferedArray< _node_type > m_nodes;
-    bool m_sorted;
+	//!The nodes
+	//array< _node_type, SuperAllocator<_node_type> > m_nodes;
+	gim_array<_node_type> m_nodes;
+	//SuperBufferedArray< _node_type > m_nodes;
+	bool m_sorted;
 
-    ///Hash table data management. The hash table has the indices to the corresponding m_nodes array
-    GUINT * m_hash_table;//!<
-    GUINT m_table_size;//!<
-    GUINT m_node_size;//!<
-    GUINT m_min_hash_table_size;
+	///Hash table data management. The hash table has the indices to the corresponding m_nodes array
+	GUINT* m_hash_table;  //!<
+	GUINT m_table_size;   //!<
+	GUINT m_node_size;    //!<
+	GUINT m_min_hash_table_size;
 
+	//! Returns the cell index
+	inline GUINT _find_cell(GUINT hashkey)
+	{
+		_node_type* nodesptr = m_nodes.pointer();
+		GUINT start_index = (hashkey % m_table_size) * m_node_size;
+		GUINT end_index = start_index + m_node_size;
 
+		while (start_index < end_index)
+		{
+			GUINT value = m_hash_table[start_index];
+			if (value != GIM_INVALID_HASH)
+			{
+				if (nodesptr[value].m_key == hashkey) return start_index;
+			}
+			start_index++;
+		}
+		return GIM_INVALID_HASH;
+	}
 
-    //! Returns the cell index
-    inline GUINT _find_cell(GUINT hashkey)
-    {
-        _node_type * nodesptr = m_nodes.pointer();
-        GUINT start_index = (hashkey%m_table_size)*m_node_size;
-        GUINT end_index = start_index + m_node_size;
+	//! Find the avaliable cell for the hashkey, and return an existing cell if it has the same hash key
+	inline GUINT _find_avaliable_cell(GUINT hashkey)
+	{
+		_node_type* nodesptr = m_nodes.pointer();
+		GUINT avaliable_index = GIM_INVALID_HASH;
+		GUINT start_index = (hashkey % m_table_size) * m_node_size;
+		GUINT end_index = start_index + m_node_size;
 
-        while(start_index<end_index)
-        {
-            GUINT value = m_hash_table[start_index];
-            if(value != GIM_INVALID_HASH)
-            {
-                if(nodesptr[value].m_key == hashkey) return start_index;
-            }
-            start_index++;
-        }
-        return GIM_INVALID_HASH;
-    }
+		while (start_index < end_index)
+		{
+			GUINT value = m_hash_table[start_index];
+			if (value == GIM_INVALID_HASH)
+			{
+				if (avaliable_index == GIM_INVALID_HASH)
+				{
+					avaliable_index = start_index;
+				}
+			}
+			else if (nodesptr[value].m_key == hashkey)
+			{
+				return start_index;
+			}
+			start_index++;
+		}
+		return avaliable_index;
+	}
 
-    //! Find the avaliable cell for the hashkey, and return an existing cell if it has the same hash key
-    inline GUINT _find_avaliable_cell(GUINT hashkey)
-    {
-        _node_type * nodesptr = m_nodes.pointer();
-        GUINT avaliable_index = GIM_INVALID_HASH;
-        GUINT start_index = (hashkey%m_table_size)*m_node_size;
-        GUINT end_index = start_index + m_node_size;
-
-        while(start_index<end_index)
-        {
-            GUINT value = m_hash_table[start_index];
-            if(value == GIM_INVALID_HASH)
-            {
-                if(avaliable_index==GIM_INVALID_HASH)
-                {
-                    avaliable_index = start_index;
-                }
-            }
-            else if(nodesptr[value].m_key == hashkey)
-            {
-                return start_index;
-            }
-            start_index++;
-        }
-        return avaliable_index;
-    }
-
-
-
-    //! reserves the memory for the hash table.
-    /*!
+	//! reserves the memory for the hash table.
+	/*!
     \pre hash table must be empty
     \post reserves the memory for the hash table, an initializes all elements to GIM_INVALID_HASH.
     */
-    inline void _reserve_table_memory(GUINT newtablesize)
-    {
-        if(newtablesize==0) return;
-        if(m_node_size==0) return;
+	inline void _reserve_table_memory(GUINT newtablesize)
+	{
+		if (newtablesize == 0) return;
+		if (m_node_size == 0) return;
 
-        //Get a Prime size
+		//Get a Prime size
 
-        m_table_size = gim_next_prime(newtablesize);
+		m_table_size = gim_next_prime(newtablesize);
 
-        GUINT datasize = m_table_size*m_node_size;
-        //Alloc the data buffer
-        m_hash_table =  (GUINT *)gim_alloc(datasize*sizeof(GUINT));
-    }
+		GUINT datasize = m_table_size * m_node_size;
+		//Alloc the data buffer
+		m_hash_table = (GUINT*)gim_alloc(datasize * sizeof(GUINT));
+	}
 
-    inline void _invalidate_keys()
-    {
-        GUINT datasize = m_table_size*m_node_size;
-        for(GUINT i=0;i<datasize;i++)
-        {
-            m_hash_table[i] = GIM_INVALID_HASH;// invalidate keys
-        }
-    }
+	inline void _invalidate_keys()
+	{
+		GUINT datasize = m_table_size * m_node_size;
+		for (GUINT i = 0; i < datasize; i++)
+		{
+			m_hash_table[i] = GIM_INVALID_HASH;  // invalidate keys
+		}
+	}
 
-    //! Clear all memory for the hash table
-    inline void _clear_table_memory()
-    {
-        if(m_hash_table==NULL) return;
-        gim_free(m_hash_table);
-        m_hash_table = NULL;
-        m_table_size = 0;
-    }
+	//! Clear all memory for the hash table
+	inline void _clear_table_memory()
+	{
+		if (m_hash_table == NULL) return;
+		gim_free(m_hash_table);
+		m_hash_table = NULL;
+		m_table_size = 0;
+	}
 
-    //! Invalidates the keys (Assigning GIM_INVALID_HASH to all) Reorders the hash keys
-    inline void _rehash()
-    {
-        _invalidate_keys();
+	//! Invalidates the keys (Assigning GIM_INVALID_HASH to all) Reorders the hash keys
+	inline void _rehash()
+	{
+		_invalidate_keys();
 
-        _node_type * nodesptr = m_nodes.pointer();
-        for(GUINT i=0;i<(GUINT)m_nodes.size();i++)
-        {
-            GUINT nodekey = nodesptr[i].m_key;
-            if(nodekey != GIM_INVALID_HASH)
-            {
-                //Search for the avaliable cell in buffer
-                GUINT index = _find_avaliable_cell(nodekey);
+		_node_type* nodesptr = m_nodes.pointer();
+		for (GUINT i = 0; i < (GUINT)m_nodes.size(); i++)
+		{
+			GUINT nodekey = nodesptr[i].m_key;
+			if (nodekey != GIM_INVALID_HASH)
+			{
+				//Search for the avaliable cell in buffer
+				GUINT index = _find_avaliable_cell(nodekey);
 
-
-				if(m_hash_table[index]!=GIM_INVALID_HASH)
-				{//The new index is alreade used... discard this new incomming object, repeated key
-				    btAssert(m_hash_table[index]==nodekey);
+				if (m_hash_table[index] != GIM_INVALID_HASH)
+				{  //The new index is alreade used... discard this new incomming object, repeated key
+					btAssert(m_hash_table[index] == nodekey);
 					nodesptr[i].m_key = GIM_INVALID_HASH;
 				}
 				else
@@ -318,585 +298,560 @@ protected:
 					//Assign the value for alloc
 					m_hash_table[index] = i;
 				}
-            }
-        }
-    }
+			}
+		}
+	}
 
-    //! Resize hash table indices
-    inline void _resize_table(GUINT newsize)
-    {
-        //Clear memory
-        _clear_table_memory();
-        //Alloc the data
-        _reserve_table_memory(newsize);
-        //Invalidate keys and rehash
-        _rehash();
-    }
+	//! Resize hash table indices
+	inline void _resize_table(GUINT newsize)
+	{
+		//Clear memory
+		_clear_table_memory();
+		//Alloc the data
+		_reserve_table_memory(newsize);
+		//Invalidate keys and rehash
+		_rehash();
+	}
 
-    //! Destroy hash table memory
-    inline void _destroy()
-    {
-        if(m_hash_table==NULL) return;
-        _clear_table_memory();
-    }
+	//! Destroy hash table memory
+	inline void _destroy()
+	{
+		if (m_hash_table == NULL) return;
+		_clear_table_memory();
+	}
 
-    //! Finds an avaliable hash table cell, and resizes the table if there isn't space
-    inline GUINT _assign_hash_table_cell(GUINT hashkey)
-    {
-        GUINT cell_index = _find_avaliable_cell(hashkey);
+	//! Finds an avaliable hash table cell, and resizes the table if there isn't space
+	inline GUINT _assign_hash_table_cell(GUINT hashkey)
+	{
+		GUINT cell_index = _find_avaliable_cell(hashkey);
 
-        if(cell_index==GIM_INVALID_HASH)
-        {
-            //rehashing
-            _resize_table(m_table_size+1);
-            GUINT cell_index = _find_avaliable_cell(hashkey);
-            btAssert(cell_index!=GIM_INVALID_HASH);
-        }
-        return cell_index;
-    }
+		if (cell_index == GIM_INVALID_HASH)
+		{
+			//rehashing
+			_resize_table(m_table_size + 1);
+			GUINT cell_index = _find_avaliable_cell(hashkey);
+			btAssert(cell_index != GIM_INVALID_HASH);
+		}
+		return cell_index;
+	}
 
-    //! erase by index in hash table
-    inline bool _erase_by_index_hash_table(GUINT index)
-    {
-        if(index >= m_nodes.size()) return false;
-        if(m_nodes[index].m_key != GIM_INVALID_HASH)
-        {
-            //Search for the avaliable cell in buffer
-            GUINT cell_index = _find_cell(m_nodes[index].m_key);
+	//! erase by index in hash table
+	inline bool _erase_by_index_hash_table(GUINT index)
+	{
+		if (index >= m_nodes.size()) return false;
+		if (m_nodes[index].m_key != GIM_INVALID_HASH)
+		{
+			//Search for the avaliable cell in buffer
+			GUINT cell_index = _find_cell(m_nodes[index].m_key);
 
-            btAssert(cell_index!=GIM_INVALID_HASH);
-            btAssert(m_hash_table[cell_index]==index);
+			btAssert(cell_index != GIM_INVALID_HASH);
+			btAssert(m_hash_table[cell_index] == index);
 
-            m_hash_table[cell_index] = GIM_INVALID_HASH;
-        }
+			m_hash_table[cell_index] = GIM_INVALID_HASH;
+		}
 
-        return this->_erase_unsorted(index);
-    }
+		return this->_erase_unsorted(index);
+	}
 
-    //! erase by key in hash table
-    inline bool _erase_hash_table(GUINT hashkey)
-    {
-        if(hashkey == GIM_INVALID_HASH) return false;
+	//! erase by key in hash table
+	inline bool _erase_hash_table(GUINT hashkey)
+	{
+		if (hashkey == GIM_INVALID_HASH) return false;
 
-        //Search for the avaliable cell in buffer
-        GUINT cell_index = _find_cell(hashkey);
-        if(cell_index ==GIM_INVALID_HASH) return false;
+		//Search for the avaliable cell in buffer
+		GUINT cell_index = _find_cell(hashkey);
+		if (cell_index == GIM_INVALID_HASH) return false;
 
-        GUINT index = m_hash_table[cell_index];
-        m_hash_table[cell_index] = GIM_INVALID_HASH;
+		GUINT index = m_hash_table[cell_index];
+		m_hash_table[cell_index] = GIM_INVALID_HASH;
 
-        return this->_erase_unsorted(index);
-    }
+		return this->_erase_unsorted(index);
+	}
 
-
-
-    //! insert an element in hash table
-    /*!
+	//! insert an element in hash table
+	/*!
     If the element exists, this won't insert the element
     \return the index in the array of the existing element,or GIM_INVALID_HASH if the element has been inserted
     If so, the element has been inserted at the last position of the array.
     */
-    inline GUINT _insert_hash_table(GUINT hashkey, const T & value)
-    {
-        if(hashkey==GIM_INVALID_HASH)
-        {
-            //Insert anyway
-            _insert_unsorted(hashkey,value);
-            return GIM_INVALID_HASH;
-        }
+	inline GUINT _insert_hash_table(GUINT hashkey, const T& value)
+	{
+		if (hashkey == GIM_INVALID_HASH)
+		{
+			//Insert anyway
+			_insert_unsorted(hashkey, value);
+			return GIM_INVALID_HASH;
+		}
 
-        GUINT cell_index = _assign_hash_table_cell(hashkey);
+		GUINT cell_index = _assign_hash_table_cell(hashkey);
 
-        GUINT value_key = m_hash_table[cell_index];
+		GUINT value_key = m_hash_table[cell_index];
 
-        if(value_key!= GIM_INVALID_HASH) return value_key;// Not overrited
+		if (value_key != GIM_INVALID_HASH) return value_key;  // Not overrited
 
-        m_hash_table[cell_index] = m_nodes.size();
+		m_hash_table[cell_index] = m_nodes.size();
 
-        _insert_unsorted(hashkey,value);
-        return GIM_INVALID_HASH;
-    }
+		_insert_unsorted(hashkey, value);
+		return GIM_INVALID_HASH;
+	}
 
-    //! insert an element in hash table.
-    /*!
+	//! insert an element in hash table.
+	/*!
     If the element exists, this replaces the element.
     \return the index in the array of the existing element,or GIM_INVALID_HASH if the element has been inserted
     If so, the element has been inserted at the last position of the array.
     */
-    inline GUINT _insert_hash_table_replace(GUINT hashkey, const T & value)
-    {
-        if(hashkey==GIM_INVALID_HASH)
-        {
-            //Insert anyway
-            _insert_unsorted(hashkey,value);
-            return GIM_INVALID_HASH;
-        }
+	inline GUINT _insert_hash_table_replace(GUINT hashkey, const T& value)
+	{
+		if (hashkey == GIM_INVALID_HASH)
+		{
+			//Insert anyway
+			_insert_unsorted(hashkey, value);
+			return GIM_INVALID_HASH;
+		}
 
-        GUINT cell_index = _assign_hash_table_cell(hashkey);
+		GUINT cell_index = _assign_hash_table_cell(hashkey);
 
-        GUINT value_key = m_hash_table[cell_index];
+		GUINT value_key = m_hash_table[cell_index];
 
-        if(value_key!= GIM_INVALID_HASH)
-        {//replaces the existing
-            m_nodes[value_key] = _node_type(hashkey,value);
-            return value_key;// index of the replaced element
-        }
+		if (value_key != GIM_INVALID_HASH)
+		{  //replaces the existing
+			m_nodes[value_key] = _node_type(hashkey, value);
+			return value_key;  // index of the replaced element
+		}
 
-        m_hash_table[cell_index] = m_nodes.size();
+		m_hash_table[cell_index] = m_nodes.size();
 
-        _insert_unsorted(hashkey,value);
-        return GIM_INVALID_HASH;
+		_insert_unsorted(hashkey, value);
+		return GIM_INVALID_HASH;
+	}
 
-    }
+	///Sorted array data management. The hash table has the indices to the corresponding m_nodes array
+	inline bool _erase_sorted(GUINT index)
+	{
+		if (index >= (GUINT)m_nodes.size()) return false;
+		m_nodes.erase_sorted(index);
+		if (m_nodes.size() < 2) m_sorted = false;
+		return true;
+	}
 
-    
-    ///Sorted array data management. The hash table has the indices to the corresponding m_nodes array
-    inline bool _erase_sorted(GUINT index)
-    {
-        if(index>=(GUINT)m_nodes.size()) return false;
-        m_nodes.erase_sorted(index);
-		if(m_nodes.size()<2) m_sorted = false;
-        return true;
-    }
+	//! faster, but unsorted
+	inline bool _erase_unsorted(GUINT index)
+	{
+		if (index >= m_nodes.size()) return false;
 
-    //! faster, but unsorted
-    inline bool _erase_unsorted(GUINT index)
-    {
-        if(index>=m_nodes.size()) return false;
-
-        GUINT lastindex = m_nodes.size()-1;
-        if(index<lastindex && m_hash_table!=0)
-        {
-			GUINT hashkey =  m_nodes[lastindex].m_key;
-			if(hashkey!=GIM_INVALID_HASH)
+		GUINT lastindex = m_nodes.size() - 1;
+		if (index < lastindex && m_hash_table != 0)
+		{
+			GUINT hashkey = m_nodes[lastindex].m_key;
+			if (hashkey != GIM_INVALID_HASH)
 			{
 				//update the new position of the last element
 				GUINT cell_index = _find_cell(hashkey);
-				btAssert(cell_index!=GIM_INVALID_HASH);
+				btAssert(cell_index != GIM_INVALID_HASH);
 				//new position of the last element which will be swaped
 				m_hash_table[cell_index] = index;
 			}
-        }
-        m_nodes.erase(index);
-        m_sorted = false;
-        return true;
-    }
+		}
+		m_nodes.erase(index);
+		m_sorted = false;
+		return true;
+	}
 
-    //! Insert in position ordered
-    /*!
+	//! Insert in position ordered
+	/*!
     Also checks if it is needed to transform this container to a hash table, by calling check_for_switching_to_hashtable
     */
-    inline void _insert_in_pos(GUINT hashkey, const T & value, GUINT pos)
-    {
-        m_nodes.insert(_node_type(hashkey,value),pos);
-        this->check_for_switching_to_hashtable();
-    }
+	inline void _insert_in_pos(GUINT hashkey, const T& value, GUINT pos)
+	{
+		m_nodes.insert(_node_type(hashkey, value), pos);
+		this->check_for_switching_to_hashtable();
+	}
 
-    //! Insert an element in an ordered array
-    inline GUINT _insert_sorted(GUINT hashkey, const T & value)
-    {
-        if(hashkey==GIM_INVALID_HASH || size()==0)
-        {
-            m_nodes.push_back(_node_type(hashkey,value));
-            return GIM_INVALID_HASH;
-        }
-        //Insert at last position
-        //Sort element
+	//! Insert an element in an ordered array
+	inline GUINT _insert_sorted(GUINT hashkey, const T& value)
+	{
+		if (hashkey == GIM_INVALID_HASH || size() == 0)
+		{
+			m_nodes.push_back(_node_type(hashkey, value));
+			return GIM_INVALID_HASH;
+		}
+		//Insert at last position
+		//Sort element
 
+		GUINT result_ind = 0;
+		GUINT last_index = m_nodes.size() - 1;
+		_node_type* ptr = m_nodes.pointer();
 
-        GUINT result_ind=0;
-        GUINT last_index = m_nodes.size()-1;
-        _node_type * ptr = m_nodes.pointer();
+		bool found = gim_binary_search_ex(
+			ptr, 0, last_index, result_ind, hashkey, GIM_HASH_NODE_CMP_KEY_MACRO());
 
-        bool found = gim_binary_search_ex(
-        	ptr,0,last_index,result_ind,hashkey,GIM_HASH_NODE_CMP_KEY_MACRO());
+		//Insert before found index
+		if (found)
+		{
+			return result_ind;
+		}
+		else
+		{
+			_insert_in_pos(hashkey, value, result_ind);
+		}
+		return GIM_INVALID_HASH;
+	}
 
+	inline GUINT _insert_sorted_replace(GUINT hashkey, const T& value)
+	{
+		if (hashkey == GIM_INVALID_HASH || size() == 0)
+		{
+			m_nodes.push_back(_node_type(hashkey, value));
+			return GIM_INVALID_HASH;
+		}
+		//Insert at last position
+		//Sort element
+		GUINT result_ind;
+		GUINT last_index = m_nodes.size() - 1;
+		_node_type* ptr = m_nodes.pointer();
 
-        //Insert before found index
-        if(found)
-        {
-            return result_ind;
-        }
-        else
-        {
-            _insert_in_pos(hashkey, value, result_ind);
-        }
-        return GIM_INVALID_HASH;
-    }
+		bool found = gim_binary_search_ex(
+			ptr, 0, last_index, result_ind, hashkey, GIM_HASH_NODE_CMP_KEY_MACRO());
 
-    inline GUINT _insert_sorted_replace(GUINT hashkey, const T & value)
-    {
-        if(hashkey==GIM_INVALID_HASH || size()==0)
-        {
-            m_nodes.push_back(_node_type(hashkey,value));
-            return GIM_INVALID_HASH;
-        }
-        //Insert at last position
-        //Sort element
-        GUINT result_ind;
-        GUINT last_index = m_nodes.size()-1;
-        _node_type * ptr = m_nodes.pointer();
+		//Insert before found index
+		if (found)
+		{
+			m_nodes[result_ind] = _node_type(hashkey, value);
+		}
+		else
+		{
+			_insert_in_pos(hashkey, value, result_ind);
+		}
+		return result_ind;
+	}
 
-        bool found = gim_binary_search_ex(
-        	ptr,0,last_index,result_ind,hashkey,GIM_HASH_NODE_CMP_KEY_MACRO());
-
-        //Insert before found index
-        if(found)
-        {
-            m_nodes[result_ind] = _node_type(hashkey,value);
-        }
-        else
-        {
-            _insert_in_pos(hashkey, value, result_ind);
-        }
-        return result_ind;
-    }
-
-    //! Fast insertion in m_nodes array
-    inline GUINT  _insert_unsorted(GUINT hashkey, const T & value)
-    {
-        m_nodes.push_back(_node_type(hashkey,value));
-        m_sorted = false;
-        return GIM_INVALID_HASH;
-    }
-
-    
+	//! Fast insertion in m_nodes array
+	inline GUINT _insert_unsorted(GUINT hashkey, const T& value)
+	{
+		m_nodes.push_back(_node_type(hashkey, value));
+		m_sorted = false;
+		return GIM_INVALID_HASH;
+	}
 
 public:
-
-    /*!
+	/*!
         <li> if node_size = 0, then this container becomes a simple sorted array allocator. reserve_size is used for reserve memory in m_nodes.
         When the array size reaches the size equivalent to 'min_hash_table_size', then it becomes a hash table by calling check_for_switching_to_hashtable.
         <li> If node_size != 0, then this container becomes a hash table for ever
         </ul>
     */
-    gim_hash_table(GUINT reserve_size = GIM_DEFAULT_HASH_TABLE_SIZE,
-                     GUINT node_size = GIM_DEFAULT_HASH_TABLE_NODE_SIZE,
-                     GUINT min_hash_table_size = GIM_INVALID_HASH)
-    {
-        m_hash_table = NULL;
-        m_table_size = 0;
-        m_sorted = false;
-        m_node_size = node_size;
-        m_min_hash_table_size = min_hash_table_size;
+	gim_hash_table(GUINT reserve_size = GIM_DEFAULT_HASH_TABLE_SIZE,
+				   GUINT node_size = GIM_DEFAULT_HASH_TABLE_NODE_SIZE,
+				   GUINT min_hash_table_size = GIM_INVALID_HASH)
+	{
+		m_hash_table = NULL;
+		m_table_size = 0;
+		m_sorted = false;
+		m_node_size = node_size;
+		m_min_hash_table_size = min_hash_table_size;
 
-        if(m_node_size!=0)
-        {
-            if(reserve_size!=0)
-            {
-                m_nodes.reserve(reserve_size);
-                _reserve_table_memory(reserve_size);
-                _invalidate_keys();
-            }
-            else
-            {
-                m_nodes.reserve(GIM_DEFAULT_HASH_TABLE_SIZE);
-                _reserve_table_memory(GIM_DEFAULT_HASH_TABLE_SIZE);
-                _invalidate_keys();
-            }
-        }
-        else if(reserve_size!=0)
-        {
-            m_nodes.reserve(reserve_size);
-        }
+		if (m_node_size != 0)
+		{
+			if (reserve_size != 0)
+			{
+				m_nodes.reserve(reserve_size);
+				_reserve_table_memory(reserve_size);
+				_invalidate_keys();
+			}
+			else
+			{
+				m_nodes.reserve(GIM_DEFAULT_HASH_TABLE_SIZE);
+				_reserve_table_memory(GIM_DEFAULT_HASH_TABLE_SIZE);
+				_invalidate_keys();
+			}
+		}
+		else if (reserve_size != 0)
+		{
+			m_nodes.reserve(reserve_size);
+		}
+	}
 
-    }
+	~gim_hash_table()
+	{
+		_destroy();
+	}
 
-    ~gim_hash_table()
-    {
-        _destroy();
-    }
+	inline bool is_hash_table()
+	{
+		if (m_hash_table) return true;
+		return false;
+	}
 
-    inline bool is_hash_table()
-    {
-        if(m_hash_table) return true;
-        return false;
-    }
+	inline bool is_sorted()
+	{
+		if (size() < 2) return true;
+		return m_sorted;
+	}
 
-    inline bool is_sorted()
-    {
-        if(size()<2) return true;
-        return m_sorted;
-    }
+	bool sort()
+	{
+		if (is_sorted()) return true;
+		if (m_nodes.size() < 2) return false;
 
-    bool sort()
-    {
-        if(is_sorted()) return true;
-        if(m_nodes.size()<2) return false;
+		_node_type* ptr = m_nodes.pointer();
+		GUINT siz = m_nodes.size();
+		gim_sort_hash_node_array(ptr, siz);
+		m_sorted = true;
 
+		if (m_hash_table)
+		{
+			_rehash();
+		}
+		return true;
+	}
 
-        _node_type * ptr = m_nodes.pointer();
-        GUINT siz = m_nodes.size();
-        gim_sort_hash_node_array(ptr,siz);
-        m_sorted=true;
+	bool switch_to_hashtable()
+	{
+		if (m_hash_table) return false;
+		if (m_node_size == 0) m_node_size = GIM_DEFAULT_HASH_TABLE_NODE_SIZE;
+		if (m_nodes.size() < GIM_DEFAULT_HASH_TABLE_SIZE)
+		{
+			_resize_table(GIM_DEFAULT_HASH_TABLE_SIZE);
+		}
+		else
+		{
+			_resize_table(m_nodes.size() + 1);
+		}
 
+		return true;
+	}
 
+	bool switch_to_sorted_array()
+	{
+		if (m_hash_table == NULL) return true;
+		_clear_table_memory();
+		return sort();
+	}
 
-        if(m_hash_table)
-        {
-            _rehash();
-        }
-        return true;
-    }
+	//!If the container reaches the
+	bool check_for_switching_to_hashtable()
+	{
+		if (this->m_hash_table) return true;
 
-    bool switch_to_hashtable()
-    {
-        if(m_hash_table) return false;
-        if(m_node_size==0) m_node_size = GIM_DEFAULT_HASH_TABLE_NODE_SIZE;
-        if(m_nodes.size()<GIM_DEFAULT_HASH_TABLE_SIZE)
-        {
-            _resize_table(GIM_DEFAULT_HASH_TABLE_SIZE);
-        }
-        else
-        {
-            _resize_table(m_nodes.size()+1);
-        }
+		if (!(m_nodes.size() < m_min_hash_table_size))
+		{
+			if (m_node_size == 0)
+			{
+				m_node_size = GIM_DEFAULT_HASH_TABLE_NODE_SIZE;
+			}
 
-        return true;
-    }
+			_resize_table(m_nodes.size() + 1);
+			return true;
+		}
+		return false;
+	}
 
-    bool switch_to_sorted_array()
-    {
-        if(m_hash_table==NULL) return true;
-        _clear_table_memory();
-        return sort();
-    }
+	inline void set_sorted(bool value)
+	{
+		m_sorted = value;
+	}
 
-    //!If the container reaches the
-    bool check_for_switching_to_hashtable()
-    {
-        if(this->m_hash_table) return true;
+	//! Retrieves the amount of keys.
+	inline GUINT size() const
+	{
+		return m_nodes.size();
+	}
 
-        if(!(m_nodes.size()< m_min_hash_table_size))
-        {
-            if(m_node_size == 0)
-            {
-                m_node_size = GIM_DEFAULT_HASH_TABLE_NODE_SIZE;
-            }
+	//! Retrieves the hash key.
+	inline GUINT get_key(GUINT index) const
+	{
+		return m_nodes[index].m_key;
+	}
 
-            _resize_table(m_nodes.size()+1);
-            return true;
-        }
-        return false;
-    }
-
-    inline void set_sorted(bool value)
-    {
-    	m_sorted = value;
-    }
-
-    //! Retrieves the amount of keys.
-    inline GUINT size() const
-    {
-        return m_nodes.size();
-    }
-
-    //! Retrieves the hash key.
-    inline GUINT get_key(GUINT index) const
-    {
-        return m_nodes[index].m_key;
-    }
-
-    //! Retrieves the value by index
-    /*!
+	//! Retrieves the value by index
+	/*!
     */
-    inline T * get_value_by_index(GUINT index)
-    {
-        return &m_nodes[index].m_data;
-    }
+	inline T* get_value_by_index(GUINT index)
+	{
+		return &m_nodes[index].m_data;
+	}
 
-    inline const T& operator[](GUINT index) const
-    {
-        return m_nodes[index].m_data;
-    }
+	inline const T& operator[](GUINT index) const
+	{
+		return m_nodes[index].m_data;
+	}
 
-    inline T& operator[](GUINT index)
-    {
-        return m_nodes[index].m_data;
-    }
+	inline T& operator[](GUINT index)
+	{
+		return m_nodes[index].m_data;
+	}
 
-    //! Finds the index of the element with the key
-    /*!
+	//! Finds the index of the element with the key
+	/*!
     \return the index in the array of the existing element,or GIM_INVALID_HASH if the element has been inserted
     If so, the element has been inserted at the last position of the array.
     */
-    inline GUINT find(GUINT hashkey)
-    {
-        if(m_hash_table)
-        {
-            GUINT cell_index = _find_cell(hashkey);
-            if(cell_index==GIM_INVALID_HASH) return GIM_INVALID_HASH;
-            return m_hash_table[cell_index];
-        }
+	inline GUINT find(GUINT hashkey)
+	{
+		if (m_hash_table)
+		{
+			GUINT cell_index = _find_cell(hashkey);
+			if (cell_index == GIM_INVALID_HASH) return GIM_INVALID_HASH;
+			return m_hash_table[cell_index];
+		}
 		GUINT last_index = m_nodes.size();
-        if(last_index<2)
-        {
-			if(last_index==0) return GIM_INVALID_HASH;
-            if(m_nodes[0].m_key == hashkey) return 0;
-            return GIM_INVALID_HASH;
-        }
-        else if(m_sorted)
-        {
-            //Binary search
-            GUINT result_ind = 0;
+		if (last_index < 2)
+		{
+			if (last_index == 0) return GIM_INVALID_HASH;
+			if (m_nodes[0].m_key == hashkey) return 0;
+			return GIM_INVALID_HASH;
+		}
+		else if (m_sorted)
+		{
+			//Binary search
+			GUINT result_ind = 0;
 			last_index--;
-            _node_type *  ptr =  m_nodes.pointer();
+			_node_type* ptr = m_nodes.pointer();
 
-            bool found = gim_binary_search_ex(ptr,0,last_index,result_ind,hashkey,GIM_HASH_NODE_CMP_KEY_MACRO());
+			bool found = gim_binary_search_ex(ptr, 0, last_index, result_ind, hashkey, GIM_HASH_NODE_CMP_KEY_MACRO());
 
+			if (found) return result_ind;
+		}
+		return GIM_INVALID_HASH;
+	}
 
-            if(found) return result_ind;
-        }
-        return GIM_INVALID_HASH;
-    }
-
-    //! Retrieves the value associated with the index
-    /*!
+	//! Retrieves the value associated with the index
+	/*!
     \return the found element, or null
     */
-    inline T * get_value(GUINT hashkey)
-    {
-        GUINT index = find(hashkey);
-        if(index == GIM_INVALID_HASH) return NULL;
-        return &m_nodes[index].m_data;
-    }
+	inline T* get_value(GUINT hashkey)
+	{
+		GUINT index = find(hashkey);
+		if (index == GIM_INVALID_HASH) return NULL;
+		return &m_nodes[index].m_data;
+	}
 
-
-    /*!
+	/*!
     */
-    inline bool erase_by_index(GUINT index)
-    {
-        if(index > m_nodes.size()) return false;
+	inline bool erase_by_index(GUINT index)
+	{
+		if (index > m_nodes.size()) return false;
 
-        if(m_hash_table == NULL)
-        {
-            if(is_sorted())
-            {
-                return this->_erase_sorted(index);
-            }
-            else
-            {
-                return this->_erase_unsorted(index);
-            }
-        }
-        else
-        {
-            return this->_erase_by_index_hash_table(index);
-        }
-        return false;
-    }
+		if (m_hash_table == NULL)
+		{
+			if (is_sorted())
+			{
+				return this->_erase_sorted(index);
+			}
+			else
+			{
+				return this->_erase_unsorted(index);
+			}
+		}
+		else
+		{
+			return this->_erase_by_index_hash_table(index);
+		}
+		return false;
+	}
 
+	inline bool erase_by_index_unsorted(GUINT index)
+	{
+		if (index > m_nodes.size()) return false;
 
+		if (m_hash_table == NULL)
+		{
+			return this->_erase_unsorted(index);
+		}
+		else
+		{
+			return this->_erase_by_index_hash_table(index);
+		}
+		return false;
+	}
 
-    inline bool erase_by_index_unsorted(GUINT index)
-    {
-        if(index > m_nodes.size()) return false;
-
-        if(m_hash_table == NULL)
-        {
-            return this->_erase_unsorted(index);
-        }
-        else
-        {
-            return this->_erase_by_index_hash_table(index);
-        }
-        return false;
-    }
-
-
-
-    /*!
+	/*!
 
     */
-    inline bool erase_by_key(GUINT hashkey)
-    {
-        if(size()==0) return false;
+	inline bool erase_by_key(GUINT hashkey)
+	{
+		if (size() == 0) return false;
 
-        if(m_hash_table)
-        {
-            return this->_erase_hash_table(hashkey);
-        }
-        //Binary search
+		if (m_hash_table)
+		{
+			return this->_erase_hash_table(hashkey);
+		}
+		//Binary search
 
-        if(is_sorted()==false) return false;
+		if (is_sorted() == false) return false;
 
-        GUINT result_ind = find(hashkey);
-        if(result_ind!= GIM_INVALID_HASH)
-        {
-            return this->_erase_sorted(result_ind);
-        }
-        return false;
-    }
+		GUINT result_ind = find(hashkey);
+		if (result_ind != GIM_INVALID_HASH)
+		{
+			return this->_erase_sorted(result_ind);
+		}
+		return false;
+	}
 
-    void clear()
-    {
-        m_nodes.clear();
+	void clear()
+	{
+		m_nodes.clear();
 
-        if(m_hash_table==NULL) return;
-        GUINT datasize = m_table_size*m_node_size;
-        //Initialize the hashkeys.
-        GUINT i;
-        for(i=0;i<datasize;i++)
-        {
-            m_hash_table[i] = GIM_INVALID_HASH;// invalidate keys
-        }
+		if (m_hash_table == NULL) return;
+		GUINT datasize = m_table_size * m_node_size;
+		//Initialize the hashkeys.
+		GUINT i;
+		for (i = 0; i < datasize; i++)
+		{
+			m_hash_table[i] = GIM_INVALID_HASH;  // invalidate keys
+		}
 		m_sorted = false;
-    }
+	}
 
-    //! Insert an element into the hash
-    /*!
+	//! Insert an element into the hash
+	/*!
     \return If GIM_INVALID_HASH, the object has been inserted succesfully. Else it returns the position
     of the existing element.
     */
-    inline GUINT insert(GUINT hashkey, const T & element)
-    {
-        if(m_hash_table)
-        {
-            return this->_insert_hash_table(hashkey,element);
-        }
-        if(this->is_sorted())
-        {
-            return this->_insert_sorted(hashkey,element);
-        }
-        return this->_insert_unsorted(hashkey,element);
-    }
+	inline GUINT insert(GUINT hashkey, const T& element)
+	{
+		if (m_hash_table)
+		{
+			return this->_insert_hash_table(hashkey, element);
+		}
+		if (this->is_sorted())
+		{
+			return this->_insert_sorted(hashkey, element);
+		}
+		return this->_insert_unsorted(hashkey, element);
+	}
 
-    //! Insert an element into the hash, and could overrite an existing object with the same hash.
-    /*!
+	//! Insert an element into the hash, and could overrite an existing object with the same hash.
+	/*!
     \return If GIM_INVALID_HASH, the object has been inserted succesfully. Else it returns the position
     of the replaced element.
     */
-    inline GUINT insert_override(GUINT hashkey, const T & element)
-    {
-        if(m_hash_table)
-        {
-            return this->_insert_hash_table_replace(hashkey,element);
-        }
-        if(this->is_sorted())
-        {
-            return this->_insert_sorted_replace(hashkey,element);
-        }
-        this->_insert_unsorted(hashkey,element);
-        return m_nodes.size();
-    }
+	inline GUINT insert_override(GUINT hashkey, const T& element)
+	{
+		if (m_hash_table)
+		{
+			return this->_insert_hash_table_replace(hashkey, element);
+		}
+		if (this->is_sorted())
+		{
+			return this->_insert_sorted_replace(hashkey, element);
+		}
+		this->_insert_unsorted(hashkey, element);
+		return m_nodes.size();
+	}
 
-
-
-    //! Insert an element into the hash,But if this container is a sorted array, this inserts it unsorted
-    /*!
+	//! Insert an element into the hash,But if this container is a sorted array, this inserts it unsorted
+	/*!
     */
-    inline GUINT insert_unsorted(GUINT hashkey,const T & element)
-    {
-        if(m_hash_table)
-        {
-            return this->_insert_hash_table(hashkey,element);
-        }
-        return this->_insert_unsorted(hashkey,element);
-    }
-
-
+	inline GUINT insert_unsorted(GUINT hashkey, const T& element)
+	{
+		if (m_hash_table)
+		{
+			return this->_insert_hash_table(hashkey, element);
+		}
+		return this->_insert_unsorted(hashkey, element);
+	}
 };
 
-
-
-#endif // GIM_CONTAINERS_H_INCLUDED
+#endif  // GIM_CONTAINERS_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/gim_linear_math.h b/thirdparty/bullet/BulletCollision/Gimpact/gim_linear_math.h
index 64f11b49543..98401a404ab 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/gim_linear_math.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/gim_linear_math.h
@@ -34,962 +34,900 @@ email: projectileman@yahoo.com
 -----------------------------------------------------------------------------
 */
 
-
 #include "gim_math.h"
 #include "gim_geom_types.h"
 
-
-
-
 //! Zero out a 2D vector
-#define VEC_ZERO_2(a)				\
-{						\
-   (a)[0] = (a)[1] = 0.0f;			\
-}\
-
+#define VEC_ZERO_2(a)           \
+	{                           \
+		(a)[0] = (a)[1] = 0.0f; \
+	}
 
 //! Zero out a 3D vector
-#define VEC_ZERO(a)				\
-{						\
-   (a)[0] = (a)[1] = (a)[2] = 0.0f;		\
-}\
-
+#define VEC_ZERO(a)                      \
+	{                                    \
+		(a)[0] = (a)[1] = (a)[2] = 0.0f; \
+	}
 
 /// Zero out a 4D vector
-#define VEC_ZERO_4(a)				\
-{						\
-   (a)[0] = (a)[1] = (a)[2] = (a)[3] = 0.0f;	\
-}\
-
+#define VEC_ZERO_4(a)                             \
+	{                                             \
+		(a)[0] = (a)[1] = (a)[2] = (a)[3] = 0.0f; \
+	}
 
 /// Vector copy
-#define VEC_COPY_2(b,a)				\
-{						\
-   (b)[0] = (a)[0];				\
-   (b)[1] = (a)[1];				\
-}\
-
+#define VEC_COPY_2(b, a) \
+	{                    \
+		(b)[0] = (a)[0]; \
+		(b)[1] = (a)[1]; \
+	}
 
 /// Copy 3D vector
-#define VEC_COPY(b,a)				\
-{						\
-   (b)[0] = (a)[0];				\
-   (b)[1] = (a)[1];				\
-   (b)[2] = (a)[2];				\
-}\
-
+#define VEC_COPY(b, a)   \
+	{                    \
+		(b)[0] = (a)[0]; \
+		(b)[1] = (a)[1]; \
+		(b)[2] = (a)[2]; \
+	}
 
 /// Copy 4D vector
-#define VEC_COPY_4(b,a)				\
-{						\
-   (b)[0] = (a)[0];				\
-   (b)[1] = (a)[1];				\
-   (b)[2] = (a)[2];				\
-   (b)[3] = (a)[3];				\
-}\
+#define VEC_COPY_4(b, a) \
+	{                    \
+		(b)[0] = (a)[0]; \
+		(b)[1] = (a)[1]; \
+		(b)[2] = (a)[2]; \
+		(b)[3] = (a)[3]; \
+	}
 
 /// VECTOR SWAP
-#define VEC_SWAP(b,a)				\
-{  \
-    GIM_SWAP_NUMBERS((b)[0],(a)[0]);\
-    GIM_SWAP_NUMBERS((b)[1],(a)[1]);\
-    GIM_SWAP_NUMBERS((b)[2],(a)[2]);\
-}\
+#define VEC_SWAP(b, a)                    \
+	{                                     \
+		GIM_SWAP_NUMBERS((b)[0], (a)[0]); \
+		GIM_SWAP_NUMBERS((b)[1], (a)[1]); \
+		GIM_SWAP_NUMBERS((b)[2], (a)[2]); \
+	}
 
 /// Vector difference
-#define VEC_DIFF_2(v21,v2,v1)			\
-{						\
-   (v21)[0] = (v2)[0] - (v1)[0];		\
-   (v21)[1] = (v2)[1] - (v1)[1];		\
-}\
-
+#define VEC_DIFF_2(v21, v2, v1)       \
+	{                                 \
+		(v21)[0] = (v2)[0] - (v1)[0]; \
+		(v21)[1] = (v2)[1] - (v1)[1]; \
+	}
 
 /// Vector difference
-#define VEC_DIFF(v21,v2,v1)			\
-{						\
-   (v21)[0] = (v2)[0] - (v1)[0];		\
-   (v21)[1] = (v2)[1] - (v1)[1];		\
-   (v21)[2] = (v2)[2] - (v1)[2];		\
-}\
-
+#define VEC_DIFF(v21, v2, v1)         \
+	{                                 \
+		(v21)[0] = (v2)[0] - (v1)[0]; \
+		(v21)[1] = (v2)[1] - (v1)[1]; \
+		(v21)[2] = (v2)[2] - (v1)[2]; \
+	}
 
 /// Vector difference
-#define VEC_DIFF_4(v21,v2,v1)			\
-{						\
-   (v21)[0] = (v2)[0] - (v1)[0];		\
-   (v21)[1] = (v2)[1] - (v1)[1];		\
-   (v21)[2] = (v2)[2] - (v1)[2];		\
-   (v21)[3] = (v2)[3] - (v1)[3];		\
-}\
-
+#define VEC_DIFF_4(v21, v2, v1)       \
+	{                                 \
+		(v21)[0] = (v2)[0] - (v1)[0]; \
+		(v21)[1] = (v2)[1] - (v1)[1]; \
+		(v21)[2] = (v2)[2] - (v1)[2]; \
+		(v21)[3] = (v2)[3] - (v1)[3]; \
+	}
 
 /// Vector sum
-#define VEC_SUM_2(v21,v2,v1)			\
-{						\
-   (v21)[0] = (v2)[0] + (v1)[0];		\
-   (v21)[1] = (v2)[1] + (v1)[1];		\
-}\
-
+#define VEC_SUM_2(v21, v2, v1)        \
+	{                                 \
+		(v21)[0] = (v2)[0] + (v1)[0]; \
+		(v21)[1] = (v2)[1] + (v1)[1]; \
+	}
 
 /// Vector sum
-#define VEC_SUM(v21,v2,v1)			\
-{						\
-   (v21)[0] = (v2)[0] + (v1)[0];		\
-   (v21)[1] = (v2)[1] + (v1)[1];		\
-   (v21)[2] = (v2)[2] + (v1)[2];		\
-}\
-
+#define VEC_SUM(v21, v2, v1)          \
+	{                                 \
+		(v21)[0] = (v2)[0] + (v1)[0]; \
+		(v21)[1] = (v2)[1] + (v1)[1]; \
+		(v21)[2] = (v2)[2] + (v1)[2]; \
+	}
 
 /// Vector sum
-#define VEC_SUM_4(v21,v2,v1)			\
-{						\
-   (v21)[0] = (v2)[0] + (v1)[0];		\
-   (v21)[1] = (v2)[1] + (v1)[1];		\
-   (v21)[2] = (v2)[2] + (v1)[2];		\
-   (v21)[3] = (v2)[3] + (v1)[3];		\
-}\
-
+#define VEC_SUM_4(v21, v2, v1)        \
+	{                                 \
+		(v21)[0] = (v2)[0] + (v1)[0]; \
+		(v21)[1] = (v2)[1] + (v1)[1]; \
+		(v21)[2] = (v2)[2] + (v1)[2]; \
+		(v21)[3] = (v2)[3] + (v1)[3]; \
+	}
 
 /// scalar times vector
-#define VEC_SCALE_2(c,a,b)			\
-{						\
-   (c)[0] = (a)*(b)[0];				\
-   (c)[1] = (a)*(b)[1];				\
-}\
-
+#define VEC_SCALE_2(c, a, b)   \
+	{                          \
+		(c)[0] = (a) * (b)[0]; \
+		(c)[1] = (a) * (b)[1]; \
+	}
 
 /// scalar times vector
-#define VEC_SCALE(c,a,b)			\
-{						\
-   (c)[0] = (a)*(b)[0];				\
-   (c)[1] = (a)*(b)[1];				\
-   (c)[2] = (a)*(b)[2];				\
-}\
-
+#define VEC_SCALE(c, a, b)     \
+	{                          \
+		(c)[0] = (a) * (b)[0]; \
+		(c)[1] = (a) * (b)[1]; \
+		(c)[2] = (a) * (b)[2]; \
+	}
 
 /// scalar times vector
-#define VEC_SCALE_4(c,a,b)			\
-{						\
-   (c)[0] = (a)*(b)[0];				\
-   (c)[1] = (a)*(b)[1];				\
-   (c)[2] = (a)*(b)[2];				\
-   (c)[3] = (a)*(b)[3];				\
-}\
-
+#define VEC_SCALE_4(c, a, b)   \
+	{                          \
+		(c)[0] = (a) * (b)[0]; \
+		(c)[1] = (a) * (b)[1]; \
+		(c)[2] = (a) * (b)[2]; \
+		(c)[3] = (a) * (b)[3]; \
+	}
 
 /// accumulate scaled vector
-#define VEC_ACCUM_2(c,a,b)			\
-{						\
-   (c)[0] += (a)*(b)[0];			\
-   (c)[1] += (a)*(b)[1];			\
-}\
-
+#define VEC_ACCUM_2(c, a, b)    \
+	{                           \
+		(c)[0] += (a) * (b)[0]; \
+		(c)[1] += (a) * (b)[1]; \
+	}
 
 /// accumulate scaled vector
-#define VEC_ACCUM(c,a,b)			\
-{						\
-   (c)[0] += (a)*(b)[0];			\
-   (c)[1] += (a)*(b)[1];			\
-   (c)[2] += (a)*(b)[2];			\
-}\
-
+#define VEC_ACCUM(c, a, b)      \
+	{                           \
+		(c)[0] += (a) * (b)[0]; \
+		(c)[1] += (a) * (b)[1]; \
+		(c)[2] += (a) * (b)[2]; \
+	}
 
 /// accumulate scaled vector
-#define VEC_ACCUM_4(c,a,b)			\
-{						\
-   (c)[0] += (a)*(b)[0];			\
-   (c)[1] += (a)*(b)[1];			\
-   (c)[2] += (a)*(b)[2];			\
-   (c)[3] += (a)*(b)[3];			\
-}\
-
+#define VEC_ACCUM_4(c, a, b)    \
+	{                           \
+		(c)[0] += (a) * (b)[0]; \
+		(c)[1] += (a) * (b)[1]; \
+		(c)[2] += (a) * (b)[2]; \
+		(c)[3] += (a) * (b)[3]; \
+	}
 
 /// Vector dot product
-#define VEC_DOT_2(a,b) ((a)[0]*(b)[0] + (a)[1]*(b)[1])
-
+#define VEC_DOT_2(a, b) ((a)[0] * (b)[0] + (a)[1] * (b)[1])
 
 /// Vector dot product
-#define VEC_DOT(a,b) ((a)[0]*(b)[0] + (a)[1]*(b)[1] + (a)[2]*(b)[2])
+#define VEC_DOT(a, b) ((a)[0] * (b)[0] + (a)[1] * (b)[1] + (a)[2] * (b)[2])
 
 /// Vector dot product
-#define VEC_DOT_4(a,b)	((a)[0]*(b)[0] + (a)[1]*(b)[1] + (a)[2]*(b)[2] + (a)[3]*(b)[3])
+#define VEC_DOT_4(a, b) ((a)[0] * (b)[0] + (a)[1] * (b)[1] + (a)[2] * (b)[2] + (a)[3] * (b)[3])
 
 /// vector impact parameter (squared)
-#define VEC_IMPACT_SQ(bsq,direction,position) {\
-   GREAL _llel_ = VEC_DOT(direction, position);\
-   bsq = VEC_DOT(position, position) - _llel_*_llel_;\
-}\
-
+#define VEC_IMPACT_SQ(bsq, direction, position)              \
+	{                                                        \
+		GREAL _llel_ = VEC_DOT(direction, position);         \
+		bsq = VEC_DOT(position, position) - _llel_ * _llel_; \
+	}
 
 /// vector impact parameter
-#define VEC_IMPACT(bsq,direction,position)	{\
-   VEC_IMPACT_SQ(bsq,direction,position);		\
-   GIM_SQRT(bsq,bsq);					\
-}\
+#define VEC_IMPACT(bsq, direction, position)     \
+	{                                            \
+		VEC_IMPACT_SQ(bsq, direction, position); \
+		GIM_SQRT(bsq, bsq);                      \
+	}
 
 /// Vector length
-#define VEC_LENGTH_2(a,l)\
-{\
-    GREAL _pp = VEC_DOT_2(a,a);\
-    GIM_SQRT(_pp,l);\
-}\
-
+#define VEC_LENGTH_2(a, l)           \
+	{                                \
+		GREAL _pp = VEC_DOT_2(a, a); \
+		GIM_SQRT(_pp, l);            \
+	}
 
 /// Vector length
-#define VEC_LENGTH(a,l)\
-{\
-    GREAL _pp = VEC_DOT(a,a);\
-    GIM_SQRT(_pp,l);\
-}\
-
+#define VEC_LENGTH(a, l)           \
+	{                              \
+		GREAL _pp = VEC_DOT(a, a); \
+		GIM_SQRT(_pp, l);          \
+	}
 
 /// Vector length
-#define VEC_LENGTH_4(a,l)\
-{\
-    GREAL _pp = VEC_DOT_4(a,a);\
-    GIM_SQRT(_pp,l);\
-}\
+#define VEC_LENGTH_4(a, l)           \
+	{                                \
+		GREAL _pp = VEC_DOT_4(a, a); \
+		GIM_SQRT(_pp, l);            \
+	}
 
 /// Vector inv length
-#define VEC_INV_LENGTH_2(a,l)\
-{\
-    GREAL _pp = VEC_DOT_2(a,a);\
-    GIM_INV_SQRT(_pp,l);\
-}\
-
+#define VEC_INV_LENGTH_2(a, l)       \
+	{                                \
+		GREAL _pp = VEC_DOT_2(a, a); \
+		GIM_INV_SQRT(_pp, l);        \
+	}
 
 /// Vector inv length
-#define VEC_INV_LENGTH(a,l)\
-{\
-    GREAL _pp = VEC_DOT(a,a);\
-    GIM_INV_SQRT(_pp,l);\
-}\
-
+#define VEC_INV_LENGTH(a, l)       \
+	{                              \
+		GREAL _pp = VEC_DOT(a, a); \
+		GIM_INV_SQRT(_pp, l);      \
+	}
 
 /// Vector inv length
-#define VEC_INV_LENGTH_4(a,l)\
-{\
-    GREAL _pp = VEC_DOT_4(a,a);\
-    GIM_INV_SQRT(_pp,l);\
-}\
-
-
+#define VEC_INV_LENGTH_4(a, l)       \
+	{                                \
+		GREAL _pp = VEC_DOT_4(a, a); \
+		GIM_INV_SQRT(_pp, l);        \
+	}
 
 /// distance between two points
-#define VEC_DISTANCE(_len,_va,_vb) {\
-    vec3f _tmp_;				\
-    VEC_DIFF(_tmp_, _vb, _va);			\
-    VEC_LENGTH(_tmp_,_len);			\
-}\
-
+#define VEC_DISTANCE(_len, _va, _vb) \
+	{                                \
+		vec3f _tmp_;                 \
+		VEC_DIFF(_tmp_, _vb, _va);   \
+		VEC_LENGTH(_tmp_, _len);     \
+	}
 
 /// Vector length
-#define VEC_CONJUGATE_LENGTH(a,l)\
-{\
-    GREAL _pp = 1.0 - a[0]*a[0] - a[1]*a[1] - a[2]*a[2];\
-    GIM_SQRT(_pp,l);\
-}\
-
+#define VEC_CONJUGATE_LENGTH(a, l)                                 \
+	{                                                              \
+		GREAL _pp = 1.0 - a[0] * a[0] - a[1] * a[1] - a[2] * a[2]; \
+		GIM_SQRT(_pp, l);                                          \
+	}
 
 /// Vector length
-#define VEC_NORMALIZE(a) {	\
-    GREAL len;\
-    VEC_INV_LENGTH(a,len); \
-    if(len<G_REAL_INFINITY)\
-    {\
-        a[0] *= len;				\
-        a[1] *= len;				\
-        a[2] *= len;				\
-    }						\
-}\
+#define VEC_NORMALIZE(a)           \
+	{                              \
+		GREAL len;                 \
+		VEC_INV_LENGTH(a, len);    \
+		if (len < G_REAL_INFINITY) \
+		{                          \
+			a[0] *= len;           \
+			a[1] *= len;           \
+			a[2] *= len;           \
+		}                          \
+	}
 
 /// Set Vector size
-#define VEC_RENORMALIZE(a,newlen) {	\
-    GREAL len;\
-    VEC_INV_LENGTH(a,len); \
-    if(len<G_REAL_INFINITY)\
-    {\
-        len *= newlen;\
-        a[0] *= len;				\
-        a[1] *= len;				\
-        a[2] *= len;				\
-    }						\
-}\
+#define VEC_RENORMALIZE(a, newlen) \
+	{                              \
+		GREAL len;                 \
+		VEC_INV_LENGTH(a, len);    \
+		if (len < G_REAL_INFINITY) \
+		{                          \
+			len *= newlen;         \
+			a[0] *= len;           \
+			a[1] *= len;           \
+			a[2] *= len;           \
+		}                          \
+	}
 
 /// Vector cross
-#define VEC_CROSS(c,a,b)		\
-{						\
-   c[0] = (a)[1] * (b)[2] - (a)[2] * (b)[1];	\
-   c[1] = (a)[2] * (b)[0] - (a)[0] * (b)[2];	\
-   c[2] = (a)[0] * (b)[1] - (a)[1] * (b)[0];	\
-}\
-
+#define VEC_CROSS(c, a, b)                        \
+	{                                             \
+		c[0] = (a)[1] * (b)[2] - (a)[2] * (b)[1]; \
+		c[1] = (a)[2] * (b)[0] - (a)[0] * (b)[2]; \
+		c[2] = (a)[0] * (b)[1] - (a)[1] * (b)[0]; \
+	}
 
 /*! Vector perp -- assumes that n is of unit length
  * accepts vector v, subtracts out any component parallel to n */
-#define VEC_PERPENDICULAR(vp,v,n)			\
-{						\
-   GREAL dot = VEC_DOT(v, n);			\
-   vp[0] = (v)[0] - dot*(n)[0];		\
-   vp[1] = (v)[1] - dot*(n)[1];		\
-   vp[2] = (v)[2] - dot*(n)[2];		\
-}\
-
+#define VEC_PERPENDICULAR(vp, v, n)    \
+	{                                  \
+		GREAL dot = VEC_DOT(v, n);     \
+		vp[0] = (v)[0] - dot * (n)[0]; \
+		vp[1] = (v)[1] - dot * (n)[1]; \
+		vp[2] = (v)[2] - dot * (n)[2]; \
+	}
 
 /*! Vector parallel -- assumes that n is of unit length */
-#define VEC_PARALLEL(vp,v,n)			\
-{						\
-   GREAL dot = VEC_DOT(v, n);			\
-   vp[0] = (dot) * (n)[0];			\
-   vp[1] = (dot) * (n)[1];			\
-   vp[2] = (dot) * (n)[2];			\
-}\
+#define VEC_PARALLEL(vp, v, n)     \
+	{                              \
+		GREAL dot = VEC_DOT(v, n); \
+		vp[0] = (dot) * (n)[0];    \
+		vp[1] = (dot) * (n)[1];    \
+		vp[2] = (dot) * (n)[2];    \
+	}
 
 /*! Same as Vector parallel --  n can have any length
  * accepts vector v, subtracts out any component perpendicular to n */
-#define VEC_PROJECT(vp,v,n)			\
-{ \
-	GREAL scalar = VEC_DOT(v, n);			\
-	scalar/= VEC_DOT(n, n); \
-	vp[0] = (scalar) * (n)[0];			\
-    vp[1] = (scalar) * (n)[1];			\
-    vp[2] = (scalar) * (n)[2];			\
-}\
-
+#define VEC_PROJECT(vp, v, n)         \
+	{                                 \
+		GREAL scalar = VEC_DOT(v, n); \
+		scalar /= VEC_DOT(n, n);      \
+		vp[0] = (scalar) * (n)[0];    \
+		vp[1] = (scalar) * (n)[1];    \
+		vp[2] = (scalar) * (n)[2];    \
+	}
 
 /*! accepts vector v*/
-#define VEC_UNPROJECT(vp,v,n)			\
-{ \
-	GREAL scalar = VEC_DOT(v, n);			\
-	scalar = VEC_DOT(n, n)/scalar; \
-	vp[0] = (scalar) * (n)[0];			\
-    vp[1] = (scalar) * (n)[1];			\
-    vp[2] = (scalar) * (n)[2];			\
-}\
-
+#define VEC_UNPROJECT(vp, v, n)          \
+	{                                    \
+		GREAL scalar = VEC_DOT(v, n);    \
+		scalar = VEC_DOT(n, n) / scalar; \
+		vp[0] = (scalar) * (n)[0];       \
+		vp[1] = (scalar) * (n)[1];       \
+		vp[2] = (scalar) * (n)[2];       \
+	}
 
 /*! Vector reflection -- assumes n is of unit length
  Takes vector v, reflects it against reflector n, and returns vr */
-#define VEC_REFLECT(vr,v,n)			\
-{						\
-   GREAL dot = VEC_DOT(v, n);			\
-   vr[0] = (v)[0] - 2.0 * (dot) * (n)[0];	\
-   vr[1] = (v)[1] - 2.0 * (dot) * (n)[1];	\
-   vr[2] = (v)[2] - 2.0 * (dot) * (n)[2];	\
-}\
-
+#define VEC_REFLECT(vr, v, n)                  \
+	{                                          \
+		GREAL dot = VEC_DOT(v, n);             \
+		vr[0] = (v)[0] - 2.0 * (dot) * (n)[0]; \
+		vr[1] = (v)[1] - 2.0 * (dot) * (n)[1]; \
+		vr[2] = (v)[2] - 2.0 * (dot) * (n)[2]; \
+	}
 
 /*! Vector blending
 Takes two vectors a, b, blends them together with two scalars */
-#define VEC_BLEND_AB(vr,sa,a,sb,b)			\
-{						\
-   vr[0] = (sa) * (a)[0] + (sb) * (b)[0];	\
-   vr[1] = (sa) * (a)[1] + (sb) * (b)[1];	\
-   vr[2] = (sa) * (a)[2] + (sb) * (b)[2];	\
-}\
+#define VEC_BLEND_AB(vr, sa, a, sb, b)         \
+	{                                          \
+		vr[0] = (sa) * (a)[0] + (sb) * (b)[0]; \
+		vr[1] = (sa) * (a)[1] + (sb) * (b)[1]; \
+		vr[2] = (sa) * (a)[2] + (sb) * (b)[2]; \
+	}
 
 /*! Vector blending
 Takes two vectors a, b, blends them together with s <=1 */
-#define VEC_BLEND(vr,a,b,s) VEC_BLEND_AB(vr,(1-s),a,s,b)
+#define VEC_BLEND(vr, a, b, s) VEC_BLEND_AB(vr, (1 - s), a, s, b)
 
-#define VEC_SET3(a,b,op,c) a[0]=b[0] op c[0]; a[1]=b[1] op c[1]; a[2]=b[2] op c[2];
+#define VEC_SET3(a, b, op, c) \
+	a[0] = b[0] op c[0];      \
+	a[1] = b[1] op c[1];      \
+	a[2] = b[2] op c[2];
 
 //! Finds the bigger cartesian coordinate from a vector
-#define VEC_MAYOR_COORD(vec, maxc)\
-{\
-	GREAL A[] = {fabs(vec[0]),fabs(vec[1]),fabs(vec[2])};\
-    maxc =  A[0]>A[1]?(A[0]>A[2]?0:2):(A[1]>A[2]?1:2);\
-}\
+#define VEC_MAYOR_COORD(vec, maxc)                                          \
+	{                                                                       \
+		GREAL A[] = {fabs(vec[0]), fabs(vec[1]), fabs(vec[2])};             \
+		maxc = A[0] > A[1] ? (A[0] > A[2] ? 0 : 2) : (A[1] > A[2] ? 1 : 2); \
+	}
 
 //! Finds the 2 smallest cartesian coordinates from a vector
-#define VEC_MINOR_AXES(vec, i0, i1)\
-{\
-	VEC_MAYOR_COORD(vec,i0);\
-	i0 = (i0+1)%3;\
-	i1 = (i0+1)%3;\
-}\
+#define VEC_MINOR_AXES(vec, i0, i1) \
+	{                               \
+		VEC_MAYOR_COORD(vec, i0);   \
+		i0 = (i0 + 1) % 3;          \
+		i1 = (i0 + 1) % 3;          \
+	}
 
+#define VEC_EQUAL(v1, v2) (v1[0] == v2[0] && v1[1] == v2[1] && v1[2] == v2[2])
 
-
-
-#define VEC_EQUAL(v1,v2) (v1[0]==v2[0]&&v1[1]==v2[1]&&v1[2]==v2[2])
-
-#define VEC_NEAR_EQUAL(v1,v2) (GIM_NEAR_EQUAL(v1[0],v2[0])&&GIM_NEAR_EQUAL(v1[1],v2[1])&&GIM_NEAR_EQUAL(v1[2],v2[2]))
-
+#define VEC_NEAR_EQUAL(v1, v2) (GIM_NEAR_EQUAL(v1[0], v2[0]) && GIM_NEAR_EQUAL(v1[1], v2[1]) && GIM_NEAR_EQUAL(v1[2], v2[2]))
 
 /// Vector cross
-#define X_AXIS_CROSS_VEC(dst,src)\
-{					   \
-	dst[0] = 0.0f;     \
-	dst[1] = -src[2];  \
-	dst[2] = src[1];  \
-}\
-
-#define Y_AXIS_CROSS_VEC(dst,src)\
-{					   \
-	dst[0] = src[2];     \
-	dst[1] = 0.0f;  \
-	dst[2] = -src[0];  \
-}\
-
-#define Z_AXIS_CROSS_VEC(dst,src)\
-{					   \
-	dst[0] = -src[1];     \
-	dst[1] = src[0];  \
-	dst[2] = 0.0f;  \
-}\
-
-
-
+#define X_AXIS_CROSS_VEC(dst, src) \
+	{                              \
+		dst[0] = 0.0f;             \
+		dst[1] = -src[2];          \
+		dst[2] = src[1];           \
+	}
 
+#define Y_AXIS_CROSS_VEC(dst, src) \
+	{                              \
+		dst[0] = src[2];           \
+		dst[1] = 0.0f;             \
+		dst[2] = -src[0];          \
+	}
 
+#define Z_AXIS_CROSS_VEC(dst, src) \
+	{                              \
+		dst[0] = -src[1];          \
+		dst[1] = src[0];           \
+		dst[2] = 0.0f;             \
+	}
 
 /// initialize matrix
-#define IDENTIFY_MATRIX_3X3(m)			\
-{						\
-   m[0][0] = 1.0;				\
-   m[0][1] = 0.0;				\
-   m[0][2] = 0.0;				\
-						\
-   m[1][0] = 0.0;				\
-   m[1][1] = 1.0;				\
-   m[1][2] = 0.0;				\
-						\
-   m[2][0] = 0.0;				\
-   m[2][1] = 0.0;				\
-   m[2][2] = 1.0;				\
-}\
+#define IDENTIFY_MATRIX_3X3(m) \
+	{                          \
+		m[0][0] = 1.0;         \
+		m[0][1] = 0.0;         \
+		m[0][2] = 0.0;         \
+                               \
+		m[1][0] = 0.0;         \
+		m[1][1] = 1.0;         \
+		m[1][2] = 0.0;         \
+                               \
+		m[2][0] = 0.0;         \
+		m[2][1] = 0.0;         \
+		m[2][2] = 1.0;         \
+	}
 
 /*! initialize matrix */
-#define IDENTIFY_MATRIX_4X4(m)			\
-{						\
-   m[0][0] = 1.0;				\
-   m[0][1] = 0.0;				\
-   m[0][2] = 0.0;				\
-   m[0][3] = 0.0;				\
-						\
-   m[1][0] = 0.0;				\
-   m[1][1] = 1.0;				\
-   m[1][2] = 0.0;				\
-   m[1][3] = 0.0;				\
-						\
-   m[2][0] = 0.0;				\
-   m[2][1] = 0.0;				\
-   m[2][2] = 1.0;				\
-   m[2][3] = 0.0;				\
-						\
-   m[3][0] = 0.0;				\
-   m[3][1] = 0.0;				\
-   m[3][2] = 0.0;				\
-   m[3][3] = 1.0;				\
-}\
+#define IDENTIFY_MATRIX_4X4(m) \
+	{                          \
+		m[0][0] = 1.0;         \
+		m[0][1] = 0.0;         \
+		m[0][2] = 0.0;         \
+		m[0][3] = 0.0;         \
+                               \
+		m[1][0] = 0.0;         \
+		m[1][1] = 1.0;         \
+		m[1][2] = 0.0;         \
+		m[1][3] = 0.0;         \
+                               \
+		m[2][0] = 0.0;         \
+		m[2][1] = 0.0;         \
+		m[2][2] = 1.0;         \
+		m[2][3] = 0.0;         \
+                               \
+		m[3][0] = 0.0;         \
+		m[3][1] = 0.0;         \
+		m[3][2] = 0.0;         \
+		m[3][3] = 1.0;         \
+	}
 
 /*! initialize matrix */
-#define ZERO_MATRIX_4X4(m)			\
-{						\
-   m[0][0] = 0.0;				\
-   m[0][1] = 0.0;				\
-   m[0][2] = 0.0;				\
-   m[0][3] = 0.0;				\
-						\
-   m[1][0] = 0.0;				\
-   m[1][1] = 0.0;				\
-   m[1][2] = 0.0;				\
-   m[1][3] = 0.0;				\
-						\
-   m[2][0] = 0.0;				\
-   m[2][1] = 0.0;				\
-   m[2][2] = 0.0;				\
-   m[2][3] = 0.0;				\
-						\
-   m[3][0] = 0.0;				\
-   m[3][1] = 0.0;				\
-   m[3][2] = 0.0;				\
-   m[3][3] = 0.0;				\
-}\
+#define ZERO_MATRIX_4X4(m) \
+	{                      \
+		m[0][0] = 0.0;     \
+		m[0][1] = 0.0;     \
+		m[0][2] = 0.0;     \
+		m[0][3] = 0.0;     \
+                           \
+		m[1][0] = 0.0;     \
+		m[1][1] = 0.0;     \
+		m[1][2] = 0.0;     \
+		m[1][3] = 0.0;     \
+                           \
+		m[2][0] = 0.0;     \
+		m[2][1] = 0.0;     \
+		m[2][2] = 0.0;     \
+		m[2][3] = 0.0;     \
+                           \
+		m[3][0] = 0.0;     \
+		m[3][1] = 0.0;     \
+		m[3][2] = 0.0;     \
+		m[3][3] = 0.0;     \
+	}
 
 /*! matrix rotation  X */
-#define ROTX_CS(m,cosine,sine)		\
-{					\
-   /* rotation about the x-axis */	\
-					\
-   m[0][0] = 1.0;			\
-   m[0][1] = 0.0;			\
-   m[0][2] = 0.0;			\
-   m[0][3] = 0.0;			\
-					\
-   m[1][0] = 0.0;			\
-   m[1][1] = (cosine);			\
-   m[1][2] = (sine);			\
-   m[1][3] = 0.0;			\
-					\
-   m[2][0] = 0.0;			\
-   m[2][1] = -(sine);			\
-   m[2][2] = (cosine);			\
-   m[2][3] = 0.0;			\
-					\
-   m[3][0] = 0.0;			\
-   m[3][1] = 0.0;			\
-   m[3][2] = 0.0;			\
-   m[3][3] = 1.0;			\
-}\
+#define ROTX_CS(m, cosine, sine)        \
+	{                                   \
+		/* rotation about the x-axis */ \
+                                        \
+		m[0][0] = 1.0;                  \
+		m[0][1] = 0.0;                  \
+		m[0][2] = 0.0;                  \
+		m[0][3] = 0.0;                  \
+                                        \
+		m[1][0] = 0.0;                  \
+		m[1][1] = (cosine);             \
+		m[1][2] = (sine);               \
+		m[1][3] = 0.0;                  \
+                                        \
+		m[2][0] = 0.0;                  \
+		m[2][1] = -(sine);              \
+		m[2][2] = (cosine);             \
+		m[2][3] = 0.0;                  \
+                                        \
+		m[3][0] = 0.0;                  \
+		m[3][1] = 0.0;                  \
+		m[3][2] = 0.0;                  \
+		m[3][3] = 1.0;                  \
+	}
 
 /*! matrix rotation  Y */
-#define ROTY_CS(m,cosine,sine)		\
-{					\
-   /* rotation about the y-axis */	\
-					\
-   m[0][0] = (cosine);			\
-   m[0][1] = 0.0;			\
-   m[0][2] = -(sine);			\
-   m[0][3] = 0.0;			\
-					\
-   m[1][0] = 0.0;			\
-   m[1][1] = 1.0;			\
-   m[1][2] = 0.0;			\
-   m[1][3] = 0.0;			\
-					\
-   m[2][0] = (sine);			\
-   m[2][1] = 0.0;			\
-   m[2][2] = (cosine);			\
-   m[2][3] = 0.0;			\
-					\
-   m[3][0] = 0.0;			\
-   m[3][1] = 0.0;			\
-   m[3][2] = 0.0;			\
-   m[3][3] = 1.0;			\
-}\
+#define ROTY_CS(m, cosine, sine)        \
+	{                                   \
+		/* rotation about the y-axis */ \
+                                        \
+		m[0][0] = (cosine);             \
+		m[0][1] = 0.0;                  \
+		m[0][2] = -(sine);              \
+		m[0][3] = 0.0;                  \
+                                        \
+		m[1][0] = 0.0;                  \
+		m[1][1] = 1.0;                  \
+		m[1][2] = 0.0;                  \
+		m[1][3] = 0.0;                  \
+                                        \
+		m[2][0] = (sine);               \
+		m[2][1] = 0.0;                  \
+		m[2][2] = (cosine);             \
+		m[2][3] = 0.0;                  \
+                                        \
+		m[3][0] = 0.0;                  \
+		m[3][1] = 0.0;                  \
+		m[3][2] = 0.0;                  \
+		m[3][3] = 1.0;                  \
+	}
 
 /*! matrix rotation  Z */
-#define ROTZ_CS(m,cosine,sine)		\
-{					\
-   /* rotation about the z-axis */	\
-					\
-   m[0][0] = (cosine);			\
-   m[0][1] = (sine);			\
-   m[0][2] = 0.0;			\
-   m[0][3] = 0.0;			\
-					\
-   m[1][0] = -(sine);			\
-   m[1][1] = (cosine);			\
-   m[1][2] = 0.0;			\
-   m[1][3] = 0.0;			\
-					\
-   m[2][0] = 0.0;			\
-   m[2][1] = 0.0;			\
-   m[2][2] = 1.0;			\
-   m[2][3] = 0.0;			\
-					\
-   m[3][0] = 0.0;			\
-   m[3][1] = 0.0;			\
-   m[3][2] = 0.0;			\
-   m[3][3] = 1.0;			\
-}\
+#define ROTZ_CS(m, cosine, sine)        \
+	{                                   \
+		/* rotation about the z-axis */ \
+                                        \
+		m[0][0] = (cosine);             \
+		m[0][1] = (sine);               \
+		m[0][2] = 0.0;                  \
+		m[0][3] = 0.0;                  \
+                                        \
+		m[1][0] = -(sine);              \
+		m[1][1] = (cosine);             \
+		m[1][2] = 0.0;                  \
+		m[1][3] = 0.0;                  \
+                                        \
+		m[2][0] = 0.0;                  \
+		m[2][1] = 0.0;                  \
+		m[2][2] = 1.0;                  \
+		m[2][3] = 0.0;                  \
+                                        \
+		m[3][0] = 0.0;                  \
+		m[3][1] = 0.0;                  \
+		m[3][2] = 0.0;                  \
+		m[3][3] = 1.0;                  \
+	}
 
 /*! matrix copy */
-#define COPY_MATRIX_2X2(b,a)	\
-{				\
-   b[0][0] = a[0][0];		\
-   b[0][1] = a[0][1];		\
-				\
-   b[1][0] = a[1][0];		\
-   b[1][1] = a[1][1];		\
-				\
-}\
-
+#define COPY_MATRIX_2X2(b, a) \
+	{                         \
+		b[0][0] = a[0][0];    \
+		b[0][1] = a[0][1];    \
+                              \
+		b[1][0] = a[1][0];    \
+		b[1][1] = a[1][1];    \
+	}
 
 /*! matrix copy */
-#define COPY_MATRIX_2X3(b,a)	\
-{				\
-   b[0][0] = a[0][0];		\
-   b[0][1] = a[0][1];		\
-   b[0][2] = a[0][2];		\
-				\
-   b[1][0] = a[1][0];		\
-   b[1][1] = a[1][1];		\
-   b[1][2] = a[1][2];		\
-}\
-
+#define COPY_MATRIX_2X3(b, a) \
+	{                         \
+		b[0][0] = a[0][0];    \
+		b[0][1] = a[0][1];    \
+		b[0][2] = a[0][2];    \
+                              \
+		b[1][0] = a[1][0];    \
+		b[1][1] = a[1][1];    \
+		b[1][2] = a[1][2];    \
+	}
 
 /*! matrix copy */
-#define COPY_MATRIX_3X3(b,a)	\
-{				\
-   b[0][0] = a[0][0];		\
-   b[0][1] = a[0][1];		\
-   b[0][2] = a[0][2];		\
-				\
-   b[1][0] = a[1][0];		\
-   b[1][1] = a[1][1];		\
-   b[1][2] = a[1][2];		\
-				\
-   b[2][0] = a[2][0];		\
-   b[2][1] = a[2][1];		\
-   b[2][2] = a[2][2];		\
-}\
-
+#define COPY_MATRIX_3X3(b, a) \
+	{                         \
+		b[0][0] = a[0][0];    \
+		b[0][1] = a[0][1];    \
+		b[0][2] = a[0][2];    \
+                              \
+		b[1][0] = a[1][0];    \
+		b[1][1] = a[1][1];    \
+		b[1][2] = a[1][2];    \
+                              \
+		b[2][0] = a[2][0];    \
+		b[2][1] = a[2][1];    \
+		b[2][2] = a[2][2];    \
+	}
 
 /*! matrix copy */
-#define COPY_MATRIX_4X4(b,a)	\
-{				\
-   b[0][0] = a[0][0];		\
-   b[0][1] = a[0][1];		\
-   b[0][2] = a[0][2];		\
-   b[0][3] = a[0][3];		\
-				\
-   b[1][0] = a[1][0];		\
-   b[1][1] = a[1][1];		\
-   b[1][2] = a[1][2];		\
-   b[1][3] = a[1][3];		\
-				\
-   b[2][0] = a[2][0];		\
-   b[2][1] = a[2][1];		\
-   b[2][2] = a[2][2];		\
-   b[2][3] = a[2][3];		\
-				\
-   b[3][0] = a[3][0];		\
-   b[3][1] = a[3][1];		\
-   b[3][2] = a[3][2];		\
-   b[3][3] = a[3][3];		\
-}\
-
+#define COPY_MATRIX_4X4(b, a) \
+	{                         \
+		b[0][0] = a[0][0];    \
+		b[0][1] = a[0][1];    \
+		b[0][2] = a[0][2];    \
+		b[0][3] = a[0][3];    \
+                              \
+		b[1][0] = a[1][0];    \
+		b[1][1] = a[1][1];    \
+		b[1][2] = a[1][2];    \
+		b[1][3] = a[1][3];    \
+                              \
+		b[2][0] = a[2][0];    \
+		b[2][1] = a[2][1];    \
+		b[2][2] = a[2][2];    \
+		b[2][3] = a[2][3];    \
+                              \
+		b[3][0] = a[3][0];    \
+		b[3][1] = a[3][1];    \
+		b[3][2] = a[3][2];    \
+		b[3][3] = a[3][3];    \
+	}
 
 /*! matrix transpose */
-#define TRANSPOSE_MATRIX_2X2(b,a)	\
-{				\
-   b[0][0] = a[0][0];		\
-   b[0][1] = a[1][0];		\
-				\
-   b[1][0] = a[0][1];		\
-   b[1][1] = a[1][1];		\
-}\
-
+#define TRANSPOSE_MATRIX_2X2(b, a) \
+	{                              \
+		b[0][0] = a[0][0];         \
+		b[0][1] = a[1][0];         \
+                                   \
+		b[1][0] = a[0][1];         \
+		b[1][1] = a[1][1];         \
+	}
 
 /*! matrix transpose */
-#define TRANSPOSE_MATRIX_3X3(b,a)	\
-{				\
-   b[0][0] = a[0][0];		\
-   b[0][1] = a[1][0];		\
-   b[0][2] = a[2][0];		\
-				\
-   b[1][0] = a[0][1];		\
-   b[1][1] = a[1][1];		\
-   b[1][2] = a[2][1];		\
-				\
-   b[2][0] = a[0][2];		\
-   b[2][1] = a[1][2];		\
-   b[2][2] = a[2][2];		\
-}\
-
+#define TRANSPOSE_MATRIX_3X3(b, a) \
+	{                              \
+		b[0][0] = a[0][0];         \
+		b[0][1] = a[1][0];         \
+		b[0][2] = a[2][0];         \
+                                   \
+		b[1][0] = a[0][1];         \
+		b[1][1] = a[1][1];         \
+		b[1][2] = a[2][1];         \
+                                   \
+		b[2][0] = a[0][2];         \
+		b[2][1] = a[1][2];         \
+		b[2][2] = a[2][2];         \
+	}
 
 /*! matrix transpose */
-#define TRANSPOSE_MATRIX_4X4(b,a)	\
-{				\
-   b[0][0] = a[0][0];		\
-   b[0][1] = a[1][0];		\
-   b[0][2] = a[2][0];		\
-   b[0][3] = a[3][0];		\
-				\
-   b[1][0] = a[0][1];		\
-   b[1][1] = a[1][1];		\
-   b[1][2] = a[2][1];		\
-   b[1][3] = a[3][1];		\
-				\
-   b[2][0] = a[0][2];		\
-   b[2][1] = a[1][2];		\
-   b[2][2] = a[2][2];		\
-   b[2][3] = a[3][2];		\
-				\
-   b[3][0] = a[0][3];		\
-   b[3][1] = a[1][3];		\
-   b[3][2] = a[2][3];		\
-   b[3][3] = a[3][3];		\
-}\
-
+#define TRANSPOSE_MATRIX_4X4(b, a) \
+	{                              \
+		b[0][0] = a[0][0];         \
+		b[0][1] = a[1][0];         \
+		b[0][2] = a[2][0];         \
+		b[0][3] = a[3][0];         \
+                                   \
+		b[1][0] = a[0][1];         \
+		b[1][1] = a[1][1];         \
+		b[1][2] = a[2][1];         \
+		b[1][3] = a[3][1];         \
+                                   \
+		b[2][0] = a[0][2];         \
+		b[2][1] = a[1][2];         \
+		b[2][2] = a[2][2];         \
+		b[2][3] = a[3][2];         \
+                                   \
+		b[3][0] = a[0][3];         \
+		b[3][1] = a[1][3];         \
+		b[3][2] = a[2][3];         \
+		b[3][3] = a[3][3];         \
+	}
 
 /*! multiply matrix by scalar */
-#define SCALE_MATRIX_2X2(b,s,a)		\
-{					\
-   b[0][0] = (s) * a[0][0];		\
-   b[0][1] = (s) * a[0][1];		\
-					\
-   b[1][0] = (s) * a[1][0];		\
-   b[1][1] = (s) * a[1][1];		\
-}\
-
+#define SCALE_MATRIX_2X2(b, s, a) \
+	{                             \
+		b[0][0] = (s)*a[0][0];    \
+		b[0][1] = (s)*a[0][1];    \
+                                  \
+		b[1][0] = (s)*a[1][0];    \
+		b[1][1] = (s)*a[1][1];    \
+	}
 
 /*! multiply matrix by scalar */
-#define SCALE_MATRIX_3X3(b,s,a)		\
-{					\
-   b[0][0] = (s) * a[0][0];		\
-   b[0][1] = (s) * a[0][1];		\
-   b[0][2] = (s) * a[0][2];		\
-					\
-   b[1][0] = (s) * a[1][0];		\
-   b[1][1] = (s) * a[1][1];		\
-   b[1][2] = (s) * a[1][2];		\
-					\
-   b[2][0] = (s) * a[2][0];		\
-   b[2][1] = (s) * a[2][1];		\
-   b[2][2] = (s) * a[2][2];		\
-}\
-
+#define SCALE_MATRIX_3X3(b, s, a) \
+	{                             \
+		b[0][0] = (s)*a[0][0];    \
+		b[0][1] = (s)*a[0][1];    \
+		b[0][2] = (s)*a[0][2];    \
+                                  \
+		b[1][0] = (s)*a[1][0];    \
+		b[1][1] = (s)*a[1][1];    \
+		b[1][2] = (s)*a[1][2];    \
+                                  \
+		b[2][0] = (s)*a[2][0];    \
+		b[2][1] = (s)*a[2][1];    \
+		b[2][2] = (s)*a[2][2];    \
+	}
 
 /*! multiply matrix by scalar */
-#define SCALE_MATRIX_4X4(b,s,a)		\
-{					\
-   b[0][0] = (s) * a[0][0];		\
-   b[0][1] = (s) * a[0][1];		\
-   b[0][2] = (s) * a[0][2];		\
-   b[0][3] = (s) * a[0][3];		\
-					\
-   b[1][0] = (s) * a[1][0];		\
-   b[1][1] = (s) * a[1][1];		\
-   b[1][2] = (s) * a[1][2];		\
-   b[1][3] = (s) * a[1][3];		\
-					\
-   b[2][0] = (s) * a[2][0];		\
-   b[2][1] = (s) * a[2][1];		\
-   b[2][2] = (s) * a[2][2];		\
-   b[2][3] = (s) * a[2][3];		\
-					\
-   b[3][0] = s * a[3][0];		\
-   b[3][1] = s * a[3][1];		\
-   b[3][2] = s * a[3][2];		\
-   b[3][3] = s * a[3][3];		\
-}\
-
+#define SCALE_MATRIX_4X4(b, s, a) \
+	{                             \
+		b[0][0] = (s)*a[0][0];    \
+		b[0][1] = (s)*a[0][1];    \
+		b[0][2] = (s)*a[0][2];    \
+		b[0][3] = (s)*a[0][3];    \
+                                  \
+		b[1][0] = (s)*a[1][0];    \
+		b[1][1] = (s)*a[1][1];    \
+		b[1][2] = (s)*a[1][2];    \
+		b[1][3] = (s)*a[1][3];    \
+                                  \
+		b[2][0] = (s)*a[2][0];    \
+		b[2][1] = (s)*a[2][1];    \
+		b[2][2] = (s)*a[2][2];    \
+		b[2][3] = (s)*a[2][3];    \
+                                  \
+		b[3][0] = s * a[3][0];    \
+		b[3][1] = s * a[3][1];    \
+		b[3][2] = s * a[3][2];    \
+		b[3][3] = s * a[3][3];    \
+	}
 
 /*! multiply matrix by scalar */
-#define SCALE_VEC_MATRIX_2X2(b,svec,a)		\
-{					\
-   b[0][0] = svec[0] * a[0][0];		\
-   b[1][0] = svec[0] * a[1][0];		\
-					\
-   b[0][1] = svec[1] * a[0][1];		\
-   b[1][1] = svec[1] * a[1][1];		\
-}\
-
+#define SCALE_VEC_MATRIX_2X2(b, svec, a) \
+	{                                    \
+		b[0][0] = svec[0] * a[0][0];     \
+		b[1][0] = svec[0] * a[1][0];     \
+                                         \
+		b[0][1] = svec[1] * a[0][1];     \
+		b[1][1] = svec[1] * a[1][1];     \
+	}
 
 /*! multiply matrix by scalar. Each columns is scaled by each scalar vector component */
-#define SCALE_VEC_MATRIX_3X3(b,svec,a)		\
-{					\
-   b[0][0] = svec[0] * a[0][0];		\
-   b[1][0] = svec[0] * a[1][0];		\
-   b[2][0] = svec[0] * a[2][0];		\
-					\
-   b[0][1] = svec[1] * a[0][1];		\
-   b[1][1] = svec[1] * a[1][1];		\
-   b[2][1] = svec[1] * a[2][1];		\
-					\
-   b[0][2] = svec[2] * a[0][2];		\
-   b[1][2] = svec[2] * a[1][2];		\
-   b[2][2] = svec[2] * a[2][2];		\
-}\
-
+#define SCALE_VEC_MATRIX_3X3(b, svec, a) \
+	{                                    \
+		b[0][0] = svec[0] * a[0][0];     \
+		b[1][0] = svec[0] * a[1][0];     \
+		b[2][0] = svec[0] * a[2][0];     \
+                                         \
+		b[0][1] = svec[1] * a[0][1];     \
+		b[1][1] = svec[1] * a[1][1];     \
+		b[2][1] = svec[1] * a[2][1];     \
+                                         \
+		b[0][2] = svec[2] * a[0][2];     \
+		b[1][2] = svec[2] * a[1][2];     \
+		b[2][2] = svec[2] * a[2][2];     \
+	}
 
 /*! multiply matrix by scalar */
-#define SCALE_VEC_MATRIX_4X4(b,svec,a)		\
-{					\
-   b[0][0] = svec[0] * a[0][0];		\
-   b[1][0] = svec[0] * a[1][0];		\
-   b[2][0] = svec[0] * a[2][0];		\
-   b[3][0] = svec[0] * a[3][0];		\
-					\
-   b[0][1] = svec[1] * a[0][1];		\
-   b[1][1] = svec[1] * a[1][1];		\
-   b[2][1] = svec[1] * a[2][1];		\
-   b[3][1] = svec[1] * a[3][1];		\
-					\
-   b[0][2] = svec[2] * a[0][2];		\
-   b[1][2] = svec[2] * a[1][2];		\
-   b[2][2] = svec[2] * a[2][2];		\
-   b[3][2] = svec[2] * a[3][2];		\
-   \
-   b[0][3] = svec[3] * a[0][3];		\
-   b[1][3] = svec[3] * a[1][3];		\
-   b[2][3] = svec[3] * a[2][3];		\
-   b[3][3] = svec[3] * a[3][3];		\
-}\
-
+#define SCALE_VEC_MATRIX_4X4(b, svec, a) \
+	{                                    \
+		b[0][0] = svec[0] * a[0][0];     \
+		b[1][0] = svec[0] * a[1][0];     \
+		b[2][0] = svec[0] * a[2][0];     \
+		b[3][0] = svec[0] * a[3][0];     \
+                                         \
+		b[0][1] = svec[1] * a[0][1];     \
+		b[1][1] = svec[1] * a[1][1];     \
+		b[2][1] = svec[1] * a[2][1];     \
+		b[3][1] = svec[1] * a[3][1];     \
+                                         \
+		b[0][2] = svec[2] * a[0][2];     \
+		b[1][2] = svec[2] * a[1][2];     \
+		b[2][2] = svec[2] * a[2][2];     \
+		b[3][2] = svec[2] * a[3][2];     \
+                                         \
+		b[0][3] = svec[3] * a[0][3];     \
+		b[1][3] = svec[3] * a[1][3];     \
+		b[2][3] = svec[3] * a[2][3];     \
+		b[3][3] = svec[3] * a[3][3];     \
+	}
 
 /*! multiply matrix by scalar */
-#define ACCUM_SCALE_MATRIX_2X2(b,s,a)		\
-{					\
-   b[0][0] += (s) * a[0][0];		\
-   b[0][1] += (s) * a[0][1];		\
-					\
-   b[1][0] += (s) * a[1][0];		\
-   b[1][1] += (s) * a[1][1];		\
-}\
-
+#define ACCUM_SCALE_MATRIX_2X2(b, s, a) \
+	{                                   \
+		b[0][0] += (s)*a[0][0];         \
+		b[0][1] += (s)*a[0][1];         \
+                                        \
+		b[1][0] += (s)*a[1][0];         \
+		b[1][1] += (s)*a[1][1];         \
+	}
 
 /*! multiply matrix by scalar */
-#define ACCUM_SCALE_MATRIX_3X3(b,s,a)		\
-{					\
-   b[0][0] += (s) * a[0][0];		\
-   b[0][1] += (s) * a[0][1];		\
-   b[0][2] += (s) * a[0][2];		\
-					\
-   b[1][0] += (s) * a[1][0];		\
-   b[1][1] += (s) * a[1][1];		\
-   b[1][2] += (s) * a[1][2];		\
-					\
-   b[2][0] += (s) * a[2][0];		\
-   b[2][1] += (s) * a[2][1];		\
-   b[2][2] += (s) * a[2][2];		\
-}\
-
+#define ACCUM_SCALE_MATRIX_3X3(b, s, a) \
+	{                                   \
+		b[0][0] += (s)*a[0][0];         \
+		b[0][1] += (s)*a[0][1];         \
+		b[0][2] += (s)*a[0][2];         \
+                                        \
+		b[1][0] += (s)*a[1][0];         \
+		b[1][1] += (s)*a[1][1];         \
+		b[1][2] += (s)*a[1][2];         \
+                                        \
+		b[2][0] += (s)*a[2][0];         \
+		b[2][1] += (s)*a[2][1];         \
+		b[2][2] += (s)*a[2][2];         \
+	}
 
 /*! multiply matrix by scalar */
-#define ACCUM_SCALE_MATRIX_4X4(b,s,a)		\
-{					\
-   b[0][0] += (s) * a[0][0];		\
-   b[0][1] += (s) * a[0][1];		\
-   b[0][2] += (s) * a[0][2];		\
-   b[0][3] += (s) * a[0][3];		\
-					\
-   b[1][0] += (s) * a[1][0];		\
-   b[1][1] += (s) * a[1][1];		\
-   b[1][2] += (s) * a[1][2];		\
-   b[1][3] += (s) * a[1][3];		\
-					\
-   b[2][0] += (s) * a[2][0];		\
-   b[2][1] += (s) * a[2][1];		\
-   b[2][2] += (s) * a[2][2];		\
-   b[2][3] += (s) * a[2][3];		\
-					\
-   b[3][0] += (s) * a[3][0];		\
-   b[3][1] += (s) * a[3][1];		\
-   b[3][2] += (s) * a[3][2];		\
-   b[3][3] += (s) * a[3][3];		\
-}\
+#define ACCUM_SCALE_MATRIX_4X4(b, s, a) \
+	{                                   \
+		b[0][0] += (s)*a[0][0];         \
+		b[0][1] += (s)*a[0][1];         \
+		b[0][2] += (s)*a[0][2];         \
+		b[0][3] += (s)*a[0][3];         \
+                                        \
+		b[1][0] += (s)*a[1][0];         \
+		b[1][1] += (s)*a[1][1];         \
+		b[1][2] += (s)*a[1][2];         \
+		b[1][3] += (s)*a[1][3];         \
+                                        \
+		b[2][0] += (s)*a[2][0];         \
+		b[2][1] += (s)*a[2][1];         \
+		b[2][2] += (s)*a[2][2];         \
+		b[2][3] += (s)*a[2][3];         \
+                                        \
+		b[3][0] += (s)*a[3][0];         \
+		b[3][1] += (s)*a[3][1];         \
+		b[3][2] += (s)*a[3][2];         \
+		b[3][3] += (s)*a[3][3];         \
+	}
 
 /*! matrix product */
 /*! c[x][y] = a[x][0]*b[0][y]+a[x][1]*b[1][y]+a[x][2]*b[2][y]+a[x][3]*b[3][y];*/
-#define MATRIX_PRODUCT_2X2(c,a,b)		\
-{						\
-   c[0][0] = a[0][0]*b[0][0]+a[0][1]*b[1][0];	\
-   c[0][1] = a[0][0]*b[0][1]+a[0][1]*b[1][1];	\
-						\
-   c[1][0] = a[1][0]*b[0][0]+a[1][1]*b[1][0];	\
-   c[1][1] = a[1][0]*b[0][1]+a[1][1]*b[1][1];	\
-						\
-}\
+#define MATRIX_PRODUCT_2X2(c, a, b)                      \
+	{                                                    \
+		c[0][0] = a[0][0] * b[0][0] + a[0][1] * b[1][0]; \
+		c[0][1] = a[0][0] * b[0][1] + a[0][1] * b[1][1]; \
+                                                         \
+		c[1][0] = a[1][0] * b[0][0] + a[1][1] * b[1][0]; \
+		c[1][1] = a[1][0] * b[0][1] + a[1][1] * b[1][1]; \
+	}
 
 /*! matrix product */
 /*! c[x][y] = a[x][0]*b[0][y]+a[x][1]*b[1][y]+a[x][2]*b[2][y]+a[x][3]*b[3][y];*/
-#define MATRIX_PRODUCT_3X3(c,a,b)				\
-{								\
-   c[0][0] = a[0][0]*b[0][0]+a[0][1]*b[1][0]+a[0][2]*b[2][0];	\
-   c[0][1] = a[0][0]*b[0][1]+a[0][1]*b[1][1]+a[0][2]*b[2][1];	\
-   c[0][2] = a[0][0]*b[0][2]+a[0][1]*b[1][2]+a[0][2]*b[2][2];	\
-								\
-   c[1][0] = a[1][0]*b[0][0]+a[1][1]*b[1][0]+a[1][2]*b[2][0];	\
-   c[1][1] = a[1][0]*b[0][1]+a[1][1]*b[1][1]+a[1][2]*b[2][1];	\
-   c[1][2] = a[1][0]*b[0][2]+a[1][1]*b[1][2]+a[1][2]*b[2][2];	\
-								\
-   c[2][0] = a[2][0]*b[0][0]+a[2][1]*b[1][0]+a[2][2]*b[2][0];	\
-   c[2][1] = a[2][0]*b[0][1]+a[2][1]*b[1][1]+a[2][2]*b[2][1];	\
-   c[2][2] = a[2][0]*b[0][2]+a[2][1]*b[1][2]+a[2][2]*b[2][2];	\
-}\
-
+#define MATRIX_PRODUCT_3X3(c, a, b)                                          \
+	{                                                                        \
+		c[0][0] = a[0][0] * b[0][0] + a[0][1] * b[1][0] + a[0][2] * b[2][0]; \
+		c[0][1] = a[0][0] * b[0][1] + a[0][1] * b[1][1] + a[0][2] * b[2][1]; \
+		c[0][2] = a[0][0] * b[0][2] + a[0][1] * b[1][2] + a[0][2] * b[2][2]; \
+                                                                             \
+		c[1][0] = a[1][0] * b[0][0] + a[1][1] * b[1][0] + a[1][2] * b[2][0]; \
+		c[1][1] = a[1][0] * b[0][1] + a[1][1] * b[1][1] + a[1][2] * b[2][1]; \
+		c[1][2] = a[1][0] * b[0][2] + a[1][1] * b[1][2] + a[1][2] * b[2][2]; \
+                                                                             \
+		c[2][0] = a[2][0] * b[0][0] + a[2][1] * b[1][0] + a[2][2] * b[2][0]; \
+		c[2][1] = a[2][0] * b[0][1] + a[2][1] * b[1][1] + a[2][2] * b[2][1]; \
+		c[2][2] = a[2][0] * b[0][2] + a[2][1] * b[1][2] + a[2][2] * b[2][2]; \
+	}
 
 /*! matrix product */
 /*! c[x][y] = a[x][0]*b[0][y]+a[x][1]*b[1][y]+a[x][2]*b[2][y]+a[x][3]*b[3][y];*/
-#define MATRIX_PRODUCT_4X4(c,a,b)		\
-{						\
-   c[0][0] = a[0][0]*b[0][0]+a[0][1]*b[1][0]+a[0][2]*b[2][0]+a[0][3]*b[3][0];\
-   c[0][1] = a[0][0]*b[0][1]+a[0][1]*b[1][1]+a[0][2]*b[2][1]+a[0][3]*b[3][1];\
-   c[0][2] = a[0][0]*b[0][2]+a[0][1]*b[1][2]+a[0][2]*b[2][2]+a[0][3]*b[3][2];\
-   c[0][3] = a[0][0]*b[0][3]+a[0][1]*b[1][3]+a[0][2]*b[2][3]+a[0][3]*b[3][3];\
-						\
-   c[1][0] = a[1][0]*b[0][0]+a[1][1]*b[1][0]+a[1][2]*b[2][0]+a[1][3]*b[3][0];\
-   c[1][1] = a[1][0]*b[0][1]+a[1][1]*b[1][1]+a[1][2]*b[2][1]+a[1][3]*b[3][1];\
-   c[1][2] = a[1][0]*b[0][2]+a[1][1]*b[1][2]+a[1][2]*b[2][2]+a[1][3]*b[3][2];\
-   c[1][3] = a[1][0]*b[0][3]+a[1][1]*b[1][3]+a[1][2]*b[2][3]+a[1][3]*b[3][3];\
-						\
-   c[2][0] = a[2][0]*b[0][0]+a[2][1]*b[1][0]+a[2][2]*b[2][0]+a[2][3]*b[3][0];\
-   c[2][1] = a[2][0]*b[0][1]+a[2][1]*b[1][1]+a[2][2]*b[2][1]+a[2][3]*b[3][1];\
-   c[2][2] = a[2][0]*b[0][2]+a[2][1]*b[1][2]+a[2][2]*b[2][2]+a[2][3]*b[3][2];\
-   c[2][3] = a[2][0]*b[0][3]+a[2][1]*b[1][3]+a[2][2]*b[2][3]+a[2][3]*b[3][3];\
-						\
-   c[3][0] = a[3][0]*b[0][0]+a[3][1]*b[1][0]+a[3][2]*b[2][0]+a[3][3]*b[3][0];\
-   c[3][1] = a[3][0]*b[0][1]+a[3][1]*b[1][1]+a[3][2]*b[2][1]+a[3][3]*b[3][1];\
-   c[3][2] = a[3][0]*b[0][2]+a[3][1]*b[1][2]+a[3][2]*b[2][2]+a[3][3]*b[3][2];\
-   c[3][3] = a[3][0]*b[0][3]+a[3][1]*b[1][3]+a[3][2]*b[2][3]+a[3][3]*b[3][3];\
-}\
-
+#define MATRIX_PRODUCT_4X4(c, a, b)                                                              \
+	{                                                                                            \
+		c[0][0] = a[0][0] * b[0][0] + a[0][1] * b[1][0] + a[0][2] * b[2][0] + a[0][3] * b[3][0]; \
+		c[0][1] = a[0][0] * b[0][1] + a[0][1] * b[1][1] + a[0][2] * b[2][1] + a[0][3] * b[3][1]; \
+		c[0][2] = a[0][0] * b[0][2] + a[0][1] * b[1][2] + a[0][2] * b[2][2] + a[0][3] * b[3][2]; \
+		c[0][3] = a[0][0] * b[0][3] + a[0][1] * b[1][3] + a[0][2] * b[2][3] + a[0][3] * b[3][3]; \
+                                                                                                 \
+		c[1][0] = a[1][0] * b[0][0] + a[1][1] * b[1][0] + a[1][2] * b[2][0] + a[1][3] * b[3][0]; \
+		c[1][1] = a[1][0] * b[0][1] + a[1][1] * b[1][1] + a[1][2] * b[2][1] + a[1][3] * b[3][1]; \
+		c[1][2] = a[1][0] * b[0][2] + a[1][1] * b[1][2] + a[1][2] * b[2][2] + a[1][3] * b[3][2]; \
+		c[1][3] = a[1][0] * b[0][3] + a[1][1] * b[1][3] + a[1][2] * b[2][3] + a[1][3] * b[3][3]; \
+                                                                                                 \
+		c[2][0] = a[2][0] * b[0][0] + a[2][1] * b[1][0] + a[2][2] * b[2][0] + a[2][3] * b[3][0]; \
+		c[2][1] = a[2][0] * b[0][1] + a[2][1] * b[1][1] + a[2][2] * b[2][1] + a[2][3] * b[3][1]; \
+		c[2][2] = a[2][0] * b[0][2] + a[2][1] * b[1][2] + a[2][2] * b[2][2] + a[2][3] * b[3][2]; \
+		c[2][3] = a[2][0] * b[0][3] + a[2][1] * b[1][3] + a[2][2] * b[2][3] + a[2][3] * b[3][3]; \
+                                                                                                 \
+		c[3][0] = a[3][0] * b[0][0] + a[3][1] * b[1][0] + a[3][2] * b[2][0] + a[3][3] * b[3][0]; \
+		c[3][1] = a[3][0] * b[0][1] + a[3][1] * b[1][1] + a[3][2] * b[2][1] + a[3][3] * b[3][1]; \
+		c[3][2] = a[3][0] * b[0][2] + a[3][1] * b[1][2] + a[3][2] * b[2][2] + a[3][3] * b[3][2]; \
+		c[3][3] = a[3][0] * b[0][3] + a[3][1] * b[1][3] + a[3][2] * b[2][3] + a[3][3] * b[3][3]; \
+	}
 
 /*! matrix times vector */
-#define MAT_DOT_VEC_2X2(p,m,v)					\
-{								\
-   p[0] = m[0][0]*v[0] + m[0][1]*v[1];				\
-   p[1] = m[1][0]*v[0] + m[1][1]*v[1];				\
-}\
-
+#define MAT_DOT_VEC_2X2(p, m, v)                \
+	{                                           \
+		p[0] = m[0][0] * v[0] + m[0][1] * v[1]; \
+		p[1] = m[1][0] * v[0] + m[1][1] * v[1]; \
+	}
 
 /*! matrix times vector */
-#define MAT_DOT_VEC_3X3(p,m,v)					\
-{								\
-   p[0] = m[0][0]*v[0] + m[0][1]*v[1] + m[0][2]*v[2];		\
-   p[1] = m[1][0]*v[0] + m[1][1]*v[1] + m[1][2]*v[2];		\
-   p[2] = m[2][0]*v[0] + m[2][1]*v[1] + m[2][2]*v[2];		\
-}\
-
+#define MAT_DOT_VEC_3X3(p, m, v)                                 \
+	{                                                            \
+		p[0] = m[0][0] * v[0] + m[0][1] * v[1] + m[0][2] * v[2]; \
+		p[1] = m[1][0] * v[0] + m[1][1] * v[1] + m[1][2] * v[2]; \
+		p[2] = m[2][0] * v[0] + m[2][1] * v[1] + m[2][2] * v[2]; \
+	}
 
 /*! matrix times vector
 v is a vec4f
 */
-#define MAT_DOT_VEC_4X4(p,m,v)					\
-{								\
-   p[0] = m[0][0]*v[0] + m[0][1]*v[1] + m[0][2]*v[2] + m[0][3]*v[3];	\
-   p[1] = m[1][0]*v[0] + m[1][1]*v[1] + m[1][2]*v[2] + m[1][3]*v[3];	\
-   p[2] = m[2][0]*v[0] + m[2][1]*v[1] + m[2][2]*v[2] + m[2][3]*v[3];	\
-   p[3] = m[3][0]*v[0] + m[3][1]*v[1] + m[3][2]*v[2] + m[3][3]*v[3];	\
-}\
+#define MAT_DOT_VEC_4X4(p, m, v)                                                  \
+	{                                                                             \
+		p[0] = m[0][0] * v[0] + m[0][1] * v[1] + m[0][2] * v[2] + m[0][3] * v[3]; \
+		p[1] = m[1][0] * v[0] + m[1][1] * v[1] + m[1][2] * v[2] + m[1][3] * v[3]; \
+		p[2] = m[2][0] * v[0] + m[2][1] * v[1] + m[2][2] * v[2] + m[2][3] * v[3]; \
+		p[3] = m[3][0] * v[0] + m[3][1] * v[1] + m[3][2] * v[2] + m[3][3] * v[3]; \
+	}
 
 /*! matrix times vector
 v is a vec3f
 and m is a mat4f<br>
 Last column is added as the position
 */
-#define MAT_DOT_VEC_3X4(p,m,v)					\
-{								\
-   p[0] = m[0][0]*v[0] + m[0][1]*v[1] + m[0][2]*v[2] + m[0][3];	\
-   p[1] = m[1][0]*v[0] + m[1][1]*v[1] + m[1][2]*v[2] + m[1][3];	\
-   p[2] = m[2][0]*v[0] + m[2][1]*v[1] + m[2][2]*v[2] + m[2][3];	\
-}\
-
+#define MAT_DOT_VEC_3X4(p, m, v)                                           \
+	{                                                                      \
+		p[0] = m[0][0] * v[0] + m[0][1] * v[1] + m[0][2] * v[2] + m[0][3]; \
+		p[1] = m[1][0] * v[0] + m[1][1] * v[1] + m[1][2] * v[2] + m[1][3]; \
+		p[2] = m[2][0] * v[0] + m[2][1] * v[1] + m[2][2] * v[2] + m[2][3]; \
+	}
 
 /*! vector transpose times matrix */
 /*! p[j] = v[0]*m[0][j] + v[1]*m[1][j] + v[2]*m[2][j]; */
-#define VEC_DOT_MAT_3X3(p,v,m)					\
-{								\
-   p[0] = v[0]*m[0][0] + v[1]*m[1][0] + v[2]*m[2][0];		\
-   p[1] = v[0]*m[0][1] + v[1]*m[1][1] + v[2]*m[2][1];		\
-   p[2] = v[0]*m[0][2] + v[1]*m[1][2] + v[2]*m[2][2];		\
-}\
-
+#define VEC_DOT_MAT_3X3(p, v, m)                                 \
+	{                                                            \
+		p[0] = v[0] * m[0][0] + v[1] * m[1][0] + v[2] * m[2][0]; \
+		p[1] = v[0] * m[0][1] + v[1] * m[1][1] + v[2] * m[2][1]; \
+		p[2] = v[0] * m[0][2] + v[1] * m[1][2] + v[2] * m[2][2]; \
+	}
 
 /*! affine matrix times vector */
 /** The matrix is assumed to be an affine matrix, with last two
  * entries representing a translation */
-#define MAT_DOT_VEC_2X3(p,m,v)					\
-{								\
-   p[0] = m[0][0]*v[0] + m[0][1]*v[1] + m[0][2];		\
-   p[1] = m[1][0]*v[0] + m[1][1]*v[1] + m[1][2];		\
-}\
+#define MAT_DOT_VEC_2X3(p, m, v)                          \
+	{                                                     \
+		p[0] = m[0][0] * v[0] + m[0][1] * v[1] + m[0][2]; \
+		p[1] = m[1][0] * v[0] + m[1][1] * v[1] + m[1][2]; \
+	}
 
 //! Transform a plane
-#define MAT_TRANSFORM_PLANE_4X4(pout,m,plane)\
-{								\
-   pout[0] = m[0][0]*plane[0] + m[0][1]*plane[1]  + m[0][2]*plane[2];\
-   pout[1] = m[1][0]*plane[0] + m[1][1]*plane[1]  + m[1][2]*plane[2];\
-   pout[2] = m[2][0]*plane[0] + m[2][1]*plane[1]  + m[2][2]*plane[2];\
-   pout[3] = m[0][3]*pout[0] + m[1][3]*pout[1]  + m[2][3]*pout[2] + plane[3];\
-}\
-
-
+#define MAT_TRANSFORM_PLANE_4X4(pout, m, plane)                                         \
+	{                                                                                   \
+		pout[0] = m[0][0] * plane[0] + m[0][1] * plane[1] + m[0][2] * plane[2];         \
+		pout[1] = m[1][0] * plane[0] + m[1][1] * plane[1] + m[1][2] * plane[2];         \
+		pout[2] = m[2][0] * plane[0] + m[2][1] * plane[1] + m[2][2] * plane[2];         \
+		pout[3] = m[0][3] * pout[0] + m[1][3] * pout[1] + m[2][3] * pout[2] + plane[3]; \
+	}
 
 /** inverse transpose of matrix times vector
  *
@@ -1000,22 +938,22 @@ Last column is added as the position
  * It will leave normals the wrong length !!!
  * See macro below for use on normals.
  */
-#define INV_TRANSP_MAT_DOT_VEC_2X2(p,m,v)			\
-{								\
-   GREAL det;						\
-								\
-   det = m[0][0]*m[1][1] - m[0][1]*m[1][0];			\
-   p[0] = m[1][1]*v[0] - m[1][0]*v[1];				\
-   p[1] = - m[0][1]*v[0] + m[0][0]*v[1];			\
-								\
-   /* if matrix not singular, and not orthonormal, then renormalize */ \
-   if ((det!=1.0f) && (det != 0.0f)) {				\
-      det = 1.0f / det;						\
-      p[0] *= det;						\
-      p[1] *= det;						\
-   }								\
-}\
-
+#define INV_TRANSP_MAT_DOT_VEC_2X2(p, m, v)                                 \
+	{                                                                       \
+		GREAL det;                                                          \
+                                                                            \
+		det = m[0][0] * m[1][1] - m[0][1] * m[1][0];                        \
+		p[0] = m[1][1] * v[0] - m[1][0] * v[1];                             \
+		p[1] = -m[0][1] * v[0] + m[0][0] * v[1];                            \
+                                                                            \
+		/* if matrix not singular, and not orthonormal, then renormalize */ \
+		if ((det != 1.0f) && (det != 0.0f))                                 \
+		{                                                                   \
+			det = 1.0f / det;                                               \
+			p[0] *= det;                                                    \
+			p[1] *= det;                                                    \
+		}                                                                   \
+	}
 
 /** transform normal vector by inverse transpose of matrix
  * and then renormalize the vector
@@ -1024,550 +962,527 @@ Last column is added as the position
  * and multiplies vector v into it, to yeild vector p
  * Vector p is then normalized.
  */
-#define NORM_XFORM_2X2(p,m,v)					\
-{								\
-   GREAL len;							\
-								\
-   /* do nothing if off-diagonals are zero and diagonals are 	\
-    * equal */							\
-   if ((m[0][1] != 0.0) || (m[1][0] != 0.0) || (m[0][0] != m[1][1])) { \
-      p[0] = m[1][1]*v[0] - m[1][0]*v[1];			\
-      p[1] = - m[0][1]*v[0] + m[0][0]*v[1];			\
-								\
-      len = p[0]*p[0] + p[1]*p[1];				\
-      GIM_INV_SQRT(len,len);					\
-      p[0] *= len;						\
-      p[1] *= len;						\
-   } else {							\
-      VEC_COPY_2 (p, v);					\
-   }								\
-}\
-
+#define NORM_XFORM_2X2(p, m, v)                                           \
+	{                                                                     \
+		GREAL len;                                                        \
+                                                                          \
+		/* do nothing if off-diagonals are zero and diagonals are 	\
+    * equal */      \
+		if ((m[0][1] != 0.0) || (m[1][0] != 0.0) || (m[0][0] != m[1][1])) \
+		{                                                                 \
+			p[0] = m[1][1] * v[0] - m[1][0] * v[1];                       \
+			p[1] = -m[0][1] * v[0] + m[0][0] * v[1];                      \
+                                                                          \
+			len = p[0] * p[0] + p[1] * p[1];                              \
+			GIM_INV_SQRT(len, len);                                       \
+			p[0] *= len;                                                  \
+			p[1] *= len;                                                  \
+		}                                                                 \
+		else                                                              \
+		{                                                                 \
+			VEC_COPY_2(p, v);                                             \
+		}                                                                 \
+	}
 
 /** outer product of vector times vector transpose
  *
  * The outer product of vector v and vector transpose t yeilds
  * dyadic matrix m.
  */
-#define OUTER_PRODUCT_2X2(m,v,t)				\
-{								\
-   m[0][0] = v[0] * t[0];					\
-   m[0][1] = v[0] * t[1];					\
-								\
-   m[1][0] = v[1] * t[0];					\
-   m[1][1] = v[1] * t[1];					\
-}\
-
+#define OUTER_PRODUCT_2X2(m, v, t) \
+	{                              \
+		m[0][0] = v[0] * t[0];     \
+		m[0][1] = v[0] * t[1];     \
+                                   \
+		m[1][0] = v[1] * t[0];     \
+		m[1][1] = v[1] * t[1];     \
+	}
 
 /** outer product of vector times vector transpose
  *
  * The outer product of vector v and vector transpose t yeilds
  * dyadic matrix m.
  */
-#define OUTER_PRODUCT_3X3(m,v,t)				\
-{								\
-   m[0][0] = v[0] * t[0];					\
-   m[0][1] = v[0] * t[1];					\
-   m[0][2] = v[0] * t[2];					\
-								\
-   m[1][0] = v[1] * t[0];					\
-   m[1][1] = v[1] * t[1];					\
-   m[1][2] = v[1] * t[2];					\
-								\
-   m[2][0] = v[2] * t[0];					\
-   m[2][1] = v[2] * t[1];					\
-   m[2][2] = v[2] * t[2];					\
-}\
-
+#define OUTER_PRODUCT_3X3(m, v, t) \
+	{                              \
+		m[0][0] = v[0] * t[0];     \
+		m[0][1] = v[0] * t[1];     \
+		m[0][2] = v[0] * t[2];     \
+                                   \
+		m[1][0] = v[1] * t[0];     \
+		m[1][1] = v[1] * t[1];     \
+		m[1][2] = v[1] * t[2];     \
+                                   \
+		m[2][0] = v[2] * t[0];     \
+		m[2][1] = v[2] * t[1];     \
+		m[2][2] = v[2] * t[2];     \
+	}
 
 /** outer product of vector times vector transpose
  *
  * The outer product of vector v and vector transpose t yeilds
  * dyadic matrix m.
  */
-#define OUTER_PRODUCT_4X4(m,v,t)				\
-{								\
-   m[0][0] = v[0] * t[0];					\
-   m[0][1] = v[0] * t[1];					\
-   m[0][2] = v[0] * t[2];					\
-   m[0][3] = v[0] * t[3];					\
-								\
-   m[1][0] = v[1] * t[0];					\
-   m[1][1] = v[1] * t[1];					\
-   m[1][2] = v[1] * t[2];					\
-   m[1][3] = v[1] * t[3];					\
-								\
-   m[2][0] = v[2] * t[0];					\
-   m[2][1] = v[2] * t[1];					\
-   m[2][2] = v[2] * t[2];					\
-   m[2][3] = v[2] * t[3];					\
-								\
-   m[3][0] = v[3] * t[0];					\
-   m[3][1] = v[3] * t[1];					\
-   m[3][2] = v[3] * t[2];					\
-   m[3][3] = v[3] * t[3];					\
-}\
-
+#define OUTER_PRODUCT_4X4(m, v, t) \
+	{                              \
+		m[0][0] = v[0] * t[0];     \
+		m[0][1] = v[0] * t[1];     \
+		m[0][2] = v[0] * t[2];     \
+		m[0][3] = v[0] * t[3];     \
+                                   \
+		m[1][0] = v[1] * t[0];     \
+		m[1][1] = v[1] * t[1];     \
+		m[1][2] = v[1] * t[2];     \
+		m[1][3] = v[1] * t[3];     \
+                                   \
+		m[2][0] = v[2] * t[0];     \
+		m[2][1] = v[2] * t[1];     \
+		m[2][2] = v[2] * t[2];     \
+		m[2][3] = v[2] * t[3];     \
+                                   \
+		m[3][0] = v[3] * t[0];     \
+		m[3][1] = v[3] * t[1];     \
+		m[3][2] = v[3] * t[2];     \
+		m[3][3] = v[3] * t[3];     \
+	}
 
 /** outer product of vector times vector transpose
  *
  * The outer product of vector v and vector transpose t yeilds
  * dyadic matrix m.
  */
-#define ACCUM_OUTER_PRODUCT_2X2(m,v,t)				\
-{								\
-   m[0][0] += v[0] * t[0];					\
-   m[0][1] += v[0] * t[1];					\
-								\
-   m[1][0] += v[1] * t[0];					\
-   m[1][1] += v[1] * t[1];					\
-}\
-
+#define ACCUM_OUTER_PRODUCT_2X2(m, v, t) \
+	{                                    \
+		m[0][0] += v[0] * t[0];          \
+		m[0][1] += v[0] * t[1];          \
+                                         \
+		m[1][0] += v[1] * t[0];          \
+		m[1][1] += v[1] * t[1];          \
+	}
 
 /** outer product of vector times vector transpose
  *
  * The outer product of vector v and vector transpose t yeilds
  * dyadic matrix m.
  */
-#define ACCUM_OUTER_PRODUCT_3X3(m,v,t)				\
-{								\
-   m[0][0] += v[0] * t[0];					\
-   m[0][1] += v[0] * t[1];					\
-   m[0][2] += v[0] * t[2];					\
-								\
-   m[1][0] += v[1] * t[0];					\
-   m[1][1] += v[1] * t[1];					\
-   m[1][2] += v[1] * t[2];					\
-								\
-   m[2][0] += v[2] * t[0];					\
-   m[2][1] += v[2] * t[1];					\
-   m[2][2] += v[2] * t[2];					\
-}\
-
+#define ACCUM_OUTER_PRODUCT_3X3(m, v, t) \
+	{                                    \
+		m[0][0] += v[0] * t[0];          \
+		m[0][1] += v[0] * t[1];          \
+		m[0][2] += v[0] * t[2];          \
+                                         \
+		m[1][0] += v[1] * t[0];          \
+		m[1][1] += v[1] * t[1];          \
+		m[1][2] += v[1] * t[2];          \
+                                         \
+		m[2][0] += v[2] * t[0];          \
+		m[2][1] += v[2] * t[1];          \
+		m[2][2] += v[2] * t[2];          \
+	}
 
 /** outer product of vector times vector transpose
  *
  * The outer product of vector v and vector transpose t yeilds
  * dyadic matrix m.
  */
-#define ACCUM_OUTER_PRODUCT_4X4(m,v,t)				\
-{								\
-   m[0][0] += v[0] * t[0];					\
-   m[0][1] += v[0] * t[1];					\
-   m[0][2] += v[0] * t[2];					\
-   m[0][3] += v[0] * t[3];					\
-								\
-   m[1][0] += v[1] * t[0];					\
-   m[1][1] += v[1] * t[1];					\
-   m[1][2] += v[1] * t[2];					\
-   m[1][3] += v[1] * t[3];					\
-								\
-   m[2][0] += v[2] * t[0];					\
-   m[2][1] += v[2] * t[1];					\
-   m[2][2] += v[2] * t[2];					\
-   m[2][3] += v[2] * t[3];					\
-								\
-   m[3][0] += v[3] * t[0];					\
-   m[3][1] += v[3] * t[1];					\
-   m[3][2] += v[3] * t[2];					\
-   m[3][3] += v[3] * t[3];					\
-}\
-
+#define ACCUM_OUTER_PRODUCT_4X4(m, v, t) \
+	{                                    \
+		m[0][0] += v[0] * t[0];          \
+		m[0][1] += v[0] * t[1];          \
+		m[0][2] += v[0] * t[2];          \
+		m[0][3] += v[0] * t[3];          \
+                                         \
+		m[1][0] += v[1] * t[0];          \
+		m[1][1] += v[1] * t[1];          \
+		m[1][2] += v[1] * t[2];          \
+		m[1][3] += v[1] * t[3];          \
+                                         \
+		m[2][0] += v[2] * t[0];          \
+		m[2][1] += v[2] * t[1];          \
+		m[2][2] += v[2] * t[2];          \
+		m[2][3] += v[2] * t[3];          \
+                                         \
+		m[3][0] += v[3] * t[0];          \
+		m[3][1] += v[3] * t[1];          \
+		m[3][2] += v[3] * t[2];          \
+		m[3][3] += v[3] * t[3];          \
+	}
 
 /** determinant of matrix
  *
  * Computes determinant of matrix m, returning d
  */
-#define DETERMINANT_2X2(d,m)					\
-{								\
-   d = m[0][0] * m[1][1] - m[0][1] * m[1][0];			\
-}\
-
+#define DETERMINANT_2X2(d, m)                      \
+	{                                              \
+		d = m[0][0] * m[1][1] - m[0][1] * m[1][0]; \
+	}
 
 /** determinant of matrix
  *
  * Computes determinant of matrix m, returning d
  */
-#define DETERMINANT_3X3(d,m)					\
-{								\
-   d = m[0][0] * (m[1][1]*m[2][2] - m[1][2] * m[2][1]);		\
-   d -= m[0][1] * (m[1][0]*m[2][2] - m[1][2] * m[2][0]);	\
-   d += m[0][2] * (m[1][0]*m[2][1] - m[1][1] * m[2][0]);	\
-}\
-
+#define DETERMINANT_3X3(d, m)                                   \
+	{                                                           \
+		d = m[0][0] * (m[1][1] * m[2][2] - m[1][2] * m[2][1]);  \
+		d -= m[0][1] * (m[1][0] * m[2][2] - m[1][2] * m[2][0]); \
+		d += m[0][2] * (m[1][0] * m[2][1] - m[1][1] * m[2][0]); \
+	}
 
 /** i,j,th cofactor of a 4x4 matrix
  *
  */
-#define COFACTOR_4X4_IJ(fac,m,i,j) 				\
-{								\
-   GUINT __ii[4], __jj[4], __k;						\
-								\
-   for (__k=0; __k<i; __k++) __ii[__k] = __k;				\
-   for (__k=i; __k<3; __k++) __ii[__k] = __k+1;				\
-   for (__k=0; __k<j; __k++) __jj[__k] = __k;				\
-   for (__k=j; __k<3; __k++) __jj[__k] = __k+1;				\
-								\
-   (fac) = m[__ii[0]][__jj[0]] * (m[__ii[1]][__jj[1]]*m[__ii[2]][__jj[2]] 	\
-                            - m[__ii[1]][__jj[2]]*m[__ii[2]][__jj[1]]); \
-   (fac) -= m[__ii[0]][__jj[1]] * (m[__ii[1]][__jj[0]]*m[__ii[2]][__jj[2]]	\
-                             - m[__ii[1]][__jj[2]]*m[__ii[2]][__jj[0]]);\
-   (fac) += m[__ii[0]][__jj[2]] * (m[__ii[1]][__jj[0]]*m[__ii[2]][__jj[1]]	\
-                             - m[__ii[1]][__jj[1]]*m[__ii[2]][__jj[0]]);\
-								\
-   __k = i+j;							\
-   if ( __k != (__k/2)*2) {						\
-      (fac) = -(fac);						\
-   }								\
-}\
-
+#define COFACTOR_4X4_IJ(fac, m, i, j)                                                                                           \
+	{                                                                                                                           \
+		GUINT __ii[4], __jj[4], __k;                                                                                            \
+                                                                                                                                \
+		for (__k = 0; __k < i; __k++) __ii[__k] = __k;                                                                          \
+		for (__k = i; __k < 3; __k++) __ii[__k] = __k + 1;                                                                      \
+		for (__k = 0; __k < j; __k++) __jj[__k] = __k;                                                                          \
+		for (__k = j; __k < 3; __k++) __jj[__k] = __k + 1;                                                                      \
+                                                                                                                                \
+		(fac) = m[__ii[0]][__jj[0]] * (m[__ii[1]][__jj[1]] * m[__ii[2]][__jj[2]] - m[__ii[1]][__jj[2]] * m[__ii[2]][__jj[1]]);  \
+		(fac) -= m[__ii[0]][__jj[1]] * (m[__ii[1]][__jj[0]] * m[__ii[2]][__jj[2]] - m[__ii[1]][__jj[2]] * m[__ii[2]][__jj[0]]); \
+		(fac) += m[__ii[0]][__jj[2]] * (m[__ii[1]][__jj[0]] * m[__ii[2]][__jj[1]] - m[__ii[1]][__jj[1]] * m[__ii[2]][__jj[0]]); \
+                                                                                                                                \
+		__k = i + j;                                                                                                            \
+		if (__k != (__k / 2) * 2)                                                                                               \
+		{                                                                                                                       \
+			(fac) = -(fac);                                                                                                     \
+		}                                                                                                                       \
+	}
 
 /** determinant of matrix
  *
  * Computes determinant of matrix m, returning d
  */
-#define DETERMINANT_4X4(d,m)					\
-{								\
-   GREAL cofac;						\
-   COFACTOR_4X4_IJ (cofac, m, 0, 0);				\
-   d = m[0][0] * cofac;						\
-   COFACTOR_4X4_IJ (cofac, m, 0, 1);				\
-   d += m[0][1] * cofac;					\
-   COFACTOR_4X4_IJ (cofac, m, 0, 2);				\
-   d += m[0][2] * cofac;					\
-   COFACTOR_4X4_IJ (cofac, m, 0, 3);				\
-   d += m[0][3] * cofac;					\
-}\
-
+#define DETERMINANT_4X4(d, m)            \
+	{                                    \
+		GREAL cofac;                     \
+		COFACTOR_4X4_IJ(cofac, m, 0, 0); \
+		d = m[0][0] * cofac;             \
+		COFACTOR_4X4_IJ(cofac, m, 0, 1); \
+		d += m[0][1] * cofac;            \
+		COFACTOR_4X4_IJ(cofac, m, 0, 2); \
+		d += m[0][2] * cofac;            \
+		COFACTOR_4X4_IJ(cofac, m, 0, 3); \
+		d += m[0][3] * cofac;            \
+	}
 
 /** cofactor of matrix
  *
  * Computes cofactor of matrix m, returning a
  */
-#define COFACTOR_2X2(a,m)					\
-{								\
-   a[0][0] = (m)[1][1];						\
-   a[0][1] = - (m)[1][0];						\
-   a[1][0] = - (m)[0][1];						\
-   a[1][1] = (m)[0][0];						\
-}\
-
+#define COFACTOR_2X2(a, m)    \
+	{                         \
+		a[0][0] = (m)[1][1];  \
+		a[0][1] = -(m)[1][0]; \
+		a[1][0] = -(m)[0][1]; \
+		a[1][1] = (m)[0][0];  \
+	}
 
 /** cofactor of matrix
  *
  * Computes cofactor of matrix m, returning a
  */
-#define COFACTOR_3X3(a,m)					\
-{								\
-   a[0][0] = m[1][1]*m[2][2] - m[1][2]*m[2][1];			\
-   a[0][1] = - (m[1][0]*m[2][2] - m[2][0]*m[1][2]);		\
-   a[0][2] = m[1][0]*m[2][1] - m[1][1]*m[2][0];			\
-   a[1][0] = - (m[0][1]*m[2][2] - m[0][2]*m[2][1]);		\
-   a[1][1] = m[0][0]*m[2][2] - m[0][2]*m[2][0];			\
-   a[1][2] = - (m[0][0]*m[2][1] - m[0][1]*m[2][0]);		\
-   a[2][0] = m[0][1]*m[1][2] - m[0][2]*m[1][1];			\
-   a[2][1] = - (m[0][0]*m[1][2] - m[0][2]*m[1][0]);		\
-   a[2][2] = m[0][0]*m[1][1] - m[0][1]*m[1][0]);		\
-}\
-
+#define COFACTOR_3X3(a, m)                                  \
+	{                                                       \
+		a[0][0] = m[1][1] * m[2][2] - m[1][2] * m[2][1];    \
+		a[0][1] = -(m[1][0] * m[2][2] - m[2][0] * m[1][2]); \
+		a[0][2] = m[1][0] * m[2][1] - m[1][1] * m[2][0];    \
+		a[1][0] = -(m[0][1] * m[2][2] - m[0][2] * m[2][1]); \
+		a[1][1] = m[0][0] * m[2][2] - m[0][2] * m[2][0];    \
+		a[1][2] = -(m[0][0] * m[2][1] - m[0][1] * m[2][0]); \
+		a[2][0] = m[0][1] * m[1][2] - m[0][2] * m[1][1];    \
+		a[2][1] = -(m[0][0] * m[1][2] - m[0][2] * m[1][0]); \
+   a[2][2] = m[0][0]*m[1][1] - m[0][1]*m[1][0]);            \
+	}
 
 /** cofactor of matrix
  *
  * Computes cofactor of matrix m, returning a
  */
-#define COFACTOR_4X4(a,m)					\
-{								\
-   int i,j;							\
-								\
-   for (i=0; i<4; i++) {					\
-      for (j=0; j<4; j++) {					\
-         COFACTOR_4X4_IJ (a[i][j], m, i, j);			\
-      }								\
-   }								\
-}\
-
+#define COFACTOR_4X4(a, m)                         \
+	{                                              \
+		int i, j;                                  \
+                                                   \
+		for (i = 0; i < 4; i++)                    \
+		{                                          \
+			for (j = 0; j < 4; j++)                \
+			{                                      \
+				COFACTOR_4X4_IJ(a[i][j], m, i, j); \
+			}                                      \
+		}                                          \
+	}
 
 /** adjoint of matrix
  *
  * Computes adjoint of matrix m, returning a
  * (Note that adjoint is just the transpose of the cofactor matrix)
  */
-#define ADJOINT_2X2(a,m)					\
-{								\
-   a[0][0] = (m)[1][1];						\
-   a[1][0] = - (m)[1][0];						\
-   a[0][1] = - (m)[0][1];						\
-   a[1][1] = (m)[0][0];						\
-}\
-
+#define ADJOINT_2X2(a, m)     \
+	{                         \
+		a[0][0] = (m)[1][1];  \
+		a[1][0] = -(m)[1][0]; \
+		a[0][1] = -(m)[0][1]; \
+		a[1][1] = (m)[0][0];  \
+	}
 
 /** adjoint of matrix
  *
  * Computes adjoint of matrix m, returning a
  * (Note that adjoint is just the transpose of the cofactor matrix)
  */
-#define ADJOINT_3X3(a,m)					\
-{								\
-   a[0][0] = m[1][1]*m[2][2] - m[1][2]*m[2][1];			\
-   a[1][0] = - (m[1][0]*m[2][2] - m[2][0]*m[1][2]);		\
-   a[2][0] = m[1][0]*m[2][1] - m[1][1]*m[2][0];			\
-   a[0][1] = - (m[0][1]*m[2][2] - m[0][2]*m[2][1]);		\
-   a[1][1] = m[0][0]*m[2][2] - m[0][2]*m[2][0];			\
-   a[2][1] = - (m[0][0]*m[2][1] - m[0][1]*m[2][0]);		\
-   a[0][2] = m[0][1]*m[1][2] - m[0][2]*m[1][1];			\
-   a[1][2] = - (m[0][0]*m[1][2] - m[0][2]*m[1][0]);		\
-   a[2][2] = m[0][0]*m[1][1] - m[0][1]*m[1][0]);		\
-}\
-
+#define ADJOINT_3X3(a, m)                                   \
+	{                                                       \
+		a[0][0] = m[1][1] * m[2][2] - m[1][2] * m[2][1];    \
+		a[1][0] = -(m[1][0] * m[2][2] - m[2][0] * m[1][2]); \
+		a[2][0] = m[1][0] * m[2][1] - m[1][1] * m[2][0];    \
+		a[0][1] = -(m[0][1] * m[2][2] - m[0][2] * m[2][1]); \
+		a[1][1] = m[0][0] * m[2][2] - m[0][2] * m[2][0];    \
+		a[2][1] = -(m[0][0] * m[2][1] - m[0][1] * m[2][0]); \
+		a[0][2] = m[0][1] * m[1][2] - m[0][2] * m[1][1];    \
+		a[1][2] = -(m[0][0] * m[1][2] - m[0][2] * m[1][0]); \
+   a[2][2] = m[0][0]*m[1][1] - m[0][1]*m[1][0]);            \
+	}
 
 /** adjoint of matrix
  *
  * Computes adjoint of matrix m, returning a
  * (Note that adjoint is just the transpose of the cofactor matrix)
  */
-#define ADJOINT_4X4(a,m)					\
-{								\
-   char _i_,_j_;							\
-								\
-   for (_i_=0; _i_<4; _i_++) {					\
-      for (_j_=0; _j_<4; _j_++) {					\
-         COFACTOR_4X4_IJ (a[_j_][_i_], m, _i_, _j_);			\
-      }								\
-   }								\
-}\
-
+#define ADJOINT_4X4(a, m)                                  \
+	{                                                      \
+		char _i_, _j_;                                     \
+                                                           \
+		for (_i_ = 0; _i_ < 4; _i_++)                      \
+		{                                                  \
+			for (_j_ = 0; _j_ < 4; _j_++)                  \
+			{                                              \
+				COFACTOR_4X4_IJ(a[_j_][_i_], m, _i_, _j_); \
+			}                                              \
+		}                                                  \
+	}
 
 /** compute adjoint of matrix and scale
  *
  * Computes adjoint of matrix m, scales it by s, returning a
  */
-#define SCALE_ADJOINT_2X2(a,s,m)				\
-{								\
-   a[0][0] = (s) * m[1][1];					\
-   a[1][0] = - (s) * m[1][0];					\
-   a[0][1] = - (s) * m[0][1];					\
-   a[1][1] = (s) * m[0][0];					\
-}\
-
+#define SCALE_ADJOINT_2X2(a, s, m) \
+	{                              \
+		a[0][0] = (s)*m[1][1];     \
+		a[1][0] = -(s)*m[1][0];    \
+		a[0][1] = -(s)*m[0][1];    \
+		a[1][1] = (s)*m[0][0];     \
+	}
 
 /** compute adjoint of matrix and scale
  *
  * Computes adjoint of matrix m, scales it by s, returning a
  */
-#define SCALE_ADJOINT_3X3(a,s,m)				\
-{								\
-   a[0][0] = (s) * (m[1][1] * m[2][2] - m[1][2] * m[2][1]);	\
-   a[1][0] = (s) * (m[1][2] * m[2][0] - m[1][0] * m[2][2]);	\
-   a[2][0] = (s) * (m[1][0] * m[2][1] - m[1][1] * m[2][0]);	\
-								\
-   a[0][1] = (s) * (m[0][2] * m[2][1] - m[0][1] * m[2][2]);	\
-   a[1][1] = (s) * (m[0][0] * m[2][2] - m[0][2] * m[2][0]);	\
-   a[2][1] = (s) * (m[0][1] * m[2][0] - m[0][0] * m[2][1]);	\
-								\
-   a[0][2] = (s) * (m[0][1] * m[1][2] - m[0][2] * m[1][1]);	\
-   a[1][2] = (s) * (m[0][2] * m[1][0] - m[0][0] * m[1][2]);	\
-   a[2][2] = (s) * (m[0][0] * m[1][1] - m[0][1] * m[1][0]);	\
-}\
-
+#define SCALE_ADJOINT_3X3(a, s, m)                               \
+	{                                                            \
+		a[0][0] = (s) * (m[1][1] * m[2][2] - m[1][2] * m[2][1]); \
+		a[1][0] = (s) * (m[1][2] * m[2][0] - m[1][0] * m[2][2]); \
+		a[2][0] = (s) * (m[1][0] * m[2][1] - m[1][1] * m[2][0]); \
+                                                                 \
+		a[0][1] = (s) * (m[0][2] * m[2][1] - m[0][1] * m[2][2]); \
+		a[1][1] = (s) * (m[0][0] * m[2][2] - m[0][2] * m[2][0]); \
+		a[2][1] = (s) * (m[0][1] * m[2][0] - m[0][0] * m[2][1]); \
+                                                                 \
+		a[0][2] = (s) * (m[0][1] * m[1][2] - m[0][2] * m[1][1]); \
+		a[1][2] = (s) * (m[0][2] * m[1][0] - m[0][0] * m[1][2]); \
+		a[2][2] = (s) * (m[0][0] * m[1][1] - m[0][1] * m[1][0]); \
+	}
 
 /** compute adjoint of matrix and scale
  *
  * Computes adjoint of matrix m, scales it by s, returning a
  */
-#define SCALE_ADJOINT_4X4(a,s,m)				\
-{								\
-   char _i_,_j_; \
-   for (_i_=0; _i_<4; _i_++) {					\
-      for (_j_=0; _j_<4; _j_++) {					\
-         COFACTOR_4X4_IJ (a[_j_][_i_], m, _i_, _j_);			\
-         a[_j_][_i_] *= s;						\
-      }								\
-   }								\
-}\
+#define SCALE_ADJOINT_4X4(a, s, m)                         \
+	{                                                      \
+		char _i_, _j_;                                     \
+		for (_i_ = 0; _i_ < 4; _i_++)                      \
+		{                                                  \
+			for (_j_ = 0; _j_ < 4; _j_++)                  \
+			{                                              \
+				COFACTOR_4X4_IJ(a[_j_][_i_], m, _i_, _j_); \
+				a[_j_][_i_] *= s;                          \
+			}                                              \
+		}                                                  \
+	}
 
 /** inverse of matrix
  *
  * Compute inverse of matrix a, returning determinant m and
  * inverse b
  */
-#define INVERT_2X2(b,det,a)			\
-{						\
-   GREAL _tmp_;					\
-   DETERMINANT_2X2 (det, a);			\
-   _tmp_ = 1.0 / (det);				\
-   SCALE_ADJOINT_2X2 (b, _tmp_, a);		\
-}\
-
+#define INVERT_2X2(b, det, a)           \
+	{                                   \
+		GREAL _tmp_;                    \
+		DETERMINANT_2X2(det, a);        \
+		_tmp_ = 1.0 / (det);            \
+		SCALE_ADJOINT_2X2(b, _tmp_, a); \
+	}
 
 /** inverse of matrix
  *
  * Compute inverse of matrix a, returning determinant m and
  * inverse b
  */
-#define INVERT_3X3(b,det,a)			\
-{						\
-   GREAL _tmp_;					\
-   DETERMINANT_3X3 (det, a);			\
-   _tmp_ = 1.0 / (det);				\
-   SCALE_ADJOINT_3X3 (b, _tmp_, a);		\
-}\
-
+#define INVERT_3X3(b, det, a)           \
+	{                                   \
+		GREAL _tmp_;                    \
+		DETERMINANT_3X3(det, a);        \
+		_tmp_ = 1.0 / (det);            \
+		SCALE_ADJOINT_3X3(b, _tmp_, a); \
+	}
 
 /** inverse of matrix
  *
  * Compute inverse of matrix a, returning determinant m and
  * inverse b
  */
-#define INVERT_4X4(b,det,a)			\
-{						\
-   GREAL _tmp_;					\
-   DETERMINANT_4X4 (det, a);			\
-   _tmp_ = 1.0 / (det);				\
-   SCALE_ADJOINT_4X4 (b, _tmp_, a);		\
-}\
+#define INVERT_4X4(b, det, a)           \
+	{                                   \
+		GREAL _tmp_;                    \
+		DETERMINANT_4X4(det, a);        \
+		_tmp_ = 1.0 / (det);            \
+		SCALE_ADJOINT_4X4(b, _tmp_, a); \
+	}
 
 //! Get the triple(3) row of a transform matrix
-#define MAT_GET_ROW(mat,vec3,rowindex)\
-{\
-    vec3[0] = mat[rowindex][0];\
-    vec3[1] = mat[rowindex][1];\
-    vec3[2] = mat[rowindex][2]; \
-}\
+#define MAT_GET_ROW(mat, vec3, rowindex) \
+	{                                    \
+		vec3[0] = mat[rowindex][0];      \
+		vec3[1] = mat[rowindex][1];      \
+		vec3[2] = mat[rowindex][2];      \
+	}
 
 //! Get the tuple(2) row of a transform matrix
-#define MAT_GET_ROW2(mat,vec2,rowindex)\
-{\
-    vec2[0] = mat[rowindex][0];\
-    vec2[1] = mat[rowindex][1];\
-}\
-
+#define MAT_GET_ROW2(mat, vec2, rowindex) \
+	{                                     \
+		vec2[0] = mat[rowindex][0];       \
+		vec2[1] = mat[rowindex][1];       \
+	}
 
 //! Get the quad (4) row of a transform matrix
-#define MAT_GET_ROW4(mat,vec4,rowindex)\
-{\
-    vec4[0] = mat[rowindex][0];\
-    vec4[1] = mat[rowindex][1];\
-    vec4[2] = mat[rowindex][2];\
-    vec4[3] = mat[rowindex][3];\
-}\
+#define MAT_GET_ROW4(mat, vec4, rowindex) \
+	{                                     \
+		vec4[0] = mat[rowindex][0];       \
+		vec4[1] = mat[rowindex][1];       \
+		vec4[2] = mat[rowindex][2];       \
+		vec4[3] = mat[rowindex][3];       \
+	}
 
 //! Get the triple(3) col of a transform matrix
-#define MAT_GET_COL(mat,vec3,colindex)\
-{\
-    vec3[0] = mat[0][colindex];\
-    vec3[1] = mat[1][colindex];\
-    vec3[2] = mat[2][colindex]; \
-}\
+#define MAT_GET_COL(mat, vec3, colindex) \
+	{                                    \
+		vec3[0] = mat[0][colindex];      \
+		vec3[1] = mat[1][colindex];      \
+		vec3[2] = mat[2][colindex];      \
+	}
 
 //! Get the tuple(2) col of a transform matrix
-#define MAT_GET_COL2(mat,vec2,colindex)\
-{\
-    vec2[0] = mat[0][colindex];\
-    vec2[1] = mat[1][colindex];\
-}\
-
+#define MAT_GET_COL2(mat, vec2, colindex) \
+	{                                     \
+		vec2[0] = mat[0][colindex];       \
+		vec2[1] = mat[1][colindex];       \
+	}
 
 //! Get the quad (4) col of a transform matrix
-#define MAT_GET_COL4(mat,vec4,colindex)\
-{\
-    vec4[0] = mat[0][colindex];\
-    vec4[1] = mat[1][colindex];\
-    vec4[2] = mat[2][colindex];\
-    vec4[3] = mat[3][colindex];\
-}\
+#define MAT_GET_COL4(mat, vec4, colindex) \
+	{                                     \
+		vec4[0] = mat[0][colindex];       \
+		vec4[1] = mat[1][colindex];       \
+		vec4[2] = mat[2][colindex];       \
+		vec4[3] = mat[3][colindex];       \
+	}
 
 //! Get the triple(3) col of a transform matrix
-#define MAT_GET_X(mat,vec3)\
-{\
-    MAT_GET_COL(mat,vec3,0);\
-}\
+#define MAT_GET_X(mat, vec3)       \
+	{                              \
+		MAT_GET_COL(mat, vec3, 0); \
+	}
 
 //! Get the triple(3) col of a transform matrix
-#define MAT_GET_Y(mat,vec3)\
-{\
-    MAT_GET_COL(mat,vec3,1);\
-}\
+#define MAT_GET_Y(mat, vec3)       \
+	{                              \
+		MAT_GET_COL(mat, vec3, 1); \
+	}
 
 //! Get the triple(3) col of a transform matrix
-#define MAT_GET_Z(mat,vec3)\
-{\
-    MAT_GET_COL(mat,vec3,2);\
-}\
-
+#define MAT_GET_Z(mat, vec3)       \
+	{                              \
+		MAT_GET_COL(mat, vec3, 2); \
+	}
 
 //! Get the triple(3) col of a transform matrix
-#define MAT_SET_X(mat,vec3)\
-{\
-    mat[0][0] = vec3[0];\
-    mat[1][0] = vec3[1];\
-    mat[2][0] = vec3[2];\
-}\
+#define MAT_SET_X(mat, vec3) \
+	{                        \
+		mat[0][0] = vec3[0]; \
+		mat[1][0] = vec3[1]; \
+		mat[2][0] = vec3[2]; \
+	}
 
 //! Get the triple(3) col of a transform matrix
-#define MAT_SET_Y(mat,vec3)\
-{\
-    mat[0][1] = vec3[0];\
-    mat[1][1] = vec3[1];\
-    mat[2][1] = vec3[2];\
-}\
+#define MAT_SET_Y(mat, vec3) \
+	{                        \
+		mat[0][1] = vec3[0]; \
+		mat[1][1] = vec3[1]; \
+		mat[2][1] = vec3[2]; \
+	}
 
 //! Get the triple(3) col of a transform matrix
-#define MAT_SET_Z(mat,vec3)\
-{\
-    mat[0][2] = vec3[0];\
-    mat[1][2] = vec3[1];\
-    mat[2][2] = vec3[2];\
-}\
-
+#define MAT_SET_Z(mat, vec3) \
+	{                        \
+		mat[0][2] = vec3[0]; \
+		mat[1][2] = vec3[1]; \
+		mat[2][2] = vec3[2]; \
+	}
 
 //! Get the triple(3) col of a transform matrix
-#define MAT_GET_TRANSLATION(mat,vec3)\
-{\
-    vec3[0] = mat[0][3];\
-    vec3[1] = mat[1][3];\
-    vec3[2] = mat[2][3]; \
-}\
+#define MAT_GET_TRANSLATION(mat, vec3) \
+	{                                  \
+		vec3[0] = mat[0][3];           \
+		vec3[1] = mat[1][3];           \
+		vec3[2] = mat[2][3];           \
+	}
 
 //! Set the triple(3) col of a transform matrix
-#define MAT_SET_TRANSLATION(mat,vec3)\
-{\
-    mat[0][3] = vec3[0];\
-    mat[1][3] = vec3[1];\
-    mat[2][3] = vec3[2]; \
-}\
-
-
+#define MAT_SET_TRANSLATION(mat, vec3) \
+	{                                  \
+		mat[0][3] = vec3[0];           \
+		mat[1][3] = vec3[1];           \
+		mat[2][3] = vec3[2];           \
+	}
 
 //! Returns the dot product between a vec3f and the row of a matrix
-#define MAT_DOT_ROW(mat,vec3,rowindex) (vec3[0]*mat[rowindex][0] + vec3[1]*mat[rowindex][1] + vec3[2]*mat[rowindex][2])
+#define MAT_DOT_ROW(mat, vec3, rowindex) (vec3[0] * mat[rowindex][0] + vec3[1] * mat[rowindex][1] + vec3[2] * mat[rowindex][2])
 
 //! Returns the dot product between a vec2f and the row of a matrix
-#define MAT_DOT_ROW2(mat,vec2,rowindex) (vec2[0]*mat[rowindex][0] + vec2[1]*mat[rowindex][1])
+#define MAT_DOT_ROW2(mat, vec2, rowindex) (vec2[0] * mat[rowindex][0] + vec2[1] * mat[rowindex][1])
 
 //! Returns the dot product between a vec4f and the row of a matrix
-#define MAT_DOT_ROW4(mat,vec4,rowindex) (vec4[0]*mat[rowindex][0] + vec4[1]*mat[rowindex][1] + vec4[2]*mat[rowindex][2] + vec4[3]*mat[rowindex][3])
-
+#define MAT_DOT_ROW4(mat, vec4, rowindex) (vec4[0] * mat[rowindex][0] + vec4[1] * mat[rowindex][1] + vec4[2] * mat[rowindex][2] + vec4[3] * mat[rowindex][3])
 
 //! Returns the dot product between a vec3f and the col of a matrix
-#define MAT_DOT_COL(mat,vec3,colindex) (vec3[0]*mat[0][colindex] + vec3[1]*mat[1][colindex] + vec3[2]*mat[2][colindex])
+#define MAT_DOT_COL(mat, vec3, colindex) (vec3[0] * mat[0][colindex] + vec3[1] * mat[1][colindex] + vec3[2] * mat[2][colindex])
 
 //! Returns the dot product between a vec2f and the col of a matrix
-#define MAT_DOT_COL2(mat,vec2,colindex) (vec2[0]*mat[0][colindex] + vec2[1]*mat[1][colindex])
+#define MAT_DOT_COL2(mat, vec2, colindex) (vec2[0] * mat[0][colindex] + vec2[1] * mat[1][colindex])
 
 //! Returns the dot product between a vec4f and the col of a matrix
-#define MAT_DOT_COL4(mat,vec4,colindex) (vec4[0]*mat[0][colindex] + vec4[1]*mat[1][colindex] + vec4[2]*mat[2][colindex] + vec4[3]*mat[3][colindex])
+#define MAT_DOT_COL4(mat, vec4, colindex) (vec4[0] * mat[0][colindex] + vec4[1] * mat[1][colindex] + vec4[2] * mat[2][colindex] + vec4[3] * mat[3][colindex])
 
 /*!Transpose matrix times vector
 v is a vec3f
 and m is a mat4f<br>
 */
-#define INV_MAT_DOT_VEC_3X3(p,m,v)					\
-{								\
-   p[0] = MAT_DOT_COL(m,v,0); \
-   p[1] = MAT_DOT_COL(m,v,1);	\
-   p[2] = MAT_DOT_COL(m,v,2);	\
-}\
+#define INV_MAT_DOT_VEC_3X3(p, m, v) \
+	{                                \
+		p[0] = MAT_DOT_COL(m, v, 0); \
+		p[1] = MAT_DOT_COL(m, v, 1); \
+		p[2] = MAT_DOT_COL(m, v, 2); \
+	}
 
-
-
-#endif // GIM_VECTOR_H_INCLUDED
+#endif  // GIM_VECTOR_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/gim_math.h b/thirdparty/bullet/BulletCollision/Gimpact/gim_math.h
index 939079e1040..3c4f821a722 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/gim_math.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/gim_math.h
@@ -34,8 +34,6 @@ email: projectileman@yahoo.com
 
 #include "LinearMath/btScalar.h"
 
-
-
 #define GREAL btScalar
 #define GREAL2 double
 #define GINT int
@@ -45,8 +43,6 @@ email: projectileman@yahoo.com
 #define GINT64 long long
 #define GUINT64 unsigned long long
 
-
-
 #define G_PI 3.14159265358979f
 #define G_HALF_PI 1.5707963f
 //267948966
@@ -54,16 +50,14 @@ email: projectileman@yahoo.com
 //71795864
 #define G_ROOT3 1.73205f
 #define G_ROOT2 1.41421f
-#define G_UINT_INFINITY 0xffffffff //!< A very very high value
+#define G_UINT_INFINITY 0xffffffff  //!< A very very high value
 #define G_REAL_INFINITY FLT_MAX
-#define	G_SIGN_BITMASK			0x80000000
+#define G_SIGN_BITMASK 0x80000000
 #define G_EPSILON SIMD_EPSILON
 
-
-
 enum GIM_SCALAR_TYPES
 {
-	G_STYPE_REAL =0,
+	G_STYPE_REAL = 0,
 	G_STYPE_REAL2,
 	G_STYPE_SHORT,
 	G_STYPE_USHORT,
@@ -73,85 +67,82 @@ enum GIM_SCALAR_TYPES
 	G_STYPE_UINT64
 };
 
-
-
-#define G_DEGTORAD(X) ((X)*3.1415926f/180.0f)
-#define G_RADTODEG(X) ((X)*180.0f/3.1415926f)
+#define G_DEGTORAD(X) ((X)*3.1415926f / 180.0f)
+#define G_RADTODEG(X) ((X)*180.0f / 3.1415926f)
 
 //! Integer representation of a floating-point value.
-#define GIM_IR(x)					((GUINT&)(x))
+#define GIM_IR(x) ((GUINT&)(x))
 
 //! Signed integer representation of a floating-point value.
-#define GIM_SIR(x)					((GINT&)(x))
+#define GIM_SIR(x) ((GINT&)(x))
 
 //! Absolute integer representation of a floating-point value
-#define GIM_AIR(x)					(GIM_IR(x)&0x7fffffff)
+#define GIM_AIR(x) (GIM_IR(x) & 0x7fffffff)
 
 //! Floating-point representation of an integer value.
-#define GIM_FR(x)					((GREAL&)(x))
+#define GIM_FR(x) ((GREAL&)(x))
 
-#define GIM_MAX(a,b) (a<b?b:a)
-#define GIM_MIN(a,b) (a>b?b:a)
+#define GIM_MAX(a, b) (a < b ? b : a)
+#define GIM_MIN(a, b) (a > b ? b : a)
 
-#define GIM_MAX3(a,b,c) GIM_MAX(a,GIM_MAX(b,c))
-#define GIM_MIN3(a,b,c) GIM_MIN(a,GIM_MIN(b,c))
+#define GIM_MAX3(a, b, c) GIM_MAX(a, GIM_MAX(b, c))
+#define GIM_MIN3(a, b, c) GIM_MIN(a, GIM_MIN(b, c))
 
-#define GIM_IS_ZERO(value) (value < G_EPSILON &&  value > -G_EPSILON)
+#define GIM_IS_ZERO(value) (value < G_EPSILON && value > -G_EPSILON)
 
 #define GIM_IS_NEGATIVE(value) (value <= -G_EPSILON)
 
 #define GIM_IS_POSISITVE(value) (value >= G_EPSILON)
 
-#define GIM_NEAR_EQUAL(v1,v2) GIM_IS_ZERO((v1-v2))
+#define GIM_NEAR_EQUAL(v1, v2) GIM_IS_ZERO((v1 - v2))
 
 ///returns a clamped number
-#define GIM_CLAMP(number,minval,maxval) (number<minval?minval:(number>maxval?maxval:number))
+#define GIM_CLAMP(number, minval, maxval) (number < minval ? minval : (number > maxval ? maxval : number))
 
-#define GIM_GREATER(x, y)	btFabs(x) > (y)
+#define GIM_GREATER(x, y) btFabs(x) > (y)
 
 ///Swap numbers
-#define GIM_SWAP_NUMBERS(a,b){ \
-    a = a+b; \
-    b = a-b; \
-    a = a-b; \
-}\
+#define GIM_SWAP_NUMBERS(a, b) \
+	{                          \
+		a = a + b;             \
+		b = a - b;             \
+		a = a - b;             \
+	}
 
-#define GIM_INV_SQRT(va,isva)\
-{\
-    if(va<=0.0000001f)\
-    {\
-        isva = G_REAL_INFINITY;\
-    }\
-    else\
-    {\
-        GREAL _x = va * 0.5f;\
-        GUINT _y = 0x5f3759df - ( GIM_IR(va) >> 1);\
-        isva = GIM_FR(_y);\
-        isva  = isva * ( 1.5f - ( _x * isva * isva ) );\
-    }\
-}\
+#define GIM_INV_SQRT(va, isva)                         \
+	{                                                  \
+		if (va <= 0.0000001f)                          \
+		{                                              \
+			isva = G_REAL_INFINITY;                    \
+		}                                              \
+		else                                           \
+		{                                              \
+			GREAL _x = va * 0.5f;                      \
+			GUINT _y = 0x5f3759df - (GIM_IR(va) >> 1); \
+			isva = GIM_FR(_y);                         \
+			isva = isva * (1.5f - (_x * isva * isva)); \
+		}                                              \
+	}
 
-#define GIM_SQRT(va,sva)\
-{\
-    GIM_INV_SQRT(va,sva);\
-    sva = 1.0f/sva;\
-}\
+#define GIM_SQRT(va, sva)      \
+	{                          \
+		GIM_INV_SQRT(va, sva); \
+		sva = 1.0f / sva;      \
+	}
 
 //! Computes 1.0f / sqrtf(x). Comes from Quake3. See http://www.magic-software.com/3DGEDInvSqrt.html
 inline GREAL gim_inv_sqrt(GREAL f)
 {
-    GREAL r;
-    GIM_INV_SQRT(f,r);
-    return r;
+	GREAL r;
+	GIM_INV_SQRT(f, r);
+	return r;
 }
 
 inline GREAL gim_sqrt(GREAL f)
 {
-    GREAL r;
-    GIM_SQRT(f,r);
-    return r;
+	GREAL r;
+	GIM_SQRT(f, r);
+	return r;
 }
 
-
-
-#endif // GIM_MATH_H_INCLUDED
+#endif  // GIM_MATH_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/gim_memory.cpp b/thirdparty/bullet/BulletCollision/Gimpact/gim_memory.cpp
index 1636eb7867c..9e29ab91d67 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/gim_memory.cpp
+++ b/thirdparty/bullet/BulletCollision/Gimpact/gim_memory.cpp
@@ -27,7 +27,6 @@ email: projectileman@yahoo.com
 -----------------------------------------------------------------------------
 */
 
-
 #include "gim_memory.h"
 #include "stdlib.h"
 
@@ -40,52 +39,49 @@ static gim_alloca_function *g_allocafn = 0;
 static gim_realloc_function *g_reallocfn = 0;
 static gim_free_function *g_freefn = 0;
 
-void gim_set_alloc_handler (gim_alloc_function *fn)
+void gim_set_alloc_handler(gim_alloc_function *fn)
 {
-  g_allocfn = fn;
+	g_allocfn = fn;
 }
 
-void gim_set_alloca_handler (gim_alloca_function *fn)
+void gim_set_alloca_handler(gim_alloca_function *fn)
 {
-  g_allocafn = fn;
+	g_allocafn = fn;
 }
 
-void gim_set_realloc_handler (gim_realloc_function *fn)
+void gim_set_realloc_handler(gim_realloc_function *fn)
 {
-  g_reallocfn = fn;
+	g_reallocfn = fn;
 }
 
-void gim_set_free_handler (gim_free_function *fn)
+void gim_set_free_handler(gim_free_function *fn)
 {
-  g_freefn = fn;
+	g_freefn = fn;
 }
 
 gim_alloc_function *gim_get_alloc_handler()
 {
-  return g_allocfn;
+	return g_allocfn;
 }
 
 gim_alloca_function *gim_get_alloca_handler()
 {
-  return g_allocafn;
+	return g_allocafn;
 }
 
-
-gim_realloc_function *gim_get_realloc_handler ()
+gim_realloc_function *gim_get_realloc_handler()
 {
-  return g_reallocfn;
+	return g_reallocfn;
 }
 
-
-gim_free_function  *gim_get_free_handler ()
+gim_free_function *gim_get_free_handler()
 {
-  return g_freefn;
+	return g_freefn;
 }
 
-
-void * gim_alloc(size_t size)
+void *gim_alloc(size_t size)
 {
-	void * ptr;
+	void *ptr;
 	if (g_allocfn)
 	{
 		ptr = g_allocfn(size);
@@ -93,27 +89,29 @@ void * gim_alloc(size_t size)
 	else
 	{
 #ifdef GIM_SIMD_MEMORY
-		ptr = btAlignedAlloc(size,16);
+		ptr = btAlignedAlloc(size, 16);
 #else
 		ptr = malloc(size);
 #endif
 	}
-  	return ptr;
+	return ptr;
 }
 
-void * gim_alloca(size_t size)
+void *gim_alloca(size_t size)
 {
-  if (g_allocafn) return g_allocafn(size); else return gim_alloc(size);
+	if (g_allocafn)
+		return g_allocafn(size);
+	else
+		return gim_alloc(size);
 }
 
-
-void * gim_realloc(void *ptr, size_t oldsize, size_t newsize)
+void *gim_realloc(void *ptr, size_t oldsize, size_t newsize)
 {
- 	void * newptr = gim_alloc(newsize);
-    size_t copysize = oldsize<newsize?oldsize:newsize;
-    gim_simd_memcpy(newptr,ptr,copysize);
-    gim_free(ptr);
-    return newptr;
+	void *newptr = gim_alloc(newsize);
+	size_t copysize = oldsize < newsize ? oldsize : newsize;
+	gim_simd_memcpy(newptr, ptr, copysize);
+	gim_free(ptr);
+	return newptr;
 }
 
 void gim_free(void *ptr)
@@ -121,15 +119,14 @@ void gim_free(void *ptr)
 	if (!ptr) return;
 	if (g_freefn)
 	{
-	   g_freefn(ptr);
+		g_freefn(ptr);
 	}
 	else
 	{
-	#ifdef GIM_SIMD_MEMORY
+#ifdef GIM_SIMD_MEMORY
 		btAlignedFree(ptr);
-	#else
+#else
 		free(ptr);
-	#endif
+#endif
 	}
 }
-
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/gim_memory.h b/thirdparty/bullet/BulletCollision/Gimpact/gim_memory.h
index e203888a1e2..fffbfa23d8e 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/gim_memory.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/gim_memory.h
@@ -32,93 +32,84 @@ email: projectileman@yahoo.com
 -----------------------------------------------------------------------------
 */
 
-
 #include "gim_math.h"
 #include <string.h>
 
 #ifdef PREFETCH
-#include <xmmintrin.h>	// for prefetch
-#define pfval	64
-#define pfval2	128
+#include <xmmintrin.h>  // for prefetch
+#define pfval 64
+#define pfval2 128
 //! Prefetch 64
-#define pf(_x,_i)	_mm_prefetch((void *)(_x + _i + pfval), 0)
+#define pf(_x, _i) _mm_prefetch((void *)(_x + _i + pfval), 0)
 //! Prefetch 128
-#define pf2(_x,_i)	_mm_prefetch((void *)(_x + _i + pfval2), 0)
+#define pf2(_x, _i) _mm_prefetch((void *)(_x + _i + pfval2), 0)
 #else
 //! Prefetch 64
-#define pf(_x,_i)
+#define pf(_x, _i)
 //! Prefetch 128
-#define pf2(_x,_i)
+#define pf2(_x, _i)
 #endif
 
-
 ///Functions for manip packed arrays of numbers
-#define GIM_COPY_ARRAYS(dest_array,source_array,element_count)\
-{\
-    for (GUINT _i_=0;_i_<element_count ;++_i_)\
-    {\
-    	dest_array[_i_] = source_array[_i_];\
-    }\
-}\
+#define GIM_COPY_ARRAYS(dest_array, source_array, element_count) \
+	{                                                            \
+		for (GUINT _i_ = 0; _i_ < element_count; ++_i_)          \
+		{                                                        \
+			dest_array[_i_] = source_array[_i_];                 \
+		}                                                        \
+	}
 
-#define GIM_COPY_ARRAYS_1(dest_array,source_array,element_count,copy_macro)\
-{\
-    for (GUINT _i_=0;_i_<element_count ;++_i_)\
-    {\
-    	copy_macro(dest_array[_i_],source_array[_i_]);\
-    }\
-}\
+#define GIM_COPY_ARRAYS_1(dest_array, source_array, element_count, copy_macro) \
+	{                                                                          \
+		for (GUINT _i_ = 0; _i_ < element_count; ++_i_)                        \
+		{                                                                      \
+			copy_macro(dest_array[_i_], source_array[_i_]);                    \
+		}                                                                      \
+	}
 
+#define GIM_ZERO_ARRAY(array, element_count)            \
+	{                                                   \
+		for (GUINT _i_ = 0; _i_ < element_count; ++_i_) \
+		{                                               \
+			array[_i_] = 0;                             \
+		}                                               \
+	}
 
-#define GIM_ZERO_ARRAY(array,element_count)\
-{\
-    for (GUINT _i_=0;_i_<element_count ;++_i_)\
-    {\
-    	array[_i_] = 0;\
-    }\
-}\
-
-#define GIM_CONSTANT_ARRAY(array,element_count,constant)\
-{\
-    for (GUINT _i_=0;_i_<element_count ;++_i_)\
-    {\
-    	array[_i_] = constant;\
-    }\
-}\
-
+#define GIM_CONSTANT_ARRAY(array, element_count, constant) \
+	{                                                      \
+		for (GUINT _i_ = 0; _i_ < element_count; ++_i_)    \
+		{                                                  \
+			array[_i_] = constant;                         \
+		}                                                  \
+	}
 
 ///Function prototypes to allocate and free memory.
-typedef void * gim_alloc_function (size_t size);
-typedef void * gim_alloca_function (size_t size);//Allocs on the heap
-typedef void * gim_realloc_function (void *ptr, size_t oldsize, size_t newsize);
-typedef void gim_free_function (void *ptr);
-
+typedef void *gim_alloc_function(size_t size);
+typedef void *gim_alloca_function(size_t size);  //Allocs on the heap
+typedef void *gim_realloc_function(void *ptr, size_t oldsize, size_t newsize);
+typedef void gim_free_function(void *ptr);
 
 ///Memory Function Handlers
 ///set new memory management functions. if fn is 0, the default handlers are used.
-void gim_set_alloc_handler (gim_alloc_function *fn);
-void gim_set_alloca_handler (gim_alloca_function *fn);
-void gim_set_realloc_handler (gim_realloc_function *fn);
-void gim_set_free_handler (gim_free_function *fn);
-
+void gim_set_alloc_handler(gim_alloc_function *fn);
+void gim_set_alloca_handler(gim_alloca_function *fn);
+void gim_set_realloc_handler(gim_realloc_function *fn);
+void gim_set_free_handler(gim_free_function *fn);
 
 ///get current memory management functions.
-gim_alloc_function *gim_get_alloc_handler (void);
+gim_alloc_function *gim_get_alloc_handler(void);
 gim_alloca_function *gim_get_alloca_handler(void);
-gim_realloc_function *gim_get_realloc_handler (void);
-gim_free_function  *gim_get_free_handler (void);
-
+gim_realloc_function *gim_get_realloc_handler(void);
+gim_free_function *gim_get_free_handler(void);
 
 ///Standar Memory functions
-void * gim_alloc(size_t size);
-void * gim_alloca(size_t size);
-void * gim_realloc(void *ptr, size_t oldsize, size_t newsize);
+void *gim_alloc(size_t size);
+void *gim_alloca(size_t size);
+void *gim_realloc(void *ptr, size_t oldsize, size_t newsize);
 void gim_free(void *ptr);
 
-
-
-#if defined (_WIN32) && !defined(__MINGW32__) && !defined(__CYGWIN__)
-    #define GIM_SIMD_MEMORY 1
+#if defined(_WIN32) && !defined(__MINGW32__) && !defined(__CYGWIN__)
+#define GIM_SIMD_MEMORY 1
 #endif
 
 //! SIMD POINTER INTEGER
@@ -126,11 +117,10 @@ void gim_free(void *ptr);
 //! SIMD INTEGER SIZE
 #define SIMD_T_SIZE sizeof(SIMD_T)
 
-
-inline void gim_simd_memcpy(void * dst, const void * src, size_t copysize)
+inline void gim_simd_memcpy(void *dst, const void *src, size_t copysize)
 {
 #ifdef GIM_SIMD_MEMORY
-/*
+	/*
 //'long long int' is incompatible with visual studio 6...
     //copy words
     SIMD_T * ui_src_ptr = (SIMD_T *)src;
@@ -143,48 +133,45 @@ inline void gim_simd_memcpy(void * dst, const void * src, size_t copysize)
     if(copysize==0) return;
 */
 
-    char * c_src_ptr = (char *)src;
-    char * c_dst_ptr = (char *)dst;
-    while(copysize>0)
-    {
-        *(c_dst_ptr++) = *(c_src_ptr++);
-        copysize--;
-    }
-    return;
+	char *c_src_ptr = (char *)src;
+	char *c_dst_ptr = (char *)dst;
+	while (copysize > 0)
+	{
+		*(c_dst_ptr++) = *(c_src_ptr++);
+		copysize--;
+	}
+	return;
 #else
-    memcpy(dst,src,copysize);
+	memcpy(dst, src, copysize);
 #endif
 }
 
-
-
-template<class T>
-inline void gim_swap_elements(T* _array,size_t _i,size_t _j)
+template <class T>
+inline void gim_swap_elements(T *_array, size_t _i, size_t _j)
 {
 	T _e_tmp_ = _array[_i];
 	_array[_i] = _array[_j];
 	_array[_j] = _e_tmp_;
 }
 
-
-template<class T>
-inline void gim_swap_elements_memcpy(T* _array,size_t _i,size_t _j)
+template <class T>
+inline void gim_swap_elements_memcpy(T *_array, size_t _i, size_t _j)
 {
 	char _e_tmp_[sizeof(T)];
-	gim_simd_memcpy(_e_tmp_,&_array[_i],sizeof(T));
-	gim_simd_memcpy(&_array[_i],&_array[_j],sizeof(T));
-	gim_simd_memcpy(&_array[_j],_e_tmp_,sizeof(T));
+	gim_simd_memcpy(_e_tmp_, &_array[_i], sizeof(T));
+	gim_simd_memcpy(&_array[_i], &_array[_j], sizeof(T));
+	gim_simd_memcpy(&_array[_j], _e_tmp_, sizeof(T));
 }
 
 template <int SIZE>
-inline void gim_swap_elements_ptr(char * _array,size_t _i,size_t _j)
+inline void gim_swap_elements_ptr(char *_array, size_t _i, size_t _j)
 {
 	char _e_tmp_[SIZE];
-	_i*=SIZE;
-	_j*=SIZE;
-	gim_simd_memcpy(_e_tmp_,_array+_i,SIZE);
-	gim_simd_memcpy(_array+_i,_array+_j,SIZE);
-	gim_simd_memcpy(_array+_j,_e_tmp_,SIZE);
+	_i *= SIZE;
+	_j *= SIZE;
+	gim_simd_memcpy(_e_tmp_, _array + _i, SIZE);
+	gim_simd_memcpy(_array + _i, _array + _j, SIZE);
+	gim_simd_memcpy(_array + _j, _e_tmp_, SIZE);
 }
 
-#endif // GIM_MEMORY_H_INCLUDED
+#endif  // GIM_MEMORY_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/gim_radixsort.h b/thirdparty/bullet/BulletCollision/Gimpact/gim_radixsort.h
index c246ef12543..ff7907adca0 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/gim_radixsort.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/gim_radixsort.h
@@ -40,24 +40,22 @@ email: projectileman@yahoo.com
 //! Prototype for comparators
 class less_comparator
 {
-	public:
-
-	template<class T,class Z>
-	inline int operator() ( const T& a, const Z& b )
+public:
+	template <class T, class Z>
+	inline int operator()(const T& a, const Z& b)
 	{
-		return ( a<b?-1:(a>b?1:0));
+		return (a < b ? -1 : (a > b ? 1 : 0));
 	}
 };
 
 //! Prototype for comparators
 class integer_comparator
 {
-	public:
-
-	template<class T>
-	inline int operator() ( const T& a, const T& b )
+public:
+	template <class T>
+	inline int operator()(const T& a, const T& b)
 	{
-		return (int)(a-b);
+		return (int)(a - b);
 	}
 };
 
@@ -65,20 +63,19 @@ class integer_comparator
 class uint_key_func
 {
 public:
-	template<class T>
-	inline GUINT operator()( const T& a)
+	template <class T>
+	inline GUINT operator()(const T& a)
 	{
 		return (GUINT)a;
 	}
 };
 
-
 //!Prototype for copying elements
 class copy_elements_func
 {
 public:
-	template<class T>
-	inline void operator()(T& a,T& b)
+	template <class T>
+	inline void operator()(T& a, T& b)
 	{
 		a = b;
 	}
@@ -88,34 +85,33 @@ public:
 class memcopy_elements_func
 {
 public:
-	template<class T>
-	inline void operator()(T& a,T& b)
+	template <class T>
+	inline void operator()(T& a, T& b)
 	{
-		gim_simd_memcpy(&a,&b,sizeof(T));
+		gim_simd_memcpy(&a, &b, sizeof(T));
 	}
 };
 
-
 //! @{
 struct GIM_RSORT_TOKEN
 {
-    GUINT m_key;
-    GUINT m_value;
-    GIM_RSORT_TOKEN()
-    {
-    }
-    GIM_RSORT_TOKEN(const GIM_RSORT_TOKEN& rtoken)
-    {
-    	m_key = rtoken.m_key;
-    	m_value = rtoken.m_value;
-    }
+	GUINT m_key;
+	GUINT m_value;
+	GIM_RSORT_TOKEN()
+	{
+	}
+	GIM_RSORT_TOKEN(const GIM_RSORT_TOKEN& rtoken)
+	{
+		m_key = rtoken.m_key;
+		m_value = rtoken.m_value;
+	}
 
-    inline bool operator <(const GIM_RSORT_TOKEN& other) const
+	inline bool operator<(const GIM_RSORT_TOKEN& other) const
 	{
 		return (m_key < other.m_key);
 	}
 
-	inline bool operator >(const GIM_RSORT_TOKEN& other) const
+	inline bool operator>(const GIM_RSORT_TOKEN& other) const
 	{
 		return (m_key > other.m_key);
 	}
@@ -124,33 +120,28 @@ struct GIM_RSORT_TOKEN
 //! Prototype for comparators
 class GIM_RSORT_TOKEN_COMPARATOR
 {
-	public:
-
-	inline int operator()( const GIM_RSORT_TOKEN& a, const GIM_RSORT_TOKEN& b )
+public:
+	inline int operator()(const GIM_RSORT_TOKEN& a, const GIM_RSORT_TOKEN& b)
 	{
 		return (int)((a.m_key) - (b.m_key));
 	}
 };
 
-
-
 #define kHist 2048
 // ---- utils for accessing 11-bit quantities
-#define D11_0(x)	(x & 0x7FF)
-#define D11_1(x)	(x >> 11 & 0x7FF)
-#define D11_2(x)	(x >> 22 )
-
-
+#define D11_0(x) (x & 0x7FF)
+#define D11_1(x) (x >> 11 & 0x7FF)
+#define D11_2(x) (x >> 22)
 
 ///Radix sort for unsigned integer keys
 inline void gim_radix_sort_rtokens(
-				GIM_RSORT_TOKEN * array,
-				GIM_RSORT_TOKEN * sorted, GUINT element_count)
+	GIM_RSORT_TOKEN* array,
+	GIM_RSORT_TOKEN* sorted, GUINT element_count)
 {
 	GUINT i;
 	GUINT b0[kHist * 3];
-	GUINT *b1 = b0 + kHist;
-	GUINT *b2 = b1 + kHist;
+	GUINT* b1 = b0 + kHist;
+	GUINT* b2 = b1 + kHist;
 	for (i = 0; i < kHist * 3; ++i)
 	{
 		b0[i] = 0;
@@ -159,10 +150,10 @@ inline void gim_radix_sort_rtokens(
 	GUINT pos;
 	for (i = 0; i < element_count; ++i)
 	{
-	    fi = array[i].m_key;
-		b0[D11_0(fi)] ++;
-		b1[D11_1(fi)] ++;
-		b2[D11_2(fi)] ++;
+		fi = array[i].m_key;
+		b0[D11_0(fi)]++;
+		b1[D11_1(fi)]++;
+		b2[D11_2(fi)]++;
 	}
 	{
 		GUINT sum0 = 0, sum1 = 0, sum2 = 0;
@@ -182,7 +173,7 @@ inline void gim_radix_sort_rtokens(
 	}
 	for (i = 0; i < element_count; ++i)
 	{
-        fi = array[i].m_key;
+		fi = array[i].m_key;
 		pos = D11_0(fi);
 		pos = ++b0[pos];
 		sorted[pos].m_key = array[i].m_key;
@@ -190,7 +181,7 @@ inline void gim_radix_sort_rtokens(
 	}
 	for (i = 0; i < element_count; ++i)
 	{
-        fi = sorted[i].m_key;
+		fi = sorted[i].m_key;
 		pos = D11_1(fi);
 		pos = ++b1[pos];
 		array[pos].m_key = sorted[i].m_key;
@@ -198,7 +189,7 @@ inline void gim_radix_sort_rtokens(
 	}
 	for (i = 0; i < element_count; ++i)
 	{
-        fi = array[i].m_key;
+		fi = array[i].m_key;
 		pos = D11_2(fi);
 		pos = ++b2[pos];
 		sorted[pos].m_key = array[i].m_key;
@@ -206,9 +197,6 @@ inline void gim_radix_sort_rtokens(
 	}
 }
 
-
-
-
 /// Get the sorted tokens from an array. For generic use. Tokens are IRR_RSORT_TOKEN
 /*!
 *\param array Array of elements to sort
@@ -216,21 +204,21 @@ inline void gim_radix_sort_rtokens(
 *\param element_count element count
 *\param uintkey_macro Functor which retrieves the integer representation of an array element
 */
-template<typename T, class GETKEY_CLASS>
+template <typename T, class GETKEY_CLASS>
 void gim_radix_sort_array_tokens(
-			T* array ,
-			GIM_RSORT_TOKEN * sorted_tokens,
-			GUINT element_count,GETKEY_CLASS uintkey_macro)
+	T* array,
+	GIM_RSORT_TOKEN* sorted_tokens,
+	GUINT element_count, GETKEY_CLASS uintkey_macro)
 {
-	GIM_RSORT_TOKEN * _unsorted = (GIM_RSORT_TOKEN *) gim_alloc(sizeof(GIM_RSORT_TOKEN)*element_count);
-    for (GUINT _i=0;_i<element_count;++_i)
-    {
-        _unsorted[_i].m_key = uintkey_macro(array[_i]);
-        _unsorted[_i].m_value = _i;
-    }
-    gim_radix_sort_rtokens(_unsorted,sorted_tokens,element_count);
-    gim_free(_unsorted);
-    gim_free(_unsorted);
+	GIM_RSORT_TOKEN* _unsorted = (GIM_RSORT_TOKEN*)gim_alloc(sizeof(GIM_RSORT_TOKEN) * element_count);
+	for (GUINT _i = 0; _i < element_count; ++_i)
+	{
+		_unsorted[_i].m_key = uintkey_macro(array[_i]);
+		_unsorted[_i].m_value = _i;
+	}
+	gim_radix_sort_rtokens(_unsorted, sorted_tokens, element_count);
+	gim_free(_unsorted);
+	gim_free(_unsorted);
 }
 
 /// Sorts array in place. For generic use
@@ -241,21 +229,21 @@ void gim_radix_sort_array_tokens(
 \param get_uintkey_macro Macro for extract the Integer value of the element. Similar to SIMPLE_GET_UINTKEY
 \param copy_elements_macro Macro for copy elements, similar to SIMPLE_COPY_ELEMENTS
 */
-template<typename T, class GETKEY_CLASS, class COPY_CLASS>
+template <typename T, class GETKEY_CLASS, class COPY_CLASS>
 void gim_radix_sort(
-	T * array, GUINT element_count,
+	T* array, GUINT element_count,
 	GETKEY_CLASS get_uintkey_macro, COPY_CLASS copy_elements_macro)
 {
-	GIM_RSORT_TOKEN * _sorted = (GIM_RSORT_TOKEN  *) gim_alloc(sizeof(GIM_RSORT_TOKEN)*element_count);
-    gim_radix_sort_array_tokens(array,_sorted,element_count,get_uintkey_macro);
-    T * _original_array = (T *) gim_alloc(sizeof(T)*element_count);
-    gim_simd_memcpy(_original_array,array,sizeof(T)*element_count);
-    for (GUINT _i=0;_i<element_count;++_i)
-    {
-        copy_elements_macro(array[_i],_original_array[_sorted[_i].m_value]);
-    }
-    gim_free(_original_array);
-    gim_free(_sorted);
+	GIM_RSORT_TOKEN* _sorted = (GIM_RSORT_TOKEN*)gim_alloc(sizeof(GIM_RSORT_TOKEN) * element_count);
+	gim_radix_sort_array_tokens(array, _sorted, element_count, get_uintkey_macro);
+	T* _original_array = (T*)gim_alloc(sizeof(T) * element_count);
+	gim_simd_memcpy(_original_array, array, sizeof(T) * element_count);
+	for (GUINT _i = 0; _i < element_count; ++_i)
+	{
+		copy_elements_macro(array[_i], _original_array[_sorted[_i].m_value]);
+	}
+	gim_free(_original_array);
+	gim_free(_sorted);
 }
 
 //! Failsafe Iterative binary search,
@@ -269,20 +257,20 @@ If the element is not found, it returns the nearest upper element position, may
 \param _found If true the value has found. Boolean
 \param _result_index the index of the found element, or if not found then it will get the index of the  closest bigger value
 */
-template<class T, typename KEYCLASS, typename COMP_CLASS>
-bool  gim_binary_search_ex(
-		const T* _array, GUINT _start_i,
-		GUINT _end_i,GUINT & _result_index,
-		const KEYCLASS & _search_key,
-		COMP_CLASS _comp_macro)
+template <class T, typename KEYCLASS, typename COMP_CLASS>
+bool gim_binary_search_ex(
+	const T* _array, GUINT _start_i,
+	GUINT _end_i, GUINT& _result_index,
+	const KEYCLASS& _search_key,
+	COMP_CLASS _comp_macro)
 {
 	GUINT _k;
 	int _comp_result;
 	GUINT _i = _start_i;
-	GUINT _j = _end_i+1;
+	GUINT _j = _end_i + 1;
 	while (_i < _j)
 	{
-		_k = (_j+_i-1)/2;
+		_k = (_j + _i - 1) / 2;
 		_comp_result = _comp_macro(_array[_k], _search_key);
 		if (_comp_result == 0)
 		{
@@ -291,7 +279,7 @@ bool  gim_binary_search_ex(
 		}
 		else if (_comp_result < 0)
 		{
-			_i = _k+1;
+			_i = _k + 1;
 		}
 		else
 		{
@@ -302,8 +290,6 @@ bool  gim_binary_search_ex(
 	return false;
 }
 
-
-
 //! Failsafe Iterative binary search,Template version
 /*!
 If the element is not found, it returns the nearest upper element position, may be the further position after the last element.
@@ -314,26 +300,26 @@ If the element is not found, it returns the nearest upper element position, may
 \param _result_index the index of the found element, or if not found then it will get the index of the  closest bigger value
 \return true if found, else false
 */
-template<class T>
+template <class T>
 bool gim_binary_search(
-	const T*_array,GUINT _start_i,
-	GUINT _end_i,const T & _search_key,
-	GUINT & _result_index)
+	const T* _array, GUINT _start_i,
+	GUINT _end_i, const T& _search_key,
+	GUINT& _result_index)
 {
 	GUINT _i = _start_i;
-	GUINT _j = _end_i+1;
+	GUINT _j = _end_i + 1;
 	GUINT _k;
-	while(_i < _j)
+	while (_i < _j)
 	{
-		_k = (_j+_i-1)/2;
-		if(_array[_k]==_search_key)
+		_k = (_j + _i - 1) / 2;
+		if (_array[_k] == _search_key)
 		{
 			_result_index = _k;
 			return true;
 		}
-		else if (_array[_k]<_search_key)
+		else if (_array[_k] < _search_key)
 		{
-			_i = _k+1;
+			_i = _k + 1;
 		}
 		else
 		{
@@ -344,27 +330,25 @@ bool gim_binary_search(
 	return false;
 }
 
-
-
 ///heap sort from http://www.csse.monash.edu.au/~lloyd/tildeAlgDS/Sort/Heap/
 template <typename T, typename COMP_CLASS>
-void gim_down_heap(T *pArr, GUINT k, GUINT n,COMP_CLASS CompareFunc)
+void gim_down_heap(T* pArr, GUINT k, GUINT n, COMP_CLASS CompareFunc)
 {
 	/*  PRE: a[k+1..N] is a heap */
 	/* POST:  a[k..N]  is a heap */
 
 	T temp = pArr[k - 1];
 	/* k has child(s) */
-	while (k <= n/2)
+	while (k <= n / 2)
 	{
-		int child = 2*k;
+		int child = 2 * k;
 
-		if ((child < (int)n) && CompareFunc(pArr[child - 1] , pArr[child])<0)
+		if ((child < (int)n) && CompareFunc(pArr[child - 1], pArr[child]) < 0)
 		{
 			child++;
 		}
 		/* pick larger child */
-		if (CompareFunc(temp , pArr[child - 1])<0)
+		if (CompareFunc(temp, pArr[child - 1]) < 0)
 		{
 			/* move child up */
 			pArr[k - 1] = pArr[child - 1];
@@ -378,29 +362,25 @@ void gim_down_heap(T *pArr, GUINT k, GUINT n,COMP_CLASS CompareFunc)
 	pArr[k - 1] = temp;
 } /*downHeap*/
 
-
 template <typename T, typename COMP_CLASS>
-void gim_heap_sort(T *pArr, GUINT element_count, COMP_CLASS CompareFunc)
+void gim_heap_sort(T* pArr, GUINT element_count, COMP_CLASS CompareFunc)
 {
 	/* sort a[0..N-1],  N.B. 0 to N-1 */
 	GUINT k;
 	GUINT n = element_count;
-	for (k = n/2; k > 0; k--)
+	for (k = n / 2; k > 0; k--)
 	{
 		gim_down_heap(pArr, k, n, CompareFunc);
 	}
 
 	/* a[1..N] is now a heap */
-	while ( n>=2 )
+	while (n >= 2)
 	{
-		gim_swap_elements(pArr,0,n-1); /* largest of a[0..n-1] */
+		gim_swap_elements(pArr, 0, n - 1); /* largest of a[0..n-1] */
 		--n;
 		/* restore a[1..i-1] heap */
 		gim_down_heap(pArr, 1, n, CompareFunc);
 	}
 }
 
-
-
-
-#endif // GIM_RADIXSORT_H_INCLUDED
+#endif  // GIM_RADIXSORT_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/gim_tri_collision.cpp b/thirdparty/bullet/BulletCollision/Gimpact/gim_tri_collision.cpp
index f9727e1d53b..8d83e95da40 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/gim_tri_collision.cpp
+++ b/thirdparty/bullet/BulletCollision/Gimpact/gim_tri_collision.cpp
@@ -33,15 +33,13 @@ email: projectileman@yahoo.com
 
 #include "gim_tri_collision.h"
 
-
 #define TRI_LOCAL_EPSILON 0.000001f
 #define MIN_EDGE_EDGE_DIS 0.00001f
 
-
 class GIM_TRIANGLE_CALCULATION_CACHE
 {
 public:
-	GREAL margin;	
+	GREAL margin;
 	btVector3 tu_vertices[3];
 	btVector3 tv_vertices[3];
 	btVector4 tu_plane;
@@ -55,46 +53,47 @@ public:
 	GREAL du0du2;
 	GREAL dv[4];
 	GREAL dv0dv1;
-	GREAL dv0dv2;	
+	GREAL dv0dv2;
 	btVector3 temp_points[MAX_TRI_CLIPPING];
 	btVector3 temp_points1[MAX_TRI_CLIPPING];
 	btVector3 contact_points[MAX_TRI_CLIPPING];
-	
-
 
 	//! if returns false, the faces are paralele
 	SIMD_FORCE_INLINE bool compute_intervals(
-					const GREAL &D0,
-					const GREAL &D1,
-					const GREAL &D2,
-					const GREAL &D0D1,
-					const GREAL &D0D2,
-					GREAL & scale_edge0,
-					GREAL & scale_edge1,
-					GUINT &edge_index0,
-					GUINT &edge_index1)
+		const GREAL &D0,
+		const GREAL &D1,
+		const GREAL &D2,
+		const GREAL &D0D1,
+		const GREAL &D0D2,
+		GREAL &scale_edge0,
+		GREAL &scale_edge1,
+		GUINT &edge_index0,
+		GUINT &edge_index1)
 	{
-		if(D0D1>0.0f)
+		if (D0D1 > 0.0f)
 		{
 			/* here we know that D0D2<=0.0 */
 			/* that is D0, D1 are on the same side, D2 on the other or on the plane */
-			scale_edge0 = -D2/(D0-D2);
-			scale_edge1 = -D1/(D2-D1);
-			edge_index0 = 2;edge_index1 = 1;
+			scale_edge0 = -D2 / (D0 - D2);
+			scale_edge1 = -D1 / (D2 - D1);
+			edge_index0 = 2;
+			edge_index1 = 1;
 		}
-		else if(D0D2>0.0f)
+		else if (D0D2 > 0.0f)
 		{
 			/* here we know that d0d1<=0.0 */
-			scale_edge0 = -D0/(D1-D0);
-			scale_edge1 = -D1/(D2-D1);
-			edge_index0 = 0;edge_index1 = 1;
+			scale_edge0 = -D0 / (D1 - D0);
+			scale_edge1 = -D1 / (D2 - D1);
+			edge_index0 = 0;
+			edge_index1 = 1;
 		}
-		else if(D1*D2>0.0f || D0!=0.0f)
+		else if (D1 * D2 > 0.0f || D0 != 0.0f)
 		{
 			/* here we know that d0d1<=0.0 or that D0!=0.0 */
-			scale_edge0 = -D0/(D1-D0);
-			scale_edge1 = -D2/(D0-D2);
-			edge_index0 = 0 ;edge_index1 = 2;
+			scale_edge0 = -D0 / (D1 - D0);
+			scale_edge1 = -D2 / (D0 - D2);
+			edge_index0 = 0;
+			edge_index1 = 2;
 		}
 		else
 		{
@@ -103,46 +102,44 @@ public:
 		return true;
 	}
 
-
 	//! clip triangle
 	/*!
 	*/
 	SIMD_FORCE_INLINE GUINT clip_triangle(
-		const btVector4 & tri_plane,
-		const btVector3 * tripoints,
-		const btVector3 * srcpoints,
-		btVector3 * clip_points)
+		const btVector4 &tri_plane,
+		const btVector3 *tripoints,
+		const btVector3 *srcpoints,
+		btVector3 *clip_points)
 	{
 		// edge 0
 
 		btVector4 edgeplane;
 
-		EDGE_PLANE(tripoints[0],tripoints[1],tri_plane,edgeplane);
+		EDGE_PLANE(tripoints[0], tripoints[1], tri_plane, edgeplane);
 
 		GUINT clipped_count = PLANE_CLIP_TRIANGLE3D(
-			edgeplane,srcpoints[0],srcpoints[1],srcpoints[2],temp_points);
+			edgeplane, srcpoints[0], srcpoints[1], srcpoints[2], temp_points);
 
-		if(clipped_count == 0) return 0;
+		if (clipped_count == 0) return 0;
 
 		// edge 1
 
-		EDGE_PLANE(tripoints[1],tripoints[2],tri_plane,edgeplane);
+		EDGE_PLANE(tripoints[1], tripoints[2], tri_plane, edgeplane);
 
 		clipped_count = PLANE_CLIP_POLYGON3D(
-			edgeplane,temp_points,clipped_count,temp_points1);
+			edgeplane, temp_points, clipped_count, temp_points1);
 
-		if(clipped_count == 0) return 0;
+		if (clipped_count == 0) return 0;
 
 		// edge 2
 
-		EDGE_PLANE(tripoints[2],tripoints[0],tri_plane,edgeplane);
+		EDGE_PLANE(tripoints[2], tripoints[0], tri_plane, edgeplane);
 
 		clipped_count = PLANE_CLIP_POLYGON3D(
-			edgeplane,temp_points1,clipped_count,clip_points);
+			edgeplane, temp_points1, clipped_count, clip_points);
 
 		return clipped_count;
 
-
 		/*GUINT i0 = (tri_plane.closestAxis()+1)%3;
 		GUINT i1 = (i0+1)%3;
 		// edge 0
@@ -172,13 +169,13 @@ public:
 	}
 
 	SIMD_FORCE_INLINE void sort_isect(
-		GREAL & isect0,GREAL & isect1,GUINT &e0,GUINT &e1,btVector3 & vec0,btVector3 & vec1)
+		GREAL &isect0, GREAL &isect1, GUINT &e0, GUINT &e1, btVector3 &vec0, btVector3 &vec1)
 	{
-		if(isect1<isect0)
+		if (isect1 < isect0)
 		{
 			//swap
-			GIM_SWAP_NUMBERS(isect0,isect1);
-			GIM_SWAP_NUMBERS(e0,e1);
+			GIM_SWAP_NUMBERS(isect0, isect1);
+			GIM_SWAP_NUMBERS(e0, e1);
 			btVector3 tmp = vec0;
 			vec0 = vec1;
 			vec1 = tmp;
@@ -202,53 +199,52 @@ public:
 		// Compute direction of intersection line
 		edge_edge_dir = tu_plane.cross(tv_plane);
 		GREAL Dlen;
-		VEC_LENGTH(edge_edge_dir,Dlen);
+		VEC_LENGTH(edge_edge_dir, Dlen);
 
-		if(Dlen<0.0001)
+		if (Dlen < 0.0001)
 		{
-			return 0; //faces near paralele
+			return 0;  //faces near paralele
 		}
 
-		edge_edge_dir*= 1/Dlen;//normalize
-
+		edge_edge_dir *= 1 / Dlen;  //normalize
 
 		// Compute interval for triangle 1
-		GUINT tu_e0,tu_e1;//edge indices
-		GREAL tu_scale_e0,tu_scale_e1;//edge scale
-		if(!compute_intervals(du[0],du[1],du[2],
-			du0du1,du0du2,tu_scale_e0,tu_scale_e1,tu_e0,tu_e1)) return 0;
+		GUINT tu_e0, tu_e1;              //edge indices
+		GREAL tu_scale_e0, tu_scale_e1;  //edge scale
+		if (!compute_intervals(du[0], du[1], du[2],
+							   du0du1, du0du2, tu_scale_e0, tu_scale_e1, tu_e0, tu_e1)) return 0;
 
 		// Compute interval for triangle 2
-		GUINT tv_e0,tv_e1;//edge indices
-		GREAL tv_scale_e0,tv_scale_e1;//edge scale
+		GUINT tv_e0, tv_e1;              //edge indices
+		GREAL tv_scale_e0, tv_scale_e1;  //edge scale
 
-		if(!compute_intervals(dv[0],dv[1],dv[2],
-			dv0dv1,dv0dv2,tv_scale_e0,tv_scale_e1,tv_e0,tv_e1)) return 0;
+		if (!compute_intervals(dv[0], dv[1], dv[2],
+							   dv0dv1, dv0dv2, tv_scale_e0, tv_scale_e1, tv_e0, tv_e1)) return 0;
 
 		//proyected vertices
-		btVector3 up_e0 = tu_vertices[tu_e0].lerp(tu_vertices[(tu_e0+1)%3],tu_scale_e0);
-		btVector3 up_e1 = tu_vertices[tu_e1].lerp(tu_vertices[(tu_e1+1)%3],tu_scale_e1);
+		btVector3 up_e0 = tu_vertices[tu_e0].lerp(tu_vertices[(tu_e0 + 1) % 3], tu_scale_e0);
+		btVector3 up_e1 = tu_vertices[tu_e1].lerp(tu_vertices[(tu_e1 + 1) % 3], tu_scale_e1);
 
-		btVector3 vp_e0 = tv_vertices[tv_e0].lerp(tv_vertices[(tv_e0+1)%3],tv_scale_e0);
-		btVector3 vp_e1 = tv_vertices[tv_e1].lerp(tv_vertices[(tv_e1+1)%3],tv_scale_e1);
+		btVector3 vp_e0 = tv_vertices[tv_e0].lerp(tv_vertices[(tv_e0 + 1) % 3], tv_scale_e0);
+		btVector3 vp_e1 = tv_vertices[tv_e1].lerp(tv_vertices[(tv_e1 + 1) % 3], tv_scale_e1);
 
 		//proyected intervals
-		GREAL isect_u[] = {up_e0.dot(edge_edge_dir),up_e1.dot(edge_edge_dir)};
-		GREAL isect_v[] = {vp_e0.dot(edge_edge_dir),vp_e1.dot(edge_edge_dir)};
+		GREAL isect_u[] = {up_e0.dot(edge_edge_dir), up_e1.dot(edge_edge_dir)};
+		GREAL isect_v[] = {vp_e0.dot(edge_edge_dir), vp_e1.dot(edge_edge_dir)};
 
-		sort_isect(isect_u[0],isect_u[1],tu_e0,tu_e1,up_e0,up_e1);
-		sort_isect(isect_v[0],isect_v[1],tv_e0,tv_e1,vp_e0,vp_e1);
+		sort_isect(isect_u[0], isect_u[1], tu_e0, tu_e1, up_e0, up_e1);
+		sort_isect(isect_v[0], isect_v[1], tv_e0, tv_e1, vp_e0, vp_e1);
 
-		const GREAL midpoint_u = 0.5f*(isect_u[0]+isect_u[1]); // midpoint
-		const GREAL midpoint_v = 0.5f*(isect_v[0]+isect_v[1]); // midpoint
+		const GREAL midpoint_u = 0.5f * (isect_u[0] + isect_u[1]);  // midpoint
+		const GREAL midpoint_v = 0.5f * (isect_v[0] + isect_v[1]);  // midpoint
 
-		if(midpoint_u<midpoint_v)
+		if (midpoint_u < midpoint_v)
 		{
-			if(isect_u[1]>=isect_v[1]) // face U casts face V
+			if (isect_u[1] >= isect_v[1])  // face U casts face V
 			{
 				return 1;
 			}
-			else if(isect_v[0]<=isect_u[0]) // face V casts face U
+			else if (isect_v[0] <= isect_u[0])  // face V casts face U
 			{
 				return 2;
 			}
@@ -257,32 +253,31 @@ public:
 			closest_point_v = vp_e0;
 			// calc edges and separation
 
-			if(isect_u[1]+ MIN_EDGE_EDGE_DIS<isect_v[0]) //calc distance between two lines instead
+			if (isect_u[1] + MIN_EDGE_EDGE_DIS < isect_v[0])  //calc distance between two lines instead
 			{
 				SEGMENT_COLLISION(
-					tu_vertices[tu_e1],tu_vertices[(tu_e1+1)%3],
-					tv_vertices[tv_e0],tv_vertices[(tv_e0+1)%3],
+					tu_vertices[tu_e1], tu_vertices[(tu_e1 + 1) % 3],
+					tv_vertices[tv_e0], tv_vertices[(tv_e0 + 1) % 3],
 					closest_point_u,
 					closest_point_v);
 
-				edge_edge_dir = closest_point_u-closest_point_v;
-				VEC_LENGTH(edge_edge_dir,distances[2]);
-				edge_edge_dir *= 1.0f/distances[2];// normalize
+				edge_edge_dir = closest_point_u - closest_point_v;
+				VEC_LENGTH(edge_edge_dir, distances[2]);
+				edge_edge_dir *= 1.0f / distances[2];  // normalize
 			}
 			else
 			{
-				distances[2] = isect_v[0]-isect_u[1];//distance negative
-				//edge_edge_dir *= -1.0f; //normal pointing from V to U
+				distances[2] = isect_v[0] - isect_u[1];  //distance negative
+														 //edge_edge_dir *= -1.0f; //normal pointing from V to U
 			}
-
 		}
 		else
 		{
-			if(isect_v[1]>=isect_u[1]) // face V casts face U
+			if (isect_v[1] >= isect_u[1])  // face V casts face U
 			{
 				return 2;
 			}
-			else if(isect_u[0]<=isect_v[0]) // face U casts face V
+			else if (isect_u[0] <= isect_v[0])  // face U casts face V
 			{
 				return 1;
 			}
@@ -291,41 +286,39 @@ public:
 			closest_point_v = vp_e1;
 			// calc edges and separation
 
-			if(isect_v[1]+MIN_EDGE_EDGE_DIS<isect_u[0]) //calc distance between two lines instead
+			if (isect_v[1] + MIN_EDGE_EDGE_DIS < isect_u[0])  //calc distance between two lines instead
 			{
 				SEGMENT_COLLISION(
-					tu_vertices[tu_e0],tu_vertices[(tu_e0+1)%3],
-					tv_vertices[tv_e1],tv_vertices[(tv_e1+1)%3],
+					tu_vertices[tu_e0], tu_vertices[(tu_e0 + 1) % 3],
+					tv_vertices[tv_e1], tv_vertices[(tv_e1 + 1) % 3],
 					closest_point_u,
 					closest_point_v);
 
-				edge_edge_dir = closest_point_u-closest_point_v;
-				VEC_LENGTH(edge_edge_dir,distances[2]);
-				edge_edge_dir *= 1.0f/distances[2];// normalize
+				edge_edge_dir = closest_point_u - closest_point_v;
+				VEC_LENGTH(edge_edge_dir, distances[2]);
+				edge_edge_dir *= 1.0f / distances[2];  // normalize
 			}
 			else
 			{
-				distances[2] = isect_u[0]-isect_v[1];//distance negative
-				//edge_edge_dir *= -1.0f; //normal pointing from V to U
+				distances[2] = isect_u[0] - isect_v[1];  //distance negative
+														 //edge_edge_dir *= -1.0f; //normal pointing from V to U
 			}
 		}
 		return 3;
 	}
 
-
 	//! collides by two sides
 	SIMD_FORCE_INLINE bool triangle_collision(
-					const btVector3 & u0,
-					const btVector3 & u1,
-					const btVector3 & u2,
-					GREAL margin_u,
-					const btVector3 & v0,
-					const btVector3 & v1,
-					const btVector3 & v2,
-					GREAL margin_v,
-					GIM_TRIANGLE_CONTACT_DATA & contacts)
+		const btVector3 &u0,
+		const btVector3 &u1,
+		const btVector3 &u2,
+		GREAL margin_u,
+		const btVector3 &v0,
+		const btVector3 &v1,
+		const btVector3 &v2,
+		GREAL margin_v,
+		GIM_TRIANGLE_CONTACT_DATA &contacts)
 	{
-
 		margin = margin_u + margin_v;
 
 		tu_vertices[0] = u0;
@@ -339,103 +332,99 @@ public:
 		//create planes
 		// plane v vs U points
 
-		TRIANGLE_PLANE(tv_vertices[0],tv_vertices[1],tv_vertices[2],tv_plane);
-
-		du[0] = DISTANCE_PLANE_POINT(tv_plane,tu_vertices[0]);
-		du[1] = DISTANCE_PLANE_POINT(tv_plane,tu_vertices[1]);
-		du[2] = DISTANCE_PLANE_POINT(tv_plane,tu_vertices[2]);
+		TRIANGLE_PLANE(tv_vertices[0], tv_vertices[1], tv_vertices[2], tv_plane);
 
+		du[0] = DISTANCE_PLANE_POINT(tv_plane, tu_vertices[0]);
+		du[1] = DISTANCE_PLANE_POINT(tv_plane, tu_vertices[1]);
+		du[2] = DISTANCE_PLANE_POINT(tv_plane, tu_vertices[2]);
 
 		du0du1 = du[0] * du[1];
 		du0du2 = du[0] * du[2];
 
-
-		if(du0du1>0.0f && du0du2>0.0f)	// same sign on all of them + not equal 0 ?
+		if (du0du1 > 0.0f && du0du2 > 0.0f)  // same sign on all of them + not equal 0 ?
 		{
-			if(du[0]<0) //we need test behind the triangle plane
+			if (du[0] < 0)  //we need test behind the triangle plane
 			{
-				distances[0] = GIM_MAX3(du[0],du[1],du[2]);
+				distances[0] = GIM_MAX3(du[0], du[1], du[2]);
 				distances[0] = -distances[0];
-				if(distances[0]>margin) return false; //never intersect
+				if (distances[0] > margin) return false;  //never intersect
 
 				//reorder triangle v
-				VEC_SWAP(tv_vertices[0],tv_vertices[1]);
-				VEC_SCALE_4(tv_plane,-1.0f,tv_plane);
+				VEC_SWAP(tv_vertices[0], tv_vertices[1]);
+				VEC_SCALE_4(tv_plane, -1.0f, tv_plane);
 			}
 			else
 			{
-				distances[0] = GIM_MIN3(du[0],du[1],du[2]);
-				if(distances[0]>margin) return false; //never intersect
+				distances[0] = GIM_MIN3(du[0], du[1], du[2]);
+				if (distances[0] > margin) return false;  //never intersect
 			}
 		}
 		else
 		{
 			//Look if we need to invert the triangle
-			distances[0] = (du[0]+du[1]+du[2])/3.0f; //centroid
+			distances[0] = (du[0] + du[1] + du[2]) / 3.0f;  //centroid
 
-			if(distances[0]<0.0f)
+			if (distances[0] < 0.0f)
 			{
 				//reorder triangle v
-				VEC_SWAP(tv_vertices[0],tv_vertices[1]);
-				VEC_SCALE_4(tv_plane,-1.0f,tv_plane);
+				VEC_SWAP(tv_vertices[0], tv_vertices[1]);
+				VEC_SCALE_4(tv_plane, -1.0f, tv_plane);
 
-				distances[0] = GIM_MAX3(du[0],du[1],du[2]);
+				distances[0] = GIM_MAX3(du[0], du[1], du[2]);
 				distances[0] = -distances[0];
 			}
 			else
 			{
-				distances[0] = GIM_MIN3(du[0],du[1],du[2]);
+				distances[0] = GIM_MIN3(du[0], du[1], du[2]);
 			}
 		}
 
-
 		// plane U vs V points
 
-		TRIANGLE_PLANE(tu_vertices[0],tu_vertices[1],tu_vertices[2],tu_plane);
+		TRIANGLE_PLANE(tu_vertices[0], tu_vertices[1], tu_vertices[2], tu_plane);
 
-		dv[0] = DISTANCE_PLANE_POINT(tu_plane,tv_vertices[0]);
-		dv[1] = DISTANCE_PLANE_POINT(tu_plane,tv_vertices[1]);
-		dv[2] = DISTANCE_PLANE_POINT(tu_plane,tv_vertices[2]);
+		dv[0] = DISTANCE_PLANE_POINT(tu_plane, tv_vertices[0]);
+		dv[1] = DISTANCE_PLANE_POINT(tu_plane, tv_vertices[1]);
+		dv[2] = DISTANCE_PLANE_POINT(tu_plane, tv_vertices[2]);
 
 		dv0dv1 = dv[0] * dv[1];
 		dv0dv2 = dv[0] * dv[2];
 
-
-		if(dv0dv1>0.0f && dv0dv2>0.0f)	// same sign on all of them + not equal 0 ?
+		if (dv0dv1 > 0.0f && dv0dv2 > 0.0f)  // same sign on all of them + not equal 0 ?
 		{
-			if(dv[0]<0) //we need test behind the triangle plane
+			if (dv[0] < 0)  //we need test behind the triangle plane
 			{
-				distances[1] = GIM_MAX3(dv[0],dv[1],dv[2]);
+				distances[1] = GIM_MAX3(dv[0], dv[1], dv[2]);
 				distances[1] = -distances[1];
-				if(distances[1]>margin) return false; //never intersect
+				if (distances[1] > margin) return false;  //never intersect
 
 				//reorder triangle u
-				VEC_SWAP(tu_vertices[0],tu_vertices[1]);
-				VEC_SCALE_4(tu_plane,-1.0f,tu_plane);
+				VEC_SWAP(tu_vertices[0], tu_vertices[1]);
+				VEC_SCALE_4(tu_plane, -1.0f, tu_plane);
 			}
 			else
 			{
-				distances[1] = GIM_MIN3(dv[0],dv[1],dv[2]);
-				if(distances[1]>margin) return false; //never intersect
+				distances[1] = GIM_MIN3(dv[0], dv[1], dv[2]);
+				if (distances[1] > margin) return false;  //never intersect
 			}
 		}
 		else
 		{
 			//Look if we need to invert the triangle
-			distances[1] = (dv[0]+dv[1]+dv[2])/3.0f; //centroid
+			distances[1] = (dv[0] + dv[1] + dv[2]) / 3.0f;  //centroid
 
-			if(distances[1]<0.0f)
+			if (distances[1] < 0.0f)
 			{
 				//reorder triangle v
-				VEC_SWAP(tu_vertices[0],tu_vertices[1]);
-				VEC_SCALE_4(tu_plane,-1.0f,tu_plane);
+				VEC_SWAP(tu_vertices[0], tu_vertices[1]);
+				VEC_SCALE_4(tu_plane, -1.0f, tu_plane);
 
-				distances[1] = GIM_MAX3(dv[0],dv[1],dv[2]);
+				distances[1] = GIM_MAX3(dv[0], dv[1], dv[2]);
 				distances[1] = -distances[1];
 			}
 			else
 			{
-				distances[1] = GIM_MIN3(dv[0],dv[1],dv[2]);
+				distances[1] = GIM_MIN3(dv[0], dv[1], dv[2]);
 			}
 		}
 
@@ -448,47 +437,44 @@ public:
 		}
 		else
 		{*/
-			bl = 0;
-			if(distances[0]<distances[1]) bl = 1;
+		bl = 0;
+		if (distances[0] < distances[1]) bl = 1;
 		//}
 
-		if(bl==2) //edge edge separation
+		if (bl == 2)  //edge edge separation
 		{
-			if(distances[2]>margin) return false;
+			if (distances[2] > margin) return false;
 
 			contacts.m_penetration_depth = -distances[2] + margin;
 			contacts.m_points[0] = closest_point_v;
 			contacts.m_point_count = 1;
-			VEC_COPY(contacts.m_separating_normal,edge_edge_dir);
+			VEC_COPY(contacts.m_separating_normal, edge_edge_dir);
 
 			return true;
 		}
 
 		//clip face against other
 
-		
 		GUINT point_count;
 		//TODO
-		if(bl == 0) //clip U points against V
+		if (bl == 0)  //clip U points against V
 		{
-			point_count = clip_triangle(tv_plane,tv_vertices,tu_vertices,contact_points);
-			if(point_count == 0) return false;						
-			contacts.merge_points(tv_plane,margin,contact_points,point_count);			
+			point_count = clip_triangle(tv_plane, tv_vertices, tu_vertices, contact_points);
+			if (point_count == 0) return false;
+			contacts.merge_points(tv_plane, margin, contact_points, point_count);
 		}
-		else //clip V points against U
+		else  //clip V points against U
 		{
-			point_count = clip_triangle(tu_plane,tu_vertices,tv_vertices,contact_points);
-			if(point_count == 0) return false;			
-			contacts.merge_points(tu_plane,margin,contact_points,point_count);
+			point_count = clip_triangle(tu_plane, tu_vertices, tv_vertices, contact_points);
+			if (point_count == 0) return false;
+			contacts.merge_points(tu_plane, margin, contact_points, point_count);
 			contacts.m_separating_normal *= -1.f;
 		}
-		if(contacts.m_point_count == 0) return false;
+		if (contacts.m_point_count == 0) return false;
 		return true;
 	}
-
 };
 
-
 /*class GIM_TRIANGLE_CALCULATION_CACHE
 {
 public:
@@ -621,20 +607,13 @@ public:
 
 };*/
 
-
-
 bool GIM_TRIANGLE::collide_triangle_hard_test(
-		const GIM_TRIANGLE & other,
-		GIM_TRIANGLE_CONTACT_DATA & contact_data) const
+	const GIM_TRIANGLE &other,
+	GIM_TRIANGLE_CONTACT_DATA &contact_data) const
 {
-	GIM_TRIANGLE_CALCULATION_CACHE calc_cache;	
+	GIM_TRIANGLE_CALCULATION_CACHE calc_cache;
 	return calc_cache.triangle_collision(
-					m_vertices[0],m_vertices[1],m_vertices[2],m_margin,
-					other.m_vertices[0],other.m_vertices[1],other.m_vertices[2],other.m_margin,
-					contact_data);
-
+		m_vertices[0], m_vertices[1], m_vertices[2], m_margin,
+		other.m_vertices[0], other.m_vertices[1], other.m_vertices[2], other.m_margin,
+		contact_data);
 }
-
-
-
-
diff --git a/thirdparty/bullet/BulletCollision/Gimpact/gim_tri_collision.h b/thirdparty/bullet/BulletCollision/Gimpact/gim_tri_collision.h
index 267f806e7e5..e6d4bf54708 100644
--- a/thirdparty/bullet/BulletCollision/Gimpact/gim_tri_collision.h
+++ b/thirdparty/bullet/BulletCollision/Gimpact/gim_tri_collision.h
@@ -36,8 +36,6 @@ email: projectileman@yahoo.com
 #include "gim_box_collision.h"
 #include "gim_clip_polygon.h"
 
-
-
 #ifndef MAX_TRI_CLIPPING
 #define MAX_TRI_CLIPPING 16
 #endif
@@ -45,18 +43,18 @@ email: projectileman@yahoo.com
 //! Structure for collision
 struct GIM_TRIANGLE_CONTACT_DATA
 {
-    GREAL m_penetration_depth;
-    GUINT m_point_count;
-    btVector4 m_separating_normal;
-    btVector3 m_points[MAX_TRI_CLIPPING];
+	GREAL m_penetration_depth;
+	GUINT m_point_count;
+	btVector4 m_separating_normal;
+	btVector3 m_points[MAX_TRI_CLIPPING];
 
-	SIMD_FORCE_INLINE void copy_from(const GIM_TRIANGLE_CONTACT_DATA& other)
+	SIMD_FORCE_INLINE void copy_from(const GIM_TRIANGLE_CONTACT_DATA &other)
 	{
 		m_penetration_depth = other.m_penetration_depth;
 		m_separating_normal = other.m_separating_normal;
 		m_point_count = other.m_point_count;
 		GUINT i = m_point_count;
-		while(i--)
+		while (i--)
 		{
 			m_points[i] = other.m_points[i];
 		}
@@ -66,39 +64,36 @@ struct GIM_TRIANGLE_CONTACT_DATA
 	{
 	}
 
-	GIM_TRIANGLE_CONTACT_DATA(const GIM_TRIANGLE_CONTACT_DATA& other)
+	GIM_TRIANGLE_CONTACT_DATA(const GIM_TRIANGLE_CONTACT_DATA &other)
 	{
 		copy_from(other);
 	}
 
-	
-	
-
-    //! classify points that are closer
-    template<typename DISTANCE_FUNC,typename CLASS_PLANE>
-    SIMD_FORCE_INLINE void mergepoints_generic(const CLASS_PLANE & plane,
-    				GREAL margin, const btVector3 * points, GUINT point_count, DISTANCE_FUNC distance_func)
-    {	
-    	m_point_count = 0;
-    	m_penetration_depth= -1000.0f;
+	//! classify points that are closer
+	template <typename DISTANCE_FUNC, typename CLASS_PLANE>
+	SIMD_FORCE_INLINE void mergepoints_generic(const CLASS_PLANE &plane,
+											   GREAL margin, const btVector3 *points, GUINT point_count, DISTANCE_FUNC distance_func)
+	{
+		m_point_count = 0;
+		m_penetration_depth = -1000.0f;
 
 		GUINT point_indices[MAX_TRI_CLIPPING];
 
 		GUINT _k;
 
-		for(_k=0;_k<point_count;_k++)
+		for (_k = 0; _k < point_count; _k++)
 		{
-			GREAL _dist = -distance_func(plane,points[_k]) + margin;
+			GREAL _dist = -distance_func(plane, points[_k]) + margin;
 
-			if(_dist>=0.0f)
+			if (_dist >= 0.0f)
 			{
-				if(_dist>m_penetration_depth)
+				if (_dist > m_penetration_depth)
 				{
 					m_penetration_depth = _dist;
 					point_indices[0] = _k;
-					m_point_count=1;
+					m_point_count = 1;
 				}
-				else if((_dist+G_EPSILON)>=m_penetration_depth)
+				else if ((_dist + G_EPSILON) >= m_penetration_depth)
 				{
 					point_indices[m_point_count] = _k;
 					m_point_count++;
@@ -106,88 +101,87 @@ struct GIM_TRIANGLE_CONTACT_DATA
 			}
 		}
 
-		for( _k=0;_k<m_point_count;_k++)
+		for (_k = 0; _k < m_point_count; _k++)
 		{
 			m_points[_k] = points[point_indices[_k]];
 		}
 	}
 
 	//! classify points that are closer
-	SIMD_FORCE_INLINE void merge_points(const btVector4 & plane, GREAL margin,
-										 const btVector3 * points, GUINT point_count)
+	SIMD_FORCE_INLINE void merge_points(const btVector4 &plane, GREAL margin,
+										const btVector3 *points, GUINT point_count)
 	{
 		m_separating_normal = plane;
 		mergepoints_generic(plane, margin, points, point_count, DISTANCE_PLANE_3D_FUNC());
 	}
 };
 
-
 //! Class for colliding triangles
 class GIM_TRIANGLE
 {
 public:
 	btScalar m_margin;
-    btVector3 m_vertices[3];
+	btVector3 m_vertices[3];
 
-    GIM_TRIANGLE():m_margin(0.1f)
-    {
-    }
+	GIM_TRIANGLE() : m_margin(0.1f)
+	{
+	}
 
-    SIMD_FORCE_INLINE GIM_AABB get_box()  const
-    {
-    	return GIM_AABB(m_vertices[0],m_vertices[1],m_vertices[2],m_margin);
-    }
+	SIMD_FORCE_INLINE GIM_AABB get_box() const
+	{
+		return GIM_AABB(m_vertices[0], m_vertices[1], m_vertices[2], m_margin);
+	}
 
-    SIMD_FORCE_INLINE void get_normal(btVector3 &normal)  const
-    {
-    	TRIANGLE_NORMAL(m_vertices[0],m_vertices[1],m_vertices[2],normal);
-    }
+	SIMD_FORCE_INLINE void get_normal(btVector3 &normal) const
+	{
+		TRIANGLE_NORMAL(m_vertices[0], m_vertices[1], m_vertices[2], normal);
+	}
 
-    SIMD_FORCE_INLINE void get_plane(btVector4 &plane)  const
-    {
-    	TRIANGLE_PLANE(m_vertices[0],m_vertices[1],m_vertices[2],plane);;
-    }
+	SIMD_FORCE_INLINE void get_plane(btVector4 &plane) const
+	{
+		TRIANGLE_PLANE(m_vertices[0], m_vertices[1], m_vertices[2], plane);
+		;
+	}
 
-    SIMD_FORCE_INLINE void apply_transform(const btTransform & trans)
-    {
-    	m_vertices[0] = trans(m_vertices[0]);
-    	m_vertices[1] = trans(m_vertices[1]);
-    	m_vertices[2] = trans(m_vertices[2]);
-    }
+	SIMD_FORCE_INLINE void apply_transform(const btTransform &trans)
+	{
+		m_vertices[0] = trans(m_vertices[0]);
+		m_vertices[1] = trans(m_vertices[1]);
+		m_vertices[2] = trans(m_vertices[2]);
+	}
 
-    SIMD_FORCE_INLINE void get_edge_plane(GUINT edge_index,const btVector3 &triangle_normal,btVector4 &plane)  const
-    {
-		const btVector3 & e0 = m_vertices[edge_index];
-		const btVector3 & e1 = m_vertices[(edge_index+1)%3];
-		EDGE_PLANE(e0,e1,triangle_normal,plane);
-    }
+	SIMD_FORCE_INLINE void get_edge_plane(GUINT edge_index, const btVector3 &triangle_normal, btVector4 &plane) const
+	{
+		const btVector3 &e0 = m_vertices[edge_index];
+		const btVector3 &e1 = m_vertices[(edge_index + 1) % 3];
+		EDGE_PLANE(e0, e1, triangle_normal, plane);
+	}
 
-    //! Gets the relative transformation of this triangle
-    /*!
+	//! Gets the relative transformation of this triangle
+	/*!
     The transformation is oriented to the triangle normal , and aligned to the 1st edge of this triangle. The position corresponds to vertice 0:
     - triangle normal corresponds to Z axis.
     - 1st normalized edge corresponds to X axis,
 
     */
-    SIMD_FORCE_INLINE void get_triangle_transform(btTransform & triangle_transform)  const
-    {
-    	btMatrix3x3 & matrix = triangle_transform.getBasis();
+	SIMD_FORCE_INLINE void get_triangle_transform(btTransform &triangle_transform) const
+	{
+		btMatrix3x3 &matrix = triangle_transform.getBasis();
 
-    	btVector3 zaxis;
-    	get_normal(zaxis);
-    	MAT_SET_Z(matrix,zaxis);
+		btVector3 zaxis;
+		get_normal(zaxis);
+		MAT_SET_Z(matrix, zaxis);
 
-    	btVector3 xaxis = m_vertices[1] - m_vertices[0];
-    	VEC_NORMALIZE(xaxis);
-    	MAT_SET_X(matrix,xaxis);
+		btVector3 xaxis = m_vertices[1] - m_vertices[0];
+		VEC_NORMALIZE(xaxis);
+		MAT_SET_X(matrix, xaxis);
 
-    	//y axis
-    	xaxis = zaxis.cross(xaxis);
-    	MAT_SET_Y(matrix,xaxis);
-
-    	triangle_transform.setOrigin(m_vertices[0]);
-    }
+		//y axis
+		xaxis = zaxis.cross(xaxis);
+		MAT_SET_Y(matrix, xaxis);
 
+		triangle_transform.setOrigin(m_vertices[0]);
+	}
 
 	//! Test triangles by finding separating axis
 	/*!
@@ -195,8 +189,8 @@ public:
 	\param contact_data Structure for holding contact points, normal and penetration depth; The normal is pointing toward this triangle from the other triangle
 	*/
 	bool collide_triangle_hard_test(
-		const GIM_TRIANGLE & other,
-		GIM_TRIANGLE_CONTACT_DATA & contact_data) const;
+		const GIM_TRIANGLE &other,
+		GIM_TRIANGLE_CONTACT_DATA &contact_data) const;
 
 	//! Test boxes before doing hard test
 	/*!
@@ -205,16 +199,16 @@ public:
 	\
 	*/
 	SIMD_FORCE_INLINE bool collide_triangle(
-		const GIM_TRIANGLE & other,
-		GIM_TRIANGLE_CONTACT_DATA & contact_data) const
+		const GIM_TRIANGLE &other,
+		GIM_TRIANGLE_CONTACT_DATA &contact_data) const
 	{
 		//test box collisioin
-		GIM_AABB boxu(m_vertices[0],m_vertices[1],m_vertices[2],m_margin);
-		GIM_AABB boxv(other.m_vertices[0],other.m_vertices[1],other.m_vertices[2],other.m_margin);
-		if(!boxu.has_collision(boxv)) return false;
+		GIM_AABB boxu(m_vertices[0], m_vertices[1], m_vertices[2], m_margin);
+		GIM_AABB boxv(other.m_vertices[0], other.m_vertices[1], other.m_vertices[2], other.m_margin);
+		if (!boxu.has_collision(boxv)) return false;
 
 		//do hard test
-		return collide_triangle_hard_test(other,contact_data);
+		return collide_triangle_hard_test(other, contact_data);
 	}
 
 	/*!
@@ -246,43 +240,43 @@ if 0.0<= u+v <=1.0 then they are inside of triangle
 	\return false if the point is outside of triangle.This function  doesn't take the margin
 	*/
 	SIMD_FORCE_INLINE bool get_uv_parameters(
-			const btVector3 & point,
-			const btVector3 & tri_plane,
-			GREAL & u, GREAL & v) const
+		const btVector3 &point,
+		const btVector3 &tri_plane,
+		GREAL &u, GREAL &v) const
 	{
-		btVector3 _axe1 = m_vertices[1]-m_vertices[0];
-		btVector3 _axe2 = m_vertices[2]-m_vertices[0];
+		btVector3 _axe1 = m_vertices[1] - m_vertices[0];
+		btVector3 _axe2 = m_vertices[2] - m_vertices[0];
 		btVector3 _vecproj = point - m_vertices[0];
-		GUINT _i1 = (tri_plane.closestAxis()+1)%3;
-		GUINT _i2 = (_i1+1)%3;
-		if(btFabs(_axe2[_i2])<G_EPSILON)
+		GUINT _i1 = (tri_plane.closestAxis() + 1) % 3;
+		GUINT _i2 = (_i1 + 1) % 3;
+		if (btFabs(_axe2[_i2]) < G_EPSILON)
 		{
-			u = (_vecproj[_i2]*_axe2[_i1] - _vecproj[_i1]*_axe2[_i2]) /(_axe1[_i2]*_axe2[_i1]  - _axe1[_i1]*_axe2[_i2]);
-			v = (_vecproj[_i1] - u*_axe1[_i1])/_axe2[_i1];
+			u = (_vecproj[_i2] * _axe2[_i1] - _vecproj[_i1] * _axe2[_i2]) / (_axe1[_i2] * _axe2[_i1] - _axe1[_i1] * _axe2[_i2]);
+			v = (_vecproj[_i1] - u * _axe1[_i1]) / _axe2[_i1];
 		}
 		else
 		{
-			u = (_vecproj[_i1]*_axe2[_i2] - _vecproj[_i2]*_axe2[_i1]) /(_axe1[_i1]*_axe2[_i2]  - _axe1[_i2]*_axe2[_i1]);
-			v = (_vecproj[_i2] - u*_axe1[_i2])/_axe2[_i2];
+			u = (_vecproj[_i1] * _axe2[_i2] - _vecproj[_i2] * _axe2[_i1]) / (_axe1[_i1] * _axe2[_i2] - _axe1[_i2] * _axe2[_i1]);
+			v = (_vecproj[_i2] - u * _axe1[_i2]) / _axe2[_i2];
 		}
 
-		if(u<-G_EPSILON)
+		if (u < -G_EPSILON)
 		{
 			return false;
 		}
-		else if(v<-G_EPSILON)
+		else if (v < -G_EPSILON)
 		{
 			return false;
 		}
 		else
 		{
 			btScalar sumuv;
-			sumuv = u+v;
-			if(sumuv<-G_EPSILON)
+			sumuv = u + v;
+			if (sumuv < -G_EPSILON)
 			{
 				return false;
 			}
-			else if(sumuv-1.0f>G_EPSILON)
+			else if (sumuv - 1.0f > G_EPSILON)
 			{
 				return false;
 			}
@@ -294,50 +288,49 @@ if 0.0<= u+v <=1.0 then they are inside of triangle
 	/*!
 	Test if point is in triangle, with m_margin tolerance
 	*/
-	SIMD_FORCE_INLINE bool is_point_inside(const btVector3 & point, const btVector3 & tri_normal) const
+	SIMD_FORCE_INLINE bool is_point_inside(const btVector3 &point, const btVector3 &tri_normal) const
 	{
 		//Test with edge 0
 		btVector4 edge_plane;
-		this->get_edge_plane(0,tri_normal,edge_plane);
-		GREAL dist = DISTANCE_PLANE_POINT(edge_plane,point);
-		if(dist-m_margin>0.0f) return false; // outside plane
+		this->get_edge_plane(0, tri_normal, edge_plane);
+		GREAL dist = DISTANCE_PLANE_POINT(edge_plane, point);
+		if (dist - m_margin > 0.0f) return false;  // outside plane
 
-		this->get_edge_plane(1,tri_normal,edge_plane);
-		dist = DISTANCE_PLANE_POINT(edge_plane,point);
-		if(dist-m_margin>0.0f) return false; // outside plane
+		this->get_edge_plane(1, tri_normal, edge_plane);
+		dist = DISTANCE_PLANE_POINT(edge_plane, point);
+		if (dist - m_margin > 0.0f) return false;  // outside plane
 
-		this->get_edge_plane(2,tri_normal,edge_plane);
-		dist = DISTANCE_PLANE_POINT(edge_plane,point);
-		if(dist-m_margin>0.0f) return false; // outside plane
+		this->get_edge_plane(2, tri_normal, edge_plane);
+		dist = DISTANCE_PLANE_POINT(edge_plane, point);
+		if (dist - m_margin > 0.0f) return false;  // outside plane
 		return true;
 	}
 
-
 	//! Bidireccional ray collision
 	SIMD_FORCE_INLINE bool ray_collision(
-		const btVector3 & vPoint,
-		const btVector3 & vDir, btVector3 & pout, btVector3 & triangle_normal,
-		GREAL & tparam, GREAL tmax = G_REAL_INFINITY)
+		const btVector3 &vPoint,
+		const btVector3 &vDir, btVector3 &pout, btVector3 &triangle_normal,
+		GREAL &tparam, GREAL tmax = G_REAL_INFINITY)
 	{
 		btVector4 faceplane;
 		{
 			btVector3 dif1 = m_vertices[1] - m_vertices[0];
 			btVector3 dif2 = m_vertices[2] - m_vertices[0];
-    		VEC_CROSS(faceplane,dif1,dif2);
-    		faceplane[3] = m_vertices[0].dot(faceplane);
+			VEC_CROSS(faceplane, dif1, dif2);
+			faceplane[3] = m_vertices[0].dot(faceplane);
 		}
 
-		GUINT res = LINE_PLANE_COLLISION(faceplane,vDir,vPoint,pout,tparam, btScalar(0), tmax);
-		if(res == 0) return false;
-		if(! is_point_inside(pout,faceplane)) return false;
+		GUINT res = LINE_PLANE_COLLISION(faceplane, vDir, vPoint, pout, tparam, btScalar(0), tmax);
+		if (res == 0) return false;
+		if (!is_point_inside(pout, faceplane)) return false;
 
-		if(res==2) //invert normal
+		if (res == 2)  //invert normal
 		{
-			triangle_normal.setValue(-faceplane[0],-faceplane[1],-faceplane[2]);
+			triangle_normal.setValue(-faceplane[0], -faceplane[1], -faceplane[2]);
 		}
 		else
 		{
-			triangle_normal.setValue(faceplane[0],faceplane[1],faceplane[2]);
+			triangle_normal.setValue(faceplane[0], faceplane[1], faceplane[2]);
 		}
 
 		VEC_NORMALIZE(triangle_normal);
@@ -345,36 +338,31 @@ if 0.0<= u+v <=1.0 then they are inside of triangle
 		return true;
 	}
 
-
 	//! one direccion ray collision
 	SIMD_FORCE_INLINE bool ray_collision_front_side(
-		const btVector3 & vPoint,
-		const btVector3 & vDir, btVector3 & pout, btVector3 & triangle_normal,
-		GREAL & tparam, GREAL tmax = G_REAL_INFINITY)
+		const btVector3 &vPoint,
+		const btVector3 &vDir, btVector3 &pout, btVector3 &triangle_normal,
+		GREAL &tparam, GREAL tmax = G_REAL_INFINITY)
 	{
 		btVector4 faceplane;
 		{
 			btVector3 dif1 = m_vertices[1] - m_vertices[0];
 			btVector3 dif2 = m_vertices[2] - m_vertices[0];
-    		VEC_CROSS(faceplane,dif1,dif2);
-    		faceplane[3] = m_vertices[0].dot(faceplane);
+			VEC_CROSS(faceplane, dif1, dif2);
+			faceplane[3] = m_vertices[0].dot(faceplane);
 		}
 
-		GUINT res = LINE_PLANE_COLLISION(faceplane,vDir,vPoint,pout,tparam, btScalar(0), tmax);
-		if(res != 1) return false;
+		GUINT res = LINE_PLANE_COLLISION(faceplane, vDir, vPoint, pout, tparam, btScalar(0), tmax);
+		if (res != 1) return false;
 
-		if(!is_point_inside(pout,faceplane)) return false;
+		if (!is_point_inside(pout, faceplane)) return false;
 
-		triangle_normal.setValue(faceplane[0],faceplane[1],faceplane[2]);
+		triangle_normal.setValue(faceplane[0], faceplane[1], faceplane[2]);
 
 		VEC_NORMALIZE(triangle_normal);
 
 		return true;
 	}
-
 };
 
-
-
-
-#endif // GIM_TRI_COLLISION_H_INCLUDED
+#endif  // GIM_TRI_COLLISION_H_INCLUDED
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btComputeGjkEpaPenetration.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btComputeGjkEpaPenetration.h
index 9eb880b8dfc..c94391d8160 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btComputeGjkEpaPenetration.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btComputeGjkEpaPenetration.h
@@ -16,163 +16,153 @@ subject to the following restrictions:
 #ifndef BT_GJK_EPA_PENETATION_CONVEX_COLLISION_H
 #define BT_GJK_EPA_PENETATION_CONVEX_COLLISION_H
 
-#include "LinearMath/btTransform.h" // Note that btVector3 might be double precision...
+#include "LinearMath/btTransform.h"  // Note that btVector3 might be double precision...
 #include "btGjkEpa3.h"
 #include "btGjkCollisionDescription.h"
 #include "BulletCollision/NarrowPhaseCollision/btVoronoiSimplexSolver.h"
 
-
-
-
-
-
 template <typename btConvexTemplate>
 bool btGjkEpaCalcPenDepth(const btConvexTemplate& a, const btConvexTemplate& b,
-                          const btGjkCollisionDescription& colDesc,
-                          btVector3& v, btVector3& wWitnessOnA, btVector3& wWitnessOnB)
+						  const btGjkCollisionDescription& colDesc,
+						  btVector3& v, btVector3& wWitnessOnA, btVector3& wWitnessOnB)
 {
-    (void)v;
-    
-    //	const btScalar				radialmargin(btScalar(0.));
-    
-    btVector3	guessVector(b.getWorldTransform().getOrigin()-a.getWorldTransform().getOrigin());//?? why not use the GJK input?
-    
-    btGjkEpaSolver3::sResults	results;
+	(void)v;
 
-    
-    if(btGjkEpaSolver3_Penetration(a,b,guessVector,results))
-        
-    {
-        //	debugDraw->drawLine(results.witnesses[1],results.witnesses[1]+results.normal,btVector3(255,0,0));
-        //resultOut->addContactPoint(results.normal,results.witnesses[1],-results.depth);
-        wWitnessOnA = results.witnesses[0];
-        wWitnessOnB = results.witnesses[1];
-        v = results.normal;
-        return true;
-    } else
-    {
-        if(btGjkEpaSolver3_Distance(a,b,guessVector,results))
-        {
-            wWitnessOnA = results.witnesses[0];
-            wWitnessOnB = results.witnesses[1];
-            v = results.normal;
-            return false;
-        }
-    }
-    return false;
+	//	const btScalar				radialmargin(btScalar(0.));
+
+	btVector3 guessVector(b.getWorldTransform().getOrigin() - a.getWorldTransform().getOrigin());  //?? why not use the GJK input?
+
+	btGjkEpaSolver3::sResults results;
+
+	if (btGjkEpaSolver3_Penetration(a, b, guessVector, results))
+
+	{
+		//	debugDraw->drawLine(results.witnesses[1],results.witnesses[1]+results.normal,btVector3(255,0,0));
+		//resultOut->addContactPoint(results.normal,results.witnesses[1],-results.depth);
+		wWitnessOnA = results.witnesses[0];
+		wWitnessOnB = results.witnesses[1];
+		v = results.normal;
+		return true;
+	}
+	else
+	{
+		if (btGjkEpaSolver3_Distance(a, b, guessVector, results))
+		{
+			wWitnessOnA = results.witnesses[0];
+			wWitnessOnB = results.witnesses[1];
+			v = results.normal;
+			return false;
+		}
+	}
+	return false;
 }
 
 template <typename btConvexTemplate, typename btGjkDistanceTemplate>
-int	btComputeGjkEpaPenetration(const btConvexTemplate& a, const btConvexTemplate& b, const btGjkCollisionDescription& colDesc, btVoronoiSimplexSolver& simplexSolver, btGjkDistanceTemplate* distInfo)
+int btComputeGjkEpaPenetration(const btConvexTemplate& a, const btConvexTemplate& b, const btGjkCollisionDescription& colDesc, btVoronoiSimplexSolver& simplexSolver, btGjkDistanceTemplate* distInfo)
 {
-    
-    bool m_catchDegeneracies  = true;
-    btScalar m_cachedSeparatingDistance = 0.f;
-    
-    btScalar distance=btScalar(0.);
-    btVector3	normalInB(btScalar(0.),btScalar(0.),btScalar(0.));
-    
-    btVector3 pointOnA,pointOnB;
-    btTransform	localTransA = a.getWorldTransform();
-    btTransform localTransB = b.getWorldTransform();
-    
-    btScalar marginA = a.getMargin();
-    btScalar marginB = b.getMargin();
-    
-    int m_curIter = 0;
-    int gGjkMaxIter = colDesc.m_maxGjkIterations;//this is to catch invalid input, perhaps check for #NaN?
-    btVector3 m_cachedSeparatingAxis = colDesc.m_firstDir;
-    
-    bool isValid = false;
-    bool checkSimplex = false;
-    bool checkPenetration = true;
-    int m_degenerateSimplex = 0;
-    
-    int m_lastUsedMethod = -1;
-    
-    {
-        btScalar squaredDistance = BT_LARGE_FLOAT;
-        btScalar delta = btScalar(0.);
-        
-        btScalar margin = marginA + marginB;
-        
-        
-        
-        simplexSolver.reset();
-        
-        for ( ; ; )
-            //while (true)
-        {
-            
-            btVector3 seperatingAxisInA = (-m_cachedSeparatingAxis)* localTransA.getBasis();
-            btVector3 seperatingAxisInB = m_cachedSeparatingAxis* localTransB.getBasis();
-            
-            btVector3 pInA = a.getLocalSupportWithoutMargin(seperatingAxisInA);
-            btVector3 qInB = b.getLocalSupportWithoutMargin(seperatingAxisInB);
-            
-            btVector3  pWorld = localTransA(pInA);
-            btVector3  qWorld = localTransB(qInB);
-            
-            
-            
-            btVector3 w	= pWorld - qWorld;
-            delta = m_cachedSeparatingAxis.dot(w);
-            
-            // potential exit, they don't overlap
-            if ((delta > btScalar(0.0)) && (delta * delta > squaredDistance * colDesc.m_maximumDistanceSquared))
-            {
-                m_degenerateSimplex = 10;
-                checkSimplex=true;
-                //checkPenetration = false;
-                break;
-            }
-            
-            //exit 0: the new point is already in the simplex, or we didn't come any closer
-            if (simplexSolver.inSimplex(w))
-            {
-                m_degenerateSimplex = 1;
-                checkSimplex = true;
-                break;
-            }
-            // are we getting any closer ?
-            btScalar f0 = squaredDistance - delta;
-            btScalar f1 = squaredDistance * colDesc.m_gjkRelError2;
-            
-            if (f0 <= f1)
-            {
-                if (f0 <= btScalar(0.))
-                {
-                    m_degenerateSimplex = 2;
-                } else
-                {
-                    m_degenerateSimplex = 11;
-                }
-                checkSimplex = true;
-                break;
-            }
-            
-            //add current vertex to simplex
-            simplexSolver.addVertex(w, pWorld, qWorld);
-            btVector3 newCachedSeparatingAxis;
-            
-            //calculate the closest point to the origin (update vector v)
-            if (!simplexSolver.closest(newCachedSeparatingAxis))
-            {
-                m_degenerateSimplex = 3;
-                checkSimplex = true;
-                break;
-            }
-            
-            if(newCachedSeparatingAxis.length2()<colDesc.m_gjkRelError2)
-            {
-                m_cachedSeparatingAxis = newCachedSeparatingAxis;
-                m_degenerateSimplex = 6;
-                checkSimplex = true;
-                break;
-            }
-            
-            btScalar previousSquaredDistance = squaredDistance;
-            squaredDistance = newCachedSeparatingAxis.length2();
+	bool m_catchDegeneracies = true;
+	btScalar m_cachedSeparatingDistance = 0.f;
+
+	btScalar distance = btScalar(0.);
+	btVector3 normalInB(btScalar(0.), btScalar(0.), btScalar(0.));
+
+	btVector3 pointOnA, pointOnB;
+	btTransform localTransA = a.getWorldTransform();
+	btTransform localTransB = b.getWorldTransform();
+
+	btScalar marginA = a.getMargin();
+	btScalar marginB = b.getMargin();
+
+	int m_curIter = 0;
+	int gGjkMaxIter = colDesc.m_maxGjkIterations;  //this is to catch invalid input, perhaps check for #NaN?
+	btVector3 m_cachedSeparatingAxis = colDesc.m_firstDir;
+
+	bool isValid = false;
+	bool checkSimplex = false;
+	bool checkPenetration = true;
+	int m_degenerateSimplex = 0;
+
+	int m_lastUsedMethod = -1;
+
+	{
+		btScalar squaredDistance = BT_LARGE_FLOAT;
+		btScalar delta = btScalar(0.);
+
+		btScalar margin = marginA + marginB;
+
+		simplexSolver.reset();
+
+		for (;;)
+		//while (true)
+		{
+			btVector3 seperatingAxisInA = (-m_cachedSeparatingAxis) * localTransA.getBasis();
+			btVector3 seperatingAxisInB = m_cachedSeparatingAxis * localTransB.getBasis();
+
+			btVector3 pInA = a.getLocalSupportWithoutMargin(seperatingAxisInA);
+			btVector3 qInB = b.getLocalSupportWithoutMargin(seperatingAxisInB);
+
+			btVector3 pWorld = localTransA(pInA);
+			btVector3 qWorld = localTransB(qInB);
+
+			btVector3 w = pWorld - qWorld;
+			delta = m_cachedSeparatingAxis.dot(w);
+
+			// potential exit, they don't overlap
+			if ((delta > btScalar(0.0)) && (delta * delta > squaredDistance * colDesc.m_maximumDistanceSquared))
+			{
+				m_degenerateSimplex = 10;
+				checkSimplex = true;
+				//checkPenetration = false;
+				break;
+			}
+
+			//exit 0: the new point is already in the simplex, or we didn't come any closer
+			if (simplexSolver.inSimplex(w))
+			{
+				m_degenerateSimplex = 1;
+				checkSimplex = true;
+				break;
+			}
+			// are we getting any closer ?
+			btScalar f0 = squaredDistance - delta;
+			btScalar f1 = squaredDistance * colDesc.m_gjkRelError2;
+
+			if (f0 <= f1)
+			{
+				if (f0 <= btScalar(0.))
+				{
+					m_degenerateSimplex = 2;
+				}
+				else
+				{
+					m_degenerateSimplex = 11;
+				}
+				checkSimplex = true;
+				break;
+			}
+
+			//add current vertex to simplex
+			simplexSolver.addVertex(w, pWorld, qWorld);
+			btVector3 newCachedSeparatingAxis;
+
+			//calculate the closest point to the origin (update vector v)
+			if (!simplexSolver.closest(newCachedSeparatingAxis))
+			{
+				m_degenerateSimplex = 3;
+				checkSimplex = true;
+				break;
+			}
+
+			if (newCachedSeparatingAxis.length2() < colDesc.m_gjkRelError2)
+			{
+				m_cachedSeparatingAxis = newCachedSeparatingAxis;
+				m_degenerateSimplex = 6;
+				checkSimplex = true;
+				break;
+			}
+
+			btScalar previousSquaredDistance = squaredDistance;
+			squaredDistance = newCachedSeparatingAxis.length2();
 #if 0
             ///warning: this termination condition leads to some problems in 2d test case see Bullet/Demos/Box2dDemo
             if (squaredDistance>previousSquaredDistance)
@@ -182,188 +172,183 @@ int	btComputeGjkEpaPenetration(const btConvexTemplate& a, const btConvexTemplate
                 checkSimplex = false;
                 break;
             }
-#endif //
-            
-            
-            //redundant m_simplexSolver->compute_points(pointOnA, pointOnB);
-            
-            //are we getting any closer ?
-            if (previousSquaredDistance - squaredDistance <= SIMD_EPSILON * previousSquaredDistance)
-            {
-                //				m_simplexSolver->backup_closest(m_cachedSeparatingAxis);
-                checkSimplex = true;
-                m_degenerateSimplex = 12;
-                
-                break;
-            }
-            
-            m_cachedSeparatingAxis = newCachedSeparatingAxis;
-            
-            //degeneracy, this is typically due to invalid/uninitialized worldtransforms for a btCollisionObject
-            if (m_curIter++ > gGjkMaxIter)
-            {
-#if defined(DEBUG) || defined (_DEBUG)
-                
-                printf("btGjkPairDetector maxIter exceeded:%i\n",m_curIter);
-                printf("sepAxis=(%f,%f,%f), squaredDistance = %f\n",
-                       m_cachedSeparatingAxis.getX(),
-                       m_cachedSeparatingAxis.getY(),
-                       m_cachedSeparatingAxis.getZ(),
-                       squaredDistance);
+#endif  //
+
+			//redundant m_simplexSolver->compute_points(pointOnA, pointOnB);
+
+			//are we getting any closer ?
+			if (previousSquaredDistance - squaredDistance <= SIMD_EPSILON * previousSquaredDistance)
+			{
+				//				m_simplexSolver->backup_closest(m_cachedSeparatingAxis);
+				checkSimplex = true;
+				m_degenerateSimplex = 12;
+
+				break;
+			}
+
+			m_cachedSeparatingAxis = newCachedSeparatingAxis;
+
+			//degeneracy, this is typically due to invalid/uninitialized worldtransforms for a btCollisionObject
+			if (m_curIter++ > gGjkMaxIter)
+			{
+#if defined(DEBUG) || defined(_DEBUG)
+
+				printf("btGjkPairDetector maxIter exceeded:%i\n", m_curIter);
+				printf("sepAxis=(%f,%f,%f), squaredDistance = %f\n",
+					   m_cachedSeparatingAxis.getX(),
+					   m_cachedSeparatingAxis.getY(),
+					   m_cachedSeparatingAxis.getZ(),
+					   squaredDistance);
 #endif
-                
-                break;
-                
-            }
-            
-            
-            bool check = (!simplexSolver.fullSimplex());
-            //bool check = (!m_simplexSolver->fullSimplex() && squaredDistance > SIMD_EPSILON * m_simplexSolver->maxVertex());
-            
-            if (!check)
-            {
-                //do we need this backup_closest here ?
-                //				m_simplexSolver->backup_closest(m_cachedSeparatingAxis);
-                m_degenerateSimplex = 13;
-                break;
-            }
-        }
-        
-        if (checkSimplex)
-        {
-            simplexSolver.compute_points(pointOnA, pointOnB);
-            normalInB = m_cachedSeparatingAxis;
-            
-            btScalar lenSqr =m_cachedSeparatingAxis.length2();
-            
-            //valid normal
-            if (lenSqr < 0.0001)
-            {
-                m_degenerateSimplex = 5;
-            }
-            if (lenSqr > SIMD_EPSILON*SIMD_EPSILON)
-            {
-                btScalar rlen = btScalar(1.) / btSqrt(lenSqr );
-                normalInB *= rlen; //normalize
-                
-                btScalar s = btSqrt(squaredDistance);
-                
-                btAssert(s > btScalar(0.0));
-                pointOnA -= m_cachedSeparatingAxis * (marginA / s);
-                pointOnB += m_cachedSeparatingAxis * (marginB / s);
-                distance = ((btScalar(1.)/rlen) - margin);
-                isValid = true;
-                
-                m_lastUsedMethod = 1;
-            } else
-            {
-                m_lastUsedMethod = 2;
-            }
-        }
-        
-        bool catchDegeneratePenetrationCase =
-        (m_catchDegeneracies &&  m_degenerateSimplex && ((distance+margin) < 0.01));
-        
-        //if (checkPenetration && !isValid)
-        if (checkPenetration && (!isValid || catchDegeneratePenetrationCase ))
-        {
-            //penetration case
-            
-            //if there is no way to handle penetrations, bail out
-            
-            // Penetration depth case.
-            btVector3 tmpPointOnA,tmpPointOnB;
-            
-            m_cachedSeparatingAxis.setZero();
-            
-            bool isValid2 = btGjkEpaCalcPenDepth(a,b,
-                                                 colDesc,
-                                                 m_cachedSeparatingAxis, tmpPointOnA, tmpPointOnB);
-            
-            if (isValid2)
-            {
-                btVector3 tmpNormalInB = tmpPointOnB-tmpPointOnA;
-                btScalar lenSqr = tmpNormalInB.length2();
-                if (lenSqr <= (SIMD_EPSILON*SIMD_EPSILON))
-                {
-                    tmpNormalInB = m_cachedSeparatingAxis;
-                    lenSqr = m_cachedSeparatingAxis.length2();
-                }
-                
-                if (lenSqr > (SIMD_EPSILON*SIMD_EPSILON))
-                {
-                    tmpNormalInB /= btSqrt(lenSqr);
-                    btScalar distance2 = -(tmpPointOnA-tmpPointOnB).length();
-                    //only replace valid penetrations when the result is deeper (check)
-                    if (!isValid || (distance2 < distance))
-                    {
-                        distance = distance2;
-                        pointOnA = tmpPointOnA;
-                        pointOnB = tmpPointOnB;
-                        normalInB = tmpNormalInB;
-                        
-                        isValid = true;
-                        m_lastUsedMethod = 3;
-                    } else
-                    {
-                        m_lastUsedMethod = 8;
-                    }
-                } else
-                {
-                    m_lastUsedMethod = 9;
-                }
-            } else
-                
-            {
-                ///this is another degenerate case, where the initial GJK calculation reports a degenerate case
-                ///EPA reports no penetration, and the second GJK (using the supporting vector without margin)
-                ///reports a valid positive distance. Use the results of the second GJK instead of failing.
-                ///thanks to Jacob.Langford for the reproduction case
-                ///http://code.google.com/p/bullet/issues/detail?id=250
-                
-                
-                if (m_cachedSeparatingAxis.length2() > btScalar(0.))
-                {
-                    btScalar distance2 = (tmpPointOnA-tmpPointOnB).length()-margin;
-                    //only replace valid distances when the distance is less
-                    if (!isValid || (distance2 < distance))
-                    {
-                        distance = distance2;
-                        pointOnA = tmpPointOnA;
-                        pointOnB = tmpPointOnB;
-                        pointOnA -= m_cachedSeparatingAxis * marginA ;
-                        pointOnB += m_cachedSeparatingAxis * marginB ;
-                        normalInB = m_cachedSeparatingAxis;
-                        normalInB.normalize();
-                        
-                        isValid = true;
-                        m_lastUsedMethod = 6;
-                    } else
-                    {
-                        m_lastUsedMethod = 5;
-                    }
-                }
-            }
-        }
-    }
-    
-    
-    
-    if (isValid && ((distance < 0) || (distance*distance < colDesc.m_maximumDistanceSquared)))
-    {
-        
-        m_cachedSeparatingAxis = normalInB;
-        m_cachedSeparatingDistance = distance;
-        distInfo->m_distance = distance;
-        distInfo->m_normalBtoA = normalInB;
-        distInfo->m_pointOnB = pointOnB;
-        distInfo->m_pointOnA = pointOnB+normalInB*distance;
-        return 0;
-    }
-    return -m_lastUsedMethod;
+
+				break;
+			}
+
+			bool check = (!simplexSolver.fullSimplex());
+			//bool check = (!m_simplexSolver->fullSimplex() && squaredDistance > SIMD_EPSILON * m_simplexSolver->maxVertex());
+
+			if (!check)
+			{
+				//do we need this backup_closest here ?
+				//				m_simplexSolver->backup_closest(m_cachedSeparatingAxis);
+				m_degenerateSimplex = 13;
+				break;
+			}
+		}
+
+		if (checkSimplex)
+		{
+			simplexSolver.compute_points(pointOnA, pointOnB);
+			normalInB = m_cachedSeparatingAxis;
+
+			btScalar lenSqr = m_cachedSeparatingAxis.length2();
+
+			//valid normal
+			if (lenSqr < 0.0001)
+			{
+				m_degenerateSimplex = 5;
+			}
+			if (lenSqr > SIMD_EPSILON * SIMD_EPSILON)
+			{
+				btScalar rlen = btScalar(1.) / btSqrt(lenSqr);
+				normalInB *= rlen;  //normalize
+
+				btScalar s = btSqrt(squaredDistance);
+
+				btAssert(s > btScalar(0.0));
+				pointOnA -= m_cachedSeparatingAxis * (marginA / s);
+				pointOnB += m_cachedSeparatingAxis * (marginB / s);
+				distance = ((btScalar(1.) / rlen) - margin);
+				isValid = true;
+
+				m_lastUsedMethod = 1;
+			}
+			else
+			{
+				m_lastUsedMethod = 2;
+			}
+		}
+
+		bool catchDegeneratePenetrationCase =
+			(m_catchDegeneracies && m_degenerateSimplex && ((distance + margin) < 0.01));
+
+		//if (checkPenetration && !isValid)
+		if (checkPenetration && (!isValid || catchDegeneratePenetrationCase))
+		{
+			//penetration case
+
+			//if there is no way to handle penetrations, bail out
+
+			// Penetration depth case.
+			btVector3 tmpPointOnA, tmpPointOnB;
+
+			m_cachedSeparatingAxis.setZero();
+
+			bool isValid2 = btGjkEpaCalcPenDepth(a, b,
+												 colDesc,
+												 m_cachedSeparatingAxis, tmpPointOnA, tmpPointOnB);
+
+			if (isValid2)
+			{
+				btVector3 tmpNormalInB = tmpPointOnB - tmpPointOnA;
+				btScalar lenSqr = tmpNormalInB.length2();
+				if (lenSqr <= (SIMD_EPSILON * SIMD_EPSILON))
+				{
+					tmpNormalInB = m_cachedSeparatingAxis;
+					lenSqr = m_cachedSeparatingAxis.length2();
+				}
+
+				if (lenSqr > (SIMD_EPSILON * SIMD_EPSILON))
+				{
+					tmpNormalInB /= btSqrt(lenSqr);
+					btScalar distance2 = -(tmpPointOnA - tmpPointOnB).length();
+					//only replace valid penetrations when the result is deeper (check)
+					if (!isValid || (distance2 < distance))
+					{
+						distance = distance2;
+						pointOnA = tmpPointOnA;
+						pointOnB = tmpPointOnB;
+						normalInB = tmpNormalInB;
+
+						isValid = true;
+						m_lastUsedMethod = 3;
+					}
+					else
+					{
+						m_lastUsedMethod = 8;
+					}
+				}
+				else
+				{
+					m_lastUsedMethod = 9;
+				}
+			}
+			else
+
+			{
+				///this is another degenerate case, where the initial GJK calculation reports a degenerate case
+				///EPA reports no penetration, and the second GJK (using the supporting vector without margin)
+				///reports a valid positive distance. Use the results of the second GJK instead of failing.
+				///thanks to Jacob.Langford for the reproduction case
+				///http://code.google.com/p/bullet/issues/detail?id=250
+
+				if (m_cachedSeparatingAxis.length2() > btScalar(0.))
+				{
+					btScalar distance2 = (tmpPointOnA - tmpPointOnB).length() - margin;
+					//only replace valid distances when the distance is less
+					if (!isValid || (distance2 < distance))
+					{
+						distance = distance2;
+						pointOnA = tmpPointOnA;
+						pointOnB = tmpPointOnB;
+						pointOnA -= m_cachedSeparatingAxis * marginA;
+						pointOnB += m_cachedSeparatingAxis * marginB;
+						normalInB = m_cachedSeparatingAxis;
+						normalInB.normalize();
+
+						isValid = true;
+						m_lastUsedMethod = 6;
+					}
+					else
+					{
+						m_lastUsedMethod = 5;
+					}
+				}
+			}
+		}
+	}
+
+	if (isValid && ((distance < 0) || (distance * distance < colDesc.m_maximumDistanceSquared)))
+	{
+		m_cachedSeparatingAxis = normalInB;
+		m_cachedSeparatingDistance = distance;
+		distInfo->m_distance = distance;
+		distInfo->m_normalBtoA = normalInB;
+		distInfo->m_pointOnB = pointOnB;
+		distInfo->m_pointOnA = pointOnB + normalInB * distance;
+		return 0;
+	}
+	return -m_lastUsedMethod;
 }
 
-
-
-
-#endif //BT_GJK_EPA_PENETATION_CONVEX_COLLISION_H
+#endif  //BT_GJK_EPA_PENETATION_CONVEX_COLLISION_H
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btContinuousConvexCollision.cpp b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btContinuousConvexCollision.cpp
index 3481fec8506..38df8d4808e 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btContinuousConvexCollision.cpp
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btContinuousConvexCollision.cpp
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btContinuousConvexCollision.h"
 #include "BulletCollision/CollisionShapes/btConvexShape.h"
 #include "BulletCollision/NarrowPhaseCollision/btSimplexSolverInterface.h"
@@ -24,59 +23,60 @@ subject to the following restrictions:
 #include "btPointCollector.h"
 #include "BulletCollision/CollisionShapes/btStaticPlaneShape.h"
 
-
-
-btContinuousConvexCollision::btContinuousConvexCollision ( const btConvexShape*	convexA,const btConvexShape*	convexB,btSimplexSolverInterface* simplexSolver, btConvexPenetrationDepthSolver* penetrationDepthSolver)
-:m_simplexSolver(simplexSolver),
-m_penetrationDepthSolver(penetrationDepthSolver),
-m_convexA(convexA),m_convexB1(convexB),m_planeShape(0)
+btContinuousConvexCollision::btContinuousConvexCollision(const btConvexShape* convexA, const btConvexShape* convexB, btSimplexSolverInterface* simplexSolver, btConvexPenetrationDepthSolver* penetrationDepthSolver)
+	: m_simplexSolver(simplexSolver),
+	  m_penetrationDepthSolver(penetrationDepthSolver),
+	  m_convexA(convexA),
+	  m_convexB1(convexB),
+	  m_planeShape(0)
 {
 }
 
-
-btContinuousConvexCollision::btContinuousConvexCollision( const btConvexShape*	convexA,const btStaticPlaneShape*	plane)
-:m_simplexSolver(0),
-m_penetrationDepthSolver(0),
-m_convexA(convexA),m_convexB1(0),m_planeShape(plane)
+btContinuousConvexCollision::btContinuousConvexCollision(const btConvexShape* convexA, const btStaticPlaneShape* plane)
+	: m_simplexSolver(0),
+	  m_penetrationDepthSolver(0),
+	  m_convexA(convexA),
+	  m_convexB1(0),
+	  m_planeShape(plane)
 {
 }
 
-
 /// This maximum should not be necessary. It allows for untested/degenerate cases in production code.
 /// You don't want your game ever to lock-up.
 #define MAX_ITERATIONS 64
 
-void btContinuousConvexCollision::computeClosestPoints( const btTransform& transA, const btTransform& transB,btPointCollector& pointCollector)
+void btContinuousConvexCollision::computeClosestPoints(const btTransform& transA, const btTransform& transB, btPointCollector& pointCollector)
 {
 	if (m_convexB1)
 	{
 		m_simplexSolver->reset();
-		btGjkPairDetector gjk(m_convexA,m_convexB1,m_convexA->getShapeType(),m_convexB1->getShapeType(),m_convexA->getMargin(),m_convexB1->getMargin(),m_simplexSolver,m_penetrationDepthSolver);		
+		btGjkPairDetector gjk(m_convexA, m_convexB1, m_convexA->getShapeType(), m_convexB1->getShapeType(), m_convexA->getMargin(), m_convexB1->getMargin(), m_simplexSolver, m_penetrationDepthSolver);
 		btGjkPairDetector::ClosestPointInput input;
 		input.m_transformA = transA;
 		input.m_transformB = transB;
-		gjk.getClosestPoints(input,pointCollector,0);
-	} else
+		gjk.getClosestPoints(input, pointCollector, 0);
+	}
+	else
 	{
 		//convex versus plane
 		const btConvexShape* convexShape = m_convexA;
 		const btStaticPlaneShape* planeShape = m_planeShape;
-		
+
 		const btVector3& planeNormal = planeShape->getPlaneNormal();
 		const btScalar& planeConstant = planeShape->getPlaneConstant();
-		
+
 		btTransform convexWorldTransform = transA;
 		btTransform convexInPlaneTrans;
-		convexInPlaneTrans= transB.inverse() * convexWorldTransform;
+		convexInPlaneTrans = transB.inverse() * convexWorldTransform;
 		btTransform planeInConvex;
-		planeInConvex= convexWorldTransform.inverse() * transB;
-		
-		btVector3 vtx = convexShape->localGetSupportingVertex(planeInConvex.getBasis()*-planeNormal);
+		planeInConvex = convexWorldTransform.inverse() * transB;
+
+		btVector3 vtx = convexShape->localGetSupportingVertex(planeInConvex.getBasis() * -planeNormal);
 
 		btVector3 vtxInPlane = convexInPlaneTrans(vtx);
 		btScalar distance = (planeNormal.dot(vtxInPlane) - planeConstant);
 
-		btVector3 vtxInPlaneProjected = vtxInPlane - distance*planeNormal;
+		btVector3 vtxInPlaneProjected = vtxInPlane - distance * planeNormal;
 		btVector3 vtxInPlaneWorld = transB * vtxInPlaneProjected;
 		btVector3 normalOnSurfaceB = transB.getBasis() * planeNormal;
 
@@ -87,36 +87,33 @@ void btContinuousConvexCollision::computeClosestPoints( const btTransform& trans
 	}
 }
 
-bool	btContinuousConvexCollision::calcTimeOfImpact(
-				const btTransform& fromA,
-				const btTransform& toA,
-				const btTransform& fromB,
-				const btTransform& toB,
-				CastResult& result)
+bool btContinuousConvexCollision::calcTimeOfImpact(
+	const btTransform& fromA,
+	const btTransform& toA,
+	const btTransform& fromB,
+	const btTransform& toB,
+	CastResult& result)
 {
-
-
 	/// compute linear and angular velocity for this interval, to interpolate
-	btVector3 linVelA,angVelA,linVelB,angVelB;
-	btTransformUtil::calculateVelocity(fromA,toA,btScalar(1.),linVelA,angVelA);
-	btTransformUtil::calculateVelocity(fromB,toB,btScalar(1.),linVelB,angVelB);
-
+	btVector3 linVelA, angVelA, linVelB, angVelB;
+	btTransformUtil::calculateVelocity(fromA, toA, btScalar(1.), linVelA, angVelA);
+	btTransformUtil::calculateVelocity(fromB, toB, btScalar(1.), linVelB, angVelB);
 
 	btScalar boundingRadiusA = m_convexA->getAngularMotionDisc();
-	btScalar boundingRadiusB = m_convexB1?m_convexB1->getAngularMotionDisc():0.f;
+	btScalar boundingRadiusB = m_convexB1 ? m_convexB1->getAngularMotionDisc() : 0.f;
 
 	btScalar maxAngularProjectedVelocity = angVelA.length() * boundingRadiusA + angVelB.length() * boundingRadiusB;
-	btVector3 relLinVel = (linVelB-linVelA);
+	btVector3 relLinVel = (linVelB - linVelA);
 
-	btScalar relLinVelocLength = (linVelB-linVelA).length();
-	
-	if ((relLinVelocLength+maxAngularProjectedVelocity) == 0.f)
+	btScalar relLinVelocLength = (linVelB - linVelA).length();
+
+	if ((relLinVelocLength + maxAngularProjectedVelocity) == 0.f)
 		return false;
 
 	btScalar lambda = btScalar(0.);
 
 	btVector3 n;
-	n.setValue(btScalar(0.),btScalar(0.),btScalar(0.));
+	n.setValue(btScalar(0.), btScalar(0.), btScalar(0.));
 	bool hasResult = false;
 	btVector3 c;
 
@@ -126,14 +123,13 @@ bool	btContinuousConvexCollision::calcTimeOfImpact(
 	int numIter = 0;
 	//first solution, using GJK
 
-
 	btScalar radius = 0.001f;
-//	result.drawCoordSystem(sphereTr);
+	//	result.drawCoordSystem(sphereTr);
 
-	btPointCollector	pointCollector1;
+	btPointCollector pointCollector1;
 
-	{	
-		computeClosestPoints(fromA,fromB,pointCollector1);
+	{
+		computeClosestPoints(fromA, fromB, pointCollector1);
 
 		hasResult = pointCollector1.m_hasResult;
 		c = pointCollector1.m_pointInWorld;
@@ -145,7 +141,7 @@ bool	btContinuousConvexCollision::calcTimeOfImpact(
 		dist = pointCollector1.m_distance + result.m_allowedPenetration;
 		n = pointCollector1.m_normalOnBInWorld;
 		btScalar projectedLinearVelocity = relLinVel.dot(n);
-		if ((projectedLinearVelocity+ maxAngularProjectedVelocity)<=SIMD_EPSILON)
+		if ((projectedLinearVelocity + maxAngularProjectedVelocity) <= SIMD_EPSILON)
 			return false;
 
 		//not close enough
@@ -153,18 +149,17 @@ bool	btContinuousConvexCollision::calcTimeOfImpact(
 		{
 			if (result.m_debugDrawer)
 			{
-				result.m_debugDrawer->drawSphere(c,0.2f,btVector3(1,1,1));
+				result.m_debugDrawer->drawSphere(c, 0.2f, btVector3(1, 1, 1));
 			}
 			btScalar dLambda = btScalar(0.);
 
 			projectedLinearVelocity = relLinVel.dot(n);
 
-			
 			//don't report time of impact for motion away from the contact normal (or causes minor penetration)
-			if ((projectedLinearVelocity+ maxAngularProjectedVelocity)<=SIMD_EPSILON)
+			if ((projectedLinearVelocity + maxAngularProjectedVelocity) <= SIMD_EPSILON)
 				return false;
-			
-			dLambda = dist / (projectedLinearVelocity+ maxAngularProjectedVelocity);
+
+			dLambda = dist / (projectedLinearVelocity + maxAngularProjectedVelocity);
 
 			lambda += dLambda;
 
@@ -181,28 +176,29 @@ bool	btContinuousConvexCollision::calcTimeOfImpact(
 			lastLambda = lambda;
 
 			//interpolate to next lambda
-			btTransform interpolatedTransA,interpolatedTransB,relativeTrans;
+			btTransform interpolatedTransA, interpolatedTransB, relativeTrans;
 
-			btTransformUtil::integrateTransform(fromA,linVelA,angVelA,lambda,interpolatedTransA);
-			btTransformUtil::integrateTransform(fromB,linVelB,angVelB,lambda,interpolatedTransB);
+			btTransformUtil::integrateTransform(fromA, linVelA, angVelA, lambda, interpolatedTransA);
+			btTransformUtil::integrateTransform(fromB, linVelB, angVelB, lambda, interpolatedTransB);
 			relativeTrans = interpolatedTransB.inverseTimes(interpolatedTransA);
 
 			if (result.m_debugDrawer)
 			{
-				result.m_debugDrawer->drawSphere(interpolatedTransA.getOrigin(),0.2f,btVector3(1,0,0));
+				result.m_debugDrawer->drawSphere(interpolatedTransA.getOrigin(), 0.2f, btVector3(1, 0, 0));
 			}
 
-			result.DebugDraw( lambda );
+			result.DebugDraw(lambda);
 
-			btPointCollector	pointCollector;
-			computeClosestPoints(interpolatedTransA,interpolatedTransB,pointCollector);
+			btPointCollector pointCollector;
+			computeClosestPoints(interpolatedTransA, interpolatedTransB, pointCollector);
 
 			if (pointCollector.m_hasResult)
 			{
-				dist = pointCollector.m_distance+result.m_allowedPenetration;
-				c = pointCollector.m_pointInWorld;		
+				dist = pointCollector.m_distance + result.m_allowedPenetration;
+				c = pointCollector.m_pointInWorld;
 				n = pointCollector.m_normalOnBInWorld;
-			} else
+			}
+			else
 			{
 				result.reportFailure(-1, numIter);
 				return false;
@@ -215,7 +211,7 @@ bool	btContinuousConvexCollision::calcTimeOfImpact(
 				return false;
 			}
 		}
-	
+
 		result.m_fraction = lambda;
 		result.m_normal = n;
 		result.m_hitPoint = c;
@@ -224,4 +220,3 @@ bool	btContinuousConvexCollision::calcTimeOfImpact(
 
 	return false;
 }
-
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btContinuousConvexCollision.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btContinuousConvexCollision.h
index 528b5e01014..67b2205c366 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btContinuousConvexCollision.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btContinuousConvexCollision.h
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef BT_CONTINUOUS_COLLISION_CONVEX_CAST_H
 #define BT_CONTINUOUS_COLLISION_CONVEX_CAST_H
 
@@ -30,30 +29,25 @@ class btStaticPlaneShape;
 class btContinuousConvexCollision : public btConvexCast
 {
 	btSimplexSolverInterface* m_simplexSolver;
-	btConvexPenetrationDepthSolver*	m_penetrationDepthSolver;
-	const btConvexShape*	m_convexA;
+	btConvexPenetrationDepthSolver* m_penetrationDepthSolver;
+	const btConvexShape* m_convexA;
 	//second object is either a convex or a plane (code sharing)
-	const btConvexShape*	m_convexB1;
-	const btStaticPlaneShape*	m_planeShape;
+	const btConvexShape* m_convexB1;
+	const btStaticPlaneShape* m_planeShape;
 
-	void computeClosestPoints( const btTransform& transA, const btTransform& transB,struct btPointCollector& pointCollector);
+	void computeClosestPoints(const btTransform& transA, const btTransform& transB, struct btPointCollector& pointCollector);
 
 public:
+	btContinuousConvexCollision(const btConvexShape* shapeA, const btConvexShape* shapeB, btSimplexSolverInterface* simplexSolver, btConvexPenetrationDepthSolver* penetrationDepthSolver);
 
-	btContinuousConvexCollision (const btConvexShape*	shapeA,const btConvexShape*	shapeB ,btSimplexSolverInterface* simplexSolver,btConvexPenetrationDepthSolver* penetrationDepthSolver);
-
-	btContinuousConvexCollision(const btConvexShape*	shapeA,const btStaticPlaneShape*	plane );
-
-	virtual bool	calcTimeOfImpact(
-				const btTransform& fromA,
-				const btTransform& toA,
-				const btTransform& fromB,
-				const btTransform& toB,
-				CastResult& result);
-
+	btContinuousConvexCollision(const btConvexShape* shapeA, const btStaticPlaneShape* plane);
 
+	virtual bool calcTimeOfImpact(
+		const btTransform& fromA,
+		const btTransform& toA,
+		const btTransform& fromB,
+		const btTransform& toB,
+		CastResult& result);
 };
 
-
-#endif //BT_CONTINUOUS_COLLISION_CONVEX_CAST_H
-
+#endif  //BT_CONTINUOUS_COLLISION_CONVEX_CAST_H
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btConvexCast.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btConvexCast.h
index bfd79d03beb..76f54699c5f 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btConvexCast.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btConvexCast.h
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef BT_CONVEX_CAST_H
 #define BT_CONVEX_CAST_H
 
@@ -23,51 +22,69 @@ subject to the following restrictions:
 class btMinkowskiSumShape;
 #include "LinearMath/btIDebugDraw.h"
 
+#ifdef BT_USE_DOUBLE_PRECISION
+#define MAX_ITERATIONS 64
+#define MAX_EPSILON (SIMD_EPSILON * 10)
+#else
+#define MAX_ITERATIONS 32
+#define MAX_EPSILON btScalar(0.0001)
+#endif
+///Typically the conservative advancement reaches solution in a few iterations, clip it to 32 for degenerate cases.
+///See discussion about this here http://continuousphysics.com/Bullet/phpBB2/viewtopic.php?t=565
+//will need to digg deeper to make the algorithm more robust
+//since, a large epsilon can cause an early termination with false
+//positive results (ray intersections that shouldn't be there)
+
 /// btConvexCast is an interface for Casting
 class btConvexCast
 {
 public:
-
-
 	virtual ~btConvexCast();
 
 	///RayResult stores the closest result
 	/// alternatively, add a callback method to decide about closest/all results
-	struct	CastResult
+	struct CastResult
 	{
 		//virtual bool	addRayResult(const btVector3& normal,btScalar	fraction) = 0;
-				
-		virtual void	DebugDraw(btScalar	fraction) {(void)fraction;}
-		virtual void	drawCoordSystem(const btTransform& trans) {(void)trans;}
-		virtual void	reportFailure(int errNo, int numIterations) {(void)errNo;(void)numIterations;}
+
+		virtual void DebugDraw(btScalar fraction) { (void)fraction; }
+		virtual void drawCoordSystem(const btTransform& trans) { (void)trans; }
+		virtual void reportFailure(int errNo, int numIterations)
+		{
+			(void)errNo;
+			(void)numIterations;
+		}
 		CastResult()
-			:m_fraction(btScalar(BT_LARGE_FLOAT)),
-			m_debugDrawer(0),
-			m_allowedPenetration(btScalar(0))
+			: m_fraction(btScalar(BT_LARGE_FLOAT)),
+			  m_debugDrawer(0),
+			  m_allowedPenetration(btScalar(0)),
+			  m_subSimplexCastMaxIterations(MAX_ITERATIONS),
+			  m_subSimplexCastEpsilon(MAX_EPSILON)
 		{
 		}
 
+		virtual ~CastResult(){};
 
-		virtual ~CastResult() {};
-
-		btTransform	m_hitTransformA;
-		btTransform	m_hitTransformB;
-		btVector3	m_normal;
-		btVector3   m_hitPoint;
-		btScalar	m_fraction; //input and output
+		btTransform m_hitTransformA;
+		btTransform m_hitTransformB;
+		btVector3 m_normal;
+		btVector3 m_hitPoint;
+		btScalar m_fraction;  //input and output
 		btIDebugDraw* m_debugDrawer;
-		btScalar	m_allowedPenetration;
+		btScalar m_allowedPenetration;
+		
+		int m_subSimplexCastMaxIterations;
+		btScalar m_subSimplexCastEpsilon;
 
 	};
 
-
 	/// cast a convex against another convex object
-	virtual bool	calcTimeOfImpact(
-					const btTransform& fromA,
-					const btTransform& toA,
-					const btTransform& fromB,
-					const btTransform& toB,
-					CastResult& result) = 0;
+	virtual bool calcTimeOfImpact(
+		const btTransform& fromA,
+		const btTransform& toA,
+		const btTransform& fromB,
+		const btTransform& toB,
+		CastResult& result) = 0;
 };
 
-#endif //BT_CONVEX_CAST_H
+#endif  //BT_CONVEX_CAST_H
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btConvexPenetrationDepthSolver.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btConvexPenetrationDepthSolver.h
index 29620abffb9..65c9df9340b 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btConvexPenetrationDepthSolver.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btConvexPenetrationDepthSolver.h
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef BT_CONVEX_PENETRATION_DEPTH_H
 #define BT_CONVEX_PENETRATION_DEPTH_H
 
@@ -25,16 +24,12 @@ class btTransform;
 ///ConvexPenetrationDepthSolver provides an interface for penetration depth calculation.
 class btConvexPenetrationDepthSolver
 {
-public:	
-	
-	virtual ~btConvexPenetrationDepthSolver() {};
-	virtual bool calcPenDepth( btSimplexSolverInterface& simplexSolver,
-		const btConvexShape* convexA,const btConvexShape* convexB,
-					const btTransform& transA,const btTransform& transB,
-				btVector3& v, btVector3& pa, btVector3& pb,
-				class btIDebugDraw* debugDraw) = 0;
-
-
+public:
+	virtual ~btConvexPenetrationDepthSolver(){};
+	virtual bool calcPenDepth(btSimplexSolverInterface& simplexSolver,
+							  const btConvexShape* convexA, const btConvexShape* convexB,
+							  const btTransform& transA, const btTransform& transB,
+							  btVector3& v, btVector3& pa, btVector3& pb,
+							  class btIDebugDraw* debugDraw) = 0;
 };
-#endif //BT_CONVEX_PENETRATION_DEPTH_H
-
+#endif  //BT_CONVEX_PENETRATION_DEPTH_H
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btDiscreteCollisionDetectorInterface.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btDiscreteCollisionDetectorInterface.h
index 0ea7b483cfb..d1bbb1a46ee 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btDiscreteCollisionDetectorInterface.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btDiscreteCollisionDetectorInterface.h
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef BT_DISCRETE_COLLISION_DETECTOR1_INTERFACE_H
 #define BT_DISCRETE_COLLISION_DETECTOR1_INTERFACE_H
 
@@ -27,64 +26,60 @@ subject to the following restrictions:
 /// by taking closestPointInA = closestPointInB + m_distance * m_normalOnSurfaceB
 struct btDiscreteCollisionDetectorInterface
 {
-	
 	struct Result
 	{
-	
-		virtual ~Result(){}	
+		virtual ~Result() {}
 
 		///setShapeIdentifiersA/B provides experimental support for per-triangle material / custom material combiner
-		virtual void setShapeIdentifiersA(int partId0,int index0)=0;
-		virtual void setShapeIdentifiersB(int partId1,int index1)=0;
-		virtual void addContactPoint(const btVector3& normalOnBInWorld,const btVector3& pointInWorld,btScalar depth)=0;
+		virtual void setShapeIdentifiersA(int partId0, int index0) = 0;
+		virtual void setShapeIdentifiersB(int partId1, int index1) = 0;
+		virtual void addContactPoint(const btVector3& normalOnBInWorld, const btVector3& pointInWorld, btScalar depth) = 0;
 	};
 
 	struct ClosestPointInput
 	{
 		ClosestPointInput()
-			:m_maximumDistanceSquared(btScalar(BT_LARGE_FLOAT))
+			: m_maximumDistanceSquared(btScalar(BT_LARGE_FLOAT))
 		{
 		}
 
 		btTransform m_transformA;
 		btTransform m_transformB;
-		btScalar	m_maximumDistanceSquared;
+		btScalar m_maximumDistanceSquared;
 	};
 
-	virtual ~btDiscreteCollisionDetectorInterface() {};
+	virtual ~btDiscreteCollisionDetectorInterface(){};
 
 	//
 	// give either closest points (distance > 0) or penetration (distance)
 	// the normal always points from B towards A
 	//
-	virtual void	getClosestPoints(const ClosestPointInput& input,Result& output,class btIDebugDraw* debugDraw,bool swapResults=false) = 0;
-
+	virtual void getClosestPoints(const ClosestPointInput& input, Result& output, class btIDebugDraw* debugDraw, bool swapResults = false) = 0;
 };
 
 struct btStorageResult : public btDiscreteCollisionDetectorInterface::Result
 {
-		btVector3	m_normalOnSurfaceB;
-		btVector3	m_closestPointInB;
-		btScalar	m_distance; //negative means penetration !
+	btVector3 m_normalOnSurfaceB;
+	btVector3 m_closestPointInB;
+	btScalar m_distance;  //negative means penetration !
 
-    protected:
-		btStorageResult() : m_distance(btScalar(BT_LARGE_FLOAT))
-		{
-		}
-		
-	public:
-		virtual ~btStorageResult() {};
+protected:
+	btStorageResult() : m_distance(btScalar(BT_LARGE_FLOAT))
+	{
+	}
 
-		virtual void addContactPoint(const btVector3& normalOnBInWorld,const btVector3& pointInWorld,btScalar depth)
+public:
+	virtual ~btStorageResult(){};
+
+	virtual void addContactPoint(const btVector3& normalOnBInWorld, const btVector3& pointInWorld, btScalar depth)
+	{
+		if (depth < m_distance)
 		{
-			if (depth < m_distance)
-			{
-				m_normalOnSurfaceB = normalOnBInWorld;
-				m_closestPointInB = pointInWorld;
-				m_distance = depth;
-			}
+			m_normalOnSurfaceB = normalOnBInWorld;
+			m_closestPointInB = pointInWorld;
+			m_distance = depth;
 		}
+	}
 };
 
-#endif //BT_DISCRETE_COLLISION_DETECTOR1_INTERFACE_H
-
+#endif  //BT_DISCRETE_COLLISION_DETECTOR1_INTERFACE_H
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkCollisionDescription.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkCollisionDescription.h
index 0b49b0ecc65..c9fd84bebfa 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkCollisionDescription.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkCollisionDescription.h
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef GJK_COLLISION_DESCRIPTION_H
 #define GJK_COLLISION_DESCRIPTION_H
 
@@ -21,21 +20,20 @@ subject to the following restrictions:
 
 struct btGjkCollisionDescription
 {
-    btVector3	m_firstDir;
-    int			m_maxGjkIterations;
-    btScalar	m_maximumDistanceSquared;
-    btScalar	m_gjkRelError2;
-    btGjkCollisionDescription()
-    :m_firstDir(0,1,0),
-    m_maxGjkIterations(1000),
-    m_maximumDistanceSquared(1e30f),
-    m_gjkRelError2(1.0e-6)
-    {
-    }
-    virtual ~btGjkCollisionDescription()
-    {
-    }
+	btVector3 m_firstDir;
+	int m_maxGjkIterations;
+	btScalar m_maximumDistanceSquared;
+	btScalar m_gjkRelError2;
+	btGjkCollisionDescription()
+		: m_firstDir(0, 1, 0),
+		  m_maxGjkIterations(1000),
+		  m_maximumDistanceSquared(1e30f),
+		  m_gjkRelError2(1.0e-6)
+	{
+	}
+	virtual ~btGjkCollisionDescription()
+	{
+	}
 };
 
-#endif //GJK_COLLISION_DESCRIPTION_H
-
+#endif  //GJK_COLLISION_DESCRIPTION_H
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkConvexCast.cpp b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkConvexCast.cpp
index bef697a0a11..9d61e75dac8 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkConvexCast.cpp
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkConvexCast.cpp
@@ -13,8 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #include "btGjkConvexCast.h"
 #include "BulletCollision/CollisionShapes/btSphereShape.h"
 #include "btGjkPairDetector.h"
@@ -27,41 +25,39 @@ subject to the following restrictions:
 #define MAX_ITERATIONS 32
 #endif
 
-btGjkConvexCast::btGjkConvexCast(const btConvexShape* convexA,const btConvexShape* convexB,btSimplexSolverInterface* simplexSolver)
-:m_simplexSolver(simplexSolver),
-m_convexA(convexA),
-m_convexB(convexB)
+btGjkConvexCast::btGjkConvexCast(const btConvexShape* convexA, const btConvexShape* convexB, btSimplexSolverInterface* simplexSolver)
+	: m_simplexSolver(simplexSolver),
+	  m_convexA(convexA),
+	  m_convexB(convexB)
 {
 }
 
-bool	btGjkConvexCast::calcTimeOfImpact(
-					const btTransform& fromA,
-					const btTransform& toA,
-					const btTransform& fromB,
-					const btTransform& toB,
-					CastResult& result)
+bool btGjkConvexCast::calcTimeOfImpact(
+	const btTransform& fromA,
+	const btTransform& toA,
+	const btTransform& fromB,
+	const btTransform& toB,
+	CastResult& result)
 {
-
-
 	m_simplexSolver->reset();
 
 	/// compute linear velocity for this interval, to interpolate
 	//assume no rotation/angular velocity, assert here?
-	btVector3 linVelA,linVelB;
-	linVelA = toA.getOrigin()-fromA.getOrigin();
-	linVelB = toB.getOrigin()-fromB.getOrigin();
+	btVector3 linVelA, linVelB;
+	linVelA = toA.getOrigin() - fromA.getOrigin();
+	linVelB = toB.getOrigin() - fromB.getOrigin();
 
 	btScalar radius = btScalar(0.001);
 	btScalar lambda = btScalar(0.);
-	btVector3 v(1,0,0);
+	btVector3 v(1, 0, 0);
 
 	int maxIter = MAX_ITERATIONS;
 
 	btVector3 n;
-	n.setValue(btScalar(0.),btScalar(0.),btScalar(0.));
+	n.setValue(btScalar(0.), btScalar(0.), btScalar(0.));
 	bool hasResult = false;
 	btVector3 c;
-	btVector3 r = (linVelA-linVelB);
+	btVector3 r = (linVelA - linVelB);
 
 	btScalar lastLambda = lambda;
 	//btScalar epsilon = btScalar(0.001);
@@ -69,17 +65,14 @@ bool	btGjkConvexCast::calcTimeOfImpact(
 	int numIter = 0;
 	//first solution, using GJK
 
-
 	btTransform identityTrans;
 	identityTrans.setIdentity();
 
+	//	result.drawCoordSystem(sphereTr);
 
-//	result.drawCoordSystem(sphereTr);
+	btPointCollector pointCollector;
 
-	btPointCollector	pointCollector;
-
-		
-	btGjkPairDetector gjk(m_convexA,m_convexB,m_simplexSolver,0);//m_penetrationDepthSolver);		
+	btGjkPairDetector gjk(m_convexA, m_convexB, m_simplexSolver, 0);  //m_penetrationDepthSolver);
 	btGjkPairDetector::ClosestPointInput input;
 
 	//we don't use margins during CCD
@@ -87,7 +80,7 @@ bool	btGjkConvexCast::calcTimeOfImpact(
 
 	input.m_transformA = fromA;
 	input.m_transformB = fromB;
-	gjk.getClosestPoints(input,pointCollector,0);
+	gjk.getClosestPoints(input, pointCollector, 0);
 
 	hasResult = pointCollector.m_hasResult;
 	c = pointCollector.m_pointInWorld;
@@ -98,20 +91,18 @@ bool	btGjkConvexCast::calcTimeOfImpact(
 		dist = pointCollector.m_distance;
 		n = pointCollector.m_normalOnBInWorld;
 
-	
-
 		//not close enough
 		while (dist > radius)
 		{
 			numIter++;
 			if (numIter > maxIter)
 			{
-				return false; //todo: report a failure
+				return false;  //todo: report a failure
 			}
 			btScalar dLambda = btScalar(0.);
 
 			btScalar projectedLinearVelocity = r.dot(n);
-			
+
 			dLambda = dist / (projectedLinearVelocity);
 
 			lambda = lambda - dLambda;
@@ -132,35 +123,35 @@ bool	btGjkConvexCast::calcTimeOfImpact(
 			lastLambda = lambda;
 
 			//interpolate to next lambda
-			result.DebugDraw( lambda );
-			input.m_transformA.getOrigin().setInterpolate3(fromA.getOrigin(),toA.getOrigin(),lambda);
-			input.m_transformB.getOrigin().setInterpolate3(fromB.getOrigin(),toB.getOrigin(),lambda);
-			
-			gjk.getClosestPoints(input,pointCollector,0);
+			result.DebugDraw(lambda);
+			input.m_transformA.getOrigin().setInterpolate3(fromA.getOrigin(), toA.getOrigin(), lambda);
+			input.m_transformB.getOrigin().setInterpolate3(fromB.getOrigin(), toB.getOrigin(), lambda);
+
+			gjk.getClosestPoints(input, pointCollector, 0);
 			if (pointCollector.m_hasResult)
 			{
 				if (pointCollector.m_distance < btScalar(0.))
 				{
 					result.m_fraction = lastLambda;
 					n = pointCollector.m_normalOnBInWorld;
-					result.m_normal=n;
+					result.m_normal = n;
 					result.m_hitPoint = pointCollector.m_pointInWorld;
 					return true;
 				}
-				c = pointCollector.m_pointInWorld;		
+				c = pointCollector.m_pointInWorld;
 				n = pointCollector.m_normalOnBInWorld;
 				dist = pointCollector.m_distance;
-			} else
+			}
+			else
 			{
 				//??
 				return false;
 			}
-
 		}
 
 		//is n normalized?
 		//don't report time of impact for motion away from the contact normal (or causes minor penetration)
-		if (n.dot(r)>=-result.m_allowedPenetration)
+		if (n.dot(r) >= -result.m_allowedPenetration)
 			return false;
 
 		result.m_fraction = lambda;
@@ -170,7 +161,4 @@ bool	btGjkConvexCast::calcTimeOfImpact(
 	}
 
 	return false;
-
-
 }
-
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkConvexCast.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkConvexCast.h
index 6a42ee63b03..ef5979173e9 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkConvexCast.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkConvexCast.h
@@ -13,8 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #ifndef BT_GJK_CONVEX_CAST_H
 #define BT_GJK_CONVEX_CAST_H
 
@@ -29,22 +27,20 @@ class btMinkowskiSumShape;
 ///GjkConvexCast performs a raycast on a convex object using support mapping.
 class btGjkConvexCast : public btConvexCast
 {
-	btSimplexSolverInterface*	m_simplexSolver;
-	const btConvexShape*	m_convexA;
-	const btConvexShape*	m_convexB;
+	btSimplexSolverInterface* m_simplexSolver;
+	const btConvexShape* m_convexA;
+	const btConvexShape* m_convexB;
 
 public:
-
-	btGjkConvexCast(const btConvexShape*	convexA,const btConvexShape* convexB,btSimplexSolverInterface* simplexSolver);
+	btGjkConvexCast(const btConvexShape* convexA, const btConvexShape* convexB, btSimplexSolverInterface* simplexSolver);
 
 	/// cast a convex against another convex object
-	virtual bool	calcTimeOfImpact(
-					const btTransform& fromA,
-					const btTransform& toA,
-					const btTransform& fromB,
-					const btTransform& toB,
-					CastResult& result);
-
+	virtual bool calcTimeOfImpact(
+		const btTransform& fromA,
+		const btTransform& toA,
+		const btTransform& fromB,
+		const btTransform& toB,
+		CastResult& result);
 };
 
-#endif //BT_GJK_CONVEX_CAST_H
+#endif  //BT_GJK_CONVEX_CAST_H
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkEpa2.cpp b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkEpa2.cpp
index eefb974bbd2..45d18171357 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkEpa2.cpp
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkEpa2.cpp
@@ -26,1010 +26,1065 @@ GJK-EPA collision solver by Nathanael Presson, 2008
 #include "BulletCollision/CollisionShapes/btSphereShape.h"
 #include "btGjkEpa2.h"
 
-#if defined(DEBUG) || defined (_DEBUG)
-#include <stdio.h> //for debug printf
+#if defined(DEBUG) || defined(_DEBUG)
+#include <stdio.h>  //for debug printf
 #ifdef __SPU__
 #include <spu_printf.h>
 #define printf spu_printf
-#endif //__SPU__
+#endif  //__SPU__
 #endif
 
 namespace gjkepa2_impl
 {
+// Config
 
-	// Config
-
-	/* GJK	*/ 
-#define GJK_MAX_ITERATIONS	128
+/* GJK	*/
+#define GJK_MAX_ITERATIONS 128
 
 #ifdef BT_USE_DOUBLE_PRECISION
-	#define GJK_ACCURACY		((btScalar)1e-12)
-	#define GJK_MIN_DISTANCE	((btScalar)1e-12)
-	#define GJK_DUPLICATED_EPS	((btScalar)1e-12)
+#define GJK_ACCURACY ((btScalar)1e-12)
+#define GJK_MIN_DISTANCE ((btScalar)1e-12)
+#define GJK_DUPLICATED_EPS ((btScalar)1e-12)
 #else
-	#define GJK_ACCURACY		((btScalar)0.0001)
-	#define GJK_MIN_DISTANCE	((btScalar)0.0001)
-	#define GJK_DUPLICATED_EPS	((btScalar)0.0001)
-#endif //BT_USE_DOUBLE_PRECISION
+#define GJK_ACCURACY ((btScalar)0.0001)
+#define GJK_MIN_DISTANCE ((btScalar)0.0001)
+#define GJK_DUPLICATED_EPS ((btScalar)0.0001)
+#endif  //BT_USE_DOUBLE_PRECISION
 
+#define GJK_SIMPLEX2_EPS ((btScalar)0.0)
+#define GJK_SIMPLEX3_EPS ((btScalar)0.0)
+#define GJK_SIMPLEX4_EPS ((btScalar)0.0)
 
-#define GJK_SIMPLEX2_EPS	((btScalar)0.0)
-#define GJK_SIMPLEX3_EPS	((btScalar)0.0)
-#define GJK_SIMPLEX4_EPS	((btScalar)0.0)
-
-	/* EPA	*/ 
-#define EPA_MAX_VERTICES	128
-#define EPA_MAX_ITERATIONS	255
+/* EPA	*/
+#define EPA_MAX_VERTICES 128
+#define EPA_MAX_ITERATIONS 255
 
 #ifdef BT_USE_DOUBLE_PRECISION
-	#define EPA_ACCURACY		((btScalar)1e-12)
-	#define EPA_PLANE_EPS		((btScalar)1e-14)
-	#define EPA_INSIDE_EPS		((btScalar)1e-9)
+#define EPA_ACCURACY ((btScalar)1e-12)
+#define EPA_PLANE_EPS ((btScalar)1e-14)
+#define EPA_INSIDE_EPS ((btScalar)1e-9)
 #else
-	#define EPA_ACCURACY		((btScalar)0.0001)
-	#define EPA_PLANE_EPS		((btScalar)0.00001)
-	#define EPA_INSIDE_EPS		((btScalar)0.01)
+#define EPA_ACCURACY ((btScalar)0.0001)
+#define EPA_PLANE_EPS ((btScalar)0.00001)
+#define EPA_INSIDE_EPS ((btScalar)0.01)
 #endif
 
-#define EPA_FALLBACK            (10*EPA_ACCURACY)
-#define EPA_MAX_FACES           (EPA_MAX_VERTICES*2)
+#define EPA_FALLBACK (10 * EPA_ACCURACY)
+#define EPA_MAX_FACES (EPA_MAX_VERTICES * 2)
 
+// Shorthands
+typedef unsigned int U;
+typedef unsigned char U1;
 
-	// Shorthands
-	typedef unsigned int	U;
-	typedef unsigned char	U1;
+// MinkowskiDiff
+struct MinkowskiDiff
+{
+	const btConvexShape* m_shapes[2];
+	btMatrix3x3 m_toshape1;
+	btTransform m_toshape0;
+#ifdef __SPU__
+	bool m_enableMargin;
+#else
+	btVector3 (btConvexShape::*Ls)(const btVector3&) const;
+#endif  //__SPU__
 
-	// MinkowskiDiff
-	struct	MinkowskiDiff
+	MinkowskiDiff()
 	{
-		const btConvexShape*	m_shapes[2];
-		btMatrix3x3				m_toshape1;
-		btTransform				m_toshape0;
+	}
 #ifdef __SPU__
-		bool					m_enableMargin;
+	void EnableMargin(bool enable)
+	{
+		m_enableMargin = enable;
+	}
+	inline btVector3 Support0(const btVector3& d) const
+	{
+		if (m_enableMargin)
+		{
+			return m_shapes[0]->localGetSupportVertexNonVirtual(d);
+		}
+		else
+		{
+			return m_shapes[0]->localGetSupportVertexWithoutMarginNonVirtual(d);
+		}
+	}
+	inline btVector3 Support1(const btVector3& d) const
+	{
+		if (m_enableMargin)
+		{
+			return m_toshape0 * (m_shapes[1]->localGetSupportVertexNonVirtual(m_toshape1 * d));
+		}
+		else
+		{
+			return m_toshape0 * (m_shapes[1]->localGetSupportVertexWithoutMarginNonVirtual(m_toshape1 * d));
+		}
+	}
 #else
-		btVector3				(btConvexShape::*Ls)(const btVector3&) const;
-#endif//__SPU__
-		
+	void EnableMargin(bool enable)
+	{
+		if (enable)
+			Ls = &btConvexShape::localGetSupportVertexNonVirtual;
+		else
+			Ls = &btConvexShape::localGetSupportVertexWithoutMarginNonVirtual;
+	}
+	inline btVector3 Support0(const btVector3& d) const
+	{
+		return (((m_shapes[0])->*(Ls))(d));
+	}
+	inline btVector3 Support1(const btVector3& d) const
+	{
+		return (m_toshape0 * ((m_shapes[1])->*(Ls))(m_toshape1 * d));
+	}
+#endif  //__SPU__
 
-		MinkowskiDiff()
-		{
+	inline btVector3 Support(const btVector3& d) const
+	{
+		return (Support0(d) - Support1(-d));
+	}
+	btVector3 Support(const btVector3& d, U index) const
+	{
+		if (index)
+			return (Support1(d));
+		else
+			return (Support0(d));
+	}
+};
 
-		}
-#ifdef __SPU__
-			void					EnableMargin(bool enable)
-		{
-			m_enableMargin = enable;
-		}	
-		inline btVector3		Support0(const btVector3& d) const
-		{
-			if (m_enableMargin)
-			{
-				return m_shapes[0]->localGetSupportVertexNonVirtual(d);
-			} else
-			{
-				return m_shapes[0]->localGetSupportVertexWithoutMarginNonVirtual(d);
-			}
-		}
-		inline btVector3		Support1(const btVector3& d) const
-		{
-			if (m_enableMargin)
-			{
-				return m_toshape0*(m_shapes[1]->localGetSupportVertexNonVirtual(m_toshape1*d));
-			} else
-			{
-				return m_toshape0*(m_shapes[1]->localGetSupportVertexWithoutMarginNonVirtual(m_toshape1*d));
-			}
-		}
-#else
-		void					EnableMargin(bool enable)
-		{
-			if(enable)
-				Ls=&btConvexShape::localGetSupportVertexNonVirtual;
-			else
-				Ls=&btConvexShape::localGetSupportVertexWithoutMarginNonVirtual;
-		}	
-		inline btVector3		Support0(const btVector3& d) const
-		{
-			return(((m_shapes[0])->*(Ls))(d));
-		}
-		inline btVector3		Support1(const btVector3& d) const
-		{
-			return(m_toshape0*((m_shapes[1])->*(Ls))(m_toshape1*d));
-		}
-#endif //__SPU__
+typedef MinkowskiDiff tShape;
 
-		inline btVector3		Support(const btVector3& d) const
-		{
-			return(Support0(d)-Support1(-d));
-		}
-		btVector3				Support(const btVector3& d,U index) const
-		{
-			if(index)
-				return(Support1(d));
-			else
-				return(Support0(d));
-		}
+// GJK
+struct GJK
+{
+	/* Types		*/
+	struct sSV
+	{
+		btVector3 d, w;
 	};
-
-	typedef	MinkowskiDiff	tShape;
-
-
-	// GJK
-	struct	GJK
+	struct sSimplex
 	{
-		/* Types		*/ 
-		struct	sSV
+		sSV* c[4];
+		btScalar p[4];
+		U rank;
+	};
+	struct eStatus
+	{
+		enum _
 		{
-			btVector3	d,w;
-		};
-		struct	sSimplex
-		{
-			sSV*		c[4];
-			btScalar	p[4];
-			U			rank;
-		};
-		struct	eStatus	{ enum _ {
 			Valid,
 			Inside,
-			Failed		};};
-			/* Fields		*/ 
-			tShape			m_shape;
-			btVector3		m_ray;
-			btScalar		m_distance;
-			sSimplex		m_simplices[2];
-			sSV				m_store[4];
-			sSV*			m_free[4];
-			U				m_nfree;
-			U				m_current;
-			sSimplex*		m_simplex;
-			eStatus::_		m_status;
-			/* Methods		*/ 
-			GJK()
-			{
-				Initialize();
-			}
-			void				Initialize()
-			{
-				m_ray		=	btVector3(0,0,0);
-				m_nfree		=	0;
-				m_status	=	eStatus::Failed;
-				m_current	=	0;
-				m_distance	=	0;
-			}
-			eStatus::_			Evaluate(const tShape& shapearg,const btVector3& guess)
-			{
-				U			iterations=0;
-				btScalar	sqdist=0;
-				btScalar	alpha=0;
-				btVector3	lastw[4];
-				U			clastw=0;
-				/* Initialize solver		*/ 
-				m_free[0]			=	&m_store[0];
-				m_free[1]			=	&m_store[1];
-				m_free[2]			=	&m_store[2];
-				m_free[3]			=	&m_store[3];
-				m_nfree				=	4;
-				m_current			=	0;
-				m_status			=	eStatus::Valid;
-				m_shape				=	shapearg;
-				m_distance			=	0;
-				/* Initialize simplex		*/ 
-				m_simplices[0].rank	=	0;
-				m_ray				=	guess;
-				const btScalar	sqrl=	m_ray.length2();
-				appendvertice(m_simplices[0],sqrl>0?-m_ray:btVector3(1,0,0));
-				m_simplices[0].p[0]	=	1;
-				m_ray				=	m_simplices[0].c[0]->w;	
-				sqdist				=	sqrl;
-				lastw[0]			=
-					lastw[1]			=
-					lastw[2]			=
-					lastw[3]			=	m_ray;
-				/* Loop						*/ 
-				do	{
-					const U		next=1-m_current;
-					sSimplex&	cs=m_simplices[m_current];
-					sSimplex&	ns=m_simplices[next];
-					/* Check zero							*/ 
-					const btScalar	rl=m_ray.length();
-					if(rl<GJK_MIN_DISTANCE)
-					{/* Touching or inside				*/ 
-						m_status=eStatus::Inside;
-						break;
-					}
-					/* Append new vertice in -'v' direction	*/ 
-					appendvertice(cs,-m_ray);
-					const btVector3&	w=cs.c[cs.rank-1]->w;
-					bool				found=false;
-					for(U i=0;i<4;++i)
-					{
-						if((w-lastw[i]).length2()<GJK_DUPLICATED_EPS)
-						{ found=true;break; }
-					}
-					if(found)
-					{/* Return old simplex				*/ 
-						removevertice(m_simplices[m_current]);
-						break;
-					}
-					else
-					{/* Update lastw					*/ 
-						lastw[clastw=(clastw+1)&3]=w;
-					}
-					/* Check for termination				*/ 
-					const btScalar	omega=btDot(m_ray,w)/rl;
-					alpha=btMax(omega,alpha);
-					if(((rl-alpha)-(GJK_ACCURACY*rl))<=0)
-					{/* Return old simplex				*/ 
-						removevertice(m_simplices[m_current]);
-						break;
-					}		
-					/* Reduce simplex						*/ 
-					btScalar	weights[4];
-					U			mask=0;
-					switch(cs.rank)
-					{
-					case	2:	sqdist=projectorigin(	cs.c[0]->w,
-									cs.c[1]->w,
-									weights,mask);break;
-					case	3:	sqdist=projectorigin(	cs.c[0]->w,
-									cs.c[1]->w,
-									cs.c[2]->w,
-									weights,mask);break;
-					case	4:	sqdist=projectorigin(	cs.c[0]->w,
-									cs.c[1]->w,
-									cs.c[2]->w,
-									cs.c[3]->w,
-									weights,mask);break;
-					}
-					if(sqdist>=0)
-					{/* Valid	*/ 
-						ns.rank		=	0;
-						m_ray		=	btVector3(0,0,0);
-						m_current	=	next;
-						for(U i=0,ni=cs.rank;i<ni;++i)
-						{
-							if(mask&(1<<i))
-							{
-								ns.c[ns.rank]		=	cs.c[i];
-								ns.p[ns.rank++]		=	weights[i];
-								m_ray				+=	cs.c[i]->w*weights[i];
-							}
-							else
-							{
-								m_free[m_nfree++]	=	cs.c[i];
-							}
-						}
-						if(mask==15) m_status=eStatus::Inside;
-					}
-					else
-					{/* Return old simplex				*/ 
-						removevertice(m_simplices[m_current]);
-						break;
-					}
-					m_status=((++iterations)<GJK_MAX_ITERATIONS)?m_status:eStatus::Failed;
-				} while(m_status==eStatus::Valid);
-				m_simplex=&m_simplices[m_current];
-				switch(m_status)
-				{
-				case	eStatus::Valid:		m_distance=m_ray.length();break;
-				case	eStatus::Inside:	m_distance=0;break;
-				default:
-					{
-					}
-				}	
-				return(m_status);
-			}
-			bool					EncloseOrigin()
-			{
-				switch(m_simplex->rank)
-				{
-				case	1:
-					{
-						for(U i=0;i<3;++i)
-						{
-							btVector3		axis=btVector3(0,0,0);
-							axis[i]=1;
-							appendvertice(*m_simplex, axis);
-							if(EncloseOrigin())	return(true);
-							removevertice(*m_simplex);
-							appendvertice(*m_simplex,-axis);
-							if(EncloseOrigin())	return(true);
-							removevertice(*m_simplex);
-						}
-					}
-					break;
-				case	2:
-					{
-						const btVector3	d=m_simplex->c[1]->w-m_simplex->c[0]->w;
-						for(U i=0;i<3;++i)
-						{
-							btVector3		axis=btVector3(0,0,0);
-							axis[i]=1;
-							const btVector3	p=btCross(d,axis);
-							if(p.length2()>0)
-							{
-								appendvertice(*m_simplex, p);
-								if(EncloseOrigin())	return(true);
-								removevertice(*m_simplex);
-								appendvertice(*m_simplex,-p);
-								if(EncloseOrigin())	return(true);
-								removevertice(*m_simplex);
-							}
-						}
-					}
-					break;
-				case	3:
-					{
-						const btVector3	n=btCross(m_simplex->c[1]->w-m_simplex->c[0]->w,
-							m_simplex->c[2]->w-m_simplex->c[0]->w);
-						if(n.length2()>0)
-						{
-							appendvertice(*m_simplex,n);
-							if(EncloseOrigin())	return(true);
-							removevertice(*m_simplex);
-							appendvertice(*m_simplex,-n);
-							if(EncloseOrigin())	return(true);
-							removevertice(*m_simplex);
-						}
-					}
-					break;
-				case	4:
-					{
-						if(btFabs(det(	m_simplex->c[0]->w-m_simplex->c[3]->w,
-							m_simplex->c[1]->w-m_simplex->c[3]->w,
-							m_simplex->c[2]->w-m_simplex->c[3]->w))>0)
-							return(true);
-					}
-					break;
-				}
-				return(false);
-			}
-			/* Internals	*/ 
-			void				getsupport(const btVector3& d,sSV& sv) const
-			{
-				sv.d	=	d/d.length();
-				sv.w	=	m_shape.Support(sv.d);
-			}
-			void				removevertice(sSimplex& simplex)
-			{
-				m_free[m_nfree++]=simplex.c[--simplex.rank];
-			}
-			void				appendvertice(sSimplex& simplex,const btVector3& v)
-			{
-				simplex.p[simplex.rank]=0;
-				simplex.c[simplex.rank]=m_free[--m_nfree];
-				getsupport(v,*simplex.c[simplex.rank++]);
-			}
-			static btScalar		det(const btVector3& a,const btVector3& b,const btVector3& c)
-			{
-				return(	a.y()*b.z()*c.x()+a.z()*b.x()*c.y()-
-					a.x()*b.z()*c.y()-a.y()*b.x()*c.z()+
-					a.x()*b.y()*c.z()-a.z()*b.y()*c.x());
-			}
-			static btScalar		projectorigin(	const btVector3& a,
-				const btVector3& b,
-				btScalar* w,U& m)
-			{
-				const btVector3	d=b-a;
-				const btScalar	l=d.length2();
-				if(l>GJK_SIMPLEX2_EPS)
-				{
-					const btScalar	t(l>0?-btDot(a,d)/l:0);
-					if(t>=1)		{ w[0]=0;w[1]=1;m=2;return(b.length2()); }
-					else if(t<=0)	{ w[0]=1;w[1]=0;m=1;return(a.length2()); }
-					else			{ w[0]=1-(w[1]=t);m=3;return((a+d*t).length2()); }
-				}
-				return(-1);
-			}
-			static btScalar		projectorigin(	const btVector3& a,
-				const btVector3& b,
-				const btVector3& c,
-				btScalar* w,U& m)
-			{
-				static const U		imd3[]={1,2,0};
-				const btVector3*	vt[]={&a,&b,&c};
-				const btVector3		dl[]={a-b,b-c,c-a};
-				const btVector3		n=btCross(dl[0],dl[1]);
-				const btScalar		l=n.length2();
-				if(l>GJK_SIMPLEX3_EPS)
-				{
-					btScalar	mindist=-1;
-					btScalar	subw[2]={0.f,0.f};
-					U			subm(0);
-					for(U i=0;i<3;++i)
-					{
-						if(btDot(*vt[i],btCross(dl[i],n))>0)
-						{
-							const U			j=imd3[i];
-							const btScalar	subd(projectorigin(*vt[i],*vt[j],subw,subm));
-							if((mindist<0)||(subd<mindist))
-							{
-								mindist		=	subd;
-								m			=	static_cast<U>(((subm&1)?1<<i:0)+((subm&2)?1<<j:0));
-								w[i]		=	subw[0];
-								w[j]		=	subw[1];
-								w[imd3[j]]	=	0;				
-							}
-						}
-					}
-					if(mindist<0)
-					{
-						const btScalar	d=btDot(a,n);	
-						const btScalar	s=btSqrt(l);
-						const btVector3	p=n*(d/l);
-						mindist	=	p.length2();
-						m		=	7;
-						w[0]	=	(btCross(dl[1],b-p)).length()/s;
-						w[1]	=	(btCross(dl[2],c-p)).length()/s;
-						w[2]	=	1-(w[0]+w[1]);
-					}
-					return(mindist);
-				}
-				return(-1);
-			}
-			static btScalar		projectorigin(	const btVector3& a,
-				const btVector3& b,
-				const btVector3& c,
-				const btVector3& d,
-				btScalar* w,U& m)
-			{
-				static const U		imd3[]={1,2,0};
-				const btVector3*	vt[]={&a,&b,&c,&d};
-				const btVector3		dl[]={a-d,b-d,c-d};
-				const btScalar		vl=det(dl[0],dl[1],dl[2]);
-				const bool			ng=(vl*btDot(a,btCross(b-c,a-b)))<=0;
-				if(ng&&(btFabs(vl)>GJK_SIMPLEX4_EPS))
-				{
-					btScalar	mindist=-1;
-					btScalar	subw[3]={0.f,0.f,0.f};
-					U			subm(0);
-					for(U i=0;i<3;++i)
-					{
-						const U			j=imd3[i];
-						const btScalar	s=vl*btDot(d,btCross(dl[i],dl[j]));
-						if(s>0)
-						{
-							const btScalar	subd=projectorigin(*vt[i],*vt[j],d,subw,subm);
-							if((mindist<0)||(subd<mindist))
-							{
-								mindist		=	subd;
-								m			=	static_cast<U>((subm&1?1<<i:0)+
-									(subm&2?1<<j:0)+
-									(subm&4?8:0));
-								w[i]		=	subw[0];
-								w[j]		=	subw[1];
-								w[imd3[j]]	=	0;
-								w[3]		=	subw[2];
-							}
-						}
-					}
-					if(mindist<0)
-					{
-						mindist	=	0;
-						m		=	15;
-						w[0]	=	det(c,b,d)/vl;
-						w[1]	=	det(a,c,d)/vl;
-						w[2]	=	det(b,a,d)/vl;
-						w[3]	=	1-(w[0]+w[1]+w[2]);
-					}
-					return(mindist);
-				}
-				return(-1);
-			}
+			Failed
+		};
 	};
-
-	// EPA
-	struct	EPA
+	/* Fields		*/
+	tShape m_shape;
+	btVector3 m_ray;
+	btScalar m_distance;
+	sSimplex m_simplices[2];
+	sSV m_store[4];
+	sSV* m_free[4];
+	U m_nfree;
+	U m_current;
+	sSimplex* m_simplex;
+	eStatus::_ m_status;
+	/* Methods		*/
+	GJK()
 	{
-		/* Types		*/ 
-		typedef	GJK::sSV	sSV;
-		struct	sFace
+		Initialize();
+	}
+	void Initialize()
+	{
+		m_ray = btVector3(0, 0, 0);
+		m_nfree = 0;
+		m_status = eStatus::Failed;
+		m_current = 0;
+		m_distance = 0;
+	}
+	eStatus::_ Evaluate(const tShape& shapearg, const btVector3& guess)
+	{
+		U iterations = 0;
+		btScalar sqdist = 0;
+		btScalar alpha = 0;
+		btVector3 lastw[4];
+		U clastw = 0;
+		/* Initialize solver		*/
+		m_free[0] = &m_store[0];
+		m_free[1] = &m_store[1];
+		m_free[2] = &m_store[2];
+		m_free[3] = &m_store[3];
+		m_nfree = 4;
+		m_current = 0;
+		m_status = eStatus::Valid;
+		m_shape = shapearg;
+		m_distance = 0;
+		/* Initialize simplex		*/
+		m_simplices[0].rank = 0;
+		m_ray = guess;
+		const btScalar sqrl = m_ray.length2();
+		appendvertice(m_simplices[0], sqrl > 0 ? -m_ray : btVector3(1, 0, 0));
+		m_simplices[0].p[0] = 1;
+		m_ray = m_simplices[0].c[0]->w;
+		sqdist = sqrl;
+		lastw[0] =
+			lastw[1] =
+				lastw[2] =
+					lastw[3] = m_ray;
+		/* Loop						*/
+		do
 		{
-			btVector3	n;
-			btScalar	d;
-			sSV*		c[3];
-			sFace*		f[3];
-			sFace*		l[2];
-			U1			e[3];
-			U1			pass;
-		};
-		struct	sList
+			const U next = 1 - m_current;
+			sSimplex& cs = m_simplices[m_current];
+			sSimplex& ns = m_simplices[next];
+			/* Check zero							*/
+			const btScalar rl = m_ray.length();
+			if (rl < GJK_MIN_DISTANCE)
+			{ /* Touching or inside				*/
+				m_status = eStatus::Inside;
+				break;
+			}
+			/* Append new vertice in -'v' direction	*/
+			appendvertice(cs, -m_ray);
+			const btVector3& w = cs.c[cs.rank - 1]->w;
+			bool found = false;
+			for (U i = 0; i < 4; ++i)
+			{
+				if ((w - lastw[i]).length2() < GJK_DUPLICATED_EPS)
+				{
+					found = true;
+					break;
+				}
+			}
+			if (found)
+			{ /* Return old simplex				*/
+				removevertice(m_simplices[m_current]);
+				break;
+			}
+			else
+			{ /* Update lastw					*/
+				lastw[clastw = (clastw + 1) & 3] = w;
+			}
+			/* Check for termination				*/
+			const btScalar omega = btDot(m_ray, w) / rl;
+			alpha = btMax(omega, alpha);
+			if (((rl - alpha) - (GJK_ACCURACY * rl)) <= 0)
+			{ /* Return old simplex				*/
+				removevertice(m_simplices[m_current]);
+				break;
+			}
+			/* Reduce simplex						*/
+			btScalar weights[4];
+			U mask = 0;
+			switch (cs.rank)
+			{
+				case 2:
+					sqdist = projectorigin(cs.c[0]->w,
+										   cs.c[1]->w,
+										   weights, mask);
+					break;
+				case 3:
+					sqdist = projectorigin(cs.c[0]->w,
+										   cs.c[1]->w,
+										   cs.c[2]->w,
+										   weights, mask);
+					break;
+				case 4:
+					sqdist = projectorigin(cs.c[0]->w,
+										   cs.c[1]->w,
+										   cs.c[2]->w,
+										   cs.c[3]->w,
+										   weights, mask);
+					break;
+			}
+			if (sqdist >= 0)
+			{ /* Valid	*/
+				ns.rank = 0;
+				m_ray = btVector3(0, 0, 0);
+				m_current = next;
+				for (U i = 0, ni = cs.rank; i < ni; ++i)
+				{
+					if (mask & (1 << i))
+					{
+						ns.c[ns.rank] = cs.c[i];
+						ns.p[ns.rank++] = weights[i];
+						m_ray += cs.c[i]->w * weights[i];
+					}
+					else
+					{
+						m_free[m_nfree++] = cs.c[i];
+					}
+				}
+				if (mask == 15) m_status = eStatus::Inside;
+			}
+			else
+			{ /* Return old simplex				*/
+				removevertice(m_simplices[m_current]);
+				break;
+			}
+			m_status = ((++iterations) < GJK_MAX_ITERATIONS) ? m_status : eStatus::Failed;
+		} while (m_status == eStatus::Valid);
+		m_simplex = &m_simplices[m_current];
+		switch (m_status)
 		{
-			sFace*		root;
-			U			count;
-			sList() : root(0),count(0)	{}
-		};
-		struct	sHorizon
+			case eStatus::Valid:
+				m_distance = m_ray.length();
+				break;
+			case eStatus::Inside:
+				m_distance = 0;
+				break;
+			default:
+			{
+			}
+		}
+		return (m_status);
+	}
+	bool EncloseOrigin()
+	{
+		switch (m_simplex->rank)
+		{
+			case 1:
+			{
+				for (U i = 0; i < 3; ++i)
+				{
+					btVector3 axis = btVector3(0, 0, 0);
+					axis[i] = 1;
+					appendvertice(*m_simplex, axis);
+					if (EncloseOrigin()) return (true);
+					removevertice(*m_simplex);
+					appendvertice(*m_simplex, -axis);
+					if (EncloseOrigin()) return (true);
+					removevertice(*m_simplex);
+				}
+			}
+			break;
+			case 2:
+			{
+				const btVector3 d = m_simplex->c[1]->w - m_simplex->c[0]->w;
+				for (U i = 0; i < 3; ++i)
+				{
+					btVector3 axis = btVector3(0, 0, 0);
+					axis[i] = 1;
+					const btVector3 p = btCross(d, axis);
+					if (p.length2() > 0)
+					{
+						appendvertice(*m_simplex, p);
+						if (EncloseOrigin()) return (true);
+						removevertice(*m_simplex);
+						appendvertice(*m_simplex, -p);
+						if (EncloseOrigin()) return (true);
+						removevertice(*m_simplex);
+					}
+				}
+			}
+			break;
+			case 3:
+			{
+				const btVector3 n = btCross(m_simplex->c[1]->w - m_simplex->c[0]->w,
+											m_simplex->c[2]->w - m_simplex->c[0]->w);
+				if (n.length2() > 0)
+				{
+					appendvertice(*m_simplex, n);
+					if (EncloseOrigin()) return (true);
+					removevertice(*m_simplex);
+					appendvertice(*m_simplex, -n);
+					if (EncloseOrigin()) return (true);
+					removevertice(*m_simplex);
+				}
+			}
+			break;
+			case 4:
+			{
+				if (btFabs(det(m_simplex->c[0]->w - m_simplex->c[3]->w,
+							   m_simplex->c[1]->w - m_simplex->c[3]->w,
+							   m_simplex->c[2]->w - m_simplex->c[3]->w)) > 0)
+					return (true);
+			}
+			break;
+		}
+		return (false);
+	}
+	/* Internals	*/
+	void getsupport(const btVector3& d, sSV& sv) const
+	{
+		sv.d = d / d.length();
+		sv.w = m_shape.Support(sv.d);
+	}
+	void removevertice(sSimplex& simplex)
+	{
+		m_free[m_nfree++] = simplex.c[--simplex.rank];
+	}
+	void appendvertice(sSimplex& simplex, const btVector3& v)
+	{
+		simplex.p[simplex.rank] = 0;
+		simplex.c[simplex.rank] = m_free[--m_nfree];
+		getsupport(v, *simplex.c[simplex.rank++]);
+	}
+	static btScalar det(const btVector3& a, const btVector3& b, const btVector3& c)
+	{
+		return (a.y() * b.z() * c.x() + a.z() * b.x() * c.y() -
+				a.x() * b.z() * c.y() - a.y() * b.x() * c.z() +
+				a.x() * b.y() * c.z() - a.z() * b.y() * c.x());
+	}
+	static btScalar projectorigin(const btVector3& a,
+								  const btVector3& b,
+								  btScalar* w, U& m)
+	{
+		const btVector3 d = b - a;
+		const btScalar l = d.length2();
+		if (l > GJK_SIMPLEX2_EPS)
+		{
+			const btScalar t(l > 0 ? -btDot(a, d) / l : 0);
+			if (t >= 1)
+			{
+				w[0] = 0;
+				w[1] = 1;
+				m = 2;
+				return (b.length2());
+			}
+			else if (t <= 0)
+			{
+				w[0] = 1;
+				w[1] = 0;
+				m = 1;
+				return (a.length2());
+			}
+			else
+			{
+				w[0] = 1 - (w[1] = t);
+				m = 3;
+				return ((a + d * t).length2());
+			}
+		}
+		return (-1);
+	}
+	static btScalar projectorigin(const btVector3& a,
+								  const btVector3& b,
+								  const btVector3& c,
+								  btScalar* w, U& m)
+	{
+		static const U imd3[] = {1, 2, 0};
+		const btVector3* vt[] = {&a, &b, &c};
+		const btVector3 dl[] = {a - b, b - c, c - a};
+		const btVector3 n = btCross(dl[0], dl[1]);
+		const btScalar l = n.length2();
+		if (l > GJK_SIMPLEX3_EPS)
+		{
+			btScalar mindist = -1;
+			btScalar subw[2] = {0.f, 0.f};
+			U subm(0);
+			for (U i = 0; i < 3; ++i)
+			{
+				if (btDot(*vt[i], btCross(dl[i], n)) > 0)
+				{
+					const U j = imd3[i];
+					const btScalar subd(projectorigin(*vt[i], *vt[j], subw, subm));
+					if ((mindist < 0) || (subd < mindist))
+					{
+						mindist = subd;
+						m = static_cast<U>(((subm & 1) ? 1 << i : 0) + ((subm & 2) ? 1 << j : 0));
+						w[i] = subw[0];
+						w[j] = subw[1];
+						w[imd3[j]] = 0;
+					}
+				}
+			}
+			if (mindist < 0)
+			{
+				const btScalar d = btDot(a, n);
+				const btScalar s = btSqrt(l);
+				const btVector3 p = n * (d / l);
+				mindist = p.length2();
+				m = 7;
+				w[0] = (btCross(dl[1], b - p)).length() / s;
+				w[1] = (btCross(dl[2], c - p)).length() / s;
+				w[2] = 1 - (w[0] + w[1]);
+			}
+			return (mindist);
+		}
+		return (-1);
+	}
+	static btScalar projectorigin(const btVector3& a,
+								  const btVector3& b,
+								  const btVector3& c,
+								  const btVector3& d,
+								  btScalar* w, U& m)
+	{
+		static const U imd3[] = {1, 2, 0};
+		const btVector3* vt[] = {&a, &b, &c, &d};
+		const btVector3 dl[] = {a - d, b - d, c - d};
+		const btScalar vl = det(dl[0], dl[1], dl[2]);
+		const bool ng = (vl * btDot(a, btCross(b - c, a - b))) <= 0;
+		if (ng && (btFabs(vl) > GJK_SIMPLEX4_EPS))
+		{
+			btScalar mindist = -1;
+			btScalar subw[3] = {0.f, 0.f, 0.f};
+			U subm(0);
+			for (U i = 0; i < 3; ++i)
+			{
+				const U j = imd3[i];
+				const btScalar s = vl * btDot(d, btCross(dl[i], dl[j]));
+				if (s > 0)
+				{
+					const btScalar subd = projectorigin(*vt[i], *vt[j], d, subw, subm);
+					if ((mindist < 0) || (subd < mindist))
+					{
+						mindist = subd;
+						m = static_cast<U>((subm & 1 ? 1 << i : 0) +
+										   (subm & 2 ? 1 << j : 0) +
+										   (subm & 4 ? 8 : 0));
+						w[i] = subw[0];
+						w[j] = subw[1];
+						w[imd3[j]] = 0;
+						w[3] = subw[2];
+					}
+				}
+			}
+			if (mindist < 0)
+			{
+				mindist = 0;
+				m = 15;
+				w[0] = det(c, b, d) / vl;
+				w[1] = det(a, c, d) / vl;
+				w[2] = det(b, a, d) / vl;
+				w[3] = 1 - (w[0] + w[1] + w[2]);
+			}
+			return (mindist);
+		}
+		return (-1);
+	}
+};
+
+// EPA
+struct EPA
+{
+	/* Types		*/
+	typedef GJK::sSV sSV;
+	struct sFace
+	{
+		btVector3 n;
+		btScalar d;
+		sSV* c[3];
+		sFace* f[3];
+		sFace* l[2];
+		U1 e[3];
+		U1 pass;
+	};
+	struct sList
+	{
+		sFace* root;
+		U count;
+		sList() : root(0), count(0) {}
+	};
+	struct sHorizon
+	{
+		sFace* cf;
+		sFace* ff;
+		U nf;
+		sHorizon() : cf(0), ff(0), nf(0) {}
+	};
+	struct eStatus
+	{
+		enum _
 		{
-			sFace*		cf;
-			sFace*		ff;
-			U			nf;
-			sHorizon() : cf(0),ff(0),nf(0)	{}
-		};
-		struct	eStatus { enum _ {
 			Valid,
 			Touching,
 			Degenerated,
 			NonConvex,
-			InvalidHull,		
+			InvalidHull,
 			OutOfFaces,
 			OutOfVertices,
 			AccuraryReached,
 			FallBack,
-			Failed		};};
-			/* Fields		*/ 
-			eStatus::_		m_status;
-			GJK::sSimplex	m_result;
-			btVector3		m_normal;
-			btScalar		m_depth;
-			sSV				m_sv_store[EPA_MAX_VERTICES];
-			sFace			m_fc_store[EPA_MAX_FACES];
-			U				m_nextsv;
-			sList			m_hull;
-			sList			m_stock;
-			/* Methods		*/ 
-			EPA()
-			{
-				Initialize();	
-			}
-
-
-			static inline void		bind(sFace* fa,U ea,sFace* fb,U eb)
-			{
-				fa->e[ea]=(U1)eb;fa->f[ea]=fb;
-				fb->e[eb]=(U1)ea;fb->f[eb]=fa;
-			}
-			static inline void		append(sList& list,sFace* face)
-			{
-				face->l[0]	=	0;
-				face->l[1]	=	list.root;
-				if(list.root) list.root->l[0]=face;
-				list.root	=	face;
-				++list.count;
-			}
-			static inline void		remove(sList& list,sFace* face)
-			{
-				if(face->l[1]) face->l[1]->l[0]=face->l[0];
-				if(face->l[0]) face->l[0]->l[1]=face->l[1];
-				if(face==list.root) list.root=face->l[1];
-				--list.count;
-			}
-
-
-			void				Initialize()
-			{
-				m_status	=	eStatus::Failed;
-				m_normal	=	btVector3(0,0,0);
-				m_depth		=	0;
-				m_nextsv	=	0;
-				for(U i=0;i<EPA_MAX_FACES;++i)
-				{
-					append(m_stock,&m_fc_store[EPA_MAX_FACES-i-1]);
-				}
-			}
-			eStatus::_			Evaluate(GJK& gjk,const btVector3& guess)
-			{
-				GJK::sSimplex&	simplex=*gjk.m_simplex;
-				if((simplex.rank>1)&&gjk.EncloseOrigin())
-				{
-
-					/* Clean up				*/ 
-					while(m_hull.root)
-					{
-						sFace*	f = m_hull.root;
-						remove(m_hull,f);
-						append(m_stock,f);
-					}
-					m_status	=	eStatus::Valid;
-					m_nextsv	=	0;
-					/* Orient simplex		*/ 
-					if(gjk.det(	simplex.c[0]->w-simplex.c[3]->w,
-						simplex.c[1]->w-simplex.c[3]->w,
-						simplex.c[2]->w-simplex.c[3]->w)<0)
-					{
-						btSwap(simplex.c[0],simplex.c[1]);
-						btSwap(simplex.p[0],simplex.p[1]);
-					}
-					/* Build initial hull	*/ 
-					sFace*	tetra[]={newface(simplex.c[0],simplex.c[1],simplex.c[2],true),
-						newface(simplex.c[1],simplex.c[0],simplex.c[3],true),
-						newface(simplex.c[2],simplex.c[1],simplex.c[3],true),
-						newface(simplex.c[0],simplex.c[2],simplex.c[3],true)};
-					if(m_hull.count==4)
-					{
-						sFace*		best=findbest();
-						sFace		outer=*best;
-						U			pass=0;
-						U			iterations=0;
-						bind(tetra[0],0,tetra[1],0);
-						bind(tetra[0],1,tetra[2],0);
-						bind(tetra[0],2,tetra[3],0);
-						bind(tetra[1],1,tetra[3],2);
-						bind(tetra[1],2,tetra[2],1);
-						bind(tetra[2],2,tetra[3],1);
-						m_status=eStatus::Valid;
-						for(;iterations<EPA_MAX_ITERATIONS;++iterations)
-						{
-							if(m_nextsv<EPA_MAX_VERTICES)
-							{	
-								sHorizon		horizon;
-								sSV*			w=&m_sv_store[m_nextsv++];
-								bool			valid=true;					
-								best->pass	=	(U1)(++pass);
-								gjk.getsupport(best->n,*w);
-								const btScalar	wdist=btDot(best->n,w->w)-best->d;
-								if(wdist>EPA_ACCURACY)
-								{
-									for(U j=0;(j<3)&&valid;++j)
-									{
-										valid&=expand(	pass,w,
-											best->f[j],best->e[j],
-											horizon);
-									}
-									if(valid&&(horizon.nf>=3))
-									{
-										bind(horizon.cf,1,horizon.ff,2);
-										remove(m_hull,best);
-										append(m_stock,best);
-										best=findbest();
-										outer=*best;
-									} else { m_status=eStatus::InvalidHull;break; }
-								} else { m_status=eStatus::AccuraryReached;break; }
-							} else { m_status=eStatus::OutOfVertices;break; }
-						}
-						const btVector3	projection=outer.n*outer.d;
-						m_normal	=	outer.n;
-						m_depth		=	outer.d;
-						m_result.rank	=	3;
-						m_result.c[0]	=	outer.c[0];
-						m_result.c[1]	=	outer.c[1];
-						m_result.c[2]	=	outer.c[2];
-						m_result.p[0]	=	btCross(	outer.c[1]->w-projection,
-							outer.c[2]->w-projection).length();
-						m_result.p[1]	=	btCross(	outer.c[2]->w-projection,
-							outer.c[0]->w-projection).length();
-						m_result.p[2]	=	btCross(	outer.c[0]->w-projection,
-							outer.c[1]->w-projection).length();
-						const btScalar	sum=m_result.p[0]+m_result.p[1]+m_result.p[2];
-						m_result.p[0]	/=	sum;
-						m_result.p[1]	/=	sum;
-						m_result.p[2]	/=	sum;
-						return(m_status);
-					}
-				}
-				/* Fallback		*/ 
-				m_status	=	eStatus::FallBack;
-				m_normal	=	-guess;
-				const btScalar	nl=m_normal.length();
-				if(nl>0)
-					m_normal	=	m_normal/nl;
-				else
-					m_normal	=	btVector3(1,0,0);
-				m_depth	=	0;
-				m_result.rank=1;
-				m_result.c[0]=simplex.c[0];
-				m_result.p[0]=1;	
-				return(m_status);
-			}
-			bool getedgedist(sFace* face, sSV* a, sSV* b, btScalar& dist)
-			{
-				const btVector3 ba = b->w - a->w;
-				const btVector3 n_ab = btCross(ba, face->n); // Outward facing edge normal direction, on triangle plane
-				const btScalar a_dot_nab = btDot(a->w, n_ab); // Only care about the sign to determine inside/outside, so not normalization required
-
-				if(a_dot_nab < 0)
-				{
-					// Outside of edge a->b
-
-					const btScalar ba_l2 = ba.length2();
-					const btScalar a_dot_ba = btDot(a->w, ba);
-					const btScalar b_dot_ba = btDot(b->w, ba);
-
-					if(a_dot_ba > 0)
-					{
-						// Pick distance vertex a
-						dist = a->w.length();
-					}
-					else if(b_dot_ba < 0)
-					{
-						// Pick distance vertex b
-						dist = b->w.length();
-					}
-					else
-					{
-						// Pick distance to edge a->b
-						const btScalar a_dot_b = btDot(a->w, b->w);
-						dist = btSqrt(btMax((a->w.length2() * b->w.length2() - a_dot_b * a_dot_b) / ba_l2, (btScalar)0));
-					}
-
-					return true;
-				}
-
-				return false;
-			}
-			sFace*				newface(sSV* a,sSV* b,sSV* c,bool forced)
-			{
-				if(m_stock.root)
-				{
-					sFace*	face=m_stock.root;
-					remove(m_stock,face);
-					append(m_hull,face);
-					face->pass	=	0;
-					face->c[0]	=	a;
-					face->c[1]	=	b;
-					face->c[2]	=	c;
-					face->n		=	btCross(b->w-a->w,c->w-a->w);
-					const btScalar	l=face->n.length();
-					const bool		v=l>EPA_ACCURACY;
-
-					if(v)
-					{
-						if(!(getedgedist(face, a, b, face->d) ||
-							 getedgedist(face, b, c, face->d) ||
-							 getedgedist(face, c, a, face->d)))
-						{
-							// Origin projects to the interior of the triangle
-							// Use distance to triangle plane
-							face->d = btDot(a->w, face->n) / l;
-						}
-
-						face->n /= l;
-						if(forced || (face->d >= -EPA_PLANE_EPS))
-						{
-							return face;
-						}
-						else
-							m_status=eStatus::NonConvex;
-					}
-					else
-						m_status=eStatus::Degenerated;
-
-					remove(m_hull, face);
-					append(m_stock, face);
-					return 0;
-
-				}
-				m_status = m_stock.root ? eStatus::OutOfVertices : eStatus::OutOfFaces;
-				return 0;
-			}
-			sFace*				findbest()
-			{
-				sFace*		minf=m_hull.root;
-				btScalar	mind=minf->d*minf->d;
-				for(sFace* f=minf->l[1];f;f=f->l[1])
-				{
-					const btScalar	sqd=f->d*f->d;
-					if(sqd<mind)
-					{
-						minf=f;
-						mind=sqd;
-					}
-				}
-				return(minf);
-			}
-			bool				expand(U pass,sSV* w,sFace* f,U e,sHorizon& horizon)
-			{
-				static const U	i1m3[]={1,2,0};
-				static const U	i2m3[]={2,0,1};
-				if(f->pass!=pass)
-				{
-					const U	e1=i1m3[e];
-					if((btDot(f->n,w->w)-f->d)<-EPA_PLANE_EPS)
-					{
-						sFace*	nf=newface(f->c[e1],f->c[e],w,false);
-						if(nf)
-						{
-							bind(nf,0,f,e);
-							if(horizon.cf) bind(horizon.cf,1,nf,2); else horizon.ff=nf;
-							horizon.cf=nf;
-							++horizon.nf;
-							return(true);
-						}
-					}
-					else
-					{
-						const U	e2=i2m3[e];
-						f->pass		=	(U1)pass;
-						if(	expand(pass,w,f->f[e1],f->e[e1],horizon)&&
-							expand(pass,w,f->f[e2],f->e[e2],horizon))
-						{
-							remove(m_hull,f);
-							append(m_stock,f);
-							return(true);
-						}
-					}
-				}
-				return(false);
-			}
-
+			Failed
+		};
 	};
-
-	//
-	static void	Initialize(	const btConvexShape* shape0,const btTransform& wtrs0,
-		const btConvexShape* shape1,const btTransform& wtrs1,
-		btGjkEpaSolver2::sResults& results,
-		tShape& shape,
-		bool withmargins)
+	/* Fields		*/
+	eStatus::_ m_status;
+	GJK::sSimplex m_result;
+	btVector3 m_normal;
+	btScalar m_depth;
+	sSV m_sv_store[EPA_MAX_VERTICES];
+	sFace m_fc_store[EPA_MAX_FACES];
+	U m_nextsv;
+	sList m_hull;
+	sList m_stock;
+	/* Methods		*/
+	EPA()
 	{
-		/* Results		*/ 
-		results.witnesses[0]	=
-			results.witnesses[1]	=	btVector3(0,0,0);
-		results.status			=	btGjkEpaSolver2::sResults::Separated;
-		/* Shape		*/ 
-		shape.m_shapes[0]		=	shape0;
-		shape.m_shapes[1]		=	shape1;
-		shape.m_toshape1		=	wtrs1.getBasis().transposeTimes(wtrs0.getBasis());
-		shape.m_toshape0		=	wtrs0.inverseTimes(wtrs1);
-		shape.EnableMargin(withmargins);
+		Initialize();
 	}
 
+	static inline void bind(sFace* fa, U ea, sFace* fb, U eb)
+	{
+		fa->e[ea] = (U1)eb;
+		fa->f[ea] = fb;
+		fb->e[eb] = (U1)ea;
+		fb->f[eb] = fa;
+	}
+	static inline void append(sList& list, sFace* face)
+	{
+		face->l[0] = 0;
+		face->l[1] = list.root;
+		if (list.root) list.root->l[0] = face;
+		list.root = face;
+		++list.count;
+	}
+	static inline void remove(sList& list, sFace* face)
+	{
+		if (face->l[1]) face->l[1]->l[0] = face->l[0];
+		if (face->l[0]) face->l[0]->l[1] = face->l[1];
+		if (face == list.root) list.root = face->l[1];
+		--list.count;
+	}
+
+	void Initialize()
+	{
+		m_status = eStatus::Failed;
+		m_normal = btVector3(0, 0, 0);
+		m_depth = 0;
+		m_nextsv = 0;
+		for (U i = 0; i < EPA_MAX_FACES; ++i)
+		{
+			append(m_stock, &m_fc_store[EPA_MAX_FACES - i - 1]);
+		}
+	}
+	eStatus::_ Evaluate(GJK& gjk, const btVector3& guess)
+	{
+		GJK::sSimplex& simplex = *gjk.m_simplex;
+		if ((simplex.rank > 1) && gjk.EncloseOrigin())
+		{
+			/* Clean up				*/
+			while (m_hull.root)
+			{
+				sFace* f = m_hull.root;
+				remove(m_hull, f);
+				append(m_stock, f);
+			}
+			m_status = eStatus::Valid;
+			m_nextsv = 0;
+			/* Orient simplex		*/
+			if (gjk.det(simplex.c[0]->w - simplex.c[3]->w,
+						simplex.c[1]->w - simplex.c[3]->w,
+						simplex.c[2]->w - simplex.c[3]->w) < 0)
+			{
+				btSwap(simplex.c[0], simplex.c[1]);
+				btSwap(simplex.p[0], simplex.p[1]);
+			}
+			/* Build initial hull	*/
+			sFace* tetra[] = {newface(simplex.c[0], simplex.c[1], simplex.c[2], true),
+							  newface(simplex.c[1], simplex.c[0], simplex.c[3], true),
+							  newface(simplex.c[2], simplex.c[1], simplex.c[3], true),
+							  newface(simplex.c[0], simplex.c[2], simplex.c[3], true)};
+			if (m_hull.count == 4)
+			{
+				sFace* best = findbest();
+				sFace outer = *best;
+				U pass = 0;
+				U iterations = 0;
+				bind(tetra[0], 0, tetra[1], 0);
+				bind(tetra[0], 1, tetra[2], 0);
+				bind(tetra[0], 2, tetra[3], 0);
+				bind(tetra[1], 1, tetra[3], 2);
+				bind(tetra[1], 2, tetra[2], 1);
+				bind(tetra[2], 2, tetra[3], 1);
+				m_status = eStatus::Valid;
+				for (; iterations < EPA_MAX_ITERATIONS; ++iterations)
+				{
+					if (m_nextsv < EPA_MAX_VERTICES)
+					{
+						sHorizon horizon;
+						sSV* w = &m_sv_store[m_nextsv++];
+						bool valid = true;
+						best->pass = (U1)(++pass);
+						gjk.getsupport(best->n, *w);
+						const btScalar wdist = btDot(best->n, w->w) - best->d;
+						if (wdist > EPA_ACCURACY)
+						{
+							for (U j = 0; (j < 3) && valid; ++j)
+							{
+								valid &= expand(pass, w,
+												best->f[j], best->e[j],
+												horizon);
+							}
+							if (valid && (horizon.nf >= 3))
+							{
+								bind(horizon.cf, 1, horizon.ff, 2);
+								remove(m_hull, best);
+								append(m_stock, best);
+								best = findbest();
+								outer = *best;
+							}
+							else
+							{
+								m_status = eStatus::InvalidHull;
+								break;
+							}
+						}
+						else
+						{
+							m_status = eStatus::AccuraryReached;
+							break;
+						}
+					}
+					else
+					{
+						m_status = eStatus::OutOfVertices;
+						break;
+					}
+				}
+				const btVector3 projection = outer.n * outer.d;
+				m_normal = outer.n;
+				m_depth = outer.d;
+				m_result.rank = 3;
+				m_result.c[0] = outer.c[0];
+				m_result.c[1] = outer.c[1];
+				m_result.c[2] = outer.c[2];
+				m_result.p[0] = btCross(outer.c[1]->w - projection,
+										outer.c[2]->w - projection)
+									.length();
+				m_result.p[1] = btCross(outer.c[2]->w - projection,
+										outer.c[0]->w - projection)
+									.length();
+				m_result.p[2] = btCross(outer.c[0]->w - projection,
+										outer.c[1]->w - projection)
+									.length();
+				const btScalar sum = m_result.p[0] + m_result.p[1] + m_result.p[2];
+				m_result.p[0] /= sum;
+				m_result.p[1] /= sum;
+				m_result.p[2] /= sum;
+				return (m_status);
+			}
+		}
+		/* Fallback		*/
+		m_status = eStatus::FallBack;
+		m_normal = -guess;
+		const btScalar nl = m_normal.length();
+		if (nl > 0)
+			m_normal = m_normal / nl;
+		else
+			m_normal = btVector3(1, 0, 0);
+		m_depth = 0;
+		m_result.rank = 1;
+		m_result.c[0] = simplex.c[0];
+		m_result.p[0] = 1;
+		return (m_status);
+	}
+	bool getedgedist(sFace* face, sSV* a, sSV* b, btScalar& dist)
+	{
+		const btVector3 ba = b->w - a->w;
+		const btVector3 n_ab = btCross(ba, face->n);   // Outward facing edge normal direction, on triangle plane
+		const btScalar a_dot_nab = btDot(a->w, n_ab);  // Only care about the sign to determine inside/outside, so not normalization required
+
+		if (a_dot_nab < 0)
+		{
+			// Outside of edge a->b
+
+			const btScalar ba_l2 = ba.length2();
+			const btScalar a_dot_ba = btDot(a->w, ba);
+			const btScalar b_dot_ba = btDot(b->w, ba);
+
+			if (a_dot_ba > 0)
+			{
+				// Pick distance vertex a
+				dist = a->w.length();
+			}
+			else if (b_dot_ba < 0)
+			{
+				// Pick distance vertex b
+				dist = b->w.length();
+			}
+			else
+			{
+				// Pick distance to edge a->b
+				const btScalar a_dot_b = btDot(a->w, b->w);
+				dist = btSqrt(btMax((a->w.length2() * b->w.length2() - a_dot_b * a_dot_b) / ba_l2, (btScalar)0));
+			}
+
+			return true;
+		}
+
+		return false;
+	}
+	sFace* newface(sSV* a, sSV* b, sSV* c, bool forced)
+	{
+		if (m_stock.root)
+		{
+			sFace* face = m_stock.root;
+			remove(m_stock, face);
+			append(m_hull, face);
+			face->pass = 0;
+			face->c[0] = a;
+			face->c[1] = b;
+			face->c[2] = c;
+			face->n = btCross(b->w - a->w, c->w - a->w);
+			const btScalar l = face->n.length();
+			const bool v = l > EPA_ACCURACY;
+
+			if (v)
+			{
+				if (!(getedgedist(face, a, b, face->d) ||
+					  getedgedist(face, b, c, face->d) ||
+					  getedgedist(face, c, a, face->d)))
+				{
+					// Origin projects to the interior of the triangle
+					// Use distance to triangle plane
+					face->d = btDot(a->w, face->n) / l;
+				}
+
+				face->n /= l;
+				if (forced || (face->d >= -EPA_PLANE_EPS))
+				{
+					return face;
+				}
+				else
+					m_status = eStatus::NonConvex;
+			}
+			else
+				m_status = eStatus::Degenerated;
+
+			remove(m_hull, face);
+			append(m_stock, face);
+			return 0;
+		}
+		m_status = m_stock.root ? eStatus::OutOfVertices : eStatus::OutOfFaces;
+		return 0;
+	}
+	sFace* findbest()
+	{
+		sFace* minf = m_hull.root;
+		btScalar mind = minf->d * minf->d;
+		for (sFace* f = minf->l[1]; f; f = f->l[1])
+		{
+			const btScalar sqd = f->d * f->d;
+			if (sqd < mind)
+			{
+				minf = f;
+				mind = sqd;
+			}
+		}
+		return (minf);
+	}
+	bool expand(U pass, sSV* w, sFace* f, U e, sHorizon& horizon)
+	{
+		static const U i1m3[] = {1, 2, 0};
+		static const U i2m3[] = {2, 0, 1};
+		if (f->pass != pass)
+		{
+			const U e1 = i1m3[e];
+			if ((btDot(f->n, w->w) - f->d) < -EPA_PLANE_EPS)
+			{
+				sFace* nf = newface(f->c[e1], f->c[e], w, false);
+				if (nf)
+				{
+					bind(nf, 0, f, e);
+					if (horizon.cf)
+						bind(horizon.cf, 1, nf, 2);
+					else
+						horizon.ff = nf;
+					horizon.cf = nf;
+					++horizon.nf;
+					return (true);
+				}
+			}
+			else
+			{
+				const U e2 = i2m3[e];
+				f->pass = (U1)pass;
+				if (expand(pass, w, f->f[e1], f->e[e1], horizon) &&
+					expand(pass, w, f->f[e2], f->e[e2], horizon))
+				{
+					remove(m_hull, f);
+					append(m_stock, f);
+					return (true);
+				}
+			}
+		}
+		return (false);
+	}
+};
+
+//
+static void Initialize(const btConvexShape* shape0, const btTransform& wtrs0,
+					   const btConvexShape* shape1, const btTransform& wtrs1,
+					   btGjkEpaSolver2::sResults& results,
+					   tShape& shape,
+					   bool withmargins)
+{
+	/* Results		*/
+	results.witnesses[0] =
+		results.witnesses[1] = btVector3(0, 0, 0);
+	results.status = btGjkEpaSolver2::sResults::Separated;
+	/* Shape		*/
+	shape.m_shapes[0] = shape0;
+	shape.m_shapes[1] = shape1;
+	shape.m_toshape1 = wtrs1.getBasis().transposeTimes(wtrs0.getBasis());
+	shape.m_toshape0 = wtrs0.inverseTimes(wtrs1);
+	shape.EnableMargin(withmargins);
 }
 
+}  // namespace gjkepa2_impl
+
 //
 // Api
 //
 
-using namespace	gjkepa2_impl;
+using namespace gjkepa2_impl;
 
 //
-int			btGjkEpaSolver2::StackSizeRequirement()
+int btGjkEpaSolver2::StackSizeRequirement()
 {
-	return(sizeof(GJK)+sizeof(EPA));
+	return (sizeof(GJK) + sizeof(EPA));
 }
 
 //
-bool		btGjkEpaSolver2::Distance(	const btConvexShape*	shape0,
-									  const btTransform&		wtrs0,
-									  const btConvexShape*	shape1,
-									  const btTransform&		wtrs1,
-									  const btVector3&		guess,
-									  sResults&				results)
+bool btGjkEpaSolver2::Distance(const btConvexShape* shape0,
+							   const btTransform& wtrs0,
+							   const btConvexShape* shape1,
+							   const btTransform& wtrs1,
+							   const btVector3& guess,
+							   sResults& results)
 {
-	tShape			shape;
-	Initialize(shape0,wtrs0,shape1,wtrs1,results,shape,false);
-	GJK				gjk;
-	GJK::eStatus::_	gjk_status=gjk.Evaluate(shape,guess);
-	if(gjk_status==GJK::eStatus::Valid)
+	tShape shape;
+	Initialize(shape0, wtrs0, shape1, wtrs1, results, shape, false);
+	GJK gjk;
+	GJK::eStatus::_ gjk_status = gjk.Evaluate(shape, guess);
+	if (gjk_status == GJK::eStatus::Valid)
 	{
-		btVector3	w0=btVector3(0,0,0);
-		btVector3	w1=btVector3(0,0,0);
-		for(U i=0;i<gjk.m_simplex->rank;++i)
+		btVector3 w0 = btVector3(0, 0, 0);
+		btVector3 w1 = btVector3(0, 0, 0);
+		for (U i = 0; i < gjk.m_simplex->rank; ++i)
 		{
-			const btScalar	p=gjk.m_simplex->p[i];
-			w0+=shape.Support( gjk.m_simplex->c[i]->d,0)*p;
-			w1+=shape.Support(-gjk.m_simplex->c[i]->d,1)*p;
+			const btScalar p = gjk.m_simplex->p[i];
+			w0 += shape.Support(gjk.m_simplex->c[i]->d, 0) * p;
+			w1 += shape.Support(-gjk.m_simplex->c[i]->d, 1) * p;
 		}
-		results.witnesses[0]	=	wtrs0*w0;
-		results.witnesses[1]	=	wtrs0*w1;
-		results.normal			=	w0-w1;
-		results.distance		=	results.normal.length();
-		results.normal			/=	results.distance>GJK_MIN_DISTANCE?results.distance:1;
-		return(true);
+		results.witnesses[0] = wtrs0 * w0;
+		results.witnesses[1] = wtrs0 * w1;
+		results.normal = w0 - w1;
+		results.distance = results.normal.length();
+		results.normal /= results.distance > GJK_MIN_DISTANCE ? results.distance : 1;
+		return (true);
 	}
 	else
 	{
-		results.status	=	gjk_status==GJK::eStatus::Inside?
-			sResults::Penetrating	:
-		sResults::GJK_Failed	;
-		return(false);
+		results.status = gjk_status == GJK::eStatus::Inside ? sResults::Penetrating : sResults::GJK_Failed;
+		return (false);
 	}
 }
 
 //
-bool	btGjkEpaSolver2::Penetration(	const btConvexShape*	shape0,
-									 const btTransform&		wtrs0,
-									 const btConvexShape*	shape1,
-									 const btTransform&		wtrs1,
-									 const btVector3&		guess,
-									 sResults&				results,
-									 bool					usemargins)
+bool btGjkEpaSolver2::Penetration(const btConvexShape* shape0,
+								  const btTransform& wtrs0,
+								  const btConvexShape* shape1,
+								  const btTransform& wtrs1,
+								  const btVector3& guess,
+								  sResults& results,
+								  bool usemargins)
 {
-	tShape			shape;
-	Initialize(shape0,wtrs0,shape1,wtrs1,results,shape,usemargins);
-	GJK				gjk;	
-	GJK::eStatus::_	gjk_status=gjk.Evaluate(shape,-guess);
-	switch(gjk_status)
+	tShape shape;
+	Initialize(shape0, wtrs0, shape1, wtrs1, results, shape, usemargins);
+	GJK gjk;
+	GJK::eStatus::_ gjk_status = gjk.Evaluate(shape, -guess);
+	switch (gjk_status)
 	{
-	case	GJK::eStatus::Inside:
+		case GJK::eStatus::Inside:
 		{
-			EPA				epa;
-			EPA::eStatus::_	epa_status=epa.Evaluate(gjk,-guess);
-			if(epa_status!=EPA::eStatus::Failed)
+			EPA epa;
+			EPA::eStatus::_ epa_status = epa.Evaluate(gjk, -guess);
+			if (epa_status != EPA::eStatus::Failed)
 			{
-				btVector3	w0=btVector3(0,0,0);
-				for(U i=0;i<epa.m_result.rank;++i)
+				btVector3 w0 = btVector3(0, 0, 0);
+				for (U i = 0; i < epa.m_result.rank; ++i)
 				{
-					w0+=shape.Support(epa.m_result.c[i]->d,0)*epa.m_result.p[i];
+					w0 += shape.Support(epa.m_result.c[i]->d, 0) * epa.m_result.p[i];
 				}
-				results.status			=	sResults::Penetrating;
-				results.witnesses[0]	=	wtrs0*w0;
-				results.witnesses[1]	=	wtrs0*(w0-epa.m_normal*epa.m_depth);
-				results.normal			=	-epa.m_normal;
-				results.distance		=	-epa.m_depth;
-				return(true);
-			} else results.status=sResults::EPA_Failed;
+				results.status = sResults::Penetrating;
+				results.witnesses[0] = wtrs0 * w0;
+				results.witnesses[1] = wtrs0 * (w0 - epa.m_normal * epa.m_depth);
+				results.normal = -epa.m_normal;
+				results.distance = -epa.m_depth;
+				return (true);
+			}
+			else
+				results.status = sResults::EPA_Failed;
 		}
 		break;
-	case	GJK::eStatus::Failed:
-		results.status=sResults::GJK_Failed;
-		break;
+		case GJK::eStatus::Failed:
+			results.status = sResults::GJK_Failed;
+			break;
 		default:
-					{
-					}
+		{
+		}
 	}
-	return(false);
+	return (false);
 }
 
 #ifndef __SPU__
 //
-btScalar	btGjkEpaSolver2::SignedDistance(const btVector3& position,
-											btScalar margin,
-											const btConvexShape* shape0,
-											const btTransform& wtrs0,
-											sResults& results)
+btScalar btGjkEpaSolver2::SignedDistance(const btVector3& position,
+										 btScalar margin,
+										 const btConvexShape* shape0,
+										 const btTransform& wtrs0,
+										 sResults& results)
 {
-	tShape			shape;
-	btSphereShape	shape1(margin);
-	btTransform		wtrs1(btQuaternion(0,0,0,1),position);
-	Initialize(shape0,wtrs0,&shape1,wtrs1,results,shape,false);
-	GJK				gjk;	
-	GJK::eStatus::_	gjk_status=gjk.Evaluate(shape,btVector3(1,1,1));
-	if(gjk_status==GJK::eStatus::Valid)
+	tShape shape;
+	btSphereShape shape1(margin);
+	btTransform wtrs1(btQuaternion(0, 0, 0, 1), position);
+	Initialize(shape0, wtrs0, &shape1, wtrs1, results, shape, false);
+	GJK gjk;
+	GJK::eStatus::_ gjk_status = gjk.Evaluate(shape, btVector3(1, 1, 1));
+	if (gjk_status == GJK::eStatus::Valid)
 	{
-		btVector3	w0=btVector3(0,0,0);
-		btVector3	w1=btVector3(0,0,0);
-		for(U i=0;i<gjk.m_simplex->rank;++i)
+		btVector3 w0 = btVector3(0, 0, 0);
+		btVector3 w1 = btVector3(0, 0, 0);
+		for (U i = 0; i < gjk.m_simplex->rank; ++i)
 		{
-			const btScalar	p=gjk.m_simplex->p[i];
-			w0+=shape.Support( gjk.m_simplex->c[i]->d,0)*p;
-			w1+=shape.Support(-gjk.m_simplex->c[i]->d,1)*p;
+			const btScalar p = gjk.m_simplex->p[i];
+			w0 += shape.Support(gjk.m_simplex->c[i]->d, 0) * p;
+			w1 += shape.Support(-gjk.m_simplex->c[i]->d, 1) * p;
 		}
-		results.witnesses[0]	=	wtrs0*w0;
-		results.witnesses[1]	=	wtrs0*w1;
-		const btVector3	delta=	results.witnesses[1]-
-			results.witnesses[0];
-		const btScalar	margin=	shape0->getMarginNonVirtual()+
-			shape1.getMarginNonVirtual();
-		const btScalar	length=	delta.length();	
-		results.normal			=	delta/length;
-		results.witnesses[0]	+=	results.normal*margin;
-		return(length-margin);
+		results.witnesses[0] = wtrs0 * w0;
+		results.witnesses[1] = wtrs0 * w1;
+		const btVector3 delta = results.witnesses[1] -
+								results.witnesses[0];
+		const btScalar margin = shape0->getMarginNonVirtual() +
+								shape1.getMarginNonVirtual();
+		const btScalar length = delta.length();
+		results.normal = delta / length;
+		results.witnesses[0] += results.normal * margin;
+		return (length - margin);
 	}
 	else
 	{
-		if(gjk_status==GJK::eStatus::Inside)
+		if (gjk_status == GJK::eStatus::Inside)
 		{
-			if(Penetration(shape0,wtrs0,&shape1,wtrs1,gjk.m_ray,results))
+			if (Penetration(shape0, wtrs0, &shape1, wtrs1, gjk.m_ray, results))
 			{
-				const btVector3	delta=	results.witnesses[0]-
-					results.witnesses[1];
-				const btScalar	length=	delta.length();
+				const btVector3 delta = results.witnesses[0] -
+										results.witnesses[1];
+				const btScalar length = delta.length();
 				if (length >= SIMD_EPSILON)
-					results.normal	=	delta/length;			
-				return(-length);
+					results.normal = delta / length;
+				return (-length);
 			}
-		}	
+		}
 	}
-	return(SIMD_INFINITY);
+	return (SIMD_INFINITY);
 }
 
 //
-bool	btGjkEpaSolver2::SignedDistance(const btConvexShape*	shape0,
-										const btTransform&		wtrs0,
-										const btConvexShape*	shape1,
-										const btTransform&		wtrs1,
-										const btVector3&		guess,
-										sResults&				results)
+bool btGjkEpaSolver2::SignedDistance(const btConvexShape* shape0,
+									 const btTransform& wtrs0,
+									 const btConvexShape* shape1,
+									 const btTransform& wtrs1,
+									 const btVector3& guess,
+									 sResults& results)
 {
-	if(!Distance(shape0,wtrs0,shape1,wtrs1,guess,results))
-		return(Penetration(shape0,wtrs0,shape1,wtrs1,guess,results,false));
+	if (!Distance(shape0, wtrs0, shape1, wtrs1, guess, results))
+		return (Penetration(shape0, wtrs0, shape1, wtrs1, guess, results, false));
 	else
-		return(true);
+		return (true);
 }
-#endif //__SPU__
+#endif  //__SPU__
 
-/* Symbols cleanup		*/ 
+/* Symbols cleanup		*/
 
 #undef GJK_MAX_ITERATIONS
 #undef GJK_ACCURACY
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkEpa2.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkEpa2.h
index ac501d5ecfe..893daea3f54 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkEpa2.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkEpa2.h
@@ -28,48 +28,46 @@ GJK-EPA collision solver by Nathanael Presson, 2008
 #include "BulletCollision/CollisionShapes/btConvexShape.h"
 
 ///btGjkEpaSolver contributed under zlib by Nathanael Presson
-struct	btGjkEpaSolver2
+struct btGjkEpaSolver2
 {
-struct	sResults
+	struct sResults
 	{
-	enum eStatus
+		enum eStatus
 		{
-		Separated,		/* Shapes doesnt penetrate												*/ 
-		Penetrating,	/* Shapes are penetrating												*/ 
-		GJK_Failed,		/* GJK phase fail, no big issue, shapes are probably just 'touching'	*/ 
-		EPA_Failed		/* EPA phase fail, bigger problem, need to save parameters, and debug	*/ 
-		}		status;
-	btVector3	witnesses[2];
-	btVector3	normal;
-	btScalar	distance;
+			Separated,   /* Shapes doesnt penetrate												*/
+			Penetrating, /* Shapes are penetrating												*/
+			GJK_Failed,  /* GJK phase fail, no big issue, shapes are probably just 'touching'	*/
+			EPA_Failed   /* EPA phase fail, bigger problem, need to save parameters, and debug	*/
+		} status;
+		btVector3 witnesses[2];
+		btVector3 normal;
+		btScalar distance;
 	};
 
-static int		StackSizeRequirement();
+	static int StackSizeRequirement();
 
-static bool		Distance(	const btConvexShape* shape0,const btTransform& wtrs0,
-							const btConvexShape* shape1,const btTransform& wtrs1,
-							const btVector3& guess,
-							sResults& results);
+	static bool Distance(const btConvexShape* shape0, const btTransform& wtrs0,
+						 const btConvexShape* shape1, const btTransform& wtrs1,
+						 const btVector3& guess,
+						 sResults& results);
 
-static bool		Penetration(const btConvexShape* shape0,const btTransform& wtrs0,
-							const btConvexShape* shape1,const btTransform& wtrs1,
+	static bool Penetration(const btConvexShape* shape0, const btTransform& wtrs0,
+							const btConvexShape* shape1, const btTransform& wtrs1,
 							const btVector3& guess,
 							sResults& results,
-							bool usemargins=true);
+							bool usemargins = true);
 #ifndef __SPU__
-static btScalar	SignedDistance(	const btVector3& position,
-								btScalar margin,
-								const btConvexShape* shape,
-								const btTransform& wtrs,
-								sResults& results);
-							
-static bool		SignedDistance(	const btConvexShape* shape0,const btTransform& wtrs0,
-								const btConvexShape* shape1,const btTransform& wtrs1,
-								const btVector3& guess,
-								sResults& results);
-#endif //__SPU__
+	static btScalar SignedDistance(const btVector3& position,
+								   btScalar margin,
+								   const btConvexShape* shape,
+								   const btTransform& wtrs,
+								   sResults& results);
 
+	static bool SignedDistance(const btConvexShape* shape0, const btTransform& wtrs0,
+							   const btConvexShape* shape1, const btTransform& wtrs1,
+							   const btVector3& guess,
+							   sResults& results);
+#endif  //__SPU__
 };
 
-#endif //BT_GJK_EPA2_H
-
+#endif  //BT_GJK_EPA2_H
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkEpa3.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkEpa3.h
index ce1f24bc50d..6fedbbb3e5f 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkEpa3.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkEpa3.h
@@ -29,915 +29,946 @@ Improvements and refactoring by Erwin Coumans, 2008-2014
 #include "LinearMath/btTransform.h"
 #include "btGjkCollisionDescription.h"
 
-
-
-struct	btGjkEpaSolver3
+struct btGjkEpaSolver3
 {
-struct	sResults
+	struct sResults
 	{
-	enum eStatus
+		enum eStatus
 		{
-		Separated,		/* Shapes doesnt penetrate												*/ 
-		Penetrating,	/* Shapes are penetrating												*/ 
-		GJK_Failed,		/* GJK phase fail, no big issue, shapes are probably just 'touching'	*/ 
-		EPA_Failed		/* EPA phase fail, bigger problem, need to save parameters, and debug	*/ 
-		}		status;
-	btVector3	witnesses[2];
-	btVector3	normal;
-	btScalar	distance;
+			Separated,   /* Shapes doesnt penetrate												*/
+			Penetrating, /* Shapes are penetrating												*/
+			GJK_Failed,  /* GJK phase fail, no big issue, shapes are probably just 'touching'	*/
+			EPA_Failed   /* EPA phase fail, bigger problem, need to save parameters, and debug	*/
+		} status;
+		btVector3 witnesses[2];
+		btVector3 normal;
+		btScalar distance;
 	};
-
-
 };
 
-
-
-#if defined(DEBUG) || defined (_DEBUG)
-#include <stdio.h> //for debug printf
+#if defined(DEBUG) || defined(_DEBUG)
+#include <stdio.h>  //for debug printf
 #ifdef __SPU__
 #include <spu_printf.h>
 #define printf spu_printf
-#endif //__SPU__
+#endif  //__SPU__
 #endif
 
+// Config
 
-    
-    // Config
-    
-    /* GJK	*/
-#define GJK_MAX_ITERATIONS	128
-#define GJK_ACCURARY		((btScalar)0.0001)
-#define GJK_MIN_DISTANCE	((btScalar)0.0001)
-#define GJK_DUPLICATED_EPS	((btScalar)0.0001)
-#define GJK_SIMPLEX2_EPS	((btScalar)0.0)
-#define GJK_SIMPLEX3_EPS	((btScalar)0.0)
-#define GJK_SIMPLEX4_EPS	((btScalar)0.0)
-    
-    /* EPA	*/
-#define EPA_MAX_VERTICES	64
-#define EPA_MAX_FACES		(EPA_MAX_VERTICES*2)
-#define EPA_MAX_ITERATIONS	255
-#define EPA_ACCURACY		((btScalar)0.0001)
-#define EPA_FALLBACK		(10*EPA_ACCURACY)
-#define EPA_PLANE_EPS		((btScalar)0.00001)
-#define EPA_INSIDE_EPS		((btScalar)0.01)
-    
-    
-    // Shorthands
-    typedef unsigned int	U;
-    typedef unsigned char	U1;
-    
-    // MinkowskiDiff
-    template <typename btConvexTemplate>
-    struct	MinkowskiDiff
-    {
-        const btConvexTemplate* m_convexAPtr;
-        const btConvexTemplate* m_convexBPtr;
-        
-        btMatrix3x3				m_toshape1;
-        btTransform				m_toshape0;
-        
-        bool					m_enableMargin;
-        
-        
-        MinkowskiDiff(const btConvexTemplate& a, const btConvexTemplate& b)
-        :m_convexAPtr(&a),
-        m_convexBPtr(&b)
-        {
-        }
-        
-        void					EnableMargin(bool enable)
-        {
-            m_enableMargin = enable;
-        }
-        inline btVector3		Support0(const btVector3& d) const
-        {
-            return m_convexAPtr->getLocalSupportWithMargin(d);
-        }
-        inline btVector3		Support1(const btVector3& d) const
-        {
-            return m_toshape0*m_convexBPtr->getLocalSupportWithMargin(m_toshape1*d);
-        }
-        
-        
-        inline btVector3		Support(const btVector3& d) const
-        {
-            return(Support0(d)-Support1(-d));
-        }
-        btVector3				Support(const btVector3& d,U index) const
-        {
-            if(index)
-                return(Support1(d));
-            else
-                return(Support0(d));
-        }
-    };
-    
-enum	eGjkStatus
-{
-    eGjkValid,
-    eGjkInside,
-    eGjkFailed
-};
+/* GJK	*/
+#define GJK_MAX_ITERATIONS 128
+#define GJK_ACCURARY ((btScalar)0.0001)
+#define GJK_MIN_DISTANCE ((btScalar)0.0001)
+#define GJK_DUPLICATED_EPS ((btScalar)0.0001)
+#define GJK_SIMPLEX2_EPS ((btScalar)0.0)
+#define GJK_SIMPLEX3_EPS ((btScalar)0.0)
+#define GJK_SIMPLEX4_EPS ((btScalar)0.0)
 
-    // GJK
-    template <typename btConvexTemplate>
-    struct	GJK
-    {
-        /* Types		*/
-        struct	sSV
-        {
-            btVector3	d,w;
-        };
-        struct	sSimplex
-        {
-            sSV*		c[4];
-            btScalar	p[4];
-            U			rank;
-        };
-        
-        /* Fields		*/
-        
-        MinkowskiDiff<btConvexTemplate>			m_shape;
-        btVector3		m_ray;
-        btScalar		m_distance;
-        sSimplex		m_simplices[2];
-        sSV				m_store[4];
-        sSV*			m_free[4];
-        U				m_nfree;
-        U				m_current;
-        sSimplex*		m_simplex;
-        eGjkStatus      m_status;
-        /* Methods		*/
-        
-        GJK(const btConvexTemplate& a, const btConvexTemplate& b)
-        :m_shape(a,b)
-        {
-            Initialize();
-        }
-        void				Initialize()
-        {
-            m_ray		=	btVector3(0,0,0);
-            m_nfree		=	0;
-            m_status	=	eGjkFailed;
-            m_current	=	0;
-            m_distance	=	0;
-        }
-        eGjkStatus			Evaluate(const MinkowskiDiff<btConvexTemplate>& shapearg,const btVector3& guess)
-        {
-            U			iterations=0;
-            btScalar	sqdist=0;
-            btScalar	alpha=0;
-            btVector3	lastw[4];
-            U			clastw=0;
-            /* Initialize solver		*/
-            m_free[0]			=	&m_store[0];
-            m_free[1]			=	&m_store[1];
-            m_free[2]			=	&m_store[2];
-            m_free[3]			=	&m_store[3];
-            m_nfree				=	4;
-            m_current			=	0;
-            m_status			=	eGjkValid;
-            m_shape				=	shapearg;
-            m_distance			=	0;
-            /* Initialize simplex		*/
-            m_simplices[0].rank	=	0;
-            m_ray				=	guess;
-            const btScalar	sqrl=	m_ray.length2();
-            appendvertice(m_simplices[0],sqrl>0?-m_ray:btVector3(1,0,0));
-            m_simplices[0].p[0]	=	1;
-            m_ray				=	m_simplices[0].c[0]->w;
-            sqdist				=	sqrl;
-            lastw[0]			=
-            lastw[1]			=
-            lastw[2]			=
-            lastw[3]			=	m_ray;
-            /* Loop						*/
-            do	{
-                const U		next=1-m_current;
-                sSimplex&	cs=m_simplices[m_current];
-                sSimplex&	ns=m_simplices[next];
-                /* Check zero							*/
-                const btScalar	rl=m_ray.length();
-                if(rl<GJK_MIN_DISTANCE)
-                {/* Touching or inside				*/
-                    m_status=eGjkInside;
-                    break;
-                }
-                /* Append new vertice in -'v' direction	*/
-                appendvertice(cs,-m_ray);
-                const btVector3&	w=cs.c[cs.rank-1]->w;
-                bool				found=false;
-                for(U i=0;i<4;++i)
-                {
-                    if((w-lastw[i]).length2()<GJK_DUPLICATED_EPS)
-                    { found=true;break; }
-                }
-                if(found)
-                {/* Return old simplex				*/
-                    removevertice(m_simplices[m_current]);
-                    break;
-                }
-                else
-                {/* Update lastw					*/
-                    lastw[clastw=(clastw+1)&3]=w;
-                }
-                /* Check for termination				*/
-                const btScalar	omega=btDot(m_ray,w)/rl;
-                alpha=btMax(omega,alpha);
-                if(((rl-alpha)-(GJK_ACCURARY*rl))<=0)
-                {/* Return old simplex				*/
-                    removevertice(m_simplices[m_current]);
-                    break;
-                }
-                /* Reduce simplex						*/
-                btScalar	weights[4];
-                U			mask=0;
-                switch(cs.rank)
-                {
-                    case	2:	sqdist=projectorigin(	cs.c[0]->w,
-                                                     cs.c[1]->w,
-                                                     weights,mask);break;
-                    case	3:	sqdist=projectorigin(	cs.c[0]->w,
-                                                     cs.c[1]->w,
-                                                     cs.c[2]->w,
-                                                     weights,mask);break;
-                    case	4:	sqdist=projectorigin(	cs.c[0]->w,
-                                                     cs.c[1]->w,
-                                                     cs.c[2]->w,
-                                                     cs.c[3]->w,
-                                                     weights,mask);break;
-                }
-                if(sqdist>=0)
-                {/* Valid	*/
-                    ns.rank		=	0;
-                    m_ray		=	btVector3(0,0,0);
-                    m_current	=	next;
-                    for(U i=0,ni=cs.rank;i<ni;++i)
-                    {
-                        if(mask&(1<<i))
-                        {
-                            ns.c[ns.rank]		=	cs.c[i];
-                            ns.p[ns.rank++]		=	weights[i];
-                            m_ray				+=	cs.c[i]->w*weights[i];
-                        }
-                        else
-                        {
-                            m_free[m_nfree++]	=	cs.c[i];
-                        }
-                    }
-                    if(mask==15) m_status=eGjkInside;
-                }
-                else
-                {/* Return old simplex				*/
-                    removevertice(m_simplices[m_current]);
-                    break;
-                }
-                m_status=((++iterations)<GJK_MAX_ITERATIONS)?m_status:eGjkFailed;
-            } while(m_status==eGjkValid);
-            m_simplex=&m_simplices[m_current];
-            switch(m_status)
-            {
-                case	eGjkValid:		m_distance=m_ray.length();break;
-                case	eGjkInside:	m_distance=0;break;
-                default:
-                {
-                }
-            }
-            return(m_status);
-        }
-        bool					EncloseOrigin()
-        {
-            switch(m_simplex->rank)
-            {
-                case	1:
-                {
-                    for(U i=0;i<3;++i)
-                    {
-                        btVector3		axis=btVector3(0,0,0);
-                        axis[i]=1;
-                        appendvertice(*m_simplex, axis);
-                        if(EncloseOrigin())	return(true);
-                        removevertice(*m_simplex);
-                        appendvertice(*m_simplex,-axis);
-                        if(EncloseOrigin())	return(true);
-                        removevertice(*m_simplex);
-                    }
-                }
-                    break;
-                case	2:
-                {
-                    const btVector3	d=m_simplex->c[1]->w-m_simplex->c[0]->w;
-                    for(U i=0;i<3;++i)
-                    {
-                        btVector3		axis=btVector3(0,0,0);
-                        axis[i]=1;
-                        const btVector3	p=btCross(d,axis);
-                        if(p.length2()>0)
-                        {
-                            appendvertice(*m_simplex, p);
-                            if(EncloseOrigin())	return(true);
-                            removevertice(*m_simplex);
-                            appendvertice(*m_simplex,-p);
-                            if(EncloseOrigin())	return(true);
-                            removevertice(*m_simplex);
-                        }
-                    }
-                }
-                    break;
-                case	3:
-                {
-                    const btVector3	n=btCross(m_simplex->c[1]->w-m_simplex->c[0]->w,
-                                              m_simplex->c[2]->w-m_simplex->c[0]->w);
-                    if(n.length2()>0)
-                    {
-                        appendvertice(*m_simplex,n);
-                        if(EncloseOrigin())	return(true);
-                        removevertice(*m_simplex);
-                        appendvertice(*m_simplex,-n);
-                        if(EncloseOrigin())	return(true);
-                        removevertice(*m_simplex);
-                    }
-                }
-                    break;
-                case	4:
-                {
-                    if(btFabs(det(	m_simplex->c[0]->w-m_simplex->c[3]->w,
-                                  m_simplex->c[1]->w-m_simplex->c[3]->w,
-                                  m_simplex->c[2]->w-m_simplex->c[3]->w))>0)
-                        return(true);
-                }
-                    break;
-            }
-            return(false);
-        }
-        /* Internals	*/
-        void				getsupport(const btVector3& d,sSV& sv) const
-        {
-            sv.d	=	d/d.length();
-            sv.w	=	m_shape.Support(sv.d);
-        }
-        void				removevertice(sSimplex& simplex)
-        {
-            m_free[m_nfree++]=simplex.c[--simplex.rank];
-        }
-        void				appendvertice(sSimplex& simplex,const btVector3& v)
-        {
-            simplex.p[simplex.rank]=0;
-            simplex.c[simplex.rank]=m_free[--m_nfree];
-            getsupport(v,*simplex.c[simplex.rank++]);
-        }
-        static btScalar		det(const btVector3& a,const btVector3& b,const btVector3& c)
-        {
-            return(	a.y()*b.z()*c.x()+a.z()*b.x()*c.y()-
-                   a.x()*b.z()*c.y()-a.y()*b.x()*c.z()+
-                   a.x()*b.y()*c.z()-a.z()*b.y()*c.x());
-        }
-        static btScalar		projectorigin(	const btVector3& a,
-                                          const btVector3& b,
-                                          btScalar* w,U& m)
-        {
-            const btVector3	d=b-a;
-            const btScalar	l=d.length2();
-            if(l>GJK_SIMPLEX2_EPS)
-            {
-                const btScalar	t(l>0?-btDot(a,d)/l:0);
-                if(t>=1)		{ w[0]=0;w[1]=1;m=2;return(b.length2()); }
-                else if(t<=0)	{ w[0]=1;w[1]=0;m=1;return(a.length2()); }
-                else			{ w[0]=1-(w[1]=t);m=3;return((a+d*t).length2()); }
-            }
-            return(-1);
-        }
-        static btScalar		projectorigin(	const btVector3& a,
-                                          const btVector3& b,
-                                          const btVector3& c,
-                                          btScalar* w,U& m)
-        {
-            static const U		imd3[]={1,2,0};
-            const btVector3*	vt[]={&a,&b,&c};
-            const btVector3		dl[]={a-b,b-c,c-a};
-            const btVector3		n=btCross(dl[0],dl[1]);
-            const btScalar		l=n.length2();
-            if(l>GJK_SIMPLEX3_EPS)
-            {
-                btScalar	mindist=-1;
-                btScalar	subw[2]={0.f,0.f};
-                U			subm(0);
-                for(U i=0;i<3;++i)
-                {
-                    if(btDot(*vt[i],btCross(dl[i],n))>0)
-                    {
-                        const U			j=imd3[i];
-                        const btScalar	subd(projectorigin(*vt[i],*vt[j],subw,subm));
-                        if((mindist<0)||(subd<mindist))
-                        {
-                            mindist		=	subd;
-                            m			=	static_cast<U>(((subm&1)?1<<i:0)+((subm&2)?1<<j:0));
-                            w[i]		=	subw[0];
-                            w[j]		=	subw[1];
-                            w[imd3[j]]	=	0;
-                        }
-                    }
-                }
-                if(mindist<0)
-                {
-                    const btScalar	d=btDot(a,n);
-                    const btScalar	s=btSqrt(l);
-                    const btVector3	p=n*(d/l);
-                    mindist	=	p.length2();
-                    m		=	7;
-                    w[0]	=	(btCross(dl[1],b-p)).length()/s;
-                    w[1]	=	(btCross(dl[2],c-p)).length()/s;
-                    w[2]	=	1-(w[0]+w[1]);
-                }
-                return(mindist);
-            }
-            return(-1);
-        }
-        static btScalar		projectorigin(	const btVector3& a,
-                                          const btVector3& b,
-                                          const btVector3& c,
-                                          const btVector3& d,
-                                          btScalar* w,U& m)
-        {
-            static const U		imd3[]={1,2,0};
-            const btVector3*	vt[]={&a,&b,&c,&d};
-            const btVector3		dl[]={a-d,b-d,c-d};
-            const btScalar		vl=det(dl[0],dl[1],dl[2]);
-            const bool			ng=(vl*btDot(a,btCross(b-c,a-b)))<=0;
-            if(ng&&(btFabs(vl)>GJK_SIMPLEX4_EPS))
-            {
-                btScalar	mindist=-1;
-                btScalar	subw[3]={0.f,0.f,0.f};
-                U			subm(0);
-                for(U i=0;i<3;++i)
-                {
-                    const U			j=imd3[i];
-                    const btScalar	s=vl*btDot(d,btCross(dl[i],dl[j]));
-                    if(s>0)
-                    {
-                        const btScalar	subd=projectorigin(*vt[i],*vt[j],d,subw,subm);
-                        if((mindist<0)||(subd<mindist))
-                        {
-                            mindist		=	subd;
-                            m			=	static_cast<U>((subm&1?1<<i:0)+
-                                                           (subm&2?1<<j:0)+
-                                                           (subm&4?8:0));
-                            w[i]		=	subw[0];
-                            w[j]		=	subw[1];
-                            w[imd3[j]]	=	0;
-                            w[3]		=	subw[2];
-                        }
-                    }
-                }
-                if(mindist<0)
-                {
-                    mindist	=	0;
-                    m		=	15;
-                    w[0]	=	det(c,b,d)/vl;
-                    w[1]	=	det(a,c,d)/vl;
-                    w[2]	=	det(b,a,d)/vl;
-                    w[3]	=	1-(w[0]+w[1]+w[2]);
-                }
-                return(mindist);
-            }
-            return(-1);
-        }
-    };
+/* EPA	*/
+#define EPA_MAX_VERTICES 64
+#define EPA_MAX_FACES (EPA_MAX_VERTICES * 2)
+#define EPA_MAX_ITERATIONS 255
+#define EPA_ACCURACY ((btScalar)0.0001)
+#define EPA_FALLBACK (10 * EPA_ACCURACY)
+#define EPA_PLANE_EPS ((btScalar)0.00001)
+#define EPA_INSIDE_EPS ((btScalar)0.01)
 
+// Shorthands
+typedef unsigned int U;
+typedef unsigned char U1;
 
-enum	eEpaStatus
-{
-    eEpaValid,
-    eEpaTouching,
-    eEpaDegenerated,
-    eEpaNonConvex,
-    eEpaInvalidHull,
-    eEpaOutOfFaces,
-    eEpaOutOfVertices,
-    eEpaAccuraryReached,
-    eEpaFallBack,
-    eEpaFailed
-};
-
-
-    // EPA
+// MinkowskiDiff
 template <typename btConvexTemplate>
-    struct	EPA
-    {
-        /* Types		*/
-       
-        struct	sFace
-        {
-            btVector3	n;
-            btScalar	d;
-            typename GJK<btConvexTemplate>::sSV*		c[3];
-            sFace*		f[3];
-            sFace*		l[2];
-            U1			e[3];
-            U1			pass;
-        };
-        struct	sList
-        {
-            sFace*		root;
-            U			count;
-            sList() : root(0),count(0)	{}
-        };
-        struct	sHorizon
-        {
-            sFace*		cf;
-            sFace*		ff;
-            U			nf;
-            sHorizon() : cf(0),ff(0),nf(0)	{}
-        };
-       
-        /* Fields		*/
-        eEpaStatus		m_status;
-        typename GJK<btConvexTemplate>::sSimplex	m_result;
-        btVector3		m_normal;
-        btScalar		m_depth;
-        typename GJK<btConvexTemplate>::sSV				m_sv_store[EPA_MAX_VERTICES];
-        sFace			m_fc_store[EPA_MAX_FACES];
-        U				m_nextsv;
-        sList			m_hull;
-        sList			m_stock;
-        /* Methods		*/
-        EPA()
-        {
-            Initialize();
-        }
-        
-        
-        static inline void		bind(sFace* fa,U ea,sFace* fb,U eb)
-        {
-            fa->e[ea]=(U1)eb;fa->f[ea]=fb;
-            fb->e[eb]=(U1)ea;fb->f[eb]=fa;
-        }
-        static inline void		append(sList& list,sFace* face)
-        {
-            face->l[0]	=	0;
-            face->l[1]	=	list.root;
-            if(list.root) list.root->l[0]=face;
-            list.root	=	face;
-            ++list.count;
-        }
-        static inline void		remove(sList& list,sFace* face)
-        {
-            if(face->l[1]) face->l[1]->l[0]=face->l[0];
-            if(face->l[0]) face->l[0]->l[1]=face->l[1];
-            if(face==list.root) list.root=face->l[1];
-            --list.count;
-        }
-        
-        
-        void				Initialize()
-        {
-            m_status	=	eEpaFailed;
-            m_normal	=	btVector3(0,0,0);
-            m_depth		=	0;
-            m_nextsv	=	0;
-            for(U i=0;i<EPA_MAX_FACES;++i)
-            {
-                append(m_stock,&m_fc_store[EPA_MAX_FACES-i-1]);
-            }
-        }
-        eEpaStatus			Evaluate(GJK<btConvexTemplate>& gjk,const btVector3& guess)
-        {
-            typename GJK<btConvexTemplate>::sSimplex&	simplex=*gjk.m_simplex;
-            if((simplex.rank>1)&&gjk.EncloseOrigin())
-            {
-                
-                /* Clean up				*/
-                while(m_hull.root)
-                {
-                    sFace*	f = m_hull.root;
-                    remove(m_hull,f);
-                    append(m_stock,f);
-                }
-                m_status	=	eEpaValid;
-                m_nextsv	=	0;
-                /* Orient simplex		*/
-                if(gjk.det(	simplex.c[0]->w-simplex.c[3]->w,
-                           simplex.c[1]->w-simplex.c[3]->w,
-                           simplex.c[2]->w-simplex.c[3]->w)<0)
-                {
-                    btSwap(simplex.c[0],simplex.c[1]);
-                    btSwap(simplex.p[0],simplex.p[1]);
-                }
-                /* Build initial hull	*/
-                sFace*	tetra[]={newface(simplex.c[0],simplex.c[1],simplex.c[2],true),
-                    newface(simplex.c[1],simplex.c[0],simplex.c[3],true),
-                    newface(simplex.c[2],simplex.c[1],simplex.c[3],true),
-                    newface(simplex.c[0],simplex.c[2],simplex.c[3],true)};
-                if(m_hull.count==4)
-                {
-                    sFace*		best=findbest();
-                    sFace		outer=*best;
-                    U			pass=0;
-                    U			iterations=0;
-                    bind(tetra[0],0,tetra[1],0);
-                    bind(tetra[0],1,tetra[2],0);
-                    bind(tetra[0],2,tetra[3],0);
-                    bind(tetra[1],1,tetra[3],2);
-                    bind(tetra[1],2,tetra[2],1);
-                    bind(tetra[2],2,tetra[3],1);
-                    m_status=eEpaValid;
-                    for(;iterations<EPA_MAX_ITERATIONS;++iterations)
-                    {
-                        if(m_nextsv<EPA_MAX_VERTICES)
-                        {
-                            sHorizon		horizon;
-                            typename GJK<btConvexTemplate>::sSV*			w=&m_sv_store[m_nextsv++];
-                            bool			valid=true;
-                            best->pass	=	(U1)(++pass);
-                            gjk.getsupport(best->n,*w);
-                            const btScalar	wdist=btDot(best->n,w->w)-best->d;
-                            if(wdist>EPA_ACCURACY)
-                            {
-                                for(U j=0;(j<3)&&valid;++j)
-                                {
-                                    valid&=expand(	pass,w,
-                                                  best->f[j],best->e[j],
-                                                  horizon);
-                                }
-                                if(valid&&(horizon.nf>=3))
-                                {
-                                    bind(horizon.cf,1,horizon.ff,2);
-                                    remove(m_hull,best);
-                                    append(m_stock,best);
-                                    best=findbest();
-                                    outer=*best;
-                                } else { m_status=eEpaInvalidHull;break; }
-                            } else { m_status=eEpaAccuraryReached;break; }
-                        } else { m_status=eEpaOutOfVertices;break; }
-                    }
-                    const btVector3	projection=outer.n*outer.d;
-                    m_normal	=	outer.n;
-                    m_depth		=	outer.d;
-                    m_result.rank	=	3;
-                    m_result.c[0]	=	outer.c[0];
-                    m_result.c[1]	=	outer.c[1];
-                    m_result.c[2]	=	outer.c[2];
-                    m_result.p[0]	=	btCross(	outer.c[1]->w-projection,
-                                                outer.c[2]->w-projection).length();
-                    m_result.p[1]	=	btCross(	outer.c[2]->w-projection,
-                                                outer.c[0]->w-projection).length();
-                    m_result.p[2]	=	btCross(	outer.c[0]->w-projection,
-                                                outer.c[1]->w-projection).length();
-                    const btScalar	sum=m_result.p[0]+m_result.p[1]+m_result.p[2];
-                    m_result.p[0]	/=	sum;
-                    m_result.p[1]	/=	sum;
-                    m_result.p[2]	/=	sum;
-                    return(m_status);
-                }
-            }
-            /* Fallback		*/
-            m_status	=	eEpaFallBack;
-            m_normal	=	-guess;
-            const btScalar	nl=m_normal.length();
-            if(nl>0)
-                m_normal	=	m_normal/nl;
-            else
-                m_normal	=	btVector3(1,0,0);
-            m_depth	=	0;
-            m_result.rank=1;
-            m_result.c[0]=simplex.c[0];
-            m_result.p[0]=1;
-            return(m_status);
-        }
-        bool getedgedist(sFace* face, typename GJK<btConvexTemplate>::sSV* a, typename GJK<btConvexTemplate>::sSV* b, btScalar& dist)
-        {
-            const btVector3 ba = b->w - a->w;
-            const btVector3 n_ab = btCross(ba, face->n); // Outward facing edge normal direction, on triangle plane
-            const btScalar a_dot_nab = btDot(a->w, n_ab); // Only care about the sign to determine inside/outside, so not normalization required
-            
-            if(a_dot_nab < 0)
-            {
-                // Outside of edge a->b
-                
-                const btScalar ba_l2 = ba.length2();
-                const btScalar a_dot_ba = btDot(a->w, ba);
-                const btScalar b_dot_ba = btDot(b->w, ba);
-                
-                if(a_dot_ba > 0)
-                {
-                    // Pick distance vertex a
-                    dist = a->w.length();
-                }
-                else if(b_dot_ba < 0)
-                {
-                    // Pick distance vertex b
-                    dist = b->w.length();
-                }
-                else
-                {
-                    // Pick distance to edge a->b
-                    const btScalar a_dot_b = btDot(a->w, b->w);
-                    dist = btSqrt(btMax((a->w.length2() * b->w.length2() - a_dot_b * a_dot_b) / ba_l2, (btScalar)0));
-                }
-                
-                return true;
-            }
-            
-            return false;
-        }
-        sFace*				newface(typename GJK<btConvexTemplate>::sSV* a,typename GJK<btConvexTemplate>::sSV* b,typename GJK<btConvexTemplate>::sSV* c,bool forced)
-        {
-            if(m_stock.root)
-            {
-                sFace*	face=m_stock.root;
-                remove(m_stock,face);
-                append(m_hull,face);
-                face->pass	=	0;
-                face->c[0]	=	a;
-                face->c[1]	=	b;
-                face->c[2]	=	c;
-                face->n		=	btCross(b->w-a->w,c->w-a->w);
-                const btScalar	l=face->n.length();
-                const bool		v=l>EPA_ACCURACY;
-                
-                if(v)
-                {
-                    if(!(getedgedist(face, a, b, face->d) ||
-                         getedgedist(face, b, c, face->d) ||
-                         getedgedist(face, c, a, face->d)))
-                    {
-                        // Origin projects to the interior of the triangle
-                        // Use distance to triangle plane
-                        face->d = btDot(a->w, face->n) / l;
-                    }
-                    
-                    face->n /= l;
-                    if(forced || (face->d >= -EPA_PLANE_EPS))
-                    {
-                        return face;
-                    }
-                    else
-                        m_status=eEpaNonConvex;
-                }
-                else
-                    m_status=eEpaDegenerated;
-                
-                remove(m_hull, face);
-                append(m_stock, face);
-                return 0;
-                
-            }
-            m_status = m_stock.root ? eEpaOutOfVertices : eEpaOutOfFaces;
-            return 0;
-        }
-        sFace*				findbest()
-        {
-            sFace*		minf=m_hull.root;
-            btScalar	mind=minf->d*minf->d;
-            for(sFace* f=minf->l[1];f;f=f->l[1])
-            {
-                const btScalar	sqd=f->d*f->d;
-                if(sqd<mind)
-                {
-                    minf=f;
-                    mind=sqd;
-                }
-            }
-            return(minf);
-        }
-        bool				expand(U pass,typename GJK<btConvexTemplate>::sSV* w,sFace* f,U e,sHorizon& horizon)
-        {
-            static const U	i1m3[]={1,2,0};
-            static const U	i2m3[]={2,0,1};
-            if(f->pass!=pass)
-            {
-                const U	e1=i1m3[e];
-                if((btDot(f->n,w->w)-f->d)<-EPA_PLANE_EPS)
-                {
-                    sFace*	nf=newface(f->c[e1],f->c[e],w,false);
-                    if(nf)
-                    {
-                        bind(nf,0,f,e);
-                        if(horizon.cf) bind(horizon.cf,1,nf,2); else horizon.ff=nf;
-                        horizon.cf=nf;
-                        ++horizon.nf;
-                        return(true);
-                    }
-                }
-                else
-                {
-                    const U	e2=i2m3[e];
-                    f->pass		=	(U1)pass;
-                    if(	expand(pass,w,f->f[e1],f->e[e1],horizon)&&
-                       expand(pass,w,f->f[e2],f->e[e2],horizon))
-                    {
-                        remove(m_hull,f);
-                        append(m_stock,f);
-                        return(true);
-                    }
-                }
-            }
-            return(false);
-        }
-        
-    };
-    
-    template <typename btConvexTemplate>
-    static void	Initialize(	const btConvexTemplate& a, const btConvexTemplate& b,
-                           btGjkEpaSolver3::sResults& results,
-                           MinkowskiDiff<btConvexTemplate>& shape)
-    {
-        /* Results		*/ 
-        results.witnesses[0]	=
-        results.witnesses[1]	=	btVector3(0,0,0);
-        results.status			=	btGjkEpaSolver3::sResults::Separated;
-        /* Shape		*/ 
-       
-        shape.m_toshape1		=	b.getWorldTransform().getBasis().transposeTimes(a.getWorldTransform().getBasis());
-        shape.m_toshape0		=	a.getWorldTransform().inverseTimes(b.getWorldTransform());
-        
-    }
-    
+struct MinkowskiDiff
+{
+	const btConvexTemplate* m_convexAPtr;
+	const btConvexTemplate* m_convexBPtr;
+
+	btMatrix3x3 m_toshape1;
+	btTransform m_toshape0;
+
+	bool m_enableMargin;
+
+	MinkowskiDiff(const btConvexTemplate& a, const btConvexTemplate& b)
+		: m_convexAPtr(&a),
+		  m_convexBPtr(&b)
+	{
+	}
+
+	void EnableMargin(bool enable)
+	{
+		m_enableMargin = enable;
+	}
+	inline btVector3 Support0(const btVector3& d) const
+	{
+		return m_convexAPtr->getLocalSupportWithMargin(d);
+	}
+	inline btVector3 Support1(const btVector3& d) const
+	{
+		return m_toshape0 * m_convexBPtr->getLocalSupportWithMargin(m_toshape1 * d);
+	}
+
+	inline btVector3 Support(const btVector3& d) const
+	{
+		return (Support0(d) - Support1(-d));
+	}
+	btVector3 Support(const btVector3& d, U index) const
+	{
+		if (index)
+			return (Support1(d));
+		else
+			return (Support0(d));
+	}
+};
+
+enum eGjkStatus
+{
+	eGjkValid,
+	eGjkInside,
+	eGjkFailed
+};
+
+// GJK
+template <typename btConvexTemplate>
+struct GJK
+{
+	/* Types		*/
+	struct sSV
+	{
+		btVector3 d, w;
+	};
+	struct sSimplex
+	{
+		sSV* c[4];
+		btScalar p[4];
+		U rank;
+	};
+
+	/* Fields		*/
+
+	MinkowskiDiff<btConvexTemplate> m_shape;
+	btVector3 m_ray;
+	btScalar m_distance;
+	sSimplex m_simplices[2];
+	sSV m_store[4];
+	sSV* m_free[4];
+	U m_nfree;
+	U m_current;
+	sSimplex* m_simplex;
+	eGjkStatus m_status;
+	/* Methods		*/
+
+	GJK(const btConvexTemplate& a, const btConvexTemplate& b)
+		: m_shape(a, b)
+	{
+		Initialize();
+	}
+	void Initialize()
+	{
+		m_ray = btVector3(0, 0, 0);
+		m_nfree = 0;
+		m_status = eGjkFailed;
+		m_current = 0;
+		m_distance = 0;
+	}
+	eGjkStatus Evaluate(const MinkowskiDiff<btConvexTemplate>& shapearg, const btVector3& guess)
+	{
+		U iterations = 0;
+		btScalar sqdist = 0;
+		btScalar alpha = 0;
+		btVector3 lastw[4];
+		U clastw = 0;
+		/* Initialize solver		*/
+		m_free[0] = &m_store[0];
+		m_free[1] = &m_store[1];
+		m_free[2] = &m_store[2];
+		m_free[3] = &m_store[3];
+		m_nfree = 4;
+		m_current = 0;
+		m_status = eGjkValid;
+		m_shape = shapearg;
+		m_distance = 0;
+		/* Initialize simplex		*/
+		m_simplices[0].rank = 0;
+		m_ray = guess;
+		const btScalar sqrl = m_ray.length2();
+		appendvertice(m_simplices[0], sqrl > 0 ? -m_ray : btVector3(1, 0, 0));
+		m_simplices[0].p[0] = 1;
+		m_ray = m_simplices[0].c[0]->w;
+		sqdist = sqrl;
+		lastw[0] =
+			lastw[1] =
+				lastw[2] =
+					lastw[3] = m_ray;
+		/* Loop						*/
+		do
+		{
+			const U next = 1 - m_current;
+			sSimplex& cs = m_simplices[m_current];
+			sSimplex& ns = m_simplices[next];
+			/* Check zero							*/
+			const btScalar rl = m_ray.length();
+			if (rl < GJK_MIN_DISTANCE)
+			{ /* Touching or inside				*/
+				m_status = eGjkInside;
+				break;
+			}
+			/* Append new vertice in -'v' direction	*/
+			appendvertice(cs, -m_ray);
+			const btVector3& w = cs.c[cs.rank - 1]->w;
+			bool found = false;
+			for (U i = 0; i < 4; ++i)
+			{
+				if ((w - lastw[i]).length2() < GJK_DUPLICATED_EPS)
+				{
+					found = true;
+					break;
+				}
+			}
+			if (found)
+			{ /* Return old simplex				*/
+				removevertice(m_simplices[m_current]);
+				break;
+			}
+			else
+			{ /* Update lastw					*/
+				lastw[clastw = (clastw + 1) & 3] = w;
+			}
+			/* Check for termination				*/
+			const btScalar omega = btDot(m_ray, w) / rl;
+			alpha = btMax(omega, alpha);
+			if (((rl - alpha) - (GJK_ACCURARY * rl)) <= 0)
+			{ /* Return old simplex				*/
+				removevertice(m_simplices[m_current]);
+				break;
+			}
+			/* Reduce simplex						*/
+			btScalar weights[4];
+			U mask = 0;
+			switch (cs.rank)
+			{
+				case 2:
+					sqdist = projectorigin(cs.c[0]->w,
+										   cs.c[1]->w,
+										   weights, mask);
+					break;
+				case 3:
+					sqdist = projectorigin(cs.c[0]->w,
+										   cs.c[1]->w,
+										   cs.c[2]->w,
+										   weights, mask);
+					break;
+				case 4:
+					sqdist = projectorigin(cs.c[0]->w,
+										   cs.c[1]->w,
+										   cs.c[2]->w,
+										   cs.c[3]->w,
+										   weights, mask);
+					break;
+			}
+			if (sqdist >= 0)
+			{ /* Valid	*/
+				ns.rank = 0;
+				m_ray = btVector3(0, 0, 0);
+				m_current = next;
+				for (U i = 0, ni = cs.rank; i < ni; ++i)
+				{
+					if (mask & (1 << i))
+					{
+						ns.c[ns.rank] = cs.c[i];
+						ns.p[ns.rank++] = weights[i];
+						m_ray += cs.c[i]->w * weights[i];
+					}
+					else
+					{
+						m_free[m_nfree++] = cs.c[i];
+					}
+				}
+				if (mask == 15) m_status = eGjkInside;
+			}
+			else
+			{ /* Return old simplex				*/
+				removevertice(m_simplices[m_current]);
+				break;
+			}
+			m_status = ((++iterations) < GJK_MAX_ITERATIONS) ? m_status : eGjkFailed;
+		} while (m_status == eGjkValid);
+		m_simplex = &m_simplices[m_current];
+		switch (m_status)
+		{
+			case eGjkValid:
+				m_distance = m_ray.length();
+				break;
+			case eGjkInside:
+				m_distance = 0;
+				break;
+			default:
+			{
+			}
+		}
+		return (m_status);
+	}
+	bool EncloseOrigin()
+	{
+		switch (m_simplex->rank)
+		{
+			case 1:
+			{
+				for (U i = 0; i < 3; ++i)
+				{
+					btVector3 axis = btVector3(0, 0, 0);
+					axis[i] = 1;
+					appendvertice(*m_simplex, axis);
+					if (EncloseOrigin()) return (true);
+					removevertice(*m_simplex);
+					appendvertice(*m_simplex, -axis);
+					if (EncloseOrigin()) return (true);
+					removevertice(*m_simplex);
+				}
+			}
+			break;
+			case 2:
+			{
+				const btVector3 d = m_simplex->c[1]->w - m_simplex->c[0]->w;
+				for (U i = 0; i < 3; ++i)
+				{
+					btVector3 axis = btVector3(0, 0, 0);
+					axis[i] = 1;
+					const btVector3 p = btCross(d, axis);
+					if (p.length2() > 0)
+					{
+						appendvertice(*m_simplex, p);
+						if (EncloseOrigin()) return (true);
+						removevertice(*m_simplex);
+						appendvertice(*m_simplex, -p);
+						if (EncloseOrigin()) return (true);
+						removevertice(*m_simplex);
+					}
+				}
+			}
+			break;
+			case 3:
+			{
+				const btVector3 n = btCross(m_simplex->c[1]->w - m_simplex->c[0]->w,
+											m_simplex->c[2]->w - m_simplex->c[0]->w);
+				if (n.length2() > 0)
+				{
+					appendvertice(*m_simplex, n);
+					if (EncloseOrigin()) return (true);
+					removevertice(*m_simplex);
+					appendvertice(*m_simplex, -n);
+					if (EncloseOrigin()) return (true);
+					removevertice(*m_simplex);
+				}
+			}
+			break;
+			case 4:
+			{
+				if (btFabs(det(m_simplex->c[0]->w - m_simplex->c[3]->w,
+							   m_simplex->c[1]->w - m_simplex->c[3]->w,
+							   m_simplex->c[2]->w - m_simplex->c[3]->w)) > 0)
+					return (true);
+			}
+			break;
+		}
+		return (false);
+	}
+	/* Internals	*/
+	void getsupport(const btVector3& d, sSV& sv) const
+	{
+		sv.d = d / d.length();
+		sv.w = m_shape.Support(sv.d);
+	}
+	void removevertice(sSimplex& simplex)
+	{
+		m_free[m_nfree++] = simplex.c[--simplex.rank];
+	}
+	void appendvertice(sSimplex& simplex, const btVector3& v)
+	{
+		simplex.p[simplex.rank] = 0;
+		simplex.c[simplex.rank] = m_free[--m_nfree];
+		getsupport(v, *simplex.c[simplex.rank++]);
+	}
+	static btScalar det(const btVector3& a, const btVector3& b, const btVector3& c)
+	{
+		return (a.y() * b.z() * c.x() + a.z() * b.x() * c.y() -
+				a.x() * b.z() * c.y() - a.y() * b.x() * c.z() +
+				a.x() * b.y() * c.z() - a.z() * b.y() * c.x());
+	}
+	static btScalar projectorigin(const btVector3& a,
+								  const btVector3& b,
+								  btScalar* w, U& m)
+	{
+		const btVector3 d = b - a;
+		const btScalar l = d.length2();
+		if (l > GJK_SIMPLEX2_EPS)
+		{
+			const btScalar t(l > 0 ? -btDot(a, d) / l : 0);
+			if (t >= 1)
+			{
+				w[0] = 0;
+				w[1] = 1;
+				m = 2;
+				return (b.length2());
+			}
+			else if (t <= 0)
+			{
+				w[0] = 1;
+				w[1] = 0;
+				m = 1;
+				return (a.length2());
+			}
+			else
+			{
+				w[0] = 1 - (w[1] = t);
+				m = 3;
+				return ((a + d * t).length2());
+			}
+		}
+		return (-1);
+	}
+	static btScalar projectorigin(const btVector3& a,
+								  const btVector3& b,
+								  const btVector3& c,
+								  btScalar* w, U& m)
+	{
+		static const U imd3[] = {1, 2, 0};
+		const btVector3* vt[] = {&a, &b, &c};
+		const btVector3 dl[] = {a - b, b - c, c - a};
+		const btVector3 n = btCross(dl[0], dl[1]);
+		const btScalar l = n.length2();
+		if (l > GJK_SIMPLEX3_EPS)
+		{
+			btScalar mindist = -1;
+			btScalar subw[2] = {0.f, 0.f};
+			U subm(0);
+			for (U i = 0; i < 3; ++i)
+			{
+				if (btDot(*vt[i], btCross(dl[i], n)) > 0)
+				{
+					const U j = imd3[i];
+					const btScalar subd(projectorigin(*vt[i], *vt[j], subw, subm));
+					if ((mindist < 0) || (subd < mindist))
+					{
+						mindist = subd;
+						m = static_cast<U>(((subm & 1) ? 1 << i : 0) + ((subm & 2) ? 1 << j : 0));
+						w[i] = subw[0];
+						w[j] = subw[1];
+						w[imd3[j]] = 0;
+					}
+				}
+			}
+			if (mindist < 0)
+			{
+				const btScalar d = btDot(a, n);
+				const btScalar s = btSqrt(l);
+				const btVector3 p = n * (d / l);
+				mindist = p.length2();
+				m = 7;
+				w[0] = (btCross(dl[1], b - p)).length() / s;
+				w[1] = (btCross(dl[2], c - p)).length() / s;
+				w[2] = 1 - (w[0] + w[1]);
+			}
+			return (mindist);
+		}
+		return (-1);
+	}
+	static btScalar projectorigin(const btVector3& a,
+								  const btVector3& b,
+								  const btVector3& c,
+								  const btVector3& d,
+								  btScalar* w, U& m)
+	{
+		static const U imd3[] = {1, 2, 0};
+		const btVector3* vt[] = {&a, &b, &c, &d};
+		const btVector3 dl[] = {a - d, b - d, c - d};
+		const btScalar vl = det(dl[0], dl[1], dl[2]);
+		const bool ng = (vl * btDot(a, btCross(b - c, a - b))) <= 0;
+		if (ng && (btFabs(vl) > GJK_SIMPLEX4_EPS))
+		{
+			btScalar mindist = -1;
+			btScalar subw[3] = {0.f, 0.f, 0.f};
+			U subm(0);
+			for (U i = 0; i < 3; ++i)
+			{
+				const U j = imd3[i];
+				const btScalar s = vl * btDot(d, btCross(dl[i], dl[j]));
+				if (s > 0)
+				{
+					const btScalar subd = projectorigin(*vt[i], *vt[j], d, subw, subm);
+					if ((mindist < 0) || (subd < mindist))
+					{
+						mindist = subd;
+						m = static_cast<U>((subm & 1 ? 1 << i : 0) +
+										   (subm & 2 ? 1 << j : 0) +
+										   (subm & 4 ? 8 : 0));
+						w[i] = subw[0];
+						w[j] = subw[1];
+						w[imd3[j]] = 0;
+						w[3] = subw[2];
+					}
+				}
+			}
+			if (mindist < 0)
+			{
+				mindist = 0;
+				m = 15;
+				w[0] = det(c, b, d) / vl;
+				w[1] = det(a, c, d) / vl;
+				w[2] = det(b, a, d) / vl;
+				w[3] = 1 - (w[0] + w[1] + w[2]);
+			}
+			return (mindist);
+		}
+		return (-1);
+	}
+};
+
+enum eEpaStatus
+{
+	eEpaValid,
+	eEpaTouching,
+	eEpaDegenerated,
+	eEpaNonConvex,
+	eEpaInvalidHull,
+	eEpaOutOfFaces,
+	eEpaOutOfVertices,
+	eEpaAccuraryReached,
+	eEpaFallBack,
+	eEpaFailed
+};
+
+// EPA
+template <typename btConvexTemplate>
+struct EPA
+{
+	/* Types		*/
+
+	struct sFace
+	{
+		btVector3 n;
+		btScalar d;
+		typename GJK<btConvexTemplate>::sSV* c[3];
+		sFace* f[3];
+		sFace* l[2];
+		U1 e[3];
+		U1 pass;
+	};
+	struct sList
+	{
+		sFace* root;
+		U count;
+		sList() : root(0), count(0) {}
+	};
+	struct sHorizon
+	{
+		sFace* cf;
+		sFace* ff;
+		U nf;
+		sHorizon() : cf(0), ff(0), nf(0) {}
+	};
+
+	/* Fields		*/
+	eEpaStatus m_status;
+	typename GJK<btConvexTemplate>::sSimplex m_result;
+	btVector3 m_normal;
+	btScalar m_depth;
+	typename GJK<btConvexTemplate>::sSV m_sv_store[EPA_MAX_VERTICES];
+	sFace m_fc_store[EPA_MAX_FACES];
+	U m_nextsv;
+	sList m_hull;
+	sList m_stock;
+	/* Methods		*/
+	EPA()
+	{
+		Initialize();
+	}
+
+	static inline void bind(sFace* fa, U ea, sFace* fb, U eb)
+	{
+		fa->e[ea] = (U1)eb;
+		fa->f[ea] = fb;
+		fb->e[eb] = (U1)ea;
+		fb->f[eb] = fa;
+	}
+	static inline void append(sList& list, sFace* face)
+	{
+		face->l[0] = 0;
+		face->l[1] = list.root;
+		if (list.root) list.root->l[0] = face;
+		list.root = face;
+		++list.count;
+	}
+	static inline void remove(sList& list, sFace* face)
+	{
+		if (face->l[1]) face->l[1]->l[0] = face->l[0];
+		if (face->l[0]) face->l[0]->l[1] = face->l[1];
+		if (face == list.root) list.root = face->l[1];
+		--list.count;
+	}
+
+	void Initialize()
+	{
+		m_status = eEpaFailed;
+		m_normal = btVector3(0, 0, 0);
+		m_depth = 0;
+		m_nextsv = 0;
+		for (U i = 0; i < EPA_MAX_FACES; ++i)
+		{
+			append(m_stock, &m_fc_store[EPA_MAX_FACES - i - 1]);
+		}
+	}
+	eEpaStatus Evaluate(GJK<btConvexTemplate>& gjk, const btVector3& guess)
+	{
+		typename GJK<btConvexTemplate>::sSimplex& simplex = *gjk.m_simplex;
+		if ((simplex.rank > 1) && gjk.EncloseOrigin())
+		{
+			/* Clean up				*/
+			while (m_hull.root)
+			{
+				sFace* f = m_hull.root;
+				remove(m_hull, f);
+				append(m_stock, f);
+			}
+			m_status = eEpaValid;
+			m_nextsv = 0;
+			/* Orient simplex		*/
+			if (gjk.det(simplex.c[0]->w - simplex.c[3]->w,
+						simplex.c[1]->w - simplex.c[3]->w,
+						simplex.c[2]->w - simplex.c[3]->w) < 0)
+			{
+				btSwap(simplex.c[0], simplex.c[1]);
+				btSwap(simplex.p[0], simplex.p[1]);
+			}
+			/* Build initial hull	*/
+			sFace* tetra[] = {newface(simplex.c[0], simplex.c[1], simplex.c[2], true),
+							  newface(simplex.c[1], simplex.c[0], simplex.c[3], true),
+							  newface(simplex.c[2], simplex.c[1], simplex.c[3], true),
+							  newface(simplex.c[0], simplex.c[2], simplex.c[3], true)};
+			if (m_hull.count == 4)
+			{
+				sFace* best = findbest();
+				sFace outer = *best;
+				U pass = 0;
+				U iterations = 0;
+				bind(tetra[0], 0, tetra[1], 0);
+				bind(tetra[0], 1, tetra[2], 0);
+				bind(tetra[0], 2, tetra[3], 0);
+				bind(tetra[1], 1, tetra[3], 2);
+				bind(tetra[1], 2, tetra[2], 1);
+				bind(tetra[2], 2, tetra[3], 1);
+				m_status = eEpaValid;
+				for (; iterations < EPA_MAX_ITERATIONS; ++iterations)
+				{
+					if (m_nextsv < EPA_MAX_VERTICES)
+					{
+						sHorizon horizon;
+						typename GJK<btConvexTemplate>::sSV* w = &m_sv_store[m_nextsv++];
+						bool valid = true;
+						best->pass = (U1)(++pass);
+						gjk.getsupport(best->n, *w);
+						const btScalar wdist = btDot(best->n, w->w) - best->d;
+						if (wdist > EPA_ACCURACY)
+						{
+							for (U j = 0; (j < 3) && valid; ++j)
+							{
+								valid &= expand(pass, w,
+												best->f[j], best->e[j],
+												horizon);
+							}
+							if (valid && (horizon.nf >= 3))
+							{
+								bind(horizon.cf, 1, horizon.ff, 2);
+								remove(m_hull, best);
+								append(m_stock, best);
+								best = findbest();
+								outer = *best;
+							}
+							else
+							{
+								m_status = eEpaInvalidHull;
+								break;
+							}
+						}
+						else
+						{
+							m_status = eEpaAccuraryReached;
+							break;
+						}
+					}
+					else
+					{
+						m_status = eEpaOutOfVertices;
+						break;
+					}
+				}
+				const btVector3 projection = outer.n * outer.d;
+				m_normal = outer.n;
+				m_depth = outer.d;
+				m_result.rank = 3;
+				m_result.c[0] = outer.c[0];
+				m_result.c[1] = outer.c[1];
+				m_result.c[2] = outer.c[2];
+				m_result.p[0] = btCross(outer.c[1]->w - projection,
+										outer.c[2]->w - projection)
+									.length();
+				m_result.p[1] = btCross(outer.c[2]->w - projection,
+										outer.c[0]->w - projection)
+									.length();
+				m_result.p[2] = btCross(outer.c[0]->w - projection,
+										outer.c[1]->w - projection)
+									.length();
+				const btScalar sum = m_result.p[0] + m_result.p[1] + m_result.p[2];
+				m_result.p[0] /= sum;
+				m_result.p[1] /= sum;
+				m_result.p[2] /= sum;
+				return (m_status);
+			}
+		}
+		/* Fallback		*/
+		m_status = eEpaFallBack;
+		m_normal = -guess;
+		const btScalar nl = m_normal.length();
+		if (nl > 0)
+			m_normal = m_normal / nl;
+		else
+			m_normal = btVector3(1, 0, 0);
+		m_depth = 0;
+		m_result.rank = 1;
+		m_result.c[0] = simplex.c[0];
+		m_result.p[0] = 1;
+		return (m_status);
+	}
+	bool getedgedist(sFace* face, typename GJK<btConvexTemplate>::sSV* a, typename GJK<btConvexTemplate>::sSV* b, btScalar& dist)
+	{
+		const btVector3 ba = b->w - a->w;
+		const btVector3 n_ab = btCross(ba, face->n);   // Outward facing edge normal direction, on triangle plane
+		const btScalar a_dot_nab = btDot(a->w, n_ab);  // Only care about the sign to determine inside/outside, so not normalization required
+
+		if (a_dot_nab < 0)
+		{
+			// Outside of edge a->b
+
+			const btScalar ba_l2 = ba.length2();
+			const btScalar a_dot_ba = btDot(a->w, ba);
+			const btScalar b_dot_ba = btDot(b->w, ba);
+
+			if (a_dot_ba > 0)
+			{
+				// Pick distance vertex a
+				dist = a->w.length();
+			}
+			else if (b_dot_ba < 0)
+			{
+				// Pick distance vertex b
+				dist = b->w.length();
+			}
+			else
+			{
+				// Pick distance to edge a->b
+				const btScalar a_dot_b = btDot(a->w, b->w);
+				dist = btSqrt(btMax((a->w.length2() * b->w.length2() - a_dot_b * a_dot_b) / ba_l2, (btScalar)0));
+			}
+
+			return true;
+		}
+
+		return false;
+	}
+	sFace* newface(typename GJK<btConvexTemplate>::sSV* a, typename GJK<btConvexTemplate>::sSV* b, typename GJK<btConvexTemplate>::sSV* c, bool forced)
+	{
+		if (m_stock.root)
+		{
+			sFace* face = m_stock.root;
+			remove(m_stock, face);
+			append(m_hull, face);
+			face->pass = 0;
+			face->c[0] = a;
+			face->c[1] = b;
+			face->c[2] = c;
+			face->n = btCross(b->w - a->w, c->w - a->w);
+			const btScalar l = face->n.length();
+			const bool v = l > EPA_ACCURACY;
+
+			if (v)
+			{
+				if (!(getedgedist(face, a, b, face->d) ||
+					  getedgedist(face, b, c, face->d) ||
+					  getedgedist(face, c, a, face->d)))
+				{
+					// Origin projects to the interior of the triangle
+					// Use distance to triangle plane
+					face->d = btDot(a->w, face->n) / l;
+				}
+
+				face->n /= l;
+				if (forced || (face->d >= -EPA_PLANE_EPS))
+				{
+					return face;
+				}
+				else
+					m_status = eEpaNonConvex;
+			}
+			else
+				m_status = eEpaDegenerated;
+
+			remove(m_hull, face);
+			append(m_stock, face);
+			return 0;
+		}
+		m_status = m_stock.root ? eEpaOutOfVertices : eEpaOutOfFaces;
+		return 0;
+	}
+	sFace* findbest()
+	{
+		sFace* minf = m_hull.root;
+		btScalar mind = minf->d * minf->d;
+		for (sFace* f = minf->l[1]; f; f = f->l[1])
+		{
+			const btScalar sqd = f->d * f->d;
+			if (sqd < mind)
+			{
+				minf = f;
+				mind = sqd;
+			}
+		}
+		return (minf);
+	}
+	bool expand(U pass, typename GJK<btConvexTemplate>::sSV* w, sFace* f, U e, sHorizon& horizon)
+	{
+		static const U i1m3[] = {1, 2, 0};
+		static const U i2m3[] = {2, 0, 1};
+		if (f->pass != pass)
+		{
+			const U e1 = i1m3[e];
+			if ((btDot(f->n, w->w) - f->d) < -EPA_PLANE_EPS)
+			{
+				sFace* nf = newface(f->c[e1], f->c[e], w, false);
+				if (nf)
+				{
+					bind(nf, 0, f, e);
+					if (horizon.cf)
+						bind(horizon.cf, 1, nf, 2);
+					else
+						horizon.ff = nf;
+					horizon.cf = nf;
+					++horizon.nf;
+					return (true);
+				}
+			}
+			else
+			{
+				const U e2 = i2m3[e];
+				f->pass = (U1)pass;
+				if (expand(pass, w, f->f[e1], f->e[e1], horizon) &&
+					expand(pass, w, f->f[e2], f->e[e2], horizon))
+				{
+					remove(m_hull, f);
+					append(m_stock, f);
+					return (true);
+				}
+			}
+		}
+		return (false);
+	}
+};
+
+template <typename btConvexTemplate>
+static void Initialize(const btConvexTemplate& a, const btConvexTemplate& b,
+					   btGjkEpaSolver3::sResults& results,
+					   MinkowskiDiff<btConvexTemplate>& shape)
+{
+	/* Results		*/
+	results.witnesses[0] =
+		results.witnesses[1] = btVector3(0, 0, 0);
+	results.status = btGjkEpaSolver3::sResults::Separated;
+	/* Shape		*/
+
+	shape.m_toshape1 = b.getWorldTransform().getBasis().transposeTimes(a.getWorldTransform().getBasis());
+	shape.m_toshape0 = a.getWorldTransform().inverseTimes(b.getWorldTransform());
+}
 
 //
 // Api
 //
 
-
-
 //
 template <typename btConvexTemplate>
-bool		btGjkEpaSolver3_Distance(const btConvexTemplate& a, const btConvexTemplate& b,
-                                      const btVector3& guess,
-                                      btGjkEpaSolver3::sResults& results)
+bool btGjkEpaSolver3_Distance(const btConvexTemplate& a, const btConvexTemplate& b,
+							  const btVector3& guess,
+							  btGjkEpaSolver3::sResults& results)
 {
-    MinkowskiDiff<btConvexTemplate>			shape(a,b);
-    Initialize(a,b,results,shape);
-    GJK<btConvexTemplate>				gjk(a,b);
-    eGjkStatus	gjk_status=gjk.Evaluate(shape,guess);
-    if(gjk_status==eGjkValid)
-    {
-        btVector3	w0=btVector3(0,0,0);
-        btVector3	w1=btVector3(0,0,0);
-        for(U i=0;i<gjk.m_simplex->rank;++i)
-        {
-            const btScalar	p=gjk.m_simplex->p[i];
-            w0+=shape.Support( gjk.m_simplex->c[i]->d,0)*p;
-            w1+=shape.Support(-gjk.m_simplex->c[i]->d,1)*p;
-        }
-        results.witnesses[0]	=	a.getWorldTransform()*w0;
-        results.witnesses[1]	=	a.getWorldTransform()*w1;
-        results.normal			=	w0-w1;
-        results.distance		=	results.normal.length();
-        results.normal			/=	results.distance>GJK_MIN_DISTANCE?results.distance:1;
-        return(true);
-    }
-    else
-    {
-        results.status	=	gjk_status==eGjkInside?
-        btGjkEpaSolver3::sResults::Penetrating	:
-        btGjkEpaSolver3::sResults::GJK_Failed	;
-        return(false);
-    }
+	MinkowskiDiff<btConvexTemplate> shape(a, b);
+	Initialize(a, b, results, shape);
+	GJK<btConvexTemplate> gjk(a, b);
+	eGjkStatus gjk_status = gjk.Evaluate(shape, guess);
+	if (gjk_status == eGjkValid)
+	{
+		btVector3 w0 = btVector3(0, 0, 0);
+		btVector3 w1 = btVector3(0, 0, 0);
+		for (U i = 0; i < gjk.m_simplex->rank; ++i)
+		{
+			const btScalar p = gjk.m_simplex->p[i];
+			w0 += shape.Support(gjk.m_simplex->c[i]->d, 0) * p;
+			w1 += shape.Support(-gjk.m_simplex->c[i]->d, 1) * p;
+		}
+		results.witnesses[0] = a.getWorldTransform() * w0;
+		results.witnesses[1] = a.getWorldTransform() * w1;
+		results.normal = w0 - w1;
+		results.distance = results.normal.length();
+		results.normal /= results.distance > GJK_MIN_DISTANCE ? results.distance : 1;
+		return (true);
+	}
+	else
+	{
+		results.status = gjk_status == eGjkInside ? btGjkEpaSolver3::sResults::Penetrating : btGjkEpaSolver3::sResults::GJK_Failed;
+		return (false);
+	}
 }
 
-
 template <typename btConvexTemplate>
-bool	btGjkEpaSolver3_Penetration(const btConvexTemplate& a,
-                                     const btConvexTemplate& b,
-                                     const btVector3& guess,
-                                     btGjkEpaSolver3::sResults& results)
+bool btGjkEpaSolver3_Penetration(const btConvexTemplate& a,
+								 const btConvexTemplate& b,
+								 const btVector3& guess,
+								 btGjkEpaSolver3::sResults& results)
 {
-    MinkowskiDiff<btConvexTemplate>			shape(a,b);
-    Initialize(a,b,results,shape);
-    GJK<btConvexTemplate>				gjk(a,b);
-    eGjkStatus	gjk_status=gjk.Evaluate(shape,-guess);
-    switch(gjk_status)
-    {
-        case	eGjkInside:
-        {
-            EPA<btConvexTemplate>				epa;
-            eEpaStatus	epa_status=epa.Evaluate(gjk,-guess);
-            if(epa_status!=eEpaFailed)
-            {
-                btVector3	w0=btVector3(0,0,0);
-                for(U i=0;i<epa.m_result.rank;++i)
-                {
-                    w0+=shape.Support(epa.m_result.c[i]->d,0)*epa.m_result.p[i];
-                }
-                results.status			=	btGjkEpaSolver3::sResults::Penetrating;
-                results.witnesses[0]	=	a.getWorldTransform()*w0;
-                results.witnesses[1]	=	a.getWorldTransform()*(w0-epa.m_normal*epa.m_depth);
-                results.normal			=	-epa.m_normal;
-                results.distance		=	-epa.m_depth;
-                return(true);
-            } else results.status=btGjkEpaSolver3::sResults::EPA_Failed;
-        }
-            break;
-        case	eGjkFailed:
-            results.status=btGjkEpaSolver3::sResults::GJK_Failed;
-            break;
-        default:
-        {
-        }
-    }
-    return(false);
+	MinkowskiDiff<btConvexTemplate> shape(a, b);
+	Initialize(a, b, results, shape);
+	GJK<btConvexTemplate> gjk(a, b);
+	eGjkStatus gjk_status = gjk.Evaluate(shape, -guess);
+	switch (gjk_status)
+	{
+		case eGjkInside:
+		{
+			EPA<btConvexTemplate> epa;
+			eEpaStatus epa_status = epa.Evaluate(gjk, -guess);
+			if (epa_status != eEpaFailed)
+			{
+				btVector3 w0 = btVector3(0, 0, 0);
+				for (U i = 0; i < epa.m_result.rank; ++i)
+				{
+					w0 += shape.Support(epa.m_result.c[i]->d, 0) * epa.m_result.p[i];
+				}
+				results.status = btGjkEpaSolver3::sResults::Penetrating;
+				results.witnesses[0] = a.getWorldTransform() * w0;
+				results.witnesses[1] = a.getWorldTransform() * (w0 - epa.m_normal * epa.m_depth);
+				results.normal = -epa.m_normal;
+				results.distance = -epa.m_depth;
+				return (true);
+			}
+			else
+				results.status = btGjkEpaSolver3::sResults::EPA_Failed;
+		}
+		break;
+		case eGjkFailed:
+			results.status = btGjkEpaSolver3::sResults::GJK_Failed;
+			break;
+		default:
+		{
+		}
+	}
+	return (false);
 }
 
 #if 0
@@ -990,28 +1021,28 @@ int	btComputeGjkEpaPenetration2(const btCollisionDescription& colDesc, btDistanc
 #endif
 
 template <typename btConvexTemplate, typename btDistanceInfoTemplate>
-int	btComputeGjkDistance(const btConvexTemplate& a, const btConvexTemplate& b,
-                         const btGjkCollisionDescription& colDesc, btDistanceInfoTemplate* distInfo)
+int btComputeGjkDistance(const btConvexTemplate& a, const btConvexTemplate& b,
+						 const btGjkCollisionDescription& colDesc, btDistanceInfoTemplate* distInfo)
 {
-    btGjkEpaSolver3::sResults results;
-    btVector3 guess = colDesc.m_firstDir;
-    
-    bool isSeparated = btGjkEpaSolver3_Distance(	a,b,
-                                                 guess,
-                                                 results);
-    if (isSeparated)
-    {
-        distInfo->m_distance = results.distance;
-        distInfo->m_pointOnA= results.witnesses[0];
-        distInfo->m_pointOnB= results.witnesses[1];
-        distInfo->m_normalBtoA= results.normal;
-        return 0;
-    }
-    
-    return -1;
+	btGjkEpaSolver3::sResults results;
+	btVector3 guess = colDesc.m_firstDir;
+
+	bool isSeparated = btGjkEpaSolver3_Distance(a, b,
+												guess,
+												results);
+	if (isSeparated)
+	{
+		distInfo->m_distance = results.distance;
+		distInfo->m_pointOnA = results.witnesses[0];
+		distInfo->m_pointOnB = results.witnesses[1];
+		distInfo->m_normalBtoA = results.normal;
+		return 0;
+	}
+
+	return -1;
 }
 
-/* Symbols cleanup		*/ 
+/* Symbols cleanup		*/
 
 #undef GJK_MAX_ITERATIONS
 #undef GJK_ACCURARY
@@ -1029,7 +1060,4 @@ int	btComputeGjkDistance(const btConvexTemplate& a, const btConvexTemplate& b,
 #undef EPA_PLANE_EPS
 #undef EPA_INSIDE_EPS
 
-
-
-#endif //BT_GJK_EPA3_H
-
+#endif  //BT_GJK_EPA3_H
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkEpaPenetrationDepthSolver.cpp b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkEpaPenetrationDepthSolver.cpp
index b79f49d6110..07629229abc 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkEpaPenetrationDepthSolver.cpp
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkEpaPenetrationDepthSolver.cpp
@@ -18,23 +18,21 @@ subject to the following restrictions:
 #include "BulletCollision/CollisionShapes/btConvexShape.h"
 #include "btGjkEpaPenetrationDepthSolver.h"
 
-
 #include "BulletCollision/NarrowPhaseCollision/btGjkEpa2.h"
 
 bool btGjkEpaPenetrationDepthSolver::calcPenDepth(btSimplexSolverInterface& simplexSolver,
-	const btConvexShape* pConvexA, const btConvexShape* pConvexB,
-	const btTransform& transformA, const btTransform& transformB,
-	btVector3& v, btVector3& wWitnessOnA, btVector3& wWitnessOnB,
-	class btIDebugDraw* debugDraw)
+												  const btConvexShape* pConvexA, const btConvexShape* pConvexB,
+												  const btTransform& transformA, const btTransform& transformB,
+												  btVector3& v, btVector3& wWitnessOnA, btVector3& wWitnessOnB,
+												  class btIDebugDraw* debugDraw)
 {
-
 	(void)debugDraw;
 	(void)v;
 	(void)simplexSolver;
 
 	btVector3 guessVectors[] = {
-		btVector3(transformB.getOrigin() - transformA.getOrigin()).normalized(),
-		btVector3(transformA.getOrigin() - transformB.getOrigin()).normalized(),
+		btVector3(transformB.getOrigin() - transformA.getOrigin()).safeNormalize(),
+		btVector3(transformA.getOrigin() - transformB.getOrigin()).safeNormalize(),
 		btVector3(0, 0, 1),
 		btVector3(0, 1, 0),
 		btVector3(1, 0, 0),
@@ -49,13 +47,13 @@ bool btGjkEpaPenetrationDepthSolver::calcPenDepth(btSimplexSolverInterface& simp
 	for (int i = 0; i < numVectors; i++)
 	{
 		simplexSolver.reset();
-		btVector3	guessVector = guessVectors[i];
+		btVector3 guessVector = guessVectors[i];
 
-		btGjkEpaSolver2::sResults	results;
+		btGjkEpaSolver2::sResults results;
 
 		if (btGjkEpaSolver2::Penetration(pConvexA, transformA,
-			pConvexB, transformB,
-			guessVector, results))
+										 pConvexB, transformB,
+										 guessVector, results))
 
 		{
 			wWitnessOnA = results.witnesses[0];
@@ -81,4 +79,3 @@ bool btGjkEpaPenetrationDepthSolver::calcPenDepth(btSimplexSolverInterface& simp
 	v.setValue(0, 0, 0);
 	return false;
 }
-
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkEpaPenetrationDepthSolver.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkEpaPenetrationDepthSolver.h
index 1ed6340af3b..92d6df1729f 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkEpaPenetrationDepthSolver.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkEpaPenetrationDepthSolver.h
@@ -23,21 +23,18 @@ subject to the following restrictions:
 ///calculate the penetration depth between two convex shapes.
 class btGjkEpaPenetrationDepthSolver : public btConvexPenetrationDepthSolver
 {
-	public :
+public:
+	btGjkEpaPenetrationDepthSolver()
+	{
+	}
 
-		btGjkEpaPenetrationDepthSolver()
-		{
-		}
-
-		bool			calcPenDepth( btSimplexSolverInterface& simplexSolver,
-									  const btConvexShape* pConvexA, const btConvexShape* pConvexB,
-									  const btTransform& transformA, const btTransform& transformB,
-									  btVector3& v, btVector3& wWitnessOnA, btVector3& wWitnessOnB,
-									  class btIDebugDraw* debugDraw);
-
-	private :
+	bool calcPenDepth(btSimplexSolverInterface& simplexSolver,
+					  const btConvexShape* pConvexA, const btConvexShape* pConvexB,
+					  const btTransform& transformA, const btTransform& transformB,
+					  btVector3& v, btVector3& wWitnessOnA, btVector3& wWitnessOnB,
+					  class btIDebugDraw* debugDraw);
 
+private:
 };
 
-#endif	// BT_GJP_EPA_PENETRATION_DEPTH_H
-
+#endif  // BT_GJP_EPA_PENETRATION_DEPTH_H
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkPairDetector.cpp b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkPairDetector.cpp
index a0b825f0e80..803f6e06715 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkPairDetector.cpp
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkPairDetector.cpp
@@ -18,75 +18,69 @@ subject to the following restrictions:
 #include "BulletCollision/NarrowPhaseCollision/btSimplexSolverInterface.h"
 #include "BulletCollision/NarrowPhaseCollision/btConvexPenetrationDepthSolver.h"
 
-
-
-#if defined(DEBUG) || defined (_DEBUG)
+#if defined(DEBUG) || defined(_DEBUG)
 //#define TEST_NON_VIRTUAL 1
-#include <stdio.h> //for debug printf
+#include <stdio.h>  //for debug printf
 #ifdef __SPU__
 #include <spu_printf.h>
 #define printf spu_printf
-#endif //__SPU__
+#endif  //__SPU__
 #endif
 
 //must be above the machine epsilon
-#ifdef  BT_USE_DOUBLE_PRECISION
-	#define REL_ERROR2 btScalar(1.0e-12)
-	btScalar gGjkEpaPenetrationTolerance = 1.0e-12;
+#ifdef BT_USE_DOUBLE_PRECISION
+#define REL_ERROR2 btScalar(1.0e-12)
+btScalar gGjkEpaPenetrationTolerance = 1.0e-12;
 #else
-	#define REL_ERROR2 btScalar(1.0e-6)
-	btScalar gGjkEpaPenetrationTolerance = 0.001;
+#define REL_ERROR2 btScalar(1.0e-6)
+btScalar gGjkEpaPenetrationTolerance = 0.001;
 #endif
 
-//temp globals, to improve GJK/EPA/penetration calculations
-int gNumDeepPenetrationChecks = 0;
-int gNumGjkChecks = 0;
 
-
-btGjkPairDetector::btGjkPairDetector(const btConvexShape* objectA,const btConvexShape* objectB,btSimplexSolverInterface* simplexSolver,btConvexPenetrationDepthSolver*	penetrationDepthSolver)
-:m_cachedSeparatingAxis(btScalar(0.),btScalar(1.),btScalar(0.)),
-m_penetrationDepthSolver(penetrationDepthSolver),
-m_simplexSolver(simplexSolver),
-m_minkowskiA(objectA),
-m_minkowskiB(objectB),
-m_shapeTypeA(objectA->getShapeType()),
-m_shapeTypeB(objectB->getShapeType()),
-m_marginA(objectA->getMargin()),
-m_marginB(objectB->getMargin()),
-m_ignoreMargin(false),
-m_lastUsedMethod(-1),
-m_catchDegeneracies(1),
-m_fixContactNormalDirection(1)
+btGjkPairDetector::btGjkPairDetector(const btConvexShape *objectA, const btConvexShape *objectB, btSimplexSolverInterface *simplexSolver, btConvexPenetrationDepthSolver *penetrationDepthSolver)
+	: m_cachedSeparatingAxis(btScalar(0.), btScalar(1.), btScalar(0.)),
+	  m_penetrationDepthSolver(penetrationDepthSolver),
+	  m_simplexSolver(simplexSolver),
+	  m_minkowskiA(objectA),
+	  m_minkowskiB(objectB),
+	  m_shapeTypeA(objectA->getShapeType()),
+	  m_shapeTypeB(objectB->getShapeType()),
+	  m_marginA(objectA->getMargin()),
+	  m_marginB(objectB->getMargin()),
+	  m_ignoreMargin(false),
+	  m_lastUsedMethod(-1),
+	  m_catchDegeneracies(1),
+	  m_fixContactNormalDirection(1)
 {
 }
-btGjkPairDetector::btGjkPairDetector(const btConvexShape* objectA,const btConvexShape* objectB,int shapeTypeA,int shapeTypeB,btScalar marginA, btScalar marginB, btSimplexSolverInterface* simplexSolver,btConvexPenetrationDepthSolver*	penetrationDepthSolver)
-:m_cachedSeparatingAxis(btScalar(0.),btScalar(1.),btScalar(0.)),
-m_penetrationDepthSolver(penetrationDepthSolver),
-m_simplexSolver(simplexSolver),
-m_minkowskiA(objectA),
-m_minkowskiB(objectB),
-m_shapeTypeA(shapeTypeA),
-m_shapeTypeB(shapeTypeB),
-m_marginA(marginA),
-m_marginB(marginB),
-m_ignoreMargin(false),
-m_lastUsedMethod(-1),
-m_catchDegeneracies(1),
-m_fixContactNormalDirection(1)
+btGjkPairDetector::btGjkPairDetector(const btConvexShape *objectA, const btConvexShape *objectB, int shapeTypeA, int shapeTypeB, btScalar marginA, btScalar marginB, btSimplexSolverInterface *simplexSolver, btConvexPenetrationDepthSolver *penetrationDepthSolver)
+	: m_cachedSeparatingAxis(btScalar(0.), btScalar(1.), btScalar(0.)),
+	  m_penetrationDepthSolver(penetrationDepthSolver),
+	  m_simplexSolver(simplexSolver),
+	  m_minkowskiA(objectA),
+	  m_minkowskiB(objectB),
+	  m_shapeTypeA(shapeTypeA),
+	  m_shapeTypeB(shapeTypeB),
+	  m_marginA(marginA),
+	  m_marginB(marginB),
+	  m_ignoreMargin(false),
+	  m_lastUsedMethod(-1),
+	  m_catchDegeneracies(1),
+	  m_fixContactNormalDirection(1)
 {
 }
 
-void	btGjkPairDetector::getClosestPoints(const ClosestPointInput& input,Result& output,class btIDebugDraw* debugDraw,bool swapResults)
+void btGjkPairDetector::getClosestPoints(const ClosestPointInput &input, Result &output, class btIDebugDraw *debugDraw, bool swapResults)
 {
 	(void)swapResults;
 
-	getClosestPointsNonVirtual(input,output,debugDraw);
+	getClosestPointsNonVirtual(input, output, debugDraw);
 }
 
-static void btComputeSupport(const btConvexShape* convexA, const btTransform& localTransA, const btConvexShape* convexB, const btTransform& localTransB, const btVector3& dir, bool check2d, btVector3& supAworld, btVector3& supBworld, btVector3& aMinb)
+static void btComputeSupport(const btConvexShape *convexA, const btTransform &localTransA, const btConvexShape *convexB, const btTransform &localTransB, const btVector3 &dir, bool check2d, btVector3 &supAworld, btVector3 &supBworld, btVector3 &aMinb)
 {
-	btVector3 seperatingAxisInA = (dir)* localTransA.getBasis();
-	btVector3 seperatingAxisInB = (-dir)* localTransB.getBasis();
+	btVector3 seperatingAxisInA = (dir)*localTransA.getBasis();
+	btVector3 seperatingAxisInB = (-dir) * localTransB.getBasis();
 
 	btVector3 pInANoMargin = convexA->localGetSupportVertexWithoutMarginNonVirtual(seperatingAxisInA);
 	btVector3 qInBNoMargin = convexB->localGetSupportVertexWithoutMarginNonVirtual(seperatingAxisInB);
@@ -106,22 +100,21 @@ static void btComputeSupport(const btConvexShape* convexA, const btTransform& lo
 	aMinb = supAworld - supBworld;
 }
 
-struct btSupportVector 
+struct btSupportVector
 {
-	btVector3 v;  //!< Support point in minkowski sum
-	btVector3 v1; //!< Support point in obj1
-	btVector3 v2; //!< Support point in obj2
+	btVector3 v;   //!< Support point in minkowski sum
+	btVector3 v1;  //!< Support point in obj1
+	btVector3 v2;  //!< Support point in obj2
 };
 
-struct btSimplex 
+struct btSimplex
 {
 	btSupportVector ps[4];
-	int last; //!< index of last added point
+	int last;  //!< index of last added point
 };
 
 static btVector3 ccd_vec3_origin(0, 0, 0);
 
-
 inline void btSimplexInit(btSimplex *s)
 {
 	s->last = -1;
@@ -142,19 +135,18 @@ inline void btSupportCopy(btSupportVector *d, const btSupportVector *s)
 	*d = *s;
 }
 
-inline  void btVec3Copy(btVector3 *v, const btVector3* w)
+inline void btVec3Copy(btVector3 *v, const btVector3 *w)
 {
 	*v = *w;
 }
 
-inline void ccdVec3Add(btVector3*v, const btVector3*w)
+inline void ccdVec3Add(btVector3 *v, const btVector3 *w)
 {
 	v->m_floats[0] += w->m_floats[0];
 	v->m_floats[1] += w->m_floats[1];
 	v->m_floats[2] += w->m_floats[2];
 }
 
-
 inline void ccdVec3Sub(btVector3 *v, const btVector3 *w)
 {
 	*v -= *w;
@@ -162,24 +154,22 @@ inline void ccdVec3Sub(btVector3 *v, const btVector3 *w)
 inline void btVec3Sub2(btVector3 *d, const btVector3 *v, const btVector3 *w)
 {
 	*d = (*v) - (*w);
-	
 }
-inline  btScalar btVec3Dot(const btVector3 *a, const btVector3 *b)
+inline btScalar btVec3Dot(const btVector3 *a, const btVector3 *b)
 {
 	btScalar dot;
 	dot = a->dot(*b);
-	
+
 	return dot;
 }
 
-inline btScalar ccdVec3Dist2(const btVector3 *a, const btVector3*b)
+inline btScalar ccdVec3Dist2(const btVector3 *a, const btVector3 *b)
 {
 	btVector3 ab;
 	btVec3Sub2(&ab, a, b);
 	return btVec3Dot(&ab, &ab);
 }
 
-
 inline void btVec3Scale(btVector3 *d, btScalar k)
 {
 	d->m_floats[0] *= k;
@@ -195,7 +185,7 @@ inline void btVec3Cross(btVector3 *d, const btVector3 *a, const btVector3 *b)
 }
 
 inline void btTripleCross(const btVector3 *a, const btVector3 *b,
-	const btVector3 *c, btVector3 *d)
+						  const btVector3 *c, btVector3 *d)
 {
 	btVector3 e;
 	btVec3Cross(&e, a, b);
@@ -213,36 +203,35 @@ inline int ccdEq(btScalar _a, btScalar _b)
 
 	a = btFabs(_a);
 	b = btFabs(_b);
-	if (b > a) {
+	if (b > a)
+	{
 		return ab < SIMD_EPSILON * b;
 	}
-	else {
+	else
+	{
 		return ab < SIMD_EPSILON * a;
 	}
 }
 
-btScalar ccdVec3X(const btVector3* v)
+btScalar ccdVec3X(const btVector3 *v)
 {
 	return v->x();
 }
 
-btScalar ccdVec3Y(const btVector3* v)
+btScalar ccdVec3Y(const btVector3 *v)
 {
 	return v->y();
 }
 
-btScalar ccdVec3Z(const btVector3* v)
+btScalar ccdVec3Z(const btVector3 *v)
 {
 	return v->z();
 }
 inline int btVec3Eq(const btVector3 *a, const btVector3 *b)
 {
-	return ccdEq(ccdVec3X(a), ccdVec3X(b))
-		&& ccdEq(ccdVec3Y(a), ccdVec3Y(b))
-		&& ccdEq(ccdVec3Z(a), ccdVec3Z(b));
+	return ccdEq(ccdVec3X(a), ccdVec3X(b)) && ccdEq(ccdVec3Y(a), ccdVec3Y(b)) && ccdEq(ccdVec3Z(a), ccdVec3Z(b));
 }
 
-
 inline void btSimplexAdd(btSimplex *s, const btSupportVector *v)
 {
 	// here is no check on boundaries in sake of speed
@@ -250,7 +239,6 @@ inline void btSimplexAdd(btSimplex *s, const btSupportVector *v)
 	btSupportCopy(s->ps + s->last, v);
 }
 
-
 inline void btSimplexSet(btSimplex *s, size_t pos, const btSupportVector *a)
 {
 	btSupportCopy(s->ps + pos, a);
@@ -268,27 +256,28 @@ inline const btSupportVector *ccdSimplexLast(const btSimplex *s)
 
 inline int ccdSign(btScalar val)
 {
-	if (btFuzzyZero(val)) {
+	if (btFuzzyZero(val))
+	{
 		return 0;
 	}
-	else if (val < btScalar(0)) {
+	else if (val < btScalar(0))
+	{
 		return -1;
 	}
 	return 1;
 }
 
-
 inline btScalar btVec3PointSegmentDist2(const btVector3 *P,
-	const btVector3 *x0,
-	const btVector3 *b,
-	btVector3 *witness)
+										const btVector3 *x0,
+										const btVector3 *b,
+										btVector3 *witness)
 {
 	// The computation comes from solving equation of segment:
 	//      S(t) = x0 + t.d
 	//          where - x0 is initial point of segment
 	//                - d is direction of segment from x0 (|d| > 0)
 	//                - t belongs to <0, 1> interval
-	// 
+	//
 	// Than, distance from a segment to some point P can be expressed:
 	//      D(t) = |x0 + t.d - P|^2
 	//          which is distance from any point on segment. Minimization
@@ -310,24 +299,29 @@ inline btScalar btVec3PointSegmentDist2(const btVector3 *P,
 	t = -btScalar(1.) * btVec3Dot(&a, &d);
 	t /= btVec3Dot(&d, &d);
 
-	if (t < btScalar(0) || btFuzzyZero(t)) {
+	if (t < btScalar(0) || btFuzzyZero(t))
+	{
 		dist = ccdVec3Dist2(x0, P);
 		if (witness)
 			btVec3Copy(witness, x0);
 	}
-	else if (t > btScalar(1) || ccdEq(t, btScalar(1))) {
+	else if (t > btScalar(1) || ccdEq(t, btScalar(1)))
+	{
 		dist = ccdVec3Dist2(b, P);
 		if (witness)
 			btVec3Copy(witness, b);
 	}
-	else {
-		if (witness) {
+	else
+	{
+		if (witness)
+		{
 			btVec3Copy(witness, &d);
 			btVec3Scale(witness, t);
 			ccdVec3Add(witness, x0);
 			dist = ccdVec3Dist2(witness, P);
 		}
-		else {
+		else
+		{
 			// recycling variables
 			btVec3Scale(&d, t);
 			ccdVec3Add(&d, &a);
@@ -338,11 +332,10 @@ inline btScalar btVec3PointSegmentDist2(const btVector3 *P,
 	return dist;
 }
 
-
 btScalar btVec3PointTriDist2(const btVector3 *P,
-	const btVector3 *x0, const btVector3 *B,
-	const btVector3 *C,
-	btVector3 *witness)
+							 const btVector3 *x0, const btVector3 *B,
+							 const btVector3 *C,
+							 btVector3 *witness)
 {
 	// Computation comes from analytic expression for triangle (x0, B, C)
 	//      T(s, t) = x0 + s.d1 + t.d2, where d1 = B - x0 and d2 = C - x0 and
@@ -372,13 +365,9 @@ btScalar btVec3PointTriDist2(const btVector3 *P,
 	s = (q * r - w * p) / (w * v - r * r);
 	t = (-s * r - q) / w;
 
-	if ((btFuzzyZero(s) || s > btScalar(0))
-		&& (ccdEq(s, btScalar(1)) || s < btScalar(1))
-		&& (btFuzzyZero(t) || t > btScalar(0))
-		&& (ccdEq(t, btScalar(1)) || t < btScalar(1))
-		&& (ccdEq(t + s, btScalar(1)) || t + s < btScalar(1))) {
-
-		if (witness) 
+	if ((btFuzzyZero(s) || s > btScalar(0)) && (ccdEq(s, btScalar(1)) || s < btScalar(1)) && (btFuzzyZero(t) || t > btScalar(0)) && (ccdEq(t, btScalar(1)) || t < btScalar(1)) && (ccdEq(t + s, btScalar(1)) || t + s < btScalar(1)))
+	{
+		if (witness)
 		{
 			btVec3Scale(&d1, s);
 			btVec3Scale(&d2, t);
@@ -388,7 +377,7 @@ btScalar btVec3PointTriDist2(const btVector3 *P,
 
 			dist = ccdVec3Dist2(witness, P);
 		}
-		else 
+		else
 		{
 			dist = s * s * v;
 			dist += t * t * w;
@@ -398,18 +387,21 @@ btScalar btVec3PointTriDist2(const btVector3 *P,
 			dist += u;
 		}
 	}
-	else {
+	else
+	{
 		dist = btVec3PointSegmentDist2(P, x0, B, witness);
 
 		dist2 = btVec3PointSegmentDist2(P, x0, C, &witness2);
-		if (dist2 < dist) {
+		if (dist2 < dist)
+		{
 			dist = dist2;
 			if (witness)
 				btVec3Copy(witness, &witness2);
 		}
 
 		dist2 = btVec3PointSegmentDist2(P, B, C, &witness2);
-		if (dist2 < dist) {
+		if (dist2 < dist)
+		{
 			dist = dist2;
 			if (witness)
 				btVec3Copy(witness, &witness2);
@@ -419,7 +411,6 @@ btScalar btVec3PointTriDist2(const btVector3 *P,
 	return dist;
 }
 
-
 static int btDoSimplex2(btSimplex *simplex, btVector3 *dir)
 {
 	const btSupportVector *A, *B;
@@ -441,18 +432,21 @@ static int btDoSimplex2(btSimplex *simplex, btVector3 *dir)
 
 	// check if origin doesn't lie on AB segment
 	btVec3Cross(&tmp, &AB, &AO);
-	if (btFuzzyZero(btVec3Dot(&tmp, &tmp)) && dot > btScalar(0)) {
+	if (btFuzzyZero(btVec3Dot(&tmp, &tmp)) && dot > btScalar(0))
+	{
 		return 1;
 	}
 
 	// check if origin is in area where AB segment is
-	if (btFuzzyZero(dot) || dot < btScalar(0)) {
+	if (btFuzzyZero(dot) || dot < btScalar(0))
+	{
 		// origin is in outside are of A
 		btSimplexSet(simplex, 0, A);
 		btSimplexSetSize(simplex, 1);
 		btVec3Copy(dir, &AO);
 	}
-	else {
+	else
+	{
 		// origin is in area where AB segment is
 
 		// keep simplex untouched and set direction to
@@ -463,8 +457,6 @@ static int btDoSimplex2(btSimplex *simplex, btVector3 *dir)
 	return 0;
 }
 
-
-
 static int btDoSimplex3(btSimplex *simplex, btVector3 *dir)
 {
 	const btSupportVector *A, *B, *C;
@@ -479,13 +471,15 @@ static int btDoSimplex3(btSimplex *simplex, btVector3 *dir)
 
 	// check touching contact
 	dist = btVec3PointTriDist2(&ccd_vec3_origin, &A->v, &B->v, &C->v, 0);
-	if (btFuzzyZero(dist)) {
+	if (btFuzzyZero(dist))
+	{
 		return 1;
 	}
 
 	// check if triangle is really triangle (has area > 0)
 	// if not simplex can't be expanded and thus no itersection is found
-	if (btVec3Eq(&A->v, &B->v) || btVec3Eq(&A->v, &C->v)) {
+	if (btVec3Eq(&A->v, &B->v) || btVec3Eq(&A->v, &C->v))
+	{
 		return -1;
 	}
 
@@ -500,54 +494,64 @@ static int btDoSimplex3(btSimplex *simplex, btVector3 *dir)
 
 	btVec3Cross(&tmp, &ABC, &AC);
 	dot = btVec3Dot(&tmp, &AO);
-	if (btFuzzyZero(dot) || dot > btScalar(0)) {
+	if (btFuzzyZero(dot) || dot > btScalar(0))
+	{
 		dot = btVec3Dot(&AC, &AO);
-		if (btFuzzyZero(dot) || dot > btScalar(0)) {
+		if (btFuzzyZero(dot) || dot > btScalar(0))
+		{
 			// C is already in place
 			btSimplexSet(simplex, 1, A);
 			btSimplexSetSize(simplex, 2);
 			btTripleCross(&AC, &AO, &AC, dir);
 		}
-		else {
-		
+		else
+		{
 			dot = btVec3Dot(&AB, &AO);
-			if (btFuzzyZero(dot) || dot > btScalar(0)) {
+			if (btFuzzyZero(dot) || dot > btScalar(0))
+			{
 				btSimplexSet(simplex, 0, B);
 				btSimplexSet(simplex, 1, A);
 				btSimplexSetSize(simplex, 2);
 				btTripleCross(&AB, &AO, &AB, dir);
 			}
-			else {
+			else
+			{
 				btSimplexSet(simplex, 0, A);
 				btSimplexSetSize(simplex, 1);
 				btVec3Copy(dir, &AO);
 			}
 		}
 	}
-	else {
+	else
+	{
 		btVec3Cross(&tmp, &AB, &ABC);
 		dot = btVec3Dot(&tmp, &AO);
-		if (btFuzzyZero(dot) || dot > btScalar(0)) 
+		if (btFuzzyZero(dot) || dot > btScalar(0))
 		{
 			dot = btVec3Dot(&AB, &AO);
-			if (btFuzzyZero(dot) || dot > btScalar(0)) {
+			if (btFuzzyZero(dot) || dot > btScalar(0))
+			{
 				btSimplexSet(simplex, 0, B);
 				btSimplexSet(simplex, 1, A);
 				btSimplexSetSize(simplex, 2);
 				btTripleCross(&AB, &AO, &AB, dir);
 			}
-			else {
+			else
+			{
 				btSimplexSet(simplex, 0, A);
 				btSimplexSetSize(simplex, 1);
 				btVec3Copy(dir, &AO);
 			}
 		}
-		else {
+		else
+		{
 			dot = btVec3Dot(&ABC, &AO);
-			if (btFuzzyZero(dot) || dot > btScalar(0)) {
+			if (btFuzzyZero(dot) || dot > btScalar(0))
+			{
 				btVec3Copy(dir, &ABC);
 			}
-			else {
+			else
+			{
 				btSupportVector tmp;
 				btSupportCopy(&tmp, C);
 				btSimplexSet(simplex, 0, B);
@@ -581,7 +585,8 @@ static int btDoSimplex4(btSimplex *simplex, btVector3 *dir)
 	// if it is not simplex can't be expanded and thus no intersection is
 	// found
 	dist = btVec3PointTriDist2(&A->v, &B->v, &C->v, &D->v, 0);
-	if (btFuzzyZero(dist)) {
+	if (btFuzzyZero(dist))
+	{
 		return -1;
 	}
 
@@ -622,12 +627,14 @@ static int btDoSimplex4(btSimplex *simplex, btVector3 *dir)
 	AC_O = ccdSign(btVec3Dot(&ADB, &AO)) == C_on_ADB;
 	AD_O = ccdSign(btVec3Dot(&ABC, &AO)) == D_on_ABC;
 
-	if (AB_O && AC_O && AD_O) {
+	if (AB_O && AC_O && AD_O)
+	{
 		// origin is in tetrahedron
 		return 1;
 		// rearrange simplex to triangle and call btDoSimplex3()
 	}
-	else if (!AB_O) {
+	else if (!AB_O)
+	{
 		// B is farthest from the origin among all of the tetrahedron's
 		// points, so remove it from the list and go on with the triangle
 		// case
@@ -636,14 +643,16 @@ static int btDoSimplex4(btSimplex *simplex, btVector3 *dir)
 		btSimplexSet(simplex, 2, A);
 		btSimplexSetSize(simplex, 3);
 	}
-	else if (!AC_O) {
+	else if (!AC_O)
+	{
 		// C is farthest
 		btSimplexSet(simplex, 1, D);
 		btSimplexSet(simplex, 0, B);
 		btSimplexSet(simplex, 2, A);
 		btSimplexSetSize(simplex, 3);
 	}
-	else { // (!AD_O)
+	else
+	{  // (!AD_O)
 		btSimplexSet(simplex, 0, C);
 		btSimplexSet(simplex, 1, B);
 		btSimplexSet(simplex, 2, A);
@@ -655,36 +664,39 @@ static int btDoSimplex4(btSimplex *simplex, btVector3 *dir)
 
 static int btDoSimplex(btSimplex *simplex, btVector3 *dir)
 {
-	if (btSimplexSize(simplex) == 2) {
+	if (btSimplexSize(simplex) == 2)
+	{
 		// simplex contains segment only one segment
 		return btDoSimplex2(simplex, dir);
 	}
-	else if (btSimplexSize(simplex) == 3) {
+	else if (btSimplexSize(simplex) == 3)
+	{
 		// simplex contains triangle
 		return btDoSimplex3(simplex, dir);
 	}
-	else { // btSimplexSize(simplex) == 4
-			// tetrahedron - this is the only shape which can encapsule origin
-			// so btDoSimplex4() also contains test on it
+	else
+	{  // btSimplexSize(simplex) == 4
+		// tetrahedron - this is the only shape which can encapsule origin
+		// so btDoSimplex4() also contains test on it
 		return btDoSimplex4(simplex, dir);
 	}
 }
 
 #ifdef __SPU__
-void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput& input,Result& output,class btIDebugDraw* debugDraw)
+void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput &input, Result &output, class btIDebugDraw *debugDraw)
 #else
-void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput& input, Result& output, class btIDebugDraw* debugDraw)
+void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput &input, Result &output, class btIDebugDraw *debugDraw)
 #endif
 {
 	m_cachedSeparatingDistance = 0.f;
 
-	btScalar distance=btScalar(0.);
-	btVector3	normalInB(btScalar(0.),btScalar(0.),btScalar(0.));
+	btScalar distance = btScalar(0.);
+	btVector3 normalInB(btScalar(0.), btScalar(0.), btScalar(0.));
 
-	btVector3 pointOnA,pointOnB;
-	btTransform	localTransA = input.m_transformA;
+	btVector3 pointOnA, pointOnB;
+	btTransform localTransA = input.m_transformA;
 	btTransform localTransB = input.m_transformB;
-	btVector3 positionOffset=(localTransA.getOrigin() + localTransB.getOrigin()) * btScalar(0.5);
+	btVector3 positionOffset = (localTransA.getOrigin() + localTransB.getOrigin()) * btScalar(0.5);
 	localTransA.getOrigin() -= positionOffset;
 	localTransB.getOrigin() -= positionOffset;
 
@@ -693,7 +705,6 @@ void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput& inpu
 	btScalar marginA = m_marginA;
 	btScalar marginB = m_marginB;
 
-	gNumGjkChecks++;
 
 	//for CCD we don't use margins
 	if (m_ignoreMargin)
@@ -703,19 +714,19 @@ void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput& inpu
 	}
 
 	m_curIter = 0;
-	int gGjkMaxIter = 1000;//this is to catch invalid input, perhaps check for #NaN?
-	m_cachedSeparatingAxis.setValue(0,1,0);
+	int gGjkMaxIter = 1000;  //this is to catch invalid input, perhaps check for #NaN?
+	m_cachedSeparatingAxis.setValue(0, 1, 0);
 
 	bool isValid = false;
 	bool checkSimplex = false;
 	bool checkPenetration = true;
 	m_degenerateSimplex = 0;
-	
+
 	m_lastUsedMethod = -1;
 	int status = -2;
 	btVector3 orgNormalInB(0, 0, 0);
 	btScalar margin = marginA + marginB;
-	
+
 	//we add a separate implementation to check if the convex shapes intersect
 	//See also "Real-time Collision Detection with Implicit Objects" by Leif Olvang
 	//Todo: integrate the simplex penetration check directly inside the Bullet btVoronoiSimplexSolver
@@ -726,22 +737,18 @@ void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput& inpu
 		btScalar squaredDistance = BT_LARGE_FLOAT;
 		btScalar delta = btScalar(0.);
 
-
-
-		
 		btSimplex simplex1;
-		btSimplex* simplex = &simplex1;
+		btSimplex *simplex = &simplex1;
 		btSimplexInit(simplex);
 
 		btVector3 dir(1, 0, 0);
 
 		{
-
 			btVector3 lastSupV;
 			btVector3 supAworld;
 			btVector3 supBworld;
 			btComputeSupport(m_minkowskiA, localTransA, m_minkowskiB, localTransB, dir, check2d, supAworld, supBworld, lastSupV);
-			
+
 			btSupportVector last;
 			last.v = lastSupV;
 			last.v1 = supAworld;
@@ -751,10 +758,8 @@ void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput& inpu
 
 			dir = -lastSupV;
 
-			
-			
 			// start iterations
-			for (int iterations = 0; iterations <gGjkMaxIter; iterations++)
+			for (int iterations = 0; iterations < gGjkMaxIter; iterations++)
 			{
 				// obtain support point
 				btComputeSupport(m_minkowskiA, localTransA, m_minkowskiB, localTransB, dir, check2d, supAworld, supBworld, lastSupV);
@@ -769,7 +774,7 @@ void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput& inpu
 					status = -1;
 					break;
 				}
-				
+
 				// add last support vector to simplex
 				last.v = lastSupV;
 				last.v1 = supAworld;
@@ -781,21 +786,21 @@ void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput& inpu
 				// intersect and 0 if algorithm should continue
 
 				btVector3 newDir;
-				int do_simplex_res = btDoSimplex(simplex, &dir); 
+				int do_simplex_res = btDoSimplex(simplex, &dir);
 
-				if (do_simplex_res == 1) 
+				if (do_simplex_res == 1)
 				{
-					status = 0; // intersection found
+					status = 0;  // intersection found
 					break;
 				}
-				else if (do_simplex_res == -1) 
+				else if (do_simplex_res == -1)
 				{
 					// intersection not found
 					status = -1;
 					break;
 				}
-				
-				if (btFuzzyZero(btVec3Dot(&dir, &dir))) 
+
+				if (btFuzzyZero(btVec3Dot(&dir, &dir)))
 				{
 					// intersection not found
 					status = -1;
@@ -815,7 +820,6 @@ void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput& inpu
 					break;
 				}
 			}
-
 		}
 
 		m_simplexSolver->reset();
@@ -825,27 +829,24 @@ void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput& inpu
 			//printf("Intersect!\n");
 		}
 
-		if (status==-1)
+		if (status == -1)
 		{
 			//printf("not intersect\n");
 		}
 		//printf("dir=%f,%f,%f\n",dir[0],dir[1],dir[2]);
 		if (1)
 		{
-			for (; ; )
-				//while (true)
+			for (;;)
+			//while (true)
 			{
-
-				btVector3 seperatingAxisInA = (-m_cachedSeparatingAxis)* localTransA.getBasis();
-				btVector3 seperatingAxisInB = m_cachedSeparatingAxis* localTransB.getBasis();
-
+				btVector3 seperatingAxisInA = (-m_cachedSeparatingAxis) * localTransA.getBasis();
+				btVector3 seperatingAxisInB = m_cachedSeparatingAxis * localTransB.getBasis();
 
 				btVector3 pInA = m_minkowskiA->localGetSupportVertexWithoutMarginNonVirtual(seperatingAxisInA);
 				btVector3 qInB = m_minkowskiB->localGetSupportVertexWithoutMarginNonVirtual(seperatingAxisInB);
 
-				btVector3  pWorld = localTransA(pInA);
-				btVector3  qWorld = localTransB(qInB);
-
+				btVector3 pWorld = localTransA(pInA);
+				btVector3 qWorld = localTransB(qInB);
 
 				if (check2d)
 				{
@@ -921,8 +922,7 @@ void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput& inpu
 					checkSimplex = false;
 					break;
 				}
-#endif //
-
+#endif  //
 
 				//redundant m_simplexSolver->compute_points(pointOnA, pointOnB);
 
@@ -938,33 +938,31 @@ void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput& inpu
 
 				m_cachedSeparatingAxis = newCachedSeparatingAxis;
 
-				//degeneracy, this is typically due to invalid/uninitialized worldtransforms for a btCollisionObject   
+				//degeneracy, this is typically due to invalid/uninitialized worldtransforms for a btCollisionObject
 				if (m_curIter++ > gGjkMaxIter)
 				{
-#if defined(DEBUG) || defined (_DEBUG)
+#if defined(DEBUG) || defined(_DEBUG)
 
 					printf("btGjkPairDetector maxIter exceeded:%i\n", m_curIter);
 					printf("sepAxis=(%f,%f,%f), squaredDistance = %f, shapeTypeA=%i,shapeTypeB=%i\n",
-						m_cachedSeparatingAxis.getX(),
-						m_cachedSeparatingAxis.getY(),
-						m_cachedSeparatingAxis.getZ(),
-						squaredDistance,
-						m_minkowskiA->getShapeType(),
-						m_minkowskiB->getShapeType());
+						   m_cachedSeparatingAxis.getX(),
+						   m_cachedSeparatingAxis.getY(),
+						   m_cachedSeparatingAxis.getZ(),
+						   squaredDistance,
+						   m_minkowskiA->getShapeType(),
+						   m_minkowskiB->getShapeType());
 
-#endif   
+#endif
 					break;
-
 				}
 
-
 				bool check = (!m_simplexSolver->fullSimplex());
 				//bool check = (!m_simplexSolver->fullSimplex() && squaredDistance > SIMD_EPSILON * m_simplexSolver->maxVertex());
 
 				if (!check)
 				{
 					//do we need this backup_closest here ?
-	//				m_simplexSolver->backup_closest(m_cachedSeparatingAxis);
+					//				m_simplexSolver->backup_closest(m_cachedSeparatingAxis);
 					m_degenerateSimplex = 13;
 					break;
 				}
@@ -972,20 +970,20 @@ void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput& inpu
 
 			if (checkSimplex)
 			{
-			m_simplexSolver->compute_points(pointOnA, pointOnB);
-			normalInB = m_cachedSeparatingAxis;
+				m_simplexSolver->compute_points(pointOnA, pointOnB);
+				normalInB = m_cachedSeparatingAxis;
+
+				btScalar lenSqr = m_cachedSeparatingAxis.length2();
 
-			btScalar lenSqr =m_cachedSeparatingAxis.length2();
-			
 				//valid normal
 				if (lenSqr < REL_ERROR2)
 				{
 					m_degenerateSimplex = 5;
 				}
-				if (lenSqr > SIMD_EPSILON*SIMD_EPSILON)
+				if (lenSqr > SIMD_EPSILON * SIMD_EPSILON)
 				{
 					btScalar rlen = btScalar(1.) / btSqrt(lenSqr);
-					normalInB *= rlen; //normalize
+					normalInB *= rlen;  //normalize
 
 					btScalar s = btSqrt(squaredDistance);
 
@@ -1005,13 +1003,11 @@ void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput& inpu
 			}
 		}
 
-
-		
-		bool catchDegeneratePenetrationCase = 
-			(m_catchDegeneracies && m_penetrationDepthSolver && m_degenerateSimplex && ((distance+margin) < gGjkEpaPenetrationTolerance));
+		bool catchDegeneratePenetrationCase =
+			(m_catchDegeneracies && m_penetrationDepthSolver && m_degenerateSimplex && ((distance + margin) < gGjkEpaPenetrationTolerance));
 
 		//if (checkPenetration && !isValid)
-		if ((checkPenetration && (!isValid || catchDegeneratePenetrationCase )) || (status == 0))
+		if ((checkPenetration && (!isValid || catchDegeneratePenetrationCase)) || (status == 0))
 		{
 			//penetration case
 
@@ -1019,19 +1015,16 @@ void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput& inpu
 			if (m_penetrationDepthSolver)
 			{
 				// Penetration depth case.
-				btVector3 tmpPointOnA,tmpPointOnB;
-				
-				gNumDeepPenetrationChecks++;
+				btVector3 tmpPointOnA, tmpPointOnB;
+
 				m_cachedSeparatingAxis.setZero();
 
-				bool isValid2 = m_penetrationDepthSolver->calcPenDepth( 
-					*m_simplexSolver, 
-					m_minkowskiA,m_minkowskiB,
-					localTransA,localTransB,
+				bool isValid2 = m_penetrationDepthSolver->calcPenDepth(
+					*m_simplexSolver,
+					m_minkowskiA, m_minkowskiB,
+					localTransA, localTransB,
 					m_cachedSeparatingAxis, tmpPointOnA, tmpPointOnB,
-					debugDraw
-					);
-
+					debugDraw);
 
 				if (m_cachedSeparatingAxis.length2())
 				{
@@ -1039,13 +1032,13 @@ void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput& inpu
 					{
 						btVector3 tmpNormalInB = tmpPointOnB - tmpPointOnA;
 						btScalar lenSqr = tmpNormalInB.length2();
-						if (lenSqr <= (SIMD_EPSILON*SIMD_EPSILON))
+						if (lenSqr <= (SIMD_EPSILON * SIMD_EPSILON))
 						{
 							tmpNormalInB = m_cachedSeparatingAxis;
 							lenSqr = m_cachedSeparatingAxis.length2();
 						}
 
-						if (lenSqr > (SIMD_EPSILON*SIMD_EPSILON))
+						if (lenSqr > (SIMD_EPSILON * SIMD_EPSILON))
 						{
 							tmpNormalInB /= btSqrt(lenSqr);
 							btScalar distance2 = -(tmpPointOnA - tmpPointOnB).length();
@@ -1058,7 +1051,6 @@ void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput& inpu
 								pointOnB = tmpPointOnB;
 								normalInB = tmpNormalInB;
 								isValid = true;
-
 							}
 							else
 							{
@@ -1079,7 +1071,6 @@ void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput& inpu
 						///thanks to Jacob.Langford for the reproduction case
 						///http://code.google.com/p/bullet/issues/detail?id=250
 
-
 						if (m_cachedSeparatingAxis.length2() > btScalar(0.))
 						{
 							btScalar distance2 = (tmpPointOnA - tmpPointOnB).length() - margin;
@@ -1103,109 +1094,90 @@ void btGjkPairDetector::getClosestPointsNonVirtual(const ClosestPointInput& inpu
 							}
 						}
 					}
-				} else
+				}
+				else
 				{
 					//printf("EPA didn't return a valid value\n");
 				}
-				
 			}
-
 		}
 	}
 
-	
-
-	if (isValid && ((distance < 0) || (distance*distance < input.m_maximumDistanceSquared)))
+	if (isValid && ((distance < 0) || (distance * distance < input.m_maximumDistanceSquared)))
 	{
-
 		m_cachedSeparatingAxis = normalInB;
 		m_cachedSeparatingDistance = distance;
 		if (1)
 		{
-		///todo: need to track down this EPA penetration solver degeneracy
-		///the penetration solver reports penetration but the contact normal
-		///connecting the contact points is pointing in the opposite direction
-		///until then, detect the issue and revert the normal
+			///todo: need to track down this EPA penetration solver degeneracy
+			///the penetration solver reports penetration but the contact normal
+			///connecting the contact points is pointing in the opposite direction
+			///until then, detect the issue and revert the normal
 
 			btScalar d2 = 0.f;
 			{
-				btVector3 seperatingAxisInA = (-orgNormalInB)* localTransA.getBasis();
-				btVector3 seperatingAxisInB = orgNormalInB* localTransB.getBasis();
-
+				btVector3 seperatingAxisInA = (-orgNormalInB) * localTransA.getBasis();
+				btVector3 seperatingAxisInB = orgNormalInB * localTransB.getBasis();
 
 				btVector3 pInA = m_minkowskiA->localGetSupportVertexWithoutMarginNonVirtual(seperatingAxisInA);
 				btVector3 qInB = m_minkowskiB->localGetSupportVertexWithoutMarginNonVirtual(seperatingAxisInB);
 
-				btVector3  pWorld = localTransA(pInA);
-				btVector3  qWorld = localTransB(qInB);
+				btVector3 pWorld = localTransA(pInA);
+				btVector3 qWorld = localTransB(qInB);
 				btVector3 w = pWorld - qWorld;
-				d2 = orgNormalInB.dot(w)- margin;
+				d2 = orgNormalInB.dot(w) - margin;
 			}
-			
-			btScalar d1=0;
+
+			btScalar d1 = 0;
 			{
-				
-				btVector3 seperatingAxisInA = (normalInB)* localTransA.getBasis();
-				btVector3 seperatingAxisInB = -normalInB* localTransB.getBasis();
-			
+				btVector3 seperatingAxisInA = (normalInB)*localTransA.getBasis();
+				btVector3 seperatingAxisInB = -normalInB * localTransB.getBasis();
 
 				btVector3 pInA = m_minkowskiA->localGetSupportVertexWithoutMarginNonVirtual(seperatingAxisInA);
 				btVector3 qInB = m_minkowskiB->localGetSupportVertexWithoutMarginNonVirtual(seperatingAxisInB);
 
-				btVector3  pWorld = localTransA(pInA);	
-				btVector3  qWorld = localTransB(qInB);
-				btVector3 w	= pWorld - qWorld;
-				d1 = (-normalInB).dot(w)- margin;
-
+				btVector3 pWorld = localTransA(pInA);
+				btVector3 qWorld = localTransB(qInB);
+				btVector3 w = pWorld - qWorld;
+				d1 = (-normalInB).dot(w) - margin;
 			}
 			btScalar d0 = 0.f;
 			{
-				btVector3 seperatingAxisInA = (-normalInB)* input.m_transformA.getBasis();
-				btVector3 seperatingAxisInB = normalInB* input.m_transformB.getBasis();
-			
+				btVector3 seperatingAxisInA = (-normalInB) * input.m_transformA.getBasis();
+				btVector3 seperatingAxisInB = normalInB * input.m_transformB.getBasis();
 
 				btVector3 pInA = m_minkowskiA->localGetSupportVertexWithoutMarginNonVirtual(seperatingAxisInA);
 				btVector3 qInB = m_minkowskiB->localGetSupportVertexWithoutMarginNonVirtual(seperatingAxisInB);
 
-				btVector3  pWorld = localTransA(pInA);	
-				btVector3  qWorld = localTransB(qInB);
-				btVector3 w	= pWorld - qWorld;
-				d0 = normalInB.dot(w)-margin;
+				btVector3 pWorld = localTransA(pInA);
+				btVector3 qWorld = localTransB(qInB);
+				btVector3 w = pWorld - qWorld;
+				d0 = normalInB.dot(w) - margin;
 			}
-			
-			if (d1>d0)
+
+			if (d1 > d0)
 			{
 				m_lastUsedMethod = 10;
-				normalInB*=-1;
-			} 
+				normalInB *= -1;
+			}
 
 			if (orgNormalInB.length2())
 			{
 				if (d2 > d0 && d2 > d1 && d2 > distance)
 				{
-
 					normalInB = orgNormalInB;
 					distance = d2;
 				}
 			}
 		}
 
-
 		output.addContactPoint(
 			normalInB,
-			pointOnB+positionOffset,
+			pointOnB + positionOffset,
 			distance);
-
 	}
 	else
 	{
 		//printf("invalid gjk query\n");
 	}
-
-
 }
-
-
-
-
-
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkPairDetector.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkPairDetector.h
index feeae686213..659b63551e2 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkPairDetector.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btGjkPairDetector.h
@@ -13,9 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
-
 #ifndef BT_GJK_PAIR_DETECTOR_H
 #define BT_GJK_PAIR_DETECTOR_H
 
@@ -29,39 +26,34 @@ class btConvexPenetrationDepthSolver;
 /// btGjkPairDetector uses GJK to implement the btDiscreteCollisionDetectorInterface
 class btGjkPairDetector : public btDiscreteCollisionDetectorInterface
 {
-	
-
-	btVector3	m_cachedSeparatingAxis;
-	btConvexPenetrationDepthSolver*	m_penetrationDepthSolver;
+	btVector3 m_cachedSeparatingAxis;
+	btConvexPenetrationDepthSolver* m_penetrationDepthSolver;
 	btSimplexSolverInterface* m_simplexSolver;
 	const btConvexShape* m_minkowskiA;
 	const btConvexShape* m_minkowskiB;
-	int	m_shapeTypeA;
+	int m_shapeTypeA;
 	int m_shapeTypeB;
-	btScalar	m_marginA;
-	btScalar	m_marginB;
+	btScalar m_marginA;
+	btScalar m_marginB;
 
-	bool		m_ignoreMargin;
-	btScalar	m_cachedSeparatingDistance;
-	
+	bool m_ignoreMargin;
+	btScalar m_cachedSeparatingDistance;
 
 public:
-
 	//some debugging to fix degeneracy problems
-	int			m_lastUsedMethod;
-	int			m_curIter;
-	int			m_degenerateSimplex;
-	int			m_catchDegeneracies;
-	int			m_fixContactNormalDirection;
+	int m_lastUsedMethod;
+	int m_curIter;
+	int m_degenerateSimplex;
+	int m_catchDegeneracies;
+	int m_fixContactNormalDirection;
 
-	btGjkPairDetector(const btConvexShape* objectA,const btConvexShape* objectB,btSimplexSolverInterface* simplexSolver,btConvexPenetrationDepthSolver*	penetrationDepthSolver);
-	btGjkPairDetector(const btConvexShape* objectA,const btConvexShape* objectB,int shapeTypeA,int shapeTypeB,btScalar marginA, btScalar marginB, btSimplexSolverInterface* simplexSolver,btConvexPenetrationDepthSolver*	penetrationDepthSolver);
-	virtual ~btGjkPairDetector() {};
+	btGjkPairDetector(const btConvexShape* objectA, const btConvexShape* objectB, btSimplexSolverInterface* simplexSolver, btConvexPenetrationDepthSolver* penetrationDepthSolver);
+	btGjkPairDetector(const btConvexShape* objectA, const btConvexShape* objectB, int shapeTypeA, int shapeTypeB, btScalar marginA, btScalar marginB, btSimplexSolverInterface* simplexSolver, btConvexPenetrationDepthSolver* penetrationDepthSolver);
+	virtual ~btGjkPairDetector(){};
 
-	virtual void	getClosestPoints(const ClosestPointInput& input,Result& output,class btIDebugDraw* debugDraw,bool swapResults=false);
+	virtual void getClosestPoints(const ClosestPointInput& input, Result& output, class btIDebugDraw* debugDraw, bool swapResults = false);
 
-	void	getClosestPointsNonVirtual(const ClosestPointInput& input,Result& output,class btIDebugDraw* debugDraw);
-	
+	void getClosestPointsNonVirtual(const ClosestPointInput& input, Result& output, class btIDebugDraw* debugDraw);
 
 	void setMinkowskiA(const btConvexShape* minkA)
 	{
@@ -81,23 +73,21 @@ public:
 	{
 		return m_cachedSeparatingAxis;
 	}
-	btScalar	getCachedSeparatingDistance() const
+	btScalar getCachedSeparatingDistance() const
 	{
 		return m_cachedSeparatingDistance;
 	}
 
-	void	setPenetrationDepthSolver(btConvexPenetrationDepthSolver*	penetrationDepthSolver)
+	void setPenetrationDepthSolver(btConvexPenetrationDepthSolver* penetrationDepthSolver)
 	{
 		m_penetrationDepthSolver = penetrationDepthSolver;
 	}
 
 	///don't use setIgnoreMargin, it's for Bullet's internal use
-	void	setIgnoreMargin(bool ignoreMargin)
+	void setIgnoreMargin(bool ignoreMargin)
 	{
 		m_ignoreMargin = ignoreMargin;
 	}
-
-
 };
 
-#endif //BT_GJK_PAIR_DETECTOR_H
+#endif  //BT_GJK_PAIR_DETECTOR_H
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btManifoldPoint.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btManifoldPoint.h
index 571ad2c5f72..6d21f662024 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btManifoldPoint.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btManifoldPoint.h
@@ -20,161 +20,152 @@ subject to the following restrictions:
 #include "LinearMath/btTransformUtil.h"
 
 #ifdef PFX_USE_FREE_VECTORMATH
-	#include "physics_effects/base_level/solver/pfx_constraint_row.h"
+#include "physics_effects/base_level/solver/pfx_constraint_row.h"
 typedef sce::PhysicsEffects::PfxConstraintRow btConstraintRow;
 #else
-	// Don't change following order of parameters
-	ATTRIBUTE_ALIGNED16(struct) btConstraintRow {
-		btScalar m_normal[3];
-		btScalar m_rhs;
-		btScalar m_jacDiagInv;
-		btScalar m_lowerLimit;
-		btScalar m_upperLimit;
-		btScalar m_accumImpulse;
-	};
-	typedef btConstraintRow PfxConstraintRow;
-#endif //PFX_USE_FREE_VECTORMATH
+// Don't change following order of parameters
+ATTRIBUTE_ALIGNED16(struct)
+btConstraintRow
+{
+	btScalar m_normal[3];
+	btScalar m_rhs;
+	btScalar m_jacDiagInv;
+	btScalar m_lowerLimit;
+	btScalar m_upperLimit;
+	btScalar m_accumImpulse;
+};
+typedef btConstraintRow PfxConstraintRow;
+#endif  //PFX_USE_FREE_VECTORMATH
 
 enum btContactPointFlags
 {
-	BT_CONTACT_FLAG_LATERAL_FRICTION_INITIALIZED=1,
-	BT_CONTACT_FLAG_HAS_CONTACT_CFM=2,
-	BT_CONTACT_FLAG_HAS_CONTACT_ERP=4,
-    BT_CONTACT_FLAG_CONTACT_STIFFNESS_DAMPING = 8,
+	BT_CONTACT_FLAG_LATERAL_FRICTION_INITIALIZED = 1,
+	BT_CONTACT_FLAG_HAS_CONTACT_CFM = 2,
+	BT_CONTACT_FLAG_HAS_CONTACT_ERP = 4,
+	BT_CONTACT_FLAG_CONTACT_STIFFNESS_DAMPING = 8,
 	BT_CONTACT_FLAG_FRICTION_ANCHOR = 16,
 };
 
 /// ManifoldContactPoint collects and maintains persistent contactpoints.
 /// used to improve stability and performance of rigidbody dynamics response.
 class btManifoldPoint
+{
+public:
+	btManifoldPoint()
+		: m_userPersistentData(0),
+		  m_contactPointFlags(0),
+		  m_appliedImpulse(0.f),
+		  m_appliedImpulseLateral1(0.f),
+		  m_appliedImpulseLateral2(0.f),
+		  m_contactMotion1(0.f),
+		  m_contactMotion2(0.f),
+		  m_contactCFM(0.f),
+		  m_contactERP(0.f),
+		  m_frictionCFM(0.f),
+		  m_lifeTime(0)
 	{
-		public:
-			btManifoldPoint()
-				:m_userPersistentData(0),
-				m_contactPointFlags(0),
-				m_appliedImpulse(0.f),
-                m_appliedImpulseLateral1(0.f),
-				m_appliedImpulseLateral2(0.f),
-				m_contactMotion1(0.f),
-				m_contactMotion2(0.f),
-				m_contactCFM(0.f),
-				m_contactERP(0.f),
-				m_frictionCFM(0.f),
-				m_lifeTime(0)
-			{
-			}
+	}
 
-			btManifoldPoint( const btVector3 &pointA, const btVector3 &pointB, 
-					const btVector3 &normal, 
-					btScalar distance ) :
-					m_localPointA( pointA ), 
-					m_localPointB( pointB ), 
-					m_normalWorldOnB( normal ), 
-					m_distance1( distance ),
-					m_combinedFriction(btScalar(0.)),
-					m_combinedRollingFriction(btScalar(0.)),
-                    m_combinedSpinningFriction(btScalar(0.)),
-                    m_combinedRestitution(btScalar(0.)),
-					m_userPersistentData(0),
-					m_contactPointFlags(0),
-					m_appliedImpulse(0.f),
-                    m_appliedImpulseLateral1(0.f),
-					m_appliedImpulseLateral2(0.f),
-					m_contactMotion1(0.f),
-					m_contactMotion2(0.f),
-					m_contactCFM(0.f),
-					m_contactERP(0.f),
-					m_frictionCFM(0.f),
-					m_lifeTime(0)
-			{
-				
-			}
+	btManifoldPoint(const btVector3& pointA, const btVector3& pointB,
+					const btVector3& normal,
+					btScalar distance) : m_localPointA(pointA),
+										 m_localPointB(pointB),
+										 m_normalWorldOnB(normal),
+										 m_distance1(distance),
+										 m_combinedFriction(btScalar(0.)),
+										 m_combinedRollingFriction(btScalar(0.)),
+										 m_combinedSpinningFriction(btScalar(0.)),
+										 m_combinedRestitution(btScalar(0.)),
+										 m_userPersistentData(0),
+										 m_contactPointFlags(0),
+										 m_appliedImpulse(0.f),
+										 m_appliedImpulseLateral1(0.f),
+										 m_appliedImpulseLateral2(0.f),
+										 m_contactMotion1(0.f),
+										 m_contactMotion2(0.f),
+										 m_contactCFM(0.f),
+										 m_contactERP(0.f),
+										 m_frictionCFM(0.f),
+										 m_lifeTime(0)
+	{
+	}
 
-			
+	btVector3 m_localPointA;
+	btVector3 m_localPointB;
+	btVector3 m_positionWorldOnB;
+	///m_positionWorldOnA is redundant information, see getPositionWorldOnA(), but for clarity
+	btVector3 m_positionWorldOnA;
+	btVector3 m_normalWorldOnB;
 
-			btVector3 m_localPointA;			
-			btVector3 m_localPointB;			
-			btVector3	m_positionWorldOnB;
-			///m_positionWorldOnA is redundant information, see getPositionWorldOnA(), but for clarity
-			btVector3	m_positionWorldOnA;
-			btVector3 m_normalWorldOnB;
-		
-			btScalar	m_distance1;
-			btScalar	m_combinedFriction;
-			btScalar	m_combinedRollingFriction;//torsional friction orthogonal to contact normal, useful to make spheres stop rolling forever
-            btScalar	m_combinedSpinningFriction;//torsional friction around contact normal, useful for grasping objects
-            btScalar	m_combinedRestitution;
+	btScalar m_distance1;
+	btScalar m_combinedFriction;
+	btScalar m_combinedRollingFriction;   //torsional friction orthogonal to contact normal, useful to make spheres stop rolling forever
+	btScalar m_combinedSpinningFriction;  //torsional friction around contact normal, useful for grasping objects
+	btScalar m_combinedRestitution;
 
-			//BP mod, store contact triangles.
-			int			m_partId0;
-			int			m_partId1;
-			int			m_index0;
-			int			m_index1;
-				
-			mutable void*	m_userPersistentData;
-			//bool			m_lateralFrictionInitialized;
-			int				m_contactPointFlags;
-			
-			btScalar		m_appliedImpulse;
-			btScalar		m_appliedImpulseLateral1;
-			btScalar		m_appliedImpulseLateral2;
-			btScalar		m_contactMotion1;
-			btScalar		m_contactMotion2;
-			
-			union
-			{
-                btScalar		m_contactCFM;
-                btScalar        m_combinedContactStiffness1;
-			};
-			
-			union
-			{
-                btScalar		m_contactERP;
-                btScalar        m_combinedContactDamping1;
-			};
+	//BP mod, store contact triangles.
+	int m_partId0;
+	int m_partId1;
+	int m_index0;
+	int m_index1;
 
-			btScalar		m_frictionCFM;
+	mutable void* m_userPersistentData;
+	//bool			m_lateralFrictionInitialized;
+	int m_contactPointFlags;
 
-			int				m_lifeTime;//lifetime of the contactpoint in frames
-			
-			btVector3		m_lateralFrictionDir1;
-			btVector3		m_lateralFrictionDir2;
-
-
-
-
-			btScalar getDistance() const
-			{
-				return m_distance1;
-			}
-			int	getLifeTime() const
-			{
-				return m_lifeTime;
-			}
-
-			const btVector3& getPositionWorldOnA() const {
-				return m_positionWorldOnA;
-//				return m_positionWorldOnB + m_normalWorldOnB * m_distance1;
-			}
-
-			const btVector3& getPositionWorldOnB() const
-			{
-				return m_positionWorldOnB;
-			}
-
-			void	setDistance(btScalar dist)
-			{
-				m_distance1 = dist;
-			}
-			
-			///this returns the most recent applied impulse, to satisfy contact constraints by the constraint solver
-			btScalar	getAppliedImpulse() const
-			{
-				return m_appliedImpulse;
-			}
-
-			
+	btScalar m_appliedImpulse;
+	btScalar m_appliedImpulseLateral1;
+	btScalar m_appliedImpulseLateral2;
+	btScalar m_contactMotion1;
+	btScalar m_contactMotion2;
 
+	union {
+		btScalar m_contactCFM;
+		btScalar m_combinedContactStiffness1;
 	};
 
-#endif //BT_MANIFOLD_CONTACT_POINT_H
+	union {
+		btScalar m_contactERP;
+		btScalar m_combinedContactDamping1;
+	};
+
+	btScalar m_frictionCFM;
+
+	int m_lifeTime;  //lifetime of the contactpoint in frames
+
+	btVector3 m_lateralFrictionDir1;
+	btVector3 m_lateralFrictionDir2;
+
+	btScalar getDistance() const
+	{
+		return m_distance1;
+	}
+	int getLifeTime() const
+	{
+		return m_lifeTime;
+	}
+
+	const btVector3& getPositionWorldOnA() const
+	{
+		return m_positionWorldOnA;
+		//				return m_positionWorldOnB + m_normalWorldOnB * m_distance1;
+	}
+
+	const btVector3& getPositionWorldOnB() const
+	{
+		return m_positionWorldOnB;
+	}
+
+	void setDistance(btScalar dist)
+	{
+		m_distance1 = dist;
+	}
+
+	///this returns the most recent applied impulse, to satisfy contact constraints by the constraint solver
+	btScalar getAppliedImpulse() const
+	{
+		return m_appliedImpulse;
+	}
+};
+
+#endif  //BT_MANIFOLD_CONTACT_POINT_H
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btMinkowskiPenetrationDepthSolver.cpp b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btMinkowskiPenetrationDepthSolver.cpp
index fa45f49037e..a77449812fd 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btMinkowskiPenetrationDepthSolver.cpp
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btMinkowskiPenetrationDepthSolver.cpp
@@ -21,42 +21,38 @@ subject to the following restrictions:
 
 #define NUM_UNITSPHERE_POINTS 42
 
-
 bool btMinkowskiPenetrationDepthSolver::calcPenDepth(btSimplexSolverInterface& simplexSolver,
-												   const btConvexShape* convexA,const btConvexShape* convexB,
-												   const btTransform& transA,const btTransform& transB,
-												   btVector3& v, btVector3& pa, btVector3& pb,
-												   class btIDebugDraw* debugDraw
-												   )
+													 const btConvexShape* convexA, const btConvexShape* convexB,
+													 const btTransform& transA, const btTransform& transB,
+													 btVector3& v, btVector3& pa, btVector3& pb,
+													 class btIDebugDraw* debugDraw)
 {
-
 	(void)v;
-	
-	bool check2d= convexA->isConvex2d() && convexB->isConvex2d();
+
+	bool check2d = convexA->isConvex2d() && convexB->isConvex2d();
 
 	struct btIntermediateResult : public btDiscreteCollisionDetectorInterface::Result
 	{
-
-		btIntermediateResult():m_hasResult(false)
+		btIntermediateResult() : m_hasResult(false)
 		{
 		}
-		
+
 		btVector3 m_normalOnBInWorld;
 		btVector3 m_pointInWorld;
 		btScalar m_depth;
-		bool	m_hasResult;
+		bool m_hasResult;
 
-		virtual void setShapeIdentifiersA(int partId0,int index0)
+		virtual void setShapeIdentifiersA(int partId0, int index0)
 		{
 			(void)partId0;
 			(void)index0;
 		}
-		virtual void setShapeIdentifiersB(int partId1,int index1)
+		virtual void setShapeIdentifiersB(int partId1, int index1)
 		{
 			(void)partId1;
 			(void)index1;
 		}
-		void addContactPoint(const btVector3& normalOnBInWorld,const btVector3& pointInWorld,btScalar depth)
+		void addContactPoint(const btVector3& normalOnBInWorld, const btVector3& pointInWorld, btScalar depth)
 		{
 			m_normalOnBInWorld = normalOnBInWorld;
 			m_pointInWorld = pointInWorld;
@@ -68,39 +64,39 @@ bool btMinkowskiPenetrationDepthSolver::calcPenDepth(btSimplexSolverInterface& s
 	//just take fixed number of orientation, and sample the penetration depth in that direction
 	btScalar minProj = btScalar(BT_LARGE_FLOAT);
 	btVector3 minNorm(btScalar(0.), btScalar(0.), btScalar(0.));
-	btVector3 minA,minB;
-	btVector3 seperatingAxisInA,seperatingAxisInB;
-	btVector3 pInA,qInB,pWorld,qWorld,w;
+	btVector3 minA, minB;
+	btVector3 seperatingAxisInA, seperatingAxisInB;
+	btVector3 pInA, qInB, pWorld, qWorld, w;
 
 #ifndef __SPU__
 #define USE_BATCHED_SUPPORT 1
 #endif
 #ifdef USE_BATCHED_SUPPORT
 
-	btVector3	supportVerticesABatch[NUM_UNITSPHERE_POINTS+MAX_PREFERRED_PENETRATION_DIRECTIONS*2];
-	btVector3	supportVerticesBBatch[NUM_UNITSPHERE_POINTS+MAX_PREFERRED_PENETRATION_DIRECTIONS*2];
-	btVector3	seperatingAxisInABatch[NUM_UNITSPHERE_POINTS+MAX_PREFERRED_PENETRATION_DIRECTIONS*2];
-	btVector3	seperatingAxisInBBatch[NUM_UNITSPHERE_POINTS+MAX_PREFERRED_PENETRATION_DIRECTIONS*2];
+	btVector3 supportVerticesABatch[NUM_UNITSPHERE_POINTS + MAX_PREFERRED_PENETRATION_DIRECTIONS * 2];
+	btVector3 supportVerticesBBatch[NUM_UNITSPHERE_POINTS + MAX_PREFERRED_PENETRATION_DIRECTIONS * 2];
+	btVector3 seperatingAxisInABatch[NUM_UNITSPHERE_POINTS + MAX_PREFERRED_PENETRATION_DIRECTIONS * 2];
+	btVector3 seperatingAxisInBBatch[NUM_UNITSPHERE_POINTS + MAX_PREFERRED_PENETRATION_DIRECTIONS * 2];
 	int i;
 
 	int numSampleDirections = NUM_UNITSPHERE_POINTS;
 
-	for (i=0;i<numSampleDirections;i++)
+	for (i = 0; i < numSampleDirections; i++)
 	{
 		btVector3 norm = getPenetrationDirections()[i];
-		seperatingAxisInABatch[i] =  (-norm) * transA.getBasis() ;
-		seperatingAxisInBBatch[i] =  norm   * transB.getBasis() ;
+		seperatingAxisInABatch[i] = (-norm) * transA.getBasis();
+		seperatingAxisInBBatch[i] = norm * transB.getBasis();
 	}
 
 	{
 		int numPDA = convexA->getNumPreferredPenetrationDirections();
 		if (numPDA)
 		{
-			for (int i=0;i<numPDA;i++)
+			for (int i = 0; i < numPDA; i++)
 			{
 				btVector3 norm;
-				convexA->getPreferredPenetrationDirection(i,norm);
-				norm  = transA.getBasis() * norm;
+				convexA->getPreferredPenetrationDirection(i, norm);
+				norm = transA.getBasis() * norm;
 				getPenetrationDirections()[numSampleDirections] = norm;
 				seperatingAxisInABatch[numSampleDirections] = (-norm) * transA.getBasis();
 				seperatingAxisInBBatch[numSampleDirections] = norm * transB.getBasis();
@@ -113,11 +109,11 @@ bool btMinkowskiPenetrationDepthSolver::calcPenDepth(btSimplexSolverInterface& s
 		int numPDB = convexB->getNumPreferredPenetrationDirections();
 		if (numPDB)
 		{
-			for (int i=0;i<numPDB;i++)
+			for (int i = 0; i < numPDB; i++)
 			{
 				btVector3 norm;
-				convexB->getPreferredPenetrationDirection(i,norm);
-				norm  = transB.getBasis() * norm;
+				convexB->getPreferredPenetrationDirection(i, norm);
+				norm = transB.getBasis() * norm;
 				getPenetrationDirections()[numSampleDirections] = norm;
 				seperatingAxisInABatch[numSampleDirections] = (-norm) * transA.getBasis();
 				seperatingAxisInBBatch[numSampleDirections] = norm * transB.getBasis();
@@ -126,29 +122,25 @@ bool btMinkowskiPenetrationDepthSolver::calcPenDepth(btSimplexSolverInterface& s
 		}
 	}
 
+	convexA->batchedUnitVectorGetSupportingVertexWithoutMargin(seperatingAxisInABatch, supportVerticesABatch, numSampleDirections);
+	convexB->batchedUnitVectorGetSupportingVertexWithoutMargin(seperatingAxisInBBatch, supportVerticesBBatch, numSampleDirections);
 
-
-
-	convexA->batchedUnitVectorGetSupportingVertexWithoutMargin(seperatingAxisInABatch,supportVerticesABatch,numSampleDirections);
-	convexB->batchedUnitVectorGetSupportingVertexWithoutMargin(seperatingAxisInBBatch,supportVerticesBBatch,numSampleDirections);
-
-	for (i=0;i<numSampleDirections;i++)
+	for (i = 0; i < numSampleDirections; i++)
 	{
 		btVector3 norm = getPenetrationDirections()[i];
 		if (check2d)
 		{
 			norm[2] = 0.f;
 		}
-		if (norm.length2()>0.01)
+		if (norm.length2() > 0.01)
 		{
-
 			seperatingAxisInA = seperatingAxisInABatch[i];
 			seperatingAxisInB = seperatingAxisInBBatch[i];
 
 			pInA = supportVerticesABatch[i];
 			qInB = supportVerticesBBatch[i];
 
-			pWorld = transA(pInA);	
+			pWorld = transA(pInA);
 			qWorld = transB(qInB);
 			if (check2d)
 			{
@@ -156,7 +148,7 @@ bool btMinkowskiPenetrationDepthSolver::calcPenDepth(btSimplexSolverInterface& s
 				qWorld[2] = 0.f;
 			}
 
-			w	= qWorld - pWorld;
+			w = qWorld - pWorld;
 			btScalar delta = norm.dot(w);
 			//find smallest delta
 			if (delta < minProj)
@@ -167,7 +159,7 @@ bool btMinkowskiPenetrationDepthSolver::calcPenDepth(btSimplexSolverInterface& s
 				minB = qWorld;
 			}
 		}
-	}	
+	}
 #else
 
 	int numSampleDirections = NUM_UNITSPHERE_POINTS;
@@ -177,11 +169,11 @@ bool btMinkowskiPenetrationDepthSolver::calcPenDepth(btSimplexSolverInterface& s
 		int numPDA = convexA->getNumPreferredPenetrationDirections();
 		if (numPDA)
 		{
-			for (int i=0;i<numPDA;i++)
+			for (int i = 0; i < numPDA; i++)
 			{
 				btVector3 norm;
-				convexA->getPreferredPenetrationDirection(i,norm);
-				norm  = transA.getBasis() * norm;
+				convexA->getPreferredPenetrationDirection(i, norm);
+				norm = transA.getBasis() * norm;
 				getPenetrationDirections()[numSampleDirections] = norm;
 				numSampleDirections++;
 			}
@@ -192,28 +184,28 @@ bool btMinkowskiPenetrationDepthSolver::calcPenDepth(btSimplexSolverInterface& s
 		int numPDB = convexB->getNumPreferredPenetrationDirections();
 		if (numPDB)
 		{
-			for (int i=0;i<numPDB;i++)
+			for (int i = 0; i < numPDB; i++)
 			{
 				btVector3 norm;
-				convexB->getPreferredPenetrationDirection(i,norm);
-				norm  = transB.getBasis() * norm;
+				convexB->getPreferredPenetrationDirection(i, norm);
+				norm = transB.getBasis() * norm;
 				getPenetrationDirections()[numSampleDirections] = norm;
 				numSampleDirections++;
 			}
 		}
 	}
-#endif // __SPU__
+#endif  // __SPU__
 
-	for (int i=0;i<numSampleDirections;i++)
+	for (int i = 0; i < numSampleDirections; i++)
 	{
 		const btVector3& norm = getPenetrationDirections()[i];
-		seperatingAxisInA = (-norm)* transA.getBasis();
-		seperatingAxisInB = norm* transB.getBasis();
+		seperatingAxisInA = (-norm) * transA.getBasis();
+		seperatingAxisInB = norm * transB.getBasis();
 		pInA = convexA->localGetSupportVertexWithoutMarginNonVirtual(seperatingAxisInA);
 		qInB = convexB->localGetSupportVertexWithoutMarginNonVirtual(seperatingAxisInB);
-		pWorld = transA(pInA);	
+		pWorld = transA(pInA);
 		qWorld = transB(qInB);
-		w	= qWorld - pWorld;
+		w = qWorld - pWorld;
 		btScalar delta = norm.dot(w);
 		//find smallest delta
 		if (delta < minProj)
@@ -224,48 +216,39 @@ bool btMinkowskiPenetrationDepthSolver::calcPenDepth(btSimplexSolverInterface& s
 			minB = qWorld;
 		}
 	}
-#endif //USE_BATCHED_SUPPORT
+#endif  //USE_BATCHED_SUPPORT
 
 	//add the margins
 
-	minA += minNorm*convexA->getMarginNonVirtual();
-	minB -= minNorm*convexB->getMarginNonVirtual();
+	minA += minNorm * convexA->getMarginNonVirtual();
+	minB -= minNorm * convexB->getMarginNonVirtual();
 	//no penetration
 	if (minProj < btScalar(0.))
 		return false;
 
-	btScalar extraSeparation = 0.5f;///scale dependent
-	minProj += extraSeparation+(convexA->getMarginNonVirtual() + convexB->getMarginNonVirtual());
-
-
-
-
+	btScalar extraSeparation = 0.5f;  ///scale dependent
+	minProj += extraSeparation + (convexA->getMarginNonVirtual() + convexB->getMarginNonVirtual());
 
 //#define DEBUG_DRAW 1
 #ifdef DEBUG_DRAW
 	if (debugDraw)
 	{
-		btVector3 color(0,1,0);
-		debugDraw->drawLine(minA,minB,color);
-		color = btVector3 (1,1,1);
-		btVector3 vec = minB-minA;
+		btVector3 color(0, 1, 0);
+		debugDraw->drawLine(minA, minB, color);
+		color = btVector3(1, 1, 1);
+		btVector3 vec = minB - minA;
 		btScalar prj2 = minNorm.dot(vec);
-		debugDraw->drawLine(minA,minA+(minNorm*minProj),color);
-
+		debugDraw->drawLine(minA, minA + (minNorm * minProj), color);
 	}
-#endif //DEBUG_DRAW
+#endif  //DEBUG_DRAW
 
-	
-
-	btGjkPairDetector gjkdet(convexA,convexB,&simplexSolver,0);
+	btGjkPairDetector gjkdet(convexA, convexB, &simplexSolver, 0);
 
 	btScalar offsetDist = minProj;
 	btVector3 offset = minNorm * offsetDist;
-	
-
 
 	btGjkPairDetector::ClosestPointInput input;
-		
+
 	btVector3 newOrg = transA.getOrigin() + offset;
 
 	btTransform displacedTrans = transA;
@@ -273,89 +256,81 @@ bool btMinkowskiPenetrationDepthSolver::calcPenDepth(btSimplexSolverInterface& s
 
 	input.m_transformA = displacedTrans;
 	input.m_transformB = transB;
-	input.m_maximumDistanceSquared = btScalar(BT_LARGE_FLOAT);//minProj;
-	
+	input.m_maximumDistanceSquared = btScalar(BT_LARGE_FLOAT);  //minProj;
+
 	btIntermediateResult res;
 	gjkdet.setCachedSeperatingAxis(-minNorm);
-	gjkdet.getClosestPoints(input,res,debugDraw);
+	gjkdet.getClosestPoints(input, res, debugDraw);
 
 	btScalar correctedMinNorm = minProj - res.m_depth;
 
-
 	//the penetration depth is over-estimated, relax it
-	btScalar penetration_relaxation= btScalar(1.);
-	minNorm*=penetration_relaxation;
-	
+	btScalar penetration_relaxation = btScalar(1.);
+	minNorm *= penetration_relaxation;
 
 	if (res.m_hasResult)
 	{
-
 		pa = res.m_pointInWorld - minNorm * correctedMinNorm;
 		pb = res.m_pointInWorld;
 		v = minNorm;
-		
+
 #ifdef DEBUG_DRAW
 		if (debugDraw)
 		{
-			btVector3 color(1,0,0);
-			debugDraw->drawLine(pa,pb,color);
+			btVector3 color(1, 0, 0);
+			debugDraw->drawLine(pa, pb, color);
 		}
-#endif//DEBUG_DRAW
-
-
+#endif  //DEBUG_DRAW
 	}
 	return res.m_hasResult;
 }
 
-btVector3*	btMinkowskiPenetrationDepthSolver::getPenetrationDirections()
+btVector3* btMinkowskiPenetrationDepthSolver::getPenetrationDirections()
 {
-	static btVector3	sPenetrationDirections[NUM_UNITSPHERE_POINTS+MAX_PREFERRED_PENETRATION_DIRECTIONS*2] = 
-	{
-	btVector3(btScalar(0.000000) , btScalar(-0.000000),btScalar(-1.000000)),
-	btVector3(btScalar(0.723608) , btScalar(-0.525725),btScalar(-0.447219)),
-	btVector3(btScalar(-0.276388) , btScalar(-0.850649),btScalar(-0.447219)),
-	btVector3(btScalar(-0.894426) , btScalar(-0.000000),btScalar(-0.447216)),
-	btVector3(btScalar(-0.276388) , btScalar(0.850649),btScalar(-0.447220)),
-	btVector3(btScalar(0.723608) , btScalar(0.525725),btScalar(-0.447219)),
-	btVector3(btScalar(0.276388) , btScalar(-0.850649),btScalar(0.447220)),
-	btVector3(btScalar(-0.723608) , btScalar(-0.525725),btScalar(0.447219)),
-	btVector3(btScalar(-0.723608) , btScalar(0.525725),btScalar(0.447219)),
-	btVector3(btScalar(0.276388) , btScalar(0.850649),btScalar(0.447219)),
-	btVector3(btScalar(0.894426) , btScalar(0.000000),btScalar(0.447216)),
-	btVector3(btScalar(-0.000000) , btScalar(0.000000),btScalar(1.000000)),
-	btVector3(btScalar(0.425323) , btScalar(-0.309011),btScalar(-0.850654)),
-	btVector3(btScalar(-0.162456) , btScalar(-0.499995),btScalar(-0.850654)),
-	btVector3(btScalar(0.262869) , btScalar(-0.809012),btScalar(-0.525738)),
-	btVector3(btScalar(0.425323) , btScalar(0.309011),btScalar(-0.850654)),
-	btVector3(btScalar(0.850648) , btScalar(-0.000000),btScalar(-0.525736)),
-	btVector3(btScalar(-0.525730) , btScalar(-0.000000),btScalar(-0.850652)),
-	btVector3(btScalar(-0.688190) , btScalar(-0.499997),btScalar(-0.525736)),
-	btVector3(btScalar(-0.162456) , btScalar(0.499995),btScalar(-0.850654)),
-	btVector3(btScalar(-0.688190) , btScalar(0.499997),btScalar(-0.525736)),
-	btVector3(btScalar(0.262869) , btScalar(0.809012),btScalar(-0.525738)),
-	btVector3(btScalar(0.951058) , btScalar(0.309013),btScalar(0.000000)),
-	btVector3(btScalar(0.951058) , btScalar(-0.309013),btScalar(0.000000)),
-	btVector3(btScalar(0.587786) , btScalar(-0.809017),btScalar(0.000000)),
-	btVector3(btScalar(0.000000) , btScalar(-1.000000),btScalar(0.000000)),
-	btVector3(btScalar(-0.587786) , btScalar(-0.809017),btScalar(0.000000)),
-	btVector3(btScalar(-0.951058) , btScalar(-0.309013),btScalar(-0.000000)),
-	btVector3(btScalar(-0.951058) , btScalar(0.309013),btScalar(-0.000000)),
-	btVector3(btScalar(-0.587786) , btScalar(0.809017),btScalar(-0.000000)),
-	btVector3(btScalar(-0.000000) , btScalar(1.000000),btScalar(-0.000000)),
-	btVector3(btScalar(0.587786) , btScalar(0.809017),btScalar(-0.000000)),
-	btVector3(btScalar(0.688190) , btScalar(-0.499997),btScalar(0.525736)),
-	btVector3(btScalar(-0.262869) , btScalar(-0.809012),btScalar(0.525738)),
-	btVector3(btScalar(-0.850648) , btScalar(0.000000),btScalar(0.525736)),
-	btVector3(btScalar(-0.262869) , btScalar(0.809012),btScalar(0.525738)),
-	btVector3(btScalar(0.688190) , btScalar(0.499997),btScalar(0.525736)),
-	btVector3(btScalar(0.525730) , btScalar(0.000000),btScalar(0.850652)),
-	btVector3(btScalar(0.162456) , btScalar(-0.499995),btScalar(0.850654)),
-	btVector3(btScalar(-0.425323) , btScalar(-0.309011),btScalar(0.850654)),
-	btVector3(btScalar(-0.425323) , btScalar(0.309011),btScalar(0.850654)),
-	btVector3(btScalar(0.162456) , btScalar(0.499995),btScalar(0.850654))
-	};
+	static btVector3 sPenetrationDirections[NUM_UNITSPHERE_POINTS + MAX_PREFERRED_PENETRATION_DIRECTIONS * 2] =
+		{
+			btVector3(btScalar(0.000000), btScalar(-0.000000), btScalar(-1.000000)),
+			btVector3(btScalar(0.723608), btScalar(-0.525725), btScalar(-0.447219)),
+			btVector3(btScalar(-0.276388), btScalar(-0.850649), btScalar(-0.447219)),
+			btVector3(btScalar(-0.894426), btScalar(-0.000000), btScalar(-0.447216)),
+			btVector3(btScalar(-0.276388), btScalar(0.850649), btScalar(-0.447220)),
+			btVector3(btScalar(0.723608), btScalar(0.525725), btScalar(-0.447219)),
+			btVector3(btScalar(0.276388), btScalar(-0.850649), btScalar(0.447220)),
+			btVector3(btScalar(-0.723608), btScalar(-0.525725), btScalar(0.447219)),
+			btVector3(btScalar(-0.723608), btScalar(0.525725), btScalar(0.447219)),
+			btVector3(btScalar(0.276388), btScalar(0.850649), btScalar(0.447219)),
+			btVector3(btScalar(0.894426), btScalar(0.000000), btScalar(0.447216)),
+			btVector3(btScalar(-0.000000), btScalar(0.000000), btScalar(1.000000)),
+			btVector3(btScalar(0.425323), btScalar(-0.309011), btScalar(-0.850654)),
+			btVector3(btScalar(-0.162456), btScalar(-0.499995), btScalar(-0.850654)),
+			btVector3(btScalar(0.262869), btScalar(-0.809012), btScalar(-0.525738)),
+			btVector3(btScalar(0.425323), btScalar(0.309011), btScalar(-0.850654)),
+			btVector3(btScalar(0.850648), btScalar(-0.000000), btScalar(-0.525736)),
+			btVector3(btScalar(-0.525730), btScalar(-0.000000), btScalar(-0.850652)),
+			btVector3(btScalar(-0.688190), btScalar(-0.499997), btScalar(-0.525736)),
+			btVector3(btScalar(-0.162456), btScalar(0.499995), btScalar(-0.850654)),
+			btVector3(btScalar(-0.688190), btScalar(0.499997), btScalar(-0.525736)),
+			btVector3(btScalar(0.262869), btScalar(0.809012), btScalar(-0.525738)),
+			btVector3(btScalar(0.951058), btScalar(0.309013), btScalar(0.000000)),
+			btVector3(btScalar(0.951058), btScalar(-0.309013), btScalar(0.000000)),
+			btVector3(btScalar(0.587786), btScalar(-0.809017), btScalar(0.000000)),
+			btVector3(btScalar(0.000000), btScalar(-1.000000), btScalar(0.000000)),
+			btVector3(btScalar(-0.587786), btScalar(-0.809017), btScalar(0.000000)),
+			btVector3(btScalar(-0.951058), btScalar(-0.309013), btScalar(-0.000000)),
+			btVector3(btScalar(-0.951058), btScalar(0.309013), btScalar(-0.000000)),
+			btVector3(btScalar(-0.587786), btScalar(0.809017), btScalar(-0.000000)),
+			btVector3(btScalar(-0.000000), btScalar(1.000000), btScalar(-0.000000)),
+			btVector3(btScalar(0.587786), btScalar(0.809017), btScalar(-0.000000)),
+			btVector3(btScalar(0.688190), btScalar(-0.499997), btScalar(0.525736)),
+			btVector3(btScalar(-0.262869), btScalar(-0.809012), btScalar(0.525738)),
+			btVector3(btScalar(-0.850648), btScalar(0.000000), btScalar(0.525736)),
+			btVector3(btScalar(-0.262869), btScalar(0.809012), btScalar(0.525738)),
+			btVector3(btScalar(0.688190), btScalar(0.499997), btScalar(0.525736)),
+			btVector3(btScalar(0.525730), btScalar(0.000000), btScalar(0.850652)),
+			btVector3(btScalar(0.162456), btScalar(-0.499995), btScalar(0.850654)),
+			btVector3(btScalar(-0.425323), btScalar(-0.309011), btScalar(0.850654)),
+			btVector3(btScalar(-0.425323), btScalar(0.309011), btScalar(0.850654)),
+			btVector3(btScalar(0.162456), btScalar(0.499995), btScalar(0.850654))};
 
 	return sPenetrationDirections;
 }
-
-
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btMinkowskiPenetrationDepthSolver.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btMinkowskiPenetrationDepthSolver.h
index fd533b4fc33..8e3e3932591 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btMinkowskiPenetrationDepthSolver.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btMinkowskiPenetrationDepthSolver.h
@@ -23,18 +23,14 @@ subject to the following restrictions:
 class btMinkowskiPenetrationDepthSolver : public btConvexPenetrationDepthSolver
 {
 protected:
-
-	static btVector3*	getPenetrationDirections();
+	static btVector3* getPenetrationDirections();
 
 public:
-
-	virtual bool calcPenDepth( btSimplexSolverInterface& simplexSolver,
-	const btConvexShape* convexA,const btConvexShape* convexB,
-				const btTransform& transA,const btTransform& transB,
-			btVector3& v, btVector3& pa, btVector3& pb,
-			class btIDebugDraw* debugDraw
-			);
+	virtual bool calcPenDepth(btSimplexSolverInterface& simplexSolver,
+							  const btConvexShape* convexA, const btConvexShape* convexB,
+							  const btTransform& transA, const btTransform& transB,
+							  btVector3& v, btVector3& pa, btVector3& pb,
+							  class btIDebugDraw* debugDraw);
 };
 
-#endif //BT_MINKOWSKI_PENETRATION_DEPTH_SOLVER_H
-
+#endif  //BT_MINKOWSKI_PENETRATION_DEPTH_SOLVER_H
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btMprPenetration.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btMprPenetration.h
index a22a0bae66b..358bc95d815 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btMprPenetration.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btMprPenetration.h
@@ -28,32 +28,31 @@
 
 //#define MPR_AVERAGE_CONTACT_POSITIONS
 
-
 struct btMprCollisionDescription
 {
-    btVector3	m_firstDir;
-    int			m_maxGjkIterations;
-    btScalar	m_maximumDistanceSquared;
-    btScalar	m_gjkRelError2;
-   
-    btMprCollisionDescription()
-    :	m_firstDir(0,1,0),
-        m_maxGjkIterations(1000),
-        m_maximumDistanceSquared(1e30f),
-        m_gjkRelError2(1.0e-6)
-    {
-    }
-    virtual ~btMprCollisionDescription()
-    {
-    }
+	btVector3 m_firstDir;
+	int m_maxGjkIterations;
+	btScalar m_maximumDistanceSquared;
+	btScalar m_gjkRelError2;
+
+	btMprCollisionDescription()
+		: m_firstDir(0, 1, 0),
+		  m_maxGjkIterations(1000),
+		  m_maximumDistanceSquared(1e30f),
+		  m_gjkRelError2(1.0e-6)
+	{
+	}
+	virtual ~btMprCollisionDescription()
+	{
+	}
 };
 
 struct btMprDistanceInfo
 {
-    btVector3	m_pointOnA;
-    btVector3	m_pointOnB;
-    btVector3	m_normalBtoA;
-    btScalar	m_distance;
+	btVector3 m_pointOnA;
+	btVector3 m_pointOnB;
+	btVector3 m_normalBtoA;
+	btScalar m_distance;
 };
 
 #ifdef __cplusplus
@@ -67,131 +66,112 @@ struct btMprDistanceInfo
 #define BT_MPR_TOLERANCE 1E-6f
 #define BT_MPR_MAX_ITERATIONS 1000
 
-struct _btMprSupport_t 
+struct _btMprSupport_t
 {
-    btVector3 v;  //!< Support point in minkowski sum
-    btVector3 v1; //!< Support point in obj1
-    btVector3 v2; //!< Support point in obj2
+	btVector3 v;   //!< Support point in minkowski sum
+	btVector3 v1;  //!< Support point in obj1
+	btVector3 v2;  //!< Support point in obj2
 };
 typedef struct _btMprSupport_t btMprSupport_t;
 
-struct _btMprSimplex_t 
+struct _btMprSimplex_t
 {
-    btMprSupport_t ps[4];
-    int last; //!< index of last added point
+	btMprSupport_t ps[4];
+	int last;  //!< index of last added point
 };
 typedef struct _btMprSimplex_t btMprSimplex_t;
 
-inline btMprSupport_t* btMprSimplexPointW(btMprSimplex_t *s, int idx)
+inline btMprSupport_t *btMprSimplexPointW(btMprSimplex_t *s, int idx)
 {
-    return &s->ps[idx];
+	return &s->ps[idx];
 }
 
 inline void btMprSimplexSetSize(btMprSimplex_t *s, int size)
 {
-    s->last = size - 1;
+	s->last = size - 1;
 }
 
 #ifdef DEBUG_MPR
-inline void btPrintPortalVertex(_btMprSimplex_t* portal, int index)
+inline void btPrintPortalVertex(_btMprSimplex_t *portal, int index)
 {
-    printf("portal[%d].v = %f,%f,%f, v1=%f,%f,%f, v2=%f,%f,%f\n", index, portal->ps[index].v.x(),portal->ps[index].v.y(),portal->ps[index].v.z(),
-           portal->ps[index].v1.x(),portal->ps[index].v1.y(),portal->ps[index].v1.z(),
-           portal->ps[index].v2.x(),portal->ps[index].v2.y(),portal->ps[index].v2.z());
+	printf("portal[%d].v = %f,%f,%f, v1=%f,%f,%f, v2=%f,%f,%f\n", index, portal->ps[index].v.x(), portal->ps[index].v.y(), portal->ps[index].v.z(),
+		   portal->ps[index].v1.x(), portal->ps[index].v1.y(), portal->ps[index].v1.z(),
+		   portal->ps[index].v2.x(), portal->ps[index].v2.y(), portal->ps[index].v2.z());
 }
-#endif //DEBUG_MPR
-
-
-
+#endif  //DEBUG_MPR
 
 inline int btMprSimplexSize(const btMprSimplex_t *s)
 {
-    return s->last + 1;
+	return s->last + 1;
 }
 
-
-inline const btMprSupport_t* btMprSimplexPoint(const btMprSimplex_t* s, int idx)
+inline const btMprSupport_t *btMprSimplexPoint(const btMprSimplex_t *s, int idx)
 {
-    // here is no check on boundaries
-    return &s->ps[idx];
+	// here is no check on boundaries
+	return &s->ps[idx];
 }
 
 inline void btMprSupportCopy(btMprSupport_t *d, const btMprSupport_t *s)
 {
-    *d = *s;
+	*d = *s;
 }
 
 inline void btMprSimplexSet(btMprSimplex_t *s, size_t pos, const btMprSupport_t *a)
 {
-    btMprSupportCopy(s->ps + pos, a);
+	btMprSupportCopy(s->ps + pos, a);
 }
 
-
 inline void btMprSimplexSwap(btMprSimplex_t *s, size_t pos1, size_t pos2)
 {
-    btMprSupport_t supp;
+	btMprSupport_t supp;
 
-    btMprSupportCopy(&supp, &s->ps[pos1]);
-    btMprSupportCopy(&s->ps[pos1], &s->ps[pos2]);
-    btMprSupportCopy(&s->ps[pos2], &supp);
+	btMprSupportCopy(&supp, &s->ps[pos1]);
+	btMprSupportCopy(&s->ps[pos1], &s->ps[pos2]);
+	btMprSupportCopy(&s->ps[pos2], &supp);
 }
 
-
 inline int btMprIsZero(float val)
 {
-    return BT_MPR_FABS(val) < FLT_EPSILON;
+	return BT_MPR_FABS(val) < FLT_EPSILON;
 }
 
-
-
 inline int btMprEq(float _a, float _b)
 {
-    float ab;
-    float a, b;
+	float ab;
+	float a, b;
 
-    ab = BT_MPR_FABS(_a - _b);
-    if (BT_MPR_FABS(ab) < FLT_EPSILON)
-        return 1;
+	ab = BT_MPR_FABS(_a - _b);
+	if (BT_MPR_FABS(ab) < FLT_EPSILON)
+		return 1;
 
-    a = BT_MPR_FABS(_a);
-    b = BT_MPR_FABS(_b);
-    if (b > a){
-        return ab < FLT_EPSILON * b;
-    }else{
-        return ab < FLT_EPSILON * a;
-    }
+	a = BT_MPR_FABS(_a);
+	b = BT_MPR_FABS(_b);
+	if (b > a)
+	{
+		return ab < FLT_EPSILON * b;
+	}
+	else
+	{
+		return ab < FLT_EPSILON * a;
+	}
 }
 
-
-inline int btMprVec3Eq(const btVector3* a, const btVector3 *b)
+inline int btMprVec3Eq(const btVector3 *a, const btVector3 *b)
 {
-    return btMprEq((*a).x(), (*b).x())
-            && btMprEq((*a).y(), (*b).y())
-            && btMprEq((*a).z(), (*b).z());
+	return btMprEq((*a).x(), (*b).x()) && btMprEq((*a).y(), (*b).y()) && btMprEq((*a).z(), (*b).z());
 }
 
-
-
-
-
-
-
-
-
-
-
 template <typename btConvexTemplate>
-inline void btFindOrigin(const btConvexTemplate& a, const btConvexTemplate& b, const btMprCollisionDescription& colDesc,btMprSupport_t *center)
+inline void btFindOrigin(const btConvexTemplate &a, const btConvexTemplate &b, const btMprCollisionDescription &colDesc, btMprSupport_t *center)
 {
-
 	center->v1 = a.getObjectCenterInWorld();
-    center->v2 = b.getObjectCenterInWorld();
-    center->v = center->v1 - center->v2;
+	center->v2 = b.getObjectCenterInWorld();
+	center->v = center->v1 - center->v2;
 }
 
 inline void btMprVec3Set(btVector3 *v, float x, float y, float z)
 {
-	v->setValue(x,y,z);
+	v->setValue(x, y, z);
 }
 
 inline void btMprVec3Add(btVector3 *v, const btVector3 *w)
@@ -201,41 +181,38 @@ inline void btMprVec3Add(btVector3 *v, const btVector3 *w)
 
 inline void btMprVec3Copy(btVector3 *v, const btVector3 *w)
 {
-    *v = *w;
+	*v = *w;
 }
 
 inline void btMprVec3Scale(btVector3 *d, float k)
 {
-    *d *= k;
+	*d *= k;
 }
 
 inline float btMprVec3Dot(const btVector3 *a, const btVector3 *b)
 {
-    float dot;
+	float dot;
 
-	dot = btDot(*a,*b);
-    return dot;
+	dot = btDot(*a, *b);
+	return dot;
 }
 
-
 inline float btMprVec3Len2(const btVector3 *v)
 {
-    return btMprVec3Dot(v, v);
+	return btMprVec3Dot(v, v);
 }
 
 inline void btMprVec3Normalize(btVector3 *d)
 {
-    float k = 1.f / BT_MPR_SQRT(btMprVec3Len2(d));
-    btMprVec3Scale(d, k);
+	float k = 1.f / BT_MPR_SQRT(btMprVec3Len2(d));
+	btMprVec3Scale(d, k);
 }
 
 inline void btMprVec3Cross(btVector3 *d, const btVector3 *a, const btVector3 *b)
 {
-	*d = btCross(*a,*b);
-	
+	*d = btCross(*a, *b);
 }
 
-
 inline void btMprVec3Sub2(btVector3 *d, const btVector3 *v, const btVector3 *w)
 {
 	*d = *v - *w;
@@ -243,89 +220,97 @@ inline void btMprVec3Sub2(btVector3 *d, const btVector3 *v, const btVector3 *w)
 
 inline void btPortalDir(const btMprSimplex_t *portal, btVector3 *dir)
 {
-    btVector3 v2v1, v3v1;
+	btVector3 v2v1, v3v1;
 
-    btMprVec3Sub2(&v2v1, &btMprSimplexPoint(portal, 2)->v,
-                       &btMprSimplexPoint(portal, 1)->v);
-    btMprVec3Sub2(&v3v1, &btMprSimplexPoint(portal, 3)->v,
-                       &btMprSimplexPoint(portal, 1)->v);
-    btMprVec3Cross(dir, &v2v1, &v3v1);
-    btMprVec3Normalize(dir);
+	btMprVec3Sub2(&v2v1, &btMprSimplexPoint(portal, 2)->v,
+				  &btMprSimplexPoint(portal, 1)->v);
+	btMprVec3Sub2(&v3v1, &btMprSimplexPoint(portal, 3)->v,
+				  &btMprSimplexPoint(portal, 1)->v);
+	btMprVec3Cross(dir, &v2v1, &v3v1);
+	btMprVec3Normalize(dir);
 }
 
-
 inline int portalEncapsulesOrigin(const btMprSimplex_t *portal,
-                                       const btVector3 *dir)
+								  const btVector3 *dir)
 {
-    float dot;
-    dot = btMprVec3Dot(dir, &btMprSimplexPoint(portal, 1)->v);
-    return btMprIsZero(dot) || dot > 0.f;
+	float dot;
+	dot = btMprVec3Dot(dir, &btMprSimplexPoint(portal, 1)->v);
+	return btMprIsZero(dot) || dot > 0.f;
 }
 
 inline int portalReachTolerance(const btMprSimplex_t *portal,
-                                     const btMprSupport_t *v4,
-                                     const btVector3 *dir)
+								const btMprSupport_t *v4,
+								const btVector3 *dir)
 {
-    float dv1, dv2, dv3, dv4;
-    float dot1, dot2, dot3;
+	float dv1, dv2, dv3, dv4;
+	float dot1, dot2, dot3;
 
-    // find the smallest dot product of dir and {v1-v4, v2-v4, v3-v4}
+	// find the smallest dot product of dir and {v1-v4, v2-v4, v3-v4}
 
-    dv1 = btMprVec3Dot(&btMprSimplexPoint(portal, 1)->v, dir);
-    dv2 = btMprVec3Dot(&btMprSimplexPoint(portal, 2)->v, dir);
-    dv3 = btMprVec3Dot(&btMprSimplexPoint(portal, 3)->v, dir);
-    dv4 = btMprVec3Dot(&v4->v, dir);
+	dv1 = btMprVec3Dot(&btMprSimplexPoint(portal, 1)->v, dir);
+	dv2 = btMprVec3Dot(&btMprSimplexPoint(portal, 2)->v, dir);
+	dv3 = btMprVec3Dot(&btMprSimplexPoint(portal, 3)->v, dir);
+	dv4 = btMprVec3Dot(&v4->v, dir);
 
-    dot1 = dv4 - dv1;
-    dot2 = dv4 - dv2;
-    dot3 = dv4 - dv3;
+	dot1 = dv4 - dv1;
+	dot2 = dv4 - dv2;
+	dot3 = dv4 - dv3;
 
-    dot1 = BT_MPR_FMIN(dot1, dot2);
-    dot1 = BT_MPR_FMIN(dot1, dot3);
+	dot1 = BT_MPR_FMIN(dot1, dot2);
+	dot1 = BT_MPR_FMIN(dot1, dot3);
 
-    return btMprEq(dot1, BT_MPR_TOLERANCE) || dot1 < BT_MPR_TOLERANCE;
+	return btMprEq(dot1, BT_MPR_TOLERANCE) || dot1 < BT_MPR_TOLERANCE;
 }
 
 inline int portalCanEncapsuleOrigin(const btMprSimplex_t *portal,
-                                         const btMprSupport_t *v4,
-                                         const btVector3 *dir)
+									const btMprSupport_t *v4,
+									const btVector3 *dir)
 {
-    float dot;
-    dot = btMprVec3Dot(&v4->v, dir);
-    return btMprIsZero(dot) || dot > 0.f;
+	float dot;
+	dot = btMprVec3Dot(&v4->v, dir);
+	return btMprIsZero(dot) || dot > 0.f;
 }
 
 inline void btExpandPortal(btMprSimplex_t *portal,
-                              const btMprSupport_t *v4)
+						   const btMprSupport_t *v4)
 {
-    float dot;
-    btVector3 v4v0;
+	float dot;
+	btVector3 v4v0;
 
-    btMprVec3Cross(&v4v0, &v4->v, &btMprSimplexPoint(portal, 0)->v);
-    dot = btMprVec3Dot(&btMprSimplexPoint(portal, 1)->v, &v4v0);
-    if (dot > 0.f){
-        dot = btMprVec3Dot(&btMprSimplexPoint(portal, 2)->v, &v4v0);
-        if (dot > 0.f){
-            btMprSimplexSet(portal, 1, v4);
-        }else{
-            btMprSimplexSet(portal, 3, v4);
-        }
-    }else{
-        dot = btMprVec3Dot(&btMprSimplexPoint(portal, 3)->v, &v4v0);
-        if (dot > 0.f){
-            btMprSimplexSet(portal, 2, v4);
-        }else{
-            btMprSimplexSet(portal, 1, v4);
-        }
-    }
+	btMprVec3Cross(&v4v0, &v4->v, &btMprSimplexPoint(portal, 0)->v);
+	dot = btMprVec3Dot(&btMprSimplexPoint(portal, 1)->v, &v4v0);
+	if (dot > 0.f)
+	{
+		dot = btMprVec3Dot(&btMprSimplexPoint(portal, 2)->v, &v4v0);
+		if (dot > 0.f)
+		{
+			btMprSimplexSet(portal, 1, v4);
+		}
+		else
+		{
+			btMprSimplexSet(portal, 3, v4);
+		}
+	}
+	else
+	{
+		dot = btMprVec3Dot(&btMprSimplexPoint(portal, 3)->v, &v4v0);
+		if (dot > 0.f)
+		{
+			btMprSimplexSet(portal, 2, v4);
+		}
+		else
+		{
+			btMprSimplexSet(portal, 1, v4);
+		}
+	}
 }
 template <typename btConvexTemplate>
-inline void btMprSupport(const btConvexTemplate& a, const btConvexTemplate& b,
-                         const btMprCollisionDescription& colDesc,
-													const btVector3& dir, btMprSupport_t *supp)
+inline void btMprSupport(const btConvexTemplate &a, const btConvexTemplate &b,
+						 const btMprCollisionDescription &colDesc,
+						 const btVector3 &dir, btMprSupport_t *supp)
 {
-	btVector3 seperatingAxisInA = dir* a.getWorldTransform().getBasis();
-	btVector3 seperatingAxisInB = -dir* b.getWorldTransform().getBasis();
+	btVector3 seperatingAxisInA = dir * a.getWorldTransform().getBasis();
+	btVector3 seperatingAxisInB = -dir * b.getWorldTransform().getBasis();
 
 	btVector3 pInA = a.getLocalSupportWithMargin(seperatingAxisInA);
 	btVector3 qInB = b.getLocalSupportWithMargin(seperatingAxisInB);
@@ -335,574 +320,565 @@ inline void btMprSupport(const btConvexTemplate& a, const btConvexTemplate& b,
 	supp->v = supp->v1 - supp->v2;
 }
 
-
 template <typename btConvexTemplate>
-static int btDiscoverPortal(const btConvexTemplate& a, const btConvexTemplate& b,
-                            const btMprCollisionDescription& colDesc,
-													btMprSimplex_t *portal)
+static int btDiscoverPortal(const btConvexTemplate &a, const btConvexTemplate &b,
+							const btMprCollisionDescription &colDesc,
+							btMprSimplex_t *portal)
 {
-    btVector3 dir, va, vb;
-    float dot;
-    int cont;
-	
-	
+	btVector3 dir, va, vb;
+	float dot;
+	int cont;
 
-    // vertex 0 is center of portal
-    btFindOrigin(a,b,colDesc, btMprSimplexPointW(portal, 0));
-   
-    
-    // vertex 0 is center of portal
-    btMprSimplexSetSize(portal, 1);
-    
+	// vertex 0 is center of portal
+	btFindOrigin(a, b, colDesc, btMprSimplexPointW(portal, 0));
 
+	// vertex 0 is center of portal
+	btMprSimplexSetSize(portal, 1);
 
-	btVector3 zero = btVector3(0,0,0);
-	btVector3* org = &zero;
+	btVector3 zero = btVector3(0, 0, 0);
+	btVector3 *org = &zero;
 
-    if (btMprVec3Eq(&btMprSimplexPoint(portal, 0)->v, org)){
-        // Portal's center lies on origin (0,0,0) => we know that objects
-        // intersect but we would need to know penetration info.
-        // So move center little bit...
-        btMprVec3Set(&va, FLT_EPSILON * 10.f, 0.f, 0.f);
-        btMprVec3Add(&btMprSimplexPointW(portal, 0)->v, &va);
-    }
+	if (btMprVec3Eq(&btMprSimplexPoint(portal, 0)->v, org))
+	{
+		// Portal's center lies on origin (0,0,0) => we know that objects
+		// intersect but we would need to know penetration info.
+		// So move center little bit...
+		btMprVec3Set(&va, FLT_EPSILON * 10.f, 0.f, 0.f);
+		btMprVec3Add(&btMprSimplexPointW(portal, 0)->v, &va);
+	}
 
+	// vertex 1 = support in direction of origin
+	btMprVec3Copy(&dir, &btMprSimplexPoint(portal, 0)->v);
+	btMprVec3Scale(&dir, -1.f);
+	btMprVec3Normalize(&dir);
 
-    // vertex 1 = support in direction of origin
-    btMprVec3Copy(&dir, &btMprSimplexPoint(portal, 0)->v);
-    btMprVec3Scale(&dir, -1.f);
-    btMprVec3Normalize(&dir);
+	btMprSupport(a, b, colDesc, dir, btMprSimplexPointW(portal, 1));
 
+	btMprSimplexSetSize(portal, 2);
 
-    btMprSupport(a,b,colDesc, dir, btMprSimplexPointW(portal, 1));
- 
-    btMprSimplexSetSize(portal, 2);
+	// test if origin isn't outside of v1
+	dot = btMprVec3Dot(&btMprSimplexPoint(portal, 1)->v, &dir);
 
-    // test if origin isn't outside of v1
-    dot = btMprVec3Dot(&btMprSimplexPoint(portal, 1)->v, &dir);
-	
+	if (btMprIsZero(dot) || dot < 0.f)
+		return -1;
 
-    if (btMprIsZero(dot) || dot < 0.f)
-        return -1;
+	// vertex 2
+	btMprVec3Cross(&dir, &btMprSimplexPoint(portal, 0)->v,
+				   &btMprSimplexPoint(portal, 1)->v);
+	if (btMprIsZero(btMprVec3Len2(&dir)))
+	{
+		if (btMprVec3Eq(&btMprSimplexPoint(portal, 1)->v, org))
+		{
+			// origin lies on v1
+			return 1;
+		}
+		else
+		{
+			// origin lies on v0-v1 segment
+			return 2;
+		}
+	}
 
+	btMprVec3Normalize(&dir);
+	btMprSupport(a, b, colDesc, dir, btMprSimplexPointW(portal, 2));
 
-    // vertex 2
-    btMprVec3Cross(&dir, &btMprSimplexPoint(portal, 0)->v,
-                       &btMprSimplexPoint(portal, 1)->v);
-    if (btMprIsZero(btMprVec3Len2(&dir))){
-        if (btMprVec3Eq(&btMprSimplexPoint(portal, 1)->v, org)){
-            // origin lies on v1
-            return 1;
-        }else{
-            // origin lies on v0-v1 segment
-            return 2;
-        }
-    }
+	dot = btMprVec3Dot(&btMprSimplexPoint(portal, 2)->v, &dir);
+	if (btMprIsZero(dot) || dot < 0.f)
+		return -1;
 
-    btMprVec3Normalize(&dir);
-    btMprSupport(a,b,colDesc, dir, btMprSimplexPointW(portal, 2));
-    
-    
-    
-    dot = btMprVec3Dot(&btMprSimplexPoint(portal, 2)->v, &dir);
-    if (btMprIsZero(dot) || dot < 0.f)
-        return -1;
+	btMprSimplexSetSize(portal, 3);
 
-    btMprSimplexSetSize(portal, 3);
+	// vertex 3 direction
+	btMprVec3Sub2(&va, &btMprSimplexPoint(portal, 1)->v,
+				  &btMprSimplexPoint(portal, 0)->v);
+	btMprVec3Sub2(&vb, &btMprSimplexPoint(portal, 2)->v,
+				  &btMprSimplexPoint(portal, 0)->v);
+	btMprVec3Cross(&dir, &va, &vb);
+	btMprVec3Normalize(&dir);
 
-    // vertex 3 direction
-    btMprVec3Sub2(&va, &btMprSimplexPoint(portal, 1)->v,
-                     &btMprSimplexPoint(portal, 0)->v);
-    btMprVec3Sub2(&vb, &btMprSimplexPoint(portal, 2)->v,
-                     &btMprSimplexPoint(portal, 0)->v);
-    btMprVec3Cross(&dir, &va, &vb);
-    btMprVec3Normalize(&dir);
+	// it is better to form portal faces to be oriented "outside" origin
+	dot = btMprVec3Dot(&dir, &btMprSimplexPoint(portal, 0)->v);
+	if (dot > 0.f)
+	{
+		btMprSimplexSwap(portal, 1, 2);
+		btMprVec3Scale(&dir, -1.f);
+	}
 
-    // it is better to form portal faces to be oriented "outside" origin
-    dot = btMprVec3Dot(&dir, &btMprSimplexPoint(portal, 0)->v);
-    if (dot > 0.f){
-        btMprSimplexSwap(portal, 1, 2);
-        btMprVec3Scale(&dir, -1.f);
-    }
+	while (btMprSimplexSize(portal) < 4)
+	{
+		btMprSupport(a, b, colDesc, dir, btMprSimplexPointW(portal, 3));
 
-    while (btMprSimplexSize(portal) < 4){
-		 btMprSupport(a,b,colDesc, dir, btMprSimplexPointW(portal, 3));
-        
-        dot = btMprVec3Dot(&btMprSimplexPoint(portal, 3)->v, &dir);
-        if (btMprIsZero(dot) || dot < 0.f)
-            return -1;
+		dot = btMprVec3Dot(&btMprSimplexPoint(portal, 3)->v, &dir);
+		if (btMprIsZero(dot) || dot < 0.f)
+			return -1;
 
-        cont = 0;
+		cont = 0;
 
-        // test if origin is outside (v1, v0, v3) - set v2 as v3 and
-        // continue
-        btMprVec3Cross(&va, &btMprSimplexPoint(portal, 1)->v,
-                          &btMprSimplexPoint(portal, 3)->v);
-        dot = btMprVec3Dot(&va, &btMprSimplexPoint(portal, 0)->v);
-        if (dot < 0.f && !btMprIsZero(dot)){
-            btMprSimplexSet(portal, 2, btMprSimplexPoint(portal, 3));
-            cont = 1;
-        }
+		// test if origin is outside (v1, v0, v3) - set v2 as v3 and
+		// continue
+		btMprVec3Cross(&va, &btMprSimplexPoint(portal, 1)->v,
+					   &btMprSimplexPoint(portal, 3)->v);
+		dot = btMprVec3Dot(&va, &btMprSimplexPoint(portal, 0)->v);
+		if (dot < 0.f && !btMprIsZero(dot))
+		{
+			btMprSimplexSet(portal, 2, btMprSimplexPoint(portal, 3));
+			cont = 1;
+		}
 
-        if (!cont){
-            // test if origin is outside (v3, v0, v2) - set v1 as v3 and
-            // continue
-            btMprVec3Cross(&va, &btMprSimplexPoint(portal, 3)->v,
-                              &btMprSimplexPoint(portal, 2)->v);
-            dot = btMprVec3Dot(&va, &btMprSimplexPoint(portal, 0)->v);
-            if (dot < 0.f && !btMprIsZero(dot)){
-                btMprSimplexSet(portal, 1, btMprSimplexPoint(portal, 3));
-                cont = 1;
-            }
-        }
+		if (!cont)
+		{
+			// test if origin is outside (v3, v0, v2) - set v1 as v3 and
+			// continue
+			btMprVec3Cross(&va, &btMprSimplexPoint(portal, 3)->v,
+						   &btMprSimplexPoint(portal, 2)->v);
+			dot = btMprVec3Dot(&va, &btMprSimplexPoint(portal, 0)->v);
+			if (dot < 0.f && !btMprIsZero(dot))
+			{
+				btMprSimplexSet(portal, 1, btMprSimplexPoint(portal, 3));
+				cont = 1;
+			}
+		}
 
-        if (cont){
-            btMprVec3Sub2(&va, &btMprSimplexPoint(portal, 1)->v,
-                             &btMprSimplexPoint(portal, 0)->v);
-            btMprVec3Sub2(&vb, &btMprSimplexPoint(portal, 2)->v,
-                             &btMprSimplexPoint(portal, 0)->v);
-            btMprVec3Cross(&dir, &va, &vb);
-            btMprVec3Normalize(&dir);
-        }else{
-            btMprSimplexSetSize(portal, 4);
-        }
-    }
+		if (cont)
+		{
+			btMprVec3Sub2(&va, &btMprSimplexPoint(portal, 1)->v,
+						  &btMprSimplexPoint(portal, 0)->v);
+			btMprVec3Sub2(&vb, &btMprSimplexPoint(portal, 2)->v,
+						  &btMprSimplexPoint(portal, 0)->v);
+			btMprVec3Cross(&dir, &va, &vb);
+			btMprVec3Normalize(&dir);
+		}
+		else
+		{
+			btMprSimplexSetSize(portal, 4);
+		}
+	}
 
-    return 0;
+	return 0;
 }
 
 template <typename btConvexTemplate>
-static int btRefinePortal(const btConvexTemplate& a, const btConvexTemplate& b,const btMprCollisionDescription& colDesc,
-							btMprSimplex_t *portal)
+static int btRefinePortal(const btConvexTemplate &a, const btConvexTemplate &b, const btMprCollisionDescription &colDesc,
+						  btMprSimplex_t *portal)
 {
-    btVector3 dir;
-    btMprSupport_t v4;
+	btVector3 dir;
+	btMprSupport_t v4;
 
-	for (int i=0;i<BT_MPR_MAX_ITERATIONS;i++)
-    //while (1)
+	for (int i = 0; i < BT_MPR_MAX_ITERATIONS; i++)
+	//while (1)
 	{
-        // compute direction outside the portal (from v0 throught v1,v2,v3
-        // face)
-        btPortalDir(portal, &dir);
+		// compute direction outside the portal (from v0 throught v1,v2,v3
+		// face)
+		btPortalDir(portal, &dir);
 
-        // test if origin is inside the portal
-        if (portalEncapsulesOrigin(portal, &dir))
-            return 0;
+		// test if origin is inside the portal
+		if (portalEncapsulesOrigin(portal, &dir))
+			return 0;
 
-        // get next support point
-        
-		 btMprSupport(a,b,colDesc, dir, &v4);
+		// get next support point
 
+		btMprSupport(a, b, colDesc, dir, &v4);
 
-        // test if v4 can expand portal to contain origin and if portal
-        // expanding doesn't reach given tolerance
-        if (!portalCanEncapsuleOrigin(portal, &v4, &dir)
-                || portalReachTolerance(portal, &v4, &dir))
+		// test if v4 can expand portal to contain origin and if portal
+		// expanding doesn't reach given tolerance
+		if (!portalCanEncapsuleOrigin(portal, &v4, &dir) || portalReachTolerance(portal, &v4, &dir))
 		{
-            return -1;
-        }
+			return -1;
+		}
 
-        // v1-v2-v3 triangle must be rearranged to face outside Minkowski
-        // difference (direction from v0).
-        btExpandPortal(portal, &v4);
-    }
+		// v1-v2-v3 triangle must be rearranged to face outside Minkowski
+		// difference (direction from v0).
+		btExpandPortal(portal, &v4);
+	}
 
-    return -1;
+	return -1;
 }
 
 static void btFindPos(const btMprSimplex_t *portal, btVector3 *pos)
 {
+	btVector3 zero = btVector3(0, 0, 0);
+	btVector3 *origin = &zero;
 
-	btVector3 zero = btVector3(0,0,0);
-	btVector3* origin = &zero;
+	btVector3 dir;
+	size_t i;
+	float b[4], sum, inv;
+	btVector3 vec, p1, p2;
 
-    btVector3 dir;
-    size_t i;
-    float b[4], sum, inv;
-    btVector3 vec, p1, p2;
+	btPortalDir(portal, &dir);
 
-    btPortalDir(portal, &dir);
+	// use barycentric coordinates of tetrahedron to find origin
+	btMprVec3Cross(&vec, &btMprSimplexPoint(portal, 1)->v,
+				   &btMprSimplexPoint(portal, 2)->v);
+	b[0] = btMprVec3Dot(&vec, &btMprSimplexPoint(portal, 3)->v);
 
-    // use barycentric coordinates of tetrahedron to find origin
-    btMprVec3Cross(&vec, &btMprSimplexPoint(portal, 1)->v,
-                       &btMprSimplexPoint(portal, 2)->v);
-    b[0] = btMprVec3Dot(&vec, &btMprSimplexPoint(portal, 3)->v);
+	btMprVec3Cross(&vec, &btMprSimplexPoint(portal, 3)->v,
+				   &btMprSimplexPoint(portal, 2)->v);
+	b[1] = btMprVec3Dot(&vec, &btMprSimplexPoint(portal, 0)->v);
 
-    btMprVec3Cross(&vec, &btMprSimplexPoint(portal, 3)->v,
-                       &btMprSimplexPoint(portal, 2)->v);
-    b[1] = btMprVec3Dot(&vec, &btMprSimplexPoint(portal, 0)->v);
+	btMprVec3Cross(&vec, &btMprSimplexPoint(portal, 0)->v,
+				   &btMprSimplexPoint(portal, 1)->v);
+	b[2] = btMprVec3Dot(&vec, &btMprSimplexPoint(portal, 3)->v);
 
-    btMprVec3Cross(&vec, &btMprSimplexPoint(portal, 0)->v,
-                       &btMprSimplexPoint(portal, 1)->v);
-    b[2] = btMprVec3Dot(&vec, &btMprSimplexPoint(portal, 3)->v);
-
-    btMprVec3Cross(&vec, &btMprSimplexPoint(portal, 2)->v,
-                       &btMprSimplexPoint(portal, 1)->v);
-    b[3] = btMprVec3Dot(&vec, &btMprSimplexPoint(portal, 0)->v);
+	btMprVec3Cross(&vec, &btMprSimplexPoint(portal, 2)->v,
+				   &btMprSimplexPoint(portal, 1)->v);
+	b[3] = btMprVec3Dot(&vec, &btMprSimplexPoint(portal, 0)->v);
 
 	sum = b[0] + b[1] + b[2] + b[3];
 
-    if (btMprIsZero(sum) || sum < 0.f){
+	if (btMprIsZero(sum) || sum < 0.f)
+	{
 		b[0] = 0.f;
 
-        btMprVec3Cross(&vec, &btMprSimplexPoint(portal, 2)->v,
-                           &btMprSimplexPoint(portal, 3)->v);
-        b[1] = btMprVec3Dot(&vec, &dir);
-        btMprVec3Cross(&vec, &btMprSimplexPoint(portal, 3)->v,
-                           &btMprSimplexPoint(portal, 1)->v);
-        b[2] = btMprVec3Dot(&vec, &dir);
-        btMprVec3Cross(&vec, &btMprSimplexPoint(portal, 1)->v,
-                           &btMprSimplexPoint(portal, 2)->v);
-        b[3] = btMprVec3Dot(&vec, &dir);
+		btMprVec3Cross(&vec, &btMprSimplexPoint(portal, 2)->v,
+					   &btMprSimplexPoint(portal, 3)->v);
+		b[1] = btMprVec3Dot(&vec, &dir);
+		btMprVec3Cross(&vec, &btMprSimplexPoint(portal, 3)->v,
+					   &btMprSimplexPoint(portal, 1)->v);
+		b[2] = btMprVec3Dot(&vec, &dir);
+		btMprVec3Cross(&vec, &btMprSimplexPoint(portal, 1)->v,
+					   &btMprSimplexPoint(portal, 2)->v);
+		b[3] = btMprVec3Dot(&vec, &dir);
 
 		sum = b[1] + b[2] + b[3];
 	}
 
 	inv = 1.f / sum;
 
-    btMprVec3Copy(&p1, origin);
-    btMprVec3Copy(&p2, origin);
-    for (i = 0; i < 4; i++){
-        btMprVec3Copy(&vec, &btMprSimplexPoint(portal, i)->v1);
-        btMprVec3Scale(&vec, b[i]);
-        btMprVec3Add(&p1, &vec);
+	btMprVec3Copy(&p1, origin);
+	btMprVec3Copy(&p2, origin);
+	for (i = 0; i < 4; i++)
+	{
+		btMprVec3Copy(&vec, &btMprSimplexPoint(portal, i)->v1);
+		btMprVec3Scale(&vec, b[i]);
+		btMprVec3Add(&p1, &vec);
 
-        btMprVec3Copy(&vec, &btMprSimplexPoint(portal, i)->v2);
-        btMprVec3Scale(&vec, b[i]);
-        btMprVec3Add(&p2, &vec);
-    }
-    btMprVec3Scale(&p1, inv);
-    btMprVec3Scale(&p2, inv);
+		btMprVec3Copy(&vec, &btMprSimplexPoint(portal, i)->v2);
+		btMprVec3Scale(&vec, b[i]);
+		btMprVec3Add(&p2, &vec);
+	}
+	btMprVec3Scale(&p1, inv);
+	btMprVec3Scale(&p2, inv);
 #ifdef MPR_AVERAGE_CONTACT_POSITIONS
-    btMprVec3Copy(pos, &p1);
-    btMprVec3Add(pos, &p2);
-    btMprVec3Scale(pos, 0.5);
+	btMprVec3Copy(pos, &p1);
+	btMprVec3Add(pos, &p2);
+	btMprVec3Scale(pos, 0.5);
 #else
-    btMprVec3Copy(pos, &p2);
-#endif//MPR_AVERAGE_CONTACT_POSITIONS
+	btMprVec3Copy(pos, &p2);
+#endif  //MPR_AVERAGE_CONTACT_POSITIONS
 }
 
 inline float btMprVec3Dist2(const btVector3 *a, const btVector3 *b)
 {
-    btVector3 ab;
-    btMprVec3Sub2(&ab, a, b);
-    return btMprVec3Len2(&ab);
+	btVector3 ab;
+	btMprVec3Sub2(&ab, a, b);
+	return btMprVec3Len2(&ab);
 }
 
 inline float _btMprVec3PointSegmentDist2(const btVector3 *P,
-                                                  const btVector3 *x0,
-                                                  const btVector3 *b,
-                                                  btVector3 *witness)
+										 const btVector3 *x0,
+										 const btVector3 *b,
+										 btVector3 *witness)
 {
-    // The computation comes from solving equation of segment:
-    //      S(t) = x0 + t.d
-    //          where - x0 is initial point of segment
-    //                - d is direction of segment from x0 (|d| > 0)
-    //                - t belongs to <0, 1> interval
-    // 
-    // Than, distance from a segment to some point P can be expressed:
-    //      D(t) = |x0 + t.d - P|^2
-    //          which is distance from any point on segment. Minimization
-    //          of this function brings distance from P to segment.
-    // Minimization of D(t) leads to simple quadratic equation that's
-    // solving is straightforward.
-    //
-    // Bonus of this method is witness point for free.
+	// The computation comes from solving equation of segment:
+	//      S(t) = x0 + t.d
+	//          where - x0 is initial point of segment
+	//                - d is direction of segment from x0 (|d| > 0)
+	//                - t belongs to <0, 1> interval
+	//
+	// Than, distance from a segment to some point P can be expressed:
+	//      D(t) = |x0 + t.d - P|^2
+	//          which is distance from any point on segment. Minimization
+	//          of this function brings distance from P to segment.
+	// Minimization of D(t) leads to simple quadratic equation that's
+	// solving is straightforward.
+	//
+	// Bonus of this method is witness point for free.
 
-    float dist, t;
-    btVector3 d, a;
+	float dist, t;
+	btVector3 d, a;
 
-    // direction of segment
-    btMprVec3Sub2(&d, b, x0);
+	// direction of segment
+	btMprVec3Sub2(&d, b, x0);
 
-    // precompute vector from P to x0
-    btMprVec3Sub2(&a, x0, P);
+	// precompute vector from P to x0
+	btMprVec3Sub2(&a, x0, P);
 
-    t  = -1.f * btMprVec3Dot(&a, &d);
-    t /= btMprVec3Len2(&d);
+	t = -1.f * btMprVec3Dot(&a, &d);
+	t /= btMprVec3Len2(&d);
 
-    if (t < 0.f || btMprIsZero(t)){
-        dist = btMprVec3Dist2(x0, P);
-        if (witness)
-            btMprVec3Copy(witness, x0);
-    }else if (t > 1.f || btMprEq(t, 1.f)){
-        dist = btMprVec3Dist2(b, P);
-        if (witness)
-            btMprVec3Copy(witness, b);
-    }else{
-        if (witness){
-            btMprVec3Copy(witness, &d);
-            btMprVec3Scale(witness, t);
-            btMprVec3Add(witness, x0);
-            dist = btMprVec3Dist2(witness, P);
-        }else{
-            // recycling variables
-            btMprVec3Scale(&d, t);
-            btMprVec3Add(&d, &a);
-            dist = btMprVec3Len2(&d);
-        }
-    }
+	if (t < 0.f || btMprIsZero(t))
+	{
+		dist = btMprVec3Dist2(x0, P);
+		if (witness)
+			btMprVec3Copy(witness, x0);
+	}
+	else if (t > 1.f || btMprEq(t, 1.f))
+	{
+		dist = btMprVec3Dist2(b, P);
+		if (witness)
+			btMprVec3Copy(witness, b);
+	}
+	else
+	{
+		if (witness)
+		{
+			btMprVec3Copy(witness, &d);
+			btMprVec3Scale(witness, t);
+			btMprVec3Add(witness, x0);
+			dist = btMprVec3Dist2(witness, P);
+		}
+		else
+		{
+			// recycling variables
+			btMprVec3Scale(&d, t);
+			btMprVec3Add(&d, &a);
+			dist = btMprVec3Len2(&d);
+		}
+	}
 
-    return dist;
+	return dist;
 }
 
-
-
 inline float btMprVec3PointTriDist2(const btVector3 *P,
-                                const btVector3 *x0, const btVector3 *B,
-                                const btVector3 *C,
-                                btVector3 *witness)
+									const btVector3 *x0, const btVector3 *B,
+									const btVector3 *C,
+									btVector3 *witness)
 {
-    // Computation comes from analytic expression for triangle (x0, B, C)
-    //      T(s, t) = x0 + s.d1 + t.d2, where d1 = B - x0 and d2 = C - x0 and
-    // Then equation for distance is:
-    //      D(s, t) = | T(s, t) - P |^2
-    // This leads to minimization of quadratic function of two variables.
-    // The solution from is taken only if s is between 0 and 1, t is
-    // between 0 and 1 and t + s < 1, otherwise distance from segment is
-    // computed.
+	// Computation comes from analytic expression for triangle (x0, B, C)
+	//      T(s, t) = x0 + s.d1 + t.d2, where d1 = B - x0 and d2 = C - x0 and
+	// Then equation for distance is:
+	//      D(s, t) = | T(s, t) - P |^2
+	// This leads to minimization of quadratic function of two variables.
+	// The solution from is taken only if s is between 0 and 1, t is
+	// between 0 and 1 and t + s < 1, otherwise distance from segment is
+	// computed.
 
-    btVector3 d1, d2, a;
-    float u, v, w, p, q, r;
-    float s, t, dist, dist2;
-    btVector3 witness2;
+	btVector3 d1, d2, a;
+	float u, v, w, p, q, r;
+	float s, t, dist, dist2;
+	btVector3 witness2;
 
-    btMprVec3Sub2(&d1, B, x0);
-    btMprVec3Sub2(&d2, C, x0);
-    btMprVec3Sub2(&a, x0, P);
+	btMprVec3Sub2(&d1, B, x0);
+	btMprVec3Sub2(&d2, C, x0);
+	btMprVec3Sub2(&a, x0, P);
 
-    u = btMprVec3Dot(&a, &a);
-    v = btMprVec3Dot(&d1, &d1);
-    w = btMprVec3Dot(&d2, &d2);
-    p = btMprVec3Dot(&a, &d1);
-    q = btMprVec3Dot(&a, &d2);
-    r = btMprVec3Dot(&d1, &d2);
+	u = btMprVec3Dot(&a, &a);
+	v = btMprVec3Dot(&d1, &d1);
+	w = btMprVec3Dot(&d2, &d2);
+	p = btMprVec3Dot(&a, &d1);
+	q = btMprVec3Dot(&a, &d2);
+	r = btMprVec3Dot(&d1, &d2);
 
 	btScalar div = (w * v - r * r);
 	if (btMprIsZero(div))
 	{
-		s=-1;
-	} else
+		s = -1;
+	}
+	else
 	{
 		s = (q * r - w * p) / div;
 		t = (-s * r - q) / w;
 	}
 
-    if ((btMprIsZero(s) || s > 0.f)
-            && (btMprEq(s, 1.f) || s < 1.f)
-            && (btMprIsZero(t) || t > 0.f)
-            && (btMprEq(t, 1.f) || t < 1.f)
-            && (btMprEq(t + s, 1.f) || t + s < 1.f)){
+	if ((btMprIsZero(s) || s > 0.f) && (btMprEq(s, 1.f) || s < 1.f) && (btMprIsZero(t) || t > 0.f) && (btMprEq(t, 1.f) || t < 1.f) && (btMprEq(t + s, 1.f) || t + s < 1.f))
+	{
+		if (witness)
+		{
+			btMprVec3Scale(&d1, s);
+			btMprVec3Scale(&d2, t);
+			btMprVec3Copy(witness, x0);
+			btMprVec3Add(witness, &d1);
+			btMprVec3Add(witness, &d2);
 
-        if (witness){
-            btMprVec3Scale(&d1, s);
-            btMprVec3Scale(&d2, t);
-            btMprVec3Copy(witness, x0);
-            btMprVec3Add(witness, &d1);
-            btMprVec3Add(witness, &d2);
+			dist = btMprVec3Dist2(witness, P);
+		}
+		else
+		{
+			dist = s * s * v;
+			dist += t * t * w;
+			dist += 2.f * s * t * r;
+			dist += 2.f * s * p;
+			dist += 2.f * t * q;
+			dist += u;
+		}
+	}
+	else
+	{
+		dist = _btMprVec3PointSegmentDist2(P, x0, B, witness);
 
-            dist = btMprVec3Dist2(witness, P);
-        }else{
-            dist  = s * s * v;
-            dist += t * t * w;
-            dist += 2.f * s * t * r;
-            dist += 2.f * s * p;
-            dist += 2.f * t * q;
-            dist += u;
-        }
-    }else{
-        dist = _btMprVec3PointSegmentDist2(P, x0, B, witness);
+		dist2 = _btMprVec3PointSegmentDist2(P, x0, C, &witness2);
+		if (dist2 < dist)
+		{
+			dist = dist2;
+			if (witness)
+				btMprVec3Copy(witness, &witness2);
+		}
 
-        dist2 = _btMprVec3PointSegmentDist2(P, x0, C, &witness2);
-        if (dist2 < dist){
-            dist = dist2;
-            if (witness)
-                btMprVec3Copy(witness, &witness2);
-        }
+		dist2 = _btMprVec3PointSegmentDist2(P, B, C, &witness2);
+		if (dist2 < dist)
+		{
+			dist = dist2;
+			if (witness)
+				btMprVec3Copy(witness, &witness2);
+		}
+	}
 
-        dist2 = _btMprVec3PointSegmentDist2(P, B, C, &witness2);
-        if (dist2 < dist){
-            dist = dist2;
-            if (witness)
-                btMprVec3Copy(witness, &witness2);
-        }
-    }
-
-    return dist;
+	return dist;
 }
 
 template <typename btConvexTemplate>
-static void btFindPenetr(const btConvexTemplate& a, const btConvexTemplate& b,
-                         const btMprCollisionDescription& colDesc,
-                         btMprSimplex_t *portal,
-                         float *depth, btVector3 *pdir, btVector3 *pos)
+static void btFindPenetr(const btConvexTemplate &a, const btConvexTemplate &b,
+						 const btMprCollisionDescription &colDesc,
+						 btMprSimplex_t *portal,
+						 float *depth, btVector3 *pdir, btVector3 *pos)
 {
-    btVector3 dir;
-    btMprSupport_t v4;
-    unsigned long iterations;
+	btVector3 dir;
+	btMprSupport_t v4;
+	unsigned long iterations;
 
-	btVector3 zero = btVector3(0,0,0);
-	btVector3* origin = &zero;
+	btVector3 zero = btVector3(0, 0, 0);
+	btVector3 *origin = &zero;
 
-
-    iterations = 1UL;
-	for (int i=0;i<BT_MPR_MAX_ITERATIONS;i++)
-    //while (1)
+	iterations = 1UL;
+	for (int i = 0; i < BT_MPR_MAX_ITERATIONS; i++)
+	//while (1)
 	{
-        // compute portal direction and obtain next support point
-        btPortalDir(portal, &dir);
-        
-		 btMprSupport(a,b,colDesc, dir, &v4);
+		// compute portal direction and obtain next support point
+		btPortalDir(portal, &dir);
 
+		btMprSupport(a, b, colDesc, dir, &v4);
 
-        // reached tolerance -> find penetration info
-        if (portalReachTolerance(portal, &v4, &dir)
-                || iterations ==BT_MPR_MAX_ITERATIONS)
+		// reached tolerance -> find penetration info
+		if (portalReachTolerance(portal, &v4, &dir) || iterations == BT_MPR_MAX_ITERATIONS)
 		{
-            *depth = btMprVec3PointTriDist2(origin,&btMprSimplexPoint(portal, 1)->v,&btMprSimplexPoint(portal, 2)->v,&btMprSimplexPoint(portal, 3)->v,pdir);
-            *depth = BT_MPR_SQRT(*depth);
-			
+			*depth = btMprVec3PointTriDist2(origin, &btMprSimplexPoint(portal, 1)->v, &btMprSimplexPoint(portal, 2)->v, &btMprSimplexPoint(portal, 3)->v, pdir);
+			*depth = BT_MPR_SQRT(*depth);
+
 			if (btMprIsZero((*pdir).x()) && btMprIsZero((*pdir).y()) && btMprIsZero((*pdir).z()))
 			{
-				
 				*pdir = dir;
-			} 
+			}
 			btMprVec3Normalize(pdir);
-			
-            // barycentric coordinates:
-            btFindPos(portal, pos);
 
+			// barycentric coordinates:
+			btFindPos(portal, pos);
 
-            return;
-        }
+			return;
+		}
 
-        btExpandPortal(portal, &v4);
+		btExpandPortal(portal, &v4);
 
-        iterations++;
-    }
+		iterations++;
+	}
 }
 
-static void btFindPenetrTouch(btMprSimplex_t *portal,float *depth, btVector3 *dir, btVector3 *pos)
+static void btFindPenetrTouch(btMprSimplex_t *portal, float *depth, btVector3 *dir, btVector3 *pos)
 {
-    // Touching contact on portal's v1 - so depth is zero and direction
-    // is unimportant and pos can be guessed
-    *depth = 0.f;
-    btVector3 zero = btVector3(0,0,0);
-	btVector3* origin = &zero;
-
+	// Touching contact on portal's v1 - so depth is zero and direction
+	// is unimportant and pos can be guessed
+	*depth = 0.f;
+	btVector3 zero = btVector3(0, 0, 0);
+	btVector3 *origin = &zero;
 
 	btMprVec3Copy(dir, origin);
 #ifdef MPR_AVERAGE_CONTACT_POSITIONS
-    btMprVec3Copy(pos, &btMprSimplexPoint(portal, 1)->v1);
-    btMprVec3Add(pos, &btMprSimplexPoint(portal, 1)->v2);
-    btMprVec3Scale(pos, 0.5);
+	btMprVec3Copy(pos, &btMprSimplexPoint(portal, 1)->v1);
+	btMprVec3Add(pos, &btMprSimplexPoint(portal, 1)->v2);
+	btMprVec3Scale(pos, 0.5);
 #else
-     btMprVec3Copy(pos, &btMprSimplexPoint(portal, 1)->v2);
+	btMprVec3Copy(pos, &btMprSimplexPoint(portal, 1)->v2);
 #endif
 }
 
 static void btFindPenetrSegment(btMprSimplex_t *portal,
-                              float *depth, btVector3 *dir, btVector3 *pos)
+								float *depth, btVector3 *dir, btVector3 *pos)
 {
-    
-    // Origin lies on v0-v1 segment.
-    // Depth is distance to v1, direction also and position must be
-    // computed
+	// Origin lies on v0-v1 segment.
+	// Depth is distance to v1, direction also and position must be
+	// computed
 #ifdef MPR_AVERAGE_CONTACT_POSITIONS
-    btMprVec3Copy(pos, &btMprSimplexPoint(portal, 1)->v1);
-    btMprVec3Add(pos, &btMprSimplexPoint(portal, 1)->v2);
-    btMprVec3Scale(pos, 0.5f);
+	btMprVec3Copy(pos, &btMprSimplexPoint(portal, 1)->v1);
+	btMprVec3Add(pos, &btMprSimplexPoint(portal, 1)->v2);
+	btMprVec3Scale(pos, 0.5f);
 #else
-     btMprVec3Copy(pos, &btMprSimplexPoint(portal, 1)->v2);
-#endif//MPR_AVERAGE_CONTACT_POSITIONS
+	btMprVec3Copy(pos, &btMprSimplexPoint(portal, 1)->v2);
+#endif  //MPR_AVERAGE_CONTACT_POSITIONS
 
-    btMprVec3Copy(dir, &btMprSimplexPoint(portal, 1)->v);
-    *depth = BT_MPR_SQRT(btMprVec3Len2(dir));
-    btMprVec3Normalize(dir);
-
-    
+	btMprVec3Copy(dir, &btMprSimplexPoint(portal, 1)->v);
+	*depth = BT_MPR_SQRT(btMprVec3Len2(dir));
+	btMprVec3Normalize(dir);
 }
 
-
 template <typename btConvexTemplate>
-inline int btMprPenetration( const btConvexTemplate& a, const btConvexTemplate& b,
-                            const btMprCollisionDescription& colDesc,
-					float *depthOut, btVector3* dirOut, btVector3* posOut)
+inline int btMprPenetration(const btConvexTemplate &a, const btConvexTemplate &b,
+							const btMprCollisionDescription &colDesc,
+							float *depthOut, btVector3 *dirOut, btVector3 *posOut)
 {
-	
-	 btMprSimplex_t portal;
+	btMprSimplex_t portal;
 
+	// Phase 1: Portal discovery
+	int result = btDiscoverPortal(a, b, colDesc, &portal);
 
-    // Phase 1: Portal discovery
-    int result = btDiscoverPortal(a,b,colDesc, &portal);
-	
-	  
 	//sepAxis[pairIndex] = *pdir;//or -dir?
 
 	switch (result)
 	{
-	case 0:
+		case 0:
 		{
 			// Phase 2: Portal refinement
-		
-			result = btRefinePortal(a,b,colDesc, &portal);
+
+			result = btRefinePortal(a, b, colDesc, &portal);
 			if (result < 0)
 				return -1;
 
 			// Phase 3. Penetration info
-			btFindPenetr(a,b,colDesc, &portal, depthOut, dirOut, posOut);
-			
-			
+			btFindPenetr(a, b, colDesc, &portal, depthOut, dirOut, posOut);
+
 			break;
 		}
-	case 1:
+		case 1:
 		{
-			 // Touching contact on portal's v1.
+			// Touching contact on portal's v1.
 			btFindPenetrTouch(&portal, depthOut, dirOut, posOut);
-			result=0;
+			result = 0;
 			break;
 		}
-	case 2:
+		case 2:
 		{
-			
-			btFindPenetrSegment( &portal, depthOut, dirOut, posOut);
-			result=0;
+			btFindPenetrSegment(&portal, depthOut, dirOut, posOut);
+			result = 0;
 			break;
 		}
-	default:
+		default:
 		{
 			//if (res < 0)
 			//{
-				// Origin isn't inside portal - no collision.
-				result = -1;
+			// Origin isn't inside portal - no collision.
+			result = -1;
 			//}
 		}
 	};
-	
+
 	return result;
 };
 
-
-template<typename btConvexTemplate, typename btMprDistanceTemplate>
-inline int	btComputeMprPenetration( const btConvexTemplate& a, const btConvexTemplate& b, const
-                                    btMprCollisionDescription& colDesc, btMprDistanceTemplate* distInfo)
+template <typename btConvexTemplate, typename btMprDistanceTemplate>
+inline int btComputeMprPenetration(const btConvexTemplate &a, const btConvexTemplate &b, const btMprCollisionDescription &colDesc, btMprDistanceTemplate *distInfo)
 {
-	btVector3 dir,pos;
+	btVector3 dir, pos;
 	float depth;
 
-	int res = btMprPenetration(a,b,colDesc,&depth, &dir, &pos);
-	if (res==0)
+	int res = btMprPenetration(a, b, colDesc, &depth, &dir, &pos);
+	if (res == 0)
 	{
 		distInfo->m_distance = -depth;
 		distInfo->m_pointOnB = pos;
 		distInfo->m_normalBtoA = -dir;
-		distInfo->m_pointOnA = pos-distInfo->m_distance*dir;
+		distInfo->m_pointOnA = pos - distInfo->m_distance * dir;
 		return 0;
 	}
 
 	return -1;
 }
 
-
-
-#endif //BT_MPR_PENETRATION_H
+#endif  //BT_MPR_PENETRATION_H
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btPersistentManifold.cpp b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btPersistentManifold.cpp
index 9603a8bbdca..f1422cad4d9 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btPersistentManifold.cpp
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btPersistentManifold.cpp
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btPersistentManifold.h"
 #include "LinearMath/btTransform.h"
 #include "LinearMath/btSerializer.h"
@@ -24,83 +23,76 @@ subject to the following restrictions:
 #define btCollisionObjectData btCollisionObjectFloatData
 #endif
 
-btScalar					gContactBreakingThreshold = btScalar(0.02);
-ContactDestroyedCallback	gContactDestroyedCallback = 0;
-ContactProcessedCallback	gContactProcessedCallback = 0;
-ContactStartedCallback		gContactStartedCallback = 0;
-ContactEndedCallback		gContactEndedCallback = 0;
+btScalar gContactBreakingThreshold = btScalar(0.02);
+ContactDestroyedCallback gContactDestroyedCallback = 0;
+ContactProcessedCallback gContactProcessedCallback = 0;
+ContactStartedCallback gContactStartedCallback = 0;
+ContactEndedCallback gContactEndedCallback = 0;
 ///gContactCalcArea3Points will approximate the convex hull area using 3 points
 ///when setting it to false, it will use 4 points to compute the area: it is more accurate but slower
-bool						gContactCalcArea3Points = true;
-
+bool gContactCalcArea3Points = true;
 
 btPersistentManifold::btPersistentManifold()
-:btTypedObject(BT_PERSISTENT_MANIFOLD_TYPE),
-m_body0(0),
-m_body1(0),
-m_cachedPoints (0),
-m_companionIdA(0),
-m_companionIdB(0),
-m_index1a(0)
+	: btTypedObject(BT_PERSISTENT_MANIFOLD_TYPE),
+	  m_body0(0),
+	  m_body1(0),
+	  m_cachedPoints(0),
+	  m_companionIdA(0),
+	  m_companionIdB(0),
+	  m_index1a(0)
 {
 }
 
-
-
-
 #ifdef DEBUG_PERSISTENCY
 #include <stdio.h>
-void	btPersistentManifold::DebugPersistency()
+void btPersistentManifold::DebugPersistency()
 {
 	int i;
-	printf("DebugPersistency : numPoints %d\n",m_cachedPoints);
-	for (i=0;i<m_cachedPoints;i++)
+	printf("DebugPersistency : numPoints %d\n", m_cachedPoints);
+	for (i = 0; i < m_cachedPoints; i++)
 	{
-		printf("m_pointCache[%d].m_userPersistentData = %x\n",i,m_pointCache[i].m_userPersistentData);
+		printf("m_pointCache[%d].m_userPersistentData = %x\n", i, m_pointCache[i].m_userPersistentData);
 	}
 }
-#endif //DEBUG_PERSISTENCY
+#endif  //DEBUG_PERSISTENCY
 
 void btPersistentManifold::clearUserCache(btManifoldPoint& pt)
 {
-
 	void* oldPtr = pt.m_userPersistentData;
 	if (oldPtr)
 	{
 #ifdef DEBUG_PERSISTENCY
 		int i;
 		int occurance = 0;
-		for (i=0;i<m_cachedPoints;i++)
+		for (i = 0; i < m_cachedPoints; i++)
 		{
 			if (m_pointCache[i].m_userPersistentData == oldPtr)
 			{
 				occurance++;
-				if (occurance>1)
+				if (occurance > 1)
 					printf("error in clearUserCache\n");
 			}
 		}
-		btAssert(occurance<=0);
-#endif //DEBUG_PERSISTENCY
+		btAssert(occurance <= 0);
+#endif  //DEBUG_PERSISTENCY
 
 		if (pt.m_userPersistentData && gContactDestroyedCallback)
 		{
 			(*gContactDestroyedCallback)(pt.m_userPersistentData);
 			pt.m_userPersistentData = 0;
 		}
-		
+
 #ifdef DEBUG_PERSISTENCY
 		DebugPersistency();
 #endif
 	}
-
-	
 }
 
-static inline btScalar calcArea4Points(const btVector3 &p0,const btVector3 &p1,const btVector3 &p2,const btVector3 &p3)
+static inline btScalar calcArea4Points(const btVector3& p0, const btVector3& p1, const btVector3& p2, const btVector3& p3)
 {
 	// It calculates possible 3 area constructed from random 4 points and returns the biggest one.
 
-	btVector3 a[3],b[3];
+	btVector3 a[3], b[3];
 	a[0] = p0 - p1;
 	a[1] = p0 - p2;
 	a[2] = p0 - p3;
@@ -113,100 +105,102 @@ static inline btScalar calcArea4Points(const btVector3 &p0,const btVector3 &p1,c
 	btVector3 tmp1 = a[1].cross(b[1]);
 	btVector3 tmp2 = a[2].cross(b[2]);
 
-	return btMax(btMax(tmp0.length2(),tmp1.length2()),tmp2.length2());
+	return btMax(btMax(tmp0.length2(), tmp1.length2()), tmp2.length2());
 }
 
-int btPersistentManifold::sortCachedPoints(const btManifoldPoint& pt) 
+int btPersistentManifold::sortCachedPoints(const btManifoldPoint& pt)
 {
-		//calculate 4 possible cases areas, and take biggest area
-		//also need to keep 'deepest'
-		
-		int maxPenetrationIndex = -1;
+	//calculate 4 possible cases areas, and take biggest area
+	//also need to keep 'deepest'
+
+	int maxPenetrationIndex = -1;
 #define KEEP_DEEPEST_POINT 1
 #ifdef KEEP_DEEPEST_POINT
-		btScalar maxPenetration = pt.getDistance();
-		for (int i=0;i<4;i++)
+	btScalar maxPenetration = pt.getDistance();
+	for (int i = 0; i < 4; i++)
+	{
+		if (m_pointCache[i].getDistance() < maxPenetration)
 		{
-			if (m_pointCache[i].getDistance() < maxPenetration)
-			{
-				maxPenetrationIndex = i;
-				maxPenetration = m_pointCache[i].getDistance();
-			}
+			maxPenetrationIndex = i;
+			maxPenetration = m_pointCache[i].getDistance();
 		}
-#endif //KEEP_DEEPEST_POINT
-		
-		btScalar res0(btScalar(0.)),res1(btScalar(0.)),res2(btScalar(0.)),res3(btScalar(0.));
+	}
+#endif  //KEEP_DEEPEST_POINT
+
+	btScalar res0(btScalar(0.)), res1(btScalar(0.)), res2(btScalar(0.)), res3(btScalar(0.));
 
 	if (gContactCalcArea3Points)
 	{
 		if (maxPenetrationIndex != 0)
 		{
-			btVector3 a0 = pt.m_localPointA-m_pointCache[1].m_localPointA;
-			btVector3 b0 = m_pointCache[3].m_localPointA-m_pointCache[2].m_localPointA;
+			btVector3 a0 = pt.m_localPointA - m_pointCache[1].m_localPointA;
+			btVector3 b0 = m_pointCache[3].m_localPointA - m_pointCache[2].m_localPointA;
 			btVector3 cross = a0.cross(b0);
 			res0 = cross.length2();
 		}
 		if (maxPenetrationIndex != 1)
 		{
-			btVector3 a1 = pt.m_localPointA-m_pointCache[0].m_localPointA;
-			btVector3 b1 = m_pointCache[3].m_localPointA-m_pointCache[2].m_localPointA;
+			btVector3 a1 = pt.m_localPointA - m_pointCache[0].m_localPointA;
+			btVector3 b1 = m_pointCache[3].m_localPointA - m_pointCache[2].m_localPointA;
 			btVector3 cross = a1.cross(b1);
 			res1 = cross.length2();
 		}
 
 		if (maxPenetrationIndex != 2)
 		{
-			btVector3 a2 = pt.m_localPointA-m_pointCache[0].m_localPointA;
-			btVector3 b2 = m_pointCache[3].m_localPointA-m_pointCache[1].m_localPointA;
+			btVector3 a2 = pt.m_localPointA - m_pointCache[0].m_localPointA;
+			btVector3 b2 = m_pointCache[3].m_localPointA - m_pointCache[1].m_localPointA;
 			btVector3 cross = a2.cross(b2);
 			res2 = cross.length2();
 		}
 
 		if (maxPenetrationIndex != 3)
 		{
-			btVector3 a3 = pt.m_localPointA-m_pointCache[0].m_localPointA;
-			btVector3 b3 = m_pointCache[2].m_localPointA-m_pointCache[1].m_localPointA;
+			btVector3 a3 = pt.m_localPointA - m_pointCache[0].m_localPointA;
+			btVector3 b3 = m_pointCache[2].m_localPointA - m_pointCache[1].m_localPointA;
 			btVector3 cross = a3.cross(b3);
 			res3 = cross.length2();
 		}
-	} 
+	}
 	else
 	{
-		if(maxPenetrationIndex != 0) {
-			res0 = calcArea4Points(pt.m_localPointA,m_pointCache[1].m_localPointA,m_pointCache[2].m_localPointA,m_pointCache[3].m_localPointA);
+		if (maxPenetrationIndex != 0)
+		{
+			res0 = calcArea4Points(pt.m_localPointA, m_pointCache[1].m_localPointA, m_pointCache[2].m_localPointA, m_pointCache[3].m_localPointA);
 		}
 
-		if(maxPenetrationIndex != 1) {
-			res1 = calcArea4Points(pt.m_localPointA,m_pointCache[0].m_localPointA,m_pointCache[2].m_localPointA,m_pointCache[3].m_localPointA);
+		if (maxPenetrationIndex != 1)
+		{
+			res1 = calcArea4Points(pt.m_localPointA, m_pointCache[0].m_localPointA, m_pointCache[2].m_localPointA, m_pointCache[3].m_localPointA);
 		}
 
-		if(maxPenetrationIndex != 2) {
-			res2 = calcArea4Points(pt.m_localPointA,m_pointCache[0].m_localPointA,m_pointCache[1].m_localPointA,m_pointCache[3].m_localPointA);
+		if (maxPenetrationIndex != 2)
+		{
+			res2 = calcArea4Points(pt.m_localPointA, m_pointCache[0].m_localPointA, m_pointCache[1].m_localPointA, m_pointCache[3].m_localPointA);
 		}
 
-		if(maxPenetrationIndex != 3) {
-			res3 = calcArea4Points(pt.m_localPointA,m_pointCache[0].m_localPointA,m_pointCache[1].m_localPointA,m_pointCache[2].m_localPointA);
+		if (maxPenetrationIndex != 3)
+		{
+			res3 = calcArea4Points(pt.m_localPointA, m_pointCache[0].m_localPointA, m_pointCache[1].m_localPointA, m_pointCache[2].m_localPointA);
 		}
 	}
-	btVector4 maxvec(res0,res1,res2,res3);
+	btVector4 maxvec(res0, res1, res2, res3);
 	int biggestarea = maxvec.closestAxis4();
 	return biggestarea;
-	
 }
 
-
 int btPersistentManifold::getCacheEntry(const btManifoldPoint& newPoint) const
 {
-	btScalar shortestDist =  getContactBreakingThreshold() * getContactBreakingThreshold();
+	btScalar shortestDist = getContactBreakingThreshold() * getContactBreakingThreshold();
 	int size = getNumContacts();
 	int nearestPoint = -1;
-	for( int i = 0; i < size; i++ )
+	for (int i = 0; i < size; i++)
 	{
-		const btManifoldPoint &mp = m_pointCache[i];
+		const btManifoldPoint& mp = m_pointCache[i];
 
-		btVector3 diffA =  mp.m_localPointA- newPoint.m_localPointA;
+		btVector3 diffA = mp.m_localPointA - newPoint.m_localPointA;
 		const btScalar distToManiPoint = diffA.dot(diffA);
-		if( distToManiPoint < shortestDist )
+		if (distToManiPoint < shortestDist)
 		{
 			shortestDist = distToManiPoint;
 			nearestPoint = i;
@@ -221,7 +215,7 @@ int btPersistentManifold::addManifoldPoint(const btManifoldPoint& newPoint, bool
 	{
 		btAssert(validContactDistance(newPoint));
 	}
-	
+
 	int insertIndex = getNumContacts();
 	if (insertIndex == MANIFOLD_CACHE_SIZE)
 	{
@@ -232,91 +226,87 @@ int btPersistentManifold::addManifoldPoint(const btManifoldPoint& newPoint, bool
 		insertIndex = 0;
 #endif
 		clearUserCache(m_pointCache[insertIndex]);
-		
-	} else
+	}
+	else
 	{
 		m_cachedPoints++;
-
-		
 	}
-	if (insertIndex<0)
-		insertIndex=0;
+	if (insertIndex < 0)
+		insertIndex = 0;
 
-	btAssert(m_pointCache[insertIndex].m_userPersistentData==0);
+	btAssert(m_pointCache[insertIndex].m_userPersistentData == 0);
 	m_pointCache[insertIndex] = newPoint;
 	return insertIndex;
 }
 
-btScalar	btPersistentManifold::getContactBreakingThreshold() const
+btScalar btPersistentManifold::getContactBreakingThreshold() const
 {
 	return m_contactBreakingThreshold;
 }
 
-
-
-void btPersistentManifold::refreshContactPoints(const btTransform& trA,const btTransform& trB)
+void btPersistentManifold::refreshContactPoints(const btTransform& trA, const btTransform& trB)
 {
 	int i;
 #ifdef DEBUG_PERSISTENCY
 	printf("refreshContactPoints posA = (%f,%f,%f) posB = (%f,%f,%f)\n",
-		trA.getOrigin().getX(),
-		trA.getOrigin().getY(),
-		trA.getOrigin().getZ(),
-		trB.getOrigin().getX(),
-		trB.getOrigin().getY(),
-		trB.getOrigin().getZ());
-#endif //DEBUG_PERSISTENCY
+		   trA.getOrigin().getX(),
+		   trA.getOrigin().getY(),
+		   trA.getOrigin().getZ(),
+		   trB.getOrigin().getX(),
+		   trB.getOrigin().getY(),
+		   trB.getOrigin().getZ());
+#endif  //DEBUG_PERSISTENCY
 	/// first refresh worldspace positions and distance
-	for (i=getNumContacts()-1;i>=0;i--)
+	for (i = getNumContacts() - 1; i >= 0; i--)
 	{
-		btManifoldPoint &manifoldPoint = m_pointCache[i];
-		manifoldPoint.m_positionWorldOnA = trA( manifoldPoint.m_localPointA );
-		manifoldPoint.m_positionWorldOnB = trB( manifoldPoint.m_localPointB );
-		manifoldPoint.m_distance1 = (manifoldPoint.m_positionWorldOnA -  manifoldPoint.m_positionWorldOnB).dot(manifoldPoint.m_normalWorldOnB);
+		btManifoldPoint& manifoldPoint = m_pointCache[i];
+		manifoldPoint.m_positionWorldOnA = trA(manifoldPoint.m_localPointA);
+		manifoldPoint.m_positionWorldOnB = trB(manifoldPoint.m_localPointB);
+		manifoldPoint.m_distance1 = (manifoldPoint.m_positionWorldOnA - manifoldPoint.m_positionWorldOnB).dot(manifoldPoint.m_normalWorldOnB);
 		manifoldPoint.m_lifeTime++;
 	}
 
-	/// then 
+	/// then
 	btScalar distance2d;
-	btVector3 projectedDifference,projectedPoint;
-	for (i=getNumContacts()-1;i>=0;i--)
+	btVector3 projectedDifference, projectedPoint;
+	for (i = getNumContacts() - 1; i >= 0; i--)
 	{
-		
-		btManifoldPoint &manifoldPoint = m_pointCache[i];
+		btManifoldPoint& manifoldPoint = m_pointCache[i];
 		//contact becomes invalid when signed distance exceeds margin (projected on contactnormal direction)
 		if (!validContactDistance(manifoldPoint))
 		{
 			removeContactPoint(i);
-		} else
+		}
+		else
 		{
-            //todo: friction anchor may require the contact to be around a bit longer
+			//todo: friction anchor may require the contact to be around a bit longer
 			//contact also becomes invalid when relative movement orthogonal to normal exceeds margin
 			projectedPoint = manifoldPoint.m_positionWorldOnA - manifoldPoint.m_normalWorldOnB * manifoldPoint.m_distance1;
 			projectedDifference = manifoldPoint.m_positionWorldOnB - projectedPoint;
 			distance2d = projectedDifference.dot(projectedDifference);
-			if (distance2d  > getContactBreakingThreshold()*getContactBreakingThreshold() )
+			if (distance2d > getContactBreakingThreshold() * getContactBreakingThreshold())
 			{
 				removeContactPoint(i);
-			} else
+			}
+			else
 			{
 				//contact point processed callback
 				if (gContactProcessedCallback)
-					(*gContactProcessedCallback)(manifoldPoint,(void*)m_body0,(void*)m_body1);
+					(*gContactProcessedCallback)(manifoldPoint, (void*)m_body0, (void*)m_body1);
 			}
 		}
 	}
 #ifdef DEBUG_PERSISTENCY
 	DebugPersistency();
-#endif //
+#endif  //
 }
 
-
-int	btPersistentManifold::calculateSerializeBufferSize()	const
+int btPersistentManifold::calculateSerializeBufferSize() const
 {
 	return sizeof(btPersistentManifoldData);
 }
 
-const char*	btPersistentManifold::serialize(const class btPersistentManifold* manifold, void* dataBuffer, class btSerializer* serializer) const
+const char* btPersistentManifold::serialize(const class btPersistentManifold* manifold, void* dataBuffer, class btSerializer* serializer) const
 {
 	btPersistentManifoldData* dataOut = (btPersistentManifoldData*)dataBuffer;
 	memset(dataOut, 0, sizeof(btPersistentManifoldData));
@@ -379,7 +369,7 @@ void btPersistentManifold::deSerialize(const struct btPersistentManifoldDoubleDa
 	for (int i = 0; i < this->getNumContacts(); i++)
 	{
 		btManifoldPoint& pt = m_pointCache[i];
-		
+
 		pt.m_appliedImpulse = manifoldDataPtr->m_pointCacheAppliedImpulse[i];
 		pt.m_appliedImpulseLateral1 = manifoldDataPtr->m_pointCacheAppliedImpulseLateral1[i];
 		pt.m_appliedImpulseLateral2 = manifoldDataPtr->m_pointCacheAppliedImpulseLateral2[i];
@@ -409,7 +399,6 @@ void btPersistentManifold::deSerialize(const struct btPersistentManifoldDoubleDa
 		pt.m_contactMotion1 = manifoldDataPtr->m_pointCacheContactMotion1[i];
 		pt.m_contactMotion2 = manifoldDataPtr->m_pointCacheContactMotion2[i];
 	}
-
 }
 
 void btPersistentManifold::deSerialize(const struct btPersistentManifoldFloatData* manifoldDataPtr)
@@ -455,5 +444,4 @@ void btPersistentManifold::deSerialize(const struct btPersistentManifoldFloatDat
 		pt.m_contactMotion1 = manifoldDataPtr->m_pointCacheContactMotion1[i];
 		pt.m_contactMotion2 = manifoldDataPtr->m_pointCacheContactMotion2[i];
 	}
-
 }
\ No newline at end of file
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btPersistentManifold.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btPersistentManifold.h
index 67be0c48eba..8a9134c95cc 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btPersistentManifold.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btPersistentManifold.h
@@ -16,7 +16,6 @@ subject to the following restrictions:
 #ifndef BT_PERSISTENT_MANIFOLD_H
 #define BT_PERSISTENT_MANIFOLD_H
 
-
 #include "LinearMath/btVector3.h"
 #include "LinearMath/btTransform.h"
 #include "btManifoldPoint.h"
@@ -34,14 +33,14 @@ extern btScalar gContactBreakingThreshold;
 class btPersistentManifold;
 
 typedef bool (*ContactDestroyedCallback)(void* userPersistentData);
-typedef bool (*ContactProcessedCallback)(btManifoldPoint& cp,void* body0,void* body1);
-typedef void (*ContactStartedCallback)(btPersistentManifold* const &manifold);
-typedef void (*ContactEndedCallback)(btPersistentManifold* const &manifold);
-extern ContactDestroyedCallback	gContactDestroyedCallback;
+typedef bool (*ContactProcessedCallback)(btManifoldPoint& cp, void* body0, void* body1);
+typedef void (*ContactStartedCallback)(btPersistentManifold* const& manifold);
+typedef void (*ContactEndedCallback)(btPersistentManifold* const& manifold);
+extern ContactDestroyedCallback gContactDestroyedCallback;
 extern ContactProcessedCallback gContactProcessedCallback;
 extern ContactStartedCallback gContactStartedCallback;
 extern ContactEndedCallback gContactEndedCallback;
-#endif //SWIG
+#endif  //SWIG
 
 //the enum starts at 1024 to avoid type conflicts with btTypedConstraint
 enum btContactManifoldTypes
@@ -60,73 +59,74 @@ enum btContactManifoldTypes
 ///the contact point with deepest penetration is always kept, and it tries to maximuze the area covered by the points
 ///note that some pairs of objects might have more then one contact manifold.
 
-
 //ATTRIBUTE_ALIGNED128( class) btPersistentManifold : public btTypedObject
-ATTRIBUTE_ALIGNED16( class) btPersistentManifold : public btTypedObject
+ATTRIBUTE_ALIGNED16(class)
+btPersistentManifold : public btTypedObject
 {
-
 	btManifoldPoint m_pointCache[MANIFOLD_CACHE_SIZE];
 
 	/// this two body pointers can point to the physics rigidbody class.
 	const btCollisionObject* m_body0;
 	const btCollisionObject* m_body1;
 
-	int	m_cachedPoints;
+	int m_cachedPoints;
 
-	btScalar	m_contactBreakingThreshold;
-	btScalar	m_contactProcessingThreshold;
+	btScalar m_contactBreakingThreshold;
+	btScalar m_contactProcessingThreshold;
 
-	
 	/// sort cached points so most isolated points come first
-	int	sortCachedPoints(const btManifoldPoint& pt);
+	int sortCachedPoints(const btManifoldPoint& pt);
 
-	int		findContactPoint(const btManifoldPoint* unUsed, int numUnused,const btManifoldPoint& pt);
+	int findContactPoint(const btManifoldPoint* unUsed, int numUnused, const btManifoldPoint& pt);
 
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-	int	m_companionIdA;
-	int	m_companionIdB;
+	int m_companionIdA;
+	int m_companionIdB;
 
 	int m_index1a;
 
 	btPersistentManifold();
 
-	btPersistentManifold(const btCollisionObject* body0,const btCollisionObject* body1,int , btScalar contactBreakingThreshold,btScalar contactProcessingThreshold)
+	btPersistentManifold(const btCollisionObject* body0, const btCollisionObject* body1, int, btScalar contactBreakingThreshold, btScalar contactProcessingThreshold)
 		: btTypedObject(BT_PERSISTENT_MANIFOLD_TYPE),
-	m_body0(body0),m_body1(body1),m_cachedPoints(0),
-		m_contactBreakingThreshold(contactBreakingThreshold),
-		m_contactProcessingThreshold(contactProcessingThreshold),
-		m_companionIdA(0),
-		m_companionIdB(0),
-		m_index1a(0)
+		  m_body0(body0),
+		  m_body1(body1),
+		  m_cachedPoints(0),
+		  m_contactBreakingThreshold(contactBreakingThreshold),
+		  m_contactProcessingThreshold(contactProcessingThreshold),
+		  m_companionIdA(0),
+		  m_companionIdB(0),
+		  m_index1a(0)
 	{
 	}
 
-	SIMD_FORCE_INLINE const btCollisionObject* getBody0() const { return m_body0;}
-	SIMD_FORCE_INLINE const btCollisionObject* getBody1() const { return m_body1;}
+	SIMD_FORCE_INLINE const btCollisionObject* getBody0() const { return m_body0; }
+	SIMD_FORCE_INLINE const btCollisionObject* getBody1() const { return m_body1; }
 
-	void	setBodies(const btCollisionObject* body0,const btCollisionObject* body1)
+	void setBodies(const btCollisionObject* body0, const btCollisionObject* body1)
 	{
 		m_body0 = body0;
 		m_body1 = body1;
 	}
 
-	void clearUserCache(btManifoldPoint& pt);
+	void clearUserCache(btManifoldPoint & pt);
 
 #ifdef DEBUG_PERSISTENCY
-	void	DebugPersistency();
-#endif //
-	
-	SIMD_FORCE_INLINE int	getNumContacts() const { return m_cachedPoints;}
+	void DebugPersistency();
+#endif  //
+
+	SIMD_FORCE_INLINE int getNumContacts() const
+	{
+		return m_cachedPoints;
+	}
 	/// the setNumContacts API is usually not used, except when you gather/fill all contacts manually
 	void setNumContacts(int cachedPoints)
 	{
 		m_cachedPoints = cachedPoints;
 	}
 
-
 	SIMD_FORCE_INLINE const btManifoldPoint& getContactPoint(int index) const
 	{
 		btAssert(index < m_cachedPoints);
@@ -140,39 +140,36 @@ public:
 	}
 
 	///@todo: get this margin from the current physics / collision environment
-	btScalar	getContactBreakingThreshold() const;
+	btScalar getContactBreakingThreshold() const;
 
-	btScalar	getContactProcessingThreshold() const
+	btScalar getContactProcessingThreshold() const
 	{
 		return m_contactProcessingThreshold;
 	}
-	
+
 	void setContactBreakingThreshold(btScalar contactBreakingThreshold)
 	{
 		m_contactBreakingThreshold = contactBreakingThreshold;
 	}
 
-	void setContactProcessingThreshold(btScalar	contactProcessingThreshold)
+	void setContactProcessingThreshold(btScalar contactProcessingThreshold)
 	{
 		m_contactProcessingThreshold = contactProcessingThreshold;
 	}
-	
-	
-
 
 	int getCacheEntry(const btManifoldPoint& newPoint) const;
 
-	int addManifoldPoint( const btManifoldPoint& newPoint, bool isPredictive=false);
+	int addManifoldPoint(const btManifoldPoint& newPoint, bool isPredictive = false);
 
-	void removeContactPoint (int index)
+	void removeContactPoint(int index)
 	{
 		clearUserCache(m_pointCache[index]);
 
 		int lastUsedIndex = getNumContacts() - 1;
-//		m_pointCache[index] = m_pointCache[lastUsedIndex];
-		if(index != lastUsedIndex) 
+		//		m_pointCache[index] = m_pointCache[lastUsedIndex];
+		if (index != lastUsedIndex)
 		{
-			m_pointCache[index] = m_pointCache[lastUsedIndex]; 
+			m_pointCache[index] = m_pointCache[lastUsedIndex];
 			//get rid of duplicated userPersistentData pointer
 			m_pointCache[lastUsedIndex].m_userPersistentData = 0;
 			m_pointCache[lastUsedIndex].m_appliedImpulse = 0.f;
@@ -182,7 +179,7 @@ public:
 			m_pointCache[lastUsedIndex].m_lifeTime = 0;
 		}
 
-		btAssert(m_pointCache[lastUsedIndex].m_userPersistentData==0);
+		btAssert(m_pointCache[lastUsedIndex].m_userPersistentData == 0);
 		m_cachedPoints--;
 
 		if (gContactEndedCallback && m_cachedPoints == 0)
@@ -243,13 +240,12 @@ public:
 		return pt.m_distance1 <= getContactBreakingThreshold();
 	}
 	/// calculated new worldspace coordinates and depth, and reject points that exceed the collision margin
-	void	refreshContactPoints(  const btTransform& trA,const btTransform& trB);
+	void refreshContactPoints(const btTransform& trA, const btTransform& trB);
 
-	
-	SIMD_FORCE_INLINE	void	clearManifold()
+	SIMD_FORCE_INLINE void clearManifold()
 	{
 		int i;
-		for (i=0;i<m_cachedPoints;i++)
+		for (i = 0; i < m_cachedPoints; i++)
 		{
 			clearUserCache(m_pointCache[i]);
 		}
@@ -261,15 +257,13 @@ public:
 		m_cachedPoints = 0;
 	}
 
-	int	calculateSerializeBufferSize()	const;
-	const char*	serialize(const class btPersistentManifold* manifold, void* dataBuffer, class btSerializer* serializer) const;
+	int calculateSerializeBufferSize() const;
+	const char* serialize(const class btPersistentManifold* manifold, void* dataBuffer, class btSerializer* serializer) const;
 	void deSerialize(const struct btPersistentManifoldDoubleData* manifoldDataPtr);
 	void deSerialize(const struct btPersistentManifoldFloatData* manifoldDataPtr);
-
-
 };
 
-
+// clang-format off
 
 struct btPersistentManifoldDoubleData
 {
@@ -362,17 +356,14 @@ struct btPersistentManifoldFloatData
 	btCollisionObjectFloatData *m_body1;
 };
 
+// clang-format on
+
 #ifdef BT_USE_DOUBLE_PRECISION
-#define btPersistentManifoldData	btPersistentManifoldDoubleData
-#define btPersistentManifoldDataName	"btPersistentManifoldDoubleData"
+#define btPersistentManifoldData btPersistentManifoldDoubleData
+#define btPersistentManifoldDataName "btPersistentManifoldDoubleData"
 #else
-#define btPersistentManifoldData	btPersistentManifoldFloatData
-#define btPersistentManifoldDataName	"btPersistentManifoldFloatData"
-#endif //BT_USE_DOUBLE_PRECISION
+#define btPersistentManifoldData btPersistentManifoldFloatData
+#define btPersistentManifoldDataName "btPersistentManifoldFloatData"
+#endif  //BT_USE_DOUBLE_PRECISION
 
-
-
-
-
-
-#endif //BT_PERSISTENT_MANIFOLD_H
+#endif  //BT_PERSISTENT_MANIFOLD_H
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btPointCollector.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btPointCollector.h
index 18da171011a..0900eb6e85a 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btPointCollector.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btPointCollector.h
@@ -18,38 +18,33 @@ subject to the following restrictions:
 
 #include "btDiscreteCollisionDetectorInterface.h"
 
-
-
 struct btPointCollector : public btDiscreteCollisionDetectorInterface::Result
 {
-	
-	
 	btVector3 m_normalOnBInWorld;
 	btVector3 m_pointInWorld;
-	btScalar	m_distance;//negative means penetration
+	btScalar m_distance;  //negative means penetration
 
-	bool	m_hasResult;
+	bool m_hasResult;
 
-	btPointCollector () 
-		: m_distance(btScalar(BT_LARGE_FLOAT)),m_hasResult(false)
+	btPointCollector()
+		: m_distance(btScalar(BT_LARGE_FLOAT)), m_hasResult(false)
 	{
 	}
 
-	virtual void setShapeIdentifiersA(int partId0,int index0)
+	virtual void setShapeIdentifiersA(int partId0, int index0)
 	{
 		(void)partId0;
 		(void)index0;
-			
 	}
-	virtual void setShapeIdentifiersB(int partId1,int index1)
+	virtual void setShapeIdentifiersB(int partId1, int index1)
 	{
 		(void)partId1;
 		(void)index1;
 	}
 
-	virtual void addContactPoint(const btVector3& normalOnBInWorld,const btVector3& pointInWorld,btScalar depth)
+	virtual void addContactPoint(const btVector3& normalOnBInWorld, const btVector3& pointInWorld, btScalar depth)
 	{
-		if (depth< m_distance)
+		if (depth < m_distance)
 		{
 			m_hasResult = true;
 			m_normalOnBInWorld = normalOnBInWorld;
@@ -60,5 +55,4 @@ struct btPointCollector : public btDiscreteCollisionDetectorInterface::Result
 	}
 };
 
-#endif //BT_POINT_COLLECTOR_H
-
+#endif  //BT_POINT_COLLECTOR_H
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btPolyhedralContactClipping.cpp b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btPolyhedralContactClipping.cpp
index ea380bc5f1b..9d1836037d5 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btPolyhedralContactClipping.cpp
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btPolyhedralContactClipping.cpp
@@ -13,45 +13,42 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 ///This file was written by Erwin Coumans
 ///Separating axis rest based on work from Pierre Terdiman, see
 ///And contact clipping based on work from Simon Hobbs
 
-
 #include "btPolyhedralContactClipping.h"
 #include "BulletCollision/CollisionShapes/btConvexPolyhedron.h"
 
-#include <float.h> //for FLT_MAX
+#include <float.h>  //for FLT_MAX
 
-int gExpectedNbTests=0;
+int gExpectedNbTests = 0;
 int gActualNbTests = 0;
 bool gUseInternalObject = true;
 
 // Clips a face to the back of a plane
-void btPolyhedralContactClipping::clipFace(const btVertexArray& pVtxIn, btVertexArray& ppVtxOut, const btVector3& planeNormalWS,btScalar planeEqWS)
+void btPolyhedralContactClipping::clipFace(const btVertexArray& pVtxIn, btVertexArray& ppVtxOut, const btVector3& planeNormalWS, btScalar planeEqWS)
 {
-	
 	int ve;
 	btScalar ds, de;
 	int numVerts = pVtxIn.size();
 	if (numVerts < 2)
 		return;
 
-	btVector3 firstVertex=pVtxIn[pVtxIn.size()-1];
+	btVector3 firstVertex = pVtxIn[pVtxIn.size() - 1];
 	btVector3 endVertex = pVtxIn[0];
-	
-	ds = planeNormalWS.dot(firstVertex)+planeEqWS;
+
+	ds = planeNormalWS.dot(firstVertex) + planeEqWS;
 
 	for (ve = 0; ve < numVerts; ve++)
 	{
-		endVertex=pVtxIn[ve];
+		endVertex = pVtxIn[ve];
 
-		de = planeNormalWS.dot(endVertex)+planeEqWS;
+		de = planeNormalWS.dot(endVertex) + planeEqWS;
 
-		if (ds<0)
+		if (ds < 0)
 		{
-			if (de<0)
+			if (de < 0)
 			{
 				// Start < 0, end < 0, so output endVertex
 				ppVtxOut.push_back(endVertex);
@@ -59,15 +56,15 @@ void btPolyhedralContactClipping::clipFace(const btVertexArray& pVtxIn, btVertex
 			else
 			{
 				// Start < 0, end >= 0, so output intersection
-				ppVtxOut.push_back( 	firstVertex.lerp(endVertex,btScalar(ds * 1.f/(ds - de))));
+				ppVtxOut.push_back(firstVertex.lerp(endVertex, btScalar(ds * 1.f / (ds - de))));
 			}
 		}
 		else
 		{
-			if (de<0)
+			if (de < 0)
 			{
 				// Start >= 0, end < 0 so output intersection and end
-				ppVtxOut.push_back(firstVertex.lerp(endVertex,btScalar(ds * 1.f/(ds - de))));
+				ppVtxOut.push_back(firstVertex.lerp(endVertex, btScalar(ds * 1.f / (ds - de))));
 				ppVtxOut.push_back(endVertex);
 			}
 		}
@@ -76,47 +73,44 @@ void btPolyhedralContactClipping::clipFace(const btVertexArray& pVtxIn, btVertex
 	}
 }
 
-
-static bool TestSepAxis(const btConvexPolyhedron& hullA, const btConvexPolyhedron& hullB, const btTransform& transA,const btTransform& transB, const btVector3& sep_axis, btScalar& depth, btVector3& witnessPointA, btVector3& witnessPointB)
+static bool TestSepAxis(const btConvexPolyhedron& hullA, const btConvexPolyhedron& hullB, const btTransform& transA, const btTransform& transB, const btVector3& sep_axis, btScalar& depth, btVector3& witnessPointA, btVector3& witnessPointB)
 {
-	btScalar Min0,Max0;
-	btScalar Min1,Max1;
-	btVector3 witnesPtMinA,witnesPtMaxA;
-	btVector3 witnesPtMinB,witnesPtMaxB;
+	btScalar Min0, Max0;
+	btScalar Min1, Max1;
+	btVector3 witnesPtMinA, witnesPtMaxA;
+	btVector3 witnesPtMinB, witnesPtMaxB;
 
-	hullA.project(transA,sep_axis, Min0, Max0,witnesPtMinA,witnesPtMaxA);
-	hullB.project(transB, sep_axis, Min1, Max1,witnesPtMinB,witnesPtMaxB);
+	hullA.project(transA, sep_axis, Min0, Max0, witnesPtMinA, witnesPtMaxA);
+	hullB.project(transB, sep_axis, Min1, Max1, witnesPtMinB, witnesPtMaxB);
 
-	if(Max0<Min1 || Max1<Min0)
+	if (Max0 < Min1 || Max1 < Min0)
 		return false;
 
 	btScalar d0 = Max0 - Min1;
-	btAssert(d0>=0.0f);
+	btAssert(d0 >= 0.0f);
 	btScalar d1 = Max1 - Min0;
-	btAssert(d1>=0.0f);
-	if (d0<d1)
+	btAssert(d1 >= 0.0f);
+	if (d0 < d1)
 	{
 		depth = d0;
 		witnessPointA = witnesPtMaxA;
 		witnessPointB = witnesPtMinB;
-
-	} else
+	}
+	else
 	{
 		depth = d1;
 		witnessPointA = witnesPtMinA;
 		witnessPointB = witnesPtMaxB;
 	}
-	
+
 	return true;
 }
 
-
-
-static int gActualSATPairTests=0;
+static int gActualSATPairTests = 0;
 
 inline bool IsAlmostZero(const btVector3& v)
 {
-	if(btFabs(v.x())>1e-6 || btFabs(v.y())>1e-6 || btFabs(v.z())>1e-6)	return false;
+	if (btFabs(v.x()) > 1e-6 || btFabs(v.y()) > 1e-6 || btFabs(v.z()) > 1e-6) return false;
 	return true;
 }
 
@@ -125,9 +119,9 @@ inline bool IsAlmostZero(const btVector3& v)
 inline void BoxSupport(const btScalar extents[3], const btScalar sv[3], btScalar p[3])
 {
 	// This version is ~11.000 cycles (4%) faster overall in one of the tests.
-//	IR(p[0]) = IR(extents[0])|(IR(sv[0])&SIGN_BITMASK);
-//	IR(p[1]) = IR(extents[1])|(IR(sv[1])&SIGN_BITMASK);
-//	IR(p[2]) = IR(extents[2])|(IR(sv[2])&SIGN_BITMASK);
+	//	IR(p[0]) = IR(extents[0])|(IR(sv[0])&SIGN_BITMASK);
+	//	IR(p[1]) = IR(extents[1])|(IR(sv[1])&SIGN_BITMASK);
+	//	IR(p[2]) = IR(extents[2])|(IR(sv[2])&SIGN_BITMASK);
 	p[0] = sv[0] < 0.0f ? -extents[0] : extents[0];
 	p[1] = sv[1] < 0.0f ? -extents[1] : extents[1];
 	p[2] = sv[2] < 0.0f ? -extents[2] : extents[2];
@@ -140,90 +134,94 @@ void InverseTransformPoint3x3(btVector3& out, const btVector3& in, const btTrans
 	const btVector3& r1 = rot[1];
 	const btVector3& r2 = rot[2];
 
-	const btScalar x = r0.x()*in.x() + r1.x()*in.y() + r2.x()*in.z();
-	const btScalar y = r0.y()*in.x() + r1.y()*in.y() + r2.y()*in.z();
-	const btScalar z = r0.z()*in.x() + r1.z()*in.y() + r2.z()*in.z();
+	const btScalar x = r0.x() * in.x() + r1.x() * in.y() + r2.x() * in.z();
+	const btScalar y = r0.y() * in.x() + r1.y() * in.y() + r2.y() * in.z();
+	const btScalar z = r0.z() * in.x() + r1.z() * in.y() + r2.z() * in.z();
 
 	out.setValue(x, y, z);
 }
 
- bool TestInternalObjects( const btTransform& trans0, const btTransform& trans1, const btVector3& delta_c, const btVector3& axis, const btConvexPolyhedron& convex0, const btConvexPolyhedron& convex1, btScalar dmin)
+bool TestInternalObjects(const btTransform& trans0, const btTransform& trans1, const btVector3& delta_c, const btVector3& axis, const btConvexPolyhedron& convex0, const btConvexPolyhedron& convex1, btScalar dmin)
 {
 	const btScalar dp = delta_c.dot(axis);
 
 	btVector3 localAxis0;
-	InverseTransformPoint3x3(localAxis0, axis,trans0);
+	InverseTransformPoint3x3(localAxis0, axis, trans0);
 	btVector3 localAxis1;
-	InverseTransformPoint3x3(localAxis1, axis,trans1);
+	InverseTransformPoint3x3(localAxis1, axis, trans1);
 
 	btScalar p0[3];
 	BoxSupport(convex0.m_extents, localAxis0, p0);
 	btScalar p1[3];
 	BoxSupport(convex1.m_extents, localAxis1, p1);
 
-	const btScalar Radius0 = p0[0]*localAxis0.x() + p0[1]*localAxis0.y() + p0[2]*localAxis0.z();
-	const btScalar Radius1 = p1[0]*localAxis1.x() + p1[1]*localAxis1.y() + p1[2]*localAxis1.z();
+	const btScalar Radius0 = p0[0] * localAxis0.x() + p0[1] * localAxis0.y() + p0[2] * localAxis0.z();
+	const btScalar Radius1 = p1[0] * localAxis1.x() + p1[1] * localAxis1.y() + p1[2] * localAxis1.z();
 
-	const btScalar MinRadius = Radius0>convex0.m_radius ? Radius0 : convex0.m_radius;
-	const btScalar MaxRadius = Radius1>convex1.m_radius ? Radius1 : convex1.m_radius;
+	const btScalar MinRadius = Radius0 > convex0.m_radius ? Radius0 : convex0.m_radius;
+	const btScalar MaxRadius = Radius1 > convex1.m_radius ? Radius1 : convex1.m_radius;
 
 	const btScalar MinMaxRadius = MaxRadius + MinRadius;
 	const btScalar d0 = MinMaxRadius + dp;
 	const btScalar d1 = MinMaxRadius - dp;
 
-	const btScalar depth = d0<d1 ? d0:d1;
-	if(depth>dmin)
+	const btScalar depth = d0 < d1 ? d0 : d1;
+	if (depth > dmin)
 		return false;
 	return true;
 }
-#endif //TEST_INTERNAL_OBJECTS
+#endif  //TEST_INTERNAL_OBJECTS
 
- 
- 
- SIMD_FORCE_INLINE void btSegmentsClosestPoints(
+SIMD_FORCE_INLINE void btSegmentsClosestPoints(
 	btVector3& ptsVector,
 	btVector3& offsetA,
 	btVector3& offsetB,
 	btScalar& tA, btScalar& tB,
 	const btVector3& translation,
 	const btVector3& dirA, btScalar hlenA,
-	const btVector3& dirB, btScalar hlenB )
+	const btVector3& dirB, btScalar hlenB)
 {
 	// compute the parameters of the closest points on each line segment
 
-	btScalar dirA_dot_dirB = btDot(dirA,dirB);
-	btScalar dirA_dot_trans = btDot(dirA,translation);
-	btScalar dirB_dot_trans = btDot(dirB,translation);
+	btScalar dirA_dot_dirB = btDot(dirA, dirB);
+	btScalar dirA_dot_trans = btDot(dirA, translation);
+	btScalar dirB_dot_trans = btDot(dirB, translation);
 
 	btScalar denom = 1.0f - dirA_dot_dirB * dirA_dot_dirB;
 
-	if ( denom == 0.0f ) {
+	if (denom == 0.0f)
+	{
 		tA = 0.0f;
-	} else {
-		tA = ( dirA_dot_trans - dirB_dot_trans * dirA_dot_dirB ) / denom;
-		if ( tA < -hlenA )
+	}
+	else
+	{
+		tA = (dirA_dot_trans - dirB_dot_trans * dirA_dot_dirB) / denom;
+		if (tA < -hlenA)
 			tA = -hlenA;
-		else if ( tA > hlenA )
+		else if (tA > hlenA)
 			tA = hlenA;
 	}
 
 	tB = tA * dirA_dot_dirB - dirB_dot_trans;
 
-	if ( tB < -hlenB ) {
+	if (tB < -hlenB)
+	{
 		tB = -hlenB;
 		tA = tB * dirA_dot_dirB + dirA_dot_trans;
 
-		if ( tA < -hlenA )
+		if (tA < -hlenA)
 			tA = -hlenA;
-		else if ( tA > hlenA )
+		else if (tA > hlenA)
 			tA = hlenA;
-	} else if ( tB > hlenB ) {
+	}
+	else if (tB > hlenB)
+	{
 		tB = hlenB;
 		tA = tB * dirA_dot_dirB + dirA_dot_trans;
 
-		if ( tA < -hlenA )
+		if (tA < -hlenA)
 			tA = -hlenA;
-		else if ( tA > hlenA )
+		else if (tA > hlenA)
 			tA = hlenA;
 	}
 
@@ -235,44 +233,42 @@ void InverseTransformPoint3x3(btVector3& out, const btVector3& in, const btTrans
 	ptsVector = translation - offsetA + offsetB;
 }
 
-
-
-bool btPolyhedralContactClipping::findSeparatingAxis(	const btConvexPolyhedron& hullA, const btConvexPolyhedron& hullB, const btTransform& transA,const btTransform& transB, btVector3& sep, btDiscreteCollisionDetectorInterface::Result& resultOut)
+bool btPolyhedralContactClipping::findSeparatingAxis(const btConvexPolyhedron& hullA, const btConvexPolyhedron& hullB, const btTransform& transA, const btTransform& transB, btVector3& sep, btDiscreteCollisionDetectorInterface::Result& resultOut)
 {
 	gActualSATPairTests++;
 
-//#ifdef TEST_INTERNAL_OBJECTS
+	//#ifdef TEST_INTERNAL_OBJECTS
 	const btVector3 c0 = transA * hullA.m_localCenter;
 	const btVector3 c1 = transB * hullB.m_localCenter;
 	const btVector3 DeltaC2 = c0 - c1;
-//#endif
+	//#endif
 
 	btScalar dmin = FLT_MAX;
-	int curPlaneTests=0;
+	int curPlaneTests = 0;
 
 	int numFacesA = hullA.m_faces.size();
 	// Test normals from hullA
-	for(int i=0;i<numFacesA;i++)
+	for (int i = 0; i < numFacesA; i++)
 	{
 		const btVector3 Normal(hullA.m_faces[i].m_plane[0], hullA.m_faces[i].m_plane[1], hullA.m_faces[i].m_plane[2]);
 		btVector3 faceANormalWS = transA.getBasis() * Normal;
-		if (DeltaC2.dot(faceANormalWS)<0)
-			faceANormalWS*=-1.f;
+		if (DeltaC2.dot(faceANormalWS) < 0)
+			faceANormalWS *= -1.f;
 
 		curPlaneTests++;
 #ifdef TEST_INTERNAL_OBJECTS
 		gExpectedNbTests++;
-		if(gUseInternalObject && !TestInternalObjects(transA,transB, DeltaC2, faceANormalWS, hullA, hullB, dmin))
+		if (gUseInternalObject && !TestInternalObjects(transA, transB, DeltaC2, faceANormalWS, hullA, hullB, dmin))
 			continue;
 		gActualNbTests++;
 #endif
 
 		btScalar d;
-		btVector3 wA,wB;
-		if(!TestSepAxis( hullA, hullB, transA,transB, faceANormalWS, d,wA,wB))
+		btVector3 wA, wB;
+		if (!TestSepAxis(hullA, hullB, transA, transB, faceANormalWS, d, wA, wB))
 			return false;
 
-		if(d<dmin)
+		if (d < dmin)
 		{
 			dmin = d;
 			sep = faceANormalWS;
@@ -281,92 +277,89 @@ bool btPolyhedralContactClipping::findSeparatingAxis(	const btConvexPolyhedron&
 
 	int numFacesB = hullB.m_faces.size();
 	// Test normals from hullB
-	for(int i=0;i<numFacesB;i++)
+	for (int i = 0; i < numFacesB; i++)
 	{
 		const btVector3 Normal(hullB.m_faces[i].m_plane[0], hullB.m_faces[i].m_plane[1], hullB.m_faces[i].m_plane[2]);
 		btVector3 WorldNormal = transB.getBasis() * Normal;
-		if (DeltaC2.dot(WorldNormal)<0)
-			WorldNormal *=-1.f;
+		if (DeltaC2.dot(WorldNormal) < 0)
+			WorldNormal *= -1.f;
 
 		curPlaneTests++;
 #ifdef TEST_INTERNAL_OBJECTS
 		gExpectedNbTests++;
-		if(gUseInternalObject && !TestInternalObjects(transA,transB,DeltaC2, WorldNormal, hullA, hullB, dmin))
+		if (gUseInternalObject && !TestInternalObjects(transA, transB, DeltaC2, WorldNormal, hullA, hullB, dmin))
 			continue;
 		gActualNbTests++;
 #endif
 
 		btScalar d;
-		btVector3 wA,wB;
-		if(!TestSepAxis(hullA, hullB,transA,transB, WorldNormal,d,wA,wB))
+		btVector3 wA, wB;
+		if (!TestSepAxis(hullA, hullB, transA, transB, WorldNormal, d, wA, wB))
 			return false;
 
-		if(d<dmin)
+		if (d < dmin)
 		{
 			dmin = d;
 			sep = WorldNormal;
 		}
 	}
 
-	btVector3 edgeAstart,edgeAend,edgeBstart,edgeBend;
-	int edgeA=-1;
-	int edgeB=-1;
+	btVector3 edgeAstart, edgeAend, edgeBstart, edgeBend;
+	int edgeA = -1;
+	int edgeB = -1;
 	btVector3 worldEdgeA;
 	btVector3 worldEdgeB;
-	btVector3 witnessPointA(0,0,0),witnessPointB(0,0,0);
-	
+	btVector3 witnessPointA(0, 0, 0), witnessPointB(0, 0, 0);
 
 	int curEdgeEdge = 0;
 	// Test edges
-	for(int e0=0;e0<hullA.m_uniqueEdges.size();e0++)
+	for (int e0 = 0; e0 < hullA.m_uniqueEdges.size(); e0++)
 	{
 		const btVector3 edge0 = hullA.m_uniqueEdges[e0];
 		const btVector3 WorldEdge0 = transA.getBasis() * edge0;
-		for(int e1=0;e1<hullB.m_uniqueEdges.size();e1++)
+		for (int e1 = 0; e1 < hullB.m_uniqueEdges.size(); e1++)
 		{
 			const btVector3 edge1 = hullB.m_uniqueEdges[e1];
 			const btVector3 WorldEdge1 = transB.getBasis() * edge1;
 
 			btVector3 Cross = WorldEdge0.cross(WorldEdge1);
 			curEdgeEdge++;
-			if(!IsAlmostZero(Cross))
+			if (!IsAlmostZero(Cross))
 			{
 				Cross = Cross.normalize();
-				if (DeltaC2.dot(Cross)<0)
+				if (DeltaC2.dot(Cross) < 0)
 					Cross *= -1.f;
 
-
 #ifdef TEST_INTERNAL_OBJECTS
 				gExpectedNbTests++;
-				if(gUseInternalObject && !TestInternalObjects(transA,transB,DeltaC2, Cross, hullA, hullB, dmin))
+				if (gUseInternalObject && !TestInternalObjects(transA, transB, DeltaC2, Cross, hullA, hullB, dmin))
 					continue;
 				gActualNbTests++;
 #endif
 
 				btScalar dist;
-				btVector3 wA,wB;
-				if(!TestSepAxis( hullA, hullB, transA,transB, Cross, dist,wA,wB))
+				btVector3 wA, wB;
+				if (!TestSepAxis(hullA, hullB, transA, transB, Cross, dist, wA, wB))
 					return false;
 
-				if(dist<dmin)
+				if (dist < dmin)
 				{
 					dmin = dist;
 					sep = Cross;
-					edgeA=e0;
-					edgeB=e1;
+					edgeA = e0;
+					edgeB = e1;
 					worldEdgeA = WorldEdge0;
 					worldEdgeB = WorldEdge1;
-					witnessPointA=wA;
-					witnessPointB=wB;
+					witnessPointA = wA;
+					witnessPointB = wB;
 				}
 			}
 		}
-
 	}
 
-	if (edgeA>=0&&edgeB>=0)
+	if (edgeA >= 0 && edgeB >= 0)
 	{
-//		printf("edge-edge\n");
+		//		printf("edge-edge\n");
 		//add an edge-edge contact
 
 		btVector3 ptsVector;
@@ -375,57 +368,55 @@ bool btPolyhedralContactClipping::findSeparatingAxis(	const btConvexPolyhedron&
 		btScalar tA;
 		btScalar tB;
 
-		btVector3 translation = witnessPointB-witnessPointA;
+		btVector3 translation = witnessPointB - witnessPointA;
 
 		btVector3 dirA = worldEdgeA;
 		btVector3 dirB = worldEdgeB;
-		
+
 		btScalar hlenB = 1e30f;
 		btScalar hlenA = 1e30f;
 
-		btSegmentsClosestPoints(ptsVector,offsetA,offsetB,tA,tB,
-			translation,
-			dirA, hlenA,
-			dirB,hlenB);
+		btSegmentsClosestPoints(ptsVector, offsetA, offsetB, tA, tB,
+								translation,
+								dirA, hlenA,
+								dirB, hlenB);
 
 		btScalar nlSqrt = ptsVector.length2();
-		if (nlSqrt>SIMD_EPSILON)
+		if (nlSqrt > SIMD_EPSILON)
 		{
 			btScalar nl = btSqrt(nlSqrt);
-			ptsVector *= 1.f/nl;
-			if (ptsVector.dot(DeltaC2)<0.f)
+			ptsVector *= 1.f / nl;
+			if (ptsVector.dot(DeltaC2) < 0.f)
 			{
-				ptsVector*=-1.f;
+				ptsVector *= -1.f;
 			}
 			btVector3 ptOnB = witnessPointB + offsetB;
 			btScalar distance = nl;
-			resultOut.addContactPoint(ptsVector, ptOnB,-distance);
+			resultOut.addContactPoint(ptsVector, ptOnB, -distance);
 		}
-
 	}
 
-
-	if((DeltaC2.dot(sep))<0.0f)
+	if ((DeltaC2.dot(sep)) < 0.0f)
 		sep = -sep;
 
 	return true;
 }
 
-void	btPolyhedralContactClipping::clipFaceAgainstHull(const btVector3& separatingNormal, const btConvexPolyhedron& hullA,  const btTransform& transA, btVertexArray& worldVertsB1,btVertexArray& worldVertsB2, const btScalar minDist, btScalar maxDist,btDiscreteCollisionDetectorInterface::Result& resultOut)
+void btPolyhedralContactClipping::clipFaceAgainstHull(const btVector3& separatingNormal, const btConvexPolyhedron& hullA, const btTransform& transA, btVertexArray& worldVertsB1, btVertexArray& worldVertsB2, const btScalar minDist, btScalar maxDist, btDiscreteCollisionDetectorInterface::Result& resultOut)
 {
 	worldVertsB2.resize(0);
 	btVertexArray* pVtxIn = &worldVertsB1;
 	btVertexArray* pVtxOut = &worldVertsB2;
 	pVtxOut->reserve(pVtxIn->size());
 
-	int closestFaceA=-1;
+	int closestFaceA = -1;
 	{
 		btScalar dmin = FLT_MAX;
-		for(int face=0;face<hullA.m_faces.size();face++)
+		for (int face = 0; face < hullA.m_faces.size(); face++)
 		{
 			const btVector3 Normal(hullA.m_faces[face].m_plane[0], hullA.m_faces[face].m_plane[1], hullA.m_faces[face].m_plane[2]);
 			const btVector3 faceANormalWS = transA.getBasis() * Normal;
-		
+
 			btScalar d = faceANormalWS.dot(separatingNormal);
 			if (d < dmin)
 			{
@@ -434,69 +425,66 @@ void	btPolyhedralContactClipping::clipFaceAgainstHull(const btVector3& separatin
 			}
 		}
 	}
-	if (closestFaceA<0)
+	if (closestFaceA < 0)
 		return;
 
 	const btFace& polyA = hullA.m_faces[closestFaceA];
 
-		// clip polygon to back of planes of all faces of hull A that are adjacent to witness face
+	// clip polygon to back of planes of all faces of hull A that are adjacent to witness face
 	int numVerticesA = polyA.m_indices.size();
-	for(int e0=0;e0<numVerticesA;e0++)
+	for (int e0 = 0; e0 < numVerticesA; e0++)
 	{
 		const btVector3& a = hullA.m_vertices[polyA.m_indices[e0]];
-		const btVector3& b = hullA.m_vertices[polyA.m_indices[(e0+1)%numVerticesA]];
+		const btVector3& b = hullA.m_vertices[polyA.m_indices[(e0 + 1) % numVerticesA]];
 		const btVector3 edge0 = a - b;
 		const btVector3 WorldEdge0 = transA.getBasis() * edge0;
-		btVector3 worldPlaneAnormal1 = transA.getBasis()* btVector3(polyA.m_plane[0],polyA.m_plane[1],polyA.m_plane[2]);
+		btVector3 worldPlaneAnormal1 = transA.getBasis() * btVector3(polyA.m_plane[0], polyA.m_plane[1], polyA.m_plane[2]);
 
-		btVector3 planeNormalWS1 = -WorldEdge0.cross(worldPlaneAnormal1);//.cross(WorldEdge0);
-		btVector3 worldA1 = transA*a;
+		btVector3 planeNormalWS1 = -WorldEdge0.cross(worldPlaneAnormal1);  //.cross(WorldEdge0);
+		btVector3 worldA1 = transA * a;
 		btScalar planeEqWS1 = -worldA1.dot(planeNormalWS1);
-		
+
 //int otherFace=0;
 #ifdef BLA1
 		int otherFace = polyA.m_connectedFaces[e0];
-		btVector3 localPlaneNormal (hullA.m_faces[otherFace].m_plane[0],hullA.m_faces[otherFace].m_plane[1],hullA.m_faces[otherFace].m_plane[2]);
+		btVector3 localPlaneNormal(hullA.m_faces[otherFace].m_plane[0], hullA.m_faces[otherFace].m_plane[1], hullA.m_faces[otherFace].m_plane[2]);
 		btScalar localPlaneEq = hullA.m_faces[otherFace].m_plane[3];
 
-		btVector3 planeNormalWS = transA.getBasis()*localPlaneNormal;
-		btScalar planeEqWS=localPlaneEq-planeNormalWS.dot(transA.getOrigin());
-#else 
+		btVector3 planeNormalWS = transA.getBasis() * localPlaneNormal;
+		btScalar planeEqWS = localPlaneEq - planeNormalWS.dot(transA.getOrigin());
+#else
 		btVector3 planeNormalWS = planeNormalWS1;
-		btScalar planeEqWS=planeEqWS1;
-		
+		btScalar planeEqWS = planeEqWS1;
+
 #endif
 		//clip face
 
-		clipFace(*pVtxIn, *pVtxOut,planeNormalWS,planeEqWS);
-		btSwap(pVtxIn,pVtxOut);
+		clipFace(*pVtxIn, *pVtxOut, planeNormalWS, planeEqWS);
+		btSwap(pVtxIn, pVtxOut);
 		pVtxOut->resize(0);
 	}
 
-
-
-//#define ONLY_REPORT_DEEPEST_POINT
+	//#define ONLY_REPORT_DEEPEST_POINT
 
 	btVector3 point;
-	
 
 	// only keep points that are behind the witness face
 	{
-		btVector3 localPlaneNormal (polyA.m_plane[0],polyA.m_plane[1],polyA.m_plane[2]);
+		btVector3 localPlaneNormal(polyA.m_plane[0], polyA.m_plane[1], polyA.m_plane[2]);
 		btScalar localPlaneEq = polyA.m_plane[3];
-		btVector3 planeNormalWS = transA.getBasis()*localPlaneNormal;
-		btScalar planeEqWS=localPlaneEq-planeNormalWS.dot(transA.getOrigin());
-		for (int i=0;i<pVtxIn->size();i++)
+		btVector3 planeNormalWS = transA.getBasis() * localPlaneNormal;
+		btScalar planeEqWS = localPlaneEq - planeNormalWS.dot(transA.getOrigin());
+		for (int i = 0; i < pVtxIn->size(); i++)
 		{
 			btVector3 vtx = pVtxIn->at(i);
-			btScalar depth = planeNormalWS.dot(vtx)+planeEqWS;
-			if (depth <=minDist)
+			btScalar depth = planeNormalWS.dot(vtx) + planeEqWS;
+			if (depth <= minDist)
 			{
-//				printf("clamped: depth=%f to minDist=%f\n",depth,minDist);
+				//				printf("clamped: depth=%f to minDist=%f\n",depth,minDist);
 				depth = minDist;
 			}
 
-			if (depth <=maxDist)
+			if (depth <= maxDist)
 			{
 				btVector3 point = pVtxIn->at(i);
 #ifdef ONLY_REPORT_DEEPEST_POINT
@@ -507,40 +495,32 @@ void	btPolyhedralContactClipping::clipFaceAgainstHull(const btVector3& separatin
 				{
 					printf("error in btPolyhedralContactClipping depth = %f\n", depth);
 					printf("likely wrong separatingNormal passed in\n");
-				} 
-#endif				
-				resultOut.addContactPoint(separatingNormal,point,depth);
+				}
+#endif
+				resultOut.addContactPoint(separatingNormal, point, depth);
 #endif
 			}
 		}
 	}
 #ifdef ONLY_REPORT_DEEPEST_POINT
-	if (curMaxDist<maxDist)
+	if (curMaxDist < maxDist)
 	{
-		resultOut.addContactPoint(separatingNormal,point,curMaxDist);
+		resultOut.addContactPoint(separatingNormal, point, curMaxDist);
 	}
-#endif //ONLY_REPORT_DEEPEST_POINT
-
+#endif  //ONLY_REPORT_DEEPEST_POINT
 }
 
-
-
-
-
-void	btPolyhedralContactClipping::clipHullAgainstHull(const btVector3& separatingNormal1, const btConvexPolyhedron& hullA, const btConvexPolyhedron& hullB, const btTransform& transA,const btTransform& transB, const btScalar minDist, btScalar maxDist,btVertexArray& worldVertsB1,btVertexArray& worldVertsB2,btDiscreteCollisionDetectorInterface::Result& resultOut)
+void btPolyhedralContactClipping::clipHullAgainstHull(const btVector3& separatingNormal1, const btConvexPolyhedron& hullA, const btConvexPolyhedron& hullB, const btTransform& transA, const btTransform& transB, const btScalar minDist, btScalar maxDist, btVertexArray& worldVertsB1, btVertexArray& worldVertsB2, btDiscreteCollisionDetectorInterface::Result& resultOut)
 {
-
 	btVector3 separatingNormal = separatingNormal1.normalized();
-//	const btVector3 c0 = transA * hullA.m_localCenter;
-//	const btVector3 c1 = transB * hullB.m_localCenter;
+	//	const btVector3 c0 = transA * hullA.m_localCenter;
+	//	const btVector3 c1 = transB * hullB.m_localCenter;
 	//const btVector3 DeltaC2 = c0 - c1;
 
-
-
-	int closestFaceB=-1;
+	int closestFaceB = -1;
 	btScalar dmax = -FLT_MAX;
 	{
-		for(int face=0;face<hullB.m_faces.size();face++)
+		for (int face = 0; face < hullB.m_faces.size(); face++)
 		{
 			const btVector3 Normal(hullB.m_faces[face].m_plane[0], hullB.m_faces[face].m_plane[1], hullB.m_faces[face].m_plane[2]);
 			const btVector3 WorldNormal = transB.getBasis() * Normal;
@@ -553,18 +533,16 @@ void	btPolyhedralContactClipping::clipHullAgainstHull(const btVector3& separatin
 		}
 	}
 	worldVertsB1.resize(0);
-				{
-					const btFace& polyB = hullB.m_faces[closestFaceB];
-					const int numVertices = polyB.m_indices.size();
-					for(int e0=0;e0<numVertices;e0++)
-					{
-						const btVector3& b = hullB.m_vertices[polyB.m_indices[e0]];
-						worldVertsB1.push_back(transB*b);
-					}
-				}
-
-	
-	if (closestFaceB>=0)
-		clipFaceAgainstHull(separatingNormal, hullA, transA,worldVertsB1, worldVertsB2,minDist, maxDist,resultOut);
+	{
+		const btFace& polyB = hullB.m_faces[closestFaceB];
+		const int numVertices = polyB.m_indices.size();
+		for (int e0 = 0; e0 < numVertices; e0++)
+		{
+			const btVector3& b = hullB.m_vertices[polyB.m_indices[e0]];
+			worldVertsB1.push_back(transB * b);
+		}
+	}
 
+	if (closestFaceB >= 0)
+		clipFaceAgainstHull(separatingNormal, hullA, transA, worldVertsB1, worldVertsB2, minDist, maxDist, resultOut);
 }
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btPolyhedralContactClipping.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btPolyhedralContactClipping.h
index 30e3db687b4..328f6424bc8 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btPolyhedralContactClipping.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btPolyhedralContactClipping.h
@@ -13,14 +13,11 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 ///This file was written by Erwin Coumans
 
-
 #ifndef BT_POLYHEDRAL_CONTACT_CLIPPING_H
 #define BT_POLYHEDRAL_CONTACT_CLIPPING_H
 
-
 #include "LinearMath/btAlignedObjectArray.h"
 #include "LinearMath/btTransform.h"
 #include "btDiscreteCollisionDetectorInterface.h"
@@ -32,18 +29,14 @@ typedef btAlignedObjectArray<btVector3> btVertexArray;
 // Clips a face to the back of a plane
 struct btPolyhedralContactClipping
 {
+	static void clipHullAgainstHull(const btVector3& separatingNormal1, const btConvexPolyhedron& hullA, const btConvexPolyhedron& hullB, const btTransform& transA, const btTransform& transB, const btScalar minDist, btScalar maxDist, btVertexArray& worldVertsB1, btVertexArray& worldVertsB2, btDiscreteCollisionDetectorInterface::Result& resultOut);
 
-	static void	clipHullAgainstHull(const btVector3& separatingNormal1, const btConvexPolyhedron& hullA, const btConvexPolyhedron& hullB, const btTransform& transA,const btTransform& transB, const btScalar minDist, btScalar maxDist,btVertexArray& worldVertsB1,btVertexArray& worldVertsB2,btDiscreteCollisionDetectorInterface::Result& resultOut);
+	static void clipFaceAgainstHull(const btVector3& separatingNormal, const btConvexPolyhedron& hullA, const btTransform& transA, btVertexArray& worldVertsB1, btVertexArray& worldVertsB2, const btScalar minDist, btScalar maxDist, btDiscreteCollisionDetectorInterface::Result& resultOut);
 
-	static void	clipFaceAgainstHull(const btVector3& separatingNormal, const btConvexPolyhedron& hullA,  const btTransform& transA, btVertexArray& worldVertsB1,btVertexArray& worldVertsB2, const btScalar minDist, btScalar maxDist,btDiscreteCollisionDetectorInterface::Result& resultOut);
-
-
-	static bool findSeparatingAxis(	const btConvexPolyhedron& hullA, const btConvexPolyhedron& hullB, const btTransform& transA,const btTransform& transB, btVector3& sep, btDiscreteCollisionDetectorInterface::Result& resultOut);
+	static bool findSeparatingAxis(const btConvexPolyhedron& hullA, const btConvexPolyhedron& hullB, const btTransform& transA, const btTransform& transB, btVector3& sep, btDiscreteCollisionDetectorInterface::Result& resultOut);
 
 	///the clipFace method is used internally
-	static void clipFace(const btVertexArray& pVtxIn, btVertexArray& ppVtxOut, const btVector3& planeNormalWS,btScalar planeEqWS);
-
+	static void clipFace(const btVertexArray& pVtxIn, btVertexArray& ppVtxOut, const btVector3& planeNormalWS, btScalar planeEqWS);
 };
 
-#endif // BT_POLYHEDRAL_CONTACT_CLIPPING_H
-
+#endif  // BT_POLYHEDRAL_CONTACT_CLIPPING_H
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btRaycastCallback.cpp b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btRaycastCallback.cpp
index 786efd18200..3d11e5bce5f 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btRaycastCallback.cpp
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btRaycastCallback.cpp
@@ -23,39 +23,38 @@ subject to the following restrictions:
 #include "BulletCollision/NarrowPhaseCollision/btGjkEpaPenetrationDepthSolver.h"
 #include "btRaycastCallback.h"
 
-btTriangleRaycastCallback::btTriangleRaycastCallback(const btVector3& from,const btVector3& to, unsigned int flags)
-	:
-	m_from(from),
-	m_to(to),
-   //@BP Mod
-   m_flags(flags),
-	m_hitFraction(btScalar(1.))
+btTriangleRaycastCallback::btTriangleRaycastCallback(const btVector3& from, const btVector3& to, unsigned int flags)
+	: m_from(from),
+	  m_to(to),
+	  //@BP Mod
+	  m_flags(flags),
+	  m_hitFraction(btScalar(1.))
 {
-
 }
 
-
-
-void btTriangleRaycastCallback::processTriangle(btVector3* triangle,int partId, int triangleIndex)
+void btTriangleRaycastCallback::processTriangle(btVector3* triangle, int partId, int triangleIndex)
 {
-	const btVector3 &vert0=triangle[0];
-	const btVector3 &vert1=triangle[1];
-	const btVector3 &vert2=triangle[2];
+	const btVector3& vert0 = triangle[0];
+	const btVector3& vert1 = triangle[1];
+	const btVector3& vert2 = triangle[2];
 
-	btVector3 v10; v10 = vert1 - vert0 ;
-	btVector3 v20; v20 = vert2 - vert0 ;
+	btVector3 v10;
+	v10 = vert1 - vert0;
+	btVector3 v20;
+	v20 = vert2 - vert0;
+
+	btVector3 triangleNormal;
+	triangleNormal = v10.cross(v20);
 
-	btVector3 triangleNormal; triangleNormal = v10.cross( v20 );
-	
 	const btScalar dist = vert0.dot(triangleNormal);
-	btScalar dist_a = triangleNormal.dot(m_from) ;
-	dist_a-= dist;
+	btScalar dist_a = triangleNormal.dot(m_from);
+	dist_a -= dist;
 	btScalar dist_b = triangleNormal.dot(m_to);
 	dist_b -= dist;
 
-	if ( dist_a * dist_b >= btScalar(0.0) )
+	if (dist_a * dist_b >= btScalar(0.0))
 	{
-		return ; // same sign
+		return;  // same sign
 	}
 
 	if (((m_flags & kF_FilterBackfaces) != 0) && (dist_a <= btScalar(0.0)))
@@ -64,52 +63,52 @@ void btTriangleRaycastCallback::processTriangle(btVector3* triangle,int partId,
 		return;
 	}
 
-	
-	const btScalar proj_length=dist_a-dist_b;
-	const btScalar distance = (dist_a)/(proj_length);
+	const btScalar proj_length = dist_a - dist_b;
+	const btScalar distance = (dist_a) / (proj_length);
 	// Now we have the intersection point on the plane, we'll see if it's inside the triangle
 	// Add an epsilon as a tolerance for the raycast,
 	// in case the ray hits exacly on the edge of the triangle.
 	// It must be scaled for the triangle size.
-	
-	if(distance < m_hitFraction)
+
+	if (distance < m_hitFraction)
 	{
-		
-
-		btScalar edge_tolerance =triangleNormal.length2();		
+		btScalar edge_tolerance = triangleNormal.length2();
 		edge_tolerance *= btScalar(-0.0001);
-		btVector3 point; point.setInterpolate3( m_from, m_to, distance);
+		btVector3 point;
+		point.setInterpolate3(m_from, m_to, distance);
 		{
-			btVector3 v0p; v0p = vert0 - point;
-			btVector3 v1p; v1p = vert1 - point;
-			btVector3 cp0; cp0 = v0p.cross( v1p );
+			btVector3 v0p;
+			v0p = vert0 - point;
+			btVector3 v1p;
+			v1p = vert1 - point;
+			btVector3 cp0;
+			cp0 = v0p.cross(v1p);
 
-			if ( (btScalar)(cp0.dot(triangleNormal)) >=edge_tolerance) 
+			if ((btScalar)(cp0.dot(triangleNormal)) >= edge_tolerance)
 			{
-						
-
-				btVector3 v2p; v2p = vert2 -  point;
+				btVector3 v2p;
+				v2p = vert2 - point;
 				btVector3 cp1;
-				cp1 = v1p.cross( v2p);
-				if ( (btScalar)(cp1.dot(triangleNormal)) >=edge_tolerance) 
+				cp1 = v1p.cross(v2p);
+				if ((btScalar)(cp1.dot(triangleNormal)) >= edge_tolerance)
 				{
 					btVector3 cp2;
 					cp2 = v2p.cross(v0p);
-					
-					if ( (btScalar)(cp2.dot(triangleNormal)) >=edge_tolerance) 
-					{
-					  //@BP Mod
-					  // Triangle normal isn't normalized
-				      triangleNormal.normalize();
 
-					 //@BP Mod - Allow for unflipped normal when raycasting against backfaces
+					if ((btScalar)(cp2.dot(triangleNormal)) >= edge_tolerance)
+					{
+						//@BP Mod
+						// Triangle normal isn't normalized
+						triangleNormal.normalize();
+
+						//@BP Mod - Allow for unflipped normal when raycasting against backfaces
 						if (((m_flags & kF_KeepUnflippedNormal) == 0) && (dist_a <= btScalar(0.0)))
 						{
-							m_hitFraction = reportHit(-triangleNormal,distance,partId,triangleIndex);
+							m_hitFraction = reportHit(-triangleNormal, distance, partId, triangleIndex);
 						}
 						else
 						{
-							m_hitFraction = reportHit(triangleNormal,distance,partId,triangleIndex);
+							m_hitFraction = reportHit(triangleNormal, distance, partId, triangleIndex);
 						}
 					}
 				}
@@ -118,8 +117,7 @@ void btTriangleRaycastCallback::processTriangle(btVector3* triangle,int partId,
 	}
 }
 
-
-btTriangleConvexcastCallback::btTriangleConvexcastCallback (const btConvexShape* convexShape, const btTransform& convexShapeFrom, const btTransform& convexShapeTo, const btTransform& triangleToWorld, const btScalar triangleCollisionMargin)
+btTriangleConvexcastCallback::btTriangleConvexcastCallback(const btConvexShape* convexShape, const btTransform& convexShapeFrom, const btTransform& convexShapeTo, const btTransform& triangleToWorld, const btScalar triangleCollisionMargin)
 {
 	m_convexShape = convexShape;
 	m_convexShapeFrom = convexShapeFrom;
@@ -130,14 +128,13 @@ btTriangleConvexcastCallback::btTriangleConvexcastCallback (const btConvexShape*
 	m_allowedPenetration = 0.f;
 }
 
-void
-btTriangleConvexcastCallback::processTriangle (btVector3* triangle, int partId, int triangleIndex)
+void btTriangleConvexcastCallback::processTriangle(btVector3* triangle, int partId, int triangleIndex)
 {
-	btTriangleShape triangleShape (triangle[0], triangle[1], triangle[2]);
-    triangleShape.setMargin(m_triangleCollisionMargin);
+	btTriangleShape triangleShape(triangle[0], triangle[1], triangle[2]);
+	triangleShape.setMargin(m_triangleCollisionMargin);
 
-	btVoronoiSimplexSolver	simplexSolver;
-	btGjkEpaPenetrationDepthSolver	gjkEpaPenetrationSolver;
+	btVoronoiSimplexSolver simplexSolver;
+	btGjkEpaPenetrationDepthSolver gjkEpaPenetrationSolver;
 
 //#define  USE_SUBSIMPLEX_CONVEX_CAST 1
 //if you reenable USE_SUBSIMPLEX_CONVEX_CAST see commented out code below
@@ -145,21 +142,21 @@ btTriangleConvexcastCallback::processTriangle (btVector3* triangle, int partId,
 	btSubsimplexConvexCast convexCaster(m_convexShape, &triangleShape, &simplexSolver);
 #else
 	//btGjkConvexCast	convexCaster(m_convexShape,&triangleShape,&simplexSolver);
-	btContinuousConvexCollision convexCaster(m_convexShape,&triangleShape,&simplexSolver,&gjkEpaPenetrationSolver);
-#endif //#USE_SUBSIMPLEX_CONVEX_CAST
-	
+	btContinuousConvexCollision convexCaster(m_convexShape, &triangleShape, &simplexSolver, &gjkEpaPenetrationSolver);
+#endif  //#USE_SUBSIMPLEX_CONVEX_CAST
+
 	btConvexCast::CastResult castResult;
 	castResult.m_fraction = btScalar(1.);
 	castResult.m_allowedPenetration = m_allowedPenetration;
-	if (convexCaster.calcTimeOfImpact(m_convexShapeFrom,m_convexShapeTo,m_triangleToWorld, m_triangleToWorld, castResult))
+	if (convexCaster.calcTimeOfImpact(m_convexShapeFrom, m_convexShapeTo, m_triangleToWorld, m_triangleToWorld, castResult))
 	{
 		//add hit
 		if (castResult.m_normal.length2() > btScalar(0.0001))
-		{					
+		{
 			if (castResult.m_fraction < m_hitFraction)
 			{
-/* btContinuousConvexCast's normal is already in world space */
-/*
+				/* btContinuousConvexCast's normal is already in world space */
+				/*
 #ifdef USE_SUBSIMPLEX_CONVEX_CAST
 				//rotate normal into worldspace
 				castResult.m_normal = m_convexShapeFrom.getBasis() * castResult.m_normal;
@@ -167,11 +164,11 @@ btTriangleConvexcastCallback::processTriangle (btVector3* triangle, int partId,
 */
 				castResult.m_normal.normalize();
 
-				reportHit (castResult.m_normal,
-							castResult.m_hitPoint,
-							castResult.m_fraction,
-							partId,
-							triangleIndex);
+				reportHit(castResult.m_normal,
+						  castResult.m_hitPoint,
+						  castResult.m_fraction,
+						  partId,
+						  triangleIndex);
 			}
 		}
 	}
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btRaycastCallback.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btRaycastCallback.h
index f2ed0cd39c4..2b2dfabec27 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btRaycastCallback.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btRaycastCallback.h
@@ -21,35 +21,33 @@ subject to the following restrictions:
 struct btBroadphaseProxy;
 class btConvexShape;
 
-class  btTriangleRaycastCallback: public btTriangleCallback
+class btTriangleRaycastCallback : public btTriangleCallback
 {
 public:
-
 	//input
 	btVector3 m_from;
 	btVector3 m_to;
 
-   //@BP Mod - allow backface filtering and unflipped normals
-   enum EFlags
-   {
-	  kF_None                 = 0,
-      kF_FilterBackfaces      = 1 << 0,
-      kF_KeepUnflippedNormal  = 1 << 1,   // Prevents returned face normal getting flipped when a ray hits a back-facing triangle
-		///SubSimplexConvexCastRaytest is the default, even if kF_None is set.
-	  kF_UseSubSimplexConvexCastRaytest = 1 << 2,   // Uses an approximate but faster ray versus convex intersection algorithm
-	  kF_UseGjkConvexCastRaytest = 1 << 3,
-      kF_Terminator        = 0xFFFFFFFF
-   };
-   unsigned int m_flags;
+	//@BP Mod - allow backface filtering and unflipped normals
+	enum EFlags
+	{
+		kF_None = 0,
+		kF_FilterBackfaces = 1 << 0,
+		kF_KeepUnflippedNormal = 1 << 1,             // Prevents returned face normal getting flipped when a ray hits a back-facing triangle
+													 ///SubSimplexConvexCastRaytest is the default, even if kF_None is set.
+		kF_UseSubSimplexConvexCastRaytest = 1 << 2,  // Uses an approximate but faster ray versus convex intersection algorithm
+		kF_UseGjkConvexCastRaytest = 1 << 3,
+		kF_Terminator = 0xFFFFFFFF
+	};
+	unsigned int m_flags;
 
-	btScalar	m_hitFraction;
+	btScalar m_hitFraction;
+
+	btTriangleRaycastCallback(const btVector3& from, const btVector3& to, unsigned int flags = 0);
 
-	btTriangleRaycastCallback(const btVector3& from,const btVector3& to, unsigned int flags=0);
-	
 	virtual void processTriangle(btVector3* triangle, int partId, int triangleIndex);
 
-	virtual btScalar reportHit(const btVector3& hitNormalLocal, btScalar hitFraction, int partId, int triangleIndex ) = 0;
-	
+	virtual btScalar reportHit(const btVector3& hitNormalLocal, btScalar hitFraction, int partId, int triangleIndex) = 0;
 };
 
 class btTriangleConvexcastCallback : public btTriangleCallback
@@ -63,12 +61,11 @@ public:
 	btScalar m_triangleCollisionMargin;
 	btScalar m_allowedPenetration;
 
-	btTriangleConvexcastCallback (const btConvexShape* convexShape, const btTransform& convexShapeFrom, const btTransform& convexShapeTo, const btTransform& triangleToWorld, const btScalar triangleCollisionMargin);
+	btTriangleConvexcastCallback(const btConvexShape* convexShape, const btTransform& convexShapeFrom, const btTransform& convexShapeTo, const btTransform& triangleToWorld, const btScalar triangleCollisionMargin);
 
-	virtual void processTriangle (btVector3* triangle, int partId, int triangleIndex);
+	virtual void processTriangle(btVector3* triangle, int partId, int triangleIndex);
 
-	virtual btScalar reportHit (const btVector3& hitNormalLocal, const btVector3& hitPointLocal, btScalar hitFraction, int partId, int triangleIndex) = 0;
+	virtual btScalar reportHit(const btVector3& hitNormalLocal, const btVector3& hitPointLocal, btScalar hitFraction, int partId, int triangleIndex) = 0;
 };
 
-#endif //BT_RAYCAST_TRI_CALLBACK_H
-
+#endif  //BT_RAYCAST_TRI_CALLBACK_H
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btSimplexSolverInterface.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btSimplexSolverInterface.h
index da8a13914c9..ccd227109d2 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btSimplexSolverInterface.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btSimplexSolverInterface.h
@@ -13,8 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #ifndef BT_SIMPLEX_SOLVER_INTERFACE_H
 #define BT_SIMPLEX_SOLVER_INTERFACE_H
 
@@ -31,33 +29,30 @@ subject to the following restrictions:
 /// voronoi regions or barycentric coordinates
 class btSimplexSolverInterface
 {
-	public:
-		virtual ~btSimplexSolverInterface() {};
+public:
+	virtual ~btSimplexSolverInterface(){};
 
 	virtual void reset() = 0;
 
 	virtual void addVertex(const btVector3& w, const btVector3& p, const btVector3& q) = 0;
-	
+
 	virtual bool closest(btVector3& v) = 0;
 
 	virtual btScalar maxVertex() = 0;
 
 	virtual bool fullSimplex() const = 0;
 
-	virtual int getSimplex(btVector3 *pBuf, btVector3 *qBuf, btVector3 *yBuf) const = 0;
+	virtual int getSimplex(btVector3* pBuf, btVector3* qBuf, btVector3* yBuf) const = 0;
 
 	virtual bool inSimplex(const btVector3& w) = 0;
-	
+
 	virtual void backup_closest(btVector3& v) = 0;
 
 	virtual bool emptySimplex() const = 0;
 
 	virtual void compute_points(btVector3& p1, btVector3& p2) = 0;
 
-	virtual int numVertices() const =0;
-
-
+	virtual int numVertices() const = 0;
 };
 #endif
-#endif //BT_SIMPLEX_SOLVER_INTERFACE_H
-
+#endif  //BT_SIMPLEX_SOLVER_INTERFACE_H
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btSubSimplexConvexCast.cpp b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btSubSimplexConvexCast.cpp
index 08d6e6de869..37458339e70 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btSubSimplexConvexCast.cpp
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btSubSimplexConvexCast.cpp
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btSubSimplexConvexCast.h"
 #include "BulletCollision/CollisionShapes/btConvexShape.h"
 
@@ -22,32 +21,26 @@ subject to the following restrictions:
 #include "btPointCollector.h"
 #include "LinearMath/btTransformUtil.h"
 
-btSubsimplexConvexCast::btSubsimplexConvexCast (const btConvexShape* convexA,const btConvexShape* convexB,btSimplexSolverInterface* simplexSolver)
-:m_simplexSolver(simplexSolver),
-m_convexA(convexA),m_convexB(convexB)
+btSubsimplexConvexCast::btSubsimplexConvexCast(const btConvexShape* convexA, const btConvexShape* convexB, btSimplexSolverInterface* simplexSolver)
+	: m_simplexSolver(simplexSolver),
+	  m_convexA(convexA),
+	  m_convexB(convexB)
 {
 }
 
-///Typically the conservative advancement reaches solution in a few iterations, clip it to 32 for degenerate cases.
-///See discussion about this here http://continuousphysics.com/Bullet/phpBB2/viewtopic.php?t=565
-#ifdef BT_USE_DOUBLE_PRECISION
-#define MAX_ITERATIONS 64
-#else
-#define MAX_ITERATIONS 32
-#endif
-bool	btSubsimplexConvexCast::calcTimeOfImpact(
-		const btTransform& fromA,
-		const btTransform& toA,
-		const btTransform& fromB,
-		const btTransform& toB,
-		CastResult& result)
-{
 
+bool btSubsimplexConvexCast::calcTimeOfImpact(
+	const btTransform& fromA,
+	const btTransform& toA,
+	const btTransform& fromB,
+	const btTransform& toB,
+	CastResult& result)
+{
 	m_simplexSolver->reset();
 
-	btVector3 linVelA,linVelB;
-	linVelA = toA.getOrigin()-fromA.getOrigin();
-	linVelB = toB.getOrigin()-fromB.getOrigin();
+	btVector3 linVelA, linVelB;
+	linVelA = toA.getOrigin() - fromA.getOrigin();
+	linVelB = toB.getOrigin() - fromB.getOrigin();
 
 	btScalar lambda = btScalar(0.);
 
@@ -55,43 +48,31 @@ bool	btSubsimplexConvexCast::calcTimeOfImpact(
 	btTransform interpolatedTransB = fromB;
 
 	///take relative motion
-	btVector3 r = (linVelA-linVelB);
+	btVector3 r = (linVelA - linVelB);
 	btVector3 v;
-	
-	btVector3 supVertexA = fromA(m_convexA->localGetSupportingVertex(-r*fromA.getBasis()));
-	btVector3 supVertexB = fromB(m_convexB->localGetSupportingVertex(r*fromB.getBasis()));
-	v = supVertexA-supVertexB;
-	int maxIter = MAX_ITERATIONS;
+
+	btVector3 supVertexA = fromA(m_convexA->localGetSupportingVertex(-r * fromA.getBasis()));
+	btVector3 supVertexB = fromB(m_convexB->localGetSupportingVertex(r * fromB.getBasis()));
+	v = supVertexA - supVertexB;
+	int maxIter = result.m_subSimplexCastMaxIterations;
 
 	btVector3 n;
-	n.setValue(btScalar(0.),btScalar(0.),btScalar(0.));
-	
+	n.setValue(btScalar(0.), btScalar(0.), btScalar(0.));
+
 	btVector3 c;
 
-	
-
-
 	btScalar dist2 = v.length2();
 
-#ifdef BT_USE_DOUBLE_PRECISION
-	btScalar epsilon = SIMD_EPSILON * 10;
-#else
-//todo: epsilon kept for backward compatibility of unit tests.
-//will need to digg deeper to make the algorithm more robust
-//since, a large epsilon can cause an early termination with false
-//positive results (ray intersections that shouldn't be there)
-	btScalar epsilon = btScalar(0.0001);
-#endif //BT_USE_DOUBLE_PRECISION
 
 
-	btVector3	w,p;
+	btVector3 w, p;
 	btScalar VdotR;
-	
-	while ( (dist2 > epsilon) && maxIter--)
+
+	while ((dist2 > result.m_subSimplexCastEpsilon) && maxIter--)
 	{
-		supVertexA = interpolatedTransA(m_convexA->localGetSupportingVertex(-v*interpolatedTransA.getBasis()));
-		supVertexB = interpolatedTransB(m_convexB->localGetSupportingVertex(v*interpolatedTransB.getBasis()));
-		w = supVertexA-supVertexB;
+		supVertexA = interpolatedTransA(m_convexA->localGetSupportingVertex(-v * interpolatedTransA.getBasis()));
+		supVertexB = interpolatedTransB(m_convexB->localGetSupportingVertex(v * interpolatedTransB.getBasis()));
+		w = supVertexA - supVertexB;
 
 		btScalar VdotW = v.dot(w);
 
@@ -100,68 +81,63 @@ bool	btSubsimplexConvexCast::calcTimeOfImpact(
 			return false;
 		}
 
-		if ( VdotW > btScalar(0.))
+		if (VdotW > btScalar(0.))
 		{
 			VdotR = v.dot(r);
 
-			if (VdotR >= -(SIMD_EPSILON*SIMD_EPSILON))
+			if (VdotR >= -(SIMD_EPSILON * SIMD_EPSILON))
 				return false;
 			else
 			{
 				lambda = lambda - VdotW / VdotR;
 				//interpolate to next lambda
 				//	x = s + lambda * r;
-				interpolatedTransA.getOrigin().setInterpolate3(fromA.getOrigin(),toA.getOrigin(),lambda);
-				interpolatedTransB.getOrigin().setInterpolate3(fromB.getOrigin(),toB.getOrigin(),lambda);
+				interpolatedTransA.getOrigin().setInterpolate3(fromA.getOrigin(), toA.getOrigin(), lambda);
+				interpolatedTransB.getOrigin().setInterpolate3(fromB.getOrigin(), toB.getOrigin(), lambda);
 				//m_simplexSolver->reset();
 				//check next line
-				 w = supVertexA-supVertexB;
-				
+				w = supVertexA - supVertexB;
+
 				n = v;
-				
 			}
-		} 
+		}
 		///Just like regular GJK only add the vertex if it isn't already (close) to current vertex, it would lead to divisions by zero and NaN etc.
 		if (!m_simplexSolver->inSimplex(w))
-			m_simplexSolver->addVertex( w, supVertexA , supVertexB);
+			m_simplexSolver->addVertex(w, supVertexA, supVertexB);
 
 		if (m_simplexSolver->closest(v))
 		{
 			dist2 = v.length2();
-			
+
 			//todo: check this normal for validity
 			//n=v;
 			//printf("V=%f , %f, %f\n",v[0],v[1],v[2]);
 			//printf("DIST2=%f\n",dist2);
 			//printf("numverts = %i\n",m_simplexSolver->numVertices());
-		} else
+		}
+		else
 		{
 			dist2 = btScalar(0.);
-		} 
+		}
 	}
 
 	//int numiter = MAX_ITERATIONS - maxIter;
-//	printf("number of iterations: %d", numiter);
-	
+	//	printf("number of iterations: %d", numiter);
+
 	//don't report a time of impact when moving 'away' from the hitnormal
-	
 
 	result.m_fraction = lambda;
-	if (n.length2() >= (SIMD_EPSILON*SIMD_EPSILON))
+	if (n.length2() >= (SIMD_EPSILON * SIMD_EPSILON))
 		result.m_normal = n.normalized();
 	else
 		result.m_normal = btVector3(btScalar(0.0), btScalar(0.0), btScalar(0.0));
 
 	//don't report time of impact for motion away from the contact normal (or causes minor penetration)
-	if (result.m_normal.dot(r)>=-result.m_allowedPenetration)
+	if (result.m_normal.dot(r) >= -result.m_allowedPenetration)
 		return false;
 
-	btVector3 hitA,hitB;
-	m_simplexSolver->compute_points(hitA,hitB);
-	result.m_hitPoint=hitB;
+	btVector3 hitA, hitB;
+	m_simplexSolver->compute_points(hitA, hitB);
+	result.m_hitPoint = hitB;
 	return true;
 }
-
-
-
-
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btSubSimplexConvexCast.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btSubSimplexConvexCast.h
index 6c8127983eb..0638a30eb1a 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btSubSimplexConvexCast.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btSubSimplexConvexCast.h
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef BT_SUBSIMPLEX_CONVEX_CAST_H
 #define BT_SUBSIMPLEX_CONVEX_CAST_H
 
@@ -28,23 +27,21 @@ class btConvexShape;
 class btSubsimplexConvexCast : public btConvexCast
 {
 	btSimplexSolverInterface* m_simplexSolver;
-	const btConvexShape*	m_convexA;
-	const btConvexShape*	m_convexB;
+	const btConvexShape* m_convexA;
+	const btConvexShape* m_convexB;
 
 public:
-
-	btSubsimplexConvexCast (const btConvexShape*	shapeA,const btConvexShape*	shapeB,btSimplexSolverInterface* simplexSolver);
+	btSubsimplexConvexCast(const btConvexShape* shapeA, const btConvexShape* shapeB, btSimplexSolverInterface* simplexSolver);
 
 	//virtual ~btSubsimplexConvexCast();
 	///SimsimplexConvexCast calculateTimeOfImpact calculates the time of impact+normal for the linear cast (sweep) between two moving objects.
 	///Precondition is that objects should not penetration/overlap at the start from the interval. Overlap can be tested using btGjkPairDetector.
-	virtual bool	calcTimeOfImpact(
-			const btTransform& fromA,
-			const btTransform& toA,
-			const btTransform& fromB,
-			const btTransform& toB,
-			CastResult& result);
-
+	virtual bool calcTimeOfImpact(
+		const btTransform& fromA,
+		const btTransform& toA,
+		const btTransform& fromB,
+		const btTransform& toB,
+		CastResult& result);
 };
 
-#endif //BT_SUBSIMPLEX_CONVEX_CAST_H
+#endif  //BT_SUBSIMPLEX_CONVEX_CAST_H
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btVoronoiSimplexSolver.cpp b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btVoronoiSimplexSolver.cpp
index 756373c9b5e..8fda94d2adf 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btVoronoiSimplexSolver.cpp
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btVoronoiSimplexSolver.cpp
@@ -23,26 +23,24 @@ subject to the following restrictions:
 		
 */
 
-
 #include "btVoronoiSimplexSolver.h"
 
-#define VERTA  0
-#define VERTB  1
-#define VERTC  2
-#define VERTD  3
+#define VERTA 0
+#define VERTB 1
+#define VERTC 2
+#define VERTD 3
 
 #define CATCH_DEGENERATE_TETRAHEDRON 1
-void	btVoronoiSimplexSolver::removeVertex(int index)
+void btVoronoiSimplexSolver::removeVertex(int index)
 {
-	
-	btAssert(m_numVertices>0);
+	btAssert(m_numVertices > 0);
 	m_numVertices--;
 	m_simplexVectorW[index] = m_simplexVectorW[m_numVertices];
 	m_simplexPointsP[index] = m_simplexPointsP[m_numVertices];
 	m_simplexPointsQ[index] = m_simplexPointsQ[m_numVertices];
 }
 
-void	btVoronoiSimplexSolver::reduceVertices (const btUsageBitfield& usedVerts)
+void btVoronoiSimplexSolver::reduceVertices(const btUsageBitfield& usedVerts)
 {
 	if ((numVertices() >= 4) && (!usedVerts.usedVertexD))
 		removeVertex(3);
@@ -52,29 +50,22 @@ void	btVoronoiSimplexSolver::reduceVertices (const btUsageBitfield& usedVerts)
 
 	if ((numVertices() >= 2) && (!usedVerts.usedVertexB))
 		removeVertex(1);
-	
+
 	if ((numVertices() >= 1) && (!usedVerts.usedVertexA))
 		removeVertex(0);
-
 }
 
-
-
-
-
 //clear the simplex, remove all the vertices
 void btVoronoiSimplexSolver::reset()
 {
 	m_cachedValidClosest = false;
 	m_numVertices = 0;
 	m_needsUpdate = true;
-	m_lastW = btVector3(btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT),btScalar(BT_LARGE_FLOAT));
+	m_lastW = btVector3(btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT), btScalar(BT_LARGE_FLOAT));
 	m_cachedBC.reset();
 }
 
-
-
-	//add a vertex
+//add a vertex
 void btVoronoiSimplexSolver::addVertex(const btVector3& w, const btVector3& p, const btVector3& q)
 {
 	m_lastW = w;
@@ -87,9 +78,8 @@ void btVoronoiSimplexSolver::addVertex(const btVector3& w, const btVector3& p, c
 	m_numVertices++;
 }
 
-bool	btVoronoiSimplexSolver::updateClosestVectorAndPoints()
+bool btVoronoiSimplexSolver::updateClosestVectorAndPoints()
 {
-	
 	if (m_needsUpdate)
 	{
 		m_cachedBC.reset();
@@ -98,127 +88,131 @@ bool	btVoronoiSimplexSolver::updateClosestVectorAndPoints()
 
 		switch (numVertices())
 		{
-		case 0:
+			case 0:
 				m_cachedValidClosest = false;
 				break;
-		case 1:
+			case 1:
 			{
 				m_cachedP1 = m_simplexPointsP[0];
 				m_cachedP2 = m_simplexPointsQ[0];
-				m_cachedV = m_cachedP1-m_cachedP2; //== m_simplexVectorW[0]
+				m_cachedV = m_cachedP1 - m_cachedP2;  //== m_simplexVectorW[0]
 				m_cachedBC.reset();
-				m_cachedBC.setBarycentricCoordinates(btScalar(1.),btScalar(0.),btScalar(0.),btScalar(0.));
+				m_cachedBC.setBarycentricCoordinates(btScalar(1.), btScalar(0.), btScalar(0.), btScalar(0.));
 				m_cachedValidClosest = m_cachedBC.isValid();
 				break;
 			};
-		case 2:
+			case 2:
 			{
-			//closest point origin from line segment
-					const btVector3& from = m_simplexVectorW[0];
-					const btVector3& to = m_simplexVectorW[1];
-					btVector3 nearest;
+				//closest point origin from line segment
+				const btVector3& from = m_simplexVectorW[0];
+				const btVector3& to = m_simplexVectorW[1];
+				btVector3 nearest;
 
-					btVector3 p (btScalar(0.),btScalar(0.),btScalar(0.));
-					btVector3 diff = p - from;
-					btVector3 v = to - from;
-					btScalar t = v.dot(diff);
-					
-					if (t > 0) {
-						btScalar dotVV = v.dot(v);
-						if (t < dotVV) {
-							t /= dotVV;
-							diff -= t*v;
-							m_cachedBC.m_usedVertices.usedVertexA = true;
-							m_cachedBC.m_usedVertices.usedVertexB = true;
-						} else {
-							t = 1;
-							diff -= v;
-							//reduce to 1 point
-							m_cachedBC.m_usedVertices.usedVertexB = true;
-						}
-					} else
+				btVector3 p(btScalar(0.), btScalar(0.), btScalar(0.));
+				btVector3 diff = p - from;
+				btVector3 v = to - from;
+				btScalar t = v.dot(diff);
+
+				if (t > 0)
+				{
+					btScalar dotVV = v.dot(v);
+					if (t < dotVV)
 					{
-						t = 0;
-						//reduce to 1 point
+						t /= dotVV;
+						diff -= t * v;
 						m_cachedBC.m_usedVertices.usedVertexA = true;
+						m_cachedBC.m_usedVertices.usedVertexB = true;
 					}
-					m_cachedBC.setBarycentricCoordinates(1-t,t);
-					nearest = from + t*v;
+					else
+					{
+						t = 1;
+						diff -= v;
+						//reduce to 1 point
+						m_cachedBC.m_usedVertices.usedVertexB = true;
+					}
+				}
+				else
+				{
+					t = 0;
+					//reduce to 1 point
+					m_cachedBC.m_usedVertices.usedVertexA = true;
+				}
+				m_cachedBC.setBarycentricCoordinates(1 - t, t);
+				nearest = from + t * v;
 
-					m_cachedP1 = m_simplexPointsP[0] + t * (m_simplexPointsP[1] - m_simplexPointsP[0]);
-					m_cachedP2 = m_simplexPointsQ[0] + t * (m_simplexPointsQ[1] - m_simplexPointsQ[0]);
-					m_cachedV = m_cachedP1 - m_cachedP2;
-					
-					reduceVertices(m_cachedBC.m_usedVertices);
+				m_cachedP1 = m_simplexPointsP[0] + t * (m_simplexPointsP[1] - m_simplexPointsP[0]);
+				m_cachedP2 = m_simplexPointsQ[0] + t * (m_simplexPointsQ[1] - m_simplexPointsQ[0]);
+				m_cachedV = m_cachedP1 - m_cachedP2;
 
-					m_cachedValidClosest = m_cachedBC.isValid();
-					break;
+				reduceVertices(m_cachedBC.m_usedVertices);
+
+				m_cachedValidClosest = m_cachedBC.isValid();
+				break;
 			}
-		case 3: 
-			{ 
-				//closest point origin from triangle 
-				btVector3 p (btScalar(0.),btScalar(0.),btScalar(0.)); 
-
-				const btVector3& a = m_simplexVectorW[0]; 
-				const btVector3& b = m_simplexVectorW[1]; 
-				const btVector3& c = m_simplexVectorW[2]; 
-
-				closestPtPointTriangle(p,a,b,c,m_cachedBC); 
-				m_cachedP1 = m_simplexPointsP[0] * m_cachedBC.m_barycentricCoords[0] + 
-				m_simplexPointsP[1] * m_cachedBC.m_barycentricCoords[1] + 
-				m_simplexPointsP[2] * m_cachedBC.m_barycentricCoords[2]; 
-
-				m_cachedP2 = m_simplexPointsQ[0] * m_cachedBC.m_barycentricCoords[0] + 
-				m_simplexPointsQ[1] * m_cachedBC.m_barycentricCoords[1] + 
-				m_simplexPointsQ[2] * m_cachedBC.m_barycentricCoords[2]; 
-
-				m_cachedV = m_cachedP1-m_cachedP2; 
-
-				reduceVertices (m_cachedBC.m_usedVertices); 
-				m_cachedValidClosest = m_cachedBC.isValid(); 
-
-				break; 
-			}
-		case 4:
+			case 3:
 			{
+				//closest point origin from triangle
+				btVector3 p(btScalar(0.), btScalar(0.), btScalar(0.));
+
+				const btVector3& a = m_simplexVectorW[0];
+				const btVector3& b = m_simplexVectorW[1];
+				const btVector3& c = m_simplexVectorW[2];
+
+				closestPtPointTriangle(p, a, b, c, m_cachedBC);
+				m_cachedP1 = m_simplexPointsP[0] * m_cachedBC.m_barycentricCoords[0] +
+							 m_simplexPointsP[1] * m_cachedBC.m_barycentricCoords[1] +
+							 m_simplexPointsP[2] * m_cachedBC.m_barycentricCoords[2];
+
+				m_cachedP2 = m_simplexPointsQ[0] * m_cachedBC.m_barycentricCoords[0] +
+							 m_simplexPointsQ[1] * m_cachedBC.m_barycentricCoords[1] +
+							 m_simplexPointsQ[2] * m_cachedBC.m_barycentricCoords[2];
+
+				m_cachedV = m_cachedP1 - m_cachedP2;
+
+				reduceVertices(m_cachedBC.m_usedVertices);
+				m_cachedValidClosest = m_cachedBC.isValid();
+
+				break;
+			}
+			case 4:
+			{
+				btVector3 p(btScalar(0.), btScalar(0.), btScalar(0.));
 
-				
-				btVector3 p (btScalar(0.),btScalar(0.),btScalar(0.));
-				
 				const btVector3& a = m_simplexVectorW[0];
 				const btVector3& b = m_simplexVectorW[1];
 				const btVector3& c = m_simplexVectorW[2];
 				const btVector3& d = m_simplexVectorW[3];
 
-				bool hasSeparation = closestPtPointTetrahedron(p,a,b,c,d,m_cachedBC);
+				bool hasSeparation = closestPtPointTetrahedron(p, a, b, c, d, m_cachedBC);
 
 				if (hasSeparation)
 				{
-
 					m_cachedP1 = m_simplexPointsP[0] * m_cachedBC.m_barycentricCoords[0] +
-						m_simplexPointsP[1] * m_cachedBC.m_barycentricCoords[1] +
-						m_simplexPointsP[2] * m_cachedBC.m_barycentricCoords[2] +
-						m_simplexPointsP[3] * m_cachedBC.m_barycentricCoords[3];
+								 m_simplexPointsP[1] * m_cachedBC.m_barycentricCoords[1] +
+								 m_simplexPointsP[2] * m_cachedBC.m_barycentricCoords[2] +
+								 m_simplexPointsP[3] * m_cachedBC.m_barycentricCoords[3];
 
 					m_cachedP2 = m_simplexPointsQ[0] * m_cachedBC.m_barycentricCoords[0] +
-						m_simplexPointsQ[1] * m_cachedBC.m_barycentricCoords[1] +
-						m_simplexPointsQ[2] * m_cachedBC.m_barycentricCoords[2] +
-						m_simplexPointsQ[3] * m_cachedBC.m_barycentricCoords[3];
+								 m_simplexPointsQ[1] * m_cachedBC.m_barycentricCoords[1] +
+								 m_simplexPointsQ[2] * m_cachedBC.m_barycentricCoords[2] +
+								 m_simplexPointsQ[3] * m_cachedBC.m_barycentricCoords[3];
 
-					m_cachedV = m_cachedP1-m_cachedP2;
-					reduceVertices (m_cachedBC.m_usedVertices);
-				} else
+					m_cachedV = m_cachedP1 - m_cachedP2;
+					reduceVertices(m_cachedBC.m_usedVertices);
+				}
+				else
 				{
-//					printf("sub distance got penetration\n");
+					//					printf("sub distance got penetration\n");
 
 					if (m_cachedBC.m_degenerate)
 					{
 						m_cachedValidClosest = false;
-					} else
+					}
+					else
 					{
 						m_cachedValidClosest = true;
 						//degenerate case == false, penetration = true + zero
-						m_cachedV.setValue(btScalar(0.),btScalar(0.),btScalar(0.));
+						m_cachedV.setValue(btScalar(0.), btScalar(0.), btScalar(0.));
 					}
 					break;
 				}
@@ -228,7 +222,7 @@ bool	btVoronoiSimplexSolver::updateClosestVectorAndPoints()
 				//closest point origin from tetrahedron
 				break;
 			}
-		default:
+			default:
 			{
 				m_cachedValidClosest = false;
 			}
@@ -236,7 +230,6 @@ bool	btVoronoiSimplexSolver::updateClosestVectorAndPoints()
 	}
 
 	return m_cachedValidClosest;
-
 }
 
 //return/calculate the closest vertex
@@ -247,13 +240,11 @@ bool btVoronoiSimplexSolver::closest(btVector3& v)
 	return succes;
 }
 
-
-
 btScalar btVoronoiSimplexSolver::maxVertex()
 {
 	int i, numverts = numVertices();
 	btScalar maxV = btScalar(0.);
-	for (i=0;i<numverts;i++)
+	for (i = 0; i < numverts; i++)
 	{
 		btScalar curLen2 = m_simplexVectorW[i].length2();
 		if (maxV < curLen2)
@@ -262,13 +253,11 @@ btScalar btVoronoiSimplexSolver::maxVertex()
 	return maxV;
 }
 
-
-
-	//return the current simplex
-int btVoronoiSimplexSolver::getSimplex(btVector3 *pBuf, btVector3 *qBuf, btVector3 *yBuf) const
+//return the current simplex
+int btVoronoiSimplexSolver::getSimplex(btVector3* pBuf, btVector3* qBuf, btVector3* yBuf) const
 {
 	int i;
-	for (i=0;i<numVertices();i++)
+	for (i = 0; i < numVertices(); i++)
 	{
 		yBuf[i] = m_simplexVectorW[i];
 		pBuf[i] = m_simplexPointsP[i];
@@ -277,20 +266,17 @@ int btVoronoiSimplexSolver::getSimplex(btVector3 *pBuf, btVector3 *qBuf, btVecto
 	return numVertices();
 }
 
-
-
-
 bool btVoronoiSimplexSolver::inSimplex(const btVector3& w)
 {
 	bool found = false;
 	int i, numverts = numVertices();
 	//btScalar maxV = btScalar(0.);
-	
+
 	//w is in the current (reduced) simplex
-	for (i=0;i<numverts;i++)
+	for (i = 0; i < numverts; i++)
 	{
 #ifdef BT_USE_EQUAL_VERTEX_THRESHOLD
-		if ( m_simplexVectorW[i].distance2(w) <= m_equalVertexThreshold)
+		if (m_simplexVectorW[i].distance2(w) <= m_equalVertexThreshold)
 #else
 		if (m_simplexVectorW[i] == w)
 #endif
@@ -303,199 +289,190 @@ bool btVoronoiSimplexSolver::inSimplex(const btVector3& w)
 	//check in case lastW is already removed
 	if (w == m_lastW)
 		return true;
-    	
+
 	return found;
 }
 
-void btVoronoiSimplexSolver::backup_closest(btVector3& v) 
+void btVoronoiSimplexSolver::backup_closest(btVector3& v)
 {
 	v = m_cachedV;
 }
 
-
-bool btVoronoiSimplexSolver::emptySimplex() const 
+bool btVoronoiSimplexSolver::emptySimplex() const
 {
 	return (numVertices() == 0);
-
 }
 
-void btVoronoiSimplexSolver::compute_points(btVector3& p1, btVector3& p2) 
+void btVoronoiSimplexSolver::compute_points(btVector3& p1, btVector3& p2)
 {
 	updateClosestVectorAndPoints();
 	p1 = m_cachedP1;
 	p2 = m_cachedP2;
-
 }
 
-
-
-
-bool	btVoronoiSimplexSolver::closestPtPointTriangle(const btVector3& p, const btVector3& a, const btVector3& b, const btVector3& c,btSubSimplexClosestResult& result)
+bool btVoronoiSimplexSolver::closestPtPointTriangle(const btVector3& p, const btVector3& a, const btVector3& b, const btVector3& c, btSubSimplexClosestResult& result)
 {
 	result.m_usedVertices.reset();
 
-    // Check if P in vertex region outside A
-    btVector3 ab = b - a;
-    btVector3 ac = c - a;
-    btVector3 ap = p - a;
-    btScalar d1 = ab.dot(ap);
-    btScalar d2 = ac.dot(ap);
-    if (d1 <= btScalar(0.0) && d2 <= btScalar(0.0)) 
+	// Check if P in vertex region outside A
+	btVector3 ab = b - a;
+	btVector3 ac = c - a;
+	btVector3 ap = p - a;
+	btScalar d1 = ab.dot(ap);
+	btScalar d2 = ac.dot(ap);
+	if (d1 <= btScalar(0.0) && d2 <= btScalar(0.0))
 	{
 		result.m_closestPointOnSimplex = a;
 		result.m_usedVertices.usedVertexA = true;
-		result.setBarycentricCoordinates(1,0,0);
-		return true;// a; // barycentric coordinates (1,0,0)
+		result.setBarycentricCoordinates(1, 0, 0);
+		return true;  // a; // barycentric coordinates (1,0,0)
 	}
 
-    // Check if P in vertex region outside B
-    btVector3 bp = p - b;
-    btScalar d3 = ab.dot(bp);
-    btScalar d4 = ac.dot(bp);
-    if (d3 >= btScalar(0.0) && d4 <= d3) 
+	// Check if P in vertex region outside B
+	btVector3 bp = p - b;
+	btScalar d3 = ab.dot(bp);
+	btScalar d4 = ac.dot(bp);
+	if (d3 >= btScalar(0.0) && d4 <= d3)
 	{
 		result.m_closestPointOnSimplex = b;
 		result.m_usedVertices.usedVertexB = true;
-		result.setBarycentricCoordinates(0,1,0);
+		result.setBarycentricCoordinates(0, 1, 0);
 
-		return true; // b; // barycentric coordinates (0,1,0)
+		return true;  // b; // barycentric coordinates (0,1,0)
 	}
-    // Check if P in edge region of AB, if so return projection of P onto AB
-    btScalar vc = d1*d4 - d3*d2;
-    if (vc <= btScalar(0.0) && d1 >= btScalar(0.0) && d3 <= btScalar(0.0)) {
-        btScalar v = d1 / (d1 - d3);
+	// Check if P in edge region of AB, if so return projection of P onto AB
+	btScalar vc = d1 * d4 - d3 * d2;
+	if (vc <= btScalar(0.0) && d1 >= btScalar(0.0) && d3 <= btScalar(0.0))
+	{
+		btScalar v = d1 / (d1 - d3);
 		result.m_closestPointOnSimplex = a + v * ab;
 		result.m_usedVertices.usedVertexA = true;
 		result.m_usedVertices.usedVertexB = true;
-		result.setBarycentricCoordinates(1-v,v,0);
+		result.setBarycentricCoordinates(1 - v, v, 0);
 		return true;
-        //return a + v * ab; // barycentric coordinates (1-v,v,0)
-    }
+		//return a + v * ab; // barycentric coordinates (1-v,v,0)
+	}
 
-    // Check if P in vertex region outside C
-    btVector3 cp = p - c;
-    btScalar d5 = ab.dot(cp);
-    btScalar d6 = ac.dot(cp);
-    if (d6 >= btScalar(0.0) && d5 <= d6) 
+	// Check if P in vertex region outside C
+	btVector3 cp = p - c;
+	btScalar d5 = ab.dot(cp);
+	btScalar d6 = ac.dot(cp);
+	if (d6 >= btScalar(0.0) && d5 <= d6)
 	{
 		result.m_closestPointOnSimplex = c;
 		result.m_usedVertices.usedVertexC = true;
-		result.setBarycentricCoordinates(0,0,1);
-		return true;//c; // barycentric coordinates (0,0,1)
+		result.setBarycentricCoordinates(0, 0, 1);
+		return true;  //c; // barycentric coordinates (0,0,1)
 	}
 
-    // Check if P in edge region of AC, if so return projection of P onto AC
-    btScalar vb = d5*d2 - d1*d6;
-    if (vb <= btScalar(0.0) && d2 >= btScalar(0.0) && d6 <= btScalar(0.0)) {
-        btScalar w = d2 / (d2 - d6);
+	// Check if P in edge region of AC, if so return projection of P onto AC
+	btScalar vb = d5 * d2 - d1 * d6;
+	if (vb <= btScalar(0.0) && d2 >= btScalar(0.0) && d6 <= btScalar(0.0))
+	{
+		btScalar w = d2 / (d2 - d6);
 		result.m_closestPointOnSimplex = a + w * ac;
 		result.m_usedVertices.usedVertexA = true;
 		result.m_usedVertices.usedVertexC = true;
-		result.setBarycentricCoordinates(1-w,0,w);
+		result.setBarycentricCoordinates(1 - w, 0, w);
 		return true;
-        //return a + w * ac; // barycentric coordinates (1-w,0,w)
-    }
+		//return a + w * ac; // barycentric coordinates (1-w,0,w)
+	}
+
+	// Check if P in edge region of BC, if so return projection of P onto BC
+	btScalar va = d3 * d6 - d5 * d4;
+	if (va <= btScalar(0.0) && (d4 - d3) >= btScalar(0.0) && (d5 - d6) >= btScalar(0.0))
+	{
+		btScalar w = (d4 - d3) / ((d4 - d3) + (d5 - d6));
 
-    // Check if P in edge region of BC, if so return projection of P onto BC
-    btScalar va = d3*d6 - d5*d4;
-    if (va <= btScalar(0.0) && (d4 - d3) >= btScalar(0.0) && (d5 - d6) >= btScalar(0.0)) {
-        btScalar w = (d4 - d3) / ((d4 - d3) + (d5 - d6));
-		
 		result.m_closestPointOnSimplex = b + w * (c - b);
 		result.m_usedVertices.usedVertexB = true;
 		result.m_usedVertices.usedVertexC = true;
-		result.setBarycentricCoordinates(0,1-w,w);
-		return true;		
-       // return b + w * (c - b); // barycentric coordinates (0,1-w,w)
-    }
+		result.setBarycentricCoordinates(0, 1 - w, w);
+		return true;
+		// return b + w * (c - b); // barycentric coordinates (0,1-w,w)
+	}
+
+	// P inside face region. Compute Q through its barycentric coordinates (u,v,w)
+	btScalar denom = btScalar(1.0) / (va + vb + vc);
+	btScalar v = vb * denom;
+	btScalar w = vc * denom;
 
-    // P inside face region. Compute Q through its barycentric coordinates (u,v,w)
-    btScalar denom = btScalar(1.0) / (va + vb + vc);
-    btScalar v = vb * denom;
-    btScalar w = vc * denom;
-    
 	result.m_closestPointOnSimplex = a + ab * v + ac * w;
 	result.m_usedVertices.usedVertexA = true;
 	result.m_usedVertices.usedVertexB = true;
 	result.m_usedVertices.usedVertexC = true;
-	result.setBarycentricCoordinates(1-v-w,v,w);
-	
+	result.setBarycentricCoordinates(1 - v - w, v, w);
+
 	return true;
-//	return a + ab * v + ac * w; // = u*a + v*b + w*c, u = va * denom = btScalar(1.0) - v - w
-
+	//	return a + ab * v + ac * w; // = u*a + v*b + w*c, u = va * denom = btScalar(1.0) - v - w
 }
 
-
-
-
-
 /// Test if point p and d lie on opposite sides of plane through abc
 int btVoronoiSimplexSolver::pointOutsideOfPlane(const btVector3& p, const btVector3& a, const btVector3& b, const btVector3& c, const btVector3& d)
 {
-	btVector3 normal = (b-a).cross(c-a);
+	btVector3 normal = (b - a).cross(c - a);
 
-    btScalar signp = (p - a).dot(normal); // [AP AB AC]
-    btScalar signd = (d - a).dot( normal); // [AD AB AC]
+	btScalar signp = (p - a).dot(normal);  // [AP AB AC]
+	btScalar signd = (d - a).dot(normal);  // [AD AB AC]
 
 #ifdef CATCH_DEGENERATE_TETRAHEDRON
 #ifdef BT_USE_DOUBLE_PRECISION
-if (signd * signd < (btScalar(1e-8) * btScalar(1e-8)))
+	if (signd * signd < (btScalar(1e-8) * btScalar(1e-8)))
 	{
 		return -1;
 	}
 #else
 	if (signd * signd < (btScalar(1e-4) * btScalar(1e-4)))
 	{
-//		printf("affine dependent/degenerate\n");//
+		//		printf("affine dependent/degenerate\n");//
 		return -1;
 	}
 #endif
 
 #endif
 	// Points on opposite sides if expression signs are opposite
-    return signp * signd < btScalar(0.);
+	return signp * signd < btScalar(0.);
 }
 
-
-bool	btVoronoiSimplexSolver::closestPtPointTetrahedron(const btVector3& p, const btVector3& a, const btVector3& b, const btVector3& c, const btVector3& d, btSubSimplexClosestResult& finalResult)
+bool btVoronoiSimplexSolver::closestPtPointTetrahedron(const btVector3& p, const btVector3& a, const btVector3& b, const btVector3& c, const btVector3& d, btSubSimplexClosestResult& finalResult)
 {
 	btSubSimplexClosestResult tempResult;
 
-    // Start out assuming point inside all halfspaces, so closest to itself
+	// Start out assuming point inside all halfspaces, so closest to itself
 	finalResult.m_closestPointOnSimplex = p;
 	finalResult.m_usedVertices.reset();
-    finalResult.m_usedVertices.usedVertexA = true;
+	finalResult.m_usedVertices.usedVertexA = true;
 	finalResult.m_usedVertices.usedVertexB = true;
 	finalResult.m_usedVertices.usedVertexC = true;
 	finalResult.m_usedVertices.usedVertexD = true;
 
-    int pointOutsideABC = pointOutsideOfPlane(p, a, b, c, d);
+	int pointOutsideABC = pointOutsideOfPlane(p, a, b, c, d);
 	int pointOutsideACD = pointOutsideOfPlane(p, a, c, d, b);
-  	int	pointOutsideADB = pointOutsideOfPlane(p, a, d, b, c);
-	int	pointOutsideBDC = pointOutsideOfPlane(p, b, d, c, a);
+	int pointOutsideADB = pointOutsideOfPlane(p, a, d, b, c);
+	int pointOutsideBDC = pointOutsideOfPlane(p, b, d, c, a);
 
-   if (pointOutsideABC < 0 || pointOutsideACD < 0 || pointOutsideADB < 0 || pointOutsideBDC < 0)
-   {
-	   finalResult.m_degenerate = true;
-	   return false;
-   }
-
-   if (!pointOutsideABC  && !pointOutsideACD && !pointOutsideADB && !pointOutsideBDC)
-	 {
-		 return false;
-	 }
-
-
-    btScalar bestSqDist = FLT_MAX;
-    // If point outside face abc then compute closest point on abc
-	if (pointOutsideABC) 
+	if (pointOutsideABC < 0 || pointOutsideACD < 0 || pointOutsideADB < 0 || pointOutsideBDC < 0)
 	{
-        closestPtPointTriangle(p, a, b, c,tempResult);
+		finalResult.m_degenerate = true;
+		return false;
+	}
+
+	if (!pointOutsideABC && !pointOutsideACD && !pointOutsideADB && !pointOutsideBDC)
+	{
+		return false;
+	}
+
+	btScalar bestSqDist = FLT_MAX;
+	// If point outside face abc then compute closest point on abc
+	if (pointOutsideABC)
+	{
+		closestPtPointTriangle(p, a, b, c, tempResult);
 		btVector3 q = tempResult.m_closestPointOnSimplex;
-		
-        btScalar sqDist = (q - p).dot( q - p);
-        // Update best closest point if (squared) distance is less than current best
-        if (sqDist < bestSqDist) {
+
+		btScalar sqDist = (q - p).dot(q - p);
+		// Update best closest point if (squared) distance is less than current best
+		if (sqDist < bestSqDist)
+		{
 			bestSqDist = sqDist;
 			finalResult.m_closestPointOnSimplex = q;
 			//convert result bitmask!
@@ -504,25 +481,22 @@ bool	btVoronoiSimplexSolver::closestPtPointTetrahedron(const btVector3& p, const
 			finalResult.m_usedVertices.usedVertexB = tempResult.m_usedVertices.usedVertexB;
 			finalResult.m_usedVertices.usedVertexC = tempResult.m_usedVertices.usedVertexC;
 			finalResult.setBarycentricCoordinates(
-					tempResult.m_barycentricCoords[VERTA],
-					tempResult.m_barycentricCoords[VERTB],
-					tempResult.m_barycentricCoords[VERTC],
-					0
-			);
-
+				tempResult.m_barycentricCoords[VERTA],
+				tempResult.m_barycentricCoords[VERTB],
+				tempResult.m_barycentricCoords[VERTC],
+				0);
 		}
-    }
-  
+	}
 
 	// Repeat test for face acd
-	if (pointOutsideACD) 
+	if (pointOutsideACD)
 	{
-        closestPtPointTriangle(p, a, c, d,tempResult);
+		closestPtPointTriangle(p, a, c, d, tempResult);
 		btVector3 q = tempResult.m_closestPointOnSimplex;
 		//convert result bitmask!
 
-        btScalar sqDist = (q - p).dot( q - p);
-        if (sqDist < bestSqDist) 
+		btScalar sqDist = (q - p).dot(q - p);
+		if (sqDist < bestSqDist)
 		{
 			bestSqDist = sqDist;
 			finalResult.m_closestPointOnSimplex = q;
@@ -532,52 +506,46 @@ bool	btVoronoiSimplexSolver::closestPtPointTetrahedron(const btVector3& p, const
 			finalResult.m_usedVertices.usedVertexC = tempResult.m_usedVertices.usedVertexB;
 			finalResult.m_usedVertices.usedVertexD = tempResult.m_usedVertices.usedVertexC;
 			finalResult.setBarycentricCoordinates(
-					tempResult.m_barycentricCoords[VERTA],
-					0,
-					tempResult.m_barycentricCoords[VERTB],
-					tempResult.m_barycentricCoords[VERTC]
-			);
-
+				tempResult.m_barycentricCoords[VERTA],
+				0,
+				tempResult.m_barycentricCoords[VERTB],
+				tempResult.m_barycentricCoords[VERTC]);
 		}
-    }
-    // Repeat test for face adb
+	}
+	// Repeat test for face adb
 
-	
 	if (pointOutsideADB)
 	{
-		closestPtPointTriangle(p, a, d, b,tempResult);
+		closestPtPointTriangle(p, a, d, b, tempResult);
 		btVector3 q = tempResult.m_closestPointOnSimplex;
 		//convert result bitmask!
 
-        btScalar sqDist = (q - p).dot( q - p);
-        if (sqDist < bestSqDist) 
+		btScalar sqDist = (q - p).dot(q - p);
+		if (sqDist < bestSqDist)
 		{
 			bestSqDist = sqDist;
 			finalResult.m_closestPointOnSimplex = q;
 			finalResult.m_usedVertices.reset();
 			finalResult.m_usedVertices.usedVertexA = tempResult.m_usedVertices.usedVertexA;
 			finalResult.m_usedVertices.usedVertexB = tempResult.m_usedVertices.usedVertexC;
-			
+
 			finalResult.m_usedVertices.usedVertexD = tempResult.m_usedVertices.usedVertexB;
 			finalResult.setBarycentricCoordinates(
-					tempResult.m_barycentricCoords[VERTA],
-					tempResult.m_barycentricCoords[VERTC],
-					0,
-					tempResult.m_barycentricCoords[VERTB]
-			);
-
+				tempResult.m_barycentricCoords[VERTA],
+				tempResult.m_barycentricCoords[VERTC],
+				0,
+				tempResult.m_barycentricCoords[VERTB]);
 		}
-    }
-    // Repeat test for face bdc
-    
+	}
+	// Repeat test for face bdc
 
 	if (pointOutsideBDC)
 	{
-        closestPtPointTriangle(p, b, d, c,tempResult);
+		closestPtPointTriangle(p, b, d, c, tempResult);
 		btVector3 q = tempResult.m_closestPointOnSimplex;
 		//convert result bitmask!
-        btScalar sqDist = (q - p).dot( q - p);
-        if (sqDist < bestSqDist) 
+		btScalar sqDist = (q - p).dot(q - p);
+		if (sqDist < bestSqDist)
 		{
 			bestSqDist = sqDist;
 			finalResult.m_closestPointOnSimplex = q;
@@ -588,25 +556,22 @@ bool	btVoronoiSimplexSolver::closestPtPointTetrahedron(const btVector3& p, const
 			finalResult.m_usedVertices.usedVertexD = tempResult.m_usedVertices.usedVertexB;
 
 			finalResult.setBarycentricCoordinates(
-					0,
-					tempResult.m_barycentricCoords[VERTA],
-					tempResult.m_barycentricCoords[VERTC],
-					tempResult.m_barycentricCoords[VERTB]
-			);
-
+				0,
+				tempResult.m_barycentricCoords[VERTA],
+				tempResult.m_barycentricCoords[VERTC],
+				tempResult.m_barycentricCoords[VERTB]);
 		}
-    }
+	}
 
 	//help! we ended up full !
-	
+
 	if (finalResult.m_usedVertices.usedVertexA &&
 		finalResult.m_usedVertices.usedVertexB &&
 		finalResult.m_usedVertices.usedVertexC &&
-		finalResult.m_usedVertices.usedVertexD) 
+		finalResult.m_usedVertices.usedVertexD)
 	{
 		return true;
 	}
 
-    return true;
+	return true;
 }
-
diff --git a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btVoronoiSimplexSolver.h b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btVoronoiSimplexSolver.h
index 80fd490f4e1..24a0a8f2df8 100644
--- a/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btVoronoiSimplexSolver.h
+++ b/thirdparty/bullet/BulletCollision/NarrowPhaseCollision/btVoronoiSimplexSolver.h
@@ -13,15 +13,11 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #ifndef BT_VORONOI_SIMPLEX_SOLVER_H
 #define BT_VORONOI_SIMPLEX_SOLVER_H
 
 #include "btSimplexSolverInterface.h"
 
-
-
 #define VORONOI_SIMPLEX_MAX_VERTS 5
 
 ///disable next define, or use defaultCollisionConfiguration->getSimplexSolver()->setEqualVertexThreshold(0.f) to disable/configure
@@ -31,9 +27,10 @@ subject to the following restrictions:
 #define VORONOI_DEFAULT_EQUAL_VERTEX_THRESHOLD 1e-12f
 #else
 #define VORONOI_DEFAULT_EQUAL_VERTEX_THRESHOLD 0.0001f
-#endif//BT_USE_DOUBLE_PRECISION
+#endif  //BT_USE_DOUBLE_PRECISION
 
-struct btUsageBitfield{
+struct btUsageBitfield
+{
 	btUsageBitfield()
 	{
 		reset();
@@ -46,140 +43,131 @@ struct btUsageBitfield{
 		usedVertexC = false;
 		usedVertexD = false;
 	}
-	unsigned short usedVertexA	: 1;
-	unsigned short usedVertexB	: 1;
-	unsigned short usedVertexC	: 1;
-	unsigned short usedVertexD	: 1;
-	unsigned short unused1		: 1;
-	unsigned short unused2		: 1;
-	unsigned short unused3		: 1;
-	unsigned short unused4		: 1;
+	unsigned short usedVertexA : 1;
+	unsigned short usedVertexB : 1;
+	unsigned short usedVertexC : 1;
+	unsigned short usedVertexD : 1;
+	unsigned short unused1 : 1;
+	unsigned short unused2 : 1;
+	unsigned short unused3 : 1;
+	unsigned short unused4 : 1;
 };
 
-
-struct	btSubSimplexClosestResult
+struct btSubSimplexClosestResult
 {
-	btVector3	m_closestPointOnSimplex;
+	btVector3 m_closestPointOnSimplex;
 	//MASK for m_usedVertices
-	//stores the simplex vertex-usage, using the MASK, 
+	//stores the simplex vertex-usage, using the MASK,
 	// if m_usedVertices & MASK then the related vertex is used
-	btUsageBitfield	m_usedVertices;
-	btScalar	m_barycentricCoords[4];
+	btUsageBitfield m_usedVertices;
+	btScalar m_barycentricCoords[4];
 	bool m_degenerate;
 
-	void	reset()
+	void reset()
 	{
 		m_degenerate = false;
 		setBarycentricCoordinates();
 		m_usedVertices.reset();
 	}
-	bool	isValid()
+	bool isValid()
 	{
 		bool valid = (m_barycentricCoords[0] >= btScalar(0.)) &&
-			(m_barycentricCoords[1] >= btScalar(0.)) &&
-			(m_barycentricCoords[2] >= btScalar(0.)) &&
-			(m_barycentricCoords[3] >= btScalar(0.));
-
+					 (m_barycentricCoords[1] >= btScalar(0.)) &&
+					 (m_barycentricCoords[2] >= btScalar(0.)) &&
+					 (m_barycentricCoords[3] >= btScalar(0.));
 
 		return valid;
 	}
-	void	setBarycentricCoordinates(btScalar a=btScalar(0.),btScalar b=btScalar(0.),btScalar c=btScalar(0.),btScalar d=btScalar(0.))
+	void setBarycentricCoordinates(btScalar a = btScalar(0.), btScalar b = btScalar(0.), btScalar c = btScalar(0.), btScalar d = btScalar(0.))
 	{
 		m_barycentricCoords[0] = a;
 		m_barycentricCoords[1] = b;
 		m_barycentricCoords[2] = c;
 		m_barycentricCoords[3] = d;
 	}
-
 };
 
 /// btVoronoiSimplexSolver is an implementation of the closest point distance algorithm from a 1-4 points simplex to the origin.
 /// Can be used with GJK, as an alternative to Johnson distance algorithm.
 #ifdef NO_VIRTUAL_INTERFACE
-ATTRIBUTE_ALIGNED16(class) btVoronoiSimplexSolver
+ATTRIBUTE_ALIGNED16(class)
+btVoronoiSimplexSolver
 #else
-ATTRIBUTE_ALIGNED16(class) btVoronoiSimplexSolver : public btSimplexSolverInterface
+ATTRIBUTE_ALIGNED16(class)
+btVoronoiSimplexSolver : public btSimplexSolverInterface
 #endif
 {
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-	int	m_numVertices;
+	int m_numVertices;
 
-	btVector3	m_simplexVectorW[VORONOI_SIMPLEX_MAX_VERTS];
-	btVector3	m_simplexPointsP[VORONOI_SIMPLEX_MAX_VERTS];
-	btVector3	m_simplexPointsQ[VORONOI_SIMPLEX_MAX_VERTS];
+	btVector3 m_simplexVectorW[VORONOI_SIMPLEX_MAX_VERTS];
+	btVector3 m_simplexPointsP[VORONOI_SIMPLEX_MAX_VERTS];
+	btVector3 m_simplexPointsQ[VORONOI_SIMPLEX_MAX_VERTS];
 
-	
-
-	btVector3	m_cachedP1;
-	btVector3	m_cachedP2;
-	btVector3	m_cachedV;
-	btVector3	m_lastW;
-	
-	btScalar	m_equalVertexThreshold;
-	bool		m_cachedValidClosest;
+	btVector3 m_cachedP1;
+	btVector3 m_cachedP2;
+	btVector3 m_cachedV;
+	btVector3 m_lastW;
 
+	btScalar m_equalVertexThreshold;
+	bool m_cachedValidClosest;
 
 	btSubSimplexClosestResult m_cachedBC;
 
-	bool	m_needsUpdate;
-	
-	void	removeVertex(int index);
-	void	reduceVertices (const btUsageBitfield& usedVerts);
-	bool	updateClosestVectorAndPoints();
+	bool m_needsUpdate;
 
-	bool	closestPtPointTetrahedron(const btVector3& p, const btVector3& a, const btVector3& b, const btVector3& c, const btVector3& d, btSubSimplexClosestResult& finalResult);
-	int		pointOutsideOfPlane(const btVector3& p, const btVector3& a, const btVector3& b, const btVector3& c, const btVector3& d);
-	bool	closestPtPointTriangle(const btVector3& p, const btVector3& a, const btVector3& b, const btVector3& c,btSubSimplexClosestResult& result);
+	void removeVertex(int index);
+	void reduceVertices(const btUsageBitfield& usedVerts);
+	bool updateClosestVectorAndPoints();
+
+	bool closestPtPointTetrahedron(const btVector3& p, const btVector3& a, const btVector3& b, const btVector3& c, const btVector3& d, btSubSimplexClosestResult& finalResult);
+	int pointOutsideOfPlane(const btVector3& p, const btVector3& a, const btVector3& b, const btVector3& c, const btVector3& d);
+	bool closestPtPointTriangle(const btVector3& p, const btVector3& a, const btVector3& b, const btVector3& c, btSubSimplexClosestResult& result);
 
 public:
-
 	btVoronoiSimplexSolver()
-		:  m_equalVertexThreshold(VORONOI_DEFAULT_EQUAL_VERTEX_THRESHOLD)
+		: m_equalVertexThreshold(VORONOI_DEFAULT_EQUAL_VERTEX_THRESHOLD)
 	{
 	}
-	 void reset();
+	void reset();
 
-	 void addVertex(const btVector3& w, const btVector3& p, const btVector3& q);
+	void addVertex(const btVector3& w, const btVector3& p, const btVector3& q);
 
-	 void	setEqualVertexThreshold(btScalar threshold)
-	 {
-		 m_equalVertexThreshold = threshold;
-	 }
+	void setEqualVertexThreshold(btScalar threshold)
+	{
+		m_equalVertexThreshold = threshold;
+	}
 
-	 btScalar	getEqualVertexThreshold() const
-	 {
-		 return m_equalVertexThreshold;
-	 }
+	btScalar getEqualVertexThreshold() const
+	{
+		return m_equalVertexThreshold;
+	}
 
-	 bool closest(btVector3& v);
+	bool closest(btVector3 & v);
 
-	 btScalar maxVertex();
+	btScalar maxVertex();
 
-	 bool fullSimplex() const
-	 {
-		 return (m_numVertices == 4);
-	 }
+	bool fullSimplex() const
+	{
+		return (m_numVertices == 4);
+	}
 
-	 int getSimplex(btVector3 *pBuf, btVector3 *qBuf, btVector3 *yBuf) const;
+	int getSimplex(btVector3 * pBuf, btVector3 * qBuf, btVector3 * yBuf) const;
 
-	 bool inSimplex(const btVector3& w);
-	
-	 void backup_closest(btVector3& v) ;
+	bool inSimplex(const btVector3& w);
 
-	 bool emptySimplex() const ;
+	void backup_closest(btVector3 & v);
 
-	 void compute_points(btVector3& p1, btVector3& p2) ;
-
-	 int numVertices() const 
-	 {
-		 return m_numVertices;
-	 }
+	bool emptySimplex() const;
 
+	void compute_points(btVector3 & p1, btVector3 & p2);
 
+	int numVertices() const
+	{
+		return m_numVertices;
+	}
 };
 
-#endif //BT_VORONOI_SIMPLEX_SOLVER_H
-
+#endif  //BT_VORONOI_SIMPLEX_SOLVER_H
diff --git a/thirdparty/bullet/BulletDynamics/Character/btCharacterControllerInterface.h b/thirdparty/bullet/BulletDynamics/Character/btCharacterControllerInterface.h
index abe24b5ca6c..2ccf317b927 100644
--- a/thirdparty/bullet/BulletDynamics/Character/btCharacterControllerInterface.h
+++ b/thirdparty/bullet/BulletDynamics/Character/btCharacterControllerInterface.h
@@ -26,22 +26,21 @@ class btCollisionWorld;
 class btCharacterControllerInterface : public btActionInterface
 {
 public:
-	btCharacterControllerInterface () {};
-	virtual ~btCharacterControllerInterface () {};
-	
-	virtual void	setWalkDirection(const btVector3& walkDirection) = 0;
-	virtual void	setVelocityForTimeInterval(const btVector3& velocity, btScalar timeInterval) = 0;
-	virtual void	reset ( btCollisionWorld* collisionWorld ) = 0;
-	virtual void	warp (const btVector3& origin) = 0;
+	btCharacterControllerInterface(){};
+	virtual ~btCharacterControllerInterface(){};
 
-	virtual void	preStep ( btCollisionWorld* collisionWorld) = 0;
-	virtual void	playerStep (btCollisionWorld* collisionWorld, btScalar dt) = 0;
-	virtual bool	canJump () const = 0;
-	virtual void	jump(const btVector3& dir = btVector3(0, 0, 0)) = 0;
+	virtual void setWalkDirection(const btVector3& walkDirection) = 0;
+	virtual void setVelocityForTimeInterval(const btVector3& velocity, btScalar timeInterval) = 0;
+	virtual void reset(btCollisionWorld* collisionWorld) = 0;
+	virtual void warp(const btVector3& origin) = 0;
 
-	virtual bool	onGround () const = 0;
-	virtual void	setUpInterpolate (bool value) = 0;
+	virtual void preStep(btCollisionWorld* collisionWorld) = 0;
+	virtual void playerStep(btCollisionWorld* collisionWorld, btScalar dt) = 0;
+	virtual bool canJump() const = 0;
+	virtual void jump(const btVector3& dir = btVector3(0, 0, 0)) = 0;
+
+	virtual bool onGround() const = 0;
+	virtual void setUpInterpolate(bool value) = 0;
 };
 
-#endif //BT_CHARACTER_CONTROLLER_INTERFACE_H
-
+#endif  //BT_CHARACTER_CONTROLLER_INTERFACE_H
diff --git a/thirdparty/bullet/BulletDynamics/Character/btKinematicCharacterController.cpp b/thirdparty/bullet/BulletDynamics/Character/btKinematicCharacterController.cpp
index cb1aa71a14e..2bbccb291c7 100644
--- a/thirdparty/bullet/BulletDynamics/Character/btKinematicCharacterController.cpp
+++ b/thirdparty/bullet/BulletDynamics/Character/btKinematicCharacterController.cpp
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include <stdio.h>
 #include "LinearMath/btIDebugDraw.h"
 #include "BulletCollision/CollisionDispatch/btGhostObject.h"
@@ -24,20 +23,19 @@ subject to the following restrictions:
 #include "LinearMath/btDefaultMotionState.h"
 #include "btKinematicCharacterController.h"
 
-
 // static helper method
 static btVector3
 getNormalizedVector(const btVector3& v)
 {
 	btVector3 n(0, 0, 0);
 
-	if (v.length() > SIMD_EPSILON) {
+	if (v.length() > SIMD_EPSILON)
+	{
 		n = v.normalized();
 	}
 	return n;
 }
 
-
 ///@todo Interact with dynamic objects,
 ///Ride kinematicly animated platforms properly
 ///More realistic (or maybe just a config option) falling
@@ -47,18 +45,19 @@ getNormalizedVector(const btVector3& v)
 class btKinematicClosestNotMeRayResultCallback : public btCollisionWorld::ClosestRayResultCallback
 {
 public:
-	btKinematicClosestNotMeRayResultCallback (btCollisionObject* me) : btCollisionWorld::ClosestRayResultCallback(btVector3(0.0, 0.0, 0.0), btVector3(0.0, 0.0, 0.0))
+	btKinematicClosestNotMeRayResultCallback(btCollisionObject* me) : btCollisionWorld::ClosestRayResultCallback(btVector3(0.0, 0.0, 0.0), btVector3(0.0, 0.0, 0.0))
 	{
 		m_me = me;
 	}
 
-	virtual btScalar addSingleResult(btCollisionWorld::LocalRayResult& rayResult,bool normalInWorldSpace)
+	virtual btScalar addSingleResult(btCollisionWorld::LocalRayResult& rayResult, bool normalInWorldSpace)
 	{
 		if (rayResult.m_collisionObject == m_me)
 			return 1.0;
 
-		return ClosestRayResultCallback::addSingleResult (rayResult, normalInWorldSpace);
+		return ClosestRayResultCallback::addSingleResult(rayResult, normalInWorldSpace);
 	}
+
 protected:
 	btCollisionObject* m_me;
 };
@@ -66,15 +65,12 @@ protected:
 class btKinematicClosestNotMeConvexResultCallback : public btCollisionWorld::ClosestConvexResultCallback
 {
 public:
-	btKinematicClosestNotMeConvexResultCallback (btCollisionObject* me, const btVector3& up, btScalar minSlopeDot)
-	: btCollisionWorld::ClosestConvexResultCallback(btVector3(0.0, 0.0, 0.0), btVector3(0.0, 0.0, 0.0))
-	, m_me(me)
-	, m_up(up)
-	, m_minSlopeDot(minSlopeDot)
+	btKinematicClosestNotMeConvexResultCallback(btCollisionObject* me, const btVector3& up, btScalar minSlopeDot)
+		: btCollisionWorld::ClosestConvexResultCallback(btVector3(0.0, 0.0, 0.0), btVector3(0.0, 0.0, 0.0)), m_me(me), m_up(up), m_minSlopeDot(minSlopeDot)
 	{
 	}
 
-	virtual btScalar addSingleResult(btCollisionWorld::LocalConvexResult& convexResult,bool normalInWorldSpace)
+	virtual btScalar addSingleResult(btCollisionWorld::LocalConvexResult& convexResult, bool normalInWorldSpace)
 	{
 		if (convexResult.m_hitCollisionObject == m_me)
 			return btScalar(1.0);
@@ -86,19 +82,22 @@ public:
 		if (normalInWorldSpace)
 		{
 			hitNormalWorld = convexResult.m_hitNormalLocal;
-		} else
+		}
+		else
 		{
 			///need to transform normal into worldspace
-			hitNormalWorld = convexResult.m_hitCollisionObject->getWorldTransform().getBasis()*convexResult.m_hitNormalLocal;
+			hitNormalWorld = convexResult.m_hitCollisionObject->getWorldTransform().getBasis() * convexResult.m_hitNormalLocal;
 		}
 
 		btScalar dotUp = m_up.dot(hitNormalWorld);
-		if (dotUp < m_minSlopeDot) {
+		if (dotUp < m_minSlopeDot)
+		{
 			return btScalar(1.0);
 		}
 
-		return ClosestConvexResultCallback::addSingleResult (convexResult, normalInWorldSpace);
+		return ClosestConvexResultCallback::addSingleResult(convexResult, normalInWorldSpace);
 	}
+
 protected:
 	btCollisionObject* m_me;
 	const btVector3 m_up;
@@ -110,7 +109,7 @@ protected:
  *
  * from: http://www-cs-students.stanford.edu/~adityagp/final/node3.html
  */
-btVector3 btKinematicCharacterController::computeReflectionDirection (const btVector3& direction, const btVector3& normal)
+btVector3 btKinematicCharacterController::computeReflectionDirection(const btVector3& direction, const btVector3& normal)
 {
 	return direction - (btScalar(2.0) * direction.dot(normal)) * normal;
 }
@@ -118,7 +117,7 @@ btVector3 btKinematicCharacterController::computeReflectionDirection (const btVe
 /*
  * Returns the portion of 'direction' that is parallel to 'normal'
  */
-btVector3 btKinematicCharacterController::parallelComponent (const btVector3& direction, const btVector3& normal)
+btVector3 btKinematicCharacterController::parallelComponent(const btVector3& direction, const btVector3& normal)
 {
 	btScalar magnitude = direction.dot(normal);
 	return normal * magnitude;
@@ -127,29 +126,29 @@ btVector3 btKinematicCharacterController::parallelComponent (const btVector3& di
 /*
  * Returns the portion of 'direction' that is perpindicular to 'normal'
  */
-btVector3 btKinematicCharacterController::perpindicularComponent (const btVector3& direction, const btVector3& normal)
+btVector3 btKinematicCharacterController::perpindicularComponent(const btVector3& direction, const btVector3& normal)
 {
 	return direction - parallelComponent(direction, normal);
 }
 
-btKinematicCharacterController::btKinematicCharacterController (btPairCachingGhostObject* ghostObject,btConvexShape* convexShape,btScalar stepHeight, const btVector3& up)
+btKinematicCharacterController::btKinematicCharacterController(btPairCachingGhostObject* ghostObject, btConvexShape* convexShape, btScalar stepHeight, const btVector3& up)
 {
 	m_ghostObject = ghostObject;
 	m_up.setValue(0.0f, 0.0f, 1.0f);
 	m_jumpAxis.setValue(0.0f, 0.0f, 1.0f);
 	m_addedMargin = 0.02;
-	m_walkDirection.setValue(0.0,0.0,0.0);
+	m_walkDirection.setValue(0.0, 0.0, 0.0);
 	m_AngVel.setValue(0.0, 0.0, 0.0);
-	m_useGhostObjectSweepTest = true;	
+	m_useGhostObjectSweepTest = true;
 	m_turnAngle = btScalar(0.0);
-	m_convexShape=convexShape;	
-	m_useWalkDirection = true;	// use walk direction by default, legacy behavior
+	m_convexShape = convexShape;
+	m_useWalkDirection = true;  // use walk direction by default, legacy behavior
 	m_velocityTimeInterval = 0.0;
 	m_verticalVelocity = 0.0;
 	m_verticalOffset = 0.0;
-	m_gravity = 9.8 * 3.0 ; // 3G acceleration.
-	m_fallSpeed = 55.0; // Terminal velocity of a sky diver in m/s.
-	m_jumpSpeed = 10.0; // ?
+	m_gravity = 9.8 * 3.0;  // 3G acceleration.
+	m_fallSpeed = 55.0;     // Terminal velocity of a sky diver in m/s.
+	m_jumpSpeed = 10.0;     // ?
 	m_SetjumpSpeed = m_jumpSpeed;
 	m_wasOnGround = false;
 	m_wasJumping = false;
@@ -166,7 +165,7 @@ btKinematicCharacterController::btKinematicCharacterController (btPairCachingGho
 	setMaxSlope(btRadians(45.0));
 }
 
-btKinematicCharacterController::~btKinematicCharacterController ()
+btKinematicCharacterController::~btKinematicCharacterController()
 {
 }
 
@@ -175,7 +174,7 @@ btPairCachingGhostObject* btKinematicCharacterController::getGhostObject()
 	return m_ghostObject;
 }
 
-bool btKinematicCharacterController::recoverFromPenetration ( btCollisionWorld* collisionWorld)
+bool btKinematicCharacterController::recoverFromPenetration(btCollisionWorld* collisionWorld)
 {
 	// Here we must refresh the overlapping paircache as the penetrating movement itself or the
 	// previous recovery iteration might have used setWorldTransform and pushed us into an object
@@ -186,19 +185,19 @@ bool btKinematicCharacterController::recoverFromPenetration ( btCollisionWorld*
 	// paircache and the ghostobject's internal paircache at the same time.    /BW
 
 	btVector3 minAabb, maxAabb;
-	m_convexShape->getAabb(m_ghostObject->getWorldTransform(), minAabb,maxAabb);
-	collisionWorld->getBroadphase()->setAabb(m_ghostObject->getBroadphaseHandle(), 
-						 minAabb, 
-						 maxAabb, 
-						 collisionWorld->getDispatcher());
-						 
+	m_convexShape->getAabb(m_ghostObject->getWorldTransform(), minAabb, maxAabb);
+	collisionWorld->getBroadphase()->setAabb(m_ghostObject->getBroadphaseHandle(),
+											 minAabb,
+											 maxAabb,
+											 collisionWorld->getDispatcher());
+
 	bool penetration = false;
 
 	collisionWorld->getDispatcher()->dispatchAllCollisionPairs(m_ghostObject->getOverlappingPairCache(), collisionWorld->getDispatchInfo(), collisionWorld->getDispatcher());
 
 	m_currentPosition = m_ghostObject->getWorldTransform().getOrigin();
-	
-//	btScalar maxPen = btScalar(0.0);
+
+	//	btScalar maxPen = btScalar(0.0);
 	for (int i = 0; i < m_ghostObject->getOverlappingPairCache()->getNumOverlappingPairs(); i++)
 	{
 		m_manifoldArray.resize(0);
@@ -206,25 +205,24 @@ bool btKinematicCharacterController::recoverFromPenetration ( btCollisionWorld*
 		btBroadphasePair* collisionPair = &m_ghostObject->getOverlappingPairCache()->getOverlappingPairArray()[i];
 
 		btCollisionObject* obj0 = static_cast<btCollisionObject*>(collisionPair->m_pProxy0->m_clientObject);
-        btCollisionObject* obj1 = static_cast<btCollisionObject*>(collisionPair->m_pProxy1->m_clientObject);
+		btCollisionObject* obj1 = static_cast<btCollisionObject*>(collisionPair->m_pProxy1->m_clientObject);
 
 		if ((obj0 && !obj0->hasContactResponse()) || (obj1 && !obj1->hasContactResponse()))
 			continue;
 
 		if (!needsCollision(obj0, obj1))
 			continue;
-		
+
 		if (collisionPair->m_algorithm)
 			collisionPair->m_algorithm->getAllContactManifolds(m_manifoldArray);
 
-		
-		for (int j=0;j<m_manifoldArray.size();j++)
+		for (int j = 0; j < m_manifoldArray.size(); j++)
 		{
 			btPersistentManifold* manifold = m_manifoldArray[j];
 			btScalar directionSign = manifold->getBody0() == m_ghostObject ? btScalar(-1.0) : btScalar(1.0);
-			for (int p=0;p<manifold->getNumContacts();p++)
+			for (int p = 0; p < manifold->getNumContacts(); p++)
 			{
-				const btManifoldPoint&pt = manifold->getContactPoint(p);
+				const btManifoldPoint& pt = manifold->getContactPoint(p);
 
 				btScalar dist = pt.getDistance();
 
@@ -239,22 +237,24 @@ bool btKinematicCharacterController::recoverFromPenetration ( btCollisionWorld*
 					//}
 					m_currentPosition += pt.m_normalWorldOnB * directionSign * dist * btScalar(0.2);
 					penetration = true;
-				} else {
+				}
+				else
+				{
 					//printf("touching %f\n", dist);
 				}
 			}
-			
+
 			//manifold->clearManifold();
 		}
 	}
 	btTransform newTrans = m_ghostObject->getWorldTransform();
 	newTrans.setOrigin(m_currentPosition);
 	m_ghostObject->setWorldTransform(newTrans);
-//	printf("m_touchingNormal = %f,%f,%f\n",m_touchingNormal[0],m_touchingNormal[1],m_touchingNormal[2]);
+	//	printf("m_touchingNormal = %f,%f,%f\n",m_touchingNormal[0],m_touchingNormal[1],m_touchingNormal[2]);
 	return penetration;
 }
 
-void btKinematicCharacterController::stepUp ( btCollisionWorld* world)
+void btKinematicCharacterController::stepUp(btCollisionWorld* world)
 {
 	btScalar stepHeight = 0.0f;
 	if (m_verticalVelocity < 0.0)
@@ -263,8 +263,8 @@ void btKinematicCharacterController::stepUp ( btCollisionWorld* world)
 	// phase 1: up
 	btTransform start, end;
 
-	start.setIdentity ();
-	end.setIdentity ();
+	start.setIdentity();
+	end.setIdentity();
 
 	/* FIXME: Handle penetration properly */
 	start.setOrigin(m_currentPosition);
@@ -272,7 +272,7 @@ void btKinematicCharacterController::stepUp ( btCollisionWorld* world)
 	m_targetPosition = m_currentPosition + m_up * (stepHeight) + m_jumpAxis * ((m_verticalOffset > 0.f ? m_verticalOffset : 0.f));
 	m_currentPosition = m_targetPosition;
 
-	end.setOrigin (m_targetPosition);
+	end.setOrigin(m_targetPosition);
 
 	start.setRotation(m_currentOrientation);
 	end.setRotation(m_targetOrientation);
@@ -280,10 +280,10 @@ void btKinematicCharacterController::stepUp ( btCollisionWorld* world)
 	btKinematicClosestNotMeConvexResultCallback callback(m_ghostObject, -m_up, m_maxSlopeCosine);
 	callback.m_collisionFilterGroup = getGhostObject()->getBroadphaseHandle()->m_collisionFilterGroup;
 	callback.m_collisionFilterMask = getGhostObject()->getBroadphaseHandle()->m_collisionFilterMask;
-	
+
 	if (m_useGhostObjectSweepTest)
 	{
-		m_ghostObject->convexSweepTest (m_convexShape, start, end, callback, world->getDispatchInfo().m_allowedCcdPenetration);
+		m_ghostObject->convexSweepTest(m_convexShape, start, end, callback, world->getDispatchInfo().m_allowedCcdPenetration);
 	}
 	else
 	{
@@ -298,7 +298,7 @@ void btKinematicCharacterController::stepUp ( btCollisionWorld* world)
 			// we moved up only a fraction of the step height
 			m_currentStepOffset = stepHeight * callback.m_closestHitFraction;
 			if (m_interpolateUp == true)
-				m_currentPosition.setInterpolate3 (m_currentPosition, m_targetPosition, callback.m_closestHitFraction);
+				m_currentPosition.setInterpolate3(m_currentPosition, m_targetPosition, callback.m_closestHitFraction);
 			else
 				m_currentPosition = m_targetPosition;
 		}
@@ -329,7 +329,9 @@ void btKinematicCharacterController::stepUp ( btCollisionWorld* world)
 			m_verticalVelocity = 0.0;
 			m_currentStepOffset = m_stepHeight;
 		}
-	} else {
+	}
+	else
+	{
 		m_currentStepOffset = stepHeight;
 		m_currentPosition = m_targetPosition;
 	}
@@ -342,43 +344,44 @@ bool btKinematicCharacterController::needsCollision(const btCollisionObject* bod
 	return collides;
 }
 
-void btKinematicCharacterController::updateTargetPositionBasedOnCollision (const btVector3& hitNormal, btScalar tangentMag, btScalar normalMag)
+void btKinematicCharacterController::updateTargetPositionBasedOnCollision(const btVector3& hitNormal, btScalar tangentMag, btScalar normalMag)
 {
 	btVector3 movementDirection = m_targetPosition - m_currentPosition;
 	btScalar movementLength = movementDirection.length();
-	if (movementLength>SIMD_EPSILON)
+	if (movementLength > SIMD_EPSILON)
 	{
 		movementDirection.normalize();
 
-		btVector3 reflectDir = computeReflectionDirection (movementDirection, hitNormal);
+		btVector3 reflectDir = computeReflectionDirection(movementDirection, hitNormal);
 		reflectDir.normalize();
 
 		btVector3 parallelDir, perpindicularDir;
 
-		parallelDir = parallelComponent (reflectDir, hitNormal);
-		perpindicularDir = perpindicularComponent (reflectDir, hitNormal);
+		parallelDir = parallelComponent(reflectDir, hitNormal);
+		perpindicularDir = perpindicularComponent(reflectDir, hitNormal);
 
 		m_targetPosition = m_currentPosition;
-		if (0)//tangentMag != 0.0)
+		if (0)  //tangentMag != 0.0)
 		{
-			btVector3 parComponent = parallelDir * btScalar (tangentMag*movementLength);
-//			printf("parComponent=%f,%f,%f\n",parComponent[0],parComponent[1],parComponent[2]);
-			m_targetPosition +=  parComponent;
+			btVector3 parComponent = parallelDir * btScalar(tangentMag * movementLength);
+			//			printf("parComponent=%f,%f,%f\n",parComponent[0],parComponent[1],parComponent[2]);
+			m_targetPosition += parComponent;
 		}
 
 		if (normalMag != 0.0)
 		{
-			btVector3 perpComponent = perpindicularDir * btScalar (normalMag*movementLength);
-//			printf("perpComponent=%f,%f,%f\n",perpComponent[0],perpComponent[1],perpComponent[2]);
+			btVector3 perpComponent = perpindicularDir * btScalar(normalMag * movementLength);
+			//			printf("perpComponent=%f,%f,%f\n",perpComponent[0],perpComponent[1],perpComponent[2]);
 			m_targetPosition += perpComponent;
 		}
-	} else
+	}
+	else
 	{
-//		printf("movementLength don't normalize a zero vector\n");
+		//		printf("movementLength don't normalize a zero vector\n");
 	}
 }
 
-void btKinematicCharacterController::stepForwardAndStrafe ( btCollisionWorld* collisionWorld, const btVector3& walkMove)
+void btKinematicCharacterController::stepForwardAndStrafe(btCollisionWorld* collisionWorld, const btVector3& walkMove)
 {
 	// printf("m_normalizedDirection=%f,%f,%f\n",
 	// 	m_normalizedDirection[0],m_normalizedDirection[1],m_normalizedDirection[2]);
@@ -387,29 +390,28 @@ void btKinematicCharacterController::stepForwardAndStrafe ( btCollisionWorld* co
 
 	m_targetPosition = m_currentPosition + walkMove;
 
-	start.setIdentity ();
-	end.setIdentity ();
-	
+	start.setIdentity();
+	end.setIdentity();
+
 	btScalar fraction = 1.0;
-	btScalar distance2 = (m_currentPosition-m_targetPosition).length2();
-//	printf("distance2=%f\n",distance2);
+	btScalar distance2 = (m_currentPosition - m_targetPosition).length2();
+	//	printf("distance2=%f\n",distance2);
 
 	int maxIter = 10;
 
 	while (fraction > btScalar(0.01) && maxIter-- > 0)
 	{
-		start.setOrigin (m_currentPosition);
-		end.setOrigin (m_targetPosition);
+		start.setOrigin(m_currentPosition);
+		end.setOrigin(m_targetPosition);
 		btVector3 sweepDirNegative(m_currentPosition - m_targetPosition);
 
 		start.setRotation(m_currentOrientation);
 		end.setRotation(m_targetOrientation);
 
-		btKinematicClosestNotMeConvexResultCallback callback (m_ghostObject, sweepDirNegative, btScalar(0.0));
+		btKinematicClosestNotMeConvexResultCallback callback(m_ghostObject, sweepDirNegative, btScalar(0.0));
 		callback.m_collisionFilterGroup = getGhostObject()->getBroadphaseHandle()->m_collisionFilterGroup;
 		callback.m_collisionFilterMask = getGhostObject()->getBroadphaseHandle()->m_collisionFilterMask;
 
-
 		btScalar margin = m_convexShape->getMargin();
 		m_convexShape->setMargin(margin + m_addedMargin);
 
@@ -426,18 +428,17 @@ void btKinematicCharacterController::stepForwardAndStrafe ( btCollisionWorld* co
 		}
 		m_convexShape->setMargin(margin);
 
-		
 		fraction -= callback.m_closestHitFraction;
 
 		if (callback.hasHit() && m_ghostObject->hasContactResponse() && needsCollision(m_ghostObject, callback.m_hitCollisionObject))
-		{	
+		{
 			// we moved only a fraction
 			//btScalar hitDistance;
 			//hitDistance = (callback.m_hitPointWorld - m_currentPosition).length();
 
-//			m_currentPosition.setInterpolate3 (m_currentPosition, m_targetPosition, callback.m_closestHitFraction);
+			//			m_currentPosition.setInterpolate3 (m_currentPosition, m_targetPosition, callback.m_closestHitFraction);
 
-			updateTargetPositionBasedOnCollision (callback.m_hitNormalWorld);
+			updateTargetPositionBasedOnCollision(callback.m_hitNormalWorld);
 			btVector3 currentDir = m_targetPosition - m_currentPosition;
 			distance2 = currentDir.length2();
 			if (distance2 > SIMD_EPSILON)
@@ -448,21 +449,21 @@ void btKinematicCharacterController::stepForwardAndStrafe ( btCollisionWorld* co
 				{
 					break;
 				}
-			} else
+			}
+			else
 			{
-//				printf("currentDir: don't normalize a zero vector\n");
+				//				printf("currentDir: don't normalize a zero vector\n");
 				break;
 			}
-
 		}
-        else
-        {
-            m_currentPosition = m_targetPosition;
+		else
+		{
+			m_currentPosition = m_targetPosition;
 		}
 	}
 }
 
-void btKinematicCharacterController::stepDown ( btCollisionWorld* collisionWorld, btScalar dt)
+void btKinematicCharacterController::stepDown(btCollisionWorld* collisionWorld, btScalar dt)
 {
 	btTransform start, end, end_double;
 	bool runonce = false;
@@ -475,64 +476,64 @@ void btKinematicCharacterController::stepDown ( btCollisionWorld* collisionWorld
 	m_targetPosition -= (step_drop + gravity_drop);*/
 
 	btVector3 orig_position = m_targetPosition;
-	
-	btScalar downVelocity = (m_verticalVelocity<0.f?-m_verticalVelocity:0.f) * dt;
+
+	btScalar downVelocity = (m_verticalVelocity < 0.f ? -m_verticalVelocity : 0.f) * dt;
 
 	if (m_verticalVelocity > 0.0)
 		return;
 
-	if(downVelocity > 0.0 && downVelocity > m_fallSpeed
-		&& (m_wasOnGround || !m_wasJumping))
+	if (downVelocity > 0.0 && downVelocity > m_fallSpeed && (m_wasOnGround || !m_wasJumping))
 		downVelocity = m_fallSpeed;
 
 	btVector3 step_drop = m_up * (m_currentStepOffset + downVelocity);
 	m_targetPosition -= step_drop;
 
 	btKinematicClosestNotMeConvexResultCallback callback(m_ghostObject, m_up, m_maxSlopeCosine);
-        callback.m_collisionFilterGroup = getGhostObject()->getBroadphaseHandle()->m_collisionFilterGroup;
-        callback.m_collisionFilterMask = getGhostObject()->getBroadphaseHandle()->m_collisionFilterMask;
+	callback.m_collisionFilterGroup = getGhostObject()->getBroadphaseHandle()->m_collisionFilterGroup;
+	callback.m_collisionFilterMask = getGhostObject()->getBroadphaseHandle()->m_collisionFilterMask;
 
 	btKinematicClosestNotMeConvexResultCallback callback2(m_ghostObject, m_up, m_maxSlopeCosine);
-        callback2.m_collisionFilterGroup = getGhostObject()->getBroadphaseHandle()->m_collisionFilterGroup;
-        callback2.m_collisionFilterMask = getGhostObject()->getBroadphaseHandle()->m_collisionFilterMask;
+	callback2.m_collisionFilterGroup = getGhostObject()->getBroadphaseHandle()->m_collisionFilterGroup;
+	callback2.m_collisionFilterMask = getGhostObject()->getBroadphaseHandle()->m_collisionFilterMask;
 
 	while (1)
 	{
-		start.setIdentity ();
-		end.setIdentity ();
+		start.setIdentity();
+		end.setIdentity();
 
-		end_double.setIdentity ();
+		end_double.setIdentity();
 
-		start.setOrigin (m_currentPosition);
-		end.setOrigin (m_targetPosition);
+		start.setOrigin(m_currentPosition);
+		end.setOrigin(m_targetPosition);
 
 		start.setRotation(m_currentOrientation);
 		end.setRotation(m_targetOrientation);
 
 		//set double test for 2x the step drop, to check for a large drop vs small drop
-		end_double.setOrigin (m_targetPosition - step_drop);
+		end_double.setOrigin(m_targetPosition - step_drop);
 
 		if (m_useGhostObjectSweepTest)
 		{
-			m_ghostObject->convexSweepTest (m_convexShape, start, end, callback, collisionWorld->getDispatchInfo().m_allowedCcdPenetration);
+			m_ghostObject->convexSweepTest(m_convexShape, start, end, callback, collisionWorld->getDispatchInfo().m_allowedCcdPenetration);
 
 			if (!callback.hasHit() && m_ghostObject->hasContactResponse())
 			{
 				//test a double fall height, to see if the character should interpolate it's fall (full) or not (partial)
-				m_ghostObject->convexSweepTest (m_convexShape, start, end_double, callback2, collisionWorld->getDispatchInfo().m_allowedCcdPenetration);
+				m_ghostObject->convexSweepTest(m_convexShape, start, end_double, callback2, collisionWorld->getDispatchInfo().m_allowedCcdPenetration);
 			}
-		} else
+		}
+		else
 		{
-			collisionWorld->convexSweepTest (m_convexShape, start, end, callback, collisionWorld->getDispatchInfo().m_allowedCcdPenetration);
+			collisionWorld->convexSweepTest(m_convexShape, start, end, callback, collisionWorld->getDispatchInfo().m_allowedCcdPenetration);
 
 			if (!callback.hasHit() && m_ghostObject->hasContactResponse())
 			{
 				//test a double fall height, to see if the character should interpolate it's fall (large) or not (small)
-				collisionWorld->convexSweepTest (m_convexShape, start, end_double, callback2, collisionWorld->getDispatchInfo().m_allowedCcdPenetration);
+				collisionWorld->convexSweepTest(m_convexShape, start, end_double, callback2, collisionWorld->getDispatchInfo().m_allowedCcdPenetration);
 			}
 		}
-	
-		btScalar downVelocity2 = (m_verticalVelocity<0.f?-m_verticalVelocity:0.f) * dt;
+
+		btScalar downVelocity2 = (m_verticalVelocity < 0.f ? -m_verticalVelocity : 0.f) * dt;
 		bool has_hit;
 		if (bounce_fix == true)
 			has_hit = (callback.hasHit() || callback2.hasHit()) && m_ghostObject->hasContactResponse() && needsCollision(m_ghostObject, callback.m_hitCollisionObject);
@@ -543,8 +544,7 @@ void btKinematicCharacterController::stepDown ( btCollisionWorld* collisionWorld
 		if (m_verticalVelocity < 0.0)
 			stepHeight = m_stepHeight;
 
-		if (downVelocity2 > 0.0 && downVelocity2 < stepHeight && has_hit == true && runonce == false
-					&& (m_wasOnGround || !m_wasJumping))
+		if (downVelocity2 > 0.0 && downVelocity2 < stepHeight && has_hit == true && runonce == false && (m_wasOnGround || !m_wasJumping))
 		{
 			//redo the velocity calculation when falling a small amount, for fast stairs motion
 			//for larger falls, use the smoother/slower interpolated movement by not touching the target position
@@ -555,7 +555,7 @@ void btKinematicCharacterController::stepDown ( btCollisionWorld* collisionWorld
 			step_drop = m_up * (m_currentStepOffset + downVelocity);
 			m_targetPosition -= step_drop;
 			runonce = true;
-			continue; //re-run previous tests
+			continue;  //re-run previous tests
 		}
 		break;
 	}
@@ -570,30 +570,32 @@ void btKinematicCharacterController::stepDown ( btCollisionWorld* collisionWorld
 		if (bounce_fix == true)
 		{
 			if (full_drop == true)
-				m_currentPosition.setInterpolate3 (m_currentPosition, m_targetPosition, callback.m_closestHitFraction);
-            else
+				m_currentPosition.setInterpolate3(m_currentPosition, m_targetPosition, callback.m_closestHitFraction);
+			else
 				//due to errors in the closestHitFraction variable when used with large polygons, calculate the hit fraction manually
-				m_currentPosition.setInterpolate3 (m_currentPosition, m_targetPosition, fraction);
+				m_currentPosition.setInterpolate3(m_currentPosition, m_targetPosition, fraction);
 		}
 		else
-			m_currentPosition.setInterpolate3 (m_currentPosition, m_targetPosition, callback.m_closestHitFraction);
+			m_currentPosition.setInterpolate3(m_currentPosition, m_targetPosition, callback.m_closestHitFraction);
 
 		full_drop = false;
 
 		m_verticalVelocity = 0.0;
 		m_verticalOffset = 0.0;
 		m_wasJumping = false;
-	} else {
+	}
+	else
+	{
 		// we dropped the full height
 
 		full_drop = true;
 
 		if (bounce_fix == true)
 		{
-			downVelocity = (m_verticalVelocity<0.f?-m_verticalVelocity:0.f) * dt;
+			downVelocity = (m_verticalVelocity < 0.f ? -m_verticalVelocity : 0.f) * dt;
 			if (downVelocity > m_fallSpeed && (m_wasOnGround || !m_wasJumping))
 			{
-				m_targetPosition += step_drop; //undo previous target change
+				m_targetPosition += step_drop;  //undo previous target change
 				downVelocity = m_fallSpeed;
 				step_drop = m_up * (m_currentStepOffset + downVelocity);
 				m_targetPosition -= step_drop;
@@ -605,30 +607,22 @@ void btKinematicCharacterController::stepDown ( btCollisionWorld* collisionWorld
 	}
 }
 
-
-
-void btKinematicCharacterController::setWalkDirection
-(
-const btVector3& walkDirection
-)
+void btKinematicCharacterController::setWalkDirection(
+	const btVector3& walkDirection)
 {
 	m_useWalkDirection = true;
 	m_walkDirection = walkDirection;
 	m_normalizedDirection = getNormalizedVector(m_walkDirection);
 }
 
-
-
-void btKinematicCharacterController::setVelocityForTimeInterval
-(
-const btVector3& velocity,
-btScalar timeInterval
-)
+void btKinematicCharacterController::setVelocityForTimeInterval(
+	const btVector3& velocity,
+	btScalar timeInterval)
 {
-//	printf("setVelocity!\n");
-//	printf("  interval: %f\n", timeInterval);
-//	printf("  velocity: (%f, %f, %f)\n",
-//		 velocity.x(), velocity.y(), velocity.z());
+	//	printf("setVelocity!\n");
+	//	printf("  interval: %f\n", timeInterval);
+	//	printf("  velocity: (%f, %f, %f)\n",
+	//		 velocity.x(), velocity.y(), velocity.z());
 
 	m_useWalkDirection = false;
 	m_walkDirection = velocity;
@@ -661,7 +655,7 @@ void btKinematicCharacterController::setLinearVelocity(const btVector3& velocity
 			btVector3 upComponent = m_up * (btSin(SIMD_HALF_PI - btAcos(c)) * m_walkDirection.length());
 			m_walkDirection -= upComponent;
 			m_verticalVelocity = (c < 0.0f ? -1 : 1) * upComponent.length();
-			
+
 			if (c > 0.0f)
 			{
 				m_wasJumping = true;
@@ -678,46 +672,45 @@ btVector3 btKinematicCharacterController::getLinearVelocity() const
 	return m_walkDirection + (m_verticalVelocity * m_up);
 }
 
-void btKinematicCharacterController::reset ( btCollisionWorld* collisionWorld )
+void btKinematicCharacterController::reset(btCollisionWorld* collisionWorld)
 {
-    m_verticalVelocity = 0.0;
-    m_verticalOffset = 0.0;
-    m_wasOnGround = false;
-    m_wasJumping = false;
-    m_walkDirection.setValue(0,0,0);
-    m_velocityTimeInterval = 0.0;
+	m_verticalVelocity = 0.0;
+	m_verticalOffset = 0.0;
+	m_wasOnGround = false;
+	m_wasJumping = false;
+	m_walkDirection.setValue(0, 0, 0);
+	m_velocityTimeInterval = 0.0;
 
-    //clear pair cache
-    btHashedOverlappingPairCache *cache = m_ghostObject->getOverlappingPairCache();
-    while (cache->getOverlappingPairArray().size() > 0)
-    {
-            cache->removeOverlappingPair(cache->getOverlappingPairArray()[0].m_pProxy0, cache->getOverlappingPairArray()[0].m_pProxy1, collisionWorld->getDispatcher());
-    }
+	//clear pair cache
+	btHashedOverlappingPairCache* cache = m_ghostObject->getOverlappingPairCache();
+	while (cache->getOverlappingPairArray().size() > 0)
+	{
+		cache->removeOverlappingPair(cache->getOverlappingPairArray()[0].m_pProxy0, cache->getOverlappingPairArray()[0].m_pProxy1, collisionWorld->getDispatcher());
+	}
 }
 
-void btKinematicCharacterController::warp (const btVector3& origin)
+void btKinematicCharacterController::warp(const btVector3& origin)
 {
 	btTransform xform;
 	xform.setIdentity();
-	xform.setOrigin (origin);
-	m_ghostObject->setWorldTransform (xform);
+	xform.setOrigin(origin);
+	m_ghostObject->setWorldTransform(xform);
 }
 
-
-void btKinematicCharacterController::preStep (  btCollisionWorld* collisionWorld)
+void btKinematicCharacterController::preStep(btCollisionWorld* collisionWorld)
 {
 	m_currentPosition = m_ghostObject->getWorldTransform().getOrigin();
 	m_targetPosition = m_currentPosition;
 
 	m_currentOrientation = m_ghostObject->getWorldTransform().getRotation();
 	m_targetOrientation = m_currentOrientation;
-//	printf("m_targetPosition=%f,%f,%f\n",m_targetPosition[0],m_targetPosition[1],m_targetPosition[2]);
+	//	printf("m_targetPosition=%f,%f,%f\n",m_targetPosition[0],m_targetPosition[1],m_targetPosition[2]);
 }
 
-void btKinematicCharacterController::playerStep (  btCollisionWorld* collisionWorld, btScalar dt)
+void btKinematicCharacterController::playerStep(btCollisionWorld* collisionWorld, btScalar dt)
 {
-//	printf("playerStep(): ");
-//	printf("  dt = %f", dt);
+	//	printf("playerStep(): ");
+	//	printf("  dt = %f", dt);
 
 	if (m_AngVel.length2() > 0.0f)
 	{
@@ -744,16 +737,17 @@ void btKinematicCharacterController::playerStep (  btCollisionWorld* collisionWo
 	}
 
 	// quick check...
-	if (!m_useWalkDirection && (m_velocityTimeInterval <= 0.0)) {
-//		printf("\n");
-		return;		// no motion
+	if (!m_useWalkDirection && (m_velocityTimeInterval <= 0.0 || m_walkDirection.fuzzyZero())) 
+	{
+		//		printf("\n");
+		return;  // no motion
 	}
 
 	m_wasOnGround = onGround();
 
 	//btVector3 lvel = m_walkDirection;
 	//btScalar c = 0.0f;
-	
+
 	if (m_walkDirection.length2() > 0)
 	{
 		// apply damping
@@ -761,7 +755,7 @@ void btKinematicCharacterController::playerStep (  btCollisionWorld* collisionWo
 	}
 
 	m_verticalVelocity *= btPow(btScalar(1) - m_linearDamping, dt);
-	
+
 	// Update fall velocity.
 	m_verticalVelocity -= m_gravity * dt;
 	if (m_verticalVelocity > 0.0 && m_verticalVelocity > m_jumpSpeed)
@@ -777,12 +771,12 @@ void btKinematicCharacterController::playerStep (  btCollisionWorld* collisionWo
 	btTransform xform;
 	xform = m_ghostObject->getWorldTransform();
 
-//	printf("walkDirection(%f,%f,%f)\n",walkDirection[0],walkDirection[1],walkDirection[2]);
-//	printf("walkSpeed=%f\n",walkSpeed);
+	//	printf("walkDirection(%f,%f,%f)\n",walkDirection[0],walkDirection[1],walkDirection[2]);
+	//	printf("walkSpeed=%f\n",walkSpeed);
 
 	stepUp(collisionWorld);
 	//todo: Experimenting with behavior of controller when it hits a ceiling..
-	//bool hitUp = stepUp (collisionWorld);	
+	//bool hitUp = stepUp (collisionWorld);
 	//if (hitUp)
 	//{
 	//	m_verticalVelocity -= m_gravity * dt;
@@ -799,9 +793,12 @@ void btKinematicCharacterController::playerStep (  btCollisionWorld* collisionWo
 	//	xform = m_ghostObject->getWorldTransform();
 	//}
 
-	if (m_useWalkDirection) {
-		stepForwardAndStrafe (collisionWorld, m_walkDirection);
-	} else {
+	if (m_useWalkDirection)
+	{
+		stepForwardAndStrafe(collisionWorld, m_walkDirection);
+	}
+	else
+	{
 		//printf("  time: %f", m_velocityTimeInterval);
 		// still have some time left for moving!
 		btScalar dtMoving =
@@ -816,7 +813,7 @@ void btKinematicCharacterController::playerStep (  btCollisionWorld* collisionWo
 		// okay, step
 		stepForwardAndStrafe(collisionWorld, move);
 	}
-	stepDown (collisionWorld, dt);
+	stepDown(collisionWorld, dt);
 
 	//todo: Experimenting with max jump height
 	//if (m_wasJumping)
@@ -827,7 +824,7 @@ void btKinematicCharacterController::playerStep (  btCollisionWorld* collisionWo
 	//		// substract the overshoot
 	//		m_currentPosition[m_upAxis] -= ds - m_maxJumpHeight;
 
-	//		// max height was reached, so potential energy is at max 
+	//		// max height was reached, so potential energy is at max
 	//		// and kinematic energy is 0, thus velocity is 0.
 	//		if (m_verticalVelocity > 0.0)
 	//			m_verticalVelocity = 0.0;
@@ -835,8 +832,8 @@ void btKinematicCharacterController::playerStep (  btCollisionWorld* collisionWo
 	//}
 	// printf("\n");
 
-	xform.setOrigin (m_currentPosition);
-	m_ghostObject->setWorldTransform (xform);
+	xform.setOrigin(m_currentPosition);
+	m_ghostObject->setWorldTransform(xform);
 
 	int numPenetrationLoops = 0;
 	m_touchingContact = false;
@@ -852,23 +849,23 @@ void btKinematicCharacterController::playerStep (  btCollisionWorld* collisionWo
 	}
 }
 
-void btKinematicCharacterController::setFallSpeed (btScalar fallSpeed)
+void btKinematicCharacterController::setFallSpeed(btScalar fallSpeed)
 {
 	m_fallSpeed = fallSpeed;
 }
 
-void btKinematicCharacterController::setJumpSpeed (btScalar jumpSpeed)
+void btKinematicCharacterController::setJumpSpeed(btScalar jumpSpeed)
 {
 	m_jumpSpeed = jumpSpeed;
 	m_SetjumpSpeed = m_jumpSpeed;
 }
 
-void btKinematicCharacterController::setMaxJumpHeight (btScalar maxJumpHeight)
+void btKinematicCharacterController::setMaxJumpHeight(btScalar maxJumpHeight)
 {
 	m_maxJumpHeight = maxJumpHeight;
 }
 
-bool btKinematicCharacterController::canJump () const
+bool btKinematicCharacterController::canJump() const
 {
 	return onGround();
 }
@@ -927,20 +924,20 @@ btScalar btKinematicCharacterController::getMaxPenetrationDepth() const
 	return m_maxPenetrationDepth;
 }
 
-bool btKinematicCharacterController::onGround () const
+bool btKinematicCharacterController::onGround() const
 {
 	return (fabs(m_verticalVelocity) < SIMD_EPSILON) && (fabs(m_verticalOffset) < SIMD_EPSILON);
 }
 
-void btKinematicCharacterController::setStepHeight(btScalar h) 
+void btKinematicCharacterController::setStepHeight(btScalar h)
 {
 	m_stepHeight = h;
 }
 
 btVector3* btKinematicCharacterController::getUpAxisDirections()
 {
-	static btVector3 sUpAxisDirection[3] = { btVector3(1.0f, 0.0f, 0.0f), btVector3(0.0f, 1.0f, 0.0f), btVector3(0.0f, 0.0f, 1.0f) };
-	
+	static btVector3 sUpAxisDirection[3] = {btVector3(1.0f, 0.0f, 0.0f), btVector3(0.0f, 1.0f, 0.0f), btVector3(0.0f, 0.0f, 1.0f)};
+
 	return sUpAxisDirection;
 }
 
@@ -997,4 +994,3 @@ btQuaternion btKinematicCharacterController::getRotation(btVector3& v0, btVector
 
 	return shortestArcQuatNormalize2(v0, v1);
 }
-
diff --git a/thirdparty/bullet/BulletDynamics/Character/btKinematicCharacterController.h b/thirdparty/bullet/BulletDynamics/Character/btKinematicCharacterController.h
index 00c59c02480..ff34fc871a5 100644
--- a/thirdparty/bullet/BulletDynamics/Character/btKinematicCharacterController.h
+++ b/thirdparty/bullet/BulletDynamics/Character/btKinematicCharacterController.h
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef BT_KINEMATIC_CHARACTER_CONTROLLER_H
 #define BT_KINEMATIC_CHARACTER_CONTROLLER_H
 
@@ -23,7 +22,6 @@ subject to the following restrictions:
 
 #include "BulletCollision/BroadphaseCollision/btCollisionAlgorithm.h"
 
-
 class btCollisionShape;
 class btConvexShape;
 class btRigidBody;
@@ -34,15 +32,15 @@ class btPairCachingGhostObject;
 ///btKinematicCharacterController is an object that supports a sliding motion in a world.
 ///It uses a ghost object and convex sweep test to test for upcoming collisions. This is combined with discrete collision detection to recover from penetrations.
 ///Interaction between btKinematicCharacterController and dynamic rigid bodies needs to be explicity implemented by the user.
-ATTRIBUTE_ALIGNED16(class) btKinematicCharacterController : public btCharacterControllerInterface
+ATTRIBUTE_ALIGNED16(class)
+btKinematicCharacterController : public btCharacterControllerInterface
 {
 protected:
-
 	btScalar m_halfHeight;
-	
+
 	btPairCachingGhostObject* m_ghostObject;
-	btConvexShape*	m_convexShape;//is also in m_ghostObject, but it needs to be convex, so we store it here to avoid upcast
-	
+	btConvexShape* m_convexShape;  //is also in m_ghostObject, but it needs to be convex, so we store it here to avoid upcast
+
 	btScalar m_maxPenetrationDepth;
 	btScalar m_verticalVelocity;
 	btScalar m_verticalOffset;
@@ -50,33 +48,33 @@ protected:
 	btScalar m_jumpSpeed;
 	btScalar m_SetjumpSpeed;
 	btScalar m_maxJumpHeight;
-	btScalar m_maxSlopeRadians; // Slope angle that is set (used for returning the exact value)
-	btScalar m_maxSlopeCosine;  // Cosine equivalent of m_maxSlopeRadians (calculated once when set, for optimization)
+	btScalar m_maxSlopeRadians;  // Slope angle that is set (used for returning the exact value)
+	btScalar m_maxSlopeCosine;   // Cosine equivalent of m_maxSlopeRadians (calculated once when set, for optimization)
 	btScalar m_gravity;
 
 	btScalar m_turnAngle;
-	
+
 	btScalar m_stepHeight;
 
-	btScalar	m_addedMargin;//@todo: remove this and fix the code
+	btScalar m_addedMargin;  //@todo: remove this and fix the code
 
 	///this is the desired walk direction, set by the user
-	btVector3	m_walkDirection;
-	btVector3	m_normalizedDirection;
-	btVector3	m_AngVel;
+	btVector3 m_walkDirection;
+	btVector3 m_normalizedDirection;
+	btVector3 m_AngVel;
 
-	btVector3	m_jumpPosition;
+	btVector3 m_jumpPosition;
 
 	//some internal variables
 	btVector3 m_currentPosition;
-	btScalar  m_currentStepOffset;
+	btScalar m_currentStepOffset;
 	btVector3 m_targetPosition;
 
 	btQuaternion m_currentOrientation;
 	btQuaternion m_targetOrientation;
 
 	///keep track of the contact manifolds
-	btManifoldArray	m_manifoldArray;
+	btManifoldArray m_manifoldArray;
 
 	bool m_touchingContact;
 	btVector3 m_touchingNormal;
@@ -84,52 +82,50 @@ protected:
 	btScalar m_linearDamping;
 	btScalar m_angularDamping;
 
-	bool  m_wasOnGround;
-	bool  m_wasJumping;
-	bool	m_useGhostObjectSweepTest;
-	bool	m_useWalkDirection;
-	btScalar	m_velocityTimeInterval;
+	bool m_wasOnGround;
+	bool m_wasJumping;
+	bool m_useGhostObjectSweepTest;
+	bool m_useWalkDirection;
+	btScalar m_velocityTimeInterval;
 	btVector3 m_up;
 	btVector3 m_jumpAxis;
 
 	static btVector3* getUpAxisDirections();
-	bool  m_interpolateUp;
-	bool  full_drop;
-	bool  bounce_fix;
+	bool m_interpolateUp;
+	bool full_drop;
+	bool bounce_fix;
 
-	btVector3 computeReflectionDirection (const btVector3& direction, const btVector3& normal);
-	btVector3 parallelComponent (const btVector3& direction, const btVector3& normal);
-	btVector3 perpindicularComponent (const btVector3& direction, const btVector3& normal);
+	btVector3 computeReflectionDirection(const btVector3& direction, const btVector3& normal);
+	btVector3 parallelComponent(const btVector3& direction, const btVector3& normal);
+	btVector3 perpindicularComponent(const btVector3& direction, const btVector3& normal);
 
-	bool recoverFromPenetration ( btCollisionWorld* collisionWorld);
-	void stepUp (btCollisionWorld* collisionWorld);
-	void updateTargetPositionBasedOnCollision (const btVector3& hit_normal, btScalar tangentMag = btScalar(0.0), btScalar normalMag = btScalar(1.0));
-	void stepForwardAndStrafe (btCollisionWorld* collisionWorld, const btVector3& walkMove);
-	void stepDown (btCollisionWorld* collisionWorld, btScalar dt);
+	bool recoverFromPenetration(btCollisionWorld * collisionWorld);
+	void stepUp(btCollisionWorld * collisionWorld);
+	void updateTargetPositionBasedOnCollision(const btVector3& hit_normal, btScalar tangentMag = btScalar(0.0), btScalar normalMag = btScalar(1.0));
+	void stepForwardAndStrafe(btCollisionWorld * collisionWorld, const btVector3& walkMove);
+	void stepDown(btCollisionWorld * collisionWorld, btScalar dt);
 
 	virtual bool needsCollision(const btCollisionObject* body0, const btCollisionObject* body1);
 
 	void setUpVector(const btVector3& up);
 
-	btQuaternion getRotation(btVector3& v0, btVector3& v1) const;
+	btQuaternion getRotation(btVector3 & v0, btVector3 & v1) const;
 
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-	btKinematicCharacterController (btPairCachingGhostObject* ghostObject,btConvexShape* convexShape,btScalar stepHeight, const btVector3& up = btVector3(1.0,0.0,0.0));
-	~btKinematicCharacterController ();
-	
+	btKinematicCharacterController(btPairCachingGhostObject * ghostObject, btConvexShape * convexShape, btScalar stepHeight, const btVector3& up = btVector3(1.0, 0.0, 0.0));
+	~btKinematicCharacterController();
 
 	///btActionInterface interface
-	virtual void updateAction( btCollisionWorld* collisionWorld,btScalar deltaTime)
+	virtual void updateAction(btCollisionWorld * collisionWorld, btScalar deltaTime)
 	{
-		preStep ( collisionWorld);
-		playerStep (collisionWorld, deltaTime);
+		preStep(collisionWorld);
+		playerStep(collisionWorld, deltaTime);
 	}
-	
+
 	///btActionInterface interface
-	void	debugDraw(btIDebugDraw* debugDrawer);
+	void debugDraw(btIDebugDraw * debugDrawer);
 
 	void setUp(const btVector3& up);
 
@@ -140,7 +136,7 @@ public:
 	///	increment the position each simulation iteration, regardless
 	///	of dt.
 	/// This call will reset any velocity set by setVelocityForTimeInterval().
-	virtual void	setWalkDirection(const btVector3& walkDirection);
+	virtual void setWalkDirection(const btVector3& walkDirection);
 
 	/// Caller provides a velocity with which the character should move for
 	///	the given time period.  After the time period, velocity is reset
@@ -148,7 +144,7 @@ public:
 	/// This call will reset any walk direction set by setWalkDirection().
 	/// Negative time intervals will result in no motion.
 	virtual void setVelocityForTimeInterval(const btVector3& velocity,
-				btScalar timeInterval);
+											btScalar timeInterval);
 
 	virtual void setAngularVelocity(const btVector3& velocity);
 	virtual const btVector3& getAngularVelocity() const;
@@ -157,24 +153,24 @@ public:
 	virtual btVector3 getLinearVelocity() const;
 
 	void setLinearDamping(btScalar d) { m_linearDamping = btClamped(d, (btScalar)btScalar(0.0), (btScalar)btScalar(1.0)); }
-	btScalar getLinearDamping() const { return  m_linearDamping; }
+	btScalar getLinearDamping() const { return m_linearDamping; }
 	void setAngularDamping(btScalar d) { m_angularDamping = btClamped(d, (btScalar)btScalar(0.0), (btScalar)btScalar(1.0)); }
-	btScalar getAngularDamping() const { return  m_angularDamping; }
+	btScalar getAngularDamping() const { return m_angularDamping; }
 
-	void reset ( btCollisionWorld* collisionWorld );
-	void warp (const btVector3& origin);
+	void reset(btCollisionWorld * collisionWorld);
+	void warp(const btVector3& origin);
 
-	void preStep (  btCollisionWorld* collisionWorld);
-	void playerStep ( btCollisionWorld* collisionWorld, btScalar dt);
+	void preStep(btCollisionWorld * collisionWorld);
+	void playerStep(btCollisionWorld * collisionWorld, btScalar dt);
 
 	void setStepHeight(btScalar h);
 	btScalar getStepHeight() const { return m_stepHeight; }
-	void setFallSpeed (btScalar fallSpeed);
+	void setFallSpeed(btScalar fallSpeed);
 	btScalar getFallSpeed() const { return m_fallSpeed; }
-	void setJumpSpeed (btScalar jumpSpeed);
+	void setJumpSpeed(btScalar jumpSpeed);
 	btScalar getJumpSpeed() const { return m_jumpSpeed; }
-	void setMaxJumpHeight (btScalar maxJumpHeight);
-	bool canJump () const;
+	void setMaxJumpHeight(btScalar maxJumpHeight);
+	bool canJump() const;
 
 	void jump(const btVector3& v = btVector3(0, 0, 0));
 
@@ -192,13 +188,13 @@ public:
 	btScalar getMaxPenetrationDepth() const;
 
 	btPairCachingGhostObject* getGhostObject();
-	void	setUseGhostSweepTest(bool useGhostObjectSweepTest)
+	void setUseGhostSweepTest(bool useGhostObjectSweepTest)
 	{
 		m_useGhostObjectSweepTest = useGhostObjectSweepTest;
 	}
 
-	bool onGround () const;
-	void setUpInterpolate (bool value);
+	bool onGround() const;
+	void setUpInterpolate(bool value);
 };
 
-#endif // BT_KINEMATIC_CHARACTER_CONTROLLER_H
+#endif  // BT_KINEMATIC_CHARACTER_CONTROLLER_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btBatchedConstraints.cpp b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btBatchedConstraints.cpp
index c82ba87f9ff..b51dfaad3cb 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btBatchedConstraints.cpp
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btBatchedConstraints.cpp
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btBatchedConstraints.h"
 
 #include "LinearMath/btIDebugDraw.h"
@@ -21,606 +20,573 @@ subject to the following restrictions:
 #include "LinearMath/btStackAlloc.h"
 #include "LinearMath/btQuickprof.h"
 
-#include <string.h> //for memset
+#include <string.h>  //for memset
 
 const int kNoMerge = -1;
 
 bool btBatchedConstraints::s_debugDrawBatches = false;
 
-
 struct btBatchedConstraintInfo
 {
-    int constraintIndex;
-    int numConstraintRows;
-    int bodyIds[2];
+	int constraintIndex;
+	int numConstraintRows;
+	int bodyIds[2];
 };
 
-
 struct btBatchInfo
 {
-    int numConstraints;
-    int mergeIndex;
+	int numConstraints;
+	int mergeIndex;
 
-    btBatchInfo() : numConstraints(0), mergeIndex(kNoMerge) {}
+	btBatchInfo() : numConstraints(0), mergeIndex(kNoMerge) {}
 };
 
-
 bool btBatchedConstraints::validate(btConstraintArray* constraints, const btAlignedObjectArray<btSolverBody>& bodies) const
 {
-    //
-    // validate: for debugging only. Verify coloring of bodies, that no body is touched by more than one batch in any given phase
-    //
-    int errors = 0;
-    const int kUnassignedBatch = -1;
+	//
+	// validate: for debugging only. Verify coloring of bodies, that no body is touched by more than one batch in any given phase
+	//
+	int errors = 0;
+	const int kUnassignedBatch = -1;
 
-    btAlignedObjectArray<int> bodyBatchId;
-    for (int iPhase = 0; iPhase < m_phases.size(); ++iPhase)
-    {
-        bodyBatchId.resizeNoInitialize(0);
-        bodyBatchId.resize( bodies.size(), kUnassignedBatch );
-        const Range& phase = m_phases[iPhase];
-        for (int iBatch = phase.begin; iBatch < phase.end; ++iBatch)
-        {
-            const Range& batch = m_batches[iBatch];
-            for (int iiCons = batch.begin; iiCons < batch.end; ++iiCons)
-            {
-                int iCons = m_constraintIndices[iiCons];
-                const btSolverConstraint& cons = constraints->at(iCons);
-                const btSolverBody& bodyA = bodies[cons.m_solverBodyIdA];
-                const btSolverBody& bodyB = bodies[cons.m_solverBodyIdB];
-                if (! bodyA.internalGetInvMass().isZero())
-                {
-                    int thisBodyBatchId = bodyBatchId[cons.m_solverBodyIdA];
-                    if (thisBodyBatchId == kUnassignedBatch)
-                    {
-                        bodyBatchId[cons.m_solverBodyIdA] = iBatch;
-                    }
-                    else if (thisBodyBatchId != iBatch)
-                    {
-                        btAssert( !"dynamic body is used in 2 different batches in the same phase" );
-                        errors++;
-                    }
-                }
-                if (! bodyB.internalGetInvMass().isZero())
-                {
-                    int thisBodyBatchId = bodyBatchId[cons.m_solverBodyIdB];
-                    if (thisBodyBatchId == kUnassignedBatch)
-                    {
-                        bodyBatchId[cons.m_solverBodyIdB] = iBatch;
-                    }
-                    else if (thisBodyBatchId != iBatch)
-                    {
-                        btAssert( !"dynamic body is used in 2 different batches in the same phase" );
-                        errors++;
-                    }
-                }
-            }
-        }
-    }
-    return errors == 0;
+	btAlignedObjectArray<int> bodyBatchId;
+	for (int iPhase = 0; iPhase < m_phases.size(); ++iPhase)
+	{
+		bodyBatchId.resizeNoInitialize(0);
+		bodyBatchId.resize(bodies.size(), kUnassignedBatch);
+		const Range& phase = m_phases[iPhase];
+		for (int iBatch = phase.begin; iBatch < phase.end; ++iBatch)
+		{
+			const Range& batch = m_batches[iBatch];
+			for (int iiCons = batch.begin; iiCons < batch.end; ++iiCons)
+			{
+				int iCons = m_constraintIndices[iiCons];
+				const btSolverConstraint& cons = constraints->at(iCons);
+				const btSolverBody& bodyA = bodies[cons.m_solverBodyIdA];
+				const btSolverBody& bodyB = bodies[cons.m_solverBodyIdB];
+				if (!bodyA.internalGetInvMass().isZero())
+				{
+					int thisBodyBatchId = bodyBatchId[cons.m_solverBodyIdA];
+					if (thisBodyBatchId == kUnassignedBatch)
+					{
+						bodyBatchId[cons.m_solverBodyIdA] = iBatch;
+					}
+					else if (thisBodyBatchId != iBatch)
+					{
+						btAssert(!"dynamic body is used in 2 different batches in the same phase");
+						errors++;
+					}
+				}
+				if (!bodyB.internalGetInvMass().isZero())
+				{
+					int thisBodyBatchId = bodyBatchId[cons.m_solverBodyIdB];
+					if (thisBodyBatchId == kUnassignedBatch)
+					{
+						bodyBatchId[cons.m_solverBodyIdB] = iBatch;
+					}
+					else if (thisBodyBatchId != iBatch)
+					{
+						btAssert(!"dynamic body is used in 2 different batches in the same phase");
+						errors++;
+					}
+				}
+			}
+		}
+	}
+	return errors == 0;
 }
 
-
-static void debugDrawSingleBatch( const btBatchedConstraints* bc,
-    btConstraintArray* constraints,
-    const btAlignedObjectArray<btSolverBody>& bodies,
-    int iBatch,
-    const btVector3& color,
-    const btVector3& offset
-    )
+static void debugDrawSingleBatch(const btBatchedConstraints* bc,
+								 btConstraintArray* constraints,
+								 const btAlignedObjectArray<btSolverBody>& bodies,
+								 int iBatch,
+								 const btVector3& color,
+								 const btVector3& offset)
 {
-    if (bc && bc->m_debugDrawer && iBatch < bc->m_batches.size())
-    {
-        const btBatchedConstraints::Range& b = bc->m_batches[iBatch];
-        for (int iiCon = b.begin; iiCon < b.end; ++iiCon)
-        {
-            int iCon = bc->m_constraintIndices[iiCon];
-            const btSolverConstraint& con = constraints->at(iCon);
-            int iBody0 = con.m_solverBodyIdA;
-            int iBody1 = con.m_solverBodyIdB;
-            btVector3 pos0 = bodies[iBody0].getWorldTransform().getOrigin() + offset;
-            btVector3 pos1 = bodies[iBody1].getWorldTransform().getOrigin() + offset;
-            bc->m_debugDrawer->drawLine(pos0, pos1, color);
-        }
-    }
+	if (bc && bc->m_debugDrawer && iBatch < bc->m_batches.size())
+	{
+		const btBatchedConstraints::Range& b = bc->m_batches[iBatch];
+		for (int iiCon = b.begin; iiCon < b.end; ++iiCon)
+		{
+			int iCon = bc->m_constraintIndices[iiCon];
+			const btSolverConstraint& con = constraints->at(iCon);
+			int iBody0 = con.m_solverBodyIdA;
+			int iBody1 = con.m_solverBodyIdB;
+			btVector3 pos0 = bodies[iBody0].getWorldTransform().getOrigin() + offset;
+			btVector3 pos1 = bodies[iBody1].getWorldTransform().getOrigin() + offset;
+			bc->m_debugDrawer->drawLine(pos0, pos1, color);
+		}
+	}
 }
 
-
-static void debugDrawPhase( const btBatchedConstraints* bc,
-    btConstraintArray* constraints,
-    const btAlignedObjectArray<btSolverBody>& bodies,
-    int iPhase,
-    const btVector3& color0,
-    const btVector3& color1,
-    const btVector3& offset
-    )
+static void debugDrawPhase(const btBatchedConstraints* bc,
+						   btConstraintArray* constraints,
+						   const btAlignedObjectArray<btSolverBody>& bodies,
+						   int iPhase,
+						   const btVector3& color0,
+						   const btVector3& color1,
+						   const btVector3& offset)
 {
-    BT_PROFILE( "debugDrawPhase" );
-    if ( bc && bc->m_debugDrawer && iPhase < bc->m_phases.size() )
-    {
-        const btBatchedConstraints::Range& phase = bc->m_phases[iPhase];
-        for (int iBatch = phase.begin; iBatch < phase.end; ++iBatch)
-        {
-            float tt = float(iBatch - phase.begin) / float(btMax(1, phase.end - phase.begin - 1));
-            btVector3 col = lerp(color0, color1, tt);
-            debugDrawSingleBatch(bc, constraints, bodies, iBatch, col, offset);
-        }
-    }
+	BT_PROFILE("debugDrawPhase");
+	if (bc && bc->m_debugDrawer && iPhase < bc->m_phases.size())
+	{
+		const btBatchedConstraints::Range& phase = bc->m_phases[iPhase];
+		for (int iBatch = phase.begin; iBatch < phase.end; ++iBatch)
+		{
+			float tt = float(iBatch - phase.begin) / float(btMax(1, phase.end - phase.begin - 1));
+			btVector3 col = lerp(color0, color1, tt);
+			debugDrawSingleBatch(bc, constraints, bodies, iBatch, col, offset);
+		}
+	}
 }
 
-
-static void debugDrawAllBatches( const btBatchedConstraints* bc,
-    btConstraintArray* constraints,
-    const btAlignedObjectArray<btSolverBody>& bodies
-    )
+static void debugDrawAllBatches(const btBatchedConstraints* bc,
+								btConstraintArray* constraints,
+								const btAlignedObjectArray<btSolverBody>& bodies)
 {
-    BT_PROFILE( "debugDrawAllBatches" );
-    if ( bc && bc->m_debugDrawer && bc->m_phases.size() > 0 )
-    {
-        btVector3 bboxMin(BT_LARGE_FLOAT, BT_LARGE_FLOAT, BT_LARGE_FLOAT);
-        btVector3 bboxMax = -bboxMin;
-        for (int iBody = 0; iBody < bodies.size(); ++iBody)
-        {
-            const btVector3& pos = bodies[iBody].getWorldTransform().getOrigin();
-            bboxMin.setMin(pos);
-            bboxMax.setMax(pos);
-        }
-        btVector3 bboxExtent = bboxMax - bboxMin;
-        btVector3 offsetBase = btVector3( 0, bboxExtent.y()*1.1f, 0 );
-        btVector3 offsetStep = btVector3( 0, 0, bboxExtent.z()*1.1f );
-        int numPhases = bc->m_phases.size();
-        for (int iPhase = 0; iPhase < numPhases; ++iPhase)
-        {
-            float b = float(iPhase)/float(numPhases-1);
-            btVector3 color0 = btVector3(1,0,b);
-            btVector3 color1 = btVector3(0,1,b);
-            btVector3 offset = offsetBase + offsetStep*(float(iPhase) - float(numPhases-1)*0.5);
-            debugDrawPhase(bc, constraints, bodies, iPhase, color0, color1, offset);
-        }
-    }
+	BT_PROFILE("debugDrawAllBatches");
+	if (bc && bc->m_debugDrawer && bc->m_phases.size() > 0)
+	{
+		btVector3 bboxMin(BT_LARGE_FLOAT, BT_LARGE_FLOAT, BT_LARGE_FLOAT);
+		btVector3 bboxMax = -bboxMin;
+		for (int iBody = 0; iBody < bodies.size(); ++iBody)
+		{
+			const btVector3& pos = bodies[iBody].getWorldTransform().getOrigin();
+			bboxMin.setMin(pos);
+			bboxMax.setMax(pos);
+		}
+		btVector3 bboxExtent = bboxMax - bboxMin;
+		btVector3 offsetBase = btVector3(0, bboxExtent.y() * 1.1f, 0);
+		btVector3 offsetStep = btVector3(0, 0, bboxExtent.z() * 1.1f);
+		int numPhases = bc->m_phases.size();
+		for (int iPhase = 0; iPhase < numPhases; ++iPhase)
+		{
+			float b = float(iPhase) / float(numPhases - 1);
+			btVector3 color0 = btVector3(1, 0, b);
+			btVector3 color1 = btVector3(0, 1, b);
+			btVector3 offset = offsetBase + offsetStep * (float(iPhase) - float(numPhases - 1) * 0.5);
+			debugDrawPhase(bc, constraints, bodies, iPhase, color0, color1, offset);
+		}
+	}
 }
 
-
 static void initBatchedBodyDynamicFlags(btAlignedObjectArray<bool>* outBodyDynamicFlags, const btAlignedObjectArray<btSolverBody>& bodies)
 {
-    BT_PROFILE("initBatchedBodyDynamicFlags");
-    btAlignedObjectArray<bool>& bodyDynamicFlags = *outBodyDynamicFlags;
-    bodyDynamicFlags.resizeNoInitialize(bodies.size());
-    for (int i = 0; i < bodies.size(); ++i)
-    {
-        const btSolverBody& body = bodies[ i ];
-        bodyDynamicFlags[i] = ( body.internalGetInvMass().x() > btScalar( 0 ) );
-    }
+	BT_PROFILE("initBatchedBodyDynamicFlags");
+	btAlignedObjectArray<bool>& bodyDynamicFlags = *outBodyDynamicFlags;
+	bodyDynamicFlags.resizeNoInitialize(bodies.size());
+	for (int i = 0; i < bodies.size(); ++i)
+	{
+		const btSolverBody& body = bodies[i];
+		bodyDynamicFlags[i] = (body.internalGetInvMass().x() > btScalar(0));
+	}
 }
 
-
 static int runLengthEncodeConstraintInfo(btBatchedConstraintInfo* outConInfos, int numConstraints)
 {
-    BT_PROFILE("runLengthEncodeConstraintInfo");
-    // detect and run-length encode constraint rows that repeat the same bodies
-    int iDest = 0;
-    int iSrc = 0;
-    while (iSrc < numConstraints)
-    {
-        const btBatchedConstraintInfo& srcConInfo = outConInfos[iSrc];
-        btBatchedConstraintInfo& conInfo = outConInfos[iDest];
-        conInfo.constraintIndex = iSrc;
-        conInfo.bodyIds[0] = srcConInfo.bodyIds[0];
-        conInfo.bodyIds[1] = srcConInfo.bodyIds[1];
-        while (iSrc < numConstraints && outConInfos[iSrc].bodyIds[0] == srcConInfo.bodyIds[0] && outConInfos[iSrc].bodyIds[1] == srcConInfo.bodyIds[1])
-        {
-            ++iSrc;
-        }
-        conInfo.numConstraintRows = iSrc - conInfo.constraintIndex;
-        ++iDest;
-    }
-    return iDest;
+	BT_PROFILE("runLengthEncodeConstraintInfo");
+	// detect and run-length encode constraint rows that repeat the same bodies
+	int iDest = 0;
+	int iSrc = 0;
+	while (iSrc < numConstraints)
+	{
+		const btBatchedConstraintInfo& srcConInfo = outConInfos[iSrc];
+		btBatchedConstraintInfo& conInfo = outConInfos[iDest];
+		conInfo.constraintIndex = iSrc;
+		conInfo.bodyIds[0] = srcConInfo.bodyIds[0];
+		conInfo.bodyIds[1] = srcConInfo.bodyIds[1];
+		while (iSrc < numConstraints && outConInfos[iSrc].bodyIds[0] == srcConInfo.bodyIds[0] && outConInfos[iSrc].bodyIds[1] == srcConInfo.bodyIds[1])
+		{
+			++iSrc;
+		}
+		conInfo.numConstraintRows = iSrc - conInfo.constraintIndex;
+		++iDest;
+	}
+	return iDest;
 }
 
-
 struct ReadSolverConstraintsLoop : public btIParallelForBody
 {
-    btBatchedConstraintInfo* m_outConInfos;
-    btConstraintArray* m_constraints;
+	btBatchedConstraintInfo* m_outConInfos;
+	btConstraintArray* m_constraints;
 
-    ReadSolverConstraintsLoop( btBatchedConstraintInfo* outConInfos, btConstraintArray* constraints )
-    {
-        m_outConInfos = outConInfos;
-        m_constraints = constraints;
-    }
-    void forLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        for (int i = iBegin; i < iEnd; ++i)
-        {
-            btBatchedConstraintInfo& conInfo = m_outConInfos[i];
-            const btSolverConstraint& con = m_constraints->at( i );
-            conInfo.bodyIds[0] = con.m_solverBodyIdA;
-            conInfo.bodyIds[1] = con.m_solverBodyIdB;
-            conInfo.constraintIndex = i;
-            conInfo.numConstraintRows = 1;
-        }
-    }
+	ReadSolverConstraintsLoop(btBatchedConstraintInfo* outConInfos, btConstraintArray* constraints)
+	{
+		m_outConInfos = outConInfos;
+		m_constraints = constraints;
+	}
+	void forLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		for (int i = iBegin; i < iEnd; ++i)
+		{
+			btBatchedConstraintInfo& conInfo = m_outConInfos[i];
+			const btSolverConstraint& con = m_constraints->at(i);
+			conInfo.bodyIds[0] = con.m_solverBodyIdA;
+			conInfo.bodyIds[1] = con.m_solverBodyIdB;
+			conInfo.constraintIndex = i;
+			conInfo.numConstraintRows = 1;
+		}
+	}
 };
 
-
 static int initBatchedConstraintInfo(btBatchedConstraintInfo* outConInfos, btConstraintArray* constraints)
 {
-    BT_PROFILE("initBatchedConstraintInfo");
-    int numConstraints = constraints->size();
-    bool inParallel = true;
-    if (inParallel)
-    {
-        ReadSolverConstraintsLoop loop(outConInfos, constraints);
-        int grainSize = 1200;
-        btParallelFor(0, numConstraints, grainSize, loop);
-    }
-    else
-    {
-        for (int i = 0; i < numConstraints; ++i)
-        {
-            btBatchedConstraintInfo& conInfo = outConInfos[i];
-            const btSolverConstraint& con = constraints->at( i );
-            conInfo.bodyIds[0] = con.m_solverBodyIdA;
-            conInfo.bodyIds[1] = con.m_solverBodyIdB;
-            conInfo.constraintIndex = i;
-            conInfo.numConstraintRows = 1;
-        }
-    }
-    bool useRunLengthEncoding = true;
-    if (useRunLengthEncoding)
-    {
-        numConstraints = runLengthEncodeConstraintInfo(outConInfos, numConstraints);
-    }
-    return numConstraints;
+	BT_PROFILE("initBatchedConstraintInfo");
+	int numConstraints = constraints->size();
+	bool inParallel = true;
+	if (inParallel)
+	{
+		ReadSolverConstraintsLoop loop(outConInfos, constraints);
+		int grainSize = 1200;
+		btParallelFor(0, numConstraints, grainSize, loop);
+	}
+	else
+	{
+		for (int i = 0; i < numConstraints; ++i)
+		{
+			btBatchedConstraintInfo& conInfo = outConInfos[i];
+			const btSolverConstraint& con = constraints->at(i);
+			conInfo.bodyIds[0] = con.m_solverBodyIdA;
+			conInfo.bodyIds[1] = con.m_solverBodyIdB;
+			conInfo.constraintIndex = i;
+			conInfo.numConstraintRows = 1;
+		}
+	}
+	bool useRunLengthEncoding = true;
+	if (useRunLengthEncoding)
+	{
+		numConstraints = runLengthEncodeConstraintInfo(outConInfos, numConstraints);
+	}
+	return numConstraints;
 }
 
-
 static void expandConstraintRowsInPlace(int* constraintBatchIds, const btBatchedConstraintInfo* conInfos, int numConstraints, int numConstraintRows)
 {
-    BT_PROFILE("expandConstraintRowsInPlace");
-    if (numConstraintRows > numConstraints)
-    {
-        // we walk the array in reverse to avoid overwriteing
-        for (int iCon = numConstraints - 1; iCon >= 0; --iCon)
-        {
-            const btBatchedConstraintInfo& conInfo = conInfos[iCon];
-            int iBatch = constraintBatchIds[iCon];
-            for (int i = conInfo.numConstraintRows - 1; i >= 0; --i)
-            {
-                int iDest = conInfo.constraintIndex + i;
-                btAssert(iDest >= iCon);
-                btAssert(iDest >= 0 && iDest < numConstraintRows);
-                constraintBatchIds[iDest] = iBatch;
-            }
-        }
-    }
+	BT_PROFILE("expandConstraintRowsInPlace");
+	if (numConstraintRows > numConstraints)
+	{
+		// we walk the array in reverse to avoid overwriteing
+		for (int iCon = numConstraints - 1; iCon >= 0; --iCon)
+		{
+			const btBatchedConstraintInfo& conInfo = conInfos[iCon];
+			int iBatch = constraintBatchIds[iCon];
+			for (int i = conInfo.numConstraintRows - 1; i >= 0; --i)
+			{
+				int iDest = conInfo.constraintIndex + i;
+				btAssert(iDest >= iCon);
+				btAssert(iDest >= 0 && iDest < numConstraintRows);
+				constraintBatchIds[iDest] = iBatch;
+			}
+		}
+	}
 }
 
-
 static void expandConstraintRows(int* destConstraintBatchIds, const int* srcConstraintBatchIds, const btBatchedConstraintInfo* conInfos, int numConstraints, int numConstraintRows)
 {
-    BT_PROFILE("expandConstraintRows");
-    for ( int iCon = 0; iCon < numConstraints; ++iCon )
-    {
-        const btBatchedConstraintInfo& conInfo = conInfos[ iCon ];
-        int iBatch = srcConstraintBatchIds[ iCon ];
-        for ( int i = 0; i < conInfo.numConstraintRows; ++i )
-        {
-            int iDest = conInfo.constraintIndex + i;
-            btAssert( iDest >= iCon );
-            btAssert( iDest >= 0 && iDest < numConstraintRows );
-            destConstraintBatchIds[ iDest ] = iBatch;
-        }
-    }
+	BT_PROFILE("expandConstraintRows");
+	for (int iCon = 0; iCon < numConstraints; ++iCon)
+	{
+		const btBatchedConstraintInfo& conInfo = conInfos[iCon];
+		int iBatch = srcConstraintBatchIds[iCon];
+		for (int i = 0; i < conInfo.numConstraintRows; ++i)
+		{
+			int iDest = conInfo.constraintIndex + i;
+			btAssert(iDest >= iCon);
+			btAssert(iDest >= 0 && iDest < numConstraintRows);
+			destConstraintBatchIds[iDest] = iBatch;
+		}
+	}
 }
 
-
 struct ExpandConstraintRowsLoop : public btIParallelForBody
 {
-    int* m_destConstraintBatchIds;
-    const int* m_srcConstraintBatchIds;
-    const btBatchedConstraintInfo* m_conInfos;
-    int m_numConstraintRows;
+	int* m_destConstraintBatchIds;
+	const int* m_srcConstraintBatchIds;
+	const btBatchedConstraintInfo* m_conInfos;
+	int m_numConstraintRows;
 
-    ExpandConstraintRowsLoop( int* destConstraintBatchIds, const int* srcConstraintBatchIds, const btBatchedConstraintInfo* conInfos, int numConstraintRows)
-    {
-        m_destConstraintBatchIds = destConstraintBatchIds;
-        m_srcConstraintBatchIds = srcConstraintBatchIds;
-        m_conInfos = conInfos;
-        m_numConstraintRows = numConstraintRows;
-    }
-    void forLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        expandConstraintRows(m_destConstraintBatchIds, m_srcConstraintBatchIds + iBegin, m_conInfos + iBegin, iEnd - iBegin, m_numConstraintRows);
-    }
+	ExpandConstraintRowsLoop(int* destConstraintBatchIds, const int* srcConstraintBatchIds, const btBatchedConstraintInfo* conInfos, int numConstraintRows)
+	{
+		m_destConstraintBatchIds = destConstraintBatchIds;
+		m_srcConstraintBatchIds = srcConstraintBatchIds;
+		m_conInfos = conInfos;
+		m_numConstraintRows = numConstraintRows;
+	}
+	void forLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		expandConstraintRows(m_destConstraintBatchIds, m_srcConstraintBatchIds + iBegin, m_conInfos + iBegin, iEnd - iBegin, m_numConstraintRows);
+	}
 };
 
-
 static void expandConstraintRowsMt(int* destConstraintBatchIds, const int* srcConstraintBatchIds, const btBatchedConstraintInfo* conInfos, int numConstraints, int numConstraintRows)
 {
-    BT_PROFILE("expandConstraintRowsMt");
-    ExpandConstraintRowsLoop loop(destConstraintBatchIds, srcConstraintBatchIds, conInfos, numConstraintRows);
-    int grainSize = 600;
-    btParallelFor(0, numConstraints, grainSize, loop);
+	BT_PROFILE("expandConstraintRowsMt");
+	ExpandConstraintRowsLoop loop(destConstraintBatchIds, srcConstraintBatchIds, conInfos, numConstraintRows);
+	int grainSize = 600;
+	btParallelFor(0, numConstraints, grainSize, loop);
 }
 
-
 static void initBatchedConstraintInfoArray(btAlignedObjectArray<btBatchedConstraintInfo>* outConInfos, btConstraintArray* constraints)
 {
-    BT_PROFILE("initBatchedConstraintInfoArray");
-    btAlignedObjectArray<btBatchedConstraintInfo>& conInfos = *outConInfos;
-    int numConstraints = constraints->size();
-    conInfos.resizeNoInitialize(numConstraints);
+	BT_PROFILE("initBatchedConstraintInfoArray");
+	btAlignedObjectArray<btBatchedConstraintInfo>& conInfos = *outConInfos;
+	int numConstraints = constraints->size();
+	conInfos.resizeNoInitialize(numConstraints);
 
-    int newSize = initBatchedConstraintInfo(&outConInfos->at(0), constraints);
-    conInfos.resizeNoInitialize(newSize);
+	int newSize = initBatchedConstraintInfo(&outConInfos->at(0), constraints);
+	conInfos.resizeNoInitialize(newSize);
 }
 
-
 static void mergeSmallBatches(btBatchInfo* batches, int iBeginBatch, int iEndBatch, int minBatchSize, int maxBatchSize)
 {
-    BT_PROFILE("mergeSmallBatches");
-    for ( int iBatch = iEndBatch - 1; iBatch >= iBeginBatch; --iBatch )
-    {
-        btBatchInfo& batch = batches[ iBatch ];
-        if ( batch.mergeIndex == kNoMerge && batch.numConstraints > 0 && batch.numConstraints < minBatchSize )
-        {
-            for ( int iDestBatch = iBatch - 1; iDestBatch >= iBeginBatch; --iDestBatch )
-            {
-                btBatchInfo& destBatch = batches[ iDestBatch ];
-                if ( destBatch.mergeIndex == kNoMerge && ( destBatch.numConstraints + batch.numConstraints ) < maxBatchSize )
-                {
-                    destBatch.numConstraints += batch.numConstraints;
-                    batch.numConstraints = 0;
-                    batch.mergeIndex = iDestBatch;
-                    break;
-                }
-            }
-        }
-    }
-    // flatten mergeIndexes
-    // e.g. in case where A was merged into B and then B was merged into C, we need A to point to C instead of B
-    // Note: loop goes forward through batches because batches always merge from higher indexes to lower,
-    //     so by going from low to high it reduces the amount of trail-following
-    for ( int iBatch = iBeginBatch; iBatch < iEndBatch; ++iBatch )
-    {
-        btBatchInfo& batch = batches[ iBatch ];
-        if ( batch.mergeIndex != kNoMerge )
-        {
-            int iMergeDest = batches[ batch.mergeIndex ].mergeIndex;
-            // follow trail of merges to the end
-            while ( iMergeDest != kNoMerge )
-            {
-                int iNext = batches[ iMergeDest ].mergeIndex;
-                if ( iNext == kNoMerge )
-                {
-                    batch.mergeIndex = iMergeDest;
-                    break;
-                }
-                iMergeDest = iNext;
-            }
-        }
-    }
+	BT_PROFILE("mergeSmallBatches");
+	for (int iBatch = iEndBatch - 1; iBatch >= iBeginBatch; --iBatch)
+	{
+		btBatchInfo& batch = batches[iBatch];
+		if (batch.mergeIndex == kNoMerge && batch.numConstraints > 0 && batch.numConstraints < minBatchSize)
+		{
+			for (int iDestBatch = iBatch - 1; iDestBatch >= iBeginBatch; --iDestBatch)
+			{
+				btBatchInfo& destBatch = batches[iDestBatch];
+				if (destBatch.mergeIndex == kNoMerge && (destBatch.numConstraints + batch.numConstraints) < maxBatchSize)
+				{
+					destBatch.numConstraints += batch.numConstraints;
+					batch.numConstraints = 0;
+					batch.mergeIndex = iDestBatch;
+					break;
+				}
+			}
+		}
+	}
+	// flatten mergeIndexes
+	// e.g. in case where A was merged into B and then B was merged into C, we need A to point to C instead of B
+	// Note: loop goes forward through batches because batches always merge from higher indexes to lower,
+	//     so by going from low to high it reduces the amount of trail-following
+	for (int iBatch = iBeginBatch; iBatch < iEndBatch; ++iBatch)
+	{
+		btBatchInfo& batch = batches[iBatch];
+		if (batch.mergeIndex != kNoMerge)
+		{
+			int iMergeDest = batches[batch.mergeIndex].mergeIndex;
+			// follow trail of merges to the end
+			while (iMergeDest != kNoMerge)
+			{
+				int iNext = batches[iMergeDest].mergeIndex;
+				if (iNext == kNoMerge)
+				{
+					batch.mergeIndex = iMergeDest;
+					break;
+				}
+				iMergeDest = iNext;
+			}
+		}
+	}
 }
 
-
 static void updateConstraintBatchIdsForMerges(int* constraintBatchIds, int numConstraints, const btBatchInfo* batches, int numBatches)
 {
-    BT_PROFILE("updateConstraintBatchIdsForMerges");
-    // update batchIds to account for merges
-    for (int i = 0; i < numConstraints; ++i)
-    {
-        int iBatch = constraintBatchIds[i];
-        btAssert(iBatch < numBatches);
-        // if this constraint references a batch that was merged into another batch
-        if (batches[iBatch].mergeIndex != kNoMerge)
-        {
-            // update batchId
-            constraintBatchIds[i] = batches[iBatch].mergeIndex;
-        }
-    }
+	BT_PROFILE("updateConstraintBatchIdsForMerges");
+	// update batchIds to account for merges
+	for (int i = 0; i < numConstraints; ++i)
+	{
+		int iBatch = constraintBatchIds[i];
+		btAssert(iBatch < numBatches);
+		// if this constraint references a batch that was merged into another batch
+		if (batches[iBatch].mergeIndex != kNoMerge)
+		{
+			// update batchId
+			constraintBatchIds[i] = batches[iBatch].mergeIndex;
+		}
+	}
 }
 
-
 struct UpdateConstraintBatchIdsForMergesLoop : public btIParallelForBody
 {
-    int* m_constraintBatchIds;
-    const btBatchInfo* m_batches;
-    int m_numBatches;
+	int* m_constraintBatchIds;
+	const btBatchInfo* m_batches;
+	int m_numBatches;
 
-    UpdateConstraintBatchIdsForMergesLoop( int* constraintBatchIds, const btBatchInfo* batches, int numBatches )
-    {
-        m_constraintBatchIds = constraintBatchIds;
-        m_batches = batches;
-        m_numBatches = numBatches;
-    }
-    void forLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        BT_PROFILE( "UpdateConstraintBatchIdsForMergesLoop" );
-        updateConstraintBatchIdsForMerges( m_constraintBatchIds + iBegin, iEnd - iBegin, m_batches, m_numBatches );
-    }
+	UpdateConstraintBatchIdsForMergesLoop(int* constraintBatchIds, const btBatchInfo* batches, int numBatches)
+	{
+		m_constraintBatchIds = constraintBatchIds;
+		m_batches = batches;
+		m_numBatches = numBatches;
+	}
+	void forLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		BT_PROFILE("UpdateConstraintBatchIdsForMergesLoop");
+		updateConstraintBatchIdsForMerges(m_constraintBatchIds + iBegin, iEnd - iBegin, m_batches, m_numBatches);
+	}
 };
 
-
 static void updateConstraintBatchIdsForMergesMt(int* constraintBatchIds, int numConstraints, const btBatchInfo* batches, int numBatches)
 {
-    BT_PROFILE( "updateConstraintBatchIdsForMergesMt" );
-    UpdateConstraintBatchIdsForMergesLoop loop(constraintBatchIds, batches, numBatches);
-    int grainSize = 800;
-    btParallelFor(0, numConstraints, grainSize, loop);
+	BT_PROFILE("updateConstraintBatchIdsForMergesMt");
+	UpdateConstraintBatchIdsForMergesLoop loop(constraintBatchIds, batches, numBatches);
+	int grainSize = 800;
+	btParallelFor(0, numConstraints, grainSize, loop);
 }
 
-
 inline bool BatchCompare(const btBatchedConstraints::Range& a, const btBatchedConstraints::Range& b)
 {
-    int lenA = a.end - a.begin;
-    int lenB = b.end - b.begin;
-    return lenA > lenB;
+	int lenA = a.end - a.begin;
+	int lenB = b.end - b.begin;
+	return lenA > lenB;
 }
 
-
 static void writeOutConstraintIndicesForRangeOfBatches(btBatchedConstraints* bc,
-    const int* constraintBatchIds,
-    int numConstraints,
-    int* constraintIdPerBatch,
-    int batchBegin,
-    int batchEnd
-    )
+													   const int* constraintBatchIds,
+													   int numConstraints,
+													   int* constraintIdPerBatch,
+													   int batchBegin,
+													   int batchEnd)
 {
-    BT_PROFILE("writeOutConstraintIndicesForRangeOfBatches");
-    for ( int iCon = 0; iCon < numConstraints; ++iCon )
-    {
-        int iBatch = constraintBatchIds[ iCon ];
-        if (iBatch >= batchBegin && iBatch < batchEnd)
-        {
-            int iDestCon = constraintIdPerBatch[ iBatch ];
-            constraintIdPerBatch[ iBatch ] = iDestCon + 1;
-            bc->m_constraintIndices[ iDestCon ] = iCon;
-        }
-    }
+	BT_PROFILE("writeOutConstraintIndicesForRangeOfBatches");
+	for (int iCon = 0; iCon < numConstraints; ++iCon)
+	{
+		int iBatch = constraintBatchIds[iCon];
+		if (iBatch >= batchBegin && iBatch < batchEnd)
+		{
+			int iDestCon = constraintIdPerBatch[iBatch];
+			constraintIdPerBatch[iBatch] = iDestCon + 1;
+			bc->m_constraintIndices[iDestCon] = iCon;
+		}
+	}
 }
 
-
 struct WriteOutConstraintIndicesLoop : public btIParallelForBody
 {
-    btBatchedConstraints* m_batchedConstraints;
-    const int* m_constraintBatchIds;
-    int m_numConstraints;
-    int* m_constraintIdPerBatch;
-    int m_maxNumBatchesPerPhase;
+	btBatchedConstraints* m_batchedConstraints;
+	const int* m_constraintBatchIds;
+	int m_numConstraints;
+	int* m_constraintIdPerBatch;
+	int m_maxNumBatchesPerPhase;
 
-    WriteOutConstraintIndicesLoop( btBatchedConstraints* bc, const int* constraintBatchIds, int numConstraints, int* constraintIdPerBatch, int maxNumBatchesPerPhase )
-    {
-        m_batchedConstraints = bc;
-        m_constraintBatchIds = constraintBatchIds;
-        m_numConstraints = numConstraints;
-        m_constraintIdPerBatch = constraintIdPerBatch;
-        m_maxNumBatchesPerPhase = maxNumBatchesPerPhase;
-    }
-    void forLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        BT_PROFILE( "WriteOutConstraintIndicesLoop" );
-        int batchBegin = iBegin * m_maxNumBatchesPerPhase;
-        int batchEnd = iEnd * m_maxNumBatchesPerPhase;
-        writeOutConstraintIndicesForRangeOfBatches(m_batchedConstraints,
-            m_constraintBatchIds,
-            m_numConstraints,
-            m_constraintIdPerBatch,
-            batchBegin,
-            batchEnd
-        );
-    }
+	WriteOutConstraintIndicesLoop(btBatchedConstraints* bc, const int* constraintBatchIds, int numConstraints, int* constraintIdPerBatch, int maxNumBatchesPerPhase)
+	{
+		m_batchedConstraints = bc;
+		m_constraintBatchIds = constraintBatchIds;
+		m_numConstraints = numConstraints;
+		m_constraintIdPerBatch = constraintIdPerBatch;
+		m_maxNumBatchesPerPhase = maxNumBatchesPerPhase;
+	}
+	void forLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		BT_PROFILE("WriteOutConstraintIndicesLoop");
+		int batchBegin = iBegin * m_maxNumBatchesPerPhase;
+		int batchEnd = iEnd * m_maxNumBatchesPerPhase;
+		writeOutConstraintIndicesForRangeOfBatches(m_batchedConstraints,
+												   m_constraintBatchIds,
+												   m_numConstraints,
+												   m_constraintIdPerBatch,
+												   batchBegin,
+												   batchEnd);
+	}
 };
 
-
 static void writeOutConstraintIndicesMt(btBatchedConstraints* bc,
-    const int* constraintBatchIds,
-    int numConstraints,
-    int* constraintIdPerBatch,
-    int maxNumBatchesPerPhase,
-    int numPhases
-    )
+										const int* constraintBatchIds,
+										int numConstraints,
+										int* constraintIdPerBatch,
+										int maxNumBatchesPerPhase,
+										int numPhases)
 {
-    BT_PROFILE("writeOutConstraintIndicesMt");
-    bool inParallel = true;
-    if (inParallel)
-    {
-        WriteOutConstraintIndicesLoop loop( bc, constraintBatchIds, numConstraints, constraintIdPerBatch, maxNumBatchesPerPhase );
-        btParallelFor( 0, numPhases, 1, loop );
-    }
-    else
-    {
-        for ( int iCon = 0; iCon < numConstraints; ++iCon )
-        {
-            int iBatch = constraintBatchIds[ iCon ];
-            int iDestCon = constraintIdPerBatch[ iBatch ];
-            constraintIdPerBatch[ iBatch ] = iDestCon + 1;
-            bc->m_constraintIndices[ iDestCon ] = iCon;
-        }
-    }
+	BT_PROFILE("writeOutConstraintIndicesMt");
+	bool inParallel = true;
+	if (inParallel)
+	{
+		WriteOutConstraintIndicesLoop loop(bc, constraintBatchIds, numConstraints, constraintIdPerBatch, maxNumBatchesPerPhase);
+		btParallelFor(0, numPhases, 1, loop);
+	}
+	else
+	{
+		for (int iCon = 0; iCon < numConstraints; ++iCon)
+		{
+			int iBatch = constraintBatchIds[iCon];
+			int iDestCon = constraintIdPerBatch[iBatch];
+			constraintIdPerBatch[iBatch] = iDestCon + 1;
+			bc->m_constraintIndices[iDestCon] = iCon;
+		}
+	}
 }
 
-
 static void writeGrainSizes(btBatchedConstraints* bc)
 {
-    typedef btBatchedConstraints::Range Range;
-    int numPhases = bc->m_phases.size();
-    bc->m_phaseGrainSize.resizeNoInitialize(numPhases);
-    int numThreads = btGetTaskScheduler()->getNumThreads();
-    for (int iPhase = 0; iPhase < numPhases; ++iPhase)
-    {
-        const Range& phase = bc->m_phases[ iPhase ];
-        int numBatches = phase.end - phase.begin;
-        float grainSize = floor((0.25f*numBatches / float(numThreads)) + 0.0f);
-        bc->m_phaseGrainSize[ iPhase ] = btMax(1, int(grainSize));
-    }
+	typedef btBatchedConstraints::Range Range;
+	int numPhases = bc->m_phases.size();
+	bc->m_phaseGrainSize.resizeNoInitialize(numPhases);
+	int numThreads = btGetTaskScheduler()->getNumThreads();
+	for (int iPhase = 0; iPhase < numPhases; ++iPhase)
+	{
+		const Range& phase = bc->m_phases[iPhase];
+		int numBatches = phase.end - phase.begin;
+		float grainSize = floor((0.25f * numBatches / float(numThreads)) + 0.0f);
+		bc->m_phaseGrainSize[iPhase] = btMax(1, int(grainSize));
+	}
 }
 
-
 static void writeOutBatches(btBatchedConstraints* bc,
-    const int* constraintBatchIds,
-    int numConstraints,
-    const btBatchInfo* batches,
-    int* batchWork,
-    int maxNumBatchesPerPhase,
-    int numPhases
-)
+							const int* constraintBatchIds,
+							int numConstraints,
+							const btBatchInfo* batches,
+							int* batchWork,
+							int maxNumBatchesPerPhase,
+							int numPhases)
 {
-    BT_PROFILE("writeOutBatches");
-    typedef btBatchedConstraints::Range Range;
-    bc->m_constraintIndices.reserve( numConstraints );
-    bc->m_batches.resizeNoInitialize( 0 );
-    bc->m_phases.resizeNoInitialize( 0 );
+	BT_PROFILE("writeOutBatches");
+	typedef btBatchedConstraints::Range Range;
+	bc->m_constraintIndices.reserve(numConstraints);
+	bc->m_batches.resizeNoInitialize(0);
+	bc->m_phases.resizeNoInitialize(0);
 
-    //int maxNumBatches = numPhases * maxNumBatchesPerPhase;
-    {
-        int* constraintIdPerBatch = batchWork;  // for each batch, keep an index into the next available slot in the m_constraintIndices array
-        int iConstraint = 0;
-        for (int iPhase = 0; iPhase < numPhases; ++iPhase)
-        {
-            int curPhaseBegin = bc->m_batches.size();
-            int iBegin = iPhase * maxNumBatchesPerPhase;
-            int iEnd = iBegin + maxNumBatchesPerPhase;
-            for ( int i = iBegin; i < iEnd; ++i )
-            {
-                const btBatchInfo& batch = batches[ i ];
-                int curBatchBegin = iConstraint;
-                constraintIdPerBatch[ i ] = curBatchBegin;  // record the start of each batch in m_constraintIndices array
-                int numConstraints = batch.numConstraints;
-                iConstraint += numConstraints;
-                if ( numConstraints > 0 )
-                {
-                    bc->m_batches.push_back( Range( curBatchBegin, iConstraint ) );
-                }
-            }
-            // if any batches were emitted this phase,
-            if ( bc->m_batches.size() > curPhaseBegin )
-            {
-                // output phase
-                bc->m_phases.push_back( Range( curPhaseBegin, bc->m_batches.size() ) );
-            }
-        }
+	//int maxNumBatches = numPhases * maxNumBatchesPerPhase;
+	{
+		int* constraintIdPerBatch = batchWork;  // for each batch, keep an index into the next available slot in the m_constraintIndices array
+		int iConstraint = 0;
+		for (int iPhase = 0; iPhase < numPhases; ++iPhase)
+		{
+			int curPhaseBegin = bc->m_batches.size();
+			int iBegin = iPhase * maxNumBatchesPerPhase;
+			int iEnd = iBegin + maxNumBatchesPerPhase;
+			for (int i = iBegin; i < iEnd; ++i)
+			{
+				const btBatchInfo& batch = batches[i];
+				int curBatchBegin = iConstraint;
+				constraintIdPerBatch[i] = curBatchBegin;  // record the start of each batch in m_constraintIndices array
+				int numConstraints = batch.numConstraints;
+				iConstraint += numConstraints;
+				if (numConstraints > 0)
+				{
+					bc->m_batches.push_back(Range(curBatchBegin, iConstraint));
+				}
+			}
+			// if any batches were emitted this phase,
+			if (bc->m_batches.size() > curPhaseBegin)
+			{
+				// output phase
+				bc->m_phases.push_back(Range(curPhaseBegin, bc->m_batches.size()));
+			}
+		}
 
-        btAssert(iConstraint == numConstraints);
-        bc->m_constraintIndices.resizeNoInitialize( numConstraints );
-        writeOutConstraintIndicesMt( bc, constraintBatchIds, numConstraints, constraintIdPerBatch, maxNumBatchesPerPhase, numPhases );
-    }
-    // for each phase
-    for (int iPhase = 0; iPhase < bc->m_phases.size(); ++iPhase)
-    {
-        // sort the batches from largest to smallest (can be helpful to some task schedulers)
-        const Range& curBatches = bc->m_phases[iPhase];
-        bc->m_batches.quickSortInternal(BatchCompare, curBatches.begin, curBatches.end-1);
-    }
-    bc->m_phaseOrder.resize(bc->m_phases.size());
-    for (int i = 0; i < bc->m_phases.size(); ++i)
-    {
-        bc->m_phaseOrder[i] = i;
-    }
-    writeGrainSizes(bc);
+		btAssert(iConstraint == numConstraints);
+		bc->m_constraintIndices.resizeNoInitialize(numConstraints);
+		writeOutConstraintIndicesMt(bc, constraintBatchIds, numConstraints, constraintIdPerBatch, maxNumBatchesPerPhase, numPhases);
+	}
+	// for each phase
+	for (int iPhase = 0; iPhase < bc->m_phases.size(); ++iPhase)
+	{
+		// sort the batches from largest to smallest (can be helpful to some task schedulers)
+		const Range& curBatches = bc->m_phases[iPhase];
+		bc->m_batches.quickSortInternal(BatchCompare, curBatches.begin, curBatches.end - 1);
+	}
+	bc->m_phaseOrder.resize(bc->m_phases.size());
+	for (int i = 0; i < bc->m_phases.size(); ++i)
+	{
+		bc->m_phaseOrder[i] = i;
+	}
+	writeGrainSizes(bc);
 }
 
-
 //
 // PreallocatedMemoryHelper -- helper object for allocating a number of chunks of memory in a single contiguous block.
 //                             It is generally more efficient to do a single larger allocation than many smaller allocations.
@@ -639,191 +605,184 @@ static void writeOutBatches(btBatchedConstraints* bc,
 template <int N>
 class PreallocatedMemoryHelper
 {
-    struct Chunk
-    {
-        void** ptr;
-        size_t size;
-    };
-    Chunk m_chunks[N];
-    int m_numChunks;
+	struct Chunk
+	{
+		void** ptr;
+		size_t size;
+	};
+	Chunk m_chunks[N];
+	int m_numChunks;
+
 public:
-    PreallocatedMemoryHelper() {m_numChunks=0;}
-    void addChunk( void** ptr, size_t sz )
-    {
-        btAssert( m_numChunks < N );
-        if ( m_numChunks < N )
-        {
-            Chunk& chunk = m_chunks[ m_numChunks ];
-            chunk.ptr = ptr;
-            chunk.size = sz;
-            m_numChunks++;
-        }
-    }
-    size_t getSizeToAllocate() const
-    {
-        size_t totalSize = 0;
-        for (int i = 0; i < m_numChunks; ++i)
-        {
-            totalSize += m_chunks[i].size;
-        }
-        return totalSize;
-    }
-    void setChunkPointers(void* mem) const
-    {
-        size_t totalSize = 0;
-        for (int i = 0; i < m_numChunks; ++i)
-        {
-            const Chunk& chunk = m_chunks[ i ];
-            char* chunkPtr = static_cast<char*>(mem) + totalSize;
-            *chunk.ptr = chunkPtr;
-            totalSize += chunk.size;
-        }
-    }
+	PreallocatedMemoryHelper() { m_numChunks = 0; }
+	void addChunk(void** ptr, size_t sz)
+	{
+		btAssert(m_numChunks < N);
+		if (m_numChunks < N)
+		{
+			Chunk& chunk = m_chunks[m_numChunks];
+			chunk.ptr = ptr;
+			chunk.size = sz;
+			m_numChunks++;
+		}
+	}
+	size_t getSizeToAllocate() const
+	{
+		size_t totalSize = 0;
+		for (int i = 0; i < m_numChunks; ++i)
+		{
+			totalSize += m_chunks[i].size;
+		}
+		return totalSize;
+	}
+	void setChunkPointers(void* mem) const
+	{
+		size_t totalSize = 0;
+		for (int i = 0; i < m_numChunks; ++i)
+		{
+			const Chunk& chunk = m_chunks[i];
+			char* chunkPtr = static_cast<char*>(mem) + totalSize;
+			*chunk.ptr = chunkPtr;
+			totalSize += chunk.size;
+		}
+	}
 };
 
-
-
 static btVector3 findMaxDynamicConstraintExtent(
-    btVector3* bodyPositions,
-    bool* bodyDynamicFlags,
-    btBatchedConstraintInfo* conInfos,
-    int numConstraints,
-    int numBodies
-    )
+	btVector3* bodyPositions,
+	bool* bodyDynamicFlags,
+	btBatchedConstraintInfo* conInfos,
+	int numConstraints,
+	int numBodies)
 {
-    BT_PROFILE("findMaxDynamicConstraintExtent");
-    btVector3 consExtent = btVector3(1,1,1) * 0.001;
-    for (int iCon = 0; iCon < numConstraints; ++iCon)
-    {
-        const btBatchedConstraintInfo& con = conInfos[ iCon ];
-        int iBody0 = con.bodyIds[0];
-        int iBody1 = con.bodyIds[1];
-        btAssert(iBody0 >= 0 && iBody0 < numBodies);
-        btAssert(iBody1 >= 0 && iBody1 < numBodies);
-        // is it a dynamic constraint?
-        if (bodyDynamicFlags[iBody0] && bodyDynamicFlags[iBody1])
-        {
-            btVector3 delta = bodyPositions[iBody1] - bodyPositions[iBody0];
-            consExtent.setMax(delta.absolute());
-        }
-    }
-    return consExtent;
+	BT_PROFILE("findMaxDynamicConstraintExtent");
+	btVector3 consExtent = btVector3(1, 1, 1) * 0.001;
+	for (int iCon = 0; iCon < numConstraints; ++iCon)
+	{
+		const btBatchedConstraintInfo& con = conInfos[iCon];
+		int iBody0 = con.bodyIds[0];
+		int iBody1 = con.bodyIds[1];
+		btAssert(iBody0 >= 0 && iBody0 < numBodies);
+		btAssert(iBody1 >= 0 && iBody1 < numBodies);
+		// is it a dynamic constraint?
+		if (bodyDynamicFlags[iBody0] && bodyDynamicFlags[iBody1])
+		{
+			btVector3 delta = bodyPositions[iBody1] - bodyPositions[iBody0];
+			consExtent.setMax(delta.absolute());
+		}
+	}
+	return consExtent;
 }
 
-
 struct btIntVec3
 {
-    int m_ints[ 3 ];
+	int m_ints[3];
 
-    SIMD_FORCE_INLINE const int& operator[](int i) const {return m_ints[i];}
-    SIMD_FORCE_INLINE int&       operator[](int i)       {return m_ints[i];}
+	SIMD_FORCE_INLINE const int& operator[](int i) const { return m_ints[i]; }
+	SIMD_FORCE_INLINE int& operator[](int i) { return m_ints[i]; }
 };
 
-
 struct AssignConstraintsToGridBatchesParams
 {
-    bool* bodyDynamicFlags;
-    btIntVec3* bodyGridCoords;
-    int numBodies;
-    btBatchedConstraintInfo* conInfos;
-    int* constraintBatchIds;
-    btIntVec3 gridChunkDim;
-    int maxNumBatchesPerPhase;
-    int numPhases;
-    int phaseMask;
+	bool* bodyDynamicFlags;
+	btIntVec3* bodyGridCoords;
+	int numBodies;
+	btBatchedConstraintInfo* conInfos;
+	int* constraintBatchIds;
+	btIntVec3 gridChunkDim;
+	int maxNumBatchesPerPhase;
+	int numPhases;
+	int phaseMask;
 
-    AssignConstraintsToGridBatchesParams()
-    {
-        memset(this, 0, sizeof(*this));
-    }
+	AssignConstraintsToGridBatchesParams()
+	{
+		memset(this, 0, sizeof(*this));
+	}
 };
 
-
 static void assignConstraintsToGridBatches(const AssignConstraintsToGridBatchesParams& params, int iConBegin, int iConEnd)
 {
-    BT_PROFILE("assignConstraintsToGridBatches");
-    // (can be done in parallel)
-    for ( int iCon = iConBegin; iCon < iConEnd; ++iCon )
-    {
-        const btBatchedConstraintInfo& con = params.conInfos[ iCon ];
-        int iBody0 = con.bodyIds[ 0 ];
-        int iBody1 = con.bodyIds[ 1 ];
-        int iPhase = iCon; //iBody0; // pseudorandom choice to distribute evenly amongst phases
-        iPhase &= params.phaseMask;
-        int gridCoord[ 3 ];
-        // is it a dynamic constraint?
-        if ( params.bodyDynamicFlags[ iBody0 ] && params.bodyDynamicFlags[ iBody1 ] )
-        {
-            const btIntVec3& body0Coords = params.bodyGridCoords[iBody0];
-            const btIntVec3& body1Coords = params.bodyGridCoords[iBody1];
-            // for each dimension x,y,z,
-            for (int i = 0; i < 3; ++i)
-            {
-                int coordMin = btMin(body0Coords.m_ints[i], body1Coords.m_ints[i]);
-                int coordMax = btMax(body0Coords.m_ints[i], body1Coords.m_ints[i]);
-                if (coordMin != coordMax)
-                {
-                    btAssert( coordMax == coordMin + 1 );
-                    if ((coordMin&1) == 0)
-                    {
-                        iPhase &= ~(1 << i); // force bit off
-                    }
-                    else
-                    {
-                        iPhase |= (1 << i); // force bit on
-                        iPhase &= params.phaseMask;
-                    }
-                }
-                gridCoord[ i ] = coordMin;
-            }
-        }
-        else
-        {
-            if ( !params.bodyDynamicFlags[ iBody0 ] )
-            {
-                iBody0 = con.bodyIds[ 1 ];
-            }
-            btAssert(params.bodyDynamicFlags[ iBody0 ]);
-            const btIntVec3& body0Coords = params.bodyGridCoords[iBody0];
-            // for each dimension x,y,z,
-            for ( int i = 0; i < 3; ++i )
-            {
-                gridCoord[ i ] = body0Coords.m_ints[ i ];
-            }
-        }
-        // calculate chunk coordinates
-        int chunkCoord[ 3 ];
-        btIntVec3 gridChunkDim = params.gridChunkDim;
-        // for each dimension x,y,z,
-        for ( int i = 0; i < 3; ++i )
-        {
-            int coordOffset = ( iPhase >> i ) & 1;
-            chunkCoord[ i ] = (gridCoord[ i ] - coordOffset)/2;
-            btClamp( chunkCoord[ i ], 0, gridChunkDim[ i ] - 1);
-            btAssert( chunkCoord[ i ] < gridChunkDim[ i ] );
-        }
-        int iBatch = iPhase * params.maxNumBatchesPerPhase + chunkCoord[ 0 ] + chunkCoord[ 1 ] * gridChunkDim[ 0 ] + chunkCoord[ 2 ] * gridChunkDim[ 0 ] * gridChunkDim[ 1 ];
-        btAssert(iBatch >= 0 && iBatch < params.maxNumBatchesPerPhase*params.numPhases);
-        params.constraintBatchIds[ iCon ] = iBatch;
-    }
+	BT_PROFILE("assignConstraintsToGridBatches");
+	// (can be done in parallel)
+	for (int iCon = iConBegin; iCon < iConEnd; ++iCon)
+	{
+		const btBatchedConstraintInfo& con = params.conInfos[iCon];
+		int iBody0 = con.bodyIds[0];
+		int iBody1 = con.bodyIds[1];
+		int iPhase = iCon;  //iBody0; // pseudorandom choice to distribute evenly amongst phases
+		iPhase &= params.phaseMask;
+		int gridCoord[3];
+		// is it a dynamic constraint?
+		if (params.bodyDynamicFlags[iBody0] && params.bodyDynamicFlags[iBody1])
+		{
+			const btIntVec3& body0Coords = params.bodyGridCoords[iBody0];
+			const btIntVec3& body1Coords = params.bodyGridCoords[iBody1];
+			// for each dimension x,y,z,
+			for (int i = 0; i < 3; ++i)
+			{
+				int coordMin = btMin(body0Coords.m_ints[i], body1Coords.m_ints[i]);
+				int coordMax = btMax(body0Coords.m_ints[i], body1Coords.m_ints[i]);
+				if (coordMin != coordMax)
+				{
+					btAssert(coordMax == coordMin + 1);
+					if ((coordMin & 1) == 0)
+					{
+						iPhase &= ~(1 << i);  // force bit off
+					}
+					else
+					{
+						iPhase |= (1 << i);  // force bit on
+						iPhase &= params.phaseMask;
+					}
+				}
+				gridCoord[i] = coordMin;
+			}
+		}
+		else
+		{
+			if (!params.bodyDynamicFlags[iBody0])
+			{
+				iBody0 = con.bodyIds[1];
+			}
+			btAssert(params.bodyDynamicFlags[iBody0]);
+			const btIntVec3& body0Coords = params.bodyGridCoords[iBody0];
+			// for each dimension x,y,z,
+			for (int i = 0; i < 3; ++i)
+			{
+				gridCoord[i] = body0Coords.m_ints[i];
+			}
+		}
+		// calculate chunk coordinates
+		int chunkCoord[3];
+		btIntVec3 gridChunkDim = params.gridChunkDim;
+		// for each dimension x,y,z,
+		for (int i = 0; i < 3; ++i)
+		{
+			int coordOffset = (iPhase >> i) & 1;
+			chunkCoord[i] = (gridCoord[i] - coordOffset) / 2;
+			btClamp(chunkCoord[i], 0, gridChunkDim[i] - 1);
+			btAssert(chunkCoord[i] < gridChunkDim[i]);
+		}
+		int iBatch = iPhase * params.maxNumBatchesPerPhase + chunkCoord[0] + chunkCoord[1] * gridChunkDim[0] + chunkCoord[2] * gridChunkDim[0] * gridChunkDim[1];
+		btAssert(iBatch >= 0 && iBatch < params.maxNumBatchesPerPhase * params.numPhases);
+		params.constraintBatchIds[iCon] = iBatch;
+	}
 }
 
-
 struct AssignConstraintsToGridBatchesLoop : public btIParallelForBody
 {
-    const AssignConstraintsToGridBatchesParams* m_params;
+	const AssignConstraintsToGridBatchesParams* m_params;
 
-    AssignConstraintsToGridBatchesLoop( const AssignConstraintsToGridBatchesParams& params )
-    {
-        m_params = &params;
-    }
-    void forLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        assignConstraintsToGridBatches(*m_params, iBegin, iEnd);
-    }
+	AssignConstraintsToGridBatchesLoop(const AssignConstraintsToGridBatchesParams& params)
+	{
+		m_params = &params;
+	}
+	void forLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		assignConstraintsToGridBatches(*m_params, iBegin, iEnd);
+	}
 };
 
-
 //
 // setupSpatialGridBatchesMt -- generate batches using a uniform 3D grid
 //
@@ -853,276 +812,269 @@ to 4. With fewer phases, there are more constraints per phase and this makes it
 */
 //
 static void setupSpatialGridBatchesMt(
-    btBatchedConstraints* batchedConstraints,
-    btAlignedObjectArray<char>* scratchMemory,
-    btConstraintArray* constraints,
-    const btAlignedObjectArray<btSolverBody>& bodies,
-    int minBatchSize,
-    int maxBatchSize,
-    bool use2DGrid
-)
+	btBatchedConstraints* batchedConstraints,
+	btAlignedObjectArray<char>* scratchMemory,
+	btConstraintArray* constraints,
+	const btAlignedObjectArray<btSolverBody>& bodies,
+	int minBatchSize,
+	int maxBatchSize,
+	bool use2DGrid)
 {
-    BT_PROFILE("setupSpatialGridBatchesMt");
-    const int numPhases = 8;
-    int numConstraints = constraints->size();
-    int numConstraintRows = constraints->size();
+	BT_PROFILE("setupSpatialGridBatchesMt");
+	const int numPhases = 8;
+	int numConstraints = constraints->size();
+	int numConstraintRows = constraints->size();
 
-    const int maxGridChunkCount = 128;
-    int allocNumBatchesPerPhase = maxGridChunkCount;
-    int minNumBatchesPerPhase = 16;
-    int allocNumBatches = allocNumBatchesPerPhase * numPhases;
+	const int maxGridChunkCount = 128;
+	int allocNumBatchesPerPhase = maxGridChunkCount;
+	int minNumBatchesPerPhase = 16;
+	int allocNumBatches = allocNumBatchesPerPhase * numPhases;
 
-    btVector3* bodyPositions = NULL;
-    bool* bodyDynamicFlags = NULL;
-    btIntVec3* bodyGridCoords = NULL;
-    btBatchInfo* batches = NULL;
-    int* batchWork = NULL;
-    btBatchedConstraintInfo* conInfos = NULL;
-    int* constraintBatchIds = NULL;
-    int* constraintRowBatchIds = NULL;
-    {
-        PreallocatedMemoryHelper<10> memHelper;
-        memHelper.addChunk( (void**) &bodyPositions, sizeof( btVector3 ) * bodies.size() );
-        memHelper.addChunk( (void**) &bodyDynamicFlags, sizeof( bool ) * bodies.size() );
-        memHelper.addChunk( (void**) &bodyGridCoords, sizeof( btIntVec3 ) * bodies.size() );
-        memHelper.addChunk( (void**) &batches, sizeof( btBatchInfo )* allocNumBatches );
-        memHelper.addChunk( (void**) &batchWork, sizeof( int )* allocNumBatches );
-        memHelper.addChunk( (void**) &conInfos, sizeof( btBatchedConstraintInfo ) * numConstraints );
-        memHelper.addChunk( (void**) &constraintBatchIds, sizeof( int ) * numConstraints );
-        memHelper.addChunk( (void**) &constraintRowBatchIds, sizeof( int ) * numConstraintRows );
-        size_t scratchSize = memHelper.getSizeToAllocate();
-        // if we need to reallocate
-        if (scratchMemory->capacity() < scratchSize)
-        {
-            // allocate 6.25% extra to avoid repeated reallocs
-            scratchMemory->reserve( scratchSize + scratchSize/16 );
-        }
-        scratchMemory->resizeNoInitialize( scratchSize );
-        char* memPtr = &scratchMemory->at(0);
-        memHelper.setChunkPointers( memPtr );
-    }
+	btVector3* bodyPositions = NULL;
+	bool* bodyDynamicFlags = NULL;
+	btIntVec3* bodyGridCoords = NULL;
+	btBatchInfo* batches = NULL;
+	int* batchWork = NULL;
+	btBatchedConstraintInfo* conInfos = NULL;
+	int* constraintBatchIds = NULL;
+	int* constraintRowBatchIds = NULL;
+	{
+		PreallocatedMemoryHelper<10> memHelper;
+		memHelper.addChunk((void**)&bodyPositions, sizeof(btVector3) * bodies.size());
+		memHelper.addChunk((void**)&bodyDynamicFlags, sizeof(bool) * bodies.size());
+		memHelper.addChunk((void**)&bodyGridCoords, sizeof(btIntVec3) * bodies.size());
+		memHelper.addChunk((void**)&batches, sizeof(btBatchInfo) * allocNumBatches);
+		memHelper.addChunk((void**)&batchWork, sizeof(int) * allocNumBatches);
+		memHelper.addChunk((void**)&conInfos, sizeof(btBatchedConstraintInfo) * numConstraints);
+		memHelper.addChunk((void**)&constraintBatchIds, sizeof(int) * numConstraints);
+		memHelper.addChunk((void**)&constraintRowBatchIds, sizeof(int) * numConstraintRows);
+		size_t scratchSize = memHelper.getSizeToAllocate();
+		// if we need to reallocate
+		if (scratchMemory->capacity() < scratchSize)
+		{
+			// allocate 6.25% extra to avoid repeated reallocs
+			scratchMemory->reserve(scratchSize + scratchSize / 16);
+		}
+		scratchMemory->resizeNoInitialize(scratchSize);
+		char* memPtr = &scratchMemory->at(0);
+		memHelper.setChunkPointers(memPtr);
+	}
 
-    numConstraints = initBatchedConstraintInfo(conInfos, constraints);
+	numConstraints = initBatchedConstraintInfo(conInfos, constraints);
 
-    // compute bounding box around all dynamic bodies
-    // (could be done in parallel)
-    btVector3 bboxMin(BT_LARGE_FLOAT, BT_LARGE_FLOAT, BT_LARGE_FLOAT);
-    btVector3 bboxMax = -bboxMin;
-    //int dynamicBodyCount = 0;
-    for (int i = 0; i < bodies.size(); ++i)
-    {
-        const btSolverBody& body = bodies[i];
-        btVector3 bodyPos = body.getWorldTransform().getOrigin();
-        bool isDynamic = ( body.internalGetInvMass().x() > btScalar( 0 ) );
-        bodyPositions[i] = bodyPos;
-        bodyDynamicFlags[i] = isDynamic;
-        if (isDynamic)
-        {
-            //dynamicBodyCount++;
-            bboxMin.setMin(bodyPos);
-            bboxMax.setMax(bodyPos);
-        }
-    }
+	// compute bounding box around all dynamic bodies
+	// (could be done in parallel)
+	btVector3 bboxMin(BT_LARGE_FLOAT, BT_LARGE_FLOAT, BT_LARGE_FLOAT);
+	btVector3 bboxMax = -bboxMin;
+	//int dynamicBodyCount = 0;
+	for (int i = 0; i < bodies.size(); ++i)
+	{
+		const btSolverBody& body = bodies[i];
+		btVector3 bodyPos = body.getWorldTransform().getOrigin();
+		bool isDynamic = (body.internalGetInvMass().x() > btScalar(0));
+		bodyPositions[i] = bodyPos;
+		bodyDynamicFlags[i] = isDynamic;
+		if (isDynamic)
+		{
+			//dynamicBodyCount++;
+			bboxMin.setMin(bodyPos);
+			bboxMax.setMax(bodyPos);
+		}
+	}
 
-    // find max extent of all dynamic constraints
-    // (could be done in parallel)
-    btVector3 consExtent = findMaxDynamicConstraintExtent(bodyPositions, bodyDynamicFlags, conInfos, numConstraints, bodies.size());
+	// find max extent of all dynamic constraints
+	// (could be done in parallel)
+	btVector3 consExtent = findMaxDynamicConstraintExtent(bodyPositions, bodyDynamicFlags, conInfos, numConstraints, bodies.size());
 
-    btVector3 gridExtent = bboxMax - bboxMin;
+	btVector3 gridExtent = bboxMax - bboxMin;
 
-    btVector3 gridCellSize = consExtent;
-    int gridDim[3];
-    gridDim[ 0 ] = int( 1.0 + gridExtent.x() / gridCellSize.x() );
-    gridDim[ 1 ] = int( 1.0 + gridExtent.y() / gridCellSize.y() );
-    gridDim[ 2 ] = int( 1.0 + gridExtent.z() / gridCellSize.z() );
+	btVector3 gridCellSize = consExtent;
+	int gridDim[3];
+	gridDim[0] = int(1.0 + gridExtent.x() / gridCellSize.x());
+	gridDim[1] = int(1.0 + gridExtent.y() / gridCellSize.y());
+	gridDim[2] = int(1.0 + gridExtent.z() / gridCellSize.z());
 
-    // if we can collapse an axis, it will cut our number of phases in half which could be more efficient
-    int phaseMask = 7;
-    bool collapseAxis = use2DGrid;
-    if ( collapseAxis )
-    {
-        // pick the smallest axis to collapse, leaving us with the greatest number of cells in our grid
-        int iAxisToCollapse = 0;
-        int axisDim = gridDim[iAxisToCollapse];
-        //for each dimension
-        for ( int i = 0; i < 3; ++i )
-        {
-            if (gridDim[i] < axisDim)
-            {
-                iAxisToCollapse = i;
-                axisDim = gridDim[i];
-            }
-        }
-        // collapse it
-        gridCellSize[iAxisToCollapse] = gridExtent[iAxisToCollapse] * 2.0f;
-        phaseMask &= ~(1 << iAxisToCollapse);
-    }
+	// if we can collapse an axis, it will cut our number of phases in half which could be more efficient
+	int phaseMask = 7;
+	bool collapseAxis = use2DGrid;
+	if (collapseAxis)
+	{
+		// pick the smallest axis to collapse, leaving us with the greatest number of cells in our grid
+		int iAxisToCollapse = 0;
+		int axisDim = gridDim[iAxisToCollapse];
+		//for each dimension
+		for (int i = 0; i < 3; ++i)
+		{
+			if (gridDim[i] < axisDim)
+			{
+				iAxisToCollapse = i;
+				axisDim = gridDim[i];
+			}
+		}
+		// collapse it
+		gridCellSize[iAxisToCollapse] = gridExtent[iAxisToCollapse] * 2.0f;
+		phaseMask &= ~(1 << iAxisToCollapse);
+	}
 
-    int numGridChunks = 0;
-    btIntVec3 gridChunkDim;  // each chunk is 2x2x2 group of cells
-    while (true)
-    {
-        gridDim[0] = int( 1.0 + gridExtent.x() / gridCellSize.x() );
-        gridDim[1] = int( 1.0 + gridExtent.y() / gridCellSize.y() );
-        gridDim[2] = int( 1.0 + gridExtent.z() / gridCellSize.z() );
-        gridChunkDim[ 0 ] = btMax( 1, ( gridDim[ 0 ] + 0 ) / 2 );
-        gridChunkDim[ 1 ] = btMax( 1, ( gridDim[ 1 ] + 0 ) / 2 );
-        gridChunkDim[ 2 ] = btMax( 1, ( gridDim[ 2 ] + 0 ) / 2 );
-        numGridChunks = gridChunkDim[ 0 ] * gridChunkDim[ 1 ] * gridChunkDim[ 2 ];
-        float nChunks = float(gridChunkDim[0]) * float(gridChunkDim[1]) * float(gridChunkDim[2]);  // suceptible to integer overflow
-        if ( numGridChunks <= maxGridChunkCount && nChunks <= maxGridChunkCount )
-        {
-            break;
-        }
-        gridCellSize *= 1.25; // should roughly cut numCells in half
-    }
-    btAssert(numGridChunks <= maxGridChunkCount );
-    int maxNumBatchesPerPhase = numGridChunks;
+	int numGridChunks = 0;
+	btIntVec3 gridChunkDim;  // each chunk is 2x2x2 group of cells
+	while (true)
+	{
+		gridDim[0] = int(1.0 + gridExtent.x() / gridCellSize.x());
+		gridDim[1] = int(1.0 + gridExtent.y() / gridCellSize.y());
+		gridDim[2] = int(1.0 + gridExtent.z() / gridCellSize.z());
+		gridChunkDim[0] = btMax(1, (gridDim[0] + 0) / 2);
+		gridChunkDim[1] = btMax(1, (gridDim[1] + 0) / 2);
+		gridChunkDim[2] = btMax(1, (gridDim[2] + 0) / 2);
+		numGridChunks = gridChunkDim[0] * gridChunkDim[1] * gridChunkDim[2];
+		float nChunks = float(gridChunkDim[0]) * float(gridChunkDim[1]) * float(gridChunkDim[2]);  // suceptible to integer overflow
+		if (numGridChunks <= maxGridChunkCount && nChunks <= maxGridChunkCount)
+		{
+			break;
+		}
+		gridCellSize *= 1.25;  // should roughly cut numCells in half
+	}
+	btAssert(numGridChunks <= maxGridChunkCount);
+	int maxNumBatchesPerPhase = numGridChunks;
 
-    // for each dynamic body, compute grid coords
-    btVector3 invGridCellSize = btVector3(1,1,1)/gridCellSize;
-    // (can be done in parallel)
-    for (int iBody = 0; iBody < bodies.size(); ++iBody)
-    {
-        btIntVec3& coords = bodyGridCoords[iBody];
-        if (bodyDynamicFlags[iBody])
-        {
-            btVector3 v = ( bodyPositions[ iBody ] - bboxMin )*invGridCellSize;
-            coords.m_ints[0] = int(v.x());
-            coords.m_ints[1] = int(v.y());
-            coords.m_ints[2] = int(v.z());
-            btAssert(coords.m_ints[0] >= 0 && coords.m_ints[0] < gridDim[0]);
-            btAssert(coords.m_ints[1] >= 0 && coords.m_ints[1] < gridDim[1]);
-            btAssert(coords.m_ints[2] >= 0 && coords.m_ints[2] < gridDim[2]);
-        }
-        else
-        {
-            coords.m_ints[0] = -1;
-            coords.m_ints[1] = -1;
-            coords.m_ints[2] = -1;
-        }
-    }
+	// for each dynamic body, compute grid coords
+	btVector3 invGridCellSize = btVector3(1, 1, 1) / gridCellSize;
+	// (can be done in parallel)
+	for (int iBody = 0; iBody < bodies.size(); ++iBody)
+	{
+		btIntVec3& coords = bodyGridCoords[iBody];
+		if (bodyDynamicFlags[iBody])
+		{
+			btVector3 v = (bodyPositions[iBody] - bboxMin) * invGridCellSize;
+			coords.m_ints[0] = int(v.x());
+			coords.m_ints[1] = int(v.y());
+			coords.m_ints[2] = int(v.z());
+			btAssert(coords.m_ints[0] >= 0 && coords.m_ints[0] < gridDim[0]);
+			btAssert(coords.m_ints[1] >= 0 && coords.m_ints[1] < gridDim[1]);
+			btAssert(coords.m_ints[2] >= 0 && coords.m_ints[2] < gridDim[2]);
+		}
+		else
+		{
+			coords.m_ints[0] = -1;
+			coords.m_ints[1] = -1;
+			coords.m_ints[2] = -1;
+		}
+	}
 
-    for (int iPhase = 0; iPhase < numPhases; ++iPhase)
-    {
-        int batchBegin = iPhase * maxNumBatchesPerPhase;
-        int batchEnd = batchBegin + maxNumBatchesPerPhase;
-        for ( int iBatch = batchBegin; iBatch < batchEnd; ++iBatch )
-        {
-            btBatchInfo& batch = batches[ iBatch ];
-            batch = btBatchInfo();
-        }
-    }
+	for (int iPhase = 0; iPhase < numPhases; ++iPhase)
+	{
+		int batchBegin = iPhase * maxNumBatchesPerPhase;
+		int batchEnd = batchBegin + maxNumBatchesPerPhase;
+		for (int iBatch = batchBegin; iBatch < batchEnd; ++iBatch)
+		{
+			btBatchInfo& batch = batches[iBatch];
+			batch = btBatchInfo();
+		}
+	}
 
-    {
-        AssignConstraintsToGridBatchesParams params;
-        params.bodyDynamicFlags = bodyDynamicFlags;
-        params.bodyGridCoords = bodyGridCoords;
-        params.numBodies = bodies.size();
-        params.conInfos = conInfos;
-        params.constraintBatchIds = constraintBatchIds;
-        params.gridChunkDim = gridChunkDim;
-        params.maxNumBatchesPerPhase = maxNumBatchesPerPhase;
-        params.numPhases = numPhases;
-        params.phaseMask = phaseMask;
-        bool inParallel = true;
-        if (inParallel)
-        {
-            AssignConstraintsToGridBatchesLoop loop(params);
-            int grainSize = 250;
-            btParallelFor(0, numConstraints, grainSize, loop);
-        }
-        else
-        {
-            assignConstraintsToGridBatches( params, 0, numConstraints );
-        }
-    }
-    for ( int iCon = 0; iCon < numConstraints; ++iCon )
-    {
-        const btBatchedConstraintInfo& con = conInfos[ iCon ];
-        int iBatch = constraintBatchIds[ iCon ];
-        btBatchInfo& batch = batches[iBatch];
-        batch.numConstraints += con.numConstraintRows;
-    }
+	{
+		AssignConstraintsToGridBatchesParams params;
+		params.bodyDynamicFlags = bodyDynamicFlags;
+		params.bodyGridCoords = bodyGridCoords;
+		params.numBodies = bodies.size();
+		params.conInfos = conInfos;
+		params.constraintBatchIds = constraintBatchIds;
+		params.gridChunkDim = gridChunkDim;
+		params.maxNumBatchesPerPhase = maxNumBatchesPerPhase;
+		params.numPhases = numPhases;
+		params.phaseMask = phaseMask;
+		bool inParallel = true;
+		if (inParallel)
+		{
+			AssignConstraintsToGridBatchesLoop loop(params);
+			int grainSize = 250;
+			btParallelFor(0, numConstraints, grainSize, loop);
+		}
+		else
+		{
+			assignConstraintsToGridBatches(params, 0, numConstraints);
+		}
+	}
+	for (int iCon = 0; iCon < numConstraints; ++iCon)
+	{
+		const btBatchedConstraintInfo& con = conInfos[iCon];
+		int iBatch = constraintBatchIds[iCon];
+		btBatchInfo& batch = batches[iBatch];
+		batch.numConstraints += con.numConstraintRows;
+	}
 
-    for (int iPhase = 0; iPhase < numPhases; ++iPhase)
-    {
-        // if phase is legit,
-        if (iPhase == (iPhase&phaseMask))
-        {
-            int iBeginBatch = iPhase * maxNumBatchesPerPhase;
-            int iEndBatch = iBeginBatch + maxNumBatchesPerPhase;
-            mergeSmallBatches( batches, iBeginBatch, iEndBatch, minBatchSize, maxBatchSize );
-        }
-    }
-    // all constraints have been assigned a batchId
-    updateConstraintBatchIdsForMergesMt(constraintBatchIds, numConstraints, batches, maxNumBatchesPerPhase*numPhases);
+	for (int iPhase = 0; iPhase < numPhases; ++iPhase)
+	{
+		// if phase is legit,
+		if (iPhase == (iPhase & phaseMask))
+		{
+			int iBeginBatch = iPhase * maxNumBatchesPerPhase;
+			int iEndBatch = iBeginBatch + maxNumBatchesPerPhase;
+			mergeSmallBatches(batches, iBeginBatch, iEndBatch, minBatchSize, maxBatchSize);
+		}
+	}
+	// all constraints have been assigned a batchId
+	updateConstraintBatchIdsForMergesMt(constraintBatchIds, numConstraints, batches, maxNumBatchesPerPhase * numPhases);
 
-    if (numConstraintRows > numConstraints)
-    {
-        expandConstraintRowsMt(&constraintRowBatchIds[0], &constraintBatchIds[0], &conInfos[0], numConstraints, numConstraintRows);
-    }
-    else
-    {
-        constraintRowBatchIds = constraintBatchIds;
-    }
+	if (numConstraintRows > numConstraints)
+	{
+		expandConstraintRowsMt(&constraintRowBatchIds[0], &constraintBatchIds[0], &conInfos[0], numConstraints, numConstraintRows);
+	}
+	else
+	{
+		constraintRowBatchIds = constraintBatchIds;
+	}
 
-    writeOutBatches(batchedConstraints, constraintRowBatchIds, numConstraintRows, batches, batchWork, maxNumBatchesPerPhase, numPhases);
-    btAssert(batchedConstraints->validate(constraints, bodies));
+	writeOutBatches(batchedConstraints, constraintRowBatchIds, numConstraintRows, batches, batchWork, maxNumBatchesPerPhase, numPhases);
+	btAssert(batchedConstraints->validate(constraints, bodies));
 }
 
-
 static void setupSingleBatch(
-    btBatchedConstraints* bc,
-    int numConstraints
-)
+	btBatchedConstraints* bc,
+	int numConstraints)
 {
-    BT_PROFILE("setupSingleBatch");
-    typedef btBatchedConstraints::Range Range;
+	BT_PROFILE("setupSingleBatch");
+	typedef btBatchedConstraints::Range Range;
 
-    bc->m_constraintIndices.resize( numConstraints );
-    for ( int i = 0; i < numConstraints; ++i )
-    {
-        bc->m_constraintIndices[ i ] = i;
-    }
+	bc->m_constraintIndices.resize(numConstraints);
+	for (int i = 0; i < numConstraints; ++i)
+	{
+		bc->m_constraintIndices[i] = i;
+	}
 
-    bc->m_batches.resizeNoInitialize( 0 );
-    bc->m_phases.resizeNoInitialize( 0 );
-    bc->m_phaseOrder.resizeNoInitialize( 0 );
-    bc->m_phaseGrainSize.resizeNoInitialize( 0 );
+	bc->m_batches.resizeNoInitialize(0);
+	bc->m_phases.resizeNoInitialize(0);
+	bc->m_phaseOrder.resizeNoInitialize(0);
+	bc->m_phaseGrainSize.resizeNoInitialize(0);
 
-    if (numConstraints > 0)
-    {
-        bc->m_batches.push_back( Range( 0, numConstraints ) );
-        bc->m_phases.push_back( Range( 0, 1 ) );
-        bc->m_phaseOrder.push_back(0);
-        bc->m_phaseGrainSize.push_back(1);
-    }
+	if (numConstraints > 0)
+	{
+		bc->m_batches.push_back(Range(0, numConstraints));
+		bc->m_phases.push_back(Range(0, 1));
+		bc->m_phaseOrder.push_back(0);
+		bc->m_phaseGrainSize.push_back(1);
+	}
 }
 
-
 void btBatchedConstraints::setup(
-    btConstraintArray* constraints,
-    const btAlignedObjectArray<btSolverBody>& bodies,
-    BatchingMethod batchingMethod,
-    int minBatchSize,
-    int maxBatchSize,
-    btAlignedObjectArray<char>* scratchMemory
-    )
+	btConstraintArray* constraints,
+	const btAlignedObjectArray<btSolverBody>& bodies,
+	BatchingMethod batchingMethod,
+	int minBatchSize,
+	int maxBatchSize,
+	btAlignedObjectArray<char>* scratchMemory)
 {
-    if (constraints->size() >= minBatchSize*4)
-    {
-        bool use2DGrid = batchingMethod == BATCHING_METHOD_SPATIAL_GRID_2D;
-        setupSpatialGridBatchesMt( this, scratchMemory, constraints, bodies, minBatchSize, maxBatchSize, use2DGrid );
-        if (s_debugDrawBatches)
-        {
-            debugDrawAllBatches( this, constraints, bodies );
-        }
-    }
-    else
-    {
-        setupSingleBatch( this, constraints->size() );
-    }
+	if (constraints->size() >= minBatchSize * 4)
+	{
+		bool use2DGrid = batchingMethod == BATCHING_METHOD_SPATIAL_GRID_2D;
+		setupSpatialGridBatchesMt(this, scratchMemory, constraints, bodies, minBatchSize, maxBatchSize, use2DGrid);
+		if (s_debugDrawBatches)
+		{
+			debugDrawAllBatches(this, constraints, bodies);
+		}
+	}
+	else
+	{
+		setupSingleBatch(this, constraints->size());
+	}
 }
-
-
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btBatchedConstraints.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btBatchedConstraints.h
index 0fd8f31dd45..5d982ca370c 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btBatchedConstraints.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btBatchedConstraints.h
@@ -21,46 +21,42 @@ subject to the following restrictions:
 #include "BulletDynamics/ConstraintSolver/btSolverBody.h"
 #include "BulletDynamics/ConstraintSolver/btSolverConstraint.h"
 
-
 class btIDebugDraw;
 
 struct btBatchedConstraints
 {
-    enum BatchingMethod
-    {
-        BATCHING_METHOD_SPATIAL_GRID_2D,
-        BATCHING_METHOD_SPATIAL_GRID_3D,
-        BATCHING_METHOD_COUNT
-    };
-    struct Range
-    {
-        int begin;
-        int end;
+	enum BatchingMethod
+	{
+		BATCHING_METHOD_SPATIAL_GRID_2D,
+		BATCHING_METHOD_SPATIAL_GRID_3D,
+		BATCHING_METHOD_COUNT
+	};
+	struct Range
+	{
+		int begin;
+		int end;
 
-        Range() : begin( 0 ), end( 0 ) {}
-        Range( int _beg, int _end ) : begin( _beg ), end( _end ) {}
-    };
+		Range() : begin(0), end(0) {}
+		Range(int _beg, int _end) : begin(_beg), end(_end) {}
+	};
 
-    btAlignedObjectArray<int> m_constraintIndices;
-    btAlignedObjectArray<Range> m_batches;  // each batch is a range of indices in the m_constraintIndices array
-    btAlignedObjectArray<Range> m_phases;  // each phase is range of indices in the m_batches array
-    btAlignedObjectArray<char> m_phaseGrainSize;  // max grain size for each phase
-    btAlignedObjectArray<int> m_phaseOrder;  // phases can be done in any order, so we can randomize the order here
-    btIDebugDraw* m_debugDrawer;
+	btAlignedObjectArray<int> m_constraintIndices;
+	btAlignedObjectArray<Range> m_batches;        // each batch is a range of indices in the m_constraintIndices array
+	btAlignedObjectArray<Range> m_phases;         // each phase is range of indices in the m_batches array
+	btAlignedObjectArray<char> m_phaseGrainSize;  // max grain size for each phase
+	btAlignedObjectArray<int> m_phaseOrder;       // phases can be done in any order, so we can randomize the order here
+	btIDebugDraw* m_debugDrawer;
 
-    static bool s_debugDrawBatches;
+	static bool s_debugDrawBatches;
 
-    btBatchedConstraints() {m_debugDrawer=NULL;}
-    void setup( btConstraintArray* constraints,
-        const btAlignedObjectArray<btSolverBody>& bodies,
-        BatchingMethod batchingMethod,
-        int minBatchSize,
-        int maxBatchSize,
-        btAlignedObjectArray<char>* scratchMemory
-    );
-    bool validate( btConstraintArray* constraints, const btAlignedObjectArray<btSolverBody>& bodies ) const;
+	btBatchedConstraints() { m_debugDrawer = NULL; }
+	void setup(btConstraintArray* constraints,
+			   const btAlignedObjectArray<btSolverBody>& bodies,
+			   BatchingMethod batchingMethod,
+			   int minBatchSize,
+			   int maxBatchSize,
+			   btAlignedObjectArray<char>* scratchMemory);
+	bool validate(btConstraintArray* constraints, const btAlignedObjectArray<btSolverBody>& bodies) const;
 };
 
-
-#endif // BT_BATCHED_CONSTRAINTS_H
-
+#endif  // BT_BATCHED_CONSTRAINTS_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btConeTwistConstraint.cpp b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btConeTwistConstraint.cpp
index 0572256f74b..10678b2a616 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btConeTwistConstraint.cpp
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btConeTwistConstraint.cpp
@@ -15,49 +15,37 @@ subject to the following restrictions:
 Written by: Marcus Hennix
 */
 
-
 #include "btConeTwistConstraint.h"
 #include "BulletDynamics/Dynamics/btRigidBody.h"
 #include "LinearMath/btTransformUtil.h"
 #include "LinearMath/btMinMax.h"
 #include <new>
 
-
-
 //#define CONETWIST_USE_OBSOLETE_SOLVER true
 #define CONETWIST_USE_OBSOLETE_SOLVER false
 #define CONETWIST_DEF_FIX_THRESH btScalar(.05f)
 
-
 SIMD_FORCE_INLINE btScalar computeAngularImpulseDenominator(const btVector3& axis, const btMatrix3x3& invInertiaWorld)
 {
 	btVector3 vec = axis * invInertiaWorld;
 	return axis.dot(vec);
 }
 
-
-
-
-btConeTwistConstraint::btConeTwistConstraint(btRigidBody& rbA,btRigidBody& rbB, 
-											 const btTransform& rbAFrame,const btTransform& rbBFrame)
-											 :btTypedConstraint(CONETWIST_CONSTRAINT_TYPE, rbA,rbB),m_rbAFrame(rbAFrame),m_rbBFrame(rbBFrame),
-											 m_angularOnly(false),
-											 m_useSolveConstraintObsolete(CONETWIST_USE_OBSOLETE_SOLVER)
+btConeTwistConstraint::btConeTwistConstraint(btRigidBody& rbA, btRigidBody& rbB,
+											 const btTransform& rbAFrame, const btTransform& rbBFrame)
+	: btTypedConstraint(CONETWIST_CONSTRAINT_TYPE, rbA, rbB), m_rbAFrame(rbAFrame), m_rbBFrame(rbBFrame), m_angularOnly(false), m_useSolveConstraintObsolete(CONETWIST_USE_OBSOLETE_SOLVER)
 {
 	init();
 }
 
-btConeTwistConstraint::btConeTwistConstraint(btRigidBody& rbA,const btTransform& rbAFrame)
-											:btTypedConstraint(CONETWIST_CONSTRAINT_TYPE,rbA),m_rbAFrame(rbAFrame),
-											 m_angularOnly(false),
-											 m_useSolveConstraintObsolete(CONETWIST_USE_OBSOLETE_SOLVER)
+btConeTwistConstraint::btConeTwistConstraint(btRigidBody& rbA, const btTransform& rbAFrame)
+	: btTypedConstraint(CONETWIST_CONSTRAINT_TYPE, rbA), m_rbAFrame(rbAFrame), m_angularOnly(false), m_useSolveConstraintObsolete(CONETWIST_USE_OBSOLETE_SOLVER)
 {
 	m_rbBFrame = m_rbAFrame;
 	m_rbBFrame.setOrigin(btVector3(0., 0., 0.));
-	init();	
+	init();
 }
 
-
 void btConeTwistConstraint::init()
 {
 	m_angularOnly = false;
@@ -75,30 +63,29 @@ void btConeTwistConstraint::init()
 	m_angCFM = btScalar(0.f);
 }
 
-
-void btConeTwistConstraint::getInfo1 (btConstraintInfo1* info)
+void btConeTwistConstraint::getInfo1(btConstraintInfo1* info)
 {
 	if (m_useSolveConstraintObsolete)
 	{
 		info->m_numConstraintRows = 0;
 		info->nub = 0;
-	} 
+	}
 	else
 	{
 		info->m_numConstraintRows = 3;
 		info->nub = 3;
-		calcAngleInfo2(m_rbA.getCenterOfMassTransform(),m_rbB.getCenterOfMassTransform(),m_rbA.getInvInertiaTensorWorld(),m_rbB.getInvInertiaTensorWorld());
-		if(m_solveSwingLimit)
+		calcAngleInfo2(m_rbA.getCenterOfMassTransform(), m_rbB.getCenterOfMassTransform(), m_rbA.getInvInertiaTensorWorld(), m_rbB.getInvInertiaTensorWorld());
+		if (m_solveSwingLimit)
 		{
 			info->m_numConstraintRows++;
 			info->nub--;
-			if((m_swingSpan1 < m_fixThresh) && (m_swingSpan2 < m_fixThresh))
+			if ((m_swingSpan1 < m_fixThresh) && (m_swingSpan2 < m_fixThresh))
 			{
 				info->m_numConstraintRows++;
 				info->nub--;
 			}
 		}
-		if(m_solveTwistLimit)
+		if (m_solveTwistLimit)
 		{
 			info->m_numConstraintRows++;
 			info->nub--;
@@ -106,90 +93,88 @@ void btConeTwistConstraint::getInfo1 (btConstraintInfo1* info)
 	}
 }
 
-void btConeTwistConstraint::getInfo1NonVirtual (btConstraintInfo1* info)
+void btConeTwistConstraint::getInfo1NonVirtual(btConstraintInfo1* info)
 {
 	//always reserve 6 rows: object transform is not available on SPU
 	info->m_numConstraintRows = 6;
 	info->nub = 0;
-		
-}
-	
-
-void btConeTwistConstraint::getInfo2 (btConstraintInfo2* info)
-{
-	getInfo2NonVirtual(info,m_rbA.getCenterOfMassTransform(),m_rbB.getCenterOfMassTransform(),m_rbA.getInvInertiaTensorWorld(),m_rbB.getInvInertiaTensorWorld());
 }
 
-void btConeTwistConstraint::getInfo2NonVirtual (btConstraintInfo2* info,const btTransform& transA,const btTransform& transB,const btMatrix3x3& invInertiaWorldA,const btMatrix3x3& invInertiaWorldB)
+void btConeTwistConstraint::getInfo2(btConstraintInfo2* info)
 {
-	calcAngleInfo2(transA,transB,invInertiaWorldA,invInertiaWorldB);
-	
+	getInfo2NonVirtual(info, m_rbA.getCenterOfMassTransform(), m_rbB.getCenterOfMassTransform(), m_rbA.getInvInertiaTensorWorld(), m_rbB.getInvInertiaTensorWorld());
+}
+
+void btConeTwistConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTransform& transA, const btTransform& transB, const btMatrix3x3& invInertiaWorldA, const btMatrix3x3& invInertiaWorldB)
+{
+	calcAngleInfo2(transA, transB, invInertiaWorldA, invInertiaWorldB);
+
 	btAssert(!m_useSolveConstraintObsolete);
-    // set jacobian
-    info->m_J1linearAxis[0] = 1;
-    info->m_J1linearAxis[info->rowskip+1] = 1;
-    info->m_J1linearAxis[2*info->rowskip+2] = 1;
+	// set jacobian
+	info->m_J1linearAxis[0] = 1;
+	info->m_J1linearAxis[info->rowskip + 1] = 1;
+	info->m_J1linearAxis[2 * info->rowskip + 2] = 1;
 	btVector3 a1 = transA.getBasis() * m_rbAFrame.getOrigin();
 	{
 		btVector3* angular0 = (btVector3*)(info->m_J1angularAxis);
-		btVector3* angular1 = (btVector3*)(info->m_J1angularAxis+info->rowskip);
-		btVector3* angular2 = (btVector3*)(info->m_J1angularAxis+2*info->rowskip);
+		btVector3* angular1 = (btVector3*)(info->m_J1angularAxis + info->rowskip);
+		btVector3* angular2 = (btVector3*)(info->m_J1angularAxis + 2 * info->rowskip);
 		btVector3 a1neg = -a1;
-		a1neg.getSkewSymmetricMatrix(angular0,angular1,angular2);
+		a1neg.getSkewSymmetricMatrix(angular0, angular1, angular2);
 	}
-    info->m_J2linearAxis[0] = -1;
-    info->m_J2linearAxis[info->rowskip+1] = -1;
-    info->m_J2linearAxis[2*info->rowskip+2] = -1;
+	info->m_J2linearAxis[0] = -1;
+	info->m_J2linearAxis[info->rowskip + 1] = -1;
+	info->m_J2linearAxis[2 * info->rowskip + 2] = -1;
 	btVector3 a2 = transB.getBasis() * m_rbBFrame.getOrigin();
 	{
 		btVector3* angular0 = (btVector3*)(info->m_J2angularAxis);
-		btVector3* angular1 = (btVector3*)(info->m_J2angularAxis+info->rowskip);
-		btVector3* angular2 = (btVector3*)(info->m_J2angularAxis+2*info->rowskip);
-		a2.getSkewSymmetricMatrix(angular0,angular1,angular2);
+		btVector3* angular1 = (btVector3*)(info->m_J2angularAxis + info->rowskip);
+		btVector3* angular2 = (btVector3*)(info->m_J2angularAxis + 2 * info->rowskip);
+		a2.getSkewSymmetricMatrix(angular0, angular1, angular2);
 	}
-    // set right hand side
+	// set right hand side
 	btScalar linERP = (m_flags & BT_CONETWIST_FLAGS_LIN_ERP) ? m_linERP : info->erp;
-    btScalar k = info->fps * linERP;
-    int j;
-	for (j=0; j<3; j++)
-    {
-        info->m_constraintError[j*info->rowskip] = k * (a2[j] + transB.getOrigin()[j] - a1[j] - transA.getOrigin()[j]);
-		info->m_lowerLimit[j*info->rowskip] = -SIMD_INFINITY;
-		info->m_upperLimit[j*info->rowskip] = SIMD_INFINITY;
-		if(m_flags & BT_CONETWIST_FLAGS_LIN_CFM)
+	btScalar k = info->fps * linERP;
+	int j;
+	for (j = 0; j < 3; j++)
+	{
+		info->m_constraintError[j * info->rowskip] = k * (a2[j] + transB.getOrigin()[j] - a1[j] - transA.getOrigin()[j]);
+		info->m_lowerLimit[j * info->rowskip] = -SIMD_INFINITY;
+		info->m_upperLimit[j * info->rowskip] = SIMD_INFINITY;
+		if (m_flags & BT_CONETWIST_FLAGS_LIN_CFM)
 		{
-			info->cfm[j*info->rowskip] = m_linCFM;
+			info->cfm[j * info->rowskip] = m_linCFM;
 		}
-    }
+	}
 	int row = 3;
-    int srow = row * info->rowskip;
+	int srow = row * info->rowskip;
 	btVector3 ax1;
 	// angular limits
-	if(m_solveSwingLimit)
+	if (m_solveSwingLimit)
 	{
-		btScalar *J1 = info->m_J1angularAxis;
-		btScalar *J2 = info->m_J2angularAxis;
-		if((m_swingSpan1 < m_fixThresh) && (m_swingSpan2 < m_fixThresh))
+		btScalar* J1 = info->m_J1angularAxis;
+		btScalar* J2 = info->m_J2angularAxis;
+		if ((m_swingSpan1 < m_fixThresh) && (m_swingSpan2 < m_fixThresh))
 		{
-			btTransform trA = transA*m_rbAFrame;
+			btTransform trA = transA * m_rbAFrame;
 			btVector3 p = trA.getBasis().getColumn(1);
 			btVector3 q = trA.getBasis().getColumn(2);
 			int srow1 = srow + info->rowskip;
-			J1[srow+0] = p[0];
-			J1[srow+1] = p[1];
-			J1[srow+2] = p[2];
-			J1[srow1+0] = q[0];
-			J1[srow1+1] = q[1];
-			J1[srow1+2] = q[2];
-			J2[srow+0] = -p[0];
-			J2[srow+1] = -p[1];
-			J2[srow+2] = -p[2];
-			J2[srow1+0] = -q[0];
-			J2[srow1+1] = -q[1];
-			J2[srow1+2] = -q[2];
+			J1[srow + 0] = p[0];
+			J1[srow + 1] = p[1];
+			J1[srow + 2] = p[2];
+			J1[srow1 + 0] = q[0];
+			J1[srow1 + 1] = q[1];
+			J1[srow1 + 2] = q[2];
+			J2[srow + 0] = -p[0];
+			J2[srow + 1] = -p[1];
+			J2[srow + 2] = -p[2];
+			J2[srow1 + 0] = -q[0];
+			J2[srow1 + 1] = -q[1];
+			J2[srow1 + 2] = -q[2];
 			btScalar fact = info->fps * m_relaxationFactor;
-			info->m_constraintError[srow] =   fact * m_swingAxis.dot(p);
-			info->m_constraintError[srow1] =  fact * m_swingAxis.dot(q);
+			info->m_constraintError[srow] = fact * m_swingAxis.dot(p);
+			info->m_constraintError[srow1] = fact * m_swingAxis.dot(q);
 			info->m_lowerLimit[srow] = -SIMD_INFINITY;
 			info->m_upperLimit[srow] = SIMD_INFINITY;
 			info->m_lowerLimit[srow1] = -SIMD_INFINITY;
@@ -199,16 +184,16 @@ void btConeTwistConstraint::getInfo2NonVirtual (btConstraintInfo2* info,const bt
 		else
 		{
 			ax1 = m_swingAxis * m_relaxationFactor * m_relaxationFactor;
-			J1[srow+0] = ax1[0];
-			J1[srow+1] = ax1[1];
-			J1[srow+2] = ax1[2];
-			J2[srow+0] = -ax1[0];
-			J2[srow+1] = -ax1[1];
-			J2[srow+2] = -ax1[2];
+			J1[srow + 0] = ax1[0];
+			J1[srow + 1] = ax1[1];
+			J1[srow + 2] = ax1[2];
+			J2[srow + 0] = -ax1[0];
+			J2[srow + 1] = -ax1[1];
+			J2[srow + 2] = -ax1[2];
 			btScalar k = info->fps * m_biasFactor;
 
 			info->m_constraintError[srow] = k * m_swingCorrection;
-			if(m_flags & BT_CONETWIST_FLAGS_ANG_CFM)
+			if (m_flags & BT_CONETWIST_FLAGS_ANG_CFM)
 			{
 				info->cfm[srow] = m_angCFM;
 			}
@@ -218,36 +203,35 @@ void btConeTwistConstraint::getInfo2NonVirtual (btConstraintInfo2* info,const bt
 			srow += info->rowskip;
 		}
 	}
-	if(m_solveTwistLimit)
+	if (m_solveTwistLimit)
 	{
 		ax1 = m_twistAxis * m_relaxationFactor * m_relaxationFactor;
-		btScalar *J1 = info->m_J1angularAxis;
-		btScalar *J2 = info->m_J2angularAxis;
-		J1[srow+0] = ax1[0];
-		J1[srow+1] = ax1[1];
-		J1[srow+2] = ax1[2];
-		J2[srow+0] = -ax1[0];
-		J2[srow+1] = -ax1[1];
-		J2[srow+2] = -ax1[2];
+		btScalar* J1 = info->m_J1angularAxis;
+		btScalar* J2 = info->m_J2angularAxis;
+		J1[srow + 0] = ax1[0];
+		J1[srow + 1] = ax1[1];
+		J1[srow + 2] = ax1[2];
+		J2[srow + 0] = -ax1[0];
+		J2[srow + 1] = -ax1[1];
+		J2[srow + 2] = -ax1[2];
 		btScalar k = info->fps * m_biasFactor;
 		info->m_constraintError[srow] = k * m_twistCorrection;
-		if(m_flags & BT_CONETWIST_FLAGS_ANG_CFM)
+		if (m_flags & BT_CONETWIST_FLAGS_ANG_CFM)
 		{
 			info->cfm[srow] = m_angCFM;
 		}
-		if(m_twistSpan > 0.0f)
+		if (m_twistSpan > 0.0f)
 		{
-
-			if(m_twistCorrection > 0.0f)
+			if (m_twistCorrection > 0.0f)
 			{
 				info->m_lowerLimit[srow] = 0;
 				info->m_upperLimit[srow] = SIMD_INFINITY;
-			} 
+			}
 			else
 			{
 				info->m_lowerLimit[srow] = -SIMD_INFINITY;
 				info->m_upperLimit[srow] = 0;
-			} 
+			}
 		}
 		else
 		{
@@ -257,22 +241,20 @@ void btConeTwistConstraint::getInfo2NonVirtual (btConstraintInfo2* info,const bt
 		srow += info->rowskip;
 	}
 }
-	
 
-
-void	btConeTwistConstraint::buildJacobian()
+void btConeTwistConstraint::buildJacobian()
 {
 	if (m_useSolveConstraintObsolete)
 	{
 		m_appliedImpulse = btScalar(0.);
 		m_accTwistLimitImpulse = btScalar(0.);
 		m_accSwingLimitImpulse = btScalar(0.);
-		m_accMotorImpulse = btVector3(0.,0.,0.);
+		m_accMotorImpulse = btVector3(0., 0., 0.);
 
 		if (!m_angularOnly)
 		{
-			btVector3 pivotAInW = m_rbA.getCenterOfMassTransform()*m_rbAFrame.getOrigin();
-			btVector3 pivotBInW = m_rbB.getCenterOfMassTransform()*m_rbBFrame.getOrigin();
+			btVector3 pivotAInW = m_rbA.getCenterOfMassTransform() * m_rbAFrame.getOrigin();
+			btVector3 pivotBInW = m_rbB.getCenterOfMassTransform() * m_rbBFrame.getOrigin();
 			btVector3 relPos = pivotBInW - pivotAInW;
 
 			btVector3 normal[3];
@@ -282,71 +264,68 @@ void	btConeTwistConstraint::buildJacobian()
 			}
 			else
 			{
-				normal[0].setValue(btScalar(1.0),0,0);
+				normal[0].setValue(btScalar(1.0), 0, 0);
 			}
 
 			btPlaneSpace1(normal[0], normal[1], normal[2]);
 
-			for (int i=0;i<3;i++)
+			for (int i = 0; i < 3; i++)
 			{
 				new (&m_jac[i]) btJacobianEntry(
-				m_rbA.getCenterOfMassTransform().getBasis().transpose(),
-				m_rbB.getCenterOfMassTransform().getBasis().transpose(),
-				pivotAInW - m_rbA.getCenterOfMassPosition(),
-				pivotBInW - m_rbB.getCenterOfMassPosition(),
-				normal[i],
-				m_rbA.getInvInertiaDiagLocal(),
-				m_rbA.getInvMass(),
-				m_rbB.getInvInertiaDiagLocal(),
-				m_rbB.getInvMass());
+					m_rbA.getCenterOfMassTransform().getBasis().transpose(),
+					m_rbB.getCenterOfMassTransform().getBasis().transpose(),
+					pivotAInW - m_rbA.getCenterOfMassPosition(),
+					pivotBInW - m_rbB.getCenterOfMassPosition(),
+					normal[i],
+					m_rbA.getInvInertiaDiagLocal(),
+					m_rbA.getInvMass(),
+					m_rbB.getInvInertiaDiagLocal(),
+					m_rbB.getInvMass());
 			}
 		}
 
-		calcAngleInfo2(m_rbA.getCenterOfMassTransform(),m_rbB.getCenterOfMassTransform(),m_rbA.getInvInertiaTensorWorld(),m_rbB.getInvInertiaTensorWorld());
+		calcAngleInfo2(m_rbA.getCenterOfMassTransform(), m_rbB.getCenterOfMassTransform(), m_rbA.getInvInertiaTensorWorld(), m_rbB.getInvInertiaTensorWorld());
 	}
 }
 
-
-
-void	btConeTwistConstraint::solveConstraintObsolete(btSolverBody& bodyA,btSolverBody& bodyB,btScalar	timeStep)
+void btConeTwistConstraint::solveConstraintObsolete(btSolverBody& bodyA, btSolverBody& bodyB, btScalar timeStep)
 {
-	#ifndef __SPU__
+#ifndef __SPU__
 	if (m_useSolveConstraintObsolete)
 	{
-		btVector3 pivotAInW = m_rbA.getCenterOfMassTransform()*m_rbAFrame.getOrigin();
-		btVector3 pivotBInW = m_rbB.getCenterOfMassTransform()*m_rbBFrame.getOrigin();
+		btVector3 pivotAInW = m_rbA.getCenterOfMassTransform() * m_rbAFrame.getOrigin();
+		btVector3 pivotBInW = m_rbB.getCenterOfMassTransform() * m_rbBFrame.getOrigin();
 
 		btScalar tau = btScalar(0.3);
 
 		//linear part
 		if (!m_angularOnly)
 		{
-			btVector3 rel_pos1 = pivotAInW - m_rbA.getCenterOfMassPosition(); 
+			btVector3 rel_pos1 = pivotAInW - m_rbA.getCenterOfMassPosition();
 			btVector3 rel_pos2 = pivotBInW - m_rbB.getCenterOfMassPosition();
 
 			btVector3 vel1;
-			bodyA.internalGetVelocityInLocalPointObsolete(rel_pos1,vel1);
+			bodyA.internalGetVelocityInLocalPointObsolete(rel_pos1, vel1);
 			btVector3 vel2;
-			bodyB.internalGetVelocityInLocalPointObsolete(rel_pos2,vel2);
+			bodyB.internalGetVelocityInLocalPointObsolete(rel_pos2, vel2);
 			btVector3 vel = vel1 - vel2;
 
-			for (int i=0;i<3;i++)
-			{		
+			for (int i = 0; i < 3; i++)
+			{
 				const btVector3& normal = m_jac[i].m_linearJointAxis;
 				btScalar jacDiagABInv = btScalar(1.) / m_jac[i].getDiagonal();
 
 				btScalar rel_vel;
 				rel_vel = normal.dot(vel);
 				//positional error (zeroth order error)
-				btScalar depth = -(pivotAInW - pivotBInW).dot(normal); //this is the error projected on the normal
-				btScalar impulse = depth*tau/timeStep  * jacDiagABInv -  rel_vel * jacDiagABInv;
+				btScalar depth = -(pivotAInW - pivotBInW).dot(normal);  //this is the error projected on the normal
+				btScalar impulse = depth * tau / timeStep * jacDiagABInv - rel_vel * jacDiagABInv;
 				m_appliedImpulse += impulse;
-				
+
 				btVector3 ftorqueAxis1 = rel_pos1.cross(normal);
 				btVector3 ftorqueAxis2 = rel_pos2.cross(normal);
-				bodyA.internalApplyImpulse(normal*m_rbA.getInvMass(), m_rbA.getInvInertiaTensorWorld()*ftorqueAxis1,impulse);
-				bodyB.internalApplyImpulse(normal*m_rbB.getInvMass(), m_rbB.getInvInertiaTensorWorld()*ftorqueAxis2,-impulse);
-		
+				bodyA.internalApplyImpulse(normal * m_rbA.getInvMass(), m_rbA.getInvInertiaTensorWorld() * ftorqueAxis1, impulse);
+				bodyB.internalApplyImpulse(normal * m_rbB.getInvMass(), m_rbB.getInvInertiaTensorWorld() * ftorqueAxis2, -impulse);
 			}
 		}
 
@@ -356,13 +335,17 @@ void	btConeTwistConstraint::solveConstraintObsolete(btSolverBody& bodyA,btSolver
 			// compute current and predicted transforms
 			btTransform trACur = m_rbA.getCenterOfMassTransform();
 			btTransform trBCur = m_rbB.getCenterOfMassTransform();
-			btVector3 omegaA; bodyA.internalGetAngularVelocity(omegaA);
-			btVector3 omegaB; bodyB.internalGetAngularVelocity(omegaB);
-			btTransform trAPred; trAPred.setIdentity(); 
-			btVector3 zerovec(0,0,0);
+			btVector3 omegaA;
+			bodyA.internalGetAngularVelocity(omegaA);
+			btVector3 omegaB;
+			bodyB.internalGetAngularVelocity(omegaB);
+			btTransform trAPred;
+			trAPred.setIdentity();
+			btVector3 zerovec(0, 0, 0);
 			btTransformUtil::integrateTransform(
 				trACur, zerovec, omegaA, timeStep, trAPred);
-			btTransform trBPred; trBPred.setIdentity(); 
+			btTransform trBPred;
+			trBPred.setIdentity();
 			btTransformUtil::integrateTransform(
 				trBCur, zerovec, omegaB, timeStep, trBPred);
 
@@ -374,7 +357,7 @@ void	btConeTwistConstraint::solveConstraintObsolete(btSolverBody& bodyA,btSolver
 
 			// compute desired omegas in world
 			btVector3 omegaADes, omegaBDes;
-			
+
 			btTransformUtil::calculateVelocity(trACur, trADes, timeStep, zerovec, omegaADes);
 			btTransformUtil::calculateVelocity(trBCur, trBDes, timeStep, zerovec, omegaBDes);
 
@@ -415,10 +398,10 @@ void	btConeTwistConstraint::solveConstraintObsolete(btSolverBody& bodyA,btSolver
 				{
 					btScalar fMaxImpulse = m_maxMotorImpulse;
 					if (m_bNormalizedMotorStrength)
-						fMaxImpulse = fMaxImpulse/kAxisAInv;
+						fMaxImpulse = fMaxImpulse / kAxisAInv;
 
 					btVector3 newUnclampedAccImpulse = m_accMotorImpulse + impulse;
-					btScalar  newUnclampedMag = newUnclampedAccImpulse.length();
+					btScalar newUnclampedMag = newUnclampedAccImpulse.length();
 					if (newUnclampedMag > fMaxImpulse)
 					{
 						newUnclampedAccImpulse.normalize();
@@ -428,31 +411,32 @@ void	btConeTwistConstraint::solveConstraintObsolete(btSolverBody& bodyA,btSolver
 					m_accMotorImpulse += impulse;
 				}
 
-				btScalar  impulseMag  = impulse.length();
-				btVector3 impulseAxis =  impulse / impulseMag;
-
-				bodyA.internalApplyImpulse(btVector3(0,0,0), m_rbA.getInvInertiaTensorWorld()*impulseAxis, impulseMag);
-				bodyB.internalApplyImpulse(btVector3(0,0,0), m_rbB.getInvInertiaTensorWorld()*impulseAxis, -impulseMag);
+				btScalar impulseMag = impulse.length();
+				btVector3 impulseAxis = impulse / impulseMag;
 
+				bodyA.internalApplyImpulse(btVector3(0, 0, 0), m_rbA.getInvInertiaTensorWorld() * impulseAxis, impulseMag);
+				bodyB.internalApplyImpulse(btVector3(0, 0, 0), m_rbB.getInvInertiaTensorWorld() * impulseAxis, -impulseMag);
 			}
 		}
-		else if (m_damping > SIMD_EPSILON) // no motor: do a little damping
+		else if (m_damping > SIMD_EPSILON)  // no motor: do a little damping
 		{
-			btVector3 angVelA; bodyA.internalGetAngularVelocity(angVelA);
-			btVector3 angVelB; bodyB.internalGetAngularVelocity(angVelB);
+			btVector3 angVelA;
+			bodyA.internalGetAngularVelocity(angVelA);
+			btVector3 angVelB;
+			bodyB.internalGetAngularVelocity(angVelB);
 			btVector3 relVel = angVelB - angVelA;
 			if (relVel.length2() > SIMD_EPSILON)
 			{
 				btVector3 relVelAxis = relVel.normalized();
-				btScalar m_kDamping =  btScalar(1.) /
-					(getRigidBodyA().computeAngularImpulseDenominator(relVelAxis) +
-					 getRigidBodyB().computeAngularImpulseDenominator(relVelAxis));
+				btScalar m_kDamping = btScalar(1.) /
+									  (getRigidBodyA().computeAngularImpulseDenominator(relVelAxis) +
+									   getRigidBodyB().computeAngularImpulseDenominator(relVelAxis));
 				btVector3 impulse = m_damping * m_kDamping * relVel;
 
-				btScalar  impulseMag  = impulse.length();
+				btScalar impulseMag = impulse.length();
 				btVector3 impulseAxis = impulse / impulseMag;
-				bodyA.internalApplyImpulse(btVector3(0,0,0), m_rbA.getInvInertiaTensorWorld()*impulseAxis, impulseMag);
-				bodyB.internalApplyImpulse(btVector3(0,0,0), m_rbB.getInvInertiaTensorWorld()*impulseAxis, -impulseMag);
+				bodyA.internalApplyImpulse(btVector3(0, 0, 0), m_rbA.getInvInertiaTensorWorld() * impulseAxis, impulseMag);
+				bodyB.internalApplyImpulse(btVector3(0, 0, 0), m_rbB.getInvInertiaTensorWorld() * impulseAxis, -impulseMag);
 			}
 		}
 
@@ -467,7 +451,7 @@ void	btConeTwistConstraint::solveConstraintObsolete(btSolverBody& bodyA,btSolver
 			// solve swing limit
 			if (m_solveSwingLimit)
 			{
-				btScalar amplitude = m_swingLimitRatio * m_swingCorrection*m_biasFactor/timeStep;
+				btScalar amplitude = m_swingLimitRatio * m_swingCorrection * m_biasFactor / timeStep;
 				btScalar relSwingVel = (angVelB - angVelA).dot(m_swingAxis);
 				if (relSwingVel > 0)
 					amplitude += m_swingLimitRatio * relSwingVel * m_relaxationFactor;
@@ -475,7 +459,7 @@ void	btConeTwistConstraint::solveConstraintObsolete(btSolverBody& bodyA,btSolver
 
 				// Clamp the accumulated impulse
 				btScalar temp = m_accSwingLimitImpulse;
-				m_accSwingLimitImpulse = btMax(m_accSwingLimitImpulse + impulseMag, btScalar(0.0) );
+				m_accSwingLimitImpulse = btMax(m_accSwingLimitImpulse + impulseMag, btScalar(0.0));
 				impulseMag = m_accSwingLimitImpulse - temp;
 
 				btVector3 impulse = m_swingAxis * impulseMag;
@@ -491,47 +475,41 @@ void	btConeTwistConstraint::solveConstraintObsolete(btSolverBody& bodyA,btSolver
 				impulseMag = impulse.length();
 				btVector3 noTwistSwingAxis = impulse / impulseMag;
 
-				bodyA.internalApplyImpulse(btVector3(0,0,0), m_rbA.getInvInertiaTensorWorld()*noTwistSwingAxis, impulseMag);
-				bodyB.internalApplyImpulse(btVector3(0,0,0), m_rbB.getInvInertiaTensorWorld()*noTwistSwingAxis, -impulseMag);
+				bodyA.internalApplyImpulse(btVector3(0, 0, 0), m_rbA.getInvInertiaTensorWorld() * noTwistSwingAxis, impulseMag);
+				bodyB.internalApplyImpulse(btVector3(0, 0, 0), m_rbB.getInvInertiaTensorWorld() * noTwistSwingAxis, -impulseMag);
 			}
 
-
 			// solve twist limit
 			if (m_solveTwistLimit)
 			{
-				btScalar amplitude = m_twistLimitRatio * m_twistCorrection*m_biasFactor/timeStep;
-				btScalar relTwistVel = (angVelB - angVelA).dot( m_twistAxis );
-				if (relTwistVel > 0) // only damp when moving towards limit (m_twistAxis flipping is important)
+				btScalar amplitude = m_twistLimitRatio * m_twistCorrection * m_biasFactor / timeStep;
+				btScalar relTwistVel = (angVelB - angVelA).dot(m_twistAxis);
+				if (relTwistVel > 0)  // only damp when moving towards limit (m_twistAxis flipping is important)
 					amplitude += m_twistLimitRatio * relTwistVel * m_relaxationFactor;
 				btScalar impulseMag = amplitude * m_kTwist;
 
 				// Clamp the accumulated impulse
 				btScalar temp = m_accTwistLimitImpulse;
-				m_accTwistLimitImpulse = btMax(m_accTwistLimitImpulse + impulseMag, btScalar(0.0) );
+				m_accTwistLimitImpulse = btMax(m_accTwistLimitImpulse + impulseMag, btScalar(0.0));
 				impulseMag = m_accTwistLimitImpulse - temp;
 
-		//		btVector3 impulse = m_twistAxis * impulseMag;
+				//		btVector3 impulse = m_twistAxis * impulseMag;
 
-				bodyA.internalApplyImpulse(btVector3(0,0,0), m_rbA.getInvInertiaTensorWorld()*m_twistAxis,impulseMag);
-				bodyB.internalApplyImpulse(btVector3(0,0,0), m_rbB.getInvInertiaTensorWorld()*m_twistAxis,-impulseMag);
-			}		
+				bodyA.internalApplyImpulse(btVector3(0, 0, 0), m_rbA.getInvInertiaTensorWorld() * m_twistAxis, impulseMag);
+				bodyB.internalApplyImpulse(btVector3(0, 0, 0), m_rbB.getInvInertiaTensorWorld() * m_twistAxis, -impulseMag);
+			}
 		}
 	}
 #else
-btAssert(0);
-#endif //__SPU__
+	btAssert(0);
+#endif  //__SPU__
 }
 
-
-
-
-void	btConeTwistConstraint::updateRHS(btScalar	timeStep)
+void btConeTwistConstraint::updateRHS(btScalar timeStep)
 {
 	(void)timeStep;
-
 }
 
-
 #ifndef __SPU__
 void btConeTwistConstraint::calcAngleInfo()
 {
@@ -540,15 +518,15 @@ void btConeTwistConstraint::calcAngleInfo()
 	m_solveTwistLimit = false;
 	m_solveSwingLimit = false;
 
-	btVector3 b1Axis1(0,0,0),b1Axis2(0,0,0),b1Axis3(0,0,0);
-	btVector3 b2Axis1(0,0,0),b2Axis2(0,0,0);
+	btVector3 b1Axis1(0, 0, 0), b1Axis2(0, 0, 0), b1Axis3(0, 0, 0);
+	btVector3 b2Axis1(0, 0, 0), b2Axis2(0, 0, 0);
 
 	b1Axis1 = getRigidBodyA().getCenterOfMassTransform().getBasis() * this->m_rbAFrame.getBasis().getColumn(0);
 	b2Axis1 = getRigidBodyB().getCenterOfMassTransform().getBasis() * this->m_rbBFrame.getBasis().getColumn(0);
 
-	btScalar swing1=btScalar(0.),swing2 = btScalar(0.);
+	btScalar swing1 = btScalar(0.), swing2 = btScalar(0.);
 
-	btScalar swx=btScalar(0.),swy = btScalar(0.);
+	btScalar swx = btScalar(0.), swy = btScalar(0.);
 	btScalar thresh = btScalar(10.);
 	btScalar fact;
 
@@ -558,33 +536,33 @@ void btConeTwistConstraint::calcAngleInfo()
 		b1Axis2 = getRigidBodyA().getCenterOfMassTransform().getBasis() * this->m_rbAFrame.getBasis().getColumn(1);
 		swx = b2Axis1.dot(b1Axis1);
 		swy = b2Axis1.dot(b1Axis2);
-		swing1  = btAtan2Fast(swy, swx);
-		fact = (swy*swy + swx*swx) * thresh * thresh;
+		swing1 = btAtan2Fast(swy, swx);
+		fact = (swy * swy + swx * swx) * thresh * thresh;
 		fact = fact / (fact + btScalar(1.0));
-		swing1 *= fact; 
+		swing1 *= fact;
 	}
 
 	if (m_swingSpan2 >= btScalar(0.05f))
 	{
-		b1Axis3 = getRigidBodyA().getCenterOfMassTransform().getBasis() * this->m_rbAFrame.getBasis().getColumn(2);			
+		b1Axis3 = getRigidBodyA().getCenterOfMassTransform().getBasis() * this->m_rbAFrame.getBasis().getColumn(2);
 		swx = b2Axis1.dot(b1Axis1);
 		swy = b2Axis1.dot(b1Axis3);
-		swing2  = btAtan2Fast(swy, swx);
-		fact = (swy*swy + swx*swx) * thresh * thresh;
+		swing2 = btAtan2Fast(swy, swx);
+		fact = (swy * swy + swx * swx) * thresh * thresh;
 		fact = fact / (fact + btScalar(1.0));
-		swing2 *= fact; 
+		swing2 *= fact;
 	}
 
-	btScalar RMaxAngle1Sq = 1.0f / (m_swingSpan1*m_swingSpan1);		
-	btScalar RMaxAngle2Sq = 1.0f / (m_swingSpan2*m_swingSpan2);	
-	btScalar EllipseAngle = btFabs(swing1*swing1)* RMaxAngle1Sq + btFabs(swing2*swing2) * RMaxAngle2Sq;
+	btScalar RMaxAngle1Sq = 1.0f / (m_swingSpan1 * m_swingSpan1);
+	btScalar RMaxAngle2Sq = 1.0f / (m_swingSpan2 * m_swingSpan2);
+	btScalar EllipseAngle = btFabs(swing1 * swing1) * RMaxAngle1Sq + btFabs(swing2 * swing2) * RMaxAngle2Sq;
 
 	if (EllipseAngle > 1.0f)
 	{
-		m_swingCorrection = EllipseAngle-1.0f;
+		m_swingCorrection = EllipseAngle - 1.0f;
 		m_solveSwingLimit = true;
 		// Calculate necessary axis & factors
-		m_swingAxis = b2Axis1.cross(b1Axis2* b2Axis1.dot(b1Axis2) + b1Axis3* b2Axis1.dot(b1Axis3));
+		m_swingAxis = b2Axis1.cross(b1Axis2 * b2Axis1.dot(b1Axis2) + b1Axis3 * b2Axis1.dot(b1Axis3));
 		m_swingAxis.normalize();
 		btScalar swingAxisSign = (b2Axis1.dot(b1Axis1) >= 0.0f) ? 1.0f : -1.0f;
 		m_swingAxis *= swingAxisSign;
@@ -594,14 +572,14 @@ void btConeTwistConstraint::calcAngleInfo()
 	if (m_twistSpan >= btScalar(0.))
 	{
 		btVector3 b2Axis2 = getRigidBodyB().getCenterOfMassTransform().getBasis() * this->m_rbBFrame.getBasis().getColumn(1);
-		btQuaternion rotationArc = shortestArcQuat(b2Axis1,b1Axis1);
-		btVector3 TwistRef = quatRotate(rotationArc,b2Axis2); 
-		btScalar twist = btAtan2Fast( TwistRef.dot(b1Axis3), TwistRef.dot(b1Axis2) );
+		btQuaternion rotationArc = shortestArcQuat(b2Axis1, b1Axis1);
+		btVector3 TwistRef = quatRotate(rotationArc, b2Axis2);
+		btScalar twist = btAtan2Fast(TwistRef.dot(b1Axis3), TwistRef.dot(b1Axis2));
 		m_twistAngle = twist;
 
-//		btScalar lockedFreeFactor = (m_twistSpan > btScalar(0.05f)) ? m_limitSoftness : btScalar(0.);
+		//		btScalar lockedFreeFactor = (m_twistSpan > btScalar(0.05f)) ? m_limitSoftness : btScalar(0.);
 		btScalar lockedFreeFactor = (m_twistSpan > btScalar(0.05f)) ? btScalar(1.0f) : btScalar(0.);
-		if (twist <= -m_twistSpan*lockedFreeFactor)
+		if (twist <= -m_twistSpan * lockedFreeFactor)
 		{
 			m_twistCorrection = -(twist + m_twistSpan);
 			m_solveTwistLimit = true;
@@ -609,7 +587,7 @@ void btConeTwistConstraint::calcAngleInfo()
 			m_twistAxis.normalize();
 			m_twistAxis *= -1.0f;
 		}
-		else if (twist >  m_twistSpan*lockedFreeFactor)
+		else if (twist > m_twistSpan * lockedFreeFactor)
 		{
 			m_twistCorrection = (twist - m_twistSpan);
 			m_solveTwistLimit = true;
@@ -618,13 +596,11 @@ void btConeTwistConstraint::calcAngleInfo()
 		}
 	}
 }
-#endif //__SPU__
+#endif  //__SPU__
 
-static btVector3 vTwist(1,0,0); // twist axis in constraint's space
+static btVector3 vTwist(1, 0, 0);  // twist axis in constraint's space
 
-
-
-void btConeTwistConstraint::calcAngleInfo2(const btTransform& transA, const btTransform& transB, const btMatrix3x3& invInertiaWorldA,const btMatrix3x3& invInertiaWorldB)
+void btConeTwistConstraint::calcAngleInfo2(const btTransform& transA, const btTransform& transB, const btMatrix3x3& invInertiaWorldA, const btMatrix3x3& invInertiaWorldB)
 {
 	m_swingCorrection = btScalar(0.);
 	m_twistLimitSign = btScalar(0.);
@@ -632,7 +608,7 @@ void btConeTwistConstraint::calcAngleInfo2(const btTransform& transA, const btTr
 	m_solveSwingLimit = false;
 	// compute rotation of A wrt B (in constraint space)
 	if (m_bMotorEnabled && (!m_useSolveConstraintObsolete))
-	{	// it is assumed that setMotorTarget() was alredy called 
+	{  // it is assumed that setMotorTarget() was alredy called
 		// and motor target m_qTarget is within constraint limits
 		// TODO : split rotation to pure swing and pure twist
 		// compute desired transforms in world
@@ -641,23 +617,22 @@ void btConeTwistConstraint::calcAngleInfo2(const btTransform& transA, const btTr
 		btTransform trB = transB * m_rbBFrame;
 		btTransform trDeltaAB = trB * trPose * trA.inverse();
 		btQuaternion qDeltaAB = trDeltaAB.getRotation();
-		btVector3 swingAxis = 	btVector3(qDeltaAB.x(), qDeltaAB.y(), qDeltaAB.z());
+		btVector3 swingAxis = btVector3(qDeltaAB.x(), qDeltaAB.y(), qDeltaAB.z());
 		btScalar swingAxisLen2 = swingAxis.length2();
-		if(btFuzzyZero(swingAxisLen2))
+		if (btFuzzyZero(swingAxisLen2))
 		{
-		   return;
+			return;
 		}
 		m_swingAxis = swingAxis;
 		m_swingAxis.normalize();
 		m_swingCorrection = qDeltaAB.getAngle();
-		if(!btFuzzyZero(m_swingCorrection))
+		if (!btFuzzyZero(m_swingCorrection))
 		{
 			m_solveSwingLimit = true;
 		}
 		return;
 	}
 
-
 	{
 		// compute rotation of A wrt B (in constraint space)
 		btQuaternion qA = transA.getRotation() * m_rbAFrame.getRotation();
@@ -665,13 +640,17 @@ void btConeTwistConstraint::calcAngleInfo2(const btTransform& transA, const btTr
 		btQuaternion qAB = qB.inverse() * qA;
 		// split rotation into cone and twist
 		// (all this is done from B's perspective. Maybe I should be averaging axes...)
-		btVector3 vConeNoTwist = quatRotate(qAB, vTwist); vConeNoTwist.normalize();
-		btQuaternion qABCone  = shortestArcQuat(vTwist, vConeNoTwist); qABCone.normalize();
-		btQuaternion qABTwist = qABCone.inverse() * qAB; qABTwist.normalize();
+		btVector3 vConeNoTwist = quatRotate(qAB, vTwist);
+		vConeNoTwist.normalize();
+		btQuaternion qABCone = shortestArcQuat(vTwist, vConeNoTwist);
+		qABCone.normalize();
+		btQuaternion qABTwist = qABCone.inverse() * qAB;
+		qABTwist.normalize();
 
 		if (m_swingSpan1 >= m_fixThresh && m_swingSpan2 >= m_fixThresh)
 		{
-			btScalar swingAngle, swingLimit = 0; btVector3 swingAxis;
+			btScalar swingAngle, swingLimit = 0;
+			btVector3 swingAxis;
 			computeConeLimitInfo(qABCone, swingAngle, swingAxis, swingLimit);
 
 			if (swingAngle > swingLimit * m_limitSoftness)
@@ -684,9 +663,9 @@ void btConeTwistConstraint::calcAngleInfo2(const btTransform& transA, const btTr
 				m_swingLimitRatio = 1.f;
 				if (swingAngle < swingLimit && m_limitSoftness < 1.f - SIMD_EPSILON)
 				{
-					m_swingLimitRatio = (swingAngle - swingLimit * m_limitSoftness)/
+					m_swingLimitRatio = (swingAngle - swingLimit * m_limitSoftness) /
 										(swingLimit - swingLimit * m_limitSoftness);
-				}				
+				}
 
 				// swing correction tries to get back to soft limit
 				m_swingCorrection = swingAngle - (swingLimit * m_limitSoftness);
@@ -694,14 +673,14 @@ void btConeTwistConstraint::calcAngleInfo2(const btTransform& transA, const btTr
 				// adjustment of swing axis (based on ellipse normal)
 				adjustSwingAxisToUseEllipseNormal(swingAxis);
 
-				// Calculate necessary axis & factors		
+				// Calculate necessary axis & factors
 				m_swingAxis = quatRotate(qB, -swingAxis);
 
-				m_twistAxisA.setValue(0,0,0);
+				m_twistAxisA.setValue(0, 0, 0);
 
-				m_kSwing =  btScalar(1.) /
-					(computeAngularImpulseDenominator(m_swingAxis,invInertiaWorldA) +
-					 computeAngularImpulseDenominator(m_swingAxis,invInertiaWorldB));
+				m_kSwing = btScalar(1.) /
+						   (computeAngularImpulseDenominator(m_swingAxis, invInertiaWorldA) +
+							computeAngularImpulseDenominator(m_swingAxis, invInertiaWorldB));
 			}
 		}
 		else
@@ -717,9 +696,9 @@ void btConeTwistConstraint::calcAngleInfo2(const btTransform& transA, const btTr
 			btScalar x = ivB.dot(ivA);
 			btScalar y = ivB.dot(jvA);
 			btScalar z = ivB.dot(kvA);
-			if((m_swingSpan1 < m_fixThresh) && (m_swingSpan2 < m_fixThresh))
-			{ // fixed. We'll need to add one more row to constraint
-				if((!btFuzzyZero(y)) || (!(btFuzzyZero(z))))
+			if ((m_swingSpan1 < m_fixThresh) && (m_swingSpan2 < m_fixThresh))
+			{  // fixed. We'll need to add one more row to constraint
+				if ((!btFuzzyZero(y)) || (!(btFuzzyZero(z))))
 				{
 					m_solveSwingLimit = true;
 					m_swingAxis = -ivB.cross(ivA);
@@ -727,47 +706,47 @@ void btConeTwistConstraint::calcAngleInfo2(const btTransform& transA, const btTr
 			}
 			else
 			{
-				if(m_swingSpan1 < m_fixThresh)
-				{ // hinge around Y axis
-//					if(!(btFuzzyZero(y)))
-					if((!(btFuzzyZero(x))) || (!(btFuzzyZero(z))))
+				if (m_swingSpan1 < m_fixThresh)
+				{  // hinge around Y axis
+					//					if(!(btFuzzyZero(y)))
+					if ((!(btFuzzyZero(x))) || (!(btFuzzyZero(z))))
 					{
 						m_solveSwingLimit = true;
-						if(m_swingSpan2 >= m_fixThresh)
+						if (m_swingSpan2 >= m_fixThresh)
 						{
 							y = btScalar(0.f);
 							btScalar span2 = btAtan2(z, x);
-							if(span2 > m_swingSpan2)
+							if (span2 > m_swingSpan2)
 							{
 								x = btCos(m_swingSpan2);
 								z = btSin(m_swingSpan2);
 							}
-							else if(span2 < -m_swingSpan2)
+							else if (span2 < -m_swingSpan2)
 							{
-								x =  btCos(m_swingSpan2);
+								x = btCos(m_swingSpan2);
 								z = -btSin(m_swingSpan2);
 							}
 						}
 					}
 				}
 				else
-				{ // hinge around Z axis
-//					if(!btFuzzyZero(z))
-					if((!(btFuzzyZero(x))) || (!(btFuzzyZero(y))))
+				{  // hinge around Z axis
+					//					if(!btFuzzyZero(z))
+					if ((!(btFuzzyZero(x))) || (!(btFuzzyZero(y))))
 					{
 						m_solveSwingLimit = true;
-						if(m_swingSpan1 >= m_fixThresh)
+						if (m_swingSpan1 >= m_fixThresh)
 						{
 							z = btScalar(0.f);
 							btScalar span1 = btAtan2(y, x);
-							if(span1 > m_swingSpan1)
+							if (span1 > m_swingSpan1)
 							{
 								x = btCos(m_swingSpan1);
 								y = btSin(m_swingSpan1);
 							}
-							else if(span1 < -m_swingSpan1)
+							else if (span1 < -m_swingSpan1)
 							{
-								x =  btCos(m_swingSpan1);
+								x = btCos(m_swingSpan1);
 								y = -btSin(m_swingSpan1);
 							}
 						}
@@ -778,10 +757,10 @@ void btConeTwistConstraint::calcAngleInfo2(const btTransform& transA, const btTr
 				target[2] = x * ivA[2] + y * jvA[2] + z * kvA[2];
 				target.normalize();
 				m_swingAxis = -ivB.cross(target);
-                                m_swingCorrection = m_swingAxis.length();
+				m_swingCorrection = m_swingAxis.length();
 
-                                if (!btFuzzyZero(m_swingCorrection))
-                                    m_swingAxis.normalize();
+				if (!btFuzzyZero(m_swingCorrection))
+					m_swingAxis.normalize();
 			}
 		}
 
@@ -790,15 +769,15 @@ void btConeTwistConstraint::calcAngleInfo2(const btTransform& transA, const btTr
 			btVector3 twistAxis;
 			computeTwistLimitInfo(qABTwist, m_twistAngle, twistAxis);
 
-			if (m_twistAngle > m_twistSpan*m_limitSoftness)
+			if (m_twistAngle > m_twistSpan * m_limitSoftness)
 			{
 				m_solveTwistLimit = true;
 
 				m_twistLimitRatio = 1.f;
 				if (m_twistAngle < m_twistSpan && m_limitSoftness < 1.f - SIMD_EPSILON)
 				{
-					m_twistLimitRatio = (m_twistAngle - m_twistSpan * m_limitSoftness)/
-										(m_twistSpan  - m_twistSpan * m_limitSoftness);
+					m_twistLimitRatio = (m_twistAngle - m_twistSpan * m_limitSoftness) /
+										(m_twistSpan - m_twistSpan * m_limitSoftness);
 				}
 
 				// twist correction tries to get back to soft limit
@@ -807,8 +786,8 @@ void btConeTwistConstraint::calcAngleInfo2(const btTransform& transA, const btTr
 				m_twistAxis = quatRotate(qB, -twistAxis);
 
 				m_kTwist = btScalar(1.) /
-					(computeAngularImpulseDenominator(m_twistAxis,invInertiaWorldA) +
-					 computeAngularImpulseDenominator(m_twistAxis,invInertiaWorldB));
+						   (computeAngularImpulseDenominator(m_twistAxis, invInertiaWorldA) +
+							computeAngularImpulseDenominator(m_twistAxis, invInertiaWorldB));
 			}
 
 			if (m_solveSwingLimit)
@@ -821,15 +800,13 @@ void btConeTwistConstraint::calcAngleInfo2(const btTransform& transA, const btTr
 	}
 }
 
-
-
 // given a cone rotation in constraint space, (pre: twist must already be removed)
 // this method computes its corresponding swing angle and axis.
 // more interestingly, it computes the cone/swing limit (angle) for this cone "pose".
 void btConeTwistConstraint::computeConeLimitInfo(const btQuaternion& qCone,
-												 btScalar& swingAngle, // out
-												 btVector3& vSwingAxis, // out
-												 btScalar& swingLimit) // out
+												 btScalar& swingAngle,   // out
+												 btVector3& vSwingAxis,  // out
+												 btScalar& swingLimit)   // out
 {
 	swingAngle = qCone.getAngle();
 	if (swingAngle > SIMD_EPSILON)
@@ -840,7 +817,7 @@ void btConeTwistConstraint::computeConeLimitInfo(const btQuaternion& qCone,
         // non-zero twist?! this should never happen.
        btAssert(fabs(vSwingAxis.x()) <= SIMD_EPSILON));
 #endif
-        
+
 		// Compute limit for given swing. tricky:
 		// Given a swing axis, we're looking for the intersection with the bounding cone ellipse.
 		// (Since we're dealing with angles, this ellipse is embedded on the surface of a sphere.)
@@ -848,7 +825,7 @@ void btConeTwistConstraint::computeConeLimitInfo(const btQuaternion& qCone,
 		// For starters, compute the direction from center to surface of ellipse.
 		// This is just the perpendicular (ie. rotate 2D vector by PI/2) of the swing axis.
 		// (vSwingAxis is the cone rotation (in z,y); change vars and rotate to (x,y) coords.)
-		btScalar xEllipse =  vSwingAxis.y();
+		btScalar xEllipse = vSwingAxis.y();
 		btScalar yEllipse = -vSwingAxis.z();
 
 		// Now, we use the slope of the vector (using x/yEllipse) and find the length
@@ -858,10 +835,10 @@ void btConeTwistConstraint::computeConeLimitInfo(const btQuaternion& qCone,
 		//  a^2   b^2
 		// Do the math and it should be clear.
 
-		swingLimit = m_swingSpan1; // if xEllipse == 0, we have a pure vSwingAxis.z rotation: just use swingspan1
+		swingLimit = m_swingSpan1;  // if xEllipse == 0, we have a pure vSwingAxis.z rotation: just use swingspan1
 		if (fabs(xEllipse) > SIMD_EPSILON)
 		{
-			btScalar surfaceSlope2 = (yEllipse*yEllipse)/(xEllipse*xEllipse);
+			btScalar surfaceSlope2 = (yEllipse * yEllipse) / (xEllipse * xEllipse);
 			btScalar norm = 1 / (m_swingSpan2 * m_swingSpan2);
 			norm += surfaceSlope2 / (m_swingSpan1 * m_swingSpan1);
 			btScalar swingLimit2 = (1 + surfaceSlope2) / norm;
@@ -887,7 +864,7 @@ void btConeTwistConstraint::computeConeLimitInfo(const btQuaternion& qCone,
 #if 0
         btAssert(0);
 #endif
- 	}
+	}
 }
 
 btVector3 btConeTwistConstraint::GetPointForAngle(btScalar fAngleInRadians, btScalar fLength) const
@@ -903,10 +880,10 @@ btVector3 btConeTwistConstraint::GetPointForAngle(btScalar fAngleInRadians, btSc
 	//  a^2   b^2
 	// Do the math and it should be clear.
 
-	btScalar swingLimit = m_swingSpan1; // if xEllipse == 0, just use axis b (1)
+	btScalar swingLimit = m_swingSpan1;  // if xEllipse == 0, just use axis b (1)
 	if (fabs(xEllipse) > SIMD_EPSILON)
 	{
-		btScalar surfaceSlope2 = (yEllipse*yEllipse)/(xEllipse*xEllipse);
+		btScalar surfaceSlope2 = (yEllipse * yEllipse) / (xEllipse * xEllipse);
 		btScalar norm = 1 / (m_swingSpan2 * m_swingSpan2);
 		norm += surfaceSlope2 / (m_swingSpan1 * m_swingSpan1);
 		btScalar swingLimit2 = (1 + surfaceSlope2) / norm;
@@ -917,20 +894,20 @@ btVector3 btConeTwistConstraint::GetPointForAngle(btScalar fAngleInRadians, btSc
 	// note: twist is x-axis, swing 1 and 2 are along the z and y axes respectively
 	btVector3 vSwingAxis(0, xEllipse, -yEllipse);
 	btQuaternion qSwing(vSwingAxis, swingLimit);
-	btVector3 vPointInConstraintSpace(fLength,0,0);
+	btVector3 vPointInConstraintSpace(fLength, 0, 0);
 	return quatRotate(qSwing, vPointInConstraintSpace);
 }
 
 // given a twist rotation in constraint space, (pre: cone must already be removed)
 // this method computes its corresponding angle and axis.
 void btConeTwistConstraint::computeTwistLimitInfo(const btQuaternion& qTwist,
-												  btScalar& twistAngle, // out
-												  btVector3& vTwistAxis) // out
+												  btScalar& twistAngle,   // out
+												  btVector3& vTwistAxis)  // out
 {
 	btQuaternion qMinTwist = qTwist;
 	twistAngle = qTwist.getAngle();
 
-	if (twistAngle > SIMD_PI) // long way around. flip quat and recalculate.
+	if (twistAngle > SIMD_PI)  // long way around. flip quat and recalculate.
 	{
 		qMinTwist = -(qTwist);
 		twistAngle = qMinTwist.getAngle();
@@ -948,80 +925,79 @@ void btConeTwistConstraint::computeTwistLimitInfo(const btQuaternion& qTwist,
 		vTwistAxis.normalize();
 }
 
-
 void btConeTwistConstraint::adjustSwingAxisToUseEllipseNormal(btVector3& vSwingAxis) const
 {
 	// the swing axis is computed as the "twist-free" cone rotation,
 	// but the cone limit is not circular, but elliptical (if swingspan1 != swingspan2).
-	// so, if we're outside the limits, the closest way back inside the cone isn't 
+	// so, if we're outside the limits, the closest way back inside the cone isn't
 	// along the vector back to the center. better (and more stable) to use the ellipse normal.
 
 	// convert swing axis to direction from center to surface of ellipse
 	// (ie. rotate 2D vector by PI/2)
 	btScalar y = -vSwingAxis.z();
-	btScalar z =  vSwingAxis.y();
+	btScalar z = vSwingAxis.y();
 
 	// do the math...
-	if (fabs(z) > SIMD_EPSILON) // avoid division by 0. and we don't need an update if z == 0.
+	if (fabs(z) > SIMD_EPSILON)  // avoid division by 0. and we don't need an update if z == 0.
 	{
 		// compute gradient/normal of ellipse surface at current "point"
-		btScalar grad = y/z;
+		btScalar grad = y / z;
 		grad *= m_swingSpan2 / m_swingSpan1;
 
 		// adjust y/z to represent normal at point (instead of vector to point)
 		if (y > 0)
-			y =  fabs(grad * z);
+			y = fabs(grad * z);
 		else
 			y = -fabs(grad * z);
 
 		// convert ellipse direction back to swing axis
 		vSwingAxis.setZ(-y);
-		vSwingAxis.setY( z);
+		vSwingAxis.setY(z);
 		vSwingAxis.normalize();
 	}
 }
 
-
-
-void btConeTwistConstraint::setMotorTarget(const btQuaternion &q)
+void btConeTwistConstraint::setMotorTarget(const btQuaternion& q)
 {
 	//btTransform trACur = m_rbA.getCenterOfMassTransform();
 	//btTransform trBCur = m_rbB.getCenterOfMassTransform();
-//	btTransform trABCur = trBCur.inverse() * trACur;
-//	btQuaternion qABCur = trABCur.getRotation();
-//	btTransform trConstraintCur = (trBCur * m_rbBFrame).inverse() * (trACur * m_rbAFrame);
+	//	btTransform trABCur = trBCur.inverse() * trACur;
+	//	btQuaternion qABCur = trABCur.getRotation();
+	//	btTransform trConstraintCur = (trBCur * m_rbBFrame).inverse() * (trACur * m_rbAFrame);
 	//btQuaternion qConstraintCur = trConstraintCur.getRotation();
 
 	btQuaternion qConstraint = m_rbBFrame.getRotation().inverse() * q * m_rbAFrame.getRotation();
 	setMotorTargetInConstraintSpace(qConstraint);
 }
 
-
-void btConeTwistConstraint::setMotorTargetInConstraintSpace(const btQuaternion &q)
+void btConeTwistConstraint::setMotorTargetInConstraintSpace(const btQuaternion& q)
 {
 	m_qTarget = q;
 
 	// clamp motor target to within limits
 	{
-		btScalar softness = 1.f;//m_limitSoftness;
+		btScalar softness = 1.f;  //m_limitSoftness;
 
 		// split into twist and cone
 		btVector3 vTwisted = quatRotate(m_qTarget, vTwist);
-		btQuaternion qTargetCone  = shortestArcQuat(vTwist, vTwisted); qTargetCone.normalize();
-		btQuaternion qTargetTwist = qTargetCone.inverse() * m_qTarget; qTargetTwist.normalize();
+		btQuaternion qTargetCone = shortestArcQuat(vTwist, vTwisted);
+		qTargetCone.normalize();
+		btQuaternion qTargetTwist = qTargetCone.inverse() * m_qTarget;
+		qTargetTwist.normalize();
 
 		// clamp cone
 		if (m_swingSpan1 >= btScalar(0.05f) && m_swingSpan2 >= btScalar(0.05f))
 		{
-			btScalar swingAngle, swingLimit; btVector3 swingAxis;
+			btScalar swingAngle, swingLimit;
+			btVector3 swingAxis;
 			computeConeLimitInfo(qTargetCone, swingAngle, swingAxis, swingLimit);
 
 			if (fabs(swingAngle) > SIMD_EPSILON)
 			{
-				if (swingAngle > swingLimit*softness)
-					swingAngle = swingLimit*softness;
-				else if (swingAngle < -swingLimit*softness)
-					swingAngle = -swingLimit*softness;
+				if (swingAngle > swingLimit * softness)
+					swingAngle = swingLimit * softness;
+				else if (swingAngle < -swingLimit * softness)
+					swingAngle = -swingLimit * softness;
 				qTargetCone = btQuaternion(swingAxis, swingAngle);
 			}
 		}
@@ -1029,16 +1005,17 @@ void btConeTwistConstraint::setMotorTargetInConstraintSpace(const btQuaternion &
 		// clamp twist
 		if (m_twistSpan >= btScalar(0.05f))
 		{
-			btScalar twistAngle; btVector3 twistAxis;
+			btScalar twistAngle;
+			btVector3 twistAxis;
 			computeTwistLimitInfo(qTargetTwist, twistAngle, twistAxis);
 
 			if (fabs(twistAngle) > SIMD_EPSILON)
 			{
 				// eddy todo: limitSoftness used here???
-				if (twistAngle > m_twistSpan*softness)
-					twistAngle = m_twistSpan*softness;
-				else if (twistAngle < -m_twistSpan*softness)
-					twistAngle = -m_twistSpan*softness;
+				if (twistAngle > m_twistSpan * softness)
+					twistAngle = m_twistSpan * softness;
+				else if (twistAngle < -m_twistSpan * softness)
+					twistAngle = -m_twistSpan * softness;
 				qTargetTwist = btQuaternion(twistAxis, twistAngle);
 			}
 		}
@@ -1047,15 +1024,15 @@ void btConeTwistConstraint::setMotorTargetInConstraintSpace(const btQuaternion &
 	}
 }
 
-///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5). 
+///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5).
 ///If no axis is provided, it uses the default axis for this constraint.
 void btConeTwistConstraint::setParam(int num, btScalar value, int axis)
 {
-	switch(num)
+	switch (num)
 	{
-		case BT_CONSTRAINT_ERP :
-		case BT_CONSTRAINT_STOP_ERP :
-			if((axis >= 0) && (axis < 3)) 
+		case BT_CONSTRAINT_ERP:
+		case BT_CONSTRAINT_STOP_ERP:
+			if ((axis >= 0) && (axis < 3))
 			{
 				m_linERP = value;
 				m_flags |= BT_CONETWIST_FLAGS_LIN_ERP;
@@ -1065,9 +1042,9 @@ void btConeTwistConstraint::setParam(int num, btScalar value, int axis)
 				m_biasFactor = value;
 			}
 			break;
-		case BT_CONSTRAINT_CFM :
-		case BT_CONSTRAINT_STOP_CFM :
-			if((axis >= 0) && (axis < 3)) 
+		case BT_CONSTRAINT_CFM:
+		case BT_CONSTRAINT_STOP_CFM:
+			if ((axis >= 0) && (axis < 3))
 			{
 				m_linCFM = value;
 				m_flags |= BT_CONETWIST_FLAGS_LIN_CFM;
@@ -1085,19 +1062,19 @@ void btConeTwistConstraint::setParam(int num, btScalar value, int axis)
 }
 
 ///return the local value of parameter
-btScalar btConeTwistConstraint::getParam(int num, int axis) const 
+btScalar btConeTwistConstraint::getParam(int num, int axis) const
 {
 	btScalar retVal = 0;
-	switch(num)
+	switch (num)
 	{
-		case BT_CONSTRAINT_ERP :
-		case BT_CONSTRAINT_STOP_ERP :
-			if((axis >= 0) && (axis < 3)) 
+		case BT_CONSTRAINT_ERP:
+		case BT_CONSTRAINT_STOP_ERP:
+			if ((axis >= 0) && (axis < 3))
 			{
 				btAssertConstrParams(m_flags & BT_CONETWIST_FLAGS_LIN_ERP);
 				retVal = m_linERP;
 			}
-			else if((axis >= 3) && (axis < 6)) 
+			else if ((axis >= 3) && (axis < 6))
 			{
 				retVal = m_biasFactor;
 			}
@@ -1106,14 +1083,14 @@ btScalar btConeTwistConstraint::getParam(int num, int axis) const
 				btAssertConstrParams(0);
 			}
 			break;
-		case BT_CONSTRAINT_CFM :
-		case BT_CONSTRAINT_STOP_CFM :
-			if((axis >= 0) && (axis < 3)) 
+		case BT_CONSTRAINT_CFM:
+		case BT_CONSTRAINT_STOP_CFM:
+			if ((axis >= 0) && (axis < 3))
 			{
 				btAssertConstrParams(m_flags & BT_CONETWIST_FLAGS_LIN_CFM);
 				retVal = m_linCFM;
 			}
-			else if((axis >= 3) && (axis < 6)) 
+			else if ((axis >= 3) && (axis < 6))
 			{
 				btAssertConstrParams(m_flags & BT_CONETWIST_FLAGS_ANG_CFM);
 				retVal = m_angCFM;
@@ -1123,21 +1100,16 @@ btScalar btConeTwistConstraint::getParam(int num, int axis) const
 				btAssertConstrParams(0);
 			}
 			break;
-		default : 
+		default:
 			btAssertConstrParams(0);
 	}
 	return retVal;
 }
 
-
-void btConeTwistConstraint::setFrames(const btTransform & frameA, const btTransform & frameB)
+void btConeTwistConstraint::setFrames(const btTransform& frameA, const btTransform& frameB)
 {
 	m_rbAFrame = frameA;
 	m_rbBFrame = frameB;
 	buildJacobian();
 	//calculateTransforms();
 }
-
- 
-
-
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btConeTwistConstraint.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btConeTwistConstraint.h
index 7a33d01d1eb..64f44df1cba 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btConeTwistConstraint.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btConeTwistConstraint.h
@@ -15,8 +15,6 @@ subject to the following restrictions:
 Written by: Marcus Hennix
 */
 
-
-
 /*
 Overview:
 
@@ -31,8 +29,6 @@ twist is along the x-axis,
 and swing 1 and 2 are along the z and y axes respectively.
 */
 
-
-
 #ifndef BT_CONETWISTCONSTRAINT_H
 #define BT_CONETWISTCONSTRAINT_H
 
@@ -41,13 +37,12 @@ and swing 1 and 2 are along the z and y axes respectively.
 #include "btTypedConstraint.h"
 
 #ifdef BT_USE_DOUBLE_PRECISION
-#define btConeTwistConstraintData2	btConeTwistConstraintDoubleData
-#define btConeTwistConstraintDataName	"btConeTwistConstraintDoubleData"
+#define btConeTwistConstraintData2 btConeTwistConstraintDoubleData
+#define btConeTwistConstraintDataName "btConeTwistConstraintDoubleData"
 #else
-#define btConeTwistConstraintData2	btConeTwistConstraintData 
-#define btConeTwistConstraintDataName	"btConeTwistConstraintData" 
-#endif //BT_USE_DOUBLE_PRECISION
-
+#define btConeTwistConstraintData2 btConeTwistConstraintData
+#define btConeTwistConstraintDataName "btConeTwistConstraintData"
+#endif  //BT_USE_DOUBLE_PRECISION
 
 class btRigidBody;
 
@@ -59,103 +54,99 @@ enum btConeTwistFlags
 };
 
 ///btConeTwistConstraint can be used to simulate ragdoll joints (upper arm, leg etc)
-ATTRIBUTE_ALIGNED16(class) btConeTwistConstraint : public btTypedConstraint
+ATTRIBUTE_ALIGNED16(class)
+btConeTwistConstraint : public btTypedConstraint
 {
 #ifdef IN_PARALLELL_SOLVER
 public:
 #endif
-	btJacobianEntry	m_jac[3]; //3 orthogonal linear constraints
+	btJacobianEntry m_jac[3];  //3 orthogonal linear constraints
 
-	btTransform m_rbAFrame; 
+	btTransform m_rbAFrame;
 	btTransform m_rbBFrame;
 
-	btScalar	m_limitSoftness;
-	btScalar	m_biasFactor;
-	btScalar	m_relaxationFactor;
+	btScalar m_limitSoftness;
+	btScalar m_biasFactor;
+	btScalar m_relaxationFactor;
 
-	btScalar	m_damping;
+	btScalar m_damping;
 
-	btScalar	m_swingSpan1;
-	btScalar	m_swingSpan2;
-	btScalar	m_twistSpan;
+	btScalar m_swingSpan1;
+	btScalar m_swingSpan2;
+	btScalar m_twistSpan;
 
-	btScalar	m_fixThresh;
+	btScalar m_fixThresh;
 
-	btVector3   m_swingAxis;
-	btVector3	m_twistAxis;
+	btVector3 m_swingAxis;
+	btVector3 m_twistAxis;
 
-	btScalar	m_kSwing;
-	btScalar	m_kTwist;
+	btScalar m_kSwing;
+	btScalar m_kTwist;
 
-	btScalar	m_twistLimitSign;
-	btScalar	m_swingCorrection;
-	btScalar	m_twistCorrection;
+	btScalar m_twistLimitSign;
+	btScalar m_swingCorrection;
+	btScalar m_twistCorrection;
 
-	btScalar	m_twistAngle;
+	btScalar m_twistAngle;
 
-	btScalar	m_accSwingLimitImpulse;
-	btScalar	m_accTwistLimitImpulse;
+	btScalar m_accSwingLimitImpulse;
+	btScalar m_accTwistLimitImpulse;
 
-	bool		m_angularOnly;
-	bool		m_solveTwistLimit;
-	bool		m_solveSwingLimit;
+	bool m_angularOnly;
+	bool m_solveTwistLimit;
+	bool m_solveSwingLimit;
 
-	bool	m_useSolveConstraintObsolete;
+	bool m_useSolveConstraintObsolete;
 
 	// not yet used...
-	btScalar	m_swingLimitRatio;
-	btScalar	m_twistLimitRatio;
-	btVector3   m_twistAxisA;
+	btScalar m_swingLimitRatio;
+	btScalar m_twistLimitRatio;
+	btVector3 m_twistAxisA;
 
 	// motor
-	bool		 m_bMotorEnabled;
-	bool		 m_bNormalizedMotorStrength;
+	bool m_bMotorEnabled;
+	bool m_bNormalizedMotorStrength;
 	btQuaternion m_qTarget;
-	btScalar	 m_maxMotorImpulse;
-	btVector3	 m_accMotorImpulse;
-	
-	// parameters
-	int			m_flags;
-	btScalar	m_linCFM;
-	btScalar	m_linERP;
-	btScalar	m_angCFM;
-	
-protected:
+	btScalar m_maxMotorImpulse;
+	btVector3 m_accMotorImpulse;
 
+	// parameters
+	int m_flags;
+	btScalar m_linCFM;
+	btScalar m_linERP;
+	btScalar m_angCFM;
+
+protected:
 	void init();
 
-	void computeConeLimitInfo(const btQuaternion& qCone, // in
-		btScalar& swingAngle, btVector3& vSwingAxis, btScalar& swingLimit); // all outs
+	void computeConeLimitInfo(const btQuaternion& qCone,                                           // in
+							  btScalar& swingAngle, btVector3& vSwingAxis, btScalar& swingLimit);  // all outs
 
-	void computeTwistLimitInfo(const btQuaternion& qTwist, // in
-		btScalar& twistAngle, btVector3& vTwistAxis); // all outs
-
-	void adjustSwingAxisToUseEllipseNormal(btVector3& vSwingAxis) const;
+	void computeTwistLimitInfo(const btQuaternion& qTwist,                    // in
+							   btScalar& twistAngle, btVector3& vTwistAxis);  // all outs
 
+	void adjustSwingAxisToUseEllipseNormal(btVector3 & vSwingAxis) const;
 
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-	btConeTwistConstraint(btRigidBody& rbA,btRigidBody& rbB,const btTransform& rbAFrame, const btTransform& rbBFrame);
-	
-	btConeTwistConstraint(btRigidBody& rbA,const btTransform& rbAFrame);
+	btConeTwistConstraint(btRigidBody & rbA, btRigidBody & rbB, const btTransform& rbAFrame, const btTransform& rbBFrame);
 
-	virtual void	buildJacobian();
+	btConeTwistConstraint(btRigidBody & rbA, const btTransform& rbAFrame);
 
-	virtual void getInfo1 (btConstraintInfo1* info);
+	virtual void buildJacobian();
 
-	void	getInfo1NonVirtual(btConstraintInfo1* info);
-	
-	virtual void getInfo2 (btConstraintInfo2* info);
-	
-	void	getInfo2NonVirtual(btConstraintInfo2* info,const btTransform& transA,const btTransform& transB,const btMatrix3x3& invInertiaWorldA,const btMatrix3x3& invInertiaWorldB);
+	virtual void getInfo1(btConstraintInfo1 * info);
 
-	virtual	void	solveConstraintObsolete(btSolverBody& bodyA,btSolverBody& bodyB,btScalar	timeStep);
+	void getInfo1NonVirtual(btConstraintInfo1 * info);
 
-    
-	void	updateRHS(btScalar	timeStep);
+	virtual void getInfo2(btConstraintInfo2 * info);
 
+	void getInfo2NonVirtual(btConstraintInfo2 * info, const btTransform& transA, const btTransform& transB, const btMatrix3x3& invInertiaWorldA, const btMatrix3x3& invInertiaWorldB);
+
+	virtual void solveConstraintObsolete(btSolverBody & bodyA, btSolverBody & bodyB, btScalar timeStep);
+
+	void updateRHS(btScalar timeStep);
 
 	const btRigidBody& getRigidBodyA() const
 	{
@@ -166,64 +157,64 @@ public:
 		return m_rbB;
 	}
 
-	void	setAngularOnly(bool angularOnly)
+	void setAngularOnly(bool angularOnly)
 	{
 		m_angularOnly = angularOnly;
 	}
-	
-	bool    getAngularOnly() const
+
+	bool getAngularOnly() const
 	{
-	    return m_angularOnly;
+		return m_angularOnly;
 	}
 
-	void	setLimit(int limitIndex,btScalar limitValue)
+	void setLimit(int limitIndex, btScalar limitValue)
 	{
 		switch (limitIndex)
 		{
-		case 3:
+			case 3:
 			{
 				m_twistSpan = limitValue;
 				break;
 			}
-		case 4:
+			case 4:
 			{
 				m_swingSpan2 = limitValue;
 				break;
 			}
-		case 5:
+			case 5:
 			{
 				m_swingSpan1 = limitValue;
 				break;
 			}
-		default:
+			default:
 			{
 			}
 		};
 	}
 
-    btScalar getLimit(int limitIndex) const
+	btScalar getLimit(int limitIndex) const
 	{
 		switch (limitIndex)
 		{
-		case 3:
+			case 3:
 			{
 				return m_twistSpan;
 				break;
 			}
-		case 4:
+			case 4:
 			{
 				return m_swingSpan2;
 				break;
 			}
-		case 5:
+			case 5:
 			{
 				return m_swingSpan1;
 				break;
 			}
-		default:
+			default:
 			{
-			    btAssert(0 && "Invalid limitIndex specified for btConeTwistConstraint");
-			    return 0.0;
+				btAssert(0 && "Invalid limitIndex specified for btConeTwistConstraint");
+				return 0.0;
 			}
 		};
 	}
@@ -239,18 +230,18 @@ public:
 	// __relaxationFactor:
 	//		0->1, recommend to stay near 1.
 	//		the lower the value, the less the constraint will fight velocities which violate the angular limits.
-	void	setLimit(btScalar _swingSpan1,btScalar _swingSpan2,btScalar _twistSpan, btScalar _softness = 1.f, btScalar _biasFactor = 0.3f, btScalar _relaxationFactor = 1.0f)
+	void setLimit(btScalar _swingSpan1, btScalar _swingSpan2, btScalar _twistSpan, btScalar _softness = 1.f, btScalar _biasFactor = 0.3f, btScalar _relaxationFactor = 1.0f)
 	{
 		m_swingSpan1 = _swingSpan1;
 		m_swingSpan2 = _swingSpan2;
-		m_twistSpan  = _twistSpan;
+		m_twistSpan = _twistSpan;
 
-		m_limitSoftness =  _softness;
+		m_limitSoftness = _softness;
 		m_biasFactor = _biasFactor;
 		m_relaxationFactor = _relaxationFactor;
 	}
 
-	const btTransform& getAFrame() const { return m_rbAFrame; };	
+	const btTransform& getAFrame() const { return m_rbAFrame; };
 	const btTransform& getBFrame() const { return m_rbBFrame; };
 
 	inline int getSolveTwistLimit()
@@ -269,7 +260,7 @@ public:
 	}
 
 	void calcAngleInfo();
-	void calcAngleInfo2(const btTransform& transA, const btTransform& transB,const btMatrix3x3& invInertiaWorldA,const btMatrix3x3& invInertiaWorldB);
+	void calcAngleInfo2(const btTransform& transA, const btTransform& transB, const btMatrix3x3& invInertiaWorldA, const btMatrix3x3& invInertiaWorldB);
 
 	inline btScalar getSwingSpan1() const
 	{
@@ -308,8 +299,16 @@ public:
 	bool isMotorEnabled() const { return m_bMotorEnabled; }
 	btScalar getMaxMotorImpulse() const { return m_maxMotorImpulse; }
 	bool isMaxMotorImpulseNormalized() const { return m_bNormalizedMotorStrength; }
-	void setMaxMotorImpulse(btScalar maxMotorImpulse) { m_maxMotorImpulse = maxMotorImpulse; m_bNormalizedMotorStrength = false; }
-	void setMaxMotorImpulseNormalized(btScalar maxMotorImpulse) { m_maxMotorImpulse = maxMotorImpulse; m_bNormalizedMotorStrength = true; }
+	void setMaxMotorImpulse(btScalar maxMotorImpulse)
+	{
+		m_maxMotorImpulse = maxMotorImpulse;
+		m_bNormalizedMotorStrength = false;
+	}
+	void setMaxMotorImpulseNormalized(btScalar maxMotorImpulse)
+	{
+		m_maxMotorImpulse = maxMotorImpulse;
+		m_bNormalizedMotorStrength = true;
+	}
 
 	btScalar getFixThresh() { return m_fixThresh; }
 	void setFixThresh(btScalar fixThresh) { m_fixThresh = fixThresh; }
@@ -318,17 +317,17 @@ public:
 	// q: the desired rotation of bodyA wrt bodyB.
 	// note: if q violates the joint limits, the internal target is clamped to avoid conflicting impulses (very bad for stability)
 	// note: don't forget to enableMotor()
-	void setMotorTarget(const btQuaternion &q);
+	void setMotorTarget(const btQuaternion& q);
 	const btQuaternion& getMotorTarget() const { return m_qTarget; }
 
 	// same as above, but q is the desired rotation of frameA wrt frameB in constraint space
-	void setMotorTargetInConstraintSpace(const btQuaternion &q);
+	void setMotorTargetInConstraintSpace(const btQuaternion& q);
 
 	btVector3 GetPointForAngle(btScalar fAngleInRadians, btScalar fLength) const;
 
-	///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5). 
+	///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5).
 	///If no axis is provided, it uses the default axis for this constraint.
-	virtual	void setParam(int num, btScalar value, int axis = -1);
+	virtual void setParam(int num, btScalar value, int axis = -1);
 
 	virtual void setFrames(const btTransform& frameA, const btTransform& frameB);
 
@@ -342,84 +341,74 @@ public:
 		return m_rbBFrame;
 	}
 
-
 	///return the local value of parameter
-	virtual	btScalar getParam(int num, int axis = -1) const;
+	virtual btScalar getParam(int num, int axis = -1) const;
 
 	int getFlags() const
 	{
 		return m_flags;
 	}
 
-	virtual	int	calculateSerializeBufferSize() const;
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
-
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 };
 
-
-	
-struct	btConeTwistConstraintDoubleData
+struct btConeTwistConstraintDoubleData
 {
-	btTypedConstraintDoubleData	m_typeConstraintData;
+	btTypedConstraintDoubleData m_typeConstraintData;
 	btTransformDoubleData m_rbAFrame;
 	btTransformDoubleData m_rbBFrame;
 
 	//limits
-	double	m_swingSpan1;
-	double	m_swingSpan2;
-	double	m_twistSpan;
-	double	m_limitSoftness;
-	double	m_biasFactor;
-	double	m_relaxationFactor;
-
-	double	m_damping;
-		
-	
+	double m_swingSpan1;
+	double m_swingSpan2;
+	double m_twistSpan;
+	double m_limitSoftness;
+	double m_biasFactor;
+	double m_relaxationFactor;
 
+	double m_damping;
 };
 
 #ifdef BT_BACKWARDS_COMPATIBLE_SERIALIZATION
 ///this structure is not used, except for loading pre-2.82 .bullet files
-struct	btConeTwistConstraintData
+struct btConeTwistConstraintData
 {
-	btTypedConstraintData	m_typeConstraintData;
+	btTypedConstraintData m_typeConstraintData;
 	btTransformFloatData m_rbAFrame;
 	btTransformFloatData m_rbBFrame;
 
 	//limits
-	float	m_swingSpan1;
-	float	m_swingSpan2;
-	float	m_twistSpan;
-	float	m_limitSoftness;
-	float	m_biasFactor;
-	float	m_relaxationFactor;
+	float m_swingSpan1;
+	float m_swingSpan2;
+	float m_twistSpan;
+	float m_limitSoftness;
+	float m_biasFactor;
+	float m_relaxationFactor;
+
+	float m_damping;
 
-	float	m_damping;
-		
 	char m_pad[4];
-
 };
-#endif //BT_BACKWARDS_COMPATIBLE_SERIALIZATION
+#endif  //BT_BACKWARDS_COMPATIBLE_SERIALIZATION
 //
 
-SIMD_FORCE_INLINE int	btConeTwistConstraint::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btConeTwistConstraint::calculateSerializeBufferSize() const
 {
 	return sizeof(btConeTwistConstraintData2);
-
 }
 
-
-	///fills the dataBuffer and returns the struct name (and 0 on failure)
-SIMD_FORCE_INLINE const char*	btConeTwistConstraint::serialize(void* dataBuffer, btSerializer* serializer) const
+///fills the dataBuffer and returns the struct name (and 0 on failure)
+SIMD_FORCE_INLINE const char* btConeTwistConstraint::serialize(void* dataBuffer, btSerializer* serializer) const
 {
-	btConeTwistConstraintData2* cone = (btConeTwistConstraintData2*) dataBuffer;
-	btTypedConstraint::serialize(&cone->m_typeConstraintData,serializer);
+	btConeTwistConstraintData2* cone = (btConeTwistConstraintData2*)dataBuffer;
+	btTypedConstraint::serialize(&cone->m_typeConstraintData, serializer);
 
 	m_rbAFrame.serialize(cone->m_rbAFrame);
 	m_rbBFrame.serialize(cone->m_rbBFrame);
-	
+
 	cone->m_swingSpan1 = m_swingSpan1;
 	cone->m_swingSpan2 = m_swingSpan2;
 	cone->m_twistSpan = m_twistSpan;
@@ -431,5 +420,4 @@ SIMD_FORCE_INLINE const char*	btConeTwistConstraint::serialize(void* dataBuffer,
 	return btConeTwistConstraintDataName;
 }
 
-
-#endif //BT_CONETWISTCONSTRAINT_H
+#endif  //BT_CONETWISTCONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btConstraintSolver.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btConstraintSolver.h
index 0491639f705..808433477c4 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btConstraintSolver.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btConstraintSolver.h
@@ -26,41 +26,33 @@ struct btContactSolverInfo;
 struct btBroadphaseProxy;
 class btIDebugDraw;
 class btStackAlloc;
-class	btDispatcher;
+class btDispatcher;
 /// btConstraintSolver provides solver interface
 
-
 enum btConstraintSolverType
 {
-	BT_SEQUENTIAL_IMPULSE_SOLVER=1,
-	BT_MLCP_SOLVER=2,
-	BT_NNCG_SOLVER=4,
-    BT_MULTIBODY_SOLVER=8,
+	BT_SEQUENTIAL_IMPULSE_SOLVER = 1,
+	BT_MLCP_SOLVER = 2,
+	BT_NNCG_SOLVER = 4,
+	BT_MULTIBODY_SOLVER = 8,
 };
 
 class btConstraintSolver
 {
-
 public:
-
 	virtual ~btConstraintSolver() {}
-	
-	virtual void prepareSolve (int /* numBodies */, int /* numManifolds */) {;}
+
+	virtual void prepareSolve(int /* numBodies */, int /* numManifolds */) { ; }
 
 	///solve a group of constraints
-	virtual btScalar solveGroup(btCollisionObject** bodies,int numBodies,btPersistentManifold** manifold,int numManifolds,btTypedConstraint** constraints,int numConstraints, const btContactSolverInfo& info,class btIDebugDraw* debugDrawer,btDispatcher* dispatcher) = 0;
+	virtual btScalar solveGroup(btCollisionObject** bodies, int numBodies, btPersistentManifold** manifold, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& info, class btIDebugDraw* debugDrawer, btDispatcher* dispatcher) = 0;
 
-	virtual void allSolved (const btContactSolverInfo& /* info */,class btIDebugDraw* /* debugDrawer */) {;}
+	virtual void allSolved(const btContactSolverInfo& /* info */, class btIDebugDraw* /* debugDrawer */) { ; }
 
 	///clear internal cached data and reset random seed
-	virtual	void	reset() = 0;
-
-	virtual btConstraintSolverType	getSolverType() const=0;
-
+	virtual void reset() = 0;
 
+	virtual btConstraintSolverType getSolverType() const = 0;
 };
 
-
-
-
-#endif //BT_CONSTRAINT_SOLVER_H
+#endif  //BT_CONSTRAINT_SOLVER_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btContactConstraint.cpp b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btContactConstraint.cpp
index 1098d0c96b6..4b22b2fff5b 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btContactConstraint.cpp
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btContactConstraint.cpp
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btContactConstraint.h"
 #include "BulletDynamics/Dynamics/btRigidBody.h"
 #include "LinearMath/btVector3.h"
@@ -22,44 +21,33 @@ subject to the following restrictions:
 #include "LinearMath/btMinMax.h"
 #include "BulletCollision/NarrowPhaseCollision/btManifoldPoint.h"
 
-
-
-btContactConstraint::btContactConstraint(btPersistentManifold* contactManifold,btRigidBody& rbA,btRigidBody& rbB)
-:btTypedConstraint(CONTACT_CONSTRAINT_TYPE,rbA,rbB),
-	m_contactManifold(*contactManifold)
+btContactConstraint::btContactConstraint(btPersistentManifold* contactManifold, btRigidBody& rbA, btRigidBody& rbB)
+	: btTypedConstraint(CONTACT_CONSTRAINT_TYPE, rbA, rbB),
+	  m_contactManifold(*contactManifold)
 {
-
 }
 
 btContactConstraint::~btContactConstraint()
 {
-
 }
 
-void	btContactConstraint::setContactManifold(btPersistentManifold* contactManifold)
+void btContactConstraint::setContactManifold(btPersistentManifold* contactManifold)
 {
 	m_contactManifold = *contactManifold;
 }
 
-void btContactConstraint::getInfo1 (btConstraintInfo1* info)
+void btContactConstraint::getInfo1(btConstraintInfo1* info)
 {
-
 }
 
-void btContactConstraint::getInfo2 (btConstraintInfo2* info)
+void btContactConstraint::getInfo2(btConstraintInfo2* info)
 {
-
 }
 
-void	btContactConstraint::buildJacobian()
+void btContactConstraint::buildJacobian()
 {
-
 }
 
-
-
-
-
 #include "btContactConstraint.h"
 #include "BulletDynamics/Dynamics/btRigidBody.h"
 #include "LinearMath/btVector3.h"
@@ -68,64 +56,59 @@ void	btContactConstraint::buildJacobian()
 #include "LinearMath/btMinMax.h"
 #include "BulletCollision/NarrowPhaseCollision/btManifoldPoint.h"
 
-
-
 //response  between two dynamic objects without friction and no restitution, assuming 0 penetration depth
 btScalar resolveSingleCollision(
-        btRigidBody* body1,
-        btCollisionObject* colObj2,
-		const btVector3& contactPositionWorld,
-		const btVector3& contactNormalOnB,
-        const btContactSolverInfo& solverInfo,
-		btScalar distance)
+	btRigidBody* body1,
+	btCollisionObject* colObj2,
+	const btVector3& contactPositionWorld,
+	const btVector3& contactNormalOnB,
+	const btContactSolverInfo& solverInfo,
+	btScalar distance)
 {
 	btRigidBody* body2 = btRigidBody::upcast(colObj2);
-    
-	
-    const btVector3& normal = contactNormalOnB;
 
-    btVector3 rel_pos1 = contactPositionWorld - body1->getWorldTransform().getOrigin(); 
-    btVector3 rel_pos2 = contactPositionWorld - colObj2->getWorldTransform().getOrigin();
-    
-    btVector3 vel1 = body1->getVelocityInLocalPoint(rel_pos1);
-	btVector3 vel2 = body2? body2->getVelocityInLocalPoint(rel_pos2) : btVector3(0,0,0);
-    btVector3 vel = vel1 - vel2;
-    btScalar rel_vel;
-    rel_vel = normal.dot(vel);
-    
-    btScalar combinedRestitution = 0.f;
-    btScalar restitution = combinedRestitution* -rel_vel;
+	const btVector3& normal = contactNormalOnB;
 
-    btScalar positionalError = solverInfo.m_erp *-distance /solverInfo.m_timeStep ;
-    btScalar velocityError = -(1.0f + restitution) * rel_vel;// * damping;
-	btScalar denom0 = body1->computeImpulseDenominator(contactPositionWorld,normal);
-	btScalar denom1 = body2? body2->computeImpulseDenominator(contactPositionWorld,normal) : 0.f;
+	btVector3 rel_pos1 = contactPositionWorld - body1->getWorldTransform().getOrigin();
+	btVector3 rel_pos2 = contactPositionWorld - colObj2->getWorldTransform().getOrigin();
+
+	btVector3 vel1 = body1->getVelocityInLocalPoint(rel_pos1);
+	btVector3 vel2 = body2 ? body2->getVelocityInLocalPoint(rel_pos2) : btVector3(0, 0, 0);
+	btVector3 vel = vel1 - vel2;
+	btScalar rel_vel;
+	rel_vel = normal.dot(vel);
+
+	btScalar combinedRestitution = 0.f;
+	btScalar restitution = combinedRestitution * -rel_vel;
+
+	btScalar positionalError = solverInfo.m_erp * -distance / solverInfo.m_timeStep;
+	btScalar velocityError = -(1.0f + restitution) * rel_vel;  // * damping;
+	btScalar denom0 = body1->computeImpulseDenominator(contactPositionWorld, normal);
+	btScalar denom1 = body2 ? body2->computeImpulseDenominator(contactPositionWorld, normal) : 0.f;
 	btScalar relaxation = 1.f;
-	btScalar jacDiagABInv = relaxation/(denom0+denom1);
+	btScalar jacDiagABInv = relaxation / (denom0 + denom1);
 
-    btScalar penetrationImpulse = positionalError * jacDiagABInv;
-    btScalar velocityImpulse = velocityError * jacDiagABInv;
+	btScalar penetrationImpulse = positionalError * jacDiagABInv;
+	btScalar velocityImpulse = velocityError * jacDiagABInv;
 
-    btScalar normalImpulse = penetrationImpulse+velocityImpulse;
-    normalImpulse = 0.f > normalImpulse ? 0.f: normalImpulse;
+	btScalar normalImpulse = penetrationImpulse + velocityImpulse;
+	normalImpulse = 0.f > normalImpulse ? 0.f : normalImpulse;
 
-	body1->applyImpulse(normal*(normalImpulse), rel_pos1);
-    if (body2)
-		body2->applyImpulse(-normal*(normalImpulse), rel_pos2);
-    
-    return normalImpulse;
+	body1->applyImpulse(normal * (normalImpulse), rel_pos1);
+	if (body2)
+		body2->applyImpulse(-normal * (normalImpulse), rel_pos2);
+
+	return normalImpulse;
 }
 
-
 //bilateral constraint between two dynamic objects
 void resolveSingleBilateral(btRigidBody& body1, const btVector3& pos1,
-                      btRigidBody& body2, const btVector3& pos2,
-                      btScalar distance, const btVector3& normal,btScalar& impulse ,btScalar timeStep)
+							btRigidBody& body2, const btVector3& pos2,
+							btScalar distance, const btVector3& normal, btScalar& impulse, btScalar timeStep)
 {
 	(void)timeStep;
 	(void)distance;
 
-
 	btScalar normalLenSqr = normal.length2();
 	btAssert(btFabs(normalLenSqr) < btScalar(1.1));
 	if (normalLenSqr > btScalar(1.1))
@@ -133,45 +116,38 @@ void resolveSingleBilateral(btRigidBody& body1, const btVector3& pos1,
 		impulse = btScalar(0.);
 		return;
 	}
-	btVector3 rel_pos1 = pos1 - body1.getCenterOfMassPosition(); 
+	btVector3 rel_pos1 = pos1 - body1.getCenterOfMassPosition();
 	btVector3 rel_pos2 = pos2 - body2.getCenterOfMassPosition();
 	//this jacobian entry could be re-used for all iterations
-	
+
 	btVector3 vel1 = body1.getVelocityInLocalPoint(rel_pos1);
 	btVector3 vel2 = body2.getVelocityInLocalPoint(rel_pos2);
 	btVector3 vel = vel1 - vel2;
-	
 
-	   btJacobianEntry jac(body1.getCenterOfMassTransform().getBasis().transpose(),
-		body2.getCenterOfMassTransform().getBasis().transpose(),
-		rel_pos1,rel_pos2,normal,body1.getInvInertiaDiagLocal(),body1.getInvMass(),
-		body2.getInvInertiaDiagLocal(),body2.getInvMass());
+	btJacobianEntry jac(body1.getCenterOfMassTransform().getBasis().transpose(),
+						body2.getCenterOfMassTransform().getBasis().transpose(),
+						rel_pos1, rel_pos2, normal, body1.getInvInertiaDiagLocal(), body1.getInvMass(),
+						body2.getInvInertiaDiagLocal(), body2.getInvMass());
 
 	btScalar jacDiagAB = jac.getDiagonal();
 	btScalar jacDiagABInv = btScalar(1.) / jacDiagAB;
-	
-	  btScalar rel_vel = jac.getRelativeVelocity(
+
+	btScalar rel_vel = jac.getRelativeVelocity(
 		body1.getLinearVelocity(),
 		body1.getCenterOfMassTransform().getBasis().transpose() * body1.getAngularVelocity(),
 		body2.getLinearVelocity(),
-		body2.getCenterOfMassTransform().getBasis().transpose() * body2.getAngularVelocity()); 
-
-
+		body2.getCenterOfMassTransform().getBasis().transpose() * body2.getAngularVelocity());
 
 	rel_vel = normal.dot(vel);
-	
+
 	//todo: move this into proper structure
 	btScalar contactDamping = btScalar(0.2);
 
 #ifdef ONLY_USE_LINEAR_MASS
 	btScalar massTerm = btScalar(1.) / (body1.getInvMass() + body2.getInvMass());
-	impulse = - contactDamping * rel_vel * massTerm;
-#else	
+	impulse = -contactDamping * rel_vel * massTerm;
+#else
 	btScalar velocityImpulse = -contactDamping * rel_vel * jacDiagABInv;
 	impulse = velocityImpulse;
 #endif
 }
-
-
-
-
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btContactConstraint.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btContactConstraint.h
index adb2268353e..255489be991 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btContactConstraint.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btContactConstraint.h
@@ -22,20 +22,17 @@ subject to the following restrictions:
 #include "BulletCollision/NarrowPhaseCollision/btPersistentManifold.h"
 
 ///btContactConstraint can be automatically created to solve contact constraints using the unified btTypedConstraint interface
-ATTRIBUTE_ALIGNED16(class) btContactConstraint : public btTypedConstraint
+ATTRIBUTE_ALIGNED16(class)
+btContactConstraint : public btTypedConstraint
 {
 protected:
-
 	btPersistentManifold m_contactManifold;
 
 protected:
-
-
-	btContactConstraint(btPersistentManifold* contactManifold,btRigidBody& rbA,btRigidBody& rbB);
+	btContactConstraint(btPersistentManifold * contactManifold, btRigidBody & rbA, btRigidBody & rbB);
 
 public:
-
-	void	setContactManifold(btPersistentManifold* contactManifold);
+	void setContactManifold(btPersistentManifold * contactManifold);
 
 	btPersistentManifold* getContactManifold()
 	{
@@ -49,25 +46,20 @@ public:
 
 	virtual ~btContactConstraint();
 
-	virtual void getInfo1 (btConstraintInfo1* info);
+	virtual void getInfo1(btConstraintInfo1 * info);
 
-	virtual void getInfo2 (btConstraintInfo2* info);
+	virtual void getInfo2(btConstraintInfo2 * info);
 
 	///obsolete methods
-	virtual void	buildJacobian();
-
-
+	virtual void buildJacobian();
 };
 
 ///very basic collision resolution without friction
-btScalar resolveSingleCollision(btRigidBody* body1, class btCollisionObject* colObj2, const btVector3& contactPositionWorld,const btVector3& contactNormalOnB, const struct btContactSolverInfo& solverInfo,btScalar distance);
-
+btScalar resolveSingleCollision(btRigidBody* body1, class btCollisionObject* colObj2, const btVector3& contactPositionWorld, const btVector3& contactNormalOnB, const struct btContactSolverInfo& solverInfo, btScalar distance);
 
 ///resolveSingleBilateral is an obsolete methods used for vehicle friction between two dynamic objects
 void resolveSingleBilateral(btRigidBody& body1, const btVector3& pos1,
-                      btRigidBody& body2, const btVector3& pos2,
-                      btScalar distance, const btVector3& normal,btScalar& impulse ,btScalar timeStep);
+							btRigidBody& body2, const btVector3& pos2,
+							btScalar distance, const btVector3& normal, btScalar& impulse, btScalar timeStep);
 
-
-
-#endif //BT_CONTACT_CONSTRAINT_H
+#endif  //BT_CONTACT_CONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btContactSolverInfo.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btContactSolverInfo.h
index 93865cbc592..89f8db8b1a2 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btContactSolverInfo.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btContactSolverInfo.h
@@ -18,7 +18,7 @@ subject to the following restrictions:
 
 #include "LinearMath/btScalar.h"
 
-enum	btSolverMode
+enum btSolverMode
 {
 	SOLVER_RANDMIZE_ORDER = 1,
 	SOLVER_FRICTION_SEPARATE = 2,
@@ -35,134 +35,129 @@ enum	btSolverMode
 
 struct btContactSolverInfoData
 {
-	
+	btScalar m_tau;
+	btScalar m_damping;  //global non-contact constraint damping, can be locally overridden by constraints during 'getInfo2'.
+	btScalar m_friction;
+	btScalar m_timeStep;
+	btScalar m_restitution;
+	int m_numIterations;
+	btScalar m_maxErrorReduction;
+	btScalar m_sor;          //successive over-relaxation term
+	btScalar m_erp;          //error reduction for non-contact constraints
+	btScalar m_erp2;         //error reduction for contact constraints
+	btScalar m_globalCfm;    //constraint force mixing for contacts and non-contacts
+	btScalar m_frictionERP;  //error reduction for friction constraints
+	btScalar m_frictionCFM;  //constraint force mixing for friction constraints
 
-	btScalar	m_tau;
-	btScalar	m_damping;//global non-contact constraint damping, can be locally overridden by constraints during 'getInfo2'.
-	btScalar	m_friction;
-	btScalar	m_timeStep;
-	btScalar	m_restitution;
-	int		m_numIterations;
-	btScalar	m_maxErrorReduction;
-	btScalar	m_sor;//successive over-relaxation term
-	btScalar	m_erp;//error reduction for non-contact constraints
-	btScalar	m_erp2;//error reduction for contact constraints
-	btScalar	m_globalCfm;//constraint force mixing for contacts and non-contacts
-	btScalar	m_frictionERP;//error reduction for friction constraints
-	btScalar	m_frictionCFM;//constraint force mixing for friction constraints
-
-	int			m_splitImpulse;
-	btScalar	m_splitImpulsePenetrationThreshold;
-	btScalar	m_splitImpulseTurnErp;
-	btScalar	m_linearSlop;
-	btScalar	m_warmstartingFactor;
-
-	int			m_solverMode;
-	int	m_restingContactRestitutionThreshold;
-	int			m_minimumSolverBatchSize;
-	btScalar	m_maxGyroscopicForce;
-	btScalar	m_singleAxisRollingFrictionThreshold;
-	btScalar	m_leastSquaresResidualThreshold;
-	btScalar	m_restitutionVelocityThreshold;
+	int m_splitImpulse;
+	btScalar m_splitImpulsePenetrationThreshold;
+	btScalar m_splitImpulseTurnErp;
+	btScalar m_linearSlop;
+	btScalar m_warmstartingFactor;
 
+	int m_solverMode;
+	int m_restingContactRestitutionThreshold;
+	int m_minimumSolverBatchSize;
+	btScalar m_maxGyroscopicForce;
+	btScalar m_singleAxisRollingFrictionThreshold;
+	btScalar m_leastSquaresResidualThreshold;
+	btScalar m_restitutionVelocityThreshold;
+	bool m_jointFeedbackInWorldSpace;
+	bool m_jointFeedbackInJointFrame;
 };
 
 struct btContactSolverInfo : public btContactSolverInfoData
 {
-
-	
-
 	inline btContactSolverInfo()
 	{
 		m_tau = btScalar(0.6);
 		m_damping = btScalar(1.0);
 		m_friction = btScalar(0.3);
-		m_timeStep = btScalar(1.f/60.f);
+		m_timeStep = btScalar(1.f / 60.f);
 		m_restitution = btScalar(0.);
 		m_maxErrorReduction = btScalar(20.);
 		m_numIterations = 10;
 		m_erp = btScalar(0.2);
 		m_erp2 = btScalar(0.2);
 		m_globalCfm = btScalar(0.);
-		m_frictionERP = btScalar(0.2);//positional friction 'anchors' are disabled by default
+		m_frictionERP = btScalar(0.2);  //positional friction 'anchors' are disabled by default
 		m_frictionCFM = btScalar(0.);
 		m_sor = btScalar(1.);
 		m_splitImpulse = true;
 		m_splitImpulsePenetrationThreshold = -.04f;
 		m_splitImpulseTurnErp = 0.1f;
 		m_linearSlop = btScalar(0.0);
-		m_warmstartingFactor=btScalar(0.85);
+		m_warmstartingFactor = btScalar(0.85);
 		//m_solverMode =  SOLVER_USE_WARMSTARTING |  SOLVER_SIMD | SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION|SOLVER_USE_2_FRICTION_DIRECTIONS|SOLVER_ENABLE_FRICTION_DIRECTION_CACHING;// | SOLVER_RANDMIZE_ORDER;
-		m_solverMode = SOLVER_USE_WARMSTARTING | SOLVER_SIMD;// | SOLVER_RANDMIZE_ORDER;
-		m_restingContactRestitutionThreshold = 2;//unused as of 2.81
-		m_minimumSolverBatchSize = 128; //try to combine islands until the amount of constraints reaches this limit
-		m_maxGyroscopicForce = 100.f; ///it is only used for 'explicit' version of gyroscopic force
-		m_singleAxisRollingFrictionThreshold = 1e30f;///if the velocity is above this threshold, it will use a single constraint row (axis), otherwise 3 rows.
+		m_solverMode = SOLVER_USE_WARMSTARTING | SOLVER_SIMD;  // | SOLVER_RANDMIZE_ORDER;
+		m_restingContactRestitutionThreshold = 2;              //unused as of 2.81
+		m_minimumSolverBatchSize = 128;                        //try to combine islands until the amount of constraints reaches this limit
+		m_maxGyroscopicForce = 100.f;                          ///it is only used for 'explicit' version of gyroscopic force
+		m_singleAxisRollingFrictionThreshold = 1e30f;          ///if the velocity is above this threshold, it will use a single constraint row (axis), otherwise 3 rows.
 		m_leastSquaresResidualThreshold = 0.f;
-		m_restitutionVelocityThreshold = 0.2f;//if the relative velocity is below this threshold, there is zero restitution
+		m_restitutionVelocityThreshold = 0.2f;  //if the relative velocity is below this threshold, there is zero restitution
+		m_jointFeedbackInWorldSpace = false;
+		m_jointFeedbackInJointFrame = false;
 	}
 };
 
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
 struct btContactSolverInfoDoubleData
 {
-	double		m_tau;
-	double		m_damping;//global non-contact constraint damping, can be locally overridden by constraints during 'getInfo2'.
-	double		m_friction;
-	double		m_timeStep;
-	double		m_restitution;
-	double		m_maxErrorReduction;
-	double		m_sor;
-	double		m_erp;//used as Baumgarte factor
-	double		m_erp2;//used in Split Impulse
-	double		m_globalCfm;//constraint force mixing
-	double		m_splitImpulsePenetrationThreshold;
-	double		m_splitImpulseTurnErp;
-	double		m_linearSlop;
-	double		m_warmstartingFactor;
-	double		m_maxGyroscopicForce;///it is only used for 'explicit' version of gyroscopic force
-	double		m_singleAxisRollingFrictionThreshold;
-
-	int			m_numIterations;
-	int			m_solverMode;
-	int			m_restingContactRestitutionThreshold;
-	int			m_minimumSolverBatchSize;
-	int			m_splitImpulse;
-	char		m_padding[4];
+	double m_tau;
+	double m_damping;  //global non-contact constraint damping, can be locally overridden by constraints during 'getInfo2'.
+	double m_friction;
+	double m_timeStep;
+	double m_restitution;
+	double m_maxErrorReduction;
+	double m_sor;
+	double m_erp;        //used as Baumgarte factor
+	double m_erp2;       //used in Split Impulse
+	double m_globalCfm;  //constraint force mixing
+	double m_splitImpulsePenetrationThreshold;
+	double m_splitImpulseTurnErp;
+	double m_linearSlop;
+	double m_warmstartingFactor;
+	double m_maxGyroscopicForce;  ///it is only used for 'explicit' version of gyroscopic force
+	double m_singleAxisRollingFrictionThreshold;
 
+	int m_numIterations;
+	int m_solverMode;
+	int m_restingContactRestitutionThreshold;
+	int m_minimumSolverBatchSize;
+	int m_splitImpulse;
+	char m_padding[4];
 };
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
 struct btContactSolverInfoFloatData
 {
-	float		m_tau;
-	float		m_damping;//global non-contact constraint damping, can be locally overridden by constraints during 'getInfo2'.
-	float		m_friction;
-	float		m_timeStep;
+	float m_tau;
+	float m_damping;  //global non-contact constraint damping, can be locally overridden by constraints during 'getInfo2'.
+	float m_friction;
+	float m_timeStep;
 
-	float		m_restitution;
-	float		m_maxErrorReduction;
-	float		m_sor;
-	float		m_erp;//used as Baumgarte factor
+	float m_restitution;
+	float m_maxErrorReduction;
+	float m_sor;
+	float m_erp;  //used as Baumgarte factor
 
-	float		m_erp2;//used in Split Impulse
-	float		m_globalCfm;//constraint force mixing
-	float		m_splitImpulsePenetrationThreshold;
-	float		m_splitImpulseTurnErp;
+	float m_erp2;       //used in Split Impulse
+	float m_globalCfm;  //constraint force mixing
+	float m_splitImpulsePenetrationThreshold;
+	float m_splitImpulseTurnErp;
 
-	float		m_linearSlop;
-	float		m_warmstartingFactor;
-	float		m_maxGyroscopicForce;
-	float		m_singleAxisRollingFrictionThreshold;
+	float m_linearSlop;
+	float m_warmstartingFactor;
+	float m_maxGyroscopicForce;
+	float m_singleAxisRollingFrictionThreshold;
 
-	int			m_numIterations;
-	int			m_solverMode;
-	int			m_restingContactRestitutionThreshold;
-	int			m_minimumSolverBatchSize;
+	int m_numIterations;
+	int m_solverMode;
+	int m_restingContactRestitutionThreshold;
+	int m_minimumSolverBatchSize;
 
-	int			m_splitImpulse;
-	char		m_padding[4];
+	int m_splitImpulse;
+	char m_padding[4];
 };
 
-
-
-#endif //BT_CONTACT_SOLVER_INFO
+#endif  //BT_CONTACT_SOLVER_INFO
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btFixedConstraint.cpp b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btFixedConstraint.cpp
index 75d81cc08c2..bba102d9052 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btFixedConstraint.cpp
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btFixedConstraint.cpp
@@ -13,25 +13,20 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btFixedConstraint.h"
 #include "BulletDynamics/Dynamics/btRigidBody.h"
 #include "LinearMath/btTransformUtil.h"
 #include <new>
 
-
-btFixedConstraint::btFixedConstraint(btRigidBody& rbA,btRigidBody& rbB, const btTransform& frameInA,const btTransform& frameInB)
-:btGeneric6DofSpring2Constraint(rbA,rbB,frameInA,frameInB)
+btFixedConstraint::btFixedConstraint(btRigidBody& rbA, btRigidBody& rbB, const btTransform& frameInA, const btTransform& frameInB)
+	: btGeneric6DofSpring2Constraint(rbA, rbB, frameInA, frameInB)
 {
-	setAngularLowerLimit(btVector3(0,0,0));
-	setAngularUpperLimit(btVector3(0,0,0));
-	setLinearLowerLimit(btVector3(0,0,0));
-	setLinearUpperLimit(btVector3(0,0,0));
+	setAngularLowerLimit(btVector3(0, 0, 0));
+	setAngularUpperLimit(btVector3(0, 0, 0));
+	setLinearLowerLimit(btVector3(0, 0, 0));
+	setLinearUpperLimit(btVector3(0, 0, 0));
 }
 
-
-
-
-btFixedConstraint::~btFixedConstraint ()
+btFixedConstraint::~btFixedConstraint()
 {
 }
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btFixedConstraint.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btFixedConstraint.h
index bff2008b283..6d474ea81de 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btFixedConstraint.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btFixedConstraint.h
@@ -18,16 +18,13 @@ subject to the following restrictions:
 
 #include "btGeneric6DofSpring2Constraint.h"
 
-
-ATTRIBUTE_ALIGNED16(class) btFixedConstraint : public btGeneric6DofSpring2Constraint
+ATTRIBUTE_ALIGNED16(class)
+btFixedConstraint : public btGeneric6DofSpring2Constraint
 {
-
 public:
-	btFixedConstraint(btRigidBody& rbA,btRigidBody& rbB, const btTransform& frameInA,const btTransform& frameInB);
+	btFixedConstraint(btRigidBody & rbA, btRigidBody & rbB, const btTransform& frameInA, const btTransform& frameInB);
 
-	
 	virtual ~btFixedConstraint();
-
 };
 
-#endif //BT_FIXED_CONSTRAINT_H
+#endif  //BT_FIXED_CONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGearConstraint.cpp b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGearConstraint.cpp
index bcd457b6731..7535c52c05f 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGearConstraint.cpp
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGearConstraint.cpp
@@ -17,38 +17,36 @@ subject to the following restrictions:
 
 #include "btGearConstraint.h"
 
-btGearConstraint::btGearConstraint(btRigidBody& rbA, btRigidBody& rbB, const btVector3& axisInA,const btVector3& axisInB, btScalar ratio)
-:btTypedConstraint(GEAR_CONSTRAINT_TYPE,rbA,rbB),
-m_axisInA(axisInA),
-m_axisInB(axisInB),
-m_ratio(ratio)
+btGearConstraint::btGearConstraint(btRigidBody& rbA, btRigidBody& rbB, const btVector3& axisInA, const btVector3& axisInB, btScalar ratio)
+	: btTypedConstraint(GEAR_CONSTRAINT_TYPE, rbA, rbB),
+	  m_axisInA(axisInA),
+	  m_axisInB(axisInB),
+	  m_ratio(ratio)
 {
 }
 
-btGearConstraint::~btGearConstraint ()
+btGearConstraint::~btGearConstraint()
 {
 }
 
-void btGearConstraint::getInfo1 (btConstraintInfo1* info)
+void btGearConstraint::getInfo1(btConstraintInfo1* info)
 {
 	info->m_numConstraintRows = 1;
 	info->nub = 1;
 }
 
-void btGearConstraint::getInfo2 (btConstraintInfo2* info)
+void btGearConstraint::getInfo2(btConstraintInfo2* info)
 {
 	btVector3 globalAxisA, globalAxisB;
 
-	globalAxisA = m_rbA.getWorldTransform().getBasis()*this->m_axisInA;
-	globalAxisB = m_rbB.getWorldTransform().getBasis()*this->m_axisInB;
+	globalAxisA = m_rbA.getWorldTransform().getBasis() * this->m_axisInA;
+	globalAxisB = m_rbB.getWorldTransform().getBasis() * this->m_axisInB;
 
 	info->m_J1angularAxis[0] = globalAxisA[0];
 	info->m_J1angularAxis[1] = globalAxisA[1];
 	info->m_J1angularAxis[2] = globalAxisA[2];
 
-	info->m_J2angularAxis[0] = m_ratio*globalAxisB[0];
-	info->m_J2angularAxis[1] = m_ratio*globalAxisB[1];
-	info->m_J2angularAxis[2] = m_ratio*globalAxisB[2];
-
+	info->m_J2angularAxis[0] = m_ratio * globalAxisB[0];
+	info->m_J2angularAxis[1] = m_ratio * globalAxisB[1];
+	info->m_J2angularAxis[2] = m_ratio * globalAxisB[2];
 }
-
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGearConstraint.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGearConstraint.h
index e4613455a2b..64b15dfbce8 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGearConstraint.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGearConstraint.h
@@ -13,45 +13,40 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #ifndef BT_GEAR_CONSTRAINT_H
 #define BT_GEAR_CONSTRAINT_H
 
 #include "BulletDynamics/ConstraintSolver/btTypedConstraint.h"
 
-
 #ifdef BT_USE_DOUBLE_PRECISION
-#define btGearConstraintData	btGearConstraintDoubleData
-#define btGearConstraintDataName	"btGearConstraintDoubleData"
+#define btGearConstraintData btGearConstraintDoubleData
+#define btGearConstraintDataName "btGearConstraintDoubleData"
 #else
-#define btGearConstraintData	btGearConstraintFloatData
-#define btGearConstraintDataName	"btGearConstraintFloatData"
-#endif //BT_USE_DOUBLE_PRECISION
-
-
+#define btGearConstraintData btGearConstraintFloatData
+#define btGearConstraintDataName "btGearConstraintFloatData"
+#endif  //BT_USE_DOUBLE_PRECISION
 
 ///The btGeatConstraint will couple the angular velocity for two bodies around given local axis and ratio.
 ///See Bullet/Demos/ConstraintDemo for an example use.
 class btGearConstraint : public btTypedConstraint
 {
 protected:
-	btVector3	m_axisInA;
-	btVector3	m_axisInB;
-	bool		m_useFrameA;
-	btScalar	m_ratio;
+	btVector3 m_axisInA;
+	btVector3 m_axisInB;
+	bool m_useFrameA;
+	btScalar m_ratio;
 
 public:
-	btGearConstraint(btRigidBody& rbA, btRigidBody& rbB, const btVector3& axisInA,const btVector3& axisInB, btScalar ratio=1.f);
-	virtual ~btGearConstraint ();
+	btGearConstraint(btRigidBody& rbA, btRigidBody& rbB, const btVector3& axisInA, const btVector3& axisInB, btScalar ratio = 1.f);
+	virtual ~btGearConstraint();
 
 	///internal method used by the constraint solver, don't use them directly
-	virtual void getInfo1 (btConstraintInfo1* info);
+	virtual void getInfo1(btConstraintInfo1* info);
 
 	///internal method used by the constraint solver, don't use them directly
-	virtual void getInfo2 (btConstraintInfo2* info);
+	virtual void getInfo2(btConstraintInfo2* info);
 
-	void setAxisA(btVector3& axisA) 
+	void setAxisA(btVector3& axisA)
 	{
 		m_axisInA = axisA;
 	}
@@ -76,68 +71,64 @@ public:
 		return m_ratio;
 	}
 
-
-	virtual	void	setParam(int num, btScalar value, int axis = -1) 
+	virtual void setParam(int num, btScalar value, int axis = -1)
 	{
-		(void) num;
-		(void) value;
-		(void) axis;
+		(void)num;
+		(void)value;
+		(void)axis;
 		btAssert(0);
 	}
 
 	///return the local value of parameter
-	virtual	btScalar getParam(int num, int axis = -1) const 
-	{ 
-		(void) num;
-		(void) axis;
+	virtual btScalar getParam(int num, int axis = -1) const
+	{
+		(void)num;
+		(void)axis;
 		btAssert(0);
 		return 0.f;
 	}
 
-	virtual	int	calculateSerializeBufferSize() const;
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 };
 
-
-
-
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
 struct btGearConstraintFloatData
 {
-	btTypedConstraintFloatData	m_typeConstraintData;
+	btTypedConstraintFloatData m_typeConstraintData;
 
-	btVector3FloatData			m_axisInA;
-	btVector3FloatData			m_axisInB;
+	btVector3FloatData m_axisInA;
+	btVector3FloatData m_axisInB;
 
-	float							m_ratio;
-	char							m_padding[4];
+	float m_ratio;
+	char m_padding[4];
 };
 
 struct btGearConstraintDoubleData
 {
-	btTypedConstraintDoubleData	m_typeConstraintData;
+	btTypedConstraintDoubleData m_typeConstraintData;
 
-	btVector3DoubleData			m_axisInA;
-	btVector3DoubleData			m_axisInB;
+	btVector3DoubleData m_axisInA;
+	btVector3DoubleData m_axisInB;
 
-	double						m_ratio;
+	double m_ratio;
 };
 
-SIMD_FORCE_INLINE	int	btGearConstraint::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btGearConstraint::calculateSerializeBufferSize() const
 {
 	return sizeof(btGearConstraintData);
 }
 
-	///fills the dataBuffer and returns the struct name (and 0 on failure)
-SIMD_FORCE_INLINE	const char*	btGearConstraint::serialize(void* dataBuffer, btSerializer* serializer) const
+///fills the dataBuffer and returns the struct name (and 0 on failure)
+SIMD_FORCE_INLINE const char* btGearConstraint::serialize(void* dataBuffer, btSerializer* serializer) const
 {
 	btGearConstraintData* gear = (btGearConstraintData*)dataBuffer;
-	btTypedConstraint::serialize(&gear->m_typeConstraintData,serializer);
+	btTypedConstraint::serialize(&gear->m_typeConstraintData, serializer);
 
-	m_axisInA.serialize( gear->m_axisInA );
-	m_axisInB.serialize( gear->m_axisInB );
+	m_axisInA.serialize(gear->m_axisInA);
+	m_axisInB.serialize(gear->m_axisInB);
 
 	gear->m_ratio = m_ratio;
 
@@ -152,9 +143,4 @@ SIMD_FORCE_INLINE	const char*	btGearConstraint::serialize(void* dataBuffer, btSe
 	return btGearConstraintDataName;
 }
 
-
-
-
-
-
-#endif //BT_GEAR_CONSTRAINT_H
+#endif  //BT_GEAR_CONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofConstraint.cpp b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofConstraint.cpp
index c38b8353f06..1f542035324 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofConstraint.cpp
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofConstraint.cpp
@@ -25,83 +25,61 @@ http://gimpact.sf.net
 #include "LinearMath/btTransformUtil.h"
 #include <new>
 
-
-
 #define D6_USE_OBSOLETE_METHOD false
 #define D6_USE_FRAME_OFFSET true
 
-
-
-
-
-
 btGeneric6DofConstraint::btGeneric6DofConstraint(btRigidBody& rbA, btRigidBody& rbB, const btTransform& frameInA, const btTransform& frameInB, bool useLinearReferenceFrameA)
-: btTypedConstraint(D6_CONSTRAINT_TYPE, rbA, rbB)
-, m_frameInA(frameInA)
-, m_frameInB(frameInB),
-m_useLinearReferenceFrameA(useLinearReferenceFrameA),
-m_useOffsetForConstraintFrame(D6_USE_FRAME_OFFSET),
-m_flags(0),
-m_useSolveConstraintObsolete(D6_USE_OBSOLETE_METHOD)
+	: btTypedConstraint(D6_CONSTRAINT_TYPE, rbA, rbB), m_frameInA(frameInA), m_frameInB(frameInB), m_useLinearReferenceFrameA(useLinearReferenceFrameA), m_useOffsetForConstraintFrame(D6_USE_FRAME_OFFSET), m_flags(0), m_useSolveConstraintObsolete(D6_USE_OBSOLETE_METHOD)
 {
 	calculateTransforms();
 }
 
-
-
 btGeneric6DofConstraint::btGeneric6DofConstraint(btRigidBody& rbB, const btTransform& frameInB, bool useLinearReferenceFrameB)
-        : btTypedConstraint(D6_CONSTRAINT_TYPE, getFixedBody(), rbB),
-		m_frameInB(frameInB),
-		m_useLinearReferenceFrameA(useLinearReferenceFrameB),
-		m_useOffsetForConstraintFrame(D6_USE_FRAME_OFFSET),
-		m_flags(0),
-		m_useSolveConstraintObsolete(false)
+	: btTypedConstraint(D6_CONSTRAINT_TYPE, getFixedBody(), rbB),
+	  m_frameInB(frameInB),
+	  m_useLinearReferenceFrameA(useLinearReferenceFrameB),
+	  m_useOffsetForConstraintFrame(D6_USE_FRAME_OFFSET),
+	  m_flags(0),
+	  m_useSolveConstraintObsolete(false)
 {
 	///not providing rigidbody A means implicitly using worldspace for body A
 	m_frameInA = rbB.getCenterOfMassTransform() * m_frameInB;
 	calculateTransforms();
 }
 
-
-
-
 #define GENERIC_D6_DISABLE_WARMSTARTING 1
 
-
-
 btScalar btGetMatrixElem(const btMatrix3x3& mat, int index);
 btScalar btGetMatrixElem(const btMatrix3x3& mat, int index)
 {
-	int i = index%3;
-	int j = index/3;
+	int i = index % 3;
+	int j = index / 3;
 	return mat[i][j];
 }
 
-
-
 ///MatrixToEulerXYZ from http://www.geometrictools.com/LibFoundation/Mathematics/Wm4Matrix3.inl.html
-bool	matrixToEulerXYZ(const btMatrix3x3& mat,btVector3& xyz);
-bool	matrixToEulerXYZ(const btMatrix3x3& mat,btVector3& xyz)
+bool matrixToEulerXYZ(const btMatrix3x3& mat, btVector3& xyz);
+bool matrixToEulerXYZ(const btMatrix3x3& mat, btVector3& xyz)
 {
 	//	// rot =  cy*cz          -cy*sz           sy
 	//	//        cz*sx*sy+cx*sz  cx*cz-sx*sy*sz -cy*sx
 	//	//       -cx*cz*sy+sx*sz  cz*sx+cx*sy*sz  cx*cy
 	//
 
-	btScalar fi = btGetMatrixElem(mat,2);
+	btScalar fi = btGetMatrixElem(mat, 2);
 	if (fi < btScalar(1.0f))
 	{
 		if (fi > btScalar(-1.0f))
 		{
-			xyz[0] = btAtan2(-btGetMatrixElem(mat,5),btGetMatrixElem(mat,8));
-			xyz[1] = btAsin(btGetMatrixElem(mat,2));
-			xyz[2] = btAtan2(-btGetMatrixElem(mat,1),btGetMatrixElem(mat,0));
+			xyz[0] = btAtan2(-btGetMatrixElem(mat, 5), btGetMatrixElem(mat, 8));
+			xyz[1] = btAsin(btGetMatrixElem(mat, 2));
+			xyz[2] = btAtan2(-btGetMatrixElem(mat, 1), btGetMatrixElem(mat, 0));
 			return true;
 		}
 		else
 		{
 			// WARNING.  Not unique.  XA - ZA = -atan2(r10,r11)
-			xyz[0] = -btAtan2(btGetMatrixElem(mat,3),btGetMatrixElem(mat,4));
+			xyz[0] = -btAtan2(btGetMatrixElem(mat, 3), btGetMatrixElem(mat, 4));
 			xyz[1] = -SIMD_HALF_PI;
 			xyz[2] = btScalar(0.0);
 			return false;
@@ -110,7 +88,7 @@ bool	matrixToEulerXYZ(const btMatrix3x3& mat,btVector3& xyz)
 	else
 	{
 		// WARNING.  Not unique.  XAngle + ZAngle = atan2(r10,r11)
-		xyz[0] = btAtan2(btGetMatrixElem(mat,3),btGetMatrixElem(mat,4));
+		xyz[0] = btAtan2(btGetMatrixElem(mat, 3), btGetMatrixElem(mat, 4));
 		xyz[1] = SIMD_HALF_PI;
 		xyz[2] = 0.0;
 	}
@@ -121,52 +99,49 @@ bool	matrixToEulerXYZ(const btMatrix3x3& mat,btVector3& xyz)
 
 int btRotationalLimitMotor::testLimitValue(btScalar test_value)
 {
-	if(m_loLimit>m_hiLimit)
+	if (m_loLimit > m_hiLimit)
 	{
-		m_currentLimit = 0;//Free from violation
+		m_currentLimit = 0;  //Free from violation
 		return 0;
 	}
 	if (test_value < m_loLimit)
 	{
-		m_currentLimit = 1;//low limit violation
-		m_currentLimitError =  test_value - m_loLimit;
-		if(m_currentLimitError>SIMD_PI) 
-			m_currentLimitError-=SIMD_2_PI;
-		else if(m_currentLimitError<-SIMD_PI) 
-			m_currentLimitError+=SIMD_2_PI;
+		m_currentLimit = 1;  //low limit violation
+		m_currentLimitError = test_value - m_loLimit;
+		if (m_currentLimitError > SIMD_PI)
+			m_currentLimitError -= SIMD_2_PI;
+		else if (m_currentLimitError < -SIMD_PI)
+			m_currentLimitError += SIMD_2_PI;
 		return 1;
 	}
-	else if (test_value> m_hiLimit)
+	else if (test_value > m_hiLimit)
 	{
-		m_currentLimit = 2;//High limit violation
+		m_currentLimit = 2;  //High limit violation
 		m_currentLimitError = test_value - m_hiLimit;
-		if(m_currentLimitError>SIMD_PI) 
-			m_currentLimitError-=SIMD_2_PI;
-		else if(m_currentLimitError<-SIMD_PI) 
-			m_currentLimitError+=SIMD_2_PI;
+		if (m_currentLimitError > SIMD_PI)
+			m_currentLimitError -= SIMD_2_PI;
+		else if (m_currentLimitError < -SIMD_PI)
+			m_currentLimitError += SIMD_2_PI;
 		return 2;
 	};
 
-	m_currentLimit = 0;//Free from violation
+	m_currentLimit = 0;  //Free from violation
 	return 0;
-
 }
 
-
-
 btScalar btRotationalLimitMotor::solveAngularLimits(
-	btScalar timeStep,btVector3& axis,btScalar jacDiagABInv,
-	btRigidBody * body0, btRigidBody * body1 )
+	btScalar timeStep, btVector3& axis, btScalar jacDiagABInv,
+	btRigidBody* body0, btRigidBody* body1)
 {
-	if (needApplyTorques()==false) return 0.0f;
+	if (needApplyTorques() == false) return 0.0f;
 
 	btScalar target_velocity = m_targetVelocity;
 	btScalar maxMotorForce = m_maxMotorForce;
 
 	//current error correction
-	if (m_currentLimit!=0)
+	if (m_currentLimit != 0)
 	{
-		target_velocity = -m_stopERP*m_currentLimitError/(timeStep);
+		target_velocity = -m_stopERP * m_currentLimitError / (timeStep);
 		maxMotorForce = m_maxLimitForce;
 	}
 
@@ -178,42 +153,37 @@ btScalar btRotationalLimitMotor::solveAngularLimits(
 	btVector3 angVelB = body1->getAngularVelocity();
 
 	btVector3 vel_diff;
-	vel_diff = angVelA-angVelB;
-
-
+	vel_diff = angVelA - angVelB;
 
 	btScalar rel_vel = axis.dot(vel_diff);
 
 	// correction velocity
-	btScalar motor_relvel = m_limitSoftness*(target_velocity  - m_damping*rel_vel);
+	btScalar motor_relvel = m_limitSoftness * (target_velocity - m_damping * rel_vel);
 
-
-	if ( motor_relvel < SIMD_EPSILON && motor_relvel > -SIMD_EPSILON  )
+	if (motor_relvel < SIMD_EPSILON && motor_relvel > -SIMD_EPSILON)
 	{
-		return 0.0f;//no need for applying force
+		return 0.0f;  //no need for applying force
 	}
 
-
 	// correction impulse
-	btScalar unclippedMotorImpulse = (1+m_bounce)*motor_relvel*jacDiagABInv;
+	btScalar unclippedMotorImpulse = (1 + m_bounce) * motor_relvel * jacDiagABInv;
 
 	// clip correction impulse
 	btScalar clippedMotorImpulse;
 
 	///@todo: should clip against accumulated impulse
-	if (unclippedMotorImpulse>0.0f)
+	if (unclippedMotorImpulse > 0.0f)
 	{
-		clippedMotorImpulse =  unclippedMotorImpulse > maxMotorForce? maxMotorForce: unclippedMotorImpulse;
+		clippedMotorImpulse = unclippedMotorImpulse > maxMotorForce ? maxMotorForce : unclippedMotorImpulse;
 	}
 	else
 	{
-		clippedMotorImpulse =  unclippedMotorImpulse < -maxMotorForce ? -maxMotorForce: unclippedMotorImpulse;
+		clippedMotorImpulse = unclippedMotorImpulse < -maxMotorForce ? -maxMotorForce : unclippedMotorImpulse;
 	}
 
-
 	// sort with accumulated impulses
-	btScalar	lo = btScalar(-BT_LARGE_FLOAT);
-	btScalar	hi = btScalar(BT_LARGE_FLOAT);
+	btScalar lo = btScalar(-BT_LARGE_FLOAT);
+	btScalar hi = btScalar(BT_LARGE_FLOAT);
 
 	btScalar oldaccumImpulse = m_accumulatedImpulse;
 	btScalar sum = oldaccumImpulse + clippedMotorImpulse;
@@ -227,59 +197,50 @@ btScalar btRotationalLimitMotor::solveAngularLimits(
 	body1->applyTorqueImpulse(-motorImp);
 
 	return clippedMotorImpulse;
-
-
 }
 
 //////////////////////////// End btRotationalLimitMotor ////////////////////////////////////
 
-
-
-
 //////////////////////////// btTranslationalLimitMotor ////////////////////////////////////
 
-
 int btTranslationalLimitMotor::testLimitValue(int limitIndex, btScalar test_value)
 {
 	btScalar loLimit = m_lowerLimit[limitIndex];
 	btScalar hiLimit = m_upperLimit[limitIndex];
-	if(loLimit > hiLimit)
+	if (loLimit > hiLimit)
 	{
-		m_currentLimit[limitIndex] = 0;//Free from violation
+		m_currentLimit[limitIndex] = 0;  //Free from violation
 		m_currentLimitError[limitIndex] = btScalar(0.f);
 		return 0;
 	}
 
 	if (test_value < loLimit)
 	{
-		m_currentLimit[limitIndex] = 2;//low limit violation
-		m_currentLimitError[limitIndex] =  test_value - loLimit;
+		m_currentLimit[limitIndex] = 2;  //low limit violation
+		m_currentLimitError[limitIndex] = test_value - loLimit;
 		return 2;
 	}
-	else if (test_value> hiLimit)
+	else if (test_value > hiLimit)
 	{
-		m_currentLimit[limitIndex] = 1;//High limit violation
+		m_currentLimit[limitIndex] = 1;  //High limit violation
 		m_currentLimitError[limitIndex] = test_value - hiLimit;
 		return 1;
 	};
 
-	m_currentLimit[limitIndex] = 0;//Free from violation
+	m_currentLimit[limitIndex] = 0;  //Free from violation
 	m_currentLimitError[limitIndex] = btScalar(0.f);
 	return 0;
 }
 
-
-
 btScalar btTranslationalLimitMotor::solveLinearAxis(
 	btScalar timeStep,
 	btScalar jacDiagABInv,
-	btRigidBody& body1,const btVector3 &pointInA,
-	btRigidBody& body2,const btVector3 &pointInB,
+	btRigidBody& body1, const btVector3& pointInA,
+	btRigidBody& body2, const btVector3& pointInB,
 	int limit_index,
-	const btVector3 & axis_normal_on_a,
-	const btVector3 & anchorPos)
+	const btVector3& axis_normal_on_a,
+	const btVector3& anchorPos)
 {
-
 	///find relative velocity
 	//    btVector3 rel_pos1 = pointInA - body1.getCenterOfMassPosition();
 	//    btVector3 rel_pos2 = pointInB - body2.getCenterOfMassPosition();
@@ -292,14 +253,12 @@ btScalar btTranslationalLimitMotor::solveLinearAxis(
 
 	btScalar rel_vel = axis_normal_on_a.dot(vel);
 
-
-
 	/// apply displacement correction
 
 	//positional error (zeroth order error)
 	btScalar depth = -(pointInA - pointInB).dot(axis_normal_on_a);
-	btScalar	lo = btScalar(-BT_LARGE_FLOAT);
-	btScalar	hi = btScalar(BT_LARGE_FLOAT);
+	btScalar lo = btScalar(-BT_LARGE_FLOAT);
+	btScalar hi = btScalar(BT_LARGE_FLOAT);
 
 	btScalar minLimit = m_lowerLimit[limit_index];
 	btScalar maxLimit = m_upperLimit[limit_index];
@@ -312,7 +271,6 @@ btScalar btTranslationalLimitMotor::solveLinearAxis(
 			{
 				depth -= maxLimit;
 				lo = btScalar(0.);
-
 			}
 			else
 			{
@@ -329,10 +287,7 @@ btScalar btTranslationalLimitMotor::solveLinearAxis(
 		}
 	}
 
-	btScalar normalImpulse= m_limitSoftness*(m_restitution*depth/timeStep - m_damping*rel_vel) * jacDiagABInv;
-
-
-
+	btScalar normalImpulse = m_limitSoftness * (m_restitution * depth / timeStep - m_damping * rel_vel) * jacDiagABInv;
 
 	btScalar oldNormalImpulse = m_accumulatedImpulse[limit_index];
 	btScalar sum = oldNormalImpulse + normalImpulse;
@@ -340,11 +295,9 @@ btScalar btTranslationalLimitMotor::solveLinearAxis(
 	normalImpulse = m_accumulatedImpulse[limit_index] - oldNormalImpulse;
 
 	btVector3 impulse_vector = axis_normal_on_a * normalImpulse;
-	body1.applyImpulse( impulse_vector, rel_pos1);
+	body1.applyImpulse(impulse_vector, rel_pos1);
 	body2.applyImpulse(-impulse_vector, rel_pos2);
 
-	
-
 	return normalImpulse;
 }
 
@@ -352,8 +305,8 @@ btScalar btTranslationalLimitMotor::solveLinearAxis(
 
 void btGeneric6DofConstraint::calculateAngleInfo()
 {
-	btMatrix3x3 relative_frame = m_calculatedTransformA.getBasis().inverse()*m_calculatedTransformB.getBasis();
-	matrixToEulerXYZ(relative_frame,m_calculatedAxisAngleDiff);
+	btMatrix3x3 relative_frame = m_calculatedTransformA.getBasis().inverse() * m_calculatedTransformB.getBasis();
+	matrixToEulerXYZ(relative_frame, m_calculatedAxisAngleDiff);
 	// in euler angle mode we do not actually constrain the angular velocity
 	// along the axes axis[0] and axis[2] (although we do use axis[1]) :
 	//
@@ -378,31 +331,30 @@ void btGeneric6DofConstraint::calculateAngleInfo()
 	m_calculatedAxis[0].normalize();
 	m_calculatedAxis[1].normalize();
 	m_calculatedAxis[2].normalize();
-
 }
 
 void btGeneric6DofConstraint::calculateTransforms()
 {
-	calculateTransforms(m_rbA.getCenterOfMassTransform(),m_rbB.getCenterOfMassTransform());
+	calculateTransforms(m_rbA.getCenterOfMassTransform(), m_rbB.getCenterOfMassTransform());
 }
 
-void btGeneric6DofConstraint::calculateTransforms(const btTransform& transA,const btTransform& transB)
+void btGeneric6DofConstraint::calculateTransforms(const btTransform& transA, const btTransform& transB)
 {
 	m_calculatedTransformA = transA * m_frameInA;
 	m_calculatedTransformB = transB * m_frameInB;
 	calculateLinearInfo();
 	calculateAngleInfo();
-	if(m_useOffsetForConstraintFrame)
-	{	//  get weight factors depending on masses
+	if (m_useOffsetForConstraintFrame)
+	{  //  get weight factors depending on masses
 		btScalar miA = getRigidBodyA().getInvMass();
 		btScalar miB = getRigidBodyB().getInvMass();
 		m_hasStaticBody = (miA < SIMD_EPSILON) || (miB < SIMD_EPSILON);
 		btScalar miS = miA + miB;
-		if(miS > btScalar(0.f))
+		if (miS > btScalar(0.f))
 		{
 			m_factA = miB / miS;
 		}
-		else 
+		else
 		{
 			m_factA = btScalar(0.5f);
 		}
@@ -410,39 +362,32 @@ void btGeneric6DofConstraint::calculateTransforms(const btTransform& transA,cons
 	}
 }
 
-
-
 void btGeneric6DofConstraint::buildLinearJacobian(
-	btJacobianEntry & jacLinear,const btVector3 & normalWorld,
-	const btVector3 & pivotAInW,const btVector3 & pivotBInW)
+	btJacobianEntry& jacLinear, const btVector3& normalWorld,
+	const btVector3& pivotAInW, const btVector3& pivotBInW)
 {
 	new (&jacLinear) btJacobianEntry(
-        m_rbA.getCenterOfMassTransform().getBasis().transpose(),
-        m_rbB.getCenterOfMassTransform().getBasis().transpose(),
-        pivotAInW - m_rbA.getCenterOfMassPosition(),
-        pivotBInW - m_rbB.getCenterOfMassPosition(),
-        normalWorld,
-        m_rbA.getInvInertiaDiagLocal(),
-        m_rbA.getInvMass(),
-        m_rbB.getInvInertiaDiagLocal(),
-        m_rbB.getInvMass());
+		m_rbA.getCenterOfMassTransform().getBasis().transpose(),
+		m_rbB.getCenterOfMassTransform().getBasis().transpose(),
+		pivotAInW - m_rbA.getCenterOfMassPosition(),
+		pivotBInW - m_rbB.getCenterOfMassPosition(),
+		normalWorld,
+		m_rbA.getInvInertiaDiagLocal(),
+		m_rbA.getInvMass(),
+		m_rbB.getInvInertiaDiagLocal(),
+		m_rbB.getInvMass());
 }
 
-
-
 void btGeneric6DofConstraint::buildAngularJacobian(
-	btJacobianEntry & jacAngular,const btVector3 & jointAxisW)
+	btJacobianEntry& jacAngular, const btVector3& jointAxisW)
 {
-	 new (&jacAngular)	btJacobianEntry(jointAxisW,
-                                      m_rbA.getCenterOfMassTransform().getBasis().transpose(),
-                                      m_rbB.getCenterOfMassTransform().getBasis().transpose(),
-                                      m_rbA.getInvInertiaDiagLocal(),
-                                      m_rbB.getInvInertiaDiagLocal());
-
+	new (&jacAngular) btJacobianEntry(jointAxisW,
+									  m_rbA.getCenterOfMassTransform().getBasis().transpose(),
+									  m_rbB.getCenterOfMassTransform().getBasis().transpose(),
+									  m_rbA.getInvInertiaDiagLocal(),
+									  m_rbB.getInvInertiaDiagLocal());
 }
 
-
-
 bool btGeneric6DofConstraint::testAngularLimitMotor(int axis_index)
 {
 	btScalar angle = m_calculatedAxisAngleDiff[axis_index];
@@ -453,23 +398,20 @@ bool btGeneric6DofConstraint::testAngularLimitMotor(int axis_index)
 	return m_angularLimits[axis_index].needApplyTorques();
 }
 
-
-
 void btGeneric6DofConstraint::buildJacobian()
 {
 #ifndef __SPU__
 	if (m_useSolveConstraintObsolete)
 	{
-
 		// Clear accumulated impulses for the next simulation step
 		m_linearLimits.m_accumulatedImpulse.setValue(btScalar(0.), btScalar(0.), btScalar(0.));
 		int i;
-		for(i = 0; i < 3; i++)
+		for (i = 0; i < 3; i++)
 		{
 			m_angularLimits[i].m_accumulatedImpulse = btScalar(0.);
 		}
 		//calculates transform
-		calculateTransforms(m_rbA.getCenterOfMassTransform(),m_rbB.getCenterOfMassTransform());
+		calculateTransforms(m_rbA.getCenterOfMassTransform(), m_rbB.getCenterOfMassTransform());
 
 		//  const btVector3& pivotAInW = m_calculatedTransformA.getOrigin();
 		//  const btVector3& pivotBInW = m_calculatedTransformB.getOrigin();
@@ -483,7 +425,7 @@ void btGeneric6DofConstraint::buildJacobian()
 
 		btVector3 normalWorld;
 		//linear part
-		for (i=0;i<3;i++)
+		for (i = 0; i < 3; i++)
 		{
 			if (m_linearLimits.isLimited(i))
 			{
@@ -493,56 +435,53 @@ void btGeneric6DofConstraint::buildJacobian()
 					normalWorld = m_calculatedTransformB.getBasis().getColumn(i);
 
 				buildLinearJacobian(
-					m_jacLinear[i],normalWorld ,
-					pivotAInW,pivotBInW);
-
+					m_jacLinear[i], normalWorld,
+					pivotAInW, pivotBInW);
 			}
 		}
 
 		// angular part
-		for (i=0;i<3;i++)
+		for (i = 0; i < 3; i++)
 		{
 			//calculates error angle
 			if (testAngularLimitMotor(i))
 			{
 				normalWorld = this->getAxis(i);
 				// Create angular atom
-				buildAngularJacobian(m_jacAng[i],normalWorld);
+				buildAngularJacobian(m_jacAng[i], normalWorld);
 			}
 		}
-
 	}
-#endif //__SPU__
-
+#endif  //__SPU__
 }
 
-
-void btGeneric6DofConstraint::getInfo1 (btConstraintInfo1* info)
+void btGeneric6DofConstraint::getInfo1(btConstraintInfo1* info)
 {
 	if (m_useSolveConstraintObsolete)
 	{
 		info->m_numConstraintRows = 0;
 		info->nub = 0;
-	} else
+	}
+	else
 	{
 		//prepare constraint
-		calculateTransforms(m_rbA.getCenterOfMassTransform(),m_rbB.getCenterOfMassTransform());
+		calculateTransforms(m_rbA.getCenterOfMassTransform(), m_rbB.getCenterOfMassTransform());
 		info->m_numConstraintRows = 0;
 		info->nub = 6;
 		int i;
 		//test linear limits
-		for(i = 0; i < 3; i++)
+		for (i = 0; i < 3; i++)
 		{
-			if(m_linearLimits.needApplyForce(i))
+			if (m_linearLimits.needApplyForce(i))
 			{
 				info->m_numConstraintRows++;
 				info->nub--;
 			}
 		}
 		//test angular limits
-		for (i=0;i<3 ;i++ )
+		for (i = 0; i < 3; i++)
 		{
-			if(testAngularLimitMotor(i))
+			if (testAngularLimitMotor(i))
 			{
 				info->m_numConstraintRows++;
 				info->nub--;
@@ -551,13 +490,14 @@ void btGeneric6DofConstraint::getInfo1 (btConstraintInfo1* info)
 	}
 }
 
-void btGeneric6DofConstraint::getInfo1NonVirtual (btConstraintInfo1* info)
+void btGeneric6DofConstraint::getInfo1NonVirtual(btConstraintInfo1* info)
 {
 	if (m_useSolveConstraintObsolete)
 	{
 		info->m_numConstraintRows = 0;
 		info->nub = 0;
-	} else
+	}
+	else
 	{
 		//pre-allocate all 6
 		info->m_numConstraintRows = 6;
@@ -565,8 +505,7 @@ void btGeneric6DofConstraint::getInfo1NonVirtual (btConstraintInfo1* info)
 	}
 }
 
-
-void btGeneric6DofConstraint::getInfo2 (btConstraintInfo2* info)
+void btGeneric6DofConstraint::getInfo2(btConstraintInfo2* info)
 {
 	btAssert(!m_useSolveConstraintObsolete);
 
@@ -577,136 +516,124 @@ void btGeneric6DofConstraint::getInfo2 (btConstraintInfo2* info)
 	const btVector3& angVelA = m_rbA.getAngularVelocity();
 	const btVector3& angVelB = m_rbB.getAngularVelocity();
 
-	if(m_useOffsetForConstraintFrame)
-	{ // for stability better to solve angular limits first
-		int row = setAngularLimits(info, 0,transA,transB,linVelA,linVelB,angVelA,angVelB);
-		setLinearLimits(info, row, transA,transB,linVelA,linVelB,angVelA,angVelB);
+	if (m_useOffsetForConstraintFrame)
+	{  // for stability better to solve angular limits first
+		int row = setAngularLimits(info, 0, transA, transB, linVelA, linVelB, angVelA, angVelB);
+		setLinearLimits(info, row, transA, transB, linVelA, linVelB, angVelA, angVelB);
 	}
 	else
-	{ // leave old version for compatibility
-		int row = setLinearLimits(info, 0, transA,transB,linVelA,linVelB,angVelA,angVelB);
-		setAngularLimits(info, row,transA,transB,linVelA,linVelB,angVelA,angVelB);
+	{  // leave old version for compatibility
+		int row = setLinearLimits(info, 0, transA, transB, linVelA, linVelB, angVelA, angVelB);
+		setAngularLimits(info, row, transA, transB, linVelA, linVelB, angVelA, angVelB);
 	}
-
 }
 
-
-void btGeneric6DofConstraint::getInfo2NonVirtual (btConstraintInfo2* info, const btTransform& transA,const btTransform& transB,const btVector3& linVelA,const btVector3& linVelB,const btVector3& angVelA,const btVector3& angVelB)
+void btGeneric6DofConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTransform& transA, const btTransform& transB, const btVector3& linVelA, const btVector3& linVelB, const btVector3& angVelA, const btVector3& angVelB)
 {
-	
 	btAssert(!m_useSolveConstraintObsolete);
 	//prepare constraint
-	calculateTransforms(transA,transB);
+	calculateTransforms(transA, transB);
 
 	int i;
-	for (i=0;i<3 ;i++ )
+	for (i = 0; i < 3; i++)
 	{
 		testAngularLimitMotor(i);
 	}
 
-	if(m_useOffsetForConstraintFrame)
-	{ // for stability better to solve angular limits first
-		int row = setAngularLimits(info, 0,transA,transB,linVelA,linVelB,angVelA,angVelB);
-		setLinearLimits(info, row, transA,transB,linVelA,linVelB,angVelA,angVelB);
+	if (m_useOffsetForConstraintFrame)
+	{  // for stability better to solve angular limits first
+		int row = setAngularLimits(info, 0, transA, transB, linVelA, linVelB, angVelA, angVelB);
+		setLinearLimits(info, row, transA, transB, linVelA, linVelB, angVelA, angVelB);
 	}
 	else
-	{ // leave old version for compatibility
-		int row = setLinearLimits(info, 0, transA,transB,linVelA,linVelB,angVelA,angVelB);
-		setAngularLimits(info, row,transA,transB,linVelA,linVelB,angVelA,angVelB);
+	{  // leave old version for compatibility
+		int row = setLinearLimits(info, 0, transA, transB, linVelA, linVelB, angVelA, angVelB);
+		setAngularLimits(info, row, transA, transB, linVelA, linVelB, angVelA, angVelB);
 	}
 }
 
-
-
-int btGeneric6DofConstraint::setLinearLimits(btConstraintInfo2* info, int row, const btTransform& transA,const btTransform& transB,const btVector3& linVelA,const btVector3& linVelB,const btVector3& angVelA,const btVector3& angVelB)
+int btGeneric6DofConstraint::setLinearLimits(btConstraintInfo2* info, int row, const btTransform& transA, const btTransform& transB, const btVector3& linVelA, const btVector3& linVelB, const btVector3& angVelA, const btVector3& angVelB)
 {
-//	int row = 0;
+	//	int row = 0;
 	//solve linear limits
 	btRotationalLimitMotor limot;
-	for (int i=0;i<3 ;i++ )
+	for (int i = 0; i < 3; i++)
 	{
-		if(m_linearLimits.needApplyForce(i))
-		{ // re-use rotational motor code
+		if (m_linearLimits.needApplyForce(i))
+		{  // re-use rotational motor code
 			limot.m_bounce = btScalar(0.f);
 			limot.m_currentLimit = m_linearLimits.m_currentLimit[i];
 			limot.m_currentPosition = m_linearLimits.m_currentLinearDiff[i];
-			limot.m_currentLimitError  = m_linearLimits.m_currentLimitError[i];
-			limot.m_damping  = m_linearLimits.m_damping;
-			limot.m_enableMotor  = m_linearLimits.m_enableMotor[i];
-			limot.m_hiLimit  = m_linearLimits.m_upperLimit[i];
-			limot.m_limitSoftness  = m_linearLimits.m_limitSoftness;
-			limot.m_loLimit  = m_linearLimits.m_lowerLimit[i];
-			limot.m_maxLimitForce  = btScalar(0.f);
-			limot.m_maxMotorForce  = m_linearLimits.m_maxMotorForce[i];
-			limot.m_targetVelocity  = m_linearLimits.m_targetVelocity[i];
+			limot.m_currentLimitError = m_linearLimits.m_currentLimitError[i];
+			limot.m_damping = m_linearLimits.m_damping;
+			limot.m_enableMotor = m_linearLimits.m_enableMotor[i];
+			limot.m_hiLimit = m_linearLimits.m_upperLimit[i];
+			limot.m_limitSoftness = m_linearLimits.m_limitSoftness;
+			limot.m_loLimit = m_linearLimits.m_lowerLimit[i];
+			limot.m_maxLimitForce = btScalar(0.f);
+			limot.m_maxMotorForce = m_linearLimits.m_maxMotorForce[i];
+			limot.m_targetVelocity = m_linearLimits.m_targetVelocity[i];
 			btVector3 axis = m_calculatedTransformA.getBasis().getColumn(i);
 			int flags = m_flags >> (i * BT_6DOF_FLAGS_AXIS_SHIFT);
-			limot.m_normalCFM	= (flags & BT_6DOF_FLAGS_CFM_NORM) ? m_linearLimits.m_normalCFM[i] : info->cfm[0];
-			limot.m_stopCFM		= (flags & BT_6DOF_FLAGS_CFM_STOP) ? m_linearLimits.m_stopCFM[i] : info->cfm[0];
-			limot.m_stopERP		= (flags & BT_6DOF_FLAGS_ERP_STOP) ? m_linearLimits.m_stopERP[i] : info->erp;
-			if(m_useOffsetForConstraintFrame)
+			limot.m_normalCFM = (flags & BT_6DOF_FLAGS_CFM_NORM) ? m_linearLimits.m_normalCFM[i] : info->cfm[0];
+			limot.m_stopCFM = (flags & BT_6DOF_FLAGS_CFM_STOP) ? m_linearLimits.m_stopCFM[i] : info->cfm[0];
+			limot.m_stopERP = (flags & BT_6DOF_FLAGS_ERP_STOP) ? m_linearLimits.m_stopERP[i] : info->erp;
+			if (m_useOffsetForConstraintFrame)
 			{
 				int indx1 = (i + 1) % 3;
 				int indx2 = (i + 2) % 3;
-				int rotAllowed = 1; // rotations around orthos to current axis
-				if(m_angularLimits[indx1].m_currentLimit && m_angularLimits[indx2].m_currentLimit)
+				int rotAllowed = 1;  // rotations around orthos to current axis
+				if (m_angularLimits[indx1].m_currentLimit && m_angularLimits[indx2].m_currentLimit)
 				{
 					rotAllowed = 0;
 				}
-				row += get_limit_motor_info2(&limot, transA,transB,linVelA,linVelB,angVelA,angVelB, info, row, axis, 0, rotAllowed);
+				row += get_limit_motor_info2(&limot, transA, transB, linVelA, linVelB, angVelA, angVelB, info, row, axis, 0, rotAllowed);
 			}
 			else
 			{
-				row += get_limit_motor_info2(&limot, transA,transB,linVelA,linVelB,angVelA,angVelB, info, row, axis, 0);
+				row += get_limit_motor_info2(&limot, transA, transB, linVelA, linVelB, angVelA, angVelB, info, row, axis, 0);
 			}
 		}
 	}
 	return row;
 }
 
-
-
-int btGeneric6DofConstraint::setAngularLimits(btConstraintInfo2 *info, int row_offset, const btTransform& transA,const btTransform& transB,const btVector3& linVelA,const btVector3& linVelB,const btVector3& angVelA,const btVector3& angVelB)
+int btGeneric6DofConstraint::setAngularLimits(btConstraintInfo2* info, int row_offset, const btTransform& transA, const btTransform& transB, const btVector3& linVelA, const btVector3& linVelB, const btVector3& angVelA, const btVector3& angVelB)
 {
-	btGeneric6DofConstraint * d6constraint = this;
+	btGeneric6DofConstraint* d6constraint = this;
 	int row = row_offset;
 	//solve angular limits
-	for (int i=0;i<3 ;i++ )
+	for (int i = 0; i < 3; i++)
 	{
-		if(d6constraint->getRotationalLimitMotor(i)->needApplyTorques())
+		if (d6constraint->getRotationalLimitMotor(i)->needApplyTorques())
 		{
 			btVector3 axis = d6constraint->getAxis(i);
 			int flags = m_flags >> ((i + 3) * BT_6DOF_FLAGS_AXIS_SHIFT);
-			if(!(flags & BT_6DOF_FLAGS_CFM_NORM))
+			if (!(flags & BT_6DOF_FLAGS_CFM_NORM))
 			{
 				m_angularLimits[i].m_normalCFM = info->cfm[0];
 			}
-			if(!(flags & BT_6DOF_FLAGS_CFM_STOP))
+			if (!(flags & BT_6DOF_FLAGS_CFM_STOP))
 			{
 				m_angularLimits[i].m_stopCFM = info->cfm[0];
 			}
-			if(!(flags & BT_6DOF_FLAGS_ERP_STOP))
+			if (!(flags & BT_6DOF_FLAGS_ERP_STOP))
 			{
 				m_angularLimits[i].m_stopERP = info->erp;
 			}
 			row += get_limit_motor_info2(d6constraint->getRotationalLimitMotor(i),
-												transA,transB,linVelA,linVelB,angVelA,angVelB, info,row,axis,1);
+										 transA, transB, linVelA, linVelB, angVelA, angVelB, info, row, axis, 1);
 		}
 	}
 
 	return row;
 }
 
-
-
-
-void	btGeneric6DofConstraint::updateRHS(btScalar	timeStep)
+void btGeneric6DofConstraint::updateRHS(btScalar timeStep)
 {
 	(void)timeStep;
-
 }
 
-
 void btGeneric6DofConstraint::setFrames(const btTransform& frameA, const btTransform& frameB)
 {
 	m_frameInA = frameA;
@@ -715,33 +642,27 @@ void btGeneric6DofConstraint::setFrames(const btTransform& frameA, const btTrans
 	calculateTransforms();
 }
 
-
-
 btVector3 btGeneric6DofConstraint::getAxis(int axis_index) const
 {
 	return m_calculatedAxis[axis_index];
 }
 
-
-btScalar	btGeneric6DofConstraint::getRelativePivotPosition(int axisIndex) const
+btScalar btGeneric6DofConstraint::getRelativePivotPosition(int axisIndex) const
 {
 	return m_calculatedLinearDiff[axisIndex];
 }
 
-
 btScalar btGeneric6DofConstraint::getAngle(int axisIndex) const
 {
 	return m_calculatedAxisAngleDiff[axisIndex];
 }
 
-
-
 void btGeneric6DofConstraint::calcAnchorPos(void)
 {
 	btScalar imA = m_rbA.getInvMass();
 	btScalar imB = m_rbB.getInvMass();
 	btScalar weight;
-	if(imB == btScalar(0.0))
+	if (imB == btScalar(0.0))
 	{
 		weight = btScalar(1.0);
 	}
@@ -755,43 +676,39 @@ void btGeneric6DofConstraint::calcAnchorPos(void)
 	return;
 }
 
-
-
 void btGeneric6DofConstraint::calculateLinearInfo()
 {
 	m_calculatedLinearDiff = m_calculatedTransformB.getOrigin() - m_calculatedTransformA.getOrigin();
 	m_calculatedLinearDiff = m_calculatedTransformA.getBasis().inverse() * m_calculatedLinearDiff;
-	for(int i = 0; i < 3; i++)
+	for (int i = 0; i < 3; i++)
 	{
 		m_linearLimits.m_currentLinearDiff[i] = m_calculatedLinearDiff[i];
 		m_linearLimits.testLimitValue(i, m_calculatedLinearDiff[i]);
 	}
 }
 
-
-
 int btGeneric6DofConstraint::get_limit_motor_info2(
-	btRotationalLimitMotor * limot,
-	const btTransform& transA,const btTransform& transB,const btVector3& linVelA,const btVector3& linVelB,const btVector3& angVelA,const btVector3& angVelB,
-	btConstraintInfo2 *info, int row, btVector3& ax1, int rotational,int rotAllowed)
+	btRotationalLimitMotor* limot,
+	const btTransform& transA, const btTransform& transB, const btVector3& linVelA, const btVector3& linVelB, const btVector3& angVelA, const btVector3& angVelB,
+	btConstraintInfo2* info, int row, btVector3& ax1, int rotational, int rotAllowed)
 {
-    int srow = row * info->rowskip;
-    bool powered = limot->m_enableMotor;
-    int limit = limot->m_currentLimit;
-    if (powered || limit)
-    {   // if the joint is powered, or has joint limits, add in the extra row
-        btScalar *J1 = rotational ? info->m_J1angularAxis : info->m_J1linearAxis;
-        btScalar *J2 = rotational ? info->m_J2angularAxis : info->m_J2linearAxis;
-        J1[srow+0] = ax1[0];
-        J1[srow+1] = ax1[1];
-        J1[srow+2] = ax1[2];
+	int srow = row * info->rowskip;
+	bool powered = limot->m_enableMotor;
+	int limit = limot->m_currentLimit;
+	if (powered || limit)
+	{  // if the joint is powered, or has joint limits, add in the extra row
+		btScalar* J1 = rotational ? info->m_J1angularAxis : info->m_J1linearAxis;
+		btScalar* J2 = rotational ? info->m_J2angularAxis : info->m_J2linearAxis;
+		J1[srow + 0] = ax1[0];
+		J1[srow + 1] = ax1[1];
+		J1[srow + 2] = ax1[2];
 
-        J2[srow+0] = -ax1[0];
-        J2[srow+1] = -ax1[1];
-        J2[srow+2] = -ax1[2];
+		J2[srow + 0] = -ax1[0];
+		J2[srow + 1] = -ax1[1];
+		J2[srow + 2] = -ax1[2];
 
-		if((!rotational))
-        {
+		if ((!rotational))
+		{
 			if (m_useOffsetForConstraintFrame)
 			{
 				btVector3 tmpA, tmpB, relA, relB;
@@ -814,55 +731,56 @@ int btGeneric6DofConstraint::get_limit_motor_info2(
 				relB = orthoB - totalDist * m_factB;
 				tmpA = relA.cross(ax1);
 				tmpB = relB.cross(ax1);
-				if(m_hasStaticBody && (!rotAllowed))
+				if (m_hasStaticBody && (!rotAllowed))
 				{
 					tmpA *= m_factA;
 					tmpB *= m_factB;
 				}
 				int i;
-				for (i=0; i<3; i++) info->m_J1angularAxis[srow+i] = tmpA[i];
-				for (i=0; i<3; i++) info->m_J2angularAxis[srow+i] = -tmpB[i];
-			} else
+				for (i = 0; i < 3; i++) info->m_J1angularAxis[srow + i] = tmpA[i];
+				for (i = 0; i < 3; i++) info->m_J2angularAxis[srow + i] = -tmpB[i];
+			}
+			else
 			{
-				btVector3 ltd;	// Linear Torque Decoupling vector
+				btVector3 ltd;  // Linear Torque Decoupling vector
 				btVector3 c = m_calculatedTransformB.getOrigin() - transA.getOrigin();
 				ltd = c.cross(ax1);
-				info->m_J1angularAxis[srow+0] = ltd[0];
-				info->m_J1angularAxis[srow+1] = ltd[1];
-				info->m_J1angularAxis[srow+2] = ltd[2];
+				info->m_J1angularAxis[srow + 0] = ltd[0];
+				info->m_J1angularAxis[srow + 1] = ltd[1];
+				info->m_J1angularAxis[srow + 2] = ltd[2];
 
 				c = m_calculatedTransformB.getOrigin() - transB.getOrigin();
 				ltd = -c.cross(ax1);
-				info->m_J2angularAxis[srow+0] = ltd[0];
-				info->m_J2angularAxis[srow+1] = ltd[1];
-				info->m_J2angularAxis[srow+2] = ltd[2];
+				info->m_J2angularAxis[srow + 0] = ltd[0];
+				info->m_J2angularAxis[srow + 1] = ltd[1];
+				info->m_J2angularAxis[srow + 2] = ltd[2];
 			}
-        }
-        // if we're limited low and high simultaneously, the joint motor is
-        // ineffective
-        if (limit && (limot->m_loLimit == limot->m_hiLimit)) powered = false;
-        info->m_constraintError[srow] = btScalar(0.f);
-        if (powered)
-        {
+		}
+		// if we're limited low and high simultaneously, the joint motor is
+		// ineffective
+		if (limit && (limot->m_loLimit == limot->m_hiLimit)) powered = false;
+		info->m_constraintError[srow] = btScalar(0.f);
+		if (powered)
+		{
 			info->cfm[srow] = limot->m_normalCFM;
-            if(!limit)
-            {
+			if (!limit)
+			{
 				btScalar tag_vel = rotational ? limot->m_targetVelocity : -limot->m_targetVelocity;
 
-				btScalar mot_fact = getMotorFactor(	limot->m_currentPosition, 
-													limot->m_loLimit,
-													limot->m_hiLimit, 
-													tag_vel, 
-													info->fps * limot->m_stopERP);
+				btScalar mot_fact = getMotorFactor(limot->m_currentPosition,
+												   limot->m_loLimit,
+												   limot->m_hiLimit,
+												   tag_vel,
+												   info->fps * limot->m_stopERP);
 				info->m_constraintError[srow] += mot_fact * limot->m_targetVelocity;
-                info->m_lowerLimit[srow] = -limot->m_maxMotorForce / info->fps;
-                info->m_upperLimit[srow] = limot->m_maxMotorForce / info->fps;
-            }
-        }
-        if(limit)
-        {
-            btScalar k = info->fps * limot->m_stopERP;
-			if(!rotational)
+				info->m_lowerLimit[srow] = -limot->m_maxMotorForce / info->fps;
+				info->m_upperLimit[srow] = limot->m_maxMotorForce / info->fps;
+			}
+		}
+		if (limit)
+		{
+			btScalar k = info->fps * limot->m_stopERP;
+			if (!rotational)
 			{
 				info->m_constraintError[srow] += k * limot->m_currentLimitError;
 			}
@@ -871,116 +789,112 @@ int btGeneric6DofConstraint::get_limit_motor_info2(
 				info->m_constraintError[srow] += -k * limot->m_currentLimitError;
 			}
 			info->cfm[srow] = limot->m_stopCFM;
-            if (limot->m_loLimit == limot->m_hiLimit)
-            {   // limited low and high simultaneously
-                info->m_lowerLimit[srow] = -SIMD_INFINITY;
-                info->m_upperLimit[srow] = SIMD_INFINITY;
-            }
-            else
-            {
-                if (limit == 1)
-                {
-                    info->m_lowerLimit[srow] = 0;
-                    info->m_upperLimit[srow] = SIMD_INFINITY;
-                }
-                else
-                {
-                    info->m_lowerLimit[srow] = -SIMD_INFINITY;
-                    info->m_upperLimit[srow] = 0;
-                }
-                // deal with bounce
-                if (limot->m_bounce > 0)
-                {
-                    // calculate joint velocity
-                    btScalar vel;
-                    if (rotational)
-                    {
-                        vel = angVelA.dot(ax1);
-//make sure that if no body -> angVelB == zero vec
-//                        if (body1)
-                            vel -= angVelB.dot(ax1);
-                    }
-                    else
-                    {
-                        vel = linVelA.dot(ax1);
-//make sure that if no body -> angVelB == zero vec
-//                        if (body1)
-                            vel -= linVelB.dot(ax1);
-                    }
-                    // only apply bounce if the velocity is incoming, and if the
-                    // resulting c[] exceeds what we already have.
-                    if (limit == 1)
-                    {
-                        if (vel < 0)
-                        {
-                            btScalar newc = -limot->m_bounce* vel;
-                            if (newc > info->m_constraintError[srow]) 
+			if (limot->m_loLimit == limot->m_hiLimit)
+			{  // limited low and high simultaneously
+				info->m_lowerLimit[srow] = -SIMD_INFINITY;
+				info->m_upperLimit[srow] = SIMD_INFINITY;
+			}
+			else
+			{
+				if (limit == 1)
+				{
+					info->m_lowerLimit[srow] = 0;
+					info->m_upperLimit[srow] = SIMD_INFINITY;
+				}
+				else
+				{
+					info->m_lowerLimit[srow] = -SIMD_INFINITY;
+					info->m_upperLimit[srow] = 0;
+				}
+				// deal with bounce
+				if (limot->m_bounce > 0)
+				{
+					// calculate joint velocity
+					btScalar vel;
+					if (rotational)
+					{
+						vel = angVelA.dot(ax1);
+						//make sure that if no body -> angVelB == zero vec
+						//                        if (body1)
+						vel -= angVelB.dot(ax1);
+					}
+					else
+					{
+						vel = linVelA.dot(ax1);
+						//make sure that if no body -> angVelB == zero vec
+						//                        if (body1)
+						vel -= linVelB.dot(ax1);
+					}
+					// only apply bounce if the velocity is incoming, and if the
+					// resulting c[] exceeds what we already have.
+					if (limit == 1)
+					{
+						if (vel < 0)
+						{
+							btScalar newc = -limot->m_bounce * vel;
+							if (newc > info->m_constraintError[srow])
 								info->m_constraintError[srow] = newc;
-                        }
-                    }
-                    else
-                    {
-                        if (vel > 0)
-                        {
-                            btScalar newc = -limot->m_bounce * vel;
-                            if (newc < info->m_constraintError[srow]) 
+						}
+					}
+					else
+					{
+						if (vel > 0)
+						{
+							btScalar newc = -limot->m_bounce * vel;
+							if (newc < info->m_constraintError[srow])
 								info->m_constraintError[srow] = newc;
-                        }
-                    }
-                }
-            }
-        }
-        return 1;
-    }
-    else return 0;
+						}
+					}
+				}
+			}
+		}
+		return 1;
+	}
+	else
+		return 0;
 }
 
-
-
-
-
-
-	///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5). 
-	///If no axis is provided, it uses the default axis for this constraint.
+///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5).
+///If no axis is provided, it uses the default axis for this constraint.
 void btGeneric6DofConstraint::setParam(int num, btScalar value, int axis)
 {
-	if((axis >= 0) && (axis < 3))
+	if ((axis >= 0) && (axis < 3))
 	{
-		switch(num)
+		switch (num)
 		{
-			case BT_CONSTRAINT_STOP_ERP : 
+			case BT_CONSTRAINT_STOP_ERP:
 				m_linearLimits.m_stopERP[axis] = value;
 				m_flags |= BT_6DOF_FLAGS_ERP_STOP << (axis * BT_6DOF_FLAGS_AXIS_SHIFT);
 				break;
-			case BT_CONSTRAINT_STOP_CFM : 
+			case BT_CONSTRAINT_STOP_CFM:
 				m_linearLimits.m_stopCFM[axis] = value;
 				m_flags |= BT_6DOF_FLAGS_CFM_STOP << (axis * BT_6DOF_FLAGS_AXIS_SHIFT);
 				break;
-			case BT_CONSTRAINT_CFM : 
+			case BT_CONSTRAINT_CFM:
 				m_linearLimits.m_normalCFM[axis] = value;
 				m_flags |= BT_6DOF_FLAGS_CFM_NORM << (axis * BT_6DOF_FLAGS_AXIS_SHIFT);
 				break;
-			default : 
+			default:
 				btAssertConstrParams(0);
 		}
 	}
-	else if((axis >=3) && (axis < 6))
+	else if ((axis >= 3) && (axis < 6))
 	{
-		switch(num)
+		switch (num)
 		{
-			case BT_CONSTRAINT_STOP_ERP : 
+			case BT_CONSTRAINT_STOP_ERP:
 				m_angularLimits[axis - 3].m_stopERP = value;
 				m_flags |= BT_6DOF_FLAGS_ERP_STOP << (axis * BT_6DOF_FLAGS_AXIS_SHIFT);
 				break;
-			case BT_CONSTRAINT_STOP_CFM : 
+			case BT_CONSTRAINT_STOP_CFM:
 				m_angularLimits[axis - 3].m_stopCFM = value;
 				m_flags |= BT_6DOF_FLAGS_CFM_STOP << (axis * BT_6DOF_FLAGS_AXIS_SHIFT);
 				break;
-			case BT_CONSTRAINT_CFM : 
+			case BT_CONSTRAINT_CFM:
 				m_angularLimits[axis - 3].m_normalCFM = value;
 				m_flags |= BT_6DOF_FLAGS_CFM_NORM << (axis * BT_6DOF_FLAGS_AXIS_SHIFT);
 				break;
-			default : 
+			default:
 				btAssertConstrParams(0);
 		}
 	}
@@ -990,47 +904,47 @@ void btGeneric6DofConstraint::setParam(int num, btScalar value, int axis)
 	}
 }
 
-	///return the local value of parameter
-btScalar btGeneric6DofConstraint::getParam(int num, int axis) const 
+///return the local value of parameter
+btScalar btGeneric6DofConstraint::getParam(int num, int axis) const
 {
 	btScalar retVal = 0;
-	if((axis >= 0) && (axis < 3))
+	if ((axis >= 0) && (axis < 3))
 	{
-		switch(num)
+		switch (num)
 		{
-			case BT_CONSTRAINT_STOP_ERP : 
+			case BT_CONSTRAINT_STOP_ERP:
 				btAssertConstrParams(m_flags & (BT_6DOF_FLAGS_ERP_STOP << (axis * BT_6DOF_FLAGS_AXIS_SHIFT)));
 				retVal = m_linearLimits.m_stopERP[axis];
 				break;
-			case BT_CONSTRAINT_STOP_CFM : 
+			case BT_CONSTRAINT_STOP_CFM:
 				btAssertConstrParams(m_flags & (BT_6DOF_FLAGS_CFM_STOP << (axis * BT_6DOF_FLAGS_AXIS_SHIFT)));
 				retVal = m_linearLimits.m_stopCFM[axis];
 				break;
-			case BT_CONSTRAINT_CFM : 
+			case BT_CONSTRAINT_CFM:
 				btAssertConstrParams(m_flags & (BT_6DOF_FLAGS_CFM_NORM << (axis * BT_6DOF_FLAGS_AXIS_SHIFT)));
 				retVal = m_linearLimits.m_normalCFM[axis];
 				break;
-			default : 
+			default:
 				btAssertConstrParams(0);
 		}
 	}
-	else if((axis >=3) && (axis < 6))
+	else if ((axis >= 3) && (axis < 6))
 	{
-		switch(num)
+		switch (num)
 		{
-			case BT_CONSTRAINT_STOP_ERP : 
+			case BT_CONSTRAINT_STOP_ERP:
 				btAssertConstrParams(m_flags & (BT_6DOF_FLAGS_ERP_STOP << (axis * BT_6DOF_FLAGS_AXIS_SHIFT)));
 				retVal = m_angularLimits[axis - 3].m_stopERP;
 				break;
-			case BT_CONSTRAINT_STOP_CFM : 
+			case BT_CONSTRAINT_STOP_CFM:
 				btAssertConstrParams(m_flags & (BT_6DOF_FLAGS_CFM_STOP << (axis * BT_6DOF_FLAGS_AXIS_SHIFT)));
 				retVal = m_angularLimits[axis - 3].m_stopCFM;
 				break;
-			case BT_CONSTRAINT_CFM : 
+			case BT_CONSTRAINT_CFM:
 				btAssertConstrParams(m_flags & (BT_6DOF_FLAGS_CFM_NORM << (axis * BT_6DOF_FLAGS_AXIS_SHIFT)));
 				retVal = m_angularLimits[axis - 3].m_normalCFM;
 				break;
-			default : 
+			default:
 				btAssertConstrParams(0);
 		}
 	}
@@ -1041,23 +955,21 @@ btScalar btGeneric6DofConstraint::getParam(int num, int axis) const
 	return retVal;
 }
 
- 
-
-void btGeneric6DofConstraint::setAxis(const btVector3& axis1,const btVector3& axis2)
+void btGeneric6DofConstraint::setAxis(const btVector3& axis1, const btVector3& axis2)
 {
 	btVector3 zAxis = axis1.normalized();
 	btVector3 yAxis = axis2.normalized();
-	btVector3 xAxis = yAxis.cross(zAxis); // we want right coordinate system
-	
+	btVector3 xAxis = yAxis.cross(zAxis);  // we want right coordinate system
+
 	btTransform frameInW;
 	frameInW.setIdentity();
-	frameInW.getBasis().setValue(	xAxis[0], yAxis[0], zAxis[0],	
-	                                xAxis[1], yAxis[1], zAxis[1],
-	                               xAxis[2], yAxis[2], zAxis[2]);
-	
+	frameInW.getBasis().setValue(xAxis[0], yAxis[0], zAxis[0],
+								 xAxis[1], yAxis[1], zAxis[1],
+								 xAxis[2], yAxis[2], zAxis[2]);
+
 	// now get constraint frame in local coordinate systems
 	m_frameInA = m_rbA.getCenterOfMassTransform().inverse() * frameInW;
 	m_frameInB = m_rbB.getCenterOfMassTransform().inverse() * frameInW;
-	
+
 	calculateTransforms();
 }
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofConstraint.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofConstraint.h
index b2ad45f749d..b9e762e175f 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofConstraint.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofConstraint.h
@@ -23,7 +23,6 @@ email: projectileman@yahoo.com
 http://gimpact.sf.net
 */
 
-
 #ifndef BT_GENERIC_6DOF_CONSTRAINT_H
 #define BT_GENERIC_6DOF_CONSTRAINT_H
 
@@ -33,96 +32,91 @@ http://gimpact.sf.net
 
 class btRigidBody;
 
-
-
 #ifdef BT_USE_DOUBLE_PRECISION
-#define btGeneric6DofConstraintData2		btGeneric6DofConstraintDoubleData2
-#define btGeneric6DofConstraintDataName	"btGeneric6DofConstraintDoubleData2"
+#define btGeneric6DofConstraintData2 btGeneric6DofConstraintDoubleData2
+#define btGeneric6DofConstraintDataName "btGeneric6DofConstraintDoubleData2"
 #else
-#define btGeneric6DofConstraintData2		btGeneric6DofConstraintData
-#define btGeneric6DofConstraintDataName	"btGeneric6DofConstraintData"
-#endif //BT_USE_DOUBLE_PRECISION
-
+#define btGeneric6DofConstraintData2 btGeneric6DofConstraintData
+#define btGeneric6DofConstraintDataName "btGeneric6DofConstraintData"
+#endif  //BT_USE_DOUBLE_PRECISION
 
 //! Rotation Limit structure for generic joints
 class btRotationalLimitMotor
 {
 public:
-    //! limit_parameters
-    //!@{
-    btScalar m_loLimit;//!< joint limit
-    btScalar m_hiLimit;//!< joint limit
-    btScalar m_targetVelocity;//!< target motor velocity
-    btScalar m_maxMotorForce;//!< max force on motor
-    btScalar m_maxLimitForce;//!< max force on limit
-    btScalar m_damping;//!< Damping.
-    btScalar m_limitSoftness;//! Relaxation factor
-    btScalar m_normalCFM;//!< Constraint force mixing factor
-    btScalar m_stopERP;//!< Error tolerance factor when joint is at limit
-    btScalar m_stopCFM;//!< Constraint force mixing factor when joint is at limit
-    btScalar m_bounce;//!< restitution factor
-    bool m_enableMotor;
+	//! limit_parameters
+	//!@{
+	btScalar m_loLimit;         //!< joint limit
+	btScalar m_hiLimit;         //!< joint limit
+	btScalar m_targetVelocity;  //!< target motor velocity
+	btScalar m_maxMotorForce;   //!< max force on motor
+	btScalar m_maxLimitForce;   //!< max force on limit
+	btScalar m_damping;         //!< Damping.
+	btScalar m_limitSoftness;   //! Relaxation factor
+	btScalar m_normalCFM;       //!< Constraint force mixing factor
+	btScalar m_stopERP;         //!< Error tolerance factor when joint is at limit
+	btScalar m_stopCFM;         //!< Constraint force mixing factor when joint is at limit
+	btScalar m_bounce;          //!< restitution factor
+	bool m_enableMotor;
 
-    //!@}
+	//!@}
 
-    //! temp_variables
-    //!@{
-    btScalar m_currentLimitError;//!  How much is violated this limit
-    btScalar m_currentPosition;     //!  current value of angle 
-    int m_currentLimit;//!< 0=free, 1=at lo limit, 2=at hi limit
-    btScalar m_accumulatedImpulse;
-    //!@}
+	//! temp_variables
+	//!@{
+	btScalar m_currentLimitError;  //!  How much is violated this limit
+	btScalar m_currentPosition;    //!  current value of angle
+	int m_currentLimit;            //!< 0=free, 1=at lo limit, 2=at hi limit
+	btScalar m_accumulatedImpulse;
+	//!@}
 
-    btRotationalLimitMotor()
-    {
-    	m_accumulatedImpulse = 0.f;
-        m_targetVelocity = 0;
-        m_maxMotorForce = 6.0f;
-        m_maxLimitForce = 300.0f;
-        m_loLimit = 1.0f;
-        m_hiLimit = -1.0f;
+	btRotationalLimitMotor()
+	{
+		m_accumulatedImpulse = 0.f;
+		m_targetVelocity = 0;
+		m_maxMotorForce = 6.0f;
+		m_maxLimitForce = 300.0f;
+		m_loLimit = 1.0f;
+		m_hiLimit = -1.0f;
 		m_normalCFM = 0.f;
 		m_stopERP = 0.2f;
 		m_stopCFM = 0.f;
-        m_bounce = 0.0f;
-        m_damping = 1.0f;
-        m_limitSoftness = 0.5f;
-        m_currentLimit = 0;
-        m_currentLimitError = 0;
-        m_enableMotor = false;
-    }
+		m_bounce = 0.0f;
+		m_damping = 1.0f;
+		m_limitSoftness = 0.5f;
+		m_currentLimit = 0;
+		m_currentLimitError = 0;
+		m_enableMotor = false;
+	}
 
-    btRotationalLimitMotor(const btRotationalLimitMotor & limot)
-    {
-        m_targetVelocity = limot.m_targetVelocity;
-        m_maxMotorForce = limot.m_maxMotorForce;
-        m_limitSoftness = limot.m_limitSoftness;
-        m_loLimit = limot.m_loLimit;
-        m_hiLimit = limot.m_hiLimit;
+	btRotationalLimitMotor(const btRotationalLimitMotor& limot)
+	{
+		m_targetVelocity = limot.m_targetVelocity;
+		m_maxMotorForce = limot.m_maxMotorForce;
+		m_limitSoftness = limot.m_limitSoftness;
+		m_loLimit = limot.m_loLimit;
+		m_hiLimit = limot.m_hiLimit;
 		m_normalCFM = limot.m_normalCFM;
 		m_stopERP = limot.m_stopERP;
-		m_stopCFM =	limot.m_stopCFM;
-        m_bounce = limot.m_bounce;
-        m_currentLimit = limot.m_currentLimit;
-        m_currentLimitError = limot.m_currentLimitError;
-        m_enableMotor = limot.m_enableMotor;
-    }
-
-
+		m_stopCFM = limot.m_stopCFM;
+		m_bounce = limot.m_bounce;
+		m_currentLimit = limot.m_currentLimit;
+		m_currentLimitError = limot.m_currentLimitError;
+		m_enableMotor = limot.m_enableMotor;
+	}
 
 	//! Is limited
-    bool isLimited() const
-    {
-    	if(m_loLimit > m_hiLimit) return false;
-    	return true;
-    }
+	bool isLimited() const
+	{
+		if (m_loLimit > m_hiLimit) return false;
+		return true;
+	}
 
 	//! Need apply correction
-    bool needApplyTorques() const
-    {
-    	if(m_currentLimit == 0 && m_enableMotor == false) return false;
-    	return true;
-    }
+	bool needApplyTorques() const
+	{
+		if (m_currentLimit == 0 && m_enableMotor == false) return false;
+		return true;
+	}
 
 	//! calculates  error
 	/*!
@@ -131,104 +125,98 @@ public:
 	int testLimitValue(btScalar test_value);
 
 	//! apply the correction impulses for two bodies
-    btScalar solveAngularLimits(btScalar timeStep,btVector3& axis, btScalar jacDiagABInv,btRigidBody * body0, btRigidBody * body1);
-
+	btScalar solveAngularLimits(btScalar timeStep, btVector3& axis, btScalar jacDiagABInv, btRigidBody* body0, btRigidBody* body1);
 };
 
-
-
 class btTranslationalLimitMotor
 {
 public:
-	btVector3 m_lowerLimit;//!< the constraint lower limits
-    btVector3 m_upperLimit;//!< the constraint upper limits
-    btVector3 m_accumulatedImpulse;
-    //! Linear_Limit_parameters
-    //!@{
-    btScalar	m_limitSoftness;//!< Softness for linear limit
-    btScalar	m_damping;//!< Damping for linear limit
-    btScalar	m_restitution;//! Bounce parameter for linear limit
-	btVector3	m_normalCFM;//!< Constraint force mixing factor
-    btVector3	m_stopERP;//!< Error tolerance factor when joint is at limit
-	btVector3	m_stopCFM;//!< Constraint force mixing factor when joint is at limit
-    //!@}
-	bool		m_enableMotor[3];
-    btVector3	m_targetVelocity;//!< target motor velocity
-    btVector3	m_maxMotorForce;//!< max force on motor
-    btVector3	m_currentLimitError;//!  How much is violated this limit
-    btVector3	m_currentLinearDiff;//!  Current relative offset of constraint frames
-    int			m_currentLimit[3];//!< 0=free, 1=at lower limit, 2=at upper limit
+	btVector3 m_lowerLimit;  //!< the constraint lower limits
+	btVector3 m_upperLimit;  //!< the constraint upper limits
+	btVector3 m_accumulatedImpulse;
+	//! Linear_Limit_parameters
+	//!@{
+	btScalar m_limitSoftness;  //!< Softness for linear limit
+	btScalar m_damping;        //!< Damping for linear limit
+	btScalar m_restitution;    //! Bounce parameter for linear limit
+	btVector3 m_normalCFM;     //!< Constraint force mixing factor
+	btVector3 m_stopERP;       //!< Error tolerance factor when joint is at limit
+	btVector3 m_stopCFM;       //!< Constraint force mixing factor when joint is at limit
+							   //!@}
+	bool m_enableMotor[3];
+	btVector3 m_targetVelocity;     //!< target motor velocity
+	btVector3 m_maxMotorForce;      //!< max force on motor
+	btVector3 m_currentLimitError;  //!  How much is violated this limit
+	btVector3 m_currentLinearDiff;  //!  Current relative offset of constraint frames
+	int m_currentLimit[3];          //!< 0=free, 1=at lower limit, 2=at upper limit
 
-    btTranslationalLimitMotor()
-    {
-    	m_lowerLimit.setValue(0.f,0.f,0.f);
-    	m_upperLimit.setValue(0.f,0.f,0.f);
-    	m_accumulatedImpulse.setValue(0.f,0.f,0.f);
+	btTranslationalLimitMotor()
+	{
+		m_lowerLimit.setValue(0.f, 0.f, 0.f);
+		m_upperLimit.setValue(0.f, 0.f, 0.f);
+		m_accumulatedImpulse.setValue(0.f, 0.f, 0.f);
 		m_normalCFM.setValue(0.f, 0.f, 0.f);
 		m_stopERP.setValue(0.2f, 0.2f, 0.2f);
 		m_stopCFM.setValue(0.f, 0.f, 0.f);
 
-    	m_limitSoftness = 0.7f;
-    	m_damping = btScalar(1.0f);
-    	m_restitution = btScalar(0.5f);
-		for(int i=0; i < 3; i++) 
+		m_limitSoftness = 0.7f;
+		m_damping = btScalar(1.0f);
+		m_restitution = btScalar(0.5f);
+		for (int i = 0; i < 3; i++)
 		{
 			m_enableMotor[i] = false;
 			m_targetVelocity[i] = btScalar(0.f);
 			m_maxMotorForce[i] = btScalar(0.f);
 		}
-    }
+	}
 
-    btTranslationalLimitMotor(const btTranslationalLimitMotor & other )
-    {
-    	m_lowerLimit = other.m_lowerLimit;
-    	m_upperLimit = other.m_upperLimit;
-    	m_accumulatedImpulse = other.m_accumulatedImpulse;
+	btTranslationalLimitMotor(const btTranslationalLimitMotor& other)
+	{
+		m_lowerLimit = other.m_lowerLimit;
+		m_upperLimit = other.m_upperLimit;
+		m_accumulatedImpulse = other.m_accumulatedImpulse;
 
-    	m_limitSoftness = other.m_limitSoftness ;
-    	m_damping = other.m_damping;
-    	m_restitution = other.m_restitution;
+		m_limitSoftness = other.m_limitSoftness;
+		m_damping = other.m_damping;
+		m_restitution = other.m_restitution;
 		m_normalCFM = other.m_normalCFM;
 		m_stopERP = other.m_stopERP;
 		m_stopCFM = other.m_stopCFM;
 
-		for(int i=0; i < 3; i++) 
+		for (int i = 0; i < 3; i++)
 		{
 			m_enableMotor[i] = other.m_enableMotor[i];
 			m_targetVelocity[i] = other.m_targetVelocity[i];
 			m_maxMotorForce[i] = other.m_maxMotorForce[i];
 		}
-    }
+	}
 
-    //! Test limit
+	//! Test limit
 	/*!
     - free means upper < lower,
     - locked means upper == lower
     - limited means upper > lower
     - limitIndex: first 3 are linear, next 3 are angular
     */
-    inline bool	isLimited(int limitIndex) const
-    {
-       return (m_upperLimit[limitIndex] >= m_lowerLimit[limitIndex]);
-    }
-    inline bool needApplyForce(int limitIndex) const
-    {
-    	if(m_currentLimit[limitIndex] == 0 && m_enableMotor[limitIndex] == false) return false;
-    	return true;
-    }
+	inline bool isLimited(int limitIndex) const
+	{
+		return (m_upperLimit[limitIndex] >= m_lowerLimit[limitIndex]);
+	}
+	inline bool needApplyForce(int limitIndex) const
+	{
+		if (m_currentLimit[limitIndex] == 0 && m_enableMotor[limitIndex] == false) return false;
+		return true;
+	}
 	int testLimitValue(int limitIndex, btScalar test_value);
 
-
-    btScalar solveLinearAxis(
-    	btScalar timeStep,
-        btScalar jacDiagABInv,
-        btRigidBody& body1,const btVector3 &pointInA,
-        btRigidBody& body2,const btVector3 &pointInB,
-        int limit_index,
-        const btVector3 & axis_normal_on_a,
-		const btVector3 & anchorPos);
-
-
+	btScalar solveLinearAxis(
+		btScalar timeStep,
+		btScalar jacDiagABInv,
+		btRigidBody& body1, const btVector3& pointInA,
+		btRigidBody& body2, const btVector3& pointInB,
+		int limit_index,
+		const btVector3& axis_normal_on_a,
+		const btVector3& anchorPos);
 };
 
 enum bt6DofFlags
@@ -237,8 +225,7 @@ enum bt6DofFlags
 	BT_6DOF_FLAGS_CFM_STOP = 2,
 	BT_6DOF_FLAGS_ERP_STOP = 4
 };
-#define BT_6DOF_FLAGS_AXIS_SHIFT 3 // bits per axis
-
+#define BT_6DOF_FLAGS_AXIS_SHIFT 3  // bits per axis
 
 /// btGeneric6DofConstraint between two rigidbodies each with a pivotpoint that descibes the axis location in local space
 /*!
@@ -276,254 +263,245 @@ This brings support for limit parameters and motors. </li>
 </ul>
 
 */
-ATTRIBUTE_ALIGNED16(class) btGeneric6DofConstraint : public btTypedConstraint
+ATTRIBUTE_ALIGNED16(class)
+btGeneric6DofConstraint : public btTypedConstraint
 {
 protected:
-
 	//! relative_frames
-    //!@{
-	btTransform	m_frameInA;//!< the constraint space w.r.t body A
-    btTransform	m_frameInB;//!< the constraint space w.r.t body B
-    //!@}
-
-    //! Jacobians
-    //!@{
-    btJacobianEntry	m_jacLinear[3];//!< 3 orthogonal linear constraints
-    btJacobianEntry	m_jacAng[3];//!< 3 orthogonal angular constraints
-    //!@}
-
-	//! Linear_Limit_parameters
-    //!@{
-    btTranslationalLimitMotor m_linearLimits;
-    //!@}
-
-
-    //! hinge_parameters
-    //!@{
-    btRotationalLimitMotor m_angularLimits[3];
+	//!@{
+	btTransform m_frameInA;  //!< the constraint space w.r.t body A
+	btTransform m_frameInB;  //!< the constraint space w.r.t body B
 	//!@}
 
+	//! Jacobians
+	//!@{
+	btJacobianEntry m_jacLinear[3];  //!< 3 orthogonal linear constraints
+	btJacobianEntry m_jacAng[3];     //!< 3 orthogonal angular constraints
+	//!@}
+
+	//! Linear_Limit_parameters
+	//!@{
+	btTranslationalLimitMotor m_linearLimits;
+	//!@}
+
+	//! hinge_parameters
+	//!@{
+	btRotationalLimitMotor m_angularLimits[3];
+	//!@}
 
 protected:
-    //! temporal variables
-    //!@{
-    btScalar m_timeStep;
-    btTransform m_calculatedTransformA;
-    btTransform m_calculatedTransformB;
-    btVector3 m_calculatedAxisAngleDiff;
-    btVector3 m_calculatedAxis[3];
-    btVector3 m_calculatedLinearDiff;
-	btScalar	m_factA;
-	btScalar	m_factB;
-	bool		m_hasStaticBody;
-    
-	btVector3 m_AnchorPos; // point betwen pivots of bodies A and B to solve linear axes
+	//! temporal variables
+	//!@{
+	btScalar m_timeStep;
+	btTransform m_calculatedTransformA;
+	btTransform m_calculatedTransformB;
+	btVector3 m_calculatedAxisAngleDiff;
+	btVector3 m_calculatedAxis[3];
+	btVector3 m_calculatedLinearDiff;
+	btScalar m_factA;
+	btScalar m_factB;
+	bool m_hasStaticBody;
 
-    bool	m_useLinearReferenceFrameA;
-	bool	m_useOffsetForConstraintFrame;
-    
-	int		m_flags;
+	btVector3 m_AnchorPos;  // point betwen pivots of bodies A and B to solve linear axes
 
-    //!@}
+	bool m_useLinearReferenceFrameA;
+	bool m_useOffsetForConstraintFrame;
 
-    btGeneric6DofConstraint&	operator=(btGeneric6DofConstraint&	other)
-    {
-        btAssert(0);
-        (void) other;
-        return *this;
-    }
+	int m_flags;
 
+	//!@}
 
-	int setAngularLimits(btConstraintInfo2 *info, int row_offset,const btTransform& transA,const btTransform& transB,const btVector3& linVelA,const btVector3& linVelB,const btVector3& angVelA,const btVector3& angVelB);
+	btGeneric6DofConstraint& operator=(btGeneric6DofConstraint& other)
+	{
+		btAssert(0);
+		(void)other;
+		return *this;
+	}
 
-	int setLinearLimits(btConstraintInfo2 *info, int row, const btTransform& transA,const btTransform& transB,const btVector3& linVelA,const btVector3& linVelB,const btVector3& angVelA,const btVector3& angVelB);
+	int setAngularLimits(btConstraintInfo2 * info, int row_offset, const btTransform& transA, const btTransform& transB, const btVector3& linVelA, const btVector3& linVelB, const btVector3& angVelA, const btVector3& angVelB);
 
-    void buildLinearJacobian(
-        btJacobianEntry & jacLinear,const btVector3 & normalWorld,
-        const btVector3 & pivotAInW,const btVector3 & pivotBInW);
+	int setLinearLimits(btConstraintInfo2 * info, int row, const btTransform& transA, const btTransform& transB, const btVector3& linVelA, const btVector3& linVelB, const btVector3& angVelA, const btVector3& angVelB);
 
-    void buildAngularJacobian(btJacobianEntry & jacAngular,const btVector3 & jointAxisW);
+	void buildLinearJacobian(
+		btJacobianEntry & jacLinear, const btVector3& normalWorld,
+		const btVector3& pivotAInW, const btVector3& pivotBInW);
+
+	void buildAngularJacobian(btJacobianEntry & jacAngular, const btVector3& jointAxisW);
 
 	// tests linear limits
 	void calculateLinearInfo();
 
 	//! calcs the euler angles between the two bodies.
-    void calculateAngleInfo();
-
-
+	void calculateAngleInfo();
 
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
-	///for backwards compatibility during the transition to 'getInfo/getInfo2'
-	bool		m_useSolveConstraintObsolete;
 
-    btGeneric6DofConstraint(btRigidBody& rbA, btRigidBody& rbB, const btTransform& frameInA, const btTransform& frameInB ,bool useLinearReferenceFrameA);
-    btGeneric6DofConstraint(btRigidBody& rbB, const btTransform& frameInB, bool useLinearReferenceFrameB);
-    
+	///for backwards compatibility during the transition to 'getInfo/getInfo2'
+	bool m_useSolveConstraintObsolete;
+
+	btGeneric6DofConstraint(btRigidBody & rbA, btRigidBody & rbB, const btTransform& frameInA, const btTransform& frameInB, bool useLinearReferenceFrameA);
+	btGeneric6DofConstraint(btRigidBody & rbB, const btTransform& frameInB, bool useLinearReferenceFrameB);
+
 	//! Calcs global transform of the offsets
 	/*!
 	Calcs the global transform for the joint offset for body A an B, and also calcs the agle differences between the bodies.
 	\sa btGeneric6DofConstraint.getCalculatedTransformA , btGeneric6DofConstraint.getCalculatedTransformB, btGeneric6DofConstraint.calculateAngleInfo
 	*/
-    void calculateTransforms(const btTransform& transA,const btTransform& transB);
+	void calculateTransforms(const btTransform& transA, const btTransform& transB);
 
 	void calculateTransforms();
 
 	//! Gets the global transform of the offset for body A
-    /*!
+	/*!
     \sa btGeneric6DofConstraint.getFrameOffsetA, btGeneric6DofConstraint.getFrameOffsetB, btGeneric6DofConstraint.calculateAngleInfo.
     */
-    const btTransform & getCalculatedTransformA() const
-    {
-    	return m_calculatedTransformA;
-    }
+	const btTransform& getCalculatedTransformA() const
+	{
+		return m_calculatedTransformA;
+	}
 
-    //! Gets the global transform of the offset for body B
-    /*!
+	//! Gets the global transform of the offset for body B
+	/*!
     \sa btGeneric6DofConstraint.getFrameOffsetA, btGeneric6DofConstraint.getFrameOffsetB, btGeneric6DofConstraint.calculateAngleInfo.
     */
-    const btTransform & getCalculatedTransformB() const
-    {
-    	return m_calculatedTransformB;
-    }
+	const btTransform& getCalculatedTransformB() const
+	{
+		return m_calculatedTransformB;
+	}
 
-    const btTransform & getFrameOffsetA() const
-    {
-    	return m_frameInA;
-    }
+	const btTransform& getFrameOffsetA() const
+	{
+		return m_frameInA;
+	}
 
-    const btTransform & getFrameOffsetB() const
-    {
-    	return m_frameInB;
-    }
+	const btTransform& getFrameOffsetB() const
+	{
+		return m_frameInB;
+	}
 
+	btTransform& getFrameOffsetA()
+	{
+		return m_frameInA;
+	}
 
-    btTransform & getFrameOffsetA()
-    {
-    	return m_frameInA;
-    }
-
-    btTransform & getFrameOffsetB()
-    {
-    	return m_frameInB;
-    }
-
+	btTransform& getFrameOffsetB()
+	{
+		return m_frameInB;
+	}
 
 	//! performs Jacobian calculation, and also calculates angle differences and axis
-    virtual void	buildJacobian();
+	virtual void buildJacobian();
 
-	virtual void getInfo1 (btConstraintInfo1* info);
+	virtual void getInfo1(btConstraintInfo1 * info);
 
-	void getInfo1NonVirtual (btConstraintInfo1* info);
+	void getInfo1NonVirtual(btConstraintInfo1 * info);
 
-	virtual void getInfo2 (btConstraintInfo2* info);
+	virtual void getInfo2(btConstraintInfo2 * info);
 
-	void getInfo2NonVirtual (btConstraintInfo2* info,const btTransform& transA,const btTransform& transB,const btVector3& linVelA,const btVector3& linVelB,const btVector3& angVelA,const btVector3& angVelB);
+	void getInfo2NonVirtual(btConstraintInfo2 * info, const btTransform& transA, const btTransform& transB, const btVector3& linVelA, const btVector3& linVelB, const btVector3& angVelA, const btVector3& angVelB);
 
-
-    void	updateRHS(btScalar	timeStep);
+	void updateRHS(btScalar timeStep);
 
 	//! Get the rotation axis in global coordinates
 	/*!
 	\pre btGeneric6DofConstraint.buildJacobian must be called previously.
 	*/
-    btVector3 getAxis(int axis_index) const;
+	btVector3 getAxis(int axis_index) const;
 
-    //! Get the relative Euler angle
-    /*!
+	//! Get the relative Euler angle
+	/*!
 	\pre btGeneric6DofConstraint::calculateTransforms() must be called previously.
 	*/
-    btScalar getAngle(int axis_index) const;
+	btScalar getAngle(int axis_index) const;
 
 	//! Get the relative position of the constraint pivot
-    /*!
+	/*!
 	\pre btGeneric6DofConstraint::calculateTransforms() must be called previously.
 	*/
 	btScalar getRelativePivotPosition(int axis_index) const;
 
-	void setFrames(const btTransform & frameA, const btTransform & frameB);
+	void setFrames(const btTransform& frameA, const btTransform& frameB);
 
 	//! Test angular limit.
 	/*!
 	Calculates angular correction and returns true if limit needs to be corrected.
 	\pre btGeneric6DofConstraint::calculateTransforms() must be called previously.
 	*/
-    bool testAngularLimitMotor(int axis_index);
+	bool testAngularLimitMotor(int axis_index);
 
-    void	setLinearLowerLimit(const btVector3& linearLower)
-    {
-    	m_linearLimits.m_lowerLimit = linearLower;
-    }
+	void setLinearLowerLimit(const btVector3& linearLower)
+	{
+		m_linearLimits.m_lowerLimit = linearLower;
+	}
 
-	void	getLinearLowerLimit(btVector3& linearLower) const
+	void getLinearLowerLimit(btVector3 & linearLower) const
 	{
 		linearLower = m_linearLimits.m_lowerLimit;
 	}
 
-	void	setLinearUpperLimit(const btVector3& linearUpper)
+	void setLinearUpperLimit(const btVector3& linearUpper)
 	{
 		m_linearLimits.m_upperLimit = linearUpper;
 	}
 
-	void	getLinearUpperLimit(btVector3& linearUpper) const
+	void getLinearUpperLimit(btVector3 & linearUpper) const
 	{
 		linearUpper = m_linearLimits.m_upperLimit;
 	}
 
-    void	setAngularLowerLimit(const btVector3& angularLower)
-    {
-		for(int i = 0; i < 3; i++) 
-			m_angularLimits[i].m_loLimit = btNormalizeAngle(angularLower[i]);
-    }
-
-	void	getAngularLowerLimit(btVector3& angularLower) const
+	void setAngularLowerLimit(const btVector3& angularLower)
 	{
-		for(int i = 0; i < 3; i++) 
+		for (int i = 0; i < 3; i++)
+			m_angularLimits[i].m_loLimit = btNormalizeAngle(angularLower[i]);
+	}
+
+	void getAngularLowerLimit(btVector3 & angularLower) const
+	{
+		for (int i = 0; i < 3; i++)
 			angularLower[i] = m_angularLimits[i].m_loLimit;
 	}
 
-    void	setAngularUpperLimit(const btVector3& angularUpper)
-    {
-		for(int i = 0; i < 3; i++)
-			m_angularLimits[i].m_hiLimit = btNormalizeAngle(angularUpper[i]);
-    }
-
-	void	getAngularUpperLimit(btVector3& angularUpper) const
+	void setAngularUpperLimit(const btVector3& angularUpper)
 	{
-		for(int i = 0; i < 3; i++)
+		for (int i = 0; i < 3; i++)
+			m_angularLimits[i].m_hiLimit = btNormalizeAngle(angularUpper[i]);
+	}
+
+	void getAngularUpperLimit(btVector3 & angularUpper) const
+	{
+		for (int i = 0; i < 3; i++)
 			angularUpper[i] = m_angularLimits[i].m_hiLimit;
 	}
 
 	//! Retrieves the angular limit informacion
-    btRotationalLimitMotor * getRotationalLimitMotor(int index)
-    {
-    	return &m_angularLimits[index];
-    }
+	btRotationalLimitMotor* getRotationalLimitMotor(int index)
+	{
+		return &m_angularLimits[index];
+	}
 
-    //! Retrieves the  limit informacion
-    btTranslationalLimitMotor * getTranslationalLimitMotor()
-    {
-    	return &m_linearLimits;
-    }
+	//! Retrieves the  limit informacion
+	btTranslationalLimitMotor* getTranslationalLimitMotor()
+	{
+		return &m_linearLimits;
+	}
 
-    //first 3 are linear, next 3 are angular
-    void setLimit(int axis, btScalar lo, btScalar hi)
-    {
-    	if(axis<3)
-    	{
-    		m_linearLimits.m_lowerLimit[axis] = lo;
-    		m_linearLimits.m_upperLimit[axis] = hi;
-    	}
-    	else
-    	{
+	//first 3 are linear, next 3 are angular
+	void setLimit(int axis, btScalar lo, btScalar hi)
+	{
+		if (axis < 3)
+		{
+			m_linearLimits.m_lowerLimit[axis] = lo;
+			m_linearLimits.m_upperLimit[axis] = hi;
+		}
+		else
+		{
 			lo = btNormalizeAngle(lo);
 			hi = btNormalizeAngle(hi);
-    		m_angularLimits[axis-3].m_loLimit = lo;
-    		m_angularLimits[axis-3].m_hiLimit = hi;
-    	}
-    }
+			m_angularLimits[axis - 3].m_loLimit = lo;
+			m_angularLimits[axis - 3].m_hiLimit = hi;
+		}
+	}
 
 	//! Test limit
 	/*!
@@ -532,116 +510,106 @@ public:
     - limited means upper > lower
     - limitIndex: first 3 are linear, next 3 are angular
     */
-    bool	isLimited(int limitIndex) const
-    {
-    	if(limitIndex<3)
-    	{
+	bool isLimited(int limitIndex) const
+	{
+		if (limitIndex < 3)
+		{
 			return m_linearLimits.isLimited(limitIndex);
+		}
+		return m_angularLimits[limitIndex - 3].isLimited();
+	}
 
-    	}
-        return m_angularLimits[limitIndex-3].isLimited();
-    }
+	virtual void calcAnchorPos(void);  // overridable
 
-	virtual void calcAnchorPos(void); // overridable
-
-	int get_limit_motor_info2(	btRotationalLimitMotor * limot,
-								const btTransform& transA,const btTransform& transB,const btVector3& linVelA,const btVector3& linVelB,const btVector3& angVelA,const btVector3& angVelB,
-								btConstraintInfo2 *info, int row, btVector3& ax1, int rotational, int rotAllowed = false);
+	int get_limit_motor_info2(btRotationalLimitMotor * limot,
+							  const btTransform& transA, const btTransform& transB, const btVector3& linVelA, const btVector3& linVelB, const btVector3& angVelA, const btVector3& angVelB,
+							  btConstraintInfo2* info, int row, btVector3& ax1, int rotational, int rotAllowed = false);
 
 	// access for UseFrameOffset
 	bool getUseFrameOffset() const { return m_useOffsetForConstraintFrame; }
 	void setUseFrameOffset(bool frameOffsetOnOff) { m_useOffsetForConstraintFrame = frameOffsetOnOff; }
-	
+
 	bool getUseLinearReferenceFrameA() const { return m_useLinearReferenceFrameA; }
 	void setUseLinearReferenceFrameA(bool linearReferenceFrameA) { m_useLinearReferenceFrameA = linearReferenceFrameA; }
 
-	///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5). 
+	///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5).
 	///If no axis is provided, it uses the default axis for this constraint.
-	virtual	void setParam(int num, btScalar value, int axis = -1);
+	virtual void setParam(int num, btScalar value, int axis = -1);
 	///return the local value of parameter
-	virtual	btScalar getParam(int num, int axis = -1) const;
+	virtual btScalar getParam(int num, int axis = -1) const;
 
-	void setAxis( const btVector3& axis1, const btVector3& axis2);
+	void setAxis(const btVector3& axis1, const btVector3& axis2);
 
-    	virtual	int getFlags() const
-    	{
-        	return m_flags;
+	virtual int getFlags() const
+	{
+		return m_flags;
 	}
 
-	virtual	int	calculateSerializeBufferSize() const;
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
-
-	
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 };
 
-
 struct btGeneric6DofConstraintData
 {
-	btTypedConstraintData	m_typeConstraintData;
-	btTransformFloatData m_rbAFrame; // constraint axii. Assumes z is hinge axis.
+	btTypedConstraintData m_typeConstraintData;
+	btTransformFloatData m_rbAFrame;  // constraint axii. Assumes z is hinge axis.
 	btTransformFloatData m_rbBFrame;
-	
-	btVector3FloatData	m_linearUpperLimit;
-	btVector3FloatData	m_linearLowerLimit;
 
-	btVector3FloatData	m_angularUpperLimit;
-	btVector3FloatData	m_angularLowerLimit;
-	
-	int	m_useLinearReferenceFrameA;
+	btVector3FloatData m_linearUpperLimit;
+	btVector3FloatData m_linearLowerLimit;
+
+	btVector3FloatData m_angularUpperLimit;
+	btVector3FloatData m_angularLowerLimit;
+
+	int m_useLinearReferenceFrameA;
 	int m_useOffsetForConstraintFrame;
 };
 
 struct btGeneric6DofConstraintDoubleData2
 {
-	btTypedConstraintDoubleData	m_typeConstraintData;
-	btTransformDoubleData m_rbAFrame; // constraint axii. Assumes z is hinge axis.
+	btTypedConstraintDoubleData m_typeConstraintData;
+	btTransformDoubleData m_rbAFrame;  // constraint axii. Assumes z is hinge axis.
 	btTransformDoubleData m_rbBFrame;
-	
-	btVector3DoubleData	m_linearUpperLimit;
-	btVector3DoubleData	m_linearLowerLimit;
 
-	btVector3DoubleData	m_angularUpperLimit;
-	btVector3DoubleData	m_angularLowerLimit;
-	
-	int	m_useLinearReferenceFrameA;
+	btVector3DoubleData m_linearUpperLimit;
+	btVector3DoubleData m_linearLowerLimit;
+
+	btVector3DoubleData m_angularUpperLimit;
+	btVector3DoubleData m_angularLowerLimit;
+
+	int m_useLinearReferenceFrameA;
 	int m_useOffsetForConstraintFrame;
 };
 
-SIMD_FORCE_INLINE	int	btGeneric6DofConstraint::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btGeneric6DofConstraint::calculateSerializeBufferSize() const
 {
 	return sizeof(btGeneric6DofConstraintData2);
 }
 
-	///fills the dataBuffer and returns the struct name (and 0 on failure)
-SIMD_FORCE_INLINE	const char*	btGeneric6DofConstraint::serialize(void* dataBuffer, btSerializer* serializer) const
+///fills the dataBuffer and returns the struct name (and 0 on failure)
+SIMD_FORCE_INLINE const char* btGeneric6DofConstraint::serialize(void* dataBuffer, btSerializer* serializer) const
 {
-
 	btGeneric6DofConstraintData2* dof = (btGeneric6DofConstraintData2*)dataBuffer;
-	btTypedConstraint::serialize(&dof->m_typeConstraintData,serializer);
+	btTypedConstraint::serialize(&dof->m_typeConstraintData, serializer);
 
 	m_frameInA.serialize(dof->m_rbAFrame);
 	m_frameInB.serialize(dof->m_rbBFrame);
 
-		
 	int i;
-	for (i=0;i<3;i++)
+	for (i = 0; i < 3; i++)
 	{
-		dof->m_angularLowerLimit.m_floats[i] =  m_angularLimits[i].m_loLimit;
-		dof->m_angularUpperLimit.m_floats[i] =  m_angularLimits[i].m_hiLimit;
+		dof->m_angularLowerLimit.m_floats[i] = m_angularLimits[i].m_loLimit;
+		dof->m_angularUpperLimit.m_floats[i] = m_angularLimits[i].m_hiLimit;
 		dof->m_linearLowerLimit.m_floats[i] = m_linearLimits.m_lowerLimit[i];
 		dof->m_linearUpperLimit.m_floats[i] = m_linearLimits.m_upperLimit[i];
 	}
-	
-	dof->m_useLinearReferenceFrameA = m_useLinearReferenceFrameA? 1 : 0;
+
+	dof->m_useLinearReferenceFrameA = m_useLinearReferenceFrameA ? 1 : 0;
 	dof->m_useOffsetForConstraintFrame = m_useOffsetForConstraintFrame ? 1 : 0;
 
 	return btGeneric6DofConstraintDataName;
 }
 
-
-
-
-
-#endif //BT_GENERIC_6DOF_CONSTRAINT_H
+#endif  //BT_GENERIC_6DOF_CONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofSpring2Constraint.cpp b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofSpring2Constraint.cpp
index 540dcd18f71..49c8d9bbf79 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofSpring2Constraint.cpp
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofSpring2Constraint.cpp
@@ -37,67 +37,54 @@ email: projectileman@yahoo.com
 http://gimpact.sf.net
 */
 
-
-
 #include "btGeneric6DofSpring2Constraint.h"
 #include "BulletDynamics/Dynamics/btRigidBody.h"
 #include "LinearMath/btTransformUtil.h"
 #include <new>
 
-
-
 btGeneric6DofSpring2Constraint::btGeneric6DofSpring2Constraint(btRigidBody& rbA, btRigidBody& rbB, const btTransform& frameInA, const btTransform& frameInB, RotateOrder rotOrder)
-	: btTypedConstraint(D6_SPRING_2_CONSTRAINT_TYPE, rbA, rbB)
-	, m_frameInA(frameInA)
-	, m_frameInB(frameInB)
-	, m_rotateOrder(rotOrder)	
-	, m_flags(0)
+	: btTypedConstraint(D6_SPRING_2_CONSTRAINT_TYPE, rbA, rbB), m_frameInA(frameInA), m_frameInB(frameInB), m_rotateOrder(rotOrder), m_flags(0)
 {
 	calculateTransforms();
 }
 
-
 btGeneric6DofSpring2Constraint::btGeneric6DofSpring2Constraint(btRigidBody& rbB, const btTransform& frameInB, RotateOrder rotOrder)
-	: btTypedConstraint(D6_SPRING_2_CONSTRAINT_TYPE, getFixedBody(), rbB)
-	, m_frameInB(frameInB)
-	, m_rotateOrder(rotOrder)
-	, m_flags(0)
+	: btTypedConstraint(D6_SPRING_2_CONSTRAINT_TYPE, getFixedBody(), rbB), m_frameInB(frameInB), m_rotateOrder(rotOrder), m_flags(0)
 {
 	///not providing rigidbody A means implicitly using worldspace for body A
 	m_frameInA = rbB.getCenterOfMassTransform() * m_frameInB;
 	calculateTransforms();
 }
 
-
 btScalar btGeneric6DofSpring2Constraint::btGetMatrixElem(const btMatrix3x3& mat, int index)
 {
-	int i = index%3;
-	int j = index/3;
+	int i = index % 3;
+	int j = index / 3;
 	return mat[i][j];
 }
 
 // MatrixToEulerXYZ from http://www.geometrictools.com/LibFoundation/Mathematics/Wm4Matrix3.inl.html
 
-bool btGeneric6DofSpring2Constraint::matrixToEulerXYZ(const btMatrix3x3& mat,btVector3& xyz)
+bool btGeneric6DofSpring2Constraint::matrixToEulerXYZ(const btMatrix3x3& mat, btVector3& xyz)
 {
 	// rot =  cy*cz          -cy*sz           sy
 	//        cz*sx*sy+cx*sz  cx*cz-sx*sy*sz -cy*sx
 	//       -cx*cz*sy+sx*sz  cz*sx+cx*sy*sz  cx*cy
 
-	btScalar fi = btGetMatrixElem(mat,2);
+	btScalar fi = btGetMatrixElem(mat, 2);
 	if (fi < btScalar(1.0f))
 	{
 		if (fi > btScalar(-1.0f))
 		{
-			xyz[0] = btAtan2(-btGetMatrixElem(mat,5),btGetMatrixElem(mat,8));
-			xyz[1] = btAsin(btGetMatrixElem(mat,2));
-			xyz[2] = btAtan2(-btGetMatrixElem(mat,1),btGetMatrixElem(mat,0));
+			xyz[0] = btAtan2(-btGetMatrixElem(mat, 5), btGetMatrixElem(mat, 8));
+			xyz[1] = btAsin(btGetMatrixElem(mat, 2));
+			xyz[2] = btAtan2(-btGetMatrixElem(mat, 1), btGetMatrixElem(mat, 0));
 			return true;
 		}
 		else
 		{
 			// WARNING.  Not unique.  XA - ZA = -atan2(r10,r11)
-			xyz[0] = -btAtan2(btGetMatrixElem(mat,3),btGetMatrixElem(mat,4));
+			xyz[0] = -btAtan2(btGetMatrixElem(mat, 3), btGetMatrixElem(mat, 4));
 			xyz[1] = -SIMD_HALF_PI;
 			xyz[2] = btScalar(0.0);
 			return false;
@@ -106,32 +93,32 @@ bool btGeneric6DofSpring2Constraint::matrixToEulerXYZ(const btMatrix3x3& mat,btV
 	else
 	{
 		// WARNING.  Not unique.  XAngle + ZAngle = atan2(r10,r11)
-		xyz[0] = btAtan2(btGetMatrixElem(mat,3),btGetMatrixElem(mat,4));
+		xyz[0] = btAtan2(btGetMatrixElem(mat, 3), btGetMatrixElem(mat, 4));
 		xyz[1] = SIMD_HALF_PI;
 		xyz[2] = 0.0;
 	}
 	return false;
 }
 
-bool btGeneric6DofSpring2Constraint::matrixToEulerXZY(const btMatrix3x3& mat,btVector3& xyz)
+bool btGeneric6DofSpring2Constraint::matrixToEulerXZY(const btMatrix3x3& mat, btVector3& xyz)
 {
 	// rot =  cy*cz          -sz           sy*cz
 	//        cy*cx*sz+sx*sy  cx*cz        sy*cx*sz-cy*sx
 	//        cy*sx*sz-cx*sy  sx*cz        sy*sx*sz+cx*cy
 
-	btScalar fi = btGetMatrixElem(mat,1);
+	btScalar fi = btGetMatrixElem(mat, 1);
 	if (fi < btScalar(1.0f))
 	{
 		if (fi > btScalar(-1.0f))
 		{
-			xyz[0] = btAtan2(btGetMatrixElem(mat,7),btGetMatrixElem(mat,4));
-			xyz[1] = btAtan2(btGetMatrixElem(mat,2),btGetMatrixElem(mat,0));
-			xyz[2] = btAsin(-btGetMatrixElem(mat,1));
+			xyz[0] = btAtan2(btGetMatrixElem(mat, 7), btGetMatrixElem(mat, 4));
+			xyz[1] = btAtan2(btGetMatrixElem(mat, 2), btGetMatrixElem(mat, 0));
+			xyz[2] = btAsin(-btGetMatrixElem(mat, 1));
 			return true;
 		}
 		else
 		{
-			xyz[0] = -btAtan2(-btGetMatrixElem(mat,6),btGetMatrixElem(mat,8));
+			xyz[0] = -btAtan2(-btGetMatrixElem(mat, 6), btGetMatrixElem(mat, 8));
 			xyz[1] = btScalar(0.0);
 			xyz[2] = SIMD_HALF_PI;
 			return false;
@@ -139,33 +126,33 @@ bool btGeneric6DofSpring2Constraint::matrixToEulerXZY(const btMatrix3x3& mat,btV
 	}
 	else
 	{
-		xyz[0] = btAtan2(-btGetMatrixElem(mat,6),btGetMatrixElem(mat,8));
+		xyz[0] = btAtan2(-btGetMatrixElem(mat, 6), btGetMatrixElem(mat, 8));
 		xyz[1] = 0.0;
 		xyz[2] = -SIMD_HALF_PI;
 	}
 	return false;
 }
 
-bool btGeneric6DofSpring2Constraint::matrixToEulerYXZ(const btMatrix3x3& mat,btVector3& xyz)
+bool btGeneric6DofSpring2Constraint::matrixToEulerYXZ(const btMatrix3x3& mat, btVector3& xyz)
 {
 	// rot =  cy*cz+sy*sx*sz  cz*sy*sx-cy*sz  cx*sy
 	//        cx*sz           cx*cz           -sx
 	//        cy*sx*sz-cz*sy  sy*sz+cy*cz*sx  cy*cx
 
-	btScalar fi = btGetMatrixElem(mat,5);
+	btScalar fi = btGetMatrixElem(mat, 5);
 	if (fi < btScalar(1.0f))
 	{
 		if (fi > btScalar(-1.0f))
 		{
-			xyz[0] = btAsin(-btGetMatrixElem(mat,5));
-			xyz[1] = btAtan2(btGetMatrixElem(mat,2),btGetMatrixElem(mat,8));
-			xyz[2] = btAtan2(btGetMatrixElem(mat,3),btGetMatrixElem(mat,4));
+			xyz[0] = btAsin(-btGetMatrixElem(mat, 5));
+			xyz[1] = btAtan2(btGetMatrixElem(mat, 2), btGetMatrixElem(mat, 8));
+			xyz[2] = btAtan2(btGetMatrixElem(mat, 3), btGetMatrixElem(mat, 4));
 			return true;
 		}
 		else
 		{
 			xyz[0] = SIMD_HALF_PI;
-			xyz[1] = -btAtan2(-btGetMatrixElem(mat,1),btGetMatrixElem(mat,0));
+			xyz[1] = -btAtan2(-btGetMatrixElem(mat, 1), btGetMatrixElem(mat, 0));
 			xyz[2] = btScalar(0.0);
 			return false;
 		}
@@ -173,32 +160,32 @@ bool btGeneric6DofSpring2Constraint::matrixToEulerYXZ(const btMatrix3x3& mat,btV
 	else
 	{
 		xyz[0] = -SIMD_HALF_PI;
-		xyz[1] = btAtan2(-btGetMatrixElem(mat,1),btGetMatrixElem(mat,0));
+		xyz[1] = btAtan2(-btGetMatrixElem(mat, 1), btGetMatrixElem(mat, 0));
 		xyz[2] = 0.0;
 	}
 	return false;
 }
 
-bool btGeneric6DofSpring2Constraint::matrixToEulerYZX(const btMatrix3x3& mat,btVector3& xyz)
+bool btGeneric6DofSpring2Constraint::matrixToEulerYZX(const btMatrix3x3& mat, btVector3& xyz)
 {
 	// rot =  cy*cz   sy*sx-cy*cx*sz   cx*sy+cy*sz*sx
 	//        sz           cz*cx           -cz*sx
 	//        -cz*sy  cy*sx+cx*sy*sz   cy*cx-sy*sz*sx
 
-	btScalar fi = btGetMatrixElem(mat,3);
+	btScalar fi = btGetMatrixElem(mat, 3);
 	if (fi < btScalar(1.0f))
 	{
 		if (fi > btScalar(-1.0f))
 		{
-			xyz[0] = btAtan2(-btGetMatrixElem(mat,5),btGetMatrixElem(mat,4));
-			xyz[1] = btAtan2(-btGetMatrixElem(mat,6),btGetMatrixElem(mat,0));
-			xyz[2] = btAsin(btGetMatrixElem(mat,3));
+			xyz[0] = btAtan2(-btGetMatrixElem(mat, 5), btGetMatrixElem(mat, 4));
+			xyz[1] = btAtan2(-btGetMatrixElem(mat, 6), btGetMatrixElem(mat, 0));
+			xyz[2] = btAsin(btGetMatrixElem(mat, 3));
 			return true;
 		}
 		else
 		{
 			xyz[0] = btScalar(0.0);
-			xyz[1] = -btAtan2(btGetMatrixElem(mat,7),btGetMatrixElem(mat,8));
+			xyz[1] = -btAtan2(btGetMatrixElem(mat, 7), btGetMatrixElem(mat, 8));
 			xyz[2] = -SIMD_HALF_PI;
 			return false;
 		}
@@ -206,33 +193,33 @@ bool btGeneric6DofSpring2Constraint::matrixToEulerYZX(const btMatrix3x3& mat,btV
 	else
 	{
 		xyz[0] = btScalar(0.0);
-		xyz[1] = btAtan2(btGetMatrixElem(mat,7),btGetMatrixElem(mat,8));
+		xyz[1] = btAtan2(btGetMatrixElem(mat, 7), btGetMatrixElem(mat, 8));
 		xyz[2] = SIMD_HALF_PI;
 	}
 	return false;
 }
 
-bool btGeneric6DofSpring2Constraint::matrixToEulerZXY(const btMatrix3x3& mat,btVector3& xyz)
+bool btGeneric6DofSpring2Constraint::matrixToEulerZXY(const btMatrix3x3& mat, btVector3& xyz)
 {
 	// rot =  cz*cy-sz*sx*sy    -cx*sz   cz*sy+cy*sz*sx
 	//        cy*sz+cz*sx*sy     cz*cx   sz*sy-cz*xy*sx
 	//        -cx*sy              sx     cx*cy
 
-	btScalar fi = btGetMatrixElem(mat,7);
+	btScalar fi = btGetMatrixElem(mat, 7);
 	if (fi < btScalar(1.0f))
 	{
 		if (fi > btScalar(-1.0f))
 		{
-			xyz[0] = btAsin(btGetMatrixElem(mat,7));
-			xyz[1] = btAtan2(-btGetMatrixElem(mat,6),btGetMatrixElem(mat,8));
-			xyz[2] = btAtan2(-btGetMatrixElem(mat,1),btGetMatrixElem(mat,4));
+			xyz[0] = btAsin(btGetMatrixElem(mat, 7));
+			xyz[1] = btAtan2(-btGetMatrixElem(mat, 6), btGetMatrixElem(mat, 8));
+			xyz[2] = btAtan2(-btGetMatrixElem(mat, 1), btGetMatrixElem(mat, 4));
 			return true;
 		}
 		else
 		{
 			xyz[0] = -SIMD_HALF_PI;
 			xyz[1] = btScalar(0.0);
-			xyz[2] = -btAtan2(btGetMatrixElem(mat,2),btGetMatrixElem(mat,0));
+			xyz[2] = -btAtan2(btGetMatrixElem(mat, 2), btGetMatrixElem(mat, 0));
 			return false;
 		}
 	}
@@ -240,32 +227,32 @@ bool btGeneric6DofSpring2Constraint::matrixToEulerZXY(const btMatrix3x3& mat,btV
 	{
 		xyz[0] = SIMD_HALF_PI;
 		xyz[1] = btScalar(0.0);
-		xyz[2] = btAtan2(btGetMatrixElem(mat,2),btGetMatrixElem(mat,0));
+		xyz[2] = btAtan2(btGetMatrixElem(mat, 2), btGetMatrixElem(mat, 0));
 	}
 	return false;
 }
 
-bool btGeneric6DofSpring2Constraint::matrixToEulerZYX(const btMatrix3x3& mat,btVector3& xyz)
+bool btGeneric6DofSpring2Constraint::matrixToEulerZYX(const btMatrix3x3& mat, btVector3& xyz)
 {
 	// rot =  cz*cy   cz*sy*sx-cx*sz   sz*sx+cz*cx*sy
 	//        cy*sz   cz*cx+sz*sy*sx   cx*sz*sy-cz*sx
 	//        -sy          cy*sx         cy*cx
 
-	btScalar fi = btGetMatrixElem(mat,6);
+	btScalar fi = btGetMatrixElem(mat, 6);
 	if (fi < btScalar(1.0f))
 	{
 		if (fi > btScalar(-1.0f))
 		{
-			xyz[0] = btAtan2(btGetMatrixElem(mat,7), btGetMatrixElem(mat,8));
-			xyz[1] = btAsin(-btGetMatrixElem(mat,6));
-			xyz[2] = btAtan2(btGetMatrixElem(mat,3),btGetMatrixElem(mat,0));
+			xyz[0] = btAtan2(btGetMatrixElem(mat, 7), btGetMatrixElem(mat, 8));
+			xyz[1] = btAsin(-btGetMatrixElem(mat, 6));
+			xyz[2] = btAtan2(btGetMatrixElem(mat, 3), btGetMatrixElem(mat, 0));
 			return true;
 		}
 		else
 		{
 			xyz[0] = btScalar(0.0);
 			xyz[1] = SIMD_HALF_PI;
-			xyz[2] = -btAtan2(btGetMatrixElem(mat,1),btGetMatrixElem(mat,2));
+			xyz[2] = -btAtan2(btGetMatrixElem(mat, 1), btGetMatrixElem(mat, 2));
 			return false;
 		}
 	}
@@ -273,23 +260,36 @@ bool btGeneric6DofSpring2Constraint::matrixToEulerZYX(const btMatrix3x3& mat,btV
 	{
 		xyz[0] = btScalar(0.0);
 		xyz[1] = -SIMD_HALF_PI;
-		xyz[2] = btAtan2(-btGetMatrixElem(mat,1),-btGetMatrixElem(mat,2));
+		xyz[2] = btAtan2(-btGetMatrixElem(mat, 1), -btGetMatrixElem(mat, 2));
 	}
 	return false;
 }
 
 void btGeneric6DofSpring2Constraint::calculateAngleInfo()
 {
-	btMatrix3x3 relative_frame = m_calculatedTransformA.getBasis().inverse()*m_calculatedTransformB.getBasis();
+	btMatrix3x3 relative_frame = m_calculatedTransformA.getBasis().inverse() * m_calculatedTransformB.getBasis();
 	switch (m_rotateOrder)
 	{
-		case RO_XYZ : matrixToEulerXYZ(relative_frame,m_calculatedAxisAngleDiff); break;
-		case RO_XZY : matrixToEulerXZY(relative_frame,m_calculatedAxisAngleDiff); break;
-		case RO_YXZ : matrixToEulerYXZ(relative_frame,m_calculatedAxisAngleDiff); break;
-		case RO_YZX : matrixToEulerYZX(relative_frame,m_calculatedAxisAngleDiff); break;
-		case RO_ZXY : matrixToEulerZXY(relative_frame,m_calculatedAxisAngleDiff); break;
-		case RO_ZYX : matrixToEulerZYX(relative_frame,m_calculatedAxisAngleDiff); break;
-		default : btAssert(false);
+		case RO_XYZ:
+			matrixToEulerXYZ(relative_frame, m_calculatedAxisAngleDiff);
+			break;
+		case RO_XZY:
+			matrixToEulerXZY(relative_frame, m_calculatedAxisAngleDiff);
+			break;
+		case RO_YXZ:
+			matrixToEulerYXZ(relative_frame, m_calculatedAxisAngleDiff);
+			break;
+		case RO_YZX:
+			matrixToEulerYZX(relative_frame, m_calculatedAxisAngleDiff);
+			break;
+		case RO_ZXY:
+			matrixToEulerZXY(relative_frame, m_calculatedAxisAngleDiff);
+			break;
+		case RO_ZYX:
+			matrixToEulerZYX(relative_frame, m_calculatedAxisAngleDiff);
+			break;
+		default:
+			btAssert(false);
 	}
 	// in euler angle mode we do not actually constrain the angular velocity
 	// along the axes axis[0] and axis[2] (although we do use axis[1]) :
@@ -307,14 +307,14 @@ void btGeneric6DofSpring2Constraint::calculateAngleInfo()
 	// to the components of w and set that to 0.
 	switch (m_rotateOrder)
 	{
-	case RO_XYZ :
+		case RO_XYZ:
 		{
 			//Is this the "line of nodes" calculation choosing planes YZ (B coordinate system) and xy (A coordinate system)? (http://en.wikipedia.org/wiki/Euler_angles)
 			//The two planes are non-homologous, so this is a Tait–Bryan angle formalism and not a proper Euler
 			//Extrinsic rotations are equal to the reversed order intrinsic rotations so the above xyz extrinsic rotations (axes are fixed) are the same as the zy'x" intrinsic rotations (axes are refreshed after each rotation)
 			//that is why xy and YZ planes are chosen (this will describe a zy'x" intrinsic rotation) (see the figure on the left at http://en.wikipedia.org/wiki/Euler_angles under Tait–Bryan angles)
 			// x' = Nperp = N.cross(axis2)
-			// y' = N = axis2.cross(axis0)	
+			// y' = N = axis2.cross(axis0)
 			// z' = z
 			//
 			// x" = X
@@ -324,7 +324,7 @@ void btGeneric6DofSpring2Constraint::calculateAngleInfo()
 			//first rotate around z
 			//second rotate around y'= z.cross(X)
 			//third rotate around x" = X
-			//Original XYZ extrinsic rotation order. 
+			//Original XYZ extrinsic rotation order.
 			//Planes: xy and YZ normals: z, X.  Plane intersection (N) is z.cross(X)
 			btVector3 axis0 = m_calculatedTransformB.getBasis().getColumn(0);
 			btVector3 axis2 = m_calculatedTransformA.getBasis().getColumn(2);
@@ -333,7 +333,7 @@ void btGeneric6DofSpring2Constraint::calculateAngleInfo()
 			m_calculatedAxis[2] = axis0.cross(m_calculatedAxis[1]);
 			break;
 		}
-	case RO_XZY :
+		case RO_XZY:
 		{
 			//planes: xz,ZY normals: y, X
 			//first rotate around y
@@ -346,7 +346,7 @@ void btGeneric6DofSpring2Constraint::calculateAngleInfo()
 			m_calculatedAxis[1] = m_calculatedAxis[2].cross(axis0);
 			break;
 		}
-	case RO_YXZ :
+		case RO_YXZ:
 		{
 			//planes: yx,XZ normals: z, Y
 			//first rotate around z
@@ -359,7 +359,7 @@ void btGeneric6DofSpring2Constraint::calculateAngleInfo()
 			m_calculatedAxis[2] = m_calculatedAxis[0].cross(axis1);
 			break;
 		}
-	case RO_YZX :
+		case RO_YZX:
 		{
 			//planes: yz,ZX normals: x, Y
 			//first rotate around x
@@ -372,7 +372,7 @@ void btGeneric6DofSpring2Constraint::calculateAngleInfo()
 			m_calculatedAxis[1] = m_calculatedAxis[2].cross(axis0);
 			break;
 		}
-	case RO_ZXY :
+		case RO_ZXY:
 		{
 			//planes: zx,XY normals: y, Z
 			//first rotate around y
@@ -385,7 +385,7 @@ void btGeneric6DofSpring2Constraint::calculateAngleInfo()
 			m_calculatedAxis[2] = m_calculatedAxis[0].cross(axis1);
 			break;
 		}
-	case RO_ZYX :
+		case RO_ZYX:
 		{
 			//planes: zy,YX normals: x, Z
 			//first rotate around x
@@ -398,22 +398,21 @@ void btGeneric6DofSpring2Constraint::calculateAngleInfo()
 			m_calculatedAxis[2] = axis0.cross(m_calculatedAxis[1]);
 			break;
 		}
-	default:
-		btAssert(false);
+		default:
+			btAssert(false);
 	}
 
 	m_calculatedAxis[0].normalize();
 	m_calculatedAxis[1].normalize();
 	m_calculatedAxis[2].normalize();
-
 }
 
 void btGeneric6DofSpring2Constraint::calculateTransforms()
 {
-	calculateTransforms(m_rbA.getCenterOfMassTransform(),m_rbB.getCenterOfMassTransform());
+	calculateTransforms(m_rbA.getCenterOfMassTransform(), m_rbB.getCenterOfMassTransform());
 }
 
-void btGeneric6DofSpring2Constraint::calculateTransforms(const btTransform& transA,const btTransform& transB)
+void btGeneric6DofSpring2Constraint::calculateTransforms(const btTransform& transA, const btTransform& transB)
 {
 	m_calculatedTransformA = transA * m_frameInA;
 	m_calculatedTransformB = transB * m_frameInB;
@@ -424,18 +423,17 @@ void btGeneric6DofSpring2Constraint::calculateTransforms(const btTransform& tran
 	btScalar miB = getRigidBodyB().getInvMass();
 	m_hasStaticBody = (miA < SIMD_EPSILON) || (miB < SIMD_EPSILON);
 	btScalar miS = miA + miB;
-	if(miS > btScalar(0.f))
+	if (miS > btScalar(0.f))
 	{
 		m_factA = miB / miS;
 	}
-	else 
+	else
 	{
 		m_factA = btScalar(0.5f);
 	}
 	m_factB = btScalar(1.0f) - m_factA;
 }
 
-
 void btGeneric6DofSpring2Constraint::testAngularLimitMotor(int axis_index)
 {
 	btScalar angle = m_calculatedAxisAngleDiff[axis_index];
@@ -444,35 +442,37 @@ void btGeneric6DofSpring2Constraint::testAngularLimitMotor(int axis_index)
 	m_angularLimits[axis_index].testLimitValue(angle);
 }
 
-
-void btGeneric6DofSpring2Constraint::getInfo1 (btConstraintInfo1* info)
+void btGeneric6DofSpring2Constraint::getInfo1(btConstraintInfo1* info)
 {
 	//prepare constraint
-	calculateTransforms(m_rbA.getCenterOfMassTransform(),m_rbB.getCenterOfMassTransform());
+	calculateTransforms(m_rbA.getCenterOfMassTransform(), m_rbB.getCenterOfMassTransform());
 	info->m_numConstraintRows = 0;
 	info->nub = 0;
 	int i;
 	//test linear limits
-	for(i = 0; i < 3; i++)
+	for (i = 0; i < 3; i++)
 	{
-		     if (m_linearLimits.m_currentLimit[i]==4) info->m_numConstraintRows += 2;
-		else if (m_linearLimits.m_currentLimit[i]!=0) info->m_numConstraintRows += 1;
-		if (m_linearLimits.m_enableMotor[i] ) info->m_numConstraintRows += 1;
+		if (m_linearLimits.m_currentLimit[i] == 4)
+			info->m_numConstraintRows += 2;
+		else if (m_linearLimits.m_currentLimit[i] != 0)
+			info->m_numConstraintRows += 1;
+		if (m_linearLimits.m_enableMotor[i]) info->m_numConstraintRows += 1;
 		if (m_linearLimits.m_enableSpring[i]) info->m_numConstraintRows += 1;
 	}
 	//test angular limits
-	for (i=0;i<3 ;i++ )
+	for (i = 0; i < 3; i++)
 	{
 		testAngularLimitMotor(i);
-		     if (m_angularLimits[i].m_currentLimit==4) info->m_numConstraintRows += 2;
-		else if (m_angularLimits[i].m_currentLimit!=0) info->m_numConstraintRows += 1;
-		if (m_angularLimits[i].m_enableMotor ) info->m_numConstraintRows += 1;
+		if (m_angularLimits[i].m_currentLimit == 4)
+			info->m_numConstraintRows += 2;
+		else if (m_angularLimits[i].m_currentLimit != 0)
+			info->m_numConstraintRows += 1;
+		if (m_angularLimits[i].m_enableMotor) info->m_numConstraintRows += 1;
 		if (m_angularLimits[i].m_enableSpring) info->m_numConstraintRows += 1;
 	}
 }
 
-
-void btGeneric6DofSpring2Constraint::getInfo2 (btConstraintInfo2* info)
+void btGeneric6DofSpring2Constraint::getInfo2(btConstraintInfo2* info)
 {
 	const btTransform& transA = m_rbA.getCenterOfMassTransform();
 	const btTransform& transB = m_rbB.getCenterOfMassTransform();
@@ -482,118 +482,138 @@ void btGeneric6DofSpring2Constraint::getInfo2 (btConstraintInfo2* info)
 	const btVector3& angVelB = m_rbB.getAngularVelocity();
 
 	// for stability better to solve angular limits first
-	int row = setAngularLimits(info, 0,transA,transB,linVelA,linVelB,angVelA,angVelB);
-	setLinearLimits(info, row, transA,transB,linVelA,linVelB,angVelA,angVelB);
+	int row = setAngularLimits(info, 0, transA, transB, linVelA, linVelB, angVelA, angVelB);
+	setLinearLimits(info, row, transA, transB, linVelA, linVelB, angVelA, angVelB);
 }
 
-
-int btGeneric6DofSpring2Constraint::setLinearLimits(btConstraintInfo2* info, int row, const btTransform& transA,const btTransform& transB,const btVector3& linVelA,const btVector3& linVelB,const btVector3& angVelA,const btVector3& angVelB)
+int btGeneric6DofSpring2Constraint::setLinearLimits(btConstraintInfo2* info, int row, const btTransform& transA, const btTransform& transB, const btVector3& linVelA, const btVector3& linVelB, const btVector3& angVelA, const btVector3& angVelB)
 {
 	//solve linear limits
 	btRotationalLimitMotor2 limot;
-	for (int i=0;i<3 ;i++ )
+	for (int i = 0; i < 3; i++)
 	{
-		if(m_linearLimits.m_currentLimit[i] || m_linearLimits.m_enableMotor[i] || m_linearLimits.m_enableSpring[i])
-		{ // re-use rotational motor code
-			limot.m_bounce                 = m_linearLimits.m_bounce[i];
-			limot.m_currentLimit           = m_linearLimits.m_currentLimit[i];
-			limot.m_currentPosition        = m_linearLimits.m_currentLinearDiff[i];
-			limot.m_currentLimitError      = m_linearLimits.m_currentLimitError[i];
-			limot.m_currentLimitErrorHi    = m_linearLimits.m_currentLimitErrorHi[i];
-			limot.m_enableMotor            = m_linearLimits.m_enableMotor[i];
-			limot.m_servoMotor             = m_linearLimits.m_servoMotor[i];
-			limot.m_servoTarget            = m_linearLimits.m_servoTarget[i];
-			limot.m_enableSpring           = m_linearLimits.m_enableSpring[i];
-			limot.m_springStiffness        = m_linearLimits.m_springStiffness[i];
+		if (m_linearLimits.m_currentLimit[i] || m_linearLimits.m_enableMotor[i] || m_linearLimits.m_enableSpring[i])
+		{  // re-use rotational motor code
+			limot.m_bounce = m_linearLimits.m_bounce[i];
+			limot.m_currentLimit = m_linearLimits.m_currentLimit[i];
+			limot.m_currentPosition = m_linearLimits.m_currentLinearDiff[i];
+			limot.m_currentLimitError = m_linearLimits.m_currentLimitError[i];
+			limot.m_currentLimitErrorHi = m_linearLimits.m_currentLimitErrorHi[i];
+			limot.m_enableMotor = m_linearLimits.m_enableMotor[i];
+			limot.m_servoMotor = m_linearLimits.m_servoMotor[i];
+			limot.m_servoTarget = m_linearLimits.m_servoTarget[i];
+			limot.m_enableSpring = m_linearLimits.m_enableSpring[i];
+			limot.m_springStiffness = m_linearLimits.m_springStiffness[i];
 			limot.m_springStiffnessLimited = m_linearLimits.m_springStiffnessLimited[i];
-			limot.m_springDamping          = m_linearLimits.m_springDamping[i];
-			limot.m_springDampingLimited   = m_linearLimits.m_springDampingLimited[i];
-			limot.m_equilibriumPoint       = m_linearLimits.m_equilibriumPoint[i];
-			limot.m_hiLimit                = m_linearLimits.m_upperLimit[i];
-			limot.m_loLimit                = m_linearLimits.m_lowerLimit[i];
-			limot.m_maxMotorForce          = m_linearLimits.m_maxMotorForce[i];
-			limot.m_targetVelocity         = m_linearLimits.m_targetVelocity[i];
+			limot.m_springDamping = m_linearLimits.m_springDamping[i];
+			limot.m_springDampingLimited = m_linearLimits.m_springDampingLimited[i];
+			limot.m_equilibriumPoint = m_linearLimits.m_equilibriumPoint[i];
+			limot.m_hiLimit = m_linearLimits.m_upperLimit[i];
+			limot.m_loLimit = m_linearLimits.m_lowerLimit[i];
+			limot.m_maxMotorForce = m_linearLimits.m_maxMotorForce[i];
+			limot.m_targetVelocity = m_linearLimits.m_targetVelocity[i];
 			btVector3 axis = m_calculatedTransformA.getBasis().getColumn(i);
 			int flags = m_flags >> (i * BT_6DOF_FLAGS_AXIS_SHIFT2);
-			limot.m_stopCFM  = (flags & BT_6DOF_FLAGS_CFM_STOP2) ? m_linearLimits.m_stopCFM[i] : info->cfm[0];
-			limot.m_stopERP  = (flags & BT_6DOF_FLAGS_ERP_STOP2) ? m_linearLimits.m_stopERP[i] : info->erp;
+			limot.m_stopCFM = (flags & BT_6DOF_FLAGS_CFM_STOP2) ? m_linearLimits.m_stopCFM[i] : info->cfm[0];
+			limot.m_stopERP = (flags & BT_6DOF_FLAGS_ERP_STOP2) ? m_linearLimits.m_stopERP[i] : info->erp;
 			limot.m_motorCFM = (flags & BT_6DOF_FLAGS_CFM_MOTO2) ? m_linearLimits.m_motorCFM[i] : info->cfm[0];
 			limot.m_motorERP = (flags & BT_6DOF_FLAGS_ERP_MOTO2) ? m_linearLimits.m_motorERP[i] : info->erp;
 
 			//rotAllowed is a bit of a magic from the original 6dof. The calculation of it here is something that imitates the original behavior as much as possible.
 			int indx1 = (i + 1) % 3;
 			int indx2 = (i + 2) % 3;
-			int rotAllowed = 1; // rotations around orthos to current axis (it is used only when one of the body is static)
-			#define D6_LIMIT_ERROR_THRESHOLD_FOR_ROTATION 1.0e-3
+			int rotAllowed = 1;  // rotations around orthos to current axis (it is used only when one of the body is static)
+#define D6_LIMIT_ERROR_THRESHOLD_FOR_ROTATION 1.0e-3
 			bool indx1Violated = m_angularLimits[indx1].m_currentLimit == 1 ||
-				m_angularLimits[indx1].m_currentLimit == 2 ||
-				( m_angularLimits[indx1].m_currentLimit == 3 && ( m_angularLimits[indx1].m_currentLimitError < -D6_LIMIT_ERROR_THRESHOLD_FOR_ROTATION || m_angularLimits[indx1].m_currentLimitError > D6_LIMIT_ERROR_THRESHOLD_FOR_ROTATION ) ) ||
-				( m_angularLimits[indx1].m_currentLimit == 4 && ( m_angularLimits[indx1].m_currentLimitError < -D6_LIMIT_ERROR_THRESHOLD_FOR_ROTATION || m_angularLimits[indx1].m_currentLimitErrorHi > D6_LIMIT_ERROR_THRESHOLD_FOR_ROTATION ) );
+								 m_angularLimits[indx1].m_currentLimit == 2 ||
+								 (m_angularLimits[indx1].m_currentLimit == 3 && (m_angularLimits[indx1].m_currentLimitError < -D6_LIMIT_ERROR_THRESHOLD_FOR_ROTATION || m_angularLimits[indx1].m_currentLimitError > D6_LIMIT_ERROR_THRESHOLD_FOR_ROTATION)) ||
+								 (m_angularLimits[indx1].m_currentLimit == 4 && (m_angularLimits[indx1].m_currentLimitError < -D6_LIMIT_ERROR_THRESHOLD_FOR_ROTATION || m_angularLimits[indx1].m_currentLimitErrorHi > D6_LIMIT_ERROR_THRESHOLD_FOR_ROTATION));
 			bool indx2Violated = m_angularLimits[indx2].m_currentLimit == 1 ||
-				m_angularLimits[indx2].m_currentLimit == 2 ||
-				( m_angularLimits[indx2].m_currentLimit == 3 && ( m_angularLimits[indx2].m_currentLimitError < -D6_LIMIT_ERROR_THRESHOLD_FOR_ROTATION || m_angularLimits[indx2].m_currentLimitError > D6_LIMIT_ERROR_THRESHOLD_FOR_ROTATION ) ) ||
-				( m_angularLimits[indx2].m_currentLimit == 4 && ( m_angularLimits[indx2].m_currentLimitError < -D6_LIMIT_ERROR_THRESHOLD_FOR_ROTATION || m_angularLimits[indx2].m_currentLimitErrorHi > D6_LIMIT_ERROR_THRESHOLD_FOR_ROTATION ) );
-			if( indx1Violated && indx2Violated )
+								 m_angularLimits[indx2].m_currentLimit == 2 ||
+								 (m_angularLimits[indx2].m_currentLimit == 3 && (m_angularLimits[indx2].m_currentLimitError < -D6_LIMIT_ERROR_THRESHOLD_FOR_ROTATION || m_angularLimits[indx2].m_currentLimitError > D6_LIMIT_ERROR_THRESHOLD_FOR_ROTATION)) ||
+								 (m_angularLimits[indx2].m_currentLimit == 4 && (m_angularLimits[indx2].m_currentLimitError < -D6_LIMIT_ERROR_THRESHOLD_FOR_ROTATION || m_angularLimits[indx2].m_currentLimitErrorHi > D6_LIMIT_ERROR_THRESHOLD_FOR_ROTATION));
+			if (indx1Violated && indx2Violated)
 			{
 				rotAllowed = 0;
 			}
-			row += get_limit_motor_info2(&limot, transA,transB,linVelA,linVelB,angVelA,angVelB, info, row, axis, 0, rotAllowed);
-
+			row += get_limit_motor_info2(&limot, transA, transB, linVelA, linVelB, angVelA, angVelB, info, row, axis, 0, rotAllowed);
 		}
 	}
 	return row;
 }
 
-
-
-int btGeneric6DofSpring2Constraint::setAngularLimits(btConstraintInfo2 *info, int row_offset, const btTransform& transA,const btTransform& transB,const btVector3& linVelA,const btVector3& linVelB,const btVector3& angVelA,const btVector3& angVelB)
+int btGeneric6DofSpring2Constraint::setAngularLimits(btConstraintInfo2* info, int row_offset, const btTransform& transA, const btTransform& transB, const btVector3& linVelA, const btVector3& linVelB, const btVector3& angVelA, const btVector3& angVelB)
 {
 	int row = row_offset;
 
 	//order of rotational constraint rows
 	int cIdx[] = {0, 1, 2};
-	switch(m_rotateOrder)
+	switch (m_rotateOrder)
 	{
-		case RO_XYZ : cIdx[0] = 0; cIdx[1] = 1; cIdx[2] = 2; break;
-		case RO_XZY : cIdx[0] = 0; cIdx[1] = 2; cIdx[2] = 1; break;
-		case RO_YXZ : cIdx[0] = 1; cIdx[1] = 0; cIdx[2] = 2; break;
-		case RO_YZX : cIdx[0] = 1; cIdx[1] = 2; cIdx[2] = 0; break;
-		case RO_ZXY : cIdx[0] = 2; cIdx[1] = 0; cIdx[2] = 1; break;
-		case RO_ZYX : cIdx[0] = 2; cIdx[1] = 1; cIdx[2] = 0; break;
-		default : btAssert(false);
+		case RO_XYZ:
+			cIdx[0] = 0;
+			cIdx[1] = 1;
+			cIdx[2] = 2;
+			break;
+		case RO_XZY:
+			cIdx[0] = 0;
+			cIdx[1] = 2;
+			cIdx[2] = 1;
+			break;
+		case RO_YXZ:
+			cIdx[0] = 1;
+			cIdx[1] = 0;
+			cIdx[2] = 2;
+			break;
+		case RO_YZX:
+			cIdx[0] = 1;
+			cIdx[1] = 2;
+			cIdx[2] = 0;
+			break;
+		case RO_ZXY:
+			cIdx[0] = 2;
+			cIdx[1] = 0;
+			cIdx[2] = 1;
+			break;
+		case RO_ZYX:
+			cIdx[0] = 2;
+			cIdx[1] = 1;
+			cIdx[2] = 0;
+			break;
+		default:
+			btAssert(false);
 	}
 
-	for (int ii = 0; ii < 3 ; ii++ )
+	for (int ii = 0; ii < 3; ii++)
 	{
 		int i = cIdx[ii];
-		if(m_angularLimits[i].m_currentLimit || m_angularLimits[i].m_enableMotor || m_angularLimits[i].m_enableSpring)
+		if (m_angularLimits[i].m_currentLimit || m_angularLimits[i].m_enableMotor || m_angularLimits[i].m_enableSpring)
 		{
 			btVector3 axis = getAxis(i);
 			int flags = m_flags >> ((i + 3) * BT_6DOF_FLAGS_AXIS_SHIFT2);
-			if(!(flags & BT_6DOF_FLAGS_CFM_STOP2))
+			if (!(flags & BT_6DOF_FLAGS_CFM_STOP2))
 			{
 				m_angularLimits[i].m_stopCFM = info->cfm[0];
 			}
-			if(!(flags & BT_6DOF_FLAGS_ERP_STOP2))
+			if (!(flags & BT_6DOF_FLAGS_ERP_STOP2))
 			{
 				m_angularLimits[i].m_stopERP = info->erp;
 			}
-			if(!(flags & BT_6DOF_FLAGS_CFM_MOTO2))
+			if (!(flags & BT_6DOF_FLAGS_CFM_MOTO2))
 			{
 				m_angularLimits[i].m_motorCFM = info->cfm[0];
 			}
-			if(!(flags & BT_6DOF_FLAGS_ERP_MOTO2))
+			if (!(flags & BT_6DOF_FLAGS_ERP_MOTO2))
 			{
 				m_angularLimits[i].m_motorERP = info->erp;
 			}
-			row += get_limit_motor_info2(&m_angularLimits[i],transA,transB,linVelA,linVelB,angVelA,angVelB, info,row,axis,1);
+			row += get_limit_motor_info2(&m_angularLimits[i], transA, transB, linVelA, linVelB, angVelA, angVelB, info, row, axis, 1);
 		}
 	}
 
 	return row;
 }
 
-
 void btGeneric6DofSpring2Constraint::setFrames(const btTransform& frameA, const btTransform& frameB)
 {
 	m_frameInA = frameA;
@@ -602,32 +622,31 @@ void btGeneric6DofSpring2Constraint::setFrames(const btTransform& frameA, const
 	calculateTransforms();
 }
 
-
 void btGeneric6DofSpring2Constraint::calculateLinearInfo()
 {
 	m_calculatedLinearDiff = m_calculatedTransformB.getOrigin() - m_calculatedTransformA.getOrigin();
 	m_calculatedLinearDiff = m_calculatedTransformA.getBasis().inverse() * m_calculatedLinearDiff;
-	for(int i = 0; i < 3; i++)
+	for (int i = 0; i < 3; i++)
 	{
 		m_linearLimits.m_currentLinearDiff[i] = m_calculatedLinearDiff[i];
 		m_linearLimits.testLimitValue(i, m_calculatedLinearDiff[i]);
 	}
 }
 
-void btGeneric6DofSpring2Constraint::calculateJacobi(btRotationalLimitMotor2 * limot, const btTransform& transA,const btTransform& transB, btConstraintInfo2 *info, int srow, btVector3& ax1, int rotational, int rotAllowed)
+void btGeneric6DofSpring2Constraint::calculateJacobi(btRotationalLimitMotor2* limot, const btTransform& transA, const btTransform& transB, btConstraintInfo2* info, int srow, btVector3& ax1, int rotational, int rotAllowed)
 {
-	btScalar *J1 = rotational ? info->m_J1angularAxis : info->m_J1linearAxis;
-	btScalar *J2 = rotational ? info->m_J2angularAxis : info->m_J2linearAxis;
+	btScalar* J1 = rotational ? info->m_J1angularAxis : info->m_J1linearAxis;
+	btScalar* J2 = rotational ? info->m_J2angularAxis : info->m_J2linearAxis;
 
-	J1[srow+0] = ax1[0];
-	J1[srow+1] = ax1[1];
-	J1[srow+2] = ax1[2];
+	J1[srow + 0] = ax1[0];
+	J1[srow + 1] = ax1[1];
+	J1[srow + 2] = ax1[2];
 
-	J2[srow+0] = -ax1[0];
-	J2[srow+1] = -ax1[1];
-	J2[srow+2] = -ax1[2];
+	J2[srow + 0] = -ax1[0];
+	J2[srow + 1] = -ax1[1];
+	J2[srow + 2] = -ax1[2];
 
-	if(!rotational)
+	if (!rotational)
 	{
 		btVector3 tmpA, tmpB, relA, relB;
 		// get vector from bodyB to frameB in WCS
@@ -636,40 +655,44 @@ void btGeneric6DofSpring2Constraint::calculateJacobi(btRotationalLimitMotor2 * l
 		relA = m_calculatedTransformA.getOrigin() - transA.getOrigin();
 		tmpA = relA.cross(ax1);
 		tmpB = relB.cross(ax1);
-		if(m_hasStaticBody && (!rotAllowed))
+		if (m_hasStaticBody && (!rotAllowed))
 		{
 			tmpA *= m_factA;
 			tmpB *= m_factB;
 		}
 		int i;
-		for (i=0; i<3; i++) info->m_J1angularAxis[srow+i] = tmpA[i];
-		for (i=0; i<3; i++) info->m_J2angularAxis[srow+i] = -tmpB[i];
+		for (i = 0; i < 3; i++) info->m_J1angularAxis[srow + i] = tmpA[i];
+		for (i = 0; i < 3; i++) info->m_J2angularAxis[srow + i] = -tmpB[i];
 	}
 }
 
-
 int btGeneric6DofSpring2Constraint::get_limit_motor_info2(
-	btRotationalLimitMotor2 * limot,
-	const btTransform& transA,const btTransform& transB,const btVector3& linVelA,const btVector3& linVelB,const btVector3& angVelA,const btVector3& angVelB,
-	btConstraintInfo2 *info, int row, btVector3& ax1, int rotational,int rotAllowed)
+	btRotationalLimitMotor2* limot,
+	const btTransform& transA, const btTransform& transB, const btVector3& linVelA, const btVector3& linVelB, const btVector3& angVelA, const btVector3& angVelB,
+	btConstraintInfo2* info, int row, btVector3& ax1, int rotational, int rotAllowed)
 {
 	int count = 0;
 	int srow = row * info->rowskip;
 
-	if (limot->m_currentLimit==4) 
+	if (limot->m_currentLimit == 4)
 	{
 		btScalar vel = rotational ? angVelA.dot(ax1) - angVelB.dot(ax1) : linVelA.dot(ax1) - linVelB.dot(ax1);
 
-		calculateJacobi(limot,transA,transB,info,srow,ax1,rotational,rotAllowed);
+		calculateJacobi(limot, transA, transB, info, srow, ax1, rotational, rotAllowed);
 		info->m_constraintError[srow] = info->fps * limot->m_stopERP * limot->m_currentLimitError * (rotational ? -1 : 1);
-		if (rotational) {
-			if (info->m_constraintError[srow]-vel*limot->m_stopERP > 0) {
-				btScalar bounceerror = -limot->m_bounce* vel;
+		if (rotational)
+		{
+			if (info->m_constraintError[srow] - vel * limot->m_stopERP > 0)
+			{
+				btScalar bounceerror = -limot->m_bounce * vel;
 				if (bounceerror > info->m_constraintError[srow]) info->m_constraintError[srow] = bounceerror;
 			}
-		} else {
-			if (info->m_constraintError[srow]-vel*limot->m_stopERP < 0) {
-				btScalar bounceerror = -limot->m_bounce* vel;
+		}
+		else
+		{
+			if (info->m_constraintError[srow] - vel * limot->m_stopERP < 0)
+			{
+				btScalar bounceerror = -limot->m_bounce * vel;
 				if (bounceerror < info->m_constraintError[srow]) info->m_constraintError[srow] = bounceerror;
 			}
 		}
@@ -679,16 +702,21 @@ int btGeneric6DofSpring2Constraint::get_limit_motor_info2(
 		srow += info->rowskip;
 		++count;
 
-		calculateJacobi(limot,transA,transB,info,srow,ax1,rotational,rotAllowed);
+		calculateJacobi(limot, transA, transB, info, srow, ax1, rotational, rotAllowed);
 		info->m_constraintError[srow] = info->fps * limot->m_stopERP * limot->m_currentLimitErrorHi * (rotational ? -1 : 1);
-		if (rotational) {
-			if (info->m_constraintError[srow]-vel*limot->m_stopERP < 0) {
-				btScalar bounceerror = -limot->m_bounce* vel;
+		if (rotational)
+		{
+			if (info->m_constraintError[srow] - vel * limot->m_stopERP < 0)
+			{
+				btScalar bounceerror = -limot->m_bounce * vel;
 				if (bounceerror < info->m_constraintError[srow]) info->m_constraintError[srow] = bounceerror;
 			}
-		} else {
-			if (info->m_constraintError[srow]-vel*limot->m_stopERP > 0) {
-				btScalar bounceerror = -limot->m_bounce* vel;
+		}
+		else
+		{
+			if (info->m_constraintError[srow] - vel * limot->m_stopERP > 0)
+			{
+				btScalar bounceerror = -limot->m_bounce * vel;
 				if (bounceerror > info->m_constraintError[srow]) info->m_constraintError[srow] = bounceerror;
 			}
 		}
@@ -697,10 +725,10 @@ int btGeneric6DofSpring2Constraint::get_limit_motor_info2(
 		info->cfm[srow] = limot->m_stopCFM;
 		srow += info->rowskip;
 		++count;
-	} else
-	if (limot->m_currentLimit==3) 
+	}
+	else if (limot->m_currentLimit == 3)
 	{
-		calculateJacobi(limot,transA,transB,info,srow,ax1,rotational,rotAllowed);
+		calculateJacobi(limot, transA, transB, info, srow, ax1, rotational, rotAllowed);
 		info->m_constraintError[srow] = info->fps * limot->m_stopERP * limot->m_currentLimitError * (rotational ? -1 : 1);
 		info->m_lowerLimit[srow] = -SIMD_INFINITY;
 		info->m_upperLimit[srow] = SIMD_INFINITY;
@@ -711,13 +739,13 @@ int btGeneric6DofSpring2Constraint::get_limit_motor_info2(
 
 	if (limot->m_enableMotor && !limot->m_servoMotor)
 	{
-		calculateJacobi(limot,transA,transB,info,srow,ax1,rotational,rotAllowed);
+		calculateJacobi(limot, transA, transB, info, srow, ax1, rotational, rotAllowed);
 		btScalar tag_vel = rotational ? limot->m_targetVelocity : -limot->m_targetVelocity;
-		btScalar mot_fact = getMotorFactor(limot->m_currentPosition, 
-			limot->m_loLimit,
-			limot->m_hiLimit,
-			tag_vel,
-			info->fps * limot->m_motorERP);
+		btScalar mot_fact = getMotorFactor(limot->m_currentPosition,
+										   limot->m_loLimit,
+										   limot->m_hiLimit,
+										   tag_vel,
+										   info->fps * limot->m_motorERP);
 		info->m_constraintError[srow] = mot_fact * limot->m_targetVelocity;
 		info->m_lowerLimit[srow] = -limot->m_maxMotorForce / info->fps;
 		info->m_upperLimit[srow] = limot->m_maxMotorForce / info->fps;
@@ -735,36 +763,36 @@ int btGeneric6DofSpring2Constraint::get_limit_motor_info2(
 			if (error > SIMD_PI)
 			{
 				error -= SIMD_2_PI;
-				curServoTarget +=SIMD_2_PI;
+				curServoTarget += SIMD_2_PI;
 			}
 			if (error < -SIMD_PI)
 			{
 				error += SIMD_2_PI;
-				curServoTarget -=SIMD_2_PI;
+				curServoTarget -= SIMD_2_PI;
 			}
 		}
 
-		calculateJacobi(limot,transA,transB,info,srow,ax1,rotational,rotAllowed);
-		btScalar targetvelocity = error<0 ? -limot->m_targetVelocity : limot->m_targetVelocity;
+		calculateJacobi(limot, transA, transB, info, srow, ax1, rotational, rotAllowed);
+		btScalar targetvelocity = error < 0 ? -limot->m_targetVelocity : limot->m_targetVelocity;
 		btScalar tag_vel = -targetvelocity;
 		btScalar mot_fact;
-		if(error != 0)
+		if (error != 0)
 		{
 			btScalar lowLimit;
 			btScalar hiLimit;
-			if(limot->m_loLimit > limot->m_hiLimit)
+			if (limot->m_loLimit > limot->m_hiLimit)
 			{
 				lowLimit = error > 0 ? curServoTarget : -SIMD_INFINITY;
-				hiLimit  = error < 0 ? curServoTarget :  SIMD_INFINITY;
+				hiLimit = error < 0 ? curServoTarget : SIMD_INFINITY;
 			}
 			else
 			{
-				lowLimit = error > 0 && curServoTarget>limot->m_loLimit ? curServoTarget : limot->m_loLimit;
-				hiLimit  = error < 0 && curServoTarget<limot->m_hiLimit ? curServoTarget : limot->m_hiLimit;
+				lowLimit = error > 0 && curServoTarget > limot->m_loLimit ? curServoTarget : limot->m_loLimit;
+				hiLimit = error < 0 && curServoTarget < limot->m_hiLimit ? curServoTarget : limot->m_hiLimit;
 			}
 			mot_fact = getMotorFactor(limot->m_currentPosition, lowLimit, hiLimit, tag_vel, info->fps * limot->m_motorERP);
-		} 
-		else 
+		}
+		else
 		{
 			mot_fact = 0;
 		}
@@ -779,7 +807,7 @@ int btGeneric6DofSpring2Constraint::get_limit_motor_info2(
 	if (limot->m_enableSpring)
 	{
 		btScalar error = limot->m_currentPosition - limot->m_equilibriumPoint;
-		calculateJacobi(limot,transA,transB,info,srow,ax1,rotational,rotAllowed);
+		calculateJacobi(limot, transA, transB, info, srow, ax1, rotational, rotAllowed);
 
 		//btScalar cfm = 1.0 / ((1.0/info->fps)*limot->m_springStiffness+ limot->m_springDamping);
 		//if(cfm > 0.99999)
@@ -792,34 +820,46 @@ int btGeneric6DofSpring2Constraint::get_limit_motor_info2(
 		btScalar dt = BT_ONE / info->fps;
 		btScalar kd = limot->m_springDamping;
 		btScalar ks = limot->m_springStiffness;
-		btScalar vel = rotational ? angVelA.dot(ax1) - angVelB.dot(ax1) : linVelA.dot(ax1) - linVelB.dot(ax1);
-//		btScalar erp = 0.1;
+		btScalar vel;
+		if (rotational)
+		{
+			vel = angVelA.dot(ax1) - angVelB.dot(ax1);
+		}
+		else
+		{
+			btVector3 tanVelA = angVelA.cross(m_calculatedTransformA.getOrigin() - transA.getOrigin());
+			btVector3 tanVelB = angVelB.cross(m_calculatedTransformB.getOrigin() - transB.getOrigin());
+			vel = (linVelA + tanVelA).dot(ax1) - (linVelB + tanVelB).dot(ax1);
+		}
 		btScalar cfm = BT_ZERO;
 		btScalar mA = BT_ONE / m_rbA.getInvMass();
 		btScalar mB = BT_ONE / m_rbB.getInvMass();
-		if (rotational) {
+		if (rotational)
+		{
 			btScalar rrA = (m_calculatedTransformA.getOrigin() - transA.getOrigin()).length2();
 			btScalar rrB = (m_calculatedTransformB.getOrigin() - transB.getOrigin()).length2();
 			if (m_rbA.getInvMass()) mA = mA * rrA + 1 / (m_rbA.getInvInertiaTensorWorld() * ax1).length();
 			if (m_rbB.getInvMass()) mB = mB * rrB + 1 / (m_rbB.getInvInertiaTensorWorld() * ax1).length();
 		}
-		btScalar m = mA > mB ? mB : mA;
+		btScalar m;
+		if (m_rbA.getInvMass() == 0) m = mB; else
+		if (m_rbB.getInvMass() == 0) m = mA; else
+			m = mA*mB / (mA + mB);
 		btScalar angularfreq = sqrt(ks / m);
 
-
 		//limit stiffness (the spring should not be sampled faster that the quarter of its angular frequency)
-		if(limot->m_springStiffnessLimited && 0.25 < angularfreq * dt)
+		if (limot->m_springStiffnessLimited && 0.25 < angularfreq * dt)
 		{
 			ks = BT_ONE / dt / dt / btScalar(16.0) * m;
 		}
 		//avoid damping that would blow up the spring
-		if(limot->m_springDampingLimited && kd * dt > m)
+		if (limot->m_springDampingLimited && kd * dt > m)
 		{
 			kd = m / dt;
 		}
 		btScalar fs = ks * error * dt;
 		btScalar fd = -kd * (vel) * (rotational ? -1 : 1) * dt;
-		btScalar f = (fs+fd);
+		btScalar f = (fs + fd);
 
 		// after the spring force affecting the body(es) the new velocity will be
 		// vel + f / m * (rotational ? -1 : 1)
@@ -828,15 +868,18 @@ int btGeneric6DofSpring2Constraint::get_limit_motor_info2(
 		// however in practice any value is fine as long as it is greater then the "proper" velocity,
 		// because the m_lowerLimit and the m_upperLimit will determinate the strength of the final pulling force
 		// so it is much simpler (and more robust) just to simply use inf (with the proper sign)
+		// (Even with our best intent the "new" velocity is only an estimation. If we underestimate
+		// the "proper" velocity that will weaken the spring, however if we overestimate it, it doesn't
+		// matter, because the solver will limit it according the force limit)
 		// you may also wonder what if the current velocity (vel) so high that the pulling force will not change its direction (in this iteration)
 		// will we not request a velocity with the wrong direction ?
-		// and the answare is not, because in practice during the solving the current velocity is subtracted from the m_constraintError
+		// and the answer is not, because in practice during the solving the current velocity is subtracted from the m_constraintError
 		// so the sign of the force that is really matters
 		info->m_constraintError[srow] = (rotational ? -1 : 1) * (f < 0 ? -SIMD_INFINITY : SIMD_INFINITY);
 
 		btScalar minf = f < fd ? f : fd;
 		btScalar maxf = f < fd ? fd : f;
-		if(!rotational)
+		if (!rotational)
 		{
 			info->m_lowerLimit[srow] = minf > 0 ? 0 : minf;
 			info->m_upperLimit[srow] = maxf < 0 ? 0 : maxf;
@@ -855,56 +898,55 @@ int btGeneric6DofSpring2Constraint::get_limit_motor_info2(
 	return count;
 }
 
-
-//override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5). 
+//override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5).
 //If no axis is provided, it uses the default axis for this constraint.
 void btGeneric6DofSpring2Constraint::setParam(int num, btScalar value, int axis)
 {
-	if((axis >= 0) && (axis < 3))
+	if ((axis >= 0) && (axis < 3))
 	{
-		switch(num)
+		switch (num)
 		{
-			case BT_CONSTRAINT_STOP_ERP : 
+			case BT_CONSTRAINT_STOP_ERP:
 				m_linearLimits.m_stopERP[axis] = value;
 				m_flags |= BT_6DOF_FLAGS_ERP_STOP2 << (axis * BT_6DOF_FLAGS_AXIS_SHIFT2);
 				break;
-			case BT_CONSTRAINT_STOP_CFM : 
+			case BT_CONSTRAINT_STOP_CFM:
 				m_linearLimits.m_stopCFM[axis] = value;
 				m_flags |= BT_6DOF_FLAGS_CFM_STOP2 << (axis * BT_6DOF_FLAGS_AXIS_SHIFT2);
 				break;
-			case BT_CONSTRAINT_ERP : 
+			case BT_CONSTRAINT_ERP:
 				m_linearLimits.m_motorERP[axis] = value;
 				m_flags |= BT_6DOF_FLAGS_ERP_MOTO2 << (axis * BT_6DOF_FLAGS_AXIS_SHIFT2);
 				break;
-			case BT_CONSTRAINT_CFM : 
+			case BT_CONSTRAINT_CFM:
 				m_linearLimits.m_motorCFM[axis] = value;
 				m_flags |= BT_6DOF_FLAGS_CFM_MOTO2 << (axis * BT_6DOF_FLAGS_AXIS_SHIFT2);
 				break;
-			default : 
+			default:
 				btAssertConstrParams(0);
 		}
 	}
-	else if((axis >=3) && (axis < 6))
+	else if ((axis >= 3) && (axis < 6))
 	{
-		switch(num)
+		switch (num)
 		{
-			case BT_CONSTRAINT_STOP_ERP : 
+			case BT_CONSTRAINT_STOP_ERP:
 				m_angularLimits[axis - 3].m_stopERP = value;
 				m_flags |= BT_6DOF_FLAGS_ERP_STOP2 << (axis * BT_6DOF_FLAGS_AXIS_SHIFT2);
 				break;
-			case BT_CONSTRAINT_STOP_CFM : 
+			case BT_CONSTRAINT_STOP_CFM:
 				m_angularLimits[axis - 3].m_stopCFM = value;
 				m_flags |= BT_6DOF_FLAGS_CFM_STOP2 << (axis * BT_6DOF_FLAGS_AXIS_SHIFT2);
 				break;
-			case BT_CONSTRAINT_ERP : 
+			case BT_CONSTRAINT_ERP:
 				m_angularLimits[axis - 3].m_motorERP = value;
 				m_flags |= BT_6DOF_FLAGS_ERP_MOTO2 << (axis * BT_6DOF_FLAGS_AXIS_SHIFT2);
 				break;
-			case BT_CONSTRAINT_CFM : 
+			case BT_CONSTRAINT_CFM:
 				m_angularLimits[axis - 3].m_motorCFM = value;
 				m_flags |= BT_6DOF_FLAGS_CFM_MOTO2 << (axis * BT_6DOF_FLAGS_AXIS_SHIFT2);
 				break;
-			default : 
+			default:
 				btAssertConstrParams(0);
 		}
 	}
@@ -915,54 +957,54 @@ void btGeneric6DofSpring2Constraint::setParam(int num, btScalar value, int axis)
 }
 
 //return the local value of parameter
-btScalar btGeneric6DofSpring2Constraint::getParam(int num, int axis) const 
+btScalar btGeneric6DofSpring2Constraint::getParam(int num, int axis) const
 {
 	btScalar retVal = 0;
-	if((axis >= 0) && (axis < 3))
+	if ((axis >= 0) && (axis < 3))
 	{
-		switch(num)
+		switch (num)
 		{
-			case BT_CONSTRAINT_STOP_ERP : 
+			case BT_CONSTRAINT_STOP_ERP:
 				btAssertConstrParams(m_flags & (BT_6DOF_FLAGS_ERP_STOP2 << (axis * BT_6DOF_FLAGS_AXIS_SHIFT2)));
 				retVal = m_linearLimits.m_stopERP[axis];
 				break;
-			case BT_CONSTRAINT_STOP_CFM : 
+			case BT_CONSTRAINT_STOP_CFM:
 				btAssertConstrParams(m_flags & (BT_6DOF_FLAGS_CFM_STOP2 << (axis * BT_6DOF_FLAGS_AXIS_SHIFT2)));
 				retVal = m_linearLimits.m_stopCFM[axis];
 				break;
-			case BT_CONSTRAINT_ERP : 
+			case BT_CONSTRAINT_ERP:
 				btAssertConstrParams(m_flags & (BT_6DOF_FLAGS_ERP_MOTO2 << (axis * BT_6DOF_FLAGS_AXIS_SHIFT2)));
 				retVal = m_linearLimits.m_motorERP[axis];
 				break;
-			case BT_CONSTRAINT_CFM : 
+			case BT_CONSTRAINT_CFM:
 				btAssertConstrParams(m_flags & (BT_6DOF_FLAGS_CFM_MOTO2 << (axis * BT_6DOF_FLAGS_AXIS_SHIFT2)));
 				retVal = m_linearLimits.m_motorCFM[axis];
 				break;
-			default : 
+			default:
 				btAssertConstrParams(0);
 		}
 	}
-	else if((axis >=3) && (axis < 6))
+	else if ((axis >= 3) && (axis < 6))
 	{
-		switch(num)
+		switch (num)
 		{
-			case BT_CONSTRAINT_STOP_ERP : 
+			case BT_CONSTRAINT_STOP_ERP:
 				btAssertConstrParams(m_flags & (BT_6DOF_FLAGS_ERP_STOP2 << (axis * BT_6DOF_FLAGS_AXIS_SHIFT2)));
 				retVal = m_angularLimits[axis - 3].m_stopERP;
 				break;
-			case BT_CONSTRAINT_STOP_CFM : 
+			case BT_CONSTRAINT_STOP_CFM:
 				btAssertConstrParams(m_flags & (BT_6DOF_FLAGS_CFM_STOP2 << (axis * BT_6DOF_FLAGS_AXIS_SHIFT2)));
 				retVal = m_angularLimits[axis - 3].m_stopCFM;
 				break;
-			case BT_CONSTRAINT_ERP : 
+			case BT_CONSTRAINT_ERP:
 				btAssertConstrParams(m_flags & (BT_6DOF_FLAGS_ERP_MOTO2 << (axis * BT_6DOF_FLAGS_AXIS_SHIFT2)));
 				retVal = m_angularLimits[axis - 3].m_motorERP;
 				break;
-			case BT_CONSTRAINT_CFM : 
+			case BT_CONSTRAINT_CFM:
 				btAssertConstrParams(m_flags & (BT_6DOF_FLAGS_CFM_MOTO2 << (axis * BT_6DOF_FLAGS_AXIS_SHIFT2)));
 				retVal = m_angularLimits[axis - 3].m_motorCFM;
 				break;
-			default : 
+			default:
 				btAssertConstrParams(0);
 		}
 	}
@@ -973,31 +1015,29 @@ btScalar btGeneric6DofSpring2Constraint::getParam(int num, int axis) const
 	return retVal;
 }
 
- 
-
-void btGeneric6DofSpring2Constraint::setAxis(const btVector3& axis1,const btVector3& axis2)
+void btGeneric6DofSpring2Constraint::setAxis(const btVector3& axis1, const btVector3& axis2)
 {
 	btVector3 zAxis = axis1.normalized();
 	btVector3 yAxis = axis2.normalized();
-	btVector3 xAxis = yAxis.cross(zAxis); // we want right coordinate system
-	
+	btVector3 xAxis = yAxis.cross(zAxis);  // we want right coordinate system
+
 	btTransform frameInW;
 	frameInW.setIdentity();
-	frameInW.getBasis().setValue( xAxis[0], yAxis[0], zAxis[0],
-	                              xAxis[1], yAxis[1], zAxis[1],
-	                              xAxis[2], yAxis[2], zAxis[2]);
-	
+	frameInW.getBasis().setValue(xAxis[0], yAxis[0], zAxis[0],
+								 xAxis[1], yAxis[1], zAxis[1],
+								 xAxis[2], yAxis[2], zAxis[2]);
+
 	// now get constraint frame in local coordinate systems
 	m_frameInA = m_rbA.getCenterOfMassTransform().inverse() * frameInW;
 	m_frameInB = m_rbB.getCenterOfMassTransform().inverse() * frameInW;
-	
+
 	calculateTransforms();
 }
 
 void btGeneric6DofSpring2Constraint::setBounce(int index, btScalar bounce)
 {
 	btAssert((index >= 0) && (index < 6));
-	if (index<3)
+	if (index < 3)
 		m_linearLimits.m_bounce[index] = bounce;
 	else
 		m_angularLimits[index - 3].m_bounce = bounce;
@@ -1006,7 +1046,7 @@ void btGeneric6DofSpring2Constraint::setBounce(int index, btScalar bounce)
 void btGeneric6DofSpring2Constraint::enableMotor(int index, bool onOff)
 {
 	btAssert((index >= 0) && (index < 6));
-	if (index<3)
+	if (index < 3)
 		m_linearLimits.m_enableMotor[index] = onOff;
 	else
 		m_angularLimits[index - 3].m_enableMotor = onOff;
@@ -1015,7 +1055,7 @@ void btGeneric6DofSpring2Constraint::enableMotor(int index, bool onOff)
 void btGeneric6DofSpring2Constraint::setServo(int index, bool onOff)
 {
 	btAssert((index >= 0) && (index < 6));
-	if (index<3)
+	if (index < 3)
 		m_linearLimits.m_servoMotor[index] = onOff;
 	else
 		m_angularLimits[index - 3].m_servoMotor = onOff;
@@ -1024,18 +1064,16 @@ void btGeneric6DofSpring2Constraint::setServo(int index, bool onOff)
 void btGeneric6DofSpring2Constraint::setTargetVelocity(int index, btScalar velocity)
 {
 	btAssert((index >= 0) && (index < 6));
-	if (index<3)
+	if (index < 3)
 		m_linearLimits.m_targetVelocity[index] = velocity;
 	else
 		m_angularLimits[index - 3].m_targetVelocity = velocity;
 }
 
-
-
 void btGeneric6DofSpring2Constraint::setServoTarget(int index, btScalar targetOrg)
 {
 	btAssert((index >= 0) && (index < 6));
-	if (index<3)
+	if (index < 3)
 	{
 		m_linearLimits.m_servoTarget[index] = targetOrg;
 	}
@@ -1044,23 +1082,24 @@ void btGeneric6DofSpring2Constraint::setServoTarget(int index, btScalar targetOr
 		//wrap between -PI and PI, see also
 		//https://stackoverflow.com/questions/4633177/c-how-to-wrap-a-float-to-the-interval-pi-pi
 
-		btScalar target = targetOrg+SIMD_PI;
+		btScalar target = targetOrg + SIMD_PI;
 		if (1)
 		{
-			btScalar m = target - SIMD_2_PI * floor(target/SIMD_2_PI);
+			btScalar m = target - SIMD_2_PI * floor(target / SIMD_2_PI);
 			// handle boundary cases resulted from floating-point cut off:
 			{
-				if (m>=SIMD_2_PI)
+				if (m >= SIMD_2_PI)
 				{
 					target = 0;
-				} else
+				}
+				else
 				{
-					if (m<0 )
+					if (m < 0)
 					{
-						if (SIMD_2_PI+m == SIMD_2_PI)
+						if (SIMD_2_PI + m == SIMD_2_PI)
 							target = 0;
 						else
-							target = SIMD_2_PI+m;
+							target = SIMD_2_PI + m;
 					}
 					else
 					{
@@ -1070,7 +1109,7 @@ void btGeneric6DofSpring2Constraint::setServoTarget(int index, btScalar targetOr
 			}
 			target -= SIMD_PI;
 		}
-		
+
 		m_angularLimits[index - 3].m_servoTarget = target;
 	}
 }
@@ -1078,7 +1117,7 @@ void btGeneric6DofSpring2Constraint::setServoTarget(int index, btScalar targetOr
 void btGeneric6DofSpring2Constraint::setMaxMotorForce(int index, btScalar force)
 {
 	btAssert((index >= 0) && (index < 6));
-	if (index<3)
+	if (index < 3)
 		m_linearLimits.m_maxMotorForce[index] = force;
 	else
 		m_angularLimits[index - 3].m_maxMotorForce = force;
@@ -1087,19 +1126,22 @@ void btGeneric6DofSpring2Constraint::setMaxMotorForce(int index, btScalar force)
 void btGeneric6DofSpring2Constraint::enableSpring(int index, bool onOff)
 {
 	btAssert((index >= 0) && (index < 6));
-	if (index<3)
+	if (index < 3)
 		m_linearLimits.m_enableSpring[index] = onOff;
 	else
-		m_angularLimits[index - 3] .m_enableSpring = onOff;
+		m_angularLimits[index - 3].m_enableSpring = onOff;
 }
 
 void btGeneric6DofSpring2Constraint::setStiffness(int index, btScalar stiffness, bool limitIfNeeded)
 {
 	btAssert((index >= 0) && (index < 6));
-	if (index<3) {
+	if (index < 3)
+	{
 		m_linearLimits.m_springStiffness[index] = stiffness;
 		m_linearLimits.m_springStiffnessLimited[index] = limitIfNeeded;
-	} else {
+	}
+	else
+	{
 		m_angularLimits[index - 3].m_springStiffness = stiffness;
 		m_angularLimits[index - 3].m_springStiffnessLimited = limitIfNeeded;
 	}
@@ -1108,10 +1150,13 @@ void btGeneric6DofSpring2Constraint::setStiffness(int index, btScalar stiffness,
 void btGeneric6DofSpring2Constraint::setDamping(int index, btScalar damping, bool limitIfNeeded)
 {
 	btAssert((index >= 0) && (index < 6));
-	if (index<3) {
+	if (index < 3)
+	{
 		m_linearLimits.m_springDamping[index] = damping;
 		m_linearLimits.m_springDampingLimited[index] = limitIfNeeded;
-	} else {
+	}
+	else
+	{
 		m_angularLimits[index - 3].m_springDamping = damping;
 		m_angularLimits[index - 3].m_springDampingLimited = limitIfNeeded;
 	}
@@ -1121,9 +1166,9 @@ void btGeneric6DofSpring2Constraint::setEquilibriumPoint()
 {
 	calculateTransforms();
 	int i;
-	for( i = 0; i < 3; i++)
+	for (i = 0; i < 3; i++)
 		m_linearLimits.m_equilibriumPoint[i] = m_calculatedLinearDiff[i];
-	for(i = 0; i < 3; i++)
+	for (i = 0; i < 3; i++)
 		m_angularLimits[i].m_equilibriumPoint = m_calculatedAxisAngleDiff[i];
 }
 
@@ -1131,35 +1176,38 @@ void btGeneric6DofSpring2Constraint::setEquilibriumPoint(int index)
 {
 	btAssert((index >= 0) && (index < 6));
 	calculateTransforms();
-	if (index<3)
+	if (index < 3)
 		m_linearLimits.m_equilibriumPoint[index] = m_calculatedLinearDiff[index];
 	else
-		m_angularLimits[index - 3] .m_equilibriumPoint = m_calculatedAxisAngleDiff[index - 3];
+		m_angularLimits[index - 3].m_equilibriumPoint = m_calculatedAxisAngleDiff[index - 3];
 }
 
 void btGeneric6DofSpring2Constraint::setEquilibriumPoint(int index, btScalar val)
 {
 	btAssert((index >= 0) && (index < 6));
-	if (index<3)
+	if (index < 3)
 		m_linearLimits.m_equilibriumPoint[index] = val;
 	else
-		m_angularLimits[index - 3] .m_equilibriumPoint = val;
+		m_angularLimits[index - 3].m_equilibriumPoint = val;
 }
 
-
 //////////////////////////// btRotationalLimitMotor2 ////////////////////////////////////
 
 void btRotationalLimitMotor2::testLimitValue(btScalar test_value)
 {
 	//we can't normalize the angles here because we would lost the sign that we use later, but it doesn't seem to be a problem
-	if(m_loLimit > m_hiLimit) {
+	if (m_loLimit > m_hiLimit)
+	{
 		m_currentLimit = 0;
 		m_currentLimitError = btScalar(0.f);
 	}
-	else if(m_loLimit == m_hiLimit) {
+	else if (m_loLimit == m_hiLimit)
+	{
 		m_currentLimitError = test_value - m_loLimit;
 		m_currentLimit = 3;
-	} else {
+	}
+	else
+	{
 		m_currentLimitError = test_value - m_loLimit;
 		m_currentLimitErrorHi = test_value - m_hiLimit;
 		m_currentLimit = 4;
@@ -1172,18 +1220,20 @@ void btTranslationalLimitMotor2::testLimitValue(int limitIndex, btScalar test_va
 {
 	btScalar loLimit = m_lowerLimit[limitIndex];
 	btScalar hiLimit = m_upperLimit[limitIndex];
-	if(loLimit > hiLimit) {
+	if (loLimit > hiLimit)
+	{
 		m_currentLimitError[limitIndex] = 0;
 		m_currentLimit[limitIndex] = 0;
 	}
-	else if(loLimit == hiLimit) {
+	else if (loLimit == hiLimit)
+	{
 		m_currentLimitError[limitIndex] = test_value - loLimit;
 		m_currentLimit[limitIndex] = 3;
-	} else {
+	}
+	else
+	{
 		m_currentLimitError[limitIndex] = test_value - loLimit;
 		m_currentLimitErrorHi[limitIndex] = test_value - hiLimit;
 		m_currentLimit[limitIndex] = 4;
 	}
 }
-
-
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofSpring2Constraint.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofSpring2Constraint.h
index 1b8d0eace9d..bc3ee6d2103 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofSpring2Constraint.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofSpring2Constraint.h
@@ -37,7 +37,6 @@ email: projectileman@yahoo.com
 http://gimpact.sf.net
 */
 
-
 #ifndef BT_GENERIC_6DOF_CONSTRAINT2_H
 #define BT_GENERIC_6DOF_CONSTRAINT2_H
 
@@ -47,18 +46,17 @@ http://gimpact.sf.net
 
 class btRigidBody;
 
-
 #ifdef BT_USE_DOUBLE_PRECISION
-#define btGeneric6DofSpring2ConstraintData2		btGeneric6DofSpring2ConstraintDoubleData2
-#define btGeneric6DofSpring2ConstraintDataName	"btGeneric6DofSpring2ConstraintDoubleData2"
+#define btGeneric6DofSpring2ConstraintData2 btGeneric6DofSpring2ConstraintDoubleData2
+#define btGeneric6DofSpring2ConstraintDataName "btGeneric6DofSpring2ConstraintDoubleData2"
 #else
-#define btGeneric6DofSpring2ConstraintData2		btGeneric6DofSpring2ConstraintData
-#define btGeneric6DofSpring2ConstraintDataName	"btGeneric6DofSpring2ConstraintData"
-#endif //BT_USE_DOUBLE_PRECISION
+#define btGeneric6DofSpring2ConstraintData2 btGeneric6DofSpring2ConstraintData
+#define btGeneric6DofSpring2ConstraintDataName "btGeneric6DofSpring2ConstraintData"
+#endif  //BT_USE_DOUBLE_PRECISION
 
 enum RotateOrder
 {
-	RO_XYZ=0,
+	RO_XYZ = 0,
 	RO_XZY,
 	RO_YXZ,
 	RO_YZX,
@@ -69,9 +67,9 @@ enum RotateOrder
 class btRotationalLimitMotor2
 {
 public:
-// upper < lower means free
-// upper == lower means locked
-// upper > lower means limited
+	// upper < lower means free
+	// upper == lower means locked
+	// upper > lower means limited
 	btScalar m_loLimit;
 	btScalar m_hiLimit;
 	btScalar m_bounce;
@@ -79,95 +77,92 @@ public:
 	btScalar m_stopCFM;
 	btScalar m_motorERP;
 	btScalar m_motorCFM;
-	bool     m_enableMotor;
+	bool m_enableMotor;
 	btScalar m_targetVelocity;
 	btScalar m_maxMotorForce;
-	bool     m_servoMotor;
+	bool m_servoMotor;
 	btScalar m_servoTarget;
-	bool     m_enableSpring;
+	bool m_enableSpring;
 	btScalar m_springStiffness;
-	bool     m_springStiffnessLimited;
+	bool m_springStiffnessLimited;
 	btScalar m_springDamping;
-	bool     m_springDampingLimited;
+	bool m_springDampingLimited;
 	btScalar m_equilibriumPoint;
 
 	btScalar m_currentLimitError;
 	btScalar m_currentLimitErrorHi;
 	btScalar m_currentPosition;
-	int      m_currentLimit;
+	int m_currentLimit;
 
 	btRotationalLimitMotor2()
 	{
-		m_loLimit                = 1.0f;
-		m_hiLimit                = -1.0f;
-		m_bounce                 = 0.0f;
-		m_stopERP                = 0.2f;
-		m_stopCFM                = 0.f;
-		m_motorERP               = 0.9f;
-		m_motorCFM               = 0.f;
-		m_enableMotor            = false;
-		m_targetVelocity         = 0;
-		m_maxMotorForce          = 6.0f;
-		m_servoMotor             = false;
-		m_servoTarget            = 0;
-		m_enableSpring           = false;
-		m_springStiffness        = 0;
+		m_loLimit = 1.0f;
+		m_hiLimit = -1.0f;
+		m_bounce = 0.0f;
+		m_stopERP = 0.2f;
+		m_stopCFM = 0.f;
+		m_motorERP = 0.9f;
+		m_motorCFM = 0.f;
+		m_enableMotor = false;
+		m_targetVelocity = 0;
+		m_maxMotorForce = 6.0f;
+		m_servoMotor = false;
+		m_servoTarget = 0;
+		m_enableSpring = false;
+		m_springStiffness = 0;
 		m_springStiffnessLimited = false;
-		m_springDamping          = 0;
-		m_springDampingLimited   = false;
-		m_equilibriumPoint       = 0;
+		m_springDamping = 0;
+		m_springDampingLimited = false;
+		m_equilibriumPoint = 0;
 
-		m_currentLimitError   = 0;
+		m_currentLimitError = 0;
 		m_currentLimitErrorHi = 0;
-		m_currentPosition     = 0;
-		m_currentLimit        = 0;
+		m_currentPosition = 0;
+		m_currentLimit = 0;
 	}
 
-	btRotationalLimitMotor2(const btRotationalLimitMotor2 & limot)
+	btRotationalLimitMotor2(const btRotationalLimitMotor2& limot)
 	{
-		m_loLimit                = limot.m_loLimit;
-		m_hiLimit                = limot.m_hiLimit;
-		m_bounce                 = limot.m_bounce;
-		m_stopERP                = limot.m_stopERP;
-		m_stopCFM                = limot.m_stopCFM;
-		m_motorERP               = limot.m_motorERP;
-		m_motorCFM               = limot.m_motorCFM;
-		m_enableMotor            = limot.m_enableMotor;
-		m_targetVelocity         = limot.m_targetVelocity;
-		m_maxMotorForce          = limot.m_maxMotorForce;
-		m_servoMotor             = limot.m_servoMotor;
-		m_servoTarget            = limot.m_servoTarget;
-		m_enableSpring           = limot.m_enableSpring;
-		m_springStiffness        = limot.m_springStiffness;
+		m_loLimit = limot.m_loLimit;
+		m_hiLimit = limot.m_hiLimit;
+		m_bounce = limot.m_bounce;
+		m_stopERP = limot.m_stopERP;
+		m_stopCFM = limot.m_stopCFM;
+		m_motorERP = limot.m_motorERP;
+		m_motorCFM = limot.m_motorCFM;
+		m_enableMotor = limot.m_enableMotor;
+		m_targetVelocity = limot.m_targetVelocity;
+		m_maxMotorForce = limot.m_maxMotorForce;
+		m_servoMotor = limot.m_servoMotor;
+		m_servoTarget = limot.m_servoTarget;
+		m_enableSpring = limot.m_enableSpring;
+		m_springStiffness = limot.m_springStiffness;
 		m_springStiffnessLimited = limot.m_springStiffnessLimited;
-		m_springDamping          = limot.m_springDamping;
-		m_springDampingLimited   = limot.m_springDampingLimited;
-		m_equilibriumPoint       = limot.m_equilibriumPoint;
+		m_springDamping = limot.m_springDamping;
+		m_springDampingLimited = limot.m_springDampingLimited;
+		m_equilibriumPoint = limot.m_equilibriumPoint;
 
-		m_currentLimitError   = limot.m_currentLimitError;
+		m_currentLimitError = limot.m_currentLimitError;
 		m_currentLimitErrorHi = limot.m_currentLimitErrorHi;
-		m_currentPosition     = limot.m_currentPosition;
-		m_currentLimit        = limot.m_currentLimit;
+		m_currentPosition = limot.m_currentPosition;
+		m_currentLimit = limot.m_currentLimit;
 	}
 
-
 	bool isLimited()
 	{
-		if(m_loLimit > m_hiLimit) return false;
+		if (m_loLimit > m_hiLimit) return false;
 		return true;
 	}
 
 	void testLimitValue(btScalar test_value);
 };
 
-
-
 class btTranslationalLimitMotor2
 {
 public:
-// upper < lower means free
-// upper == lower means locked
-// upper > lower means limited
+	// upper < lower means free
+	// upper == lower means locked
+	// upper > lower means limited
 	btVector3 m_lowerLimit;
 	btVector3 m_upperLimit;
 	btVector3 m_bounce;
@@ -175,14 +170,14 @@ public:
 	btVector3 m_stopCFM;
 	btVector3 m_motorERP;
 	btVector3 m_motorCFM;
-	bool      m_enableMotor[3];
-	bool      m_servoMotor[3];
-	bool      m_enableSpring[3];
+	bool m_enableMotor[3];
+	bool m_servoMotor[3];
+	bool m_enableSpring[3];
 	btVector3 m_servoTarget;
 	btVector3 m_springStiffness;
-	bool      m_springStiffnessLimited[3];
+	bool m_springStiffnessLimited[3];
 	btVector3 m_springDamping;
-	bool      m_springDampingLimited[3];
+	bool m_springDampingLimited[3];
 	btVector3 m_equilibriumPoint;
 	btVector3 m_targetVelocity;
 	btVector3 m_maxMotorForce;
@@ -190,69 +185,69 @@ public:
 	btVector3 m_currentLimitError;
 	btVector3 m_currentLimitErrorHi;
 	btVector3 m_currentLinearDiff;
-	int       m_currentLimit[3];
+	int m_currentLimit[3];
 
 	btTranslationalLimitMotor2()
 	{
-		m_lowerLimit         .setValue(0.f , 0.f , 0.f );
-		m_upperLimit         .setValue(0.f , 0.f , 0.f );
-		m_bounce             .setValue(0.f , 0.f , 0.f );
-		m_stopERP            .setValue(0.2f, 0.2f, 0.2f);
-		m_stopCFM            .setValue(0.f , 0.f , 0.f );
-		m_motorERP           .setValue(0.9f, 0.9f, 0.9f);
-		m_motorCFM           .setValue(0.f , 0.f , 0.f );
+		m_lowerLimit.setValue(0.f, 0.f, 0.f);
+		m_upperLimit.setValue(0.f, 0.f, 0.f);
+		m_bounce.setValue(0.f, 0.f, 0.f);
+		m_stopERP.setValue(0.2f, 0.2f, 0.2f);
+		m_stopCFM.setValue(0.f, 0.f, 0.f);
+		m_motorERP.setValue(0.9f, 0.9f, 0.9f);
+		m_motorCFM.setValue(0.f, 0.f, 0.f);
 
-		m_currentLimitError  .setValue(0.f , 0.f , 0.f );
-		m_currentLimitErrorHi.setValue(0.f , 0.f , 0.f );
-		m_currentLinearDiff  .setValue(0.f , 0.f , 0.f );
+		m_currentLimitError.setValue(0.f, 0.f, 0.f);
+		m_currentLimitErrorHi.setValue(0.f, 0.f, 0.f);
+		m_currentLinearDiff.setValue(0.f, 0.f, 0.f);
 
-		for(int i=0; i < 3; i++) 
+		for (int i = 0; i < 3; i++)
 		{
-			m_enableMotor[i]            = false;
-			m_servoMotor[i]             = false;
-			m_enableSpring[i]           = false;
-			m_servoTarget[i]            = btScalar(0.f);
-			m_springStiffness[i]        = btScalar(0.f);
+			m_enableMotor[i] = false;
+			m_servoMotor[i] = false;
+			m_enableSpring[i] = false;
+			m_servoTarget[i] = btScalar(0.f);
+			m_springStiffness[i] = btScalar(0.f);
 			m_springStiffnessLimited[i] = false;
-			m_springDamping[i]          = btScalar(0.f);
-			m_springDampingLimited[i]   = false;
-			m_equilibriumPoint[i]       = btScalar(0.f);
-			m_targetVelocity[i]         = btScalar(0.f);
-			m_maxMotorForce[i]          = btScalar(0.f);
-			
-			m_currentLimit[i]     = 0;
+			m_springDamping[i] = btScalar(0.f);
+			m_springDampingLimited[i] = false;
+			m_equilibriumPoint[i] = btScalar(0.f);
+			m_targetVelocity[i] = btScalar(0.f);
+			m_maxMotorForce[i] = btScalar(0.f);
+
+			m_currentLimit[i] = 0;
 		}
 	}
 
-	btTranslationalLimitMotor2(const btTranslationalLimitMotor2 & other )
+	btTranslationalLimitMotor2(const btTranslationalLimitMotor2& other)
 	{
-		m_lowerLimit          = other.m_lowerLimit;
-		m_upperLimit          = other.m_upperLimit;
-		m_bounce              = other.m_bounce;
-		m_stopERP             = other.m_stopERP;
-		m_stopCFM             = other.m_stopCFM;
-		m_motorERP            = other.m_motorERP;
-		m_motorCFM            = other.m_motorCFM;
-		
-		m_currentLimitError   = other.m_currentLimitError;
+		m_lowerLimit = other.m_lowerLimit;
+		m_upperLimit = other.m_upperLimit;
+		m_bounce = other.m_bounce;
+		m_stopERP = other.m_stopERP;
+		m_stopCFM = other.m_stopCFM;
+		m_motorERP = other.m_motorERP;
+		m_motorCFM = other.m_motorCFM;
+
+		m_currentLimitError = other.m_currentLimitError;
 		m_currentLimitErrorHi = other.m_currentLimitErrorHi;
-		m_currentLinearDiff   = other.m_currentLinearDiff;
+		m_currentLinearDiff = other.m_currentLinearDiff;
 
-		for(int i=0; i < 3; i++) 
+		for (int i = 0; i < 3; i++)
 		{
-			m_enableMotor[i]            = other.m_enableMotor[i];
-			m_servoMotor[i]             = other.m_servoMotor[i];
-			m_enableSpring[i]           = other.m_enableSpring[i];
-			m_servoTarget[i]            = other.m_servoTarget[i];
-			m_springStiffness[i]        = other.m_springStiffness[i];
+			m_enableMotor[i] = other.m_enableMotor[i];
+			m_servoMotor[i] = other.m_servoMotor[i];
+			m_enableSpring[i] = other.m_enableSpring[i];
+			m_servoTarget[i] = other.m_servoTarget[i];
+			m_springStiffness[i] = other.m_springStiffness[i];
 			m_springStiffnessLimited[i] = other.m_springStiffnessLimited[i];
-			m_springDamping[i]          = other.m_springDamping[i];
-			m_springDampingLimited[i]   = other.m_springDampingLimited[i];
-			m_equilibriumPoint[i]       = other.m_equilibriumPoint[i];
-			m_targetVelocity[i]         = other.m_targetVelocity[i];
-			m_maxMotorForce[i]          = other.m_maxMotorForce[i];
+			m_springDamping[i] = other.m_springDamping[i];
+			m_springDampingLimited[i] = other.m_springDampingLimited[i];
+			m_equilibriumPoint[i] = other.m_equilibriumPoint[i];
+			m_targetVelocity[i] = other.m_targetVelocity[i];
+			m_maxMotorForce[i] = other.m_maxMotorForce[i];
 
-			m_currentLimit[i]     = other.m_currentLimit[i];
+			m_currentLimit[i] = other.m_currentLimit[i];
 		}
 	}
 
@@ -269,15 +264,14 @@ enum bt6DofFlags2
 	BT_6DOF_FLAGS_CFM_STOP2 = 1,
 	BT_6DOF_FLAGS_ERP_STOP2 = 2,
 	BT_6DOF_FLAGS_CFM_MOTO2 = 4,
-	BT_6DOF_FLAGS_ERP_MOTO2 = 8
+	BT_6DOF_FLAGS_ERP_MOTO2 = 8,
 };
-#define BT_6DOF_FLAGS_AXIS_SHIFT2 4 // bits per axis
+#define BT_6DOF_FLAGS_AXIS_SHIFT2 4  // bits per axis
 
-
-ATTRIBUTE_ALIGNED16(class) btGeneric6DofSpring2Constraint : public btTypedConstraint
+ATTRIBUTE_ALIGNED16(class)
+btGeneric6DofSpring2Constraint : public btTypedConstraint
 {
 protected:
-
 	btTransform m_frameInA;
 	btTransform m_frameInB;
 
@@ -290,45 +284,43 @@ protected:
 	RotateOrder m_rotateOrder;
 
 protected:
+	btTransform m_calculatedTransformA;
+	btTransform m_calculatedTransformB;
+	btVector3 m_calculatedAxisAngleDiff;
+	btVector3 m_calculatedAxis[3];
+	btVector3 m_calculatedLinearDiff;
+	btScalar m_factA;
+	btScalar m_factB;
+	bool m_hasStaticBody;
+	int m_flags;
 
-	btTransform  m_calculatedTransformA;
-	btTransform  m_calculatedTransformB;
-	btVector3    m_calculatedAxisAngleDiff;
-	btVector3    m_calculatedAxis[3];
-	btVector3    m_calculatedLinearDiff;
-	btScalar     m_factA;
-	btScalar     m_factB;
-	bool         m_hasStaticBody;
-	int          m_flags;
-
-	btGeneric6DofSpring2Constraint&	operator=(btGeneric6DofSpring2Constraint&)
+	btGeneric6DofSpring2Constraint& operator=(btGeneric6DofSpring2Constraint&)
 	{
 		btAssert(0);
 		return *this;
 	}
 
-	int setAngularLimits(btConstraintInfo2 *info, int row_offset,const btTransform& transA,const btTransform& transB,const btVector3& linVelA,const btVector3& linVelB,const btVector3& angVelA,const btVector3& angVelB);
-	int setLinearLimits(btConstraintInfo2 *info, int row, const btTransform& transA,const btTransform& transB,const btVector3& linVelA,const btVector3& linVelB,const btVector3& angVelA,const btVector3& angVelB);
+	int setAngularLimits(btConstraintInfo2 * info, int row_offset, const btTransform& transA, const btTransform& transB, const btVector3& linVelA, const btVector3& linVelB, const btVector3& angVelA, const btVector3& angVelB);
+	int setLinearLimits(btConstraintInfo2 * info, int row, const btTransform& transA, const btTransform& transB, const btVector3& linVelA, const btVector3& linVelB, const btVector3& angVelA, const btVector3& angVelB);
 
 	void calculateLinearInfo();
 	void calculateAngleInfo();
 	void testAngularLimitMotor(int axis_index);
 
-	void calculateJacobi(btRotationalLimitMotor2* limot, const btTransform& transA,const btTransform& transB, btConstraintInfo2* info, int srow, btVector3& ax1, int rotational, int rotAllowed);
-	int get_limit_motor_info2(btRotationalLimitMotor2* limot,
-		const btTransform& transA,const btTransform& transB,const btVector3& linVelA,const btVector3& linVelB,const btVector3& angVelA,const btVector3& angVelB,
-		btConstraintInfo2* info, int row, btVector3& ax1, int rotational, int rotAllowed = false);
+	void calculateJacobi(btRotationalLimitMotor2 * limot, const btTransform& transA, const btTransform& transB, btConstraintInfo2* info, int srow, btVector3& ax1, int rotational, int rotAllowed);
+	int get_limit_motor_info2(btRotationalLimitMotor2 * limot,
+							  const btTransform& transA, const btTransform& transB, const btVector3& linVelA, const btVector3& linVelB, const btVector3& angVelA, const btVector3& angVelB,
+							  btConstraintInfo2* info, int row, btVector3& ax1, int rotational, int rotAllowed = false);
 
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-    btGeneric6DofSpring2Constraint(btRigidBody& rbA, btRigidBody& rbB, const btTransform& frameInA, const btTransform& frameInB, RotateOrder rotOrder = RO_XYZ);
-    btGeneric6DofSpring2Constraint(btRigidBody& rbB, const btTransform& frameInB, RotateOrder rotOrder = RO_XYZ);
+	btGeneric6DofSpring2Constraint(btRigidBody & rbA, btRigidBody & rbB, const btTransform& frameInA, const btTransform& frameInB, RotateOrder rotOrder = RO_XYZ);
+	btGeneric6DofSpring2Constraint(btRigidBody & rbB, const btTransform& frameInB, RotateOrder rotOrder = RO_XYZ);
 
 	virtual void buildJacobian() {}
-	virtual void getInfo1 (btConstraintInfo1* info);
-	virtual void getInfo2 (btConstraintInfo2* info);
+	virtual void getInfo1(btConstraintInfo1 * info);
+	virtual void getInfo2(btConstraintInfo2 * info);
 	virtual int calculateSerializeBufferSize() const;
 	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 
@@ -336,19 +328,19 @@ public:
 	btTranslationalLimitMotor2* getTranslationalLimitMotor() { return &m_linearLimits; }
 
 	// Calculates the global transform for the joint offset for body A an B, and also calculates the angle differences between the bodies.
-	void calculateTransforms(const btTransform& transA,const btTransform& transB);
+	void calculateTransforms(const btTransform& transA, const btTransform& transB);
 	void calculateTransforms();
 
 	// Gets the global transform of the offset for body A
-	const btTransform & getCalculatedTransformA() const { return m_calculatedTransformA; }
+	const btTransform& getCalculatedTransformA() const { return m_calculatedTransformA; }
 	// Gets the global transform of the offset for body B
-	const btTransform & getCalculatedTransformB() const { return m_calculatedTransformB; }
+	const btTransform& getCalculatedTransformB() const { return m_calculatedTransformB; }
 
-	const btTransform & getFrameOffsetA() const { return m_frameInA; }
-	const btTransform & getFrameOffsetB() const { return m_frameInB; }
+	const btTransform& getFrameOffsetA() const { return m_frameInA; }
+	const btTransform& getFrameOffsetB() const { return m_frameInB; }
 
-	btTransform & getFrameOffsetA() { return m_frameInA; }
-	btTransform & getFrameOffsetB() { return m_frameInB; }
+	btTransform& getFrameOffsetA() { return m_frameInA; }
+	btTransform& getFrameOffsetB() { return m_frameInB; }
 
 	// Get the rotation axis in global coordinates ( btGeneric6DofSpring2Constraint::calculateTransforms() must be called previously )
 	btVector3 getAxis(int axis_index) const { return m_calculatedAxis[axis_index]; }
@@ -359,58 +351,58 @@ public:
 	// Get the relative position of the constraint pivot ( btGeneric6DofSpring2Constraint::calculateTransforms() must be called previously )
 	btScalar getRelativePivotPosition(int axis_index) const { return m_calculatedLinearDiff[axis_index]; }
 
-	void setFrames(const btTransform & frameA, const btTransform & frameB);
+	void setFrames(const btTransform& frameA, const btTransform& frameB);
 
 	void setLinearLowerLimit(const btVector3& linearLower) { m_linearLimits.m_lowerLimit = linearLower; }
-	void getLinearLowerLimit(btVector3& linearLower) { linearLower = m_linearLimits.m_lowerLimit; }
+	void getLinearLowerLimit(btVector3 & linearLower) { linearLower = m_linearLimits.m_lowerLimit; }
 	void setLinearUpperLimit(const btVector3& linearUpper) { m_linearLimits.m_upperLimit = linearUpper; }
-	void getLinearUpperLimit(btVector3& linearUpper) { linearUpper = m_linearLimits.m_upperLimit; }
+	void getLinearUpperLimit(btVector3 & linearUpper) { linearUpper = m_linearLimits.m_upperLimit; }
 
 	void setAngularLowerLimit(const btVector3& angularLower)
 	{
-		for(int i = 0; i < 3; i++) 
+		for (int i = 0; i < 3; i++)
 			m_angularLimits[i].m_loLimit = btNormalizeAngle(angularLower[i]);
 	}
 
 	void setAngularLowerLimitReversed(const btVector3& angularLower)
 	{
-		for(int i = 0; i < 3; i++) 
+		for (int i = 0; i < 3; i++)
 			m_angularLimits[i].m_hiLimit = btNormalizeAngle(-angularLower[i]);
 	}
 
-	void getAngularLowerLimit(btVector3& angularLower)
+	void getAngularLowerLimit(btVector3 & angularLower)
 	{
-		for(int i = 0; i < 3; i++) 
+		for (int i = 0; i < 3; i++)
 			angularLower[i] = m_angularLimits[i].m_loLimit;
 	}
 
-	void getAngularLowerLimitReversed(btVector3& angularLower)
+	void getAngularLowerLimitReversed(btVector3 & angularLower)
 	{
-		for(int i = 0; i < 3; i++)
+		for (int i = 0; i < 3; i++)
 			angularLower[i] = -m_angularLimits[i].m_hiLimit;
 	}
 
 	void setAngularUpperLimit(const btVector3& angularUpper)
 	{
-		for(int i = 0; i < 3; i++)
+		for (int i = 0; i < 3; i++)
 			m_angularLimits[i].m_hiLimit = btNormalizeAngle(angularUpper[i]);
 	}
 
 	void setAngularUpperLimitReversed(const btVector3& angularUpper)
 	{
-		for(int i = 0; i < 3; i++)
+		for (int i = 0; i < 3; i++)
 			m_angularLimits[i].m_loLimit = btNormalizeAngle(-angularUpper[i]);
 	}
 
-	void getAngularUpperLimit(btVector3& angularUpper)
+	void getAngularUpperLimit(btVector3 & angularUpper)
 	{
-		for(int i = 0; i < 3; i++)
+		for (int i = 0; i < 3; i++)
 			angularUpper[i] = m_angularLimits[i].m_hiLimit;
 	}
 
-	void getAngularUpperLimitReversed(btVector3& angularUpper)
+	void getAngularUpperLimitReversed(btVector3 & angularUpper)
 	{
-		for(int i = 0; i < 3; i++)
+		for (int i = 0; i < 3; i++)
 			angularUpper[i] = -m_angularLimits[i].m_loLimit;
 	}
 
@@ -418,7 +410,7 @@ public:
 
 	void setLimit(int axis, btScalar lo, btScalar hi)
 	{
-		if(axis<3)
+		if (axis < 3)
 		{
 			m_linearLimits.m_lowerLimit[axis] = lo;
 			m_linearLimits.m_upperLimit[axis] = hi;
@@ -427,14 +419,14 @@ public:
 		{
 			lo = btNormalizeAngle(lo);
 			hi = btNormalizeAngle(hi);
-			m_angularLimits[axis-3].m_loLimit = lo;
-			m_angularLimits[axis-3].m_hiLimit = hi;
+			m_angularLimits[axis - 3].m_loLimit = lo;
+			m_angularLimits[axis - 3].m_hiLimit = hi;
 		}
 	}
 
 	void setLimitReversed(int axis, btScalar lo, btScalar hi)
 	{
-		if(axis<3)
+		if (axis < 3)
 		{
 			m_linearLimits.m_lowerLimit[axis] = lo;
 			m_linearLimits.m_upperLimit[axis] = hi;
@@ -443,54 +435,53 @@ public:
 		{
 			lo = btNormalizeAngle(lo);
 			hi = btNormalizeAngle(hi);
-			m_angularLimits[axis-3].m_hiLimit = -lo;
-			m_angularLimits[axis-3].m_loLimit = -hi;
+			m_angularLimits[axis - 3].m_hiLimit = -lo;
+			m_angularLimits[axis - 3].m_loLimit = -hi;
 		}
 	}
 
 	bool isLimited(int limitIndex)
 	{
-		if(limitIndex<3)
+		if (limitIndex < 3)
 		{
 			return m_linearLimits.isLimited(limitIndex);
 		}
-		return m_angularLimits[limitIndex-3].isLimited();
+		return m_angularLimits[limitIndex - 3].isLimited();
 	}
 
 	void setRotationOrder(RotateOrder order) { m_rotateOrder = order; }
 	RotateOrder getRotationOrder() { return m_rotateOrder; }
 
-	void setAxis( const btVector3& axis1, const btVector3& axis2);
+	void setAxis(const btVector3& axis1, const btVector3& axis2);
 
 	void setBounce(int index, btScalar bounce);
 
 	void enableMotor(int index, bool onOff);
-	void setServo(int index, bool onOff); // set the type of the motor (servo or not) (the motor has to be turned on for servo also)
+	void setServo(int index, bool onOff);  // set the type of the motor (servo or not) (the motor has to be turned on for servo also)
 	void setTargetVelocity(int index, btScalar velocity);
 	void setServoTarget(int index, btScalar target);
 	void setMaxMotorForce(int index, btScalar force);
 
 	void enableSpring(int index, bool onOff);
-	void setStiffness(int index, btScalar stiffness, bool limitIfNeeded = true); // if limitIfNeeded is true the system will automatically limit the stiffness in necessary situations where otherwise the spring would move unrealistically too widely
-	void setDamping(int index, btScalar damping, bool limitIfNeeded = true); // if limitIfNeeded is true the system will automatically limit the damping in necessary situations where otherwise the spring would blow up
-	void setEquilibriumPoint(); // set the current constraint position/orientation as an equilibrium point for all DOF
-	void setEquilibriumPoint(int index);  // set the current constraint position/orientation as an equilibrium point for given DOF
+	void setStiffness(int index, btScalar stiffness, bool limitIfNeeded = true);  // if limitIfNeeded is true the system will automatically limit the stiffness in necessary situations where otherwise the spring would move unrealistically too widely
+	void setDamping(int index, btScalar damping, bool limitIfNeeded = true);      // if limitIfNeeded is true the system will automatically limit the damping in necessary situations where otherwise the spring would blow up
+	void setEquilibriumPoint();                                                   // set the current constraint position/orientation as an equilibrium point for all DOF
+	void setEquilibriumPoint(int index);                                          // set the current constraint position/orientation as an equilibrium point for given DOF
 	void setEquilibriumPoint(int index, btScalar val);
 
-	//override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5). 
+	//override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5).
 	//If no axis is provided, it uses the default axis for this constraint.
 	virtual void setParam(int num, btScalar value, int axis = -1);
 	virtual btScalar getParam(int num, int axis = -1) const;
-    
-    static btScalar btGetMatrixElem(const btMatrix3x3& mat, int index);
-    static bool matrixToEulerXYZ(const btMatrix3x3& mat,btVector3& xyz);
-    static bool matrixToEulerXZY(const btMatrix3x3& mat,btVector3& xyz);
-    static bool matrixToEulerYXZ(const btMatrix3x3& mat,btVector3& xyz);
-    static bool matrixToEulerYZX(const btMatrix3x3& mat,btVector3& xyz);
-    static bool matrixToEulerZXY(const btMatrix3x3& mat,btVector3& xyz);
-    static bool matrixToEulerZYX(const btMatrix3x3& mat,btVector3& xyz);
-};
 
+	static btScalar btGetMatrixElem(const btMatrix3x3& mat, int index);
+	static bool matrixToEulerXYZ(const btMatrix3x3& mat, btVector3& xyz);
+	static bool matrixToEulerXZY(const btMatrix3x3& mat, btVector3& xyz);
+	static bool matrixToEulerYXZ(const btMatrix3x3& mat, btVector3& xyz);
+	static bool matrixToEulerYZX(const btMatrix3x3& mat, btVector3& xyz);
+	static bool matrixToEulerZXY(const btMatrix3x3& mat, btVector3& xyz);
+	static bool matrixToEulerZYX(const btMatrix3x3& mat, btVector3& xyz);
+};
 
 struct btGeneric6DofSpring2ConstraintData
 {
@@ -511,12 +502,12 @@ struct btGeneric6DofSpring2ConstraintData
 	btVector3FloatData m_linearSpringStiffness;
 	btVector3FloatData m_linearSpringDamping;
 	btVector3FloatData m_linearEquilibriumPoint;
-	char               m_linearEnableMotor[4];
-	char               m_linearServoMotor[4];
-	char               m_linearEnableSpring[4];
-	char               m_linearSpringStiffnessLimited[4];
-	char               m_linearSpringDampingLimited[4];
-	char               m_padding1[4];
+	char m_linearEnableMotor[4];
+	char m_linearServoMotor[4];
+	char m_linearEnableSpring[4];
+	char m_linearSpringStiffnessLimited[4];
+	char m_linearSpringDampingLimited[4];
+	char m_padding1[4];
 
 	btVector3FloatData m_angularUpperLimit;
 	btVector3FloatData m_angularLowerLimit;
@@ -531,13 +522,13 @@ struct btGeneric6DofSpring2ConstraintData
 	btVector3FloatData m_angularSpringStiffness;
 	btVector3FloatData m_angularSpringDamping;
 	btVector3FloatData m_angularEquilibriumPoint;
-	char               m_angularEnableMotor[4];
-	char               m_angularServoMotor[4];
-	char               m_angularEnableSpring[4];
-	char               m_angularSpringStiffnessLimited[4];
-	char               m_angularSpringDampingLimited[4];
+	char m_angularEnableMotor[4];
+	char m_angularServoMotor[4];
+	char m_angularEnableSpring[4];
+	char m_angularSpringStiffnessLimited[4];
+	char m_angularSpringDampingLimited[4];
 
-	int                m_rotateOrder;
+	int m_rotateOrder;
 };
 
 struct btGeneric6DofSpring2ConstraintDoubleData2
@@ -559,12 +550,12 @@ struct btGeneric6DofSpring2ConstraintDoubleData2
 	btVector3DoubleData m_linearSpringStiffness;
 	btVector3DoubleData m_linearSpringDamping;
 	btVector3DoubleData m_linearEquilibriumPoint;
-	char                m_linearEnableMotor[4];
-	char                m_linearServoMotor[4];
-	char                m_linearEnableSpring[4];
-	char                m_linearSpringStiffnessLimited[4];
-	char                m_linearSpringDampingLimited[4];
-	char                m_padding1[4];
+	char m_linearEnableMotor[4];
+	char m_linearServoMotor[4];
+	char m_linearEnableSpring[4];
+	char m_linearSpringStiffnessLimited[4];
+	char m_linearSpringDampingLimited[4];
+	char m_padding1[4];
 
 	btVector3DoubleData m_angularUpperLimit;
 	btVector3DoubleData m_angularLowerLimit;
@@ -579,13 +570,13 @@ struct btGeneric6DofSpring2ConstraintDoubleData2
 	btVector3DoubleData m_angularSpringStiffness;
 	btVector3DoubleData m_angularSpringDamping;
 	btVector3DoubleData m_angularEquilibriumPoint;
-	char                m_angularEnableMotor[4];
-	char                m_angularServoMotor[4];
-	char                m_angularEnableSpring[4];
-	char                m_angularSpringStiffnessLimited[4];
-	char                m_angularSpringDampingLimited[4];
+	char m_angularEnableMotor[4];
+	char m_angularServoMotor[4];
+	char m_angularEnableSpring[4];
+	char m_angularSpringStiffnessLimited[4];
+	char m_angularSpringDampingLimited[4];
 
-	int                 m_rotateOrder;
+	int m_rotateOrder;
 };
 
 SIMD_FORCE_INLINE int btGeneric6DofSpring2Constraint::calculateSerializeBufferSize() const
@@ -596,70 +587,70 @@ SIMD_FORCE_INLINE int btGeneric6DofSpring2Constraint::calculateSerializeBufferSi
 SIMD_FORCE_INLINE const char* btGeneric6DofSpring2Constraint::serialize(void* dataBuffer, btSerializer* serializer) const
 {
 	btGeneric6DofSpring2ConstraintData2* dof = (btGeneric6DofSpring2ConstraintData2*)dataBuffer;
-	btTypedConstraint::serialize(&dof->m_typeConstraintData,serializer);
+	btTypedConstraint::serialize(&dof->m_typeConstraintData, serializer);
 
 	m_frameInA.serialize(dof->m_rbAFrame);
 	m_frameInB.serialize(dof->m_rbBFrame);
 
 	int i;
-	for (i=0;i<3;i++)
+	for (i = 0; i < 3; i++)
 	{
-		dof->m_angularLowerLimit.m_floats[i]       = m_angularLimits[i].m_loLimit;
-		dof->m_angularUpperLimit.m_floats[i]       = m_angularLimits[i].m_hiLimit;
-		dof->m_angularBounce.m_floats[i]           = m_angularLimits[i].m_bounce;
-		dof->m_angularStopERP.m_floats[i]          = m_angularLimits[i].m_stopERP;
-		dof->m_angularStopCFM.m_floats[i]          = m_angularLimits[i].m_stopCFM;
-		dof->m_angularMotorERP.m_floats[i]         = m_angularLimits[i].m_motorERP;
-		dof->m_angularMotorCFM.m_floats[i]         = m_angularLimits[i].m_motorCFM;
-		dof->m_angularTargetVelocity.m_floats[i]   = m_angularLimits[i].m_targetVelocity;
-		dof->m_angularMaxMotorForce.m_floats[i]    = m_angularLimits[i].m_maxMotorForce;
-		dof->m_angularServoTarget.m_floats[i]      = m_angularLimits[i].m_servoTarget;
-		dof->m_angularSpringStiffness.m_floats[i]  = m_angularLimits[i].m_springStiffness;
-		dof->m_angularSpringDamping.m_floats[i]    = m_angularLimits[i].m_springDamping;
+		dof->m_angularLowerLimit.m_floats[i] = m_angularLimits[i].m_loLimit;
+		dof->m_angularUpperLimit.m_floats[i] = m_angularLimits[i].m_hiLimit;
+		dof->m_angularBounce.m_floats[i] = m_angularLimits[i].m_bounce;
+		dof->m_angularStopERP.m_floats[i] = m_angularLimits[i].m_stopERP;
+		dof->m_angularStopCFM.m_floats[i] = m_angularLimits[i].m_stopCFM;
+		dof->m_angularMotorERP.m_floats[i] = m_angularLimits[i].m_motorERP;
+		dof->m_angularMotorCFM.m_floats[i] = m_angularLimits[i].m_motorCFM;
+		dof->m_angularTargetVelocity.m_floats[i] = m_angularLimits[i].m_targetVelocity;
+		dof->m_angularMaxMotorForce.m_floats[i] = m_angularLimits[i].m_maxMotorForce;
+		dof->m_angularServoTarget.m_floats[i] = m_angularLimits[i].m_servoTarget;
+		dof->m_angularSpringStiffness.m_floats[i] = m_angularLimits[i].m_springStiffness;
+		dof->m_angularSpringDamping.m_floats[i] = m_angularLimits[i].m_springDamping;
 		dof->m_angularEquilibriumPoint.m_floats[i] = m_angularLimits[i].m_equilibriumPoint;
 	}
-	dof->m_angularLowerLimit.m_floats[3]       = 0;
-	dof->m_angularUpperLimit.m_floats[3]       = 0;
-	dof->m_angularBounce.m_floats[3]           = 0;
-	dof->m_angularStopERP.m_floats[3]          = 0;
-	dof->m_angularStopCFM.m_floats[3]          = 0;
-	dof->m_angularMotorERP.m_floats[3]         = 0;
-	dof->m_angularMotorCFM.m_floats[3]         = 0;
-	dof->m_angularTargetVelocity.m_floats[3]   = 0;
-	dof->m_angularMaxMotorForce.m_floats[3]    = 0;
-	dof->m_angularServoTarget.m_floats[3]      = 0;
-	dof->m_angularSpringStiffness.m_floats[3]  = 0;
-	dof->m_angularSpringDamping.m_floats[3]    = 0;
+	dof->m_angularLowerLimit.m_floats[3] = 0;
+	dof->m_angularUpperLimit.m_floats[3] = 0;
+	dof->m_angularBounce.m_floats[3] = 0;
+	dof->m_angularStopERP.m_floats[3] = 0;
+	dof->m_angularStopCFM.m_floats[3] = 0;
+	dof->m_angularMotorERP.m_floats[3] = 0;
+	dof->m_angularMotorCFM.m_floats[3] = 0;
+	dof->m_angularTargetVelocity.m_floats[3] = 0;
+	dof->m_angularMaxMotorForce.m_floats[3] = 0;
+	dof->m_angularServoTarget.m_floats[3] = 0;
+	dof->m_angularSpringStiffness.m_floats[3] = 0;
+	dof->m_angularSpringDamping.m_floats[3] = 0;
 	dof->m_angularEquilibriumPoint.m_floats[3] = 0;
-	for (i=0;i<4;i++)
+	for (i = 0; i < 4; i++)
 	{
-		dof->m_angularEnableMotor[i]            = i < 3 ? ( m_angularLimits[i].m_enableMotor ? 1 : 0 ) : 0;
-		dof->m_angularServoMotor[i]             = i < 3 ? ( m_angularLimits[i].m_servoMotor ? 1 : 0 ) : 0;
-		dof->m_angularEnableSpring[i]           = i < 3 ? ( m_angularLimits[i].m_enableSpring ? 1 : 0 ) : 0;
-		dof->m_angularSpringStiffnessLimited[i] = i < 3 ? ( m_angularLimits[i].m_springStiffnessLimited ? 1 : 0 ) : 0;
-		dof->m_angularSpringDampingLimited[i]   = i < 3 ? ( m_angularLimits[i].m_springDampingLimited ? 1 : 0 ) : 0;
+		dof->m_angularEnableMotor[i] = i < 3 ? (m_angularLimits[i].m_enableMotor ? 1 : 0) : 0;
+		dof->m_angularServoMotor[i] = i < 3 ? (m_angularLimits[i].m_servoMotor ? 1 : 0) : 0;
+		dof->m_angularEnableSpring[i] = i < 3 ? (m_angularLimits[i].m_enableSpring ? 1 : 0) : 0;
+		dof->m_angularSpringStiffnessLimited[i] = i < 3 ? (m_angularLimits[i].m_springStiffnessLimited ? 1 : 0) : 0;
+		dof->m_angularSpringDampingLimited[i] = i < 3 ? (m_angularLimits[i].m_springDampingLimited ? 1 : 0) : 0;
 	}
 
-	m_linearLimits.m_lowerLimit.serialize( dof->m_linearLowerLimit );
-	m_linearLimits.m_upperLimit.serialize( dof->m_linearUpperLimit );
-	m_linearLimits.m_bounce.serialize( dof->m_linearBounce );
-	m_linearLimits.m_stopERP.serialize( dof->m_linearStopERP );
-	m_linearLimits.m_stopCFM.serialize( dof->m_linearStopCFM );
-	m_linearLimits.m_motorERP.serialize( dof->m_linearMotorERP );
-	m_linearLimits.m_motorCFM.serialize( dof->m_linearMotorCFM );
-	m_linearLimits.m_targetVelocity.serialize( dof->m_linearTargetVelocity );
-	m_linearLimits.m_maxMotorForce.serialize( dof->m_linearMaxMotorForce );
-	m_linearLimits.m_servoTarget.serialize( dof->m_linearServoTarget );
-	m_linearLimits.m_springStiffness.serialize( dof->m_linearSpringStiffness );
-	m_linearLimits.m_springDamping.serialize( dof->m_linearSpringDamping );
-	m_linearLimits.m_equilibriumPoint.serialize( dof->m_linearEquilibriumPoint );
-	for (i=0;i<4;i++)
+	m_linearLimits.m_lowerLimit.serialize(dof->m_linearLowerLimit);
+	m_linearLimits.m_upperLimit.serialize(dof->m_linearUpperLimit);
+	m_linearLimits.m_bounce.serialize(dof->m_linearBounce);
+	m_linearLimits.m_stopERP.serialize(dof->m_linearStopERP);
+	m_linearLimits.m_stopCFM.serialize(dof->m_linearStopCFM);
+	m_linearLimits.m_motorERP.serialize(dof->m_linearMotorERP);
+	m_linearLimits.m_motorCFM.serialize(dof->m_linearMotorCFM);
+	m_linearLimits.m_targetVelocity.serialize(dof->m_linearTargetVelocity);
+	m_linearLimits.m_maxMotorForce.serialize(dof->m_linearMaxMotorForce);
+	m_linearLimits.m_servoTarget.serialize(dof->m_linearServoTarget);
+	m_linearLimits.m_springStiffness.serialize(dof->m_linearSpringStiffness);
+	m_linearLimits.m_springDamping.serialize(dof->m_linearSpringDamping);
+	m_linearLimits.m_equilibriumPoint.serialize(dof->m_linearEquilibriumPoint);
+	for (i = 0; i < 4; i++)
 	{
-		dof->m_linearEnableMotor[i]            = i < 3 ? ( m_linearLimits.m_enableMotor[i] ? 1 : 0 ) : 0;
-		dof->m_linearServoMotor[i]             = i < 3 ? ( m_linearLimits.m_servoMotor[i] ? 1 : 0 ) : 0;
-		dof->m_linearEnableSpring[i]           = i < 3 ? ( m_linearLimits.m_enableSpring[i] ? 1 : 0 ) : 0;
-		dof->m_linearSpringStiffnessLimited[i] = i < 3 ? ( m_linearLimits.m_springStiffnessLimited[i] ? 1 : 0 ) : 0;
-		dof->m_linearSpringDampingLimited[i]   = i < 3 ? ( m_linearLimits.m_springDampingLimited[i] ? 1 : 0 ) : 0;
+		dof->m_linearEnableMotor[i] = i < 3 ? (m_linearLimits.m_enableMotor[i] ? 1 : 0) : 0;
+		dof->m_linearServoMotor[i] = i < 3 ? (m_linearLimits.m_servoMotor[i] ? 1 : 0) : 0;
+		dof->m_linearEnableSpring[i] = i < 3 ? (m_linearLimits.m_enableSpring[i] ? 1 : 0) : 0;
+		dof->m_linearSpringStiffnessLimited[i] = i < 3 ? (m_linearLimits.m_springStiffnessLimited[i] ? 1 : 0) : 0;
+		dof->m_linearSpringDampingLimited[i] = i < 3 ? (m_linearLimits.m_springDampingLimited[i] ? 1 : 0) : 0;
 	}
 
 	dof->m_rotateOrder = m_rotateOrder;
@@ -672,8 +663,4 @@ SIMD_FORCE_INLINE const char* btGeneric6DofSpring2Constraint::serialize(void* da
 	return btGeneric6DofSpring2ConstraintDataName;
 }
 
-
-
-
-
-#endif //BT_GENERIC_6DOF_CONSTRAINT_H
+#endif  //BT_GENERIC_6DOF_CONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofSpringConstraint.cpp b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofSpringConstraint.cpp
index 3f875989ea5..8baf52bcd19 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofSpringConstraint.cpp
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofSpringConstraint.cpp
@@ -17,26 +17,23 @@ subject to the following restrictions:
 #include "BulletDynamics/Dynamics/btRigidBody.h"
 #include "LinearMath/btTransformUtil.h"
 
-
-btGeneric6DofSpringConstraint::btGeneric6DofSpringConstraint(btRigidBody& rbA, btRigidBody& rbB, const btTransform& frameInA, const btTransform& frameInB ,bool useLinearReferenceFrameA)
+btGeneric6DofSpringConstraint::btGeneric6DofSpringConstraint(btRigidBody& rbA, btRigidBody& rbB, const btTransform& frameInA, const btTransform& frameInB, bool useLinearReferenceFrameA)
 	: btGeneric6DofConstraint(rbA, rbB, frameInA, frameInB, useLinearReferenceFrameA)
 {
-    init();
+	init();
 }
 
-
 btGeneric6DofSpringConstraint::btGeneric6DofSpringConstraint(btRigidBody& rbB, const btTransform& frameInB, bool useLinearReferenceFrameB)
-        : btGeneric6DofConstraint(rbB, frameInB, useLinearReferenceFrameB)
+	: btGeneric6DofConstraint(rbB, frameInB, useLinearReferenceFrameB)
 {
-    init();
+	init();
 }
 
-
 void btGeneric6DofSpringConstraint::init()
 {
 	m_objectType = D6_SPRING_CONSTRAINT_TYPE;
 
-	for(int i = 0; i < 6; i++)
+	for (int i = 0; i < 6; i++)
 	{
 		m_springEnabled[i] = false;
 		m_equilibriumPoint[i] = btScalar(0.f);
@@ -45,12 +42,11 @@ void btGeneric6DofSpringConstraint::init()
 	}
 }
 
-
 void btGeneric6DofSpringConstraint::enableSpring(int index, bool onOff)
 {
 	btAssert((index >= 0) && (index < 6));
 	m_springEnabled[index] = onOff;
-	if(index < 3)
+	if (index < 3)
 	{
 		m_linearLimits.m_enableMotor[index] = onOff;
 	}
@@ -60,44 +56,38 @@ void btGeneric6DofSpringConstraint::enableSpring(int index, bool onOff)
 	}
 }
 
-
-
 void btGeneric6DofSpringConstraint::setStiffness(int index, btScalar stiffness)
 {
 	btAssert((index >= 0) && (index < 6));
 	m_springStiffness[index] = stiffness;
 }
 
-
 void btGeneric6DofSpringConstraint::setDamping(int index, btScalar damping)
 {
 	btAssert((index >= 0) && (index < 6));
 	m_springDamping[index] = damping;
 }
 
-
 void btGeneric6DofSpringConstraint::setEquilibriumPoint()
 {
 	calculateTransforms();
 	int i;
 
-	for( i = 0; i < 3; i++)
+	for (i = 0; i < 3; i++)
 	{
 		m_equilibriumPoint[i] = m_calculatedLinearDiff[i];
 	}
-	for(i = 0; i < 3; i++)
+	for (i = 0; i < 3; i++)
 	{
 		m_equilibriumPoint[i + 3] = m_calculatedAxisAngleDiff[i];
 	}
 }
 
-
-
 void btGeneric6DofSpringConstraint::setEquilibriumPoint(int index)
 {
 	btAssert((index >= 0) && (index < 6));
 	calculateTransforms();
-	if(index < 3)
+	if (index < 3)
 	{
 		m_equilibriumPoint[index] = m_calculatedLinearDiff[index];
 	}
@@ -113,15 +103,14 @@ void btGeneric6DofSpringConstraint::setEquilibriumPoint(int index, btScalar val)
 	m_equilibriumPoint[index] = val;
 }
 
-
 void btGeneric6DofSpringConstraint::internalUpdateSprings(btConstraintInfo2* info)
 {
 	// it is assumed that calculateTransforms() have been called before this call
 	int i;
 	//btVector3 relVel = m_rbB.getLinearVelocity() - m_rbA.getLinearVelocity();
-	for(i = 0; i < 3; i++)
+	for (i = 0; i < 3; i++)
 	{
-		if(m_springEnabled[i])
+		if (m_springEnabled[i])
 		{
 			// get current position of constraint
 			btScalar currPos = m_calculatedLinearDiff[i];
@@ -130,28 +119,27 @@ void btGeneric6DofSpringConstraint::internalUpdateSprings(btConstraintInfo2* inf
 			// spring force is (delta * m_stiffness) according to Hooke's Law
 			btScalar force = delta * m_springStiffness[i];
 			btScalar velFactor = info->fps * m_springDamping[i] / btScalar(info->m_numIterations);
-			m_linearLimits.m_targetVelocity[i] =  velFactor * force;
-			m_linearLimits.m_maxMotorForce[i] =  btFabs(force);
+			m_linearLimits.m_targetVelocity[i] = velFactor * force;
+			m_linearLimits.m_maxMotorForce[i] = btFabs(force);
 		}
 	}
-	for(i = 0; i < 3; i++)
+	for (i = 0; i < 3; i++)
 	{
-		if(m_springEnabled[i + 3])
+		if (m_springEnabled[i + 3])
 		{
 			// get current position of constraint
 			btScalar currPos = m_calculatedAxisAngleDiff[i];
 			// calculate difference
-			btScalar delta = currPos - m_equilibriumPoint[i+3];
+			btScalar delta = currPos - m_equilibriumPoint[i + 3];
 			// spring force is (-delta * m_stiffness) according to Hooke's Law
-			btScalar force = -delta * m_springStiffness[i+3];
-			btScalar velFactor = info->fps * m_springDamping[i+3] / btScalar(info->m_numIterations);
+			btScalar force = -delta * m_springStiffness[i + 3];
+			btScalar velFactor = info->fps * m_springDamping[i + 3] / btScalar(info->m_numIterations);
 			m_angularLimits[i].m_targetVelocity = velFactor * force;
 			m_angularLimits[i].m_maxMotorForce = btFabs(force);
 		}
 	}
 }
 
-
 void btGeneric6DofSpringConstraint::getInfo2(btConstraintInfo2* info)
 {
 	// this will be called by constraint solver at the constraint setup stage
@@ -161,25 +149,21 @@ void btGeneric6DofSpringConstraint::getInfo2(btConstraintInfo2* info)
 	btGeneric6DofConstraint::getInfo2(info);
 }
 
-
-void btGeneric6DofSpringConstraint::setAxis(const btVector3& axis1,const btVector3& axis2)
+void btGeneric6DofSpringConstraint::setAxis(const btVector3& axis1, const btVector3& axis2)
 {
 	btVector3 zAxis = axis1.normalized();
 	btVector3 yAxis = axis2.normalized();
-	btVector3 xAxis = yAxis.cross(zAxis); // we want right coordinate system
+	btVector3 xAxis = yAxis.cross(zAxis);  // we want right coordinate system
 
 	btTransform frameInW;
 	frameInW.setIdentity();
-	frameInW.getBasis().setValue(	xAxis[0], yAxis[0], zAxis[0],	
-                                xAxis[1], yAxis[1], zAxis[1],
-                                xAxis[2], yAxis[2], zAxis[2]);
+	frameInW.getBasis().setValue(xAxis[0], yAxis[0], zAxis[0],
+								 xAxis[1], yAxis[1], zAxis[1],
+								 xAxis[2], yAxis[2], zAxis[2]);
 
 	// now get constraint frame in local coordinate systems
 	m_frameInA = m_rbA.getCenterOfMassTransform().inverse() * frameInW;
 	m_frameInB = m_rbB.getCenterOfMassTransform().inverse() * frameInW;
 
-  calculateTransforms();
+	calculateTransforms();
 }
-
-
-
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofSpringConstraint.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofSpringConstraint.h
index dac59c6889d..02b9d4d05d1 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofSpringConstraint.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btGeneric6DofSpringConstraint.h
@@ -16,20 +16,17 @@ subject to the following restrictions:
 #ifndef BT_GENERIC_6DOF_SPRING_CONSTRAINT_H
 #define BT_GENERIC_6DOF_SPRING_CONSTRAINT_H
 
-
 #include "LinearMath/btVector3.h"
 #include "btTypedConstraint.h"
 #include "btGeneric6DofConstraint.h"
 
 #ifdef BT_USE_DOUBLE_PRECISION
-#define btGeneric6DofSpringConstraintData2		btGeneric6DofSpringConstraintDoubleData2
-#define btGeneric6DofSpringConstraintDataName	"btGeneric6DofSpringConstraintDoubleData2"
+#define btGeneric6DofSpringConstraintData2 btGeneric6DofSpringConstraintDoubleData2
+#define btGeneric6DofSpringConstraintDataName "btGeneric6DofSpringConstraintDoubleData2"
 #else
-#define btGeneric6DofSpringConstraintData2		btGeneric6DofSpringConstraintData
-#define btGeneric6DofSpringConstraintDataName	"btGeneric6DofSpringConstraintData"
-#endif //BT_USE_DOUBLE_PRECISION
-
-
+#define btGeneric6DofSpringConstraintData2 btGeneric6DofSpringConstraintData
+#define btGeneric6DofSpringConstraintDataName "btGeneric6DofSpringConstraintData"
+#endif  //BT_USE_DOUBLE_PRECISION
 
 /// Generic 6 DOF constraint that allows to set spring motors to any translational and rotational DOF
 
@@ -41,101 +38,98 @@ subject to the following restrictions:
 /// 4 : rotation Y (2nd Euler rotational around new position of Y axis, range [-PI/2+epsilon, PI/2-epsilon] )
 /// 5 : rotation Z (1st Euler rotational around Z axis, range [-PI+epsilon, PI-epsilon] )
 
-ATTRIBUTE_ALIGNED16(class) btGeneric6DofSpringConstraint : public btGeneric6DofConstraint
+ATTRIBUTE_ALIGNED16(class)
+btGeneric6DofSpringConstraint : public btGeneric6DofConstraint
 {
 protected:
-	bool		m_springEnabled[6];
-	btScalar	m_equilibriumPoint[6];
-	btScalar	m_springStiffness[6];
-	btScalar	m_springDamping[6]; // between 0 and 1 (1 == no damping)
+	bool m_springEnabled[6];
+	btScalar m_equilibriumPoint[6];
+	btScalar m_springStiffness[6];
+	btScalar m_springDamping[6];  // between 0 and 1 (1 == no damping)
 	void init();
-	void internalUpdateSprings(btConstraintInfo2* info);
-public: 
-	
+	void internalUpdateSprings(btConstraintInfo2 * info);
+
+public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
-    btGeneric6DofSpringConstraint(btRigidBody& rbA, btRigidBody& rbB, const btTransform& frameInA, const btTransform& frameInB ,bool useLinearReferenceFrameA);
-    btGeneric6DofSpringConstraint(btRigidBody& rbB, const btTransform& frameInB, bool useLinearReferenceFrameB);
+
+	btGeneric6DofSpringConstraint(btRigidBody & rbA, btRigidBody & rbB, const btTransform& frameInA, const btTransform& frameInB, bool useLinearReferenceFrameA);
+	btGeneric6DofSpringConstraint(btRigidBody & rbB, const btTransform& frameInB, bool useLinearReferenceFrameB);
 	void enableSpring(int index, bool onOff);
 	void setStiffness(int index, btScalar stiffness);
 	void setDamping(int index, btScalar damping);
-	void setEquilibriumPoint(); // set the current constraint position/orientation as an equilibrium point for all DOF
+	void setEquilibriumPoint();           // set the current constraint position/orientation as an equilibrium point for all DOF
 	void setEquilibriumPoint(int index);  // set the current constraint position/orientation as an equilibrium point for given DOF
 	void setEquilibriumPoint(int index, btScalar val);
 
 	bool isSpringEnabled(int index) const
 	{
-	    return m_springEnabled[index];
+		return m_springEnabled[index];
 	}
 
 	btScalar getStiffness(int index) const
 	{
-	    return m_springStiffness[index];
+		return m_springStiffness[index];
 	}
 
 	btScalar getDamping(int index) const
 	{
-	    return m_springDamping[index];
+		return m_springDamping[index];
 	}
 
 	btScalar getEquilibriumPoint(int index) const
 	{
-	    return m_equilibriumPoint[index];
+		return m_equilibriumPoint[index];
 	}
 
-	virtual void setAxis( const btVector3& axis1, const btVector3& axis2);
+	virtual void setAxis(const btVector3& axis1, const btVector3& axis2);
 
-	virtual void getInfo2 (btConstraintInfo2* info);
+	virtual void getInfo2(btConstraintInfo2 * info);
 
-	virtual	int	calculateSerializeBufferSize() const;
+	virtual int calculateSerializeBufferSize() const;
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
-
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 };
 
-
 struct btGeneric6DofSpringConstraintData
 {
-	btGeneric6DofConstraintData	m_6dofData;
-	
-	int			m_springEnabled[6];
-	float		m_equilibriumPoint[6];
-	float		m_springStiffness[6];
-	float		m_springDamping[6];
+	btGeneric6DofConstraintData m_6dofData;
+
+	int m_springEnabled[6];
+	float m_equilibriumPoint[6];
+	float m_springStiffness[6];
+	float m_springDamping[6];
 };
 
 struct btGeneric6DofSpringConstraintDoubleData2
 {
-	btGeneric6DofConstraintDoubleData2	m_6dofData;
-	
-	int			m_springEnabled[6];
-	double		m_equilibriumPoint[6];
-	double		m_springStiffness[6];
-	double		m_springDamping[6];
+	btGeneric6DofConstraintDoubleData2 m_6dofData;
+
+	int m_springEnabled[6];
+	double m_equilibriumPoint[6];
+	double m_springStiffness[6];
+	double m_springDamping[6];
 };
 
-
-SIMD_FORCE_INLINE	int	btGeneric6DofSpringConstraint::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btGeneric6DofSpringConstraint::calculateSerializeBufferSize() const
 {
 	return sizeof(btGeneric6DofSpringConstraintData2);
 }
 
-	///fills the dataBuffer and returns the struct name (and 0 on failure)
-SIMD_FORCE_INLINE	const char*	btGeneric6DofSpringConstraint::serialize(void* dataBuffer, btSerializer* serializer) const
+///fills the dataBuffer and returns the struct name (and 0 on failure)
+SIMD_FORCE_INLINE const char* btGeneric6DofSpringConstraint::serialize(void* dataBuffer, btSerializer* serializer) const
 {
 	btGeneric6DofSpringConstraintData2* dof = (btGeneric6DofSpringConstraintData2*)dataBuffer;
-	btGeneric6DofConstraint::serialize(&dof->m_6dofData,serializer);
+	btGeneric6DofConstraint::serialize(&dof->m_6dofData, serializer);
 
 	int i;
-	for (i=0;i<6;i++)
+	for (i = 0; i < 6; i++)
 	{
 		dof->m_equilibriumPoint[i] = m_equilibriumPoint[i];
 		dof->m_springDamping[i] = m_springDamping[i];
-		dof->m_springEnabled[i] = m_springEnabled[i]? 1 : 0;
+		dof->m_springEnabled[i] = m_springEnabled[i] ? 1 : 0;
 		dof->m_springStiffness[i] = m_springStiffness[i];
 	}
 	return btGeneric6DofSpringConstraintDataName;
 }
 
-#endif // BT_GENERIC_6DOF_SPRING_CONSTRAINT_H
-
+#endif  // BT_GENERIC_6DOF_SPRING_CONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btHinge2Constraint.cpp b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btHinge2Constraint.cpp
index 4be2aabe4d3..6507e1a0a74 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btHinge2Constraint.cpp
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btHinge2Constraint.cpp
@@ -13,54 +13,49 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #include "btHinge2Constraint.h"
 #include "BulletDynamics/Dynamics/btRigidBody.h"
 #include "LinearMath/btTransformUtil.h"
 
-
-
 // constructor
 // anchor, axis1 and axis2 are in world coordinate system
 // axis1 must be orthogonal to axis2
 btHinge2Constraint::btHinge2Constraint(btRigidBody& rbA, btRigidBody& rbB, btVector3& anchor, btVector3& axis1, btVector3& axis2)
-: btGeneric6DofSpring2Constraint(rbA, rbB, btTransform::getIdentity(), btTransform::getIdentity(),RO_XYZ),
- m_anchor(anchor),
- m_axis1(axis1),
- m_axis2(axis2)
+	: btGeneric6DofSpring2Constraint(rbA, rbB, btTransform::getIdentity(), btTransform::getIdentity(), RO_XYZ),
+	  m_anchor(anchor),
+	  m_axis1(axis1),
+	  m_axis2(axis2)
 {
 	// build frame basis
 	// 6DOF constraint uses Euler angles and to define limits
 	// it is assumed that rotational order is :
 	// Z - first, allowed limits are (-PI,PI);
-	// new position of Y - second (allowed limits are (-PI/2 + epsilon, PI/2 - epsilon), where epsilon is a small positive number 
+	// new position of Y - second (allowed limits are (-PI/2 + epsilon, PI/2 - epsilon), where epsilon is a small positive number
 	// used to prevent constraint from instability on poles;
 	// new position of X, allowed limits are (-PI,PI);
 	// So to simulate ODE Universal joint we should use parent axis as Z, child axis as Y and limit all other DOFs
 	// Build the frame in world coordinate system first
 	btVector3 zAxis = axis1.normalize();
 	btVector3 xAxis = axis2.normalize();
-	btVector3 yAxis = zAxis.cross(xAxis); // we want right coordinate system
+	btVector3 yAxis = zAxis.cross(xAxis);  // we want right coordinate system
 	btTransform frameInW;
 	frameInW.setIdentity();
-	frameInW.getBasis().setValue(	xAxis[0], yAxis[0], zAxis[0],	
-									xAxis[1], yAxis[1], zAxis[1],
-									xAxis[2], yAxis[2], zAxis[2]);
+	frameInW.getBasis().setValue(xAxis[0], yAxis[0], zAxis[0],
+								 xAxis[1], yAxis[1], zAxis[1],
+								 xAxis[2], yAxis[2], zAxis[2]);
 	frameInW.setOrigin(anchor);
 	// now get constraint frame in local coordinate systems
 	m_frameInA = rbA.getCenterOfMassTransform().inverse() * frameInW;
 	m_frameInB = rbB.getCenterOfMassTransform().inverse() * frameInW;
 	// sei limits
 	setLinearLowerLimit(btVector3(0.f, 0.f, -1.f));
-	setLinearUpperLimit(btVector3(0.f, 0.f,  1.f));
+	setLinearUpperLimit(btVector3(0.f, 0.f, 1.f));
 	// like front wheels of a car
-	setAngularLowerLimit(btVector3(1.f,  0.f, -SIMD_HALF_PI * 0.5f)); 
-	setAngularUpperLimit(btVector3(-1.f, 0.f,  SIMD_HALF_PI * 0.5f));
+	setAngularLowerLimit(btVector3(1.f, 0.f, -SIMD_HALF_PI * 0.5f));
+	setAngularUpperLimit(btVector3(-1.f, 0.f, SIMD_HALF_PI * 0.5f));
 	// enable suspension
 	enableSpring(2, true);
 	setStiffness(2, SIMD_PI * SIMD_PI * 4.f);
 	setDamping(2, 0.01f);
 	setEquilibriumPoint();
 }
-
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btHinge2Constraint.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btHinge2Constraint.h
index 06a8e3ecd14..95f604a8906 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btHinge2Constraint.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btHinge2Constraint.h
@@ -16,32 +16,30 @@ subject to the following restrictions:
 #ifndef BT_HINGE2_CONSTRAINT_H
 #define BT_HINGE2_CONSTRAINT_H
 
-
-
 #include "LinearMath/btVector3.h"
 #include "btTypedConstraint.h"
 #include "btGeneric6DofSpring2Constraint.h"
 
-
-
 // Constraint similar to ODE Hinge2 Joint
 // has 3 degrees of frredom:
 // 2 rotational degrees of freedom, similar to Euler rotations around Z (axis 1) and X (axis 2)
 // 1 translational (along axis Z) with suspension spring
 
-ATTRIBUTE_ALIGNED16(class) btHinge2Constraint : public btGeneric6DofSpring2Constraint
+ATTRIBUTE_ALIGNED16(class)
+btHinge2Constraint : public btGeneric6DofSpring2Constraint
 {
 protected:
-	btVector3	m_anchor;
-	btVector3	m_axis1;
-	btVector3	m_axis2;
+	btVector3 m_anchor;
+	btVector3 m_axis1;
+	btVector3 m_axis2;
+
 public:
-		BT_DECLARE_ALIGNED_ALLOCATOR();
-		
+	BT_DECLARE_ALIGNED_ALLOCATOR();
+
 	// constructor
 	// anchor, axis1 and axis2 are in world coordinate system
 	// axis1 must be orthogonal to axis2
-    btHinge2Constraint(btRigidBody& rbA, btRigidBody& rbB, btVector3& anchor, btVector3& axis1, btVector3& axis2);
+	btHinge2Constraint(btRigidBody & rbA, btRigidBody & rbB, btVector3 & anchor, btVector3 & axis1, btVector3 & axis2);
 	// access
 	const btVector3& getAnchor() { return m_calculatedTransformA.getOrigin(); }
 	const btVector3& getAnchor2() { return m_calculatedTransformB.getOrigin(); }
@@ -51,10 +49,7 @@ public:
 	btScalar getAngle2() { return getAngle(0); }
 	// limits
 	void setUpperLimit(btScalar ang1max) { setAngularUpperLimit(btVector3(-1.f, 0.f, ang1max)); }
-	void setLowerLimit(btScalar ang1min) { setAngularLowerLimit(btVector3( 1.f, 0.f, ang1min)); }
+	void setLowerLimit(btScalar ang1min) { setAngularLowerLimit(btVector3(1.f, 0.f, ang1min)); }
 };
 
-
-
-#endif // BT_HINGE2_CONSTRAINT_H
-
+#endif  // BT_HINGE2_CONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btHingeConstraint.cpp b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btHingeConstraint.cpp
index 7e5e6f9e54c..aa6f69000d2 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btHingeConstraint.cpp
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btHingeConstraint.cpp
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btHingeConstraint.h"
 #include "BulletDynamics/Dynamics/btRigidBody.h"
 #include "LinearMath/btTransformUtil.h"
@@ -21,8 +20,6 @@ subject to the following restrictions:
 #include <new>
 #include "btSolverBody.h"
 
-
-
 //#define HINGE_USE_OBSOLETE_SOLVER false
 #define HINGE_USE_OBSOLETE_SOLVER false
 
@@ -30,59 +27,60 @@ subject to the following restrictions:
 
 #ifndef __SPU__
 
-
-
-
-
-btHingeConstraint::btHingeConstraint(btRigidBody& rbA,btRigidBody& rbB, const btVector3& pivotInA,const btVector3& pivotInB,
-									 const btVector3& axisInA,const btVector3& axisInB, bool useReferenceFrameA)
-									 :btTypedConstraint(HINGE_CONSTRAINT_TYPE, rbA,rbB),
+btHingeConstraint::btHingeConstraint(btRigidBody& rbA, btRigidBody& rbB, const btVector3& pivotInA, const btVector3& pivotInB,
+									 const btVector3& axisInA, const btVector3& axisInB, bool useReferenceFrameA)
+	: btTypedConstraint(HINGE_CONSTRAINT_TYPE, rbA, rbB),
 #ifdef _BT_USE_CENTER_LIMIT_
-									 m_limit(),
+	  m_limit(),
 #endif
-									 m_angularOnly(false),
-									 m_enableAngularMotor(false),
-									 m_useSolveConstraintObsolete(HINGE_USE_OBSOLETE_SOLVER),
-									 m_useOffsetForConstraintFrame(HINGE_USE_FRAME_OFFSET),
-									 m_useReferenceFrameA(useReferenceFrameA),
-									 m_flags(0),
-									 m_normalCFM(0),
-									 m_normalERP(0),
-									 m_stopCFM(0),
-									 m_stopERP(0)
+	  m_angularOnly(false),
+	  m_enableAngularMotor(false),
+	  m_useSolveConstraintObsolete(HINGE_USE_OBSOLETE_SOLVER),
+	  m_useOffsetForConstraintFrame(HINGE_USE_FRAME_OFFSET),
+	  m_useReferenceFrameA(useReferenceFrameA),
+	  m_flags(0),
+	  m_normalCFM(0),
+	  m_normalERP(0),
+	  m_stopCFM(0),
+	  m_stopERP(0)
 {
 	m_rbAFrame.getOrigin() = pivotInA;
-	
+
 	// since no frame is given, assume this to be zero angle and just pick rb transform axis
 	btVector3 rbAxisA1 = rbA.getCenterOfMassTransform().getBasis().getColumn(0);
 
 	btVector3 rbAxisA2;
 	btScalar projection = axisInA.dot(rbAxisA1);
-	if (projection >= 1.0f - SIMD_EPSILON) {
+	if (projection >= 1.0f - SIMD_EPSILON)
+	{
 		rbAxisA1 = -rbA.getCenterOfMassTransform().getBasis().getColumn(2);
 		rbAxisA2 = rbA.getCenterOfMassTransform().getBasis().getColumn(1);
-	} else if (projection <= -1.0f + SIMD_EPSILON) {
+	}
+	else if (projection <= -1.0f + SIMD_EPSILON)
+	{
 		rbAxisA1 = rbA.getCenterOfMassTransform().getBasis().getColumn(2);
-		rbAxisA2 = rbA.getCenterOfMassTransform().getBasis().getColumn(1);      
-	} else {
+		rbAxisA2 = rbA.getCenterOfMassTransform().getBasis().getColumn(1);
+	}
+	else
+	{
 		rbAxisA2 = axisInA.cross(rbAxisA1);
 		rbAxisA1 = rbAxisA2.cross(axisInA);
 	}
 
-	m_rbAFrame.getBasis().setValue( rbAxisA1.getX(),rbAxisA2.getX(),axisInA.getX(),
-									rbAxisA1.getY(),rbAxisA2.getY(),axisInA.getY(),
-									rbAxisA1.getZ(),rbAxisA2.getZ(),axisInA.getZ() );
+	m_rbAFrame.getBasis().setValue(rbAxisA1.getX(), rbAxisA2.getX(), axisInA.getX(),
+								   rbAxisA1.getY(), rbAxisA2.getY(), axisInA.getY(),
+								   rbAxisA1.getZ(), rbAxisA2.getZ(), axisInA.getZ());
+
+	btQuaternion rotationArc = shortestArcQuat(axisInA, axisInB);
+	btVector3 rbAxisB1 = quatRotate(rotationArc, rbAxisA1);
+	btVector3 rbAxisB2 = axisInB.cross(rbAxisB1);
 
-	btQuaternion rotationArc = shortestArcQuat(axisInA,axisInB);
-	btVector3 rbAxisB1 =  quatRotate(rotationArc,rbAxisA1);
-	btVector3 rbAxisB2 =  axisInB.cross(rbAxisB1);	
-	
 	m_rbBFrame.getOrigin() = pivotInB;
-	m_rbBFrame.getBasis().setValue( rbAxisB1.getX(),rbAxisB2.getX(),axisInB.getX(),
-									rbAxisB1.getY(),rbAxisB2.getY(),axisInB.getY(),
-									rbAxisB1.getZ(),rbAxisB2.getZ(),axisInB.getZ() );
-	
-#ifndef	_BT_USE_CENTER_LIMIT_
+	m_rbBFrame.getBasis().setValue(rbAxisB1.getX(), rbAxisB2.getX(), axisInB.getX(),
+								   rbAxisB1.getY(), rbAxisB2.getY(), axisInB.getY(),
+								   rbAxisB1.getZ(), rbAxisB2.getZ(), axisInB.getZ());
+
+#ifndef _BT_USE_CENTER_LIMIT_
 	//start with free
 	m_lowerLimit = btScalar(1.0f);
 	m_upperLimit = btScalar(-1.0f);
@@ -94,47 +92,44 @@ btHingeConstraint::btHingeConstraint(btRigidBody& rbA,btRigidBody& rbB, const bt
 	m_referenceSign = m_useReferenceFrameA ? btScalar(-1.f) : btScalar(1.f);
 }
 
-
-
-btHingeConstraint::btHingeConstraint(btRigidBody& rbA,const btVector3& pivotInA,const btVector3& axisInA, bool useReferenceFrameA)
-:btTypedConstraint(HINGE_CONSTRAINT_TYPE, rbA),
+btHingeConstraint::btHingeConstraint(btRigidBody& rbA, const btVector3& pivotInA, const btVector3& axisInA, bool useReferenceFrameA)
+	: btTypedConstraint(HINGE_CONSTRAINT_TYPE, rbA),
 #ifdef _BT_USE_CENTER_LIMIT_
-m_limit(),
+	  m_limit(),
 #endif
-m_angularOnly(false), m_enableAngularMotor(false), 
-m_useSolveConstraintObsolete(HINGE_USE_OBSOLETE_SOLVER),
-m_useOffsetForConstraintFrame(HINGE_USE_FRAME_OFFSET),
-m_useReferenceFrameA(useReferenceFrameA),
-m_flags(0),
-m_normalCFM(0),
-m_normalERP(0),
-m_stopCFM(0),
-m_stopERP(0)
+	  m_angularOnly(false),
+	  m_enableAngularMotor(false),
+	  m_useSolveConstraintObsolete(HINGE_USE_OBSOLETE_SOLVER),
+	  m_useOffsetForConstraintFrame(HINGE_USE_FRAME_OFFSET),
+	  m_useReferenceFrameA(useReferenceFrameA),
+	  m_flags(0),
+	  m_normalCFM(0),
+	  m_normalERP(0),
+	  m_stopCFM(0),
+	  m_stopERP(0)
 {
-
 	// since no frame is given, assume this to be zero angle and just pick rb transform axis
 	// fixed axis in worldspace
 	btVector3 rbAxisA1, rbAxisA2;
 	btPlaneSpace1(axisInA, rbAxisA1, rbAxisA2);
 
 	m_rbAFrame.getOrigin() = pivotInA;
-	m_rbAFrame.getBasis().setValue( rbAxisA1.getX(),rbAxisA2.getX(),axisInA.getX(),
-									rbAxisA1.getY(),rbAxisA2.getY(),axisInA.getY(),
-									rbAxisA1.getZ(),rbAxisA2.getZ(),axisInA.getZ() );
+	m_rbAFrame.getBasis().setValue(rbAxisA1.getX(), rbAxisA2.getX(), axisInA.getX(),
+								   rbAxisA1.getY(), rbAxisA2.getY(), axisInA.getY(),
+								   rbAxisA1.getZ(), rbAxisA2.getZ(), axisInA.getZ());
 
 	btVector3 axisInB = rbA.getCenterOfMassTransform().getBasis() * axisInA;
 
-	btQuaternion rotationArc = shortestArcQuat(axisInA,axisInB);
-	btVector3 rbAxisB1 =  quatRotate(rotationArc,rbAxisA1);
+	btQuaternion rotationArc = shortestArcQuat(axisInA, axisInB);
+	btVector3 rbAxisB1 = quatRotate(rotationArc, rbAxisA1);
 	btVector3 rbAxisB2 = axisInB.cross(rbAxisB1);
 
-
 	m_rbBFrame.getOrigin() = rbA.getCenterOfMassTransform()(pivotInA);
-	m_rbBFrame.getBasis().setValue( rbAxisB1.getX(),rbAxisB2.getX(),axisInB.getX(),
-									rbAxisB1.getY(),rbAxisB2.getY(),axisInB.getY(),
-									rbAxisB1.getZ(),rbAxisB2.getZ(),axisInB.getZ() );
-	
-#ifndef	_BT_USE_CENTER_LIMIT_
+	m_rbBFrame.getBasis().setValue(rbAxisB1.getX(), rbAxisB2.getX(), axisInB.getX(),
+								   rbAxisB1.getY(), rbAxisB2.getY(), axisInB.getY(),
+								   rbAxisB1.getZ(), rbAxisB2.getZ(), axisInB.getZ());
+
+#ifndef _BT_USE_CENTER_LIMIT_
 	//start with free
 	m_lowerLimit = btScalar(1.0f);
 	m_upperLimit = btScalar(-1.0f);
@@ -146,26 +141,24 @@ m_stopERP(0)
 	m_referenceSign = m_useReferenceFrameA ? btScalar(-1.f) : btScalar(1.f);
 }
 
-
-
-btHingeConstraint::btHingeConstraint(btRigidBody& rbA,btRigidBody& rbB, 
-								     const btTransform& rbAFrame, const btTransform& rbBFrame, bool useReferenceFrameA)
-:btTypedConstraint(HINGE_CONSTRAINT_TYPE, rbA,rbB),m_rbAFrame(rbAFrame),m_rbBFrame(rbBFrame),
+btHingeConstraint::btHingeConstraint(btRigidBody& rbA, btRigidBody& rbB,
+									 const btTransform& rbAFrame, const btTransform& rbBFrame, bool useReferenceFrameA)
+	: btTypedConstraint(HINGE_CONSTRAINT_TYPE, rbA, rbB), m_rbAFrame(rbAFrame), m_rbBFrame(rbBFrame),
 #ifdef _BT_USE_CENTER_LIMIT_
-m_limit(),
+	  m_limit(),
 #endif
-m_angularOnly(false),
-m_enableAngularMotor(false),
-m_useSolveConstraintObsolete(HINGE_USE_OBSOLETE_SOLVER),
-m_useOffsetForConstraintFrame(HINGE_USE_FRAME_OFFSET),
-m_useReferenceFrameA(useReferenceFrameA),
-m_flags(0),
-m_normalCFM(0),
-m_normalERP(0),
-m_stopCFM(0),
-m_stopERP(0)
+	  m_angularOnly(false),
+	  m_enableAngularMotor(false),
+	  m_useSolveConstraintObsolete(HINGE_USE_OBSOLETE_SOLVER),
+	  m_useOffsetForConstraintFrame(HINGE_USE_FRAME_OFFSET),
+	  m_useReferenceFrameA(useReferenceFrameA),
+	  m_flags(0),
+	  m_normalCFM(0),
+	  m_normalERP(0),
+	  m_stopCFM(0),
+	  m_stopERP(0)
 {
-#ifndef	_BT_USE_CENTER_LIMIT_
+#ifndef _BT_USE_CENTER_LIMIT_
 	//start with free
 	m_lowerLimit = btScalar(1.0f);
 	m_upperLimit = btScalar(-1.0f);
@@ -175,30 +168,28 @@ m_stopERP(0)
 	m_solveLimit = false;
 #endif
 	m_referenceSign = m_useReferenceFrameA ? btScalar(-1.f) : btScalar(1.f);
-}			
-
-
+}
 
 btHingeConstraint::btHingeConstraint(btRigidBody& rbA, const btTransform& rbAFrame, bool useReferenceFrameA)
-:btTypedConstraint(HINGE_CONSTRAINT_TYPE, rbA),m_rbAFrame(rbAFrame),m_rbBFrame(rbAFrame),
+	: btTypedConstraint(HINGE_CONSTRAINT_TYPE, rbA), m_rbAFrame(rbAFrame), m_rbBFrame(rbAFrame),
 #ifdef _BT_USE_CENTER_LIMIT_
-m_limit(),
+	  m_limit(),
 #endif
-m_angularOnly(false),
-m_enableAngularMotor(false),
-m_useSolveConstraintObsolete(HINGE_USE_OBSOLETE_SOLVER),
-m_useOffsetForConstraintFrame(HINGE_USE_FRAME_OFFSET),
-m_useReferenceFrameA(useReferenceFrameA),
-m_flags(0),
-m_normalCFM(0),
-m_normalERP(0),
-m_stopCFM(0),
-m_stopERP(0)
+	  m_angularOnly(false),
+	  m_enableAngularMotor(false),
+	  m_useSolveConstraintObsolete(HINGE_USE_OBSOLETE_SOLVER),
+	  m_useOffsetForConstraintFrame(HINGE_USE_FRAME_OFFSET),
+	  m_useReferenceFrameA(useReferenceFrameA),
+	  m_flags(0),
+	  m_normalCFM(0),
+	  m_normalERP(0),
+	  m_stopCFM(0),
+	  m_stopERP(0)
 {
 	///not providing rigidbody B means implicitly using worldspace for body B
 
 	m_rbBFrame.getOrigin() = m_rbA.getCenterOfMassTransform()(m_rbAFrame.getOrigin());
-#ifndef	_BT_USE_CENTER_LIMIT_
+#ifndef _BT_USE_CENTER_LIMIT_
 	//start with free
 	m_lowerLimit = btScalar(1.0f);
 	m_upperLimit = btScalar(-1.0f);
@@ -210,9 +201,7 @@ m_stopERP(0)
 	m_referenceSign = m_useReferenceFrameA ? btScalar(-1.f) : btScalar(1.f);
 }
 
-
-
-void	btHingeConstraint::buildJacobian()
+void btHingeConstraint::buildJacobian()
 {
 	if (m_useSolveConstraintObsolete)
 	{
@@ -221,8 +210,8 @@ void	btHingeConstraint::buildJacobian()
 
 		if (!m_angularOnly)
 		{
-			btVector3 pivotAInW = m_rbA.getCenterOfMassTransform()*m_rbAFrame.getOrigin();
-			btVector3 pivotBInW = m_rbB.getCenterOfMassTransform()*m_rbBFrame.getOrigin();
+			btVector3 pivotAInW = m_rbA.getCenterOfMassTransform() * m_rbAFrame.getOrigin();
+			btVector3 pivotBInW = m_rbB.getCenterOfMassTransform() * m_rbBFrame.getOrigin();
 			btVector3 relPos = pivotBInW - pivotAInW;
 
 			btVector3 normal[3];
@@ -232,23 +221,23 @@ void	btHingeConstraint::buildJacobian()
 			}
 			else
 			{
-				normal[0].setValue(btScalar(1.0),0,0);
+				normal[0].setValue(btScalar(1.0), 0, 0);
 			}
 
 			btPlaneSpace1(normal[0], normal[1], normal[2]);
 
-			for (int i=0;i<3;i++)
+			for (int i = 0; i < 3; i++)
 			{
 				new (&m_jac[i]) btJacobianEntry(
-				m_rbA.getCenterOfMassTransform().getBasis().transpose(),
-				m_rbB.getCenterOfMassTransform().getBasis().transpose(),
-				pivotAInW - m_rbA.getCenterOfMassPosition(),
-				pivotBInW - m_rbB.getCenterOfMassPosition(),
-				normal[i],
-				m_rbA.getInvInertiaDiagLocal(),
-				m_rbA.getInvMass(),
-				m_rbB.getInvInertiaDiagLocal(),
-				m_rbB.getInvMass());
+					m_rbA.getCenterOfMassTransform().getBasis().transpose(),
+					m_rbB.getCenterOfMassTransform().getBasis().transpose(),
+					pivotAInW - m_rbA.getCenterOfMassPosition(),
+					pivotBInW - m_rbB.getCenterOfMassPosition(),
+					normal[i],
+					m_rbA.getInvInertiaDiagLocal(),
+					m_rbA.getInvMass(),
+					m_rbB.getInvInertiaDiagLocal(),
+					m_rbB.getInvMass());
 			}
 		}
 
@@ -258,60 +247,55 @@ void	btHingeConstraint::buildJacobian()
 		//this is unused for now, it's a todo
 		btVector3 jointAxis0local;
 		btVector3 jointAxis1local;
-		
-		btPlaneSpace1(m_rbAFrame.getBasis().getColumn(2),jointAxis0local,jointAxis1local);
+
+		btPlaneSpace1(m_rbAFrame.getBasis().getColumn(2), jointAxis0local, jointAxis1local);
 
 		btVector3 jointAxis0 = getRigidBodyA().getCenterOfMassTransform().getBasis() * jointAxis0local;
 		btVector3 jointAxis1 = getRigidBodyA().getCenterOfMassTransform().getBasis() * jointAxis1local;
 		btVector3 hingeAxisWorld = getRigidBodyA().getCenterOfMassTransform().getBasis() * m_rbAFrame.getBasis().getColumn(2);
-			
-		new (&m_jacAng[0])	btJacobianEntry(jointAxis0,
-			m_rbA.getCenterOfMassTransform().getBasis().transpose(),
-			m_rbB.getCenterOfMassTransform().getBasis().transpose(),
-			m_rbA.getInvInertiaDiagLocal(),
-			m_rbB.getInvInertiaDiagLocal());
 
-		new (&m_jacAng[1])	btJacobianEntry(jointAxis1,
-			m_rbA.getCenterOfMassTransform().getBasis().transpose(),
-			m_rbB.getCenterOfMassTransform().getBasis().transpose(),
-			m_rbA.getInvInertiaDiagLocal(),
-			m_rbB.getInvInertiaDiagLocal());
+		new (&m_jacAng[0]) btJacobianEntry(jointAxis0,
+										   m_rbA.getCenterOfMassTransform().getBasis().transpose(),
+										   m_rbB.getCenterOfMassTransform().getBasis().transpose(),
+										   m_rbA.getInvInertiaDiagLocal(),
+										   m_rbB.getInvInertiaDiagLocal());
 
-		new (&m_jacAng[2])	btJacobianEntry(hingeAxisWorld,
-			m_rbA.getCenterOfMassTransform().getBasis().transpose(),
-			m_rbB.getCenterOfMassTransform().getBasis().transpose(),
-			m_rbA.getInvInertiaDiagLocal(),
-			m_rbB.getInvInertiaDiagLocal());
+		new (&m_jacAng[1]) btJacobianEntry(jointAxis1,
+										   m_rbA.getCenterOfMassTransform().getBasis().transpose(),
+										   m_rbB.getCenterOfMassTransform().getBasis().transpose(),
+										   m_rbA.getInvInertiaDiagLocal(),
+										   m_rbB.getInvInertiaDiagLocal());
 
-			// clear accumulator
-			m_accLimitImpulse = btScalar(0.);
+		new (&m_jacAng[2]) btJacobianEntry(hingeAxisWorld,
+										   m_rbA.getCenterOfMassTransform().getBasis().transpose(),
+										   m_rbB.getCenterOfMassTransform().getBasis().transpose(),
+										   m_rbA.getInvInertiaDiagLocal(),
+										   m_rbB.getInvInertiaDiagLocal());
 
-			// test angular limit
-			testLimit(m_rbA.getCenterOfMassTransform(),m_rbB.getCenterOfMassTransform());
+		// clear accumulator
+		m_accLimitImpulse = btScalar(0.);
+
+		// test angular limit
+		testLimit(m_rbA.getCenterOfMassTransform(), m_rbB.getCenterOfMassTransform());
 
 		//Compute K = J*W*J' for hinge axis
-		btVector3 axisA =  getRigidBodyA().getCenterOfMassTransform().getBasis() *  m_rbAFrame.getBasis().getColumn(2);
-		m_kHinge =   1.0f / (getRigidBodyA().computeAngularImpulseDenominator(axisA) +
-							 getRigidBodyB().computeAngularImpulseDenominator(axisA));
-
+		btVector3 axisA = getRigidBodyA().getCenterOfMassTransform().getBasis() * m_rbAFrame.getBasis().getColumn(2);
+		m_kHinge = 1.0f / (getRigidBodyA().computeAngularImpulseDenominator(axisA) +
+						   getRigidBodyB().computeAngularImpulseDenominator(axisA));
 	}
 }
 
-
-#endif //__SPU__
-
+#endif  //__SPU__
 
 static inline btScalar btNormalizeAnglePositive(btScalar angle)
 {
-  return btFmod(btFmod(angle, btScalar(2.0*SIMD_PI)) + btScalar(2.0*SIMD_PI), btScalar(2.0*SIMD_PI));
+	return btFmod(btFmod(angle, btScalar(2.0 * SIMD_PI)) + btScalar(2.0 * SIMD_PI), btScalar(2.0 * SIMD_PI));
 }
 
-
-
 static btScalar btShortestAngularDistance(btScalar accAngle, btScalar curAngle)
 {
 	btScalar result = btNormalizeAngle(btNormalizeAnglePositive(btNormalizeAnglePositive(curAngle) -
-	btNormalizeAnglePositive(accAngle)));
+																btNormalizeAnglePositive(accAngle)));
 	return result;
 }
 
@@ -320,41 +304,36 @@ static btScalar btShortestAngleUpdate(btScalar accAngle, btScalar curAngle)
 	btScalar tol(0.3);
 	btScalar result = btShortestAngularDistance(accAngle, curAngle);
 
-	  if (btFabs(result) > tol)
+	if (btFabs(result) > tol)
 		return curAngle;
-	  else
+	else
 		return accAngle + result;
 
 	return curAngle;
 }
 
-
 btScalar btHingeAccumulatedAngleConstraint::getAccumulatedHingeAngle()
 {
 	btScalar hingeAngle = getHingeAngle();
-	m_accumulatedAngle = btShortestAngleUpdate(m_accumulatedAngle,hingeAngle);
+	m_accumulatedAngle = btShortestAngleUpdate(m_accumulatedAngle, hingeAngle);
 	return m_accumulatedAngle;
 }
-void	btHingeAccumulatedAngleConstraint::setAccumulatedHingeAngle(btScalar accAngle)
+void btHingeAccumulatedAngleConstraint::setAccumulatedHingeAngle(btScalar accAngle)
 {
-	m_accumulatedAngle  = accAngle;
+	m_accumulatedAngle = accAngle;
 }
 
 void btHingeAccumulatedAngleConstraint::getInfo1(btConstraintInfo1* info)
 {
 	//update m_accumulatedAngle
 	btScalar curHingeAngle = getHingeAngle();
-	m_accumulatedAngle = btShortestAngleUpdate(m_accumulatedAngle,curHingeAngle);
+	m_accumulatedAngle = btShortestAngleUpdate(m_accumulatedAngle, curHingeAngle);
 
 	btHingeConstraint::getInfo1(info);
-	
 }
 
-
 void btHingeConstraint::getInfo1(btConstraintInfo1* info)
 {
-
-
 	if (m_useSolveConstraintObsolete)
 	{
 		info->m_numConstraintRows = 0;
@@ -362,17 +341,16 @@ void btHingeConstraint::getInfo1(btConstraintInfo1* info)
 	}
 	else
 	{
-		info->m_numConstraintRows = 5; // Fixed 3 linear + 2 angular
-		info->nub = 1; 
+		info->m_numConstraintRows = 5;  // Fixed 3 linear + 2 angular
+		info->nub = 1;
 		//always add the row, to avoid computation (data is not available yet)
 		//prepare constraint
-		testLimit(m_rbA.getCenterOfMassTransform(),m_rbB.getCenterOfMassTransform());
-		if(getSolveLimit() || getEnableAngularMotor())
+		testLimit(m_rbA.getCenterOfMassTransform(), m_rbB.getCenterOfMassTransform());
+		if (getSolveLimit() || getEnableAngularMotor())
 		{
-			info->m_numConstraintRows++; // limit 3rd anguar as well
-			info->nub--; 
+			info->m_numConstraintRows++;  // limit 3rd anguar as well
+			info->nub--;
 		}
-
 	}
 }
 
@@ -386,41 +364,38 @@ void btHingeConstraint::getInfo1NonVirtual(btConstraintInfo1* info)
 	else
 	{
 		//always add the 'limit' row, to avoid computation (data is not available yet)
-		info->m_numConstraintRows = 6; // Fixed 3 linear + 2 angular
-		info->nub = 0; 
+		info->m_numConstraintRows = 6;  // Fixed 3 linear + 2 angular
+		info->nub = 0;
 	}
 }
 
-void btHingeConstraint::getInfo2 (btConstraintInfo2* info)
+void btHingeConstraint::getInfo2(btConstraintInfo2* info)
 {
-	if(m_useOffsetForConstraintFrame)
+	if (m_useOffsetForConstraintFrame)
 	{
-		getInfo2InternalUsingFrameOffset(info, m_rbA.getCenterOfMassTransform(),m_rbB.getCenterOfMassTransform(),m_rbA.getAngularVelocity(),m_rbB.getAngularVelocity());
+		getInfo2InternalUsingFrameOffset(info, m_rbA.getCenterOfMassTransform(), m_rbB.getCenterOfMassTransform(), m_rbA.getAngularVelocity(), m_rbB.getAngularVelocity());
 	}
 	else
 	{
-		getInfo2Internal(info, m_rbA.getCenterOfMassTransform(),m_rbB.getCenterOfMassTransform(),m_rbA.getAngularVelocity(),m_rbB.getAngularVelocity());
+		getInfo2Internal(info, m_rbA.getCenterOfMassTransform(), m_rbB.getCenterOfMassTransform(), m_rbA.getAngularVelocity(), m_rbB.getAngularVelocity());
 	}
 }
 
-
-void	btHingeConstraint::getInfo2NonVirtual (btConstraintInfo2* info,const btTransform& transA,const btTransform& transB,const btVector3& angVelA,const btVector3& angVelB)
+void btHingeConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTransform& transA, const btTransform& transB, const btVector3& angVelA, const btVector3& angVelB)
 {
 	///the regular (virtual) implementation getInfo2 already performs 'testLimit' during getInfo1, so we need to do it now
-	testLimit(transA,transB);
+	testLimit(transA, transB);
 
-	getInfo2Internal(info,transA,transB,angVelA,angVelB);
+	getInfo2Internal(info, transA, transB, angVelA, angVelB);
 }
 
-
-void btHingeConstraint::getInfo2Internal(btConstraintInfo2* info, const btTransform& transA,const btTransform& transB,const btVector3& angVelA,const btVector3& angVelB)
+void btHingeConstraint::getInfo2Internal(btConstraintInfo2* info, const btTransform& transA, const btTransform& transB, const btVector3& angVelA, const btVector3& angVelB)
 {
-
 	btAssert(!m_useSolveConstraintObsolete);
 	int i, skip = info->rowskip;
 	// transforms in world space
-	btTransform trA = transA*m_rbAFrame;
-	btTransform trB = transB*m_rbBFrame;
+	btTransform trA = transA * m_rbAFrame;
+	btTransform trB = transB * m_rbBFrame;
 	// pivot point
 	btVector3 pivotAInW = trA.getOrigin();
 	btVector3 pivotBInW = trB.getOrigin();
@@ -448,7 +423,7 @@ void btHingeConstraint::getInfo2Internal(btConstraintInfo2* info, const btTransf
 			info->m_constraintError[i*skip]=0.f;
 		}
 	}
-#endif //#if 0
+#endif  //#if 0
 	// linear (all fixed)
 
 	if (!m_angularOnly)
@@ -460,10 +435,7 @@ void btHingeConstraint::getInfo2Internal(btConstraintInfo2* info, const btTransf
 		info->m_J2linearAxis[0] = -1;
 		info->m_J2linearAxis[skip + 1] = -1;
 		info->m_J2linearAxis[2 * skip + 2] = -1;
-	}	
-
-
-
+	}
 
 	btVector3 a1 = pivotAInW - transA.getOrigin();
 	{
@@ -471,22 +443,22 @@ void btHingeConstraint::getInfo2Internal(btConstraintInfo2* info, const btTransf
 		btVector3* angular1 = (btVector3*)(info->m_J1angularAxis + skip);
 		btVector3* angular2 = (btVector3*)(info->m_J1angularAxis + 2 * skip);
 		btVector3 a1neg = -a1;
-		a1neg.getSkewSymmetricMatrix(angular0,angular1,angular2);
+		a1neg.getSkewSymmetricMatrix(angular0, angular1, angular2);
 	}
 	btVector3 a2 = pivotBInW - transB.getOrigin();
 	{
 		btVector3* angular0 = (btVector3*)(info->m_J2angularAxis);
 		btVector3* angular1 = (btVector3*)(info->m_J2angularAxis + skip);
 		btVector3* angular2 = (btVector3*)(info->m_J2angularAxis + 2 * skip);
-		a2.getSkewSymmetricMatrix(angular0,angular1,angular2);
+		a2.getSkewSymmetricMatrix(angular0, angular1, angular2);
 	}
 	// linear RHS
 	btScalar normalErp = (m_flags & BT_HINGE_FLAGS_ERP_NORM) ? m_normalERP : info->erp;
 
-    btScalar k = info->fps * normalErp;
+	btScalar k = info->fps * normalErp;
 	if (!m_angularOnly)
 	{
-		for(i = 0; i < 3; i++)
+		for (i = 0; i < 3; i++)
 		{
 			info->m_constraintError[i * skip] = k * (pivotBInW[i] - pivotAInW[i]);
 		}
@@ -504,9 +476,9 @@ void btHingeConstraint::getInfo2Internal(btConstraintInfo2* info, const btTransf
 	// get 2 orthos to hinge axis (X, Y)
 	btVector3 p = trA.getBasis().getColumn(0);
 	btVector3 q = trA.getBasis().getColumn(1);
-	// set the two hinge angular rows 
-    int s3 = 3 * info->rowskip;
-    int s4 = 4 * info->rowskip;
+	// set the two hinge angular rows
+	int s3 = 3 * info->rowskip;
+	int s4 = 4 * info->rowskip;
 
 	info->m_J1angularAxis[s3 + 0] = p[0];
 	info->m_J1angularAxis[s3 + 1] = p[1];
@@ -521,181 +493,172 @@ void btHingeConstraint::getInfo2Internal(btConstraintInfo2* info, const btTransf
 	info->m_J2angularAxis[s4 + 0] = -q[0];
 	info->m_J2angularAxis[s4 + 1] = -q[1];
 	info->m_J2angularAxis[s4 + 2] = -q[2];
-    // compute the right hand side of the constraint equation. set relative
-    // body velocities along p and q to bring the hinge back into alignment.
-    // if ax1,ax2 are the unit length hinge axes as computed from body1 and
-    // body2, we need to rotate both bodies along the axis u = (ax1 x ax2).
-    // if `theta' is the angle between ax1 and ax2, we need an angular velocity
-    // along u to cover angle erp*theta in one step :
-    //   |angular_velocity| = angle/time = erp*theta / stepsize
-    //                      = (erp*fps) * theta
-    //    angular_velocity  = |angular_velocity| * (ax1 x ax2) / |ax1 x ax2|
-    //                      = (erp*fps) * theta * (ax1 x ax2) / sin(theta)
-    // ...as ax1 and ax2 are unit length. if theta is smallish,
-    // theta ~= sin(theta), so
-    //    angular_velocity  = (erp*fps) * (ax1 x ax2)
-    // ax1 x ax2 is in the plane space of ax1, so we project the angular
-    // velocity to p and q to find the right hand side.
-    btVector3 ax2 = trB.getBasis().getColumn(2);
+	// compute the right hand side of the constraint equation. set relative
+	// body velocities along p and q to bring the hinge back into alignment.
+	// if ax1,ax2 are the unit length hinge axes as computed from body1 and
+	// body2, we need to rotate both bodies along the axis u = (ax1 x ax2).
+	// if `theta' is the angle between ax1 and ax2, we need an angular velocity
+	// along u to cover angle erp*theta in one step :
+	//   |angular_velocity| = angle/time = erp*theta / stepsize
+	//                      = (erp*fps) * theta
+	//    angular_velocity  = |angular_velocity| * (ax1 x ax2) / |ax1 x ax2|
+	//                      = (erp*fps) * theta * (ax1 x ax2) / sin(theta)
+	// ...as ax1 and ax2 are unit length. if theta is smallish,
+	// theta ~= sin(theta), so
+	//    angular_velocity  = (erp*fps) * (ax1 x ax2)
+	// ax1 x ax2 is in the plane space of ax1, so we project the angular
+	// velocity to p and q to find the right hand side.
+	btVector3 ax2 = trB.getBasis().getColumn(2);
 	btVector3 u = ax1.cross(ax2);
 	info->m_constraintError[s3] = k * u.dot(p);
 	info->m_constraintError[s4] = k * u.dot(q);
 	// check angular limits
-	int nrow = 4; // last filled row
+	int nrow = 4;  // last filled row
 	int srow;
 	btScalar limit_err = btScalar(0.0);
 	int limit = 0;
-	if(getSolveLimit())
+	if (getSolveLimit())
 	{
-#ifdef	_BT_USE_CENTER_LIMIT_
-	limit_err = m_limit.getCorrection() * m_referenceSign;
+#ifdef _BT_USE_CENTER_LIMIT_
+		limit_err = m_limit.getCorrection() * m_referenceSign;
 #else
-	limit_err = m_correction * m_referenceSign;
+		limit_err = m_correction * m_referenceSign;
 #endif
-	limit = (limit_err > btScalar(0.0)) ? 1 : 2;
-
+		limit = (limit_err > btScalar(0.0)) ? 1 : 2;
 	}
 	// if the hinge has joint limits or motor, add in the extra row
 	bool powered = getEnableAngularMotor();
-	if(limit || powered)
+	if (limit || powered)
 	{
 		nrow++;
 		srow = nrow * info->rowskip;
-		info->m_J1angularAxis[srow+0] = ax1[0];
-		info->m_J1angularAxis[srow+1] = ax1[1];
-		info->m_J1angularAxis[srow+2] = ax1[2];
+		info->m_J1angularAxis[srow + 0] = ax1[0];
+		info->m_J1angularAxis[srow + 1] = ax1[1];
+		info->m_J1angularAxis[srow + 2] = ax1[2];
 
-		info->m_J2angularAxis[srow+0] = -ax1[0];
-		info->m_J2angularAxis[srow+1] = -ax1[1];
-		info->m_J2angularAxis[srow+2] = -ax1[2];
+		info->m_J2angularAxis[srow + 0] = -ax1[0];
+		info->m_J2angularAxis[srow + 1] = -ax1[1];
+		info->m_J2angularAxis[srow + 2] = -ax1[2];
 
 		btScalar lostop = getLowerLimit();
 		btScalar histop = getUpperLimit();
-		if(limit && (lostop == histop))
+		if (limit && (lostop == histop))
 		{  // the joint motor is ineffective
 			powered = false;
 		}
 		info->m_constraintError[srow] = btScalar(0.0f);
 		btScalar currERP = (m_flags & BT_HINGE_FLAGS_ERP_STOP) ? m_stopERP : normalErp;
-		if(powered)
+		if (powered)
 		{
-			if(m_flags & BT_HINGE_FLAGS_CFM_NORM)
+			if (m_flags & BT_HINGE_FLAGS_CFM_NORM)
 			{
 				info->cfm[srow] = m_normalCFM;
 			}
 			btScalar mot_fact = getMotorFactor(m_hingeAngle, lostop, histop, m_motorTargetVelocity, info->fps * currERP);
 			info->m_constraintError[srow] += mot_fact * m_motorTargetVelocity * m_referenceSign;
-			info->m_lowerLimit[srow] = - m_maxMotorImpulse;
-			info->m_upperLimit[srow] =   m_maxMotorImpulse;
+			info->m_lowerLimit[srow] = -m_maxMotorImpulse;
+			info->m_upperLimit[srow] = m_maxMotorImpulse;
 		}
-		if(limit)
+		if (limit)
 		{
 			k = info->fps * currERP;
 			info->m_constraintError[srow] += k * limit_err;
-			if(m_flags & BT_HINGE_FLAGS_CFM_STOP)
+			if (m_flags & BT_HINGE_FLAGS_CFM_STOP)
 			{
 				info->cfm[srow] = m_stopCFM;
 			}
-			if(lostop == histop) 
+			if (lostop == histop)
 			{
 				// limited low and high simultaneously
 				info->m_lowerLimit[srow] = -SIMD_INFINITY;
 				info->m_upperLimit[srow] = SIMD_INFINITY;
 			}
-			else if(limit == 1) 
-			{ // low limit
+			else if (limit == 1)
+			{  // low limit
 				info->m_lowerLimit[srow] = 0;
 				info->m_upperLimit[srow] = SIMD_INFINITY;
 			}
-			else 
-			{ // high limit
+			else
+			{  // high limit
 				info->m_lowerLimit[srow] = -SIMD_INFINITY;
 				info->m_upperLimit[srow] = 0;
 			}
 			// bounce (we'll use slider parameter abs(1.0 - m_dampingLimAng) for that)
-#ifdef	_BT_USE_CENTER_LIMIT_
+#ifdef _BT_USE_CENTER_LIMIT_
 			btScalar bounce = m_limit.getRelaxationFactor();
 #else
 			btScalar bounce = m_relaxationFactor;
 #endif
-			if(bounce > btScalar(0.0))
+			if (bounce > btScalar(0.0))
 			{
 				btScalar vel = angVelA.dot(ax1);
 				vel -= angVelB.dot(ax1);
 				// only apply bounce if the velocity is incoming, and if the
 				// resulting c[] exceeds what we already have.
-				if(limit == 1)
-				{	// low limit
-					if(vel < 0)
+				if (limit == 1)
+				{  // low limit
+					if (vel < 0)
 					{
 						btScalar newc = -bounce * vel;
-						if(newc > info->m_constraintError[srow])
+						if (newc > info->m_constraintError[srow])
 						{
 							info->m_constraintError[srow] = newc;
 						}
 					}
 				}
 				else
-				{	// high limit - all those computations are reversed
-					if(vel > 0)
+				{  // high limit - all those computations are reversed
+					if (vel > 0)
 					{
 						btScalar newc = -bounce * vel;
-						if(newc < info->m_constraintError[srow])
+						if (newc < info->m_constraintError[srow])
 						{
 							info->m_constraintError[srow] = newc;
 						}
 					}
 				}
 			}
-#ifdef	_BT_USE_CENTER_LIMIT_
+#ifdef _BT_USE_CENTER_LIMIT_
 			info->m_constraintError[srow] *= m_limit.getBiasFactor();
 #else
 			info->m_constraintError[srow] *= m_biasFactor;
 #endif
-		} // if(limit)
-	} // if angular limit or powered
+		}  // if(limit)
+	}      // if angular limit or powered
 }
 
-
-void btHingeConstraint::setFrames(const btTransform & frameA, const btTransform & frameB)
+void btHingeConstraint::setFrames(const btTransform& frameA, const btTransform& frameB)
 {
 	m_rbAFrame = frameA;
 	m_rbBFrame = frameB;
 	buildJacobian();
 }
 
-
-void	btHingeConstraint::updateRHS(btScalar	timeStep)
+void btHingeConstraint::updateRHS(btScalar timeStep)
 {
 	(void)timeStep;
-
 }
 
-
-
-
 btScalar btHingeConstraint::getHingeAngle()
 {
-	return getHingeAngle(m_rbA.getCenterOfMassTransform(),m_rbB.getCenterOfMassTransform());
+	return getHingeAngle(m_rbA.getCenterOfMassTransform(), m_rbB.getCenterOfMassTransform());
 }
 
-btScalar btHingeConstraint::getHingeAngle(const btTransform& transA,const btTransform& transB)
+btScalar btHingeConstraint::getHingeAngle(const btTransform& transA, const btTransform& transB)
 {
-	const btVector3 refAxis0  = transA.getBasis() * m_rbAFrame.getBasis().getColumn(0);
-	const btVector3 refAxis1  = transA.getBasis() * m_rbAFrame.getBasis().getColumn(1);
+	const btVector3 refAxis0 = transA.getBasis() * m_rbAFrame.getBasis().getColumn(0);
+	const btVector3 refAxis1 = transA.getBasis() * m_rbAFrame.getBasis().getColumn(1);
 	const btVector3 swingAxis = transB.getBasis() * m_rbBFrame.getBasis().getColumn(1);
-//	btScalar angle = btAtan2Fast(swingAxis.dot(refAxis0), swingAxis.dot(refAxis1));
+	//	btScalar angle = btAtan2Fast(swingAxis.dot(refAxis0), swingAxis.dot(refAxis1));
 	btScalar angle = btAtan2(swingAxis.dot(refAxis0), swingAxis.dot(refAxis1));
 	return m_referenceSign * angle;
 }
 
-
-
-void btHingeConstraint::testLimit(const btTransform& transA,const btTransform& transB)
+void btHingeConstraint::testLimit(const btTransform& transA, const btTransform& transB)
 {
 	// Compute limit information
-	m_hingeAngle = getHingeAngle(transA,transB);
-#ifdef	_BT_USE_CENTER_LIMIT_
+	m_hingeAngle = getHingeAngle(transA, transB);
+#ifdef _BT_USE_CENTER_LIMIT_
 	m_limit.test(m_hingeAngle);
 #else
 	m_correction = btScalar(0.);
@@ -709,7 +672,7 @@ void btHingeConstraint::testLimit(const btTransform& transA,const btTransform& t
 			m_correction = (m_lowerLimit - m_hingeAngle);
 			m_limitSign = 1.0f;
 			m_solveLimit = true;
-		} 
+		}
 		else if (m_hingeAngle >= m_upperLimit)
 		{
 			m_correction = m_upperLimit - m_hingeAngle;
@@ -721,7 +684,6 @@ void btHingeConstraint::testLimit(const btTransform& transA,const btTransform& t
 	return;
 }
 
-
 static btVector3 vHinge(0, 0, btScalar(1));
 
 void btHingeConstraint::setMotorTarget(const btQuaternion& qAinB, btScalar dt)
@@ -731,14 +693,15 @@ void btHingeConstraint::setMotorTarget(const btQuaternion& qAinB, btScalar dt)
 	qConstraint.normalize();
 
 	// extract "pure" hinge component
-	btVector3 vNoHinge = quatRotate(qConstraint, vHinge); vNoHinge.normalize();
+	btVector3 vNoHinge = quatRotate(qConstraint, vHinge);
+	vNoHinge.normalize();
 	btQuaternion qNoHinge = shortestArcQuat(vHinge, vNoHinge);
 	btQuaternion qHinge = qNoHinge.inverse() * qConstraint;
 	qHinge.normalize();
 
 	// compute angular target, clamped to limits
 	btScalar targetAngle = qHinge.getAngle();
-	if (targetAngle > SIMD_PI) // long way around. flip quat and recalculate.
+	if (targetAngle > SIMD_PI)  // long way around. flip quat and recalculate.
 	{
 		qHinge = -(qHinge);
 		targetAngle = qHinge.getAngle();
@@ -751,7 +714,7 @@ void btHingeConstraint::setMotorTarget(const btQuaternion& qAinB, btScalar dt)
 
 void btHingeConstraint::setMotorTarget(btScalar targetAngle, btScalar dt)
 {
-#ifdef	_BT_USE_CENTER_LIMIT_
+#ifdef _BT_USE_CENTER_LIMIT_
 	m_limit.fit(targetAngle);
 #else
 	if (m_lowerLimit < m_upperLimit)
@@ -763,20 +726,18 @@ void btHingeConstraint::setMotorTarget(btScalar targetAngle, btScalar dt)
 	}
 #endif
 	// compute angular velocity
-	btScalar curAngle  = getHingeAngle(m_rbA.getCenterOfMassTransform(),m_rbB.getCenterOfMassTransform());
+	btScalar curAngle = getHingeAngle(m_rbA.getCenterOfMassTransform(), m_rbB.getCenterOfMassTransform());
 	btScalar dAngle = targetAngle - curAngle;
 	m_motorTargetVelocity = dAngle / dt;
 }
 
-
-
-void btHingeConstraint::getInfo2InternalUsingFrameOffset(btConstraintInfo2* info, const btTransform& transA,const btTransform& transB,const btVector3& angVelA,const btVector3& angVelB)
+void btHingeConstraint::getInfo2InternalUsingFrameOffset(btConstraintInfo2* info, const btTransform& transA, const btTransform& transB, const btVector3& angVelA, const btVector3& angVelB)
 {
 	btAssert(!m_useSolveConstraintObsolete);
 	int i, s = info->rowskip;
 	// transforms in world space
-	btTransform trA = transA*m_rbAFrame;
-	btTransform trB = transB*m_rbBFrame;
+	btTransform trA = transA * m_rbAFrame;
+	btTransform trB = transB * m_rbBFrame;
 	// pivot point
 //	btVector3 pivotAInW = trA.getOrigin();
 //	btVector3 pivotBInW = trB.getOrigin();
@@ -789,11 +750,11 @@ void btHingeConstraint::getInfo2InternalUsingFrameOffset(btConstraintInfo2* info
 	bool hasStaticBody = (miA < SIMD_EPSILON) || (miB < SIMD_EPSILON);
 	btScalar miS = miA + miB;
 	btScalar factA, factB;
-	if(miS > btScalar(0.f))
+	if (miS > btScalar(0.f))
 	{
 		factA = miB / miS;
 	}
-	else 
+	else
 	{
 		factA = btScalar(0.5f);
 	}
@@ -803,15 +764,21 @@ void btHingeConstraint::getInfo2InternalUsingFrameOffset(btConstraintInfo2* info
 	btVector3 ax1A = trA.getBasis().getColumn(2);
 	btVector3 ax1B = trB.getBasis().getColumn(2);
 	btVector3 ax1 = ax1A * factA + ax1B * factB;
+	if (ax1.length2()<SIMD_EPSILON)
+	{
+		factA=0.f;
+		factB=1.f;
+		ax1 = ax1A * factA + ax1B * factB;
+	}
 	ax1.normalize();
-	// fill first 3 rows 
+	// fill first 3 rows
 	// we want: velA + wA x relA == velB + wB x relB
 	btTransform bodyA_trans = transA;
 	btTransform bodyB_trans = transB;
 	int s0 = 0;
 	int s1 = s;
 	int s2 = s * 2;
-	int nrow = 2; // last filled row
+	int nrow = 2;  // last filled row
 	btVector3 tmpA, tmpB, relA, relB, p, q;
 	// get vector from bodyB to frameB in WCS
 	relB = trB.getOrigin() - bodyB_trans.getOrigin();
@@ -830,7 +797,7 @@ void btHingeConstraint::getInfo2InternalUsingFrameOffset(btConstraintInfo2* info
 	// now choose average ortho to hinge axis
 	p = orthoB * factA + orthoA * factB;
 	btScalar len2 = p.length2();
-	if(len2 > SIMD_EPSILON)
+	if (len2 > SIMD_EPSILON)
 	{
 		p /= btSqrt(len2);
 	}
@@ -843,44 +810,44 @@ void btHingeConstraint::getInfo2InternalUsingFrameOffset(btConstraintInfo2* info
 	// fill three rows
 	tmpA = relA.cross(p);
 	tmpB = relB.cross(p);
-    for (i=0; i<3; i++) info->m_J1angularAxis[s0+i] = tmpA[i];
-    for (i=0; i<3; i++) info->m_J2angularAxis[s0+i] = -tmpB[i];
+	for (i = 0; i < 3; i++) info->m_J1angularAxis[s0 + i] = tmpA[i];
+	for (i = 0; i < 3; i++) info->m_J2angularAxis[s0 + i] = -tmpB[i];
 	tmpA = relA.cross(q);
 	tmpB = relB.cross(q);
-	if(hasStaticBody && getSolveLimit())
-	{ // to make constraint between static and dynamic objects more rigid
+	if (hasStaticBody && getSolveLimit())
+	{  // to make constraint between static and dynamic objects more rigid
 		// remove wA (or wB) from equation if angular limit is hit
 		tmpB *= factB;
 		tmpA *= factA;
 	}
-	for (i=0; i<3; i++) info->m_J1angularAxis[s1+i] = tmpA[i];
-    for (i=0; i<3; i++) info->m_J2angularAxis[s1+i] = -tmpB[i];
+	for (i = 0; i < 3; i++) info->m_J1angularAxis[s1 + i] = tmpA[i];
+	for (i = 0; i < 3; i++) info->m_J2angularAxis[s1 + i] = -tmpB[i];
 	tmpA = relA.cross(ax1);
 	tmpB = relB.cross(ax1);
-	if(hasStaticBody)
-	{ // to make constraint between static and dynamic objects more rigid
+	if (hasStaticBody)
+	{  // to make constraint between static and dynamic objects more rigid
 		// remove wA (or wB) from equation
 		tmpB *= factB;
 		tmpA *= factA;
 	}
-	for (i=0; i<3; i++) info->m_J1angularAxis[s2+i] = tmpA[i];
-    for (i=0; i<3; i++) info->m_J2angularAxis[s2+i] = -tmpB[i];
+	for (i = 0; i < 3; i++) info->m_J1angularAxis[s2 + i] = tmpA[i];
+	for (i = 0; i < 3; i++) info->m_J2angularAxis[s2 + i] = -tmpB[i];
 
-	btScalar normalErp = (m_flags & BT_HINGE_FLAGS_ERP_NORM)? m_normalERP : info->erp;
+	btScalar normalErp = (m_flags & BT_HINGE_FLAGS_ERP_NORM) ? m_normalERP : info->erp;
 	btScalar k = info->fps * normalErp;
 
 	if (!m_angularOnly)
 	{
-		for (i=0; i<3; i++) info->m_J1linearAxis[s0+i] = p[i];
-		for (i=0; i<3; i++) info->m_J1linearAxis[s1+i] = q[i];
-		for (i=0; i<3; i++) info->m_J1linearAxis[s2+i] = ax1[i];
+		for (i = 0; i < 3; i++) info->m_J1linearAxis[s0 + i] = p[i];
+		for (i = 0; i < 3; i++) info->m_J1linearAxis[s1 + i] = q[i];
+		for (i = 0; i < 3; i++) info->m_J1linearAxis[s2 + i] = ax1[i];
 
-		for (i=0; i<3; i++) info->m_J2linearAxis[s0+i] = -p[i];
-		for (i=0; i<3; i++) info->m_J2linearAxis[s1+i] = -q[i];
-		for (i=0; i<3; i++) info->m_J2linearAxis[s2+i] = -ax1[i];
+		for (i = 0; i < 3; i++) info->m_J2linearAxis[s0 + i] = -p[i];
+		for (i = 0; i < 3; i++) info->m_J2linearAxis[s1 + i] = -q[i];
+		for (i = 0; i < 3; i++) info->m_J2linearAxis[s2 + i] = -ax1[i];
+
+		// compute three elements of right hand side
 
-	// compute three elements of right hand side
-	
 		btScalar rhs = k * p.dot(ofs);
 		info->m_constraintError[s0] = rhs;
 		rhs = k * q.dot(ofs);
@@ -925,146 +892,144 @@ void btHingeConstraint::getInfo2InternalUsingFrameOffset(btConstraintInfo2* info
 	//    angular_velocity  = (erp*fps) * (ax1 x ax2)
 	// ax1 x ax2 is in the plane space of ax1, so we project the angular
 	// velocity to p and q to find the right hand side.
-	k = info->fps * normalErp;//??
+	k = info->fps * normalErp;  //??
 
 	btVector3 u = ax1A.cross(ax1B);
 	info->m_constraintError[s3] = k * u.dot(p);
 	info->m_constraintError[s4] = k * u.dot(q);
 #endif
 	// check angular limits
-	nrow = 4; // last filled row
+	nrow = 4;  // last filled row
 	int srow;
 	btScalar limit_err = btScalar(0.0);
 	int limit = 0;
-	if(getSolveLimit())
+	if (getSolveLimit())
 	{
-#ifdef	_BT_USE_CENTER_LIMIT_
-	limit_err = m_limit.getCorrection() * m_referenceSign;
+#ifdef _BT_USE_CENTER_LIMIT_
+		limit_err = m_limit.getCorrection() * m_referenceSign;
 #else
-	limit_err = m_correction * m_referenceSign;
+		limit_err = m_correction * m_referenceSign;
 #endif
-	limit = (limit_err > btScalar(0.0)) ? 1 : 2;
-
+		limit = (limit_err > btScalar(0.0)) ? 1 : 2;
 	}
 	// if the hinge has joint limits or motor, add in the extra row
 	bool powered = getEnableAngularMotor();
-	if(limit || powered)
+	if (limit || powered)
 	{
 		nrow++;
 		srow = nrow * info->rowskip;
-		info->m_J1angularAxis[srow+0] = ax1[0];
-		info->m_J1angularAxis[srow+1] = ax1[1];
-		info->m_J1angularAxis[srow+2] = ax1[2];
+		info->m_J1angularAxis[srow + 0] = ax1[0];
+		info->m_J1angularAxis[srow + 1] = ax1[1];
+		info->m_J1angularAxis[srow + 2] = ax1[2];
 
-		info->m_J2angularAxis[srow+0] = -ax1[0];
-		info->m_J2angularAxis[srow+1] = -ax1[1];
-		info->m_J2angularAxis[srow+2] = -ax1[2];
+		info->m_J2angularAxis[srow + 0] = -ax1[0];
+		info->m_J2angularAxis[srow + 1] = -ax1[1];
+		info->m_J2angularAxis[srow + 2] = -ax1[2];
 
 		btScalar lostop = getLowerLimit();
 		btScalar histop = getUpperLimit();
-		if(limit && (lostop == histop))
+		if (limit && (lostop == histop))
 		{  // the joint motor is ineffective
 			powered = false;
 		}
 		info->m_constraintError[srow] = btScalar(0.0f);
 		btScalar currERP = (m_flags & BT_HINGE_FLAGS_ERP_STOP) ? m_stopERP : normalErp;
-		if(powered)
+		if (powered)
 		{
-			if(m_flags & BT_HINGE_FLAGS_CFM_NORM)
+			if (m_flags & BT_HINGE_FLAGS_CFM_NORM)
 			{
 				info->cfm[srow] = m_normalCFM;
 			}
 			btScalar mot_fact = getMotorFactor(m_hingeAngle, lostop, histop, m_motorTargetVelocity, info->fps * currERP);
 			info->m_constraintError[srow] += mot_fact * m_motorTargetVelocity * m_referenceSign;
-			info->m_lowerLimit[srow] = - m_maxMotorImpulse;
-			info->m_upperLimit[srow] =   m_maxMotorImpulse;
+			info->m_lowerLimit[srow] = -m_maxMotorImpulse;
+			info->m_upperLimit[srow] = m_maxMotorImpulse;
 		}
-		if(limit)
+		if (limit)
 		{
 			k = info->fps * currERP;
 			info->m_constraintError[srow] += k * limit_err;
-			if(m_flags & BT_HINGE_FLAGS_CFM_STOP)
+			if (m_flags & BT_HINGE_FLAGS_CFM_STOP)
 			{
 				info->cfm[srow] = m_stopCFM;
 			}
-			if(lostop == histop) 
+			if (lostop == histop)
 			{
 				// limited low and high simultaneously
 				info->m_lowerLimit[srow] = -SIMD_INFINITY;
 				info->m_upperLimit[srow] = SIMD_INFINITY;
 			}
-			else if(limit == 1) 
-			{ // low limit
+			else if (limit == 1)
+			{  // low limit
 				info->m_lowerLimit[srow] = 0;
 				info->m_upperLimit[srow] = SIMD_INFINITY;
 			}
-			else 
-			{ // high limit
+			else
+			{  // high limit
 				info->m_lowerLimit[srow] = -SIMD_INFINITY;
 				info->m_upperLimit[srow] = 0;
 			}
 			// bounce (we'll use slider parameter abs(1.0 - m_dampingLimAng) for that)
-#ifdef	_BT_USE_CENTER_LIMIT_
+#ifdef _BT_USE_CENTER_LIMIT_
 			btScalar bounce = m_limit.getRelaxationFactor();
 #else
 			btScalar bounce = m_relaxationFactor;
 #endif
-			if(bounce > btScalar(0.0))
+			if (bounce > btScalar(0.0))
 			{
 				btScalar vel = angVelA.dot(ax1);
 				vel -= angVelB.dot(ax1);
 				// only apply bounce if the velocity is incoming, and if the
 				// resulting c[] exceeds what we already have.
-				if(limit == 1)
-				{	// low limit
-					if(vel < 0)
+				if (limit == 1)
+				{  // low limit
+					if (vel < 0)
 					{
 						btScalar newc = -bounce * vel;
-						if(newc > info->m_constraintError[srow])
+						if (newc > info->m_constraintError[srow])
 						{
 							info->m_constraintError[srow] = newc;
 						}
 					}
 				}
 				else
-				{	// high limit - all those computations are reversed
-					if(vel > 0)
+				{  // high limit - all those computations are reversed
+					if (vel > 0)
 					{
 						btScalar newc = -bounce * vel;
-						if(newc < info->m_constraintError[srow])
+						if (newc < info->m_constraintError[srow])
 						{
 							info->m_constraintError[srow] = newc;
 						}
 					}
 				}
 			}
-#ifdef	_BT_USE_CENTER_LIMIT_
+#ifdef _BT_USE_CENTER_LIMIT_
 			info->m_constraintError[srow] *= m_limit.getBiasFactor();
 #else
 			info->m_constraintError[srow] *= m_biasFactor;
 #endif
-		} // if(limit)
-	} // if angular limit or powered
+		}  // if(limit)
+	}      // if angular limit or powered
 }
 
-
-///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5). 
+///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5).
 ///If no axis is provided, it uses the default axis for this constraint.
 void btHingeConstraint::setParam(int num, btScalar value, int axis)
 {
-	if((axis == -1) || (axis == 5))
+	if ((axis == -1) || (axis == 5))
 	{
-		switch(num)
-		{	
-			case BT_CONSTRAINT_STOP_ERP :
+		switch (num)
+		{
+			case BT_CONSTRAINT_STOP_ERP:
 				m_stopERP = value;
 				m_flags |= BT_HINGE_FLAGS_ERP_STOP;
 				break;
-			case BT_CONSTRAINT_STOP_CFM :
+			case BT_CONSTRAINT_STOP_CFM:
 				m_stopCFM = value;
 				m_flags |= BT_HINGE_FLAGS_CFM_STOP;
 				break;
-			case BT_CONSTRAINT_CFM :
+			case BT_CONSTRAINT_CFM:
 				m_normalCFM = value;
 				m_flags |= BT_HINGE_FLAGS_CFM_NORM;
 				break;
@@ -1072,7 +1037,7 @@ void btHingeConstraint::setParam(int num, btScalar value, int axis)
 				m_normalERP = value;
 				m_flags |= BT_HINGE_FLAGS_ERP_NORM;
 				break;
-			default : 
+			default:
 				btAssertConstrParams(0);
 		}
 	}
@@ -1083,22 +1048,22 @@ void btHingeConstraint::setParam(int num, btScalar value, int axis)
 }
 
 ///return the local value of parameter
-btScalar btHingeConstraint::getParam(int num, int axis) const 
+btScalar btHingeConstraint::getParam(int num, int axis) const
 {
 	btScalar retVal = 0;
-	if((axis == -1) || (axis == 5))
+	if ((axis == -1) || (axis == 5))
 	{
-		switch(num)
-		{	
-			case BT_CONSTRAINT_STOP_ERP :
+		switch (num)
+		{
+			case BT_CONSTRAINT_STOP_ERP:
 				btAssertConstrParams(m_flags & BT_HINGE_FLAGS_ERP_STOP);
 				retVal = m_stopERP;
 				break;
-			case BT_CONSTRAINT_STOP_CFM :
+			case BT_CONSTRAINT_STOP_CFM:
 				btAssertConstrParams(m_flags & BT_HINGE_FLAGS_CFM_STOP);
 				retVal = m_stopCFM;
 				break;
-			case BT_CONSTRAINT_CFM :
+			case BT_CONSTRAINT_CFM:
 				btAssertConstrParams(m_flags & BT_HINGE_FLAGS_CFM_NORM);
 				retVal = m_normalCFM;
 				break;
@@ -1106,7 +1071,7 @@ btScalar btHingeConstraint::getParam(int num, int axis) const
 				btAssertConstrParams(m_flags & BT_HINGE_FLAGS_ERP_NORM);
 				retVal = m_normalERP;
 				break;
-			default : 
+			default:
 				btAssertConstrParams(0);
 		}
 	}
@@ -1116,5 +1081,3 @@ btScalar btHingeConstraint::getParam(int num, int axis) const
 	}
 	return retVal;
 }
-
-
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btHingeConstraint.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btHingeConstraint.h
index 3c3df24dba2..c7509e30afb 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btHingeConstraint.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btHingeConstraint.h
@@ -20,7 +20,6 @@ subject to the following restrictions:
 
 #define _BT_USE_CENTER_LIMIT_ 1
 
-
 #include "LinearMath/btVector3.h"
 #include "btJacobianEntry.h"
 #include "btTypedConstraint.h"
@@ -28,14 +27,12 @@ subject to the following restrictions:
 class btRigidBody;
 
 #ifdef BT_USE_DOUBLE_PRECISION
-#define btHingeConstraintData	btHingeConstraintDoubleData2 //rename to 2 for backwards compatibility, so we can still load the 'btHingeConstraintDoubleData' version
-#define btHingeConstraintDataName	"btHingeConstraintDoubleData2" 
+#define btHingeConstraintData btHingeConstraintDoubleData2  //rename to 2 for backwards compatibility, so we can still load the 'btHingeConstraintDoubleData' version
+#define btHingeConstraintDataName "btHingeConstraintDoubleData2"
 #else
-#define btHingeConstraintData	btHingeConstraintFloatData
-#define btHingeConstraintDataName	"btHingeConstraintFloatData"
-#endif //BT_USE_DOUBLE_PRECISION
-
-
+#define btHingeConstraintData btHingeConstraintFloatData
+#define btHingeConstraintDataName "btHingeConstraintFloatData"
+#endif  //BT_USE_DOUBLE_PRECISION
 
 enum btHingeFlags
 {
@@ -45,89 +42,83 @@ enum btHingeFlags
 	BT_HINGE_FLAGS_ERP_NORM = 8
 };
 
-
 /// hinge constraint between two rigidbodies each with a pivotpoint that descibes the axis location in local space
 /// axis defines the orientation of the hinge axis
-ATTRIBUTE_ALIGNED16(class) btHingeConstraint : public btTypedConstraint
+ATTRIBUTE_ALIGNED16(class)
+btHingeConstraint : public btTypedConstraint
 {
 #ifdef IN_PARALLELL_SOLVER
 public:
 #endif
-	btJacobianEntry	m_jac[3]; //3 orthogonal linear constraints
-	btJacobianEntry	m_jacAng[3]; //2 orthogonal angular constraints+ 1 for limit/motor
+	btJacobianEntry m_jac[3];     //3 orthogonal linear constraints
+	btJacobianEntry m_jacAng[3];  //2 orthogonal angular constraints+ 1 for limit/motor
 
-	btTransform m_rbAFrame; // constraint axii. Assumes z is hinge axis.
+	btTransform m_rbAFrame;  // constraint axii. Assumes z is hinge axis.
 	btTransform m_rbBFrame;
 
-	btScalar	m_motorTargetVelocity;
-	btScalar	m_maxMotorImpulse;
+	btScalar m_motorTargetVelocity;
+	btScalar m_maxMotorImpulse;
 
-
-#ifdef	_BT_USE_CENTER_LIMIT_
-	btAngularLimit	m_limit;
+#ifdef _BT_USE_CENTER_LIMIT_
+	btAngularLimit m_limit;
 #else
-	btScalar	m_lowerLimit;	
-	btScalar	m_upperLimit;	
-	btScalar	m_limitSign;
-	btScalar	m_correction;
+	btScalar m_lowerLimit;
+	btScalar m_upperLimit;
+	btScalar m_limitSign;
+	btScalar m_correction;
 
-	btScalar	m_limitSoftness; 
-	btScalar	m_biasFactor; 
-	btScalar	m_relaxationFactor; 
+	btScalar m_limitSoftness;
+	btScalar m_biasFactor;
+	btScalar m_relaxationFactor;
 
-	bool		m_solveLimit;
+	bool m_solveLimit;
 #endif
 
-	btScalar	m_kHinge;
+	btScalar m_kHinge;
 
+	btScalar m_accLimitImpulse;
+	btScalar m_hingeAngle;
+	btScalar m_referenceSign;
 
-	btScalar	m_accLimitImpulse;
-	btScalar	m_hingeAngle;
-	btScalar	m_referenceSign;
+	bool m_angularOnly;
+	bool m_enableAngularMotor;
+	bool m_useSolveConstraintObsolete;
+	bool m_useOffsetForConstraintFrame;
+	bool m_useReferenceFrameA;
 
-	bool		m_angularOnly;
-	bool		m_enableAngularMotor;
-	bool		m_useSolveConstraintObsolete;
-	bool		m_useOffsetForConstraintFrame;
-	bool		m_useReferenceFrameA;
+	btScalar m_accMotorImpulse;
 
-	btScalar	m_accMotorImpulse;
+	int m_flags;
+	btScalar m_normalCFM;
+	btScalar m_normalERP;
+	btScalar m_stopCFM;
+	btScalar m_stopERP;
 
-	int			m_flags;
-	btScalar	m_normalCFM;
-	btScalar	m_normalERP;
-	btScalar	m_stopCFM;
-	btScalar	m_stopERP;
-
-	
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
-	btHingeConstraint(btRigidBody& rbA,btRigidBody& rbB, const btVector3& pivotInA,const btVector3& pivotInB, const btVector3& axisInA,const btVector3& axisInB, bool useReferenceFrameA = false);
 
-	btHingeConstraint(btRigidBody& rbA,const btVector3& pivotInA,const btVector3& axisInA, bool useReferenceFrameA = false);
-	
-	btHingeConstraint(btRigidBody& rbA,btRigidBody& rbB, const btTransform& rbAFrame, const btTransform& rbBFrame, bool useReferenceFrameA = false);
+	btHingeConstraint(btRigidBody & rbA, btRigidBody & rbB, const btVector3& pivotInA, const btVector3& pivotInB, const btVector3& axisInA, const btVector3& axisInB, bool useReferenceFrameA = false);
 
-	btHingeConstraint(btRigidBody& rbA,const btTransform& rbAFrame, bool useReferenceFrameA = false);
+	btHingeConstraint(btRigidBody & rbA, const btVector3& pivotInA, const btVector3& axisInA, bool useReferenceFrameA = false);
 
+	btHingeConstraint(btRigidBody & rbA, btRigidBody & rbB, const btTransform& rbAFrame, const btTransform& rbBFrame, bool useReferenceFrameA = false);
 
-	virtual void	buildJacobian();
+	btHingeConstraint(btRigidBody & rbA, const btTransform& rbAFrame, bool useReferenceFrameA = false);
 
-	virtual void getInfo1 (btConstraintInfo1* info);
+	virtual void buildJacobian();
 
-	void getInfo1NonVirtual(btConstraintInfo1* info);
+	virtual void getInfo1(btConstraintInfo1 * info);
 
-	virtual void getInfo2 (btConstraintInfo2* info);
+	void getInfo1NonVirtual(btConstraintInfo1 * info);
 
-	void	getInfo2NonVirtual(btConstraintInfo2* info,const btTransform& transA,const btTransform& transB,const btVector3& angVelA,const btVector3& angVelB);
+	virtual void getInfo2(btConstraintInfo2 * info);
 
-	void	getInfo2Internal(btConstraintInfo2* info,const btTransform& transA,const btTransform& transB,const btVector3& angVelA,const btVector3& angVelB);
-	void	getInfo2InternalUsingFrameOffset(btConstraintInfo2* info,const btTransform& transA,const btTransform& transB,const btVector3& angVelA,const btVector3& angVelB);
-		
+	void getInfo2NonVirtual(btConstraintInfo2 * info, const btTransform& transA, const btTransform& transB, const btVector3& angVelA, const btVector3& angVelB);
 
-	void	updateRHS(btScalar	timeStep);
+	void getInfo2Internal(btConstraintInfo2 * info, const btTransform& transA, const btTransform& transB, const btVector3& angVelA, const btVector3& angVelB);
+	void getInfo2InternalUsingFrameOffset(btConstraintInfo2 * info, const btTransform& transA, const btTransform& transB, const btVector3& angVelA, const btVector3& angVelB);
+
+	void updateRHS(btScalar timeStep);
 
 	const btRigidBody& getRigidBodyA() const
 	{
@@ -138,19 +129,19 @@ public:
 		return m_rbB;
 	}
 
-	btRigidBody& getRigidBodyA()	
-	{		
-		return m_rbA;	
-	}	
+	btRigidBody& getRigidBodyA()
+	{
+		return m_rbA;
+	}
 
-	btRigidBody& getRigidBodyB()	
-	{		
-		return m_rbB;	
+	btRigidBody& getRigidBodyB()
+	{
+		return m_rbB;
 	}
 
 	btTransform& getFrameOffsetA()
 	{
-	return m_rbAFrame;
+		return m_rbAFrame;
 	}
 
 	btTransform& getFrameOffsetB()
@@ -159,15 +150,15 @@ public:
 	}
 
 	void setFrames(const btTransform& frameA, const btTransform& frameB);
-	
-	void	setAngularOnly(bool angularOnly)
+
+	void setAngularOnly(bool angularOnly)
 	{
 		m_angularOnly = angularOnly;
 	}
 
-	void	enableAngularMotor(bool enableMotor,btScalar targetVelocity,btScalar maxMotorImpulse)
+	void enableAngularMotor(bool enableMotor, btScalar targetVelocity, btScalar maxMotorImpulse)
 	{
-		m_enableAngularMotor  = enableMotor;
+		m_enableAngularMotor = enableMotor;
 		m_motorTargetVelocity = targetVelocity;
 		m_maxMotorImpulse = maxMotorImpulse;
 	}
@@ -175,29 +166,28 @@ public:
 	// extra motor API, including ability to set a target rotation (as opposed to angular velocity)
 	// note: setMotorTarget sets angular velocity under the hood, so you must call it every tick to
 	//       maintain a given angular target.
-	void enableMotor(bool enableMotor) 	{ m_enableAngularMotor = enableMotor; }
+	void enableMotor(bool enableMotor) { m_enableAngularMotor = enableMotor; }
 	void setMaxMotorImpulse(btScalar maxMotorImpulse) { m_maxMotorImpulse = maxMotorImpulse; }
 	void setMotorTargetVelocity(btScalar motorTargetVelocity) { m_motorTargetVelocity = motorTargetVelocity; }
-	void setMotorTarget(const btQuaternion& qAinB, btScalar dt); // qAinB is rotation of body A wrt body B.
+	void setMotorTarget(const btQuaternion& qAinB, btScalar dt);  // qAinB is rotation of body A wrt body B.
 	void setMotorTarget(btScalar targetAngle, btScalar dt);
 
-
-	void	setLimit(btScalar low,btScalar high,btScalar _softness = 0.9f, btScalar _biasFactor = 0.3f, btScalar _relaxationFactor = 1.0f)
+	void setLimit(btScalar low, btScalar high, btScalar _softness = 0.9f, btScalar _biasFactor = 0.3f, btScalar _relaxationFactor = 1.0f)
 	{
-#ifdef	_BT_USE_CENTER_LIMIT_
+#ifdef _BT_USE_CENTER_LIMIT_
 		m_limit.set(low, high, _softness, _biasFactor, _relaxationFactor);
 #else
 		m_lowerLimit = btNormalizeAngle(low);
 		m_upperLimit = btNormalizeAngle(high);
-		m_limitSoftness =  _softness;
+		m_limitSoftness = _softness;
 		m_biasFactor = _biasFactor;
 		m_relaxationFactor = _relaxationFactor;
 #endif
 	}
-	
+
 	btScalar getLimitSoftness() const
 	{
-#ifdef	_BT_USE_CENTER_LIMIT_
+#ifdef _BT_USE_CENTER_LIMIT_
 		return m_limit.getSoftness();
 #else
 		return m_limitSoftness;
@@ -206,7 +196,7 @@ public:
 
 	btScalar getLimitBiasFactor() const
 	{
-#ifdef	_BT_USE_CENTER_LIMIT_
+#ifdef _BT_USE_CENTER_LIMIT_
 		return m_limit.getBiasFactor();
 #else
 		return m_biasFactor;
@@ -215,112 +205,110 @@ public:
 
 	btScalar getLimitRelaxationFactor() const
 	{
-#ifdef	_BT_USE_CENTER_LIMIT_
+#ifdef _BT_USE_CENTER_LIMIT_
 		return m_limit.getRelaxationFactor();
 #else
 		return m_relaxationFactor;
 #endif
 	}
 
-	void	setAxis(btVector3& axisInA)
+	void setAxis(btVector3 & axisInA)
 	{
 		btVector3 rbAxisA1, rbAxisA2;
 		btPlaneSpace1(axisInA, rbAxisA1, rbAxisA2);
 		btVector3 pivotInA = m_rbAFrame.getOrigin();
-//		m_rbAFrame.getOrigin() = pivotInA;
-		m_rbAFrame.getBasis().setValue( rbAxisA1.getX(),rbAxisA2.getX(),axisInA.getX(),
-										rbAxisA1.getY(),rbAxisA2.getY(),axisInA.getY(),
-										rbAxisA1.getZ(),rbAxisA2.getZ(),axisInA.getZ() );
+		//		m_rbAFrame.getOrigin() = pivotInA;
+		m_rbAFrame.getBasis().setValue(rbAxisA1.getX(), rbAxisA2.getX(), axisInA.getX(),
+									   rbAxisA1.getY(), rbAxisA2.getY(), axisInA.getY(),
+									   rbAxisA1.getZ(), rbAxisA2.getZ(), axisInA.getZ());
 
 		btVector3 axisInB = m_rbA.getCenterOfMassTransform().getBasis() * axisInA;
 
-		btQuaternion rotationArc = shortestArcQuat(axisInA,axisInB);
-		btVector3 rbAxisB1 =  quatRotate(rotationArc,rbAxisA1);
+		btQuaternion rotationArc = shortestArcQuat(axisInA, axisInB);
+		btVector3 rbAxisB1 = quatRotate(rotationArc, rbAxisA1);
 		btVector3 rbAxisB2 = axisInB.cross(rbAxisB1);
 
 		m_rbBFrame.getOrigin() = m_rbB.getCenterOfMassTransform().inverse()(m_rbA.getCenterOfMassTransform()(pivotInA));
 
-		m_rbBFrame.getBasis().setValue( rbAxisB1.getX(),rbAxisB2.getX(),axisInB.getX(),
-										rbAxisB1.getY(),rbAxisB2.getY(),axisInB.getY(),
-										rbAxisB1.getZ(),rbAxisB2.getZ(),axisInB.getZ() );
+		m_rbBFrame.getBasis().setValue(rbAxisB1.getX(), rbAxisB2.getX(), axisInB.getX(),
+									   rbAxisB1.getY(), rbAxisB2.getY(), axisInB.getY(),
+									   rbAxisB1.getZ(), rbAxisB2.getZ(), axisInB.getZ());
 		m_rbBFrame.getBasis() = m_rbB.getCenterOfMassTransform().getBasis().inverse() * m_rbBFrame.getBasis();
-
 	}
 
-    bool hasLimit() const {
-#ifdef  _BT_USE_CENTER_LIMIT_
-        return m_limit.getHalfRange() > 0;
-#else
-        return m_lowerLimit <= m_upperLimit;
-#endif
-    }
-
-	btScalar	getLowerLimit() const
+	bool hasLimit() const
 	{
-#ifdef	_BT_USE_CENTER_LIMIT_
-	return m_limit.getLow();
+#ifdef _BT_USE_CENTER_LIMIT_
+		return m_limit.getHalfRange() > 0;
 #else
-	return m_lowerLimit;
+		return m_lowerLimit <= m_upperLimit;
 #endif
 	}
 
-	btScalar	getUpperLimit() const
+	btScalar getLowerLimit() const
 	{
-#ifdef	_BT_USE_CENTER_LIMIT_
-	return m_limit.getHigh();
-#else		
-	return m_upperLimit;
+#ifdef _BT_USE_CENTER_LIMIT_
+		return m_limit.getLow();
+#else
+		return m_lowerLimit;
 #endif
 	}
 
+	btScalar getUpperLimit() const
+	{
+#ifdef _BT_USE_CENTER_LIMIT_
+		return m_limit.getHigh();
+#else
+		return m_upperLimit;
+#endif
+	}
 
 	///The getHingeAngle gives the hinge angle in range [-PI,PI]
 	btScalar getHingeAngle();
 
-	btScalar getHingeAngle(const btTransform& transA,const btTransform& transB);
+	btScalar getHingeAngle(const btTransform& transA, const btTransform& transB);
 
-	void testLimit(const btTransform& transA,const btTransform& transB);
+	void testLimit(const btTransform& transA, const btTransform& transB);
 
-
-	const btTransform& getAFrame() const { return m_rbAFrame; };	
+	const btTransform& getAFrame() const { return m_rbAFrame; };
 	const btTransform& getBFrame() const { return m_rbBFrame; };
 
-	btTransform& getAFrame() { return m_rbAFrame; };	
+	btTransform& getAFrame() { return m_rbAFrame; };
 	btTransform& getBFrame() { return m_rbBFrame; };
 
 	inline int getSolveLimit()
 	{
-#ifdef	_BT_USE_CENTER_LIMIT_
-	return m_limit.isLimit();
+#ifdef _BT_USE_CENTER_LIMIT_
+		return m_limit.isLimit();
 #else
-	return m_solveLimit;
+		return m_solveLimit;
 #endif
 	}
 
 	inline btScalar getLimitSign()
 	{
-#ifdef	_BT_USE_CENTER_LIMIT_
-	return m_limit.getSign();
+#ifdef _BT_USE_CENTER_LIMIT_
+		return m_limit.getSign();
 #else
 		return m_limitSign;
 #endif
 	}
 
-	inline bool getAngularOnly() 
-	{ 
-		return m_angularOnly; 
+	inline bool getAngularOnly()
+	{
+		return m_angularOnly;
 	}
-	inline bool getEnableAngularMotor() 
-	{ 
-		return m_enableAngularMotor; 
+	inline bool getEnableAngularMotor()
+	{
+		return m_enableAngularMotor;
 	}
-	inline btScalar getMotorTargetVelocity() 
-	{ 
-		return m_motorTargetVelocity; 
+	inline btScalar getMotorTargetVelocity()
+	{
+		return m_motorTargetVelocity;
 	}
-	inline btScalar getMaxMotorImpulse() 
-	{ 
-		return m_maxMotorImpulse; 
+	inline btScalar getMaxMotorImpulse()
+	{
+		return m_maxMotorImpulse;
 	}
 	// access for UseFrameOffset
 	bool getUseFrameOffset() { return m_useOffsetForConstraintFrame; }
@@ -329,143 +317,132 @@ public:
 	bool getUseReferenceFrameA() const { return m_useReferenceFrameA; }
 	void setUseReferenceFrameA(bool useReferenceFrameA) { m_useReferenceFrameA = useReferenceFrameA; }
 
-	///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5). 
+	///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5).
 	///If no axis is provided, it uses the default axis for this constraint.
-	virtual	void	setParam(int num, btScalar value, int axis = -1);
+	virtual void setParam(int num, btScalar value, int axis = -1);
 	///return the local value of parameter
-	virtual	btScalar getParam(int num, int axis = -1) const;
-	
-	virtual	int getFlags() const
+	virtual btScalar getParam(int num, int axis = -1) const;
+
+	virtual int getFlags() const
 	{
-  	    return m_flags;
+		return m_flags;
 	}
 
-	virtual	int	calculateSerializeBufferSize() const;
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
-
-
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 };
 
-
 //only for backward compatibility
 #ifdef BT_BACKWARDS_COMPATIBLE_SERIALIZATION
 ///this structure is not used, except for loading pre-2.82 .bullet files
-struct	btHingeConstraintDoubleData
+struct btHingeConstraintDoubleData
 {
-	btTypedConstraintData	m_typeConstraintData;
-	btTransformDoubleData m_rbAFrame; // constraint axii. Assumes z is hinge axis.
+	btTypedConstraintData m_typeConstraintData;
+	btTransformDoubleData m_rbAFrame;  // constraint axii. Assumes z is hinge axis.
 	btTransformDoubleData m_rbBFrame;
-	int			m_useReferenceFrameA;
-	int			m_angularOnly;
-	int			m_enableAngularMotor;
-	float	m_motorTargetVelocity;
-	float	m_maxMotorImpulse;
-
-	float	m_lowerLimit;
-	float	m_upperLimit;
-	float	m_limitSoftness;
-	float	m_biasFactor;
-	float	m_relaxationFactor;
+	int m_useReferenceFrameA;
+	int m_angularOnly;
+	int m_enableAngularMotor;
+	float m_motorTargetVelocity;
+	float m_maxMotorImpulse;
 
+	float m_lowerLimit;
+	float m_upperLimit;
+	float m_limitSoftness;
+	float m_biasFactor;
+	float m_relaxationFactor;
 };
-#endif //BT_BACKWARDS_COMPATIBLE_SERIALIZATION
+#endif  //BT_BACKWARDS_COMPATIBLE_SERIALIZATION
 
 ///The getAccumulatedHingeAngle returns the accumulated hinge angle, taking rotation across the -PI/PI boundary into account
-ATTRIBUTE_ALIGNED16(class) btHingeAccumulatedAngleConstraint : public btHingeConstraint
+ATTRIBUTE_ALIGNED16(class)
+btHingeAccumulatedAngleConstraint : public btHingeConstraint
 {
 protected:
-	btScalar	m_accumulatedAngle;
+	btScalar m_accumulatedAngle;
+
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
-	btHingeAccumulatedAngleConstraint(btRigidBody& rbA,btRigidBody& rbB, const btVector3& pivotInA,const btVector3& pivotInB, const btVector3& axisInA,const btVector3& axisInB, bool useReferenceFrameA = false)
-	:btHingeConstraint(rbA,rbB,pivotInA,pivotInB, axisInA,axisInB, useReferenceFrameA )
+
+	btHingeAccumulatedAngleConstraint(btRigidBody & rbA, btRigidBody & rbB, const btVector3& pivotInA, const btVector3& pivotInB, const btVector3& axisInA, const btVector3& axisInB, bool useReferenceFrameA = false)
+		: btHingeConstraint(rbA, rbB, pivotInA, pivotInB, axisInA, axisInB, useReferenceFrameA)
 	{
-		m_accumulatedAngle=getHingeAngle();
+		m_accumulatedAngle = getHingeAngle();
 	}
 
-	btHingeAccumulatedAngleConstraint(btRigidBody& rbA,const btVector3& pivotInA,const btVector3& axisInA, bool useReferenceFrameA = false)
-	:btHingeConstraint(rbA,pivotInA,axisInA, useReferenceFrameA)
+	btHingeAccumulatedAngleConstraint(btRigidBody & rbA, const btVector3& pivotInA, const btVector3& axisInA, bool useReferenceFrameA = false)
+		: btHingeConstraint(rbA, pivotInA, axisInA, useReferenceFrameA)
 	{
-		m_accumulatedAngle=getHingeAngle();
-	}
-	
-	btHingeAccumulatedAngleConstraint(btRigidBody& rbA,btRigidBody& rbB, const btTransform& rbAFrame, const btTransform& rbBFrame, bool useReferenceFrameA = false)
-	:btHingeConstraint(rbA,rbB, rbAFrame, rbBFrame, useReferenceFrameA )
-	{
-		m_accumulatedAngle=getHingeAngle();
+		m_accumulatedAngle = getHingeAngle();
 	}
 
-	btHingeAccumulatedAngleConstraint(btRigidBody& rbA,const btTransform& rbAFrame, bool useReferenceFrameA = false)
-	:btHingeConstraint(rbA,rbAFrame, useReferenceFrameA )
+	btHingeAccumulatedAngleConstraint(btRigidBody & rbA, btRigidBody & rbB, const btTransform& rbAFrame, const btTransform& rbBFrame, bool useReferenceFrameA = false)
+		: btHingeConstraint(rbA, rbB, rbAFrame, rbBFrame, useReferenceFrameA)
 	{
-		m_accumulatedAngle=getHingeAngle();
+		m_accumulatedAngle = getHingeAngle();
+	}
+
+	btHingeAccumulatedAngleConstraint(btRigidBody & rbA, const btTransform& rbAFrame, bool useReferenceFrameA = false)
+		: btHingeConstraint(rbA, rbAFrame, useReferenceFrameA)
+	{
+		m_accumulatedAngle = getHingeAngle();
 	}
 	btScalar getAccumulatedHingeAngle();
-	void	setAccumulatedHingeAngle(btScalar accAngle);
-	virtual void getInfo1 (btConstraintInfo1* info);
-
+	void setAccumulatedHingeAngle(btScalar accAngle);
+	virtual void getInfo1(btConstraintInfo1 * info);
 };
 
-struct	btHingeConstraintFloatData
+struct btHingeConstraintFloatData
 {
-	btTypedConstraintData	m_typeConstraintData;
-	btTransformFloatData m_rbAFrame; // constraint axii. Assumes z is hinge axis.
+	btTypedConstraintData m_typeConstraintData;
+	btTransformFloatData m_rbAFrame;  // constraint axii. Assumes z is hinge axis.
 	btTransformFloatData m_rbBFrame;
-	int			m_useReferenceFrameA;
-	int			m_angularOnly;
-	
-	int			m_enableAngularMotor;
-	float	m_motorTargetVelocity;
-	float	m_maxMotorImpulse;
+	int m_useReferenceFrameA;
+	int m_angularOnly;
 
-	float	m_lowerLimit;
-	float	m_upperLimit;
-	float	m_limitSoftness;
-	float	m_biasFactor;
-	float	m_relaxationFactor;
+	int m_enableAngularMotor;
+	float m_motorTargetVelocity;
+	float m_maxMotorImpulse;
 
+	float m_lowerLimit;
+	float m_upperLimit;
+	float m_limitSoftness;
+	float m_biasFactor;
+	float m_relaxationFactor;
 };
 
-
-
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
-struct	btHingeConstraintDoubleData2
+struct btHingeConstraintDoubleData2
 {
-	btTypedConstraintDoubleData	m_typeConstraintData;
-	btTransformDoubleData m_rbAFrame; // constraint axii. Assumes z is hinge axis.
+	btTypedConstraintDoubleData m_typeConstraintData;
+	btTransformDoubleData m_rbAFrame;  // constraint axii. Assumes z is hinge axis.
 	btTransformDoubleData m_rbBFrame;
-	int			m_useReferenceFrameA;
-	int			m_angularOnly;
-	int			m_enableAngularMotor;
-	double		m_motorTargetVelocity;
-	double		m_maxMotorImpulse;
-
-	double		m_lowerLimit;
-	double		m_upperLimit;
-	double		m_limitSoftness;
-	double		m_biasFactor;
-	double		m_relaxationFactor;
-	char	m_padding1[4];
+	int m_useReferenceFrameA;
+	int m_angularOnly;
+	int m_enableAngularMotor;
+	double m_motorTargetVelocity;
+	double m_maxMotorImpulse;
 
+	double m_lowerLimit;
+	double m_upperLimit;
+	double m_limitSoftness;
+	double m_biasFactor;
+	double m_relaxationFactor;
+	char m_padding1[4];
 };
 
-
-
-
-SIMD_FORCE_INLINE	int	btHingeConstraint::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btHingeConstraint::calculateSerializeBufferSize() const
 {
 	return sizeof(btHingeConstraintData);
 }
 
-	///fills the dataBuffer and returns the struct name (and 0 on failure)
-SIMD_FORCE_INLINE	const char*	btHingeConstraint::serialize(void* dataBuffer, btSerializer* serializer) const
+///fills the dataBuffer and returns the struct name (and 0 on failure)
+SIMD_FORCE_INLINE const char* btHingeConstraint::serialize(void* dataBuffer, btSerializer* serializer) const
 {
 	btHingeConstraintData* hingeData = (btHingeConstraintData*)dataBuffer;
-	btTypedConstraint::serialize(&hingeData->m_typeConstraintData,serializer);
+	btTypedConstraint::serialize(&hingeData->m_typeConstraintData, serializer);
 
 	m_rbAFrame.serialize(hingeData->m_rbAFrame);
 	m_rbBFrame.serialize(hingeData->m_rbBFrame);
@@ -475,7 +452,7 @@ SIMD_FORCE_INLINE	const char*	btHingeConstraint::serialize(void* dataBuffer, btS
 	hingeData->m_maxMotorImpulse = float(m_maxMotorImpulse);
 	hingeData->m_motorTargetVelocity = float(m_motorTargetVelocity);
 	hingeData->m_useReferenceFrameA = m_useReferenceFrameA;
-#ifdef	_BT_USE_CENTER_LIMIT_
+#ifdef _BT_USE_CENTER_LIMIT_
 	hingeData->m_lowerLimit = float(m_limit.getLow());
 	hingeData->m_upperLimit = float(m_limit.getHigh());
 	hingeData->m_limitSoftness = float(m_limit.getSoftness());
@@ -500,4 +477,4 @@ SIMD_FORCE_INLINE	const char*	btHingeConstraint::serialize(void* dataBuffer, btS
 	return btHingeConstraintDataName;
 }
 
-#endif //BT_HINGECONSTRAINT_H
+#endif  //BT_HINGECONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btJacobianEntry.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btJacobianEntry.h
index 125580d1998..438456fe512 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btJacobianEntry.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btJacobianEntry.h
@@ -18,7 +18,6 @@ subject to the following restrictions:
 
 #include "LinearMath/btMatrix3x3.h"
 
-
 //notes:
 // Another memory optimization would be to store m_1MinvJt in the remaining 3 w components
 // which makes the btJacobianEntry memory layout 16 bytes
@@ -27,25 +26,26 @@ subject to the following restrictions:
 /// Jacobian entry is an abstraction that allows to describe constraints
 /// it can be used in combination with a constraint solver
 /// Can be used to relate the effect of an impulse to the constraint error
-ATTRIBUTE_ALIGNED16(class) btJacobianEntry
+ATTRIBUTE_ALIGNED16(class)
+btJacobianEntry
 {
 public:
-	btJacobianEntry() {};
+	btJacobianEntry(){};
 	//constraint between two different rigidbodies
 	btJacobianEntry(
 		const btMatrix3x3& world2A,
 		const btMatrix3x3& world2B,
-		const btVector3& rel_pos1,const btVector3& rel_pos2,
+		const btVector3& rel_pos1, const btVector3& rel_pos2,
 		const btVector3& jointAxis,
-		const btVector3& inertiaInvA, 
+		const btVector3& inertiaInvA,
 		const btScalar massInvA,
 		const btVector3& inertiaInvB,
 		const btScalar massInvB)
-		:m_linearJointAxis(jointAxis)
+		: m_linearJointAxis(jointAxis)
 	{
-		m_aJ = world2A*(rel_pos1.cross(m_linearJointAxis));
-		m_bJ = world2B*(rel_pos2.cross(-m_linearJointAxis));
-		m_0MinvJt	= inertiaInvA * m_aJ;
+		m_aJ = world2A * (rel_pos1.cross(m_linearJointAxis));
+		m_bJ = world2B * (rel_pos2.cross(-m_linearJointAxis));
+		m_0MinvJt = inertiaInvA * m_aJ;
 		m_1MinvJt = inertiaInvB * m_bJ;
 		m_Adiag = massInvA + m_0MinvJt.dot(m_aJ) + massInvB + m_1MinvJt.dot(m_bJ);
 
@@ -54,33 +54,31 @@ public:
 
 	//angular constraint between two different rigidbodies
 	btJacobianEntry(const btVector3& jointAxis,
-		const btMatrix3x3& world2A,
-		const btMatrix3x3& world2B,
-		const btVector3& inertiaInvA,
-		const btVector3& inertiaInvB)
-		:m_linearJointAxis(btVector3(btScalar(0.),btScalar(0.),btScalar(0.)))
+					const btMatrix3x3& world2A,
+					const btMatrix3x3& world2B,
+					const btVector3& inertiaInvA,
+					const btVector3& inertiaInvB)
+		: m_linearJointAxis(btVector3(btScalar(0.), btScalar(0.), btScalar(0.)))
 	{
-		m_aJ= world2A*jointAxis;
-		m_bJ = world2B*-jointAxis;
-		m_0MinvJt	= inertiaInvA * m_aJ;
+		m_aJ = world2A * jointAxis;
+		m_bJ = world2B * -jointAxis;
+		m_0MinvJt = inertiaInvA * m_aJ;
 		m_1MinvJt = inertiaInvB * m_bJ;
-		m_Adiag =  m_0MinvJt.dot(m_aJ) + m_1MinvJt.dot(m_bJ);
+		m_Adiag = m_0MinvJt.dot(m_aJ) + m_1MinvJt.dot(m_bJ);
 
 		btAssert(m_Adiag > btScalar(0.0));
 	}
 
 	//angular constraint between two different rigidbodies
 	btJacobianEntry(const btVector3& axisInA,
-		const btVector3& axisInB,
-		const btVector3& inertiaInvA,
-		const btVector3& inertiaInvB)
-		: m_linearJointAxis(btVector3(btScalar(0.),btScalar(0.),btScalar(0.)))
-		, m_aJ(axisInA)
-		, m_bJ(-axisInB)
+					const btVector3& axisInB,
+					const btVector3& inertiaInvA,
+					const btVector3& inertiaInvB)
+		: m_linearJointAxis(btVector3(btScalar(0.), btScalar(0.), btScalar(0.))), m_aJ(axisInA), m_bJ(-axisInB)
 	{
-		m_0MinvJt	= inertiaInvA * m_aJ;
+		m_0MinvJt = inertiaInvA * m_aJ;
 		m_1MinvJt = inertiaInvB * m_bJ;
-		m_Adiag =  m_0MinvJt.dot(m_aJ) + m_1MinvJt.dot(m_bJ);
+		m_Adiag = m_0MinvJt.dot(m_aJ) + m_1MinvJt.dot(m_bJ);
 
 		btAssert(m_Adiag > btScalar(0.0));
 	}
@@ -88,25 +86,25 @@ public:
 	//constraint on one rigidbody
 	btJacobianEntry(
 		const btMatrix3x3& world2A,
-		const btVector3& rel_pos1,const btVector3& rel_pos2,
+		const btVector3& rel_pos1, const btVector3& rel_pos2,
 		const btVector3& jointAxis,
-		const btVector3& inertiaInvA, 
+		const btVector3& inertiaInvA,
 		const btScalar massInvA)
-		:m_linearJointAxis(jointAxis)
+		: m_linearJointAxis(jointAxis)
 	{
-		m_aJ= world2A*(rel_pos1.cross(jointAxis));
-		m_bJ = world2A*(rel_pos2.cross(-jointAxis));
-		m_0MinvJt	= inertiaInvA * m_aJ;
-		m_1MinvJt = btVector3(btScalar(0.),btScalar(0.),btScalar(0.));
+		m_aJ = world2A * (rel_pos1.cross(jointAxis));
+		m_bJ = world2A * (rel_pos2.cross(-jointAxis));
+		m_0MinvJt = inertiaInvA * m_aJ;
+		m_1MinvJt = btVector3(btScalar(0.), btScalar(0.), btScalar(0.));
 		m_Adiag = massInvA + m_0MinvJt.dot(m_aJ);
 
 		btAssert(m_Adiag > btScalar(0.0));
 	}
 
-	btScalar	getDiagonal() const { return m_Adiag; }
+	btScalar getDiagonal() const { return m_Adiag; }
 
 	// for two constraints on the same rigidbody (for example vehicle friction)
-	btScalar	getNonDiagonal(const btJacobianEntry& jacB, const btScalar massInvA) const
+	btScalar getNonDiagonal(const btJacobianEntry& jacB, const btScalar massInvA) const
 	{
 		const btJacobianEntry& jacA = *this;
 		btScalar lin = massInvA * jacA.m_linearJointAxis.dot(jacB.m_linearJointAxis);
@@ -114,42 +112,39 @@ public:
 		return lin + ang;
 	}
 
-	
-
 	// for two constraints on sharing two same rigidbodies (for example two contact points between two rigidbodies)
-	btScalar	getNonDiagonal(const btJacobianEntry& jacB,const btScalar massInvA,const btScalar massInvB) const
+	btScalar getNonDiagonal(const btJacobianEntry& jacB, const btScalar massInvA, const btScalar massInvB) const
 	{
 		const btJacobianEntry& jacA = *this;
 		btVector3 lin = jacA.m_linearJointAxis * jacB.m_linearJointAxis;
 		btVector3 ang0 = jacA.m_0MinvJt * jacB.m_aJ;
 		btVector3 ang1 = jacA.m_1MinvJt * jacB.m_bJ;
-		btVector3 lin0 = massInvA * lin ;
+		btVector3 lin0 = massInvA * lin;
 		btVector3 lin1 = massInvB * lin;
-		btVector3 sum = ang0+ang1+lin0+lin1;
-		return sum[0]+sum[1]+sum[2];
+		btVector3 sum = ang0 + ang1 + lin0 + lin1;
+		return sum[0] + sum[1] + sum[2];
 	}
 
-	btScalar getRelativeVelocity(const btVector3& linvelA,const btVector3& angvelA,const btVector3& linvelB,const btVector3& angvelB)
+	btScalar getRelativeVelocity(const btVector3& linvelA, const btVector3& angvelA, const btVector3& linvelB, const btVector3& angvelB)
 	{
 		btVector3 linrel = linvelA - linvelB;
-		btVector3 angvela  = angvelA * m_aJ;
-		btVector3 angvelb  = angvelB * m_bJ;
+		btVector3 angvela = angvelA * m_aJ;
+		btVector3 angvelb = angvelB * m_bJ;
 		linrel *= m_linearJointAxis;
 		angvela += angvelb;
 		angvela += linrel;
-		btScalar rel_vel2 = angvela[0]+angvela[1]+angvela[2];
+		btScalar rel_vel2 = angvela[0] + angvela[1] + angvela[2];
 		return rel_vel2 + SIMD_EPSILON;
 	}
-//private:
+	//private:
 
-	btVector3	m_linearJointAxis;
-	btVector3	m_aJ;
-	btVector3	m_bJ;
-	btVector3	m_0MinvJt;
-	btVector3	m_1MinvJt;
+	btVector3 m_linearJointAxis;
+	btVector3 m_aJ;
+	btVector3 m_bJ;
+	btVector3 m_0MinvJt;
+	btVector3 m_1MinvJt;
 	//Optimization: can be stored in the w/last component of one of the vectors
-	btScalar	m_Adiag;
-
+	btScalar m_Adiag;
 };
 
-#endif //BT_JACOBIAN_ENTRY_H
+#endif  //BT_JACOBIAN_ENTRY_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btNNCGConstraintSolver.cpp b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btNNCGConstraintSolver.cpp
index f3979be358a..ccf89160491 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btNNCGConstraintSolver.cpp
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btNNCGConstraintSolver.cpp
@@ -15,14 +15,9 @@ subject to the following restrictions:
 
 #include "btNNCGConstraintSolver.h"
 
-
-
-
-
-
-btScalar btNNCGConstraintSolver::solveGroupCacheFriendlySetup(btCollisionObject** bodies,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer)
+btScalar btNNCGConstraintSolver::solveGroupCacheFriendlySetup(btCollisionObject** bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer)
 {
-	btScalar val = btSequentialImpulseConstraintSolver::solveGroupCacheFriendlySetup( bodies,numBodies,manifoldPtr, numManifolds, constraints,numConstraints,infoGlobal,debugDrawer);
+	btScalar val = btSequentialImpulseConstraintSolver::solveGroupCacheFriendlySetup(bodies, numBodies, manifoldPtr, numManifolds, constraints, numConstraints, infoGlobal, debugDrawer);
 
 	m_pNC.resizeNoInitialize(m_tmpSolverNonContactConstraintPool.size());
 	m_pC.resizeNoInitialize(m_tmpSolverContactConstraintPool.size());
@@ -37,38 +32,39 @@ btScalar btNNCGConstraintSolver::solveGroupCacheFriendlySetup(btCollisionObject*
 	return val;
 }
 
-btScalar btNNCGConstraintSolver::solveSingleIteration(int iteration, btCollisionObject** /*bodies */,int /*numBodies*/,btPersistentManifold** /*manifoldPtr*/, int /*numManifolds*/,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* /*debugDrawer*/)
+btScalar btNNCGConstraintSolver::solveSingleIteration(int iteration, btCollisionObject** /*bodies */, int /*numBodies*/, btPersistentManifold** /*manifoldPtr*/, int /*numManifolds*/, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* /*debugDrawer*/)
 {
-
 	int numNonContactPool = m_tmpSolverNonContactConstraintPool.size();
 	int numConstraintPool = m_tmpSolverContactConstraintPool.size();
 	int numFrictionPool = m_tmpSolverContactFrictionConstraintPool.size();
 
 	if (infoGlobal.m_solverMode & SOLVER_RANDMIZE_ORDER)
 	{
-		if (1)			// uncomment this for a bit less random ((iteration & 7) == 0)
+		if (1)  // uncomment this for a bit less random ((iteration & 7) == 0)
 		{
-
-			for (int j=0; j<numNonContactPool; ++j) {
+			for (int j = 0; j < numNonContactPool; ++j)
+			{
 				int tmp = m_orderNonContactConstraintPool[j];
-				int swapi = btRandInt2(j+1);
+				int swapi = btRandInt2(j + 1);
 				m_orderNonContactConstraintPool[j] = m_orderNonContactConstraintPool[swapi];
 				m_orderNonContactConstraintPool[swapi] = tmp;
 			}
 
-			//contact/friction constraints are not solved more than 
-			if (iteration< infoGlobal.m_numIterations)
+			//contact/friction constraints are not solved more than
+			if (iteration < infoGlobal.m_numIterations)
 			{
-				for (int j=0; j<numConstraintPool; ++j) {
+				for (int j = 0; j < numConstraintPool; ++j)
+				{
 					int tmp = m_orderTmpConstraintPool[j];
-					int swapi = btRandInt2(j+1);
+					int swapi = btRandInt2(j + 1);
 					m_orderTmpConstraintPool[j] = m_orderTmpConstraintPool[swapi];
 					m_orderTmpConstraintPool[swapi] = tmp;
 				}
 
-				for (int j=0; j<numFrictionPool; ++j) {
+				for (int j = 0; j < numFrictionPool; ++j)
+				{
 					int tmp = m_orderFrictionConstraintPool[j];
-					int swapi = btRandInt2(j+1);
+					int swapi = btRandInt2(j + 1);
 					m_orderFrictionConstraintPool[j] = m_orderFrictionConstraintPool[swapi];
 					m_orderFrictionConstraintPool[swapi] = tmp;
 				}
@@ -76,39 +72,40 @@ btScalar btNNCGConstraintSolver::solveSingleIteration(int iteration, btCollision
 		}
 	}
 
-
 	btScalar deltaflengthsqr = 0;
 	{
-		for (int j=0;j<m_tmpSolverNonContactConstraintPool.size();j++)
+		for (int j = 0; j < m_tmpSolverNonContactConstraintPool.size(); j++)
 		{
 			btSolverConstraint& constraint = m_tmpSolverNonContactConstraintPool[m_orderNonContactConstraintPool[j]];
-			if (iteration < constraint.m_overrideNumSolverIterations) 
+			if (iteration < constraint.m_overrideNumSolverIterations)
 			{
-				btScalar deltaf = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[constraint.m_solverBodyIdA],m_tmpSolverBodyPool[constraint.m_solverBodyIdB],constraint);
+				btScalar deltaf = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[constraint.m_solverBodyIdA], m_tmpSolverBodyPool[constraint.m_solverBodyIdB], constraint);
 				m_deltafNC[j] = deltaf;
 				deltaflengthsqr += deltaf * deltaf;
 			}
 		}
 	}
 
-
-	if (m_onlyForNoneContact) 
+	if (m_onlyForNoneContact)
 	{
-		if (iteration==0) 
+		if (iteration == 0)
+		{
+			for (int j = 0; j < m_tmpSolverNonContactConstraintPool.size(); j++) m_pNC[j] = m_deltafNC[j];
+		}
+		else
 		{
-			for (int j=0;j<m_tmpSolverNonContactConstraintPool.size();j++) m_pNC[j] = m_deltafNC[j];
-		} else {
 			// deltaflengthsqrprev can be 0 only if the solver solved the problem exactly in the previous iteration. In this case we should have quit, but mainly for debug reason with this 'hack' it is now allowed to continue the calculation
-			btScalar beta = m_deltafLengthSqrPrev>0 ? deltaflengthsqr / m_deltafLengthSqrPrev : 2;
-			if (beta>1) 
+			btScalar beta = m_deltafLengthSqrPrev > 0 ? deltaflengthsqr / m_deltafLengthSqrPrev : 2;
+			if (beta > 1)
 			{
-				for (int j=0;j<m_tmpSolverNonContactConstraintPool.size();j++) m_pNC[j] = 0;
-			} else 
+				for (int j = 0; j < m_tmpSolverNonContactConstraintPool.size(); j++) m_pNC[j] = 0;
+			}
+			else
 			{
-				for (int j=0;j<m_tmpSolverNonContactConstraintPool.size();j++)
+				for (int j = 0; j < m_tmpSolverNonContactConstraintPool.size(); j++)
 				{
 					btSolverConstraint& constraint = m_tmpSolverNonContactConstraintPool[m_orderNonContactConstraintPool[j]];
-					if (iteration < constraint.m_overrideNumSolverIterations) 
+					if (iteration < constraint.m_overrideNumSolverIterations)
 					{
 						btScalar additionaldeltaimpulse = beta * m_pNC[j];
 						constraint.m_appliedImpulse = btScalar(constraint.m_appliedImpulse) + additionaldeltaimpulse;
@@ -116,8 +113,8 @@ btScalar btNNCGConstraintSolver::solveSingleIteration(int iteration, btCollision
 						btSolverBody& body1 = m_tmpSolverBodyPool[constraint.m_solverBodyIdA];
 						btSolverBody& body2 = m_tmpSolverBodyPool[constraint.m_solverBodyIdB];
 						const btSolverConstraint& c = constraint;
-						body1.internalApplyImpulse(c.m_contactNormal1*body1.internalGetInvMass(),c.m_angularComponentA,additionaldeltaimpulse);
-						body2.internalApplyImpulse(c.m_contactNormal2*body2.internalGetInvMass(),c.m_angularComponentB,additionaldeltaimpulse);
+						body1.internalApplyImpulse(c.m_contactNormal1 * body1.internalGetInvMass(), c.m_angularComponentA, additionaldeltaimpulse);
+						body2.internalApplyImpulse(c.m_contactNormal2 * body2.internalGetInvMass(), c.m_angularComponentB, additionaldeltaimpulse);
 					}
 				}
 			}
@@ -125,21 +122,18 @@ btScalar btNNCGConstraintSolver::solveSingleIteration(int iteration, btCollision
 		m_deltafLengthSqrPrev = deltaflengthsqr;
 	}
 
-
-
 	{
-
-		if (iteration< infoGlobal.m_numIterations)
+		if (iteration < infoGlobal.m_numIterations)
 		{
-			for (int j=0;j<numConstraints;j++)
+			for (int j = 0; j < numConstraints; j++)
 			{
 				if (constraints[j]->isEnabled())
 				{
-					int bodyAid = getOrInitSolverBody(constraints[j]->getRigidBodyA(),infoGlobal.m_timeStep);
-					int bodyBid = getOrInitSolverBody(constraints[j]->getRigidBodyB(),infoGlobal.m_timeStep);
+					int bodyAid = getOrInitSolverBody(constraints[j]->getRigidBodyA(), infoGlobal.m_timeStep);
+					int bodyBid = getOrInitSolverBody(constraints[j]->getRigidBodyB(), infoGlobal.m_timeStep);
 					btSolverBody& bodyA = m_tmpSolverBodyPool[bodyAid];
 					btSolverBody& bodyB = m_tmpSolverBodyPool[bodyBid];
-					constraints[j]->solveConstraintObsolete(bodyA,bodyB,infoGlobal.m_timeStep);
+					constraints[j]->solveConstraintObsolete(bodyA, bodyB, infoGlobal.m_timeStep);
 				}
 			}
 
@@ -147,203 +141,206 @@ btScalar btNNCGConstraintSolver::solveSingleIteration(int iteration, btCollision
 			if (infoGlobal.m_solverMode & SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS)
 			{
 				int numPoolConstraints = m_tmpSolverContactConstraintPool.size();
-				int multiplier = (infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS)? 2 : 1;
+				int multiplier = (infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS) ? 2 : 1;
 
-				for (int c=0;c<numPoolConstraints;c++)
+				for (int c = 0; c < numPoolConstraints; c++)
 				{
-					btScalar totalImpulse =0;
+					btScalar totalImpulse = 0;
 
 					{
 						const btSolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[m_orderTmpConstraintPool[c]];
-						btScalar deltaf = resolveSingleConstraintRowLowerLimit(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA],m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB],solveManifold);
+						btScalar deltaf = resolveSingleConstraintRowLowerLimit(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
 						m_deltafC[c] = deltaf;
-						deltaflengthsqr += deltaf*deltaf;
+						deltaflengthsqr += deltaf * deltaf;
 						totalImpulse = solveManifold.m_appliedImpulse;
 					}
 					bool applyFriction = true;
 					if (applyFriction)
 					{
 						{
+							btSolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[m_orderFrictionConstraintPool[c * multiplier]];
 
-							btSolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[m_orderFrictionConstraintPool[c*multiplier]];
-
-							if (totalImpulse>btScalar(0))
+							if (totalImpulse > btScalar(0))
 							{
-								solveManifold.m_lowerLimit = -(solveManifold.m_friction*totalImpulse);
-								solveManifold.m_upperLimit = solveManifold.m_friction*totalImpulse;
-								btScalar deltaf = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA],m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB],solveManifold);
-								m_deltafCF[c*multiplier] = deltaf;
-								deltaflengthsqr += deltaf*deltaf;
-							} else {
-								m_deltafCF[c*multiplier] = 0;
+								solveManifold.m_lowerLimit = -(solveManifold.m_friction * totalImpulse);
+								solveManifold.m_upperLimit = solveManifold.m_friction * totalImpulse;
+								btScalar deltaf = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
+								m_deltafCF[c * multiplier] = deltaf;
+								deltaflengthsqr += deltaf * deltaf;
+							}
+							else
+							{
+								m_deltafCF[c * multiplier] = 0;
 							}
 						}
 
 						if (infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS)
 						{
+							btSolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[m_orderFrictionConstraintPool[c * multiplier + 1]];
 
-							btSolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[m_orderFrictionConstraintPool[c*multiplier+1]];
-
-							if (totalImpulse>btScalar(0))
+							if (totalImpulse > btScalar(0))
 							{
-								solveManifold.m_lowerLimit = -(solveManifold.m_friction*totalImpulse);
-								solveManifold.m_upperLimit = solveManifold.m_friction*totalImpulse;
-								btScalar deltaf = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA],m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB],solveManifold);
-								m_deltafCF[c*multiplier+1] = deltaf;
-								deltaflengthsqr += deltaf*deltaf;
-							} else {
-								m_deltafCF[c*multiplier+1] = 0;
+								solveManifold.m_lowerLimit = -(solveManifold.m_friction * totalImpulse);
+								solveManifold.m_upperLimit = solveManifold.m_friction * totalImpulse;
+								btScalar deltaf = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
+								m_deltafCF[c * multiplier + 1] = deltaf;
+								deltaflengthsqr += deltaf * deltaf;
+							}
+							else
+							{
+								m_deltafCF[c * multiplier + 1] = 0;
 							}
 						}
 					}
 				}
-
 			}
-			else//SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS
+			else  //SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS
 			{
 				//solve the friction constraints after all contact constraints, don't interleave them
 				int numPoolConstraints = m_tmpSolverContactConstraintPool.size();
 				int j;
 
-				for (j=0;j<numPoolConstraints;j++)
+				for (j = 0; j < numPoolConstraints; j++)
 				{
 					const btSolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[m_orderTmpConstraintPool[j]];
-					btScalar deltaf = resolveSingleConstraintRowLowerLimit(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA],m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB],solveManifold);
+					btScalar deltaf = resolveSingleConstraintRowLowerLimit(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
 					m_deltafC[j] = deltaf;
-					deltaflengthsqr += deltaf*deltaf;
+					deltaflengthsqr += deltaf * deltaf;
 				}
 
-
-
 				///solve all friction constraints
 
 				int numFrictionPoolConstraints = m_tmpSolverContactFrictionConstraintPool.size();
-				for (j=0;j<numFrictionPoolConstraints;j++)
+				for (j = 0; j < numFrictionPoolConstraints; j++)
 				{
 					btSolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[m_orderFrictionConstraintPool[j]];
 					btScalar totalImpulse = m_tmpSolverContactConstraintPool[solveManifold.m_frictionIndex].m_appliedImpulse;
 
-					if (totalImpulse>btScalar(0))
+					if (totalImpulse > btScalar(0))
 					{
-						solveManifold.m_lowerLimit = -(solveManifold.m_friction*totalImpulse);
-						solveManifold.m_upperLimit = solveManifold.m_friction*totalImpulse;
+						solveManifold.m_lowerLimit = -(solveManifold.m_friction * totalImpulse);
+						solveManifold.m_upperLimit = solveManifold.m_friction * totalImpulse;
 
-						btScalar deltaf = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA],m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB],solveManifold);
+						btScalar deltaf = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
 						m_deltafCF[j] = deltaf;
-						deltaflengthsqr += deltaf*deltaf;
-					} else {
+						deltaflengthsqr += deltaf * deltaf;
+					}
+					else
+					{
 						m_deltafCF[j] = 0;
 					}
 				}
 			}
 
-            {
+			{
 				int numRollingFrictionPoolConstraints = m_tmpSolverContactRollingFrictionConstraintPool.size();
-				for (int j=0;j<numRollingFrictionPoolConstraints;j++)
+				for (int j = 0; j < numRollingFrictionPoolConstraints; j++)
 				{
-
 					btSolverConstraint& rollingFrictionConstraint = m_tmpSolverContactRollingFrictionConstraintPool[j];
 					btScalar totalImpulse = m_tmpSolverContactConstraintPool[rollingFrictionConstraint.m_frictionIndex].m_appliedImpulse;
-					if (totalImpulse>btScalar(0))
+					if (totalImpulse > btScalar(0))
 					{
-						btScalar rollingFrictionMagnitude = rollingFrictionConstraint.m_friction*totalImpulse;
-						if (rollingFrictionMagnitude>rollingFrictionConstraint.m_friction)
+						btScalar rollingFrictionMagnitude = rollingFrictionConstraint.m_friction * totalImpulse;
+						if (rollingFrictionMagnitude > rollingFrictionConstraint.m_friction)
 							rollingFrictionMagnitude = rollingFrictionConstraint.m_friction;
 
 						rollingFrictionConstraint.m_lowerLimit = -rollingFrictionMagnitude;
 						rollingFrictionConstraint.m_upperLimit = rollingFrictionMagnitude;
 
-						btScalar deltaf = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdA],m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdB],rollingFrictionConstraint);
+						btScalar deltaf = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdA], m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdB], rollingFrictionConstraint);
 						m_deltafCRF[j] = deltaf;
-						deltaflengthsqr += deltaf*deltaf;
-					} else {
+						deltaflengthsqr += deltaf * deltaf;
+					}
+					else
+					{
 						m_deltafCRF[j] = 0;
 					}
 				}
-            }
-
+			}
 		}
-
-
-
 	}
 
-
-
-
-	if (!m_onlyForNoneContact) 
+	if (!m_onlyForNoneContact)
 	{
-		if (iteration==0) 
+		if (iteration == 0)
 		{
-			for (int j=0;j<m_tmpSolverNonContactConstraintPool.size();j++) m_pNC[j] = m_deltafNC[j];
-			for (int j=0;j<m_tmpSolverContactConstraintPool.size();j++) m_pC[j] = m_deltafC[j];
-			for (int j=0;j<m_tmpSolverContactFrictionConstraintPool.size();j++) m_pCF[j] = m_deltafCF[j];
-			for (int j=0;j<m_tmpSolverContactRollingFrictionConstraintPool.size();j++) m_pCRF[j] = m_deltafCRF[j];
-		} else 
+			for (int j = 0; j < m_tmpSolverNonContactConstraintPool.size(); j++) m_pNC[j] = m_deltafNC[j];
+			for (int j = 0; j < m_tmpSolverContactConstraintPool.size(); j++) m_pC[j] = m_deltafC[j];
+			for (int j = 0; j < m_tmpSolverContactFrictionConstraintPool.size(); j++) m_pCF[j] = m_deltafCF[j];
+			for (int j = 0; j < m_tmpSolverContactRollingFrictionConstraintPool.size(); j++) m_pCRF[j] = m_deltafCRF[j];
+		}
+		else
 		{
 			// deltaflengthsqrprev can be 0 only if the solver solved the problem exactly in the previous iteration. In this case we should have quit, but mainly for debug reason with this 'hack' it is now allowed to continue the calculation
-			btScalar beta = m_deltafLengthSqrPrev>0 ? deltaflengthsqr / m_deltafLengthSqrPrev : 2;
-			if (beta>1) {
-				for (int j=0;j<m_tmpSolverNonContactConstraintPool.size();j++) m_pNC[j] = 0;
-				for (int j=0;j<m_tmpSolverContactConstraintPool.size();j++) m_pC[j] = 0;
-				for (int j=0;j<m_tmpSolverContactFrictionConstraintPool.size();j++) m_pCF[j] = 0;
-				for (int j=0;j<m_tmpSolverContactRollingFrictionConstraintPool.size();j++) m_pCRF[j] = 0;
-			} else {
-				for (int j=0;j<m_tmpSolverNonContactConstraintPool.size();j++)
+			btScalar beta = m_deltafLengthSqrPrev > 0 ? deltaflengthsqr / m_deltafLengthSqrPrev : 2;
+			if (beta > 1)
+			{
+				for (int j = 0; j < m_tmpSolverNonContactConstraintPool.size(); j++) m_pNC[j] = 0;
+				for (int j = 0; j < m_tmpSolverContactConstraintPool.size(); j++) m_pC[j] = 0;
+				for (int j = 0; j < m_tmpSolverContactFrictionConstraintPool.size(); j++) m_pCF[j] = 0;
+				for (int j = 0; j < m_tmpSolverContactRollingFrictionConstraintPool.size(); j++) m_pCRF[j] = 0;
+			}
+			else
+			{
+				for (int j = 0; j < m_tmpSolverNonContactConstraintPool.size(); j++)
 				{
 					btSolverConstraint& constraint = m_tmpSolverNonContactConstraintPool[m_orderNonContactConstraintPool[j]];
-					if (iteration < constraint.m_overrideNumSolverIterations) {
+					if (iteration < constraint.m_overrideNumSolverIterations)
+					{
 						btScalar additionaldeltaimpulse = beta * m_pNC[j];
 						constraint.m_appliedImpulse = btScalar(constraint.m_appliedImpulse) + additionaldeltaimpulse;
 						m_pNC[j] = beta * m_pNC[j] + m_deltafNC[j];
 						btSolverBody& body1 = m_tmpSolverBodyPool[constraint.m_solverBodyIdA];
 						btSolverBody& body2 = m_tmpSolverBodyPool[constraint.m_solverBodyIdB];
 						const btSolverConstraint& c = constraint;
-						body1.internalApplyImpulse(c.m_contactNormal1*body1.internalGetInvMass(),c.m_angularComponentA,additionaldeltaimpulse);
-						body2.internalApplyImpulse(c.m_contactNormal2*body2.internalGetInvMass(),c.m_angularComponentB,additionaldeltaimpulse);
+						body1.internalApplyImpulse(c.m_contactNormal1 * body1.internalGetInvMass(), c.m_angularComponentA, additionaldeltaimpulse);
+						body2.internalApplyImpulse(c.m_contactNormal2 * body2.internalGetInvMass(), c.m_angularComponentB, additionaldeltaimpulse);
 					}
 				}
-				for (int j=0;j<m_tmpSolverContactConstraintPool.size();j++)
+				for (int j = 0; j < m_tmpSolverContactConstraintPool.size(); j++)
 				{
 					btSolverConstraint& constraint = m_tmpSolverContactConstraintPool[m_orderTmpConstraintPool[j]];
-					if (iteration< infoGlobal.m_numIterations) {
+					if (iteration < infoGlobal.m_numIterations)
+					{
 						btScalar additionaldeltaimpulse = beta * m_pC[j];
 						constraint.m_appliedImpulse = btScalar(constraint.m_appliedImpulse) + additionaldeltaimpulse;
 						m_pC[j] = beta * m_pC[j] + m_deltafC[j];
 						btSolverBody& body1 = m_tmpSolverBodyPool[constraint.m_solverBodyIdA];
 						btSolverBody& body2 = m_tmpSolverBodyPool[constraint.m_solverBodyIdB];
 						const btSolverConstraint& c = constraint;
-						body1.internalApplyImpulse(c.m_contactNormal1*body1.internalGetInvMass(),c.m_angularComponentA,additionaldeltaimpulse);
-						body2.internalApplyImpulse(c.m_contactNormal2*body2.internalGetInvMass(),c.m_angularComponentB,additionaldeltaimpulse);
+						body1.internalApplyImpulse(c.m_contactNormal1 * body1.internalGetInvMass(), c.m_angularComponentA, additionaldeltaimpulse);
+						body2.internalApplyImpulse(c.m_contactNormal2 * body2.internalGetInvMass(), c.m_angularComponentB, additionaldeltaimpulse);
 					}
 				}
-				for (int j=0;j<m_tmpSolverContactFrictionConstraintPool.size();j++)
+				for (int j = 0; j < m_tmpSolverContactFrictionConstraintPool.size(); j++)
 				{
 					btSolverConstraint& constraint = m_tmpSolverContactFrictionConstraintPool[m_orderFrictionConstraintPool[j]];
-					if (iteration< infoGlobal.m_numIterations) {
+					if (iteration < infoGlobal.m_numIterations)
+					{
 						btScalar additionaldeltaimpulse = beta * m_pCF[j];
 						constraint.m_appliedImpulse = btScalar(constraint.m_appliedImpulse) + additionaldeltaimpulse;
 						m_pCF[j] = beta * m_pCF[j] + m_deltafCF[j];
 						btSolverBody& body1 = m_tmpSolverBodyPool[constraint.m_solverBodyIdA];
 						btSolverBody& body2 = m_tmpSolverBodyPool[constraint.m_solverBodyIdB];
 						const btSolverConstraint& c = constraint;
-						body1.internalApplyImpulse(c.m_contactNormal1*body1.internalGetInvMass(),c.m_angularComponentA,additionaldeltaimpulse);
-						body2.internalApplyImpulse(c.m_contactNormal2*body2.internalGetInvMass(),c.m_angularComponentB,additionaldeltaimpulse);
+						body1.internalApplyImpulse(c.m_contactNormal1 * body1.internalGetInvMass(), c.m_angularComponentA, additionaldeltaimpulse);
+						body2.internalApplyImpulse(c.m_contactNormal2 * body2.internalGetInvMass(), c.m_angularComponentB, additionaldeltaimpulse);
 					}
 				}
 				{
-					for (int j=0;j<m_tmpSolverContactRollingFrictionConstraintPool.size();j++)
+					for (int j = 0; j < m_tmpSolverContactRollingFrictionConstraintPool.size(); j++)
 					{
 						btSolverConstraint& constraint = m_tmpSolverContactRollingFrictionConstraintPool[j];
-						if (iteration< infoGlobal.m_numIterations) {
+						if (iteration < infoGlobal.m_numIterations)
+						{
 							btScalar additionaldeltaimpulse = beta * m_pCRF[j];
 							constraint.m_appliedImpulse = btScalar(constraint.m_appliedImpulse) + additionaldeltaimpulse;
 							m_pCRF[j] = beta * m_pCRF[j] + m_deltafCRF[j];
 							btSolverBody& body1 = m_tmpSolverBodyPool[constraint.m_solverBodyIdA];
 							btSolverBody& body2 = m_tmpSolverBodyPool[constraint.m_solverBodyIdB];
 							const btSolverConstraint& c = constraint;
-							body1.internalApplyImpulse(c.m_contactNormal1*body1.internalGetInvMass(),c.m_angularComponentA,additionaldeltaimpulse);
-							body2.internalApplyImpulse(c.m_contactNormal2*body2.internalGetInvMass(),c.m_angularComponentB,additionaldeltaimpulse);
+							body1.internalApplyImpulse(c.m_contactNormal1 * body1.internalGetInvMass(), c.m_angularComponentA, additionaldeltaimpulse);
+							body2.internalApplyImpulse(c.m_contactNormal2 * body2.internalGetInvMass(), c.m_angularComponentB, additionaldeltaimpulse);
 						}
 					}
 				}
@@ -355,7 +352,7 @@ btScalar btNNCGConstraintSolver::solveSingleIteration(int iteration, btCollision
 	return deltaflengthsqr;
 }
 
-btScalar btNNCGConstraintSolver::solveGroupCacheFriendlyFinish(btCollisionObject** bodies,int numBodies,const btContactSolverInfo& infoGlobal)
+btScalar btNNCGConstraintSolver::solveGroupCacheFriendlyFinish(btCollisionObject** bodies, int numBodies, const btContactSolverInfo& infoGlobal)
 {
 	m_pNC.resizeNoInitialize(0);
 	m_pC.resizeNoInitialize(0);
@@ -369,6 +366,3 @@ btScalar btNNCGConstraintSolver::solveGroupCacheFriendlyFinish(btCollisionObject
 
 	return btSequentialImpulseConstraintSolver::solveGroupCacheFriendlyFinish(bodies, numBodies, infoGlobal);
 }
-
-
-
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btNNCGConstraintSolver.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btNNCGConstraintSolver.h
index a300929cd5c..c84f274a991 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btNNCGConstraintSolver.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btNNCGConstraintSolver.h
@@ -18,33 +18,30 @@ subject to the following restrictions:
 
 #include "btSequentialImpulseConstraintSolver.h"
 
-ATTRIBUTE_ALIGNED16(class) btNNCGConstraintSolver : public btSequentialImpulseConstraintSolver
+ATTRIBUTE_ALIGNED16(class)
+btNNCGConstraintSolver : public btSequentialImpulseConstraintSolver
 {
 protected:
-
 	btScalar m_deltafLengthSqrPrev;
 
-	btAlignedObjectArray<btScalar> m_pNC;  // p for None Contact constraints
-	btAlignedObjectArray<btScalar> m_pC;   // p for Contact constraints
-	btAlignedObjectArray<btScalar> m_pCF;  // p for ContactFriction constraints
-	btAlignedObjectArray<btScalar> m_pCRF; // p for ContactRollingFriction constraints
+	btAlignedObjectArray<btScalar> m_pNC;   // p for None Contact constraints
+	btAlignedObjectArray<btScalar> m_pC;    // p for Contact constraints
+	btAlignedObjectArray<btScalar> m_pCF;   // p for ContactFriction constraints
+	btAlignedObjectArray<btScalar> m_pCRF;  // p for ContactRollingFriction constraints
 
 	//These are recalculated in every iterations. We just keep these to prevent reallocation in each iteration.
-	btAlignedObjectArray<btScalar> m_deltafNC;  // deltaf for NoneContact constraints
-	btAlignedObjectArray<btScalar> m_deltafC;   // deltaf for Contact constraints
-	btAlignedObjectArray<btScalar> m_deltafCF;  // deltaf for ContactFriction constraints
-	btAlignedObjectArray<btScalar> m_deltafCRF; // deltaf for ContactRollingFriction constraints
+	btAlignedObjectArray<btScalar> m_deltafNC;   // deltaf for NoneContact constraints
+	btAlignedObjectArray<btScalar> m_deltafC;    // deltaf for Contact constraints
+	btAlignedObjectArray<btScalar> m_deltafCF;   // deltaf for ContactFriction constraints
+	btAlignedObjectArray<btScalar> m_deltafCRF;  // deltaf for ContactRollingFriction constraints
 
-		
 protected:
+	virtual btScalar solveGroupCacheFriendlyFinish(btCollisionObject * *bodies, int numBodies, const btContactSolverInfo& infoGlobal);
+	virtual btScalar solveSingleIteration(int iteration, btCollisionObject** bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer);
 
-	virtual btScalar solveGroupCacheFriendlyFinish(btCollisionObject** bodies,int numBodies,const btContactSolverInfo& infoGlobal);
-	virtual btScalar solveSingleIteration(int iteration, btCollisionObject** bodies ,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer);
-
-	virtual btScalar solveGroupCacheFriendlySetup(btCollisionObject** bodies,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer);
+	virtual btScalar solveGroupCacheFriendlySetup(btCollisionObject * *bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer);
 
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
 	btNNCGConstraintSolver() : btSequentialImpulseConstraintSolver(), m_onlyForNoneContact(false) {}
@@ -57,8 +54,4 @@ public:
 	bool m_onlyForNoneContact;
 };
 
-
-
-
-#endif //BT_NNCG_CONSTRAINT_SOLVER_H
-
+#endif  //BT_NNCG_CONSTRAINT_SOLVER_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btPoint2PointConstraint.cpp b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btPoint2PointConstraint.cpp
index 3c0430b903f..ad399dc57f7 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btPoint2PointConstraint.cpp
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btPoint2PointConstraint.cpp
@@ -13,217 +13,193 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btPoint2PointConstraint.h"
 #include "BulletDynamics/Dynamics/btRigidBody.h"
 #include <new>
 
-
-
-
-
-btPoint2PointConstraint::btPoint2PointConstraint(btRigidBody& rbA,btRigidBody& rbB, const btVector3& pivotInA,const btVector3& pivotInB)
-:btTypedConstraint(POINT2POINT_CONSTRAINT_TYPE,rbA,rbB),m_pivotInA(pivotInA),m_pivotInB(pivotInB),
-m_flags(0),
-m_useSolveConstraintObsolete(false)
+btPoint2PointConstraint::btPoint2PointConstraint(btRigidBody& rbA, btRigidBody& rbB, const btVector3& pivotInA, const btVector3& pivotInB)
+	: btTypedConstraint(POINT2POINT_CONSTRAINT_TYPE, rbA, rbB), m_pivotInA(pivotInA), m_pivotInB(pivotInB), m_flags(0), m_useSolveConstraintObsolete(false)
 {
-
 }
 
-
-btPoint2PointConstraint::btPoint2PointConstraint(btRigidBody& rbA,const btVector3& pivotInA)
-:btTypedConstraint(POINT2POINT_CONSTRAINT_TYPE,rbA),m_pivotInA(pivotInA),m_pivotInB(rbA.getCenterOfMassTransform()(pivotInA)),
-m_flags(0),
-m_useSolveConstraintObsolete(false)
+btPoint2PointConstraint::btPoint2PointConstraint(btRigidBody& rbA, const btVector3& pivotInA)
+	: btTypedConstraint(POINT2POINT_CONSTRAINT_TYPE, rbA), m_pivotInA(pivotInA), m_pivotInB(rbA.getCenterOfMassTransform()(pivotInA)), m_flags(0), m_useSolveConstraintObsolete(false)
 {
-	
 }
 
-void	btPoint2PointConstraint::buildJacobian()
+void btPoint2PointConstraint::buildJacobian()
 {
-
 	///we need it for both methods
 	{
 		m_appliedImpulse = btScalar(0.);
 
-		btVector3	normal(0,0,0);
+		btVector3 normal(0, 0, 0);
 
-		for (int i=0;i<3;i++)
+		for (int i = 0; i < 3; i++)
 		{
 			normal[i] = 1;
 			new (&m_jac[i]) btJacobianEntry(
-			m_rbA.getCenterOfMassTransform().getBasis().transpose(),
-			m_rbB.getCenterOfMassTransform().getBasis().transpose(),
-			m_rbA.getCenterOfMassTransform()*m_pivotInA - m_rbA.getCenterOfMassPosition(),
-			m_rbB.getCenterOfMassTransform()*m_pivotInB - m_rbB.getCenterOfMassPosition(),
-			normal,
-			m_rbA.getInvInertiaDiagLocal(),
-			m_rbA.getInvMass(),
-			m_rbB.getInvInertiaDiagLocal(),
-			m_rbB.getInvMass());
-		normal[i] = 0;
+				m_rbA.getCenterOfMassTransform().getBasis().transpose(),
+				m_rbB.getCenterOfMassTransform().getBasis().transpose(),
+				m_rbA.getCenterOfMassTransform() * m_pivotInA - m_rbA.getCenterOfMassPosition(),
+				m_rbB.getCenterOfMassTransform() * m_pivotInB - m_rbB.getCenterOfMassPosition(),
+				normal,
+				m_rbA.getInvInertiaDiagLocal(),
+				m_rbA.getInvMass(),
+				m_rbB.getInvInertiaDiagLocal(),
+				m_rbB.getInvMass());
+			normal[i] = 0;
 		}
 	}
-
-
 }
 
-void btPoint2PointConstraint::getInfo1 (btConstraintInfo1* info)
+void btPoint2PointConstraint::getInfo1(btConstraintInfo1* info)
 {
 	getInfo1NonVirtual(info);
 }
 
-void btPoint2PointConstraint::getInfo1NonVirtual (btConstraintInfo1* info)
+void btPoint2PointConstraint::getInfo1NonVirtual(btConstraintInfo1* info)
 {
 	if (m_useSolveConstraintObsolete)
 	{
 		info->m_numConstraintRows = 0;
 		info->nub = 0;
-	} else
+	}
+	else
 	{
 		info->m_numConstraintRows = 3;
 		info->nub = 3;
 	}
 }
 
-
-
-
-void btPoint2PointConstraint::getInfo2 (btConstraintInfo2* info)
+void btPoint2PointConstraint::getInfo2(btConstraintInfo2* info)
 {
-	getInfo2NonVirtual(info, m_rbA.getCenterOfMassTransform(),m_rbB.getCenterOfMassTransform());
+	getInfo2NonVirtual(info, m_rbA.getCenterOfMassTransform(), m_rbB.getCenterOfMassTransform());
 }
 
-void btPoint2PointConstraint::getInfo2NonVirtual (btConstraintInfo2* info, const btTransform& body0_trans, const btTransform& body1_trans)
+void btPoint2PointConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTransform& body0_trans, const btTransform& body1_trans)
 {
 	btAssert(!m_useSolveConstraintObsolete);
 
-	 //retrieve matrices
+	//retrieve matrices
 
 	// anchor points in global coordinates with respect to body PORs.
-   
-    // set jacobian
-    info->m_J1linearAxis[0] = 1;
-	info->m_J1linearAxis[info->rowskip+1] = 1;
-	info->m_J1linearAxis[2*info->rowskip+2] = 1;
 
-	btVector3 a1 = body0_trans.getBasis()*getPivotInA();
+	// set jacobian
+	info->m_J1linearAxis[0] = 1;
+	info->m_J1linearAxis[info->rowskip + 1] = 1;
+	info->m_J1linearAxis[2 * info->rowskip + 2] = 1;
+
+	btVector3 a1 = body0_trans.getBasis() * getPivotInA();
 	{
 		btVector3* angular0 = (btVector3*)(info->m_J1angularAxis);
-		btVector3* angular1 = (btVector3*)(info->m_J1angularAxis+info->rowskip);
-		btVector3* angular2 = (btVector3*)(info->m_J1angularAxis+2*info->rowskip);
+		btVector3* angular1 = (btVector3*)(info->m_J1angularAxis + info->rowskip);
+		btVector3* angular2 = (btVector3*)(info->m_J1angularAxis + 2 * info->rowskip);
 		btVector3 a1neg = -a1;
-		a1neg.getSkewSymmetricMatrix(angular0,angular1,angular2);
+		a1neg.getSkewSymmetricMatrix(angular0, angular1, angular2);
 	}
-    
+
 	info->m_J2linearAxis[0] = -1;
-    info->m_J2linearAxis[info->rowskip+1] = -1;
-    info->m_J2linearAxis[2*info->rowskip+2] = -1;
-	
-	btVector3 a2 = body1_trans.getBasis()*getPivotInB();
-   
+	info->m_J2linearAxis[info->rowskip + 1] = -1;
+	info->m_J2linearAxis[2 * info->rowskip + 2] = -1;
+
+	btVector3 a2 = body1_trans.getBasis() * getPivotInB();
+
 	{
-	//	btVector3 a2n = -a2;
+		//	btVector3 a2n = -a2;
 		btVector3* angular0 = (btVector3*)(info->m_J2angularAxis);
-		btVector3* angular1 = (btVector3*)(info->m_J2angularAxis+info->rowskip);
-		btVector3* angular2 = (btVector3*)(info->m_J2angularAxis+2*info->rowskip);
-		a2.getSkewSymmetricMatrix(angular0,angular1,angular2);
+		btVector3* angular1 = (btVector3*)(info->m_J2angularAxis + info->rowskip);
+		btVector3* angular2 = (btVector3*)(info->m_J2angularAxis + 2 * info->rowskip);
+		a2.getSkewSymmetricMatrix(angular0, angular1, angular2);
 	}
-    
 
-
-    // set right hand side
+	// set right hand side
 	btScalar currERP = (m_flags & BT_P2P_FLAGS_ERP) ? m_erp : info->erp;
-    btScalar k = info->fps * currERP;
-    int j;
-	for (j=0; j<3; j++)
-    {
-        info->m_constraintError[j*info->rowskip] = k * (a2[j] + body1_trans.getOrigin()[j] - a1[j] - body0_trans.getOrigin()[j]);
-		//printf("info->m_constraintError[%d]=%f\n",j,info->m_constraintError[j]);
-    }
-	if(m_flags & BT_P2P_FLAGS_CFM)
+	btScalar k = info->fps * currERP;
+	int j;
+	for (j = 0; j < 3; j++)
 	{
-		for (j=0; j<3; j++)
+		info->m_constraintError[j * info->rowskip] = k * (a2[j] + body1_trans.getOrigin()[j] - a1[j] - body0_trans.getOrigin()[j]);
+		//printf("info->m_constraintError[%d]=%f\n",j,info->m_constraintError[j]);
+	}
+	if (m_flags & BT_P2P_FLAGS_CFM)
+	{
+		for (j = 0; j < 3; j++)
 		{
-			info->cfm[j*info->rowskip] = m_cfm;
+			info->cfm[j * info->rowskip] = m_cfm;
 		}
 	}
 
-	btScalar impulseClamp = m_setting.m_impulseClamp;//
-	for (j=0; j<3; j++)
-    {
+	btScalar impulseClamp = m_setting.m_impulseClamp;  //
+	for (j = 0; j < 3; j++)
+	{
 		if (m_setting.m_impulseClamp > 0)
 		{
-			info->m_lowerLimit[j*info->rowskip] = -impulseClamp;
-			info->m_upperLimit[j*info->rowskip] = impulseClamp;
+			info->m_lowerLimit[j * info->rowskip] = -impulseClamp;
+			info->m_upperLimit[j * info->rowskip] = impulseClamp;
 		}
 	}
 	info->m_damping = m_setting.m_damping;
-	
 }
 
-
-
-void	btPoint2PointConstraint::updateRHS(btScalar	timeStep)
+void btPoint2PointConstraint::updateRHS(btScalar timeStep)
 {
 	(void)timeStep;
-
 }
 
-///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5). 
+///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5).
 ///If no axis is provided, it uses the default axis for this constraint.
 void btPoint2PointConstraint::setParam(int num, btScalar value, int axis)
 {
-	if(axis != -1)
+	if (axis != -1)
 	{
 		btAssertConstrParams(0);
 	}
 	else
 	{
-		switch(num)
+		switch (num)
 		{
-			case BT_CONSTRAINT_ERP :
-			case BT_CONSTRAINT_STOP_ERP :
-				m_erp = value; 
+			case BT_CONSTRAINT_ERP:
+			case BT_CONSTRAINT_STOP_ERP:
+				m_erp = value;
 				m_flags |= BT_P2P_FLAGS_ERP;
 				break;
-			case BT_CONSTRAINT_CFM :
-			case BT_CONSTRAINT_STOP_CFM :
-				m_cfm = value; 
+			case BT_CONSTRAINT_CFM:
+			case BT_CONSTRAINT_STOP_CFM:
+				m_cfm = value;
 				m_flags |= BT_P2P_FLAGS_CFM;
 				break;
-			default: 
+			default:
 				btAssertConstrParams(0);
 		}
 	}
 }
 
 ///return the local value of parameter
-btScalar btPoint2PointConstraint::getParam(int num, int axis) const 
+btScalar btPoint2PointConstraint::getParam(int num, int axis) const
 {
 	btScalar retVal(SIMD_INFINITY);
-	if(axis != -1)
+	if (axis != -1)
 	{
 		btAssertConstrParams(0);
 	}
 	else
 	{
-		switch(num)
+		switch (num)
 		{
-			case BT_CONSTRAINT_ERP :
-			case BT_CONSTRAINT_STOP_ERP :
+			case BT_CONSTRAINT_ERP:
+			case BT_CONSTRAINT_STOP_ERP:
 				btAssertConstrParams(m_flags & BT_P2P_FLAGS_ERP);
-				retVal = m_erp; 
+				retVal = m_erp;
 				break;
-			case BT_CONSTRAINT_CFM :
-			case BT_CONSTRAINT_STOP_CFM :
+			case BT_CONSTRAINT_CFM:
+			case BT_CONSTRAINT_STOP_CFM:
 				btAssertConstrParams(m_flags & BT_P2P_FLAGS_CFM);
-				retVal = m_cfm; 
+				retVal = m_cfm;
 				break;
-			default: 
+			default:
 				btAssertConstrParams(0);
 		}
 	}
 	return retVal;
 }
-	
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btPoint2PointConstraint.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btPoint2PointConstraint.h
index 8fa03d719d7..4717e198004 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btPoint2PointConstraint.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btPoint2PointConstraint.h
@@ -22,26 +22,24 @@ subject to the following restrictions:
 
 class btRigidBody;
 
-
 #ifdef BT_USE_DOUBLE_PRECISION
-#define btPoint2PointConstraintData2	btPoint2PointConstraintDoubleData2
-#define btPoint2PointConstraintDataName	"btPoint2PointConstraintDoubleData2"
+#define btPoint2PointConstraintData2 btPoint2PointConstraintDoubleData2
+#define btPoint2PointConstraintDataName "btPoint2PointConstraintDoubleData2"
 #else
-#define btPoint2PointConstraintData2	btPoint2PointConstraintFloatData
-#define btPoint2PointConstraintDataName	"btPoint2PointConstraintFloatData"
-#endif //BT_USE_DOUBLE_PRECISION
+#define btPoint2PointConstraintData2 btPoint2PointConstraintFloatData
+#define btPoint2PointConstraintDataName "btPoint2PointConstraintFloatData"
+#endif  //BT_USE_DOUBLE_PRECISION
 
-struct	btConstraintSetting
+struct btConstraintSetting
 {
-	btConstraintSetting()	:
-		m_tau(btScalar(0.3)),
-		m_damping(btScalar(1.)),
-		m_impulseClamp(btScalar(0.))
+	btConstraintSetting() : m_tau(btScalar(0.3)),
+							m_damping(btScalar(1.)),
+							m_impulseClamp(btScalar(0.))
 	{
 	}
-	btScalar		m_tau;
-	btScalar		m_damping;
-	btScalar		m_impulseClamp;
+	btScalar m_tau;
+	btScalar m_damping;
+	btScalar m_impulseClamp;
 };
 
 enum btPoint2PointFlags
@@ -51,52 +49,51 @@ enum btPoint2PointFlags
 };
 
 /// point to point constraint between two rigidbodies each with a pivotpoint that descibes the 'ballsocket' location in local space
-ATTRIBUTE_ALIGNED16(class) btPoint2PointConstraint : public btTypedConstraint
+ATTRIBUTE_ALIGNED16(class)
+btPoint2PointConstraint : public btTypedConstraint
 {
 #ifdef IN_PARALLELL_SOLVER
 public:
 #endif
-	btJacobianEntry	m_jac[3]; //3 orthogonal linear constraints
-	
-	btVector3	m_pivotInA;
-	btVector3	m_pivotInB;
-	
-	int			m_flags;
-	btScalar	m_erp;
-	btScalar	m_cfm;
-	
-public:
+	btJacobianEntry m_jac[3];  //3 orthogonal linear constraints
 
+	btVector3 m_pivotInA;
+	btVector3 m_pivotInB;
+
+	int m_flags;
+	btScalar m_erp;
+	btScalar m_cfm;
+
+public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
 	///for backwards compatibility during the transition to 'getInfo/getInfo2'
-	bool		m_useSolveConstraintObsolete;
+	bool m_useSolveConstraintObsolete;
 
-	btConstraintSetting	m_setting;
+	btConstraintSetting m_setting;
 
-	btPoint2PointConstraint(btRigidBody& rbA,btRigidBody& rbB, const btVector3& pivotInA,const btVector3& pivotInB);
+	btPoint2PointConstraint(btRigidBody & rbA, btRigidBody & rbB, const btVector3& pivotInA, const btVector3& pivotInB);
 
-	btPoint2PointConstraint(btRigidBody& rbA,const btVector3& pivotInA);
+	btPoint2PointConstraint(btRigidBody & rbA, const btVector3& pivotInA);
 
+	virtual void buildJacobian();
 
-	virtual void	buildJacobian();
+	virtual void getInfo1(btConstraintInfo1 * info);
 
-	virtual void getInfo1 (btConstraintInfo1* info);
+	void getInfo1NonVirtual(btConstraintInfo1 * info);
 
-	void getInfo1NonVirtual (btConstraintInfo1* info);
+	virtual void getInfo2(btConstraintInfo2 * info);
 
-	virtual void getInfo2 (btConstraintInfo2* info);
+	void getInfo2NonVirtual(btConstraintInfo2 * info, const btTransform& body0_trans, const btTransform& body1_trans);
 
-	void getInfo2NonVirtual (btConstraintInfo2* info, const btTransform& body0_trans, const btTransform& body1_trans);
+	void updateRHS(btScalar timeStep);
 
-	void	updateRHS(btScalar	timeStep);
-
-	void	setPivotA(const btVector3& pivotA)
+	void setPivotA(const btVector3& pivotA)
 	{
 		m_pivotInA = pivotA;
 	}
 
-	void	setPivotB(const btVector3& pivotB)
+	void setPivotB(const btVector3& pivotB)
 	{
 		m_pivotInB = pivotB;
 	}
@@ -111,70 +108,66 @@ public:
 		return m_pivotInB;
 	}
 
-	///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5). 
+	///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5).
 	///If no axis is provided, it uses the default axis for this constraint.
-	virtual	void	setParam(int num, btScalar value, int axis = -1);
+	virtual void setParam(int num, btScalar value, int axis = -1);
 	///return the local value of parameter
-	virtual	btScalar getParam(int num, int axis = -1) const;
-	
-	virtual	int getFlags() const
-	{
-        	return m_flags;
-    	}
+	virtual btScalar getParam(int num, int axis = -1) const;
 
-	virtual	int	calculateSerializeBufferSize() const;
+	virtual int getFlags() const
+	{
+		return m_flags;
+	}
+
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
-
-
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 };
 
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
-struct	btPoint2PointConstraintFloatData
+struct btPoint2PointConstraintFloatData
 {
-	btTypedConstraintData	m_typeConstraintData;
-	btVector3FloatData	m_pivotInA;
-	btVector3FloatData	m_pivotInB;
+	btTypedConstraintData m_typeConstraintData;
+	btVector3FloatData m_pivotInA;
+	btVector3FloatData m_pivotInB;
 };
 
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
-struct	btPoint2PointConstraintDoubleData2
+struct btPoint2PointConstraintDoubleData2
 {
-	btTypedConstraintDoubleData	m_typeConstraintData;
-	btVector3DoubleData	m_pivotInA;
-	btVector3DoubleData	m_pivotInB;
+	btTypedConstraintDoubleData m_typeConstraintData;
+	btVector3DoubleData m_pivotInA;
+	btVector3DoubleData m_pivotInB;
 };
 
 #ifdef BT_BACKWARDS_COMPATIBLE_SERIALIZATION
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
 ///this structure is not used, except for loading pre-2.82 .bullet files
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
-struct	btPoint2PointConstraintDoubleData
+struct btPoint2PointConstraintDoubleData
 {
-	btTypedConstraintData	m_typeConstraintData;
-	btVector3DoubleData	m_pivotInA;
-	btVector3DoubleData	m_pivotInB;
+	btTypedConstraintData m_typeConstraintData;
+	btVector3DoubleData m_pivotInA;
+	btVector3DoubleData m_pivotInB;
 };
-#endif //BT_BACKWARDS_COMPATIBLE_SERIALIZATION
+#endif  //BT_BACKWARDS_COMPATIBLE_SERIALIZATION
 
-
-SIMD_FORCE_INLINE	int	btPoint2PointConstraint::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btPoint2PointConstraint::calculateSerializeBufferSize() const
 {
 	return sizeof(btPoint2PointConstraintData2);
-
 }
 
-	///fills the dataBuffer and returns the struct name (and 0 on failure)
-SIMD_FORCE_INLINE	const char*	btPoint2PointConstraint::serialize(void* dataBuffer, btSerializer* serializer) const
+///fills the dataBuffer and returns the struct name (and 0 on failure)
+SIMD_FORCE_INLINE const char* btPoint2PointConstraint::serialize(void* dataBuffer, btSerializer* serializer) const
 {
 	btPoint2PointConstraintData2* p2pData = (btPoint2PointConstraintData2*)dataBuffer;
 
-	btTypedConstraint::serialize(&p2pData->m_typeConstraintData,serializer);
+	btTypedConstraint::serialize(&p2pData->m_typeConstraintData, serializer);
 	m_pivotInA.serialize(p2pData->m_pivotInA);
 	m_pivotInB.serialize(p2pData->m_pivotInB);
 
 	return btPoint2PointConstraintDataName;
 }
 
-#endif //BT_POINT2POINTCONSTRAINT_H
+#endif  //BT_POINT2POINTCONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSequentialImpulseConstraintSolver.cpp b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSequentialImpulseConstraintSolver.cpp
index 63174a6ec0e..def3227b438 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSequentialImpulseConstraintSolver.cpp
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSequentialImpulseConstraintSolver.cpp
@@ -21,7 +21,6 @@ subject to the following restrictions:
 #include "btSequentialImpulseConstraintSolver.h"
 #include "BulletCollision/NarrowPhaseCollision/btPersistentManifold.h"
 
-
 #include "LinearMath/btIDebugDraw.h"
 #include "LinearMath/btCpuFeatureUtility.h"
 
@@ -34,9 +33,9 @@ subject to the following restrictions:
 //#include "btSolverBody.h"
 //#include "btSolverConstraint.h"
 #include "LinearMath/btAlignedObjectArray.h"
-#include <string.h> //for memset
+#include <string.h>  //for memset
 
-int		gNumSplitImpulseRecoveries = 0;
+int gNumSplitImpulseRecoveries = 0;
 
 #include "BulletDynamics/Dynamics/btRigidBody.h"
 
@@ -45,13 +44,13 @@ int		gNumSplitImpulseRecoveries = 0;
 ///Below are optional SSE2 and SSE4/FMA3 versions. We assume most hardware has SSE2. For SSE4/FMA3 we perform a CPU feature check.
 static btScalar gResolveSingleConstraintRowGeneric_scalar_reference(btSolverBody& bodyA, btSolverBody& bodyB, const btSolverConstraint& c)
 {
-	btScalar deltaImpulse = c.m_rhs - btScalar(c.m_appliedImpulse)*c.m_cfm;
+	btScalar deltaImpulse = c.m_rhs - btScalar(c.m_appliedImpulse) * c.m_cfm;
 	const btScalar deltaVel1Dotn = c.m_contactNormal1.dot(bodyA.internalGetDeltaLinearVelocity()) + c.m_relpos1CrossNormal.dot(bodyA.internalGetDeltaAngularVelocity());
 	const btScalar deltaVel2Dotn = c.m_contactNormal2.dot(bodyB.internalGetDeltaLinearVelocity()) + c.m_relpos2CrossNormal.dot(bodyB.internalGetDeltaAngularVelocity());
 
 	//	const btScalar delta_rel_vel	=	deltaVel1Dotn-deltaVel2Dotn;
-	deltaImpulse -= deltaVel1Dotn*c.m_jacDiagABInv;
-	deltaImpulse -= deltaVel2Dotn*c.m_jacDiagABInv;
+	deltaImpulse -= deltaVel1Dotn * c.m_jacDiagABInv;
+	deltaImpulse -= deltaVel2Dotn * c.m_jacDiagABInv;
 
 	const btScalar sum = btScalar(c.m_appliedImpulse) + deltaImpulse;
 	if (sum < c.m_lowerLimit)
@@ -69,21 +68,20 @@ static btScalar gResolveSingleConstraintRowGeneric_scalar_reference(btSolverBody
 		c.m_appliedImpulse = sum;
 	}
 
-	bodyA.internalApplyImpulse(c.m_contactNormal1*bodyA.internalGetInvMass(), c.m_angularComponentA, deltaImpulse);
-	bodyB.internalApplyImpulse(c.m_contactNormal2*bodyB.internalGetInvMass(), c.m_angularComponentB, deltaImpulse);
+	bodyA.internalApplyImpulse(c.m_contactNormal1 * bodyA.internalGetInvMass(), c.m_angularComponentA, deltaImpulse);
+	bodyB.internalApplyImpulse(c.m_contactNormal2 * bodyB.internalGetInvMass(), c.m_angularComponentB, deltaImpulse);
 
-	return deltaImpulse*(1./c.m_jacDiagABInv);
+	return deltaImpulse * (1. / c.m_jacDiagABInv);
 }
 
-
 static btScalar gResolveSingleConstraintRowLowerLimit_scalar_reference(btSolverBody& bodyA, btSolverBody& bodyB, const btSolverConstraint& c)
 {
-	btScalar deltaImpulse = c.m_rhs - btScalar(c.m_appliedImpulse)*c.m_cfm;
+	btScalar deltaImpulse = c.m_rhs - btScalar(c.m_appliedImpulse) * c.m_cfm;
 	const btScalar deltaVel1Dotn = c.m_contactNormal1.dot(bodyA.internalGetDeltaLinearVelocity()) + c.m_relpos1CrossNormal.dot(bodyA.internalGetDeltaAngularVelocity());
 	const btScalar deltaVel2Dotn = c.m_contactNormal2.dot(bodyB.internalGetDeltaLinearVelocity()) + c.m_relpos2CrossNormal.dot(bodyB.internalGetDeltaAngularVelocity());
 
-	deltaImpulse -= deltaVel1Dotn*c.m_jacDiagABInv;
-	deltaImpulse -= deltaVel2Dotn*c.m_jacDiagABInv;
+	deltaImpulse -= deltaVel1Dotn * c.m_jacDiagABInv;
+	deltaImpulse -= deltaVel2Dotn * c.m_jacDiagABInv;
 	const btScalar sum = btScalar(c.m_appliedImpulse) + deltaImpulse;
 	if (sum < c.m_lowerLimit)
 	{
@@ -94,58 +92,55 @@ static btScalar gResolveSingleConstraintRowLowerLimit_scalar_reference(btSolverB
 	{
 		c.m_appliedImpulse = sum;
 	}
-	bodyA.internalApplyImpulse(c.m_contactNormal1*bodyA.internalGetInvMass(), c.m_angularComponentA, deltaImpulse);
-	bodyB.internalApplyImpulse(c.m_contactNormal2*bodyB.internalGetInvMass(), c.m_angularComponentB, deltaImpulse);
+	bodyA.internalApplyImpulse(c.m_contactNormal1 * bodyA.internalGetInvMass(), c.m_angularComponentA, deltaImpulse);
+	bodyB.internalApplyImpulse(c.m_contactNormal2 * bodyB.internalGetInvMass(), c.m_angularComponentB, deltaImpulse);
 
-	return deltaImpulse*(1./c.m_jacDiagABInv);
+	return deltaImpulse * (1. / c.m_jacDiagABInv);
 }
 
-
-
 #ifdef USE_SIMD
 #include <emmintrin.h>
 
-
-#define btVecSplat(x, e) _mm_shuffle_ps(x, x, _MM_SHUFFLE(e,e,e,e))
-static inline __m128 btSimdDot3( __m128 vec0, __m128 vec1 )
+#define btVecSplat(x, e) _mm_shuffle_ps(x, x, _MM_SHUFFLE(e, e, e, e))
+static inline __m128 btSimdDot3(__m128 vec0, __m128 vec1)
 {
-	__m128 result = _mm_mul_ps( vec0, vec1);
-	return _mm_add_ps( btVecSplat( result, 0 ), _mm_add_ps( btVecSplat( result, 1 ), btVecSplat( result, 2 ) ) );
+	__m128 result = _mm_mul_ps(vec0, vec1);
+	return _mm_add_ps(btVecSplat(result, 0), _mm_add_ps(btVecSplat(result, 1), btVecSplat(result, 2)));
 }
 
-#if defined (BT_ALLOW_SSE4)
+#if defined(BT_ALLOW_SSE4)
 #include <intrin.h>
 
-#define USE_FMA					1
-#define USE_FMA3_INSTEAD_FMA4	1
-#define USE_SSE4_DOT			1
+#define USE_FMA 1
+#define USE_FMA3_INSTEAD_FMA4 1
+#define USE_SSE4_DOT 1
 
-#define SSE4_DP(a, b)			_mm_dp_ps(a, b, 0x7f)
-#define SSE4_DP_FP(a, b)		_mm_cvtss_f32(_mm_dp_ps(a, b, 0x7f))
+#define SSE4_DP(a, b) _mm_dp_ps(a, b, 0x7f)
+#define SSE4_DP_FP(a, b) _mm_cvtss_f32(_mm_dp_ps(a, b, 0x7f))
 
 #if USE_SSE4_DOT
-#define DOT_PRODUCT(a, b)		SSE4_DP(a, b)
+#define DOT_PRODUCT(a, b) SSE4_DP(a, b)
 #else
-#define DOT_PRODUCT(a, b)		btSimdDot3(a, b)
+#define DOT_PRODUCT(a, b) btSimdDot3(a, b)
 #endif
 
 #if USE_FMA
 #if USE_FMA3_INSTEAD_FMA4
 // a*b + c
-#define FMADD(a, b, c)		_mm_fmadd_ps(a, b, c)
+#define FMADD(a, b, c) _mm_fmadd_ps(a, b, c)
 // -(a*b) + c
-#define FMNADD(a, b, c)		_mm_fnmadd_ps(a, b, c)
-#else // USE_FMA3
+#define FMNADD(a, b, c) _mm_fnmadd_ps(a, b, c)
+#else  // USE_FMA3
 // a*b + c
-#define FMADD(a, b, c)		_mm_macc_ps(a, b, c)
+#define FMADD(a, b, c) _mm_macc_ps(a, b, c)
 // -(a*b) + c
-#define FMNADD(a, b, c)		_mm_nmacc_ps(a, b, c)
+#define FMNADD(a, b, c) _mm_nmacc_ps(a, b, c)
 #endif
-#else // USE_FMA
+#else  // USE_FMA
 // c + a*b
-#define FMADD(a, b, c)		_mm_add_ps(c, _mm_mul_ps(a, b))
+#define FMADD(a, b, c) _mm_add_ps(c, _mm_mul_ps(a, b))
 // c - a*b
-#define FMNADD(a, b, c)		_mm_sub_ps(c, _mm_mul_ps(a, b))
+#define FMNADD(a, b, c) _mm_sub_ps(c, _mm_mul_ps(a, b))
 #endif
 #endif
 
@@ -153,8 +148,8 @@ static inline __m128 btSimdDot3( __m128 vec0, __m128 vec1 )
 static btScalar gResolveSingleConstraintRowGeneric_sse2(btSolverBody& bodyA, btSolverBody& bodyB, const btSolverConstraint& c)
 {
 	__m128 cpAppliedImp = _mm_set1_ps(c.m_appliedImpulse);
-	__m128	lowerLimit1 = _mm_set1_ps(c.m_lowerLimit);
-	__m128	upperLimit1 = _mm_set1_ps(c.m_upperLimit);
+	__m128 lowerLimit1 = _mm_set1_ps(c.m_lowerLimit);
+	__m128 upperLimit1 = _mm_set1_ps(c.m_upperLimit);
 	btSimdScalar deltaImpulse = _mm_sub_ps(_mm_set1_ps(c.m_rhs), _mm_mul_ps(_mm_set1_ps(c.m_appliedImpulse), _mm_set1_ps(c.m_cfm)));
 	__m128 deltaVel1Dotn = _mm_add_ps(btSimdDot3(c.m_contactNormal1.mVec128, bodyA.internalGetDeltaLinearVelocity().mVec128), btSimdDot3(c.m_relpos1CrossNormal.mVec128, bodyA.internalGetDeltaAngularVelocity().mVec128));
 	__m128 deltaVel2Dotn = _mm_add_ps(btSimdDot3(c.m_contactNormal2.mVec128, bodyB.internalGetDeltaLinearVelocity().mVec128), btSimdDot3(c.m_relpos2CrossNormal.mVec128, bodyB.internalGetDeltaAngularVelocity().mVec128));
@@ -170,52 +165,49 @@ static btScalar gResolveSingleConstraintRowGeneric_sse2(btSolverBody& bodyA, btS
 	__m128 upperMinApplied = _mm_sub_ps(upperLimit1, cpAppliedImp);
 	deltaImpulse = _mm_or_ps(_mm_and_ps(resultUpperLess, deltaImpulse), _mm_andnot_ps(resultUpperLess, upperMinApplied));
 	c.m_appliedImpulse = _mm_or_ps(_mm_and_ps(resultUpperLess, c.m_appliedImpulse), _mm_andnot_ps(resultUpperLess, upperLimit1));
-	__m128	linearComponentA = _mm_mul_ps(c.m_contactNormal1.mVec128, bodyA.internalGetInvMass().mVec128);
-	__m128	linearComponentB = _mm_mul_ps((c.m_contactNormal2).mVec128, bodyB.internalGetInvMass().mVec128);
+	__m128 linearComponentA = _mm_mul_ps(c.m_contactNormal1.mVec128, bodyA.internalGetInvMass().mVec128);
+	__m128 linearComponentB = _mm_mul_ps((c.m_contactNormal2).mVec128, bodyB.internalGetInvMass().mVec128);
 	__m128 impulseMagnitude = deltaImpulse;
 	bodyA.internalGetDeltaLinearVelocity().mVec128 = _mm_add_ps(bodyA.internalGetDeltaLinearVelocity().mVec128, _mm_mul_ps(linearComponentA, impulseMagnitude));
 	bodyA.internalGetDeltaAngularVelocity().mVec128 = _mm_add_ps(bodyA.internalGetDeltaAngularVelocity().mVec128, _mm_mul_ps(c.m_angularComponentA.mVec128, impulseMagnitude));
 	bodyB.internalGetDeltaLinearVelocity().mVec128 = _mm_add_ps(bodyB.internalGetDeltaLinearVelocity().mVec128, _mm_mul_ps(linearComponentB, impulseMagnitude));
 	bodyB.internalGetDeltaAngularVelocity().mVec128 = _mm_add_ps(bodyB.internalGetDeltaAngularVelocity().mVec128, _mm_mul_ps(c.m_angularComponentB.mVec128, impulseMagnitude));
-	return  deltaImpulse.m_floats[0]/c.m_jacDiagABInv;
+	return deltaImpulse.m_floats[0] / c.m_jacDiagABInv;
 }
 
-
 // Enhanced version of gResolveSingleConstraintRowGeneric_sse2 with SSE4.1 and FMA3
 static btScalar gResolveSingleConstraintRowGeneric_sse4_1_fma3(btSolverBody& bodyA, btSolverBody& bodyB, const btSolverConstraint& c)
 {
-#if defined (BT_ALLOW_SSE4)
-	__m128 tmp					= _mm_set_ps1(c.m_jacDiagABInv);
-	__m128 deltaImpulse			= _mm_set_ps1(c.m_rhs - btScalar(c.m_appliedImpulse)*c.m_cfm);
-	const __m128 lowerLimit		= _mm_set_ps1(c.m_lowerLimit);
-	const __m128 upperLimit		= _mm_set_ps1(c.m_upperLimit);
-	const __m128 deltaVel1Dotn	= _mm_add_ps(DOT_PRODUCT(c.m_contactNormal1.mVec128, bodyA.internalGetDeltaLinearVelocity().mVec128), DOT_PRODUCT(c.m_relpos1CrossNormal.mVec128, bodyA.internalGetDeltaAngularVelocity().mVec128));
-	const __m128 deltaVel2Dotn	= _mm_add_ps(DOT_PRODUCT(c.m_contactNormal2.mVec128, bodyB.internalGetDeltaLinearVelocity().mVec128), DOT_PRODUCT(c.m_relpos2CrossNormal.mVec128, bodyB.internalGetDeltaAngularVelocity().mVec128));
-	deltaImpulse				= FMNADD(deltaVel1Dotn, tmp, deltaImpulse);
-	deltaImpulse				= FMNADD(deltaVel2Dotn, tmp, deltaImpulse);
-	tmp							= _mm_add_ps(c.m_appliedImpulse, deltaImpulse); // sum
-	const __m128 maskLower		= _mm_cmpgt_ps(tmp, lowerLimit);
-	const __m128 maskUpper		= _mm_cmpgt_ps(upperLimit, tmp);
-	deltaImpulse				= _mm_blendv_ps(_mm_sub_ps(lowerLimit, c.m_appliedImpulse), _mm_blendv_ps(_mm_sub_ps(upperLimit, c.m_appliedImpulse), deltaImpulse, maskUpper), maskLower);
-	c.m_appliedImpulse			= _mm_blendv_ps(lowerLimit, _mm_blendv_ps(upperLimit, tmp, maskUpper), maskLower);
-	bodyA.internalGetDeltaLinearVelocity().mVec128	= FMADD(_mm_mul_ps(c.m_contactNormal1.mVec128, bodyA.internalGetInvMass().mVec128), deltaImpulse, bodyA.internalGetDeltaLinearVelocity().mVec128);
+#if defined(BT_ALLOW_SSE4)
+	__m128 tmp = _mm_set_ps1(c.m_jacDiagABInv);
+	__m128 deltaImpulse = _mm_set_ps1(c.m_rhs - btScalar(c.m_appliedImpulse) * c.m_cfm);
+	const __m128 lowerLimit = _mm_set_ps1(c.m_lowerLimit);
+	const __m128 upperLimit = _mm_set_ps1(c.m_upperLimit);
+	const __m128 deltaVel1Dotn = _mm_add_ps(DOT_PRODUCT(c.m_contactNormal1.mVec128, bodyA.internalGetDeltaLinearVelocity().mVec128), DOT_PRODUCT(c.m_relpos1CrossNormal.mVec128, bodyA.internalGetDeltaAngularVelocity().mVec128));
+	const __m128 deltaVel2Dotn = _mm_add_ps(DOT_PRODUCT(c.m_contactNormal2.mVec128, bodyB.internalGetDeltaLinearVelocity().mVec128), DOT_PRODUCT(c.m_relpos2CrossNormal.mVec128, bodyB.internalGetDeltaAngularVelocity().mVec128));
+	deltaImpulse = FMNADD(deltaVel1Dotn, tmp, deltaImpulse);
+	deltaImpulse = FMNADD(deltaVel2Dotn, tmp, deltaImpulse);
+	tmp = _mm_add_ps(c.m_appliedImpulse, deltaImpulse);  // sum
+	const __m128 maskLower = _mm_cmpgt_ps(tmp, lowerLimit);
+	const __m128 maskUpper = _mm_cmpgt_ps(upperLimit, tmp);
+	deltaImpulse = _mm_blendv_ps(_mm_sub_ps(lowerLimit, c.m_appliedImpulse), _mm_blendv_ps(_mm_sub_ps(upperLimit, c.m_appliedImpulse), deltaImpulse, maskUpper), maskLower);
+	c.m_appliedImpulse = _mm_blendv_ps(lowerLimit, _mm_blendv_ps(upperLimit, tmp, maskUpper), maskLower);
+	bodyA.internalGetDeltaLinearVelocity().mVec128 = FMADD(_mm_mul_ps(c.m_contactNormal1.mVec128, bodyA.internalGetInvMass().mVec128), deltaImpulse, bodyA.internalGetDeltaLinearVelocity().mVec128);
 	bodyA.internalGetDeltaAngularVelocity().mVec128 = FMADD(c.m_angularComponentA.mVec128, deltaImpulse, bodyA.internalGetDeltaAngularVelocity().mVec128);
-	bodyB.internalGetDeltaLinearVelocity().mVec128	= FMADD(_mm_mul_ps(c.m_contactNormal2.mVec128, bodyB.internalGetInvMass().mVec128), deltaImpulse, bodyB.internalGetDeltaLinearVelocity().mVec128);
+	bodyB.internalGetDeltaLinearVelocity().mVec128 = FMADD(_mm_mul_ps(c.m_contactNormal2.mVec128, bodyB.internalGetInvMass().mVec128), deltaImpulse, bodyB.internalGetDeltaLinearVelocity().mVec128);
 	bodyB.internalGetDeltaAngularVelocity().mVec128 = FMADD(c.m_angularComponentB.mVec128, deltaImpulse, bodyB.internalGetDeltaAngularVelocity().mVec128);
 	btSimdScalar deltaImp = deltaImpulse;
-	return deltaImp.m_floats[0]*(1./c.m_jacDiagABInv);
+	return deltaImp.m_floats[0] * (1. / c.m_jacDiagABInv);
 #else
-	return gResolveSingleConstraintRowGeneric_sse2(bodyA,bodyB,c);
+	return gResolveSingleConstraintRowGeneric_sse2(bodyA, bodyB, c);
 #endif
 }
 
-
-
 static btScalar gResolveSingleConstraintRowLowerLimit_sse2(btSolverBody& bodyA, btSolverBody& bodyB, const btSolverConstraint& c)
 {
 	__m128 cpAppliedImp = _mm_set1_ps(c.m_appliedImpulse);
-	__m128	lowerLimit1 = _mm_set1_ps(c.m_lowerLimit);
-	__m128	upperLimit1 = _mm_set1_ps(c.m_upperLimit);
+	__m128 lowerLimit1 = _mm_set1_ps(c.m_lowerLimit);
+	__m128 upperLimit1 = _mm_set1_ps(c.m_upperLimit);
 	btSimdScalar deltaImpulse = _mm_sub_ps(_mm_set1_ps(c.m_rhs), _mm_mul_ps(_mm_set1_ps(c.m_appliedImpulse), _mm_set1_ps(c.m_cfm)));
 	__m128 deltaVel1Dotn = _mm_add_ps(btSimdDot3(c.m_contactNormal1.mVec128, bodyA.internalGetDeltaLinearVelocity().mVec128), btSimdDot3(c.m_relpos1CrossNormal.mVec128, bodyA.internalGetDeltaAngularVelocity().mVec128));
 	__m128 deltaVel2Dotn = _mm_add_ps(btSimdDot3(c.m_contactNormal2.mVec128, bodyB.internalGetDeltaLinearVelocity().mVec128), btSimdDot3(c.m_relpos2CrossNormal.mVec128, bodyB.internalGetDeltaAngularVelocity().mVec128));
@@ -228,104 +220,98 @@ static btScalar gResolveSingleConstraintRowLowerLimit_sse2(btSolverBody& bodyA,
 	__m128 lowMinApplied = _mm_sub_ps(lowerLimit1, cpAppliedImp);
 	deltaImpulse = _mm_or_ps(_mm_and_ps(resultLowerLess, lowMinApplied), _mm_andnot_ps(resultLowerLess, deltaImpulse));
 	c.m_appliedImpulse = _mm_or_ps(_mm_and_ps(resultLowerLess, lowerLimit1), _mm_andnot_ps(resultLowerLess, sum));
-	__m128	linearComponentA = _mm_mul_ps(c.m_contactNormal1.mVec128, bodyA.internalGetInvMass().mVec128);
-	__m128	linearComponentB = _mm_mul_ps(c.m_contactNormal2.mVec128, bodyB.internalGetInvMass().mVec128);
+	__m128 linearComponentA = _mm_mul_ps(c.m_contactNormal1.mVec128, bodyA.internalGetInvMass().mVec128);
+	__m128 linearComponentB = _mm_mul_ps(c.m_contactNormal2.mVec128, bodyB.internalGetInvMass().mVec128);
 	__m128 impulseMagnitude = deltaImpulse;
 	bodyA.internalGetDeltaLinearVelocity().mVec128 = _mm_add_ps(bodyA.internalGetDeltaLinearVelocity().mVec128, _mm_mul_ps(linearComponentA, impulseMagnitude));
 	bodyA.internalGetDeltaAngularVelocity().mVec128 = _mm_add_ps(bodyA.internalGetDeltaAngularVelocity().mVec128, _mm_mul_ps(c.m_angularComponentA.mVec128, impulseMagnitude));
 	bodyB.internalGetDeltaLinearVelocity().mVec128 = _mm_add_ps(bodyB.internalGetDeltaLinearVelocity().mVec128, _mm_mul_ps(linearComponentB, impulseMagnitude));
 	bodyB.internalGetDeltaAngularVelocity().mVec128 = _mm_add_ps(bodyB.internalGetDeltaAngularVelocity().mVec128, _mm_mul_ps(c.m_angularComponentB.mVec128, impulseMagnitude));
-	return  deltaImpulse.m_floats[0]/c.m_jacDiagABInv;
+	return deltaImpulse.m_floats[0] / c.m_jacDiagABInv;
 }
 
-
 // Enhanced version of gResolveSingleConstraintRowGeneric_sse2 with SSE4.1 and FMA3
 static btScalar gResolveSingleConstraintRowLowerLimit_sse4_1_fma3(btSolverBody& bodyA, btSolverBody& bodyB, const btSolverConstraint& c)
 {
 #ifdef BT_ALLOW_SSE4
-	__m128 tmp					= _mm_set_ps1(c.m_jacDiagABInv);
-	__m128 deltaImpulse			= _mm_set_ps1(c.m_rhs - btScalar(c.m_appliedImpulse)*c.m_cfm);
-	const __m128 lowerLimit		= _mm_set_ps1(c.m_lowerLimit);
-	const __m128 deltaVel1Dotn	= _mm_add_ps(DOT_PRODUCT(c.m_contactNormal1.mVec128, bodyA.internalGetDeltaLinearVelocity().mVec128), DOT_PRODUCT(c.m_relpos1CrossNormal.mVec128, bodyA.internalGetDeltaAngularVelocity().mVec128));
-	const __m128 deltaVel2Dotn	= _mm_add_ps(DOT_PRODUCT(c.m_contactNormal2.mVec128, bodyB.internalGetDeltaLinearVelocity().mVec128), DOT_PRODUCT(c.m_relpos2CrossNormal.mVec128, bodyB.internalGetDeltaAngularVelocity().mVec128));
-	deltaImpulse				= FMNADD(deltaVel1Dotn, tmp, deltaImpulse);
-	deltaImpulse				= FMNADD(deltaVel2Dotn, tmp, deltaImpulse);
-	tmp							= _mm_add_ps(c.m_appliedImpulse, deltaImpulse);
-	const __m128 mask			= _mm_cmpgt_ps(tmp, lowerLimit);
-	deltaImpulse				= _mm_blendv_ps(_mm_sub_ps(lowerLimit, c.m_appliedImpulse), deltaImpulse, mask);
-	c.m_appliedImpulse			= _mm_blendv_ps(lowerLimit, tmp, mask);
-	bodyA.internalGetDeltaLinearVelocity().mVec128	= FMADD(_mm_mul_ps(c.m_contactNormal1.mVec128, bodyA.internalGetInvMass().mVec128), deltaImpulse, bodyA.internalGetDeltaLinearVelocity().mVec128);
+	__m128 tmp = _mm_set_ps1(c.m_jacDiagABInv);
+	__m128 deltaImpulse = _mm_set_ps1(c.m_rhs - btScalar(c.m_appliedImpulse) * c.m_cfm);
+	const __m128 lowerLimit = _mm_set_ps1(c.m_lowerLimit);
+	const __m128 deltaVel1Dotn = _mm_add_ps(DOT_PRODUCT(c.m_contactNormal1.mVec128, bodyA.internalGetDeltaLinearVelocity().mVec128), DOT_PRODUCT(c.m_relpos1CrossNormal.mVec128, bodyA.internalGetDeltaAngularVelocity().mVec128));
+	const __m128 deltaVel2Dotn = _mm_add_ps(DOT_PRODUCT(c.m_contactNormal2.mVec128, bodyB.internalGetDeltaLinearVelocity().mVec128), DOT_PRODUCT(c.m_relpos2CrossNormal.mVec128, bodyB.internalGetDeltaAngularVelocity().mVec128));
+	deltaImpulse = FMNADD(deltaVel1Dotn, tmp, deltaImpulse);
+	deltaImpulse = FMNADD(deltaVel2Dotn, tmp, deltaImpulse);
+	tmp = _mm_add_ps(c.m_appliedImpulse, deltaImpulse);
+	const __m128 mask = _mm_cmpgt_ps(tmp, lowerLimit);
+	deltaImpulse = _mm_blendv_ps(_mm_sub_ps(lowerLimit, c.m_appliedImpulse), deltaImpulse, mask);
+	c.m_appliedImpulse = _mm_blendv_ps(lowerLimit, tmp, mask);
+	bodyA.internalGetDeltaLinearVelocity().mVec128 = FMADD(_mm_mul_ps(c.m_contactNormal1.mVec128, bodyA.internalGetInvMass().mVec128), deltaImpulse, bodyA.internalGetDeltaLinearVelocity().mVec128);
 	bodyA.internalGetDeltaAngularVelocity().mVec128 = FMADD(c.m_angularComponentA.mVec128, deltaImpulse, bodyA.internalGetDeltaAngularVelocity().mVec128);
-	bodyB.internalGetDeltaLinearVelocity().mVec128	= FMADD(_mm_mul_ps(c.m_contactNormal2.mVec128, bodyB.internalGetInvMass().mVec128), deltaImpulse, bodyB.internalGetDeltaLinearVelocity().mVec128);
+	bodyB.internalGetDeltaLinearVelocity().mVec128 = FMADD(_mm_mul_ps(c.m_contactNormal2.mVec128, bodyB.internalGetInvMass().mVec128), deltaImpulse, bodyB.internalGetDeltaLinearVelocity().mVec128);
 	bodyB.internalGetDeltaAngularVelocity().mVec128 = FMADD(c.m_angularComponentB.mVec128, deltaImpulse, bodyB.internalGetDeltaAngularVelocity().mVec128);
 	btSimdScalar deltaImp = deltaImpulse;
-	return deltaImp.m_floats[0]*(1./c.m_jacDiagABInv);
+	return deltaImp.m_floats[0] * (1. / c.m_jacDiagABInv);
 #else
-	return gResolveSingleConstraintRowLowerLimit_sse2(bodyA,bodyB,c);
-#endif //BT_ALLOW_SSE4
+	return gResolveSingleConstraintRowLowerLimit_sse2(bodyA, bodyB, c);
+#endif  //BT_ALLOW_SSE4
 }
 
+#endif  //USE_SIMD
 
-#endif //USE_SIMD
-
-
-
-btScalar btSequentialImpulseConstraintSolver::resolveSingleConstraintRowGenericSIMD(btSolverBody& bodyA,btSolverBody& bodyB,const btSolverConstraint& c)
+btScalar btSequentialImpulseConstraintSolver::resolveSingleConstraintRowGenericSIMD(btSolverBody& bodyA, btSolverBody& bodyB, const btSolverConstraint& c)
 {
 	return m_resolveSingleConstraintRowGeneric(bodyA, bodyB, c);
 }
 
 // Project Gauss Seidel or the equivalent Sequential Impulse
-btScalar btSequentialImpulseConstraintSolver::resolveSingleConstraintRowGeneric(btSolverBody& bodyA,btSolverBody& bodyB,const btSolverConstraint& c)
+btScalar btSequentialImpulseConstraintSolver::resolveSingleConstraintRowGeneric(btSolverBody& bodyA, btSolverBody& bodyB, const btSolverConstraint& c)
 {
 	return m_resolveSingleConstraintRowGeneric(bodyA, bodyB, c);
 }
 
-btScalar btSequentialImpulseConstraintSolver::resolveSingleConstraintRowLowerLimitSIMD(btSolverBody& bodyA,btSolverBody& bodyB,const btSolverConstraint& c)
+btScalar btSequentialImpulseConstraintSolver::resolveSingleConstraintRowLowerLimitSIMD(btSolverBody& bodyA, btSolverBody& bodyB, const btSolverConstraint& c)
 {
 	return m_resolveSingleConstraintRowLowerLimit(bodyA, bodyB, c);
 }
 
-
-btScalar btSequentialImpulseConstraintSolver::resolveSingleConstraintRowLowerLimit(btSolverBody& bodyA,btSolverBody& bodyB,const btSolverConstraint& c)
+btScalar btSequentialImpulseConstraintSolver::resolveSingleConstraintRowLowerLimit(btSolverBody& bodyA, btSolverBody& bodyB, const btSolverConstraint& c)
 {
 	return m_resolveSingleConstraintRowLowerLimit(bodyA, bodyB, c);
 }
 
-
 static btScalar gResolveSplitPenetrationImpulse_scalar_reference(
-        btSolverBody& bodyA,
-        btSolverBody& bodyB,
-        const btSolverConstraint& c)
+	btSolverBody& bodyA,
+	btSolverBody& bodyB,
+	const btSolverConstraint& c)
 {
 	btScalar deltaImpulse = 0.f;
 
-		if (c.m_rhsPenetration)
-        {
-			gNumSplitImpulseRecoveries++;
-			deltaImpulse = c.m_rhsPenetration-btScalar(c.m_appliedPushImpulse)*c.m_cfm;
-			const btScalar deltaVel1Dotn	=	c.m_contactNormal1.dot(bodyA.internalGetPushVelocity()) 	+ c.m_relpos1CrossNormal.dot(bodyA.internalGetTurnVelocity());
-			const btScalar deltaVel2Dotn	=	c.m_contactNormal2.dot(bodyB.internalGetPushVelocity())		+ c.m_relpos2CrossNormal.dot(bodyB.internalGetTurnVelocity());
+	if (c.m_rhsPenetration)
+	{
+		gNumSplitImpulseRecoveries++;
+		deltaImpulse = c.m_rhsPenetration - btScalar(c.m_appliedPushImpulse) * c.m_cfm;
+		const btScalar deltaVel1Dotn = c.m_contactNormal1.dot(bodyA.internalGetPushVelocity()) + c.m_relpos1CrossNormal.dot(bodyA.internalGetTurnVelocity());
+		const btScalar deltaVel2Dotn = c.m_contactNormal2.dot(bodyB.internalGetPushVelocity()) + c.m_relpos2CrossNormal.dot(bodyB.internalGetTurnVelocity());
 
-			deltaImpulse	-=	deltaVel1Dotn*c.m_jacDiagABInv;
-			deltaImpulse	-=	deltaVel2Dotn*c.m_jacDiagABInv;
-			const btScalar sum = btScalar(c.m_appliedPushImpulse) + deltaImpulse;
-			if (sum < c.m_lowerLimit)
-			{
-				deltaImpulse = c.m_lowerLimit-c.m_appliedPushImpulse;
-				c.m_appliedPushImpulse = c.m_lowerLimit;
-			}
-			else
-			{
-				c.m_appliedPushImpulse = sum;
-			}
-			bodyA.internalApplyPushImpulse(c.m_contactNormal1*bodyA.internalGetInvMass(),c.m_angularComponentA,deltaImpulse);
-			bodyB.internalApplyPushImpulse(c.m_contactNormal2*bodyB.internalGetInvMass(),c.m_angularComponentB,deltaImpulse);
-        }
-		return deltaImpulse*(1./c.m_jacDiagABInv);
+		deltaImpulse -= deltaVel1Dotn * c.m_jacDiagABInv;
+		deltaImpulse -= deltaVel2Dotn * c.m_jacDiagABInv;
+		const btScalar sum = btScalar(c.m_appliedPushImpulse) + deltaImpulse;
+		if (sum < c.m_lowerLimit)
+		{
+			deltaImpulse = c.m_lowerLimit - c.m_appliedPushImpulse;
+			c.m_appliedPushImpulse = c.m_lowerLimit;
+		}
+		else
+		{
+			c.m_appliedPushImpulse = sum;
+		}
+		bodyA.internalApplyPushImpulse(c.m_contactNormal1 * bodyA.internalGetInvMass(), c.m_angularComponentA, deltaImpulse);
+		bodyB.internalApplyPushImpulse(c.m_contactNormal2 * bodyB.internalGetInvMass(), c.m_angularComponentB, deltaImpulse);
+	}
+	return deltaImpulse * (1. / c.m_jacDiagABInv);
 }
 
-static btScalar gResolveSplitPenetrationImpulse_sse2(btSolverBody& bodyA,btSolverBody& bodyB,const btSolverConstraint& c)
+static btScalar gResolveSplitPenetrationImpulse_sse2(btSolverBody& bodyA, btSolverBody& bodyB, const btSolverConstraint& c)
 {
 #ifdef USE_SIMD
 	if (!c.m_rhsPenetration)
@@ -334,113 +320,109 @@ static btScalar gResolveSplitPenetrationImpulse_sse2(btSolverBody& bodyA,btSolve
 	gNumSplitImpulseRecoveries++;
 
 	__m128 cpAppliedImp = _mm_set1_ps(c.m_appliedPushImpulse);
-	__m128	lowerLimit1 = _mm_set1_ps(c.m_lowerLimit);
-	__m128	upperLimit1 = _mm_set1_ps(c.m_upperLimit);
-	__m128 deltaImpulse = _mm_sub_ps(_mm_set1_ps(c.m_rhsPenetration), _mm_mul_ps(_mm_set1_ps(c.m_appliedPushImpulse),_mm_set1_ps(c.m_cfm)));
-	__m128 deltaVel1Dotn	=	_mm_add_ps(btSimdDot3(c.m_contactNormal1.mVec128,bodyA.internalGetPushVelocity().mVec128), btSimdDot3(c.m_relpos1CrossNormal.mVec128,bodyA.internalGetTurnVelocity().mVec128));
-	__m128 deltaVel2Dotn	=	_mm_add_ps(btSimdDot3(c.m_contactNormal2.mVec128,bodyB.internalGetPushVelocity().mVec128), btSimdDot3(c.m_relpos2CrossNormal.mVec128,bodyB.internalGetTurnVelocity().mVec128));
-	deltaImpulse	=	_mm_sub_ps(deltaImpulse,_mm_mul_ps(deltaVel1Dotn,_mm_set1_ps(c.m_jacDiagABInv)));
-	deltaImpulse	=	_mm_sub_ps(deltaImpulse,_mm_mul_ps(deltaVel2Dotn,_mm_set1_ps(c.m_jacDiagABInv)));
-	btSimdScalar sum = _mm_add_ps(cpAppliedImp,deltaImpulse);
-	btSimdScalar resultLowerLess,resultUpperLess;
-	resultLowerLess = _mm_cmplt_ps(sum,lowerLimit1);
-	resultUpperLess = _mm_cmplt_ps(sum,upperLimit1);
-	__m128 lowMinApplied = _mm_sub_ps(lowerLimit1,cpAppliedImp);
-	deltaImpulse = _mm_or_ps( _mm_and_ps(resultLowerLess, lowMinApplied), _mm_andnot_ps(resultLowerLess, deltaImpulse) );
-	c.m_appliedPushImpulse = _mm_or_ps( _mm_and_ps(resultLowerLess, lowerLimit1), _mm_andnot_ps(resultLowerLess, sum) );
-	__m128	linearComponentA = _mm_mul_ps(c.m_contactNormal1.mVec128,bodyA.internalGetInvMass().mVec128);
-	__m128	linearComponentB = _mm_mul_ps(c.m_contactNormal2.mVec128,bodyB.internalGetInvMass().mVec128);
+	__m128 lowerLimit1 = _mm_set1_ps(c.m_lowerLimit);
+	__m128 upperLimit1 = _mm_set1_ps(c.m_upperLimit);
+	__m128 deltaImpulse = _mm_sub_ps(_mm_set1_ps(c.m_rhsPenetration), _mm_mul_ps(_mm_set1_ps(c.m_appliedPushImpulse), _mm_set1_ps(c.m_cfm)));
+	__m128 deltaVel1Dotn = _mm_add_ps(btSimdDot3(c.m_contactNormal1.mVec128, bodyA.internalGetPushVelocity().mVec128), btSimdDot3(c.m_relpos1CrossNormal.mVec128, bodyA.internalGetTurnVelocity().mVec128));
+	__m128 deltaVel2Dotn = _mm_add_ps(btSimdDot3(c.m_contactNormal2.mVec128, bodyB.internalGetPushVelocity().mVec128), btSimdDot3(c.m_relpos2CrossNormal.mVec128, bodyB.internalGetTurnVelocity().mVec128));
+	deltaImpulse = _mm_sub_ps(deltaImpulse, _mm_mul_ps(deltaVel1Dotn, _mm_set1_ps(c.m_jacDiagABInv)));
+	deltaImpulse = _mm_sub_ps(deltaImpulse, _mm_mul_ps(deltaVel2Dotn, _mm_set1_ps(c.m_jacDiagABInv)));
+	btSimdScalar sum = _mm_add_ps(cpAppliedImp, deltaImpulse);
+	btSimdScalar resultLowerLess, resultUpperLess;
+	resultLowerLess = _mm_cmplt_ps(sum, lowerLimit1);
+	resultUpperLess = _mm_cmplt_ps(sum, upperLimit1);
+	__m128 lowMinApplied = _mm_sub_ps(lowerLimit1, cpAppliedImp);
+	deltaImpulse = _mm_or_ps(_mm_and_ps(resultLowerLess, lowMinApplied), _mm_andnot_ps(resultLowerLess, deltaImpulse));
+	c.m_appliedPushImpulse = _mm_or_ps(_mm_and_ps(resultLowerLess, lowerLimit1), _mm_andnot_ps(resultLowerLess, sum));
+	__m128 linearComponentA = _mm_mul_ps(c.m_contactNormal1.mVec128, bodyA.internalGetInvMass().mVec128);
+	__m128 linearComponentB = _mm_mul_ps(c.m_contactNormal2.mVec128, bodyB.internalGetInvMass().mVec128);
 	__m128 impulseMagnitude = deltaImpulse;
-	bodyA.internalGetPushVelocity().mVec128 = _mm_add_ps(bodyA.internalGetPushVelocity().mVec128,_mm_mul_ps(linearComponentA,impulseMagnitude));
-	bodyA.internalGetTurnVelocity().mVec128 = _mm_add_ps(bodyA.internalGetTurnVelocity().mVec128 ,_mm_mul_ps(c.m_angularComponentA.mVec128,impulseMagnitude));
-	bodyB.internalGetPushVelocity().mVec128 = _mm_add_ps(bodyB.internalGetPushVelocity().mVec128,_mm_mul_ps(linearComponentB,impulseMagnitude));
-	bodyB.internalGetTurnVelocity().mVec128 = _mm_add_ps(bodyB.internalGetTurnVelocity().mVec128 ,_mm_mul_ps(c.m_angularComponentB.mVec128,impulseMagnitude));
+	bodyA.internalGetPushVelocity().mVec128 = _mm_add_ps(bodyA.internalGetPushVelocity().mVec128, _mm_mul_ps(linearComponentA, impulseMagnitude));
+	bodyA.internalGetTurnVelocity().mVec128 = _mm_add_ps(bodyA.internalGetTurnVelocity().mVec128, _mm_mul_ps(c.m_angularComponentA.mVec128, impulseMagnitude));
+	bodyB.internalGetPushVelocity().mVec128 = _mm_add_ps(bodyB.internalGetPushVelocity().mVec128, _mm_mul_ps(linearComponentB, impulseMagnitude));
+	bodyB.internalGetTurnVelocity().mVec128 = _mm_add_ps(bodyB.internalGetTurnVelocity().mVec128, _mm_mul_ps(c.m_angularComponentB.mVec128, impulseMagnitude));
 	btSimdScalar deltaImp = deltaImpulse;
 	return deltaImp.m_floats[0] * (1. / c.m_jacDiagABInv);
 #else
-	return gResolveSplitPenetrationImpulse_scalar_reference(bodyA,bodyB,c);
+	return gResolveSplitPenetrationImpulse_scalar_reference(bodyA, bodyB, c);
 #endif
 }
 
-
 btSequentialImpulseConstraintSolver::btSequentialImpulseConstraintSolver()
 {
-    m_btSeed2 = 0;
-    m_cachedSolverMode = 0;
-    setupSolverFunctions( false );
+	m_btSeed2 = 0;
+	m_cachedSolverMode = 0;
+	setupSolverFunctions(false);
 }
 
-void btSequentialImpulseConstraintSolver::setupSolverFunctions( bool useSimd )
+void btSequentialImpulseConstraintSolver::setupSolverFunctions(bool useSimd)
 {
-    m_resolveSingleConstraintRowGeneric = gResolveSingleConstraintRowGeneric_scalar_reference;
-    m_resolveSingleConstraintRowLowerLimit = gResolveSingleConstraintRowLowerLimit_scalar_reference;
-    m_resolveSplitPenetrationImpulse = gResolveSplitPenetrationImpulse_scalar_reference;
+	m_resolveSingleConstraintRowGeneric = gResolveSingleConstraintRowGeneric_scalar_reference;
+	m_resolveSingleConstraintRowLowerLimit = gResolveSingleConstraintRowLowerLimit_scalar_reference;
+	m_resolveSplitPenetrationImpulse = gResolveSplitPenetrationImpulse_scalar_reference;
 
-    if ( useSimd )
-    {
+	if (useSimd)
+	{
 #ifdef USE_SIMD
-        m_resolveSingleConstraintRowGeneric = gResolveSingleConstraintRowGeneric_sse2;
-        m_resolveSingleConstraintRowLowerLimit = gResolveSingleConstraintRowLowerLimit_sse2;
-        m_resolveSplitPenetrationImpulse = gResolveSplitPenetrationImpulse_sse2;
+		m_resolveSingleConstraintRowGeneric = gResolveSingleConstraintRowGeneric_sse2;
+		m_resolveSingleConstraintRowLowerLimit = gResolveSingleConstraintRowLowerLimit_sse2;
+		m_resolveSplitPenetrationImpulse = gResolveSplitPenetrationImpulse_sse2;
 
 #ifdef BT_ALLOW_SSE4
-        int cpuFeatures = btCpuFeatureUtility::getCpuFeatures();
-        if ((cpuFeatures & btCpuFeatureUtility::CPU_FEATURE_FMA3) && (cpuFeatures & btCpuFeatureUtility::CPU_FEATURE_SSE4_1))
-        {
-            m_resolveSingleConstraintRowGeneric = gResolveSingleConstraintRowGeneric_sse4_1_fma3;
-            m_resolveSingleConstraintRowLowerLimit = gResolveSingleConstraintRowLowerLimit_sse4_1_fma3;
-        }
-#endif//BT_ALLOW_SSE4
-#endif //USE_SIMD
-    }
+		int cpuFeatures = btCpuFeatureUtility::getCpuFeatures();
+		if ((cpuFeatures & btCpuFeatureUtility::CPU_FEATURE_FMA3) && (cpuFeatures & btCpuFeatureUtility::CPU_FEATURE_SSE4_1))
+		{
+			m_resolveSingleConstraintRowGeneric = gResolveSingleConstraintRowGeneric_sse4_1_fma3;
+			m_resolveSingleConstraintRowLowerLimit = gResolveSingleConstraintRowLowerLimit_sse4_1_fma3;
+		}
+#endif  //BT_ALLOW_SSE4
+#endif  //USE_SIMD
+	}
 }
 
- btSequentialImpulseConstraintSolver::~btSequentialImpulseConstraintSolver()
- {
- }
+btSequentialImpulseConstraintSolver::~btSequentialImpulseConstraintSolver()
+{
+}
 
- btSingleConstraintRowSolver	btSequentialImpulseConstraintSolver::getScalarConstraintRowSolverGeneric()
- {
-	 return gResolveSingleConstraintRowGeneric_scalar_reference;
- }
-
- btSingleConstraintRowSolver	btSequentialImpulseConstraintSolver::getScalarConstraintRowSolverLowerLimit()
- {
-	 return gResolveSingleConstraintRowLowerLimit_scalar_reference;
- }
+btSingleConstraintRowSolver btSequentialImpulseConstraintSolver::getScalarConstraintRowSolverGeneric()
+{
+	return gResolveSingleConstraintRowGeneric_scalar_reference;
+}
 
+btSingleConstraintRowSolver btSequentialImpulseConstraintSolver::getScalarConstraintRowSolverLowerLimit()
+{
+	return gResolveSingleConstraintRowLowerLimit_scalar_reference;
+}
 
 #ifdef USE_SIMD
- btSingleConstraintRowSolver	btSequentialImpulseConstraintSolver::getSSE2ConstraintRowSolverGeneric()
- {
-	 return gResolveSingleConstraintRowGeneric_sse2;
- }
- btSingleConstraintRowSolver	btSequentialImpulseConstraintSolver::getSSE2ConstraintRowSolverLowerLimit()
- {
-	 return gResolveSingleConstraintRowLowerLimit_sse2;
- }
+btSingleConstraintRowSolver btSequentialImpulseConstraintSolver::getSSE2ConstraintRowSolverGeneric()
+{
+	return gResolveSingleConstraintRowGeneric_sse2;
+}
+btSingleConstraintRowSolver btSequentialImpulseConstraintSolver::getSSE2ConstraintRowSolverLowerLimit()
+{
+	return gResolveSingleConstraintRowLowerLimit_sse2;
+}
 #ifdef BT_ALLOW_SSE4
- btSingleConstraintRowSolver	btSequentialImpulseConstraintSolver::getSSE4_1ConstraintRowSolverGeneric()
- {
-	 return gResolveSingleConstraintRowGeneric_sse4_1_fma3;
- }
- btSingleConstraintRowSolver	btSequentialImpulseConstraintSolver::getSSE4_1ConstraintRowSolverLowerLimit()
- {
-	 return gResolveSingleConstraintRowLowerLimit_sse4_1_fma3;
- }
-#endif //BT_ALLOW_SSE4
-#endif //USE_SIMD
+btSingleConstraintRowSolver btSequentialImpulseConstraintSolver::getSSE4_1ConstraintRowSolverGeneric()
+{
+	return gResolveSingleConstraintRowGeneric_sse4_1_fma3;
+}
+btSingleConstraintRowSolver btSequentialImpulseConstraintSolver::getSSE4_1ConstraintRowSolverLowerLimit()
+{
+	return gResolveSingleConstraintRowLowerLimit_sse4_1_fma3;
+}
+#endif  //BT_ALLOW_SSE4
+#endif  //USE_SIMD
 
 unsigned long btSequentialImpulseConstraintSolver::btRand2()
 {
-	m_btSeed2 = (1664525L*m_btSeed2 + 1013904223L) & 0xffffffff;
+	m_btSeed2 = (1664525L * m_btSeed2 + 1013904223L) & 0xffffffff;
 	return m_btSeed2;
 }
 
-
-
 //See ODE: adam's all-int straightforward(?) dRandInt (0..n-1)
-int btSequentialImpulseConstraintSolver::btRandInt2 (int n)
+int btSequentialImpulseConstraintSolver::btRandInt2(int n)
 {
 	// seems good; xor-fold and modulus
 	const unsigned long un = static_cast<unsigned long>(n);
@@ -448,15 +430,20 @@ int btSequentialImpulseConstraintSolver::btRandInt2 (int n)
 
 	// note: probably more aggressive than it needs to be -- might be
 	//       able to get away without one or two of the innermost branches.
-	if (un <= 0x00010000UL) {
+	if (un <= 0x00010000UL)
+	{
 		r ^= (r >> 16);
-		if (un <= 0x00000100UL) {
+		if (un <= 0x00000100UL)
+		{
 			r ^= (r >> 8);
-			if (un <= 0x00000010UL) {
+			if (un <= 0x00000010UL)
+			{
 				r ^= (r >> 4);
-				if (un <= 0x00000004UL) {
+				if (un <= 0x00000004UL)
+				{
 					r ^= (r >> 2);
-					if (un <= 0x00000002UL) {
+					if (un <= 0x00000002UL)
+					{
 						r ^= (r >> 1);
 					}
 				}
@@ -464,70 +451,56 @@ int btSequentialImpulseConstraintSolver::btRandInt2 (int n)
 		}
 	}
 
-	return (int) (r % un);
+	return (int)(r % un);
 }
 
-
-
-void	btSequentialImpulseConstraintSolver::initSolverBody(btSolverBody* solverBody, btCollisionObject* collisionObject, btScalar timeStep)
+void btSequentialImpulseConstraintSolver::initSolverBody(btSolverBody* solverBody, btCollisionObject* collisionObject, btScalar timeStep)
 {
+	btRigidBody* rb = collisionObject ? btRigidBody::upcast(collisionObject) : 0;
 
-	btRigidBody* rb = collisionObject? btRigidBody::upcast(collisionObject) : 0;
-
-	solverBody->internalGetDeltaLinearVelocity().setValue(0.f,0.f,0.f);
-	solverBody->internalGetDeltaAngularVelocity().setValue(0.f,0.f,0.f);
-	solverBody->internalGetPushVelocity().setValue(0.f,0.f,0.f);
-	solverBody->internalGetTurnVelocity().setValue(0.f,0.f,0.f);
+	solverBody->internalGetDeltaLinearVelocity().setValue(0.f, 0.f, 0.f);
+	solverBody->internalGetDeltaAngularVelocity().setValue(0.f, 0.f, 0.f);
+	solverBody->internalGetPushVelocity().setValue(0.f, 0.f, 0.f);
+	solverBody->internalGetTurnVelocity().setValue(0.f, 0.f, 0.f);
 
 	if (rb)
 	{
 		solverBody->m_worldTransform = rb->getWorldTransform();
-		solverBody->internalSetInvMass(btVector3(rb->getInvMass(),rb->getInvMass(),rb->getInvMass())*rb->getLinearFactor());
+		solverBody->internalSetInvMass(btVector3(rb->getInvMass(), rb->getInvMass(), rb->getInvMass()) * rb->getLinearFactor());
 		solverBody->m_originalBody = rb;
 		solverBody->m_angularFactor = rb->getAngularFactor();
 		solverBody->m_linearFactor = rb->getLinearFactor();
 		solverBody->m_linearVelocity = rb->getLinearVelocity();
 		solverBody->m_angularVelocity = rb->getAngularVelocity();
-		solverBody->m_externalForceImpulse = rb->getTotalForce()*rb->getInvMass()*timeStep;
-		solverBody->m_externalTorqueImpulse = rb->getTotalTorque()*rb->getInvInertiaTensorWorld()*timeStep ;
-
-	} else
+		solverBody->m_externalForceImpulse = rb->getTotalForce() * rb->getInvMass() * timeStep;
+		solverBody->m_externalTorqueImpulse = rb->getTotalTorque() * rb->getInvInertiaTensorWorld() * timeStep;
+	}
+	else
 	{
 		solverBody->m_worldTransform.setIdentity();
-		solverBody->internalSetInvMass(btVector3(0,0,0));
+		solverBody->internalSetInvMass(btVector3(0, 0, 0));
 		solverBody->m_originalBody = 0;
-		solverBody->m_angularFactor.setValue(1,1,1);
-		solverBody->m_linearFactor.setValue(1,1,1);
-		solverBody->m_linearVelocity.setValue(0,0,0);
-		solverBody->m_angularVelocity.setValue(0,0,0);
-		solverBody->m_externalForceImpulse.setValue(0,0,0);
-		solverBody->m_externalTorqueImpulse.setValue(0,0,0);
+		solverBody->m_angularFactor.setValue(1, 1, 1);
+		solverBody->m_linearFactor.setValue(1, 1, 1);
+		solverBody->m_linearVelocity.setValue(0, 0, 0);
+		solverBody->m_angularVelocity.setValue(0, 0, 0);
+		solverBody->m_externalForceImpulse.setValue(0, 0, 0);
+		solverBody->m_externalTorqueImpulse.setValue(0, 0, 0);
 	}
-
-
 }
 
-
-
-
-
-
 btScalar btSequentialImpulseConstraintSolver::restitutionCurve(btScalar rel_vel, btScalar restitution, btScalar velocityThreshold)
 {
 	//printf("rel_vel =%f\n", rel_vel);
-	if (btFabs(rel_vel)<velocityThreshold)
+	if (btFabs(rel_vel) < velocityThreshold)
 		return 0.;
 
 	btScalar rest = restitution * -rel_vel;
 	return rest;
 }
 
-
-
-void	btSequentialImpulseConstraintSolver::applyAnisotropicFriction(btCollisionObject* colObj,btVector3& frictionDirection, int frictionMode)
+void btSequentialImpulseConstraintSolver::applyAnisotropicFriction(btCollisionObject* colObj, btVector3& frictionDirection, int frictionMode)
 {
-
-
 	if (colObj && colObj->hasAnisotropicFriction(frictionMode))
 	{
 		// transform to local coordinates
@@ -538,16 +511,10 @@ void	btSequentialImpulseConstraintSolver::applyAnisotropicFriction(btCollisionOb
 		// ... and transform it back to global coordinates
 		frictionDirection = colObj->getWorldTransform().getBasis() * loc_lateral;
 	}
-
 }
 
-
-
-
-void btSequentialImpulseConstraintSolver::setupFrictionConstraint(btSolverConstraint& solverConstraint, const btVector3& normalAxis,int  solverBodyIdA,int solverBodyIdB,btManifoldPoint& cp,const btVector3& rel_pos1,const btVector3& rel_pos2,btCollisionObject* colObj0,btCollisionObject* colObj1, btScalar relaxation, const btContactSolverInfo& infoGlobal, btScalar desiredVelocity, btScalar cfmSlip)
+void btSequentialImpulseConstraintSolver::setupFrictionConstraint(btSolverConstraint& solverConstraint, const btVector3& normalAxis, int solverBodyIdA, int solverBodyIdB, btManifoldPoint& cp, const btVector3& rel_pos1, const btVector3& rel_pos2, btCollisionObject* colObj0, btCollisionObject* colObj1, btScalar relaxation, const btContactSolverInfo& infoGlobal, btScalar desiredVelocity, btScalar cfmSlip)
 {
-
-
 	btSolverBody& solverBodyA = m_tmpSolverBodyPool[solverBodyIdA];
 	btSolverBody& solverBodyB = m_tmpSolverBodyPool[solverBodyIdB];
 
@@ -568,12 +535,13 @@ void btSequentialImpulseConstraintSolver::setupFrictionConstraint(btSolverConstr
 		solverConstraint.m_contactNormal1 = normalAxis;
 		btVector3 ftorqueAxis1 = rel_pos1.cross(solverConstraint.m_contactNormal1);
 		solverConstraint.m_relpos1CrossNormal = ftorqueAxis1;
-		solverConstraint.m_angularComponentA = body0->getInvInertiaTensorWorld()*ftorqueAxis1*body0->getAngularFactor();
-	}else
+		solverConstraint.m_angularComponentA = body0->getInvInertiaTensorWorld() * ftorqueAxis1 * body0->getAngularFactor();
+	}
+	else
 	{
 		solverConstraint.m_contactNormal1.setZero();
 		solverConstraint.m_relpos1CrossNormal.setZero();
-		solverConstraint.m_angularComponentA .setZero();
+		solverConstraint.m_angularComponentA.setZero();
 	}
 
 	if (bodyA)
@@ -581,8 +549,9 @@ void btSequentialImpulseConstraintSolver::setupFrictionConstraint(btSolverConstr
 		solverConstraint.m_contactNormal2 = -normalAxis;
 		btVector3 ftorqueAxis1 = rel_pos2.cross(solverConstraint.m_contactNormal2);
 		solverConstraint.m_relpos2CrossNormal = ftorqueAxis1;
-		solverConstraint.m_angularComponentB = bodyA->getInvInertiaTensorWorld()*ftorqueAxis1*bodyA->getAngularFactor();
-	} else
+		solverConstraint.m_angularComponentB = bodyA->getInvInertiaTensorWorld() * ftorqueAxis1 * bodyA->getAngularFactor();
+	}
+	else
 	{
 		solverConstraint.m_contactNormal2.setZero();
 		solverConstraint.m_relpos2CrossNormal.setZero();
@@ -595,32 +564,28 @@ void btSequentialImpulseConstraintSolver::setupFrictionConstraint(btSolverConstr
 		btScalar denom1 = 0.f;
 		if (body0)
 		{
-			vec = ( solverConstraint.m_angularComponentA).cross(rel_pos1);
+			vec = (solverConstraint.m_angularComponentA).cross(rel_pos1);
 			denom0 = body0->getInvMass() + normalAxis.dot(vec);
 		}
 		if (bodyA)
 		{
-			vec = ( -solverConstraint.m_angularComponentB).cross(rel_pos2);
+			vec = (-solverConstraint.m_angularComponentB).cross(rel_pos2);
 			denom1 = bodyA->getInvMass() + normalAxis.dot(vec);
 		}
-		btScalar denom = relaxation/(denom0+denom1);
+		btScalar denom = relaxation / (denom0 + denom1);
 		solverConstraint.m_jacDiagABInv = denom;
 	}
 
 	{
-
-
 		btScalar rel_vel;
-		btScalar vel1Dotn = solverConstraint.m_contactNormal1.dot(body0?solverBodyA.m_linearVelocity+solverBodyA.m_externalForceImpulse:btVector3(0,0,0))
-			+ solverConstraint.m_relpos1CrossNormal.dot(body0?solverBodyA.m_angularVelocity:btVector3(0,0,0));
-		btScalar vel2Dotn = solverConstraint.m_contactNormal2.dot(bodyA?solverBodyB.m_linearVelocity+solverBodyB.m_externalForceImpulse:btVector3(0,0,0))
-			+ solverConstraint.m_relpos2CrossNormal.dot(bodyA?solverBodyB.m_angularVelocity:btVector3(0,0,0));
+		btScalar vel1Dotn = solverConstraint.m_contactNormal1.dot(body0 ? solverBodyA.m_linearVelocity + solverBodyA.m_externalForceImpulse : btVector3(0, 0, 0)) + solverConstraint.m_relpos1CrossNormal.dot(body0 ? solverBodyA.m_angularVelocity : btVector3(0, 0, 0));
+		btScalar vel2Dotn = solverConstraint.m_contactNormal2.dot(bodyA ? solverBodyB.m_linearVelocity + solverBodyB.m_externalForceImpulse : btVector3(0, 0, 0)) + solverConstraint.m_relpos2CrossNormal.dot(bodyA ? solverBodyB.m_angularVelocity : btVector3(0, 0, 0));
 
-		rel_vel = vel1Dotn+vel2Dotn;
+		rel_vel = vel1Dotn + vel2Dotn;
 
-//		btScalar positionalError = 0.f;
+		//		btScalar positionalError = 0.f;
 
-		btScalar velocityError =  desiredVelocity - rel_vel;
+		btScalar velocityError = desiredVelocity - rel_vel;
 		btScalar velocityImpulse = velocityError * solverConstraint.m_jacDiagABInv;
 
 		btScalar penetrationImpulse = btScalar(0);
@@ -628,8 +593,8 @@ void btSequentialImpulseConstraintSolver::setupFrictionConstraint(btSolverConstr
 		if (cp.m_contactPointFlags & BT_CONTACT_FLAG_FRICTION_ANCHOR)
 		{
 			btScalar distance = (cp.getPositionWorldOnA() - cp.getPositionWorldOnB()).dot(normalAxis);
-			btScalar positionalError = -distance * infoGlobal.m_frictionERP/infoGlobal.m_timeStep;
-			penetrationImpulse = positionalError*solverConstraint.m_jacDiagABInv;
+			btScalar positionalError = -distance * infoGlobal.m_frictionERP / infoGlobal.m_timeStep;
+			penetrationImpulse = positionalError * solverConstraint.m_jacDiagABInv;
 		}
 
 		solverConstraint.m_rhs = penetrationImpulse + velocityImpulse;
@@ -637,11 +602,10 @@ void btSequentialImpulseConstraintSolver::setupFrictionConstraint(btSolverConstr
 		solverConstraint.m_cfm = cfmSlip;
 		solverConstraint.m_lowerLimit = -solverConstraint.m_friction;
 		solverConstraint.m_upperLimit = solverConstraint.m_friction;
-
 	}
 }
 
-btSolverConstraint&	btSequentialImpulseConstraintSolver::addFrictionConstraint(const btVector3& normalAxis,int solverBodyIdA,int solverBodyIdB,int frictionIndex,btManifoldPoint& cp,const btVector3& rel_pos1,const btVector3& rel_pos2,btCollisionObject* colObj0,btCollisionObject* colObj1, btScalar relaxation, const btContactSolverInfo& infoGlobal, btScalar desiredVelocity, btScalar cfmSlip)
+btSolverConstraint& btSequentialImpulseConstraintSolver::addFrictionConstraint(const btVector3& normalAxis, int solverBodyIdA, int solverBodyIdB, int frictionIndex, btManifoldPoint& cp, const btVector3& rel_pos1, const btVector3& rel_pos2, btCollisionObject* colObj0, btCollisionObject* colObj1, btScalar relaxation, const btContactSolverInfo& infoGlobal, btScalar desiredVelocity, btScalar cfmSlip)
 {
 	btSolverConstraint& solverConstraint = m_tmpSolverContactFrictionConstraintPool.expandNonInitializing();
 	solverConstraint.m_frictionIndex = frictionIndex;
@@ -650,15 +614,13 @@ btSolverConstraint&	btSequentialImpulseConstraintSolver::addFrictionConstraint(c
 	return solverConstraint;
 }
 
-
-void btSequentialImpulseConstraintSolver::setupTorsionalFrictionConstraint(	btSolverConstraint& solverConstraint, const btVector3& normalAxis1,int solverBodyIdA,int  solverBodyIdB,
-									btManifoldPoint& cp,btScalar combinedTorsionalFriction, const btVector3& rel_pos1,const btVector3& rel_pos2,
-									btCollisionObject* colObj0,btCollisionObject* colObj1, btScalar relaxation,
-									btScalar desiredVelocity, btScalar cfmSlip)
+void btSequentialImpulseConstraintSolver::setupTorsionalFrictionConstraint(btSolverConstraint& solverConstraint, const btVector3& normalAxis1, int solverBodyIdA, int solverBodyIdB,
+																		   btManifoldPoint& cp, btScalar combinedTorsionalFriction, const btVector3& rel_pos1, const btVector3& rel_pos2,
+																		   btCollisionObject* colObj0, btCollisionObject* colObj1, btScalar relaxation,
+																		   btScalar desiredVelocity, btScalar cfmSlip)
 
 {
-	btVector3 normalAxis(0,0,0);
-
+	btVector3 normalAxis(0, 0, 0);
 
 	solverConstraint.m_contactNormal1 = normalAxis;
 	solverConstraint.m_contactNormal2 = -normalAxis;
@@ -671,8 +633,8 @@ void btSequentialImpulseConstraintSolver::setupTorsionalFrictionConstraint(	btSo
 	solverConstraint.m_solverBodyIdA = solverBodyIdA;
 	solverConstraint.m_solverBodyIdB = solverBodyIdB;
 
-    solverConstraint.m_friction = combinedTorsionalFriction;
-    solverConstraint.m_originalContactPoint = 0;
+	solverConstraint.m_friction = combinedTorsionalFriction;
+	solverConstraint.m_originalContactPoint = 0;
 
 	solverConstraint.m_appliedImpulse = 0.f;
 	solverConstraint.m_appliedPushImpulse = 0.f;
@@ -680,138 +642,125 @@ void btSequentialImpulseConstraintSolver::setupTorsionalFrictionConstraint(	btSo
 	{
 		btVector3 ftorqueAxis1 = -normalAxis1;
 		solverConstraint.m_relpos1CrossNormal = ftorqueAxis1;
-		solverConstraint.m_angularComponentA = body0 ? body0->getInvInertiaTensorWorld()*ftorqueAxis1*body0->getAngularFactor() : btVector3(0,0,0);
+		solverConstraint.m_angularComponentA = body0 ? body0->getInvInertiaTensorWorld() * ftorqueAxis1 * body0->getAngularFactor() : btVector3(0, 0, 0);
 	}
 	{
 		btVector3 ftorqueAxis1 = normalAxis1;
 		solverConstraint.m_relpos2CrossNormal = ftorqueAxis1;
-		solverConstraint.m_angularComponentB = bodyA ? bodyA->getInvInertiaTensorWorld()*ftorqueAxis1*bodyA->getAngularFactor() : btVector3(0,0,0);
+		solverConstraint.m_angularComponentB = bodyA ? bodyA->getInvInertiaTensorWorld() * ftorqueAxis1 * bodyA->getAngularFactor() : btVector3(0, 0, 0);
 	}
 
-
 	{
-		btVector3 iMJaA = body0?body0->getInvInertiaTensorWorld()*solverConstraint.m_relpos1CrossNormal:btVector3(0,0,0);
-		btVector3 iMJaB = bodyA?bodyA->getInvInertiaTensorWorld()*solverConstraint.m_relpos2CrossNormal:btVector3(0,0,0);
+		btVector3 iMJaA = body0 ? body0->getInvInertiaTensorWorld() * solverConstraint.m_relpos1CrossNormal : btVector3(0, 0, 0);
+		btVector3 iMJaB = bodyA ? bodyA->getInvInertiaTensorWorld() * solverConstraint.m_relpos2CrossNormal : btVector3(0, 0, 0);
 		btScalar sum = 0;
 		sum += iMJaA.dot(solverConstraint.m_relpos1CrossNormal);
 		sum += iMJaB.dot(solverConstraint.m_relpos2CrossNormal);
-		solverConstraint.m_jacDiagABInv = btScalar(1.)/sum;
+		solverConstraint.m_jacDiagABInv = btScalar(1.) / sum;
 	}
 
 	{
-
-
 		btScalar rel_vel;
-		btScalar vel1Dotn = solverConstraint.m_contactNormal1.dot(body0?solverBodyA.m_linearVelocity+solverBodyA.m_externalForceImpulse:btVector3(0,0,0))
-			+ solverConstraint.m_relpos1CrossNormal.dot(body0?solverBodyA.m_angularVelocity:btVector3(0,0,0));
-		btScalar vel2Dotn = solverConstraint.m_contactNormal2.dot(bodyA?solverBodyB.m_linearVelocity+solverBodyB.m_externalForceImpulse:btVector3(0,0,0))
-			+ solverConstraint.m_relpos2CrossNormal.dot(bodyA?solverBodyB.m_angularVelocity:btVector3(0,0,0));
+		btScalar vel1Dotn = solverConstraint.m_contactNormal1.dot(body0 ? solverBodyA.m_linearVelocity + solverBodyA.m_externalForceImpulse : btVector3(0, 0, 0)) + solverConstraint.m_relpos1CrossNormal.dot(body0 ? solverBodyA.m_angularVelocity : btVector3(0, 0, 0));
+		btScalar vel2Dotn = solverConstraint.m_contactNormal2.dot(bodyA ? solverBodyB.m_linearVelocity + solverBodyB.m_externalForceImpulse : btVector3(0, 0, 0)) + solverConstraint.m_relpos2CrossNormal.dot(bodyA ? solverBodyB.m_angularVelocity : btVector3(0, 0, 0));
 
-		rel_vel = vel1Dotn+vel2Dotn;
+		rel_vel = vel1Dotn + vel2Dotn;
 
-//		btScalar positionalError = 0.f;
+		//		btScalar positionalError = 0.f;
 
-		btSimdScalar velocityError =  desiredVelocity - rel_vel;
-		btSimdScalar	velocityImpulse = velocityError * btSimdScalar(solverConstraint.m_jacDiagABInv);
+		btSimdScalar velocityError = desiredVelocity - rel_vel;
+		btSimdScalar velocityImpulse = velocityError * btSimdScalar(solverConstraint.m_jacDiagABInv);
 		solverConstraint.m_rhs = velocityImpulse;
 		solverConstraint.m_cfm = cfmSlip;
 		solverConstraint.m_lowerLimit = -solverConstraint.m_friction;
 		solverConstraint.m_upperLimit = solverConstraint.m_friction;
-
 	}
 }
 
-
-
-
-
-
-
-
-btSolverConstraint&	btSequentialImpulseConstraintSolver::addTorsionalFrictionConstraint(const btVector3& normalAxis,int solverBodyIdA,int solverBodyIdB,int frictionIndex,btManifoldPoint& cp,btScalar combinedTorsionalFriction, const btVector3& rel_pos1,const btVector3& rel_pos2,btCollisionObject* colObj0,btCollisionObject* colObj1, btScalar relaxation, btScalar desiredVelocity, btScalar cfmSlip)
+btSolverConstraint& btSequentialImpulseConstraintSolver::addTorsionalFrictionConstraint(const btVector3& normalAxis, int solverBodyIdA, int solverBodyIdB, int frictionIndex, btManifoldPoint& cp, btScalar combinedTorsionalFriction, const btVector3& rel_pos1, const btVector3& rel_pos2, btCollisionObject* colObj0, btCollisionObject* colObj1, btScalar relaxation, btScalar desiredVelocity, btScalar cfmSlip)
 {
 	btSolverConstraint& solverConstraint = m_tmpSolverContactRollingFrictionConstraintPool.expandNonInitializing();
 	solverConstraint.m_frictionIndex = frictionIndex;
-	setupTorsionalFrictionConstraint(solverConstraint, normalAxis, solverBodyIdA, solverBodyIdB, cp, combinedTorsionalFriction,rel_pos1, rel_pos2,
-							colObj0, colObj1, relaxation, desiredVelocity, cfmSlip);
+	setupTorsionalFrictionConstraint(solverConstraint, normalAxis, solverBodyIdA, solverBodyIdB, cp, combinedTorsionalFriction, rel_pos1, rel_pos2,
+									 colObj0, colObj1, relaxation, desiredVelocity, cfmSlip);
 	return solverConstraint;
 }
 
-
-int	btSequentialImpulseConstraintSolver::getOrInitSolverBody(btCollisionObject& body,btScalar timeStep)
+int btSequentialImpulseConstraintSolver::getOrInitSolverBody(btCollisionObject& body, btScalar timeStep)
 {
 #if BT_THREADSAFE
-    int solverBodyId = -1;
-    bool isRigidBodyType = btRigidBody::upcast( &body ) != NULL;
-    if ( isRigidBodyType && !body.isStaticOrKinematicObject() )
-    {
-        // dynamic body
-        // Dynamic bodies can only be in one island, so it's safe to write to the companionId
-        solverBodyId = body.getCompanionId();
-        if ( solverBodyId < 0 )
-        {
-            solverBodyId = m_tmpSolverBodyPool.size();
-            btSolverBody& solverBody = m_tmpSolverBodyPool.expand();
-            initSolverBody( &solverBody, &body, timeStep );
-            body.setCompanionId( solverBodyId );
-        }
-    }
-    else if (isRigidBodyType && body.isKinematicObject())
-    {
-        //
-        // NOTE: must test for kinematic before static because some kinematic objects also
-        //   identify as "static"
-        //
-        // Kinematic bodies can be in multiple islands at once, so it is a
-        // race condition to write to them, so we use an alternate method
-        // to record the solverBodyId
-        int uniqueId = body.getWorldArrayIndex();
-        const int INVALID_SOLVER_BODY_ID = -1;
-        if (uniqueId >= m_kinematicBodyUniqueIdToSolverBodyTable.size())
-        {
-            m_kinematicBodyUniqueIdToSolverBodyTable.resize(uniqueId + 1, INVALID_SOLVER_BODY_ID);
-        }
-        solverBodyId = m_kinematicBodyUniqueIdToSolverBodyTable[ uniqueId ];
-        // if no table entry yet,
-        if ( solverBodyId == INVALID_SOLVER_BODY_ID )
-        {
-            // create a table entry for this body
-            solverBodyId = m_tmpSolverBodyPool.size();
-            btSolverBody& solverBody = m_tmpSolverBodyPool.expand();
-            initSolverBody( &solverBody, &body, timeStep );
-            m_kinematicBodyUniqueIdToSolverBodyTable[ uniqueId ] = solverBodyId;
-        }
-    }
-    else
-    {
-		bool isMultiBodyType = (body.getInternalType()&btCollisionObject::CO_FEATHERSTONE_LINK);
-        // Incorrectly set collision object flags can degrade performance in various ways.
+	int solverBodyId = -1;
+	bool isRigidBodyType = btRigidBody::upcast(&body) != NULL;
+	if (isRigidBodyType && !body.isStaticOrKinematicObject())
+	{
+		// dynamic body
+		// Dynamic bodies can only be in one island, so it's safe to write to the companionId
+		solverBodyId = body.getCompanionId();
+		if (solverBodyId < 0)
+		{
+			solverBodyId = m_tmpSolverBodyPool.size();
+			btSolverBody& solverBody = m_tmpSolverBodyPool.expand();
+			initSolverBody(&solverBody, &body, timeStep);
+			body.setCompanionId(solverBodyId);
+		}
+	}
+	else if (isRigidBodyType && body.isKinematicObject())
+	{
+		//
+		// NOTE: must test for kinematic before static because some kinematic objects also
+		//   identify as "static"
+		//
+		// Kinematic bodies can be in multiple islands at once, so it is a
+		// race condition to write to them, so we use an alternate method
+		// to record the solverBodyId
+		int uniqueId = body.getWorldArrayIndex();
+		const int INVALID_SOLVER_BODY_ID = -1;
+		if (uniqueId >= m_kinematicBodyUniqueIdToSolverBodyTable.size())
+		{
+			m_kinematicBodyUniqueIdToSolverBodyTable.resize(uniqueId + 1, INVALID_SOLVER_BODY_ID);
+		}
+		solverBodyId = m_kinematicBodyUniqueIdToSolverBodyTable[uniqueId];
+		// if no table entry yet,
+		if (solverBodyId == INVALID_SOLVER_BODY_ID)
+		{
+			// create a table entry for this body
+			solverBodyId = m_tmpSolverBodyPool.size();
+			btSolverBody& solverBody = m_tmpSolverBodyPool.expand();
+			initSolverBody(&solverBody, &body, timeStep);
+			m_kinematicBodyUniqueIdToSolverBodyTable[uniqueId] = solverBodyId;
+		}
+	}
+	else
+	{
+		bool isMultiBodyType = (body.getInternalType() & btCollisionObject::CO_FEATHERSTONE_LINK);
+		// Incorrectly set collision object flags can degrade performance in various ways.
 		if (!isMultiBodyType)
 		{
-			btAssert( body.isStaticOrKinematicObject() );
+			btAssert(body.isStaticOrKinematicObject());
 		}
-        //it could be a multibody link collider
-        // all fixed bodies (inf mass) get mapped to a single solver id
-        if ( m_fixedBodyId < 0 )
-        {
-            m_fixedBodyId = m_tmpSolverBodyPool.size();
-            btSolverBody& fixedBody = m_tmpSolverBodyPool.expand();
-            initSolverBody( &fixedBody, 0, timeStep );
-        }
-        solverBodyId = m_fixedBodyId;
-    }
-    btAssert( solverBodyId >= 0 && solverBodyId < m_tmpSolverBodyPool.size() );
+		//it could be a multibody link collider
+		// all fixed bodies (inf mass) get mapped to a single solver id
+		if (m_fixedBodyId < 0)
+		{
+			m_fixedBodyId = m_tmpSolverBodyPool.size();
+			btSolverBody& fixedBody = m_tmpSolverBodyPool.expand();
+			initSolverBody(&fixedBody, 0, timeStep);
+		}
+		solverBodyId = m_fixedBodyId;
+	}
+	btAssert(solverBodyId >= 0 && solverBodyId < m_tmpSolverBodyPool.size());
 	return solverBodyId;
-#else // BT_THREADSAFE
+#else   // BT_THREADSAFE
 
-    int solverBodyIdA = -1;
+	int solverBodyIdA = -1;
 
 	if (body.getCompanionId() >= 0)
 	{
 		//body has already been converted
 		solverBodyIdA = body.getCompanionId();
-        btAssert(solverBodyIdA < m_tmpSolverBodyPool.size());
-	} else
+		btAssert(solverBodyIdA < m_tmpSolverBodyPool.size());
+	}
+	else
 	{
 		btRigidBody* rb = btRigidBody::upcast(&body);
 		//convert both active and kinematic objects (for their velocity)
@@ -819,233 +768,216 @@ int	btSequentialImpulseConstraintSolver::getOrInitSolverBody(btCollisionObject&
 		{
 			solverBodyIdA = m_tmpSolverBodyPool.size();
 			btSolverBody& solverBody = m_tmpSolverBodyPool.expand();
-			initSolverBody(&solverBody,&body,timeStep);
+			initSolverBody(&solverBody, &body, timeStep);
 			body.setCompanionId(solverBodyIdA);
-		} else
+		}
+		else
 		{
-
-			if (m_fixedBodyId<0)
+			if (m_fixedBodyId < 0)
 			{
 				m_fixedBodyId = m_tmpSolverBodyPool.size();
 				btSolverBody& fixedBody = m_tmpSolverBodyPool.expand();
-				initSolverBody(&fixedBody,0,timeStep);
+				initSolverBody(&fixedBody, 0, timeStep);
 			}
 			return m_fixedBodyId;
-//			return 0;//assume first one is a fixed solver body
+			//			return 0;//assume first one is a fixed solver body
 		}
 	}
 
 	return solverBodyIdA;
-#endif // BT_THREADSAFE
-
+#endif  // BT_THREADSAFE
 }
 #include <stdio.h>
 
-
 void btSequentialImpulseConstraintSolver::setupContactConstraint(btSolverConstraint& solverConstraint,
 																 int solverBodyIdA, int solverBodyIdB,
 																 btManifoldPoint& cp, const btContactSolverInfo& infoGlobal,
 																 btScalar& relaxation,
 																 const btVector3& rel_pos1, const btVector3& rel_pos2)
 {
+	//	const btVector3& pos1 = cp.getPositionWorldOnA();
+	//	const btVector3& pos2 = cp.getPositionWorldOnB();
 
-		//	const btVector3& pos1 = cp.getPositionWorldOnA();
-		//	const btVector3& pos2 = cp.getPositionWorldOnB();
+	btSolverBody* bodyA = &m_tmpSolverBodyPool[solverBodyIdA];
+	btSolverBody* bodyB = &m_tmpSolverBodyPool[solverBodyIdB];
 
-			btSolverBody* bodyA = &m_tmpSolverBodyPool[solverBodyIdA];
-			btSolverBody* bodyB = &m_tmpSolverBodyPool[solverBodyIdB];
+	btRigidBody* rb0 = bodyA->m_originalBody;
+	btRigidBody* rb1 = bodyB->m_originalBody;
 
-			btRigidBody* rb0 = bodyA->m_originalBody;
-			btRigidBody* rb1 = bodyB->m_originalBody;
+	//			btVector3 rel_pos1 = pos1 - colObj0->getWorldTransform().getOrigin();
+	//			btVector3 rel_pos2 = pos2 - colObj1->getWorldTransform().getOrigin();
+	//rel_pos1 = pos1 - bodyA->getWorldTransform().getOrigin();
+	//rel_pos2 = pos2 - bodyB->getWorldTransform().getOrigin();
 
-//			btVector3 rel_pos1 = pos1 - colObj0->getWorldTransform().getOrigin();
-//			btVector3 rel_pos2 = pos2 - colObj1->getWorldTransform().getOrigin();
-			//rel_pos1 = pos1 - bodyA->getWorldTransform().getOrigin();
-			//rel_pos2 = pos2 - bodyB->getWorldTransform().getOrigin();
+	relaxation = infoGlobal.m_sor;
+	btScalar invTimeStep = btScalar(1) / infoGlobal.m_timeStep;
 
-			relaxation = infoGlobal.m_sor;
-			btScalar invTimeStep = btScalar(1)/infoGlobal.m_timeStep;
+	//cfm = 1 /       ( dt * kp + kd )
+	//erp = dt * kp / ( dt * kp + kd )
 
-            //cfm = 1 /       ( dt * kp + kd )
-            //erp = dt * kp / ( dt * kp + kd )
-            
-            btScalar cfm = infoGlobal.m_globalCfm;
-            btScalar erp = infoGlobal.m_erp2;
+	btScalar cfm = infoGlobal.m_globalCfm;
+	btScalar erp = infoGlobal.m_erp2;
 
-            if ((cp.m_contactPointFlags&BT_CONTACT_FLAG_HAS_CONTACT_CFM) || (cp.m_contactPointFlags&BT_CONTACT_FLAG_HAS_CONTACT_ERP))
-            {
-                if (cp.m_contactPointFlags&BT_CONTACT_FLAG_HAS_CONTACT_CFM)
-                    cfm  = cp.m_contactCFM;
-                if (cp.m_contactPointFlags&BT_CONTACT_FLAG_HAS_CONTACT_ERP)
-                    erp = cp.m_contactERP;                
-            } else
-            {
-                if (cp.m_contactPointFlags & BT_CONTACT_FLAG_CONTACT_STIFFNESS_DAMPING)
-                {
-                    btScalar denom = ( infoGlobal.m_timeStep * cp.m_combinedContactStiffness1 + cp.m_combinedContactDamping1 );
-                    if (denom < SIMD_EPSILON)
-                    {
-                        denom = SIMD_EPSILON;
-                    }
-                    cfm = btScalar(1) / denom; 
-                    erp = (infoGlobal.m_timeStep * cp.m_combinedContactStiffness1) / denom;
-                }
-            }
-            
-            cfm *= invTimeStep;
-            
+	if ((cp.m_contactPointFlags & BT_CONTACT_FLAG_HAS_CONTACT_CFM) || (cp.m_contactPointFlags & BT_CONTACT_FLAG_HAS_CONTACT_ERP))
+	{
+		if (cp.m_contactPointFlags & BT_CONTACT_FLAG_HAS_CONTACT_CFM)
+			cfm = cp.m_contactCFM;
+		if (cp.m_contactPointFlags & BT_CONTACT_FLAG_HAS_CONTACT_ERP)
+			erp = cp.m_contactERP;
+	}
+	else
+	{
+		if (cp.m_contactPointFlags & BT_CONTACT_FLAG_CONTACT_STIFFNESS_DAMPING)
+		{
+			btScalar denom = (infoGlobal.m_timeStep * cp.m_combinedContactStiffness1 + cp.m_combinedContactDamping1);
+			if (denom < SIMD_EPSILON)
+			{
+				denom = SIMD_EPSILON;
+			}
+			cfm = btScalar(1) / denom;
+			erp = (infoGlobal.m_timeStep * cp.m_combinedContactStiffness1) / denom;
+		}
+	}
 
-			btVector3 torqueAxis0 = rel_pos1.cross(cp.m_normalWorldOnB);
-			solverConstraint.m_angularComponentA = rb0 ? rb0->getInvInertiaTensorWorld()*torqueAxis0*rb0->getAngularFactor() : btVector3(0,0,0);
-			btVector3 torqueAxis1 = rel_pos2.cross(cp.m_normalWorldOnB);
-			solverConstraint.m_angularComponentB = rb1 ? rb1->getInvInertiaTensorWorld()*-torqueAxis1*rb1->getAngularFactor() : btVector3(0,0,0);
+	cfm *= invTimeStep;
 
-				{
+	btVector3 torqueAxis0 = rel_pos1.cross(cp.m_normalWorldOnB);
+	solverConstraint.m_angularComponentA = rb0 ? rb0->getInvInertiaTensorWorld() * torqueAxis0 * rb0->getAngularFactor() : btVector3(0, 0, 0);
+	btVector3 torqueAxis1 = rel_pos2.cross(cp.m_normalWorldOnB);
+	solverConstraint.m_angularComponentB = rb1 ? rb1->getInvInertiaTensorWorld() * -torqueAxis1 * rb1->getAngularFactor() : btVector3(0, 0, 0);
+
+	{
 #ifdef COMPUTE_IMPULSE_DENOM
-					btScalar denom0 = rb0->computeImpulseDenominator(pos1,cp.m_normalWorldOnB);
-					btScalar denom1 = rb1->computeImpulseDenominator(pos2,cp.m_normalWorldOnB);
+		btScalar denom0 = rb0->computeImpulseDenominator(pos1, cp.m_normalWorldOnB);
+		btScalar denom1 = rb1->computeImpulseDenominator(pos2, cp.m_normalWorldOnB);
 #else
-					btVector3 vec;
-					btScalar denom0 = 0.f;
-					btScalar denom1 = 0.f;
-					if (rb0)
-					{
-						vec = ( solverConstraint.m_angularComponentA).cross(rel_pos1);
-						denom0 = rb0->getInvMass() + cp.m_normalWorldOnB.dot(vec);
-					}
-					if (rb1)
-					{
-						vec = ( -solverConstraint.m_angularComponentB).cross(rel_pos2);
-						denom1 = rb1->getInvMass() + cp.m_normalWorldOnB.dot(vec);
-					}
-#endif //COMPUTE_IMPULSE_DENOM
+		btVector3 vec;
+		btScalar denom0 = 0.f;
+		btScalar denom1 = 0.f;
+		if (rb0)
+		{
+			vec = (solverConstraint.m_angularComponentA).cross(rel_pos1);
+			denom0 = rb0->getInvMass() + cp.m_normalWorldOnB.dot(vec);
+		}
+		if (rb1)
+		{
+			vec = (-solverConstraint.m_angularComponentB).cross(rel_pos2);
+			denom1 = rb1->getInvMass() + cp.m_normalWorldOnB.dot(vec);
+		}
+#endif  //COMPUTE_IMPULSE_DENOM
 
-					btScalar denom = relaxation/(denom0+denom1+cfm);
-					solverConstraint.m_jacDiagABInv = denom;
-				}
+		btScalar denom = relaxation / (denom0 + denom1 + cfm);
+		solverConstraint.m_jacDiagABInv = denom;
+	}
 
-				if (rb0)
-				{
-					solverConstraint.m_contactNormal1 = cp.m_normalWorldOnB;
-					solverConstraint.m_relpos1CrossNormal = torqueAxis0;
-				} else
-				{
-					solverConstraint.m_contactNormal1.setZero();
-					solverConstraint.m_relpos1CrossNormal.setZero();
-				}
-				if (rb1)
-				{
-					solverConstraint.m_contactNormal2 = -cp.m_normalWorldOnB;
-					solverConstraint.m_relpos2CrossNormal = -torqueAxis1;
-				}else
-				{
-					solverConstraint.m_contactNormal2.setZero();
-					solverConstraint.m_relpos2CrossNormal.setZero();
-				}
+	if (rb0)
+	{
+		solverConstraint.m_contactNormal1 = cp.m_normalWorldOnB;
+		solverConstraint.m_relpos1CrossNormal = torqueAxis0;
+	}
+	else
+	{
+		solverConstraint.m_contactNormal1.setZero();
+		solverConstraint.m_relpos1CrossNormal.setZero();
+	}
+	if (rb1)
+	{
+		solverConstraint.m_contactNormal2 = -cp.m_normalWorldOnB;
+		solverConstraint.m_relpos2CrossNormal = -torqueAxis1;
+	}
+	else
+	{
+		solverConstraint.m_contactNormal2.setZero();
+		solverConstraint.m_relpos2CrossNormal.setZero();
+	}
 
-				btScalar restitution = 0.f;
-				btScalar penetration = cp.getDistance()+infoGlobal.m_linearSlop;
+	btScalar restitution = 0.f;
+	btScalar penetration = cp.getDistance() + infoGlobal.m_linearSlop;
 
-				{
-					btVector3 vel1,vel2;
+	{
+		btVector3 vel1, vel2;
 
-					vel1 = rb0? rb0->getVelocityInLocalPoint(rel_pos1) : btVector3(0,0,0);
-					vel2 = rb1? rb1->getVelocityInLocalPoint(rel_pos2) : btVector3(0,0,0);
+		vel1 = rb0 ? rb0->getVelocityInLocalPoint(rel_pos1) : btVector3(0, 0, 0);
+		vel2 = rb1 ? rb1->getVelocityInLocalPoint(rel_pos2) : btVector3(0, 0, 0);
 
-	//			btVector3 vel2 = rb1 ? rb1->getVelocityInLocalPoint(rel_pos2) : btVector3(0,0,0);
-					btVector3 vel  = vel1 - vel2;
-					btScalar rel_vel = cp.m_normalWorldOnB.dot(vel);
+		//			btVector3 vel2 = rb1 ? rb1->getVelocityInLocalPoint(rel_pos2) : btVector3(0,0,0);
+		btVector3 vel = vel1 - vel2;
+		btScalar rel_vel = cp.m_normalWorldOnB.dot(vel);
 
+		solverConstraint.m_friction = cp.m_combinedFriction;
 
+		restitution = restitutionCurve(rel_vel, cp.m_combinedRestitution, infoGlobal.m_restitutionVelocityThreshold);
+		if (restitution <= btScalar(0.))
+		{
+			restitution = 0.f;
+		};
+	}
 
-					solverConstraint.m_friction = cp.m_combinedFriction;
+	///warm starting (or zero if disabled)
+	if (infoGlobal.m_solverMode & SOLVER_USE_WARMSTARTING)
+	{
+		solverConstraint.m_appliedImpulse = cp.m_appliedImpulse * infoGlobal.m_warmstartingFactor;
+		if (rb0)
+			bodyA->internalApplyImpulse(solverConstraint.m_contactNormal1 * bodyA->internalGetInvMass(), solverConstraint.m_angularComponentA, solverConstraint.m_appliedImpulse);
+		if (rb1)
+			bodyB->internalApplyImpulse(-solverConstraint.m_contactNormal2 * bodyB->internalGetInvMass() , -solverConstraint.m_angularComponentB, -(btScalar)solverConstraint.m_appliedImpulse);
+	}
+	else
+	{
+		solverConstraint.m_appliedImpulse = 0.f;
+	}
 
+	solverConstraint.m_appliedPushImpulse = 0.f;
 
-					restitution =  restitutionCurve(rel_vel, cp.m_combinedRestitution, infoGlobal.m_restitutionVelocityThreshold);
-					if (restitution <= btScalar(0.))
-					{
-						restitution = 0.f;
-					};
-				}
+	{
+		btVector3 externalForceImpulseA = bodyA->m_originalBody ? bodyA->m_externalForceImpulse : btVector3(0, 0, 0);
+		btVector3 externalTorqueImpulseA = bodyA->m_originalBody ? bodyA->m_externalTorqueImpulse : btVector3(0, 0, 0);
+		btVector3 externalForceImpulseB = bodyB->m_originalBody ? bodyB->m_externalForceImpulse : btVector3(0, 0, 0);
+		btVector3 externalTorqueImpulseB = bodyB->m_originalBody ? bodyB->m_externalTorqueImpulse : btVector3(0, 0, 0);
 
+		btScalar vel1Dotn = solverConstraint.m_contactNormal1.dot(bodyA->m_linearVelocity + externalForceImpulseA) + solverConstraint.m_relpos1CrossNormal.dot(bodyA->m_angularVelocity + externalTorqueImpulseA);
+		btScalar vel2Dotn = solverConstraint.m_contactNormal2.dot(bodyB->m_linearVelocity + externalForceImpulseB) + solverConstraint.m_relpos2CrossNormal.dot(bodyB->m_angularVelocity + externalTorqueImpulseB);
+		btScalar rel_vel = vel1Dotn + vel2Dotn;
 
-				///warm starting (or zero if disabled)
-				if (infoGlobal.m_solverMode & SOLVER_USE_WARMSTARTING)
-				{
-					solverConstraint.m_appliedImpulse = cp.m_appliedImpulse * infoGlobal.m_warmstartingFactor;
-					if (rb0)
-						bodyA->internalApplyImpulse(solverConstraint.m_contactNormal1*bodyA->internalGetInvMass()*rb0->getLinearFactor(),solverConstraint.m_angularComponentA,solverConstraint.m_appliedImpulse);
-					if (rb1)
-						bodyB->internalApplyImpulse(-solverConstraint.m_contactNormal2*bodyB->internalGetInvMass()*rb1->getLinearFactor(),-solverConstraint.m_angularComponentB,-(btScalar)solverConstraint.m_appliedImpulse);
-				} else
-				{
-					solverConstraint.m_appliedImpulse = 0.f;
-				}
-
-				solverConstraint.m_appliedPushImpulse = 0.f;
-
-				{
-
-					btVector3 externalForceImpulseA = bodyA->m_originalBody ? bodyA->m_externalForceImpulse: btVector3(0,0,0);
-					btVector3 externalTorqueImpulseA = bodyA->m_originalBody ? bodyA->m_externalTorqueImpulse: btVector3(0,0,0);
-					btVector3 externalForceImpulseB = bodyB->m_originalBody ? bodyB->m_externalForceImpulse: btVector3(0,0,0);
-					btVector3 externalTorqueImpulseB = bodyB->m_originalBody ?bodyB->m_externalTorqueImpulse : btVector3(0,0,0);
-
-
-					btScalar vel1Dotn = solverConstraint.m_contactNormal1.dot(bodyA->m_linearVelocity+externalForceImpulseA)
-						+ solverConstraint.m_relpos1CrossNormal.dot(bodyA->m_angularVelocity+externalTorqueImpulseA);
-					btScalar vel2Dotn = solverConstraint.m_contactNormal2.dot(bodyB->m_linearVelocity+externalForceImpulseB)
-						+ solverConstraint.m_relpos2CrossNormal.dot(bodyB->m_angularVelocity+externalTorqueImpulseB);
-					btScalar rel_vel = vel1Dotn+vel2Dotn;
-
-					btScalar positionalError = 0.f;
-					btScalar	velocityError = restitution - rel_vel;// * damping;
-
-
-				
-					if (penetration>0)
-					{
-						positionalError = 0;
-
-						velocityError -= penetration *invTimeStep;
-					} else
-					{
-						positionalError = -penetration * erp*invTimeStep;
-
-					}
-
-					btScalar  penetrationImpulse = positionalError*solverConstraint.m_jacDiagABInv;
-					btScalar velocityImpulse = velocityError *solverConstraint.m_jacDiagABInv;
-
-					if (!infoGlobal.m_splitImpulse || (penetration > infoGlobal.m_splitImpulsePenetrationThreshold))
-					{
-						//combine position and velocity into rhs
-						solverConstraint.m_rhs = penetrationImpulse+velocityImpulse;//-solverConstraint.m_contactNormal1.dot(bodyA->m_externalForce*bodyA->m_invMass-bodyB->m_externalForce/bodyB->m_invMass)*solverConstraint.m_jacDiagABInv;
-						solverConstraint.m_rhsPenetration = 0.f;
-
-					} else
-					{
-						//split position and velocity into rhs and m_rhsPenetration
-						solverConstraint.m_rhs = velocityImpulse;
-						solverConstraint.m_rhsPenetration = penetrationImpulse;
-					}
-					solverConstraint.m_cfm = cfm*solverConstraint.m_jacDiagABInv;
-					solverConstraint.m_lowerLimit = 0;
-					solverConstraint.m_upperLimit = 1e10f;
-				}
+		btScalar positionalError = 0.f;
+		btScalar velocityError = restitution - rel_vel;  // * damping;
 
+		if (penetration > 0)
+		{
+			positionalError = 0;
 
+			velocityError -= penetration * invTimeStep;
+		}
+		else
+		{
+			positionalError = -penetration * erp * invTimeStep;
+		}
 
+		btScalar penetrationImpulse = positionalError * solverConstraint.m_jacDiagABInv;
+		btScalar velocityImpulse = velocityError * solverConstraint.m_jacDiagABInv;
 
+		if (!infoGlobal.m_splitImpulse || (penetration > infoGlobal.m_splitImpulsePenetrationThreshold))
+		{
+			//combine position and velocity into rhs
+			solverConstraint.m_rhs = penetrationImpulse + velocityImpulse;  //-solverConstraint.m_contactNormal1.dot(bodyA->m_externalForce*bodyA->m_invMass-bodyB->m_externalForce/bodyB->m_invMass)*solverConstraint.m_jacDiagABInv;
+			solverConstraint.m_rhsPenetration = 0.f;
+		}
+		else
+		{
+			//split position and velocity into rhs and m_rhsPenetration
+			solverConstraint.m_rhs = velocityImpulse;
+			solverConstraint.m_rhsPenetration = penetrationImpulse;
+		}
+		solverConstraint.m_cfm = cfm * solverConstraint.m_jacDiagABInv;
+		solverConstraint.m_lowerLimit = 0;
+		solverConstraint.m_upperLimit = 1e10f;
+	}
 }
 
-
-
-void btSequentialImpulseConstraintSolver::setFrictionConstraintImpulse( btSolverConstraint& solverConstraint,
-																		int solverBodyIdA, int solverBodyIdB,
-																 btManifoldPoint& cp, const btContactSolverInfo& infoGlobal)
+void btSequentialImpulseConstraintSolver::setFrictionConstraintImpulse(btSolverConstraint& solverConstraint,
+																	   int solverBodyIdA, int solverBodyIdB,
+																	   btManifoldPoint& cp, const btContactSolverInfo& infoGlobal)
 {
-
 	btSolverBody* bodyA = &m_tmpSolverBodyPool[solverBodyIdA];
 	btSolverBody* bodyB = &m_tmpSolverBodyPool[solverBodyIdB];
 
@@ -1058,10 +990,11 @@ void btSequentialImpulseConstraintSolver::setFrictionConstraintImpulse( btSolver
 		{
 			frictionConstraint1.m_appliedImpulse = cp.m_appliedImpulseLateral1 * infoGlobal.m_warmstartingFactor;
 			if (rb0)
-				bodyA->internalApplyImpulse(frictionConstraint1.m_contactNormal1*rb0->getInvMass()*rb0->getLinearFactor(),frictionConstraint1.m_angularComponentA,frictionConstraint1.m_appliedImpulse);
+				bodyA->internalApplyImpulse(frictionConstraint1.m_contactNormal1 * rb0->getInvMass() , frictionConstraint1.m_angularComponentA, frictionConstraint1.m_appliedImpulse);
 			if (rb1)
-				bodyB->internalApplyImpulse(-frictionConstraint1.m_contactNormal2*rb1->getInvMass()*rb1->getLinearFactor(),-frictionConstraint1.m_angularComponentB,-(btScalar)frictionConstraint1.m_appliedImpulse);
-		} else
+				bodyB->internalApplyImpulse(-frictionConstraint1.m_contactNormal2 * rb1->getInvMass() , -frictionConstraint1.m_angularComponentB, -(btScalar)frictionConstraint1.m_appliedImpulse);
+		}
+		else
 		{
 			frictionConstraint1.m_appliedImpulse = 0.f;
 		}
@@ -1069,50 +1002,45 @@ void btSequentialImpulseConstraintSolver::setFrictionConstraintImpulse( btSolver
 
 	if ((infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS))
 	{
-		btSolverConstraint& frictionConstraint2 = m_tmpSolverContactFrictionConstraintPool[solverConstraint.m_frictionIndex+1];
+		btSolverConstraint& frictionConstraint2 = m_tmpSolverContactFrictionConstraintPool[solverConstraint.m_frictionIndex + 1];
 		if (infoGlobal.m_solverMode & SOLVER_USE_WARMSTARTING)
 		{
-			frictionConstraint2.m_appliedImpulse = cp.m_appliedImpulseLateral2  * infoGlobal.m_warmstartingFactor;
+			frictionConstraint2.m_appliedImpulse = cp.m_appliedImpulseLateral2 * infoGlobal.m_warmstartingFactor;
 			if (rb0)
-				bodyA->internalApplyImpulse(frictionConstraint2.m_contactNormal1*rb0->getInvMass(),frictionConstraint2.m_angularComponentA,frictionConstraint2.m_appliedImpulse);
+				bodyA->internalApplyImpulse(frictionConstraint2.m_contactNormal1 * rb0->getInvMass(), frictionConstraint2.m_angularComponentA, frictionConstraint2.m_appliedImpulse);
 			if (rb1)
-				bodyB->internalApplyImpulse(-frictionConstraint2.m_contactNormal2*rb1->getInvMass(),-frictionConstraint2.m_angularComponentB,-(btScalar)frictionConstraint2.m_appliedImpulse);
-		} else
+				bodyB->internalApplyImpulse(-frictionConstraint2.m_contactNormal2 * rb1->getInvMass(), -frictionConstraint2.m_angularComponentB, -(btScalar)frictionConstraint2.m_appliedImpulse);
+		}
+		else
 		{
 			frictionConstraint2.m_appliedImpulse = 0.f;
 		}
 	}
 }
 
-
-
-
-void	btSequentialImpulseConstraintSolver::convertContact(btPersistentManifold* manifold,const btContactSolverInfo& infoGlobal)
+void btSequentialImpulseConstraintSolver::convertContact(btPersistentManifold* manifold, const btContactSolverInfo& infoGlobal)
 {
-	btCollisionObject* colObj0=0,*colObj1=0;
+	btCollisionObject *colObj0 = 0, *colObj1 = 0;
 
 	colObj0 = (btCollisionObject*)manifold->getBody0();
 	colObj1 = (btCollisionObject*)manifold->getBody1();
 
-	int solverBodyIdA = getOrInitSolverBody(*colObj0,infoGlobal.m_timeStep);
-	int solverBodyIdB = getOrInitSolverBody(*colObj1,infoGlobal.m_timeStep);
+	int solverBodyIdA = getOrInitSolverBody(*colObj0, infoGlobal.m_timeStep);
+	int solverBodyIdB = getOrInitSolverBody(*colObj1, infoGlobal.m_timeStep);
 
-//	btRigidBody* bodyA = btRigidBody::upcast(colObj0);
-//	btRigidBody* bodyB = btRigidBody::upcast(colObj1);
+	//	btRigidBody* bodyA = btRigidBody::upcast(colObj0);
+	//	btRigidBody* bodyB = btRigidBody::upcast(colObj1);
 
 	btSolverBody* solverBodyA = &m_tmpSolverBodyPool[solverBodyIdA];
 	btSolverBody* solverBodyB = &m_tmpSolverBodyPool[solverBodyIdB];
 
-
-
 	///avoid collision response between two static objects
 	if (!solverBodyA || (solverBodyA->m_invMass.fuzzyZero() && (!solverBodyB || solverBodyB->m_invMass.fuzzyZero())))
 		return;
 
-	int rollingFriction=1;
-	for (int j=0;j<manifold->getNumContacts();j++)
+	int rollingFriction = 1;
+	for (int j = 0; j < manifold->getNumContacts(); j++)
 	{
-
 		btManifoldPoint& cp = manifold->getContactPoint(j);
 
 		if (cp.getDistance() <= manifold->getContactProcessingThreshold())
@@ -1121,7 +1049,6 @@ void	btSequentialImpulseConstraintSolver::convertContact(btPersistentManifold* m
 			btVector3 rel_pos2;
 			btScalar relaxation;
 
-
 			int frictionIndex = m_tmpSolverContactConstraintPool.size();
 			btSolverConstraint& solverConstraint = m_tmpSolverContactConstraintPool.expandNonInitializing();
 			solverConstraint.m_solverBodyIdA = solverBodyIdA;
@@ -1137,43 +1064,38 @@ void	btSequentialImpulseConstraintSolver::convertContact(btPersistentManifold* m
 
 			btVector3 vel1;
 			btVector3 vel2;
-			
-			solverBodyA->getVelocityInLocalPointNoDelta(rel_pos1,vel1);
-			solverBodyB->getVelocityInLocalPointNoDelta(rel_pos2,vel2 );
 
-			btVector3 vel  = vel1 - vel2;
+			solverBodyA->getVelocityInLocalPointNoDelta(rel_pos1, vel1);
+			solverBodyB->getVelocityInLocalPointNoDelta(rel_pos2, vel2);
+
+			btVector3 vel = vel1 - vel2;
 			btScalar rel_vel = cp.m_normalWorldOnB.dot(vel);
 
 			setupContactConstraint(solverConstraint, solverBodyIdA, solverBodyIdB, cp, infoGlobal, relaxation, rel_pos1, rel_pos2);
 
-
-
-
 			/////setup the friction constraints
 
 			solverConstraint.m_frictionIndex = m_tmpSolverContactFrictionConstraintPool.size();
 
-			if ((cp.m_combinedRollingFriction>0.f) && (rollingFriction>0))
+			if ((cp.m_combinedRollingFriction > 0.f) && (rollingFriction > 0))
 			{
-               
 				{
-					addTorsionalFrictionConstraint(cp.m_normalWorldOnB,solverBodyIdA,solverBodyIdB,frictionIndex,cp,cp.m_combinedSpinningFriction, rel_pos1,rel_pos2,colObj0,colObj1, relaxation);
-					btVector3 axis0,axis1;
-					btPlaneSpace1(cp.m_normalWorldOnB,axis0,axis1);
+					addTorsionalFrictionConstraint(cp.m_normalWorldOnB, solverBodyIdA, solverBodyIdB, frictionIndex, cp, cp.m_combinedSpinningFriction, rel_pos1, rel_pos2, colObj0, colObj1, relaxation);
+					btVector3 axis0, axis1;
+					btPlaneSpace1(cp.m_normalWorldOnB, axis0, axis1);
 					axis0.normalize();
 					axis1.normalize();
-					
-					applyAnisotropicFriction(colObj0,axis0,btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION);
-					applyAnisotropicFriction(colObj1,axis0,btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION);
-					applyAnisotropicFriction(colObj0,axis1,btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION);
-					applyAnisotropicFriction(colObj1,axis1,btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION);
-					if (axis0.length()>0.001)
-						addTorsionalFrictionConstraint(axis0,solverBodyIdA,solverBodyIdB,frictionIndex,cp,
-                                                       cp.m_combinedRollingFriction, rel_pos1,rel_pos2,colObj0,colObj1, relaxation);
-					if (axis1.length()>0.001)
-						addTorsionalFrictionConstraint(axis1,solverBodyIdA,solverBodyIdB,frictionIndex,cp,
-                                                       cp.m_combinedRollingFriction, rel_pos1,rel_pos2,colObj0,colObj1, relaxation);
 
+					applyAnisotropicFriction(colObj0, axis0, btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION);
+					applyAnisotropicFriction(colObj1, axis0, btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION);
+					applyAnisotropicFriction(colObj0, axis1, btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION);
+					applyAnisotropicFriction(colObj1, axis1, btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION);
+					if (axis0.length() > 0.001)
+						addTorsionalFrictionConstraint(axis0, solverBodyIdA, solverBodyIdB, frictionIndex, cp,
+													   cp.m_combinedRollingFriction, rel_pos1, rel_pos2, colObj0, colObj1, relaxation);
+					if (axis1.length() > 0.001)
+						addTorsionalFrictionConstraint(axis1, solverBodyIdA, solverBodyIdB, frictionIndex, cp,
+													   cp.m_combinedRollingFriction, rel_pos1, rel_pos2, colObj0, colObj1, relaxation);
 				}
 			}
 
@@ -1188,106 +1110,97 @@ void	btSequentialImpulseConstraintSolver::convertContact(btPersistentManifold* m
 			///If you choose SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION, then the friction will be independent from the relative projected velocity.
 			///
 			///The user can manually override the friction directions for certain contacts using a contact callback,
-			///and set the cp.m_lateralFrictionInitialized to true
+			///and use contactPoint.m_contactPointFlags |= BT_CONTACT_FLAG_LATERAL_FRICTION_INITIALIZED
 			///In that case, you can set the target relative motion in each friction direction (cp.m_contactMotion1 and cp.m_contactMotion2)
 			///this will give a conveyor belt effect
 			///
-		
-			if (!(infoGlobal.m_solverMode & SOLVER_ENABLE_FRICTION_DIRECTION_CACHING) || !(cp.m_contactPointFlags&BT_CONTACT_FLAG_LATERAL_FRICTION_INITIALIZED))
+
+			if (!(infoGlobal.m_solverMode & SOLVER_ENABLE_FRICTION_DIRECTION_CACHING) || !(cp.m_contactPointFlags & BT_CONTACT_FLAG_LATERAL_FRICTION_INITIALIZED))
 			{
 				cp.m_lateralFrictionDir1 = vel - cp.m_normalWorldOnB * rel_vel;
 				btScalar lat_rel_vel = cp.m_lateralFrictionDir1.length2();
 				if (!(infoGlobal.m_solverMode & SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION) && lat_rel_vel > SIMD_EPSILON)
 				{
-					cp.m_lateralFrictionDir1 *= 1.f/btSqrt(lat_rel_vel);
-					applyAnisotropicFriction(colObj0,cp.m_lateralFrictionDir1,btCollisionObject::CF_ANISOTROPIC_FRICTION);
-					applyAnisotropicFriction(colObj1,cp.m_lateralFrictionDir1,btCollisionObject::CF_ANISOTROPIC_FRICTION);
-					addFrictionConstraint(cp.m_lateralFrictionDir1,solverBodyIdA,solverBodyIdB,frictionIndex,cp,rel_pos1,rel_pos2,colObj0,colObj1, relaxation,infoGlobal);
-
-					if((infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS))
-					{
-						cp.m_lateralFrictionDir2 = cp.m_lateralFrictionDir1.cross(cp.m_normalWorldOnB);
-						cp.m_lateralFrictionDir2.normalize();//??
-						applyAnisotropicFriction(colObj0,cp.m_lateralFrictionDir2,btCollisionObject::CF_ANISOTROPIC_FRICTION);
-						applyAnisotropicFriction(colObj1,cp.m_lateralFrictionDir2,btCollisionObject::CF_ANISOTROPIC_FRICTION);
-						addFrictionConstraint(cp.m_lateralFrictionDir2,solverBodyIdA,solverBodyIdB,frictionIndex,cp,rel_pos1,rel_pos2,colObj0,colObj1, relaxation, infoGlobal);
-					}
-
-				} else
-				{
-					btPlaneSpace1(cp.m_normalWorldOnB,cp.m_lateralFrictionDir1,cp.m_lateralFrictionDir2);
-
-					applyAnisotropicFriction(colObj0,cp.m_lateralFrictionDir1,btCollisionObject::CF_ANISOTROPIC_FRICTION);
-					applyAnisotropicFriction(colObj1,cp.m_lateralFrictionDir1,btCollisionObject::CF_ANISOTROPIC_FRICTION);
-					addFrictionConstraint(cp.m_lateralFrictionDir1,solverBodyIdA,solverBodyIdB,frictionIndex,cp,rel_pos1,rel_pos2,colObj0,colObj1, relaxation, infoGlobal);
+					cp.m_lateralFrictionDir1 *= 1.f / btSqrt(lat_rel_vel);
+					applyAnisotropicFriction(colObj0, cp.m_lateralFrictionDir1, btCollisionObject::CF_ANISOTROPIC_FRICTION);
+					applyAnisotropicFriction(colObj1, cp.m_lateralFrictionDir1, btCollisionObject::CF_ANISOTROPIC_FRICTION);
+					addFrictionConstraint(cp.m_lateralFrictionDir1, solverBodyIdA, solverBodyIdB, frictionIndex, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, infoGlobal);
 
 					if ((infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS))
 					{
-						applyAnisotropicFriction(colObj0,cp.m_lateralFrictionDir2,btCollisionObject::CF_ANISOTROPIC_FRICTION);
-						applyAnisotropicFriction(colObj1,cp.m_lateralFrictionDir2,btCollisionObject::CF_ANISOTROPIC_FRICTION);
-						addFrictionConstraint(cp.m_lateralFrictionDir2,solverBodyIdA,solverBodyIdB,frictionIndex,cp,rel_pos1,rel_pos2,colObj0,colObj1, relaxation, infoGlobal);
+						cp.m_lateralFrictionDir2 = cp.m_lateralFrictionDir1.cross(cp.m_normalWorldOnB);
+						cp.m_lateralFrictionDir2.normalize();  //??
+						applyAnisotropicFriction(colObj0, cp.m_lateralFrictionDir2, btCollisionObject::CF_ANISOTROPIC_FRICTION);
+						applyAnisotropicFriction(colObj1, cp.m_lateralFrictionDir2, btCollisionObject::CF_ANISOTROPIC_FRICTION);
+						addFrictionConstraint(cp.m_lateralFrictionDir2, solverBodyIdA, solverBodyIdB, frictionIndex, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, infoGlobal);
 					}
+				}
+				else
+				{
+					btPlaneSpace1(cp.m_normalWorldOnB, cp.m_lateralFrictionDir1, cp.m_lateralFrictionDir2);
 
+					applyAnisotropicFriction(colObj0, cp.m_lateralFrictionDir1, btCollisionObject::CF_ANISOTROPIC_FRICTION);
+					applyAnisotropicFriction(colObj1, cp.m_lateralFrictionDir1, btCollisionObject::CF_ANISOTROPIC_FRICTION);
+					addFrictionConstraint(cp.m_lateralFrictionDir1, solverBodyIdA, solverBodyIdB, frictionIndex, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, infoGlobal);
+
+					if ((infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS))
+					{
+						applyAnisotropicFriction(colObj0, cp.m_lateralFrictionDir2, btCollisionObject::CF_ANISOTROPIC_FRICTION);
+						applyAnisotropicFriction(colObj1, cp.m_lateralFrictionDir2, btCollisionObject::CF_ANISOTROPIC_FRICTION);
+						addFrictionConstraint(cp.m_lateralFrictionDir2, solverBodyIdA, solverBodyIdB, frictionIndex, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, infoGlobal);
+					}
 
 					if ((infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS) && (infoGlobal.m_solverMode & SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION))
 					{
-						cp.m_contactPointFlags|=BT_CONTACT_FLAG_LATERAL_FRICTION_INITIALIZED;
+						cp.m_contactPointFlags |= BT_CONTACT_FLAG_LATERAL_FRICTION_INITIALIZED;
 					}
 				}
-
-			} else
+			}
+			else
 			{
-				addFrictionConstraint(cp.m_lateralFrictionDir1,solverBodyIdA,solverBodyIdB,frictionIndex,cp,rel_pos1,rel_pos2,colObj0,colObj1, relaxation, infoGlobal, cp.m_contactMotion1, cp.m_frictionCFM);
+				addFrictionConstraint(cp.m_lateralFrictionDir1, solverBodyIdA, solverBodyIdB, frictionIndex, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, infoGlobal, cp.m_contactMotion1, cp.m_frictionCFM);
 
 				if ((infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS))
-					addFrictionConstraint(cp.m_lateralFrictionDir2,solverBodyIdA,solverBodyIdB,frictionIndex,cp,rel_pos1,rel_pos2,colObj0,colObj1, relaxation, infoGlobal, cp.m_contactMotion2, cp.m_frictionCFM);
-
+					addFrictionConstraint(cp.m_lateralFrictionDir2, solverBodyIdA, solverBodyIdB, frictionIndex, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, infoGlobal, cp.m_contactMotion2, cp.m_frictionCFM);
 			}
-			setFrictionConstraintImpulse( solverConstraint, solverBodyIdA, solverBodyIdB, cp, infoGlobal);
-
-
-
-
+			setFrictionConstraintImpulse(solverConstraint, solverBodyIdA, solverBodyIdB, cp, infoGlobal);
 		}
 	}
 }
 
-void btSequentialImpulseConstraintSolver::convertContacts(btPersistentManifold** manifoldPtr,int numManifolds, const btContactSolverInfo& infoGlobal)
+void btSequentialImpulseConstraintSolver::convertContacts(btPersistentManifold** manifoldPtr, int numManifolds, const btContactSolverInfo& infoGlobal)
 {
 	int i;
 	btPersistentManifold* manifold = 0;
-//			btCollisionObject* colObj0=0,*colObj1=0;
+	//			btCollisionObject* colObj0=0,*colObj1=0;
 
-
-	for (i=0;i<numManifolds;i++)
+	for (i = 0; i < numManifolds; i++)
 	{
 		manifold = manifoldPtr[i];
-		convertContact(manifold,infoGlobal);
+		convertContact(manifold, infoGlobal);
 	}
 }
 
-
 void btSequentialImpulseConstraintSolver::convertJoint(btSolverConstraint* currentConstraintRow,
-    btTypedConstraint* constraint,
-    const btTypedConstraint::btConstraintInfo1& info1,
-    int solverBodyIdA,
-    int solverBodyIdB,
-    const btContactSolverInfo& infoGlobal
-    )
+													   btTypedConstraint* constraint,
+													   const btTypedConstraint::btConstraintInfo1& info1,
+													   int solverBodyIdA,
+													   int solverBodyIdB,
+													   const btContactSolverInfo& infoGlobal)
 {
 	const btRigidBody& rbA = constraint->getRigidBodyA();
 	const btRigidBody& rbB = constraint->getRigidBodyB();
 
-    const btSolverBody* bodyAPtr = &m_tmpSolverBodyPool[solverBodyIdA];
-    const btSolverBody* bodyBPtr = &m_tmpSolverBodyPool[solverBodyIdB];
+	const btSolverBody* bodyAPtr = &m_tmpSolverBodyPool[solverBodyIdA];
+	const btSolverBody* bodyBPtr = &m_tmpSolverBodyPool[solverBodyIdB];
 
 	int overrideNumSolverIterations = constraint->getOverrideNumSolverIterations() > 0 ? constraint->getOverrideNumSolverIterations() : infoGlobal.m_numIterations;
-	if (overrideNumSolverIterations>m_maxOverrideNumSolverIterations)
+	if (overrideNumSolverIterations > m_maxOverrideNumSolverIterations)
 		m_maxOverrideNumSolverIterations = overrideNumSolverIterations;
 
-	for (int j=0;j<info1.m_numConstraintRows;j++)
+	for (int j = 0; j < info1.m_numConstraintRows; j++)
 	{
-		memset(&currentConstraintRow[j],0,sizeof(btSolverConstraint));
+		memset(&currentConstraintRow[j], 0, sizeof(btSolverConstraint));
 		currentConstraintRow[j].m_lowerLimit = -SIMD_INFINITY;
 		currentConstraintRow[j].m_upperLimit = SIMD_INFINITY;
 		currentConstraintRow[j].m_appliedImpulse = 0.f;
@@ -1297,15 +1210,15 @@ void btSequentialImpulseConstraintSolver::convertJoint(btSolverConstraint* curre
 		currentConstraintRow[j].m_overrideNumSolverIterations = overrideNumSolverIterations;
 	}
 
-    // these vectors are already cleared in initSolverBody, no need to redundantly clear again
-    btAssert(bodyAPtr->getDeltaLinearVelocity().isZero());
-    btAssert(bodyAPtr->getDeltaAngularVelocity().isZero());
-    btAssert(bodyAPtr->getPushVelocity().isZero());
-    btAssert(bodyAPtr->getTurnVelocity().isZero());
-    btAssert(bodyBPtr->getDeltaLinearVelocity().isZero());
-    btAssert(bodyBPtr->getDeltaAngularVelocity().isZero());
-    btAssert(bodyBPtr->getPushVelocity().isZero());
-    btAssert(bodyBPtr->getTurnVelocity().isZero());
+	// these vectors are already cleared in initSolverBody, no need to redundantly clear again
+	btAssert(bodyAPtr->getDeltaLinearVelocity().isZero());
+	btAssert(bodyAPtr->getDeltaAngularVelocity().isZero());
+	btAssert(bodyAPtr->getPushVelocity().isZero());
+	btAssert(bodyAPtr->getTurnVelocity().isZero());
+	btAssert(bodyBPtr->getDeltaLinearVelocity().isZero());
+	btAssert(bodyBPtr->getDeltaAngularVelocity().isZero());
+	btAssert(bodyBPtr->getPushVelocity().isZero());
+	btAssert(bodyBPtr->getTurnVelocity().isZero());
 	//bodyAPtr->internalGetDeltaLinearVelocity().setValue(0.f,0.f,0.f);
 	//bodyAPtr->internalGetDeltaAngularVelocity().setValue(0.f,0.f,0.f);
 	//bodyAPtr->internalGetPushVelocity().setValue(0.f,0.f,0.f);
@@ -1315,17 +1228,16 @@ void btSequentialImpulseConstraintSolver::convertJoint(btSolverConstraint* curre
 	//bodyBPtr->internalGetPushVelocity().setValue(0.f,0.f,0.f);
 	//bodyBPtr->internalGetTurnVelocity().setValue(0.f,0.f,0.f);
 
-
 	btTypedConstraint::btConstraintInfo2 info2;
-	info2.fps = 1.f/infoGlobal.m_timeStep;
+	info2.fps = 1.f / infoGlobal.m_timeStep;
 	info2.erp = infoGlobal.m_erp;
 	info2.m_J1linearAxis = currentConstraintRow->m_contactNormal1;
 	info2.m_J1angularAxis = currentConstraintRow->m_relpos1CrossNormal;
 	info2.m_J2linearAxis = currentConstraintRow->m_contactNormal2;
 	info2.m_J2angularAxis = currentConstraintRow->m_relpos2CrossNormal;
-	info2.rowskip = sizeof(btSolverConstraint)/sizeof(btScalar);//check this
+	info2.rowskip = sizeof(btSolverConstraint) / sizeof(btScalar);  //check this
 	///the size of btSolverConstraint needs be a multiple of btScalar
-	btAssert(info2.rowskip*sizeof(btScalar)== sizeof(btSolverConstraint));
+	btAssert(info2.rowskip * sizeof(btScalar) == sizeof(btSolverConstraint));
 	info2.m_constraintError = &currentConstraintRow->m_rhs;
 	currentConstraintRow->m_cfm = infoGlobal.m_globalCfm;
 	info2.m_damping = infoGlobal.m_damping;
@@ -1336,16 +1248,16 @@ void btSequentialImpulseConstraintSolver::convertJoint(btSolverConstraint* curre
 	constraint->getInfo2(&info2);
 
 	///finalize the constraint setup
-	for (int j=0;j<info1.m_numConstraintRows;j++)
+	for (int j = 0; j < info1.m_numConstraintRows; j++)
 	{
 		btSolverConstraint& solverConstraint = currentConstraintRow[j];
 
-		if (solverConstraint.m_upperLimit>=constraint->getBreakingImpulseThreshold())
+		if (solverConstraint.m_upperLimit >= constraint->getBreakingImpulseThreshold())
 		{
 			solverConstraint.m_upperLimit = constraint->getBreakingImpulseThreshold();
 		}
 
-		if (solverConstraint.m_lowerLimit<=-constraint->getBreakingImpulseThreshold())
+		if (solverConstraint.m_lowerLimit <= -constraint->getBreakingImpulseThreshold())
 		{
 			solverConstraint.m_lowerLimit = -constraint->getBreakingImpulseThreshold();
 		}
@@ -1354,18 +1266,18 @@ void btSequentialImpulseConstraintSolver::convertJoint(btSolverConstraint* curre
 
 		{
 			const btVector3& ftorqueAxis1 = solverConstraint.m_relpos1CrossNormal;
-			solverConstraint.m_angularComponentA = constraint->getRigidBodyA().getInvInertiaTensorWorld()*ftorqueAxis1*constraint->getRigidBodyA().getAngularFactor();
+			solverConstraint.m_angularComponentA = constraint->getRigidBodyA().getInvInertiaTensorWorld() * ftorqueAxis1 * constraint->getRigidBodyA().getAngularFactor();
 		}
 		{
 			const btVector3& ftorqueAxis2 = solverConstraint.m_relpos2CrossNormal;
-			solverConstraint.m_angularComponentB = constraint->getRigidBodyB().getInvInertiaTensorWorld()*ftorqueAxis2*constraint->getRigidBodyB().getAngularFactor();
+			solverConstraint.m_angularComponentB = constraint->getRigidBodyB().getInvInertiaTensorWorld() * ftorqueAxis2 * constraint->getRigidBodyB().getAngularFactor();
 		}
 
 		{
-			btVector3 iMJlA = solverConstraint.m_contactNormal1*rbA.getInvMass();
-			btVector3 iMJaA = rbA.getInvInertiaTensorWorld()*solverConstraint.m_relpos1CrossNormal;
-			btVector3 iMJlB = solverConstraint.m_contactNormal2*rbB.getInvMass();//sign of normal?
-			btVector3 iMJaB = rbB.getInvInertiaTensorWorld()*solverConstraint.m_relpos2CrossNormal;
+			btVector3 iMJlA = solverConstraint.m_contactNormal1 * rbA.getInvMass();
+			btVector3 iMJaA = rbA.getInvInertiaTensorWorld() * solverConstraint.m_relpos1CrossNormal;
+			btVector3 iMJlB = solverConstraint.m_contactNormal2 * rbB.getInvMass();  //sign of normal?
+			btVector3 iMJaB = rbB.getInvInertiaTensorWorld() * solverConstraint.m_relpos2CrossNormal;
 
 			btScalar sum = iMJlA.dot(solverConstraint.m_contactNormal1);
 			sum += iMJaA.dot(solverConstraint.m_relpos1CrossNormal);
@@ -1373,41 +1285,38 @@ void btSequentialImpulseConstraintSolver::convertJoint(btSolverConstraint* curre
 			sum += iMJaB.dot(solverConstraint.m_relpos2CrossNormal);
 			btScalar fsum = btFabs(sum);
 			btAssert(fsum > SIMD_EPSILON);
-			btScalar sorRelaxation = 1.f;//todo: get from globalInfo?
-			solverConstraint.m_jacDiagABInv = fsum>SIMD_EPSILON?sorRelaxation/sum : 0.f;
+			btScalar sorRelaxation = 1.f;  //todo: get from globalInfo?
+			solverConstraint.m_jacDiagABInv = fsum > SIMD_EPSILON ? sorRelaxation / sum : 0.f;
 		}
 
 		{
 			btScalar rel_vel;
-			btVector3 externalForceImpulseA = bodyAPtr->m_originalBody ? bodyAPtr->m_externalForceImpulse : btVector3(0,0,0);
-			btVector3 externalTorqueImpulseA = bodyAPtr->m_originalBody ? bodyAPtr->m_externalTorqueImpulse : btVector3(0,0,0);
+			btVector3 externalForceImpulseA = bodyAPtr->m_originalBody ? bodyAPtr->m_externalForceImpulse : btVector3(0, 0, 0);
+			btVector3 externalTorqueImpulseA = bodyAPtr->m_originalBody ? bodyAPtr->m_externalTorqueImpulse : btVector3(0, 0, 0);
 
-			btVector3 externalForceImpulseB = bodyBPtr->m_originalBody ? bodyBPtr->m_externalForceImpulse : btVector3(0,0,0);
-			btVector3 externalTorqueImpulseB = bodyBPtr->m_originalBody ?bodyBPtr->m_externalTorqueImpulse : btVector3(0,0,0);
+			btVector3 externalForceImpulseB = bodyBPtr->m_originalBody ? bodyBPtr->m_externalForceImpulse : btVector3(0, 0, 0);
+			btVector3 externalTorqueImpulseB = bodyBPtr->m_originalBody ? bodyBPtr->m_externalTorqueImpulse : btVector3(0, 0, 0);
 
-			btScalar vel1Dotn = solverConstraint.m_contactNormal1.dot(rbA.getLinearVelocity()+externalForceImpulseA)
-								+ solverConstraint.m_relpos1CrossNormal.dot(rbA.getAngularVelocity()+externalTorqueImpulseA);
+			btScalar vel1Dotn = solverConstraint.m_contactNormal1.dot(rbA.getLinearVelocity() + externalForceImpulseA) + solverConstraint.m_relpos1CrossNormal.dot(rbA.getAngularVelocity() + externalTorqueImpulseA);
 
-			btScalar vel2Dotn = solverConstraint.m_contactNormal2.dot(rbB.getLinearVelocity()+externalForceImpulseB)
-												+ solverConstraint.m_relpos2CrossNormal.dot(rbB.getAngularVelocity()+externalTorqueImpulseB);
+			btScalar vel2Dotn = solverConstraint.m_contactNormal2.dot(rbB.getLinearVelocity() + externalForceImpulseB) + solverConstraint.m_relpos2CrossNormal.dot(rbB.getAngularVelocity() + externalTorqueImpulseB);
 
-			rel_vel = vel1Dotn+vel2Dotn;
+			rel_vel = vel1Dotn + vel2Dotn;
 			btScalar restitution = 0.f;
-			btScalar positionalError = solverConstraint.m_rhs;//already filled in by getConstraintInfo2
-			btScalar	velocityError = restitution - rel_vel * info2.m_damping;
-			btScalar	penetrationImpulse = positionalError*solverConstraint.m_jacDiagABInv;
-			btScalar	velocityImpulse = velocityError *solverConstraint.m_jacDiagABInv;
-			solverConstraint.m_rhs = penetrationImpulse+velocityImpulse;
+			btScalar positionalError = solverConstraint.m_rhs;  //already filled in by getConstraintInfo2
+			btScalar velocityError = restitution - rel_vel * info2.m_damping;
+			btScalar penetrationImpulse = positionalError * solverConstraint.m_jacDiagABInv;
+			btScalar velocityImpulse = velocityError * solverConstraint.m_jacDiagABInv;
+			solverConstraint.m_rhs = penetrationImpulse + velocityImpulse;
 			solverConstraint.m_appliedImpulse = 0.f;
 		}
 	}
 }
 
-
-void btSequentialImpulseConstraintSolver::convertJoints(btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal)
+void btSequentialImpulseConstraintSolver::convertJoints(btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal)
 {
-    BT_PROFILE("convertJoints");
-	for (int j=0;j<numConstraints;j++)
+	BT_PROFILE("convertJoints");
+	for (int j = 0; j < numConstraints; j++)
 	{
 		btTypedConstraint* constraint = constraints[j];
 		constraint->buildJacobian();
@@ -1418,7 +1327,7 @@ void btSequentialImpulseConstraintSolver::convertJoints(btTypedConstraint** cons
 
 	m_tmpConstraintSizesPool.resizeNoInitialize(numConstraints);
 	//calculate the total number of contraint rows
-	for (int i=0;i<numConstraints;i++)
+	for (int i = 0; i < numConstraints; i++)
 	{
 		btTypedConstraint::btConstraintInfo1& info1 = m_tmpConstraintSizesPool[i];
 		btJointFeedback* fb = constraints[i]->getJointFeedback();
@@ -1433,7 +1342,8 @@ void btSequentialImpulseConstraintSolver::convertJoints(btTypedConstraint** cons
 		if (constraints[i]->isEnabled())
 		{
 			constraints[i]->getInfo1(&info1);
-		} else
+		}
+		else
 		{
 			info1.m_numConstraintRows = 0;
 			info1.nub = 0;
@@ -1442,110 +1352,105 @@ void btSequentialImpulseConstraintSolver::convertJoints(btTypedConstraint** cons
 	}
 	m_tmpSolverNonContactConstraintPool.resizeNoInitialize(totalNumRows);
 
-
 	///setup the btSolverConstraints
 	int currentRow = 0;
 
-	for (int i=0;i<numConstraints;i++)
+	for (int i = 0; i < numConstraints; i++)
 	{
 		const btTypedConstraint::btConstraintInfo1& info1 = m_tmpConstraintSizesPool[i];
 
 		if (info1.m_numConstraintRows)
 		{
-			btAssert(currentRow<totalNumRows);
+			btAssert(currentRow < totalNumRows);
 
 			btSolverConstraint* currentConstraintRow = &m_tmpSolverNonContactConstraintPool[currentRow];
 			btTypedConstraint* constraint = constraints[i];
 			btRigidBody& rbA = constraint->getRigidBodyA();
 			btRigidBody& rbB = constraint->getRigidBodyB();
 
-			int solverBodyIdA = getOrInitSolverBody(rbA,infoGlobal.m_timeStep);
-            int solverBodyIdB = getOrInitSolverBody(rbB,infoGlobal.m_timeStep);
+			int solverBodyIdA = getOrInitSolverBody(rbA, infoGlobal.m_timeStep);
+			int solverBodyIdB = getOrInitSolverBody(rbB, infoGlobal.m_timeStep);
 
-            convertJoint(currentConstraintRow, constraint, info1, solverBodyIdA, solverBodyIdB, infoGlobal);
-        }
-		currentRow+=info1.m_numConstraintRows;
+			convertJoint(currentConstraintRow, constraint, info1, solverBodyIdA, solverBodyIdB, infoGlobal);
+		}
+		currentRow += info1.m_numConstraintRows;
 	}
 }
 
-
 void btSequentialImpulseConstraintSolver::convertBodies(btCollisionObject** bodies, int numBodies, const btContactSolverInfo& infoGlobal)
 {
-    BT_PROFILE("convertBodies");
+	BT_PROFILE("convertBodies");
 	for (int i = 0; i < numBodies; i++)
 	{
 		bodies[i]->setCompanionId(-1);
 	}
 #if BT_THREADSAFE
-    m_kinematicBodyUniqueIdToSolverBodyTable.resize( 0 );
-#endif // BT_THREADSAFE
+	m_kinematicBodyUniqueIdToSolverBodyTable.resize(0);
+#endif  // BT_THREADSAFE
 
-	m_tmpSolverBodyPool.reserve(numBodies+1);
+	m_tmpSolverBodyPool.reserve(numBodies + 1);
 	m_tmpSolverBodyPool.resize(0);
 
 	//btSolverBody& fixedBody = m_tmpSolverBodyPool.expand();
-    //initSolverBody(&fixedBody,0);
+	//initSolverBody(&fixedBody,0);
 
-    for (int i=0;i<numBodies;i++)
+	for (int i = 0; i < numBodies; i++)
 	{
-		int bodyId = getOrInitSolverBody(*bodies[i],infoGlobal.m_timeStep);
+		int bodyId = getOrInitSolverBody(*bodies[i], infoGlobal.m_timeStep);
 
 		btRigidBody* body = btRigidBody::upcast(bodies[i]);
 		if (body && body->getInvMass())
 		{
 			btSolverBody& solverBody = m_tmpSolverBodyPool[bodyId];
-			btVector3 gyroForce (0,0,0);
-			if (body->getFlags()&BT_ENABLE_GYROSCOPIC_FORCE_EXPLICIT)
+			btVector3 gyroForce(0, 0, 0);
+			if (body->getFlags() & BT_ENABLE_GYROSCOPIC_FORCE_EXPLICIT)
 			{
 				gyroForce = body->computeGyroscopicForceExplicit(infoGlobal.m_maxGyroscopicForce);
-				solverBody.m_externalTorqueImpulse -= gyroForce*body->getInvInertiaTensorWorld()*infoGlobal.m_timeStep;
+				solverBody.m_externalTorqueImpulse -= gyroForce * body->getInvInertiaTensorWorld() * infoGlobal.m_timeStep;
 			}
-			if (body->getFlags()&BT_ENABLE_GYROSCOPIC_FORCE_IMPLICIT_WORLD)
+			if (body->getFlags() & BT_ENABLE_GYROSCOPIC_FORCE_IMPLICIT_WORLD)
 			{
 				gyroForce = body->computeGyroscopicImpulseImplicit_World(infoGlobal.m_timeStep);
 				solverBody.m_externalTorqueImpulse += gyroForce;
 			}
-			if (body->getFlags()&BT_ENABLE_GYROSCOPIC_FORCE_IMPLICIT_BODY)
+			if (body->getFlags() & BT_ENABLE_GYROSCOPIC_FORCE_IMPLICIT_BODY)
 			{
 				gyroForce = body->computeGyroscopicImpulseImplicit_Body(infoGlobal.m_timeStep);
 				solverBody.m_externalTorqueImpulse += gyroForce;
-
 			}
 		}
 	}
 }
 
-
-btScalar btSequentialImpulseConstraintSolver::solveGroupCacheFriendlySetup(btCollisionObject** bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer)
+btScalar btSequentialImpulseConstraintSolver::solveGroupCacheFriendlySetup(btCollisionObject** bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer)
 {
 	m_fixedBodyId = -1;
 	BT_PROFILE("solveGroupCacheFriendlySetup");
 	(void)debugDrawer;
 
-    // if solver mode has changed,
-    if ( infoGlobal.m_solverMode != m_cachedSolverMode )
-    {
-        // update solver functions to use SIMD or non-SIMD
-        bool useSimd = !!( infoGlobal.m_solverMode & SOLVER_SIMD );
-        setupSolverFunctions( useSimd );
-        m_cachedSolverMode = infoGlobal.m_solverMode;
-    }
+	// if solver mode has changed,
+	if (infoGlobal.m_solverMode != m_cachedSolverMode)
+	{
+		// update solver functions to use SIMD or non-SIMD
+		bool useSimd = !!(infoGlobal.m_solverMode & SOLVER_SIMD);
+		setupSolverFunctions(useSimd);
+		m_cachedSolverMode = infoGlobal.m_solverMode;
+	}
 	m_maxOverrideNumSolverIterations = 0;
 
 #ifdef BT_ADDITIONAL_DEBUG
-	 //make sure that dynamic bodies exist for all (enabled) constraints
-	for (int i=0;i<numConstraints;i++)
+	//make sure that dynamic bodies exist for all (enabled) constraints
+	for (int i = 0; i < numConstraints; i++)
 	{
 		btTypedConstraint* constraint = constraints[i];
 		if (constraint->isEnabled())
 		{
 			if (!constraint->getRigidBodyA().isStaticOrKinematicObject())
 			{
-				bool found=false;
-				for (int b=0;b<numBodies;b++)
+				bool found = false;
+				for (int b = 0; b < numBodies; b++)
 				{
-
-					if (&constraint->getRigidBodyA()==bodies[b])
+					if (&constraint->getRigidBodyA() == bodies[b])
 					{
 						found = true;
 						break;
@@ -1555,10 +1460,10 @@ btScalar btSequentialImpulseConstraintSolver::solveGroupCacheFriendlySetup(btCol
 			}
 			if (!constraint->getRigidBodyB().isStaticOrKinematicObject())
 			{
-				bool found=false;
-				for (int b=0;b<numBodies;b++)
+				bool found = false;
+				for (int b = 0; b < numBodies; b++)
 				{
-					if (&constraint->getRigidBodyB()==bodies[b])
+					if (&constraint->getRigidBodyB() == bodies[b])
 					{
 						found = true;
 						break;
@@ -1568,50 +1473,46 @@ btScalar btSequentialImpulseConstraintSolver::solveGroupCacheFriendlySetup(btCol
 			}
 		}
 	}
-    //make sure that dynamic bodies exist for all contact manifolds
-    for (int i=0;i<numManifolds;i++)
-    {
-        if (!manifoldPtr[i]->getBody0()->isStaticOrKinematicObject())
-        {
-            bool found=false;
-            for (int b=0;b<numBodies;b++)
-            {
-
-                if (manifoldPtr[i]->getBody0()==bodies[b])
-                {
-                    found = true;
-                    break;
-                }
-            }
-            btAssert(found);
-        }
-        if (!manifoldPtr[i]->getBody1()->isStaticOrKinematicObject())
-        {
-            bool found=false;
-            for (int b=0;b<numBodies;b++)
-            {
-                if (manifoldPtr[i]->getBody1()==bodies[b])
-                {
-                    found = true;
-                    break;
-                }
-            }
-            btAssert(found);
-        }
-    }
-#endif //BT_ADDITIONAL_DEBUG
-
+	//make sure that dynamic bodies exist for all contact manifolds
+	for (int i = 0; i < numManifolds; i++)
+	{
+		if (!manifoldPtr[i]->getBody0()->isStaticOrKinematicObject())
+		{
+			bool found = false;
+			for (int b = 0; b < numBodies; b++)
+			{
+				if (manifoldPtr[i]->getBody0() == bodies[b])
+				{
+					found = true;
+					break;
+				}
+			}
+			btAssert(found);
+		}
+		if (!manifoldPtr[i]->getBody1()->isStaticOrKinematicObject())
+		{
+			bool found = false;
+			for (int b = 0; b < numBodies; b++)
+			{
+				if (manifoldPtr[i]->getBody1() == bodies[b])
+				{
+					found = true;
+					break;
+				}
+			}
+			btAssert(found);
+		}
+	}
+#endif  //BT_ADDITIONAL_DEBUG
 
 	//convert all bodies
-    convertBodies(bodies, numBodies, infoGlobal);
+	convertBodies(bodies, numBodies, infoGlobal);
 
-    convertJoints(constraints, numConstraints, infoGlobal);
+	convertJoints(constraints, numConstraints, infoGlobal);
 
-	convertContacts(manifoldPtr,numManifolds,infoGlobal);
-
-
-//	btContactSolverInfo info = infoGlobal;
+	convertContacts(manifoldPtr, numManifolds, infoGlobal);
 
+	//	btContactSolverInfo info = infoGlobal;
 
 	int numNonContactPool = m_tmpSolverNonContactConstraintPool.size();
 	int numConstraintPool = m_tmpSolverContactConstraintPool.size();
@@ -1620,35 +1521,33 @@ btScalar btSequentialImpulseConstraintSolver::solveGroupCacheFriendlySetup(btCol
 	///@todo: use stack allocator for such temporarily memory, same for solver bodies/constraints
 	m_orderNonContactConstraintPool.resizeNoInitialize(numNonContactPool);
 	if ((infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS))
-		m_orderTmpConstraintPool.resizeNoInitialize(numConstraintPool*2);
+		m_orderTmpConstraintPool.resizeNoInitialize(numConstraintPool * 2);
 	else
 		m_orderTmpConstraintPool.resizeNoInitialize(numConstraintPool);
 
 	m_orderFrictionConstraintPool.resizeNoInitialize(numFrictionPool);
 	{
 		int i;
-		for (i=0;i<numNonContactPool;i++)
+		for (i = 0; i < numNonContactPool; i++)
 		{
 			m_orderNonContactConstraintPool[i] = i;
 		}
-		for (i=0;i<numConstraintPool;i++)
+		for (i = 0; i < numConstraintPool; i++)
 		{
 			m_orderTmpConstraintPool[i] = i;
 		}
-		for (i=0;i<numFrictionPool;i++)
+		for (i = 0; i < numFrictionPool; i++)
 		{
 			m_orderFrictionConstraintPool[i] = i;
 		}
 	}
 
 	return 0.f;
-
 }
 
-
-btScalar btSequentialImpulseConstraintSolver::solveSingleIteration(int iteration, btCollisionObject** /*bodies */,int /*numBodies*/,btPersistentManifold** /*manifoldPtr*/, int /*numManifolds*/,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* /*debugDrawer*/)
+btScalar btSequentialImpulseConstraintSolver::solveSingleIteration(int iteration, btCollisionObject** /*bodies */, int /*numBodies*/, btPersistentManifold** /*manifoldPtr*/, int /*numManifolds*/, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* /*debugDrawer*/)
 {
-    BT_PROFILE("solveSingleIteration");
+	BT_PROFILE("solveSingleIteration");
 	btScalar leastSquaresResidual = 0.f;
 
 	int numNonContactPool = m_tmpSolverNonContactConstraintPool.size();
@@ -1657,29 +1556,31 @@ btScalar btSequentialImpulseConstraintSolver::solveSingleIteration(int iteration
 
 	if (infoGlobal.m_solverMode & SOLVER_RANDMIZE_ORDER)
 	{
-		if (1)			// uncomment this for a bit less random ((iteration & 7) == 0)
+		if (1)  // uncomment this for a bit less random ((iteration & 7) == 0)
 		{
-
-			for (int j=0; j<numNonContactPool; ++j) {
+			for (int j = 0; j < numNonContactPool; ++j)
+			{
 				int tmp = m_orderNonContactConstraintPool[j];
-				int swapi = btRandInt2(j+1);
+				int swapi = btRandInt2(j + 1);
 				m_orderNonContactConstraintPool[j] = m_orderNonContactConstraintPool[swapi];
 				m_orderNonContactConstraintPool[swapi] = tmp;
 			}
 
 			//contact/friction constraints are not solved more than
-			if (iteration< infoGlobal.m_numIterations)
+			if (iteration < infoGlobal.m_numIterations)
 			{
-				for (int j=0; j<numConstraintPool; ++j) {
+				for (int j = 0; j < numConstraintPool; ++j)
+				{
 					int tmp = m_orderTmpConstraintPool[j];
-					int swapi = btRandInt2(j+1);
+					int swapi = btRandInt2(j + 1);
 					m_orderTmpConstraintPool[j] = m_orderTmpConstraintPool[swapi];
 					m_orderTmpConstraintPool[swapi] = tmp;
 				}
 
-				for (int j=0; j<numFrictionPool; ++j) {
+				for (int j = 0; j < numFrictionPool; ++j)
+				{
 					int tmp = m_orderFrictionConstraintPool[j];
-					int swapi = btRandInt2(j+1);
+					int swapi = btRandInt2(j + 1);
 					m_orderFrictionConstraintPool[j] = m_orderFrictionConstraintPool[swapi];
 					m_orderFrictionConstraintPool[swapi] = tmp;
 				}
@@ -1687,253 +1588,240 @@ btScalar btSequentialImpulseConstraintSolver::solveSingleIteration(int iteration
 		}
 	}
 
-		///solve all joint constraints
-		for (int j=0;j<m_tmpSolverNonContactConstraintPool.size();j++)
+	///solve all joint constraints
+	for (int j = 0; j < m_tmpSolverNonContactConstraintPool.size(); j++)
+	{
+		btSolverConstraint& constraint = m_tmpSolverNonContactConstraintPool[m_orderNonContactConstraintPool[j]];
+		if (iteration < constraint.m_overrideNumSolverIterations)
 		{
-			btSolverConstraint& constraint = m_tmpSolverNonContactConstraintPool[m_orderNonContactConstraintPool[j]];
-			if (iteration < constraint.m_overrideNumSolverIterations)
+			btScalar residual = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[constraint.m_solverBodyIdA], m_tmpSolverBodyPool[constraint.m_solverBodyIdB], constraint);
+			leastSquaresResidual = btMax(leastSquaresResidual, residual * residual);
+		}
+	}
+
+	if (iteration < infoGlobal.m_numIterations)
+	{
+		for (int j = 0; j < numConstraints; j++)
+		{
+			if (constraints[j]->isEnabled())
 			{
-				btScalar residual = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[constraint.m_solverBodyIdA],m_tmpSolverBodyPool[constraint.m_solverBodyIdB],constraint);
-				leastSquaresResidual = btMax(leastSquaresResidual, residual*residual);
+				int bodyAid = getOrInitSolverBody(constraints[j]->getRigidBodyA(), infoGlobal.m_timeStep);
+				int bodyBid = getOrInitSolverBody(constraints[j]->getRigidBodyB(), infoGlobal.m_timeStep);
+				btSolverBody& bodyA = m_tmpSolverBodyPool[bodyAid];
+				btSolverBody& bodyB = m_tmpSolverBodyPool[bodyBid];
+				constraints[j]->solveConstraintObsolete(bodyA, bodyB, infoGlobal.m_timeStep);
 			}
 		}
 
-		if (iteration< infoGlobal.m_numIterations)
+		///solve all contact constraints
+		if (infoGlobal.m_solverMode & SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS)
 		{
-			for (int j=0;j<numConstraints;j++)
+			int numPoolConstraints = m_tmpSolverContactConstraintPool.size();
+			int multiplier = (infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS) ? 2 : 1;
+
+			for (int c = 0; c < numPoolConstraints; c++)
 			{
-				if (constraints[j]->isEnabled())
+				btScalar totalImpulse = 0;
+
 				{
-					int bodyAid = getOrInitSolverBody(constraints[j]->getRigidBodyA(),infoGlobal.m_timeStep);
-					int bodyBid = getOrInitSolverBody(constraints[j]->getRigidBodyB(),infoGlobal.m_timeStep);
-					btSolverBody& bodyA = m_tmpSolverBodyPool[bodyAid];
-					btSolverBody& bodyB = m_tmpSolverBodyPool[bodyBid];
-					constraints[j]->solveConstraintObsolete(bodyA,bodyB,infoGlobal.m_timeStep);
+					const btSolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[m_orderTmpConstraintPool[c]];
+					btScalar residual = resolveSingleConstraintRowLowerLimit(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
+					leastSquaresResidual = btMax(leastSquaresResidual, residual * residual);
+
+					totalImpulse = solveManifold.m_appliedImpulse;
 				}
-			}
-
-			///solve all contact constraints
-			if (infoGlobal.m_solverMode & SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS)
-			{
-				int numPoolConstraints = m_tmpSolverContactConstraintPool.size();
-				int multiplier = (infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS)? 2 : 1;
-
-				for (int c=0;c<numPoolConstraints;c++)
+				bool applyFriction = true;
+				if (applyFriction)
 				{
-					btScalar totalImpulse =0;
-
 					{
-						const btSolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[m_orderTmpConstraintPool[c]];
-						btScalar residual = resolveSingleConstraintRowLowerLimit(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA],m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB],solveManifold);
-						leastSquaresResidual = btMax(leastSquaresResidual, residual*residual);
+						btSolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[m_orderFrictionConstraintPool[c * multiplier]];
 
-						totalImpulse = solveManifold.m_appliedImpulse;
-					}
-					bool applyFriction = true;
-					if (applyFriction)
-					{
+						if (totalImpulse > btScalar(0))
 						{
+							solveManifold.m_lowerLimit = -(solveManifold.m_friction * totalImpulse);
+							solveManifold.m_upperLimit = solveManifold.m_friction * totalImpulse;
 
-							btSolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[m_orderFrictionConstraintPool[c*multiplier]];
-
-							if (totalImpulse>btScalar(0))
-							{
-								solveManifold.m_lowerLimit = -(solveManifold.m_friction*totalImpulse);
-								solveManifold.m_upperLimit = solveManifold.m_friction*totalImpulse;
-
-								btScalar residual = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA],m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB],solveManifold);
-								leastSquaresResidual = btMax(leastSquaresResidual, residual*residual);
-							}
+							btScalar residual = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
+							leastSquaresResidual = btMax(leastSquaresResidual, residual * residual);
 						}
+					}
 
-						if (infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS)
+					if (infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS)
+					{
+						btSolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[m_orderFrictionConstraintPool[c * multiplier + 1]];
+
+						if (totalImpulse > btScalar(0))
 						{
+							solveManifold.m_lowerLimit = -(solveManifold.m_friction * totalImpulse);
+							solveManifold.m_upperLimit = solveManifold.m_friction * totalImpulse;
 
-							btSolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[m_orderFrictionConstraintPool[c*multiplier+1]];
-
-							if (totalImpulse>btScalar(0))
-							{
-								solveManifold.m_lowerLimit = -(solveManifold.m_friction*totalImpulse);
-								solveManifold.m_upperLimit = solveManifold.m_friction*totalImpulse;
-
-								btScalar residual = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA],m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB],solveManifold);
-								leastSquaresResidual = btMax(leastSquaresResidual, residual*residual);
-							}
+							btScalar residual = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
+							leastSquaresResidual = btMax(leastSquaresResidual, residual * residual);
 						}
 					}
 				}
-
 			}
-			else//SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS
-			{
-				//solve the friction constraints after all contact constraints, don't interleave them
-				int numPoolConstraints = m_tmpSolverContactConstraintPool.size();
-				int j;
-
-				for (j=0;j<numPoolConstraints;j++)
-				{
-					const btSolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[m_orderTmpConstraintPool[j]];
-					btScalar residual = resolveSingleConstraintRowLowerLimit(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA],m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB],solveManifold);
-					leastSquaresResidual = btMax(leastSquaresResidual, residual*residual);
-				}
-
-
-
-				///solve all friction constraints
-
-				int numFrictionPoolConstraints = m_tmpSolverContactFrictionConstraintPool.size();
-				for (j=0;j<numFrictionPoolConstraints;j++)
-				{
-					btSolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[m_orderFrictionConstraintPool[j]];
-					btScalar totalImpulse = m_tmpSolverContactConstraintPool[solveManifold.m_frictionIndex].m_appliedImpulse;
-
-					if (totalImpulse>btScalar(0))
-					{
-						solveManifold.m_lowerLimit = -(solveManifold.m_friction*totalImpulse);
-						solveManifold.m_upperLimit = solveManifold.m_friction*totalImpulse;
-
-						btScalar residual = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA],m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB],solveManifold);
-						leastSquaresResidual = btMax(leastSquaresResidual, residual*residual);
-					}
-				}
-			}
-
-
-				int numRollingFrictionPoolConstraints = m_tmpSolverContactRollingFrictionConstraintPool.size();
-				for (int j=0;j<numRollingFrictionPoolConstraints;j++)
-				{
-
-					btSolverConstraint& rollingFrictionConstraint = m_tmpSolverContactRollingFrictionConstraintPool[j];
-					btScalar totalImpulse = m_tmpSolverContactConstraintPool[rollingFrictionConstraint.m_frictionIndex].m_appliedImpulse;
-					if (totalImpulse>btScalar(0))
-					{
-						btScalar rollingFrictionMagnitude = rollingFrictionConstraint.m_friction*totalImpulse;
-						if (rollingFrictionMagnitude>rollingFrictionConstraint.m_friction)
-							rollingFrictionMagnitude = rollingFrictionConstraint.m_friction;
-
-						rollingFrictionConstraint.m_lowerLimit = -rollingFrictionMagnitude;
-						rollingFrictionConstraint.m_upperLimit = rollingFrictionMagnitude;
-
-						btScalar residual = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdA],m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdB],rollingFrictionConstraint);
-						leastSquaresResidual = btMax(leastSquaresResidual, residual*residual);
-					}
-				}
-
-
 		}
+		else  //SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS
+		{
+			//solve the friction constraints after all contact constraints, don't interleave them
+			int numPoolConstraints = m_tmpSolverContactConstraintPool.size();
+			int j;
+
+			for (j = 0; j < numPoolConstraints; j++)
+			{
+				const btSolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[m_orderTmpConstraintPool[j]];
+				btScalar residual = resolveSingleConstraintRowLowerLimit(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
+				leastSquaresResidual = btMax(leastSquaresResidual, residual * residual);
+			}
+
+			///solve all friction constraints
+
+			int numFrictionPoolConstraints = m_tmpSolverContactFrictionConstraintPool.size();
+			for (j = 0; j < numFrictionPoolConstraints; j++)
+			{
+				btSolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[m_orderFrictionConstraintPool[j]];
+				btScalar totalImpulse = m_tmpSolverContactConstraintPool[solveManifold.m_frictionIndex].m_appliedImpulse;
+
+				if (totalImpulse > btScalar(0))
+				{
+					solveManifold.m_lowerLimit = -(solveManifold.m_friction * totalImpulse);
+					solveManifold.m_upperLimit = solveManifold.m_friction * totalImpulse;
+
+					btScalar residual = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
+					leastSquaresResidual = btMax(leastSquaresResidual, residual * residual);
+				}
+			}
+		}
+
+		int numRollingFrictionPoolConstraints = m_tmpSolverContactRollingFrictionConstraintPool.size();
+		for (int j = 0; j < numRollingFrictionPoolConstraints; j++)
+		{
+			btSolverConstraint& rollingFrictionConstraint = m_tmpSolverContactRollingFrictionConstraintPool[j];
+			btScalar totalImpulse = m_tmpSolverContactConstraintPool[rollingFrictionConstraint.m_frictionIndex].m_appliedImpulse;
+			if (totalImpulse > btScalar(0))
+			{
+				btScalar rollingFrictionMagnitude = rollingFrictionConstraint.m_friction * totalImpulse;
+				if (rollingFrictionMagnitude > rollingFrictionConstraint.m_friction)
+					rollingFrictionMagnitude = rollingFrictionConstraint.m_friction;
+
+				rollingFrictionConstraint.m_lowerLimit = -rollingFrictionMagnitude;
+				rollingFrictionConstraint.m_upperLimit = rollingFrictionMagnitude;
+
+				btScalar residual = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdA], m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdB], rollingFrictionConstraint);
+				leastSquaresResidual = btMax(leastSquaresResidual, residual * residual);
+			}
+		}
+	}
 	return leastSquaresResidual;
 }
 
-
-void btSequentialImpulseConstraintSolver::solveGroupCacheFriendlySplitImpulseIterations(btCollisionObject** bodies,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer)
+void btSequentialImpulseConstraintSolver::solveGroupCacheFriendlySplitImpulseIterations(btCollisionObject** bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer)
 {
 	BT_PROFILE("solveGroupCacheFriendlySplitImpulseIterations");
 	int iteration;
 	if (infoGlobal.m_splitImpulse)
 	{
 		{
-			for ( iteration = 0;iteration<infoGlobal.m_numIterations;iteration++)
+			for (iteration = 0; iteration < infoGlobal.m_numIterations; iteration++)
 			{
-				btScalar leastSquaresResidual =0.f;
+				btScalar leastSquaresResidual = 0.f;
 				{
 					int numPoolConstraints = m_tmpSolverContactConstraintPool.size();
 					int j;
-					for (j=0;j<numPoolConstraints;j++)
+					for (j = 0; j < numPoolConstraints; j++)
 					{
 						const btSolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[m_orderTmpConstraintPool[j]];
 
-						btScalar residual = resolveSplitPenetrationImpulse(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA],m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB],solveManifold);
-						leastSquaresResidual = btMax(leastSquaresResidual, residual*residual);
+						btScalar residual = resolveSplitPenetrationImpulse(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
+						leastSquaresResidual = btMax(leastSquaresResidual, residual * residual);
 					}
 				}
-				if (leastSquaresResidual <= infoGlobal.m_leastSquaresResidualThreshold || iteration>=(infoGlobal.m_numIterations-1))
+				if (leastSquaresResidual <= infoGlobal.m_leastSquaresResidualThreshold || iteration >= (infoGlobal.m_numIterations - 1))
 				{
 #ifdef VERBOSE_RESIDUAL_PRINTF
-					printf("residual = %f at iteration #%d\n",leastSquaresResidual,iteration);
+					printf("residual = %f at iteration #%d\n", leastSquaresResidual, iteration);
 #endif
 					break;
 				}
 			}
-        }
+		}
 	}
 }
 
-btScalar btSequentialImpulseConstraintSolver::solveGroupCacheFriendlyIterations(btCollisionObject** bodies ,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer)
+btScalar btSequentialImpulseConstraintSolver::solveGroupCacheFriendlyIterations(btCollisionObject** bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer)
 {
 	BT_PROFILE("solveGroupCacheFriendlyIterations");
 
 	{
 		///this is a special step to resolve penetrations (just for contacts)
-		solveGroupCacheFriendlySplitImpulseIterations(bodies ,numBodies,manifoldPtr, numManifolds,constraints,numConstraints,infoGlobal,debugDrawer);
+		solveGroupCacheFriendlySplitImpulseIterations(bodies, numBodies, manifoldPtr, numManifolds, constraints, numConstraints, infoGlobal, debugDrawer);
 
-		int maxIterations = m_maxOverrideNumSolverIterations > infoGlobal.m_numIterations? m_maxOverrideNumSolverIterations : infoGlobal.m_numIterations;
+		int maxIterations = m_maxOverrideNumSolverIterations > infoGlobal.m_numIterations ? m_maxOverrideNumSolverIterations : infoGlobal.m_numIterations;
 
-		for ( int iteration = 0 ; iteration< maxIterations ; iteration++)
+		for (int iteration = 0; iteration < maxIterations; iteration++)
 		//for ( int iteration = maxIterations-1  ; iteration >= 0;iteration--)
 		{
-			m_leastSquaresResidual = solveSingleIteration(iteration, bodies ,numBodies,manifoldPtr, numManifolds,constraints,numConstraints,infoGlobal,debugDrawer);
+			m_leastSquaresResidual = solveSingleIteration(iteration, bodies, numBodies, manifoldPtr, numManifolds, constraints, numConstraints, infoGlobal, debugDrawer);
 
-			if (m_leastSquaresResidual <= infoGlobal.m_leastSquaresResidualThreshold || (iteration>= (maxIterations-1)))
+			if (m_leastSquaresResidual <= infoGlobal.m_leastSquaresResidualThreshold || (iteration >= (maxIterations - 1)))
 			{
 #ifdef VERBOSE_RESIDUAL_PRINTF
-						printf("residual = %f at iteration #%d\n",m_leastSquaresResidual,iteration);
+				printf("residual = %f at iteration #%d\n", m_leastSquaresResidual, iteration);
 #endif
 				break;
 			}
 		}
-
 	}
 	return 0.f;
 }
 
 void btSequentialImpulseConstraintSolver::writeBackContacts(int iBegin, int iEnd, const btContactSolverInfo& infoGlobal)
 {
-		for (int j=iBegin; j<iEnd; j++)
-		{
-			const btSolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[j];
-			btManifoldPoint* pt = (btManifoldPoint*) solveManifold.m_originalContactPoint;
-			btAssert(pt);
-			pt->m_appliedImpulse = solveManifold.m_appliedImpulse;
+	for (int j = iBegin; j < iEnd; j++)
+	{
+		const btSolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[j];
+		btManifoldPoint* pt = (btManifoldPoint*)solveManifold.m_originalContactPoint;
+		btAssert(pt);
+		pt->m_appliedImpulse = solveManifold.m_appliedImpulse;
 		//	float f = m_tmpSolverContactFrictionConstraintPool[solveManifold.m_frictionIndex].m_appliedImpulse;
-			//	printf("pt->m_appliedImpulseLateral1 = %f\n", f);
-			pt->m_appliedImpulseLateral1 = m_tmpSolverContactFrictionConstraintPool[solveManifold.m_frictionIndex].m_appliedImpulse;
-			//printf("pt->m_appliedImpulseLateral1 = %f\n", pt->m_appliedImpulseLateral1);
-			if ((infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS))
-			{
-				pt->m_appliedImpulseLateral2 = m_tmpSolverContactFrictionConstraintPool[solveManifold.m_frictionIndex+1].m_appliedImpulse;
-			}
-			//do a callback here?
+		//	printf("pt->m_appliedImpulseLateral1 = %f\n", f);
+		pt->m_appliedImpulseLateral1 = m_tmpSolverContactFrictionConstraintPool[solveManifold.m_frictionIndex].m_appliedImpulse;
+		//printf("pt->m_appliedImpulseLateral1 = %f\n", pt->m_appliedImpulseLateral1);
+		if ((infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS))
+		{
+			pt->m_appliedImpulseLateral2 = m_tmpSolverContactFrictionConstraintPool[solveManifold.m_frictionIndex + 1].m_appliedImpulse;
 		}
+		//do a callback here?
+	}
 }
 
 void btSequentialImpulseConstraintSolver::writeBackJoints(int iBegin, int iEnd, const btContactSolverInfo& infoGlobal)
 {
-	for (int j=iBegin; j<iEnd; j++)
+	for (int j = iBegin; j < iEnd; j++)
 	{
 		const btSolverConstraint& solverConstr = m_tmpSolverNonContactConstraintPool[j];
 		btTypedConstraint* constr = (btTypedConstraint*)solverConstr.m_originalContactPoint;
 		btJointFeedback* fb = constr->getJointFeedback();
 		if (fb)
 		{
-			fb->m_appliedForceBodyA += solverConstr.m_contactNormal1*solverConstr.m_appliedImpulse*constr->getRigidBodyA().getLinearFactor()/infoGlobal.m_timeStep;
-			fb->m_appliedForceBodyB += solverConstr.m_contactNormal2*solverConstr.m_appliedImpulse*constr->getRigidBodyB().getLinearFactor()/infoGlobal.m_timeStep;
-			fb->m_appliedTorqueBodyA += solverConstr.m_relpos1CrossNormal* constr->getRigidBodyA().getAngularFactor()*solverConstr.m_appliedImpulse/infoGlobal.m_timeStep;
-			fb->m_appliedTorqueBodyB += solverConstr.m_relpos2CrossNormal* constr->getRigidBodyB().getAngularFactor()*solverConstr.m_appliedImpulse/infoGlobal.m_timeStep; /*RGM ???? */
-
+			fb->m_appliedForceBodyA += solverConstr.m_contactNormal1 * solverConstr.m_appliedImpulse * constr->getRigidBodyA().getLinearFactor() / infoGlobal.m_timeStep;
+			fb->m_appliedForceBodyB += solverConstr.m_contactNormal2 * solverConstr.m_appliedImpulse * constr->getRigidBodyB().getLinearFactor() / infoGlobal.m_timeStep;
+			fb->m_appliedTorqueBodyA += solverConstr.m_relpos1CrossNormal * constr->getRigidBodyA().getAngularFactor() * solverConstr.m_appliedImpulse / infoGlobal.m_timeStep;
+			fb->m_appliedTorqueBodyB += solverConstr.m_relpos2CrossNormal * constr->getRigidBodyB().getAngularFactor() * solverConstr.m_appliedImpulse / infoGlobal.m_timeStep; /*RGM ???? */
 		}
 
 		constr->internalSetAppliedImpulse(solverConstr.m_appliedImpulse);
-		if (btFabs(solverConstr.m_appliedImpulse)>=constr->getBreakingImpulseThreshold())
+		if (btFabs(solverConstr.m_appliedImpulse) >= constr->getBreakingImpulseThreshold())
 		{
 			constr->setEnabled(false);
 		}
 	}
 }
 
-
 void btSequentialImpulseConstraintSolver::writeBackBodies(int iBegin, int iEnd, const btContactSolverInfo& infoGlobal)
 {
-	for (int i=iBegin; i<iEnd; i++)
+	for (int i = iBegin; i < iEnd; i++)
 	{
 		btRigidBody* body = m_tmpSolverBodyPool[i].m_originalBody;
 		if (body)
@@ -1944,11 +1832,11 @@ void btSequentialImpulseConstraintSolver::writeBackBodies(int iBegin, int iEnd,
 				m_tmpSolverBodyPool[i].writebackVelocity();
 
 			m_tmpSolverBodyPool[i].m_originalBody->setLinearVelocity(
-				m_tmpSolverBodyPool[i].m_linearVelocity+
+				m_tmpSolverBodyPool[i].m_linearVelocity +
 				m_tmpSolverBodyPool[i].m_externalForceImpulse);
 
 			m_tmpSolverBodyPool[i].m_originalBody->setAngularVelocity(
-				m_tmpSolverBodyPool[i].m_angularVelocity+
+				m_tmpSolverBodyPool[i].m_angularVelocity +
 				m_tmpSolverBodyPool[i].m_externalTorqueImpulse);
 
 			if (infoGlobal.m_splitImpulse)
@@ -1959,17 +1847,17 @@ void btSequentialImpulseConstraintSolver::writeBackBodies(int iBegin, int iEnd,
 	}
 }
 
-btScalar btSequentialImpulseConstraintSolver::solveGroupCacheFriendlyFinish(btCollisionObject** bodies,int numBodies,const btContactSolverInfo& infoGlobal)
+btScalar btSequentialImpulseConstraintSolver::solveGroupCacheFriendlyFinish(btCollisionObject** bodies, int numBodies, const btContactSolverInfo& infoGlobal)
 {
 	BT_PROFILE("solveGroupCacheFriendlyFinish");
 
 	if (infoGlobal.m_solverMode & SOLVER_USE_WARMSTARTING)
 	{
-        writeBackContacts(0, m_tmpSolverContactConstraintPool.size(), infoGlobal);
+		writeBackContacts(0, m_tmpSolverContactConstraintPool.size(), infoGlobal);
 	}
 
-    writeBackJoints(0, m_tmpSolverNonContactConstraintPool.size(), infoGlobal);
-    writeBackBodies(0, m_tmpSolverBodyPool.size(), infoGlobal);
+	writeBackJoints(0, m_tmpSolverNonContactConstraintPool.size(), infoGlobal);
+	writeBackBodies(0, m_tmpSolverBodyPool.size(), infoGlobal);
 
 	m_tmpSolverContactConstraintPool.resizeNoInitialize(0);
 	m_tmpSolverNonContactConstraintPool.resizeNoInitialize(0);
@@ -1980,25 +1868,22 @@ btScalar btSequentialImpulseConstraintSolver::solveGroupCacheFriendlyFinish(btCo
 	return 0.f;
 }
 
-
-
 /// btSequentialImpulseConstraintSolver Sequentially applies impulses
-btScalar btSequentialImpulseConstraintSolver::solveGroup(btCollisionObject** bodies,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer,btDispatcher* /*dispatcher*/)
+btScalar btSequentialImpulseConstraintSolver::solveGroup(btCollisionObject** bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer, btDispatcher* /*dispatcher*/)
 {
-
 	BT_PROFILE("solveGroup");
 	//you need to provide at least some bodies
 
-	solveGroupCacheFriendlySetup( bodies, numBodies, manifoldPtr,  numManifolds,constraints, numConstraints,infoGlobal,debugDrawer);
+	solveGroupCacheFriendlySetup(bodies, numBodies, manifoldPtr, numManifolds, constraints, numConstraints, infoGlobal, debugDrawer);
 
-	solveGroupCacheFriendlyIterations(bodies, numBodies, manifoldPtr,  numManifolds,constraints, numConstraints,infoGlobal,debugDrawer);
+	solveGroupCacheFriendlyIterations(bodies, numBodies, manifoldPtr, numManifolds, constraints, numConstraints, infoGlobal, debugDrawer);
 
 	solveGroupCacheFriendlyFinish(bodies, numBodies, infoGlobal);
 
 	return 0.f;
 }
 
-void	btSequentialImpulseConstraintSolver::reset()
+void btSequentialImpulseConstraintSolver::reset()
 {
 	m_btSeed2 = 0;
 }
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSequentialImpulseConstraintSolver.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSequentialImpulseConstraintSolver.h
index b834c3dac3b..70db83b063d 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSequentialImpulseConstraintSolver.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSequentialImpulseConstraintSolver.h
@@ -27,147 +27,142 @@ class btCollisionObject;
 #include "BulletCollision/NarrowPhaseCollision/btManifoldPoint.h"
 #include "BulletDynamics/ConstraintSolver/btConstraintSolver.h"
 
-typedef btScalar(*btSingleConstraintRowSolver)(btSolverBody&, btSolverBody&, const btSolverConstraint&);
+typedef btScalar (*btSingleConstraintRowSolver)(btSolverBody&, btSolverBody&, const btSolverConstraint&);
 
 ///The btSequentialImpulseConstraintSolver is a fast SIMD implementation of the Projected Gauss Seidel (iterative LCP) method.
-ATTRIBUTE_ALIGNED16(class) btSequentialImpulseConstraintSolver : public btConstraintSolver
+ATTRIBUTE_ALIGNED16(class)
+btSequentialImpulseConstraintSolver : public btConstraintSolver
 {
 protected:
-	btAlignedObjectArray<btSolverBody>      m_tmpSolverBodyPool;
-	btConstraintArray			m_tmpSolverContactConstraintPool;
-	btConstraintArray			m_tmpSolverNonContactConstraintPool;
-	btConstraintArray			m_tmpSolverContactFrictionConstraintPool;
-	btConstraintArray			m_tmpSolverContactRollingFrictionConstraintPool;
+	btAlignedObjectArray<btSolverBody> m_tmpSolverBodyPool;
+	btConstraintArray m_tmpSolverContactConstraintPool;
+	btConstraintArray m_tmpSolverNonContactConstraintPool;
+	btConstraintArray m_tmpSolverContactFrictionConstraintPool;
+	btConstraintArray m_tmpSolverContactRollingFrictionConstraintPool;
 
-	btAlignedObjectArray<int>	m_orderTmpConstraintPool;
-	btAlignedObjectArray<int>	m_orderNonContactConstraintPool;
-	btAlignedObjectArray<int>	m_orderFrictionConstraintPool;
+	btAlignedObjectArray<int> m_orderTmpConstraintPool;
+	btAlignedObjectArray<int> m_orderNonContactConstraintPool;
+	btAlignedObjectArray<int> m_orderFrictionConstraintPool;
 	btAlignedObjectArray<btTypedConstraint::btConstraintInfo1> m_tmpConstraintSizesPool;
-	int							m_maxOverrideNumSolverIterations;
+	int m_maxOverrideNumSolverIterations;
 	int m_fixedBodyId;
-    // When running solvers on multiple threads, a race condition exists for Kinematic objects that
-    // participate in more than one solver.
-    // The getOrInitSolverBody() function writes the companionId of each body (storing the index of the solver body
-    // for the current solver). For normal dynamic bodies it isn't an issue because they can only be in one island
-    // (and therefore one thread) at a time. But kinematic bodies can be in multiple islands at once.
-    // To avoid this race condition, this solver does not write the companionId, instead it stores the solver body
-    // index in this solver-local table, indexed by the uniqueId of the body.
-    btAlignedObjectArray<int>	m_kinematicBodyUniqueIdToSolverBodyTable;  // only used for multithreading
+	// When running solvers on multiple threads, a race condition exists for Kinematic objects that
+	// participate in more than one solver.
+	// The getOrInitSolverBody() function writes the companionId of each body (storing the index of the solver body
+	// for the current solver). For normal dynamic bodies it isn't an issue because they can only be in one island
+	// (and therefore one thread) at a time. But kinematic bodies can be in multiple islands at once.
+	// To avoid this race condition, this solver does not write the companionId, instead it stores the solver body
+	// index in this solver-local table, indexed by the uniqueId of the body.
+	btAlignedObjectArray<int> m_kinematicBodyUniqueIdToSolverBodyTable;  // only used for multithreading
 
 	btSingleConstraintRowSolver m_resolveSingleConstraintRowGeneric;
 	btSingleConstraintRowSolver m_resolveSingleConstraintRowLowerLimit;
-    btSingleConstraintRowSolver m_resolveSplitPenetrationImpulse;
-    int m_cachedSolverMode;  // used to check if SOLVER_SIMD flag has been changed
-    void setupSolverFunctions( bool useSimd );
+	btSingleConstraintRowSolver m_resolveSplitPenetrationImpulse;
+	int m_cachedSolverMode;  // used to check if SOLVER_SIMD flag has been changed
+	void setupSolverFunctions(bool useSimd);
 
-	btScalar	m_leastSquaresResidual;
+	btScalar m_leastSquaresResidual;
 
-	void setupFrictionConstraint(	btSolverConstraint& solverConstraint, const btVector3& normalAxis,int solverBodyIdA,int  solverBodyIdB,
-									btManifoldPoint& cp,const btVector3& rel_pos1,const btVector3& rel_pos2,
-									btCollisionObject* colObj0,btCollisionObject* colObj1, btScalar relaxation,
-									const btContactSolverInfo& infoGlobal,
-									btScalar desiredVelocity=0., btScalar cfmSlip=0.);
+	void setupFrictionConstraint(btSolverConstraint & solverConstraint, const btVector3& normalAxis, int solverBodyIdA, int solverBodyIdB,
+								 btManifoldPoint& cp, const btVector3& rel_pos1, const btVector3& rel_pos2,
+								 btCollisionObject* colObj0, btCollisionObject* colObj1, btScalar relaxation,
+								 const btContactSolverInfo& infoGlobal,
+								 btScalar desiredVelocity = 0., btScalar cfmSlip = 0.);
 
-	void setupTorsionalFrictionConstraint(	btSolverConstraint& solverConstraint, const btVector3& normalAxis,int solverBodyIdA,int  solverBodyIdB,
-									btManifoldPoint& cp,btScalar combinedTorsionalFriction, const btVector3& rel_pos1,const btVector3& rel_pos2,
-									btCollisionObject* colObj0,btCollisionObject* colObj1, btScalar relaxation,
-									btScalar desiredVelocity=0., btScalar cfmSlip=0.);
+	void setupTorsionalFrictionConstraint(btSolverConstraint & solverConstraint, const btVector3& normalAxis, int solverBodyIdA, int solverBodyIdB,
+										  btManifoldPoint& cp, btScalar combinedTorsionalFriction, const btVector3& rel_pos1, const btVector3& rel_pos2,
+										  btCollisionObject* colObj0, btCollisionObject* colObj1, btScalar relaxation,
+										  btScalar desiredVelocity = 0., btScalar cfmSlip = 0.);
 
-	btSolverConstraint&	addFrictionConstraint(const btVector3& normalAxis,int solverBodyIdA,int solverBodyIdB,int frictionIndex,btManifoldPoint& cp,const btVector3& rel_pos1,const btVector3& rel_pos2,btCollisionObject* colObj0,btCollisionObject* colObj1, btScalar relaxation, const btContactSolverInfo& infoGlobal, btScalar desiredVelocity=0., btScalar cfmSlip=0.);
-	btSolverConstraint&	addTorsionalFrictionConstraint(const btVector3& normalAxis,int solverBodyIdA,int solverBodyIdB,int frictionIndex,btManifoldPoint& cp,btScalar torsionalFriction, const btVector3& rel_pos1,const btVector3& rel_pos2,btCollisionObject* colObj0,btCollisionObject* colObj1, btScalar relaxation, btScalar desiredVelocity=0, btScalar cfmSlip=0.f);
+	btSolverConstraint& addFrictionConstraint(const btVector3& normalAxis, int solverBodyIdA, int solverBodyIdB, int frictionIndex, btManifoldPoint& cp, const btVector3& rel_pos1, const btVector3& rel_pos2, btCollisionObject* colObj0, btCollisionObject* colObj1, btScalar relaxation, const btContactSolverInfo& infoGlobal, btScalar desiredVelocity = 0., btScalar cfmSlip = 0.);
+	btSolverConstraint& addTorsionalFrictionConstraint(const btVector3& normalAxis, int solverBodyIdA, int solverBodyIdB, int frictionIndex, btManifoldPoint& cp, btScalar torsionalFriction, const btVector3& rel_pos1, const btVector3& rel_pos2, btCollisionObject* colObj0, btCollisionObject* colObj1, btScalar relaxation, btScalar desiredVelocity = 0, btScalar cfmSlip = 0.f);
 
+	void setupContactConstraint(btSolverConstraint & solverConstraint, int solverBodyIdA, int solverBodyIdB, btManifoldPoint& cp,
+								const btContactSolverInfo& infoGlobal, btScalar& relaxation, const btVector3& rel_pos1, const btVector3& rel_pos2);
 
-	void setupContactConstraint(btSolverConstraint& solverConstraint, int solverBodyIdA, int solverBodyIdB, btManifoldPoint& cp,
-								const btContactSolverInfo& infoGlobal,btScalar& relaxation, const btVector3& rel_pos1, const btVector3& rel_pos2);
+	static void applyAnisotropicFriction(btCollisionObject * colObj, btVector3 & frictionDirection, int frictionMode);
 
-	static void	applyAnisotropicFriction(btCollisionObject* colObj,btVector3& frictionDirection, int frictionMode);
-
-	void setFrictionConstraintImpulse( btSolverConstraint& solverConstraint, int solverBodyIdA,int solverBodyIdB,
-										 btManifoldPoint& cp, const btContactSolverInfo& infoGlobal);
+	void setFrictionConstraintImpulse(btSolverConstraint & solverConstraint, int solverBodyIdA, int solverBodyIdB,
+									  btManifoldPoint& cp, const btContactSolverInfo& infoGlobal);
 
 	///m_btSeed2 is used for re-arranging the constraint rows. improves convergence/quality of friction
-	unsigned long	m_btSeed2;
-
+	unsigned long m_btSeed2;
 
 	btScalar restitutionCurve(btScalar rel_vel, btScalar restitution, btScalar velocityThreshold);
 
-	virtual void convertContacts(btPersistentManifold** manifoldPtr, int numManifolds, const btContactSolverInfo& infoGlobal);
+	virtual void convertContacts(btPersistentManifold * *manifoldPtr, int numManifolds, const btContactSolverInfo& infoGlobal);
 
-	void	convertContact(btPersistentManifold* manifold,const btContactSolverInfo& infoGlobal);
+	void convertContact(btPersistentManifold * manifold, const btContactSolverInfo& infoGlobal);
 
-    virtual void convertJoints(btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal);
-    void convertJoint(btSolverConstraint* currentConstraintRow, btTypedConstraint* constraint, const btTypedConstraint::btConstraintInfo1& info1, int solverBodyIdA, int solverBodyIdB, const btContactSolverInfo& infoGlobal);
+	virtual void convertJoints(btTypedConstraint * *constraints, int numConstraints, const btContactSolverInfo& infoGlobal);
+	void convertJoint(btSolverConstraint * currentConstraintRow, btTypedConstraint * constraint, const btTypedConstraint::btConstraintInfo1& info1, int solverBodyIdA, int solverBodyIdB, const btContactSolverInfo& infoGlobal);
 
-    virtual void convertBodies(btCollisionObject** bodies, int numBodies, const btContactSolverInfo& infoGlobal);
+	virtual void convertBodies(btCollisionObject * *bodies, int numBodies, const btContactSolverInfo& infoGlobal);
 
-	btScalar	resolveSplitPenetrationSIMD(btSolverBody& bodyA,btSolverBody& bodyB, const btSolverConstraint& contactConstraint)
-    {
-        return m_resolveSplitPenetrationImpulse( bodyA, bodyB, contactConstraint );
-    }
+	btScalar resolveSplitPenetrationSIMD(btSolverBody & bodyA, btSolverBody & bodyB, const btSolverConstraint& contactConstraint)
+	{
+		return m_resolveSplitPenetrationImpulse(bodyA, bodyB, contactConstraint);
+	}
 
-	btScalar	resolveSplitPenetrationImpulseCacheFriendly(btSolverBody& bodyA,btSolverBody& bodyB, const btSolverConstraint& contactConstraint)
-    {
-        return m_resolveSplitPenetrationImpulse( bodyA, bodyB, contactConstraint );
-    }
+	btScalar resolveSplitPenetrationImpulseCacheFriendly(btSolverBody & bodyA, btSolverBody & bodyB, const btSolverConstraint& contactConstraint)
+	{
+		return m_resolveSplitPenetrationImpulse(bodyA, bodyB, contactConstraint);
+	}
 
 	//internal method
-	int		getOrInitSolverBody(btCollisionObject& body,btScalar timeStep);
-	void	initSolverBody(btSolverBody* solverBody, btCollisionObject* collisionObject, btScalar timeStep);
+	int getOrInitSolverBody(btCollisionObject & body, btScalar timeStep);
+	void initSolverBody(btSolverBody * solverBody, btCollisionObject * collisionObject, btScalar timeStep);
 
-	btScalar	resolveSingleConstraintRowGeneric(btSolverBody& bodyA,btSolverBody& bodyB,const btSolverConstraint& contactConstraint);
-	btScalar	resolveSingleConstraintRowGenericSIMD(btSolverBody& bodyA,btSolverBody& bodyB,const btSolverConstraint& contactConstraint);
-	btScalar	resolveSingleConstraintRowLowerLimit(btSolverBody& bodyA,btSolverBody& bodyB,const btSolverConstraint& contactConstraint);
-	btScalar	resolveSingleConstraintRowLowerLimitSIMD(btSolverBody& bodyA,btSolverBody& bodyB,const btSolverConstraint& contactConstraint);
-	btScalar	resolveSplitPenetrationImpulse(btSolverBody& bodyA,btSolverBody& bodyB, const btSolverConstraint& contactConstraint)
-    {
-        return m_resolveSplitPenetrationImpulse( bodyA, bodyB, contactConstraint );
-    }
+	btScalar resolveSingleConstraintRowGeneric(btSolverBody & bodyA, btSolverBody & bodyB, const btSolverConstraint& contactConstraint);
+	btScalar resolveSingleConstraintRowGenericSIMD(btSolverBody & bodyA, btSolverBody & bodyB, const btSolverConstraint& contactConstraint);
+	btScalar resolveSingleConstraintRowLowerLimit(btSolverBody & bodyA, btSolverBody & bodyB, const btSolverConstraint& contactConstraint);
+	btScalar resolveSingleConstraintRowLowerLimitSIMD(btSolverBody & bodyA, btSolverBody & bodyB, const btSolverConstraint& contactConstraint);
+	btScalar resolveSplitPenetrationImpulse(btSolverBody & bodyA, btSolverBody & bodyB, const btSolverConstraint& contactConstraint)
+	{
+		return m_resolveSplitPenetrationImpulse(bodyA, bodyB, contactConstraint);
+	}
 
 protected:
+	void writeBackContacts(int iBegin, int iEnd, const btContactSolverInfo& infoGlobal);
+	void writeBackJoints(int iBegin, int iEnd, const btContactSolverInfo& infoGlobal);
+	void writeBackBodies(int iBegin, int iEnd, const btContactSolverInfo& infoGlobal);
+	virtual void solveGroupCacheFriendlySplitImpulseIterations(btCollisionObject * *bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer);
+	virtual btScalar solveGroupCacheFriendlyFinish(btCollisionObject * *bodies, int numBodies, const btContactSolverInfo& infoGlobal);
+	virtual btScalar solveSingleIteration(int iteration, btCollisionObject** bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer);
 
-    void writeBackContacts(int iBegin, int iEnd, const btContactSolverInfo& infoGlobal);
-    void writeBackJoints(int iBegin, int iEnd, const btContactSolverInfo& infoGlobal);
-    void writeBackBodies(int iBegin, int iEnd, const btContactSolverInfo& infoGlobal);
-	virtual void solveGroupCacheFriendlySplitImpulseIterations(btCollisionObject** bodies,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer);
-	virtual btScalar solveGroupCacheFriendlyFinish(btCollisionObject** bodies,int numBodies,const btContactSolverInfo& infoGlobal);
-	virtual btScalar solveSingleIteration(int iteration, btCollisionObject** bodies ,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer);
-
-	virtual btScalar solveGroupCacheFriendlySetup(btCollisionObject** bodies,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer);
-	virtual btScalar solveGroupCacheFriendlyIterations(btCollisionObject** bodies,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer);
-
+	virtual btScalar solveGroupCacheFriendlySetup(btCollisionObject * *bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer);
+	virtual btScalar solveGroupCacheFriendlyIterations(btCollisionObject * *bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer);
 
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
 	btSequentialImpulseConstraintSolver();
 	virtual ~btSequentialImpulseConstraintSolver();
 
-	virtual btScalar solveGroup(btCollisionObject** bodies,int numBodies,btPersistentManifold** manifold,int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& info, btIDebugDraw* debugDrawer,btDispatcher* dispatcher);
+	virtual btScalar solveGroup(btCollisionObject * *bodies, int numBodies, btPersistentManifold** manifold, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& info, btIDebugDraw* debugDrawer, btDispatcher* dispatcher);
 
 	///clear internal cached data and reset random seed
-	virtual	void	reset();
+	virtual void reset();
 
 	unsigned long btRand2();
 
-	int btRandInt2 (int n);
+	int btRandInt2(int n);
 
-	void	setRandSeed(unsigned long seed)
+	void setRandSeed(unsigned long seed)
 	{
 		m_btSeed2 = seed;
 	}
-	unsigned long	getRandSeed() const
+	unsigned long getRandSeed() const
 	{
 		return m_btSeed2;
 	}
 
-
-	virtual btConstraintSolverType	getSolverType() const
+	virtual btConstraintSolverType getSolverType() const
 	{
 		return BT_SEQUENTIAL_IMPULSE_SOLVER;
 	}
 
-	btSingleConstraintRowSolver	getActiveConstraintRowSolverGeneric()
+	btSingleConstraintRowSolver getActiveConstraintRowSolverGeneric()
 	{
 		return m_resolveSingleConstraintRowGeneric;
 	}
@@ -175,7 +170,7 @@ public:
 	{
 		m_resolveSingleConstraintRowGeneric = rowSolver;
 	}
-	btSingleConstraintRowSolver	getActiveConstraintRowSolverLowerLimit()
+	btSingleConstraintRowSolver getActiveConstraintRowSolverLowerLimit()
 	{
 		return m_resolveSingleConstraintRowLowerLimit;
 	}
@@ -185,18 +180,14 @@ public:
 	}
 
 	///Various implementations of solving a single constraint row using a generic equality constraint, using scalar reference, SSE2 or SSE4
-	btSingleConstraintRowSolver	getScalarConstraintRowSolverGeneric();
-	btSingleConstraintRowSolver	getSSE2ConstraintRowSolverGeneric();
-	btSingleConstraintRowSolver	getSSE4_1ConstraintRowSolverGeneric();
+	btSingleConstraintRowSolver getScalarConstraintRowSolverGeneric();
+	btSingleConstraintRowSolver getSSE2ConstraintRowSolverGeneric();
+	btSingleConstraintRowSolver getSSE4_1ConstraintRowSolverGeneric();
 
 	///Various implementations of solving a single constraint row using an inequality (lower limit) constraint, using scalar reference, SSE2 or SSE4
-	btSingleConstraintRowSolver	getScalarConstraintRowSolverLowerLimit();
-	btSingleConstraintRowSolver	getSSE2ConstraintRowSolverLowerLimit();
-	btSingleConstraintRowSolver	getSSE4_1ConstraintRowSolverLowerLimit();
+	btSingleConstraintRowSolver getScalarConstraintRowSolverLowerLimit();
+	btSingleConstraintRowSolver getSSE2ConstraintRowSolverLowerLimit();
+	btSingleConstraintRowSolver getSSE4_1ConstraintRowSolverLowerLimit();
 };
 
-
-
-
-#endif //BT_SEQUENTIAL_IMPULSE_CONSTRAINT_SOLVER_H
-
+#endif  //BT_SEQUENTIAL_IMPULSE_CONSTRAINT_SOLVER_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSequentialImpulseConstraintSolverMt.cpp b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSequentialImpulseConstraintSolverMt.cpp
index 4306c37e495..2718da4a501 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSequentialImpulseConstraintSolverMt.cpp
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSequentialImpulseConstraintSolverMt.cpp
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btSequentialImpulseConstraintSolverMt.h"
 
 #include "LinearMath/btQuickprof.h"
@@ -23,8 +22,6 @@ subject to the following restrictions:
 #include "BulletDynamics/ConstraintSolver/btTypedConstraint.h"
 #include "BulletDynamics/Dynamics/btRigidBody.h"
 
-
-
 bool btSequentialImpulseConstraintSolverMt::s_allowNestedParallelForLoops = false;  // some task schedulers don't like nested loops
 int btSequentialImpulseConstraintSolverMt::s_minimumContactManifoldsForBatching = 250;
 int btSequentialImpulseConstraintSolverMt::s_minBatchSize = 50;
@@ -32,613 +29,594 @@ int btSequentialImpulseConstraintSolverMt::s_maxBatchSize = 100;
 btBatchedConstraints::BatchingMethod btSequentialImpulseConstraintSolverMt::s_contactBatchingMethod = btBatchedConstraints::BATCHING_METHOD_SPATIAL_GRID_2D;
 btBatchedConstraints::BatchingMethod btSequentialImpulseConstraintSolverMt::s_jointBatchingMethod = btBatchedConstraints::BATCHING_METHOD_SPATIAL_GRID_2D;
 
-
 btSequentialImpulseConstraintSolverMt::btSequentialImpulseConstraintSolverMt()
 {
-    m_numFrictionDirections = 1;
-    m_useBatching = false;
-    m_useObsoleteJointConstraints = false;
+	m_numFrictionDirections = 1;
+	m_useBatching = false;
+	m_useObsoleteJointConstraints = false;
 }
 
-
 btSequentialImpulseConstraintSolverMt::~btSequentialImpulseConstraintSolverMt()
 {
 }
 
-
 void btSequentialImpulseConstraintSolverMt::setupBatchedContactConstraints()
 {
-    BT_PROFILE("setupBatchedContactConstraints");
-    m_batchedContactConstraints.setup( &m_tmpSolverContactConstraintPool,
-        m_tmpSolverBodyPool,
-        s_contactBatchingMethod,
-        s_minBatchSize,
-        s_maxBatchSize,
-        &m_scratchMemory
-    );
+	BT_PROFILE("setupBatchedContactConstraints");
+	m_batchedContactConstraints.setup(&m_tmpSolverContactConstraintPool,
+									  m_tmpSolverBodyPool,
+									  s_contactBatchingMethod,
+									  s_minBatchSize,
+									  s_maxBatchSize,
+									  &m_scratchMemory);
 }
 
-
 void btSequentialImpulseConstraintSolverMt::setupBatchedJointConstraints()
 {
-    BT_PROFILE("setupBatchedJointConstraints");
-    m_batchedJointConstraints.setup( &m_tmpSolverNonContactConstraintPool,
-        m_tmpSolverBodyPool,
-        s_jointBatchingMethod,
-        s_minBatchSize,
-        s_maxBatchSize,
-        &m_scratchMemory
-    );
+	BT_PROFILE("setupBatchedJointConstraints");
+	m_batchedJointConstraints.setup(&m_tmpSolverNonContactConstraintPool,
+									m_tmpSolverBodyPool,
+									s_jointBatchingMethod,
+									s_minBatchSize,
+									s_maxBatchSize,
+									&m_scratchMemory);
 }
 
-
 void btSequentialImpulseConstraintSolverMt::internalSetupContactConstraints(int iContactConstraint, const btContactSolverInfo& infoGlobal)
 {
-    btSolverConstraint& contactConstraint = m_tmpSolverContactConstraintPool[iContactConstraint];
+	btSolverConstraint& contactConstraint = m_tmpSolverContactConstraintPool[iContactConstraint];
 
-    btVector3 rel_pos1;
-    btVector3 rel_pos2;
-    btScalar relaxation;
+	btVector3 rel_pos1;
+	btVector3 rel_pos2;
+	btScalar relaxation;
 
-    int solverBodyIdA = contactConstraint.m_solverBodyIdA;
-    int solverBodyIdB = contactConstraint.m_solverBodyIdB;
+	int solverBodyIdA = contactConstraint.m_solverBodyIdA;
+	int solverBodyIdB = contactConstraint.m_solverBodyIdB;
 
-    btSolverBody* solverBodyA = &m_tmpSolverBodyPool[ solverBodyIdA ];
-    btSolverBody* solverBodyB = &m_tmpSolverBodyPool[ solverBodyIdB ];
+	btSolverBody* solverBodyA = &m_tmpSolverBodyPool[solverBodyIdA];
+	btSolverBody* solverBodyB = &m_tmpSolverBodyPool[solverBodyIdB];
 
-    btRigidBody* colObj0 = solverBodyA->m_originalBody;
-    btRigidBody* colObj1 = solverBodyB->m_originalBody;
+	btRigidBody* colObj0 = solverBodyA->m_originalBody;
+	btRigidBody* colObj1 = solverBodyB->m_originalBody;
 
-    btManifoldPoint& cp = *static_cast<btManifoldPoint*>( contactConstraint.m_originalContactPoint );
+	btManifoldPoint& cp = *static_cast<btManifoldPoint*>(contactConstraint.m_originalContactPoint);
 
-    const btVector3& pos1 = cp.getPositionWorldOnA();
-    const btVector3& pos2 = cp.getPositionWorldOnB();
+	const btVector3& pos1 = cp.getPositionWorldOnA();
+	const btVector3& pos2 = cp.getPositionWorldOnB();
 
-    rel_pos1 = pos1 - solverBodyA->getWorldTransform().getOrigin();
-    rel_pos2 = pos2 - solverBodyB->getWorldTransform().getOrigin();
+	rel_pos1 = pos1 - solverBodyA->getWorldTransform().getOrigin();
+	rel_pos2 = pos2 - solverBodyB->getWorldTransform().getOrigin();
 
-    btVector3 vel1;
-    btVector3 vel2;
+	btVector3 vel1;
+	btVector3 vel2;
 
-    solverBodyA->getVelocityInLocalPointNoDelta( rel_pos1, vel1 );
-    solverBodyB->getVelocityInLocalPointNoDelta( rel_pos2, vel2 );
+	solverBodyA->getVelocityInLocalPointNoDelta(rel_pos1, vel1);
+	solverBodyB->getVelocityInLocalPointNoDelta(rel_pos2, vel2);
 
-    btVector3 vel = vel1 - vel2;
-    btScalar rel_vel = cp.m_normalWorldOnB.dot( vel );
+	btVector3 vel = vel1 - vel2;
+	btScalar rel_vel = cp.m_normalWorldOnB.dot(vel);
 
-    setupContactConstraint( contactConstraint, solverBodyIdA, solverBodyIdB, cp, infoGlobal, relaxation, rel_pos1, rel_pos2 );
+	setupContactConstraint(contactConstraint, solverBodyIdA, solverBodyIdB, cp, infoGlobal, relaxation, rel_pos1, rel_pos2);
 
-    // setup rolling friction constraints
-    int rollingFrictionIndex = m_rollingFrictionIndexTable[iContactConstraint];
-    if (rollingFrictionIndex >= 0)
-    {
-        btSolverConstraint& spinningFrictionConstraint = m_tmpSolverContactRollingFrictionConstraintPool[ rollingFrictionIndex ];
-        btAssert( spinningFrictionConstraint.m_frictionIndex == iContactConstraint );
-        setupTorsionalFrictionConstraint( spinningFrictionConstraint,
-            cp.m_normalWorldOnB,
-            solverBodyIdA,
-            solverBodyIdB,
-            cp,
-            cp.m_combinedSpinningFriction,
-            rel_pos1,
-            rel_pos2,
-            colObj0,
-            colObj1,
-            relaxation,
-            0.0f,
-            0.0f
-        );
-        btVector3 axis[2];
-        btPlaneSpace1( cp.m_normalWorldOnB, axis[0], axis[1] );
-        axis[0].normalize();
-        axis[1].normalize();
+	// setup rolling friction constraints
+	int rollingFrictionIndex = m_rollingFrictionIndexTable[iContactConstraint];
+	if (rollingFrictionIndex >= 0)
+	{
+		btSolverConstraint& spinningFrictionConstraint = m_tmpSolverContactRollingFrictionConstraintPool[rollingFrictionIndex];
+		btAssert(spinningFrictionConstraint.m_frictionIndex == iContactConstraint);
+		setupTorsionalFrictionConstraint(spinningFrictionConstraint,
+										 cp.m_normalWorldOnB,
+										 solverBodyIdA,
+										 solverBodyIdB,
+										 cp,
+										 cp.m_combinedSpinningFriction,
+										 rel_pos1,
+										 rel_pos2,
+										 colObj0,
+										 colObj1,
+										 relaxation,
+										 0.0f,
+										 0.0f);
+		btVector3 axis[2];
+		btPlaneSpace1(cp.m_normalWorldOnB, axis[0], axis[1]);
+		axis[0].normalize();
+		axis[1].normalize();
 
-        applyAnisotropicFriction( colObj0, axis[0], btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION );
-        applyAnisotropicFriction( colObj1, axis[0], btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION );
-        applyAnisotropicFriction( colObj0, axis[1], btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION );
-        applyAnisotropicFriction( colObj1, axis[1], btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION );
-        // put the largest axis first
-        if (axis[1].length2() > axis[0].length2())
-        {
-            btSwap(axis[0], axis[1]);
-        }
-        const btScalar kRollingFrictionThreshold = 0.001f;
-        for (int i = 0; i < 2; ++i)
-        {
-            int iRollingFric = rollingFrictionIndex + 1 + i;
-            btSolverConstraint& rollingFrictionConstraint = m_tmpSolverContactRollingFrictionConstraintPool[ iRollingFric ];
-            btAssert(rollingFrictionConstraint.m_frictionIndex == iContactConstraint);
-            btVector3 dir = axis[i];
-            if ( dir.length() > kRollingFrictionThreshold )
-            {
-                setupTorsionalFrictionConstraint( rollingFrictionConstraint,
-                    dir,
-                    solverBodyIdA,
-                    solverBodyIdB,
-                    cp,
-                    cp.m_combinedRollingFriction,
-                    rel_pos1,
-                    rel_pos2,
-                    colObj0,
-                    colObj1,
-                    relaxation,
-                    0.0f,
-                    0.0f
-                );
-            }
-            else
-            {
-                rollingFrictionConstraint.m_frictionIndex = -1;  // disable constraint
-            }
-        }
-    }
+		applyAnisotropicFriction(colObj0, axis[0], btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION);
+		applyAnisotropicFriction(colObj1, axis[0], btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION);
+		applyAnisotropicFriction(colObj0, axis[1], btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION);
+		applyAnisotropicFriction(colObj1, axis[1], btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION);
+		// put the largest axis first
+		if (axis[1].length2() > axis[0].length2())
+		{
+			btSwap(axis[0], axis[1]);
+		}
+		const btScalar kRollingFrictionThreshold = 0.001f;
+		for (int i = 0; i < 2; ++i)
+		{
+			int iRollingFric = rollingFrictionIndex + 1 + i;
+			btSolverConstraint& rollingFrictionConstraint = m_tmpSolverContactRollingFrictionConstraintPool[iRollingFric];
+			btAssert(rollingFrictionConstraint.m_frictionIndex == iContactConstraint);
+			btVector3 dir = axis[i];
+			if (dir.length() > kRollingFrictionThreshold)
+			{
+				setupTorsionalFrictionConstraint(rollingFrictionConstraint,
+												 dir,
+												 solverBodyIdA,
+												 solverBodyIdB,
+												 cp,
+												 cp.m_combinedRollingFriction,
+												 rel_pos1,
+												 rel_pos2,
+												 colObj0,
+												 colObj1,
+												 relaxation,
+												 0.0f,
+												 0.0f);
+			}
+			else
+			{
+				rollingFrictionConstraint.m_frictionIndex = -1;  // disable constraint
+			}
+		}
+	}
 
-    // setup friction constraints
-    //	setupFrictionConstraint(solverConstraint, normalAxis, solverBodyIdA, solverBodyIdB, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, infoGlobal, desiredVelocity, cfmSlip);
-    {
-        ///Bullet has several options to set the friction directions
-        ///By default, each contact has only a single friction direction that is recomputed automatically very frame
-        ///based on the relative linear velocity.
-        ///If the relative velocity it zero, it will automatically compute a friction direction.
+	// setup friction constraints
+	//	setupFrictionConstraint(solverConstraint, normalAxis, solverBodyIdA, solverBodyIdB, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, infoGlobal, desiredVelocity, cfmSlip);
+	{
+		///Bullet has several options to set the friction directions
+		///By default, each contact has only a single friction direction that is recomputed automatically very frame
+		///based on the relative linear velocity.
+		///If the relative velocity it zero, it will automatically compute a friction direction.
 
-        ///You can also enable two friction directions, using the SOLVER_USE_2_FRICTION_DIRECTIONS.
-        ///In that case, the second friction direction will be orthogonal to both contact normal and first friction direction.
-        ///
-        ///If you choose SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION, then the friction will be independent from the relative projected velocity.
-        ///
-        ///The user can manually override the friction directions for certain contacts using a contact callback,
-        ///and set the cp.m_lateralFrictionInitialized to true
-        ///In that case, you can set the target relative motion in each friction direction (cp.m_contactMotion1 and cp.m_contactMotion2)
-        ///this will give a conveyor belt effect
-        ///
-	    btSolverConstraint* frictionConstraint1 = &m_tmpSolverContactFrictionConstraintPool[contactConstraint.m_frictionIndex];
-        btAssert(frictionConstraint1->m_frictionIndex == iContactConstraint);
+		///You can also enable two friction directions, using the SOLVER_USE_2_FRICTION_DIRECTIONS.
+		///In that case, the second friction direction will be orthogonal to both contact normal and first friction direction.
+		///
+		///If you choose SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION, then the friction will be independent from the relative projected velocity.
+		///
+		///The user can manually override the friction directions for certain contacts using a contact callback,
+		///and set the cp.m_lateralFrictionInitialized to true
+		///In that case, you can set the target relative motion in each friction direction (cp.m_contactMotion1 and cp.m_contactMotion2)
+		///this will give a conveyor belt effect
+		///
+		btSolverConstraint* frictionConstraint1 = &m_tmpSolverContactFrictionConstraintPool[contactConstraint.m_frictionIndex];
+		btAssert(frictionConstraint1->m_frictionIndex == iContactConstraint);
 
-        btSolverConstraint* frictionConstraint2 = NULL;
-        if ( infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS )
-        {
-            frictionConstraint2 = &m_tmpSolverContactFrictionConstraintPool[contactConstraint.m_frictionIndex + 1];
-            btAssert( frictionConstraint2->m_frictionIndex == iContactConstraint );
-        }
+		btSolverConstraint* frictionConstraint2 = NULL;
+		if (infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS)
+		{
+			frictionConstraint2 = &m_tmpSolverContactFrictionConstraintPool[contactConstraint.m_frictionIndex + 1];
+			btAssert(frictionConstraint2->m_frictionIndex == iContactConstraint);
+		}
 
-        if ( !( infoGlobal.m_solverMode & SOLVER_ENABLE_FRICTION_DIRECTION_CACHING ) || !( cp.m_contactPointFlags&BT_CONTACT_FLAG_LATERAL_FRICTION_INITIALIZED ) )
-        {
-            cp.m_lateralFrictionDir1 = vel - cp.m_normalWorldOnB * rel_vel;
-            btScalar lat_rel_vel = cp.m_lateralFrictionDir1.length2();
-            if ( !( infoGlobal.m_solverMode & SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION ) && lat_rel_vel > SIMD_EPSILON )
-            {
-                cp.m_lateralFrictionDir1 *= 1.f / btSqrt( lat_rel_vel );
-                applyAnisotropicFriction( colObj0, cp.m_lateralFrictionDir1, btCollisionObject::CF_ANISOTROPIC_FRICTION );
-                applyAnisotropicFriction( colObj1, cp.m_lateralFrictionDir1, btCollisionObject::CF_ANISOTROPIC_FRICTION );
-                setupFrictionConstraint( *frictionConstraint1, cp.m_lateralFrictionDir1, solverBodyIdA, solverBodyIdB, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, infoGlobal );
+		if (!(infoGlobal.m_solverMode & SOLVER_ENABLE_FRICTION_DIRECTION_CACHING) || !(cp.m_contactPointFlags & BT_CONTACT_FLAG_LATERAL_FRICTION_INITIALIZED))
+		{
+			cp.m_lateralFrictionDir1 = vel - cp.m_normalWorldOnB * rel_vel;
+			btScalar lat_rel_vel = cp.m_lateralFrictionDir1.length2();
+			if (!(infoGlobal.m_solverMode & SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION) && lat_rel_vel > SIMD_EPSILON)
+			{
+				cp.m_lateralFrictionDir1 *= 1.f / btSqrt(lat_rel_vel);
+				applyAnisotropicFriction(colObj0, cp.m_lateralFrictionDir1, btCollisionObject::CF_ANISOTROPIC_FRICTION);
+				applyAnisotropicFriction(colObj1, cp.m_lateralFrictionDir1, btCollisionObject::CF_ANISOTROPIC_FRICTION);
+				setupFrictionConstraint(*frictionConstraint1, cp.m_lateralFrictionDir1, solverBodyIdA, solverBodyIdB, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, infoGlobal);
 
-                if ( frictionConstraint2 )
-                {
-                    cp.m_lateralFrictionDir2 = cp.m_lateralFrictionDir1.cross( cp.m_normalWorldOnB );
-                    cp.m_lateralFrictionDir2.normalize();//??
-                    applyAnisotropicFriction( colObj0, cp.m_lateralFrictionDir2, btCollisionObject::CF_ANISOTROPIC_FRICTION );
-                    applyAnisotropicFriction( colObj1, cp.m_lateralFrictionDir2, btCollisionObject::CF_ANISOTROPIC_FRICTION );
-                    setupFrictionConstraint( *frictionConstraint2, cp.m_lateralFrictionDir2, solverBodyIdA, solverBodyIdB, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, infoGlobal );
-                }
-            }
-            else
-            {
-                btPlaneSpace1( cp.m_normalWorldOnB, cp.m_lateralFrictionDir1, cp.m_lateralFrictionDir2 );
+				if (frictionConstraint2)
+				{
+					cp.m_lateralFrictionDir2 = cp.m_lateralFrictionDir1.cross(cp.m_normalWorldOnB);
+					cp.m_lateralFrictionDir2.normalize();  //??
+					applyAnisotropicFriction(colObj0, cp.m_lateralFrictionDir2, btCollisionObject::CF_ANISOTROPIC_FRICTION);
+					applyAnisotropicFriction(colObj1, cp.m_lateralFrictionDir2, btCollisionObject::CF_ANISOTROPIC_FRICTION);
+					setupFrictionConstraint(*frictionConstraint2, cp.m_lateralFrictionDir2, solverBodyIdA, solverBodyIdB, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, infoGlobal);
+				}
+			}
+			else
+			{
+				btPlaneSpace1(cp.m_normalWorldOnB, cp.m_lateralFrictionDir1, cp.m_lateralFrictionDir2);
 
-                applyAnisotropicFriction( colObj0, cp.m_lateralFrictionDir1, btCollisionObject::CF_ANISOTROPIC_FRICTION );
-                applyAnisotropicFriction( colObj1, cp.m_lateralFrictionDir1, btCollisionObject::CF_ANISOTROPIC_FRICTION );
-                setupFrictionConstraint( *frictionConstraint1, cp.m_lateralFrictionDir1, solverBodyIdA, solverBodyIdB, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, infoGlobal );
+				applyAnisotropicFriction(colObj0, cp.m_lateralFrictionDir1, btCollisionObject::CF_ANISOTROPIC_FRICTION);
+				applyAnisotropicFriction(colObj1, cp.m_lateralFrictionDir1, btCollisionObject::CF_ANISOTROPIC_FRICTION);
+				setupFrictionConstraint(*frictionConstraint1, cp.m_lateralFrictionDir1, solverBodyIdA, solverBodyIdB, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, infoGlobal);
 
-                if ( frictionConstraint2 )
-                {
-                    applyAnisotropicFriction( colObj0, cp.m_lateralFrictionDir2, btCollisionObject::CF_ANISOTROPIC_FRICTION );
-                    applyAnisotropicFriction( colObj1, cp.m_lateralFrictionDir2, btCollisionObject::CF_ANISOTROPIC_FRICTION );
-                    setupFrictionConstraint( *frictionConstraint2, cp.m_lateralFrictionDir2, solverBodyIdA, solverBodyIdB, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, infoGlobal );
-                }
+				if (frictionConstraint2)
+				{
+					applyAnisotropicFriction(colObj0, cp.m_lateralFrictionDir2, btCollisionObject::CF_ANISOTROPIC_FRICTION);
+					applyAnisotropicFriction(colObj1, cp.m_lateralFrictionDir2, btCollisionObject::CF_ANISOTROPIC_FRICTION);
+					setupFrictionConstraint(*frictionConstraint2, cp.m_lateralFrictionDir2, solverBodyIdA, solverBodyIdB, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, infoGlobal);
+				}
 
-                if ( ( infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS ) && ( infoGlobal.m_solverMode & SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION ) )
-                {
-                    cp.m_contactPointFlags |= BT_CONTACT_FLAG_LATERAL_FRICTION_INITIALIZED;
-                }
-            }
-        }
-        else
-        {
-            setupFrictionConstraint( *frictionConstraint1, cp.m_lateralFrictionDir1, solverBodyIdA, solverBodyIdB, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, infoGlobal, cp.m_contactMotion1, cp.m_frictionCFM );
-            if ( frictionConstraint2 )
-            {
-                setupFrictionConstraint( *frictionConstraint2, cp.m_lateralFrictionDir2, solverBodyIdA, solverBodyIdB, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, infoGlobal, cp.m_contactMotion2, cp.m_frictionCFM );
-            }
-        }
-    }
+				if ((infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS) && (infoGlobal.m_solverMode & SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION))
+				{
+					cp.m_contactPointFlags |= BT_CONTACT_FLAG_LATERAL_FRICTION_INITIALIZED;
+				}
+			}
+		}
+		else
+		{
+			setupFrictionConstraint(*frictionConstraint1, cp.m_lateralFrictionDir1, solverBodyIdA, solverBodyIdB, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, infoGlobal, cp.m_contactMotion1, cp.m_frictionCFM);
+			if (frictionConstraint2)
+			{
+				setupFrictionConstraint(*frictionConstraint2, cp.m_lateralFrictionDir2, solverBodyIdA, solverBodyIdB, cp, rel_pos1, rel_pos2, colObj0, colObj1, relaxation, infoGlobal, cp.m_contactMotion2, cp.m_frictionCFM);
+			}
+		}
+	}
 
-    setFrictionConstraintImpulse( contactConstraint, solverBodyIdA, solverBodyIdB, cp, infoGlobal );
+	setFrictionConstraintImpulse(contactConstraint, solverBodyIdA, solverBodyIdB, cp, infoGlobal);
 }
 
-
 struct SetupContactConstraintsLoop : public btIParallelForBody
 {
-    btSequentialImpulseConstraintSolverMt* m_solver;
-    const btBatchedConstraints* m_bc;
-    const btContactSolverInfo* m_infoGlobal;
+	btSequentialImpulseConstraintSolverMt* m_solver;
+	const btBatchedConstraints* m_bc;
+	const btContactSolverInfo* m_infoGlobal;
 
-    SetupContactConstraintsLoop( btSequentialImpulseConstraintSolverMt* solver, const btBatchedConstraints* bc, const btContactSolverInfo& infoGlobal )
-    {
-        m_solver = solver;
-        m_bc = bc;
-        m_infoGlobal = &infoGlobal;
-    }
-    void forLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        BT_PROFILE( "SetupContactConstraintsLoop" );
-        for ( int iBatch = iBegin; iBatch < iEnd; ++iBatch )
-        {
-            const btBatchedConstraints::Range& batch = m_bc->m_batches[ iBatch ];
-            for (int i = batch.begin; i < batch.end; ++i)
-            {
-                int iContact = m_bc->m_constraintIndices[i];
-                m_solver->internalSetupContactConstraints( iContact, *m_infoGlobal );
-            }
-        }
-    }
+	SetupContactConstraintsLoop(btSequentialImpulseConstraintSolverMt* solver, const btBatchedConstraints* bc, const btContactSolverInfo& infoGlobal)
+	{
+		m_solver = solver;
+		m_bc = bc;
+		m_infoGlobal = &infoGlobal;
+	}
+	void forLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		BT_PROFILE("SetupContactConstraintsLoop");
+		for (int iBatch = iBegin; iBatch < iEnd; ++iBatch)
+		{
+			const btBatchedConstraints::Range& batch = m_bc->m_batches[iBatch];
+			for (int i = batch.begin; i < batch.end; ++i)
+			{
+				int iContact = m_bc->m_constraintIndices[i];
+				m_solver->internalSetupContactConstraints(iContact, *m_infoGlobal);
+			}
+		}
+	}
 };
 
-
 void btSequentialImpulseConstraintSolverMt::setupAllContactConstraints(const btContactSolverInfo& infoGlobal)
 {
-    BT_PROFILE( "setupAllContactConstraints" );
-    if ( m_useBatching )
-    {
-        const btBatchedConstraints& batchedCons = m_batchedContactConstraints;
-        SetupContactConstraintsLoop loop( this, &batchedCons, infoGlobal );
-        for ( int iiPhase = 0; iiPhase < batchedCons.m_phases.size(); ++iiPhase )
-        {
-            int iPhase = batchedCons.m_phaseOrder[ iiPhase ];
-            const btBatchedConstraints::Range& phase = batchedCons.m_phases[ iPhase ];
-            int grainSize = 1;
-            btParallelFor( phase.begin, phase.end, grainSize, loop );
-        }
-    }
-    else
-    {
-        for ( int i = 0; i < m_tmpSolverContactConstraintPool.size(); ++i )
-        {
-            internalSetupContactConstraints( i, infoGlobal );
-        }
-    }
+	BT_PROFILE("setupAllContactConstraints");
+	if (m_useBatching)
+	{
+		const btBatchedConstraints& batchedCons = m_batchedContactConstraints;
+		SetupContactConstraintsLoop loop(this, &batchedCons, infoGlobal);
+		for (int iiPhase = 0; iiPhase < batchedCons.m_phases.size(); ++iiPhase)
+		{
+			int iPhase = batchedCons.m_phaseOrder[iiPhase];
+			const btBatchedConstraints::Range& phase = batchedCons.m_phases[iPhase];
+			int grainSize = 1;
+			btParallelFor(phase.begin, phase.end, grainSize, loop);
+		}
+	}
+	else
+	{
+		for (int i = 0; i < m_tmpSolverContactConstraintPool.size(); ++i)
+		{
+			internalSetupContactConstraints(i, infoGlobal);
+		}
+	}
 }
 
-
-int	btSequentialImpulseConstraintSolverMt::getOrInitSolverBodyThreadsafe(btCollisionObject& body,btScalar timeStep)
+int btSequentialImpulseConstraintSolverMt::getOrInitSolverBodyThreadsafe(btCollisionObject& body, btScalar timeStep)
 {
-    //
-    // getOrInitSolverBody is threadsafe only for a single thread per solver (with potentially multiple solvers)
-    //
-    // getOrInitSolverBodyThreadsafe -- attempts to be fully threadsafe (however may affect determinism)
-    //
-    int solverBodyId = -1;
-    bool isRigidBodyType = btRigidBody::upcast( &body ) != NULL;
-    if ( isRigidBodyType && !body.isStaticOrKinematicObject() )
-    {
-        // dynamic body
-        // Dynamic bodies can only be in one island, so it's safe to write to the companionId
-        solverBodyId = body.getCompanionId();
-        if ( solverBodyId < 0 )
-        {
-            m_bodySolverArrayMutex.lock();
-            // now that we have the lock, check again
-            solverBodyId = body.getCompanionId();
-            if ( solverBodyId < 0 )
-            {
-                solverBodyId = m_tmpSolverBodyPool.size();
-                btSolverBody& solverBody = m_tmpSolverBodyPool.expand();
-                initSolverBody( &solverBody, &body, timeStep );
-                body.setCompanionId( solverBodyId );
-            }
-            m_bodySolverArrayMutex.unlock();
-        }
-    }
-    else if (isRigidBodyType && body.isKinematicObject())
-    {
-        //
-        // NOTE: must test for kinematic before static because some kinematic objects also
-        //   identify as "static"
-        //
-        // Kinematic bodies can be in multiple islands at once, so it is a
-        // race condition to write to them, so we use an alternate method
-        // to record the solverBodyId
-        int uniqueId = body.getWorldArrayIndex();
-        const int INVALID_SOLVER_BODY_ID = -1;
-        if (m_kinematicBodyUniqueIdToSolverBodyTable.size() <= uniqueId )
-        {
-            m_kinematicBodyUniqueIdToSolverBodyTableMutex.lock();
-            // now that we have the lock, check again
-            if ( m_kinematicBodyUniqueIdToSolverBodyTable.size() <= uniqueId )
-            {
-                m_kinematicBodyUniqueIdToSolverBodyTable.resize( uniqueId + 1, INVALID_SOLVER_BODY_ID );
-            }
-            m_kinematicBodyUniqueIdToSolverBodyTableMutex.unlock();
-        }
-        solverBodyId = m_kinematicBodyUniqueIdToSolverBodyTable[ uniqueId ];
-        // if no table entry yet,
-        if ( INVALID_SOLVER_BODY_ID == solverBodyId )
-        {
-            // need to acquire both locks
-            m_kinematicBodyUniqueIdToSolverBodyTableMutex.lock();
-            m_bodySolverArrayMutex.lock();
-            // now that we have the lock, check again
-            solverBodyId = m_kinematicBodyUniqueIdToSolverBodyTable[ uniqueId ];
-            if ( INVALID_SOLVER_BODY_ID == solverBodyId )
-            {
-                // create a table entry for this body
-                solverBodyId = m_tmpSolverBodyPool.size();
-                btSolverBody& solverBody = m_tmpSolverBodyPool.expand();
-                initSolverBody( &solverBody, &body, timeStep );
-                m_kinematicBodyUniqueIdToSolverBodyTable[ uniqueId ] = solverBodyId;
-            }
-            m_bodySolverArrayMutex.unlock();
-            m_kinematicBodyUniqueIdToSolverBodyTableMutex.unlock();
-        }
-    }
-    else
-    {
-        // all fixed bodies (inf mass) get mapped to a single solver id
-        if ( m_fixedBodyId < 0 )
-        {
-            m_bodySolverArrayMutex.lock();
-            // now that we have the lock, check again
-            if ( m_fixedBodyId < 0 )
-            {
-                m_fixedBodyId = m_tmpSolverBodyPool.size();
-                btSolverBody& fixedBody = m_tmpSolverBodyPool.expand();
-                initSolverBody( &fixedBody, 0, timeStep );
-            }
-            m_bodySolverArrayMutex.unlock();
-        }
-        solverBodyId = m_fixedBodyId;
-    }
-    btAssert( solverBodyId >= 0 && solverBodyId < m_tmpSolverBodyPool.size() );
+	//
+	// getOrInitSolverBody is threadsafe only for a single thread per solver (with potentially multiple solvers)
+	//
+	// getOrInitSolverBodyThreadsafe -- attempts to be fully threadsafe (however may affect determinism)
+	//
+	int solverBodyId = -1;
+	bool isRigidBodyType = btRigidBody::upcast(&body) != NULL;
+	if (isRigidBodyType && !body.isStaticOrKinematicObject())
+	{
+		// dynamic body
+		// Dynamic bodies can only be in one island, so it's safe to write to the companionId
+		solverBodyId = body.getCompanionId();
+		if (solverBodyId < 0)
+		{
+			m_bodySolverArrayMutex.lock();
+			// now that we have the lock, check again
+			solverBodyId = body.getCompanionId();
+			if (solverBodyId < 0)
+			{
+				solverBodyId = m_tmpSolverBodyPool.size();
+				btSolverBody& solverBody = m_tmpSolverBodyPool.expand();
+				initSolverBody(&solverBody, &body, timeStep);
+				body.setCompanionId(solverBodyId);
+			}
+			m_bodySolverArrayMutex.unlock();
+		}
+	}
+	else if (isRigidBodyType && body.isKinematicObject())
+	{
+		//
+		// NOTE: must test for kinematic before static because some kinematic objects also
+		//   identify as "static"
+		//
+		// Kinematic bodies can be in multiple islands at once, so it is a
+		// race condition to write to them, so we use an alternate method
+		// to record the solverBodyId
+		int uniqueId = body.getWorldArrayIndex();
+		const int INVALID_SOLVER_BODY_ID = -1;
+		if (m_kinematicBodyUniqueIdToSolverBodyTable.size() <= uniqueId)
+		{
+			m_kinematicBodyUniqueIdToSolverBodyTableMutex.lock();
+			// now that we have the lock, check again
+			if (m_kinematicBodyUniqueIdToSolverBodyTable.size() <= uniqueId)
+			{
+				m_kinematicBodyUniqueIdToSolverBodyTable.resize(uniqueId + 1, INVALID_SOLVER_BODY_ID);
+			}
+			m_kinematicBodyUniqueIdToSolverBodyTableMutex.unlock();
+		}
+		solverBodyId = m_kinematicBodyUniqueIdToSolverBodyTable[uniqueId];
+		// if no table entry yet,
+		if (INVALID_SOLVER_BODY_ID == solverBodyId)
+		{
+			// need to acquire both locks
+			m_kinematicBodyUniqueIdToSolverBodyTableMutex.lock();
+			m_bodySolverArrayMutex.lock();
+			// now that we have the lock, check again
+			solverBodyId = m_kinematicBodyUniqueIdToSolverBodyTable[uniqueId];
+			if (INVALID_SOLVER_BODY_ID == solverBodyId)
+			{
+				// create a table entry for this body
+				solverBodyId = m_tmpSolverBodyPool.size();
+				btSolverBody& solverBody = m_tmpSolverBodyPool.expand();
+				initSolverBody(&solverBody, &body, timeStep);
+				m_kinematicBodyUniqueIdToSolverBodyTable[uniqueId] = solverBodyId;
+			}
+			m_bodySolverArrayMutex.unlock();
+			m_kinematicBodyUniqueIdToSolverBodyTableMutex.unlock();
+		}
+	}
+	else
+	{
+		// all fixed bodies (inf mass) get mapped to a single solver id
+		if (m_fixedBodyId < 0)
+		{
+			m_bodySolverArrayMutex.lock();
+			// now that we have the lock, check again
+			if (m_fixedBodyId < 0)
+			{
+				m_fixedBodyId = m_tmpSolverBodyPool.size();
+				btSolverBody& fixedBody = m_tmpSolverBodyPool.expand();
+				initSolverBody(&fixedBody, 0, timeStep);
+			}
+			m_bodySolverArrayMutex.unlock();
+		}
+		solverBodyId = m_fixedBodyId;
+	}
+	btAssert(solverBodyId >= 0 && solverBodyId < m_tmpSolverBodyPool.size());
 	return solverBodyId;
 }
 
-
 void btSequentialImpulseConstraintSolverMt::internalCollectContactManifoldCachedInfo(btContactManifoldCachedInfo* cachedInfoArray, btPersistentManifold** manifoldPtr, int numManifolds, const btContactSolverInfo& infoGlobal)
 {
-    BT_PROFILE("internalCollectContactManifoldCachedInfo");
-    for (int i = 0; i < numManifolds; ++i)
-    {
-        btContactManifoldCachedInfo* cachedInfo = &cachedInfoArray[i];
-        btPersistentManifold* manifold = manifoldPtr[i];
-        btCollisionObject* colObj0 = (btCollisionObject*) manifold->getBody0();
-        btCollisionObject* colObj1 = (btCollisionObject*) manifold->getBody1();
+	BT_PROFILE("internalCollectContactManifoldCachedInfo");
+	for (int i = 0; i < numManifolds; ++i)
+	{
+		btContactManifoldCachedInfo* cachedInfo = &cachedInfoArray[i];
+		btPersistentManifold* manifold = manifoldPtr[i];
+		btCollisionObject* colObj0 = (btCollisionObject*)manifold->getBody0();
+		btCollisionObject* colObj1 = (btCollisionObject*)manifold->getBody1();
 
-        int solverBodyIdA = getOrInitSolverBodyThreadsafe( *colObj0, infoGlobal.m_timeStep );
-        int solverBodyIdB = getOrInitSolverBodyThreadsafe( *colObj1, infoGlobal.m_timeStep );
+		int solverBodyIdA = getOrInitSolverBodyThreadsafe(*colObj0, infoGlobal.m_timeStep);
+		int solverBodyIdB = getOrInitSolverBodyThreadsafe(*colObj1, infoGlobal.m_timeStep);
 
-        cachedInfo->solverBodyIds[ 0 ] = solverBodyIdA;
-        cachedInfo->solverBodyIds[ 1 ] = solverBodyIdB;
-        cachedInfo->numTouchingContacts = 0;
+		cachedInfo->solverBodyIds[0] = solverBodyIdA;
+		cachedInfo->solverBodyIds[1] = solverBodyIdB;
+		cachedInfo->numTouchingContacts = 0;
 
-        btSolverBody* solverBodyA = &m_tmpSolverBodyPool[ solverBodyIdA ];
-        btSolverBody* solverBodyB = &m_tmpSolverBodyPool[ solverBodyIdB ];
+		btSolverBody* solverBodyA = &m_tmpSolverBodyPool[solverBodyIdA];
+		btSolverBody* solverBodyB = &m_tmpSolverBodyPool[solverBodyIdB];
 
-        // A contact manifold between 2 static object should not exist!
-        // check the collision flags of your objects if this assert fires.
-        // Incorrectly set collision object flags can degrade performance in various ways.
-        btAssert( !m_tmpSolverBodyPool[ solverBodyIdA ].m_invMass.isZero() || !m_tmpSolverBodyPool[ solverBodyIdB ].m_invMass.isZero() );
+		// A contact manifold between 2 static object should not exist!
+		// check the collision flags of your objects if this assert fires.
+		// Incorrectly set collision object flags can degrade performance in various ways.
+		btAssert(!m_tmpSolverBodyPool[solverBodyIdA].m_invMass.isZero() || !m_tmpSolverBodyPool[solverBodyIdB].m_invMass.isZero());
 
-        int iContact = 0;
-        for ( int j = 0; j < manifold->getNumContacts(); j++ )
-        {
-            btManifoldPoint& cp = manifold->getContactPoint( j );
+		int iContact = 0;
+		for (int j = 0; j < manifold->getNumContacts(); j++)
+		{
+			btManifoldPoint& cp = manifold->getContactPoint(j);
 
-            if ( cp.getDistance() <= manifold->getContactProcessingThreshold() )
-            {
-                cachedInfo->contactPoints[ iContact ] = &cp;
-                cachedInfo->contactHasRollingFriction[ iContact ] = ( cp.m_combinedRollingFriction > 0.f );
-                iContact++;
-            }
-        }
-        cachedInfo->numTouchingContacts = iContact;
-    }
+			if (cp.getDistance() <= manifold->getContactProcessingThreshold())
+			{
+				cachedInfo->contactPoints[iContact] = &cp;
+				cachedInfo->contactHasRollingFriction[iContact] = (cp.m_combinedRollingFriction > 0.f);
+				iContact++;
+			}
+		}
+		cachedInfo->numTouchingContacts = iContact;
+	}
 }
 
-
 struct CollectContactManifoldCachedInfoLoop : public btIParallelForBody
 {
-    btSequentialImpulseConstraintSolverMt* m_solver;
-    btSequentialImpulseConstraintSolverMt::btContactManifoldCachedInfo* m_cachedInfoArray;
-    btPersistentManifold** m_manifoldPtr;
-    const btContactSolverInfo* m_infoGlobal;
+	btSequentialImpulseConstraintSolverMt* m_solver;
+	btSequentialImpulseConstraintSolverMt::btContactManifoldCachedInfo* m_cachedInfoArray;
+	btPersistentManifold** m_manifoldPtr;
+	const btContactSolverInfo* m_infoGlobal;
 
-    CollectContactManifoldCachedInfoLoop( btSequentialImpulseConstraintSolverMt* solver, btSequentialImpulseConstraintSolverMt::btContactManifoldCachedInfo* cachedInfoArray, btPersistentManifold** manifoldPtr, const btContactSolverInfo& infoGlobal )
-    {
-        m_solver = solver;
-        m_cachedInfoArray = cachedInfoArray;
-        m_manifoldPtr = manifoldPtr;
-        m_infoGlobal = &infoGlobal;
-    }
-    void forLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        m_solver->internalCollectContactManifoldCachedInfo( m_cachedInfoArray + iBegin, m_manifoldPtr + iBegin, iEnd - iBegin, *m_infoGlobal );
-    }
+	CollectContactManifoldCachedInfoLoop(btSequentialImpulseConstraintSolverMt* solver, btSequentialImpulseConstraintSolverMt::btContactManifoldCachedInfo* cachedInfoArray, btPersistentManifold** manifoldPtr, const btContactSolverInfo& infoGlobal)
+	{
+		m_solver = solver;
+		m_cachedInfoArray = cachedInfoArray;
+		m_manifoldPtr = manifoldPtr;
+		m_infoGlobal = &infoGlobal;
+	}
+	void forLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		m_solver->internalCollectContactManifoldCachedInfo(m_cachedInfoArray + iBegin, m_manifoldPtr + iBegin, iEnd - iBegin, *m_infoGlobal);
+	}
 };
 
-
 void btSequentialImpulseConstraintSolverMt::internalAllocContactConstraints(const btContactManifoldCachedInfo* cachedInfoArray, int numManifolds)
 {
-    BT_PROFILE("internalAllocContactConstraints");
-    // possibly parallel part
-    for ( int iManifold = 0; iManifold < numManifolds; ++iManifold )
-    {
-        const btContactManifoldCachedInfo& cachedInfo = cachedInfoArray[ iManifold ];
-        int contactIndex = cachedInfo.contactIndex;
-        int frictionIndex = contactIndex * m_numFrictionDirections;
-        int rollingFrictionIndex = cachedInfo.rollingFrictionIndex;
-        for ( int i = 0; i < cachedInfo.numTouchingContacts; i++ )
-        {
-            btSolverConstraint& contactConstraint = m_tmpSolverContactConstraintPool[contactIndex];
-            contactConstraint.m_solverBodyIdA = cachedInfo.solverBodyIds[ 0 ];
-            contactConstraint.m_solverBodyIdB = cachedInfo.solverBodyIds[ 1 ];
-            contactConstraint.m_originalContactPoint = cachedInfo.contactPoints[ i ];
+	BT_PROFILE("internalAllocContactConstraints");
+	// possibly parallel part
+	for (int iManifold = 0; iManifold < numManifolds; ++iManifold)
+	{
+		const btContactManifoldCachedInfo& cachedInfo = cachedInfoArray[iManifold];
+		int contactIndex = cachedInfo.contactIndex;
+		int frictionIndex = contactIndex * m_numFrictionDirections;
+		int rollingFrictionIndex = cachedInfo.rollingFrictionIndex;
+		for (int i = 0; i < cachedInfo.numTouchingContacts; i++)
+		{
+			btSolverConstraint& contactConstraint = m_tmpSolverContactConstraintPool[contactIndex];
+			contactConstraint.m_solverBodyIdA = cachedInfo.solverBodyIds[0];
+			contactConstraint.m_solverBodyIdB = cachedInfo.solverBodyIds[1];
+			contactConstraint.m_originalContactPoint = cachedInfo.contactPoints[i];
 
-            // allocate the friction constraints
-            contactConstraint.m_frictionIndex = frictionIndex;
-            for ( int iDir = 0; iDir < m_numFrictionDirections; ++iDir )
-            {
-                btSolverConstraint& frictionConstraint = m_tmpSolverContactFrictionConstraintPool[frictionIndex];
-                frictionConstraint.m_frictionIndex = contactIndex;
-                frictionIndex++;
-            }
+			// allocate the friction constraints
+			contactConstraint.m_frictionIndex = frictionIndex;
+			for (int iDir = 0; iDir < m_numFrictionDirections; ++iDir)
+			{
+				btSolverConstraint& frictionConstraint = m_tmpSolverContactFrictionConstraintPool[frictionIndex];
+				frictionConstraint.m_frictionIndex = contactIndex;
+				frictionIndex++;
+			}
 
-            // allocate rolling friction constraints
-            if ( cachedInfo.contactHasRollingFriction[ i ] )
-            {
-                m_rollingFrictionIndexTable[ contactIndex ] = rollingFrictionIndex;
-                // allocate 3 (although we may use only 2 sometimes)
-                for ( int i = 0; i < 3; i++ )
-                {
-                    m_tmpSolverContactRollingFrictionConstraintPool[ rollingFrictionIndex ].m_frictionIndex = contactIndex;
-                    rollingFrictionIndex++;
-                }
-            }
-            else
-            {
-                // indicate there is no rolling friction for this contact point
-                m_rollingFrictionIndexTable[ contactIndex ] = -1;
-            }
-            contactIndex++;
-        }
-    }
+			// allocate rolling friction constraints
+			if (cachedInfo.contactHasRollingFriction[i])
+			{
+				m_rollingFrictionIndexTable[contactIndex] = rollingFrictionIndex;
+				// allocate 3 (although we may use only 2 sometimes)
+				for (int i = 0; i < 3; i++)
+				{
+					m_tmpSolverContactRollingFrictionConstraintPool[rollingFrictionIndex].m_frictionIndex = contactIndex;
+					rollingFrictionIndex++;
+				}
+			}
+			else
+			{
+				// indicate there is no rolling friction for this contact point
+				m_rollingFrictionIndexTable[contactIndex] = -1;
+			}
+			contactIndex++;
+		}
+	}
 }
 
-
 struct AllocContactConstraintsLoop : public btIParallelForBody
 {
-    btSequentialImpulseConstraintSolverMt* m_solver;
-    const btSequentialImpulseConstraintSolverMt::btContactManifoldCachedInfo* m_cachedInfoArray;
+	btSequentialImpulseConstraintSolverMt* m_solver;
+	const btSequentialImpulseConstraintSolverMt::btContactManifoldCachedInfo* m_cachedInfoArray;
 
-    AllocContactConstraintsLoop( btSequentialImpulseConstraintSolverMt* solver, btSequentialImpulseConstraintSolverMt::btContactManifoldCachedInfo* cachedInfoArray )
-    {
-        m_solver = solver;
-        m_cachedInfoArray = cachedInfoArray;
-    }
-    void forLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        m_solver->internalAllocContactConstraints( m_cachedInfoArray + iBegin, iEnd - iBegin );
-    }
+	AllocContactConstraintsLoop(btSequentialImpulseConstraintSolverMt* solver, btSequentialImpulseConstraintSolverMt::btContactManifoldCachedInfo* cachedInfoArray)
+	{
+		m_solver = solver;
+		m_cachedInfoArray = cachedInfoArray;
+	}
+	void forLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		m_solver->internalAllocContactConstraints(m_cachedInfoArray + iBegin, iEnd - iBegin);
+	}
 };
 
-
 void btSequentialImpulseConstraintSolverMt::allocAllContactConstraints(btPersistentManifold** manifoldPtr, int numManifolds, const btContactSolverInfo& infoGlobal)
 {
-    BT_PROFILE( "allocAllContactConstraints" );
-    btAlignedObjectArray<btContactManifoldCachedInfo> cachedInfoArray; // = m_manifoldCachedInfoArray;
-    cachedInfoArray.resizeNoInitialize( numManifolds );
-    if (/* DISABLES CODE */ (false))
-    {
-        // sequential
-        internalCollectContactManifoldCachedInfo(&cachedInfoArray[ 0 ], manifoldPtr, numManifolds, infoGlobal);
-    }
-    else
-    {
-        // may alter ordering of bodies which affects determinism
-        CollectContactManifoldCachedInfoLoop loop( this, &cachedInfoArray[ 0 ], manifoldPtr, infoGlobal );
-        int grainSize = 200;
-        btParallelFor( 0, numManifolds, grainSize, loop );
-    }
+	BT_PROFILE("allocAllContactConstraints");
+	btAlignedObjectArray<btContactManifoldCachedInfo> cachedInfoArray;  // = m_manifoldCachedInfoArray;
+	cachedInfoArray.resizeNoInitialize(numManifolds);
+	if (/* DISABLES CODE */ (false))
+	{
+		// sequential
+		internalCollectContactManifoldCachedInfo(&cachedInfoArray[0], manifoldPtr, numManifolds, infoGlobal);
+	}
+	else
+	{
+		// may alter ordering of bodies which affects determinism
+		CollectContactManifoldCachedInfoLoop loop(this, &cachedInfoArray[0], manifoldPtr, infoGlobal);
+		int grainSize = 200;
+		btParallelFor(0, numManifolds, grainSize, loop);
+	}
 
-    {
-        // serial part
-        int numContacts = 0;
-        int numRollingFrictionConstraints = 0;
-        for ( int iManifold = 0; iManifold < numManifolds; ++iManifold )
-        {
-            btContactManifoldCachedInfo& cachedInfo = cachedInfoArray[ iManifold ];
-            cachedInfo.contactIndex = numContacts;
-            cachedInfo.rollingFrictionIndex = numRollingFrictionConstraints;
-            numContacts += cachedInfo.numTouchingContacts;
-            for (int i = 0; i < cachedInfo.numTouchingContacts; ++i)
-            {
-                if (cachedInfo.contactHasRollingFriction[i])
-                {
-                    numRollingFrictionConstraints += 3;
-                }
-            }
-        }
-        {
-            BT_PROFILE( "allocPools" );
-            if ( m_tmpSolverContactConstraintPool.capacity() < numContacts )
-            {
-                // if we need to reallocate, reserve some extra so we don't have to reallocate again next frame
-                int extraReserve = numContacts / 16;
-                m_tmpSolverContactConstraintPool.reserve( numContacts + extraReserve );
-                m_rollingFrictionIndexTable.reserve( numContacts + extraReserve );
-                m_tmpSolverContactFrictionConstraintPool.reserve( ( numContacts + extraReserve )*m_numFrictionDirections );
-                m_tmpSolverContactRollingFrictionConstraintPool.reserve( numRollingFrictionConstraints + extraReserve );
-            }
-            m_tmpSolverContactConstraintPool.resizeNoInitialize( numContacts );
-            m_rollingFrictionIndexTable.resizeNoInitialize( numContacts );
-            m_tmpSolverContactFrictionConstraintPool.resizeNoInitialize( numContacts*m_numFrictionDirections );
-            m_tmpSolverContactRollingFrictionConstraintPool.resizeNoInitialize( numRollingFrictionConstraints );
-        }
-    }
-    {
-        AllocContactConstraintsLoop loop(this, &cachedInfoArray[0]);
-        int grainSize = 200;
-        btParallelFor( 0, numManifolds, grainSize, loop );
-    }
+	{
+		// serial part
+		int numContacts = 0;
+		int numRollingFrictionConstraints = 0;
+		for (int iManifold = 0; iManifold < numManifolds; ++iManifold)
+		{
+			btContactManifoldCachedInfo& cachedInfo = cachedInfoArray[iManifold];
+			cachedInfo.contactIndex = numContacts;
+			cachedInfo.rollingFrictionIndex = numRollingFrictionConstraints;
+			numContacts += cachedInfo.numTouchingContacts;
+			for (int i = 0; i < cachedInfo.numTouchingContacts; ++i)
+			{
+				if (cachedInfo.contactHasRollingFriction[i])
+				{
+					numRollingFrictionConstraints += 3;
+				}
+			}
+		}
+		{
+			BT_PROFILE("allocPools");
+			if (m_tmpSolverContactConstraintPool.capacity() < numContacts)
+			{
+				// if we need to reallocate, reserve some extra so we don't have to reallocate again next frame
+				int extraReserve = numContacts / 16;
+				m_tmpSolverContactConstraintPool.reserve(numContacts + extraReserve);
+				m_rollingFrictionIndexTable.reserve(numContacts + extraReserve);
+				m_tmpSolverContactFrictionConstraintPool.reserve((numContacts + extraReserve) * m_numFrictionDirections);
+				m_tmpSolverContactRollingFrictionConstraintPool.reserve(numRollingFrictionConstraints + extraReserve);
+			}
+			m_tmpSolverContactConstraintPool.resizeNoInitialize(numContacts);
+			m_rollingFrictionIndexTable.resizeNoInitialize(numContacts);
+			m_tmpSolverContactFrictionConstraintPool.resizeNoInitialize(numContacts * m_numFrictionDirections);
+			m_tmpSolverContactRollingFrictionConstraintPool.resizeNoInitialize(numRollingFrictionConstraints);
+		}
+	}
+	{
+		AllocContactConstraintsLoop loop(this, &cachedInfoArray[0]);
+		int grainSize = 200;
+		btParallelFor(0, numManifolds, grainSize, loop);
+	}
 }
 
-
 void btSequentialImpulseConstraintSolverMt::convertContacts(btPersistentManifold** manifoldPtr, int numManifolds, const btContactSolverInfo& infoGlobal)
 {
-    if (!m_useBatching)
-    {
-        btSequentialImpulseConstraintSolver::convertContacts(manifoldPtr, numManifolds, infoGlobal);
-        return;
-    }
-    BT_PROFILE( "convertContacts" );
-    if (numManifolds > 0)
-    {
-        if ( m_fixedBodyId < 0 )
-        {
-            m_fixedBodyId = m_tmpSolverBodyPool.size();
-            btSolverBody& fixedBody = m_tmpSolverBodyPool.expand();
-            initSolverBody( &fixedBody, 0, infoGlobal.m_timeStep );
-        }
-        allocAllContactConstraints( manifoldPtr, numManifolds, infoGlobal );
-        if ( m_useBatching )
-        {
-            setupBatchedContactConstraints();
-        }
-        setupAllContactConstraints( infoGlobal );
-    }
+	if (!m_useBatching)
+	{
+		btSequentialImpulseConstraintSolver::convertContacts(manifoldPtr, numManifolds, infoGlobal);
+		return;
+	}
+	BT_PROFILE("convertContacts");
+	if (numManifolds > 0)
+	{
+		if (m_fixedBodyId < 0)
+		{
+			m_fixedBodyId = m_tmpSolverBodyPool.size();
+			btSolverBody& fixedBody = m_tmpSolverBodyPool.expand();
+			initSolverBody(&fixedBody, 0, infoGlobal.m_timeStep);
+		}
+		allocAllContactConstraints(manifoldPtr, numManifolds, infoGlobal);
+		if (m_useBatching)
+		{
+			setupBatchedContactConstraints();
+		}
+		setupAllContactConstraints(infoGlobal);
+	}
 }
 
-
-void btSequentialImpulseConstraintSolverMt::internalInitMultipleJoints( btTypedConstraint** constraints, int iBegin, int iEnd )
+void btSequentialImpulseConstraintSolverMt::internalInitMultipleJoints(btTypedConstraint** constraints, int iBegin, int iEnd)
 {
-    BT_PROFILE("internalInitMultipleJoints");
-    for ( int i = iBegin; i < iEnd; i++ )
+	BT_PROFILE("internalInitMultipleJoints");
+	for (int i = iBegin; i < iEnd; i++)
 	{
 		btTypedConstraint* constraint = constraints[i];
 		btTypedConstraint::btConstraintInfo1& info1 = m_tmpConstraintSizesPool[i];
 		if (constraint->isEnabled())
-        {
-            constraint->buildJacobian();
-            constraint->internalSetAppliedImpulse( 0.0f );
-            btJointFeedback* fb = constraint->getJointFeedback();
-            if ( fb )
-            {
-                fb->m_appliedForceBodyA.setZero();
-                fb->m_appliedTorqueBodyA.setZero();
-                fb->m_appliedForceBodyB.setZero();
-                fb->m_appliedTorqueBodyB.setZero();
-            }
-            constraint->getInfo1( &info1 );
-        }
-        else
+		{
+			constraint->buildJacobian();
+			constraint->internalSetAppliedImpulse(0.0f);
+			btJointFeedback* fb = constraint->getJointFeedback();
+			if (fb)
+			{
+				fb->m_appliedForceBodyA.setZero();
+				fb->m_appliedTorqueBodyA.setZero();
+				fb->m_appliedForceBodyB.setZero();
+				fb->m_appliedTorqueBodyB.setZero();
+			}
+			constraint->getInfo1(&info1);
+		}
+		else
 		{
 			info1.m_numConstraintRows = 0;
 			info1.nub = 0;
@@ -646,158 +624,151 @@ void btSequentialImpulseConstraintSolverMt::internalInitMultipleJoints( btTypedC
 	}
 }
 
-
 struct InitJointsLoop : public btIParallelForBody
 {
-    btSequentialImpulseConstraintSolverMt* m_solver;
-    btTypedConstraint** m_constraints;
+	btSequentialImpulseConstraintSolverMt* m_solver;
+	btTypedConstraint** m_constraints;
 
-    InitJointsLoop( btSequentialImpulseConstraintSolverMt* solver, btTypedConstraint** constraints )
-    {
-        m_solver = solver;
-        m_constraints = constraints;
-    }
-    void forLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        m_solver->internalInitMultipleJoints( m_constraints, iBegin, iEnd );
-    }
+	InitJointsLoop(btSequentialImpulseConstraintSolverMt* solver, btTypedConstraint** constraints)
+	{
+		m_solver = solver;
+		m_constraints = constraints;
+	}
+	void forLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		m_solver->internalInitMultipleJoints(m_constraints, iBegin, iEnd);
+	}
 };
 
-
-void btSequentialImpulseConstraintSolverMt::internalConvertMultipleJoints( const btAlignedObjectArray<JointParams>& jointParamsArray, btTypedConstraint** constraints, int iBegin, int iEnd, const btContactSolverInfo& infoGlobal )
+void btSequentialImpulseConstraintSolverMt::internalConvertMultipleJoints(const btAlignedObjectArray<JointParams>& jointParamsArray, btTypedConstraint** constraints, int iBegin, int iEnd, const btContactSolverInfo& infoGlobal)
 {
-    BT_PROFILE("internalConvertMultipleJoints");
-    for ( int i = iBegin; i < iEnd; ++i )
-    {
-        const JointParams& jointParams = jointParamsArray[ i ];
-        int currentRow = jointParams.m_solverConstraint;
-        if ( currentRow != -1 )
-        {
-            const btTypedConstraint::btConstraintInfo1& info1 = m_tmpConstraintSizesPool[ i ];
-            btAssert( currentRow < m_tmpSolverNonContactConstraintPool.size() );
-            btAssert( info1.m_numConstraintRows > 0 );
+	BT_PROFILE("internalConvertMultipleJoints");
+	for (int i = iBegin; i < iEnd; ++i)
+	{
+		const JointParams& jointParams = jointParamsArray[i];
+		int currentRow = jointParams.m_solverConstraint;
+		if (currentRow != -1)
+		{
+			const btTypedConstraint::btConstraintInfo1& info1 = m_tmpConstraintSizesPool[i];
+			btAssert(currentRow < m_tmpSolverNonContactConstraintPool.size());
+			btAssert(info1.m_numConstraintRows > 0);
 
-            btSolverConstraint* currentConstraintRow = &m_tmpSolverNonContactConstraintPool[ currentRow ];
-            btTypedConstraint* constraint = constraints[ i ];
+			btSolverConstraint* currentConstraintRow = &m_tmpSolverNonContactConstraintPool[currentRow];
+			btTypedConstraint* constraint = constraints[i];
 
-            convertJoint( currentConstraintRow, constraint, info1, jointParams.m_solverBodyA, jointParams.m_solverBodyB, infoGlobal );
-        }
-    }
+			convertJoint(currentConstraintRow, constraint, info1, jointParams.m_solverBodyA, jointParams.m_solverBodyB, infoGlobal);
+		}
+	}
 }
 
-
 struct ConvertJointsLoop : public btIParallelForBody
 {
-    btSequentialImpulseConstraintSolverMt* m_solver;
-    const btAlignedObjectArray<btSequentialImpulseConstraintSolverMt::JointParams>& m_jointParamsArray;
-    btTypedConstraint** m_srcConstraints;
-    const btContactSolverInfo& m_infoGlobal;
+	btSequentialImpulseConstraintSolverMt* m_solver;
+	const btAlignedObjectArray<btSequentialImpulseConstraintSolverMt::JointParams>& m_jointParamsArray;
+	btTypedConstraint** m_srcConstraints;
+	const btContactSolverInfo& m_infoGlobal;
 
-    ConvertJointsLoop( btSequentialImpulseConstraintSolverMt* solver,
-        const btAlignedObjectArray<btSequentialImpulseConstraintSolverMt::JointParams>& jointParamsArray,
-        btTypedConstraint** srcConstraints,
-        const btContactSolverInfo& infoGlobal
-    ) :
-        m_jointParamsArray(jointParamsArray),
-        m_infoGlobal(infoGlobal)
-    {
-        m_solver = solver;
-        m_srcConstraints = srcConstraints;
-    }
-    void forLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        m_solver->internalConvertMultipleJoints( m_jointParamsArray, m_srcConstraints, iBegin, iEnd, m_infoGlobal );
-    }
+	ConvertJointsLoop(btSequentialImpulseConstraintSolverMt* solver,
+					  const btAlignedObjectArray<btSequentialImpulseConstraintSolverMt::JointParams>& jointParamsArray,
+					  btTypedConstraint** srcConstraints,
+					  const btContactSolverInfo& infoGlobal) : m_jointParamsArray(jointParamsArray),
+															   m_infoGlobal(infoGlobal)
+	{
+		m_solver = solver;
+		m_srcConstraints = srcConstraints;
+	}
+	void forLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		m_solver->internalConvertMultipleJoints(m_jointParamsArray, m_srcConstraints, iBegin, iEnd, m_infoGlobal);
+	}
 };
 
-
 void btSequentialImpulseConstraintSolverMt::convertJoints(btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal)
 {
-    if ( !m_useBatching )
-    {
-        btSequentialImpulseConstraintSolver::convertJoints(constraints, numConstraints, infoGlobal);
-        return;
-    }
-    BT_PROFILE("convertJoints");
-    bool parallelJointSetup = true;
+	if (!m_useBatching)
+	{
+		btSequentialImpulseConstraintSolver::convertJoints(constraints, numConstraints, infoGlobal);
+		return;
+	}
+	BT_PROFILE("convertJoints");
+	bool parallelJointSetup = true;
 	m_tmpConstraintSizesPool.resizeNoInitialize(numConstraints);
-    if (parallelJointSetup)
-    {
-        InitJointsLoop loop(this, constraints);
-        int grainSize = 40;
-        btParallelFor(0, numConstraints, grainSize, loop);
-    }
-    else
-    {
-        internalInitMultipleJoints( constraints, 0, numConstraints );
-    }
+	if (parallelJointSetup)
+	{
+		InitJointsLoop loop(this, constraints);
+		int grainSize = 40;
+		btParallelFor(0, numConstraints, grainSize, loop);
+	}
+	else
+	{
+		internalInitMultipleJoints(constraints, 0, numConstraints);
+	}
 
 	int totalNumRows = 0;
-    btAlignedObjectArray<JointParams> jointParamsArray;
-    jointParamsArray.resizeNoInitialize(numConstraints);
+	btAlignedObjectArray<JointParams> jointParamsArray;
+	jointParamsArray.resizeNoInitialize(numConstraints);
 
 	//calculate the total number of contraint rows
-	for (int i=0;i<numConstraints;i++)
+	for (int i = 0; i < numConstraints; i++)
 	{
-        btTypedConstraint* constraint = constraints[ i ];
+		btTypedConstraint* constraint = constraints[i];
 
-        JointParams& params = jointParamsArray[ i ];
+		JointParams& params = jointParamsArray[i];
 		const btTypedConstraint::btConstraintInfo1& info1 = m_tmpConstraintSizesPool[i];
 
 		if (info1.m_numConstraintRows)
 		{
-            params.m_solverConstraint = totalNumRows;
-            params.m_solverBodyA = getOrInitSolverBody( constraint->getRigidBodyA(), infoGlobal.m_timeStep );
-            params.m_solverBodyB = getOrInitSolverBody( constraint->getRigidBodyB(), infoGlobal.m_timeStep );
+			params.m_solverConstraint = totalNumRows;
+			params.m_solverBodyA = getOrInitSolverBody(constraint->getRigidBodyA(), infoGlobal.m_timeStep);
+			params.m_solverBodyB = getOrInitSolverBody(constraint->getRigidBodyB(), infoGlobal.m_timeStep);
 		}
-        else
+		else
 		{
-            params.m_solverConstraint = -1;
+			params.m_solverConstraint = -1;
 		}
 		totalNumRows += info1.m_numConstraintRows;
 	}
 	m_tmpSolverNonContactConstraintPool.resizeNoInitialize(totalNumRows);
 
 	///setup the btSolverConstraints
-    if ( parallelJointSetup )
-    {
-        ConvertJointsLoop loop(this, jointParamsArray, constraints, infoGlobal);
-        int grainSize = 20;
-        btParallelFor(0, numConstraints, grainSize, loop);
-    }
-    else
-    {
-        internalConvertMultipleJoints( jointParamsArray, constraints, 0, numConstraints, infoGlobal );
-    }
-    setupBatchedJointConstraints();
+	if (parallelJointSetup)
+	{
+		ConvertJointsLoop loop(this, jointParamsArray, constraints, infoGlobal);
+		int grainSize = 20;
+		btParallelFor(0, numConstraints, grainSize, loop);
+	}
+	else
+	{
+		internalConvertMultipleJoints(jointParamsArray, constraints, 0, numConstraints, infoGlobal);
+	}
+	setupBatchedJointConstraints();
 }
 
-
 void btSequentialImpulseConstraintSolverMt::internalConvertBodies(btCollisionObject** bodies, int iBegin, int iEnd, const btContactSolverInfo& infoGlobal)
 {
-    BT_PROFILE("internalConvertBodies");
-    for (int i=iBegin; i < iEnd; i++)
+	BT_PROFILE("internalConvertBodies");
+	for (int i = iBegin; i < iEnd; i++)
 	{
-        btCollisionObject* obj = bodies[i];
+		btCollisionObject* obj = bodies[i];
 		obj->setCompanionId(i);
 		btSolverBody& solverBody = m_tmpSolverBodyPool[i];
-        initSolverBody(&solverBody, obj, infoGlobal.m_timeStep);
+		initSolverBody(&solverBody, obj, infoGlobal.m_timeStep);
 
 		btRigidBody* body = btRigidBody::upcast(obj);
 		if (body && body->getInvMass())
 		{
-			btVector3 gyroForce (0,0,0);
-			if (body->getFlags()&BT_ENABLE_GYROSCOPIC_FORCE_EXPLICIT)
+			btVector3 gyroForce(0, 0, 0);
+			if (body->getFlags() & BT_ENABLE_GYROSCOPIC_FORCE_EXPLICIT)
 			{
 				gyroForce = body->computeGyroscopicForceExplicit(infoGlobal.m_maxGyroscopicForce);
-				solverBody.m_externalTorqueImpulse -= gyroForce*body->getInvInertiaTensorWorld()*infoGlobal.m_timeStep;
+				solverBody.m_externalTorqueImpulse -= gyroForce * body->getInvInertiaTensorWorld() * infoGlobal.m_timeStep;
 			}
-			if (body->getFlags()&BT_ENABLE_GYROSCOPIC_FORCE_IMPLICIT_WORLD)
+			if (body->getFlags() & BT_ENABLE_GYROSCOPIC_FORCE_IMPLICIT_WORLD)
 			{
 				gyroForce = body->computeGyroscopicImpulseImplicit_World(infoGlobal.m_timeStep);
 				solverBody.m_externalTorqueImpulse += gyroForce;
 			}
-			if (body->getFlags()&BT_ENABLE_GYROSCOPIC_FORCE_IMPLICIT_BODY)
+			if (body->getFlags() & BT_ENABLE_GYROSCOPIC_FORCE_IMPLICIT_BODY)
 			{
 				gyroForce = body->computeGyroscopicImpulseImplicit_Body(infoGlobal.m_timeStep);
 				solverBody.m_externalTorqueImpulse += gyroForce;
@@ -806,809 +777,772 @@ void btSequentialImpulseConstraintSolverMt::internalConvertBodies(btCollisionObj
 	}
 }
 
-
 struct ConvertBodiesLoop : public btIParallelForBody
 {
-    btSequentialImpulseConstraintSolverMt* m_solver;
-    btCollisionObject** m_bodies;
-    int m_numBodies;
-    const btContactSolverInfo& m_infoGlobal;
+	btSequentialImpulseConstraintSolverMt* m_solver;
+	btCollisionObject** m_bodies;
+	int m_numBodies;
+	const btContactSolverInfo& m_infoGlobal;
 
-    ConvertBodiesLoop( btSequentialImpulseConstraintSolverMt* solver,
-        btCollisionObject** bodies,
-        int numBodies,
-        const btContactSolverInfo& infoGlobal
-    ) :
-        m_infoGlobal(infoGlobal)
-    {
-        m_solver = solver;
-        m_bodies = bodies;
-        m_numBodies = numBodies;
-    }
-    void forLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        m_solver->internalConvertBodies( m_bodies, iBegin, iEnd, m_infoGlobal );
-    }
+	ConvertBodiesLoop(btSequentialImpulseConstraintSolverMt* solver,
+					  btCollisionObject** bodies,
+					  int numBodies,
+					  const btContactSolverInfo& infoGlobal) : m_infoGlobal(infoGlobal)
+	{
+		m_solver = solver;
+		m_bodies = bodies;
+		m_numBodies = numBodies;
+	}
+	void forLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		m_solver->internalConvertBodies(m_bodies, iBegin, iEnd, m_infoGlobal);
+	}
 };
 
-
 void btSequentialImpulseConstraintSolverMt::convertBodies(btCollisionObject** bodies, int numBodies, const btContactSolverInfo& infoGlobal)
 {
-    BT_PROFILE("convertBodies");
-    m_kinematicBodyUniqueIdToSolverBodyTable.resize( 0 );
+	BT_PROFILE("convertBodies");
+	m_kinematicBodyUniqueIdToSolverBodyTable.resize(0);
 
-	m_tmpSolverBodyPool.resizeNoInitialize(numBodies+1);
+	m_tmpSolverBodyPool.resizeNoInitialize(numBodies + 1);
 
-    m_fixedBodyId = numBodies;
-    {
-        btSolverBody& fixedBody = m_tmpSolverBodyPool[ m_fixedBodyId ];
-        initSolverBody( &fixedBody, NULL, infoGlobal.m_timeStep );
-    }
+	m_fixedBodyId = numBodies;
+	{
+		btSolverBody& fixedBody = m_tmpSolverBodyPool[m_fixedBodyId];
+		initSolverBody(&fixedBody, NULL, infoGlobal.m_timeStep);
+	}
 
-    bool parallelBodySetup = true;
-    if (parallelBodySetup)
-    {
-        ConvertBodiesLoop loop(this, bodies, numBodies, infoGlobal);
-        int grainSize = 40;
-        btParallelFor(0, numBodies, grainSize, loop);
-    }
-    else
-    {
-        internalConvertBodies( bodies, 0, numBodies, infoGlobal );
-    }
+	bool parallelBodySetup = true;
+	if (parallelBodySetup)
+	{
+		ConvertBodiesLoop loop(this, bodies, numBodies, infoGlobal);
+		int grainSize = 40;
+		btParallelFor(0, numBodies, grainSize, loop);
+	}
+	else
+	{
+		internalConvertBodies(bodies, 0, numBodies, infoGlobal);
+	}
 }
 
-
 btScalar btSequentialImpulseConstraintSolverMt::solveGroupCacheFriendlySetup(
-     btCollisionObject** bodies,
-     int numBodies,
-     btPersistentManifold** manifoldPtr,
-     int numManifolds,
-     btTypedConstraint** constraints,
-     int numConstraints,
-     const btContactSolverInfo& infoGlobal,
-     btIDebugDraw* debugDrawer
-     )
+	btCollisionObject** bodies,
+	int numBodies,
+	btPersistentManifold** manifoldPtr,
+	int numManifolds,
+	btTypedConstraint** constraints,
+	int numConstraints,
+	const btContactSolverInfo& infoGlobal,
+	btIDebugDraw* debugDrawer)
 {
-    m_numFrictionDirections = (infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS) ? 2 : 1;
-    m_useBatching = false;
-    if ( numManifolds >= s_minimumContactManifoldsForBatching &&
-        (s_allowNestedParallelForLoops || !btThreadsAreRunning())
-        )
-    {
-        m_useBatching = true;
-        m_batchedContactConstraints.m_debugDrawer = debugDrawer;
-        m_batchedJointConstraints.m_debugDrawer = debugDrawer;
-    }
-    btSequentialImpulseConstraintSolver::solveGroupCacheFriendlySetup( bodies,
-                                                                       numBodies,
-                                                                       manifoldPtr,
-                                                                       numManifolds,
-                                                                       constraints,
-                                                                       numConstraints,
-                                                                       infoGlobal,
-                                                                       debugDrawer
-                                                                       );
-    return 0.0f;
+	m_numFrictionDirections = (infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS) ? 2 : 1;
+	m_useBatching = false;
+	if (numManifolds >= s_minimumContactManifoldsForBatching &&
+		(s_allowNestedParallelForLoops || !btThreadsAreRunning()))
+	{
+		m_useBatching = true;
+		m_batchedContactConstraints.m_debugDrawer = debugDrawer;
+		m_batchedJointConstraints.m_debugDrawer = debugDrawer;
+	}
+	btSequentialImpulseConstraintSolver::solveGroupCacheFriendlySetup(bodies,
+																	  numBodies,
+																	  manifoldPtr,
+																	  numManifolds,
+																	  constraints,
+																	  numConstraints,
+																	  infoGlobal,
+																	  debugDrawer);
+	return 0.0f;
 }
 
-
-btScalar btSequentialImpulseConstraintSolverMt::resolveMultipleContactSplitPenetrationImpulseConstraints( const btAlignedObjectArray<int>& consIndices, int batchBegin, int batchEnd )
+btScalar btSequentialImpulseConstraintSolverMt::resolveMultipleContactSplitPenetrationImpulseConstraints(const btAlignedObjectArray<int>& consIndices, int batchBegin, int batchEnd)
 {
-    btScalar leastSquaresResidual = 0.f;
-    for ( int iiCons = batchBegin; iiCons < batchEnd; ++iiCons )
-    {
-        int iCons = consIndices[ iiCons ];
-        const btSolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[ iCons ];
-        btSolverBody& bodyA = m_tmpSolverBodyPool[ solveManifold.m_solverBodyIdA ];
-        btSolverBody& bodyB = m_tmpSolverBodyPool[ solveManifold.m_solverBodyIdB ];
-        btScalar residual = resolveSplitPenetrationImpulse( bodyA, bodyB, solveManifold );
-        leastSquaresResidual += residual*residual;
-    }
-    return leastSquaresResidual;
+	btScalar leastSquaresResidual = 0.f;
+	for (int iiCons = batchBegin; iiCons < batchEnd; ++iiCons)
+	{
+		int iCons = consIndices[iiCons];
+		const btSolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[iCons];
+		btSolverBody& bodyA = m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA];
+		btSolverBody& bodyB = m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB];
+		btScalar residual = resolveSplitPenetrationImpulse(bodyA, bodyB, solveManifold);
+		leastSquaresResidual += residual * residual;
+	}
+	return leastSquaresResidual;
 }
 
-
 struct ContactSplitPenetrationImpulseSolverLoop : public btIParallelSumBody
 {
-    btSequentialImpulseConstraintSolverMt* m_solver;
-    const btBatchedConstraints* m_bc;
+	btSequentialImpulseConstraintSolverMt* m_solver;
+	const btBatchedConstraints* m_bc;
 
-    ContactSplitPenetrationImpulseSolverLoop( btSequentialImpulseConstraintSolverMt* solver, const btBatchedConstraints* bc )
-    {
-        m_solver = solver;
-        m_bc = bc;
-    }
-    btScalar sumLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        BT_PROFILE( "ContactSplitPenetrationImpulseSolverLoop" );
-        btScalar sum = 0;
-        for ( int iBatch = iBegin; iBatch < iEnd; ++iBatch )
-        {
-            const btBatchedConstraints::Range& batch = m_bc->m_batches[ iBatch ];
-            sum += m_solver->resolveMultipleContactSplitPenetrationImpulseConstraints( m_bc->m_constraintIndices, batch.begin, batch.end );
-        }
-        return sum;
-    }
+	ContactSplitPenetrationImpulseSolverLoop(btSequentialImpulseConstraintSolverMt* solver, const btBatchedConstraints* bc)
+	{
+		m_solver = solver;
+		m_bc = bc;
+	}
+	btScalar sumLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		BT_PROFILE("ContactSplitPenetrationImpulseSolverLoop");
+		btScalar sum = 0;
+		for (int iBatch = iBegin; iBatch < iEnd; ++iBatch)
+		{
+			const btBatchedConstraints::Range& batch = m_bc->m_batches[iBatch];
+			sum += m_solver->resolveMultipleContactSplitPenetrationImpulseConstraints(m_bc->m_constraintIndices, batch.begin, batch.end);
+		}
+		return sum;
+	}
 };
 
-
-void btSequentialImpulseConstraintSolverMt::solveGroupCacheFriendlySplitImpulseIterations(btCollisionObject** bodies,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer)
+void btSequentialImpulseConstraintSolverMt::solveGroupCacheFriendlySplitImpulseIterations(btCollisionObject** bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer)
 {
 	BT_PROFILE("solveGroupCacheFriendlySplitImpulseIterations");
 	if (infoGlobal.m_splitImpulse)
 	{
-        for ( int iteration = 0; iteration < infoGlobal.m_numIterations; iteration++ )
-        {
-            btScalar leastSquaresResidual = 0.f;
-            if (m_useBatching)
-            {
-                const btBatchedConstraints& batchedCons = m_batchedContactConstraints;
-                ContactSplitPenetrationImpulseSolverLoop loop( this, &batchedCons );
-                btScalar leastSquaresResidual = 0.f;
-                for ( int iiPhase = 0; iiPhase < batchedCons.m_phases.size(); ++iiPhase )
-                {
-                    int iPhase = batchedCons.m_phaseOrder[ iiPhase ];
-                    const btBatchedConstraints::Range& phase = batchedCons.m_phases[ iPhase ];
-                    int grainSize = batchedCons.m_phaseGrainSize[iPhase];
-                    leastSquaresResidual += btParallelSum( phase.begin, phase.end, grainSize, loop );
-                }
-            }
-            else
-            {
-                // non-batched
-                leastSquaresResidual = resolveMultipleContactSplitPenetrationImpulseConstraints(m_orderTmpConstraintPool, 0, m_tmpSolverContactConstraintPool.size());
-            }
-            if ( leastSquaresResidual <= infoGlobal.m_leastSquaresResidualThreshold || iteration >= ( infoGlobal.m_numIterations - 1 ) )
-            {
+		for (int iteration = 0; iteration < infoGlobal.m_numIterations; iteration++)
+		{
+			btScalar leastSquaresResidual = 0.f;
+			if (m_useBatching)
+			{
+				const btBatchedConstraints& batchedCons = m_batchedContactConstraints;
+				ContactSplitPenetrationImpulseSolverLoop loop(this, &batchedCons);
+				btScalar leastSquaresResidual = 0.f;
+				for (int iiPhase = 0; iiPhase < batchedCons.m_phases.size(); ++iiPhase)
+				{
+					int iPhase = batchedCons.m_phaseOrder[iiPhase];
+					const btBatchedConstraints::Range& phase = batchedCons.m_phases[iPhase];
+					int grainSize = batchedCons.m_phaseGrainSize[iPhase];
+					leastSquaresResidual += btParallelSum(phase.begin, phase.end, grainSize, loop);
+				}
+			}
+			else
+			{
+				// non-batched
+				leastSquaresResidual = resolveMultipleContactSplitPenetrationImpulseConstraints(m_orderTmpConstraintPool, 0, m_tmpSolverContactConstraintPool.size());
+			}
+			if (leastSquaresResidual <= infoGlobal.m_leastSquaresResidualThreshold || iteration >= (infoGlobal.m_numIterations - 1))
+			{
 #ifdef VERBOSE_RESIDUAL_PRINTF
-                printf( "residual = %f at iteration #%d\n", leastSquaresResidual, iteration );
+				printf("residual = %f at iteration #%d\n", leastSquaresResidual, iteration);
 #endif
-                break;
-            }
-        }
+				break;
+			}
+		}
 	}
 }
 
-
-btScalar btSequentialImpulseConstraintSolverMt::solveSingleIteration(int iteration, btCollisionObject** bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer)
+btScalar btSequentialImpulseConstraintSolverMt::solveSingleIteration(int iteration, btCollisionObject** bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer)
 {
-    if ( !m_useBatching )
-    {
-        return btSequentialImpulseConstraintSolver::solveSingleIteration( iteration, bodies, numBodies, manifoldPtr, numManifolds, constraints, numConstraints, infoGlobal, debugDrawer );
-    }
-    BT_PROFILE( "solveSingleIterationMt" );
-    btScalar leastSquaresResidual = 0.f;
+	if (!m_useBatching)
+	{
+		return btSequentialImpulseConstraintSolver::solveSingleIteration(iteration, bodies, numBodies, manifoldPtr, numManifolds, constraints, numConstraints, infoGlobal, debugDrawer);
+	}
+	BT_PROFILE("solveSingleIterationMt");
+	btScalar leastSquaresResidual = 0.f;
 
 	if (infoGlobal.m_solverMode & SOLVER_RANDMIZE_ORDER)
 	{
-		if (1)			// uncomment this for a bit less random ((iteration & 7) == 0)
+		if (1)  // uncomment this for a bit less random ((iteration & 7) == 0)
 		{
-            randomizeConstraintOrdering(iteration, infoGlobal.m_numIterations);
+			randomizeConstraintOrdering(iteration, infoGlobal.m_numIterations);
 		}
 	}
 
 	{
 		///solve all joint constraints
-        leastSquaresResidual += resolveAllJointConstraints(iteration);
+		leastSquaresResidual += resolveAllJointConstraints(iteration);
 
-		if (iteration< infoGlobal.m_numIterations)
+		if (iteration < infoGlobal.m_numIterations)
 		{
-            // this loop is only used for cone-twist constraints,
-            // it would be nice to skip this loop if none of the constraints need it
-            if ( m_useObsoleteJointConstraints )
-            {
-                for ( int j = 0; j<numConstraints; j++ )
-                {
-                    if ( constraints[ j ]->isEnabled() )
-                    {
-                        int bodyAid = getOrInitSolverBody( constraints[ j ]->getRigidBodyA(), infoGlobal.m_timeStep );
-                        int bodyBid = getOrInitSolverBody( constraints[ j ]->getRigidBodyB(), infoGlobal.m_timeStep );
-                        btSolverBody& bodyA = m_tmpSolverBodyPool[ bodyAid ];
-                        btSolverBody& bodyB = m_tmpSolverBodyPool[ bodyBid ];
-                        constraints[ j ]->solveConstraintObsolete( bodyA, bodyB, infoGlobal.m_timeStep );
-                    }
-                }
-            }
+			// this loop is only used for cone-twist constraints,
+			// it would be nice to skip this loop if none of the constraints need it
+			if (m_useObsoleteJointConstraints)
+			{
+				for (int j = 0; j < numConstraints; j++)
+				{
+					if (constraints[j]->isEnabled())
+					{
+						int bodyAid = getOrInitSolverBody(constraints[j]->getRigidBodyA(), infoGlobal.m_timeStep);
+						int bodyBid = getOrInitSolverBody(constraints[j]->getRigidBodyB(), infoGlobal.m_timeStep);
+						btSolverBody& bodyA = m_tmpSolverBodyPool[bodyAid];
+						btSolverBody& bodyB = m_tmpSolverBodyPool[bodyBid];
+						constraints[j]->solveConstraintObsolete(bodyA, bodyB, infoGlobal.m_timeStep);
+					}
+				}
+			}
 
 			if (infoGlobal.m_solverMode & SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS)
 			{
-                // solve all contact, contact-friction, and rolling friction constraints interleaved
-                leastSquaresResidual += resolveAllContactConstraintsInterleaved();
+				// solve all contact, contact-friction, and rolling friction constraints interleaved
+				leastSquaresResidual += resolveAllContactConstraintsInterleaved();
 			}
-			else//SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS
+			else  //SOLVER_INTERLEAVE_CONTACT_AND_FRICTION_CONSTRAINTS
 			{
-                // don't interleave them
+				// don't interleave them
 				// solve all contact constraints
-                leastSquaresResidual += resolveAllContactConstraints();
+				leastSquaresResidual += resolveAllContactConstraints();
 
 				// solve all contact friction constraints
-                leastSquaresResidual += resolveAllContactFrictionConstraints();
+				leastSquaresResidual += resolveAllContactFrictionConstraints();
 
-                // solve all rolling friction constraints
-                leastSquaresResidual += resolveAllRollingFrictionConstraints();
+				// solve all rolling friction constraints
+				leastSquaresResidual += resolveAllRollingFrictionConstraints();
 			}
 		}
 	}
-    return leastSquaresResidual;
+	return leastSquaresResidual;
 }
 
-
-btScalar btSequentialImpulseConstraintSolverMt::resolveMultipleJointConstraints( const btAlignedObjectArray<int>& consIndices, int batchBegin, int batchEnd, int iteration )
+btScalar btSequentialImpulseConstraintSolverMt::resolveMultipleJointConstraints(const btAlignedObjectArray<int>& consIndices, int batchBegin, int batchEnd, int iteration)
 {
-    btScalar leastSquaresResidual = 0.f;
-    for ( int iiCons = batchBegin; iiCons < batchEnd; ++iiCons )
-    {
-        int iCons = consIndices[ iiCons ];
-        const btSolverConstraint& constraint = m_tmpSolverNonContactConstraintPool[ iCons ];
-        if ( iteration < constraint.m_overrideNumSolverIterations )
-        {
-            btSolverBody& bodyA = m_tmpSolverBodyPool[ constraint.m_solverBodyIdA ];
-            btSolverBody& bodyB = m_tmpSolverBodyPool[ constraint.m_solverBodyIdB ];
-            btScalar residual = resolveSingleConstraintRowGeneric( bodyA, bodyB, constraint );
-            leastSquaresResidual += residual*residual;
-        }
-    }
-    return leastSquaresResidual;
+	btScalar leastSquaresResidual = 0.f;
+	for (int iiCons = batchBegin; iiCons < batchEnd; ++iiCons)
+	{
+		int iCons = consIndices[iiCons];
+		const btSolverConstraint& constraint = m_tmpSolverNonContactConstraintPool[iCons];
+		if (iteration < constraint.m_overrideNumSolverIterations)
+		{
+			btSolverBody& bodyA = m_tmpSolverBodyPool[constraint.m_solverBodyIdA];
+			btSolverBody& bodyB = m_tmpSolverBodyPool[constraint.m_solverBodyIdB];
+			btScalar residual = resolveSingleConstraintRowGeneric(bodyA, bodyB, constraint);
+			leastSquaresResidual += residual * residual;
+		}
+	}
+	return leastSquaresResidual;
 }
 
-
-btScalar btSequentialImpulseConstraintSolverMt::resolveMultipleContactConstraints( const btAlignedObjectArray<int>& consIndices, int batchBegin, int batchEnd )
+btScalar btSequentialImpulseConstraintSolverMt::resolveMultipleContactConstraints(const btAlignedObjectArray<int>& consIndices, int batchBegin, int batchEnd)
 {
-    btScalar leastSquaresResidual = 0.f;
-    for ( int iiCons = batchBegin; iiCons < batchEnd; ++iiCons )
-    {
-        int iCons = consIndices[ iiCons ];
-        const btSolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[ iCons ];
-        btSolverBody& bodyA = m_tmpSolverBodyPool[ solveManifold.m_solverBodyIdA ];
-        btSolverBody& bodyB = m_tmpSolverBodyPool[ solveManifold.m_solverBodyIdB ];
-        btScalar residual = resolveSingleConstraintRowLowerLimit( bodyA, bodyB, solveManifold );
-        leastSquaresResidual += residual*residual;
-    }
-    return leastSquaresResidual;
+	btScalar leastSquaresResidual = 0.f;
+	for (int iiCons = batchBegin; iiCons < batchEnd; ++iiCons)
+	{
+		int iCons = consIndices[iiCons];
+		const btSolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[iCons];
+		btSolverBody& bodyA = m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA];
+		btSolverBody& bodyB = m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB];
+		btScalar residual = resolveSingleConstraintRowLowerLimit(bodyA, bodyB, solveManifold);
+		leastSquaresResidual += residual * residual;
+	}
+	return leastSquaresResidual;
 }
 
-
-btScalar btSequentialImpulseConstraintSolverMt::resolveMultipleContactFrictionConstraints( const btAlignedObjectArray<int>& consIndices, int batchBegin, int batchEnd )
+btScalar btSequentialImpulseConstraintSolverMt::resolveMultipleContactFrictionConstraints(const btAlignedObjectArray<int>& consIndices, int batchBegin, int batchEnd)
 {
-    btScalar leastSquaresResidual = 0.f;
-    for ( int iiCons = batchBegin; iiCons < batchEnd; ++iiCons )
-    {
-        int iContact = consIndices[ iiCons ];
-        btScalar totalImpulse = m_tmpSolverContactConstraintPool[ iContact ].m_appliedImpulse;
+	btScalar leastSquaresResidual = 0.f;
+	for (int iiCons = batchBegin; iiCons < batchEnd; ++iiCons)
+	{
+		int iContact = consIndices[iiCons];
+		btScalar totalImpulse = m_tmpSolverContactConstraintPool[iContact].m_appliedImpulse;
 
-        // apply sliding friction
-        if ( totalImpulse > 0.0f )
-        {
-            int iBegin = iContact * m_numFrictionDirections;
-            int iEnd = iBegin + m_numFrictionDirections;
-            for ( int iFriction = iBegin; iFriction < iEnd; ++iFriction )
-            {
-                btSolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[ iFriction++ ];
-                btAssert( solveManifold.m_frictionIndex == iContact );
+		// apply sliding friction
+		if (totalImpulse > 0.0f)
+		{
+			int iBegin = iContact * m_numFrictionDirections;
+			int iEnd = iBegin + m_numFrictionDirections;
+			for (int iFriction = iBegin; iFriction < iEnd; ++iFriction)
+			{
+				btSolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[iFriction++];
+				btAssert(solveManifold.m_frictionIndex == iContact);
 
-                solveManifold.m_lowerLimit = -( solveManifold.m_friction*totalImpulse );
-                solveManifold.m_upperLimit = solveManifold.m_friction*totalImpulse;
+				solveManifold.m_lowerLimit = -(solveManifold.m_friction * totalImpulse);
+				solveManifold.m_upperLimit = solveManifold.m_friction * totalImpulse;
 
-                btSolverBody& bodyA = m_tmpSolverBodyPool[ solveManifold.m_solverBodyIdA ];
-                btSolverBody& bodyB = m_tmpSolverBodyPool[ solveManifold.m_solverBodyIdB ];
-                btScalar residual = resolveSingleConstraintRowGeneric( bodyA, bodyB, solveManifold );
-                leastSquaresResidual += residual*residual;
-            }
-        }
-    }
-    return leastSquaresResidual;
+				btSolverBody& bodyA = m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA];
+				btSolverBody& bodyB = m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB];
+				btScalar residual = resolveSingleConstraintRowGeneric(bodyA, bodyB, solveManifold);
+				leastSquaresResidual += residual * residual;
+			}
+		}
+	}
+	return leastSquaresResidual;
 }
 
-
-btScalar btSequentialImpulseConstraintSolverMt::resolveMultipleContactRollingFrictionConstraints( const btAlignedObjectArray<int>& consIndices, int batchBegin, int batchEnd )
+btScalar btSequentialImpulseConstraintSolverMt::resolveMultipleContactRollingFrictionConstraints(const btAlignedObjectArray<int>& consIndices, int batchBegin, int batchEnd)
 {
-    btScalar leastSquaresResidual = 0.f;
-    for ( int iiCons = batchBegin; iiCons < batchEnd; ++iiCons )
-    {
-        int iContact = consIndices[ iiCons ];
-        int iFirstRollingFriction = m_rollingFrictionIndexTable[ iContact ];
-        if ( iFirstRollingFriction >= 0 )
-        {
-            btScalar totalImpulse = m_tmpSolverContactConstraintPool[ iContact ].m_appliedImpulse;
-            // apply rolling friction
-            if ( totalImpulse > 0.0f )
-            {
-                int iBegin = iFirstRollingFriction;
-                int iEnd = iBegin + 3;
-                for ( int iRollingFric = iBegin; iRollingFric < iEnd; ++iRollingFric )
-                {
-                    btSolverConstraint& rollingFrictionConstraint = m_tmpSolverContactRollingFrictionConstraintPool[ iRollingFric ];
-                    if ( rollingFrictionConstraint.m_frictionIndex != iContact )
-                    {
-                        break;
-                    }
-                    btScalar rollingFrictionMagnitude = rollingFrictionConstraint.m_friction*totalImpulse;
-                    if ( rollingFrictionMagnitude > rollingFrictionConstraint.m_friction )
-                    {
-                        rollingFrictionMagnitude = rollingFrictionConstraint.m_friction;
-                    }
+	btScalar leastSquaresResidual = 0.f;
+	for (int iiCons = batchBegin; iiCons < batchEnd; ++iiCons)
+	{
+		int iContact = consIndices[iiCons];
+		int iFirstRollingFriction = m_rollingFrictionIndexTable[iContact];
+		if (iFirstRollingFriction >= 0)
+		{
+			btScalar totalImpulse = m_tmpSolverContactConstraintPool[iContact].m_appliedImpulse;
+			// apply rolling friction
+			if (totalImpulse > 0.0f)
+			{
+				int iBegin = iFirstRollingFriction;
+				int iEnd = iBegin + 3;
+				for (int iRollingFric = iBegin; iRollingFric < iEnd; ++iRollingFric)
+				{
+					btSolverConstraint& rollingFrictionConstraint = m_tmpSolverContactRollingFrictionConstraintPool[iRollingFric];
+					if (rollingFrictionConstraint.m_frictionIndex != iContact)
+					{
+						break;
+					}
+					btScalar rollingFrictionMagnitude = rollingFrictionConstraint.m_friction * totalImpulse;
+					if (rollingFrictionMagnitude > rollingFrictionConstraint.m_friction)
+					{
+						rollingFrictionMagnitude = rollingFrictionConstraint.m_friction;
+					}
 
-                    rollingFrictionConstraint.m_lowerLimit = -rollingFrictionMagnitude;
-                    rollingFrictionConstraint.m_upperLimit = rollingFrictionMagnitude;
+					rollingFrictionConstraint.m_lowerLimit = -rollingFrictionMagnitude;
+					rollingFrictionConstraint.m_upperLimit = rollingFrictionMagnitude;
 
-                    btScalar residual = resolveSingleConstraintRowGeneric( m_tmpSolverBodyPool[ rollingFrictionConstraint.m_solverBodyIdA ], m_tmpSolverBodyPool[ rollingFrictionConstraint.m_solverBodyIdB ], rollingFrictionConstraint );
-                    leastSquaresResidual += residual*residual;
-                }
-            }
-        }
-    }
-    return leastSquaresResidual;
+					btScalar residual = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdA], m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdB], rollingFrictionConstraint);
+					leastSquaresResidual += residual * residual;
+				}
+			}
+		}
+	}
+	return leastSquaresResidual;
 }
 
-
-btScalar btSequentialImpulseConstraintSolverMt::resolveMultipleContactConstraintsInterleaved( const btAlignedObjectArray<int>& contactIndices,
-                                                                                          int batchBegin,
-                                                                                          int batchEnd
-                                                                                          )
+btScalar btSequentialImpulseConstraintSolverMt::resolveMultipleContactConstraintsInterleaved(const btAlignedObjectArray<int>& contactIndices,
+																							 int batchBegin,
+																							 int batchEnd)
 {
-    btScalar leastSquaresResidual = 0.f;
-    int numPoolConstraints = m_tmpSolverContactConstraintPool.size();
+	btScalar leastSquaresResidual = 0.f;
+	int numPoolConstraints = m_tmpSolverContactConstraintPool.size();
 
-    for ( int iiCons = batchBegin; iiCons < batchEnd; iiCons++ )
-    {
-        btScalar totalImpulse = 0;
-        int iContact = contactIndices[ iiCons ];
-        // apply penetration constraint
-        {
-            const btSolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[ iContact ];
-            btScalar residual = resolveSingleConstraintRowLowerLimit( m_tmpSolverBodyPool[ solveManifold.m_solverBodyIdA ], m_tmpSolverBodyPool[ solveManifold.m_solverBodyIdB ], solveManifold );
-            leastSquaresResidual += residual*residual;
-            totalImpulse = solveManifold.m_appliedImpulse;
-        }
+	for (int iiCons = batchBegin; iiCons < batchEnd; iiCons++)
+	{
+		btScalar totalImpulse = 0;
+		int iContact = contactIndices[iiCons];
+		// apply penetration constraint
+		{
+			const btSolverConstraint& solveManifold = m_tmpSolverContactConstraintPool[iContact];
+			btScalar residual = resolveSingleConstraintRowLowerLimit(m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA], m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB], solveManifold);
+			leastSquaresResidual += residual * residual;
+			totalImpulse = solveManifold.m_appliedImpulse;
+		}
 
-        // apply sliding friction
-        if ( totalImpulse > 0.0f )
-        {
-            int iBegin = iContact * m_numFrictionDirections;
-            int iEnd = iBegin + m_numFrictionDirections;
-            for ( int iFriction = iBegin; iFriction < iEnd; ++iFriction )
-            {
-                btSolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[ iFriction ];
-                btAssert( solveManifold.m_frictionIndex == iContact );
+		// apply sliding friction
+		if (totalImpulse > 0.0f)
+		{
+			int iBegin = iContact * m_numFrictionDirections;
+			int iEnd = iBegin + m_numFrictionDirections;
+			for (int iFriction = iBegin; iFriction < iEnd; ++iFriction)
+			{
+				btSolverConstraint& solveManifold = m_tmpSolverContactFrictionConstraintPool[iFriction];
+				btAssert(solveManifold.m_frictionIndex == iContact);
 
-                solveManifold.m_lowerLimit = -( solveManifold.m_friction*totalImpulse );
-                solveManifold.m_upperLimit = solveManifold.m_friction*totalImpulse;
+				solveManifold.m_lowerLimit = -(solveManifold.m_friction * totalImpulse);
+				solveManifold.m_upperLimit = solveManifold.m_friction * totalImpulse;
 
-                btSolverBody& bodyA = m_tmpSolverBodyPool[ solveManifold.m_solverBodyIdA ];
-                btSolverBody& bodyB = m_tmpSolverBodyPool[ solveManifold.m_solverBodyIdB ];
-                btScalar residual = resolveSingleConstraintRowGeneric( bodyA, bodyB, solveManifold );
-                leastSquaresResidual += residual*residual;
-            }
-        }
+				btSolverBody& bodyA = m_tmpSolverBodyPool[solveManifold.m_solverBodyIdA];
+				btSolverBody& bodyB = m_tmpSolverBodyPool[solveManifold.m_solverBodyIdB];
+				btScalar residual = resolveSingleConstraintRowGeneric(bodyA, bodyB, solveManifold);
+				leastSquaresResidual += residual * residual;
+			}
+		}
 
-        // apply rolling friction
-        int iFirstRollingFriction = m_rollingFrictionIndexTable[ iContact ];
-        if ( totalImpulse > 0.0f && iFirstRollingFriction >= 0)
-        {
-            int iBegin = iFirstRollingFriction;
-            int iEnd = iBegin + 3;
-            for ( int iRollingFric = iBegin; iRollingFric < iEnd; ++iRollingFric )
-            {
-                btSolverConstraint& rollingFrictionConstraint = m_tmpSolverContactRollingFrictionConstraintPool[ iRollingFric ];
-                if ( rollingFrictionConstraint.m_frictionIndex != iContact )
-                {
-                    break;
-                }
-                btScalar rollingFrictionMagnitude = rollingFrictionConstraint.m_friction*totalImpulse;
-                if ( rollingFrictionMagnitude > rollingFrictionConstraint.m_friction )
-                {
-                    rollingFrictionMagnitude = rollingFrictionConstraint.m_friction;
-                }
+		// apply rolling friction
+		int iFirstRollingFriction = m_rollingFrictionIndexTable[iContact];
+		if (totalImpulse > 0.0f && iFirstRollingFriction >= 0)
+		{
+			int iBegin = iFirstRollingFriction;
+			int iEnd = iBegin + 3;
+			for (int iRollingFric = iBegin; iRollingFric < iEnd; ++iRollingFric)
+			{
+				btSolverConstraint& rollingFrictionConstraint = m_tmpSolverContactRollingFrictionConstraintPool[iRollingFric];
+				if (rollingFrictionConstraint.m_frictionIndex != iContact)
+				{
+					break;
+				}
+				btScalar rollingFrictionMagnitude = rollingFrictionConstraint.m_friction * totalImpulse;
+				if (rollingFrictionMagnitude > rollingFrictionConstraint.m_friction)
+				{
+					rollingFrictionMagnitude = rollingFrictionConstraint.m_friction;
+				}
 
-                rollingFrictionConstraint.m_lowerLimit = -rollingFrictionMagnitude;
-                rollingFrictionConstraint.m_upperLimit = rollingFrictionMagnitude;
+				rollingFrictionConstraint.m_lowerLimit = -rollingFrictionMagnitude;
+				rollingFrictionConstraint.m_upperLimit = rollingFrictionMagnitude;
 
-                btScalar residual = resolveSingleConstraintRowGeneric( m_tmpSolverBodyPool[ rollingFrictionConstraint.m_solverBodyIdA ], m_tmpSolverBodyPool[ rollingFrictionConstraint.m_solverBodyIdB ], rollingFrictionConstraint );
-                leastSquaresResidual += residual*residual;
-            }
-        }
-    }
-    return leastSquaresResidual;
+				btScalar residual = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdA], m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdB], rollingFrictionConstraint);
+				leastSquaresResidual += residual * residual;
+			}
+		}
+	}
+	return leastSquaresResidual;
 }
 
-
-void btSequentialImpulseConstraintSolverMt::randomizeBatchedConstraintOrdering( btBatchedConstraints* batchedConstraints )
+void btSequentialImpulseConstraintSolverMt::randomizeBatchedConstraintOrdering(btBatchedConstraints* batchedConstraints)
 {
-    btBatchedConstraints& bc = *batchedConstraints;
-    // randomize ordering of phases
-    for ( int ii = 1; ii < bc.m_phaseOrder.size(); ++ii )
-    {
-        int iSwap = btRandInt2( ii + 1 );
-        bc.m_phaseOrder.swap( ii, iSwap );
-    }
+	btBatchedConstraints& bc = *batchedConstraints;
+	// randomize ordering of phases
+	for (int ii = 1; ii < bc.m_phaseOrder.size(); ++ii)
+	{
+		int iSwap = btRandInt2(ii + 1);
+		bc.m_phaseOrder.swap(ii, iSwap);
+	}
 
-    // for each batch,
-    for ( int iBatch = 0; iBatch < bc.m_batches.size(); ++iBatch )
-    {
-        // randomize ordering of constraints within the batch
-        const btBatchedConstraints::Range& batch = bc.m_batches[ iBatch ];
-        for ( int iiCons = batch.begin; iiCons < batch.end; ++iiCons )
-        {
-            int iSwap = batch.begin + btRandInt2( iiCons - batch.begin + 1 );
-            btAssert(iSwap >= batch.begin && iSwap < batch.end);
-            bc.m_constraintIndices.swap( iiCons, iSwap );
-        }
-    }
+	// for each batch,
+	for (int iBatch = 0; iBatch < bc.m_batches.size(); ++iBatch)
+	{
+		// randomize ordering of constraints within the batch
+		const btBatchedConstraints::Range& batch = bc.m_batches[iBatch];
+		for (int iiCons = batch.begin; iiCons < batch.end; ++iiCons)
+		{
+			int iSwap = batch.begin + btRandInt2(iiCons - batch.begin + 1);
+			btAssert(iSwap >= batch.begin && iSwap < batch.end);
+			bc.m_constraintIndices.swap(iiCons, iSwap);
+		}
+	}
 }
 
-
 void btSequentialImpulseConstraintSolverMt::randomizeConstraintOrdering(int iteration, int numIterations)
 {
-    // randomize ordering of joint constraints
-    randomizeBatchedConstraintOrdering( &m_batchedJointConstraints );
+	// randomize ordering of joint constraints
+	randomizeBatchedConstraintOrdering(&m_batchedJointConstraints);
 
-    //contact/friction constraints are not solved more than numIterations
-    if ( iteration < numIterations )
-    {
-        randomizeBatchedConstraintOrdering( &m_batchedContactConstraints );
-    }
+	//contact/friction constraints are not solved more than numIterations
+	if (iteration < numIterations)
+	{
+		randomizeBatchedConstraintOrdering(&m_batchedContactConstraints);
+	}
 }
 
-
 struct JointSolverLoop : public btIParallelSumBody
 {
-    btSequentialImpulseConstraintSolverMt* m_solver;
-    const btBatchedConstraints* m_bc;
-    int m_iteration;
+	btSequentialImpulseConstraintSolverMt* m_solver;
+	const btBatchedConstraints* m_bc;
+	int m_iteration;
 
-    JointSolverLoop( btSequentialImpulseConstraintSolverMt* solver, const btBatchedConstraints* bc, int iteration )
-    {
-        m_solver = solver;
-        m_bc = bc;
-        m_iteration = iteration;
-    }
-    btScalar sumLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        BT_PROFILE( "JointSolverLoop" );
-        btScalar sum = 0;
-        for ( int iBatch = iBegin; iBatch < iEnd; ++iBatch )
-        {
-            const btBatchedConstraints::Range& batch = m_bc->m_batches[ iBatch ];
-            sum += m_solver->resolveMultipleJointConstraints( m_bc->m_constraintIndices, batch.begin, batch.end, m_iteration );
-        }
-        return sum;
-    }
+	JointSolverLoop(btSequentialImpulseConstraintSolverMt* solver, const btBatchedConstraints* bc, int iteration)
+	{
+		m_solver = solver;
+		m_bc = bc;
+		m_iteration = iteration;
+	}
+	btScalar sumLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		BT_PROFILE("JointSolverLoop");
+		btScalar sum = 0;
+		for (int iBatch = iBegin; iBatch < iEnd; ++iBatch)
+		{
+			const btBatchedConstraints::Range& batch = m_bc->m_batches[iBatch];
+			sum += m_solver->resolveMultipleJointConstraints(m_bc->m_constraintIndices, batch.begin, batch.end, m_iteration);
+		}
+		return sum;
+	}
 };
 
-
 btScalar btSequentialImpulseConstraintSolverMt::resolveAllJointConstraints(int iteration)
 {
-    BT_PROFILE( "resolveAllJointConstraints" );
-    const btBatchedConstraints& batchedCons = m_batchedJointConstraints;
-    JointSolverLoop loop( this, &batchedCons, iteration );
-    btScalar leastSquaresResidual = 0.f;
-    for ( int iiPhase = 0; iiPhase < batchedCons.m_phases.size(); ++iiPhase )
-    {
-        int iPhase = batchedCons.m_phaseOrder[ iiPhase ];
-        const btBatchedConstraints::Range& phase = batchedCons.m_phases[ iPhase ];
-        int grainSize = 1;
-        leastSquaresResidual += btParallelSum( phase.begin, phase.end, grainSize, loop );
-    }
-    return leastSquaresResidual;
+	BT_PROFILE("resolveAllJointConstraints");
+	const btBatchedConstraints& batchedCons = m_batchedJointConstraints;
+	JointSolverLoop loop(this, &batchedCons, iteration);
+	btScalar leastSquaresResidual = 0.f;
+	for (int iiPhase = 0; iiPhase < batchedCons.m_phases.size(); ++iiPhase)
+	{
+		int iPhase = batchedCons.m_phaseOrder[iiPhase];
+		const btBatchedConstraints::Range& phase = batchedCons.m_phases[iPhase];
+		int grainSize = 1;
+		leastSquaresResidual += btParallelSum(phase.begin, phase.end, grainSize, loop);
+	}
+	return leastSquaresResidual;
 }
 
-
 struct ContactSolverLoop : public btIParallelSumBody
 {
-    btSequentialImpulseConstraintSolverMt* m_solver;
-    const btBatchedConstraints* m_bc;
+	btSequentialImpulseConstraintSolverMt* m_solver;
+	const btBatchedConstraints* m_bc;
 
-    ContactSolverLoop( btSequentialImpulseConstraintSolverMt* solver, const btBatchedConstraints* bc )
-    {
-        m_solver = solver;
-        m_bc = bc;
-    }
-    btScalar sumLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        BT_PROFILE( "ContactSolverLoop" );
-        btScalar sum = 0;
-        for ( int iBatch = iBegin; iBatch < iEnd; ++iBatch )
-        {
-            const btBatchedConstraints::Range& batch = m_bc->m_batches[ iBatch ];
-            sum += m_solver->resolveMultipleContactConstraints( m_bc->m_constraintIndices, batch.begin, batch.end );
-        }
-        return sum;
-    }
+	ContactSolverLoop(btSequentialImpulseConstraintSolverMt* solver, const btBatchedConstraints* bc)
+	{
+		m_solver = solver;
+		m_bc = bc;
+	}
+	btScalar sumLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		BT_PROFILE("ContactSolverLoop");
+		btScalar sum = 0;
+		for (int iBatch = iBegin; iBatch < iEnd; ++iBatch)
+		{
+			const btBatchedConstraints::Range& batch = m_bc->m_batches[iBatch];
+			sum += m_solver->resolveMultipleContactConstraints(m_bc->m_constraintIndices, batch.begin, batch.end);
+		}
+		return sum;
+	}
 };
 
-
 btScalar btSequentialImpulseConstraintSolverMt::resolveAllContactConstraints()
 {
-    BT_PROFILE( "resolveAllContactConstraints" );
-    const btBatchedConstraints& batchedCons = m_batchedContactConstraints;
-    ContactSolverLoop loop( this, &batchedCons );
-    btScalar leastSquaresResidual = 0.f;
-    for ( int iiPhase = 0; iiPhase < batchedCons.m_phases.size(); ++iiPhase )
-    {
-        int iPhase = batchedCons.m_phaseOrder[ iiPhase ];
-        const btBatchedConstraints::Range& phase = batchedCons.m_phases[ iPhase ];
-        int grainSize = batchedCons.m_phaseGrainSize[iPhase];
-        leastSquaresResidual += btParallelSum( phase.begin, phase.end, grainSize, loop );
-    }
-    return leastSquaresResidual;
+	BT_PROFILE("resolveAllContactConstraints");
+	const btBatchedConstraints& batchedCons = m_batchedContactConstraints;
+	ContactSolverLoop loop(this, &batchedCons);
+	btScalar leastSquaresResidual = 0.f;
+	for (int iiPhase = 0; iiPhase < batchedCons.m_phases.size(); ++iiPhase)
+	{
+		int iPhase = batchedCons.m_phaseOrder[iiPhase];
+		const btBatchedConstraints::Range& phase = batchedCons.m_phases[iPhase];
+		int grainSize = batchedCons.m_phaseGrainSize[iPhase];
+		leastSquaresResidual += btParallelSum(phase.begin, phase.end, grainSize, loop);
+	}
+	return leastSquaresResidual;
 }
 
-
 struct ContactFrictionSolverLoop : public btIParallelSumBody
 {
-    btSequentialImpulseConstraintSolverMt* m_solver;
-    const btBatchedConstraints* m_bc;
+	btSequentialImpulseConstraintSolverMt* m_solver;
+	const btBatchedConstraints* m_bc;
 
-    ContactFrictionSolverLoop( btSequentialImpulseConstraintSolverMt* solver, const btBatchedConstraints* bc )
-    {
-        m_solver = solver;
-        m_bc = bc;
-    }
-    btScalar sumLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        BT_PROFILE( "ContactFrictionSolverLoop" );
-        btScalar sum = 0;
-        for ( int iBatch = iBegin; iBatch < iEnd; ++iBatch )
-        {
-            const btBatchedConstraints::Range& batch = m_bc->m_batches[ iBatch ];
-            sum += m_solver->resolveMultipleContactFrictionConstraints( m_bc->m_constraintIndices, batch.begin, batch.end );
-        }
-        return sum;
-    }
+	ContactFrictionSolverLoop(btSequentialImpulseConstraintSolverMt* solver, const btBatchedConstraints* bc)
+	{
+		m_solver = solver;
+		m_bc = bc;
+	}
+	btScalar sumLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		BT_PROFILE("ContactFrictionSolverLoop");
+		btScalar sum = 0;
+		for (int iBatch = iBegin; iBatch < iEnd; ++iBatch)
+		{
+			const btBatchedConstraints::Range& batch = m_bc->m_batches[iBatch];
+			sum += m_solver->resolveMultipleContactFrictionConstraints(m_bc->m_constraintIndices, batch.begin, batch.end);
+		}
+		return sum;
+	}
 };
 
-
 btScalar btSequentialImpulseConstraintSolverMt::resolveAllContactFrictionConstraints()
 {
-    BT_PROFILE( "resolveAllContactFrictionConstraints" );
-    const btBatchedConstraints& batchedCons = m_batchedContactConstraints;
-    ContactFrictionSolverLoop loop( this, &batchedCons );
-    btScalar leastSquaresResidual = 0.f;
-    for ( int iiPhase = 0; iiPhase < batchedCons.m_phases.size(); ++iiPhase )
-    {
-        int iPhase = batchedCons.m_phaseOrder[ iiPhase ];
-        const btBatchedConstraints::Range& phase = batchedCons.m_phases[ iPhase ];
-        int grainSize = batchedCons.m_phaseGrainSize[iPhase];
-        leastSquaresResidual += btParallelSum( phase.begin, phase.end, grainSize, loop );
-    }
-    return leastSquaresResidual;
+	BT_PROFILE("resolveAllContactFrictionConstraints");
+	const btBatchedConstraints& batchedCons = m_batchedContactConstraints;
+	ContactFrictionSolverLoop loop(this, &batchedCons);
+	btScalar leastSquaresResidual = 0.f;
+	for (int iiPhase = 0; iiPhase < batchedCons.m_phases.size(); ++iiPhase)
+	{
+		int iPhase = batchedCons.m_phaseOrder[iiPhase];
+		const btBatchedConstraints::Range& phase = batchedCons.m_phases[iPhase];
+		int grainSize = batchedCons.m_phaseGrainSize[iPhase];
+		leastSquaresResidual += btParallelSum(phase.begin, phase.end, grainSize, loop);
+	}
+	return leastSquaresResidual;
 }
 
-
 struct InterleavedContactSolverLoop : public btIParallelSumBody
 {
-    btSequentialImpulseConstraintSolverMt* m_solver;
-    const btBatchedConstraints* m_bc;
+	btSequentialImpulseConstraintSolverMt* m_solver;
+	const btBatchedConstraints* m_bc;
 
-    InterleavedContactSolverLoop( btSequentialImpulseConstraintSolverMt* solver, const btBatchedConstraints* bc )
-    {
-        m_solver = solver;
-        m_bc = bc;
-    }
-    btScalar sumLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        BT_PROFILE( "InterleavedContactSolverLoop" );
-        btScalar sum = 0;
-        for ( int iBatch = iBegin; iBatch < iEnd; ++iBatch )
-        {
-            const btBatchedConstraints::Range& batch = m_bc->m_batches[ iBatch ];
-            sum += m_solver->resolveMultipleContactConstraintsInterleaved( m_bc->m_constraintIndices, batch.begin, batch.end );
-        }
-        return sum;
-    }
+	InterleavedContactSolverLoop(btSequentialImpulseConstraintSolverMt* solver, const btBatchedConstraints* bc)
+	{
+		m_solver = solver;
+		m_bc = bc;
+	}
+	btScalar sumLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		BT_PROFILE("InterleavedContactSolverLoop");
+		btScalar sum = 0;
+		for (int iBatch = iBegin; iBatch < iEnd; ++iBatch)
+		{
+			const btBatchedConstraints::Range& batch = m_bc->m_batches[iBatch];
+			sum += m_solver->resolveMultipleContactConstraintsInterleaved(m_bc->m_constraintIndices, batch.begin, batch.end);
+		}
+		return sum;
+	}
 };
 
-
 btScalar btSequentialImpulseConstraintSolverMt::resolveAllContactConstraintsInterleaved()
 {
-    BT_PROFILE( "resolveAllContactConstraintsInterleaved" );
-    const btBatchedConstraints& batchedCons = m_batchedContactConstraints;
-    InterleavedContactSolverLoop loop( this, &batchedCons );
-    btScalar leastSquaresResidual = 0.f;
-    for ( int iiPhase = 0; iiPhase < batchedCons.m_phases.size(); ++iiPhase )
-    {
-        int iPhase = batchedCons.m_phaseOrder[ iiPhase ];
-        const btBatchedConstraints::Range& phase = batchedCons.m_phases[ iPhase ];
-        int grainSize = 1;
-        leastSquaresResidual += btParallelSum( phase.begin, phase.end, grainSize, loop );
-    }
-    return leastSquaresResidual;
+	BT_PROFILE("resolveAllContactConstraintsInterleaved");
+	const btBatchedConstraints& batchedCons = m_batchedContactConstraints;
+	InterleavedContactSolverLoop loop(this, &batchedCons);
+	btScalar leastSquaresResidual = 0.f;
+	for (int iiPhase = 0; iiPhase < batchedCons.m_phases.size(); ++iiPhase)
+	{
+		int iPhase = batchedCons.m_phaseOrder[iiPhase];
+		const btBatchedConstraints::Range& phase = batchedCons.m_phases[iPhase];
+		int grainSize = 1;
+		leastSquaresResidual += btParallelSum(phase.begin, phase.end, grainSize, loop);
+	}
+	return leastSquaresResidual;
 }
 
-
 struct ContactRollingFrictionSolverLoop : public btIParallelSumBody
 {
-    btSequentialImpulseConstraintSolverMt* m_solver;
-    const btBatchedConstraints* m_bc;
+	btSequentialImpulseConstraintSolverMt* m_solver;
+	const btBatchedConstraints* m_bc;
 
-    ContactRollingFrictionSolverLoop( btSequentialImpulseConstraintSolverMt* solver, const btBatchedConstraints* bc )
-    {
-        m_solver = solver;
-        m_bc = bc;
-    }
-    btScalar sumLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        BT_PROFILE( "ContactFrictionSolverLoop" );
-        btScalar sum = 0;
-        for ( int iBatch = iBegin; iBatch < iEnd; ++iBatch )
-        {
-            const btBatchedConstraints::Range& batch = m_bc->m_batches[ iBatch ];
-            sum += m_solver->resolveMultipleContactRollingFrictionConstraints( m_bc->m_constraintIndices, batch.begin, batch.end );
-        }
-        return sum;
-    }
+	ContactRollingFrictionSolverLoop(btSequentialImpulseConstraintSolverMt* solver, const btBatchedConstraints* bc)
+	{
+		m_solver = solver;
+		m_bc = bc;
+	}
+	btScalar sumLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		BT_PROFILE("ContactFrictionSolverLoop");
+		btScalar sum = 0;
+		for (int iBatch = iBegin; iBatch < iEnd; ++iBatch)
+		{
+			const btBatchedConstraints::Range& batch = m_bc->m_batches[iBatch];
+			sum += m_solver->resolveMultipleContactRollingFrictionConstraints(m_bc->m_constraintIndices, batch.begin, batch.end);
+		}
+		return sum;
+	}
 };
 
-
 btScalar btSequentialImpulseConstraintSolverMt::resolveAllRollingFrictionConstraints()
 {
-    BT_PROFILE( "resolveAllRollingFrictionConstraints" );
-    btScalar leastSquaresResidual = 0.f;
-    //
-    // We do not generate batches for rolling friction constraints. We assume that
-    // one of two cases is true:
-    //
-    //  1. either most bodies in the simulation have rolling friction, in which case we can use the
-    //     batches for contacts and use a lookup table to translate contact indices to rolling friction
-    //     (ignoring any contact indices that don't map to a rolling friction constraint). As long as
-    //     most contacts have a corresponding rolling friction constraint, this should parallelize well.
-    //
-    //  -OR-
-    //
-    //  2. few bodies in the simulation have rolling friction, so it is not worth trying to use the
-    //     batches from contacts as most of the contacts won't have corresponding rolling friction
-    //     constraints and most threads would end up doing very little work. Most of the time would
-    //     go to threading overhead, so we don't bother with threading.
-    //
-    int numRollingFrictionPoolConstraints = m_tmpSolverContactRollingFrictionConstraintPool.size();
-    if (numRollingFrictionPoolConstraints >= m_tmpSolverContactConstraintPool.size())
-    {
-        // use batching if there are many rolling friction constraints
-        const btBatchedConstraints& batchedCons = m_batchedContactConstraints;
-        ContactRollingFrictionSolverLoop loop( this, &batchedCons );
-        btScalar leastSquaresResidual = 0.f;
-        for ( int iiPhase = 0; iiPhase < batchedCons.m_phases.size(); ++iiPhase )
-        {
-            int iPhase = batchedCons.m_phaseOrder[ iiPhase ];
-            const btBatchedConstraints::Range& phase = batchedCons.m_phases[ iPhase ];
-            int grainSize = 1;
-            leastSquaresResidual += btParallelSum( phase.begin, phase.end, grainSize, loop );
-        }
-    }
-    else
-    {
-        // no batching, also ignores SOLVER_RANDMIZE_ORDER
-        for ( int j = 0; j < numRollingFrictionPoolConstraints; j++ )
-        {
-            btSolverConstraint& rollingFrictionConstraint = m_tmpSolverContactRollingFrictionConstraintPool[ j ];
-            if ( rollingFrictionConstraint.m_frictionIndex >= 0 )
-            {
-                btScalar totalImpulse = m_tmpSolverContactConstraintPool[ rollingFrictionConstraint.m_frictionIndex ].m_appliedImpulse;
-                if ( totalImpulse > 0.0f )
-                {
-                    btScalar rollingFrictionMagnitude = rollingFrictionConstraint.m_friction*totalImpulse;
-                    if ( rollingFrictionMagnitude > rollingFrictionConstraint.m_friction )
-                        rollingFrictionMagnitude = rollingFrictionConstraint.m_friction;
+	BT_PROFILE("resolveAllRollingFrictionConstraints");
+	btScalar leastSquaresResidual = 0.f;
+	//
+	// We do not generate batches for rolling friction constraints. We assume that
+	// one of two cases is true:
+	//
+	//  1. either most bodies in the simulation have rolling friction, in which case we can use the
+	//     batches for contacts and use a lookup table to translate contact indices to rolling friction
+	//     (ignoring any contact indices that don't map to a rolling friction constraint). As long as
+	//     most contacts have a corresponding rolling friction constraint, this should parallelize well.
+	//
+	//  -OR-
+	//
+	//  2. few bodies in the simulation have rolling friction, so it is not worth trying to use the
+	//     batches from contacts as most of the contacts won't have corresponding rolling friction
+	//     constraints and most threads would end up doing very little work. Most of the time would
+	//     go to threading overhead, so we don't bother with threading.
+	//
+	int numRollingFrictionPoolConstraints = m_tmpSolverContactRollingFrictionConstraintPool.size();
+	if (numRollingFrictionPoolConstraints >= m_tmpSolverContactConstraintPool.size())
+	{
+		// use batching if there are many rolling friction constraints
+		const btBatchedConstraints& batchedCons = m_batchedContactConstraints;
+		ContactRollingFrictionSolverLoop loop(this, &batchedCons);
+		btScalar leastSquaresResidual = 0.f;
+		for (int iiPhase = 0; iiPhase < batchedCons.m_phases.size(); ++iiPhase)
+		{
+			int iPhase = batchedCons.m_phaseOrder[iiPhase];
+			const btBatchedConstraints::Range& phase = batchedCons.m_phases[iPhase];
+			int grainSize = 1;
+			leastSquaresResidual += btParallelSum(phase.begin, phase.end, grainSize, loop);
+		}
+	}
+	else
+	{
+		// no batching, also ignores SOLVER_RANDMIZE_ORDER
+		for (int j = 0; j < numRollingFrictionPoolConstraints; j++)
+		{
+			btSolverConstraint& rollingFrictionConstraint = m_tmpSolverContactRollingFrictionConstraintPool[j];
+			if (rollingFrictionConstraint.m_frictionIndex >= 0)
+			{
+				btScalar totalImpulse = m_tmpSolverContactConstraintPool[rollingFrictionConstraint.m_frictionIndex].m_appliedImpulse;
+				if (totalImpulse > 0.0f)
+				{
+					btScalar rollingFrictionMagnitude = rollingFrictionConstraint.m_friction * totalImpulse;
+					if (rollingFrictionMagnitude > rollingFrictionConstraint.m_friction)
+						rollingFrictionMagnitude = rollingFrictionConstraint.m_friction;
 
-                    rollingFrictionConstraint.m_lowerLimit = -rollingFrictionMagnitude;
-                    rollingFrictionConstraint.m_upperLimit = rollingFrictionMagnitude;
+					rollingFrictionConstraint.m_lowerLimit = -rollingFrictionMagnitude;
+					rollingFrictionConstraint.m_upperLimit = rollingFrictionMagnitude;
 
-                    btScalar residual = resolveSingleConstraintRowGeneric( m_tmpSolverBodyPool[ rollingFrictionConstraint.m_solverBodyIdA ], m_tmpSolverBodyPool[ rollingFrictionConstraint.m_solverBodyIdB ], rollingFrictionConstraint );
-                    leastSquaresResidual += residual*residual;
-                }
-            }
-        }
-    }
-    return leastSquaresResidual;
+					btScalar residual = resolveSingleConstraintRowGeneric(m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdA], m_tmpSolverBodyPool[rollingFrictionConstraint.m_solverBodyIdB], rollingFrictionConstraint);
+					leastSquaresResidual += residual * residual;
+				}
+			}
+		}
+	}
+	return leastSquaresResidual;
 }
 
-
-void btSequentialImpulseConstraintSolverMt::internalWriteBackContacts( int iBegin, int iEnd, const btContactSolverInfo& infoGlobal )
+void btSequentialImpulseConstraintSolverMt::internalWriteBackContacts(int iBegin, int iEnd, const btContactSolverInfo& infoGlobal)
 {
-    BT_PROFILE("internalWriteBackContacts");
-    writeBackContacts(iBegin, iEnd, infoGlobal);
-    //for ( int iContact = iBegin; iContact < iEnd; ++iContact)
-    //{
-    //    const btSolverConstraint& contactConstraint = m_tmpSolverContactConstraintPool[ iContact ];
-    //    btManifoldPoint* pt = (btManifoldPoint*) contactConstraint.m_originalContactPoint;
-    //    btAssert( pt );
-    //    pt->m_appliedImpulse = contactConstraint.m_appliedImpulse;
-    //    pt->m_appliedImpulseLateral1 = m_tmpSolverContactFrictionConstraintPool[ contactConstraint.m_frictionIndex ].m_appliedImpulse;
-    //    if ( m_numFrictionDirections == 2 )
-    //    {
-    //        pt->m_appliedImpulseLateral2 = m_tmpSolverContactFrictionConstraintPool[ contactConstraint.m_frictionIndex + 1 ].m_appliedImpulse;
-    //    }
-    //}
+	BT_PROFILE("internalWriteBackContacts");
+	writeBackContacts(iBegin, iEnd, infoGlobal);
+	//for ( int iContact = iBegin; iContact < iEnd; ++iContact)
+	//{
+	//    const btSolverConstraint& contactConstraint = m_tmpSolverContactConstraintPool[ iContact ];
+	//    btManifoldPoint* pt = (btManifoldPoint*) contactConstraint.m_originalContactPoint;
+	//    btAssert( pt );
+	//    pt->m_appliedImpulse = contactConstraint.m_appliedImpulse;
+	//    pt->m_appliedImpulseLateral1 = m_tmpSolverContactFrictionConstraintPool[ contactConstraint.m_frictionIndex ].m_appliedImpulse;
+	//    if ( m_numFrictionDirections == 2 )
+	//    {
+	//        pt->m_appliedImpulseLateral2 = m_tmpSolverContactFrictionConstraintPool[ contactConstraint.m_frictionIndex + 1 ].m_appliedImpulse;
+	//    }
+	//}
 }
 
-
-void btSequentialImpulseConstraintSolverMt::internalWriteBackJoints( int iBegin, int iEnd, const btContactSolverInfo& infoGlobal )
+void btSequentialImpulseConstraintSolverMt::internalWriteBackJoints(int iBegin, int iEnd, const btContactSolverInfo& infoGlobal)
 {
 	BT_PROFILE("internalWriteBackJoints");
-    writeBackJoints(iBegin, iEnd, infoGlobal);
+	writeBackJoints(iBegin, iEnd, infoGlobal);
 }
 
-
-void btSequentialImpulseConstraintSolverMt::internalWriteBackBodies( int iBegin, int iEnd, const btContactSolverInfo& infoGlobal )
+void btSequentialImpulseConstraintSolverMt::internalWriteBackBodies(int iBegin, int iEnd, const btContactSolverInfo& infoGlobal)
 {
 	BT_PROFILE("internalWriteBackBodies");
-    writeBackBodies( iBegin, iEnd, infoGlobal );
+	writeBackBodies(iBegin, iEnd, infoGlobal);
 }
 
-
 struct WriteContactPointsLoop : public btIParallelForBody
 {
-    btSequentialImpulseConstraintSolverMt* m_solver;
-    const btContactSolverInfo* m_infoGlobal;
+	btSequentialImpulseConstraintSolverMt* m_solver;
+	const btContactSolverInfo* m_infoGlobal;
 
-    WriteContactPointsLoop( btSequentialImpulseConstraintSolverMt* solver, const btContactSolverInfo& infoGlobal )
-    {
-        m_solver = solver;
-        m_infoGlobal = &infoGlobal;
-    }
-    void forLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        m_solver->internalWriteBackContacts( iBegin, iEnd, *m_infoGlobal );
-    }
+	WriteContactPointsLoop(btSequentialImpulseConstraintSolverMt* solver, const btContactSolverInfo& infoGlobal)
+	{
+		m_solver = solver;
+		m_infoGlobal = &infoGlobal;
+	}
+	void forLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		m_solver->internalWriteBackContacts(iBegin, iEnd, *m_infoGlobal);
+	}
 };
 
-
 struct WriteJointsLoop : public btIParallelForBody
 {
-    btSequentialImpulseConstraintSolverMt* m_solver;
-    const btContactSolverInfo* m_infoGlobal;
+	btSequentialImpulseConstraintSolverMt* m_solver;
+	const btContactSolverInfo* m_infoGlobal;
 
-    WriteJointsLoop( btSequentialImpulseConstraintSolverMt* solver, const btContactSolverInfo& infoGlobal )
-    {
-        m_solver = solver;
-        m_infoGlobal = &infoGlobal;
-    }
-    void forLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        m_solver->internalWriteBackJoints( iBegin, iEnd, *m_infoGlobal );
-    }
+	WriteJointsLoop(btSequentialImpulseConstraintSolverMt* solver, const btContactSolverInfo& infoGlobal)
+	{
+		m_solver = solver;
+		m_infoGlobal = &infoGlobal;
+	}
+	void forLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		m_solver->internalWriteBackJoints(iBegin, iEnd, *m_infoGlobal);
+	}
 };
 
-
 struct WriteBodiesLoop : public btIParallelForBody
 {
-    btSequentialImpulseConstraintSolverMt* m_solver;
-    const btContactSolverInfo* m_infoGlobal;
+	btSequentialImpulseConstraintSolverMt* m_solver;
+	const btContactSolverInfo* m_infoGlobal;
 
-    WriteBodiesLoop( btSequentialImpulseConstraintSolverMt* solver, const btContactSolverInfo& infoGlobal )
-    {
-        m_solver = solver;
-        m_infoGlobal = &infoGlobal;
-    }
-    void forLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        m_solver->internalWriteBackBodies( iBegin, iEnd, *m_infoGlobal );
-    }
+	WriteBodiesLoop(btSequentialImpulseConstraintSolverMt* solver, const btContactSolverInfo& infoGlobal)
+	{
+		m_solver = solver;
+		m_infoGlobal = &infoGlobal;
+	}
+	void forLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		m_solver->internalWriteBackBodies(iBegin, iEnd, *m_infoGlobal);
+	}
 };
 
-
 btScalar btSequentialImpulseConstraintSolverMt::solveGroupCacheFriendlyFinish(btCollisionObject** bodies, int numBodies, const btContactSolverInfo& infoGlobal)
 {
 	BT_PROFILE("solveGroupCacheFriendlyFinish");
 
 	if (infoGlobal.m_solverMode & SOLVER_USE_WARMSTARTING)
-    {
-        WriteContactPointsLoop loop( this, infoGlobal );
-        int grainSize = 500;
-        btParallelFor( 0, m_tmpSolverContactConstraintPool.size(), grainSize, loop );
-    }
+	{
+		WriteContactPointsLoop loop(this, infoGlobal);
+		int grainSize = 500;
+		btParallelFor(0, m_tmpSolverContactConstraintPool.size(), grainSize, loop);
+	}
 
-    {
-        WriteJointsLoop loop( this, infoGlobal );
-        int grainSize = 400;
-        btParallelFor( 0, m_tmpSolverNonContactConstraintPool.size(), grainSize, loop );
-    }
-    {
-        WriteBodiesLoop loop( this, infoGlobal );
-        int grainSize = 100;
-        btParallelFor( 0, m_tmpSolverBodyPool.size(), grainSize, loop );
-    }
+	{
+		WriteJointsLoop loop(this, infoGlobal);
+		int grainSize = 400;
+		btParallelFor(0, m_tmpSolverNonContactConstraintPool.size(), grainSize, loop);
+	}
+	{
+		WriteBodiesLoop loop(this, infoGlobal);
+		int grainSize = 100;
+		btParallelFor(0, m_tmpSolverBodyPool.size(), grainSize, loop);
+	}
 
 	m_tmpSolverContactConstraintPool.resizeNoInitialize(0);
 	m_tmpSolverNonContactConstraintPool.resizeNoInitialize(0);
@@ -1618,4 +1552,3 @@ btScalar btSequentialImpulseConstraintSolverMt::solveGroupCacheFriendlyFinish(bt
 	m_tmpSolverBodyPool.resizeNoInitialize(0);
 	return 0.f;
 }
-
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSequentialImpulseConstraintSolverMt.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSequentialImpulseConstraintSolverMt.h
index 55d53474c47..1861ddd7d76 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSequentialImpulseConstraintSolverMt.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSequentialImpulseConstraintSolverMt.h
@@ -53,102 +53,98 @@ subject to the following restrictions:
 ///  because floating point addition is not associative due to rounding errors.
 ///  The task scheduler can and should ensure that the result of any parallelSum operation is deterministic.
 ///
-ATTRIBUTE_ALIGNED16(class) btSequentialImpulseConstraintSolverMt : public btSequentialImpulseConstraintSolver
+ATTRIBUTE_ALIGNED16(class)
+btSequentialImpulseConstraintSolverMt : public btSequentialImpulseConstraintSolver
 {
 public:
-	virtual void solveGroupCacheFriendlySplitImpulseIterations(btCollisionObject** bodies,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer) BT_OVERRIDE;
-	virtual btScalar solveSingleIteration(int iteration, btCollisionObject** bodies ,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer) BT_OVERRIDE;
-	virtual btScalar solveGroupCacheFriendlySetup(btCollisionObject** bodies,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer) BT_OVERRIDE;
-	virtual btScalar solveGroupCacheFriendlyFinish(btCollisionObject** bodies, int numBodies, const btContactSolverInfo& infoGlobal) BT_OVERRIDE;
+	virtual void solveGroupCacheFriendlySplitImpulseIterations(btCollisionObject * *bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer) BT_OVERRIDE;
+	virtual btScalar solveSingleIteration(int iteration, btCollisionObject** bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer) BT_OVERRIDE;
+	virtual btScalar solveGroupCacheFriendlySetup(btCollisionObject * *bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer) BT_OVERRIDE;
+	virtual btScalar solveGroupCacheFriendlyFinish(btCollisionObject * *bodies, int numBodies, const btContactSolverInfo& infoGlobal) BT_OVERRIDE;
 
-    // temp struct used to collect info from persistent manifolds into a cache-friendly struct using multiple threads
-    struct btContactManifoldCachedInfo
-    {
-        static const int MAX_NUM_CONTACT_POINTS = 4;
+	// temp struct used to collect info from persistent manifolds into a cache-friendly struct using multiple threads
+	struct btContactManifoldCachedInfo
+	{
+		static const int MAX_NUM_CONTACT_POINTS = 4;
 
-        int numTouchingContacts;
-        int solverBodyIds[ 2 ];
-        int contactIndex;
-        int rollingFrictionIndex;
-        bool contactHasRollingFriction[ MAX_NUM_CONTACT_POINTS ];
-        btManifoldPoint* contactPoints[ MAX_NUM_CONTACT_POINTS ];
-    };
-    // temp struct used for setting up joint constraints in parallel
-    struct JointParams
-    {
-        int m_solverConstraint;
-        int m_solverBodyA;
-        int m_solverBodyB;
-    };
-    void internalInitMultipleJoints(btTypedConstraint** constraints, int iBegin, int iEnd);
-    void internalConvertMultipleJoints( const btAlignedObjectArray<JointParams>& jointParamsArray, btTypedConstraint** constraints, int iBegin, int iEnd, const btContactSolverInfo& infoGlobal );
+		int numTouchingContacts;
+		int solverBodyIds[2];
+		int contactIndex;
+		int rollingFrictionIndex;
+		bool contactHasRollingFriction[MAX_NUM_CONTACT_POINTS];
+		btManifoldPoint* contactPoints[MAX_NUM_CONTACT_POINTS];
+	};
+	// temp struct used for setting up joint constraints in parallel
+	struct JointParams
+	{
+		int m_solverConstraint;
+		int m_solverBodyA;
+		int m_solverBodyB;
+	};
+	void internalInitMultipleJoints(btTypedConstraint * *constraints, int iBegin, int iEnd);
+	void internalConvertMultipleJoints(const btAlignedObjectArray<JointParams>& jointParamsArray, btTypedConstraint** constraints, int iBegin, int iEnd, const btContactSolverInfo& infoGlobal);
 
-    // parameters to control batching
-    static bool s_allowNestedParallelForLoops;        // whether to allow nested parallel operations
-    static int s_minimumContactManifoldsForBatching;  // don't even try to batch if fewer manifolds than this
-    static btBatchedConstraints::BatchingMethod s_contactBatchingMethod;
-    static btBatchedConstraints::BatchingMethod s_jointBatchingMethod;
-    static int s_minBatchSize;  // desired number of constraints per batch
-    static int s_maxBatchSize;
+	// parameters to control batching
+	static bool s_allowNestedParallelForLoops;        // whether to allow nested parallel operations
+	static int s_minimumContactManifoldsForBatching;  // don't even try to batch if fewer manifolds than this
+	static btBatchedConstraints::BatchingMethod s_contactBatchingMethod;
+	static btBatchedConstraints::BatchingMethod s_jointBatchingMethod;
+	static int s_minBatchSize;  // desired number of constraints per batch
+	static int s_maxBatchSize;
 
 protected:
-    static const int CACHE_LINE_SIZE = 64;
+	static const int CACHE_LINE_SIZE = 64;
 
-    btBatchedConstraints m_batchedContactConstraints;
-    btBatchedConstraints m_batchedJointConstraints;
-    int m_numFrictionDirections;
-    bool m_useBatching;
-    bool m_useObsoleteJointConstraints;
-    btAlignedObjectArray<btContactManifoldCachedInfo> m_manifoldCachedInfoArray;
-    btAlignedObjectArray<int> m_rollingFrictionIndexTable;  // lookup table mapping contact index to rolling friction index
-    btSpinMutex m_bodySolverArrayMutex;
-    char m_antiFalseSharingPadding[CACHE_LINE_SIZE]; // padding to keep mutexes in separate cachelines
-    btSpinMutex m_kinematicBodyUniqueIdToSolverBodyTableMutex;
-    btAlignedObjectArray<char> m_scratchMemory;
+	btBatchedConstraints m_batchedContactConstraints;
+	btBatchedConstraints m_batchedJointConstraints;
+	int m_numFrictionDirections;
+	bool m_useBatching;
+	bool m_useObsoleteJointConstraints;
+	btAlignedObjectArray<btContactManifoldCachedInfo> m_manifoldCachedInfoArray;
+	btAlignedObjectArray<int> m_rollingFrictionIndexTable;  // lookup table mapping contact index to rolling friction index
+	btSpinMutex m_bodySolverArrayMutex;
+	char m_antiFalseSharingPadding[CACHE_LINE_SIZE];  // padding to keep mutexes in separate cachelines
+	btSpinMutex m_kinematicBodyUniqueIdToSolverBodyTableMutex;
+	btAlignedObjectArray<char> m_scratchMemory;
 
-    virtual void randomizeConstraintOrdering( int iteration, int numIterations );
-    virtual btScalar resolveAllJointConstraints( int iteration );
-    virtual btScalar resolveAllContactConstraints();
-    virtual btScalar resolveAllContactFrictionConstraints();
-    virtual btScalar resolveAllContactConstraintsInterleaved();
-    virtual btScalar resolveAllRollingFrictionConstraints();
+	virtual void randomizeConstraintOrdering(int iteration, int numIterations);
+	virtual btScalar resolveAllJointConstraints(int iteration);
+	virtual btScalar resolveAllContactConstraints();
+	virtual btScalar resolveAllContactFrictionConstraints();
+	virtual btScalar resolveAllContactConstraintsInterleaved();
+	virtual btScalar resolveAllRollingFrictionConstraints();
 
-    virtual void setupBatchedContactConstraints();
-    virtual void setupBatchedJointConstraints();
-    virtual void convertJoints(btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal) BT_OVERRIDE;
-	virtual void convertContacts(btPersistentManifold** manifoldPtr, int numManifolds, const btContactSolverInfo& infoGlobal) BT_OVERRIDE;
-    virtual void convertBodies(btCollisionObject** bodies, int numBodies, const btContactSolverInfo& infoGlobal) BT_OVERRIDE;
+	virtual void setupBatchedContactConstraints();
+	virtual void setupBatchedJointConstraints();
+	virtual void convertJoints(btTypedConstraint * *constraints, int numConstraints, const btContactSolverInfo& infoGlobal) BT_OVERRIDE;
+	virtual void convertContacts(btPersistentManifold * *manifoldPtr, int numManifolds, const btContactSolverInfo& infoGlobal) BT_OVERRIDE;
+	virtual void convertBodies(btCollisionObject * *bodies, int numBodies, const btContactSolverInfo& infoGlobal) BT_OVERRIDE;
 
-	int getOrInitSolverBodyThreadsafe(btCollisionObject& body, btScalar timeStep);
-    void allocAllContactConstraints(btPersistentManifold** manifoldPtr, int numManifolds, const btContactSolverInfo& infoGlobal);
-    void setupAllContactConstraints(const btContactSolverInfo& infoGlobal);
-    void randomizeBatchedConstraintOrdering( btBatchedConstraints* batchedConstraints );
+	int getOrInitSolverBodyThreadsafe(btCollisionObject & body, btScalar timeStep);
+	void allocAllContactConstraints(btPersistentManifold * *manifoldPtr, int numManifolds, const btContactSolverInfo& infoGlobal);
+	void setupAllContactConstraints(const btContactSolverInfo& infoGlobal);
+	void randomizeBatchedConstraintOrdering(btBatchedConstraints * batchedConstraints);
 
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
 	btSequentialImpulseConstraintSolverMt();
 	virtual ~btSequentialImpulseConstraintSolverMt();
 
-    btScalar resolveMultipleJointConstraints( const btAlignedObjectArray<int>& consIndices, int batchBegin, int batchEnd, int iteration );
-    btScalar resolveMultipleContactConstraints( const btAlignedObjectArray<int>& consIndices, int batchBegin, int batchEnd );
-    btScalar resolveMultipleContactSplitPenetrationImpulseConstraints( const btAlignedObjectArray<int>& consIndices, int batchBegin, int batchEnd );
-    btScalar resolveMultipleContactFrictionConstraints( const btAlignedObjectArray<int>& consIndices, int batchBegin, int batchEnd );
-    btScalar resolveMultipleContactRollingFrictionConstraints( const btAlignedObjectArray<int>& consIndices, int batchBegin, int batchEnd );
-    btScalar resolveMultipleContactConstraintsInterleaved( const btAlignedObjectArray<int>& contactIndices, int batchBegin, int batchEnd );
+	btScalar resolveMultipleJointConstraints(const btAlignedObjectArray<int>& consIndices, int batchBegin, int batchEnd, int iteration);
+	btScalar resolveMultipleContactConstraints(const btAlignedObjectArray<int>& consIndices, int batchBegin, int batchEnd);
+	btScalar resolveMultipleContactSplitPenetrationImpulseConstraints(const btAlignedObjectArray<int>& consIndices, int batchBegin, int batchEnd);
+	btScalar resolveMultipleContactFrictionConstraints(const btAlignedObjectArray<int>& consIndices, int batchBegin, int batchEnd);
+	btScalar resolveMultipleContactRollingFrictionConstraints(const btAlignedObjectArray<int>& consIndices, int batchBegin, int batchEnd);
+	btScalar resolveMultipleContactConstraintsInterleaved(const btAlignedObjectArray<int>& contactIndices, int batchBegin, int batchEnd);
 
-    void internalCollectContactManifoldCachedInfo(btContactManifoldCachedInfo* cachedInfoArray, btPersistentManifold** manifoldPtr, int numManifolds, const btContactSolverInfo& infoGlobal);
-    void internalAllocContactConstraints(const btContactManifoldCachedInfo* cachedInfoArray, int numManifolds);
-    void internalSetupContactConstraints(int iContactConstraint, const btContactSolverInfo& infoGlobal);
-    void internalConvertBodies(btCollisionObject** bodies, int iBegin, int iEnd, const btContactSolverInfo& infoGlobal);
-    void internalWriteBackContacts(int iBegin, int iEnd, const btContactSolverInfo& infoGlobal);
-    void internalWriteBackJoints(int iBegin, int iEnd, const btContactSolverInfo& infoGlobal);
-    void internalWriteBackBodies(int iBegin, int iEnd, const btContactSolverInfo& infoGlobal);
+	void internalCollectContactManifoldCachedInfo(btContactManifoldCachedInfo * cachedInfoArray, btPersistentManifold * *manifoldPtr, int numManifolds, const btContactSolverInfo& infoGlobal);
+	void internalAllocContactConstraints(const btContactManifoldCachedInfo* cachedInfoArray, int numManifolds);
+	void internalSetupContactConstraints(int iContactConstraint, const btContactSolverInfo& infoGlobal);
+	void internalConvertBodies(btCollisionObject * *bodies, int iBegin, int iEnd, const btContactSolverInfo& infoGlobal);
+	void internalWriteBackContacts(int iBegin, int iEnd, const btContactSolverInfo& infoGlobal);
+	void internalWriteBackJoints(int iBegin, int iEnd, const btContactSolverInfo& infoGlobal);
+	void internalWriteBackBodies(int iBegin, int iEnd, const btContactSolverInfo& infoGlobal);
 };
 
-
-
-
-#endif //BT_SEQUENTIAL_IMPULSE_CONSTRAINT_SOLVER_MT_H
-
+#endif  //BT_SEQUENTIAL_IMPULSE_CONSTRAINT_SOLVER_MT_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSliderConstraint.cpp b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSliderConstraint.cpp
index d63cef03162..cac5302a73e 100755
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSliderConstraint.cpp
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSliderConstraint.cpp
@@ -18,8 +18,6 @@ Added by Roman Ponomarev (rponom@gmail.com)
 April 04, 2008
 */
 
-
-
 #include "btSliderConstraint.h"
 #include "BulletDynamics/Dynamics/btRigidBody.h"
 #include "LinearMath/btTransformUtil.h"
@@ -29,10 +27,10 @@ April 04, 2008
 
 void btSliderConstraint::initParams()
 {
-    m_lowerLinLimit = btScalar(1.0);
-    m_upperLinLimit = btScalar(-1.0);
-    m_lowerAngLimit = btScalar(0.);
-    m_upperAngLimit = btScalar(0.);
+	m_lowerLinLimit = btScalar(1.0);
+	m_upperLinLimit = btScalar(-1.0);
+	m_lowerAngLimit = btScalar(0.);
+	m_upperAngLimit = btScalar(0.);
 	m_softnessDirLin = SLIDER_CONSTRAINT_DEF_SOFTNESS;
 	m_restitutionDirLin = SLIDER_CONSTRAINT_DEF_RESTITUTION;
 	m_dampingDirLin = btScalar(0.);
@@ -59,13 +57,13 @@ void btSliderConstraint::initParams()
 	m_cfmLimAng = SLIDER_CONSTRAINT_DEF_CFM;
 
 	m_poweredLinMotor = false;
-    m_targetLinMotorVelocity = btScalar(0.);
-    m_maxLinMotorForce = btScalar(0.);
+	m_targetLinMotorVelocity = btScalar(0.);
+	m_maxLinMotorForce = btScalar(0.);
 	m_accumulatedLinMotorImpulse = btScalar(0.0);
 
 	m_poweredAngMotor = false;
-    m_targetAngMotorVelocity = btScalar(0.);
-    m_maxAngMotorForce = btScalar(0.);
+	m_targetAngMotorVelocity = btScalar(0.);
+	m_maxAngMotorForce = btScalar(0.);
 	m_accumulatedAngMotorImpulse = btScalar(0.0);
 
 	m_flags = 0;
@@ -73,43 +71,32 @@ void btSliderConstraint::initParams()
 
 	m_useOffsetForConstraintFrame = USE_OFFSET_FOR_CONSTANT_FRAME;
 
-	calculateTransforms(m_rbA.getCenterOfMassTransform(),m_rbB.getCenterOfMassTransform());
+	calculateTransforms(m_rbA.getCenterOfMassTransform(), m_rbB.getCenterOfMassTransform());
 }
 
-
-
-
-
 btSliderConstraint::btSliderConstraint(btRigidBody& rbA, btRigidBody& rbB, const btTransform& frameInA, const btTransform& frameInB, bool useLinearReferenceFrameA)
-        : btTypedConstraint(SLIDER_CONSTRAINT_TYPE, rbA, rbB),
-		m_useSolveConstraintObsolete(false),
-		m_frameInA(frameInA),
-        m_frameInB(frameInB),
-		m_useLinearReferenceFrameA(useLinearReferenceFrameA)
+	: btTypedConstraint(SLIDER_CONSTRAINT_TYPE, rbA, rbB),
+	  m_useSolveConstraintObsolete(false),
+	  m_frameInA(frameInA),
+	  m_frameInB(frameInB),
+	  m_useLinearReferenceFrameA(useLinearReferenceFrameA)
 {
 	initParams();
 }
 
-
-
 btSliderConstraint::btSliderConstraint(btRigidBody& rbB, const btTransform& frameInB, bool useLinearReferenceFrameA)
-        : btTypedConstraint(SLIDER_CONSTRAINT_TYPE, getFixedBody(), rbB),
-		m_useSolveConstraintObsolete(false),
-		m_frameInB(frameInB),
-		m_useLinearReferenceFrameA(useLinearReferenceFrameA)
+	: btTypedConstraint(SLIDER_CONSTRAINT_TYPE, getFixedBody(), rbB),
+	  m_useSolveConstraintObsolete(false),
+	  m_frameInB(frameInB),
+	  m_useLinearReferenceFrameA(useLinearReferenceFrameA)
 {
 	///not providing rigidbody A means implicitly using worldspace for body A
 	m_frameInA = rbB.getCenterOfMassTransform() * m_frameInB;
-//	m_frameInA.getOrigin() = m_rbA.getCenterOfMassTransform()(m_frameInA.getOrigin());
+	//	m_frameInA.getOrigin() = m_rbA.getCenterOfMassTransform()(m_frameInA.getOrigin());
 
 	initParams();
 }
 
-
-
-
-
-
 void btSliderConstraint::getInfo1(btConstraintInfo1* info)
 {
 	if (m_useSolveConstraintObsolete)
@@ -119,46 +106,39 @@ void btSliderConstraint::getInfo1(btConstraintInfo1* info)
 	}
 	else
 	{
-		info->m_numConstraintRows = 4; // Fixed 2 linear + 2 angular
-		info->nub = 2; 
+		info->m_numConstraintRows = 4;  // Fixed 2 linear + 2 angular
+		info->nub = 2;
 		//prepare constraint
-		calculateTransforms(m_rbA.getCenterOfMassTransform(),m_rbB.getCenterOfMassTransform());
+		calculateTransforms(m_rbA.getCenterOfMassTransform(), m_rbB.getCenterOfMassTransform());
 		testAngLimits();
 		testLinLimits();
-		if(getSolveLinLimit() || getPoweredLinMotor())
+		if (getSolveLinLimit() || getPoweredLinMotor())
 		{
-			info->m_numConstraintRows++; // limit 3rd linear as well
-			info->nub--; 
+			info->m_numConstraintRows++;  // limit 3rd linear as well
+			info->nub--;
 		}
-		if(getSolveAngLimit() || getPoweredAngMotor())
+		if (getSolveAngLimit() || getPoweredAngMotor())
 		{
-			info->m_numConstraintRows++; // limit 3rd angular as well
-			info->nub--; 
+			info->m_numConstraintRows++;  // limit 3rd angular as well
+			info->nub--;
 		}
 	}
 }
 
 void btSliderConstraint::getInfo1NonVirtual(btConstraintInfo1* info)
 {
-
-	info->m_numConstraintRows = 6; // Fixed 2 linear + 2 angular + 1 limit (even if not used)
-	info->nub = 0; 
+	info->m_numConstraintRows = 6;  // Fixed 2 linear + 2 angular + 1 limit (even if not used)
+	info->nub = 0;
 }
 
 void btSliderConstraint::getInfo2(btConstraintInfo2* info)
 {
-	getInfo2NonVirtual(info,m_rbA.getCenterOfMassTransform(),m_rbB.getCenterOfMassTransform(), m_rbA.getLinearVelocity(),m_rbB.getLinearVelocity(), m_rbA.getInvMass(),m_rbB.getInvMass());
+	getInfo2NonVirtual(info, m_rbA.getCenterOfMassTransform(), m_rbB.getCenterOfMassTransform(), m_rbA.getLinearVelocity(), m_rbB.getLinearVelocity(), m_rbA.getInvMass(), m_rbB.getInvMass());
 }
 
-
-
-
-
-
-
-void btSliderConstraint::calculateTransforms(const btTransform& transA,const btTransform& transB)
+void btSliderConstraint::calculateTransforms(const btTransform& transA, const btTransform& transB)
 {
-	if(m_useLinearReferenceFrameA || (!m_useSolveConstraintObsolete))
+	if (m_useLinearReferenceFrameA || (!m_useSolveConstraintObsolete))
 	{
 		m_calculatedTransformA = transA * m_frameInA;
 		m_calculatedTransformB = transB * m_frameInB;
@@ -170,8 +150,8 @@ void btSliderConstraint::calculateTransforms(const btTransform& transA,const btT
 	}
 	m_realPivotAInW = m_calculatedTransformA.getOrigin();
 	m_realPivotBInW = m_calculatedTransformB.getOrigin();
-	m_sliderAxis = m_calculatedTransformA.getBasis().getColumn(0); // along X
-	if(m_useLinearReferenceFrameA || m_useSolveConstraintObsolete)
+	m_sliderAxis = m_calculatedTransformA.getBasis().getColumn(0);  // along X
+	if (m_useLinearReferenceFrameA || m_useSolveConstraintObsolete)
 	{
 		m_delta = m_realPivotBInW - m_realPivotAInW;
 	}
@@ -180,30 +160,28 @@ void btSliderConstraint::calculateTransforms(const btTransform& transA,const btT
 		m_delta = m_realPivotAInW - m_realPivotBInW;
 	}
 	m_projPivotInW = m_realPivotAInW + m_sliderAxis.dot(m_delta) * m_sliderAxis;
-    btVector3 normalWorld;
-    int i;
-    //linear part
-    for(i = 0; i < 3; i++)
-    {
+	btVector3 normalWorld;
+	int i;
+	//linear part
+	for (i = 0; i < 3; i++)
+	{
 		normalWorld = m_calculatedTransformA.getBasis().getColumn(i);
 		m_depth[i] = m_delta.dot(normalWorld);
-    }
+	}
 }
- 
-
 
 void btSliderConstraint::testLinLimits(void)
 {
 	m_solveLinLim = false;
 	m_linPos = m_depth[0];
-	if(m_lowerLinLimit <= m_upperLinLimit)
+	if (m_lowerLinLimit <= m_upperLinLimit)
 	{
-		if(m_depth[0] > m_upperLinLimit)
+		if (m_depth[0] > m_upperLinLimit)
 		{
 			m_depth[0] -= m_upperLinLimit;
 			m_solveLinLim = true;
 		}
-		else if(m_depth[0] < m_lowerLinLimit)
+		else if (m_depth[0] < m_lowerLinLimit)
 		{
 			m_depth[0] -= m_lowerLinLimit;
 			m_solveLinLim = true;
@@ -219,27 +197,25 @@ void btSliderConstraint::testLinLimits(void)
 	}
 }
 
-
-
 void btSliderConstraint::testAngLimits(void)
 {
 	m_angDepth = btScalar(0.);
 	m_solveAngLim = false;
-	if(m_lowerAngLimit <= m_upperAngLimit)
+	if (m_lowerAngLimit <= m_upperAngLimit)
 	{
 		const btVector3 axisA0 = m_calculatedTransformA.getBasis().getColumn(1);
 		const btVector3 axisA1 = m_calculatedTransformA.getBasis().getColumn(2);
 		const btVector3 axisB0 = m_calculatedTransformB.getBasis().getColumn(1);
-//		btScalar rot = btAtan2Fast(axisB0.dot(axisA1), axisB0.dot(axisA0));  
-		btScalar rot = btAtan2(axisB0.dot(axisA1), axisB0.dot(axisA0));  
+		//		btScalar rot = btAtan2Fast(axisB0.dot(axisA1), axisB0.dot(axisA0));
+		btScalar rot = btAtan2(axisB0.dot(axisA1), axisB0.dot(axisA0));
 		rot = btAdjustAngleToLimits(rot, m_lowerAngLimit, m_upperAngLimit);
 		m_angPos = rot;
-		if(rot < m_lowerAngLimit)
+		if (rot < m_lowerAngLimit)
 		{
 			m_angDepth = rot - m_lowerAngLimit;
 			m_solveAngLim = true;
-		} 
-		else if(rot > m_upperAngLimit)
+		}
+		else if (rot > m_upperAngLimit)
 		{
 			m_angDepth = rot - m_upperAngLimit;
 			m_solveAngLim = true;
@@ -255,8 +231,6 @@ btVector3 btSliderConstraint::getAncorInA(void)
 	return ancorInA;
 }
 
-
-
 btVector3 btSliderConstraint::getAncorInB(void)
 {
 	btVector3 ancorInB;
@@ -264,17 +238,16 @@ btVector3 btSliderConstraint::getAncorInB(void)
 	return ancorInB;
 }
 
-
-void btSliderConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTransform& transA,const btTransform& transB, const btVector3& linVelA,const btVector3& linVelB, btScalar rbAinvMass,btScalar rbBinvMass  )
+void btSliderConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTransform& transA, const btTransform& transB, const btVector3& linVelA, const btVector3& linVelB, btScalar rbAinvMass, btScalar rbBinvMass)
 {
 	const btTransform& trA = getCalculatedTransformA();
 	const btTransform& trB = getCalculatedTransformB();
-	
+
 	btAssert(!m_useSolveConstraintObsolete);
 	int i, s = info->rowskip;
-	
+
 	btScalar signFact = m_useLinearReferenceFrameA ? btScalar(1.0f) : btScalar(-1.0f);
-	
+
 	// difference between frames in WCS
 	btVector3 ofs = trB.getOrigin() - trA.getOrigin();
 	// now get weight factors depending on masses
@@ -283,11 +256,11 @@ void btSliderConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTra
 	bool hasStaticBody = (miA < SIMD_EPSILON) || (miB < SIMD_EPSILON);
 	btScalar miS = miA + miB;
 	btScalar factA, factB;
-	if(miS > btScalar(0.f))
+	if (miS > btScalar(0.f))
 	{
 		factA = miB / miS;
 	}
-	else 
+	else
 	{
 		factA = btScalar(0.5f);
 	}
@@ -295,17 +268,17 @@ void btSliderConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTra
 	btVector3 ax1, p, q;
 	btVector3 ax1A = trA.getBasis().getColumn(0);
 	btVector3 ax1B = trB.getBasis().getColumn(0);
-	if(m_useOffsetForConstraintFrame)
+	if (m_useOffsetForConstraintFrame)
 	{
 		// get the desired direction of slider axis
 		// as weighted sum of X-orthos of frameA and frameB in WCS
 		ax1 = ax1A * factA + ax1B * factB;
 		ax1.normalize();
 		// construct two orthos to slider axis
-		btPlaneSpace1 (ax1, p, q);
+		btPlaneSpace1(ax1, p, q);
 	}
 	else
-	{ // old way - use frameA
+	{  // old way - use frameA
 		ax1 = trA.getBasis().getColumn(0);
 		// get 2 orthos to slider axis (Y, Z)
 		p = trA.getBasis().getColumn(1);
@@ -322,16 +295,16 @@ void btSliderConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTra
 	info->m_J1angularAxis[0] = p[0];
 	info->m_J1angularAxis[1] = p[1];
 	info->m_J1angularAxis[2] = p[2];
-	info->m_J1angularAxis[s+0] = q[0];
-	info->m_J1angularAxis[s+1] = q[1];
-	info->m_J1angularAxis[s+2] = q[2];
+	info->m_J1angularAxis[s + 0] = q[0];
+	info->m_J1angularAxis[s + 1] = q[1];
+	info->m_J1angularAxis[s + 2] = q[2];
 
 	info->m_J2angularAxis[0] = -p[0];
 	info->m_J2angularAxis[1] = -p[1];
 	info->m_J2angularAxis[2] = -p[2];
-	info->m_J2angularAxis[s+0] = -q[0];
-	info->m_J2angularAxis[s+1] = -q[1];
-	info->m_J2angularAxis[s+2] = -q[2];
+	info->m_J2angularAxis[s + 0] = -q[0];
+	info->m_J2angularAxis[s + 1] = -q[1];
+	info->m_J2angularAxis[s + 2] = -q[2];
 	// compute the right hand side of the constraint equation. set relative
 	// body velocities along p and q to bring the slider back into alignment.
 	// if ax1A,ax1B are the unit length slider axes as computed from bodyA and
@@ -347,25 +320,25 @@ void btSliderConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTra
 	//    angular_velocity  = (erp*fps) * (ax1 x ax2)
 	// ax1 x ax2 is in the plane space of ax1, so we project the angular
 	// velocity to p and q to find the right hand side.
-//	btScalar k = info->fps * info->erp * getSoftnessOrthoAng();
+	//	btScalar k = info->fps * info->erp * getSoftnessOrthoAng();
 	btScalar currERP = (m_flags & BT_SLIDER_FLAGS_ERP_ORTANG) ? m_softnessOrthoAng : m_softnessOrthoAng * info->erp;
 	btScalar k = info->fps * currERP;
 
 	btVector3 u = ax1A.cross(ax1B);
 	info->m_constraintError[0] = k * u.dot(p);
 	info->m_constraintError[s] = k * u.dot(q);
-	if(m_flags & BT_SLIDER_FLAGS_CFM_ORTANG)
+	if (m_flags & BT_SLIDER_FLAGS_CFM_ORTANG)
 	{
 		info->cfm[0] = m_cfmOrthoAng;
 		info->cfm[s] = m_cfmOrthoAng;
 	}
 
-	int nrow = 1; // last filled row
+	int nrow = 1;  // last filled row
 	int srow;
 	btScalar limit_err;
 	int limit;
 
-	// next two rows. 
+	// next two rows.
 	// we want: velA + wA x relA == velB + wB x relB ... but this would
 	// result in three equations, so we project along two orthos to the slider axis
 
@@ -375,8 +348,8 @@ void btSliderConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTra
 	int s2 = nrow * s;
 	nrow++;
 	int s3 = nrow * s;
-	btVector3 tmpA(0,0,0), tmpB(0,0,0), relA(0,0,0), relB(0,0,0), c(0,0,0);
-	if(m_useOffsetForConstraintFrame)
+	btVector3 tmpA(0, 0, 0), tmpB(0, 0, 0), relA(0, 0, 0), relB(0, 0, 0), c(0, 0, 0);
+	if (m_useOffsetForConstraintFrame)
 	{
 		// get vector from bodyB to frameB in WCS
 		relB = trB.getOrigin() - bodyB_trans.getOrigin();
@@ -398,7 +371,7 @@ void btSliderConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTra
 		// now choose average ortho to slider axis
 		p = orthoB * factA + orthoA * factB;
 		btScalar len2 = p.length2();
-		if(len2 > SIMD_EPSILON)
+		if (len2 > SIMD_EPSILON)
 		{
 			p /= btSqrt(len2);
 		}
@@ -411,38 +384,38 @@ void btSliderConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTra
 		// fill two rows
 		tmpA = relA.cross(p);
 		tmpB = relB.cross(p);
-		for (i=0; i<3; i++) info->m_J1angularAxis[s2+i] = tmpA[i];
-		for (i=0; i<3; i++) info->m_J2angularAxis[s2+i] = -tmpB[i];
+		for (i = 0; i < 3; i++) info->m_J1angularAxis[s2 + i] = tmpA[i];
+		for (i = 0; i < 3; i++) info->m_J2angularAxis[s2 + i] = -tmpB[i];
 		tmpA = relA.cross(q);
 		tmpB = relB.cross(q);
-		if(hasStaticBody && getSolveAngLimit())
-		{ // to make constraint between static and dynamic objects more rigid
+		if (hasStaticBody && getSolveAngLimit())
+		{  // to make constraint between static and dynamic objects more rigid
 			// remove wA (or wB) from equation if angular limit is hit
 			tmpB *= factB;
 			tmpA *= factA;
 		}
-		for (i=0; i<3; i++) info->m_J1angularAxis[s3+i] = tmpA[i];
-		for (i=0; i<3; i++) info->m_J2angularAxis[s3+i] = -tmpB[i];
-		for (i=0; i<3; i++) info->m_J1linearAxis[s2+i] = p[i];
-		for (i=0; i<3; i++) info->m_J1linearAxis[s3+i] = q[i];
-		for (i=0; i<3; i++) info->m_J2linearAxis[s2+i] = -p[i];
-		for (i=0; i<3; i++) info->m_J2linearAxis[s3+i] = -q[i];
+		for (i = 0; i < 3; i++) info->m_J1angularAxis[s3 + i] = tmpA[i];
+		for (i = 0; i < 3; i++) info->m_J2angularAxis[s3 + i] = -tmpB[i];
+		for (i = 0; i < 3; i++) info->m_J1linearAxis[s2 + i] = p[i];
+		for (i = 0; i < 3; i++) info->m_J1linearAxis[s3 + i] = q[i];
+		for (i = 0; i < 3; i++) info->m_J2linearAxis[s2 + i] = -p[i];
+		for (i = 0; i < 3; i++) info->m_J2linearAxis[s3 + i] = -q[i];
 	}
 	else
-	{	// old way - maybe incorrect if bodies are not on the slider axis
+	{  // old way - maybe incorrect if bodies are not on the slider axis
 		// see discussion "Bug in slider constraint" http://bulletphysics.org/Bullet/phpBB3/viewtopic.php?f=9&t=4024&start=0
 		c = bodyB_trans.getOrigin() - bodyA_trans.getOrigin();
 		btVector3 tmp = c.cross(p);
-		for (i=0; i<3; i++) info->m_J1angularAxis[s2+i] = factA*tmp[i];
-		for (i=0; i<3; i++) info->m_J2angularAxis[s2+i] = factB*tmp[i];
+		for (i = 0; i < 3; i++) info->m_J1angularAxis[s2 + i] = factA * tmp[i];
+		for (i = 0; i < 3; i++) info->m_J2angularAxis[s2 + i] = factB * tmp[i];
 		tmp = c.cross(q);
-		for (i=0; i<3; i++) info->m_J1angularAxis[s3+i] = factA*tmp[i];
-		for (i=0; i<3; i++) info->m_J2angularAxis[s3+i] = factB*tmp[i];
+		for (i = 0; i < 3; i++) info->m_J1angularAxis[s3 + i] = factA * tmp[i];
+		for (i = 0; i < 3; i++) info->m_J2angularAxis[s3 + i] = factB * tmp[i];
 
-		for (i=0; i<3; i++) info->m_J1linearAxis[s2+i] = p[i];
-		for (i=0; i<3; i++) info->m_J1linearAxis[s3+i] = q[i];
-		for (i=0; i<3; i++) info->m_J2linearAxis[s2+i] = -p[i];
-		for (i=0; i<3; i++) info->m_J2linearAxis[s3+i] = -q[i];
+		for (i = 0; i < 3; i++) info->m_J1linearAxis[s2 + i] = p[i];
+		for (i = 0; i < 3; i++) info->m_J1linearAxis[s3 + i] = q[i];
+		for (i = 0; i < 3; i++) info->m_J2linearAxis[s2 + i] = -p[i];
+		for (i = 0; i < 3; i++) info->m_J2linearAxis[s3 + i] = -q[i];
 	}
 	// compute two elements of right hand side
 
@@ -454,19 +427,18 @@ void btSliderConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTra
 	info->m_constraintError[s2] = rhs;
 	rhs = k * q.dot(ofs);
 	info->m_constraintError[s3] = rhs;
-	if(m_flags & BT_SLIDER_FLAGS_CFM_ORTLIN)
+	if (m_flags & BT_SLIDER_FLAGS_CFM_ORTLIN)
 	{
 		info->cfm[s2] = m_cfmOrthoLin;
 		info->cfm[s3] = m_cfmOrthoLin;
 	}
 
-
 	// check linear limits
 	limit_err = btScalar(0.0);
 	limit = 0;
-	if(getSolveLinLimit())
+	if (getSolveLinLimit())
 	{
-		limit_err = getLinDepth() *  signFact;
+		limit_err = getLinDepth() * signFact;
 		limit = (limit_err > btScalar(0.0)) ? 2 : 1;
 	}
 	bool powered = getPoweredLinMotor();
@@ -475,12 +447,12 @@ void btSliderConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTra
 	{
 		nrow++;
 		srow = nrow * info->rowskip;
-		info->m_J1linearAxis[srow+0] = ax1[0];
-		info->m_J1linearAxis[srow+1] = ax1[1];
-		info->m_J1linearAxis[srow+2] = ax1[2];
-		info->m_J2linearAxis[srow+0] = -ax1[0];
-		info->m_J2linearAxis[srow+1] = -ax1[1];
-		info->m_J2linearAxis[srow+2] = -ax1[2];
+		info->m_J1linearAxis[srow + 0] = ax1[0];
+		info->m_J1linearAxis[srow + 1] = ax1[1];
+		info->m_J1linearAxis[srow + 2] = ax1[2];
+		info->m_J2linearAxis[srow + 0] = -ax1[0];
+		info->m_J2linearAxis[srow + 1] = -ax1[1];
+		info->m_J2linearAxis[srow + 2] = -ax1[2];
 		// linear torque decoupling step:
 		//
 		// we have to be careful that the linear constraint forces (+/- ax1) applied to the two bodies
@@ -488,36 +460,36 @@ void btSliderConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTra
 		// constraint force is applied at must lie along the same ax1 axis.
 		// a torque couple will result in limited slider-jointed free
 		// bodies from gaining angular momentum.
-		if(m_useOffsetForConstraintFrame)
+		if (m_useOffsetForConstraintFrame)
 		{
 			// this is needed only when bodyA and bodyB are both dynamic.
-			if(!hasStaticBody)
+			if (!hasStaticBody)
 			{
 				tmpA = relA.cross(ax1);
 				tmpB = relB.cross(ax1);
-				info->m_J1angularAxis[srow+0] = tmpA[0];
-				info->m_J1angularAxis[srow+1] = tmpA[1];
-				info->m_J1angularAxis[srow+2] = tmpA[2];
-				info->m_J2angularAxis[srow+0] = -tmpB[0];
-				info->m_J2angularAxis[srow+1] = -tmpB[1];
-				info->m_J2angularAxis[srow+2] = -tmpB[2];
+				info->m_J1angularAxis[srow + 0] = tmpA[0];
+				info->m_J1angularAxis[srow + 1] = tmpA[1];
+				info->m_J1angularAxis[srow + 2] = tmpA[2];
+				info->m_J2angularAxis[srow + 0] = -tmpB[0];
+				info->m_J2angularAxis[srow + 1] = -tmpB[1];
+				info->m_J2angularAxis[srow + 2] = -tmpB[2];
 			}
 		}
 		else
-		{ // The old way. May be incorrect if bodies are not on the slider axis
-			btVector3 ltd;	// Linear Torque Decoupling vector (a torque)
+		{                   // The old way. May be incorrect if bodies are not on the slider axis
+			btVector3 ltd;  // Linear Torque Decoupling vector (a torque)
 			ltd = c.cross(ax1);
-			info->m_J1angularAxis[srow+0] = factA*ltd[0];
-			info->m_J1angularAxis[srow+1] = factA*ltd[1];
-			info->m_J1angularAxis[srow+2] = factA*ltd[2];
-			info->m_J2angularAxis[srow+0] = factB*ltd[0];
-			info->m_J2angularAxis[srow+1] = factB*ltd[1];
-			info->m_J2angularAxis[srow+2] = factB*ltd[2];
+			info->m_J1angularAxis[srow + 0] = factA * ltd[0];
+			info->m_J1angularAxis[srow + 1] = factA * ltd[1];
+			info->m_J1angularAxis[srow + 2] = factA * ltd[2];
+			info->m_J2angularAxis[srow + 0] = factB * ltd[0];
+			info->m_J2angularAxis[srow + 1] = factB * ltd[1];
+			info->m_J2angularAxis[srow + 2] = factB * ltd[2];
 		}
 		// right-hand part
 		btScalar lostop = getLowerLinLimit();
 		btScalar histop = getUpperLinLimit();
-		if(limit && (lostop == histop))
+		if (limit && (lostop == histop))
 		{  // the joint motor is ineffective
 			powered = false;
 		}
@@ -525,9 +497,9 @@ void btSliderConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTra
 		info->m_lowerLimit[srow] = 0.;
 		info->m_upperLimit[srow] = 0.;
 		currERP = (m_flags & BT_SLIDER_FLAGS_ERP_LIMLIN) ? m_softnessLimLin : info->erp;
-		if(powered)
+		if (powered)
 		{
-			if(m_flags & BT_SLIDER_FLAGS_CFM_DIRLIN)
+			if (m_flags & BT_SLIDER_FLAGS_CFM_DIRLIN)
 			{
 				info->cfm[srow] = m_cfmDirLin;
 			}
@@ -537,41 +509,41 @@ void btSliderConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTra
 			info->m_lowerLimit[srow] += -getMaxLinMotorForce() / info->fps;
 			info->m_upperLimit[srow] += getMaxLinMotorForce() / info->fps;
 		}
-		if(limit)
+		if (limit)
 		{
 			k = info->fps * currERP;
 			info->m_constraintError[srow] += k * limit_err;
-			if(m_flags & BT_SLIDER_FLAGS_CFM_LIMLIN)
+			if (m_flags & BT_SLIDER_FLAGS_CFM_LIMLIN)
 			{
 				info->cfm[srow] = m_cfmLimLin;
 			}
-			if(lostop == histop) 
-			{	// limited low and high simultaneously
+			if (lostop == histop)
+			{  // limited low and high simultaneously
 				info->m_lowerLimit[srow] = -SIMD_INFINITY;
 				info->m_upperLimit[srow] = SIMD_INFINITY;
 			}
-			else if(limit == 1) 
-			{ // low limit
+			else if (limit == 1)
+			{  // low limit
 				info->m_lowerLimit[srow] = -SIMD_INFINITY;
 				info->m_upperLimit[srow] = 0;
 			}
-			else 
-			{ // high limit
+			else
+			{  // high limit
 				info->m_lowerLimit[srow] = 0;
 				info->m_upperLimit[srow] = SIMD_INFINITY;
 			}
 			// bounce (we'll use slider parameter abs(1.0 - m_dampingLimLin) for that)
 			btScalar bounce = btFabs(btScalar(1.0) - getDampingLimLin());
-			if(bounce > btScalar(0.0))
+			if (bounce > btScalar(0.0))
 			{
 				btScalar vel = linVelA.dot(ax1);
 				vel -= linVelB.dot(ax1);
 				vel *= signFact;
 				// only apply bounce if the velocity is incoming, and if the
 				// resulting c[] exceeds what we already have.
-				if(limit == 1)
-				{	// low limit
-					if(vel < 0)
+				if (limit == 1)
+				{  // low limit
+					if (vel < 0)
 					{
 						btScalar newc = -bounce * vel;
 						if (newc > info->m_constraintError[srow])
@@ -581,11 +553,11 @@ void btSliderConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTra
 					}
 				}
 				else
-				{ // high limit - all those computations are reversed
-					if(vel > 0)
+				{  // high limit - all those computations are reversed
+					if (vel > 0)
 					{
 						btScalar newc = -bounce * vel;
-						if(newc < info->m_constraintError[srow]) 
+						if (newc < info->m_constraintError[srow])
 						{
 							info->m_constraintError[srow] = newc;
 						}
@@ -593,40 +565,40 @@ void btSliderConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTra
 				}
 			}
 			info->m_constraintError[srow] *= getSoftnessLimLin();
-		} // if(limit)
-	} // if linear limit
+		}  // if(limit)
+	}      // if linear limit
 	// check angular limits
 	limit_err = btScalar(0.0);
 	limit = 0;
-	if(getSolveAngLimit())
+	if (getSolveAngLimit())
 	{
 		limit_err = getAngDepth();
 		limit = (limit_err > btScalar(0.0)) ? 1 : 2;
 	}
 	// if the slider has joint limits, add in the extra row
 	powered = getPoweredAngMotor();
-	if(limit || powered)
+	if (limit || powered)
 	{
 		nrow++;
 		srow = nrow * info->rowskip;
-		info->m_J1angularAxis[srow+0] = ax1[0];
-		info->m_J1angularAxis[srow+1] = ax1[1];
-		info->m_J1angularAxis[srow+2] = ax1[2];
+		info->m_J1angularAxis[srow + 0] = ax1[0];
+		info->m_J1angularAxis[srow + 1] = ax1[1];
+		info->m_J1angularAxis[srow + 2] = ax1[2];
 
-		info->m_J2angularAxis[srow+0] = -ax1[0];
-		info->m_J2angularAxis[srow+1] = -ax1[1];
-		info->m_J2angularAxis[srow+2] = -ax1[2];
+		info->m_J2angularAxis[srow + 0] = -ax1[0];
+		info->m_J2angularAxis[srow + 1] = -ax1[1];
+		info->m_J2angularAxis[srow + 2] = -ax1[2];
 
 		btScalar lostop = getLowerAngLimit();
 		btScalar histop = getUpperAngLimit();
-		if(limit && (lostop == histop))
+		if (limit && (lostop == histop))
 		{  // the joint motor is ineffective
 			powered = false;
 		}
 		currERP = (m_flags & BT_SLIDER_FLAGS_ERP_LIMANG) ? m_softnessLimAng : info->erp;
-		if(powered)
+		if (powered)
 		{
-			if(m_flags & BT_SLIDER_FLAGS_CFM_DIRANG)
+			if (m_flags & BT_SLIDER_FLAGS_CFM_DIRANG)
 			{
 				info->cfm[srow] = m_cfmDirAng;
 			}
@@ -635,55 +607,55 @@ void btSliderConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTra
 			info->m_lowerLimit[srow] = -getMaxAngMotorForce() / info->fps;
 			info->m_upperLimit[srow] = getMaxAngMotorForce() / info->fps;
 		}
-		if(limit)
+		if (limit)
 		{
 			k = info->fps * currERP;
 			info->m_constraintError[srow] += k * limit_err;
-			if(m_flags & BT_SLIDER_FLAGS_CFM_LIMANG)
+			if (m_flags & BT_SLIDER_FLAGS_CFM_LIMANG)
 			{
 				info->cfm[srow] = m_cfmLimAng;
 			}
-			if(lostop == histop) 
+			if (lostop == histop)
 			{
 				// limited low and high simultaneously
 				info->m_lowerLimit[srow] = -SIMD_INFINITY;
 				info->m_upperLimit[srow] = SIMD_INFINITY;
 			}
-			else if(limit == 1) 
-			{ // low limit
+			else if (limit == 1)
+			{  // low limit
 				info->m_lowerLimit[srow] = 0;
 				info->m_upperLimit[srow] = SIMD_INFINITY;
 			}
-			else 
-			{ // high limit
+			else
+			{  // high limit
 				info->m_lowerLimit[srow] = -SIMD_INFINITY;
 				info->m_upperLimit[srow] = 0;
 			}
 			// bounce (we'll use slider parameter abs(1.0 - m_dampingLimAng) for that)
 			btScalar bounce = btFabs(btScalar(1.0) - getDampingLimAng());
-			if(bounce > btScalar(0.0))
+			if (bounce > btScalar(0.0))
 			{
 				btScalar vel = m_rbA.getAngularVelocity().dot(ax1);
 				vel -= m_rbB.getAngularVelocity().dot(ax1);
 				// only apply bounce if the velocity is incoming, and if the
 				// resulting c[] exceeds what we already have.
-				if(limit == 1)
-				{	// low limit
-					if(vel < 0)
+				if (limit == 1)
+				{  // low limit
+					if (vel < 0)
 					{
 						btScalar newc = -bounce * vel;
-						if(newc > info->m_constraintError[srow])
+						if (newc > info->m_constraintError[srow])
 						{
 							info->m_constraintError[srow] = newc;
 						}
 					}
 				}
 				else
-				{	// high limit - all those computations are reversed
-					if(vel > 0)
+				{  // high limit - all those computations are reversed
+					if (vel > 0)
 					{
 						btScalar newc = -bounce * vel;
-						if(newc < info->m_constraintError[srow])
+						if (newc < info->m_constraintError[srow])
 						{
 							info->m_constraintError[srow] = newc;
 						}
@@ -691,165 +663,161 @@ void btSliderConstraint::getInfo2NonVirtual(btConstraintInfo2* info, const btTra
 				}
 			}
 			info->m_constraintError[srow] *= getSoftnessLimAng();
-		} // if(limit)
-	} // if angular limit or powered
+		}  // if(limit)
+	}      // if angular limit or powered
 }
 
-
-///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5). 
+///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5).
 ///If no axis is provided, it uses the default axis for this constraint.
 void btSliderConstraint::setParam(int num, btScalar value, int axis)
 {
-	switch(num)
+	switch (num)
 	{
-	case BT_CONSTRAINT_STOP_ERP :
-		if(axis < 1)
-		{
-			m_softnessLimLin = value;
-			m_flags |= BT_SLIDER_FLAGS_ERP_LIMLIN;
-		}
-		else if(axis < 3)
-		{
-			m_softnessOrthoLin = value;
-			m_flags |= BT_SLIDER_FLAGS_ERP_ORTLIN;
-		}
-		else if(axis == 3)
-		{
-			m_softnessLimAng = value;
-			m_flags |= BT_SLIDER_FLAGS_ERP_LIMANG;
-		}
-		else if(axis < 6)
-		{
-			m_softnessOrthoAng = value;
-			m_flags |= BT_SLIDER_FLAGS_ERP_ORTANG;
-		}
-		else
-		{
-			btAssertConstrParams(0);
-		}
-		break;
-	case BT_CONSTRAINT_CFM :
-		if(axis < 1)
-		{
-			m_cfmDirLin = value;
-			m_flags |= BT_SLIDER_FLAGS_CFM_DIRLIN;
-		}
-		else if(axis == 3)
-		{
-			m_cfmDirAng = value;
-			m_flags |= BT_SLIDER_FLAGS_CFM_DIRANG;
-		}
-		else
-		{
-			btAssertConstrParams(0);
-		}
-		break;
-	case BT_CONSTRAINT_STOP_CFM :
-		if(axis < 1)
-		{
-			m_cfmLimLin = value;
-			m_flags |= BT_SLIDER_FLAGS_CFM_LIMLIN;
-		}
-		else if(axis < 3)
-		{
-			m_cfmOrthoLin = value;
-			m_flags |= BT_SLIDER_FLAGS_CFM_ORTLIN;
-		}
-		else if(axis == 3)
-		{
-			m_cfmLimAng = value;
-			m_flags |= BT_SLIDER_FLAGS_CFM_LIMANG;
-		}
-		else if(axis < 6)
-		{
-			m_cfmOrthoAng = value;
-			m_flags |= BT_SLIDER_FLAGS_CFM_ORTANG;
-		}
-		else
-		{
-			btAssertConstrParams(0);
-		}
-		break;
+		case BT_CONSTRAINT_STOP_ERP:
+			if (axis < 1)
+			{
+				m_softnessLimLin = value;
+				m_flags |= BT_SLIDER_FLAGS_ERP_LIMLIN;
+			}
+			else if (axis < 3)
+			{
+				m_softnessOrthoLin = value;
+				m_flags |= BT_SLIDER_FLAGS_ERP_ORTLIN;
+			}
+			else if (axis == 3)
+			{
+				m_softnessLimAng = value;
+				m_flags |= BT_SLIDER_FLAGS_ERP_LIMANG;
+			}
+			else if (axis < 6)
+			{
+				m_softnessOrthoAng = value;
+				m_flags |= BT_SLIDER_FLAGS_ERP_ORTANG;
+			}
+			else
+			{
+				btAssertConstrParams(0);
+			}
+			break;
+		case BT_CONSTRAINT_CFM:
+			if (axis < 1)
+			{
+				m_cfmDirLin = value;
+				m_flags |= BT_SLIDER_FLAGS_CFM_DIRLIN;
+			}
+			else if (axis == 3)
+			{
+				m_cfmDirAng = value;
+				m_flags |= BT_SLIDER_FLAGS_CFM_DIRANG;
+			}
+			else
+			{
+				btAssertConstrParams(0);
+			}
+			break;
+		case BT_CONSTRAINT_STOP_CFM:
+			if (axis < 1)
+			{
+				m_cfmLimLin = value;
+				m_flags |= BT_SLIDER_FLAGS_CFM_LIMLIN;
+			}
+			else if (axis < 3)
+			{
+				m_cfmOrthoLin = value;
+				m_flags |= BT_SLIDER_FLAGS_CFM_ORTLIN;
+			}
+			else if (axis == 3)
+			{
+				m_cfmLimAng = value;
+				m_flags |= BT_SLIDER_FLAGS_CFM_LIMANG;
+			}
+			else if (axis < 6)
+			{
+				m_cfmOrthoAng = value;
+				m_flags |= BT_SLIDER_FLAGS_CFM_ORTANG;
+			}
+			else
+			{
+				btAssertConstrParams(0);
+			}
+			break;
 	}
 }
 
 ///return the local value of parameter
-btScalar btSliderConstraint::getParam(int num, int axis) const 
+btScalar btSliderConstraint::getParam(int num, int axis) const
 {
 	btScalar retVal(SIMD_INFINITY);
-	switch(num)
+	switch (num)
 	{
-	case BT_CONSTRAINT_STOP_ERP :
-		if(axis < 1)
-		{
-			btAssertConstrParams(m_flags & BT_SLIDER_FLAGS_ERP_LIMLIN);
-			retVal = m_softnessLimLin;
-		}
-		else if(axis < 3)
-		{
-			btAssertConstrParams(m_flags & BT_SLIDER_FLAGS_ERP_ORTLIN);
-			retVal = m_softnessOrthoLin;
-		}
-		else if(axis == 3)
-		{
-			btAssertConstrParams(m_flags & BT_SLIDER_FLAGS_ERP_LIMANG);
-			retVal = m_softnessLimAng;
-		}
-		else if(axis < 6)
-		{
-			btAssertConstrParams(m_flags & BT_SLIDER_FLAGS_ERP_ORTANG);
-			retVal = m_softnessOrthoAng;
-		}
-		else
-		{
-			btAssertConstrParams(0);
-		}
-		break;
-	case BT_CONSTRAINT_CFM :
-		if(axis < 1)
-		{
-			btAssertConstrParams(m_flags & BT_SLIDER_FLAGS_CFM_DIRLIN);
-			retVal = m_cfmDirLin;
-		}
-		else if(axis == 3)
-		{
-			btAssertConstrParams(m_flags & BT_SLIDER_FLAGS_CFM_DIRANG);
-			retVal = m_cfmDirAng;
-		}
-		else
-		{
-			btAssertConstrParams(0);
-		}
-		break;
-	case BT_CONSTRAINT_STOP_CFM :
-		if(axis < 1)
-		{
-			btAssertConstrParams(m_flags & BT_SLIDER_FLAGS_CFM_LIMLIN);
-			retVal = m_cfmLimLin;
-		}
-		else if(axis < 3)
-		{
-			btAssertConstrParams(m_flags & BT_SLIDER_FLAGS_CFM_ORTLIN);
-			retVal = m_cfmOrthoLin;
-		}
-		else if(axis == 3)
-		{
-			btAssertConstrParams(m_flags & BT_SLIDER_FLAGS_CFM_LIMANG);
-			retVal = m_cfmLimAng;
-		}
-		else if(axis < 6)
-		{
-			btAssertConstrParams(m_flags & BT_SLIDER_FLAGS_CFM_ORTANG);
-			retVal = m_cfmOrthoAng;
-		}
-		else
-		{
-			btAssertConstrParams(0);
-		}
-		break;
+		case BT_CONSTRAINT_STOP_ERP:
+			if (axis < 1)
+			{
+				btAssertConstrParams(m_flags & BT_SLIDER_FLAGS_ERP_LIMLIN);
+				retVal = m_softnessLimLin;
+			}
+			else if (axis < 3)
+			{
+				btAssertConstrParams(m_flags & BT_SLIDER_FLAGS_ERP_ORTLIN);
+				retVal = m_softnessOrthoLin;
+			}
+			else if (axis == 3)
+			{
+				btAssertConstrParams(m_flags & BT_SLIDER_FLAGS_ERP_LIMANG);
+				retVal = m_softnessLimAng;
+			}
+			else if (axis < 6)
+			{
+				btAssertConstrParams(m_flags & BT_SLIDER_FLAGS_ERP_ORTANG);
+				retVal = m_softnessOrthoAng;
+			}
+			else
+			{
+				btAssertConstrParams(0);
+			}
+			break;
+		case BT_CONSTRAINT_CFM:
+			if (axis < 1)
+			{
+				btAssertConstrParams(m_flags & BT_SLIDER_FLAGS_CFM_DIRLIN);
+				retVal = m_cfmDirLin;
+			}
+			else if (axis == 3)
+			{
+				btAssertConstrParams(m_flags & BT_SLIDER_FLAGS_CFM_DIRANG);
+				retVal = m_cfmDirAng;
+			}
+			else
+			{
+				btAssertConstrParams(0);
+			}
+			break;
+		case BT_CONSTRAINT_STOP_CFM:
+			if (axis < 1)
+			{
+				btAssertConstrParams(m_flags & BT_SLIDER_FLAGS_CFM_LIMLIN);
+				retVal = m_cfmLimLin;
+			}
+			else if (axis < 3)
+			{
+				btAssertConstrParams(m_flags & BT_SLIDER_FLAGS_CFM_ORTLIN);
+				retVal = m_cfmOrthoLin;
+			}
+			else if (axis == 3)
+			{
+				btAssertConstrParams(m_flags & BT_SLIDER_FLAGS_CFM_LIMANG);
+				retVal = m_cfmLimAng;
+			}
+			else if (axis < 6)
+			{
+				btAssertConstrParams(m_flags & BT_SLIDER_FLAGS_CFM_ORTANG);
+				retVal = m_cfmOrthoAng;
+			}
+			else
+			{
+				btAssertConstrParams(0);
+			}
+			break;
 	}
 	return retVal;
 }
-
-
-
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSliderConstraint.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSliderConstraint.h
index 1957f08a96c..75ca34e978f 100755
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSliderConstraint.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSliderConstraint.h
@@ -25,31 +25,26 @@ TODO:
 #ifndef BT_SLIDER_CONSTRAINT_H
 #define BT_SLIDER_CONSTRAINT_H
 
-#include "LinearMath/btScalar.h"//for BT_USE_DOUBLE_PRECISION
+#include "LinearMath/btScalar.h"  //for BT_USE_DOUBLE_PRECISION
 
 #ifdef BT_USE_DOUBLE_PRECISION
-#define btSliderConstraintData2		btSliderConstraintDoubleData
-#define btSliderConstraintDataName  "btSliderConstraintDoubleData"
+#define btSliderConstraintData2 btSliderConstraintDoubleData
+#define btSliderConstraintDataName "btSliderConstraintDoubleData"
 #else
-#define btSliderConstraintData2		btSliderConstraintData 
-#define btSliderConstraintDataName	"btSliderConstraintData"
-#endif //BT_USE_DOUBLE_PRECISION
+#define btSliderConstraintData2 btSliderConstraintData
+#define btSliderConstraintDataName "btSliderConstraintData"
+#endif  //BT_USE_DOUBLE_PRECISION
 
 #include "LinearMath/btVector3.h"
 #include "btJacobianEntry.h"
 #include "btTypedConstraint.h"
 
-
-
 class btRigidBody;
 
-
-
-#define SLIDER_CONSTRAINT_DEF_SOFTNESS		(btScalar(1.0))
-#define SLIDER_CONSTRAINT_DEF_DAMPING		(btScalar(1.0))
-#define SLIDER_CONSTRAINT_DEF_RESTITUTION	(btScalar(0.7))
-#define SLIDER_CONSTRAINT_DEF_CFM			(btScalar(0.f))
-
+#define SLIDER_CONSTRAINT_DEF_SOFTNESS (btScalar(1.0))
+#define SLIDER_CONSTRAINT_DEF_DAMPING (btScalar(1.0))
+#define SLIDER_CONSTRAINT_DEF_RESTITUTION (btScalar(0.7))
+#define SLIDER_CONSTRAINT_DEF_CFM (btScalar(0.f))
 
 enum btSliderFlags
 {
@@ -67,15 +62,15 @@ enum btSliderFlags
 	BT_SLIDER_FLAGS_ERP_LIMANG = (1 << 11)
 };
 
-
-ATTRIBUTE_ALIGNED16(class) btSliderConstraint : public btTypedConstraint
+ATTRIBUTE_ALIGNED16(class)
+btSliderConstraint : public btTypedConstraint
 {
 protected:
 	///for backwards compatibility during the transition to 'getInfo/getInfo2'
-	bool		m_useSolveConstraintObsolete;
-	bool		m_useOffsetForConstraintFrame;
-	btTransform	m_frameInA;
-    btTransform	m_frameInB;
+	bool m_useSolveConstraintObsolete;
+	bool m_useOffsetForConstraintFrame;
+	btTransform m_frameInA;
+	btTransform m_frameInB;
 	// use frameA fo define limits, if true
 	bool m_useLinearReferenceFrameA;
 	// linear limits
@@ -119,21 +114,21 @@ protected:
 	btScalar m_restitutionOrthoAng;
 	btScalar m_dampingOrthoAng;
 	btScalar m_cfmOrthoAng;
-	
+
 	// for interlal use
 	bool m_solveLinLim;
 	bool m_solveAngLim;
 
 	int m_flags;
 
-	btJacobianEntry	m_jacLin[3];
-	btScalar		m_jacLinDiagABInv[3];
+	btJacobianEntry m_jacLin[3];
+	btScalar m_jacLinDiagABInv[3];
 
-    btJacobianEntry	m_jacAng[3];
+	btJacobianEntry m_jacAng[3];
 
 	btScalar m_timeStep;
-    btTransform m_calculatedTransformA;
-    btTransform m_calculatedTransformB;
+	btTransform m_calculatedTransformA;
+	btTransform m_calculatedTransformB;
 
 	btVector3 m_sliderAxis;
 	btVector3 m_realPivotAInW;
@@ -150,57 +145,57 @@ protected:
 	btScalar m_angDepth;
 	btScalar m_kAngle;
 
-	bool	 m_poweredLinMotor;
-    btScalar m_targetLinMotorVelocity;
-    btScalar m_maxLinMotorForce;
-    btScalar m_accumulatedLinMotorImpulse;
-	
-	bool	 m_poweredAngMotor;
-    btScalar m_targetAngMotorVelocity;
-    btScalar m_maxAngMotorForce;
-    btScalar m_accumulatedAngMotorImpulse;
+	bool m_poweredLinMotor;
+	btScalar m_targetLinMotorVelocity;
+	btScalar m_maxLinMotorForce;
+	btScalar m_accumulatedLinMotorImpulse;
 
-	//------------------------    
+	bool m_poweredAngMotor;
+	btScalar m_targetAngMotorVelocity;
+	btScalar m_maxAngMotorForce;
+	btScalar m_accumulatedAngMotorImpulse;
+
+	//------------------------
 	void initParams();
+
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
+
 	// constructors
-    btSliderConstraint(btRigidBody& rbA, btRigidBody& rbB, const btTransform& frameInA, const btTransform& frameInB ,bool useLinearReferenceFrameA);
-    btSliderConstraint(btRigidBody& rbB, const btTransform& frameInB, bool useLinearReferenceFrameA);
+	btSliderConstraint(btRigidBody & rbA, btRigidBody & rbB, const btTransform& frameInA, const btTransform& frameInB, bool useLinearReferenceFrameA);
+	btSliderConstraint(btRigidBody & rbB, const btTransform& frameInB, bool useLinearReferenceFrameA);
 
 	// overrides
 
-    virtual void getInfo1 (btConstraintInfo1* info);
+	virtual void getInfo1(btConstraintInfo1 * info);
 
-	void getInfo1NonVirtual(btConstraintInfo1* info);
-	
-	virtual void getInfo2 (btConstraintInfo2* info);
+	void getInfo1NonVirtual(btConstraintInfo1 * info);
 
-	void getInfo2NonVirtual(btConstraintInfo2* info, const btTransform& transA, const btTransform& transB,const btVector3& linVelA,const btVector3& linVelB, btScalar rbAinvMass,btScalar rbBinvMass);
+	virtual void getInfo2(btConstraintInfo2 * info);
 
+	void getInfo2NonVirtual(btConstraintInfo2 * info, const btTransform& transA, const btTransform& transB, const btVector3& linVelA, const btVector3& linVelB, btScalar rbAinvMass, btScalar rbBinvMass);
 
 	// access
-    const btRigidBody& getRigidBodyA() const { return m_rbA; }
-    const btRigidBody& getRigidBodyB() const { return m_rbB; }
-    const btTransform & getCalculatedTransformA() const { return m_calculatedTransformA; }
-    const btTransform & getCalculatedTransformB() const { return m_calculatedTransformB; }
-    const btTransform & getFrameOffsetA() const { return m_frameInA; }
-    const btTransform & getFrameOffsetB() const { return m_frameInB; }
-    btTransform & getFrameOffsetA() { return m_frameInA; }
-    btTransform & getFrameOffsetB() { return m_frameInB; }
-    btScalar getLowerLinLimit() { return m_lowerLinLimit; }
-    void setLowerLinLimit(btScalar lowerLimit) { m_lowerLinLimit = lowerLimit; }
-    btScalar getUpperLinLimit() { return m_upperLinLimit; }
-    void setUpperLinLimit(btScalar upperLimit) { m_upperLinLimit = upperLimit; }
-    btScalar getLowerAngLimit() { return m_lowerAngLimit; }
-    void setLowerAngLimit(btScalar lowerLimit) { m_lowerAngLimit = btNormalizeAngle(lowerLimit); }
-    btScalar getUpperAngLimit() { return m_upperAngLimit; }
-    void setUpperAngLimit(btScalar upperLimit) { m_upperAngLimit = btNormalizeAngle(upperLimit); }
+	const btRigidBody& getRigidBodyA() const { return m_rbA; }
+	const btRigidBody& getRigidBodyB() const { return m_rbB; }
+	const btTransform& getCalculatedTransformA() const { return m_calculatedTransformA; }
+	const btTransform& getCalculatedTransformB() const { return m_calculatedTransformB; }
+	const btTransform& getFrameOffsetA() const { return m_frameInA; }
+	const btTransform& getFrameOffsetB() const { return m_frameInB; }
+	btTransform& getFrameOffsetA() { return m_frameInA; }
+	btTransform& getFrameOffsetB() { return m_frameInB; }
+	btScalar getLowerLinLimit() { return m_lowerLinLimit; }
+	void setLowerLinLimit(btScalar lowerLimit) { m_lowerLinLimit = lowerLimit; }
+	btScalar getUpperLinLimit() { return m_upperLinLimit; }
+	void setUpperLinLimit(btScalar upperLimit) { m_upperLinLimit = upperLimit; }
+	btScalar getLowerAngLimit() { return m_lowerAngLimit; }
+	void setLowerAngLimit(btScalar lowerLimit) { m_lowerAngLimit = btNormalizeAngle(lowerLimit); }
+	btScalar getUpperAngLimit() { return m_upperAngLimit; }
+	void setUpperAngLimit(btScalar upperLimit) { m_upperAngLimit = btNormalizeAngle(upperLimit); }
 	bool getUseLinearReferenceFrameA() { return m_useLinearReferenceFrameA; }
 	btScalar getSoftnessDirLin() { return m_softnessDirLin; }
 	btScalar getRestitutionDirLin() { return m_restitutionDirLin; }
-	btScalar getDampingDirLin() { return m_dampingDirLin ; }
+	btScalar getDampingDirLin() { return m_dampingDirLin; }
 	btScalar getSoftnessDirAng() { return m_softnessDirAng; }
 	btScalar getRestitutionDirAng() { return m_restitutionDirAng; }
 	btScalar getDampingDirAng() { return m_dampingDirAng; }
@@ -249,8 +244,6 @@ public:
 
 	btScalar getLinearPos() const { return m_linPos; }
 	btScalar getAngularPos() const { return m_angPos; }
-	
-	
 
 	// access for ODE solver
 	bool getSolveLinLimit() { return m_solveLinLim; }
@@ -258,9 +251,9 @@ public:
 	bool getSolveAngLimit() { return m_solveAngLim; }
 	btScalar getAngDepth() { return m_angDepth; }
 	// shared code used by ODE solver
-	void	calculateTransforms(const btTransform& transA,const btTransform& transB);
-	void	testLinLimits();
-	void	testAngLimits();
+	void calculateTransforms(const btTransform& transA, const btTransform& transB);
+	void testLinLimits();
+	void testAngLimits();
 	// access for PE Solver
 	btVector3 getAncorInA();
 	btVector3 getAncorInB();
@@ -268,84 +261,75 @@ public:
 	bool getUseFrameOffset() { return m_useOffsetForConstraintFrame; }
 	void setUseFrameOffset(bool frameOffsetOnOff) { m_useOffsetForConstraintFrame = frameOffsetOnOff; }
 
-	void setFrames(const btTransform& frameA, const btTransform& frameB) 
-	{ 
-		m_frameInA=frameA; 
-		m_frameInB=frameB;
-		calculateTransforms(m_rbA.getCenterOfMassTransform(),m_rbB.getCenterOfMassTransform());
+	void setFrames(const btTransform& frameA, const btTransform& frameB)
+	{
+		m_frameInA = frameA;
+		m_frameInB = frameB;
+		calculateTransforms(m_rbA.getCenterOfMassTransform(), m_rbB.getCenterOfMassTransform());
 		buildJacobian();
-	} 
+	}
 
-
-	///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5). 
+	///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5).
 	///If no axis is provided, it uses the default axis for this constraint.
-	virtual	void	setParam(int num, btScalar value, int axis = -1);
+	virtual void setParam(int num, btScalar value, int axis = -1);
 	///return the local value of parameter
-	virtual	btScalar getParam(int num, int axis = -1) const;
-	
-	virtual	int getFlags() const
-    	{
+	virtual btScalar getParam(int num, int axis = -1) const;
+
+	virtual int getFlags() const
+	{
 		return m_flags;
 	}
 
-	virtual	int	calculateSerializeBufferSize() const;
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
-
-
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 };
 
-
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
 
-
 struct btSliderConstraintData
 {
-	btTypedConstraintData	m_typeConstraintData;
-	btTransformFloatData m_rbAFrame; // constraint axii. Assumes z is hinge axis.
+	btTypedConstraintData m_typeConstraintData;
+	btTransformFloatData m_rbAFrame;  // constraint axii. Assumes z is hinge axis.
 	btTransformFloatData m_rbBFrame;
-	
-	float	m_linearUpperLimit;
-	float	m_linearLowerLimit;
 
-	float	m_angularUpperLimit;
-	float	m_angularLowerLimit;
+	float m_linearUpperLimit;
+	float m_linearLowerLimit;
 
-	int	m_useLinearReferenceFrameA;
+	float m_angularUpperLimit;
+	float m_angularLowerLimit;
+
+	int m_useLinearReferenceFrameA;
 	int m_useOffsetForConstraintFrame;
-
 };
 
-
 struct btSliderConstraintDoubleData
 {
-	btTypedConstraintDoubleData	m_typeConstraintData;
-	btTransformDoubleData m_rbAFrame; // constraint axii. Assumes z is hinge axis.
+	btTypedConstraintDoubleData m_typeConstraintData;
+	btTransformDoubleData m_rbAFrame;  // constraint axii. Assumes z is hinge axis.
 	btTransformDoubleData m_rbBFrame;
-	
-	double	m_linearUpperLimit;
-	double	m_linearLowerLimit;
 
-	double	m_angularUpperLimit;
-	double	m_angularLowerLimit;
+	double m_linearUpperLimit;
+	double m_linearLowerLimit;
 
-	int	m_useLinearReferenceFrameA;
+	double m_angularUpperLimit;
+	double m_angularLowerLimit;
+
+	int m_useLinearReferenceFrameA;
 	int m_useOffsetForConstraintFrame;
-
 };
 
-SIMD_FORCE_INLINE		int	btSliderConstraint::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btSliderConstraint::calculateSerializeBufferSize() const
 {
 	return sizeof(btSliderConstraintData2);
 }
 
-	///fills the dataBuffer and returns the struct name (and 0 on failure)
-SIMD_FORCE_INLINE	const char*	btSliderConstraint::serialize(void* dataBuffer, btSerializer* serializer) const
+///fills the dataBuffer and returns the struct name (and 0 on failure)
+SIMD_FORCE_INLINE const char* btSliderConstraint::serialize(void* dataBuffer, btSerializer* serializer) const
 {
-
-	btSliderConstraintData2* sliderData = (btSliderConstraintData2*) dataBuffer;
-	btTypedConstraint::serialize(&sliderData->m_typeConstraintData,serializer);
+	btSliderConstraintData2* sliderData = (btSliderConstraintData2*)dataBuffer;
+	btTypedConstraint::serialize(&sliderData->m_typeConstraintData, serializer);
 
 	m_frameInA.serialize(sliderData->m_rbAFrame);
 	m_frameInB.serialize(sliderData->m_rbBFrame);
@@ -362,7 +346,4 @@ SIMD_FORCE_INLINE	const char*	btSliderConstraint::serialize(void* dataBuffer, bt
 	return btSliderConstraintDataName;
 }
 
-
-
-#endif //BT_SLIDER_CONSTRAINT_H
-
+#endif  //BT_SLIDER_CONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSolve2LinearConstraint.cpp b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSolve2LinearConstraint.cpp
index 0c7dbd668bb..1ea20edcb2b 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSolve2LinearConstraint.cpp
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSolve2LinearConstraint.cpp
@@ -13,43 +13,38 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #include "btSolve2LinearConstraint.h"
 
 #include "BulletDynamics/Dynamics/btRigidBody.h"
 #include "LinearMath/btVector3.h"
 #include "btJacobianEntry.h"
 
-
 void btSolve2LinearConstraint::resolveUnilateralPairConstraint(
-												   btRigidBody* body1,
-		btRigidBody* body2,
+	btRigidBody* body1,
+	btRigidBody* body2,
 
-						const btMatrix3x3& world2A,
-						const btMatrix3x3& world2B,
-						
-						const btVector3& invInertiaADiag,
-						const btScalar invMassA,
-						const btVector3& linvelA,const btVector3& angvelA,
-						const btVector3& rel_posA1,
-						const btVector3& invInertiaBDiag,
-						const btScalar invMassB,
-						const btVector3& linvelB,const btVector3& angvelB,
-						const btVector3& rel_posA2,
+	const btMatrix3x3& world2A,
+	const btMatrix3x3& world2B,
 
-					  btScalar depthA, const btVector3& normalA, 
-					  const btVector3& rel_posB1,const btVector3& rel_posB2,
-					  btScalar depthB, const btVector3& normalB, 
-					  btScalar& imp0,btScalar& imp1)
+	const btVector3& invInertiaADiag,
+	const btScalar invMassA,
+	const btVector3& linvelA, const btVector3& angvelA,
+	const btVector3& rel_posA1,
+	const btVector3& invInertiaBDiag,
+	const btScalar invMassB,
+	const btVector3& linvelB, const btVector3& angvelB,
+	const btVector3& rel_posA2,
+
+	btScalar depthA, const btVector3& normalA,
+	const btVector3& rel_posB1, const btVector3& rel_posB2,
+	btScalar depthB, const btVector3& normalB,
+	btScalar& imp0, btScalar& imp1)
 {
 	(void)linvelA;
 	(void)linvelB;
 	(void)angvelB;
 	(void)angvelA;
 
-
-
 	imp0 = btScalar(0.);
 	imp1 = btScalar(0.);
 
@@ -59,86 +54,76 @@ void btSolve2LinearConstraint::resolveUnilateralPairConstraint(
 
 	btAssert(len < SIMD_EPSILON);
 
-
 	//this jacobian entry could be re-used for all iterations
-	btJacobianEntry jacA(world2A,world2B,rel_posA1,rel_posA2,normalA,invInertiaADiag,invMassA,
-		invInertiaBDiag,invMassB);
-	btJacobianEntry jacB(world2A,world2B,rel_posB1,rel_posB2,normalB,invInertiaADiag,invMassA,
-		invInertiaBDiag,invMassB);
-	
+	btJacobianEntry jacA(world2A, world2B, rel_posA1, rel_posA2, normalA, invInertiaADiag, invMassA,
+						 invInertiaBDiag, invMassB);
+	btJacobianEntry jacB(world2A, world2B, rel_posB1, rel_posB2, normalB, invInertiaADiag, invMassA,
+						 invInertiaBDiag, invMassB);
+
 	//const btScalar vel0 = jacA.getRelativeVelocity(linvelA,angvelA,linvelB,angvelB);
 	//const btScalar vel1 = jacB.getRelativeVelocity(linvelA,angvelA,linvelB,angvelB);
 
-	const btScalar vel0 = normalA.dot(body1->getVelocityInLocalPoint(rel_posA1)-body2->getVelocityInLocalPoint(rel_posA1));
-	const btScalar vel1 = normalB.dot(body1->getVelocityInLocalPoint(rel_posB1)-body2->getVelocityInLocalPoint(rel_posB1));
+	const btScalar vel0 = normalA.dot(body1->getVelocityInLocalPoint(rel_posA1) - body2->getVelocityInLocalPoint(rel_posA1));
+	const btScalar vel1 = normalB.dot(body1->getVelocityInLocalPoint(rel_posB1) - body2->getVelocityInLocalPoint(rel_posB1));
 
-//	btScalar penetrationImpulse = (depth*contactTau*timeCorrection)  * massTerm;//jacDiagABInv
+	//	btScalar penetrationImpulse = (depth*contactTau*timeCorrection)  * massTerm;//jacDiagABInv
 	btScalar massTerm = btScalar(1.) / (invMassA + invMassB);
 
-
 	// calculate rhs (or error) terms
-	const btScalar dv0 = depthA  * m_tau * massTerm - vel0 * m_damping;
-	const btScalar dv1 = depthB  * m_tau * massTerm - vel1 * m_damping;
-
+	const btScalar dv0 = depthA * m_tau * massTerm - vel0 * m_damping;
+	const btScalar dv1 = depthB * m_tau * massTerm - vel1 * m_damping;
 
 	// dC/dv * dv = -C
-	
+
 	// jacobian * impulse = -error
 	//
 
 	//impulse = jacobianInverse * -error
 
 	// inverting 2x2 symmetric system (offdiagonal are equal!)
-	// 
+	//
 
+	btScalar nonDiag = jacA.getNonDiagonal(jacB, invMassA, invMassB);
+	btScalar invDet = btScalar(1.0) / (jacA.getDiagonal() * jacB.getDiagonal() - nonDiag * nonDiag);
 
-	btScalar nonDiag = jacA.getNonDiagonal(jacB,invMassA,invMassB);
-	btScalar	invDet = btScalar(1.0) / (jacA.getDiagonal() * jacB.getDiagonal() - nonDiag * nonDiag );
-	
 	//imp0 = dv0 * jacA.getDiagonal() * invDet + dv1 * -nonDiag * invDet;
 	//imp1 = dv1 * jacB.getDiagonal() * invDet + dv0 * - nonDiag * invDet;
 
 	imp0 = dv0 * jacA.getDiagonal() * invDet + dv1 * -nonDiag * invDet;
-	imp1 = dv1 * jacB.getDiagonal() * invDet + dv0 * - nonDiag * invDet;
+	imp1 = dv1 * jacB.getDiagonal() * invDet + dv0 * -nonDiag * invDet;
 
 	//[a b]								  [d -c]
 	//[c d] inverse = (1 / determinant) * [-b a] where determinant is (ad - bc)
 
 	//[jA nD] * [imp0] = [dv0]
 	//[nD jB]   [imp1]   [dv1]
-
 }
 
-
-
 void btSolve2LinearConstraint::resolveBilateralPairConstraint(
-						btRigidBody* body1,
-						btRigidBody* body2,
-						const btMatrix3x3& world2A,
-						const btMatrix3x3& world2B,
-						
-						const btVector3& invInertiaADiag,
-						const btScalar invMassA,
-						const btVector3& linvelA,const btVector3& angvelA,
-						const btVector3& rel_posA1,
-						const btVector3& invInertiaBDiag,
-						const btScalar invMassB,
-						const btVector3& linvelB,const btVector3& angvelB,
-						const btVector3& rel_posA2,
+	btRigidBody* body1,
+	btRigidBody* body2,
+	const btMatrix3x3& world2A,
+	const btMatrix3x3& world2B,
 
-					  btScalar depthA, const btVector3& normalA, 
-					  const btVector3& rel_posB1,const btVector3& rel_posB2,
-					  btScalar depthB, const btVector3& normalB, 
-					  btScalar& imp0,btScalar& imp1)
+	const btVector3& invInertiaADiag,
+	const btScalar invMassA,
+	const btVector3& linvelA, const btVector3& angvelA,
+	const btVector3& rel_posA1,
+	const btVector3& invInertiaBDiag,
+	const btScalar invMassB,
+	const btVector3& linvelB, const btVector3& angvelB,
+	const btVector3& rel_posA2,
+
+	btScalar depthA, const btVector3& normalA,
+	const btVector3& rel_posB1, const btVector3& rel_posB2,
+	btScalar depthB, const btVector3& normalB,
+	btScalar& imp0, btScalar& imp1)
 {
-
 	(void)linvelA;
 	(void)linvelB;
 	(void)angvelA;
 	(void)angvelB;
 
-
-
 	imp0 = btScalar(0.);
 	imp1 = btScalar(0.);
 
@@ -148,42 +133,40 @@ void btSolve2LinearConstraint::resolveBilateralPairConstraint(
 
 	btAssert(len < SIMD_EPSILON);
 
-
 	//this jacobian entry could be re-used for all iterations
-	btJacobianEntry jacA(world2A,world2B,rel_posA1,rel_posA2,normalA,invInertiaADiag,invMassA,
-		invInertiaBDiag,invMassB);
-	btJacobianEntry jacB(world2A,world2B,rel_posB1,rel_posB2,normalB,invInertiaADiag,invMassA,
-		invInertiaBDiag,invMassB);
-	
+	btJacobianEntry jacA(world2A, world2B, rel_posA1, rel_posA2, normalA, invInertiaADiag, invMassA,
+						 invInertiaBDiag, invMassB);
+	btJacobianEntry jacB(world2A, world2B, rel_posB1, rel_posB2, normalB, invInertiaADiag, invMassA,
+						 invInertiaBDiag, invMassB);
+
 	//const btScalar vel0 = jacA.getRelativeVelocity(linvelA,angvelA,linvelB,angvelB);
 	//const btScalar vel1 = jacB.getRelativeVelocity(linvelA,angvelA,linvelB,angvelB);
 
-	const btScalar vel0 = normalA.dot(body1->getVelocityInLocalPoint(rel_posA1)-body2->getVelocityInLocalPoint(rel_posA1));
-	const btScalar vel1 = normalB.dot(body1->getVelocityInLocalPoint(rel_posB1)-body2->getVelocityInLocalPoint(rel_posB1));
+	const btScalar vel0 = normalA.dot(body1->getVelocityInLocalPoint(rel_posA1) - body2->getVelocityInLocalPoint(rel_posA1));
+	const btScalar vel1 = normalB.dot(body1->getVelocityInLocalPoint(rel_posB1) - body2->getVelocityInLocalPoint(rel_posB1));
 
 	// calculate rhs (or error) terms
-	const btScalar dv0 = depthA  * m_tau - vel0 * m_damping;
-	const btScalar dv1 = depthB  * m_tau - vel1 * m_damping;
+	const btScalar dv0 = depthA * m_tau - vel0 * m_damping;
+	const btScalar dv1 = depthB * m_tau - vel1 * m_damping;
 
 	// dC/dv * dv = -C
-	
+
 	// jacobian * impulse = -error
 	//
 
 	//impulse = jacobianInverse * -error
 
 	// inverting 2x2 symmetric system (offdiagonal are equal!)
-	// 
+	//
 
+	btScalar nonDiag = jacA.getNonDiagonal(jacB, invMassA, invMassB);
+	btScalar invDet = btScalar(1.0) / (jacA.getDiagonal() * jacB.getDiagonal() - nonDiag * nonDiag);
 
-	btScalar nonDiag = jacA.getNonDiagonal(jacB,invMassA,invMassB);
-	btScalar	invDet = btScalar(1.0) / (jacA.getDiagonal() * jacB.getDiagonal() - nonDiag * nonDiag );
-	
 	//imp0 = dv0 * jacA.getDiagonal() * invDet + dv1 * -nonDiag * invDet;
 	//imp1 = dv1 * jacB.getDiagonal() * invDet + dv0 * - nonDiag * invDet;
 
 	imp0 = dv0 * jacA.getDiagonal() * invDet + dv1 * -nonDiag * invDet;
-	imp1 = dv1 * jacB.getDiagonal() * invDet + dv0 * - nonDiag * invDet;
+	imp1 = dv1 * jacB.getDiagonal() * invDet + dv0 * -nonDiag * invDet;
 
 	//[a b]								  [d -c]
 	//[c d] inverse = (1 / determinant) * [-b a] where determinant is (ad - bc)
@@ -191,9 +174,9 @@ void btSolve2LinearConstraint::resolveBilateralPairConstraint(
 	//[jA nD] * [imp0] = [dv0]
 	//[nD jB]   [imp1]   [dv1]
 
-	if ( imp0 > btScalar(0.0))
+	if (imp0 > btScalar(0.0))
 	{
-		if ( imp1 > btScalar(0.0) )
+		if (imp1 > btScalar(0.0))
 		{
 			//both positive
 		}
@@ -203,9 +186,10 @@ void btSolve2LinearConstraint::resolveBilateralPairConstraint(
 
 			// now imp0>0 imp1<0
 			imp0 = dv0 / jacA.getDiagonal();
-			if ( imp0 > btScalar(0.0) )
+			if (imp0 > btScalar(0.0))
 			{
-			} else
+			}
+			else
 			{
 				imp0 = btScalar(0.);
 			}
@@ -216,24 +200,25 @@ void btSolve2LinearConstraint::resolveBilateralPairConstraint(
 		imp0 = btScalar(0.);
 
 		imp1 = dv1 / jacB.getDiagonal();
-		if ( imp1 <= btScalar(0.0) )
+		if (imp1 <= btScalar(0.0))
 		{
 			imp1 = btScalar(0.);
 			// now imp0>0 imp1<0
 			imp0 = dv0 / jacA.getDiagonal();
-			if ( imp0 > btScalar(0.0) )
+			if (imp0 > btScalar(0.0))
 			{
-			} else
+			}
+			else
 			{
 				imp0 = btScalar(0.);
 			}
-		} else
+		}
+		else
 		{
 		}
 	}
 }
 
-
 /*
 void btSolve2LinearConstraint::resolveAngularConstraint(	const btMatrix3x3& invInertiaAWS,
 											const btScalar invMassA,
@@ -252,4 +237,3 @@ void btSolve2LinearConstraint::resolveAngularConstraint(	const btMatrix3x3& invI
 
 }
 */
-
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSolve2LinearConstraint.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSolve2LinearConstraint.h
index e8bfabf864e..fca8ecec81d 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSolve2LinearConstraint.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSolve2LinearConstraint.h
@@ -19,20 +19,16 @@ subject to the following restrictions:
 #include "LinearMath/btMatrix3x3.h"
 #include "LinearMath/btVector3.h"
 
-
 class btRigidBody;
 
-
-
 /// constraint class used for lateral tyre friction.
-class	btSolve2LinearConstraint
+class btSolve2LinearConstraint
 {
-	btScalar	m_tau;
-	btScalar	m_damping;
+	btScalar m_tau;
+	btScalar m_damping;
 
 public:
-
-	btSolve2LinearConstraint(btScalar tau,btScalar damping)
+	btSolve2LinearConstraint(btScalar tau, btScalar damping)
 	{
 		m_tau = tau;
 		m_damping = damping;
@@ -40,52 +36,51 @@ public:
 	//
 	// solve unilateral constraint (equality, direct method)
 	//
-	void resolveUnilateralPairConstraint(		
-														   btRigidBody* body0,
+	void resolveUnilateralPairConstraint(
+		btRigidBody* body0,
 		btRigidBody* body1,
 
 		const btMatrix3x3& world2A,
-						const btMatrix3x3& world2B,
-						
-						const btVector3& invInertiaADiag,
-						const btScalar invMassA,
-						const btVector3& linvelA,const btVector3& angvelA,
-						const btVector3& rel_posA1,
-						const btVector3& invInertiaBDiag,
-						const btScalar invMassB,
-						const btVector3& linvelB,const btVector3& angvelB,
-						const btVector3& rel_posA2,
+		const btMatrix3x3& world2B,
 
-					  btScalar depthA, const btVector3& normalA, 
-					  const btVector3& rel_posB1,const btVector3& rel_posB2,
-					  btScalar depthB, const btVector3& normalB, 
-					  btScalar& imp0,btScalar& imp1);
+		const btVector3& invInertiaADiag,
+		const btScalar invMassA,
+		const btVector3& linvelA, const btVector3& angvelA,
+		const btVector3& rel_posA1,
+		const btVector3& invInertiaBDiag,
+		const btScalar invMassB,
+		const btVector3& linvelB, const btVector3& angvelB,
+		const btVector3& rel_posA2,
 
+		btScalar depthA, const btVector3& normalA,
+		const btVector3& rel_posB1, const btVector3& rel_posB2,
+		btScalar depthB, const btVector3& normalB,
+		btScalar& imp0, btScalar& imp1);
 
 	//
 	// solving 2x2 lcp problem (inequality, direct solution )
 	//
 	void resolveBilateralPairConstraint(
-			btRigidBody* body0,
-						btRigidBody* body1,
+		btRigidBody* body0,
+		btRigidBody* body1,
 		const btMatrix3x3& world2A,
-						const btMatrix3x3& world2B,
-						
-						const btVector3& invInertiaADiag,
-						const btScalar invMassA,
-						const btVector3& linvelA,const btVector3& angvelA,
-						const btVector3& rel_posA1,
-						const btVector3& invInertiaBDiag,
-						const btScalar invMassB,
-						const btVector3& linvelB,const btVector3& angvelB,
-						const btVector3& rel_posA2,
+		const btMatrix3x3& world2B,
 
-					  btScalar depthA, const btVector3& normalA, 
-					  const btVector3& rel_posB1,const btVector3& rel_posB2,
-					  btScalar depthB, const btVector3& normalB, 
-					  btScalar& imp0,btScalar& imp1);
+		const btVector3& invInertiaADiag,
+		const btScalar invMassA,
+		const btVector3& linvelA, const btVector3& angvelA,
+		const btVector3& rel_posA1,
+		const btVector3& invInertiaBDiag,
+		const btScalar invMassB,
+		const btVector3& linvelB, const btVector3& angvelB,
+		const btVector3& rel_posA2,
 
-/*
+		btScalar depthA, const btVector3& normalA,
+		const btVector3& rel_posB1, const btVector3& rel_posB2,
+		btScalar depthB, const btVector3& normalB,
+		btScalar& imp0, btScalar& imp1);
+
+	/*
 	void resolveAngularConstraint(	const btMatrix3x3& invInertiaAWS,
 						const btScalar invMassA,
 						const btVector3& linvelA,const btVector3& angvelA,
@@ -101,7 +96,6 @@ public:
 					  btScalar& imp0,btScalar& imp1);
 
 */
-
 };
 
-#endif //BT_SOLVE_2LINEAR_CONSTRAINT_H
+#endif  //BT_SOLVE_2LINEAR_CONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSolverBody.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSolverBody.h
index 27ccefe4169..409aa8a08cb 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSolverBody.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSolverBody.h
@@ -16,7 +16,7 @@ subject to the following restrictions:
 #ifndef BT_SOLVER_BODY_H
 #define BT_SOLVER_BODY_H
 
-class	btRigidBody;
+class btRigidBody;
 #include "LinearMath/btVector3.h"
 #include "LinearMath/btMatrix3x3.h"
 
@@ -26,103 +26,99 @@ class	btRigidBody;
 ///Until we get other contributions, only use SIMD on Windows, when using Visual Studio 2008 or later, and not double precision
 #ifdef BT_USE_SSE
 #define USE_SIMD 1
-#endif //
-
+#endif  //
 
 #ifdef USE_SIMD
 
-struct	btSimdScalar
+struct btSimdScalar
 {
-	SIMD_FORCE_INLINE	btSimdScalar()
-	{
-
-	}
-
-	SIMD_FORCE_INLINE	btSimdScalar(float	fl)
-	:m_vec128 (_mm_set1_ps(fl))
+	SIMD_FORCE_INLINE btSimdScalar()
 	{
 	}
 
-	SIMD_FORCE_INLINE	btSimdScalar(__m128 v128)
-		:m_vec128(v128)
+	SIMD_FORCE_INLINE btSimdScalar(float fl)
+		: m_vec128(_mm_set1_ps(fl))
 	{
 	}
-	union
+
+	SIMD_FORCE_INLINE btSimdScalar(__m128 v128)
+		: m_vec128(v128)
 	{
-		__m128		m_vec128;
-		float		m_floats[4];
-		int			m_ints[4];
-		btScalar	m_unusedPadding;
+	}
+	union {
+		__m128 m_vec128;
+		float m_floats[4];
+		int m_ints[4];
+		btScalar m_unusedPadding;
 	};
-	SIMD_FORCE_INLINE	__m128	get128()
+	SIMD_FORCE_INLINE __m128 get128()
 	{
 		return m_vec128;
 	}
 
-	SIMD_FORCE_INLINE	const __m128	get128() const
+	SIMD_FORCE_INLINE const __m128 get128() const
 	{
 		return m_vec128;
 	}
 
-	SIMD_FORCE_INLINE	void	set128(__m128 v128)
+	SIMD_FORCE_INLINE void set128(__m128 v128)
 	{
 		m_vec128 = v128;
 	}
 
-	SIMD_FORCE_INLINE	operator       __m128()       
-	{ 
-		return m_vec128; 
+	SIMD_FORCE_INLINE operator __m128()
+	{
+		return m_vec128;
 	}
-	SIMD_FORCE_INLINE	operator const __m128() const 
-	{ 
-		return m_vec128; 
-	}
-	
-	SIMD_FORCE_INLINE	operator float() const 
-	{ 
-		return m_floats[0]; 
+	SIMD_FORCE_INLINE operator const __m128() const
+	{
+		return m_vec128;
 	}
 
+	SIMD_FORCE_INLINE operator float() const
+	{
+		return m_floats[0];
+	}
 };
 
 ///@brief Return the elementwise product of two btSimdScalar
-SIMD_FORCE_INLINE btSimdScalar 
-operator*(const btSimdScalar& v1, const btSimdScalar& v2) 
+SIMD_FORCE_INLINE btSimdScalar
+operator*(const btSimdScalar& v1, const btSimdScalar& v2)
 {
-	return btSimdScalar(_mm_mul_ps(v1.get128(),v2.get128()));
+	return btSimdScalar(_mm_mul_ps(v1.get128(), v2.get128()));
 }
 
 ///@brief Return the elementwise product of two btSimdScalar
-SIMD_FORCE_INLINE btSimdScalar 
-operator+(const btSimdScalar& v1, const btSimdScalar& v2) 
+SIMD_FORCE_INLINE btSimdScalar
+operator+(const btSimdScalar& v1, const btSimdScalar& v2)
 {
-	return btSimdScalar(_mm_add_ps(v1.get128(),v2.get128()));
+	return btSimdScalar(_mm_add_ps(v1.get128(), v2.get128()));
 }
 
-
 #else
 #define btSimdScalar btScalar
 #endif
 
 ///The btSolverBody is an internal datastructure for the constraint solver. Only necessary data is packed to increase cache coherence/performance.
-ATTRIBUTE_ALIGNED16 (struct)	btSolverBody
+ATTRIBUTE_ALIGNED16(struct)
+btSolverBody
 {
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	btTransform		m_worldTransform;
-	btVector3		m_deltaLinearVelocity;
-	btVector3		m_deltaAngularVelocity;
-	btVector3		m_angularFactor;
-	btVector3		m_linearFactor;
-	btVector3		m_invMass;
-	btVector3		m_pushVelocity;
-	btVector3		m_turnVelocity;
-	btVector3		m_linearVelocity;
-	btVector3		m_angularVelocity;
-	btVector3		m_externalForceImpulse;
-	btVector3		m_externalTorqueImpulse;
+	btTransform m_worldTransform;
+	btVector3 m_deltaLinearVelocity;
+	btVector3 m_deltaAngularVelocity;
+	btVector3 m_angularFactor;
+	btVector3 m_linearFactor;
+	btVector3 m_invMass;
+	btVector3 m_pushVelocity;
+	btVector3 m_turnVelocity;
+	btVector3 m_linearVelocity;
+	btVector3 m_angularVelocity;
+	btVector3 m_externalForceImpulse;
+	btVector3 m_externalTorqueImpulse;
 
-	btRigidBody*	m_originalBody;
-	void	setWorldTransform(const btTransform& worldTransform)
+	btRigidBody* m_originalBody;
+	void setWorldTransform(const btTransform& worldTransform)
 	{
 		m_worldTransform = worldTransform;
 	}
@@ -131,56 +127,50 @@ ATTRIBUTE_ALIGNED16 (struct)	btSolverBody
 	{
 		return m_worldTransform;
 	}
-	
-	
 
-	SIMD_FORCE_INLINE void	getVelocityInLocalPointNoDelta(const btVector3& rel_pos, btVector3& velocity ) const
+	SIMD_FORCE_INLINE void getVelocityInLocalPointNoDelta(const btVector3& rel_pos, btVector3& velocity) const
 	{
 		if (m_originalBody)
-			velocity = m_linearVelocity + m_externalForceImpulse + (m_angularVelocity+m_externalTorqueImpulse).cross(rel_pos);
+			velocity = m_linearVelocity + m_externalForceImpulse + (m_angularVelocity + m_externalTorqueImpulse).cross(rel_pos);
 		else
-			velocity.setValue(0,0,0);
+			velocity.setValue(0, 0, 0);
 	}
 
-
-	SIMD_FORCE_INLINE void	getVelocityInLocalPointObsolete(const btVector3& rel_pos, btVector3& velocity ) const
+	SIMD_FORCE_INLINE void getVelocityInLocalPointObsolete(const btVector3& rel_pos, btVector3& velocity) const
 	{
 		if (m_originalBody)
-			velocity = m_linearVelocity+m_deltaLinearVelocity + (m_angularVelocity+m_deltaAngularVelocity).cross(rel_pos);
+			velocity = m_linearVelocity + m_deltaLinearVelocity + (m_angularVelocity + m_deltaAngularVelocity).cross(rel_pos);
 		else
-			velocity.setValue(0,0,0);
+			velocity.setValue(0, 0, 0);
 	}
 
-	SIMD_FORCE_INLINE void	getAngularVelocity(btVector3& angVel) const
+	SIMD_FORCE_INLINE void getAngularVelocity(btVector3 & angVel) const
 	{
 		if (m_originalBody)
-			angVel =m_angularVelocity+m_deltaAngularVelocity;
+			angVel = m_angularVelocity + m_deltaAngularVelocity;
 		else
-			angVel.setValue(0,0,0);
+			angVel.setValue(0, 0, 0);
 	}
 
-
 	//Optimization for the iterative solver: avoid calculating constant terms involving inertia, normal, relative position
-	SIMD_FORCE_INLINE void applyImpulse(const btVector3& linearComponent, const btVector3& angularComponent,const btScalar impulseMagnitude)
+	SIMD_FORCE_INLINE void applyImpulse(const btVector3& linearComponent, const btVector3& angularComponent, const btScalar impulseMagnitude)
 	{
 		if (m_originalBody)
 		{
-			m_deltaLinearVelocity += linearComponent*impulseMagnitude*m_linearFactor;
-			m_deltaAngularVelocity += angularComponent*(impulseMagnitude*m_angularFactor);
+			m_deltaLinearVelocity += linearComponent * impulseMagnitude * m_linearFactor;
+			m_deltaAngularVelocity += angularComponent * (impulseMagnitude * m_angularFactor);
 		}
 	}
 
-	SIMD_FORCE_INLINE void internalApplyPushImpulse(const btVector3& linearComponent, const btVector3& angularComponent,btScalar impulseMagnitude)
+	SIMD_FORCE_INLINE void internalApplyPushImpulse(const btVector3& linearComponent, const btVector3& angularComponent, btScalar impulseMagnitude)
 	{
 		if (m_originalBody)
 		{
-			m_pushVelocity += linearComponent*impulseMagnitude*m_linearFactor;
-			m_turnVelocity += angularComponent*(impulseMagnitude*m_angularFactor);
+			m_pushVelocity += linearComponent * impulseMagnitude * m_linearFactor;
+			m_turnVelocity += angularComponent * (impulseMagnitude * m_angularFactor);
 		}
 	}
 
-
-
 	const btVector3& getDeltaLinearVelocity() const
 	{
 		return m_deltaLinearVelocity;
@@ -191,20 +181,19 @@ ATTRIBUTE_ALIGNED16 (struct)	btSolverBody
 		return m_deltaAngularVelocity;
 	}
 
-	const btVector3& getPushVelocity() const 
+	const btVector3& getPushVelocity() const
 	{
 		return m_pushVelocity;
 	}
 
-	const btVector3& getTurnVelocity() const 
+	const btVector3& getTurnVelocity() const
 	{
 		return m_turnVelocity;
 	}
 
-
 	////////////////////////////////////////////////
 	///some internal methods, don't use them
-		
+
 	btVector3& internalGetDeltaLinearVelocity()
 	{
 		return m_deltaLinearVelocity;
@@ -229,7 +218,7 @@ ATTRIBUTE_ALIGNED16 (struct)	btSolverBody
 	{
 		m_invMass = invMass;
 	}
-	
+
 	btVector3& internalGetPushVelocity()
 	{
 		return m_pushVelocity;
@@ -240,67 +229,57 @@ ATTRIBUTE_ALIGNED16 (struct)	btSolverBody
 		return m_turnVelocity;
 	}
 
-	SIMD_FORCE_INLINE void	internalGetVelocityInLocalPointObsolete(const btVector3& rel_pos, btVector3& velocity ) const
+	SIMD_FORCE_INLINE void internalGetVelocityInLocalPointObsolete(const btVector3& rel_pos, btVector3& velocity) const
 	{
-		velocity = m_linearVelocity+m_deltaLinearVelocity + (m_angularVelocity+m_deltaAngularVelocity).cross(rel_pos);
+		velocity = m_linearVelocity + m_deltaLinearVelocity + (m_angularVelocity + m_deltaAngularVelocity).cross(rel_pos);
 	}
 
-	SIMD_FORCE_INLINE void	internalGetAngularVelocity(btVector3& angVel) const
+	SIMD_FORCE_INLINE void internalGetAngularVelocity(btVector3 & angVel) const
 	{
-		angVel = m_angularVelocity+m_deltaAngularVelocity;
+		angVel = m_angularVelocity + m_deltaAngularVelocity;
 	}
 
-
 	//Optimization for the iterative solver: avoid calculating constant terms involving inertia, normal, relative position
-	SIMD_FORCE_INLINE void internalApplyImpulse(const btVector3& linearComponent, const btVector3& angularComponent,const btScalar impulseMagnitude)
+	SIMD_FORCE_INLINE void internalApplyImpulse(const btVector3& linearComponent, const btVector3& angularComponent, const btScalar impulseMagnitude)
 	{
 		if (m_originalBody)
 		{
-			m_deltaLinearVelocity += linearComponent*impulseMagnitude*m_linearFactor;
-			m_deltaAngularVelocity += angularComponent*(impulseMagnitude*m_angularFactor);
-		}
-	}
-		
-	
-	
-
-	void	writebackVelocity()
-	{
-		if (m_originalBody)
-		{
-			m_linearVelocity +=m_deltaLinearVelocity;
-			m_angularVelocity += m_deltaAngularVelocity;
-			
-			//m_originalBody->setCompanionId(-1);
+			m_deltaLinearVelocity += linearComponent * impulseMagnitude * m_linearFactor;
+			m_deltaAngularVelocity += angularComponent * (impulseMagnitude * m_angularFactor);
 		}
 	}
 
-
-	void	writebackVelocityAndTransform(btScalar timeStep, btScalar splitImpulseTurnErp)
+	void writebackVelocity()
 	{
-        (void) timeStep;
 		if (m_originalBody)
 		{
 			m_linearVelocity += m_deltaLinearVelocity;
 			m_angularVelocity += m_deltaAngularVelocity;
-			
+
+			//m_originalBody->setCompanionId(-1);
+		}
+	}
+
+	void writebackVelocityAndTransform(btScalar timeStep, btScalar splitImpulseTurnErp)
+	{
+		(void)timeStep;
+		if (m_originalBody)
+		{
+			m_linearVelocity += m_deltaLinearVelocity;
+			m_angularVelocity += m_deltaAngularVelocity;
+
 			//correct the position/orientation based on push/turn recovery
 			btTransform newTransform;
-			if (m_pushVelocity[0]!=0.f || m_pushVelocity[1]!=0 || m_pushVelocity[2]!=0 || m_turnVelocity[0]!=0.f || m_turnVelocity[1]!=0 || m_turnVelocity[2]!=0)
+			if (m_pushVelocity[0] != 0.f || m_pushVelocity[1] != 0 || m_pushVelocity[2] != 0 || m_turnVelocity[0] != 0.f || m_turnVelocity[1] != 0 || m_turnVelocity[2] != 0)
 			{
-			//	btQuaternion orn = m_worldTransform.getRotation();
-				btTransformUtil::integrateTransform(m_worldTransform,m_pushVelocity,m_turnVelocity*splitImpulseTurnErp,timeStep,newTransform);
+				//	btQuaternion orn = m_worldTransform.getRotation();
+				btTransformUtil::integrateTransform(m_worldTransform, m_pushVelocity, m_turnVelocity * splitImpulseTurnErp, timeStep, newTransform);
 				m_worldTransform = newTransform;
 			}
 			//m_worldTransform.setRotation(orn);
 			//m_originalBody->setCompanionId(-1);
 		}
 	}
-	
-
-
 };
 
-#endif //BT_SOLVER_BODY_H
-
-
+#endif  //BT_SOLVER_BODY_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSolverConstraint.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSolverConstraint.h
index 5515e6b311c..c7938df8675 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSolverConstraint.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btSolverConstraint.h
@@ -16,7 +16,7 @@ subject to the following restrictions:
 #ifndef BT_SOLVER_CONSTRAINT_H
 #define BT_SOLVER_CONSTRAINT_H
 
-class	btRigidBody;
+class btRigidBody;
 #include "LinearMath/btVector3.h"
 #include "LinearMath/btMatrix3x3.h"
 #include "btJacobianEntry.h"
@@ -25,56 +25,50 @@ class	btRigidBody;
 //#define NO_FRICTION_TANGENTIALS 1
 #include "btSolverBody.h"
 
-
 ///1D constraint along a normal axis between bodyA and bodyB. It can be combined to solve contact and friction constraints.
-ATTRIBUTE_ALIGNED16 (struct)	btSolverConstraint
+ATTRIBUTE_ALIGNED16(struct)
+btSolverConstraint
 {
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-	btVector3		m_relpos1CrossNormal;
-	btVector3		m_contactNormal1;
+	btVector3 m_relpos1CrossNormal;
+	btVector3 m_contactNormal1;
 
-	btVector3		m_relpos2CrossNormal;
-	btVector3		m_contactNormal2; //usually m_contactNormal2 == -m_contactNormal1, but not always
+	btVector3 m_relpos2CrossNormal;
+	btVector3 m_contactNormal2;  //usually m_contactNormal2 == -m_contactNormal1, but not always
 
-	btVector3		m_angularComponentA;
-	btVector3		m_angularComponentB;
-	
-	mutable btSimdScalar	m_appliedPushImpulse;
-	mutable btSimdScalar	m_appliedImpulse;
+	btVector3 m_angularComponentA;
+	btVector3 m_angularComponentB;
 
-	btScalar	m_friction;
-	btScalar	m_jacDiagABInv;
-	btScalar		m_rhs;
-	btScalar		m_cfm;
-	
-    btScalar		m_lowerLimit;
-	btScalar		m_upperLimit;
-	btScalar		m_rhsPenetration;
-    union
-	{
-		void*		m_originalContactPoint;
-		btScalar	m_unusedPadding4;
-		int			m_numRowsForNonContactConstraint;
+	mutable btSimdScalar m_appliedPushImpulse;
+	mutable btSimdScalar m_appliedImpulse;
+
+	btScalar m_friction;
+	btScalar m_jacDiagABInv;
+	btScalar m_rhs;
+	btScalar m_cfm;
+
+	btScalar m_lowerLimit;
+	btScalar m_upperLimit;
+	btScalar m_rhsPenetration;
+	union {
+		void* m_originalContactPoint;
+		btScalar m_unusedPadding4;
+		int m_numRowsForNonContactConstraint;
 	};
 
-	int	m_overrideNumSolverIterations;
-    int			m_frictionIndex;
+	int m_overrideNumSolverIterations;
+	int m_frictionIndex;
 	int m_solverBodyIdA;
 	int m_solverBodyIdB;
 
-    
-	enum		btSolverConstraintType
+	enum btSolverConstraintType
 	{
 		BT_SOLVER_CONTACT_1D = 0,
 		BT_SOLVER_FRICTION_1D
 	};
 };
 
-typedef btAlignedObjectArray<btSolverConstraint>	btConstraintArray;
-
-
-#endif //BT_SOLVER_CONSTRAINT_H
-
-
+typedef btAlignedObjectArray<btSolverConstraint> btConstraintArray;
 
+#endif  //BT_SOLVER_CONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btTypedConstraint.cpp b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btTypedConstraint.cpp
index 9f04f280539..ebe679c449d 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btTypedConstraint.cpp
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btTypedConstraint.cpp
@@ -13,69 +13,63 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btTypedConstraint.h"
 #include "BulletDynamics/Dynamics/btRigidBody.h"
 #include "LinearMath/btSerializer.h"
 
-
 #define DEFAULT_DEBUGDRAW_SIZE btScalar(0.05f)
 
 btTypedConstraint::btTypedConstraint(btTypedConstraintType type, btRigidBody& rbA)
-:btTypedObject(type),
-m_userConstraintType(-1),
-m_userConstraintPtr((void*)-1),
-m_breakingImpulseThreshold(SIMD_INFINITY),
-m_isEnabled(true),
-m_needsFeedback(false),
-m_overrideNumSolverIterations(-1),
-m_rbA(rbA),
-m_rbB(getFixedBody()),
-m_appliedImpulse(btScalar(0.)),
-m_dbgDrawSize(DEFAULT_DEBUGDRAW_SIZE),
-m_jointFeedback(0)
+	: btTypedObject(type),
+	  m_userConstraintType(-1),
+	  m_userConstraintPtr((void*)-1),
+	  m_breakingImpulseThreshold(SIMD_INFINITY),
+	  m_isEnabled(true),
+	  m_needsFeedback(false),
+	  m_overrideNumSolverIterations(-1),
+	  m_rbA(rbA),
+	  m_rbB(getFixedBody()),
+	  m_appliedImpulse(btScalar(0.)),
+	  m_dbgDrawSize(DEFAULT_DEBUGDRAW_SIZE),
+	  m_jointFeedback(0)
 {
 }
 
-
-btTypedConstraint::btTypedConstraint(btTypedConstraintType type, btRigidBody& rbA,btRigidBody& rbB)
-:btTypedObject(type),
-m_userConstraintType(-1),
-m_userConstraintPtr((void*)-1),
-m_breakingImpulseThreshold(SIMD_INFINITY),
-m_isEnabled(true),
-m_needsFeedback(false),
-m_overrideNumSolverIterations(-1),
-m_rbA(rbA),
-m_rbB(rbB),
-m_appliedImpulse(btScalar(0.)),
-m_dbgDrawSize(DEFAULT_DEBUGDRAW_SIZE),
-m_jointFeedback(0)
+btTypedConstraint::btTypedConstraint(btTypedConstraintType type, btRigidBody& rbA, btRigidBody& rbB)
+	: btTypedObject(type),
+	  m_userConstraintType(-1),
+	  m_userConstraintPtr((void*)-1),
+	  m_breakingImpulseThreshold(SIMD_INFINITY),
+	  m_isEnabled(true),
+	  m_needsFeedback(false),
+	  m_overrideNumSolverIterations(-1),
+	  m_rbA(rbA),
+	  m_rbB(rbB),
+	  m_appliedImpulse(btScalar(0.)),
+	  m_dbgDrawSize(DEFAULT_DEBUGDRAW_SIZE),
+	  m_jointFeedback(0)
 {
 }
 
-
-
-
 btScalar btTypedConstraint::getMotorFactor(btScalar pos, btScalar lowLim, btScalar uppLim, btScalar vel, btScalar timeFact)
 {
-	if(lowLim > uppLim)
+	if (lowLim > uppLim)
 	{
 		return btScalar(1.0f);
 	}
-	else if(lowLim == uppLim)
+	else if (lowLim == uppLim)
 	{
 		return btScalar(0.0f);
 	}
 	btScalar lim_fact = btScalar(1.0f);
 	btScalar delta_max = vel / timeFact;
-	if(delta_max < btScalar(0.0f))
+	if (delta_max < btScalar(0.0f))
 	{
-		if((pos >= lowLim) && (pos < (lowLim - delta_max)))
+		if ((pos >= lowLim) && (pos < (lowLim - delta_max)))
 		{
 			lim_fact = (lowLim - pos) / delta_max;
 		}
-		else if(pos  < lowLim)
+		else if (pos < lowLim)
 		{
 			lim_fact = btScalar(0.0f);
 		}
@@ -84,13 +78,13 @@ btScalar btTypedConstraint::getMotorFactor(btScalar pos, btScalar lowLim, btScal
 			lim_fact = btScalar(1.0f);
 		}
 	}
-	else if(delta_max > btScalar(0.0f))
+	else if (delta_max > btScalar(0.0f))
 	{
-		if((pos <= uppLim) && (pos > (uppLim - delta_max)))
+		if ((pos <= uppLim) && (pos > (uppLim - delta_max)))
 		{
 			lim_fact = (uppLim - pos) / delta_max;
 		}
-		else if(pos  > uppLim)
+		else if (pos > uppLim)
 		{
 			lim_fact = btScalar(0.0f);
 		}
@@ -101,19 +95,19 @@ btScalar btTypedConstraint::getMotorFactor(btScalar pos, btScalar lowLim, btScal
 	}
 	else
 	{
-			lim_fact = btScalar(0.0f);
+		lim_fact = btScalar(0.0f);
 	}
 	return lim_fact;
 }
 
 ///fills the dataBuffer and returns the struct name (and 0 on failure)
-const char*	btTypedConstraint::serialize(void* dataBuffer, btSerializer* serializer) const
+const char* btTypedConstraint::serialize(void* dataBuffer, btSerializer* serializer) const
 {
-	btTypedConstraintData2* tcd = (btTypedConstraintData2*) dataBuffer;
+	btTypedConstraintData2* tcd = (btTypedConstraintData2*)dataBuffer;
 
 	tcd->m_rbA = (btRigidBodyData*)serializer->getUniquePointer(&m_rbA);
 	tcd->m_rbB = (btRigidBodyData*)serializer->getUniquePointer(&m_rbB);
-	char* name = (char*) serializer->findNameForPointer(this);
+	char* name = (char*)serializer->findNameForPointer(this);
 	tcd->m_name = (char*)serializer->getUniquePointer(name);
 	if (tcd->m_name)
 	{
@@ -124,10 +118,10 @@ const char*	btTypedConstraint::serialize(void* dataBuffer, btSerializer* seriali
 	tcd->m_needsFeedback = m_needsFeedback;
 	tcd->m_overrideNumSolverIterations = m_overrideNumSolverIterations;
 	tcd->m_breakingImpulseThreshold = m_breakingImpulseThreshold;
-	tcd->m_isEnabled = m_isEnabled? 1: 0;
-	
-	tcd->m_userConstraintId =m_userConstraintId;
-	tcd->m_userConstraintType =m_userConstraintType;
+	tcd->m_isEnabled = m_isEnabled ? 1 : 0;
+
+	tcd->m_userConstraintId = m_userConstraintId;
+	tcd->m_userConstraintType = m_userConstraintType;
 
 	tcd->m_appliedImpulse = m_appliedImpulse;
 	tcd->m_dbgDrawSize = m_dbgDrawSize;
@@ -135,10 +129,10 @@ const char*	btTypedConstraint::serialize(void* dataBuffer, btSerializer* seriali
 	tcd->m_disableCollisionsBetweenLinkedBodies = false;
 
 	int i;
-	for (i=0;i<m_rbA.getNumConstraintRefs();i++)
+	for (i = 0; i < m_rbA.getNumConstraintRefs(); i++)
 		if (m_rbA.getConstraintRef(i) == this)
 			tcd->m_disableCollisionsBetweenLinkedBodies = true;
-	for (i=0;i<m_rbB.getNumConstraintRefs();i++)
+	for (i = 0; i < m_rbB.getNumConstraintRefs(); i++)
 		if (m_rbB.getConstraintRef(i) == this)
 			tcd->m_disableCollisionsBetweenLinkedBodies = true;
 
@@ -147,17 +141,16 @@ const char*	btTypedConstraint::serialize(void* dataBuffer, btSerializer* seriali
 
 btRigidBody& btTypedConstraint::getFixedBody()
 {
-	static btRigidBody s_fixed(0, 0,0);
-	s_fixed.setMassProps(btScalar(0.),btVector3(btScalar(0.),btScalar(0.),btScalar(0.)));
+	static btRigidBody s_fixed(0, 0, 0);
+	s_fixed.setMassProps(btScalar(0.), btVector3(btScalar(0.), btScalar(0.), btScalar(0.)));
 	return s_fixed;
 }
 
-
 void btAngularLimit::set(btScalar low, btScalar high, btScalar _softness, btScalar _biasFactor, btScalar _relaxationFactor)
 {
 	m_halfRange = (high - low) / 2.0f;
 	m_center = btNormalizeAngle(low + m_halfRange);
-	m_softness =  _softness;
+	m_softness = _softness;
 	m_biasFactor = _biasFactor;
 	m_relaxationFactor = _relaxationFactor;
 }
@@ -174,7 +167,7 @@ void btAngularLimit::test(const btScalar angle)
 		if (deviation < -m_halfRange)
 		{
 			m_solveLimit = true;
-			m_correction = - (deviation + m_halfRange);
+			m_correction = -(deviation + m_halfRange);
 			m_sign = +1.0f;
 		}
 		else if (deviation > m_halfRange)
@@ -186,7 +179,6 @@ void btAngularLimit::test(const btScalar angle)
 	}
 }
 
-
 btScalar btAngularLimit::getError() const
 {
 	return m_correction * m_sign;
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btTypedConstraint.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btTypedConstraint.h
index 8a2a2d1ae7e..d30f3dee5c5 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btTypedConstraint.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btTypedConstraint.h
@@ -16,26 +16,24 @@ subject to the following restrictions:
 #ifndef BT_TYPED_CONSTRAINT_H
 #define BT_TYPED_CONSTRAINT_H
 
-
 #include "LinearMath/btScalar.h"
 #include "btSolverConstraint.h"
 #include "BulletDynamics/Dynamics/btRigidBody.h"
 
 #ifdef BT_USE_DOUBLE_PRECISION
-#define btTypedConstraintData2		btTypedConstraintDoubleData
-#define btTypedConstraintDataName	"btTypedConstraintDoubleData"
+#define btTypedConstraintData2 btTypedConstraintDoubleData
+#define btTypedConstraintDataName "btTypedConstraintDoubleData"
 #else
-#define btTypedConstraintData2 		btTypedConstraintFloatData
-#define btTypedConstraintDataName  "btTypedConstraintFloatData" 
-#endif //BT_USE_DOUBLE_PRECISION
-
+#define btTypedConstraintData2 btTypedConstraintFloatData
+#define btTypedConstraintDataName "btTypedConstraintFloatData"
+#endif  //BT_USE_DOUBLE_PRECISION
 
 class btSerializer;
 
 //Don't change any of the existing enum values, so add enum types at the end for serialization compatibility
 enum btTypedConstraintType
 {
-	POINT2POINT_CONSTRAINT_TYPE=3,
+	POINT2POINT_CONSTRAINT_TYPE = 3,
 	HINGE_CONSTRAINT_TYPE,
 	CONETWIST_CONSTRAINT_TYPE,
 	D6_CONSTRAINT_TYPE,
@@ -48,91 +46,88 @@ enum btTypedConstraintType
 	MAX_CONSTRAINT_TYPE
 };
 
-
 enum btConstraintParams
 {
-	BT_CONSTRAINT_ERP=1,
+	BT_CONSTRAINT_ERP = 1,
 	BT_CONSTRAINT_STOP_ERP,
 	BT_CONSTRAINT_CFM,
 	BT_CONSTRAINT_STOP_CFM
 };
 
 #if 1
-	#define btAssertConstrParams(_par) btAssert(_par) 
+#define btAssertConstrParams(_par) btAssert(_par)
 #else
-	#define btAssertConstrParams(_par)
+#define btAssertConstrParams(_par)
 #endif
 
-
-ATTRIBUTE_ALIGNED16(struct)	btJointFeedback
+ATTRIBUTE_ALIGNED16(struct)
+btJointFeedback
 {
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	btVector3	m_appliedForceBodyA;
-	btVector3	m_appliedTorqueBodyA;
-	btVector3	m_appliedForceBodyB;
-	btVector3	m_appliedTorqueBodyB;
+	btVector3 m_appliedForceBodyA;
+	btVector3 m_appliedTorqueBodyA;
+	btVector3 m_appliedForceBodyB;
+	btVector3 m_appliedTorqueBodyB;
 };
 
-
 ///TypedConstraint is the baseclass for Bullet constraints and vehicles
-ATTRIBUTE_ALIGNED16(class) btTypedConstraint : public btTypedObject
+ATTRIBUTE_ALIGNED16(class)
+btTypedConstraint : public btTypedObject
 {
-	int	m_userConstraintType;
+	int m_userConstraintType;
 
-	union
-	{
-		int	m_userConstraintId;
+	union {
+		int m_userConstraintId;
 		void* m_userConstraintPtr;
 	};
 
-	btScalar	m_breakingImpulseThreshold;
-	bool		m_isEnabled;
-	bool		m_needsFeedback;
-	int			m_overrideNumSolverIterations;
+	btScalar m_breakingImpulseThreshold;
+	bool m_isEnabled;
+	bool m_needsFeedback;
+	int m_overrideNumSolverIterations;
 
-
-	btTypedConstraint&	operator=(btTypedConstraint&	other)
+	btTypedConstraint& operator=(btTypedConstraint& other)
 	{
 		btAssert(0);
-		(void) other;
+		(void)other;
 		return *this;
 	}
 
 protected:
-	btRigidBody&	m_rbA;
-	btRigidBody&	m_rbB;
-	btScalar	m_appliedImpulse;
-	btScalar	m_dbgDrawSize;
-	btJointFeedback*	m_jointFeedback;
+	btRigidBody& m_rbA;
+	btRigidBody& m_rbB;
+	btScalar m_appliedImpulse;
+	btScalar m_dbgDrawSize;
+	btJointFeedback* m_jointFeedback;
 
 	///internal method used by the constraint solver, don't use them directly
 	btScalar getMotorFactor(btScalar pos, btScalar lowLim, btScalar uppLim, btScalar vel, btScalar timeFact);
-	
 
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-	virtual ~btTypedConstraint() {};
-	btTypedConstraint(btTypedConstraintType type, btRigidBody& rbA);
-	btTypedConstraint(btTypedConstraintType type, btRigidBody& rbA,btRigidBody& rbB);
+	virtual ~btTypedConstraint(){};
+	btTypedConstraint(btTypedConstraintType type, btRigidBody & rbA);
+	btTypedConstraint(btTypedConstraintType type, btRigidBody & rbA, btRigidBody & rbB);
 
-	struct btConstraintInfo1 {
-		int m_numConstraintRows,nub;
+	struct btConstraintInfo1
+	{
+		int m_numConstraintRows, nub;
 	};
 
 	static btRigidBody& getFixedBody();
 
-	struct btConstraintInfo2 {
+	struct btConstraintInfo2
+	{
 		// integrator parameters: frames per second (1/stepsize), default error
 		// reduction parameter (0..1).
-		btScalar fps,erp;
+		btScalar fps, erp;
 
 		// for the first and second body, pointers to two (linear and angular)
 		// n*3 jacobian sub matrices, stored by rows. these matrices will have
 		// been initialized to 0 on entry. if the second body is zero then the
 		// J2xx pointers may be 0.
-		btScalar *m_J1linearAxis,*m_J1angularAxis,*m_J2linearAxis,*m_J2angularAxis;
+		btScalar *m_J1linearAxis, *m_J1angularAxis, *m_J2linearAxis, *m_J2angularAxis;
 
 		// elements to jump from one row to the next in J's
 		int rowskip;
@@ -140,19 +135,19 @@ public:
 		// right hand sides of the equation J*v = c + cfm * lambda. cfm is the
 		// "constraint force mixing" vector. c is set to zero on entry, cfm is
 		// set to a constant value (typically very small or zero) value on entry.
-		btScalar *m_constraintError,*cfm;
+		btScalar *m_constraintError, *cfm;
 
 		// lo and hi limits for variables (set to -/+ infinity on entry).
-		btScalar *m_lowerLimit,*m_upperLimit;
+		btScalar *m_lowerLimit, *m_upperLimit;
 
 		// number of solver iterations
 		int m_numIterations;
 
 		//damping of the velocity
-		btScalar	m_damping;
+		btScalar m_damping;
 	};
 
-	int	getOverrideNumSolverIterations() const
+	int getOverrideNumSolverIterations() const
 	{
 		return m_overrideNumSolverIterations;
 	}
@@ -165,60 +160,57 @@ public:
 	}
 
 	///internal method used by the constraint solver, don't use them directly
-	virtual void	buildJacobian() {};
+	virtual void buildJacobian(){};
 
 	///internal method used by the constraint solver, don't use them directly
-	virtual	void	setupSolverConstraint(btConstraintArray& ca, int solverBodyA,int solverBodyB, btScalar timeStep)
+	virtual void setupSolverConstraint(btConstraintArray & ca, int solverBodyA, int solverBodyB, btScalar timeStep)
 	{
-        (void)ca;
-        (void)solverBodyA;
-        (void)solverBodyB;
-        (void)timeStep;
+		(void)ca;
+		(void)solverBodyA;
+		(void)solverBodyB;
+		(void)timeStep;
 	}
-	
-	///internal method used by the constraint solver, don't use them directly
-	virtual void getInfo1 (btConstraintInfo1* info)=0;
 
 	///internal method used by the constraint solver, don't use them directly
-	virtual void getInfo2 (btConstraintInfo2* info)=0;
+	virtual void getInfo1(btConstraintInfo1 * info) = 0;
 
 	///internal method used by the constraint solver, don't use them directly
-	void	internalSetAppliedImpulse(btScalar appliedImpulse)
+	virtual void getInfo2(btConstraintInfo2 * info) = 0;
+
+	///internal method used by the constraint solver, don't use them directly
+	void internalSetAppliedImpulse(btScalar appliedImpulse)
 	{
 		m_appliedImpulse = appliedImpulse;
 	}
 	///internal method used by the constraint solver, don't use them directly
-	btScalar	internalGetAppliedImpulse()
+	btScalar internalGetAppliedImpulse()
 	{
 		return m_appliedImpulse;
 	}
 
-
-	btScalar	getBreakingImpulseThreshold() const
+	btScalar getBreakingImpulseThreshold() const
 	{
-		return 	m_breakingImpulseThreshold;
+		return m_breakingImpulseThreshold;
 	}
 
-	void	setBreakingImpulseThreshold(btScalar threshold)
+	void setBreakingImpulseThreshold(btScalar threshold)
 	{
 		m_breakingImpulseThreshold = threshold;
 	}
 
-	bool	isEnabled() const
+	bool isEnabled() const
 	{
 		return m_isEnabled;
 	}
 
-	void	setEnabled(bool enabled)
+	void setEnabled(bool enabled)
 	{
-		m_isEnabled=enabled;
+		m_isEnabled = enabled;
 	}
 
-
 	///internal method used by the constraint solver, don't use them directly
-	virtual	void	solveConstraintObsolete(btSolverBody& /*bodyA*/,btSolverBody& /*bodyB*/,btScalar	/*timeStep*/) {};
+	virtual void solveConstraintObsolete(btSolverBody& /*bodyA*/, btSolverBody& /*bodyB*/, btScalar /*timeStep*/){};
 
-	
 	const btRigidBody& getRigidBodyA() const
 	{
 		return m_rbA;
@@ -228,7 +220,7 @@ public:
 		return m_rbB;
 	}
 
-	btRigidBody& getRigidBodyA() 
+	btRigidBody& getRigidBodyA()
 	{
 		return m_rbA;
 	}
@@ -239,15 +231,15 @@ public:
 
 	int getUserConstraintType() const
 	{
-		return m_userConstraintType ;
+		return m_userConstraintType;
 	}
 
-	void	setUserConstraintType(int userConstraintType)
+	void setUserConstraintType(int userConstraintType)
 	{
 		m_userConstraintType = userConstraintType;
 	};
 
-	void	setUserConstraintId(int uid)
+	void setUserConstraintId(int uid)
 	{
 		m_userConstraintId = uid;
 	}
@@ -257,17 +249,17 @@ public:
 		return m_userConstraintId;
 	}
 
-	void	setUserConstraintPtr(void* ptr)
+	void setUserConstraintPtr(void* ptr)
 	{
 		m_userConstraintPtr = ptr;
 	}
 
-	void*	getUserConstraintPtr()
+	void* getUserConstraintPtr()
 	{
 		return m_userConstraintPtr;
 	}
 
-	void	setJointFeedback(btJointFeedback* jointFeedback)
+	void setJointFeedback(btJointFeedback * jointFeedback)
 	{
 		m_jointFeedback = jointFeedback;
 	}
@@ -282,37 +274,36 @@ public:
 		return m_jointFeedback;
 	}
 
-
 	int getUid() const
 	{
-		return m_userConstraintId;   
-	} 
+		return m_userConstraintId;
+	}
 
-	bool	needsFeedback() const
+	bool needsFeedback() const
 	{
 		return m_needsFeedback;
 	}
 
 	///enableFeedback will allow to read the applied linear and angular impulse
 	///use getAppliedImpulse, getAppliedLinearImpulse and getAppliedAngularImpulse to read feedback information
-	void	enableFeedback(bool needsFeedback)
+	void enableFeedback(bool needsFeedback)
 	{
 		m_needsFeedback = needsFeedback;
 	}
 
-	///getAppliedImpulse is an estimated total applied impulse. 
+	///getAppliedImpulse is an estimated total applied impulse.
 	///This feedback could be used to determine breaking constraints or playing sounds.
-	btScalar	getAppliedImpulse() const
+	btScalar getAppliedImpulse() const
 	{
 		btAssert(m_needsFeedback);
 		return m_appliedImpulse;
 	}
 
-	btTypedConstraintType getConstraintType () const
+	btTypedConstraintType getConstraintType() const
 	{
 		return btTypedConstraintType(m_objectType);
 	}
-	
+
 	void setDbgDrawSize(btScalar dbgDrawSize)
 	{
 		m_dbgDrawSize = dbgDrawSize;
@@ -322,35 +313,34 @@ public:
 		return m_dbgDrawSize;
 	}
 
-	///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5). 
+	///override the default global value of a parameter (such as ERP or CFM), optionally provide the axis (0..5).
 	///If no axis is provided, it uses the default axis for this constraint.
-	virtual	void	setParam(int num, btScalar value, int axis = -1) = 0;
+	virtual void setParam(int num, btScalar value, int axis = -1) = 0;
 
 	///return the local value of parameter
-	virtual	btScalar getParam(int num, int axis = -1) const = 0;
-	
-	virtual	int	calculateSerializeBufferSize() const;
+	virtual btScalar getParam(int num, int axis = -1) const = 0;
+
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer, btSerializer* serializer) const;
-
+	virtual const char* serialize(void* dataBuffer, btSerializer* serializer) const;
 };
 
-// returns angle in range [-SIMD_2_PI, SIMD_2_PI], closest to one of the limits 
+// returns angle in range [-SIMD_2_PI, SIMD_2_PI], closest to one of the limits
 // all arguments should be normalized angles (i.e. in range [-SIMD_PI, SIMD_PI])
 SIMD_FORCE_INLINE btScalar btAdjustAngleToLimits(btScalar angleInRadians, btScalar angleLowerLimitInRadians, btScalar angleUpperLimitInRadians)
 {
-	if(angleLowerLimitInRadians >= angleUpperLimitInRadians)
+	if (angleLowerLimitInRadians >= angleUpperLimitInRadians)
 	{
 		return angleInRadians;
 	}
-	else if(angleInRadians < angleLowerLimitInRadians)
+	else if (angleInRadians < angleLowerLimitInRadians)
 	{
 		btScalar diffLo = btFabs(btNormalizeAngle(angleLowerLimitInRadians - angleInRadians));
 		btScalar diffHi = btFabs(btNormalizeAngle(angleUpperLimitInRadians - angleInRadians));
 		return (diffLo < diffHi) ? angleInRadians : (angleInRadians + SIMD_2_PI);
 	}
-	else if(angleInRadians > angleUpperLimitInRadians)
+	else if (angleInRadians > angleUpperLimitInRadians)
 	{
 		btScalar diffHi = btFabs(btNormalizeAngle(angleInRadians - angleUpperLimitInRadians));
 		btScalar diffLo = btFabs(btNormalizeAngle(angleInRadians - angleLowerLimitInRadians));
@@ -362,6 +352,8 @@ SIMD_FORCE_INLINE btScalar btAdjustAngleToLimits(btScalar angleInRadians, btScal
 	}
 }
 
+// clang-format off
+
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
 struct	btTypedConstraintFloatData
 {
@@ -385,6 +377,8 @@ struct	btTypedConstraintFloatData
 	
 };
 
+
+
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
 
 #define BT_BACKWARDS_COMPATIBLE_SERIALIZATION
@@ -436,18 +430,17 @@ struct	btTypedConstraintDoubleData
 	
 };
 
+// clang-format on
 
-SIMD_FORCE_INLINE	int	btTypedConstraint::calculateSerializeBufferSize() const
+SIMD_FORCE_INLINE int btTypedConstraint::calculateSerializeBufferSize() const
 {
 	return sizeof(btTypedConstraintData2);
 }
 
-
-
 class btAngularLimit
 {
 private:
-	btScalar 
+	btScalar
 		m_center,
 		m_halfRange,
 		m_softness,
@@ -462,15 +455,16 @@ private:
 public:
 	/// Default constructor initializes limit as inactive, allowing free constraint movement
 	btAngularLimit()
-		:m_center(0.0f),
-		m_halfRange(-1.0f),
-		m_softness(0.9f),
-		m_biasFactor(0.3f),
-		m_relaxationFactor(1.0f),
-		m_correction(0.0f),
-		m_sign(0.0f),
-		m_solveLimit(false)
-	{}
+		: m_center(0.0f),
+		  m_halfRange(-1.0f),
+		  m_softness(0.9f),
+		  m_biasFactor(0.3f),
+		  m_relaxationFactor(1.0f),
+		  m_correction(0.0f),
+		  m_sign(0.0f),
+		  m_solveLimit(false)
+	{
+	}
 
 	/// Sets all limit's parameters.
 	/// When low > high limit becomes inactive.
@@ -499,13 +493,13 @@ public:
 		return m_relaxationFactor;
 	}
 
-	/// Returns correction value evaluated when test() was invoked 
+	/// Returns correction value evaluated when test() was invoked
 	inline btScalar getCorrection() const
 	{
 		return m_correction;
 	}
 
-	/// Returns sign value evaluated when test() was invoked 
+	/// Returns sign value evaluated when test() was invoked
 	inline btScalar getSign() const
 	{
 		return m_sign;
@@ -533,9 +527,6 @@ public:
 	btScalar getLow() const;
 
 	btScalar getHigh() const;
-
 };
 
-
-
-#endif //BT_TYPED_CONSTRAINT_H
+#endif  //BT_TYPED_CONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btUniversalConstraint.cpp b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btUniversalConstraint.cpp
index b009f41aec8..42ed1fbb878 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btUniversalConstraint.cpp
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btUniversalConstraint.cpp
@@ -13,43 +13,38 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #include "btUniversalConstraint.h"
 #include "BulletDynamics/Dynamics/btRigidBody.h"
 #include "LinearMath/btTransformUtil.h"
 
-
-
 #define UNIV_EPS btScalar(0.01f)
 
-
 // constructor
 // anchor, axis1 and axis2 are in world coordinate system
 // axis1 must be orthogonal to axis2
 btUniversalConstraint::btUniversalConstraint(btRigidBody& rbA, btRigidBody& rbB, const btVector3& anchor, const btVector3& axis1, const btVector3& axis2)
-: btGeneric6DofConstraint(rbA, rbB, btTransform::getIdentity(), btTransform::getIdentity(), true),
- m_anchor(anchor),
- m_axis1(axis1),
- m_axis2(axis2)
+	: btGeneric6DofConstraint(rbA, rbB, btTransform::getIdentity(), btTransform::getIdentity(), true),
+	  m_anchor(anchor),
+	  m_axis1(axis1),
+	  m_axis2(axis2)
 {
 	// build frame basis
 	// 6DOF constraint uses Euler angles and to define limits
 	// it is assumed that rotational order is :
 	// Z - first, allowed limits are (-PI,PI);
-	// new position of Y - second (allowed limits are (-PI/2 + epsilon, PI/2 - epsilon), where epsilon is a small positive number 
+	// new position of Y - second (allowed limits are (-PI/2 + epsilon, PI/2 - epsilon), where epsilon is a small positive number
 	// used to prevent constraint from instability on poles;
 	// new position of X, allowed limits are (-PI,PI);
 	// So to simulate ODE Universal joint we should use parent axis as Z, child axis as Y and limit all other DOFs
 	// Build the frame in world coordinate system first
 	btVector3 zAxis = m_axis1.normalize();
 	btVector3 yAxis = m_axis2.normalize();
-	btVector3 xAxis = yAxis.cross(zAxis); // we want right coordinate system
+	btVector3 xAxis = yAxis.cross(zAxis);  // we want right coordinate system
 	btTransform frameInW;
 	frameInW.setIdentity();
-	frameInW.getBasis().setValue(	xAxis[0], yAxis[0], zAxis[0],	
-									xAxis[1], yAxis[1], zAxis[1],
-									xAxis[2], yAxis[2], zAxis[2]);
+	frameInW.getBasis().setValue(xAxis[0], yAxis[0], zAxis[0],
+								 xAxis[1], yAxis[1], zAxis[1],
+								 xAxis[2], yAxis[2], zAxis[2]);
 	frameInW.setOrigin(anchor);
 	// now get constraint frame in local coordinate systems
 	m_frameInA = rbA.getCenterOfMassTransform().inverse() * frameInW;
@@ -58,30 +53,28 @@ btUniversalConstraint::btUniversalConstraint(btRigidBody& rbA, btRigidBody& rbB,
 	setLinearLowerLimit(btVector3(0., 0., 0.));
 	setLinearUpperLimit(btVector3(0., 0., 0.));
 	setAngularLowerLimit(btVector3(0.f, -SIMD_HALF_PI + UNIV_EPS, -SIMD_PI + UNIV_EPS));
-	setAngularUpperLimit(btVector3(0.f,  SIMD_HALF_PI - UNIV_EPS,  SIMD_PI - UNIV_EPS));
+	setAngularUpperLimit(btVector3(0.f, SIMD_HALF_PI - UNIV_EPS, SIMD_PI - UNIV_EPS));
 }
 
-void btUniversalConstraint::setAxis(const btVector3& axis1,const btVector3& axis2)
+void btUniversalConstraint::setAxis(const btVector3& axis1, const btVector3& axis2)
 {
-  m_axis1 = axis1;
-  m_axis2 = axis2;
+	m_axis1 = axis1;
+	m_axis2 = axis2;
 
 	btVector3 zAxis = axis1.normalized();
 	btVector3 yAxis = axis2.normalized();
-	btVector3 xAxis = yAxis.cross(zAxis); // we want right coordinate system
+	btVector3 xAxis = yAxis.cross(zAxis);  // we want right coordinate system
 
 	btTransform frameInW;
 	frameInW.setIdentity();
-	frameInW.getBasis().setValue(	xAxis[0], yAxis[0], zAxis[0],	
-                                xAxis[1], yAxis[1], zAxis[1],
-                                xAxis[2], yAxis[2], zAxis[2]);
+	frameInW.getBasis().setValue(xAxis[0], yAxis[0], zAxis[0],
+								 xAxis[1], yAxis[1], zAxis[1],
+								 xAxis[2], yAxis[2], zAxis[2]);
 	frameInW.setOrigin(m_anchor);
 
 	// now get constraint frame in local coordinate systems
 	m_frameInA = m_rbA.getCenterOfMassTransform().inverse() * frameInW;
 	m_frameInB = m_rbB.getCenterOfMassTransform().inverse() * frameInW;
 
-  calculateTransforms();
+	calculateTransforms();
 }
-
-
diff --git a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btUniversalConstraint.h b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btUniversalConstraint.h
index 9e708410430..8c24d93a641 100644
--- a/thirdparty/bullet/BulletDynamics/ConstraintSolver/btUniversalConstraint.h
+++ b/thirdparty/bullet/BulletDynamics/ConstraintSolver/btUniversalConstraint.h
@@ -16,35 +16,32 @@ subject to the following restrictions:
 #ifndef BT_UNIVERSAL_CONSTRAINT_H
 #define BT_UNIVERSAL_CONSTRAINT_H
 
-
-
 #include "LinearMath/btVector3.h"
 #include "btTypedConstraint.h"
 #include "btGeneric6DofConstraint.h"
 
-
-
 /// Constraint similar to ODE Universal Joint
 /// has 2 rotatioonal degrees of freedom, similar to Euler rotations around Z (axis 1)
 /// and Y (axis 2)
-/// Description from ODE manual : 
-/// "Given axis 1 on body 1, and axis 2 on body 2 that is perpendicular to axis 1, it keeps them perpendicular. 
+/// Description from ODE manual :
+/// "Given axis 1 on body 1, and axis 2 on body 2 that is perpendicular to axis 1, it keeps them perpendicular.
 /// In other words, rotation of the two bodies about the direction perpendicular to the two axes will be equal."
 
-ATTRIBUTE_ALIGNED16(class) btUniversalConstraint : public btGeneric6DofConstraint
+ATTRIBUTE_ALIGNED16(class)
+btUniversalConstraint : public btGeneric6DofConstraint
 {
 protected:
-	btVector3	m_anchor;
-	btVector3	m_axis1;
-	btVector3	m_axis2;
+	btVector3 m_anchor;
+	btVector3 m_axis1;
+	btVector3 m_axis2;
+
 public:
-	
 	BT_DECLARE_ALIGNED_ALLOCATOR();
-	
+
 	// constructor
 	// anchor, axis1 and axis2 are in world coordinate system
 	// axis1 must be orthogonal to axis2
-    btUniversalConstraint(btRigidBody& rbA, btRigidBody& rbB, const btVector3& anchor, const btVector3& axis1, const btVector3& axis2);
+	btUniversalConstraint(btRigidBody & rbA, btRigidBody & rbB, const btVector3& anchor, const btVector3& axis1, const btVector3& axis2);
 	// access
 	const btVector3& getAnchor() { return m_calculatedTransformA.getOrigin(); }
 	const btVector3& getAnchor2() { return m_calculatedTransformB.getOrigin(); }
@@ -56,10 +53,7 @@ public:
 	void setUpperLimit(btScalar ang1max, btScalar ang2max) { setAngularUpperLimit(btVector3(0.f, ang1max, ang2max)); }
 	void setLowerLimit(btScalar ang1min, btScalar ang2min) { setAngularLowerLimit(btVector3(0.f, ang1min, ang2min)); }
 
-	void setAxis( const btVector3& axis1, const btVector3& axis2);
+	void setAxis(const btVector3& axis1, const btVector3& axis2);
 };
 
-
-
-#endif // BT_UNIVERSAL_CONSTRAINT_H
-
+#endif  // BT_UNIVERSAL_CONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/Dynamics/btActionInterface.h b/thirdparty/bullet/BulletDynamics/Dynamics/btActionInterface.h
index e1fea3a49c0..b5cac56cdca 100644
--- a/thirdparty/bullet/BulletDynamics/Dynamics/btActionInterface.h
+++ b/thirdparty/bullet/BulletDynamics/Dynamics/btActionInterface.h
@@ -26,21 +26,16 @@ class btCollisionWorld;
 class btActionInterface
 {
 protected:
-
 	static btRigidBody& getFixedBody();
-	
-	
-public:
 
+public:
 	virtual ~btActionInterface()
 	{
 	}
 
-	virtual void updateAction( btCollisionWorld* collisionWorld, btScalar deltaTimeStep)=0;
+	virtual void updateAction(btCollisionWorld* collisionWorld, btScalar deltaTimeStep) = 0;
 
 	virtual void debugDraw(btIDebugDraw* debugDrawer) = 0;
-
 };
 
-#endif //_BT_ACTION_INTERFACE_H
-
+#endif  //_BT_ACTION_INTERFACE_H
diff --git a/thirdparty/bullet/BulletDynamics/Dynamics/btDiscreteDynamicsWorld.cpp b/thirdparty/bullet/BulletDynamics/Dynamics/btDiscreteDynamicsWorld.cpp
index b9944c138b0..dfbbdb154f7 100644
--- a/thirdparty/bullet/BulletDynamics/Dynamics/btDiscreteDynamicsWorld.cpp
+++ b/thirdparty/bullet/BulletDynamics/Dynamics/btDiscreteDynamicsWorld.cpp
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btDiscreteDynamicsWorld.h"
 
 //collision detection
@@ -38,11 +37,9 @@ subject to the following restrictions:
 #include "BulletDynamics/ConstraintSolver/btSliderConstraint.h"
 #include "BulletDynamics/ConstraintSolver/btContactConstraint.h"
 
-
 #include "LinearMath/btIDebugDraw.h"
 #include "BulletCollision/CollisionShapes/btSphereShape.h"
 
-
 #include "BulletDynamics/Dynamics/btActionInterface.h"
 #include "LinearMath/btQuickprof.h"
 #include "LinearMath/btMotionState.h"
@@ -56,57 +53,52 @@ int startHit=2;
 int firstHit=startHit;
 #endif
 
-SIMD_FORCE_INLINE	int	btGetConstraintIslandId(const btTypedConstraint* lhs)
+SIMD_FORCE_INLINE int btGetConstraintIslandId(const btTypedConstraint* lhs)
 {
 	int islandId;
 
 	const btCollisionObject& rcolObj0 = lhs->getRigidBodyA();
 	const btCollisionObject& rcolObj1 = lhs->getRigidBodyB();
-	islandId= rcolObj0.getIslandTag()>=0?rcolObj0.getIslandTag():rcolObj1.getIslandTag();
+	islandId = rcolObj0.getIslandTag() >= 0 ? rcolObj0.getIslandTag() : rcolObj1.getIslandTag();
 	return islandId;
-
 }
 
-
 class btSortConstraintOnIslandPredicate
 {
-	public:
-
-		bool operator() ( const btTypedConstraint* lhs, const btTypedConstraint* rhs ) const
-		{
-			int rIslandId0,lIslandId0;
-			rIslandId0 = btGetConstraintIslandId(rhs);
-			lIslandId0 = btGetConstraintIslandId(lhs);
-			return lIslandId0 < rIslandId0;
-		}
+public:
+	bool operator()(const btTypedConstraint* lhs, const btTypedConstraint* rhs) const
+	{
+		int rIslandId0, lIslandId0;
+		rIslandId0 = btGetConstraintIslandId(rhs);
+		lIslandId0 = btGetConstraintIslandId(lhs);
+		return lIslandId0 < rIslandId0;
+	}
 };
 
 struct InplaceSolverIslandCallback : public btSimulationIslandManager::IslandCallback
 {
-	btContactSolverInfo*	m_solverInfo;
-	btConstraintSolver*		m_solver;
-	btTypedConstraint**		m_sortedConstraints;
-	int						m_numConstraints;
-	btIDebugDraw*			m_debugDrawer;
-	btDispatcher*			m_dispatcher;
+	btContactSolverInfo* m_solverInfo;
+	btConstraintSolver* m_solver;
+	btTypedConstraint** m_sortedConstraints;
+	int m_numConstraints;
+	btIDebugDraw* m_debugDrawer;
+	btDispatcher* m_dispatcher;
 
 	btAlignedObjectArray<btCollisionObject*> m_bodies;
 	btAlignedObjectArray<btPersistentManifold*> m_manifolds;
 	btAlignedObjectArray<btTypedConstraint*> m_constraints;
 
-
 	InplaceSolverIslandCallback(
-		btConstraintSolver*	solver,
+		btConstraintSolver* solver,
 		btStackAlloc* stackAlloc,
 		btDispatcher* dispatcher)
-		:m_solverInfo(NULL),
-		m_solver(solver),
-		m_sortedConstraints(NULL),
-		m_numConstraints(0),
-		m_debugDrawer(NULL),
-		m_dispatcher(dispatcher)
+		: m_solverInfo(NULL),
+		  m_solver(solver),
+		  m_sortedConstraints(NULL),
+		  m_numConstraints(0),
+		  m_debugDrawer(NULL),
+		  m_dispatcher(dispatcher)
 	{
-
 	}
 
 	InplaceSolverIslandCallback& operator=(InplaceSolverIslandCallback& other)
@@ -116,34 +108,34 @@ struct InplaceSolverIslandCallback : public btSimulationIslandManager::IslandCal
 		return *this;
 	}
 
-	SIMD_FORCE_INLINE void setup ( btContactSolverInfo* solverInfo, btTypedConstraint** sortedConstraints,	int	numConstraints,	btIDebugDraw* debugDrawer)
+	SIMD_FORCE_INLINE void setup(btContactSolverInfo* solverInfo, btTypedConstraint** sortedConstraints, int numConstraints, btIDebugDraw* debugDrawer)
 	{
 		btAssert(solverInfo);
 		m_solverInfo = solverInfo;
 		m_sortedConstraints = sortedConstraints;
 		m_numConstraints = numConstraints;
 		m_debugDrawer = debugDrawer;
-		m_bodies.resize (0);
-		m_manifolds.resize (0);
-		m_constraints.resize (0);
+		m_bodies.resize(0);
+		m_manifolds.resize(0);
+		m_constraints.resize(0);
 	}
 
-
-	virtual	void	processIsland(btCollisionObject** bodies,int numBodies,btPersistentManifold**	manifolds,int numManifolds, int islandId)
+	virtual void processIsland(btCollisionObject** bodies, int numBodies, btPersistentManifold** manifolds, int numManifolds, int islandId)
 	{
-		if (islandId<0)
+		if (islandId < 0)
 		{
 			///we don't split islands, so all constraints/contact manifolds/bodies are passed into the solver regardless the island id
-			m_solver->solveGroup( bodies,numBodies,manifolds, numManifolds,&m_sortedConstraints[0],m_numConstraints,*m_solverInfo,m_debugDrawer,m_dispatcher);
-		} else
+			m_solver->solveGroup(bodies, numBodies, manifolds, numManifolds, &m_sortedConstraints[0], m_numConstraints, *m_solverInfo, m_debugDrawer, m_dispatcher);
+		}
+		else
 		{
-				//also add all non-contact constraints/joints for this island
+			//also add all non-contact constraints/joints for this island
 			btTypedConstraint** startConstraint = 0;
 			int numCurConstraints = 0;
 			int i;
 
 			//find the first constraint for this island
-			for (i=0;i<m_numConstraints;i++)
+			for (i = 0; i < m_numConstraints; i++)
 			{
 				if (btGetConstraintIslandId(m_sortedConstraints[i]) == islandId)
 				{
@@ -152,7 +144,7 @@ struct InplaceSolverIslandCallback : public btSimulationIslandManager::IslandCal
 				}
 			}
 			//count the number of constraints in this island
-			for (;i<m_numConstraints;i++)
+			for (; i < m_numConstraints; i++)
 			{
 				if (btGetConstraintIslandId(m_sortedConstraints[i]) == islandId)
 				{
@@ -160,91 +152,87 @@ struct InplaceSolverIslandCallback : public btSimulationIslandManager::IslandCal
 				}
 			}
 
-			if (m_solverInfo->m_minimumSolverBatchSize<=1)
+			if (m_solverInfo->m_minimumSolverBatchSize <= 1)
 			{
-				m_solver->solveGroup( bodies,numBodies,manifolds, numManifolds,startConstraint,numCurConstraints,*m_solverInfo,m_debugDrawer,m_dispatcher);
-			} else
+				m_solver->solveGroup(bodies, numBodies, manifolds, numManifolds, startConstraint, numCurConstraints, *m_solverInfo, m_debugDrawer, m_dispatcher);
+			}
+			else
 			{
-
-				for (i=0;i<numBodies;i++)
+				for (i = 0; i < numBodies; i++)
 					m_bodies.push_back(bodies[i]);
-				for (i=0;i<numManifolds;i++)
+				for (i = 0; i < numManifolds; i++)
 					m_manifolds.push_back(manifolds[i]);
-				for (i=0;i<numCurConstraints;i++)
+				for (i = 0; i < numCurConstraints; i++)
 					m_constraints.push_back(startConstraint[i]);
-				if ((m_constraints.size()+m_manifolds.size())>m_solverInfo->m_minimumSolverBatchSize)
+				if ((m_constraints.size() + m_manifolds.size()) > m_solverInfo->m_minimumSolverBatchSize)
 				{
 					processConstraints();
-				} else
+				}
+				else
 				{
 					//printf("deferred\n");
 				}
 			}
 		}
 	}
-	void	processConstraints()
+	void processConstraints()
 	{
+		btCollisionObject** bodies = m_bodies.size() ? &m_bodies[0] : 0;
+		btPersistentManifold** manifold = m_manifolds.size() ? &m_manifolds[0] : 0;
+		btTypedConstraint** constraints = m_constraints.size() ? &m_constraints[0] : 0;
 
-		btCollisionObject** bodies = m_bodies.size()? &m_bodies[0]:0;
-		btPersistentManifold** manifold = m_manifolds.size()?&m_manifolds[0]:0;
-		btTypedConstraint** constraints = m_constraints.size()?&m_constraints[0]:0;
-
-		m_solver->solveGroup( bodies,m_bodies.size(),manifold, m_manifolds.size(),constraints, m_constraints.size() ,*m_solverInfo,m_debugDrawer,m_dispatcher);
+		m_solver->solveGroup(bodies, m_bodies.size(), manifold, m_manifolds.size(), constraints, m_constraints.size(), *m_solverInfo, m_debugDrawer, m_dispatcher);
 		m_bodies.resize(0);
 		m_manifolds.resize(0);
 		m_constraints.resize(0);
-
 	}
-
 };
 
-
-
-btDiscreteDynamicsWorld::btDiscreteDynamicsWorld(btDispatcher* dispatcher,btBroadphaseInterface* pairCache,btConstraintSolver* constraintSolver, btCollisionConfiguration* collisionConfiguration)
-:btDynamicsWorld(dispatcher,pairCache,collisionConfiguration),
-m_sortedConstraints	(),
-m_solverIslandCallback ( NULL ),
-m_constraintSolver(constraintSolver),
-m_gravity(0,-10,0),
-m_localTime(0),
-m_fixedTimeStep(0),
-m_synchronizeAllMotionStates(false),
-m_applySpeculativeContactRestitution(false),
-m_profileTimings(0),
-m_latencyMotionStateInterpolation(true)
+btDiscreteDynamicsWorld::btDiscreteDynamicsWorld(btDispatcher* dispatcher, btBroadphaseInterface* pairCache, btConstraintSolver* constraintSolver, btCollisionConfiguration* collisionConfiguration)
+	: btDynamicsWorld(dispatcher, pairCache, collisionConfiguration),
+	  m_sortedConstraints(),
+	  m_solverIslandCallback(NULL),
+	  m_constraintSolver(constraintSolver),
+	  m_gravity(0, -10, 0),
+	  m_localTime(0),
+	  m_fixedTimeStep(0),
+	  m_synchronizeAllMotionStates(false),
+	  m_applySpeculativeContactRestitution(false),
+	  m_profileTimings(0),
+	  m_latencyMotionStateInterpolation(true)
 
 {
 	if (!m_constraintSolver)
 	{
-		void* mem = btAlignedAlloc(sizeof(btSequentialImpulseConstraintSolver),16);
+		void* mem = btAlignedAlloc(sizeof(btSequentialImpulseConstraintSolver), 16);
 		m_constraintSolver = new (mem) btSequentialImpulseConstraintSolver;
 		m_ownsConstraintSolver = true;
-	} else
+	}
+	else
 	{
 		m_ownsConstraintSolver = false;
 	}
 
 	{
-		void* mem = btAlignedAlloc(sizeof(btSimulationIslandManager),16);
+		void* mem = btAlignedAlloc(sizeof(btSimulationIslandManager), 16);
 		m_islandManager = new (mem) btSimulationIslandManager();
 	}
 
 	m_ownsIslandManager = true;
 
 	{
-		void* mem = btAlignedAlloc(sizeof(InplaceSolverIslandCallback),16);
-		m_solverIslandCallback = new (mem) InplaceSolverIslandCallback (m_constraintSolver, 0, dispatcher);
+		void* mem = btAlignedAlloc(sizeof(InplaceSolverIslandCallback), 16);
+		m_solverIslandCallback = new (mem) InplaceSolverIslandCallback(m_constraintSolver, 0, dispatcher);
 	}
 }
 
-
 btDiscreteDynamicsWorld::~btDiscreteDynamicsWorld()
 {
 	//only delete it when we created it
 	if (m_ownsIslandManager)
 	{
 		m_islandManager->~btSimulationIslandManager();
-		btAlignedFree( m_islandManager);
+		btAlignedFree(m_islandManager);
 	}
 	if (m_solverIslandCallback)
 	{
@@ -253,18 +241,17 @@ btDiscreteDynamicsWorld::~btDiscreteDynamicsWorld()
 	}
 	if (m_ownsConstraintSolver)
 	{
-
 		m_constraintSolver->~btConstraintSolver();
 		btAlignedFree(m_constraintSolver);
 	}
 }
 
-void	btDiscreteDynamicsWorld::saveKinematicState(btScalar timeStep)
+void btDiscreteDynamicsWorld::saveKinematicState(btScalar timeStep)
 {
-///would like to iterate over m_nonStaticRigidBodies, but unfortunately old API allows
-///to switch status _after_ adding kinematic objects to the world
-///fix it for Bullet 3.x release
-	for (int i=0;i<m_collisionObjects.size();i++)
+	///would like to iterate over m_nonStaticRigidBodies, but unfortunately old API allows
+	///to switch status _after_ adding kinematic objects to the world
+	///fix it for Bullet 3.x release
+	for (int i = 0; i < m_collisionObjects.size(); i++)
 	{
 		btCollisionObject* colObj = m_collisionObjects[i];
 		btRigidBody* body = btRigidBody::upcast(colObj);
@@ -277,10 +264,9 @@ void	btDiscreteDynamicsWorld::saveKinematicState(btScalar timeStep)
 			}
 		}
 	}
-
 }
 
-void	btDiscreteDynamicsWorld::debugDrawWorld()
+void btDiscreteDynamicsWorld::debugDrawWorld()
 {
 	BT_PROFILE("debugDrawWorld");
 
@@ -290,43 +276,40 @@ void	btDiscreteDynamicsWorld::debugDrawWorld()
 	if (getDebugDrawer())
 	{
 		int mode = getDebugDrawer()->getDebugMode();
-		if(mode  & (btIDebugDraw::DBG_DrawConstraints | btIDebugDraw::DBG_DrawConstraintLimits))
+		if (mode & (btIDebugDraw::DBG_DrawConstraints | btIDebugDraw::DBG_DrawConstraintLimits))
 		{
 			drawConstraints = true;
 		}
 	}
-	if(drawConstraints)
+	if (drawConstraints)
 	{
-		for(int i = getNumConstraints()-1; i>=0 ;i--)
+		for (int i = getNumConstraints() - 1; i >= 0; i--)
 		{
 			btTypedConstraint* constraint = getConstraint(i);
 			debugDrawConstraint(constraint);
 		}
 	}
 
-
-
-    if (getDebugDrawer() && (getDebugDrawer()->getDebugMode() & (btIDebugDraw::DBG_DrawWireframe | btIDebugDraw::DBG_DrawAabb | btIDebugDraw::DBG_DrawNormals)))
+	if (getDebugDrawer() && (getDebugDrawer()->getDebugMode() & (btIDebugDraw::DBG_DrawWireframe | btIDebugDraw::DBG_DrawAabb | btIDebugDraw::DBG_DrawNormals)))
 	{
 		int i;
 
 		if (getDebugDrawer() && getDebugDrawer()->getDebugMode())
 		{
-			for (i=0;i<m_actions.size();i++)
+			for (i = 0; i < m_actions.size(); i++)
 			{
 				m_actions[i]->debugDraw(m_debugDrawer);
 			}
 		}
 	}
-    if (getDebugDrawer())
-        getDebugDrawer()->flushLines();
-
+	if (getDebugDrawer())
+		getDebugDrawer()->flushLines();
 }
 
-void	btDiscreteDynamicsWorld::clearForces()
+void btDiscreteDynamicsWorld::clearForces()
 {
 	///@todo: iterate over awake simulation islands!
-	for ( int i=0;i<m_nonStaticRigidBodies.size();i++)
+	for (int i = 0; i < m_nonStaticRigidBodies.size(); i++)
 	{
 		btRigidBody* body = m_nonStaticRigidBodies[i];
 		//need to check if next line is ok
@@ -336,10 +319,10 @@ void	btDiscreteDynamicsWorld::clearForces()
 }
 
 ///apply gravity, call this once per timestep
-void	btDiscreteDynamicsWorld::applyGravity()
+void btDiscreteDynamicsWorld::applyGravity()
 {
 	///@todo: iterate over awake simulation islands!
-	for ( int i=0;i<m_nonStaticRigidBodies.size();i++)
+	for (int i = 0; i < m_nonStaticRigidBodies.size(); i++)
 	{
 		btRigidBody* body = m_nonStaticRigidBodies[i];
 		if (body->isActive())
@@ -349,8 +332,7 @@ void	btDiscreteDynamicsWorld::applyGravity()
 	}
 }
 
-
-void	btDiscreteDynamicsWorld::synchronizeSingleMotionState(btRigidBody* body)
+void btDiscreteDynamicsWorld::synchronizeSingleMotionState(btRigidBody* body)
 {
 	btAssert(body);
 
@@ -363,32 +345,32 @@ void	btDiscreteDynamicsWorld::synchronizeSingleMotionState(btRigidBody* body)
 		{
 			btTransform interpolatedTransform;
 			btTransformUtil::integrateTransform(body->getInterpolationWorldTransform(),
-				body->getInterpolationLinearVelocity(),body->getInterpolationAngularVelocity(),
-				(m_latencyMotionStateInterpolation && m_fixedTimeStep) ? m_localTime - m_fixedTimeStep : m_localTime*body->getHitFraction(),
-				interpolatedTransform);
+												body->getInterpolationLinearVelocity(), body->getInterpolationAngularVelocity(),
+												(m_latencyMotionStateInterpolation && m_fixedTimeStep) ? m_localTime - m_fixedTimeStep : m_localTime * body->getHitFraction(),
+												interpolatedTransform);
 			body->getMotionState()->setWorldTransform(interpolatedTransform);
 		}
 	}
 }
 
-
-void	btDiscreteDynamicsWorld::synchronizeMotionStates()
+void btDiscreteDynamicsWorld::synchronizeMotionStates()
 {
-//	BT_PROFILE("synchronizeMotionStates");
+	//	BT_PROFILE("synchronizeMotionStates");
 	if (m_synchronizeAllMotionStates)
 	{
 		//iterate  over all collision objects
-		for ( int i=0;i<m_collisionObjects.size();i++)
+		for (int i = 0; i < m_collisionObjects.size(); i++)
 		{
 			btCollisionObject* colObj = m_collisionObjects[i];
 			btRigidBody* body = btRigidBody::upcast(colObj);
 			if (body)
 				synchronizeSingleMotionState(body);
 		}
-	} else
+	}
+	else
 	{
 		//iterate over all active rigid bodies
-		for ( int i=0;i<m_nonStaticRigidBodies.size();i++)
+		for (int i = 0; i < m_nonStaticRigidBodies.size(); i++)
 		{
 			btRigidBody* body = m_nonStaticRigidBodies[i];
 			if (body->isActive())
@@ -397,12 +379,10 @@ void	btDiscreteDynamicsWorld::synchronizeMotionStates()
 	}
 }
 
-
-int	btDiscreteDynamicsWorld::stepSimulation( btScalar timeStep,int maxSubSteps, btScalar fixedTimeStep)
+int btDiscreteDynamicsWorld::stepSimulation(btScalar timeStep, int maxSubSteps, btScalar fixedTimeStep)
 {
 	startProfiling(timeStep);
 
-
 	int numSimulationSubSteps = 0;
 
 	if (maxSubSteps)
@@ -412,10 +392,11 @@ int	btDiscreteDynamicsWorld::stepSimulation( btScalar timeStep,int maxSubSteps,
 		m_localTime += timeStep;
 		if (m_localTime >= fixedTimeStep)
 		{
-			numSimulationSubSteps = int( m_localTime / fixedTimeStep);
+			numSimulationSubSteps = int(m_localTime / fixedTimeStep);
 			m_localTime -= numSimulationSubSteps * fixedTimeStep;
 		}
-	} else
+	}
+	else
 	{
 		//variable timestep
 		fixedTimeStep = timeStep;
@@ -425,7 +406,8 @@ int	btDiscreteDynamicsWorld::stepSimulation( btScalar timeStep,int maxSubSteps,
 		{
 			numSimulationSubSteps = 0;
 			maxSubSteps = 0;
-		} else
+		}
+		else
 		{
 			numSimulationSubSteps = 1;
 			maxSubSteps = 1;
@@ -435,28 +417,25 @@ int	btDiscreteDynamicsWorld::stepSimulation( btScalar timeStep,int maxSubSteps,
 	//process some debugging flags
 	if (getDebugDrawer())
 	{
-		btIDebugDraw* debugDrawer = getDebugDrawer ();
+		btIDebugDraw* debugDrawer = getDebugDrawer();
 		gDisableDeactivation = (debugDrawer->getDebugMode() & btIDebugDraw::DBG_NoDeactivation) != 0;
 	}
 	if (numSimulationSubSteps)
 	{
-
 		//clamp the number of substeps, to prevent simulation grinding spiralling down to a halt
-		int clampedSimulationSteps = (numSimulationSubSteps > maxSubSteps)? maxSubSteps : numSimulationSubSteps;
+		int clampedSimulationSteps = (numSimulationSubSteps > maxSubSteps) ? maxSubSteps : numSimulationSubSteps;
 
-		saveKinematicState(fixedTimeStep*clampedSimulationSteps);
+		saveKinematicState(fixedTimeStep * clampedSimulationSteps);
 
 		applyGravity();
 
-
-
-		for (int i=0;i<clampedSimulationSteps;i++)
+		for (int i = 0; i < clampedSimulationSteps; i++)
 		{
 			internalSingleStepSimulation(fixedTimeStep);
 			synchronizeMotionStates();
 		}
-
-	} else
+	}
+	else
 	{
 		synchronizeMotionStates();
 	}
@@ -465,17 +444,17 @@ int	btDiscreteDynamicsWorld::stepSimulation( btScalar timeStep,int maxSubSteps,
 
 #ifndef BT_NO_PROFILE
 	CProfileManager::Increment_Frame_Counter();
-#endif //BT_NO_PROFILE
+#endif  //BT_NO_PROFILE
 
 	return numSimulationSubSteps;
 }
 
-void	btDiscreteDynamicsWorld::internalSingleStepSimulation(btScalar timeStep)
+void btDiscreteDynamicsWorld::internalSingleStepSimulation(btScalar timeStep)
 {
-
 	BT_PROFILE("internalSingleStepSimulation");
 
-	if(0 != m_internalPreTickCallback) {
+	if (0 != m_internalPreTickCallback)
+	{
 		(*m_internalPreTickCallback)(this, timeStep);
 	}
 
@@ -488,19 +467,15 @@ void	btDiscreteDynamicsWorld::internalSingleStepSimulation(btScalar timeStep)
 	dispatchInfo.m_stepCount = 0;
 	dispatchInfo.m_debugDraw = getDebugDrawer();
 
-
-    createPredictiveContacts(timeStep);
+	createPredictiveContacts(timeStep);
 
 	///perform collision detection
 	performDiscreteCollisionDetection();
 
 	calculateSimulationIslands();
 
-
 	getSolverInfo().m_timeStep = timeStep;
 
-
-
 	///solve contact and other joint constraints
 	solveConstraints(getSolverInfo());
 
@@ -513,37 +488,38 @@ void	btDiscreteDynamicsWorld::internalSingleStepSimulation(btScalar timeStep)
 	///update vehicle simulation
 	updateActions(timeStep);
 
-	updateActivationState( timeStep );
+	updateActivationState(timeStep);
 
-	if(0 != m_internalTickCallback) {
+	if (0 != m_internalTickCallback)
+	{
 		(*m_internalTickCallback)(this, timeStep);
 	}
 }
 
-void	btDiscreteDynamicsWorld::setGravity(const btVector3& gravity)
+void btDiscreteDynamicsWorld::setGravity(const btVector3& gravity)
 {
 	m_gravity = gravity;
-	for ( int i=0;i<m_nonStaticRigidBodies.size();i++)
+	for (int i = 0; i < m_nonStaticRigidBodies.size(); i++)
 	{
 		btRigidBody* body = m_nonStaticRigidBodies[i];
-		if (body->isActive() && !(body->getFlags() &BT_DISABLE_WORLD_GRAVITY))
+		if (body->isActive() && !(body->getFlags() & BT_DISABLE_WORLD_GRAVITY))
 		{
 			body->setGravity(gravity);
 		}
 	}
 }
 
-btVector3 btDiscreteDynamicsWorld::getGravity () const
+btVector3 btDiscreteDynamicsWorld::getGravity() const
 {
 	return m_gravity;
 }
 
-void	btDiscreteDynamicsWorld::addCollisionObject(btCollisionObject* collisionObject, int collisionFilterGroup, int collisionFilterMask)
+void btDiscreteDynamicsWorld::addCollisionObject(btCollisionObject* collisionObject, int collisionFilterGroup, int collisionFilterMask)
 {
-	btCollisionWorld::addCollisionObject(collisionObject,collisionFilterGroup,collisionFilterMask);
+	btCollisionWorld::addCollisionObject(collisionObject, collisionFilterGroup, collisionFilterMask);
 }
 
-void	btDiscreteDynamicsWorld::removeCollisionObject(btCollisionObject* collisionObject)
+void btDiscreteDynamicsWorld::removeCollisionObject(btCollisionObject* collisionObject)
 {
 	btRigidBody* body = btRigidBody::upcast(collisionObject);
 	if (body)
@@ -552,16 +528,15 @@ void	btDiscreteDynamicsWorld::removeCollisionObject(btCollisionObject* collision
 		btCollisionWorld::removeCollisionObject(collisionObject);
 }
 
-void	btDiscreteDynamicsWorld::removeRigidBody(btRigidBody* body)
+void btDiscreteDynamicsWorld::removeRigidBody(btRigidBody* body)
 {
 	m_nonStaticRigidBodies.remove(body);
 	btCollisionWorld::removeCollisionObject(body);
 }
 
-
-void	btDiscreteDynamicsWorld::addRigidBody(btRigidBody* body)
+void btDiscreteDynamicsWorld::addRigidBody(btRigidBody* body)
 {
-	if (!body->isStaticOrKinematicObject() && !(body->getFlags() &BT_DISABLE_WORLD_GRAVITY))
+	if (!body->isStaticOrKinematicObject() && !(body->getFlags() & BT_DISABLE_WORLD_GRAVITY))
 	{
 		body->setGravity(m_gravity);
 	}
@@ -571,22 +546,23 @@ void	btDiscreteDynamicsWorld::addRigidBody(btRigidBody* body)
 		if (!body->isStaticObject())
 		{
 			m_nonStaticRigidBodies.push_back(body);
-		} else
+		}
+		else
 		{
 			body->setActivationState(ISLAND_SLEEPING);
 		}
 
 		bool isDynamic = !(body->isStaticObject() || body->isKinematicObject());
-		int collisionFilterGroup = isDynamic? int(btBroadphaseProxy::DefaultFilter) : int(btBroadphaseProxy::StaticFilter);
-		int collisionFilterMask = isDynamic? 	int(btBroadphaseProxy::AllFilter) : 	int(btBroadphaseProxy::AllFilter ^ btBroadphaseProxy::StaticFilter);
+		int collisionFilterGroup = isDynamic ? int(btBroadphaseProxy::DefaultFilter) : int(btBroadphaseProxy::StaticFilter);
+		int collisionFilterMask = isDynamic ? int(btBroadphaseProxy::AllFilter) : int(btBroadphaseProxy::AllFilter ^ btBroadphaseProxy::StaticFilter);
 
-		addCollisionObject(body,collisionFilterGroup,collisionFilterMask);
+		addCollisionObject(body, collisionFilterGroup, collisionFilterMask);
 	}
 }
 
-void	btDiscreteDynamicsWorld::addRigidBody(btRigidBody* body, int group, int mask)
+void btDiscreteDynamicsWorld::addRigidBody(btRigidBody* body, int group, int mask)
 {
-	if (!body->isStaticOrKinematicObject() && !(body->getFlags() &BT_DISABLE_WORLD_GRAVITY))
+	if (!body->isStaticOrKinematicObject() && !(body->getFlags() & BT_DISABLE_WORLD_GRAVITY))
 	{
 		body->setGravity(m_gravity);
 	}
@@ -597,31 +573,29 @@ void	btDiscreteDynamicsWorld::addRigidBody(btRigidBody* body, int group, int mas
 		{
 			m_nonStaticRigidBodies.push_back(body);
 		}
-		 else
+		else
 		{
 			body->setActivationState(ISLAND_SLEEPING);
 		}
-		addCollisionObject(body,group,mask);
+		addCollisionObject(body, group, mask);
 	}
 }
 
-
-void	btDiscreteDynamicsWorld::updateActions(btScalar timeStep)
+void btDiscreteDynamicsWorld::updateActions(btScalar timeStep)
 {
 	BT_PROFILE("updateActions");
 
-	for ( int i=0;i<m_actions.size();i++)
+	for (int i = 0; i < m_actions.size(); i++)
 	{
-		m_actions[i]->updateAction( this, timeStep);
+		m_actions[i]->updateAction(this, timeStep);
 	}
 }
 
-
-void	btDiscreteDynamicsWorld::updateActivationState(btScalar timeStep)
+void btDiscreteDynamicsWorld::updateActivationState(btScalar timeStep)
 {
 	BT_PROFILE("updateActivationState");
 
-	for ( int i=0;i<m_nonStaticRigidBodies.size();i++)
+	for (int i = 0; i < m_nonStaticRigidBodies.size(); i++)
 	{
 		btRigidBody* body = m_nonStaticRigidBodies[i];
 		if (body)
@@ -633,32 +607,33 @@ void	btDiscreteDynamicsWorld::updateActivationState(btScalar timeStep)
 				if (body->isStaticOrKinematicObject())
 				{
 					body->setActivationState(ISLAND_SLEEPING);
-				} else
+				}
+				else
 				{
 					if (body->getActivationState() == ACTIVE_TAG)
-						body->setActivationState( WANTS_DEACTIVATION );
+						body->setActivationState(WANTS_DEACTIVATION);
 					if (body->getActivationState() == ISLAND_SLEEPING)
 					{
-						body->setAngularVelocity(btVector3(0,0,0));
-						body->setLinearVelocity(btVector3(0,0,0));
+						body->setAngularVelocity(btVector3(0, 0, 0));
+						body->setLinearVelocity(btVector3(0, 0, 0));
 					}
-
 				}
-			} else
+			}
+			else
 			{
 				if (body->getActivationState() != DISABLE_DEACTIVATION)
-					body->setActivationState( ACTIVE_TAG );
+					body->setActivationState(ACTIVE_TAG);
 			}
 		}
 	}
 }
 
-void	btDiscreteDynamicsWorld::addConstraint(btTypedConstraint* constraint,bool disableCollisionsBetweenLinkedBodies)
+void btDiscreteDynamicsWorld::addConstraint(btTypedConstraint* constraint, bool disableCollisionsBetweenLinkedBodies)
 {
 	m_constraints.push_back(constraint);
-    //Make sure the two bodies of a type constraint are different (possibly add this to the btTypedConstraint constructor?)
-    btAssert(&constraint->getRigidBodyA()!=&constraint->getRigidBodyB());
-    
+	//Make sure the two bodies of a type constraint are different (possibly add this to the btTypedConstraint constructor?)
+	btAssert(&constraint->getRigidBodyA() != &constraint->getRigidBodyB());
+
 	if (disableCollisionsBetweenLinkedBodies)
 	{
 		constraint->getRigidBodyA().addConstraintRef(constraint);
@@ -666,105 +641,98 @@ void	btDiscreteDynamicsWorld::addConstraint(btTypedConstraint* constraint,bool d
 	}
 }
 
-void	btDiscreteDynamicsWorld::removeConstraint(btTypedConstraint* constraint)
+void btDiscreteDynamicsWorld::removeConstraint(btTypedConstraint* constraint)
 {
 	m_constraints.remove(constraint);
 	constraint->getRigidBodyA().removeConstraintRef(constraint);
 	constraint->getRigidBodyB().removeConstraintRef(constraint);
 }
 
-void	btDiscreteDynamicsWorld::addAction(btActionInterface* action)
+void btDiscreteDynamicsWorld::addAction(btActionInterface* action)
 {
 	m_actions.push_back(action);
 }
 
-void	btDiscreteDynamicsWorld::removeAction(btActionInterface* action)
+void btDiscreteDynamicsWorld::removeAction(btActionInterface* action)
 {
 	m_actions.remove(action);
 }
 
-
-void	btDiscreteDynamicsWorld::addVehicle(btActionInterface* vehicle)
+void btDiscreteDynamicsWorld::addVehicle(btActionInterface* vehicle)
 {
 	addAction(vehicle);
 }
 
-void	btDiscreteDynamicsWorld::removeVehicle(btActionInterface* vehicle)
+void btDiscreteDynamicsWorld::removeVehicle(btActionInterface* vehicle)
 {
 	removeAction(vehicle);
 }
 
-void	btDiscreteDynamicsWorld::addCharacter(btActionInterface* character)
+void btDiscreteDynamicsWorld::addCharacter(btActionInterface* character)
 {
 	addAction(character);
 }
 
-void	btDiscreteDynamicsWorld::removeCharacter(btActionInterface* character)
+void btDiscreteDynamicsWorld::removeCharacter(btActionInterface* character)
 {
 	removeAction(character);
 }
 
-
-
-
-void	btDiscreteDynamicsWorld::solveConstraints(btContactSolverInfo& solverInfo)
+void btDiscreteDynamicsWorld::solveConstraints(btContactSolverInfo& solverInfo)
 {
 	BT_PROFILE("solveConstraints");
 
-	m_sortedConstraints.resize( m_constraints.size());
+	m_sortedConstraints.resize(m_constraints.size());
 	int i;
-	for (i=0;i<getNumConstraints();i++)
+	for (i = 0; i < getNumConstraints(); i++)
 	{
 		m_sortedConstraints[i] = m_constraints[i];
 	}
 
-//	btAssert(0);
-
-
+	//	btAssert(0);
 
 	m_sortedConstraints.quickSort(btSortConstraintOnIslandPredicate());
 
 	btTypedConstraint** constraintsPtr = getNumConstraints() ? &m_sortedConstraints[0] : 0;
 
-	m_solverIslandCallback->setup(&solverInfo,constraintsPtr,m_sortedConstraints.size(),getDebugDrawer());
+	m_solverIslandCallback->setup(&solverInfo, constraintsPtr, m_sortedConstraints.size(), getDebugDrawer());
 	m_constraintSolver->prepareSolve(getCollisionWorld()->getNumCollisionObjects(), getCollisionWorld()->getDispatcher()->getNumManifolds());
 
 	/// solve all the constraints for this island
-	m_islandManager->buildAndProcessIslands(getCollisionWorld()->getDispatcher(),getCollisionWorld(),m_solverIslandCallback);
+	m_islandManager->buildAndProcessIslands(getCollisionWorld()->getDispatcher(), getCollisionWorld(), m_solverIslandCallback);
 
 	m_solverIslandCallback->processConstraints();
 
 	m_constraintSolver->allSolved(solverInfo, m_debugDrawer);
 }
 
-
-void	btDiscreteDynamicsWorld::calculateSimulationIslands()
+void btDiscreteDynamicsWorld::calculateSimulationIslands()
 {
 	BT_PROFILE("calculateSimulationIslands");
 
-	getSimulationIslandManager()->updateActivationState(getCollisionWorld(),getCollisionWorld()->getDispatcher());
+	getSimulationIslandManager()->updateActivationState(getCollisionWorld(), getCollisionWorld()->getDispatcher());
 
-    {
-        //merge islands based on speculative contact manifolds too
-        for (int i=0;i<this->m_predictiveManifolds.size();i++)
-        {
-            btPersistentManifold* manifold = m_predictiveManifolds[i];
+	{
+		//merge islands based on speculative contact manifolds too
+		for (int i = 0; i < this->m_predictiveManifolds.size(); i++)
+		{
+			btPersistentManifold* manifold = m_predictiveManifolds[i];
 
-            const btCollisionObject* colObj0 = manifold->getBody0();
-            const btCollisionObject* colObj1 = manifold->getBody1();
+			const btCollisionObject* colObj0 = manifold->getBody0();
+			const btCollisionObject* colObj1 = manifold->getBody1();
 
-            if (((colObj0) && (!(colObj0)->isStaticOrKinematicObject())) &&
-                ((colObj1) && (!(colObj1)->isStaticOrKinematicObject())))
-            {
-				getSimulationIslandManager()->getUnionFind().unite((colObj0)->getIslandTag(),(colObj1)->getIslandTag());
-            }
-        }
-    }
+			if (((colObj0) && (!(colObj0)->isStaticOrKinematicObject())) &&
+				((colObj1) && (!(colObj1)->isStaticOrKinematicObject())))
+			{
+				getSimulationIslandManager()->getUnionFind().unite((colObj0)->getIslandTag(), (colObj1)->getIslandTag());
+			}
+		}
+	}
 
 	{
 		int i;
 		int numConstraints = int(m_constraints.size());
-		for (i=0;i< numConstraints ; i++ )
+		for (i = 0; i < numConstraints; i++)
 		{
 			btTypedConstraint* constraint = m_constraints[i];
 			if (constraint->isEnabled())
@@ -775,7 +743,7 @@ void	btDiscreteDynamicsWorld::calculateSimulationIslands()
 				if (((colObj0) && (!(colObj0)->isStaticOrKinematicObject())) &&
 					((colObj1) && (!(colObj1)->isStaticOrKinematicObject())))
 				{
-					getSimulationIslandManager()->getUnionFind().unite((colObj0)->getIslandTag(),(colObj1)->getIslandTag());
+					getSimulationIslandManager()->getUnionFind().unite((colObj0)->getIslandTag(), (colObj1)->getIslandTag());
 				}
 			}
 		}
@@ -783,51 +751,44 @@ void	btDiscreteDynamicsWorld::calculateSimulationIslands()
 
 	//Store the island id in each body
 	getSimulationIslandManager()->storeIslandActivationState(getCollisionWorld());
-
-
 }
 
-
-
-
 class btClosestNotMeConvexResultCallback : public btCollisionWorld::ClosestConvexResultCallback
 {
 public:
-
 	btCollisionObject* m_me;
 	btScalar m_allowedPenetration;
 	btOverlappingPairCache* m_pairCache;
 	btDispatcher* m_dispatcher;
 
 public:
-	btClosestNotMeConvexResultCallback (btCollisionObject* me,const btVector3& fromA,const btVector3& toA,btOverlappingPairCache* pairCache,btDispatcher* dispatcher) :
-	  btCollisionWorld::ClosestConvexResultCallback(fromA,toA),
-		m_me(me),
-		m_allowedPenetration(0.0f),
-		m_pairCache(pairCache),
-		m_dispatcher(dispatcher)
+	btClosestNotMeConvexResultCallback(btCollisionObject* me, const btVector3& fromA, const btVector3& toA, btOverlappingPairCache* pairCache, btDispatcher* dispatcher) : btCollisionWorld::ClosestConvexResultCallback(fromA, toA),
+																																										   m_me(me),
+																																										   m_allowedPenetration(0.0f),
+																																										   m_pairCache(pairCache),
+																																										   m_dispatcher(dispatcher)
 	{
 	}
 
-	virtual btScalar addSingleResult(btCollisionWorld::LocalConvexResult& convexResult,bool normalInWorldSpace)
+	virtual btScalar addSingleResult(btCollisionWorld::LocalConvexResult& convexResult, bool normalInWorldSpace)
 	{
 		if (convexResult.m_hitCollisionObject == m_me)
 			return 1.0f;
 
 		//ignore result if there is no contact response
-		if(!convexResult.m_hitCollisionObject->hasContactResponse())
+		if (!convexResult.m_hitCollisionObject->hasContactResponse())
 			return 1.0f;
 
-		btVector3 linVelA,linVelB;
-		linVelA = m_convexToWorld-m_convexFromWorld;
-		linVelB = btVector3(0,0,0);//toB.getOrigin()-fromB.getOrigin();
+		btVector3 linVelA, linVelB;
+		linVelA = m_convexToWorld - m_convexFromWorld;
+		linVelB = btVector3(0, 0, 0);  //toB.getOrigin()-fromB.getOrigin();
 
-		btVector3 relativeVelocity = (linVelA-linVelB);
+		btVector3 relativeVelocity = (linVelA - linVelB);
 		//don't report time of impact for motion away from the contact normal (or causes minor penetration)
-		if (convexResult.m_hitNormalLocal.dot(relativeVelocity)>=-m_allowedPenetration)
+		if (convexResult.m_hitNormalLocal.dot(relativeVelocity) >= -m_allowedPenetration)
 			return 1.f;
 
-		return ClosestConvexResultCallback::addSingleResult (convexResult, normalInWorldSpace);
+		return ClosestConvexResultCallback::addSingleResult(convexResult, normalInWorldSpace);
 	}
 
 	virtual bool needsCollision(btBroadphaseProxy* proxy0) const
@@ -840,13 +801,13 @@ public:
 		if (!ClosestConvexResultCallback::needsCollision(proxy0))
 			return false;
 
-		btCollisionObject* otherObj = (btCollisionObject*) proxy0->m_clientObject;
+		btCollisionObject* otherObj = (btCollisionObject*)proxy0->m_clientObject;
 
-		if(!m_dispatcher->needsCollision(m_me, otherObj))
+		if (!m_dispatcher->needsCollision(m_me, otherObj))
 			return false;
 
 		//call needsResponse, see http://code.google.com/p/bullet/issues/detail?id=179
-		if (m_dispatcher->needsResponse(m_me,otherObj))
+		if (m_dispatcher->needsResponse(m_me, otherObj))
 		{
 #if 0
 			///don't do CCD when there are already contact points (touching contact/penetration)
@@ -872,28 +833,24 @@ public:
 
 		return false;
 	}
-
-
 };
 
 ///internal debugging variable. this value shouldn't be too high
-int gNumClampedCcdMotions=0;
+int gNumClampedCcdMotions = 0;
 
-
-void btDiscreteDynamicsWorld::createPredictiveContactsInternal( btRigidBody** bodies, int numBodies, btScalar timeStep)
+void btDiscreteDynamicsWorld::createPredictiveContactsInternal(btRigidBody** bodies, int numBodies, btScalar timeStep)
 {
 	btTransform predictedTrans;
-	for ( int i=0;i<numBodies;i++)
+	for (int i = 0; i < numBodies; i++)
 	{
 		btRigidBody* body = bodies[i];
 		body->setHitFraction(1.f);
 
 		if (body->isActive() && (!body->isStaticOrKinematicObject()))
 		{
-
 			body->predictIntegratedTransform(timeStep, predictedTrans);
 
-			btScalar squareMotion = (predictedTrans.getOrigin()-body->getWorldTransform().getOrigin()).length2();
+			btScalar squareMotion = (predictedTrans.getOrigin() - body->getWorldTransform().getOrigin()).length2();
 
 			if (getDispatchInfo().m_useContinuous && body->getCcdSquareMotionThreshold() && body->getCcdSquareMotionThreshold() < squareMotion)
 			{
@@ -905,60 +862,55 @@ void btDiscreteDynamicsWorld::createPredictiveContactsInternal( btRigidBody** bo
 					class StaticOnlyCallback : public btClosestNotMeConvexResultCallback
 					{
 					public:
-
-						StaticOnlyCallback (btCollisionObject* me,const btVector3& fromA,const btVector3& toA,btOverlappingPairCache* pairCache,btDispatcher* dispatcher) :
-						  btClosestNotMeConvexResultCallback(me,fromA,toA,pairCache,dispatcher)
+						StaticOnlyCallback(btCollisionObject* me, const btVector3& fromA, const btVector3& toA, btOverlappingPairCache* pairCache, btDispatcher* dispatcher) : btClosestNotMeConvexResultCallback(me, fromA, toA, pairCache, dispatcher)
 						{
 						}
 
-					  	virtual bool needsCollision(btBroadphaseProxy* proxy0) const
+						virtual bool needsCollision(btBroadphaseProxy* proxy0) const
 						{
-							btCollisionObject* otherObj = (btCollisionObject*) proxy0->m_clientObject;
+							btCollisionObject* otherObj = (btCollisionObject*)proxy0->m_clientObject;
 							if (!otherObj->isStaticOrKinematicObject())
 								return false;
 							return btClosestNotMeConvexResultCallback::needsCollision(proxy0);
 						}
 					};
 
-					StaticOnlyCallback sweepResults(body,body->getWorldTransform().getOrigin(),predictedTrans.getOrigin(),getBroadphase()->getOverlappingPairCache(),getDispatcher());
+					StaticOnlyCallback sweepResults(body, body->getWorldTransform().getOrigin(), predictedTrans.getOrigin(), getBroadphase()->getOverlappingPairCache(), getDispatcher());
 #else
-					btClosestNotMeConvexResultCallback sweepResults(body,body->getWorldTransform().getOrigin(),predictedTrans.getOrigin(),getBroadphase()->getOverlappingPairCache(),getDispatcher());
+					btClosestNotMeConvexResultCallback sweepResults(body, body->getWorldTransform().getOrigin(), predictedTrans.getOrigin(), getBroadphase()->getOverlappingPairCache(), getDispatcher());
 #endif
 					//btConvexShape* convexShape = static_cast<btConvexShape*>(body->getCollisionShape());
-					btSphereShape tmpSphere(body->getCcdSweptSphereRadius());//btConvexShape* convexShape = static_cast<btConvexShape*>(body->getCollisionShape());
-					sweepResults.m_allowedPenetration=getDispatchInfo().m_allowedCcdPenetration;
+					btSphereShape tmpSphere(body->getCcdSweptSphereRadius());  //btConvexShape* convexShape = static_cast<btConvexShape*>(body->getCollisionShape());
+					sweepResults.m_allowedPenetration = getDispatchInfo().m_allowedCcdPenetration;
 
 					sweepResults.m_collisionFilterGroup = body->getBroadphaseProxy()->m_collisionFilterGroup;
-					sweepResults.m_collisionFilterMask  = body->getBroadphaseProxy()->m_collisionFilterMask;
+					sweepResults.m_collisionFilterMask = body->getBroadphaseProxy()->m_collisionFilterMask;
 					btTransform modifiedPredictedTrans = predictedTrans;
 					modifiedPredictedTrans.setBasis(body->getWorldTransform().getBasis());
 
-					convexSweepTest(&tmpSphere,body->getWorldTransform(),modifiedPredictedTrans,sweepResults);
+					convexSweepTest(&tmpSphere, body->getWorldTransform(), modifiedPredictedTrans, sweepResults);
 					if (sweepResults.hasHit() && (sweepResults.m_closestHitFraction < 1.f))
 					{
-
-						btVector3 distVec = (predictedTrans.getOrigin()-body->getWorldTransform().getOrigin())*sweepResults.m_closestHitFraction;
+						btVector3 distVec = (predictedTrans.getOrigin() - body->getWorldTransform().getOrigin()) * sweepResults.m_closestHitFraction;
 						btScalar distance = distVec.dot(-sweepResults.m_hitNormalWorld);
 
-
-						btPersistentManifold* manifold = m_dispatcher1->getNewManifold(body,sweepResults.m_hitCollisionObject);
-                        btMutexLock( &m_predictiveManifoldsMutex );
+						btPersistentManifold* manifold = m_dispatcher1->getNewManifold(body, sweepResults.m_hitCollisionObject);
+						btMutexLock(&m_predictiveManifoldsMutex);
 						m_predictiveManifolds.push_back(manifold);
-                        btMutexUnlock( &m_predictiveManifoldsMutex );
+						btMutexUnlock(&m_predictiveManifoldsMutex);
 
-						btVector3 worldPointB = body->getWorldTransform().getOrigin()+distVec;
-						btVector3 localPointB = sweepResults.m_hitCollisionObject->getWorldTransform().inverse()*worldPointB;
+						btVector3 worldPointB = body->getWorldTransform().getOrigin() + distVec;
+						btVector3 localPointB = sweepResults.m_hitCollisionObject->getWorldTransform().inverse() * worldPointB;
 
-						btManifoldPoint newPoint(btVector3(0,0,0), localPointB,sweepResults.m_hitNormalWorld,distance);
+						btManifoldPoint newPoint(btVector3(0, 0, 0), localPointB, sweepResults.m_hitNormalWorld, distance);
 
 						bool isPredictive = true;
 						int index = manifold->addManifoldPoint(newPoint, isPredictive);
 						btManifoldPoint& pt = manifold->getContactPoint(index);
 						pt.m_combinedRestitution = 0;
-						pt.m_combinedFriction = gCalculateCombinedFrictionCallback(body,sweepResults.m_hitCollisionObject);
+						pt.m_combinedFriction = gCalculateCombinedFrictionCallback(body, sweepResults.m_hitCollisionObject);
 						pt.m_positionWorldOnA = body->getWorldTransform().getOrigin();
 						pt.m_positionWorldOnB = worldPointB;
-
 					}
 				}
 			}
@@ -968,42 +920,39 @@ void btDiscreteDynamicsWorld::createPredictiveContactsInternal( btRigidBody** bo
 
 void btDiscreteDynamicsWorld::releasePredictiveContacts()
 {
-    BT_PROFILE( "release predictive contact manifolds" );
+	BT_PROFILE("release predictive contact manifolds");
 
-    for ( int i = 0; i < m_predictiveManifolds.size(); i++ )
-    {
-        btPersistentManifold* manifold = m_predictiveManifolds[ i ];
-        this->m_dispatcher1->releaseManifold( manifold );
-    }
-    m_predictiveManifolds.clear();
+	for (int i = 0; i < m_predictiveManifolds.size(); i++)
+	{
+		btPersistentManifold* manifold = m_predictiveManifolds[i];
+		this->m_dispatcher1->releaseManifold(manifold);
+	}
+	m_predictiveManifolds.clear();
 }
 
 void btDiscreteDynamicsWorld::createPredictiveContacts(btScalar timeStep)
 {
 	BT_PROFILE("createPredictiveContacts");
-    releasePredictiveContacts();
-    if (m_nonStaticRigidBodies.size() > 0)
-    {
-        createPredictiveContactsInternal( &m_nonStaticRigidBodies[ 0 ], m_nonStaticRigidBodies.size(), timeStep );
-    }
+	releasePredictiveContacts();
+	if (m_nonStaticRigidBodies.size() > 0)
+	{
+		createPredictiveContactsInternal(&m_nonStaticRigidBodies[0], m_nonStaticRigidBodies.size(), timeStep);
+	}
 }
 
-void btDiscreteDynamicsWorld::integrateTransformsInternal( btRigidBody** bodies, int numBodies, btScalar timeStep )
+void btDiscreteDynamicsWorld::integrateTransformsInternal(btRigidBody** bodies, int numBodies, btScalar timeStep)
 {
 	btTransform predictedTrans;
-	for (int i=0;i<numBodies;i++)
+	for (int i = 0; i < numBodies; i++)
 	{
 		btRigidBody* body = bodies[i];
 		body->setHitFraction(1.f);
 
 		if (body->isActive() && (!body->isStaticOrKinematicObject()))
 		{
-
 			body->predictIntegratedTransform(timeStep, predictedTrans);
 
-			btScalar squareMotion = (predictedTrans.getOrigin()-body->getWorldTransform().getOrigin()).length2();
-
-
+			btScalar squareMotion = (predictedTrans.getOrigin() - body->getWorldTransform().getOrigin()).length2();
 
 			if (getDispatchInfo().m_useContinuous && body->getCcdSquareMotionThreshold() && body->getCcdSquareMotionThreshold() < squareMotion)
 			{
@@ -1015,43 +964,40 @@ void btDiscreteDynamicsWorld::integrateTransformsInternal( btRigidBody** bodies,
 					class StaticOnlyCallback : public btClosestNotMeConvexResultCallback
 					{
 					public:
-
-						StaticOnlyCallback (btCollisionObject* me,const btVector3& fromA,const btVector3& toA,btOverlappingPairCache* pairCache,btDispatcher* dispatcher) :
-						  btClosestNotMeConvexResultCallback(me,fromA,toA,pairCache,dispatcher)
+						StaticOnlyCallback(btCollisionObject* me, const btVector3& fromA, const btVector3& toA, btOverlappingPairCache* pairCache, btDispatcher* dispatcher) : btClosestNotMeConvexResultCallback(me, fromA, toA, pairCache, dispatcher)
 						{
 						}
 
-					  	virtual bool needsCollision(btBroadphaseProxy* proxy0) const
+						virtual bool needsCollision(btBroadphaseProxy* proxy0) const
 						{
-							btCollisionObject* otherObj = (btCollisionObject*) proxy0->m_clientObject;
+							btCollisionObject* otherObj = (btCollisionObject*)proxy0->m_clientObject;
 							if (!otherObj->isStaticOrKinematicObject())
 								return false;
 							return btClosestNotMeConvexResultCallback::needsCollision(proxy0);
 						}
 					};
 
-					StaticOnlyCallback sweepResults(body,body->getWorldTransform().getOrigin(),predictedTrans.getOrigin(),getBroadphase()->getOverlappingPairCache(),getDispatcher());
+					StaticOnlyCallback sweepResults(body, body->getWorldTransform().getOrigin(), predictedTrans.getOrigin(), getBroadphase()->getOverlappingPairCache(), getDispatcher());
 #else
-					btClosestNotMeConvexResultCallback sweepResults(body,body->getWorldTransform().getOrigin(),predictedTrans.getOrigin(),getBroadphase()->getOverlappingPairCache(),getDispatcher());
+					btClosestNotMeConvexResultCallback sweepResults(body, body->getWorldTransform().getOrigin(), predictedTrans.getOrigin(), getBroadphase()->getOverlappingPairCache(), getDispatcher());
 #endif
 					//btConvexShape* convexShape = static_cast<btConvexShape*>(body->getCollisionShape());
-					btSphereShape tmpSphere(body->getCcdSweptSphereRadius());//btConvexShape* convexShape = static_cast<btConvexShape*>(body->getCollisionShape());
-					sweepResults.m_allowedPenetration=getDispatchInfo().m_allowedCcdPenetration;
+					btSphereShape tmpSphere(body->getCcdSweptSphereRadius());  //btConvexShape* convexShape = static_cast<btConvexShape*>(body->getCollisionShape());
+					sweepResults.m_allowedPenetration = getDispatchInfo().m_allowedCcdPenetration;
 
 					sweepResults.m_collisionFilterGroup = body->getBroadphaseProxy()->m_collisionFilterGroup;
-					sweepResults.m_collisionFilterMask  = body->getBroadphaseProxy()->m_collisionFilterMask;
+					sweepResults.m_collisionFilterMask = body->getBroadphaseProxy()->m_collisionFilterMask;
 					btTransform modifiedPredictedTrans = predictedTrans;
 					modifiedPredictedTrans.setBasis(body->getWorldTransform().getBasis());
 
-					convexSweepTest(&tmpSphere,body->getWorldTransform(),modifiedPredictedTrans,sweepResults);
+					convexSweepTest(&tmpSphere, body->getWorldTransform(), modifiedPredictedTrans, sweepResults);
 					if (sweepResults.hasHit() && (sweepResults.m_closestHitFraction < 1.f))
 					{
-
 						//printf("clamped integration to hit fraction = %f\n",fraction);
 						body->setHitFraction(sweepResults.m_closestHitFraction);
-						body->predictIntegratedTransform(timeStep*body->getHitFraction(), predictedTrans);
+						body->predictIntegratedTransform(timeStep * body->getHitFraction(), predictedTrans);
 						body->setHitFraction(0.f);
-						body->proceedToTransform( predictedTrans);
+						body->proceedToTransform(predictedTrans);
 
 #if 0
 						btVector3 linVel = body->getLinearVelocity();
@@ -1078,50 +1024,45 @@ void btDiscreteDynamicsWorld::integrateTransformsInternal( btRigidBody** bodies,
 						//btScalar depth = 0.f;
 						//appliedImpulse = resolveSingleCollision(body,(btCollisionObject*)sweepResults.m_hitCollisionObject,sweepResults.m_hitPointWorld,sweepResults.m_hitNormalWorld,getSolverInfo(), depth);
 
-
 #endif
 
-        				continue;
+						continue;
 					}
 				}
 			}
 
-
-			body->proceedToTransform( predictedTrans);
-
+			body->proceedToTransform(predictedTrans);
 		}
-
 	}
-
 }
 
 void btDiscreteDynamicsWorld::integrateTransforms(btScalar timeStep)
 {
 	BT_PROFILE("integrateTransforms");
-    if (m_nonStaticRigidBodies.size() > 0)
-    {
-        integrateTransformsInternal(&m_nonStaticRigidBodies[0], m_nonStaticRigidBodies.size(), timeStep);
-    }
+	if (m_nonStaticRigidBodies.size() > 0)
+	{
+		integrateTransformsInternal(&m_nonStaticRigidBodies[0], m_nonStaticRigidBodies.size(), timeStep);
+	}
 
-    ///this should probably be switched on by default, but it is not well tested yet
+	///this should probably be switched on by default, but it is not well tested yet
 	if (m_applySpeculativeContactRestitution)
 	{
 		BT_PROFILE("apply speculative contact restitution");
-		for (int i=0;i<m_predictiveManifolds.size();i++)
+		for (int i = 0; i < m_predictiveManifolds.size(); i++)
 		{
 			btPersistentManifold* manifold = m_predictiveManifolds[i];
 			btRigidBody* body0 = btRigidBody::upcast((btCollisionObject*)manifold->getBody0());
 			btRigidBody* body1 = btRigidBody::upcast((btCollisionObject*)manifold->getBody1());
 
-			for (int p=0;p<manifold->getNumContacts();p++)
+			for (int p = 0; p < manifold->getNumContacts(); p++)
 			{
 				const btManifoldPoint& pt = manifold->getContactPoint(p);
 				btScalar combinedRestitution = gCalculateCombinedRestitutionCallback(body0, body1);
 
-				if (combinedRestitution>0 && pt.m_appliedImpulse != 0.f)
+				if (combinedRestitution > 0 && pt.m_appliedImpulse != 0.f)
 				//if (pt.getDistance()>0 && combinedRestitution>0 && pt.m_appliedImpulse != 0.f)
 				{
-					btVector3 imp = -pt.m_normalWorldOnB * pt.m_appliedImpulse* combinedRestitution;
+					btVector3 imp = -pt.m_normalWorldOnB * pt.m_appliedImpulse * combinedRestitution;
 
 					const btVector3& pos1 = pt.getPositionWorldOnA();
 					const btVector3& pos2 = pt.getPositionWorldOnB();
@@ -1130,23 +1071,19 @@ void btDiscreteDynamicsWorld::integrateTransforms(btScalar timeStep)
 					btVector3 rel_pos1 = pos2 - body1->getWorldTransform().getOrigin();
 
 					if (body0)
-						body0->applyImpulse(imp,rel_pos0);
+						body0->applyImpulse(imp, rel_pos0);
 					if (body1)
-						body1->applyImpulse(-imp,rel_pos1);
+						body1->applyImpulse(-imp, rel_pos1);
 				}
 			}
 		}
 	}
 }
 
-
-
-
-
-void	btDiscreteDynamicsWorld::predictUnconstraintMotion(btScalar timeStep)
+void btDiscreteDynamicsWorld::predictUnconstraintMotion(btScalar timeStep)
 {
 	BT_PROFILE("predictUnconstraintMotion");
-	for ( int i=0;i<m_nonStaticRigidBodies.size();i++)
+	for (int i = 0; i < m_nonStaticRigidBodies.size(); i++)
 	{
 		btRigidBody* body = m_nonStaticRigidBodies[i];
 		if (!body->isStaticOrKinematicObject())
@@ -1155,179 +1092,171 @@ void	btDiscreteDynamicsWorld::predictUnconstraintMotion(btScalar timeStep)
 
 			body->applyDamping(timeStep);
 
-			body->predictIntegratedTransform(timeStep,body->getInterpolationWorldTransform());
+			body->predictIntegratedTransform(timeStep, body->getInterpolationWorldTransform());
 		}
 	}
 }
 
-
-void	btDiscreteDynamicsWorld::startProfiling(btScalar timeStep)
+void btDiscreteDynamicsWorld::startProfiling(btScalar timeStep)
 {
 	(void)timeStep;
 
 #ifndef BT_NO_PROFILE
 	CProfileManager::Reset();
-#endif //BT_NO_PROFILE
-
+#endif  //BT_NO_PROFILE
 }
 
-
-
-
-
-
 void btDiscreteDynamicsWorld::debugDrawConstraint(btTypedConstraint* constraint)
 {
 	bool drawFrames = (getDebugDrawer()->getDebugMode() & btIDebugDraw::DBG_DrawConstraints) != 0;
 	bool drawLimits = (getDebugDrawer()->getDebugMode() & btIDebugDraw::DBG_DrawConstraintLimits) != 0;
 	btScalar dbgDrawSize = constraint->getDbgDrawSize();
-	if(dbgDrawSize <= btScalar(0.f))
+	if (dbgDrawSize <= btScalar(0.f))
 	{
 		return;
 	}
 
-	switch(constraint->getConstraintType())
+	switch (constraint->getConstraintType())
 	{
 		case POINT2POINT_CONSTRAINT_TYPE:
-			{
-				btPoint2PointConstraint* p2pC = (btPoint2PointConstraint*)constraint;
-				btTransform tr;
-				tr.setIdentity();
-				btVector3 pivot = p2pC->getPivotInA();
-				pivot = p2pC->getRigidBodyA().getCenterOfMassTransform() * pivot;
-				tr.setOrigin(pivot);
-				getDebugDrawer()->drawTransform(tr, dbgDrawSize);
-				// that ideally should draw the same frame
-				pivot = p2pC->getPivotInB();
-				pivot = p2pC->getRigidBodyB().getCenterOfMassTransform() * pivot;
-				tr.setOrigin(pivot);
-				if(drawFrames) getDebugDrawer()->drawTransform(tr, dbgDrawSize);
-			}
-			break;
+		{
+			btPoint2PointConstraint* p2pC = (btPoint2PointConstraint*)constraint;
+			btTransform tr;
+			tr.setIdentity();
+			btVector3 pivot = p2pC->getPivotInA();
+			pivot = p2pC->getRigidBodyA().getCenterOfMassTransform() * pivot;
+			tr.setOrigin(pivot);
+			getDebugDrawer()->drawTransform(tr, dbgDrawSize);
+			// that ideally should draw the same frame
+			pivot = p2pC->getPivotInB();
+			pivot = p2pC->getRigidBodyB().getCenterOfMassTransform() * pivot;
+			tr.setOrigin(pivot);
+			if (drawFrames) getDebugDrawer()->drawTransform(tr, dbgDrawSize);
+		}
+		break;
 		case HINGE_CONSTRAINT_TYPE:
+		{
+			btHingeConstraint* pHinge = (btHingeConstraint*)constraint;
+			btTransform tr = pHinge->getRigidBodyA().getCenterOfMassTransform() * pHinge->getAFrame();
+			if (drawFrames) getDebugDrawer()->drawTransform(tr, dbgDrawSize);
+			tr = pHinge->getRigidBodyB().getCenterOfMassTransform() * pHinge->getBFrame();
+			if (drawFrames) getDebugDrawer()->drawTransform(tr, dbgDrawSize);
+			btScalar minAng = pHinge->getLowerLimit();
+			btScalar maxAng = pHinge->getUpperLimit();
+			if (minAng == maxAng)
 			{
-				btHingeConstraint* pHinge = (btHingeConstraint*)constraint;
-				btTransform tr = pHinge->getRigidBodyA().getCenterOfMassTransform() * pHinge->getAFrame();
-				if(drawFrames) getDebugDrawer()->drawTransform(tr, dbgDrawSize);
-				tr = pHinge->getRigidBodyB().getCenterOfMassTransform() * pHinge->getBFrame();
-				if(drawFrames) getDebugDrawer()->drawTransform(tr, dbgDrawSize);
-				btScalar minAng = pHinge->getLowerLimit();
-				btScalar maxAng = pHinge->getUpperLimit();
-				if(minAng == maxAng)
-				{
-					break;
-				}
-				bool drawSect = true;
-				if(!pHinge->hasLimit())
-				{
-					minAng = btScalar(0.f);
-					maxAng = SIMD_2_PI;
-					drawSect = false;
-				}
-				if(drawLimits)
-				{
-					btVector3& center = tr.getOrigin();
-					btVector3 normal = tr.getBasis().getColumn(2);
-					btVector3 axis = tr.getBasis().getColumn(0);
-					getDebugDrawer()->drawArc(center, normal, axis, dbgDrawSize, dbgDrawSize, minAng, maxAng, btVector3(0,0,0), drawSect);
-				}
+				break;
 			}
-			break;
+			bool drawSect = true;
+			if (!pHinge->hasLimit())
+			{
+				minAng = btScalar(0.f);
+				maxAng = SIMD_2_PI;
+				drawSect = false;
+			}
+			if (drawLimits)
+			{
+				btVector3& center = tr.getOrigin();
+				btVector3 normal = tr.getBasis().getColumn(2);
+				btVector3 axis = tr.getBasis().getColumn(0);
+				getDebugDrawer()->drawArc(center, normal, axis, dbgDrawSize, dbgDrawSize, minAng, maxAng, btVector3(0, 0, 0), drawSect);
+			}
+		}
+		break;
 		case CONETWIST_CONSTRAINT_TYPE:
+		{
+			btConeTwistConstraint* pCT = (btConeTwistConstraint*)constraint;
+			btTransform tr = pCT->getRigidBodyA().getCenterOfMassTransform() * pCT->getAFrame();
+			if (drawFrames) getDebugDrawer()->drawTransform(tr, dbgDrawSize);
+			tr = pCT->getRigidBodyB().getCenterOfMassTransform() * pCT->getBFrame();
+			if (drawFrames) getDebugDrawer()->drawTransform(tr, dbgDrawSize);
+			if (drawLimits)
 			{
-				btConeTwistConstraint* pCT = (btConeTwistConstraint*)constraint;
-				btTransform tr = pCT->getRigidBodyA().getCenterOfMassTransform() * pCT->getAFrame();
-				if(drawFrames) getDebugDrawer()->drawTransform(tr, dbgDrawSize);
-				tr = pCT->getRigidBodyB().getCenterOfMassTransform() * pCT->getBFrame();
-				if(drawFrames) getDebugDrawer()->drawTransform(tr, dbgDrawSize);
-				if(drawLimits)
+				//const btScalar length = btScalar(5);
+				const btScalar length = dbgDrawSize;
+				static int nSegments = 8 * 4;
+				btScalar fAngleInRadians = btScalar(2. * 3.1415926) * (btScalar)(nSegments - 1) / btScalar(nSegments);
+				btVector3 pPrev = pCT->GetPointForAngle(fAngleInRadians, length);
+				pPrev = tr * pPrev;
+				for (int i = 0; i < nSegments; i++)
 				{
-					//const btScalar length = btScalar(5);
-					const btScalar length = dbgDrawSize;
-					static int nSegments = 8*4;
-					btScalar fAngleInRadians = btScalar(2.*3.1415926) * (btScalar)(nSegments-1)/btScalar(nSegments);
-					btVector3 pPrev = pCT->GetPointForAngle(fAngleInRadians, length);
-					pPrev = tr * pPrev;
-					for (int i=0; i<nSegments; i++)
-					{
-						fAngleInRadians = btScalar(2.*3.1415926) * (btScalar)i/btScalar(nSegments);
-						btVector3 pCur = pCT->GetPointForAngle(fAngleInRadians, length);
-						pCur = tr * pCur;
-						getDebugDrawer()->drawLine(pPrev, pCur, btVector3(0,0,0));
+					fAngleInRadians = btScalar(2. * 3.1415926) * (btScalar)i / btScalar(nSegments);
+					btVector3 pCur = pCT->GetPointForAngle(fAngleInRadians, length);
+					pCur = tr * pCur;
+					getDebugDrawer()->drawLine(pPrev, pCur, btVector3(0, 0, 0));
 
-						if (i%(nSegments/8) == 0)
-							getDebugDrawer()->drawLine(tr.getOrigin(), pCur, btVector3(0,0,0));
-
-						pPrev = pCur;
-					}
-					btScalar tws = pCT->getTwistSpan();
-					btScalar twa = pCT->getTwistAngle();
-					bool useFrameB = (pCT->getRigidBodyB().getInvMass() > btScalar(0.f));
-					if(useFrameB)
-					{
-						tr = pCT->getRigidBodyB().getCenterOfMassTransform() * pCT->getBFrame();
-					}
-					else
-					{
-						tr = pCT->getRigidBodyA().getCenterOfMassTransform() * pCT->getAFrame();
-					}
-					btVector3 pivot = tr.getOrigin();
-					btVector3 normal = tr.getBasis().getColumn(0);
-					btVector3 axis1 = tr.getBasis().getColumn(1);
-					getDebugDrawer()->drawArc(pivot, normal, axis1, dbgDrawSize, dbgDrawSize, -twa-tws, -twa+tws, btVector3(0,0,0), true);
+					if (i % (nSegments / 8) == 0)
+						getDebugDrawer()->drawLine(tr.getOrigin(), pCur, btVector3(0, 0, 0));
 
+					pPrev = pCur;
 				}
+				btScalar tws = pCT->getTwistSpan();
+				btScalar twa = pCT->getTwistAngle();
+				bool useFrameB = (pCT->getRigidBodyB().getInvMass() > btScalar(0.f));
+				if (useFrameB)
+				{
+					tr = pCT->getRigidBodyB().getCenterOfMassTransform() * pCT->getBFrame();
+				}
+				else
+				{
+					tr = pCT->getRigidBodyA().getCenterOfMassTransform() * pCT->getAFrame();
+				}
+				btVector3 pivot = tr.getOrigin();
+				btVector3 normal = tr.getBasis().getColumn(0);
+				btVector3 axis1 = tr.getBasis().getColumn(1);
+				getDebugDrawer()->drawArc(pivot, normal, axis1, dbgDrawSize, dbgDrawSize, -twa - tws, -twa + tws, btVector3(0, 0, 0), true);
 			}
-			break;
+		}
+		break;
 		case D6_SPRING_CONSTRAINT_TYPE:
 		case D6_CONSTRAINT_TYPE:
+		{
+			btGeneric6DofConstraint* p6DOF = (btGeneric6DofConstraint*)constraint;
+			btTransform tr = p6DOF->getCalculatedTransformA();
+			if (drawFrames) getDebugDrawer()->drawTransform(tr, dbgDrawSize);
+			tr = p6DOF->getCalculatedTransformB();
+			if (drawFrames) getDebugDrawer()->drawTransform(tr, dbgDrawSize);
+			if (drawLimits)
 			{
-				btGeneric6DofConstraint* p6DOF = (btGeneric6DofConstraint*)constraint;
-				btTransform tr = p6DOF->getCalculatedTransformA();
-				if(drawFrames) getDebugDrawer()->drawTransform(tr, dbgDrawSize);
+				tr = p6DOF->getCalculatedTransformA();
+				const btVector3& center = p6DOF->getCalculatedTransformB().getOrigin();
+				btVector3 up = tr.getBasis().getColumn(2);
+				btVector3 axis = tr.getBasis().getColumn(0);
+				btScalar minTh = p6DOF->getRotationalLimitMotor(1)->m_loLimit;
+				btScalar maxTh = p6DOF->getRotationalLimitMotor(1)->m_hiLimit;
+				btScalar minPs = p6DOF->getRotationalLimitMotor(2)->m_loLimit;
+				btScalar maxPs = p6DOF->getRotationalLimitMotor(2)->m_hiLimit;
+				getDebugDrawer()->drawSpherePatch(center, up, axis, dbgDrawSize * btScalar(.9f), minTh, maxTh, minPs, maxPs, btVector3(0, 0, 0));
+				axis = tr.getBasis().getColumn(1);
+				btScalar ay = p6DOF->getAngle(1);
+				btScalar az = p6DOF->getAngle(2);
+				btScalar cy = btCos(ay);
+				btScalar sy = btSin(ay);
+				btScalar cz = btCos(az);
+				btScalar sz = btSin(az);
+				btVector3 ref;
+				ref[0] = cy * cz * axis[0] + cy * sz * axis[1] - sy * axis[2];
+				ref[1] = -sz * axis[0] + cz * axis[1];
+				ref[2] = cz * sy * axis[0] + sz * sy * axis[1] + cy * axis[2];
 				tr = p6DOF->getCalculatedTransformB();
-				if(drawFrames) getDebugDrawer()->drawTransform(tr, dbgDrawSize);
-				if(drawLimits)
+				btVector3 normal = -tr.getBasis().getColumn(0);
+				btScalar minFi = p6DOF->getRotationalLimitMotor(0)->m_loLimit;
+				btScalar maxFi = p6DOF->getRotationalLimitMotor(0)->m_hiLimit;
+				if (minFi > maxFi)
 				{
-					tr = p6DOF->getCalculatedTransformA();
-					const btVector3& center = p6DOF->getCalculatedTransformB().getOrigin();
-					btVector3 up = tr.getBasis().getColumn(2);
-					btVector3 axis = tr.getBasis().getColumn(0);
-					btScalar minTh = p6DOF->getRotationalLimitMotor(1)->m_loLimit;
-					btScalar maxTh = p6DOF->getRotationalLimitMotor(1)->m_hiLimit;
-					btScalar minPs = p6DOF->getRotationalLimitMotor(2)->m_loLimit;
-					btScalar maxPs = p6DOF->getRotationalLimitMotor(2)->m_hiLimit;
-					getDebugDrawer()->drawSpherePatch(center, up, axis, dbgDrawSize * btScalar(.9f), minTh, maxTh, minPs, maxPs, btVector3(0,0,0));
-					axis = tr.getBasis().getColumn(1);
-					btScalar ay = p6DOF->getAngle(1);
-					btScalar az = p6DOF->getAngle(2);
-					btScalar cy = btCos(ay);
-					btScalar sy = btSin(ay);
-					btScalar cz = btCos(az);
-					btScalar sz = btSin(az);
-					btVector3 ref;
-					ref[0] = cy*cz*axis[0] + cy*sz*axis[1] - sy*axis[2];
-					ref[1] = -sz*axis[0] + cz*axis[1];
-					ref[2] = cz*sy*axis[0] + sz*sy*axis[1] + cy*axis[2];
-					tr = p6DOF->getCalculatedTransformB();
-					btVector3 normal = -tr.getBasis().getColumn(0);
-					btScalar minFi = p6DOF->getRotationalLimitMotor(0)->m_loLimit;
-					btScalar maxFi = p6DOF->getRotationalLimitMotor(0)->m_hiLimit;
-					if(minFi > maxFi)
-					{
-						getDebugDrawer()->drawArc(center, normal, ref, dbgDrawSize, dbgDrawSize, -SIMD_PI, SIMD_PI, btVector3(0,0,0), false);
-					}
-					else if(minFi < maxFi)
-					{
-						getDebugDrawer()->drawArc(center, normal, ref, dbgDrawSize, dbgDrawSize, minFi, maxFi, btVector3(0,0,0), true);
-					}
-					tr = p6DOF->getCalculatedTransformA();
-					btVector3 bbMin = p6DOF->getTranslationalLimitMotor()->m_lowerLimit;
-					btVector3 bbMax = p6DOF->getTranslationalLimitMotor()->m_upperLimit;
-					getDebugDrawer()->drawBox(bbMin, bbMax, tr, btVector3(0,0,0));
+					getDebugDrawer()->drawArc(center, normal, ref, dbgDrawSize, dbgDrawSize, -SIMD_PI, SIMD_PI, btVector3(0, 0, 0), false);
 				}
+				else if (minFi < maxFi)
+				{
+					getDebugDrawer()->drawArc(center, normal, ref, dbgDrawSize, dbgDrawSize, minFi, maxFi, btVector3(0, 0, 0), true);
+				}
+				tr = p6DOF->getCalculatedTransformA();
+				btVector3 bbMin = p6DOF->getTranslationalLimitMotor()->m_lowerLimit;
+				btVector3 bbMax = p6DOF->getTranslationalLimitMotor()->m_upperLimit;
+				getDebugDrawer()->drawBox(bbMin, bbMax, tr, btVector3(0, 0, 0));
 			}
-			break;
+		}
+		break;
 		///note: the code for D6_SPRING_2_CONSTRAINT_TYPE is identical to D6_CONSTRAINT_TYPE, the D6_CONSTRAINT_TYPE+D6_SPRING_CONSTRAINT_TYPE will likely become obsolete/deprecated at some stage
 		case D6_SPRING_2_CONSTRAINT_TYPE:
 		{
@@ -1359,9 +1288,9 @@ void btDiscreteDynamicsWorld::debugDrawConstraint(btTypedConstraint* constraint)
 					btScalar cz = btCos(az);
 					btScalar sz = btSin(az);
 					btVector3 ref;
-					ref[0] = cy*cz*axis[0] + cy*sz*axis[1] - sy*axis[2];
-					ref[1] = -sz*axis[0] + cz*axis[1];
-					ref[2] = cz*sy*axis[0] + sz*sy*axis[1] + cy*axis[2];
+					ref[0] = cy * cz * axis[0] + cy * sz * axis[1] - sy * axis[2];
+					ref[1] = -sz * axis[0] + cz * axis[1];
+					ref[2] = cz * sy * axis[0] + sz * sy * axis[1] + cy * axis[2];
 					tr = p6DOF->getCalculatedTransformB();
 					btVector3 normal = -tr.getBasis().getColumn(0);
 					btScalar minFi = p6DOF->getRotationalLimitMotor(0)->m_loLimit;
@@ -1383,42 +1312,38 @@ void btDiscreteDynamicsWorld::debugDrawConstraint(btTypedConstraint* constraint)
 			break;
 		}
 		case SLIDER_CONSTRAINT_TYPE:
+		{
+			btSliderConstraint* pSlider = (btSliderConstraint*)constraint;
+			btTransform tr = pSlider->getCalculatedTransformA();
+			if (drawFrames) getDebugDrawer()->drawTransform(tr, dbgDrawSize);
+			tr = pSlider->getCalculatedTransformB();
+			if (drawFrames) getDebugDrawer()->drawTransform(tr, dbgDrawSize);
+			if (drawLimits)
 			{
-				btSliderConstraint* pSlider = (btSliderConstraint*)constraint;
-				btTransform tr = pSlider->getCalculatedTransformA();
-				if(drawFrames) getDebugDrawer()->drawTransform(tr, dbgDrawSize);
-				tr = pSlider->getCalculatedTransformB();
-				if(drawFrames) getDebugDrawer()->drawTransform(tr, dbgDrawSize);
-				if(drawLimits)
-				{
-					btTransform tr = pSlider->getUseLinearReferenceFrameA() ? pSlider->getCalculatedTransformA() : pSlider->getCalculatedTransformB();
-					btVector3 li_min = tr * btVector3(pSlider->getLowerLinLimit(), 0.f, 0.f);
-					btVector3 li_max = tr * btVector3(pSlider->getUpperLinLimit(), 0.f, 0.f);
-					getDebugDrawer()->drawLine(li_min, li_max, btVector3(0, 0, 0));
-					btVector3 normal = tr.getBasis().getColumn(0);
-					btVector3 axis = tr.getBasis().getColumn(1);
-					btScalar a_min = pSlider->getLowerAngLimit();
-					btScalar a_max = pSlider->getUpperAngLimit();
-					const btVector3& center = pSlider->getCalculatedTransformB().getOrigin();
-					getDebugDrawer()->drawArc(center, normal, axis, dbgDrawSize, dbgDrawSize, a_min, a_max, btVector3(0,0,0), true);
-				}
+				btTransform tr = pSlider->getUseLinearReferenceFrameA() ? pSlider->getCalculatedTransformA() : pSlider->getCalculatedTransformB();
+				btVector3 li_min = tr * btVector3(pSlider->getLowerLinLimit(), 0.f, 0.f);
+				btVector3 li_max = tr * btVector3(pSlider->getUpperLinLimit(), 0.f, 0.f);
+				getDebugDrawer()->drawLine(li_min, li_max, btVector3(0, 0, 0));
+				btVector3 normal = tr.getBasis().getColumn(0);
+				btVector3 axis = tr.getBasis().getColumn(1);
+				btScalar a_min = pSlider->getLowerAngLimit();
+				btScalar a_max = pSlider->getUpperAngLimit();
+				const btVector3& center = pSlider->getCalculatedTransformB().getOrigin();
+				getDebugDrawer()->drawArc(center, normal, axis, dbgDrawSize, dbgDrawSize, a_min, a_max, btVector3(0, 0, 0), true);
 			}
-			break;
-		default :
+		}
+		break;
+		default:
 			break;
 	}
 	return;
 }
 
-
-
-
-
-void	btDiscreteDynamicsWorld::setConstraintSolver(btConstraintSolver* solver)
+void btDiscreteDynamicsWorld::setConstraintSolver(btConstraintSolver* solver)
 {
 	if (m_ownsConstraintSolver)
 	{
-		btAlignedFree( m_constraintSolver);
+		btAlignedFree(m_constraintSolver);
 	}
 	m_ownsConstraintSolver = false;
 	m_constraintSolver = solver;
@@ -1430,8 +1355,7 @@ btConstraintSolver* btDiscreteDynamicsWorld::getConstraintSolver()
 	return m_constraintSolver;
 }
 
-
-int		btDiscreteDynamicsWorld::getNumConstraints() const
+int btDiscreteDynamicsWorld::getNumConstraints() const
 {
 	return int(m_constraints.size());
 }
@@ -1444,93 +1368,87 @@ const btTypedConstraint* btDiscreteDynamicsWorld::getConstraint(int index) const
 	return m_constraints[index];
 }
 
-
-
-void	btDiscreteDynamicsWorld::serializeRigidBodies(btSerializer* serializer)
+void btDiscreteDynamicsWorld::serializeRigidBodies(btSerializer* serializer)
 {
 	int i;
 	//serialize all collision objects
-	for (i=0;i<m_collisionObjects.size();i++)
+	for (i = 0; i < m_collisionObjects.size(); i++)
 	{
 		btCollisionObject* colObj = m_collisionObjects[i];
 		if (colObj->getInternalType() & btCollisionObject::CO_RIGID_BODY)
 		{
 			int len = colObj->calculateSerializeBufferSize();
-			btChunk* chunk = serializer->allocate(len,1);
+			btChunk* chunk = serializer->allocate(len, 1);
 			const char* structType = colObj->serialize(chunk->m_oldPtr, serializer);
-			serializer->finalizeChunk(chunk,structType,BT_RIGIDBODY_CODE,colObj);
+			serializer->finalizeChunk(chunk, structType, BT_RIGIDBODY_CODE, colObj);
 		}
 	}
 
-	for (i=0;i<m_constraints.size();i++)
+	for (i = 0; i < m_constraints.size(); i++)
 	{
 		btTypedConstraint* constraint = m_constraints[i];
 		int size = constraint->calculateSerializeBufferSize();
-		btChunk* chunk = serializer->allocate(size,1);
-		const char* structType = constraint->serialize(chunk->m_oldPtr,serializer);
-		serializer->finalizeChunk(chunk,structType,BT_CONSTRAINT_CODE,constraint);
+		btChunk* chunk = serializer->allocate(size, 1);
+		const char* structType = constraint->serialize(chunk->m_oldPtr, serializer);
+		serializer->finalizeChunk(chunk, structType, BT_CONSTRAINT_CODE, constraint);
 	}
 }
 
-
-
-
-void	btDiscreteDynamicsWorld::serializeDynamicsWorldInfo(btSerializer* serializer)
+void btDiscreteDynamicsWorld::serializeDynamicsWorldInfo(btSerializer* serializer)
 {
 #ifdef BT_USE_DOUBLE_PRECISION
-		int len = sizeof(btDynamicsWorldDoubleData);
-		btChunk* chunk = serializer->allocate(len,1);
-		btDynamicsWorldDoubleData* worldInfo = (btDynamicsWorldDoubleData*)chunk->m_oldPtr;
-#else//BT_USE_DOUBLE_PRECISION
-		int len = sizeof(btDynamicsWorldFloatData);
-		btChunk* chunk = serializer->allocate(len,1);
-		btDynamicsWorldFloatData* worldInfo = (btDynamicsWorldFloatData*)chunk->m_oldPtr;
-#endif//BT_USE_DOUBLE_PRECISION
+	int len = sizeof(btDynamicsWorldDoubleData);
+	btChunk* chunk = serializer->allocate(len, 1);
+	btDynamicsWorldDoubleData* worldInfo = (btDynamicsWorldDoubleData*)chunk->m_oldPtr;
+#else   //BT_USE_DOUBLE_PRECISION
+	int len = sizeof(btDynamicsWorldFloatData);
+	btChunk* chunk = serializer->allocate(len, 1);
+	btDynamicsWorldFloatData* worldInfo = (btDynamicsWorldFloatData*)chunk->m_oldPtr;
+#endif  //BT_USE_DOUBLE_PRECISION
 
-		memset(worldInfo ,0x00,len);
+	memset(worldInfo, 0x00, len);
 
-		m_gravity.serialize(worldInfo->m_gravity);
-		worldInfo->m_solverInfo.m_tau = getSolverInfo().m_tau;
-		worldInfo->m_solverInfo.m_damping = getSolverInfo().m_damping;
-		worldInfo->m_solverInfo.m_friction = getSolverInfo().m_friction;
-		worldInfo->m_solverInfo.m_timeStep = getSolverInfo().m_timeStep;
+	m_gravity.serialize(worldInfo->m_gravity);
+	worldInfo->m_solverInfo.m_tau = getSolverInfo().m_tau;
+	worldInfo->m_solverInfo.m_damping = getSolverInfo().m_damping;
+	worldInfo->m_solverInfo.m_friction = getSolverInfo().m_friction;
+	worldInfo->m_solverInfo.m_timeStep = getSolverInfo().m_timeStep;
 
-		worldInfo->m_solverInfo.m_restitution = getSolverInfo().m_restitution;
-		worldInfo->m_solverInfo.m_maxErrorReduction = getSolverInfo().m_maxErrorReduction;
-		worldInfo->m_solverInfo.m_sor = getSolverInfo().m_sor;
-		worldInfo->m_solverInfo.m_erp = getSolverInfo().m_erp;
+	worldInfo->m_solverInfo.m_restitution = getSolverInfo().m_restitution;
+	worldInfo->m_solverInfo.m_maxErrorReduction = getSolverInfo().m_maxErrorReduction;
+	worldInfo->m_solverInfo.m_sor = getSolverInfo().m_sor;
+	worldInfo->m_solverInfo.m_erp = getSolverInfo().m_erp;
 
-		worldInfo->m_solverInfo.m_erp2 = getSolverInfo().m_erp2;
-		worldInfo->m_solverInfo.m_globalCfm = getSolverInfo().m_globalCfm;
-		worldInfo->m_solverInfo.m_splitImpulsePenetrationThreshold = getSolverInfo().m_splitImpulsePenetrationThreshold;
-		worldInfo->m_solverInfo.m_splitImpulseTurnErp = getSolverInfo().m_splitImpulseTurnErp;
+	worldInfo->m_solverInfo.m_erp2 = getSolverInfo().m_erp2;
+	worldInfo->m_solverInfo.m_globalCfm = getSolverInfo().m_globalCfm;
+	worldInfo->m_solverInfo.m_splitImpulsePenetrationThreshold = getSolverInfo().m_splitImpulsePenetrationThreshold;
+	worldInfo->m_solverInfo.m_splitImpulseTurnErp = getSolverInfo().m_splitImpulseTurnErp;
 
-		worldInfo->m_solverInfo.m_linearSlop = getSolverInfo().m_linearSlop;
-		worldInfo->m_solverInfo.m_warmstartingFactor = getSolverInfo().m_warmstartingFactor;
-		worldInfo->m_solverInfo.m_maxGyroscopicForce = getSolverInfo().m_maxGyroscopicForce;
-		worldInfo->m_solverInfo.m_singleAxisRollingFrictionThreshold = getSolverInfo().m_singleAxisRollingFrictionThreshold;
+	worldInfo->m_solverInfo.m_linearSlop = getSolverInfo().m_linearSlop;
+	worldInfo->m_solverInfo.m_warmstartingFactor = getSolverInfo().m_warmstartingFactor;
+	worldInfo->m_solverInfo.m_maxGyroscopicForce = getSolverInfo().m_maxGyroscopicForce;
+	worldInfo->m_solverInfo.m_singleAxisRollingFrictionThreshold = getSolverInfo().m_singleAxisRollingFrictionThreshold;
 
-		worldInfo->m_solverInfo.m_numIterations = getSolverInfo().m_numIterations;
-		worldInfo->m_solverInfo.m_solverMode = getSolverInfo().m_solverMode;
-		worldInfo->m_solverInfo.m_restingContactRestitutionThreshold = getSolverInfo().m_restingContactRestitutionThreshold;
-		worldInfo->m_solverInfo.m_minimumSolverBatchSize = getSolverInfo().m_minimumSolverBatchSize;
+	worldInfo->m_solverInfo.m_numIterations = getSolverInfo().m_numIterations;
+	worldInfo->m_solverInfo.m_solverMode = getSolverInfo().m_solverMode;
+	worldInfo->m_solverInfo.m_restingContactRestitutionThreshold = getSolverInfo().m_restingContactRestitutionThreshold;
+	worldInfo->m_solverInfo.m_minimumSolverBatchSize = getSolverInfo().m_minimumSolverBatchSize;
 
-		worldInfo->m_solverInfo.m_splitImpulse = getSolverInfo().m_splitImpulse;
+	worldInfo->m_solverInfo.m_splitImpulse = getSolverInfo().m_splitImpulse;
 
-		// Fill padding with zeros to appease msan.
-		memset(worldInfo->m_solverInfo.m_padding, 0, sizeof(worldInfo->m_solverInfo.m_padding));
+	// Fill padding with zeros to appease msan.
+	memset(worldInfo->m_solverInfo.m_padding, 0, sizeof(worldInfo->m_solverInfo.m_padding));
 
 #ifdef BT_USE_DOUBLE_PRECISION
-		const char* structType = "btDynamicsWorldDoubleData";
-#else//BT_USE_DOUBLE_PRECISION
-		const char* structType = "btDynamicsWorldFloatData";
-#endif//BT_USE_DOUBLE_PRECISION
-		serializer->finalizeChunk(chunk,structType,BT_DYNAMICSWORLD_CODE,worldInfo);
+	const char* structType = "btDynamicsWorldDoubleData";
+#else   //BT_USE_DOUBLE_PRECISION
+	const char* structType = "btDynamicsWorldFloatData";
+#endif  //BT_USE_DOUBLE_PRECISION
+	serializer->finalizeChunk(chunk, structType, BT_DYNAMICSWORLD_CODE, worldInfo);
 }
 
-void	btDiscreteDynamicsWorld::serialize(btSerializer* serializer)
+void btDiscreteDynamicsWorld::serialize(btSerializer* serializer)
 {
-
 	serializer->startSerialization();
 
 	serializeDynamicsWorldInfo(serializer);
@@ -1543,4 +1461,3 @@ void	btDiscreteDynamicsWorld::serialize(btSerializer* serializer)
 
 	serializer->finishSerialization();
 }
-
diff --git a/thirdparty/bullet/BulletDynamics/Dynamics/btDiscreteDynamicsWorld.h b/thirdparty/bullet/BulletDynamics/Dynamics/btDiscreteDynamicsWorld.h
index b0d19f48a37..7fe96192138 100644
--- a/thirdparty/bullet/BulletDynamics/Dynamics/btDiscreteDynamicsWorld.h
+++ b/thirdparty/bullet/BulletDynamics/Dynamics/btDiscreteDynamicsWorld.h
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef BT_DISCRETE_DYNAMICS_WORLD_H
 #define BT_DISCRETE_DYNAMICS_WORLD_H
 
@@ -32,159 +31,153 @@ struct InplaceSolverIslandCallback;
 #include "LinearMath/btAlignedObjectArray.h"
 #include "LinearMath/btThreads.h"
 
-
 ///btDiscreteDynamicsWorld provides discrete rigid body simulation
 ///those classes replace the obsolete CcdPhysicsEnvironment/CcdPhysicsController
-ATTRIBUTE_ALIGNED16(class) btDiscreteDynamicsWorld : public btDynamicsWorld
+ATTRIBUTE_ALIGNED16(class)
+btDiscreteDynamicsWorld : public btDynamicsWorld
 {
 protected:
-	
-    btAlignedObjectArray<btTypedConstraint*>	m_sortedConstraints;
-	InplaceSolverIslandCallback* 	m_solverIslandCallback;
+	btAlignedObjectArray<btTypedConstraint*> m_sortedConstraints;
+	InplaceSolverIslandCallback* m_solverIslandCallback;
 
-	btConstraintSolver*	m_constraintSolver;
+	btConstraintSolver* m_constraintSolver;
 
-	btSimulationIslandManager*	m_islandManager;
+	btSimulationIslandManager* m_islandManager;
 
 	btAlignedObjectArray<btTypedConstraint*> m_constraints;
 
 	btAlignedObjectArray<btRigidBody*> m_nonStaticRigidBodies;
 
-	btVector3	m_gravity;
+	btVector3 m_gravity;
 
 	//for variable timesteps
-	btScalar	m_localTime;
-	btScalar	m_fixedTimeStep;
+	btScalar m_localTime;
+	btScalar m_fixedTimeStep;
 	//for variable timesteps
 
-	bool	m_ownsIslandManager;
-	bool	m_ownsConstraintSolver;
-	bool	m_synchronizeAllMotionStates;
-	bool	m_applySpeculativeContactRestitution;
+	bool m_ownsIslandManager;
+	bool m_ownsConstraintSolver;
+	bool m_synchronizeAllMotionStates;
+	bool m_applySpeculativeContactRestitution;
 
-	btAlignedObjectArray<btActionInterface*>	m_actions;
-	
-	int	m_profileTimings;
+	btAlignedObjectArray<btActionInterface*> m_actions;
 
-	bool	m_latencyMotionStateInterpolation;
+	int m_profileTimings;
 
-	btAlignedObjectArray<btPersistentManifold*>	m_predictiveManifolds;
-    btSpinMutex m_predictiveManifoldsMutex;  // used to synchronize threads creating predictive contacts
+	bool m_latencyMotionStateInterpolation;
 
-	virtual void	predictUnconstraintMotion(btScalar timeStep);
-	
-    void integrateTransformsInternal( btRigidBody** bodies, int numBodies, btScalar timeStep );  // can be called in parallel
-	virtual void	integrateTransforms(btScalar timeStep);
-		
-	virtual void	calculateSimulationIslands();
+	btAlignedObjectArray<btPersistentManifold*> m_predictiveManifolds;
+	btSpinMutex m_predictiveManifoldsMutex;  // used to synchronize threads creating predictive contacts
 
-	virtual void	solveConstraints(btContactSolverInfo& solverInfo);
-	
-	virtual void	updateActivationState(btScalar timeStep);
+	virtual void predictUnconstraintMotion(btScalar timeStep);
 
-	void	updateActions(btScalar timeStep);
+	void integrateTransformsInternal(btRigidBody * *bodies, int numBodies, btScalar timeStep);  // can be called in parallel
+	virtual void integrateTransforms(btScalar timeStep);
 
-	void	startProfiling(btScalar timeStep);
+	virtual void calculateSimulationIslands();
 
-	virtual void	internalSingleStepSimulation( btScalar timeStep);
+	virtual void solveConstraints(btContactSolverInfo & solverInfo);
 
-    void releasePredictiveContacts();
-    void createPredictiveContactsInternal( btRigidBody** bodies, int numBodies, btScalar timeStep );  // can be called in parallel
-	virtual void	createPredictiveContacts(btScalar timeStep);
+	virtual void updateActivationState(btScalar timeStep);
 
-	virtual void	saveKinematicState(btScalar timeStep);
+	void updateActions(btScalar timeStep);
 
-	void	serializeRigidBodies(btSerializer* serializer);
+	void startProfiling(btScalar timeStep);
 
-	void	serializeDynamicsWorldInfo(btSerializer* serializer);
+	virtual void internalSingleStepSimulation(btScalar timeStep);
+
+	void releasePredictiveContacts();
+	void createPredictiveContactsInternal(btRigidBody * *bodies, int numBodies, btScalar timeStep);  // can be called in parallel
+	virtual void createPredictiveContacts(btScalar timeStep);
+
+	virtual void saveKinematicState(btScalar timeStep);
+
+	void serializeRigidBodies(btSerializer * serializer);
+
+	void serializeDynamicsWorldInfo(btSerializer * serializer);
 
 public:
-
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
 	///this btDiscreteDynamicsWorld constructor gets created objects from the user, and will not delete those
-	btDiscreteDynamicsWorld(btDispatcher* dispatcher,btBroadphaseInterface* pairCache,btConstraintSolver* constraintSolver,btCollisionConfiguration* collisionConfiguration);
+	btDiscreteDynamicsWorld(btDispatcher * dispatcher, btBroadphaseInterface * pairCache, btConstraintSolver * constraintSolver, btCollisionConfiguration * collisionConfiguration);
 
 	virtual ~btDiscreteDynamicsWorld();
 
 	///if maxSubSteps > 0, it will interpolate motion between fixedTimeStep's
-	virtual int	stepSimulation( btScalar timeStep,int maxSubSteps=1, btScalar fixedTimeStep=btScalar(1.)/btScalar(60.));
+	virtual int stepSimulation(btScalar timeStep, int maxSubSteps = 1, btScalar fixedTimeStep = btScalar(1.) / btScalar(60.));
 
-
-	virtual void	synchronizeMotionStates();
+	virtual void synchronizeMotionStates();
 
 	///this can be useful to synchronize a single rigid body -> graphics object
-	void	synchronizeSingleMotionState(btRigidBody* body);
+	void synchronizeSingleMotionState(btRigidBody * body);
 
-	virtual void	addConstraint(btTypedConstraint* constraint, bool disableCollisionsBetweenLinkedBodies=false);
+	virtual void addConstraint(btTypedConstraint * constraint, bool disableCollisionsBetweenLinkedBodies = false);
 
-	virtual void	removeConstraint(btTypedConstraint* constraint);
+	virtual void removeConstraint(btTypedConstraint * constraint);
 
-	virtual void	addAction(btActionInterface*);
+	virtual void addAction(btActionInterface*);
 
-	virtual void	removeAction(btActionInterface*);
-	
-	btSimulationIslandManager*	getSimulationIslandManager()
+	virtual void removeAction(btActionInterface*);
+
+	btSimulationIslandManager* getSimulationIslandManager()
 	{
 		return m_islandManager;
 	}
 
-	const btSimulationIslandManager*	getSimulationIslandManager() const 
+	const btSimulationIslandManager* getSimulationIslandManager() const
 	{
 		return m_islandManager;
 	}
 
-	btCollisionWorld*	getCollisionWorld()
+	btCollisionWorld* getCollisionWorld()
 	{
 		return this;
 	}
 
-	virtual void	setGravity(const btVector3& gravity);
+	virtual void setGravity(const btVector3& gravity);
 
-	virtual btVector3 getGravity () const;
+	virtual btVector3 getGravity() const;
 
-	virtual void	addCollisionObject(btCollisionObject* collisionObject, int collisionFilterGroup=btBroadphaseProxy::StaticFilter, int collisionFilterMask=btBroadphaseProxy::AllFilter ^ btBroadphaseProxy::StaticFilter);
+	virtual void addCollisionObject(btCollisionObject * collisionObject, int collisionFilterGroup = btBroadphaseProxy::StaticFilter, int collisionFilterMask = btBroadphaseProxy::AllFilter ^ btBroadphaseProxy::StaticFilter);
 
-	virtual void	addRigidBody(btRigidBody* body);
+	virtual void addRigidBody(btRigidBody * body);
 
-	virtual void	addRigidBody(btRigidBody* body, int group, int mask);
+	virtual void addRigidBody(btRigidBody * body, int group, int mask);
 
-	virtual void	removeRigidBody(btRigidBody* body);
+	virtual void removeRigidBody(btRigidBody * body);
 
 	///removeCollisionObject will first check if it is a rigid body, if so call removeRigidBody otherwise call btCollisionWorld::removeCollisionObject
-	virtual void	removeCollisionObject(btCollisionObject* collisionObject);
+	virtual void removeCollisionObject(btCollisionObject * collisionObject);
 
+	virtual void debugDrawConstraint(btTypedConstraint * constraint);
 
-	virtual void	debugDrawConstraint(btTypedConstraint* constraint);
+	virtual void debugDrawWorld();
 
-	virtual void	debugDrawWorld();
-
-	virtual void	setConstraintSolver(btConstraintSolver* solver);
+	virtual void setConstraintSolver(btConstraintSolver * solver);
 
 	virtual btConstraintSolver* getConstraintSolver();
-	
-	virtual	int		getNumConstraints() const;
 
-	virtual btTypedConstraint* getConstraint(int index)	;
+	virtual int getNumConstraints() const;
+
+	virtual btTypedConstraint* getConstraint(int index);
 
 	virtual const btTypedConstraint* getConstraint(int index) const;
 
-	
-	virtual btDynamicsWorldType	getWorldType() const
+	virtual btDynamicsWorldType getWorldType() const
 	{
 		return BT_DISCRETE_DYNAMICS_WORLD;
 	}
-	
+
 	///the forces on each rigidbody is accumulating together with gravity. clear this after each timestep.
-	virtual void	clearForces();
+	virtual void clearForces();
 
 	///apply gravity, call this once per timestep
-	virtual void	applyGravity();
+	virtual void applyGravity();
 
-	virtual void	setNumTasks(int numTasks)
+	virtual void setNumTasks(int numTasks)
 	{
-        (void) numTasks;
+		(void)numTasks;
 	}
 
 	///obsolete, use updateActions instead
@@ -194,15 +187,15 @@ public:
 	}
 
 	///obsolete, use addAction instead
-	virtual void	addVehicle(btActionInterface* vehicle);
+	virtual void addVehicle(btActionInterface * vehicle);
 	///obsolete, use removeAction instead
-	virtual void	removeVehicle(btActionInterface* vehicle);
+	virtual void removeVehicle(btActionInterface * vehicle);
 	///obsolete, use addAction instead
-	virtual void	addCharacter(btActionInterface* character);
+	virtual void addCharacter(btActionInterface * character);
 	///obsolete, use removeAction instead
-	virtual void	removeCharacter(btActionInterface* character);
+	virtual void removeCharacter(btActionInterface * character);
 
-	void	setSynchronizeAllMotionStates(bool synchronizeAll)
+	void setSynchronizeAllMotionStates(bool synchronizeAll)
 	{
 		m_synchronizeAllMotionStates = synchronizeAll;
 	}
@@ -215,18 +208,18 @@ public:
 	{
 		m_applySpeculativeContactRestitution = enable;
 	}
-	
+
 	bool getApplySpeculativeContactRestitution() const
 	{
 		return m_applySpeculativeContactRestitution;
 	}
 
 	///Preliminary serialization test for Bullet 2.76. Loading those files requires a separate parser (see Bullet/Demos/SerializeDemo)
-	virtual	void	serialize(btSerializer* serializer);
+	virtual void serialize(btSerializer * serializer);
 
 	///Interpolate motion state between previous and current transform, instead of current and next transform.
 	///This can relieve discontinuities in the rendering, due to penetrations
-	void setLatencyMotionStateInterpolation(bool latencyInterpolation )
+	void setLatencyMotionStateInterpolation(bool latencyInterpolation)
 	{
 		m_latencyMotionStateInterpolation = latencyInterpolation;
 	}
@@ -236,4 +229,4 @@ public:
 	}
 };
 
-#endif //BT_DISCRETE_DYNAMICS_WORLD_H
+#endif  //BT_DISCRETE_DYNAMICS_WORLD_H
diff --git a/thirdparty/bullet/BulletDynamics/Dynamics/btDiscreteDynamicsWorldMt.cpp b/thirdparty/bullet/BulletDynamics/Dynamics/btDiscreteDynamicsWorldMt.cpp
index d705bf23817..8207b471356 100644
--- a/thirdparty/bullet/BulletDynamics/Dynamics/btDiscreteDynamicsWorldMt.cpp
+++ b/thirdparty/bullet/BulletDynamics/Dynamics/btDiscreteDynamicsWorldMt.cpp
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btDiscreteDynamicsWorldMt.h"
 
 //collision detection
@@ -38,148 +37,139 @@ subject to the following restrictions:
 #include "BulletDynamics/ConstraintSolver/btSliderConstraint.h"
 #include "BulletDynamics/ConstraintSolver/btContactConstraint.h"
 
-
 #include "LinearMath/btIDebugDraw.h"
 #include "BulletCollision/CollisionShapes/btSphereShape.h"
 
-
 #include "BulletDynamics/Dynamics/btActionInterface.h"
 #include "LinearMath/btQuickprof.h"
 #include "LinearMath/btMotionState.h"
 
 #include "LinearMath/btSerializer.h"
 
-
-
 ///
 /// btConstraintSolverPoolMt
 ///
 
 btConstraintSolverPoolMt::ThreadSolver* btConstraintSolverPoolMt::getAndLockThreadSolver()
 {
-    int i = 0;
+	int i = 0;
 #if BT_THREADSAFE
-    i = btGetCurrentThreadIndex() % m_solvers.size();
-#endif // #if BT_THREADSAFE
-    while ( true )
-    {
-        ThreadSolver& solver = m_solvers[ i ];
-        if ( solver.mutex.tryLock() )
-        {
-            return &solver;
-        }
-        // failed, try the next one
-        i = ( i + 1 ) % m_solvers.size();
-    }
-    return NULL;
+	i = btGetCurrentThreadIndex() % m_solvers.size();
+#endif  // #if BT_THREADSAFE
+	while (true)
+	{
+		ThreadSolver& solver = m_solvers[i];
+		if (solver.mutex.tryLock())
+		{
+			return &solver;
+		}
+		// failed, try the next one
+		i = (i + 1) % m_solvers.size();
+	}
+	return NULL;
 }
 
-void btConstraintSolverPoolMt::init( btConstraintSolver** solvers, int numSolvers )
+void btConstraintSolverPoolMt::init(btConstraintSolver** solvers, int numSolvers)
 {
-    m_solverType = BT_SEQUENTIAL_IMPULSE_SOLVER;
-    m_solvers.resize( numSolvers );
-    for ( int i = 0; i < numSolvers; ++i )
-    {
-        m_solvers[ i ].solver = solvers[ i ];
-    }
-    if ( numSolvers > 0 )
-    {
-        m_solverType = solvers[ 0 ]->getSolverType();
-    }
+	m_solverType = BT_SEQUENTIAL_IMPULSE_SOLVER;
+	m_solvers.resize(numSolvers);
+	for (int i = 0; i < numSolvers; ++i)
+	{
+		m_solvers[i].solver = solvers[i];
+	}
+	if (numSolvers > 0)
+	{
+		m_solverType = solvers[0]->getSolverType();
+	}
 }
 
 // create the solvers for me
-btConstraintSolverPoolMt::btConstraintSolverPoolMt( int numSolvers )
+btConstraintSolverPoolMt::btConstraintSolverPoolMt(int numSolvers)
 {
-    btAlignedObjectArray<btConstraintSolver*> solvers;
-    solvers.reserve( numSolvers );
-    for ( int i = 0; i < numSolvers; ++i )
-    {
-        btConstraintSolver* solver = new btSequentialImpulseConstraintSolver();
-        solvers.push_back( solver );
-    }
-    init( &solvers[ 0 ], numSolvers );
+	btAlignedObjectArray<btConstraintSolver*> solvers;
+	solvers.reserve(numSolvers);
+	for (int i = 0; i < numSolvers; ++i)
+	{
+		btConstraintSolver* solver = new btSequentialImpulseConstraintSolver();
+		solvers.push_back(solver);
+	}
+	init(&solvers[0], numSolvers);
 }
 
 // pass in fully constructed solvers (destructor will delete them)
-btConstraintSolverPoolMt::btConstraintSolverPoolMt( btConstraintSolver** solvers, int numSolvers )
+btConstraintSolverPoolMt::btConstraintSolverPoolMt(btConstraintSolver** solvers, int numSolvers)
 {
-    init( solvers, numSolvers );
+	init(solvers, numSolvers);
 }
 
 btConstraintSolverPoolMt::~btConstraintSolverPoolMt()
 {
-    // delete all solvers
-    for ( int i = 0; i < m_solvers.size(); ++i )
-    {
-        ThreadSolver& solver = m_solvers[ i ];
-        delete solver.solver;
-        solver.solver = NULL;
-    }
+	// delete all solvers
+	for (int i = 0; i < m_solvers.size(); ++i)
+	{
+		ThreadSolver& solver = m_solvers[i];
+		delete solver.solver;
+		solver.solver = NULL;
+	}
 }
 
 ///solve a group of constraints
-btScalar btConstraintSolverPoolMt::solveGroup( btCollisionObject** bodies,
-    int numBodies,
-    btPersistentManifold** manifolds,
-    int numManifolds,
-    btTypedConstraint** constraints,
-    int numConstraints,
-    const btContactSolverInfo& info,
-    btIDebugDraw* debugDrawer,
-    btDispatcher* dispatcher
-)
+btScalar btConstraintSolverPoolMt::solveGroup(btCollisionObject** bodies,
+											  int numBodies,
+											  btPersistentManifold** manifolds,
+											  int numManifolds,
+											  btTypedConstraint** constraints,
+											  int numConstraints,
+											  const btContactSolverInfo& info,
+											  btIDebugDraw* debugDrawer,
+											  btDispatcher* dispatcher)
 {
-    ThreadSolver* ts = getAndLockThreadSolver();
-    ts->solver->solveGroup( bodies, numBodies, manifolds, numManifolds, constraints, numConstraints, info, debugDrawer, dispatcher );
-    ts->mutex.unlock();
-    return 0.0f;
+	ThreadSolver* ts = getAndLockThreadSolver();
+	ts->solver->solveGroup(bodies, numBodies, manifolds, numManifolds, constraints, numConstraints, info, debugDrawer, dispatcher);
+	ts->mutex.unlock();
+	return 0.0f;
 }
 
 void btConstraintSolverPoolMt::reset()
 {
-    for ( int i = 0; i < m_solvers.size(); ++i )
-    {
-        ThreadSolver& solver = m_solvers[ i ];
-        solver.mutex.lock();
-        solver.solver->reset();
-        solver.mutex.unlock();
-    }
+	for (int i = 0; i < m_solvers.size(); ++i)
+	{
+		ThreadSolver& solver = m_solvers[i];
+		solver.mutex.lock();
+		solver.solver->reset();
+		solver.mutex.unlock();
+	}
 }
 
-
 ///
 /// btDiscreteDynamicsWorldMt
 ///
 
 btDiscreteDynamicsWorldMt::btDiscreteDynamicsWorldMt(btDispatcher* dispatcher,
-    btBroadphaseInterface* pairCache,
-    btConstraintSolverPoolMt* constraintSolver,
-    btConstraintSolver* constraintSolverMt,
-    btCollisionConfiguration* collisionConfiguration
-)
-: btDiscreteDynamicsWorld(dispatcher,pairCache,constraintSolver,collisionConfiguration)
+													 btBroadphaseInterface* pairCache,
+													 btConstraintSolverPoolMt* solverPool,
+													 btConstraintSolver* constraintSolverMt,
+													 btCollisionConfiguration* collisionConfiguration)
+	: btDiscreteDynamicsWorld(dispatcher, pairCache, solverPool, collisionConfiguration)
 {
 	if (m_ownsIslandManager)
 	{
 		m_islandManager->~btSimulationIslandManager();
-		btAlignedFree( m_islandManager);
+		btAlignedFree(m_islandManager);
 	}
 	{
-		void* mem = btAlignedAlloc(sizeof(btSimulationIslandManagerMt),16);
+		void* mem = btAlignedAlloc(sizeof(btSimulationIslandManagerMt), 16);
 		btSimulationIslandManagerMt* im = new (mem) btSimulationIslandManagerMt();
-        im->setMinimumSolverBatchSize( m_solverInfo.m_minimumSolverBatchSize );
-        m_islandManager = im;
+		im->setMinimumSolverBatchSize(m_solverInfo.m_minimumSolverBatchSize);
+		m_islandManager = im;
 	}
-    m_constraintSolverMt = constraintSolverMt;
+	m_constraintSolverMt = constraintSolverMt;
 }
 
-
 btDiscreteDynamicsWorldMt::~btDiscreteDynamicsWorldMt()
 {
 }
 
-
 void btDiscreteDynamicsWorldMt::solveConstraints(btContactSolverInfo& solverInfo)
 {
 	BT_PROFILE("solveConstraints");
@@ -187,92 +177,87 @@ void btDiscreteDynamicsWorldMt::solveConstraints(btContactSolverInfo& solverInfo
 	m_constraintSolver->prepareSolve(getCollisionWorld()->getNumCollisionObjects(), getCollisionWorld()->getDispatcher()->getNumManifolds());
 
 	/// solve all the constraints for this island
-    btSimulationIslandManagerMt* im = static_cast<btSimulationIslandManagerMt*>(m_islandManager);
-    btSimulationIslandManagerMt::SolverParams solverParams;
-    solverParams.m_solverPool = m_constraintSolver;
-    solverParams.m_solverMt = m_constraintSolverMt;
-    solverParams.m_solverInfo = &solverInfo;
-    solverParams.m_debugDrawer = m_debugDrawer;
-    solverParams.m_dispatcher = getCollisionWorld()->getDispatcher();
-    im->buildAndProcessIslands( getCollisionWorld()->getDispatcher(), getCollisionWorld(), m_constraints, solverParams );
+	btSimulationIslandManagerMt* im = static_cast<btSimulationIslandManagerMt*>(m_islandManager);
+	btSimulationIslandManagerMt::SolverParams solverParams;
+	solverParams.m_solverPool = m_constraintSolver;
+	solverParams.m_solverMt = m_constraintSolverMt;
+	solverParams.m_solverInfo = &solverInfo;
+	solverParams.m_debugDrawer = m_debugDrawer;
+	solverParams.m_dispatcher = getCollisionWorld()->getDispatcher();
+	im->buildAndProcessIslands(getCollisionWorld()->getDispatcher(), getCollisionWorld(), m_constraints, solverParams);
 
 	m_constraintSolver->allSolved(solverInfo, m_debugDrawer);
 }
 
-
 struct UpdaterUnconstrainedMotion : public btIParallelForBody
 {
-    btScalar timeStep;
-    btRigidBody** rigidBodies;
+	btScalar timeStep;
+	btRigidBody** rigidBodies;
 
-    void forLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        for ( int i = iBegin; i < iEnd; ++i )
-        {
-            btRigidBody* body = rigidBodies[ i ];
-            if ( !body->isStaticOrKinematicObject() )
-            {
-                //don't integrate/update velocities here, it happens in the constraint solver
-                body->applyDamping( timeStep );
-                body->predictIntegratedTransform( timeStep, body->getInterpolationWorldTransform() );
-            }
-        }
-    }
+	void forLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		for (int i = iBegin; i < iEnd; ++i)
+		{
+			btRigidBody* body = rigidBodies[i];
+			if (!body->isStaticOrKinematicObject())
+			{
+				//don't integrate/update velocities here, it happens in the constraint solver
+				body->applyDamping(timeStep);
+				body->predictIntegratedTransform(timeStep, body->getInterpolationWorldTransform());
+			}
+		}
+	}
 };
 
-
-void btDiscreteDynamicsWorldMt::predictUnconstraintMotion( btScalar timeStep )
+void btDiscreteDynamicsWorldMt::predictUnconstraintMotion(btScalar timeStep)
 {
-    BT_PROFILE( "predictUnconstraintMotion" );
-    if ( m_nonStaticRigidBodies.size() > 0 )
-    {
-        UpdaterUnconstrainedMotion update;
-        update.timeStep = timeStep;
-        update.rigidBodies = &m_nonStaticRigidBodies[ 0 ];
-        int grainSize = 50;  // num of iterations per task for task scheduler
-        btParallelFor( 0, m_nonStaticRigidBodies.size(), grainSize, update );
-    }
+	BT_PROFILE("predictUnconstraintMotion");
+	if (m_nonStaticRigidBodies.size() > 0)
+	{
+		UpdaterUnconstrainedMotion update;
+		update.timeStep = timeStep;
+		update.rigidBodies = &m_nonStaticRigidBodies[0];
+		int grainSize = 50;  // num of iterations per task for task scheduler
+		btParallelFor(0, m_nonStaticRigidBodies.size(), grainSize, update);
+	}
 }
 
-
-void btDiscreteDynamicsWorldMt::createPredictiveContacts( btScalar timeStep )
+void btDiscreteDynamicsWorldMt::createPredictiveContacts(btScalar timeStep)
 {
-    BT_PROFILE( "createPredictiveContacts" );
-    releasePredictiveContacts();
-    if ( m_nonStaticRigidBodies.size() > 0 )
-    {
-        UpdaterCreatePredictiveContacts update;
-        update.world = this;
-        update.timeStep = timeStep;
-        update.rigidBodies = &m_nonStaticRigidBodies[ 0 ];
-        int grainSize = 50;  // num of iterations per task for task scheduler
-        btParallelFor( 0, m_nonStaticRigidBodies.size(), grainSize, update );
-    }
+	BT_PROFILE("createPredictiveContacts");
+	releasePredictiveContacts();
+	if (m_nonStaticRigidBodies.size() > 0)
+	{
+		UpdaterCreatePredictiveContacts update;
+		update.world = this;
+		update.timeStep = timeStep;
+		update.rigidBodies = &m_nonStaticRigidBodies[0];
+		int grainSize = 50;  // num of iterations per task for task scheduler
+		btParallelFor(0, m_nonStaticRigidBodies.size(), grainSize, update);
+	}
 }
 
-
-void btDiscreteDynamicsWorldMt::integrateTransforms( btScalar timeStep )
+void btDiscreteDynamicsWorldMt::integrateTransforms(btScalar timeStep)
 {
-    BT_PROFILE( "integrateTransforms" );
-    if ( m_nonStaticRigidBodies.size() > 0 )
-    {
-        UpdaterIntegrateTransforms update;
-        update.world = this;
-        update.timeStep = timeStep;
-        update.rigidBodies = &m_nonStaticRigidBodies[ 0 ];
-        int grainSize = 50;  // num of iterations per task for task scheduler
-        btParallelFor( 0, m_nonStaticRigidBodies.size(), grainSize, update );
-    }
+	BT_PROFILE("integrateTransforms");
+	if (m_nonStaticRigidBodies.size() > 0)
+	{
+		UpdaterIntegrateTransforms update;
+		update.world = this;
+		update.timeStep = timeStep;
+		update.rigidBodies = &m_nonStaticRigidBodies[0];
+		int grainSize = 50;  // num of iterations per task for task scheduler
+		btParallelFor(0, m_nonStaticRigidBodies.size(), grainSize, update);
+	}
 }
 
-
-int	btDiscreteDynamicsWorldMt::stepSimulation( btScalar timeStep, int maxSubSteps, btScalar fixedTimeStep )
+int btDiscreteDynamicsWorldMt::stepSimulation(btScalar timeStep, int maxSubSteps, btScalar fixedTimeStep)
 {
-    int numSubSteps = btDiscreteDynamicsWorld::stepSimulation(timeStep, maxSubSteps, fixedTimeStep);
-    if (btITaskScheduler* scheduler = btGetTaskScheduler())
-    {
-        // tell Bullet's threads to sleep, so other threads can run
-        scheduler->sleepWorkerThreadsHint();
-    }
-    return numSubSteps;
+	int numSubSteps = btDiscreteDynamicsWorld::stepSimulation(timeStep, maxSubSteps, fixedTimeStep);
+	if (btITaskScheduler* scheduler = btGetTaskScheduler())
+	{
+		// tell Bullet's threads to sleep, so other threads can run
+		scheduler->sleepWorkerThreadsHint();
+	}
+	return numSubSteps;
 }
diff --git a/thirdparty/bullet/BulletDynamics/Dynamics/btDiscreteDynamicsWorldMt.h b/thirdparty/bullet/BulletDynamics/Dynamics/btDiscreteDynamicsWorldMt.h
index 667fe5800e5..dccf35d7a7a 100644
--- a/thirdparty/bullet/BulletDynamics/Dynamics/btDiscreteDynamicsWorldMt.h
+++ b/thirdparty/bullet/BulletDynamics/Dynamics/btDiscreteDynamicsWorldMt.h
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef BT_DISCRETE_DYNAMICS_WORLD_MT_H
 #define BT_DISCRETE_DYNAMICS_WORLD_MT_H
 
@@ -21,7 +20,6 @@ subject to the following restrictions:
 #include "btSimulationIslandManagerMt.h"
 #include "BulletDynamics/ConstraintSolver/btConstraintSolver.h"
 
-
 ///
 /// btConstraintSolverPoolMt - masquerades as a constraint solver, but really it is a threadsafe pool of them.
 ///
@@ -34,46 +32,43 @@ subject to the following restrictions:
 class btConstraintSolverPoolMt : public btConstraintSolver
 {
 public:
-    // create the solvers for me
-    explicit btConstraintSolverPoolMt( int numSolvers );
+	// create the solvers for me
+	explicit btConstraintSolverPoolMt(int numSolvers);
 
-    // pass in fully constructed solvers (destructor will delete them)
-    btConstraintSolverPoolMt( btConstraintSolver** solvers, int numSolvers );
+	// pass in fully constructed solvers (destructor will delete them)
+	btConstraintSolverPoolMt(btConstraintSolver** solvers, int numSolvers);
 
-    virtual ~btConstraintSolverPoolMt();
+	virtual ~btConstraintSolverPoolMt();
 
-    ///solve a group of constraints
-    virtual btScalar solveGroup( btCollisionObject** bodies,
-        int numBodies,
-        btPersistentManifold** manifolds,
-        int numManifolds,
-        btTypedConstraint** constraints,
-        int numConstraints,
-        const btContactSolverInfo& info,
-        btIDebugDraw* debugDrawer,
-        btDispatcher* dispatcher
-    ) BT_OVERRIDE;
+	///solve a group of constraints
+	virtual btScalar solveGroup(btCollisionObject** bodies,
+								int numBodies,
+								btPersistentManifold** manifolds,
+								int numManifolds,
+								btTypedConstraint** constraints,
+								int numConstraints,
+								const btContactSolverInfo& info,
+								btIDebugDraw* debugDrawer,
+								btDispatcher* dispatcher) BT_OVERRIDE;
 
-    virtual	void reset() BT_OVERRIDE;
-    virtual btConstraintSolverType getSolverType() const BT_OVERRIDE { return m_solverType; }
+	virtual void reset() BT_OVERRIDE;
+	virtual btConstraintSolverType getSolverType() const BT_OVERRIDE { return m_solverType; }
 
 private:
-    const static size_t kCacheLineSize = 128;
-    struct ThreadSolver
-    {
-        btConstraintSolver* solver;
-        btSpinMutex mutex;
-        char _cachelinePadding[ kCacheLineSize - sizeof( btSpinMutex ) - sizeof( void* ) ];  // keep mutexes from sharing a cache line
-    };
-    btAlignedObjectArray<ThreadSolver> m_solvers;
-    btConstraintSolverType m_solverType;
+	const static size_t kCacheLineSize = 128;
+	struct ThreadSolver
+	{
+		btConstraintSolver* solver;
+		btSpinMutex mutex;
+		char _cachelinePadding[kCacheLineSize - sizeof(btSpinMutex) - sizeof(void*)];  // keep mutexes from sharing a cache line
+	};
+	btAlignedObjectArray<ThreadSolver> m_solvers;
+	btConstraintSolverType m_solverType;
 
-    ThreadSolver* getAndLockThreadSolver();
-    void init( btConstraintSolver** solvers, int numSolvers );
+	ThreadSolver* getAndLockThreadSolver();
+	void init(btConstraintSolver** solvers, int numSolvers);
 };
 
-
-
 ///
 /// btDiscreteDynamicsWorldMt -- a version of DiscreteDynamicsWorld with some minor changes to support
 ///                              solving simulation islands on multiple threads.
@@ -84,53 +79,53 @@ private:
 ///     - integrateTransforms
 ///     - createPredictiveContacts
 ///
-ATTRIBUTE_ALIGNED16(class) btDiscreteDynamicsWorldMt : public btDiscreteDynamicsWorld
+ATTRIBUTE_ALIGNED16(class)
+btDiscreteDynamicsWorldMt : public btDiscreteDynamicsWorld
 {
 protected:
-    btConstraintSolver* m_constraintSolverMt;
+	btConstraintSolver* m_constraintSolverMt;
 
-    virtual void solveConstraints(btContactSolverInfo& solverInfo) BT_OVERRIDE;
+	virtual void solveConstraints(btContactSolverInfo & solverInfo) BT_OVERRIDE;
 
-    virtual void predictUnconstraintMotion( btScalar timeStep ) BT_OVERRIDE;
+	virtual void predictUnconstraintMotion(btScalar timeStep) BT_OVERRIDE;
 
-    struct UpdaterCreatePredictiveContacts : public btIParallelForBody
-    {
-        btScalar timeStep;
-        btRigidBody** rigidBodies;
-        btDiscreteDynamicsWorldMt* world;
+	struct UpdaterCreatePredictiveContacts : public btIParallelForBody
+	{
+		btScalar timeStep;
+		btRigidBody** rigidBodies;
+		btDiscreteDynamicsWorldMt* world;
 
-        void forLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-        {
-            world->createPredictiveContactsInternal( &rigidBodies[ iBegin ], iEnd - iBegin, timeStep );
-        }
-    };
-    virtual void createPredictiveContacts( btScalar timeStep ) BT_OVERRIDE;
+		void forLoop(int iBegin, int iEnd) const BT_OVERRIDE
+		{
+			world->createPredictiveContactsInternal(&rigidBodies[iBegin], iEnd - iBegin, timeStep);
+		}
+	};
+	virtual void createPredictiveContacts(btScalar timeStep) BT_OVERRIDE;
 
-    struct UpdaterIntegrateTransforms : public btIParallelForBody
-    {
-        btScalar timeStep;
-        btRigidBody** rigidBodies;
-        btDiscreteDynamicsWorldMt* world;
+	struct UpdaterIntegrateTransforms : public btIParallelForBody
+	{
+		btScalar timeStep;
+		btRigidBody** rigidBodies;
+		btDiscreteDynamicsWorldMt* world;
 
-        void forLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-        {
-            world->integrateTransformsInternal( &rigidBodies[ iBegin ], iEnd - iBegin, timeStep );
-        }
-    };
-    virtual void integrateTransforms( btScalar timeStep ) BT_OVERRIDE;
+		void forLoop(int iBegin, int iEnd) const BT_OVERRIDE
+		{
+			world->integrateTransformsInternal(&rigidBodies[iBegin], iEnd - iBegin, timeStep);
+		}
+	};
+	virtual void integrateTransforms(btScalar timeStep) BT_OVERRIDE;
 
 public:
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-	btDiscreteDynamicsWorldMt(btDispatcher* dispatcher,
-        btBroadphaseInterface* pairCache,
-        btConstraintSolverPoolMt* constraintSolver,   // Note this should be a solver-pool for multi-threading
-        btConstraintSolver* constraintSolverMt,    // single multi-threaded solver for large islands (or NULL)
-        btCollisionConfiguration* collisionConfiguration
-    );
+	btDiscreteDynamicsWorldMt(btDispatcher * dispatcher,
+							  btBroadphaseInterface * pairCache,
+							  btConstraintSolverPoolMt * solverPool,        // Note this should be a solver-pool for multi-threading
+							  btConstraintSolver * constraintSolverMt,      // single multi-threaded solver for large islands (or NULL)
+							  btCollisionConfiguration * collisionConfiguration);
 	virtual ~btDiscreteDynamicsWorldMt();
 
-    virtual int	stepSimulation( btScalar timeStep, int maxSubSteps, btScalar fixedTimeStep ) BT_OVERRIDE;
+	virtual int stepSimulation(btScalar timeStep, int maxSubSteps, btScalar fixedTimeStep) BT_OVERRIDE;
 };
 
-#endif //BT_DISCRETE_DYNAMICS_WORLD_H
+#endif  //BT_DISCRETE_DYNAMICS_WORLD_H
diff --git a/thirdparty/bullet/BulletDynamics/Dynamics/btDynamicsWorld.h b/thirdparty/bullet/BulletDynamics/Dynamics/btDynamicsWorld.h
index 42d8fc0de3f..eadd8c12e73 100644
--- a/thirdparty/bullet/BulletDynamics/Dynamics/btDynamicsWorld.h
+++ b/thirdparty/bullet/BulletDynamics/Dynamics/btDynamicsWorld.h
@@ -24,150 +24,150 @@ class btActionInterface;
 class btConstraintSolver;
 class btDynamicsWorld;
 
-
 /// Type for the callback for each tick
-typedef void (*btInternalTickCallback)(btDynamicsWorld *world, btScalar timeStep);
+typedef void (*btInternalTickCallback)(btDynamicsWorld* world, btScalar timeStep);
 
 enum btDynamicsWorldType
 {
-	BT_SIMPLE_DYNAMICS_WORLD=1,
-	BT_DISCRETE_DYNAMICS_WORLD=2,
-	BT_CONTINUOUS_DYNAMICS_WORLD=3,
-	BT_SOFT_RIGID_DYNAMICS_WORLD=4,
-	BT_GPU_DYNAMICS_WORLD=5,
-    BT_SOFT_MULTIBODY_DYNAMICS_WORLD=6
+	BT_SIMPLE_DYNAMICS_WORLD = 1,
+	BT_DISCRETE_DYNAMICS_WORLD = 2,
+	BT_CONTINUOUS_DYNAMICS_WORLD = 3,
+	BT_SOFT_RIGID_DYNAMICS_WORLD = 4,
+	BT_GPU_DYNAMICS_WORLD = 5,
+	BT_SOFT_MULTIBODY_DYNAMICS_WORLD = 6
 };
 
 ///The btDynamicsWorld is the interface class for several dynamics implementation, basic, discrete, parallel, and continuous etc.
 class btDynamicsWorld : public btCollisionWorld
 {
-
 protected:
-		btInternalTickCallback m_internalTickCallback;
-		btInternalTickCallback m_internalPreTickCallback;
-		void*	m_worldUserInfo;
+	btInternalTickCallback m_internalTickCallback;
+	btInternalTickCallback m_internalPreTickCallback;
+	void* m_worldUserInfo;
 
-		btContactSolverInfo	m_solverInfo;
+	btContactSolverInfo m_solverInfo;
 
 public:
-		
+	btDynamicsWorld(btDispatcher* dispatcher, btBroadphaseInterface* broadphase, btCollisionConfiguration* collisionConfiguration)
+		: btCollisionWorld(dispatcher, broadphase, collisionConfiguration), m_internalTickCallback(0), m_internalPreTickCallback(0), m_worldUserInfo(0)
+	{
+	}
 
-		btDynamicsWorld(btDispatcher* dispatcher,btBroadphaseInterface* broadphase,btCollisionConfiguration* collisionConfiguration)
-		:btCollisionWorld(dispatcher,broadphase,collisionConfiguration), m_internalTickCallback(0),m_internalPreTickCallback(0), m_worldUserInfo(0)
+	virtual ~btDynamicsWorld()
+	{
+	}
+
+	///stepSimulation proceeds the simulation over 'timeStep', units in preferably in seconds.
+	///By default, Bullet will subdivide the timestep in constant substeps of each 'fixedTimeStep'.
+	///in order to keep the simulation real-time, the maximum number of substeps can be clamped to 'maxSubSteps'.
+	///You can disable subdividing the timestep/substepping by passing maxSubSteps=0 as second argument to stepSimulation, but in that case you have to keep the timeStep constant.
+	virtual int stepSimulation(btScalar timeStep, int maxSubSteps = 1, btScalar fixedTimeStep = btScalar(1.) / btScalar(60.)) = 0;
+
+	virtual void debugDrawWorld() = 0;
+
+	virtual void addConstraint(btTypedConstraint* constraint, bool disableCollisionsBetweenLinkedBodies = false)
+	{
+		(void)constraint;
+		(void)disableCollisionsBetweenLinkedBodies;
+	}
+
+	virtual void removeConstraint(btTypedConstraint* constraint) { (void)constraint; }
+
+	virtual void addAction(btActionInterface* action) = 0;
+
+	virtual void removeAction(btActionInterface* action) = 0;
+
+	//once a rigidbody is added to the dynamics world, it will get this gravity assigned
+	//existing rigidbodies in the world get gravity assigned too, during this method
+	virtual void setGravity(const btVector3& gravity) = 0;
+	virtual btVector3 getGravity() const = 0;
+
+	virtual void synchronizeMotionStates() = 0;
+
+	virtual void addRigidBody(btRigidBody* body) = 0;
+
+	virtual void addRigidBody(btRigidBody* body, int group, int mask) = 0;
+
+	virtual void removeRigidBody(btRigidBody* body) = 0;
+
+	virtual void setConstraintSolver(btConstraintSolver* solver) = 0;
+
+	virtual btConstraintSolver* getConstraintSolver() = 0;
+
+	virtual int getNumConstraints() const { return 0; }
+
+	virtual btTypedConstraint* getConstraint(int index)
+	{
+		(void)index;
+		return 0;
+	}
+
+	virtual const btTypedConstraint* getConstraint(int index) const
+	{
+		(void)index;
+		return 0;
+	}
+
+	virtual btDynamicsWorldType getWorldType() const = 0;
+
+	virtual void clearForces() = 0;
+
+	/// Set the callback for when an internal tick (simulation substep) happens, optional user info
+	void setInternalTickCallback(btInternalTickCallback cb, void* worldUserInfo = 0, bool isPreTick = false)
+	{
+		if (isPreTick)
 		{
+			m_internalPreTickCallback = cb;
 		}
-
-		virtual ~btDynamicsWorld()
+		else
 		{
+			m_internalTickCallback = cb;
 		}
-		
-		///stepSimulation proceeds the simulation over 'timeStep', units in preferably in seconds.
-		///By default, Bullet will subdivide the timestep in constant substeps of each 'fixedTimeStep'.
-		///in order to keep the simulation real-time, the maximum number of substeps can be clamped to 'maxSubSteps'.
-		///You can disable subdividing the timestep/substepping by passing maxSubSteps=0 as second argument to stepSimulation, but in that case you have to keep the timeStep constant.
-		virtual int		stepSimulation( btScalar timeStep,int maxSubSteps=1, btScalar fixedTimeStep=btScalar(1.)/btScalar(60.))=0;
-			
-		virtual void	debugDrawWorld() = 0;
-				
-		virtual void	addConstraint(btTypedConstraint* constraint, bool disableCollisionsBetweenLinkedBodies=false) 
-		{ 
-			(void)constraint; (void)disableCollisionsBetweenLinkedBodies;
-		}
+		m_worldUserInfo = worldUserInfo;
+	}
 
-		virtual void	removeConstraint(btTypedConstraint* constraint) {(void)constraint;}
+	void setWorldUserInfo(void* worldUserInfo)
+	{
+		m_worldUserInfo = worldUserInfo;
+	}
 
-		virtual void	addAction(btActionInterface* action) = 0;
+	void* getWorldUserInfo() const
+	{
+		return m_worldUserInfo;
+	}
 
-		virtual void	removeAction(btActionInterface* action) = 0;
-
-		//once a rigidbody is added to the dynamics world, it will get this gravity assigned
-		//existing rigidbodies in the world get gravity assigned too, during this method
-		virtual void	setGravity(const btVector3& gravity) = 0;
-		virtual btVector3 getGravity () const = 0;
-
-		virtual void	synchronizeMotionStates() = 0;
-
-		virtual void	addRigidBody(btRigidBody* body) = 0;
-
-		virtual void	addRigidBody(btRigidBody* body, int group, int mask) = 0;
-
-		virtual void	removeRigidBody(btRigidBody* body) = 0;
-
-		virtual void	setConstraintSolver(btConstraintSolver* solver) = 0;
-
-		virtual btConstraintSolver* getConstraintSolver() = 0;
-		
-		virtual	int		getNumConstraints() const {	return 0;		}
-		
-		virtual btTypedConstraint* getConstraint(int index)		{	(void)index;		return 0;		}
-		
-		virtual const btTypedConstraint* getConstraint(int index) const	{	(void)index;	return 0;	}
-
-		virtual btDynamicsWorldType	getWorldType() const=0;
-
-		virtual void	clearForces() = 0;
-
-		/// Set the callback for when an internal tick (simulation substep) happens, optional user info
-		void setInternalTickCallback(btInternalTickCallback cb,	void* worldUserInfo=0,bool isPreTick=false) 
-		{ 
-			if (isPreTick)
-			{
-				m_internalPreTickCallback = cb;
-			} else
-			{
-				m_internalTickCallback = cb; 
-			}
-			m_worldUserInfo = worldUserInfo;
-		}
-
-		void	setWorldUserInfo(void* worldUserInfo)
-		{
-			m_worldUserInfo = worldUserInfo;
-		}
-
-		void*	getWorldUserInfo() const
-		{
-			return m_worldUserInfo;
-		}
-
-		btContactSolverInfo& getSolverInfo()
-		{
-			return m_solverInfo;
-		}
-
-		const btContactSolverInfo& getSolverInfo() const
-		{
-			return m_solverInfo;
-		}
-
-
-		///obsolete, use addAction instead.
-		virtual void	addVehicle(btActionInterface* vehicle) {(void)vehicle;}
-		///obsolete, use removeAction instead
-		virtual void	removeVehicle(btActionInterface* vehicle) {(void)vehicle;}
-		///obsolete, use addAction instead.
-		virtual void	addCharacter(btActionInterface* character) {(void)character;}
-		///obsolete, use removeAction instead
-		virtual void	removeCharacter(btActionInterface* character) {(void)character;}
+	btContactSolverInfo& getSolverInfo()
+	{
+		return m_solverInfo;
+	}
 
+	const btContactSolverInfo& getSolverInfo() const
+	{
+		return m_solverInfo;
+	}
 
+	///obsolete, use addAction instead.
+	virtual void addVehicle(btActionInterface* vehicle) { (void)vehicle; }
+	///obsolete, use removeAction instead
+	virtual void removeVehicle(btActionInterface* vehicle) { (void)vehicle; }
+	///obsolete, use addAction instead.
+	virtual void addCharacter(btActionInterface* character) { (void)character; }
+	///obsolete, use removeAction instead
+	virtual void removeCharacter(btActionInterface* character) { (void)character; }
 };
 
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
 struct btDynamicsWorldDoubleData
 {
-	btContactSolverInfoDoubleData	m_solverInfo;
-	btVector3DoubleData	m_gravity;
+	btContactSolverInfoDoubleData m_solverInfo;
+	btVector3DoubleData m_gravity;
 };
 
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
 struct btDynamicsWorldFloatData
 {
-	btContactSolverInfoFloatData	m_solverInfo;
-	btVector3FloatData	m_gravity;
+	btContactSolverInfoFloatData m_solverInfo;
+	btVector3FloatData m_gravity;
 };
 
-
-#endif //BT_DYNAMICS_WORLD_H
-
-
+#endif  //BT_DYNAMICS_WORLD_H
diff --git a/thirdparty/bullet/BulletDynamics/Dynamics/btRigidBody.cpp b/thirdparty/bullet/BulletDynamics/Dynamics/btRigidBody.cpp
index ca0714fcfa8..f4bcabada27 100644
--- a/thirdparty/bullet/BulletDynamics/Dynamics/btRigidBody.cpp
+++ b/thirdparty/bullet/BulletDynamics/Dynamics/btRigidBody.cpp
@@ -22,36 +22,34 @@ subject to the following restrictions:
 #include "LinearMath/btSerializer.h"
 
 //'temporarily' global variables
-btScalar	gDeactivationTime = btScalar(2.);
-bool	gDisableDeactivation = false;
+btScalar gDeactivationTime = btScalar(2.);
+bool gDisableDeactivation = false;
 static int uniqueId = 0;
 
-
 btRigidBody::btRigidBody(const btRigidBody::btRigidBodyConstructionInfo& constructionInfo)
 {
 	setupRigidBody(constructionInfo);
 }
 
-btRigidBody::btRigidBody(btScalar mass, btMotionState *motionState, btCollisionShape *collisionShape, const btVector3 &localInertia)
+btRigidBody::btRigidBody(btScalar mass, btMotionState* motionState, btCollisionShape* collisionShape, const btVector3& localInertia)
 {
-	btRigidBodyConstructionInfo cinfo(mass,motionState,collisionShape,localInertia);
+	btRigidBodyConstructionInfo cinfo(mass, motionState, collisionShape, localInertia);
 	setupRigidBody(cinfo);
 }
 
-void	btRigidBody::setupRigidBody(const btRigidBody::btRigidBodyConstructionInfo& constructionInfo)
+void btRigidBody::setupRigidBody(const btRigidBody::btRigidBodyConstructionInfo& constructionInfo)
 {
-
-	m_internalType=CO_RIGID_BODY;
+	m_internalType = CO_RIGID_BODY;
 
 	m_linearVelocity.setValue(btScalar(0.0), btScalar(0.0), btScalar(0.0));
-	m_angularVelocity.setValue(btScalar(0.),btScalar(0.),btScalar(0.));
-	m_angularFactor.setValue(1,1,1);
-	m_linearFactor.setValue(1,1,1);
+	m_angularVelocity.setValue(btScalar(0.), btScalar(0.), btScalar(0.));
+	m_angularFactor.setValue(1, 1, 1);
+	m_linearFactor.setValue(1, 1, 1);
 	m_gravity.setValue(btScalar(0.0), btScalar(0.0), btScalar(0.0));
 	m_gravity_acceleration.setValue(btScalar(0.0), btScalar(0.0), btScalar(0.0));
 	m_totalForce.setValue(btScalar(0.0), btScalar(0.0), btScalar(0.0));
 	m_totalTorque.setValue(btScalar(0.0), btScalar(0.0), btScalar(0.0)),
-    setDamping(constructionInfo.m_linearDamping, constructionInfo.m_angularDamping);
+		setDamping(constructionInfo.m_linearDamping, constructionInfo.m_angularDamping);
 
 	m_linearSleepingThreshold = constructionInfo.m_linearSleepingThreshold;
 	m_angularSleepingThreshold = constructionInfo.m_angularSleepingThreshold;
@@ -67,48 +65,44 @@ void	btRigidBody::setupRigidBody(const btRigidBody::btRigidBodyConstructionInfo&
 	if (m_optionalMotionState)
 	{
 		m_optionalMotionState->getWorldTransform(m_worldTransform);
-	} else
+	}
+	else
 	{
 		m_worldTransform = constructionInfo.m_startWorldTransform;
 	}
 
 	m_interpolationWorldTransform = m_worldTransform;
-	m_interpolationLinearVelocity.setValue(0,0,0);
-	m_interpolationAngularVelocity.setValue(0,0,0);
-	
+	m_interpolationLinearVelocity.setValue(0, 0, 0);
+	m_interpolationAngularVelocity.setValue(0, 0, 0);
+
 	//moved to btCollisionObject
 	m_friction = constructionInfo.m_friction;
 	m_rollingFriction = constructionInfo.m_rollingFriction;
-    m_spinningFriction = constructionInfo.m_spinningFriction;
-    
+	m_spinningFriction = constructionInfo.m_spinningFriction;
+
 	m_restitution = constructionInfo.m_restitution;
 
-	setCollisionShape( constructionInfo.m_collisionShape );
+	setCollisionShape(constructionInfo.m_collisionShape);
 	m_debugBodyId = uniqueId++;
-	
+
 	setMassProps(constructionInfo.m_mass, constructionInfo.m_localInertia);
 	updateInertiaTensor();
 
 	m_rigidbodyFlags = BT_ENABLE_GYROSCOPIC_FORCE_IMPLICIT_BODY;
 
-
 	m_deltaLinearVelocity.setZero();
 	m_deltaAngularVelocity.setZero();
-	m_invMass = m_inverseMass*m_linearFactor;
+	m_invMass = m_inverseMass * m_linearFactor;
 	m_pushVelocity.setZero();
 	m_turnVelocity.setZero();
-
-	
-
 }
 
-
-void btRigidBody::predictIntegratedTransform(btScalar timeStep,btTransform& predictedTransform) 
+void btRigidBody::predictIntegratedTransform(btScalar timeStep, btTransform& predictedTransform)
 {
-	btTransformUtil::integrateTransform(m_worldTransform,m_linearVelocity,m_angularVelocity,timeStep,predictedTransform);
+	btTransformUtil::integrateTransform(m_worldTransform, m_linearVelocity, m_angularVelocity, timeStep, predictedTransform);
 }
 
-void			btRigidBody::saveKinematicState(btScalar timeStep)
+void btRigidBody::saveKinematicState(btScalar timeStep)
 {
 	//todo: clamp to some (user definable) safe minimum timestep, to limit maximum angular/linear velocities
 	if (timeStep != btScalar(0.))
@@ -116,25 +110,22 @@ void			btRigidBody::saveKinematicState(btScalar timeStep)
 		//if we use motionstate to synchronize world transforms, get the new kinematic/animated world transform
 		if (getMotionState())
 			getMotionState()->getWorldTransform(m_worldTransform);
-		btVector3 linVel,angVel;
-		
-		btTransformUtil::calculateVelocity(m_interpolationWorldTransform,m_worldTransform,timeStep,m_linearVelocity,m_angularVelocity);
+		btVector3 linVel, angVel;
+
+		btTransformUtil::calculateVelocity(m_interpolationWorldTransform, m_worldTransform, timeStep, m_linearVelocity, m_angularVelocity);
 		m_interpolationLinearVelocity = m_linearVelocity;
 		m_interpolationAngularVelocity = m_angularVelocity;
 		m_interpolationWorldTransform = m_worldTransform;
 		//printf("angular = %f %f %f\n",m_angularVelocity.getX(),m_angularVelocity.getY(),m_angularVelocity.getZ());
 	}
 }
-	
-void	btRigidBody::getAabb(btVector3& aabbMin,btVector3& aabbMax) const
+
+void btRigidBody::getAabb(btVector3& aabbMin, btVector3& aabbMax) const
 {
-	getCollisionShape()->getAabb(m_worldTransform,aabbMin,aabbMax);
+	getCollisionShape()->getAabb(m_worldTransform, aabbMin, aabbMax);
 }
 
-
-
-
-void btRigidBody::setGravity(const btVector3& acceleration) 
+void btRigidBody::setGravity(const btVector3& acceleration)
 {
 	if (m_inverseMass != btScalar(0.0))
 	{
@@ -143,22 +134,14 @@ void btRigidBody::setGravity(const btVector3& acceleration)
 	m_gravity_acceleration = acceleration;
 }
 
-
-
-
-
-
 void btRigidBody::setDamping(btScalar lin_damping, btScalar ang_damping)
 {
 	m_linearDamping = btClamped(lin_damping, (btScalar)btScalar(0.0), (btScalar)btScalar(1.0));
 	m_angularDamping = btClamped(ang_damping, (btScalar)btScalar(0.0), (btScalar)btScalar(1.0));
 }
 
-
-
-
 ///applyDamping damps the velocity, using the given m_linearDamping and m_angularDamping
-void			btRigidBody::applyDamping(btScalar timeStep)
+void btRigidBody::applyDamping(btScalar timeStep)
 {
 	//On new damping: see discussion/issue report here: http://code.google.com/p/bullet/issues/detail?id=74
 	//todo: do some performance comparisons (but other parts of the engine are probably bottleneck anyway
@@ -168,8 +151,8 @@ void			btRigidBody::applyDamping(btScalar timeStep)
 	m_linearVelocity *= GEN_clamped((btScalar(1.) - timeStep * m_linearDamping), (btScalar)btScalar(0.0), (btScalar)btScalar(1.0));
 	m_angularVelocity *= GEN_clamped((btScalar(1.) - timeStep * m_angularDamping), (btScalar)btScalar(0.0), (btScalar)btScalar(1.0));
 #else
-	m_linearVelocity *= btPow(btScalar(1)-m_linearDamping, timeStep);
-	m_angularVelocity *= btPow(btScalar(1)-m_angularDamping, timeStep);
+	m_linearVelocity *= btPow(btScalar(1) - m_linearDamping, timeStep);
+	m_angularVelocity *= btPow(btScalar(1) - m_angularDamping, timeStep);
 #endif
 
 	if (m_additionalDamping)
@@ -182,7 +165,6 @@ void			btRigidBody::applyDamping(btScalar timeStep)
 			m_angularVelocity *= m_additionalDampingFactor;
 			m_linearVelocity *= m_additionalDampingFactor;
 		}
-	
 
 		btScalar speed = m_linearVelocity.length();
 		if (speed < m_linearDamping)
@@ -191,10 +173,11 @@ void			btRigidBody::applyDamping(btScalar timeStep)
 			if (speed > dampVel)
 			{
 				btVector3 dir = m_linearVelocity.normalized();
-				m_linearVelocity -=  dir * dampVel;
-			} else
+				m_linearVelocity -= dir * dampVel;
+			}
+			else
 			{
-				m_linearVelocity.setValue(btScalar(0.),btScalar(0.),btScalar(0.));
+				m_linearVelocity.setValue(btScalar(0.), btScalar(0.), btScalar(0.));
 			}
 		}
 
@@ -205,30 +188,28 @@ void			btRigidBody::applyDamping(btScalar timeStep)
 			if (angSpeed > angDampVel)
 			{
 				btVector3 dir = m_angularVelocity.normalized();
-				m_angularVelocity -=  dir * angDampVel;
-			} else
+				m_angularVelocity -= dir * angDampVel;
+			}
+			else
 			{
-				m_angularVelocity.setValue(btScalar(0.),btScalar(0.),btScalar(0.));
+				m_angularVelocity.setValue(btScalar(0.), btScalar(0.), btScalar(0.));
 			}
 		}
 	}
 }
 
-
 void btRigidBody::applyGravity()
 {
 	if (isStaticOrKinematicObject())
 		return;
-	
-	applyCentralForce(m_gravity);	
 
+	applyCentralForce(m_gravity);
 }
 
 void btRigidBody::proceedToTransform(const btTransform& newTrans)
 {
-	setCenterOfMassTransform( newTrans );
+	setCenterOfMassTransform(newTrans);
 }
-	
 
 void btRigidBody::setMassProps(btScalar mass, const btVector3& inertia)
 {
@@ -236,7 +217,8 @@ void btRigidBody::setMassProps(btScalar mass, const btVector3& inertia)
 	{
 		m_collisionFlags |= btCollisionObject::CF_STATIC_OBJECT;
 		m_inverseMass = btScalar(0.);
-	} else
+	}
+	else
 	{
 		m_collisionFlags &= (~btCollisionObject::CF_STATIC_OBJECT);
 		m_inverseMass = btScalar(1.0) / mass;
@@ -244,50 +226,45 @@ void btRigidBody::setMassProps(btScalar mass, const btVector3& inertia)
 
 	//Fg = m * a
 	m_gravity = mass * m_gravity_acceleration;
-	
-	m_invInertiaLocal.setValue(inertia.x() != btScalar(0.0) ? btScalar(1.0) / inertia.x(): btScalar(0.0),
-				   inertia.y() != btScalar(0.0) ? btScalar(1.0) / inertia.y(): btScalar(0.0),
-				   inertia.z() != btScalar(0.0) ? btScalar(1.0) / inertia.z(): btScalar(0.0));
 
-	m_invMass = m_linearFactor*m_inverseMass;
+	m_invInertiaLocal.setValue(inertia.x() != btScalar(0.0) ? btScalar(1.0) / inertia.x() : btScalar(0.0),
+							   inertia.y() != btScalar(0.0) ? btScalar(1.0) / inertia.y() : btScalar(0.0),
+							   inertia.z() != btScalar(0.0) ? btScalar(1.0) / inertia.z() : btScalar(0.0));
+
+	m_invMass = m_linearFactor * m_inverseMass;
 }
 
-	
-void btRigidBody::updateInertiaTensor() 
+void btRigidBody::updateInertiaTensor()
 {
 	m_invInertiaTensorWorld = m_worldTransform.getBasis().scaled(m_invInertiaLocal) * m_worldTransform.getBasis().transpose();
 }
 
-
-
 btVector3 btRigidBody::getLocalInertia() const
 {
-
 	btVector3 inertiaLocal;
 	const btVector3 inertia = m_invInertiaLocal;
 	inertiaLocal.setValue(inertia.x() != btScalar(0.0) ? btScalar(1.0) / inertia.x() : btScalar(0.0),
-		inertia.y() != btScalar(0.0) ? btScalar(1.0) / inertia.y() : btScalar(0.0),
-		inertia.z() != btScalar(0.0) ? btScalar(1.0) / inertia.z() : btScalar(0.0));
+						  inertia.y() != btScalar(0.0) ? btScalar(1.0) / inertia.y() : btScalar(0.0),
+						  inertia.z() != btScalar(0.0) ? btScalar(1.0) / inertia.z() : btScalar(0.0));
 	return inertiaLocal;
 }
 
 inline btVector3 evalEulerEqn(const btVector3& w1, const btVector3& w0, const btVector3& T, const btScalar dt,
-	const btMatrix3x3 &I)
+							  const btMatrix3x3& I)
 {
-	const btVector3 w2 = I*w1 + w1.cross(I*w1)*dt - (T*dt + I*w0);
+	const btVector3 w2 = I * w1 + w1.cross(I * w1) * dt - (T * dt + I * w0);
 	return w2;
 }
 
 inline btMatrix3x3 evalEulerEqnDeriv(const btVector3& w1, const btVector3& w0, const btScalar dt,
-	const btMatrix3x3 &I)
+									 const btMatrix3x3& I)
 {
-
 	btMatrix3x3 w1x, Iw1x;
-	const btVector3 Iwi = (I*w1);
+	const btVector3 Iwi = (I * w1);
 	w1.getSkewSymmetricMatrix(&w1x[0], &w1x[1], &w1x[2]);
 	Iwi.getSkewSymmetricMatrix(&Iw1x[0], &Iw1x[1], &Iw1x[2]);
 
-	const btMatrix3x3 dfw1 = I + (w1x*I - Iw1x)*dt;
+	const btMatrix3x3 dfw1 = I + (w1x * I - Iw1x) * dt;
 	return dfw1;
 }
 
@@ -295,58 +272,55 @@ btVector3 btRigidBody::computeGyroscopicForceExplicit(btScalar maxGyroscopicForc
 {
 	btVector3 inertiaLocal = getLocalInertia();
 	btMatrix3x3 inertiaTensorWorld = getWorldTransform().getBasis().scaled(inertiaLocal) * getWorldTransform().getBasis().transpose();
-	btVector3 tmp = inertiaTensorWorld*getAngularVelocity();
+	btVector3 tmp = inertiaTensorWorld * getAngularVelocity();
 	btVector3 gf = getAngularVelocity().cross(tmp);
 	btScalar l2 = gf.length2();
-	if (l2>maxGyroscopicForce*maxGyroscopicForce)
+	if (l2 > maxGyroscopicForce * maxGyroscopicForce)
 	{
-		gf *= btScalar(1.)/btSqrt(l2)*maxGyroscopicForce;
+		gf *= btScalar(1.) / btSqrt(l2) * maxGyroscopicForce;
 	}
 	return gf;
 }
 
-
 btVector3 btRigidBody::computeGyroscopicImpulseImplicit_Body(btScalar step) const
-{	
+{
 	btVector3 idl = getLocalInertia();
 	btVector3 omega1 = getAngularVelocity();
 	btQuaternion q = getWorldTransform().getRotation();
-	
+
 	// Convert to body coordinates
 	btVector3 omegab = quatRotate(q.inverse(), omega1);
 	btMatrix3x3 Ib;
-	Ib.setValue(idl.x(),0,0,
-				0,idl.y(),0,
-				0,0,idl.z());
-	
-	btVector3 ibo = Ib*omegab;
+	Ib.setValue(idl.x(), 0, 0,
+				0, idl.y(), 0,
+				0, 0, idl.z());
+
+	btVector3 ibo = Ib * omegab;
 
 	// Residual vector
 	btVector3 f = step * omegab.cross(ibo);
-	
+
 	btMatrix3x3 skew0;
 	omegab.getSkewSymmetricMatrix(&skew0[0], &skew0[1], &skew0[2]);
-	btVector3 om = Ib*omegab;
+	btVector3 om = Ib * omegab;
 	btMatrix3x3 skew1;
-	om.getSkewSymmetricMatrix(&skew1[0],&skew1[1],&skew1[2]);
-	
+	om.getSkewSymmetricMatrix(&skew1[0], &skew1[1], &skew1[2]);
+
 	// Jacobian
-	btMatrix3x3 J = Ib +  (skew0*Ib - skew1)*step;
-	
-//	btMatrix3x3 Jinv = J.inverse();
-//	btVector3 omega_div = Jinv*f;
+	btMatrix3x3 J = Ib + (skew0 * Ib - skew1) * step;
+
+	//	btMatrix3x3 Jinv = J.inverse();
+	//	btVector3 omega_div = Jinv*f;
 	btVector3 omega_div = J.solve33(f);
-	
+
 	// Single Newton-Raphson update
-	omegab = omegab - omega_div;//Solve33(J, f);
+	omegab = omegab - omega_div;  //Solve33(J, f);
 	// Back to world coordinates
-	btVector3 omega2 = quatRotate(q,omegab);
-	btVector3 gf = omega2-omega1;
+	btVector3 omega2 = quatRotate(q, omegab);
+	btVector3 gf = omega2 - omega1;
 	return gf;
 }
 
-
-
 btVector3 btRigidBody::computeGyroscopicImpulseImplicit_World(btScalar step) const
 {
 	// use full newton-euler equations.  common practice to drop the wxIw term. want it for better tumbling behavior.
@@ -361,7 +335,7 @@ btVector3 btRigidBody::computeGyroscopicImpulseImplicit_World(btScalar step) con
 		m_worldTransform.getBasis().transpose();
 
 	// use newtons method to find implicit solution for new angular velocity (w')
-	// f(w') = -(T*step + Iw) + Iw' + w' + w'xIw'*step = 0 
+	// f(w') = -(T*step + Iw) + Iw' + w' + w'xIw'*step = 0
 	// df/dw' = I + 1xIw'*step + w'xI*step
 
 	btVector3 w1 = w0;
@@ -383,8 +357,7 @@ btVector3 btRigidBody::computeGyroscopicImpulseImplicit_World(btScalar step) con
 	return gf;
 }
 
-
-void btRigidBody::integrateVelocities(btScalar step) 
+void btRigidBody::integrateVelocities(btScalar step)
 {
 	if (isStaticOrKinematicObject())
 		return;
@@ -393,30 +366,28 @@ void btRigidBody::integrateVelocities(btScalar step)
 	m_angularVelocity += m_invInertiaTensorWorld * m_totalTorque * step;
 
 #define MAX_ANGVEL SIMD_HALF_PI
-	/// clamp angular velocity. collision calculations will fail on higher angular velocities	
+	/// clamp angular velocity. collision calculations will fail on higher angular velocities
 	btScalar angvel = m_angularVelocity.length();
-	if (angvel*step > MAX_ANGVEL)
+	if (angvel * step > MAX_ANGVEL)
 	{
-		m_angularVelocity *= (MAX_ANGVEL/step) /angvel;
+		m_angularVelocity *= (MAX_ANGVEL / step) / angvel;
 	}
-
 }
 
 btQuaternion btRigidBody::getOrientation() const
 {
-		btQuaternion orn;
-		m_worldTransform.getBasis().getRotation(orn);
-		return orn;
+	btQuaternion orn;
+	m_worldTransform.getBasis().getRotation(orn);
+	return orn;
 }
-	
-	
+
 void btRigidBody::setCenterOfMassTransform(const btTransform& xform)
 {
-
 	if (isKinematicObject())
 	{
 		m_interpolationWorldTransform = m_worldTransform;
-	} else
+	}
+	else
 	{
 		m_interpolationWorldTransform = xform;
 	}
@@ -426,10 +397,6 @@ void btRigidBody::setCenterOfMassTransform(const btTransform& xform)
 	updateInertiaTensor();
 }
 
-
-
-
-
 void btRigidBody::addConstraintRef(btTypedConstraint* c)
 {
 	///disable collision with the 'other' body
@@ -450,39 +417,39 @@ void btRigidBody::addConstraintRef(btTypedConstraint* c)
 		{
 			colObjB->setIgnoreCollisionCheck(colObjA, true);
 		}
-	} 
+	}
 }
 
 void btRigidBody::removeConstraintRef(btTypedConstraint* c)
 {
 	int index = m_constraintRefs.findLinearSearch(c);
 	//don't remove constraints that are not referenced
-	if(index < m_constraintRefs.size())
-    {
-        m_constraintRefs.remove(c);
-        btCollisionObject* colObjA = &c->getRigidBodyA();
-        btCollisionObject* colObjB = &c->getRigidBodyB();
-        if (colObjA == this)
-        {
-            colObjA->setIgnoreCollisionCheck(colObjB, false);
-        }
-        else
-        {
-            colObjB->setIgnoreCollisionCheck(colObjA, false);
-        }
-    }
+	if (index < m_constraintRefs.size())
+	{
+		m_constraintRefs.remove(c);
+		btCollisionObject* colObjA = &c->getRigidBodyA();
+		btCollisionObject* colObjB = &c->getRigidBodyB();
+		if (colObjA == this)
+		{
+			colObjA->setIgnoreCollisionCheck(colObjB, false);
+		}
+		else
+		{
+			colObjB->setIgnoreCollisionCheck(colObjA, false);
+		}
+	}
 }
 
-int	btRigidBody::calculateSerializeBufferSize()	const
+int btRigidBody::calculateSerializeBufferSize() const
 {
 	int sz = sizeof(btRigidBodyData);
 	return sz;
 }
 
-	///fills the dataBuffer and returns the struct name (and 0 on failure)
-const char*	btRigidBody::serialize(void* dataBuffer, class btSerializer* serializer) const
+///fills the dataBuffer and returns the struct name (and 0 on failure)
+const char* btRigidBody::serialize(void* dataBuffer, class btSerializer* serializer) const
 {
-	btRigidBodyData* rbd = (btRigidBodyData*) dataBuffer;
+	btRigidBodyData* rbd = (btRigidBodyData*)dataBuffer;
 
 	btCollisionObject::serialize(&rbd->m_collisionObjectData, serializer);
 
@@ -504,7 +471,7 @@ const char*	btRigidBody::serialize(void* dataBuffer, class btSerializer* seriali
 	rbd->m_additionalLinearDampingThresholdSqr = m_additionalLinearDampingThresholdSqr;
 	rbd->m_additionalAngularDampingThresholdSqr = m_additionalAngularDampingThresholdSqr;
 	rbd->m_additionalAngularDampingFactor = m_additionalAngularDampingFactor;
-	rbd->m_linearSleepingThreshold=m_linearSleepingThreshold;
+	rbd->m_linearSleepingThreshold = m_linearSleepingThreshold;
 	rbd->m_angularSleepingThreshold = m_angularSleepingThreshold;
 
 	// Fill padding with zeros to appease msan.
@@ -515,13 +482,9 @@ const char*	btRigidBody::serialize(void* dataBuffer, class btSerializer* seriali
 	return btRigidBodyDataName;
 }
 
-
-
 void btRigidBody::serializeSingleObject(class btSerializer* serializer) const
 {
-	btChunk* chunk = serializer->allocate(calculateSerializeBufferSize(),1);
+	btChunk* chunk = serializer->allocate(calculateSerializeBufferSize(), 1);
 	const char* structType = serialize(chunk->m_oldPtr, serializer);
-	serializer->finalizeChunk(chunk,structType,BT_RIGIDBODY_CODE,(void*)this);
+	serializer->finalizeChunk(chunk, structType, BT_RIGIDBODY_CODE, (void*)this);
 }
-
-
diff --git a/thirdparty/bullet/BulletDynamics/Dynamics/btRigidBody.h b/thirdparty/bullet/BulletDynamics/Dynamics/btRigidBody.h
index 372245031b1..05f270a4b8e 100644
--- a/thirdparty/bullet/BulletDynamics/Dynamics/btRigidBody.h
+++ b/thirdparty/bullet/BulletDynamics/Dynamics/btRigidBody.h
@@ -25,209 +25,195 @@ class btCollisionShape;
 class btMotionState;
 class btTypedConstraint;
 
-
 extern btScalar gDeactivationTime;
 extern bool gDisableDeactivation;
 
 #ifdef BT_USE_DOUBLE_PRECISION
-#define btRigidBodyData	btRigidBodyDoubleData
-#define btRigidBodyDataName	"btRigidBodyDoubleData"
+#define btRigidBodyData btRigidBodyDoubleData
+#define btRigidBodyDataName "btRigidBodyDoubleData"
 #else
-#define btRigidBodyData	btRigidBodyFloatData
-#define btRigidBodyDataName	"btRigidBodyFloatData"
-#endif //BT_USE_DOUBLE_PRECISION
+#define btRigidBodyData btRigidBodyFloatData
+#define btRigidBodyDataName "btRigidBodyFloatData"
+#endif  //BT_USE_DOUBLE_PRECISION
 
-
-enum	btRigidBodyFlags
+enum btRigidBodyFlags
 {
 	BT_DISABLE_WORLD_GRAVITY = 1,
 	///BT_ENABLE_GYROPSCOPIC_FORCE flags is enabled by default in Bullet 2.83 and onwards.
 	///and it BT_ENABLE_GYROPSCOPIC_FORCE becomes equivalent to BT_ENABLE_GYROSCOPIC_FORCE_IMPLICIT_BODY
 	///See Demos/GyroscopicDemo and computeGyroscopicImpulseImplicit
 	BT_ENABLE_GYROSCOPIC_FORCE_EXPLICIT = 2,
-	BT_ENABLE_GYROSCOPIC_FORCE_IMPLICIT_WORLD=4,
-	BT_ENABLE_GYROSCOPIC_FORCE_IMPLICIT_BODY=8,
+	BT_ENABLE_GYROSCOPIC_FORCE_IMPLICIT_WORLD = 4,
+	BT_ENABLE_GYROSCOPIC_FORCE_IMPLICIT_BODY = 8,
 	BT_ENABLE_GYROPSCOPIC_FORCE = BT_ENABLE_GYROSCOPIC_FORCE_IMPLICIT_BODY,
 };
 
-
 ///The btRigidBody is the main class for rigid body objects. It is derived from btCollisionObject, so it keeps a pointer to a btCollisionShape.
 ///It is recommended for performance and memory use to share btCollisionShape objects whenever possible.
-///There are 3 types of rigid bodies: 
+///There are 3 types of rigid bodies:
 ///- A) Dynamic rigid bodies, with positive mass. Motion is controlled by rigid body dynamics.
 ///- B) Fixed objects with zero mass. They are not moving (basically collision objects)
-///- C) Kinematic objects, which are objects without mass, but the user can move them. There is on-way interaction, and Bullet calculates a velocity based on the timestep and previous and current world transform.
+///- C) Kinematic objects, which are objects without mass, but the user can move them. There is one-way interaction, and Bullet calculates a velocity based on the timestep and previous and current world transform.
 ///Bullet automatically deactivates dynamic rigid bodies, when the velocity is below a threshold for a given time.
 ///Deactivated (sleeping) rigid bodies don't take any processing time, except a minor broadphase collision detection impact (to allow active objects to activate/wake up sleeping objects)
-class btRigidBody  : public btCollisionObject
+class btRigidBody : public btCollisionObject
 {
+	btMatrix3x3 m_invInertiaTensorWorld;
+	btVector3 m_linearVelocity;
+	btVector3 m_angularVelocity;
+	btScalar m_inverseMass;
+	btVector3 m_linearFactor;
 
-	btMatrix3x3	m_invInertiaTensorWorld;
-	btVector3		m_linearVelocity;
-	btVector3		m_angularVelocity;
-	btScalar		m_inverseMass;
-	btVector3		m_linearFactor;
+	btVector3 m_gravity;
+	btVector3 m_gravity_acceleration;
+	btVector3 m_invInertiaLocal;
+	btVector3 m_totalForce;
+	btVector3 m_totalTorque;
 
-	btVector3		m_gravity;	
-	btVector3		m_gravity_acceleration;
-	btVector3		m_invInertiaLocal;
-	btVector3		m_totalForce;
-	btVector3		m_totalTorque;
-	
-	btScalar		m_linearDamping;
-	btScalar		m_angularDamping;
+	btScalar m_linearDamping;
+	btScalar m_angularDamping;
 
-	bool			m_additionalDamping;
-	btScalar		m_additionalDampingFactor;
-	btScalar		m_additionalLinearDampingThresholdSqr;
-	btScalar		m_additionalAngularDampingThresholdSqr;
-	btScalar		m_additionalAngularDampingFactor;
+	bool m_additionalDamping;
+	btScalar m_additionalDampingFactor;
+	btScalar m_additionalLinearDampingThresholdSqr;
+	btScalar m_additionalAngularDampingThresholdSqr;
+	btScalar m_additionalAngularDampingFactor;
 
-
-	btScalar		m_linearSleepingThreshold;
-	btScalar		m_angularSleepingThreshold;
+	btScalar m_linearSleepingThreshold;
+	btScalar m_angularSleepingThreshold;
 
 	//m_optionalMotionState allows to automatic synchronize the world transform for active objects
-	btMotionState*	m_optionalMotionState;
+	btMotionState* m_optionalMotionState;
 
 	//keep track of typed constraints referencing this rigid body, to disable collision between linked bodies
 	btAlignedObjectArray<btTypedConstraint*> m_constraintRefs;
 
-	int				m_rigidbodyFlags;
-	
-	int				m_debugBodyId;
-	
+	int m_rigidbodyFlags;
+
+	int m_debugBodyId;
 
 protected:
-
-	ATTRIBUTE_ALIGNED16(btVector3		m_deltaLinearVelocity);
-	btVector3		m_deltaAngularVelocity;
-	btVector3		m_angularFactor;
-	btVector3		m_invMass;
-	btVector3		m_pushVelocity;
-	btVector3		m_turnVelocity;
-
+	ATTRIBUTE_ALIGNED16(btVector3 m_deltaLinearVelocity);
+	btVector3 m_deltaAngularVelocity;
+	btVector3 m_angularFactor;
+	btVector3 m_invMass;
+	btVector3 m_pushVelocity;
+	btVector3 m_turnVelocity;
 
 public:
-
-
 	///The btRigidBodyConstructionInfo structure provides information to create a rigid body. Setting mass to zero creates a fixed (non-dynamic) rigid body.
 	///For dynamic objects, you can use the collision shape to approximate the local inertia tensor, otherwise use the zero vector (default argument)
-	///You can use the motion state to synchronize the world transform between physics and graphics objects. 
+	///You can use the motion state to synchronize the world transform between physics and graphics objects.
 	///And if the motion state is provided, the rigid body will initialize its initial world transform from the motion state,
 	///m_startWorldTransform is only used when you don't provide a motion state.
-	struct	btRigidBodyConstructionInfo
+	struct btRigidBodyConstructionInfo
 	{
-		btScalar			m_mass;
+		btScalar m_mass;
 
 		///When a motionState is provided, the rigid body will initialize its world transform from the motion state
 		///In this case, m_startWorldTransform is ignored.
-		btMotionState*		m_motionState;
-		btTransform	m_startWorldTransform;
+		btMotionState* m_motionState;
+		btTransform m_startWorldTransform;
 
-		btCollisionShape*	m_collisionShape;
-		btVector3			m_localInertia;
-		btScalar			m_linearDamping;
-		btScalar			m_angularDamping;
+		btCollisionShape* m_collisionShape;
+		btVector3 m_localInertia;
+		btScalar m_linearDamping;
+		btScalar m_angularDamping;
 
 		///best simulation results when friction is non-zero
-		btScalar			m_friction;
+		btScalar m_friction;
 		///the m_rollingFriction prevents rounded shapes, such as spheres, cylinders and capsules from rolling forever.
 		///See Bullet/Demos/RollingFrictionDemo for usage
-		btScalar			m_rollingFriction;
-        btScalar			m_spinningFriction;//torsional friction around contact normal
-        
-		///best simulation results using zero restitution.
-		btScalar			m_restitution;
+		btScalar m_rollingFriction;
+		btScalar m_spinningFriction;  //torsional friction around contact normal
 
-		btScalar			m_linearSleepingThreshold;
-		btScalar			m_angularSleepingThreshold;
+		///best simulation results using zero restitution.
+		btScalar m_restitution;
+
+		btScalar m_linearSleepingThreshold;
+		btScalar m_angularSleepingThreshold;
 
 		//Additional damping can help avoiding lowpass jitter motion, help stability for ragdolls etc.
 		//Such damping is undesirable, so once the overall simulation quality of the rigid body dynamics system has improved, this should become obsolete
-		bool				m_additionalDamping;
-		btScalar			m_additionalDampingFactor;
-		btScalar			m_additionalLinearDampingThresholdSqr;
-		btScalar			m_additionalAngularDampingThresholdSqr;
-		btScalar			m_additionalAngularDampingFactor;
+		bool m_additionalDamping;
+		btScalar m_additionalDampingFactor;
+		btScalar m_additionalLinearDampingThresholdSqr;
+		btScalar m_additionalAngularDampingThresholdSqr;
+		btScalar m_additionalAngularDampingFactor;
 
-		btRigidBodyConstructionInfo(	btScalar mass, btMotionState* motionState, btCollisionShape* collisionShape, const btVector3& localInertia=btVector3(0,0,0)):
-		m_mass(mass),
-			m_motionState(motionState),
-			m_collisionShape(collisionShape),
-			m_localInertia(localInertia),
-			m_linearDamping(btScalar(0.)),
-			m_angularDamping(btScalar(0.)),
-			m_friction(btScalar(0.5)),
-			m_rollingFriction(btScalar(0)),
-            m_spinningFriction(btScalar(0)),
-			m_restitution(btScalar(0.)),
-			m_linearSleepingThreshold(btScalar(0.8)),
-			m_angularSleepingThreshold(btScalar(1.f)),
-			m_additionalDamping(false),
-			m_additionalDampingFactor(btScalar(0.005)),
-			m_additionalLinearDampingThresholdSqr(btScalar(0.01)),
-			m_additionalAngularDampingThresholdSqr(btScalar(0.01)),
-			m_additionalAngularDampingFactor(btScalar(0.01))
+		btRigidBodyConstructionInfo(btScalar mass, btMotionState* motionState, btCollisionShape* collisionShape, const btVector3& localInertia = btVector3(0, 0, 0)) : m_mass(mass),
+																																									   m_motionState(motionState),
+																																									   m_collisionShape(collisionShape),
+																																									   m_localInertia(localInertia),
+																																									   m_linearDamping(btScalar(0.)),
+																																									   m_angularDamping(btScalar(0.)),
+																																									   m_friction(btScalar(0.5)),
+																																									   m_rollingFriction(btScalar(0)),
+																																									   m_spinningFriction(btScalar(0)),
+																																									   m_restitution(btScalar(0.)),
+																																									   m_linearSleepingThreshold(btScalar(0.8)),
+																																									   m_angularSleepingThreshold(btScalar(1.f)),
+																																									   m_additionalDamping(false),
+																																									   m_additionalDampingFactor(btScalar(0.005)),
+																																									   m_additionalLinearDampingThresholdSqr(btScalar(0.01)),
+																																									   m_additionalAngularDampingThresholdSqr(btScalar(0.01)),
+																																									   m_additionalAngularDampingFactor(btScalar(0.01))
 		{
 			m_startWorldTransform.setIdentity();
 		}
 	};
 
 	///btRigidBody constructor using construction info
-	btRigidBody(	const btRigidBodyConstructionInfo& constructionInfo);
+	btRigidBody(const btRigidBodyConstructionInfo& constructionInfo);
 
-	///btRigidBody constructor for backwards compatibility. 
+	///btRigidBody constructor for backwards compatibility.
 	///To specify friction (etc) during rigid body construction, please use the other constructor (using btRigidBodyConstructionInfo)
-	btRigidBody(	btScalar mass, btMotionState* motionState, btCollisionShape* collisionShape, const btVector3& localInertia=btVector3(0,0,0));
-
+	btRigidBody(btScalar mass, btMotionState* motionState, btCollisionShape* collisionShape, const btVector3& localInertia = btVector3(0, 0, 0));
 
 	virtual ~btRigidBody()
-        { 
-                //No constraints should point to this rigidbody
-		//Remove constraints from the dynamics world before you delete the related rigidbodies. 
-                btAssert(m_constraintRefs.size()==0); 
-        }
+	{
+		//No constraints should point to this rigidbody
+		//Remove constraints from the dynamics world before you delete the related rigidbodies.
+		btAssert(m_constraintRefs.size() == 0);
+	}
 
 protected:
-
 	///setupRigidBody is only used internally by the constructor
-	void	setupRigidBody(const btRigidBodyConstructionInfo& constructionInfo);
+	void setupRigidBody(const btRigidBodyConstructionInfo& constructionInfo);
 
 public:
+	void proceedToTransform(const btTransform& newTrans);
 
-	void			proceedToTransform(const btTransform& newTrans); 
-	
 	///to keep collision detection and dynamics separate we don't store a rigidbody pointer
 	///but a rigidbody is derived from btCollisionObject, so we can safely perform an upcast
-	static const btRigidBody*	upcast(const btCollisionObject* colObj)
+	static const btRigidBody* upcast(const btCollisionObject* colObj)
 	{
-		if (colObj->getInternalType()&btCollisionObject::CO_RIGID_BODY)
+		if (colObj->getInternalType() & btCollisionObject::CO_RIGID_BODY)
 			return (const btRigidBody*)colObj;
 		return 0;
 	}
-	static btRigidBody*	upcast(btCollisionObject* colObj)
+	static btRigidBody* upcast(btCollisionObject* colObj)
 	{
-		if (colObj->getInternalType()&btCollisionObject::CO_RIGID_BODY)
+		if (colObj->getInternalType() & btCollisionObject::CO_RIGID_BODY)
 			return (btRigidBody*)colObj;
 		return 0;
 	}
-	
-	/// continuous collision detection needs prediction
-	void			predictIntegratedTransform(btScalar step, btTransform& predictedTransform) ;
-	
-	void			saveKinematicState(btScalar step);
-	
-	void			applyGravity();
-	
-	void			setGravity(const btVector3& acceleration);  
 
-	const btVector3&	getGravity() const
+	/// continuous collision detection needs prediction
+	void predictIntegratedTransform(btScalar step, btTransform& predictedTransform);
+
+	void saveKinematicState(btScalar step);
+
+	void applyGravity();
+
+	void setGravity(const btVector3& acceleration);
+
+	const btVector3& getGravity() const
 	{
 		return m_gravity_acceleration;
 	}
 
-	void			setDamping(btScalar lin_damping, btScalar ang_damping);
+	void setDamping(btScalar lin_damping, btScalar ang_damping);
 
 	btScalar getLinearDamping() const
 	{
@@ -249,18 +235,20 @@ public:
 		return m_angularSleepingThreshold;
 	}
 
-	void			applyDamping(btScalar timeStep);
+	void applyDamping(btScalar timeStep);
 
-	SIMD_FORCE_INLINE const btCollisionShape*	getCollisionShape() const {
+	SIMD_FORCE_INLINE const btCollisionShape* getCollisionShape() const
+	{
 		return m_collisionShape;
 	}
 
-	SIMD_FORCE_INLINE btCollisionShape*	getCollisionShape() {
-			return m_collisionShape;
+	SIMD_FORCE_INLINE btCollisionShape* getCollisionShape()
+	{
+		return m_collisionShape;
 	}
-	
-	void			setMassProps(btScalar mass, const btVector3& inertia);
-	
+
+	void setMassProps(btScalar mass, const btVector3& inertia);
+
 	const btVector3& getLinearFactor() const
 	{
 		return m_linearFactor;
@@ -268,20 +256,21 @@ public:
 	void setLinearFactor(const btVector3& linearFactor)
 	{
 		m_linearFactor = linearFactor;
-		m_invMass = m_linearFactor*m_inverseMass;
+		m_invMass = m_linearFactor * m_inverseMass;
 	}
-	btScalar		getInvMass() const { return m_inverseMass; }
-	const btMatrix3x3& getInvInertiaTensorWorld() const { 
-		return m_invInertiaTensorWorld; 
-	}
-		
-	void			integrateVelocities(btScalar step);
-
-	void			setCenterOfMassTransform(const btTransform& xform);
-
-	void			applyCentralForce(const btVector3& force)
+	btScalar getInvMass() const { return m_inverseMass; }
+	const btMatrix3x3& getInvInertiaTensorWorld() const
 	{
-		m_totalForce += force*m_linearFactor;
+		return m_invInertiaTensorWorld;
+	}
+
+	void integrateVelocities(btScalar step);
+
+	void setCenterOfMassTransform(const btTransform& xform);
+
+	void applyCentralForce(const btVector3& force)
+	{
+		m_totalForce += force * m_linearFactor;
 	}
 
 	const btVector3& getTotalForce() const
@@ -293,90 +282,93 @@ public:
 	{
 		return m_totalTorque;
 	};
-    
+
 	const btVector3& getInvInertiaDiagLocal() const
 	{
 		return m_invInertiaLocal;
 	};
 
-	void	setInvInertiaDiagLocal(const btVector3& diagInvInertia)
+	void setInvInertiaDiagLocal(const btVector3& diagInvInertia)
 	{
 		m_invInertiaLocal = diagInvInertia;
 	}
 
-	void	setSleepingThresholds(btScalar linear,btScalar angular)
+	void setSleepingThresholds(btScalar linear, btScalar angular)
 	{
 		m_linearSleepingThreshold = linear;
 		m_angularSleepingThreshold = angular;
 	}
 
-	void	applyTorque(const btVector3& torque)
+	void applyTorque(const btVector3& torque)
 	{
-		m_totalTorque += torque*m_angularFactor;
+		m_totalTorque += torque * m_angularFactor;
 	}
-	
-	void	applyForce(const btVector3& force, const btVector3& rel_pos) 
+
+	void applyForce(const btVector3& force, const btVector3& rel_pos)
 	{
 		applyCentralForce(force);
-		applyTorque(rel_pos.cross(force*m_linearFactor));
+		applyTorque(rel_pos.cross(force * m_linearFactor));
 	}
-	
+
 	void applyCentralImpulse(const btVector3& impulse)
 	{
-		m_linearVelocity += impulse *m_linearFactor * m_inverseMass;
+		m_linearVelocity += impulse * m_linearFactor * m_inverseMass;
 	}
-	
-  	void applyTorqueImpulse(const btVector3& torque)
+
+	void applyTorqueImpulse(const btVector3& torque)
 	{
-			m_angularVelocity += m_invInertiaTensorWorld * torque * m_angularFactor;
+		m_angularVelocity += m_invInertiaTensorWorld * torque * m_angularFactor;
 	}
-	
-	void applyImpulse(const btVector3& impulse, const btVector3& rel_pos) 
+
+	void applyImpulse(const btVector3& impulse, const btVector3& rel_pos)
 	{
 		if (m_inverseMass != btScalar(0.))
 		{
 			applyCentralImpulse(impulse);
 			if (m_angularFactor)
 			{
-				applyTorqueImpulse(rel_pos.cross(impulse*m_linearFactor));
+				applyTorqueImpulse(rel_pos.cross(impulse * m_linearFactor));
 			}
 		}
 	}
 
-	void clearForces() 
+	void clearForces()
 	{
 		m_totalForce.setValue(btScalar(0.0), btScalar(0.0), btScalar(0.0));
 		m_totalTorque.setValue(btScalar(0.0), btScalar(0.0), btScalar(0.0));
 	}
-	
-	void updateInertiaTensor();    
-	
-	const btVector3&     getCenterOfMassPosition() const { 
-		return m_worldTransform.getOrigin(); 
+
+	void updateInertiaTensor();
+
+	const btVector3& getCenterOfMassPosition() const
+	{
+		return m_worldTransform.getOrigin();
 	}
 	btQuaternion getOrientation() const;
-	
-	const btTransform&  getCenterOfMassTransform() const { 
-		return m_worldTransform; 
+
+	const btTransform& getCenterOfMassTransform() const
+	{
+		return m_worldTransform;
 	}
-	const btVector3&   getLinearVelocity() const { 
-		return m_linearVelocity; 
+	const btVector3& getLinearVelocity() const
+	{
+		return m_linearVelocity;
 	}
-	const btVector3&    getAngularVelocity() const { 
-		return m_angularVelocity; 
+	const btVector3& getAngularVelocity() const
+	{
+		return m_angularVelocity;
 	}
-	
 
 	inline void setLinearVelocity(const btVector3& lin_vel)
-	{ 
+	{
 		m_updateRevision++;
-		m_linearVelocity = lin_vel; 
+		m_linearVelocity = lin_vel;
 	}
 
-	inline void setAngularVelocity(const btVector3& ang_vel) 
-	{ 
+	inline void setAngularVelocity(const btVector3& ang_vel)
+	{
 		m_updateRevision++;
-		m_angularVelocity = ang_vel; 
+		m_angularVelocity = ang_vel;
 	}
 
 	btVector3 getVelocityInLocalPoint(const btVector3& rel_pos) const
@@ -388,18 +380,13 @@ public:
 		//		return 	(m_worldTransform(rel_pos) - m_interpolationWorldTransform(rel_pos)) / m_kinematicTimeStep;
 	}
 
-	void translate(const btVector3& v) 
+	void translate(const btVector3& v)
 	{
-		m_worldTransform.getOrigin() += v; 
+		m_worldTransform.getOrigin() += v;
 	}
 
-	
-	void	getAabb(btVector3& aabbMin,btVector3& aabbMax) const;
+	void getAabb(btVector3& aabbMin, btVector3& aabbMax) const;
 
-
-
-
-	
 	SIMD_FORCE_INLINE btScalar computeImpulseDenominator(const btVector3& pos, const btVector3& normal) const
 	{
 		btVector3 r0 = pos - getCenterOfMassPosition();
@@ -409,7 +396,6 @@ public:
 		btVector3 vec = (c0 * getInvInertiaTensorWorld()).cross(r0);
 
 		return m_inverseMass + normal.dot(vec);
-
 	}
 
 	SIMD_FORCE_INLINE btScalar computeAngularImpulseDenominator(const btVector3& axis) const
@@ -418,26 +404,25 @@ public:
 		return axis.dot(vec);
 	}
 
-	SIMD_FORCE_INLINE void	updateDeactivation(btScalar timeStep)
+	SIMD_FORCE_INLINE void updateDeactivation(btScalar timeStep)
 	{
-		if ( (getActivationState() == ISLAND_SLEEPING) || (getActivationState() == DISABLE_DEACTIVATION))
+		if ((getActivationState() == ISLAND_SLEEPING) || (getActivationState() == DISABLE_DEACTIVATION))
 			return;
 
-		if ((getLinearVelocity().length2() < m_linearSleepingThreshold*m_linearSleepingThreshold) &&
-			(getAngularVelocity().length2() < m_angularSleepingThreshold*m_angularSleepingThreshold))
+		if ((getLinearVelocity().length2() < m_linearSleepingThreshold * m_linearSleepingThreshold) &&
+			(getAngularVelocity().length2() < m_angularSleepingThreshold * m_angularSleepingThreshold))
 		{
 			m_deactivationTime += timeStep;
-		} else
+		}
+		else
 		{
-			m_deactivationTime=btScalar(0.);
+			m_deactivationTime = btScalar(0.);
 			setActivationState(0);
 		}
-
 	}
 
-	SIMD_FORCE_INLINE bool	wantsSleeping()
+	SIMD_FORCE_INLINE bool wantsSleeping()
 	{
-
 		if (getActivationState() == DISABLE_DEACTIVATION)
 			return false;
 
@@ -445,41 +430,39 @@ public:
 		if (gDisableDeactivation || (gDeactivationTime == btScalar(0.)))
 			return false;
 
-		if ( (getActivationState() == ISLAND_SLEEPING) || (getActivationState() == WANTS_DEACTIVATION))
+		if ((getActivationState() == ISLAND_SLEEPING) || (getActivationState() == WANTS_DEACTIVATION))
 			return true;
 
-		if (m_deactivationTime> gDeactivationTime)
+		if (m_deactivationTime > gDeactivationTime)
 		{
 			return true;
 		}
 		return false;
 	}
 
-
-	
-	const btBroadphaseProxy*	getBroadphaseProxy() const
+	const btBroadphaseProxy* getBroadphaseProxy() const
 	{
 		return m_broadphaseHandle;
 	}
-	btBroadphaseProxy*	getBroadphaseProxy() 
+	btBroadphaseProxy* getBroadphaseProxy()
 	{
 		return m_broadphaseHandle;
 	}
-	void	setNewBroadphaseProxy(btBroadphaseProxy* broadphaseProxy)
+	void setNewBroadphaseProxy(btBroadphaseProxy* broadphaseProxy)
 	{
 		m_broadphaseHandle = broadphaseProxy;
 	}
 
 	//btMotionState allows to automatic synchronize the world transform for active objects
-	btMotionState*	getMotionState()
+	btMotionState* getMotionState()
 	{
 		return m_optionalMotionState;
 	}
-	const btMotionState*	getMotionState() const
+	const btMotionState* getMotionState() const
 	{
 		return m_optionalMotionState;
 	}
-	void	setMotionState(btMotionState* motionState)
+	void setMotionState(btMotionState* motionState)
 	{
 		m_optionalMotionState = motionState;
 		if (m_optionalMotionState)
@@ -487,27 +470,27 @@ public:
 	}
 
 	//for experimental overriding of friction/contact solver func
-	int	m_contactSolverType;
-	int	m_frictionSolverType;
+	int m_contactSolverType;
+	int m_frictionSolverType;
 
-	void	setAngularFactor(const btVector3& angFac)
+	void setAngularFactor(const btVector3& angFac)
 	{
 		m_updateRevision++;
 		m_angularFactor = angFac;
 	}
 
-	void	setAngularFactor(btScalar angFac)
+	void setAngularFactor(btScalar angFac)
 	{
 		m_updateRevision++;
-		m_angularFactor.setValue(angFac,angFac,angFac);
+		m_angularFactor.setValue(angFac, angFac, angFac);
 	}
-	const btVector3&	getAngularFactor() const
+	const btVector3& getAngularFactor() const
 	{
 		return m_angularFactor;
 	}
 
 	//is this rigidbody added to a btCollisionWorld/btDynamicsWorld/btBroadphase?
-	bool	isInWorld() const
+	bool isInWorld() const
 	{
 		return (getBroadphaseProxy() != 0);
 	}
@@ -525,7 +508,7 @@ public:
 		return m_constraintRefs.size();
 	}
 
-	void	setFlags(int flags)
+	void setFlags(int flags)
 	{
 		m_rigidbodyFlags = flags;
 	}
@@ -535,12 +518,9 @@ public:
 		return m_rigidbodyFlags;
 	}
 
-
-	
-
 	///perform implicit force computation in world space
 	btVector3 computeGyroscopicImpulseImplicit_World(btScalar dt) const;
-	
+
 	///perform implicit force computation in body space (inertial frame)
 	btVector3 computeGyroscopicImpulseImplicit_Body(btScalar step) const;
 
@@ -550,70 +530,66 @@ public:
 
 	///////////////////////////////////////////////
 
-	virtual	int	calculateSerializeBufferSize()	const;
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer,  class btSerializer* serializer) const;
+	virtual const char* serialize(void* dataBuffer, class btSerializer* serializer) const;
 
 	virtual void serializeSingleObject(class btSerializer* serializer) const;
-
 };
 
 //@todo add m_optionalMotionState and m_constraintRefs to btRigidBodyData
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
-struct	btRigidBodyFloatData
+struct btRigidBodyFloatData
 {
-	btCollisionObjectFloatData	m_collisionObjectData;
-	btMatrix3x3FloatData		m_invInertiaTensorWorld;
-	btVector3FloatData		m_linearVelocity;
-	btVector3FloatData		m_angularVelocity;
-	btVector3FloatData		m_angularFactor;
-	btVector3FloatData		m_linearFactor;
-	btVector3FloatData		m_gravity;	
-	btVector3FloatData		m_gravity_acceleration;
-	btVector3FloatData		m_invInertiaLocal;
-	btVector3FloatData		m_totalForce;
-	btVector3FloatData		m_totalTorque;
-	float					m_inverseMass;
-	float					m_linearDamping;
-	float					m_angularDamping;
-	float					m_additionalDampingFactor;
-	float					m_additionalLinearDampingThresholdSqr;
-	float					m_additionalAngularDampingThresholdSqr;
-	float					m_additionalAngularDampingFactor;
-	float					m_linearSleepingThreshold;
-	float					m_angularSleepingThreshold;
-	int						m_additionalDamping;
+	btCollisionObjectFloatData m_collisionObjectData;
+	btMatrix3x3FloatData m_invInertiaTensorWorld;
+	btVector3FloatData m_linearVelocity;
+	btVector3FloatData m_angularVelocity;
+	btVector3FloatData m_angularFactor;
+	btVector3FloatData m_linearFactor;
+	btVector3FloatData m_gravity;
+	btVector3FloatData m_gravity_acceleration;
+	btVector3FloatData m_invInertiaLocal;
+	btVector3FloatData m_totalForce;
+	btVector3FloatData m_totalTorque;
+	float m_inverseMass;
+	float m_linearDamping;
+	float m_angularDamping;
+	float m_additionalDampingFactor;
+	float m_additionalLinearDampingThresholdSqr;
+	float m_additionalAngularDampingThresholdSqr;
+	float m_additionalAngularDampingFactor;
+	float m_linearSleepingThreshold;
+	float m_angularSleepingThreshold;
+	int m_additionalDamping;
 };
 
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
-struct	btRigidBodyDoubleData
+struct btRigidBodyDoubleData
 {
-	btCollisionObjectDoubleData	m_collisionObjectData;
-	btMatrix3x3DoubleData		m_invInertiaTensorWorld;
-	btVector3DoubleData		m_linearVelocity;
-	btVector3DoubleData		m_angularVelocity;
-	btVector3DoubleData		m_angularFactor;
-	btVector3DoubleData		m_linearFactor;
-	btVector3DoubleData		m_gravity;	
-	btVector3DoubleData		m_gravity_acceleration;
-	btVector3DoubleData		m_invInertiaLocal;
-	btVector3DoubleData		m_totalForce;
-	btVector3DoubleData		m_totalTorque;
-	double					m_inverseMass;
-	double					m_linearDamping;
-	double					m_angularDamping;
-	double					m_additionalDampingFactor;
-	double					m_additionalLinearDampingThresholdSqr;
-	double					m_additionalAngularDampingThresholdSqr;
-	double					m_additionalAngularDampingFactor;
-	double					m_linearSleepingThreshold;
-	double					m_angularSleepingThreshold;
-	int						m_additionalDamping;
-	char	m_padding[4];
+	btCollisionObjectDoubleData m_collisionObjectData;
+	btMatrix3x3DoubleData m_invInertiaTensorWorld;
+	btVector3DoubleData m_linearVelocity;
+	btVector3DoubleData m_angularVelocity;
+	btVector3DoubleData m_angularFactor;
+	btVector3DoubleData m_linearFactor;
+	btVector3DoubleData m_gravity;
+	btVector3DoubleData m_gravity_acceleration;
+	btVector3DoubleData m_invInertiaLocal;
+	btVector3DoubleData m_totalForce;
+	btVector3DoubleData m_totalTorque;
+	double m_inverseMass;
+	double m_linearDamping;
+	double m_angularDamping;
+	double m_additionalDampingFactor;
+	double m_additionalLinearDampingThresholdSqr;
+	double m_additionalAngularDampingThresholdSqr;
+	double m_additionalAngularDampingFactor;
+	double m_linearSleepingThreshold;
+	double m_angularSleepingThreshold;
+	int m_additionalDamping;
+	char m_padding[4];
 };
 
-
-
-#endif //BT_RIGIDBODY_H
-
+#endif  //BT_RIGIDBODY_H
diff --git a/thirdparty/bullet/BulletDynamics/Dynamics/btSimpleDynamicsWorld.cpp b/thirdparty/bullet/BulletDynamics/Dynamics/btSimpleDynamicsWorld.cpp
index 6f63b87c80a..8103390fb1d 100644
--- a/thirdparty/bullet/BulletDynamics/Dynamics/btSimpleDynamicsWorld.cpp
+++ b/thirdparty/bullet/BulletDynamics/Dynamics/btSimpleDynamicsWorld.cpp
@@ -21,47 +21,40 @@ subject to the following restrictions:
 #include "BulletDynamics/ConstraintSolver/btSequentialImpulseConstraintSolver.h"
 #include "BulletDynamics/ConstraintSolver/btContactSolverInfo.h"
 
-
 /*
   Make sure this dummy function never changes so that it
   can be used by probes that are checking whether the
   library is actually installed.
 */
-extern "C" 
+extern "C"
 {
-	void btBulletDynamicsProbe ();
-	void btBulletDynamicsProbe () {}
+	void btBulletDynamicsProbe();
+	void btBulletDynamicsProbe() {}
 }
 
-
-
-
-btSimpleDynamicsWorld::btSimpleDynamicsWorld(btDispatcher* dispatcher,btBroadphaseInterface* pairCache,btConstraintSolver* constraintSolver,btCollisionConfiguration* collisionConfiguration)
-:btDynamicsWorld(dispatcher,pairCache,collisionConfiguration),
-m_constraintSolver(constraintSolver),
-m_ownsConstraintSolver(false),
-m_gravity(0,0,-10)
+btSimpleDynamicsWorld::btSimpleDynamicsWorld(btDispatcher* dispatcher, btBroadphaseInterface* pairCache, btConstraintSolver* constraintSolver, btCollisionConfiguration* collisionConfiguration)
+	: btDynamicsWorld(dispatcher, pairCache, collisionConfiguration),
+	  m_constraintSolver(constraintSolver),
+	  m_ownsConstraintSolver(false),
+	  m_gravity(0, 0, -10)
 {
-
 }
 
-
 btSimpleDynamicsWorld::~btSimpleDynamicsWorld()
 {
 	if (m_ownsConstraintSolver)
-		btAlignedFree( m_constraintSolver);
+		btAlignedFree(m_constraintSolver);
 }
 
-int		btSimpleDynamicsWorld::stepSimulation( btScalar timeStep,int maxSubSteps, btScalar fixedTimeStep)
+int btSimpleDynamicsWorld::stepSimulation(btScalar timeStep, int maxSubSteps, btScalar fixedTimeStep)
 {
 	(void)fixedTimeStep;
 	(void)maxSubSteps;
 
-
 	///apply gravity, predict motion
 	predictUnconstraintMotion(timeStep);
 
-	btDispatcherInfo&	dispatchInfo = getDispatchInfo();
+	btDispatcherInfo& dispatchInfo = getDispatchInfo();
 	dispatchInfo.m_timeStep = timeStep;
 	dispatchInfo.m_stepCount = 0;
 	dispatchInfo.m_debugDraw = getDebugDrawer();
@@ -74,17 +67,17 @@ int		btSimpleDynamicsWorld::stepSimulation( btScalar timeStep,int maxSubSteps, b
 	if (numManifolds)
 	{
 		btPersistentManifold** manifoldPtr = ((btCollisionDispatcher*)m_dispatcher1)->getInternalManifoldPointer();
-		
+
 		btContactSolverInfo infoGlobal;
 		infoGlobal.m_timeStep = timeStep;
-		m_constraintSolver->prepareSolve(0,numManifolds);
-		m_constraintSolver->solveGroup(&getCollisionObjectArray()[0],getNumCollisionObjects(),manifoldPtr, numManifolds,0,0,infoGlobal,m_debugDrawer, m_dispatcher1);
-		m_constraintSolver->allSolved(infoGlobal,m_debugDrawer);
+		m_constraintSolver->prepareSolve(0, numManifolds);
+		m_constraintSolver->solveGroup(&getCollisionObjectArray()[0], getNumCollisionObjects(), manifoldPtr, numManifolds, 0, 0, infoGlobal, m_debugDrawer, m_dispatcher1);
+		m_constraintSolver->allSolved(infoGlobal, m_debugDrawer);
 	}
 
 	///integrate transforms
 	integrateTransforms(timeStep);
-		
+
 	updateAabbs();
 
 	synchronizeMotionStates();
@@ -92,29 +85,27 @@ int		btSimpleDynamicsWorld::stepSimulation( btScalar timeStep,int maxSubSteps, b
 	clearForces();
 
 	return 1;
-
 }
 
-void	btSimpleDynamicsWorld::clearForces()
+void btSimpleDynamicsWorld::clearForces()
 {
 	///@todo: iterate over awake simulation islands!
-	for ( int i=0;i<m_collisionObjects.size();i++)
+	for (int i = 0; i < m_collisionObjects.size(); i++)
 	{
 		btCollisionObject* colObj = m_collisionObjects[i];
-		
+
 		btRigidBody* body = btRigidBody::upcast(colObj);
 		if (body)
 		{
 			body->clearForces();
 		}
 	}
-}	
+}
 
-
-void	btSimpleDynamicsWorld::setGravity(const btVector3& gravity)
+void btSimpleDynamicsWorld::setGravity(const btVector3& gravity)
 {
 	m_gravity = gravity;
-	for ( int i=0;i<m_collisionObjects.size();i++)
+	for (int i = 0; i < m_collisionObjects.size(); i++)
 	{
 		btCollisionObject* colObj = m_collisionObjects[i];
 		btRigidBody* body = btRigidBody::upcast(colObj);
@@ -125,17 +116,17 @@ void	btSimpleDynamicsWorld::setGravity(const btVector3& gravity)
 	}
 }
 
-btVector3 btSimpleDynamicsWorld::getGravity () const
+btVector3 btSimpleDynamicsWorld::getGravity() const
 {
 	return m_gravity;
 }
 
-void	btSimpleDynamicsWorld::removeRigidBody(btRigidBody* body)
+void btSimpleDynamicsWorld::removeRigidBody(btRigidBody* body)
 {
 	btCollisionWorld::removeCollisionObject(body);
 }
 
-void	btSimpleDynamicsWorld::removeCollisionObject(btCollisionObject* collisionObject)
+void btSimpleDynamicsWorld::removeCollisionObject(btCollisionObject* collisionObject)
 {
 	btRigidBody* body = btRigidBody::upcast(collisionObject);
 	if (body)
@@ -144,8 +135,7 @@ void	btSimpleDynamicsWorld::removeCollisionObject(btCollisionObject* collisionOb
 		btCollisionWorld::removeCollisionObject(collisionObject);
 }
 
-
-void	btSimpleDynamicsWorld::addRigidBody(btRigidBody* body)
+void btSimpleDynamicsWorld::addRigidBody(btRigidBody* body)
 {
 	body->setGravity(m_gravity);
 
@@ -155,37 +145,32 @@ void	btSimpleDynamicsWorld::addRigidBody(btRigidBody* body)
 	}
 }
 
-void	btSimpleDynamicsWorld::addRigidBody(btRigidBody* body, int group, int mask)
+void btSimpleDynamicsWorld::addRigidBody(btRigidBody* body, int group, int mask)
 {
 	body->setGravity(m_gravity);
 
 	if (body->getCollisionShape())
 	{
-		addCollisionObject(body,group,mask);
+		addCollisionObject(body, group, mask);
 	}
 }
 
-
-void	btSimpleDynamicsWorld::debugDrawWorld()
+void btSimpleDynamicsWorld::debugDrawWorld()
 {
-
-}
-				
-void	btSimpleDynamicsWorld::addAction(btActionInterface* action)
-{
-
 }
 
-void	btSimpleDynamicsWorld::removeAction(btActionInterface* action)
+void btSimpleDynamicsWorld::addAction(btActionInterface* action)
 {
-
 }
 
+void btSimpleDynamicsWorld::removeAction(btActionInterface* action)
+{
+}
 
-void	btSimpleDynamicsWorld::updateAabbs()
+void btSimpleDynamicsWorld::updateAabbs()
 {
 	btTransform predictedTrans;
-	for ( int i=0;i<m_collisionObjects.size();i++)
+	for (int i = 0; i < m_collisionObjects.size(); i++)
 	{
 		btCollisionObject* colObj = m_collisionObjects[i];
 		btRigidBody* body = btRigidBody::upcast(colObj);
@@ -193,19 +178,19 @@ void	btSimpleDynamicsWorld::updateAabbs()
 		{
 			if (body->isActive() && (!body->isStaticObject()))
 			{
-				btVector3 minAabb,maxAabb;
-				colObj->getCollisionShape()->getAabb(colObj->getWorldTransform(), minAabb,maxAabb);
+				btVector3 minAabb, maxAabb;
+				colObj->getCollisionShape()->getAabb(colObj->getWorldTransform(), minAabb, maxAabb);
 				btBroadphaseInterface* bp = getBroadphase();
-				bp->setAabb(body->getBroadphaseHandle(),minAabb,maxAabb, m_dispatcher1);
+				bp->setAabb(body->getBroadphaseHandle(), minAabb, maxAabb, m_dispatcher1);
 			}
 		}
 	}
 }
 
-void	btSimpleDynamicsWorld::integrateTransforms(btScalar timeStep)
+void btSimpleDynamicsWorld::integrateTransforms(btScalar timeStep)
 {
 	btTransform predictedTrans;
-	for ( int i=0;i<m_collisionObjects.size();i++)
+	for (int i = 0; i < m_collisionObjects.size(); i++)
 	{
 		btCollisionObject* colObj = m_collisionObjects[i];
 		btRigidBody* body = btRigidBody::upcast(colObj);
@@ -214,17 +199,15 @@ void	btSimpleDynamicsWorld::integrateTransforms(btScalar timeStep)
 			if (body->isActive() && (!body->isStaticObject()))
 			{
 				body->predictIntegratedTransform(timeStep, predictedTrans);
-				body->proceedToTransform( predictedTrans);
+				body->proceedToTransform(predictedTrans);
 			}
 		}
 	}
 }
 
-
-
-void	btSimpleDynamicsWorld::predictUnconstraintMotion(btScalar timeStep)
+void btSimpleDynamicsWorld::predictUnconstraintMotion(btScalar timeStep)
 {
-	for ( int i=0;i<m_collisionObjects.size();i++)
+	for (int i = 0; i < m_collisionObjects.size(); i++)
 	{
 		btCollisionObject* colObj = m_collisionObjects[i];
 		btRigidBody* body = btRigidBody::upcast(colObj);
@@ -235,20 +218,19 @@ void	btSimpleDynamicsWorld::predictUnconstraintMotion(btScalar timeStep)
 				if (body->isActive())
 				{
 					body->applyGravity();
-					body->integrateVelocities( timeStep);
+					body->integrateVelocities(timeStep);
 					body->applyDamping(timeStep);
-					body->predictIntegratedTransform(timeStep,body->getInterpolationWorldTransform());
+					body->predictIntegratedTransform(timeStep, body->getInterpolationWorldTransform());
 				}
 			}
 		}
 	}
 }
 
-
-void	btSimpleDynamicsWorld::synchronizeMotionStates()
+void btSimpleDynamicsWorld::synchronizeMotionStates()
 {
 	///@todo: iterate over awake simulation islands!
-	for ( int i=0;i<m_collisionObjects.size();i++)
+	for (int i = 0; i < m_collisionObjects.size(); i++)
 	{
 		btCollisionObject* colObj = m_collisionObjects[i];
 		btRigidBody* body = btRigidBody::upcast(colObj);
@@ -260,11 +242,9 @@ void	btSimpleDynamicsWorld::synchronizeMotionStates()
 			}
 		}
 	}
-
 }
 
-
-void	btSimpleDynamicsWorld::setConstraintSolver(btConstraintSolver* solver)
+void btSimpleDynamicsWorld::setConstraintSolver(btConstraintSolver* solver)
 {
 	if (m_ownsConstraintSolver)
 	{
diff --git a/thirdparty/bullet/BulletDynamics/Dynamics/btSimpleDynamicsWorld.h b/thirdparty/bullet/BulletDynamics/Dynamics/btSimpleDynamicsWorld.h
index 44b7e7fb34b..12be231c7f1 100644
--- a/thirdparty/bullet/BulletDynamics/Dynamics/btSimpleDynamicsWorld.h
+++ b/thirdparty/bullet/BulletDynamics/Dynamics/btSimpleDynamicsWorld.h
@@ -27,63 +27,58 @@ class btConstraintSolver;
 class btSimpleDynamicsWorld : public btDynamicsWorld
 {
 protected:
+	btConstraintSolver* m_constraintSolver;
 
-	btConstraintSolver*	m_constraintSolver;
+	bool m_ownsConstraintSolver;
 
-	bool	m_ownsConstraintSolver;
+	void predictUnconstraintMotion(btScalar timeStep);
+
+	void integrateTransforms(btScalar timeStep);
+
+	btVector3 m_gravity;
 
-	void	predictUnconstraintMotion(btScalar timeStep);
-	
-	void	integrateTransforms(btScalar timeStep);
-		
-	btVector3	m_gravity;
-	
 public:
-
-
-
 	///this btSimpleDynamicsWorld constructor creates dispatcher, broadphase pairCache and constraintSolver
-	btSimpleDynamicsWorld(btDispatcher* dispatcher,btBroadphaseInterface* pairCache,btConstraintSolver* constraintSolver,btCollisionConfiguration* collisionConfiguration);
+	btSimpleDynamicsWorld(btDispatcher* dispatcher, btBroadphaseInterface* pairCache, btConstraintSolver* constraintSolver, btCollisionConfiguration* collisionConfiguration);
 
 	virtual ~btSimpleDynamicsWorld();
-		
+
 	///maxSubSteps/fixedTimeStep for interpolation is currently ignored for btSimpleDynamicsWorld, use btDiscreteDynamicsWorld instead
-	virtual int	stepSimulation( btScalar timeStep,int maxSubSteps=1, btScalar fixedTimeStep=btScalar(1.)/btScalar(60.));
+	virtual int stepSimulation(btScalar timeStep, int maxSubSteps = 1, btScalar fixedTimeStep = btScalar(1.) / btScalar(60.));
 
-	virtual void	setGravity(const btVector3& gravity);
+	virtual void setGravity(const btVector3& gravity);
 
-	virtual btVector3 getGravity () const;
+	virtual btVector3 getGravity() const;
 
-	virtual void	addRigidBody(btRigidBody* body);
+	virtual void addRigidBody(btRigidBody* body);
 
-	virtual void	addRigidBody(btRigidBody* body, int group, int mask);
+	virtual void addRigidBody(btRigidBody* body, int group, int mask);
 
-	virtual void	removeRigidBody(btRigidBody* body);
+	virtual void removeRigidBody(btRigidBody* body);
 
-	virtual void	debugDrawWorld();
-				
-	virtual void	addAction(btActionInterface* action);
+	virtual void debugDrawWorld();
 
-	virtual void	removeAction(btActionInterface* action);
+	virtual void addAction(btActionInterface* action);
+
+	virtual void removeAction(btActionInterface* action);
 
 	///removeCollisionObject will first check if it is a rigid body, if so call removeRigidBody otherwise call btCollisionWorld::removeCollisionObject
-	virtual void	removeCollisionObject(btCollisionObject* collisionObject);
-	
-	virtual void	updateAabbs();
+	virtual void removeCollisionObject(btCollisionObject* collisionObject);
 
-	virtual void	synchronizeMotionStates();
+	virtual void updateAabbs();
 
-	virtual void	setConstraintSolver(btConstraintSolver* solver);
+	virtual void synchronizeMotionStates();
+
+	virtual void setConstraintSolver(btConstraintSolver* solver);
 
 	virtual btConstraintSolver* getConstraintSolver();
 
-	virtual btDynamicsWorldType	getWorldType() const
+	virtual btDynamicsWorldType getWorldType() const
 	{
 		return BT_SIMPLE_DYNAMICS_WORLD;
 	}
 
-	virtual void	clearForces();
-
+	virtual void clearForces();
 };
 
-#endif //BT_SIMPLE_DYNAMICS_WORLD_H
+#endif  //BT_SIMPLE_DYNAMICS_WORLD_H
diff --git a/thirdparty/bullet/BulletDynamics/Dynamics/btSimulationIslandManagerMt.cpp b/thirdparty/bullet/BulletDynamics/Dynamics/btSimulationIslandManagerMt.cpp
index fc54f0ba6ea..17287aa82a4 100644
--- a/thirdparty/bullet/BulletDynamics/Dynamics/btSimulationIslandManagerMt.cpp
+++ b/thirdparty/bullet/BulletDynamics/Dynamics/btSimulationIslandManagerMt.cpp
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "LinearMath/btScalar.h"
 #include "LinearMath/btThreads.h"
 #include "btSimulationIslandManagerMt.h"
@@ -27,273 +26,259 @@ subject to the following restrictions:
 //#include <stdio.h>
 #include "LinearMath/btQuickprof.h"
 
-
-SIMD_FORCE_INLINE int calcBatchCost( int bodies, int manifolds, int constraints )
+SIMD_FORCE_INLINE int calcBatchCost(int bodies, int manifolds, int constraints)
 {
-    // rough estimate of the cost of a batch, used for merging
-    int batchCost = bodies + 8 * manifolds + 4 * constraints;
-    return batchCost;
+	// rough estimate of the cost of a batch, used for merging
+	int batchCost = bodies + 8 * manifolds + 4 * constraints;
+	return batchCost;
 }
 
-
-SIMD_FORCE_INLINE int calcBatchCost( const btSimulationIslandManagerMt::Island* island )
+SIMD_FORCE_INLINE int calcBatchCost(const btSimulationIslandManagerMt::Island* island)
 {
-    return calcBatchCost( island->bodyArray.size(), island->manifoldArray.size(), island->constraintArray.size() );
+	return calcBatchCost(island->bodyArray.size(), island->manifoldArray.size(), island->constraintArray.size());
 }
 
-
 btSimulationIslandManagerMt::btSimulationIslandManagerMt()
 {
-    m_minimumSolverBatchSize = calcBatchCost(0, 128, 0);
-    m_batchIslandMinBodyCount = 32;
-    m_islandDispatch = parallelIslandDispatch;
-    m_batchIsland = NULL;
+	m_minimumSolverBatchSize = calcBatchCost(0, 128, 0);
+	m_batchIslandMinBodyCount = 32;
+	m_islandDispatch = parallelIslandDispatch;
+	m_batchIsland = NULL;
 }
 
-
 btSimulationIslandManagerMt::~btSimulationIslandManagerMt()
 {
-    for ( int i = 0; i < m_allocatedIslands.size(); ++i )
-    {
-        delete m_allocatedIslands[ i ];
-    }
-    m_allocatedIslands.resize( 0 );
-    m_activeIslands.resize( 0 );
-    m_freeIslands.resize( 0 );
+	for (int i = 0; i < m_allocatedIslands.size(); ++i)
+	{
+		delete m_allocatedIslands[i];
+	}
+	m_allocatedIslands.resize(0);
+	m_activeIslands.resize(0);
+	m_freeIslands.resize(0);
 }
 
-
-inline	int	getIslandId(const btPersistentManifold* lhs)
+inline int getIslandId(const btPersistentManifold* lhs)
 {
 	const btCollisionObject* rcolObj0 = static_cast<const btCollisionObject*>(lhs->getBody0());
 	const btCollisionObject* rcolObj1 = static_cast<const btCollisionObject*>(lhs->getBody1());
-    int islandId = rcolObj0->getIslandTag() >= 0 ? rcolObj0->getIslandTag() : rcolObj1->getIslandTag();
+	int islandId = rcolObj0->getIslandTag() >= 0 ? rcolObj0->getIslandTag() : rcolObj1->getIslandTag();
 	return islandId;
 }
 
-
-SIMD_FORCE_INLINE	int	btGetConstraintIslandId( const btTypedConstraint* lhs )
+SIMD_FORCE_INLINE int btGetConstraintIslandId(const btTypedConstraint* lhs)
 {
-    const btCollisionObject& rcolObj0 = lhs->getRigidBodyA();
-    const btCollisionObject& rcolObj1 = lhs->getRigidBodyB();
-    int islandId = rcolObj0.getIslandTag() >= 0 ? rcolObj0.getIslandTag() : rcolObj1.getIslandTag();
-    return islandId;
+	const btCollisionObject& rcolObj0 = lhs->getRigidBodyA();
+	const btCollisionObject& rcolObj1 = lhs->getRigidBodyB();
+	int islandId = rcolObj0.getIslandTag() >= 0 ? rcolObj0.getIslandTag() : rcolObj1.getIslandTag();
+	return islandId;
 }
 
 /// function object that routes calls to operator<
 class IslandBatchSizeSortPredicate
 {
 public:
-    bool operator() ( const btSimulationIslandManagerMt::Island* lhs, const btSimulationIslandManagerMt::Island* rhs ) const
-    {
-        int lCost = calcBatchCost( lhs );
-        int rCost = calcBatchCost( rhs );
-        return lCost > rCost;
-    }
+	bool operator()(const btSimulationIslandManagerMt::Island* lhs, const btSimulationIslandManagerMt::Island* rhs) const
+	{
+		int lCost = calcBatchCost(lhs);
+		int rCost = calcBatchCost(rhs);
+		return lCost > rCost;
+	}
 };
 
-
 class IslandBodyCapacitySortPredicate
 {
 public:
-    bool operator() ( const btSimulationIslandManagerMt::Island* lhs, const btSimulationIslandManagerMt::Island* rhs ) const
-    {
-        return lhs->bodyArray.capacity() > rhs->bodyArray.capacity();
-    }
+	bool operator()(const btSimulationIslandManagerMt::Island* lhs, const btSimulationIslandManagerMt::Island* rhs) const
+	{
+		return lhs->bodyArray.capacity() > rhs->bodyArray.capacity();
+	}
 };
 
-
-void btSimulationIslandManagerMt::Island::append( const Island& other )
+void btSimulationIslandManagerMt::Island::append(const Island& other)
 {
-    // append bodies
-    for ( int i = 0; i < other.bodyArray.size(); ++i )
-    {
-        bodyArray.push_back( other.bodyArray[ i ] );
-    }
-    // append manifolds
-    for ( int i = 0; i < other.manifoldArray.size(); ++i )
-    {
-        manifoldArray.push_back( other.manifoldArray[ i ] );
-    }
-    // append constraints
-    for ( int i = 0; i < other.constraintArray.size(); ++i )
-    {
-        constraintArray.push_back( other.constraintArray[ i ] );
-    }
+	// append bodies
+	for (int i = 0; i < other.bodyArray.size(); ++i)
+	{
+		bodyArray.push_back(other.bodyArray[i]);
+	}
+	// append manifolds
+	for (int i = 0; i < other.manifoldArray.size(); ++i)
+	{
+		manifoldArray.push_back(other.manifoldArray[i]);
+	}
+	// append constraints
+	for (int i = 0; i < other.constraintArray.size(); ++i)
+	{
+		constraintArray.push_back(other.constraintArray[i]);
+	}
 }
 
-
-bool btIsBodyInIsland( const btSimulationIslandManagerMt::Island& island, const btCollisionObject* obj )
+bool btIsBodyInIsland(const btSimulationIslandManagerMt::Island& island, const btCollisionObject* obj)
 {
-    for ( int i = 0; i < island.bodyArray.size(); ++i )
-    {
-        if ( island.bodyArray[ i ] == obj )
-        {
-            return true;
-        }
-    }
-    return false;
+	for (int i = 0; i < island.bodyArray.size(); ++i)
+	{
+		if (island.bodyArray[i] == obj)
+		{
+			return true;
+		}
+	}
+	return false;
 }
 
-
 void btSimulationIslandManagerMt::initIslandPools()
 {
-    // reset island pools
-    int numElem = getUnionFind().getNumElements();
-    m_lookupIslandFromId.resize( numElem );
-    for ( int i = 0; i < m_lookupIslandFromId.size(); ++i )
-    {
-        m_lookupIslandFromId[ i ] = NULL;
-    }
-    m_activeIslands.resize( 0 );
-    m_freeIslands.resize( 0 );
-    // check whether allocated islands are sorted by body capacity (largest to smallest)
-    int lastCapacity = 0;
-    bool isSorted = true;
-    for ( int i = 0; i < m_allocatedIslands.size(); ++i )
-    {
-        Island* island = m_allocatedIslands[ i ];
-        int cap = island->bodyArray.capacity();
-        if ( cap > lastCapacity )
-        {
-            isSorted = false;
-            break;
-        }
-        lastCapacity = cap;
-    }
-    if ( !isSorted )
-    {
-        m_allocatedIslands.quickSort( IslandBodyCapacitySortPredicate() );
-    }
+	// reset island pools
+	int numElem = getUnionFind().getNumElements();
+	m_lookupIslandFromId.resize(numElem);
+	for (int i = 0; i < m_lookupIslandFromId.size(); ++i)
+	{
+		m_lookupIslandFromId[i] = NULL;
+	}
+	m_activeIslands.resize(0);
+	m_freeIslands.resize(0);
+	// check whether allocated islands are sorted by body capacity (largest to smallest)
+	int lastCapacity = 0;
+	bool isSorted = true;
+	for (int i = 0; i < m_allocatedIslands.size(); ++i)
+	{
+		Island* island = m_allocatedIslands[i];
+		int cap = island->bodyArray.capacity();
+		if (cap > lastCapacity)
+		{
+			isSorted = false;
+			break;
+		}
+		lastCapacity = cap;
+	}
+	if (!isSorted)
+	{
+		m_allocatedIslands.quickSort(IslandBodyCapacitySortPredicate());
+	}
 
-    m_batchIsland = NULL;
-    // mark all islands free (but avoid deallocation)
-    for ( int i = 0; i < m_allocatedIslands.size(); ++i )
-    {
-        Island* island = m_allocatedIslands[ i ];
-        island->bodyArray.resize( 0 );
-        island->manifoldArray.resize( 0 );
-        island->constraintArray.resize( 0 );
-        island->id = -1;
-        island->isSleeping = true;
-        m_freeIslands.push_back( island );
-    }
+	m_batchIsland = NULL;
+	// mark all islands free (but avoid deallocation)
+	for (int i = 0; i < m_allocatedIslands.size(); ++i)
+	{
+		Island* island = m_allocatedIslands[i];
+		island->bodyArray.resize(0);
+		island->manifoldArray.resize(0);
+		island->constraintArray.resize(0);
+		island->id = -1;
+		island->isSleeping = true;
+		m_freeIslands.push_back(island);
+	}
 }
 
-
-btSimulationIslandManagerMt::Island* btSimulationIslandManagerMt::getIsland( int id )
+btSimulationIslandManagerMt::Island* btSimulationIslandManagerMt::getIsland(int id)
 {
-    Island* island = m_lookupIslandFromId[ id ];
-    if ( island == NULL )
-    {
-        // search for existing island
-        for ( int i = 0; i < m_activeIslands.size(); ++i )
-        {
-            if ( m_activeIslands[ i ]->id == id )
-            {
-                island = m_activeIslands[ i ];
-                break;
-            }
-        }
-        m_lookupIslandFromId[ id ] = island;
-    }
-    return island;
+	Island* island = m_lookupIslandFromId[id];
+	if (island == NULL)
+	{
+		// search for existing island
+		for (int i = 0; i < m_activeIslands.size(); ++i)
+		{
+			if (m_activeIslands[i]->id == id)
+			{
+				island = m_activeIslands[i];
+				break;
+			}
+		}
+		m_lookupIslandFromId[id] = island;
+	}
+	return island;
 }
 
-
-btSimulationIslandManagerMt::Island* btSimulationIslandManagerMt::allocateIsland( int id, int numBodies )
+btSimulationIslandManagerMt::Island* btSimulationIslandManagerMt::allocateIsland(int id, int numBodies)
 {
-    Island* island = NULL;
-    int allocSize = numBodies;
-    if ( numBodies < m_batchIslandMinBodyCount )
-    {
-        if ( m_batchIsland )
-        {
-            island = m_batchIsland;
-            m_lookupIslandFromId[ id ] = island;
-            // if we've made a large enough batch,
-            if ( island->bodyArray.size() + numBodies >= m_batchIslandMinBodyCount )
-            {
-                // next time start a new batch
-                m_batchIsland = NULL;
-            }
-            return island;
-        }
-        else
-        {
-            // need to allocate a batch island
-            allocSize = m_batchIslandMinBodyCount * 2;
-        }
-    }
-    btAlignedObjectArray<Island*>& freeIslands = m_freeIslands;
+	Island* island = NULL;
+	int allocSize = numBodies;
+	if (numBodies < m_batchIslandMinBodyCount)
+	{
+		if (m_batchIsland)
+		{
+			island = m_batchIsland;
+			m_lookupIslandFromId[id] = island;
+			// if we've made a large enough batch,
+			if (island->bodyArray.size() + numBodies >= m_batchIslandMinBodyCount)
+			{
+				// next time start a new batch
+				m_batchIsland = NULL;
+			}
+			return island;
+		}
+		else
+		{
+			// need to allocate a batch island
+			allocSize = m_batchIslandMinBodyCount * 2;
+		}
+	}
+	btAlignedObjectArray<Island*>& freeIslands = m_freeIslands;
 
-    // search for free island
-    if ( freeIslands.size() > 0 )
-    {
-        // try to reuse a previously allocated island
-        int iFound = freeIslands.size();
-        // linear search for smallest island that can hold our bodies
-        for ( int i = freeIslands.size() - 1; i >= 0; --i )
-        {
-            if ( freeIslands[ i ]->bodyArray.capacity() >= allocSize )
-            {
-                iFound = i;
-                island = freeIslands[ i ];
-                island->id = id;
-                break;
-            }
-        }
-        // if found, shrink array while maintaining ordering
-        if ( island )
-        {
-            int iDest = iFound;
-            int iSrc = iDest + 1;
-            while ( iSrc < freeIslands.size() )
-            {
-                freeIslands[ iDest++ ] = freeIslands[ iSrc++ ];
-            }
-            freeIslands.pop_back();
-        }
-    }
-    if ( island == NULL )
-    {
-        // no free island found, allocate
-        island = new Island();  // TODO: change this to use the pool allocator
-        island->id = id;
-        island->bodyArray.reserve( allocSize );
-        m_allocatedIslands.push_back( island );
-    }
-    m_lookupIslandFromId[ id ] = island;
-    if ( numBodies < m_batchIslandMinBodyCount )
-    {
-        m_batchIsland = island;
-    }
-    m_activeIslands.push_back( island );
-    return island;
+	// search for free island
+	if (freeIslands.size() > 0)
+	{
+		// try to reuse a previously allocated island
+		int iFound = freeIslands.size();
+		// linear search for smallest island that can hold our bodies
+		for (int i = freeIslands.size() - 1; i >= 0; --i)
+		{
+			if (freeIslands[i]->bodyArray.capacity() >= allocSize)
+			{
+				iFound = i;
+				island = freeIslands[i];
+				island->id = id;
+				break;
+			}
+		}
+		// if found, shrink array while maintaining ordering
+		if (island)
+		{
+			int iDest = iFound;
+			int iSrc = iDest + 1;
+			while (iSrc < freeIslands.size())
+			{
+				freeIslands[iDest++] = freeIslands[iSrc++];
+			}
+			freeIslands.pop_back();
+		}
+	}
+	if (island == NULL)
+	{
+		// no free island found, allocate
+		island = new Island();  // TODO: change this to use the pool allocator
+		island->id = id;
+		island->bodyArray.reserve(allocSize);
+		m_allocatedIslands.push_back(island);
+	}
+	m_lookupIslandFromId[id] = island;
+	if (numBodies < m_batchIslandMinBodyCount)
+	{
+		m_batchIsland = island;
+	}
+	m_activeIslands.push_back(island);
+	return island;
 }
 
-
-void btSimulationIslandManagerMt::buildIslands( btDispatcher* dispatcher, btCollisionWorld* collisionWorld )
+void btSimulationIslandManagerMt::buildIslands(btDispatcher* dispatcher, btCollisionWorld* collisionWorld)
 {
-
 	BT_PROFILE("buildIslands");
-	
+
 	btCollisionObjectArray& collisionObjects = collisionWorld->getCollisionObjectArray();
 
 	//we are going to sort the unionfind array, and store the element id in the size
 	//afterwards, we clean unionfind, to make sure no-one uses it anymore
-	
+
 	getUnionFind().sortIslands();
 	int numElem = getUnionFind().getNumElements();
 
-	int endIslandIndex=1;
+	int endIslandIndex = 1;
 	int startIslandIndex;
 
 	//update the sleeping state for bodies, if all are sleeping
-	for ( startIslandIndex=0;startIslandIndex<numElem;startIslandIndex = endIslandIndex)
+	for (startIslandIndex = 0; startIslandIndex < numElem; startIslandIndex = endIslandIndex)
 	{
 		int islandId = getUnionFind().getElement(startIslandIndex).m_id;
-		for (endIslandIndex = startIslandIndex+1;(endIslandIndex<numElem) && (getUnionFind().getElement(endIslandIndex).m_id == islandId);endIslandIndex++)
+		for (endIslandIndex = startIslandIndex + 1; (endIslandIndex < numElem) && (getUnionFind().getElement(endIslandIndex).m_id == islandId); endIslandIndex++)
 		{
 		}
 
@@ -302,21 +287,21 @@ void btSimulationIslandManagerMt::buildIslands( btDispatcher* dispatcher, btColl
 		bool allSleeping = true;
 
 		int idx;
-		for (idx=startIslandIndex;idx<endIslandIndex;idx++)
+		for (idx = startIslandIndex; idx < endIslandIndex; idx++)
 		{
 			int i = getUnionFind().getElement(idx).m_sz;
 
 			btCollisionObject* colObj0 = collisionObjects[i];
 			if ((colObj0->getIslandTag() != islandId) && (colObj0->getIslandTag() != -1))
 			{
-//				printf("error in island management\n");
+				//				printf("error in island management\n");
 			}
 
 			btAssert((colObj0->getIslandTag() == islandId) || (colObj0->getIslandTag() == -1));
 			if (colObj0->getIslandTag() == islandId)
 			{
-				if (colObj0->getActivationState()== ACTIVE_TAG ||
-				   colObj0->getActivationState()== DISABLE_DEACTIVATION)
+				if (colObj0->getActivationState() == ACTIVE_TAG ||
+					colObj0->getActivationState() == DISABLE_DEACTIVATION)
 				{
 					allSleeping = false;
 					break;
@@ -327,43 +312,43 @@ void btSimulationIslandManagerMt::buildIslands( btDispatcher* dispatcher, btColl
 		if (allSleeping)
 		{
 			int idx;
-			for (idx=startIslandIndex;idx<endIslandIndex;idx++)
+			for (idx = startIslandIndex; idx < endIslandIndex; idx++)
 			{
 				int i = getUnionFind().getElement(idx).m_sz;
 				btCollisionObject* colObj0 = collisionObjects[i];
 				if ((colObj0->getIslandTag() != islandId) && (colObj0->getIslandTag() != -1))
 				{
-//					printf("error in island management\n");
+					//					printf("error in island management\n");
 				}
 
 				btAssert((colObj0->getIslandTag() == islandId) || (colObj0->getIslandTag() == -1));
 
 				if (colObj0->getIslandTag() == islandId)
 				{
-					colObj0->setActivationState( ISLAND_SLEEPING );
+					colObj0->setActivationState(ISLAND_SLEEPING);
 				}
 			}
-		} else
+		}
+		else
 		{
-
 			int idx;
-			for (idx=startIslandIndex;idx<endIslandIndex;idx++)
+			for (idx = startIslandIndex; idx < endIslandIndex; idx++)
 			{
 				int i = getUnionFind().getElement(idx).m_sz;
 
 				btCollisionObject* colObj0 = collisionObjects[i];
 				if ((colObj0->getIslandTag() != islandId) && (colObj0->getIslandTag() != -1))
 				{
-//					printf("error in island management\n");
+					//					printf("error in island management\n");
 				}
 
 				btAssert((colObj0->getIslandTag() == islandId) || (colObj0->getIslandTag() == -1));
 
 				if (colObj0->getIslandTag() == islandId)
 				{
-					if ( colObj0->getActivationState() == ISLAND_SLEEPING)
+					if (colObj0->getActivationState() == ISLAND_SLEEPING)
 					{
-						colObj0->setActivationState( WANTS_DEACTIVATION);
+						colObj0->setActivationState(WANTS_DEACTIVATION);
 						colObj0->setDeactivationTime(0.f);
 					}
 				}
@@ -372,352 +357,338 @@ void btSimulationIslandManagerMt::buildIslands( btDispatcher* dispatcher, btColl
 	}
 }
 
-
-void btSimulationIslandManagerMt::addBodiesToIslands( btCollisionWorld* collisionWorld )
+void btSimulationIslandManagerMt::addBodiesToIslands(btCollisionWorld* collisionWorld)
 {
-    btCollisionObjectArray& collisionObjects = collisionWorld->getCollisionObjectArray();
-    int endIslandIndex = 1;
-    int startIslandIndex;
-    int numElem = getUnionFind().getNumElements();
+	btCollisionObjectArray& collisionObjects = collisionWorld->getCollisionObjectArray();
+	int endIslandIndex = 1;
+	int startIslandIndex;
+	int numElem = getUnionFind().getNumElements();
 
-    // create explicit islands and add bodies to each
-    for ( startIslandIndex = 0; startIslandIndex < numElem; startIslandIndex = endIslandIndex )
-    {
-        int islandId = getUnionFind().getElement( startIslandIndex ).m_id;
+	// create explicit islands and add bodies to each
+	for (startIslandIndex = 0; startIslandIndex < numElem; startIslandIndex = endIslandIndex)
+	{
+		int islandId = getUnionFind().getElement(startIslandIndex).m_id;
 
-        // find end index
-        for ( endIslandIndex = startIslandIndex; ( endIslandIndex < numElem ) && ( getUnionFind().getElement( endIslandIndex ).m_id == islandId ); endIslandIndex++ )
-        {
-        }
-        // check if island is sleeping
-        bool islandSleeping = true;
-        for ( int iElem = startIslandIndex; iElem < endIslandIndex; iElem++ )
-        {
-            int i = getUnionFind().getElement( iElem ).m_sz;
-            btCollisionObject* colObj = collisionObjects[ i ];
-            if ( colObj->isActive() )
-            {
-                islandSleeping = false;
-            }
-        }
-        if ( !islandSleeping )
-        {
-            // want to count the number of bodies before allocating the island to optimize memory usage of the Island structures
-            int numBodies = endIslandIndex - startIslandIndex;
-            Island* island = allocateIsland( islandId, numBodies );
-            island->isSleeping = false;
-
-            // add bodies to island
-            for ( int iElem = startIslandIndex; iElem < endIslandIndex; iElem++ )
-            {
-                int i = getUnionFind().getElement( iElem ).m_sz;
-                btCollisionObject* colObj = collisionObjects[ i ];
-                island->bodyArray.push_back( colObj );
-            }
-        }
-    }
+		// find end index
+		for (endIslandIndex = startIslandIndex; (endIslandIndex < numElem) && (getUnionFind().getElement(endIslandIndex).m_id == islandId); endIslandIndex++)
+		{
+		}
+		// check if island is sleeping
+		bool islandSleeping = true;
+		for (int iElem = startIslandIndex; iElem < endIslandIndex; iElem++)
+		{
+			int i = getUnionFind().getElement(iElem).m_sz;
+			btCollisionObject* colObj = collisionObjects[i];
+			if (colObj->isActive())
+			{
+				islandSleeping = false;
+			}
+		}
+		if (!islandSleeping)
+		{
+			// want to count the number of bodies before allocating the island to optimize memory usage of the Island structures
+			int numBodies = endIslandIndex - startIslandIndex;
+			Island* island = allocateIsland(islandId, numBodies);
+			island->isSleeping = false;
 
+			// add bodies to island
+			for (int iElem = startIslandIndex; iElem < endIslandIndex; iElem++)
+			{
+				int i = getUnionFind().getElement(iElem).m_sz;
+				btCollisionObject* colObj = collisionObjects[i];
+				island->bodyArray.push_back(colObj);
+			}
+		}
+	}
 }
 
-
-void btSimulationIslandManagerMt::addManifoldsToIslands( btDispatcher* dispatcher )
+void btSimulationIslandManagerMt::addManifoldsToIslands(btDispatcher* dispatcher)
 {
-    // walk all the manifolds, activating bodies touched by kinematic objects, and add each manifold to its Island
-    int maxNumManifolds = dispatcher->getNumManifolds();
-    for ( int i = 0; i < maxNumManifolds; i++ )
-    {
-        btPersistentManifold* manifold = dispatcher->getManifoldByIndexInternal( i );
+	// walk all the manifolds, activating bodies touched by kinematic objects, and add each manifold to its Island
+	int maxNumManifolds = dispatcher->getNumManifolds();
+	for (int i = 0; i < maxNumManifolds; i++)
+	{
+		btPersistentManifold* manifold = dispatcher->getManifoldByIndexInternal(i);
 
-        const btCollisionObject* colObj0 = static_cast<const btCollisionObject*>( manifold->getBody0() );
-        const btCollisionObject* colObj1 = static_cast<const btCollisionObject*>( manifold->getBody1() );
+		const btCollisionObject* colObj0 = static_cast<const btCollisionObject*>(manifold->getBody0());
+		const btCollisionObject* colObj1 = static_cast<const btCollisionObject*>(manifold->getBody1());
 
-        ///@todo: check sleeping conditions!
-        if ( ( ( colObj0 ) && colObj0->getActivationState() != ISLAND_SLEEPING ) ||
-             ( ( colObj1 ) && colObj1->getActivationState() != ISLAND_SLEEPING ) )
-        {
-
-            //kinematic objects don't merge islands, but wake up all connected objects
-            if ( colObj0->isKinematicObject() && colObj0->getActivationState() != ISLAND_SLEEPING )
-            {
-                if ( colObj0->hasContactResponse() )
-                    colObj1->activate();
-            }
-            if ( colObj1->isKinematicObject() && colObj1->getActivationState() != ISLAND_SLEEPING )
-            {
-                if ( colObj1->hasContactResponse() )
-                    colObj0->activate();
-            }
-            //filtering for response
-            if ( dispatcher->needsResponse( colObj0, colObj1 ) )
-            {
-                // scatter manifolds into various islands
-                int islandId = getIslandId( manifold );
-                // if island not sleeping,
-                if ( Island* island = getIsland( islandId ) )
-                {
-                    island->manifoldArray.push_back( manifold );
-                }
-            }
-        }
-    }
+		///@todo: check sleeping conditions!
+		if (((colObj0) && colObj0->getActivationState() != ISLAND_SLEEPING) ||
+			((colObj1) && colObj1->getActivationState() != ISLAND_SLEEPING))
+		{
+			//kinematic objects don't merge islands, but wake up all connected objects
+			if (colObj0->isKinematicObject() && colObj0->getActivationState() != ISLAND_SLEEPING)
+			{
+				if (colObj0->hasContactResponse())
+					colObj1->activate();
+			}
+			if (colObj1->isKinematicObject() && colObj1->getActivationState() != ISLAND_SLEEPING)
+			{
+				if (colObj1->hasContactResponse())
+					colObj0->activate();
+			}
+			//filtering for response
+			if (dispatcher->needsResponse(colObj0, colObj1))
+			{
+				// scatter manifolds into various islands
+				int islandId = getIslandId(manifold);
+				// if island not sleeping,
+				if (Island* island = getIsland(islandId))
+				{
+					island->manifoldArray.push_back(manifold);
+				}
+			}
+		}
+	}
 }
 
-
-void btSimulationIslandManagerMt::addConstraintsToIslands( btAlignedObjectArray<btTypedConstraint*>& constraints )
+void btSimulationIslandManagerMt::addConstraintsToIslands(btAlignedObjectArray<btTypedConstraint*>& constraints)
 {
-    // walk constraints
-    for ( int i = 0; i < constraints.size(); i++ )
-    {
-        // scatter constraints into various islands
-        btTypedConstraint* constraint = constraints[ i ];
-        if ( constraint->isEnabled() )
-        {
-            int islandId = btGetConstraintIslandId( constraint );
-            // if island is not sleeping,
-            if ( Island* island = getIsland( islandId ) )
-            {
-                island->constraintArray.push_back( constraint );
-            }
-        }
-    }
+	// walk constraints
+	for (int i = 0; i < constraints.size(); i++)
+	{
+		// scatter constraints into various islands
+		btTypedConstraint* constraint = constraints[i];
+		if (constraint->isEnabled())
+		{
+			int islandId = btGetConstraintIslandId(constraint);
+			// if island is not sleeping,
+			if (Island* island = getIsland(islandId))
+			{
+				island->constraintArray.push_back(constraint);
+			}
+		}
+	}
 }
 
-
 void btSimulationIslandManagerMt::mergeIslands()
 {
-    // sort islands in order of decreasing batch size
-    m_activeIslands.quickSort( IslandBatchSizeSortPredicate() );
+	// sort islands in order of decreasing batch size
+	m_activeIslands.quickSort(IslandBatchSizeSortPredicate());
 
-    // merge small islands to satisfy minimum batch size
-    // find first small batch island
-    int destIslandIndex = m_activeIslands.size();
-    for ( int i = 0; i < m_activeIslands.size(); ++i )
-    {
-        Island* island = m_activeIslands[ i ];
-        int batchSize = calcBatchCost( island );
-        if ( batchSize < m_minimumSolverBatchSize )
-        {
-            destIslandIndex = i;
-            break;
-        }
-    }
-    int lastIndex = m_activeIslands.size() - 1;
-    while ( destIslandIndex < lastIndex )
-    {
-        // merge islands from the back of the list
-        Island* island = m_activeIslands[ destIslandIndex ];
-        int numBodies = island->bodyArray.size();
-        int numManifolds = island->manifoldArray.size();
-        int numConstraints = island->constraintArray.size();
-        int firstIndex = lastIndex;
-        // figure out how many islands we want to merge and find out how many bodies, manifolds and constraints we will have
-        while ( true )
-        {
-            Island* src = m_activeIslands[ firstIndex ];
-            numBodies += src->bodyArray.size();
-            numManifolds += src->manifoldArray.size();
-            numConstraints += src->constraintArray.size();
-            int batchCost = calcBatchCost( numBodies, numManifolds, numConstraints );
-            if ( batchCost >= m_minimumSolverBatchSize )
-            {
-                break;
-            }
-            if ( firstIndex - 1 == destIslandIndex )
-            {
-                break;
-            }
-            firstIndex--;
-        }
-        // reserve space for these pointers to minimize reallocation
-        island->bodyArray.reserve( numBodies );
-        island->manifoldArray.reserve( numManifolds );
-        island->constraintArray.reserve( numConstraints );
-        // merge islands
-        for ( int i = firstIndex; i <= lastIndex; ++i )
-        {
-            island->append( *m_activeIslands[ i ] );
-        }
-        // shrink array to exclude the islands that were merged from
-        m_activeIslands.resize( firstIndex );
-        lastIndex = firstIndex - 1;
-        destIslandIndex++;
-    }
+	// merge small islands to satisfy minimum batch size
+	// find first small batch island
+	int destIslandIndex = m_activeIslands.size();
+	for (int i = 0; i < m_activeIslands.size(); ++i)
+	{
+		Island* island = m_activeIslands[i];
+		int batchSize = calcBatchCost(island);
+		if (batchSize < m_minimumSolverBatchSize)
+		{
+			destIslandIndex = i;
+			break;
+		}
+	}
+	int lastIndex = m_activeIslands.size() - 1;
+	while (destIslandIndex < lastIndex)
+	{
+		// merge islands from the back of the list
+		Island* island = m_activeIslands[destIslandIndex];
+		int numBodies = island->bodyArray.size();
+		int numManifolds = island->manifoldArray.size();
+		int numConstraints = island->constraintArray.size();
+		int firstIndex = lastIndex;
+		// figure out how many islands we want to merge and find out how many bodies, manifolds and constraints we will have
+		while (true)
+		{
+			Island* src = m_activeIslands[firstIndex];
+			numBodies += src->bodyArray.size();
+			numManifolds += src->manifoldArray.size();
+			numConstraints += src->constraintArray.size();
+			int batchCost = calcBatchCost(numBodies, numManifolds, numConstraints);
+			if (batchCost >= m_minimumSolverBatchSize)
+			{
+				break;
+			}
+			if (firstIndex - 1 == destIslandIndex)
+			{
+				break;
+			}
+			firstIndex--;
+		}
+		// reserve space for these pointers to minimize reallocation
+		island->bodyArray.reserve(numBodies);
+		island->manifoldArray.reserve(numManifolds);
+		island->constraintArray.reserve(numConstraints);
+		// merge islands
+		for (int i = firstIndex; i <= lastIndex; ++i)
+		{
+			island->append(*m_activeIslands[i]);
+		}
+		// shrink array to exclude the islands that were merged from
+		m_activeIslands.resize(firstIndex);
+		lastIndex = firstIndex - 1;
+		destIslandIndex++;
+	}
 }
 
-
 void btSimulationIslandManagerMt::solveIsland(btConstraintSolver* solver, Island& island, const SolverParams& solverParams)
 {
-    btPersistentManifold** manifolds = island.manifoldArray.size() ? &island.manifoldArray[ 0 ] : NULL;
-    btTypedConstraint** constraintsPtr = island.constraintArray.size() ? &island.constraintArray[ 0 ] : NULL;
-    solver->solveGroup( &island.bodyArray[ 0 ],
-        island.bodyArray.size(),
-        manifolds,
-        island.manifoldArray.size(),
-        constraintsPtr,
-        island.constraintArray.size(),
-        *solverParams.m_solverInfo,
-        solverParams.m_debugDrawer,
-        solverParams.m_dispatcher
-    );
+	btPersistentManifold** manifolds = island.manifoldArray.size() ? &island.manifoldArray[0] : NULL;
+	btTypedConstraint** constraintsPtr = island.constraintArray.size() ? &island.constraintArray[0] : NULL;
+	solver->solveGroup(&island.bodyArray[0],
+					   island.bodyArray.size(),
+					   manifolds,
+					   island.manifoldArray.size(),
+					   constraintsPtr,
+					   island.constraintArray.size(),
+					   *solverParams.m_solverInfo,
+					   solverParams.m_debugDrawer,
+					   solverParams.m_dispatcher);
 }
 
-
-void btSimulationIslandManagerMt::serialIslandDispatch( btAlignedObjectArray<Island*>* islandsPtr, const SolverParams& solverParams )
+void btSimulationIslandManagerMt::serialIslandDispatch(btAlignedObjectArray<Island*>* islandsPtr, const SolverParams& solverParams)
 {
-    BT_PROFILE( "serialIslandDispatch" );
-    // serial dispatch
-    btAlignedObjectArray<Island*>& islands = *islandsPtr;
-    btConstraintSolver* solver = solverParams.m_solverMt ? solverParams.m_solverMt : solverParams.m_solverPool;
-    for ( int i = 0; i < islands.size(); ++i )
-    {
-        solveIsland(solver, *islands[ i ], solverParams);
-    }
+	BT_PROFILE("serialIslandDispatch");
+	// serial dispatch
+	btAlignedObjectArray<Island*>& islands = *islandsPtr;
+	btConstraintSolver* solver = solverParams.m_solverMt ? solverParams.m_solverMt : solverParams.m_solverPool;
+	for (int i = 0; i < islands.size(); ++i)
+	{
+		solveIsland(solver, *islands[i], solverParams);
+	}
 }
 
-
 struct UpdateIslandDispatcher : public btIParallelForBody
 {
-    btAlignedObjectArray<btSimulationIslandManagerMt::Island*>& m_islandsPtr;
-    const btSimulationIslandManagerMt::SolverParams& m_solverParams;
+	btAlignedObjectArray<btSimulationIslandManagerMt::Island*>& m_islandsPtr;
+	const btSimulationIslandManagerMt::SolverParams& m_solverParams;
 
-    UpdateIslandDispatcher(btAlignedObjectArray<btSimulationIslandManagerMt::Island*>& islandsPtr, const btSimulationIslandManagerMt::SolverParams& solverParams)
-        : m_islandsPtr(islandsPtr), m_solverParams(solverParams)
-    {}
+	UpdateIslandDispatcher(btAlignedObjectArray<btSimulationIslandManagerMt::Island*>& islandsPtr, const btSimulationIslandManagerMt::SolverParams& solverParams)
+		: m_islandsPtr(islandsPtr), m_solverParams(solverParams)
+	{
+	}
 
-    void forLoop( int iBegin, int iEnd ) const BT_OVERRIDE
-    {
-        btConstraintSolver* solver = m_solverParams.m_solverPool;
-        for ( int i = iBegin; i < iEnd; ++i )
-        {
-            btSimulationIslandManagerMt::Island* island = m_islandsPtr[ i ];
-            btSimulationIslandManagerMt::solveIsland( solver, *island, m_solverParams );
-        }
-    }
+	void forLoop(int iBegin, int iEnd) const BT_OVERRIDE
+	{
+		btConstraintSolver* solver = m_solverParams.m_solverPool;
+		for (int i = iBegin; i < iEnd; ++i)
+		{
+			btSimulationIslandManagerMt::Island* island = m_islandsPtr[i];
+			btSimulationIslandManagerMt::solveIsland(solver, *island, m_solverParams);
+		}
+	}
 };
 
-
-void btSimulationIslandManagerMt::parallelIslandDispatch( btAlignedObjectArray<Island*>* islandsPtr, const SolverParams& solverParams )
+void btSimulationIslandManagerMt::parallelIslandDispatch(btAlignedObjectArray<Island*>* islandsPtr, const SolverParams& solverParams)
 {
-    BT_PROFILE( "parallelIslandDispatch" );
-    //
-    // if there are islands with many contacts, it may be faster to submit these
-    // large islands *serially* to a single parallel constraint solver, and then later
-    // submit the remaining smaller islands in parallel to multiple sequential solvers.
-    //
-    // Some task schedulers do not deal well with nested parallelFor loops. One implementation
-    // of OpenMP was actually slower than doing everything single-threaded. Intel TBB
-    // on the other hand, seems to do a pretty respectable job with it.
-    //
-    // When solving islands in parallel, the worst case performance happens when there
-    // is one very large island and then perhaps a smattering of very small
-    // islands -- one worker thread takes the large island and the remaining workers
-    // tear through the smaller islands and then sit idle waiting for the first worker
-    // to finish. Solving islands in parallel works best when there are numerous small
-    // islands, roughly equal in size.
-    //
-    // By contrast, the other approach -- the parallel constraint solver -- is only
-    // able to deliver a worthwhile speedup when the island is large. For smaller islands,
-    // it is difficult to extract a useful amount of parallelism -- the overhead of grouping
-    // the constraints into batches and sending the batches to worker threads can nullify
-    // any gains from parallelism.
-    //
+	BT_PROFILE("parallelIslandDispatch");
+	//
+	// if there are islands with many contacts, it may be faster to submit these
+	// large islands *serially* to a single parallel constraint solver, and then later
+	// submit the remaining smaller islands in parallel to multiple sequential solvers.
+	//
+	// Some task schedulers do not deal well with nested parallelFor loops. One implementation
+	// of OpenMP was actually slower than doing everything single-threaded. Intel TBB
+	// on the other hand, seems to do a pretty respectable job with it.
+	//
+	// When solving islands in parallel, the worst case performance happens when there
+	// is one very large island and then perhaps a smattering of very small
+	// islands -- one worker thread takes the large island and the remaining workers
+	// tear through the smaller islands and then sit idle waiting for the first worker
+	// to finish. Solving islands in parallel works best when there are numerous small
+	// islands, roughly equal in size.
+	//
+	// By contrast, the other approach -- the parallel constraint solver -- is only
+	// able to deliver a worthwhile speedup when the island is large. For smaller islands,
+	// it is difficult to extract a useful amount of parallelism -- the overhead of grouping
+	// the constraints into batches and sending the batches to worker threads can nullify
+	// any gains from parallelism.
+	//
 
-    UpdateIslandDispatcher dispatcher(*islandsPtr, solverParams);
-    // We take advantage of the fact the islands are sorted in order of decreasing size
-    int iBegin = 0;
-    if (solverParams.m_solverMt)
-    {
-        while ( iBegin < islandsPtr->size() )
-        {
-            btSimulationIslandManagerMt::Island* island = ( *islandsPtr )[ iBegin ];
-            if ( island->manifoldArray.size() < btSequentialImpulseConstraintSolverMt::s_minimumContactManifoldsForBatching )
-            {
-                // OK to submit the rest of the array in parallel
-                break;
-            }
-            // serial dispatch to parallel solver for large islands (if any)
-            solveIsland(solverParams.m_solverMt, *island, solverParams);
-            ++iBegin;
-        }
-    }
-    // parallel dispatch to sequential solvers for rest
-    btParallelFor( iBegin, islandsPtr->size(), 1, dispatcher );
+	UpdateIslandDispatcher dispatcher(*islandsPtr, solverParams);
+	// We take advantage of the fact the islands are sorted in order of decreasing size
+	int iBegin = 0;
+	if (solverParams.m_solverMt)
+	{
+		while (iBegin < islandsPtr->size())
+		{
+			btSimulationIslandManagerMt::Island* island = (*islandsPtr)[iBegin];
+			if (island->manifoldArray.size() < btSequentialImpulseConstraintSolverMt::s_minimumContactManifoldsForBatching)
+			{
+				// OK to submit the rest of the array in parallel
+				break;
+			}
+			// serial dispatch to parallel solver for large islands (if any)
+			solveIsland(solverParams.m_solverMt, *island, solverParams);
+			++iBegin;
+		}
+	}
+	// parallel dispatch to sequential solvers for rest
+	btParallelFor(iBegin, islandsPtr->size(), 1, dispatcher);
 }
 
-
 ///@todo: this is random access, it can be walked 'cache friendly'!
-void btSimulationIslandManagerMt::buildAndProcessIslands( btDispatcher* dispatcher,
-                                                        btCollisionWorld* collisionWorld,
-                                                        btAlignedObjectArray<btTypedConstraint*>& constraints,
-                                                        const SolverParams& solverParams
-                                                        )
+void btSimulationIslandManagerMt::buildAndProcessIslands(btDispatcher* dispatcher,
+														 btCollisionWorld* collisionWorld,
+														 btAlignedObjectArray<btTypedConstraint*>& constraints,
+														 const SolverParams& solverParams)
 {
 	BT_PROFILE("buildAndProcessIslands");
 	btCollisionObjectArray& collisionObjects = collisionWorld->getCollisionObjectArray();
 
-	buildIslands(dispatcher,collisionWorld);
+	buildIslands(dispatcher, collisionWorld);
 
-	if(!getSplitIslands())
+	if (!getSplitIslands())
 	{
-        btPersistentManifold** manifolds = dispatcher->getInternalManifoldPointer();
-        int maxNumManifolds = dispatcher->getNumManifolds();
+		btPersistentManifold** manifolds = dispatcher->getInternalManifoldPointer();
+		int maxNumManifolds = dispatcher->getNumManifolds();
 
-        for ( int i = 0; i < maxNumManifolds; i++ )
-        {
-            btPersistentManifold* manifold = manifolds[ i ];
+		for (int i = 0; i < maxNumManifolds; i++)
+		{
+			btPersistentManifold* manifold = manifolds[i];
 
-            const btCollisionObject* colObj0 = static_cast<const btCollisionObject*>( manifold->getBody0() );
-            const btCollisionObject* colObj1 = static_cast<const btCollisionObject*>( manifold->getBody1() );
+			const btCollisionObject* colObj0 = static_cast<const btCollisionObject*>(manifold->getBody0());
+			const btCollisionObject* colObj1 = static_cast<const btCollisionObject*>(manifold->getBody1());
 
-            ///@todo: check sleeping conditions!
-            if ( ( ( colObj0 ) && colObj0->getActivationState() != ISLAND_SLEEPING ) ||
-                 ( ( colObj1 ) && colObj1->getActivationState() != ISLAND_SLEEPING ) )
-            {
-
-                //kinematic objects don't merge islands, but wake up all connected objects
-                if ( colObj0->isKinematicObject() && colObj0->getActivationState() != ISLAND_SLEEPING )
-                {
-                    if ( colObj0->hasContactResponse() )
-                        colObj1->activate();
-                }
-                if ( colObj1->isKinematicObject() && colObj1->getActivationState() != ISLAND_SLEEPING )
-                {
-                    if ( colObj1->hasContactResponse() )
-                        colObj0->activate();
-                }
-            }
-        }
-        btTypedConstraint** constraintsPtr = constraints.size() ? &constraints[ 0 ] : NULL;
-        btConstraintSolver* solver = solverParams.m_solverMt ? solverParams.m_solverMt : solverParams.m_solverPool;
-        solver->solveGroup(&collisionObjects[0],
-                           collisionObjects.size(),
-                           manifolds,
-                           maxNumManifolds,
-                           constraintsPtr,
-                           constraints.size(),
-                           *solverParams.m_solverInfo,
-                           solverParams.m_debugDrawer,
-                           solverParams.m_dispatcher
-                           );
+			///@todo: check sleeping conditions!
+			if (((colObj0) && colObj0->getActivationState() != ISLAND_SLEEPING) ||
+				((colObj1) && colObj1->getActivationState() != ISLAND_SLEEPING))
+			{
+				//kinematic objects don't merge islands, but wake up all connected objects
+				if (colObj0->isKinematicObject() && colObj0->getActivationState() != ISLAND_SLEEPING)
+				{
+					if (colObj0->hasContactResponse())
+						colObj1->activate();
+				}
+				if (colObj1->isKinematicObject() && colObj1->getActivationState() != ISLAND_SLEEPING)
+				{
+					if (colObj1->hasContactResponse())
+						colObj0->activate();
+				}
+			}
+		}
+		btTypedConstraint** constraintsPtr = constraints.size() ? &constraints[0] : NULL;
+		btConstraintSolver* solver = solverParams.m_solverMt ? solverParams.m_solverMt : solverParams.m_solverPool;
+		solver->solveGroup(&collisionObjects[0],
+						   collisionObjects.size(),
+						   manifolds,
+						   maxNumManifolds,
+						   constraintsPtr,
+						   constraints.size(),
+						   *solverParams.m_solverInfo,
+						   solverParams.m_debugDrawer,
+						   solverParams.m_dispatcher);
 	}
 	else
 	{
-        initIslandPools();
+		initIslandPools();
 
-        //traverse the simulation islands, and call the solver, unless all objects are sleeping/deactivated
-        addBodiesToIslands( collisionWorld );
-        addManifoldsToIslands( dispatcher );
-        addConstraintsToIslands( constraints );
+		//traverse the simulation islands, and call the solver, unless all objects are sleeping/deactivated
+		addBodiesToIslands(collisionWorld);
+		addManifoldsToIslands(dispatcher);
+		addConstraintsToIslands(constraints);
 
-        // m_activeIslands array should now contain all non-sleeping Islands, and each Island should
-        // have all the necessary bodies, manifolds and constraints.
+		// m_activeIslands array should now contain all non-sleeping Islands, and each Island should
+		// have all the necessary bodies, manifolds and constraints.
 
-        // if we want to merge islands with small batch counts,
-        if ( m_minimumSolverBatchSize > 1 )
-        {
-            mergeIslands();
-        }
-        // dispatch islands to solver
-        m_islandDispatch( &m_activeIslands, solverParams );
+		// if we want to merge islands with small batch counts,
+		if (m_minimumSolverBatchSize > 1)
+		{
+			mergeIslands();
+		}
+		// dispatch islands to solver
+		m_islandDispatch(&m_activeIslands, solverParams);
 	}
 }
diff --git a/thirdparty/bullet/BulletDynamics/Dynamics/btSimulationIslandManagerMt.h b/thirdparty/bullet/BulletDynamics/Dynamics/btSimulationIslandManagerMt.h
index 563577a6f43..ab73a899f14 100644
--- a/thirdparty/bullet/BulletDynamics/Dynamics/btSimulationIslandManagerMt.h
+++ b/thirdparty/bullet/BulletDynamics/Dynamics/btSimulationIslandManagerMt.h
@@ -35,80 +35,78 @@ class btIDebugDraw;
 class btSimulationIslandManagerMt : public btSimulationIslandManager
 {
 public:
-    struct Island
-    {
-        // a simulation island consisting of bodies, manifolds and constraints,
-        // to be passed into a constraint solver.
-        btAlignedObjectArray<btCollisionObject*> bodyArray;
-        btAlignedObjectArray<btPersistentManifold*> manifoldArray;
-        btAlignedObjectArray<btTypedConstraint*> constraintArray;
-        int id;  // island id
-        bool isSleeping;
+	struct Island
+	{
+		// a simulation island consisting of bodies, manifolds and constraints,
+		// to be passed into a constraint solver.
+		btAlignedObjectArray<btCollisionObject*> bodyArray;
+		btAlignedObjectArray<btPersistentManifold*> manifoldArray;
+		btAlignedObjectArray<btTypedConstraint*> constraintArray;
+		int id;  // island id
+		bool isSleeping;
 
-        void append( const Island& other );  // add bodies, manifolds, constraints to my own
-    };
-    struct SolverParams
-    {
-        btConstraintSolver*		m_solverPool;
-        btConstraintSolver*		m_solverMt;
-        btContactSolverInfo*	m_solverInfo;
-        btIDebugDraw*			m_debugDrawer;
-        btDispatcher*			m_dispatcher;
-    };
-    static void solveIsland(btConstraintSolver* solver, Island& island, const SolverParams& solverParams);
+		void append(const Island& other);  // add bodies, manifolds, constraints to my own
+	};
+	struct SolverParams
+	{
+		btConstraintSolver* m_solverPool;
+		btConstraintSolver* m_solverMt;
+		btContactSolverInfo* m_solverInfo;
+		btIDebugDraw* m_debugDrawer;
+		btDispatcher* m_dispatcher;
+	};
+	static void solveIsland(btConstraintSolver* solver, Island& island, const SolverParams& solverParams);
+
+	typedef void (*IslandDispatchFunc)(btAlignedObjectArray<Island*>* islands, const SolverParams& solverParams);
+	static void serialIslandDispatch(btAlignedObjectArray<Island*>* islandsPtr, const SolverParams& solverParams);
+	static void parallelIslandDispatch(btAlignedObjectArray<Island*>* islandsPtr, const SolverParams& solverParams);
 
-    typedef void( *IslandDispatchFunc ) ( btAlignedObjectArray<Island*>* islands, const SolverParams& solverParams );
-    static void serialIslandDispatch( btAlignedObjectArray<Island*>* islandsPtr, const SolverParams& solverParams );
-    static void parallelIslandDispatch( btAlignedObjectArray<Island*>* islandsPtr, const SolverParams& solverParams );
 protected:
-    btAlignedObjectArray<Island*> m_allocatedIslands;  // owner of all Islands
-    btAlignedObjectArray<Island*> m_activeIslands;  // islands actively in use
-    btAlignedObjectArray<Island*> m_freeIslands;  // islands ready to be reused
-    btAlignedObjectArray<Island*> m_lookupIslandFromId;  // big lookup table to map islandId to Island pointer
-    Island* m_batchIsland;
-    int m_minimumSolverBatchSize;
-    int m_batchIslandMinBodyCount;
-    IslandDispatchFunc m_islandDispatch;
+	btAlignedObjectArray<Island*> m_allocatedIslands;    // owner of all Islands
+	btAlignedObjectArray<Island*> m_activeIslands;       // islands actively in use
+	btAlignedObjectArray<Island*> m_freeIslands;         // islands ready to be reused
+	btAlignedObjectArray<Island*> m_lookupIslandFromId;  // big lookup table to map islandId to Island pointer
+	Island* m_batchIsland;
+	int m_minimumSolverBatchSize;
+	int m_batchIslandMinBodyCount;
+	IslandDispatchFunc m_islandDispatch;
+
+	Island* getIsland(int id);
+	virtual Island* allocateIsland(int id, int numBodies);
+	virtual void initIslandPools();
+	virtual void addBodiesToIslands(btCollisionWorld* collisionWorld);
+	virtual void addManifoldsToIslands(btDispatcher* dispatcher);
+	virtual void addConstraintsToIslands(btAlignedObjectArray<btTypedConstraint*>& constraints);
+	virtual void mergeIslands();
 
-    Island* getIsland( int id );
-    virtual Island* allocateIsland( int id, int numBodies );
-    virtual void initIslandPools();
-    virtual void addBodiesToIslands( btCollisionWorld* collisionWorld );
-    virtual void addManifoldsToIslands( btDispatcher* dispatcher );
-    virtual void addConstraintsToIslands( btAlignedObjectArray<btTypedConstraint*>& constraints );
-    virtual void mergeIslands();
-	
 public:
 	btSimulationIslandManagerMt();
 	virtual ~btSimulationIslandManagerMt();
 
-    virtual void buildAndProcessIslands( btDispatcher* dispatcher,
-        btCollisionWorld* collisionWorld,
-        btAlignedObjectArray<btTypedConstraint*>& constraints,
-        const SolverParams& solverParams
-    );
+	virtual void buildAndProcessIslands(btDispatcher* dispatcher,
+										btCollisionWorld* collisionWorld,
+										btAlignedObjectArray<btTypedConstraint*>& constraints,
+										const SolverParams& solverParams);
 
-	virtual void buildIslands(btDispatcher* dispatcher,btCollisionWorld* colWorld);
+	virtual void buildIslands(btDispatcher* dispatcher, btCollisionWorld* colWorld);
 
-    int getMinimumSolverBatchSize() const
-    {
-        return m_minimumSolverBatchSize;
-    }
-    void setMinimumSolverBatchSize( int sz )
-    {
-        m_minimumSolverBatchSize = sz;
-    }
-    IslandDispatchFunc getIslandDispatchFunction() const
-    {
-        return m_islandDispatch;
-    }
-    // allow users to set their own dispatch function for multithreaded dispatch
-    void setIslandDispatchFunction( IslandDispatchFunc func )
-    {
-        m_islandDispatch = func;
-    }
+	int getMinimumSolverBatchSize() const
+	{
+		return m_minimumSolverBatchSize;
+	}
+	void setMinimumSolverBatchSize(int sz)
+	{
+		m_minimumSolverBatchSize = sz;
+	}
+	IslandDispatchFunc getIslandDispatchFunction() const
+	{
+		return m_islandDispatch;
+	}
+	// allow users to set their own dispatch function for multithreaded dispatch
+	void setIslandDispatchFunction(IslandDispatchFunc func)
+	{
+		m_islandDispatch = func;
+	}
 };
 
-
-#endif //BT_SIMULATION_ISLAND_MANAGER_H
-
+#endif  //BT_SIMULATION_ISLAND_MANAGER_H
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBody.cpp b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBody.cpp
index 0e85b55728f..53fc48d4b95 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBody.cpp
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBody.cpp
@@ -21,7 +21,6 @@
  
  */
 
-
 #include "btMultiBody.h"
 #include "btMultiBodyLink.h"
 #include "btMultiBodyLinkCollider.h"
@@ -29,28 +28,29 @@
 #include "LinearMath/btTransformUtil.h"
 #include "LinearMath/btSerializer.h"
 //#include "Bullet3Common/b3Logging.h"
-// #define INCLUDE_GYRO_TERM 
+// #define INCLUDE_GYRO_TERM
 
-///todo: determine if we need these options. If so, make a proper API, otherwise delete those globals
-bool gJointFeedbackInWorldSpace = false;
-bool gJointFeedbackInJointFrame = false;
 
-namespace {
-    const btScalar SLEEP_EPSILON = btScalar(0.05);  // this is a squared velocity (m^2 s^-2)
-    const btScalar SLEEP_TIMEOUT = btScalar(2);     // in seconds
+namespace
+{
+const btScalar SLEEP_EPSILON = btScalar(0.05);  // this is a squared velocity (m^2 s^-2)
+const btScalar SLEEP_TIMEOUT = btScalar(2);     // in seconds
+}  // namespace
+
+void btMultiBody::spatialTransform(const btMatrix3x3 &rotation_matrix,  // rotates vectors in 'from' frame to vectors in 'to' frame
+	const btVector3 &displacement,       // vector from origin of 'from' frame to origin of 'to' frame, in 'to' coordinates
+	const btVector3 &top_in,             // top part of input vector
+	const btVector3 &bottom_in,          // bottom part of input vector
+	btVector3 &top_out,                  // top part of output vector
+	btVector3 &bottom_out)               // bottom part of output vector
+{
+	top_out = rotation_matrix * top_in;
+	bottom_out = -displacement.cross(top_out) + rotation_matrix * bottom_in;
 }
 
-namespace {
-    void SpatialTransform(const btMatrix3x3 &rotation_matrix,  // rotates vectors in 'from' frame to vectors in 'to' frame
-                          const btVector3 &displacement,     // vector from origin of 'from' frame to origin of 'to' frame, in 'to' coordinates
-                          const btVector3 &top_in,       // top part of input vector
-                          const btVector3 &bottom_in,    // bottom part of input vector
-                          btVector3 &top_out,         // top part of output vector
-                          btVector3 &bottom_out)      // bottom part of output vector
-    {
-        top_out = rotation_matrix * top_in;
-        bottom_out = -displacement.cross(top_out) + rotation_matrix * bottom_in;
-    }
+namespace
+{
+
 
 #if 0
     void InverseSpatialTransform(const btMatrix3x3 &rotation_matrix,
@@ -83,60 +83,58 @@ namespace {
 		bottom_out = a_bottom.cross(b_top) + a_top.cross(b_bottom);
 	}
 #endif
-	
-}
 
+}  // namespace
 
 //
 // Implementation of class btMultiBody
 //
 
 btMultiBody::btMultiBody(int n_links,
-                     btScalar mass,
-                     const btVector3 &inertia,
-                     bool fixedBase,
-                     bool canSleep,
-		     bool /*deprecatedUseMultiDof*/)
-    : 
-    	m_baseCollider(0),
-		m_baseName(0),
-    	m_basePos(0,0,0),
-    	m_baseQuat(0, 0, 0, 1),
-      m_baseMass(mass),
-      m_baseInertia(inertia),
-    
-		m_fixedBase(fixedBase),
-		m_awake(true),
-		m_canSleep(canSleep),
-		m_sleepTimer(0),
-		m_userObjectPointer(0),
-		m_userIndex2(-1),
-		m_userIndex(-1),
-		m_companionId(-1),
-		m_linearDamping(0.04f),
-		m_angularDamping(0.04f),
-		m_useGyroTerm(true),
-		m_maxAppliedImpulse(1000.f),
-		m_maxCoordinateVelocity(100.f),
-		m_hasSelfCollision(true),		
-		__posUpdated(false),
-		m_dofCount(0),
-		m_posVarCnt(0),
-		m_useRK4(false), 	
-		m_useGlobalVelocities(false),
-		m_internalNeedsJointFeedback(false)
+						 btScalar mass,
+						 const btVector3 &inertia,
+						 bool fixedBase,
+						 bool canSleep,
+						 bool /*deprecatedUseMultiDof*/)
+	: m_baseCollider(0),
+	  m_baseName(0),
+	  m_basePos(0, 0, 0),
+	  m_baseQuat(0, 0, 0, 1),
+	  m_baseMass(mass),
+	  m_baseInertia(inertia),
+
+	  m_fixedBase(fixedBase),
+	  m_awake(true),
+	  m_canSleep(canSleep),
+	  m_sleepTimer(0),
+	  m_userObjectPointer(0),
+	  m_userIndex2(-1),
+	  m_userIndex(-1),
+	  m_companionId(-1),
+	  m_linearDamping(0.04f),
+	  m_angularDamping(0.04f),
+	  m_useGyroTerm(true),
+	  m_maxAppliedImpulse(1000.f),
+	  m_maxCoordinateVelocity(100.f),
+	  m_hasSelfCollision(true),
+	  __posUpdated(false),
+	  m_dofCount(0),
+	  m_posVarCnt(0),
+	  m_useRK4(false),
+	  m_useGlobalVelocities(false),
+	  m_internalNeedsJointFeedback(false)
 {
-	m_cachedInertiaTopLeft.setValue(0,0,0,0,0,0,0,0,0);
-	m_cachedInertiaTopRight.setValue(0,0,0,0,0,0,0,0,0);
-	m_cachedInertiaLowerLeft.setValue(0,0,0,0,0,0,0,0,0);
-	m_cachedInertiaLowerRight.setValue(0,0,0,0,0,0,0,0,0);
-	m_cachedInertiaValid=false;
+	m_cachedInertiaTopLeft.setValue(0, 0, 0, 0, 0, 0, 0, 0, 0);
+	m_cachedInertiaTopRight.setValue(0, 0, 0, 0, 0, 0, 0, 0, 0);
+	m_cachedInertiaLowerLeft.setValue(0, 0, 0, 0, 0, 0, 0, 0, 0);
+	m_cachedInertiaLowerRight.setValue(0, 0, 0, 0, 0, 0, 0, 0, 0);
+	m_cachedInertiaValid = false;
 
 	m_links.resize(n_links);
 	m_matrixBuf.resize(n_links + 1);
 
-    m_baseForce.setValue(0, 0, 0);
-    m_baseTorque.setValue(0, 0, 0);
+	m_baseForce.setValue(0, 0, 0);
+	m_baseTorque.setValue(0, 0, 0);
 
 	clearConstraintForces();
 	clearForcesAndTorques();
@@ -147,131 +145,125 @@ btMultiBody::~btMultiBody()
 }
 
 void btMultiBody::setupFixed(int i,
-						   btScalar mass,
-						   const btVector3 &inertia,
-						   int parent,
-						   const btQuaternion &rotParentToThis,
-						   const btVector3 &parentComToThisPivotOffset,
-                           const btVector3 &thisPivotToThisComOffset, bool /*deprecatedDisableParentCollision*/)
+							 btScalar mass,
+							 const btVector3 &inertia,
+							 int parent,
+							 const btQuaternion &rotParentToThis,
+							 const btVector3 &parentComToThisPivotOffset,
+							 const btVector3 &thisPivotToThisComOffset, bool /*deprecatedDisableParentCollision*/)
 {
-	
 	m_links[i].m_mass = mass;
-    m_links[i].m_inertiaLocal = inertia;
-    m_links[i].m_parent = parent;
-	 m_links[i].setAxisTop(0, 0., 0., 0.);
-    m_links[i].setAxisBottom(0, btVector3(0,0,0));
-    m_links[i].m_zeroRotParentToThis = rotParentToThis;
+	m_links[i].m_inertiaLocal = inertia;
+	m_links[i].m_parent = parent;
+	m_links[i].setAxisTop(0, 0., 0., 0.);
+	m_links[i].setAxisBottom(0, btVector3(0, 0, 0));
+	m_links[i].m_zeroRotParentToThis = rotParentToThis;
 	m_links[i].m_dVector = thisPivotToThisComOffset;
-    m_links[i].m_eVector = parentComToThisPivotOffset;    
+	m_links[i].m_eVector = parentComToThisPivotOffset;
 
 	m_links[i].m_jointType = btMultibodyLink::eFixed;
 	m_links[i].m_dofCount = 0;
 	m_links[i].m_posVarCount = 0;
 
-	m_links[i].m_flags |=BT_MULTIBODYLINKFLAGS_DISABLE_PARENT_COLLISION;    
-	
+	m_links[i].m_flags |= BT_MULTIBODYLINKFLAGS_DISABLE_PARENT_COLLISION;
+
 	m_links[i].updateCacheMultiDof();
 
 	updateLinksDofOffsets();
-
 }
 
-
 void btMultiBody::setupPrismatic(int i,
-                               btScalar mass,
-                               const btVector3 &inertia,
-                               int parent,
-                               const btQuaternion &rotParentToThis,
-                               const btVector3 &jointAxis,
-                               const btVector3 &parentComToThisPivotOffset,
-							   const btVector3 &thisPivotToThisComOffset,
-							   bool disableParentCollision)
+								 btScalar mass,
+								 const btVector3 &inertia,
+								 int parent,
+								 const btQuaternion &rotParentToThis,
+								 const btVector3 &jointAxis,
+								 const btVector3 &parentComToThisPivotOffset,
+								 const btVector3 &thisPivotToThisComOffset,
+								 bool disableParentCollision)
 {
 	m_dofCount += 1;
 	m_posVarCnt += 1;
-	
-    m_links[i].m_mass = mass;
-    m_links[i].m_inertiaLocal = inertia;
-    m_links[i].m_parent = parent;
-    m_links[i].m_zeroRotParentToThis = rotParentToThis;
-    m_links[i].setAxisTop(0, 0., 0., 0.);
-    m_links[i].setAxisBottom(0, jointAxis);
-    m_links[i].m_eVector = parentComToThisPivotOffset;
+
+	m_links[i].m_mass = mass;
+	m_links[i].m_inertiaLocal = inertia;
+	m_links[i].m_parent = parent;
+	m_links[i].m_zeroRotParentToThis = rotParentToThis;
+	m_links[i].setAxisTop(0, 0., 0., 0.);
+	m_links[i].setAxisBottom(0, jointAxis);
+	m_links[i].m_eVector = parentComToThisPivotOffset;
 	m_links[i].m_dVector = thisPivotToThisComOffset;
-    m_links[i].m_cachedRotParentToThis = rotParentToThis;
+	m_links[i].m_cachedRotParentToThis = rotParentToThis;
 
 	m_links[i].m_jointType = btMultibodyLink::ePrismatic;
 	m_links[i].m_dofCount = 1;
-	m_links[i].m_posVarCount = 1;	
+	m_links[i].m_posVarCount = 1;
 	m_links[i].m_jointPos[0] = 0.f;
 	m_links[i].m_jointTorque[0] = 0.f;
 
 	if (disableParentCollision)
-		m_links[i].m_flags |=BT_MULTIBODYLINKFLAGS_DISABLE_PARENT_COLLISION;
+		m_links[i].m_flags |= BT_MULTIBODYLINKFLAGS_DISABLE_PARENT_COLLISION;
 	//
-	
+
 	m_links[i].updateCacheMultiDof();
-	
+
 	updateLinksDofOffsets();
 }
 
 void btMultiBody::setupRevolute(int i,
-                              btScalar mass,
-                              const btVector3 &inertia,
-                              int parent,
-                              const btQuaternion &rotParentToThis,
-                              const btVector3 &jointAxis,
-                              const btVector3 &parentComToThisPivotOffset,
-                              const btVector3 &thisPivotToThisComOffset,
-							  bool disableParentCollision)
+								btScalar mass,
+								const btVector3 &inertia,
+								int parent,
+								const btQuaternion &rotParentToThis,
+								const btVector3 &jointAxis,
+								const btVector3 &parentComToThisPivotOffset,
+								const btVector3 &thisPivotToThisComOffset,
+								bool disableParentCollision)
 {
 	m_dofCount += 1;
 	m_posVarCnt += 1;
-	
-    m_links[i].m_mass = mass;
-    m_links[i].m_inertiaLocal = inertia;
-    m_links[i].m_parent = parent;
-    m_links[i].m_zeroRotParentToThis = rotParentToThis;
-    m_links[i].setAxisTop(0, jointAxis);
-    m_links[i].setAxisBottom(0, jointAxis.cross(thisPivotToThisComOffset));
-    m_links[i].m_dVector = thisPivotToThisComOffset;
-    m_links[i].m_eVector = parentComToThisPivotOffset;
+
+	m_links[i].m_mass = mass;
+	m_links[i].m_inertiaLocal = inertia;
+	m_links[i].m_parent = parent;
+	m_links[i].m_zeroRotParentToThis = rotParentToThis;
+	m_links[i].setAxisTop(0, jointAxis);
+	m_links[i].setAxisBottom(0, jointAxis.cross(thisPivotToThisComOffset));
+	m_links[i].m_dVector = thisPivotToThisComOffset;
+	m_links[i].m_eVector = parentComToThisPivotOffset;
 
 	m_links[i].m_jointType = btMultibodyLink::eRevolute;
 	m_links[i].m_dofCount = 1;
-	m_links[i].m_posVarCount = 1;	
+	m_links[i].m_posVarCount = 1;
 	m_links[i].m_jointPos[0] = 0.f;
 	m_links[i].m_jointTorque[0] = 0.f;
 
 	if (disableParentCollision)
-		m_links[i].m_flags |=BT_MULTIBODYLINKFLAGS_DISABLE_PARENT_COLLISION;
-    //
+		m_links[i].m_flags |= BT_MULTIBODYLINKFLAGS_DISABLE_PARENT_COLLISION;
+	//
 	m_links[i].updateCacheMultiDof();
 	//
 	updateLinksDofOffsets();
 }
 
-
-
 void btMultiBody::setupSpherical(int i,
-						   btScalar mass,
-						   const btVector3 &inertia,
-						   int parent,
-						   const btQuaternion &rotParentToThis,
-						   const btVector3 &parentComToThisPivotOffset,
-						   const btVector3 &thisPivotToThisComOffset,
-						   bool disableParentCollision)
+								 btScalar mass,
+								 const btVector3 &inertia,
+								 int parent,
+								 const btQuaternion &rotParentToThis,
+								 const btVector3 &parentComToThisPivotOffset,
+								 const btVector3 &thisPivotToThisComOffset,
+								 bool disableParentCollision)
 {
-	
 	m_dofCount += 3;
 	m_posVarCnt += 4;
 
 	m_links[i].m_mass = mass;
-    m_links[i].m_inertiaLocal = inertia;
-    m_links[i].m_parent = parent;
-    m_links[i].m_zeroRotParentToThis = rotParentToThis;    
-    m_links[i].m_dVector = thisPivotToThisComOffset;
-    m_links[i].m_eVector = parentComToThisPivotOffset;    
+	m_links[i].m_inertiaLocal = inertia;
+	m_links[i].m_parent = parent;
+	m_links[i].m_zeroRotParentToThis = rotParentToThis;
+	m_links[i].m_dVector = thisPivotToThisComOffset;
+	m_links[i].m_eVector = parentComToThisPivotOffset;
 
 	m_links[i].m_jointType = btMultibodyLink::eSpherical;
 	m_links[i].m_dofCount = 3;
@@ -282,281 +274,297 @@ void btMultiBody::setupSpherical(int i,
 	m_links[i].setAxisBottom(0, m_links[i].getAxisTop(0).cross(thisPivotToThisComOffset));
 	m_links[i].setAxisBottom(1, m_links[i].getAxisTop(1).cross(thisPivotToThisComOffset));
 	m_links[i].setAxisBottom(2, m_links[i].getAxisTop(2).cross(thisPivotToThisComOffset));
-	m_links[i].m_jointPos[0] = m_links[i].m_jointPos[1] = m_links[i].m_jointPos[2] = 0.f; m_links[i].m_jointPos[3] = 1.f;
+	m_links[i].m_jointPos[0] = m_links[i].m_jointPos[1] = m_links[i].m_jointPos[2] = 0.f;
+	m_links[i].m_jointPos[3] = 1.f;
 	m_links[i].m_jointTorque[0] = m_links[i].m_jointTorque[1] = m_links[i].m_jointTorque[2] = 0.f;
 
-
 	if (disableParentCollision)
-		m_links[i].m_flags |=BT_MULTIBODYLINKFLAGS_DISABLE_PARENT_COLLISION;    
+		m_links[i].m_flags |= BT_MULTIBODYLINKFLAGS_DISABLE_PARENT_COLLISION;
 	//
-	m_links[i].updateCacheMultiDof();	
+	m_links[i].updateCacheMultiDof();
 	//
 	updateLinksDofOffsets();
 }
 
 void btMultiBody::setupPlanar(int i,
-						   btScalar mass,
-						   const btVector3 &inertia,
-						   int parent,
-						   const btQuaternion &rotParentToThis,
-						   const btVector3 &rotationAxis,
-						   const btVector3 &parentComToThisComOffset,						   
-						   bool disableParentCollision)
+							  btScalar mass,
+							  const btVector3 &inertia,
+							  int parent,
+							  const btQuaternion &rotParentToThis,
+							  const btVector3 &rotationAxis,
+							  const btVector3 &parentComToThisComOffset,
+							  bool disableParentCollision)
 {
-	
 	m_dofCount += 3;
 	m_posVarCnt += 3;
 
 	m_links[i].m_mass = mass;
-    m_links[i].m_inertiaLocal = inertia;
-    m_links[i].m_parent = parent;
-    m_links[i].m_zeroRotParentToThis = rotParentToThis;    
+	m_links[i].m_inertiaLocal = inertia;
+	m_links[i].m_parent = parent;
+	m_links[i].m_zeroRotParentToThis = rotParentToThis;
 	m_links[i].m_dVector.setZero();
-    m_links[i].m_eVector = parentComToThisComOffset;
+	m_links[i].m_eVector = parentComToThisComOffset;
 
 	//
 	btVector3 vecNonParallelToRotAxis(1, 0, 0);
-	if(rotationAxis.normalized().dot(vecNonParallelToRotAxis) > 0.999)
+	if (rotationAxis.normalized().dot(vecNonParallelToRotAxis) > 0.999)
 		vecNonParallelToRotAxis.setValue(0, 1, 0);
 	//
 
 	m_links[i].m_jointType = btMultibodyLink::ePlanar;
 	m_links[i].m_dofCount = 3;
 	m_links[i].m_posVarCount = 3;
-	btVector3 n=rotationAxis.normalized();
-	m_links[i].setAxisTop(0, n[0],n[1],n[2]);
-	m_links[i].setAxisTop(1,0,0,0);
-	m_links[i].setAxisTop(2,0,0,0);
-	m_links[i].setAxisBottom(0,0,0,0);
+	btVector3 n = rotationAxis.normalized();
+	m_links[i].setAxisTop(0, n[0], n[1], n[2]);
+	m_links[i].setAxisTop(1, 0, 0, 0);
+	m_links[i].setAxisTop(2, 0, 0, 0);
+	m_links[i].setAxisBottom(0, 0, 0, 0);
 	btVector3 cr = m_links[i].getAxisTop(0).cross(vecNonParallelToRotAxis);
-	m_links[i].setAxisBottom(1,cr[0],cr[1],cr[2]);
+	m_links[i].setAxisBottom(1, cr[0], cr[1], cr[2]);
 	cr = m_links[i].getAxisBottom(1).cross(m_links[i].getAxisTop(0));
-	m_links[i].setAxisBottom(2,cr[0],cr[1],cr[2]);
+	m_links[i].setAxisBottom(2, cr[0], cr[1], cr[2]);
 	m_links[i].m_jointPos[0] = m_links[i].m_jointPos[1] = m_links[i].m_jointPos[2] = 0.f;
 	m_links[i].m_jointTorque[0] = m_links[i].m_jointTorque[1] = m_links[i].m_jointTorque[2] = 0.f;
 
 	if (disableParentCollision)
-		m_links[i].m_flags |=BT_MULTIBODYLINKFLAGS_DISABLE_PARENT_COLLISION;
-    //
+		m_links[i].m_flags |= BT_MULTIBODYLINKFLAGS_DISABLE_PARENT_COLLISION;
+	//
 	m_links[i].updateCacheMultiDof();
 	//
 	updateLinksDofOffsets();
+
+	m_links[i].setAxisBottom(1, m_links[i].getAxisBottom(1).normalized());
+	m_links[i].setAxisBottom(2, m_links[i].getAxisBottom(2).normalized());
 }
 
 void btMultiBody::finalizeMultiDof()
 {
 	m_deltaV.resize(0);
 	m_deltaV.resize(6 + m_dofCount);
-	m_realBuf.resize(6 + m_dofCount + m_dofCount*m_dofCount + 6 + m_dofCount);			//m_dofCount for joint-space vels + m_dofCount^2 for "D" matrices + delta-pos vector (6 base "vels" + joint "vels")
-	m_vectorBuf.resize(2 * m_dofCount);													//two 3-vectors (i.e. one six-vector) for each system dof	("h" matrices)
-	for (int i=0;i<m_vectorBuf.size();i++)
+	m_realBuf.resize(6 + m_dofCount + m_dofCount * m_dofCount + 6 + m_dofCount);  //m_dofCount for joint-space vels + m_dofCount^2 for "D" matrices + delta-pos vector (6 base "vels" + joint "vels")
+	m_vectorBuf.resize(2 * m_dofCount);                                           //two 3-vectors (i.e. one six-vector) for each system dof	("h" matrices)
+	for (int i = 0; i < m_vectorBuf.size(); i++)
 	{
-		m_vectorBuf[i].setValue(0,0,0);
+		m_vectorBuf[i].setValue(0, 0, 0);
 	}
 	updateLinksDofOffsets();
 }
-	
+
 int btMultiBody::getParent(int i) const
 {
-    return m_links[i].m_parent;
+	return m_links[i].m_parent;
 }
 
 btScalar btMultiBody::getLinkMass(int i) const
 {
-    return m_links[i].m_mass;
+	return m_links[i].m_mass;
 }
 
-const btVector3 & btMultiBody::getLinkInertia(int i) const
+const btVector3 &btMultiBody::getLinkInertia(int i) const
 {
-    return m_links[i].m_inertiaLocal;
+	return m_links[i].m_inertiaLocal;
 }
 
 btScalar btMultiBody::getJointPos(int i) const
 {
-    return m_links[i].m_jointPos[0];
+	return m_links[i].m_jointPos[0];
 }
 
 btScalar btMultiBody::getJointVel(int i) const
 {
-    return m_realBuf[6 + m_links[i].m_dofOffset];
+	return m_realBuf[6 + m_links[i].m_dofOffset];
 }
 
-btScalar * btMultiBody::getJointPosMultiDof(int i)
+btScalar *btMultiBody::getJointPosMultiDof(int i)
 {
 	return &m_links[i].m_jointPos[0];
 }
 
-btScalar * btMultiBody::getJointVelMultiDof(int i)
+btScalar *btMultiBody::getJointVelMultiDof(int i)
 {
 	return &m_realBuf[6 + m_links[i].m_dofOffset];
 }
 
-const btScalar * btMultiBody::getJointPosMultiDof(int i) const 
+const btScalar *btMultiBody::getJointPosMultiDof(int i) const
 {
 	return &m_links[i].m_jointPos[0];
 }
 
-const btScalar * btMultiBody::getJointVelMultiDof(int i) const 
+const btScalar *btMultiBody::getJointVelMultiDof(int i) const
 {
 	return &m_realBuf[6 + m_links[i].m_dofOffset];
 }
 
-
 void btMultiBody::setJointPos(int i, btScalar q)
 {
-    m_links[i].m_jointPos[0] = q;
-    m_links[i].updateCacheMultiDof();
+	m_links[i].m_jointPos[0] = q;
+	m_links[i].updateCacheMultiDof();
 }
 
-void btMultiBody::setJointPosMultiDof(int i, btScalar *q)
+
+void btMultiBody::setJointPosMultiDof(int i, const double *q)
 {
-	for(int pos = 0; pos < m_links[i].m_posVarCount; ++pos)
-		m_links[i].m_jointPos[pos] = q[pos];
+	for (int pos = 0; pos < m_links[i].m_posVarCount; ++pos)
+		m_links[i].m_jointPos[pos] = (btScalar)q[pos];
 
-    m_links[i].updateCacheMultiDof();
+	m_links[i].updateCacheMultiDof();
 }
 
+void btMultiBody::setJointPosMultiDof(int i, const float *q)
+{
+	for (int pos = 0; pos < m_links[i].m_posVarCount; ++pos)
+		m_links[i].m_jointPos[pos] = (btScalar)q[pos];
+
+	m_links[i].updateCacheMultiDof();
+}
+
+
+
 void btMultiBody::setJointVel(int i, btScalar qdot)
 {
-    m_realBuf[6 + m_links[i].m_dofOffset] = qdot;
+	m_realBuf[6 + m_links[i].m_dofOffset] = qdot;
 }
 
-void btMultiBody::setJointVelMultiDof(int i, btScalar *qdot)
+void btMultiBody::setJointVelMultiDof(int i, const double *qdot)
 {
-	for(int dof = 0; dof < m_links[i].m_dofCount; ++dof)
-		m_realBuf[6 + m_links[i].m_dofOffset + dof] = qdot[dof];
+	for (int dof = 0; dof < m_links[i].m_dofCount; ++dof)
+		m_realBuf[6 + m_links[i].m_dofOffset + dof] = (btScalar)qdot[dof];
 }
 
-const btVector3 & btMultiBody::getRVector(int i) const
+void btMultiBody::setJointVelMultiDof(int i, const float* qdot)
 {
-    return m_links[i].m_cachedRVector;
+	for (int dof = 0; dof < m_links[i].m_dofCount; ++dof)
+		m_realBuf[6 + m_links[i].m_dofOffset + dof] = (btScalar)qdot[dof];
 }
 
-const btQuaternion & btMultiBody::getParentToLocalRot(int i) const
+const btVector3 &btMultiBody::getRVector(int i) const
 {
-    return m_links[i].m_cachedRotParentToThis;
+	return m_links[i].m_cachedRVector;
+}
+
+const btQuaternion &btMultiBody::getParentToLocalRot(int i) const
+{
+	return m_links[i].m_cachedRotParentToThis;
 }
 
 btVector3 btMultiBody::localPosToWorld(int i, const btVector3 &local_pos) const
 {
-	btAssert(i>=-1);
-	btAssert(i<m_links.size());
-	if ((i<-1) || (i>=m_links.size()))
+	btAssert(i >= -1);
+	btAssert(i < m_links.size());
+	if ((i < -1) || (i >= m_links.size()))
 	{
-		return btVector3(SIMD_INFINITY,SIMD_INFINITY,SIMD_INFINITY);
+		return btVector3(SIMD_INFINITY, SIMD_INFINITY, SIMD_INFINITY);
 	}
 
-    btVector3 result = local_pos;
-    while (i != -1) {
-        // 'result' is in frame i. transform it to frame parent(i)
-        result += getRVector(i);
-        result = quatRotate(getParentToLocalRot(i).inverse(),result);
-        i = getParent(i);
-    }
+	btVector3 result = local_pos;
+	while (i != -1)
+	{
+		// 'result' is in frame i. transform it to frame parent(i)
+		result += getRVector(i);
+		result = quatRotate(getParentToLocalRot(i).inverse(), result);
+		i = getParent(i);
+	}
 
-    // 'result' is now in the base frame. transform it to world frame
-    result = quatRotate(getWorldToBaseRot().inverse() ,result);
-    result += getBasePos();
+	// 'result' is now in the base frame. transform it to world frame
+	result = quatRotate(getWorldToBaseRot().inverse(), result);
+	result += getBasePos();
 
-    return result;
+	return result;
 }
 
 btVector3 btMultiBody::worldPosToLocal(int i, const btVector3 &world_pos) const
 {
-	btAssert(i>=-1);
-	btAssert(i<m_links.size());
-	if ((i<-1) || (i>=m_links.size()))
+	btAssert(i >= -1);
+	btAssert(i < m_links.size());
+	if ((i < -1) || (i >= m_links.size()))
 	{
-		return btVector3(SIMD_INFINITY,SIMD_INFINITY,SIMD_INFINITY);
+		return btVector3(SIMD_INFINITY, SIMD_INFINITY, SIMD_INFINITY);
 	}
 
-    if (i == -1) {
-        // world to base
-        return quatRotate(getWorldToBaseRot(),(world_pos - getBasePos()));
-    } else {
-        // find position in parent frame, then transform to current frame
-        return quatRotate(getParentToLocalRot(i),worldPosToLocal(getParent(i), world_pos)) - getRVector(i);
-    }
+	if (i == -1)
+	{
+		// world to base
+		return quatRotate(getWorldToBaseRot(), (world_pos - getBasePos()));
+	}
+	else
+	{
+		// find position in parent frame, then transform to current frame
+		return quatRotate(getParentToLocalRot(i), worldPosToLocal(getParent(i), world_pos)) - getRVector(i);
+	}
 }
 
 btVector3 btMultiBody::localDirToWorld(int i, const btVector3 &local_dir) const
 {
-	btAssert(i>=-1);
-	btAssert(i<m_links.size());
-	if ((i<-1) || (i>=m_links.size()))
+	btAssert(i >= -1);
+	btAssert(i < m_links.size());
+	if ((i < -1) || (i >= m_links.size()))
 	{
-		return btVector3(SIMD_INFINITY,SIMD_INFINITY,SIMD_INFINITY);
+		return btVector3(SIMD_INFINITY, SIMD_INFINITY, SIMD_INFINITY);
 	}
 
-
-    btVector3 result = local_dir;
-    while (i != -1) {
-        result = quatRotate(getParentToLocalRot(i).inverse() , result);
-        i = getParent(i);
-    }
-    result = quatRotate(getWorldToBaseRot().inverse() , result);
-    return result;
+	btVector3 result = local_dir;
+	while (i != -1)
+	{
+		result = quatRotate(getParentToLocalRot(i).inverse(), result);
+		i = getParent(i);
+	}
+	result = quatRotate(getWorldToBaseRot().inverse(), result);
+	return result;
 }
 
 btVector3 btMultiBody::worldDirToLocal(int i, const btVector3 &world_dir) const
 {
-	btAssert(i>=-1);
-	btAssert(i<m_links.size());
-	if ((i<-1) || (i>=m_links.size()))
+	btAssert(i >= -1);
+	btAssert(i < m_links.size());
+	if ((i < -1) || (i >= m_links.size()))
 	{
-		return btVector3(SIMD_INFINITY,SIMD_INFINITY,SIMD_INFINITY);
+		return btVector3(SIMD_INFINITY, SIMD_INFINITY, SIMD_INFINITY);
 	}
 
-    if (i == -1) {
-        return quatRotate(getWorldToBaseRot(), world_dir);
-    } else {
-        return quatRotate(getParentToLocalRot(i) ,worldDirToLocal(getParent(i), world_dir));
-    }
+	if (i == -1)
+	{
+		return quatRotate(getWorldToBaseRot(), world_dir);
+	}
+	else
+	{
+		return quatRotate(getParentToLocalRot(i), worldDirToLocal(getParent(i), world_dir));
+	}
 }
 
 btMatrix3x3 btMultiBody::localFrameToWorld(int i, const btMatrix3x3 &local_frame) const
 {
-    btMatrix3x3 result = local_frame;
-    btVector3 frameInWorld0 = localDirToWorld(i, local_frame.getColumn(0));
-    btVector3 frameInWorld1 = localDirToWorld(i, local_frame.getColumn(1));
-    btVector3 frameInWorld2 = localDirToWorld(i, local_frame.getColumn(2));
-    result.setValue(frameInWorld0[0], frameInWorld1[0], frameInWorld2[0], frameInWorld0[1], frameInWorld1[1], frameInWorld2[1], frameInWorld0[2], frameInWorld1[2], frameInWorld2[2]);
-    return result;
+	btMatrix3x3 result = local_frame;
+	btVector3 frameInWorld0 = localDirToWorld(i, local_frame.getColumn(0));
+	btVector3 frameInWorld1 = localDirToWorld(i, local_frame.getColumn(1));
+	btVector3 frameInWorld2 = localDirToWorld(i, local_frame.getColumn(2));
+	result.setValue(frameInWorld0[0], frameInWorld1[0], frameInWorld2[0], frameInWorld0[1], frameInWorld1[1], frameInWorld2[1], frameInWorld0[2], frameInWorld1[2], frameInWorld2[2]);
+	return result;
 }
 
 void btMultiBody::compTreeLinkVelocities(btVector3 *omega, btVector3 *vel) const
 {
 	int num_links = getNumLinks();
-    // Calculates the velocities of each link (and the base) in its local frame
-    omega[0] = quatRotate(m_baseQuat ,getBaseOmega());
-    vel[0] = quatRotate(m_baseQuat ,getBaseVel());
-    
-    for (int i = 0; i < num_links; ++i) 
+	// Calculates the velocities of each link (and the base) in its local frame
+	const btQuaternion& base_rot = getWorldToBaseRot();
+	omega[0] = quatRotate(base_rot, getBaseOmega());
+	vel[0] = quatRotate(base_rot, getBaseVel());
+
+	for (int i = 0; i < num_links; ++i)
 	{
-        const int parent = m_links[i].m_parent;
+		const btMultibodyLink& link = getLink(i);
+		const int parent = link.m_parent;
 
-        // transform parent vel into this frame, store in omega[i+1], vel[i+1]
-        SpatialTransform(btMatrix3x3(m_links[i].m_cachedRotParentToThis), m_links[i].m_cachedRVector,
-                         omega[parent+1], vel[parent+1],
-                         omega[i+1], vel[i+1]);
+		// transform parent vel into this frame, store in omega[i+1], vel[i+1]
+		spatialTransform(btMatrix3x3(link.m_cachedRotParentToThis), link.m_cachedRVector,
+			omega[parent + 1], vel[parent + 1],
+			omega[i + 1], vel[i + 1]);
 
-        // now add qidot * shat_i
-		//only supported for revolute/prismatic joints, todo: spherical and planar joints
-		switch(m_links[i].m_jointType)
+		// now add qidot * shat_i
+		const btScalar* jointVel = getJointVelMultiDof(i);
+		for (int dof = 0; dof < link.m_dofCount; ++dof)
 		{
-			case btMultibodyLink::ePrismatic:
-			case btMultibodyLink::eRevolute:
-			{
-				btVector3 axisTop = m_links[i].getAxisTop(0);
-				btVector3 axisBottom = m_links[i].getAxisBottom(0);
-				btScalar jointVel = getJointVel(i);
-				omega[i+1] += jointVel * axisTop;
-				vel[i+1] += jointVel * axisBottom;
-				break;
-			}
-			default:
-			{
-			}
+			omega[i + 1] += jointVel[dof] * link.getAxisTop(dof);
+			vel[i + 1] += jointVel[dof] * link.getAxisBottom(dof);
 		}
 	}
 }
@@ -564,41 +572,48 @@ void btMultiBody::compTreeLinkVelocities(btVector3 *omega, btVector3 *vel) const
 btScalar btMultiBody::getKineticEnergy() const
 {
 	int num_links = getNumLinks();
-    // TODO: would be better not to allocate memory here
-    btAlignedObjectArray<btVector3> omega;omega.resize(num_links+1);
-	btAlignedObjectArray<btVector3> vel;vel.resize(num_links+1);
-    compTreeLinkVelocities(&omega[0], &vel[0]);
+	// TODO: would be better not to allocate memory here
+	btAlignedObjectArray<btVector3> omega;
+	omega.resize(num_links + 1);
+	btAlignedObjectArray<btVector3> vel;
+	vel.resize(num_links + 1);
+	compTreeLinkVelocities(&omega[0], &vel[0]);
 
-    // we will do the factor of 0.5 at the end
-    btScalar result = m_baseMass * vel[0].dot(vel[0]);
-    result += omega[0].dot(m_baseInertia * omega[0]);
-    
-    for (int i = 0; i < num_links; ++i) {
-        result += m_links[i].m_mass * vel[i+1].dot(vel[i+1]);
-        result += omega[i+1].dot(m_links[i].m_inertiaLocal * omega[i+1]);
-    }
+	// we will do the factor of 0.5 at the end
+	btScalar result = m_baseMass * vel[0].dot(vel[0]);
+	result += omega[0].dot(m_baseInertia * omega[0]);
 
-    return 0.5f * result;
+	for (int i = 0; i < num_links; ++i)
+	{
+		result += m_links[i].m_mass * vel[i + 1].dot(vel[i + 1]);
+		result += omega[i + 1].dot(m_links[i].m_inertiaLocal * omega[i + 1]);
+	}
+
+	return 0.5f * result;
 }
 
 btVector3 btMultiBody::getAngularMomentum() const
 {
 	int num_links = getNumLinks();
-    // TODO: would be better not to allocate memory here
-    btAlignedObjectArray<btVector3> omega;omega.resize(num_links+1);
-	btAlignedObjectArray<btVector3> vel;vel.resize(num_links+1);
-    btAlignedObjectArray<btQuaternion> rot_from_world;rot_from_world.resize(num_links+1);
-    compTreeLinkVelocities(&omega[0], &vel[0]);
+	// TODO: would be better not to allocate memory here
+	btAlignedObjectArray<btVector3> omega;
+	omega.resize(num_links + 1);
+	btAlignedObjectArray<btVector3> vel;
+	vel.resize(num_links + 1);
+	btAlignedObjectArray<btQuaternion> rot_from_world;
+	rot_from_world.resize(num_links + 1);
+	compTreeLinkVelocities(&omega[0], &vel[0]);
 
-    rot_from_world[0] = m_baseQuat;
-    btVector3 result = quatRotate(rot_from_world[0].inverse() , (m_baseInertia * omega[0]));
+	rot_from_world[0] = m_baseQuat;
+	btVector3 result = quatRotate(rot_from_world[0].inverse(), (m_baseInertia * omega[0]));
 
-    for (int i = 0; i < num_links; ++i) {
-        rot_from_world[i+1] = m_links[i].m_cachedRotParentToThis * rot_from_world[m_links[i].m_parent+1];
-        result += (quatRotate(rot_from_world[i+1].inverse() , (m_links[i].m_inertiaLocal * omega[i+1])));
-    }
+	for (int i = 0; i < num_links; ++i)
+	{
+		rot_from_world[i + 1] = m_links[i].m_cachedRotParentToThis * rot_from_world[m_links[i].m_parent + 1];
+		result += (quatRotate(rot_from_world[i + 1].inverse(), (m_links[i].m_inertiaLocal * omega[i + 1])));
+	}
 
-    return result;
+	return result;
 }
 
 void btMultiBody::clearConstraintForces()
@@ -606,57 +621,55 @@ void btMultiBody::clearConstraintForces()
 	m_baseConstraintForce.setValue(0, 0, 0);
 	m_baseConstraintTorque.setValue(0, 0, 0);
 
-
-    for (int i = 0; i < getNumLinks(); ++i) {
-        m_links[i].m_appliedConstraintForce.setValue(0, 0, 0);
-        m_links[i].m_appliedConstraintTorque.setValue(0, 0, 0);
-    }
+	for (int i = 0; i < getNumLinks(); ++i)
+	{
+		m_links[i].m_appliedConstraintForce.setValue(0, 0, 0);
+		m_links[i].m_appliedConstraintTorque.setValue(0, 0, 0);
+	}
 }
 void btMultiBody::clearForcesAndTorques()
 {
-    m_baseForce.setValue(0, 0, 0);
-    m_baseTorque.setValue(0, 0, 0);
+	m_baseForce.setValue(0, 0, 0);
+	m_baseTorque.setValue(0, 0, 0);
 
-	
-    for (int i = 0; i < getNumLinks(); ++i) {
-        m_links[i].m_appliedForce.setValue(0, 0, 0);
-        m_links[i].m_appliedTorque.setValue(0, 0, 0);
+	for (int i = 0; i < getNumLinks(); ++i)
+	{
+		m_links[i].m_appliedForce.setValue(0, 0, 0);
+		m_links[i].m_appliedTorque.setValue(0, 0, 0);
 		m_links[i].m_jointTorque[0] = m_links[i].m_jointTorque[1] = m_links[i].m_jointTorque[2] = m_links[i].m_jointTorque[3] = m_links[i].m_jointTorque[4] = m_links[i].m_jointTorque[5] = 0.f;
-    }
+	}
 }
 
 void btMultiBody::clearVelocities()
 {
-	for (int i = 0; i < 6 + getNumDofs(); ++i) 
+	for (int i = 0; i < 6 + getNumDofs(); ++i)
 	{
 		m_realBuf[i] = 0.f;
 	}
 }
 void btMultiBody::addLinkForce(int i, const btVector3 &f)
 {
-    m_links[i].m_appliedForce += f;
+	m_links[i].m_appliedForce += f;
 }
 
 void btMultiBody::addLinkTorque(int i, const btVector3 &t)
 {
-    m_links[i].m_appliedTorque += t;
+	m_links[i].m_appliedTorque += t;
 }
 
 void btMultiBody::addLinkConstraintForce(int i, const btVector3 &f)
 {
-    m_links[i].m_appliedConstraintForce += f;
+	m_links[i].m_appliedConstraintForce += f;
 }
 
 void btMultiBody::addLinkConstraintTorque(int i, const btVector3 &t)
 {
-    m_links[i].m_appliedConstraintTorque += t;
+	m_links[i].m_appliedConstraintTorque += t;
 }
 
-
-
 void btMultiBody::addJointTorque(int i, btScalar Q)
 {
-    m_links[i].m_jointTorque[0] += Q;
+	m_links[i].m_jointTorque[0] += Q;
 }
 
 void btMultiBody::addJointTorqueMultiDof(int i, int dof, btScalar Q)
@@ -666,70 +679,72 @@ void btMultiBody::addJointTorqueMultiDof(int i, int dof, btScalar Q)
 
 void btMultiBody::addJointTorqueMultiDof(int i, const btScalar *Q)
 {
-	for(int dof = 0; dof < m_links[i].m_dofCount; ++dof)
+	for (int dof = 0; dof < m_links[i].m_dofCount; ++dof)
 		m_links[i].m_jointTorque[dof] = Q[dof];
 }
 
-const btVector3 & btMultiBody::getLinkForce(int i) const
+const btVector3 &btMultiBody::getLinkForce(int i) const
 {
-    return m_links[i].m_appliedForce;
+	return m_links[i].m_appliedForce;
 }
 
-const btVector3 & btMultiBody::getLinkTorque(int i) const
+const btVector3 &btMultiBody::getLinkTorque(int i) const
 {
-    return m_links[i].m_appliedTorque;
+	return m_links[i].m_appliedTorque;
 }
 
 btScalar btMultiBody::getJointTorque(int i) const
 {
-    return m_links[i].m_jointTorque[0];
+	return m_links[i].m_jointTorque[0];
 }
 
-btScalar * btMultiBody::getJointTorqueMultiDof(int i)
+btScalar *btMultiBody::getJointTorqueMultiDof(int i)
 {
-    return &m_links[i].m_jointTorque[0];
+	return &m_links[i].m_jointTorque[0];
 }
 
-inline btMatrix3x3 outerProduct(const btVector3& v0, const btVector3& v1)				//renamed it from vecMulVecTranspose (http://en.wikipedia.org/wiki/Outer_product); maybe it should be moved to btVector3 like dot and cross?
+inline btMatrix3x3 outerProduct(const btVector3 &v0, const btVector3 &v1)  //renamed it from vecMulVecTranspose (http://en.wikipedia.org/wiki/Outer_product); maybe it should be moved to btVector3 like dot and cross?
 {
-		btVector3 row0 = btVector3( 
-			v0.x() * v1.x(),
-			v0.x() * v1.y(),
-			v0.x() * v1.z());
-		btVector3 row1 = btVector3( 
-			v0.y() * v1.x(),
-			v0.y() * v1.y(),
-			v0.y() * v1.z());
-		btVector3 row2 = btVector3( 
-			v0.z() * v1.x(),
-			v0.z() * v1.y(),
-			v0.z() * v1.z());
+	btVector3 row0 = btVector3(
+		v0.x() * v1.x(),
+		v0.x() * v1.y(),
+		v0.x() * v1.z());
+	btVector3 row1 = btVector3(
+		v0.y() * v1.x(),
+		v0.y() * v1.y(),
+		v0.y() * v1.z());
+	btVector3 row2 = btVector3(
+		v0.z() * v1.x(),
+		v0.z() * v1.y(),
+		v0.z() * v1.z());
 
-        btMatrix3x3 m(row0[0],row0[1],row0[2],
-						row1[0],row1[1],row1[2],
-						row2[0],row2[1],row2[2]);
-		return m;
+	btMatrix3x3 m(row0[0], row0[1], row0[2],
+				  row1[0], row1[1], row1[2],
+				  row2[0], row2[1], row2[2]);
+	return m;
 }
 
 #define vecMulVecTranspose(v0, v1Transposed) outerProduct(v0, v1Transposed)
 //
 
 void btMultiBody::computeAccelerationsArticulatedBodyAlgorithmMultiDof(btScalar dt,
-                               btAlignedObjectArray<btScalar> &scratch_r,
-                               btAlignedObjectArray<btVector3> &scratch_v,
-                               btAlignedObjectArray<btMatrix3x3> &scratch_m,
-				bool isConstraintPass)
+    btAlignedObjectArray<btScalar> &scratch_r,
+    btAlignedObjectArray<btVector3> &scratch_v,
+    btAlignedObjectArray<btMatrix3x3> &scratch_m,
+	bool isConstraintPass,
+	bool jointFeedbackInWorldSpace,
+	bool jointFeedbackInJointFrame)
 {
-    // Implement Featherstone's algorithm to calculate joint accelerations (q_double_dot)
-    // and the base linear & angular accelerations.
+	// Implement Featherstone's algorithm to calculate joint accelerations (q_double_dot)
+	// and the base linear & angular accelerations.
 
-    // We apply damping forces in this routine as well as any external forces specified by the 
-    // caller (via addBaseForce etc).
+	// We apply damping forces in this routine as well as any external forces specified by the
+	// caller (via addBaseForce etc).
+
+	// output should point to an array of 6 + num_links reals.
+	// Format is: 3 angular accelerations (in world frame), 3 linear accelerations (in world frame),
+	// num_links joint acceleration values.
 
-    // output should point to an array of 6 + num_links reals.
-    // Format is: 3 angular accelerations (in world frame), 3 linear accelerations (in world frame),
-    // num_links joint acceleration values.
-    
 	// We added support for multi degree of freedom (multi dof) joints.
 	// In addition we also can compute the joint reaction forces. This is performed in a second pass,
 	// so that we can include the effect of the constraint solver forces (computed in the PGS LCP solver)
@@ -738,96 +753,96 @@ void btMultiBody::computeAccelerationsArticulatedBodyAlgorithmMultiDof(btScalar
 
 	int num_links = getNumLinks();
 
-    const btScalar DAMPING_K1_LINEAR = m_linearDamping;
+	const btScalar DAMPING_K1_LINEAR = m_linearDamping;
 	const btScalar DAMPING_K2_LINEAR = m_linearDamping;
 
 	const btScalar DAMPING_K1_ANGULAR = m_angularDamping;
-	const btScalar DAMPING_K2_ANGULAR= m_angularDamping;
+	const btScalar DAMPING_K2_ANGULAR = m_angularDamping;
 
 	const btVector3 base_vel = getBaseVel();
 	const btVector3 base_omega = getBaseOmega();
 
-    // Temporary matrices/vectors -- use scratch space from caller
-    // so that we don't have to keep reallocating every frame
+	// Temporary matrices/vectors -- use scratch space from caller
+	// so that we don't have to keep reallocating every frame
 
-    scratch_r.resize(2*m_dofCount + 7);				//multidof? ("Y"s use it and it is used to store qdd) => 2 x m_dofCount
-    scratch_v.resize(8*num_links + 6);
-    scratch_m.resize(4*num_links + 4);
+	scratch_r.resize(2 * m_dofCount + 7);  //multidof? ("Y"s use it and it is used to store qdd) => 2 x m_dofCount
+	scratch_v.resize(8 * num_links + 6);
+	scratch_m.resize(4 * num_links + 4);
 
 	//btScalar * r_ptr = &scratch_r[0];
-    btScalar * output = &scratch_r[m_dofCount];  // "output" holds the q_double_dot results
-    btVector3 * v_ptr = &scratch_v[0];
-    
-    // vhat_i  (top = angular, bottom = linear part)	
+	btScalar *output = &scratch_r[m_dofCount];  // "output" holds the q_double_dot results
+	btVector3 *v_ptr = &scratch_v[0];
+
+	// vhat_i  (top = angular, bottom = linear part)
 	btSpatialMotionVector *spatVel = (btSpatialMotionVector *)v_ptr;
 	v_ptr += num_links * 2 + 2;
 	//
-    // zhat_i^A    
-	btSpatialForceVector * zeroAccSpatFrc = (btSpatialForceVector *)v_ptr;
+	// zhat_i^A
+	btSpatialForceVector *zeroAccSpatFrc = (btSpatialForceVector *)v_ptr;
 	v_ptr += num_links * 2 + 2;
 	//
-    // chat_i  (note NOT defined for the base)    
-	btSpatialMotionVector * spatCoriolisAcc = (btSpatialMotionVector *)v_ptr;
+	// chat_i  (note NOT defined for the base)
+	btSpatialMotionVector *spatCoriolisAcc = (btSpatialMotionVector *)v_ptr;
 	v_ptr += num_links * 2;
 	//
-    // Ihat_i^A.    
-	btSymmetricSpatialDyad * spatInertia = (btSymmetricSpatialDyad *)&scratch_m[num_links + 1];
+	// Ihat_i^A.
+	btSymmetricSpatialDyad *spatInertia = (btSymmetricSpatialDyad *)&scratch_m[num_links + 1];
 
-    // Cached 3x3 rotation matrices from parent frame to this frame.
-    btMatrix3x3 * rot_from_parent = &m_matrixBuf[0];
-    btMatrix3x3 * rot_from_world = &scratch_m[0];
+	// Cached 3x3 rotation matrices from parent frame to this frame.
+	btMatrix3x3 *rot_from_parent = &m_matrixBuf[0];
+	btMatrix3x3 *rot_from_world = &scratch_m[0];
 
-    // hhat_i, ahat_i
-    // hhat is NOT stored for the base (but ahat is)    
-	btSpatialForceVector * h = (btSpatialForceVector *)(m_dofCount > 0 ? &m_vectorBuf[0] : 0);
-	btSpatialMotionVector * spatAcc = (btSpatialMotionVector *)v_ptr;
-	v_ptr += num_links * 2 + 2; 
+	// hhat_i, ahat_i
+	// hhat is NOT stored for the base (but ahat is)
+	btSpatialForceVector *h = (btSpatialForceVector *)(m_dofCount > 0 ? &m_vectorBuf[0] : 0);
+	btSpatialMotionVector *spatAcc = (btSpatialMotionVector *)v_ptr;
+	v_ptr += num_links * 2 + 2;
 	//
-    // Y_i, invD_i
-    btScalar * invD = m_dofCount > 0 ? &m_realBuf[6 + m_dofCount] : 0;
-	btScalar * Y = &scratch_r[0];
+	// Y_i, invD_i
+	btScalar *invD = m_dofCount > 0 ? &m_realBuf[6 + m_dofCount] : 0;
+	btScalar *Y = &scratch_r[0];
 	//
-	//aux variables	
-	btSpatialMotionVector spatJointVel;					//spatial velocity due to the joint motion (i.e. without predecessors' influence)
-	btScalar D[36];										//"D" matrix; it's dofxdof for each body so asingle 6x6 D matrix will do	
-	btScalar invD_times_Y[6];							//D^{-1} * Y [dofxdof x dofx1 = dofx1] <=> D^{-1} * u; better moved to buffers since it is recalced in calcAccelerationDeltasMultiDof; num_dof of btScalar would cover all bodies	
-	btSpatialMotionVector result;							//holds results of the SolveImatrix op; it is a spatial motion vector (accel)
-	btScalar Y_minus_hT_a[6];							//Y - h^{T} * a; it's dofx1 for each body so a single 6x1 temp is enough	
-	btSpatialForceVector spatForceVecTemps[6];				//6 temporary spatial force vectors
-	btSpatialTransformationMatrix fromParent;				//spatial transform from parent to child
-	btSymmetricSpatialDyad dyadTemp;						//inertia matrix temp
+	//aux variables
+	btSpatialMotionVector spatJointVel;         //spatial velocity due to the joint motion (i.e. without predecessors' influence)
+	btScalar D[36];                             //"D" matrix; it's dofxdof for each body so asingle 6x6 D matrix will do
+	btScalar invD_times_Y[6];                   //D^{-1} * Y [dofxdof x dofx1 = dofx1] <=> D^{-1} * u; better moved to buffers since it is recalced in calcAccelerationDeltasMultiDof; num_dof of btScalar would cover all bodies
+	btSpatialMotionVector result;               //holds results of the SolveImatrix op; it is a spatial motion vector (accel)
+	btScalar Y_minus_hT_a[6];                   //Y - h^{T} * a; it's dofx1 for each body so a single 6x1 temp is enough
+	btSpatialForceVector spatForceVecTemps[6];  //6 temporary spatial force vectors
+	btSpatialTransformationMatrix fromParent;   //spatial transform from parent to child
+	btSymmetricSpatialDyad dyadTemp;            //inertia matrix temp
 	btSpatialTransformationMatrix fromWorld;
 	fromWorld.m_trnVec.setZero();
 	/////////////////
 
-    // ptr to the joint accel part of the output
-    btScalar * joint_accel = output + 6;
+	// ptr to the joint accel part of the output
+	btScalar *joint_accel = output + 6;
 
-    // Start of the algorithm proper.
-    
-    // First 'upward' loop.
-    // Combines CompTreeLinkVelocities and InitTreeLinks from Mirtich.
+	// Start of the algorithm proper.
 
-    rot_from_parent[0] = btMatrix3x3(m_baseQuat);				//m_baseQuat assumed to be alias!?
+	// First 'upward' loop.
+	// Combines CompTreeLinkVelocities and InitTreeLinks from Mirtich.
+
+	rot_from_parent[0] = btMatrix3x3(m_baseQuat);  //m_baseQuat assumed to be alias!?
 
 	//create the vector of spatial velocity of the base by transforming global-coor linear and angular velocities into base-local coordinates
 	spatVel[0].setVector(rot_from_parent[0] * base_omega, rot_from_parent[0] * base_vel);
 
-    if (m_fixedBase) 
-	{        
-		zeroAccSpatFrc[0].setZero();
-    }
-	else 
+	if (m_fixedBase)
 	{
-		const btVector3& baseForce = isConstraintPass? m_baseConstraintForce : m_baseForce;
-		const btVector3& baseTorque = isConstraintPass? m_baseConstraintTorque : m_baseTorque;
-		//external forces		
-		zeroAccSpatFrc[0].setVector(-(rot_from_parent[0] * baseTorque), -(rot_from_parent[0] * baseForce));	
+		zeroAccSpatFrc[0].setZero();
+	}
+	else
+	{
+		const btVector3 &baseForce = isConstraintPass ? m_baseConstraintForce : m_baseForce;
+		const btVector3 &baseTorque = isConstraintPass ? m_baseConstraintTorque : m_baseTorque;
+		//external forces
+		zeroAccSpatFrc[0].setVector(-(rot_from_parent[0] * baseTorque), -(rot_from_parent[0] * baseForce));
 
 		//adding damping terms (only)
 		const btScalar linDampMult = 1., angDampMult = 1.;
 		zeroAccSpatFrc[0].addVector(angDampMult * m_baseInertia * spatVel[0].getAngular() * (DAMPING_K1_ANGULAR + DAMPING_K2_ANGULAR * spatVel[0].getAngular().safeNorm()),
-								   linDampMult * m_baseMass * spatVel[0].getLinear() * (DAMPING_K1_LINEAR + DAMPING_K2_LINEAR * spatVel[0].getLinear().safeNorm()));
+									linDampMult * m_baseMass * spatVel[0].getLinear() * (DAMPING_K1_LINEAR + DAMPING_K2_LINEAR * spatVel[0].getLinear().safeNorm()));
 
 		//
 		//p += vhat x Ihat vhat - done in a simpler way
@@ -835,67 +850,66 @@ void btMultiBody::computeAccelerationsArticulatedBodyAlgorithmMultiDof(btScalar
 			zeroAccSpatFrc[0].addAngular(spatVel[0].getAngular().cross(m_baseInertia * spatVel[0].getAngular()));
 		//
 		zeroAccSpatFrc[0].addLinear(m_baseMass * spatVel[0].getAngular().cross(spatVel[0].getLinear()));
-    }
-
+	}
 
 	//init the spatial AB inertia (it has the simple form thanks to choosing local body frames origins at their COMs)
-	spatInertia[0].setMatrix(	btMatrix3x3(0,0,0,0,0,0,0,0,0),
-								//
-								btMatrix3x3(m_baseMass, 0, 0,
-											0, m_baseMass, 0,
-											0, 0, m_baseMass),
-								//
-								btMatrix3x3(m_baseInertia[0], 0, 0,
-											0, m_baseInertia[1], 0,
-											0, 0, m_baseInertia[2])
-							);
+	spatInertia[0].setMatrix(btMatrix3x3(0, 0, 0, 0, 0, 0, 0, 0, 0),
+							 //
+							 btMatrix3x3(m_baseMass, 0, 0,
+										 0, m_baseMass, 0,
+										 0, 0, m_baseMass),
+							 //
+							 btMatrix3x3(m_baseInertia[0], 0, 0,
+										 0, m_baseInertia[1], 0,
+										 0, 0, m_baseInertia[2]));
 
-    rot_from_world[0] = rot_from_parent[0];
+	rot_from_world[0] = rot_from_parent[0];
 
 	//
-    for (int i = 0; i < num_links; ++i) {		
-        const int parent = m_links[i].m_parent;
-        rot_from_parent[i+1] = btMatrix3x3(m_links[i].m_cachedRotParentToThis);
-        rot_from_world[i+1] = rot_from_parent[i+1] * rot_from_world[parent+1];
+	for (int i = 0; i < num_links; ++i)
+	{
+		const int parent = m_links[i].m_parent;
+		rot_from_parent[i + 1] = btMatrix3x3(m_links[i].m_cachedRotParentToThis);
+		rot_from_world[i + 1] = rot_from_parent[i + 1] * rot_from_world[parent + 1];
 
-		fromParent.m_rotMat = rot_from_parent[i+1]; fromParent.m_trnVec = m_links[i].m_cachedRVector;
-		fromWorld.m_rotMat = rot_from_world[i+1];
-		fromParent.transform(spatVel[parent+1], spatVel[i+1]);
+		fromParent.m_rotMat = rot_from_parent[i + 1];
+		fromParent.m_trnVec = m_links[i].m_cachedRVector;
+		fromWorld.m_rotMat = rot_from_world[i + 1];
+		fromParent.transform(spatVel[parent + 1], spatVel[i + 1]);
 
 		// now set vhat_i to its true value by doing
-        // vhat_i += qidot * shat_i			
-		if(!m_useGlobalVelocities)
+		// vhat_i += qidot * shat_i
+		if (!m_useGlobalVelocities)
 		{
 			spatJointVel.setZero();
 
-			for(int dof = 0; dof < m_links[i].m_dofCount; ++dof)		
+			for (int dof = 0; dof < m_links[i].m_dofCount; ++dof)
 				spatJointVel += m_links[i].m_axes[dof] * getJointVelMultiDof(i)[dof];
 
 			// remember vhat_i is really vhat_p(i) (but in current frame) at this point	=> we need to add velocity across the inboard joint
-			spatVel[i+1] += spatJointVel;
+			spatVel[i + 1] += spatJointVel;
 
 			//
 			// vhat_i is vhat_p(i) transformed to local coors + the velocity across the i-th inboard joint
 			//spatVel[i+1] = fromParent * spatVel[parent+1] + spatJointVel;
-
 		}
 		else
 		{
-			fromWorld.transformRotationOnly(m_links[i].m_absFrameTotVelocity, spatVel[i+1]);
+			fromWorld.transformRotationOnly(m_links[i].m_absFrameTotVelocity, spatVel[i + 1]);
 			fromWorld.transformRotationOnly(m_links[i].m_absFrameLocVelocity, spatJointVel);
 		}
 
-		// we can now calculate chat_i 		
-		spatVel[i+1].cross(spatJointVel, spatCoriolisAcc[i]);		
+		// we can now calculate chat_i
+		spatVel[i + 1].cross(spatJointVel, spatCoriolisAcc[i]);
 
-        // calculate zhat_i^A
+		// calculate zhat_i^A
 		//
-		//external forces		
-		btVector3 linkAppliedForce = isConstraintPass? m_links[i].m_appliedConstraintForce : m_links[i].m_appliedForce;
-		btVector3 linkAppliedTorque =isConstraintPass ? m_links[i].m_appliedConstraintTorque : m_links[i].m_appliedTorque;
- 
-		zeroAccSpatFrc[i+1].setVector(-(rot_from_world[i+1] * linkAppliedTorque), -(rot_from_world[i+1] * linkAppliedForce ));
-	
+		//external forces
+		btVector3 linkAppliedForce = isConstraintPass ? m_links[i].m_appliedConstraintForce : m_links[i].m_appliedForce;
+		btVector3 linkAppliedTorque = isConstraintPass ? m_links[i].m_appliedConstraintTorque : m_links[i].m_appliedTorque;
+
+		zeroAccSpatFrc[i + 1].setVector(-(rot_from_world[i + 1] * linkAppliedTorque), -(rot_from_world[i + 1] * linkAppliedForce));
+
 #if 0	
 		{
 
@@ -913,27 +927,26 @@ void btMultiBody::computeAccelerationsArticulatedBodyAlgorithmMultiDof(btScalar
 		//
 		//adding damping terms (only)
 		btScalar linDampMult = 1., angDampMult = 1.;
-		zeroAccSpatFrc[i+1].addVector(angDampMult * m_links[i].m_inertiaLocal * spatVel[i+1].getAngular() * (DAMPING_K1_ANGULAR + DAMPING_K2_ANGULAR * spatVel[i+1].getAngular().safeNorm()),
-									 linDampMult * m_links[i].m_mass * spatVel[i+1].getLinear() * (DAMPING_K1_LINEAR + DAMPING_K2_LINEAR * spatVel[i+1].getLinear().safeNorm()));
-		
-        // calculate Ihat_i^A
+		zeroAccSpatFrc[i + 1].addVector(angDampMult * m_links[i].m_inertiaLocal * spatVel[i + 1].getAngular() * (DAMPING_K1_ANGULAR + DAMPING_K2_ANGULAR * spatVel[i + 1].getAngular().safeNorm()),
+										linDampMult * m_links[i].m_mass * spatVel[i + 1].getLinear() * (DAMPING_K1_LINEAR + DAMPING_K2_LINEAR * spatVel[i + 1].getLinear().safeNorm()));
+
+		// calculate Ihat_i^A
 		//init the spatial AB inertia (it has the simple form thanks to choosing local body frames origins at their COMs)
-		spatInertia[i+1].setMatrix(	btMatrix3x3(0,0,0,0,0,0,0,0,0),
-									//
-									btMatrix3x3(m_links[i].m_mass, 0, 0,
-												0, m_links[i].m_mass, 0,
-												0, 0, m_links[i].m_mass),
-									//
-									btMatrix3x3(m_links[i].m_inertiaLocal[0], 0, 0,
-												0, m_links[i].m_inertiaLocal[1], 0,
-												0, 0, m_links[i].m_inertiaLocal[2])
-								);
+		spatInertia[i + 1].setMatrix(btMatrix3x3(0, 0, 0, 0, 0, 0, 0, 0, 0),
+									 //
+									 btMatrix3x3(m_links[i].m_mass, 0, 0,
+												 0, m_links[i].m_mass, 0,
+												 0, 0, m_links[i].m_mass),
+									 //
+									 btMatrix3x3(m_links[i].m_inertiaLocal[0], 0, 0,
+												 0, m_links[i].m_inertiaLocal[1], 0,
+												 0, 0, m_links[i].m_inertiaLocal[2]));
 		//
 		//p += vhat x Ihat vhat - done in a simpler way
-		if(m_useGyroTerm)
-			zeroAccSpatFrc[i+1].addAngular(spatVel[i+1].getAngular().cross(m_links[i].m_inertiaLocal * spatVel[i+1].getAngular()));			
-		//		
-		zeroAccSpatFrc[i+1].addLinear(m_links[i].m_mass * spatVel[i+1].getAngular().cross(spatVel[i+1].getLinear()));
+		if (m_useGyroTerm)
+			zeroAccSpatFrc[i + 1].addAngular(spatVel[i + 1].getAngular().cross(m_links[i].m_inertiaLocal * spatVel[i + 1].getAngular()));
+		//
+		zeroAccSpatFrc[i + 1].addLinear(m_links[i].m_mass * spatVel[i + 1].getAngular().cross(spatVel[i + 1].getLinear()));
 		//btVector3 temp = m_links[i].m_mass * spatVel[i+1].getAngular().cross(spatVel[i+1].getLinear());
 		////clamp parent's omega
 		//btScalar parOmegaMod = temp.length();
@@ -944,52 +957,49 @@ void btMultiBody::computeAccelerationsArticulatedBodyAlgorithmMultiDof(btScalar
 		//printf("|zeroAccSpatFrc[%d]| = %.4f\n", i+1, temp.length());
 		//temp = spatCoriolisAcc[i].getLinear();
 		//printf("|spatCoriolisAcc[%d]| = %.4f\n", i+1, temp.length());
-		
-		
 
 		//printf("w[%d] = [%.4f %.4f %.4f]\n", i, vel_top_angular[i+1].x(), vel_top_angular[i+1].y(), vel_top_angular[i+1].z());
-		//printf("v[%d] = [%.4f %.4f %.4f]\n", i, vel_bottom_linear[i+1].x(), vel_bottom_linear[i+1].y(), vel_bottom_linear[i+1].z());		
+		//printf("v[%d] = [%.4f %.4f %.4f]\n", i, vel_bottom_linear[i+1].x(), vel_bottom_linear[i+1].y(), vel_bottom_linear[i+1].z());
 		//printf("c[%d] = [%.4f %.4f %.4f]\n", i, coriolis_bottom_linear[i].x(), coriolis_bottom_linear[i].y(), coriolis_bottom_linear[i].z());
-    }
-	
-    // 'Downward' loop.
-    // (part of TreeForwardDynamics in Mirtich.)
-    for (int i = num_links - 1; i >= 0; --i)
+	}
+
+	// 'Downward' loop.
+	// (part of TreeForwardDynamics in Mirtich.)
+	for (int i = num_links - 1; i >= 0; --i)
 	{
 		const int parent = m_links[i].m_parent;
-		fromParent.m_rotMat = rot_from_parent[i+1]; fromParent.m_trnVec = m_links[i].m_cachedRVector;
+		fromParent.m_rotMat = rot_from_parent[i + 1];
+		fromParent.m_trnVec = m_links[i].m_cachedRVector;
 
-		for(int dof = 0; dof < m_links[i].m_dofCount; ++dof)
+		for (int dof = 0; dof < m_links[i].m_dofCount; ++dof)
 		{
 			btSpatialForceVector &hDof = h[m_links[i].m_dofOffset + dof];
 			//
-			hDof = spatInertia[i+1] * m_links[i].m_axes[dof];
+			hDof = spatInertia[i + 1] * m_links[i].m_axes[dof];
 			//
-			Y[m_links[i].m_dofOffset + dof] = m_links[i].m_jointTorque[dof]
-			- m_links[i].m_axes[dof].dot(zeroAccSpatFrc[i+1])
-			- spatCoriolisAcc[i].dot(hDof);
-
+			Y[m_links[i].m_dofOffset + dof] = m_links[i].m_jointTorque[dof] - m_links[i].m_axes[dof].dot(zeroAccSpatFrc[i + 1]) - spatCoriolisAcc[i].dot(hDof);
 		}
- 		for(int dof = 0; dof < m_links[i].m_dofCount; ++dof)
-		{	
+		for (int dof = 0; dof < m_links[i].m_dofCount; ++dof)
+		{
 			btScalar *D_row = &D[dof * m_links[i].m_dofCount];
-			for(int dof2 = 0; dof2 < m_links[i].m_dofCount; ++dof2)
+			for (int dof2 = 0; dof2 < m_links[i].m_dofCount; ++dof2)
 			{
 				const btSpatialForceVector &hDof2 = h[m_links[i].m_dofOffset + dof2];
 				D_row[dof2] = m_links[i].m_axes[dof].dot(hDof2);
 			}
 		}
 
-        btScalar *invDi = &invD[m_links[i].m_dofOffset*m_links[i].m_dofOffset];
-		switch(m_links[i].m_jointType)
+		btScalar *invDi = &invD[m_links[i].m_dofOffset * m_links[i].m_dofOffset];
+		switch (m_links[i].m_jointType)
 		{
 			case btMultibodyLink::ePrismatic:
 			case btMultibodyLink::eRevolute:
 			{
-				if (D[0]>=SIMD_EPSILON)
+				if (D[0] >= SIMD_EPSILON)
 				{
 					invDi[0] = 1.0f / D[0];
-				} else
+				}
+				else
 				{
 					invDi[0] = 0;
 				}
@@ -1002,10 +1012,10 @@ void btMultiBody::computeAccelerationsArticulatedBodyAlgorithmMultiDof(btScalar
 				const btMatrix3x3 invD3x3(D3x3.inverse());
 
 				//unroll the loop?
-				for(int row = 0; row < 3; ++row)
+				for (int row = 0; row < 3; ++row)
 				{
-					for(int col = 0; col < 3; ++col)
-					{						
+					for (int col = 0; col < 3; ++col)
+					{
 						invDi[row * 3 + col] = invD3x3[row][col];
 					}
 				}
@@ -1014,86 +1024,82 @@ void btMultiBody::computeAccelerationsArticulatedBodyAlgorithmMultiDof(btScalar
 			}
 			default:
 			{
-			
 			}
 		}
 
 		//determine h*D^{-1}
-		for(int dof = 0; dof < m_links[i].m_dofCount; ++dof)
+		for (int dof = 0; dof < m_links[i].m_dofCount; ++dof)
 		{
 			spatForceVecTemps[dof].setZero();
 
-			for(int dof2 = 0; dof2 < m_links[i].m_dofCount; ++dof2)
-			{				
+			for (int dof2 = 0; dof2 < m_links[i].m_dofCount; ++dof2)
+			{
 				const btSpatialForceVector &hDof2 = h[m_links[i].m_dofOffset + dof2];
-				//				
+				//
 				spatForceVecTemps[dof] += hDof2 * invDi[dof2 * m_links[i].m_dofCount + dof];
 			}
 		}
 
-		dyadTemp = spatInertia[i+1];
+		dyadTemp = spatInertia[i + 1];
 
 		//determine (h*D^{-1}) * h^{T}
-		for(int dof = 0; dof < m_links[i].m_dofCount; ++dof)
-		{			
+		for (int dof = 0; dof < m_links[i].m_dofCount; ++dof)
+		{
 			const btSpatialForceVector &hDof = h[m_links[i].m_dofOffset + dof];
 			//
 			dyadTemp -= symmetricSpatialOuterProduct(hDof, spatForceVecTemps[dof]);
 		}
 
-		fromParent.transformInverse(dyadTemp, spatInertia[parent+1], btSpatialTransformationMatrix::Add);
-        
-		for(int dof = 0; dof < m_links[i].m_dofCount; ++dof)
+		fromParent.transformInverse(dyadTemp, spatInertia[parent + 1], btSpatialTransformationMatrix::Add);
+
+		for (int dof = 0; dof < m_links[i].m_dofCount; ++dof)
 		{
 			invD_times_Y[dof] = 0.f;
 
-			for(int dof2 = 0; dof2 < m_links[i].m_dofCount; ++dof2)
+			for (int dof2 = 0; dof2 < m_links[i].m_dofCount; ++dof2)
 			{
-				invD_times_Y[dof] += invDi[dof * m_links[i].m_dofCount + dof2] * Y[m_links[i].m_dofOffset + dof2];				
-			}	
+				invD_times_Y[dof] += invDi[dof * m_links[i].m_dofCount + dof2] * Y[m_links[i].m_dofOffset + dof2];
+			}
 		}
-		
-		spatForceVecTemps[0] = zeroAccSpatFrc[i+1] + spatInertia[i+1] * spatCoriolisAcc[i];		
 
-		for(int dof = 0; dof < m_links[i].m_dofCount; ++dof)
-		{				
+		spatForceVecTemps[0] = zeroAccSpatFrc[i + 1] + spatInertia[i + 1] * spatCoriolisAcc[i];
+
+		for (int dof = 0; dof < m_links[i].m_dofCount; ++dof)
+		{
 			const btSpatialForceVector &hDof = h[m_links[i].m_dofOffset + dof];
 			//
-			spatForceVecTemps[0] += hDof * invD_times_Y[dof];			
+			spatForceVecTemps[0] += hDof * invD_times_Y[dof];
 		}
-		
+
 		fromParent.transformInverse(spatForceVecTemps[0], spatForceVecTemps[1]);
-		
-		zeroAccSpatFrc[parent+1] += spatForceVecTemps[1];
-    }
 
+		zeroAccSpatFrc[parent + 1] += spatForceVecTemps[1];
+	}
 
-    // Second 'upward' loop
-    // (part of TreeForwardDynamics in Mirtich)
+	// Second 'upward' loop
+	// (part of TreeForwardDynamics in Mirtich)
 
-    if (m_fixedBase) 
+	if (m_fixedBase)
 	{
-        spatAcc[0].setZero();
-    } 
-	else 
+		spatAcc[0].setZero();
+	}
+	else
 	{
-        if (num_links > 0) 
+		if (num_links > 0)
 		{
 			m_cachedInertiaValid = true;
 			m_cachedInertiaTopLeft = spatInertia[0].m_topLeftMat;
 			m_cachedInertiaTopRight = spatInertia[0].m_topRightMat;
 			m_cachedInertiaLowerLeft = spatInertia[0].m_bottomLeftMat;
-			m_cachedInertiaLowerRight= spatInertia[0].m_topLeftMat.transpose();
+			m_cachedInertiaLowerRight = spatInertia[0].m_topLeftMat.transpose();
+		}
 
-        }		
-		
 		solveImatrix(zeroAccSpatFrc[0], result);
 		spatAcc[0] = -result;
-    }
-	
-	
-    // now do the loop over the m_links
-    for (int i = 0; i < num_links; ++i) 
+	}
+
+	// now do the loop over the m_links
+	for (int i = 0; i < num_links; ++i)
 	{
 		//	qdd = D^{-1} * (Y - h^{T}*apar) = (S^{T}*I*S)^{-1} * (tau - S^{T}*I*cor - S^{T}*zeroAccFrc - S^{T}*I*apar)
 		//	a = apar + cor + Sqdd
@@ -1101,73 +1107,73 @@ void btMultiBody::computeAccelerationsArticulatedBodyAlgorithmMultiDof(btScalar
 		//	qdd = D^{-1} * (Y - h^{T}*(apar+cor))
 		//	a = apar + Sqdd
 
-        const int parent = m_links[i].m_parent;
-		fromParent.m_rotMat = rot_from_parent[i+1]; fromParent.m_trnVec = m_links[i].m_cachedRVector;
+		const int parent = m_links[i].m_parent;
+		fromParent.m_rotMat = rot_from_parent[i + 1];
+		fromParent.m_trnVec = m_links[i].m_cachedRVector;
 
-		fromParent.transform(spatAcc[parent+1], spatAcc[i+1]);
-		
-		for(int dof = 0; dof < m_links[i].m_dofCount; ++dof)
+		fromParent.transform(spatAcc[parent + 1], spatAcc[i + 1]);
+
+		for (int dof = 0; dof < m_links[i].m_dofCount; ++dof)
 		{
 			const btSpatialForceVector &hDof = h[m_links[i].m_dofOffset + dof];
-			//			
-			Y_minus_hT_a[dof] = Y[m_links[i].m_dofOffset + dof] - spatAcc[i+1].dot(hDof);			
+			//
+			Y_minus_hT_a[dof] = Y[m_links[i].m_dofOffset + dof] - spatAcc[i + 1].dot(hDof);
 		}
 
-		btScalar *invDi = &invD[m_links[i].m_dofOffset*m_links[i].m_dofOffset];
+		btScalar *invDi = &invD[m_links[i].m_dofOffset * m_links[i].m_dofOffset];
 		//D^{-1} * (Y - h^{T}*apar)
 		mulMatrix(invDi, Y_minus_hT_a, m_links[i].m_dofCount, m_links[i].m_dofCount, m_links[i].m_dofCount, 1, &joint_accel[m_links[i].m_dofOffset]);
 
-		spatAcc[i+1] += spatCoriolisAcc[i];		
+		spatAcc[i + 1] += spatCoriolisAcc[i];
 
-		for(int dof = 0; dof < m_links[i].m_dofCount; ++dof)		
-			spatAcc[i+1] += m_links[i].m_axes[dof] * joint_accel[m_links[i].m_dofOffset + dof];
+		for (int dof = 0; dof < m_links[i].m_dofCount; ++dof)
+			spatAcc[i + 1] += m_links[i].m_axes[dof] * joint_accel[m_links[i].m_dofOffset + dof];
 
 		if (m_links[i].m_jointFeedback)
 		{
 			m_internalNeedsJointFeedback = true;
 
-			btVector3 angularBotVec = (spatInertia[i+1]*spatAcc[i+1]+zeroAccSpatFrc[i+1]).m_bottomVec;
-			btVector3 linearTopVec = (spatInertia[i+1]*spatAcc[i+1]+zeroAccSpatFrc[i+1]).m_topVec;
+			btVector3 angularBotVec = (spatInertia[i + 1] * spatAcc[i + 1] + zeroAccSpatFrc[i + 1]).m_bottomVec;
+			btVector3 linearTopVec = (spatInertia[i + 1] * spatAcc[i + 1] + zeroAccSpatFrc[i + 1]).m_topVec;
 
-			if (gJointFeedbackInJointFrame)
+			if (jointFeedbackInJointFrame)
 			{
 				//shift the reaction forces to the joint frame
 				//linear (force) component is the same
 				//shift the angular (torque, moment) component using the relative position,  m_links[i].m_dVector
-				 angularBotVec = angularBotVec - linearTopVec.cross(m_links[i].m_dVector);
+				angularBotVec = angularBotVec - linearTopVec.cross(m_links[i].m_dVector);
 			}
-			
 
-			if (gJointFeedbackInWorldSpace)
+			if (jointFeedbackInWorldSpace)
 			{
 				if (isConstraintPass)
 				{
- m_links[i].m_jointFeedback->m_reactionForces.m_bottomVec += m_links[i].m_cachedWorldTransform.getBasis()*angularBotVec;
-                                        m_links[i].m_jointFeedback->m_reactionForces.m_topVec += m_links[i].m_cachedWorldTransform.getBasis()*linearTopVec;
-				} else
-				{
-					m_links[i].m_jointFeedback->m_reactionForces.m_bottomVec = m_links[i].m_cachedWorldTransform.getBasis()*angularBotVec;
-					m_links[i].m_jointFeedback->m_reactionForces.m_topVec = m_links[i].m_cachedWorldTransform.getBasis()*linearTopVec;
-				}
-			} else
-			{
-				if (isConstraintPass)
-				{
-					  m_links[i].m_jointFeedback->m_reactionForces.m_bottomVec += angularBotVec;                        
-                                m_links[i].m_jointFeedback->m_reactionForces.m_topVec += linearTopVec;
-
+					m_links[i].m_jointFeedback->m_reactionForces.m_bottomVec += m_links[i].m_cachedWorldTransform.getBasis() * angularBotVec;
+					m_links[i].m_jointFeedback->m_reactionForces.m_topVec += m_links[i].m_cachedWorldTransform.getBasis() * linearTopVec;
 				}
 				else
 				{
-				m_links[i].m_jointFeedback->m_reactionForces.m_bottomVec = angularBotVec;
-				m_links[i].m_jointFeedback->m_reactionForces.m_topVec = linearTopVec;
-				}		
-			}	
+					m_links[i].m_jointFeedback->m_reactionForces.m_bottomVec = m_links[i].m_cachedWorldTransform.getBasis() * angularBotVec;
+					m_links[i].m_jointFeedback->m_reactionForces.m_topVec = m_links[i].m_cachedWorldTransform.getBasis() * linearTopVec;
+				}
+			}
+			else
+			{
+				if (isConstraintPass)
+				{
+					m_links[i].m_jointFeedback->m_reactionForces.m_bottomVec += angularBotVec;
+					m_links[i].m_jointFeedback->m_reactionForces.m_topVec += linearTopVec;
+				}
+				else
+				{
+					m_links[i].m_jointFeedback->m_reactionForces.m_bottomVec = angularBotVec;
+					m_links[i].m_jointFeedback->m_reactionForces.m_topVec = linearTopVec;
+				}
+			}
+		}
 	}
 
-    }
-
-    // transform base accelerations back to the world frame.
+	// transform base accelerations back to the world frame.
 	const btVector3 omegadot_out = rot_from_parent[0].transpose() * spatAcc[0].getAngular();
 	output[0] = omegadot_out[0];
 	output[1] = omegadot_out[1];
@@ -1196,26 +1202,25 @@ void btMultiBody::computeAccelerationsArticulatedBodyAlgorithmMultiDof(btScalar
 	//printf("]\n");
 	/////////////////
 
-    // Final step: add the accelerations (times dt) to the velocities.
+	// Final step: add the accelerations (times dt) to the velocities.
 
 	if (!isConstraintPass)
 	{
-	if(dt > 0.)
-		applyDeltaVeeMultiDof(output, dt);
-
+		if (dt > 0.)
+			applyDeltaVeeMultiDof(output, dt);
 	}
 	/////
 	//btScalar angularThres = 1;
-	//btScalar maxAngVel = 0.;		
+	//btScalar maxAngVel = 0.;
 	//bool scaleDown = 1.;
 	//for(int link = 0; link < m_links.size(); ++link)
-	//{		
+	//{
 	//	if(spatVel[link+1].getAngular().length() > maxAngVel)
 	//	{
 	//		maxAngVel = spatVel[link+1].getAngular().length();
 	//		scaleDown = angularThres / spatVel[link+1].getAngular().length();
 	//		break;
-	//	}		
+	//	}
 	//}
 
 	//if(scaleDown != 1.)
@@ -1232,77 +1237,77 @@ void btMultiBody::computeAccelerationsArticulatedBodyAlgorithmMultiDof(btScalar
 	/////
 
 	/////////////////////
-	if(m_useGlobalVelocities)
+	if (m_useGlobalVelocities)
 	{
-		for (int i = 0; i < num_links; ++i) 
+		for (int i = 0; i < num_links; ++i)
 		{
 			const int parent = m_links[i].m_parent;
 			//rot_from_parent[i+1] = btMatrix3x3(m_links[i].m_cachedRotParentToThis);    /// <- done
 			//rot_from_world[i+1] = rot_from_parent[i+1] * rot_from_world[parent+1];		/// <- done
-		
-			fromParent.m_rotMat = rot_from_parent[i+1]; fromParent.m_trnVec = m_links[i].m_cachedRVector;
-			fromWorld.m_rotMat = rot_from_world[i+1];			
-        
-			// vhat_i = i_xhat_p(i) * vhat_p(i)		
-			fromParent.transform(spatVel[parent+1], spatVel[i+1]);
+
+			fromParent.m_rotMat = rot_from_parent[i + 1];
+			fromParent.m_trnVec = m_links[i].m_cachedRVector;
+			fromWorld.m_rotMat = rot_from_world[i + 1];
+
+			// vhat_i = i_xhat_p(i) * vhat_p(i)
+			fromParent.transform(spatVel[parent + 1], spatVel[i + 1]);
 			//nice alternative below (using operator *) but it generates temps
 			/////////////////////////////////////////////////////////////
 
 			// now set vhat_i to its true value by doing
-			// vhat_i += qidot * shat_i			
+			// vhat_i += qidot * shat_i
 			spatJointVel.setZero();
 
-			for(int dof = 0; dof < m_links[i].m_dofCount; ++dof)		
+			for (int dof = 0; dof < m_links[i].m_dofCount; ++dof)
 				spatJointVel += m_links[i].m_axes[dof] * getJointVelMultiDof(i)[dof];
-		
+
 			// remember vhat_i is really vhat_p(i) (but in current frame) at this point	=> we need to add velocity across the inboard joint
-			spatVel[i+1] += spatJointVel;
+			spatVel[i + 1] += spatJointVel;
 
-
-			fromWorld.transformInverseRotationOnly(spatVel[i+1], m_links[i].m_absFrameTotVelocity);
+			fromWorld.transformInverseRotationOnly(spatVel[i + 1], m_links[i].m_absFrameTotVelocity);
 			fromWorld.transformInverseRotationOnly(spatJointVel, m_links[i].m_absFrameLocVelocity);
 		}
 	}
-	
 }
 
-
-
-void btMultiBody::solveImatrix(const btVector3& rhs_top, const btVector3& rhs_bot, btScalar result[6]) const
+void btMultiBody::solveImatrix(const btVector3 &rhs_top, const btVector3 &rhs_bot, btScalar result[6]) const
 {
 	int num_links = getNumLinks();
 	///solve I * x = rhs, so the result = invI * rhs
-    if (num_links == 0) 
+	if (num_links == 0)
 	{
 		// in the case of 0 m_links (i.e. a plain rigid body, not a multibody) rhs * invI is easier
-   
-	if ((m_baseInertia[0] >= SIMD_EPSILON) && (m_baseInertia[1] >= SIMD_EPSILON) && (m_baseInertia[2] >= SIMD_EPSILON))
-	{
-		result[0] = rhs_bot[0] / m_baseInertia[0];
-		result[1] = rhs_bot[1] / m_baseInertia[1];
-        result[2] = rhs_bot[2] / m_baseInertia[2];
-	} else
-	{
-		result[0] = 0;
-		result[1] = 0;
-		result[2] = 0;
+
+		if ((m_baseInertia[0] >= SIMD_EPSILON) && (m_baseInertia[1] >= SIMD_EPSILON) && (m_baseInertia[2] >= SIMD_EPSILON))
+		{
+			result[0] = rhs_bot[0] / m_baseInertia[0];
+			result[1] = rhs_bot[1] / m_baseInertia[1];
+			result[2] = rhs_bot[2] / m_baseInertia[2];
+		}
+		else
+		{
+			result[0] = 0;
+			result[1] = 0;
+			result[2] = 0;
+		}
+		if (m_baseMass >= SIMD_EPSILON)
+		{
+			result[3] = rhs_top[0] / m_baseMass;
+			result[4] = rhs_top[1] / m_baseMass;
+			result[5] = rhs_top[2] / m_baseMass;
+		}
+		else
+		{
+			result[3] = 0;
+			result[4] = 0;
+			result[5] = 0;
+		}
 	}
-	if (m_baseMass>=SIMD_EPSILON)
-	{
-        	result[3] = rhs_top[0] / m_baseMass;
-        	result[4] = rhs_top[1] / m_baseMass;
-        	result[5] = rhs_top[2] / m_baseMass;
-	} else
-	{
-		result[3] = 0;
-		result[4] = 0;
-		result[5] = 0;
-	}
-    } else 
+	else
 	{
 		if (!m_cachedInertiaValid)
 		{
-			for (int i=0;i<6;i++)
+			for (int i = 0; i < 6; i++)
 			{
 				result[i] = 0.f;
 			}
@@ -1310,94 +1315,95 @@ void btMultiBody::solveImatrix(const btVector3& rhs_top, const btVector3& rhs_bo
 		}
 		/// Special routine for calculating the inverse of a spatial inertia matrix
 		///the 6x6 matrix is stored as 4 blocks of 3x3 matrices
-		btMatrix3x3 Binv = m_cachedInertiaTopRight.inverse()*-1.f;
+		btMatrix3x3 Binv = m_cachedInertiaTopRight.inverse() * -1.f;
 		btMatrix3x3 tmp = m_cachedInertiaLowerRight * Binv;
 		btMatrix3x3 invIupper_right = (tmp * m_cachedInertiaTopLeft + m_cachedInertiaLowerLeft).inverse();
 		tmp = invIupper_right * m_cachedInertiaLowerRight;
 		btMatrix3x3 invI_upper_left = (tmp * Binv);
 		btMatrix3x3 invI_lower_right = (invI_upper_left).transpose();
-		tmp = m_cachedInertiaTopLeft  * invI_upper_left;
-		tmp[0][0]-= 1.0;
-		tmp[1][1]-= 1.0;
-		tmp[2][2]-= 1.0;
+		tmp = m_cachedInertiaTopLeft * invI_upper_left;
+		tmp[0][0] -= 1.0;
+		tmp[1][1] -= 1.0;
+		tmp[2][2] -= 1.0;
 		btMatrix3x3 invI_lower_left = (Binv * tmp);
 
 		//multiply result = invI * rhs
 		{
-		  btVector3 vtop = invI_upper_left*rhs_top;
-		  btVector3 tmp;
-		  tmp = invIupper_right * rhs_bot;
-		  vtop += tmp;
-		  btVector3 vbot = invI_lower_left*rhs_top;
-		  tmp = invI_lower_right * rhs_bot;
-		  vbot += tmp;
-		  result[0] = vtop[0];
-		  result[1] = vtop[1];
-		  result[2] = vtop[2];
-		  result[3] = vbot[0];
-		  result[4] = vbot[1];
-		  result[5] = vbot[2];
+			btVector3 vtop = invI_upper_left * rhs_top;
+			btVector3 tmp;
+			tmp = invIupper_right * rhs_bot;
+			vtop += tmp;
+			btVector3 vbot = invI_lower_left * rhs_top;
+			tmp = invI_lower_right * rhs_bot;
+			vbot += tmp;
+			result[0] = vtop[0];
+			result[1] = vtop[1];
+			result[2] = vtop[2];
+			result[3] = vbot[0];
+			result[4] = vbot[1];
+			result[5] = vbot[2];
 		}
-		
-    }
+	}
 }
 void btMultiBody::solveImatrix(const btSpatialForceVector &rhs, btSpatialMotionVector &result) const
 {
 	int num_links = getNumLinks();
 	///solve I * x = rhs, so the result = invI * rhs
-    if (num_links == 0) 
+	if (num_links == 0)
 	{
 		// in the case of 0 m_links (i.e. a plain rigid body, not a multibody) rhs * invI is easier
 		if ((m_baseInertia[0] >= SIMD_EPSILON) && (m_baseInertia[1] >= SIMD_EPSILON) && (m_baseInertia[2] >= SIMD_EPSILON))
-        	{
-			result.setAngular(rhs.getAngular() / m_baseInertia);       
-        	} else  
-        	{       
-                	result.setAngular(btVector3(0,0,0));
-        	}
-        	if (m_baseMass>=SIMD_EPSILON)
-        	{       
-			result.setLinear(rhs.getLinear() / m_baseMass);               	 	
-        	} else  
-        	{       
-                	result.setLinear(btVector3(0,0,0));
-        	}
-    	} else 
+		{
+			result.setAngular(rhs.getAngular() / m_baseInertia);
+		}
+		else
+		{
+			result.setAngular(btVector3(0, 0, 0));
+		}
+		if (m_baseMass >= SIMD_EPSILON)
+		{
+			result.setLinear(rhs.getLinear() / m_baseMass);
+		}
+		else
+		{
+			result.setLinear(btVector3(0, 0, 0));
+		}
+	}
+	else
 	{
 		/// Special routine for calculating the inverse of a spatial inertia matrix
 		///the 6x6 matrix is stored as 4 blocks of 3x3 matrices
 		if (!m_cachedInertiaValid)
 		{
-			result.setLinear(btVector3(0,0,0));
-			result.setAngular(btVector3(0,0,0));
-			result.setVector(btVector3(0,0,0),btVector3(0,0,0));
+			result.setLinear(btVector3(0, 0, 0));
+			result.setAngular(btVector3(0, 0, 0));
+			result.setVector(btVector3(0, 0, 0), btVector3(0, 0, 0));
 			return;
 		}
-		btMatrix3x3 Binv = m_cachedInertiaTopRight.inverse()*-1.f;
+		btMatrix3x3 Binv = m_cachedInertiaTopRight.inverse() * -1.f;
 		btMatrix3x3 tmp = m_cachedInertiaLowerRight * Binv;
 		btMatrix3x3 invIupper_right = (tmp * m_cachedInertiaTopLeft + m_cachedInertiaLowerLeft).inverse();
 		tmp = invIupper_right * m_cachedInertiaLowerRight;
 		btMatrix3x3 invI_upper_left = (tmp * Binv);
 		btMatrix3x3 invI_lower_right = (invI_upper_left).transpose();
-		tmp = m_cachedInertiaTopLeft  * invI_upper_left;
-		tmp[0][0]-= 1.0;
-		tmp[1][1]-= 1.0;
-		tmp[2][2]-= 1.0;
+		tmp = m_cachedInertiaTopLeft * invI_upper_left;
+		tmp[0][0] -= 1.0;
+		tmp[1][1] -= 1.0;
+		tmp[2][2] -= 1.0;
 		btMatrix3x3 invI_lower_left = (Binv * tmp);
 
 		//multiply result = invI * rhs
 		{
-		  btVector3 vtop = invI_upper_left*rhs.getLinear();
-		  btVector3 tmp;
-		  tmp = invIupper_right * rhs.getAngular();
-		  vtop += tmp;
-		  btVector3 vbot = invI_lower_left*rhs.getLinear();
-		  tmp = invI_lower_right * rhs.getAngular();
-		  vbot += tmp;
-		  result.setVector(vtop, vbot);		  
+			btVector3 vtop = invI_upper_left * rhs.getLinear();
+			btVector3 tmp;
+			tmp = invIupper_right * rhs.getAngular();
+			vtop += tmp;
+			btVector3 vbot = invI_lower_left * rhs.getLinear();
+			tmp = invI_lower_right * rhs.getAngular();
+			vbot += tmp;
+			result.setVector(vtop, vbot);
 		}
-		
-    }
+	}
 }
 
 void btMultiBody::mulMatrix(btScalar *pA, btScalar *pB, int rowsA, int colsA, int rowsB, int colsB, btScalar *pC) const
@@ -1416,155 +1422,152 @@ void btMultiBody::mulMatrix(btScalar *pA, btScalar *pB, int rowsA, int colsA, in
 }
 
 void btMultiBody::calcAccelerationDeltasMultiDof(const btScalar *force, btScalar *output,
-                                       btAlignedObjectArray<btScalar> &scratch_r, btAlignedObjectArray<btVector3> &scratch_v) const
+												 btAlignedObjectArray<btScalar> &scratch_r, btAlignedObjectArray<btVector3> &scratch_v) const
 {
-    // Temporary matrices/vectors -- use scratch space from caller
-    // so that we don't have to keep reallocating every frame
+	// Temporary matrices/vectors -- use scratch space from caller
+	// so that we don't have to keep reallocating every frame
 
-	
-	int num_links = getNumLinks();	
-    scratch_r.resize(m_dofCount);
-    scratch_v.resize(4*num_links + 4);	    
+	int num_links = getNumLinks();
+	scratch_r.resize(m_dofCount);
+	scratch_v.resize(4 * num_links + 4);
 
-    btScalar * r_ptr = m_dofCount ? &scratch_r[0] : 0;
-    btVector3 * v_ptr = &scratch_v[0];
+	btScalar *r_ptr = m_dofCount ? &scratch_r[0] : 0;
+	btVector3 *v_ptr = &scratch_v[0];
 
-    // zhat_i^A (scratch space)
-    btSpatialForceVector * zeroAccSpatFrc = (btSpatialForceVector *)v_ptr;
+	// zhat_i^A (scratch space)
+	btSpatialForceVector *zeroAccSpatFrc = (btSpatialForceVector *)v_ptr;
 	v_ptr += num_links * 2 + 2;
 
-    // rot_from_parent (cached from calcAccelerations)
-    const btMatrix3x3 * rot_from_parent = &m_matrixBuf[0];
+	// rot_from_parent (cached from calcAccelerations)
+	const btMatrix3x3 *rot_from_parent = &m_matrixBuf[0];
 
-    // hhat (cached), accel (scratch)
-    // hhat is NOT stored for the base (but ahat is) 
-	const btSpatialForceVector * h = (btSpatialForceVector *)(m_dofCount > 0 ? &m_vectorBuf[0] : 0);
-	btSpatialMotionVector * spatAcc = (btSpatialMotionVector *)v_ptr;
+	// hhat (cached), accel (scratch)
+	// hhat is NOT stored for the base (but ahat is)
+	const btSpatialForceVector *h = (btSpatialForceVector *)(m_dofCount > 0 ? &m_vectorBuf[0] : 0);
+	btSpatialMotionVector *spatAcc = (btSpatialMotionVector *)v_ptr;
 	v_ptr += num_links * 2 + 2;
 
-    // Y_i (scratch), invD_i (cached)
-    const btScalar * invD = m_dofCount > 0 ? &m_realBuf[6 + m_dofCount] : 0;
-	btScalar * Y = r_ptr; 
+	// Y_i (scratch), invD_i (cached)
+	const btScalar *invD = m_dofCount > 0 ? &m_realBuf[6 + m_dofCount] : 0;
+	btScalar *Y = r_ptr;
 	////////////////
 	//aux variables
-	btScalar invD_times_Y[6];							//D^{-1} * Y [dofxdof x dofx1 = dofx1] <=> D^{-1} * u; better moved to buffers since it is recalced in calcAccelerationDeltasMultiDof; num_dof of btScalar would cover all bodies
-	btSpatialMotionVector result;							//holds results of the SolveImatrix op; it is a spatial motion vector (accel)
-	btScalar Y_minus_hT_a[6];							//Y - h^{T} * a; it's dofx1 for each body so a single 6x1 temp is enough	
-	btSpatialForceVector spatForceVecTemps[6];				//6 temporary spatial force vectors
-	btSpatialTransformationMatrix fromParent;	
+	btScalar invD_times_Y[6];                   //D^{-1} * Y [dofxdof x dofx1 = dofx1] <=> D^{-1} * u; better moved to buffers since it is recalced in calcAccelerationDeltasMultiDof; num_dof of btScalar would cover all bodies
+	btSpatialMotionVector result;               //holds results of the SolveImatrix op; it is a spatial motion vector (accel)
+	btScalar Y_minus_hT_a[6];                   //Y - h^{T} * a; it's dofx1 for each body so a single 6x1 temp is enough
+	btSpatialForceVector spatForceVecTemps[6];  //6 temporary spatial force vectors
+	btSpatialTransformationMatrix fromParent;
 	/////////////////
 
-    // First 'upward' loop.
-    // Combines CompTreeLinkVelocities and InitTreeLinks from Mirtich.
-	
+	// First 'upward' loop.
+	// Combines CompTreeLinkVelocities and InitTreeLinks from Mirtich.
+
 	// Fill in zero_acc
-    // -- set to force/torque on the base, zero otherwise
-    if (m_fixedBase) 
+	// -- set to force/torque on the base, zero otherwise
+	if (m_fixedBase)
+	{
+		zeroAccSpatFrc[0].setZero();
+	}
+	else
 	{
-        zeroAccSpatFrc[0].setZero();
-    } else 
-	{	
 		//test forces
 		fromParent.m_rotMat = rot_from_parent[0];
-		fromParent.transformRotationOnly(btSpatialForceVector(-force[0],-force[1],-force[2], -force[3],-force[4],-force[5]), zeroAccSpatFrc[0]);
-    }
-    for (int i = 0; i < num_links; ++i) 
+		fromParent.transformRotationOnly(btSpatialForceVector(-force[0], -force[1], -force[2], -force[3], -force[4], -force[5]), zeroAccSpatFrc[0]);
+	}
+	for (int i = 0; i < num_links; ++i)
 	{
-		zeroAccSpatFrc[i+1].setZero();
-    }    
+		zeroAccSpatFrc[i + 1].setZero();
+	}
 
 	// 'Downward' loop.
-    // (part of TreeForwardDynamics in Mirtich.)
-    for (int i = num_links - 1; i >= 0; --i)
+	// (part of TreeForwardDynamics in Mirtich.)
+	for (int i = num_links - 1; i >= 0; --i)
 	{
 		const int parent = m_links[i].m_parent;
-		fromParent.m_rotMat = rot_from_parent[i+1]; fromParent.m_trnVec = m_links[i].m_cachedRVector;
+		fromParent.m_rotMat = rot_from_parent[i + 1];
+		fromParent.m_trnVec = m_links[i].m_cachedRVector;
 
-		for(int dof = 0; dof < m_links[i].m_dofCount; ++dof)
+		for (int dof = 0; dof < m_links[i].m_dofCount; ++dof)
 		{
-			Y[m_links[i].m_dofOffset + dof] = force[6 + m_links[i].m_dofOffset + dof]
-											- m_links[i].m_axes[dof].dot(zeroAccSpatFrc[i+1])
-											;
+			Y[m_links[i].m_dofOffset + dof] = force[6 + m_links[i].m_dofOffset + dof] - m_links[i].m_axes[dof].dot(zeroAccSpatFrc[i + 1]);
 		}
 
 		btVector3 in_top, in_bottom, out_top, out_bottom;
-		const btScalar *invDi = &invD[m_links[i].m_dofOffset*m_links[i].m_dofOffset];
-		
-		for(int dof = 0; dof < m_links[i].m_dofCount; ++dof)
+		const btScalar *invDi = &invD[m_links[i].m_dofOffset * m_links[i].m_dofOffset];
+
+		for (int dof = 0; dof < m_links[i].m_dofCount; ++dof)
 		{
 			invD_times_Y[dof] = 0.f;
 
-			for(int dof2 = 0; dof2 < m_links[i].m_dofCount; ++dof2)
+			for (int dof2 = 0; dof2 < m_links[i].m_dofCount; ++dof2)
 			{
-				invD_times_Y[dof] += invDi[dof * m_links[i].m_dofCount + dof2] * Y[m_links[i].m_dofOffset + dof2];				
-			}	
+				invD_times_Y[dof] += invDi[dof * m_links[i].m_dofCount + dof2] * Y[m_links[i].m_dofOffset + dof2];
+			}
 		}
-		
-		 // Zp += pXi * (Zi + hi*Yi/Di)
-		spatForceVecTemps[0] = zeroAccSpatFrc[i+1];
 
-		for(int dof = 0; dof < m_links[i].m_dofCount; ++dof)
+		// Zp += pXi * (Zi + hi*Yi/Di)
+		spatForceVecTemps[0] = zeroAccSpatFrc[i + 1];
+
+		for (int dof = 0; dof < m_links[i].m_dofCount; ++dof)
 		{
 			const btSpatialForceVector &hDof = h[m_links[i].m_dofOffset + dof];
 			//
-			spatForceVecTemps[0] += hDof * invD_times_Y[dof];		
+			spatForceVecTemps[0] += hDof * invD_times_Y[dof];
 		}
-		
 
 		fromParent.transformInverse(spatForceVecTemps[0], spatForceVecTemps[1]);
-			
-		zeroAccSpatFrc[parent+1] += spatForceVecTemps[1];
-    }
-	
+
+		zeroAccSpatFrc[parent + 1] += spatForceVecTemps[1];
+	}
+
 	// ptr to the joint accel part of the output
-    btScalar * joint_accel = output + 6;
+	btScalar *joint_accel = output + 6;
 
+	// Second 'upward' loop
+	// (part of TreeForwardDynamics in Mirtich)
 
-    // Second 'upward' loop
-    // (part of TreeForwardDynamics in Mirtich)
-
-    if (m_fixedBase) 
+	if (m_fixedBase)
 	{
-        spatAcc[0].setZero();
-    } 
-	else 
+		spatAcc[0].setZero();
+	}
+	else
 	{
 		solveImatrix(zeroAccSpatFrc[0], result);
 		spatAcc[0] = -result;
+	}
 
-    }
-	
-    // now do the loop over the m_links
-    for (int i = 0; i < num_links; ++i)
+	// now do the loop over the m_links
+	for (int i = 0; i < num_links; ++i)
 	{
-        const int parent = m_links[i].m_parent;
-		fromParent.m_rotMat = rot_from_parent[i+1]; fromParent.m_trnVec = m_links[i].m_cachedRVector;
+		const int parent = m_links[i].m_parent;
+		fromParent.m_rotMat = rot_from_parent[i + 1];
+		fromParent.m_trnVec = m_links[i].m_cachedRVector;
 
-		fromParent.transform(spatAcc[parent+1], spatAcc[i+1]);
-		
-		for(int dof = 0; dof < m_links[i].m_dofCount; ++dof)
+		fromParent.transform(spatAcc[parent + 1], spatAcc[i + 1]);
+
+		for (int dof = 0; dof < m_links[i].m_dofCount; ++dof)
 		{
 			const btSpatialForceVector &hDof = h[m_links[i].m_dofOffset + dof];
-			//			
-			Y_minus_hT_a[dof] = Y[m_links[i].m_dofOffset + dof] - spatAcc[i+1].dot(hDof);
+			//
+			Y_minus_hT_a[dof] = Y[m_links[i].m_dofOffset + dof] - spatAcc[i + 1].dot(hDof);
 		}
 
-		const btScalar *invDi = &invD[m_links[i].m_dofOffset*m_links[i].m_dofOffset];
-		mulMatrix(const_cast<btScalar*>(invDi), Y_minus_hT_a, m_links[i].m_dofCount, m_links[i].m_dofCount, m_links[i].m_dofCount, 1, &joint_accel[m_links[i].m_dofOffset]);
+		const btScalar *invDi = &invD[m_links[i].m_dofOffset * m_links[i].m_dofOffset];
+		mulMatrix(const_cast<btScalar *>(invDi), Y_minus_hT_a, m_links[i].m_dofCount, m_links[i].m_dofCount, m_links[i].m_dofCount, 1, &joint_accel[m_links[i].m_dofOffset]);
 
-		for(int dof = 0; dof < m_links[i].m_dofCount; ++dof)		
-			spatAcc[i+1] += m_links[i].m_axes[dof] * joint_accel[m_links[i].m_dofOffset + dof];      
-    }
+		for (int dof = 0; dof < m_links[i].m_dofCount; ++dof)
+			spatAcc[i + 1] += m_links[i].m_axes[dof] * joint_accel[m_links[i].m_dofOffset + dof];
+	}
 
-    // transform base accelerations back to the world frame.
-    btVector3 omegadot_out;
-    omegadot_out = rot_from_parent[0].transpose() * spatAcc[0].getAngular();
+	// transform base accelerations back to the world frame.
+	btVector3 omegadot_out;
+	omegadot_out = rot_from_parent[0].transpose() * spatAcc[0].getAngular();
 	output[0] = omegadot_out[0];
 	output[1] = omegadot_out[1];
 	output[2] = omegadot_out[2];
 
-    btVector3 vdot_out;
-    vdot_out = rot_from_parent[0].transpose() * spatAcc[0].getLinear();
+	btVector3 vdot_out;
+	vdot_out = rot_from_parent[0].transpose() * spatAcc[0].getLinear();
 	output[3] = vdot_out[0];
 	output[4] = vdot_out[1];
 	output[5] = vdot_out[2];
@@ -1577,19 +1580,16 @@ void btMultiBody::calcAccelerationDeltasMultiDof(const btScalar *force, btScalar
 	/////////////////
 }
 
-
-
-
 void btMultiBody::stepPositionsMultiDof(btScalar dt, btScalar *pq, btScalar *pqd)
-{	
+{
 	int num_links = getNumLinks();
-    // step position by adding dt * velocity
-	//btVector3 v = getBaseVel();	
-    //m_basePos += dt * v;
+	// step position by adding dt * velocity
+	//btVector3 v = getBaseVel();
+	//m_basePos += dt * v;
 	//
 	btScalar *pBasePos = (pq ? &pq[4] : m_basePos);
-	btScalar *pBaseVel = (pqd ? &pqd[3] : &m_realBuf[3]);			//note: the !pqd case assumes m_realBuf holds with base velocity at 3,4,5 (should be wrapped for safety)
-	//	
+	btScalar *pBaseVel = (pqd ? &pqd[3] : &m_realBuf[3]);  //note: the !pqd case assumes m_realBuf holds with base velocity at 3,4,5 (should be wrapped for safety)
+	//
 	pBasePos[0] += dt * pBaseVel[0];
 	pBasePos[1] += dt * pBaseVel[1];
 	pBasePos[2] += dt * pBaseVel[2];
@@ -1599,92 +1599,98 @@ void btMultiBody::stepPositionsMultiDof(btScalar dt, btScalar *pq, btScalar *pqd
 	struct
 	{
 		//"exponential map" based on btTransformUtil::integrateTransform(..)
-		void operator() (const btVector3 &omega, btQuaternion &quat, bool baseBody, btScalar dt)
+		void operator()(const btVector3 &omega, btQuaternion &quat, bool baseBody, btScalar dt)
 		{
 			//baseBody	=>	quat is alias and omega is global coor
-			//!baseBody	=>	quat is alibi and omega is local coor	
-			
+			//!baseBody	=>	quat is alibi and omega is local coor
+
 			btVector3 axis;
 			btVector3 angvel;
 
-			if(!baseBody)			
-				angvel = quatRotate(quat, omega);				//if quat is not m_baseQuat, it is alibi => ok			
+			if (!baseBody)
+				angvel = quatRotate(quat, omega);  //if quat is not m_baseQuat, it is alibi => ok
 			else
 				angvel = omega;
-		
-			btScalar fAngle = angvel.length(); 		
+
+			btScalar fAngle = angvel.length();
 			//limit the angular motion
 			if (fAngle * dt > ANGULAR_MOTION_THRESHOLD)
 			{
-				fAngle = btScalar(0.5)*SIMD_HALF_PI / dt;
+				fAngle = btScalar(0.5) * SIMD_HALF_PI / dt;
 			}
 
-			if ( fAngle < btScalar(0.001) )
+			if (fAngle < btScalar(0.001))
 			{
 				// use Taylor's expansions of sync function
-				axis   = angvel*( btScalar(0.5)*dt-(dt*dt*dt)*(btScalar(0.020833333333))*fAngle*fAngle );
+				axis = angvel * (btScalar(0.5) * dt - (dt * dt * dt) * (btScalar(0.020833333333)) * fAngle * fAngle);
 			}
 			else
 			{
 				// sync(fAngle) = sin(c*fAngle)/t
-				axis   = angvel*( btSin(btScalar(0.5)*fAngle*dt)/fAngle );
+				axis = angvel * (btSin(btScalar(0.5) * fAngle * dt) / fAngle);
 			}
-			
-			if(!baseBody)				
-				quat = btQuaternion(axis.x(),axis.y(),axis.z(),btCos( fAngle*dt*btScalar(0.5) )) * quat;			
-			else			
-				quat = quat * btQuaternion(-axis.x(),-axis.y(),-axis.z(),btCos( fAngle*dt*btScalar(0.5) ));
-				//equivalent to: quat = (btQuaternion(axis.x(),axis.y(),axis.z(),btCos( fAngle*dt*btScalar(0.5) )) * quat.inverse()).inverse();			
-		
+
+			if (!baseBody)
+				quat = btQuaternion(axis.x(), axis.y(), axis.z(), btCos(fAngle * dt * btScalar(0.5))) * quat;
+			else
+				quat = quat * btQuaternion(-axis.x(), -axis.y(), -axis.z(), btCos(fAngle * dt * btScalar(0.5)));
+			//equivalent to: quat = (btQuaternion(axis.x(),axis.y(),axis.z(),btCos( fAngle*dt*btScalar(0.5) )) * quat.inverse()).inverse();
+
 			quat.normalize();
 		}
 	} pQuatUpdateFun;
 	///////////////////////////////
 
 	//pQuatUpdateFun(getBaseOmega(), m_baseQuat, true, dt);
-	//	
-	btScalar *pBaseQuat = pq ? pq : m_baseQuat;	
-	btScalar *pBaseOmega = pqd ? pqd : &m_realBuf[0];		//note: the !pqd case assumes m_realBuf starts with base omega (should be wrapped for safety)
 	//
-	btQuaternion baseQuat; baseQuat.setValue(pBaseQuat[0], pBaseQuat[1], pBaseQuat[2], pBaseQuat[3]);
-	btVector3 baseOmega; baseOmega.setValue(pBaseOmega[0], pBaseOmega[1], pBaseOmega[2]);
+	btScalar *pBaseQuat = pq ? pq : m_baseQuat;
+	btScalar *pBaseOmega = pqd ? pqd : &m_realBuf[0];  //note: the !pqd case assumes m_realBuf starts with base omega (should be wrapped for safety)
+	//
+	btQuaternion baseQuat;
+	baseQuat.setValue(pBaseQuat[0], pBaseQuat[1], pBaseQuat[2], pBaseQuat[3]);
+	btVector3 baseOmega;
+	baseOmega.setValue(pBaseOmega[0], pBaseOmega[1], pBaseOmega[2]);
 	pQuatUpdateFun(baseOmega, baseQuat, true, dt);
 	pBaseQuat[0] = baseQuat.x();
 	pBaseQuat[1] = baseQuat.y();
 	pBaseQuat[2] = baseQuat.z();
 	pBaseQuat[3] = baseQuat.w();
 
-
 	//printf("pBaseOmega = %.4f %.4f %.4f\n", pBaseOmega->x(), pBaseOmega->y(), pBaseOmega->z());
 	//printf("pBaseVel = %.4f %.4f %.4f\n", pBaseVel->x(), pBaseVel->y(), pBaseVel->z());
 	//printf("baseQuat = %.4f %.4f %.4f %.4f\n", pBaseQuat->x(), pBaseQuat->y(), pBaseQuat->z(), pBaseQuat->w());
 
-	if(pq)		
+	if (pq)
 		pq += 7;
-	if(pqd)
+	if (pqd)
 		pqd += 6;
 
 	// Finally we can update m_jointPos for each of the m_links
-    for (int i = 0; i < num_links; ++i) 
+	for (int i = 0; i < num_links; ++i)
 	{
-		btScalar *pJointPos = (pq ? pq : &m_links[i].m_jointPos[0]);		
+		btScalar *pJointPos = (pq ? pq : &m_links[i].m_jointPos[0]);
 		btScalar *pJointVel = (pqd ? pqd : getJointVelMultiDof(i));
 
-		switch(m_links[i].m_jointType)
+		switch (m_links[i].m_jointType)
 		{
 			case btMultibodyLink::ePrismatic:
 			case btMultibodyLink::eRevolute:
 			{
-				btScalar jointVel = pJointVel[0];	
+				btScalar jointVel = pJointVel[0];
 				pJointPos[0] += dt * jointVel;
 				break;
 			}
 			case btMultibodyLink::eSpherical:
 			{
-				btVector3 jointVel; jointVel.setValue(pJointVel[0], pJointVel[1], pJointVel[2]);
-				btQuaternion jointOri; jointOri.setValue(pJointPos[0], pJointPos[1], pJointPos[2], pJointPos[3]);
+				btVector3 jointVel;
+				jointVel.setValue(pJointVel[0], pJointVel[1], pJointVel[2]);
+				btQuaternion jointOri;
+				jointOri.setValue(pJointPos[0], pJointPos[1], pJointPos[2], pJointPos[3]);
 				pQuatUpdateFun(jointVel, jointOri, false, dt);
-				pJointPos[0] = jointOri.x(); pJointPos[1] = jointOri.y(); pJointPos[2] = jointOri.z(); pJointPos[3] = jointOri.w();
+				pJointPos[0] = jointOri.x();
+				pJointPos[1] = jointOri.y();
+				pJointPos[2] = jointOri.z();
+				pJointPos[3] = jointOri.w();
 				break;
 			}
 			case btMultibodyLink::ePlanar:
@@ -1701,122 +1707,135 @@ void btMultiBody::stepPositionsMultiDof(btScalar dt, btScalar *pq, btScalar *pqd
 			default:
 			{
 			}
-
 		}
 
 		m_links[i].updateCacheMultiDof(pq);
 
-		if(pq)		
+		if (pq)
 			pq += m_links[i].m_posVarCount;
-		if(pqd)
+		if (pqd)
 			pqd += m_links[i].m_dofCount;
-    }
+	}
 }
 
 void btMultiBody::fillConstraintJacobianMultiDof(int link,
-                                    const btVector3 &contact_point,
-									const btVector3 &normal_ang,
-                                    const btVector3 &normal_lin,
-                                    btScalar *jac,
-                                    btAlignedObjectArray<btScalar> &scratch_r,
-                                    btAlignedObjectArray<btVector3> &scratch_v,
-                                    btAlignedObjectArray<btMatrix3x3> &scratch_m) const
+												 const btVector3 &contact_point,
+												 const btVector3 &normal_ang,
+												 const btVector3 &normal_lin,
+												 btScalar *jac,
+												 btAlignedObjectArray<btScalar> &scratch_r1,
+												 btAlignedObjectArray<btVector3> &scratch_v,
+												 btAlignedObjectArray<btMatrix3x3> &scratch_m) const
 {
-    // temporary space
+	// temporary space
 	int num_links = getNumLinks();
 	int m_dofCount = getNumDofs();
-    scratch_v.resize(3*num_links + 3);			//(num_links + base) offsets + (num_links + base) normals_lin + (num_links + base) normals_ang
-    scratch_m.resize(num_links + 1);
+	scratch_v.resize(3 * num_links + 3);  //(num_links + base) offsets + (num_links + base) normals_lin + (num_links + base) normals_ang
+	scratch_m.resize(num_links + 1);
 
-    btVector3 * v_ptr = &scratch_v[0];
-    btVector3 * p_minus_com_local = v_ptr; v_ptr += num_links + 1;
-    btVector3 * n_local_lin = v_ptr; v_ptr += num_links + 1;
-	btVector3 * n_local_ang = v_ptr; v_ptr += num_links + 1;
-    btAssert(v_ptr - &scratch_v[0] == scratch_v.size());
+	btVector3 *v_ptr = &scratch_v[0];
+	btVector3 *p_minus_com_local = v_ptr;
+	v_ptr += num_links + 1;
+	btVector3 *n_local_lin = v_ptr;
+	v_ptr += num_links + 1;
+	btVector3 *n_local_ang = v_ptr;
+	v_ptr += num_links + 1;
+	btAssert(v_ptr - &scratch_v[0] == scratch_v.size());
 
-    scratch_r.resize(m_dofCount);
-    btScalar * results = m_dofCount > 0 ? &scratch_r[0] : 0;
+	//scratch_r.resize(m_dofCount);
+	//btScalar *results = m_dofCount > 0 ? &scratch_r[0] : 0;
 
-    btMatrix3x3 * rot_from_world = &scratch_m[0];
+    scratch_r1.resize(m_dofCount+num_links);
+    btScalar * results = m_dofCount > 0 ? &scratch_r1[0] : 0;
+    btScalar* links = num_links? &scratch_r1[m_dofCount] : 0;
+    int numLinksChildToRoot=0;
+    int l = link;
+    while (l != -1)
+    {
+        links[numLinksChildToRoot++]=l;
+        l = m_links[l].m_parent;
+    }
+    
+	btMatrix3x3 *rot_from_world = &scratch_m[0];
 
-    const btVector3 p_minus_com_world = contact_point - m_basePos;
-	const btVector3 &normal_lin_world = normal_lin;							//convenience
+	const btVector3 p_minus_com_world = contact_point - m_basePos;
+	const btVector3 &normal_lin_world = normal_lin;  //convenience
 	const btVector3 &normal_ang_world = normal_ang;
 
-    rot_from_world[0] = btMatrix3x3(m_baseQuat);    
-    
-    // omega coeffients first.
-    btVector3 omega_coeffs_world;
-    omega_coeffs_world = p_minus_com_world.cross(normal_lin_world);
+	rot_from_world[0] = btMatrix3x3(m_baseQuat);
+
+	// omega coeffients first.
+	btVector3 omega_coeffs_world;
+	omega_coeffs_world = p_minus_com_world.cross(normal_lin_world);
 	jac[0] = omega_coeffs_world[0] + normal_ang_world[0];
 	jac[1] = omega_coeffs_world[1] + normal_ang_world[1];
 	jac[2] = omega_coeffs_world[2] + normal_ang_world[2];
-    // then v coefficients
-    jac[3] = normal_lin_world[0];
-    jac[4] = normal_lin_world[1];
-    jac[5] = normal_lin_world[2];
+	// then v coefficients
+	jac[3] = normal_lin_world[0];
+	jac[4] = normal_lin_world[1];
+	jac[5] = normal_lin_world[2];
 
 	//create link-local versions of p_minus_com and normal
 	p_minus_com_local[0] = rot_from_world[0] * p_minus_com_world;
-    n_local_lin[0] = rot_from_world[0] * normal_lin_world;
+	n_local_lin[0] = rot_from_world[0] * normal_lin_world;
 	n_local_ang[0] = rot_from_world[0] * normal_ang_world;
 
-    // Set remaining jac values to zero for now.
-    for (int i = 6; i < 6 + m_dofCount; ++i) 
+	// Set remaining jac values to zero for now.
+	for (int i = 6; i < 6 + m_dofCount; ++i)
 	{
-        jac[i] = 0;
-    }
+		jac[i] = 0;
+	}
 
-    // Qdot coefficients, if necessary.
-    if (num_links > 0 && link > -1) {
+	// Qdot coefficients, if necessary.
+	if (num_links > 0 && link > -1)
+	{
+        // TODO: (Also, we are making 3 separate calls to this function, for the normal & the 2 friction directions,
+		// which is resulting in repeated work being done...)
 
-        // TODO: speed this up -- don't calculate for m_links we don't need.
-        // (Also, we are making 3 separate calls to this function, for the normal & the 2 friction directions,
-        // which is resulting in repeated work being done...)
+		// calculate required normals & positions in the local frames.
+        for (int a = 0; a < numLinksChildToRoot; a++)
+        {
+            int i = links[numLinksChildToRoot-1-a];
+        	// transform to local frame
+			const int parent = m_links[i].m_parent;
+			const btMatrix3x3 mtx(m_links[i].m_cachedRotParentToThis);
+			rot_from_world[i + 1] = mtx * rot_from_world[parent + 1];
 
-        // calculate required normals & positions in the local frames.
-        for (int i = 0; i < num_links; ++i) {
-
-            // transform to local frame
-            const int parent = m_links[i].m_parent;
-            const btMatrix3x3 mtx(m_links[i].m_cachedRotParentToThis);
-            rot_from_world[i+1] = mtx * rot_from_world[parent+1];
-
-            n_local_lin[i+1] = mtx * n_local_lin[parent+1];
-			n_local_ang[i+1] = mtx * n_local_ang[parent+1];
-            p_minus_com_local[i+1] = mtx * p_minus_com_local[parent+1] - m_links[i].m_cachedRVector;
+			n_local_lin[i + 1] = mtx * n_local_lin[parent + 1];
+			n_local_ang[i + 1] = mtx * n_local_ang[parent + 1];
+			p_minus_com_local[i + 1] = mtx * p_minus_com_local[parent + 1] - m_links[i].m_cachedRVector;
 
 			// calculate the jacobian entry
-			switch(m_links[i].m_jointType)
+			switch (m_links[i].m_jointType)
 			{
 				case btMultibodyLink::eRevolute:
 				{
-					results[m_links[i].m_dofOffset] = n_local_lin[i+1].dot(m_links[i].getAxisTop(0).cross(p_minus_com_local[i+1]) + m_links[i].getAxisBottom(0));
-					results[m_links[i].m_dofOffset] += n_local_ang[i+1].dot(m_links[i].getAxisTop(0));
+					results[m_links[i].m_dofOffset] = n_local_lin[i + 1].dot(m_links[i].getAxisTop(0).cross(p_minus_com_local[i + 1]) + m_links[i].getAxisBottom(0));
+					results[m_links[i].m_dofOffset] += n_local_ang[i + 1].dot(m_links[i].getAxisTop(0));
 					break;
 				}
 				case btMultibodyLink::ePrismatic:
 				{
-					results[m_links[i].m_dofOffset] = n_local_lin[i+1].dot(m_links[i].getAxisBottom(0));
+					results[m_links[i].m_dofOffset] = n_local_lin[i + 1].dot(m_links[i].getAxisBottom(0));
 					break;
 				}
 				case btMultibodyLink::eSpherical:
 				{
-					results[m_links[i].m_dofOffset + 0] = n_local_lin[i+1].dot(m_links[i].getAxisTop(0).cross(p_minus_com_local[i+1]) + m_links[i].getAxisBottom(0));
-					results[m_links[i].m_dofOffset + 1] = n_local_lin[i+1].dot(m_links[i].getAxisTop(1).cross(p_minus_com_local[i+1]) + m_links[i].getAxisBottom(1));
-					results[m_links[i].m_dofOffset + 2] = n_local_lin[i+1].dot(m_links[i].getAxisTop(2).cross(p_minus_com_local[i+1]) + m_links[i].getAxisBottom(2));
-										
-					results[m_links[i].m_dofOffset + 0] += n_local_ang[i+1].dot(m_links[i].getAxisTop(0));
-					results[m_links[i].m_dofOffset + 1] += n_local_ang[i+1].dot(m_links[i].getAxisTop(1));
-					results[m_links[i].m_dofOffset + 2] += n_local_ang[i+1].dot(m_links[i].getAxisTop(2));
+					results[m_links[i].m_dofOffset + 0] = n_local_lin[i + 1].dot(m_links[i].getAxisTop(0).cross(p_minus_com_local[i + 1]) + m_links[i].getAxisBottom(0));
+					results[m_links[i].m_dofOffset + 1] = n_local_lin[i + 1].dot(m_links[i].getAxisTop(1).cross(p_minus_com_local[i + 1]) + m_links[i].getAxisBottom(1));
+					results[m_links[i].m_dofOffset + 2] = n_local_lin[i + 1].dot(m_links[i].getAxisTop(2).cross(p_minus_com_local[i + 1]) + m_links[i].getAxisBottom(2));
+
+					results[m_links[i].m_dofOffset + 0] += n_local_ang[i + 1].dot(m_links[i].getAxisTop(0));
+					results[m_links[i].m_dofOffset + 1] += n_local_ang[i + 1].dot(m_links[i].getAxisTop(1));
+					results[m_links[i].m_dofOffset + 2] += n_local_ang[i + 1].dot(m_links[i].getAxisTop(2));
 
 					break;
 				}
 				case btMultibodyLink::ePlanar:
 				{
-					results[m_links[i].m_dofOffset + 0] = n_local_lin[i+1].dot(m_links[i].getAxisTop(0).cross(p_minus_com_local[i+1]));// + m_links[i].getAxisBottom(0));
-					results[m_links[i].m_dofOffset + 1] = n_local_lin[i+1].dot(m_links[i].getAxisBottom(1));
-					results[m_links[i].m_dofOffset + 2] = n_local_lin[i+1].dot(m_links[i].getAxisBottom(2));
+					results[m_links[i].m_dofOffset + 0] = n_local_lin[i + 1].dot(m_links[i].getAxisTop(0).cross(p_minus_com_local[i + 1]));  // + m_links[i].getAxisBottom(0));
+					results[m_links[i].m_dofOffset + 1] = n_local_lin[i + 1].dot(m_links[i].getAxisBottom(1));
+					results[m_links[i].m_dofOffset + 2] = n_local_lin[i + 1].dot(m_links[i].getAxisBottom(2));
 
 					break;
 				}
@@ -1824,269 +1843,260 @@ void btMultiBody::fillConstraintJacobianMultiDof(int link,
 				{
 				}
 			}
-            
-        }
+		}
 
-        // Now copy through to output.
+		// Now copy through to output.
 		//printf("jac[%d] = ", link);
-        while (link != -1) 
+		while (link != -1)
 		{
-			for(int dof = 0; dof < m_links[link].m_dofCount; ++dof)
+			for (int dof = 0; dof < m_links[link].m_dofCount; ++dof)
 			{
 				jac[6 + m_links[link].m_dofOffset + dof] = results[m_links[link].m_dofOffset + dof];
 				//printf("%.2f\t", jac[6 + m_links[link].m_dofOffset + dof]);
 			}
-            
-			link = m_links[link].m_parent;
-        }
-		//printf("]\n");
-    }
-}
 
+			link = m_links[link].m_parent;
+		}
+		//printf("]\n");
+	}
+}
 
 void btMultiBody::wakeUp()
 {
 	m_sleepTimer = 0;
-    m_awake = true;
+	m_awake = true;
 }
 
 void btMultiBody::goToSleep()
 {
-    m_awake = false;
+	m_awake = false;
 }
 
 void btMultiBody::checkMotionAndSleepIfRequired(btScalar timestep)
 {
 	extern bool gDisableDeactivation;
-    if (!m_canSleep || gDisableDeactivation) 
+	if (!m_canSleep || gDisableDeactivation)
 	{
 		m_awake = true;
 		m_sleepTimer = 0;
 		return;
 	}
 
-    // motion is computed as omega^2 + v^2 + (sum of squares of joint velocities)
-    btScalar motion = 0;
+	// motion is computed as omega^2 + v^2 + (sum of squares of joint velocities)
+	btScalar motion = 0;
 	{
-		for (int i = 0; i < 6 + m_dofCount; ++i) 		
+		for (int i = 0; i < 6 + m_dofCount; ++i)
 			motion += m_realBuf[i] * m_realBuf[i];
 	}
-	
 
-    if (motion < SLEEP_EPSILON) {
-        m_sleepTimer += timestep;
-        if (m_sleepTimer > SLEEP_TIMEOUT) {
-            goToSleep();
-        }
-    } else {
-        m_sleepTimer = 0;
+	if (motion < SLEEP_EPSILON)
+	{
+		m_sleepTimer += timestep;
+		if (m_sleepTimer > SLEEP_TIMEOUT)
+		{
+			goToSleep();
+		}
+	}
+	else
+	{
+		m_sleepTimer = 0;
 		if (!m_awake)
 			wakeUp();
-    }
+	}
 }
 
-
-void	btMultiBody::forwardKinematics(btAlignedObjectArray<btQuaternion>& world_to_local,btAlignedObjectArray<btVector3>& local_origin)
+void btMultiBody::forwardKinematics(btAlignedObjectArray<btQuaternion> &world_to_local, btAlignedObjectArray<btVector3> &local_origin)
 {
-	
 	int num_links = getNumLinks();
 
 	// Cached 3x3 rotation matrices from parent frame to this frame.
-	btMatrix3x3* rot_from_parent =(btMatrix3x3 *) &m_matrixBuf[0];
+	btMatrix3x3 *rot_from_parent = (btMatrix3x3 *)&m_matrixBuf[0];
 
-	rot_from_parent[0] = btMatrix3x3(m_baseQuat);				//m_baseQuat assumed to be alias!?
-	
-	for (int i = 0; i < num_links; ++i) 
+	rot_from_parent[0] = btMatrix3x3(m_baseQuat);  //m_baseQuat assumed to be alias!?
+
+	for (int i = 0; i < num_links; ++i)
 	{
-		rot_from_parent[i+1] = btMatrix3x3(m_links[i].m_cachedRotParentToThis);
+		rot_from_parent[i + 1] = btMatrix3x3(m_links[i].m_cachedRotParentToThis);
 	}
-		
+
 	int nLinks = getNumLinks();
 	///base + num m_links
-	world_to_local.resize(nLinks+1);
-	local_origin.resize(nLinks+1);
+	world_to_local.resize(nLinks + 1);
+	local_origin.resize(nLinks + 1);
 
 	world_to_local[0] = getWorldToBaseRot();
 	local_origin[0] = getBasePos();
-	
-	for (int k=0;k<getNumLinks();k++)
+
+	for (int k = 0; k < getNumLinks(); k++)
 	{
 		const int parent = getParent(k);
-		world_to_local[k+1] = getParentToLocalRot(k) * world_to_local[parent+1];
-		local_origin[k+1] = local_origin[parent+1] + (quatRotate(world_to_local[k+1].inverse() , getRVector(k)));
+		world_to_local[k + 1] = getParentToLocalRot(k) * world_to_local[parent + 1];
+		local_origin[k + 1] = local_origin[parent + 1] + (quatRotate(world_to_local[k + 1].inverse(), getRVector(k)));
 	}
 
-	for (int link=0;link<getNumLinks();link++)
+	for (int link = 0; link < getNumLinks(); link++)
 	{
-		int index = link+1;
+		int index = link + 1;
 
 		btVector3 posr = local_origin[index];
-		btScalar quat[4]={-world_to_local[index].x(),-world_to_local[index].y(),-world_to_local[index].z(),world_to_local[index].w()};
+		btScalar quat[4] = {-world_to_local[index].x(), -world_to_local[index].y(), -world_to_local[index].z(), world_to_local[index].w()};
 		btTransform tr;
 		tr.setIdentity();
 		tr.setOrigin(posr);
-		tr.setRotation(btQuaternion(quat[0],quat[1],quat[2],quat[3]));
+		tr.setRotation(btQuaternion(quat[0], quat[1], quat[2], quat[3]));
 		getLink(link).m_cachedWorldTransform = tr;
-			
 	}
-
 }
 
-void	btMultiBody::updateCollisionObjectWorldTransforms(btAlignedObjectArray<btQuaternion>& world_to_local,btAlignedObjectArray<btVector3>& local_origin)
+void btMultiBody::updateCollisionObjectWorldTransforms(btAlignedObjectArray<btQuaternion> &world_to_local, btAlignedObjectArray<btVector3> &local_origin)
 {
-	world_to_local.resize(getNumLinks()+1);
-	local_origin.resize(getNumLinks()+1);
-	
+	world_to_local.resize(getNumLinks() + 1);
+	local_origin.resize(getNumLinks() + 1);
+
 	world_to_local[0] = getWorldToBaseRot();
 	local_origin[0] = getBasePos();
-	
+
 	if (getBaseCollider())
 	{
 		btVector3 posr = local_origin[0];
 		//	float pos[4]={posr.x(),posr.y(),posr.z(),1};
-		btScalar quat[4]={-world_to_local[0].x(),-world_to_local[0].y(),-world_to_local[0].z(),world_to_local[0].w()};
+		btScalar quat[4] = {-world_to_local[0].x(), -world_to_local[0].y(), -world_to_local[0].z(), world_to_local[0].w()};
 		btTransform tr;
 		tr.setIdentity();
 		tr.setOrigin(posr);
-		tr.setRotation(btQuaternion(quat[0],quat[1],quat[2],quat[3]));
-		
+		tr.setRotation(btQuaternion(quat[0], quat[1], quat[2], quat[3]));
+
 		getBaseCollider()->setWorldTransform(tr);
-		
 	}
-	
-	for (int k=0;k<getNumLinks();k++)
+
+	for (int k = 0; k < getNumLinks(); k++)
 	{
 		const int parent = getParent(k);
-		world_to_local[k+1] = getParentToLocalRot(k) * world_to_local[parent+1];
-		local_origin[k+1] = local_origin[parent+1] + (quatRotate(world_to_local[k+1].inverse() , getRVector(k)));
+		world_to_local[k + 1] = getParentToLocalRot(k) * world_to_local[parent + 1];
+		local_origin[k + 1] = local_origin[parent + 1] + (quatRotate(world_to_local[k + 1].inverse(), getRVector(k)));
 	}
-	
-	
-	for (int m=0;m<getNumLinks();m++)
+
+	for (int m = 0; m < getNumLinks(); m++)
 	{
-		btMultiBodyLinkCollider* col = getLink(m).m_collider;
+		btMultiBodyLinkCollider *col = getLink(m).m_collider;
 		if (col)
 		{
 			int link = col->m_link;
 			btAssert(link == m);
-			
-			int index = link+1;
-			
+
+			int index = link + 1;
+
 			btVector3 posr = local_origin[index];
 			//			float pos[4]={posr.x(),posr.y(),posr.z(),1};
-			btScalar quat[4]={-world_to_local[index].x(),-world_to_local[index].y(),-world_to_local[index].z(),world_to_local[index].w()};
+			btScalar quat[4] = {-world_to_local[index].x(), -world_to_local[index].y(), -world_to_local[index].z(), world_to_local[index].w()};
 			btTransform tr;
 			tr.setIdentity();
 			tr.setOrigin(posr);
-			tr.setRotation(btQuaternion(quat[0],quat[1],quat[2],quat[3]));
-			
+			tr.setRotation(btQuaternion(quat[0], quat[1], quat[2], quat[3]));
+
 			col->setWorldTransform(tr);
 		}
 	}
 }
 
-int	btMultiBody::calculateSerializeBufferSize()	const
+int btMultiBody::calculateSerializeBufferSize() const
 {
 	int sz = sizeof(btMultiBodyData);
 	return sz;
 }
 
-	///fills the dataBuffer and returns the struct name (and 0 on failure)
-const char*	btMultiBody::serialize(void* dataBuffer, class btSerializer* serializer) const
+///fills the dataBuffer and returns the struct name (and 0 on failure)
+const char *btMultiBody::serialize(void *dataBuffer, class btSerializer *serializer) const
 {
-		btMultiBodyData* mbd = (btMultiBodyData*) dataBuffer;
-		getBasePos().serialize(mbd->m_baseWorldPosition);
-		getWorldToBaseRot().inverse().serialize(mbd->m_baseWorldOrientation);
-		getBaseVel().serialize(mbd->m_baseLinearVelocity);
-		getBaseOmega().serialize(mbd->m_baseAngularVelocity);
+	btMultiBodyData *mbd = (btMultiBodyData *)dataBuffer;
+	getBasePos().serialize(mbd->m_baseWorldPosition);
+	getWorldToBaseRot().inverse().serialize(mbd->m_baseWorldOrientation);
+	getBaseVel().serialize(mbd->m_baseLinearVelocity);
+	getBaseOmega().serialize(mbd->m_baseAngularVelocity);
 
-		mbd->m_baseMass = this->getBaseMass();
-		getBaseInertia().serialize(mbd->m_baseInertia);
+	mbd->m_baseMass = this->getBaseMass();
+	getBaseInertia().serialize(mbd->m_baseInertia);
+	{
+		char *name = (char *)serializer->findNameForPointer(m_baseName);
+		mbd->m_baseName = (char *)serializer->getUniquePointer(name);
+		if (mbd->m_baseName)
 		{
-			char* name = (char*) serializer->findNameForPointer(m_baseName);
-			mbd->m_baseName = (char*)serializer->getUniquePointer(name);
-			if (mbd->m_baseName)
-			{
-				serializer->serializeName(name);
-			}
+			serializer->serializeName(name);
 		}
-		mbd->m_numLinks = this->getNumLinks();
-		if (mbd->m_numLinks)
+	}
+	mbd->m_numLinks = this->getNumLinks();
+	if (mbd->m_numLinks)
+	{
+		int sz = sizeof(btMultiBodyLinkData);
+		int numElem = mbd->m_numLinks;
+		btChunk *chunk = serializer->allocate(sz, numElem);
+		btMultiBodyLinkData *memPtr = (btMultiBodyLinkData *)chunk->m_oldPtr;
+		for (int i = 0; i < numElem; i++, memPtr++)
 		{
-			int sz = sizeof(btMultiBodyLinkData);
-			int numElem = mbd->m_numLinks;
-			btChunk* chunk = serializer->allocate(sz,numElem);
-			btMultiBodyLinkData* memPtr = (btMultiBodyLinkData*)chunk->m_oldPtr;
-			for (int i=0;i<numElem;i++,memPtr++)
+			memPtr->m_jointType = getLink(i).m_jointType;
+			memPtr->m_dofCount = getLink(i).m_dofCount;
+			memPtr->m_posVarCount = getLink(i).m_posVarCount;
+
+			getLink(i).m_inertiaLocal.serialize(memPtr->m_linkInertia);
+
+			getLink(i).m_absFrameTotVelocity.m_topVec.serialize(memPtr->m_absFrameTotVelocityTop);
+			getLink(i).m_absFrameTotVelocity.m_bottomVec.serialize(memPtr->m_absFrameTotVelocityBottom);
+			getLink(i).m_absFrameLocVelocity.m_topVec.serialize(memPtr->m_absFrameLocVelocityTop);
+			getLink(i).m_absFrameLocVelocity.m_bottomVec.serialize(memPtr->m_absFrameLocVelocityBottom);
+
+			memPtr->m_linkMass = getLink(i).m_mass;
+			memPtr->m_parentIndex = getLink(i).m_parent;
+			memPtr->m_jointDamping = getLink(i).m_jointDamping;
+			memPtr->m_jointFriction = getLink(i).m_jointFriction;
+			memPtr->m_jointLowerLimit = getLink(i).m_jointLowerLimit;
+			memPtr->m_jointUpperLimit = getLink(i).m_jointUpperLimit;
+			memPtr->m_jointMaxForce = getLink(i).m_jointMaxForce;
+			memPtr->m_jointMaxVelocity = getLink(i).m_jointMaxVelocity;
+
+			getLink(i).m_eVector.serialize(memPtr->m_parentComToThisPivotOffset);
+			getLink(i).m_dVector.serialize(memPtr->m_thisPivotToThisComOffset);
+			getLink(i).m_zeroRotParentToThis.serialize(memPtr->m_zeroRotParentToThis);
+			btAssert(memPtr->m_dofCount <= 3);
+			for (int dof = 0; dof < getLink(i).m_dofCount; dof++)
 			{
+				getLink(i).getAxisBottom(dof).serialize(memPtr->m_jointAxisBottom[dof]);
+				getLink(i).getAxisTop(dof).serialize(memPtr->m_jointAxisTop[dof]);
 
-				memPtr->m_jointType = getLink(i).m_jointType;
-				memPtr->m_dofCount = getLink(i).m_dofCount;
-				memPtr->m_posVarCount = getLink(i).m_posVarCount;
-				
-				getLink(i).m_inertiaLocal.serialize(memPtr->m_linkInertia);
-
-				getLink(i).m_absFrameTotVelocity.m_topVec.serialize(memPtr->m_absFrameTotVelocityTop);
-				getLink(i).m_absFrameTotVelocity.m_bottomVec.serialize(memPtr->m_absFrameTotVelocityBottom);
-				getLink(i).m_absFrameLocVelocity.m_topVec.serialize(memPtr->m_absFrameLocVelocityTop);
-				getLink(i).m_absFrameLocVelocity.m_bottomVec.serialize(memPtr->m_absFrameLocVelocityBottom);
-
-				memPtr->m_linkMass = getLink(i).m_mass;
-				memPtr->m_parentIndex = getLink(i).m_parent;
-				memPtr->m_jointDamping = getLink(i).m_jointDamping;
-				memPtr->m_jointFriction = getLink(i).m_jointFriction;
-				memPtr->m_jointLowerLimit = getLink(i).m_jointLowerLimit;
-				memPtr->m_jointUpperLimit = getLink(i).m_jointUpperLimit;
-				memPtr->m_jointMaxForce = getLink(i).m_jointMaxForce;
-				memPtr->m_jointMaxVelocity = getLink(i).m_jointMaxVelocity;
-
-				getLink(i).m_eVector.serialize(memPtr->m_parentComToThisPivotOffset);
-				getLink(i).m_dVector.serialize(memPtr->m_thisPivotToThisComOffset);
-				getLink(i).m_zeroRotParentToThis.serialize(memPtr->m_zeroRotParentToThis);
-				btAssert(memPtr->m_dofCount<=3);
-				for (int dof = 0;dof<getLink(i).m_dofCount;dof++)
-				{
-					getLink(i).getAxisBottom(dof).serialize(memPtr->m_jointAxisBottom[dof]);
-					getLink(i).getAxisTop(dof).serialize(memPtr->m_jointAxisTop[dof]);
-					
-					memPtr->m_jointTorque[dof] = getLink(i).m_jointTorque[dof];
-					memPtr->m_jointVel[dof] = getJointVelMultiDof(i)[dof];
-
-				}
-				int numPosVar = getLink(i).m_posVarCount;
-				for (int posvar = 0; posvar < numPosVar;posvar++)
-				{
-					memPtr->m_jointPos[posvar] = getLink(i).m_jointPos[posvar];
-				}
-				
-				
-				{
-					char* name = (char*) serializer->findNameForPointer(m_links[i].m_linkName);
-					memPtr->m_linkName = (char*)serializer->getUniquePointer(name);
-					if (memPtr->m_linkName)
-					{
-						serializer->serializeName(name);
-					}
-				}
-				{
-					char* name = (char*) serializer->findNameForPointer(m_links[i].m_jointName);
-					memPtr->m_jointName = (char*)serializer->getUniquePointer(name);
-					if (memPtr->m_jointName)
-					{
-						serializer->serializeName(name);
-					}
-				}
-				memPtr->m_linkCollider = (btCollisionObjectData*)serializer->getUniquePointer(getLink(i).m_collider);
-
+				memPtr->m_jointTorque[dof] = getLink(i).m_jointTorque[dof];
+				memPtr->m_jointVel[dof] = getJointVelMultiDof(i)[dof];
+			}
+			int numPosVar = getLink(i).m_posVarCount;
+			for (int posvar = 0; posvar < numPosVar; posvar++)
+			{
+				memPtr->m_jointPos[posvar] = getLink(i).m_jointPos[posvar];
 			}
-			serializer->finalizeChunk(chunk,btMultiBodyLinkDataName,BT_ARRAY_CODE,(void*) &m_links[0]);
-		}
-		mbd->m_links = mbd->m_numLinks? (btMultiBodyLinkData*) serializer->getUniquePointer((void*)&m_links[0]):0;
 
-		// Fill padding with zeros to appease msan.
+			{
+				char *name = (char *)serializer->findNameForPointer(m_links[i].m_linkName);
+				memPtr->m_linkName = (char *)serializer->getUniquePointer(name);
+				if (memPtr->m_linkName)
+				{
+					serializer->serializeName(name);
+				}
+			}
+			{
+				char *name = (char *)serializer->findNameForPointer(m_links[i].m_jointName);
+				memPtr->m_jointName = (char *)serializer->getUniquePointer(name);
+				if (memPtr->m_jointName)
+				{
+					serializer->serializeName(name);
+				}
+			}
+			memPtr->m_linkCollider = (btCollisionObjectData *)serializer->getUniquePointer(getLink(i).m_collider);
+		}
+		serializer->finalizeChunk(chunk, btMultiBodyLinkDataName, BT_ARRAY_CODE, (void *)&m_links[0]);
+	}
+	mbd->m_links = mbd->m_numLinks ? (btMultiBodyLinkData *)serializer->getUniquePointer((void *)&m_links[0]) : 0;
+
+	// Fill padding with zeros to appease msan.
 #ifdef BT_USE_DOUBLE_PRECISION
-		memset(mbd->m_padding, 0, sizeof(mbd->m_padding));
+	memset(mbd->m_padding, 0, sizeof(mbd->m_padding));
 #endif
 
-		return btMultiBodyDataName;
+	return btMultiBodyDataName;
 }
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBody.h b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBody.h
index 5cd00e51737..e5c0f1806bd 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBody.h
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBody.h
@@ -21,7 +21,6 @@
  
  */
 
-
 #ifndef BT_MULTIBODY_H
 #define BT_MULTIBODY_H
 
@@ -31,116 +30,111 @@
 #include "LinearMath/btMatrix3x3.h"
 #include "LinearMath/btAlignedObjectArray.h"
 
-
 ///serialization data, don't change them if you are not familiar with the details of the serialization mechanisms
 #ifdef BT_USE_DOUBLE_PRECISION
-	#define btMultiBodyData	btMultiBodyDoubleData
-	#define btMultiBodyDataName	"btMultiBodyDoubleData"
-	#define btMultiBodyLinkData btMultiBodyLinkDoubleData
-	#define btMultiBodyLinkDataName	"btMultiBodyLinkDoubleData"
+#define btMultiBodyData btMultiBodyDoubleData
+#define btMultiBodyDataName "btMultiBodyDoubleData"
+#define btMultiBodyLinkData btMultiBodyLinkDoubleData
+#define btMultiBodyLinkDataName "btMultiBodyLinkDoubleData"
 #else
-	#define btMultiBodyData	btMultiBodyFloatData
-	#define btMultiBodyDataName	"btMultiBodyFloatData"
-	#define btMultiBodyLinkData btMultiBodyLinkFloatData
-	#define btMultiBodyLinkDataName	"btMultiBodyLinkFloatData"
-#endif //BT_USE_DOUBLE_PRECISION
+#define btMultiBodyData btMultiBodyFloatData
+#define btMultiBodyDataName "btMultiBodyFloatData"
+#define btMultiBodyLinkData btMultiBodyLinkFloatData
+#define btMultiBodyLinkDataName "btMultiBodyLinkFloatData"
+#endif  //BT_USE_DOUBLE_PRECISION
 
 #include "btMultiBodyLink.h"
 class btMultiBodyLinkCollider;
 
-ATTRIBUTE_ALIGNED16(class) btMultiBody 
+ATTRIBUTE_ALIGNED16(class)
+btMultiBody
 {
 public:
-
-    
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-    //
-    // initialization
-    //
-    
-	btMultiBody(int n_links,             // NOT including the base
-		btScalar mass,                // mass of base
-		const btVector3 &inertia,    // inertia of base, in base frame; assumed diagonal
-		bool fixedBase,           // whether the base is fixed (true) or can move (false)
-		bool canSleep, bool deprecatedMultiDof=true);
+	//
+	// initialization
+	//
 
+	btMultiBody(int n_links,               // NOT including the base
+				btScalar mass,             // mass of base
+				const btVector3 &inertia,  // inertia of base, in base frame; assumed diagonal
+				bool fixedBase,            // whether the base is fixed (true) or can move (false)
+				bool canSleep, bool deprecatedMultiDof = true);
 
 	virtual ~btMultiBody();
-    
+
 	//note: fixed link collision with parent is always disabled
 	void setupFixed(int linkIndex,
-						   btScalar mass,
-						   const btVector3 &inertia,
-						   int parent,
-						   const btQuaternion &rotParentToThis,
-						   const btVector3 &parentComToThisPivotOffset,
-                           const btVector3 &thisPivotToThisComOffset, bool deprecatedDisableParentCollision=true);
+					btScalar mass,
+					const btVector3 &inertia,
+					int parent,
+					const btQuaternion &rotParentToThis,
+					const btVector3 &parentComToThisPivotOffset,
+					const btVector3 &thisPivotToThisComOffset, bool deprecatedDisableParentCollision = true);
 
-						
 	void setupPrismatic(int i,
-                               btScalar mass,
-                               const btVector3 &inertia,
-                               int parent,
-                               const btQuaternion &rotParentToThis,
-                               const btVector3 &jointAxis,
-                               const btVector3 &parentComToThisPivotOffset,
-							   const btVector3 &thisPivotToThisComOffset,
-							   bool disableParentCollision);
+						btScalar mass,
+						const btVector3 &inertia,
+						int parent,
+						const btQuaternion &rotParentToThis,
+						const btVector3 &jointAxis,
+						const btVector3 &parentComToThisPivotOffset,
+						const btVector3 &thisPivotToThisComOffset,
+						bool disableParentCollision);
 
-    void setupRevolute(int linkIndex,            // 0 to num_links-1
-                       btScalar mass,
-                       const btVector3 &inertia,
-                       int parentIndex,
-                       const btQuaternion &rotParentToThis,  // rotate points in parent frame to this frame, when q = 0
-                       const btVector3 &jointAxis,    // in my frame
-                       const btVector3 &parentComToThisPivotOffset,    // vector from parent COM to joint axis, in PARENT frame
-                       const btVector3 &thisPivotToThisComOffset,       // vector from joint axis to my COM, in MY frame
-					   bool disableParentCollision=false);
+	void setupRevolute(int linkIndex,  // 0 to num_links-1
+					   btScalar mass,
+					   const btVector3 &inertia,
+					   int parentIndex,
+					   const btQuaternion &rotParentToThis,          // rotate points in parent frame to this frame, when q = 0
+					   const btVector3 &jointAxis,                   // in my frame
+					   const btVector3 &parentComToThisPivotOffset,  // vector from parent COM to joint axis, in PARENT frame
+					   const btVector3 &thisPivotToThisComOffset,    // vector from joint axis to my COM, in MY frame
+					   bool disableParentCollision = false);
 
-	void setupSpherical(int linkIndex,											// 0 to num_links-1
-                       btScalar mass,
-                       const btVector3 &inertia,
-                       int parent,
-                       const btQuaternion &rotParentToThis,		// rotate points in parent frame to this frame, when q = 0                       
-                       const btVector3 &parentComToThisPivotOffset,			// vector from parent COM to joint axis, in PARENT frame
-                       const btVector3 &thisPivotToThisComOffset,				// vector from joint axis to my COM, in MY frame
-					   bool disableParentCollision=false);		
+	void setupSpherical(int linkIndex,  // 0 to num_links-1
+						btScalar mass,
+						const btVector3 &inertia,
+						int parent,
+						const btQuaternion &rotParentToThis,          // rotate points in parent frame to this frame, when q = 0
+						const btVector3 &parentComToThisPivotOffset,  // vector from parent COM to joint axis, in PARENT frame
+						const btVector3 &thisPivotToThisComOffset,    // vector from joint axis to my COM, in MY frame
+						bool disableParentCollision = false);
 
-	void setupPlanar(int i,											// 0 to num_links-1
-                       btScalar mass,
-                       const btVector3 &inertia,
-                       int parent,
-                       const btQuaternion &rotParentToThis,		// rotate points in parent frame to this frame, when q = 0                       
-					   const btVector3 &rotationAxis,
-                       const btVector3 &parentComToThisComOffset,			// vector from parent COM to this COM, in PARENT frame                       
-					   bool disableParentCollision=false);		
-	
-	const btMultibodyLink& getLink(int index) const
+	void setupPlanar(int i,  // 0 to num_links-1
+					 btScalar mass,
+					 const btVector3 &inertia,
+					 int parent,
+					 const btQuaternion &rotParentToThis,  // rotate points in parent frame to this frame, when q = 0
+					 const btVector3 &rotationAxis,
+					 const btVector3 &parentComToThisComOffset,  // vector from parent COM to this COM, in PARENT frame
+					 bool disableParentCollision = false);
+
+	const btMultibodyLink &getLink(int index) const
 	{
 		return m_links[index];
 	}
 
-	btMultibodyLink& getLink(int index)
+	btMultibodyLink &getLink(int index)
 	{
 		return m_links[index];
 	}
 
-
-	void setBaseCollider(btMultiBodyLinkCollider* collider)//collider can be NULL to disable collision for the base
+	void setBaseCollider(btMultiBodyLinkCollider * collider)  //collider can be NULL to disable collision for the base
 	{
 		m_baseCollider = collider;
 	}
-	const btMultiBodyLinkCollider* getBaseCollider() const
+	const btMultiBodyLinkCollider *getBaseCollider() const
 	{
 		return m_baseCollider;
 	}
-	btMultiBodyLinkCollider* getBaseCollider()
+	btMultiBodyLinkCollider *getBaseCollider()
 	{
 		return m_baseCollider;
 	}
 
-	btMultiBodyLinkCollider* getLinkCollider(int index)
+	const btMultiBodyLinkCollider *getLinkCollider(int index) const
 	{
 		if (index >= 0 && index < getNumLinks())
 		{
@@ -149,61 +143,65 @@ public:
 		return 0;
 	}
 
-    //
-    // get parent
-    // input: link num from 0 to num_links-1
-    // output: link num from 0 to num_links-1, OR -1 to mean the base.
-    //
-    int getParent(int link_num) const;
-    
-    
-    //
-    // get number of m_links, masses, moments of inertia
-    //
-
-    int getNumLinks() const { return m_links.size(); }
-	int getNumDofs() const { return m_dofCount; }
-	int getNumPosVars() const { return m_posVarCnt; }
-    btScalar getBaseMass() const { return m_baseMass; }
-    const btVector3 & getBaseInertia() const { return m_baseInertia; }
-    btScalar getLinkMass(int i) const;
-    const btVector3 & getLinkInertia(int i) const;
-    
-	
-
-    //
-    // change mass (incomplete: can only change base mass and inertia at present)
-    //
-
-    void setBaseMass(btScalar mass) { m_baseMass = mass; }
-    void setBaseInertia(const btVector3 &inertia) { m_baseInertia = inertia; }
-
-
-    //
-    // get/set pos/vel/rot/omega for the base link
-    //
-
-    const btVector3 & getBasePos() const { return m_basePos; }    // in world frame
-    const btVector3 getBaseVel() const 
-	{ 
-		return btVector3(m_realBuf[3],m_realBuf[4],m_realBuf[5]); 
-	}     // in world frame
-    const btQuaternion & getWorldToBaseRot() const 
-	{ 
-		return m_baseQuat; 
-	}     // rotates world vectors into base frame
-    btVector3 getBaseOmega() const { return btVector3(m_realBuf[0],m_realBuf[1],m_realBuf[2]); }   // in world frame
-
-    void setBasePos(const btVector3 &pos) 
-	{ 
-		m_basePos = pos; 
+	btMultiBodyLinkCollider *getLinkCollider(int index)
+	{
+		if (index >= 0 && index < getNumLinks())
+		{
+			return getLink(index).m_collider;
+		}
+		return 0;
 	}
 
-	void setBaseWorldTransform(const btTransform& tr)
+	//
+	// get parent
+	// input: link num from 0 to num_links-1
+	// output: link num from 0 to num_links-1, OR -1 to mean the base.
+	//
+	int getParent(int link_num) const;
+
+	//
+	// get number of m_links, masses, moments of inertia
+	//
+
+	int getNumLinks() const { return m_links.size(); }
+	int getNumDofs() const { return m_dofCount; }
+	int getNumPosVars() const { return m_posVarCnt; }
+	btScalar getBaseMass() const { return m_baseMass; }
+	const btVector3 &getBaseInertia() const { return m_baseInertia; }
+	btScalar getLinkMass(int i) const;
+	const btVector3 &getLinkInertia(int i) const;
+
+	//
+	// change mass (incomplete: can only change base mass and inertia at present)
+	//
+
+	void setBaseMass(btScalar mass) { m_baseMass = mass; }
+	void setBaseInertia(const btVector3 &inertia) { m_baseInertia = inertia; }
+
+	//
+	// get/set pos/vel/rot/omega for the base link
+	//
+
+	const btVector3 &getBasePos() const { return m_basePos; }  // in world frame
+	const btVector3 getBaseVel() const
+	{
+		return btVector3(m_realBuf[3], m_realBuf[4], m_realBuf[5]);
+	}  // in world frame
+	const btQuaternion &getWorldToBaseRot() const
+	{
+		return m_baseQuat;
+	}                                                                                               // rotates world vectors into base frame
+	btVector3 getBaseOmega() const { return btVector3(m_realBuf[0], m_realBuf[1], m_realBuf[2]); }  // in world frame
+
+	void setBasePos(const btVector3 &pos)
+	{
+		m_basePos = pos;
+	}
+
+	void setBaseWorldTransform(const btTransform &tr)
 	{
 		setBasePos(tr.getOrigin());
 		setWorldToBaseRot(tr.getRotation().inverse());
-
 	}
 
 	btTransform getBaseWorldTransform() const
@@ -214,190 +212,186 @@ public:
 		return tr;
 	}
 
-    void setBaseVel(const btVector3 &vel) 
-	{ 
-
-		m_realBuf[3]=vel[0]; m_realBuf[4]=vel[1]; m_realBuf[5]=vel[2]; 
+	void setBaseVel(const btVector3 &vel)
+	{
+		m_realBuf[3] = vel[0];
+		m_realBuf[4] = vel[1];
+		m_realBuf[5] = vel[2];
 	}
-    void setWorldToBaseRot(const btQuaternion &rot) 
-	{ 
-		m_baseQuat = rot;					//m_baseQuat asumed to ba alias!?
+	void setWorldToBaseRot(const btQuaternion &rot)
+	{
+		m_baseQuat = rot;  //m_baseQuat asumed to ba alias!?
 	}
-    void setBaseOmega(const btVector3 &omega) 
-	{ 
-		m_realBuf[0]=omega[0]; 
-		m_realBuf[1]=omega[1]; 
-		m_realBuf[2]=omega[2]; 
+	void setBaseOmega(const btVector3 &omega)
+	{
+		m_realBuf[0] = omega[0];
+		m_realBuf[1] = omega[1];
+		m_realBuf[2] = omega[2];
 	}
 
+	//
+	// get/set pos/vel for child m_links (i = 0 to num_links-1)
+	//
 
-    //
-    // get/set pos/vel for child m_links (i = 0 to num_links-1)
-    //
+	btScalar getJointPos(int i) const;
+	btScalar getJointVel(int i) const;
 
-    btScalar getJointPos(int i) const;
-    btScalar getJointVel(int i) const;
+	btScalar *getJointVelMultiDof(int i);
+	btScalar *getJointPosMultiDof(int i);
 
-	btScalar * getJointVelMultiDof(int i);
-	btScalar * getJointPosMultiDof(int i);
+	const btScalar *getJointVelMultiDof(int i) const;
+	const btScalar *getJointPosMultiDof(int i) const;
 
-	const btScalar * getJointVelMultiDof(int i) const ;
-	const btScalar * getJointPosMultiDof(int i) const ;
+	void setJointPos(int i, btScalar q);
+	void setJointVel(int i, btScalar qdot);
+	void setJointPosMultiDof(int i, const double *q);
+	void setJointVelMultiDof(int i, const double *qdot);
+	void setJointPosMultiDof(int i, const float *q);
+	void setJointVelMultiDof(int i, const float *qdot);
 
-    void setJointPos(int i, btScalar q);
-    void setJointVel(int i, btScalar qdot);
-	void setJointPosMultiDof(int i, btScalar *q);
-    void setJointVelMultiDof(int i, btScalar *qdot);	
-
-
-
-    //
-    // direct access to velocities as a vector of 6 + num_links elements.
-    // (omega first, then v, then joint velocities.)
-    //
-    const btScalar * getVelocityVector() const 
-	{ 
-		return &m_realBuf[0]; 
+	//
+	// direct access to velocities as a vector of 6 + num_links elements.
+	// (omega first, then v, then joint velocities.)
+	//
+	const btScalar *getVelocityVector() const
+	{
+		return &m_realBuf[0];
 	}
-/*    btScalar * getVelocityVector() 
+	/*    btScalar * getVelocityVector() 
 	{ 
 		return &real_buf[0]; 
 	}
-  */  
+  */
 
-    //
-    // get the frames of reference (positions and orientations) of the child m_links
-    // (i = 0 to num_links-1)
-    //
+	//
+	// get the frames of reference (positions and orientations) of the child m_links
+	// (i = 0 to num_links-1)
+	//
 
-    const btVector3 & getRVector(int i) const;   // vector from COM(parent(i)) to COM(i), in frame i's coords
-    const btQuaternion & getParentToLocalRot(int i) const;   // rotates vectors in frame parent(i) to vectors in frame i.
+	const btVector3 &getRVector(int i) const;              // vector from COM(parent(i)) to COM(i), in frame i's coords
+	const btQuaternion &getParentToLocalRot(int i) const;  // rotates vectors in frame parent(i) to vectors in frame i.
 
+	//
+	// transform vectors in local frame of link i to world frame (or vice versa)
+	//
+	btVector3 localPosToWorld(int i, const btVector3 &vec) const;
+	btVector3 localDirToWorld(int i, const btVector3 &vec) const;
+	btVector3 worldPosToLocal(int i, const btVector3 &vec) const;
+	btVector3 worldDirToLocal(int i, const btVector3 &vec) const;
 
-    //
-    // transform vectors in local frame of link i to world frame (or vice versa)
-    //
-    btVector3 localPosToWorld(int i, const btVector3 &vec) const;
-    btVector3 localDirToWorld(int i, const btVector3 &vec) const;
-    btVector3 worldPosToLocal(int i, const btVector3 &vec) const;
-    btVector3 worldDirToLocal(int i, const btVector3 &vec) const;
+	//
+	// transform a frame in local coordinate to a frame in world coordinate
+	//
+	btMatrix3x3 localFrameToWorld(int i, const btMatrix3x3 &mat) const;
 
-    //
-    // transform a frame in local coordinate to a frame in world coordinate
-    //
-    btMatrix3x3 localFrameToWorld(int i, const btMatrix3x3 &mat) const;
+	//
+	// calculate kinetic energy and angular momentum
+	// useful for debugging.
+	//
 
-    //
-    // calculate kinetic energy and angular momentum
-    // useful for debugging.
-    //
+	btScalar getKineticEnergy() const;
+	btVector3 getAngularMomentum() const;
 
-    btScalar getKineticEnergy() const;
-    btVector3 getAngularMomentum() const;
-    
+	//
+	// set external forces and torques. Note all external forces/torques are given in the WORLD frame.
+	//
 
-    //
-    // set external forces and torques. Note all external forces/torques are given in the WORLD frame.
-    //
-
-    void clearForcesAndTorques();
-   void clearConstraintForces();
+	void clearForcesAndTorques();
+	void clearConstraintForces();
 
 	void clearVelocities();
 
-    void addBaseForce(const btVector3 &f) 
-	{ 
-		m_baseForce += f; 
+	void addBaseForce(const btVector3 &f)
+	{
+		m_baseForce += f;
 	}
-    void addBaseTorque(const btVector3 &t) { m_baseTorque += t; }
-    void addLinkForce(int i, const btVector3 &f);
-    void addLinkTorque(int i, const btVector3 &t);
+	void addBaseTorque(const btVector3 &t) { m_baseTorque += t; }
+	void addLinkForce(int i, const btVector3 &f);
+	void addLinkTorque(int i, const btVector3 &t);
 
- void addBaseConstraintForce(const btVector3 &f)
-        {
-                m_baseConstraintForce += f;
-        }
-    void addBaseConstraintTorque(const btVector3 &t) { m_baseConstraintTorque += t; }
-    void addLinkConstraintForce(int i, const btVector3 &f);
-    void addLinkConstraintTorque(int i, const btVector3 &t);
-       
+	void addBaseConstraintForce(const btVector3 &f)
+	{
+		m_baseConstraintForce += f;
+	}
+	void addBaseConstraintTorque(const btVector3 &t) { m_baseConstraintTorque += t; }
+	void addLinkConstraintForce(int i, const btVector3 &f);
+	void addLinkConstraintTorque(int i, const btVector3 &t);
 
-void addJointTorque(int i, btScalar Q);
+	void addJointTorque(int i, btScalar Q);
 	void addJointTorqueMultiDof(int i, int dof, btScalar Q);
 	void addJointTorqueMultiDof(int i, const btScalar *Q);
 
-    const btVector3 & getBaseForce() const { return m_baseForce; }
-    const btVector3 & getBaseTorque() const { return m_baseTorque; }
-    const btVector3 & getLinkForce(int i) const;
-    const btVector3 & getLinkTorque(int i) const;
-    btScalar getJointTorque(int i) const;
-	btScalar * getJointTorqueMultiDof(int i);
+	const btVector3 &getBaseForce() const { return m_baseForce; }
+	const btVector3 &getBaseTorque() const { return m_baseTorque; }
+	const btVector3 &getLinkForce(int i) const;
+	const btVector3 &getLinkTorque(int i) const;
+	btScalar getJointTorque(int i) const;
+	btScalar *getJointTorqueMultiDof(int i);
 
+	//
+	// dynamics routines.
+	//
 
-    //
-    // dynamics routines.
-    //
+	// timestep the velocities (given the external forces/torques set using addBaseForce etc).
+	// also sets up caches for calcAccelerationDeltas.
+	//
+	// Note: the caller must provide three vectors which are used as
+	// temporary scratch space. The idea here is to reduce dynamic
+	// memory allocation: the same scratch vectors can be re-used
+	// again and again for different Multibodies, instead of each
+	// btMultiBody allocating (and then deallocating) their own
+	// individual scratch buffers. This gives a considerable speed
+	// improvement, at least on Windows (where dynamic memory
+	// allocation appears to be fairly slow).
+	//
 
-    // timestep the velocities (given the external forces/torques set using addBaseForce etc).
-    // also sets up caches for calcAccelerationDeltas.
-    //
-    // Note: the caller must provide three vectors which are used as
-    // temporary scratch space. The idea here is to reduce dynamic
-    // memory allocation: the same scratch vectors can be re-used
-    // again and again for different Multibodies, instead of each
-    // btMultiBody allocating (and then deallocating) their own
-    // individual scratch buffers. This gives a considerable speed
-    // improvement, at least on Windows (where dynamic memory
-    // allocation appears to be fairly slow).
-    //
-   
- 
 	void computeAccelerationsArticulatedBodyAlgorithmMultiDof(btScalar dt,
-                        btAlignedObjectArray<btScalar> &scratch_r,
-                        btAlignedObjectArray<btVector3> &scratch_v,
-                        btAlignedObjectArray<btMatrix3x3> &scratch_m,
-			bool isConstraintPass=false
-		);
+															  btAlignedObjectArray<btScalar> & scratch_r,
+															  btAlignedObjectArray<btVector3> & scratch_v,
+															  btAlignedObjectArray<btMatrix3x3> & scratch_m,
+															  bool isConstraintPass,
+                                                              bool jointFeedbackInWorldSpace,
+                                                              bool jointFeedbackInJointFrame
+                                                              );
 
-///stepVelocitiesMultiDof is deprecated, use computeAccelerationsArticulatedBodyAlgorithmMultiDof instead
-        void stepVelocitiesMultiDof(btScalar dt,
-                        btAlignedObjectArray<btScalar> &scratch_r,
-                        btAlignedObjectArray<btVector3> &scratch_v,
-                        btAlignedObjectArray<btMatrix3x3> &scratch_m,
-                        bool isConstraintPass=false)
-	{
-		computeAccelerationsArticulatedBodyAlgorithmMultiDof(dt,scratch_r,scratch_v,scratch_m,isConstraintPass);
-        }
+	///stepVelocitiesMultiDof is deprecated, use computeAccelerationsArticulatedBodyAlgorithmMultiDof instead
+	//void stepVelocitiesMultiDof(btScalar dt,
+	//							btAlignedObjectArray<btScalar> & scratch_r,
+	//							btAlignedObjectArray<btVector3> & scratch_v,
+	//							btAlignedObjectArray<btMatrix3x3> & scratch_m,
+	//							bool isConstraintPass = false)
+	//{
+	//	computeAccelerationsArticulatedBodyAlgorithmMultiDof(dt, scratch_r, scratch_v, scratch_m, isConstraintPass, false, false);
+	//}
 
-    // calcAccelerationDeltasMultiDof
-    // input: force vector (in same format as jacobian, i.e.:
-    //                      3 torque values, 3 force values, num_links joint torque values)
-    // output: 3 omegadot values, 3 vdot values, num_links q_double_dot values
-    // (existing contents of output array are replaced)
-    // calcAccelerationDeltasMultiDof must have been called first.
+	// calcAccelerationDeltasMultiDof
+	// input: force vector (in same format as jacobian, i.e.:
+	//                      3 torque values, 3 force values, num_links joint torque values)
+	// output: 3 omegadot values, 3 vdot values, num_links q_double_dot values
+	// (existing contents of output array are replaced)
+	// calcAccelerationDeltasMultiDof must have been called first.
 	void calcAccelerationDeltasMultiDof(const btScalar *force, btScalar *output,
-                                btAlignedObjectArray<btScalar> &scratch_r,
-                                btAlignedObjectArray<btVector3> &scratch_v) const;
-	
-  
-	void applyDeltaVeeMultiDof2(const btScalar * delta_vee, btScalar multiplier)
+										btAlignedObjectArray<btScalar> &scratch_r,
+										btAlignedObjectArray<btVector3> &scratch_v) const;
+
+	void applyDeltaVeeMultiDof2(const btScalar *delta_vee, btScalar multiplier)
 	{
 		for (int dof = 0; dof < 6 + getNumDofs(); ++dof)
-                {
-                        m_deltaV[dof] += delta_vee[dof] * multiplier;
-                }
+		{
+			m_deltaV[dof] += delta_vee[dof] * multiplier;
+		}
 	}
 	void processDeltaVeeMultiDof2()
 	{
-		applyDeltaVeeMultiDof(&m_deltaV[0],1);
+		applyDeltaVeeMultiDof(&m_deltaV[0], 1);
 
 		for (int dof = 0; dof < 6 + getNumDofs(); ++dof)
-                {
+		{
 			m_deltaV[dof] = 0.f;
 		}
 	}
 
-	void applyDeltaVeeMultiDof(const btScalar * delta_vee, btScalar multiplier) 
+	void applyDeltaVeeMultiDof(const btScalar *delta_vee, btScalar multiplier)
 	{
 		//for (int dof = 0; dof < 6 + getNumDofs(); ++dof)
 		//	printf("%.4f ", delta_vee[dof]*multiplier);
@@ -418,65 +412,61 @@ void addJointTorque(int i, btScalar Q);
 		for (int dof = 0; dof < 6 + getNumDofs(); ++dof)
 		{
 			m_realBuf[dof] += delta_vee[dof] * multiplier;
-			btClamp(m_realBuf[dof],-m_maxCoordinateVelocity,m_maxCoordinateVelocity);
+			btClamp(m_realBuf[dof], -m_maxCoordinateVelocity, m_maxCoordinateVelocity);
 		}
-    }
+	}
 
-	
-	
-    // timestep the positions (given current velocities).
+	// timestep the positions (given current velocities).
 	void stepPositionsMultiDof(btScalar dt, btScalar *pq = 0, btScalar *pqd = 0);
 
+	//
+	// contacts
+	//
 
-    //
-    // contacts
-    //
+	// This routine fills out a contact constraint jacobian for this body.
+	// the 'normal' supplied must be -n for body1 or +n for body2 of the contact.
+	// 'normal' & 'contact_point' are both given in world coordinates.
 
-    // This routine fills out a contact constraint jacobian for this body.
-    // the 'normal' supplied must be -n for body1 or +n for body2 of the contact.
-    // 'normal' & 'contact_point' are both given in world coordinates.
-	
 	void fillContactJacobianMultiDof(int link,
-                             const btVector3 &contact_point,
-                             const btVector3 &normal,
-                             btScalar *jac,
-                             btAlignedObjectArray<btScalar> &scratch_r,
-                             btAlignedObjectArray<btVector3> &scratch_v,
-							 btAlignedObjectArray<btMatrix3x3> &scratch_m) const { fillConstraintJacobianMultiDof(link, contact_point, btVector3(0, 0, 0), normal, jac, scratch_r, scratch_v, scratch_m); }
+									 const btVector3 &contact_point,
+									 const btVector3 &normal,
+									 btScalar *jac,
+									 btAlignedObjectArray<btScalar> &scratch_r,
+									 btAlignedObjectArray<btVector3> &scratch_v,
+									 btAlignedObjectArray<btMatrix3x3> &scratch_m) const { fillConstraintJacobianMultiDof(link, contact_point, btVector3(0, 0, 0), normal, jac, scratch_r, scratch_v, scratch_m); }
 
 	//a more general version of fillContactJacobianMultiDof which does not assume..
 	//.. that the constraint in question is contact or, to be more precise, constrains linear velocity only
 	void fillConstraintJacobianMultiDof(int link,
-                             const btVector3 &contact_point,
-							 const btVector3 &normal_ang,
-                             const btVector3 &normal_lin,
-                             btScalar *jac,
-                             btAlignedObjectArray<btScalar> &scratch_r,
-                             btAlignedObjectArray<btVector3> &scratch_v,
-                             btAlignedObjectArray<btMatrix3x3> &scratch_m) const;
+										const btVector3 &contact_point,
+										const btVector3 &normal_ang,
+										const btVector3 &normal_lin,
+										btScalar *jac,
+										btAlignedObjectArray<btScalar> &scratch_r,
+										btAlignedObjectArray<btVector3> &scratch_v,
+										btAlignedObjectArray<btMatrix3x3> &scratch_m) const;
 
-
-    //
-    // sleeping
-    //
-	void	setCanSleep(bool canSleep)
+	//
+	// sleeping
+	//
+	void setCanSleep(bool canSleep)
 	{
 		m_canSleep = canSleep;
 	}
 
-	bool getCanSleep()const
+	bool getCanSleep() const
 	{
 		return m_canSleep;
 	}
 
-    bool isAwake() const { return m_awake; }
-    void wakeUp();
-    void goToSleep();
-    void checkMotionAndSleepIfRequired(btScalar timestep);
-    
+	bool isAwake() const { return m_awake; }
+	void wakeUp();
+	void goToSleep();
+	void checkMotionAndSleepIfRequired(btScalar timestep);
+
 	bool hasFixedBase() const
 	{
-		    return m_fixedBase;
+		return m_fixedBase;
 	}
 
 	int getCompanionId() const
@@ -489,16 +479,16 @@ void addJointTorque(int i, btScalar Q);
 		m_companionId = id;
 	}
 
-	void setNumLinks(int numLinks)//careful: when changing the number of m_links, make sure to re-initialize or update existing m_links
+	void setNumLinks(int numLinks)  //careful: when changing the number of m_links, make sure to re-initialize or update existing m_links
 	{
 		m_links.resize(numLinks);
 	}
 
 	btScalar getLinearDamping() const
 	{
-			return m_linearDamping;
+		return m_linearDamping;
 	}
-	void setLinearDamping( btScalar damp)
+	void setLinearDamping(btScalar damp)
 	{
 		m_linearDamping = damp;
 	}
@@ -506,11 +496,11 @@ void addJointTorque(int i, btScalar Q);
 	{
 		return m_angularDamping;
 	}
-	void setAngularDamping( btScalar damp)
+	void setAngularDamping(btScalar damp)
 	{
 		m_angularDamping = damp;
 	}
-		
+
 	bool getUseGyroTerm() const
 	{
 		return m_useGyroTerm;
@@ -519,24 +509,24 @@ void addJointTorque(int i, btScalar Q);
 	{
 		m_useGyroTerm = useGyro;
 	}
-	btScalar	getMaxCoordinateVelocity() const
+	btScalar getMaxCoordinateVelocity() const
 	{
-		return m_maxCoordinateVelocity ;
+		return m_maxCoordinateVelocity;
 	}
-	void	setMaxCoordinateVelocity(btScalar maxVel)
+	void setMaxCoordinateVelocity(btScalar maxVel)
 	{
 		m_maxCoordinateVelocity = maxVel;
 	}
 
-	btScalar	getMaxAppliedImpulse() const
+	btScalar getMaxAppliedImpulse() const
 	{
 		return m_maxAppliedImpulse;
 	}
-	void	setMaxAppliedImpulse(btScalar maxImp)
+	void setMaxAppliedImpulse(btScalar maxImp)
 	{
 		m_maxAppliedImpulse = maxImp;
 	}
-	void	setHasSelfCollision(bool hasSelfCollision)
+	void setHasSelfCollision(bool hasSelfCollision)
 	{
 		m_hasSelfCollision = hasSelfCollision;
 	}
@@ -545,7 +535,6 @@ void addJointTorque(int i, btScalar Q);
 		return m_hasSelfCollision;
 	}
 
-	
 	void finalizeMultiDof();
 
 	void useRK4Integration(bool use) { m_useRK4 = use; }
@@ -561,126 +550,132 @@ void addJointTorque(int i, btScalar Q);
 	{
 		__posUpdated = updated;
 	}
-	
+
 	//internalNeedsJointFeedback is for internal use only
 	bool internalNeedsJointFeedback() const
 	{
 		return m_internalNeedsJointFeedback;
 	}
-	void	forwardKinematics(btAlignedObjectArray<btQuaternion>& scratch_q,btAlignedObjectArray<btVector3>& scratch_m);
+	void forwardKinematics(btAlignedObjectArray<btQuaternion> & scratch_q, btAlignedObjectArray<btVector3> & scratch_m);
 
-	void compTreeLinkVelocities(btVector3 *omega, btVector3 *vel) const;
+	void compTreeLinkVelocities(btVector3 * omega, btVector3 * vel) const;
 
-	void	updateCollisionObjectWorldTransforms(btAlignedObjectArray<btQuaternion>& scratch_q,btAlignedObjectArray<btVector3>& scratch_m);
-	
-	virtual	int	calculateSerializeBufferSize()	const;
+	void updateCollisionObjectWorldTransforms(btAlignedObjectArray<btQuaternion> & scratch_q, btAlignedObjectArray<btVector3> & scratch_m);
+
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer,  class btSerializer* serializer) const;
+	virtual const char *serialize(void *dataBuffer, class btSerializer *serializer) const;
 
-	const char*				getBaseName() const
+	const char *getBaseName() const
 	{
 		return m_baseName;
 	}
 	///memory of setBaseName needs to be manager by user
-	void	setBaseName(const char* name)
+	void setBaseName(const char *name)
 	{
 		m_baseName = name;
 	}
 
 	///users can point to their objects, userPointer is not used by Bullet
-	void*	getUserPointer() const
+	void *getUserPointer() const
 	{
 		return m_userObjectPointer;
 	}
 
-	int	getUserIndex() const
+	int getUserIndex() const
 	{
 		return m_userIndex;
 	}
 
-	int	getUserIndex2() const
+	int getUserIndex2() const
 	{
 		return m_userIndex2;
 	}
 	///users can point to their objects, userPointer is not used by Bullet
-	void	setUserPointer(void* userPointer)
+	void setUserPointer(void *userPointer)
 	{
 		m_userObjectPointer = userPointer;
 	}
 
 	///users can point to their objects, userPointer is not used by Bullet
-	void	setUserIndex(int index)
+	void setUserIndex(int index)
 	{
 		m_userIndex = index;
 	}
 
-	void	setUserIndex2(int index)
+	void setUserIndex2(int index)
 	{
 		m_userIndex2 = index;
 	}
 
+	static void spatialTransform(const btMatrix3x3 &rotation_matrix,  // rotates vectors in 'from' frame to vectors in 'to' frame
+		const btVector3 &displacement,     // vector from origin of 'from' frame to origin of 'to' frame, in 'to' coordinates
+		const btVector3 &top_in,       // top part of input vector
+		const btVector3 &bottom_in,    // bottom part of input vector
+		btVector3 &top_out,         // top part of output vector
+		btVector3 &bottom_out);      // bottom part of output vector
+
+
+
 private:
-    btMultiBody(const btMultiBody &);  // not implemented
-    void operator=(const btMultiBody &);  // not implemented
+	btMultiBody(const btMultiBody &);     // not implemented
+	void operator=(const btMultiBody &);  // not implemented
 
-
-	void solveImatrix(const btVector3& rhs_top, const btVector3& rhs_bot, btScalar result[6]) const;
+	void solveImatrix(const btVector3 &rhs_top, const btVector3 &rhs_bot, btScalar result[6]) const;
 	void solveImatrix(const btSpatialForceVector &rhs, btSpatialMotionVector &result) const;
-	
+
 	void updateLinksDofOffsets()
 	{
 		int dofOffset = 0, cfgOffset = 0;
-		for(int bidx = 0; bidx < m_links.size(); ++bidx)
+		for (int bidx = 0; bidx < m_links.size(); ++bidx)
 		{
-			m_links[bidx].m_dofOffset = dofOffset; m_links[bidx].m_cfgOffset = cfgOffset;
-			dofOffset += m_links[bidx].m_dofCount; cfgOffset += m_links[bidx].m_posVarCount;
+			m_links[bidx].m_dofOffset = dofOffset;
+			m_links[bidx].m_cfgOffset = cfgOffset;
+			dofOffset += m_links[bidx].m_dofCount;
+			cfgOffset += m_links[bidx].m_posVarCount;
 		}
 	}
 
-	void mulMatrix(btScalar *pA, btScalar *pB, int rowsA, int colsA, int rowsB, int colsB, btScalar *pC) const;
-	
-	
+	void mulMatrix(btScalar * pA, btScalar * pB, int rowsA, int colsA, int rowsB, int colsB, btScalar *pC) const;
+
 private:
+	btMultiBodyLinkCollider *m_baseCollider;  //can be NULL
+	const char *m_baseName;                   //memory needs to be manager by user!
 
-	btMultiBodyLinkCollider* m_baseCollider;//can be NULL
-	const char*				m_baseName;//memory needs to be manager by user!
+	btVector3 m_basePos;      // position of COM of base (world frame)
+	btQuaternion m_baseQuat;  // rotates world points into base frame
 
-    btVector3 m_basePos;       // position of COM of base (world frame)
-    btQuaternion m_baseQuat;   // rotates world points into base frame
+	btScalar m_baseMass;      // mass of the base
+	btVector3 m_baseInertia;  // inertia of the base (in local frame; diagonal)
 
-    btScalar m_baseMass;         // mass of the base
-    btVector3 m_baseInertia;   // inertia of the base (in local frame; diagonal)
+	btVector3 m_baseForce;   // external force applied to base. World frame.
+	btVector3 m_baseTorque;  // external torque applied to base. World frame.
 
-    btVector3 m_baseForce;     // external force applied to base. World frame.
-    btVector3 m_baseTorque;    // external torque applied to base. World frame.
-   
-    btVector3 m_baseConstraintForce;     // external force applied to base. World frame.
-    btVector3 m_baseConstraintTorque;    // external torque applied to base. World frame.
- 
-    btAlignedObjectArray<btMultibodyLink> m_links;    // array of m_links, excluding the base. index from 0 to num_links-1.
+	btVector3 m_baseConstraintForce;   // external force applied to base. World frame.
+	btVector3 m_baseConstraintTorque;  // external torque applied to base. World frame.
 
-    
-    //
-    // realBuf:
-    //  offset         size            array
-    //   0              6 + num_links   v (base_omega; base_vel; joint_vels)					MULTIDOF [sysdof x sysdof for D matrices (TOO MUCH!) + pos_delta which is sys-cfg sized]
-    //   6+num_links    num_links       D
-    //
-    // vectorBuf:
-    //  offset         size         array
-    //   0              num_links    h_top
-    //   num_links      num_links    h_bottom
-    //
-    // matrixBuf:
-    //  offset         size         array
-    //   0              num_links+1  rot_from_parent
-    //
-   btAlignedObjectArray<btScalar> m_deltaV; 
-    btAlignedObjectArray<btScalar> m_realBuf;
-    btAlignedObjectArray<btVector3> m_vectorBuf;
-    btAlignedObjectArray<btMatrix3x3> m_matrixBuf;
+	btAlignedObjectArray<btMultibodyLink> m_links;  // array of m_links, excluding the base. index from 0 to num_links-1.
 
+	//
+	// realBuf:
+	//  offset         size            array
+	//   0              6 + num_links   v (base_omega; base_vel; joint_vels)					MULTIDOF [sysdof x sysdof for D matrices (TOO MUCH!) + pos_delta which is sys-cfg sized]
+	//   6+num_links    num_links       D
+	//
+	// vectorBuf:
+	//  offset         size         array
+	//   0              num_links    h_top
+	//   num_links      num_links    h_bottom
+	//
+	// matrixBuf:
+	//  offset         size         array
+	//   0              num_links+1  rot_from_parent
+	//
+	btAlignedObjectArray<btScalar> m_deltaV;
+	btAlignedObjectArray<btScalar> m_realBuf;
+	btAlignedObjectArray<btVector3> m_vectorBuf;
+	btAlignedObjectArray<btMatrix3x3> m_matrixBuf;
 
 	btMatrix3x3 m_cachedInertiaTopLeft;
 	btMatrix3x3 m_cachedInertiaTopRight;
@@ -688,25 +683,25 @@ private:
 	btMatrix3x3 m_cachedInertiaLowerRight;
 	bool m_cachedInertiaValid;
 
-    bool m_fixedBase;
+	bool m_fixedBase;
 
-    // Sleep parameters.
-    bool m_awake;
-    bool m_canSleep;
-    btScalar m_sleepTimer;
+	// Sleep parameters.
+	bool m_awake;
+	bool m_canSleep;
+	btScalar m_sleepTimer;
 
-	void* m_userObjectPointer;
+	void *m_userObjectPointer;
 	int m_userIndex2;
 	int m_userIndex;
 
-	int	m_companionId;
-	btScalar	m_linearDamping;
-	btScalar	m_angularDamping;
-	bool		m_useGyroTerm;
-	btScalar	m_maxAppliedImpulse;
-	btScalar	m_maxCoordinateVelocity;
-	bool		m_hasSelfCollision;
-	
+	int m_companionId;
+	btScalar m_linearDamping;
+	btScalar m_angularDamping;
+	bool m_useGyroTerm;
+	btScalar m_maxAppliedImpulse;
+	btScalar m_maxCoordinateVelocity;
+	bool m_hasSelfCollision;
+
 	bool __posUpdated;
 	int m_dofCount, m_posVarCnt;
 
@@ -720,117 +715,108 @@ private:
 
 struct btMultiBodyLinkDoubleData
 {
-	btQuaternionDoubleData	m_zeroRotParentToThis;
-	btVector3DoubleData		m_parentComToThisPivotOffset;
-	btVector3DoubleData		m_thisPivotToThisComOffset;
-	btVector3DoubleData		m_jointAxisTop[6];
-	btVector3DoubleData		m_jointAxisBottom[6];
+	btQuaternionDoubleData m_zeroRotParentToThis;
+	btVector3DoubleData m_parentComToThisPivotOffset;
+	btVector3DoubleData m_thisPivotToThisComOffset;
+	btVector3DoubleData m_jointAxisTop[6];
+	btVector3DoubleData m_jointAxisBottom[6];
 
-	btVector3DoubleData		m_linkInertia;   // inertia of the base (in local frame; diagonal)
-	btVector3DoubleData		m_absFrameTotVelocityTop;
-	btVector3DoubleData		m_absFrameTotVelocityBottom;
-	btVector3DoubleData		m_absFrameLocVelocityTop;
-	btVector3DoubleData		m_absFrameLocVelocityBottom;
+	btVector3DoubleData m_linkInertia;  // inertia of the base (in local frame; diagonal)
+	btVector3DoubleData m_absFrameTotVelocityTop;
+	btVector3DoubleData m_absFrameTotVelocityBottom;
+	btVector3DoubleData m_absFrameLocVelocityTop;
+	btVector3DoubleData m_absFrameLocVelocityBottom;
 
-	double					m_linkMass;
-	int						m_parentIndex;
-	int						m_jointType;
+	double m_linkMass;
+	int m_parentIndex;
+	int m_jointType;
 
-	int						m_dofCount;
-	int						m_posVarCount;
-	double					m_jointPos[7];
-	double					m_jointVel[6];
-	double					m_jointTorque[6];
+	int m_dofCount;
+	int m_posVarCount;
+	double m_jointPos[7];
+	double m_jointVel[6];
+	double m_jointTorque[6];
 
-	double					m_jointDamping;
-	double					m_jointFriction;
-	double					m_jointLowerLimit;
-	double					m_jointUpperLimit;
-	double					m_jointMaxForce;
-	double					m_jointMaxVelocity;
-	
-	char					*m_linkName;
-	char					*m_jointName;
-	btCollisionObjectDoubleData	*m_linkCollider;
-	char					*m_paddingPtr;
+	double m_jointDamping;
+	double m_jointFriction;
+	double m_jointLowerLimit;
+	double m_jointUpperLimit;
+	double m_jointMaxForce;
+	double m_jointMaxVelocity;
 
+	char *m_linkName;
+	char *m_jointName;
+	btCollisionObjectDoubleData *m_linkCollider;
+	char *m_paddingPtr;
 };
 
 struct btMultiBodyLinkFloatData
 {
-	btQuaternionFloatData	m_zeroRotParentToThis;
-	btVector3FloatData		m_parentComToThisPivotOffset;
-	btVector3FloatData		m_thisPivotToThisComOffset;
-	btVector3FloatData		m_jointAxisTop[6];
-	btVector3FloatData		m_jointAxisBottom[6];
-	btVector3FloatData		m_linkInertia;   // inertia of the base (in local frame; diagonal)
-	btVector3FloatData		m_absFrameTotVelocityTop;
-	btVector3FloatData		m_absFrameTotVelocityBottom;
-	btVector3FloatData		m_absFrameLocVelocityTop;
-	btVector3FloatData		m_absFrameLocVelocityBottom;
+	btQuaternionFloatData m_zeroRotParentToThis;
+	btVector3FloatData m_parentComToThisPivotOffset;
+	btVector3FloatData m_thisPivotToThisComOffset;
+	btVector3FloatData m_jointAxisTop[6];
+	btVector3FloatData m_jointAxisBottom[6];
+	btVector3FloatData m_linkInertia;  // inertia of the base (in local frame; diagonal)
+	btVector3FloatData m_absFrameTotVelocityTop;
+	btVector3FloatData m_absFrameTotVelocityBottom;
+	btVector3FloatData m_absFrameLocVelocityTop;
+	btVector3FloatData m_absFrameLocVelocityBottom;
 
-	int						m_dofCount;
-	float				m_linkMass;
-	int					m_parentIndex;
-	int					m_jointType;
-	
+	int m_dofCount;
+	float m_linkMass;
+	int m_parentIndex;
+	int m_jointType;
 
-		
-	float					m_jointPos[7];
-	float					m_jointVel[6];
-	float					m_jointTorque[6];
-	int						m_posVarCount;
-	float					m_jointDamping;
-	float					m_jointFriction;
-	float					m_jointLowerLimit;
-	float					m_jointUpperLimit;
-	float					m_jointMaxForce;
-	float					m_jointMaxVelocity;
-	
-	char				*m_linkName;
-	char				*m_jointName;
-	btCollisionObjectFloatData	*m_linkCollider;
-	char				*m_paddingPtr;
+	float m_jointPos[7];
+	float m_jointVel[6];
+	float m_jointTorque[6];
+	int m_posVarCount;
+	float m_jointDamping;
+	float m_jointFriction;
+	float m_jointLowerLimit;
+	float m_jointUpperLimit;
+	float m_jointMaxForce;
+	float m_jointMaxVelocity;
 
+	char *m_linkName;
+	char *m_jointName;
+	btCollisionObjectFloatData *m_linkCollider;
+	char *m_paddingPtr;
 };
 
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
-struct	btMultiBodyDoubleData
+struct btMultiBodyDoubleData
 {
 	btVector3DoubleData m_baseWorldPosition;
 	btQuaternionDoubleData m_baseWorldOrientation;
 	btVector3DoubleData m_baseLinearVelocity;
 	btVector3DoubleData m_baseAngularVelocity;
-	btVector3DoubleData m_baseInertia;   // inertia of the base (in local frame; diagonal)
-	double	m_baseMass;
-	int		m_numLinks;
-	char	m_padding[4];
+	btVector3DoubleData m_baseInertia;  // inertia of the base (in local frame; diagonal)
+	double m_baseMass;
+	int m_numLinks;
+	char m_padding[4];
 
-	char	*m_baseName;
-	btMultiBodyLinkDoubleData	*m_links;
-	btCollisionObjectDoubleData	*m_baseCollider;
-	
-	
+	char *m_baseName;
+	btMultiBodyLinkDoubleData *m_links;
+	btCollisionObjectDoubleData *m_baseCollider;
 };
 
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
-struct	btMultiBodyFloatData
+struct btMultiBodyFloatData
 {
 	btVector3FloatData m_baseWorldPosition;
 	btQuaternionFloatData m_baseWorldOrientation;
 	btVector3FloatData m_baseLinearVelocity;
 	btVector3FloatData m_baseAngularVelocity;
 
-	btVector3FloatData m_baseInertia;   // inertia of the base (in local frame; diagonal)
-	float	m_baseMass;
-	int		m_numLinks;
-
-	char	*m_baseName;
-	btMultiBodyLinkFloatData	*m_links;
-	btCollisionObjectFloatData	*m_baseCollider;
+	btVector3FloatData m_baseInertia;  // inertia of the base (in local frame; diagonal)
+	float m_baseMass;
+	int m_numLinks;
 
+	char *m_baseName;
+	btMultiBodyLinkFloatData *m_links;
+	btCollisionObjectFloatData *m_baseCollider;
 };
 
-
-
 #endif
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyConstraint.cpp b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyConstraint.cpp
index 9f61874b837..e17ab94d98d 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyConstraint.cpp
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyConstraint.cpp
@@ -1,32 +1,29 @@
 #include "btMultiBodyConstraint.h"
 #include "BulletDynamics/Dynamics/btRigidBody.h"
-#include "btMultiBodyPoint2Point.h"				//for testing (BTMBP2PCONSTRAINT_BLOCK_ANGULAR_MOTION_TEST macro)
+#include "btMultiBodyPoint2Point.h"  //for testing (BTMBP2PCONSTRAINT_BLOCK_ANGULAR_MOTION_TEST macro)
 
-
-
-btMultiBodyConstraint::btMultiBodyConstraint(btMultiBody* bodyA,btMultiBody* bodyB,int linkA, int linkB, int numRows, bool isUnilateral)
-	:m_bodyA(bodyA),
-	m_bodyB(bodyB),
-	m_linkA(linkA),
-	m_linkB(linkB),
-	m_numRows(numRows),
-	m_jacSizeA(0),
-	m_jacSizeBoth(0),
-	m_isUnilateral(isUnilateral),
-	m_numDofsFinalized(-1),
-	m_maxAppliedImpulse(100)
+btMultiBodyConstraint::btMultiBodyConstraint(btMultiBody* bodyA, btMultiBody* bodyB, int linkA, int linkB, int numRows, bool isUnilateral)
+	: m_bodyA(bodyA),
+	  m_bodyB(bodyB),
+	  m_linkA(linkA),
+	  m_linkB(linkB),
+	  m_numRows(numRows),
+	  m_jacSizeA(0),
+	  m_jacSizeBoth(0),
+	  m_isUnilateral(isUnilateral),
+	  m_numDofsFinalized(-1),
+	  m_maxAppliedImpulse(100)
 {
-
 }
 
 void btMultiBodyConstraint::updateJacobianSizes()
 {
-    if(m_bodyA)
+	if (m_bodyA)
 	{
 		m_jacSizeA = (6 + m_bodyA->getNumDofs());
 	}
 
-	if(m_bodyB)
+	if (m_bodyB)
 	{
 		m_jacSizeBoth = m_jacSizeA + 6 + m_bodyB->getNumDofs();
 	}
@@ -38,7 +35,7 @@ void btMultiBodyConstraint::allocateJacobiansMultiDof()
 {
 	updateJacobianSizes();
 
-	m_posOffset = ((1 + m_jacSizeBoth)*m_numRows);
+	m_posOffset = ((1 + m_jacSizeBoth) * m_numRows);
 	m_data.resize((2 + m_jacSizeBoth) * m_numRows);
 }
 
@@ -46,298 +43,307 @@ btMultiBodyConstraint::~btMultiBodyConstraint()
 {
 }
 
-void	btMultiBodyConstraint::applyDeltaVee(btMultiBodyJacobianData& data, btScalar* delta_vee, btScalar impulse, int velocityIndex, int ndof)
+void btMultiBodyConstraint::applyDeltaVee(btMultiBodyJacobianData& data, btScalar* delta_vee, btScalar impulse, int velocityIndex, int ndof)
 {
 	for (int i = 0; i < ndof; ++i)
-		data.m_deltaVelocities[velocityIndex+i] += delta_vee[i] * impulse;
+		data.m_deltaVelocities[velocityIndex + i] += delta_vee[i] * impulse;
 }
 
-btScalar btMultiBodyConstraint::fillMultiBodyConstraint(	btMultiBodySolverConstraint& solverConstraint,
-                                                        btMultiBodyJacobianData& data,
-                                                        btScalar* jacOrgA, btScalar* jacOrgB,
-                                                        const btVector3& constraintNormalAng,
-                                                        const btVector3& constraintNormalLin,
-                                                        const btVector3& posAworld, const btVector3& posBworld,
-                                                        btScalar posError,
-                                                        const btContactSolverInfo& infoGlobal,
-                                                        btScalar lowerLimit, btScalar upperLimit,
-                                                        bool angConstraint,
-                                                        btScalar relaxation,
-                                                        bool isFriction, btScalar desiredVelocity, btScalar cfmSlip)
+btScalar btMultiBodyConstraint::fillMultiBodyConstraint(btMultiBodySolverConstraint& solverConstraint,
+														btMultiBodyJacobianData& data,
+														btScalar* jacOrgA, btScalar* jacOrgB,
+														const btVector3& constraintNormalAng,
+														const btVector3& constraintNormalLin,
+														const btVector3& posAworld, const btVector3& posBworld,
+														btScalar posError,
+														const btContactSolverInfo& infoGlobal,
+														btScalar lowerLimit, btScalar upperLimit,
+														bool angConstraint,
+														btScalar relaxation,
+														bool isFriction, btScalar desiredVelocity, btScalar cfmSlip)
 {
-    solverConstraint.m_multiBodyA = m_bodyA;
-    solverConstraint.m_multiBodyB = m_bodyB;
-    solverConstraint.m_linkA = m_linkA;
-    solverConstraint.m_linkB = m_linkB;
-    
-    btMultiBody* multiBodyA = solverConstraint.m_multiBodyA;
-    btMultiBody* multiBodyB = solverConstraint.m_multiBodyB;
-    
-    btSolverBody* bodyA = multiBodyA ? 0 : &data.m_solverBodyPool->at(solverConstraint.m_solverBodyIdA);
-    btSolverBody* bodyB = multiBodyB ? 0 : &data.m_solverBodyPool->at(solverConstraint.m_solverBodyIdB);
-    
-    btRigidBody* rb0 = multiBodyA ? 0 : bodyA->m_originalBody;
-    btRigidBody* rb1 = multiBodyB ? 0 : bodyB->m_originalBody;
-    
-    btVector3 rel_pos1, rel_pos2;				//these two used to be inited to posAworld and posBworld (respectively) but it does not seem necessary
-    if (bodyA)
-        rel_pos1 = posAworld - bodyA->getWorldTransform().getOrigin();
-    if (bodyB)
-        rel_pos2 = posBworld - bodyB->getWorldTransform().getOrigin();
-    
-    if (multiBodyA)
-    {
-        if (solverConstraint.m_linkA<0)
-        {
-            rel_pos1 = posAworld - multiBodyA->getBasePos();
-        } else
-        {
-            rel_pos1 = posAworld - multiBodyA->getLink(solverConstraint.m_linkA).m_cachedWorldTransform.getOrigin();
-        }
-        
-        const int ndofA  = multiBodyA->getNumDofs() + 6;
-        
-        solverConstraint.m_deltaVelAindex = multiBodyA->getCompanionId();
-        
-        if (solverConstraint.m_deltaVelAindex <0)
-        {
-            solverConstraint.m_deltaVelAindex = data.m_deltaVelocities.size();
-            multiBodyA->setCompanionId(solverConstraint.m_deltaVelAindex);
-            data.m_deltaVelocities.resize(data.m_deltaVelocities.size()+ndofA);
-        } else
-        {
-            btAssert(data.m_deltaVelocities.size() >= solverConstraint.m_deltaVelAindex+ndofA);
-        }
-        
-        //determine jacobian of this 1D constraint in terms of multibodyA's degrees of freedom
-        //resize..
-        solverConstraint.m_jacAindex = data.m_jacobians.size();
-        data.m_jacobians.resize(data.m_jacobians.size()+ndofA);
-        //copy/determine
-        if(jacOrgA)
-        {
-            for (int i=0;i<ndofA;i++)
-                data.m_jacobians[solverConstraint.m_jacAindex+i] = jacOrgA[i];
-        }
-        else
-        {
-            btScalar* jac1=&data.m_jacobians[solverConstraint.m_jacAindex];
-            //multiBodyA->fillContactJacobianMultiDof(solverConstraint.m_linkA, posAworld, constraintNormalLin, jac1, data.scratch_r, data.scratch_v, data.scratch_m);
-            multiBodyA->fillConstraintJacobianMultiDof(solverConstraint.m_linkA, posAworld, constraintNormalAng, constraintNormalLin, jac1, data.scratch_r, data.scratch_v, data.scratch_m);
-        }
-        
-        //determine the velocity response of multibodyA to reaction impulses of this constraint (i.e. A[i,i] for i=1,...n_con: multibody's inverse inertia with respect to this 1D constraint)
-        //resize..
-        data.m_deltaVelocitiesUnitImpulse.resize(data.m_deltaVelocitiesUnitImpulse.size()+ndofA);		//=> each constraint row has the constrained tree dofs allocated in m_deltaVelocitiesUnitImpulse
-        btAssert(data.m_jacobians.size() == data.m_deltaVelocitiesUnitImpulse.size());
-        btScalar* delta = &data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacAindex];
-        //determine..
-        multiBodyA->calcAccelerationDeltasMultiDof(&data.m_jacobians[solverConstraint.m_jacAindex],delta,data.scratch_r, data.scratch_v);
-        
-        btVector3 torqueAxis0;
-        if (angConstraint) {
-            torqueAxis0 = constraintNormalAng;
-        }
-        else {
-            torqueAxis0 = rel_pos1.cross(constraintNormalLin);
-            
-        }
-        solverConstraint.m_relpos1CrossNormal = torqueAxis0;
-        solverConstraint.m_contactNormal1 = constraintNormalLin;
-    }
-    else //if(rb0)
-    {
-        btVector3 torqueAxis0;
-        if (angConstraint) {
-            torqueAxis0 = constraintNormalAng;
-        }
-        else {
-            torqueAxis0 = rel_pos1.cross(constraintNormalLin);
-        }
-        solverConstraint.m_angularComponentA = rb0 ? rb0->getInvInertiaTensorWorld()*torqueAxis0*rb0->getAngularFactor() : btVector3(0,0,0);
-        solverConstraint.m_relpos1CrossNormal = torqueAxis0;
-        solverConstraint.m_contactNormal1 = constraintNormalLin;
-    }
-    
-    if (multiBodyB)
-    {
-        if (solverConstraint.m_linkB<0)
-        {
-            rel_pos2 = posBworld - multiBodyB->getBasePos();
-        } else
-        {
-            rel_pos2 = posBworld - multiBodyB->getLink(solverConstraint.m_linkB).m_cachedWorldTransform.getOrigin();
-        }
-        
-        const int ndofB  = multiBodyB->getNumDofs() + 6;
-        
-        solverConstraint.m_deltaVelBindex = multiBodyB->getCompanionId();
-        if (solverConstraint.m_deltaVelBindex <0)
-        {
-            solverConstraint.m_deltaVelBindex = data.m_deltaVelocities.size();
-            multiBodyB->setCompanionId(solverConstraint.m_deltaVelBindex);
-            data.m_deltaVelocities.resize(data.m_deltaVelocities.size()+ndofB);
-        }
-        
-        //determine jacobian of this 1D constraint in terms of multibodyB's degrees of freedom
-        //resize..
-        solverConstraint.m_jacBindex = data.m_jacobians.size();
-        data.m_jacobians.resize(data.m_jacobians.size()+ndofB);
-        //copy/determine..
-        if(jacOrgB)
-        {
-            for (int i=0;i<ndofB;i++)
-                data.m_jacobians[solverConstraint.m_jacBindex+i] = jacOrgB[i];
-        }
-        else
-        {
-            //multiBodyB->fillContactJacobianMultiDof(solverConstraint.m_linkB, posBworld, -constraintNormalLin, &data.m_jacobians[solverConstraint.m_jacBindex], data.scratch_r, data.scratch_v, data.scratch_m);
-            multiBodyB->fillConstraintJacobianMultiDof(solverConstraint.m_linkB, posBworld, -constraintNormalAng, -constraintNormalLin, &data.m_jacobians[solverConstraint.m_jacBindex], data.scratch_r, data.scratch_v, data.scratch_m);
-        }
-        
-        //determine velocity response of multibodyB to reaction impulses of this constraint (i.e. A[i,i] for i=1,...n_con: multibody's inverse inertia with respect to this 1D constraint)
-        //resize..
-        data.m_deltaVelocitiesUnitImpulse.resize(data.m_deltaVelocitiesUnitImpulse.size()+ndofB);
-        btAssert(data.m_jacobians.size() == data.m_deltaVelocitiesUnitImpulse.size());
-        btScalar* delta = &data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacBindex];
-        //determine..
-        multiBodyB->calcAccelerationDeltasMultiDof(&data.m_jacobians[solverConstraint.m_jacBindex],delta,data.scratch_r, data.scratch_v);
-        
-        btVector3 torqueAxis1;
-        if (angConstraint) {
-            torqueAxis1 = constraintNormalAng;
-        }
-        else {
-            torqueAxis1 = rel_pos2.cross(constraintNormalLin);
-        }
-        solverConstraint.m_relpos2CrossNormal = -torqueAxis1;
-        solverConstraint.m_contactNormal2 = -constraintNormalLin;
-    }
-    else //if(rb1)
-    {
-        btVector3 torqueAxis1;
-        if (angConstraint) {
-            torqueAxis1 = constraintNormalAng;
-        }
-        else {
-            torqueAxis1 = rel_pos2.cross(constraintNormalLin);
-        }
-        solverConstraint.m_angularComponentB = rb1 ? rb1->getInvInertiaTensorWorld()*-torqueAxis1*rb1->getAngularFactor() : btVector3(0,0,0);
-        solverConstraint.m_relpos2CrossNormal = -torqueAxis1;
-        solverConstraint.m_contactNormal2 = -constraintNormalLin;
-    }
-    {
-        
-        btVector3 vec;
-        btScalar denom0 = 0.f;
-        btScalar denom1 = 0.f;
-        btScalar* jacB = 0;
-        btScalar* jacA = 0;
-        btScalar* deltaVelA = 0;
-        btScalar* deltaVelB = 0;
-        int ndofA  = 0;
-        //determine the "effective mass" of the constrained multibodyA with respect to this 1D constraint (i.e. 1/A[i,i])
-        if (multiBodyA)
-        {
-            ndofA = multiBodyA->getNumDofs() + 6;
-            jacA = &data.m_jacobians[solverConstraint.m_jacAindex];
-            deltaVelA = &data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacAindex];
-            for (int i = 0; i < ndofA; ++i)
-            {
-                btScalar j = jacA[i] ;
-                btScalar l = deltaVelA[i];
-                denom0 += j*l;
-            }
-        }
-        else if(rb0)
-        {
-            vec = ( solverConstraint.m_angularComponentA).cross(rel_pos1);
-            if (angConstraint) {
+	solverConstraint.m_multiBodyA = m_bodyA;
+	solverConstraint.m_multiBodyB = m_bodyB;
+	solverConstraint.m_linkA = m_linkA;
+	solverConstraint.m_linkB = m_linkB;
+
+	btMultiBody* multiBodyA = solverConstraint.m_multiBodyA;
+	btMultiBody* multiBodyB = solverConstraint.m_multiBodyB;
+
+	btSolverBody* bodyA = multiBodyA ? 0 : &data.m_solverBodyPool->at(solverConstraint.m_solverBodyIdA);
+	btSolverBody* bodyB = multiBodyB ? 0 : &data.m_solverBodyPool->at(solverConstraint.m_solverBodyIdB);
+
+	btRigidBody* rb0 = multiBodyA ? 0 : bodyA->m_originalBody;
+	btRigidBody* rb1 = multiBodyB ? 0 : bodyB->m_originalBody;
+
+	btVector3 rel_pos1, rel_pos2;  //these two used to be inited to posAworld and posBworld (respectively) but it does not seem necessary
+	if (bodyA)
+		rel_pos1 = posAworld - bodyA->getWorldTransform().getOrigin();
+	if (bodyB)
+		rel_pos2 = posBworld - bodyB->getWorldTransform().getOrigin();
+
+	if (multiBodyA)
+	{
+		if (solverConstraint.m_linkA < 0)
+		{
+			rel_pos1 = posAworld - multiBodyA->getBasePos();
+		}
+		else
+		{
+			rel_pos1 = posAworld - multiBodyA->getLink(solverConstraint.m_linkA).m_cachedWorldTransform.getOrigin();
+		}
+
+		const int ndofA = multiBodyA->getNumDofs() + 6;
+
+		solverConstraint.m_deltaVelAindex = multiBodyA->getCompanionId();
+
+		if (solverConstraint.m_deltaVelAindex < 0)
+		{
+			solverConstraint.m_deltaVelAindex = data.m_deltaVelocities.size();
+			multiBodyA->setCompanionId(solverConstraint.m_deltaVelAindex);
+			data.m_deltaVelocities.resize(data.m_deltaVelocities.size() + ndofA);
+		}
+		else
+		{
+			btAssert(data.m_deltaVelocities.size() >= solverConstraint.m_deltaVelAindex + ndofA);
+		}
+
+		//determine jacobian of this 1D constraint in terms of multibodyA's degrees of freedom
+		//resize..
+		solverConstraint.m_jacAindex = data.m_jacobians.size();
+		data.m_jacobians.resize(data.m_jacobians.size() + ndofA);
+		//copy/determine
+		if (jacOrgA)
+		{
+			for (int i = 0; i < ndofA; i++)
+				data.m_jacobians[solverConstraint.m_jacAindex + i] = jacOrgA[i];
+		}
+		else
+		{
+			btScalar* jac1 = &data.m_jacobians[solverConstraint.m_jacAindex];
+			//multiBodyA->fillContactJacobianMultiDof(solverConstraint.m_linkA, posAworld, constraintNormalLin, jac1, data.scratch_r, data.scratch_v, data.scratch_m);
+			multiBodyA->fillConstraintJacobianMultiDof(solverConstraint.m_linkA, posAworld, constraintNormalAng, constraintNormalLin, jac1, data.scratch_r, data.scratch_v, data.scratch_m);
+		}
+
+		//determine the velocity response of multibodyA to reaction impulses of this constraint (i.e. A[i,i] for i=1,...n_con: multibody's inverse inertia with respect to this 1D constraint)
+		//resize..
+		data.m_deltaVelocitiesUnitImpulse.resize(data.m_deltaVelocitiesUnitImpulse.size() + ndofA);  //=> each constraint row has the constrained tree dofs allocated in m_deltaVelocitiesUnitImpulse
+		btAssert(data.m_jacobians.size() == data.m_deltaVelocitiesUnitImpulse.size());
+		btScalar* delta = &data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacAindex];
+		//determine..
+		multiBodyA->calcAccelerationDeltasMultiDof(&data.m_jacobians[solverConstraint.m_jacAindex], delta, data.scratch_r, data.scratch_v);
+
+		btVector3 torqueAxis0;
+		if (angConstraint)
+		{
+			torqueAxis0 = constraintNormalAng;
+		}
+		else
+		{
+			torqueAxis0 = rel_pos1.cross(constraintNormalLin);
+		}
+		solverConstraint.m_relpos1CrossNormal = torqueAxis0;
+		solverConstraint.m_contactNormal1 = constraintNormalLin;
+	}
+	else  //if(rb0)
+	{
+		btVector3 torqueAxis0;
+		if (angConstraint)
+		{
+			torqueAxis0 = constraintNormalAng;
+		}
+		else
+		{
+			torqueAxis0 = rel_pos1.cross(constraintNormalLin);
+		}
+		solverConstraint.m_angularComponentA = rb0 ? rb0->getInvInertiaTensorWorld() * torqueAxis0 * rb0->getAngularFactor() : btVector3(0, 0, 0);
+		solverConstraint.m_relpos1CrossNormal = torqueAxis0;
+		solverConstraint.m_contactNormal1 = constraintNormalLin;
+	}
+
+	if (multiBodyB)
+	{
+		if (solverConstraint.m_linkB < 0)
+		{
+			rel_pos2 = posBworld - multiBodyB->getBasePos();
+		}
+		else
+		{
+			rel_pos2 = posBworld - multiBodyB->getLink(solverConstraint.m_linkB).m_cachedWorldTransform.getOrigin();
+		}
+
+		const int ndofB = multiBodyB->getNumDofs() + 6;
+
+		solverConstraint.m_deltaVelBindex = multiBodyB->getCompanionId();
+		if (solverConstraint.m_deltaVelBindex < 0)
+		{
+			solverConstraint.m_deltaVelBindex = data.m_deltaVelocities.size();
+			multiBodyB->setCompanionId(solverConstraint.m_deltaVelBindex);
+			data.m_deltaVelocities.resize(data.m_deltaVelocities.size() + ndofB);
+		}
+
+		//determine jacobian of this 1D constraint in terms of multibodyB's degrees of freedom
+		//resize..
+		solverConstraint.m_jacBindex = data.m_jacobians.size();
+		data.m_jacobians.resize(data.m_jacobians.size() + ndofB);
+		//copy/determine..
+		if (jacOrgB)
+		{
+			for (int i = 0; i < ndofB; i++)
+				data.m_jacobians[solverConstraint.m_jacBindex + i] = jacOrgB[i];
+		}
+		else
+		{
+			//multiBodyB->fillContactJacobianMultiDof(solverConstraint.m_linkB, posBworld, -constraintNormalLin, &data.m_jacobians[solverConstraint.m_jacBindex], data.scratch_r, data.scratch_v, data.scratch_m);
+			multiBodyB->fillConstraintJacobianMultiDof(solverConstraint.m_linkB, posBworld, -constraintNormalAng, -constraintNormalLin, &data.m_jacobians[solverConstraint.m_jacBindex], data.scratch_r, data.scratch_v, data.scratch_m);
+		}
+
+		//determine velocity response of multibodyB to reaction impulses of this constraint (i.e. A[i,i] for i=1,...n_con: multibody's inverse inertia with respect to this 1D constraint)
+		//resize..
+		data.m_deltaVelocitiesUnitImpulse.resize(data.m_deltaVelocitiesUnitImpulse.size() + ndofB);
+		btAssert(data.m_jacobians.size() == data.m_deltaVelocitiesUnitImpulse.size());
+		btScalar* delta = &data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacBindex];
+		//determine..
+		multiBodyB->calcAccelerationDeltasMultiDof(&data.m_jacobians[solverConstraint.m_jacBindex], delta, data.scratch_r, data.scratch_v);
+
+		btVector3 torqueAxis1;
+		if (angConstraint)
+		{
+			torqueAxis1 = constraintNormalAng;
+		}
+		else
+		{
+			torqueAxis1 = rel_pos2.cross(constraintNormalLin);
+		}
+		solverConstraint.m_relpos2CrossNormal = -torqueAxis1;
+		solverConstraint.m_contactNormal2 = -constraintNormalLin;
+	}
+	else  //if(rb1)
+	{
+		btVector3 torqueAxis1;
+		if (angConstraint)
+		{
+			torqueAxis1 = constraintNormalAng;
+		}
+		else
+		{
+			torqueAxis1 = rel_pos2.cross(constraintNormalLin);
+		}
+		solverConstraint.m_angularComponentB = rb1 ? rb1->getInvInertiaTensorWorld() * -torqueAxis1 * rb1->getAngularFactor() : btVector3(0, 0, 0);
+		solverConstraint.m_relpos2CrossNormal = -torqueAxis1;
+		solverConstraint.m_contactNormal2 = -constraintNormalLin;
+	}
+	{
+		btVector3 vec;
+		btScalar denom0 = 0.f;
+		btScalar denom1 = 0.f;
+		btScalar* jacB = 0;
+		btScalar* jacA = 0;
+		btScalar* deltaVelA = 0;
+		btScalar* deltaVelB = 0;
+		int ndofA = 0;
+		//determine the "effective mass" of the constrained multibodyA with respect to this 1D constraint (i.e. 1/A[i,i])
+		if (multiBodyA)
+		{
+			ndofA = multiBodyA->getNumDofs() + 6;
+			jacA = &data.m_jacobians[solverConstraint.m_jacAindex];
+			deltaVelA = &data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacAindex];
+			for (int i = 0; i < ndofA; ++i)
+			{
+				btScalar j = jacA[i];
+				btScalar l = deltaVelA[i];
+				denom0 += j * l;
+			}
+		}
+		else if (rb0)
+		{
+			vec = (solverConstraint.m_angularComponentA).cross(rel_pos1);
+			if (angConstraint)
+			{
 				denom0 = constraintNormalAng.dot(solverConstraint.m_angularComponentA);
-            }
-            else {
-                denom0 = rb0->getInvMass() + constraintNormalLin.dot(vec);
-            }
-        }
-        //
-        if (multiBodyB)
-        {
-            const int ndofB = multiBodyB->getNumDofs() + 6;
-            jacB = &data.m_jacobians[solverConstraint.m_jacBindex];
-            deltaVelB = &data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacBindex];
-            for (int i = 0; i < ndofB; ++i)
-            {
-                btScalar j = jacB[i] ;
-                btScalar l = deltaVelB[i];
-                denom1 += j*l;
-            }
-            
-        }
-        else if(rb1)
-        {
-            vec = ( -solverConstraint.m_angularComponentB).cross(rel_pos2);
-            if (angConstraint) {
+			}
+			else
+			{
+				denom0 = rb0->getInvMass() + constraintNormalLin.dot(vec);
+			}
+		}
+		//
+		if (multiBodyB)
+		{
+			const int ndofB = multiBodyB->getNumDofs() + 6;
+			jacB = &data.m_jacobians[solverConstraint.m_jacBindex];
+			deltaVelB = &data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacBindex];
+			for (int i = 0; i < ndofB; ++i)
+			{
+				btScalar j = jacB[i];
+				btScalar l = deltaVelB[i];
+				denom1 += j * l;
+			}
+		}
+		else if (rb1)
+		{
+			vec = (-solverConstraint.m_angularComponentB).cross(rel_pos2);
+			if (angConstraint)
+			{
 				denom1 = constraintNormalAng.dot(-solverConstraint.m_angularComponentB);
-            }
-            else {
-                denom1 = rb1->getInvMass() + constraintNormalLin.dot(vec);
-            }
-        }
-        
-        //
-        btScalar d = denom0+denom1;
-        if (d>SIMD_EPSILON)
-        {
-            solverConstraint.m_jacDiagABInv = relaxation/(d);
-        }
-        else
-        {
-            //disable the constraint row to handle singularity/redundant constraint
-            solverConstraint.m_jacDiagABInv  = 0.f;
-        }
-    }
-    
-    
-    //compute rhs and remaining solverConstraint fields
-    btScalar penetration = isFriction? 0 : posError;
-    
-    btScalar rel_vel = 0.f;
-    int ndofA  = 0;
-    int ndofB  = 0;
-    {
-        btVector3 vel1,vel2;
-        if (multiBodyA)
-        {
-            ndofA = multiBodyA->getNumDofs() + 6;
-            btScalar* jacA = &data.m_jacobians[solverConstraint.m_jacAindex];
-            for (int i = 0; i < ndofA ; ++i)
-                rel_vel += multiBodyA->getVelocityVector()[i] * jacA[i];
-        }
-        else if(rb0)
-        {
+			}
+			else
+			{
+				denom1 = rb1->getInvMass() + constraintNormalLin.dot(vec);
+			}
+		}
+
+		//
+		btScalar d = denom0 + denom1;
+		if (d > SIMD_EPSILON)
+		{
+			solverConstraint.m_jacDiagABInv = relaxation / (d);
+		}
+		else
+		{
+			//disable the constraint row to handle singularity/redundant constraint
+			solverConstraint.m_jacDiagABInv = 0.f;
+		}
+	}
+
+	//compute rhs and remaining solverConstraint fields
+	btScalar penetration = isFriction ? 0 : posError;
+
+	btScalar rel_vel = 0.f;
+	int ndofA = 0;
+	int ndofB = 0;
+	{
+		btVector3 vel1, vel2;
+		if (multiBodyA)
+		{
+			ndofA = multiBodyA->getNumDofs() + 6;
+			btScalar* jacA = &data.m_jacobians[solverConstraint.m_jacAindex];
+			for (int i = 0; i < ndofA; ++i)
+				rel_vel += multiBodyA->getVelocityVector()[i] * jacA[i];
+		}
+		else if (rb0)
+		{
 			rel_vel += rb0->getLinearVelocity().dot(solverConstraint.m_contactNormal1);
 			rel_vel += rb0->getAngularVelocity().dot(solverConstraint.m_relpos1CrossNormal);
-        }
-        if (multiBodyB)
-        {
-            ndofB = multiBodyB->getNumDofs() + 6;
-            btScalar* jacB = &data.m_jacobians[solverConstraint.m_jacBindex];
-            for (int i = 0; i < ndofB ; ++i)
-                rel_vel += multiBodyB->getVelocityVector()[i] * jacB[i];
-            
-        }
-        else if(rb1)
-        {
+		}
+		if (multiBodyB)
+		{
+			ndofB = multiBodyB->getNumDofs() + 6;
+			btScalar* jacB = &data.m_jacobians[solverConstraint.m_jacBindex];
+			for (int i = 0; i < ndofB; ++i)
+				rel_vel += multiBodyB->getVelocityVector()[i] * jacB[i];
+		}
+		else if (rb1)
+		{
 			rel_vel += rb1->getLinearVelocity().dot(solverConstraint.m_contactNormal2);
 			rel_vel += rb1->getAngularVelocity().dot(solverConstraint.m_relpos2CrossNormal);
-        }
-        
-        solverConstraint.m_friction = 0.f;//cp.m_combinedFriction;
-    }
-    
-    
-    ///warm starting (or zero if disabled)
-    /*
+		}
+
+		solverConstraint.m_friction = 0.f;  //cp.m_combinedFriction;
+	}
+
+	///warm starting (or zero if disabled)
+	/*
      if (infoGlobal.m_solverMode & SOLVER_USE_WARMSTARTING)
      {
      solverConstraint.m_appliedImpulse = isFriction ? 0 : cp.m_appliedImpulse * infoGlobal.m_warmstartingFactor;
@@ -369,38 +375,35 @@ btScalar btMultiBodyConstraint::fillMultiBodyConstraint(	btMultiBodySolverConstr
      }
      } else
      */
-    
-    solverConstraint.m_appliedImpulse = 0.f;
-    solverConstraint.m_appliedPushImpulse = 0.f;
-    
-    {
-        
-        btScalar positionalError = 0.f;
-        btScalar	velocityError = desiredVelocity - rel_vel;// * damping;
-        
-        
-        btScalar erp = infoGlobal.m_erp2;
-		
+
+	solverConstraint.m_appliedImpulse = 0.f;
+	solverConstraint.m_appliedPushImpulse = 0.f;
+
+	{
+		btScalar positionalError = 0.f;
+		btScalar velocityError = desiredVelocity - rel_vel;  // * damping;
+
+		btScalar erp = infoGlobal.m_erp2;
+
 		//split impulse is not implemented yet for btMultiBody*
 		//if (!infoGlobal.m_splitImpulse || (penetration > infoGlobal.m_splitImpulsePenetrationThreshold))
-        {
-            erp = infoGlobal.m_erp;
-        }
-        
-        positionalError = -penetration * erp/infoGlobal.m_timeStep;
-        
-        btScalar  penetrationImpulse = positionalError*solverConstraint.m_jacDiagABInv;
-        btScalar velocityImpulse = velocityError *solverConstraint.m_jacDiagABInv;
-        
+		{
+			erp = infoGlobal.m_erp;
+		}
+
+		positionalError = -penetration * erp / infoGlobal.m_timeStep;
+
+		btScalar penetrationImpulse = positionalError * solverConstraint.m_jacDiagABInv;
+		btScalar velocityImpulse = velocityError * solverConstraint.m_jacDiagABInv;
+
 		//split impulse is not implemented yet for btMultiBody*
 
-      //  if (!infoGlobal.m_splitImpulse || (penetration > infoGlobal.m_splitImpulsePenetrationThreshold))
-        {
-            //combine position and velocity into rhs
-            solverConstraint.m_rhs = penetrationImpulse+velocityImpulse;
-            solverConstraint.m_rhsPenetration = 0.f;
-            
-        } 
+		//  if (!infoGlobal.m_splitImpulse || (penetration > infoGlobal.m_splitImpulsePenetrationThreshold))
+		{
+			//combine position and velocity into rhs
+			solverConstraint.m_rhs = penetrationImpulse + velocityImpulse;
+			solverConstraint.m_rhsPenetration = 0.f;
+		}
 		/*else
         {
             //split position and velocity into rhs and m_rhsPenetration
@@ -409,11 +412,10 @@ btScalar btMultiBodyConstraint::fillMultiBodyConstraint(	btMultiBodySolverConstr
         }
         */
 
-        solverConstraint.m_cfm = 0.f;
-        solverConstraint.m_lowerLimit = lowerLimit;
-        solverConstraint.m_upperLimit = upperLimit;
-    }
-    
-    return rel_vel;
-    
+		solverConstraint.m_cfm = 0.f;
+		solverConstraint.m_lowerLimit = lowerLimit;
+		solverConstraint.m_upperLimit = upperLimit;
+	}
+
+	return rel_vel;
 }
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyConstraint.h b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyConstraint.h
index a2ae5712733..5c15f3e8512 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyConstraint.h
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyConstraint.h
@@ -27,66 +27,62 @@ struct btSolverInfo;
 
 struct btMultiBodyJacobianData
 {
-	btAlignedObjectArray<btScalar>		m_jacobians;
-	btAlignedObjectArray<btScalar>		m_deltaVelocitiesUnitImpulse;	//holds the joint-space response of the corresp. tree to the test impulse in each constraint space dimension
-	btAlignedObjectArray<btScalar>		m_deltaVelocities;				//holds joint-space vectors of all the constrained trees accumulating the effect of corrective impulses applied in SI
-	btAlignedObjectArray<btScalar>		scratch_r;
-	btAlignedObjectArray<btVector3>		scratch_v;
-	btAlignedObjectArray<btMatrix3x3>	scratch_m;
-	btAlignedObjectArray<btSolverBody>*	m_solverBodyPool;
-	int									m_fixedBodyId;
-
+	btAlignedObjectArray<btScalar> m_jacobians;
+	btAlignedObjectArray<btScalar> m_deltaVelocitiesUnitImpulse;  //holds the joint-space response of the corresp. tree to the test impulse in each constraint space dimension
+	btAlignedObjectArray<btScalar> m_deltaVelocities;             //holds joint-space vectors of all the constrained trees accumulating the effect of corrective impulses applied in SI
+	btAlignedObjectArray<btScalar> scratch_r;
+	btAlignedObjectArray<btVector3> scratch_v;
+	btAlignedObjectArray<btMatrix3x3> scratch_m;
+	btAlignedObjectArray<btSolverBody>* m_solverBodyPool;
+	int m_fixedBodyId;
 };
 
-
-ATTRIBUTE_ALIGNED16(class) btMultiBodyConstraint
+ATTRIBUTE_ALIGNED16(class)
+btMultiBodyConstraint
 {
 protected:
+	btMultiBody* m_bodyA;
+	btMultiBody* m_bodyB;
+	int m_linkA;
+	int m_linkB;
 
-	btMultiBody*	m_bodyA;
-    btMultiBody*	m_bodyB;
-    int				m_linkA;
-    int				m_linkB;
+	int m_numRows;
+	int m_jacSizeA;
+	int m_jacSizeBoth;
+	int m_posOffset;
 
-    int				m_numRows;
-    int				m_jacSizeA;
-    int				m_jacSizeBoth;
-    int				m_posOffset;
+	bool m_isUnilateral;
+	int m_numDofsFinalized;
+	btScalar m_maxAppliedImpulse;
 
-	bool			m_isUnilateral;
-	int				m_numDofsFinalized;
-	btScalar		m_maxAppliedImpulse;
+	// warning: the data block lay out is not consistent for all constraints
+	// data block laid out as follows:
+	// cached impulses. (one per row.)
+	// jacobians. (interleaved, row1 body1 then row1 body2 then row2 body 1 etc)
+	// positions. (one per row.)
+	btAlignedObjectArray<btScalar> m_data;
 
+	void applyDeltaVee(btMultiBodyJacobianData & data, btScalar * delta_vee, btScalar impulse, int velocityIndex, int ndof);
 
-    // warning: the data block lay out is not consistent for all constraints
-    // data block laid out as follows:
-    // cached impulses. (one per row.)
-    // jacobians. (interleaved, row1 body1 then row1 body2 then row2 body 1 etc)
-    // positions. (one per row.)
-    btAlignedObjectArray<btScalar> m_data;
+	btScalar fillMultiBodyConstraint(btMultiBodySolverConstraint & solverConstraint,
+									 btMultiBodyJacobianData & data,
+									 btScalar * jacOrgA, btScalar * jacOrgB,
+									 const btVector3& constraintNormalAng,
 
-	void	applyDeltaVee(btMultiBodyJacobianData& data, btScalar* delta_vee, btScalar impulse, int velocityIndex, int ndof);
-    
-	btScalar fillMultiBodyConstraint(btMultiBodySolverConstraint& solverConstraint,
-																btMultiBodyJacobianData& data,
-                                     btScalar* jacOrgA, btScalar* jacOrgB,
-                                     const btVector3& constraintNormalAng,
-                                     
-																const btVector3& constraintNormalLin,
-																const btVector3& posAworld, const btVector3& posBworld,
-																btScalar posError,
-																const btContactSolverInfo& infoGlobal,
-                                     btScalar lowerLimit, btScalar upperLimit,
-                                     bool angConstraint = false,
-                                     
-																btScalar relaxation = 1.f,
-																bool isFriction = false, btScalar desiredVelocity=0, btScalar cfmSlip=0);
+									 const btVector3& constraintNormalLin,
+									 const btVector3& posAworld, const btVector3& posBworld,
+									 btScalar posError,
+									 const btContactSolverInfo& infoGlobal,
+									 btScalar lowerLimit, btScalar upperLimit,
+									 bool angConstraint = false,
+
+									 btScalar relaxation = 1.f,
+									 bool isFriction = false, btScalar desiredVelocity = 0, btScalar cfmSlip = 0);
 
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-	btMultiBodyConstraint(btMultiBody* bodyA,btMultiBody* bodyB,int linkA, int linkB, int numRows, bool isUnilateral);
+	btMultiBodyConstraint(btMultiBody * bodyA, btMultiBody * bodyB, int linkA, int linkB, int numRows, bool isUnilateral);
 	virtual ~btMultiBodyConstraint();
 
 	void updateJacobianSizes();
@@ -94,27 +90,27 @@ public:
 
 	//many constraints have setFrameInB/setPivotInB. Will use 'getConstraintType' later.
 	virtual void setFrameInB(const btMatrix3x3& frameInB) {}
-	virtual void setPivotInB(const btVector3& pivotInB){}
+	virtual void setPivotInB(const btVector3& pivotInB) {}
 
-	virtual void finalizeMultiDof()=0;
+	virtual void finalizeMultiDof() = 0;
 
-	virtual int getIslandIdA() const =0;
-	virtual int getIslandIdB() const =0;
+	virtual int getIslandIdA() const = 0;
+	virtual int getIslandIdB() const = 0;
 
-	virtual void createConstraintRows(btMultiBodyConstraintArray& constraintRows,
-		btMultiBodyJacobianData& data,
-		const btContactSolverInfo& infoGlobal)=0;
+	virtual void createConstraintRows(btMultiBodyConstraintArray & constraintRows,
+									  btMultiBodyJacobianData & data,
+									  const btContactSolverInfo& infoGlobal) = 0;
 
-	int	getNumRows() const
+	int getNumRows() const
 	{
 		return m_numRows;
 	}
 
-	btMultiBody*	getMultiBodyA()
+	btMultiBody* getMultiBodyA()
 	{
 		return m_bodyA;
 	}
-    btMultiBody*	getMultiBodyB()
+	btMultiBody* getMultiBodyB()
 	{
 		return m_bodyB;
 	}
@@ -127,77 +123,72 @@ public:
 	{
 		return m_linkB;
 	}
-	void	internalSetAppliedImpulse(int dof, btScalar appliedImpulse)
+	void internalSetAppliedImpulse(int dof, btScalar appliedImpulse)
 	{
-		btAssert(dof>=0);
+		btAssert(dof >= 0);
 		btAssert(dof < getNumRows());
 		m_data[dof] = appliedImpulse;
-        
 	}
-	
-	btScalar	getAppliedImpulse(int dof)
+
+	btScalar getAppliedImpulse(int dof)
 	{
-		btAssert(dof>=0);
+		btAssert(dof >= 0);
 		btAssert(dof < getNumRows());
 		return m_data[dof];
 	}
 	// current constraint position
-    // constraint is pos >= 0 for unilateral, or pos = 0 for bilateral
-    // NOTE: ignored position for friction rows.
-    btScalar getPosition(int row) const
+	// constraint is pos >= 0 for unilateral, or pos = 0 for bilateral
+	// NOTE: ignored position for friction rows.
+	btScalar getPosition(int row) const
 	{
 		return m_data[m_posOffset + row];
 	}
 
-    void setPosition(int row, btScalar pos)
+	void setPosition(int row, btScalar pos)
 	{
 		m_data[m_posOffset + row] = pos;
 	}
 
-
 	bool isUnilateral() const
 	{
 		return m_isUnilateral;
 	}
 
 	// jacobian blocks.
-    // each of size 6 + num_links. (jacobian2 is null if no body2.)
-    // format: 3 'omega' coefficients, 3 'v' coefficients, then the 'qdot' coefficients.
-    btScalar* jacobianA(int row)
+	// each of size 6 + num_links. (jacobian2 is null if no body2.)
+	// format: 3 'omega' coefficients, 3 'v' coefficients, then the 'qdot' coefficients.
+	btScalar* jacobianA(int row)
 	{
 		return &m_data[m_numRows + row * m_jacSizeBoth];
 	}
-    const btScalar* jacobianA(int row) const
+	const btScalar* jacobianA(int row) const
 	{
 		return &m_data[m_numRows + (row * m_jacSizeBoth)];
 	}
-    btScalar* jacobianB(int row)
+	btScalar* jacobianB(int row)
 	{
 		return &m_data[m_numRows + (row * m_jacSizeBoth) + m_jacSizeA];
 	}
-    const btScalar* jacobianB(int row) const
+	const btScalar* jacobianB(int row) const
 	{
 		return &m_data[m_numRows + (row * m_jacSizeBoth) + m_jacSizeA];
 	}
 
-	btScalar	getMaxAppliedImpulse() const
+	btScalar getMaxAppliedImpulse() const
 	{
 		return m_maxAppliedImpulse;
 	}
-	void	setMaxAppliedImpulse(btScalar maxImp)
+	void setMaxAppliedImpulse(btScalar maxImp)
 	{
 		m_maxAppliedImpulse = maxImp;
 	}
 
-	virtual void debugDraw(class btIDebugDraw* drawer)=0;
+	virtual void debugDraw(class btIDebugDraw * drawer) = 0;
 
 	virtual void setGearRatio(btScalar ratio) {}
 	virtual void setGearAuxLink(int gearAuxLink) {}
-	virtual void setRelativePositionTarget(btScalar relPosTarget){}
-	virtual void setErp(btScalar erp){}
-	
-	
+	virtual void setRelativePositionTarget(btScalar relPosTarget) {}
+	virtual void setErp(btScalar erp) {}
 };
 
-#endif //BT_MULTIBODY_CONSTRAINT_H
-
+#endif  //BT_MULTIBODY_CONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyConstraintSolver.cpp b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyConstraintSolver.cpp
index cd84826e1a2..e97bd71cc41 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyConstraintSolver.cpp
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyConstraintSolver.cpp
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btMultiBodyConstraintSolver.h"
 #include "BulletCollision/NarrowPhaseCollision/btPersistentManifold.h"
 #include "btMultiBodyLinkCollider.h"
@@ -24,33 +23,33 @@ subject to the following restrictions:
 
 #include "LinearMath/btQuickprof.h"
 
-btScalar btMultiBodyConstraintSolver::solveSingleIteration(int iteration, btCollisionObject** bodies ,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer)
+btScalar btMultiBodyConstraintSolver::solveSingleIteration(int iteration, btCollisionObject** bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer)
 {
-	btScalar leastSquaredResidual  = btSequentialImpulseConstraintSolver::solveSingleIteration(iteration, bodies ,numBodies,manifoldPtr, numManifolds,constraints,numConstraints,infoGlobal,debugDrawer);
-	
+	btScalar leastSquaredResidual = btSequentialImpulseConstraintSolver::solveSingleIteration(iteration, bodies, numBodies, manifoldPtr, numManifolds, constraints, numConstraints, infoGlobal, debugDrawer);
+
 	//solve featherstone non-contact constraints
 
 	//printf("m_multiBodyNonContactConstraints = %d\n",m_multiBodyNonContactConstraints.size());
 
-	for (int j=0;j<m_multiBodyNonContactConstraints.size();j++)
+	for (int j = 0; j < m_multiBodyNonContactConstraints.size(); j++)
 	{
-		int index = iteration&1? j : m_multiBodyNonContactConstraints.size()-1-j;
+		int index = iteration & 1 ? j : m_multiBodyNonContactConstraints.size() - 1 - j;
 
 		btMultiBodySolverConstraint& constraint = m_multiBodyNonContactConstraints[index];
-		
-		btScalar residual = resolveSingleConstraintRowGeneric(constraint);
-		leastSquaredResidual = btMax(leastSquaredResidual,residual*residual);
 
-		if(constraint.m_multiBodyA) 
+		btScalar residual = resolveSingleConstraintRowGeneric(constraint);
+		leastSquaredResidual = btMax(leastSquaredResidual, residual * residual);
+
+		if (constraint.m_multiBodyA)
 			constraint.m_multiBodyA->setPosUpdated(false);
-		if(constraint.m_multiBodyB) 
+		if (constraint.m_multiBodyB)
 			constraint.m_multiBodyB->setPosUpdated(false);
 	}
 
 	//solve featherstone normal contact
-	for (int j0=0;j0<m_multiBodyNormalContactConstraints.size();j0++)
+	for (int j0 = 0; j0 < m_multiBodyNormalContactConstraints.size(); j0++)
 	{
-		int index = j0;//iteration&1? j0 : m_multiBodyNormalContactConstraints.size()-1-j0;
+		int index = j0;  //iteration&1? j0 : m_multiBodyNormalContactConstraints.size()-1-j0;
 
 		btMultiBodySolverConstraint& constraint = m_multiBodyNormalContactConstraints[index];
 		btScalar residual = 0.f;
@@ -60,32 +59,32 @@ btScalar btMultiBodyConstraintSolver::solveSingleIteration(int iteration, btColl
 			residual = resolveSingleConstraintRowGeneric(constraint);
 		}
 
-		leastSquaredResidual = btMax(leastSquaredResidual,residual*residual);
- 
-		if(constraint.m_multiBodyA) 
+		leastSquaredResidual = btMax(leastSquaredResidual, residual * residual);
+
+		if (constraint.m_multiBodyA)
 			constraint.m_multiBodyA->setPosUpdated(false);
-		if(constraint.m_multiBodyB) 
+		if (constraint.m_multiBodyB)
 			constraint.m_multiBodyB->setPosUpdated(false);
 	}
 
 	//solve featherstone frictional contact
-	if (infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS && ((infoGlobal.m_solverMode&SOLVER_DISABLE_IMPLICIT_CONE_FRICTION) == 0))
+	if (infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS && ((infoGlobal.m_solverMode & SOLVER_DISABLE_IMPLICIT_CONE_FRICTION) == 0))
 	{
-		for (int j1 = 0; j1<this->m_multiBodyTorsionalFrictionContactConstraints.size(); j1++)
+		for (int j1 = 0; j1 < this->m_multiBodyTorsionalFrictionContactConstraints.size(); j1++)
 		{
 			if (iteration < infoGlobal.m_numIterations)
 			{
-				int index = j1;//iteration&1? j1 : m_multiBodyTorsionalFrictionContactConstraints.size()-1-j1;
+				int index = j1;  //iteration&1? j1 : m_multiBodyTorsionalFrictionContactConstraints.size()-1-j1;
 
 				btMultiBodySolverConstraint& frictionConstraint = m_multiBodyTorsionalFrictionContactConstraints[index];
 				btScalar totalImpulse = m_multiBodyNormalContactConstraints[frictionConstraint.m_frictionIndex].m_appliedImpulse;
 				//adjust friction limits here
-				if (totalImpulse>btScalar(0))
+				if (totalImpulse > btScalar(0))
 				{
-					frictionConstraint.m_lowerLimit = -(frictionConstraint.m_friction*totalImpulse);
-					frictionConstraint.m_upperLimit = frictionConstraint.m_friction*totalImpulse;
+					frictionConstraint.m_lowerLimit = -(frictionConstraint.m_friction * totalImpulse);
+					frictionConstraint.m_upperLimit = frictionConstraint.m_friction * totalImpulse;
 					btScalar residual = resolveSingleConstraintRowGeneric(frictionConstraint);
-					leastSquaredResidual = btMax(leastSquaredResidual , residual*residual);
+					leastSquaredResidual = btMax(leastSquaredResidual, residual * residual);
 
 					if (frictionConstraint.m_multiBodyA)
 						frictionConstraint.m_multiBodyA->setPosUpdated(false);
@@ -99,29 +98,29 @@ btScalar btMultiBodyConstraintSolver::solveSingleIteration(int iteration, btColl
 		{
 			if (iteration < infoGlobal.m_numIterations)
 			{
-				int index = j1;//iteration&1? j1 : m_multiBodyFrictionContactConstraints.size()-1-j1;
+				int index = j1;  //iteration&1? j1 : m_multiBodyFrictionContactConstraints.size()-1-j1;
 				btMultiBodySolverConstraint& frictionConstraint = m_multiBodyFrictionContactConstraints[index];
 
 				btScalar totalImpulse = m_multiBodyNormalContactConstraints[frictionConstraint.m_frictionIndex].m_appliedImpulse;
 				j1++;
-				int index2 = j1;//iteration&1? j1 : m_multiBodyFrictionContactConstraints.size()-1-j1;
+				int index2 = j1;  //iteration&1? j1 : m_multiBodyFrictionContactConstraints.size()-1-j1;
 				btMultiBodySolverConstraint& frictionConstraintB = m_multiBodyFrictionContactConstraints[index2];
 				btAssert(frictionConstraint.m_frictionIndex == frictionConstraintB.m_frictionIndex);
 
 				if (frictionConstraint.m_frictionIndex == frictionConstraintB.m_frictionIndex)
 				{
-					frictionConstraint.m_lowerLimit = -(frictionConstraint.m_friction*totalImpulse);
-					frictionConstraint.m_upperLimit = frictionConstraint.m_friction*totalImpulse;
-					frictionConstraintB.m_lowerLimit = -(frictionConstraintB.m_friction*totalImpulse);
-					frictionConstraintB.m_upperLimit = frictionConstraintB.m_friction*totalImpulse;
+					frictionConstraint.m_lowerLimit = -(frictionConstraint.m_friction * totalImpulse);
+					frictionConstraint.m_upperLimit = frictionConstraint.m_friction * totalImpulse;
+					frictionConstraintB.m_lowerLimit = -(frictionConstraintB.m_friction * totalImpulse);
+					frictionConstraintB.m_upperLimit = frictionConstraintB.m_friction * totalImpulse;
 					btScalar residual = resolveConeFrictionConstraintRows(frictionConstraint, frictionConstraintB);
-					leastSquaredResidual = btMax(leastSquaredResidual, residual*residual);
-					
+					leastSquaredResidual = btMax(leastSquaredResidual, residual * residual);
+
 					if (frictionConstraintB.m_multiBodyA)
 						frictionConstraintB.m_multiBodyA->setPosUpdated(false);
 					if (frictionConstraintB.m_multiBodyB)
 						frictionConstraintB.m_multiBodyB->setPosUpdated(false);
-				
+
 					if (frictionConstraint.m_multiBodyA)
 						frictionConstraint.m_multiBodyA->setPosUpdated(false);
 					if (frictionConstraint.m_multiBodyB)
@@ -129,26 +128,24 @@ btScalar btMultiBodyConstraintSolver::solveSingleIteration(int iteration, btColl
 				}
 			}
 		}
-
-		
 	}
 	else
 	{
-		for (int j1 = 0; j1<this->m_multiBodyFrictionContactConstraints.size(); j1++)
+		for (int j1 = 0; j1 < this->m_multiBodyFrictionContactConstraints.size(); j1++)
 		{
 			if (iteration < infoGlobal.m_numIterations)
 			{
-				int index = j1;//iteration&1? j1 : m_multiBodyFrictionContactConstraints.size()-1-j1;
+				int index = j1;  //iteration&1? j1 : m_multiBodyFrictionContactConstraints.size()-1-j1;
 
 				btMultiBodySolverConstraint& frictionConstraint = m_multiBodyFrictionContactConstraints[index];
 				btScalar totalImpulse = m_multiBodyNormalContactConstraints[frictionConstraint.m_frictionIndex].m_appliedImpulse;
 				//adjust friction limits here
-				if (totalImpulse>btScalar(0))
+				if (totalImpulse > btScalar(0))
 				{
-					frictionConstraint.m_lowerLimit = -(frictionConstraint.m_friction*totalImpulse);
-					frictionConstraint.m_upperLimit = frictionConstraint.m_friction*totalImpulse;
+					frictionConstraint.m_lowerLimit = -(frictionConstraint.m_friction * totalImpulse);
+					frictionConstraint.m_upperLimit = frictionConstraint.m_friction * totalImpulse;
 					btScalar residual = resolveSingleConstraintRowGeneric(frictionConstraint);
-					leastSquaredResidual = btMax(leastSquaredResidual, residual*residual);
+					leastSquaredResidual = btMax(leastSquaredResidual, residual * residual);
 
 					if (frictionConstraint.m_multiBodyA)
 						frictionConstraint.m_multiBodyA->setPosUpdated(false);
@@ -161,18 +158,18 @@ btScalar btMultiBodyConstraintSolver::solveSingleIteration(int iteration, btColl
 	return leastSquaredResidual;
 }
 
-btScalar btMultiBodyConstraintSolver::solveGroupCacheFriendlySetup(btCollisionObject** bodies,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer)
+btScalar btMultiBodyConstraintSolver::solveGroupCacheFriendlySetup(btCollisionObject** bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer)
 {
 	m_multiBodyNonContactConstraints.resize(0);
 	m_multiBodyNormalContactConstraints.resize(0);
 	m_multiBodyFrictionContactConstraints.resize(0);
 	m_multiBodyTorsionalFrictionContactConstraints.resize(0);
-	
+
 	m_data.m_jacobians.resize(0);
 	m_data.m_deltaVelocitiesUnitImpulse.resize(0);
 	m_data.m_deltaVelocities.resize(0);
 
-	for (int i=0;i<numBodies;i++)
+	for (int i = 0; i < numBodies; i++)
 	{
 		const btMultiBodyLinkCollider* fcA = btMultiBodyLinkCollider::upcast(bodies[i]);
 		if (fcA)
@@ -181,21 +178,20 @@ btScalar btMultiBodyConstraintSolver::solveGroupCacheFriendlySetup(btCollisionOb
 		}
 	}
 
-	btScalar val = btSequentialImpulseConstraintSolver::solveGroupCacheFriendlySetup( bodies,numBodies,manifoldPtr, numManifolds, constraints,numConstraints,infoGlobal,debugDrawer);
+	btScalar val = btSequentialImpulseConstraintSolver::solveGroupCacheFriendlySetup(bodies, numBodies, manifoldPtr, numManifolds, constraints, numConstraints, infoGlobal, debugDrawer);
 
 	return val;
 }
 
-void	btMultiBodyConstraintSolver::applyDeltaVee(btScalar* delta_vee, btScalar impulse, int velocityIndex, int ndof)
+void btMultiBodyConstraintSolver::applyDeltaVee(btScalar* delta_vee, btScalar impulse, int velocityIndex, int ndof)
 {
-    for (int i = 0; i < ndof; ++i) 
-		m_data.m_deltaVelocities[velocityIndex+i] += delta_vee[i] * impulse;
+	for (int i = 0; i < ndof; ++i)
+		m_data.m_deltaVelocities[velocityIndex + i] += delta_vee[i] * impulse;
 }
 
 btScalar btMultiBodyConstraintSolver::resolveSingleConstraintRowGeneric(const btMultiBodySolverConstraint& c)
 {
-
-	btScalar deltaImpulse = c.m_rhs - btScalar(c.m_appliedImpulse)*c.m_cfm;
+	btScalar deltaImpulse = c.m_rhs - btScalar(c.m_appliedImpulse) * c.m_cfm;
 	btScalar deltaVelADotn = 0;
 	btScalar deltaVelBDotn = 0;
 	btSolverBody* bodyA = 0;
@@ -227,9 +223,8 @@ btScalar btMultiBodyConstraintSolver::resolveSingleConstraintRowGeneric(const bt
 		deltaVelBDotn += c.m_contactNormal2.dot(bodyB->internalGetDeltaLinearVelocity()) + c.m_relpos2CrossNormal.dot(bodyB->internalGetDeltaAngularVelocity());
 	}
 
-
-	deltaImpulse -= deltaVelADotn*c.m_jacDiagABInv;//m_jacDiagABInv = 1./denom
-	deltaImpulse -= deltaVelBDotn*c.m_jacDiagABInv;
+	deltaImpulse -= deltaVelADotn * c.m_jacDiagABInv;  //m_jacDiagABInv = 1./denom
+	deltaImpulse -= deltaVelBDotn * c.m_jacDiagABInv;
 	const btScalar sum = btScalar(c.m_appliedImpulse) + deltaImpulse;
 
 	if (sum < c.m_lowerLimit)
@@ -246,7 +241,7 @@ btScalar btMultiBodyConstraintSolver::resolveSingleConstraintRowGeneric(const bt
 	{
 		c.m_appliedImpulse = sum;
 	}
-	
+
 	if (c.m_multiBodyA)
 	{
 		applyDeltaVee(&m_data.m_deltaVelocitiesUnitImpulse[c.m_jacAindex], deltaImpulse, c.m_deltaVelAindex, ndofA);
@@ -254,12 +249,11 @@ btScalar btMultiBodyConstraintSolver::resolveSingleConstraintRowGeneric(const bt
 		//note: update of the actual velocities (below) in the multibody does not have to happen now since m_deltaVelocities can be applied after all iterations
 		//it would make the multibody solver more like the regular one with m_deltaVelocities being equivalent to btSolverBody::m_deltaLinearVelocity/m_deltaAngularVelocity
 		c.m_multiBodyA->applyDeltaVeeMultiDof2(&m_data.m_deltaVelocitiesUnitImpulse[c.m_jacAindex], deltaImpulse);
-#endif //DIRECTLY_UPDATE_VELOCITY_DURING_SOLVER_ITERATIONS
+#endif  //DIRECTLY_UPDATE_VELOCITY_DURING_SOLVER_ITERATIONS
 	}
 	else if (c.m_solverBodyIdA >= 0)
 	{
-		bodyA->internalApplyImpulse(c.m_contactNormal1*bodyA->internalGetInvMass(), c.m_angularComponentA, deltaImpulse);
-
+		bodyA->internalApplyImpulse(c.m_contactNormal1 * bodyA->internalGetInvMass(), c.m_angularComponentA, deltaImpulse);
 	}
 	if (c.m_multiBodyB)
 	{
@@ -268,54 +262,54 @@ btScalar btMultiBodyConstraintSolver::resolveSingleConstraintRowGeneric(const bt
 		//note: update of the actual velocities (below) in the multibody does not have to happen now since m_deltaVelocities can be applied after all iterations
 		//it would make the multibody solver more like the regular one with m_deltaVelocities being equivalent to btSolverBody::m_deltaLinearVelocity/m_deltaAngularVelocity
 		c.m_multiBodyB->applyDeltaVeeMultiDof2(&m_data.m_deltaVelocitiesUnitImpulse[c.m_jacBindex], deltaImpulse);
-#endif //DIRECTLY_UPDATE_VELOCITY_DURING_SOLVER_ITERATIONS
+#endif  //DIRECTLY_UPDATE_VELOCITY_DURING_SOLVER_ITERATIONS
 	}
 	else if (c.m_solverBodyIdB >= 0)
 	{
-		bodyB->internalApplyImpulse(c.m_contactNormal2*bodyB->internalGetInvMass(), c.m_angularComponentB, deltaImpulse);
+		bodyB->internalApplyImpulse(c.m_contactNormal2 * bodyB->internalGetInvMass(), c.m_angularComponentB, deltaImpulse);
 	}
-    btScalar deltaVel =deltaImpulse/c.m_jacDiagABInv;
+	btScalar deltaVel = deltaImpulse / c.m_jacDiagABInv;
 	return deltaVel;
 }
 
-
-btScalar btMultiBodyConstraintSolver::resolveConeFrictionConstraintRows(const btMultiBodySolverConstraint& cA1,const btMultiBodySolverConstraint& cB)
+btScalar btMultiBodyConstraintSolver::resolveConeFrictionConstraintRows(const btMultiBodySolverConstraint& cA1, const btMultiBodySolverConstraint& cB)
 {
-	int ndofA=0;
-	int ndofB=0;
+	int ndofA = 0;
+	int ndofB = 0;
 	btSolverBody* bodyA = 0;
 	btSolverBody* bodyB = 0;
 	btScalar deltaImpulseB = 0.f;
 	btScalar sumB = 0.f;
 	{
-		deltaImpulseB = cB.m_rhs-btScalar(cB.m_appliedImpulse)*cB.m_cfm;
-		btScalar deltaVelADotn=0;
-		btScalar deltaVelBDotn=0;
+		deltaImpulseB = cB.m_rhs - btScalar(cB.m_appliedImpulse) * cB.m_cfm;
+		btScalar deltaVelADotn = 0;
+		btScalar deltaVelBDotn = 0;
 		if (cB.m_multiBodyA)
 		{
-			ndofA  = cB.m_multiBodyA->getNumDofs() + 6;
-			for (int i = 0; i < ndofA; ++i) 
-				deltaVelADotn += m_data.m_jacobians[cB.m_jacAindex+i] * m_data.m_deltaVelocities[cB.m_deltaVelAindex+i];
-		} else if(cB.m_solverBodyIdA >= 0)
+			ndofA = cB.m_multiBodyA->getNumDofs() + 6;
+			for (int i = 0; i < ndofA; ++i)
+				deltaVelADotn += m_data.m_jacobians[cB.m_jacAindex + i] * m_data.m_deltaVelocities[cB.m_deltaVelAindex + i];
+		}
+		else if (cB.m_solverBodyIdA >= 0)
 		{
 			bodyA = &m_tmpSolverBodyPool[cB.m_solverBodyIdA];
-			deltaVelADotn += cB.m_contactNormal1.dot(bodyA->internalGetDeltaLinearVelocity()) 	+ cB.m_relpos1CrossNormal.dot(bodyA->internalGetDeltaAngularVelocity());
+			deltaVelADotn += cB.m_contactNormal1.dot(bodyA->internalGetDeltaLinearVelocity()) + cB.m_relpos1CrossNormal.dot(bodyA->internalGetDeltaAngularVelocity());
 		}
 
 		if (cB.m_multiBodyB)
 		{
-			ndofB  = cB.m_multiBodyB->getNumDofs() + 6;
-			for (int i = 0; i < ndofB; ++i) 
-				deltaVelBDotn += m_data.m_jacobians[cB.m_jacBindex+i] * m_data.m_deltaVelocities[cB.m_deltaVelBindex+i];
-		} else if(cB.m_solverBodyIdB >= 0)
+			ndofB = cB.m_multiBodyB->getNumDofs() + 6;
+			for (int i = 0; i < ndofB; ++i)
+				deltaVelBDotn += m_data.m_jacobians[cB.m_jacBindex + i] * m_data.m_deltaVelocities[cB.m_deltaVelBindex + i];
+		}
+		else if (cB.m_solverBodyIdB >= 0)
 		{
 			bodyB = &m_tmpSolverBodyPool[cB.m_solverBodyIdB];
-			deltaVelBDotn += cB.m_contactNormal2.dot(bodyB->internalGetDeltaLinearVelocity())  + cB.m_relpos2CrossNormal.dot(bodyB->internalGetDeltaAngularVelocity());
+			deltaVelBDotn += cB.m_contactNormal2.dot(bodyB->internalGetDeltaLinearVelocity()) + cB.m_relpos2CrossNormal.dot(bodyB->internalGetDeltaAngularVelocity());
 		}
 
-	
-		deltaImpulseB	-=	deltaVelADotn*cB.m_jacDiagABInv;//m_jacDiagABInv = 1./denom
-		deltaImpulseB	-=	deltaVelBDotn*cB.m_jacDiagABInv;
+		deltaImpulseB -= deltaVelADotn * cB.m_jacDiagABInv;  //m_jacDiagABInv = 1./denom
+		deltaImpulseB -= deltaVelBDotn * cB.m_jacDiagABInv;
 		sumB = btScalar(cB.m_appliedImpulse) + deltaImpulseB;
 	}
 
@@ -324,45 +318,45 @@ btScalar btMultiBodyConstraintSolver::resolveConeFrictionConstraintRows(const bt
 	const btMultiBodySolverConstraint& cA = cA1;
 	{
 		{
-			deltaImpulseA = cA.m_rhs-btScalar(cA.m_appliedImpulse)*cA.m_cfm;
-			btScalar deltaVelADotn=0;
-			btScalar deltaVelBDotn=0;
+			deltaImpulseA = cA.m_rhs - btScalar(cA.m_appliedImpulse) * cA.m_cfm;
+			btScalar deltaVelADotn = 0;
+			btScalar deltaVelBDotn = 0;
 			if (cA.m_multiBodyA)
 			{
-				ndofA  = cA.m_multiBodyA->getNumDofs() + 6;
-				for (int i = 0; i < ndofA; ++i) 
-					deltaVelADotn += m_data.m_jacobians[cA.m_jacAindex+i] * m_data.m_deltaVelocities[cA.m_deltaVelAindex+i];
-			} else if(cA.m_solverBodyIdA >= 0)
+				ndofA = cA.m_multiBodyA->getNumDofs() + 6;
+				for (int i = 0; i < ndofA; ++i)
+					deltaVelADotn += m_data.m_jacobians[cA.m_jacAindex + i] * m_data.m_deltaVelocities[cA.m_deltaVelAindex + i];
+			}
+			else if (cA.m_solverBodyIdA >= 0)
 			{
 				bodyA = &m_tmpSolverBodyPool[cA.m_solverBodyIdA];
-				deltaVelADotn += cA.m_contactNormal1.dot(bodyA->internalGetDeltaLinearVelocity()) 	+ cA.m_relpos1CrossNormal.dot(bodyA->internalGetDeltaAngularVelocity());
+				deltaVelADotn += cA.m_contactNormal1.dot(bodyA->internalGetDeltaLinearVelocity()) + cA.m_relpos1CrossNormal.dot(bodyA->internalGetDeltaAngularVelocity());
 			}
 
 			if (cA.m_multiBodyB)
 			{
-				ndofB  = cA.m_multiBodyB->getNumDofs() + 6;
-				for (int i = 0; i < ndofB; ++i) 
-					deltaVelBDotn += m_data.m_jacobians[cA.m_jacBindex+i] * m_data.m_deltaVelocities[cA.m_deltaVelBindex+i];
-			} else if(cA.m_solverBodyIdB >= 0)
+				ndofB = cA.m_multiBodyB->getNumDofs() + 6;
+				for (int i = 0; i < ndofB; ++i)
+					deltaVelBDotn += m_data.m_jacobians[cA.m_jacBindex + i] * m_data.m_deltaVelocities[cA.m_deltaVelBindex + i];
+			}
+			else if (cA.m_solverBodyIdB >= 0)
 			{
 				bodyB = &m_tmpSolverBodyPool[cA.m_solverBodyIdB];
-				deltaVelBDotn += cA.m_contactNormal2.dot(bodyB->internalGetDeltaLinearVelocity())  + cA.m_relpos2CrossNormal.dot(bodyB->internalGetDeltaAngularVelocity());
+				deltaVelBDotn += cA.m_contactNormal2.dot(bodyB->internalGetDeltaLinearVelocity()) + cA.m_relpos2CrossNormal.dot(bodyB->internalGetDeltaAngularVelocity());
 			}
 
-	
-			deltaImpulseA	-=	deltaVelADotn*cA.m_jacDiagABInv;//m_jacDiagABInv = 1./denom
-			deltaImpulseA	-=	deltaVelBDotn*cA.m_jacDiagABInv;
+			deltaImpulseA -= deltaVelADotn * cA.m_jacDiagABInv;  //m_jacDiagABInv = 1./denom
+			deltaImpulseA -= deltaVelBDotn * cA.m_jacDiagABInv;
 			sumA = btScalar(cA.m_appliedImpulse) + deltaImpulseA;
 		}
 	}
 
-	if (sumA*sumA+sumB*sumB>=cA.m_lowerLimit*cB.m_lowerLimit)
+	if (sumA * sumA + sumB * sumB >= cA.m_lowerLimit * cB.m_lowerLimit)
 	{
-		btScalar angle = btAtan2(sumA,sumB);
-		btScalar sumAclipped = btFabs(cA.m_lowerLimit*btSin(angle));
-		btScalar sumBclipped = btFabs(cB.m_lowerLimit*btCos(angle));
+		btScalar angle = btAtan2(sumA, sumB);
+		btScalar sumAclipped = btFabs(cA.m_lowerLimit * btSin(angle));
+		btScalar sumBclipped = btFabs(cB.m_lowerLimit * btCos(angle));
 
-		
 		if (sumA < -sumAclipped)
 		{
 			deltaImpulseA = -sumAclipped - cA.m_appliedImpulse;
@@ -396,78 +390,77 @@ btScalar btMultiBodyConstraintSolver::resolveConeFrictionConstraintRows(const bt
 		//cA.m_appliedImpulse = sumAclipped;
 		//deltaImpulseB = sumBclipped-cB.m_appliedImpulse;
 		//cB.m_appliedImpulse = sumBclipped;
-	} 
+	}
 	else
 	{
 		cA.m_appliedImpulse = sumA;
 		cB.m_appliedImpulse = sumB;
 	}
-	
+
 	if (cA.m_multiBodyA)
 	{
-		applyDeltaVee(&m_data.m_deltaVelocitiesUnitImpulse[cA.m_jacAindex],deltaImpulseA,cA.m_deltaVelAindex,ndofA);
+		applyDeltaVee(&m_data.m_deltaVelocitiesUnitImpulse[cA.m_jacAindex], deltaImpulseA, cA.m_deltaVelAindex, ndofA);
 #ifdef DIRECTLY_UPDATE_VELOCITY_DURING_SOLVER_ITERATIONS
 		//note: update of the actual velocities (below) in the multibody does not have to happen now since m_deltaVelocities can be applied after all iterations
 		//it would make the multibody solver more like the regular one with m_deltaVelocities being equivalent to btSolverBody::m_deltaLinearVelocity/m_deltaAngularVelocity
-		cA.m_multiBodyA->applyDeltaVeeMultiDof2(&m_data.m_deltaVelocitiesUnitImpulse[cA.m_jacAindex],deltaImpulseA);
-#endif //DIRECTLY_UPDATE_VELOCITY_DURING_SOLVER_ITERATIONS
-	} else if(cA.m_solverBodyIdA >= 0)
+		cA.m_multiBodyA->applyDeltaVeeMultiDof2(&m_data.m_deltaVelocitiesUnitImpulse[cA.m_jacAindex], deltaImpulseA);
+#endif  //DIRECTLY_UPDATE_VELOCITY_DURING_SOLVER_ITERATIONS
+	}
+	else if (cA.m_solverBodyIdA >= 0)
 	{
-		bodyA->internalApplyImpulse(cA.m_contactNormal1*bodyA->internalGetInvMass(),cA.m_angularComponentA,deltaImpulseA);
-
+		bodyA->internalApplyImpulse(cA.m_contactNormal1 * bodyA->internalGetInvMass(), cA.m_angularComponentA, deltaImpulseA);
 	}
 	if (cA.m_multiBodyB)
 	{
-		applyDeltaVee(&m_data.m_deltaVelocitiesUnitImpulse[cA.m_jacBindex],deltaImpulseA,cA.m_deltaVelBindex,ndofB);
+		applyDeltaVee(&m_data.m_deltaVelocitiesUnitImpulse[cA.m_jacBindex], deltaImpulseA, cA.m_deltaVelBindex, ndofB);
 #ifdef DIRECTLY_UPDATE_VELOCITY_DURING_SOLVER_ITERATIONS
 		//note: update of the actual velocities (below) in the multibody does not have to happen now since m_deltaVelocities can be applied after all iterations
 		//it would make the multibody solver more like the regular one with m_deltaVelocities being equivalent to btSolverBody::m_deltaLinearVelocity/m_deltaAngularVelocity
-		cA.m_multiBodyB->applyDeltaVeeMultiDof2(&m_data.m_deltaVelocitiesUnitImpulse[cA.m_jacBindex],deltaImpulseA);
-#endif //DIRECTLY_UPDATE_VELOCITY_DURING_SOLVER_ITERATIONS
-	} else if(cA.m_solverBodyIdB >= 0)
+		cA.m_multiBodyB->applyDeltaVeeMultiDof2(&m_data.m_deltaVelocitiesUnitImpulse[cA.m_jacBindex], deltaImpulseA);
+#endif  //DIRECTLY_UPDATE_VELOCITY_DURING_SOLVER_ITERATIONS
+	}
+	else if (cA.m_solverBodyIdB >= 0)
 	{
-		bodyB->internalApplyImpulse(cA.m_contactNormal2*bodyB->internalGetInvMass(),cA.m_angularComponentB,deltaImpulseA);
+		bodyB->internalApplyImpulse(cA.m_contactNormal2 * bodyB->internalGetInvMass(), cA.m_angularComponentB, deltaImpulseA);
 	}
 
 	if (cB.m_multiBodyA)
 	{
-		applyDeltaVee(&m_data.m_deltaVelocitiesUnitImpulse[cB.m_jacAindex],deltaImpulseB,cB.m_deltaVelAindex,ndofA);
+		applyDeltaVee(&m_data.m_deltaVelocitiesUnitImpulse[cB.m_jacAindex], deltaImpulseB, cB.m_deltaVelAindex, ndofA);
 #ifdef DIRECTLY_UPDATE_VELOCITY_DURING_SOLVER_ITERATIONS
 		//note: update of the actual velocities (below) in the multibody does not have to happen now since m_deltaVelocities can be applied after all iterations
 		//it would make the multibody solver more like the regular one with m_deltaVelocities being equivalent to btSolverBody::m_deltaLinearVelocity/m_deltaAngularVelocity
-		cB.m_multiBodyA->applyDeltaVeeMultiDof2(&m_data.m_deltaVelocitiesUnitImpulse[cB.m_jacAindex],deltaImpulseB);
-#endif //DIRECTLY_UPDATE_VELOCITY_DURING_SOLVER_ITERATIONS
-	} else if(cB.m_solverBodyIdA >= 0)
+		cB.m_multiBodyA->applyDeltaVeeMultiDof2(&m_data.m_deltaVelocitiesUnitImpulse[cB.m_jacAindex], deltaImpulseB);
+#endif  //DIRECTLY_UPDATE_VELOCITY_DURING_SOLVER_ITERATIONS
+	}
+	else if (cB.m_solverBodyIdA >= 0)
 	{
-		bodyA->internalApplyImpulse(cB.m_contactNormal1*bodyA->internalGetInvMass(),cB.m_angularComponentA,deltaImpulseB);
+		bodyA->internalApplyImpulse(cB.m_contactNormal1 * bodyA->internalGetInvMass(), cB.m_angularComponentA, deltaImpulseB);
 	}
 	if (cB.m_multiBodyB)
 	{
-		applyDeltaVee(&m_data.m_deltaVelocitiesUnitImpulse[cB.m_jacBindex],deltaImpulseB,cB.m_deltaVelBindex,ndofB);
+		applyDeltaVee(&m_data.m_deltaVelocitiesUnitImpulse[cB.m_jacBindex], deltaImpulseB, cB.m_deltaVelBindex, ndofB);
 #ifdef DIRECTLY_UPDATE_VELOCITY_DURING_SOLVER_ITERATIONS
 		//note: update of the actual velocities (below) in the multibody does not have to happen now since m_deltaVelocities can be applied after all iterations
 		//it would make the multibody solver more like the regular one with m_deltaVelocities being equivalent to btSolverBody::m_deltaLinearVelocity/m_deltaAngularVelocity
-		cB.m_multiBodyB->applyDeltaVeeMultiDof2(&m_data.m_deltaVelocitiesUnitImpulse[cB.m_jacBindex],deltaImpulseB);
-#endif //DIRECTLY_UPDATE_VELOCITY_DURING_SOLVER_ITERATIONS
-	} else if(cB.m_solverBodyIdB >= 0)
+		cB.m_multiBodyB->applyDeltaVeeMultiDof2(&m_data.m_deltaVelocitiesUnitImpulse[cB.m_jacBindex], deltaImpulseB);
+#endif  //DIRECTLY_UPDATE_VELOCITY_DURING_SOLVER_ITERATIONS
+	}
+	else if (cB.m_solverBodyIdB >= 0)
 	{
-		bodyB->internalApplyImpulse(cB.m_contactNormal2*bodyB->internalGetInvMass(),cB.m_angularComponentB,deltaImpulseB);
+		bodyB->internalApplyImpulse(cB.m_contactNormal2 * bodyB->internalGetInvMass(), cB.m_angularComponentB, deltaImpulseB);
 	}
 
-    btScalar deltaVel =deltaImpulseA/cA.m_jacDiagABInv+deltaImpulseB/cB.m_jacDiagABInv;
-    return deltaVel;
+	btScalar deltaVel = deltaImpulseA / cA.m_jacDiagABInv + deltaImpulseB / cB.m_jacDiagABInv;
+	return deltaVel;
 }
 
-
-
-
-void btMultiBodyConstraintSolver::setupMultiBodyContactConstraint(btMultiBodySolverConstraint& solverConstraint, 
-																 const btVector3& contactNormal,
-																 btManifoldPoint& cp, const btContactSolverInfo& infoGlobal,
-																 btScalar& relaxation,
-																 bool isFriction, btScalar desiredVelocity, btScalar cfmSlip)
+void btMultiBodyConstraintSolver::setupMultiBodyContactConstraint(btMultiBodySolverConstraint& solverConstraint,
+																  const btVector3& contactNormal,
+																  btManifoldPoint& cp, const btContactSolverInfo& infoGlobal,
+																  btScalar& relaxation,
+																  bool isFriction, btScalar desiredVelocity, btScalar cfmSlip)
 {
-			
 	BT_PROFILE("setupMultiBodyContactConstraint");
 	btVector3 rel_pos1;
 	btVector3 rel_pos2;
@@ -485,44 +478,46 @@ void btMultiBodyConstraintSolver::setupMultiBodyContactConstraint(btMultiBodySol
 	btRigidBody* rb1 = multiBodyB ? 0 : bodyB->m_originalBody;
 
 	if (bodyA)
-		rel_pos1 = pos1 - bodyA->getWorldTransform().getOrigin(); 
+		rel_pos1 = pos1 - bodyA->getWorldTransform().getOrigin();
 	if (bodyB)
 		rel_pos2 = pos2 - bodyB->getWorldTransform().getOrigin();
 
 	relaxation = infoGlobal.m_sor;
-	
-	btScalar invTimeStep = btScalar(1)/infoGlobal.m_timeStep;
-	
-	 //cfm = 1 /       ( dt * kp + kd )
-    //erp = dt * kp / ( dt * kp + kd )
-    
-    btScalar cfm;
+
+	btScalar invTimeStep = btScalar(1) / infoGlobal.m_timeStep;
+
+	//cfm = 1 /       ( dt * kp + kd )
+	//erp = dt * kp / ( dt * kp + kd )
+
+	btScalar cfm;
 	btScalar erp;
 	if (isFriction)
 	{
 		cfm = infoGlobal.m_frictionCFM;
 		erp = infoGlobal.m_frictionERP;
-	} else
+	}
+	else
 	{
 		cfm = infoGlobal.m_globalCfm;
 		erp = infoGlobal.m_erp2;
 
-		if ((cp.m_contactPointFlags&BT_CONTACT_FLAG_HAS_CONTACT_CFM) || (cp.m_contactPointFlags&BT_CONTACT_FLAG_HAS_CONTACT_ERP))
+		if ((cp.m_contactPointFlags & BT_CONTACT_FLAG_HAS_CONTACT_CFM) || (cp.m_contactPointFlags & BT_CONTACT_FLAG_HAS_CONTACT_ERP))
 		{
-			if (cp.m_contactPointFlags&BT_CONTACT_FLAG_HAS_CONTACT_CFM)
-				cfm  = cp.m_contactCFM;
-			if (cp.m_contactPointFlags&BT_CONTACT_FLAG_HAS_CONTACT_ERP)
-				erp = cp.m_contactERP;                
-		} else
+			if (cp.m_contactPointFlags & BT_CONTACT_FLAG_HAS_CONTACT_CFM)
+				cfm = cp.m_contactCFM;
+			if (cp.m_contactPointFlags & BT_CONTACT_FLAG_HAS_CONTACT_ERP)
+				erp = cp.m_contactERP;
+		}
+		else
 		{
 			if (cp.m_contactPointFlags & BT_CONTACT_FLAG_CONTACT_STIFFNESS_DAMPING)
 			{
-				btScalar denom = ( infoGlobal.m_timeStep * cp.m_combinedContactStiffness1 + cp.m_combinedContactDamping1 );
+				btScalar denom = (infoGlobal.m_timeStep * cp.m_combinedContactStiffness1 + cp.m_combinedContactDamping1);
 				if (denom < SIMD_EPSILON)
 				{
 					denom = SIMD_EPSILON;
 				}
-				cfm = btScalar(1) / denom; 
+				cfm = btScalar(1) / denom;
 				erp = (infoGlobal.m_timeStep * cp.m_combinedContactStiffness1) / denom;
 			}
 		}
@@ -532,218 +527,217 @@ void btMultiBodyConstraintSolver::setupMultiBodyContactConstraint(btMultiBodySol
 
 	if (multiBodyA)
 	{
-		if (solverConstraint.m_linkA<0)
+		if (solverConstraint.m_linkA < 0)
 		{
 			rel_pos1 = pos1 - multiBodyA->getBasePos();
-		} else
+		}
+		else
 		{
 			rel_pos1 = pos1 - multiBodyA->getLink(solverConstraint.m_linkA).m_cachedWorldTransform.getOrigin();
 		}
-		const int ndofA  = multiBodyA->getNumDofs() + 6;
+		const int ndofA = multiBodyA->getNumDofs() + 6;
 
 		solverConstraint.m_deltaVelAindex = multiBodyA->getCompanionId();
 
-		if (solverConstraint.m_deltaVelAindex <0)
+		if (solverConstraint.m_deltaVelAindex < 0)
 		{
 			solverConstraint.m_deltaVelAindex = m_data.m_deltaVelocities.size();
 			multiBodyA->setCompanionId(solverConstraint.m_deltaVelAindex);
-			m_data.m_deltaVelocities.resize(m_data.m_deltaVelocities.size()+ndofA);
-		} else
+			m_data.m_deltaVelocities.resize(m_data.m_deltaVelocities.size() + ndofA);
+		}
+		else
 		{
-			btAssert(m_data.m_deltaVelocities.size() >= solverConstraint.m_deltaVelAindex+ndofA);
+			btAssert(m_data.m_deltaVelocities.size() >= solverConstraint.m_deltaVelAindex + ndofA);
 		}
 
 		solverConstraint.m_jacAindex = m_data.m_jacobians.size();
-		m_data.m_jacobians.resize(m_data.m_jacobians.size()+ndofA);
-		m_data.m_deltaVelocitiesUnitImpulse.resize(m_data.m_deltaVelocitiesUnitImpulse.size()+ndofA);
+		m_data.m_jacobians.resize(m_data.m_jacobians.size() + ndofA);
+		m_data.m_deltaVelocitiesUnitImpulse.resize(m_data.m_deltaVelocitiesUnitImpulse.size() + ndofA);
 		btAssert(m_data.m_jacobians.size() == m_data.m_deltaVelocitiesUnitImpulse.size());
 
-		btScalar* jac1=&m_data.m_jacobians[solverConstraint.m_jacAindex];
+		btScalar* jac1 = &m_data.m_jacobians[solverConstraint.m_jacAindex];
 		multiBodyA->fillContactJacobianMultiDof(solverConstraint.m_linkA, cp.getPositionWorldOnA(), contactNormal, jac1, m_data.scratch_r, m_data.scratch_v, m_data.scratch_m);
 		btScalar* delta = &m_data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacAindex];
-		multiBodyA->calcAccelerationDeltasMultiDof(&m_data.m_jacobians[solverConstraint.m_jacAindex],delta,m_data.scratch_r, m_data.scratch_v);
+		multiBodyA->calcAccelerationDeltasMultiDof(&m_data.m_jacobians[solverConstraint.m_jacAindex], delta, m_data.scratch_r, m_data.scratch_v);
 
 		btVector3 torqueAxis0 = rel_pos1.cross(contactNormal);
 		solverConstraint.m_relpos1CrossNormal = torqueAxis0;
 		solverConstraint.m_contactNormal1 = contactNormal;
-	} else
+	}
+	else
 	{
 		btVector3 torqueAxis0 = rel_pos1.cross(contactNormal);
 		solverConstraint.m_relpos1CrossNormal = torqueAxis0;
 		solverConstraint.m_contactNormal1 = contactNormal;
-		solverConstraint.m_angularComponentA = rb0 ? rb0->getInvInertiaTensorWorld()*torqueAxis0*rb0->getAngularFactor() : btVector3(0,0,0);
+		solverConstraint.m_angularComponentA = rb0 ? rb0->getInvInertiaTensorWorld() * torqueAxis0 * rb0->getAngularFactor() : btVector3(0, 0, 0);
 	}
 
-	
-
 	if (multiBodyB)
 	{
-		if (solverConstraint.m_linkB<0)
+		if (solverConstraint.m_linkB < 0)
 		{
 			rel_pos2 = pos2 - multiBodyB->getBasePos();
-		} else
+		}
+		else
 		{
 			rel_pos2 = pos2 - multiBodyB->getLink(solverConstraint.m_linkB).m_cachedWorldTransform.getOrigin();
 		}
 
-		const int ndofB  = multiBodyB->getNumDofs() + 6;
+		const int ndofB = multiBodyB->getNumDofs() + 6;
 
 		solverConstraint.m_deltaVelBindex = multiBodyB->getCompanionId();
-		if (solverConstraint.m_deltaVelBindex <0)
+		if (solverConstraint.m_deltaVelBindex < 0)
 		{
 			solverConstraint.m_deltaVelBindex = m_data.m_deltaVelocities.size();
 			multiBodyB->setCompanionId(solverConstraint.m_deltaVelBindex);
-			m_data.m_deltaVelocities.resize(m_data.m_deltaVelocities.size()+ndofB);
+			m_data.m_deltaVelocities.resize(m_data.m_deltaVelocities.size() + ndofB);
 		}
 
 		solverConstraint.m_jacBindex = m_data.m_jacobians.size();
 
-		m_data.m_jacobians.resize(m_data.m_jacobians.size()+ndofB);
-		m_data.m_deltaVelocitiesUnitImpulse.resize(m_data.m_deltaVelocitiesUnitImpulse.size()+ndofB);
+		m_data.m_jacobians.resize(m_data.m_jacobians.size() + ndofB);
+		m_data.m_deltaVelocitiesUnitImpulse.resize(m_data.m_deltaVelocitiesUnitImpulse.size() + ndofB);
 		btAssert(m_data.m_jacobians.size() == m_data.m_deltaVelocitiesUnitImpulse.size());
 
 		multiBodyB->fillContactJacobianMultiDof(solverConstraint.m_linkB, cp.getPositionWorldOnB(), -contactNormal, &m_data.m_jacobians[solverConstraint.m_jacBindex], m_data.scratch_r, m_data.scratch_v, m_data.scratch_m);
-		multiBodyB->calcAccelerationDeltasMultiDof(&m_data.m_jacobians[solverConstraint.m_jacBindex],&m_data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacBindex],m_data.scratch_r, m_data.scratch_v);
-		
-		btVector3 torqueAxis1 = rel_pos2.cross(contactNormal);		
+		multiBodyB->calcAccelerationDeltasMultiDof(&m_data.m_jacobians[solverConstraint.m_jacBindex], &m_data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacBindex], m_data.scratch_r, m_data.scratch_v);
+
+		btVector3 torqueAxis1 = rel_pos2.cross(contactNormal);
 		solverConstraint.m_relpos2CrossNormal = -torqueAxis1;
 		solverConstraint.m_contactNormal2 = -contactNormal;
-	
-	} else
+	}
+	else
 	{
-		btVector3 torqueAxis1 = rel_pos2.cross(contactNormal);		
+		btVector3 torqueAxis1 = rel_pos2.cross(contactNormal);
 		solverConstraint.m_relpos2CrossNormal = -torqueAxis1;
 		solverConstraint.m_contactNormal2 = -contactNormal;
-	
-		solverConstraint.m_angularComponentB = rb1 ? rb1->getInvInertiaTensorWorld()*-torqueAxis1*rb1->getAngularFactor() : btVector3(0,0,0);
+
+		solverConstraint.m_angularComponentB = rb1 ? rb1->getInvInertiaTensorWorld() * -torqueAxis1 * rb1->getAngularFactor() : btVector3(0, 0, 0);
 	}
 
 	{
-						
 		btVector3 vec;
 		btScalar denom0 = 0.f;
 		btScalar denom1 = 0.f;
 		btScalar* jacB = 0;
 		btScalar* jacA = 0;
-		btScalar* lambdaA =0;
-		btScalar* lambdaB =0;
-		int ndofA  = 0;
+		btScalar* lambdaA = 0;
+		btScalar* lambdaB = 0;
+		int ndofA = 0;
 		if (multiBodyA)
 		{
-			ndofA  = multiBodyA->getNumDofs() + 6;
+			ndofA = multiBodyA->getNumDofs() + 6;
 			jacA = &m_data.m_jacobians[solverConstraint.m_jacAindex];
 			lambdaA = &m_data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacAindex];
 			for (int i = 0; i < ndofA; ++i)
 			{
-				btScalar j = jacA[i] ;
-				btScalar l =lambdaA[i];
-				denom0 += j*l;
+				btScalar j = jacA[i];
+				btScalar l = lambdaA[i];
+				denom0 += j * l;
 			}
-		} else
+		}
+		else
 		{
 			if (rb0)
 			{
-				vec = ( solverConstraint.m_angularComponentA).cross(rel_pos1);
+				vec = (solverConstraint.m_angularComponentA).cross(rel_pos1);
 				denom0 = rb0->getInvMass() + contactNormal.dot(vec);
 			}
 		}
 		if (multiBodyB)
 		{
-			const int ndofB  = multiBodyB->getNumDofs() + 6;
+			const int ndofB = multiBodyB->getNumDofs() + 6;
 			jacB = &m_data.m_jacobians[solverConstraint.m_jacBindex];
 			lambdaB = &m_data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacBindex];
 			for (int i = 0; i < ndofB; ++i)
 			{
-				btScalar j = jacB[i] ;
-				btScalar l =lambdaB[i];
-				denom1 += j*l;
+				btScalar j = jacB[i];
+				btScalar l = lambdaB[i];
+				denom1 += j * l;
 			}
-
-		} else
+		}
+		else
 		{
 			if (rb1)
 			{
-				vec = ( -solverConstraint.m_angularComponentB).cross(rel_pos2);
+				vec = (-solverConstraint.m_angularComponentB).cross(rel_pos2);
 				denom1 = rb1->getInvMass() + contactNormal.dot(vec);
 			}
 		}
 
-		 
-
-		 btScalar d = denom0+denom1+cfm;
-		 if (d>SIMD_EPSILON)
-		 {
-			solverConstraint.m_jacDiagABInv = relaxation/(d);
-		 } else
-		 {
+		btScalar d = denom0 + denom1 + cfm;
+		if (d > SIMD_EPSILON)
+		{
+			solverConstraint.m_jacDiagABInv = relaxation / (d);
+		}
+		else
+		{
 			//disable the constraint row to handle singularity/redundant constraint
-			solverConstraint.m_jacDiagABInv  = 0.f;
-		 }
-		
+			solverConstraint.m_jacDiagABInv = 0.f;
+		}
 	}
 
-	
 	//compute rhs and remaining solverConstraint fields
 
-	
-
 	btScalar restitution = 0.f;
-    btScalar distance = 0;
-    if (!isFriction)
-    {
-        distance = cp.getDistance()+infoGlobal.m_linearSlop;
-    } else
-    {
-        if (cp.m_contactPointFlags & BT_CONTACT_FLAG_FRICTION_ANCHOR)
-        {
-          distance = (cp.getPositionWorldOnA() - cp.getPositionWorldOnB()).dot(contactNormal);
-        }
-    }
-  
-    
-	btScalar rel_vel = 0.f;
-	int ndofA  = 0;
-	int ndofB  = 0;
+	btScalar distance = 0;
+	if (!isFriction)
 	{
+		distance = cp.getDistance() + infoGlobal.m_linearSlop;
+	}
+	else
+	{
+		if (cp.m_contactPointFlags & BT_CONTACT_FLAG_FRICTION_ANCHOR)
+		{
+			distance = (cp.getPositionWorldOnA() - cp.getPositionWorldOnB()).dot(contactNormal);
+		}
+	}
 
-		btVector3 vel1,vel2;
+	btScalar rel_vel = 0.f;
+	int ndofA = 0;
+	int ndofB = 0;
+	{
+		btVector3 vel1, vel2;
 		if (multiBodyA)
 		{
-			ndofA  = multiBodyA->getNumDofs() + 6;
+			ndofA = multiBodyA->getNumDofs() + 6;
 			btScalar* jacA = &m_data.m_jacobians[solverConstraint.m_jacAindex];
-			for (int i = 0; i < ndofA ; ++i) 
+			for (int i = 0; i < ndofA; ++i)
 				rel_vel += multiBodyA->getVelocityVector()[i] * jacA[i];
-		} else
+		}
+		else
 		{
 			if (rb0)
 			{
-				rel_vel += (rb0->getVelocityInLocalPoint(rel_pos1) + 
-							(rb0->getTotalTorque()*rb0->getInvInertiaTensorWorld()*infoGlobal.m_timeStep).cross(rel_pos1)+
-							rb0->getTotalForce()*rb0->getInvMass()*infoGlobal.m_timeStep).dot(solverConstraint.m_contactNormal1);
+				rel_vel += (rb0->getVelocityInLocalPoint(rel_pos1) +
+							(rb0->getTotalTorque() * rb0->getInvInertiaTensorWorld() * infoGlobal.m_timeStep).cross(rel_pos1) +
+							rb0->getTotalForce() * rb0->getInvMass() * infoGlobal.m_timeStep)
+							   .dot(solverConstraint.m_contactNormal1);
 			}
 		}
 		if (multiBodyB)
 		{
-			ndofB  = multiBodyB->getNumDofs() + 6;
+			ndofB = multiBodyB->getNumDofs() + 6;
 			btScalar* jacB = &m_data.m_jacobians[solverConstraint.m_jacBindex];
-			for (int i = 0; i < ndofB ; ++i) 
+			for (int i = 0; i < ndofB; ++i)
 				rel_vel += multiBodyB->getVelocityVector()[i] * jacB[i];
-
-		} else
+		}
+		else
 		{
 			if (rb1)
 			{
-				rel_vel += (rb1->getVelocityInLocalPoint(rel_pos2)+
-					(rb1->getTotalTorque()*rb1->getInvInertiaTensorWorld()*infoGlobal.m_timeStep).cross(rel_pos2) +
-					rb1->getTotalForce()*rb1->getInvMass()*infoGlobal.m_timeStep).dot(solverConstraint.m_contactNormal2);
+				rel_vel += (rb1->getVelocityInLocalPoint(rel_pos2) +
+							(rb1->getTotalTorque() * rb1->getInvInertiaTensorWorld() * infoGlobal.m_timeStep).cross(rel_pos2) +
+							rb1->getTotalForce() * rb1->getInvMass() * infoGlobal.m_timeStep)
+							   .dot(solverConstraint.m_contactNormal2);
 			}
 		}
 
 		solverConstraint.m_friction = cp.m_combinedFriction;
 
-		if(!isFriction)
+		if (!isFriction)
 		{
-			restitution =  restitutionCurve(rel_vel, cp.m_combinedRestitution, infoGlobal.m_restitutionVelocityThreshold);	
+			restitution = restitutionCurve(rel_vel, cp.m_combinedRestitution, infoGlobal.m_restitutionVelocityThreshold);
 			if (restitution <= btScalar(0.))
 			{
 				restitution = 0.f;
@@ -751,10 +745,9 @@ void btMultiBodyConstraintSolver::setupMultiBodyContactConstraint(btMultiBodySol
 		}
 	}
 
-
 	///warm starting (or zero if disabled)
 	//disable warmstarting for btMultiBody, it has issues gaining energy (==explosion)
-	if (0)//infoGlobal.m_solverMode & SOLVER_USE_WARMSTARTING)
+	if (0)  //infoGlobal.m_solverMode & SOLVER_USE_WARMSTARTING)
 	{
 		solverConstraint.m_appliedImpulse = isFriction ? 0 : cp.m_appliedImpulse * infoGlobal.m_warmstartingFactor;
 
@@ -764,27 +757,30 @@ void btMultiBodyConstraintSolver::setupMultiBodyContactConstraint(btMultiBodySol
 			{
 				btScalar impulse = solverConstraint.m_appliedImpulse;
 				btScalar* deltaV = &m_data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacAindex];
-				multiBodyA->applyDeltaVeeMultiDof(deltaV,impulse);
-				
-				applyDeltaVee(deltaV,impulse,solverConstraint.m_deltaVelAindex,ndofA);
-			} else
+				multiBodyA->applyDeltaVeeMultiDof(deltaV, impulse);
+
+				applyDeltaVee(deltaV, impulse, solverConstraint.m_deltaVelAindex, ndofA);
+			}
+			else
 			{
 				if (rb0)
-					bodyA->internalApplyImpulse(solverConstraint.m_contactNormal1*bodyA->internalGetInvMass()*rb0->getLinearFactor(),solverConstraint.m_angularComponentA,solverConstraint.m_appliedImpulse);
+					bodyA->internalApplyImpulse(solverConstraint.m_contactNormal1 * bodyA->internalGetInvMass() * rb0->getLinearFactor(), solverConstraint.m_angularComponentA, solverConstraint.m_appliedImpulse);
 			}
 			if (multiBodyB)
 			{
 				btScalar impulse = solverConstraint.m_appliedImpulse;
 				btScalar* deltaV = &m_data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacBindex];
-				multiBodyB->applyDeltaVeeMultiDof(deltaV,impulse);
-				applyDeltaVee(deltaV,impulse,solverConstraint.m_deltaVelBindex,ndofB);
-			} else
+				multiBodyB->applyDeltaVeeMultiDof(deltaV, impulse);
+				applyDeltaVee(deltaV, impulse, solverConstraint.m_deltaVelBindex, ndofB);
+			}
+			else
 			{
 				if (rb1)
-					bodyB->internalApplyImpulse(-solverConstraint.m_contactNormal2*bodyB->internalGetInvMass()*rb1->getLinearFactor(),-solverConstraint.m_angularComponentB,-(btScalar)solverConstraint.m_appliedImpulse);
+					bodyB->internalApplyImpulse(-solverConstraint.m_contactNormal2 * bodyB->internalGetInvMass() * rb1->getLinearFactor(), -solverConstraint.m_angularComponentB, -(btScalar)solverConstraint.m_appliedImpulse);
 			}
 		}
-	} else
+	}
+	else
 	{
 		solverConstraint.m_appliedImpulse = 0.f;
 	}
@@ -792,38 +788,37 @@ void btMultiBodyConstraintSolver::setupMultiBodyContactConstraint(btMultiBodySol
 	solverConstraint.m_appliedPushImpulse = 0.f;
 
 	{
-
 		btScalar positionalError = 0.f;
-		btScalar velocityError = restitution - rel_vel;// * damping;	//note for friction restitution is always set to 0 (check above) so it is acutally velocityError = -rel_vel for friction
+		btScalar velocityError = restitution - rel_vel;  // * damping;	//note for friction restitution is always set to 0 (check above) so it is acutally velocityError = -rel_vel for friction
 		if (isFriction)
 		{
-			positionalError = -distance * erp/infoGlobal.m_timeStep;
-		} else
+			positionalError = -distance * erp / infoGlobal.m_timeStep;
+		}
+		else
 		{
-    			if (distance>0)
+			if (distance > 0)
 			{
 				positionalError = 0;
 				velocityError -= distance / infoGlobal.m_timeStep;
-
-			} else
+			}
+			else
 			{
-				positionalError = -distance * erp/infoGlobal.m_timeStep;
+				positionalError = -distance * erp / infoGlobal.m_timeStep;
 			}
 		}
 
-		btScalar  penetrationImpulse = positionalError*solverConstraint.m_jacDiagABInv;
-		btScalar velocityImpulse = velocityError *solverConstraint.m_jacDiagABInv;
+		btScalar penetrationImpulse = positionalError * solverConstraint.m_jacDiagABInv;
+		btScalar velocityImpulse = velocityError * solverConstraint.m_jacDiagABInv;
 
-		if(!isFriction)
+		if (!isFriction)
 		{
-		//	if (!infoGlobal.m_splitImpulse || (penetration > infoGlobal.m_splitImpulsePenetrationThreshold))
+			//	if (!infoGlobal.m_splitImpulse || (penetration > infoGlobal.m_splitImpulsePenetrationThreshold))
 			{
 				//combine position and velocity into rhs
-				solverConstraint.m_rhs = penetrationImpulse+velocityImpulse;
+				solverConstraint.m_rhs = penetrationImpulse + velocityImpulse;
 				solverConstraint.m_rhsPenetration = 0.f;
-
 			}
-		/*else
+			/*else
 			{
 				//split position and velocity into rhs and m_rhsPenetration
 				solverConstraint.m_rhs = velocityImpulse;
@@ -835,309 +830,288 @@ void btMultiBodyConstraintSolver::setupMultiBodyContactConstraint(btMultiBodySol
 		}
 		else
 		{
-			solverConstraint.m_rhs = penetrationImpulse+velocityImpulse;
+			solverConstraint.m_rhs = penetrationImpulse + velocityImpulse;
 			solverConstraint.m_rhsPenetration = 0.f;
 			solverConstraint.m_lowerLimit = -solverConstraint.m_friction;
 			solverConstraint.m_upperLimit = solverConstraint.m_friction;
 		}
 
-		solverConstraint.m_cfm = cfm*solverConstraint.m_jacDiagABInv;
-		
-
-
+		solverConstraint.m_cfm = cfm * solverConstraint.m_jacDiagABInv;
 	}
-
 }
 
 void btMultiBodyConstraintSolver::setupMultiBodyTorsionalFrictionConstraint(btMultiBodySolverConstraint& solverConstraint,
-                                                                  const btVector3& constraintNormal,
-                                                                  btManifoldPoint& cp,
-                                                                    btScalar combinedTorsionalFriction,
-                                                                    const btContactSolverInfo& infoGlobal,
-                                                                  btScalar& relaxation,
-                                                                  bool isFriction, btScalar desiredVelocity, btScalar cfmSlip)
+																			const btVector3& constraintNormal,
+																			btManifoldPoint& cp,
+																			btScalar combinedTorsionalFriction,
+																			const btContactSolverInfo& infoGlobal,
+																			btScalar& relaxation,
+																			bool isFriction, btScalar desiredVelocity, btScalar cfmSlip)
 {
-    
-    BT_PROFILE("setupMultiBodyRollingFrictionConstraint");
-    btVector3 rel_pos1;
-    btVector3 rel_pos2;
-    
-    btMultiBody* multiBodyA = solverConstraint.m_multiBodyA;
-    btMultiBody* multiBodyB = solverConstraint.m_multiBodyB;
-    
-    const btVector3& pos1 = cp.getPositionWorldOnA();
-    const btVector3& pos2 = cp.getPositionWorldOnB();
-    
-    btSolverBody* bodyA = multiBodyA ? 0 : &m_tmpSolverBodyPool[solverConstraint.m_solverBodyIdA];
-    btSolverBody* bodyB = multiBodyB ? 0 : &m_tmpSolverBodyPool[solverConstraint.m_solverBodyIdB];
-    
-    btRigidBody* rb0 = multiBodyA ? 0 : bodyA->m_originalBody;
-    btRigidBody* rb1 = multiBodyB ? 0 : bodyB->m_originalBody;
-    
-    if (bodyA)
-        rel_pos1 = pos1 - bodyA->getWorldTransform().getOrigin();
-    if (bodyB)
-        rel_pos2 = pos2 - bodyB->getWorldTransform().getOrigin();
-    
-    relaxation = infoGlobal.m_sor;
-    
-   // btScalar invTimeStep = btScalar(1)/infoGlobal.m_timeStep;
-    
-    
-    if (multiBodyA)
-    {
-        if (solverConstraint.m_linkA<0)
-        {
-            rel_pos1 = pos1 - multiBodyA->getBasePos();
-        } else
-        {
-            rel_pos1 = pos1 - multiBodyA->getLink(solverConstraint.m_linkA).m_cachedWorldTransform.getOrigin();
-        }
-        const int ndofA  = multiBodyA->getNumDofs() + 6;
-        
-        solverConstraint.m_deltaVelAindex = multiBodyA->getCompanionId();
-        
-        if (solverConstraint.m_deltaVelAindex <0)
-        {
-            solverConstraint.m_deltaVelAindex = m_data.m_deltaVelocities.size();
-            multiBodyA->setCompanionId(solverConstraint.m_deltaVelAindex);
-            m_data.m_deltaVelocities.resize(m_data.m_deltaVelocities.size()+ndofA);
-        } else
-        {
-            btAssert(m_data.m_deltaVelocities.size() >= solverConstraint.m_deltaVelAindex+ndofA);
-        }
-        
-        solverConstraint.m_jacAindex = m_data.m_jacobians.size();
-        m_data.m_jacobians.resize(m_data.m_jacobians.size()+ndofA);
-        m_data.m_deltaVelocitiesUnitImpulse.resize(m_data.m_deltaVelocitiesUnitImpulse.size()+ndofA);
-        btAssert(m_data.m_jacobians.size() == m_data.m_deltaVelocitiesUnitImpulse.size());
-        
-        btScalar* jac1=&m_data.m_jacobians[solverConstraint.m_jacAindex];
-        multiBodyA->fillConstraintJacobianMultiDof(solverConstraint.m_linkA, cp.getPositionWorldOnA(), constraintNormal, btVector3(0,0,0), jac1, m_data.scratch_r, m_data.scratch_v, m_data.scratch_m);
-        btScalar* delta = &m_data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacAindex];
-        multiBodyA->calcAccelerationDeltasMultiDof(&m_data.m_jacobians[solverConstraint.m_jacAindex],delta,m_data.scratch_r, m_data.scratch_v);
-        
-        btVector3 torqueAxis0 = -constraintNormal;
-        solverConstraint.m_relpos1CrossNormal = torqueAxis0;
-        solverConstraint.m_contactNormal1 = btVector3(0,0,0);
-    } else
-    {
-        btVector3 torqueAxis0 = -constraintNormal;
-        solverConstraint.m_relpos1CrossNormal = torqueAxis0;
-        solverConstraint.m_contactNormal1 = btVector3(0,0,0);
-        solverConstraint.m_angularComponentA = rb0 ? rb0->getInvInertiaTensorWorld()*torqueAxis0*rb0->getAngularFactor() : btVector3(0,0,0);
-    }
-    
-    
-    
-    if (multiBodyB)
-    {
-        if (solverConstraint.m_linkB<0)
-        {
-            rel_pos2 = pos2 - multiBodyB->getBasePos();
-        } else
-        {
-            rel_pos2 = pos2 - multiBodyB->getLink(solverConstraint.m_linkB).m_cachedWorldTransform.getOrigin();
-        }
-        
-        const int ndofB  = multiBodyB->getNumDofs() + 6;
-        
-        solverConstraint.m_deltaVelBindex = multiBodyB->getCompanionId();
-        if (solverConstraint.m_deltaVelBindex <0)
-        {
-            solverConstraint.m_deltaVelBindex = m_data.m_deltaVelocities.size();
-            multiBodyB->setCompanionId(solverConstraint.m_deltaVelBindex);
-            m_data.m_deltaVelocities.resize(m_data.m_deltaVelocities.size()+ndofB);
-        }
-        
-        solverConstraint.m_jacBindex = m_data.m_jacobians.size();
-        
-        m_data.m_jacobians.resize(m_data.m_jacobians.size()+ndofB);
-        m_data.m_deltaVelocitiesUnitImpulse.resize(m_data.m_deltaVelocitiesUnitImpulse.size()+ndofB);
-        btAssert(m_data.m_jacobians.size() == m_data.m_deltaVelocitiesUnitImpulse.size());
-        
-        multiBodyB->fillConstraintJacobianMultiDof(solverConstraint.m_linkB, cp.getPositionWorldOnB(), -constraintNormal, btVector3(0,0,0), &m_data.m_jacobians[solverConstraint.m_jacBindex], m_data.scratch_r, m_data.scratch_v, m_data.scratch_m);
-        multiBodyB->calcAccelerationDeltasMultiDof(&m_data.m_jacobians[solverConstraint.m_jacBindex],&m_data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacBindex],m_data.scratch_r, m_data.scratch_v);
-        
-        btVector3 torqueAxis1 = constraintNormal;
-        solverConstraint.m_relpos2CrossNormal = torqueAxis1;
-        solverConstraint.m_contactNormal2 = -btVector3(0,0,0);
-        
-    } else
-    {
-        btVector3 torqueAxis1 = constraintNormal;
-        solverConstraint.m_relpos2CrossNormal = torqueAxis1;
-        solverConstraint.m_contactNormal2 = -btVector3(0,0,0);
-        
-        solverConstraint.m_angularComponentB = rb1 ? rb1->getInvInertiaTensorWorld()*torqueAxis1*rb1->getAngularFactor() : btVector3(0,0,0);
-    }
-    
-    {
-        
-        btScalar denom0 = 0.f;
-        btScalar denom1 = 0.f;
-        btScalar* jacB = 0;
-        btScalar* jacA = 0;
-        btScalar* lambdaA =0;
-        btScalar* lambdaB =0;
-        int ndofA  = 0;
-        if (multiBodyA)
-        {
-            ndofA  = multiBodyA->getNumDofs() + 6;
-            jacA = &m_data.m_jacobians[solverConstraint.m_jacAindex];
-            lambdaA = &m_data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacAindex];
-            for (int i = 0; i < ndofA; ++i)
-            {
-                btScalar j = jacA[i] ;
-                btScalar l =lambdaA[i];
-                denom0 += j*l;
-            }
-        } else
-        {
-            if (rb0)
-            {
-				btVector3 iMJaA = rb0?rb0->getInvInertiaTensorWorld()*solverConstraint.m_relpos1CrossNormal:btVector3(0,0,0);
+	BT_PROFILE("setupMultiBodyRollingFrictionConstraint");
+	btVector3 rel_pos1;
+	btVector3 rel_pos2;
+
+	btMultiBody* multiBodyA = solverConstraint.m_multiBodyA;
+	btMultiBody* multiBodyB = solverConstraint.m_multiBodyB;
+
+	const btVector3& pos1 = cp.getPositionWorldOnA();
+	const btVector3& pos2 = cp.getPositionWorldOnB();
+
+	btSolverBody* bodyA = multiBodyA ? 0 : &m_tmpSolverBodyPool[solverConstraint.m_solverBodyIdA];
+	btSolverBody* bodyB = multiBodyB ? 0 : &m_tmpSolverBodyPool[solverConstraint.m_solverBodyIdB];
+
+	btRigidBody* rb0 = multiBodyA ? 0 : bodyA->m_originalBody;
+	btRigidBody* rb1 = multiBodyB ? 0 : bodyB->m_originalBody;
+
+	if (bodyA)
+		rel_pos1 = pos1 - bodyA->getWorldTransform().getOrigin();
+	if (bodyB)
+		rel_pos2 = pos2 - bodyB->getWorldTransform().getOrigin();
+
+	relaxation = infoGlobal.m_sor;
+
+	// btScalar invTimeStep = btScalar(1)/infoGlobal.m_timeStep;
+
+	if (multiBodyA)
+	{
+		if (solverConstraint.m_linkA < 0)
+		{
+			rel_pos1 = pos1 - multiBodyA->getBasePos();
+		}
+		else
+		{
+			rel_pos1 = pos1 - multiBodyA->getLink(solverConstraint.m_linkA).m_cachedWorldTransform.getOrigin();
+		}
+		const int ndofA = multiBodyA->getNumDofs() + 6;
+
+		solverConstraint.m_deltaVelAindex = multiBodyA->getCompanionId();
+
+		if (solverConstraint.m_deltaVelAindex < 0)
+		{
+			solverConstraint.m_deltaVelAindex = m_data.m_deltaVelocities.size();
+			multiBodyA->setCompanionId(solverConstraint.m_deltaVelAindex);
+			m_data.m_deltaVelocities.resize(m_data.m_deltaVelocities.size() + ndofA);
+		}
+		else
+		{
+			btAssert(m_data.m_deltaVelocities.size() >= solverConstraint.m_deltaVelAindex + ndofA);
+		}
+
+		solverConstraint.m_jacAindex = m_data.m_jacobians.size();
+		m_data.m_jacobians.resize(m_data.m_jacobians.size() + ndofA);
+		m_data.m_deltaVelocitiesUnitImpulse.resize(m_data.m_deltaVelocitiesUnitImpulse.size() + ndofA);
+		btAssert(m_data.m_jacobians.size() == m_data.m_deltaVelocitiesUnitImpulse.size());
+
+		btScalar* jac1 = &m_data.m_jacobians[solverConstraint.m_jacAindex];
+		multiBodyA->fillConstraintJacobianMultiDof(solverConstraint.m_linkA, cp.getPositionWorldOnA(), constraintNormal, btVector3(0, 0, 0), jac1, m_data.scratch_r, m_data.scratch_v, m_data.scratch_m);
+		btScalar* delta = &m_data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacAindex];
+		multiBodyA->calcAccelerationDeltasMultiDof(&m_data.m_jacobians[solverConstraint.m_jacAindex], delta, m_data.scratch_r, m_data.scratch_v);
+
+		btVector3 torqueAxis0 = -constraintNormal;
+		solverConstraint.m_relpos1CrossNormal = torqueAxis0;
+		solverConstraint.m_contactNormal1 = btVector3(0, 0, 0);
+	}
+	else
+	{
+		btVector3 torqueAxis0 = -constraintNormal;
+		solverConstraint.m_relpos1CrossNormal = torqueAxis0;
+		solverConstraint.m_contactNormal1 = btVector3(0, 0, 0);
+		solverConstraint.m_angularComponentA = rb0 ? rb0->getInvInertiaTensorWorld() * torqueAxis0 * rb0->getAngularFactor() : btVector3(0, 0, 0);
+	}
+
+	if (multiBodyB)
+	{
+		if (solverConstraint.m_linkB < 0)
+		{
+			rel_pos2 = pos2 - multiBodyB->getBasePos();
+		}
+		else
+		{
+			rel_pos2 = pos2 - multiBodyB->getLink(solverConstraint.m_linkB).m_cachedWorldTransform.getOrigin();
+		}
+
+		const int ndofB = multiBodyB->getNumDofs() + 6;
+
+		solverConstraint.m_deltaVelBindex = multiBodyB->getCompanionId();
+		if (solverConstraint.m_deltaVelBindex < 0)
+		{
+			solverConstraint.m_deltaVelBindex = m_data.m_deltaVelocities.size();
+			multiBodyB->setCompanionId(solverConstraint.m_deltaVelBindex);
+			m_data.m_deltaVelocities.resize(m_data.m_deltaVelocities.size() + ndofB);
+		}
+
+		solverConstraint.m_jacBindex = m_data.m_jacobians.size();
+
+		m_data.m_jacobians.resize(m_data.m_jacobians.size() + ndofB);
+		m_data.m_deltaVelocitiesUnitImpulse.resize(m_data.m_deltaVelocitiesUnitImpulse.size() + ndofB);
+		btAssert(m_data.m_jacobians.size() == m_data.m_deltaVelocitiesUnitImpulse.size());
+
+		multiBodyB->fillConstraintJacobianMultiDof(solverConstraint.m_linkB, cp.getPositionWorldOnB(), -constraintNormal, btVector3(0, 0, 0), &m_data.m_jacobians[solverConstraint.m_jacBindex], m_data.scratch_r, m_data.scratch_v, m_data.scratch_m);
+		multiBodyB->calcAccelerationDeltasMultiDof(&m_data.m_jacobians[solverConstraint.m_jacBindex], &m_data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacBindex], m_data.scratch_r, m_data.scratch_v);
+
+		btVector3 torqueAxis1 = constraintNormal;
+		solverConstraint.m_relpos2CrossNormal = torqueAxis1;
+		solverConstraint.m_contactNormal2 = -btVector3(0, 0, 0);
+	}
+	else
+	{
+		btVector3 torqueAxis1 = constraintNormal;
+		solverConstraint.m_relpos2CrossNormal = torqueAxis1;
+		solverConstraint.m_contactNormal2 = -btVector3(0, 0, 0);
+
+		solverConstraint.m_angularComponentB = rb1 ? rb1->getInvInertiaTensorWorld() * torqueAxis1 * rb1->getAngularFactor() : btVector3(0, 0, 0);
+	}
+
+	{
+		btScalar denom0 = 0.f;
+		btScalar denom1 = 0.f;
+		btScalar* jacB = 0;
+		btScalar* jacA = 0;
+		btScalar* lambdaA = 0;
+		btScalar* lambdaB = 0;
+		int ndofA = 0;
+		if (multiBodyA)
+		{
+			ndofA = multiBodyA->getNumDofs() + 6;
+			jacA = &m_data.m_jacobians[solverConstraint.m_jacAindex];
+			lambdaA = &m_data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacAindex];
+			for (int i = 0; i < ndofA; ++i)
+			{
+				btScalar j = jacA[i];
+				btScalar l = lambdaA[i];
+				denom0 += j * l;
+			}
+		}
+		else
+		{
+			if (rb0)
+			{
+				btVector3 iMJaA = rb0 ? rb0->getInvInertiaTensorWorld() * solverConstraint.m_relpos1CrossNormal : btVector3(0, 0, 0);
 				denom0 = iMJaA.dot(solverConstraint.m_relpos1CrossNormal);
-            }
-        }
-        if (multiBodyB)
-        {
-            const int ndofB  = multiBodyB->getNumDofs() + 6;
-            jacB = &m_data.m_jacobians[solverConstraint.m_jacBindex];
-            lambdaB = &m_data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacBindex];
-            for (int i = 0; i < ndofB; ++i)
-            {
-                btScalar j = jacB[i] ;
-                btScalar l =lambdaB[i];
-                denom1 += j*l;
-            }
-            
-        } else
-        {
-            if (rb1)
-            {
-				btVector3 iMJaB = rb1?rb1->getInvInertiaTensorWorld()*solverConstraint.m_relpos2CrossNormal:btVector3(0,0,0);
+			}
+		}
+		if (multiBodyB)
+		{
+			const int ndofB = multiBodyB->getNumDofs() + 6;
+			jacB = &m_data.m_jacobians[solverConstraint.m_jacBindex];
+			lambdaB = &m_data.m_deltaVelocitiesUnitImpulse[solverConstraint.m_jacBindex];
+			for (int i = 0; i < ndofB; ++i)
+			{
+				btScalar j = jacB[i];
+				btScalar l = lambdaB[i];
+				denom1 += j * l;
+			}
+		}
+		else
+		{
+			if (rb1)
+			{
+				btVector3 iMJaB = rb1 ? rb1->getInvInertiaTensorWorld() * solverConstraint.m_relpos2CrossNormal : btVector3(0, 0, 0);
 				denom1 = iMJaB.dot(solverConstraint.m_relpos2CrossNormal);
-            }
-        }
-        
-        
-        
-        btScalar d = denom0+denom1+infoGlobal.m_globalCfm;
-        if (d>SIMD_EPSILON)
-        {
-            solverConstraint.m_jacDiagABInv = relaxation/(d);
-        } else
-        {
-            //disable the constraint row to handle singularity/redundant constraint
-            solverConstraint.m_jacDiagABInv  = 0.f;
-        }
-        
-    }
-    
-    
-    //compute rhs and remaining solverConstraint fields
-    
-    
-    
-    btScalar restitution = 0.f;
-    btScalar penetration = isFriction? 0 : cp.getDistance();
-    
-    btScalar rel_vel = 0.f;
-    int ndofA  = 0;
-    int ndofB  = 0;
-    {
-        
-        btVector3 vel1,vel2;
-        if (multiBodyA)
-        {
-            ndofA  = multiBodyA->getNumDofs() + 6;
-            btScalar* jacA = &m_data.m_jacobians[solverConstraint.m_jacAindex];
-            for (int i = 0; i < ndofA ; ++i)
-                rel_vel += multiBodyA->getVelocityVector()[i] * jacA[i];
-        } else
-        {
-            if (rb0)
-            {
+			}
+		}
+
+		btScalar d = denom0 + denom1 + infoGlobal.m_globalCfm;
+		if (d > SIMD_EPSILON)
+		{
+			solverConstraint.m_jacDiagABInv = relaxation / (d);
+		}
+		else
+		{
+			//disable the constraint row to handle singularity/redundant constraint
+			solverConstraint.m_jacDiagABInv = 0.f;
+		}
+	}
+
+	//compute rhs and remaining solverConstraint fields
+
+	btScalar restitution = 0.f;
+	btScalar penetration = isFriction ? 0 : cp.getDistance();
+
+	btScalar rel_vel = 0.f;
+	int ndofA = 0;
+	int ndofB = 0;
+	{
+		btVector3 vel1, vel2;
+		if (multiBodyA)
+		{
+			ndofA = multiBodyA->getNumDofs() + 6;
+			btScalar* jacA = &m_data.m_jacobians[solverConstraint.m_jacAindex];
+			for (int i = 0; i < ndofA; ++i)
+				rel_vel += multiBodyA->getVelocityVector()[i] * jacA[i];
+		}
+		else
+		{
+			if (rb0)
+			{
 				btSolverBody* solverBodyA = &m_tmpSolverBodyPool[solverConstraint.m_solverBodyIdA];
-				rel_vel += solverConstraint.m_contactNormal1.dot(rb0?solverBodyA->m_linearVelocity+solverBodyA->m_externalForceImpulse:btVector3(0,0,0))
-				+ solverConstraint.m_relpos1CrossNormal.dot(rb0?solverBodyA->m_angularVelocity:btVector3(0,0,0));
-	
-            }
-        }
-        if (multiBodyB)
-        {
-            ndofB  = multiBodyB->getNumDofs() + 6;
-            btScalar* jacB = &m_data.m_jacobians[solverConstraint.m_jacBindex];
-            for (int i = 0; i < ndofB ; ++i)
-                rel_vel += multiBodyB->getVelocityVector()[i] * jacB[i];
-            
-        } else
-        {
-            if (rb1)
-            {
+				rel_vel += solverConstraint.m_contactNormal1.dot(rb0 ? solverBodyA->m_linearVelocity + solverBodyA->m_externalForceImpulse : btVector3(0, 0, 0)) + solverConstraint.m_relpos1CrossNormal.dot(rb0 ? solverBodyA->m_angularVelocity : btVector3(0, 0, 0));
+			}
+		}
+		if (multiBodyB)
+		{
+			ndofB = multiBodyB->getNumDofs() + 6;
+			btScalar* jacB = &m_data.m_jacobians[solverConstraint.m_jacBindex];
+			for (int i = 0; i < ndofB; ++i)
+				rel_vel += multiBodyB->getVelocityVector()[i] * jacB[i];
+		}
+		else
+		{
+			if (rb1)
+			{
 				btSolverBody* solverBodyB = &m_tmpSolverBodyPool[solverConstraint.m_solverBodyIdB];
-				rel_vel += solverConstraint.m_contactNormal2.dot(rb1?solverBodyB->m_linearVelocity+solverBodyB->m_externalForceImpulse:btVector3(0,0,0))
-			+ solverConstraint.m_relpos2CrossNormal.dot(rb1?solverBodyB->m_angularVelocity:btVector3(0,0,0));
+				rel_vel += solverConstraint.m_contactNormal2.dot(rb1 ? solverBodyB->m_linearVelocity + solverBodyB->m_externalForceImpulse : btVector3(0, 0, 0)) + solverConstraint.m_relpos2CrossNormal.dot(rb1 ? solverBodyB->m_angularVelocity : btVector3(0, 0, 0));
+			}
+		}
 
-            }
-        }
+		solverConstraint.m_friction = combinedTorsionalFriction;
 
-        solverConstraint.m_friction =combinedTorsionalFriction;
-        
-        if(!isFriction)
-        {
-            restitution =  restitutionCurve(rel_vel, cp.m_combinedRestitution, infoGlobal.m_restitutionVelocityThreshold);
-            if (restitution <= btScalar(0.))
-            {
-                restitution = 0.f;
-            }
-        }
-    }
-    
-    
-    solverConstraint.m_appliedImpulse = 0.f;
-    solverConstraint.m_appliedPushImpulse = 0.f;
-    
-    {
-        
-        btScalar velocityError = 0 - rel_vel;// * damping;	//note for friction restitution is always set to 0 (check above) so it is acutally velocityError = -rel_vel for friction
-        
-       
-        
-        btScalar velocityImpulse = velocityError*solverConstraint.m_jacDiagABInv;
-        
-        solverConstraint.m_rhs = velocityImpulse;
-        solverConstraint.m_rhsPenetration = 0.f;
-        solverConstraint.m_lowerLimit = -solverConstraint.m_friction;
-        solverConstraint.m_upperLimit = solverConstraint.m_friction;
-        
-        solverConstraint.m_cfm = infoGlobal.m_globalCfm*solverConstraint.m_jacDiagABInv;
-        
-        
-        
-    }
-    
+		if (!isFriction)
+		{
+			restitution = restitutionCurve(rel_vel, cp.m_combinedRestitution, infoGlobal.m_restitutionVelocityThreshold);
+			if (restitution <= btScalar(0.))
+			{
+				restitution = 0.f;
+			}
+		}
+	}
+
+	solverConstraint.m_appliedImpulse = 0.f;
+	solverConstraint.m_appliedPushImpulse = 0.f;
+
+	{
+		btScalar velocityError = 0 - rel_vel;  // * damping;	//note for friction restitution is always set to 0 (check above) so it is acutally velocityError = -rel_vel for friction
+
+		btScalar velocityImpulse = velocityError * solverConstraint.m_jacDiagABInv;
+
+		solverConstraint.m_rhs = velocityImpulse;
+		solverConstraint.m_rhsPenetration = 0.f;
+		solverConstraint.m_lowerLimit = -solverConstraint.m_friction;
+		solverConstraint.m_upperLimit = solverConstraint.m_friction;
+
+		solverConstraint.m_cfm = infoGlobal.m_globalCfm * solverConstraint.m_jacDiagABInv;
+	}
 }
 
-btMultiBodySolverConstraint&	btMultiBodyConstraintSolver::addMultiBodyFrictionConstraint(const btVector3& normalAxis,btPersistentManifold* manifold,int frictionIndex,btManifoldPoint& cp,btCollisionObject* colObj0,btCollisionObject* colObj1, btScalar relaxation, const btContactSolverInfo& infoGlobal, btScalar desiredVelocity, btScalar cfmSlip)
+btMultiBodySolverConstraint& btMultiBodyConstraintSolver::addMultiBodyFrictionConstraint(const btVector3& normalAxis, btPersistentManifold* manifold, int frictionIndex, btManifoldPoint& cp, btCollisionObject* colObj0, btCollisionObject* colObj1, btScalar relaxation, const btContactSolverInfo& infoGlobal, btScalar desiredVelocity, btScalar cfmSlip)
 {
 	BT_PROFILE("addMultiBodyFrictionConstraint");
 	btMultiBodySolverConstraint& solverConstraint = m_multiBodyFrictionContactConstraints.expandNonInitializing();
-    solverConstraint.m_orgConstraint = 0;
-    solverConstraint.m_orgDofIndex = -1;
-    
+	solverConstraint.m_orgConstraint = 0;
+	solverConstraint.m_orgDofIndex = -1;
+
 	solverConstraint.m_frictionIndex = frictionIndex;
 	bool isFriction = true;
 
 	const btMultiBodyLinkCollider* fcA = btMultiBodyLinkCollider::upcast(manifold->getBody0());
 	const btMultiBodyLinkCollider* fcB = btMultiBodyLinkCollider::upcast(manifold->getBody1());
-	
-	btMultiBody* mbA = fcA? fcA->m_multiBody : 0;
-	btMultiBody* mbB = fcB? fcB->m_multiBody : 0;
 
-	int solverBodyIdA = mbA? -1 : getOrInitSolverBody(*colObj0,infoGlobal.m_timeStep);
-	int solverBodyIdB = mbB ? -1 : getOrInitSolverBody(*colObj1,infoGlobal.m_timeStep);
+	btMultiBody* mbA = fcA ? fcA->m_multiBody : 0;
+	btMultiBody* mbB = fcB ? fcB->m_multiBody : 0;
+
+	int solverBodyIdA = mbA ? -1 : getOrInitSolverBody(*colObj0, infoGlobal.m_timeStep);
+	int solverBodyIdB = mbB ? -1 : getOrInitSolverBody(*colObj1, infoGlobal.m_timeStep);
 
 	solverConstraint.m_solverBodyIdA = solverBodyIdA;
 	solverConstraint.m_solverBodyIdB = solverBodyIdB;
@@ -1151,95 +1125,92 @@ btMultiBodySolverConstraint&	btMultiBodyConstraintSolver::addMultiBodyFrictionCo
 
 	solverConstraint.m_originalContactPoint = &cp;
 
-	setupMultiBodyContactConstraint(solverConstraint, normalAxis, cp, infoGlobal,relaxation,isFriction, desiredVelocity, cfmSlip);
+	setupMultiBodyContactConstraint(solverConstraint, normalAxis, cp, infoGlobal, relaxation, isFriction, desiredVelocity, cfmSlip);
 	return solverConstraint;
 }
 
-btMultiBodySolverConstraint&	btMultiBodyConstraintSolver::addMultiBodyTorsionalFrictionConstraint(const btVector3& normalAxis,btPersistentManifold* manifold,int frictionIndex,btManifoldPoint& cp,
-                                                                btScalar combinedTorsionalFriction,
-                                                                                                     btCollisionObject* colObj0,btCollisionObject* colObj1, btScalar relaxation, const btContactSolverInfo& infoGlobal, btScalar desiredVelocity, btScalar cfmSlip)
+btMultiBodySolverConstraint& btMultiBodyConstraintSolver::addMultiBodyTorsionalFrictionConstraint(const btVector3& normalAxis, btPersistentManifold* manifold, int frictionIndex, btManifoldPoint& cp,
+																								  btScalar combinedTorsionalFriction,
+																								  btCollisionObject* colObj0, btCollisionObject* colObj1, btScalar relaxation, const btContactSolverInfo& infoGlobal, btScalar desiredVelocity, btScalar cfmSlip)
 {
-    BT_PROFILE("addMultiBodyRollingFrictionConstraint");
+	BT_PROFILE("addMultiBodyRollingFrictionConstraint");
 
-	bool useTorsionalAndConeFriction = (infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS && ((infoGlobal.m_solverMode&SOLVER_DISABLE_IMPLICIT_CONE_FRICTION) == 0));
+	bool useTorsionalAndConeFriction = (infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS && ((infoGlobal.m_solverMode & SOLVER_DISABLE_IMPLICIT_CONE_FRICTION) == 0));
 
-    btMultiBodySolverConstraint& solverConstraint = useTorsionalAndConeFriction? m_multiBodyTorsionalFrictionContactConstraints.expandNonInitializing() : m_multiBodyFrictionContactConstraints.expandNonInitializing();
-    solverConstraint.m_orgConstraint = 0;
-    solverConstraint.m_orgDofIndex = -1;
-    
-    solverConstraint.m_frictionIndex = frictionIndex;
-    bool isFriction = true;
-    
-    const btMultiBodyLinkCollider* fcA = btMultiBodyLinkCollider::upcast(manifold->getBody0());
-    const btMultiBodyLinkCollider* fcB = btMultiBodyLinkCollider::upcast(manifold->getBody1());
-    
-    btMultiBody* mbA = fcA? fcA->m_multiBody : 0;
-    btMultiBody* mbB = fcB? fcB->m_multiBody : 0;
-    
-    int solverBodyIdA = mbA? -1 : getOrInitSolverBody(*colObj0,infoGlobal.m_timeStep);
-    int solverBodyIdB = mbB ? -1 : getOrInitSolverBody(*colObj1,infoGlobal.m_timeStep);
-    
-    solverConstraint.m_solverBodyIdA = solverBodyIdA;
-    solverConstraint.m_solverBodyIdB = solverBodyIdB;
-    solverConstraint.m_multiBodyA = mbA;
-    if (mbA)
-        solverConstraint.m_linkA = fcA->m_link;
-    
-    solverConstraint.m_multiBodyB = mbB;
-    if (mbB)
-        solverConstraint.m_linkB = fcB->m_link;
-    
-    solverConstraint.m_originalContactPoint = &cp;
-    
-    setupMultiBodyTorsionalFrictionConstraint(solverConstraint, normalAxis, cp, combinedTorsionalFriction,infoGlobal,relaxation,isFriction, desiredVelocity, cfmSlip);
-    return solverConstraint;
+	btMultiBodySolverConstraint& solverConstraint = useTorsionalAndConeFriction ? m_multiBodyTorsionalFrictionContactConstraints.expandNonInitializing() : m_multiBodyFrictionContactConstraints.expandNonInitializing();
+	solverConstraint.m_orgConstraint = 0;
+	solverConstraint.m_orgDofIndex = -1;
+
+	solverConstraint.m_frictionIndex = frictionIndex;
+	bool isFriction = true;
+
+	const btMultiBodyLinkCollider* fcA = btMultiBodyLinkCollider::upcast(manifold->getBody0());
+	const btMultiBodyLinkCollider* fcB = btMultiBodyLinkCollider::upcast(manifold->getBody1());
+
+	btMultiBody* mbA = fcA ? fcA->m_multiBody : 0;
+	btMultiBody* mbB = fcB ? fcB->m_multiBody : 0;
+
+	int solverBodyIdA = mbA ? -1 : getOrInitSolverBody(*colObj0, infoGlobal.m_timeStep);
+	int solverBodyIdB = mbB ? -1 : getOrInitSolverBody(*colObj1, infoGlobal.m_timeStep);
+
+	solverConstraint.m_solverBodyIdA = solverBodyIdA;
+	solverConstraint.m_solverBodyIdB = solverBodyIdB;
+	solverConstraint.m_multiBodyA = mbA;
+	if (mbA)
+		solverConstraint.m_linkA = fcA->m_link;
+
+	solverConstraint.m_multiBodyB = mbB;
+	if (mbB)
+		solverConstraint.m_linkB = fcB->m_link;
+
+	solverConstraint.m_originalContactPoint = &cp;
+
+	setupMultiBodyTorsionalFrictionConstraint(solverConstraint, normalAxis, cp, combinedTorsionalFriction, infoGlobal, relaxation, isFriction, desiredVelocity, cfmSlip);
+	return solverConstraint;
 }
 
-void	btMultiBodyConstraintSolver::convertMultiBodyContact(btPersistentManifold* manifold,const btContactSolverInfo& infoGlobal)
+void btMultiBodyConstraintSolver::convertMultiBodyContact(btPersistentManifold* manifold, const btContactSolverInfo& infoGlobal)
 {
 	const btMultiBodyLinkCollider* fcA = btMultiBodyLinkCollider::upcast(manifold->getBody0());
 	const btMultiBodyLinkCollider* fcB = btMultiBodyLinkCollider::upcast(manifold->getBody1());
-	
-	btMultiBody* mbA = fcA? fcA->m_multiBody : 0;
-	btMultiBody* mbB = fcB? fcB->m_multiBody : 0;
 
-	btCollisionObject* colObj0=0,*colObj1=0;
+	btMultiBody* mbA = fcA ? fcA->m_multiBody : 0;
+	btMultiBody* mbB = fcB ? fcB->m_multiBody : 0;
+
+	btCollisionObject *colObj0 = 0, *colObj1 = 0;
 
 	colObj0 = (btCollisionObject*)manifold->getBody0();
 	colObj1 = (btCollisionObject*)manifold->getBody1();
 
-	int solverBodyIdA = mbA? -1 : getOrInitSolverBody(*colObj0,infoGlobal.m_timeStep);
-	int solverBodyIdB = mbB ? -1 : getOrInitSolverBody(*colObj1,infoGlobal.m_timeStep);
-
-//	btSolverBody* solverBodyA = mbA ? 0 : &m_tmpSolverBodyPool[solverBodyIdA];
-//	btSolverBody* solverBodyB = mbB ? 0 : &m_tmpSolverBodyPool[solverBodyIdB];
+	int solverBodyIdA = mbA ? -1 : getOrInitSolverBody(*colObj0, infoGlobal.m_timeStep);
+	int solverBodyIdB = mbB ? -1 : getOrInitSolverBody(*colObj1, infoGlobal.m_timeStep);
 
+	//	btSolverBody* solverBodyA = mbA ? 0 : &m_tmpSolverBodyPool[solverBodyIdA];
+	//	btSolverBody* solverBodyB = mbB ? 0 : &m_tmpSolverBodyPool[solverBodyIdB];
 
 	///avoid collision response between two static objects
-//	if (!solverBodyA || (solverBodyA->m_invMass.isZero() && (!solverBodyB || solverBodyB->m_invMass.isZero())))
+	//	if (!solverBodyA || (solverBodyA->m_invMass.isZero() && (!solverBodyB || solverBodyB->m_invMass.isZero())))
 	//	return;
 
-    //only a single rollingFriction per manifold
-    int rollingFriction=1;
-    
-	for (int j=0;j<manifold->getNumContacts();j++)
-	{
+	//only a single rollingFriction per manifold
+	int rollingFriction = 1;
 
+	for (int j = 0; j < manifold->getNumContacts(); j++)
+	{
 		btManifoldPoint& cp = manifold->getContactPoint(j);
 
 		if (cp.getDistance() <= manifold->getContactProcessingThreshold())
 		{
-		
 			btScalar relaxation;
 
 			int frictionIndex = m_multiBodyNormalContactConstraints.size();
 
 			btMultiBodySolverConstraint& solverConstraint = m_multiBodyNormalContactConstraints.expandNonInitializing();
 
-	//		btRigidBody* rb0 = btRigidBody::upcast(colObj0);
-	//		btRigidBody* rb1 = btRigidBody::upcast(colObj1);
-            solverConstraint.m_orgConstraint = 0;
-            solverConstraint.m_orgDofIndex = -1;
+			//		btRigidBody* rb0 = btRigidBody::upcast(colObj0);
+			//		btRigidBody* rb1 = btRigidBody::upcast(colObj1);
+			solverConstraint.m_orgConstraint = 0;
+			solverConstraint.m_orgDofIndex = -1;
 			solverConstraint.m_solverBodyIdA = solverBodyIdA;
 			solverConstraint.m_solverBodyIdB = solverBodyIdB;
 			solverConstraint.m_multiBodyA = mbA;
@@ -1253,60 +1224,59 @@ void	btMultiBodyConstraintSolver::convertMultiBodyContact(btPersistentManifold*
 			solverConstraint.m_originalContactPoint = &cp;
 
 			bool isFriction = false;
-			setupMultiBodyContactConstraint(solverConstraint, cp.m_normalWorldOnB,cp, infoGlobal, relaxation, isFriction);
+			setupMultiBodyContactConstraint(solverConstraint, cp.m_normalWorldOnB, cp, infoGlobal, relaxation, isFriction);
 
-//			const btVector3& pos1 = cp.getPositionWorldOnA();
-//			const btVector3& pos2 = cp.getPositionWorldOnB();
+			//			const btVector3& pos1 = cp.getPositionWorldOnA();
+			//			const btVector3& pos2 = cp.getPositionWorldOnB();
 
 			/////setup the friction constraints
 #define ENABLE_FRICTION
 #ifdef ENABLE_FRICTION
-			solverConstraint.m_frictionIndex = frictionIndex;
+			solverConstraint.m_frictionIndex = m_multiBodyFrictionContactConstraints.size();
 
 			///Bullet has several options to set the friction directions
 			///By default, each contact has only a single friction direction that is recomputed automatically every frame
 			///based on the relative linear velocity.
 			///If the relative velocity is zero, it will automatically compute a friction direction.
-			
+
 			///You can also enable two friction directions, using the SOLVER_USE_2_FRICTION_DIRECTIONS.
 			///In that case, the second friction direction will be orthogonal to both contact normal and first friction direction.
 			///
 			///If you choose SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION, then the friction will be independent from the relative projected velocity.
 			///
-			///The user can manually override the friction directions for certain contacts using a contact callback, 
+			///The user can manually override the friction directions for certain contacts using a contact callback,
 			///and set the cp.m_lateralFrictionInitialized to true
 			///In that case, you can set the target relative motion in each friction direction (cp.m_contactMotion1 and cp.m_contactMotion2)
 			///this will give a conveyor belt effect
 			///
 
-			btPlaneSpace1(cp.m_normalWorldOnB,cp.m_lateralFrictionDir1,cp.m_lateralFrictionDir2);
+			btPlaneSpace1(cp.m_normalWorldOnB, cp.m_lateralFrictionDir1, cp.m_lateralFrictionDir2);
 			cp.m_lateralFrictionDir1.normalize();
 			cp.m_lateralFrictionDir2.normalize();
 
-            if (rollingFriction > 0 )
-             {
-                if (cp.m_combinedSpinningFriction>0)
-                {
-                    addMultiBodyTorsionalFrictionConstraint(cp.m_normalWorldOnB,manifold,frictionIndex,cp,cp.m_combinedSpinningFriction, colObj0,colObj1, relaxation,infoGlobal);
-                }
-                if (cp.m_combinedRollingFriction>0)
-                {
+			if (rollingFriction > 0)
+			{
+				if (cp.m_combinedSpinningFriction > 0)
+				{
+					addMultiBodyTorsionalFrictionConstraint(cp.m_normalWorldOnB, manifold, frictionIndex, cp, cp.m_combinedSpinningFriction, colObj0, colObj1, relaxation, infoGlobal);
+				}
+				if (cp.m_combinedRollingFriction > 0)
+				{
+					applyAnisotropicFriction(colObj0, cp.m_lateralFrictionDir1, btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION);
+					applyAnisotropicFriction(colObj1, cp.m_lateralFrictionDir1, btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION);
+					applyAnisotropicFriction(colObj0, cp.m_lateralFrictionDir2, btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION);
+					applyAnisotropicFriction(colObj1, cp.m_lateralFrictionDir2, btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION);
 
-					applyAnisotropicFriction(colObj0,cp.m_lateralFrictionDir1,btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION);
-					applyAnisotropicFriction(colObj1,cp.m_lateralFrictionDir1,btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION);
-					applyAnisotropicFriction(colObj0,cp.m_lateralFrictionDir2,btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION);
-					applyAnisotropicFriction(colObj1,cp.m_lateralFrictionDir2,btCollisionObject::CF_ANISOTROPIC_ROLLING_FRICTION);
-					
-					if (cp.m_lateralFrictionDir1.length()>0.001)
-	                    addMultiBodyTorsionalFrictionConstraint(cp.m_lateralFrictionDir1,manifold,frictionIndex,cp,cp.m_combinedRollingFriction, colObj0,colObj1, relaxation,infoGlobal);
+					if (cp.m_lateralFrictionDir1.length() > 0.001)
+						addMultiBodyTorsionalFrictionConstraint(cp.m_lateralFrictionDir1, manifold, frictionIndex, cp, cp.m_combinedRollingFriction, colObj0, colObj1, relaxation, infoGlobal);
 
-					if (cp.m_lateralFrictionDir2.length()>0.001)
-                    addMultiBodyTorsionalFrictionConstraint(cp.m_lateralFrictionDir2,manifold,frictionIndex,cp,cp.m_combinedRollingFriction, colObj0,colObj1, relaxation,infoGlobal);
-                }
-              rollingFriction--;
-            }
-			if (!(infoGlobal.m_solverMode & SOLVER_ENABLE_FRICTION_DIRECTION_CACHING) || !(cp.m_contactPointFlags&BT_CONTACT_FLAG_LATERAL_FRICTION_INITIALIZED))
-			{/*
+					if (cp.m_lateralFrictionDir2.length() > 0.001)
+						addMultiBodyTorsionalFrictionConstraint(cp.m_lateralFrictionDir2, manifold, frictionIndex, cp, cp.m_combinedRollingFriction, colObj0, colObj1, relaxation, infoGlobal);
+				}
+				rollingFriction--;
+			}
+			if (!(infoGlobal.m_solverMode & SOLVER_ENABLE_FRICTION_DIRECTION_CACHING) || !(cp.m_contactPointFlags & BT_CONTACT_FLAG_LATERAL_FRICTION_INITIALIZED))
+			{ /*
 				cp.m_lateralFrictionDir1 = vel - cp.m_normalWorldOnB * rel_vel;
 				btScalar lat_rel_vel = cp.m_lateralFrictionDir1.length2();
 				if (!(infoGlobal.m_solverMode & SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION) && lat_rel_vel > SIMD_EPSILON)
@@ -1329,85 +1299,77 @@ void	btMultiBodyConstraintSolver::convertMultiBodyContact(btPersistentManifold*
 				} else
 				*/
 				{
-					
-
-					applyAnisotropicFriction(colObj0,cp.m_lateralFrictionDir1,btCollisionObject::CF_ANISOTROPIC_FRICTION);
-					applyAnisotropicFriction(colObj1,cp.m_lateralFrictionDir1,btCollisionObject::CF_ANISOTROPIC_FRICTION);
-					addMultiBodyFrictionConstraint(cp.m_lateralFrictionDir1,manifold,frictionIndex,cp,colObj0,colObj1, relaxation,infoGlobal);
-                    
+					applyAnisotropicFriction(colObj0, cp.m_lateralFrictionDir1, btCollisionObject::CF_ANISOTROPIC_FRICTION);
+					applyAnisotropicFriction(colObj1, cp.m_lateralFrictionDir1, btCollisionObject::CF_ANISOTROPIC_FRICTION);
+					addMultiBodyFrictionConstraint(cp.m_lateralFrictionDir1, manifold, frictionIndex, cp, colObj0, colObj1, relaxation, infoGlobal);
 
 					if ((infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS))
 					{
-						applyAnisotropicFriction(colObj0,cp.m_lateralFrictionDir2,btCollisionObject::CF_ANISOTROPIC_FRICTION);
-						applyAnisotropicFriction(colObj1,cp.m_lateralFrictionDir2,btCollisionObject::CF_ANISOTROPIC_FRICTION);
-						addMultiBodyFrictionConstraint(cp.m_lateralFrictionDir2,manifold,frictionIndex,cp,colObj0,colObj1, relaxation,infoGlobal);
+						applyAnisotropicFriction(colObj0, cp.m_lateralFrictionDir2, btCollisionObject::CF_ANISOTROPIC_FRICTION);
+						applyAnisotropicFriction(colObj1, cp.m_lateralFrictionDir2, btCollisionObject::CF_ANISOTROPIC_FRICTION);
+						addMultiBodyFrictionConstraint(cp.m_lateralFrictionDir2, manifold, frictionIndex, cp, colObj0, colObj1, relaxation, infoGlobal);
 					}
 
 					if ((infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS) && (infoGlobal.m_solverMode & SOLVER_DISABLE_VELOCITY_DEPENDENT_FRICTION_DIRECTION))
 					{
-						cp.m_contactPointFlags|=BT_CONTACT_FLAG_LATERAL_FRICTION_INITIALIZED;
+						cp.m_contactPointFlags |= BT_CONTACT_FLAG_LATERAL_FRICTION_INITIALIZED;
 					}
 				}
-
-			} else
+			}
+			else
 			{
-				addMultiBodyFrictionConstraint(cp.m_lateralFrictionDir1,manifold,frictionIndex,cp,colObj0,colObj1, relaxation,infoGlobal,cp.m_contactMotion1, cp.m_frictionCFM);
+				addMultiBodyFrictionConstraint(cp.m_lateralFrictionDir1, manifold, frictionIndex, cp, colObj0, colObj1, relaxation, infoGlobal, cp.m_contactMotion1, cp.m_frictionCFM);
 
 				if ((infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS))
-					addMultiBodyFrictionConstraint(cp.m_lateralFrictionDir2,manifold,frictionIndex,cp,colObj0,colObj1, relaxation, infoGlobal,cp.m_contactMotion2, cp.m_frictionCFM);
+					addMultiBodyFrictionConstraint(cp.m_lateralFrictionDir2, manifold, frictionIndex, cp, colObj0, colObj1, relaxation, infoGlobal, cp.m_contactMotion2, cp.m_frictionCFM);
 
 				//setMultiBodyFrictionConstraintImpulse( solverConstraint, solverBodyIdA, solverBodyIdB, cp, infoGlobal);
 				//todo:
 				solverConstraint.m_appliedImpulse = 0.f;
 				solverConstraint.m_appliedPushImpulse = 0.f;
 			}
-		
-
-#endif //ENABLE_FRICTION
 
+#endif  //ENABLE_FRICTION
 		}
 	}
 }
 
-void btMultiBodyConstraintSolver::convertContacts(btPersistentManifold** manifoldPtr,int numManifolds, const btContactSolverInfo& infoGlobal)
+void btMultiBodyConstraintSolver::convertContacts(btPersistentManifold** manifoldPtr, int numManifolds, const btContactSolverInfo& infoGlobal)
 {
 	//btPersistentManifold* manifold = 0;
 
-	for (int i=0;i<numManifolds;i++)
+	for (int i = 0; i < numManifolds; i++)
 	{
-		btPersistentManifold* manifold= manifoldPtr[i];
+		btPersistentManifold* manifold = manifoldPtr[i];
 		const btMultiBodyLinkCollider* fcA = btMultiBodyLinkCollider::upcast(manifold->getBody0());
 		const btMultiBodyLinkCollider* fcB = btMultiBodyLinkCollider::upcast(manifold->getBody1());
 		if (!fcA && !fcB)
 		{
 			//the contact doesn't involve any Featherstone btMultiBody, so deal with the regular btRigidBody/btCollisionObject case
-			convertContact(manifold,infoGlobal);
-		} else
+			convertContact(manifold, infoGlobal);
+		}
+		else
 		{
-			convertMultiBodyContact(manifold,infoGlobal);
+			convertMultiBodyContact(manifold, infoGlobal);
 		}
 	}
 
 	//also convert the multibody constraints, if any
 
-	
-	for (int i=0;i<m_tmpNumMultiBodyConstraints;i++)
+	for (int i = 0; i < m_tmpNumMultiBodyConstraints; i++)
 	{
 		btMultiBodyConstraint* c = m_tmpMultiBodyConstraints[i];
 		m_data.m_solverBodyPool = &m_tmpSolverBodyPool;
 		m_data.m_fixedBodyId = m_fixedBodyId;
-		
-		c->createConstraintRows(m_multiBodyNonContactConstraints,m_data,	infoGlobal);
-	}
 
+		c->createConstraintRows(m_multiBodyNonContactConstraints, m_data, infoGlobal);
+	}
 }
 
-
-
-btScalar btMultiBodyConstraintSolver::solveGroup(btCollisionObject** bodies,int numBodies,btPersistentManifold** manifold,int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& info, btIDebugDraw* debugDrawer,btDispatcher* dispatcher)
+btScalar btMultiBodyConstraintSolver::solveGroup(btCollisionObject** bodies, int numBodies, btPersistentManifold** manifold, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& info, btIDebugDraw* debugDrawer, btDispatcher* dispatcher)
 {
 	//printf("btMultiBodyConstraintSolver::solveGroup: numBodies=%d, numConstraints=%d\n", numBodies, numConstraints);
-	return btSequentialImpulseConstraintSolver::solveGroup(bodies,numBodies,manifold,numManifolds,constraints,numConstraints,info,debugDrawer,dispatcher);
+	return btSequentialImpulseConstraintSolver::solveGroup(bodies, numBodies, manifold, numManifolds, constraints, numConstraints, info, debugDrawer, dispatcher);
 }
 
 #if 0
@@ -1431,56 +1393,54 @@ static void applyJointFeedback(btMultiBodyJacobianData& data, const btMultiBodyS
 }
 #endif
 
-
 void btMultiBodyConstraintSolver::writeBackSolverBodyToMultiBody(btMultiBodySolverConstraint& c, btScalar deltaTime)
 {
-#if 1 
-	
+#if 1
+
 	//bod->addBaseForce(m_gravity * bod->getBaseMass());
 	//bod->addLinkForce(j, m_gravity * bod->getLinkMass(j));
 
 	if (c.m_orgConstraint)
 	{
-		c.m_orgConstraint->internalSetAppliedImpulse(c.m_orgDofIndex,c.m_appliedImpulse);
+		c.m_orgConstraint->internalSetAppliedImpulse(c.m_orgDofIndex, c.m_appliedImpulse);
 	}
-	
 
 	if (c.m_multiBodyA)
 	{
-		
 		c.m_multiBodyA->setCompanionId(-1);
-		btVector3 force = c.m_contactNormal1*(c.m_appliedImpulse/deltaTime);
-		btVector3 torque = c.m_relpos1CrossNormal*(c.m_appliedImpulse/deltaTime);
-		if (c.m_linkA<0)
+		btVector3 force = c.m_contactNormal1 * (c.m_appliedImpulse / deltaTime);
+		btVector3 torque = c.m_relpos1CrossNormal * (c.m_appliedImpulse / deltaTime);
+		if (c.m_linkA < 0)
 		{
 			c.m_multiBodyA->addBaseConstraintForce(force);
 			c.m_multiBodyA->addBaseConstraintTorque(torque);
-		} else
+		}
+		else
 		{
-			c.m_multiBodyA->addLinkConstraintForce(c.m_linkA,force);
-				//b3Printf("force = %f,%f,%f\n",force[0],force[1],force[2]);//[0],torque[1],torque[2]);
-			c.m_multiBodyA->addLinkConstraintTorque(c.m_linkA,torque);
+			c.m_multiBodyA->addLinkConstraintForce(c.m_linkA, force);
+			//b3Printf("force = %f,%f,%f\n",force[0],force[1],force[2]);//[0],torque[1],torque[2]);
+			c.m_multiBodyA->addLinkConstraintTorque(c.m_linkA, torque);
 		}
 	}
-	
+
 	if (c.m_multiBodyB)
 	{
 		{
 			c.m_multiBodyB->setCompanionId(-1);
-			btVector3 force = c.m_contactNormal2*(c.m_appliedImpulse/deltaTime);
-			btVector3 torque = c.m_relpos2CrossNormal*(c.m_appliedImpulse/deltaTime);
-			if (c.m_linkB<0)
+			btVector3 force = c.m_contactNormal2 * (c.m_appliedImpulse / deltaTime);
+			btVector3 torque = c.m_relpos2CrossNormal * (c.m_appliedImpulse / deltaTime);
+			if (c.m_linkB < 0)
 			{
 				c.m_multiBodyB->addBaseConstraintForce(force);
 				c.m_multiBodyB->addBaseConstraintTorque(torque);
-			} else
+			}
+			else
 			{
 				{
-					c.m_multiBodyB->addLinkConstraintForce(c.m_linkB,force);
+					c.m_multiBodyB->addLinkConstraintForce(c.m_linkB, force);
 					//b3Printf("t = %f,%f,%f\n",force[0],force[1],force[2]);//[0],torque[1],torque[2]);
-					c.m_multiBodyB->addLinkConstraintTorque(c.m_linkB,torque);
+					c.m_multiBodyB->addLinkConstraintTorque(c.m_linkB, torque);
 				}
-					
 			}
 		}
 	}
@@ -1490,66 +1450,64 @@ void btMultiBodyConstraintSolver::writeBackSolverBodyToMultiBody(btMultiBodySolv
 
 	if (c.m_multiBodyA)
 	{
-		c.m_multiBodyA->applyDeltaVeeMultiDof(&m_data.m_deltaVelocitiesUnitImpulse[c.m_jacAindex],c.m_appliedImpulse);
+		c.m_multiBodyA->applyDeltaVeeMultiDof(&m_data.m_deltaVelocitiesUnitImpulse[c.m_jacAindex], c.m_appliedImpulse);
 	}
-	
+
 	if (c.m_multiBodyB)
 	{
-		c.m_multiBodyB->applyDeltaVeeMultiDof(&m_data.m_deltaVelocitiesUnitImpulse[c.m_jacBindex],c.m_appliedImpulse);
+		c.m_multiBodyB->applyDeltaVeeMultiDof(&m_data.m_deltaVelocitiesUnitImpulse[c.m_jacBindex], c.m_appliedImpulse);
 	}
 #endif
-
-
-
 }
 
-btScalar btMultiBodyConstraintSolver::solveGroupCacheFriendlyFinish(btCollisionObject** bodies,int numBodies,const btContactSolverInfo& infoGlobal)
+btScalar btMultiBodyConstraintSolver::solveGroupCacheFriendlyFinish(btCollisionObject** bodies, int numBodies, const btContactSolverInfo& infoGlobal)
 {
 	BT_PROFILE("btMultiBodyConstraintSolver::solveGroupCacheFriendlyFinish");
 	int numPoolConstraints = m_multiBodyNormalContactConstraints.size();
-	
 
-	//write back the delta v to the multi bodies, either as applied impulse (direct velocity change) 
+	//write back the delta v to the multi bodies, either as applied impulse (direct velocity change)
 	//or as applied force, so we can measure the joint reaction forces easier
-	for (int i=0;i<numPoolConstraints;i++)
+	for (int i = 0; i < numPoolConstraints; i++)
 	{
 		btMultiBodySolverConstraint& solverConstraint = m_multiBodyNormalContactConstraints[i];
-		writeBackSolverBodyToMultiBody(solverConstraint,infoGlobal.m_timeStep);
+		writeBackSolverBodyToMultiBody(solverConstraint, infoGlobal.m_timeStep);
 
-		writeBackSolverBodyToMultiBody(m_multiBodyFrictionContactConstraints[solverConstraint.m_frictionIndex],infoGlobal.m_timeStep);
+		writeBackSolverBodyToMultiBody(m_multiBodyFrictionContactConstraints[solverConstraint.m_frictionIndex], infoGlobal.m_timeStep);
 
 		if ((infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS))
 		{
-			writeBackSolverBodyToMultiBody(m_multiBodyFrictionContactConstraints[solverConstraint.m_frictionIndex+1],infoGlobal.m_timeStep);
+			writeBackSolverBodyToMultiBody(m_multiBodyFrictionContactConstraints[solverConstraint.m_frictionIndex + 1], infoGlobal.m_timeStep);
 		}
 	}
 
-
-	for (int i=0;i<m_multiBodyNonContactConstraints.size();i++)
+	for (int i = 0; i < m_multiBodyNonContactConstraints.size(); i++)
 	{
 		btMultiBodySolverConstraint& solverConstraint = m_multiBodyNonContactConstraints[i];
-		writeBackSolverBodyToMultiBody(solverConstraint,infoGlobal.m_timeStep);
+		writeBackSolverBodyToMultiBody(solverConstraint, infoGlobal.m_timeStep);
 	}
 
-	
 	if (infoGlobal.m_solverMode & SOLVER_USE_WARMSTARTING)
 	{
 		BT_PROFILE("warm starting write back");
-		for (int j=0;j<numPoolConstraints;j++)
+		for (int j = 0; j < numPoolConstraints; j++)
 		{
 			const btMultiBodySolverConstraint& solverConstraint = m_multiBodyNormalContactConstraints[j];
-			btManifoldPoint* pt = (btManifoldPoint*) solverConstraint.m_originalContactPoint;
+			btManifoldPoint* pt = (btManifoldPoint*)solverConstraint.m_originalContactPoint;
 			btAssert(pt);
 			pt->m_appliedImpulse = solverConstraint.m_appliedImpulse;
 			pt->m_appliedImpulseLateral1 = m_multiBodyFrictionContactConstraints[solverConstraint.m_frictionIndex].m_appliedImpulse;
-			
+
 			//printf("pt->m_appliedImpulseLateral1 = %f\n", pt->m_appliedImpulseLateral1);
 			if ((infoGlobal.m_solverMode & SOLVER_USE_2_FRICTION_DIRECTIONS))
 			{
-				pt->m_appliedImpulseLateral2 = m_multiBodyFrictionContactConstraints[solverConstraint.m_frictionIndex+1].m_appliedImpulse;
+				pt->m_appliedImpulseLateral2 = m_multiBodyFrictionContactConstraints[solverConstraint.m_frictionIndex + 1].m_appliedImpulse;
+			} else
+			{
+				pt->m_appliedImpulseLateral2 = 0;
 			}
-			//do a callback here?
 		}
+		
+			//do a callback here?
 	}
 #if 0
 	//multibody joint feedback
@@ -1648,25 +1606,22 @@ btScalar btMultiBodyConstraintSolver::solveGroupCacheFriendlyFinish(btCollisionO
 		}
 
 	}
-#endif 
+#endif
 #endif
 
-	return btSequentialImpulseConstraintSolver::solveGroupCacheFriendlyFinish(bodies,numBodies,infoGlobal);
+	return btSequentialImpulseConstraintSolver::solveGroupCacheFriendlyFinish(bodies, numBodies, infoGlobal);
 }
 
-
-void  btMultiBodyConstraintSolver::solveMultiBodyGroup(btCollisionObject** bodies,int numBodies,btPersistentManifold** manifold,int numManifolds,btTypedConstraint** constraints,int numConstraints,btMultiBodyConstraint** multiBodyConstraints, int numMultiBodyConstraints, const btContactSolverInfo& info, btIDebugDraw* debugDrawer,btDispatcher* dispatcher)
+void btMultiBodyConstraintSolver::solveMultiBodyGroup(btCollisionObject** bodies, int numBodies, btPersistentManifold** manifold, int numManifolds, btTypedConstraint** constraints, int numConstraints, btMultiBodyConstraint** multiBodyConstraints, int numMultiBodyConstraints, const btContactSolverInfo& info, btIDebugDraw* debugDrawer, btDispatcher* dispatcher)
 {
 	//printf("solveMultiBodyGroup: numBodies=%d, numConstraints=%d, numManifolds=%d, numMultiBodyConstraints=%d\n", numBodies, numConstraints, numManifolds, numMultiBodyConstraints);
 
 	//printf("solveMultiBodyGroup start\n");
 	m_tmpMultiBodyConstraints = multiBodyConstraints;
 	m_tmpNumMultiBodyConstraints = numMultiBodyConstraints;
-	
-	btSequentialImpulseConstraintSolver::solveGroup(bodies,numBodies,manifold,numManifolds,constraints,numConstraints,info,debugDrawer,dispatcher);
+
+	btSequentialImpulseConstraintSolver::solveGroup(bodies, numBodies, manifold, numManifolds, constraints, numConstraints, info, debugDrawer, dispatcher);
 
 	m_tmpMultiBodyConstraints = 0;
 	m_tmpNumMultiBodyConstraints = 0;
-	
-
 }
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyConstraintSolver.h b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyConstraintSolver.h
index 29f484e1d8d..f39f2879fc7 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyConstraintSolver.h
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyConstraintSolver.h
@@ -25,80 +25,71 @@ class btMultiBody;
 
 #include "btMultiBodyConstraint.h"
 
-
-
-ATTRIBUTE_ALIGNED16(class) btMultiBodyConstraintSolver : public btSequentialImpulseConstraintSolver
+ATTRIBUTE_ALIGNED16(class)
+btMultiBodyConstraintSolver : public btSequentialImpulseConstraintSolver
 {
-
 protected:
+	btMultiBodyConstraintArray m_multiBodyNonContactConstraints;
 
-	btMultiBodyConstraintArray			m_multiBodyNonContactConstraints;
+	btMultiBodyConstraintArray m_multiBodyNormalContactConstraints;
+	btMultiBodyConstraintArray m_multiBodyFrictionContactConstraints;
+	btMultiBodyConstraintArray m_multiBodyTorsionalFrictionContactConstraints;
 
-	btMultiBodyConstraintArray			m_multiBodyNormalContactConstraints;
-	btMultiBodyConstraintArray			m_multiBodyFrictionContactConstraints;
-	btMultiBodyConstraintArray			m_multiBodyTorsionalFrictionContactConstraints;
+	btMultiBodyJacobianData m_data;
 
-	btMultiBodyJacobianData				m_data;
-	
 	//temp storage for multi body constraints for a specific island/group called by 'solveGroup'
-	btMultiBodyConstraint**					m_tmpMultiBodyConstraints;
-	int										m_tmpNumMultiBodyConstraints;
+	btMultiBodyConstraint** m_tmpMultiBodyConstraints;
+	int m_tmpNumMultiBodyConstraints;
 
 	btScalar resolveSingleConstraintRowGeneric(const btMultiBodySolverConstraint& c);
-	
+
 	//solve 2 friction directions and clamp against the implicit friction cone
 	btScalar resolveConeFrictionConstraintRows(const btMultiBodySolverConstraint& cA1, const btMultiBodySolverConstraint& cB);
-	
 
-	void convertContacts(btPersistentManifold** manifoldPtr,int numManifolds, const btContactSolverInfo& infoGlobal);
-    
-    btMultiBodySolverConstraint&	addMultiBodyFrictionConstraint(const btVector3& normalAxis,btPersistentManifold* manifold,int frictionIndex,btManifoldPoint& cp,btCollisionObject* colObj0,btCollisionObject* colObj1, btScalar relaxation, const btContactSolverInfo& infoGlobal, btScalar desiredVelocity=0, btScalar cfmSlip=0);
+	void convertContacts(btPersistentManifold * *manifoldPtr, int numManifolds, const btContactSolverInfo& infoGlobal);
 
-    btMultiBodySolverConstraint&	addMultiBodyTorsionalFrictionConstraint(const btVector3& normalAxis,btPersistentManifold* manifold,int frictionIndex,btManifoldPoint& cp,
-                                                            btScalar combinedTorsionalFriction,
-                                                            btCollisionObject* colObj0,btCollisionObject* colObj1, btScalar relaxation, const btContactSolverInfo& infoGlobal, btScalar desiredVelocity=0, btScalar cfmSlip=0);
+	btMultiBodySolverConstraint& addMultiBodyFrictionConstraint(const btVector3& normalAxis, btPersistentManifold* manifold, int frictionIndex, btManifoldPoint& cp, btCollisionObject* colObj0, btCollisionObject* colObj1, btScalar relaxation, const btContactSolverInfo& infoGlobal, btScalar desiredVelocity = 0, btScalar cfmSlip = 0);
 
-	void setupMultiBodyJointLimitConstraint(btMultiBodySolverConstraint& constraintRow, 
-																 btScalar* jacA,btScalar* jacB,
-																 btScalar penetration,btScalar combinedFrictionCoeff, btScalar combinedRestitutionCoeff,
-																 const btContactSolverInfo& infoGlobal);
+	btMultiBodySolverConstraint& addMultiBodyTorsionalFrictionConstraint(const btVector3& normalAxis, btPersistentManifold* manifold, int frictionIndex, btManifoldPoint& cp,
+																		 btScalar combinedTorsionalFriction,
+																		 btCollisionObject* colObj0, btCollisionObject* colObj1, btScalar relaxation, const btContactSolverInfo& infoGlobal, btScalar desiredVelocity = 0, btScalar cfmSlip = 0);
 
-	void setupMultiBodyContactConstraint(btMultiBodySolverConstraint& solverConstraint, 
-																 const btVector3& contactNormal,
-																 btManifoldPoint& cp, const btContactSolverInfo& infoGlobal,
-																 btScalar& relaxation,
-																 bool isFriction, btScalar desiredVelocity=0, btScalar cfmSlip=0);
-    
-    //either rolling or spinning friction
-    void setupMultiBodyTorsionalFrictionConstraint(btMultiBodySolverConstraint& solverConstraint,
-                                         const btVector3& contactNormal,
-                                         btManifoldPoint& cp,
-                                        btScalar combinedTorsionalFriction,
-                                        const btContactSolverInfo& infoGlobal,
-                                         btScalar& relaxation,
-                                         bool isFriction, btScalar desiredVelocity=0, btScalar cfmSlip=0);
+	void setupMultiBodyJointLimitConstraint(btMultiBodySolverConstraint & constraintRow,
+											btScalar * jacA, btScalar * jacB,
+											btScalar penetration, btScalar combinedFrictionCoeff, btScalar combinedRestitutionCoeff,
+											const btContactSolverInfo& infoGlobal);
 
-	void convertMultiBodyContact(btPersistentManifold* manifold,const btContactSolverInfo& infoGlobal);
-	virtual btScalar solveGroupCacheFriendlySetup(btCollisionObject** bodies,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer);
-//	virtual btScalar solveGroupCacheFriendlyIterations(btCollisionObject** bodies,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer);
+	void setupMultiBodyContactConstraint(btMultiBodySolverConstraint & solverConstraint,
+										 const btVector3& contactNormal,
+										 btManifoldPoint& cp, const btContactSolverInfo& infoGlobal,
+										 btScalar& relaxation,
+										 bool isFriction, btScalar desiredVelocity = 0, btScalar cfmSlip = 0);
+
+	//either rolling or spinning friction
+	void setupMultiBodyTorsionalFrictionConstraint(btMultiBodySolverConstraint & solverConstraint,
+												   const btVector3& contactNormal,
+												   btManifoldPoint& cp,
+												   btScalar combinedTorsionalFriction,
+												   const btContactSolverInfo& infoGlobal,
+												   btScalar& relaxation,
+												   bool isFriction, btScalar desiredVelocity = 0, btScalar cfmSlip = 0);
+
+	void convertMultiBodyContact(btPersistentManifold * manifold, const btContactSolverInfo& infoGlobal);
+	virtual btScalar solveGroupCacheFriendlySetup(btCollisionObject * *bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer);
+	//	virtual btScalar solveGroupCacheFriendlyIterations(btCollisionObject** bodies,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer);
+
+	virtual btScalar solveSingleIteration(int iteration, btCollisionObject** bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer);
+	void applyDeltaVee(btScalar * deltaV, btScalar impulse, int velocityIndex, int ndof);
+	void writeBackSolverBodyToMultiBody(btMultiBodySolverConstraint & constraint, btScalar deltaTime);
 
-	virtual btScalar solveSingleIteration(int iteration, btCollisionObject** bodies ,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer);
-	void	applyDeltaVee(btScalar* deltaV, btScalar impulse, int velocityIndex, int ndof);
-	void writeBackSolverBodyToMultiBody(btMultiBodySolverConstraint& constraint, btScalar deltaTime);
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
 	///this method should not be called, it was just used during porting/integration of Featherstone btMultiBody, providing backwards compatibility but no support for btMultiBodyConstraint (only contact constraints)
-	virtual btScalar solveGroup(btCollisionObject** bodies,int numBodies,btPersistentManifold** manifold,int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& info, btIDebugDraw* debugDrawer,btDispatcher* dispatcher);
-	virtual btScalar solveGroupCacheFriendlyFinish(btCollisionObject** bodies,int numBodies,const btContactSolverInfo& infoGlobal);
-	
-	virtual void solveMultiBodyGroup(btCollisionObject** bodies,int numBodies,btPersistentManifold** manifold,int numManifolds,btTypedConstraint** constraints,int numConstraints,btMultiBodyConstraint** multiBodyConstraints, int numMultiBodyConstraints, const btContactSolverInfo& info, btIDebugDraw* debugDrawer,btDispatcher* dispatcher);
+	virtual btScalar solveGroup(btCollisionObject * *bodies, int numBodies, btPersistentManifold** manifold, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& info, btIDebugDraw* debugDrawer, btDispatcher* dispatcher);
+	virtual btScalar solveGroupCacheFriendlyFinish(btCollisionObject * *bodies, int numBodies, const btContactSolverInfo& infoGlobal);
+
+	virtual void solveMultiBodyGroup(btCollisionObject * *bodies, int numBodies, btPersistentManifold** manifold, int numManifolds, btTypedConstraint** constraints, int numConstraints, btMultiBodyConstraint** multiBodyConstraints, int numMultiBodyConstraints, const btContactSolverInfo& info, btIDebugDraw* debugDrawer, btDispatcher* dispatcher);
 };
 
-	
-	
-
-
-#endif //BT_MULTIBODY_CONSTRAINT_SOLVER_H
-
+#endif  //BT_MULTIBODY_CONSTRAINT_SOLVER_H
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyDynamicsWorld.cpp b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyDynamicsWorld.cpp
index 9c5f3ad8a97..1557987bc34 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyDynamicsWorld.cpp
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyDynamicsWorld.cpp
@@ -23,45 +23,43 @@ subject to the following restrictions:
 #include "LinearMath/btIDebugDraw.h"
 #include "LinearMath/btSerializer.h"
 
-
-void	btMultiBodyDynamicsWorld::addMultiBody(btMultiBody* body, int group, int mask)
+void btMultiBodyDynamicsWorld::addMultiBody(btMultiBody* body, int group, int mask)
 {
 	m_multiBodies.push_back(body);
-
 }
 
-void	btMultiBodyDynamicsWorld::removeMultiBody(btMultiBody* body)
+void btMultiBodyDynamicsWorld::removeMultiBody(btMultiBody* body)
 {
 	m_multiBodies.remove(body);
 }
 
-void	btMultiBodyDynamicsWorld::calculateSimulationIslands()
+void btMultiBodyDynamicsWorld::calculateSimulationIslands()
 {
 	BT_PROFILE("calculateSimulationIslands");
 
-	getSimulationIslandManager()->updateActivationState(getCollisionWorld(),getCollisionWorld()->getDispatcher());
+	getSimulationIslandManager()->updateActivationState(getCollisionWorld(), getCollisionWorld()->getDispatcher());
+
+	{
+		//merge islands based on speculative contact manifolds too
+		for (int i = 0; i < this->m_predictiveManifolds.size(); i++)
+		{
+			btPersistentManifold* manifold = m_predictiveManifolds[i];
+
+			const btCollisionObject* colObj0 = manifold->getBody0();
+			const btCollisionObject* colObj1 = manifold->getBody1();
+
+			if (((colObj0) && (!(colObj0)->isStaticOrKinematicObject())) &&
+				((colObj1) && (!(colObj1)->isStaticOrKinematicObject())))
+			{
+				getSimulationIslandManager()->getUnionFind().unite((colObj0)->getIslandTag(), (colObj1)->getIslandTag());
+			}
+		}
+	}
 
-    {
-        //merge islands based on speculative contact manifolds too
-        for (int i=0;i<this->m_predictiveManifolds.size();i++)
-        {
-            btPersistentManifold* manifold = m_predictiveManifolds[i];
-            
-            const btCollisionObject* colObj0 = manifold->getBody0();
-            const btCollisionObject* colObj1 = manifold->getBody1();
-            
-            if (((colObj0) && (!(colObj0)->isStaticOrKinematicObject())) &&
-                ((colObj1) && (!(colObj1)->isStaticOrKinematicObject())))
-            {
-				getSimulationIslandManager()->getUnionFind().unite((colObj0)->getIslandTag(),(colObj1)->getIslandTag());
-            }
-        }
-    }
-    
 	{
 		int i;
 		int numConstraints = int(m_constraints.size());
-		for (i=0;i< numConstraints ; i++ )
+		for (i = 0; i < numConstraints; i++)
 		{
 			btTypedConstraint* constraint = m_constraints[i];
 			if (constraint->isEnabled())
@@ -72,23 +70,23 @@ void	btMultiBodyDynamicsWorld::calculateSimulationIslands()
 				if (((colObj0) && (!(colObj0)->isStaticOrKinematicObject())) &&
 					((colObj1) && (!(colObj1)->isStaticOrKinematicObject())))
 				{
-					getSimulationIslandManager()->getUnionFind().unite((colObj0)->getIslandTag(),(colObj1)->getIslandTag());
+					getSimulationIslandManager()->getUnionFind().unite((colObj0)->getIslandTag(), (colObj1)->getIslandTag());
 				}
 			}
 		}
 	}
 
 	//merge islands linked by Featherstone link colliders
-	for (int i=0;i<m_multiBodies.size();i++)
+	for (int i = 0; i < m_multiBodies.size(); i++)
 	{
 		btMultiBody* body = m_multiBodies[i];
 		{
 			btMultiBodyLinkCollider* prev = body->getBaseCollider();
 
-			for (int b=0;b<body->getNumLinks();b++)
+			for (int b = 0; b < body->getNumLinks(); b++)
 			{
 				btMultiBodyLinkCollider* cur = body->getLink(b).m_collider;
-				
+
 				if (((cur) && (!(cur)->isStaticOrKinematicObject())) &&
 					((prev) && (!(prev)->isStaticOrKinematicObject())))
 				{
@@ -98,36 +96,31 @@ void	btMultiBodyDynamicsWorld::calculateSimulationIslands()
 				}
 				if (cur && !cur->isStaticOrKinematicObject())
 					prev = cur;
-				
 			}
 		}
 	}
 
 	//merge islands linked by multibody constraints
 	{
-		for (int i=0;i<this->m_multiBodyConstraints.size();i++)
+		for (int i = 0; i < this->m_multiBodyConstraints.size(); i++)
 		{
 			btMultiBodyConstraint* c = m_multiBodyConstraints[i];
 			int tagA = c->getIslandIdA();
 			int tagB = c->getIslandIdB();
-			if (tagA>=0 && tagB>=0)
+			if (tagA >= 0 && tagB >= 0)
 				getSimulationIslandManager()->getUnionFind().unite(tagA, tagB);
 		}
 	}
 
 	//Store the island id in each body
 	getSimulationIslandManager()->storeIslandActivationState(getCollisionWorld());
-
 }
 
-
-void	btMultiBodyDynamicsWorld::updateActivationState(btScalar timeStep)
+void btMultiBodyDynamicsWorld::updateActivationState(btScalar timeStep)
 {
 	BT_PROFILE("btMultiBodyDynamicsWorld::updateActivationState");
 
-	
-	
-	for ( int i=0;i<m_multiBodies.size();i++)
+	for (int i = 0; i < m_multiBodies.size(); i++)
 	{
 		btMultiBody* body = m_multiBodies[i];
 		if (body)
@@ -138,119 +131,108 @@ void	btMultiBodyDynamicsWorld::updateActivationState(btScalar timeStep)
 				btMultiBodyLinkCollider* col = body->getBaseCollider();
 				if (col && col->getActivationState() == ACTIVE_TAG)
 				{
-					col->setActivationState( WANTS_DEACTIVATION);
+					col->setActivationState(WANTS_DEACTIVATION);
 					col->setDeactivationTime(0.f);
 				}
-				for (int b=0;b<body->getNumLinks();b++)
+				for (int b = 0; b < body->getNumLinks(); b++)
 				{
 					btMultiBodyLinkCollider* col = body->getLink(b).m_collider;
 					if (col && col->getActivationState() == ACTIVE_TAG)
 					{
-						col->setActivationState( WANTS_DEACTIVATION);
+						col->setActivationState(WANTS_DEACTIVATION);
 						col->setDeactivationTime(0.f);
 					}
 				}
-			} else
+			}
+			else
 			{
 				btMultiBodyLinkCollider* col = body->getBaseCollider();
 				if (col && col->getActivationState() != DISABLE_DEACTIVATION)
-					col->setActivationState( ACTIVE_TAG );
+					col->setActivationState(ACTIVE_TAG);
 
-				for (int b=0;b<body->getNumLinks();b++)
+				for (int b = 0; b < body->getNumLinks(); b++)
 				{
 					btMultiBodyLinkCollider* col = body->getLink(b).m_collider;
 					if (col && col->getActivationState() != DISABLE_DEACTIVATION)
-						col->setActivationState( ACTIVE_TAG );
+						col->setActivationState(ACTIVE_TAG);
 				}
 			}
-
 		}
 	}
 
 	btDiscreteDynamicsWorld::updateActivationState(timeStep);
 }
 
-
-SIMD_FORCE_INLINE	int	btGetConstraintIslandId2(const btTypedConstraint* lhs)
+SIMD_FORCE_INLINE int btGetConstraintIslandId2(const btTypedConstraint* lhs)
 {
 	int islandId;
-	
+
 	const btCollisionObject& rcolObj0 = lhs->getRigidBodyA();
 	const btCollisionObject& rcolObj1 = lhs->getRigidBodyB();
-	islandId= rcolObj0.getIslandTag()>=0?rcolObj0.getIslandTag():rcolObj1.getIslandTag();
+	islandId = rcolObj0.getIslandTag() >= 0 ? rcolObj0.getIslandTag() : rcolObj1.getIslandTag();
 	return islandId;
-
 }
 
-
 class btSortConstraintOnIslandPredicate2
 {
-	public:
-
-		bool operator() ( const btTypedConstraint* lhs, const btTypedConstraint* rhs ) const
-		{
-			int rIslandId0,lIslandId0;
-			rIslandId0 = btGetConstraintIslandId2(rhs);
-			lIslandId0 = btGetConstraintIslandId2(lhs);
-			return lIslandId0 < rIslandId0;
-		}
+public:
+	bool operator()(const btTypedConstraint* lhs, const btTypedConstraint* rhs) const
+	{
+		int rIslandId0, lIslandId0;
+		rIslandId0 = btGetConstraintIslandId2(rhs);
+		lIslandId0 = btGetConstraintIslandId2(lhs);
+		return lIslandId0 < rIslandId0;
+	}
 };
 
-
-
-SIMD_FORCE_INLINE	int	btGetMultiBodyConstraintIslandId(const btMultiBodyConstraint* lhs)
+SIMD_FORCE_INLINE int btGetMultiBodyConstraintIslandId(const btMultiBodyConstraint* lhs)
 {
 	int islandId;
-	
+
 	int islandTagA = lhs->getIslandIdA();
 	int islandTagB = lhs->getIslandIdB();
-	islandId= islandTagA>=0?islandTagA:islandTagB;
+	islandId = islandTagA >= 0 ? islandTagA : islandTagB;
 	return islandId;
-
 }
 
-
 class btSortMultiBodyConstraintOnIslandPredicate
 {
-	public:
-
-		bool operator() ( const btMultiBodyConstraint* lhs, const btMultiBodyConstraint* rhs ) const
-		{
-			int rIslandId0,lIslandId0;
-			rIslandId0 = btGetMultiBodyConstraintIslandId(rhs);
-			lIslandId0 = btGetMultiBodyConstraintIslandId(lhs);
-			return lIslandId0 < rIslandId0;
-		}
+public:
+	bool operator()(const btMultiBodyConstraint* lhs, const btMultiBodyConstraint* rhs) const
+	{
+		int rIslandId0, lIslandId0;
+		rIslandId0 = btGetMultiBodyConstraintIslandId(rhs);
+		lIslandId0 = btGetMultiBodyConstraintIslandId(lhs);
+		return lIslandId0 < rIslandId0;
+	}
 };
 
 struct MultiBodyInplaceSolverIslandCallback : public btSimulationIslandManager::IslandCallback
 {
-	btContactSolverInfo*	m_solverInfo;
-	btMultiBodyConstraintSolver*		m_solver;
-	btMultiBodyConstraint**		m_multiBodySortedConstraints;
-	int							m_numMultiBodyConstraints;
-	
-	btTypedConstraint**		m_sortedConstraints;
-	int						m_numConstraints;
-	btIDebugDraw*			m_debugDrawer;
-	btDispatcher*			m_dispatcher;
-	
+	btContactSolverInfo* m_solverInfo;
+	btMultiBodyConstraintSolver* m_solver;
+	btMultiBodyConstraint** m_multiBodySortedConstraints;
+	int m_numMultiBodyConstraints;
+
+	btTypedConstraint** m_sortedConstraints;
+	int m_numConstraints;
+	btIDebugDraw* m_debugDrawer;
+	btDispatcher* m_dispatcher;
+
 	btAlignedObjectArray<btCollisionObject*> m_bodies;
 	btAlignedObjectArray<btPersistentManifold*> m_manifolds;
 	btAlignedObjectArray<btTypedConstraint*> m_constraints;
 	btAlignedObjectArray<btMultiBodyConstraint*> m_multiBodyConstraints;
 
-
-	MultiBodyInplaceSolverIslandCallback(	btMultiBodyConstraintSolver*	solver,
-									btDispatcher* dispatcher)
-		:m_solverInfo(NULL),
-		m_solver(solver),
-		m_multiBodySortedConstraints(NULL),
-		m_numConstraints(0),
-		m_debugDrawer(NULL),
-		m_dispatcher(dispatcher)
+	MultiBodyInplaceSolverIslandCallback(btMultiBodyConstraintSolver* solver,
+										 btDispatcher* dispatcher)
+		: m_solverInfo(NULL),
+		  m_solver(solver),
+		  m_multiBodySortedConstraints(NULL),
+		  m_numConstraints(0),
+		  m_debugDrawer(NULL),
+		  m_dispatcher(dispatcher)
 	{
-
 	}
 
 	MultiBodyInplaceSolverIslandCallback& operator=(MultiBodyInplaceSolverIslandCallback& other)
@@ -260,7 +242,7 @@ struct MultiBodyInplaceSolverIslandCallback : public btSimulationIslandManager::
 		return *this;
 	}
 
-	SIMD_FORCE_INLINE void setup ( btContactSolverInfo* solverInfo, btTypedConstraint** sortedConstraints, int numConstraints, btMultiBodyConstraint** sortedMultiBodyConstraints,	int	numMultiBodyConstraints,	btIDebugDraw* debugDrawer)
+	SIMD_FORCE_INLINE void setup(btContactSolverInfo* solverInfo, btTypedConstraint** sortedConstraints, int numConstraints, btMultiBodyConstraint** sortedMultiBodyConstraints, int numMultiBodyConstraints, btIDebugDraw* debugDrawer)
 	{
 		btAssert(solverInfo);
 		m_solverInfo = solverInfo;
@@ -271,26 +253,27 @@ struct MultiBodyInplaceSolverIslandCallback : public btSimulationIslandManager::
 		m_numConstraints = numConstraints;
 
 		m_debugDrawer = debugDrawer;
-		m_bodies.resize (0);
-		m_manifolds.resize (0);
-		m_constraints.resize (0);
+		m_bodies.resize(0);
+		m_manifolds.resize(0);
+		m_constraints.resize(0);
 		m_multiBodyConstraints.resize(0);
 	}
 
-    void    setMultiBodyConstraintSolver(btMultiBodyConstraintSolver* solver)
-    {
-        m_solver = solver;
-    }
-    
-	virtual	void	processIsland(btCollisionObject** bodies,int numBodies,btPersistentManifold**	manifolds,int numManifolds, int islandId)
+	void setMultiBodyConstraintSolver(btMultiBodyConstraintSolver* solver)
 	{
-		if (islandId<0)
+		m_solver = solver;
+	}
+
+	virtual void processIsland(btCollisionObject** bodies, int numBodies, btPersistentManifold** manifolds, int numManifolds, int islandId)
+	{
+		if (islandId < 0)
 		{
 			///we don't split islands, so all constraints/contact manifolds/bodies are passed into the solver regardless the island id
-			m_solver->solveMultiBodyGroup( bodies,numBodies,manifolds, numManifolds,m_sortedConstraints, m_numConstraints, &m_multiBodySortedConstraints[0],m_numConstraints,*m_solverInfo,m_debugDrawer,m_dispatcher);
-		} else
+			m_solver->solveMultiBodyGroup(bodies, numBodies, manifolds, numManifolds, m_sortedConstraints, m_numConstraints, &m_multiBodySortedConstraints[0], m_numConstraints, *m_solverInfo, m_debugDrawer, m_dispatcher);
+		}
+		else
 		{
-				//also add all non-contact constraints/joints for this island
+			//also add all non-contact constraints/joints for this island
 			btTypedConstraint** startConstraint = 0;
 			btMultiBodyConstraint** startMultiBodyConstraint = 0;
 
@@ -298,10 +281,10 @@ struct MultiBodyInplaceSolverIslandCallback : public btSimulationIslandManager::
 			int numCurMultiBodyConstraints = 0;
 
 			int i;
-			
+
 			//find the first constraint for this island
 
-			for (i=0;i<m_numConstraints;i++)
+			for (i = 0; i < m_numConstraints; i++)
 			{
 				if (btGetConstraintIslandId2(m_sortedConstraints[i]) == islandId)
 				{
@@ -310,7 +293,7 @@ struct MultiBodyInplaceSolverIslandCallback : public btSimulationIslandManager::
 				}
 			}
 			//count the number of constraints in this island
-			for (;i<m_numConstraints;i++)
+			for (; i < m_numConstraints; i++)
 			{
 				if (btGetConstraintIslandId2(m_sortedConstraints[i]) == islandId)
 				{
@@ -318,17 +301,16 @@ struct MultiBodyInplaceSolverIslandCallback : public btSimulationIslandManager::
 				}
 			}
 
-			for (i=0;i<m_numMultiBodyConstraints;i++)
+			for (i = 0; i < m_numMultiBodyConstraints; i++)
 			{
 				if (btGetMultiBodyConstraintIslandId(m_multiBodySortedConstraints[i]) == islandId)
 				{
-					
 					startMultiBodyConstraint = &m_multiBodySortedConstraints[i];
 					break;
 				}
 			}
 			//count the number of multi body constraints in this island
-			for (;i<m_numMultiBodyConstraints;i++)
+			for (; i < m_numMultiBodyConstraints; i++)
 			{
 				if (btGetMultiBodyConstraintIslandId(m_multiBodySortedConstraints[i]) == islandId)
 				{
@@ -341,101 +323,94 @@ struct MultiBodyInplaceSolverIslandCallback : public btSimulationIslandManager::
 			//	m_solver->solveGroup( bodies,numBodies,manifolds, numManifolds,startConstraint,numCurConstraints,*m_solverInfo,m_debugDrawer,m_dispatcher);
 			//} else
 			{
-				
-				for (i=0;i<numBodies;i++)
+				for (i = 0; i < numBodies; i++)
 					m_bodies.push_back(bodies[i]);
-				for (i=0;i<numManifolds;i++)
+				for (i = 0; i < numManifolds; i++)
 					m_manifolds.push_back(manifolds[i]);
-				for (i=0;i<numCurConstraints;i++)
+				for (i = 0; i < numCurConstraints; i++)
 					m_constraints.push_back(startConstraint[i]);
-				
-				for (i=0;i<numCurMultiBodyConstraints;i++)
+
+				for (i = 0; i < numCurMultiBodyConstraints; i++)
 					m_multiBodyConstraints.push_back(startMultiBodyConstraint[i]);
-				
-				if ((m_multiBodyConstraints.size()+m_constraints.size()+m_manifolds.size())>m_solverInfo->m_minimumSolverBatchSize)
+
+				if ((m_multiBodyConstraints.size() + m_constraints.size() + m_manifolds.size()) > m_solverInfo->m_minimumSolverBatchSize)
 				{
 					processConstraints();
-				} else
+				}
+				else
 				{
 					//printf("deferred\n");
 				}
 			}
 		}
 	}
-	void	processConstraints()
+	void processConstraints()
 	{
-
-		btCollisionObject** bodies = m_bodies.size()? &m_bodies[0]:0;
-		btPersistentManifold** manifold = m_manifolds.size()?&m_manifolds[0]:0;
-		btTypedConstraint** constraints = m_constraints.size()?&m_constraints[0]:0;
-		btMultiBodyConstraint** multiBodyConstraints = m_multiBodyConstraints.size() ? &m_multiBodyConstraints[0] : 0;			
+		btCollisionObject** bodies = m_bodies.size() ? &m_bodies[0] : 0;
+		btPersistentManifold** manifold = m_manifolds.size() ? &m_manifolds[0] : 0;
+		btTypedConstraint** constraints = m_constraints.size() ? &m_constraints[0] : 0;
+		btMultiBodyConstraint** multiBodyConstraints = m_multiBodyConstraints.size() ? &m_multiBodyConstraints[0] : 0;
 
 		//printf("mb contacts = %d, mb constraints = %d\n", mbContacts, m_multiBodyConstraints.size());
-	
-		m_solver->solveMultiBodyGroup( bodies,m_bodies.size(),manifold, m_manifolds.size(),constraints, m_constraints.size() ,multiBodyConstraints, m_multiBodyConstraints.size(), *m_solverInfo,m_debugDrawer,m_dispatcher);
+
+		m_solver->solveMultiBodyGroup(bodies, m_bodies.size(), manifold, m_manifolds.size(), constraints, m_constraints.size(), multiBodyConstraints, m_multiBodyConstraints.size(), *m_solverInfo, m_debugDrawer, m_dispatcher);
 		m_bodies.resize(0);
 		m_manifolds.resize(0);
 		m_constraints.resize(0);
 		m_multiBodyConstraints.resize(0);
 	}
-
 };
 
-
-
-btMultiBodyDynamicsWorld::btMultiBodyDynamicsWorld(btDispatcher* dispatcher,btBroadphaseInterface* pairCache,btMultiBodyConstraintSolver* constraintSolver,btCollisionConfiguration* collisionConfiguration)
-	:btDiscreteDynamicsWorld(dispatcher,pairCache,constraintSolver,collisionConfiguration),
-	m_multiBodyConstraintSolver(constraintSolver)
+btMultiBodyDynamicsWorld::btMultiBodyDynamicsWorld(btDispatcher* dispatcher, btBroadphaseInterface* pairCache, btMultiBodyConstraintSolver* constraintSolver, btCollisionConfiguration* collisionConfiguration)
+	: btDiscreteDynamicsWorld(dispatcher, pairCache, constraintSolver, collisionConfiguration),
+	  m_multiBodyConstraintSolver(constraintSolver)
 {
 	//split impulse is not yet supported for Featherstone hierarchies
-//	getSolverInfo().m_splitImpulse = false;
-	getSolverInfo().m_solverMode |=SOLVER_USE_2_FRICTION_DIRECTIONS;
-	m_solverMultiBodyIslandCallback = new MultiBodyInplaceSolverIslandCallback(constraintSolver,dispatcher);
+	//	getSolverInfo().m_splitImpulse = false;
+	getSolverInfo().m_solverMode |= SOLVER_USE_2_FRICTION_DIRECTIONS;
+	m_solverMultiBodyIslandCallback = new MultiBodyInplaceSolverIslandCallback(constraintSolver, dispatcher);
 }
 
-btMultiBodyDynamicsWorld::~btMultiBodyDynamicsWorld ()
+btMultiBodyDynamicsWorld::~btMultiBodyDynamicsWorld()
 {
 	delete m_solverMultiBodyIslandCallback;
 }
 
-void    btMultiBodyDynamicsWorld::setMultiBodyConstraintSolver(btMultiBodyConstraintSolver* solver)
+void btMultiBodyDynamicsWorld::setMultiBodyConstraintSolver(btMultiBodyConstraintSolver* solver)
 {
-    m_multiBodyConstraintSolver = solver;
-    m_solverMultiBodyIslandCallback->setMultiBodyConstraintSolver(solver);
-    btDiscreteDynamicsWorld::setConstraintSolver(solver);
+	m_multiBodyConstraintSolver = solver;
+	m_solverMultiBodyIslandCallback->setMultiBodyConstraintSolver(solver);
+	btDiscreteDynamicsWorld::setConstraintSolver(solver);
 }
 
-void    btMultiBodyDynamicsWorld::setConstraintSolver(btConstraintSolver* solver)
+void btMultiBodyDynamicsWorld::setConstraintSolver(btConstraintSolver* solver)
 {
-    if (solver->getSolverType()==BT_MULTIBODY_SOLVER)
-    {
-        m_multiBodyConstraintSolver = (btMultiBodyConstraintSolver*)solver;
-    }
-    btDiscreteDynamicsWorld::setConstraintSolver(solver);
+	if (solver->getSolverType() == BT_MULTIBODY_SOLVER)
+	{
+		m_multiBodyConstraintSolver = (btMultiBodyConstraintSolver*)solver;
+	}
+	btDiscreteDynamicsWorld::setConstraintSolver(solver);
 }
 
-void	btMultiBodyDynamicsWorld::forwardKinematics()
+void btMultiBodyDynamicsWorld::forwardKinematics()
 {
-
-	for (int b=0;b<m_multiBodies.size();b++)
+	for (int b = 0; b < m_multiBodies.size(); b++)
 	{
 		btMultiBody* bod = m_multiBodies[b];
-		bod->forwardKinematics(m_scratch_world_to_local,m_scratch_local_origin);
+		bod->forwardKinematics(m_scratch_world_to_local, m_scratch_local_origin);
 	}
 }
-void	btMultiBodyDynamicsWorld::solveConstraints(btContactSolverInfo& solverInfo)
+void btMultiBodyDynamicsWorld::solveConstraints(btContactSolverInfo& solverInfo)
 {
 	forwardKinematics();
 
-
-
 	BT_PROFILE("solveConstraints");
-	
+
 	clearMultiBodyConstraintForces();
 
-	m_sortedConstraints.resize( m_constraints.size());
-	int i; 
-	for (i=0;i<getNumConstraints();i++)
+	m_sortedConstraints.resize(m_constraints.size());
+	int i;
+	for (i = 0; i < getNumConstraints(); i++)
 	{
 		m_sortedConstraints[i] = m_constraints[i];
 	}
@@ -443,109 +418,120 @@ void	btMultiBodyDynamicsWorld::solveConstraints(btContactSolverInfo& solverInfo)
 	btTypedConstraint** constraintsPtr = getNumConstraints() ? &m_sortedConstraints[0] : 0;
 
 	m_sortedMultiBodyConstraints.resize(m_multiBodyConstraints.size());
-	for (i=0;i<m_multiBodyConstraints.size();i++)
+	for (i = 0; i < m_multiBodyConstraints.size(); i++)
 	{
 		m_sortedMultiBodyConstraints[i] = m_multiBodyConstraints[i];
 	}
 	m_sortedMultiBodyConstraints.quickSort(btSortMultiBodyConstraintOnIslandPredicate());
 
-	btMultiBodyConstraint** sortedMultiBodyConstraints = m_sortedMultiBodyConstraints.size() ?  &m_sortedMultiBodyConstraints[0] : 0;
-	
+	btMultiBodyConstraint** sortedMultiBodyConstraints = m_sortedMultiBodyConstraints.size() ? &m_sortedMultiBodyConstraints[0] : 0;
 
-	m_solverMultiBodyIslandCallback->setup(&solverInfo,constraintsPtr,m_sortedConstraints.size(),sortedMultiBodyConstraints,m_sortedMultiBodyConstraints.size(), getDebugDrawer());
+	m_solverMultiBodyIslandCallback->setup(&solverInfo, constraintsPtr, m_sortedConstraints.size(), sortedMultiBodyConstraints, m_sortedMultiBodyConstraints.size(), getDebugDrawer());
 	m_constraintSolver->prepareSolve(getCollisionWorld()->getNumCollisionObjects(), getCollisionWorld()->getDispatcher()->getNumManifolds());
-	
 
 #ifndef BT_USE_VIRTUAL_CLEARFORCES_AND_GRAVITY
 	{
 		BT_PROFILE("btMultiBody addForce");
-		for (int i=0;i<this->m_multiBodies.size();i++)
+		for (int i = 0; i < this->m_multiBodies.size(); i++)
 		{
 			btMultiBody* bod = m_multiBodies[i];
 
 			bool isSleeping = false;
-			
+
 			if (bod->getBaseCollider() && bod->getBaseCollider()->getActivationState() == ISLAND_SLEEPING)
 			{
 				isSleeping = true;
-			} 
-			for (int b=0;b<bod->getNumLinks();b++)
+			}
+			for (int b = 0; b < bod->getNumLinks(); b++)
 			{
-				if (bod->getLink(b).m_collider && bod->getLink(b).m_collider->getActivationState()==ISLAND_SLEEPING)
+				if (bod->getLink(b).m_collider && bod->getLink(b).m_collider->getActivationState() == ISLAND_SLEEPING)
 					isSleeping = true;
-			} 
+			}
 
 			if (!isSleeping)
 			{
 				//useless? they get resized in stepVelocities once again (AND DIFFERENTLY)
-				m_scratch_r.resize(bod->getNumLinks()+1);			//multidof? ("Y"s use it and it is used to store qdd)
-				m_scratch_v.resize(bod->getNumLinks()+1);
-				m_scratch_m.resize(bod->getNumLinks()+1);
+				m_scratch_r.resize(bod->getNumLinks() + 1);  //multidof? ("Y"s use it and it is used to store qdd)
+				m_scratch_v.resize(bod->getNumLinks() + 1);
+				m_scratch_m.resize(bod->getNumLinks() + 1);
 
 				bod->addBaseForce(m_gravity * bod->getBaseMass());
 
-				for (int j = 0; j < bod->getNumLinks(); ++j) 
+				for (int j = 0; j < bod->getNumLinks(); ++j)
 				{
 					bod->addLinkForce(j, m_gravity * bod->getLinkMass(j));
 				}
-			}//if (!isSleeping)
+			}  //if (!isSleeping)
 		}
 	}
-#endif //BT_USE_VIRTUAL_CLEARFORCES_AND_GRAVITY
-	
+#endif  //BT_USE_VIRTUAL_CLEARFORCES_AND_GRAVITY
 
 	{
 		BT_PROFILE("btMultiBody stepVelocities");
-		for (int i=0;i<this->m_multiBodies.size();i++)
+		for (int i = 0; i < this->m_multiBodies.size(); i++)
 		{
 			btMultiBody* bod = m_multiBodies[i];
 
 			bool isSleeping = false;
-			
+
 			if (bod->getBaseCollider() && bod->getBaseCollider()->getActivationState() == ISLAND_SLEEPING)
 			{
 				isSleeping = true;
-			} 
-			for (int b=0;b<bod->getNumLinks();b++)
+			}
+			for (int b = 0; b < bod->getNumLinks(); b++)
 			{
-				if (bod->getLink(b).m_collider && bod->getLink(b).m_collider->getActivationState()==ISLAND_SLEEPING)
+				if (bod->getLink(b).m_collider && bod->getLink(b).m_collider->getActivationState() == ISLAND_SLEEPING)
 					isSleeping = true;
-			} 
+			}
 
 			if (!isSleeping)
 			{
 				//useless? they get resized in stepVelocities once again (AND DIFFERENTLY)
-				m_scratch_r.resize(bod->getNumLinks()+1);			//multidof? ("Y"s use it and it is used to store qdd)
-				m_scratch_v.resize(bod->getNumLinks()+1);
-				m_scratch_m.resize(bod->getNumLinks()+1);
+				m_scratch_r.resize(bod->getNumLinks() + 1);  //multidof? ("Y"s use it and it is used to store qdd)
+				m_scratch_v.resize(bod->getNumLinks() + 1);
+				m_scratch_m.resize(bod->getNumLinks() + 1);
 				bool doNotUpdatePos = false;
-
+                bool isConstraintPass = false;
 				{
-					if(!bod->isUsingRK4Integration())
+					if (!bod->isUsingRK4Integration())
 					{
-						bod->computeAccelerationsArticulatedBodyAlgorithmMultiDof(solverInfo.m_timeStep, m_scratch_r, m_scratch_v, m_scratch_m);
+						bod->computeAccelerationsArticulatedBodyAlgorithmMultiDof(solverInfo.m_timeStep,
+						m_scratch_r, m_scratch_v, m_scratch_m,isConstraintPass,
+						getSolverInfo().m_jointFeedbackInWorldSpace,
+						getSolverInfo().m_jointFeedbackInJointFrame);
 					}
 					else
-					{						
+					{
 						//
 						int numDofs = bod->getNumDofs() + 6;
 						int numPosVars = bod->getNumPosVars() + 7;
-						btAlignedObjectArray<btScalar> scratch_r2; scratch_r2.resize(2*numPosVars + 8*numDofs);
+						btAlignedObjectArray<btScalar> scratch_r2;
+						scratch_r2.resize(2 * numPosVars + 8 * numDofs);
 						//convenience
-						btScalar *pMem = &scratch_r2[0];
-						btScalar *scratch_q0 = pMem; pMem += numPosVars;
-						btScalar *scratch_qx = pMem; pMem += numPosVars;
-						btScalar *scratch_qd0 = pMem; pMem += numDofs;
-						btScalar *scratch_qd1 = pMem; pMem += numDofs;
-						btScalar *scratch_qd2 = pMem; pMem += numDofs;
-						btScalar *scratch_qd3 = pMem; pMem += numDofs;
-						btScalar *scratch_qdd0 = pMem; pMem += numDofs;
-						btScalar *scratch_qdd1 = pMem; pMem += numDofs;
-						btScalar *scratch_qdd2 = pMem; pMem += numDofs;
-						btScalar *scratch_qdd3 = pMem; pMem += numDofs;
-						btAssert((pMem - (2*numPosVars + 8*numDofs)) == &scratch_r2[0]);
+						btScalar* pMem = &scratch_r2[0];
+						btScalar* scratch_q0 = pMem;
+						pMem += numPosVars;
+						btScalar* scratch_qx = pMem;
+						pMem += numPosVars;
+						btScalar* scratch_qd0 = pMem;
+						pMem += numDofs;
+						btScalar* scratch_qd1 = pMem;
+						pMem += numDofs;
+						btScalar* scratch_qd2 = pMem;
+						pMem += numDofs;
+						btScalar* scratch_qd3 = pMem;
+						pMem += numDofs;
+						btScalar* scratch_qdd0 = pMem;
+						pMem += numDofs;
+						btScalar* scratch_qdd1 = pMem;
+						pMem += numDofs;
+						btScalar* scratch_qdd2 = pMem;
+						pMem += numDofs;
+						btScalar* scratch_qdd3 = pMem;
+						pMem += numDofs;
+						btAssert((pMem - (2 * numPosVars + 8 * numDofs)) == &scratch_r2[0]);
 
-						/////						
+						/////
 						//copy q0 to scratch_q0 and qd0 to scratch_qd0
 						scratch_q0[0] = bod->getWorldToBaseRot().x();
 						scratch_q0[1] = bod->getWorldToBaseRot().y();
@@ -555,83 +541,88 @@ void	btMultiBodyDynamicsWorld::solveConstraints(btContactSolverInfo& solverInfo)
 						scratch_q0[5] = bod->getBasePos().y();
 						scratch_q0[6] = bod->getBasePos().z();
 						//
-						for(int link = 0; link < bod->getNumLinks(); ++link)
+						for (int link = 0; link < bod->getNumLinks(); ++link)
 						{
-							for(int dof = 0; dof < bod->getLink(link).m_posVarCount; ++dof)
-								scratch_q0[7 + bod->getLink(link).m_cfgOffset + dof] = bod->getLink(link).m_jointPos[dof];							
+							for (int dof = 0; dof < bod->getLink(link).m_posVarCount; ++dof)
+								scratch_q0[7 + bod->getLink(link).m_cfgOffset + dof] = bod->getLink(link).m_jointPos[dof];
 						}
 						//
-						for(int dof = 0; dof < numDofs; ++dof)								
+						for (int dof = 0; dof < numDofs; ++dof)
 							scratch_qd0[dof] = bod->getVelocityVector()[dof];
 						////
 						struct
 						{
-						    btMultiBody *bod;
-                            btScalar *scratch_qx, *scratch_q0;
+							btMultiBody* bod;
+							btScalar *scratch_qx, *scratch_q0;
 
-						    void operator()()
-						    {
-						        for(int dof = 0; dof < bod->getNumPosVars() + 7; ++dof)
-                                    scratch_qx[dof] = scratch_q0[dof];
-						    }
+							void operator()()
+							{
+								for (int dof = 0; dof < bod->getNumPosVars() + 7; ++dof)
+									scratch_qx[dof] = scratch_q0[dof];
+							}
 						} pResetQx = {bod, scratch_qx, scratch_q0};
 						//
 						struct
 						{
-						    void operator()(btScalar dt, const btScalar *pDer, const btScalar *pCurVal, btScalar *pVal, int size)
-						    {
-						        for(int i = 0; i < size; ++i)
-                                    pVal[i] = pCurVal[i] + dt * pDer[i];
-						    }
+							void operator()(btScalar dt, const btScalar* pDer, const btScalar* pCurVal, btScalar* pVal, int size)
+							{
+								for (int i = 0; i < size; ++i)
+									pVal[i] = pCurVal[i] + dt * pDer[i];
+							}
 
 						} pEulerIntegrate;
 						//
 						struct
-                        {
-                            void operator()(btMultiBody *pBody, const btScalar *pData)
-                            {
-                                btScalar *pVel = const_cast<btScalar*>(pBody->getVelocityVector());
-
-                                for(int i = 0; i < pBody->getNumDofs() + 6; ++i)
-                                    pVel[i] = pData[i];
-
-                            }
-                        } pCopyToVelocityVector;
-						//
-                        struct
 						{
-						    void operator()(const btScalar *pSrc, btScalar *pDst, int start, int size)
-						    {
-						        for(int i = 0; i < size; ++i)
-                                    pDst[i] = pSrc[start + i];
-						    }
+							void operator()(btMultiBody* pBody, const btScalar* pData)
+							{
+								btScalar* pVel = const_cast<btScalar*>(pBody->getVelocityVector());
+
+								for (int i = 0; i < pBody->getNumDofs() + 6; ++i)
+									pVel[i] = pData[i];
+							}
+						} pCopyToVelocityVector;
+						//
+						struct
+						{
+							void operator()(const btScalar* pSrc, btScalar* pDst, int start, int size)
+							{
+								for (int i = 0; i < size; ++i)
+									pDst[i] = pSrc[start + i];
+							}
 						} pCopy;
 						//
 
 						btScalar h = solverInfo.m_timeStep;
-						#define output &m_scratch_r[bod->getNumDofs()]
-						//calc qdd0 from: q0 & qd0	
-						bod->computeAccelerationsArticulatedBodyAlgorithmMultiDof(0., m_scratch_r, m_scratch_v, m_scratch_m);
+#define output &m_scratch_r[bod->getNumDofs()]
+						//calc qdd0 from: q0 & qd0
+						bod->computeAccelerationsArticulatedBodyAlgorithmMultiDof(0., m_scratch_r, m_scratch_v, m_scratch_m,
+						isConstraintPass,getSolverInfo().m_jointFeedbackInWorldSpace,
+						getSolverInfo().m_jointFeedbackInJointFrame);
 						pCopy(output, scratch_qdd0, 0, numDofs);
 						//calc q1 = q0 + h/2 * qd0
 						pResetQx();
-						bod->stepPositionsMultiDof(btScalar(.5)*h, scratch_qx, scratch_qd0);
+						bod->stepPositionsMultiDof(btScalar(.5) * h, scratch_qx, scratch_qd0);
 						//calc qd1 = qd0 + h/2 * qdd0
-						pEulerIntegrate(btScalar(.5)*h, scratch_qdd0, scratch_qd0, scratch_qd1, numDofs);
+						pEulerIntegrate(btScalar(.5) * h, scratch_qdd0, scratch_qd0, scratch_qd1, numDofs);
 						//
 						//calc qdd1 from: q1 & qd1
 						pCopyToVelocityVector(bod, scratch_qd1);
-						bod->computeAccelerationsArticulatedBodyAlgorithmMultiDof(0., m_scratch_r, m_scratch_v, m_scratch_m);
+						bod->computeAccelerationsArticulatedBodyAlgorithmMultiDof(0., m_scratch_r, m_scratch_v, m_scratch_m,
+						isConstraintPass,getSolverInfo().m_jointFeedbackInWorldSpace,
+						getSolverInfo().m_jointFeedbackInJointFrame);
 						pCopy(output, scratch_qdd1, 0, numDofs);
 						//calc q2 = q0 + h/2 * qd1
 						pResetQx();
-						bod->stepPositionsMultiDof(btScalar(.5)*h, scratch_qx, scratch_qd1);
+						bod->stepPositionsMultiDof(btScalar(.5) * h, scratch_qx, scratch_qd1);
 						//calc qd2 = qd0 + h/2 * qdd1
-						pEulerIntegrate(btScalar(.5)*h, scratch_qdd1, scratch_qd0, scratch_qd2, numDofs);
+						pEulerIntegrate(btScalar(.5) * h, scratch_qdd1, scratch_qd0, scratch_qd2, numDofs);
 						//
 						//calc qdd2 from: q2 & qd2
 						pCopyToVelocityVector(bod, scratch_qd2);
-						bod->computeAccelerationsArticulatedBodyAlgorithmMultiDof(0., m_scratch_r, m_scratch_v, m_scratch_m);
+						bod->computeAccelerationsArticulatedBodyAlgorithmMultiDof(0., m_scratch_r, m_scratch_v, m_scratch_m,
+						isConstraintPass,getSolverInfo().m_jointFeedbackInWorldSpace,
+						getSolverInfo().m_jointFeedbackInJointFrame);
 						pCopy(output, scratch_qdd2, 0, numDofs);
 						//calc q3 = q0 + h * qd2
 						pResetQx();
@@ -641,156 +632,158 @@ void	btMultiBodyDynamicsWorld::solveConstraints(btContactSolverInfo& solverInfo)
 						//
 						//calc qdd3 from: q3 & qd3
 						pCopyToVelocityVector(bod, scratch_qd3);
-						bod->computeAccelerationsArticulatedBodyAlgorithmMultiDof(0., m_scratch_r, m_scratch_v, m_scratch_m);
+						bod->computeAccelerationsArticulatedBodyAlgorithmMultiDof(0., m_scratch_r, m_scratch_v, m_scratch_m,
+						isConstraintPass,getSolverInfo().m_jointFeedbackInWorldSpace,
+						getSolverInfo().m_jointFeedbackInJointFrame);
 						pCopy(output, scratch_qdd3, 0, numDofs);
 
 						//
 						//calc q = q0 + h/6(qd0 + 2*(qd1 + qd2) + qd3)
-						//calc qd = qd0 + h/6(qdd0 + 2*(qdd1 + qdd2) + qdd3)						
-						btAlignedObjectArray<btScalar> delta_q; delta_q.resize(numDofs);
-						btAlignedObjectArray<btScalar> delta_qd; delta_qd.resize(numDofs);
-						for(int i = 0; i < numDofs; ++i)
+						//calc qd = qd0 + h/6(qdd0 + 2*(qdd1 + qdd2) + qdd3)
+						btAlignedObjectArray<btScalar> delta_q;
+						delta_q.resize(numDofs);
+						btAlignedObjectArray<btScalar> delta_qd;
+						delta_qd.resize(numDofs);
+						for (int i = 0; i < numDofs; ++i)
 						{
-							delta_q[i] = h/btScalar(6.)*(scratch_qd0[i] + 2*scratch_qd1[i] + 2*scratch_qd2[i] + scratch_qd3[i]);
-							delta_qd[i] = h/btScalar(6.)*(scratch_qdd0[i] + 2*scratch_qdd1[i] + 2*scratch_qdd2[i] + scratch_qdd3[i]);							
+							delta_q[i] = h / btScalar(6.) * (scratch_qd0[i] + 2 * scratch_qd1[i] + 2 * scratch_qd2[i] + scratch_qd3[i]);
+							delta_qd[i] = h / btScalar(6.) * (scratch_qdd0[i] + 2 * scratch_qdd1[i] + 2 * scratch_qdd2[i] + scratch_qdd3[i]);
 							//delta_q[i] = h*scratch_qd0[i];
 							//delta_qd[i] = h*scratch_qdd0[i];
 						}
 						//
 						pCopyToVelocityVector(bod, scratch_qd0);
-						bod->applyDeltaVeeMultiDof(&delta_qd[0], 1);						
+						bod->applyDeltaVeeMultiDof(&delta_qd[0], 1);
 						//
-						if(!doNotUpdatePos)
+						if (!doNotUpdatePos)
 						{
-							btScalar *pRealBuf = const_cast<btScalar *>(bod->getVelocityVector());
-							pRealBuf += 6 + bod->getNumDofs() + bod->getNumDofs()*bod->getNumDofs();
+							btScalar* pRealBuf = const_cast<btScalar*>(bod->getVelocityVector());
+							pRealBuf += 6 + bod->getNumDofs() + bod->getNumDofs() * bod->getNumDofs();
 
-							for(int i = 0; i < numDofs; ++i)
+							for (int i = 0; i < numDofs; ++i)
 								pRealBuf[i] = delta_q[i];
 
 							//bod->stepPositionsMultiDof(1, 0, &delta_q[0]);
-							bod->setPosUpdated(true);							
+							bod->setPosUpdated(true);
 						}
 
 						//ugly hack which resets the cached data to t0 (needed for constraint solver)
 						{
-							for(int link = 0; link < bod->getNumLinks(); ++link)
+							for (int link = 0; link < bod->getNumLinks(); ++link)
 								bod->getLink(link).updateCacheMultiDof();
-							bod->computeAccelerationsArticulatedBodyAlgorithmMultiDof(0, m_scratch_r, m_scratch_v, m_scratch_m);
+							bod->computeAccelerationsArticulatedBodyAlgorithmMultiDof(0, m_scratch_r, m_scratch_v, m_scratch_m,
+							isConstraintPass,getSolverInfo().m_jointFeedbackInWorldSpace,
+							getSolverInfo().m_jointFeedbackInJointFrame);
 						}
-						
 					}
 				}
-				
+
 #ifndef BT_USE_VIRTUAL_CLEARFORCES_AND_GRAVITY
 				bod->clearForcesAndTorques();
-#endif //BT_USE_VIRTUAL_CLEARFORCES_AND_GRAVITY
-			}//if (!isSleeping)
+#endif         //BT_USE_VIRTUAL_CLEARFORCES_AND_GRAVITY
+			}  //if (!isSleeping)
 		}
 	}
 
 	/// solve all the constraints for this island
 	m_islandManager->buildAndProcessIslands(getCollisionWorld()->getDispatcher(), getCollisionWorld(), m_solverMultiBodyIslandCallback);
 
-
 	m_solverMultiBodyIslandCallback->processConstraints();
-	
+
 	m_constraintSolver->allSolved(solverInfo, m_debugDrawer);
 
 	{
-                BT_PROFILE("btMultiBody stepVelocities");
-                for (int i=0;i<this->m_multiBodies.size();i++)
-                {
-                        btMultiBody* bod = m_multiBodies[i];
+		BT_PROFILE("btMultiBody stepVelocities");
+		for (int i = 0; i < this->m_multiBodies.size(); i++)
+		{
+			btMultiBody* bod = m_multiBodies[i];
 
-                        bool isSleeping = false;
+			bool isSleeping = false;
 
-                        if (bod->getBaseCollider() && bod->getBaseCollider()->getActivationState() == ISLAND_SLEEPING)
-                        {
-                                isSleeping = true;
-                        }
-                        for (int b=0;b<bod->getNumLinks();b++)
-                        {
-                                if (bod->getLink(b).m_collider && bod->getLink(b).m_collider->getActivationState()==ISLAND_SLEEPING)
-                                        isSleeping = true;
-                        }
+			if (bod->getBaseCollider() && bod->getBaseCollider()->getActivationState() == ISLAND_SLEEPING)
+			{
+				isSleeping = true;
+			}
+			for (int b = 0; b < bod->getNumLinks(); b++)
+			{
+				if (bod->getLink(b).m_collider && bod->getLink(b).m_collider->getActivationState() == ISLAND_SLEEPING)
+					isSleeping = true;
+			}
 
-                        if (!isSleeping)
-                        {
-                                //useless? they get resized in stepVelocities once again (AND DIFFERENTLY)
-                                m_scratch_r.resize(bod->getNumLinks()+1);                 //multidof? ("Y"s use it and it is used to store qdd)
-                                m_scratch_v.resize(bod->getNumLinks()+1);
-                                m_scratch_m.resize(bod->getNumLinks()+1);
+			if (!isSleeping)
+			{
+				//useless? they get resized in stepVelocities once again (AND DIFFERENTLY)
+				m_scratch_r.resize(bod->getNumLinks() + 1);  //multidof? ("Y"s use it and it is used to store qdd)
+				m_scratch_v.resize(bod->getNumLinks() + 1);
+				m_scratch_m.resize(bod->getNumLinks() + 1);
 
-                                
-                            {
-                                if(!bod->isUsingRK4Integration())
-                                {
-									bool isConstraintPass = true;
-                                    bod->computeAccelerationsArticulatedBodyAlgorithmMultiDof(solverInfo.m_timeStep, m_scratch_r, m_scratch_v, m_scratch_m, isConstraintPass);
-                                }
+				{
+					if (!bod->isUsingRK4Integration())
+					{
+						bool isConstraintPass = true;
+						bod->computeAccelerationsArticulatedBodyAlgorithmMultiDof(solverInfo.m_timeStep, m_scratch_r, m_scratch_v, m_scratch_m, isConstraintPass,
+						getSolverInfo().m_jointFeedbackInWorldSpace,
+						getSolverInfo().m_jointFeedbackInJointFrame);
+					}
 				}
 			}
 		}
 	}
 
-	for (int i=0;i<this->m_multiBodies.size();i++)
+	for (int i = 0; i < this->m_multiBodies.size(); i++)
 	{
 		btMultiBody* bod = m_multiBodies[i];
 		bod->processDeltaVeeMultiDof2();
 	}
-
 }
 
-void	btMultiBodyDynamicsWorld::integrateTransforms(btScalar timeStep)
+void btMultiBodyDynamicsWorld::integrateTransforms(btScalar timeStep)
 {
 	btDiscreteDynamicsWorld::integrateTransforms(timeStep);
 
 	{
 		BT_PROFILE("btMultiBody stepPositions");
 		//integrate and update the Featherstone hierarchies
-	
-		for (int b=0;b<m_multiBodies.size();b++)
+
+		for (int b = 0; b < m_multiBodies.size(); b++)
 		{
 			btMultiBody* bod = m_multiBodies[b];
 			bool isSleeping = false;
 			if (bod->getBaseCollider() && bod->getBaseCollider()->getActivationState() == ISLAND_SLEEPING)
 			{
 				isSleeping = true;
-			} 
-			for (int b=0;b<bod->getNumLinks();b++)
+			}
+			for (int b = 0; b < bod->getNumLinks(); b++)
 			{
-				if (bod->getLink(b).m_collider && bod->getLink(b).m_collider->getActivationState()==ISLAND_SLEEPING)
+				if (bod->getLink(b).m_collider && bod->getLink(b).m_collider->getActivationState() == ISLAND_SLEEPING)
 					isSleeping = true;
 			}
 
-
 			if (!isSleeping)
 			{
 				int nLinks = bod->getNumLinks();
 
 				///base + num m_links
-			
-				
+
 				{
-					if(!bod->isPosUpdated())
+					if (!bod->isPosUpdated())
 						bod->stepPositionsMultiDof(timeStep);
 					else
 					{
-						btScalar *pRealBuf = const_cast<btScalar *>(bod->getVelocityVector());
-						pRealBuf += 6 + bod->getNumDofs() + bod->getNumDofs()*bod->getNumDofs();
+						btScalar* pRealBuf = const_cast<btScalar*>(bod->getVelocityVector());
+						pRealBuf += 6 + bod->getNumDofs() + bod->getNumDofs() * bod->getNumDofs();
 
 						bod->stepPositionsMultiDof(1, 0, pRealBuf);
 						bod->setPosUpdated(false);
 					}
 				}
-				
-				m_scratch_world_to_local.resize(nLinks+1);
-				m_scratch_local_origin.resize(nLinks+1);
 
-				bod->updateCollisionObjectWorldTransforms(m_scratch_world_to_local,m_scratch_local_origin);
-				
-			} else
+				m_scratch_world_to_local.resize(nLinks + 1);
+				m_scratch_local_origin.resize(nLinks + 1);
+
+				bod->updateCollisionObjectWorldTransforms(m_scratch_world_to_local, m_scratch_local_origin);
+			}
+			else
 			{
 				bod->clearVelocities();
 			}
@@ -798,14 +791,12 @@ void	btMultiBodyDynamicsWorld::integrateTransforms(btScalar timeStep)
 	}
 }
 
-
-
-void	btMultiBodyDynamicsWorld::addMultiBodyConstraint( btMultiBodyConstraint* constraint)
+void btMultiBodyDynamicsWorld::addMultiBodyConstraint(btMultiBodyConstraint* constraint)
 {
 	m_multiBodyConstraints.push_back(constraint);
 }
 
-void	btMultiBodyDynamicsWorld::removeMultiBodyConstraint( btMultiBodyConstraint* constraint)
+void btMultiBodyDynamicsWorld::removeMultiBodyConstraint(btMultiBodyConstraint* constraint)
 {
 	m_multiBodyConstraints.remove(constraint);
 }
@@ -815,8 +806,7 @@ void btMultiBodyDynamicsWorld::debugDrawMultiBodyConstraint(btMultiBodyConstrain
 	constraint->debugDraw(getDebugDrawer());
 }
 
-
-void	btMultiBodyDynamicsWorld::debugDrawWorld()
+void btMultiBodyDynamicsWorld::debugDrawWorld()
 {
 	BT_PROFILE("btMultiBodyDynamicsWorld debugDrawWorld");
 
@@ -826,7 +816,7 @@ void	btMultiBodyDynamicsWorld::debugDrawWorld()
 	if (getDebugDrawer())
 	{
 		int mode = getDebugDrawer()->getDebugMode();
-		if (mode  & (btIDebugDraw::DBG_DrawConstraints | btIDebugDraw::DBG_DrawConstraintLimits))
+		if (mode & (btIDebugDraw::DBG_DrawConstraints | btIDebugDraw::DBG_DrawConstraintLimits))
 		{
 			drawConstraints = true;
 		}
@@ -834,160 +824,148 @@ void	btMultiBodyDynamicsWorld::debugDrawWorld()
 		if (drawConstraints)
 		{
 			BT_PROFILE("btMultiBody debugDrawWorld");
-			
 
-			for (int c=0;c<m_multiBodyConstraints.size();c++)
+			for (int c = 0; c < m_multiBodyConstraints.size(); c++)
 			{
 				btMultiBodyConstraint* constraint = m_multiBodyConstraints[c];
 				debugDrawMultiBodyConstraint(constraint);
 			}
 
-			for (int b = 0; b<m_multiBodies.size(); b++)
+			for (int b = 0; b < m_multiBodies.size(); b++)
 			{
 				btMultiBody* bod = m_multiBodies[b];
-				bod->forwardKinematics(m_scratch_world_to_local1,m_scratch_local_origin1);
-		
-				if (mode  & btIDebugDraw::DBG_DrawFrames)
+				bod->forwardKinematics(m_scratch_world_to_local1, m_scratch_local_origin1);
+
+				if (mode & btIDebugDraw::DBG_DrawFrames)
 				{
 					getDebugDrawer()->drawTransform(bod->getBaseWorldTransform(), 0.1);
 				}
 
-				for (int m = 0; m<bod->getNumLinks(); m++)
+				for (int m = 0; m < bod->getNumLinks(); m++)
 				{
-					
 					const btTransform& tr = bod->getLink(m).m_cachedWorldTransform;
-					if (mode  & btIDebugDraw::DBG_DrawFrames)
+					if (mode & btIDebugDraw::DBG_DrawFrames)
 					{
 						getDebugDrawer()->drawTransform(tr, 0.1);
 					}
-						//draw the joint axis
-					if (bod->getLink(m).m_jointType==btMultibodyLink::eRevolute)
+					//draw the joint axis
+					if (bod->getLink(m).m_jointType == btMultibodyLink::eRevolute)
 					{
-						btVector3 vec = quatRotate(tr.getRotation(),bod->getLink(m).m_axes[0].m_topVec)*0.1;
-					
-						btVector4 color(0,0,0,1);//1,1,1);
-						btVector3 from = vec+tr.getOrigin()-quatRotate(tr.getRotation(),bod->getLink(m).m_dVector);
-						btVector3 to = tr.getOrigin()-quatRotate(tr.getRotation(),bod->getLink(m).m_dVector);
-						getDebugDrawer()->drawLine(from,to,color);
+						btVector3 vec = quatRotate(tr.getRotation(), bod->getLink(m).m_axes[0].m_topVec) * 0.1;
+
+						btVector4 color(0, 0, 0, 1);  //1,1,1);
+						btVector3 from = vec + tr.getOrigin() - quatRotate(tr.getRotation(), bod->getLink(m).m_dVector);
+						btVector3 to = tr.getOrigin() - quatRotate(tr.getRotation(), bod->getLink(m).m_dVector);
+						getDebugDrawer()->drawLine(from, to, color);
 					}
-					if (bod->getLink(m).m_jointType==btMultibodyLink::eFixed)
+					if (bod->getLink(m).m_jointType == btMultibodyLink::eFixed)
 					{
-						btVector3 vec = quatRotate(tr.getRotation(),bod->getLink(m).m_axes[0].m_bottomVec)*0.1;
-					
-						btVector4 color(0,0,0,1);//1,1,1);
-						btVector3 from = vec+tr.getOrigin()-quatRotate(tr.getRotation(),bod->getLink(m).m_dVector);
-						btVector3 to = tr.getOrigin()-quatRotate(tr.getRotation(),bod->getLink(m).m_dVector);
-						getDebugDrawer()->drawLine(from,to,color);
+						btVector3 vec = quatRotate(tr.getRotation(), bod->getLink(m).m_axes[0].m_bottomVec) * 0.1;
+
+						btVector4 color(0, 0, 0, 1);  //1,1,1);
+						btVector3 from = vec + tr.getOrigin() - quatRotate(tr.getRotation(), bod->getLink(m).m_dVector);
+						btVector3 to = tr.getOrigin() - quatRotate(tr.getRotation(), bod->getLink(m).m_dVector);
+						getDebugDrawer()->drawLine(from, to, color);
 					}
-					if (bod->getLink(m).m_jointType==btMultibodyLink::ePrismatic)
+					if (bod->getLink(m).m_jointType == btMultibodyLink::ePrismatic)
 					{
-						btVector3 vec = quatRotate(tr.getRotation(),bod->getLink(m).m_axes[0].m_bottomVec)*0.1;
-					
-						btVector4 color(0,0,0,1);//1,1,1);
-						btVector3 from = vec+tr.getOrigin()-quatRotate(tr.getRotation(),bod->getLink(m).m_dVector);
-						btVector3 to = tr.getOrigin()-quatRotate(tr.getRotation(),bod->getLink(m).m_dVector);
-						getDebugDrawer()->drawLine(from,to,color);
+						btVector3 vec = quatRotate(tr.getRotation(), bod->getLink(m).m_axes[0].m_bottomVec) * 0.1;
+
+						btVector4 color(0, 0, 0, 1);  //1,1,1);
+						btVector3 from = vec + tr.getOrigin() - quatRotate(tr.getRotation(), bod->getLink(m).m_dVector);
+						btVector3 to = tr.getOrigin() - quatRotate(tr.getRotation(), bod->getLink(m).m_dVector);
+						getDebugDrawer()->drawLine(from, to, color);
 					}
-					
 				}
 			}
 		}
 	}
-
-	
 }
 
-
-
 void btMultiBodyDynamicsWorld::applyGravity()
 {
-        btDiscreteDynamicsWorld::applyGravity();
+	btDiscreteDynamicsWorld::applyGravity();
 #ifdef BT_USE_VIRTUAL_CLEARFORCES_AND_GRAVITY
-        BT_PROFILE("btMultiBody addGravity");
-        for (int i=0;i<this->m_multiBodies.size();i++)
-        {
-                btMultiBody* bod = m_multiBodies[i];
+	BT_PROFILE("btMultiBody addGravity");
+	for (int i = 0; i < this->m_multiBodies.size(); i++)
+	{
+		btMultiBody* bod = m_multiBodies[i];
 
-                bool isSleeping = false;
+		bool isSleeping = false;
 
-                if (bod->getBaseCollider() && bod->getBaseCollider()->getActivationState() == ISLAND_SLEEPING)
-                {
-                        isSleeping = true;
-                }
-                for (int b=0;b<bod->getNumLinks();b++)
-                {
-                        if (bod->getLink(b).m_collider && bod->getLink(b).m_collider->getActivationState()==ISLAND_SLEEPING)
-                                isSleeping = true;
-                }
+		if (bod->getBaseCollider() && bod->getBaseCollider()->getActivationState() == ISLAND_SLEEPING)
+		{
+			isSleeping = true;
+		}
+		for (int b = 0; b < bod->getNumLinks(); b++)
+		{
+			if (bod->getLink(b).m_collider && bod->getLink(b).m_collider->getActivationState() == ISLAND_SLEEPING)
+				isSleeping = true;
+		}
 
-                if (!isSleeping)
-                {
-                        bod->addBaseForce(m_gravity * bod->getBaseMass());
+		if (!isSleeping)
+		{
+			bod->addBaseForce(m_gravity * bod->getBaseMass());
 
-                        for (int j = 0; j < bod->getNumLinks(); ++j)
-                        {
-                                bod->addLinkForce(j, m_gravity * bod->getLinkMass(j));
-                        }
-                }//if (!isSleeping)
-        }
-#endif //BT_USE_VIRTUAL_CLEARFORCES_AND_GRAVITY
+			for (int j = 0; j < bod->getNumLinks(); ++j)
+			{
+				bod->addLinkForce(j, m_gravity * bod->getLinkMass(j));
+			}
+		}  //if (!isSleeping)
+	}
+#endif  //BT_USE_VIRTUAL_CLEARFORCES_AND_GRAVITY
 }
 
 void btMultiBodyDynamicsWorld::clearMultiBodyConstraintForces()
-{ 
-  for (int i=0;i<this->m_multiBodies.size();i++)
-                {       
-                        btMultiBody* bod = m_multiBodies[i];
-			bod->clearConstraintForces();
-                  } 
+{
+	for (int i = 0; i < this->m_multiBodies.size(); i++)
+	{
+		btMultiBody* bod = m_multiBodies[i];
+		bod->clearConstraintForces();
+	}
 }
 void btMultiBodyDynamicsWorld::clearMultiBodyForces()
 {
-              {
-               // BT_PROFILE("clearMultiBodyForces");
-                for (int i=0;i<this->m_multiBodies.size();i++)
-                {
-                        btMultiBody* bod = m_multiBodies[i];
+	{
+		// BT_PROFILE("clearMultiBodyForces");
+		for (int i = 0; i < this->m_multiBodies.size(); i++)
+		{
+			btMultiBody* bod = m_multiBodies[i];
 
-                        bool isSleeping = false;
+			bool isSleeping = false;
 
-                        if (bod->getBaseCollider() && bod->getBaseCollider()->getActivationState() == ISLAND_SLEEPING)
-                        {       
-                                isSleeping = true;
-                        }
-                        for (int b=0;b<bod->getNumLinks();b++)
-                        {       
-                                if (bod->getLink(b).m_collider && bod->getLink(b).m_collider->getActivationState()==ISLAND_SLEEPING)     
-                                        isSleeping = true;
-                        }
+			if (bod->getBaseCollider() && bod->getBaseCollider()->getActivationState() == ISLAND_SLEEPING)
+			{
+				isSleeping = true;
+			}
+			for (int b = 0; b < bod->getNumLinks(); b++)
+			{
+				if (bod->getLink(b).m_collider && bod->getLink(b).m_collider->getActivationState() == ISLAND_SLEEPING)
+					isSleeping = true;
+			}
 
-                        if (!isSleeping)
-                        {
-                        btMultiBody* bod = m_multiBodies[i];
-                        bod->clearForcesAndTorques();
-                	}
+			if (!isSleeping)
+			{
+				btMultiBody* bod = m_multiBodies[i];
+				bod->clearForcesAndTorques();
+			}
 		}
 	}
-
 }
 void btMultiBodyDynamicsWorld::clearForces()
 {
-        btDiscreteDynamicsWorld::clearForces();
+	btDiscreteDynamicsWorld::clearForces();
 
 #ifdef BT_USE_VIRTUAL_CLEARFORCES_AND_GRAVITY
 	clearMultiBodyForces();
 #endif
 }
 
-
-
-
-void	btMultiBodyDynamicsWorld::serialize(btSerializer* serializer)
+void btMultiBodyDynamicsWorld::serialize(btSerializer* serializer)
 {
-
 	serializer->startSerialization();
 
-	serializeDynamicsWorldInfo( serializer);
+	serializeDynamicsWorldInfo(serializer);
 
 	serializeMultiBodies(serializer);
 
@@ -1000,32 +978,31 @@ void	btMultiBodyDynamicsWorld::serialize(btSerializer* serializer)
 	serializer->finishSerialization();
 }
 
-void	btMultiBodyDynamicsWorld::serializeMultiBodies(btSerializer* serializer)
+void btMultiBodyDynamicsWorld::serializeMultiBodies(btSerializer* serializer)
 {
 	int i;
 	//serialize all collision objects
-	for (i=0;i<m_multiBodies.size();i++)
+	for (i = 0; i < m_multiBodies.size(); i++)
 	{
 		btMultiBody* mb = m_multiBodies[i];
 		{
 			int len = mb->calculateSerializeBufferSize();
-			btChunk* chunk = serializer->allocate(len,1);
+			btChunk* chunk = serializer->allocate(len, 1);
 			const char* structType = mb->serialize(chunk->m_oldPtr, serializer);
-			serializer->finalizeChunk(chunk,structType,BT_MULTIBODY_CODE,mb);
+			serializer->finalizeChunk(chunk, structType, BT_MULTIBODY_CODE, mb);
 		}
 	}
 
 	//serialize all multibody links (collision objects)
-	for (i=0;i<m_collisionObjects.size();i++)
+	for (i = 0; i < m_collisionObjects.size(); i++)
 	{
 		btCollisionObject* colObj = m_collisionObjects[i];
 		if (colObj->getInternalType() == btCollisionObject::CO_FEATHERSTONE_LINK)
 		{
 			int len = colObj->calculateSerializeBufferSize();
-			btChunk* chunk = serializer->allocate(len,1);
+			btChunk* chunk = serializer->allocate(len, 1);
 			const char* structType = colObj->serialize(chunk->m_oldPtr, serializer);
-			serializer->finalizeChunk(chunk,structType,BT_MB_LINKCOLLIDER_CODE,colObj);
+			serializer->finalizeChunk(chunk, structType, BT_MB_LINKCOLLIDER_CODE, colObj);
 		}
 	}
-
 }
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyDynamicsWorld.h b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyDynamicsWorld.h
index 2fbf089d810..641238f3bb9 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyDynamicsWorld.h
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyDynamicsWorld.h
@@ -33,8 +33,8 @@ protected:
 	btAlignedObjectArray<btMultiBody*> m_multiBodies;
 	btAlignedObjectArray<btMultiBodyConstraint*> m_multiBodyConstraints;
 	btAlignedObjectArray<btMultiBodyConstraint*> m_sortedMultiBodyConstraints;
-	btMultiBodyConstraintSolver*	m_multiBodyConstraintSolver;
-	MultiBodyInplaceSolverIslandCallback*	m_solverMultiBodyIslandCallback;
+	btMultiBodyConstraintSolver* m_multiBodyConstraintSolver;
+	MultiBodyInplaceSolverIslandCallback* m_solverMultiBodyIslandCallback;
 
 	//cached data to avoid memory allocations
 	btAlignedObjectArray<btQuaternion> m_scratch_world_to_local;
@@ -45,72 +45,69 @@ protected:
 	btAlignedObjectArray<btVector3> m_scratch_v;
 	btAlignedObjectArray<btMatrix3x3> m_scratch_m;
 
-	
-	virtual void	calculateSimulationIslands();
-	virtual void	updateActivationState(btScalar timeStep);
-	virtual void	solveConstraints(btContactSolverInfo& solverInfo);
-	
-	virtual void	serializeMultiBodies(btSerializer* serializer);
+	virtual void calculateSimulationIslands();
+	virtual void updateActivationState(btScalar timeStep);
+	virtual void solveConstraints(btContactSolverInfo& solverInfo);
+
+	virtual void serializeMultiBodies(btSerializer* serializer);
 
 public:
+	btMultiBodyDynamicsWorld(btDispatcher* dispatcher, btBroadphaseInterface* pairCache, btMultiBodyConstraintSolver* constraintSolver, btCollisionConfiguration* collisionConfiguration);
 
-	btMultiBodyDynamicsWorld(btDispatcher* dispatcher,btBroadphaseInterface* pairCache,btMultiBodyConstraintSolver* constraintSolver,btCollisionConfiguration* collisionConfiguration);
+	virtual ~btMultiBodyDynamicsWorld();
 
-	virtual ~btMultiBodyDynamicsWorld ();
+	virtual void addMultiBody(btMultiBody* body, int group = btBroadphaseProxy::DefaultFilter, int mask = btBroadphaseProxy::AllFilter);
 
-	virtual void	addMultiBody(btMultiBody* body, int group= btBroadphaseProxy::DefaultFilter, int mask=btBroadphaseProxy::AllFilter);
+	virtual void removeMultiBody(btMultiBody* body);
 
-	virtual void	removeMultiBody(btMultiBody* body);
-
-	virtual int		getNumMultibodies() const
+	virtual int getNumMultibodies() const
 	{
 		return m_multiBodies.size();
 	}
 
-	btMultiBody*	getMultiBody(int mbIndex)
+	btMultiBody* getMultiBody(int mbIndex)
 	{
 		return m_multiBodies[mbIndex];
 	}
 
-	const btMultiBody*	getMultiBody(int mbIndex) const
+	const btMultiBody* getMultiBody(int mbIndex) const
 	{
 		return m_multiBodies[mbIndex];
 	}
 
-	virtual void	addMultiBodyConstraint( btMultiBodyConstraint* constraint);
+	virtual void addMultiBodyConstraint(btMultiBodyConstraint* constraint);
 
-	virtual int     getNumMultiBodyConstraints() const
+	virtual int getNumMultiBodyConstraints() const
 	{
-        return m_multiBodyConstraints.size();
+		return m_multiBodyConstraints.size();
 	}
 
-	virtual btMultiBodyConstraint*	getMultiBodyConstraint( int constraintIndex)
+	virtual btMultiBodyConstraint* getMultiBodyConstraint(int constraintIndex)
 	{
-        return m_multiBodyConstraints[constraintIndex];
+		return m_multiBodyConstraints[constraintIndex];
 	}
 
-	virtual const btMultiBodyConstraint*	getMultiBodyConstraint( int constraintIndex) const
+	virtual const btMultiBodyConstraint* getMultiBodyConstraint(int constraintIndex) const
 	{
-        return m_multiBodyConstraints[constraintIndex];
+		return m_multiBodyConstraints[constraintIndex];
 	}
 
-	virtual void	removeMultiBodyConstraint( btMultiBodyConstraint* constraint);
+	virtual void removeMultiBodyConstraint(btMultiBodyConstraint* constraint);
 
-	virtual void	integrateTransforms(btScalar timeStep);
+	virtual void integrateTransforms(btScalar timeStep);
 
-	virtual void	debugDrawWorld();
+	virtual void debugDrawWorld();
 
-	virtual void	debugDrawMultiBodyConstraint(btMultiBodyConstraint* constraint);
-	
-	void	forwardKinematics();
+	virtual void debugDrawMultiBodyConstraint(btMultiBodyConstraint* constraint);
+
+	void forwardKinematics();
 	virtual void clearForces();
 	virtual void clearMultiBodyConstraintForces();
 	virtual void clearMultiBodyForces();
 	virtual void applyGravity();
-	
-	virtual	void	serialize(btSerializer* serializer);
-	virtual void	setMultiBodyConstraintSolver(btMultiBodyConstraintSolver* solver);
-    virtual void    setConstraintSolver(btConstraintSolver* solver);
 
+	virtual void serialize(btSerializer* serializer);
+	virtual void setMultiBodyConstraintSolver(btMultiBodyConstraintSolver* solver);
+	virtual void setConstraintSolver(btConstraintSolver* solver);
 };
-#endif //BT_MULTIBODY_DYNAMICS_WORLD_H
+#endif  //BT_MULTIBODY_DYNAMICS_WORLD_H
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyFixedConstraint.cpp b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyFixedConstraint.cpp
index af48e94a83a..5ef9444c2f1 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyFixedConstraint.cpp
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyFixedConstraint.cpp
@@ -24,27 +24,27 @@ subject to the following restrictions:
 #define BTMBFIXEDCONSTRAINT_DIM 6
 
 btMultiBodyFixedConstraint::btMultiBodyFixedConstraint(btMultiBody* body, int link, btRigidBody* bodyB, const btVector3& pivotInA, const btVector3& pivotInB, const btMatrix3x3& frameInA, const btMatrix3x3& frameInB)
-	:btMultiBodyConstraint(body,0,link,-1,BTMBFIXEDCONSTRAINT_DIM,false),
-	m_rigidBodyA(0),
-	m_rigidBodyB(bodyB),
-	m_pivotInA(pivotInA),
-	m_pivotInB(pivotInB),
-    m_frameInA(frameInA),
-    m_frameInB(frameInB)
+	: btMultiBodyConstraint(body, 0, link, -1, BTMBFIXEDCONSTRAINT_DIM, false),
+	  m_rigidBodyA(0),
+	  m_rigidBodyB(bodyB),
+	  m_pivotInA(pivotInA),
+	  m_pivotInB(pivotInB),
+	  m_frameInA(frameInA),
+	  m_frameInB(frameInB)
 {
-    m_data.resize(BTMBFIXEDCONSTRAINT_DIM);//at least store the applied impulses
+	m_data.resize(BTMBFIXEDCONSTRAINT_DIM);  //at least store the applied impulses
 }
 
 btMultiBodyFixedConstraint::btMultiBodyFixedConstraint(btMultiBody* bodyA, int linkA, btMultiBody* bodyB, int linkB, const btVector3& pivotInA, const btVector3& pivotInB, const btMatrix3x3& frameInA, const btMatrix3x3& frameInB)
-	:btMultiBodyConstraint(bodyA,bodyB,linkA,linkB,BTMBFIXEDCONSTRAINT_DIM,false),
-	m_rigidBodyA(0),
-	m_rigidBodyB(0),
-	m_pivotInA(pivotInA),
-	m_pivotInB(pivotInB),
-    m_frameInA(frameInA),
-    m_frameInB(frameInB)
+	: btMultiBodyConstraint(bodyA, bodyB, linkA, linkB, BTMBFIXEDCONSTRAINT_DIM, false),
+	  m_rigidBodyA(0),
+	  m_rigidBodyB(0),
+	  m_pivotInA(pivotInA),
+	  m_pivotInB(pivotInB),
+	  m_frameInA(frameInA),
+	  m_frameInB(frameInB)
 {
-    m_data.resize(BTMBFIXEDCONSTRAINT_DIM);//at least store the applied impulses
+	m_data.resize(BTMBFIXEDCONSTRAINT_DIM);  //at least store the applied impulses
 }
 
 void btMultiBodyFixedConstraint::finalizeMultiDof()
@@ -57,7 +57,6 @@ btMultiBodyFixedConstraint::~btMultiBodyFixedConstraint()
 {
 }
 
-
 int btMultiBodyFixedConstraint::getIslandIdA() const
 {
 	if (m_rigidBodyA)
@@ -103,82 +102,83 @@ int btMultiBodyFixedConstraint::getIslandIdB() const
 
 void btMultiBodyFixedConstraint::createConstraintRows(btMultiBodyConstraintArray& constraintRows, btMultiBodyJacobianData& data, const btContactSolverInfo& infoGlobal)
 {
-    int numDim = BTMBFIXEDCONSTRAINT_DIM;
-    for (int i=0;i<numDim;i++)
+	int numDim = BTMBFIXEDCONSTRAINT_DIM;
+	for (int i = 0; i < numDim; i++)
 	{
-        btMultiBodySolverConstraint& constraintRow = constraintRows.expandNonInitializing();
-        constraintRow.m_orgConstraint = this;
-        constraintRow.m_orgDofIndex = i;
-        constraintRow.m_relpos1CrossNormal.setValue(0,0,0);
-        constraintRow.m_contactNormal1.setValue(0,0,0);
-        constraintRow.m_relpos2CrossNormal.setValue(0,0,0);
-        constraintRow.m_contactNormal2.setValue(0,0,0);
-        constraintRow.m_angularComponentA.setValue(0,0,0);
-        constraintRow.m_angularComponentB.setValue(0,0,0);
-        
-        constraintRow.m_solverBodyIdA = data.m_fixedBodyId;
-        constraintRow.m_solverBodyIdB = data.m_fixedBodyId;
-        
-        // Convert local points back to world
-        btVector3 pivotAworld = m_pivotInA;
-        btMatrix3x3 frameAworld = m_frameInA;
-        if (m_rigidBodyA)
-        {
-            
-            constraintRow.m_solverBodyIdA = m_rigidBodyA->getCompanionId();
-            pivotAworld = m_rigidBodyA->getCenterOfMassTransform()*m_pivotInA;
-            frameAworld = frameAworld.transpose()*btMatrix3x3(m_rigidBodyA->getOrientation());
-            
-        } else
-        {
-            if (m_bodyA) {
-                pivotAworld = m_bodyA->localPosToWorld(m_linkA, m_pivotInA);
-                frameAworld = m_bodyA->localFrameToWorld(m_linkA, frameAworld);
-            }
-        }
-        btVector3 pivotBworld = m_pivotInB;
-        btMatrix3x3 frameBworld = m_frameInB;
-        if (m_rigidBodyB)
-        {
-            constraintRow.m_solverBodyIdB = m_rigidBodyB->getCompanionId();
-            pivotBworld = m_rigidBodyB->getCenterOfMassTransform()*m_pivotInB;
-            frameBworld = frameBworld.transpose()*btMatrix3x3(m_rigidBodyB->getOrientation());
-             
-        } else
-        {
-            if (m_bodyB) {
-                pivotBworld = m_bodyB->localPosToWorld(m_linkB, m_pivotInB);
-                frameBworld = m_bodyB->localFrameToWorld(m_linkB, frameBworld);
-            }
-        }
-        
-        btMatrix3x3 relRot = frameAworld.inverse()*frameBworld;
-        btVector3 angleDiff;
-        btGeneric6DofSpring2Constraint::matrixToEulerXYZ(relRot,angleDiff);
-        
-        btVector3 constraintNormalLin(0,0,0);
-        btVector3 constraintNormalAng(0,0,0);
-        btScalar posError = 0.0;
-        if (i < 3) {
-            constraintNormalLin[i] = 1;
-            posError = (pivotAworld-pivotBworld).dot(constraintNormalLin);
-            fillMultiBodyConstraint(constraintRow, data, 0, 0, constraintNormalAng,
-                                    constraintNormalLin, pivotAworld, pivotBworld,
-                                    posError,
-                                    infoGlobal,
-                                    -m_maxAppliedImpulse, m_maxAppliedImpulse
-                                    );
-        }
-        else { //i>=3
-            constraintNormalAng = frameAworld.getColumn(i%3);
-            posError = angleDiff[i%3];
-            fillMultiBodyConstraint(constraintRow, data, 0, 0, constraintNormalAng,
-                                    constraintNormalLin, pivotAworld, pivotBworld,
-                                    posError,
-                                    infoGlobal,
-                                    -m_maxAppliedImpulse, m_maxAppliedImpulse, true
-                                    );
-        }
+		btMultiBodySolverConstraint& constraintRow = constraintRows.expandNonInitializing();
+		constraintRow.m_orgConstraint = this;
+		constraintRow.m_orgDofIndex = i;
+		constraintRow.m_relpos1CrossNormal.setValue(0, 0, 0);
+		constraintRow.m_contactNormal1.setValue(0, 0, 0);
+		constraintRow.m_relpos2CrossNormal.setValue(0, 0, 0);
+		constraintRow.m_contactNormal2.setValue(0, 0, 0);
+		constraintRow.m_angularComponentA.setValue(0, 0, 0);
+		constraintRow.m_angularComponentB.setValue(0, 0, 0);
+
+		constraintRow.m_solverBodyIdA = data.m_fixedBodyId;
+		constraintRow.m_solverBodyIdB = data.m_fixedBodyId;
+
+		// Convert local points back to world
+		btVector3 pivotAworld = m_pivotInA;
+		btMatrix3x3 frameAworld = m_frameInA;
+		if (m_rigidBodyA)
+		{
+			constraintRow.m_solverBodyIdA = m_rigidBodyA->getCompanionId();
+			pivotAworld = m_rigidBodyA->getCenterOfMassTransform() * m_pivotInA;
+			frameAworld = frameAworld.transpose() * btMatrix3x3(m_rigidBodyA->getOrientation());
+		}
+		else
+		{
+			if (m_bodyA)
+			{
+				pivotAworld = m_bodyA->localPosToWorld(m_linkA, m_pivotInA);
+				frameAworld = m_bodyA->localFrameToWorld(m_linkA, frameAworld);
+			}
+		}
+		btVector3 pivotBworld = m_pivotInB;
+		btMatrix3x3 frameBworld = m_frameInB;
+		if (m_rigidBodyB)
+		{
+			constraintRow.m_solverBodyIdB = m_rigidBodyB->getCompanionId();
+			pivotBworld = m_rigidBodyB->getCenterOfMassTransform() * m_pivotInB;
+			frameBworld = frameBworld.transpose() * btMatrix3x3(m_rigidBodyB->getOrientation());
+		}
+		else
+		{
+			if (m_bodyB)
+			{
+				pivotBworld = m_bodyB->localPosToWorld(m_linkB, m_pivotInB);
+				frameBworld = m_bodyB->localFrameToWorld(m_linkB, frameBworld);
+			}
+		}
+
+		btMatrix3x3 relRot = frameAworld.inverse() * frameBworld;
+		btVector3 angleDiff;
+		btGeneric6DofSpring2Constraint::matrixToEulerXYZ(relRot, angleDiff);
+
+		btVector3 constraintNormalLin(0, 0, 0);
+		btVector3 constraintNormalAng(0, 0, 0);
+		btScalar posError = 0.0;
+		if (i < 3)
+		{
+			constraintNormalLin[i] = 1;
+			posError = (pivotAworld - pivotBworld).dot(constraintNormalLin);
+			fillMultiBodyConstraint(constraintRow, data, 0, 0, constraintNormalAng,
+									constraintNormalLin, pivotAworld, pivotBworld,
+									posError,
+									infoGlobal,
+									-m_maxAppliedImpulse, m_maxAppliedImpulse);
+		}
+		else
+		{  //i>=3
+			constraintNormalAng = frameAworld.getColumn(i % 3);
+			posError = angleDiff[i % 3];
+			fillMultiBodyConstraint(constraintRow, data, 0, 0, constraintNormalAng,
+									constraintNormalLin, pivotAworld, pivotBworld,
+									posError,
+									infoGlobal,
+									-m_maxAppliedImpulse, m_maxAppliedImpulse, true);
+		}
 	}
 }
 
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyFixedConstraint.h b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyFixedConstraint.h
index 036025136ed..adb1cb47dad 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyFixedConstraint.h
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyFixedConstraint.h
@@ -23,16 +23,14 @@ subject to the following restrictions:
 class btMultiBodyFixedConstraint : public btMultiBodyConstraint
 {
 protected:
-
-	btRigidBody*	m_rigidBodyA;
-	btRigidBody*	m_rigidBodyB;
-	btVector3		m_pivotInA;
-	btVector3		m_pivotInB;
-    btMatrix3x3     m_frameInA;
-    btMatrix3x3     m_frameInB;
+	btRigidBody* m_rigidBodyA;
+	btRigidBody* m_rigidBodyB;
+	btVector3 m_pivotInA;
+	btVector3 m_pivotInB;
+	btMatrix3x3 m_frameInA;
+	btMatrix3x3 m_frameInB;
 
 public:
-
 	btMultiBodyFixedConstraint(btMultiBody* body, int link, btRigidBody* bodyB, const btVector3& pivotInA, const btVector3& pivotInB, const btMatrix3x3& frameInA, const btMatrix3x3& frameInB);
 	btMultiBodyFixedConstraint(btMultiBody* bodyA, int linkA, btMultiBody* bodyB, int linkB, const btVector3& pivotInA, const btVector3& pivotInB, const btMatrix3x3& frameInA, const btMatrix3x3& frameInB);
 
@@ -44,18 +42,18 @@ public:
 	virtual int getIslandIdB() const;
 
 	virtual void createConstraintRows(btMultiBodyConstraintArray& constraintRows,
-		btMultiBodyJacobianData& data,
-		const btContactSolverInfo& infoGlobal);
+									  btMultiBodyJacobianData& data,
+									  const btContactSolverInfo& infoGlobal);
 
-    const btVector3& getPivotInA() const
-    {
-        return m_pivotInA;
-    }
-    
-    void setPivotInA(const btVector3& pivotInA)
-    {
-        m_pivotInA = pivotInA;
-    }
+	const btVector3& getPivotInA() const
+	{
+		return m_pivotInA;
+	}
+
+	void setPivotInA(const btVector3& pivotInA)
+	{
+		m_pivotInA = pivotInA;
+	}
 
 	const btVector3& getPivotInB() const
 	{
@@ -66,29 +64,28 @@ public:
 	{
 		m_pivotInB = pivotInB;
 	}
-    
-    const btMatrix3x3& getFrameInA() const
-    {
-        return m_frameInA;
-    }
-    
-    void setFrameInA(const btMatrix3x3& frameInA)
-    {
-        m_frameInA = frameInA;
-    }
-    
-    const btMatrix3x3& getFrameInB() const
-    {
-        return m_frameInB;
-    }
-    
-    virtual void setFrameInB(const btMatrix3x3& frameInB)
-    {
-        m_frameInB = frameInB;
-    }
+
+	const btMatrix3x3& getFrameInA() const
+	{
+		return m_frameInA;
+	}
+
+	void setFrameInA(const btMatrix3x3& frameInA)
+	{
+		m_frameInA = frameInA;
+	}
+
+	const btMatrix3x3& getFrameInB() const
+	{
+		return m_frameInB;
+	}
+
+	virtual void setFrameInB(const btMatrix3x3& frameInB)
+	{
+		m_frameInB = frameInB;
+	}
 
 	virtual void debugDraw(class btIDebugDraw* drawer);
-
 };
 
-#endif //BT_MULTIBODY_FIXED_CONSTRAINT_H
+#endif  //BT_MULTIBODY_FIXED_CONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyGearConstraint.cpp b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyGearConstraint.cpp
index 09ddd65cd86..bf6b811d26e 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyGearConstraint.cpp
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyGearConstraint.cpp
@@ -21,20 +21,18 @@ subject to the following restrictions:
 #include "BulletCollision/CollisionDispatch/btCollisionObject.h"
 
 btMultiBodyGearConstraint::btMultiBodyGearConstraint(btMultiBody* bodyA, int linkA, btMultiBody* bodyB, int linkB, const btVector3& pivotInA, const btVector3& pivotInB, const btMatrix3x3& frameInA, const btMatrix3x3& frameInB)
-	:btMultiBodyConstraint(bodyA,bodyB,linkA,linkB,1,false),
-	m_gearRatio(1),
-	m_gearAuxLink(-1),
-	m_erp(0),
-	m_relativePositionTarget(0)
+	: btMultiBodyConstraint(bodyA, bodyB, linkA, linkB, 1, false),
+	  m_gearRatio(1),
+	  m_gearAuxLink(-1),
+	  m_erp(0),
+	  m_relativePositionTarget(0)
 {
-   
 }
 
 void btMultiBodyGearConstraint::finalizeMultiDof()
 {
-	
 	allocateJacobiansMultiDof();
-	
+
 	m_numDofsFinalized = m_jacSizeBoth;
 }
 
@@ -42,7 +40,6 @@ btMultiBodyGearConstraint::~btMultiBodyGearConstraint()
 {
 }
 
-
 int btMultiBodyGearConstraint::getIslandIdA() const
 {
 	if (m_bodyA)
@@ -81,27 +78,25 @@ int btMultiBodyGearConstraint::getIslandIdB() const
 	return -1;
 }
 
-
 void btMultiBodyGearConstraint::createConstraintRows(btMultiBodyConstraintArray& constraintRows,
-		btMultiBodyJacobianData& data,
-		const btContactSolverInfo& infoGlobal)
+													 btMultiBodyJacobianData& data,
+													 const btContactSolverInfo& infoGlobal)
 {
-    // only positions need to be updated -- data.m_jacobians and force
-    // directions were set in the ctor and never change.
-	
+	// only positions need to be updated -- data.m_jacobians and force
+	// directions were set in the ctor and never change.
+
 	if (m_numDofsFinalized != m_jacSizeBoth)
 	{
-        finalizeMultiDof();
+		finalizeMultiDof();
 	}
 
 	//don't crash
 	if (m_numDofsFinalized != m_jacSizeBoth)
 		return;
 
-	
-	if (m_maxAppliedImpulse==0.f)
+	if (m_maxAppliedImpulse == 0.f)
 		return;
-	
+
 	// note: we rely on the fact that data.m_jacobians are
 	// always initialized to zero by the Constraint ctor
 	int linkDoF = 0;
@@ -114,67 +109,66 @@ void btMultiBodyGearConstraint::createConstraintRows(btMultiBodyConstraintArray&
 
 	btScalar posError = 0;
 	const btVector3 dummy(0, 0, 0);
-	
+
 	btScalar kp = 1;
 	btScalar kd = 1;
 	int numRows = getNumRows();
 
-	for (int row=0;row<numRows;row++)
+	for (int row = 0; row < numRows; row++)
 	{
 		btMultiBodySolverConstraint& constraintRow = constraintRows.expandNonInitializing();
 
-
-        int dof = 0;
-        btScalar currentPosition = m_bodyA->getJointPosMultiDof(m_linkA)[dof];
-        btScalar currentVelocity = m_bodyA->getJointVelMultiDof(m_linkA)[dof];
+		int dof = 0;
+		btScalar currentPosition = m_bodyA->getJointPosMultiDof(m_linkA)[dof];
+		btScalar currentVelocity = m_bodyA->getJointVelMultiDof(m_linkA)[dof];
 		btScalar auxVel = 0;
-		
-		if (m_gearAuxLink>=0)
+
+		if (m_gearAuxLink >= 0)
 		{
 			auxVel = m_bodyA->getJointVelMultiDof(m_gearAuxLink)[dof];
 		}
 		currentVelocity += auxVel;
-		if (m_erp!=0)
+		if (m_erp != 0)
 		{
 			btScalar currentPositionA = m_bodyA->getJointPosMultiDof(m_linkA)[dof];
 			if (m_gearAuxLink >= 0)
 			{
 				currentPositionA -= m_bodyA->getJointPosMultiDof(m_gearAuxLink)[dof];
 			}
-			btScalar currentPositionB = m_gearRatio*m_bodyA->getJointPosMultiDof(m_linkB)[dof];
-			btScalar diff = currentPositionB+currentPositionA;
+			btScalar currentPositionB = m_gearRatio * m_bodyA->getJointPosMultiDof(m_linkB)[dof];
+			btScalar diff = currentPositionB + currentPositionA;
 			btScalar desiredPositionDiff = this->m_relativePositionTarget;
-			posError = -m_erp*(desiredPositionDiff - diff);
+			posError = -m_erp * (desiredPositionDiff - diff);
 		}
-		
-        btScalar desiredRelativeVelocity =   auxVel;
-    
-		fillMultiBodyConstraint(constraintRow,data,jacobianA(row),jacobianB(row),dummy,dummy,dummy,dummy,posError,infoGlobal,-m_maxAppliedImpulse,m_maxAppliedImpulse,false,1,false,desiredRelativeVelocity);
+
+		btScalar desiredRelativeVelocity = auxVel;
+
+		fillMultiBodyConstraint(constraintRow, data, jacobianA(row), jacobianB(row), dummy, dummy, dummy, dummy, posError, infoGlobal, -m_maxAppliedImpulse, m_maxAppliedImpulse, false, 1, false, desiredRelativeVelocity);
 
 		constraintRow.m_orgConstraint = this;
 		constraintRow.m_orgDofIndex = row;
 		{
 			//expect either prismatic or revolute joint type for now
-			btAssert((m_bodyA->getLink(m_linkA).m_jointType == btMultibodyLink::eRevolute)||(m_bodyA->getLink(m_linkA).m_jointType == btMultibodyLink::ePrismatic));
+			btAssert((m_bodyA->getLink(m_linkA).m_jointType == btMultibodyLink::eRevolute) || (m_bodyA->getLink(m_linkA).m_jointType == btMultibodyLink::ePrismatic));
 			switch (m_bodyA->getLink(m_linkA).m_jointType)
 			{
 				case btMultibodyLink::eRevolute:
 				{
 					constraintRow.m_contactNormal1.setZero();
 					constraintRow.m_contactNormal2.setZero();
-					btVector3 revoluteAxisInWorld = quatRotate(m_bodyA->getLink(m_linkA).m_cachedWorldTransform.getRotation(),m_bodyA->getLink(m_linkA).m_axes[0].m_topVec);
-					constraintRow.m_relpos1CrossNormal=revoluteAxisInWorld;
-					constraintRow.m_relpos2CrossNormal=-revoluteAxisInWorld;
-					
+					btVector3 revoluteAxisInWorld = quatRotate(m_bodyA->getLink(m_linkA).m_cachedWorldTransform.getRotation(), m_bodyA->getLink(m_linkA).m_axes[0].m_topVec);
+					constraintRow.m_relpos1CrossNormal = revoluteAxisInWorld;
+					constraintRow.m_relpos2CrossNormal = -revoluteAxisInWorld;
+
 					break;
 				}
 				case btMultibodyLink::ePrismatic:
 				{
-					btVector3 prismaticAxisInWorld = quatRotate(m_bodyA->getLink(m_linkA).m_cachedWorldTransform.getRotation(),m_bodyA->getLink(m_linkA).m_axes[0].m_bottomVec);
-					constraintRow.m_contactNormal1=prismaticAxisInWorld;
-					constraintRow.m_contactNormal2=-prismaticAxisInWorld;
+					btVector3 prismaticAxisInWorld = quatRotate(m_bodyA->getLink(m_linkA).m_cachedWorldTransform.getRotation(), m_bodyA->getLink(m_linkA).m_axes[0].m_bottomVec);
+					constraintRow.m_contactNormal1 = prismaticAxisInWorld;
+					constraintRow.m_contactNormal2 = -prismaticAxisInWorld;
 					constraintRow.m_relpos1CrossNormal.setZero();
-					constraintRow.m_relpos2CrossNormal.setZero();					
+					constraintRow.m_relpos2CrossNormal.setZero();
 					break;
 				}
 				default:
@@ -182,10 +176,6 @@ void btMultiBodyGearConstraint::createConstraintRows(btMultiBodyConstraintArray&
 					btAssert(0);
 				}
 			};
-			
 		}
-
 	}
-
 }
-
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyGearConstraint.h b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyGearConstraint.h
index 0115de6241c..31888fbc682 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyGearConstraint.h
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyGearConstraint.h
@@ -23,20 +23,18 @@ subject to the following restrictions:
 class btMultiBodyGearConstraint : public btMultiBodyConstraint
 {
 protected:
+	btRigidBody* m_rigidBodyA;
+	btRigidBody* m_rigidBodyB;
+	btVector3 m_pivotInA;
+	btVector3 m_pivotInB;
+	btMatrix3x3 m_frameInA;
+	btMatrix3x3 m_frameInB;
+	btScalar m_gearRatio;
+	int m_gearAuxLink;
+	btScalar m_erp;
+	btScalar m_relativePositionTarget;
 
-	btRigidBody*	m_rigidBodyA;
-	btRigidBody*	m_rigidBodyB;
-	btVector3		m_pivotInA;
-	btVector3		m_pivotInB;
-	btMatrix3x3     m_frameInA;
-	btMatrix3x3     m_frameInB;
-	btScalar			m_gearRatio;
-	int			m_gearAuxLink;
-	btScalar	m_erp;
-	btScalar	m_relativePositionTarget;
-	
 public:
-
 	//btMultiBodyGearConstraint(btMultiBody* body, int link, btRigidBody* bodyB, const btVector3& pivotInA, const btVector3& pivotInB, const btMatrix3x3& frameInA, const btMatrix3x3& frameInB);
 	btMultiBodyGearConstraint(btMultiBody* bodyA, int linkA, btMultiBody* bodyB, int linkB, const btVector3& pivotInA, const btVector3& pivotInB, const btMatrix3x3& frameInA, const btMatrix3x3& frameInB);
 
@@ -48,18 +46,18 @@ public:
 	virtual int getIslandIdB() const;
 
 	virtual void createConstraintRows(btMultiBodyConstraintArray& constraintRows,
-		btMultiBodyJacobianData& data,
-		const btContactSolverInfo& infoGlobal);
+									  btMultiBodyJacobianData& data,
+									  const btContactSolverInfo& infoGlobal);
 
-    const btVector3& getPivotInA() const
-    {
-        return m_pivotInA;
-    }
-    
-    void setPivotInA(const btVector3& pivotInA)
-    {
-        m_pivotInA = pivotInA;
-    }
+	const btVector3& getPivotInA() const
+	{
+		return m_pivotInA;
+	}
+
+	void setPivotInA(const btVector3& pivotInA)
+	{
+		m_pivotInA = pivotInA;
+	}
 
 	const btVector3& getPivotInB() const
 	{
@@ -70,32 +68,32 @@ public:
 	{
 		m_pivotInB = pivotInB;
 	}
-    
-    const btMatrix3x3& getFrameInA() const
-    {
-        return m_frameInA;
-    }
-    
-    void setFrameInA(const btMatrix3x3& frameInA)
-    {
-        m_frameInA = frameInA;
-    }
-    
-    const btMatrix3x3& getFrameInB() const
-    {
-        return m_frameInB;
-    }
-    
-    virtual void setFrameInB(const btMatrix3x3& frameInB)
-    {
-        m_frameInB = frameInB;
-    }
+
+	const btMatrix3x3& getFrameInA() const
+	{
+		return m_frameInA;
+	}
+
+	void setFrameInA(const btMatrix3x3& frameInA)
+	{
+		m_frameInA = frameInA;
+	}
+
+	const btMatrix3x3& getFrameInB() const
+	{
+		return m_frameInB;
+	}
+
+	virtual void setFrameInB(const btMatrix3x3& frameInB)
+	{
+		m_frameInB = frameInB;
+	}
 
 	virtual void debugDraw(class btIDebugDraw* drawer)
 	{
 		//todo(erwincoumans)
 	}
-	
+
 	virtual void setGearRatio(btScalar gearRatio)
 	{
 		m_gearRatio = gearRatio;
@@ -114,4 +112,4 @@ public:
 	}
 };
 
-#endif //BT_MULTIBODY_GEAR_CONSTRAINT_H
+#endif  //BT_MULTIBODY_GEAR_CONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyJointFeedback.h b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyJointFeedback.h
index 5c2fa8ed5b9..d943019e711 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyJointFeedback.h
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyJointFeedback.h
@@ -12,8 +12,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #ifndef BT_MULTIBODY_JOINT_FEEDBACK_H
 #define BT_MULTIBODY_JOINT_FEEDBACK_H
 
@@ -21,7 +19,7 @@ subject to the following restrictions:
 
 struct btMultiBodyJointFeedback
 {
-	btSpatialForceVector	m_reactionForces;
+	btSpatialForceVector m_reactionForces;
 };
 
-#endif //BT_MULTIBODY_JOINT_FEEDBACK_H
+#endif  //BT_MULTIBODY_JOINT_FEEDBACK_H
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyJointLimitConstraint.cpp b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyJointLimitConstraint.cpp
index 35c929f7cec..8791ad28688 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyJointLimitConstraint.cpp
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyJointLimitConstraint.cpp
@@ -20,21 +20,18 @@ subject to the following restrictions:
 #include "btMultiBodyLinkCollider.h"
 #include "BulletCollision/CollisionDispatch/btCollisionObject.h"
 
-
-
 btMultiBodyJointLimitConstraint::btMultiBodyJointLimitConstraint(btMultiBody* body, int link, btScalar lower, btScalar upper)
 	//:btMultiBodyConstraint(body,0,link,-1,2,true),
-	:btMultiBodyConstraint(body,body,link,body->getLink(link).m_parent,2,true),
-	m_lowerBound(lower),
-	m_upperBound(upper)
+	: btMultiBodyConstraint(body, body, link, body->getLink(link).m_parent, 2, true),
+	  m_lowerBound(lower),
+	  m_upperBound(upper)
 {
-
 }
 
 void btMultiBodyJointLimitConstraint::finalizeMultiDof()
 {
 	// the data.m_jacobians never change, so may as well
-    // initialize them here
+	// initialize them here
 
 	allocateJacobiansMultiDof();
 
@@ -53,10 +50,8 @@ btMultiBodyJointLimitConstraint::~btMultiBodyJointLimitConstraint()
 {
 }
 
-
 int btMultiBodyJointLimitConstraint::getIslandIdA() const
 {
-
 	if (m_bodyA)
 	{
 		if (m_linkA < 0)
@@ -93,72 +88,69 @@ int btMultiBodyJointLimitConstraint::getIslandIdB() const
 	return -1;
 }
 
-
 void btMultiBodyJointLimitConstraint::createConstraintRows(btMultiBodyConstraintArray& constraintRows,
-		btMultiBodyJacobianData& data,
-		const btContactSolverInfo& infoGlobal)
+														   btMultiBodyJacobianData& data,
+														   const btContactSolverInfo& infoGlobal)
 {
-	
-    // only positions need to be updated -- data.m_jacobians and force
-    // directions were set in the ctor and never change.
+	// only positions need to be updated -- data.m_jacobians and force
+	// directions were set in the ctor and never change.
 
 	if (m_numDofsFinalized != m_jacSizeBoth)
 	{
-        finalizeMultiDof();
+		finalizeMultiDof();
 	}
 
+	// row 0: the lower bound
+	setPosition(0, m_bodyA->getJointPos(m_linkA) - m_lowerBound);  //multidof: this is joint-type dependent
 
-    // row 0: the lower bound
-    setPosition(0, m_bodyA->getJointPos(m_linkA) - m_lowerBound);			//multidof: this is joint-type dependent
+	// row 1: the upper bound
+	setPosition(1, m_upperBound - m_bodyA->getJointPos(m_linkA));
 
-    // row 1: the upper bound
-    setPosition(1, m_upperBound - m_bodyA->getJointPos(m_linkA));
-	
-	for (int row=0;row<getNumRows();row++)
+	for (int row = 0; row < getNumRows(); row++)
 	{
 		btScalar penetration = getPosition(row);
 
 		//todo: consider adding some safety threshold here
-		if (penetration>0)
+		if (penetration > 0)
 		{
 			continue;
 		}
-		btScalar direction = row? -1 : 1;
+		btScalar direction = row ? -1 : 1;
 
 		btMultiBodySolverConstraint& constraintRow = constraintRows.expandNonInitializing();
-        constraintRow.m_orgConstraint = this;
-        constraintRow.m_orgDofIndex = row;
-        
+		constraintRow.m_orgConstraint = this;
+		constraintRow.m_orgDofIndex = row;
+
 		constraintRow.m_multiBodyA = m_bodyA;
 		constraintRow.m_multiBodyB = m_bodyB;
-		const btScalar posError = 0;						//why assume it's zero?
+		const btScalar posError = 0;  //why assume it's zero?
 		const btVector3 dummy(0, 0, 0);
 
-		btScalar rel_vel = fillMultiBodyConstraint(constraintRow,data,jacobianA(row),jacobianB(row),dummy,dummy,dummy,dummy,posError,infoGlobal,0,m_maxAppliedImpulse);
+		btScalar rel_vel = fillMultiBodyConstraint(constraintRow, data, jacobianA(row), jacobianB(row), dummy, dummy, dummy, dummy, posError, infoGlobal, 0, m_maxAppliedImpulse);
 
 		{
 			//expect either prismatic or revolute joint type for now
-			btAssert((m_bodyA->getLink(m_linkA).m_jointType == btMultibodyLink::eRevolute)||(m_bodyA->getLink(m_linkA).m_jointType == btMultibodyLink::ePrismatic));
+			btAssert((m_bodyA->getLink(m_linkA).m_jointType == btMultibodyLink::eRevolute) || (m_bodyA->getLink(m_linkA).m_jointType == btMultibodyLink::ePrismatic));
 			switch (m_bodyA->getLink(m_linkA).m_jointType)
 			{
 				case btMultibodyLink::eRevolute:
 				{
 					constraintRow.m_contactNormal1.setZero();
 					constraintRow.m_contactNormal2.setZero();
-					btVector3 revoluteAxisInWorld = direction*quatRotate(m_bodyA->getLink(m_linkA).m_cachedWorldTransform.getRotation(),m_bodyA->getLink(m_linkA).m_axes[0].m_topVec);
-					constraintRow.m_relpos1CrossNormal=revoluteAxisInWorld;
-					constraintRow.m_relpos2CrossNormal=-revoluteAxisInWorld;
-					
+					btVector3 revoluteAxisInWorld = direction * quatRotate(m_bodyA->getLink(m_linkA).m_cachedWorldTransform.getRotation(), m_bodyA->getLink(m_linkA).m_axes[0].m_topVec);
+					constraintRow.m_relpos1CrossNormal = revoluteAxisInWorld;
+					constraintRow.m_relpos2CrossNormal = -revoluteAxisInWorld;
+
 					break;
 				}
 				case btMultibodyLink::ePrismatic:
 				{
-					btVector3 prismaticAxisInWorld = direction* quatRotate(m_bodyA->getLink(m_linkA).m_cachedWorldTransform.getRotation(),m_bodyA->getLink(m_linkA).m_axes[0].m_bottomVec);
-					constraintRow.m_contactNormal1=prismaticAxisInWorld;
-					constraintRow.m_contactNormal2=-prismaticAxisInWorld;
+					btVector3 prismaticAxisInWorld = direction * quatRotate(m_bodyA->getLink(m_linkA).m_cachedWorldTransform.getRotation(), m_bodyA->getLink(m_linkA).m_axes[0].m_bottomVec);
+					constraintRow.m_contactNormal1 = prismaticAxisInWorld;
+					constraintRow.m_contactNormal2 = -prismaticAxisInWorld;
 					constraintRow.m_relpos1CrossNormal.setZero();
 					constraintRow.m_relpos2CrossNormal.setZero();
-					
+
 					break;
 				}
 				default:
@@ -166,36 +158,35 @@ void btMultiBodyJointLimitConstraint::createConstraintRows(btMultiBodyConstraint
 					btAssert(0);
 				}
 			};
-			
 		}
 
 		{
-			
 			btScalar positionalError = 0.f;
-			btScalar	velocityError =  - rel_vel;// * damping;
+			btScalar velocityError = -rel_vel;  // * damping;
 			btScalar erp = infoGlobal.m_erp2;
 			if (!infoGlobal.m_splitImpulse || (penetration > infoGlobal.m_splitImpulsePenetrationThreshold))
 			{
 				erp = infoGlobal.m_erp;
 			}
-			if (penetration>0)
+			if (penetration > 0)
 			{
 				positionalError = 0;
 				velocityError = -penetration / infoGlobal.m_timeStep;
-			} else
+			}
+			else
 			{
-				positionalError = -penetration * erp/infoGlobal.m_timeStep;
+				positionalError = -penetration * erp / infoGlobal.m_timeStep;
 			}
 
-			btScalar  penetrationImpulse = positionalError*constraintRow.m_jacDiagABInv;
-			btScalar velocityImpulse = velocityError *constraintRow.m_jacDiagABInv;
+			btScalar penetrationImpulse = positionalError * constraintRow.m_jacDiagABInv;
+			btScalar velocityImpulse = velocityError * constraintRow.m_jacDiagABInv;
 			if (!infoGlobal.m_splitImpulse || (penetration > infoGlobal.m_splitImpulsePenetrationThreshold))
 			{
 				//combine position and velocity into rhs
-				constraintRow.m_rhs = penetrationImpulse+velocityImpulse;
+				constraintRow.m_rhs = penetrationImpulse + velocityImpulse;
 				constraintRow.m_rhsPenetration = 0.f;
-
-			} else
+			}
+			else
 			{
 				//split position and velocity into rhs and m_rhsPenetration
 				constraintRow.m_rhs = velocityImpulse;
@@ -203,9 +194,4 @@ void btMultiBodyJointLimitConstraint::createConstraintRows(btMultiBodyConstraint
 			}
 		}
 	}
-
 }
-
-
-
-
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyJointLimitConstraint.h b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyJointLimitConstraint.h
index 55b8d122b97..6716ba490f8 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyJointLimitConstraint.h
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyJointLimitConstraint.h
@@ -22,11 +22,10 @@ struct btSolverInfo;
 class btMultiBodyJointLimitConstraint : public btMultiBodyConstraint
 {
 protected:
+	btScalar m_lowerBound;
+	btScalar m_upperBound;
 
-	btScalar	m_lowerBound;
-	btScalar	m_upperBound;
 public:
-
 	btMultiBodyJointLimitConstraint(btMultiBody* body, int link, btScalar lower, btScalar upper);
 	virtual ~btMultiBodyJointLimitConstraint();
 
@@ -36,15 +35,13 @@ public:
 	virtual int getIslandIdB() const;
 
 	virtual void createConstraintRows(btMultiBodyConstraintArray& constraintRows,
-		btMultiBodyJacobianData& data,
-		const btContactSolverInfo& infoGlobal);
+									  btMultiBodyJacobianData& data,
+									  const btContactSolverInfo& infoGlobal);
 
 	virtual void debugDraw(class btIDebugDraw* drawer)
 	{
 		//todo(erwincoumans)
 	}
-
 };
 
-#endif //BT_MULTIBODY_JOINT_LIMIT_CONSTRAINT_H
-
+#endif  //BT_MULTIBODY_JOINT_LIMIT_CONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyJointMotor.cpp b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyJointMotor.cpp
index 2a70ea97e57..5c816c49876 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyJointMotor.cpp
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyJointMotor.cpp
@@ -20,22 +20,18 @@ subject to the following restrictions:
 #include "btMultiBodyLinkCollider.h"
 #include "BulletCollision/CollisionDispatch/btCollisionObject.h"
 
-
 btMultiBodyJointMotor::btMultiBodyJointMotor(btMultiBody* body, int link, btScalar desiredVelocity, btScalar maxMotorImpulse)
-	:btMultiBodyConstraint(body,body,link,body->getLink(link).m_parent,1,true),
-	m_desiredVelocity(desiredVelocity),
-	m_desiredPosition(0),
-	m_kd(1.),
-	m_kp(0),
-	m_erp(1),
-	m_rhsClamp(SIMD_INFINITY)
+	: btMultiBodyConstraint(body, body, link, body->getLink(link).m_parent, 1, true),
+	  m_desiredVelocity(desiredVelocity),
+	  m_desiredPosition(0),
+	  m_kd(1.),
+	  m_kp(0),
+	  m_erp(1),
+	  m_rhsClamp(SIMD_INFINITY)
 {
-
 	m_maxAppliedImpulse = maxMotorImpulse;
 	// the data.m_jacobians never change, so may as well
-    // initialize them here
-
-
+	// initialize them here
 }
 
 void btMultiBodyJointMotor::finalizeMultiDof()
@@ -55,18 +51,17 @@ void btMultiBodyJointMotor::finalizeMultiDof()
 
 btMultiBodyJointMotor::btMultiBodyJointMotor(btMultiBody* body, int link, int linkDoF, btScalar desiredVelocity, btScalar maxMotorImpulse)
 	//:btMultiBodyConstraint(body,0,link,-1,1,true),
-	:btMultiBodyConstraint(body,body,link,body->getLink(link).m_parent,1,true),
-	m_desiredVelocity(desiredVelocity),
-	m_desiredPosition(0),
-	m_kd(1.),
-	m_kp(0),
-    m_erp(1),
-	m_rhsClamp(SIMD_INFINITY)
+	: btMultiBodyConstraint(body, body, link, body->getLink(link).m_parent, 1, true),
+	  m_desiredVelocity(desiredVelocity),
+	  m_desiredPosition(0),
+	  m_kd(1.),
+	  m_kp(0),
+	  m_erp(1),
+	  m_rhsClamp(SIMD_INFINITY)
 {
 	btAssert(linkDoF < body->getLink(link).m_dofCount);
 
 	m_maxAppliedImpulse = maxMotorImpulse;
-
 }
 btMultiBodyJointMotor::~btMultiBodyJointMotor()
 {
@@ -108,76 +103,74 @@ int btMultiBodyJointMotor::getIslandIdB() const
 	return -1;
 }
 
-
 void btMultiBodyJointMotor::createConstraintRows(btMultiBodyConstraintArray& constraintRows,
-		btMultiBodyJacobianData& data,
-		const btContactSolverInfo& infoGlobal)
+												 btMultiBodyJacobianData& data,
+												 const btContactSolverInfo& infoGlobal)
 {
-    // only positions need to be updated -- data.m_jacobians and force
-    // directions were set in the ctor and never change.
-	
+	// only positions need to be updated -- data.m_jacobians and force
+	// directions were set in the ctor and never change.
+
 	if (m_numDofsFinalized != m_jacSizeBoth)
 	{
-        finalizeMultiDof();
+		finalizeMultiDof();
 	}
 
 	//don't crash
 	if (m_numDofsFinalized != m_jacSizeBoth)
 		return;
 
-	if (m_maxAppliedImpulse==0.f)
+	if (m_maxAppliedImpulse == 0.f)
 		return;
 
 	const btScalar posError = 0;
 	const btVector3 dummy(0, 0, 0);
 
-	for (int row=0;row<getNumRows();row++)
+	for (int row = 0; row < getNumRows(); row++)
 	{
 		btMultiBodySolverConstraint& constraintRow = constraintRows.expandNonInitializing();
 
-        int dof = 0;
-        btScalar currentPosition = m_bodyA->getJointPosMultiDof(m_linkA)[dof];
-        btScalar currentVelocity = m_bodyA->getJointVelMultiDof(m_linkA)[dof];
-        btScalar positionStabiliationTerm = m_erp*(m_desiredPosition-currentPosition)/infoGlobal.m_timeStep;
-		
-        btScalar velocityError = (m_desiredVelocity - currentVelocity);
-        btScalar rhs =   m_kp * positionStabiliationTerm + currentVelocity+m_kd * velocityError;
-		if (rhs>m_rhsClamp)
+		int dof = 0;
+		btScalar currentPosition = m_bodyA->getJointPosMultiDof(m_linkA)[dof];
+		btScalar currentVelocity = m_bodyA->getJointVelMultiDof(m_linkA)[dof];
+		btScalar positionStabiliationTerm = m_erp * (m_desiredPosition - currentPosition) / infoGlobal.m_timeStep;
+
+		btScalar velocityError = (m_desiredVelocity - currentVelocity);
+		btScalar rhs = m_kp * positionStabiliationTerm + currentVelocity + m_kd * velocityError;
+		if (rhs > m_rhsClamp)
 		{
-			rhs=m_rhsClamp;
+			rhs = m_rhsClamp;
 		}
-		if (rhs<-m_rhsClamp)
+		if (rhs < -m_rhsClamp)
 		{
-			rhs=-m_rhsClamp;
+			rhs = -m_rhsClamp;
 		}
-        
-        
-		fillMultiBodyConstraint(constraintRow,data,jacobianA(row),jacobianB(row),dummy,dummy,dummy,dummy,posError,infoGlobal,-m_maxAppliedImpulse,m_maxAppliedImpulse,false,1,false,rhs);
+
+		fillMultiBodyConstraint(constraintRow, data, jacobianA(row), jacobianB(row), dummy, dummy, dummy, dummy, posError, infoGlobal, -m_maxAppliedImpulse, m_maxAppliedImpulse, false, 1, false, rhs);
 		constraintRow.m_orgConstraint = this;
 		constraintRow.m_orgDofIndex = row;
 		{
 			//expect either prismatic or revolute joint type for now
-			btAssert((m_bodyA->getLink(m_linkA).m_jointType == btMultibodyLink::eRevolute)||(m_bodyA->getLink(m_linkA).m_jointType == btMultibodyLink::ePrismatic));
+			btAssert((m_bodyA->getLink(m_linkA).m_jointType == btMultibodyLink::eRevolute) || (m_bodyA->getLink(m_linkA).m_jointType == btMultibodyLink::ePrismatic));
 			switch (m_bodyA->getLink(m_linkA).m_jointType)
 			{
 				case btMultibodyLink::eRevolute:
 				{
 					constraintRow.m_contactNormal1.setZero();
 					constraintRow.m_contactNormal2.setZero();
-					btVector3 revoluteAxisInWorld = quatRotate(m_bodyA->getLink(m_linkA).m_cachedWorldTransform.getRotation(),m_bodyA->getLink(m_linkA).m_axes[0].m_topVec);
-					constraintRow.m_relpos1CrossNormal=revoluteAxisInWorld;
-					constraintRow.m_relpos2CrossNormal=-revoluteAxisInWorld;
-					
+					btVector3 revoluteAxisInWorld = quatRotate(m_bodyA->getLink(m_linkA).m_cachedWorldTransform.getRotation(), m_bodyA->getLink(m_linkA).m_axes[0].m_topVec);
+					constraintRow.m_relpos1CrossNormal = revoluteAxisInWorld;
+					constraintRow.m_relpos2CrossNormal = -revoluteAxisInWorld;
+
 					break;
 				}
 				case btMultibodyLink::ePrismatic:
 				{
-					btVector3 prismaticAxisInWorld = quatRotate(m_bodyA->getLink(m_linkA).m_cachedWorldTransform.getRotation(),m_bodyA->getLink(m_linkA).m_axes[0].m_bottomVec);
-					constraintRow.m_contactNormal1=prismaticAxisInWorld;
-					constraintRow.m_contactNormal2=-prismaticAxisInWorld;
+					btVector3 prismaticAxisInWorld = quatRotate(m_bodyA->getLink(m_linkA).m_cachedWorldTransform.getRotation(), m_bodyA->getLink(m_linkA).m_axes[0].m_bottomVec);
+					constraintRow.m_contactNormal1 = prismaticAxisInWorld;
+					constraintRow.m_contactNormal2 = -prismaticAxisInWorld;
 					constraintRow.m_relpos1CrossNormal.setZero();
 					constraintRow.m_relpos2CrossNormal.setZero();
-					
+
 					break;
 				}
 				default:
@@ -185,10 +178,6 @@ void btMultiBodyJointMotor::createConstraintRows(btMultiBodyConstraintArray& con
 					btAssert(0);
 				}
 			};
-			
 		}
-
 	}
-
 }
-
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyJointMotor.h b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyJointMotor.h
index 4063bed79af..1aca36352e7 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyJointMotor.h
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyJointMotor.h
@@ -24,41 +24,38 @@ struct btSolverInfo;
 class btMultiBodyJointMotor : public btMultiBodyConstraint
 {
 protected:
-
-	btScalar	m_desiredVelocity;
-	btScalar	m_desiredPosition;
-	btScalar    m_kd;
-	btScalar    m_kp;
-	btScalar	m_erp;
-	btScalar	m_rhsClamp;//maximum error
-	
+	btScalar m_desiredVelocity;
+	btScalar m_desiredPosition;
+	btScalar m_kd;
+	btScalar m_kp;
+	btScalar m_erp;
+	btScalar m_rhsClamp;  //maximum error
 
 public:
-
 	btMultiBodyJointMotor(btMultiBody* body, int link, btScalar desiredVelocity, btScalar maxMotorImpulse);
 	btMultiBodyJointMotor(btMultiBody* body, int link, int linkDoF, btScalar desiredVelocity, btScalar maxMotorImpulse);
 	virtual ~btMultiBodyJointMotor();
-    virtual void finalizeMultiDof();
+	virtual void finalizeMultiDof();
 
 	virtual int getIslandIdA() const;
 	virtual int getIslandIdB() const;
 
 	virtual void createConstraintRows(btMultiBodyConstraintArray& constraintRows,
-		btMultiBodyJacobianData& data,
-		const btContactSolverInfo& infoGlobal);
+									  btMultiBodyJacobianData& data,
+									  const btContactSolverInfo& infoGlobal);
 
-    virtual void setVelocityTarget(btScalar velTarget, btScalar kd = 1.f)
-    {
-        m_desiredVelocity = velTarget;
-        m_kd = kd;
-    }
+	virtual void setVelocityTarget(btScalar velTarget, btScalar kd = 1.f)
+	{
+		m_desiredVelocity = velTarget;
+		m_kd = kd;
+	}
+
+	virtual void setPositionTarget(btScalar posTarget, btScalar kp = 1.f)
+	{
+		m_desiredPosition = posTarget;
+		m_kp = kp;
+	}
 
-    virtual void setPositionTarget(btScalar posTarget, btScalar kp = 1.f)
-    {
-        m_desiredPosition = posTarget;
-        m_kp = kp;
-    }
-    
 	virtual void setErp(btScalar erp)
 	{
 		m_erp = erp;
@@ -77,5 +74,4 @@ public:
 	}
 };
 
-#endif //BT_MULTIBODY_JOINT_MOTOR_H
-
+#endif  //BT_MULTIBODY_JOINT_MOTOR_H
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyLink.h b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyLink.h
index 21c9e7a557c..92d41dfac21 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyLink.h
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyLink.h
@@ -20,7 +20,7 @@ subject to the following restrictions:
 #include "LinearMath/btVector3.h"
 #include "BulletCollision/CollisionDispatch/btCollisionObject.h"
 
-enum	btMultiBodyLinkFlags
+enum btMultiBodyLinkFlags
 {
 	BT_MULTIBODYLINKFLAGS_DISABLE_PARENT_COLLISION = 1,
 	BT_MULTIBODYLINKFLAGS_DISABLE_ALL_PARENT_COLLISION = 2,
@@ -36,7 +36,6 @@ enum	btMultiBodyLinkFlags
 
 //namespace {
 
-
 #include "LinearMath/btSpatialAlgebra.h"
 
 //}
@@ -45,27 +44,26 @@ enum	btMultiBodyLinkFlags
 // Link struct
 //
 
-struct btMultibodyLink 
+struct btMultibodyLink
 {
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-    btScalar m_mass;         // mass of link
-    btVector3 m_inertiaLocal;   // inertia of link (local frame; diagonal)
+	btScalar m_mass;           // mass of link
+	btVector3 m_inertiaLocal;  // inertia of link (local frame; diagonal)
 
-    int m_parent;         // index of the parent link (assumed to be < index of this link), or -1 if parent is the base link.
+	int m_parent;  // index of the parent link (assumed to be < index of this link), or -1 if parent is the base link.
 
-    btQuaternion m_zeroRotParentToThis;    // rotates vectors in parent-frame to vectors in local-frame (when q=0). constant.
+	btQuaternion m_zeroRotParentToThis;  // rotates vectors in parent-frame to vectors in local-frame (when q=0). constant.
 
-    btVector3 m_dVector;   // vector from the inboard joint pos to this link's COM. (local frame.) constant.
-	//this is set to zero for planar joint (see also m_eVector comment)
-	
-    // m_eVector is constant, but depends on the joint type:
-    // revolute, fixed, prismatic, spherical: vector from parent's COM to the pivot point, in PARENT's frame.
+	btVector3 m_dVector;  // vector from the inboard joint pos to this link's COM. (local frame.) constant.
+						  //this is set to zero for planar joint (see also m_eVector comment)
+
+	// m_eVector is constant, but depends on the joint type:
+	// revolute, fixed, prismatic, spherical: vector from parent's COM to the pivot point, in PARENT's frame.
 	// planar: vector from COM of parent to COM of this link, WHEN Q = 0. (local frame.)
 	// todo: fix the planar so it is consistent with the other joints
-	
-    btVector3 m_eVector;
+
+	btVector3 m_eVector;
 
 	btSpatialMotionVector m_absFrameTotVelocity, m_absFrameLocVelocity;
 
@@ -79,13 +77,11 @@ struct btMultibodyLink
 		eInvalid
 	};
 
-	
-
 	// "axis" = spatial joint axis (Mirtich Defn 9 p104). (expressed in local frame.) constant.
-    // for prismatic: m_axesTop[0] = zero;
-    //                m_axesBottom[0] = unit vector along the joint axis.
-    // for revolute: m_axesTop[0] = unit vector along the rotation axis (u);
-    //               m_axesBottom[0] = u cross m_dVector (i.e. COM linear motion due to the rotation at the joint)
+	// for prismatic: m_axesTop[0] = zero;
+	//                m_axesBottom[0] = unit vector along the joint axis.
+	// for revolute: m_axesTop[0] = unit vector along the rotation axis (u);
+	//               m_axesBottom[0] = u cross m_dVector (i.e. COM linear motion due to the rotation at the joint)
 	//
 	// for spherical: m_axesTop[0][1][2] (u1,u2,u3) form a 3x3 identity matrix (3 rotation axes)
 	//				  m_axesBottom[0][1][2] cross u1,u2,u3 (i.e. COM linear motion due to the rotation at the joint)
@@ -93,143 +89,141 @@ struct btMultibodyLink
 	// for planar: m_axesTop[0] = unit vector along the rotation axis (u); defines the plane of motion
 	//			   m_axesTop[1][2] = zero
 	//			   m_axesBottom[0] = zero
-	//			   m_axesBottom[1][2] = unit vectors along the translational axes on that plane		
+	//			   m_axesBottom[1][2] = unit vectors along the translational axes on that plane
 	btSpatialMotionVector m_axes[6];
 	void setAxisTop(int dof, const btVector3 &axis) { m_axes[dof].m_topVec = axis; }
-	void setAxisBottom(int dof, const btVector3 &axis) 
-	{ 
-		m_axes[dof].m_bottomVec = axis; 
-	}
-	void setAxisTop(int dof, const btScalar &x, const btScalar &y, const btScalar &z) 
+	void setAxisBottom(int dof, const btVector3 &axis)
 	{
-		m_axes[dof].m_topVec.setValue(x, y, z); 
+		m_axes[dof].m_bottomVec = axis;
 	}
-	void setAxisBottom(int dof, const btScalar &x, const btScalar &y, const btScalar &z) 
-	{ 
-		m_axes[dof].m_bottomVec.setValue(x, y, z); 
+	void setAxisTop(int dof, const btScalar &x, const btScalar &y, const btScalar &z)
+	{
+		m_axes[dof].m_topVec.setValue(x, y, z);
 	}
-	const btVector3 & getAxisTop(int dof) const { return m_axes[dof].m_topVec; }
-	const btVector3 & getAxisBottom(int dof) const { return m_axes[dof].m_bottomVec; }
+	void setAxisBottom(int dof, const btScalar &x, const btScalar &y, const btScalar &z)
+	{
+		m_axes[dof].m_bottomVec.setValue(x, y, z);
+	}
+	const btVector3 &getAxisTop(int dof) const { return m_axes[dof].m_topVec; }
+	const btVector3 &getAxisBottom(int dof) const { return m_axes[dof].m_bottomVec; }
 
 	int m_dofOffset, m_cfgOffset;
 
-    btQuaternion m_cachedRotParentToThis;   // rotates vectors in parent frame to vectors in local frame
-    btVector3 m_cachedRVector;                // vector from COM of parent to COM of this link, in local frame.
+	btQuaternion m_cachedRotParentToThis;  // rotates vectors in parent frame to vectors in local frame
+	btVector3 m_cachedRVector;             // vector from COM of parent to COM of this link, in local frame.
 
-    btVector3 m_appliedForce;    // In WORLD frame
-    btVector3 m_appliedTorque;   // In WORLD frame
+	btVector3 m_appliedForce;   // In WORLD frame
+	btVector3 m_appliedTorque;  // In WORLD frame
 
-btVector3 m_appliedConstraintForce;    // In WORLD frame
-    btVector3 m_appliedConstraintTorque;   // In WORLD frame
+	btVector3 m_appliedConstraintForce;   // In WORLD frame
+	btVector3 m_appliedConstraintTorque;  // In WORLD frame
 
 	btScalar m_jointPos[7];
-    
-    //m_jointTorque is the joint torque applied by the user using 'addJointTorque'.
-    //It gets set to zero after each internal stepSimulation call
+
+	//m_jointTorque is the joint torque applied by the user using 'addJointTorque'.
+	//It gets set to zero after each internal stepSimulation call
 	btScalar m_jointTorque[6];
-    
-	class btMultiBodyLinkCollider* m_collider;
+
+	class btMultiBodyLinkCollider *m_collider;
 	int m_flags;
-	
-	
-	int m_dofCount, m_posVarCount;				//redundant but handy
-	
+
+	int m_dofCount, m_posVarCount;  //redundant but handy
+
 	eFeatherstoneJointType m_jointType;
-	
-	struct btMultiBodyJointFeedback*	m_jointFeedback;
 
-	btTransform	m_cachedWorldTransform;//this cache is updated when calling btMultiBody::forwardKinematics
+	struct btMultiBodyJointFeedback *m_jointFeedback;
+
+	btTransform m_cachedWorldTransform;  //this cache is updated when calling btMultiBody::forwardKinematics
+
+	const char *m_linkName;   //m_linkName memory needs to be managed by the developer/user!
+	const char *m_jointName;  //m_jointName memory needs to be managed by the developer/user!
+	const void *m_userPtr;    //m_userPtr ptr needs to be managed by the developer/user!
+
+	btScalar m_jointDamping;      //todo: implement this internally. It is unused for now, it is set by a URDF loader. User can apply manual damping.
+	btScalar m_jointFriction;     //todo: implement this internally. It is unused for now, it is set by a URDF loader. User can apply manual friction using a velocity motor.
+	btScalar m_jointLowerLimit;   //todo: implement this internally. It is unused for now, it is set by a URDF loader.
+	btScalar m_jointUpperLimit;   //todo: implement this internally. It is unused for now, it is set by a URDF loader.
+	btScalar m_jointMaxForce;     //todo: implement this internally. It is unused for now, it is set by a URDF loader.
+	btScalar m_jointMaxVelocity;  //todo: implement this internally. It is unused for now, it is set by a URDF loader.
 
-	const char* m_linkName;//m_linkName memory needs to be managed by the developer/user!
-	const char* m_jointName;//m_jointName memory needs to be managed by the developer/user!
-    const void* m_userPtr;//m_userPtr ptr needs to be managed by the developer/user!
-    
-	btScalar m_jointDamping; //todo: implement this internally. It is unused for now, it is set by a URDF loader. User can apply manual damping.
-	btScalar m_jointFriction; //todo: implement this internally. It is unused for now, it is set by a URDF loader. User can apply manual friction using a velocity motor.
-	btScalar m_jointLowerLimit; //todo: implement this internally. It is unused for now, it is set by a URDF loader. 
-	btScalar m_jointUpperLimit; //todo: implement this internally. It is unused for now, it is set by a URDF loader.
-	btScalar m_jointMaxForce; //todo: implement this internally. It is unused for now, it is set by a URDF loader. 
-	btScalar m_jointMaxVelocity;//todo: implement this internally. It is unused for now, it is set by a URDF loader. 
-	
 	// ctor: set some sensible defaults
 	btMultibodyLink()
-		: 	m_mass(1),
-			m_parent(-1),
-			m_zeroRotParentToThis(0, 0, 0, 1),
-			m_cachedRotParentToThis(0, 0, 0, 1),
-			m_collider(0),
-			m_flags(0),
-			m_dofCount(0),
-			m_posVarCount(0),
-			m_jointType(btMultibodyLink::eInvalid),
-			m_jointFeedback(0),
-			m_linkName(0),
-			m_jointName(0),
-            m_userPtr(0),
-			m_jointDamping(0),
-			m_jointFriction(0),
-			m_jointLowerLimit(0),
-			m_jointUpperLimit(0),
-			m_jointMaxForce(0),
-			m_jointMaxVelocity(0)
+		: m_mass(1),
+		  m_parent(-1),
+		  m_zeroRotParentToThis(0, 0, 0, 1),
+		  m_cachedRotParentToThis(0, 0, 0, 1),
+		  m_collider(0),
+		  m_flags(0),
+		  m_dofCount(0),
+		  m_posVarCount(0),
+		  m_jointType(btMultibodyLink::eInvalid),
+		  m_jointFeedback(0),
+		  m_linkName(0),
+		  m_jointName(0),
+		  m_userPtr(0),
+		  m_jointDamping(0),
+		  m_jointFriction(0),
+		  m_jointLowerLimit(0),
+		  m_jointUpperLimit(0),
+		  m_jointMaxForce(0),
+		  m_jointMaxVelocity(0)
 	{
-		
 		m_inertiaLocal.setValue(1, 1, 1);
 		setAxisTop(0, 0., 0., 0.);
 		setAxisBottom(0, 1., 0., 0.);
 		m_dVector.setValue(0, 0, 0);
 		m_eVector.setValue(0, 0, 0);
 		m_cachedRVector.setValue(0, 0, 0);
-		m_appliedForce.setValue( 0, 0, 0);
+		m_appliedForce.setValue(0, 0, 0);
 		m_appliedTorque.setValue(0, 0, 0);
-		m_appliedConstraintForce.setValue(0,0,0);
-		m_appliedConstraintTorque.setValue(0,0,0);
-		//		
+		m_appliedConstraintForce.setValue(0, 0, 0);
+		m_appliedConstraintTorque.setValue(0, 0, 0);
+		//
 		m_jointPos[0] = m_jointPos[1] = m_jointPos[2] = m_jointPos[4] = m_jointPos[5] = m_jointPos[6] = 0.f;
-		m_jointPos[3] = 1.f;			//"quat.w"
+		m_jointPos[3] = 1.f;  //"quat.w"
 		m_jointTorque[0] = m_jointTorque[1] = m_jointTorque[2] = m_jointTorque[3] = m_jointTorque[4] = m_jointTorque[5] = 0.f;
 		m_cachedWorldTransform.setIdentity();
 	}
 
-    // routine to update m_cachedRotParentToThis and m_cachedRVector
+	// routine to update m_cachedRotParentToThis and m_cachedRVector
 	void updateCacheMultiDof(btScalar *pq = 0)
 	{
 		btScalar *pJointPos = (pq ? pq : &m_jointPos[0]);
 
-		switch(m_jointType)
+		switch (m_jointType)
 		{
 			case eRevolute:
 			{
-				m_cachedRotParentToThis = btQuaternion(getAxisTop(0),-pJointPos[0]) * m_zeroRotParentToThis;
-				m_cachedRVector = m_dVector + quatRotate(m_cachedRotParentToThis,m_eVector);
+				m_cachedRotParentToThis = btQuaternion(getAxisTop(0), -pJointPos[0]) * m_zeroRotParentToThis;
+				m_cachedRVector = m_dVector + quatRotate(m_cachedRotParentToThis, m_eVector);
 
 				break;
 			}
 			case ePrismatic:
 			{
 				// m_cachedRotParentToThis never changes, so no need to update
-				m_cachedRVector = m_dVector + quatRotate(m_cachedRotParentToThis,m_eVector) + pJointPos[0] * getAxisBottom(0);
+				m_cachedRVector = m_dVector + quatRotate(m_cachedRotParentToThis, m_eVector) + pJointPos[0] * getAxisBottom(0);
 
 				break;
 			}
 			case eSpherical:
 			{
 				m_cachedRotParentToThis = btQuaternion(pJointPos[0], pJointPos[1], pJointPos[2], -pJointPos[3]) * m_zeroRotParentToThis;
-				m_cachedRVector = m_dVector + quatRotate(m_cachedRotParentToThis,m_eVector);
+				m_cachedRVector = m_dVector + quatRotate(m_cachedRotParentToThis, m_eVector);
 
 				break;
 			}
 			case ePlanar:
 			{
-				m_cachedRotParentToThis = btQuaternion(getAxisTop(0),-pJointPos[0]) * m_zeroRotParentToThis;				
-				m_cachedRVector = quatRotate(btQuaternion(getAxisTop(0),-pJointPos[0]), pJointPos[1] * getAxisBottom(1) + pJointPos[2] * getAxisBottom(2)) + quatRotate(m_cachedRotParentToThis,m_eVector);				
+				m_cachedRotParentToThis = btQuaternion(getAxisTop(0), -pJointPos[0]) * m_zeroRotParentToThis;
+				m_cachedRVector = quatRotate(btQuaternion(getAxisTop(0), -pJointPos[0]), pJointPos[1] * getAxisBottom(1) + pJointPos[2] * getAxisBottom(2)) + quatRotate(m_cachedRotParentToThis, m_eVector);
 
 				break;
 			}
 			case eFixed:
 			{
 				m_cachedRotParentToThis = m_zeroRotParentToThis;
-				m_cachedRVector = m_dVector + quatRotate(m_cachedRotParentToThis,m_eVector);
+				m_cachedRVector = m_dVector + quatRotate(m_cachedRotParentToThis, m_eVector);
 
 				break;
 			}
@@ -242,5 +236,4 @@ btVector3 m_appliedConstraintForce;    // In WORLD frame
 	}
 };
 
-
-#endif //BT_MULTIBODY_LINK_H
+#endif  //BT_MULTIBODY_LINK_H
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyLinkCollider.h b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyLinkCollider.h
index 7092e62b5ae..f91c001f127 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyLinkCollider.h
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyLinkCollider.h
@@ -29,21 +29,18 @@ subject to the following restrictions:
 #define btMultiBodyLinkColliderDataName "btMultiBodyLinkColliderFloatData"
 #endif
 
-
 class btMultiBodyLinkCollider : public btCollisionObject
 {
-//protected:
+	//protected:
 public:
-
 	btMultiBody* m_multiBody;
 	int m_link;
 
-
-	btMultiBodyLinkCollider (btMultiBody* multiBody,int link)
-		:m_multiBody(multiBody),
-		m_link(link)
+	btMultiBodyLinkCollider(btMultiBody* multiBody, int link)
+		: m_multiBody(multiBody),
+		  m_link(link)
 	{
-		m_checkCollideWith =  true;
+		m_checkCollideWith = true;
 		//we need to remove the 'CF_STATIC_OBJECT' flag, otherwise links/base doesn't merge islands
 		//this means that some constraints might point to bodies that are not in the islands, causing crashes
 		//if (link>=0 || (multiBody && !multiBody->hasFixedBase()))
@@ -59,18 +56,18 @@ public:
 	}
 	static btMultiBodyLinkCollider* upcast(btCollisionObject* colObj)
 	{
-		if (colObj->getInternalType()&btCollisionObject::CO_FEATHERSTONE_LINK)
+		if (colObj->getInternalType() & btCollisionObject::CO_FEATHERSTONE_LINK)
 			return (btMultiBodyLinkCollider*)colObj;
 		return 0;
 	}
 	static const btMultiBodyLinkCollider* upcast(const btCollisionObject* colObj)
 	{
-		if (colObj->getInternalType()&btCollisionObject::CO_FEATHERSTONE_LINK)
+		if (colObj->getInternalType() & btCollisionObject::CO_FEATHERSTONE_LINK)
 			return (btMultiBodyLinkCollider*)colObj;
 		return 0;
 	}
 
-	virtual bool checkCollideWithOverride(const  btCollisionObject* co) const
+	virtual bool checkCollideWithOverride(const btCollisionObject* co) const
 	{
 		const btMultiBodyLinkCollider* other = btMultiBodyLinkCollider::upcast(co);
 		if (!other)
@@ -81,47 +78,46 @@ public:
 			return false;
 
 		//check if 'link' has collision disabled
-		if (m_link>=0)
+		if (m_link >= 0)
 		{
 			const btMultibodyLink& link = m_multiBody->getLink(this->m_link);
-			if (link.m_flags&BT_MULTIBODYLINKFLAGS_DISABLE_ALL_PARENT_COLLISION)
+			if (link.m_flags & BT_MULTIBODYLINKFLAGS_DISABLE_ALL_PARENT_COLLISION)
 			{
 				int parent_of_this = m_link;
 				while (1)
 				{
-					if (parent_of_this==-1)
+					if (parent_of_this == -1)
 						break;
 					parent_of_this = m_multiBody->getLink(parent_of_this).m_parent;
-					if (parent_of_this==other->m_link)
+					if (parent_of_this == other->m_link)
 					{
 						return false;
 					}
 				}
 			}
-			else if (link.m_flags&BT_MULTIBODYLINKFLAGS_DISABLE_PARENT_COLLISION)
+			else if (link.m_flags & BT_MULTIBODYLINKFLAGS_DISABLE_PARENT_COLLISION)
 			{
-				if ( link.m_parent == other->m_link)
+				if (link.m_parent == other->m_link)
 					return false;
 			}
-
 		}
 
-		if (other->m_link>=0)
+		if (other->m_link >= 0)
 		{
 			const btMultibodyLink& otherLink = other->m_multiBody->getLink(other->m_link);
-			if (otherLink.m_flags& BT_MULTIBODYLINKFLAGS_DISABLE_ALL_PARENT_COLLISION)
+			if (otherLink.m_flags & BT_MULTIBODYLINKFLAGS_DISABLE_ALL_PARENT_COLLISION)
 			{
 				int parent_of_other = other->m_link;
 				while (1)
 				{
-					if (parent_of_other==-1)
+					if (parent_of_other == -1)
 						break;
 					parent_of_other = m_multiBody->getLink(parent_of_other).m_parent;
-					if (parent_of_other==this->m_link)
+					if (parent_of_other == this->m_link)
 						return false;
 				}
 			}
-			else if (otherLink.m_flags& BT_MULTIBODYLINKFLAGS_DISABLE_PARENT_COLLISION)
+			else if (otherLink.m_flags & BT_MULTIBODYLINKFLAGS_DISABLE_PARENT_COLLISION)
 			{
 				if (otherLink.m_parent == this->m_link)
 					return false;
@@ -130,13 +126,13 @@ public:
 		return true;
 	}
 
-	virtual	int	calculateSerializeBufferSize()	const;
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer,  class btSerializer* serializer) const;
-
+	virtual const char* serialize(void* dataBuffer, class btSerializer* serializer) const;
 };
 
+// clang-format off
 
 struct	btMultiBodyLinkColliderFloatData
 {
@@ -154,16 +150,18 @@ struct	btMultiBodyLinkColliderDoubleData
 	char		m_padding[4];
 };
 
-SIMD_FORCE_INLINE	int	btMultiBodyLinkCollider::calculateSerializeBufferSize() const
+// clang-format on
+
+SIMD_FORCE_INLINE int btMultiBodyLinkCollider::calculateSerializeBufferSize() const
 {
 	return sizeof(btMultiBodyLinkColliderData);
 }
 
-SIMD_FORCE_INLINE	const char*	btMultiBodyLinkCollider::serialize(void* dataBuffer,  class btSerializer* serializer) const
+SIMD_FORCE_INLINE const char* btMultiBodyLinkCollider::serialize(void* dataBuffer, class btSerializer* serializer) const
 {
 	btMultiBodyLinkColliderData* dataOut = (btMultiBodyLinkColliderData*)dataBuffer;
-	btCollisionObject::serialize(&dataOut->m_colObjData,serializer);
-	
+	btCollisionObject::serialize(&dataOut->m_colObjData, serializer);
+
 	dataOut->m_link = this->m_link;
 	dataOut->m_multiBody = (btMultiBodyData*)serializer->getUniquePointer(m_multiBody);
 
@@ -173,5 +171,4 @@ SIMD_FORCE_INLINE	const char*	btMultiBodyLinkCollider::serialize(void* dataBuffe
 	return btMultiBodyLinkColliderDataName;
 }
 
-#endif //BT_FEATHERSTONE_LINK_COLLIDER_H
-
+#endif  //BT_FEATHERSTONE_LINK_COLLIDER_H
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyPoint2Point.cpp b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyPoint2Point.cpp
index 2b59f0b7a66..37d3aede378 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyPoint2Point.cpp
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyPoint2Point.cpp
@@ -21,29 +21,29 @@ subject to the following restrictions:
 #include "LinearMath/btIDebugDraw.h"
 
 #ifndef BTMBP2PCONSTRAINT_BLOCK_ANGULAR_MOTION_TEST
-	#define BTMBP2PCONSTRAINT_DIM 3
+#define BTMBP2PCONSTRAINT_DIM 3
 #else
-	#define BTMBP2PCONSTRAINT_DIM 6
+#define BTMBP2PCONSTRAINT_DIM 6
 #endif
 
 btMultiBodyPoint2Point::btMultiBodyPoint2Point(btMultiBody* body, int link, btRigidBody* bodyB, const btVector3& pivotInA, const btVector3& pivotInB)
-	:btMultiBodyConstraint(body,0,link,-1,BTMBP2PCONSTRAINT_DIM,false),
-	m_rigidBodyA(0),
-	m_rigidBodyB(bodyB),
-	m_pivotInA(pivotInA),
-	m_pivotInB(pivotInB)
+	: btMultiBodyConstraint(body, 0, link, -1, BTMBP2PCONSTRAINT_DIM, false),
+	  m_rigidBodyA(0),
+	  m_rigidBodyB(bodyB),
+	  m_pivotInA(pivotInA),
+	  m_pivotInB(pivotInB)
 {
-    m_data.resize(BTMBP2PCONSTRAINT_DIM);//at least store the applied impulses
+	m_data.resize(BTMBP2PCONSTRAINT_DIM);  //at least store the applied impulses
 }
 
 btMultiBodyPoint2Point::btMultiBodyPoint2Point(btMultiBody* bodyA, int linkA, btMultiBody* bodyB, int linkB, const btVector3& pivotInA, const btVector3& pivotInB)
-	:btMultiBodyConstraint(bodyA,bodyB,linkA,linkB,BTMBP2PCONSTRAINT_DIM,false),
-	m_rigidBodyA(0),
-	m_rigidBodyB(0),
-	m_pivotInA(pivotInA),
-	m_pivotInB(pivotInB)
+	: btMultiBodyConstraint(bodyA, bodyB, linkA, linkB, BTMBP2PCONSTRAINT_DIM, false),
+	  m_rigidBodyA(0),
+	  m_rigidBodyB(0),
+	  m_pivotInA(pivotInA),
+	  m_pivotInB(pivotInB)
 {
-    m_data.resize(BTMBP2PCONSTRAINT_DIM);//at least store the applied impulses
+	m_data.resize(BTMBP2PCONSTRAINT_DIM);  //at least store the applied impulses
 }
 
 void btMultiBodyPoint2Point::finalizeMultiDof()
@@ -56,7 +56,6 @@ btMultiBodyPoint2Point::~btMultiBodyPoint2Point()
 {
 }
 
-
 int btMultiBodyPoint2Point::getIslandIdA() const
 {
 	if (m_rigidBodyA)
@@ -73,7 +72,7 @@ int btMultiBodyPoint2Point::getIslandIdA() const
 		else
 		{
 			if (m_bodyA->getLink(m_linkA).m_collider)
-				return m_bodyA->getLink(m_linkA).m_collider->getIslandTag();			
+				return m_bodyA->getLink(m_linkA).m_collider->getIslandTag();
 		}
 	}
 	return -1;
@@ -100,48 +99,43 @@ int btMultiBodyPoint2Point::getIslandIdB() const
 	return -1;
 }
 
-
-
 void btMultiBodyPoint2Point::createConstraintRows(btMultiBodyConstraintArray& constraintRows,
-		btMultiBodyJacobianData& data,
-		const btContactSolverInfo& infoGlobal)
+												  btMultiBodyJacobianData& data,
+												  const btContactSolverInfo& infoGlobal)
 {
-
-//	int i=1;
-int numDim = BTMBP2PCONSTRAINT_DIM;
-	for (int i=0;i<numDim;i++)
+	//	int i=1;
+	int numDim = BTMBP2PCONSTRAINT_DIM;
+	for (int i = 0; i < numDim; i++)
 	{
-
 		btMultiBodySolverConstraint& constraintRow = constraintRows.expandNonInitializing();
-        //memset(&constraintRow,0xffffffff,sizeof(btMultiBodySolverConstraint));
-	constraintRow.m_orgConstraint = this;
-	constraintRow.m_orgDofIndex = i;
-        constraintRow.m_relpos1CrossNormal.setValue(0,0,0);
-        constraintRow.m_contactNormal1.setValue(0,0,0);
-        constraintRow.m_relpos2CrossNormal.setValue(0,0,0);
-        constraintRow.m_contactNormal2.setValue(0,0,0);
-        constraintRow.m_angularComponentA.setValue(0,0,0);
-        constraintRow.m_angularComponentB.setValue(0,0,0);
+		//memset(&constraintRow,0xffffffff,sizeof(btMultiBodySolverConstraint));
+		constraintRow.m_orgConstraint = this;
+		constraintRow.m_orgDofIndex = i;
+		constraintRow.m_relpos1CrossNormal.setValue(0, 0, 0);
+		constraintRow.m_contactNormal1.setValue(0, 0, 0);
+		constraintRow.m_relpos2CrossNormal.setValue(0, 0, 0);
+		constraintRow.m_contactNormal2.setValue(0, 0, 0);
+		constraintRow.m_angularComponentA.setValue(0, 0, 0);
+		constraintRow.m_angularComponentB.setValue(0, 0, 0);
 
 		constraintRow.m_solverBodyIdA = data.m_fixedBodyId;
 		constraintRow.m_solverBodyIdB = data.m_fixedBodyId;
 
-		btVector3 contactNormalOnB(0,0,0);
+		btVector3 contactNormalOnB(0, 0, 0);
 #ifndef BTMBP2PCONSTRAINT_BLOCK_ANGULAR_MOTION_TEST
 		contactNormalOnB[i] = -1;
 #else
-		contactNormalOnB[i%3] = -1;
+		contactNormalOnB[i % 3] = -1;
 #endif
 
-
-		 // Convert local points back to world
+		// Convert local points back to world
 		btVector3 pivotAworld = m_pivotInA;
 		if (m_rigidBodyA)
 		{
-
 			constraintRow.m_solverBodyIdA = m_rigidBodyA->getCompanionId();
-			pivotAworld = m_rigidBodyA->getCenterOfMassTransform()*m_pivotInA;
-		} else
+			pivotAworld = m_rigidBodyA->getCenterOfMassTransform() * m_pivotInA;
+		}
+		else
 		{
 			if (m_bodyA)
 				pivotAworld = m_bodyA->localPosToWorld(m_linkA, m_pivotInA);
@@ -150,44 +144,41 @@ int numDim = BTMBP2PCONSTRAINT_DIM;
 		if (m_rigidBodyB)
 		{
 			constraintRow.m_solverBodyIdB = m_rigidBodyB->getCompanionId();
-			pivotBworld = m_rigidBodyB->getCenterOfMassTransform()*m_pivotInB;
-		} else
+			pivotBworld = m_rigidBodyB->getCenterOfMassTransform() * m_pivotInB;
+		}
+		else
 		{
 			if (m_bodyB)
 				pivotBworld = m_bodyB->localPosToWorld(m_linkB, m_pivotInB);
-
 		}
 
-		btScalar posError = i < 3 ? (pivotAworld-pivotBworld).dot(contactNormalOnB) : 0;
+		btScalar posError = i < 3 ? (pivotAworld - pivotBworld).dot(contactNormalOnB) : 0;
 
 #ifndef BTMBP2PCONSTRAINT_BLOCK_ANGULAR_MOTION_TEST
 
-
-		fillMultiBodyConstraint(constraintRow, data, 0, 0, btVector3(0,0,0),
-															contactNormalOnB, pivotAworld, pivotBworld,						//sucks but let it be this way "for the time being"
-															posError,
-															infoGlobal,
-															-m_maxAppliedImpulse, m_maxAppliedImpulse
-															);
-    //@todo: support the case of btMultiBody versus btRigidBody,
-    //see btPoint2PointConstraint::getInfo2NonVirtual
+		fillMultiBodyConstraint(constraintRow, data, 0, 0, btVector3(0, 0, 0),
+								contactNormalOnB, pivotAworld, pivotBworld,  //sucks but let it be this way "for the time being"
+								posError,
+								infoGlobal,
+								-m_maxAppliedImpulse, m_maxAppliedImpulse);
+		//@todo: support the case of btMultiBody versus btRigidBody,
+		//see btPoint2PointConstraint::getInfo2NonVirtual
 #else
 		const btVector3 dummy(0, 0, 0);
 
 		btAssert(m_bodyA->isMultiDof());
 
 		btScalar* jac1 = jacobianA(i);
-		const btVector3 &normalAng = i >= 3 ? contactNormalOnB : dummy;
-		const btVector3 &normalLin = i < 3 ? contactNormalOnB : dummy;
+		const btVector3& normalAng = i >= 3 ? contactNormalOnB : dummy;
+		const btVector3& normalLin = i < 3 ? contactNormalOnB : dummy;
 
 		m_bodyA->filConstraintJacobianMultiDof(m_linkA, pivotAworld, normalAng, normalLin, jac1, data.scratch_r, data.scratch_v, data.scratch_m);
 
 		fillMultiBodyConstraint(constraintRow, data, jac1, 0,
-													dummy, dummy, dummy,						//sucks but let it be this way "for the time being"
-													posError,
-													infoGlobal,
-													-m_maxAppliedImpulse, m_maxAppliedImpulse
-													);
+								dummy, dummy, dummy,  //sucks but let it be this way "for the time being"
+								posError,
+								infoGlobal,
+								-m_maxAppliedImpulse, m_maxAppliedImpulse);
 #endif
 	}
 }
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyPoint2Point.h b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyPoint2Point.h
index bf39acc5b95..ef03a557ec5 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyPoint2Point.h
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodyPoint2Point.h
@@ -22,22 +22,20 @@ subject to the following restrictions:
 
 //#define BTMBP2PCONSTRAINT_BLOCK_ANGULAR_MOTION_TEST
 
-ATTRIBUTE_ALIGNED16(class) btMultiBodyPoint2Point : public btMultiBodyConstraint
+ATTRIBUTE_ALIGNED16(class)
+btMultiBodyPoint2Point : public btMultiBodyConstraint
 {
 protected:
-
-	btRigidBody*	m_rigidBodyA;
-	btRigidBody*	m_rigidBodyB;
-	btVector3		m_pivotInA;
-	btVector3		m_pivotInB;
-
+	btRigidBody* m_rigidBodyA;
+	btRigidBody* m_rigidBodyB;
+	btVector3 m_pivotInA;
+	btVector3 m_pivotInB;
 
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-	btMultiBodyPoint2Point(btMultiBody* body, int link, btRigidBody* bodyB, const btVector3& pivotInA, const btVector3& pivotInB);
-	btMultiBodyPoint2Point(btMultiBody* bodyA, int linkA, btMultiBody* bodyB, int linkB, const btVector3& pivotInA, const btVector3& pivotInB);
+	btMultiBodyPoint2Point(btMultiBody * body, int link, btRigidBody* bodyB, const btVector3& pivotInA, const btVector3& pivotInB);
+	btMultiBodyPoint2Point(btMultiBody * bodyA, int linkA, btMultiBody* bodyB, int linkB, const btVector3& pivotInA, const btVector3& pivotInB);
 
 	virtual ~btMultiBodyPoint2Point();
 
@@ -46,9 +44,9 @@ public:
 	virtual int getIslandIdA() const;
 	virtual int getIslandIdB() const;
 
-	virtual void createConstraintRows(btMultiBodyConstraintArray& constraintRows,
-		btMultiBodyJacobianData& data,
-		const btContactSolverInfo& infoGlobal);
+	virtual void createConstraintRows(btMultiBodyConstraintArray & constraintRows,
+									  btMultiBodyJacobianData & data,
+									  const btContactSolverInfo& infoGlobal);
 
 	const btVector3& getPivotInB() const
 	{
@@ -60,9 +58,7 @@ public:
 		m_pivotInB = pivotInB;
 	}
 
-
-	virtual void debugDraw(class btIDebugDraw* drawer);
-
+	virtual void debugDraw(class btIDebugDraw * drawer);
 };
 
-#endif //BT_MULTIBODY_POINT2POINT_H
+#endif  //BT_MULTIBODY_POINT2POINT_H
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodySliderConstraint.cpp b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodySliderConstraint.cpp
index 43f26f9833c..e025302ce6c 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodySliderConstraint.cpp
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodySliderConstraint.cpp
@@ -25,29 +25,29 @@ subject to the following restrictions:
 #define EPSILON 0.000001
 
 btMultiBodySliderConstraint::btMultiBodySliderConstraint(btMultiBody* body, int link, btRigidBody* bodyB, const btVector3& pivotInA, const btVector3& pivotInB, const btMatrix3x3& frameInA, const btMatrix3x3& frameInB, const btVector3& jointAxis)
-	:btMultiBodyConstraint(body,0,link,-1,BTMBSLIDERCONSTRAINT_DIM,false),
-	m_rigidBodyA(0),
-	m_rigidBodyB(bodyB),
-	m_pivotInA(pivotInA),
-	m_pivotInB(pivotInB),
-    m_frameInA(frameInA),
-    m_frameInB(frameInB),
-    m_jointAxis(jointAxis)
+	: btMultiBodyConstraint(body, 0, link, -1, BTMBSLIDERCONSTRAINT_DIM, false),
+	  m_rigidBodyA(0),
+	  m_rigidBodyB(bodyB),
+	  m_pivotInA(pivotInA),
+	  m_pivotInB(pivotInB),
+	  m_frameInA(frameInA),
+	  m_frameInB(frameInB),
+	  m_jointAxis(jointAxis)
 {
-    m_data.resize(BTMBSLIDERCONSTRAINT_DIM);//at least store the applied impulses
+	m_data.resize(BTMBSLIDERCONSTRAINT_DIM);  //at least store the applied impulses
 }
 
 btMultiBodySliderConstraint::btMultiBodySliderConstraint(btMultiBody* bodyA, int linkA, btMultiBody* bodyB, int linkB, const btVector3& pivotInA, const btVector3& pivotInB, const btMatrix3x3& frameInA, const btMatrix3x3& frameInB, const btVector3& jointAxis)
-	:btMultiBodyConstraint(bodyA,bodyB,linkA,linkB,BTMBSLIDERCONSTRAINT_DIM,false),
-	m_rigidBodyA(0),
-	m_rigidBodyB(0),
-	m_pivotInA(pivotInA),
-	m_pivotInB(pivotInB),
-    m_frameInA(frameInA),
-    m_frameInB(frameInB),
-    m_jointAxis(jointAxis)
+	: btMultiBodyConstraint(bodyA, bodyB, linkA, linkB, BTMBSLIDERCONSTRAINT_DIM, false),
+	  m_rigidBodyA(0),
+	  m_rigidBodyB(0),
+	  m_pivotInA(pivotInA),
+	  m_pivotInB(pivotInB),
+	  m_frameInA(frameInA),
+	  m_frameInB(frameInB),
+	  m_jointAxis(jointAxis)
 {
-    m_data.resize(BTMBSLIDERCONSTRAINT_DIM);//at least store the applied impulses
+	m_data.resize(BTMBSLIDERCONSTRAINT_DIM);  //at least store the applied impulses
 }
 
 void btMultiBodySliderConstraint::finalizeMultiDof()
@@ -60,7 +60,6 @@ btMultiBodySliderConstraint::~btMultiBodySliderConstraint()
 {
 }
 
-
 int btMultiBodySliderConstraint::getIslandIdA() const
 {
 	if (m_rigidBodyA)
@@ -105,98 +104,100 @@ int btMultiBodySliderConstraint::getIslandIdB() const
 }
 void btMultiBodySliderConstraint::createConstraintRows(btMultiBodyConstraintArray& constraintRows, btMultiBodyJacobianData& data, const btContactSolverInfo& infoGlobal)
 {
-    // Convert local points back to world
-    btVector3 pivotAworld = m_pivotInA;
-    btMatrix3x3 frameAworld = m_frameInA;
-    btVector3 jointAxis = m_jointAxis;
-    if (m_rigidBodyA)
-    {
-        pivotAworld = m_rigidBodyA->getCenterOfMassTransform()*m_pivotInA;
-        frameAworld = m_frameInA.transpose()*btMatrix3x3(m_rigidBodyA->getOrientation());
-        jointAxis = quatRotate(m_rigidBodyA->getOrientation(),m_jointAxis);
-        
-    } else if (m_bodyA) {
-        pivotAworld = m_bodyA->localPosToWorld(m_linkA, m_pivotInA);
-        frameAworld = m_bodyA->localFrameToWorld(m_linkA, m_frameInA);
-        jointAxis = m_bodyA->localDirToWorld(m_linkA, m_jointAxis);
-    }
-    btVector3 pivotBworld = m_pivotInB;
-    btMatrix3x3 frameBworld = m_frameInB;
-    if (m_rigidBodyB)
-    {
-        pivotBworld = m_rigidBodyB->getCenterOfMassTransform()*m_pivotInB;
-        frameBworld = m_frameInB.transpose()*btMatrix3x3(m_rigidBodyB->getOrientation());
-        
-    } else if (m_bodyB) {
-        pivotBworld = m_bodyB->localPosToWorld(m_linkB, m_pivotInB);
-        frameBworld = m_bodyB->localFrameToWorld(m_linkB, m_frameInB);
-    }
-    
-    btVector3 constraintAxis[2];
-    for (int i = 0; i < 3; ++i)
-    {
-        constraintAxis[0] = frameAworld.getColumn(i).cross(jointAxis);
-        if (constraintAxis[0].safeNorm() > EPSILON)
-        {
-            constraintAxis[0] = constraintAxis[0].normalized();
-            constraintAxis[1] = jointAxis.cross(constraintAxis[0]);
-            constraintAxis[1] = constraintAxis[1].normalized();
-            break;
-        }
-    }
-    
-    btMatrix3x3 relRot = frameAworld.inverse()*frameBworld;
-    btVector3 angleDiff;
-    btGeneric6DofSpring2Constraint::matrixToEulerXYZ(relRot,angleDiff);
-    
-    int numDim = BTMBSLIDERCONSTRAINT_DIM;
-    for (int i=0;i<numDim;i++)
+	// Convert local points back to world
+	btVector3 pivotAworld = m_pivotInA;
+	btMatrix3x3 frameAworld = m_frameInA;
+	btVector3 jointAxis = m_jointAxis;
+	if (m_rigidBodyA)
 	{
-        btMultiBodySolverConstraint& constraintRow = constraintRows.expandNonInitializing();
-        constraintRow.m_orgConstraint = this;
-        constraintRow.m_orgDofIndex = i;
-        constraintRow.m_relpos1CrossNormal.setValue(0,0,0);
-        constraintRow.m_contactNormal1.setValue(0,0,0);
-        constraintRow.m_relpos2CrossNormal.setValue(0,0,0);
-        constraintRow.m_contactNormal2.setValue(0,0,0);
-        constraintRow.m_angularComponentA.setValue(0,0,0);
-        constraintRow.m_angularComponentB.setValue(0,0,0);
-        
-        constraintRow.m_solverBodyIdA = data.m_fixedBodyId;
-        constraintRow.m_solverBodyIdB = data.m_fixedBodyId;
-        
-        if (m_rigidBodyA)
-        {
-            constraintRow.m_solverBodyIdA = m_rigidBodyA->getCompanionId();
-        }
-        if (m_rigidBodyB)
-        {
-            constraintRow.m_solverBodyIdB = m_rigidBodyB->getCompanionId();
-        }
-        
-        btVector3 constraintNormalLin(0,0,0);
-        btVector3 constraintNormalAng(0,0,0);
-        btScalar posError = 0.0;
-        if (i < 2) {
-            constraintNormalLin = constraintAxis[i];
-            posError = (pivotAworld-pivotBworld).dot(constraintNormalLin);
-            fillMultiBodyConstraint(constraintRow, data, 0, 0, constraintNormalAng,
-                                    constraintNormalLin, pivotAworld, pivotBworld,
-                                    posError,
-                                    infoGlobal,
-                                    -m_maxAppliedImpulse, m_maxAppliedImpulse
-                                    );
-        }
-        else { //i>=2
-            constraintNormalAng = frameAworld.getColumn(i%3);
-            posError = angleDiff[i%3];
-            fillMultiBodyConstraint(constraintRow, data, 0, 0, constraintNormalAng,
-                                    constraintNormalLin, pivotAworld, pivotBworld,
-                                    posError,
-                                    infoGlobal,
-                                    -m_maxAppliedImpulse, m_maxAppliedImpulse, true
-                                    );
-        }
+		pivotAworld = m_rigidBodyA->getCenterOfMassTransform() * m_pivotInA;
+		frameAworld = m_frameInA.transpose() * btMatrix3x3(m_rigidBodyA->getOrientation());
+		jointAxis = quatRotate(m_rigidBodyA->getOrientation(), m_jointAxis);
+	}
+	else if (m_bodyA)
+	{
+		pivotAworld = m_bodyA->localPosToWorld(m_linkA, m_pivotInA);
+		frameAworld = m_bodyA->localFrameToWorld(m_linkA, m_frameInA);
+		jointAxis = m_bodyA->localDirToWorld(m_linkA, m_jointAxis);
+	}
+	btVector3 pivotBworld = m_pivotInB;
+	btMatrix3x3 frameBworld = m_frameInB;
+	if (m_rigidBodyB)
+	{
+		pivotBworld = m_rigidBodyB->getCenterOfMassTransform() * m_pivotInB;
+		frameBworld = m_frameInB.transpose() * btMatrix3x3(m_rigidBodyB->getOrientation());
+	}
+	else if (m_bodyB)
+	{
+		pivotBworld = m_bodyB->localPosToWorld(m_linkB, m_pivotInB);
+		frameBworld = m_bodyB->localFrameToWorld(m_linkB, m_frameInB);
+	}
+
+	btVector3 constraintAxis[2];
+	for (int i = 0; i < 3; ++i)
+	{
+		constraintAxis[0] = frameAworld.getColumn(i).cross(jointAxis);
+		if (constraintAxis[0].safeNorm() > EPSILON)
+		{
+			constraintAxis[0] = constraintAxis[0].normalized();
+			constraintAxis[1] = jointAxis.cross(constraintAxis[0]);
+			constraintAxis[1] = constraintAxis[1].normalized();
+			break;
+		}
+	}
+
+	btMatrix3x3 relRot = frameAworld.inverse() * frameBworld;
+	btVector3 angleDiff;
+	btGeneric6DofSpring2Constraint::matrixToEulerXYZ(relRot, angleDiff);
+
+	int numDim = BTMBSLIDERCONSTRAINT_DIM;
+	for (int i = 0; i < numDim; i++)
+	{
+		btMultiBodySolverConstraint& constraintRow = constraintRows.expandNonInitializing();
+		constraintRow.m_orgConstraint = this;
+		constraintRow.m_orgDofIndex = i;
+		constraintRow.m_relpos1CrossNormal.setValue(0, 0, 0);
+		constraintRow.m_contactNormal1.setValue(0, 0, 0);
+		constraintRow.m_relpos2CrossNormal.setValue(0, 0, 0);
+		constraintRow.m_contactNormal2.setValue(0, 0, 0);
+		constraintRow.m_angularComponentA.setValue(0, 0, 0);
+		constraintRow.m_angularComponentB.setValue(0, 0, 0);
+
+		constraintRow.m_solverBodyIdA = data.m_fixedBodyId;
+		constraintRow.m_solverBodyIdB = data.m_fixedBodyId;
+
+		if (m_rigidBodyA)
+		{
+			constraintRow.m_solverBodyIdA = m_rigidBodyA->getCompanionId();
+		}
+		if (m_rigidBodyB)
+		{
+			constraintRow.m_solverBodyIdB = m_rigidBodyB->getCompanionId();
+		}
+
+		btVector3 constraintNormalLin(0, 0, 0);
+		btVector3 constraintNormalAng(0, 0, 0);
+		btScalar posError = 0.0;
+		if (i < 2)
+		{
+			constraintNormalLin = constraintAxis[i];
+			posError = (pivotAworld - pivotBworld).dot(constraintNormalLin);
+			fillMultiBodyConstraint(constraintRow, data, 0, 0, constraintNormalAng,
+									constraintNormalLin, pivotAworld, pivotBworld,
+									posError,
+									infoGlobal,
+									-m_maxAppliedImpulse, m_maxAppliedImpulse);
+		}
+		else
+		{  //i>=2
+			constraintNormalAng = frameAworld.getColumn(i % 3);
+			posError = angleDiff[i % 3];
+			fillMultiBodyConstraint(constraintRow, data, 0, 0, constraintNormalAng,
+									constraintNormalLin, pivotAworld, pivotBworld,
+									posError,
+									infoGlobal,
+									-m_maxAppliedImpulse, m_maxAppliedImpulse, true);
+		}
 	}
 }
 
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodySliderConstraint.h b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodySliderConstraint.h
index 0a6cf3df12b..b192b6f8f37 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodySliderConstraint.h
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodySliderConstraint.h
@@ -23,17 +23,15 @@ subject to the following restrictions:
 class btMultiBodySliderConstraint : public btMultiBodyConstraint
 {
 protected:
-
-	btRigidBody*	m_rigidBodyA;
-	btRigidBody*	m_rigidBodyB;
-	btVector3		m_pivotInA;
-	btVector3		m_pivotInB;
-    btMatrix3x3     m_frameInA;
-    btMatrix3x3     m_frameInB;
-    btVector3       m_jointAxis;
+	btRigidBody* m_rigidBodyA;
+	btRigidBody* m_rigidBodyB;
+	btVector3 m_pivotInA;
+	btVector3 m_pivotInB;
+	btMatrix3x3 m_frameInA;
+	btMatrix3x3 m_frameInB;
+	btVector3 m_jointAxis;
 
 public:
-
 	btMultiBodySliderConstraint(btMultiBody* body, int link, btRigidBody* bodyB, const btVector3& pivotInA, const btVector3& pivotInB, const btMatrix3x3& frameInA, const btMatrix3x3& frameInB, const btVector3& jointAxis);
 	btMultiBodySliderConstraint(btMultiBody* bodyA, int linkA, btMultiBody* bodyB, int linkB, const btVector3& pivotInA, const btVector3& pivotInB, const btMatrix3x3& frameInA, const btMatrix3x3& frameInB, const btVector3& jointAxis);
 
@@ -45,18 +43,18 @@ public:
 	virtual int getIslandIdB() const;
 
 	virtual void createConstraintRows(btMultiBodyConstraintArray& constraintRows,
-		btMultiBodyJacobianData& data,
-		const btContactSolverInfo& infoGlobal);
+									  btMultiBodyJacobianData& data,
+									  const btContactSolverInfo& infoGlobal);
 
-    const btVector3& getPivotInA() const
-    {
-        return m_pivotInA;
-    }
-    
-    void setPivotInA(const btVector3& pivotInA)
-    {
-        m_pivotInA = pivotInA;
-    }
+	const btVector3& getPivotInA() const
+	{
+		return m_pivotInA;
+	}
+
+	void setPivotInA(const btVector3& pivotInA)
+	{
+		m_pivotInA = pivotInA;
+	}
 
 	const btVector3& getPivotInB() const
 	{
@@ -67,39 +65,38 @@ public:
 	{
 		m_pivotInB = pivotInB;
 	}
-    
-    const btMatrix3x3& getFrameInA() const
-    {
-        return m_frameInA;
-    }
-    
-    void setFrameInA(const btMatrix3x3& frameInA)
-    {
-        m_frameInA = frameInA;
-    }
-    
-    const btMatrix3x3& getFrameInB() const
-    {
-        return m_frameInB;
-    }
-    
-    virtual void setFrameInB(const btMatrix3x3& frameInB)
-    {
-        m_frameInB = frameInB;
-    }
-    
-    const btVector3& getJointAxis() const
-    {
-        return m_jointAxis;
-    }
-    
-    void setJointAxis(const btVector3& jointAxis)
-    {
-        m_jointAxis = jointAxis;
-    }
+
+	const btMatrix3x3& getFrameInA() const
+	{
+		return m_frameInA;
+	}
+
+	void setFrameInA(const btMatrix3x3& frameInA)
+	{
+		m_frameInA = frameInA;
+	}
+
+	const btMatrix3x3& getFrameInB() const
+	{
+		return m_frameInB;
+	}
+
+	virtual void setFrameInB(const btMatrix3x3& frameInB)
+	{
+		m_frameInB = frameInB;
+	}
+
+	const btVector3& getJointAxis() const
+	{
+		return m_jointAxis;
+	}
+
+	void setJointAxis(const btVector3& jointAxis)
+	{
+		m_jointAxis = jointAxis;
+	}
 
 	virtual void debugDraw(class btIDebugDraw* drawer);
-
 };
 
-#endif //BT_MULTIBODY_SLIDER_CONSTRAINT_H
+#endif  //BT_MULTIBODY_SLIDER_CONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodySolverConstraint.h b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodySolverConstraint.h
index 6fa1550e9e6..deed3e2a123 100644
--- a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodySolverConstraint.h
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodySolverConstraint.h
@@ -25,66 +25,66 @@ class btMultiBodyConstraint;
 #include "BulletDynamics/ConstraintSolver/btContactSolverInfo.h"
 
 ///1D constraint along a normal axis between bodyA and bodyB. It can be combined to solve contact and friction constraints.
-ATTRIBUTE_ALIGNED16 (struct)	btMultiBodySolverConstraint
+ATTRIBUTE_ALIGNED16(struct)
+btMultiBodySolverConstraint
 {
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-	btMultiBodySolverConstraint() : m_solverBodyIdA(-1), m_multiBodyA(0), m_linkA(-1), m_solverBodyIdB(-1), m_multiBodyB(0), m_linkB(-1),m_orgConstraint(0), m_orgDofIndex(-1)
-	{}
-
-	int				m_deltaVelAindex;//more generic version of m_relpos1CrossNormal/m_contactNormal1
-	int				m_jacAindex;
-	int				m_deltaVelBindex;
-	int				m_jacBindex;
-
-	btVector3		m_relpos1CrossNormal;
-	btVector3		m_contactNormal1;	
-	btVector3		m_relpos2CrossNormal;
-	btVector3		m_contactNormal2; //usually m_contactNormal2 == -m_contactNormal1, but not always
-	
-
-	btVector3		m_angularComponentA;
-	btVector3		m_angularComponentB;
-	
-	mutable btSimdScalar	m_appliedPushImpulse;
-	mutable btSimdScalar	m_appliedImpulse;
-
-	btScalar	m_friction;
-	btScalar	m_jacDiagABInv;
-	btScalar		m_rhs;
-	btScalar		m_cfm;
-	
-    btScalar		m_lowerLimit;
-	btScalar		m_upperLimit;
-	btScalar		m_rhsPenetration;
-    union
+	btMultiBodySolverConstraint() : m_solverBodyIdA(-1), m_multiBodyA(0), m_linkA(-1), m_solverBodyIdB(-1), m_multiBodyB(0), m_linkB(-1), m_orgConstraint(0), m_orgDofIndex(-1)
 	{
-		void*		m_originalContactPoint;
-		btScalar	m_unusedPadding4;
+	}
+
+	int m_deltaVelAindex;  //more generic version of m_relpos1CrossNormal/m_contactNormal1
+	int m_jacAindex;
+	int m_deltaVelBindex;
+	int m_jacBindex;
+
+	btVector3 m_relpos1CrossNormal;
+	btVector3 m_contactNormal1;
+	btVector3 m_relpos2CrossNormal;
+	btVector3 m_contactNormal2;  //usually m_contactNormal2 == -m_contactNormal1, but not always
+
+	btVector3 m_angularComponentA;
+	btVector3 m_angularComponentB;
+
+	mutable btSimdScalar m_appliedPushImpulse;
+	mutable btSimdScalar m_appliedImpulse;
+
+	btScalar m_friction;
+	btScalar m_jacDiagABInv;
+	btScalar m_rhs;
+	btScalar m_cfm;
+
+	btScalar m_lowerLimit;
+	btScalar m_upperLimit;
+	btScalar m_rhsPenetration;
+	union {
+		void* m_originalContactPoint;
+		btScalar m_unusedPadding4;
 	};
 
-	int	m_overrideNumSolverIterations;
-    int			m_frictionIndex;
+	int m_overrideNumSolverIterations;
+	int m_frictionIndex;
 
 	int m_solverBodyIdA;
 	btMultiBody* m_multiBodyA;
-	int			m_linkA;
-	
+	int m_linkA;
+
 	int m_solverBodyIdB;
 	btMultiBody* m_multiBodyB;
-	int			m_linkB;
+	int m_linkB;
 
 	//for writing back applied impulses
-	btMultiBodyConstraint*	m_orgConstraint;
+	btMultiBodyConstraint* m_orgConstraint;
 	int m_orgDofIndex;
 
-	enum		btSolverConstraintType
+	enum btSolverConstraintType
 	{
 		BT_SOLVER_CONTACT_1D = 0,
 		BT_SOLVER_FRICTION_1D
 	};
 };
 
-typedef btAlignedObjectArray<btMultiBodySolverConstraint>	btMultiBodyConstraintArray;
+typedef btAlignedObjectArray<btMultiBodySolverConstraint> btMultiBodyConstraintArray;
 
-#endif //BT_MULTIBODY_SOLVER_CONSTRAINT_H
+#endif  //BT_MULTIBODY_SOLVER_CONSTRAINT_H
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodySphericalJointMotor.cpp b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodySphericalJointMotor.cpp
new file mode 100644
index 00000000000..3e5aa30f283
--- /dev/null
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodySphericalJointMotor.cpp
@@ -0,0 +1,172 @@
+/*
+Bullet Continuous Collision Detection and Physics Library
+Copyright (c) 2018 Erwin Coumans  http://bulletphysics.org
+
+This software is provided 'as-is', without any express or implied warranty.
+In no event will the authors be held liable for any damages arising from the use of this software.
+Permission is granted to anyone to use this software for any purpose,
+including commercial applications, and to alter it and redistribute it freely,
+subject to the following restrictions:
+
+1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.
+2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.
+3. This notice may not be removed or altered from any source distribution.
+*/
+
+///This file was written by Erwin Coumans
+
+#include "btMultiBodySphericalJointMotor.h"
+#include "btMultiBody.h"
+#include "btMultiBodyLinkCollider.h"
+#include "BulletCollision/CollisionDispatch/btCollisionObject.h"
+#include "LinearMath/btTransformUtil.h"
+#include "BulletDynamics/ConstraintSolver/btGeneric6DofSpring2Constraint.h"
+
+btMultiBodySphericalJointMotor::btMultiBodySphericalJointMotor(btMultiBody* body, int link, btScalar maxMotorImpulse)
+	: btMultiBodyConstraint(body, body, link, body->getLink(link).m_parent, 3, true),
+	m_desiredVelocity(0, 0, 0),
+	m_desiredPosition(0,0,0,1),
+	m_kd(1.),
+	m_kp(0.2),
+	m_erp(1),
+	m_rhsClamp(SIMD_INFINITY)
+{
+
+	m_maxAppliedImpulse = maxMotorImpulse;
+}
+
+
+void btMultiBodySphericalJointMotor::finalizeMultiDof()
+{
+	allocateJacobiansMultiDof();
+	// note: we rely on the fact that data.m_jacobians are
+	// always initialized to zero by the Constraint ctor
+	int linkDoF = 0;
+	unsigned int offset = 6 + (m_bodyA->getLink(m_linkA).m_dofOffset + linkDoF);
+
+	// row 0: the lower bound
+	// row 0: the lower bound
+	jacobianA(0)[offset] = 1;
+
+	m_numDofsFinalized = m_jacSizeBoth;
+}
+
+
+btMultiBodySphericalJointMotor::~btMultiBodySphericalJointMotor()
+{
+}
+
+int btMultiBodySphericalJointMotor::getIslandIdA() const
+{
+	if (this->m_linkA < 0)
+	{
+		btMultiBodyLinkCollider* col = m_bodyA->getBaseCollider();
+		if (col)
+			return col->getIslandTag();
+	}
+	else
+	{
+		if (m_bodyA->getLink(m_linkA).m_collider)
+		{
+			return m_bodyA->getLink(m_linkA).m_collider->getIslandTag();
+		}
+	}
+	return -1;
+}
+
+int btMultiBodySphericalJointMotor::getIslandIdB() const
+{
+	if (m_linkB < 0)
+	{
+		btMultiBodyLinkCollider* col = m_bodyB->getBaseCollider();
+		if (col)
+			return col->getIslandTag();
+	}
+	else
+	{
+		if (m_bodyB->getLink(m_linkB).m_collider)
+		{
+			return m_bodyB->getLink(m_linkB).m_collider->getIslandTag();
+		}
+	}
+	return -1;
+}
+
+void btMultiBodySphericalJointMotor::createConstraintRows(btMultiBodyConstraintArray& constraintRows,
+												 btMultiBodyJacobianData& data,
+												 const btContactSolverInfo& infoGlobal)
+{
+	// only positions need to be updated -- data.m_jacobians and force
+	// directions were set in the ctor and never change.
+
+	if (m_numDofsFinalized != m_jacSizeBoth)
+	{
+		finalizeMultiDof();
+	}
+
+	//don't crash
+	if (m_numDofsFinalized != m_jacSizeBoth)
+		return;
+	
+
+	if (m_maxAppliedImpulse == 0.f)
+		return;
+
+	const btScalar posError = 0;
+	const btVector3 dummy(0, 0, 0);
+
+	
+	btVector3 axis[3] = { btVector3(1, 0, 0), btVector3(0, 1, 0), btVector3(0, 0, 1) };
+	
+	btQuaternion desiredQuat = m_desiredPosition;
+	btQuaternion currentQuat(m_bodyA->getJointPosMultiDof(m_linkA)[0],
+		m_bodyA->getJointPosMultiDof(m_linkA)[1],
+		m_bodyA->getJointPosMultiDof(m_linkA)[2],
+		m_bodyA->getJointPosMultiDof(m_linkA)[3]);
+
+btQuaternion relRot = currentQuat.inverse() * desiredQuat;
+	btVector3 angleDiff;
+	btGeneric6DofSpring2Constraint::matrixToEulerXYZ(btMatrix3x3(relRot), angleDiff);
+
+
+
+	for (int row = 0; row < getNumRows(); row++)
+	{
+		btMultiBodySolverConstraint& constraintRow = constraintRows.expandNonInitializing();
+
+		int dof = row;
+		
+		btScalar currentVelocity = m_bodyA->getJointVelMultiDof(m_linkA)[dof];
+		btScalar desiredVelocity = this->m_desiredVelocity[row];
+		
+		btScalar velocityError = desiredVelocity - currentVelocity;
+
+		btMatrix3x3 frameAworld;
+		frameAworld.setIdentity();
+		frameAworld = m_bodyA->localFrameToWorld(m_linkA, frameAworld);
+		btScalar posError = 0;
+		{
+			btAssert(m_bodyA->getLink(m_linkA).m_jointType == btMultibodyLink::eSpherical);
+			switch (m_bodyA->getLink(m_linkA).m_jointType)
+			{
+				case btMultibodyLink::eSpherical:
+				{
+					btVector3 constraintNormalAng = frameAworld.getColumn(row % 3);
+					posError = m_kp*angleDiff[row % 3];
+					fillMultiBodyConstraint(constraintRow, data, 0, 0, constraintNormalAng,
+						btVector3(0,0,0), dummy, dummy,
+						posError,
+						infoGlobal,
+						-m_maxAppliedImpulse, m_maxAppliedImpulse, true);
+					constraintRow.m_orgConstraint = this;
+					constraintRow.m_orgDofIndex = row;
+					break;
+				}
+				default:
+				{
+					btAssert(0);
+				}
+			};
+		}
+	}
+}
diff --git a/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodySphericalJointMotor.h b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodySphericalJointMotor.h
new file mode 100644
index 00000000000..621beab5a42
--- /dev/null
+++ b/thirdparty/bullet/BulletDynamics/Featherstone/btMultiBodySphericalJointMotor.h
@@ -0,0 +1,77 @@
+/*
+Bullet Continuous Collision Detection and Physics Library
+Copyright (c) 2018 Erwin Coumans  http://bulletphysics.org
+
+This software is provided 'as-is', without any express or implied warranty.
+In no event will the authors be held liable for any damages arising from the use of this software.
+Permission is granted to anyone to use this software for any purpose,
+including commercial applications, and to alter it and redistribute it freely,
+subject to the following restrictions:
+
+1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.
+2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.
+3. This notice may not be removed or altered from any source distribution.
+*/
+
+///This file was written by Erwin Coumans
+
+#ifndef BT_MULTIBODY_SPHERICAL_JOINT_MOTOR_H
+#define BT_MULTIBODY_SPHERICAL_JOINT_MOTOR_H
+
+#include "btMultiBodyConstraint.h"
+struct btSolverInfo;
+
+class btMultiBodySphericalJointMotor : public btMultiBodyConstraint
+{
+protected:
+	btVector3 m_desiredVelocity;
+	btQuaternion m_desiredPosition;
+	btScalar m_kd;
+	btScalar m_kp;
+	btScalar m_erp;
+	btScalar m_rhsClamp;  //maximum error
+
+public:
+	btMultiBodySphericalJointMotor(btMultiBody* body, int link, btScalar maxMotorImpulse);
+	
+	virtual ~btMultiBodySphericalJointMotor();
+	virtual void finalizeMultiDof();
+
+	virtual int getIslandIdA() const;
+	virtual int getIslandIdB() const;
+
+	virtual void createConstraintRows(btMultiBodyConstraintArray& constraintRows,
+									  btMultiBodyJacobianData& data,
+									  const btContactSolverInfo& infoGlobal);
+
+	virtual void setVelocityTarget(const btVector3& velTarget, btScalar kd = 1.f)
+	{
+		m_desiredVelocity = velTarget;
+		m_kd = kd;
+	}
+
+	virtual void setPositionTarget(const btQuaternion& posTarget, btScalar kp = 1.f)
+	{
+		m_desiredPosition = posTarget;
+		m_kp = kp;
+	}
+
+	virtual void setErp(btScalar erp)
+	{
+		m_erp = erp;
+	}
+	virtual btScalar getErp() const
+	{
+		return m_erp;
+	}
+	virtual void setRhsClamp(btScalar rhsClamp)
+	{
+		m_rhsClamp = rhsClamp;
+	}
+	virtual void debugDraw(class btIDebugDraw* drawer)
+	{
+		//todo(erwincoumans)
+	}
+};
+
+#endif  //BT_MULTIBODY_SPHERICAL_JOINT_MOTOR_H
diff --git a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btDantzigLCP.cpp b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btDantzigLCP.cpp
index 986f2148701..98ecdc07947 100644
--- a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btDantzigLCP.cpp
+++ b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btDantzigLCP.cpp
@@ -108,18 +108,16 @@ rows/columns and manipulate C.
 
 */
 
-
 #include "btDantzigLCP.h"
 
-#include <string.h>//memcpy
+#include <string.h>  //memcpy
 
 bool s_error = false;
 
 //***************************************************************************
 // code generation parameters
 
-
-#define btLCP_FAST		// use fast btLCP object
+#define btLCP_FAST  // use fast btLCP object
 
 // option 1 : matrix row pointers (less data copying)
 #define BTROWPTRS
@@ -133,8 +131,6 @@ bool s_error = false;
 
 #define BTNUB_OPTIMIZATIONS
 
-
-
 /* solve L*X=B, with B containing 1 right hand sides.
  * L is an n*n lower triangular matrix with ones on the diagonal.
  * L is stored by rows and its leading dimension is lskip.
@@ -145,66 +141,69 @@ bool s_error = false;
  * if this is in the factorizer source file, n must be a multiple of 2.
  */
 
-static void btSolveL1_1 (const btScalar *L, btScalar *B, int n, int lskip1)
-{  
-  /* declare variables - Z matrix, p and q vectors, etc */
-  btScalar Z11,m11,Z21,m21,p1,q1,p2,*ex;
-  const btScalar *ell;
-  int i,j;
-  /* compute all 2 x 1 blocks of X */
-  for (i=0; i < n; i+=2) {
-    /* compute all 2 x 1 block of X, from rows i..i+2-1 */
-    /* set the Z matrix to 0 */
-    Z11=0;
-    Z21=0;
-    ell = L + i*lskip1;
-    ex = B;
-    /* the inner loop that computes outer products and adds them to Z */
-    for (j=i-2; j >= 0; j -= 2) {
-      /* compute outer product and add it to the Z matrix */
-      p1=ell[0];
-      q1=ex[0];
-      m11 = p1 * q1;
-      p2=ell[lskip1];
-      m21 = p2 * q1;
-      Z11 += m11;
-      Z21 += m21;
-      /* compute outer product and add it to the Z matrix */
-      p1=ell[1];
-      q1=ex[1];
-      m11 = p1 * q1;
-      p2=ell[1+lskip1];
-      m21 = p2 * q1;
-      /* advance pointers */
-      ell += 2;
-      ex += 2;
-      Z11 += m11;
-      Z21 += m21;
-      /* end of inner loop */
-    }
-    /* compute left-over iterations */
-    j += 2;
-    for (; j > 0; j--) {
-      /* compute outer product and add it to the Z matrix */
-      p1=ell[0];
-      q1=ex[0];
-      m11 = p1 * q1;
-      p2=ell[lskip1];
-      m21 = p2 * q1;
-      /* advance pointers */
-      ell += 1;
-      ex += 1;
-      Z11 += m11;
-      Z21 += m21;
-    }
-    /* finish computing the X(i) block */
-    Z11 = ex[0] - Z11;
-    ex[0] = Z11;
-    p1 = ell[lskip1];
-    Z21 = ex[1] - Z21 - p1*Z11;
-    ex[1] = Z21;
-    /* end of outer loop */
-  }
+static void btSolveL1_1(const btScalar *L, btScalar *B, int n, int lskip1)
+{
+	/* declare variables - Z matrix, p and q vectors, etc */
+	btScalar Z11, m11, Z21, m21, p1, q1, p2, *ex;
+	const btScalar *ell;
+	int i, j;
+	/* compute all 2 x 1 blocks of X */
+	for (i = 0; i < n; i += 2)
+	{
+		/* compute all 2 x 1 block of X, from rows i..i+2-1 */
+		/* set the Z matrix to 0 */
+		Z11 = 0;
+		Z21 = 0;
+		ell = L + i * lskip1;
+		ex = B;
+		/* the inner loop that computes outer products and adds them to Z */
+		for (j = i - 2; j >= 0; j -= 2)
+		{
+			/* compute outer product and add it to the Z matrix */
+			p1 = ell[0];
+			q1 = ex[0];
+			m11 = p1 * q1;
+			p2 = ell[lskip1];
+			m21 = p2 * q1;
+			Z11 += m11;
+			Z21 += m21;
+			/* compute outer product and add it to the Z matrix */
+			p1 = ell[1];
+			q1 = ex[1];
+			m11 = p1 * q1;
+			p2 = ell[1 + lskip1];
+			m21 = p2 * q1;
+			/* advance pointers */
+			ell += 2;
+			ex += 2;
+			Z11 += m11;
+			Z21 += m21;
+			/* end of inner loop */
+		}
+		/* compute left-over iterations */
+		j += 2;
+		for (; j > 0; j--)
+		{
+			/* compute outer product and add it to the Z matrix */
+			p1 = ell[0];
+			q1 = ex[0];
+			m11 = p1 * q1;
+			p2 = ell[lskip1];
+			m21 = p2 * q1;
+			/* advance pointers */
+			ell += 1;
+			ex += 1;
+			Z11 += m11;
+			Z21 += m21;
+		}
+		/* finish computing the X(i) block */
+		Z11 = ex[0] - Z11;
+		ex[0] = Z11;
+		p1 = ell[lskip1];
+		Z21 = ex[1] - Z21 - p1 * Z11;
+		ex[1] = Z21;
+		/* end of outer loop */
+	}
 }
 
 /* solve L*X=B, with B containing 2 right hand sides.
@@ -217,300 +216,308 @@ static void btSolveL1_1 (const btScalar *L, btScalar *B, int n, int lskip1)
  * if this is in the factorizer source file, n must be a multiple of 2.
  */
 
-static void btSolveL1_2 (const btScalar *L, btScalar *B, int n, int lskip1)
-{  
-  /* declare variables - Z matrix, p and q vectors, etc */
-  btScalar Z11,m11,Z12,m12,Z21,m21,Z22,m22,p1,q1,p2,q2,*ex;
-  const btScalar *ell;
-  int i,j;
-  /* compute all 2 x 2 blocks of X */
-  for (i=0; i < n; i+=2) {
-    /* compute all 2 x 2 block of X, from rows i..i+2-1 */
-    /* set the Z matrix to 0 */
-    Z11=0;
-    Z12=0;
-    Z21=0;
-    Z22=0;
-    ell = L + i*lskip1;
-    ex = B;
-    /* the inner loop that computes outer products and adds them to Z */
-    for (j=i-2; j >= 0; j -= 2) {
-      /* compute outer product and add it to the Z matrix */
-      p1=ell[0];
-      q1=ex[0];
-      m11 = p1 * q1;
-      q2=ex[lskip1];
-      m12 = p1 * q2;
-      p2=ell[lskip1];
-      m21 = p2 * q1;
-      m22 = p2 * q2;
-      Z11 += m11;
-      Z12 += m12;
-      Z21 += m21;
-      Z22 += m22;
-      /* compute outer product and add it to the Z matrix */
-      p1=ell[1];
-      q1=ex[1];
-      m11 = p1 * q1;
-      q2=ex[1+lskip1];
-      m12 = p1 * q2;
-      p2=ell[1+lskip1];
-      m21 = p2 * q1;
-      m22 = p2 * q2;
-      /* advance pointers */
-      ell += 2;
-      ex += 2;
-      Z11 += m11;
-      Z12 += m12;
-      Z21 += m21;
-      Z22 += m22;
-      /* end of inner loop */
-    }
-    /* compute left-over iterations */
-    j += 2;
-    for (; j > 0; j--) {
-      /* compute outer product and add it to the Z matrix */
-      p1=ell[0];
-      q1=ex[0];
-      m11 = p1 * q1;
-      q2=ex[lskip1];
-      m12 = p1 * q2;
-      p2=ell[lskip1];
-      m21 = p2 * q1;
-      m22 = p2 * q2;
-      /* advance pointers */
-      ell += 1;
-      ex += 1;
-      Z11 += m11;
-      Z12 += m12;
-      Z21 += m21;
-      Z22 += m22;
-    }
-    /* finish computing the X(i) block */
-    Z11 = ex[0] - Z11;
-    ex[0] = Z11;
-    Z12 = ex[lskip1] - Z12;
-    ex[lskip1] = Z12;
-    p1 = ell[lskip1];
-    Z21 = ex[1] - Z21 - p1*Z11;
-    ex[1] = Z21;
-    Z22 = ex[1+lskip1] - Z22 - p1*Z12;
-    ex[1+lskip1] = Z22;
-    /* end of outer loop */
-  }
+static void btSolveL1_2(const btScalar *L, btScalar *B, int n, int lskip1)
+{
+	/* declare variables - Z matrix, p and q vectors, etc */
+	btScalar Z11, m11, Z12, m12, Z21, m21, Z22, m22, p1, q1, p2, q2, *ex;
+	const btScalar *ell;
+	int i, j;
+	/* compute all 2 x 2 blocks of X */
+	for (i = 0; i < n; i += 2)
+	{
+		/* compute all 2 x 2 block of X, from rows i..i+2-1 */
+		/* set the Z matrix to 0 */
+		Z11 = 0;
+		Z12 = 0;
+		Z21 = 0;
+		Z22 = 0;
+		ell = L + i * lskip1;
+		ex = B;
+		/* the inner loop that computes outer products and adds them to Z */
+		for (j = i - 2; j >= 0; j -= 2)
+		{
+			/* compute outer product and add it to the Z matrix */
+			p1 = ell[0];
+			q1 = ex[0];
+			m11 = p1 * q1;
+			q2 = ex[lskip1];
+			m12 = p1 * q2;
+			p2 = ell[lskip1];
+			m21 = p2 * q1;
+			m22 = p2 * q2;
+			Z11 += m11;
+			Z12 += m12;
+			Z21 += m21;
+			Z22 += m22;
+			/* compute outer product and add it to the Z matrix */
+			p1 = ell[1];
+			q1 = ex[1];
+			m11 = p1 * q1;
+			q2 = ex[1 + lskip1];
+			m12 = p1 * q2;
+			p2 = ell[1 + lskip1];
+			m21 = p2 * q1;
+			m22 = p2 * q2;
+			/* advance pointers */
+			ell += 2;
+			ex += 2;
+			Z11 += m11;
+			Z12 += m12;
+			Z21 += m21;
+			Z22 += m22;
+			/* end of inner loop */
+		}
+		/* compute left-over iterations */
+		j += 2;
+		for (; j > 0; j--)
+		{
+			/* compute outer product and add it to the Z matrix */
+			p1 = ell[0];
+			q1 = ex[0];
+			m11 = p1 * q1;
+			q2 = ex[lskip1];
+			m12 = p1 * q2;
+			p2 = ell[lskip1];
+			m21 = p2 * q1;
+			m22 = p2 * q2;
+			/* advance pointers */
+			ell += 1;
+			ex += 1;
+			Z11 += m11;
+			Z12 += m12;
+			Z21 += m21;
+			Z22 += m22;
+		}
+		/* finish computing the X(i) block */
+		Z11 = ex[0] - Z11;
+		ex[0] = Z11;
+		Z12 = ex[lskip1] - Z12;
+		ex[lskip1] = Z12;
+		p1 = ell[lskip1];
+		Z21 = ex[1] - Z21 - p1 * Z11;
+		ex[1] = Z21;
+		Z22 = ex[1 + lskip1] - Z22 - p1 * Z12;
+		ex[1 + lskip1] = Z22;
+		/* end of outer loop */
+	}
 }
 
+void btFactorLDLT(btScalar *A, btScalar *d, int n, int nskip1)
+{
+	int i, j;
+	btScalar sum, *ell, *dee, dd, p1, p2, q1, q2, Z11, m11, Z21, m21, Z22, m22;
+	if (n < 1) return;
 
-void btFactorLDLT (btScalar *A, btScalar *d, int n, int nskip1)
-{  
-  int i,j;
-  btScalar sum,*ell,*dee,dd,p1,p2,q1,q2,Z11,m11,Z21,m21,Z22,m22;
-  if (n < 1) return;
-  
-  for (i=0; i<=n-2; i += 2) {
-    /* solve L*(D*l)=a, l is scaled elements in 2 x i block at A(i,0) */
-    btSolveL1_2 (A,A+i*nskip1,i,nskip1);
-    /* scale the elements in a 2 x i block at A(i,0), and also */
-    /* compute Z = the outer product matrix that we'll need. */
-    Z11 = 0;
-    Z21 = 0;
-    Z22 = 0;
-    ell = A+i*nskip1;
-    dee = d;
-    for (j=i-6; j >= 0; j -= 6) {
-      p1 = ell[0];
-      p2 = ell[nskip1];
-      dd = dee[0];
-      q1 = p1*dd;
-      q2 = p2*dd;
-      ell[0] = q1;
-      ell[nskip1] = q2;
-      m11 = p1*q1;
-      m21 = p2*q1;
-      m22 = p2*q2;
-      Z11 += m11;
-      Z21 += m21;
-      Z22 += m22;
-      p1 = ell[1];
-      p2 = ell[1+nskip1];
-      dd = dee[1];
-      q1 = p1*dd;
-      q2 = p2*dd;
-      ell[1] = q1;
-      ell[1+nskip1] = q2;
-      m11 = p1*q1;
-      m21 = p2*q1;
-      m22 = p2*q2;
-      Z11 += m11;
-      Z21 += m21;
-      Z22 += m22;
-      p1 = ell[2];
-      p2 = ell[2+nskip1];
-      dd = dee[2];
-      q1 = p1*dd;
-      q2 = p2*dd;
-      ell[2] = q1;
-      ell[2+nskip1] = q2;
-      m11 = p1*q1;
-      m21 = p2*q1;
-      m22 = p2*q2;
-      Z11 += m11;
-      Z21 += m21;
-      Z22 += m22;
-      p1 = ell[3];
-      p2 = ell[3+nskip1];
-      dd = dee[3];
-      q1 = p1*dd;
-      q2 = p2*dd;
-      ell[3] = q1;
-      ell[3+nskip1] = q2;
-      m11 = p1*q1;
-      m21 = p2*q1;
-      m22 = p2*q2;
-      Z11 += m11;
-      Z21 += m21;
-      Z22 += m22;
-      p1 = ell[4];
-      p2 = ell[4+nskip1];
-      dd = dee[4];
-      q1 = p1*dd;
-      q2 = p2*dd;
-      ell[4] = q1;
-      ell[4+nskip1] = q2;
-      m11 = p1*q1;
-      m21 = p2*q1;
-      m22 = p2*q2;
-      Z11 += m11;
-      Z21 += m21;
-      Z22 += m22;
-      p1 = ell[5];
-      p2 = ell[5+nskip1];
-      dd = dee[5];
-      q1 = p1*dd;
-      q2 = p2*dd;
-      ell[5] = q1;
-      ell[5+nskip1] = q2;
-      m11 = p1*q1;
-      m21 = p2*q1;
-      m22 = p2*q2;
-      Z11 += m11;
-      Z21 += m21;
-      Z22 += m22;
-      ell += 6;
-      dee += 6;
-    }
-    /* compute left-over iterations */
-    j += 6;
-    for (; j > 0; j--) {
-      p1 = ell[0];
-      p2 = ell[nskip1];
-      dd = dee[0];
-      q1 = p1*dd;
-      q2 = p2*dd;
-      ell[0] = q1;
-      ell[nskip1] = q2;
-      m11 = p1*q1;
-      m21 = p2*q1;
-      m22 = p2*q2;
-      Z11 += m11;
-      Z21 += m21;
-      Z22 += m22;
-      ell++;
-      dee++;
-    }
-    /* solve for diagonal 2 x 2 block at A(i,i) */
-    Z11 = ell[0] - Z11;
-    Z21 = ell[nskip1] - Z21;
-    Z22 = ell[1+nskip1] - Z22;
-    dee = d + i;
-    /* factorize 2 x 2 block Z,dee */
-    /* factorize row 1 */
-    dee[0] = btRecip(Z11);
-    /* factorize row 2 */
-    sum = 0;
-    q1 = Z21;
-    q2 = q1 * dee[0];
-    Z21 = q2;
-    sum += q1*q2;
-    dee[1] = btRecip(Z22 - sum);
-    /* done factorizing 2 x 2 block */
-    ell[nskip1] = Z21;
-  }
-  /* compute the (less than 2) rows at the bottom */
-  switch (n-i) {
-    case 0:
-    break;
-    
-    case 1:
-    btSolveL1_1 (A,A+i*nskip1,i,nskip1);
-    /* scale the elements in a 1 x i block at A(i,0), and also */
-    /* compute Z = the outer product matrix that we'll need. */
-    Z11 = 0;
-    ell = A+i*nskip1;
-    dee = d;
-    for (j=i-6; j >= 0; j -= 6) {
-      p1 = ell[0];
-      dd = dee[0];
-      q1 = p1*dd;
-      ell[0] = q1;
-      m11 = p1*q1;
-      Z11 += m11;
-      p1 = ell[1];
-      dd = dee[1];
-      q1 = p1*dd;
-      ell[1] = q1;
-      m11 = p1*q1;
-      Z11 += m11;
-      p1 = ell[2];
-      dd = dee[2];
-      q1 = p1*dd;
-      ell[2] = q1;
-      m11 = p1*q1;
-      Z11 += m11;
-      p1 = ell[3];
-      dd = dee[3];
-      q1 = p1*dd;
-      ell[3] = q1;
-      m11 = p1*q1;
-      Z11 += m11;
-      p1 = ell[4];
-      dd = dee[4];
-      q1 = p1*dd;
-      ell[4] = q1;
-      m11 = p1*q1;
-      Z11 += m11;
-      p1 = ell[5];
-      dd = dee[5];
-      q1 = p1*dd;
-      ell[5] = q1;
-      m11 = p1*q1;
-      Z11 += m11;
-      ell += 6;
-      dee += 6;
-    }
-    /* compute left-over iterations */
-    j += 6;
-    for (; j > 0; j--) {
-      p1 = ell[0];
-      dd = dee[0];
-      q1 = p1*dd;
-      ell[0] = q1;
-      m11 = p1*q1;
-      Z11 += m11;
-      ell++;
-      dee++;
-    }
-    /* solve for diagonal 1 x 1 block at A(i,i) */
-    Z11 = ell[0] - Z11;
-    dee = d + i;
-    /* factorize 1 x 1 block Z,dee */
-    /* factorize row 1 */
-    dee[0] = btRecip(Z11);
-    /* done factorizing 1 x 1 block */
-    break;
-    
-    //default: *((char*)0)=0;  /* this should never happen! */
-  }
+	for (i = 0; i <= n - 2; i += 2)
+	{
+		/* solve L*(D*l)=a, l is scaled elements in 2 x i block at A(i,0) */
+		btSolveL1_2(A, A + i * nskip1, i, nskip1);
+		/* scale the elements in a 2 x i block at A(i,0), and also */
+		/* compute Z = the outer product matrix that we'll need. */
+		Z11 = 0;
+		Z21 = 0;
+		Z22 = 0;
+		ell = A + i * nskip1;
+		dee = d;
+		for (j = i - 6; j >= 0; j -= 6)
+		{
+			p1 = ell[0];
+			p2 = ell[nskip1];
+			dd = dee[0];
+			q1 = p1 * dd;
+			q2 = p2 * dd;
+			ell[0] = q1;
+			ell[nskip1] = q2;
+			m11 = p1 * q1;
+			m21 = p2 * q1;
+			m22 = p2 * q2;
+			Z11 += m11;
+			Z21 += m21;
+			Z22 += m22;
+			p1 = ell[1];
+			p2 = ell[1 + nskip1];
+			dd = dee[1];
+			q1 = p1 * dd;
+			q2 = p2 * dd;
+			ell[1] = q1;
+			ell[1 + nskip1] = q2;
+			m11 = p1 * q1;
+			m21 = p2 * q1;
+			m22 = p2 * q2;
+			Z11 += m11;
+			Z21 += m21;
+			Z22 += m22;
+			p1 = ell[2];
+			p2 = ell[2 + nskip1];
+			dd = dee[2];
+			q1 = p1 * dd;
+			q2 = p2 * dd;
+			ell[2] = q1;
+			ell[2 + nskip1] = q2;
+			m11 = p1 * q1;
+			m21 = p2 * q1;
+			m22 = p2 * q2;
+			Z11 += m11;
+			Z21 += m21;
+			Z22 += m22;
+			p1 = ell[3];
+			p2 = ell[3 + nskip1];
+			dd = dee[3];
+			q1 = p1 * dd;
+			q2 = p2 * dd;
+			ell[3] = q1;
+			ell[3 + nskip1] = q2;
+			m11 = p1 * q1;
+			m21 = p2 * q1;
+			m22 = p2 * q2;
+			Z11 += m11;
+			Z21 += m21;
+			Z22 += m22;
+			p1 = ell[4];
+			p2 = ell[4 + nskip1];
+			dd = dee[4];
+			q1 = p1 * dd;
+			q2 = p2 * dd;
+			ell[4] = q1;
+			ell[4 + nskip1] = q2;
+			m11 = p1 * q1;
+			m21 = p2 * q1;
+			m22 = p2 * q2;
+			Z11 += m11;
+			Z21 += m21;
+			Z22 += m22;
+			p1 = ell[5];
+			p2 = ell[5 + nskip1];
+			dd = dee[5];
+			q1 = p1 * dd;
+			q2 = p2 * dd;
+			ell[5] = q1;
+			ell[5 + nskip1] = q2;
+			m11 = p1 * q1;
+			m21 = p2 * q1;
+			m22 = p2 * q2;
+			Z11 += m11;
+			Z21 += m21;
+			Z22 += m22;
+			ell += 6;
+			dee += 6;
+		}
+		/* compute left-over iterations */
+		j += 6;
+		for (; j > 0; j--)
+		{
+			p1 = ell[0];
+			p2 = ell[nskip1];
+			dd = dee[0];
+			q1 = p1 * dd;
+			q2 = p2 * dd;
+			ell[0] = q1;
+			ell[nskip1] = q2;
+			m11 = p1 * q1;
+			m21 = p2 * q1;
+			m22 = p2 * q2;
+			Z11 += m11;
+			Z21 += m21;
+			Z22 += m22;
+			ell++;
+			dee++;
+		}
+		/* solve for diagonal 2 x 2 block at A(i,i) */
+		Z11 = ell[0] - Z11;
+		Z21 = ell[nskip1] - Z21;
+		Z22 = ell[1 + nskip1] - Z22;
+		dee = d + i;
+		/* factorize 2 x 2 block Z,dee */
+		/* factorize row 1 */
+		dee[0] = btRecip(Z11);
+		/* factorize row 2 */
+		sum = 0;
+		q1 = Z21;
+		q2 = q1 * dee[0];
+		Z21 = q2;
+		sum += q1 * q2;
+		dee[1] = btRecip(Z22 - sum);
+		/* done factorizing 2 x 2 block */
+		ell[nskip1] = Z21;
+	}
+	/* compute the (less than 2) rows at the bottom */
+	switch (n - i)
+	{
+		case 0:
+			break;
+
+		case 1:
+			btSolveL1_1(A, A + i * nskip1, i, nskip1);
+			/* scale the elements in a 1 x i block at A(i,0), and also */
+			/* compute Z = the outer product matrix that we'll need. */
+			Z11 = 0;
+			ell = A + i * nskip1;
+			dee = d;
+			for (j = i - 6; j >= 0; j -= 6)
+			{
+				p1 = ell[0];
+				dd = dee[0];
+				q1 = p1 * dd;
+				ell[0] = q1;
+				m11 = p1 * q1;
+				Z11 += m11;
+				p1 = ell[1];
+				dd = dee[1];
+				q1 = p1 * dd;
+				ell[1] = q1;
+				m11 = p1 * q1;
+				Z11 += m11;
+				p1 = ell[2];
+				dd = dee[2];
+				q1 = p1 * dd;
+				ell[2] = q1;
+				m11 = p1 * q1;
+				Z11 += m11;
+				p1 = ell[3];
+				dd = dee[3];
+				q1 = p1 * dd;
+				ell[3] = q1;
+				m11 = p1 * q1;
+				Z11 += m11;
+				p1 = ell[4];
+				dd = dee[4];
+				q1 = p1 * dd;
+				ell[4] = q1;
+				m11 = p1 * q1;
+				Z11 += m11;
+				p1 = ell[5];
+				dd = dee[5];
+				q1 = p1 * dd;
+				ell[5] = q1;
+				m11 = p1 * q1;
+				Z11 += m11;
+				ell += 6;
+				dee += 6;
+			}
+			/* compute left-over iterations */
+			j += 6;
+			for (; j > 0; j--)
+			{
+				p1 = ell[0];
+				dd = dee[0];
+				q1 = p1 * dd;
+				ell[0] = q1;
+				m11 = p1 * q1;
+				Z11 += m11;
+				ell++;
+				dee++;
+			}
+			/* solve for diagonal 1 x 1 block at A(i,i) */
+			Z11 = ell[0] - Z11;
+			dee = d + i;
+			/* factorize 1 x 1 block Z,dee */
+			/* factorize row 1 */
+			dee[0] = btRecip(Z11);
+			/* done factorizing 1 x 1 block */
+			break;
+
+			//default: *((char*)0)=0;  /* this should never happen! */
+	}
 }
 
 /* solve L*X=B, with B containing 1 right hand sides.
@@ -523,289 +530,295 @@ void btFactorLDLT (btScalar *A, btScalar *d, int n, int nskip1)
  * if this is in the factorizer source file, n must be a multiple of 4.
  */
 
-void btSolveL1 (const btScalar *L, btScalar *B, int n, int lskip1)
-{  
-  /* declare variables - Z matrix, p and q vectors, etc */
-  btScalar Z11,Z21,Z31,Z41,p1,q1,p2,p3,p4,*ex;
-  const btScalar *ell;
-  int lskip2,lskip3,i,j;
-  /* compute lskip values */
-  lskip2 = 2*lskip1;
-  lskip3 = 3*lskip1;
-  /* compute all 4 x 1 blocks of X */
-  for (i=0; i <= n-4; i+=4) {
-    /* compute all 4 x 1 block of X, from rows i..i+4-1 */
-    /* set the Z matrix to 0 */
-    Z11=0;
-    Z21=0;
-    Z31=0;
-    Z41=0;
-    ell = L + i*lskip1;
-    ex = B;
-    /* the inner loop that computes outer products and adds them to Z */
-    for (j=i-12; j >= 0; j -= 12) {
-      /* load p and q values */
-      p1=ell[0];
-      q1=ex[0];
-      p2=ell[lskip1];
-      p3=ell[lskip2];
-      p4=ell[lskip3];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      Z21 += p2 * q1;
-      Z31 += p3 * q1;
-      Z41 += p4 * q1;
-      /* load p and q values */
-      p1=ell[1];
-      q1=ex[1];
-      p2=ell[1+lskip1];
-      p3=ell[1+lskip2];
-      p4=ell[1+lskip3];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      Z21 += p2 * q1;
-      Z31 += p3 * q1;
-      Z41 += p4 * q1;
-      /* load p and q values */
-      p1=ell[2];
-      q1=ex[2];
-      p2=ell[2+lskip1];
-      p3=ell[2+lskip2];
-      p4=ell[2+lskip3];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      Z21 += p2 * q1;
-      Z31 += p3 * q1;
-      Z41 += p4 * q1;
-      /* load p and q values */
-      p1=ell[3];
-      q1=ex[3];
-      p2=ell[3+lskip1];
-      p3=ell[3+lskip2];
-      p4=ell[3+lskip3];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      Z21 += p2 * q1;
-      Z31 += p3 * q1;
-      Z41 += p4 * q1;
-      /* load p and q values */
-      p1=ell[4];
-      q1=ex[4];
-      p2=ell[4+lskip1];
-      p3=ell[4+lskip2];
-      p4=ell[4+lskip3];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      Z21 += p2 * q1;
-      Z31 += p3 * q1;
-      Z41 += p4 * q1;
-      /* load p and q values */
-      p1=ell[5];
-      q1=ex[5];
-      p2=ell[5+lskip1];
-      p3=ell[5+lskip2];
-      p4=ell[5+lskip3];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      Z21 += p2 * q1;
-      Z31 += p3 * q1;
-      Z41 += p4 * q1;
-      /* load p and q values */
-      p1=ell[6];
-      q1=ex[6];
-      p2=ell[6+lskip1];
-      p3=ell[6+lskip2];
-      p4=ell[6+lskip3];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      Z21 += p2 * q1;
-      Z31 += p3 * q1;
-      Z41 += p4 * q1;
-      /* load p and q values */
-      p1=ell[7];
-      q1=ex[7];
-      p2=ell[7+lskip1];
-      p3=ell[7+lskip2];
-      p4=ell[7+lskip3];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      Z21 += p2 * q1;
-      Z31 += p3 * q1;
-      Z41 += p4 * q1;
-      /* load p and q values */
-      p1=ell[8];
-      q1=ex[8];
-      p2=ell[8+lskip1];
-      p3=ell[8+lskip2];
-      p4=ell[8+lskip3];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      Z21 += p2 * q1;
-      Z31 += p3 * q1;
-      Z41 += p4 * q1;
-      /* load p and q values */
-      p1=ell[9];
-      q1=ex[9];
-      p2=ell[9+lskip1];
-      p3=ell[9+lskip2];
-      p4=ell[9+lskip3];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      Z21 += p2 * q1;
-      Z31 += p3 * q1;
-      Z41 += p4 * q1;
-      /* load p and q values */
-      p1=ell[10];
-      q1=ex[10];
-      p2=ell[10+lskip1];
-      p3=ell[10+lskip2];
-      p4=ell[10+lskip3];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      Z21 += p2 * q1;
-      Z31 += p3 * q1;
-      Z41 += p4 * q1;
-      /* load p and q values */
-      p1=ell[11];
-      q1=ex[11];
-      p2=ell[11+lskip1];
-      p3=ell[11+lskip2];
-      p4=ell[11+lskip3];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      Z21 += p2 * q1;
-      Z31 += p3 * q1;
-      Z41 += p4 * q1;
-      /* advance pointers */
-      ell += 12;
-      ex += 12;
-      /* end of inner loop */
-    }
-    /* compute left-over iterations */
-    j += 12;
-    for (; j > 0; j--) {
-      /* load p and q values */
-      p1=ell[0];
-      q1=ex[0];
-      p2=ell[lskip1];
-      p3=ell[lskip2];
-      p4=ell[lskip3];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      Z21 += p2 * q1;
-      Z31 += p3 * q1;
-      Z41 += p4 * q1;
-      /* advance pointers */
-      ell += 1;
-      ex += 1;
-    }
-    /* finish computing the X(i) block */
-    Z11 = ex[0] - Z11;
-    ex[0] = Z11;
-    p1 = ell[lskip1];
-    Z21 = ex[1] - Z21 - p1*Z11;
-    ex[1] = Z21;
-    p1 = ell[lskip2];
-    p2 = ell[1+lskip2];
-    Z31 = ex[2] - Z31 - p1*Z11 - p2*Z21;
-    ex[2] = Z31;
-    p1 = ell[lskip3];
-    p2 = ell[1+lskip3];
-    p3 = ell[2+lskip3];
-    Z41 = ex[3] - Z41 - p1*Z11 - p2*Z21 - p3*Z31;
-    ex[3] = Z41;
-    /* end of outer loop */
-  }
-  /* compute rows at end that are not a multiple of block size */
-  for (; i < n; i++) {
-    /* compute all 1 x 1 block of X, from rows i..i+1-1 */
-    /* set the Z matrix to 0 */
-    Z11=0;
-    ell = L + i*lskip1;
-    ex = B;
-    /* the inner loop that computes outer products and adds them to Z */
-    for (j=i-12; j >= 0; j -= 12) {
-      /* load p and q values */
-      p1=ell[0];
-      q1=ex[0];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      /* load p and q values */
-      p1=ell[1];
-      q1=ex[1];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      /* load p and q values */
-      p1=ell[2];
-      q1=ex[2];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      /* load p and q values */
-      p1=ell[3];
-      q1=ex[3];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      /* load p and q values */
-      p1=ell[4];
-      q1=ex[4];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      /* load p and q values */
-      p1=ell[5];
-      q1=ex[5];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      /* load p and q values */
-      p1=ell[6];
-      q1=ex[6];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      /* load p and q values */
-      p1=ell[7];
-      q1=ex[7];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      /* load p and q values */
-      p1=ell[8];
-      q1=ex[8];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      /* load p and q values */
-      p1=ell[9];
-      q1=ex[9];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      /* load p and q values */
-      p1=ell[10];
-      q1=ex[10];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      /* load p and q values */
-      p1=ell[11];
-      q1=ex[11];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      /* advance pointers */
-      ell += 12;
-      ex += 12;
-      /* end of inner loop */
-    }
-    /* compute left-over iterations */
-    j += 12;
-    for (; j > 0; j--) {
-      /* load p and q values */
-      p1=ell[0];
-      q1=ex[0];
-      /* compute outer product and add it to the Z matrix */
-      Z11 += p1 * q1;
-      /* advance pointers */
-      ell += 1;
-      ex += 1;
-    }
-    /* finish computing the X(i) block */
-    Z11 = ex[0] - Z11;
-    ex[0] = Z11;
-  }
+void btSolveL1(const btScalar *L, btScalar *B, int n, int lskip1)
+{
+	/* declare variables - Z matrix, p and q vectors, etc */
+	btScalar Z11, Z21, Z31, Z41, p1, q1, p2, p3, p4, *ex;
+	const btScalar *ell;
+	int lskip2, lskip3, i, j;
+	/* compute lskip values */
+	lskip2 = 2 * lskip1;
+	lskip3 = 3 * lskip1;
+	/* compute all 4 x 1 blocks of X */
+	for (i = 0; i <= n - 4; i += 4)
+	{
+		/* compute all 4 x 1 block of X, from rows i..i+4-1 */
+		/* set the Z matrix to 0 */
+		Z11 = 0;
+		Z21 = 0;
+		Z31 = 0;
+		Z41 = 0;
+		ell = L + i * lskip1;
+		ex = B;
+		/* the inner loop that computes outer products and adds them to Z */
+		for (j = i - 12; j >= 0; j -= 12)
+		{
+			/* load p and q values */
+			p1 = ell[0];
+			q1 = ex[0];
+			p2 = ell[lskip1];
+			p3 = ell[lskip2];
+			p4 = ell[lskip3];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			Z21 += p2 * q1;
+			Z31 += p3 * q1;
+			Z41 += p4 * q1;
+			/* load p and q values */
+			p1 = ell[1];
+			q1 = ex[1];
+			p2 = ell[1 + lskip1];
+			p3 = ell[1 + lskip2];
+			p4 = ell[1 + lskip3];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			Z21 += p2 * q1;
+			Z31 += p3 * q1;
+			Z41 += p4 * q1;
+			/* load p and q values */
+			p1 = ell[2];
+			q1 = ex[2];
+			p2 = ell[2 + lskip1];
+			p3 = ell[2 + lskip2];
+			p4 = ell[2 + lskip3];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			Z21 += p2 * q1;
+			Z31 += p3 * q1;
+			Z41 += p4 * q1;
+			/* load p and q values */
+			p1 = ell[3];
+			q1 = ex[3];
+			p2 = ell[3 + lskip1];
+			p3 = ell[3 + lskip2];
+			p4 = ell[3 + lskip3];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			Z21 += p2 * q1;
+			Z31 += p3 * q1;
+			Z41 += p4 * q1;
+			/* load p and q values */
+			p1 = ell[4];
+			q1 = ex[4];
+			p2 = ell[4 + lskip1];
+			p3 = ell[4 + lskip2];
+			p4 = ell[4 + lskip3];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			Z21 += p2 * q1;
+			Z31 += p3 * q1;
+			Z41 += p4 * q1;
+			/* load p and q values */
+			p1 = ell[5];
+			q1 = ex[5];
+			p2 = ell[5 + lskip1];
+			p3 = ell[5 + lskip2];
+			p4 = ell[5 + lskip3];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			Z21 += p2 * q1;
+			Z31 += p3 * q1;
+			Z41 += p4 * q1;
+			/* load p and q values */
+			p1 = ell[6];
+			q1 = ex[6];
+			p2 = ell[6 + lskip1];
+			p3 = ell[6 + lskip2];
+			p4 = ell[6 + lskip3];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			Z21 += p2 * q1;
+			Z31 += p3 * q1;
+			Z41 += p4 * q1;
+			/* load p and q values */
+			p1 = ell[7];
+			q1 = ex[7];
+			p2 = ell[7 + lskip1];
+			p3 = ell[7 + lskip2];
+			p4 = ell[7 + lskip3];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			Z21 += p2 * q1;
+			Z31 += p3 * q1;
+			Z41 += p4 * q1;
+			/* load p and q values */
+			p1 = ell[8];
+			q1 = ex[8];
+			p2 = ell[8 + lskip1];
+			p3 = ell[8 + lskip2];
+			p4 = ell[8 + lskip3];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			Z21 += p2 * q1;
+			Z31 += p3 * q1;
+			Z41 += p4 * q1;
+			/* load p and q values */
+			p1 = ell[9];
+			q1 = ex[9];
+			p2 = ell[9 + lskip1];
+			p3 = ell[9 + lskip2];
+			p4 = ell[9 + lskip3];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			Z21 += p2 * q1;
+			Z31 += p3 * q1;
+			Z41 += p4 * q1;
+			/* load p and q values */
+			p1 = ell[10];
+			q1 = ex[10];
+			p2 = ell[10 + lskip1];
+			p3 = ell[10 + lskip2];
+			p4 = ell[10 + lskip3];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			Z21 += p2 * q1;
+			Z31 += p3 * q1;
+			Z41 += p4 * q1;
+			/* load p and q values */
+			p1 = ell[11];
+			q1 = ex[11];
+			p2 = ell[11 + lskip1];
+			p3 = ell[11 + lskip2];
+			p4 = ell[11 + lskip3];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			Z21 += p2 * q1;
+			Z31 += p3 * q1;
+			Z41 += p4 * q1;
+			/* advance pointers */
+			ell += 12;
+			ex += 12;
+			/* end of inner loop */
+		}
+		/* compute left-over iterations */
+		j += 12;
+		for (; j > 0; j--)
+		{
+			/* load p and q values */
+			p1 = ell[0];
+			q1 = ex[0];
+			p2 = ell[lskip1];
+			p3 = ell[lskip2];
+			p4 = ell[lskip3];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			Z21 += p2 * q1;
+			Z31 += p3 * q1;
+			Z41 += p4 * q1;
+			/* advance pointers */
+			ell += 1;
+			ex += 1;
+		}
+		/* finish computing the X(i) block */
+		Z11 = ex[0] - Z11;
+		ex[0] = Z11;
+		p1 = ell[lskip1];
+		Z21 = ex[1] - Z21 - p1 * Z11;
+		ex[1] = Z21;
+		p1 = ell[lskip2];
+		p2 = ell[1 + lskip2];
+		Z31 = ex[2] - Z31 - p1 * Z11 - p2 * Z21;
+		ex[2] = Z31;
+		p1 = ell[lskip3];
+		p2 = ell[1 + lskip3];
+		p3 = ell[2 + lskip3];
+		Z41 = ex[3] - Z41 - p1 * Z11 - p2 * Z21 - p3 * Z31;
+		ex[3] = Z41;
+		/* end of outer loop */
+	}
+	/* compute rows at end that are not a multiple of block size */
+	for (; i < n; i++)
+	{
+		/* compute all 1 x 1 block of X, from rows i..i+1-1 */
+		/* set the Z matrix to 0 */
+		Z11 = 0;
+		ell = L + i * lskip1;
+		ex = B;
+		/* the inner loop that computes outer products and adds them to Z */
+		for (j = i - 12; j >= 0; j -= 12)
+		{
+			/* load p and q values */
+			p1 = ell[0];
+			q1 = ex[0];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			/* load p and q values */
+			p1 = ell[1];
+			q1 = ex[1];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			/* load p and q values */
+			p1 = ell[2];
+			q1 = ex[2];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			/* load p and q values */
+			p1 = ell[3];
+			q1 = ex[3];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			/* load p and q values */
+			p1 = ell[4];
+			q1 = ex[4];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			/* load p and q values */
+			p1 = ell[5];
+			q1 = ex[5];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			/* load p and q values */
+			p1 = ell[6];
+			q1 = ex[6];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			/* load p and q values */
+			p1 = ell[7];
+			q1 = ex[7];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			/* load p and q values */
+			p1 = ell[8];
+			q1 = ex[8];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			/* load p and q values */
+			p1 = ell[9];
+			q1 = ex[9];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			/* load p and q values */
+			p1 = ell[10];
+			q1 = ex[10];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			/* load p and q values */
+			p1 = ell[11];
+			q1 = ex[11];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			/* advance pointers */
+			ell += 12;
+			ex += 12;
+			/* end of inner loop */
+		}
+		/* compute left-over iterations */
+		j += 12;
+		for (; j > 0; j--)
+		{
+			/* load p and q values */
+			p1 = ell[0];
+			q1 = ex[0];
+			/* compute outer product and add it to the Z matrix */
+			Z11 += p1 * q1;
+			/* advance pointers */
+			ell += 1;
+			ex += 1;
+		}
+		/* finish computing the X(i) block */
+		Z11 = ex[0] - Z11;
+		ex[0] = Z11;
+	}
 }
 
 /* solve L^T * x=b, with b containing 1 right hand side.
@@ -816,214 +829,217 @@ void btSolveL1 (const btScalar *L, btScalar *B, int n, int lskip1)
  * this processes blocks of 4.
  */
 
-void btSolveL1T (const btScalar *L, btScalar *B, int n, int lskip1)
-{  
-  /* declare variables - Z matrix, p and q vectors, etc */
-  btScalar Z11,m11,Z21,m21,Z31,m31,Z41,m41,p1,q1,p2,p3,p4,*ex;
-  const btScalar *ell;
-  int lskip2,i,j;
-//  int lskip3;
-  /* special handling for L and B because we're solving L1 *transpose* */
-  L = L + (n-1)*(lskip1+1);
-  B = B + n-1;
-  lskip1 = -lskip1;
-  /* compute lskip values */
-  lskip2 = 2*lskip1;
-  //lskip3 = 3*lskip1;
-  /* compute all 4 x 1 blocks of X */
-  for (i=0; i <= n-4; i+=4) {
-    /* compute all 4 x 1 block of X, from rows i..i+4-1 */
-    /* set the Z matrix to 0 */
-    Z11=0;
-    Z21=0;
-    Z31=0;
-    Z41=0;
-    ell = L - i;
-    ex = B;
-    /* the inner loop that computes outer products and adds them to Z */
-    for (j=i-4; j >= 0; j -= 4) {
-      /* load p and q values */
-      p1=ell[0];
-      q1=ex[0];
-      p2=ell[-1];
-      p3=ell[-2];
-      p4=ell[-3];
-      /* compute outer product and add it to the Z matrix */
-      m11 = p1 * q1;
-      m21 = p2 * q1;
-      m31 = p3 * q1;
-      m41 = p4 * q1;
-      ell += lskip1;
-      Z11 += m11;
-      Z21 += m21;
-      Z31 += m31;
-      Z41 += m41;
-      /* load p and q values */
-      p1=ell[0];
-      q1=ex[-1];
-      p2=ell[-1];
-      p3=ell[-2];
-      p4=ell[-3];
-      /* compute outer product and add it to the Z matrix */
-      m11 = p1 * q1;
-      m21 = p2 * q1;
-      m31 = p3 * q1;
-      m41 = p4 * q1;
-      ell += lskip1;
-      Z11 += m11;
-      Z21 += m21;
-      Z31 += m31;
-      Z41 += m41;
-      /* load p and q values */
-      p1=ell[0];
-      q1=ex[-2];
-      p2=ell[-1];
-      p3=ell[-2];
-      p4=ell[-3];
-      /* compute outer product and add it to the Z matrix */
-      m11 = p1 * q1;
-      m21 = p2 * q1;
-      m31 = p3 * q1;
-      m41 = p4 * q1;
-      ell += lskip1;
-      Z11 += m11;
-      Z21 += m21;
-      Z31 += m31;
-      Z41 += m41;
-      /* load p and q values */
-      p1=ell[0];
-      q1=ex[-3];
-      p2=ell[-1];
-      p3=ell[-2];
-      p4=ell[-3];
-      /* compute outer product and add it to the Z matrix */
-      m11 = p1 * q1;
-      m21 = p2 * q1;
-      m31 = p3 * q1;
-      m41 = p4 * q1;
-      ell += lskip1;
-      ex -= 4;
-      Z11 += m11;
-      Z21 += m21;
-      Z31 += m31;
-      Z41 += m41;
-      /* end of inner loop */
-    }
-    /* compute left-over iterations */
-    j += 4;
-    for (; j > 0; j--) {
-      /* load p and q values */
-      p1=ell[0];
-      q1=ex[0];
-      p2=ell[-1];
-      p3=ell[-2];
-      p4=ell[-3];
-      /* compute outer product and add it to the Z matrix */
-      m11 = p1 * q1;
-      m21 = p2 * q1;
-      m31 = p3 * q1;
-      m41 = p4 * q1;
-      ell += lskip1;
-      ex -= 1;
-      Z11 += m11;
-      Z21 += m21;
-      Z31 += m31;
-      Z41 += m41;
-    }
-    /* finish computing the X(i) block */
-    Z11 = ex[0] - Z11;
-    ex[0] = Z11;
-    p1 = ell[-1];
-    Z21 = ex[-1] - Z21 - p1*Z11;
-    ex[-1] = Z21;
-    p1 = ell[-2];
-    p2 = ell[-2+lskip1];
-    Z31 = ex[-2] - Z31 - p1*Z11 - p2*Z21;
-    ex[-2] = Z31;
-    p1 = ell[-3];
-    p2 = ell[-3+lskip1];
-    p3 = ell[-3+lskip2];
-    Z41 = ex[-3] - Z41 - p1*Z11 - p2*Z21 - p3*Z31;
-    ex[-3] = Z41;
-    /* end of outer loop */
-  }
-  /* compute rows at end that are not a multiple of block size */
-  for (; i < n; i++) {
-    /* compute all 1 x 1 block of X, from rows i..i+1-1 */
-    /* set the Z matrix to 0 */
-    Z11=0;
-    ell = L - i;
-    ex = B;
-    /* the inner loop that computes outer products and adds them to Z */
-    for (j=i-4; j >= 0; j -= 4) {
-      /* load p and q values */
-      p1=ell[0];
-      q1=ex[0];
-      /* compute outer product and add it to the Z matrix */
-      m11 = p1 * q1;
-      ell += lskip1;
-      Z11 += m11;
-      /* load p and q values */
-      p1=ell[0];
-      q1=ex[-1];
-      /* compute outer product and add it to the Z matrix */
-      m11 = p1 * q1;
-      ell += lskip1;
-      Z11 += m11;
-      /* load p and q values */
-      p1=ell[0];
-      q1=ex[-2];
-      /* compute outer product and add it to the Z matrix */
-      m11 = p1 * q1;
-      ell += lskip1;
-      Z11 += m11;
-      /* load p and q values */
-      p1=ell[0];
-      q1=ex[-3];
-      /* compute outer product and add it to the Z matrix */
-      m11 = p1 * q1;
-      ell += lskip1;
-      ex -= 4;
-      Z11 += m11;
-      /* end of inner loop */
-    }
-    /* compute left-over iterations */
-    j += 4;
-    for (; j > 0; j--) {
-      /* load p and q values */
-      p1=ell[0];
-      q1=ex[0];
-      /* compute outer product and add it to the Z matrix */
-      m11 = p1 * q1;
-      ell += lskip1;
-      ex -= 1;
-      Z11 += m11;
-    }
-    /* finish computing the X(i) block */
-    Z11 = ex[0] - Z11;
-    ex[0] = Z11;
-  }
-}
-
-
-
-void btVectorScale (btScalar *a, const btScalar *d, int n)
+void btSolveL1T(const btScalar *L, btScalar *B, int n, int lskip1)
 {
-  btAssert (a && d && n >= 0);
-  for (int i=0; i<n; i++) {
-    a[i] *= d[i];
-  }
+	/* declare variables - Z matrix, p and q vectors, etc */
+	btScalar Z11, m11, Z21, m21, Z31, m31, Z41, m41, p1, q1, p2, p3, p4, *ex;
+	const btScalar *ell;
+	int lskip2, i, j;
+	//  int lskip3;
+	/* special handling for L and B because we're solving L1 *transpose* */
+	L = L + (n - 1) * (lskip1 + 1);
+	B = B + n - 1;
+	lskip1 = -lskip1;
+	/* compute lskip values */
+	lskip2 = 2 * lskip1;
+	//lskip3 = 3*lskip1;
+	/* compute all 4 x 1 blocks of X */
+	for (i = 0; i <= n - 4; i += 4)
+	{
+		/* compute all 4 x 1 block of X, from rows i..i+4-1 */
+		/* set the Z matrix to 0 */
+		Z11 = 0;
+		Z21 = 0;
+		Z31 = 0;
+		Z41 = 0;
+		ell = L - i;
+		ex = B;
+		/* the inner loop that computes outer products and adds them to Z */
+		for (j = i - 4; j >= 0; j -= 4)
+		{
+			/* load p and q values */
+			p1 = ell[0];
+			q1 = ex[0];
+			p2 = ell[-1];
+			p3 = ell[-2];
+			p4 = ell[-3];
+			/* compute outer product and add it to the Z matrix */
+			m11 = p1 * q1;
+			m21 = p2 * q1;
+			m31 = p3 * q1;
+			m41 = p4 * q1;
+			ell += lskip1;
+			Z11 += m11;
+			Z21 += m21;
+			Z31 += m31;
+			Z41 += m41;
+			/* load p and q values */
+			p1 = ell[0];
+			q1 = ex[-1];
+			p2 = ell[-1];
+			p3 = ell[-2];
+			p4 = ell[-3];
+			/* compute outer product and add it to the Z matrix */
+			m11 = p1 * q1;
+			m21 = p2 * q1;
+			m31 = p3 * q1;
+			m41 = p4 * q1;
+			ell += lskip1;
+			Z11 += m11;
+			Z21 += m21;
+			Z31 += m31;
+			Z41 += m41;
+			/* load p and q values */
+			p1 = ell[0];
+			q1 = ex[-2];
+			p2 = ell[-1];
+			p3 = ell[-2];
+			p4 = ell[-3];
+			/* compute outer product and add it to the Z matrix */
+			m11 = p1 * q1;
+			m21 = p2 * q1;
+			m31 = p3 * q1;
+			m41 = p4 * q1;
+			ell += lskip1;
+			Z11 += m11;
+			Z21 += m21;
+			Z31 += m31;
+			Z41 += m41;
+			/* load p and q values */
+			p1 = ell[0];
+			q1 = ex[-3];
+			p2 = ell[-1];
+			p3 = ell[-2];
+			p4 = ell[-3];
+			/* compute outer product and add it to the Z matrix */
+			m11 = p1 * q1;
+			m21 = p2 * q1;
+			m31 = p3 * q1;
+			m41 = p4 * q1;
+			ell += lskip1;
+			ex -= 4;
+			Z11 += m11;
+			Z21 += m21;
+			Z31 += m31;
+			Z41 += m41;
+			/* end of inner loop */
+		}
+		/* compute left-over iterations */
+		j += 4;
+		for (; j > 0; j--)
+		{
+			/* load p and q values */
+			p1 = ell[0];
+			q1 = ex[0];
+			p2 = ell[-1];
+			p3 = ell[-2];
+			p4 = ell[-3];
+			/* compute outer product and add it to the Z matrix */
+			m11 = p1 * q1;
+			m21 = p2 * q1;
+			m31 = p3 * q1;
+			m41 = p4 * q1;
+			ell += lskip1;
+			ex -= 1;
+			Z11 += m11;
+			Z21 += m21;
+			Z31 += m31;
+			Z41 += m41;
+		}
+		/* finish computing the X(i) block */
+		Z11 = ex[0] - Z11;
+		ex[0] = Z11;
+		p1 = ell[-1];
+		Z21 = ex[-1] - Z21 - p1 * Z11;
+		ex[-1] = Z21;
+		p1 = ell[-2];
+		p2 = ell[-2 + lskip1];
+		Z31 = ex[-2] - Z31 - p1 * Z11 - p2 * Z21;
+		ex[-2] = Z31;
+		p1 = ell[-3];
+		p2 = ell[-3 + lskip1];
+		p3 = ell[-3 + lskip2];
+		Z41 = ex[-3] - Z41 - p1 * Z11 - p2 * Z21 - p3 * Z31;
+		ex[-3] = Z41;
+		/* end of outer loop */
+	}
+	/* compute rows at end that are not a multiple of block size */
+	for (; i < n; i++)
+	{
+		/* compute all 1 x 1 block of X, from rows i..i+1-1 */
+		/* set the Z matrix to 0 */
+		Z11 = 0;
+		ell = L - i;
+		ex = B;
+		/* the inner loop that computes outer products and adds them to Z */
+		for (j = i - 4; j >= 0; j -= 4)
+		{
+			/* load p and q values */
+			p1 = ell[0];
+			q1 = ex[0];
+			/* compute outer product and add it to the Z matrix */
+			m11 = p1 * q1;
+			ell += lskip1;
+			Z11 += m11;
+			/* load p and q values */
+			p1 = ell[0];
+			q1 = ex[-1];
+			/* compute outer product and add it to the Z matrix */
+			m11 = p1 * q1;
+			ell += lskip1;
+			Z11 += m11;
+			/* load p and q values */
+			p1 = ell[0];
+			q1 = ex[-2];
+			/* compute outer product and add it to the Z matrix */
+			m11 = p1 * q1;
+			ell += lskip1;
+			Z11 += m11;
+			/* load p and q values */
+			p1 = ell[0];
+			q1 = ex[-3];
+			/* compute outer product and add it to the Z matrix */
+			m11 = p1 * q1;
+			ell += lskip1;
+			ex -= 4;
+			Z11 += m11;
+			/* end of inner loop */
+		}
+		/* compute left-over iterations */
+		j += 4;
+		for (; j > 0; j--)
+		{
+			/* load p and q values */
+			p1 = ell[0];
+			q1 = ex[0];
+			/* compute outer product and add it to the Z matrix */
+			m11 = p1 * q1;
+			ell += lskip1;
+			ex -= 1;
+			Z11 += m11;
+		}
+		/* finish computing the X(i) block */
+		Z11 = ex[0] - Z11;
+		ex[0] = Z11;
+	}
 }
 
-void btSolveLDLT (const btScalar *L, const btScalar *d, btScalar *b, int n, int nskip)
+void btVectorScale(btScalar *a, const btScalar *d, int n)
 {
-  btAssert (L && d && b && n > 0 && nskip >= n);
-  btSolveL1 (L,b,n,nskip);
-  btVectorScale (b,d,n);
-  btSolveL1T (L,b,n,nskip);
+	btAssert(a && d && n >= 0);
+	for (int i = 0; i < n; i++)
+	{
+		a[i] *= d[i];
+	}
 }
 
-
+void btSolveLDLT(const btScalar *L, const btScalar *d, btScalar *b, int n, int nskip)
+{
+	btAssert(L && d && b && n > 0 && nskip >= n);
+	btSolveL1(L, b, n, nskip);
+	btVectorScale(b, d, n);
+	btSolveL1T(L, b, n, nskip);
+}
 
 //***************************************************************************
 
@@ -1033,125 +1049,130 @@ void btSolveLDLT (const btScalar *L, const btScalar *d, btScalar *b, int n, int
 // rows will be swapped by exchanging row pointers. otherwise the data will
 // be copied.
 
-static void btSwapRowsAndCols (BTATYPE A, int n, int i1, int i2, int nskip, 
-  int do_fast_row_swaps)
+static void btSwapRowsAndCols(BTATYPE A, int n, int i1, int i2, int nskip,
+							  int do_fast_row_swaps)
 {
-  btAssert (A && n > 0 && i1 >= 0 && i2 >= 0 && i1 < n && i2 < n &&
-    nskip >= n && i1 < i2);
+	btAssert(A && n > 0 && i1 >= 0 && i2 >= 0 && i1 < n && i2 < n &&
+			 nskip >= n && i1 < i2);
 
-# ifdef BTROWPTRS
-  btScalar *A_i1 = A[i1];
-  btScalar *A_i2 = A[i2];
-  for (int i=i1+1; i<i2; ++i) {
-    btScalar *A_i_i1 = A[i] + i1;
-    A_i1[i] = *A_i_i1;
-    *A_i_i1 = A_i2[i];
-  }
-  A_i1[i2] = A_i1[i1];
-  A_i1[i1] = A_i2[i1];
-  A_i2[i1] = A_i2[i2];
-  // swap rows, by swapping row pointers
-  if (do_fast_row_swaps) {
-    A[i1] = A_i2;
-    A[i2] = A_i1;
-  }
-  else {
-    // Only swap till i2 column to match A plain storage variant.
-    for (int k = 0; k <= i2; ++k) {
-      btScalar tmp = A_i1[k];
-      A_i1[k] = A_i2[k];
-      A_i2[k] = tmp;
-    }
-  }
-  // swap columns the hard way
-  for (int j=i2+1; j<n; ++j) {
-    btScalar *A_j = A[j];
-    btScalar tmp = A_j[i1];
-    A_j[i1] = A_j[i2];
-    A_j[i2] = tmp;
-  }
-# else
-  btScalar *A_i1 = A+i1*nskip;
-  btScalar *A_i2 = A+i2*nskip;
-  for (int k = 0; k < i1; ++k) {
-    btScalar tmp = A_i1[k];
-    A_i1[k] = A_i2[k];
-    A_i2[k] = tmp;
-  }
-  btScalar *A_i = A_i1 + nskip;
-  for (int i=i1+1; i<i2; A_i+=nskip, ++i) {
-    btScalar tmp = A_i2[i];
-    A_i2[i] = A_i[i1];
-    A_i[i1] = tmp;
-  }
-  {
-    btScalar tmp = A_i1[i1];
-    A_i1[i1] = A_i2[i2];
-    A_i2[i2] = tmp;
-  }
-  btScalar *A_j = A_i2 + nskip;
-  for (int j=i2+1; j<n; A_j+=nskip, ++j) {
-    btScalar tmp = A_j[i1];
-    A_j[i1] = A_j[i2];
-    A_j[i2] = tmp;
-  }
-# endif
+#ifdef BTROWPTRS
+	btScalar *A_i1 = A[i1];
+	btScalar *A_i2 = A[i2];
+	for (int i = i1 + 1; i < i2; ++i)
+	{
+		btScalar *A_i_i1 = A[i] + i1;
+		A_i1[i] = *A_i_i1;
+		*A_i_i1 = A_i2[i];
+	}
+	A_i1[i2] = A_i1[i1];
+	A_i1[i1] = A_i2[i1];
+	A_i2[i1] = A_i2[i2];
+	// swap rows, by swapping row pointers
+	if (do_fast_row_swaps)
+	{
+		A[i1] = A_i2;
+		A[i2] = A_i1;
+	}
+	else
+	{
+		// Only swap till i2 column to match A plain storage variant.
+		for (int k = 0; k <= i2; ++k)
+		{
+			btScalar tmp = A_i1[k];
+			A_i1[k] = A_i2[k];
+			A_i2[k] = tmp;
+		}
+	}
+	// swap columns the hard way
+	for (int j = i2 + 1; j < n; ++j)
+	{
+		btScalar *A_j = A[j];
+		btScalar tmp = A_j[i1];
+		A_j[i1] = A_j[i2];
+		A_j[i2] = tmp;
+	}
+#else
+	btScalar *A_i1 = A + i1 * nskip;
+	btScalar *A_i2 = A + i2 * nskip;
+	for (int k = 0; k < i1; ++k)
+	{
+		btScalar tmp = A_i1[k];
+		A_i1[k] = A_i2[k];
+		A_i2[k] = tmp;
+	}
+	btScalar *A_i = A_i1 + nskip;
+	for (int i = i1 + 1; i < i2; A_i += nskip, ++i)
+	{
+		btScalar tmp = A_i2[i];
+		A_i2[i] = A_i[i1];
+		A_i[i1] = tmp;
+	}
+	{
+		btScalar tmp = A_i1[i1];
+		A_i1[i1] = A_i2[i2];
+		A_i2[i2] = tmp;
+	}
+	btScalar *A_j = A_i2 + nskip;
+	for (int j = i2 + 1; j < n; A_j += nskip, ++j)
+	{
+		btScalar tmp = A_j[i1];
+		A_j[i1] = A_j[i2];
+		A_j[i2] = tmp;
+	}
+#endif
 }
 
-
 // swap two indexes in the n*n LCP problem. i1 must be <= i2.
 
-static void btSwapProblem (BTATYPE A, btScalar *x, btScalar *b, btScalar *w, btScalar *lo,
-                         btScalar *hi, int *p, bool *state, int *findex,
-                         int n, int i1, int i2, int nskip,
-                         int do_fast_row_swaps)
+static void btSwapProblem(BTATYPE A, btScalar *x, btScalar *b, btScalar *w, btScalar *lo,
+						  btScalar *hi, int *p, bool *state, int *findex,
+						  int n, int i1, int i2, int nskip,
+						  int do_fast_row_swaps)
 {
-  btScalar tmpr;
-  int tmpi;
-  bool tmpb;
-  btAssert (n>0 && i1 >=0 && i2 >= 0 && i1 < n && i2 < n && nskip >= n && i1 <= i2);
-  if (i1==i2) return;
-  
-  btSwapRowsAndCols (A,n,i1,i2,nskip,do_fast_row_swaps);
-  
-  tmpr = x[i1];
-  x[i1] = x[i2];
-  x[i2] = tmpr;
-  
-  tmpr = b[i1];
-  b[i1] = b[i2];
-  b[i2] = tmpr;
-  
-  tmpr = w[i1];
-  w[i1] = w[i2];
-  w[i2] = tmpr;
-  
-  tmpr = lo[i1];
-  lo[i1] = lo[i2];
-  lo[i2] = tmpr;
+	btScalar tmpr;
+	int tmpi;
+	bool tmpb;
+	btAssert(n > 0 && i1 >= 0 && i2 >= 0 && i1 < n && i2 < n && nskip >= n && i1 <= i2);
+	if (i1 == i2) return;
 
-  tmpr = hi[i1];
-  hi[i1] = hi[i2];
-  hi[i2] = tmpr;
+	btSwapRowsAndCols(A, n, i1, i2, nskip, do_fast_row_swaps);
 
-  tmpi = p[i1];
-  p[i1] = p[i2];
-  p[i2] = tmpi;
+	tmpr = x[i1];
+	x[i1] = x[i2];
+	x[i2] = tmpr;
 
-  tmpb = state[i1];
-  state[i1] = state[i2];
-  state[i2] = tmpb;
+	tmpr = b[i1];
+	b[i1] = b[i2];
+	b[i2] = tmpr;
 
-  if (findex) {
-    tmpi = findex[i1];
-    findex[i1] = findex[i2];
-    findex[i2] = tmpi;
-  }
+	tmpr = w[i1];
+	w[i1] = w[i2];
+	w[i2] = tmpr;
+
+	tmpr = lo[i1];
+	lo[i1] = lo[i2];
+	lo[i2] = tmpr;
+
+	tmpr = hi[i1];
+	hi[i1] = hi[i2];
+	hi[i2] = tmpr;
+
+	tmpi = p[i1];
+	p[i1] = p[i2];
+	p[i2] = tmpi;
+
+	tmpb = state[i1];
+	state[i1] = state[i2];
+	state[i2] = tmpb;
+
+	if (findex)
+	{
+		tmpi = findex[i1];
+		findex[i1] = findex[i2];
+		findex[i2] = tmpi;
+	}
 }
 
-
-
-
 //***************************************************************************
 // btLCP manipulator object. this represents an n*n LCP problem.
 //
@@ -1186,79 +1207,88 @@ static void btSwapProblem (BTATYPE A, btScalar *x, btScalar *b, btScalar *w, btS
 
 #ifdef btLCP_FAST
 
-struct btLCP 
+struct btLCP
 {
 	const int m_n;
 	const int m_nskip;
 	int m_nub;
-	int m_nC, m_nN;				// size of each index set
-	BTATYPE const m_A;				// A rows
-	btScalar *const m_x, * const m_b, *const m_w, *const m_lo,* const m_hi;	// permuted LCP problem data
-	btScalar *const m_L, *const m_d;				// L*D*L' factorization of set C
+	int m_nC, m_nN;                                                         // size of each index set
+	BTATYPE const m_A;                                                      // A rows
+	btScalar *const m_x, *const m_b, *const m_w, *const m_lo, *const m_hi;  // permuted LCP problem data
+	btScalar *const m_L, *const m_d;                                        // L*D*L' factorization of set C
 	btScalar *const m_Dell, *const m_ell, *const m_tmp;
 	bool *const m_state;
 	int *const m_findex, *const m_p, *const m_C;
 
-	btLCP (int _n, int _nskip, int _nub, btScalar *_Adata, btScalar *_x, btScalar *_b, btScalar *_w,
-		btScalar *_lo, btScalar *_hi, btScalar *l, btScalar *_d,
-		btScalar *_Dell, btScalar *_ell, btScalar *_tmp,
-		bool *_state, int *_findex, int *p, int *c, btScalar **Arows);
+	btLCP(int _n, int _nskip, int _nub, btScalar *_Adata, btScalar *_x, btScalar *_b, btScalar *_w,
+		  btScalar *_lo, btScalar *_hi, btScalar *l, btScalar *_d,
+		  btScalar *_Dell, btScalar *_ell, btScalar *_tmp,
+		  bool *_state, int *_findex, int *p, int *c, btScalar **Arows);
 	int getNub() const { return m_nub; }
-	void transfer_i_to_C (int i);
-	void transfer_i_to_N (int i) { m_nN++; }			// because we can assume C and N span 1:i-1
-	void transfer_i_from_N_to_C (int i);
-	void transfer_i_from_C_to_N (int i, btAlignedObjectArray<btScalar>& scratch);
+	void transfer_i_to_C(int i);
+	void transfer_i_to_N(int i) { m_nN++; }  // because we can assume C and N span 1:i-1
+	void transfer_i_from_N_to_C(int i);
+	void transfer_i_from_C_to_N(int i, btAlignedObjectArray<btScalar> &scratch);
 	int numC() const { return m_nC; }
 	int numN() const { return m_nN; }
-	int indexC (int i) const { return i; }
-	int indexN (int i) const { return i+m_nC; }
-	btScalar Aii (int i) const  { return BTAROW(i)[i]; }
-	btScalar AiC_times_qC (int i, btScalar *q) const { return btLargeDot (BTAROW(i), q, m_nC); }
-	btScalar AiN_times_qN (int i, btScalar *q) const { return btLargeDot (BTAROW(i)+m_nC, q+m_nC, m_nN); }
-	void pN_equals_ANC_times_qC (btScalar *p, btScalar *q);
-	void pN_plusequals_ANi (btScalar *p, int i, int sign=1);
-	void pC_plusequals_s_times_qC (btScalar *p, btScalar s, btScalar *q);
-	void pN_plusequals_s_times_qN (btScalar *p, btScalar s, btScalar *q);
-	void solve1 (btScalar *a, int i, int dir=1, int only_transfer=0);
+	int indexC(int i) const { return i; }
+	int indexN(int i) const { return i + m_nC; }
+	btScalar Aii(int i) const { return BTAROW(i)[i]; }
+	btScalar AiC_times_qC(int i, btScalar *q) const { return btLargeDot(BTAROW(i), q, m_nC); }
+	btScalar AiN_times_qN(int i, btScalar *q) const { return btLargeDot(BTAROW(i) + m_nC, q + m_nC, m_nN); }
+	void pN_equals_ANC_times_qC(btScalar *p, btScalar *q);
+	void pN_plusequals_ANi(btScalar *p, int i, int sign = 1);
+	void pC_plusequals_s_times_qC(btScalar *p, btScalar s, btScalar *q);
+	void pN_plusequals_s_times_qN(btScalar *p, btScalar s, btScalar *q);
+	void solve1(btScalar *a, int i, int dir = 1, int only_transfer = 0);
 	void unpermute();
 };
 
-
-btLCP::btLCP (int _n, int _nskip, int _nub, btScalar *_Adata, btScalar *_x, btScalar *_b, btScalar *_w,
-            btScalar *_lo, btScalar *_hi, btScalar *l, btScalar *_d,
-            btScalar *_Dell, btScalar *_ell, btScalar *_tmp,
-            bool *_state, int *_findex, int *p, int *c, btScalar **Arows):
-  m_n(_n), m_nskip(_nskip), m_nub(_nub), m_nC(0), m_nN(0),
-# ifdef BTROWPTRS
-  m_A(Arows),
+btLCP::btLCP(int _n, int _nskip, int _nub, btScalar *_Adata, btScalar *_x, btScalar *_b, btScalar *_w,
+			 btScalar *_lo, btScalar *_hi, btScalar *l, btScalar *_d,
+			 btScalar *_Dell, btScalar *_ell, btScalar *_tmp,
+			 bool *_state, int *_findex, int *p, int *c, btScalar **Arows) : m_n(_n), m_nskip(_nskip), m_nub(_nub), m_nC(0), m_nN(0),
+#ifdef BTROWPTRS
+																			 m_A(Arows),
 #else
-  m_A(_Adata),
+																			 m_A(_Adata),
 #endif
-  m_x(_x), m_b(_b), m_w(_w), m_lo(_lo), m_hi(_hi),
-  m_L(l), m_d(_d), m_Dell(_Dell), m_ell(_ell), m_tmp(_tmp),
-  m_state(_state), m_findex(_findex), m_p(p), m_C(c)
+																			 m_x(_x),
+																			 m_b(_b),
+																			 m_w(_w),
+																			 m_lo(_lo),
+																			 m_hi(_hi),
+																			 m_L(l),
+																			 m_d(_d),
+																			 m_Dell(_Dell),
+																			 m_ell(_ell),
+																			 m_tmp(_tmp),
+																			 m_state(_state),
+																			 m_findex(_findex),
+																			 m_p(p),
+																			 m_C(c)
 {
-  {
-    btSetZero (m_x,m_n);
-  }
+	{
+		btSetZero(m_x, m_n);
+	}
 
-  {
-# ifdef BTROWPTRS
-    // make matrix row pointers
-    btScalar *aptr = _Adata;
-    BTATYPE A = m_A;
-    const int n = m_n, nskip = m_nskip;
-    for (int k=0; k<n; aptr+=nskip, ++k) A[k] = aptr;
-# endif
-  }
+	{
+#ifdef BTROWPTRS
+		// make matrix row pointers
+		btScalar *aptr = _Adata;
+		BTATYPE A = m_A;
+		const int n = m_n, nskip = m_nskip;
+		for (int k = 0; k < n; aptr += nskip, ++k) A[k] = aptr;
+#endif
+	}
 
-  {
-    int *p = m_p;
-    const int n = m_n;
-    for (int k=0; k<n; ++k) p[k]=k;		// initially unpermuted
-  }
+	{
+		int *p = m_p;
+		const int n = m_n;
+		for (int k = 0; k < n; ++k) p[k] = k;  // initially unpermuted
+	}
 
-  /*
+	/*
   // for testing, we can do some random swaps in the area i > nub
   {
     const int n = m_n;
@@ -1277,63 +1307,69 @@ btLCP::btLCP (int _n, int _nskip, int _nub, btScalar *_Adata, btScalar *_x, btSc
   }
   */
 
-  // permute the problem so that *all* the unbounded variables are at the
-  // start, i.e. look for unbounded variables not included in `nub'. we can
-  // potentially push up `nub' this way and get a bigger initial factorization.
-  // note that when we swap rows/cols here we must not just swap row pointers,
-  // as the initial factorization relies on the data being all in one chunk.
-  // variables that have findex >= 0 are *not* considered to be unbounded even
-  // if lo=-inf and hi=inf - this is because these limits may change during the
-  // solution process.
+	// permute the problem so that *all* the unbounded variables are at the
+	// start, i.e. look for unbounded variables not included in `nub'. we can
+	// potentially push up `nub' this way and get a bigger initial factorization.
+	// note that when we swap rows/cols here we must not just swap row pointers,
+	// as the initial factorization relies on the data being all in one chunk.
+	// variables that have findex >= 0 are *not* considered to be unbounded even
+	// if lo=-inf and hi=inf - this is because these limits may change during the
+	// solution process.
 
-  {
-    int *findex = m_findex;
-    btScalar *lo = m_lo, *hi = m_hi;
-    const int n = m_n;
-    for (int k = m_nub; k<n; ++k) {
-      if (findex && findex[k] >= 0) continue;
-      if (lo[k]==-BT_INFINITY && hi[k]==BT_INFINITY) {
-        btSwapProblem (m_A,m_x,m_b,m_w,lo,hi,m_p,m_state,findex,n,m_nub,k,m_nskip,0);
-        m_nub++;
-      }
-    }
-  }
+	{
+		int *findex = m_findex;
+		btScalar *lo = m_lo, *hi = m_hi;
+		const int n = m_n;
+		for (int k = m_nub; k < n; ++k)
+		{
+			if (findex && findex[k] >= 0) continue;
+			if (lo[k] == -BT_INFINITY && hi[k] == BT_INFINITY)
+			{
+				btSwapProblem(m_A, m_x, m_b, m_w, lo, hi, m_p, m_state, findex, n, m_nub, k, m_nskip, 0);
+				m_nub++;
+			}
+		}
+	}
 
-  // if there are unbounded variables at the start, factorize A up to that
-  // point and solve for x. this puts all indexes 0..nub-1 into C.
-  if (m_nub > 0) {
-    const int nub = m_nub;
-    {
-      btScalar *Lrow = m_L;
-      const int nskip = m_nskip;
-      for (int j=0; j<nub; Lrow+=nskip, ++j) memcpy(Lrow,BTAROW(j),(j+1)*sizeof(btScalar));
-    }
-    btFactorLDLT (m_L,m_d,nub,m_nskip);
-    memcpy (m_x,m_b,nub*sizeof(btScalar));
-    btSolveLDLT (m_L,m_d,m_x,nub,m_nskip);
-    btSetZero (m_w,nub);
-    {
-      int *C = m_C;
-      for (int k=0; k<nub; ++k) C[k] = k;
-    }
-    m_nC = nub;
-  }
+	// if there are unbounded variables at the start, factorize A up to that
+	// point and solve for x. this puts all indexes 0..nub-1 into C.
+	if (m_nub > 0)
+	{
+		const int nub = m_nub;
+		{
+			btScalar *Lrow = m_L;
+			const int nskip = m_nskip;
+			for (int j = 0; j < nub; Lrow += nskip, ++j) memcpy(Lrow, BTAROW(j), (j + 1) * sizeof(btScalar));
+		}
+		btFactorLDLT(m_L, m_d, nub, m_nskip);
+		memcpy(m_x, m_b, nub * sizeof(btScalar));
+		btSolveLDLT(m_L, m_d, m_x, nub, m_nskip);
+		btSetZero(m_w, nub);
+		{
+			int *C = m_C;
+			for (int k = 0; k < nub; ++k) C[k] = k;
+		}
+		m_nC = nub;
+	}
 
-  // permute the indexes > nub such that all findex variables are at the end
-  if (m_findex) {
-    const int nub = m_nub;
-    int *findex = m_findex;
-    int num_at_end = 0;
-    for (int k=m_n-1; k >= nub; k--) {
-      if (findex[k] >= 0) {
-        btSwapProblem (m_A,m_x,m_b,m_w,m_lo,m_hi,m_p,m_state,findex,m_n,k,m_n-1-num_at_end,m_nskip,1);
-        num_at_end++;
-      }
-    }
-  }
+	// permute the indexes > nub such that all findex variables are at the end
+	if (m_findex)
+	{
+		const int nub = m_nub;
+		int *findex = m_findex;
+		int num_at_end = 0;
+		for (int k = m_n - 1; k >= nub; k--)
+		{
+			if (findex[k] >= 0)
+			{
+				btSwapProblem(m_A, m_x, m_b, m_w, m_lo, m_hi, m_p, m_state, findex, m_n, k, m_n - 1 - num_at_end, m_nskip, 1);
+				num_at_end++;
+			}
+		}
+	}
 
-  // print info about indexes
-  /*
+	// print info about indexes
+	/*
   {
     const int n = m_n;
     const int nub = m_nub;
@@ -1347,734 +1383,776 @@ btLCP::btLCP (int _n, int _nskip, int _nub, btScalar *_Adata, btScalar *_x, btSc
   */
 }
 
-
-void btLCP::transfer_i_to_C (int i)
+void btLCP::transfer_i_to_C(int i)
 {
-  {
-    if (m_nC > 0) {
-      // ell,Dell were computed by solve1(). note, ell = D \ L1solve (L,A(i,C))
-      {
-        const int nC = m_nC;
-        btScalar *const Ltgt = m_L + nC*m_nskip, *ell = m_ell;
-        for (int j=0; j<nC; ++j) Ltgt[j] = ell[j];
-      }
-      const int nC = m_nC;
-      m_d[nC] = btRecip (BTAROW(i)[i] - btLargeDot(m_ell,m_Dell,nC));
-    }
-    else {
-      m_d[0] = btRecip (BTAROW(i)[i]);
-    }
+	{
+		if (m_nC > 0)
+		{
+			// ell,Dell were computed by solve1(). note, ell = D \ L1solve (L,A(i,C))
+			{
+				const int nC = m_nC;
+				btScalar *const Ltgt = m_L + nC * m_nskip, *ell = m_ell;
+				for (int j = 0; j < nC; ++j) Ltgt[j] = ell[j];
+			}
+			const int nC = m_nC;
+			m_d[nC] = btRecip(BTAROW(i)[i] - btLargeDot(m_ell, m_Dell, nC));
+		}
+		else
+		{
+			m_d[0] = btRecip(BTAROW(i)[i]);
+		}
 
-    btSwapProblem (m_A,m_x,m_b,m_w,m_lo,m_hi,m_p,m_state,m_findex,m_n,m_nC,i,m_nskip,1);
-
-    const int nC = m_nC;
-    m_C[nC] = nC;
-    m_nC = nC + 1; // nC value is outdated after this line
-  }
+		btSwapProblem(m_A, m_x, m_b, m_w, m_lo, m_hi, m_p, m_state, m_findex, m_n, m_nC, i, m_nskip, 1);
 
+		const int nC = m_nC;
+		m_C[nC] = nC;
+		m_nC = nC + 1;  // nC value is outdated after this line
+	}
 }
 
-
-void btLCP::transfer_i_from_N_to_C (int i)
+void btLCP::transfer_i_from_N_to_C(int i)
 {
-  {
-    if (m_nC > 0) {
-      {
-        btScalar *const aptr = BTAROW(i);
-        btScalar *Dell = m_Dell;
-        const int *C = m_C;
-#   ifdef BTNUB_OPTIMIZATIONS
-        // if nub>0, initial part of aptr unpermuted
-        const int nub = m_nub;
-        int j=0;
-        for ( ; j<nub; ++j) Dell[j] = aptr[j];
-        const int nC = m_nC;
-        for ( ; j<nC; ++j) Dell[j] = aptr[C[j]];
-#   else
-        const int nC = m_nC;
-        for (int j=0; j<nC; ++j) Dell[j] = aptr[C[j]];
-#   endif
-      }
-      btSolveL1 (m_L,m_Dell,m_nC,m_nskip);
-      {
-        const int nC = m_nC;
-        btScalar *const Ltgt = m_L + nC*m_nskip;
-        btScalar *ell = m_ell, *Dell = m_Dell, *d = m_d;
-        for (int j=0; j<nC; ++j) Ltgt[j] = ell[j] = Dell[j] * d[j];
-      }
-      const int nC = m_nC;
-      m_d[nC] = btRecip (BTAROW(i)[i] - btLargeDot(m_ell,m_Dell,nC));
-    }
-    else {
-      m_d[0] = btRecip (BTAROW(i)[i]);
-    }
+	{
+		if (m_nC > 0)
+		{
+			{
+				btScalar *const aptr = BTAROW(i);
+				btScalar *Dell = m_Dell;
+				const int *C = m_C;
+#ifdef BTNUB_OPTIMIZATIONS
+				// if nub>0, initial part of aptr unpermuted
+				const int nub = m_nub;
+				int j = 0;
+				for (; j < nub; ++j) Dell[j] = aptr[j];
+				const int nC = m_nC;
+				for (; j < nC; ++j) Dell[j] = aptr[C[j]];
+#else
+				const int nC = m_nC;
+				for (int j = 0; j < nC; ++j) Dell[j] = aptr[C[j]];
+#endif
+			}
+			btSolveL1(m_L, m_Dell, m_nC, m_nskip);
+			{
+				const int nC = m_nC;
+				btScalar *const Ltgt = m_L + nC * m_nskip;
+				btScalar *ell = m_ell, *Dell = m_Dell, *d = m_d;
+				for (int j = 0; j < nC; ++j) Ltgt[j] = ell[j] = Dell[j] * d[j];
+			}
+			const int nC = m_nC;
+			m_d[nC] = btRecip(BTAROW(i)[i] - btLargeDot(m_ell, m_Dell, nC));
+		}
+		else
+		{
+			m_d[0] = btRecip(BTAROW(i)[i]);
+		}
 
-    btSwapProblem (m_A,m_x,m_b,m_w,m_lo,m_hi,m_p,m_state,m_findex,m_n,m_nC,i,m_nskip,1);
+		btSwapProblem(m_A, m_x, m_b, m_w, m_lo, m_hi, m_p, m_state, m_findex, m_n, m_nC, i, m_nskip, 1);
 
-    const int nC = m_nC;
-    m_C[nC] = nC;
-    m_nN--;
-    m_nC = nC + 1; // nC value is outdated after this line
-  }
-
-  // @@@ TO DO LATER
-  // if we just finish here then we'll go back and re-solve for
-  // delta_x. but actually we can be more efficient and incrementally
-  // update delta_x here. but if we do this, we wont have ell and Dell
-  // to use in updating the factorization later.
+		const int nC = m_nC;
+		m_C[nC] = nC;
+		m_nN--;
+		m_nC = nC + 1;  // nC value is outdated after this line
+	}
 
+	// @@@ TO DO LATER
+	// if we just finish here then we'll go back and re-solve for
+	// delta_x. but actually we can be more efficient and incrementally
+	// update delta_x here. but if we do this, we wont have ell and Dell
+	// to use in updating the factorization later.
 }
 
-void btRemoveRowCol (btScalar *A, int n, int nskip, int r)
+void btRemoveRowCol(btScalar *A, int n, int nskip, int r)
 {
-  btAssert(A && n > 0 && nskip >= n && r >= 0 && r < n);
-  if (r >= n-1) return;
-  if (r > 0) {
-    {
-      const size_t move_size = (n-r-1)*sizeof(btScalar);
-      btScalar *Adst = A + r;
-      for (int i=0; i<r; Adst+=nskip,++i) {
-        btScalar *Asrc = Adst + 1;
-        memmove (Adst,Asrc,move_size);
-      }
-    }
-    {
-      const size_t cpy_size = r*sizeof(btScalar);
-      btScalar *Adst = A + r * nskip;
-      for (int i=r; i<(n-1); ++i) {
-        btScalar *Asrc = Adst + nskip;
-        memcpy (Adst,Asrc,cpy_size);
-        Adst = Asrc;
-      }
-    }
-  }
-  {
-    const size_t cpy_size = (n-r-1)*sizeof(btScalar);
-    btScalar *Adst = A + r * (nskip + 1);
-    for (int i=r; i<(n-1); ++i) {
-      btScalar *Asrc = Adst + (nskip + 1);
-      memcpy (Adst,Asrc,cpy_size);
-      Adst = Asrc - 1;
-    }
-  }
+	btAssert(A && n > 0 && nskip >= n && r >= 0 && r < n);
+	if (r >= n - 1) return;
+	if (r > 0)
+	{
+		{
+			const size_t move_size = (n - r - 1) * sizeof(btScalar);
+			btScalar *Adst = A + r;
+			for (int i = 0; i < r; Adst += nskip, ++i)
+			{
+				btScalar *Asrc = Adst + 1;
+				memmove(Adst, Asrc, move_size);
+			}
+		}
+		{
+			const size_t cpy_size = r * sizeof(btScalar);
+			btScalar *Adst = A + r * nskip;
+			for (int i = r; i < (n - 1); ++i)
+			{
+				btScalar *Asrc = Adst + nskip;
+				memcpy(Adst, Asrc, cpy_size);
+				Adst = Asrc;
+			}
+		}
+	}
+	{
+		const size_t cpy_size = (n - r - 1) * sizeof(btScalar);
+		btScalar *Adst = A + r * (nskip + 1);
+		for (int i = r; i < (n - 1); ++i)
+		{
+			btScalar *Asrc = Adst + (nskip + 1);
+			memcpy(Adst, Asrc, cpy_size);
+			Adst = Asrc - 1;
+		}
+	}
 }
 
-
-
-
-void btLDLTAddTL (btScalar *L, btScalar *d, const btScalar *a, int n, int nskip, btAlignedObjectArray<btScalar>& scratch)
+void btLDLTAddTL(btScalar *L, btScalar *d, const btScalar *a, int n, int nskip, btAlignedObjectArray<btScalar> &scratch)
 {
-  btAssert (L && d && a && n > 0 && nskip >= n);
+	btAssert(L && d && a && n > 0 && nskip >= n);
 
-  if (n < 2) return;
-  scratch.resize(2*nskip);
-  btScalar *W1 = &scratch[0];
-  
-  btScalar *W2 = W1 + nskip;
+	if (n < 2) return;
+	scratch.resize(2 * nskip);
+	btScalar *W1 = &scratch[0];
 
-  W1[0] = btScalar(0.0);
-  W2[0] = btScalar(0.0);
-  for (int j=1; j<n; ++j) {
-    W1[j] = W2[j] = (btScalar) (a[j] * SIMDSQRT12);
-  }
-  btScalar W11 = (btScalar) ((btScalar(0.5)*a[0]+1)*SIMDSQRT12);
-  btScalar W21 = (btScalar) ((btScalar(0.5)*a[0]-1)*SIMDSQRT12);
+	btScalar *W2 = W1 + nskip;
 
-  btScalar alpha1 = btScalar(1.0);
-  btScalar alpha2 = btScalar(1.0);
+	W1[0] = btScalar(0.0);
+	W2[0] = btScalar(0.0);
+	for (int j = 1; j < n; ++j)
+	{
+		W1[j] = W2[j] = (btScalar)(a[j] * SIMDSQRT12);
+	}
+	btScalar W11 = (btScalar)((btScalar(0.5) * a[0] + 1) * SIMDSQRT12);
+	btScalar W21 = (btScalar)((btScalar(0.5) * a[0] - 1) * SIMDSQRT12);
 
-  {
-    btScalar dee = d[0];
-    btScalar alphanew = alpha1 + (W11*W11)*dee;
-    btAssert(alphanew != btScalar(0.0));
-    dee /= alphanew;
-    btScalar gamma1 = W11 * dee;
-    dee *= alpha1;
-    alpha1 = alphanew;
-    alphanew = alpha2 - (W21*W21)*dee;
-    dee /= alphanew;
-    //btScalar gamma2 = W21 * dee;
-    alpha2 = alphanew;
-    btScalar k1 = btScalar(1.0) - W21*gamma1;
-    btScalar k2 = W21*gamma1*W11 - W21;
-    btScalar *ll = L + nskip;
-    for (int p=1; p<n; ll+=nskip, ++p) {
-      btScalar Wp = W1[p];
-      btScalar ell = *ll;
-      W1[p] =    Wp - W11*ell;
-      W2[p] = k1*Wp +  k2*ell;
-    }
-  }
+	btScalar alpha1 = btScalar(1.0);
+	btScalar alpha2 = btScalar(1.0);
 
-  btScalar *ll = L + (nskip + 1);
-  for (int j=1; j<n; ll+=nskip+1, ++j) {
-    btScalar k1 = W1[j];
-    btScalar k2 = W2[j];
+	{
+		btScalar dee = d[0];
+		btScalar alphanew = alpha1 + (W11 * W11) * dee;
+		btAssert(alphanew != btScalar(0.0));
+		dee /= alphanew;
+		btScalar gamma1 = W11 * dee;
+		dee *= alpha1;
+		alpha1 = alphanew;
+		alphanew = alpha2 - (W21 * W21) * dee;
+		dee /= alphanew;
+		//btScalar gamma2 = W21 * dee;
+		alpha2 = alphanew;
+		btScalar k1 = btScalar(1.0) - W21 * gamma1;
+		btScalar k2 = W21 * gamma1 * W11 - W21;
+		btScalar *ll = L + nskip;
+		for (int p = 1; p < n; ll += nskip, ++p)
+		{
+			btScalar Wp = W1[p];
+			btScalar ell = *ll;
+			W1[p] = Wp - W11 * ell;
+			W2[p] = k1 * Wp + k2 * ell;
+		}
+	}
 
-    btScalar dee = d[j];
-    btScalar alphanew = alpha1 + (k1*k1)*dee;
-    btAssert(alphanew != btScalar(0.0));
-    dee /= alphanew;
-    btScalar gamma1 = k1 * dee;
-    dee *= alpha1;
-    alpha1 = alphanew;
-    alphanew = alpha2 - (k2*k2)*dee;
-    dee /= alphanew;
-    btScalar gamma2 = k2 * dee;
-    dee *= alpha2;
-    d[j] = dee;
-    alpha2 = alphanew;
+	btScalar *ll = L + (nskip + 1);
+	for (int j = 1; j < n; ll += nskip + 1, ++j)
+	{
+		btScalar k1 = W1[j];
+		btScalar k2 = W2[j];
 
-    btScalar *l = ll + nskip;
-    for (int p=j+1; p<n; l+=nskip, ++p) {
-      btScalar ell = *l;
-      btScalar Wp = W1[p] - k1 * ell;
-      ell += gamma1 * Wp;
-      W1[p] = Wp;
-      Wp = W2[p] - k2 * ell;
-      ell -= gamma2 * Wp;
-      W2[p] = Wp;
-      *l = ell;
-    }
-  }
+		btScalar dee = d[j];
+		btScalar alphanew = alpha1 + (k1 * k1) * dee;
+		btAssert(alphanew != btScalar(0.0));
+		dee /= alphanew;
+		btScalar gamma1 = k1 * dee;
+		dee *= alpha1;
+		alpha1 = alphanew;
+		alphanew = alpha2 - (k2 * k2) * dee;
+		dee /= alphanew;
+		btScalar gamma2 = k2 * dee;
+		dee *= alpha2;
+		d[j] = dee;
+		alpha2 = alphanew;
+
+		btScalar *l = ll + nskip;
+		for (int p = j + 1; p < n; l += nskip, ++p)
+		{
+			btScalar ell = *l;
+			btScalar Wp = W1[p] - k1 * ell;
+			ell += gamma1 * Wp;
+			W1[p] = Wp;
+			Wp = W2[p] - k2 * ell;
+			ell -= gamma2 * Wp;
+			W2[p] = Wp;
+			*l = ell;
+		}
+	}
 }
 
-
-#define _BTGETA(i,j) (A[i][j])
+#define _BTGETA(i, j) (A[i][j])
 //#define _GETA(i,j) (A[(i)*nskip+(j)])
-#define BTGETA(i,j) ((i > j) ? _BTGETA(i,j) : _BTGETA(j,i))
+#define BTGETA(i, j) ((i > j) ? _BTGETA(i, j) : _BTGETA(j, i))
 
 inline size_t btEstimateLDLTAddTLTmpbufSize(int nskip)
 {
-  return nskip * 2 * sizeof(btScalar);
+	return nskip * 2 * sizeof(btScalar);
 }
 
-
-void btLDLTRemove (btScalar **A, const int *p, btScalar *L, btScalar *d,
-    int n1, int n2, int r, int nskip, btAlignedObjectArray<btScalar>& scratch)
+void btLDLTRemove(btScalar **A, const int *p, btScalar *L, btScalar *d,
+				  int n1, int n2, int r, int nskip, btAlignedObjectArray<btScalar> &scratch)
 {
-  btAssert(A && p && L && d && n1 > 0 && n2 > 0 && r >= 0 && r < n2 &&
-	   n1 >= n2 && nskip >= n1);
-  #ifdef BT_DEBUG
-	for (int i=0; i<n2; ++i) 
+	btAssert(A && p && L && d && n1 > 0 && n2 > 0 && r >= 0 && r < n2 &&
+			 n1 >= n2 && nskip >= n1);
+#ifdef BT_DEBUG
+	for (int i = 0; i < n2; ++i)
 		btAssert(p[i] >= 0 && p[i] < n1);
-  #endif
+#endif
 
-  if (r==n2-1) {
-    return;		// deleting last row/col is easy
-  }
-  else {
-    size_t LDLTAddTL_size = btEstimateLDLTAddTLTmpbufSize(nskip);
-    btAssert(LDLTAddTL_size % sizeof(btScalar) == 0);
-	scratch.resize(nskip * 2+n2);
-    btScalar *tmp = &scratch[0];
-    if (r==0) {
-      btScalar *a = (btScalar *)((char *)tmp + LDLTAddTL_size);
-      const int p_0 = p[0];
-      for (int i=0; i<n2; ++i) {
-        a[i] = -BTGETA(p[i],p_0);
-      }
-      a[0] += btScalar(1.0);
-      btLDLTAddTL (L,d,a,n2,nskip,scratch);
-    }
-    else {
-      btScalar *t = (btScalar *)((char *)tmp + LDLTAddTL_size);
-      {
-        btScalar *Lcurr = L + r*nskip;
-        for (int i=0; i<r; ++Lcurr, ++i) {
-          btAssert(d[i] != btScalar(0.0));
-          t[i] = *Lcurr / d[i];
-        }
-      }
-      btScalar *a = t + r;
-      {
-        btScalar *Lcurr = L + r*nskip;
-        const int *pp_r = p + r, p_r = *pp_r;
-        const int n2_minus_r = n2-r;
-        for (int i=0; i<n2_minus_r; Lcurr+=nskip,++i) {
-          a[i] = btLargeDot(Lcurr,t,r) - BTGETA(pp_r[i],p_r);
-        }
-      }
-      a[0] += btScalar(1.0);
-      btLDLTAddTL (L + r*nskip+r, d+r, a, n2-r, nskip, scratch);
-    }
-  }
+	if (r == n2 - 1)
+	{
+		return;  // deleting last row/col is easy
+	}
+	else
+	{
+		size_t LDLTAddTL_size = btEstimateLDLTAddTLTmpbufSize(nskip);
+		btAssert(LDLTAddTL_size % sizeof(btScalar) == 0);
+		scratch.resize(nskip * 2 + n2);
+		btScalar *tmp = &scratch[0];
+		if (r == 0)
+		{
+			btScalar *a = (btScalar *)((char *)tmp + LDLTAddTL_size);
+			const int p_0 = p[0];
+			for (int i = 0; i < n2; ++i)
+			{
+				a[i] = -BTGETA(p[i], p_0);
+			}
+			a[0] += btScalar(1.0);
+			btLDLTAddTL(L, d, a, n2, nskip, scratch);
+		}
+		else
+		{
+			btScalar *t = (btScalar *)((char *)tmp + LDLTAddTL_size);
+			{
+				btScalar *Lcurr = L + r * nskip;
+				for (int i = 0; i < r; ++Lcurr, ++i)
+				{
+					btAssert(d[i] != btScalar(0.0));
+					t[i] = *Lcurr / d[i];
+				}
+			}
+			btScalar *a = t + r;
+			{
+				btScalar *Lcurr = L + r * nskip;
+				const int *pp_r = p + r, p_r = *pp_r;
+				const int n2_minus_r = n2 - r;
+				for (int i = 0; i < n2_minus_r; Lcurr += nskip, ++i)
+				{
+					a[i] = btLargeDot(Lcurr, t, r) - BTGETA(pp_r[i], p_r);
+				}
+			}
+			a[0] += btScalar(1.0);
+			btLDLTAddTL(L + r * nskip + r, d + r, a, n2 - r, nskip, scratch);
+		}
+	}
 
-  // snip out row/column r from L and d
-  btRemoveRowCol (L,n2,nskip,r);
-  if (r < (n2-1)) memmove (d+r,d+r+1,(n2-r-1)*sizeof(btScalar));
+	// snip out row/column r from L and d
+	btRemoveRowCol(L, n2, nskip, r);
+	if (r < (n2 - 1)) memmove(d + r, d + r + 1, (n2 - r - 1) * sizeof(btScalar));
 }
 
-
-void btLCP::transfer_i_from_C_to_N (int i, btAlignedObjectArray<btScalar>& scratch)
+void btLCP::transfer_i_from_C_to_N(int i, btAlignedObjectArray<btScalar> &scratch)
 {
-  {
-    int *C = m_C;
-    // remove a row/column from the factorization, and adjust the
-    // indexes (black magic!)
-    int last_idx = -1;
-    const int nC = m_nC;
-    int j = 0;
-    for ( ; j<nC; ++j) {
-      if (C[j]==nC-1) {
-        last_idx = j;
-      }
-      if (C[j]==i) {
-        btLDLTRemove (m_A,C,m_L,m_d,m_n,nC,j,m_nskip,scratch);
-        int k;
-        if (last_idx == -1) {
-          for (k=j+1 ; k<nC; ++k) {
-            if (C[k]==nC-1) {
-              break;
-            }
-          }
-          btAssert (k < nC);
-        }
-        else {
-          k = last_idx;
-        }
-        C[k] = C[j];
-        if (j < (nC-1)) memmove (C+j,C+j+1,(nC-j-1)*sizeof(int));
-        break;
-      }
-    }
-    btAssert (j < nC);
+	{
+		int *C = m_C;
+		// remove a row/column from the factorization, and adjust the
+		// indexes (black magic!)
+		int last_idx = -1;
+		const int nC = m_nC;
+		int j = 0;
+		for (; j < nC; ++j)
+		{
+			if (C[j] == nC - 1)
+			{
+				last_idx = j;
+			}
+			if (C[j] == i)
+			{
+				btLDLTRemove(m_A, C, m_L, m_d, m_n, nC, j, m_nskip, scratch);
+				int k;
+				if (last_idx == -1)
+				{
+					for (k = j + 1; k < nC; ++k)
+					{
+						if (C[k] == nC - 1)
+						{
+							break;
+						}
+					}
+					btAssert(k < nC);
+				}
+				else
+				{
+					k = last_idx;
+				}
+				C[k] = C[j];
+				if (j < (nC - 1)) memmove(C + j, C + j + 1, (nC - j - 1) * sizeof(int));
+				break;
+			}
+		}
+		btAssert(j < nC);
 
-    btSwapProblem (m_A,m_x,m_b,m_w,m_lo,m_hi,m_p,m_state,m_findex,m_n,i,nC-1,m_nskip,1);
-
-    m_nN++;
-    m_nC = nC - 1; // nC value is outdated after this line
-  }
+		btSwapProblem(m_A, m_x, m_b, m_w, m_lo, m_hi, m_p, m_state, m_findex, m_n, i, nC - 1, m_nskip, 1);
 
+		m_nN++;
+		m_nC = nC - 1;  // nC value is outdated after this line
+	}
 }
 
-
-void btLCP::pN_equals_ANC_times_qC (btScalar *p, btScalar *q)
+void btLCP::pN_equals_ANC_times_qC(btScalar *p, btScalar *q)
 {
-  // we could try to make this matrix-vector multiplication faster using
-  // outer product matrix tricks, e.g. with the dMultidotX() functions.
-  // but i tried it and it actually made things slower on random 100x100
-  // problems because of the overhead involved. so we'll stick with the
-  // simple method for now.
-  const int nC = m_nC;
-  btScalar *ptgt = p + nC;
-  const int nN = m_nN;
-  for (int i=0; i<nN; ++i) {
-    ptgt[i] = btLargeDot (BTAROW(i+nC),q,nC);
-  }
+	// we could try to make this matrix-vector multiplication faster using
+	// outer product matrix tricks, e.g. with the dMultidotX() functions.
+	// but i tried it and it actually made things slower on random 100x100
+	// problems because of the overhead involved. so we'll stick with the
+	// simple method for now.
+	const int nC = m_nC;
+	btScalar *ptgt = p + nC;
+	const int nN = m_nN;
+	for (int i = 0; i < nN; ++i)
+	{
+		ptgt[i] = btLargeDot(BTAROW(i + nC), q, nC);
+	}
 }
 
-
-void btLCP::pN_plusequals_ANi (btScalar *p, int i, int sign)
+void btLCP::pN_plusequals_ANi(btScalar *p, int i, int sign)
 {
-  const int nC = m_nC;
-  btScalar *aptr = BTAROW(i) + nC;
-  btScalar *ptgt = p + nC;
-  if (sign > 0) {
-    const int nN = m_nN;
-    for (int j=0; j<nN; ++j) ptgt[j] += aptr[j];
-  }
-  else {
-    const int nN = m_nN;
-    for (int j=0; j<nN; ++j) ptgt[j] -= aptr[j];
-  }
+	const int nC = m_nC;
+	btScalar *aptr = BTAROW(i) + nC;
+	btScalar *ptgt = p + nC;
+	if (sign > 0)
+	{
+		const int nN = m_nN;
+		for (int j = 0; j < nN; ++j) ptgt[j] += aptr[j];
+	}
+	else
+	{
+		const int nN = m_nN;
+		for (int j = 0; j < nN; ++j) ptgt[j] -= aptr[j];
+	}
 }
 
-void btLCP::pC_plusequals_s_times_qC (btScalar *p, btScalar s, btScalar *q)
+void btLCP::pC_plusequals_s_times_qC(btScalar *p, btScalar s, btScalar *q)
 {
-  const int nC = m_nC;
-  for (int i=0; i<nC; ++i) {
-    p[i] += s*q[i];
-  }
+	const int nC = m_nC;
+	for (int i = 0; i < nC; ++i)
+	{
+		p[i] += s * q[i];
+	}
 }
 
-void btLCP::pN_plusequals_s_times_qN (btScalar *p, btScalar s, btScalar *q)
+void btLCP::pN_plusequals_s_times_qN(btScalar *p, btScalar s, btScalar *q)
 {
-  const int nC = m_nC;
-  btScalar *ptgt = p + nC, *qsrc = q + nC;
-  const int nN = m_nN;
-  for (int i=0; i<nN; ++i) {
-    ptgt[i] += s*qsrc[i];
-  }
+	const int nC = m_nC;
+	btScalar *ptgt = p + nC, *qsrc = q + nC;
+	const int nN = m_nN;
+	for (int i = 0; i < nN; ++i)
+	{
+		ptgt[i] += s * qsrc[i];
+	}
 }
 
-void btLCP::solve1 (btScalar *a, int i, int dir, int only_transfer)
+void btLCP::solve1(btScalar *a, int i, int dir, int only_transfer)
 {
-  // the `Dell' and `ell' that are computed here are saved. if index i is
-  // later added to the factorization then they can be reused.
-  //
-  // @@@ question: do we need to solve for entire delta_x??? yes, but
-  //     only if an x goes below 0 during the step.
+	// the `Dell' and `ell' that are computed here are saved. if index i is
+	// later added to the factorization then they can be reused.
+	//
+	// @@@ question: do we need to solve for entire delta_x??? yes, but
+	//     only if an x goes below 0 during the step.
 
-  if (m_nC > 0) {
-    {
-      btScalar *Dell = m_Dell;
-      int *C = m_C;
-      btScalar *aptr = BTAROW(i);
-#   ifdef BTNUB_OPTIMIZATIONS
-      // if nub>0, initial part of aptr[] is guaranteed unpermuted
-      const int nub = m_nub;
-      int j=0;
-      for ( ; j<nub; ++j) Dell[j] = aptr[j];
-      const int nC = m_nC;
-      for ( ; j<nC; ++j) Dell[j] = aptr[C[j]];
-#   else
-      const int nC = m_nC;
-      for (int j=0; j<nC; ++j) Dell[j] = aptr[C[j]];
-#   endif
-    }
-    btSolveL1 (m_L,m_Dell,m_nC,m_nskip);
-    {
-      btScalar *ell = m_ell, *Dell = m_Dell, *d = m_d;
-      const int nC = m_nC;
-      for (int j=0; j<nC; ++j) ell[j] = Dell[j] * d[j];
-    }
+	if (m_nC > 0)
+	{
+		{
+			btScalar *Dell = m_Dell;
+			int *C = m_C;
+			btScalar *aptr = BTAROW(i);
+#ifdef BTNUB_OPTIMIZATIONS
+			// if nub>0, initial part of aptr[] is guaranteed unpermuted
+			const int nub = m_nub;
+			int j = 0;
+			for (; j < nub; ++j) Dell[j] = aptr[j];
+			const int nC = m_nC;
+			for (; j < nC; ++j) Dell[j] = aptr[C[j]];
+#else
+			const int nC = m_nC;
+			for (int j = 0; j < nC; ++j) Dell[j] = aptr[C[j]];
+#endif
+		}
+		btSolveL1(m_L, m_Dell, m_nC, m_nskip);
+		{
+			btScalar *ell = m_ell, *Dell = m_Dell, *d = m_d;
+			const int nC = m_nC;
+			for (int j = 0; j < nC; ++j) ell[j] = Dell[j] * d[j];
+		}
 
-    if (!only_transfer) {
-      btScalar *tmp = m_tmp, *ell = m_ell;
-      {
-        const int nC = m_nC;
-        for (int j=0; j<nC; ++j) tmp[j] = ell[j];
-      }
-      btSolveL1T (m_L,tmp,m_nC,m_nskip);
-      if (dir > 0) {
-        int *C = m_C;
-        btScalar *tmp = m_tmp;
-        const int nC = m_nC;
-        for (int j=0; j<nC; ++j) a[C[j]] = -tmp[j];
-      } else {
-        int *C = m_C;
-        btScalar *tmp = m_tmp;
-        const int nC = m_nC;
-        for (int j=0; j<nC; ++j) a[C[j]] = tmp[j];
-      }
-    }
-  }
+		if (!only_transfer)
+		{
+			btScalar *tmp = m_tmp, *ell = m_ell;
+			{
+				const int nC = m_nC;
+				for (int j = 0; j < nC; ++j) tmp[j] = ell[j];
+			}
+			btSolveL1T(m_L, tmp, m_nC, m_nskip);
+			if (dir > 0)
+			{
+				int *C = m_C;
+				btScalar *tmp = m_tmp;
+				const int nC = m_nC;
+				for (int j = 0; j < nC; ++j) a[C[j]] = -tmp[j];
+			}
+			else
+			{
+				int *C = m_C;
+				btScalar *tmp = m_tmp;
+				const int nC = m_nC;
+				for (int j = 0; j < nC; ++j) a[C[j]] = tmp[j];
+			}
+		}
+	}
 }
 
-
 void btLCP::unpermute()
 {
-  // now we have to un-permute x and w
-  {
-    memcpy (m_tmp,m_x,m_n*sizeof(btScalar));
-    btScalar *x = m_x, *tmp = m_tmp;
-    const int *p = m_p;
-    const int n = m_n;
-    for (int j=0; j<n; ++j) x[p[j]] = tmp[j];
-  }
-  {
-    memcpy (m_tmp,m_w,m_n*sizeof(btScalar));
-    btScalar *w = m_w, *tmp = m_tmp;
-    const int *p = m_p;
-    const int n = m_n;
-    for (int j=0; j<n; ++j) w[p[j]] = tmp[j];
-  }
+	// now we have to un-permute x and w
+	{
+		memcpy(m_tmp, m_x, m_n * sizeof(btScalar));
+		btScalar *x = m_x, *tmp = m_tmp;
+		const int *p = m_p;
+		const int n = m_n;
+		for (int j = 0; j < n; ++j) x[p[j]] = tmp[j];
+	}
+	{
+		memcpy(m_tmp, m_w, m_n * sizeof(btScalar));
+		btScalar *w = m_w, *tmp = m_tmp;
+		const int *p = m_p;
+		const int n = m_n;
+		for (int j = 0; j < n; ++j) w[p[j]] = tmp[j];
+	}
 }
 
-#endif // btLCP_FAST
-
+#endif  // btLCP_FAST
 
 //***************************************************************************
 // an optimized Dantzig LCP driver routine for the lo-hi LCP problem.
 
-bool btSolveDantzigLCP (int n, btScalar *A, btScalar *x, btScalar *b,
-                btScalar* outer_w, int nub, btScalar *lo, btScalar *hi, int *findex, btDantzigScratchMemory& scratchMem)
+bool btSolveDantzigLCP(int n, btScalar *A, btScalar *x, btScalar *b,
+					   btScalar *outer_w, int nub, btScalar *lo, btScalar *hi, int *findex, btDantzigScratchMemory &scratchMem)
 {
 	s_error = false;
 
-//	printf("btSolveDantzigLCP n=%d\n",n);
-  btAssert (n>0 && A && x && b && lo && hi && nub >= 0 && nub <= n);
-  btAssert(outer_w);
+	//	printf("btSolveDantzigLCP n=%d\n",n);
+	btAssert(n > 0 && A && x && b && lo && hi && nub >= 0 && nub <= n);
+	btAssert(outer_w);
 
 #ifdef BT_DEBUG
-  {
-    // check restrictions on lo and hi
-    for (int k=0; k<n; ++k) 
-		btAssert (lo[k] <= 0 && hi[k] >= 0);
-  }
-# endif
-
-
-  // if all the variables are unbounded then we can just factor, solve,
-  // and return
-  if (nub >= n) 
-  {
-   
-
-    int nskip = (n);
-    btFactorLDLT (A, outer_w, n, nskip);
-    btSolveLDLT (A, outer_w, b, n, nskip);
-    memcpy (x, b, n*sizeof(btScalar));
-
-    return !s_error;
-  }
-
-  const int nskip = (n);
-  scratchMem.L.resize(n*nskip);
-
-  scratchMem.d.resize(n);
-
-  btScalar *w = outer_w;
-  scratchMem.delta_w.resize(n);
-  scratchMem.delta_x.resize(n);
-  scratchMem.Dell.resize(n);
-  scratchMem.ell.resize(n);
-  scratchMem.Arows.resize(n);
-  scratchMem.p.resize(n);
-  scratchMem.C.resize(n);
-
-  // for i in N, state[i] is 0 if x(i)==lo(i) or 1 if x(i)==hi(i)
-  scratchMem.state.resize(n);
-
-
-  // create LCP object. note that tmp is set to delta_w to save space, this
-  // optimization relies on knowledge of how tmp is used, so be careful!
-  btLCP lcp(n,nskip,nub,A,x,b,w,lo,hi,&scratchMem.L[0],&scratchMem.d[0],&scratchMem.Dell[0],&scratchMem.ell[0],&scratchMem.delta_w[0],&scratchMem.state[0],findex,&scratchMem.p[0],&scratchMem.C[0],&scratchMem.Arows[0]);
-  int adj_nub = lcp.getNub();
-
-  // loop over all indexes adj_nub..n-1. for index i, if x(i),w(i) satisfy the
-  // LCP conditions then i is added to the appropriate index set. otherwise
-  // x(i),w(i) is driven either +ve or -ve to force it to the valid region.
-  // as we drive x(i), x(C) is also adjusted to keep w(C) at zero.
-  // while driving x(i) we maintain the LCP conditions on the other variables
-  // 0..i-1. we do this by watching out for other x(i),w(i) values going
-  // outside the valid region, and then switching them between index sets
-  // when that happens.
-
-  bool hit_first_friction_index = false;
-  for (int i=adj_nub; i<n; ++i) 
-  {
-    s_error = false;
-    // the index i is the driving index and indexes i+1..n-1 are "dont care",
-    // i.e. when we make changes to the system those x's will be zero and we
-    // don't care what happens to those w's. in other words, we only consider
-    // an (i+1)*(i+1) sub-problem of A*x=b+w.
-
-    // if we've hit the first friction index, we have to compute the lo and
-    // hi values based on the values of x already computed. we have been
-    // permuting the indexes, so the values stored in the findex vector are
-    // no longer valid. thus we have to temporarily unpermute the x vector. 
-    // for the purposes of this computation, 0*infinity = 0 ... so if the
-    // contact constraint's normal force is 0, there should be no tangential
-    // force applied.
-
-    if (!hit_first_friction_index && findex && findex[i] >= 0) {
-      // un-permute x into delta_w, which is not being used at the moment
-      for (int j=0; j<n; ++j) scratchMem.delta_w[scratchMem.p[j]] = x[j];
-
-      // set lo and hi values
-      for (int k=i; k<n; ++k) {
-        btScalar wfk = scratchMem.delta_w[findex[k]];
-        if (wfk == 0) {
-          hi[k] = 0;
-          lo[k] = 0;
-        }
-        else {
-          hi[k] = btFabs (hi[k] * wfk);
-          lo[k] = -hi[k];
-        }
-      }
-      hit_first_friction_index = true;
-    }
-
-    // thus far we have not even been computing the w values for indexes
-    // greater than i, so compute w[i] now.
-    w[i] = lcp.AiC_times_qC (i,x) + lcp.AiN_times_qN (i,x) - b[i];
-
-    // if lo=hi=0 (which can happen for tangential friction when normals are
-    // 0) then the index will be assigned to set N with some state. however,
-    // set C's line has zero size, so the index will always remain in set N.
-    // with the "normal" switching logic, if w changed sign then the index
-    // would have to switch to set C and then back to set N with an inverted
-    // state. this is pointless, and also computationally expensive. to
-    // prevent this from happening, we use the rule that indexes with lo=hi=0
-    // will never be checked for set changes. this means that the state for
-    // these indexes may be incorrect, but that doesn't matter.
-
-    // see if x(i),w(i) is in a valid region
-    if (lo[i]==0 && w[i] >= 0) {
-      lcp.transfer_i_to_N (i);
-      scratchMem.state[i] = false;
-    }
-    else if (hi[i]==0 && w[i] <= 0) {
-      lcp.transfer_i_to_N (i);
-      scratchMem.state[i] = true;
-    }
-    else if (w[i]==0) {
-      // this is a degenerate case. by the time we get to this test we know
-      // that lo != 0, which means that lo < 0 as lo is not allowed to be +ve,
-      // and similarly that hi > 0. this means that the line segment
-      // corresponding to set C is at least finite in extent, and we are on it.
-      // NOTE: we must call lcp.solve1() before lcp.transfer_i_to_C()
-      lcp.solve1 (&scratchMem.delta_x[0],i,0,1);
-
-      lcp.transfer_i_to_C (i);
-    }
-    else {
-      // we must push x(i) and w(i)
-      for (;;) {
-        int dir;
-        btScalar dirf;
-        // find direction to push on x(i)
-        if (w[i] <= 0) {
-          dir = 1;
-          dirf = btScalar(1.0);
-        }
-        else {
-          dir = -1;
-          dirf = btScalar(-1.0);
-        }
-
-        // compute: delta_x(C) = -dir*A(C,C)\A(C,i)
-        lcp.solve1 (&scratchMem.delta_x[0],i,dir);
-
-        // note that delta_x[i] = dirf, but we wont bother to set it
-
-        // compute: delta_w = A*delta_x ... note we only care about
-        // delta_w(N) and delta_w(i), the rest is ignored
-        lcp.pN_equals_ANC_times_qC (&scratchMem.delta_w[0],&scratchMem.delta_x[0]);
-        lcp.pN_plusequals_ANi (&scratchMem.delta_w[0],i,dir);
-        scratchMem.delta_w[i] = lcp.AiC_times_qC (i,&scratchMem.delta_x[0]) + lcp.Aii(i)*dirf;
-
-        // find largest step we can take (size=s), either to drive x(i),w(i)
-        // to the valid LCP region or to drive an already-valid variable
-        // outside the valid region.
-
-        int cmd = 1;		// index switching command
-        int si = 0;		// si = index to switch if cmd>3
-        btScalar s = -w[i]/scratchMem.delta_w[i];
-        if (dir > 0) {
-          if (hi[i] < BT_INFINITY) {
-            btScalar s2 = (hi[i]-x[i])*dirf;	// was (hi[i]-x[i])/dirf	// step to x(i)=hi(i)
-            if (s2 < s) {
-              s = s2;
-              cmd = 3;
-            }
-          }
-        }
-        else {
-          if (lo[i] > -BT_INFINITY) {
-            btScalar s2 = (lo[i]-x[i])*dirf;	// was (lo[i]-x[i])/dirf	// step to x(i)=lo(i)
-            if (s2 < s) {
-              s = s2;
-              cmd = 2;
-            }
-          }
-        }
-
-        {
-          const int numN = lcp.numN();
-          for (int k=0; k < numN; ++k) {
-            const int indexN_k = lcp.indexN(k);
-            if (!scratchMem.state[indexN_k] ? scratchMem.delta_w[indexN_k] < 0 : scratchMem.delta_w[indexN_k] > 0) {
-                // don't bother checking if lo=hi=0
-                if (lo[indexN_k] == 0 && hi[indexN_k] == 0) continue;
-                btScalar s2 = -w[indexN_k] / scratchMem.delta_w[indexN_k];
-                if (s2 < s) {
-                  s = s2;
-                  cmd = 4;
-                  si = indexN_k;
-                }
-            }
-          }
-        }
-
-        {
-          const int numC = lcp.numC();
-          for (int k=adj_nub; k < numC; ++k) {
-            const int indexC_k = lcp.indexC(k);
-            if (scratchMem.delta_x[indexC_k] < 0 && lo[indexC_k] > -BT_INFINITY) {
-              btScalar s2 = (lo[indexC_k]-x[indexC_k]) / scratchMem.delta_x[indexC_k];
-              if (s2 < s) {
-                s = s2;
-                cmd = 5;
-                si = indexC_k;
-              }
-            }
-            if (scratchMem.delta_x[indexC_k] > 0 && hi[indexC_k] < BT_INFINITY) {
-              btScalar s2 = (hi[indexC_k]-x[indexC_k]) / scratchMem.delta_x[indexC_k];
-              if (s2 < s) {
-                s = s2;
-                cmd = 6;
-                si = indexC_k;
-              }
-            }
-          }
-        }
-
-        //static char* cmdstring[8] = {0,"->C","->NL","->NH","N->C",
-        //			     "C->NL","C->NH"};
-        //printf ("cmd=%d (%s), si=%d\n",cmd,cmdstring[cmd],(cmd>3) ? si : i);
-
-        // if s <= 0 then we've got a problem. if we just keep going then
-        // we're going to get stuck in an infinite loop. instead, just cross
-        // our fingers and exit with the current solution.
-        if (s <= btScalar(0.0)) 
-		{
-//          printf("LCP internal error, s <= 0 (s=%.4e)",(double)s);
-          if (i < n) {
-            btSetZero (x+i,n-i);
-            btSetZero (w+i,n-i);
-          }
-          s_error = true;
-          break;
-        }
-
-        // apply x = x + s * delta_x
-        lcp.pC_plusequals_s_times_qC (x, s, &scratchMem.delta_x[0]);
-        x[i] += s * dirf;
-
-        // apply w = w + s * delta_w
-        lcp.pN_plusequals_s_times_qN (w, s, &scratchMem.delta_w[0]);
-        w[i] += s * scratchMem.delta_w[i];
-
-//        void *tmpbuf;
-        // switch indexes between sets if necessary
-        switch (cmd) {
-        case 1:		// done
-          w[i] = 0;
-          lcp.transfer_i_to_C (i);
-          break;
-        case 2:		// done
-          x[i] = lo[i];
-          scratchMem.state[i] = false;
-          lcp.transfer_i_to_N (i);
-          break;
-        case 3:		// done
-          x[i] = hi[i];
-          scratchMem.state[i] = true;
-          lcp.transfer_i_to_N (i);
-          break;
-        case 4:		// keep going
-          w[si] = 0;
-          lcp.transfer_i_from_N_to_C (si);
-          break;
-        case 5:		// keep going
-          x[si] = lo[si];
-          scratchMem.state[si] = false;
-		  lcp.transfer_i_from_C_to_N (si, scratchMem.m_scratch);
-          break;
-        case 6:		// keep going
-          x[si] = hi[si];
-          scratchMem.state[si] = true;
-          lcp.transfer_i_from_C_to_N (si, scratchMem.m_scratch);
-          break;
-        }
-
-        if (cmd <= 3) break;
-      } // for (;;)
-    } // else
-
-    if (s_error) 
 	{
-      break;
-    }
-  } // for (int i=adj_nub; i<n; ++i)
+		// check restrictions on lo and hi
+		for (int k = 0; k < n; ++k)
+			btAssert(lo[k] <= 0 && hi[k] >= 0);
+	}
+#endif
 
-  lcp.unpermute();
+	// if all the variables are unbounded then we can just factor, solve,
+	// and return
+	if (nub >= n)
+	{
+		int nskip = (n);
+		btFactorLDLT(A, outer_w, n, nskip);
+		btSolveLDLT(A, outer_w, b, n, nskip);
+		memcpy(x, b, n * sizeof(btScalar));
 
+		return !s_error;
+	}
 
-  return !s_error;
+	const int nskip = (n);
+	scratchMem.L.resize(n * nskip);
+
+	scratchMem.d.resize(n);
+
+	btScalar *w = outer_w;
+	scratchMem.delta_w.resize(n);
+	scratchMem.delta_x.resize(n);
+	scratchMem.Dell.resize(n);
+	scratchMem.ell.resize(n);
+	scratchMem.Arows.resize(n);
+	scratchMem.p.resize(n);
+	scratchMem.C.resize(n);
+
+	// for i in N, state[i] is 0 if x(i)==lo(i) or 1 if x(i)==hi(i)
+	scratchMem.state.resize(n);
+
+	// create LCP object. note that tmp is set to delta_w to save space, this
+	// optimization relies on knowledge of how tmp is used, so be careful!
+	btLCP lcp(n, nskip, nub, A, x, b, w, lo, hi, &scratchMem.L[0], &scratchMem.d[0], &scratchMem.Dell[0], &scratchMem.ell[0], &scratchMem.delta_w[0], &scratchMem.state[0], findex, &scratchMem.p[0], &scratchMem.C[0], &scratchMem.Arows[0]);
+	int adj_nub = lcp.getNub();
+
+	// loop over all indexes adj_nub..n-1. for index i, if x(i),w(i) satisfy the
+	// LCP conditions then i is added to the appropriate index set. otherwise
+	// x(i),w(i) is driven either +ve or -ve to force it to the valid region.
+	// as we drive x(i), x(C) is also adjusted to keep w(C) at zero.
+	// while driving x(i) we maintain the LCP conditions on the other variables
+	// 0..i-1. we do this by watching out for other x(i),w(i) values going
+	// outside the valid region, and then switching them between index sets
+	// when that happens.
+
+	bool hit_first_friction_index = false;
+	for (int i = adj_nub; i < n; ++i)
+	{
+		s_error = false;
+		// the index i is the driving index and indexes i+1..n-1 are "dont care",
+		// i.e. when we make changes to the system those x's will be zero and we
+		// don't care what happens to those w's. in other words, we only consider
+		// an (i+1)*(i+1) sub-problem of A*x=b+w.
+
+		// if we've hit the first friction index, we have to compute the lo and
+		// hi values based on the values of x already computed. we have been
+		// permuting the indexes, so the values stored in the findex vector are
+		// no longer valid. thus we have to temporarily unpermute the x vector.
+		// for the purposes of this computation, 0*infinity = 0 ... so if the
+		// contact constraint's normal force is 0, there should be no tangential
+		// force applied.
+
+		if (!hit_first_friction_index && findex && findex[i] >= 0)
+		{
+			// un-permute x into delta_w, which is not being used at the moment
+			for (int j = 0; j < n; ++j) scratchMem.delta_w[scratchMem.p[j]] = x[j];
+
+			// set lo and hi values
+			for (int k = i; k < n; ++k)
+			{
+				btScalar wfk = scratchMem.delta_w[findex[k]];
+				if (wfk == 0)
+				{
+					hi[k] = 0;
+					lo[k] = 0;
+				}
+				else
+				{
+					hi[k] = btFabs(hi[k] * wfk);
+					lo[k] = -hi[k];
+				}
+			}
+			hit_first_friction_index = true;
+		}
+
+		// thus far we have not even been computing the w values for indexes
+		// greater than i, so compute w[i] now.
+		w[i] = lcp.AiC_times_qC(i, x) + lcp.AiN_times_qN(i, x) - b[i];
+
+		// if lo=hi=0 (which can happen for tangential friction when normals are
+		// 0) then the index will be assigned to set N with some state. however,
+		// set C's line has zero size, so the index will always remain in set N.
+		// with the "normal" switching logic, if w changed sign then the index
+		// would have to switch to set C and then back to set N with an inverted
+		// state. this is pointless, and also computationally expensive. to
+		// prevent this from happening, we use the rule that indexes with lo=hi=0
+		// will never be checked for set changes. this means that the state for
+		// these indexes may be incorrect, but that doesn't matter.
+
+		// see if x(i),w(i) is in a valid region
+		if (lo[i] == 0 && w[i] >= 0)
+		{
+			lcp.transfer_i_to_N(i);
+			scratchMem.state[i] = false;
+		}
+		else if (hi[i] == 0 && w[i] <= 0)
+		{
+			lcp.transfer_i_to_N(i);
+			scratchMem.state[i] = true;
+		}
+		else if (w[i] == 0)
+		{
+			// this is a degenerate case. by the time we get to this test we know
+			// that lo != 0, which means that lo < 0 as lo is not allowed to be +ve,
+			// and similarly that hi > 0. this means that the line segment
+			// corresponding to set C is at least finite in extent, and we are on it.
+			// NOTE: we must call lcp.solve1() before lcp.transfer_i_to_C()
+			lcp.solve1(&scratchMem.delta_x[0], i, 0, 1);
+
+			lcp.transfer_i_to_C(i);
+		}
+		else
+		{
+			// we must push x(i) and w(i)
+			for (;;)
+			{
+				int dir;
+				btScalar dirf;
+				// find direction to push on x(i)
+				if (w[i] <= 0)
+				{
+					dir = 1;
+					dirf = btScalar(1.0);
+				}
+				else
+				{
+					dir = -1;
+					dirf = btScalar(-1.0);
+				}
+
+				// compute: delta_x(C) = -dir*A(C,C)\A(C,i)
+				lcp.solve1(&scratchMem.delta_x[0], i, dir);
+
+				// note that delta_x[i] = dirf, but we wont bother to set it
+
+				// compute: delta_w = A*delta_x ... note we only care about
+				// delta_w(N) and delta_w(i), the rest is ignored
+				lcp.pN_equals_ANC_times_qC(&scratchMem.delta_w[0], &scratchMem.delta_x[0]);
+				lcp.pN_plusequals_ANi(&scratchMem.delta_w[0], i, dir);
+				scratchMem.delta_w[i] = lcp.AiC_times_qC(i, &scratchMem.delta_x[0]) + lcp.Aii(i) * dirf;
+
+				// find largest step we can take (size=s), either to drive x(i),w(i)
+				// to the valid LCP region or to drive an already-valid variable
+				// outside the valid region.
+
+				int cmd = 1;  // index switching command
+				int si = 0;   // si = index to switch if cmd>3
+				btScalar s = -w[i] / scratchMem.delta_w[i];
+				if (dir > 0)
+				{
+					if (hi[i] < BT_INFINITY)
+					{
+						btScalar s2 = (hi[i] - x[i]) * dirf;  // was (hi[i]-x[i])/dirf	// step to x(i)=hi(i)
+						if (s2 < s)
+						{
+							s = s2;
+							cmd = 3;
+						}
+					}
+				}
+				else
+				{
+					if (lo[i] > -BT_INFINITY)
+					{
+						btScalar s2 = (lo[i] - x[i]) * dirf;  // was (lo[i]-x[i])/dirf	// step to x(i)=lo(i)
+						if (s2 < s)
+						{
+							s = s2;
+							cmd = 2;
+						}
+					}
+				}
+
+				{
+					const int numN = lcp.numN();
+					for (int k = 0; k < numN; ++k)
+					{
+						const int indexN_k = lcp.indexN(k);
+						if (!scratchMem.state[indexN_k] ? scratchMem.delta_w[indexN_k] < 0 : scratchMem.delta_w[indexN_k] > 0)
+						{
+							// don't bother checking if lo=hi=0
+							if (lo[indexN_k] == 0 && hi[indexN_k] == 0) continue;
+							btScalar s2 = -w[indexN_k] / scratchMem.delta_w[indexN_k];
+							if (s2 < s)
+							{
+								s = s2;
+								cmd = 4;
+								si = indexN_k;
+							}
+						}
+					}
+				}
+
+				{
+					const int numC = lcp.numC();
+					for (int k = adj_nub; k < numC; ++k)
+					{
+						const int indexC_k = lcp.indexC(k);
+						if (scratchMem.delta_x[indexC_k] < 0 && lo[indexC_k] > -BT_INFINITY)
+						{
+							btScalar s2 = (lo[indexC_k] - x[indexC_k]) / scratchMem.delta_x[indexC_k];
+							if (s2 < s)
+							{
+								s = s2;
+								cmd = 5;
+								si = indexC_k;
+							}
+						}
+						if (scratchMem.delta_x[indexC_k] > 0 && hi[indexC_k] < BT_INFINITY)
+						{
+							btScalar s2 = (hi[indexC_k] - x[indexC_k]) / scratchMem.delta_x[indexC_k];
+							if (s2 < s)
+							{
+								s = s2;
+								cmd = 6;
+								si = indexC_k;
+							}
+						}
+					}
+				}
+
+				//static char* cmdstring[8] = {0,"->C","->NL","->NH","N->C",
+				//			     "C->NL","C->NH"};
+				//printf ("cmd=%d (%s), si=%d\n",cmd,cmdstring[cmd],(cmd>3) ? si : i);
+
+				// if s <= 0 then we've got a problem. if we just keep going then
+				// we're going to get stuck in an infinite loop. instead, just cross
+				// our fingers and exit with the current solution.
+				if (s <= btScalar(0.0))
+				{
+					//          printf("LCP internal error, s <= 0 (s=%.4e)",(double)s);
+					if (i < n)
+					{
+						btSetZero(x + i, n - i);
+						btSetZero(w + i, n - i);
+					}
+					s_error = true;
+					break;
+				}
+
+				// apply x = x + s * delta_x
+				lcp.pC_plusequals_s_times_qC(x, s, &scratchMem.delta_x[0]);
+				x[i] += s * dirf;
+
+				// apply w = w + s * delta_w
+				lcp.pN_plusequals_s_times_qN(w, s, &scratchMem.delta_w[0]);
+				w[i] += s * scratchMem.delta_w[i];
+
+				//        void *tmpbuf;
+				// switch indexes between sets if necessary
+				switch (cmd)
+				{
+					case 1:  // done
+						w[i] = 0;
+						lcp.transfer_i_to_C(i);
+						break;
+					case 2:  // done
+						x[i] = lo[i];
+						scratchMem.state[i] = false;
+						lcp.transfer_i_to_N(i);
+						break;
+					case 3:  // done
+						x[i] = hi[i];
+						scratchMem.state[i] = true;
+						lcp.transfer_i_to_N(i);
+						break;
+					case 4:  // keep going
+						w[si] = 0;
+						lcp.transfer_i_from_N_to_C(si);
+						break;
+					case 5:  // keep going
+						x[si] = lo[si];
+						scratchMem.state[si] = false;
+						lcp.transfer_i_from_C_to_N(si, scratchMem.m_scratch);
+						break;
+					case 6:  // keep going
+						x[si] = hi[si];
+						scratchMem.state[si] = true;
+						lcp.transfer_i_from_C_to_N(si, scratchMem.m_scratch);
+						break;
+				}
+
+				if (cmd <= 3) break;
+			}  // for (;;)
+		}      // else
+
+		if (s_error)
+		{
+			break;
+		}
+	}  // for (int i=adj_nub; i<n; ++i)
+
+	lcp.unpermute();
+
+	return !s_error;
 }
-
diff --git a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btDantzigLCP.h b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btDantzigLCP.h
index 903832770ae..8d9b2a13e99 100644
--- a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btDantzigLCP.h
+++ b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btDantzigLCP.h
@@ -41,7 +41,6 @@ to be implemented. the first `nub' variables are assumed to have findex < 0.
 
 */
 
-
 #ifndef _BT_LCP_H_
 #define _BT_LCP_H_
 
@@ -49,7 +48,6 @@ to be implemented. the first `nub' variables are assumed to have findex < 0.
 #include <stdio.h>
 #include <assert.h>
 
-
 #include "LinearMath/btScalar.h"
 #include "LinearMath/btAlignedObjectArray.h"
 
@@ -62,16 +60,14 @@ struct btDantzigScratchMemory
 	btAlignedObjectArray<btScalar> delta_x;
 	btAlignedObjectArray<btScalar> Dell;
 	btAlignedObjectArray<btScalar> ell;
-	btAlignedObjectArray<btScalar*> Arows;
+	btAlignedObjectArray<btScalar *> Arows;
 	btAlignedObjectArray<int> p;
 	btAlignedObjectArray<int> C;
 	btAlignedObjectArray<bool> state;
 };
 
 //return false if solving failed
-bool btSolveDantzigLCP (int n, btScalar *A, btScalar *x, btScalar *b, btScalar *w,
-	int nub, btScalar *lo, btScalar *hi, int *findex,btDantzigScratchMemory& scratch);
+bool btSolveDantzigLCP(int n, btScalar *A, btScalar *x, btScalar *b, btScalar *w,
+					   int nub, btScalar *lo, btScalar *hi, int *findex, btDantzigScratchMemory &scratch);
 
-
-
-#endif //_BT_LCP_H_
+#endif  //_BT_LCP_H_
diff --git a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btDantzigSolver.h b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btDantzigSolver.h
index 2a2f2d3d32d..1f669751ce7 100644
--- a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btDantzigSolver.h
+++ b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btDantzigSolver.h
@@ -20,30 +20,28 @@ subject to the following restrictions:
 #include "btMLCPSolverInterface.h"
 #include "btDantzigLCP.h"
 
-
 class btDantzigSolver : public btMLCPSolverInterface
 {
 protected:
-
 	btScalar m_acceptableUpperLimitSolution;
 
-	btAlignedObjectArray<char>	m_tempBuffer;
+	btAlignedObjectArray<char> m_tempBuffer;
 
 	btAlignedObjectArray<btScalar> m_A;
 	btAlignedObjectArray<btScalar> m_b;
 	btAlignedObjectArray<btScalar> m_x;
 	btAlignedObjectArray<btScalar> m_lo;
 	btAlignedObjectArray<btScalar> m_hi;
-	btAlignedObjectArray<int>	m_dependencies;
+	btAlignedObjectArray<int> m_dependencies;
 	btDantzigScratchMemory m_scratchMemory;
-public:
 
+public:
 	btDantzigSolver()
-		:m_acceptableUpperLimitSolution(btScalar(1000))
+		: m_acceptableUpperLimitSolution(btScalar(1000))
 	{
 	}
 
-	virtual bool solveMLCP(const btMatrixXu & A, const btVectorXu & b, btVectorXu& x, const btVectorXu & lo,const btVectorXu & hi,const btAlignedObjectArray<int>& limitDependency, int numIterations, bool useSparsity = true)
+	virtual bool solveMLCP(const btMatrixXu& A, const btVectorXu& b, btVectorXu& x, const btVectorXu& lo, const btVectorXu& hi, const btAlignedObjectArray<int>& limitDependency, int numIterations, bool useSparsity = true)
 	{
 		bool result = true;
 		int n = b.rows();
@@ -52,14 +50,12 @@ public:
 			int nub = 0;
 			btAlignedObjectArray<btScalar> ww;
 			ww.resize(n);
-	
 
 			const btScalar* Aptr = A.getBufferPointer();
-			m_A.resize(n*n);
-			for (int i=0;i<n*n;i++)
+			m_A.resize(n * n);
+			for (int i = 0; i < n * n; i++)
 			{
 				m_A[i] = Aptr[i];
-
 			}
 
 			m_b.resize(n);
@@ -67,7 +63,7 @@ public:
 			m_lo.resize(n);
 			m_hi.resize(n);
 			m_dependencies.resize(n);
-			for (int i=0;i<n;i++)
+			for (int i = 0; i < n; i++)
 			{
 				m_lo[i] = lo[i];
 				m_hi[i] = hi[i];
@@ -76,13 +72,12 @@ public:
 				m_dependencies[i] = limitDependency[i];
 			}
 
-
-			result = btSolveDantzigLCP (n,&m_A[0],&m_x[0],&m_b[0],&ww[0],nub,&m_lo[0],&m_hi[0],&m_dependencies[0],m_scratchMemory);
+			result = btSolveDantzigLCP(n, &m_A[0], &m_x[0], &m_b[0], &ww[0], nub, &m_lo[0], &m_hi[0], &m_dependencies[0], m_scratchMemory);
 			if (!result)
 				return result;
 
-//			printf("numAllocas = %d\n",numAllocas);
-			for (int i=0;i<n;i++)
+			//			printf("numAllocas = %d\n",numAllocas);
+			for (int i = 0; i < n; i++)
 			{
 				volatile btScalar xx = m_x[i];
 				if (xx != m_x[i])
@@ -98,15 +93,14 @@ public:
 				}
 			}
 
-			for (int i=0;i<n;i++)
+			for (int i = 0; i < n; i++)
 			{
 				x[i] = m_x[i];
 			}
-			
 		}
 
 		return result;
 	}
 };
 
-#endif //BT_DANTZIG_SOLVER_H
+#endif  //BT_DANTZIG_SOLVER_H
diff --git a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btLemkeAlgorithm.cpp b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btLemkeAlgorithm.cpp
index 1f4015c7c72..954ffaed759 100644
--- a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btLemkeAlgorithm.cpp
+++ b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btLemkeAlgorithm.cpp
@@ -19,64 +19,60 @@ subject to the following restrictions:
 //Math library was replaced from fmatvec to a the file src/LinearMath/btMatrixX.h
 //STL/std::vector replaced by btAlignedObjectArray
 
-
-
 #include "btLemkeAlgorithm.h"
 
 #undef BT_DEBUG_OSTREAM
 #ifdef BT_DEBUG_OSTREAM
 using namespace std;
-#endif //BT_DEBUG_OSTREAM
+#endif  //BT_DEBUG_OSTREAM
 
 btScalar btMachEps()
 {
-	static bool calculated=false;
+	static bool calculated = false;
 	static btScalar machEps = btScalar(1.);
 	if (!calculated)
 	{
-		do {
+		do
+		{
 			machEps /= btScalar(2.0);
 			// If next epsilon yields 1, then break, because current
 			// epsilon is the machine epsilon.
-		}
-		while ((btScalar)(1.0 + (machEps/btScalar(2.0))) != btScalar(1.0));
-//		printf( "\nCalculated Machine epsilon: %G\n", machEps );
-		calculated=true;
+		} while ((btScalar)(1.0 + (machEps / btScalar(2.0))) != btScalar(1.0));
+		//		printf( "\nCalculated Machine epsilon: %G\n", machEps );
+		calculated = true;
 	}
 	return machEps;
 }
 
-btScalar btEpsRoot() {
-	
+btScalar btEpsRoot()
+{
 	static btScalar epsroot = 0.;
 	static bool alreadyCalculated = false;
-	
-	if (!alreadyCalculated) {
+
+	if (!alreadyCalculated)
+	{
 		epsroot = btSqrt(btMachEps());
 		alreadyCalculated = true;
 	}
 	return epsroot;
 }
 
-	 
-
-  btVectorXu btLemkeAlgorithm::solve(unsigned int maxloops /* = 0*/)
+btVectorXu btLemkeAlgorithm::solve(unsigned int maxloops /* = 0*/)
 {
-  
-    
-    steps = 0;
+	steps = 0;
 
-    int dim = m_q.size();
+	int dim = m_q.size();
 #ifdef BT_DEBUG_OSTREAM
-    if(DEBUGLEVEL >= 1) {
-      cout << "Dimension = " << dim << endl;
-    }
-#endif //BT_DEBUG_OSTREAM
+	if (DEBUGLEVEL >= 1)
+	{
+		cout << "Dimension = " << dim << endl;
+	}
+#endif  //BT_DEBUG_OSTREAM
 
 	btVectorXu solutionVector(2 * dim);
 	solutionVector.setZero();
-	  
-	  //, INIT, 0.);
+
+	//, INIT, 0.);
 
 	btMatrixXu ident(dim, dim);
 	ident.setIdentity();
@@ -85,287 +81,289 @@ btScalar btEpsRoot() {
 #endif
 
 	btMatrixXu mNeg = m_M.negative();
-	  
-    btMatrixXu A(dim, 2 * dim + 2);
+
+	btMatrixXu A(dim, 2 * dim + 2);
 	//
-	A.setSubMatrix(0, 0, dim - 1, dim - 1,ident);
-	A.setSubMatrix(0, dim, dim - 1, 2 * dim - 1,mNeg);
+	A.setSubMatrix(0, 0, dim - 1, dim - 1, ident);
+	A.setSubMatrix(0, dim, dim - 1, 2 * dim - 1, mNeg);
 	A.setSubMatrix(0, 2 * dim, dim - 1, 2 * dim, -1.f);
-	A.setSubMatrix(0, 2 * dim + 1, dim - 1, 2 * dim + 1,m_q);
+	A.setSubMatrix(0, 2 * dim + 1, dim - 1, 2 * dim + 1, m_q);
 
 #ifdef BT_DEBUG_OSTREAM
 	cout << A << std::endl;
-#endif //BT_DEBUG_OSTREAM
+#endif  //BT_DEBUG_OSTREAM
 
+	//   btVectorXu q_;
+	//   q_ >> A(0, 2 * dim + 1, dim - 1, 2 * dim + 1);
 
- //   btVectorXu q_;
- //   q_ >> A(0, 2 * dim + 1, dim - 1, 2 * dim + 1);
-
-    btAlignedObjectArray<int> basis;
-    //At first, all w-values are in the basis
-    for (int i = 0; i < dim; i++)
-      basis.push_back(i);
+	btAlignedObjectArray<int> basis;
+	//At first, all w-values are in the basis
+	for (int i = 0; i < dim; i++)
+		basis.push_back(i);
 
 	int pivotRowIndex = -1;
 	btScalar minValue = 1e30f;
 	bool greaterZero = true;
-	for (int i=0;i<dim;i++)
+	for (int i = 0; i < dim; i++)
 	{
-		btScalar v =A(i,2*dim+1);
-		if (v<minValue)
+		btScalar v = A(i, 2 * dim + 1);
+		if (v < minValue)
 		{
-			minValue=v;
+			minValue = v;
 			pivotRowIndex = i;
 		}
-		if (v<0)
+		if (v < 0)
 			greaterZero = false;
 	}
-	
 
-	
-  //  int pivotRowIndex = q_.minIndex();//minIndex(q_);     // first row is that with lowest q-value
-    int z0Row = pivotRowIndex;           // remember the col of z0 for ending algorithm afterwards
-    int pivotColIndex = 2 * dim;         // first col is that of z0
+	//  int pivotRowIndex = q_.minIndex();//minIndex(q_);     // first row is that with lowest q-value
+	int z0Row = pivotRowIndex;    // remember the col of z0 for ending algorithm afterwards
+	int pivotColIndex = 2 * dim;  // first col is that of z0
 
 #ifdef BT_DEBUG_OSTREAM
-    if (DEBUGLEVEL >= 3)
+	if (DEBUGLEVEL >= 3)
 	{
-    //  cout << "A: " << A << endl;
-      cout << "pivotRowIndex " << pivotRowIndex << endl;
-      cout << "pivotColIndex " << pivotColIndex << endl;
-      cout << "Basis: ";
-      for (int i = 0; i < basis.size(); i++)
-        cout << basis[i] << " ";
-      cout << endl;
-    }
-#endif //BT_DEBUG_OSTREAM
+		//  cout << "A: " << A << endl;
+		cout << "pivotRowIndex " << pivotRowIndex << endl;
+		cout << "pivotColIndex " << pivotColIndex << endl;
+		cout << "Basis: ";
+		for (int i = 0; i < basis.size(); i++)
+			cout << basis[i] << " ";
+		cout << endl;
+	}
+#endif  //BT_DEBUG_OSTREAM
 
 	if (!greaterZero)
 	{
+		if (maxloops == 0)
+		{
+			maxloops = 100;
+			//        maxloops = UINT_MAX; //TODO: not a really nice way, problem is: maxloops should be 2^dim (=1<<dim), but this could exceed UINT_MAX and thus the result would be 0 and therefore the lemke algorithm wouldn't start but probably would find a solution within less then UINT_MAX steps. Therefore this constant is used as a upper border right now...
+		}
 
-      if (maxloops == 0) {
-		  maxloops = 100;
-//        maxloops = UINT_MAX; //TODO: not a really nice way, problem is: maxloops should be 2^dim (=1<<dim), but this could exceed UINT_MAX and thus the result would be 0 and therefore the lemke algorithm wouldn't start but probably would find a solution within less then UINT_MAX steps. Therefore this constant is used as a upper border right now...
-      }
-
-      /*start looping*/
-      for(steps = 0; steps < maxloops; steps++) {
-
-        GaussJordanEliminationStep(A, pivotRowIndex, pivotColIndex, basis);
+		/*start looping*/
+		for (steps = 0; steps < maxloops; steps++)
+		{
+			GaussJordanEliminationStep(A, pivotRowIndex, pivotColIndex, basis);
 #ifdef BT_DEBUG_OSTREAM
-        if (DEBUGLEVEL >= 3) {
-        //  cout << "A: " << A << endl;
-          cout << "pivotRowIndex " << pivotRowIndex << endl;
-          cout << "pivotColIndex " << pivotColIndex << endl;
-          cout << "Basis: ";
-          for (int i = 0; i < basis.size(); i++)
-            cout << basis[i] << " ";
-          cout << endl;
-        }
-#endif //BT_DEBUG_OSTREAM
+			if (DEBUGLEVEL >= 3)
+			{
+				//  cout << "A: " << A << endl;
+				cout << "pivotRowIndex " << pivotRowIndex << endl;
+				cout << "pivotColIndex " << pivotColIndex << endl;
+				cout << "Basis: ";
+				for (int i = 0; i < basis.size(); i++)
+					cout << basis[i] << " ";
+				cout << endl;
+			}
+#endif  //BT_DEBUG_OSTREAM
 
-        int pivotColIndexOld = pivotColIndex;
+			int pivotColIndexOld = pivotColIndex;
 
-        /*find new column index */
-        if (basis[pivotRowIndex] < dim) //if a w-value left the basis get in the correspondent z-value
-          pivotColIndex = basis[pivotRowIndex] + dim;
-        else
-          //else do it the other way round and get in the corresponding w-value
-          pivotColIndex = basis[pivotRowIndex] - dim;
+			/*find new column index */
+			if (basis[pivotRowIndex] < dim)  //if a w-value left the basis get in the correspondent z-value
+				pivotColIndex = basis[pivotRowIndex] + dim;
+			else
+				//else do it the other way round and get in the corresponding w-value
+				pivotColIndex = basis[pivotRowIndex] - dim;
 
-        /*the column becomes part of the basis*/
-        basis[pivotRowIndex] = pivotColIndexOld;
+			/*the column becomes part of the basis*/
+			basis[pivotRowIndex] = pivotColIndexOld;
 
-        pivotRowIndex = findLexicographicMinimum(A, pivotColIndex);
+			pivotRowIndex = findLexicographicMinimum(A, pivotColIndex);
 
-        if(z0Row == pivotRowIndex) { //if z0 leaves the basis the solution is found --> one last elimination step is necessary
-          GaussJordanEliminationStep(A, pivotRowIndex, pivotColIndex, basis);
-          basis[pivotRowIndex] = pivotColIndex; //update basis
-          break;
-      }
-
-      }
+			if (z0Row == pivotRowIndex)
+			{  //if z0 leaves the basis the solution is found --> one last elimination step is necessary
+				GaussJordanEliminationStep(A, pivotRowIndex, pivotColIndex, basis);
+				basis[pivotRowIndex] = pivotColIndex;  //update basis
+				break;
+			}
+		}
 #ifdef BT_DEBUG_OSTREAM
-      if(DEBUGLEVEL >= 1) {
-        cout << "Number of loops: " << steps << endl;
-        cout << "Number of maximal loops: " << maxloops << endl;
-      }
-#endif //BT_DEBUG_OSTREAM
+		if (DEBUGLEVEL >= 1)
+		{
+			cout << "Number of loops: " << steps << endl;
+			cout << "Number of maximal loops: " << maxloops << endl;
+		}
+#endif  //BT_DEBUG_OSTREAM
 
-      if(!validBasis(basis)) {
-        info = -1;
+		if (!validBasis(basis))
+		{
+			info = -1;
 #ifdef BT_DEBUG_OSTREAM
-        if(DEBUGLEVEL >= 1)
-          cerr << "Lemke-Algorithm ended with Ray-Termination (no valid solution)." << endl;
-#endif //BT_DEBUG_OSTREAM
+			if (DEBUGLEVEL >= 1)
+				cerr << "Lemke-Algorithm ended with Ray-Termination (no valid solution)." << endl;
+#endif  //BT_DEBUG_OSTREAM
 
-        return solutionVector;
-      }
-
-    }
+			return solutionVector;
+		}
+	}
 #ifdef BT_DEBUG_OSTREAM
-    if (DEBUGLEVEL >= 2) {
-     // cout << "A: " << A << endl;
-      cout << "pivotRowIndex " << pivotRowIndex << endl;
-      cout << "pivotColIndex " << pivotColIndex << endl;
-    }
-#endif //BT_DEBUG_OSTREAM
-
-    for (int i = 0; i < basis.size(); i++)
+	if (DEBUGLEVEL >= 2)
 	{
-      solutionVector[basis[i]] = A(i,2*dim+1);//q_[i];
+		// cout << "A: " << A << endl;
+		cout << "pivotRowIndex " << pivotRowIndex << endl;
+		cout << "pivotColIndex " << pivotColIndex << endl;
+	}
+#endif  //BT_DEBUG_OSTREAM
+
+	for (int i = 0; i < basis.size(); i++)
+	{
+		solutionVector[basis[i]] = A(i, 2 * dim + 1);  //q_[i];
 	}
 
-    info = 0;
+	info = 0;
 
-    return solutionVector;
-  }
+	return solutionVector;
+}
 
-  int btLemkeAlgorithm::findLexicographicMinimum(const btMatrixXu& A, const int & pivotColIndex) {
-	  int RowIndex = 0;
-	  int dim = A.rows();
-	  btAlignedObjectArray<btVectorXu> Rows;
-	  for (int row = 0; row < dim; row++) 
-	  {
-
-		  btVectorXu vec(dim + 1);
-		  vec.setZero();//, INIT, 0.)
-		  Rows.push_back(vec);
-		  btScalar a = A(row, pivotColIndex);
-		  if (a > 0) {
-			  Rows[row][0] = A(row, 2 * dim + 1) / a;
-			  Rows[row][1] = A(row, 2 * dim) / a;
-			  for (int j = 2; j < dim + 1; j++)
-				  Rows[row][j] = A(row, j - 1) / a;
+int btLemkeAlgorithm::findLexicographicMinimum(const btMatrixXu& A, const int& pivotColIndex)
+{
+	int RowIndex = 0;
+	int dim = A.rows();
+	btAlignedObjectArray<btVectorXu> Rows;
+	for (int row = 0; row < dim; row++)
+	{
+		btVectorXu vec(dim + 1);
+		vec.setZero();  //, INIT, 0.)
+		Rows.push_back(vec);
+		btScalar a = A(row, pivotColIndex);
+		if (a > 0)
+		{
+			Rows[row][0] = A(row, 2 * dim + 1) / a;
+			Rows[row][1] = A(row, 2 * dim) / a;
+			for (int j = 2; j < dim + 1; j++)
+				Rows[row][j] = A(row, j - 1) / a;
 
 #ifdef BT_DEBUG_OSTREAM
-		//		if (DEBUGLEVEL) {
-			//	  cout << "Rows(" << row << ") = " << Rows[row] << endl;
+				//		if (DEBUGLEVEL) {
+				//	  cout << "Rows(" << row << ") = " << Rows[row] << endl;
 				// }
-#endif 
-		  }
-	  }
+#endif
+		}
+	}
 
-	  for (int i = 0; i < Rows.size(); i++) 
-	  {
-		  if (Rows[i].nrm2() > 0.) {
+	for (int i = 0; i < Rows.size(); i++)
+	{
+		if (Rows[i].nrm2() > 0.)
+		{
+			int j = 0;
+			for (; j < Rows.size(); j++)
+			{
+				if (i != j)
+				{
+					if (Rows[j].nrm2() > 0.)
+					{
+						btVectorXu test(dim + 1);
+						for (int ii = 0; ii < dim + 1; ii++)
+						{
+							test[ii] = Rows[j][ii] - Rows[i][ii];
+						}
 
-			  int j = 0;
-			  for (; j < Rows.size(); j++) 
-			  {
-				  if(i != j)
-				  {
-					  if(Rows[j].nrm2() > 0.)
-					  {
-						  btVectorXu test(dim + 1);
-						  for (int ii=0;ii<dim+1;ii++)
-						  {
-							  test[ii] = Rows[j][ii] - Rows[i][ii];
-						  }
+						//=Rows[j] - Rows[i]
+						if (!LexicographicPositive(test))
+							break;
+					}
+				}
+			}
 
-						  //=Rows[j] - Rows[i]
-						  if (! LexicographicPositive(test))
-							  break;
-					  }
-				  }
-			  }
+			if (j == Rows.size())
+			{
+				RowIndex += i;
+				break;
+			}
+		}
+	}
 
-			  if (j == Rows.size()) 
-			  {
-				  RowIndex += i;
-				  break;
-			  }
-		  }
-	  }
+	return RowIndex;
+}
 
-	  return RowIndex;
-  }
-
-  bool btLemkeAlgorithm::LexicographicPositive(const btVectorXu & v)
+bool btLemkeAlgorithm::LexicographicPositive(const btVectorXu& v)
 {
-    int i = 0;
-  //  if (DEBUGLEVEL)
-    //  cout << "v " << v << endl;
+	int i = 0;
+	//  if (DEBUGLEVEL)
+	//  cout << "v " << v << endl;
 
-    while(i < v.size()-1 && fabs(v[i]) < btMachEps())
-      i++;
-    if (v[i] > 0)
-      return true;
+	while (i < v.size() - 1 && fabs(v[i]) < btMachEps())
+		i++;
+	if (v[i] > 0)
+		return true;
 
-    return false;
-  }
+	return false;
+}
 
-void btLemkeAlgorithm::GaussJordanEliminationStep(btMatrixXu& A, int pivotRowIndex, int pivotColumnIndex, const btAlignedObjectArray<int>& basis) 
+void btLemkeAlgorithm::GaussJordanEliminationStep(btMatrixXu& A, int pivotRowIndex, int pivotColumnIndex, const btAlignedObjectArray<int>& basis)
 {
-
 	btScalar a = -1 / A(pivotRowIndex, pivotColumnIndex);
 #ifdef BT_DEBUG_OSTREAM
 	cout << A << std::endl;
 #endif
 
-    for (int i = 0; i < A.rows(); i++)
+	for (int i = 0; i < A.rows(); i++)
 	{
-      if (i != pivotRowIndex)
-	  {
-        for (int j = 0; j < A.cols(); j++)
+		if (i != pivotRowIndex)
 		{
-          if (j != pivotColumnIndex)
-		  {
-			  btScalar v = A(i, j);
-			  v += A(pivotRowIndex, j) * A(i, pivotColumnIndex) * a;
-            A.setElem(i, j, v);
-		  }
+			for (int j = 0; j < A.cols(); j++)
+			{
+				if (j != pivotColumnIndex)
+				{
+					btScalar v = A(i, j);
+					v += A(pivotRowIndex, j) * A(i, pivotColumnIndex) * a;
+					A.setElem(i, j, v);
+				}
+			}
 		}
-	  }
 	}
 
 #ifdef BT_DEBUG_OSTREAM
 	cout << A << std::endl;
-#endif //BT_DEBUG_OSTREAM
-    for (int i = 0; i < A.cols(); i++) 
+#endif  //BT_DEBUG_OSTREAM
+	for (int i = 0; i < A.cols(); i++)
 	{
-      A.mulElem(pivotRowIndex, i,-a);
-    }
-#ifdef BT_DEBUG_OSTREAM
-	cout << A << std::endl;
-#endif //#ifdef BT_DEBUG_OSTREAM
-
-    for (int i = 0; i < A.rows(); i++)
-	{
-      if (i != pivotRowIndex)
-	  {
-        A.setElem(i, pivotColumnIndex,0);
-	  }
+		A.mulElem(pivotRowIndex, i, -a);
 	}
 #ifdef BT_DEBUG_OSTREAM
 	cout << A << std::endl;
-#endif //#ifdef BT_DEBUG_OSTREAM
-  }
+#endif  //#ifdef BT_DEBUG_OSTREAM
 
-  bool btLemkeAlgorithm::greaterZero(const btVectorXu & vector)
+	for (int i = 0; i < A.rows(); i++)
+	{
+		if (i != pivotRowIndex)
+		{
+			A.setElem(i, pivotColumnIndex, 0);
+		}
+	}
+#ifdef BT_DEBUG_OSTREAM
+	cout << A << std::endl;
+#endif  //#ifdef BT_DEBUG_OSTREAM
+}
+
+bool btLemkeAlgorithm::greaterZero(const btVectorXu& vector)
 {
-    bool isGreater = true;
-    for (int i = 0; i < vector.size(); i++) {
-      if (vector[i] < 0) {
-        isGreater = false;
-        break;
-      }
-    }
+	bool isGreater = true;
+	for (int i = 0; i < vector.size(); i++)
+	{
+		if (vector[i] < 0)
+		{
+			isGreater = false;
+			break;
+		}
+	}
 
-    return isGreater;
-  }
-
-  bool btLemkeAlgorithm::validBasis(const btAlignedObjectArray<int>& basis) 
-  {
-    bool isValid = true;
-    for (int i = 0; i < basis.size(); i++) {
-      if (basis[i] >= basis.size() * 2) { //then z0 is in the base
-        isValid = false;
-        break;
-      }
-    }
-
-    return isValid;
-  }
+	return isGreater;
+}
 
+bool btLemkeAlgorithm::validBasis(const btAlignedObjectArray<int>& basis)
+{
+	bool isValid = true;
+	for (int i = 0; i < basis.size(); i++)
+	{
+		if (basis[i] >= basis.size() * 2)
+		{  //then z0 is in the base
+			isValid = false;
+			break;
+		}
+	}
 
+	return isValid;
+}
diff --git a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btLemkeAlgorithm.h b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btLemkeAlgorithm.h
index 7555cd9d207..3c6bf72a233 100644
--- a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btLemkeAlgorithm.h
+++ b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btLemkeAlgorithm.h
@@ -19,90 +19,84 @@ subject to the following restrictions:
 //Math library was replaced from fmatvec to a the file src/LinearMath/btMatrixX.h
 //STL/std::vector replaced by btAlignedObjectArray
 
-
-
 #ifndef BT_NUMERICS_LEMKE_ALGORITHM_H_
 #define BT_NUMERICS_LEMKE_ALGORITHM_H_
 
 #include "LinearMath/btMatrixX.h"
 
-
-#include <vector> //todo: replace by btAlignedObjectArray
+#include <vector>  //todo: replace by btAlignedObjectArray
 
 class btLemkeAlgorithm
 {
 public:
- 
+	btLemkeAlgorithm(const btMatrixXu& M_, const btVectorXu& q_, const int& DEBUGLEVEL_ = 0) : DEBUGLEVEL(DEBUGLEVEL_)
+	{
+		setSystem(M_, q_);
+	}
 
-  btLemkeAlgorithm(const btMatrixXu& M_, const btVectorXu& q_, const int & DEBUGLEVEL_ = 0) :
-	  DEBUGLEVEL(DEBUGLEVEL_)
-  {
-	setSystem(M_, q_);
-  }
-
-  /* GETTER / SETTER */
-  /**
+	/* GETTER / SETTER */
+	/**
    * \brief return info of solution process
    */
-  int getInfo() {
-	return info;
-  }
+	int getInfo()
+	{
+		return info;
+	}
 
-  /**
+	/**
    * \brief get the number of steps until the solution was found
    */
-  int getSteps(void) {
-	return steps;
-  }
+	int getSteps(void)
+	{
+		return steps;
+	}
 
-
-
-  /**
+	/**
    * \brief set system with Matrix M and vector q
    */
-  void setSystem(const btMatrixXu & M_, const btVectorXu & q_)
+	void setSystem(const btMatrixXu& M_, const btVectorXu& q_)
 	{
 		m_M = M_;
 		m_q = q_;
-  }
-  /***************************************************/
+	}
+	/***************************************************/
 
-  /**
+	/**
    * \brief solve algorithm adapted from : Fast Implementation of Lemke’s Algorithm for Rigid Body Contact Simulation (John E. Lloyd)
    */
-  btVectorXu solve(unsigned int maxloops = 0);
+	btVectorXu solve(unsigned int maxloops = 0);
 
-  virtual ~btLemkeAlgorithm() {
-  }
+	virtual ~btLemkeAlgorithm()
+	{
+	}
 
 protected:
-  int findLexicographicMinimum(const btMatrixXu &A, const int & pivotColIndex);
-  bool LexicographicPositive(const btVectorXu & v);
-  void GaussJordanEliminationStep(btMatrixXu &A, int pivotRowIndex, int pivotColumnIndex, const btAlignedObjectArray<int>& basis);
-  bool greaterZero(const btVectorXu & vector);
-  bool validBasis(const btAlignedObjectArray<int>& basis);
+	int findLexicographicMinimum(const btMatrixXu& A, const int& pivotColIndex);
+	bool LexicographicPositive(const btVectorXu& v);
+	void GaussJordanEliminationStep(btMatrixXu& A, int pivotRowIndex, int pivotColumnIndex, const btAlignedObjectArray<int>& basis);
+	bool greaterZero(const btVectorXu& vector);
+	bool validBasis(const btAlignedObjectArray<int>& basis);
 
-  btMatrixXu m_M;
-  btVectorXu m_q;
+	btMatrixXu m_M;
+	btVectorXu m_q;
 
-  /**
+	/**
    * \brief number of steps until the Lemke algorithm found a solution
    */
-  unsigned int steps;
+	unsigned int steps;
 
-  /**
+	/**
    * \brief define level of debug output
    */
-  int DEBUGLEVEL;
+	int DEBUGLEVEL;
 
-  /**
+	/**
    * \brief did the algorithm find a solution
    *
    * -1 : not successful
    *  0 : successful
    */
-  int info;
+	int info;
 };
 
-
 #endif /* BT_NUMERICS_LEMKE_ALGORITHM_H_ */
diff --git a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btLemkeSolver.h b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btLemkeSolver.h
index 98484c37964..3f215e56bbb 100644
--- a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btLemkeSolver.h
+++ b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btLemkeSolver.h
@@ -17,334 +17,322 @@ subject to the following restrictions:
 #ifndef BT_LEMKE_SOLVER_H
 #define BT_LEMKE_SOLVER_H
 
-
 #include "btMLCPSolverInterface.h"
 #include "btLemkeAlgorithm.h"
 
-
-
-
 ///The btLemkeSolver is based on "Fast Implementation of Lemke’s Algorithm for Rigid Body Contact Simulation (John E. Lloyd) "
 ///It is a slower but more accurate solver. Increase the m_maxLoops for better convergence, at the cost of more CPU time.
 ///The original implementation of the btLemkeAlgorithm was done by Kilian Grundl from the MBSim team
 class btLemkeSolver : public btMLCPSolverInterface
 {
 protected:
-	
 public:
-
-	btScalar	m_maxValue;
-	int			m_debugLevel;
-	int			m_maxLoops;
-	bool		m_useLoHighBounds;
-	
-	
+	btScalar m_maxValue;
+	int m_debugLevel;
+	int m_maxLoops;
+	bool m_useLoHighBounds;
 
 	btLemkeSolver()
-		:m_maxValue(100000),
-		m_debugLevel(0),
-		m_maxLoops(1000),
-		m_useLoHighBounds(true)
+		: m_maxValue(100000),
+		  m_debugLevel(0),
+		  m_maxLoops(1000),
+		  m_useLoHighBounds(true)
 	{
 	}
-	virtual bool solveMLCP(const btMatrixXu & A, const btVectorXu & b, btVectorXu& x, const btVectorXu & lo,const btVectorXu & hi,const btAlignedObjectArray<int>& limitDependency, int numIterations, bool useSparsity = true)
+	virtual bool solveMLCP(const btMatrixXu& A, const btVectorXu& b, btVectorXu& x, const btVectorXu& lo, const btVectorXu& hi, const btAlignedObjectArray<int>& limitDependency, int numIterations, bool useSparsity = true)
 	{
-		
 		if (m_useLoHighBounds)
 		{
+			BT_PROFILE("btLemkeSolver::solveMLCP");
+			int n = A.rows();
+			if (0 == n)
+				return true;
 
-		BT_PROFILE("btLemkeSolver::solveMLCP");
-		int n = A.rows();
-		if (0==n)
-			return true;
-		
-		bool fail = false;
+			bool fail = false;
 
-		btVectorXu solution(n);
-		btVectorXu q1;
-		q1.resize(n);
-		for (int row=0;row<n;row++)
-		{
-			q1[row] = -b[row];
-		}
+			btVectorXu solution(n);
+			btVectorXu q1;
+			q1.resize(n);
+			for (int row = 0; row < n; row++)
+			{
+				q1[row] = -b[row];
+			}
 
-	//		cout << "A" << endl;
-	//		cout << A << endl;
+			//		cout << "A" << endl;
+			//		cout << A << endl;
 
 			/////////////////////////////////////
 
 			//slow matrix inversion, replace with LU decomposition
 			btMatrixXu A1;
-			btMatrixXu B(n,n);
+			btMatrixXu B(n, n);
 			{
 				BT_PROFILE("inverse(slow)");
-				A1.resize(A.rows(),A.cols());
-				for (int row=0;row<A.rows();row++)
+				A1.resize(A.rows(), A.cols());
+				for (int row = 0; row < A.rows(); row++)
 				{
-					for (int col=0;col<A.cols();col++)
+					for (int col = 0; col < A.cols(); col++)
 					{
-						A1.setElem(row,col,A(row,col));
+						A1.setElem(row, col, A(row, col));
 					}
 				}
 
 				btMatrixXu matrix;
-				matrix.resize(n,2*n);
-				for (int row=0;row<n;row++)
+				matrix.resize(n, 2 * n);
+				for (int row = 0; row < n; row++)
 				{
-					for (int col=0;col<n;col++)
+					for (int col = 0; col < n; col++)
 					{
-						matrix.setElem(row,col,A1(row,col));
+						matrix.setElem(row, col, A1(row, col));
 					}
 				}
 
-
-				btScalar ratio,a;
-				int i,j,k;
-				for(i = 0; i < n; i++){
-				for(j = n; j < 2*n; j++){
-					if(i==(j-n))
-						matrix.setElem(i,j,1.0);
-					else
-						matrix.setElem(i,j,0.0);
-				}
-			}
-			for(i = 0; i < n; i++){
-				for(j = 0; j < n; j++){
-					if(i!=j)
+				btScalar ratio, a;
+				int i, j, k;
+				for (i = 0; i < n; i++)
+				{
+					for (j = n; j < 2 * n; j++)
 					{
-						btScalar v = matrix(i,i);
-						if (btFuzzyZero(v))
+						if (i == (j - n))
+							matrix.setElem(i, j, 1.0);
+						else
+							matrix.setElem(i, j, 0.0);
+					}
+				}
+				for (i = 0; i < n; i++)
+				{
+					for (j = 0; j < n; j++)
+					{
+						if (i != j)
 						{
-							a = 0.000001f;
-						}
-						ratio = matrix(j,i)/matrix(i,i);
-						for(k = 0; k < 2*n; k++){
-							matrix.addElem(j,k,- ratio * matrix(i,k));
+							btScalar v = matrix(i, i);
+							if (btFuzzyZero(v))
+							{
+								a = 0.000001f;
+							}
+							ratio = matrix(j, i) / matrix(i, i);
+							for (k = 0; k < 2 * n; k++)
+							{
+								matrix.addElem(j, k, -ratio * matrix(i, k));
+							}
 						}
 					}
 				}
-			}
-			for(i = 0; i < n; i++){
-				a = matrix(i,i);
-				if (btFuzzyZero(a))
+				for (i = 0; i < n; i++)
 				{
-					a = 0.000001f;
-				}
-				btScalar invA = 1.f/a;
-				for(j = 0; j < 2*n; j++){
-					matrix.mulElem(i,j,invA);
-				}
-			}
-
-	
-
-	
-
-			for (int row=0;row<n;row++)
-				{
-					for (int col=0;col<n;col++)
+					a = matrix(i, i);
+					if (btFuzzyZero(a))
 					{
-						B.setElem(row,col,matrix(row,n+col));
+						a = 0.000001f;
+					}
+					btScalar invA = 1.f / a;
+					for (j = 0; j < 2 * n; j++)
+					{
+						matrix.mulElem(i, j, invA);
+					}
+				}
+
+				for (int row = 0; row < n; row++)
+				{
+					for (int col = 0; col < n; col++)
+					{
+						B.setElem(row, col, matrix(row, n + col));
 					}
 				}
 			}
 
-		btMatrixXu b1(n,1);
+			btMatrixXu b1(n, 1);
 
-		btMatrixXu M(n*2,n*2);
-		for (int row=0;row<n;row++)
-		{
-			b1.setElem(row,0,-b[row]);
-			for (int col=0;col<n;col++)
+			btMatrixXu M(n * 2, n * 2);
+			for (int row = 0; row < n; row++)
 			{
-				btScalar v =B(row,col);
-				M.setElem(row,col,v);
-				M.setElem(n+row,n+col,v);
-				M.setElem(n+row,col,-v);
-				M.setElem(row,n+col,-v);
-
-			}
-		}
-
-		btMatrixXu Bb1 = B*b1;
-//		q = [ (-B*b1 - lo)'   (hi + B*b1)' ]'
-
-		btVectorXu qq;
-		qq.resize(n*2);
-		for (int row=0;row<n;row++)
-		{
-			qq[row] = -Bb1(row,0)-lo[row];
-			qq[n+row] = Bb1(row,0)+hi[row];
-		}
-
-		btVectorXu z1;
-
-		btMatrixXu y1;
-		y1.resize(n,1);
-		btLemkeAlgorithm lemke(M,qq,m_debugLevel);
-		{
-			BT_PROFILE("lemke.solve");
-			lemke.setSystem(M,qq);
-			z1  = lemke.solve(m_maxLoops);
-		}
-		for (int row=0;row<n;row++)
-		{
-			y1.setElem(row,0,z1[2*n+row]-z1[3*n+row]);
-		}
-		btMatrixXu y1_b1(n,1);
-		for (int i=0;i<n;i++)
-		{
-			y1_b1.setElem(i,0,y1(i,0)-b1(i,0));
-		}
-
-		btMatrixXu x1;
-
-		x1 = B*(y1_b1);
-			
-		for (int row=0;row<n;row++)
-		{
-			solution[row] = x1(row,0);//n];
-		}
-
-		int errorIndexMax = -1;
-		int errorIndexMin = -1;
-		float errorValueMax = -1e30;
-		float errorValueMin = 1e30;
-		
-		for (int i=0;i<n;i++)
-		{
-			x[i] = solution[i];
-			volatile btScalar check = x[i];
-			if (x[i] != check)
-			{
-				//printf("Lemke result is #NAN\n");
-				x.setZero();
-				return false;
-			}
-			
-			//this is some hack/safety mechanism, to discard invalid solutions from the Lemke solver 
-			//we need to figure out why it happens, and fix it, or detect it properly)
-			if (x[i]>m_maxValue)
-			{
-				if (x[i]> errorValueMax)
+				b1.setElem(row, 0, -b[row]);
+				for (int col = 0; col < n; col++)
 				{
-					fail = true;
-					errorIndexMax = i;
-					errorValueMax = x[i];
-				}
-				////printf("x[i] = %f,",x[i]);
-			}
-			if (x[i]<-m_maxValue)
-			{
-				if (x[i]<errorValueMin)
-				{
-					errorIndexMin = i;
-					errorValueMin = x[i];
-					fail = true;
-					//printf("x[i] = %f,",x[i]);
+					btScalar v = B(row, col);
+					M.setElem(row, col, v);
+					M.setElem(n + row, n + col, v);
+					M.setElem(n + row, col, -v);
+					M.setElem(row, n + col, -v);
 				}
 			}
-		}
-		if (fail)
-		{
-			int m_errorCountTimes = 0;
-			if (errorIndexMin<0)
-				errorValueMin = 0.f;
-			if (errorIndexMax<0)
-				errorValueMax = 0.f;
-			m_errorCountTimes++;
-		//	printf("Error (x[%d] = %f, x[%d] = %f), resetting %d times\n", errorIndexMin,errorValueMin, errorIndexMax, errorValueMax, errorCountTimes++);
-			for (int i=0;i<n;i++)
+
+			btMatrixXu Bb1 = B * b1;
+			//		q = [ (-B*b1 - lo)'   (hi + B*b1)' ]'
+
+			btVectorXu qq;
+			qq.resize(n * 2);
+			for (int row = 0; row < n; row++)
 			{
-				x[i]=0.f;
+				qq[row] = -Bb1(row, 0) - lo[row];
+				qq[n + row] = Bb1(row, 0) + hi[row];
 			}
+
+			btVectorXu z1;
+
+			btMatrixXu y1;
+			y1.resize(n, 1);
+			btLemkeAlgorithm lemke(M, qq, m_debugLevel);
+			{
+				BT_PROFILE("lemke.solve");
+				lemke.setSystem(M, qq);
+				z1 = lemke.solve(m_maxLoops);
+			}
+			for (int row = 0; row < n; row++)
+			{
+				y1.setElem(row, 0, z1[2 * n + row] - z1[3 * n + row]);
+			}
+			btMatrixXu y1_b1(n, 1);
+			for (int i = 0; i < n; i++)
+			{
+				y1_b1.setElem(i, 0, y1(i, 0) - b1(i, 0));
+			}
+
+			btMatrixXu x1;
+
+			x1 = B * (y1_b1);
+
+			for (int row = 0; row < n; row++)
+			{
+				solution[row] = x1(row, 0);  //n];
+			}
+
+			int errorIndexMax = -1;
+			int errorIndexMin = -1;
+			float errorValueMax = -1e30;
+			float errorValueMin = 1e30;
+
+			for (int i = 0; i < n; i++)
+			{
+				x[i] = solution[i];
+				volatile btScalar check = x[i];
+				if (x[i] != check)
+				{
+					//printf("Lemke result is #NAN\n");
+					x.setZero();
+					return false;
+				}
+
+				//this is some hack/safety mechanism, to discard invalid solutions from the Lemke solver
+				//we need to figure out why it happens, and fix it, or detect it properly)
+				if (x[i] > m_maxValue)
+				{
+					if (x[i] > errorValueMax)
+					{
+						fail = true;
+						errorIndexMax = i;
+						errorValueMax = x[i];
+					}
+					////printf("x[i] = %f,",x[i]);
+				}
+				if (x[i] < -m_maxValue)
+				{
+					if (x[i] < errorValueMin)
+					{
+						errorIndexMin = i;
+						errorValueMin = x[i];
+						fail = true;
+						//printf("x[i] = %f,",x[i]);
+					}
+				}
+			}
+			if (fail)
+			{
+				int m_errorCountTimes = 0;
+				if (errorIndexMin < 0)
+					errorValueMin = 0.f;
+				if (errorIndexMax < 0)
+					errorValueMax = 0.f;
+				m_errorCountTimes++;
+				//	printf("Error (x[%d] = %f, x[%d] = %f), resetting %d times\n", errorIndexMin,errorValueMin, errorIndexMax, errorValueMax, errorCountTimes++);
+				for (int i = 0; i < n; i++)
+				{
+					x[i] = 0.f;
+				}
+			}
+			return !fail;
 		}
-		return !fail;
-	} else
-		
-	{
+		else
+
+		{
 			int dimension = A.rows();
-		if (0==dimension)
-			return true;
-		
-//		printf("================ solving using Lemke/Newton/Fixpoint\n");
+			if (0 == dimension)
+				return true;
 
-		btVectorXu q;
-		q.resize(dimension);
-		for (int row=0;row<dimension;row++)
-		{
-			q[row] = -b[row];
-		}
-		
-		btLemkeAlgorithm lemke(A,q,m_debugLevel);
-		
-		
-		lemke.setSystem(A,q);
-		
-		btVectorXu solution = lemke.solve(m_maxLoops);
-		
-		//check solution
-		
-		bool fail = false;
-		int errorIndexMax = -1;
-		int errorIndexMin = -1;
-		float errorValueMax = -1e30;
-		float errorValueMin = 1e30;
-		
-		for (int i=0;i<dimension;i++)
-		{
-			x[i] = solution[i+dimension];
-			volatile btScalar check = x[i];
-			if (x[i] != check)
+			//		printf("================ solving using Lemke/Newton/Fixpoint\n");
+
+			btVectorXu q;
+			q.resize(dimension);
+			for (int row = 0; row < dimension; row++)
 			{
-				x.setZero();
-				return false;
+				q[row] = -b[row];
 			}
-			
-			//this is some hack/safety mechanism, to discard invalid solutions from the Lemke solver 
-			//we need to figure out why it happens, and fix it, or detect it properly)
-			if (x[i]>m_maxValue)
+
+			btLemkeAlgorithm lemke(A, q, m_debugLevel);
+
+			lemke.setSystem(A, q);
+
+			btVectorXu solution = lemke.solve(m_maxLoops);
+
+			//check solution
+
+			bool fail = false;
+			int errorIndexMax = -1;
+			int errorIndexMin = -1;
+			float errorValueMax = -1e30;
+			float errorValueMin = 1e30;
+
+			for (int i = 0; i < dimension; i++)
 			{
-				if (x[i]> errorValueMax)
+				x[i] = solution[i + dimension];
+				volatile btScalar check = x[i];
+				if (x[i] != check)
 				{
-					fail = true;
-					errorIndexMax = i;
-					errorValueMax = x[i];
+					x.setZero();
+					return false;
 				}
-				////printf("x[i] = %f,",x[i]);
-			}
-			if (x[i]<-m_maxValue)
-			{
-				if (x[i]<errorValueMin)
+
+				//this is some hack/safety mechanism, to discard invalid solutions from the Lemke solver
+				//we need to figure out why it happens, and fix it, or detect it properly)
+				if (x[i] > m_maxValue)
 				{
-					errorIndexMin = i;
-					errorValueMin = x[i];
-					fail = true;
-					//printf("x[i] = %f,",x[i]);
+					if (x[i] > errorValueMax)
+					{
+						fail = true;
+						errorIndexMax = i;
+						errorValueMax = x[i];
+					}
+					////printf("x[i] = %f,",x[i]);
+				}
+				if (x[i] < -m_maxValue)
+				{
+					if (x[i] < errorValueMin)
+					{
+						errorIndexMin = i;
+						errorValueMin = x[i];
+						fail = true;
+						//printf("x[i] = %f,",x[i]);
+					}
 				}
 			}
-		}
-		if (fail)
-		{
-			static int errorCountTimes = 0;
-			if (errorIndexMin<0)
-				errorValueMin = 0.f;
-			if (errorIndexMax<0)
-				errorValueMax = 0.f;
-			printf("Error (x[%d] = %f, x[%d] = %f), resetting %d times\n", errorIndexMin,errorValueMin, errorIndexMax, errorValueMax, errorCountTimes++);
-			for (int i=0;i<dimension;i++)
+			if (fail)
 			{
-				x[i]=0.f;
+				static int errorCountTimes = 0;
+				if (errorIndexMin < 0)
+					errorValueMin = 0.f;
+				if (errorIndexMax < 0)
+					errorValueMax = 0.f;
+				printf("Error (x[%d] = %f, x[%d] = %f), resetting %d times\n", errorIndexMin, errorValueMin, errorIndexMax, errorValueMax, errorCountTimes++);
+				for (int i = 0; i < dimension; i++)
+				{
+					x[i] = 0.f;
+				}
 			}
+
+			return !fail;
 		}
-
-
-		return !fail;
+		return true;
 	}
-	return true;
-
-	}
-
 };
 
-#endif //BT_LEMKE_SOLVER_H
+#endif  //BT_LEMKE_SOLVER_H
diff --git a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btMLCPSolver.cpp b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btMLCPSolver.cpp
index 8f54c526269..5d864f27578 100644
--- a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btMLCPSolver.cpp
+++ b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btMLCPSolver.cpp
@@ -19,10 +19,9 @@ subject to the following restrictions:
 #include "LinearMath/btQuickprof.h"
 #include "btSolveProjectedGaussSeidel.h"
 
-
-btMLCPSolver::btMLCPSolver(	 btMLCPSolverInterface* solver)
-:m_solver(solver),
-m_fallback(0)
+btMLCPSolver::btMLCPSolver(btMLCPSolverInterface* solver)
+	: m_solver(solver),
+	  m_fallback(0)
 {
 }
 
@@ -33,67 +32,65 @@ btMLCPSolver::~btMLCPSolver()
 bool gUseMatrixMultiply = false;
 bool interleaveContactAndFriction = false;
 
-btScalar btMLCPSolver::solveGroupCacheFriendlySetup(btCollisionObject** bodies, int numBodiesUnUsed, btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer)
+btScalar btMLCPSolver::solveGroupCacheFriendlySetup(btCollisionObject** bodies, int numBodiesUnUsed, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer)
 {
-	btSequentialImpulseConstraintSolver::solveGroupCacheFriendlySetup( bodies, numBodiesUnUsed, manifoldPtr, numManifolds,constraints,numConstraints,infoGlobal,debugDrawer);
+	btSequentialImpulseConstraintSolver::solveGroupCacheFriendlySetup(bodies, numBodiesUnUsed, manifoldPtr, numManifolds, constraints, numConstraints, infoGlobal, debugDrawer);
 
 	{
 		BT_PROFILE("gather constraint data");
 
-		int numFrictionPerContact = m_tmpSolverContactConstraintPool.size()==m_tmpSolverContactFrictionConstraintPool.size()? 1 : 2;
+		int numFrictionPerContact = m_tmpSolverContactConstraintPool.size() == m_tmpSolverContactFrictionConstraintPool.size() ? 1 : 2;
 
-
-	//	int numBodies = m_tmpSolverBodyPool.size();
+		//	int numBodies = m_tmpSolverBodyPool.size();
 		m_allConstraintPtrArray.resize(0);
-		m_limitDependencies.resize(m_tmpSolverNonContactConstraintPool.size()+m_tmpSolverContactConstraintPool.size()+m_tmpSolverContactFrictionConstraintPool.size());
-		btAssert(m_limitDependencies.size() == m_tmpSolverNonContactConstraintPool.size()+m_tmpSolverContactConstraintPool.size()+m_tmpSolverContactFrictionConstraintPool.size());
-	//	printf("m_limitDependencies.size() = %d\n",m_limitDependencies.size());
+		m_limitDependencies.resize(m_tmpSolverNonContactConstraintPool.size() + m_tmpSolverContactConstraintPool.size() + m_tmpSolverContactFrictionConstraintPool.size());
+		btAssert(m_limitDependencies.size() == m_tmpSolverNonContactConstraintPool.size() + m_tmpSolverContactConstraintPool.size() + m_tmpSolverContactFrictionConstraintPool.size());
+		//	printf("m_limitDependencies.size() = %d\n",m_limitDependencies.size());
 
 		int dindex = 0;
-		for (int i=0;i<m_tmpSolverNonContactConstraintPool.size();i++)
+		for (int i = 0; i < m_tmpSolverNonContactConstraintPool.size(); i++)
 		{
 			m_allConstraintPtrArray.push_back(&m_tmpSolverNonContactConstraintPool[i]);
 			m_limitDependencies[dindex++] = -1;
 		}
- 
+
 		///The btSequentialImpulseConstraintSolver moves all friction constraints at the very end, we can also interleave them instead
-		
-		int firstContactConstraintOffset=dindex;
+
+		int firstContactConstraintOffset = dindex;
 
 		if (interleaveContactAndFriction)
 		{
-			for (int i=0;i<m_tmpSolverContactConstraintPool.size();i++)
+			for (int i = 0; i < m_tmpSolverContactConstraintPool.size(); i++)
 			{
 				m_allConstraintPtrArray.push_back(&m_tmpSolverContactConstraintPool[i]);
 				m_limitDependencies[dindex++] = -1;
-				m_allConstraintPtrArray.push_back(&m_tmpSolverContactFrictionConstraintPool[i*numFrictionPerContact]);
-				int findex = (m_tmpSolverContactFrictionConstraintPool[i*numFrictionPerContact].m_frictionIndex*(1+numFrictionPerContact));
-				m_limitDependencies[dindex++] = findex +firstContactConstraintOffset;
-				if (numFrictionPerContact==2)
+				m_allConstraintPtrArray.push_back(&m_tmpSolverContactFrictionConstraintPool[i * numFrictionPerContact]);
+				int findex = (m_tmpSolverContactFrictionConstraintPool[i * numFrictionPerContact].m_frictionIndex * (1 + numFrictionPerContact));
+				m_limitDependencies[dindex++] = findex + firstContactConstraintOffset;
+				if (numFrictionPerContact == 2)
 				{
-					m_allConstraintPtrArray.push_back(&m_tmpSolverContactFrictionConstraintPool[i*numFrictionPerContact+1]);
-					m_limitDependencies[dindex++] = findex+firstContactConstraintOffset;
+					m_allConstraintPtrArray.push_back(&m_tmpSolverContactFrictionConstraintPool[i * numFrictionPerContact + 1]);
+					m_limitDependencies[dindex++] = findex + firstContactConstraintOffset;
 				}
 			}
-		} else
+		}
+		else
 		{
-			for (int i=0;i<m_tmpSolverContactConstraintPool.size();i++)
+			for (int i = 0; i < m_tmpSolverContactConstraintPool.size(); i++)
 			{
 				m_allConstraintPtrArray.push_back(&m_tmpSolverContactConstraintPool[i]);
 				m_limitDependencies[dindex++] = -1;
 			}
-			for (int i=0;i<m_tmpSolverContactFrictionConstraintPool.size();i++)
+			for (int i = 0; i < m_tmpSolverContactFrictionConstraintPool.size(); i++)
 			{
 				m_allConstraintPtrArray.push_back(&m_tmpSolverContactFrictionConstraintPool[i]);
-				m_limitDependencies[dindex++] = m_tmpSolverContactFrictionConstraintPool[i].m_frictionIndex+firstContactConstraintOffset;
+				m_limitDependencies[dindex++] = m_tmpSolverContactFrictionConstraintPool[i].m_frictionIndex + firstContactConstraintOffset;
 			}
-			
 		}
 
-
 		if (!m_allConstraintPtrArray.size())
 		{
-			m_A.resize(0,0);
+			m_A.resize(0, 0);
 			m_b.resize(0);
 			m_x.resize(0);
 			m_lo.resize(0);
@@ -102,7 +99,6 @@ btScalar btMLCPSolver::solveGroupCacheFriendlySetup(btCollisionObject** bodies,
 		}
 	}
 
-	
 	if (gUseMatrixMultiply)
 	{
 		BT_PROFILE("createMLCP");
@@ -121,7 +117,7 @@ bool btMLCPSolver::solveMLCP(const btContactSolverInfo& infoGlobal)
 {
 	bool result = true;
 
-	if (m_A.rows()==0)
+	if (m_A.rows() == 0)
 		return true;
 
 	//if using split impulse, we solve 2 separate (M)LCPs
@@ -129,28 +125,26 @@ bool btMLCPSolver::solveMLCP(const btContactSolverInfo& infoGlobal)
 	{
 		btMatrixXu Acopy = m_A;
 		btAlignedObjectArray<int> limitDependenciesCopy = m_limitDependencies;
-//		printf("solve first LCP\n");
-		result = m_solver->solveMLCP(m_A, m_b, m_x, m_lo,m_hi, m_limitDependencies,infoGlobal.m_numIterations );
+		//		printf("solve first LCP\n");
+		result = m_solver->solveMLCP(m_A, m_b, m_x, m_lo, m_hi, m_limitDependencies, infoGlobal.m_numIterations);
 		if (result)
-			result = m_solver->solveMLCP(Acopy, m_bSplit, m_xSplit, m_lo,m_hi, limitDependenciesCopy,infoGlobal.m_numIterations );
-
-	} else
+			result = m_solver->solveMLCP(Acopy, m_bSplit, m_xSplit, m_lo, m_hi, limitDependenciesCopy, infoGlobal.m_numIterations);
+	}
+	else
 	{
-		result = m_solver->solveMLCP(m_A, m_b, m_x, m_lo,m_hi, m_limitDependencies,infoGlobal.m_numIterations );
+		result = m_solver->solveMLCP(m_A, m_b, m_x, m_lo, m_hi, m_limitDependencies, infoGlobal.m_numIterations);
 	}
 	return result;
 }
 
 struct btJointNode
 {
-	int jointIndex;     // pointer to enclosing dxJoint object
-	int otherBodyIndex;       // *other* body this joint is connected to
-	int nextJointNodeIndex;//-1 for null
+	int jointIndex;          // pointer to enclosing dxJoint object
+	int otherBodyIndex;      // *other* body this joint is connected to
+	int nextJointNodeIndex;  //-1 for null
 	int constraintRowIndex;
 };
 
-
-
 void btMLCPSolver::createMLCPFast(const btContactSolverInfo& infoGlobal)
 {
 	int numContactRows = interleaveContactAndFriction ? 3 : 1;
@@ -163,36 +157,36 @@ void btMLCPSolver::createMLCPFast(const btContactSolverInfo& infoGlobal)
 		m_bSplit.resize(numConstraintRows);
 		m_b.setZero();
 		m_bSplit.setZero();
-		for (int i=0;i<numConstraintRows ;i++)
+		for (int i = 0; i < numConstraintRows; i++)
 		{
 			btScalar jacDiag = m_allConstraintPtrArray[i]->m_jacDiagABInv;
 			if (!btFuzzyZero(jacDiag))
 			{
 				btScalar rhs = m_allConstraintPtrArray[i]->m_rhs;
 				btScalar rhsPenetration = m_allConstraintPtrArray[i]->m_rhsPenetration;
-				m_b[i]=rhs/jacDiag;
-				m_bSplit[i] = rhsPenetration/jacDiag;
+				m_b[i] = rhs / jacDiag;
+				m_bSplit[i] = rhsPenetration / jacDiag;
 			}
-
 		}
 	}
 
-//	btScalar* w = 0;
-//	int nub = 0;
+	//	btScalar* w = 0;
+	//	int nub = 0;
 
 	m_lo.resize(numConstraintRows);
 	m_hi.resize(numConstraintRows);
- 
+
 	{
 		BT_PROFILE("init lo/ho");
 
-		for (int i=0;i<numConstraintRows;i++)
+		for (int i = 0; i < numConstraintRows; i++)
 		{
-			if (0)//m_limitDependencies[i]>=0)
+			if (0)  //m_limitDependencies[i]>=0)
 			{
 				m_lo[i] = -BT_INFINITY;
 				m_hi[i] = BT_INFINITY;
-			} else
+			}
+			else
 			{
 				m_lo[i] = m_allConstraintPtrArray[i]->m_lowerLimit;
 				m_hi[i] = m_allConstraintPtrArray[i]->m_upperLimit;
@@ -201,48 +195,48 @@ void btMLCPSolver::createMLCPFast(const btContactSolverInfo& infoGlobal)
 	}
 
 	//
-	int m=m_allConstraintPtrArray.size();
+	int m = m_allConstraintPtrArray.size();
 
 	int numBodies = m_tmpSolverBodyPool.size();
 	btAlignedObjectArray<int> bodyJointNodeArray;
 	{
 		BT_PROFILE("bodyJointNodeArray.resize");
-		bodyJointNodeArray.resize(numBodies,-1);
+		bodyJointNodeArray.resize(numBodies, -1);
 	}
 	btAlignedObjectArray<btJointNode> jointNodeArray;
 	{
 		BT_PROFILE("jointNodeArray.reserve");
-		jointNodeArray.reserve(2*m_allConstraintPtrArray.size());
+		jointNodeArray.reserve(2 * m_allConstraintPtrArray.size());
 	}
 
-    btMatrixXu& J3 = m_scratchJ3;
+	btMatrixXu& J3 = m_scratchJ3;
 	{
 		BT_PROFILE("J3.resize");
-		J3.resize(2*m,8);
+		J3.resize(2 * m, 8);
 	}
-    btMatrixXu& JinvM3 = m_scratchJInvM3;
+	btMatrixXu& JinvM3 = m_scratchJInvM3;
 	{
 		BT_PROFILE("JinvM3.resize/setZero");
 
-		JinvM3.resize(2*m,8);
+		JinvM3.resize(2 * m, 8);
 		JinvM3.setZero();
 		J3.setZero();
 	}
-	int cur=0;
+	int cur = 0;
 	int rowOffset = 0;
-    btAlignedObjectArray<int>& ofs = m_scratchOfs;
+	btAlignedObjectArray<int>& ofs = m_scratchOfs;
 	{
 		BT_PROFILE("ofs resize");
 		ofs.resize(0);
 		ofs.resizeNoInitialize(m_allConstraintPtrArray.size());
-	}				
+	}
 	{
 		BT_PROFILE("Compute J and JinvM");
-		int c=0;
+		int c = 0;
 
 		int numRows = 0;
 
-		for (int i=0;i<m_allConstraintPtrArray.size();i+=numRows,c++)
+		for (int i = 0; i < m_allConstraintPtrArray.size(); i += numRows, c++)
 		{
 			ofs[c] = rowOffset;
 			int sbA = m_allConstraintPtrArray[i]->m_solverBodyIdA;
@@ -250,14 +244,14 @@ void btMLCPSolver::createMLCPFast(const btContactSolverInfo& infoGlobal)
 			btRigidBody* orgBodyA = m_tmpSolverBodyPool[sbA].m_originalBody;
 			btRigidBody* orgBodyB = m_tmpSolverBodyPool[sbB].m_originalBody;
 
-			numRows = i<m_tmpSolverNonContactConstraintPool.size() ? m_tmpConstraintSizesPool[c].m_numConstraintRows : numContactRows ;
+			numRows = i < m_tmpSolverNonContactConstraintPool.size() ? m_tmpConstraintSizesPool[c].m_numConstraintRows : numContactRows;
 			if (orgBodyA)
 			{
 				{
-					int slotA=-1;
+					int slotA = -1;
 					//find free jointNode slot for sbA
-					slotA =jointNodeArray.size();
-					jointNodeArray.expand();//NonInitializing();
+					slotA = jointNodeArray.size();
+					jointNodeArray.expand();  //NonInitializing();
 					int prevSlot = bodyJointNodeArray[sbA];
 					bodyJointNodeArray[sbA] = slotA;
 					jointNodeArray[slotA].nextJointNodeIndex = prevSlot;
@@ -265,35 +259,35 @@ void btMLCPSolver::createMLCPFast(const btContactSolverInfo& infoGlobal)
 					jointNodeArray[slotA].constraintRowIndex = i;
 					jointNodeArray[slotA].otherBodyIndex = orgBodyB ? sbB : -1;
 				}
-				for (int row=0;row<numRows;row++,cur++)
+				for (int row = 0; row < numRows; row++, cur++)
 				{
-					btVector3 normalInvMass =				m_allConstraintPtrArray[i+row]->m_contactNormal1 *		orgBodyA->getInvMass();
-					btVector3 relPosCrossNormalInvInertia = m_allConstraintPtrArray[i+row]->m_relpos1CrossNormal *	orgBodyA->getInvInertiaTensorWorld();
+					btVector3 normalInvMass = m_allConstraintPtrArray[i + row]->m_contactNormal1 * orgBodyA->getInvMass();
+					btVector3 relPosCrossNormalInvInertia = m_allConstraintPtrArray[i + row]->m_relpos1CrossNormal * orgBodyA->getInvInertiaTensorWorld();
 
-					for (int r=0;r<3;r++)
+					for (int r = 0; r < 3; r++)
 					{
-						J3.setElem(cur,r,m_allConstraintPtrArray[i+row]->m_contactNormal1[r]);
-						J3.setElem(cur,r+4,m_allConstraintPtrArray[i+row]->m_relpos1CrossNormal[r]);
-						JinvM3.setElem(cur,r,normalInvMass[r]);
-						JinvM3.setElem(cur,r+4,relPosCrossNormalInvInertia[r]);
+						J3.setElem(cur, r, m_allConstraintPtrArray[i + row]->m_contactNormal1[r]);
+						J3.setElem(cur, r + 4, m_allConstraintPtrArray[i + row]->m_relpos1CrossNormal[r]);
+						JinvM3.setElem(cur, r, normalInvMass[r]);
+						JinvM3.setElem(cur, r + 4, relPosCrossNormalInvInertia[r]);
 					}
-					J3.setElem(cur,3,0);
-					JinvM3.setElem(cur,3,0);
-					J3.setElem(cur,7,0);
-					JinvM3.setElem(cur,7,0);
+					J3.setElem(cur, 3, 0);
+					JinvM3.setElem(cur, 3, 0);
+					J3.setElem(cur, 7, 0);
+					JinvM3.setElem(cur, 7, 0);
 				}
-			} else
+			}
+			else
 			{
 				cur += numRows;
 			}
 			if (orgBodyB)
 			{
-
 				{
-					int slotB=-1;
+					int slotB = -1;
 					//find free jointNode slot for sbA
-					slotB =jointNodeArray.size();
-					jointNodeArray.expand();//NonInitializing();
+					slotB = jointNodeArray.size();
+					jointNodeArray.expand();  //NonInitializing();
 					int prevSlot = bodyJointNodeArray[sbB];
 					bodyJointNodeArray[sbB] = slotB;
 					jointNodeArray[slotB].nextJointNodeIndex = prevSlot;
@@ -302,78 +296,74 @@ void btMLCPSolver::createMLCPFast(const btContactSolverInfo& infoGlobal)
 					jointNodeArray[slotB].constraintRowIndex = i;
 				}
 
-				for (int row=0;row<numRows;row++,cur++)
+				for (int row = 0; row < numRows; row++, cur++)
 				{
-					btVector3 normalInvMassB = m_allConstraintPtrArray[i+row]->m_contactNormal2*orgBodyB->getInvMass();
-					btVector3 relPosInvInertiaB = m_allConstraintPtrArray[i+row]->m_relpos2CrossNormal * orgBodyB->getInvInertiaTensorWorld();
+					btVector3 normalInvMassB = m_allConstraintPtrArray[i + row]->m_contactNormal2 * orgBodyB->getInvMass();
+					btVector3 relPosInvInertiaB = m_allConstraintPtrArray[i + row]->m_relpos2CrossNormal * orgBodyB->getInvInertiaTensorWorld();
 
-					for (int r=0;r<3;r++)
+					for (int r = 0; r < 3; r++)
 					{
-						J3.setElem(cur,r,m_allConstraintPtrArray[i+row]->m_contactNormal2[r]);
-						J3.setElem(cur,r+4,m_allConstraintPtrArray[i+row]->m_relpos2CrossNormal[r]);
-						JinvM3.setElem(cur,r,normalInvMassB[r]);
-						JinvM3.setElem(cur,r+4,relPosInvInertiaB[r]);
+						J3.setElem(cur, r, m_allConstraintPtrArray[i + row]->m_contactNormal2[r]);
+						J3.setElem(cur, r + 4, m_allConstraintPtrArray[i + row]->m_relpos2CrossNormal[r]);
+						JinvM3.setElem(cur, r, normalInvMassB[r]);
+						JinvM3.setElem(cur, r + 4, relPosInvInertiaB[r]);
 					}
-					J3.setElem(cur,3,0);
-					JinvM3.setElem(cur,3,0);
-					J3.setElem(cur,7,0);
-					JinvM3.setElem(cur,7,0);
+					J3.setElem(cur, 3, 0);
+					JinvM3.setElem(cur, 3, 0);
+					J3.setElem(cur, 7, 0);
+					JinvM3.setElem(cur, 7, 0);
 				}
 			}
 			else
 			{
 				cur += numRows;
 			}
-			rowOffset+=numRows;
-
+			rowOffset += numRows;
 		}
-		
 	}
 
-
 	//compute JinvM = J*invM.
 	const btScalar* JinvM = JinvM3.getBufferPointer();
 
 	const btScalar* Jptr = J3.getBufferPointer();
 	{
 		BT_PROFILE("m_A.resize");
-		m_A.resize(n,n);
+		m_A.resize(n, n);
 	}
-	
+
 	{
 		BT_PROFILE("m_A.setZero");
 		m_A.setZero();
 	}
-	int c=0;
+	int c = 0;
 	{
 		int numRows = 0;
 		BT_PROFILE("Compute A");
-		for (int i=0;i<m_allConstraintPtrArray.size();i+= numRows,c++)
+		for (int i = 0; i < m_allConstraintPtrArray.size(); i += numRows, c++)
 		{
 			int row__ = ofs[c];
 			int sbA = m_allConstraintPtrArray[i]->m_solverBodyIdA;
 			int sbB = m_allConstraintPtrArray[i]->m_solverBodyIdB;
-		//	btRigidBody* orgBodyA = m_tmpSolverBodyPool[sbA].m_originalBody;
-		//	btRigidBody* orgBodyB = m_tmpSolverBodyPool[sbB].m_originalBody;
+			//	btRigidBody* orgBodyA = m_tmpSolverBodyPool[sbA].m_originalBody;
+			//	btRigidBody* orgBodyB = m_tmpSolverBodyPool[sbB].m_originalBody;
 
-			numRows = i<m_tmpSolverNonContactConstraintPool.size() ? m_tmpConstraintSizesPool[c].m_numConstraintRows : numContactRows ;
-					
-			const btScalar *JinvMrow = JinvM + 2*8*(size_t)row__;
+			numRows = i < m_tmpSolverNonContactConstraintPool.size() ? m_tmpConstraintSizesPool[c].m_numConstraintRows : numContactRows;
+
+			const btScalar* JinvMrow = JinvM + 2 * 8 * (size_t)row__;
 
 			{
 				int startJointNodeA = bodyJointNodeArray[sbA];
-				while (startJointNodeA>=0)
+				while (startJointNodeA >= 0)
 				{
 					int j0 = jointNodeArray[startJointNodeA].jointIndex;
 					int cr0 = jointNodeArray[startJointNodeA].constraintRowIndex;
-					if (j0<c)
+					if (j0 < c)
 					{
-								 
 						int numRowsOther = cr0 < m_tmpSolverNonContactConstraintPool.size() ? m_tmpConstraintSizesPool[j0].m_numConstraintRows : numContactRows;
-						size_t ofsother = (m_allConstraintPtrArray[cr0]->m_solverBodyIdB == sbA) ? 8*numRowsOther  : 0;
+						size_t ofsother = (m_allConstraintPtrArray[cr0]->m_solverBodyIdB == sbA) ? 8 * numRowsOther : 0;
 						//printf("%d joint i %d and j0: %d: ",count++,i,j0);
-						m_A.multiplyAdd2_p8r ( JinvMrow, 
-						Jptr + 2*8*(size_t)ofs[j0] + ofsother, numRows, numRowsOther,  row__,ofs[j0]);
+						m_A.multiplyAdd2_p8r(JinvMrow,
+											 Jptr + 2 * 8 * (size_t)ofs[j0] + ofsother, numRows, numRowsOther, row__, ofs[j0]);
 					}
 					startJointNodeA = jointNodeArray[startJointNodeA].nextJointNodeIndex;
 				}
@@ -381,17 +371,17 @@ void btMLCPSolver::createMLCPFast(const btContactSolverInfo& infoGlobal)
 
 			{
 				int startJointNodeB = bodyJointNodeArray[sbB];
-				while (startJointNodeB>=0)
+				while (startJointNodeB >= 0)
 				{
 					int j1 = jointNodeArray[startJointNodeB].jointIndex;
 					int cj1 = jointNodeArray[startJointNodeB].constraintRowIndex;
 
-					if (j1<c)
+					if (j1 < c)
 					{
-						int numRowsOther =  cj1 < m_tmpSolverNonContactConstraintPool.size() ? m_tmpConstraintSizesPool[j1].m_numConstraintRows : numContactRows;
-						size_t ofsother = (m_allConstraintPtrArray[cj1]->m_solverBodyIdB == sbB) ? 8*numRowsOther  : 0;
-						m_A.multiplyAdd2_p8r ( JinvMrow + 8*(size_t)numRows, 
-						Jptr + 2*8*(size_t)ofs[j1] + ofsother, numRows, numRowsOther, row__,ofs[j1]);
+						int numRowsOther = cj1 < m_tmpSolverNonContactConstraintPool.size() ? m_tmpConstraintSizesPool[j1].m_numConstraintRows : numContactRows;
+						size_t ofsother = (m_allConstraintPtrArray[cj1]->m_solverBodyIdB == sbB) ? 8 * numRowsOther : 0;
+						m_A.multiplyAdd2_p8r(JinvMrow + 8 * (size_t)numRows,
+											 Jptr + 2 * 8 * (size_t)ofs[j1] + ofsother, numRows, numRowsOther, row__, ofs[j1]);
 					}
 					startJointNodeB = jointNodeArray[startJointNodeB].nextJointNodeIndex;
 				}
@@ -402,27 +392,25 @@ void btMLCPSolver::createMLCPFast(const btContactSolverInfo& infoGlobal)
 			BT_PROFILE("compute diagonal");
 			// compute diagonal blocks of m_A
 
-			int  row__ = 0;
+			int row__ = 0;
 			int numJointRows = m_allConstraintPtrArray.size();
 
-			int jj=0;
-			for (;row__<numJointRows;)
+			int jj = 0;
+			for (; row__ < numJointRows;)
 			{
-
 				//int sbA = m_allConstraintPtrArray[row__]->m_solverBodyIdA;
 				int sbB = m_allConstraintPtrArray[row__]->m_solverBodyIdB;
-			//	btRigidBody* orgBodyA = m_tmpSolverBodyPool[sbA].m_originalBody;
+				//	btRigidBody* orgBodyA = m_tmpSolverBodyPool[sbA].m_originalBody;
 				btRigidBody* orgBodyB = m_tmpSolverBodyPool[sbB].m_originalBody;
 
+				const unsigned int infom = row__ < m_tmpSolverNonContactConstraintPool.size() ? m_tmpConstraintSizesPool[jj].m_numConstraintRows : numContactRows;
 
-				const unsigned int infom =  row__ < m_tmpSolverNonContactConstraintPool.size() ? m_tmpConstraintSizesPool[jj].m_numConstraintRows : numContactRows;
-				
-				const btScalar *JinvMrow = JinvM + 2*8*(size_t)row__;
-				const btScalar *Jrow = Jptr + 2*8*(size_t)row__;
-				m_A.multiply2_p8r (JinvMrow, Jrow, infom, infom, row__,row__);
-				if (orgBodyB) 
+				const btScalar* JinvMrow = JinvM + 2 * 8 * (size_t)row__;
+				const btScalar* Jrow = Jptr + 2 * 8 * (size_t)row__;
+				m_A.multiply2_p8r(JinvMrow, Jrow, infom, infom, row__, row__);
+				if (orgBodyB)
 				{
-					m_A.multiplyAdd2_p8r (JinvMrow + 8*(size_t)infom, Jrow + 8*(size_t)infom, infom, infom,  row__,row__);
+					m_A.multiplyAdd2_p8r(JinvMrow + 8 * (size_t)infom, Jrow + 8 * (size_t)infom, infom, infom, row__, row__);
 				}
 				row__ += infom;
 				jj++;
@@ -433,12 +421,12 @@ void btMLCPSolver::createMLCPFast(const btContactSolverInfo& infoGlobal)
 	if (1)
 	{
 		// add cfm to the diagonal of m_A
-		for ( int i=0; i<m_A.rows(); ++i) 
+		for (int i = 0; i < m_A.rows(); ++i)
 		{
-			m_A.setElem(i,i,m_A(i,i)+ infoGlobal.m_globalCfm/ infoGlobal.m_timeStep);
+			m_A.setElem(i, i, m_A(i, i) + infoGlobal.m_globalCfm / infoGlobal.m_timeStep);
 		}
 	}
-				   
+
 	///fill the upper triangle of the matrix, to make it symmetric
 	{
 		BT_PROFILE("fill the upper triangle ");
@@ -450,21 +438,21 @@ void btMLCPSolver::createMLCPFast(const btContactSolverInfo& infoGlobal)
 		m_x.resize(numConstraintRows);
 		m_xSplit.resize(numConstraintRows);
 
-		if (infoGlobal.m_solverMode&SOLVER_USE_WARMSTARTING)
+		if (infoGlobal.m_solverMode & SOLVER_USE_WARMSTARTING)
 		{
-			for (int i=0;i<m_allConstraintPtrArray.size();i++)
+			for (int i = 0; i < m_allConstraintPtrArray.size(); i++)
 			{
 				const btSolverConstraint& c = *m_allConstraintPtrArray[i];
-				m_x[i]=c.m_appliedImpulse;
+				m_x[i] = c.m_appliedImpulse;
 				m_xSplit[i] = c.m_appliedPushImpulse;
 			}
-		} else
+		}
+		else
 		{
 			m_x.setZero();
 			m_xSplit.setZero();
 		}
 	}
-
 }
 
 void btMLCPSolver::createMLCP(const btContactSolverInfo& infoGlobal)
@@ -475,120 +463,116 @@ void btMLCPSolver::createMLCP(const btContactSolverInfo& infoGlobal)
 	m_b.resize(numConstraintRows);
 	if (infoGlobal.m_splitImpulse)
 		m_bSplit.resize(numConstraintRows);
- 
+
 	m_bSplit.setZero();
 	m_b.setZero();
 
-	for (int i=0;i<numConstraintRows ;i++)
+	for (int i = 0; i < numConstraintRows; i++)
 	{
 		if (m_allConstraintPtrArray[i]->m_jacDiagABInv)
 		{
-			m_b[i]=m_allConstraintPtrArray[i]->m_rhs/m_allConstraintPtrArray[i]->m_jacDiagABInv;
+			m_b[i] = m_allConstraintPtrArray[i]->m_rhs / m_allConstraintPtrArray[i]->m_jacDiagABInv;
 			if (infoGlobal.m_splitImpulse)
-				m_bSplit[i] = m_allConstraintPtrArray[i]->m_rhsPenetration/m_allConstraintPtrArray[i]->m_jacDiagABInv;
+				m_bSplit[i] = m_allConstraintPtrArray[i]->m_rhsPenetration / m_allConstraintPtrArray[i]->m_jacDiagABInv;
 		}
 	}
- 
-    btMatrixXu& Minv = m_scratchMInv;
-	Minv.resize(6*numBodies,6*numBodies);
+
+	btMatrixXu& Minv = m_scratchMInv;
+	Minv.resize(6 * numBodies, 6 * numBodies);
 	Minv.setZero();
-	for (int i=0;i<numBodies;i++)
+	for (int i = 0; i < numBodies; i++)
 	{
 		const btSolverBody& rb = m_tmpSolverBodyPool[i];
 		const btVector3& invMass = rb.m_invMass;
-		setElem(Minv,i*6+0,i*6+0,invMass[0]);
-		setElem(Minv,i*6+1,i*6+1,invMass[1]);
-		setElem(Minv,i*6+2,i*6+2,invMass[2]);
+		setElem(Minv, i * 6 + 0, i * 6 + 0, invMass[0]);
+		setElem(Minv, i * 6 + 1, i * 6 + 1, invMass[1]);
+		setElem(Minv, i * 6 + 2, i * 6 + 2, invMass[2]);
 		btRigidBody* orgBody = m_tmpSolverBodyPool[i].m_originalBody;
- 
-		for (int r=0;r<3;r++)
-			for (int c=0;c<3;c++)
-				setElem(Minv,i*6+3+r,i*6+3+c,orgBody? orgBody->getInvInertiaTensorWorld()[r][c] : 0);
+
+		for (int r = 0; r < 3; r++)
+			for (int c = 0; c < 3; c++)
+				setElem(Minv, i * 6 + 3 + r, i * 6 + 3 + c, orgBody ? orgBody->getInvInertiaTensorWorld()[r][c] : 0);
 	}
- 
-    btMatrixXu& J = m_scratchJ;
-	J.resize(numConstraintRows,6*numBodies);
+
+	btMatrixXu& J = m_scratchJ;
+	J.resize(numConstraintRows, 6 * numBodies);
 	J.setZero();
- 
+
 	m_lo.resize(numConstraintRows);
 	m_hi.resize(numConstraintRows);
- 
-	for (int i=0;i<numConstraintRows;i++)
-	{
 
+	for (int i = 0; i < numConstraintRows; i++)
+	{
 		m_lo[i] = m_allConstraintPtrArray[i]->m_lowerLimit;
 		m_hi[i] = m_allConstraintPtrArray[i]->m_upperLimit;
- 
+
 		int bodyIndex0 = m_allConstraintPtrArray[i]->m_solverBodyIdA;
 		int bodyIndex1 = m_allConstraintPtrArray[i]->m_solverBodyIdB;
 		if (m_tmpSolverBodyPool[bodyIndex0].m_originalBody)
 		{
-			setElem(J,i,6*bodyIndex0+0,m_allConstraintPtrArray[i]->m_contactNormal1[0]);
-			setElem(J,i,6*bodyIndex0+1,m_allConstraintPtrArray[i]->m_contactNormal1[1]);
-			setElem(J,i,6*bodyIndex0+2,m_allConstraintPtrArray[i]->m_contactNormal1[2]);
-			setElem(J,i,6*bodyIndex0+3,m_allConstraintPtrArray[i]->m_relpos1CrossNormal[0]);
-			setElem(J,i,6*bodyIndex0+4,m_allConstraintPtrArray[i]->m_relpos1CrossNormal[1]);
-			setElem(J,i,6*bodyIndex0+5,m_allConstraintPtrArray[i]->m_relpos1CrossNormal[2]);
+			setElem(J, i, 6 * bodyIndex0 + 0, m_allConstraintPtrArray[i]->m_contactNormal1[0]);
+			setElem(J, i, 6 * bodyIndex0 + 1, m_allConstraintPtrArray[i]->m_contactNormal1[1]);
+			setElem(J, i, 6 * bodyIndex0 + 2, m_allConstraintPtrArray[i]->m_contactNormal1[2]);
+			setElem(J, i, 6 * bodyIndex0 + 3, m_allConstraintPtrArray[i]->m_relpos1CrossNormal[0]);
+			setElem(J, i, 6 * bodyIndex0 + 4, m_allConstraintPtrArray[i]->m_relpos1CrossNormal[1]);
+			setElem(J, i, 6 * bodyIndex0 + 5, m_allConstraintPtrArray[i]->m_relpos1CrossNormal[2]);
 		}
 		if (m_tmpSolverBodyPool[bodyIndex1].m_originalBody)
 		{
-			setElem(J,i,6*bodyIndex1+0,m_allConstraintPtrArray[i]->m_contactNormal2[0]);
-			setElem(J,i,6*bodyIndex1+1,m_allConstraintPtrArray[i]->m_contactNormal2[1]);
-			setElem(J,i,6*bodyIndex1+2,m_allConstraintPtrArray[i]->m_contactNormal2[2]);
-			setElem(J,i,6*bodyIndex1+3,m_allConstraintPtrArray[i]->m_relpos2CrossNormal[0]);
-			setElem(J,i,6*bodyIndex1+4,m_allConstraintPtrArray[i]->m_relpos2CrossNormal[1]);
-			setElem(J,i,6*bodyIndex1+5,m_allConstraintPtrArray[i]->m_relpos2CrossNormal[2]);
+			setElem(J, i, 6 * bodyIndex1 + 0, m_allConstraintPtrArray[i]->m_contactNormal2[0]);
+			setElem(J, i, 6 * bodyIndex1 + 1, m_allConstraintPtrArray[i]->m_contactNormal2[1]);
+			setElem(J, i, 6 * bodyIndex1 + 2, m_allConstraintPtrArray[i]->m_contactNormal2[2]);
+			setElem(J, i, 6 * bodyIndex1 + 3, m_allConstraintPtrArray[i]->m_relpos2CrossNormal[0]);
+			setElem(J, i, 6 * bodyIndex1 + 4, m_allConstraintPtrArray[i]->m_relpos2CrossNormal[1]);
+			setElem(J, i, 6 * bodyIndex1 + 5, m_allConstraintPtrArray[i]->m_relpos2CrossNormal[2]);
 		}
 	}
- 
-    btMatrixXu& J_transpose = m_scratchJTranspose;
-	J_transpose= J.transpose();
 
-    btMatrixXu& tmp = m_scratchTmp;
+	btMatrixXu& J_transpose = m_scratchJTranspose;
+	J_transpose = J.transpose();
+
+	btMatrixXu& tmp = m_scratchTmp;
 
 	{
 		{
 			BT_PROFILE("J*Minv");
-			tmp = J*Minv;
-
+			tmp = J * Minv;
 		}
 		{
 			BT_PROFILE("J*tmp");
-			m_A = tmp*J_transpose;
+			m_A = tmp * J_transpose;
 		}
 	}
 
 	if (1)
 	{
 		// add cfm to the diagonal of m_A
-		for ( int i=0; i<m_A.rows(); ++i) 
+		for (int i = 0; i < m_A.rows(); ++i)
 		{
-			m_A.setElem(i,i,m_A(i,i)+ infoGlobal.m_globalCfm / infoGlobal.m_timeStep);
+			m_A.setElem(i, i, m_A(i, i) + infoGlobal.m_globalCfm / infoGlobal.m_timeStep);
 		}
 	}
 
 	m_x.resize(numConstraintRows);
 	if (infoGlobal.m_splitImpulse)
 		m_xSplit.resize(numConstraintRows);
-//	m_x.setZero();
+	//	m_x.setZero();
 
-	for (int i=0;i<m_allConstraintPtrArray.size();i++)
+	for (int i = 0; i < m_allConstraintPtrArray.size(); i++)
 	{
 		const btSolverConstraint& c = *m_allConstraintPtrArray[i];
-		m_x[i]=c.m_appliedImpulse;
+		m_x[i] = c.m_appliedImpulse;
 		if (infoGlobal.m_splitImpulse)
 			m_xSplit[i] = c.m_appliedPushImpulse;
 	}
-
 }
 
-
-btScalar btMLCPSolver::solveGroupCacheFriendlyIterations(btCollisionObject** bodies ,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer)
+btScalar btMLCPSolver::solveGroupCacheFriendlyIterations(btCollisionObject** bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer)
 {
 	bool result = true;
 	{
 		BT_PROFILE("solveMLCP");
-//		printf("m_A(%d,%d)\n", m_A.rows(),m_A.cols());
+		//		printf("m_A(%d,%d)\n", m_A.rows(),m_A.cols());
 		result = solveMLCP(infoGlobal);
 	}
 
@@ -596,44 +580,41 @@ btScalar btMLCPSolver::solveGroupCacheFriendlyIterations(btCollisionObject** bod
 	if (result)
 	{
 		BT_PROFILE("process MLCP results");
-		for (int i=0;i<m_allConstraintPtrArray.size();i++)
+		for (int i = 0; i < m_allConstraintPtrArray.size(); i++)
 		{
 			{
 				btSolverConstraint& c = *m_allConstraintPtrArray[i];
 				int sbA = c.m_solverBodyIdA;
 				int sbB = c.m_solverBodyIdB;
 				//btRigidBody* orgBodyA = m_tmpSolverBodyPool[sbA].m_originalBody;
-			//	btRigidBody* orgBodyB = m_tmpSolverBodyPool[sbB].m_originalBody;
+				//	btRigidBody* orgBodyB = m_tmpSolverBodyPool[sbB].m_originalBody;
 
 				btSolverBody& solverBodyA = m_tmpSolverBodyPool[sbA];
 				btSolverBody& solverBodyB = m_tmpSolverBodyPool[sbB];
- 
+
 				{
-					btScalar deltaImpulse = m_x[i]-c.m_appliedImpulse;
+					btScalar deltaImpulse = m_x[i] - c.m_appliedImpulse;
 					c.m_appliedImpulse = m_x[i];
-					solverBodyA.internalApplyImpulse(c.m_contactNormal1*solverBodyA.internalGetInvMass(),c.m_angularComponentA,deltaImpulse);
-					solverBodyB.internalApplyImpulse(c.m_contactNormal2*solverBodyB.internalGetInvMass(),c.m_angularComponentB,deltaImpulse);
+					solverBodyA.internalApplyImpulse(c.m_contactNormal1 * solverBodyA.internalGetInvMass(), c.m_angularComponentA, deltaImpulse);
+					solverBodyB.internalApplyImpulse(c.m_contactNormal2 * solverBodyB.internalGetInvMass(), c.m_angularComponentB, deltaImpulse);
 				}
 
 				if (infoGlobal.m_splitImpulse)
 				{
 					btScalar deltaImpulse = m_xSplit[i] - c.m_appliedPushImpulse;
-					solverBodyA.internalApplyPushImpulse(c.m_contactNormal1*solverBodyA.internalGetInvMass(),c.m_angularComponentA,deltaImpulse);
-					solverBodyB.internalApplyPushImpulse(c.m_contactNormal2*solverBodyB.internalGetInvMass(),c.m_angularComponentB,deltaImpulse);
+					solverBodyA.internalApplyPushImpulse(c.m_contactNormal1 * solverBodyA.internalGetInvMass(), c.m_angularComponentA, deltaImpulse);
+					solverBodyB.internalApplyPushImpulse(c.m_contactNormal2 * solverBodyB.internalGetInvMass(), c.m_angularComponentB, deltaImpulse);
 					c.m_appliedPushImpulse = m_xSplit[i];
 				}
-				
 			}
 		}
 	}
 	else
 	{
-	//	printf("m_fallback = %d\n",m_fallback);
+		//	printf("m_fallback = %d\n",m_fallback);
 		m_fallback++;
-		btSequentialImpulseConstraintSolver::solveGroupCacheFriendlyIterations(bodies ,numBodies,manifoldPtr, numManifolds,constraints,numConstraints,infoGlobal,debugDrawer);
+		btSequentialImpulseConstraintSolver::solveGroupCacheFriendlyIterations(bodies, numBodies, manifoldPtr, numManifolds, constraints, numConstraints, infoGlobal, debugDrawer);
 	}
 
 	return 0.f;
 }
-
-
diff --git a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btMLCPSolver.h b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btMLCPSolver.h
index 26b482ddc10..510ae59e58b 100644
--- a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btMLCPSolver.h
+++ b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btMLCPSolver.h
@@ -23,15 +23,13 @@ subject to the following restrictions:
 
 class btMLCPSolver : public btSequentialImpulseConstraintSolver
 {
-
 protected:
-	
 	btMatrixXu m_A;
 	btVectorXu m_b;
 	btVectorXu m_x;
 	btVectorXu m_lo;
 	btVectorXu m_hi;
-	
+
 	///when using 'split impulse' we solve two separate (M)LCPs
 	btVectorXu m_bSplit;
 	btVectorXu m_xSplit;
@@ -39,24 +37,23 @@ protected:
 	btVectorXu m_xSplit2;
 
 	btAlignedObjectArray<int> m_limitDependencies;
-	btAlignedObjectArray<btSolverConstraint*>	m_allConstraintPtrArray;
+	btAlignedObjectArray<btSolverConstraint*> m_allConstraintPtrArray;
 	btMLCPSolverInterface* m_solver;
 	int m_fallback;
 
-    /// The following scratch variables are not stateful -- contents are cleared prior to each use.
-    /// They are only cached here to avoid extra memory allocations and deallocations and to ensure
-    /// that multiple instances of the solver can be run in parallel.
-    btMatrixXu m_scratchJ3;
-    btMatrixXu m_scratchJInvM3;
-    btAlignedObjectArray<int> m_scratchOfs;
-    btMatrixXu m_scratchMInv;
-    btMatrixXu m_scratchJ;
-    btMatrixXu m_scratchJTranspose;
-    btMatrixXu m_scratchTmp;
-
-	virtual btScalar solveGroupCacheFriendlySetup(btCollisionObject** bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer);
-	virtual btScalar solveGroupCacheFriendlyIterations(btCollisionObject** bodies ,int numBodies,btPersistentManifold** manifoldPtr, int numManifolds,btTypedConstraint** constraints,int numConstraints,const btContactSolverInfo& infoGlobal,btIDebugDraw* debugDrawer);
+	/// The following scratch variables are not stateful -- contents are cleared prior to each use.
+	/// They are only cached here to avoid extra memory allocations and deallocations and to ensure
+	/// that multiple instances of the solver can be run in parallel.
+	btMatrixXu m_scratchJ3;
+	btMatrixXu m_scratchJInvM3;
+	btAlignedObjectArray<int> m_scratchOfs;
+	btMatrixXu m_scratchMInv;
+	btMatrixXu m_scratchJ;
+	btMatrixXu m_scratchJTranspose;
+	btMatrixXu m_scratchTmp;
 
+	virtual btScalar solveGroupCacheFriendlySetup(btCollisionObject** bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer);
+	virtual btScalar solveGroupCacheFriendlyIterations(btCollisionObject** bodies, int numBodies, btPersistentManifold** manifoldPtr, int numManifolds, btTypedConstraint** constraints, int numConstraints, const btContactSolverInfo& infoGlobal, btIDebugDraw* debugDrawer);
 
 	virtual void createMLCP(const btContactSolverInfo& infoGlobal);
 	virtual void createMLCPFast(const btContactSolverInfo& infoGlobal);
@@ -65,8 +62,7 @@ protected:
 	virtual bool solveMLCP(const btContactSolverInfo& infoGlobal);
 
 public:
-
-	btMLCPSolver(	 btMLCPSolverInterface* solver);
+	btMLCPSolver(btMLCPSolverInterface* solver);
 	virtual ~btMLCPSolver();
 
 	void setMLCPSolver(btMLCPSolverInterface* solver)
@@ -83,12 +79,10 @@ public:
 		m_fallback = num;
 	}
 
-	virtual btConstraintSolverType	getSolverType() const
+	virtual btConstraintSolverType getSolverType() const
 	{
 		return BT_MLCP_SOLVER;
 	}
-
 };
 
-
-#endif //BT_MLCP_SOLVER_H
+#endif  //BT_MLCP_SOLVER_H
diff --git a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btMLCPSolverInterface.h b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btMLCPSolverInterface.h
index 25bb3f6d327..6b0465b88d7 100644
--- a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btMLCPSolverInterface.h
+++ b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btMLCPSolverInterface.h
@@ -27,7 +27,7 @@ public:
 	}
 
 	//return true is it solves the problem successfully
-	virtual bool solveMLCP(const btMatrixXu & A, const btVectorXu & b, btVectorXu& x, const btVectorXu & lo,const btVectorXu & hi,const btAlignedObjectArray<int>& limitDependency, int numIterations, bool useSparsity = true)=0;
+	virtual bool solveMLCP(const btMatrixXu& A, const btVectorXu& b, btVectorXu& x, const btVectorXu& lo, const btVectorXu& hi, const btAlignedObjectArray<int>& limitDependency, int numIterations, bool useSparsity = true) = 0;
 };
 
-#endif //BT_MLCP_SOLVER_INTERFACE_H
+#endif  //BT_MLCP_SOLVER_INTERFACE_H
diff --git a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btPATHSolver.h b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btPATHSolver.h
index 9ec31a6d4e4..7f8eec3f6e3 100644
--- a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btPATHSolver.h
+++ b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btPATHSolver.h
@@ -14,38 +14,35 @@ subject to the following restrictions:
 */
 ///original version written by Erwin Coumans, October 2013
 
-
 #ifndef BT_PATH_SOLVER_H
 #define BT_PATH_SOLVER_H
 
 //#define BT_USE_PATH
 #ifdef BT_USE_PATH
 
-extern "C" {
+extern "C"
+{
 #include "PATH/SimpleLCP.h"
 #include "PATH/License.h"
 #include "PATH/Error_Interface.h"
 };
-  void __stdcall MyError(Void *data, Char *msg)
+void __stdcall MyError(Void *data, Char *msg)
 {
-	printf("Path Error: %s\n",msg);
+	printf("Path Error: %s\n", msg);
 }
-  void __stdcall MyWarning(Void *data, Char *msg)
+void __stdcall MyWarning(Void *data, Char *msg)
 {
-	printf("Path Warning: %s\n",msg);
+	printf("Path Warning: %s\n", msg);
 }
 
 Error_Interface e;
 
-
-
 #include "btMLCPSolverInterface.h"
 #include "Dantzig/lcp.h"
 
 class btPathSolver : public btMLCPSolverInterface
 {
 public:
-
 	btPathSolver()
 	{
 		License_SetString("2069810742&Courtesy_License&&&USR&2013&14_12_2011&1000&PATH&GEN&31_12_2013&0_0_0&0&0_0");
@@ -55,17 +52,15 @@ public:
 		Error_SetInterface(&e);
 	}
 
-
-	virtual bool solveMLCP(const btMatrixXu & A, const btVectorXu & b, btVectorXu& x, const btVectorXu & lo,const btVectorXu & hi,const btAlignedObjectArray<int>& limitDependency, int numIterations, bool useSparsity = true)
+	virtual bool solveMLCP(const btMatrixXu &A, const btVectorXu &b, btVectorXu &x, const btVectorXu &lo, const btVectorXu &hi, const btAlignedObjectArray<int> &limitDependency, int numIterations, bool useSparsity = true)
 	{
 		MCP_Termination status;
-		
 
 		int numVariables = b.rows();
-		if (0==numVariables)
+		if (0 == numVariables)
 			return true;
 
-			/*	 - variables - the number of variables in the problem
+		/*	 - variables - the number of variables in the problem
 			- m_nnz - the number of nonzeros in the M matrix
 			- m_i - a vector of size m_nnz containing the row indices for M
 			- m_j - a vector of size m_nnz containing the column indices for M
@@ -78,16 +73,16 @@ public:
 		btAlignedObjectArray<int> rowIndices;
 		btAlignedObjectArray<int> colIndices;
 
-		for (int i=0;i<A.rows();i++)
+		for (int i = 0; i < A.rows(); i++)
 		{
-			for (int j=0;j<A.cols();j++)
+			for (int j = 0; j < A.cols(); j++)
 			{
-				if (A(i,j)!=0.f)
+				if (A(i, j) != 0.f)
 				{
 					//add 1, because Path starts at 1, instead of 0
-					rowIndices.push_back(i+1);
-					colIndices.push_back(j+1);
-					values.push_back(A(i,j));
+					rowIndices.push_back(i + 1);
+					colIndices.push_back(j + 1);
+					values.push_back(A(i, j));
 				}
 			}
 		}
@@ -97,19 +92,18 @@ public:
 		btAlignedObjectArray<double> rhs;
 		btAlignedObjectArray<double> upperBounds;
 		btAlignedObjectArray<double> lowerBounds;
-		for (int i=0;i<numVariables;i++)
+		for (int i = 0; i < numVariables; i++)
 		{
 			upperBounds.push_back(hi[i]);
 			lowerBounds.push_back(lo[i]);
 			rhs.push_back(-b[i]);
 		}
 
-
-		SimpleLCP(numVariables,numNonZero,&rowIndices[0],&colIndices[0],&values[0],&rhs[0],&lowerBounds[0],&upperBounds[0], &status, &zResult[0]);
+		SimpleLCP(numVariables, numNonZero, &rowIndices[0], &colIndices[0], &values[0], &rhs[0], &lowerBounds[0], &upperBounds[0], &status, &zResult[0]);
 
 		if (status != MCP_Solved)
 		{
-			static const char* gReturnMsgs[] = {
+			static const char *gReturnMsgs[] = {
 				"Invalid return",
 				"MCP_Solved: The problem was solved",
 				"MCP_NoProgress: A stationary point was found",
@@ -122,16 +116,16 @@ public:
 				"MCP_Infeasible: Problem has no solution",
 				"MCP_Error: An error occurred within the code",
 				"MCP_LicenseError: License could not be found",
-				"MCP_OK"
-			};
+				"MCP_OK"};
 
-			printf("ERROR: The PATH MCP solver failed: %s\n", gReturnMsgs[(unsigned int)status]);// << std::endl;
+			printf("ERROR: The PATH MCP solver failed: %s\n", gReturnMsgs[(unsigned int)status]);  // << std::endl;
 			printf("using Projected Gauss Seidel fallback\n");
-			
+
 			return false;
-		} else
+		}
+		else
 		{
-			for (int i=0;i<numVariables;i++)
+			for (int i = 0; i < numVariables; i++)
 			{
 				x[i] = zResult[i];
 				//check for #NAN
@@ -139,13 +133,10 @@ public:
 					return false;
 			}
 			return true;
-
 		}
-
 	}
 };
 
-#endif //BT_USE_PATH
+#endif  //BT_USE_PATH
 
-
-#endif //BT_PATH_SOLVER_H
+#endif  //BT_PATH_SOLVER_H
diff --git a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btSolveProjectedGaussSeidel.h b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btSolveProjectedGaussSeidel.h
index c0b40ffd9f8..c3f4ec39974 100644
--- a/thirdparty/bullet/BulletDynamics/MLCPSolvers/btSolveProjectedGaussSeidel.h
+++ b/thirdparty/bullet/BulletDynamics/MLCPSolvers/btSolveProjectedGaussSeidel.h
@@ -17,25 +17,22 @@ subject to the following restrictions:
 #ifndef BT_SOLVE_PROJECTED_GAUSS_SEIDEL_H
 #define BT_SOLVE_PROJECTED_GAUSS_SEIDEL_H
 
-
 #include "btMLCPSolverInterface.h"
 
 ///This solver is mainly for debug/learning purposes: it is functionally equivalent to the btSequentialImpulseConstraintSolver solver, but much slower (it builds the full LCP matrix)
 class btSolveProjectedGaussSeidel : public btMLCPSolverInterface
 {
-
 public:
-
 	btScalar m_leastSquaresResidualThreshold;
 	btScalar m_leastSquaresResidual;
 
 	btSolveProjectedGaussSeidel()
-		:m_leastSquaresResidualThreshold(0),
-		m_leastSquaresResidual(0)
+		: m_leastSquaresResidualThreshold(0),
+		  m_leastSquaresResidual(0)
 	{
 	}
 
-	virtual bool solveMLCP(const btMatrixXu & A, const btVectorXu & b, btVectorXu& x, const btVectorXu & lo,const btVectorXu & hi,const btAlignedObjectArray<int>& limitDependency, int numIterations, bool useSparsity = true)
+	virtual bool solveMLCP(const btMatrixXu& A, const btVectorXu& b, btVectorXu& x, const btVectorXu& lo, const btVectorXu& hi, const btAlignedObjectArray<int>& limitDependency, int numIterations, bool useSparsity = true)
 	{
 		if (!A.rows())
 			return true;
@@ -46,65 +43,65 @@ public:
 		btAssert(A.rows() == b.rows());
 
 		int i, j, numRows = A.rows();
-	
+
 		btScalar delta;
 
-		for (int k = 0; k <numIterations; k++)
+		for (int k = 0; k < numIterations; k++)
 		{
 			m_leastSquaresResidual = 0.f;
-			for (i = 0; i <numRows; i++)
+			for (i = 0; i < numRows; i++)
 			{
 				delta = 0.0f;
 				if (useSparsity)
 				{
-					for (int h=0;h<A.m_rowNonZeroElements1[i].size();h++)
+					for (int h = 0; h < A.m_rowNonZeroElements1[i].size(); h++)
 					{
-						int j = A.m_rowNonZeroElements1[i][h];
-						if (j != i)//skip main diagonal
+						j = A.m_rowNonZeroElements1[i][h];
+						if (j != i)  //skip main diagonal
 						{
-							delta += A(i,j) * x[j];
+							delta += A(i, j) * x[j];
 						}
 					}
-				} else
+				}
+				else
 				{
-					for (j = 0; j <i; j++) 
-						delta += A(i,j) * x[j];
-					for (j = i+1; j<numRows; j++) 
-						delta += A(i,j) * x[j];
+					for (j = 0; j < i; j++)
+						delta += A(i, j) * x[j];
+					for (j = i + 1; j < numRows; j++)
+						delta += A(i, j) * x[j];
 				}
 
-				btScalar aDiag = A(i,i);
+				btScalar aDiag = A(i, i);
 				btScalar xOld = x[i];
-				x [i] = (b [i] - delta) / aDiag;
+				x[i] = (b[i] - delta) / aDiag;
 				btScalar s = 1.f;
 
-				if (limitDependency[i]>=0)
+				if (limitDependency[i] >= 0)
 				{
 					s = x[limitDependency[i]];
-					if (s<0)
-						s=1;
+					if (s < 0)
+						s = 1;
 				}
-			
-				if (x[i]<lo[i]*s)
-					x[i]=lo[i]*s;
-				if (x[i]>hi[i]*s)
-					x[i]=hi[i]*s;
+
+				if (x[i] < lo[i] * s)
+					x[i] = lo[i] * s;
+				if (x[i] > hi[i] * s)
+					x[i] = hi[i] * s;
 				btScalar diff = x[i] - xOld;
-				m_leastSquaresResidual += diff*diff;
+				m_leastSquaresResidual += diff * diff;
 			}
 
-			btScalar eps  = m_leastSquaresResidualThreshold;
-			if ((m_leastSquaresResidual < eps) || (k >=(numIterations-1)))
+			btScalar eps = m_leastSquaresResidualThreshold;
+			if ((m_leastSquaresResidual < eps) || (k >= (numIterations - 1)))
 			{
 #ifdef VERBOSE_PRINTF_RESIDUAL
-				printf("totalLenSqr = %f at iteration #%d\n", m_leastSquaresResidual,k);
+				printf("totalLenSqr = %f at iteration #%d\n", m_leastSquaresResidual, k);
 #endif
 				break;
 			}
 		}
 		return true;
 	}
-
 };
 
-#endif //BT_SOLVE_PROJECTED_GAUSS_SEIDEL_H
+#endif  //BT_SOLVE_PROJECTED_GAUSS_SEIDEL_H
diff --git a/thirdparty/bullet/BulletDynamics/Vehicle/btRaycastVehicle.cpp b/thirdparty/bullet/BulletDynamics/Vehicle/btRaycastVehicle.cpp
index f299aa34e8d..fc70d8e637d 100644
--- a/thirdparty/bullet/BulletDynamics/Vehicle/btRaycastVehicle.cpp
+++ b/thirdparty/bullet/BulletDynamics/Vehicle/btRaycastVehicle.cpp
@@ -24,17 +24,16 @@
 
 #define ROLLING_INFLUENCE_FIX
 
-
 btRigidBody& btActionInterface::getFixedBody()
 {
-	static btRigidBody s_fixed(0, 0,0);
-	s_fixed.setMassProps(btScalar(0.),btVector3(btScalar(0.),btScalar(0.),btScalar(0.)));
+	static btRigidBody s_fixed(0, 0, 0);
+	s_fixed.setMassProps(btScalar(0.), btVector3(btScalar(0.), btScalar(0.), btScalar(0.)));
 	return s_fixed;
 }
 
-btRaycastVehicle::btRaycastVehicle(const btVehicleTuning& tuning,btRigidBody* chassis,	btVehicleRaycaster* raycaster )
-:m_vehicleRaycaster(raycaster),
-m_pitchControl(btScalar(0.))
+btRaycastVehicle::btRaycastVehicle(const btVehicleTuning& tuning, btRigidBody* chassis, btVehicleRaycaster* raycaster)
+	: m_vehicleRaycaster(raycaster),
+	  m_pitchControl(btScalar(0.))
 {
 	m_chassisBody = chassis;
 	m_indexRightAxis = 0;
@@ -43,28 +42,22 @@ m_pitchControl(btScalar(0.))
 	defaultInit(tuning);
 }
 
-
 void btRaycastVehicle::defaultInit(const btVehicleTuning& tuning)
 {
 	(void)tuning;
 	m_currentVehicleSpeedKmHour = btScalar(0.);
 	m_steeringValue = btScalar(0.);
-	
 }
 
-	
-
 btRaycastVehicle::~btRaycastVehicle()
 {
 }
 
-
 //
 // basically most of the code is general for 2 or 4 wheel vehicles, but some of it needs to be reviewed
 //
-btWheelInfo&	btRaycastVehicle::addWheel( const btVector3& connectionPointCS, const btVector3& wheelDirectionCS0,const btVector3& wheelAxleCS, btScalar suspensionRestLength, btScalar wheelRadius,const btVehicleTuning& tuning, bool isFrontWheel)
+btWheelInfo& btRaycastVehicle::addWheel(const btVector3& connectionPointCS, const btVector3& wheelDirectionCS0, const btVector3& wheelAxleCS, btScalar suspensionRestLength, btScalar wheelRadius, const btVehicleTuning& tuning, bool isFrontWheel)
 {
-
 	btWheelInfoConstructionInfo ci;
 
 	ci.m_chassisConnectionCS = connectionPointCS;
@@ -80,83 +73,76 @@ btWheelInfo&	btRaycastVehicle::addWheel( const btVector3& connectionPointCS, con
 	ci.m_maxSuspensionTravelCm = tuning.m_maxSuspensionTravelCm;
 	ci.m_maxSuspensionForce = tuning.m_maxSuspensionForce;
 
-	m_wheelInfo.push_back( btWheelInfo(ci));
-	
-	btWheelInfo& wheel = m_wheelInfo[getNumWheels()-1];
-	
-	updateWheelTransformsWS( wheel , false );
-	updateWheelTransform(getNumWheels()-1,false);
+	m_wheelInfo.push_back(btWheelInfo(ci));
+
+	btWheelInfo& wheel = m_wheelInfo[getNumWheels() - 1];
+
+	updateWheelTransformsWS(wheel, false);
+	updateWheelTransform(getNumWheels() - 1, false);
 	return wheel;
 }
 
-
-
-
-const btTransform&	btRaycastVehicle::getWheelTransformWS( int wheelIndex ) const
+const btTransform& btRaycastVehicle::getWheelTransformWS(int wheelIndex) const
 {
 	btAssert(wheelIndex < getNumWheels());
 	const btWheelInfo& wheel = m_wheelInfo[wheelIndex];
 	return wheel.m_worldTransform;
-
 }
 
-void	btRaycastVehicle::updateWheelTransform( int wheelIndex , bool interpolatedTransform)
+void btRaycastVehicle::updateWheelTransform(int wheelIndex, bool interpolatedTransform)
 {
-	
-	btWheelInfo& wheel = m_wheelInfo[ wheelIndex ];
-	updateWheelTransformsWS(wheel,interpolatedTransform);
+	btWheelInfo& wheel = m_wheelInfo[wheelIndex];
+	updateWheelTransformsWS(wheel, interpolatedTransform);
 	btVector3 up = -wheel.m_raycastInfo.m_wheelDirectionWS;
 	const btVector3& right = wheel.m_raycastInfo.m_wheelAxleWS;
 	btVector3 fwd = up.cross(right);
 	fwd = fwd.normalize();
-//	up = right.cross(fwd);
-//	up.normalize();
+	//	up = right.cross(fwd);
+	//	up.normalize();
 
 	//rotate around steering over de wheelAxleWS
 	btScalar steering = wheel.m_steering;
-	
-	btQuaternion steeringOrn(up,steering);//wheel.m_steering);
+
+	btQuaternion steeringOrn(up, steering);  //wheel.m_steering);
 	btMatrix3x3 steeringMat(steeringOrn);
 
-	btQuaternion rotatingOrn(right,-wheel.m_rotation);
+	btQuaternion rotatingOrn(right, -wheel.m_rotation);
 	btMatrix3x3 rotatingMat(rotatingOrn);
 
-    btMatrix3x3 basis2;
-    basis2[0][m_indexRightAxis] = -right[0];
-    basis2[1][m_indexRightAxis] = -right[1];
-    basis2[2][m_indexRightAxis] = -right[2];
+	btMatrix3x3 basis2;
+	basis2[0][m_indexRightAxis] = -right[0];
+	basis2[1][m_indexRightAxis] = -right[1];
+	basis2[2][m_indexRightAxis] = -right[2];
 
-    basis2[0][m_indexUpAxis] = up[0];
-    basis2[1][m_indexUpAxis] = up[1];
-    basis2[2][m_indexUpAxis] = up[2];
+	basis2[0][m_indexUpAxis] = up[0];
+	basis2[1][m_indexUpAxis] = up[1];
+	basis2[2][m_indexUpAxis] = up[2];
 
-    basis2[0][m_indexForwardAxis] = fwd[0];
-    basis2[1][m_indexForwardAxis] = fwd[1];
-    basis2[2][m_indexForwardAxis] = fwd[2];
+	basis2[0][m_indexForwardAxis] = fwd[0];
+	basis2[1][m_indexForwardAxis] = fwd[1];
+	basis2[2][m_indexForwardAxis] = fwd[2];
 
 	wheel.m_worldTransform.setBasis(steeringMat * rotatingMat * basis2);
 	wheel.m_worldTransform.setOrigin(
-		wheel.m_raycastInfo.m_hardPointWS + wheel.m_raycastInfo.m_wheelDirectionWS * wheel.m_raycastInfo.m_suspensionLength
-	);
+		wheel.m_raycastInfo.m_hardPointWS + wheel.m_raycastInfo.m_wheelDirectionWS * wheel.m_raycastInfo.m_suspensionLength);
 }
 
 void btRaycastVehicle::resetSuspension()
 {
-
 	int i;
-	for (i=0;i<m_wheelInfo.size();	i++)
+	for (i = 0; i < m_wheelInfo.size(); i++)
 	{
-			btWheelInfo& wheel = m_wheelInfo[i];
-			wheel.m_raycastInfo.m_suspensionLength = wheel.getSuspensionRestLength();
-			wheel.m_suspensionRelativeVelocity = btScalar(0.0);
-			
-			wheel.m_raycastInfo.m_contactNormalWS = - wheel.m_raycastInfo.m_wheelDirectionWS;
-			//wheel_info.setContactFriction(btScalar(0.0));
-			wheel.m_clippedInvContactDotSuspension = btScalar(1.0);
+		btWheelInfo& wheel = m_wheelInfo[i];
+		wheel.m_raycastInfo.m_suspensionLength = wheel.getSuspensionRestLength();
+		wheel.m_suspensionRelativeVelocity = btScalar(0.0);
+
+		wheel.m_raycastInfo.m_contactNormalWS = -wheel.m_raycastInfo.m_wheelDirectionWS;
+		//wheel_info.setContactFriction(btScalar(0.0));
+		wheel.m_clippedInvContactDotSuspension = btScalar(1.0);
 	}
 }
 
-void	btRaycastVehicle::updateWheelTransformsWS(btWheelInfo& wheel , bool interpolatedTransform)
+void btRaycastVehicle::updateWheelTransformsWS(btWheelInfo& wheel, bool interpolatedTransform)
 {
 	wheel.m_raycastInfo.m_isInContact = false;
 
@@ -166,19 +152,18 @@ void	btRaycastVehicle::updateWheelTransformsWS(btWheelInfo& wheel , bool interpo
 		getRigidBody()->getMotionState()->getWorldTransform(chassisTrans);
 	}
 
-	wheel.m_raycastInfo.m_hardPointWS = chassisTrans( wheel.m_chassisConnectionPointCS );
-	wheel.m_raycastInfo.m_wheelDirectionWS = chassisTrans.getBasis() *  wheel.m_wheelDirectionCS ;
+	wheel.m_raycastInfo.m_hardPointWS = chassisTrans(wheel.m_chassisConnectionPointCS);
+	wheel.m_raycastInfo.m_wheelDirectionWS = chassisTrans.getBasis() * wheel.m_wheelDirectionCS;
 	wheel.m_raycastInfo.m_wheelAxleWS = chassisTrans.getBasis() * wheel.m_wheelAxleCS;
 }
 
 btScalar btRaycastVehicle::rayCast(btWheelInfo& wheel)
 {
-	updateWheelTransformsWS( wheel,false);
+	updateWheelTransformsWS(wheel, false);
 
-	
 	btScalar depth = -1;
-	
-	btScalar raylen = wheel.getSuspensionRestLength()+wheel.m_wheelsRadius;
+
+	btScalar raylen = wheel.getSuspensionRestLength() + wheel.m_wheelsRadius;
 
 	btVector3 rayvector = wheel.m_raycastInfo.m_wheelDirectionWS * (raylen);
 	const btVector3& source = wheel.m_raycastInfo.m_hardPointWS;
@@ -186,12 +171,12 @@ btScalar btRaycastVehicle::rayCast(btWheelInfo& wheel)
 	const btVector3& target = wheel.m_raycastInfo.m_contactPointWS;
 
 	btScalar param = btScalar(0.);
-	
-	btVehicleRaycaster::btVehicleRaycasterResult	rayResults;
+
+	btVehicleRaycaster::btVehicleRaycasterResult rayResults;
 
 	btAssert(m_vehicleRaycaster);
 
-	void* object = m_vehicleRaycaster->castRay(source,target,rayResults);
+	void* object = m_vehicleRaycaster->castRay(source, target, rayResults);
 
 	wheel.m_raycastInfo.m_groundObject = 0;
 
@@ -199,19 +184,18 @@ btScalar btRaycastVehicle::rayCast(btWheelInfo& wheel)
 	{
 		param = rayResults.m_distFraction;
 		depth = raylen * rayResults.m_distFraction;
-		wheel.m_raycastInfo.m_contactNormalWS  = rayResults.m_hitNormalInWorld;
+		wheel.m_raycastInfo.m_contactNormalWS = rayResults.m_hitNormalInWorld;
 		wheel.m_raycastInfo.m_isInContact = true;
-		
-		wheel.m_raycastInfo.m_groundObject = &getFixedBody();///@todo for driving on dynamic/movable objects!;
+
+		wheel.m_raycastInfo.m_groundObject = &getFixedBody();  ///@todo for driving on dynamic/movable objects!;
 		//wheel.m_raycastInfo.m_groundObject = object;
 
-
-		btScalar hitDistance = param*raylen;
+		btScalar hitDistance = param * raylen;
 		wheel.m_raycastInfo.m_suspensionLength = hitDistance - wheel.m_wheelsRadius;
 		//clamp on max suspension travel
 
-		btScalar  minSuspensionLength = wheel.getSuspensionRestLength() - wheel.m_maxSuspensionTravelCm*btScalar(0.01);
-		btScalar maxSuspensionLength = wheel.getSuspensionRestLength()+ wheel.m_maxSuspensionTravelCm*btScalar(0.01);
+		btScalar minSuspensionLength = wheel.getSuspensionRestLength() - wheel.m_maxSuspensionTravelCm * btScalar(0.01);
+		btScalar maxSuspensionLength = wheel.getSuspensionRestLength() + wheel.m_maxSuspensionTravelCm * btScalar(0.01);
 		if (wheel.m_raycastInfo.m_suspensionLength < minSuspensionLength)
 		{
 			wheel.m_raycastInfo.m_suspensionLength = minSuspensionLength;
@@ -223,16 +207,16 @@ btScalar btRaycastVehicle::rayCast(btWheelInfo& wheel)
 
 		wheel.m_raycastInfo.m_contactPointWS = rayResults.m_hitPointInWorld;
 
-		btScalar denominator= wheel.m_raycastInfo.m_contactNormalWS.dot( wheel.m_raycastInfo.m_wheelDirectionWS );
+		btScalar denominator = wheel.m_raycastInfo.m_contactNormalWS.dot(wheel.m_raycastInfo.m_wheelDirectionWS);
 
 		btVector3 chassis_velocity_at_contactPoint;
-		btVector3 relpos = wheel.m_raycastInfo.m_contactPointWS-getRigidBody()->getCenterOfMassPosition();
+		btVector3 relpos = wheel.m_raycastInfo.m_contactPointWS - getRigidBody()->getCenterOfMassPosition();
 
 		chassis_velocity_at_contactPoint = getRigidBody()->getVelocityInLocalPoint(relpos);
 
-		btScalar projVel = wheel.m_raycastInfo.m_contactNormalWS.dot( chassis_velocity_at_contactPoint );
+		btScalar projVel = wheel.m_raycastInfo.m_contactNormalWS.dot(chassis_velocity_at_contactPoint);
 
-		if ( denominator >= btScalar(-0.1))
+		if (denominator >= btScalar(-0.1))
 		{
 			wheel.m_suspensionRelativeVelocity = btScalar(0.0);
 			wheel.m_clippedInvContactDotSuspension = btScalar(1.0) / btScalar(0.1);
@@ -243,20 +227,19 @@ btScalar btRaycastVehicle::rayCast(btWheelInfo& wheel)
 			wheel.m_suspensionRelativeVelocity = projVel * inv;
 			wheel.m_clippedInvContactDotSuspension = inv;
 		}
-			
-	} else
+	}
+	else
 	{
 		//put wheel info as in rest position
 		wheel.m_raycastInfo.m_suspensionLength = wheel.getSuspensionRestLength();
 		wheel.m_suspensionRelativeVelocity = btScalar(0.0);
-		wheel.m_raycastInfo.m_contactNormalWS = - wheel.m_raycastInfo.m_wheelDirectionWS;
+		wheel.m_raycastInfo.m_contactNormalWS = -wheel.m_raycastInfo.m_wheelDirectionWS;
 		wheel.m_clippedInvContactDotSuspension = btScalar(1.0);
 	}
 
 	return depth;
 }
 
-
 const btTransform& btRaycastVehicle::getChassisWorldTransform() const
 {
 	/*if (getRigidBody()->getMotionState())
@@ -267,26 +250,23 @@ const btTransform& btRaycastVehicle::getChassisWorldTransform() const
 	}
 	*/
 
-	
 	return getRigidBody()->getCenterOfMassTransform();
 }
 
-
-void btRaycastVehicle::updateVehicle( btScalar step )
+void btRaycastVehicle::updateVehicle(btScalar step)
 {
 	{
-		for (int i=0;i<getNumWheels();i++)
+		for (int i = 0; i < getNumWheels(); i++)
 		{
-			updateWheelTransform(i,false);
+			updateWheelTransform(i, false);
 		}
 	}
 
-
 	m_currentVehicleSpeedKmHour = btScalar(3.6) * getRigidBody()->getLinearVelocity().length();
-	
+
 	const btTransform& chassisTrans = getChassisWorldTransform();
 
-	btVector3 forwardW (
+	btVector3 forwardW(
 		chassisTrans.getBasis()[0][m_indexForwardAxis],
 		chassisTrans.getBasis()[1][m_indexForwardAxis],
 		chassisTrans.getBasis()[2][m_indexForwardAxis]);
@@ -299,52 +279,47 @@ void btRaycastVehicle::updateVehicle( btScalar step )
 	//
 	// simulate suspension
 	//
-	
-	int i=0;
-	for (i=0;i<m_wheelInfo.size();i++)
+
+	int i = 0;
+	for (i = 0; i < m_wheelInfo.size(); i++)
 	{
-		//btScalar depth; 
-		//depth = 
-		rayCast( m_wheelInfo[i]);
+		//btScalar depth;
+		//depth =
+		rayCast(m_wheelInfo[i]);
 	}
 
 	updateSuspension(step);
 
-	
-	for (i=0;i<m_wheelInfo.size();i++)
+	for (i = 0; i < m_wheelInfo.size(); i++)
 	{
 		//apply suspension force
 		btWheelInfo& wheel = m_wheelInfo[i];
-		
+
 		btScalar suspensionForce = wheel.m_wheelsSuspensionForce;
-		
+
 		if (suspensionForce > wheel.m_maxSuspensionForce)
 		{
 			suspensionForce = wheel.m_maxSuspensionForce;
 		}
 		btVector3 impulse = wheel.m_raycastInfo.m_contactNormalWS * suspensionForce * step;
 		btVector3 relpos = wheel.m_raycastInfo.m_contactPointWS - getRigidBody()->getCenterOfMassPosition();
-		
+
 		getRigidBody()->applyImpulse(impulse, relpos);
-	
 	}
-	
 
-	
-	updateFriction( step);
+	updateFriction(step);
 
-	
-	for (i=0;i<m_wheelInfo.size();i++)
+	for (i = 0; i < m_wheelInfo.size(); i++)
 	{
 		btWheelInfo& wheel = m_wheelInfo[i];
 		btVector3 relpos = wheel.m_raycastInfo.m_hardPointWS - getRigidBody()->getCenterOfMassPosition();
-		btVector3 vel = getRigidBody()->getVelocityInLocalPoint( relpos );
+		btVector3 vel = getRigidBody()->getVelocityInLocalPoint(relpos);
 
 		if (wheel.m_raycastInfo.m_isInContact)
 		{
-			const btTransform&	chassisWorldTransform = getChassisWorldTransform();
+			const btTransform& chassisWorldTransform = getChassisWorldTransform();
 
-			btVector3 fwd (
+			btVector3 fwd(
 				chassisWorldTransform.getBasis()[0][m_indexForwardAxis],
 				chassisWorldTransform.getBasis()[1][m_indexForwardAxis],
 				chassisWorldTransform.getBasis()[2][m_indexForwardAxis]);
@@ -353,99 +328,88 @@ void btRaycastVehicle::updateVehicle( btScalar step )
 			fwd -= wheel.m_raycastInfo.m_contactNormalWS * proj;
 
 			btScalar proj2 = fwd.dot(vel);
-			
+
 			wheel.m_deltaRotation = (proj2 * step) / (wheel.m_wheelsRadius);
 			wheel.m_rotation += wheel.m_deltaRotation;
-
-		} else
+		}
+		else
 		{
 			wheel.m_rotation += wheel.m_deltaRotation;
 		}
-		
-		wheel.m_deltaRotation *= btScalar(0.99);//damping of rotation when not in contact
 
+		wheel.m_deltaRotation *= btScalar(0.99);  //damping of rotation when not in contact
 	}
-
-
-
 }
 
-
-void	btRaycastVehicle::setSteeringValue(btScalar steering,int wheel)
+void btRaycastVehicle::setSteeringValue(btScalar steering, int wheel)
 {
-	btAssert(wheel>=0 && wheel < getNumWheels());
+	btAssert(wheel >= 0 && wheel < getNumWheels());
 
 	btWheelInfo& wheelInfo = getWheelInfo(wheel);
 	wheelInfo.m_steering = steering;
 }
 
-
-
-btScalar	btRaycastVehicle::getSteeringValue(int wheel) const
+btScalar btRaycastVehicle::getSteeringValue(int wheel) const
 {
 	return getWheelInfo(wheel).m_steering;
 }
 
-
-void	btRaycastVehicle::applyEngineForce(btScalar force, int wheel)
+void btRaycastVehicle::applyEngineForce(btScalar force, int wheel)
 {
-	btAssert(wheel>=0 && wheel < getNumWheels());
+	btAssert(wheel >= 0 && wheel < getNumWheels());
 	btWheelInfo& wheelInfo = getWheelInfo(wheel);
 	wheelInfo.m_engineForce = force;
 }
 
-
-const btWheelInfo&	btRaycastVehicle::getWheelInfo(int index) const
+const btWheelInfo& btRaycastVehicle::getWheelInfo(int index) const
 {
-	btAssert((index >= 0) && (index < 	getNumWheels()));
-	
+	btAssert((index >= 0) && (index < getNumWheels()));
+
 	return m_wheelInfo[index];
 }
 
-btWheelInfo&	btRaycastVehicle::getWheelInfo(int index) 
+btWheelInfo& btRaycastVehicle::getWheelInfo(int index)
 {
-	btAssert((index >= 0) && (index < 	getNumWheels()));
-	
+	btAssert((index >= 0) && (index < getNumWheels()));
+
 	return m_wheelInfo[index];
 }
 
-void btRaycastVehicle::setBrake(btScalar brake,int wheelIndex)
+void btRaycastVehicle::setBrake(btScalar brake, int wheelIndex)
 {
-	btAssert((wheelIndex >= 0) && (wheelIndex < 	getNumWheels()));
+	btAssert((wheelIndex >= 0) && (wheelIndex < getNumWheels()));
 	getWheelInfo(wheelIndex).m_brake = brake;
 }
 
-
-void	btRaycastVehicle::updateSuspension(btScalar deltaTime)
+void btRaycastVehicle::updateSuspension(btScalar deltaTime)
 {
 	(void)deltaTime;
 
 	btScalar chassisMass = btScalar(1.) / m_chassisBody->getInvMass();
-	
-	for (int w_it=0; w_it<getNumWheels(); w_it++)
+
+	for (int w_it = 0; w_it < getNumWheels(); w_it++)
 	{
-		btWheelInfo &wheel_info = m_wheelInfo[w_it];
-		
-		if ( wheel_info.m_raycastInfo.m_isInContact )
+		btWheelInfo& wheel_info = m_wheelInfo[w_it];
+
+		if (wheel_info.m_raycastInfo.m_isInContact)
 		{
 			btScalar force;
 			//	Spring
 			{
-				btScalar	susp_length			= wheel_info.getSuspensionRestLength();
-				btScalar	current_length = wheel_info.m_raycastInfo.m_suspensionLength;
+				btScalar susp_length = wheel_info.getSuspensionRestLength();
+				btScalar current_length = wheel_info.m_raycastInfo.m_suspensionLength;
 
 				btScalar length_diff = (susp_length - current_length);
 
-				force = wheel_info.m_suspensionStiffness
-					* length_diff * wheel_info.m_clippedInvContactDotSuspension;
+				force = wheel_info.m_suspensionStiffness * length_diff * wheel_info.m_clippedInvContactDotSuspension;
 			}
-		
+
 			// Damper
 			{
 				btScalar projected_rel_vel = wheel_info.m_suspensionRelativeVelocity;
 				{
-					btScalar	susp_damping;
-					if ( projected_rel_vel < btScalar(0.0) )
+					btScalar susp_damping;
+					if (projected_rel_vel < btScalar(0.0))
 					{
 						susp_damping = wheel_info.m_wheelsDampingCompression;
 					}
@@ -469,50 +433,43 @@ void	btRaycastVehicle::updateSuspension(btScalar deltaTime)
 			wheel_info.m_wheelsSuspensionForce = btScalar(0.0);
 		}
 	}
-
 }
 
-
 struct btWheelContactPoint
 {
 	btRigidBody* m_body0;
 	btRigidBody* m_body1;
-	btVector3	m_frictionPositionWorld;
-	btVector3	m_frictionDirectionWorld;
-	btScalar	m_jacDiagABInv;
-	btScalar	m_maxImpulse;
+	btVector3 m_frictionPositionWorld;
+	btVector3 m_frictionDirectionWorld;
+	btScalar m_jacDiagABInv;
+	btScalar m_maxImpulse;
 
-
-	btWheelContactPoint(btRigidBody* body0,btRigidBody* body1,const btVector3& frictionPosWorld,const btVector3& frictionDirectionWorld, btScalar maxImpulse)
-		:m_body0(body0),
-		m_body1(body1),
-		m_frictionPositionWorld(frictionPosWorld),
-		m_frictionDirectionWorld(frictionDirectionWorld),
-		m_maxImpulse(maxImpulse)
+	btWheelContactPoint(btRigidBody* body0, btRigidBody* body1, const btVector3& frictionPosWorld, const btVector3& frictionDirectionWorld, btScalar maxImpulse)
+		: m_body0(body0),
+		  m_body1(body1),
+		  m_frictionPositionWorld(frictionPosWorld),
+		  m_frictionDirectionWorld(frictionDirectionWorld),
+		  m_maxImpulse(maxImpulse)
 	{
-		btScalar denom0 = body0->computeImpulseDenominator(frictionPosWorld,frictionDirectionWorld);
-		btScalar denom1 = body1->computeImpulseDenominator(frictionPosWorld,frictionDirectionWorld);
-		btScalar	relaxation = 1.f;
-		m_jacDiagABInv = relaxation/(denom0+denom1);
+		btScalar denom0 = body0->computeImpulseDenominator(frictionPosWorld, frictionDirectionWorld);
+		btScalar denom1 = body1->computeImpulseDenominator(frictionPosWorld, frictionDirectionWorld);
+		btScalar relaxation = 1.f;
+		m_jacDiagABInv = relaxation / (denom0 + denom1);
 	}
-
-
-
 };
 
 btScalar calcRollingFriction(btWheelContactPoint& contactPoint, int numWheelsOnGround);
 btScalar calcRollingFriction(btWheelContactPoint& contactPoint, int numWheelsOnGround)
 {
-
-	btScalar j1=0.f;
+	btScalar j1 = 0.f;
 
 	const btVector3& contactPosWorld = contactPoint.m_frictionPositionWorld;
 
-	btVector3 rel_pos1 = contactPosWorld - contactPoint.m_body0->getCenterOfMassPosition(); 
+	btVector3 rel_pos1 = contactPosWorld - contactPoint.m_body0->getCenterOfMassPosition();
 	btVector3 rel_pos2 = contactPosWorld - contactPoint.m_body1->getCenterOfMassPosition();
-	
-	btScalar maxImpulse  = contactPoint.m_maxImpulse;
-	
+
+	btScalar maxImpulse = contactPoint.m_maxImpulse;
+
 	btVector3 vel1 = contactPoint.m_body0->getVelocityInLocalPoint(rel_pos1);
 	btVector3 vel2 = contactPoint.m_body1->getVelocityInLocalPoint(rel_pos2);
 	btVector3 vel = vel1 - vel2;
@@ -520,253 +477,225 @@ btScalar calcRollingFriction(btWheelContactPoint& contactPoint, int numWheelsOnG
 	btScalar vrel = contactPoint.m_frictionDirectionWorld.dot(vel);
 
 	// calculate j that moves us to zero relative velocity
-	j1 = -vrel * contactPoint.m_jacDiagABInv/btScalar(numWheelsOnGround);
+	j1 = -vrel * contactPoint.m_jacDiagABInv / btScalar(numWheelsOnGround);
 	btSetMin(j1, maxImpulse);
 	btSetMax(j1, -maxImpulse);
 
 	return j1;
 }
 
-
-
-
 btScalar sideFrictionStiffness2 = btScalar(1.0);
-void	btRaycastVehicle::updateFriction(btScalar	timeStep)
+void btRaycastVehicle::updateFriction(btScalar timeStep)
 {
+	//calculate the impulse, so that the wheels don't move sidewards
+	int numWheel = getNumWheels();
+	if (!numWheel)
+		return;
 
-		//calculate the impulse, so that the wheels don't move sidewards
-		int numWheel = getNumWheels();
-		if (!numWheel)
-			return;
+	m_forwardWS.resize(numWheel);
+	m_axle.resize(numWheel);
+	m_forwardImpulse.resize(numWheel);
+	m_sideImpulse.resize(numWheel);
 
-		m_forwardWS.resize(numWheel);
-		m_axle.resize(numWheel);
-		m_forwardImpulse.resize(numWheel);
-		m_sideImpulse.resize(numWheel);
-		
-		int numWheelsOnGround = 0;
-	
+	int numWheelsOnGround = 0;
 
-		//collapse all those loops into one!
-		for (int i=0;i<getNumWheels();i++)
+	//collapse all those loops into one!
+	for (int i = 0; i < getNumWheels(); i++)
+	{
+		btWheelInfo& wheelInfo = m_wheelInfo[i];
+		class btRigidBody* groundObject = (class btRigidBody*)wheelInfo.m_raycastInfo.m_groundObject;
+		if (groundObject)
+			numWheelsOnGround++;
+		m_sideImpulse[i] = btScalar(0.);
+		m_forwardImpulse[i] = btScalar(0.);
+	}
+
+	{
+		for (int i = 0; i < getNumWheels(); i++)
 		{
 			btWheelInfo& wheelInfo = m_wheelInfo[i];
-			class btRigidBody* groundObject = (class btRigidBody*) wheelInfo.m_raycastInfo.m_groundObject;
+
+			class btRigidBody* groundObject = (class btRigidBody*)wheelInfo.m_raycastInfo.m_groundObject;
+
 			if (groundObject)
-				numWheelsOnGround++;
-			m_sideImpulse[i] = btScalar(0.);
-			m_forwardImpulse[i] = btScalar(0.);
-
-		}
-	
-		{
-	
-			for (int i=0;i<getNumWheels();i++)
 			{
+				const btTransform& wheelTrans = getWheelTransformWS(i);
 
-				btWheelInfo& wheelInfo = m_wheelInfo[i];
-					
-				class btRigidBody* groundObject = (class btRigidBody*) wheelInfo.m_raycastInfo.m_groundObject;
+				btMatrix3x3 wheelBasis0 = wheelTrans.getBasis();
+				m_axle[i] = -btVector3(
+					wheelBasis0[0][m_indexRightAxis],
+					wheelBasis0[1][m_indexRightAxis],
+					wheelBasis0[2][m_indexRightAxis]);
 
-				if (groundObject)
-				{
+				const btVector3& surfNormalWS = wheelInfo.m_raycastInfo.m_contactNormalWS;
+				btScalar proj = m_axle[i].dot(surfNormalWS);
+				m_axle[i] -= surfNormalWS * proj;
+				m_axle[i] = m_axle[i].normalize();
 
-					const btTransform& wheelTrans = getWheelTransformWS( i );
+				m_forwardWS[i] = surfNormalWS.cross(m_axle[i]);
+				m_forwardWS[i].normalize();
 
-					btMatrix3x3 wheelBasis0 = wheelTrans.getBasis();
-					m_axle[i] = -btVector3(	
-						wheelBasis0[0][m_indexRightAxis],
-						wheelBasis0[1][m_indexRightAxis],
-						wheelBasis0[2][m_indexRightAxis]);
-					
-					const btVector3& surfNormalWS = wheelInfo.m_raycastInfo.m_contactNormalWS;
-					btScalar proj = m_axle[i].dot(surfNormalWS);
-					m_axle[i] -= surfNormalWS * proj;
-					m_axle[i] = m_axle[i].normalize();
-					
-					m_forwardWS[i] = surfNormalWS.cross(m_axle[i]);
-					m_forwardWS[i].normalize();
-
-				
-					resolveSingleBilateral(*m_chassisBody, wheelInfo.m_raycastInfo.m_contactPointWS,
-							  *groundObject, wheelInfo.m_raycastInfo.m_contactPointWS,
-							  btScalar(0.), m_axle[i],m_sideImpulse[i],timeStep);
-
-					m_sideImpulse[i] *= sideFrictionStiffness2;
-						
-				}
-				
+				resolveSingleBilateral(*m_chassisBody, wheelInfo.m_raycastInfo.m_contactPointWS,
+									   *groundObject, wheelInfo.m_raycastInfo.m_contactPointWS,
+									   btScalar(0.), m_axle[i], m_sideImpulse[i], timeStep);
 
+				m_sideImpulse[i] *= sideFrictionStiffness2;
 			}
 		}
+	}
 
 	btScalar sideFactor = btScalar(1.);
 	btScalar fwdFactor = 0.5;
 
 	bool sliding = false;
 	{
-		for (int wheel =0;wheel <getNumWheels();wheel++)
+		for (int wheel = 0; wheel < getNumWheels(); wheel++)
 		{
 			btWheelInfo& wheelInfo = m_wheelInfo[wheel];
-			class btRigidBody* groundObject = (class btRigidBody*) wheelInfo.m_raycastInfo.m_groundObject;
+			class btRigidBody* groundObject = (class btRigidBody*)wheelInfo.m_raycastInfo.m_groundObject;
 
-			btScalar	rollingFriction = 0.f;
+			btScalar rollingFriction = 0.f;
 
 			if (groundObject)
 			{
 				if (wheelInfo.m_engineForce != 0.f)
 				{
-					rollingFriction = wheelInfo.m_engineForce* timeStep;
-				} else
+					rollingFriction = wheelInfo.m_engineForce * timeStep;
+				}
+				else
 				{
 					btScalar defaultRollingFrictionImpulse = 0.f;
 					btScalar maxImpulse = wheelInfo.m_brake ? wheelInfo.m_brake : defaultRollingFrictionImpulse;
-					btWheelContactPoint contactPt(m_chassisBody,groundObject,wheelInfo.m_raycastInfo.m_contactPointWS,m_forwardWS[wheel],maxImpulse);
+					btWheelContactPoint contactPt(m_chassisBody, groundObject, wheelInfo.m_raycastInfo.m_contactPointWS, m_forwardWS[wheel], maxImpulse);
 					btAssert(numWheelsOnGround > 0);
 					rollingFriction = calcRollingFriction(contactPt, numWheelsOnGround);
 				}
 			}
 
 			//switch between active rolling (throttle), braking and non-active rolling friction (no throttle/break)
-			
-
-
 
 			m_forwardImpulse[wheel] = btScalar(0.);
-			m_wheelInfo[wheel].m_skidInfo= btScalar(1.);
+			m_wheelInfo[wheel].m_skidInfo = btScalar(1.);
 
 			if (groundObject)
 			{
-				m_wheelInfo[wheel].m_skidInfo= btScalar(1.);
-				
+				m_wheelInfo[wheel].m_skidInfo = btScalar(1.);
+
 				btScalar maximp = wheelInfo.m_wheelsSuspensionForce * timeStep * wheelInfo.m_frictionSlip;
 				btScalar maximpSide = maximp;
 
 				btScalar maximpSquared = maximp * maximpSide;
-			
 
-				m_forwardImpulse[wheel] = rollingFriction;//wheelInfo.m_engineForce* timeStep;
+				m_forwardImpulse[wheel] = rollingFriction;  //wheelInfo.m_engineForce* timeStep;
 
-				btScalar x = (m_forwardImpulse[wheel] ) * fwdFactor;
-				btScalar y = (m_sideImpulse[wheel] ) * sideFactor;
-				
-				btScalar impulseSquared = (x*x + y*y);
+				btScalar x = (m_forwardImpulse[wheel]) * fwdFactor;
+				btScalar y = (m_sideImpulse[wheel]) * sideFactor;
+
+				btScalar impulseSquared = (x * x + y * y);
 
 				if (impulseSquared > maximpSquared)
 				{
 					sliding = true;
-					
+
 					btScalar factor = maximp / btSqrt(impulseSquared);
-					
+
 					m_wheelInfo[wheel].m_skidInfo *= factor;
 				}
-			} 
-
+			}
 		}
 	}
 
-	
-
-
-		if (sliding)
+	if (sliding)
+	{
+		for (int wheel = 0; wheel < getNumWheels(); wheel++)
 		{
-			for (int wheel = 0;wheel < getNumWheels(); wheel++)
+			if (m_sideImpulse[wheel] != btScalar(0.))
 			{
-				if (m_sideImpulse[wheel] != btScalar(0.))
+				if (m_wheelInfo[wheel].m_skidInfo < btScalar(1.))
 				{
-					if (m_wheelInfo[wheel].m_skidInfo< btScalar(1.))
-					{
-						m_forwardImpulse[wheel] *=	m_wheelInfo[wheel].m_skidInfo;
-						m_sideImpulse[wheel] *= m_wheelInfo[wheel].m_skidInfo;
-					}
+					m_forwardImpulse[wheel] *= m_wheelInfo[wheel].m_skidInfo;
+					m_sideImpulse[wheel] *= m_wheelInfo[wheel].m_skidInfo;
 				}
 			}
 		}
+	}
 
-		// apply the impulses
+	// apply the impulses
+	{
+		for (int wheel = 0; wheel < getNumWheels(); wheel++)
 		{
-			for (int wheel = 0;wheel<getNumWheels() ; wheel++)
+			btWheelInfo& wheelInfo = m_wheelInfo[wheel];
+
+			btVector3 rel_pos = wheelInfo.m_raycastInfo.m_contactPointWS -
+								m_chassisBody->getCenterOfMassPosition();
+
+			if (m_forwardImpulse[wheel] != btScalar(0.))
 			{
-				btWheelInfo& wheelInfo = m_wheelInfo[wheel];
+				m_chassisBody->applyImpulse(m_forwardWS[wheel] * (m_forwardImpulse[wheel]), rel_pos);
+			}
+			if (m_sideImpulse[wheel] != btScalar(0.))
+			{
+				class btRigidBody* groundObject = (class btRigidBody*)m_wheelInfo[wheel].m_raycastInfo.m_groundObject;
 
-				btVector3 rel_pos = wheelInfo.m_raycastInfo.m_contactPointWS - 
-						m_chassisBody->getCenterOfMassPosition();
+				btVector3 rel_pos2 = wheelInfo.m_raycastInfo.m_contactPointWS -
+									 groundObject->getCenterOfMassPosition();
 
-				if (m_forwardImpulse[wheel] != btScalar(0.))
-				{
-					m_chassisBody->applyImpulse(m_forwardWS[wheel]*(m_forwardImpulse[wheel]),rel_pos);
-				}
-				if (m_sideImpulse[wheel] != btScalar(0.))
-				{
-					class btRigidBody* groundObject = (class btRigidBody*) m_wheelInfo[wheel].m_raycastInfo.m_groundObject;
+				btVector3 sideImp = m_axle[wheel] * m_sideImpulse[wheel];
 
-					btVector3 rel_pos2 = wheelInfo.m_raycastInfo.m_contactPointWS - 
-						groundObject->getCenterOfMassPosition();
-
-					
-					btVector3 sideImp = m_axle[wheel] * m_sideImpulse[wheel];
-
-#if defined ROLLING_INFLUENCE_FIX // fix. It only worked if car's up was along Y - VT.
-					btVector3 vChassisWorldUp = getRigidBody()->getCenterOfMassTransform().getBasis().getColumn(m_indexUpAxis);
-					rel_pos -= vChassisWorldUp * (vChassisWorldUp.dot(rel_pos) * (1.f-wheelInfo.m_rollInfluence));
+#if defined ROLLING_INFLUENCE_FIX  // fix. It only worked if car's up was along Y - VT.
+				btVector3 vChassisWorldUp = getRigidBody()->getCenterOfMassTransform().getBasis().getColumn(m_indexUpAxis);
+				rel_pos -= vChassisWorldUp * (vChassisWorldUp.dot(rel_pos) * (1.f - wheelInfo.m_rollInfluence));
 #else
-					rel_pos[m_indexUpAxis] *= wheelInfo.m_rollInfluence;
+				rel_pos[m_indexUpAxis] *= wheelInfo.m_rollInfluence;
 #endif
-					m_chassisBody->applyImpulse(sideImp,rel_pos);
+				m_chassisBody->applyImpulse(sideImp, rel_pos);
 
-					//apply friction impulse on the ground
-					groundObject->applyImpulse(-sideImp,rel_pos2);
-				}
+				//apply friction impulse on the ground
+				groundObject->applyImpulse(-sideImp, rel_pos2);
 			}
 		}
-
-	
+	}
 }
 
-
-
-void	btRaycastVehicle::debugDraw(btIDebugDraw* debugDrawer)
+void btRaycastVehicle::debugDraw(btIDebugDraw* debugDrawer)
 {
-
-	for (int v=0;v<this->getNumWheels();v++)
+	for (int v = 0; v < this->getNumWheels(); v++)
 	{
-		btVector3 wheelColor(0,1,1);
+		btVector3 wheelColor(0, 1, 1);
 		if (getWheelInfo(v).m_raycastInfo.m_isInContact)
 		{
-			wheelColor.setValue(0,0,1);
-		} else
+			wheelColor.setValue(0, 0, 1);
+		}
+		else
 		{
-			wheelColor.setValue(1,0,1);
+			wheelColor.setValue(1, 0, 1);
 		}
 
 		btVector3 wheelPosWS = getWheelInfo(v).m_worldTransform.getOrigin();
 
-		btVector3 axle = btVector3(	
+		btVector3 axle = btVector3(
 			getWheelInfo(v).m_worldTransform.getBasis()[0][getRightAxis()],
 			getWheelInfo(v).m_worldTransform.getBasis()[1][getRightAxis()],
 			getWheelInfo(v).m_worldTransform.getBasis()[2][getRightAxis()]);
 
 		//debug wheels (cylinders)
-		debugDrawer->drawLine(wheelPosWS,wheelPosWS+axle,wheelColor);
-		debugDrawer->drawLine(wheelPosWS,getWheelInfo(v).m_raycastInfo.m_contactPointWS,wheelColor);
-
+		debugDrawer->drawLine(wheelPosWS, wheelPosWS + axle, wheelColor);
+		debugDrawer->drawLine(wheelPosWS, getWheelInfo(v).m_raycastInfo.m_contactPointWS, wheelColor);
 	}
 }
 
-
-void* btDefaultVehicleRaycaster::castRay(const btVector3& from,const btVector3& to, btVehicleRaycasterResult& result)
+void* btDefaultVehicleRaycaster::castRay(const btVector3& from, const btVector3& to, btVehicleRaycasterResult& result)
 {
-//	RayResultCallback& resultCallback;
+	//	RayResultCallback& resultCallback;
 
-	btCollisionWorld::ClosestRayResultCallback rayCallback(from,to);
+	btCollisionWorld::ClosestRayResultCallback rayCallback(from, to);
 
 	m_dynamicsWorld->rayTest(from, to, rayCallback);
 
 	if (rayCallback.hasHit())
 	{
-		
 		const btRigidBody* body = btRigidBody::upcast(rayCallback.m_collisionObject);
-        if (body && body->hasContactResponse())
+		if (body && body->hasContactResponse())
 		{
 			result.m_hitPointInWorld = rayCallback.m_hitPointWorld;
 			result.m_hitNormalInWorld = rayCallback.m_hitNormalWorld;
@@ -777,4 +706,3 @@ void* btDefaultVehicleRaycaster::castRay(const btVector3& from,const btVector3&
 	}
 	return 0;
 }
-
diff --git a/thirdparty/bullet/BulletDynamics/Vehicle/btRaycastVehicle.h b/thirdparty/bullet/BulletDynamics/Vehicle/btRaycastVehicle.h
index 04656b912c7..99d6894e56e 100644
--- a/thirdparty/bullet/BulletDynamics/Vehicle/btRaycastVehicle.h
+++ b/thirdparty/bullet/BulletDynamics/Vehicle/btRaycastVehicle.h
@@ -24,122 +24,111 @@ class btDynamicsWorld;
 ///rayCast vehicle, very special constraint that turn a rigidbody into a vehicle.
 class btRaycastVehicle : public btActionInterface
 {
+	btAlignedObjectArray<btVector3> m_forwardWS;
+	btAlignedObjectArray<btVector3> m_axle;
+	btAlignedObjectArray<btScalar> m_forwardImpulse;
+	btAlignedObjectArray<btScalar> m_sideImpulse;
 
-		btAlignedObjectArray<btVector3>	m_forwardWS;
-		btAlignedObjectArray<btVector3>	m_axle;
-		btAlignedObjectArray<btScalar>	m_forwardImpulse;
-		btAlignedObjectArray<btScalar>	m_sideImpulse;
-	
-		///backwards compatibility
-		int	m_userConstraintType;
-		int	m_userConstraintId;
+	///backwards compatibility
+	int m_userConstraintType;
+	int m_userConstraintId;
 
 public:
 	class btVehicleTuning
+	{
+	public:
+		btVehicleTuning()
+			: m_suspensionStiffness(btScalar(5.88)),
+			  m_suspensionCompression(btScalar(0.83)),
+			  m_suspensionDamping(btScalar(0.88)),
+			  m_maxSuspensionTravelCm(btScalar(500.)),
+			  m_frictionSlip(btScalar(10.5)),
+			  m_maxSuspensionForce(btScalar(6000.))
 		{
-			public:
+		}
+		btScalar m_suspensionStiffness;
+		btScalar m_suspensionCompression;
+		btScalar m_suspensionDamping;
+		btScalar m_maxSuspensionTravelCm;
+		btScalar m_frictionSlip;
+		btScalar m_maxSuspensionForce;
+	};
 
-			btVehicleTuning()
-				:m_suspensionStiffness(btScalar(5.88)),
-				m_suspensionCompression(btScalar(0.83)),
-				m_suspensionDamping(btScalar(0.88)),
-				m_maxSuspensionTravelCm(btScalar(500.)),
-				m_frictionSlip(btScalar(10.5)),
-				m_maxSuspensionForce(btScalar(6000.))
-			{
-			}
-			btScalar	m_suspensionStiffness;
-			btScalar	m_suspensionCompression;
-			btScalar	m_suspensionDamping;
-			btScalar	m_maxSuspensionTravelCm;
-			btScalar	m_frictionSlip;
-			btScalar	m_maxSuspensionForce;
-
-		};
 private:
-
-	btVehicleRaycaster*	m_vehicleRaycaster;
-	btScalar		m_pitchControl;
-	btScalar	m_steeringValue; 
+	btVehicleRaycaster* m_vehicleRaycaster;
+	btScalar m_pitchControl;
+	btScalar m_steeringValue;
 	btScalar m_currentVehicleSpeedKmHour;
 
 	btRigidBody* m_chassisBody;
 
 	int m_indexRightAxis;
 	int m_indexUpAxis;
-	int	m_indexForwardAxis;
+	int m_indexForwardAxis;
 
 	void defaultInit(const btVehicleTuning& tuning);
 
 public:
-
 	//constructor to create a car from an existing rigidbody
-	btRaycastVehicle(const btVehicleTuning& tuning,btRigidBody* chassis,	btVehicleRaycaster* raycaster );
-
-	virtual ~btRaycastVehicle() ;
+	btRaycastVehicle(const btVehicleTuning& tuning, btRigidBody* chassis, btVehicleRaycaster* raycaster);
 
+	virtual ~btRaycastVehicle();
 
 	///btActionInterface interface
-	virtual void updateAction( btCollisionWorld* collisionWorld, btScalar step)
+	virtual void updateAction(btCollisionWorld* collisionWorld, btScalar step)
 	{
-        (void) collisionWorld;
+		(void)collisionWorld;
 		updateVehicle(step);
 	}
-	
 
 	///btActionInterface interface
-	void	debugDraw(btIDebugDraw* debugDrawer);
-			
+	void debugDraw(btIDebugDraw* debugDrawer);
+
 	const btTransform& getChassisWorldTransform() const;
-	
+
 	btScalar rayCast(btWheelInfo& wheel);
 
 	virtual void updateVehicle(btScalar step);
-	
-	
+
 	void resetSuspension();
 
-	btScalar	getSteeringValue(int wheel) const;
+	btScalar getSteeringValue(int wheel) const;
 
-	void	setSteeringValue(btScalar steering,int wheel);
+	void setSteeringValue(btScalar steering, int wheel);
 
+	void applyEngineForce(btScalar force, int wheel);
 
-	void	applyEngineForce(btScalar force, int wheel);
+	const btTransform& getWheelTransformWS(int wheelIndex) const;
 
-	const btTransform&	getWheelTransformWS( int wheelIndex ) const;
+	void updateWheelTransform(int wheelIndex, bool interpolatedTransform = true);
 
-	void	updateWheelTransform( int wheelIndex, bool interpolatedTransform = true );
-	
-//	void	setRaycastWheelInfo( int wheelIndex , bool isInContact, const btVector3& hitPoint, const btVector3& hitNormal,btScalar depth);
+	//	void	setRaycastWheelInfo( int wheelIndex , bool isInContact, const btVector3& hitPoint, const btVector3& hitNormal,btScalar depth);
 
-	btWheelInfo&	addWheel( const btVector3& connectionPointCS0, const btVector3& wheelDirectionCS0,const btVector3& wheelAxleCS,btScalar suspensionRestLength,btScalar wheelRadius,const btVehicleTuning& tuning, bool isFrontWheel);
+	btWheelInfo& addWheel(const btVector3& connectionPointCS0, const btVector3& wheelDirectionCS0, const btVector3& wheelAxleCS, btScalar suspensionRestLength, btScalar wheelRadius, const btVehicleTuning& tuning, bool isFrontWheel);
 
-	inline int		getNumWheels() const {
-		return int (m_wheelInfo.size());
+	inline int getNumWheels() const
+	{
+		return int(m_wheelInfo.size());
 	}
-	
-	btAlignedObjectArray<btWheelInfo>	m_wheelInfo;
 
+	btAlignedObjectArray<btWheelInfo> m_wheelInfo;
 
-	const btWheelInfo&	getWheelInfo(int index) const;
+	const btWheelInfo& getWheelInfo(int index) const;
 
-	btWheelInfo&	getWheelInfo(int index);
+	btWheelInfo& getWheelInfo(int index);
 
-	void	updateWheelTransformsWS(btWheelInfo& wheel , bool interpolatedTransform = true);
+	void updateWheelTransformsWS(btWheelInfo& wheel, bool interpolatedTransform = true);
 
-	
-	void setBrake(btScalar brake,int wheelIndex);
+	void setBrake(btScalar brake, int wheelIndex);
 
-	void	setPitchControl(btScalar pitch)
+	void setPitchControl(btScalar pitch)
 	{
 		m_pitchControl = pitch;
 	}
-	
-	void	updateSuspension(btScalar deltaTime);
-
-	virtual void	updateFriction(btScalar	timeStep);
 
+	void updateSuspension(btScalar deltaTime);
 
+	virtual void updateFriction(btScalar timeStep);
 
 	inline btRigidBody* getRigidBody()
 	{
@@ -151,7 +140,7 @@ public:
 		return m_chassisBody;
 	}
 
-	inline int	getRightAxis() const
+	inline int getRightAxis() const
 	{
 		return m_indexRightAxis;
 	}
@@ -165,46 +154,44 @@ public:
 		return m_indexForwardAxis;
 	}
 
-	
 	///Worldspace forward vector
 	btVector3 getForwardVector() const
 	{
-		const btTransform& chassisTrans = getChassisWorldTransform(); 
+		const btTransform& chassisTrans = getChassisWorldTransform();
 
-		btVector3 forwardW ( 
-			  chassisTrans.getBasis()[0][m_indexForwardAxis], 
-			  chassisTrans.getBasis()[1][m_indexForwardAxis], 
-			  chassisTrans.getBasis()[2][m_indexForwardAxis]); 
+		btVector3 forwardW(
+			chassisTrans.getBasis()[0][m_indexForwardAxis],
+			chassisTrans.getBasis()[1][m_indexForwardAxis],
+			chassisTrans.getBasis()[2][m_indexForwardAxis]);
 
 		return forwardW;
 	}
 
 	///Velocity of vehicle (positive if velocity vector has same direction as foward vector)
-	btScalar	getCurrentSpeedKmHour() const
+	btScalar getCurrentSpeedKmHour() const
 	{
 		return m_currentVehicleSpeedKmHour;
 	}
 
-	virtual void	setCoordinateSystem(int rightIndex,int upIndex,int forwardIndex)
+	virtual void setCoordinateSystem(int rightIndex, int upIndex, int forwardIndex)
 	{
 		m_indexRightAxis = rightIndex;
 		m_indexUpAxis = upIndex;
 		m_indexForwardAxis = forwardIndex;
 	}
 
-
 	///backwards compatibility
 	int getUserConstraintType() const
 	{
-		return m_userConstraintType ;
+		return m_userConstraintType;
 	}
 
-	void	setUserConstraintType(int userConstraintType)
+	void setUserConstraintType(int userConstraintType)
 	{
 		m_userConstraintType = userConstraintType;
 	};
 
-	void	setUserConstraintId(int uid)
+	void setUserConstraintId(int uid)
 	{
 		m_userConstraintId = uid;
 	}
@@ -213,22 +200,19 @@ public:
 	{
 		return m_userConstraintId;
 	}
-
 };
 
 class btDefaultVehicleRaycaster : public btVehicleRaycaster
 {
-	btDynamicsWorld*	m_dynamicsWorld;
+	btDynamicsWorld* m_dynamicsWorld;
+
 public:
 	btDefaultVehicleRaycaster(btDynamicsWorld* world)
-		:m_dynamicsWorld(world)
+		: m_dynamicsWorld(world)
 	{
 	}
 
-	virtual void* castRay(const btVector3& from,const btVector3& to, btVehicleRaycasterResult& result);
-
+	virtual void* castRay(const btVector3& from, const btVector3& to, btVehicleRaycasterResult& result);
 };
 
-
-#endif //BT_RAYCASTVEHICLE_H
-
+#endif  //BT_RAYCASTVEHICLE_H
diff --git a/thirdparty/bullet/BulletDynamics/Vehicle/btVehicleRaycaster.h b/thirdparty/bullet/BulletDynamics/Vehicle/btVehicleRaycaster.h
index 3cc909c6530..2c44ce546c7 100644
--- a/thirdparty/bullet/BulletDynamics/Vehicle/btVehicleRaycaster.h
+++ b/thirdparty/bullet/BulletDynamics/Vehicle/btVehicleRaycaster.h
@@ -16,20 +16,18 @@
 /// btVehicleRaycaster is provides interface for between vehicle simulation and raycasting
 struct btVehicleRaycaster
 {
-virtual ~btVehicleRaycaster()
-{
-}
+	virtual ~btVehicleRaycaster()
+	{
+	}
 	struct btVehicleRaycasterResult
 	{
-		btVehicleRaycasterResult() :m_distFraction(btScalar(-1.)){};
-		btVector3	m_hitPointInWorld;
-		btVector3	m_hitNormalInWorld;
-		btScalar	m_distFraction;
+		btVehicleRaycasterResult() : m_distFraction(btScalar(-1.)){};
+		btVector3 m_hitPointInWorld;
+		btVector3 m_hitNormalInWorld;
+		btScalar m_distFraction;
 	};
 
-	virtual void* castRay(const btVector3& from,const btVector3& to, btVehicleRaycasterResult& result) = 0;
-
+	virtual void* castRay(const btVector3& from, const btVector3& to, btVehicleRaycasterResult& result) = 0;
 };
 
-#endif //BT_VEHICLE_RAYCASTER_H
-
+#endif  //BT_VEHICLE_RAYCASTER_H
diff --git a/thirdparty/bullet/BulletDynamics/Vehicle/btWheelInfo.cpp b/thirdparty/bullet/BulletDynamics/Vehicle/btWheelInfo.cpp
index ef93c16fffc..d5c12f223b5 100644
--- a/thirdparty/bullet/BulletDynamics/Vehicle/btWheelInfo.cpp
+++ b/thirdparty/bullet/BulletDynamics/Vehicle/btWheelInfo.cpp
@@ -9,30 +9,26 @@
  * It is provided "as is" without express or implied warranty.
 */
 #include "btWheelInfo.h"
-#include "BulletDynamics/Dynamics/btRigidBody.h" // for pointvelocity
-
+#include "BulletDynamics/Dynamics/btRigidBody.h"  // for pointvelocity
 
 btScalar btWheelInfo::getSuspensionRestLength() const
 {
-
 	return m_suspensionRestLength1;
-
 }
 
-void	btWheelInfo::updateWheel(const btRigidBody& chassis,RaycastInfo& raycastInfo)
+void btWheelInfo::updateWheel(const btRigidBody& chassis, RaycastInfo& raycastInfo)
 {
 	(void)raycastInfo;
 
-	
 	if (m_raycastInfo.m_isInContact)
 
 	{
-		btScalar	project= m_raycastInfo.m_contactNormalWS.dot( m_raycastInfo.m_wheelDirectionWS );
-		btVector3	 chassis_velocity_at_contactPoint;
+		btScalar project = m_raycastInfo.m_contactNormalWS.dot(m_raycastInfo.m_wheelDirectionWS);
+		btVector3 chassis_velocity_at_contactPoint;
 		btVector3 relpos = m_raycastInfo.m_contactPointWS - chassis.getCenterOfMassPosition();
-		chassis_velocity_at_contactPoint = chassis.getVelocityInLocalPoint( relpos );
-		btScalar projVel = m_raycastInfo.m_contactNormalWS.dot( chassis_velocity_at_contactPoint );
-		if ( project >= btScalar(-0.1))
+		chassis_velocity_at_contactPoint = chassis.getVelocityInLocalPoint(relpos);
+		btScalar projVel = m_raycastInfo.m_contactNormalWS.dot(chassis_velocity_at_contactPoint);
+		if (project >= btScalar(-0.1))
 		{
 			m_suspensionRelativeVelocity = btScalar(0.0);
 			m_clippedInvContactDotSuspension = btScalar(1.0) / btScalar(0.1);
@@ -43,10 +39,9 @@ void	btWheelInfo::updateWheel(const btRigidBody& chassis,RaycastInfo& raycastInf
 			m_suspensionRelativeVelocity = projVel * inv;
 			m_clippedInvContactDotSuspension = inv;
 		}
-		
 	}
 
-	else	// Not in contact : position wheel in a nice (rest length) position
+	else  // Not in contact : position wheel in a nice (rest length) position
 	{
 		m_raycastInfo.m_suspensionLength = this->getSuspensionRestLength();
 		m_suspensionRelativeVelocity = btScalar(0.0);
diff --git a/thirdparty/bullet/BulletDynamics/Vehicle/btWheelInfo.h b/thirdparty/bullet/BulletDynamics/Vehicle/btWheelInfo.h
index f991a57b69e..af88b8ff837 100644
--- a/thirdparty/bullet/BulletDynamics/Vehicle/btWheelInfo.h
+++ b/thirdparty/bullet/BulletDynamics/Vehicle/btWheelInfo.h
@@ -18,20 +18,19 @@ class btRigidBody;
 
 struct btWheelInfoConstructionInfo
 {
-	btVector3	m_chassisConnectionCS;
-	btVector3	m_wheelDirectionCS;
-	btVector3	m_wheelAxleCS;
-	btScalar	m_suspensionRestLength;
-	btScalar	m_maxSuspensionTravelCm;
-	btScalar	m_wheelRadius;
-	
-	btScalar		m_suspensionStiffness;
-	btScalar		m_wheelsDampingCompression;
-	btScalar		m_wheelsDampingRelaxation;
-	btScalar		m_frictionSlip;
-	btScalar		m_maxSuspensionForce;
+	btVector3 m_chassisConnectionCS;
+	btVector3 m_wheelDirectionCS;
+	btVector3 m_wheelAxleCS;
+	btScalar m_suspensionRestLength;
+	btScalar m_maxSuspensionTravelCm;
+	btScalar m_wheelRadius;
+
+	btScalar m_suspensionStiffness;
+	btScalar m_wheelsDampingCompression;
+	btScalar m_wheelsDampingRelaxation;
+	btScalar m_frictionSlip;
+	btScalar m_maxSuspensionForce;
 	bool m_bIsFrontWheel;
-	
 };
 
 /// btWheelInfo contains information per wheel about friction and suspension.
@@ -40,51 +39,50 @@ struct btWheelInfo
 	struct RaycastInfo
 	{
 		//set by raycaster
-		btVector3	m_contactNormalWS;//contactnormal
-		btVector3	m_contactPointWS;//raycast hitpoint
-		btScalar	m_suspensionLength;
-		btVector3	m_hardPointWS;//raycast starting point
-		btVector3	m_wheelDirectionWS; //direction in worldspace
-		btVector3	m_wheelAxleWS; // axle in worldspace
-		bool		m_isInContact;
-		void*		m_groundObject; //could be general void* ptr
+		btVector3 m_contactNormalWS;  //contactnormal
+		btVector3 m_contactPointWS;   //raycast hitpoint
+		btScalar m_suspensionLength;
+		btVector3 m_hardPointWS;       //raycast starting point
+		btVector3 m_wheelDirectionWS;  //direction in worldspace
+		btVector3 m_wheelAxleWS;       // axle in worldspace
+		bool m_isInContact;
+		void* m_groundObject;  //could be general void* ptr
 	};
 
-	RaycastInfo	m_raycastInfo;
+	RaycastInfo m_raycastInfo;
 
-	btTransform	m_worldTransform;
-	
-	btVector3	m_chassisConnectionPointCS; //const
-	btVector3	m_wheelDirectionCS;//const
-	btVector3	m_wheelAxleCS; // const or modified by steering
-	btScalar	m_suspensionRestLength1;//const
-	btScalar	m_maxSuspensionTravelCm;
+	btTransform m_worldTransform;
+
+	btVector3 m_chassisConnectionPointCS;  //const
+	btVector3 m_wheelDirectionCS;          //const
+	btVector3 m_wheelAxleCS;               // const or modified by steering
+	btScalar m_suspensionRestLength1;      //const
+	btScalar m_maxSuspensionTravelCm;
 	btScalar getSuspensionRestLength() const;
-	btScalar	m_wheelsRadius;//const
-	btScalar	m_suspensionStiffness;//const
-	btScalar	m_wheelsDampingCompression;//const
-	btScalar	m_wheelsDampingRelaxation;//const
-	btScalar	m_frictionSlip;
-	btScalar	m_steering;
-	btScalar	m_rotation;
-	btScalar	m_deltaRotation;
-	btScalar	m_rollInfluence;
-	btScalar	m_maxSuspensionForce;
+	btScalar m_wheelsRadius;              //const
+	btScalar m_suspensionStiffness;       //const
+	btScalar m_wheelsDampingCompression;  //const
+	btScalar m_wheelsDampingRelaxation;   //const
+	btScalar m_frictionSlip;
+	btScalar m_steering;
+	btScalar m_rotation;
+	btScalar m_deltaRotation;
+	btScalar m_rollInfluence;
+	btScalar m_maxSuspensionForce;
 
-	btScalar	m_engineForce;
+	btScalar m_engineForce;
+
+	btScalar m_brake;
 
-	btScalar	m_brake;
-	
 	bool m_bIsFrontWheel;
-	
-	void*		m_clientInfo;//can be used to store pointer to sync transforms...
+
+	void* m_clientInfo;  //can be used to store pointer to sync transforms...
 
 	btWheelInfo() {}
 
 	btWheelInfo(btWheelInfoConstructionInfo& ci)
 
 	{
-
 		m_suspensionRestLength1 = ci.m_suspensionRestLength;
 		m_maxSuspensionTravelCm = ci.m_maxSuspensionTravelCm;
 
@@ -104,18 +102,15 @@ struct btWheelInfo
 		m_rollInfluence = btScalar(0.1);
 		m_bIsFrontWheel = ci.m_bIsFrontWheel;
 		m_maxSuspensionForce = ci.m_maxSuspensionForce;
-
 	}
 
-	void	updateWheel(const btRigidBody& chassis,RaycastInfo& raycastInfo);
+	void updateWheel(const btRigidBody& chassis, RaycastInfo& raycastInfo);
 
-	btScalar	m_clippedInvContactDotSuspension;
-	btScalar	m_suspensionRelativeVelocity;
+	btScalar m_clippedInvContactDotSuspension;
+	btScalar m_suspensionRelativeVelocity;
 	//calculated by suspension
-	btScalar	m_wheelsSuspensionForce;
-	btScalar	m_skidInfo;
-
+	btScalar m_wheelsSuspensionForce;
+	btScalar m_skidInfo;
 };
 
-#endif //BT_WHEEL_INFO_H
-
+#endif  //BT_WHEEL_INFO_H
diff --git a/thirdparty/bullet/BulletInverseDynamics/IDConfig.hpp b/thirdparty/bullet/BulletInverseDynamics/IDConfig.hpp
index ebb10e7a16e..b662b801528 100644
--- a/thirdparty/bullet/BulletInverseDynamics/IDConfig.hpp
+++ b/thirdparty/bullet/BulletInverseDynamics/IDConfig.hpp
@@ -32,10 +32,10 @@
 #define BT_ID_POW(x, y) btPow(x, y)
 #define BT_ID_PI SIMD_PI
 #ifdef _WIN32
-	#define BT_ID_SNPRINTF _snprintf
+#define BT_ID_SNPRINTF _snprintf
 #else
-	#define BT_ID_SNPRINTF snprintf
-#endif //
+#define BT_ID_SNPRINTF snprintf
+#endif  //
 #endif
 // error messages
 #include "IDErrorMessages.hpp"
@@ -52,8 +52,8 @@
 #error "custom inverse dynamics config, but no custom namespace defined"
 #endif
 
-#define BT_ID_MAX(a,b) std::max(a,b)
-#define BT_ID_MIN(a,b) std::min(a,b)
+#define BT_ID_MAX(a, b) std::max(a, b)
+#define BT_ID_MIN(a, b) std::min(a, b)
 
 #else
 #define btInverseDynamics btInverseDynamicsBullet3
@@ -62,8 +62,8 @@
 #include "LinearMath/btScalar.h"
 typedef btScalar idScalar;
 #include "LinearMath/btMinMax.h"
-#define BT_ID_MAX(a,b) btMax(a,b)
-#define BT_ID_MIN(a,b) btMin(a,b)
+#define BT_ID_MAX(a, b) btMax(a, b)
+#define BT_ID_MIN(a, b) btMin(a, b)
 
 #ifdef BT_USE_DOUBLE_PRECISION
 #define BT_ID_USE_DOUBLE_PRECISION
@@ -71,31 +71,31 @@ typedef btScalar idScalar;
 
 #ifndef BT_USE_INVERSE_DYNAMICS_WITH_BULLET2
 
-
 // use bullet types for arrays and array indices
 #include "Bullet3Common/b3AlignedObjectArray.h"
 // this is to make it work with C++2003, otherwise we could do this:
 // template <typename T>
 // using idArray = b3AlignedObjectArray<T>;
 template <typename T>
-struct idArray {
+struct idArray
+{
 	typedef b3AlignedObjectArray<T> type;
 };
 typedef int idArrayIdx;
 #define ID_DECLARE_ALIGNED_ALLOCATOR() B3_DECLARE_ALIGNED_ALLOCATOR()
 
-#else // BT_USE_INVERSE_DYNAMICS_WITH_BULLET2
+#else  // BT_USE_INVERSE_DYNAMICS_WITH_BULLET2
 
 #include "LinearMath/btAlignedObjectArray.h"
 template <typename T>
-struct idArray {
+struct idArray
+{
 	typedef btAlignedObjectArray<T> type;
 };
 typedef int idArrayIdx;
 #define ID_DECLARE_ALIGNED_ALLOCATOR() BT_DECLARE_ALIGNED_ALLOCATOR()
 
-#endif // BT_USE_INVERSE_DYNAMICS_WITH_BULLET2
-
+#endif  // BT_USE_INVERSE_DYNAMICS_WITH_BULLET2
 
 // use bullet's allocator functions
 #define idMalloc btAllocFunc
diff --git a/thirdparty/bullet/BulletInverseDynamics/IDConfigBuiltin.hpp b/thirdparty/bullet/BulletInverseDynamics/IDConfigBuiltin.hpp
index 130c19c6d6f..63923679245 100644
--- a/thirdparty/bullet/BulletInverseDynamics/IDConfigBuiltin.hpp
+++ b/thirdparty/bullet/BulletInverseDynamics/IDConfigBuiltin.hpp
@@ -14,7 +14,8 @@ typedef float idScalar;
 // template <typename T>
 // using idArray = std::vector<T>;
 template <typename T>
-struct idArray {
+struct idArray
+{
 	typedef std::vector<T> type;
 };
 typedef std::vector<int>::size_type idArrayIdx;
@@ -23,14 +24,14 @@ typedef std::vector<int>::size_type idArrayIdx;
 #define idMalloc ::malloc
 #define idFree ::free
 // currently not aligned at all...
-#define ID_DECLARE_ALIGNED_ALLOCATOR()															 \
-	inline void* operator new(std::size_t sizeInBytes) { return idMalloc(sizeInBytes); }		   \
-	inline void operator delete(void* ptr) { idFree(ptr); }										\
-	inline void* operator new(std::size_t, void* ptr) { return ptr; }							  \
-	inline void operator delete(void*, void*) {}												   \
-	inline void* operator new[](std::size_t sizeInBytes) { return idMalloc(sizeInBytes); }		 \
-	inline void operator delete[](void* ptr) { idFree(ptr); }									  \
-	inline void* operator new[](std::size_t, void* ptr) { return ptr; }							\
+#define ID_DECLARE_ALIGNED_ALLOCATOR()                                                     \
+	inline void* operator new(std::size_t sizeInBytes) { return idMalloc(sizeInBytes); }   \
+	inline void operator delete(void* ptr) { idFree(ptr); }                                \
+	inline void* operator new(std::size_t, void* ptr) { return ptr; }                      \
+	inline void operator delete(void*, void*) {}                                           \
+	inline void* operator new[](std::size_t sizeInBytes) { return idMalloc(sizeInBytes); } \
+	inline void operator delete[](void* ptr) { idFree(ptr); }                              \
+	inline void* operator new[](std::size_t, void* ptr) { return ptr; }                    \
 	inline void operator delete[](void*, void*) {}
 
 #include "details/IDMatVec.hpp"
diff --git a/thirdparty/bullet/BulletInverseDynamics/IDConfigEigen.hpp b/thirdparty/bullet/BulletInverseDynamics/IDConfigEigen.hpp
index cbd7e8a9c4e..cfb308ee554 100644
--- a/thirdparty/bullet/BulletInverseDynamics/IDConfigEigen.hpp
+++ b/thirdparty/bullet/BulletInverseDynamics/IDConfigEigen.hpp
@@ -15,7 +15,8 @@ typedef float idScalar;
 // template <typename T>
 // using idArray = std::vector<T>;
 template <typename T>
-struct idArray {
+struct idArray
+{
 	typedef std::vector<T> type;
 };
 typedef std::vector<int>::size_type idArrayIdx;
diff --git a/thirdparty/bullet/BulletInverseDynamics/IDErrorMessages.hpp b/thirdparty/bullet/BulletInverseDynamics/IDErrorMessages.hpp
index 1b3fd268a0b..5a98f014987 100644
--- a/thirdparty/bullet/BulletInverseDynamics/IDErrorMessages.hpp
+++ b/thirdparty/bullet/BulletInverseDynamics/IDErrorMessages.hpp
@@ -13,16 +13,18 @@
 #else  // BT_ID_WO_BULLET
 #include <cstdio>
 /// print error message with file/line information
-#define bt_id_error_message(...)																		 \
-	do {																						   \
-		fprintf(stderr, "[Error:%s:%d] ", __INVDYN_FILE_WO_DIR__, __LINE__);					   \
-		fprintf(stderr, __VA_ARGS__);															  \
+#define bt_id_error_message(...)                                             \
+	do                                                                       \
+	{                                                                        \
+		fprintf(stderr, "[Error:%s:%d] ", __INVDYN_FILE_WO_DIR__, __LINE__); \
+		fprintf(stderr, __VA_ARGS__);                                        \
 	} while (0)
 /// print warning message with file/line information
-#define bt_id_warning_message(...)																	   \
-	do {																						   \
-		fprintf(stderr, "[Warning:%s:%d] ", __INVDYN_FILE_WO_DIR__, __LINE__);					 \
-		fprintf(stderr, __VA_ARGS__);															  \
+#define bt_id_warning_message(...)                                             \
+	do                                                                         \
+	{                                                                          \
+		fprintf(stderr, "[Warning:%s:%d] ", __INVDYN_FILE_WO_DIR__, __LINE__); \
+		fprintf(stderr, __VA_ARGS__);                                          \
 	} while (0)
 #define id_printf(...) printf(__VA_ARGS__)
 #endif  // BT_ID_WO_BULLET
diff --git a/thirdparty/bullet/BulletInverseDynamics/IDMath.cpp b/thirdparty/bullet/BulletInverseDynamics/IDMath.cpp
index d279d3435c6..2f120ed489d 100644
--- a/thirdparty/bullet/BulletInverseDynamics/IDMath.cpp
+++ b/thirdparty/bullet/BulletInverseDynamics/IDMath.cpp
@@ -3,25 +3,30 @@
 #include <cmath>
 #include <limits>
 
-namespace btInverseDynamics {
+namespace btInverseDynamics
+{
 static const idScalar kIsZero = 5 * std::numeric_limits<idScalar>::epsilon();
 // requirements for axis length deviation from 1.0
 // experimentally set from random euler angle rotation matrices
 static const idScalar kAxisLengthEpsilon = 10 * kIsZero;
 
-void setZero(vec3 &v) {
+void setZero(vec3 &v)
+{
 	v(0) = 0;
 	v(1) = 0;
 	v(2) = 0;
 }
 
-void setZero(vecx &v) {
-	for (int i = 0; i < v.size(); i++) {
+void setZero(vecx &v)
+{
+	for (int i = 0; i < v.size(); i++)
+	{
 		v(i) = 0;
 	}
 }
 
-void setZero(mat33 &m) {
+void setZero(mat33 &m)
+{
 	m(0, 0) = 0;
 	m(0, 1) = 0;
 	m(0, 2) = 0;
@@ -33,7 +38,8 @@ void setZero(mat33 &m) {
 	m(2, 2) = 0;
 }
 
-void skew(vec3& v, mat33* result) {
+void skew(vec3 &v, mat33 *result)
+{
 	(*result)(0, 0) = 0.0;
 	(*result)(0, 1) = -v(2);
 	(*result)(0, 2) = v(1);
@@ -45,22 +51,28 @@ void skew(vec3& v, mat33* result) {
 	(*result)(2, 2) = 0.0;
 }
 
-idScalar maxAbs(const vecx &v) {
+idScalar maxAbs(const vecx &v)
+{
 	idScalar result = 0.0;
-	for (int i = 0; i < v.size(); i++) {
+	for (int i = 0; i < v.size(); i++)
+	{
 		const idScalar tmp = BT_ID_FABS(v(i));
-		if (tmp > result) {
+		if (tmp > result)
+		{
 			result = tmp;
 		}
 	}
 	return result;
 }
 
-idScalar maxAbs(const vec3 &v) {
+idScalar maxAbs(const vec3 &v)
+{
 	idScalar result = 0.0;
-	for (int i = 0; i < 3; i++) {
+	for (int i = 0; i < 3; i++)
+	{
 		const idScalar tmp = BT_ID_FABS(v(i));
-		if (tmp > result) {
+		if (tmp > result)
+		{
 			result = tmp;
 		}
 	}
@@ -68,60 +80,75 @@ idScalar maxAbs(const vec3 &v) {
 }
 
 #if (defined BT_ID_HAVE_MAT3X)
-idScalar maxAbsMat3x(const mat3x &m) {
-    // only used for tests -- so just loop here for portability
-    idScalar result = 0.0;
-    for (idArrayIdx col = 0; col < m.cols(); col++) {
-        for (idArrayIdx row = 0; row < 3; row++) {
-            result = BT_ID_MAX(result, std::fabs(m(row, col)));
-        }
-    }
-    return result;
+idScalar maxAbsMat3x(const mat3x &m)
+{
+	// only used for tests -- so just loop here for portability
+	idScalar result = 0.0;
+	for (idArrayIdx col = 0; col < m.cols(); col++)
+	{
+		for (idArrayIdx row = 0; row < 3; row++)
+		{
+			result = BT_ID_MAX(result, std::fabs(m(row, col)));
+		}
+	}
+	return result;
 }
 
-void mul(const mat33 &a, const mat3x &b, mat3x *result) {
-    if (b.cols() != result->cols()) {
-        bt_id_error_message("size missmatch. b.cols()= %d, result->cols()= %d\n",
-                      static_cast<int>(b.cols()), static_cast<int>(result->cols()));
-        abort();
-    }
+void mul(const mat33 &a, const mat3x &b, mat3x *result)
+{
+	if (b.cols() != result->cols())
+	{
+		bt_id_error_message("size missmatch. b.cols()= %d, result->cols()= %d\n",
+							static_cast<int>(b.cols()), static_cast<int>(result->cols()));
+		abort();
+	}
 
-    for (idArrayIdx col = 0; col < b.cols(); col++) {
-        const idScalar x = a(0,0)*b(0,col)+a(0,1)*b(1,col)+a(0,2)*b(2,col);
-        const idScalar y = a(1,0)*b(0,col)+a(1,1)*b(1,col)+a(1,2)*b(2,col);
-        const idScalar z = a(2,0)*b(0,col)+a(2,1)*b(1,col)+a(2,2)*b(2,col);
-        setMat3xElem(0, col, x, result);
-        setMat3xElem(1, col, y, result);
-        setMat3xElem(2, col, z, result);
-    }
+	for (idArrayIdx col = 0; col < b.cols(); col++)
+	{
+		const idScalar x = a(0, 0) * b(0, col) + a(0, 1) * b(1, col) + a(0, 2) * b(2, col);
+		const idScalar y = a(1, 0) * b(0, col) + a(1, 1) * b(1, col) + a(1, 2) * b(2, col);
+		const idScalar z = a(2, 0) * b(0, col) + a(2, 1) * b(1, col) + a(2, 2) * b(2, col);
+		setMat3xElem(0, col, x, result);
+		setMat3xElem(1, col, y, result);
+		setMat3xElem(2, col, z, result);
+	}
 }
-void add(const mat3x &a, const mat3x &b, mat3x *result) {
-    if (a.cols() != b.cols()) {
-        bt_id_error_message("size missmatch. a.cols()= %d, b.cols()= %d\n",
-                      static_cast<int>(a.cols()), static_cast<int>(b.cols()));
-        abort();
-    }
-    for (idArrayIdx col = 0; col < b.cols(); col++) {
-        for (idArrayIdx row = 0; row < 3; row++) {
-            setMat3xElem(row, col, a(row, col) + b(row, col), result);
-        }
-    }
+void add(const mat3x &a, const mat3x &b, mat3x *result)
+{
+	if (a.cols() != b.cols())
+	{
+		bt_id_error_message("size missmatch. a.cols()= %d, b.cols()= %d\n",
+							static_cast<int>(a.cols()), static_cast<int>(b.cols()));
+		abort();
+	}
+	for (idArrayIdx col = 0; col < b.cols(); col++)
+	{
+		for (idArrayIdx row = 0; row < 3; row++)
+		{
+			setMat3xElem(row, col, a(row, col) + b(row, col), result);
+		}
+	}
 }
-void sub(const mat3x &a, const mat3x &b, mat3x *result) {
-    if (a.cols() != b.cols()) {
-        bt_id_error_message("size missmatch. a.cols()= %d, b.cols()= %d\n",
-                      static_cast<int>(a.cols()), static_cast<int>(b.cols()));
-        abort();
-    }
-    for (idArrayIdx col = 0; col < b.cols(); col++) {
-        for (idArrayIdx row = 0; row < 3; row++) {
-            setMat3xElem(row, col, a(row, col) - b(row, col), result);
-        }
-    }
+void sub(const mat3x &a, const mat3x &b, mat3x *result)
+{
+	if (a.cols() != b.cols())
+	{
+		bt_id_error_message("size missmatch. a.cols()= %d, b.cols()= %d\n",
+							static_cast<int>(a.cols()), static_cast<int>(b.cols()));
+		abort();
+	}
+	for (idArrayIdx col = 0; col < b.cols(); col++)
+	{
+		for (idArrayIdx row = 0; row < 3; row++)
+		{
+			setMat3xElem(row, col, a(row, col) - b(row, col), result);
+		}
+	}
 }
 #endif
 
-mat33 transformX(const idScalar &alpha) {
+mat33 transformX(const idScalar &alpha)
+{
 	mat33 T;
 	const idScalar cos_alpha = BT_ID_COS(alpha);
 	const idScalar sin_alpha = BT_ID_SIN(alpha);
@@ -143,7 +170,8 @@ mat33 transformX(const idScalar &alpha) {
 	return T;
 }
 
-mat33 transformY(const idScalar &beta) {
+mat33 transformY(const idScalar &beta)
+{
 	mat33 T;
 	const idScalar cos_beta = BT_ID_COS(beta);
 	const idScalar sin_beta = BT_ID_SIN(beta);
@@ -165,7 +193,8 @@ mat33 transformY(const idScalar &beta) {
 	return T;
 }
 
-mat33 transformZ(const idScalar &gamma) {
+mat33 transformZ(const idScalar &gamma)
+{
 	mat33 T;
 	const idScalar cos_gamma = BT_ID_COS(gamma);
 	const idScalar sin_gamma = BT_ID_SIN(gamma);
@@ -187,7 +216,8 @@ mat33 transformZ(const idScalar &gamma) {
 	return T;
 }
 
-mat33 tildeOperator(const vec3 &v) {
+mat33 tildeOperator(const vec3 &v)
+{
 	mat33 m;
 	m(0, 0) = 0.0;
 	m(0, 1) = -v(2);
@@ -201,7 +231,8 @@ mat33 tildeOperator(const vec3 &v) {
 	return m;
 }
 
-void getVecMatFromDH(idScalar theta, idScalar d, idScalar a, idScalar alpha, vec3 *r, mat33 *T) {
+void getVecMatFromDH(idScalar theta, idScalar d, idScalar a, idScalar alpha, vec3 *r, mat33 *T)
+{
 	const idScalar sa = BT_ID_SIN(alpha);
 	const idScalar ca = BT_ID_COS(alpha);
 	const idScalar st = BT_ID_SIN(theta);
@@ -224,7 +255,8 @@ void getVecMatFromDH(idScalar theta, idScalar d, idScalar a, idScalar alpha, vec
 	(*T)(2, 2) = ca;
 }
 
-void bodyTParentFromAxisAngle(const vec3 &axis, const idScalar &angle, mat33 *T) {
+void bodyTParentFromAxisAngle(const vec3 &axis, const idScalar &angle, mat33 *T)
+{
 	const idScalar c = BT_ID_COS(angle);
 	const idScalar s = -BT_ID_SIN(angle);
 	const idScalar one_m_c = 1.0 - c;
@@ -246,175 +278,214 @@ void bodyTParentFromAxisAngle(const vec3 &axis, const idScalar &angle, mat33 *T)
 	(*T)(2, 2) = z * z * one_m_c + c;
 }
 
-bool isPositiveDefinite(const mat33 &m) {
+bool isPositiveDefinite(const mat33 &m)
+{
 	// test if all upper left determinants are positive
-	if (m(0, 0) <= 0) {  // upper 1x1
+	if (m(0, 0) <= 0)
+	{  // upper 1x1
 		return false;
 	}
-	if (m(0, 0) * m(1, 1) - m(0, 1) * m(1, 0) <= 0) {  // upper 2x2
+	if (m(0, 0) * m(1, 1) - m(0, 1) * m(1, 0) <= 0)
+	{  // upper 2x2
 		return false;
 	}
 	if ((m(0, 0) * (m(1, 1) * m(2, 2) - m(1, 2) * m(2, 1)) -
 		 m(0, 1) * (m(1, 0) * m(2, 2) - m(1, 2) * m(2, 0)) +
-		 m(0, 2) * (m(1, 0) * m(2, 1) - m(1, 1) * m(2, 0))) < 0) {
+		 m(0, 2) * (m(1, 0) * m(2, 1) - m(1, 1) * m(2, 0))) < 0)
+	{
 		return false;
 	}
 	return true;
 }
 
-bool isPositiveSemiDefinite(const mat33 &m) {
+bool isPositiveSemiDefinite(const mat33 &m)
+{
 	// test if all upper left determinants are positive
-	if (m(0, 0) < 0) {  // upper 1x1
+	if (m(0, 0) < 0)
+	{  // upper 1x1
 		return false;
 	}
-	if (m(0, 0) * m(1, 1) - m(0, 1) * m(1, 0) < 0) {  // upper 2x2
+	if (m(0, 0) * m(1, 1) - m(0, 1) * m(1, 0) < 0)
+	{  // upper 2x2
 		return false;
 	}
 	if ((m(0, 0) * (m(1, 1) * m(2, 2) - m(1, 2) * m(2, 1)) -
 		 m(0, 1) * (m(1, 0) * m(2, 2) - m(1, 2) * m(2, 0)) +
-		 m(0, 2) * (m(1, 0) * m(2, 1) - m(1, 1) * m(2, 0))) < 0) {
+		 m(0, 2) * (m(1, 0) * m(2, 1) - m(1, 1) * m(2, 0))) < 0)
+	{
 		return false;
 	}
 	return true;
 }
 
-bool isPositiveSemiDefiniteFuzzy(const mat33 &m) {
+bool isPositiveSemiDefiniteFuzzy(const mat33 &m)
+{
 	// test if all upper left determinants are positive
-	if (m(0, 0) < -kIsZero) {  // upper 1x1
+	if (m(0, 0) < -kIsZero)
+	{  // upper 1x1
 		return false;
 	}
-	if (m(0, 0) * m(1, 1) - m(0, 1) * m(1, 0) < -kIsZero) {  // upper 2x2
+	if (m(0, 0) * m(1, 1) - m(0, 1) * m(1, 0) < -kIsZero)
+	{  // upper 2x2
 		return false;
 	}
 	if ((m(0, 0) * (m(1, 1) * m(2, 2) - m(1, 2) * m(2, 1)) -
 		 m(0, 1) * (m(1, 0) * m(2, 2) - m(1, 2) * m(2, 0)) +
-		 m(0, 2) * (m(1, 0) * m(2, 1) - m(1, 1) * m(2, 0))) < -kIsZero) {
+		 m(0, 2) * (m(1, 0) * m(2, 1) - m(1, 1) * m(2, 0))) < -kIsZero)
+	{
 		return false;
 	}
 	return true;
 }
 
-idScalar determinant(const mat33 &m) {
+idScalar determinant(const mat33 &m)
+{
 	return m(0, 0) * m(1, 1) * m(2, 2) + m(0, 1) * m(1, 2) * m(2, 0) + m(0, 2) * m(1, 0) * m(2, 1) -
 		   m(0, 2) * m(1, 1) * m(2, 0) - m(0, 0) * m(1, 2) * m(2, 1) - m(0, 1) * m(1, 0) * m(2, 2);
 }
 
-bool isValidInertiaMatrix(const mat33 &I, const int index, bool has_fixed_joint) {
+bool isValidInertiaMatrix(const mat33 &I, const int index, bool has_fixed_joint)
+{
 	// TODO(Thomas) do we really want this?
 	//			  in cases where the inertia tensor about the center of mass is zero,
 	//			  the determinant of the inertia tensor about the joint axis is almost
 	//			  zero and can have a very small negative value.
-	if (!isPositiveSemiDefiniteFuzzy(I)) {
-		bt_id_error_message("invalid inertia matrix for body %d, not positive definite "
-					  "(fixed joint)\n",
-					  index);
-		bt_id_error_message("matrix is:\n"
-					  "[%.20e %.20e %.20e;\n"
-					  "%.20e %.20e %.20e;\n"
-					  "%.20e %.20e %.20e]\n",
-					  I(0, 0), I(0, 1), I(0, 2), I(1, 0), I(1, 1), I(1, 2), I(2, 0), I(2, 1),
-					  I(2, 2));
+	if (!isPositiveSemiDefiniteFuzzy(I))
+	{
+		bt_id_error_message(
+			"invalid inertia matrix for body %d, not positive definite "
+			"(fixed joint)\n",
+			index);
+		bt_id_error_message(
+			"matrix is:\n"
+			"[%.20e %.20e %.20e;\n"
+			"%.20e %.20e %.20e;\n"
+			"%.20e %.20e %.20e]\n",
+			I(0, 0), I(0, 1), I(0, 2), I(1, 0), I(1, 1), I(1, 2), I(2, 0), I(2, 1),
+			I(2, 2));
 
 		return false;
 	}
 
 	// check triangle inequality, must have I(i,i)+I(j,j)>=I(k,k)
-	if (!has_fixed_joint) {
-		if (I(0, 0) + I(1, 1) < I(2, 2)) {
+	if (!has_fixed_joint)
+	{
+		if (I(0, 0) + I(1, 1) < I(2, 2))
+		{
 			bt_id_error_message("invalid inertia tensor for body %d, I(0,0) + I(1,1) < I(2,2)\n", index);
-			bt_id_error_message("matrix is:\n"
-						  "[%.20e %.20e %.20e;\n"
-						  "%.20e %.20e %.20e;\n"
-						  "%.20e %.20e %.20e]\n",
-						  I(0, 0), I(0, 1), I(0, 2), I(1, 0), I(1, 1), I(1, 2), I(2, 0), I(2, 1),
-						  I(2, 2));
+			bt_id_error_message(
+				"matrix is:\n"
+				"[%.20e %.20e %.20e;\n"
+				"%.20e %.20e %.20e;\n"
+				"%.20e %.20e %.20e]\n",
+				I(0, 0), I(0, 1), I(0, 2), I(1, 0), I(1, 1), I(1, 2), I(2, 0), I(2, 1),
+				I(2, 2));
 			return false;
 		}
-		if (I(0, 0) + I(1, 1) < I(2, 2)) {
+		if (I(0, 0) + I(1, 1) < I(2, 2))
+		{
 			bt_id_error_message("invalid inertia tensor for body %d, I(0,0) + I(1,1) < I(2,2)\n", index);
-			bt_id_error_message("matrix is:\n"
-						  "[%.20e %.20e %.20e;\n"
-						  "%.20e %.20e %.20e;\n"
-						  "%.20e %.20e %.20e]\n",
-						  I(0, 0), I(0, 1), I(0, 2), I(1, 0), I(1, 1), I(1, 2), I(2, 0), I(2, 1),
-						  I(2, 2));
+			bt_id_error_message(
+				"matrix is:\n"
+				"[%.20e %.20e %.20e;\n"
+				"%.20e %.20e %.20e;\n"
+				"%.20e %.20e %.20e]\n",
+				I(0, 0), I(0, 1), I(0, 2), I(1, 0), I(1, 1), I(1, 2), I(2, 0), I(2, 1),
+				I(2, 2));
 			return false;
 		}
-		if (I(1, 1) + I(2, 2) < I(0, 0)) {
+		if (I(1, 1) + I(2, 2) < I(0, 0))
+		{
 			bt_id_error_message("invalid inertia tensor for body %d, I(1,1) + I(2,2) < I(0,0)\n", index);
-			bt_id_error_message("matrix is:\n"
-						  "[%.20e %.20e %.20e;\n"
-						  "%.20e %.20e %.20e;\n"
-						  "%.20e %.20e %.20e]\n",
-						  I(0, 0), I(0, 1), I(0, 2), I(1, 0), I(1, 1), I(1, 2), I(2, 0), I(2, 1),
-						  I(2, 2));
+			bt_id_error_message(
+				"matrix is:\n"
+				"[%.20e %.20e %.20e;\n"
+				"%.20e %.20e %.20e;\n"
+				"%.20e %.20e %.20e]\n",
+				I(0, 0), I(0, 1), I(0, 2), I(1, 0), I(1, 1), I(1, 2), I(2, 0), I(2, 1),
+				I(2, 2));
 			return false;
 		}
 	}
 	// check positive/zero diagonal elements
-	for (int i = 0; i < 3; i++) {
-		if (I(i, i) < 0) {  // accept zero
+	for (int i = 0; i < 3; i++)
+	{
+		if (I(i, i) < 0)
+		{  // accept zero
 			bt_id_error_message("invalid inertia tensor, I(%d,%d)= %e <0\n", i, i, I(i, i));
 			return false;
 		}
 	}
 	// check symmetry
-	if (BT_ID_FABS(I(1, 0) - I(0, 1)) > kIsZero) {
-		bt_id_error_message("invalid inertia tensor for body %d I(1,0)!=I(0,1). I(1,0)-I(0,1)= "
-					  "%e\n",
-					  index, I(1, 0) - I(0, 1));
+	if (BT_ID_FABS(I(1, 0) - I(0, 1)) > kIsZero)
+	{
+		bt_id_error_message(
+			"invalid inertia tensor for body %d I(1,0)!=I(0,1). I(1,0)-I(0,1)= "
+			"%e\n",
+			index, I(1, 0) - I(0, 1));
 		return false;
 	}
-	if (BT_ID_FABS(I(2, 0) - I(0, 2)) > kIsZero) {
-		bt_id_error_message("invalid inertia tensor for body %d I(2,0)!=I(0,2). I(2,0)-I(0,2)= "
-					  "%e\n",
-					  index, I(2, 0) - I(0, 2));
+	if (BT_ID_FABS(I(2, 0) - I(0, 2)) > kIsZero)
+	{
+		bt_id_error_message(
+			"invalid inertia tensor for body %d I(2,0)!=I(0,2). I(2,0)-I(0,2)= "
+			"%e\n",
+			index, I(2, 0) - I(0, 2));
 		return false;
 	}
-	if (BT_ID_FABS(I(1, 2) - I(2, 1)) > kIsZero) {
+	if (BT_ID_FABS(I(1, 2) - I(2, 1)) > kIsZero)
+	{
 		bt_id_error_message("invalid inertia tensor body %d I(1,2)!=I(2,1). I(1,2)-I(2,1)= %e\n", index,
-					  I(1, 2) - I(2, 1));
+							I(1, 2) - I(2, 1));
 		return false;
 	}
 	return true;
 }
 
-bool isValidTransformMatrix(const mat33 &m) {
-#define print_mat(x)																			   \
-	bt_id_error_message("matrix is [%e, %e, %e; %e, %e, %e; %e, %e, %e]\n", x(0, 0), x(0, 1), x(0, 2),   \
-				  x(1, 0), x(1, 1), x(1, 2), x(2, 0), x(2, 1), x(2, 2))
+bool isValidTransformMatrix(const mat33 &m)
+{
+#define print_mat(x)                                                                                   \
+	bt_id_error_message("matrix is [%e, %e, %e; %e, %e, %e; %e, %e, %e]\n", x(0, 0), x(0, 1), x(0, 2), \
+						x(1, 0), x(1, 1), x(1, 2), x(2, 0), x(2, 1), x(2, 2))
 
 	// check for unit length column vectors
-	for (int i = 0; i < 3; i++) {
+	for (int i = 0; i < 3; i++)
+	{
 		const idScalar length_minus_1 =
 			BT_ID_FABS(m(0, i) * m(0, i) + m(1, i) * m(1, i) + m(2, i) * m(2, i) - 1.0);
-		if (length_minus_1 > kAxisLengthEpsilon) {
-			bt_id_error_message("Not a valid rotation matrix (column %d not unit length)\n"
-						  "column = [%.18e %.18e %.18e]\n"
-						  "length-1.0= %.18e\n",
-						  i, m(0, i), m(1, i), m(2, i), length_minus_1);
+		if (length_minus_1 > kAxisLengthEpsilon)
+		{
+			bt_id_error_message(
+				"Not a valid rotation matrix (column %d not unit length)\n"
+				"column = [%.18e %.18e %.18e]\n"
+				"length-1.0= %.18e\n",
+				i, m(0, i), m(1, i), m(2, i), length_minus_1);
 			print_mat(m);
 			return false;
 		}
 	}
 	// check for orthogonal column vectors
-	if (BT_ID_FABS(m(0, 0) * m(0, 1) + m(1, 0) * m(1, 1) + m(2, 0) * m(2, 1)) > kAxisLengthEpsilon) {
+	if (BT_ID_FABS(m(0, 0) * m(0, 1) + m(1, 0) * m(1, 1) + m(2, 0) * m(2, 1)) > kAxisLengthEpsilon)
+	{
 		bt_id_error_message("Not a valid rotation matrix (columns 0 and 1 not orthogonal)\n");
 		print_mat(m);
 		return false;
 	}
-	if (BT_ID_FABS(m(0, 0) * m(0, 2) + m(1, 0) * m(1, 2) + m(2, 0) * m(2, 2)) > kAxisLengthEpsilon) {
+	if (BT_ID_FABS(m(0, 0) * m(0, 2) + m(1, 0) * m(1, 2) + m(2, 0) * m(2, 2)) > kAxisLengthEpsilon)
+	{
 		bt_id_error_message("Not a valid rotation matrix (columns 0 and 2 not orthogonal)\n");
 		print_mat(m);
 		return false;
 	}
-	if (BT_ID_FABS(m(0, 1) * m(0, 2) + m(1, 1) * m(1, 2) + m(2, 1) * m(2, 2)) > kAxisLengthEpsilon) {
+	if (BT_ID_FABS(m(0, 1) * m(0, 2) + m(1, 1) * m(1, 2) + m(2, 1) * m(2, 2)) > kAxisLengthEpsilon)
+	{
 		bt_id_error_message("Not a valid rotation matrix (columns 0 and 2 not orthogonal)\n");
 		print_mat(m);
 		return false;
 	}
 	// check determinant (rotation not reflection)
-	if (determinant(m) <= 0) {
+	if (determinant(m) <= 0)
+	{
 		bt_id_error_message("Not a valid rotation matrix (determinant <=0)\n");
 		print_mat(m);
 		return false;
@@ -422,16 +493,18 @@ bool isValidTransformMatrix(const mat33 &m) {
 	return true;
 }
 
-bool isUnitVector(const vec3 &vector) {
+bool isUnitVector(const vec3 &vector)
+{
 	return BT_ID_FABS(vector(0) * vector(0) + vector(1) * vector(1) + vector(2) * vector(2) - 1.0) <
 		   kIsZero;
 }
 
-vec3 rpyFromMatrix(const mat33 &rot) {
+vec3 rpyFromMatrix(const mat33 &rot)
+{
 	vec3 rpy;
 	rpy(2) = BT_ID_ATAN2(-rot(1, 0), rot(0, 0));
 	rpy(1) = BT_ID_ATAN2(rot(2, 0), BT_ID_COS(rpy(2)) * rot(0, 0) - BT_ID_SIN(rpy(0)) * rot(1, 0));
 	rpy(0) = BT_ID_ATAN2(-rot(2, 0), rot(2, 2));
 	return rpy;
 }
-}
+}  // namespace btInverseDynamics
diff --git a/thirdparty/bullet/BulletInverseDynamics/IDMath.hpp b/thirdparty/bullet/BulletInverseDynamics/IDMath.hpp
index b355474d44e..40bee5302b9 100644
--- a/thirdparty/bullet/BulletInverseDynamics/IDMath.hpp
+++ b/thirdparty/bullet/BulletInverseDynamics/IDMath.hpp
@@ -5,7 +5,8 @@
 #define IDMATH_HPP_
 #include "IDConfig.hpp"
 
-namespace btInverseDynamics {
+namespace btInverseDynamics
+{
 /// set all elements to zero
 void setZero(vec3& v);
 /// set all elements to zero
@@ -23,11 +24,11 @@ idScalar maxAbs(const vec3& v);
 
 #if (defined BT_ID_HAVE_MAT3X)
 idScalar maxAbsMat3x(const mat3x& m);
-void setZero(mat3x&m);
+void setZero(mat3x& m);
 // define math functions on mat3x here to avoid allocations in operators.
-void mul(const mat33&a, const mat3x&b, mat3x* result);
-void add(const mat3x&a, const mat3x&b, mat3x* result);
-void sub(const mat3x&a, const mat3x&b, mat3x* result);
+void mul(const mat33& a, const mat3x& b, mat3x* result);
+void add(const mat3x& a, const mat3x& b, mat3x* result);
+void sub(const mat3x& a, const mat3x& b, mat3x* result);
 #endif
 
 /// get offset vector & transform matrix from DH parameters
@@ -94,6 +95,6 @@ mat33 transformZ(const idScalar& gamma);
 ///calculate rpy angles (x-y-z Euler angles) from a given rotation matrix
 /// @param rot rotation matrix
 /// @returns x-y-z Euler angles
-vec3 rpyFromMatrix(const mat33&rot);
-}
+vec3 rpyFromMatrix(const mat33& rot);
+}  // namespace btInverseDynamics
 #endif  // IDMATH_HPP_
diff --git a/thirdparty/bullet/BulletInverseDynamics/MultiBodyTree.cpp b/thirdparty/bullet/BulletInverseDynamics/MultiBodyTree.cpp
index becfe0f4a21..f150b5ae4ca 100644
--- a/thirdparty/bullet/BulletInverseDynamics/MultiBodyTree.cpp
+++ b/thirdparty/bullet/BulletInverseDynamics/MultiBodyTree.cpp
@@ -8,69 +8,84 @@
 #include "details/MultiBodyTreeImpl.hpp"
 #include "details/MultiBodyTreeInitCache.hpp"
 
-namespace btInverseDynamics {
-
+namespace btInverseDynamics
+{
 MultiBodyTree::MultiBodyTree()
 	: m_is_finalized(false),
 	  m_mass_parameters_are_valid(true),
 	  m_accept_invalid_mass_parameters(false),
 	  m_impl(0x0),
-	  m_init_cache(0x0) {
+	  m_init_cache(0x0)
+{
 	m_init_cache = new InitCache();
 }
 
-MultiBodyTree::~MultiBodyTree() {
+MultiBodyTree::~MultiBodyTree()
+{
 	delete m_impl;
 	delete m_init_cache;
 }
 
-void MultiBodyTree::setAcceptInvalidMassParameters(bool flag) {
+void MultiBodyTree::setAcceptInvalidMassParameters(bool flag)
+{
 	m_accept_invalid_mass_parameters = flag;
 }
 
-bool MultiBodyTree::getAcceptInvalidMassProperties() const {
+bool MultiBodyTree::getAcceptInvalidMassProperties() const
+{
 	return m_accept_invalid_mass_parameters;
 }
 
-int MultiBodyTree::getBodyOrigin(const int body_index, vec3 *world_origin) const {
+int MultiBodyTree::getBodyOrigin(const int body_index, vec3 *world_origin) const
+{
 	return m_impl->getBodyOrigin(body_index, world_origin);
 }
 
-int MultiBodyTree::getBodyCoM(const int body_index, vec3 *world_com) const {
+int MultiBodyTree::getBodyCoM(const int body_index, vec3 *world_com) const
+{
 	return m_impl->getBodyCoM(body_index, world_com);
 }
 
-int MultiBodyTree::getBodyTransform(const int body_index, mat33 *world_T_body) const {
+int MultiBodyTree::getBodyTransform(const int body_index, mat33 *world_T_body) const
+{
 	return m_impl->getBodyTransform(body_index, world_T_body);
 }
-int MultiBodyTree::getBodyAngularVelocity(const int body_index, vec3 *world_omega) const {
+int MultiBodyTree::getBodyAngularVelocity(const int body_index, vec3 *world_omega) const
+{
 	return m_impl->getBodyAngularVelocity(body_index, world_omega);
 }
-int MultiBodyTree::getBodyLinearVelocity(const int body_index, vec3 *world_velocity) const {
+int MultiBodyTree::getBodyLinearVelocity(const int body_index, vec3 *world_velocity) const
+{
 	return m_impl->getBodyLinearVelocity(body_index, world_velocity);
 }
 
-int MultiBodyTree::getBodyLinearVelocityCoM(const int body_index, vec3 *world_velocity) const {
+int MultiBodyTree::getBodyLinearVelocityCoM(const int body_index, vec3 *world_velocity) const
+{
 	return m_impl->getBodyLinearVelocityCoM(body_index, world_velocity);
 }
 
-int MultiBodyTree::getBodyAngularAcceleration(const int body_index, vec3 *world_dot_omega) const {
+int MultiBodyTree::getBodyAngularAcceleration(const int body_index, vec3 *world_dot_omega) const
+{
 	return m_impl->getBodyAngularAcceleration(body_index, world_dot_omega);
 }
-int MultiBodyTree::getBodyLinearAcceleration(const int body_index, vec3 *world_acceleration) const {
+int MultiBodyTree::getBodyLinearAcceleration(const int body_index, vec3 *world_acceleration) const
+{
 	return m_impl->getBodyLinearAcceleration(body_index, world_acceleration);
 }
 
-int MultiBodyTree::getParentRParentBodyRef(const int body_index, vec3* r) const {
-    return m_impl->getParentRParentBodyRef(body_index, r);
+int MultiBodyTree::getParentRParentBodyRef(const int body_index, vec3 *r) const
+{
+	return m_impl->getParentRParentBodyRef(body_index, r);
 }
 
-int MultiBodyTree::getBodyTParentRef(const int body_index, mat33* T) const {
-    return m_impl->getBodyTParentRef(body_index, T);
+int MultiBodyTree::getBodyTParentRef(const int body_index, mat33 *T) const
+{
+	return m_impl->getBodyTParentRef(body_index, T);
 }
 
-int MultiBodyTree::getBodyAxisOfMotion(const int body_index, vec3* axis) const {
-    return m_impl->getBodyAxisOfMotion(body_index, axis);
+int MultiBodyTree::getBodyAxisOfMotion(const int body_index, vec3 *axis) const
+{
+	return m_impl->getBodyAxisOfMotion(body_index, axis);
 }
 
 void MultiBodyTree::printTree() { m_impl->printTree(); }
@@ -81,12 +96,15 @@ int MultiBodyTree::numBodies() const { return m_impl->m_num_bodies; }
 int MultiBodyTree::numDoFs() const { return m_impl->m_num_dofs; }
 
 int MultiBodyTree::calculateInverseDynamics(const vecx &q, const vecx &u, const vecx &dot_u,
-											vecx *joint_forces) {
-	if (false == m_is_finalized) {
+											vecx *joint_forces)
+{
+	if (false == m_is_finalized)
+	{
 		bt_id_error_message("system has not been initialized\n");
 		return -1;
 	}
-	if (-1 == m_impl->calculateInverseDynamics(q, u, dot_u, joint_forces)) {
+	if (-1 == m_impl->calculateInverseDynamics(q, u, dot_u, joint_forces))
+	{
 		bt_id_error_message("error in inverse dynamics calculation\n");
 		return -1;
 	}
@@ -95,141 +113,164 @@ int MultiBodyTree::calculateInverseDynamics(const vecx &q, const vecx &u, const
 
 int MultiBodyTree::calculateMassMatrix(const vecx &q, const bool update_kinematics,
 									   const bool initialize_matrix,
-									   const bool set_lower_triangular_matrix, matxx *mass_matrix) {
-	if (false == m_is_finalized) {
+									   const bool set_lower_triangular_matrix, matxx *mass_matrix)
+{
+	if (false == m_is_finalized)
+	{
 		bt_id_error_message("system has not been initialized\n");
 		return -1;
 	}
 	if (-1 ==
 		m_impl->calculateMassMatrix(q, update_kinematics, initialize_matrix,
-									set_lower_triangular_matrix, mass_matrix)) {
+									set_lower_triangular_matrix, mass_matrix))
+	{
 		bt_id_error_message("error in mass matrix calculation\n");
 		return -1;
 	}
 	return 0;
 }
 
-int MultiBodyTree::calculateMassMatrix(const vecx &q, matxx *mass_matrix) {
+int MultiBodyTree::calculateMassMatrix(const vecx &q, matxx *mass_matrix)
+{
 	return calculateMassMatrix(q, true, true, true, mass_matrix);
 }
 
+int MultiBodyTree::calculateKinematics(const vecx &q, const vecx &u, const vecx &dot_u)
+{
+	vec3 world_gravity(m_impl->m_world_gravity);
+	// temporarily set gravity to zero, to ensure we get the actual accelerations
+	setZero(m_impl->m_world_gravity);
 
+	if (false == m_is_finalized)
+	{
+		bt_id_error_message("system has not been initialized\n");
+		return -1;
+	}
+	if (-1 == m_impl->calculateKinematics(q, u, dot_u,
+										  MultiBodyTree::MultiBodyImpl::POSITION_VELOCITY_ACCELERATION))
+	{
+		bt_id_error_message("error in kinematics calculation\n");
+		return -1;
+	}
 
-int MultiBodyTree::calculateKinematics(const vecx& q, const vecx& u, const vecx& dot_u) {
-    vec3 world_gravity(m_impl->m_world_gravity);
-    // temporarily set gravity to zero, to ensure we get the actual accelerations
-    setZero(m_impl->m_world_gravity);
-
-    if (false == m_is_finalized) {
-        bt_id_error_message("system has not been initialized\n");
-        return -1;
-    }
-    if (-1 == m_impl->calculateKinematics(q, u, dot_u,
-                                          MultiBodyTree::MultiBodyImpl::POSITION_VELOCITY_ACCELERATION)) {
-        bt_id_error_message("error in kinematics calculation\n");
-        return -1;
-    }
-
-    m_impl->m_world_gravity=world_gravity;
-    return 0;
+	m_impl->m_world_gravity = world_gravity;
+	return 0;
 }
 
-
-int MultiBodyTree::calculatePositionKinematics(const vecx& q) {
-	if (false == m_is_finalized) {
+int MultiBodyTree::calculatePositionKinematics(const vecx &q)
+{
+	if (false == m_is_finalized)
+	{
 		bt_id_error_message("system has not been initialized\n");
 		return -1;
 	}
 	if (-1 == m_impl->calculateKinematics(q, q, q,
-                                              MultiBodyTree::MultiBodyImpl::POSITION_VELOCITY)) {
+										  MultiBodyTree::MultiBodyImpl::POSITION_VELOCITY))
+	{
 		bt_id_error_message("error in kinematics calculation\n");
 		return -1;
 	}
 	return 0;
 }
 
-int MultiBodyTree::calculatePositionAndVelocityKinematics(const vecx& q, const vecx& u) {
-	if (false == m_is_finalized) {
+int MultiBodyTree::calculatePositionAndVelocityKinematics(const vecx &q, const vecx &u)
+{
+	if (false == m_is_finalized)
+	{
 		bt_id_error_message("system has not been initialized\n");
 		return -1;
 	}
 	if (-1 == m_impl->calculateKinematics(q, u, u,
-                                              MultiBodyTree::MultiBodyImpl::POSITION_VELOCITY)) {
+										  MultiBodyTree::MultiBodyImpl::POSITION_VELOCITY))
+	{
 		bt_id_error_message("error in kinematics calculation\n");
 		return -1;
 	}
 	return 0;
 }
 
-
 #if (defined BT_ID_HAVE_MAT3X) && (defined BT_ID_WITH_JACOBIANS)
-int MultiBodyTree::calculateJacobians(const vecx& q, const vecx& u) {
-    if (false == m_is_finalized) {
-        bt_id_error_message("system has not been initialized\n");
-        return -1;
-    }
-    if (-1 == m_impl->calculateJacobians(q, u,
-                                         MultiBodyTree::MultiBodyImpl::POSITION_VELOCITY)) {
-        bt_id_error_message("error in jacobian calculation\n");
-        return -1;
-    }
-    return 0;
+int MultiBodyTree::calculateJacobians(const vecx &q, const vecx &u)
+{
+	if (false == m_is_finalized)
+	{
+		bt_id_error_message("system has not been initialized\n");
+		return -1;
+	}
+	if (-1 == m_impl->calculateJacobians(q, u,
+										 MultiBodyTree::MultiBodyImpl::POSITION_VELOCITY))
+	{
+		bt_id_error_message("error in jacobian calculation\n");
+		return -1;
+	}
+	return 0;
 }
 
-int MultiBodyTree::calculateJacobians(const vecx& q){
-    if (false == m_is_finalized) {
-        bt_id_error_message("system has not been initialized\n");
-        return -1;
-    }
-    if (-1 == m_impl->calculateJacobians(q, q,
-                                         MultiBodyTree::MultiBodyImpl::POSITION_ONLY)) {
-        bt_id_error_message("error in jacobian calculation\n");
-        return -1;
-    }
-    return 0;
+int MultiBodyTree::calculateJacobians(const vecx &q)
+{
+	if (false == m_is_finalized)
+	{
+		bt_id_error_message("system has not been initialized\n");
+		return -1;
+	}
+	if (-1 == m_impl->calculateJacobians(q, q,
+										 MultiBodyTree::MultiBodyImpl::POSITION_ONLY))
+	{
+		bt_id_error_message("error in jacobian calculation\n");
+		return -1;
+	}
+	return 0;
 }
 
-int MultiBodyTree::getBodyDotJacobianTransU(const int body_index, vec3* world_dot_jac_trans_u) const {
-    return m_impl->getBodyDotJacobianTransU(body_index,world_dot_jac_trans_u);
+int MultiBodyTree::getBodyDotJacobianTransU(const int body_index, vec3 *world_dot_jac_trans_u) const
+{
+	return m_impl->getBodyDotJacobianTransU(body_index, world_dot_jac_trans_u);
 }
 
-int MultiBodyTree::getBodyDotJacobianRotU(const int body_index, vec3* world_dot_jac_rot_u) const {
-    return m_impl->getBodyDotJacobianRotU(body_index,world_dot_jac_rot_u);
+int MultiBodyTree::getBodyDotJacobianRotU(const int body_index, vec3 *world_dot_jac_rot_u) const
+{
+	return m_impl->getBodyDotJacobianRotU(body_index, world_dot_jac_rot_u);
 }
 
-int MultiBodyTree::getBodyJacobianTrans(const int body_index, mat3x* world_jac_trans) const {
-    return m_impl->getBodyJacobianTrans(body_index,world_jac_trans);
+int MultiBodyTree::getBodyJacobianTrans(const int body_index, mat3x *world_jac_trans) const
+{
+	return m_impl->getBodyJacobianTrans(body_index, world_jac_trans);
 }
 
-int MultiBodyTree::getBodyJacobianRot(const int body_index, mat3x* world_jac_rot) const  {
-    return m_impl->getBodyJacobianRot(body_index,world_jac_rot);
+int MultiBodyTree::getBodyJacobianRot(const int body_index, mat3x *world_jac_rot) const
+{
+	return m_impl->getBodyJacobianRot(body_index, world_jac_rot);
 }
 
-
 #endif
 
 int MultiBodyTree::addBody(int body_index, int parent_index, JointType joint_type,
 						   const vec3 &parent_r_parent_body_ref, const mat33 &body_T_parent_ref,
 						   const vec3 &body_axis_of_motion_, idScalar mass,
 						   const vec3 &body_r_body_com, const mat33 &body_I_body,
-						   const int user_int, void *user_ptr) {
-	if (body_index < 0) {
+						   const int user_int, void *user_ptr)
+{
+	if (body_index < 0)
+	{
 		bt_id_error_message("body index must be positive (got %d)\n", body_index);
 		return -1;
 	}
 	vec3 body_axis_of_motion(body_axis_of_motion_);
-	switch (joint_type) {
+	switch (joint_type)
+	{
 		case REVOLUTE:
 		case PRISMATIC:
 			// check if axis is unit vector
-			if (!isUnitVector(body_axis_of_motion)) {
+			if (!isUnitVector(body_axis_of_motion))
+			{
 				bt_id_warning_message(
 					"axis of motion not a unit axis ([%f %f %f]), will use normalized vector\n",
 					body_axis_of_motion(0), body_axis_of_motion(1), body_axis_of_motion(2));
 				idScalar length = BT_ID_SQRT(BT_ID_POW(body_axis_of_motion(0), 2) +
-									   BT_ID_POW(body_axis_of_motion(1), 2) +
-									   BT_ID_POW(body_axis_of_motion(2), 2));
-				if (length < BT_ID_SQRT(std::numeric_limits<idScalar>::min())) {
+											 BT_ID_POW(body_axis_of_motion(1), 2) +
+											 BT_ID_POW(body_axis_of_motion(2), 2));
+				if (length < BT_ID_SQRT(std::numeric_limits<idScalar>::min()))
+				{
 					bt_id_error_message("axis of motion vector too short (%e)\n", length);
 					return -1;
 				}
@@ -240,29 +281,36 @@ int MultiBodyTree::addBody(int body_index, int parent_index, JointType joint_typ
 			break;
 		case FLOATING:
 			break;
+		case SPHERICAL:
+			break;
 		default:
 			bt_id_error_message("unknown joint type %d\n", joint_type);
 			return -1;
 	}
 
 	// sanity check for mass properties. Zero mass is OK.
-	if (mass < 0) {
+	if (mass < 0)
+	{
 		m_mass_parameters_are_valid = false;
 		bt_id_error_message("Body %d has invalid mass %e\n", body_index, mass);
-		if (!m_accept_invalid_mass_parameters) {
+		if (!m_accept_invalid_mass_parameters)
+		{
 			return -1;
 		}
 	}
 
-	if (!isValidInertiaMatrix(body_I_body, body_index, FIXED == joint_type)) {
+	if (!isValidInertiaMatrix(body_I_body, body_index, FIXED == joint_type))
+	{
 		m_mass_parameters_are_valid = false;
 		// error message printed in function call
-		if (!m_accept_invalid_mass_parameters) {
+		if (!m_accept_invalid_mass_parameters)
+		{
 			return -1;
 		}
 	}
 
-	if (!isValidTransformMatrix(body_T_parent_ref)) {
+	if (!isValidTransformMatrix(body_T_parent_ref))
+	{
 		return -1;
 	}
 
@@ -271,52 +319,63 @@ int MultiBodyTree::addBody(int body_index, int parent_index, JointType joint_typ
 								 body_I_body, user_int, user_ptr);
 }
 
-int MultiBodyTree::getParentIndex(const int body_index, int *parent_index) const {
+int MultiBodyTree::getParentIndex(const int body_index, int *parent_index) const
+{
 	return m_impl->getParentIndex(body_index, parent_index);
 }
 
-int MultiBodyTree::getUserInt(const int body_index, int *user_int) const {
+int MultiBodyTree::getUserInt(const int body_index, int *user_int) const
+{
 	return m_impl->getUserInt(body_index, user_int);
 }
 
-int MultiBodyTree::getUserPtr(const int body_index, void **user_ptr) const {
+int MultiBodyTree::getUserPtr(const int body_index, void **user_ptr) const
+{
 	return m_impl->getUserPtr(body_index, user_ptr);
 }
 
-int MultiBodyTree::setUserInt(const int body_index, const int user_int) {
+int MultiBodyTree::setUserInt(const int body_index, const int user_int)
+{
 	return m_impl->setUserInt(body_index, user_int);
 }
 
-int MultiBodyTree::setUserPtr(const int body_index, void *const user_ptr) {
+int MultiBodyTree::setUserPtr(const int body_index, void *const user_ptr)
+{
 	return m_impl->setUserPtr(body_index, user_ptr);
 }
 
-int MultiBodyTree::finalize() {
+int MultiBodyTree::finalize()
+{
 	const int &num_bodies = m_init_cache->numBodies();
 	const int &num_dofs = m_init_cache->numDoFs();
 
-        if(num_dofs<=0) {
-            bt_id_error_message("Need num_dofs>=1, but num_dofs= %d\n", num_dofs);
-            //return -1;
-        }
+	if (num_dofs <= 0)
+	{
+		bt_id_error_message("Need num_dofs>=1, but num_dofs= %d\n", num_dofs);
+		//return -1;
+	}
 
 	// 1 allocate internal MultiBody structure
 	m_impl = new MultiBodyImpl(num_bodies, num_dofs);
 
 	// 2 build new index set assuring index(parent) < index(child)
-	if (-1 == m_init_cache->buildIndexSets()) {
+	if (-1 == m_init_cache->buildIndexSets())
+	{
 		return -1;
 	}
 	m_init_cache->getParentIndexArray(&m_impl->m_parent_index);
 
 	// 3 setup internal kinematic and dynamic data
-	for (int index = 0; index < num_bodies; index++) {
+	for (int index = 0; index < num_bodies; index++)
+	{
 		InertiaData inertia;
 		JointData joint;
-		if (-1 == m_init_cache->getInertiaData(index, &inertia)) {
+		if (-1 == m_init_cache->getInertiaData(index, &inertia))
+		{
 			return -1;
 		}
-		if (-1 == m_init_cache->getJointData(index, &joint)) {
+		if (-1 == m_init_cache->getJointData(index, &joint))
+		{
 			return -1;
 		}
 
@@ -332,24 +391,29 @@ int MultiBodyTree::finalize() {
 		rigid_body.m_joint_type = joint.m_type;
 
 		int user_int;
-		if (-1 == m_init_cache->getUserInt(index, &user_int)) {
+		if (-1 == m_init_cache->getUserInt(index, &user_int))
+		{
 			return -1;
 		}
-		if (-1 == m_impl->setUserInt(index, user_int)) {
+		if (-1 == m_impl->setUserInt(index, user_int))
+		{
 			return -1;
 		}
 
-		void* user_ptr;
-		if (-1 == m_init_cache->getUserPtr(index, &user_ptr)) {
+		void *user_ptr;
+		if (-1 == m_init_cache->getUserPtr(index, &user_ptr))
+		{
 			return -1;
 		}
-		if (-1 == m_impl->setUserPtr(index, user_ptr)) {
+		if (-1 == m_impl->setUserPtr(index, user_ptr))
+		{
 			return -1;
 		}
 
 		// Set joint Jacobians. Note that the dimension is always 3x1 here to avoid variable sized
 		// matrices.
-		switch (rigid_body.m_joint_type) {
+		switch (rigid_body.m_joint_type)
+		{
 			case REVOLUTE:
 				rigid_body.m_Jac_JR(0) = joint.m_child_axis_of_motion(0);
 				rigid_body.m_Jac_JR(1) = joint.m_child_axis_of_motion(1);
@@ -375,6 +439,16 @@ int MultiBodyTree::finalize() {
 				rigid_body.m_Jac_JT(1) = 0.0;
 				rigid_body.m_Jac_JT(2) = 0.0;
 				break;
+			case SPHERICAL:
+				// NOTE/TODO: this is not really correct.
+				// the Jacobians should be 3x3 matrices here !
+				rigid_body.m_Jac_JR(0) = 0.0;
+				rigid_body.m_Jac_JR(1) = 0.0;
+				rigid_body.m_Jac_JR(2) = 0.0;
+				rigid_body.m_Jac_JT(0) = 0.0;
+				rigid_body.m_Jac_JT(1) = 0.0;
+				rigid_body.m_Jac_JT(2) = 0.0;
+				break;
 			case FLOATING:
 				// NOTE/TODO: this is not really correct.
 				// the Jacobians should be 3x3 matrices here !
@@ -392,7 +466,8 @@ int MultiBodyTree::finalize() {
 	}
 
 	// 4 assign degree of freedom indices & build per-joint-type index arrays
-	if (-1 == m_impl->generateIndexSets()) {
+	if (-1 == m_impl->generateIndexSets())
+	{
 		bt_id_error_message("generating index sets\n");
 		return -1;
 	}
@@ -408,54 +483,66 @@ int MultiBodyTree::finalize() {
 	return 0;
 }
 
-int MultiBodyTree::setGravityInWorldFrame(const vec3 &gravity) {
+int MultiBodyTree::setGravityInWorldFrame(const vec3 &gravity)
+{
 	return m_impl->setGravityInWorldFrame(gravity);
 }
 
-int MultiBodyTree::getJointType(const int body_index, JointType *joint_type) const {
+int MultiBodyTree::getJointType(const int body_index, JointType *joint_type) const
+{
 	return m_impl->getJointType(body_index, joint_type);
 }
 
-int MultiBodyTree::getJointTypeStr(const int body_index, const char **joint_type) const {
+int MultiBodyTree::getJointTypeStr(const int body_index, const char **joint_type) const
+{
 	return m_impl->getJointTypeStr(body_index, joint_type);
 }
 
-int MultiBodyTree::getDoFOffset(const int body_index, int *q_offset) const {
+int MultiBodyTree::getDoFOffset(const int body_index, int *q_offset) const
+{
 	return m_impl->getDoFOffset(body_index, q_offset);
 }
 
-int MultiBodyTree::setBodyMass(const int body_index, idScalar mass) {
+int MultiBodyTree::setBodyMass(const int body_index, idScalar mass)
+{
 	return m_impl->setBodyMass(body_index, mass);
 }
 
-int MultiBodyTree::setBodyFirstMassMoment(const int body_index, const vec3& first_mass_moment) {
+int MultiBodyTree::setBodyFirstMassMoment(const int body_index, const vec3 &first_mass_moment)
+{
 	return m_impl->setBodyFirstMassMoment(body_index, first_mass_moment);
 }
 
-int MultiBodyTree::setBodySecondMassMoment(const int body_index, const mat33& second_mass_moment) {
+int MultiBodyTree::setBodySecondMassMoment(const int body_index, const mat33 &second_mass_moment)
+{
 	return m_impl->setBodySecondMassMoment(body_index, second_mass_moment);
 }
 
-int MultiBodyTree::getBodyMass(const int body_index, idScalar *mass) const {
+int MultiBodyTree::getBodyMass(const int body_index, idScalar *mass) const
+{
 	return m_impl->getBodyMass(body_index, mass);
 }
 
-int MultiBodyTree::getBodyFirstMassMoment(const int body_index, vec3 *first_mass_moment) const {
+int MultiBodyTree::getBodyFirstMassMoment(const int body_index, vec3 *first_mass_moment) const
+{
 	return m_impl->getBodyFirstMassMoment(body_index, first_mass_moment);
 }
 
-int MultiBodyTree::getBodySecondMassMoment(const int body_index, mat33 *second_mass_moment) const {
+int MultiBodyTree::getBodySecondMassMoment(const int body_index, mat33 *second_mass_moment) const
+{
 	return m_impl->getBodySecondMassMoment(body_index, second_mass_moment);
 }
 
 void MultiBodyTree::clearAllUserForcesAndMoments() { m_impl->clearAllUserForcesAndMoments(); }
 
-int MultiBodyTree::addUserForce(const int body_index, const vec3 &body_force) {
+int MultiBodyTree::addUserForce(const int body_index, const vec3 &body_force)
+{
 	return m_impl->addUserForce(body_index, body_force);
 }
 
-int MultiBodyTree::addUserMoment(const int body_index, const vec3 &body_moment) {
+int MultiBodyTree::addUserMoment(const int body_index, const vec3 &body_moment)
+{
 	return m_impl->addUserMoment(body_index, body_moment);
 }
 
-}
+}  // namespace btInverseDynamics
diff --git a/thirdparty/bullet/BulletInverseDynamics/MultiBodyTree.hpp b/thirdparty/bullet/BulletInverseDynamics/MultiBodyTree.hpp
index d235aa6e761..7b852f976c1 100644
--- a/thirdparty/bullet/BulletInverseDynamics/MultiBodyTree.hpp
+++ b/thirdparty/bullet/BulletInverseDynamics/MultiBodyTree.hpp
@@ -4,10 +4,11 @@
 #include "IDConfig.hpp"
 #include "IDMath.hpp"
 
-namespace btInverseDynamics {
-
+namespace btInverseDynamics
+{
 /// Enumeration of supported joint types
-enum JointType {
+enum JointType
+{
 	/// no degree of freedom, moves with parent
 	FIXED = 0,
 	/// one rotational degree of freedom relative to parent
@@ -15,7 +16,9 @@ enum JointType {
 	/// one translational degree of freedom relative to parent
 	PRISMATIC,
 	/// six degrees of freedom relative to parent
-	FLOATING
+	FLOATING,
+	/// three degrees of freedom, relative to parent
+	SPHERICAL
 };
 
 /// Interface class for calculating inverse dynamics for tree structured
@@ -30,12 +33,14 @@ enum JointType {
 ///	- PRISMATIC: displacement [m]
 ///	- FLOATING:  Euler x-y-z angles [rad] and displacement in body-fixed frame of parent [m]
 ///				 (in that order)
+/// - SPHERICAL: Euler x-y-z angles [rad]
 /// The u vector contains the generalized speeds, which are
 ///	- FIXED:	 none
 ///	- REVOLUTE:  time derivative of angle of rotation [rad/s]
 ///	- PRISMATIC: time derivative of displacement [m/s]
 ///	- FLOATING:  angular velocity [rad/s] (*not* time derivative of rpy angles)
 ///				 and time derivative of displacement in parent frame [m/s]
+//	- SPHERICAL:  angular velocity [rad/s]
 ///
 /// The q and u vectors are obtained by stacking contributions of all bodies in one
 /// vector in the order of body indices.
@@ -46,12 +51,13 @@ enum JointType {
 ///	 - PRISMATIC: force  [N], along joint axis
 ///	 - FLOATING:  moment vector [Nm] and force vector [N], both in body-fixed frame
 ///				  (in that order)
-///
+///  - SPHERICAL: moment vector [Nm] 
 /// TODO - force element interface (friction, springs, dampers, etc)
 ///	  - gears and motor inertia
-class MultiBodyTree {
+class MultiBodyTree
+{
 public:
-        ID_DECLARE_ALIGNED_ALLOCATOR();
+	ID_DECLARE_ALIGNED_ALLOCATOR();
 	/// The contructor.
 	/// Initialization & allocation is via addBody and buildSystem calls.
 	MultiBodyTree();
@@ -119,9 +125,9 @@ public:
 	/// print tree data to stdout
 	void printTreeData();
 	/// Calculate joint forces for given generalized state & derivatives.
-        /// This also updates kinematic terms computed in calculateKinematics.
-        /// If gravity is not set to zero, acceleration terms will contain
-        /// gravitational acceleration.
+	/// This also updates kinematic terms computed in calculateKinematics.
+	/// If gravity is not set to zero, acceleration terms will contain
+	/// gravitational acceleration.
 	/// @param q generalized coordinates
 	/// @param u generalized velocities. In the general case, u=T(q)*dot(q) and dim(q)>=dim(u)
 	/// @param dot_u time derivative of u
@@ -152,30 +158,28 @@ public:
 	/// @return -1 on error, 0 on success
 	int calculateMassMatrix(const vecx& q, matxx* mass_matrix);
 
-
-        /// Calculates kinematics also calculated in calculateInverseDynamics,
-        /// but not dynamics.
-        /// This function ensures that correct accelerations are computed that do not
-        /// contain gravitational acceleration terms.
-        /// Does not calculate Jacobians, but only vector quantities (positions, velocities & accelerations)
-        int calculateKinematics(const vecx& q, const vecx& u, const vecx& dot_u);
-        /// Calculate position kinematics
-        int calculatePositionKinematics(const vecx& q);
-         /// Calculate position and velocity kinematics
-        int calculatePositionAndVelocityKinematics(const vecx& q, const vecx& u);
+	/// Calculates kinematics also calculated in calculateInverseDynamics,
+	/// but not dynamics.
+	/// This function ensures that correct accelerations are computed that do not
+	/// contain gravitational acceleration terms.
+	/// Does not calculate Jacobians, but only vector quantities (positions, velocities & accelerations)
+	int calculateKinematics(const vecx& q, const vecx& u, const vecx& dot_u);
+	/// Calculate position kinematics
+	int calculatePositionKinematics(const vecx& q);
+	/// Calculate position and velocity kinematics
+	int calculatePositionAndVelocityKinematics(const vecx& q, const vecx& u);
 
 #if (defined BT_ID_HAVE_MAT3X) && (defined BT_ID_WITH_JACOBIANS)
-        /// Calculate Jacobians (dvel/du), as well as velocity-dependent accelearation components
-        /// d(Jacobian)/dt*u
-        /// This function assumes that calculateInverseDynamics was called, or calculateKinematics,
-        /// or calculatePositionAndVelocityKinematics
-        int calculateJacobians(const vecx& q, const vecx& u);
-        /// Calculate Jacobians (dvel/du)
-        /// This function assumes that calculateInverseDynamics was called, or
-        /// one of the calculateKineamtics functions
-        int calculateJacobians(const vecx& q);
-#endif // BT_ID_HAVE_MAT3X
-
+	/// Calculate Jacobians (dvel/du), as well as velocity-dependent accelearation components
+	/// d(Jacobian)/dt*u
+	/// This function assumes that calculateInverseDynamics was called, or calculateKinematics,
+	/// or calculatePositionAndVelocityKinematics
+	int calculateJacobians(const vecx& q, const vecx& u);
+	/// Calculate Jacobians (dvel/du)
+	/// This function assumes that calculateInverseDynamics was called, or
+	/// one of the calculateKineamtics functions
+	int calculateJacobians(const vecx& q);
+#endif  // BT_ID_HAVE_MAT3X
 
 	/// set gravitational acceleration
 	/// the default is [0;0;-9.8] in the world frame
@@ -231,15 +235,15 @@ public:
 	int getBodyLinearAcceleration(const int body_index, vec3* world_acceleration) const;
 
 #if (defined BT_ID_HAVE_MAT3X) && (defined BT_ID_WITH_JACOBIANS)
-        // get translational jacobian, in world frame (dworld_velocity/du)
-        int getBodyJacobianTrans(const int body_index, mat3x* world_jac_trans) const;
-        // get rotational jacobian, in world frame (dworld_omega/du)
-        int getBodyJacobianRot(const int body_index, mat3x* world_jac_rot) const;
-        // get product of translational jacobian derivative * generatlized velocities
-        int getBodyDotJacobianTransU(const int body_index, vec3* world_dot_jac_trans_u) const;
-        // get product of rotational jacobian derivative * generatlized velocities
-        int getBodyDotJacobianRotU(const int body_index, vec3* world_dot_jac_rot_u) const;
-#endif // BT_ID_HAVE_MAT3X
+	// get translational jacobian, in world frame (dworld_velocity/du)
+	int getBodyJacobianTrans(const int body_index, mat3x* world_jac_trans) const;
+	// get rotational jacobian, in world frame (dworld_omega/du)
+	int getBodyJacobianRot(const int body_index, mat3x* world_jac_rot) const;
+	// get product of translational jacobian derivative * generatlized velocities
+	int getBodyDotJacobianTransU(const int body_index, vec3* world_dot_jac_trans_u) const;
+	// get product of rotational jacobian derivative * generatlized velocities
+	int getBodyDotJacobianRotU(const int body_index, vec3* world_dot_jac_rot_u) const;
+#endif  // BT_ID_HAVE_MAT3X
 
 	/// returns the (internal) index of body
 	/// @param body_index is the index of a body
@@ -256,21 +260,21 @@ public:
 	/// @param joint_type string naming the corresponding joint type
 	/// @return 0 on success, -1 on failure
 	int getJointTypeStr(const int body_index, const char** joint_type) const;
-    	/// get offset translation to parent body (see addBody)
+	/// get offset translation to parent body (see addBody)
 	/// @param body_index index of the body
 	/// @param r the offset translation (see above)
 	/// @return 0 on success, -1 on failure
-        int getParentRParentBodyRef(const int body_index, vec3* r) const;
+	int getParentRParentBodyRef(const int body_index, vec3* r) const;
 	/// get offset rotation to parent body (see addBody)
 	/// @param body_index index of the body
 	/// @param T the transform (see above)
 	/// @return 0 on success, -1 on failure
-        int getBodyTParentRef(const int body_index, mat33* T) const;
+	int getBodyTParentRef(const int body_index, mat33* T) const;
 	/// get axis of motion (see addBody)
 	/// @param body_index index of the body
 	/// @param axis the axis (see above)
 	/// @return 0 on success, -1 on failure
-        int getBodyAxisOfMotion(const int body_index, vec3* axis) const;
+	int getBodyAxisOfMotion(const int body_index, vec3* axis) const;
 	/// get offset for degrees of freedom of this body into the q-vector
 	/// @param body_index index of the body
 	/// @param q_offset offset the q vector
diff --git a/thirdparty/bullet/BulletInverseDynamics/details/IDEigenInterface.hpp b/thirdparty/bullet/BulletInverseDynamics/details/IDEigenInterface.hpp
index 836395cea22..fe4f1025133 100644
--- a/thirdparty/bullet/BulletInverseDynamics/details/IDEigenInterface.hpp
+++ b/thirdparty/bullet/BulletInverseDynamics/details/IDEigenInterface.hpp
@@ -1,8 +1,8 @@
 #ifndef INVDYNEIGENINTERFACE_HPP_
 #define INVDYNEIGENINTERFACE_HPP_
 #include "../IDConfig.hpp"
-namespace btInverseDynamics {
-
+namespace btInverseDynamics
+{
 #define BT_ID_HAVE_MAT3X
 
 #ifdef BT_USE_DOUBLE_PRECISION
@@ -19,18 +19,21 @@ typedef Eigen::Matrix<float, Eigen::Dynamic, Eigen::Dynamic, Eigen::DontAlign> m
 typedef Eigen::Matrix<float, 3, Eigen::Dynamic, Eigen::DontAlign> mat3x;
 #endif
 
-inline void resize(mat3x &m, Eigen::Index size) {
-    m.resize(3, size);
-    m.setZero();
+inline void resize(mat3x &m, Eigen::Index size)
+{
+	m.resize(3, size);
+	m.setZero();
 }
 
-inline void setMatxxElem(const idArrayIdx row, const idArrayIdx col, const idScalar val, matxx*m){
-    (*m)(row, col) = val;
+inline void setMatxxElem(const idArrayIdx row, const idArrayIdx col, const idScalar val, matxx *m)
+{
+	(*m)(row, col) = val;
 }
 
-inline void setMat3xElem(const idArrayIdx row, const idArrayIdx col, const idScalar val, mat3x*m){
-    (*m)(row, col) = val;
+inline void setMat3xElem(const idArrayIdx row, const idArrayIdx col, const idScalar val, mat3x *m)
+{
+	(*m)(row, col) = val;
 }
 
-}
+}  // namespace btInverseDynamics
 #endif  // INVDYNEIGENINTERFACE_HPP_
diff --git a/thirdparty/bullet/BulletInverseDynamics/details/IDLinearMathInterface.hpp b/thirdparty/bullet/BulletInverseDynamics/details/IDLinearMathInterface.hpp
index c179daeec66..0c398a37274 100644
--- a/thirdparty/bullet/BulletInverseDynamics/details/IDLinearMathInterface.hpp
+++ b/thirdparty/bullet/BulletInverseDynamics/details/IDLinearMathInterface.hpp
@@ -10,32 +10,37 @@
 #include "../../LinearMath/btMatrixX.h"
 #define BT_ID_HAVE_MAT3X
 
-namespace btInverseDynamics {
+namespace btInverseDynamics
+{
 class vec3;
 class vecx;
 class mat33;
 typedef btMatrixX<idScalar> matxx;
 
-class vec3 : public btVector3 {
+class vec3 : public btVector3
+{
 public:
 	vec3() : btVector3() {}
 	vec3(const btVector3& btv) { *this = btv; }
 	idScalar& operator()(int i) { return (*this)[i]; }
 	const idScalar& operator()(int i) const { return (*this)[i]; }
 	int size() const { return 3; }
-	const vec3& operator=(const btVector3& rhs) {
+	const vec3& operator=(const btVector3& rhs)
+	{
 		*static_cast<btVector3*>(this) = rhs;
 		return *this;
 	}
 };
 
-class mat33 : public btMatrix3x3 {
+class mat33 : public btMatrix3x3
+{
 public:
 	mat33() : btMatrix3x3() {}
 	mat33(const btMatrix3x3& btm) { *this = btm; }
 	idScalar& operator()(int i, int j) { return (*this)[i][j]; }
 	const idScalar& operator()(int i, int j) const { return (*this)[i][j]; }
-	const mat33& operator=(const btMatrix3x3& rhs) {
+	const mat33& operator=(const btMatrix3x3& rhs)
+	{
 		*static_cast<btMatrix3x3*>(this) = rhs;
 		return *this;
 	}
@@ -47,10 +52,12 @@ inline mat33 operator/(const mat33& a, const idScalar& s) { return a * (1.0 / s)
 
 inline mat33 operator*(const idScalar& s, const mat33& a) { return a * s; }
 
-class vecx : public btVectorX<idScalar> {
+class vecx : public btVectorX<idScalar>
+{
 public:
 	vecx(int size) : btVectorX<idScalar>(size) {}
-	const vecx& operator=(const btVectorX<idScalar>& rhs) {
+	const vecx& operator=(const btVectorX<idScalar>& rhs)
+	{
 		*static_cast<btVectorX<idScalar>*>(this) = rhs;
 		return *this;
 	}
@@ -66,43 +73,53 @@ public:
 	friend vecx operator/(const vecx& a, const idScalar& s);
 };
 
-inline vecx operator*(const vecx& a, const idScalar& s) {
+inline vecx operator*(const vecx& a, const idScalar& s)
+{
 	vecx result(a.size());
-	for (int i = 0; i < result.size(); i++) {
+	for (int i = 0; i < result.size(); i++)
+	{
 		result(i) = a(i) * s;
 	}
 	return result;
 }
 inline vecx operator*(const idScalar& s, const vecx& a) { return a * s; }
-inline vecx operator+(const vecx& a, const vecx& b) {
+inline vecx operator+(const vecx& a, const vecx& b)
+{
 	vecx result(a.size());
 	// TODO: error handling for a.size() != b.size()??
-	if (a.size() != b.size()) {
+	if (a.size() != b.size())
+	{
 		bt_id_error_message("size missmatch. a.size()= %d, b.size()= %d\n", a.size(), b.size());
 		abort();
 	}
-	for (int i = 0; i < a.size(); i++) {
+	for (int i = 0; i < a.size(); i++)
+	{
 		result(i) = a(i) + b(i);
 	}
 
 	return result;
 }
 
-inline vecx operator-(const vecx& a, const vecx& b) {
+inline vecx operator-(const vecx& a, const vecx& b)
+{
 	vecx result(a.size());
 	// TODO: error handling for a.size() != b.size()??
-	if (a.size() != b.size()) {
+	if (a.size() != b.size())
+	{
 		bt_id_error_message("size missmatch. a.size()= %d, b.size()= %d\n", a.size(), b.size());
 		abort();
 	}
-	for (int i = 0; i < a.size(); i++) {
+	for (int i = 0; i < a.size(); i++)
+	{
 		result(i) = a(i) - b(i);
 	}
 	return result;
 }
-inline vecx operator/(const vecx& a, const idScalar& s) {
+inline vecx operator/(const vecx& a, const idScalar& s)
+{
 	vecx result(a.size());
-	for (int i = 0; i < result.size(); i++) {
+	for (int i = 0; i < result.size(); i++)
+	{
 		result(i) = a(i) / s;
 	}
 
@@ -110,63 +127,76 @@ inline vecx operator/(const vecx& a, const idScalar& s) {
 }
 
 // use btMatrixX to implement 3xX matrix
-class mat3x : public matxx {
+class mat3x : public matxx
+{
 public:
-    mat3x(){}
-    mat3x(const mat3x&rhs) {
-        matxx::resize(rhs.rows(), rhs.cols());
-        *this = rhs;
-    }
-    mat3x(int rows, int cols): matxx(3,cols) {
-    }
-    void operator=(const mat3x& rhs) {
-	if (m_cols != rhs.m_cols) {
-            bt_id_error_message("size missmatch, cols= %d but rhs.cols= %d\n", cols(), rhs.cols());
-            abort();
+	mat3x() {}
+	mat3x(const mat3x& rhs)
+	{
+		matxx::resize(rhs.rows(), rhs.cols());
+		*this = rhs;
+	}
+	mat3x(int rows, int cols) : matxx(3, cols)
+	{
+	}
+	void operator=(const mat3x& rhs)
+	{
+		if (m_cols != rhs.m_cols)
+		{
+			bt_id_error_message("size missmatch, cols= %d but rhs.cols= %d\n", cols(), rhs.cols());
+			abort();
+		}
+		for (int i = 0; i < rows(); i++)
+		{
+			for (int k = 0; k < cols(); k++)
+			{
+				setElem(i, k, rhs(i, k));
+			}
+		}
+	}
+	void setZero()
+	{
+		matxx::setZero();
 	}
-        for(int i=0;i<rows();i++) {
-            for(int k=0;k<cols();k++) {
-                setElem(i,k,rhs(i,k));
-            }
-        }
-    }
-    void setZero() {
-        matxx::setZero();
-    }
 };
 
-
-inline vec3 operator*(const mat3x& a, const vecx& b) {
-    vec3 result;
-    if (a.cols() != b.size()) {
-        bt_id_error_message("size missmatch. a.cols()= %d, b.size()= %d\n", a.cols(), b.size());
-        abort();
-    }
-    result(0)=0.0;
-    result(1)=0.0;
-    result(2)=0.0;
-    for(int i=0;i<b.size();i++) {
-        for(int k=0;k<3;k++) {
-            result(k)+=a(k,i)*b(i);
-        }
-    }
-    return result;
+inline vec3 operator*(const mat3x& a, const vecx& b)
+{
+	vec3 result;
+	if (a.cols() != b.size())
+	{
+		bt_id_error_message("size missmatch. a.cols()= %d, b.size()= %d\n", a.cols(), b.size());
+		abort();
+	}
+	result(0) = 0.0;
+	result(1) = 0.0;
+	result(2) = 0.0;
+	for (int i = 0; i < b.size(); i++)
+	{
+		for (int k = 0; k < 3; k++)
+		{
+			result(k) += a(k, i) * b(i);
+		}
+	}
+	return result;
 }
 
-
-inline void resize(mat3x &m, idArrayIdx size) {
-    m.resize(3, size);
-    m.setZero();
+inline void resize(mat3x& m, idArrayIdx size)
+{
+	m.resize(3, size);
+	m.setZero();
 }
 
-inline void setMatxxElem(const idArrayIdx row, const idArrayIdx col, const idScalar val, matxx*m){
-    m->setElem(row, col, val);
+inline void setMatxxElem(const idArrayIdx row, const idArrayIdx col, const idScalar val, matxx* m)
+{
+	m->setElem(row, col, val);
 }
 
-inline void setMat3xElem(const idArrayIdx row, const idArrayIdx col, const idScalar val, mat3x*m){
-    m->setElem(row, col, val);
+inline void setMat3xElem(const idArrayIdx row, const idArrayIdx col, const idScalar val, mat3x* m)
+{
+	m->setElem(row, col, val);
 }
 
-}
+}  // namespace btInverseDynamics
 
 #endif  // IDLINEARMATHINTERFACE_HPP_
diff --git a/thirdparty/bullet/BulletInverseDynamics/details/IDMatVec.hpp b/thirdparty/bullet/BulletInverseDynamics/details/IDMatVec.hpp
index c89db5e123e..1c786095e7e 100644
--- a/thirdparty/bullet/BulletInverseDynamics/details/IDMatVec.hpp
+++ b/thirdparty/bullet/BulletInverseDynamics/details/IDMatVec.hpp
@@ -7,7 +7,8 @@
 #include "../IDConfig.hpp"
 #define BT_ID_HAVE_MAT3X
 
-namespace btInverseDynamics {
+namespace btInverseDynamics
+{
 class vec3;
 class vecx;
 class mat33;
@@ -17,7 +18,8 @@ class mat3x;
 /// This is a very basic implementation to enable stand-alone use of the library.
 /// The implementation is not really optimized and misses many features that you would
 /// want from a "fully featured" linear math library.
-class vec3 {
+class vec3
+{
 public:
 	idScalar& operator()(int i) { return m_data[i]; }
 	const idScalar& operator()(int i) const { return m_data[i]; }
@@ -40,7 +42,8 @@ private:
 	idScalar m_data[3];
 };
 
-class mat33 {
+class mat33
+{
 public:
 	idScalar& operator()(int i, int j) { return m_data[3 * i + j]; }
 	const idScalar& operator()(int i, int j) const { return m_data[3 * i + j]; }
@@ -62,9 +65,11 @@ private:
 	idScalar m_data[9];
 };
 
-class vecx {
+class vecx
+{
 public:
-	vecx(int size) : m_size(size) {
+	vecx(int size) : m_size(size)
+	{
 		m_data = static_cast<idScalar*>(idMalloc(sizeof(idScalar) * size));
 	}
 	~vecx() { idFree(m_data); }
@@ -85,14 +90,17 @@ private:
 	idScalar* m_data;
 };
 
-class matxx {
+class matxx
+{
 public:
-    matxx() {
-        m_data = 0x0;
-        m_cols=0;
-        m_rows=0;
-    }
-	matxx(int rows, int cols) : m_rows(rows), m_cols(cols) {
+	matxx()
+	{
+		m_data = 0x0;
+		m_cols = 0;
+		m_rows = 0;
+	}
+	matxx(int rows, int cols) : m_rows(rows), m_cols(cols)
+	{
 		m_data = static_cast<idScalar*>(idMalloc(sizeof(idScalar) * rows * cols));
 	}
 	~matxx() { idFree(m_data); }
@@ -107,69 +115,86 @@ private:
 	idScalar* m_data;
 };
 
-class mat3x {
+class mat3x
+{
 public:
-    mat3x() {
-        m_data = 0x0;
-        m_cols=0;
-    }
-    mat3x(const mat3x&rhs) {
-        m_cols=rhs.m_cols;
-        allocate();
-        *this = rhs;
-    }
-    mat3x(int rows, int cols): m_cols(cols) {
-        allocate();
-    };
-    void operator=(const mat3x& rhs) {
-	if (m_cols != rhs.m_cols) {
-            bt_id_error_message("size missmatch, cols= %d but rhs.cols= %d\n", cols(), rhs.cols());
-            abort();
+	mat3x()
+	{
+		m_data = 0x0;
+		m_cols = 0;
+	}
+	mat3x(const mat3x& rhs)
+	{
+		m_cols = rhs.m_cols;
+		allocate();
+		*this = rhs;
+	}
+	mat3x(int rows, int cols) : m_cols(cols)
+	{
+		allocate();
+	};
+	void operator=(const mat3x& rhs)
+	{
+		if (m_cols != rhs.m_cols)
+		{
+			bt_id_error_message("size missmatch, cols= %d but rhs.cols= %d\n", cols(), rhs.cols());
+			abort();
+		}
+		for (int i = 0; i < 3 * m_cols; i++)
+		{
+			m_data[i] = rhs.m_data[i];
+		}
 	}
-        for(int i=0;i<3*m_cols;i++) {
-            m_data[i] = rhs.m_data[i];
-        }
-    }
 
-    ~mat3x() {
-        free();
-    }
-    idScalar& operator()(int row, int col) { return m_data[row * m_cols + col]; }
-    const idScalar& operator()(int row, int col) const { return m_data[row * m_cols + col]; }
-    int rows() const { return m_rows; }
-    const int& cols() const { return m_cols; }
-    void resize(int rows, int cols) {
-        m_cols=cols;
-        free();
-        allocate();
-    }
-    void setZero() {
-        memset(m_data,0x0,sizeof(idScalar)*m_rows*m_cols);
-    }
-    // avoid operators that would allocate -- use functions sub/add/mul in IDMath.hpp instead
+	~mat3x()
+	{
+		free();
+	}
+	idScalar& operator()(int row, int col) { return m_data[row * m_cols + col]; }
+	const idScalar& operator()(int row, int col) const { return m_data[row * m_cols + col]; }
+	int rows() const { return m_rows; }
+	const int& cols() const { return m_cols; }
+	void resize(int rows, int cols)
+	{
+		m_cols = cols;
+		free();
+		allocate();
+	}
+	void setZero()
+	{
+		memset(m_data, 0x0, sizeof(idScalar) * m_rows * m_cols);
+	}
+	// avoid operators that would allocate -- use functions sub/add/mul in IDMath.hpp instead
 private:
-    void allocate(){m_data = static_cast<idScalar*>(idMalloc(sizeof(idScalar) * m_rows * m_cols));}
-    void free() { idFree(m_data);}
-    enum {m_rows=3};
-    int m_cols;
-    idScalar* m_data;
+	void allocate() { m_data = static_cast<idScalar*>(idMalloc(sizeof(idScalar) * m_rows * m_cols)); }
+	void free() { idFree(m_data); }
+	enum
+	{
+		m_rows = 3
+	};
+	int m_cols;
+	idScalar* m_data;
 };
 
-inline void resize(mat3x &m, idArrayIdx size) {
-    m.resize(3, size);
-    m.setZero();
+inline void resize(mat3x& m, idArrayIdx size)
+{
+	m.resize(3, size);
+	m.setZero();
 }
 
 //////////////////////////////////////////////////
 // Implementations
-inline const vec3& vec3::operator=(const vec3& rhs) {
-	if (&rhs != this) {
+inline const vec3& vec3::operator=(const vec3& rhs)
+{
+	if (&rhs != this)
+	{
 		memcpy(m_data, rhs.m_data, 3 * sizeof(idScalar));
 	}
 	return *this;
 }
 
-inline vec3 vec3::cross(const vec3& b) const {
+inline vec3 vec3::cross(const vec3& b) const
+{
 	vec3 result;
 	result.m_data[0] = m_data[1] * b.m_data[2] - m_data[2] * b.m_data[1];
 	result.m_data[1] = m_data[2] * b.m_data[0] - m_data[0] * b.m_data[2];
@@ -178,17 +203,21 @@ inline vec3 vec3::cross(const vec3& b) const {
 	return result;
 }
 
-inline idScalar vec3::dot(const vec3& b) const {
+inline idScalar vec3::dot(const vec3& b) const
+{
 	return m_data[0] * b.m_data[0] + m_data[1] * b.m_data[1] + m_data[2] * b.m_data[2];
 }
 
-inline const mat33& mat33::operator=(const mat33& rhs) {
-	if (&rhs != this) {
+inline const mat33& mat33::operator=(const mat33& rhs)
+{
+	if (&rhs != this)
+	{
 		memcpy(m_data, rhs.m_data, 9 * sizeof(idScalar));
 	}
 	return *this;
 }
-inline mat33 mat33::transpose() const {
+inline mat33 mat33::transpose() const
+{
 	mat33 result;
 	result.m_data[0] = m_data[0];
 	result.m_data[1] = m_data[3];
@@ -203,7 +232,8 @@ inline mat33 mat33::transpose() const {
 	return result;
 }
 
-inline mat33 operator*(const mat33& a, const mat33& b) {
+inline mat33 operator*(const mat33& a, const mat33& b)
+{
 	mat33 result;
 	result.m_data[0] =
 		a.m_data[0] * b.m_data[0] + a.m_data[1] * b.m_data[3] + a.m_data[2] * b.m_data[6];
@@ -227,22 +257,27 @@ inline mat33 operator*(const mat33& a, const mat33& b) {
 	return result;
 }
 
-inline const mat33& mat33::operator+=(const mat33& b) {
-	for (int i = 0; i < 9; i++) {
+inline const mat33& mat33::operator+=(const mat33& b)
+{
+	for (int i = 0; i < 9; i++)
+	{
 		m_data[i] += b.m_data[i];
 	}
 
 	return *this;
 }
 
-inline const mat33& mat33::operator-=(const mat33& b) {
-	for (int i = 0; i < 9; i++) {
+inline const mat33& mat33::operator-=(const mat33& b)
+{
+	for (int i = 0; i < 9; i++)
+	{
 		m_data[i] -= b.m_data[i];
 	}
 	return *this;
 }
 
-inline vec3 operator*(const mat33& a, const vec3& b) {
+inline vec3 operator*(const mat33& a, const vec3& b)
+{
 	vec3 result;
 
 	result.m_data[0] =
@@ -255,23 +290,29 @@ inline vec3 operator*(const mat33& a, const vec3& b) {
 	return result;
 }
 
-inline const vec3& vec3::operator+=(const vec3& b) {
-	for (int i = 0; i < 3; i++) {
+inline const vec3& vec3::operator+=(const vec3& b)
+{
+	for (int i = 0; i < 3; i++)
+	{
 		m_data[i] += b.m_data[i];
 	}
 	return *this;
 }
 
-inline const vec3& vec3::operator-=(const vec3& b) {
-	for (int i = 0; i < 3; i++) {
+inline const vec3& vec3::operator-=(const vec3& b)
+{
+	for (int i = 0; i < 3; i++)
+	{
 		m_data[i] -= b.m_data[i];
 	}
 	return *this;
 }
 
-inline mat33 operator*(const mat33& a, const idScalar& s) {
+inline mat33 operator*(const mat33& a, const idScalar& s)
+{
 	mat33 result;
-	for (int i = 0; i < 9; i++) {
+	for (int i = 0; i < 9; i++)
+	{
 		result.m_data[i] = a.m_data[i] * s;
 	}
 	return result;
@@ -279,137 +320,170 @@ inline mat33 operator*(const mat33& a, const idScalar& s) {
 
 inline mat33 operator*(const idScalar& s, const mat33& a) { return a * s; }
 
-inline vec3 operator*(const vec3& a, const idScalar& s) {
+inline vec3 operator*(const vec3& a, const idScalar& s)
+{
 	vec3 result;
-	for (int i = 0; i < 3; i++) {
+	for (int i = 0; i < 3; i++)
+	{
 		result.m_data[i] = a.m_data[i] * s;
 	}
 	return result;
 }
 inline vec3 operator*(const idScalar& s, const vec3& a) { return a * s; }
 
-inline mat33 operator+(const mat33& a, const mat33& b) {
+inline mat33 operator+(const mat33& a, const mat33& b)
+{
 	mat33 result;
-	for (int i = 0; i < 9; i++) {
+	for (int i = 0; i < 9; i++)
+	{
 		result.m_data[i] = a.m_data[i] + b.m_data[i];
 	}
 	return result;
 }
-inline vec3 operator+(const vec3& a, const vec3& b) {
+inline vec3 operator+(const vec3& a, const vec3& b)
+{
 	vec3 result;
-	for (int i = 0; i < 3; i++) {
+	for (int i = 0; i < 3; i++)
+	{
 		result.m_data[i] = a.m_data[i] + b.m_data[i];
 	}
 	return result;
 }
 
-inline mat33 operator-(const mat33& a, const mat33& b) {
+inline mat33 operator-(const mat33& a, const mat33& b)
+{
 	mat33 result;
-	for (int i = 0; i < 9; i++) {
+	for (int i = 0; i < 9; i++)
+	{
 		result.m_data[i] = a.m_data[i] - b.m_data[i];
 	}
 	return result;
 }
-inline vec3 operator-(const vec3& a, const vec3& b) {
+inline vec3 operator-(const vec3& a, const vec3& b)
+{
 	vec3 result;
-	for (int i = 0; i < 3; i++) {
+	for (int i = 0; i < 3; i++)
+	{
 		result.m_data[i] = a.m_data[i] - b.m_data[i];
 	}
 	return result;
 }
 
-inline mat33 operator/(const mat33& a, const idScalar& s) {
+inline mat33 operator/(const mat33& a, const idScalar& s)
+{
 	mat33 result;
-	for (int i = 0; i < 9; i++) {
+	for (int i = 0; i < 9; i++)
+	{
 		result.m_data[i] = a.m_data[i] / s;
 	}
 	return result;
 }
 
-inline vec3 operator/(const vec3& a, const idScalar& s) {
+inline vec3 operator/(const vec3& a, const idScalar& s)
+{
 	vec3 result;
-	for (int i = 0; i < 3; i++) {
+	for (int i = 0; i < 3; i++)
+	{
 		result.m_data[i] = a.m_data[i] / s;
 	}
 	return result;
 }
 
-inline const vecx& vecx::operator=(const vecx& rhs) {
-	if (size() != rhs.size()) {
+inline const vecx& vecx::operator=(const vecx& rhs)
+{
+	if (size() != rhs.size())
+	{
 		bt_id_error_message("size missmatch, size()= %d but rhs.size()= %d\n", size(), rhs.size());
 		abort();
 	}
-	if (&rhs != this) {
+	if (&rhs != this)
+	{
 		memcpy(m_data, rhs.m_data, rhs.size() * sizeof(idScalar));
 	}
 	return *this;
 }
-inline vecx operator*(const vecx& a, const idScalar& s) {
+inline vecx operator*(const vecx& a, const idScalar& s)
+{
 	vecx result(a.size());
-	for (int i = 0; i < result.size(); i++) {
+	for (int i = 0; i < result.size(); i++)
+	{
 		result.m_data[i] = a.m_data[i] * s;
 	}
 	return result;
 }
 inline vecx operator*(const idScalar& s, const vecx& a) { return a * s; }
-inline vecx operator+(const vecx& a, const vecx& b) {
+inline vecx operator+(const vecx& a, const vecx& b)
+{
 	vecx result(a.size());
 	// TODO: error handling for a.size() != b.size()??
-	if (a.size() != b.size()) {
+	if (a.size() != b.size())
+	{
 		bt_id_error_message("size missmatch. a.size()= %d, b.size()= %d\n", a.size(), b.size());
 		abort();
 	}
-	for (int i = 0; i < a.size(); i++) {
+	for (int i = 0; i < a.size(); i++)
+	{
 		result.m_data[i] = a.m_data[i] + b.m_data[i];
 	}
 
 	return result;
 }
-inline vecx operator-(const vecx& a, const vecx& b) {
+inline vecx operator-(const vecx& a, const vecx& b)
+{
 	vecx result(a.size());
 	// TODO: error handling for a.size() != b.size()??
-	if (a.size() != b.size()) {
+	if (a.size() != b.size())
+	{
 		bt_id_error_message("size missmatch. a.size()= %d, b.size()= %d\n", a.size(), b.size());
 		abort();
 	}
-	for (int i = 0; i < a.size(); i++) {
+	for (int i = 0; i < a.size(); i++)
+	{
 		result.m_data[i] = a.m_data[i] - b.m_data[i];
 	}
 	return result;
 }
-inline vecx operator/(const vecx& a, const idScalar& s) {
+inline vecx operator/(const vecx& a, const idScalar& s)
+{
 	vecx result(a.size());
-	for (int i = 0; i < result.size(); i++) {
+	for (int i = 0; i < result.size(); i++)
+	{
 		result.m_data[i] = a.m_data[i] / s;
 	}
 
 	return result;
 }
 
-inline vec3 operator*(const mat3x& a, const vecx& b) {
-    vec3 result;
-    if (a.cols() != b.size()) {
-        bt_id_error_message("size missmatch. a.cols()= %d, b.size()= %d\n", a.cols(), b.size());
-        abort();
-    }
-    result(0)=0.0;
-    result(1)=0.0;
-    result(2)=0.0;
-    for(int i=0;i<b.size();i++) {
-        for(int k=0;k<3;k++) {
-            result(k)+=a(k,i)*b(i);
-        }
-    }
-    return result;
+inline vec3 operator*(const mat3x& a, const vecx& b)
+{
+	vec3 result;
+	if (a.cols() != b.size())
+	{
+		bt_id_error_message("size missmatch. a.cols()= %d, b.size()= %d\n", a.cols(), b.size());
+		abort();
+	}
+	result(0) = 0.0;
+	result(1) = 0.0;
+	result(2) = 0.0;
+	for (int i = 0; i < b.size(); i++)
+	{
+		for (int k = 0; k < 3; k++)
+		{
+			result(k) += a(k, i) * b(i);
+		}
+	}
+	return result;
 }
 
-inline void setMatxxElem(const idArrayIdx row, const idArrayIdx col, const idScalar val, matxx*m){
-    (*m)(row, col) = val;
+inline void setMatxxElem(const idArrayIdx row, const idArrayIdx col, const idScalar val, matxx* m)
+{
+	(*m)(row, col) = val;
 }
 
-inline void setMat3xElem(const idArrayIdx row, const idArrayIdx col, const idScalar val, mat3x*m){
-    (*m)(row, col) = val;
+inline void setMat3xElem(const idArrayIdx row, const idArrayIdx col, const idScalar val, mat3x* m)
+{
+	(*m)(row, col) = val;
 }
 
-} // namespace btInverseDynamcis
+}  // namespace btInverseDynamics
 #endif
diff --git a/thirdparty/bullet/BulletInverseDynamics/details/MultiBodyTreeImpl.cpp b/thirdparty/bullet/BulletInverseDynamics/details/MultiBodyTreeImpl.cpp
index e8563238c32..befbc2e2a4a 100644
--- a/thirdparty/bullet/BulletInverseDynamics/details/MultiBodyTreeImpl.cpp
+++ b/thirdparty/bullet/BulletInverseDynamics/details/MultiBodyTreeImpl.cpp
@@ -1,16 +1,16 @@
 #include "MultiBodyTreeImpl.hpp"
 
-namespace btInverseDynamics {
-
+namespace btInverseDynamics
+{
 MultiBodyTree::MultiBodyImpl::MultiBodyImpl(int num_bodies_, int num_dofs_)
 	: m_num_bodies(num_bodies_), m_num_dofs(num_dofs_)
 #if (defined BT_ID_HAVE_MAT3X) && (defined BT_ID_WITH_JACOBIANS)
-        ,m_m3x(3,m_num_dofs)
+	  ,
+	  m_m3x(3, m_num_dofs)
 #endif
 {
-
 #if (defined BT_ID_HAVE_MAT3X) && (defined BT_ID_WITH_JACOBIANS)
-        resize(m_m3x,m_num_dofs);
+	resize(m_m3x, m_num_dofs);
 #endif
 	m_body_list.resize(num_bodies_);
 	m_parent_index.resize(num_bodies_);
@@ -23,8 +23,10 @@ MultiBodyTree::MultiBodyImpl::MultiBodyImpl(int num_bodies_, int num_dofs_)
 	m_world_gravity(2) = -9.8;
 }
 
-const char *MultiBodyTree::MultiBodyImpl::jointTypeToString(const JointType &type) const {
-	switch (type) {
+const char *MultiBodyTree::MultiBodyImpl::jointTypeToString(const JointType &type) const
+{
+	switch (type)
+	{
 		case FIXED:
 			return "fixed";
 		case REVOLUTE:
@@ -33,22 +35,28 @@ const char *MultiBodyTree::MultiBodyImpl::jointTypeToString(const JointType &typ
 			return "prismatic";
 		case FLOATING:
 			return "floating";
+		case SPHERICAL:
+			return "spherical";
 	}
 	return "error: invalid";
 }
 
-inline void indent(const int &level) {
+inline void indent(const int &level)
+{
 	for (int j = 0; j < level; j++)
 		id_printf("  ");  // indent
 }
 
-void MultiBodyTree::MultiBodyImpl::printTree() {
+void MultiBodyTree::MultiBodyImpl::printTree()
+{
 	id_printf("body %.2d[%s]: root\n", 0, jointTypeToString(m_body_list[0].m_joint_type));
 	printTree(0, 0);
 }
 
-void MultiBodyTree::MultiBodyImpl::printTreeData() {
-	for (idArrayIdx i = 0; i < m_body_list.size(); i++) {
+void MultiBodyTree::MultiBodyImpl::printTreeData()
+{
+	for (idArrayIdx i = 0; i < m_body_list.size(); i++)
+	{
 		RigidBody &body = m_body_list[i];
 		id_printf("body: %d\n", static_cast<int>(i));
 		id_printf("type: %s\n", jointTypeToString(body.m_joint_type));
@@ -59,19 +67,22 @@ void MultiBodyTree::MultiBodyImpl::printTreeData() {
 		id_printf("mass = %f\n", body.m_mass);
 		id_printf("mass * com = [%f %f %f]\n", body.m_body_mass_com(0), body.m_body_mass_com(1),
 				  body.m_body_mass_com(2));
-		id_printf("I_o= [%f %f %f;\n"
-				  "	  %f %f %f;\n"
-				  "	  %f %f %f]\n",
-				  body.m_body_I_body(0, 0), body.m_body_I_body(0, 1), body.m_body_I_body(0, 2),
-				  body.m_body_I_body(1, 0), body.m_body_I_body(1, 1), body.m_body_I_body(1, 2),
-				  body.m_body_I_body(2, 0), body.m_body_I_body(2, 1), body.m_body_I_body(2, 2));
+		id_printf(
+			"I_o= [%f %f %f;\n"
+			"	  %f %f %f;\n"
+			"	  %f %f %f]\n",
+			body.m_body_I_body(0, 0), body.m_body_I_body(0, 1), body.m_body_I_body(0, 2),
+			body.m_body_I_body(1, 0), body.m_body_I_body(1, 1), body.m_body_I_body(1, 2),
+			body.m_body_I_body(2, 0), body.m_body_I_body(2, 1), body.m_body_I_body(2, 2));
 
 		id_printf("parent_pos_parent_body_ref= [%f %f %f]\n", body.m_parent_pos_parent_body_ref(0),
 				  body.m_parent_pos_parent_body_ref(1), body.m_parent_pos_parent_body_ref(2));
 	}
 }
-int MultiBodyTree::MultiBodyImpl::bodyNumDoFs(const JointType &type) const {
-	switch (type) {
+int MultiBodyTree::MultiBodyImpl::bodyNumDoFs(const JointType &type) const
+{
+	switch (type)
+	{
 		case FIXED:
 			return 0;
 		case REVOLUTE:
@@ -79,12 +90,15 @@ int MultiBodyTree::MultiBodyImpl::bodyNumDoFs(const JointType &type) const {
 			return 1;
 		case FLOATING:
 			return 6;
+		case SPHERICAL:
+			return 3;
 	}
 	bt_id_error_message("unknown joint type %d\n", type);
 	return 0;
 }
 
-void MultiBodyTree::MultiBodyImpl::printTree(int index, int indentation) {
+void MultiBodyTree::MultiBodyImpl::printTree(int index, int indentation)
+{
 	// this is adapted from URDF2Bullet.
 	// TODO: fix this and print proper graph (similar to git --log --graph)
 	int num_children = m_child_indices[index].size();
@@ -92,7 +106,8 @@ void MultiBodyTree::MultiBodyImpl::printTree(int index, int indentation) {
 	indentation += 2;
 	int count = 0;
 
-	for (int i = 0; i < num_children; i++) {
+	for (int i = 0; i < num_children; i++)
+	{
 		int child_index = m_child_indices[index][i];
 		indent(indentation);
 		id_printf("body %.2d[%s]: %.2d is child no. %d (qi= %d .. %d) \n", index,
@@ -104,19 +119,23 @@ void MultiBodyTree::MultiBodyImpl::printTree(int index, int indentation) {
 	}
 }
 
-int MultiBodyTree::MultiBodyImpl::setGravityInWorldFrame(const vec3 &gravity) {
+int MultiBodyTree::MultiBodyImpl::setGravityInWorldFrame(const vec3 &gravity)
+{
 	m_world_gravity = gravity;
 	return 0;
 }
 
-int MultiBodyTree::MultiBodyImpl::generateIndexSets() {
+int MultiBodyTree::MultiBodyImpl::generateIndexSets()
+{
 	m_body_revolute_list.resize(0);
 	m_body_prismatic_list.resize(0);
 	int q_index = 0;
-	for (idArrayIdx i = 0; i < m_body_list.size(); i++) {
+	for (idArrayIdx i = 0; i < m_body_list.size(); i++)
+	{
 		RigidBody &body = m_body_list[i];
 		body.m_q_index = -1;
-		switch (body.m_joint_type) {
+		switch (body.m_joint_type)
+		{
 			case REVOLUTE:
 				m_body_revolute_list.push_back(i);
 				body.m_q_index = q_index;
@@ -135,28 +154,41 @@ int MultiBodyTree::MultiBodyImpl::generateIndexSets() {
 				body.m_q_index = q_index;
 				q_index += 6;
 				break;
+			case SPHERICAL:
+				m_body_spherical_list.push_back(i);
+				body.m_q_index = q_index;
+				q_index += 3;
+				break;
 			default:
 				bt_id_error_message("unsupported joint type %d\n", body.m_joint_type);
 				return -1;
 		}
 	}
 	// sanity check
-	if (q_index != m_num_dofs) {
+	if (q_index != m_num_dofs)
+	{
 		bt_id_error_message("internal error, q_index= %d but num_dofs %d\n", q_index, m_num_dofs);
 		return -1;
 	}
 
 	m_child_indices.resize(m_body_list.size());
 
-	for (idArrayIdx child = 1; child < m_parent_index.size(); child++) {
+	for (idArrayIdx child = 1; child < m_parent_index.size(); child++)
+	{
 		const int &parent = m_parent_index[child];
-		if (parent >= 0 && parent < (static_cast<int>(m_parent_index.size()) - 1)) {
+		if (parent >= 0 && parent < (static_cast<int>(m_parent_index.size()) - 1))
+		{
 			m_child_indices[parent].push_back(child);
-		} else {
-			if (-1 == parent) {
+		}
+		else
+		{
+			if (-1 == parent)
+			{
 				// multiple bodies are directly linked to the environment, ie, not a single root
 				bt_id_error_message("building index sets parent(%zu)= -1 (multiple roots)\n", child);
-			} else {
+			}
+			else
+			{
 				// should never happen
 				bt_id_error_message(
 					"building index sets. parent_index[%zu]= %d, but m_parent_index.size()= %d\n",
@@ -169,11 +201,14 @@ int MultiBodyTree::MultiBodyImpl::generateIndexSets() {
 	return 0;
 }
 
-void MultiBodyTree::MultiBodyImpl::calculateStaticData() {
+void MultiBodyTree::MultiBodyImpl::calculateStaticData()
+{
 	// relative kinematics that are not a function of q, u, dot_u
-	for (idArrayIdx i = 0; i < m_body_list.size(); i++) {
+	for (idArrayIdx i = 0; i < m_body_list.size(); i++)
+	{
 		RigidBody &body = m_body_list[i];
-		switch (body.m_joint_type) {
+		switch (body.m_joint_type)
+		{
 			case REVOLUTE:
 				body.m_parent_vel_rel(0) = 0;
 				body.m_parent_vel_rel(1) = 0;
@@ -212,41 +247,56 @@ void MultiBodyTree::MultiBodyImpl::calculateStaticData() {
 			case FLOATING:
 				// no static data
 				break;
+			case SPHERICAL:
+				//todo: review
+				body.m_parent_pos_parent_body = body.m_parent_pos_parent_body_ref;
+				body.m_parent_vel_rel(0) = 0;
+				body.m_parent_vel_rel(1) = 0;
+				body.m_parent_vel_rel(2) = 0;
+				body.m_parent_acc_rel(0) = 0;
+				body.m_parent_acc_rel(1) = 0;
+				body.m_parent_acc_rel(2) = 0;
+				break;
 		}
 
-       // resize & initialize jacobians to zero.
+			// resize & initialize jacobians to zero.
 #if (defined BT_ID_HAVE_MAT3X) && (defined BT_ID_WITH_JACOBIANS)
-        body.m_body_dot_Jac_T_u(0) = 0.0;
-        body.m_body_dot_Jac_T_u(1) = 0.0;
-        body.m_body_dot_Jac_T_u(2) = 0.0;
-        body.m_body_dot_Jac_R_u(0) = 0.0;
-        body.m_body_dot_Jac_R_u(1) = 0.0;
-        body.m_body_dot_Jac_R_u(2) = 0.0;
-        resize(body.m_body_Jac_T,m_num_dofs);
-        resize(body.m_body_Jac_R,m_num_dofs);
-        body.m_body_Jac_T.setZero();
-        body.m_body_Jac_R.setZero();
-#endif //
+		body.m_body_dot_Jac_T_u(0) = 0.0;
+		body.m_body_dot_Jac_T_u(1) = 0.0;
+		body.m_body_dot_Jac_T_u(2) = 0.0;
+		body.m_body_dot_Jac_R_u(0) = 0.0;
+		body.m_body_dot_Jac_R_u(1) = 0.0;
+		body.m_body_dot_Jac_R_u(2) = 0.0;
+		resize(body.m_body_Jac_T, m_num_dofs);
+		resize(body.m_body_Jac_R, m_num_dofs);
+		body.m_body_Jac_T.setZero();
+		body.m_body_Jac_R.setZero();
+#endif  //
 	}
 }
 
 int MultiBodyTree::MultiBodyImpl::calculateInverseDynamics(const vecx &q, const vecx &u,
-														   const vecx &dot_u, vecx *joint_forces) {
+														   const vecx &dot_u, vecx *joint_forces)
+{
 	if (q.size() != m_num_dofs || u.size() != m_num_dofs || dot_u.size() != m_num_dofs ||
-		joint_forces->size() != m_num_dofs) {
-		bt_id_error_message("wrong vector dimension. system has %d DOFs,\n"
-					  "but dim(q)= %d, dim(u)= %d, dim(dot_u)= %d, dim(joint_forces)= %d\n",
-					  m_num_dofs, static_cast<int>(q.size()), static_cast<int>(u.size()),
-					  static_cast<int>(dot_u.size()), static_cast<int>(joint_forces->size()));
+		joint_forces->size() != m_num_dofs)
+	{
+		bt_id_error_message(
+			"wrong vector dimension. system has %d DOFs,\n"
+			"but dim(q)= %d, dim(u)= %d, dim(dot_u)= %d, dim(joint_forces)= %d\n",
+			m_num_dofs, static_cast<int>(q.size()), static_cast<int>(u.size()),
+			static_cast<int>(dot_u.size()), static_cast<int>(joint_forces->size()));
 		return -1;
 	}
 	// 1. relative kinematics
-        if(-1 == calculateKinematics(q,u,dot_u, POSITION_VELOCITY_ACCELERATION)) {
-            bt_id_error_message("error in calculateKinematics\n");
-            return -1;
-        }
-        // 2. update contributions to equations of motion for every body.
-	for (idArrayIdx i = 0; i < m_body_list.size(); i++) {
+	if (-1 == calculateKinematics(q, u, dot_u, POSITION_VELOCITY_ACCELERATION))
+	{
+		bt_id_error_message("error in calculateKinematics\n");
+		return -1;
+	}
+	// 2. update contributions to equations of motion for every body.
+	for (idArrayIdx i = 0; i < m_body_list.size(); i++)
+	{
 		RigidBody &body = m_body_list[i];
 		// 3.4 update dynamic terms (rate of change of angular & linear momentum)
 		body.m_eom_lhs_rotational =
@@ -268,14 +318,16 @@ int MultiBodyTree::MultiBodyImpl::calculateInverseDynamics(const vecx &q, const
 	// Also, this enables adding zero weight bodies as a way to calculate frame poses
 	// for force elements, etc.
 
-	for (int body_idx = m_body_list.size() - 1; body_idx >= 0; body_idx--) {
+	for (int body_idx = m_body_list.size() - 1; body_idx >= 0; body_idx--)
+	{
 		// sum of forces and moments acting on this body from its children
 		vec3 sum_f_children;
 		vec3 sum_m_children;
 		setZero(sum_f_children);
 		setZero(sum_m_children);
 		for (idArrayIdx child_list_idx = 0; child_list_idx < m_child_indices[body_idx].size();
-			 child_list_idx++) {
+			 child_list_idx++)
+		{
 			const RigidBody &child = m_body_list[m_child_indices[body_idx][child_list_idx]];
 			vec3 child_joint_force_in_this_frame =
 				child.m_body_T_parent.transpose() * child.m_force_at_joint;
@@ -293,19 +345,22 @@ int MultiBodyTree::MultiBodyImpl::calculateInverseDynamics(const vecx &q, const
 	// These are the components of force_at_joint/moment_at_joint
 	// in the free directions given by Jac_JT/Jac_JR
 	// 4.1 revolute joints
-	for (idArrayIdx i = 0; i < m_body_revolute_list.size(); i++) {
+	for (idArrayIdx i = 0; i < m_body_revolute_list.size(); i++)
+	{
 		RigidBody &body = m_body_list[m_body_revolute_list[i]];
 		// (*joint_forces)(body.m_q_index) = body.m_Jac_JR.transpose() * body.m_moment_at_joint;
 		(*joint_forces)(body.m_q_index) = body.m_Jac_JR.dot(body.m_moment_at_joint);
 	}
 	// 4.2 for prismatic joints
-	for (idArrayIdx i = 0; i < m_body_prismatic_list.size(); i++) {
+	for (idArrayIdx i = 0; i < m_body_prismatic_list.size(); i++)
+	{
 		RigidBody &body = m_body_list[m_body_prismatic_list[i]];
 		// (*joint_forces)(body.m_q_index) = body.m_Jac_JT.transpose() * body.m_force_at_joint;
 		(*joint_forces)(body.m_q_index) = body.m_Jac_JT.dot(body.m_force_at_joint);
 	}
 	// 4.3 floating bodies (6-DoF joints)
-	for (idArrayIdx i = 0; i < m_body_floating_list.size(); i++) {
+	for (idArrayIdx i = 0; i < m_body_floating_list.size(); i++)
+	{
 		RigidBody &body = m_body_list[m_body_floating_list[i]];
 		(*joint_forces)(body.m_q_index + 0) = body.m_moment_at_joint(0);
 		(*joint_forces)(body.m_q_index + 1) = body.m_moment_at_joint(1);
@@ -316,84 +371,133 @@ int MultiBodyTree::MultiBodyImpl::calculateInverseDynamics(const vecx &q, const
 		(*joint_forces)(body.m_q_index + 5) = body.m_force_at_joint(2);
 	}
 
+	// 4.4 spherical bodies (3-DoF joints)
+	for (idArrayIdx i = 0; i < m_body_spherical_list.size(); i++)
+	{
+		//todo: review
+		RigidBody &body = m_body_list[m_body_spherical_list[i]];
+		(*joint_forces)(body.m_q_index + 0) = body.m_moment_at_joint(0);
+		(*joint_forces)(body.m_q_index + 1) = body.m_moment_at_joint(1);
+		(*joint_forces)(body.m_q_index + 2) = body.m_moment_at_joint(2);
+	}
 	return 0;
 }
 
-int MultiBodyTree::MultiBodyImpl::calculateKinematics(const vecx &q, const vecx &u, const vecx& dot_u,
-                                                      const KinUpdateType type) {
-    	if (q.size() != m_num_dofs || u.size() != m_num_dofs || dot_u.size() != m_num_dofs ) {
-		bt_id_error_message("wrong vector dimension. system has %d DOFs,\n"
-					  "but dim(q)= %d, dim(u)= %d, dim(dot_u)= %d\n",
-					  m_num_dofs, static_cast<int>(q.size()), static_cast<int>(u.size()),
-					  static_cast<int>(dot_u.size()));
+int MultiBodyTree::MultiBodyImpl::calculateKinematics(const vecx &q, const vecx &u, const vecx &dot_u,
+													  const KinUpdateType type)
+{
+	if (q.size() != m_num_dofs || u.size() != m_num_dofs || dot_u.size() != m_num_dofs)
+	{
+		bt_id_error_message(
+			"wrong vector dimension. system has %d DOFs,\n"
+			"but dim(q)= %d, dim(u)= %d, dim(dot_u)= %d\n",
+			m_num_dofs, static_cast<int>(q.size()), static_cast<int>(u.size()),
+			static_cast<int>(dot_u.size()));
+		return -1;
+	}
+	if (type != POSITION_ONLY && type != POSITION_VELOCITY && type != POSITION_VELOCITY_ACCELERATION)
+	{
+		bt_id_error_message("invalid type %d\n", type);
 		return -1;
 	}
-        if(type != POSITION_ONLY && type != POSITION_VELOCITY && type != POSITION_VELOCITY_ACCELERATION) {
-            bt_id_error_message("invalid type %d\n", type);
-            return -1;
-        }
 
 	// 1. update relative kinematics
 	// 1.1 for revolute
-	for (idArrayIdx i = 0; i < m_body_revolute_list.size(); i++) {
+	for (idArrayIdx i = 0; i < m_body_revolute_list.size(); i++)
+	{
 		RigidBody &body = m_body_list[m_body_revolute_list[i]];
 		mat33 T;
 		bodyTParentFromAxisAngle(body.m_Jac_JR, q(body.m_q_index), &T);
 		body.m_body_T_parent = T * body.m_body_T_parent_ref;
-                if(type >= POSITION_VELOCITY) {
-                    body.m_body_ang_vel_rel = body.m_Jac_JR * u(body.m_q_index);
-                }
-                if(type >= POSITION_VELOCITY_ACCELERATION) {
-                    body.m_body_ang_acc_rel = body.m_Jac_JR * dot_u(body.m_q_index);
-                }
+		if (type >= POSITION_VELOCITY)
+		{
+			body.m_body_ang_vel_rel = body.m_Jac_JR * u(body.m_q_index);
+		}
+		if (type >= POSITION_VELOCITY_ACCELERATION)
+		{
+			body.m_body_ang_acc_rel = body.m_Jac_JR * dot_u(body.m_q_index);
+		}
 	}
 	// 1.2 for prismatic
-	for (idArrayIdx i = 0; i < m_body_prismatic_list.size(); i++) {
+	for (idArrayIdx i = 0; i < m_body_prismatic_list.size(); i++)
+	{
 		RigidBody &body = m_body_list[m_body_prismatic_list[i]];
 		body.m_parent_pos_parent_body =
 			body.m_parent_pos_parent_body_ref + body.m_parent_Jac_JT * q(body.m_q_index);
-                if(type >= POSITION_VELOCITY) {
-                    body.m_parent_vel_rel =
-			body.m_body_T_parent_ref.transpose() * body.m_Jac_JT * u(body.m_q_index);
-                }
-                if(type >= POSITION_VELOCITY_ACCELERATION) {
-                    body.m_parent_acc_rel = body.m_parent_Jac_JT * dot_u(body.m_q_index);
-                }
+		if (type >= POSITION_VELOCITY)
+		{
+			body.m_parent_vel_rel =
+				body.m_body_T_parent_ref.transpose() * body.m_Jac_JT * u(body.m_q_index);
+		}
+		if (type >= POSITION_VELOCITY_ACCELERATION)
+		{
+			body.m_parent_acc_rel = body.m_parent_Jac_JT * dot_u(body.m_q_index);
+		}
 	}
 	// 1.3 fixed joints: nothing to do
 	// 1.4 6dof joints:
-	for (idArrayIdx i = 0; i < m_body_floating_list.size(); i++) {
+	for (idArrayIdx i = 0; i < m_body_floating_list.size(); i++)
+	{
 		RigidBody &body = m_body_list[m_body_floating_list[i]];
 
 		body.m_body_T_parent = transformZ(q(body.m_q_index + 2)) *
-							   transformY(q(body.m_q_index + 1)) * transformX(q(body.m_q_index));
+							   transformY(q(body.m_q_index + 1)) *
+							   transformX(q(body.m_q_index));
 		body.m_parent_pos_parent_body(0) = q(body.m_q_index + 3);
 		body.m_parent_pos_parent_body(1) = q(body.m_q_index + 4);
 		body.m_parent_pos_parent_body(2) = q(body.m_q_index + 5);
 		body.m_parent_pos_parent_body = body.m_body_T_parent * body.m_parent_pos_parent_body;
 
-                if(type >= POSITION_VELOCITY) {
-                    body.m_body_ang_vel_rel(0) = u(body.m_q_index + 0);
-                    body.m_body_ang_vel_rel(1) = u(body.m_q_index + 1);
-                    body.m_body_ang_vel_rel(2) = u(body.m_q_index + 2);
+		if (type >= POSITION_VELOCITY)
+		{
+			body.m_body_ang_vel_rel(0) = u(body.m_q_index + 0);
+			body.m_body_ang_vel_rel(1) = u(body.m_q_index + 1);
+			body.m_body_ang_vel_rel(2) = u(body.m_q_index + 2);
 
-                    body.m_parent_vel_rel(0) = u(body.m_q_index + 3);
-                    body.m_parent_vel_rel(1) = u(body.m_q_index + 4);
-                    body.m_parent_vel_rel(2) = u(body.m_q_index + 5);
+			body.m_parent_vel_rel(0) = u(body.m_q_index + 3);
+			body.m_parent_vel_rel(1) = u(body.m_q_index + 4);
+			body.m_parent_vel_rel(2) = u(body.m_q_index + 5);
 
-                    body.m_parent_vel_rel = body.m_body_T_parent.transpose() * body.m_parent_vel_rel;
-                }
-                if(type >= POSITION_VELOCITY_ACCELERATION) {
-                    body.m_body_ang_acc_rel(0) = dot_u(body.m_q_index + 0);
-                    body.m_body_ang_acc_rel(1) = dot_u(body.m_q_index + 1);
-                    body.m_body_ang_acc_rel(2) = dot_u(body.m_q_index + 2);
+			body.m_parent_vel_rel = body.m_body_T_parent.transpose() * body.m_parent_vel_rel;
+		}
+		if (type >= POSITION_VELOCITY_ACCELERATION)
+		{
+			body.m_body_ang_acc_rel(0) = dot_u(body.m_q_index + 0);
+			body.m_body_ang_acc_rel(1) = dot_u(body.m_q_index + 1);
+			body.m_body_ang_acc_rel(2) = dot_u(body.m_q_index + 2);
 
-                    body.m_parent_acc_rel(0) = dot_u(body.m_q_index + 3);
-                    body.m_parent_acc_rel(1) = dot_u(body.m_q_index + 4);
-                    body.m_parent_acc_rel(2) = dot_u(body.m_q_index + 5);
+			body.m_parent_acc_rel(0) = dot_u(body.m_q_index + 3);
+			body.m_parent_acc_rel(1) = dot_u(body.m_q_index + 4);
+			body.m_parent_acc_rel(2) = dot_u(body.m_q_index + 5);
 
-                    body.m_parent_acc_rel = body.m_body_T_parent.transpose() * body.m_parent_acc_rel;
-                }
+			body.m_parent_acc_rel = body.m_body_T_parent.transpose() * body.m_parent_acc_rel;
+		}
+	}
+	
+	for (idArrayIdx i = 0; i < m_body_spherical_list.size(); i++)
+	{
+		//todo: review
+		RigidBody &body = m_body_list[m_body_spherical_list[i]];
+
+		body.m_body_T_parent = transformZ(q(body.m_q_index + 2)) *
+							   transformY(q(body.m_q_index + 1)) *
+							   transformX(q(body.m_q_index));
+		body.m_parent_pos_parent_body = body.m_body_T_parent * body.m_parent_pos_parent_body;
+
+		if (type >= POSITION_VELOCITY)
+		{
+			body.m_body_ang_vel_rel(0) = u(body.m_q_index + 0);
+			body.m_body_ang_vel_rel(1) = u(body.m_q_index + 1);
+			body.m_body_ang_vel_rel(2) = u(body.m_q_index + 2);
+			body.m_parent_vel_rel = body.m_body_T_parent.transpose() * body.m_parent_vel_rel;
+		}
+		if (type >= POSITION_VELOCITY_ACCELERATION)
+		{
+			body.m_body_ang_acc_rel(0) = dot_u(body.m_q_index + 0);
+			body.m_body_ang_acc_rel(1) = dot_u(body.m_q_index + 1);
+			body.m_body_ang_acc_rel(2) = dot_u(body.m_q_index + 2);
+			body.m_parent_acc_rel = body.m_body_T_parent.transpose() * body.m_parent_acc_rel;
+		}
 	}
 
 	// 2. absolute kinematic quantities (vector valued)
@@ -410,26 +514,29 @@ int MultiBodyTree::MultiBodyImpl::calculateKinematics(const vecx &q, const vecx
 		body.m_body_pos = body.m_body_T_parent * body.m_parent_pos_parent_body;
 		body.m_body_T_world = body.m_body_T_parent;
 
-                if(type >= POSITION_VELOCITY) {
-                    // 3.2 update absolute velocities
-                    body.m_body_ang_vel = body.m_body_ang_vel_rel;
-                    body.m_body_vel = body.m_parent_vel_rel;
-                }
-                if(type >= POSITION_VELOCITY_ACCELERATION) {
-                    // 3.3 update absolute accelerations
-                    // NOTE: assumption: dot(J_JR) = 0; true here, but not for general joints
-                    body.m_body_ang_acc = body.m_body_ang_acc_rel;
-                    body.m_body_acc = body.m_body_T_parent * body.m_parent_acc_rel;
-                    // add gravitational acceleration to root body
-                    // this is an efficient way to add gravitational terms,
-                    // but it does mean that the kinematics are no longer
-                    // correct at the acceleration level
-                    // NOTE: To get correct acceleration kinematics, just set world_gravity to zero
-                    body.m_body_acc = body.m_body_acc - body.m_body_T_parent * m_world_gravity;
-                }
+		if (type >= POSITION_VELOCITY)
+		{
+			// 3.2 update absolute velocities
+			body.m_body_ang_vel = body.m_body_ang_vel_rel;
+			body.m_body_vel = body.m_parent_vel_rel;
+		}
+		if (type >= POSITION_VELOCITY_ACCELERATION)
+		{
+			// 3.3 update absolute accelerations
+			// NOTE: assumption: dot(J_JR) = 0; true here, but not for general joints
+			body.m_body_ang_acc = body.m_body_ang_acc_rel;
+			body.m_body_acc = body.m_body_T_parent * body.m_parent_acc_rel;
+			// add gravitational acceleration to root body
+			// this is an efficient way to add gravitational terms,
+			// but it does mean that the kinematics are no longer
+			// correct at the acceleration level
+			// NOTE: To get correct acceleration kinematics, just set world_gravity to zero
+			body.m_body_acc = body.m_body_acc - body.m_body_T_parent * m_world_gravity;
+		}
 	}
 
-	for (idArrayIdx i = 1; i < m_body_list.size(); i++) {
+	for (idArrayIdx i = 1; i < m_body_list.size(); i++)
+	{
 		RigidBody &body = m_body_list[i];
 		RigidBody &parent = m_body_list[m_parent_index[i]];
 		// 2.1 update absolute positions and orientations:
@@ -439,121 +546,159 @@ int MultiBodyTree::MultiBodyImpl::calculateKinematics(const vecx &q, const vecx
 			body.m_body_T_parent * (parent.m_body_pos + body.m_parent_pos_parent_body);
 		body.m_body_T_world = body.m_body_T_parent * parent.m_body_T_world;
 
-                if(type >= POSITION_VELOCITY) {
-                    // 2.2 update absolute velocities
-                    body.m_body_ang_vel =
-			body.m_body_T_parent * parent.m_body_ang_vel + body.m_body_ang_vel_rel;
+		if (type >= POSITION_VELOCITY)
+		{
+			// 2.2 update absolute velocities
+			body.m_body_ang_vel =
+				body.m_body_T_parent * parent.m_body_ang_vel + body.m_body_ang_vel_rel;
 
-                    body.m_body_vel =
-			body.m_body_T_parent *
-			(parent.m_body_vel + parent.m_body_ang_vel.cross(body.m_parent_pos_parent_body) +
-			 body.m_parent_vel_rel);
-                }
-                if(type >= POSITION_VELOCITY_ACCELERATION) {
-                    // 2.3 update absolute accelerations
-                    // NOTE: assumption: dot(J_JR) = 0; true here, but not for general joints
-                    body.m_body_ang_acc =
-			body.m_body_T_parent * parent.m_body_ang_acc -
-			body.m_body_ang_vel_rel.cross(body.m_body_T_parent * parent.m_body_ang_vel) +
-			body.m_body_ang_acc_rel;
-                    body.m_body_acc =
-			body.m_body_T_parent *
-			(parent.m_body_acc + parent.m_body_ang_acc.cross(body.m_parent_pos_parent_body) +
-			 parent.m_body_ang_vel.cross(parent.m_body_ang_vel.cross(body.m_parent_pos_parent_body)) +
-			 2.0 * parent.m_body_ang_vel.cross(body.m_parent_vel_rel) + body.m_parent_acc_rel);
-                }
+			body.m_body_vel =
+				body.m_body_T_parent *
+				(parent.m_body_vel + parent.m_body_ang_vel.cross(body.m_parent_pos_parent_body) +
+				 body.m_parent_vel_rel);
+		}
+		if (type >= POSITION_VELOCITY_ACCELERATION)
+		{
+			// 2.3 update absolute accelerations
+			// NOTE: assumption: dot(J_JR) = 0; true here, but not for general joints
+			body.m_body_ang_acc =
+				body.m_body_T_parent * parent.m_body_ang_acc -
+				body.m_body_ang_vel_rel.cross(body.m_body_T_parent * parent.m_body_ang_vel) +
+				body.m_body_ang_acc_rel;
+			body.m_body_acc =
+				body.m_body_T_parent *
+				(parent.m_body_acc + parent.m_body_ang_acc.cross(body.m_parent_pos_parent_body) +
+				 parent.m_body_ang_vel.cross(parent.m_body_ang_vel.cross(body.m_parent_pos_parent_body)) +
+				 2.0 * parent.m_body_ang_vel.cross(body.m_parent_vel_rel) + body.m_parent_acc_rel);
+		}
 	}
 
-    return 0;
+	return 0;
 }
 
 #if (defined BT_ID_HAVE_MAT3X) && (defined BT_ID_WITH_JACOBIANS)
 
-void MultiBodyTree::MultiBodyImpl::addRelativeJacobianComponent(RigidBody&body) {
-    const int& idx=body.m_q_index;
-    switch(body.m_joint_type) {
-    case FIXED:
-        break;
-    case REVOLUTE:
-        setMat3xElem(0,idx, body.m_Jac_JR(0), &body.m_body_Jac_R);
-        setMat3xElem(1,idx, body.m_Jac_JR(1), &body.m_body_Jac_R);
-        setMat3xElem(2,idx, body.m_Jac_JR(2), &body.m_body_Jac_R);
-        break;
-    case PRISMATIC:
-        setMat3xElem(0,idx, body.m_body_T_parent_ref(0,0)*body.m_Jac_JT(0)
-                     +body.m_body_T_parent_ref(1,0)*body.m_Jac_JT(1)
-                     +body.m_body_T_parent_ref(2,0)*body.m_Jac_JT(2),
-                     &body.m_body_Jac_T);
-        setMat3xElem(1,idx,body.m_body_T_parent_ref(0,1)*body.m_Jac_JT(0)
-                     +body.m_body_T_parent_ref(1,1)*body.m_Jac_JT(1)
-                     +body.m_body_T_parent_ref(2,1)*body.m_Jac_JT(2),
-                     &body.m_body_Jac_T);
-        setMat3xElem(2,idx, body.m_body_T_parent_ref(0,2)*body.m_Jac_JT(0)
-                     +body.m_body_T_parent_ref(1,2)*body.m_Jac_JT(1)
-                     +body.m_body_T_parent_ref(2,2)*body.m_Jac_JT(2),
-                     &body.m_body_Jac_T);
-        break;
-    case FLOATING:
-        setMat3xElem(0,idx+0, 1.0, &body.m_body_Jac_R);
-        setMat3xElem(1,idx+1, 1.0, &body.m_body_Jac_R);
-        setMat3xElem(2,idx+2, 1.0, &body.m_body_Jac_R);
-        // body_Jac_T = body_T_parent.transpose();
-        setMat3xElem(0,idx+3, body.m_body_T_parent(0,0), &body.m_body_Jac_T);
-        setMat3xElem(0,idx+4, body.m_body_T_parent(1,0), &body.m_body_Jac_T);
-        setMat3xElem(0,idx+5, body.m_body_T_parent(2,0), &body.m_body_Jac_T);
+void MultiBodyTree::MultiBodyImpl::addRelativeJacobianComponent(RigidBody &body)
+{
+	const int &idx = body.m_q_index;
+	switch (body.m_joint_type)
+	{
+		case FIXED:
+			break;
+		case REVOLUTE:
+			setMat3xElem(0, idx, body.m_Jac_JR(0), &body.m_body_Jac_R);
+			setMat3xElem(1, idx, body.m_Jac_JR(1), &body.m_body_Jac_R);
+			setMat3xElem(2, idx, body.m_Jac_JR(2), &body.m_body_Jac_R);
+			break;
+		case PRISMATIC:
+			setMat3xElem(0, idx, body.m_body_T_parent_ref(0, 0) * body.m_Jac_JT(0) + body.m_body_T_parent_ref(1, 0) * body.m_Jac_JT(1) + body.m_body_T_parent_ref(2, 0) * body.m_Jac_JT(2),
+						 &body.m_body_Jac_T);
+			setMat3xElem(1, idx, body.m_body_T_parent_ref(0, 1) * body.m_Jac_JT(0) + body.m_body_T_parent_ref(1, 1) * body.m_Jac_JT(1) + body.m_body_T_parent_ref(2, 1) * body.m_Jac_JT(2),
+						 &body.m_body_Jac_T);
+			setMat3xElem(2, idx, body.m_body_T_parent_ref(0, 2) * body.m_Jac_JT(0) + body.m_body_T_parent_ref(1, 2) * body.m_Jac_JT(1) + body.m_body_T_parent_ref(2, 2) * body.m_Jac_JT(2),
+						 &body.m_body_Jac_T);
+			break;
+		case FLOATING:
+			setMat3xElem(0, idx + 0, 1.0, &body.m_body_Jac_R);
+			setMat3xElem(1, idx + 1, 1.0, &body.m_body_Jac_R);
+			setMat3xElem(2, idx + 2, 1.0, &body.m_body_Jac_R);
+			// body_Jac_T = body_T_parent.transpose();
+			setMat3xElem(0, idx + 3, body.m_body_T_parent(0, 0), &body.m_body_Jac_T);
+			setMat3xElem(0, idx + 4, body.m_body_T_parent(1, 0), &body.m_body_Jac_T);
+			setMat3xElem(0, idx + 5, body.m_body_T_parent(2, 0), &body.m_body_Jac_T);
 
-        setMat3xElem(1,idx+3, body.m_body_T_parent(0,1), &body.m_body_Jac_T);
-        setMat3xElem(1,idx+4, body.m_body_T_parent(1,1), &body.m_body_Jac_T);
-        setMat3xElem(1,idx+5, body.m_body_T_parent(2,1), &body.m_body_Jac_T);
+			setMat3xElem(1, idx + 3, body.m_body_T_parent(0, 1), &body.m_body_Jac_T);
+			setMat3xElem(1, idx + 4, body.m_body_T_parent(1, 1), &body.m_body_Jac_T);
+			setMat3xElem(1, idx + 5, body.m_body_T_parent(2, 1), &body.m_body_Jac_T);
 
-        setMat3xElem(2,idx+3, body.m_body_T_parent(0,2), &body.m_body_Jac_T);
-        setMat3xElem(2,idx+4, body.m_body_T_parent(1,2), &body.m_body_Jac_T);
-        setMat3xElem(2,idx+5, body.m_body_T_parent(2,2), &body.m_body_Jac_T);
+			setMat3xElem(2, idx + 3, body.m_body_T_parent(0, 2), &body.m_body_Jac_T);
+			setMat3xElem(2, idx + 4, body.m_body_T_parent(1, 2), &body.m_body_Jac_T);
+			setMat3xElem(2, idx + 5, body.m_body_T_parent(2, 2), &body.m_body_Jac_T);
 
-        break;
-    }
+			break;
+		case SPHERICAL:
+			//todo: review
+			setMat3xElem(0, idx + 0, 1.0, &body.m_body_Jac_R);
+			setMat3xElem(1, idx + 1, 1.0, &body.m_body_Jac_R);
+			setMat3xElem(2, idx + 2, 1.0, &body.m_body_Jac_R);
+			break;
+	}
 }
 
-int MultiBodyTree::MultiBodyImpl::calculateJacobians(const vecx& q, const vecx& u, const KinUpdateType type) {
-    if (q.size() != m_num_dofs || u.size() != m_num_dofs) {
-        bt_id_error_message("wrong vector dimension. system has %d DOFs,\n"
-                      "but dim(q)= %d, dim(u)= %d\n",
-                      m_num_dofs, static_cast<int>(q.size()), static_cast<int>(u.size()));
-        return -1;
-    }
-    if(type != POSITION_ONLY && type != POSITION_VELOCITY) {
-        bt_id_error_message("invalid type %d\n", type);
-        return -1;
-    }
+int MultiBodyTree::MultiBodyImpl::calculateJacobians(const vecx &q, const vecx &u, const KinUpdateType type)
+{
+	if (q.size() != m_num_dofs || u.size() != m_num_dofs)
+	{
+		bt_id_error_message(
+			"wrong vector dimension. system has %d DOFs,\n"
+			"but dim(q)= %d, dim(u)= %d\n",
+			m_num_dofs, static_cast<int>(q.size()), static_cast<int>(u.size()));
+		return -1;
+	}
+	if (type != POSITION_ONLY && type != POSITION_VELOCITY)
+	{
+		bt_id_error_message("invalid type %d\n", type);
+		return -1;
+	}
 
-    addRelativeJacobianComponent(m_body_list[0]);
-    for (idArrayIdx i = 1; i < m_body_list.size(); i++) {
-        RigidBody &body = m_body_list[i];
-        RigidBody &parent = m_body_list[m_parent_index[i]];
+	addRelativeJacobianComponent(m_body_list[0]);
+	for (idArrayIdx i = 1; i < m_body_list.size(); i++)
+	{
+		RigidBody &body = m_body_list[i];
+		RigidBody &parent = m_body_list[m_parent_index[i]];
 
-        mul(body.m_body_T_parent, parent.m_body_Jac_R,& body.m_body_Jac_R);
-        body.m_body_Jac_T = parent.m_body_Jac_T;
-        mul(tildeOperator(body.m_parent_pos_parent_body),parent.m_body_Jac_R,&m_m3x);
-        sub(body.m_body_Jac_T,m_m3x, &body.m_body_Jac_T);
+		mul(body.m_body_T_parent, parent.m_body_Jac_R, &body.m_body_Jac_R);
+		body.m_body_Jac_T = parent.m_body_Jac_T;
+		mul(tildeOperator(body.m_parent_pos_parent_body), parent.m_body_Jac_R, &m_m3x);
+		sub(body.m_body_Jac_T, m_m3x, &body.m_body_Jac_T);
 
-        addRelativeJacobianComponent(body);
-        mul(body.m_body_T_parent, body.m_body_Jac_T,&body.m_body_Jac_T);
+		addRelativeJacobianComponent(body);
+		mul(body.m_body_T_parent, body.m_body_Jac_T, &body.m_body_Jac_T);
 
-        if(type >= POSITION_VELOCITY) {
-            body.m_body_dot_Jac_R_u = body.m_body_T_parent * parent.m_body_dot_Jac_R_u -
-                body.m_body_ang_vel_rel.cross(body.m_body_T_parent * parent.m_body_ang_vel);
-            body.m_body_dot_Jac_T_u = body.m_body_T_parent *
-                (parent.m_body_dot_Jac_T_u + parent.m_body_dot_Jac_R_u.cross(body.m_parent_pos_parent_body) +
-                 parent.m_body_ang_vel.cross(parent.m_body_ang_vel.cross(body.m_parent_pos_parent_body)) +
-                 2.0 * parent.m_body_ang_vel.cross(body.m_parent_vel_rel));
-        }
-    }
-    return 0;
+		if (type >= POSITION_VELOCITY)
+		{
+			body.m_body_dot_Jac_R_u = body.m_body_T_parent * parent.m_body_dot_Jac_R_u -
+									  body.m_body_ang_vel_rel.cross(body.m_body_T_parent * parent.m_body_ang_vel);
+			body.m_body_dot_Jac_T_u = body.m_body_T_parent *
+									  (parent.m_body_dot_Jac_T_u + parent.m_body_dot_Jac_R_u.cross(body.m_parent_pos_parent_body) +
+									   parent.m_body_ang_vel.cross(parent.m_body_ang_vel.cross(body.m_parent_pos_parent_body)) +
+									   2.0 * parent.m_body_ang_vel.cross(body.m_parent_vel_rel));
+		}
+	}
+	return 0;
 }
 #endif
 
-static inline void setSixDoFJacobians(const int dof, vec3 &Jac_JR, vec3 &Jac_JT) {
-	switch (dof) {
+static inline void setThreeDoFJacobians(const int dof, vec3 &Jac_JR, vec3 &Jac_JT)
+{
+	switch (dof)
+	{
+		// rotational part
+		case 0:
+			Jac_JR(0) = 1;
+			Jac_JR(1) = 0;
+			Jac_JR(2) = 0;
+			setZero(Jac_JT);
+			break;
+		case 1:
+			Jac_JR(0) = 0;
+			Jac_JR(1) = 1;
+			Jac_JR(2) = 0;
+			setZero(Jac_JT);
+			break;
+		case 2:
+			Jac_JR(0) = 0;
+			Jac_JR(1) = 0;
+			Jac_JR(2) = 1;
+			setZero(Jac_JT);
+			break;
+	}
+}
+
+static inline void setSixDoFJacobians(const int dof, vec3 &Jac_JR, vec3 &Jac_JT)
+{
+	switch (dof)
+	{
 		// rotational part
 		case 0:
 			Jac_JR(0) = 1;
@@ -595,8 +740,10 @@ static inline void setSixDoFJacobians(const int dof, vec3 &Jac_JR, vec3 &Jac_JT)
 	}
 }
 
-static inline int jointNumDoFs(const JointType &type) {
-	switch (type) {
+static inline int jointNumDoFs(const JointType &type)
+{
+	switch (type)
+	{
 		case FIXED:
 			return 0;
 		case REVOLUTE:
@@ -604,6 +751,8 @@ static inline int jointNumDoFs(const JointType &type) {
 			return 1;
 		case FLOATING:
 			return 6;
+		case SPHERICAL:
+			return 3;
 	}
 	// this should never happen
 	bt_id_error_message("invalid joint type\n");
@@ -615,37 +764,45 @@ static inline int jointNumDoFs(const JointType &type) {
 int MultiBodyTree::MultiBodyImpl::calculateMassMatrix(const vecx &q, const bool update_kinematics,
 													  const bool initialize_matrix,
 													  const bool set_lower_triangular_matrix,
-													  matxx *mass_matrix) {
-// This calculates the joint space mass matrix for the multibody system.
-// The algorithm is essentially an implementation of "method 3"
-// in "Efficient Dynamic Simulation of Robotic Mechanisms" (Walker and Orin, 1982)
-// (Later named "Composite Rigid Body Algorithm" by Featherstone).
-//
-// This implementation, however, handles branched systems and uses a formulation centered
-// on the origin of the body-fixed frame to avoid re-computing various quantities at the com.
+													  matxx *mass_matrix)
+{
+	// This calculates the joint space mass matrix for the multibody system.
+	// The algorithm is essentially an implementation of "method 3"
+	// in "Efficient Dynamic Simulation of Robotic Mechanisms" (Walker and Orin, 1982)
+	// (Later named "Composite Rigid Body Algorithm" by Featherstone).
+	//
+	// This implementation, however, handles branched systems and uses a formulation centered
+	// on the origin of the body-fixed frame to avoid re-computing various quantities at the com.
 
 	if (q.size() != m_num_dofs || mass_matrix->rows() != m_num_dofs ||
-		mass_matrix->cols() != m_num_dofs) {
-		bt_id_error_message("Dimension error. System has %d DOFs,\n"
-					  "but dim(q)= %d, dim(mass_matrix)= %d x %d\n",
-					  m_num_dofs, static_cast<int>(q.size()), static_cast<int>(mass_matrix->rows()),
-					  static_cast<int>(mass_matrix->cols()));
+		mass_matrix->cols() != m_num_dofs)
+	{
+		bt_id_error_message(
+			"Dimension error. System has %d DOFs,\n"
+			"but dim(q)= %d, dim(mass_matrix)= %d x %d\n",
+			m_num_dofs, static_cast<int>(q.size()), static_cast<int>(mass_matrix->rows()),
+			static_cast<int>(mass_matrix->cols()));
 		return -1;
 	}
 
 	// TODO add optimized zeroing function?
-	if (initialize_matrix) {
-		for (int i = 0; i < m_num_dofs; i++) {
-			for (int j = 0; j < m_num_dofs; j++) {
-                            setMatxxElem(i, j, 0.0, mass_matrix);
+	if (initialize_matrix)
+	{
+		for (int i = 0; i < m_num_dofs; i++)
+		{
+			for (int j = 0; j < m_num_dofs; j++)
+			{
+				setMatxxElem(i, j, 0.0, mass_matrix);
 			}
 		}
 	}
 
-	if (update_kinematics) {
+	if (update_kinematics)
+	{
 		// 1. update relative kinematics
 		// 1.1 for revolute joints
-		for (idArrayIdx i = 0; i < m_body_revolute_list.size(); i++) {
+		for (idArrayIdx i = 0; i < m_body_revolute_list.size(); i++)
+		{
 			RigidBody &body = m_body_list[m_body_revolute_list[i]];
 			// from reference orientation (q=0) of body-fixed frame to current orientation
 			mat33 body_T_body_ref;
@@ -653,7 +810,8 @@ int MultiBodyTree::MultiBodyImpl::calculateMassMatrix(const vecx &q, const bool
 			body.m_body_T_parent = body_T_body_ref * body.m_body_T_parent_ref;
 		}
 		// 1.2 for prismatic joints
-		for (idArrayIdx i = 0; i < m_body_prismatic_list.size(); i++) {
+		for (idArrayIdx i = 0; i < m_body_prismatic_list.size(); i++)
+		{
 			RigidBody &body = m_body_list[m_body_prismatic_list[i]];
 			// body.m_body_T_parent= fixed
 			body.m_parent_pos_parent_body =
@@ -661,7 +819,8 @@ int MultiBodyTree::MultiBodyImpl::calculateMassMatrix(const vecx &q, const bool
 		}
 		// 1.3 fixed joints: nothing to do
 		// 1.4 6dof joints:
-		for (idArrayIdx i = 0; i < m_body_floating_list.size(); i++) {
+		for (idArrayIdx i = 0; i < m_body_floating_list.size(); i++)
+		{
 			RigidBody &body = m_body_list[m_body_floating_list[i]];
 
 			body.m_body_T_parent = transformZ(q(body.m_q_index + 2)) *
@@ -674,7 +833,8 @@ int MultiBodyTree::MultiBodyImpl::calculateMassMatrix(const vecx &q, const bool
 			body.m_parent_pos_parent_body = body.m_body_T_parent * body.m_parent_pos_parent_body;
 		}
 	}
-	for (int i = m_body_list.size() - 1; i >= 0; i--) {
+	for (int i = m_body_list.size() - 1; i >= 0; i--)
+	{
 		RigidBody &body = m_body_list[i];
 		// calculate mass, center of mass and inertia of "composite rigid body",
 		// ie, sub-tree starting at current body
@@ -682,7 +842,8 @@ int MultiBodyTree::MultiBodyImpl::calculateMassMatrix(const vecx &q, const bool
 		body.m_body_subtree_mass_com = body.m_body_mass_com;
 		body.m_body_subtree_I_body = body.m_body_I_body;
 
-		for (idArrayIdx c = 0; c < m_child_indices[i].size(); c++) {
+		for (idArrayIdx c = 0; c < m_child_indices[i].size(); c++)
+		{
 			RigidBody &child = m_body_list[m_child_indices[i][c]];
 			mat33 body_T_child = child.m_body_T_parent.transpose();
 
@@ -692,7 +853,8 @@ int MultiBodyTree::MultiBodyImpl::calculateMassMatrix(const vecx &q, const bool
 			body.m_body_subtree_I_body +=
 				body_T_child * child.m_body_subtree_I_body * child.m_body_T_parent;
 
-			if (child.m_subtree_mass > 0) {
+			if (child.m_subtree_mass > 0)
+			{
 				// Shift the reference point for the child subtree inertia using the
 				// Huygens-Steiner ("parallel axis") theorem.
 				// (First shift from child origin to child com, then from there to this body's
@@ -707,7 +869,8 @@ int MultiBodyTree::MultiBodyImpl::calculateMassMatrix(const vecx &q, const bool
 		}
 	}
 
-	for (int i = m_body_list.size() - 1; i >= 0; i--) {
+	for (int i = m_body_list.size() - 1; i >= 0; i--)
+	{
 		const RigidBody &body = m_body_list[i];
 
 		// determine DoF-range for body
@@ -717,11 +880,18 @@ int MultiBodyTree::MultiBodyImpl::calculateMassMatrix(const vecx &q, const bool
 		// local joint jacobians (ok as is for 1-DoF joints)
 		vec3 Jac_JR = body.m_Jac_JR;
 		vec3 Jac_JT = body.m_Jac_JT;
-		for (int col = q_index_max; col >= q_index_min; col--) {
+		for (int col = q_index_max; col >= q_index_min; col--)
+		{
 			// set jacobians for 6-DoF joints
-			if (FLOATING == body.m_joint_type) {
+			if (FLOATING == body.m_joint_type)
+			{
 				setSixDoFJacobians(col - q_index_min, Jac_JR, Jac_JT);
 			}
+			if (SPHERICAL == body.m_joint_type)
+			{
+				//todo: review
+				setThreeDoFJacobians(col - q_index_min, Jac_JR, Jac_JT);
+			}
 
 			vec3 body_eom_rot =
 				body.m_body_subtree_I_body * Jac_JR + body.m_body_subtree_mass_com.cross(Jac_JT);
@@ -732,19 +902,27 @@ int MultiBodyTree::MultiBodyImpl::calculateMassMatrix(const vecx &q, const bool
 			// rest of the mass matrix column upwards
 			{
 				// 1. for multi-dof joints, rest of the dofs of this body
-				for (int row = col - 1; row >= q_index_min; row--) {
-					if (FLOATING != body.m_joint_type) {
-						bt_id_error_message("??\n");
-						return -1;
+				for (int row = col - 1; row >= q_index_min; row--)
+				{
+					if (SPHERICAL == body.m_joint_type)
+					{
+						//todo: review
+						setThreeDoFJacobians(row - q_index_min, Jac_JR, Jac_JT);
+						const double Mrc = Jac_JR.dot(body_eom_rot) + Jac_JT.dot(body_eom_trans);
+						setMatxxElem(col, row, Mrc, mass_matrix);
+					}
+					if (FLOATING == body.m_joint_type)
+					{
+						setSixDoFJacobians(row - q_index_min, Jac_JR, Jac_JT);
+						const double Mrc = Jac_JR.dot(body_eom_rot) + Jac_JT.dot(body_eom_trans);
+						setMatxxElem(col, row, Mrc, mass_matrix);
 					}
-					setSixDoFJacobians(row - q_index_min, Jac_JR, Jac_JT);
-					const double Mrc = Jac_JR.dot(body_eom_rot) + Jac_JT.dot(body_eom_trans);
-					setMatxxElem(col, row, Mrc, mass_matrix);
 				}
 				// 2. ancestor dofs
 				int child_idx = i;
 				int parent_idx = m_parent_index[i];
-				while (parent_idx >= 0) {
+				while (parent_idx >= 0)
+				{
 					const RigidBody &child_body = m_body_list[child_idx];
 					const RigidBody &parent_body = m_body_list[parent_idx];
 
@@ -758,9 +936,16 @@ int MultiBodyTree::MultiBodyImpl::calculateMassMatrix(const vecx &q, const bool
 						parent_body_q_index_min + jointNumDoFs(parent_body.m_joint_type) - 1;
 					vec3 Jac_JR = parent_body.m_Jac_JR;
 					vec3 Jac_JT = parent_body.m_Jac_JT;
-					for (int row = parent_body_q_index_max; row >= parent_body_q_index_min; row--) {
+					for (int row = parent_body_q_index_max; row >= parent_body_q_index_min; row--)
+					{
+						if (SPHERICAL == parent_body.m_joint_type)
+						{
+							//todo: review
+							setThreeDoFJacobians(row - parent_body_q_index_min, Jac_JR, Jac_JT);
+						}
 						// set jacobians for 6-DoF joints
-						if (FLOATING == parent_body.m_joint_type) {
+						if (FLOATING == parent_body.m_joint_type)
+						{
 							setSixDoFJacobians(row - parent_body_q_index_min, Jac_JR, Jac_JT);
 						}
 						const double Mrc = Jac_JR.dot(body_eom_rot) + Jac_JT.dot(body_eom_trans);
@@ -774,10 +959,13 @@ int MultiBodyTree::MultiBodyImpl::calculateMassMatrix(const vecx &q, const bool
 		}
 	}
 
-	if (set_lower_triangular_matrix) {
-		for (int col = 0; col < m_num_dofs; col++) {
-			for (int row = 0; row < col; row++) {
-                            setMatxxElem(row, col, (*mass_matrix)(col, row), mass_matrix);
+	if (set_lower_triangular_matrix)
+	{
+		for (int col = 0; col < m_num_dofs; col++)
+		{
+			for (int row = 0; row < col; row++)
+			{
+				setMatxxElem(row, col, (*mass_matrix)(col, row), mass_matrix);
 			}
 		}
 	}
@@ -785,76 +973,91 @@ int MultiBodyTree::MultiBodyImpl::calculateMassMatrix(const vecx &q, const bool
 }
 
 // utility macro
-#define CHECK_IF_BODY_INDEX_IS_VALID(index)														\
-	do {																						   \
-		if (index < 0 || index >= m_num_bodies) {												  \
-			bt_id_error_message("invalid index %d (num_bodies= %d)\n", index, m_num_bodies);			 \
-			return -1;																			 \
-		}																						  \
+#define CHECK_IF_BODY_INDEX_IS_VALID(index)                                                  \
+	do                                                                                       \
+	{                                                                                        \
+		if (index < 0 || index >= m_num_bodies)                                              \
+		{                                                                                    \
+			bt_id_error_message("invalid index %d (num_bodies= %d)\n", index, m_num_bodies); \
+			return -1;                                                                       \
+		}                                                                                    \
 	} while (0)
 
-int MultiBodyTree::MultiBodyImpl::getParentIndex(const int body_index, int *p) {
+int MultiBodyTree::MultiBodyImpl::getParentIndex(const int body_index, int *p)
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	*p = m_parent_index[body_index];
 	return 0;
 }
 
-int MultiBodyTree::MultiBodyImpl::getUserInt(const int body_index, int *user_int) const {
+int MultiBodyTree::MultiBodyImpl::getUserInt(const int body_index, int *user_int) const
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	*user_int = m_user_int[body_index];
 	return 0;
 }
-int MultiBodyTree::MultiBodyImpl::getUserPtr(const int body_index, void **user_ptr) const {
+int MultiBodyTree::MultiBodyImpl::getUserPtr(const int body_index, void **user_ptr) const
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	*user_ptr = m_user_ptr[body_index];
 	return 0;
 }
 
-int MultiBodyTree::MultiBodyImpl::setUserInt(const int body_index, const int user_int) {
+int MultiBodyTree::MultiBodyImpl::setUserInt(const int body_index, const int user_int)
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	m_user_int[body_index] = user_int;
 	return 0;
 }
 
-int MultiBodyTree::MultiBodyImpl::setUserPtr(const int body_index, void *const user_ptr) {
+int MultiBodyTree::MultiBodyImpl::setUserPtr(const int body_index, void *const user_ptr)
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	m_user_ptr[body_index] = user_ptr;
 	return 0;
 }
 
-int MultiBodyTree::MultiBodyImpl::getBodyOrigin(int body_index, vec3 *world_origin) const {
+int MultiBodyTree::MultiBodyImpl::getBodyOrigin(int body_index, vec3 *world_origin) const
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	const RigidBody &body = m_body_list[body_index];
 	*world_origin = body.m_body_T_world.transpose() * body.m_body_pos;
 	return 0;
 }
 
-int MultiBodyTree::MultiBodyImpl::getBodyCoM(int body_index, vec3 *world_com) const {
+int MultiBodyTree::MultiBodyImpl::getBodyCoM(int body_index, vec3 *world_com) const
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	const RigidBody &body = m_body_list[body_index];
-	if (body.m_mass > 0) {
+	if (body.m_mass > 0)
+	{
 		*world_com = body.m_body_T_world.transpose() *
 					 (body.m_body_pos + body.m_body_mass_com / body.m_mass);
-	} else {
+	}
+	else
+	{
 		*world_com = body.m_body_T_world.transpose() * (body.m_body_pos);
 	}
 	return 0;
 }
 
-int MultiBodyTree::MultiBodyImpl::getBodyTransform(int body_index, mat33 *world_T_body) const {
+int MultiBodyTree::MultiBodyImpl::getBodyTransform(int body_index, mat33 *world_T_body) const
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	const RigidBody &body = m_body_list[body_index];
 	*world_T_body = body.m_body_T_world.transpose();
 	return 0;
 }
-int MultiBodyTree::MultiBodyImpl::getBodyAngularVelocity(int body_index, vec3 *world_omega) const {
+int MultiBodyTree::MultiBodyImpl::getBodyAngularVelocity(int body_index, vec3 *world_omega) const
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	const RigidBody &body = m_body_list[body_index];
 	*world_omega = body.m_body_T_world.transpose() * body.m_body_ang_vel;
 	return 0;
 }
 int MultiBodyTree::MultiBodyImpl::getBodyLinearVelocity(int body_index,
-														vec3 *world_velocity) const {
+														vec3 *world_velocity) const
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	const RigidBody &body = m_body_list[body_index];
 	*world_velocity = body.m_body_T_world.transpose() * body.m_body_vel;
@@ -862,13 +1065,17 @@ int MultiBodyTree::MultiBodyImpl::getBodyLinearVelocity(int body_index,
 }
 
 int MultiBodyTree::MultiBodyImpl::getBodyLinearVelocityCoM(int body_index,
-														   vec3 *world_velocity) const {
+														   vec3 *world_velocity) const
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	const RigidBody &body = m_body_list[body_index];
 	vec3 com;
-	if (body.m_mass > 0) {
+	if (body.m_mass > 0)
+	{
 		com = body.m_body_mass_com / body.m_mass;
-	} else {
+	}
+	else
+	{
 		com(0) = 0;
 		com(1) = 0;
 		com(2) = 0;
@@ -880,149 +1087,173 @@ int MultiBodyTree::MultiBodyImpl::getBodyLinearVelocityCoM(int body_index,
 }
 
 int MultiBodyTree::MultiBodyImpl::getBodyAngularAcceleration(int body_index,
-															 vec3 *world_dot_omega) const {
+															 vec3 *world_dot_omega) const
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	const RigidBody &body = m_body_list[body_index];
 	*world_dot_omega = body.m_body_T_world.transpose() * body.m_body_ang_acc;
 	return 0;
 }
 int MultiBodyTree::MultiBodyImpl::getBodyLinearAcceleration(int body_index,
-															vec3 *world_acceleration) const {
+															vec3 *world_acceleration) const
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	const RigidBody &body = m_body_list[body_index];
 	*world_acceleration = body.m_body_T_world.transpose() * body.m_body_acc;
 	return 0;
 }
 
-int MultiBodyTree::MultiBodyImpl::getJointType(const int body_index, JointType *joint_type) const {
+int MultiBodyTree::MultiBodyImpl::getJointType(const int body_index, JointType *joint_type) const
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	*joint_type = m_body_list[body_index].m_joint_type;
 	return 0;
 }
 
 int MultiBodyTree::MultiBodyImpl::getJointTypeStr(const int body_index,
-												  const char **joint_type) const {
+												  const char **joint_type) const
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	*joint_type = jointTypeToString(m_body_list[body_index].m_joint_type);
 	return 0;
 }
 
-int MultiBodyTree::MultiBodyImpl::getParentRParentBodyRef(const int body_index, vec3* r)  const{
-    CHECK_IF_BODY_INDEX_IS_VALID(body_index);
-    *r=m_body_list[body_index].m_parent_pos_parent_body_ref;
-    return 0;
+int MultiBodyTree::MultiBodyImpl::getParentRParentBodyRef(const int body_index, vec3 *r) const
+{
+	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
+	*r = m_body_list[body_index].m_parent_pos_parent_body_ref;
+	return 0;
 }
 
-int MultiBodyTree::MultiBodyImpl::getBodyTParentRef(const int body_index, mat33* T)  const{
-    CHECK_IF_BODY_INDEX_IS_VALID(body_index);
-    *T=m_body_list[body_index].m_body_T_parent_ref;
-    return 0;
+int MultiBodyTree::MultiBodyImpl::getBodyTParentRef(const int body_index, mat33 *T) const
+{
+	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
+	*T = m_body_list[body_index].m_body_T_parent_ref;
+	return 0;
 }
 
-int MultiBodyTree::MultiBodyImpl::getBodyAxisOfMotion(const int body_index, vec3* axis)  const{
-    CHECK_IF_BODY_INDEX_IS_VALID(body_index);
-    if(m_body_list[body_index].m_joint_type == REVOLUTE) {
-        *axis = m_body_list[body_index].m_Jac_JR;
-        return 0;
-    }
-    if(m_body_list[body_index].m_joint_type == PRISMATIC) {
-        *axis = m_body_list[body_index].m_Jac_JT;
-        return 0;
-    }
-    setZero(*axis);
-    return 0;
+int MultiBodyTree::MultiBodyImpl::getBodyAxisOfMotion(const int body_index, vec3 *axis) const
+{
+	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
+	if (m_body_list[body_index].m_joint_type == REVOLUTE)
+	{
+		*axis = m_body_list[body_index].m_Jac_JR;
+		return 0;
+	}
+	if (m_body_list[body_index].m_joint_type == PRISMATIC)
+	{
+		*axis = m_body_list[body_index].m_Jac_JT;
+		return 0;
+	}
+	setZero(*axis);
+	return 0;
 }
 
-int MultiBodyTree::MultiBodyImpl::getDoFOffset(const int body_index, int *q_index) const {
+int MultiBodyTree::MultiBodyImpl::getDoFOffset(const int body_index, int *q_index) const
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	*q_index = m_body_list[body_index].m_q_index;
 	return 0;
 }
 
-int MultiBodyTree::MultiBodyImpl::setBodyMass(const int body_index, const idScalar mass) {
+int MultiBodyTree::MultiBodyImpl::setBodyMass(const int body_index, const idScalar mass)
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	m_body_list[body_index].m_mass = mass;
 	return 0;
 }
 
 int MultiBodyTree::MultiBodyImpl::setBodyFirstMassMoment(const int body_index,
-														 const vec3& first_mass_moment) {
+														 const vec3 &first_mass_moment)
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	m_body_list[body_index].m_body_mass_com = first_mass_moment;
 	return 0;
 }
 int MultiBodyTree::MultiBodyImpl::setBodySecondMassMoment(const int body_index,
-														  const mat33& second_mass_moment) {
+														  const mat33 &second_mass_moment)
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	m_body_list[body_index].m_body_I_body = second_mass_moment;
 	return 0;
 }
-int MultiBodyTree::MultiBodyImpl::getBodyMass(const int body_index, idScalar *mass) const {
+int MultiBodyTree::MultiBodyImpl::getBodyMass(const int body_index, idScalar *mass) const
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	*mass = m_body_list[body_index].m_mass;
 	return 0;
 }
 int MultiBodyTree::MultiBodyImpl::getBodyFirstMassMoment(const int body_index,
-														 vec3 *first_mass_moment) const {
+														 vec3 *first_mass_moment) const
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	*first_mass_moment = m_body_list[body_index].m_body_mass_com;
 	return 0;
 }
 int MultiBodyTree::MultiBodyImpl::getBodySecondMassMoment(const int body_index,
-														  mat33 *second_mass_moment) const {
+														  mat33 *second_mass_moment) const
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	*second_mass_moment = m_body_list[body_index].m_body_I_body;
 	return 0;
 }
 
-void MultiBodyTree::MultiBodyImpl::clearAllUserForcesAndMoments() {
-	for (int index = 0; index < m_num_bodies; index++) {
+void MultiBodyTree::MultiBodyImpl::clearAllUserForcesAndMoments()
+{
+	for (int index = 0; index < m_num_bodies; index++)
+	{
 		RigidBody &body = m_body_list[index];
 		setZero(body.m_body_force_user);
 		setZero(body.m_body_moment_user);
 	}
 }
 
-int MultiBodyTree::MultiBodyImpl::addUserForce(const int body_index, const vec3 &body_force) {
+int MultiBodyTree::MultiBodyImpl::addUserForce(const int body_index, const vec3 &body_force)
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	m_body_list[body_index].m_body_force_user += body_force;
 	return 0;
 }
 
-int MultiBodyTree::MultiBodyImpl::addUserMoment(const int body_index, const vec3 &body_moment) {
+int MultiBodyTree::MultiBodyImpl::addUserMoment(const int body_index, const vec3 &body_moment)
+{
 	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
 	m_body_list[body_index].m_body_moment_user += body_moment;
 	return 0;
 }
 
 #if (defined BT_ID_HAVE_MAT3X) && (defined BT_ID_WITH_JACOBIANS)
-int  MultiBodyTree::MultiBodyImpl::getBodyDotJacobianTransU(const int body_index, vec3* world_dot_jac_trans_u) const {
-    CHECK_IF_BODY_INDEX_IS_VALID(body_index);
-    const RigidBody &body = m_body_list[body_index];
-    *world_dot_jac_trans_u = body.m_body_T_world.transpose() * body.m_body_dot_Jac_T_u;
-    return 0;
+int MultiBodyTree::MultiBodyImpl::getBodyDotJacobianTransU(const int body_index, vec3 *world_dot_jac_trans_u) const
+{
+	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
+	const RigidBody &body = m_body_list[body_index];
+	*world_dot_jac_trans_u = body.m_body_T_world.transpose() * body.m_body_dot_Jac_T_u;
+	return 0;
 }
 
-int  MultiBodyTree::MultiBodyImpl::getBodyDotJacobianRotU(const int body_index, vec3* world_dot_jac_rot_u) const{
-     CHECK_IF_BODY_INDEX_IS_VALID(body_index);
-    const RigidBody &body = m_body_list[body_index];
-    *world_dot_jac_rot_u = body.m_body_T_world.transpose() * body.m_body_dot_Jac_R_u;
-    return 0;
+int MultiBodyTree::MultiBodyImpl::getBodyDotJacobianRotU(const int body_index, vec3 *world_dot_jac_rot_u) const
+{
+	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
+	const RigidBody &body = m_body_list[body_index];
+	*world_dot_jac_rot_u = body.m_body_T_world.transpose() * body.m_body_dot_Jac_R_u;
+	return 0;
 }
 
-int  MultiBodyTree::MultiBodyImpl::getBodyJacobianTrans(const int body_index, mat3x* world_jac_trans) const{
-    CHECK_IF_BODY_INDEX_IS_VALID(body_index);
-    const RigidBody &body = m_body_list[body_index];
-    mul(body.m_body_T_world.transpose(), body.m_body_Jac_T, world_jac_trans);
-    return 0;
+int MultiBodyTree::MultiBodyImpl::getBodyJacobianTrans(const int body_index, mat3x *world_jac_trans) const
+{
+	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
+	const RigidBody &body = m_body_list[body_index];
+	mul(body.m_body_T_world.transpose(), body.m_body_Jac_T, world_jac_trans);
+	return 0;
 }
 
-int  MultiBodyTree::MultiBodyImpl::getBodyJacobianRot(const int body_index, mat3x* world_jac_rot) const{
-    CHECK_IF_BODY_INDEX_IS_VALID(body_index);
-    const RigidBody &body = m_body_list[body_index];
-    mul(body.m_body_T_world.transpose(), body.m_body_Jac_R,world_jac_rot);
-    return 0;
+int MultiBodyTree::MultiBodyImpl::getBodyJacobianRot(const int body_index, mat3x *world_jac_rot) const
+{
+	CHECK_IF_BODY_INDEX_IS_VALID(body_index);
+	const RigidBody &body = m_body_list[body_index];
+	mul(body.m_body_T_world.transpose(), body.m_body_Jac_R, world_jac_rot);
+	return 0;
 }
 
 #endif
-}
+}  // namespace btInverseDynamics
diff --git a/thirdparty/bullet/BulletInverseDynamics/details/MultiBodyTreeImpl.hpp b/thirdparty/bullet/BulletInverseDynamics/details/MultiBodyTreeImpl.hpp
index 3efe9d04926..eabdbe161b4 100644
--- a/thirdparty/bullet/BulletInverseDynamics/details/MultiBodyTreeImpl.hpp
+++ b/thirdparty/bullet/BulletInverseDynamics/details/MultiBodyTreeImpl.hpp
@@ -8,12 +8,13 @@
 #include "../IDConfig.hpp"
 #include "../MultiBodyTree.hpp"
 
-namespace btInverseDynamics {
-
+namespace btInverseDynamics
+{
 /// Structure for for rigid body mass properties, connectivity and kinematic state
 /// all vectors and matrices are in body-fixed frame, if not indicated otherwise.
 /// The body-fixed frame is located in the joint connecting the body to its parent.
-struct RigidBody {
+struct RigidBody
+{
 	ID_DECLARE_ALIGNED_ALLOCATOR();
 	// 1 Inertial properties
 	/// Mass
@@ -112,31 +113,33 @@ struct RigidBody {
 	mat33 m_body_subtree_I_body;
 
 #if (defined BT_ID_HAVE_MAT3X) && (defined BT_ID_WITH_JACOBIANS)
-    /// translational jacobian in body-fixed frame d(m_body_vel)/du
-    mat3x m_body_Jac_T;
-    /// rotationsl jacobian in body-fixed frame d(m_body_ang_vel)/du
-    mat3x m_body_Jac_R;
-    /// components of linear acceleration depending on u
-    /// (same as is d(m_Jac_T)/dt*u)
-    vec3 m_body_dot_Jac_T_u;
-    /// components of angular acceleration depending on u
-    /// (same as is d(m_Jac_T)/dt*u)
-    vec3 m_body_dot_Jac_R_u;
+	/// translational jacobian in body-fixed frame d(m_body_vel)/du
+	mat3x m_body_Jac_T;
+	/// rotationsl jacobian in body-fixed frame d(m_body_ang_vel)/du
+	mat3x m_body_Jac_R;
+	/// components of linear acceleration depending on u
+	/// (same as is d(m_Jac_T)/dt*u)
+	vec3 m_body_dot_Jac_T_u;
+	/// components of angular acceleration depending on u
+	/// (same as is d(m_Jac_T)/dt*u)
+	vec3 m_body_dot_Jac_R_u;
 #endif
 };
 
 /// The MBS implements a tree structured multibody system
-class MultiBodyTree::MultiBodyImpl {
+class MultiBodyTree::MultiBodyImpl
+{
 	friend class MultiBodyTree;
 
 public:
 	ID_DECLARE_ALIGNED_ALLOCATOR();
 
-        enum KinUpdateType {
-            POSITION_ONLY,
-            POSITION_VELOCITY,
-            POSITION_VELOCITY_ACCELERATION
-        };
+	enum KinUpdateType
+	{
+		POSITION_ONLY,
+		POSITION_VELOCITY,
+		POSITION_VELOCITY_ACCELERATION
+	};
 
 	/// constructor
 	/// @param num_bodies the number of bodies in the system
@@ -150,24 +153,24 @@ public:
 	int calculateMassMatrix(const vecx& q, const bool update_kinematics,
 							const bool initialize_matrix, const bool set_lower_triangular_matrix,
 							matxx* mass_matrix);
-    	/// calculate kinematics (vector quantities)
-        /// Depending on type, update positions only, positions & velocities, or positions, velocities
-        /// and accelerations.
-        int calculateKinematics(const vecx& q, const vecx& u, const vecx& dot_u, const KinUpdateType type);
+	/// calculate kinematics (vector quantities)
+	/// Depending on type, update positions only, positions & velocities, or positions, velocities
+	/// and accelerations.
+	int calculateKinematics(const vecx& q, const vecx& u, const vecx& dot_u, const KinUpdateType type);
 #if (defined BT_ID_HAVE_MAT3X) && (defined BT_ID_WITH_JACOBIANS)
-    	/// calculate jacobians and (if type == POSITION_VELOCITY), also velocity-dependent accelration terms.
-    int calculateJacobians(const vecx& q, const vecx& u, const KinUpdateType type);
-    /// \copydoc MultiBodyTree::getBodyDotJacobianTransU
-    int getBodyDotJacobianTransU(const int body_index, vec3* world_dot_jac_trans_u) const ;
-    /// \copydoc MultiBodyTree::getBodyDotJacobianRotU
-    int getBodyDotJacobianRotU(const int body_index, vec3* world_dot_jac_rot_u) const;
-    /// \copydoc MultiBodyTree::getBodyJacobianTrans
-    int getBodyJacobianTrans(const int body_index, mat3x* world_jac_trans) const ;
-    /// \copydoc MultiBodyTree::getBodyJacobianRot
-    int getBodyJacobianRot(const int body_index, mat3x* world_jac_rot) const;
-    /// Add relative Jacobian component from motion relative to parent body
-    /// @param body the body to add the Jacobian component for
-    void addRelativeJacobianComponent(RigidBody&body);
+	/// calculate jacobians and (if type == POSITION_VELOCITY), also velocity-dependent accelration terms.
+	int calculateJacobians(const vecx& q, const vecx& u, const KinUpdateType type);
+	/// \copydoc MultiBodyTree::getBodyDotJacobianTransU
+	int getBodyDotJacobianTransU(const int body_index, vec3* world_dot_jac_trans_u) const;
+	/// \copydoc MultiBodyTree::getBodyDotJacobianRotU
+	int getBodyDotJacobianRotU(const int body_index, vec3* world_dot_jac_rot_u) const;
+	/// \copydoc MultiBodyTree::getBodyJacobianTrans
+	int getBodyJacobianTrans(const int body_index, mat3x* world_jac_trans) const;
+	/// \copydoc MultiBodyTree::getBodyJacobianRot
+	int getBodyJacobianRot(const int body_index, mat3x* world_jac_rot) const;
+	/// Add relative Jacobian component from motion relative to parent body
+	/// @param body the body to add the Jacobian component for
+	void addRelativeJacobianComponent(RigidBody& body);
 #endif
 	/// generate additional index sets from the parent_index array
 	/// @return -1 on error, 0 on success
@@ -190,12 +193,12 @@ public:
 	int getJointType(const int body_index, JointType* joint_type) const;
 	/// \copydoc MultiBodyTree::getJointTypeStr
 	int getJointTypeStr(const int body_index, const char** joint_type) const;
-        /// \copydoc MultiBodyTree::getParentRParentBodyRef
-        int getParentRParentBodyRef(const int body_index, vec3* r) const;
-        /// \copydoc MultiBodyTree::getBodyTParentRef
-        int getBodyTParentRef(const int body_index, mat33* T) const;
-        /// \copydoc MultiBodyTree::getBodyAxisOfMotion
-        int getBodyAxisOfMotion(const int body_index, vec3* axis) const;
+	/// \copydoc MultiBodyTree::getParentRParentBodyRef
+	int getParentRParentBodyRef(const int body_index, vec3* r) const;
+	/// \copydoc MultiBodyTree::getBodyTParentRef
+	int getBodyTParentRef(const int body_index, mat33* T) const;
+	/// \copydoc MultiBodyTree::getBodyAxisOfMotion
+	int getBodyAxisOfMotion(const int body_index, vec3* axis) const;
 	/// \copydoc MultiBodyTree:getDoFOffset
 	int getDoFOffset(const int body_index, int* q_index) const;
 	/// \copydoc MultiBodyTree::getBodyOrigin
@@ -271,13 +274,15 @@ private:
 	idArray<int>::type m_body_prismatic_list;
 	// Indices of floating joints
 	idArray<int>::type m_body_floating_list;
+	// Indices of spherical joints
+	idArray<int>::type m_body_spherical_list;
 	// a user-provided integer
 	idArray<int>::type m_user_int;
 	// a user-provided pointer
 	idArray<void*>::type m_user_ptr;
 #if (defined BT_ID_HAVE_MAT3X) && (defined BT_ID_WITH_JACOBIANS)
-        mat3x m_m3x;
+	mat3x m_m3x;
 #endif
 };
-}
+}  // namespace btInverseDynamics
 #endif
diff --git a/thirdparty/bullet/BulletInverseDynamics/details/MultiBodyTreeInitCache.cpp b/thirdparty/bullet/BulletInverseDynamics/details/MultiBodyTreeInitCache.cpp
index e9511b70763..a718db051ef 100644
--- a/thirdparty/bullet/BulletInverseDynamics/details/MultiBodyTreeInitCache.cpp
+++ b/thirdparty/bullet/BulletInverseDynamics/details/MultiBodyTreeInitCache.cpp
@@ -1,12 +1,13 @@
 #include "MultiBodyTreeInitCache.hpp"
 
-namespace btInverseDynamics {
-
-MultiBodyTree::InitCache::InitCache() {
+namespace btInverseDynamics
+{
+MultiBodyTree::InitCache::InitCache()
+{
 	m_inertias.resize(0);
 	m_joints.resize(0);
 	m_num_dofs = 0;
-	m_root_index=-1;
+	m_root_index = -1;
 }
 
 int MultiBodyTree::InitCache::addBody(const int body_index, const int parent_index,
@@ -15,8 +16,10 @@ int MultiBodyTree::InitCache::addBody(const int body_index, const int parent_ind
 									  const mat33& body_T_parent_ref,
 									  const vec3& body_axis_of_motion, const idScalar mass,
 									  const vec3& body_r_body_com, const mat33& body_I_body,
-									  const int user_int, void* user_ptr) {
-	switch (joint_type) {
+									  const int user_int, void* user_ptr)
+{
+	switch (joint_type)
+	{
 		case REVOLUTE:
 		case PRISMATIC:
 			m_num_dofs += 1;
@@ -25,6 +28,9 @@ int MultiBodyTree::InitCache::addBody(const int body_index, const int parent_ind
 			// does not add a degree of freedom
 			// m_num_dofs+=0;
 			break;
+		case SPHERICAL:
+			m_num_dofs += 3;
+			break;
 		case FLOATING:
 			m_num_dofs += 6;
 			break;
@@ -33,13 +39,15 @@ int MultiBodyTree::InitCache::addBody(const int body_index, const int parent_ind
 			return -1;
 	}
 
-	if(-1 == parent_index) {
-		if(m_root_index>=0) {
+	if (-1 == parent_index)
+	{
+		if (m_root_index >= 0)
+		{
 			bt_id_error_message("trying to add body %d as root, but already added %d as root body\n",
-						  body_index, m_root_index);
+								body_index, m_root_index);
 			return -1;
 		}
-		m_root_index=body_index;
+		m_root_index = body_index;
 	}
 
 	JointData joint;
@@ -61,8 +69,10 @@ int MultiBodyTree::InitCache::addBody(const int body_index, const int parent_ind
 	m_user_ptr.push_back(user_ptr);
 	return 0;
 }
-int MultiBodyTree::InitCache::getInertiaData(const int index, InertiaData* inertia) const {
-	if (index < 0 || index > static_cast<int>(m_inertias.size())) {
+int MultiBodyTree::InitCache::getInertiaData(const int index, InertiaData* inertia) const
+{
+	if (index < 0 || index > static_cast<int>(m_inertias.size()))
+	{
 		bt_id_error_message("index out of range\n");
 		return -1;
 	}
@@ -71,8 +81,10 @@ int MultiBodyTree::InitCache::getInertiaData(const int index, InertiaData* inert
 	return 0;
 }
 
-int MultiBodyTree::InitCache::getUserInt(const int index, int* user_int) const {
-	if (index < 0 || index > static_cast<int>(m_user_int.size())) {
+int MultiBodyTree::InitCache::getUserInt(const int index, int* user_int) const
+{
+	if (index < 0 || index > static_cast<int>(m_user_int.size()))
+	{
 		bt_id_error_message("index out of range\n");
 		return -1;
 	}
@@ -80,8 +92,10 @@ int MultiBodyTree::InitCache::getUserInt(const int index, int* user_int) const {
 	return 0;
 }
 
-int MultiBodyTree::InitCache::getUserPtr(const int index, void** user_ptr) const {
-	if (index < 0 || index > static_cast<int>(m_user_ptr.size())) {
+int MultiBodyTree::InitCache::getUserPtr(const int index, void** user_ptr) const
+{
+	if (index < 0 || index > static_cast<int>(m_user_ptr.size()))
+	{
 		bt_id_error_message("index out of range\n");
 		return -1;
 	}
@@ -89,8 +103,10 @@ int MultiBodyTree::InitCache::getUserPtr(const int index, void** user_ptr) const
 	return 0;
 }
 
-int MultiBodyTree::InitCache::getJointData(const int index, JointData* joint) const {
-	if (index < 0 || index > static_cast<int>(m_joints.size())) {
+int MultiBodyTree::InitCache::getJointData(const int index, JointData* joint) const
+{
+	if (index < 0 || index > static_cast<int>(m_joints.size()))
+	{
 		bt_id_error_message("index out of range\n");
 		return -1;
 	}
@@ -98,16 +114,18 @@ int MultiBodyTree::InitCache::getJointData(const int index, JointData* joint) co
 	return 0;
 }
 
-int MultiBodyTree::InitCache::buildIndexSets() {
+int MultiBodyTree::InitCache::buildIndexSets()
+{
 	// NOTE: This function assumes that proper indices were provided
 	//	   User2InternalIndex from utils can be used to facilitate this.
 
 	m_parent_index.resize(numBodies());
-	for (idArrayIdx j = 0; j < m_joints.size(); j++) {
+	for (idArrayIdx j = 0; j < m_joints.size(); j++)
+	{
 		const JointData& joint = m_joints[j];
 		m_parent_index[joint.m_child] = joint.m_parent;
 	}
 
 	return 0;
 }
-}
+}  // namespace btInverseDynamics
diff --git a/thirdparty/bullet/BulletInverseDynamics/details/MultiBodyTreeInitCache.hpp b/thirdparty/bullet/BulletInverseDynamics/details/MultiBodyTreeInitCache.hpp
index 0d2aa4a0718..dbdb3ff6040 100644
--- a/thirdparty/bullet/BulletInverseDynamics/details/MultiBodyTreeInitCache.hpp
+++ b/thirdparty/bullet/BulletInverseDynamics/details/MultiBodyTreeInitCache.hpp
@@ -5,9 +5,11 @@
 #include "../IDMath.hpp"
 #include "../MultiBodyTree.hpp"
 
-namespace btInverseDynamics {
+namespace btInverseDynamics
+{
 /// Mass properties of a rigid body
-struct InertiaData {
+struct InertiaData
+{
 	ID_DECLARE_ALIGNED_ALLOCATOR();
 
 	/// mass
@@ -21,7 +23,8 @@ struct InertiaData {
 };
 
 /// Joint properties
-struct JointData {
+struct JointData
+{
 	ID_DECLARE_ALIGNED_ALLOCATOR();
 
 	/// type of joint
@@ -48,7 +51,8 @@ struct JointData {
 
 /// Data structure to store data passed by the user.
 /// This is used in MultiBodyTree::finalize to build internal data structures.
-class MultiBodyTree::InitCache {
+class MultiBodyTree::InitCache
+{
 public:
 	ID_DECLARE_ALIGNED_ALLOCATOR();
 	/// constructor
@@ -105,5 +109,5 @@ private:
 	// index of root body (or -1 if not set)
 	int m_root_index;
 };
-}
+}  // namespace btInverseDynamics
 #endif  // MULTIBODYTREEINITCACHE_HPP_
diff --git a/thirdparty/bullet/BulletSoftBody/btDefaultSoftBodySolver.cpp b/thirdparty/bullet/BulletSoftBody/btDefaultSoftBodySolver.cpp
index 9c204030743..8b7ff9abcd8 100644
--- a/thirdparty/bullet/BulletSoftBody/btDefaultSoftBodySolver.cpp
+++ b/thirdparty/bullet/BulletSoftBody/btDefaultSoftBodySolver.cpp
@@ -21,11 +21,10 @@ subject to the following restrictions:
 #include "BulletCollision/CollisionShapes/btCapsuleShape.h"
 #include "BulletSoftBody/btSoftBody.h"
 
-
 btDefaultSoftBodySolver::btDefaultSoftBodySolver()
 {
 	// Initial we will clearly need to update solver constants
-	// For now this is global for the cloths linked with this solver - we should probably make this body specific 
+	// For now this is global for the cloths linked with this solver - we should probably make this body specific
 	// for performance in future once we understand more clearly when constants need to be updated
 	m_updateSolverConstants = true;
 }
@@ -37,67 +36,65 @@ btDefaultSoftBodySolver::~btDefaultSoftBodySolver()
 // In this case the data is already in the soft bodies so there is no need for us to do anything
 void btDefaultSoftBodySolver::copyBackToSoftBodies(bool bMove)
 {
-
 }
 
-void btDefaultSoftBodySolver::optimize( btAlignedObjectArray< btSoftBody * > &softBodies , bool forceUpdate)
+void btDefaultSoftBodySolver::optimize(btAlignedObjectArray<btSoftBody *> &softBodies, bool forceUpdate)
 {
-	m_softBodySet.copyFromArray( softBodies );
+	m_softBodySet.copyFromArray(softBodies);
 }
 
-void btDefaultSoftBodySolver::updateSoftBodies( )
+void btDefaultSoftBodySolver::updateSoftBodies()
 {
-	for ( int i=0; i < m_softBodySet.size(); i++)
+	for (int i = 0; i < m_softBodySet.size(); i++)
 	{
-		btSoftBody*	psb=(btSoftBody*)m_softBodySet[i];
+		btSoftBody *psb = (btSoftBody *)m_softBodySet[i];
 		if (psb->isActive())
 		{
-			psb->integrateMotion();	
+			psb->integrateMotion();
 		}
 	}
-} // updateSoftBodies
+}  // updateSoftBodies
 
 bool btDefaultSoftBodySolver::checkInitialized()
 {
 	return true;
 }
 
-void btDefaultSoftBodySolver::solveConstraints( float solverdt )
+void btDefaultSoftBodySolver::solveConstraints(float solverdt)
 {
 	// Solve constraints for non-solver softbodies
-	for(int i=0; i < m_softBodySet.size(); ++i)
+	for (int i = 0; i < m_softBodySet.size(); ++i)
 	{
-		btSoftBody*	psb = static_cast<btSoftBody*>(m_softBodySet[i]);
+		btSoftBody *psb = static_cast<btSoftBody *>(m_softBodySet[i]);
 		if (psb->isActive())
 		{
 			psb->solveConstraints();
 		}
-	}	
-} // btDefaultSoftBodySolver::solveConstraints
+	}
+}  // btDefaultSoftBodySolver::solveConstraints
 
-
-void btDefaultSoftBodySolver::copySoftBodyToVertexBuffer( const btSoftBody *const softBody, btVertexBufferDescriptor *vertexBuffer )
+void btDefaultSoftBodySolver::copySoftBodyToVertexBuffer(const btSoftBody *const softBody, btVertexBufferDescriptor *vertexBuffer)
 {
 	// Currently only support CPU output buffers
 	// TODO: check for DX11 buffers. Take all offsets into the same DX11 buffer
 	// and use them together on a single kernel call if possible by setting up a
 	// per-cloth target buffer array for the copy kernel.
 
-	if( vertexBuffer->getBufferType() == btVertexBufferDescriptor::CPU_BUFFER )
+	if (vertexBuffer->getBufferType() == btVertexBufferDescriptor::CPU_BUFFER)
 	{
-		const btAlignedObjectArray<btSoftBody::Node> &clothVertices( softBody->m_nodes );
+		const btAlignedObjectArray<btSoftBody::Node> &clothVertices(softBody->m_nodes);
 		int numVertices = clothVertices.size();
 
-		const btCPUVertexBufferDescriptor *cpuVertexBuffer = static_cast< btCPUVertexBufferDescriptor* >(vertexBuffer);						
-		float *basePointer = cpuVertexBuffer->getBasePointer();						
+		const btCPUVertexBufferDescriptor *cpuVertexBuffer = static_cast<btCPUVertexBufferDescriptor *>(vertexBuffer);
+		float *basePointer = cpuVertexBuffer->getBasePointer();
 
-		if( vertexBuffer->hasVertexPositions() )
+		if (vertexBuffer->hasVertexPositions())
 		{
 			const int vertexOffset = cpuVertexBuffer->getVertexOffset();
 			const int vertexStride = cpuVertexBuffer->getVertexStride();
 			float *vertexPointer = basePointer + vertexOffset;
 
-			for( int vertexIndex = 0; vertexIndex < numVertices; ++vertexIndex )
+			for (int vertexIndex = 0; vertexIndex < numVertices; ++vertexIndex)
 			{
 				btVector3 position = clothVertices[vertexIndex].m_x;
 				*(vertexPointer + 0) = (float)position.getX();
@@ -106,13 +103,13 @@ void btDefaultSoftBodySolver::copySoftBodyToVertexBuffer( const btSoftBody *cons
 				vertexPointer += vertexStride;
 			}
 		}
-		if( vertexBuffer->hasNormals() )
+		if (vertexBuffer->hasNormals())
 		{
 			const int normalOffset = cpuVertexBuffer->getNormalOffset();
 			const int normalStride = cpuVertexBuffer->getNormalStride();
 			float *normalPointer = basePointer + normalOffset;
 
-			for( int vertexIndex = 0; vertexIndex < numVertices; ++vertexIndex )
+			for (int vertexIndex = 0; vertexIndex < numVertices; ++vertexIndex)
 			{
 				btVector3 normal = clothVertices[vertexIndex].m_n;
 				*(normalPointer + 0) = (float)normal.getX();
@@ -122,30 +119,28 @@ void btDefaultSoftBodySolver::copySoftBodyToVertexBuffer( const btSoftBody *cons
 			}
 		}
 	}
-} // btDefaultSoftBodySolver::copySoftBodyToVertexBuffer
+}  // btDefaultSoftBodySolver::copySoftBodyToVertexBuffer
 
-void btDefaultSoftBodySolver::processCollision( btSoftBody* softBody, btSoftBody* otherSoftBody)
+void btDefaultSoftBodySolver::processCollision(btSoftBody *softBody, btSoftBody *otherSoftBody)
 {
-	softBody->defaultCollisionHandler( otherSoftBody);
+	softBody->defaultCollisionHandler(otherSoftBody);
 }
 
 // For the default solver just leave the soft body to do its collision processing
-void btDefaultSoftBodySolver::processCollision( btSoftBody *softBody, const btCollisionObjectWrapper* collisionObjectWrap )
+void btDefaultSoftBodySolver::processCollision(btSoftBody *softBody, const btCollisionObjectWrapper *collisionObjectWrap)
 {
-	softBody->defaultCollisionHandler( collisionObjectWrap );
-} // btDefaultSoftBodySolver::processCollision
+	softBody->defaultCollisionHandler(collisionObjectWrap);
+}  // btDefaultSoftBodySolver::processCollision
 
-
-void btDefaultSoftBodySolver::predictMotion( float timeStep )
+void btDefaultSoftBodySolver::predictMotion(float timeStep)
 {
-	for ( int i=0; i < m_softBodySet.size(); ++i)
+	for (int i = 0; i < m_softBodySet.size(); ++i)
 	{
-		btSoftBody*	psb = m_softBodySet[i];
+		btSoftBody *psb = m_softBodySet[i];
 
 		if (psb->isActive())
 		{
-			psb->predictMotion(timeStep);		
+			psb->predictMotion(timeStep);
 		}
 	}
 }
-
diff --git a/thirdparty/bullet/BulletSoftBody/btDefaultSoftBodySolver.h b/thirdparty/bullet/BulletSoftBody/btDefaultSoftBodySolver.h
index 1c17ffcbb2b..50bd7351656 100644
--- a/thirdparty/bullet/BulletSoftBody/btDefaultSoftBodySolver.h
+++ b/thirdparty/bullet/BulletSoftBody/btDefaultSoftBodySolver.h
@@ -16,25 +16,23 @@ subject to the following restrictions:
 #ifndef BT_SOFT_BODY_DEFAULT_SOLVER_H
 #define BT_SOFT_BODY_DEFAULT_SOLVER_H
 
-
 #include "BulletSoftBody/btSoftBodySolvers.h"
 #include "btSoftBodySolverVertexBuffer.h"
 struct btCollisionObjectWrapper;
 
 class btDefaultSoftBodySolver : public btSoftBodySolver
 {
-protected:		
+protected:
 	/** Variable to define whether we need to update solver constants on the next iteration */
 	bool m_updateSolverConstants;
 
-	btAlignedObjectArray< btSoftBody * > m_softBodySet;
-
+	btAlignedObjectArray<btSoftBody *> m_softBodySet;
 
 public:
 	btDefaultSoftBodySolver();
-	
+
 	virtual ~btDefaultSoftBodySolver();
-	
+
 	virtual SolverTypes getSolverType() const
 	{
 		return DEFAULT_SOLVER;
@@ -42,22 +40,21 @@ public:
 
 	virtual bool checkInitialized();
 
-	virtual void updateSoftBodies( );
+	virtual void updateSoftBodies();
 
-	virtual void optimize( btAlignedObjectArray< btSoftBody * > &softBodies,bool forceUpdate=false );
+	virtual void optimize(btAlignedObjectArray<btSoftBody *> &softBodies, bool forceUpdate = false);
 
 	virtual void copyBackToSoftBodies(bool bMove = true);
 
-	virtual void solveConstraints( float solverdt );
+	virtual void solveConstraints(float solverdt);
 
-	virtual void predictMotion( float solverdt );
+	virtual void predictMotion(float solverdt);
 
-	virtual void copySoftBodyToVertexBuffer( const btSoftBody *const softBody, btVertexBufferDescriptor *vertexBuffer );
+	virtual void copySoftBodyToVertexBuffer(const btSoftBody *const softBody, btVertexBufferDescriptor *vertexBuffer);
 
-	virtual void processCollision( btSoftBody *, const btCollisionObjectWrapper* );
-
-	virtual void processCollision( btSoftBody*, btSoftBody* );
+	virtual void processCollision(btSoftBody *, const btCollisionObjectWrapper *);
 
+	virtual void processCollision(btSoftBody *, btSoftBody *);
 };
 
-#endif // #ifndef BT_ACCELERATED_SOFT_BODY_CPU_SOLVER_H
+#endif  // #ifndef BT_ACCELERATED_SOFT_BODY_CPU_SOLVER_H
diff --git a/thirdparty/bullet/BulletSoftBody/btSoftBody.cpp b/thirdparty/bullet/BulletSoftBody/btSoftBody.cpp
index 48efb0d8d48..58796a88d00 100644
--- a/thirdparty/bullet/BulletSoftBody/btSoftBody.cpp
+++ b/thirdparty/bullet/BulletSoftBody/btSoftBody.cpp
@@ -21,97 +21,94 @@ subject to the following restrictions:
 #include "BulletDynamics/Featherstone/btMultiBodyLinkCollider.h"
 #include "BulletDynamics/Featherstone/btMultiBodyConstraint.h"
 
-
 //
-btSoftBody::btSoftBody(btSoftBodyWorldInfo*	worldInfo,int node_count,  const btVector3* x,  const btScalar* m)
-:m_softBodySolver(0),m_worldInfo(worldInfo)
-{	
-	/* Init		*/ 
+btSoftBody::btSoftBody(btSoftBodyWorldInfo* worldInfo, int node_count, const btVector3* x, const btScalar* m)
+	: m_softBodySolver(0), m_worldInfo(worldInfo)
+{
+	/* Init		*/
 	initDefaults();
 
-	/* Default material	*/ 
-	Material*	pm=appendMaterial();
-	pm->m_kLST	=	1;
-	pm->m_kAST	=	1;
-	pm->m_kVST	=	1;
-	pm->m_flags	=	fMaterial::Default;
+	/* Default material	*/
+	Material* pm = appendMaterial();
+	pm->m_kLST = 1;
+	pm->m_kAST = 1;
+	pm->m_kVST = 1;
+	pm->m_flags = fMaterial::Default;
 
-	/* Nodes			*/ 
-	const btScalar		margin=getCollisionShape()->getMargin();
+	/* Nodes			*/
+	const btScalar margin = getCollisionShape()->getMargin();
 	m_nodes.resize(node_count);
-	for(int i=0,ni=node_count;i<ni;++i)
-	{	
-		Node&	n=m_nodes[i];
+	for (int i = 0, ni = node_count; i < ni; ++i)
+	{
+		Node& n = m_nodes[i];
 		ZeroInitialize(n);
-		n.m_x		=	x?*x++:btVector3(0,0,0);
-		n.m_q		=	n.m_x;
-		n.m_im		=	m?*m++:1;
-		n.m_im		=	n.m_im>0?1/n.m_im:0;
-		n.m_leaf	=	m_ndbvt.insert(btDbvtVolume::FromCR(n.m_x,margin),&n);
-		n.m_material=	pm;
+		n.m_x = x ? *x++ : btVector3(0, 0, 0);
+		n.m_q = n.m_x;
+		n.m_im = m ? *m++ : 1;
+		n.m_im = n.m_im > 0 ? 1 / n.m_im : 0;
+		n.m_leaf = m_ndbvt.insert(btDbvtVolume::FromCR(n.m_x, margin), &n);
+		n.m_material = pm;
 	}
-	updateBounds();	
-
+	updateBounds();
 }
 
-btSoftBody::btSoftBody(btSoftBodyWorldInfo*	worldInfo)
-:m_worldInfo(worldInfo)
+btSoftBody::btSoftBody(btSoftBodyWorldInfo* worldInfo)
+	: m_worldInfo(worldInfo)
 {
 	initDefaults();
 }
 
-
-void	btSoftBody::initDefaults()
+void btSoftBody::initDefaults()
 {
-	m_internalType		=	CO_SOFT_BODY;
-	m_cfg.aeromodel		=	eAeroModel::V_Point;
-	m_cfg.kVCF			=	1;
-	m_cfg.kDG			=	0;
-	m_cfg.kLF			=	0;
-	m_cfg.kDP			=	0;
-	m_cfg.kPR			=	0;
-	m_cfg.kVC			=	0;
-	m_cfg.kDF			=	(btScalar)0.2;
-	m_cfg.kMT			=	0;
-	m_cfg.kCHR			=	(btScalar)1.0;
-	m_cfg.kKHR			=	(btScalar)0.1;
-	m_cfg.kSHR			=	(btScalar)1.0;
-	m_cfg.kAHR			=	(btScalar)0.7;
-	m_cfg.kSRHR_CL		=	(btScalar)0.1;
-	m_cfg.kSKHR_CL		=	(btScalar)1;
-	m_cfg.kSSHR_CL		=	(btScalar)0.5;
-	m_cfg.kSR_SPLT_CL	=	(btScalar)0.5;
-	m_cfg.kSK_SPLT_CL	=	(btScalar)0.5;
-	m_cfg.kSS_SPLT_CL	=	(btScalar)0.5;
-	m_cfg.maxvolume		=	(btScalar)1;
-	m_cfg.timescale		=	1;
-	m_cfg.viterations	=	0;
-	m_cfg.piterations	=	1;	
-	m_cfg.diterations	=	0;
-	m_cfg.citerations	=	4;
-	m_cfg.collisions	=	fCollision::Default;
-	m_pose.m_bvolume	=	false;
-	m_pose.m_bframe		=	false;
-	m_pose.m_volume		=	0;
-	m_pose.m_com		=	btVector3(0,0,0);
+	m_internalType = CO_SOFT_BODY;
+	m_cfg.aeromodel = eAeroModel::V_Point;
+	m_cfg.kVCF = 1;
+	m_cfg.kDG = 0;
+	m_cfg.kLF = 0;
+	m_cfg.kDP = 0;
+	m_cfg.kPR = 0;
+	m_cfg.kVC = 0;
+	m_cfg.kDF = (btScalar)0.2;
+	m_cfg.kMT = 0;
+	m_cfg.kCHR = (btScalar)1.0;
+	m_cfg.kKHR = (btScalar)0.1;
+	m_cfg.kSHR = (btScalar)1.0;
+	m_cfg.kAHR = (btScalar)0.7;
+	m_cfg.kSRHR_CL = (btScalar)0.1;
+	m_cfg.kSKHR_CL = (btScalar)1;
+	m_cfg.kSSHR_CL = (btScalar)0.5;
+	m_cfg.kSR_SPLT_CL = (btScalar)0.5;
+	m_cfg.kSK_SPLT_CL = (btScalar)0.5;
+	m_cfg.kSS_SPLT_CL = (btScalar)0.5;
+	m_cfg.maxvolume = (btScalar)1;
+	m_cfg.timescale = 1;
+	m_cfg.viterations = 0;
+	m_cfg.piterations = 1;
+	m_cfg.diterations = 0;
+	m_cfg.citerations = 4;
+	m_cfg.collisions = fCollision::Default;
+	m_pose.m_bvolume = false;
+	m_pose.m_bframe = false;
+	m_pose.m_volume = 0;
+	m_pose.m_com = btVector3(0, 0, 0);
 	m_pose.m_rot.setIdentity();
 	m_pose.m_scl.setIdentity();
-	m_tag				=	0;
-	m_timeacc			=	0;
-	m_bUpdateRtCst		=	true;
-	m_bounds[0]			=	btVector3(0,0,0);
-	m_bounds[1]			=	btVector3(0,0,0);
+	m_tag = 0;
+	m_timeacc = 0;
+	m_bUpdateRtCst = true;
+	m_bounds[0] = btVector3(0, 0, 0);
+	m_bounds[1] = btVector3(0, 0, 0);
 	m_worldTransform.setIdentity();
 	setSolver(eSolverPresets::Positions);
-	
-	/* Collision shape	*/ 
+
+	/* Collision shape	*/
 	///for now, create a collision shape internally
 	m_collisionShape = new btSoftBodyCollisionShape(this);
 	m_collisionShape->setMargin(0.25f);
-	
+
 	m_initialWorldTransform.setIdentity();
 
-	m_windVelocity = btVector3(0,0,0);
+	m_windVelocity = btVector3(0, 0, 0);
 	m_restLengthScale = btScalar(1.0);
 }
 
@@ -119,343 +116,361 @@ void	btSoftBody::initDefaults()
 btSoftBody::~btSoftBody()
 {
 	//for now, delete the internal shape
-	delete m_collisionShape;	
+	delete m_collisionShape;
 	int i;
 
 	releaseClusters();
-	for(i=0;i<m_materials.size();++i) 
+	for (i = 0; i < m_materials.size(); ++i)
 		btAlignedFree(m_materials[i]);
-	for(i=0;i<m_joints.size();++i) 
+	for (i = 0; i < m_joints.size(); ++i)
 		btAlignedFree(m_joints[i]);
 }
 
 //
-bool			btSoftBody::checkLink(int node0,int node1) const
+bool btSoftBody::checkLink(int node0, int node1) const
 {
-	return(checkLink(&m_nodes[node0],&m_nodes[node1]));
+	return (checkLink(&m_nodes[node0], &m_nodes[node1]));
 }
 
 //
-bool			btSoftBody::checkLink(const Node* node0,const Node* node1) const
+bool btSoftBody::checkLink(const Node* node0, const Node* node1) const
 {
-	const Node*	n[]={node0,node1};
-	for(int i=0,ni=m_links.size();i<ni;++i)
+	const Node* n[] = {node0, node1};
+	for (int i = 0, ni = m_links.size(); i < ni; ++i)
 	{
-		const Link&	l=m_links[i];
-		if(	(l.m_n[0]==n[0]&&l.m_n[1]==n[1])||
-			(l.m_n[0]==n[1]&&l.m_n[1]==n[0]))
+		const Link& l = m_links[i];
+		if ((l.m_n[0] == n[0] && l.m_n[1] == n[1]) ||
+			(l.m_n[0] == n[1] && l.m_n[1] == n[0]))
 		{
-			return(true);
+			return (true);
 		}
 	}
-	return(false);
+	return (false);
 }
 
 //
-bool			btSoftBody::checkFace(int node0,int node1,int node2) const
+bool btSoftBody::checkFace(int node0, int node1, int node2) const
 {
-	const Node*	n[]={	&m_nodes[node0],
-		&m_nodes[node1],
-		&m_nodes[node2]};
-	for(int i=0,ni=m_faces.size();i<ni;++i)
+	const Node* n[] = {&m_nodes[node0],
+					   &m_nodes[node1],
+					   &m_nodes[node2]};
+	for (int i = 0, ni = m_faces.size(); i < ni; ++i)
 	{
-		const Face&	f=m_faces[i];
-		int			c=0;
-		for(int j=0;j<3;++j)
+		const Face& f = m_faces[i];
+		int c = 0;
+		for (int j = 0; j < 3; ++j)
 		{
-			if(	(f.m_n[j]==n[0])||
-				(f.m_n[j]==n[1])||
-				(f.m_n[j]==n[2])) c|=1<<j; else break;
+			if ((f.m_n[j] == n[0]) ||
+				(f.m_n[j] == n[1]) ||
+				(f.m_n[j] == n[2]))
+				c |= 1 << j;
+			else
+				break;
 		}
-		if(c==7) return(true);
+		if (c == 7) return (true);
 	}
-	return(false);
+	return (false);
 }
 
 //
-btSoftBody::Material*		btSoftBody::appendMaterial()
+btSoftBody::Material* btSoftBody::appendMaterial()
 {
-	Material*	pm=new(btAlignedAlloc(sizeof(Material),16)) Material();
-	if(m_materials.size()>0)
-		*pm=*m_materials[0];
+	Material* pm = new (btAlignedAlloc(sizeof(Material), 16)) Material();
+	if (m_materials.size() > 0)
+		*pm = *m_materials[0];
 	else
 		ZeroInitialize(*pm);
 	m_materials.push_back(pm);
-	return(pm);
+	return (pm);
 }
 
 //
-void			btSoftBody::appendNote(	const char* text,
-									   const btVector3& o,
-									   const btVector4& c,
-									   Node* n0,
-									   Node* n1,
-									   Node* n2,
-									   Node* n3)
+void btSoftBody::appendNote(const char* text,
+							const btVector3& o,
+							const btVector4& c,
+							Node* n0,
+							Node* n1,
+							Node* n2,
+							Node* n3)
 {
-	Note	n;
+	Note n;
 	ZeroInitialize(n);
-	n.m_rank		=	0;
-	n.m_text		=	text;
-	n.m_offset		=	o;
-	n.m_coords[0]	=	c.x();
-	n.m_coords[1]	=	c.y();
-	n.m_coords[2]	=	c.z();
-	n.m_coords[3]	=	c.w();
-	n.m_nodes[0]	=	n0;n.m_rank+=n0?1:0;
-	n.m_nodes[1]	=	n1;n.m_rank+=n1?1:0;
-	n.m_nodes[2]	=	n2;n.m_rank+=n2?1:0;
-	n.m_nodes[3]	=	n3;n.m_rank+=n3?1:0;
+	n.m_rank = 0;
+	n.m_text = text;
+	n.m_offset = o;
+	n.m_coords[0] = c.x();
+	n.m_coords[1] = c.y();
+	n.m_coords[2] = c.z();
+	n.m_coords[3] = c.w();
+	n.m_nodes[0] = n0;
+	n.m_rank += n0 ? 1 : 0;
+	n.m_nodes[1] = n1;
+	n.m_rank += n1 ? 1 : 0;
+	n.m_nodes[2] = n2;
+	n.m_rank += n2 ? 1 : 0;
+	n.m_nodes[3] = n3;
+	n.m_rank += n3 ? 1 : 0;
 	m_notes.push_back(n);
 }
 
 //
-void			btSoftBody::appendNote(	const char* text,
-									   const btVector3& o,
-									   Node* feature)
+void btSoftBody::appendNote(const char* text,
+							const btVector3& o,
+							Node* feature)
 {
-	appendNote(text,o,btVector4(1,0,0,0),feature);
+	appendNote(text, o, btVector4(1, 0, 0, 0), feature);
 }
 
 //
-void			btSoftBody::appendNote(	const char* text,
-									   const btVector3& o,
-									   Link* feature)
+void btSoftBody::appendNote(const char* text,
+							const btVector3& o,
+							Link* feature)
 {
-	static const btScalar	w=1/(btScalar)2;
-	appendNote(text,o,btVector4(w,w,0,0),	feature->m_n[0],
-		feature->m_n[1]);
+	static const btScalar w = 1 / (btScalar)2;
+	appendNote(text, o, btVector4(w, w, 0, 0), feature->m_n[0],
+			   feature->m_n[1]);
 }
 
 //
-void			btSoftBody::appendNote(	const char* text,
-									   const btVector3& o,
-									   Face* feature)
+void btSoftBody::appendNote(const char* text,
+							const btVector3& o,
+							Face* feature)
 {
-	static const btScalar	w=1/(btScalar)3;
-	appendNote(text,o,btVector4(w,w,w,0),	feature->m_n[0],
-		feature->m_n[1],
-		feature->m_n[2]);
+	static const btScalar w = 1 / (btScalar)3;
+	appendNote(text, o, btVector4(w, w, w, 0), feature->m_n[0],
+			   feature->m_n[1],
+			   feature->m_n[2]);
 }
 
 //
-void			btSoftBody::appendNode(	const btVector3& x,btScalar m)
+void btSoftBody::appendNode(const btVector3& x, btScalar m)
 {
-	if(m_nodes.capacity()==m_nodes.size())
+	if (m_nodes.capacity() == m_nodes.size())
 	{
 		pointersToIndices();
-		m_nodes.reserve(m_nodes.size()*2+1);
+		m_nodes.reserve(m_nodes.size() * 2 + 1);
 		indicesToPointers();
 	}
-	const btScalar	margin=getCollisionShape()->getMargin();
+	const btScalar margin = getCollisionShape()->getMargin();
 	m_nodes.push_back(Node());
-	Node&			n=m_nodes[m_nodes.size()-1];
+	Node& n = m_nodes[m_nodes.size() - 1];
 	ZeroInitialize(n);
-	n.m_x			=	x;
-	n.m_q			=	n.m_x;
-	n.m_im			=	m>0?1/m:0;
-	n.m_material	=	m_materials[0];
-	n.m_leaf		=	m_ndbvt.insert(btDbvtVolume::FromCR(n.m_x,margin),&n);
+	n.m_x = x;
+	n.m_q = n.m_x;
+	n.m_im = m > 0 ? 1 / m : 0;
+	n.m_material = m_materials[0];
+	n.m_leaf = m_ndbvt.insert(btDbvtVolume::FromCR(n.m_x, margin), &n);
 }
 
 //
-void			btSoftBody::appendLink(int model,Material* mat)
+void btSoftBody::appendLink(int model, Material* mat)
 {
-	Link	l;
-	if(model>=0)
-		l=m_links[model];
+	Link l;
+	if (model >= 0)
+		l = m_links[model];
 	else
-	{ ZeroInitialize(l);l.m_material=mat?mat:m_materials[0]; }
+	{
+		ZeroInitialize(l);
+		l.m_material = mat ? mat : m_materials[0];
+	}
 	m_links.push_back(l);
 }
 
 //
-void			btSoftBody::appendLink(	int node0,
-									   int node1,
-									   Material* mat,
-									   bool bcheckexist)
+void btSoftBody::appendLink(int node0,
+							int node1,
+							Material* mat,
+							bool bcheckexist)
 {
-	appendLink(&m_nodes[node0],&m_nodes[node1],mat,bcheckexist);
+	appendLink(&m_nodes[node0], &m_nodes[node1], mat, bcheckexist);
 }
 
 //
-void			btSoftBody::appendLink(	Node* node0,
-									   Node* node1,
-									   Material* mat,
-									   bool bcheckexist)
+void btSoftBody::appendLink(Node* node0,
+							Node* node1,
+							Material* mat,
+							bool bcheckexist)
 {
-	if((!bcheckexist)||(!checkLink(node0,node1)))
+	if ((!bcheckexist) || (!checkLink(node0, node1)))
 	{
-		appendLink(-1,mat);
-		Link&	l=m_links[m_links.size()-1];
-		l.m_n[0]		=	node0;
-		l.m_n[1]		=	node1;
-		l.m_rl			=	(l.m_n[0]->m_x-l.m_n[1]->m_x).length();
-		m_bUpdateRtCst=true;
+		appendLink(-1, mat);
+		Link& l = m_links[m_links.size() - 1];
+		l.m_n[0] = node0;
+		l.m_n[1] = node1;
+		l.m_rl = (l.m_n[0]->m_x - l.m_n[1]->m_x).length();
+		m_bUpdateRtCst = true;
 	}
 }
 
 //
-void			btSoftBody::appendFace(int model,Material* mat)
+void btSoftBody::appendFace(int model, Material* mat)
 {
-	Face	f;
-	if(model>=0)
-	{ f=m_faces[model]; }
+	Face f;
+	if (model >= 0)
+	{
+		f = m_faces[model];
+	}
 	else
-	{ ZeroInitialize(f);f.m_material=mat?mat:m_materials[0]; }
+	{
+		ZeroInitialize(f);
+		f.m_material = mat ? mat : m_materials[0];
+	}
 	m_faces.push_back(f);
 }
 
 //
-void			btSoftBody::appendFace(int node0,int node1,int node2,Material* mat)
+void btSoftBody::appendFace(int node0, int node1, int node2, Material* mat)
 {
-	if (node0==node1)
+	if (node0 == node1)
 		return;
-	if (node1==node2)
+	if (node1 == node2)
 		return;
-	if (node2==node0)
+	if (node2 == node0)
 		return;
 
-	appendFace(-1,mat);
-	Face&	f=m_faces[m_faces.size()-1];
-	btAssert(node0!=node1);
-	btAssert(node1!=node2);
-	btAssert(node2!=node0);
-	f.m_n[0]	=	&m_nodes[node0];
-	f.m_n[1]	=	&m_nodes[node1];
-	f.m_n[2]	=	&m_nodes[node2];
-	f.m_ra		=	AreaOf(	f.m_n[0]->m_x,
-		f.m_n[1]->m_x,
-		f.m_n[2]->m_x);	
-	m_bUpdateRtCst=true;
+	appendFace(-1, mat);
+	Face& f = m_faces[m_faces.size() - 1];
+	btAssert(node0 != node1);
+	btAssert(node1 != node2);
+	btAssert(node2 != node0);
+	f.m_n[0] = &m_nodes[node0];
+	f.m_n[1] = &m_nodes[node1];
+	f.m_n[2] = &m_nodes[node2];
+	f.m_ra = AreaOf(f.m_n[0]->m_x,
+					f.m_n[1]->m_x,
+					f.m_n[2]->m_x);
+	m_bUpdateRtCst = true;
 }
 
 //
-void			btSoftBody::appendTetra(int model,Material* mat)
+void btSoftBody::appendTetra(int model, Material* mat)
 {
-Tetra	t;
-if(model>=0)
-	t=m_tetras[model];
+	Tetra t;
+	if (model >= 0)
+		t = m_tetras[model];
 	else
-	{ ZeroInitialize(t);t.m_material=mat?mat:m_materials[0]; }
-m_tetras.push_back(t);
+	{
+		ZeroInitialize(t);
+		t.m_material = mat ? mat : m_materials[0];
+	}
+	m_tetras.push_back(t);
 }
 
 //
-void			btSoftBody::appendTetra(int node0,
-										int node1,
-										int node2,
-										int node3,
-										Material* mat)
+void btSoftBody::appendTetra(int node0,
+							 int node1,
+							 int node2,
+							 int node3,
+							 Material* mat)
 {
-	appendTetra(-1,mat);
-	Tetra&	t=m_tetras[m_tetras.size()-1];
-	t.m_n[0]	=	&m_nodes[node0];
-	t.m_n[1]	=	&m_nodes[node1];
-	t.m_n[2]	=	&m_nodes[node2];
-	t.m_n[3]	=	&m_nodes[node3];
-	t.m_rv		=	VolumeOf(t.m_n[0]->m_x,t.m_n[1]->m_x,t.m_n[2]->m_x,t.m_n[3]->m_x);
-	m_bUpdateRtCst=true;
+	appendTetra(-1, mat);
+	Tetra& t = m_tetras[m_tetras.size() - 1];
+	t.m_n[0] = &m_nodes[node0];
+	t.m_n[1] = &m_nodes[node1];
+	t.m_n[2] = &m_nodes[node2];
+	t.m_n[3] = &m_nodes[node3];
+	t.m_rv = VolumeOf(t.m_n[0]->m_x, t.m_n[1]->m_x, t.m_n[2]->m_x, t.m_n[3]->m_x);
+	m_bUpdateRtCst = true;
 }
 
 //
 
-void			btSoftBody::appendAnchor(int node,btRigidBody* body, bool disableCollisionBetweenLinkedBodies,btScalar influence)
+void btSoftBody::appendAnchor(int node, btRigidBody* body, bool disableCollisionBetweenLinkedBodies, btScalar influence)
 {
-	btVector3 local = body->getWorldTransform().inverse()*m_nodes[node].m_x;
-	appendAnchor(node,body,local,disableCollisionBetweenLinkedBodies,influence);
+	btVector3 local = body->getWorldTransform().inverse() * m_nodes[node].m_x;
+	appendAnchor(node, body, local, disableCollisionBetweenLinkedBodies, influence);
 }
 
 //
-void			btSoftBody::appendAnchor(int node,btRigidBody* body, const btVector3& localPivot,bool disableCollisionBetweenLinkedBodies,btScalar influence)
+void btSoftBody::appendAnchor(int node, btRigidBody* body, const btVector3& localPivot, bool disableCollisionBetweenLinkedBodies, btScalar influence)
 {
 	if (disableCollisionBetweenLinkedBodies)
 	{
-		if (m_collisionDisabledObjects.findLinearSearch(body)==m_collisionDisabledObjects.size())
+		if (m_collisionDisabledObjects.findLinearSearch(body) == m_collisionDisabledObjects.size())
 		{
 			m_collisionDisabledObjects.push_back(body);
 		}
 	}
 
-	Anchor	a;
-	a.m_node			=	&m_nodes[node];
-	a.m_body			=	body;
-	a.m_local			=	localPivot;
-	a.m_node->m_battach	=	1;
+	Anchor a;
+	a.m_node = &m_nodes[node];
+	a.m_body = body;
+	a.m_local = localPivot;
+	a.m_node->m_battach = 1;
 	a.m_influence = influence;
 	m_anchors.push_back(a);
 }
 
 //
-void			btSoftBody::appendLinearJoint(const LJoint::Specs& specs,Cluster* body0,Body body1)
+void btSoftBody::appendLinearJoint(const LJoint::Specs& specs, Cluster* body0, Body body1)
 {
-	LJoint*		pj	=	new(btAlignedAlloc(sizeof(LJoint),16)) LJoint();
-	pj->m_bodies[0]	=	body0;
-	pj->m_bodies[1]	=	body1;
-	pj->m_refs[0]	=	pj->m_bodies[0].xform().inverse()*specs.position;
-	pj->m_refs[1]	=	pj->m_bodies[1].xform().inverse()*specs.position;
-	pj->m_cfm		=	specs.cfm;
-	pj->m_erp		=	specs.erp;
-	pj->m_split		=	specs.split;
+	LJoint* pj = new (btAlignedAlloc(sizeof(LJoint), 16)) LJoint();
+	pj->m_bodies[0] = body0;
+	pj->m_bodies[1] = body1;
+	pj->m_refs[0] = pj->m_bodies[0].xform().inverse() * specs.position;
+	pj->m_refs[1] = pj->m_bodies[1].xform().inverse() * specs.position;
+	pj->m_cfm = specs.cfm;
+	pj->m_erp = specs.erp;
+	pj->m_split = specs.split;
 	m_joints.push_back(pj);
 }
 
 //
-void			btSoftBody::appendLinearJoint(const LJoint::Specs& specs,Body body)
+void btSoftBody::appendLinearJoint(const LJoint::Specs& specs, Body body)
 {
-	appendLinearJoint(specs,m_clusters[0],body);
+	appendLinearJoint(specs, m_clusters[0], body);
 }
 
 //
-void			btSoftBody::appendLinearJoint(const LJoint::Specs& specs,btSoftBody* body)
+void btSoftBody::appendLinearJoint(const LJoint::Specs& specs, btSoftBody* body)
 {
-	appendLinearJoint(specs,m_clusters[0],body->m_clusters[0]);
+	appendLinearJoint(specs, m_clusters[0], body->m_clusters[0]);
 }
 
 //
-void			btSoftBody::appendAngularJoint(const AJoint::Specs& specs,Cluster* body0,Body body1)
+void btSoftBody::appendAngularJoint(const AJoint::Specs& specs, Cluster* body0, Body body1)
 {
-	AJoint*		pj	=	new(btAlignedAlloc(sizeof(AJoint),16)) AJoint();
-	pj->m_bodies[0]	=	body0;
-	pj->m_bodies[1]	=	body1;
-	pj->m_refs[0]	=	pj->m_bodies[0].xform().inverse().getBasis()*specs.axis;
-	pj->m_refs[1]	=	pj->m_bodies[1].xform().inverse().getBasis()*specs.axis;
-	pj->m_cfm		=	specs.cfm;
-	pj->m_erp		=	specs.erp;
-	pj->m_split		=	specs.split;
-	pj->m_icontrol	=	specs.icontrol;
+	AJoint* pj = new (btAlignedAlloc(sizeof(AJoint), 16)) AJoint();
+	pj->m_bodies[0] = body0;
+	pj->m_bodies[1] = body1;
+	pj->m_refs[0] = pj->m_bodies[0].xform().inverse().getBasis() * specs.axis;
+	pj->m_refs[1] = pj->m_bodies[1].xform().inverse().getBasis() * specs.axis;
+	pj->m_cfm = specs.cfm;
+	pj->m_erp = specs.erp;
+	pj->m_split = specs.split;
+	pj->m_icontrol = specs.icontrol;
 	m_joints.push_back(pj);
 }
 
 //
-void			btSoftBody::appendAngularJoint(const AJoint::Specs& specs,Body body)
+void btSoftBody::appendAngularJoint(const AJoint::Specs& specs, Body body)
 {
-	appendAngularJoint(specs,m_clusters[0],body);
+	appendAngularJoint(specs, m_clusters[0], body);
 }
 
 //
-void			btSoftBody::appendAngularJoint(const AJoint::Specs& specs,btSoftBody* body)
+void btSoftBody::appendAngularJoint(const AJoint::Specs& specs, btSoftBody* body)
 {
-	appendAngularJoint(specs,m_clusters[0],body->m_clusters[0]);
+	appendAngularJoint(specs, m_clusters[0], body->m_clusters[0]);
 }
 
 //
-void			btSoftBody::addForce(const btVector3& force)
+void btSoftBody::addForce(const btVector3& force)
 {
-	for(int i=0,ni=m_nodes.size();i<ni;++i) addForce(force,i);
+	for (int i = 0, ni = m_nodes.size(); i < ni; ++i) addForce(force, i);
 }
 
 //
-void			btSoftBody::addForce(const btVector3& force,int node)
+void btSoftBody::addForce(const btVector3& force, int node)
 {
-	Node&	n=m_nodes[node];
-	if(n.m_im>0)
+	Node& n = m_nodes[node];
+	if (n.m_im > 0)
 	{
-		n.m_f	+=	force;
+		n.m_f += force;
 	}
 }
 
-void			btSoftBody::addAeroForceToNode(const btVector3& windVelocity,int nodeIndex)
+void btSoftBody::addAeroForceToNode(const btVector3& windVelocity, int nodeIndex)
 {
 	btAssert(nodeIndex >= 0 && nodeIndex < m_nodes.size());
 
@@ -464,51 +479,51 @@ void			btSoftBody::addAeroForceToNode(const btVector3& windVelocity,int nodeInde
 	const btScalar kDG = m_cfg.kDG;
 	//const btScalar kPR = m_cfg.kPR;
 	//const btScalar kVC = m_cfg.kVC;
-	const bool as_lift = kLF>0;
-	const bool as_drag = kDG>0;
+	const bool as_lift = kLF > 0;
+	const bool as_drag = kDG > 0;
 	const bool as_aero = as_lift || as_drag;
 	const bool as_vaero = as_aero && (m_cfg.aeromodel < btSoftBody::eAeroModel::F_TwoSided);
 
 	Node& n = m_nodes[nodeIndex];
 
-	if( n.m_im>0 )
+	if (n.m_im > 0)
 	{
-		btSoftBody::sMedium	medium;
+		btSoftBody::sMedium medium;
 
 		EvaluateMedium(m_worldInfo, n.m_x, medium);
 		medium.m_velocity = windVelocity;
 		medium.m_density = m_worldInfo->air_density;
 
-		/* Aerodynamics			*/ 
-		if(as_vaero)
-		{				
-			const btVector3	rel_v = n.m_v - medium.m_velocity;					
+		/* Aerodynamics			*/
+		if (as_vaero)
+		{
+			const btVector3 rel_v = n.m_v - medium.m_velocity;
 			const btScalar rel_v_len = rel_v.length();
-			const btScalar	rel_v2 = rel_v.length2();
+			const btScalar rel_v2 = rel_v.length2();
 
-			if(rel_v2>SIMD_EPSILON)
+			if (rel_v2 > SIMD_EPSILON)
 			{
 				const btVector3 rel_v_nrm = rel_v.normalized();
-				btVector3	nrm = n.m_n;						
+				btVector3 nrm = n.m_n;
 
 				if (m_cfg.aeromodel == btSoftBody::eAeroModel::V_TwoSidedLiftDrag)
 				{
-					nrm *= (btScalar)( (btDot(nrm,rel_v) < 0) ? -1 : +1);
+					nrm *= (btScalar)((btDot(nrm, rel_v) < 0) ? -1 : +1);
 					btVector3 fDrag(0, 0, 0);
 					btVector3 fLift(0, 0, 0);
 
 					btScalar n_dot_v = nrm.dot(rel_v_nrm);
 					btScalar tri_area = 0.5f * n.m_area;
-							
+
 					fDrag = 0.5f * kDG * medium.m_density * rel_v2 * tri_area * n_dot_v * (-rel_v_nrm);
-							
+
 					// Check angle of attack
 					// cos(10º) = 0.98480
-					if ( 0 < n_dot_v && n_dot_v < 0.98480f)
-						fLift = 0.5f * kLF * medium.m_density * rel_v_len * tri_area * btSqrt(1.0f-n_dot_v*n_dot_v) * (nrm.cross(rel_v_nrm).cross(rel_v_nrm));
+					if (0 < n_dot_v && n_dot_v < 0.98480f)
+						fLift = 0.5f * kLF * medium.m_density * rel_v_len * tri_area * btSqrt(1.0f - n_dot_v * n_dot_v) * (nrm.cross(rel_v_nrm).cross(rel_v_nrm));
 
 					// Check if the velocity change resulted by aero drag force exceeds the current velocity of the node.
-					btVector3 del_v_by_fDrag = fDrag*n.m_im*m_sst.sdt;										
+					btVector3 del_v_by_fDrag = fDrag * n.m_im * m_sst.sdt;
 					btScalar del_v_by_fDrag_len2 = del_v_by_fDrag.length2();
 					btScalar v_len2 = n.m_v.length2();
 
@@ -516,7 +531,7 @@ void			btSoftBody::addAeroForceToNode(const btVector3& windVelocity,int nodeInde
 					{
 						btScalar del_v_by_fDrag_len = del_v_by_fDrag.length();
 						btScalar v_len = n.m_v.length();
-						fDrag *= btScalar(0.8)*(v_len / del_v_by_fDrag_len);
+						fDrag *= btScalar(0.8) * (v_len / del_v_by_fDrag_len);
 					}
 
 					n.m_f += fDrag;
@@ -525,83 +540,83 @@ void			btSoftBody::addAeroForceToNode(const btVector3& windVelocity,int nodeInde
 				else if (m_cfg.aeromodel == btSoftBody::eAeroModel::V_Point || m_cfg.aeromodel == btSoftBody::eAeroModel::V_OneSided || m_cfg.aeromodel == btSoftBody::eAeroModel::V_TwoSided)
 				{
 					if (m_cfg.aeromodel == btSoftBody::eAeroModel::V_TwoSided)
-						nrm *= (btScalar)( (btDot(nrm,rel_v) < 0) ? -1 : +1);
+						nrm *= (btScalar)((btDot(nrm, rel_v) < 0) ? -1 : +1);
 
-					const btScalar dvn = btDot(rel_v,nrm);
-					/* Compute forces	*/ 
-					if(dvn>0)
+					const btScalar dvn = btDot(rel_v, nrm);
+					/* Compute forces	*/
+					if (dvn > 0)
 					{
-						btVector3		force(0,0,0);
-						const btScalar	c0	=	n.m_area * dvn * rel_v2/2;
-						const btScalar	c1	=	c0 * medium.m_density;
-						force	+=	nrm*(-c1*kLF);
-						force	+=	rel_v.normalized() * (-c1 * kDG);
+						btVector3 force(0, 0, 0);
+						const btScalar c0 = n.m_area * dvn * rel_v2 / 2;
+						const btScalar c1 = c0 * medium.m_density;
+						force += nrm * (-c1 * kLF);
+						force += rel_v.normalized() * (-c1 * kDG);
 						ApplyClampedForce(n, force, dt);
 					}
-				}	
+				}
 			}
 		}
 	}
 }
 
-void			btSoftBody::addAeroForceToFace(const btVector3& windVelocity,int faceIndex)
+void btSoftBody::addAeroForceToFace(const btVector3& windVelocity, int faceIndex)
 {
 	const btScalar dt = m_sst.sdt;
 	const btScalar kLF = m_cfg.kLF;
 	const btScalar kDG = m_cfg.kDG;
-//	const btScalar kPR = m_cfg.kPR;
-//	const btScalar kVC = m_cfg.kVC;
-	const bool as_lift = kLF>0;
-	const bool as_drag = kDG>0;
+	//	const btScalar kPR = m_cfg.kPR;
+	//	const btScalar kVC = m_cfg.kVC;
+	const bool as_lift = kLF > 0;
+	const bool as_drag = kDG > 0;
 	const bool as_aero = as_lift || as_drag;
 	const bool as_faero = as_aero && (m_cfg.aeromodel >= btSoftBody::eAeroModel::F_TwoSided);
 
-	if(as_faero)
+	if (as_faero)
 	{
-		btSoftBody::Face&	f=m_faces[faceIndex];
+		btSoftBody::Face& f = m_faces[faceIndex];
 
-		btSoftBody::sMedium	medium;
-		
-		const btVector3	v=(f.m_n[0]->m_v+f.m_n[1]->m_v+f.m_n[2]->m_v)/3;
-		const btVector3	x=(f.m_n[0]->m_x+f.m_n[1]->m_x+f.m_n[2]->m_x)/3;
-		EvaluateMedium(m_worldInfo,x,medium);
+		btSoftBody::sMedium medium;
+
+		const btVector3 v = (f.m_n[0]->m_v + f.m_n[1]->m_v + f.m_n[2]->m_v) / 3;
+		const btVector3 x = (f.m_n[0]->m_x + f.m_n[1]->m_x + f.m_n[2]->m_x) / 3;
+		EvaluateMedium(m_worldInfo, x, medium);
 		medium.m_velocity = windVelocity;
 		medium.m_density = m_worldInfo->air_density;
-		const btVector3	rel_v=v-medium.m_velocity;
+		const btVector3 rel_v = v - medium.m_velocity;
 		const btScalar rel_v_len = rel_v.length();
-		const btScalar	rel_v2=rel_v.length2();
+		const btScalar rel_v2 = rel_v.length2();
 
-		if(rel_v2>SIMD_EPSILON)
+		if (rel_v2 > SIMD_EPSILON)
 		{
 			const btVector3 rel_v_nrm = rel_v.normalized();
-			btVector3	nrm = f.m_normal;
+			btVector3 nrm = f.m_normal;
 
 			if (m_cfg.aeromodel == btSoftBody::eAeroModel::F_TwoSidedLiftDrag)
 			{
-				nrm *= (btScalar)( (btDot(nrm,rel_v) < 0) ? -1 : +1);
+				nrm *= (btScalar)((btDot(nrm, rel_v) < 0) ? -1 : +1);
 
 				btVector3 fDrag(0, 0, 0);
 				btVector3 fLift(0, 0, 0);
 
 				btScalar n_dot_v = nrm.dot(rel_v_nrm);
 				btScalar tri_area = 0.5f * f.m_ra;
-					
+
 				fDrag = 0.5f * kDG * medium.m_density * rel_v2 * tri_area * n_dot_v * (-rel_v_nrm);
 
 				// Check angle of attack
 				// cos(10º) = 0.98480
-				if ( 0 < n_dot_v && n_dot_v < 0.98480f)
-					fLift = 0.5f * kLF * medium.m_density * rel_v_len * tri_area * btSqrt(1.0f-n_dot_v*n_dot_v) * (nrm.cross(rel_v_nrm).cross(rel_v_nrm));
+				if (0 < n_dot_v && n_dot_v < 0.98480f)
+					fLift = 0.5f * kLF * medium.m_density * rel_v_len * tri_area * btSqrt(1.0f - n_dot_v * n_dot_v) * (nrm.cross(rel_v_nrm).cross(rel_v_nrm));
 
 				fDrag /= 3;
 				fLift /= 3;
 
-				for(int j=0;j<3;++j) 
+				for (int j = 0; j < 3; ++j)
 				{
-					if (f.m_n[j]->m_im>0)
+					if (f.m_n[j]->m_im > 0)
 					{
 						// Check if the velocity change resulted by aero drag force exceeds the current velocity of the node.
-						btVector3 del_v_by_fDrag = fDrag*f.m_n[j]->m_im*m_sst.sdt;										
+						btVector3 del_v_by_fDrag = fDrag * f.m_n[j]->m_im * m_sst.sdt;
 						btScalar del_v_by_fDrag_len2 = del_v_by_fDrag.length2();
 						btScalar v_len2 = f.m_n[j]->m_v.length2();
 
@@ -609,10 +624,10 @@ void			btSoftBody::addAeroForceToFace(const btVector3& windVelocity,int faceInde
 						{
 							btScalar del_v_by_fDrag_len = del_v_by_fDrag.length();
 							btScalar v_len = f.m_n[j]->m_v.length();
-							fDrag *= btScalar(0.8)*(v_len / del_v_by_fDrag_len);
+							fDrag *= btScalar(0.8) * (v_len / del_v_by_fDrag_len);
 						}
 
-						f.m_n[j]->m_f += fDrag; 
+						f.m_n[j]->m_f += fDrag;
 						f.m_n[j]->m_f += fLift;
 					}
 				}
@@ -620,183 +635,181 @@ void			btSoftBody::addAeroForceToFace(const btVector3& windVelocity,int faceInde
 			else if (m_cfg.aeromodel == btSoftBody::eAeroModel::F_OneSided || m_cfg.aeromodel == btSoftBody::eAeroModel::F_TwoSided)
 			{
 				if (m_cfg.aeromodel == btSoftBody::eAeroModel::F_TwoSided)
-					nrm *= (btScalar)( (btDot(nrm,rel_v) < 0) ? -1 : +1);
+					nrm *= (btScalar)((btDot(nrm, rel_v) < 0) ? -1 : +1);
 
-				const btScalar	dvn=btDot(rel_v,nrm);
-				/* Compute forces	*/ 
-				if(dvn>0)
+				const btScalar dvn = btDot(rel_v, nrm);
+				/* Compute forces	*/
+				if (dvn > 0)
 				{
-					btVector3		force(0,0,0);
-					const btScalar	c0	=	f.m_ra*dvn*rel_v2;
-					const btScalar	c1	=	c0*medium.m_density;
-					force	+=	nrm*(-c1*kLF);
-					force	+=	rel_v.normalized()*(-c1*kDG);
-					force	/=	3;
-					for(int j=0;j<3;++j) ApplyClampedForce(*f.m_n[j],force,dt);
+					btVector3 force(0, 0, 0);
+					const btScalar c0 = f.m_ra * dvn * rel_v2;
+					const btScalar c1 = c0 * medium.m_density;
+					force += nrm * (-c1 * kLF);
+					force += rel_v.normalized() * (-c1 * kDG);
+					force /= 3;
+					for (int j = 0; j < 3; ++j) ApplyClampedForce(*f.m_n[j], force, dt);
 				}
 			}
 		}
 	}
-
 }
 
 //
-void			btSoftBody::addVelocity(const btVector3& velocity)
+void btSoftBody::addVelocity(const btVector3& velocity)
 {
-	for(int i=0,ni=m_nodes.size();i<ni;++i) addVelocity(velocity,i);
+	for (int i = 0, ni = m_nodes.size(); i < ni; ++i) addVelocity(velocity, i);
 }
 
-/* Set velocity for the entire body										*/ 
-void				btSoftBody::setVelocity(	const btVector3& velocity)
+/* Set velocity for the entire body										*/
+void btSoftBody::setVelocity(const btVector3& velocity)
 {
-	for(int i=0,ni=m_nodes.size();i<ni;++i) 
+	for (int i = 0, ni = m_nodes.size(); i < ni; ++i)
 	{
-		Node&	n=m_nodes[i];
-		if(n.m_im>0)
+		Node& n = m_nodes[i];
+		if (n.m_im > 0)
 		{
-			n.m_v	=	velocity;
+			n.m_v = velocity;
 		}
 	}
 }
 
-
 //
-void			btSoftBody::addVelocity(const btVector3& velocity,int node)
+void btSoftBody::addVelocity(const btVector3& velocity, int node)
 {
-	Node&	n=m_nodes[node];
-	if(n.m_im>0)
+	Node& n = m_nodes[node];
+	if (n.m_im > 0)
 	{
-		n.m_v	+=	velocity;
+		n.m_v += velocity;
 	}
 }
 
 //
-void			btSoftBody::setMass(int node,btScalar mass)
+void btSoftBody::setMass(int node, btScalar mass)
 {
-	m_nodes[node].m_im=mass>0?1/mass:0;
-	m_bUpdateRtCst=true;
+	m_nodes[node].m_im = mass > 0 ? 1 / mass : 0;
+	m_bUpdateRtCst = true;
 }
 
 //
-btScalar		btSoftBody::getMass(int node) const
+btScalar btSoftBody::getMass(int node) const
 {
-	return(m_nodes[node].m_im>0?1/m_nodes[node].m_im:0);
+	return (m_nodes[node].m_im > 0 ? 1 / m_nodes[node].m_im : 0);
 }
 
 //
-btScalar		btSoftBody::getTotalMass() const
+btScalar btSoftBody::getTotalMass() const
 {
-	btScalar	mass=0;
-	for(int i=0;i<m_nodes.size();++i)
+	btScalar mass = 0;
+	for (int i = 0; i < m_nodes.size(); ++i)
 	{
-		mass+=getMass(i);
+		mass += getMass(i);
 	}
-	return(mass);
+	return (mass);
 }
 
 //
-void			btSoftBody::setTotalMass(btScalar mass,bool fromfaces)
+void btSoftBody::setTotalMass(btScalar mass, bool fromfaces)
 {
 	int i;
 
-	if(fromfaces)
+	if (fromfaces)
 	{
-
-		for(i=0;i<m_nodes.size();++i)
+		for (i = 0; i < m_nodes.size(); ++i)
 		{
-			m_nodes[i].m_im=0;
+			m_nodes[i].m_im = 0;
 		}
-		for(i=0;i<m_faces.size();++i)
+		for (i = 0; i < m_faces.size(); ++i)
 		{
-			const Face&		f=m_faces[i];
-			const btScalar	twicearea=AreaOf(	f.m_n[0]->m_x,
-				f.m_n[1]->m_x,
-				f.m_n[2]->m_x);
-			for(int j=0;j<3;++j)
+			const Face& f = m_faces[i];
+			const btScalar twicearea = AreaOf(f.m_n[0]->m_x,
+											  f.m_n[1]->m_x,
+											  f.m_n[2]->m_x);
+			for (int j = 0; j < 3; ++j)
 			{
-				f.m_n[j]->m_im+=twicearea;
+				f.m_n[j]->m_im += twicearea;
 			}
 		}
-		for( i=0;i<m_nodes.size();++i)
+		for (i = 0; i < m_nodes.size(); ++i)
 		{
-			m_nodes[i].m_im=1/m_nodes[i].m_im;
+			m_nodes[i].m_im = 1 / m_nodes[i].m_im;
 		}
 	}
-	const btScalar	tm=getTotalMass();
-	const btScalar	itm=1/tm;
-	for( i=0;i<m_nodes.size();++i)
+	const btScalar tm = getTotalMass();
+	const btScalar itm = 1 / tm;
+	for (i = 0; i < m_nodes.size(); ++i)
 	{
-		m_nodes[i].m_im/=itm*mass;
+		m_nodes[i].m_im /= itm * mass;
 	}
-	m_bUpdateRtCst=true;
+	m_bUpdateRtCst = true;
 }
 
 //
-void			btSoftBody::setTotalDensity(btScalar density)
+void btSoftBody::setTotalDensity(btScalar density)
 {
-	setTotalMass(getVolume()*density,true);
+	setTotalMass(getVolume() * density, true);
 }
 
 //
-void			btSoftBody::setVolumeMass(btScalar mass)
+void btSoftBody::setVolumeMass(btScalar mass)
 {
-btAlignedObjectArray<btScalar>	ranks;
-ranks.resize(m_nodes.size(),0);
-int i;
+	btAlignedObjectArray<btScalar> ranks;
+	ranks.resize(m_nodes.size(), 0);
+	int i;
 
-for(i=0;i<m_nodes.size();++i)
+	for (i = 0; i < m_nodes.size(); ++i)
 	{
-	m_nodes[i].m_im=0;
+		m_nodes[i].m_im = 0;
 	}
-for(i=0;i<m_tetras.size();++i)
+	for (i = 0; i < m_tetras.size(); ++i)
 	{
-	const Tetra& t=m_tetras[i];
-	for(int j=0;j<4;++j)
+		const Tetra& t = m_tetras[i];
+		for (int j = 0; j < 4; ++j)
 		{
-		t.m_n[j]->m_im+=btFabs(t.m_rv);
-		ranks[int(t.m_n[j]-&m_nodes[0])]+=1;
+			t.m_n[j]->m_im += btFabs(t.m_rv);
+			ranks[int(t.m_n[j] - &m_nodes[0])] += 1;
 		}
 	}
-for( i=0;i<m_nodes.size();++i)
+	for (i = 0; i < m_nodes.size(); ++i)
 	{
-	if(m_nodes[i].m_im>0)
+		if (m_nodes[i].m_im > 0)
 		{
-		m_nodes[i].m_im=ranks[i]/m_nodes[i].m_im;
+			m_nodes[i].m_im = ranks[i] / m_nodes[i].m_im;
 		}
 	}
-setTotalMass(mass,false);
+	setTotalMass(mass, false);
 }
 
 //
-void			btSoftBody::setVolumeDensity(btScalar density)
+void btSoftBody::setVolumeDensity(btScalar density)
 {
-btScalar	volume=0;
-for(int i=0;i<m_tetras.size();++i)
+	btScalar volume = 0;
+	for (int i = 0; i < m_tetras.size(); ++i)
 	{
-	const Tetra& t=m_tetras[i];
-	for(int j=0;j<4;++j)
+		const Tetra& t = m_tetras[i];
+		for (int j = 0; j < 4; ++j)
 		{
-		volume+=btFabs(t.m_rv);
+			volume += btFabs(t.m_rv);
 		}
 	}
-setVolumeMass(volume*density/6);
+	setVolumeMass(volume * density / 6);
 }
 
 //
-void			btSoftBody::transform(const btTransform& trs)
+void btSoftBody::transform(const btTransform& trs)
 {
-	const btScalar	margin=getCollisionShape()->getMargin();
-	ATTRIBUTE_ALIGNED16(btDbvtVolume)	vol;
-	
-	for(int i=0,ni=m_nodes.size();i<ni;++i)
+	const btScalar margin = getCollisionShape()->getMargin();
+	ATTRIBUTE_ALIGNED16(btDbvtVolume)
+	vol;
+
+	for (int i = 0, ni = m_nodes.size(); i < ni; ++i)
 	{
-		Node&	n=m_nodes[i];
-		n.m_x=trs*n.m_x;
-		n.m_q=trs*n.m_q;
-		n.m_n=trs.getBasis()*n.m_n;
-		vol = btDbvtVolume::FromCR(n.m_x,margin);
-		
-		m_ndbvt.update(n.m_leaf,vol);
+		Node& n = m_nodes[i];
+		n.m_x = trs * n.m_x;
+		n.m_q = trs * n.m_q;
+		n.m_n = trs.getBasis() * n.m_n;
+		vol = btDbvtVolume::FromCR(n.m_x, margin);
+
+		m_ndbvt.update(n.m_leaf, vol);
 	}
 	updateNormals();
 	updateBounds();
@@ -805,37 +818,37 @@ void			btSoftBody::transform(const btTransform& trs)
 }
 
 //
-void			btSoftBody::translate(const btVector3& trs)
+void btSoftBody::translate(const btVector3& trs)
 {
-	btTransform	t;
+	btTransform t;
 	t.setIdentity();
 	t.setOrigin(trs);
 	transform(t);
 }
 
 //
-void			btSoftBody::rotate(	const btQuaternion& rot)
+void btSoftBody::rotate(const btQuaternion& rot)
 {
-	btTransform	t;
+	btTransform t;
 	t.setIdentity();
 	t.setRotation(rot);
 	transform(t);
 }
 
 //
-void			btSoftBody::scale(const btVector3& scl)
+void btSoftBody::scale(const btVector3& scl)
 {
+	const btScalar margin = getCollisionShape()->getMargin();
+	ATTRIBUTE_ALIGNED16(btDbvtVolume)
+	vol;
 
-	const btScalar	margin=getCollisionShape()->getMargin();
-	ATTRIBUTE_ALIGNED16(btDbvtVolume)	vol;
-	
-	for(int i=0,ni=m_nodes.size();i<ni;++i)
+	for (int i = 0, ni = m_nodes.size(); i < ni; ++i)
 	{
-		Node&	n=m_nodes[i];
-		n.m_x*=scl;
-		n.m_q*=scl;
-		vol = btDbvtVolume::FromCR(n.m_x,margin);
-		m_ndbvt.update(n.m_leaf,vol);
+		Node& n = m_nodes[i];
+		n.m_x *= scl;
+		n.m_q *= scl;
+		vol = btDbvtVolume::FromCR(n.m_x, margin);
+		m_ndbvt.update(n.m_leaf, vol);
 	}
 	updateNormals();
 	updateBounds();
@@ -851,437 +864,432 @@ btScalar btSoftBody::getRestLengthScale()
 //
 void btSoftBody::setRestLengthScale(btScalar restLengthScale)
 {
-	for(int i=0, ni=m_links.size(); i<ni; ++i)
+	for (int i = 0, ni = m_links.size(); i < ni; ++i)
 	{
-		Link&		l=m_links[i];
-		l.m_rl	=	l.m_rl / m_restLengthScale * restLengthScale;
-		l.m_c1	=	l.m_rl*l.m_rl;
+		Link& l = m_links[i];
+		l.m_rl = l.m_rl / m_restLengthScale * restLengthScale;
+		l.m_c1 = l.m_rl * l.m_rl;
 	}
 	m_restLengthScale = restLengthScale;
-	
+
 	if (getActivationState() == ISLAND_SLEEPING)
 		activate();
 }
 
 //
-void			btSoftBody::setPose(bool bvolume,bool bframe)
+void btSoftBody::setPose(bool bvolume, bool bframe)
 {
-	m_pose.m_bvolume	=	bvolume;
-	m_pose.m_bframe		=	bframe;
-	int i,ni;
+	m_pose.m_bvolume = bvolume;
+	m_pose.m_bframe = bframe;
+	int i, ni;
 
-	/* Weights		*/ 
-	const btScalar	omass=getTotalMass();
-	const btScalar	kmass=omass*m_nodes.size()*1000;
-	btScalar		tmass=omass;
+	/* Weights		*/
+	const btScalar omass = getTotalMass();
+	const btScalar kmass = omass * m_nodes.size() * 1000;
+	btScalar tmass = omass;
 	m_pose.m_wgh.resize(m_nodes.size());
-	for(i=0,ni=m_nodes.size();i<ni;++i)
+	for (i = 0, ni = m_nodes.size(); i < ni; ++i)
 	{
-		if(m_nodes[i].m_im<=0) tmass+=kmass;
+		if (m_nodes[i].m_im <= 0) tmass += kmass;
 	}
-	for( i=0,ni=m_nodes.size();i<ni;++i)
+	for (i = 0, ni = m_nodes.size(); i < ni; ++i)
 	{
-		Node&	n=m_nodes[i];
-		m_pose.m_wgh[i]=	n.m_im>0					?
-			1/(m_nodes[i].m_im*tmass)	:
-		kmass/tmass;
+		Node& n = m_nodes[i];
+		m_pose.m_wgh[i] = n.m_im > 0 ? 1 / (m_nodes[i].m_im * tmass) : kmass / tmass;
 	}
-	/* Pos		*/ 
-	const btVector3	com=evaluateCom();
+	/* Pos		*/
+	const btVector3 com = evaluateCom();
 	m_pose.m_pos.resize(m_nodes.size());
-	for( i=0,ni=m_nodes.size();i<ni;++i)
+	for (i = 0, ni = m_nodes.size(); i < ni; ++i)
 	{
-		m_pose.m_pos[i]=m_nodes[i].m_x-com;
+		m_pose.m_pos[i] = m_nodes[i].m_x - com;
 	}
-	m_pose.m_volume	=	bvolume?getVolume():0;
-	m_pose.m_com	=	com;
+	m_pose.m_volume = bvolume ? getVolume() : 0;
+	m_pose.m_com = com;
 	m_pose.m_rot.setIdentity();
 	m_pose.m_scl.setIdentity();
-	/* Aqq		*/ 
-	m_pose.m_aqq[0]	=
-		m_pose.m_aqq[1]	=
-		m_pose.m_aqq[2]	=	btVector3(0,0,0);
-	for( i=0,ni=m_nodes.size();i<ni;++i)
+	/* Aqq		*/
+	m_pose.m_aqq[0] =
+		m_pose.m_aqq[1] =
+			m_pose.m_aqq[2] = btVector3(0, 0, 0);
+	for (i = 0, ni = m_nodes.size(); i < ni; ++i)
 	{
-		const btVector3&	q=m_pose.m_pos[i];
-		const btVector3		mq=m_pose.m_wgh[i]*q;
-		m_pose.m_aqq[0]+=mq.x()*q;
-		m_pose.m_aqq[1]+=mq.y()*q;
-		m_pose.m_aqq[2]+=mq.z()*q;
+		const btVector3& q = m_pose.m_pos[i];
+		const btVector3 mq = m_pose.m_wgh[i] * q;
+		m_pose.m_aqq[0] += mq.x() * q;
+		m_pose.m_aqq[1] += mq.y() * q;
+		m_pose.m_aqq[2] += mq.z() * q;
 	}
-	m_pose.m_aqq=m_pose.m_aqq.inverse();
-	
+	m_pose.m_aqq = m_pose.m_aqq.inverse();
+
 	updateConstants();
 }
 
-void				btSoftBody::resetLinkRestLengths()
+void btSoftBody::resetLinkRestLengths()
 {
-	for(int i=0, ni=m_links.size();i<ni;++i)
+	for (int i = 0, ni = m_links.size(); i < ni; ++i)
 	{
-		Link& l =	m_links[i];
-		l.m_rl	=	(l.m_n[0]->m_x-l.m_n[1]->m_x).length();
-		l.m_c1	=	l.m_rl*l.m_rl;
+		Link& l = m_links[i];
+		l.m_rl = (l.m_n[0]->m_x - l.m_n[1]->m_x).length();
+		l.m_c1 = l.m_rl * l.m_rl;
 	}
 }
 
 //
-btScalar		btSoftBody::getVolume() const
+btScalar btSoftBody::getVolume() const
 {
-	btScalar	vol=0;
-	if(m_nodes.size()>0)
+	btScalar vol = 0;
+	if (m_nodes.size() > 0)
 	{
-		int i,ni;
+		int i, ni;
 
-		const btVector3	org=m_nodes[0].m_x;
-		for(i=0,ni=m_faces.size();i<ni;++i)
+		const btVector3 org = m_nodes[0].m_x;
+		for (i = 0, ni = m_faces.size(); i < ni; ++i)
 		{
-			const Face&	f=m_faces[i];
-			vol+=btDot(f.m_n[0]->m_x-org,btCross(f.m_n[1]->m_x-org,f.m_n[2]->m_x-org));
+			const Face& f = m_faces[i];
+			vol += btDot(f.m_n[0]->m_x - org, btCross(f.m_n[1]->m_x - org, f.m_n[2]->m_x - org));
 		}
-		vol/=(btScalar)6;
+		vol /= (btScalar)6;
 	}
-	return(vol);
+	return (vol);
 }
 
 //
-int				btSoftBody::clusterCount() const
+int btSoftBody::clusterCount() const
 {
-	return(m_clusters.size());
+	return (m_clusters.size());
 }
 
 //
-btVector3		btSoftBody::clusterCom(const Cluster* cluster)
+btVector3 btSoftBody::clusterCom(const Cluster* cluster)
 {
-	btVector3		com(0,0,0);
-	for(int i=0,ni=cluster->m_nodes.size();i<ni;++i)
+	btVector3 com(0, 0, 0);
+	for (int i = 0, ni = cluster->m_nodes.size(); i < ni; ++i)
 	{
-		com+=cluster->m_nodes[i]->m_x*cluster->m_masses[i];
+		com += cluster->m_nodes[i]->m_x * cluster->m_masses[i];
 	}
-	return(com*cluster->m_imass);
+	return (com * cluster->m_imass);
 }
 
 //
-btVector3		btSoftBody::clusterCom(int cluster) const
+btVector3 btSoftBody::clusterCom(int cluster) const
 {
-	return(clusterCom(m_clusters[cluster]));
+	return (clusterCom(m_clusters[cluster]));
 }
 
 //
-btVector3		btSoftBody::clusterVelocity(const Cluster* cluster,const btVector3& rpos)
+btVector3 btSoftBody::clusterVelocity(const Cluster* cluster, const btVector3& rpos)
 {
-	return(cluster->m_lv+btCross(cluster->m_av,rpos));
+	return (cluster->m_lv + btCross(cluster->m_av, rpos));
 }
 
 //
-void			btSoftBody::clusterVImpulse(Cluster* cluster,const btVector3& rpos,const btVector3& impulse)
+void btSoftBody::clusterVImpulse(Cluster* cluster, const btVector3& rpos, const btVector3& impulse)
 {
-	const btVector3	li=cluster->m_imass*impulse;
-	const btVector3	ai=cluster->m_invwi*btCross(rpos,impulse);
-	cluster->m_vimpulses[0]+=li;cluster->m_lv+=li;
-	cluster->m_vimpulses[1]+=ai;cluster->m_av+=ai;
+	const btVector3 li = cluster->m_imass * impulse;
+	const btVector3 ai = cluster->m_invwi * btCross(rpos, impulse);
+	cluster->m_vimpulses[0] += li;
+	cluster->m_lv += li;
+	cluster->m_vimpulses[1] += ai;
+	cluster->m_av += ai;
 	cluster->m_nvimpulses++;
 }
 
 //
-void			btSoftBody::clusterDImpulse(Cluster* cluster,const btVector3& rpos,const btVector3& impulse)
+void btSoftBody::clusterDImpulse(Cluster* cluster, const btVector3& rpos, const btVector3& impulse)
 {
-	const btVector3	li=cluster->m_imass*impulse;
-	const btVector3	ai=cluster->m_invwi*btCross(rpos,impulse);
-	cluster->m_dimpulses[0]+=li;
-	cluster->m_dimpulses[1]+=ai;
+	const btVector3 li = cluster->m_imass * impulse;
+	const btVector3 ai = cluster->m_invwi * btCross(rpos, impulse);
+	cluster->m_dimpulses[0] += li;
+	cluster->m_dimpulses[1] += ai;
 	cluster->m_ndimpulses++;
 }
 
 //
-void			btSoftBody::clusterImpulse(Cluster* cluster,const btVector3& rpos,const Impulse& impulse)
+void btSoftBody::clusterImpulse(Cluster* cluster, const btVector3& rpos, const Impulse& impulse)
 {
-	if(impulse.m_asVelocity)	clusterVImpulse(cluster,rpos,impulse.m_velocity);
-	if(impulse.m_asDrift)		clusterDImpulse(cluster,rpos,impulse.m_drift);
+	if (impulse.m_asVelocity) clusterVImpulse(cluster, rpos, impulse.m_velocity);
+	if (impulse.m_asDrift) clusterDImpulse(cluster, rpos, impulse.m_drift);
 }
 
 //
-void			btSoftBody::clusterVAImpulse(Cluster* cluster,const btVector3& impulse)
+void btSoftBody::clusterVAImpulse(Cluster* cluster, const btVector3& impulse)
 {
-	const btVector3	ai=cluster->m_invwi*impulse;
-	cluster->m_vimpulses[1]+=ai;cluster->m_av+=ai;
+	const btVector3 ai = cluster->m_invwi * impulse;
+	cluster->m_vimpulses[1] += ai;
+	cluster->m_av += ai;
 	cluster->m_nvimpulses++;
 }
 
 //
-void			btSoftBody::clusterDAImpulse(Cluster* cluster,const btVector3& impulse)
+void btSoftBody::clusterDAImpulse(Cluster* cluster, const btVector3& impulse)
 {
-	const btVector3	ai=cluster->m_invwi*impulse;
-	cluster->m_dimpulses[1]+=ai;
+	const btVector3 ai = cluster->m_invwi * impulse;
+	cluster->m_dimpulses[1] += ai;
 	cluster->m_ndimpulses++;
 }
 
 //
-void			btSoftBody::clusterAImpulse(Cluster* cluster,const Impulse& impulse)
+void btSoftBody::clusterAImpulse(Cluster* cluster, const Impulse& impulse)
 {
-	if(impulse.m_asVelocity)	clusterVAImpulse(cluster,impulse.m_velocity);
-	if(impulse.m_asDrift)		clusterDAImpulse(cluster,impulse.m_drift);
+	if (impulse.m_asVelocity) clusterVAImpulse(cluster, impulse.m_velocity);
+	if (impulse.m_asDrift) clusterDAImpulse(cluster, impulse.m_drift);
 }
 
 //
-void			btSoftBody::clusterDCImpulse(Cluster* cluster,const btVector3& impulse)
+void btSoftBody::clusterDCImpulse(Cluster* cluster, const btVector3& impulse)
 {
-	cluster->m_dimpulses[0]+=impulse*cluster->m_imass;
+	cluster->m_dimpulses[0] += impulse * cluster->m_imass;
 	cluster->m_ndimpulses++;
 }
 
 struct NodeLinks
 {
-    btAlignedObjectArray<int> m_links;
+	btAlignedObjectArray<int> m_links;
 };
 
-
-
 //
-int				btSoftBody::generateBendingConstraints(int distance,Material* mat)
+int btSoftBody::generateBendingConstraints(int distance, Material* mat)
 {
-	int i,j;
+	int i, j;
 
-	if(distance>1)
+	if (distance > 1)
 	{
-		/* Build graph	*/ 
-		const int		n=m_nodes.size();
-		const unsigned	inf=(~(unsigned)0)>>1;
-		unsigned*		adj=new unsigned[n*n];
-		
+		/* Build graph	*/
+		const int n = m_nodes.size();
+		const unsigned inf = (~(unsigned)0) >> 1;
+		unsigned* adj = new unsigned[n * n];
 
-#define IDX(_x_,_y_)	((_y_)*n+(_x_))
-		for(j=0;j<n;++j)
+#define IDX(_x_, _y_) ((_y_)*n + (_x_))
+		for (j = 0; j < n; ++j)
 		{
-			for(i=0;i<n;++i)
+			for (i = 0; i < n; ++i)
 			{
-				if(i!=j)
+				if (i != j)
 				{
-					adj[IDX(i,j)]=adj[IDX(j,i)]=inf;
+					adj[IDX(i, j)] = adj[IDX(j, i)] = inf;
 				}
 				else
 				{
-					adj[IDX(i,j)]=adj[IDX(j,i)]=0;
+					adj[IDX(i, j)] = adj[IDX(j, i)] = 0;
 				}
 			}
 		}
-		for( i=0;i<m_links.size();++i)
+		for (i = 0; i < m_links.size(); ++i)
 		{
-			const int	ia=(int)(m_links[i].m_n[0]-&m_nodes[0]);
-			const int	ib=(int)(m_links[i].m_n[1]-&m_nodes[0]);
-			adj[IDX(ia,ib)]=1;
-			adj[IDX(ib,ia)]=1;
+			const int ia = (int)(m_links[i].m_n[0] - &m_nodes[0]);
+			const int ib = (int)(m_links[i].m_n[1] - &m_nodes[0]);
+			adj[IDX(ia, ib)] = 1;
+			adj[IDX(ib, ia)] = 1;
 		}
 
-
 		//special optimized case for distance == 2
 		if (distance == 2)
 		{
-
 			btAlignedObjectArray<NodeLinks> nodeLinks;
 
-
 			/* Build node links */
 			nodeLinks.resize(m_nodes.size());
 
-			for( i=0;i<m_links.size();++i)
+			for (i = 0; i < m_links.size(); ++i)
 			{
-				const int	ia=(int)(m_links[i].m_n[0]-&m_nodes[0]);
-				const int	ib=(int)(m_links[i].m_n[1]-&m_nodes[0]);
-				if (nodeLinks[ia].m_links.findLinearSearch(ib)==nodeLinks[ia].m_links.size())
+				const int ia = (int)(m_links[i].m_n[0] - &m_nodes[0]);
+				const int ib = (int)(m_links[i].m_n[1] - &m_nodes[0]);
+				if (nodeLinks[ia].m_links.findLinearSearch(ib) == nodeLinks[ia].m_links.size())
 					nodeLinks[ia].m_links.push_back(ib);
 
-				if (nodeLinks[ib].m_links.findLinearSearch(ia)==nodeLinks[ib].m_links.size())
+				if (nodeLinks[ib].m_links.findLinearSearch(ia) == nodeLinks[ib].m_links.size())
 					nodeLinks[ib].m_links.push_back(ia);
 			}
-			for (int ii=0;ii<nodeLinks.size();ii++)
+			for (int ii = 0; ii < nodeLinks.size(); ii++)
 			{
-				int i=ii;
+				int i = ii;
 
-				for (int jj=0;jj<nodeLinks[ii].m_links.size();jj++)
+				for (int jj = 0; jj < nodeLinks[ii].m_links.size(); jj++)
 				{
 					int k = nodeLinks[ii].m_links[jj];
-					for (int kk=0;kk<nodeLinks[k].m_links.size();kk++)
+					for (int kk = 0; kk < nodeLinks[k].m_links.size(); kk++)
 					{
 						int j = nodeLinks[k].m_links[kk];
-						if (i!=j)
+						if (i != j)
 						{
-							const unsigned	sum=adj[IDX(i,k)]+adj[IDX(k,j)];
-							btAssert(sum==2);
-							if(adj[IDX(i,j)]>sum)
+							const unsigned sum = adj[IDX(i, k)] + adj[IDX(k, j)];
+							btAssert(sum == 2);
+							if (adj[IDX(i, j)] > sum)
 							{
-								adj[IDX(i,j)]=adj[IDX(j,i)]=sum;
+								adj[IDX(i, j)] = adj[IDX(j, i)] = sum;
 							}
 						}
-
 					}
 				}
 			}
-		} else
+		}
+		else
 		{
 			///generic Floyd's algorithm
-			for(int k=0;k<n;++k)
+			for (int k = 0; k < n; ++k)
 			{
-				for(j=0;j<n;++j)
+				for (j = 0; j < n; ++j)
 				{
-					for(i=j+1;i<n;++i)
+					for (i = j + 1; i < n; ++i)
 					{
-						const unsigned	sum=adj[IDX(i,k)]+adj[IDX(k,j)];
-						if(adj[IDX(i,j)]>sum)
+						const unsigned sum = adj[IDX(i, k)] + adj[IDX(k, j)];
+						if (adj[IDX(i, j)] > sum)
 						{
-							adj[IDX(i,j)]=adj[IDX(j,i)]=sum;
+							adj[IDX(i, j)] = adj[IDX(j, i)] = sum;
 						}
 					}
 				}
 			}
 		}
 
-
-		/* Build links	*/ 
-		int	nlinks=0;
-		for(j=0;j<n;++j)
+		/* Build links	*/
+		int nlinks = 0;
+		for (j = 0; j < n; ++j)
 		{
-			for(i=j+1;i<n;++i)
+			for (i = j + 1; i < n; ++i)
 			{
-				if(adj[IDX(i,j)]==(unsigned)distance)
+				if (adj[IDX(i, j)] == (unsigned)distance)
 				{
-					appendLink(i,j,mat);
-					m_links[m_links.size()-1].m_bbending=1;
+					appendLink(i, j, mat);
+					m_links[m_links.size() - 1].m_bbending = 1;
 					++nlinks;
 				}
 			}
 		}
-		delete[] adj;		
-		return(nlinks);
+		delete[] adj;
+		return (nlinks);
 	}
-	return(0);
+	return (0);
 }
 
 //
-void			btSoftBody::randomizeConstraints()
+void btSoftBody::randomizeConstraints()
 {
-	unsigned long	seed=243703;
-#define NEXTRAND (seed=(1664525L*seed+1013904223L)&0xffffffff)
-	int i,ni;
+	unsigned long seed = 243703;
+#define NEXTRAND (seed = (1664525L * seed + 1013904223L) & 0xffffffff)
+	int i, ni;
 
-	for(i=0,ni=m_links.size();i<ni;++i)
+	for (i = 0, ni = m_links.size(); i < ni; ++i)
 	{
-		btSwap(m_links[i],m_links[NEXTRAND%ni]);
+		btSwap(m_links[i], m_links[NEXTRAND % ni]);
 	}
-	for(i=0,ni=m_faces.size();i<ni;++i)
+	for (i = 0, ni = m_faces.size(); i < ni; ++i)
 	{
-		btSwap(m_faces[i],m_faces[NEXTRAND%ni]);
+		btSwap(m_faces[i], m_faces[NEXTRAND % ni]);
 	}
 #undef NEXTRAND
 }
 
 //
-void			btSoftBody::releaseCluster(int index)
+void btSoftBody::releaseCluster(int index)
 {
-	Cluster*	c=m_clusters[index];
-	if(c->m_leaf) m_cdbvt.remove(c->m_leaf);
+	Cluster* c = m_clusters[index];
+	if (c->m_leaf) m_cdbvt.remove(c->m_leaf);
 	c->~Cluster();
 	btAlignedFree(c);
 	m_clusters.remove(c);
 }
 
 //
-void			btSoftBody::releaseClusters()
+void btSoftBody::releaseClusters()
 {
-	while(m_clusters.size()>0) releaseCluster(0);
+	while (m_clusters.size() > 0) releaseCluster(0);
 }
 
 //
-int				btSoftBody::generateClusters(int k,int maxiterations)
+int btSoftBody::generateClusters(int k, int maxiterations)
 {
 	int i;
 	releaseClusters();
-	m_clusters.resize(btMin(k,m_nodes.size()));
-	for(i=0;i<m_clusters.size();++i)
+	m_clusters.resize(btMin(k, m_nodes.size()));
+	for (i = 0; i < m_clusters.size(); ++i)
 	{
-		m_clusters[i]			=	new(btAlignedAlloc(sizeof(Cluster),16)) Cluster();
-		m_clusters[i]->m_collide=	true;
+		m_clusters[i] = new (btAlignedAlloc(sizeof(Cluster), 16)) Cluster();
+		m_clusters[i]->m_collide = true;
 	}
-	k=m_clusters.size();
-	if(k>0)
+	k = m_clusters.size();
+	if (k > 0)
 	{
-		/* Initialize		*/ 
-		btAlignedObjectArray<btVector3>	centers;
-		btVector3						cog(0,0,0);
-		int								i;
-		for(i=0;i<m_nodes.size();++i)
+		/* Initialize		*/
+		btAlignedObjectArray<btVector3> centers;
+		btVector3 cog(0, 0, 0);
+		int i;
+		for (i = 0; i < m_nodes.size(); ++i)
 		{
-			cog+=m_nodes[i].m_x;
-			m_clusters[(i*29873)%m_clusters.size()]->m_nodes.push_back(&m_nodes[i]);
+			cog += m_nodes[i].m_x;
+			m_clusters[(i * 29873) % m_clusters.size()]->m_nodes.push_back(&m_nodes[i]);
 		}
-		cog/=(btScalar)m_nodes.size();
-		centers.resize(k,cog);
-		/* Iterate			*/ 
-		const btScalar	slope=16;
-		bool			changed;
-		int				iterations=0;
-		do	{
-			const btScalar	w=2-btMin<btScalar>(1,iterations/slope);
-			changed=false;
-			iterations++;	
+		cog /= (btScalar)m_nodes.size();
+		centers.resize(k, cog);
+		/* Iterate			*/
+		const btScalar slope = 16;
+		bool changed;
+		int iterations = 0;
+		do
+		{
+			const btScalar w = 2 - btMin<btScalar>(1, iterations / slope);
+			changed = false;
+			iterations++;
 			int i;
 
-			for(i=0;i<k;++i)
+			for (i = 0; i < k; ++i)
 			{
-				btVector3	c(0,0,0);
-				for(int j=0;j<m_clusters[i]->m_nodes.size();++j)
+				btVector3 c(0, 0, 0);
+				for (int j = 0; j < m_clusters[i]->m_nodes.size(); ++j)
 				{
-					c+=m_clusters[i]->m_nodes[j]->m_x;
+					c += m_clusters[i]->m_nodes[j]->m_x;
 				}
-				if(m_clusters[i]->m_nodes.size())
+				if (m_clusters[i]->m_nodes.size())
 				{
-					c			/=	(btScalar)m_clusters[i]->m_nodes.size();
-					c			=	centers[i]+(c-centers[i])*w;
-					changed		|=	((c-centers[i]).length2()>SIMD_EPSILON);
-					centers[i]	=	c;
+					c /= (btScalar)m_clusters[i]->m_nodes.size();
+					c = centers[i] + (c - centers[i]) * w;
+					changed |= ((c - centers[i]).length2() > SIMD_EPSILON);
+					centers[i] = c;
 					m_clusters[i]->m_nodes.resize(0);
-				}			
+				}
 			}
-			for(i=0;i<m_nodes.size();++i)
+			for (i = 0; i < m_nodes.size(); ++i)
 			{
-				const btVector3	nx=m_nodes[i].m_x;
-				int				kbest=0;
-				btScalar		kdist=ClusterMetric(centers[0],nx);
-				for(int j=1;j<k;++j)
+				const btVector3 nx = m_nodes[i].m_x;
+				int kbest = 0;
+				btScalar kdist = ClusterMetric(centers[0], nx);
+				for (int j = 1; j < k; ++j)
 				{
-					const btScalar	d=ClusterMetric(centers[j],nx);
-					if(d<kdist)
+					const btScalar d = ClusterMetric(centers[j], nx);
+					if (d < kdist)
 					{
-						kbest=j;
-						kdist=d;
+						kbest = j;
+						kdist = d;
 					}
 				}
 				m_clusters[kbest]->m_nodes.push_back(&m_nodes[i]);
-			}		
-		} while(changed&&(iterations<maxiterations));
-		/* Merge		*/ 
-		btAlignedObjectArray<int>	cids;
-		cids.resize(m_nodes.size(),-1);
-		for(i=0;i<m_clusters.size();++i)
+			}
+		} while (changed && (iterations < maxiterations));
+		/* Merge		*/
+		btAlignedObjectArray<int> cids;
+		cids.resize(m_nodes.size(), -1);
+		for (i = 0; i < m_clusters.size(); ++i)
 		{
-			for(int j=0;j<m_clusters[i]->m_nodes.size();++j)
+			for (int j = 0; j < m_clusters[i]->m_nodes.size(); ++j)
 			{
-				cids[int(m_clusters[i]->m_nodes[j]-&m_nodes[0])]=i;
+				cids[int(m_clusters[i]->m_nodes[j] - &m_nodes[0])] = i;
 			}
 		}
-		for(i=0;i<m_faces.size();++i)
+		for (i = 0; i < m_faces.size(); ++i)
 		{
-			const int idx[]={	int(m_faces[i].m_n[0]-&m_nodes[0]),
-				int(m_faces[i].m_n[1]-&m_nodes[0]),
-				int(m_faces[i].m_n[2]-&m_nodes[0])};
-			for(int j=0;j<3;++j)
+			const int idx[] = {int(m_faces[i].m_n[0] - &m_nodes[0]),
+							   int(m_faces[i].m_n[1] - &m_nodes[0]),
+							   int(m_faces[i].m_n[2] - &m_nodes[0])};
+			for (int j = 0; j < 3; ++j)
 			{
-				const int cid=cids[idx[j]];
-				for(int q=1;q<3;++q)
+				const int cid = cids[idx[j]];
+				for (int q = 1; q < 3; ++q)
 				{
-					const int kid=idx[(j+q)%3];
-					if(cids[kid]!=cid)
+					const int kid = idx[(j + q) % 3];
+					if (cids[kid] != cid)
 					{
-						if(m_clusters[cid]->m_nodes.findLinearSearch(&m_nodes[kid])==m_clusters[cid]->m_nodes.size())
+						if (m_clusters[cid]->m_nodes.findLinearSearch(&m_nodes[kid]) == m_clusters[cid]->m_nodes.size())
 						{
 							m_clusters[cid]->m_nodes.push_back(&m_nodes[kid]);
 						}
@@ -1289,55 +1297,56 @@ int				btSoftBody::generateClusters(int k,int maxiterations)
 				}
 			}
 		}
-		/* Master		*/ 
-		if(m_clusters.size()>1)
+		/* Master		*/
+		if (m_clusters.size() > 1)
 		{
-			Cluster*	pmaster=new(btAlignedAlloc(sizeof(Cluster),16)) Cluster();
-			pmaster->m_collide	=	false;
+			Cluster* pmaster = new (btAlignedAlloc(sizeof(Cluster), 16)) Cluster();
+			pmaster->m_collide = false;
 			pmaster->m_nodes.reserve(m_nodes.size());
-			for(int i=0;i<m_nodes.size();++i) pmaster->m_nodes.push_back(&m_nodes[i]);
+			for (int i = 0; i < m_nodes.size(); ++i) pmaster->m_nodes.push_back(&m_nodes[i]);
 			m_clusters.push_back(pmaster);
-			btSwap(m_clusters[0],m_clusters[m_clusters.size()-1]);
+			btSwap(m_clusters[0], m_clusters[m_clusters.size() - 1]);
 		}
-		/* Terminate	*/ 
-		for(i=0;i<m_clusters.size();++i)
+		/* Terminate	*/
+		for (i = 0; i < m_clusters.size(); ++i)
 		{
-			if(m_clusters[i]->m_nodes.size()==0)
+			if (m_clusters[i]->m_nodes.size() == 0)
 			{
 				releaseCluster(i--);
 			}
 		}
-	} else
+	}
+	else
 	{
 		//create a cluster for each tetrahedron (if tetrahedra exist) or each face
 		if (m_tetras.size())
 		{
 			m_clusters.resize(m_tetras.size());
-			for(i=0;i<m_clusters.size();++i)
+			for (i = 0; i < m_clusters.size(); ++i)
 			{
-				m_clusters[i]			=	new(btAlignedAlloc(sizeof(Cluster),16)) Cluster();
-				m_clusters[i]->m_collide=	true;
+				m_clusters[i] = new (btAlignedAlloc(sizeof(Cluster), 16)) Cluster();
+				m_clusters[i]->m_collide = true;
 			}
-			for (i=0;i<m_tetras.size();i++)
+			for (i = 0; i < m_tetras.size(); i++)
 			{
-				for (int j=0;j<4;j++)
+				for (int j = 0; j < 4; j++)
 				{
 					m_clusters[i]->m_nodes.push_back(m_tetras[i].m_n[j]);
 				}
 			}
-
-		} else
+		}
+		else
 		{
 			m_clusters.resize(m_faces.size());
-			for(i=0;i<m_clusters.size();++i)
+			for (i = 0; i < m_clusters.size(); ++i)
 			{
-				m_clusters[i]			=	new(btAlignedAlloc(sizeof(Cluster),16)) Cluster();
-				m_clusters[i]->m_collide=	true;
+				m_clusters[i] = new (btAlignedAlloc(sizeof(Cluster), 16)) Cluster();
+				m_clusters[i]->m_collide = true;
 			}
 
-			for(i=0;i<m_faces.size();++i)
+			for (i = 0; i < m_faces.size(); ++i)
 			{
-				for(int j=0;j<3;++j)
+				for (int j = 0; j < 3; ++j)
 				{
 					m_clusters[i]->m_nodes.push_back(m_faces[i].m_n[j]);
 				}
@@ -1350,261 +1359,272 @@ int				btSoftBody::generateClusters(int k,int maxiterations)
 		initializeClusters();
 		updateClusters();
 
-
 		//for self-collision
-		m_clusterConnectivity.resize(m_clusters.size()*m_clusters.size());
+		m_clusterConnectivity.resize(m_clusters.size() * m_clusters.size());
 		{
-			for (int c0=0;c0<m_clusters.size();c0++)
+			for (int c0 = 0; c0 < m_clusters.size(); c0++)
 			{
-				m_clusters[c0]->m_clusterIndex=c0;
-				for (int c1=0;c1<m_clusters.size();c1++)
+				m_clusters[c0]->m_clusterIndex = c0;
+				for (int c1 = 0; c1 < m_clusters.size(); c1++)
 				{
-					
-					bool connected=false;
+					bool connected = false;
 					Cluster* cla = m_clusters[c0];
 					Cluster* clb = m_clusters[c1];
-					for (int i=0;!connected&&i<cla->m_nodes.size();i++)
+					for (int i = 0; !connected && i < cla->m_nodes.size(); i++)
 					{
-						for (int j=0;j<clb->m_nodes.size();j++)
+						for (int j = 0; j < clb->m_nodes.size(); j++)
 						{
 							if (cla->m_nodes[i] == clb->m_nodes[j])
 							{
-								connected=true;
+								connected = true;
 								break;
 							}
 						}
 					}
-					m_clusterConnectivity[c0+c1*m_clusters.size()]=connected;
+					m_clusterConnectivity[c0 + c1 * m_clusters.size()] = connected;
 				}
 			}
 		}
 	}
 
-	return(m_clusters.size());
+	return (m_clusters.size());
 }
 
 //
-void			btSoftBody::refine(ImplicitFn* ifn,btScalar accurary,bool cut)
+void btSoftBody::refine(ImplicitFn* ifn, btScalar accurary, bool cut)
 {
-	const Node*			nbase = &m_nodes[0];
-	int					ncount = m_nodes.size();
-	btSymMatrix<int>	edges(ncount,-2);
-	int					newnodes=0;
-	int i,j,k,ni;
+	const Node* nbase = &m_nodes[0];
+	int ncount = m_nodes.size();
+	btSymMatrix<int> edges(ncount, -2);
+	int newnodes = 0;
+	int i, j, k, ni;
 
-	/* Filter out		*/ 
-	for(i=0;i<m_links.size();++i)
+	/* Filter out		*/
+	for (i = 0; i < m_links.size(); ++i)
 	{
-		Link&	l=m_links[i];
-		if(l.m_bbending)
+		Link& l = m_links[i];
+		if (l.m_bbending)
 		{
-			if(!SameSign(ifn->Eval(l.m_n[0]->m_x),ifn->Eval(l.m_n[1]->m_x)))
+			if (!SameSign(ifn->Eval(l.m_n[0]->m_x), ifn->Eval(l.m_n[1]->m_x)))
 			{
-				btSwap(m_links[i],m_links[m_links.size()-1]);
-				m_links.pop_back();--i;
+				btSwap(m_links[i], m_links[m_links.size() - 1]);
+				m_links.pop_back();
+				--i;
 			}
-		}	
+		}
 	}
-	/* Fill edges		*/ 
-	for(i=0;i<m_links.size();++i)
+	/* Fill edges		*/
+	for (i = 0; i < m_links.size(); ++i)
 	{
-		Link&	l=m_links[i];
-		edges(int(l.m_n[0]-nbase),int(l.m_n[1]-nbase))=-1;
+		Link& l = m_links[i];
+		edges(int(l.m_n[0] - nbase), int(l.m_n[1] - nbase)) = -1;
 	}
-	for(i=0;i<m_faces.size();++i)
-	{	
-		Face&	f=m_faces[i];
-		edges(int(f.m_n[0]-nbase),int(f.m_n[1]-nbase))=-1;
-		edges(int(f.m_n[1]-nbase),int(f.m_n[2]-nbase))=-1;
-		edges(int(f.m_n[2]-nbase),int(f.m_n[0]-nbase))=-1;
-	}
-	/* Intersect		*/ 
-	for(i=0;i<ncount;++i)
+	for (i = 0; i < m_faces.size(); ++i)
 	{
-		for(j=i+1;j<ncount;++j)
+		Face& f = m_faces[i];
+		edges(int(f.m_n[0] - nbase), int(f.m_n[1] - nbase)) = -1;
+		edges(int(f.m_n[1] - nbase), int(f.m_n[2] - nbase)) = -1;
+		edges(int(f.m_n[2] - nbase), int(f.m_n[0] - nbase)) = -1;
+	}
+	/* Intersect		*/
+	for (i = 0; i < ncount; ++i)
+	{
+		for (j = i + 1; j < ncount; ++j)
 		{
-			if(edges(i,j)==-1)
+			if (edges(i, j) == -1)
 			{
-				Node&			a=m_nodes[i];
-				Node&			b=m_nodes[j];
-				const btScalar	t=ImplicitSolve(ifn,a.m_x,b.m_x,accurary);
-				if(t>0)
+				Node& a = m_nodes[i];
+				Node& b = m_nodes[j];
+				const btScalar t = ImplicitSolve(ifn, a.m_x, b.m_x, accurary);
+				if (t > 0)
 				{
-					const btVector3	x=Lerp(a.m_x,b.m_x,t);
-					const btVector3	v=Lerp(a.m_v,b.m_v,t);
-					btScalar		m=0;
-					if(a.m_im>0)
+					const btVector3 x = Lerp(a.m_x, b.m_x, t);
+					const btVector3 v = Lerp(a.m_v, b.m_v, t);
+					btScalar m = 0;
+					if (a.m_im > 0)
 					{
-						if(b.m_im>0)
+						if (b.m_im > 0)
 						{
-							const btScalar	ma=1/a.m_im;
-							const btScalar	mb=1/b.m_im;
-							const btScalar	mc=Lerp(ma,mb,t);
-							const btScalar	f=(ma+mb)/(ma+mb+mc);
-							a.m_im=1/(ma*f);
-							b.m_im=1/(mb*f);
-							m=mc*f;
+							const btScalar ma = 1 / a.m_im;
+							const btScalar mb = 1 / b.m_im;
+							const btScalar mc = Lerp(ma, mb, t);
+							const btScalar f = (ma + mb) / (ma + mb + mc);
+							a.m_im = 1 / (ma * f);
+							b.m_im = 1 / (mb * f);
+							m = mc * f;
 						}
 						else
-						{ a.m_im/=0.5f;m=1/a.m_im; }
+						{
+							a.m_im /= 0.5f;
+							m = 1 / a.m_im;
+						}
 					}
 					else
 					{
-						if(b.m_im>0)
-						{ b.m_im/=0.5f;m=1/b.m_im; }
+						if (b.m_im > 0)
+						{
+							b.m_im /= 0.5f;
+							m = 1 / b.m_im;
+						}
 						else
-							m=0;
+							m = 0;
 					}
-					appendNode(x,m);
-					edges(i,j)=m_nodes.size()-1;
-					m_nodes[edges(i,j)].m_v=v;
+					appendNode(x, m);
+					edges(i, j) = m_nodes.size() - 1;
+					m_nodes[edges(i, j)].m_v = v;
 					++newnodes;
 				}
 			}
 		}
 	}
-	nbase=&m_nodes[0];
-	/* Refine links		*/ 
-	for(i=0,ni=m_links.size();i<ni;++i)
+	nbase = &m_nodes[0];
+	/* Refine links		*/
+	for (i = 0, ni = m_links.size(); i < ni; ++i)
 	{
-		Link&		feat=m_links[i];
-		const int	idx[]={	int(feat.m_n[0]-nbase),
-			int(feat.m_n[1]-nbase)};
-		if((idx[0]<ncount)&&(idx[1]<ncount))
+		Link& feat = m_links[i];
+		const int idx[] = {int(feat.m_n[0] - nbase),
+						   int(feat.m_n[1] - nbase)};
+		if ((idx[0] < ncount) && (idx[1] < ncount))
 		{
-			const int ni=edges(idx[0],idx[1]);
-			if(ni>0)
+			const int ni = edges(idx[0], idx[1]);
+			if (ni > 0)
 			{
 				appendLink(i);
-				Link*		pft[]={	&m_links[i],
-					&m_links[m_links.size()-1]};			
-				pft[0]->m_n[0]=&m_nodes[idx[0]];
-				pft[0]->m_n[1]=&m_nodes[ni];
-				pft[1]->m_n[0]=&m_nodes[ni];
-				pft[1]->m_n[1]=&m_nodes[idx[1]];
+				Link* pft[] = {&m_links[i],
+							   &m_links[m_links.size() - 1]};
+				pft[0]->m_n[0] = &m_nodes[idx[0]];
+				pft[0]->m_n[1] = &m_nodes[ni];
+				pft[1]->m_n[0] = &m_nodes[ni];
+				pft[1]->m_n[1] = &m_nodes[idx[1]];
 			}
 		}
 	}
-	/* Refine faces		*/ 
-	for(i=0;i<m_faces.size();++i)
+	/* Refine faces		*/
+	for (i = 0; i < m_faces.size(); ++i)
 	{
-		const Face&	feat=m_faces[i];
-		const int	idx[]={	int(feat.m_n[0]-nbase),
-			int(feat.m_n[1]-nbase),
-			int(feat.m_n[2]-nbase)};
-		for(j=2,k=0;k<3;j=k++)
+		const Face& feat = m_faces[i];
+		const int idx[] = {int(feat.m_n[0] - nbase),
+						   int(feat.m_n[1] - nbase),
+						   int(feat.m_n[2] - nbase)};
+		for (j = 2, k = 0; k < 3; j = k++)
 		{
-			if((idx[j]<ncount)&&(idx[k]<ncount))
+			if ((idx[j] < ncount) && (idx[k] < ncount))
 			{
-				const int ni=edges(idx[j],idx[k]);
-				if(ni>0)
+				const int ni = edges(idx[j], idx[k]);
+				if (ni > 0)
 				{
 					appendFace(i);
-					const int	l=(k+1)%3;
-					Face*		pft[]={	&m_faces[i],
-						&m_faces[m_faces.size()-1]};
-					pft[0]->m_n[0]=&m_nodes[idx[l]];
-					pft[0]->m_n[1]=&m_nodes[idx[j]];
-					pft[0]->m_n[2]=&m_nodes[ni];
-					pft[1]->m_n[0]=&m_nodes[ni];
-					pft[1]->m_n[1]=&m_nodes[idx[k]];
-					pft[1]->m_n[2]=&m_nodes[idx[l]];
-					appendLink(ni,idx[l],pft[0]->m_material);
-					--i;break;
+					const int l = (k + 1) % 3;
+					Face* pft[] = {&m_faces[i],
+								   &m_faces[m_faces.size() - 1]};
+					pft[0]->m_n[0] = &m_nodes[idx[l]];
+					pft[0]->m_n[1] = &m_nodes[idx[j]];
+					pft[0]->m_n[2] = &m_nodes[ni];
+					pft[1]->m_n[0] = &m_nodes[ni];
+					pft[1]->m_n[1] = &m_nodes[idx[k]];
+					pft[1]->m_n[2] = &m_nodes[idx[l]];
+					appendLink(ni, idx[l], pft[0]->m_material);
+					--i;
+					break;
 				}
 			}
 		}
 	}
-	/* Cut				*/ 
-	if(cut)
-	{	
-		btAlignedObjectArray<int>	cnodes;
-		const int					pcount=ncount;
-		int							i;
-		ncount=m_nodes.size();
-		cnodes.resize(ncount,0);
-		/* Nodes		*/ 
-		for(i=0;i<ncount;++i)
+	/* Cut				*/
+	if (cut)
+	{
+		btAlignedObjectArray<int> cnodes;
+		const int pcount = ncount;
+		int i;
+		ncount = m_nodes.size();
+		cnodes.resize(ncount, 0);
+		/* Nodes		*/
+		for (i = 0; i < ncount; ++i)
 		{
-			const btVector3	x=m_nodes[i].m_x;
-			if((i>=pcount)||(btFabs(ifn->Eval(x))<accurary))
+			const btVector3 x = m_nodes[i].m_x;
+			if ((i >= pcount) || (btFabs(ifn->Eval(x)) < accurary))
 			{
-				const btVector3	v=m_nodes[i].m_v;
-				btScalar		m=getMass(i);
-				if(m>0) { m*=0.5f;m_nodes[i].m_im/=0.5f; }
-				appendNode(x,m);
-				cnodes[i]=m_nodes.size()-1;
-				m_nodes[cnodes[i]].m_v=v;
+				const btVector3 v = m_nodes[i].m_v;
+				btScalar m = getMass(i);
+				if (m > 0)
+				{
+					m *= 0.5f;
+					m_nodes[i].m_im /= 0.5f;
+				}
+				appendNode(x, m);
+				cnodes[i] = m_nodes.size() - 1;
+				m_nodes[cnodes[i]].m_v = v;
 			}
 		}
-		nbase=&m_nodes[0];
-		/* Links		*/ 
-		for(i=0,ni=m_links.size();i<ni;++i)
+		nbase = &m_nodes[0];
+		/* Links		*/
+		for (i = 0, ni = m_links.size(); i < ni; ++i)
 		{
-			const int		id[]={	int(m_links[i].m_n[0]-nbase),
-				int(m_links[i].m_n[1]-nbase)};
-			int				todetach=0;
-			if(cnodes[id[0]]&&cnodes[id[1]])
+			const int id[] = {int(m_links[i].m_n[0] - nbase),
+							  int(m_links[i].m_n[1] - nbase)};
+			int todetach = 0;
+			if (cnodes[id[0]] && cnodes[id[1]])
 			{
 				appendLink(i);
-				todetach=m_links.size()-1;
+				todetach = m_links.size() - 1;
 			}
 			else
 			{
-				if((	(ifn->Eval(m_nodes[id[0]].m_x)<accurary)&&
-					(ifn->Eval(m_nodes[id[1]].m_x)<accurary)))
-					todetach=i;
+				if (((ifn->Eval(m_nodes[id[0]].m_x) < accurary) &&
+					 (ifn->Eval(m_nodes[id[1]].m_x) < accurary)))
+					todetach = i;
 			}
-			if(todetach)
+			if (todetach)
 			{
-				Link&	l=m_links[todetach];
-				for(int j=0;j<2;++j)
+				Link& l = m_links[todetach];
+				for (int j = 0; j < 2; ++j)
 				{
-					int cn=cnodes[int(l.m_n[j]-nbase)];
-					if(cn) l.m_n[j]=&m_nodes[cn];
-				}			
-			}
-		}
-		/* Faces		*/ 
-		for(i=0,ni=m_faces.size();i<ni;++i)
-		{
-			Node**			n=	m_faces[i].m_n;
-			if(	(ifn->Eval(n[0]->m_x)<accurary)&&
-				(ifn->Eval(n[1]->m_x)<accurary)&&
-				(ifn->Eval(n[2]->m_x)<accurary))
-			{
-				for(int j=0;j<3;++j)
-				{
-					int cn=cnodes[int(n[j]-nbase)];
-					if(cn) n[j]=&m_nodes[cn];
+					int cn = cnodes[int(l.m_n[j] - nbase)];
+					if (cn) l.m_n[j] = &m_nodes[cn];
 				}
 			}
 		}
-		/* Clean orphans	*/ 
-		int							nnodes=m_nodes.size();
-		btAlignedObjectArray<int>	ranks;
-		btAlignedObjectArray<int>	todelete;
-		ranks.resize(nnodes,0);
-		for(i=0,ni=m_links.size();i<ni;++i)
+		/* Faces		*/
+		for (i = 0, ni = m_faces.size(); i < ni; ++i)
 		{
-			for(int j=0;j<2;++j) ranks[int(m_links[i].m_n[j]-nbase)]++;
+			Node** n = m_faces[i].m_n;
+			if ((ifn->Eval(n[0]->m_x) < accurary) &&
+				(ifn->Eval(n[1]->m_x) < accurary) &&
+				(ifn->Eval(n[2]->m_x) < accurary))
+			{
+				for (int j = 0; j < 3; ++j)
+				{
+					int cn = cnodes[int(n[j] - nbase)];
+					if (cn) n[j] = &m_nodes[cn];
+				}
+			}
 		}
-		for(i=0,ni=m_faces.size();i<ni;++i)
+		/* Clean orphans	*/
+		int nnodes = m_nodes.size();
+		btAlignedObjectArray<int> ranks;
+		btAlignedObjectArray<int> todelete;
+		ranks.resize(nnodes, 0);
+		for (i = 0, ni = m_links.size(); i < ni; ++i)
 		{
-			for(int j=0;j<3;++j) ranks[int(m_faces[i].m_n[j]-nbase)]++;
+			for (int j = 0; j < 2; ++j) ranks[int(m_links[i].m_n[j] - nbase)]++;
 		}
-		for(i=0;i<m_links.size();++i)
+		for (i = 0, ni = m_faces.size(); i < ni; ++i)
 		{
-			const int	id[]={	int(m_links[i].m_n[0]-nbase),
-				int(m_links[i].m_n[1]-nbase)};
-			const bool	sg[]={	ranks[id[0]]==1,
-				ranks[id[1]]==1};
-			if(sg[0]||sg[1])
+			for (int j = 0; j < 3; ++j) ranks[int(m_faces[i].m_n[j] - nbase)]++;
+		}
+		for (i = 0; i < m_links.size(); ++i)
+		{
+			const int id[] = {int(m_links[i].m_n[0] - nbase),
+							  int(m_links[i].m_n[1] - nbase)};
+			const bool sg[] = {ranks[id[0]] == 1,
+							   ranks[id[1]] == 1};
+			if (sg[0] || sg[1])
 			{
 				--ranks[id[0]];
 				--ranks[id[1]];
-				btSwap(m_links[i],m_links[m_links.size()-1]);
-				m_links.pop_back();--i;
+				btSwap(m_links[i], m_links[m_links.size() - 1]);
+				m_links.pop_back();
+				--i;
 			}
 		}
 #if 0	
@@ -1631,679 +1651,675 @@ void			btSoftBody::refine(ImplicitFn* ifn,btScalar accurary,bool cut)
 		}
 #endif
 	}
-	m_bUpdateRtCst=true;
+	m_bUpdateRtCst = true;
 }
 
 //
-bool			btSoftBody::cutLink(const Node* node0,const Node* node1,btScalar position)
+bool btSoftBody::cutLink(const Node* node0, const Node* node1, btScalar position)
 {
-	return(cutLink(int(node0-&m_nodes[0]),int(node1-&m_nodes[0]),position));
+	return (cutLink(int(node0 - &m_nodes[0]), int(node1 - &m_nodes[0]), position));
 }
 
 //
-bool			btSoftBody::cutLink(int node0,int node1,btScalar position)
+bool btSoftBody::cutLink(int node0, int node1, btScalar position)
 {
-	bool			done=false;
-	int i,ni;
-//	const btVector3	d=m_nodes[node0].m_x-m_nodes[node1].m_x;
-	const btVector3	x=Lerp(m_nodes[node0].m_x,m_nodes[node1].m_x,position);
-	const btVector3	v=Lerp(m_nodes[node0].m_v,m_nodes[node1].m_v,position);
-	const btScalar	m=1;
-	appendNode(x,m);
-	appendNode(x,m);
-	Node*			pa=&m_nodes[node0];
-	Node*			pb=&m_nodes[node1];
-	Node*			pn[2]={	&m_nodes[m_nodes.size()-2],
-		&m_nodes[m_nodes.size()-1]};
-	pn[0]->m_v=v;
-	pn[1]->m_v=v;
-	for(i=0,ni=m_links.size();i<ni;++i)
+	bool done = false;
+	int i, ni;
+	//	const btVector3	d=m_nodes[node0].m_x-m_nodes[node1].m_x;
+	const btVector3 x = Lerp(m_nodes[node0].m_x, m_nodes[node1].m_x, position);
+	const btVector3 v = Lerp(m_nodes[node0].m_v, m_nodes[node1].m_v, position);
+	const btScalar m = 1;
+	appendNode(x, m);
+	appendNode(x, m);
+	Node* pa = &m_nodes[node0];
+	Node* pb = &m_nodes[node1];
+	Node* pn[2] = {&m_nodes[m_nodes.size() - 2],
+				   &m_nodes[m_nodes.size() - 1]};
+	pn[0]->m_v = v;
+	pn[1]->m_v = v;
+	for (i = 0, ni = m_links.size(); i < ni; ++i)
 	{
-		const int mtch=MatchEdge(m_links[i].m_n[0],m_links[i].m_n[1],pa,pb);
-		if(mtch!=-1)
+		const int mtch = MatchEdge(m_links[i].m_n[0], m_links[i].m_n[1], pa, pb);
+		if (mtch != -1)
 		{
 			appendLink(i);
-			Link*	pft[]={&m_links[i],&m_links[m_links.size()-1]};
-			pft[0]->m_n[1]=pn[mtch];
-			pft[1]->m_n[0]=pn[1-mtch];
-			done=true;
+			Link* pft[] = {&m_links[i], &m_links[m_links.size() - 1]};
+			pft[0]->m_n[1] = pn[mtch];
+			pft[1]->m_n[0] = pn[1 - mtch];
+			done = true;
 		}
 	}
-	for(i=0,ni=m_faces.size();i<ni;++i)
+	for (i = 0, ni = m_faces.size(); i < ni; ++i)
 	{
-		for(int k=2,l=0;l<3;k=l++)
+		for (int k = 2, l = 0; l < 3; k = l++)
 		{
-			const int mtch=MatchEdge(m_faces[i].m_n[k],m_faces[i].m_n[l],pa,pb);
-			if(mtch!=-1)
+			const int mtch = MatchEdge(m_faces[i].m_n[k], m_faces[i].m_n[l], pa, pb);
+			if (mtch != -1)
 			{
 				appendFace(i);
-				Face*	pft[]={&m_faces[i],&m_faces[m_faces.size()-1]};
-				pft[0]->m_n[l]=pn[mtch];
-				pft[1]->m_n[k]=pn[1-mtch];
-				appendLink(pn[0],pft[0]->m_n[(l+1)%3],pft[0]->m_material,true);
-				appendLink(pn[1],pft[0]->m_n[(l+1)%3],pft[0]->m_material,true);
+				Face* pft[] = {&m_faces[i], &m_faces[m_faces.size() - 1]};
+				pft[0]->m_n[l] = pn[mtch];
+				pft[1]->m_n[k] = pn[1 - mtch];
+				appendLink(pn[0], pft[0]->m_n[(l + 1) % 3], pft[0]->m_material, true);
+				appendLink(pn[1], pft[0]->m_n[(l + 1) % 3], pft[0]->m_material, true);
 			}
 		}
 	}
-	if(!done)
+	if (!done)
 	{
 		m_ndbvt.remove(pn[0]->m_leaf);
 		m_ndbvt.remove(pn[1]->m_leaf);
 		m_nodes.pop_back();
 		m_nodes.pop_back();
 	}
-	return(done);
+	return (done);
 }
 
 //
-bool			btSoftBody::rayTest(const btVector3& rayFrom,
-									const btVector3& rayTo,
-									sRayCast& results)
+bool btSoftBody::rayTest(const btVector3& rayFrom,
+						 const btVector3& rayTo,
+						 sRayCast& results)
 {
-	if(m_faces.size()&&m_fdbvt.empty()) 
+	if (m_faces.size() && m_fdbvt.empty())
 		initializeFaceTree();
 
-	results.body	=	this;
+	results.body = this;
 	results.fraction = 1.f;
-	results.feature	=	eFeature::None;
-	results.index	=	-1;
+	results.feature = eFeature::None;
+	results.index = -1;
 
-	return(rayTest(rayFrom,rayTo,results.fraction,results.feature,results.index,false)!=0);
+	return (rayTest(rayFrom, rayTo, results.fraction, results.feature, results.index, false) != 0);
 }
 
 //
-void			btSoftBody::setSolver(eSolverPresets::_ preset)
+void btSoftBody::setSolver(eSolverPresets::_ preset)
 {
 	m_cfg.m_vsequence.clear();
 	m_cfg.m_psequence.clear();
 	m_cfg.m_dsequence.clear();
-	switch(preset)
+	switch (preset)
 	{
-	case	eSolverPresets::Positions:
-		m_cfg.m_psequence.push_back(ePSolver::Anchors);
-		m_cfg.m_psequence.push_back(ePSolver::RContacts);
-		m_cfg.m_psequence.push_back(ePSolver::SContacts);
-		m_cfg.m_psequence.push_back(ePSolver::Linear);	
-		break;	
-	case	eSolverPresets::Velocities:
-		m_cfg.m_vsequence.push_back(eVSolver::Linear);
+		case eSolverPresets::Positions:
+			m_cfg.m_psequence.push_back(ePSolver::Anchors);
+			m_cfg.m_psequence.push_back(ePSolver::RContacts);
+			m_cfg.m_psequence.push_back(ePSolver::SContacts);
+			m_cfg.m_psequence.push_back(ePSolver::Linear);
+			break;
+		case eSolverPresets::Velocities:
+			m_cfg.m_vsequence.push_back(eVSolver::Linear);
 
-		m_cfg.m_psequence.push_back(ePSolver::Anchors);
-		m_cfg.m_psequence.push_back(ePSolver::RContacts);
-		m_cfg.m_psequence.push_back(ePSolver::SContacts);
+			m_cfg.m_psequence.push_back(ePSolver::Anchors);
+			m_cfg.m_psequence.push_back(ePSolver::RContacts);
+			m_cfg.m_psequence.push_back(ePSolver::SContacts);
 
-		m_cfg.m_dsequence.push_back(ePSolver::Linear);
-		break;
+			m_cfg.m_dsequence.push_back(ePSolver::Linear);
+			break;
 	}
 }
 
 //
-void			btSoftBody::predictMotion(btScalar dt)
+void btSoftBody::predictMotion(btScalar dt)
 {
+	int i, ni;
 
-	int i,ni;
-
-	/* Update				*/ 
-	if(m_bUpdateRtCst)
+	/* Update				*/
+	if (m_bUpdateRtCst)
 	{
-		m_bUpdateRtCst=false;
+		m_bUpdateRtCst = false;
 		updateConstants();
 		m_fdbvt.clear();
-		if(m_cfg.collisions&fCollision::VF_SS)
+		if (m_cfg.collisions & fCollision::VF_SS)
 		{
-			initializeFaceTree();			
+			initializeFaceTree();
 		}
 	}
 
-	/* Prepare				*/ 
-	m_sst.sdt		=	dt*m_cfg.timescale;
-	m_sst.isdt		=	1/m_sst.sdt;
-	m_sst.velmrg	=	m_sst.sdt*3;
-	m_sst.radmrg	=	getCollisionShape()->getMargin();
-	m_sst.updmrg	=	m_sst.radmrg*(btScalar)0.25;
-	/* Forces				*/ 
-	addVelocity(m_worldInfo->m_gravity*m_sst.sdt);
+	/* Prepare				*/
+	m_sst.sdt = dt * m_cfg.timescale;
+	m_sst.isdt = 1 / m_sst.sdt;
+	m_sst.velmrg = m_sst.sdt * 3;
+	m_sst.radmrg = getCollisionShape()->getMargin();
+	m_sst.updmrg = m_sst.radmrg * (btScalar)0.25;
+	/* Forces				*/
+	addVelocity(m_worldInfo->m_gravity * m_sst.sdt);
 	applyForces();
-	/* Integrate			*/ 
-	for(i=0,ni=m_nodes.size();i<ni;++i)
+	/* Integrate			*/
+	for (i = 0, ni = m_nodes.size(); i < ni; ++i)
 	{
-		Node&	n=m_nodes[i];
-		n.m_q	=	n.m_x;
-		btVector3 deltaV = n.m_f*n.m_im*m_sst.sdt;
+		Node& n = m_nodes[i];
+		n.m_q = n.m_x;
+		btVector3 deltaV = n.m_f * n.m_im * m_sst.sdt;
 		{
 			btScalar maxDisplacement = m_worldInfo->m_maxDisplacement;
-			btScalar clampDeltaV = maxDisplacement/m_sst.sdt;
-			for (int c=0;c<3;c++)
+			btScalar clampDeltaV = maxDisplacement / m_sst.sdt;
+			for (int c = 0; c < 3; c++)
 			{
-				if (deltaV[c]>clampDeltaV)
+				if (deltaV[c] > clampDeltaV)
 				{
 					deltaV[c] = clampDeltaV;
 				}
-				if (deltaV[c]<-clampDeltaV)
+				if (deltaV[c] < -clampDeltaV)
 				{
-					deltaV[c]=-clampDeltaV;
+					deltaV[c] = -clampDeltaV;
 				}
 			}
 		}
-		n.m_v	+=	deltaV;
-		n.m_x	+=	n.m_v*m_sst.sdt;
-		n.m_f	=	btVector3(0,0,0);
+		n.m_v += deltaV;
+		n.m_x += n.m_v * m_sst.sdt;
+		n.m_f = btVector3(0, 0, 0);
 	}
-	/* Clusters				*/ 
+	/* Clusters				*/
 	updateClusters();
-	/* Bounds				*/ 
-	updateBounds();	
-	/* Nodes				*/ 
-	ATTRIBUTE_ALIGNED16(btDbvtVolume)	vol;
-	for(i=0,ni=m_nodes.size();i<ni;++i)
+	/* Bounds				*/
+	updateBounds();
+	/* Nodes				*/
+	ATTRIBUTE_ALIGNED16(btDbvtVolume)
+	vol;
+	for (i = 0, ni = m_nodes.size(); i < ni; ++i)
 	{
-		Node&	n=m_nodes[i];
-		vol = btDbvtVolume::FromCR(n.m_x,m_sst.radmrg);
-		m_ndbvt.update(	n.m_leaf,
-			vol,
-			n.m_v*m_sst.velmrg,
-			m_sst.updmrg);
+		Node& n = m_nodes[i];
+		vol = btDbvtVolume::FromCR(n.m_x, m_sst.radmrg);
+		m_ndbvt.update(n.m_leaf,
+					   vol,
+					   n.m_v * m_sst.velmrg,
+					   m_sst.updmrg);
 	}
-	/* Faces				*/ 
-	if(!m_fdbvt.empty())
+	/* Faces				*/
+	if (!m_fdbvt.empty())
 	{
-		for(int i=0;i<m_faces.size();++i)
+		for (int i = 0; i < m_faces.size(); ++i)
 		{
-			Face&			f=m_faces[i];
-			const btVector3	v=(	f.m_n[0]->m_v+
-				f.m_n[1]->m_v+
-				f.m_n[2]->m_v)/3;
-			vol = VolumeOf(f,m_sst.radmrg);
-			m_fdbvt.update(	f.m_leaf,
-				vol,
-				v*m_sst.velmrg,
-				m_sst.updmrg);
+			Face& f = m_faces[i];
+			const btVector3 v = (f.m_n[0]->m_v +
+								 f.m_n[1]->m_v +
+								 f.m_n[2]->m_v) /
+								3;
+			vol = VolumeOf(f, m_sst.radmrg);
+			m_fdbvt.update(f.m_leaf,
+						   vol,
+						   v * m_sst.velmrg,
+						   m_sst.updmrg);
 		}
 	}
-	/* Pose					*/ 
+	/* Pose					*/
 	updatePose();
-	/* Match				*/ 
-	if(m_pose.m_bframe&&(m_cfg.kMT>0))
+	/* Match				*/
+	if (m_pose.m_bframe && (m_cfg.kMT > 0))
 	{
-		const btMatrix3x3	posetrs=m_pose.m_rot;
-		for(int i=0,ni=m_nodes.size();i<ni;++i)
+		const btMatrix3x3 posetrs = m_pose.m_rot;
+		for (int i = 0, ni = m_nodes.size(); i < ni; ++i)
 		{
-			Node&	n=m_nodes[i];
-			if(n.m_im>0)
+			Node& n = m_nodes[i];
+			if (n.m_im > 0)
 			{
-				const btVector3	x=posetrs*m_pose.m_pos[i]+m_pose.m_com;
-				n.m_x=Lerp(n.m_x,x,m_cfg.kMT);
+				const btVector3 x = posetrs * m_pose.m_pos[i] + m_pose.m_com;
+				n.m_x = Lerp(n.m_x, x, m_cfg.kMT);
 			}
 		}
 	}
-	/* Clear contacts		*/ 
+	/* Clear contacts		*/
 	m_rcontacts.resize(0);
 	m_scontacts.resize(0);
-	/* Optimize dbvt's		*/ 
+	/* Optimize dbvt's		*/
 	m_ndbvt.optimizeIncremental(1);
 	m_fdbvt.optimizeIncremental(1);
 	m_cdbvt.optimizeIncremental(1);
 }
 
 //
-void			btSoftBody::solveConstraints()
+void btSoftBody::solveConstraints()
 {
-
-	/* Apply clusters		*/ 
+	/* Apply clusters		*/
 	applyClusters(false);
-	/* Prepare links		*/ 
+	/* Prepare links		*/
 
-	int i,ni;
+	int i, ni;
 
-	for(i=0,ni=m_links.size();i<ni;++i)
+	for (i = 0, ni = m_links.size(); i < ni; ++i)
 	{
-		Link&	l=m_links[i];
-		l.m_c3		=	l.m_n[1]->m_q-l.m_n[0]->m_q;
-		l.m_c2		=	1/(l.m_c3.length2()*l.m_c0);
+		Link& l = m_links[i];
+		l.m_c3 = l.m_n[1]->m_q - l.m_n[0]->m_q;
+		l.m_c2 = 1 / (l.m_c3.length2() * l.m_c0);
 	}
-	/* Prepare anchors		*/ 
-	for(i=0,ni=m_anchors.size();i<ni;++i)
+	/* Prepare anchors		*/
+	for (i = 0, ni = m_anchors.size(); i < ni; ++i)
 	{
-		Anchor&			a=m_anchors[i];
-		const btVector3	ra=a.m_body->getWorldTransform().getBasis()*a.m_local;
-		a.m_c0	=	ImpulseMatrix(	m_sst.sdt,
-			a.m_node->m_im,
-			a.m_body->getInvMass(),
-			a.m_body->getInvInertiaTensorWorld(),
-			ra);
-		a.m_c1	=	ra;
-		a.m_c2	=	m_sst.sdt*a.m_node->m_im;
+		Anchor& a = m_anchors[i];
+		const btVector3 ra = a.m_body->getWorldTransform().getBasis() * a.m_local;
+		a.m_c0 = ImpulseMatrix(m_sst.sdt,
+							   a.m_node->m_im,
+							   a.m_body->getInvMass(),
+							   a.m_body->getInvInertiaTensorWorld(),
+							   ra);
+		a.m_c1 = ra;
+		a.m_c2 = m_sst.sdt * a.m_node->m_im;
 		a.m_body->activate();
 	}
-	/* Solve velocities		*/ 
-	if(m_cfg.viterations>0)
+	/* Solve velocities		*/
+	if (m_cfg.viterations > 0)
 	{
-		/* Solve			*/ 
-		for(int isolve=0;isolve<m_cfg.viterations;++isolve)
+		/* Solve			*/
+		for (int isolve = 0; isolve < m_cfg.viterations; ++isolve)
 		{
-			for(int iseq=0;iseq<m_cfg.m_vsequence.size();++iseq)
+			for (int iseq = 0; iseq < m_cfg.m_vsequence.size(); ++iseq)
 			{
-				getSolver(m_cfg.m_vsequence[iseq])(this,1);
+				getSolver(m_cfg.m_vsequence[iseq])(this, 1);
 			}
 		}
-		/* Update			*/ 
-		for(i=0,ni=m_nodes.size();i<ni;++i)
+		/* Update			*/
+		for (i = 0, ni = m_nodes.size(); i < ni; ++i)
 		{
-			Node&	n=m_nodes[i];
-			n.m_x	=	n.m_q+n.m_v*m_sst.sdt;
+			Node& n = m_nodes[i];
+			n.m_x = n.m_q + n.m_v * m_sst.sdt;
 		}
 	}
-	/* Solve positions		*/ 
-	if(m_cfg.piterations>0)
+	/* Solve positions		*/
+	if (m_cfg.piterations > 0)
 	{
-		for(int isolve=0;isolve<m_cfg.piterations;++isolve)
+		for (int isolve = 0; isolve < m_cfg.piterations; ++isolve)
 		{
-			const btScalar ti=isolve/(btScalar)m_cfg.piterations;
-			for(int iseq=0;iseq<m_cfg.m_psequence.size();++iseq)
+			const btScalar ti = isolve / (btScalar)m_cfg.piterations;
+			for (int iseq = 0; iseq < m_cfg.m_psequence.size(); ++iseq)
 			{
-				getSolver(m_cfg.m_psequence[iseq])(this,1,ti);
+				getSolver(m_cfg.m_psequence[iseq])(this, 1, ti);
 			}
 		}
-		const btScalar	vc=m_sst.isdt*(1-m_cfg.kDP);
-		for(i=0,ni=m_nodes.size();i<ni;++i)
+		const btScalar vc = m_sst.isdt * (1 - m_cfg.kDP);
+		for (i = 0, ni = m_nodes.size(); i < ni; ++i)
 		{
-			Node&	n=m_nodes[i];
-			n.m_v	=	(n.m_x-n.m_q)*vc;
-			n.m_f	=	btVector3(0,0,0);		
+			Node& n = m_nodes[i];
+			n.m_v = (n.m_x - n.m_q) * vc;
+			n.m_f = btVector3(0, 0, 0);
 		}
 	}
-	/* Solve drift			*/ 
-	if(m_cfg.diterations>0)
+	/* Solve drift			*/
+	if (m_cfg.diterations > 0)
 	{
-		const btScalar	vcf=m_cfg.kVCF*m_sst.isdt;
-		for(i=0,ni=m_nodes.size();i<ni;++i)
+		const btScalar vcf = m_cfg.kVCF * m_sst.isdt;
+		for (i = 0, ni = m_nodes.size(); i < ni; ++i)
 		{
-			Node&	n=m_nodes[i];
-			n.m_q	=	n.m_x;
+			Node& n = m_nodes[i];
+			n.m_q = n.m_x;
 		}
-		for(int idrift=0;idrift<m_cfg.diterations;++idrift)
+		for (int idrift = 0; idrift < m_cfg.diterations; ++idrift)
 		{
-			for(int iseq=0;iseq<m_cfg.m_dsequence.size();++iseq)
+			for (int iseq = 0; iseq < m_cfg.m_dsequence.size(); ++iseq)
 			{
-				getSolver(m_cfg.m_dsequence[iseq])(this,1,0);
+				getSolver(m_cfg.m_dsequence[iseq])(this, 1, 0);
 			}
 		}
-		for(int i=0,ni=m_nodes.size();i<ni;++i)
+		for (int i = 0, ni = m_nodes.size(); i < ni; ++i)
 		{
-			Node&	n=m_nodes[i];
-			n.m_v	+=	(n.m_x-n.m_q)*vcf;
+			Node& n = m_nodes[i];
+			n.m_v += (n.m_x - n.m_q) * vcf;
 		}
 	}
-	/* Apply clusters		*/ 
+	/* Apply clusters		*/
 	dampClusters();
 	applyClusters(true);
 }
 
 //
-void			btSoftBody::staticSolve(int iterations)
+void btSoftBody::staticSolve(int iterations)
 {
-	for(int isolve=0;isolve<iterations;++isolve)
+	for (int isolve = 0; isolve < iterations; ++isolve)
 	{
-		for(int iseq=0;iseq<m_cfg.m_psequence.size();++iseq)
+		for (int iseq = 0; iseq < m_cfg.m_psequence.size(); ++iseq)
 		{
-			getSolver(m_cfg.m_psequence[iseq])(this,1,0);
+			getSolver(m_cfg.m_psequence[iseq])(this, 1, 0);
 		}
 	}
 }
 
 //
-void			btSoftBody::solveCommonConstraints(btSoftBody** /*bodies*/,int /*count*/,int /*iterations*/)
+void btSoftBody::solveCommonConstraints(btSoftBody** /*bodies*/, int /*count*/, int /*iterations*/)
 {
 	/// placeholder
 }
 
 //
-void			btSoftBody::solveClusters(const btAlignedObjectArray<btSoftBody*>& bodies)
+void btSoftBody::solveClusters(const btAlignedObjectArray<btSoftBody*>& bodies)
 {
-	const int	nb=bodies.size();
-	int			iterations=0;
+	const int nb = bodies.size();
+	int iterations = 0;
 	int i;
 
-	for(i=0;i<nb;++i)
+	for (i = 0; i < nb; ++i)
 	{
-		iterations=btMax(iterations,bodies[i]->m_cfg.citerations);
+		iterations = btMax(iterations, bodies[i]->m_cfg.citerations);
 	}
-	for(i=0;i<nb;++i)
+	for (i = 0; i < nb; ++i)
 	{
 		bodies[i]->prepareClusters(iterations);
 	}
-	for(i=0;i<iterations;++i)
+	for (i = 0; i < iterations; ++i)
 	{
-		const btScalar sor=1;
-		for(int j=0;j<nb;++j)
+		const btScalar sor = 1;
+		for (int j = 0; j < nb; ++j)
 		{
 			bodies[j]->solveClusters(sor);
 		}
 	}
-	for(i=0;i<nb;++i)
+	for (i = 0; i < nb; ++i)
 	{
 		bodies[i]->cleanupClusters();
 	}
 }
 
 //
-void			btSoftBody::integrateMotion()
+void btSoftBody::integrateMotion()
 {
-	/* Update			*/ 
+	/* Update			*/
 	updateNormals();
 }
 
 //
-btSoftBody::RayFromToCaster::RayFromToCaster(const btVector3& rayFrom,const btVector3& rayTo,btScalar mxt)
+btSoftBody::RayFromToCaster::RayFromToCaster(const btVector3& rayFrom, const btVector3& rayTo, btScalar mxt)
 {
 	m_rayFrom = rayFrom;
-	m_rayNormalizedDirection = (rayTo-rayFrom);
+	m_rayNormalizedDirection = (rayTo - rayFrom);
 	m_rayTo = rayTo;
-	m_mint	=	mxt;
-	m_face	=	0;
-	m_tests	=	0;
+	m_mint = mxt;
+	m_face = 0;
+	m_tests = 0;
 }
 
 //
-void				btSoftBody::RayFromToCaster::Process(const btDbvtNode* leaf)
+void btSoftBody::RayFromToCaster::Process(const btDbvtNode* leaf)
 {
-	btSoftBody::Face&	f=*(btSoftBody::Face*)leaf->data;
-	const btScalar		t=rayFromToTriangle(	m_rayFrom,m_rayTo,m_rayNormalizedDirection,
-		f.m_n[0]->m_x,
-		f.m_n[1]->m_x,
-		f.m_n[2]->m_x,
-		m_mint);
-	if((t>0)&&(t<m_mint)) 
-	{ 
-		m_mint=t;m_face=&f; 
+	btSoftBody::Face& f = *(btSoftBody::Face*)leaf->data;
+	const btScalar t = rayFromToTriangle(m_rayFrom, m_rayTo, m_rayNormalizedDirection,
+										 f.m_n[0]->m_x,
+										 f.m_n[1]->m_x,
+										 f.m_n[2]->m_x,
+										 m_mint);
+	if ((t > 0) && (t < m_mint))
+	{
+		m_mint = t;
+		m_face = &f;
 	}
 	++m_tests;
 }
 
 //
-btScalar			btSoftBody::RayFromToCaster::rayFromToTriangle(	const btVector3& rayFrom,
-																   const btVector3& rayTo,
-																   const btVector3& rayNormalizedDirection,
-																   const btVector3& a,
-																   const btVector3& b,
-																   const btVector3& c,
-																   btScalar maxt)
+btScalar btSoftBody::RayFromToCaster::rayFromToTriangle(const btVector3& rayFrom,
+														const btVector3& rayTo,
+														const btVector3& rayNormalizedDirection,
+														const btVector3& a,
+														const btVector3& b,
+														const btVector3& c,
+														btScalar maxt)
 {
-	static const btScalar	ceps=-SIMD_EPSILON*10;
-	static const btScalar	teps=SIMD_EPSILON*10;
+	static const btScalar ceps = -SIMD_EPSILON * 10;
+	static const btScalar teps = SIMD_EPSILON * 10;
 
-	const btVector3			n=btCross(b-a,c-a);
-	const btScalar			d=btDot(a,n);
-	const btScalar			den=btDot(rayNormalizedDirection,n);
-	if(!btFuzzyZero(den))
+	const btVector3 n = btCross(b - a, c - a);
+	const btScalar d = btDot(a, n);
+	const btScalar den = btDot(rayNormalizedDirection, n);
+	if (!btFuzzyZero(den))
 	{
-		const btScalar		num=btDot(rayFrom,n)-d;
-		const btScalar		t=-num/den;
-		if((t>teps)&&(t<maxt))
+		const btScalar num = btDot(rayFrom, n) - d;
+		const btScalar t = -num / den;
+		if ((t > teps) && (t < maxt))
 		{
-			const btVector3	hit=rayFrom+rayNormalizedDirection*t;
-			if(	(btDot(n,btCross(a-hit,b-hit))>ceps)	&&			
-				(btDot(n,btCross(b-hit,c-hit))>ceps)	&&
-				(btDot(n,btCross(c-hit,a-hit))>ceps))
+			const btVector3 hit = rayFrom + rayNormalizedDirection * t;
+			if ((btDot(n, btCross(a - hit, b - hit)) > ceps) &&
+				(btDot(n, btCross(b - hit, c - hit)) > ceps) &&
+				(btDot(n, btCross(c - hit, a - hit)) > ceps))
 			{
-				return(t);
+				return (t);
 			}
 		}
 	}
-	return(-1);
+	return (-1);
 }
 
 //
-void				btSoftBody::pointersToIndices()
+void btSoftBody::pointersToIndices()
 {
-#define	PTR2IDX(_p_,_b_)	reinterpret_cast<btSoftBody::Node*>((_p_)-(_b_))
-	btSoftBody::Node*	base=m_nodes.size() ? &m_nodes[0] : 0;
-	int i,ni;
+#define PTR2IDX(_p_, _b_) reinterpret_cast<btSoftBody::Node*>((_p_) - (_b_))
+	btSoftBody::Node* base = m_nodes.size() ? &m_nodes[0] : 0;
+	int i, ni;
 
-	for(i=0,ni=m_nodes.size();i<ni;++i)
+	for (i = 0, ni = m_nodes.size(); i < ni; ++i)
 	{
-		if(m_nodes[i].m_leaf)
+		if (m_nodes[i].m_leaf)
 		{
-			m_nodes[i].m_leaf->data=*(void**)&i;
+			m_nodes[i].m_leaf->data = *(void**)&i;
 		}
 	}
-	for(i=0,ni=m_links.size();i<ni;++i)
+	for (i = 0, ni = m_links.size(); i < ni; ++i)
 	{
-		m_links[i].m_n[0]=PTR2IDX(m_links[i].m_n[0],base);
-		m_links[i].m_n[1]=PTR2IDX(m_links[i].m_n[1],base);
+		m_links[i].m_n[0] = PTR2IDX(m_links[i].m_n[0], base);
+		m_links[i].m_n[1] = PTR2IDX(m_links[i].m_n[1], base);
 	}
-	for(i=0,ni=m_faces.size();i<ni;++i)
+	for (i = 0, ni = m_faces.size(); i < ni; ++i)
 	{
-		m_faces[i].m_n[0]=PTR2IDX(m_faces[i].m_n[0],base);
-		m_faces[i].m_n[1]=PTR2IDX(m_faces[i].m_n[1],base);
-		m_faces[i].m_n[2]=PTR2IDX(m_faces[i].m_n[2],base);
-		if(m_faces[i].m_leaf)
+		m_faces[i].m_n[0] = PTR2IDX(m_faces[i].m_n[0], base);
+		m_faces[i].m_n[1] = PTR2IDX(m_faces[i].m_n[1], base);
+		m_faces[i].m_n[2] = PTR2IDX(m_faces[i].m_n[2], base);
+		if (m_faces[i].m_leaf)
 		{
-			m_faces[i].m_leaf->data=*(void**)&i;
+			m_faces[i].m_leaf->data = *(void**)&i;
 		}
 	}
-	for(i=0,ni=m_anchors.size();i<ni;++i)
+	for (i = 0, ni = m_anchors.size(); i < ni; ++i)
 	{
-		m_anchors[i].m_node=PTR2IDX(m_anchors[i].m_node,base);
+		m_anchors[i].m_node = PTR2IDX(m_anchors[i].m_node, base);
 	}
-	for(i=0,ni=m_notes.size();i<ni;++i)
+	for (i = 0, ni = m_notes.size(); i < ni; ++i)
 	{
-		for(int j=0;j<m_notes[i].m_rank;++j)
+		for (int j = 0; j < m_notes[i].m_rank; ++j)
 		{
-			m_notes[i].m_nodes[j]=PTR2IDX(m_notes[i].m_nodes[j],base);
+			m_notes[i].m_nodes[j] = PTR2IDX(m_notes[i].m_nodes[j], base);
 		}
 	}
-#undef	PTR2IDX
+#undef PTR2IDX
 }
 
 //
-void				btSoftBody::indicesToPointers(const int* map)
+void btSoftBody::indicesToPointers(const int* map)
 {
-#define	IDX2PTR(_p_,_b_)	map?(&(_b_)[map[(((char*)_p_)-(char*)0)]]):	\
-	(&(_b_)[(((char*)_p_)-(char*)0)])
-	btSoftBody::Node*	base=m_nodes.size() ? &m_nodes[0]:0;
-	int i,ni;
+#define IDX2PTR(_p_, _b_) map ? (&(_b_)[map[(((char*)_p_) - (char*)0)]]) : (&(_b_)[(((char*)_p_) - (char*)0)])
+	btSoftBody::Node* base = m_nodes.size() ? &m_nodes[0] : 0;
+	int i, ni;
 
-	for(i=0,ni=m_nodes.size();i<ni;++i)
+	for (i = 0, ni = m_nodes.size(); i < ni; ++i)
 	{
-		if(m_nodes[i].m_leaf)
+		if (m_nodes[i].m_leaf)
 		{
-			m_nodes[i].m_leaf->data=&m_nodes[i];
+			m_nodes[i].m_leaf->data = &m_nodes[i];
 		}
 	}
-	for(i=0,ni=m_links.size();i<ni;++i)
+	for (i = 0, ni = m_links.size(); i < ni; ++i)
 	{
-		m_links[i].m_n[0]=IDX2PTR(m_links[i].m_n[0],base);
-		m_links[i].m_n[1]=IDX2PTR(m_links[i].m_n[1],base);
+		m_links[i].m_n[0] = IDX2PTR(m_links[i].m_n[0], base);
+		m_links[i].m_n[1] = IDX2PTR(m_links[i].m_n[1], base);
 	}
-	for(i=0,ni=m_faces.size();i<ni;++i)
+	for (i = 0, ni = m_faces.size(); i < ni; ++i)
 	{
-		m_faces[i].m_n[0]=IDX2PTR(m_faces[i].m_n[0],base);
-		m_faces[i].m_n[1]=IDX2PTR(m_faces[i].m_n[1],base);
-		m_faces[i].m_n[2]=IDX2PTR(m_faces[i].m_n[2],base);
-		if(m_faces[i].m_leaf)
+		m_faces[i].m_n[0] = IDX2PTR(m_faces[i].m_n[0], base);
+		m_faces[i].m_n[1] = IDX2PTR(m_faces[i].m_n[1], base);
+		m_faces[i].m_n[2] = IDX2PTR(m_faces[i].m_n[2], base);
+		if (m_faces[i].m_leaf)
 		{
-			m_faces[i].m_leaf->data=&m_faces[i];
+			m_faces[i].m_leaf->data = &m_faces[i];
 		}
 	}
-	for(i=0,ni=m_anchors.size();i<ni;++i)
+	for (i = 0, ni = m_anchors.size(); i < ni; ++i)
 	{
-		m_anchors[i].m_node=IDX2PTR(m_anchors[i].m_node,base);
+		m_anchors[i].m_node = IDX2PTR(m_anchors[i].m_node, base);
 	}
-	for(i=0,ni=m_notes.size();i<ni;++i)
+	for (i = 0, ni = m_notes.size(); i < ni; ++i)
 	{
-		for(int j=0;j<m_notes[i].m_rank;++j)
+		for (int j = 0; j < m_notes[i].m_rank; ++j)
 		{
-			m_notes[i].m_nodes[j]=IDX2PTR(m_notes[i].m_nodes[j],base);
+			m_notes[i].m_nodes[j] = IDX2PTR(m_notes[i].m_nodes[j], base);
 		}
 	}
-#undef	IDX2PTR
+#undef IDX2PTR
 }
 
 //
-int					btSoftBody::rayTest(const btVector3& rayFrom,const btVector3& rayTo,
-										btScalar& mint,eFeature::_& feature,int& index,bool bcountonly) const
+int btSoftBody::rayTest(const btVector3& rayFrom, const btVector3& rayTo,
+						btScalar& mint, eFeature::_& feature, int& index, bool bcountonly) const
 {
-	int	cnt=0;
-	btVector3 dir = rayTo-rayFrom;
-	
+	int cnt = 0;
+	btVector3 dir = rayTo - rayFrom;
 
-	if(bcountonly||m_fdbvt.empty())
-	{/* Full search	*/ 
-		
-		for(int i=0,ni=m_faces.size();i<ni;++i)
+	if (bcountonly || m_fdbvt.empty())
+	{ /* Full search	*/
+
+		for (int i = 0, ni = m_faces.size(); i < ni; ++i)
 		{
-			const btSoftBody::Face&	f=m_faces[i];
+			const btSoftBody::Face& f = m_faces[i];
 
-			const btScalar			t=RayFromToCaster::rayFromToTriangle(	rayFrom,rayTo,dir,
-				f.m_n[0]->m_x,
-				f.m_n[1]->m_x,
-				f.m_n[2]->m_x,
-				mint);
-			if(t>0)
+			const btScalar t = RayFromToCaster::rayFromToTriangle(rayFrom, rayTo, dir,
+																  f.m_n[0]->m_x,
+																  f.m_n[1]->m_x,
+																  f.m_n[2]->m_x,
+																  mint);
+			if (t > 0)
 			{
 				++cnt;
-				if(!bcountonly)
+				if (!bcountonly)
 				{
-					feature=btSoftBody::eFeature::Face;
-					index=i;
-					mint=t;
+					feature = btSoftBody::eFeature::Face;
+					index = i;
+					mint = t;
 				}
 			}
 		}
 	}
 	else
-	{/* Use dbvt	*/ 
-		RayFromToCaster	collider(rayFrom,rayTo,mint);
+	{ /* Use dbvt	*/
+		RayFromToCaster collider(rayFrom, rayTo, mint);
 
-		btDbvt::rayTest(m_fdbvt.m_root,rayFrom,rayTo,collider);
-		if(collider.m_face)
+		btDbvt::rayTest(m_fdbvt.m_root, rayFrom, rayTo, collider);
+		if (collider.m_face)
 		{
-			mint=collider.m_mint;
-			feature=btSoftBody::eFeature::Face;
-			index=(int)(collider.m_face-&m_faces[0]);
-			cnt=1;
+			mint = collider.m_mint;
+			feature = btSoftBody::eFeature::Face;
+			index = (int)(collider.m_face - &m_faces[0]);
+			cnt = 1;
 		}
 	}
 
-	for (int i=0;i<m_tetras.size();i++)
+	for (int i = 0; i < m_tetras.size(); i++)
 	{
 		const btSoftBody::Tetra& tet = m_tetras[i];
-		int tetfaces[4][3] = {{0,1,2},{0,1,3},{1,2,3},{0,2,3}};
-		for (int f=0;f<4;f++)
+		int tetfaces[4][3] = {{0, 1, 2}, {0, 1, 3}, {1, 2, 3}, {0, 2, 3}};
+		for (int f = 0; f < 4; f++)
 		{
+			int index0 = tetfaces[f][0];
+			int index1 = tetfaces[f][1];
+			int index2 = tetfaces[f][2];
+			btVector3 v0 = tet.m_n[index0]->m_x;
+			btVector3 v1 = tet.m_n[index1]->m_x;
+			btVector3 v2 = tet.m_n[index2]->m_x;
 
-			int index0=tetfaces[f][0];
-			int index1=tetfaces[f][1];
-			int index2=tetfaces[f][2];
-			btVector3 v0=tet.m_n[index0]->m_x;
-			btVector3 v1=tet.m_n[index1]->m_x;
-			btVector3 v2=tet.m_n[index2]->m_x;
-
-
-		const btScalar			t=RayFromToCaster::rayFromToTriangle(	rayFrom,rayTo,dir,
-			v0,v1,v2,
-				mint);
-		if(t>0)
+			const btScalar t = RayFromToCaster::rayFromToTriangle(rayFrom, rayTo, dir,
+																  v0, v1, v2,
+																  mint);
+			if (t > 0)
 			{
 				++cnt;
-				if(!bcountonly)
+				if (!bcountonly)
 				{
-					feature=btSoftBody::eFeature::Tetra;
-					index=i;
-					mint=t;
+					feature = btSoftBody::eFeature::Tetra;
+					index = i;
+					mint = t;
 				}
 			}
 		}
 	}
-	return(cnt);
+	return (cnt);
 }
 
 //
-void			btSoftBody::initializeFaceTree()
+void btSoftBody::initializeFaceTree()
 {
 	m_fdbvt.clear();
-	for(int i=0;i<m_faces.size();++i)
+	for (int i = 0; i < m_faces.size(); ++i)
 	{
-		Face&	f=m_faces[i];
-		f.m_leaf=m_fdbvt.insert(VolumeOf(f,0),&f);
+		Face& f = m_faces[i];
+		f.m_leaf = m_fdbvt.insert(VolumeOf(f, 0), &f);
 	}
 }
 
 //
-btVector3		btSoftBody::evaluateCom() const
+btVector3 btSoftBody::evaluateCom() const
 {
-	btVector3	com(0,0,0);
-	if(m_pose.m_bframe)
+	btVector3 com(0, 0, 0);
+	if (m_pose.m_bframe)
 	{
-		for(int i=0,ni=m_nodes.size();i<ni;++i)
+		for (int i = 0, ni = m_nodes.size(); i < ni; ++i)
 		{
-			com+=m_nodes[i].m_x*m_pose.m_wgh[i];
+			com += m_nodes[i].m_x * m_pose.m_wgh[i];
 		}
 	}
-	return(com);
+	return (com);
 }
 
 //
-bool				btSoftBody::checkContact(	const btCollisionObjectWrapper* colObjWrap,
-											 const btVector3& x,
-											 btScalar margin,
-											 btSoftBody::sCti& cti) const
+bool btSoftBody::checkContact(const btCollisionObjectWrapper* colObjWrap,
+							  const btVector3& x,
+							  btScalar margin,
+							  btSoftBody::sCti& cti) const
 {
 	btVector3 nrm;
-	const btCollisionShape *shp = colObjWrap->getCollisionShape();
-//	const btRigidBody *tmpRigid = btRigidBody::upcast(colObjWrap->getCollisionObject());
+	const btCollisionShape* shp = colObjWrap->getCollisionShape();
+	//	const btRigidBody *tmpRigid = btRigidBody::upcast(colObjWrap->getCollisionObject());
 	//const btTransform &wtr = tmpRigid ? tmpRigid->getWorldTransform() : colObjWrap->getWorldTransform();
-	const btTransform &wtr = colObjWrap->getWorldTransform();
+	const btTransform& wtr = colObjWrap->getWorldTransform();
 	//todo: check which transform is needed here
 
-	btScalar dst = 
-		m_worldInfo->m_sparsesdf.Evaluate(	
+	btScalar dst =
+		m_worldInfo->m_sparsesdf.Evaluate(
 			wtr.invXform(x),
 			shp,
 			nrm,
 			margin);
-	if(dst<0)
+	if (dst < 0)
 	{
 		cti.m_colObj = colObjWrap->getCollisionObject();
-		cti.m_normal = wtr.getBasis()*nrm;
-		cti.m_offset = -btDot( cti.m_normal, x - cti.m_normal * dst );
-		return(true);
+		cti.m_normal = wtr.getBasis() * nrm;
+		cti.m_offset = -btDot(cti.m_normal, x - cti.m_normal * dst);
+		return (true);
 	}
-	return(false);
+	return (false);
 }
 
 //
-void					btSoftBody::updateNormals()
+void btSoftBody::updateNormals()
 {
+	const btVector3 zv(0, 0, 0);
+	int i, ni;
 
-	const btVector3	zv(0,0,0);
-	int i,ni;
-
-	for(i=0,ni=m_nodes.size();i<ni;++i)
+	for (i = 0, ni = m_nodes.size(); i < ni; ++i)
 	{
-		m_nodes[i].m_n=zv;
+		m_nodes[i].m_n = zv;
 	}
-	for(i=0,ni=m_faces.size();i<ni;++i)
+	for (i = 0, ni = m_faces.size(); i < ni; ++i)
 	{
-		btSoftBody::Face&	f=m_faces[i];
-		const btVector3		n=btCross(f.m_n[1]->m_x-f.m_n[0]->m_x,
-			f.m_n[2]->m_x-f.m_n[0]->m_x);
-		f.m_normal=n.normalized();
-		f.m_n[0]->m_n+=n;
-		f.m_n[1]->m_n+=n;
-		f.m_n[2]->m_n+=n;
+		btSoftBody::Face& f = m_faces[i];
+		const btVector3 n = btCross(f.m_n[1]->m_x - f.m_n[0]->m_x,
+									f.m_n[2]->m_x - f.m_n[0]->m_x);
+		f.m_normal = n.normalized();
+		f.m_n[0]->m_n += n;
+		f.m_n[1]->m_n += n;
+		f.m_n[2]->m_n += n;
 	}
-	for(i=0,ni=m_nodes.size();i<ni;++i)
+	for (i = 0, ni = m_nodes.size(); i < ni; ++i)
 	{
 		btScalar len = m_nodes[i].m_n.length();
-		if (len>SIMD_EPSILON)
+		if (len > SIMD_EPSILON)
 			m_nodes[i].m_n /= len;
 	}
 }
 
 //
-void					btSoftBody::updateBounds()
+void btSoftBody::updateBounds()
 {
 	/*if( m_acceleratedSoftBody )
 	{
@@ -2317,258 +2333,259 @@ void					btSoftBody::updateBounds()
 		m_bounds[1] = btVector3(1000, 1000, 1000);
 
 	} else {*/
-		if(m_ndbvt.m_root)
+	if (m_ndbvt.m_root)
+	{
+		const btVector3& mins = m_ndbvt.m_root->volume.Mins();
+		const btVector3& maxs = m_ndbvt.m_root->volume.Maxs();
+		const btScalar csm = getCollisionShape()->getMargin();
+		const btVector3 mrg = btVector3(csm,
+										csm,
+										csm) *
+							  1;  // ??? to investigate...
+		m_bounds[0] = mins - mrg;
+		m_bounds[1] = maxs + mrg;
+		if (0 != getBroadphaseHandle())
 		{
-			const btVector3&	mins=m_ndbvt.m_root->volume.Mins();
-			const btVector3&	maxs=m_ndbvt.m_root->volume.Maxs();
-			const btScalar		csm=getCollisionShape()->getMargin();
-			const btVector3		mrg=btVector3(	csm,
-				csm,
-				csm)*1; // ??? to investigate...
-			m_bounds[0]=mins-mrg;
-			m_bounds[1]=maxs+mrg;
-			if(0!=getBroadphaseHandle())
-			{					
-				m_worldInfo->m_broadphase->setAabb(	getBroadphaseHandle(),
-					m_bounds[0],
-					m_bounds[1],
-					m_worldInfo->m_dispatcher);
-			}
+			m_worldInfo->m_broadphase->setAabb(getBroadphaseHandle(),
+											   m_bounds[0],
+											   m_bounds[1],
+											   m_worldInfo->m_dispatcher);
 		}
-		else
-		{
-			m_bounds[0]=
-				m_bounds[1]=btVector3(0,0,0);
-		}		
+	}
+	else
+	{
+		m_bounds[0] =
+			m_bounds[1] = btVector3(0, 0, 0);
+	}
 	//}
 }
 
-
 //
-void					btSoftBody::updatePose()
+void btSoftBody::updatePose()
 {
-	if(m_pose.m_bframe)
+	if (m_pose.m_bframe)
 	{
-		btSoftBody::Pose&	pose=m_pose;
-		const btVector3		com=evaluateCom();
-		/* Com			*/ 
-		pose.m_com	=	com;
-		/* Rotation		*/ 
-		btMatrix3x3		Apq;
-		const btScalar	eps=SIMD_EPSILON;
-		Apq[0]=Apq[1]=Apq[2]=btVector3(0,0,0);
-		Apq[0].setX(eps);Apq[1].setY(eps*2);Apq[2].setZ(eps*3);
-		for(int i=0,ni=m_nodes.size();i<ni;++i)
+		btSoftBody::Pose& pose = m_pose;
+		const btVector3 com = evaluateCom();
+		/* Com			*/
+		pose.m_com = com;
+		/* Rotation		*/
+		btMatrix3x3 Apq;
+		const btScalar eps = SIMD_EPSILON;
+		Apq[0] = Apq[1] = Apq[2] = btVector3(0, 0, 0);
+		Apq[0].setX(eps);
+		Apq[1].setY(eps * 2);
+		Apq[2].setZ(eps * 3);
+		for (int i = 0, ni = m_nodes.size(); i < ni; ++i)
 		{
-			const btVector3		a=pose.m_wgh[i]*(m_nodes[i].m_x-com);
-			const btVector3&	b=pose.m_pos[i];
-			Apq[0]+=a.x()*b;
-			Apq[1]+=a.y()*b;
-			Apq[2]+=a.z()*b;
+			const btVector3 a = pose.m_wgh[i] * (m_nodes[i].m_x - com);
+			const btVector3& b = pose.m_pos[i];
+			Apq[0] += a.x() * b;
+			Apq[1] += a.y() * b;
+			Apq[2] += a.z() * b;
 		}
-		btMatrix3x3		r,s;
-		PolarDecompose(Apq,r,s);
-		pose.m_rot=r;
-		pose.m_scl=pose.m_aqq*r.transpose()*Apq;
-		if(m_cfg.maxvolume>1)
+		btMatrix3x3 r, s;
+		PolarDecompose(Apq, r, s);
+		pose.m_rot = r;
+		pose.m_scl = pose.m_aqq * r.transpose() * Apq;
+		if (m_cfg.maxvolume > 1)
 		{
-			const btScalar	idet=Clamp<btScalar>(	1/pose.m_scl.determinant(),
-				1,m_cfg.maxvolume);
-			pose.m_scl=Mul(pose.m_scl,idet);
+			const btScalar idet = Clamp<btScalar>(1 / pose.m_scl.determinant(),
+												  1, m_cfg.maxvolume);
+			pose.m_scl = Mul(pose.m_scl, idet);
 		}
-
 	}
 }
 
 //
-void				btSoftBody::updateArea(bool averageArea)
+void btSoftBody::updateArea(bool averageArea)
 {
-	int i,ni;
+	int i, ni;
 
-	/* Face area		*/ 
-	for(i=0,ni=m_faces.size();i<ni;++i)
+	/* Face area		*/
+	for (i = 0, ni = m_faces.size(); i < ni; ++i)
 	{
-		Face&		f=m_faces[i];
-		f.m_ra	=	AreaOf(f.m_n[0]->m_x,f.m_n[1]->m_x,f.m_n[2]->m_x);
+		Face& f = m_faces[i];
+		f.m_ra = AreaOf(f.m_n[0]->m_x, f.m_n[1]->m_x, f.m_n[2]->m_x);
 	}
-	
-	/* Node area		*/ 
+
+	/* Node area		*/
 
 	if (averageArea)
 	{
-		btAlignedObjectArray<int>	counts;
-		counts.resize(m_nodes.size(),0);
-		for(i=0,ni=m_nodes.size();i<ni;++i)
+		btAlignedObjectArray<int> counts;
+		counts.resize(m_nodes.size(), 0);
+		for (i = 0, ni = m_nodes.size(); i < ni; ++i)
 		{
-			m_nodes[i].m_area	=	0;
+			m_nodes[i].m_area = 0;
 		}
-		for(i=0,ni=m_faces.size();i<ni;++i)
+		for (i = 0, ni = m_faces.size(); i < ni; ++i)
 		{
-			btSoftBody::Face&	f=m_faces[i];
-			for(int j=0;j<3;++j)
+			btSoftBody::Face& f = m_faces[i];
+			for (int j = 0; j < 3; ++j)
 			{
-				const int index=(int)(f.m_n[j]-&m_nodes[0]);
+				const int index = (int)(f.m_n[j] - &m_nodes[0]);
 				counts[index]++;
-				f.m_n[j]->m_area+=btFabs(f.m_ra);
+				f.m_n[j]->m_area += btFabs(f.m_ra);
 			}
 		}
-		for(i=0,ni=m_nodes.size();i<ni;++i)
+		for (i = 0, ni = m_nodes.size(); i < ni; ++i)
 		{
-			if(counts[i]>0)
-				m_nodes[i].m_area/=(btScalar)counts[i];
+			if (counts[i] > 0)
+				m_nodes[i].m_area /= (btScalar)counts[i];
 			else
-				m_nodes[i].m_area=0;
+				m_nodes[i].m_area = 0;
 		}
 	}
 	else
 	{
 		// initialize node area as zero
-		for(i=0,ni=m_nodes.size();i<ni;++i)
+		for (i = 0, ni = m_nodes.size(); i < ni; ++i)
 		{
-			m_nodes[i].m_area=0;	
+			m_nodes[i].m_area = 0;
 		}
 
-		for(i=0,ni=m_faces.size();i<ni;++i)
+		for (i = 0, ni = m_faces.size(); i < ni; ++i)
 		{
-			btSoftBody::Face&	f=m_faces[i];
+			btSoftBody::Face& f = m_faces[i];
 
-			for(int j=0;j<3;++j)
+			for (int j = 0; j < 3; ++j)
 			{
 				f.m_n[j]->m_area += f.m_ra;
 			}
 		}
 
-		for(i=0,ni=m_nodes.size();i<ni;++i)
+		for (i = 0, ni = m_nodes.size(); i < ni; ++i)
 		{
 			m_nodes[i].m_area *= 0.3333333f;
 		}
 	}
 }
 
+void btSoftBody::updateLinkConstants()
+{
+	int i, ni;
 
-void				btSoftBody::updateLinkConstants()
-{	
-	int i,ni;
-
-	/* Links		*/ 
-	for(i=0,ni=m_links.size();i<ni;++i)
+	/* Links		*/
+	for (i = 0, ni = m_links.size(); i < ni; ++i)
 	{
-		Link&		l=m_links[i];
-		Material&	m=*l.m_material;
-		l.m_c0	=	(l.m_n[0]->m_im+l.m_n[1]->m_im)/m.m_kLST;
+		Link& l = m_links[i];
+		Material& m = *l.m_material;
+		l.m_c0 = (l.m_n[0]->m_im + l.m_n[1]->m_im) / m.m_kLST;
 	}
 }
 
-void				btSoftBody::updateConstants()
+void btSoftBody::updateConstants()
 {
 	resetLinkRestLengths();
 	updateLinkConstants();
 	updateArea();
 }
 
-
-
 //
-void					btSoftBody::initializeClusters()
+void btSoftBody::initializeClusters()
 {
 	int i;
 
-	for( i=0;i<m_clusters.size();++i)
+	for (i = 0; i < m_clusters.size(); ++i)
 	{
-		Cluster&	c=*m_clusters[i];
-		c.m_imass=0;
+		Cluster& c = *m_clusters[i];
+		c.m_imass = 0;
 		c.m_masses.resize(c.m_nodes.size());
-		for(int j=0;j<c.m_nodes.size();++j)
+		for (int j = 0; j < c.m_nodes.size(); ++j)
 		{
-			if (c.m_nodes[j]->m_im==0)
+			if (c.m_nodes[j]->m_im == 0)
 			{
 				c.m_containsAnchor = true;
-				c.m_masses[j]	=	BT_LARGE_FLOAT;
-			} else
-			{
-				c.m_masses[j]	=	btScalar(1.)/c.m_nodes[j]->m_im;
+				c.m_masses[j] = BT_LARGE_FLOAT;
 			}
-			c.m_imass		+=	c.m_masses[j];
+			else
+			{
+				c.m_masses[j] = btScalar(1.) / c.m_nodes[j]->m_im;
+			}
+			c.m_imass += c.m_masses[j];
 		}
-		c.m_imass		=	btScalar(1.)/c.m_imass;
-		c.m_com			=	btSoftBody::clusterCom(&c);
-		c.m_lv			=	btVector3(0,0,0);
-		c.m_av			=	btVector3(0,0,0);
-		c.m_leaf		=	0;
-		/* Inertia	*/ 
-		btMatrix3x3&	ii=c.m_locii;
-		ii[0]=ii[1]=ii[2]=btVector3(0,0,0);
+		c.m_imass = btScalar(1.) / c.m_imass;
+		c.m_com = btSoftBody::clusterCom(&c);
+		c.m_lv = btVector3(0, 0, 0);
+		c.m_av = btVector3(0, 0, 0);
+		c.m_leaf = 0;
+		/* Inertia	*/
+		btMatrix3x3& ii = c.m_locii;
+		ii[0] = ii[1] = ii[2] = btVector3(0, 0, 0);
 		{
-			int i,ni;
+			int i, ni;
 
-			for(i=0,ni=c.m_nodes.size();i<ni;++i)
+			for (i = 0, ni = c.m_nodes.size(); i < ni; ++i)
 			{
-				const btVector3	k=c.m_nodes[i]->m_x-c.m_com;
-				const btVector3	q=k*k;
-				const btScalar	m=c.m_masses[i];
-				ii[0][0]	+=	m*(q[1]+q[2]);
-				ii[1][1]	+=	m*(q[0]+q[2]);
-				ii[2][2]	+=	m*(q[0]+q[1]);
-				ii[0][1]	-=	m*k[0]*k[1];
-				ii[0][2]	-=	m*k[0]*k[2];
-				ii[1][2]	-=	m*k[1]*k[2];
+				const btVector3 k = c.m_nodes[i]->m_x - c.m_com;
+				const btVector3 q = k * k;
+				const btScalar m = c.m_masses[i];
+				ii[0][0] += m * (q[1] + q[2]);
+				ii[1][1] += m * (q[0] + q[2]);
+				ii[2][2] += m * (q[0] + q[1]);
+				ii[0][1] -= m * k[0] * k[1];
+				ii[0][2] -= m * k[0] * k[2];
+				ii[1][2] -= m * k[1] * k[2];
 			}
 		}
-		ii[1][0]=ii[0][1];
-		ii[2][0]=ii[0][2];
-		ii[2][1]=ii[1][2];
-		
+		ii[1][0] = ii[0][1];
+		ii[2][0] = ii[0][2];
+		ii[2][1] = ii[1][2];
+
 		ii = ii.inverse();
 
-		/* Frame	*/ 
+		/* Frame	*/
 		c.m_framexform.setIdentity();
 		c.m_framexform.setOrigin(c.m_com);
 		c.m_framerefs.resize(c.m_nodes.size());
 		{
 			int i;
-			for(i=0;i<c.m_framerefs.size();++i)
+			for (i = 0; i < c.m_framerefs.size(); ++i)
 			{
-				c.m_framerefs[i]=c.m_nodes[i]->m_x-c.m_com;
+				c.m_framerefs[i] = c.m_nodes[i]->m_x - c.m_com;
 			}
 		}
 	}
 }
 
 //
-void					btSoftBody::updateClusters()
+void btSoftBody::updateClusters()
 {
 	BT_PROFILE("UpdateClusters");
 	int i;
 
-	for(i=0;i<m_clusters.size();++i)
+	for (i = 0; i < m_clusters.size(); ++i)
 	{
-		btSoftBody::Cluster&	c=*m_clusters[i];
-		const int				n=c.m_nodes.size();
+		btSoftBody::Cluster& c = *m_clusters[i];
+		const int n = c.m_nodes.size();
 		//const btScalar			invn=1/(btScalar)n;
-		if(n)
+		if (n)
 		{
-			/* Frame				*/ 
-			const btScalar	eps=btScalar(0.0001);
-			btMatrix3x3		m,r,s;
-			m[0]=m[1]=m[2]=btVector3(0,0,0);
-			m[0][0]=eps*1;
-			m[1][1]=eps*2;
-			m[2][2]=eps*3;
-			c.m_com=clusterCom(&c);
-			for(int i=0;i<c.m_nodes.size();++i)
+			/* Frame				*/
+			const btScalar eps = btScalar(0.0001);
+			btMatrix3x3 m, r, s;
+			m[0] = m[1] = m[2] = btVector3(0, 0, 0);
+			m[0][0] = eps * 1;
+			m[1][1] = eps * 2;
+			m[2][2] = eps * 3;
+			c.m_com = clusterCom(&c);
+			for (int i = 0; i < c.m_nodes.size(); ++i)
 			{
-				const btVector3		a=c.m_nodes[i]->m_x-c.m_com;
-				const btVector3&	b=c.m_framerefs[i];
-				m[0]+=a[0]*b;m[1]+=a[1]*b;m[2]+=a[2]*b;
+				const btVector3 a = c.m_nodes[i]->m_x - c.m_com;
+				const btVector3& b = c.m_framerefs[i];
+				m[0] += a[0] * b;
+				m[1] += a[1] * b;
+				m[2] += a[2] * b;
 			}
-			PolarDecompose(m,r,s);
+			PolarDecompose(m, r, s);
 			c.m_framexform.setOrigin(c.m_com);
-			c.m_framexform.setBasis(r);		
-			/* Inertia			*/ 
-#if 1/* Constant	*/ 
-			c.m_invwi=c.m_framexform.getBasis()*c.m_locii*c.m_framexform.getBasis().transpose();
+			c.m_framexform.setBasis(r);
+			/* Inertia			*/
+#if 1 /* Constant	*/
+			c.m_invwi = c.m_framexform.getBasis() * c.m_locii * c.m_framexform.getBasis().transpose();
 #else
-#if 0/* Sphere	*/ 
+#if 0 /* Sphere	*/ 
 			const btScalar	rk=(2*c.m_extents.length2())/(5*c.m_imass);
 			const btVector3	inertia(rk,rk,rk);
 			const btVector3	iin(btFabs(inertia[0])>SIMD_EPSILON?1/inertia[0]:0,
@@ -2576,186 +2593,181 @@ void					btSoftBody::updateClusters()
 				btFabs(inertia[2])>SIMD_EPSILON?1/inertia[2]:0);
 
 			c.m_invwi=c.m_xform.getBasis().scaled(iin)*c.m_xform.getBasis().transpose();
-#else/* Actual	*/ 		
-			c.m_invwi[0]=c.m_invwi[1]=c.m_invwi[2]=btVector3(0,0,0);
-			for(int i=0;i<n;++i)
+#else /* Actual	*/
+			c.m_invwi[0] = c.m_invwi[1] = c.m_invwi[2] = btVector3(0, 0, 0);
+			for (int i = 0; i < n; ++i)
 			{
-				const btVector3	k=c.m_nodes[i]->m_x-c.m_com;
-				const btVector3		q=k*k;
-				const btScalar		m=1/c.m_nodes[i]->m_im;
-				c.m_invwi[0][0]	+=	m*(q[1]+q[2]);
-				c.m_invwi[1][1]	+=	m*(q[0]+q[2]);
-				c.m_invwi[2][2]	+=	m*(q[0]+q[1]);
-				c.m_invwi[0][1]	-=	m*k[0]*k[1];
-				c.m_invwi[0][2]	-=	m*k[0]*k[2];
-				c.m_invwi[1][2]	-=	m*k[1]*k[2];
+				const btVector3 k = c.m_nodes[i]->m_x - c.m_com;
+				const btVector3 q = k * k;
+				const btScalar m = 1 / c.m_nodes[i]->m_im;
+				c.m_invwi[0][0] += m * (q[1] + q[2]);
+				c.m_invwi[1][1] += m * (q[0] + q[2]);
+				c.m_invwi[2][2] += m * (q[0] + q[1]);
+				c.m_invwi[0][1] -= m * k[0] * k[1];
+				c.m_invwi[0][2] -= m * k[0] * k[2];
+				c.m_invwi[1][2] -= m * k[1] * k[2];
 			}
-			c.m_invwi[1][0]=c.m_invwi[0][1];
-			c.m_invwi[2][0]=c.m_invwi[0][2];
-			c.m_invwi[2][1]=c.m_invwi[1][2];
-			c.m_invwi=c.m_invwi.inverse();
+			c.m_invwi[1][0] = c.m_invwi[0][1];
+			c.m_invwi[2][0] = c.m_invwi[0][2];
+			c.m_invwi[2][1] = c.m_invwi[1][2];
+			c.m_invwi = c.m_invwi.inverse();
 #endif
 #endif
-			/* Velocities			*/ 
-			c.m_lv=btVector3(0,0,0);
-			c.m_av=btVector3(0,0,0);
+			/* Velocities			*/
+			c.m_lv = btVector3(0, 0, 0);
+			c.m_av = btVector3(0, 0, 0);
 			{
 				int i;
 
-				for(i=0;i<n;++i)
+				for (i = 0; i < n; ++i)
 				{
-					const btVector3	v=c.m_nodes[i]->m_v*c.m_masses[i];
-					c.m_lv	+=	v;
-					c.m_av	+=	btCross(c.m_nodes[i]->m_x-c.m_com,v);
+					const btVector3 v = c.m_nodes[i]->m_v * c.m_masses[i];
+					c.m_lv += v;
+					c.m_av += btCross(c.m_nodes[i]->m_x - c.m_com, v);
 				}
 			}
-			c.m_lv=c.m_imass*c.m_lv*(1-c.m_ldamping);
-			c.m_av=c.m_invwi*c.m_av*(1-c.m_adamping);
-			c.m_vimpulses[0]	=
-				c.m_vimpulses[1]	= btVector3(0,0,0);
-			c.m_dimpulses[0]	=
-				c.m_dimpulses[1]	= btVector3(0,0,0);
-			c.m_nvimpulses		= 0;
-			c.m_ndimpulses		= 0;
-			/* Matching				*/ 
-			if(c.m_matching>0)
+			c.m_lv = c.m_imass * c.m_lv * (1 - c.m_ldamping);
+			c.m_av = c.m_invwi * c.m_av * (1 - c.m_adamping);
+			c.m_vimpulses[0] =
+				c.m_vimpulses[1] = btVector3(0, 0, 0);
+			c.m_dimpulses[0] =
+				c.m_dimpulses[1] = btVector3(0, 0, 0);
+			c.m_nvimpulses = 0;
+			c.m_ndimpulses = 0;
+			/* Matching				*/
+			if (c.m_matching > 0)
 			{
-				for(int j=0;j<c.m_nodes.size();++j)
+				for (int j = 0; j < c.m_nodes.size(); ++j)
 				{
-					Node&			n=*c.m_nodes[j];
-					const btVector3	x=c.m_framexform*c.m_framerefs[j];
-					n.m_x=Lerp(n.m_x,x,c.m_matching);
+					Node& n = *c.m_nodes[j];
+					const btVector3 x = c.m_framexform * c.m_framerefs[j];
+					n.m_x = Lerp(n.m_x, x, c.m_matching);
 				}
-			}			
-			/* Dbvt					*/ 
-			if(c.m_collide)
+			}
+			/* Dbvt					*/
+			if (c.m_collide)
 			{
-				btVector3	mi=c.m_nodes[0]->m_x;
-				btVector3	mx=mi;
-				for(int j=1;j<n;++j)
+				btVector3 mi = c.m_nodes[0]->m_x;
+				btVector3 mx = mi;
+				for (int j = 1; j < n; ++j)
 				{
 					mi.setMin(c.m_nodes[j]->m_x);
 					mx.setMax(c.m_nodes[j]->m_x);
-				}			
-				ATTRIBUTE_ALIGNED16(btDbvtVolume)	bounds=btDbvtVolume::FromMM(mi,mx);
-				if(c.m_leaf)
-					m_cdbvt.update(c.m_leaf,bounds,c.m_lv*m_sst.sdt*3,m_sst.radmrg);
+				}
+				ATTRIBUTE_ALIGNED16(btDbvtVolume)
+				bounds = btDbvtVolume::FromMM(mi, mx);
+				if (c.m_leaf)
+					m_cdbvt.update(c.m_leaf, bounds, c.m_lv * m_sst.sdt * 3, m_sst.radmrg);
 				else
-					c.m_leaf=m_cdbvt.insert(bounds,&c);
+					c.m_leaf = m_cdbvt.insert(bounds, &c);
 			}
 		}
 	}
-
-
 }
 
-
-
-
 //
-void					btSoftBody::cleanupClusters()
+void btSoftBody::cleanupClusters()
 {
-	for(int i=0;i<m_joints.size();++i)
+	for (int i = 0; i < m_joints.size(); ++i)
 	{
 		m_joints[i]->Terminate(m_sst.sdt);
-		if(m_joints[i]->m_delete)
+		if (m_joints[i]->m_delete)
 		{
 			btAlignedFree(m_joints[i]);
 			m_joints.remove(m_joints[i--]);
-		}	
+		}
 	}
 }
 
 //
-void					btSoftBody::prepareClusters(int iterations)
+void btSoftBody::prepareClusters(int iterations)
 {
-	for(int i=0;i<m_joints.size();++i)
+	for (int i = 0; i < m_joints.size(); ++i)
 	{
-		m_joints[i]->Prepare(m_sst.sdt,iterations);
+		m_joints[i]->Prepare(m_sst.sdt, iterations);
 	}
 }
 
-
 //
-void					btSoftBody::solveClusters(btScalar sor)
+void btSoftBody::solveClusters(btScalar sor)
 {
-	for(int i=0,ni=m_joints.size();i<ni;++i)
+	for (int i = 0, ni = m_joints.size(); i < ni; ++i)
 	{
-		m_joints[i]->Solve(m_sst.sdt,sor);
+		m_joints[i]->Solve(m_sst.sdt, sor);
 	}
 }
 
 //
-void					btSoftBody::applyClusters(bool drift)
+void btSoftBody::applyClusters(bool drift)
 {
 	BT_PROFILE("ApplyClusters");
-//	const btScalar					f0=m_sst.sdt;
+	//	const btScalar					f0=m_sst.sdt;
 	//const btScalar					f1=f0/2;
 	btAlignedObjectArray<btVector3> deltas;
 	btAlignedObjectArray<btScalar> weights;
-	deltas.resize(m_nodes.size(),btVector3(0,0,0));
-	weights.resize(m_nodes.size(),0);
+	deltas.resize(m_nodes.size(), btVector3(0, 0, 0));
+	weights.resize(m_nodes.size(), 0);
 	int i;
 
-	if(drift)
+	if (drift)
 	{
-		for(i=0;i<m_clusters.size();++i)
+		for (i = 0; i < m_clusters.size(); ++i)
 		{
-			Cluster&	c=*m_clusters[i];
-			if(c.m_ndimpulses)
+			Cluster& c = *m_clusters[i];
+			if (c.m_ndimpulses)
 			{
-				c.m_dimpulses[0]/=(btScalar)c.m_ndimpulses;
-				c.m_dimpulses[1]/=(btScalar)c.m_ndimpulses;
+				c.m_dimpulses[0] /= (btScalar)c.m_ndimpulses;
+				c.m_dimpulses[1] /= (btScalar)c.m_ndimpulses;
 			}
 		}
 	}
-	
-	for(i=0;i<m_clusters.size();++i)
+
+	for (i = 0; i < m_clusters.size(); ++i)
 	{
-		Cluster&	c=*m_clusters[i];	
-		if(0<(drift?c.m_ndimpulses:c.m_nvimpulses))
+		Cluster& c = *m_clusters[i];
+		if (0 < (drift ? c.m_ndimpulses : c.m_nvimpulses))
 		{
-			const btVector3		v=(drift?c.m_dimpulses[0]:c.m_vimpulses[0])*m_sst.sdt;
-			const btVector3		w=(drift?c.m_dimpulses[1]:c.m_vimpulses[1])*m_sst.sdt;
-			for(int j=0;j<c.m_nodes.size();++j)
+			const btVector3 v = (drift ? c.m_dimpulses[0] : c.m_vimpulses[0]) * m_sst.sdt;
+			const btVector3 w = (drift ? c.m_dimpulses[1] : c.m_vimpulses[1]) * m_sst.sdt;
+			for (int j = 0; j < c.m_nodes.size(); ++j)
 			{
-				const int			idx=int(c.m_nodes[j]-&m_nodes[0]);
-				const btVector3&	x=c.m_nodes[j]->m_x;
-				const btScalar		q=c.m_masses[j];
-				deltas[idx]		+=	(v+btCross(w,x-c.m_com))*q;
-				weights[idx]	+=	q;
+				const int idx = int(c.m_nodes[j] - &m_nodes[0]);
+				const btVector3& x = c.m_nodes[j]->m_x;
+				const btScalar q = c.m_masses[j];
+				deltas[idx] += (v + btCross(w, x - c.m_com)) * q;
+				weights[idx] += q;
 			}
 		}
 	}
-	for(i=0;i<deltas.size();++i)
+	for (i = 0; i < deltas.size(); ++i)
 	{
-		if(weights[i]>0) 
+		if (weights[i] > 0)
 		{
-			m_nodes[i].m_x+=deltas[i]/weights[i];
+			m_nodes[i].m_x += deltas[i] / weights[i];
 		}
 	}
 }
 
 //
-void					btSoftBody::dampClusters()
+void btSoftBody::dampClusters()
 {
 	int i;
 
-	for(i=0;i<m_clusters.size();++i)
+	for (i = 0; i < m_clusters.size(); ++i)
 	{
-		Cluster&	c=*m_clusters[i];	
-		if(c.m_ndamping>0)
+		Cluster& c = *m_clusters[i];
+		if (c.m_ndamping > 0)
 		{
-			for(int j=0;j<c.m_nodes.size();++j)
+			for (int j = 0; j < c.m_nodes.size(); ++j)
 			{
-				Node&			n=*c.m_nodes[j];
-				if(n.m_im>0)
+				Node& n = *c.m_nodes[j];
+				if (n.m_im > 0)
 				{
-					const btVector3	vx=c.m_lv+btCross(c.m_av,c.m_nodes[j]->m_q-c.m_com);
-					if(vx.length2()<=n.m_v.length2())
-						{
-						n.m_v	+=	c.m_ndamping*(vx-n.m_v);
-						}
+					const btVector3 vx = c.m_lv + btCross(c.m_av, c.m_nodes[j]->m_q - c.m_com);
+					if (vx.length2() <= n.m_v.length2())
+					{
+						n.m_v += c.m_ndamping * (vx - n.m_v);
+					}
 				}
 			}
 		}
@@ -2763,710 +2775,696 @@ void					btSoftBody::dampClusters()
 }
 
 //
-void				btSoftBody::Joint::Prepare(btScalar dt,int)
+void btSoftBody::Joint::Prepare(btScalar dt, int)
 {
 	m_bodies[0].activate();
 	m_bodies[1].activate();
 }
 
 //
-void				btSoftBody::LJoint::Prepare(btScalar dt,int iterations)
+void btSoftBody::LJoint::Prepare(btScalar dt, int iterations)
 {
-	static const btScalar	maxdrift=4;
-	Joint::Prepare(dt,iterations);
-	m_rpos[0]		=	m_bodies[0].xform()*m_refs[0];
-	m_rpos[1]		=	m_bodies[1].xform()*m_refs[1];
-	m_drift			=	Clamp(m_rpos[0]-m_rpos[1],maxdrift)*m_erp/dt;
-	m_rpos[0]		-=	m_bodies[0].xform().getOrigin();
-	m_rpos[1]		-=	m_bodies[1].xform().getOrigin();
-	m_massmatrix	=	ImpulseMatrix(	m_bodies[0].invMass(),m_bodies[0].invWorldInertia(),m_rpos[0],
-		m_bodies[1].invMass(),m_bodies[1].invWorldInertia(),m_rpos[1]);
-	if(m_split>0)
+	static const btScalar maxdrift = 4;
+	Joint::Prepare(dt, iterations);
+	m_rpos[0] = m_bodies[0].xform() * m_refs[0];
+	m_rpos[1] = m_bodies[1].xform() * m_refs[1];
+	m_drift = Clamp(m_rpos[0] - m_rpos[1], maxdrift) * m_erp / dt;
+	m_rpos[0] -= m_bodies[0].xform().getOrigin();
+	m_rpos[1] -= m_bodies[1].xform().getOrigin();
+	m_massmatrix = ImpulseMatrix(m_bodies[0].invMass(), m_bodies[0].invWorldInertia(), m_rpos[0],
+								 m_bodies[1].invMass(), m_bodies[1].invWorldInertia(), m_rpos[1]);
+	if (m_split > 0)
 	{
-		m_sdrift	=	m_massmatrix*(m_drift*m_split);
-		m_drift		*=	1-m_split;
+		m_sdrift = m_massmatrix * (m_drift * m_split);
+		m_drift *= 1 - m_split;
 	}
-	m_drift	/=(btScalar)iterations;
+	m_drift /= (btScalar)iterations;
 }
 
 //
-void				btSoftBody::LJoint::Solve(btScalar dt,btScalar sor)
+void btSoftBody::LJoint::Solve(btScalar dt, btScalar sor)
 {
-	const btVector3		va=m_bodies[0].velocity(m_rpos[0]);
-	const btVector3		vb=m_bodies[1].velocity(m_rpos[1]);
-	const btVector3		vr=va-vb;
-	btSoftBody::Impulse	impulse;
-	impulse.m_asVelocity	=	1;
-	impulse.m_velocity		=	m_massmatrix*(m_drift+vr*m_cfm)*sor;
-	m_bodies[0].applyImpulse(-impulse,m_rpos[0]);
-	m_bodies[1].applyImpulse( impulse,m_rpos[1]);
+	const btVector3 va = m_bodies[0].velocity(m_rpos[0]);
+	const btVector3 vb = m_bodies[1].velocity(m_rpos[1]);
+	const btVector3 vr = va - vb;
+	btSoftBody::Impulse impulse;
+	impulse.m_asVelocity = 1;
+	impulse.m_velocity = m_massmatrix * (m_drift + vr * m_cfm) * sor;
+	m_bodies[0].applyImpulse(-impulse, m_rpos[0]);
+	m_bodies[1].applyImpulse(impulse, m_rpos[1]);
 }
 
 //
-void				btSoftBody::LJoint::Terminate(btScalar dt)
+void btSoftBody::LJoint::Terminate(btScalar dt)
 {
-	if(m_split>0)
+	if (m_split > 0)
 	{
-		m_bodies[0].applyDImpulse(-m_sdrift,m_rpos[0]);
-		m_bodies[1].applyDImpulse( m_sdrift,m_rpos[1]);
+		m_bodies[0].applyDImpulse(-m_sdrift, m_rpos[0]);
+		m_bodies[1].applyDImpulse(m_sdrift, m_rpos[1]);
 	}
 }
 
 //
-void				btSoftBody::AJoint::Prepare(btScalar dt,int iterations)
+void btSoftBody::AJoint::Prepare(btScalar dt, int iterations)
 {
-	static const btScalar	maxdrift=SIMD_PI/16;
+	static const btScalar maxdrift = SIMD_PI / 16;
 	m_icontrol->Prepare(this);
-	Joint::Prepare(dt,iterations);
-	m_axis[0]	=	m_bodies[0].xform().getBasis()*m_refs[0];
-	m_axis[1]	=	m_bodies[1].xform().getBasis()*m_refs[1];
-	m_drift		=	NormalizeAny(btCross(m_axis[1],m_axis[0]));
-	m_drift		*=	btMin(maxdrift,btAcos(Clamp<btScalar>(btDot(m_axis[0],m_axis[1]),-1,+1)));
-	m_drift		*=	m_erp/dt;
-	m_massmatrix=	AngularImpulseMatrix(m_bodies[0].invWorldInertia(),m_bodies[1].invWorldInertia());
-	if(m_split>0)
+	Joint::Prepare(dt, iterations);
+	m_axis[0] = m_bodies[0].xform().getBasis() * m_refs[0];
+	m_axis[1] = m_bodies[1].xform().getBasis() * m_refs[1];
+	m_drift = NormalizeAny(btCross(m_axis[1], m_axis[0]));
+	m_drift *= btMin(maxdrift, btAcos(Clamp<btScalar>(btDot(m_axis[0], m_axis[1]), -1, +1)));
+	m_drift *= m_erp / dt;
+	m_massmatrix = AngularImpulseMatrix(m_bodies[0].invWorldInertia(), m_bodies[1].invWorldInertia());
+	if (m_split > 0)
 	{
-		m_sdrift	=	m_massmatrix*(m_drift*m_split);
-		m_drift		*=	1-m_split;
+		m_sdrift = m_massmatrix * (m_drift * m_split);
+		m_drift *= 1 - m_split;
 	}
-	m_drift	/=(btScalar)iterations;
+	m_drift /= (btScalar)iterations;
 }
 
 //
-void				btSoftBody::AJoint::Solve(btScalar dt,btScalar sor)
+void btSoftBody::AJoint::Solve(btScalar dt, btScalar sor)
 {
-	const btVector3		va=m_bodies[0].angularVelocity();
-	const btVector3		vb=m_bodies[1].angularVelocity();
-	const btVector3		vr=va-vb;
-	const btScalar		sp=btDot(vr,m_axis[0]);
-	const btVector3		vc=vr-m_axis[0]*m_icontrol->Speed(this,sp);
-	btSoftBody::Impulse	impulse;
-	impulse.m_asVelocity	=	1;
-	impulse.m_velocity		=	m_massmatrix*(m_drift+vc*m_cfm)*sor;
+	const btVector3 va = m_bodies[0].angularVelocity();
+	const btVector3 vb = m_bodies[1].angularVelocity();
+	const btVector3 vr = va - vb;
+	const btScalar sp = btDot(vr, m_axis[0]);
+	const btVector3 vc = vr - m_axis[0] * m_icontrol->Speed(this, sp);
+	btSoftBody::Impulse impulse;
+	impulse.m_asVelocity = 1;
+	impulse.m_velocity = m_massmatrix * (m_drift + vc * m_cfm) * sor;
 	m_bodies[0].applyAImpulse(-impulse);
-	m_bodies[1].applyAImpulse( impulse);
+	m_bodies[1].applyAImpulse(impulse);
 }
 
 //
-void				btSoftBody::AJoint::Terminate(btScalar dt)
+void btSoftBody::AJoint::Terminate(btScalar dt)
 {
-	if(m_split>0)
+	if (m_split > 0)
 	{
 		m_bodies[0].applyDAImpulse(-m_sdrift);
-		m_bodies[1].applyDAImpulse( m_sdrift);
+		m_bodies[1].applyDAImpulse(m_sdrift);
 	}
 }
 
 //
-void				btSoftBody::CJoint::Prepare(btScalar dt,int iterations)
+void btSoftBody::CJoint::Prepare(btScalar dt, int iterations)
 {
-	Joint::Prepare(dt,iterations);
-	const bool	dodrift=(m_life==0);
-	m_delete=(++m_life)>m_maxlife;
-	if(dodrift)
+	Joint::Prepare(dt, iterations);
+	const bool dodrift = (m_life == 0);
+	m_delete = (++m_life) > m_maxlife;
+	if (dodrift)
 	{
-		m_drift=m_drift*m_erp/dt;
-		if(m_split>0)
+		m_drift = m_drift * m_erp / dt;
+		if (m_split > 0)
 		{
-			m_sdrift	=	m_massmatrix*(m_drift*m_split);
-			m_drift		*=	1-m_split;
+			m_sdrift = m_massmatrix * (m_drift * m_split);
+			m_drift *= 1 - m_split;
 		}
-		m_drift/=(btScalar)iterations;
+		m_drift /= (btScalar)iterations;
 	}
 	else
 	{
-		m_drift=m_sdrift=btVector3(0,0,0);
+		m_drift = m_sdrift = btVector3(0, 0, 0);
 	}
 }
 
 //
-void				btSoftBody::CJoint::Solve(btScalar dt,btScalar sor)
+void btSoftBody::CJoint::Solve(btScalar dt, btScalar sor)
 {
-	const btVector3		va=m_bodies[0].velocity(m_rpos[0]);
-	const btVector3		vb=m_bodies[1].velocity(m_rpos[1]);
-	const btVector3		vrel=va-vb;
-	const btScalar		rvac=btDot(vrel,m_normal);
-	btSoftBody::Impulse	impulse;
-	impulse.m_asVelocity	=	1;
-	impulse.m_velocity		=	m_drift;
-	if(rvac<0)
+	const btVector3 va = m_bodies[0].velocity(m_rpos[0]);
+	const btVector3 vb = m_bodies[1].velocity(m_rpos[1]);
+	const btVector3 vrel = va - vb;
+	const btScalar rvac = btDot(vrel, m_normal);
+	btSoftBody::Impulse impulse;
+	impulse.m_asVelocity = 1;
+	impulse.m_velocity = m_drift;
+	if (rvac < 0)
 	{
-		const btVector3	iv=m_normal*rvac;
-		const btVector3	fv=vrel-iv;
-		impulse.m_velocity	+=	iv+fv*m_friction;
+		const btVector3 iv = m_normal * rvac;
+		const btVector3 fv = vrel - iv;
+		impulse.m_velocity += iv + fv * m_friction;
 	}
-	impulse.m_velocity=m_massmatrix*impulse.m_velocity*sor;
-	
-	if (m_bodies[0].m_soft==m_bodies[1].m_soft)
+	impulse.m_velocity = m_massmatrix * impulse.m_velocity * sor;
+
+	if (m_bodies[0].m_soft == m_bodies[1].m_soft)
 	{
-		if ((impulse.m_velocity.getX() ==impulse.m_velocity.getX())&&(impulse.m_velocity.getY() ==impulse.m_velocity.getY())&&
-			(impulse.m_velocity.getZ() ==impulse.m_velocity.getZ()))
+		if ((impulse.m_velocity.getX() == impulse.m_velocity.getX()) && (impulse.m_velocity.getY() == impulse.m_velocity.getY()) &&
+			(impulse.m_velocity.getZ() == impulse.m_velocity.getZ()))
 		{
 			if (impulse.m_asVelocity)
 			{
-				if (impulse.m_velocity.length() <m_bodies[0].m_soft->m_maxSelfCollisionImpulse)
+				if (impulse.m_velocity.length() < m_bodies[0].m_soft->m_maxSelfCollisionImpulse)
 				{
-					
-				} else
+				}
+				else
 				{
-					m_bodies[0].applyImpulse(-impulse*m_bodies[0].m_soft->m_selfCollisionImpulseFactor,m_rpos[0]);
-					m_bodies[1].applyImpulse( impulse*m_bodies[0].m_soft->m_selfCollisionImpulseFactor,m_rpos[1]);
+					m_bodies[0].applyImpulse(-impulse * m_bodies[0].m_soft->m_selfCollisionImpulseFactor, m_rpos[0]);
+					m_bodies[1].applyImpulse(impulse * m_bodies[0].m_soft->m_selfCollisionImpulseFactor, m_rpos[1]);
 				}
 			}
 		}
-	} else
+	}
+	else
 	{
-		m_bodies[0].applyImpulse(-impulse,m_rpos[0]);
-		m_bodies[1].applyImpulse( impulse,m_rpos[1]);
+		m_bodies[0].applyImpulse(-impulse, m_rpos[0]);
+		m_bodies[1].applyImpulse(impulse, m_rpos[1]);
 	}
 }
 
 //
-void				btSoftBody::CJoint::Terminate(btScalar dt)
+void btSoftBody::CJoint::Terminate(btScalar dt)
 {
-	if(m_split>0)
+	if (m_split > 0)
 	{
-		m_bodies[0].applyDImpulse(-m_sdrift,m_rpos[0]);
-		m_bodies[1].applyDImpulse( m_sdrift,m_rpos[1]);
+		m_bodies[0].applyDImpulse(-m_sdrift, m_rpos[0]);
+		m_bodies[1].applyDImpulse(m_sdrift, m_rpos[1]);
 	}
 }
 
 //
-void				btSoftBody::applyForces()
+void btSoftBody::applyForces()
 {
-
 	BT_PROFILE("SoftBody applyForces");
-//	const btScalar					dt =			m_sst.sdt;
-	const btScalar					kLF =			m_cfg.kLF;
-	const btScalar					kDG =			m_cfg.kDG;
-	const btScalar					kPR =			m_cfg.kPR;
-	const btScalar					kVC =			m_cfg.kVC;
-	const bool						as_lift =		kLF>0;
-	const bool						as_drag =		kDG>0;
-	const bool						as_pressure =	kPR!=0;
-	const bool						as_volume =		kVC>0;
-	const bool						as_aero =		as_lift	||
-													as_drag		;
+	//	const btScalar					dt =			m_sst.sdt;
+	const btScalar kLF = m_cfg.kLF;
+	const btScalar kDG = m_cfg.kDG;
+	const btScalar kPR = m_cfg.kPR;
+	const btScalar kVC = m_cfg.kVC;
+	const bool as_lift = kLF > 0;
+	const bool as_drag = kDG > 0;
+	const bool as_pressure = kPR != 0;
+	const bool as_volume = kVC > 0;
+	const bool as_aero = as_lift ||
+						 as_drag;
 	//const bool						as_vaero =		as_aero	&&
 	//												(m_cfg.aeromodel < btSoftBody::eAeroModel::F_TwoSided);
 	//const bool						as_faero =		as_aero	&&
 	//												(m_cfg.aeromodel >= btSoftBody::eAeroModel::F_TwoSided);
-	const bool						use_medium =	as_aero;
-	const bool						use_volume =	as_pressure	||
-		as_volume	;
-	btScalar						volume =		0;
-	btScalar						ivolumetp =		0;
-	btScalar						dvolumetv =		0;
-	btSoftBody::sMedium	medium;
-	if(use_volume)
+	const bool use_medium = as_aero;
+	const bool use_volume = as_pressure ||
+							as_volume;
+	btScalar volume = 0;
+	btScalar ivolumetp = 0;
+	btScalar dvolumetv = 0;
+	btSoftBody::sMedium medium;
+	if (use_volume)
 	{
-		volume		=	getVolume();
-		ivolumetp	=	1/btFabs(volume)*kPR;
-		dvolumetv	=	(m_pose.m_volume-volume)*kVC;
+		volume = getVolume();
+		ivolumetp = 1 / btFabs(volume) * kPR;
+		dvolumetv = (m_pose.m_volume - volume) * kVC;
 	}
-	/* Per vertex forces			*/ 
-	int i,ni;
+	/* Per vertex forces			*/
+	int i, ni;
 
-	for(i=0,ni=m_nodes.size();i<ni;++i)
+	for (i = 0, ni = m_nodes.size(); i < ni; ++i)
 	{
-		btSoftBody::Node&	n=m_nodes[i];
-		if(n.m_im>0)
+		btSoftBody::Node& n = m_nodes[i];
+		if (n.m_im > 0)
 		{
-			if(use_medium)
+			if (use_medium)
 			{
-				/* Aerodynamics			*/ 
+				/* Aerodynamics			*/
 				addAeroForceToNode(m_windVelocity, i);
 			}
-			/* Pressure				*/ 
-			if(as_pressure)
+			/* Pressure				*/
+			if (as_pressure)
 			{
-				n.m_f	+=	n.m_n*(n.m_area*ivolumetp);
+				n.m_f += n.m_n * (n.m_area * ivolumetp);
 			}
-			/* Volume				*/ 
-			if(as_volume)
+			/* Volume				*/
+			if (as_volume)
 			{
-				n.m_f	+=	n.m_n*(n.m_area*dvolumetv);
+				n.m_f += n.m_n * (n.m_area * dvolumetv);
 			}
 		}
 	}
 
-	/* Per face forces				*/ 
-	for(i=0,ni=m_faces.size();i<ni;++i)
+	/* Per face forces				*/
+	for (i = 0, ni = m_faces.size(); i < ni; ++i)
 	{
-	//	btSoftBody::Face&	f=m_faces[i];
+		//	btSoftBody::Face&	f=m_faces[i];
 
-		/* Aerodynamics			*/ 
-		addAeroForceToFace(m_windVelocity, i);	
+		/* Aerodynamics			*/
+		addAeroForceToFace(m_windVelocity, i);
 	}
 }
 
 //
-void				btSoftBody::PSolve_Anchors(btSoftBody* psb,btScalar kst,btScalar ti)
+void btSoftBody::PSolve_Anchors(btSoftBody* psb, btScalar kst, btScalar ti)
 {
 	BT_PROFILE("PSolve_Anchors");
-	const btScalar	kAHR=psb->m_cfg.kAHR*kst;
-	const btScalar	dt=psb->m_sst.sdt;
-	for(int i=0,ni=psb->m_anchors.size();i<ni;++i)
+	const btScalar kAHR = psb->m_cfg.kAHR * kst;
+	const btScalar dt = psb->m_sst.sdt;
+	for (int i = 0, ni = psb->m_anchors.size(); i < ni; ++i)
 	{
-		const Anchor&		a=psb->m_anchors[i];
-		const btTransform&	t=a.m_body->getWorldTransform();
-		Node&				n=*a.m_node;
-		const btVector3		wa=t*a.m_local;
-		const btVector3		va=a.m_body->getVelocityInLocalPoint(a.m_c1)*dt;
-		const btVector3		vb=n.m_x-n.m_q;
-		const btVector3		vr=(va-vb)+(wa-n.m_x)*kAHR;
-		const btVector3		impulse=a.m_c0*vr*a.m_influence;
-		n.m_x+=impulse*a.m_c2;
-		a.m_body->applyImpulse(-impulse,a.m_c1);
+		const Anchor& a = psb->m_anchors[i];
+		const btTransform& t = a.m_body->getWorldTransform();
+		Node& n = *a.m_node;
+		const btVector3 wa = t * a.m_local;
+		const btVector3 va = a.m_body->getVelocityInLocalPoint(a.m_c1) * dt;
+		const btVector3 vb = n.m_x - n.m_q;
+		const btVector3 vr = (va - vb) + (wa - n.m_x) * kAHR;
+		const btVector3 impulse = a.m_c0 * vr * a.m_influence;
+		n.m_x += impulse * a.m_c2;
+		a.m_body->applyImpulse(-impulse, a.m_c1);
 	}
 }
 
-
 //
 void btSoftBody::PSolve_RContacts(btSoftBody* psb, btScalar kst, btScalar ti)
 {
 	BT_PROFILE("PSolve_RContacts");
-	const btScalar	dt = psb->m_sst.sdt;
-	const btScalar	mrg = psb->getCollisionShape()->getMargin();
+	const btScalar dt = psb->m_sst.sdt;
+	const btScalar mrg = psb->getCollisionShape()->getMargin();
 	btMultiBodyJacobianData jacobianData;
-	for(int i=0,ni=psb->m_rcontacts.size();i<ni;++i)
+	for (int i = 0, ni = psb->m_rcontacts.size(); i < ni; ++i)
 	{
-		const RContact&		c = psb->m_rcontacts[i];
-		const sCti&			cti = c.m_cti;	
-		if (cti.m_colObj->hasContactResponse()) 
+		const RContact& c = psb->m_rcontacts[i];
+		const sCti& cti = c.m_cti;
+		if (cti.m_colObj->hasContactResponse())
 		{
-            btVector3 va(0,0,0);
-            btRigidBody* rigidCol=0;
-            btMultiBodyLinkCollider* multibodyLinkCol=0;
-            btScalar* deltaV;
-            
-            if (cti.m_colObj->getInternalType() == btCollisionObject::CO_RIGID_BODY)
-            {
-                rigidCol = (btRigidBody*)btRigidBody::upcast(cti.m_colObj);
-                va = rigidCol ? rigidCol->getVelocityInLocalPoint(c.m_c1)*dt : btVector3(0,0,0);
-            }
-			else if (cti.m_colObj->getInternalType() == btCollisionObject::CO_FEATHERSTONE_LINK)
-            {
-                multibodyLinkCol = (btMultiBodyLinkCollider*)btMultiBodyLinkCollider::upcast(cti.m_colObj);
-                if (multibodyLinkCol)
-                {
-                    const int ndof  = multibodyLinkCol->m_multiBody->getNumDofs() + 6;
-                    jacobianData.m_jacobians.resize(ndof);
-                    jacobianData.m_deltaVelocitiesUnitImpulse.resize(ndof);
-                    btScalar* jac=&jacobianData.m_jacobians[0];
-                    
-                    multibodyLinkCol->m_multiBody->fillContactJacobianMultiDof(multibodyLinkCol->m_link, c.m_node->m_x, cti.m_normal, jac, jacobianData.scratch_r, jacobianData.scratch_v, jacobianData.scratch_m);
-                    deltaV = &jacobianData.m_deltaVelocitiesUnitImpulse[0];
-                    multibodyLinkCol->m_multiBody->calcAccelerationDeltasMultiDof(&jacobianData.m_jacobians[0],deltaV,jacobianData.scratch_r, jacobianData.scratch_v);
-                    
-                    btScalar vel = 0.0;
-                    for (int j = 0; j < ndof ; ++j) {
-                        vel += multibodyLinkCol->m_multiBody->getVelocityVector()[j] * jac[j];
-                    }
-                    va = cti.m_normal*vel*dt;
-                }
-            }
-            
-			const btVector3		vb = c.m_node->m_x-c.m_node->m_q;
-			const btVector3		vr = vb-va;
-			const btScalar		dn = btDot(vr, cti.m_normal);		
-			if(dn<=SIMD_EPSILON)
+			btVector3 va(0, 0, 0);
+			btRigidBody* rigidCol = 0;
+			btMultiBodyLinkCollider* multibodyLinkCol = 0;
+			btScalar* deltaV;
+
+			if (cti.m_colObj->getInternalType() == btCollisionObject::CO_RIGID_BODY)
 			{
-				const btScalar		dp = btMin( (btDot(c.m_node->m_x, cti.m_normal) + cti.m_offset), mrg );
-				const btVector3		fv = vr - (cti.m_normal * dn);
+				rigidCol = (btRigidBody*)btRigidBody::upcast(cti.m_colObj);
+				va = rigidCol ? rigidCol->getVelocityInLocalPoint(c.m_c1) * dt : btVector3(0, 0, 0);
+			}
+			else if (cti.m_colObj->getInternalType() == btCollisionObject::CO_FEATHERSTONE_LINK)
+			{
+				multibodyLinkCol = (btMultiBodyLinkCollider*)btMultiBodyLinkCollider::upcast(cti.m_colObj);
+				if (multibodyLinkCol)
+				{
+					const int ndof = multibodyLinkCol->m_multiBody->getNumDofs() + 6;
+					jacobianData.m_jacobians.resize(ndof);
+					jacobianData.m_deltaVelocitiesUnitImpulse.resize(ndof);
+					btScalar* jac = &jacobianData.m_jacobians[0];
+
+					multibodyLinkCol->m_multiBody->fillContactJacobianMultiDof(multibodyLinkCol->m_link, c.m_node->m_x, cti.m_normal, jac, jacobianData.scratch_r, jacobianData.scratch_v, jacobianData.scratch_m);
+					deltaV = &jacobianData.m_deltaVelocitiesUnitImpulse[0];
+					multibodyLinkCol->m_multiBody->calcAccelerationDeltasMultiDof(&jacobianData.m_jacobians[0], deltaV, jacobianData.scratch_r, jacobianData.scratch_v);
+
+					btScalar vel = 0.0;
+					for (int j = 0; j < ndof; ++j)
+					{
+						vel += multibodyLinkCol->m_multiBody->getVelocityVector()[j] * jac[j];
+					}
+					va = cti.m_normal * vel * dt;
+				}
+			}
+
+			const btVector3 vb = c.m_node->m_x - c.m_node->m_q;
+			const btVector3 vr = vb - va;
+			const btScalar dn = btDot(vr, cti.m_normal);
+			if (dn <= SIMD_EPSILON)
+			{
+				const btScalar dp = btMin((btDot(c.m_node->m_x, cti.m_normal) + cti.m_offset), mrg);
+				const btVector3 fv = vr - (cti.m_normal * dn);
 				// c0 is the impulse matrix, c3 is 1 - the friction coefficient or 0, c4 is the contact hardness coefficient
-				const btVector3		impulse = c.m_c0 * ( (vr - (fv * c.m_c3) + (cti.m_normal * (dp * c.m_c4))) * kst );
+				const btVector3 impulse = c.m_c0 * ((vr - (fv * c.m_c3) + (cti.m_normal * (dp * c.m_c4))) * kst);
 				c.m_node->m_x -= impulse * c.m_c2;
-                
-                if (cti.m_colObj->getInternalType() == btCollisionObject::CO_RIGID_BODY)
-                {
-                    if (rigidCol)
-                        rigidCol->applyImpulse(impulse,c.m_c1);
-                }
-                else if (cti.m_colObj->getInternalType() == btCollisionObject::CO_FEATHERSTONE_LINK)
-                {
-                    if (multibodyLinkCol)
-                    {
-                        double multiplier = 0.5;
-                        multibodyLinkCol->m_multiBody->applyDeltaVeeMultiDof(deltaV,-impulse.length()*multiplier);
-                    }
-                }
+
+				if (cti.m_colObj->getInternalType() == btCollisionObject::CO_RIGID_BODY)
+				{
+					if (rigidCol)
+						rigidCol->applyImpulse(impulse, c.m_c1);
+				}
+				else if (cti.m_colObj->getInternalType() == btCollisionObject::CO_FEATHERSTONE_LINK)
+				{
+					if (multibodyLinkCol)
+					{
+						double multiplier = 0.5;
+						multibodyLinkCol->m_multiBody->applyDeltaVeeMultiDof(deltaV, -impulse.length() * multiplier);
+					}
+				}
 			}
 		}
 	}
 }
 
 //
-void				btSoftBody::PSolve_SContacts(btSoftBody* psb,btScalar,btScalar ti)
+void btSoftBody::PSolve_SContacts(btSoftBody* psb, btScalar, btScalar ti)
 {
 	BT_PROFILE("PSolve_SContacts");
-	
-	for(int i=0,ni=psb->m_scontacts.size();i<ni;++i)
+
+	for (int i = 0, ni = psb->m_scontacts.size(); i < ni; ++i)
 	{
-		const SContact&		c=psb->m_scontacts[i];
-		const btVector3&	nr=c.m_normal;
-		Node&				n=*c.m_node;
-		Face&				f=*c.m_face;
-		const btVector3		p=BaryEval(	f.m_n[0]->m_x,
-			f.m_n[1]->m_x,
-			f.m_n[2]->m_x,
-			c.m_weights);
-		const btVector3		q=BaryEval(	f.m_n[0]->m_q,
-			f.m_n[1]->m_q,
-			f.m_n[2]->m_q,
-			c.m_weights);											
-		const btVector3		vr=(n.m_x-n.m_q)-(p-q);
-		btVector3			corr(0,0,0);
-		btScalar dot = btDot(vr,nr);
-		if(dot<0)
+		const SContact& c = psb->m_scontacts[i];
+		const btVector3& nr = c.m_normal;
+		Node& n = *c.m_node;
+		Face& f = *c.m_face;
+		const btVector3 p = BaryEval(f.m_n[0]->m_x,
+									 f.m_n[1]->m_x,
+									 f.m_n[2]->m_x,
+									 c.m_weights);
+		const btVector3 q = BaryEval(f.m_n[0]->m_q,
+									 f.m_n[1]->m_q,
+									 f.m_n[2]->m_q,
+									 c.m_weights);
+		const btVector3 vr = (n.m_x - n.m_q) - (p - q);
+		btVector3 corr(0, 0, 0);
+		btScalar dot = btDot(vr, nr);
+		if (dot < 0)
 		{
-			const btScalar	j=c.m_margin-(btDot(nr,n.m_x)-btDot(nr,p));
-			corr+=c.m_normal*j;
+			const btScalar j = c.m_margin - (btDot(nr, n.m_x) - btDot(nr, p));
+			corr += c.m_normal * j;
 		}
-		corr			-=	ProjectOnPlane(vr,nr)*c.m_friction;
-		n.m_x			+=	corr*c.m_cfm[0];
-		f.m_n[0]->m_x	-=	corr*(c.m_cfm[1]*c.m_weights.x());
-		f.m_n[1]->m_x	-=	corr*(c.m_cfm[1]*c.m_weights.y());
-		f.m_n[2]->m_x	-=	corr*(c.m_cfm[1]*c.m_weights.z());
+		corr -= ProjectOnPlane(vr, nr) * c.m_friction;
+		n.m_x += corr * c.m_cfm[0];
+		f.m_n[0]->m_x -= corr * (c.m_cfm[1] * c.m_weights.x());
+		f.m_n[1]->m_x -= corr * (c.m_cfm[1] * c.m_weights.y());
+		f.m_n[2]->m_x -= corr * (c.m_cfm[1] * c.m_weights.z());
 	}
 }
 
 //
-void				btSoftBody::PSolve_Links(btSoftBody* psb,btScalar kst,btScalar ti)
+void btSoftBody::PSolve_Links(btSoftBody* psb, btScalar kst, btScalar ti)
 {
-BT_PROFILE("PSolve_Links");
-	for(int i=0,ni=psb->m_links.size();i<ni;++i)
-	{			
-		Link&	l=psb->m_links[i];
-		if(l.m_c0>0)
+	BT_PROFILE("PSolve_Links");
+	for (int i = 0, ni = psb->m_links.size(); i < ni; ++i)
+	{
+		Link& l = psb->m_links[i];
+		if (l.m_c0 > 0)
 		{
-			Node&			a=*l.m_n[0];
-			Node&			b=*l.m_n[1];
-			const btVector3	del=b.m_x-a.m_x;
-			const btScalar	len=del.length2();
-			if (l.m_c1+len > SIMD_EPSILON)
+			Node& a = *l.m_n[0];
+			Node& b = *l.m_n[1];
+			const btVector3 del = b.m_x - a.m_x;
+			const btScalar len = del.length2();
+			if (l.m_c1 + len > SIMD_EPSILON)
 			{
-				const btScalar	k=((l.m_c1-len)/(l.m_c0*(l.m_c1+len)))*kst;
-				a.m_x-=del*(k*a.m_im);
-				b.m_x+=del*(k*b.m_im);
+				const btScalar k = ((l.m_c1 - len) / (l.m_c0 * (l.m_c1 + len))) * kst;
+				a.m_x -= del * (k * a.m_im);
+				b.m_x += del * (k * b.m_im);
 			}
 		}
 	}
 }
 
 //
-void				btSoftBody::VSolve_Links(btSoftBody* psb,btScalar kst)
+void btSoftBody::VSolve_Links(btSoftBody* psb, btScalar kst)
 {
 	BT_PROFILE("VSolve_Links");
-	for(int i=0,ni=psb->m_links.size();i<ni;++i)
-	{			
-		Link&			l=psb->m_links[i];
-		Node**			n=l.m_n;
-		const btScalar	j=-btDot(l.m_c3,n[0]->m_v-n[1]->m_v)*l.m_c2*kst;
-		n[0]->m_v+=	l.m_c3*(j*n[0]->m_im);
-		n[1]->m_v-=	l.m_c3*(j*n[1]->m_im);
+	for (int i = 0, ni = psb->m_links.size(); i < ni; ++i)
+	{
+		Link& l = psb->m_links[i];
+		Node** n = l.m_n;
+		const btScalar j = -btDot(l.m_c3, n[0]->m_v - n[1]->m_v) * l.m_c2 * kst;
+		n[0]->m_v += l.m_c3 * (j * n[0]->m_im);
+		n[1]->m_v -= l.m_c3 * (j * n[1]->m_im);
 	}
 }
 
 //
-btSoftBody::psolver_t	btSoftBody::getSolver(ePSolver::_ solver)
+btSoftBody::psolver_t btSoftBody::getSolver(ePSolver::_ solver)
 {
-	switch(solver)
+	switch (solver)
 	{
-	case	ePSolver::Anchors:		
-		return(&btSoftBody::PSolve_Anchors);
-	case	ePSolver::Linear:		
-		return(&btSoftBody::PSolve_Links);
-	case	ePSolver::RContacts:	
-		return(&btSoftBody::PSolve_RContacts);
-	case	ePSolver::SContacts:	
-		return(&btSoftBody::PSolve_SContacts);	
+		case ePSolver::Anchors:
+			return (&btSoftBody::PSolve_Anchors);
+		case ePSolver::Linear:
+			return (&btSoftBody::PSolve_Links);
+		case ePSolver::RContacts:
+			return (&btSoftBody::PSolve_RContacts);
+		case ePSolver::SContacts:
+			return (&btSoftBody::PSolve_SContacts);
 		default:
 		{
 		}
 	}
-	return(0);
+	return (0);
 }
 
 //
-btSoftBody::vsolver_t	btSoftBody::getSolver(eVSolver::_ solver)
+btSoftBody::vsolver_t btSoftBody::getSolver(eVSolver::_ solver)
 {
-	switch(solver)
+	switch (solver)
 	{
-	case	eVSolver::Linear:		return(&btSoftBody::VSolve_Links);
+		case eVSolver::Linear:
+			return (&btSoftBody::VSolve_Links);
 		default:
 		{
 		}
 	}
-	return(0);
+	return (0);
 }
 
 //
-void			btSoftBody::defaultCollisionHandler(const btCollisionObjectWrapper* pcoWrap)
+void btSoftBody::defaultCollisionHandler(const btCollisionObjectWrapper* pcoWrap)
 {
-
-	switch(m_cfg.collisions&fCollision::RVSmask)
+	switch (m_cfg.collisions & fCollision::RVSmask)
 	{
-	case	fCollision::SDF_RS:
+		case fCollision::SDF_RS:
 		{
-			btSoftColliders::CollideSDF_RS	docollide;		
-			btRigidBody*		prb1=(btRigidBody*) btRigidBody::upcast(pcoWrap->getCollisionObject());
-			btTransform	wtr=pcoWrap->getWorldTransform();
+			btSoftColliders::CollideSDF_RS docollide;
+			btRigidBody* prb1 = (btRigidBody*)btRigidBody::upcast(pcoWrap->getCollisionObject());
+			btTransform wtr = pcoWrap->getWorldTransform();
 
-			const btTransform	ctr=pcoWrap->getWorldTransform();
-			const btScalar		timemargin=(wtr.getOrigin()-ctr.getOrigin()).length();
-			const btScalar		basemargin=getCollisionShape()->getMargin();
-			btVector3			mins;
-			btVector3			maxs;
-			ATTRIBUTE_ALIGNED16(btDbvtVolume)		volume;
-			pcoWrap->getCollisionShape()->getAabb(	pcoWrap->getWorldTransform(),
-				mins,
-				maxs);
-			volume=btDbvtVolume::FromMM(mins,maxs);
-			volume.Expand(btVector3(basemargin,basemargin,basemargin));		
-			docollide.psb		=	this;
+			const btTransform ctr = pcoWrap->getWorldTransform();
+			const btScalar timemargin = (wtr.getOrigin() - ctr.getOrigin()).length();
+			const btScalar basemargin = getCollisionShape()->getMargin();
+			btVector3 mins;
+			btVector3 maxs;
+			ATTRIBUTE_ALIGNED16(btDbvtVolume)
+			volume;
+			pcoWrap->getCollisionShape()->getAabb(pcoWrap->getWorldTransform(),
+												  mins,
+												  maxs);
+			volume = btDbvtVolume::FromMM(mins, maxs);
+			volume.Expand(btVector3(basemargin, basemargin, basemargin));
+			docollide.psb = this;
 			docollide.m_colObj1Wrap = pcoWrap;
 			docollide.m_rigidBody = prb1;
 
-			docollide.dynmargin	=	basemargin+timemargin;
-			docollide.stamargin	=	basemargin;
-			m_ndbvt.collideTV(m_ndbvt.m_root,volume,docollide);
+			docollide.dynmargin = basemargin + timemargin;
+			docollide.stamargin = basemargin;
+			m_ndbvt.collideTV(m_ndbvt.m_root, volume, docollide);
 		}
 		break;
-	case	fCollision::CL_RS:
+		case fCollision::CL_RS:
 		{
-			btSoftColliders::CollideCL_RS	collider;
-			collider.ProcessColObj(this,pcoWrap);
+			btSoftColliders::CollideCL_RS collider;
+			collider.ProcessColObj(this, pcoWrap);
 		}
 		break;
 	}
 }
 
 //
-void			btSoftBody::defaultCollisionHandler(btSoftBody* psb)
+void btSoftBody::defaultCollisionHandler(btSoftBody* psb)
 {
-	const int cf=m_cfg.collisions&psb->m_cfg.collisions;
-	switch(cf&fCollision::SVSmask)
+	const int cf = m_cfg.collisions & psb->m_cfg.collisions;
+	switch (cf & fCollision::SVSmask)
 	{
-	case	fCollision::CL_SS:
+		case fCollision::CL_SS:
 		{
-			
 			//support self-collision if CL_SELF flag set
-			if (this!=psb || psb->m_cfg.collisions&fCollision::CL_SELF)
+			if (this != psb || psb->m_cfg.collisions & fCollision::CL_SELF)
 			{
-				btSoftColliders::CollideCL_SS	docollide;
-				docollide.ProcessSoftSoft(this,psb);
+				btSoftColliders::CollideCL_SS docollide;
+				docollide.ProcessSoftSoft(this, psb);
 			}
-			
 		}
 		break;
-	case	fCollision::VF_SS:
+		case fCollision::VF_SS:
 		{
 			//only self-collision for Cluster, not Vertex-Face yet
-			if (this!=psb)
+			if (this != psb)
 			{
-				btSoftColliders::CollideVF_SS	docollide;
-				/* common					*/ 
-				docollide.mrg=	getCollisionShape()->getMargin()+
-					psb->getCollisionShape()->getMargin();
-				/* psb0 nodes vs psb1 faces	*/ 
-				docollide.psb[0]=this;
-				docollide.psb[1]=psb;
-				docollide.psb[0]->m_ndbvt.collideTT(	docollide.psb[0]->m_ndbvt.m_root,
-					docollide.psb[1]->m_fdbvt.m_root,
-					docollide);
-				/* psb1 nodes vs psb0 faces	*/ 
-				docollide.psb[0]=psb;
-				docollide.psb[1]=this;
-				docollide.psb[0]->m_ndbvt.collideTT(	docollide.psb[0]->m_ndbvt.m_root,
-					docollide.psb[1]->m_fdbvt.m_root,
-					docollide);
+				btSoftColliders::CollideVF_SS docollide;
+				/* common					*/
+				docollide.mrg = getCollisionShape()->getMargin() +
+								psb->getCollisionShape()->getMargin();
+				/* psb0 nodes vs psb1 faces	*/
+				docollide.psb[0] = this;
+				docollide.psb[1] = psb;
+				docollide.psb[0]->m_ndbvt.collideTT(docollide.psb[0]->m_ndbvt.m_root,
+													docollide.psb[1]->m_fdbvt.m_root,
+													docollide);
+				/* psb1 nodes vs psb0 faces	*/
+				docollide.psb[0] = psb;
+				docollide.psb[1] = this;
+				docollide.psb[0]->m_ndbvt.collideTT(docollide.psb[0]->m_ndbvt.m_root,
+													docollide.psb[1]->m_fdbvt.m_root,
+													docollide);
 			}
 		}
 		break;
-	default:
+		default:
 		{
-			
 		}
 	}
 }
 
-
-
-void btSoftBody::setWindVelocity( const btVector3 &velocity )
+void btSoftBody::setWindVelocity(const btVector3& velocity)
 {
 	m_windVelocity = velocity;
 }
 
-
 const btVector3& btSoftBody::getWindVelocity()
 {
 	return m_windVelocity;
 }
 
-
-
-int	btSoftBody::calculateSerializeBufferSize()	const
+int btSoftBody::calculateSerializeBufferSize() const
 {
 	int sz = sizeof(btSoftBodyData);
 	return sz;
 }
 
-	///fills the dataBuffer and returns the struct name (and 0 on failure)
-const char*	btSoftBody::serialize(void* dataBuffer, class btSerializer* serializer) const
+///fills the dataBuffer and returns the struct name (and 0 on failure)
+const char* btSoftBody::serialize(void* dataBuffer, class btSerializer* serializer) const
 {
-	btSoftBodyData* sbd = (btSoftBodyData*) dataBuffer;
+	btSoftBodyData* sbd = (btSoftBodyData*)dataBuffer;
 
 	btCollisionObject::serialize(&sbd->m_collisionObjectData, serializer);
 
-	btHashMap<btHashPtr,int>	m_nodeIndexMap;
+	btHashMap<btHashPtr, int> m_nodeIndexMap;
 
 	sbd->m_numMaterials = m_materials.size();
-	sbd->m_materials = sbd->m_numMaterials? (SoftBodyMaterialData**) serializer->getUniquePointer((void*)&m_materials): 0;
+	sbd->m_materials = sbd->m_numMaterials ? (SoftBodyMaterialData**)serializer->getUniquePointer((void*)&m_materials) : 0;
 
 	if (sbd->m_materials)
 	{
 		int sz = sizeof(SoftBodyMaterialData*);
 		int numElem = sbd->m_numMaterials;
-		btChunk* chunk = serializer->allocate(sz,numElem);
+		btChunk* chunk = serializer->allocate(sz, numElem);
 		//SoftBodyMaterialData** memPtr = chunk->m_oldPtr;
 		SoftBodyMaterialData** memPtr = (SoftBodyMaterialData**)chunk->m_oldPtr;
-		for (int i=0;i<numElem;i++,memPtr++)
+		for (int i = 0; i < numElem; i++, memPtr++)
 		{
 			btSoftBody::Material* mat = m_materials[i];
 			*memPtr = mat ? (SoftBodyMaterialData*)serializer->getUniquePointer((void*)mat) : 0;
 			if (!serializer->findPointer(mat))
 			{
 				//serialize it here
-				btChunk* chunk = serializer->allocate(sizeof(SoftBodyMaterialData),1);
+				btChunk* chunk = serializer->allocate(sizeof(SoftBodyMaterialData), 1);
 				SoftBodyMaterialData* memPtr = (SoftBodyMaterialData*)chunk->m_oldPtr;
 				memPtr->m_flags = mat->m_flags;
 				memPtr->m_angularStiffness = mat->m_kAST;
 				memPtr->m_linearStiffness = mat->m_kLST;
 				memPtr->m_volumeStiffness = mat->m_kVST;
-				serializer->finalizeChunk(chunk,"SoftBodyMaterialData",BT_SBMATERIAL_CODE,mat);
+				serializer->finalizeChunk(chunk, "SoftBodyMaterialData", BT_SBMATERIAL_CODE, mat);
 			}
 		}
-		serializer->finalizeChunk(chunk,"SoftBodyMaterialData",BT_ARRAY_CODE,(void*) &m_materials);
+		serializer->finalizeChunk(chunk, "SoftBodyMaterialData", BT_ARRAY_CODE, (void*)&m_materials);
 	}
 
-
-	
-
 	sbd->m_numNodes = m_nodes.size();
-	sbd->m_nodes = sbd->m_numNodes ? (SoftBodyNodeData*)serializer->getUniquePointer((void*)&m_nodes): 0;
+	sbd->m_nodes = sbd->m_numNodes ? (SoftBodyNodeData*)serializer->getUniquePointer((void*)&m_nodes) : 0;
 	if (sbd->m_nodes)
 	{
 		int sz = sizeof(SoftBodyNodeData);
 		int numElem = sbd->m_numNodes;
-		btChunk* chunk = serializer->allocate(sz,numElem);
+		btChunk* chunk = serializer->allocate(sz, numElem);
 		SoftBodyNodeData* memPtr = (SoftBodyNodeData*)chunk->m_oldPtr;
-		for (int i=0;i<numElem;i++,memPtr++)
+		for (int i = 0; i < numElem; i++, memPtr++)
 		{
-			m_nodes[i].m_f.serializeFloat( memPtr->m_accumulatedForce);
+			m_nodes[i].m_f.serializeFloat(memPtr->m_accumulatedForce);
 			memPtr->m_area = m_nodes[i].m_area;
 			memPtr->m_attach = m_nodes[i].m_battach;
 			memPtr->m_inverseMass = m_nodes[i].m_im;
-			memPtr->m_material = m_nodes[i].m_material? (SoftBodyMaterialData*)serializer->getUniquePointer((void*) m_nodes[i].m_material):0;
+			memPtr->m_material = m_nodes[i].m_material ? (SoftBodyMaterialData*)serializer->getUniquePointer((void*)m_nodes[i].m_material) : 0;
 			m_nodes[i].m_n.serializeFloat(memPtr->m_normal);
 			m_nodes[i].m_x.serializeFloat(memPtr->m_position);
 			m_nodes[i].m_q.serializeFloat(memPtr->m_previousPosition);
 			m_nodes[i].m_v.serializeFloat(memPtr->m_velocity);
-			m_nodeIndexMap.insert(&m_nodes[i],i);
+			m_nodeIndexMap.insert(&m_nodes[i], i);
 		}
-		serializer->finalizeChunk(chunk,"SoftBodyNodeData",BT_SBNODE_CODE,(void*) &m_nodes);
+		serializer->finalizeChunk(chunk, "SoftBodyNodeData", BT_SBNODE_CODE, (void*)&m_nodes);
 	}
 
 	sbd->m_numLinks = m_links.size();
-	sbd->m_links = sbd->m_numLinks? (SoftBodyLinkData*) serializer->getUniquePointer((void*)&m_links[0]):0;
+	sbd->m_links = sbd->m_numLinks ? (SoftBodyLinkData*)serializer->getUniquePointer((void*)&m_links[0]) : 0;
 	if (sbd->m_links)
 	{
 		int sz = sizeof(SoftBodyLinkData);
 		int numElem = sbd->m_numLinks;
-		btChunk* chunk = serializer->allocate(sz,numElem);
+		btChunk* chunk = serializer->allocate(sz, numElem);
 		SoftBodyLinkData* memPtr = (SoftBodyLinkData*)chunk->m_oldPtr;
-		for (int i=0;i<numElem;i++,memPtr++)
+		for (int i = 0; i < numElem; i++, memPtr++)
 		{
 			memPtr->m_bbending = m_links[i].m_bbending;
-			memPtr->m_material = m_links[i].m_material? (SoftBodyMaterialData*)serializer->getUniquePointer((void*) m_links[i].m_material):0;
-			memPtr->m_nodeIndices[0] = m_links[i].m_n[0] ? m_links[i].m_n[0] - &m_nodes[0]: -1;
-			memPtr->m_nodeIndices[1] = m_links[i].m_n[1] ? m_links[i].m_n[1] - &m_nodes[0]: -1;
-			btAssert(memPtr->m_nodeIndices[0]<m_nodes.size());
-			btAssert(memPtr->m_nodeIndices[1]<m_nodes.size());
+			memPtr->m_material = m_links[i].m_material ? (SoftBodyMaterialData*)serializer->getUniquePointer((void*)m_links[i].m_material) : 0;
+			memPtr->m_nodeIndices[0] = m_links[i].m_n[0] ? m_links[i].m_n[0] - &m_nodes[0] : -1;
+			memPtr->m_nodeIndices[1] = m_links[i].m_n[1] ? m_links[i].m_n[1] - &m_nodes[0] : -1;
+			btAssert(memPtr->m_nodeIndices[0] < m_nodes.size());
+			btAssert(memPtr->m_nodeIndices[1] < m_nodes.size());
 			memPtr->m_restLength = m_links[i].m_rl;
 		}
-		serializer->finalizeChunk(chunk,"SoftBodyLinkData",BT_ARRAY_CODE,(void*) &m_links[0]);
-
+		serializer->finalizeChunk(chunk, "SoftBodyLinkData", BT_ARRAY_CODE, (void*)&m_links[0]);
 	}
 
-
 	sbd->m_numFaces = m_faces.size();
-	sbd->m_faces = sbd->m_numFaces? (SoftBodyFaceData*) serializer->getUniquePointer((void*)&m_faces[0]):0;
+	sbd->m_faces = sbd->m_numFaces ? (SoftBodyFaceData*)serializer->getUniquePointer((void*)&m_faces[0]) : 0;
 	if (sbd->m_faces)
 	{
 		int sz = sizeof(SoftBodyFaceData);
 		int numElem = sbd->m_numFaces;
-		btChunk* chunk = serializer->allocate(sz,numElem);
+		btChunk* chunk = serializer->allocate(sz, numElem);
 		SoftBodyFaceData* memPtr = (SoftBodyFaceData*)chunk->m_oldPtr;
-		for (int i=0;i<numElem;i++,memPtr++)
+		for (int i = 0; i < numElem; i++, memPtr++)
 		{
-			memPtr->m_material = m_faces[i].m_material ?  (SoftBodyMaterialData*) serializer->getUniquePointer((void*)m_faces[i].m_material): 0;
-			m_faces[i].m_normal.serializeFloat(	memPtr->m_normal);
-			for (int j=0;j<3;j++)
+			memPtr->m_material = m_faces[i].m_material ? (SoftBodyMaterialData*)serializer->getUniquePointer((void*)m_faces[i].m_material) : 0;
+			m_faces[i].m_normal.serializeFloat(memPtr->m_normal);
+			for (int j = 0; j < 3; j++)
 			{
-				memPtr->m_nodeIndices[j] = m_faces[i].m_n[j]? m_faces[i].m_n[j] - &m_nodes[0]: -1;
+				memPtr->m_nodeIndices[j] = m_faces[i].m_n[j] ? m_faces[i].m_n[j] - &m_nodes[0] : -1;
 			}
 			memPtr->m_restArea = m_faces[i].m_ra;
 		}
-		serializer->finalizeChunk(chunk,"SoftBodyFaceData",BT_ARRAY_CODE,(void*) &m_faces[0]);
+		serializer->finalizeChunk(chunk, "SoftBodyFaceData", BT_ARRAY_CODE, (void*)&m_faces[0]);
 	}
 
-
 	sbd->m_numTetrahedra = m_tetras.size();
-	sbd->m_tetrahedra = sbd->m_numTetrahedra ? (SoftBodyTetraData*) serializer->getUniquePointer((void*)&m_tetras[0]):0;
+	sbd->m_tetrahedra = sbd->m_numTetrahedra ? (SoftBodyTetraData*)serializer->getUniquePointer((void*)&m_tetras[0]) : 0;
 	if (sbd->m_tetrahedra)
 	{
 		int sz = sizeof(SoftBodyTetraData);
 		int numElem = sbd->m_numTetrahedra;
-		btChunk* chunk = serializer->allocate(sz,numElem);
+		btChunk* chunk = serializer->allocate(sz, numElem);
 		SoftBodyTetraData* memPtr = (SoftBodyTetraData*)chunk->m_oldPtr;
-		for (int i=0;i<numElem;i++,memPtr++)
+		for (int i = 0; i < numElem; i++, memPtr++)
 		{
-			for (int j=0;j<4;j++)
+			for (int j = 0; j < 4; j++)
 			{
-				m_tetras[i].m_c0[j].serializeFloat(	memPtr->m_c0[j] );
-				memPtr->m_nodeIndices[j] = m_tetras[j].m_n[j]? m_tetras[j].m_n[j]-&m_nodes[0] : -1;
+				m_tetras[i].m_c0[j].serializeFloat(memPtr->m_c0[j]);
+				memPtr->m_nodeIndices[j] = m_tetras[j].m_n[j] ? m_tetras[j].m_n[j] - &m_nodes[0] : -1;
 			}
 			memPtr->m_c1 = m_tetras[i].m_c1;
 			memPtr->m_c2 = m_tetras[i].m_c2;
-			memPtr->m_material = m_tetras[i].m_material ? (SoftBodyMaterialData*)serializer->getUniquePointer((void*) m_tetras[i].m_material): 0;
+			memPtr->m_material = m_tetras[i].m_material ? (SoftBodyMaterialData*)serializer->getUniquePointer((void*)m_tetras[i].m_material) : 0;
 			memPtr->m_restVolume = m_tetras[i].m_rv;
 		}
-		serializer->finalizeChunk(chunk,"SoftBodyTetraData",BT_ARRAY_CODE,(void*) &m_tetras[0]);
+		serializer->finalizeChunk(chunk, "SoftBodyTetraData", BT_ARRAY_CODE, (void*)&m_tetras[0]);
 	}
 
 	sbd->m_numAnchors = m_anchors.size();
-	sbd->m_anchors = sbd->m_numAnchors ? (SoftRigidAnchorData*) serializer->getUniquePointer((void*)&m_anchors[0]):0;
+	sbd->m_anchors = sbd->m_numAnchors ? (SoftRigidAnchorData*)serializer->getUniquePointer((void*)&m_anchors[0]) : 0;
 	if (sbd->m_anchors)
 	{
 		int sz = sizeof(SoftRigidAnchorData);
 		int numElem = sbd->m_numAnchors;
-		btChunk* chunk = serializer->allocate(sz,numElem);
+		btChunk* chunk = serializer->allocate(sz, numElem);
 		SoftRigidAnchorData* memPtr = (SoftRigidAnchorData*)chunk->m_oldPtr;
-		for (int i=0;i<numElem;i++,memPtr++)
+		for (int i = 0; i < numElem; i++, memPtr++)
 		{
 			m_anchors[i].m_c0.serializeFloat(memPtr->m_c0);
 			m_anchors[i].m_c1.serializeFloat(memPtr->m_c1);
 			memPtr->m_c2 = m_anchors[i].m_c2;
 			m_anchors[i].m_local.serializeFloat(memPtr->m_localFrame);
-			memPtr->m_nodeIndex = m_anchors[i].m_node? m_anchors[i].m_node-&m_nodes[0]: -1;
-			
-			memPtr->m_rigidBody = m_anchors[i].m_body? (btRigidBodyData*)  serializer->getUniquePointer((void*)m_anchors[i].m_body): 0;
+			memPtr->m_nodeIndex = m_anchors[i].m_node ? m_anchors[i].m_node - &m_nodes[0] : -1;
+
+			memPtr->m_rigidBody = m_anchors[i].m_body ? (btRigidBodyData*)serializer->getUniquePointer((void*)m_anchors[i].m_body) : 0;
 			btAssert(memPtr->m_nodeIndex < m_nodes.size());
 		}
-		serializer->finalizeChunk(chunk,"SoftRigidAnchorData",BT_ARRAY_CODE,(void*) &m_anchors[0]);
+		serializer->finalizeChunk(chunk, "SoftRigidAnchorData", BT_ARRAY_CODE, (void*)&m_anchors[0]);
 	}
-	
 
 	sbd->m_config.m_dynamicFriction = m_cfg.kDF;
 	sbd->m_config.m_baumgarte = m_cfg.kVCF;
@@ -3501,64 +3499,63 @@ const char*	btSoftBody::serialize(void* dataBuffer, class btSerializer* serializ
 		sbd->m_pose = (SoftBodyPoseData*)serializer->getUniquePointer((void*)&m_pose);
 
 		int sz = sizeof(SoftBodyPoseData);
-		btChunk* chunk = serializer->allocate(sz,1);
+		btChunk* chunk = serializer->allocate(sz, 1);
 		SoftBodyPoseData* memPtr = (SoftBodyPoseData*)chunk->m_oldPtr;
-		
+
 		m_pose.m_aqq.serializeFloat(memPtr->m_aqq);
 		memPtr->m_bframe = m_pose.m_bframe;
 		memPtr->m_bvolume = m_pose.m_bvolume;
 		m_pose.m_com.serializeFloat(memPtr->m_com);
-		
+
 		memPtr->m_numPositions = m_pose.m_pos.size();
-		memPtr->m_positions = memPtr->m_numPositions ? (btVector3FloatData*)serializer->getUniquePointer((void*)&m_pose.m_pos[0]): 0;
+		memPtr->m_positions = memPtr->m_numPositions ? (btVector3FloatData*)serializer->getUniquePointer((void*)&m_pose.m_pos[0]) : 0;
 		if (memPtr->m_numPositions)
 		{
 			int numElem = memPtr->m_numPositions;
 			int sz = sizeof(btVector3Data);
-			btChunk* chunk = serializer->allocate(sz,numElem);
+			btChunk* chunk = serializer->allocate(sz, numElem);
 			btVector3FloatData* memPtr = (btVector3FloatData*)chunk->m_oldPtr;
-			for (int i=0;i<numElem;i++,memPtr++)
+			for (int i = 0; i < numElem; i++, memPtr++)
 			{
 				m_pose.m_pos[i].serializeFloat(*memPtr);
 			}
-			serializer->finalizeChunk(chunk,"btVector3FloatData",BT_ARRAY_CODE,(void*)&m_pose.m_pos[0]);
+			serializer->finalizeChunk(chunk, "btVector3FloatData", BT_ARRAY_CODE, (void*)&m_pose.m_pos[0]);
 		}
 		memPtr->m_restVolume = m_pose.m_volume;
 		m_pose.m_rot.serializeFloat(memPtr->m_rot);
 		m_pose.m_scl.serializeFloat(memPtr->m_scale);
 
 		memPtr->m_numWeigts = m_pose.m_wgh.size();
-		memPtr->m_weights = memPtr->m_numWeigts? (float*) serializer->getUniquePointer((void*) &m_pose.m_wgh[0]) : 0;
+		memPtr->m_weights = memPtr->m_numWeigts ? (float*)serializer->getUniquePointer((void*)&m_pose.m_wgh[0]) : 0;
 		if (memPtr->m_numWeigts)
 		{
-			
 			int numElem = memPtr->m_numWeigts;
 			int sz = sizeof(float);
-			btChunk* chunk = serializer->allocate(sz,numElem);
-			float* memPtr = (float*) chunk->m_oldPtr;
-			for (int i=0;i<numElem;i++,memPtr++)
+			btChunk* chunk = serializer->allocate(sz, numElem);
+			float* memPtr = (float*)chunk->m_oldPtr;
+			for (int i = 0; i < numElem; i++, memPtr++)
 			{
 				*memPtr = m_pose.m_wgh[i];
 			}
-			serializer->finalizeChunk(chunk,"float",BT_ARRAY_CODE,(void*)&m_pose.m_wgh[0]);
+			serializer->finalizeChunk(chunk, "float", BT_ARRAY_CODE, (void*)&m_pose.m_wgh[0]);
 		}
 
-		serializer->finalizeChunk(chunk,"SoftBodyPoseData",BT_ARRAY_CODE,(void*)&m_pose);
+		serializer->finalizeChunk(chunk, "SoftBodyPoseData", BT_ARRAY_CODE, (void*)&m_pose);
 	}
 
 	//clusters for convex-cluster collision detection
 
 	sbd->m_numClusters = m_clusters.size();
-	sbd->m_clusters = sbd->m_numClusters? (SoftBodyClusterData*) serializer->getUniquePointer((void*)m_clusters[0]) : 0;
+	sbd->m_clusters = sbd->m_numClusters ? (SoftBodyClusterData*)serializer->getUniquePointer((void*)m_clusters[0]) : 0;
 	if (sbd->m_numClusters)
 	{
 		int numElem = sbd->m_numClusters;
 		int sz = sizeof(SoftBodyClusterData);
-		btChunk* chunk = serializer->allocate(sz,numElem);
-		SoftBodyClusterData* memPtr = (SoftBodyClusterData*) chunk->m_oldPtr;
-		for (int i=0;i<numElem;i++,memPtr++)
+		btChunk* chunk = serializer->allocate(sz, numElem);
+		SoftBodyClusterData* memPtr = (SoftBodyClusterData*)chunk->m_oldPtr;
+		for (int i = 0; i < numElem; i++, memPtr++)
 		{
-			memPtr->m_adamping= m_clusters[i]->m_adamping;
+			memPtr->m_adamping = m_clusters[i]->m_adamping;
 			m_clusters[i]->m_av.serializeFloat(memPtr->m_av);
 			memPtr->m_clusterIndex = m_clusters[i]->m_clusterIndex;
 			memPtr->m_collide = m_clusters[i]->m_collide;
@@ -3589,69 +3586,64 @@ const char*	btSoftBody::serialize(void* dataBuffer, class btSerializer* serializ
 			m_clusters[i]->m_vimpulses[1].serializeFloat(memPtr->m_vimpulses[1]);
 			memPtr->m_ndimpulses = m_clusters[i]->m_ndimpulses;
 
-			
-
-			memPtr->m_framerefs = memPtr->m_numFrameRefs? (btVector3FloatData*)serializer->getUniquePointer((void*)&m_clusters[i]->m_framerefs[0]) : 0;
+			memPtr->m_framerefs = memPtr->m_numFrameRefs ? (btVector3FloatData*)serializer->getUniquePointer((void*)&m_clusters[i]->m_framerefs[0]) : 0;
 			if (memPtr->m_framerefs)
 			{
 				int numElem = memPtr->m_numFrameRefs;
 				int sz = sizeof(btVector3FloatData);
-				btChunk* chunk = serializer->allocate(sz,numElem);
-				btVector3FloatData* memPtr = (btVector3FloatData*) chunk->m_oldPtr;
-				for (int j=0;j<numElem;j++,memPtr++)
+				btChunk* chunk = serializer->allocate(sz, numElem);
+				btVector3FloatData* memPtr = (btVector3FloatData*)chunk->m_oldPtr;
+				for (int j = 0; j < numElem; j++, memPtr++)
 				{
 					m_clusters[i]->m_framerefs[j].serializeFloat(*memPtr);
 				}
-				serializer->finalizeChunk(chunk,"btVector3FloatData",BT_ARRAY_CODE,(void*)&m_clusters[i]->m_framerefs[0]);
+				serializer->finalizeChunk(chunk, "btVector3FloatData", BT_ARRAY_CODE, (void*)&m_clusters[i]->m_framerefs[0]);
 			}
-			
-			memPtr->m_masses = memPtr->m_numMasses ? (float*) serializer->getUniquePointer((void*)&m_clusters[i]->m_masses[0]): 0;
+
+			memPtr->m_masses = memPtr->m_numMasses ? (float*)serializer->getUniquePointer((void*)&m_clusters[i]->m_masses[0]) : 0;
 			if (memPtr->m_masses)
 			{
 				int numElem = memPtr->m_numMasses;
 				int sz = sizeof(float);
-				btChunk* chunk = serializer->allocate(sz,numElem);
-				float* memPtr = (float*) chunk->m_oldPtr;
-				for (int j=0;j<numElem;j++,memPtr++)
+				btChunk* chunk = serializer->allocate(sz, numElem);
+				float* memPtr = (float*)chunk->m_oldPtr;
+				for (int j = 0; j < numElem; j++, memPtr++)
 				{
 					*memPtr = m_clusters[i]->m_masses[j];
 				}
-				serializer->finalizeChunk(chunk,"float",BT_ARRAY_CODE,(void*)&m_clusters[i]->m_masses[0]);
+				serializer->finalizeChunk(chunk, "float", BT_ARRAY_CODE, (void*)&m_clusters[i]->m_masses[0]);
 			}
 
-			memPtr->m_nodeIndices  = memPtr->m_numNodes ? (int*) serializer->getUniquePointer((void*) &m_clusters[i]->m_nodes) : 0;
-			if (memPtr->m_nodeIndices )
+			memPtr->m_nodeIndices = memPtr->m_numNodes ? (int*)serializer->getUniquePointer((void*)&m_clusters[i]->m_nodes) : 0;
+			if (memPtr->m_nodeIndices)
 			{
 				int numElem = memPtr->m_numMasses;
 				int sz = sizeof(int);
-				btChunk* chunk = serializer->allocate(sz,numElem);
-				int* memPtr = (int*) chunk->m_oldPtr;
-				for (int j=0;j<numElem;j++,memPtr++)
+				btChunk* chunk = serializer->allocate(sz, numElem);
+				int* memPtr = (int*)chunk->m_oldPtr;
+				for (int j = 0; j < numElem; j++, memPtr++)
 				{
 					int* indexPtr = m_nodeIndexMap.find(m_clusters[i]->m_nodes[j]);
 					btAssert(indexPtr);
 					*memPtr = *indexPtr;
 				}
-				serializer->finalizeChunk(chunk,"int",BT_ARRAY_CODE,(void*)&m_clusters[i]->m_nodes);
+				serializer->finalizeChunk(chunk, "int", BT_ARRAY_CODE, (void*)&m_clusters[i]->m_nodes);
 			}
 		}
-		serializer->finalizeChunk(chunk,"SoftBodyClusterData",BT_ARRAY_CODE,(void*)m_clusters[0]);
-
+		serializer->finalizeChunk(chunk, "SoftBodyClusterData", BT_ARRAY_CODE, (void*)m_clusters[0]);
 	}
-	
 
-	
 	sbd->m_numJoints = m_joints.size();
-	sbd->m_joints = m_joints.size()? (btSoftBodyJointData*) serializer->getUniquePointer((void*)&m_joints[0]) : 0;
+	sbd->m_joints = m_joints.size() ? (btSoftBodyJointData*)serializer->getUniquePointer((void*)&m_joints[0]) : 0;
 
 	if (sbd->m_joints)
 	{
 		int sz = sizeof(btSoftBodyJointData);
 		int numElem = m_joints.size();
-		btChunk* chunk = serializer->allocate(sz,numElem);
+		btChunk* chunk = serializer->allocate(sz, numElem);
 		btSoftBodyJointData* memPtr = (btSoftBodyJointData*)chunk->m_oldPtr;
 
-		for (int i=0;i<numElem;i++,memPtr++)
+		for (int i = 0; i < numElem; i++, memPtr++)
 		{
 			memPtr->m_jointType = (int)m_joints[i]->Type();
 			m_joints[i]->m_refs[0].serializeFloat(memPtr->m_refs[0]);
@@ -3660,8 +3652,8 @@ const char*	btSoftBody::serialize(void* dataBuffer, class btSerializer* serializ
 			memPtr->m_erp = float(m_joints[i]->m_erp);
 			memPtr->m_split = float(m_joints[i]->m_split);
 			memPtr->m_delete = m_joints[i]->m_delete;
-			
-			for (int j=0;j<4;j++)
+
+			for (int j = 0; j < 4; j++)
 			{
 				memPtr->m_relPosition[0].m_floats[j] = 0.f;
 				memPtr->m_relPosition[1].m_floats[j] = 0.f;
@@ -3700,10 +3692,8 @@ const char*	btSoftBody::serialize(void* dataBuffer, class btSerializer* serializ
 				memPtr->m_bodyB = serializer->getUniquePointer((void*)m_joints[i]->m_bodies[1].m_rigid);
 			}
 		}
-		serializer->finalizeChunk(chunk,"btSoftBodyJointData",BT_ARRAY_CODE,(void*) &m_joints[0]);
+		serializer->finalizeChunk(chunk, "btSoftBodyJointData", BT_ARRAY_CODE, (void*)&m_joints[0]);
 	}
 
-
 	return btSoftBodyDataName;
 }
-
diff --git a/thirdparty/bullet/BulletSoftBody/btSoftBody.h b/thirdparty/bullet/BulletSoftBody/btSoftBody.h
index ada0dfd1a5c..9b35b799d80 100644
--- a/thirdparty/bullet/BulletSoftBody/btSoftBody.h
+++ b/thirdparty/bullet/BulletSoftBody/btSoftBody.h
@@ -31,862 +31,890 @@ subject to the following restrictions:
 //#define btRigidBodyData	btRigidBodyDoubleData
 //#define btRigidBodyDataName	"btRigidBodyDoubleData"
 //#else
-#define btSoftBodyData	btSoftBodyFloatData
-#define btSoftBodyDataName	"btSoftBodyFloatData"
+#define btSoftBodyData btSoftBodyFloatData
+#define btSoftBodyDataName "btSoftBodyFloatData"
 //#endif //BT_USE_DOUBLE_PRECISION
 
 class btBroadphaseInterface;
 class btDispatcher;
 class btSoftBodySolver;
 
-/* btSoftBodyWorldInfo	*/ 
-struct	btSoftBodyWorldInfo
+/* btSoftBodyWorldInfo	*/
+struct btSoftBodyWorldInfo
 {
-	btScalar				air_density;
-	btScalar				water_density;
-	btScalar				water_offset;
-	btScalar				m_maxDisplacement;
-	btVector3				water_normal;
-	btBroadphaseInterface*	m_broadphase;
-	btDispatcher*	m_dispatcher;
-	btVector3				m_gravity;
-	btSparseSdf<3>			m_sparsesdf;
+	btScalar air_density;
+	btScalar water_density;
+	btScalar water_offset;
+	btScalar m_maxDisplacement;
+	btVector3 water_normal;
+	btBroadphaseInterface* m_broadphase;
+	btDispatcher* m_dispatcher;
+	btVector3 m_gravity;
+	btSparseSdf<3> m_sparsesdf;
 
 	btSoftBodyWorldInfo()
-		:air_density((btScalar)1.2),
-		water_density(0),
-		water_offset(0),
-		m_maxDisplacement(1000.f),//avoid soft body from 'exploding' so use some upper threshold of maximum motion that a node can travel per frame
-		water_normal(0,0,0),
-		m_broadphase(0),
-		m_dispatcher(0),
-		m_gravity(0,-10,0)
+		: air_density((btScalar)1.2),
+		  water_density(0),
+		  water_offset(0),
+		  m_maxDisplacement(1000.f),  //avoid soft body from 'exploding' so use some upper threshold of maximum motion that a node can travel per frame
+		  water_normal(0, 0, 0),
+		  m_broadphase(0),
+		  m_dispatcher(0),
+		  m_gravity(0, -10, 0)
 	{
 	}
-};	
+};
 
-
-///The btSoftBody is an class to simulate cloth and volumetric soft bodies. 
+///The btSoftBody is an class to simulate cloth and volumetric soft bodies.
 ///There is two-way interaction between btSoftBody and btRigidBody/btCollisionObject.
-class	btSoftBody : public btCollisionObject
+class btSoftBody : public btCollisionObject
 {
 public:
 	btAlignedObjectArray<const class btCollisionObject*> m_collisionDisabledObjects;
 
 	// The solver object that handles this soft body
-	btSoftBodySolver *m_softBodySolver;
+	btSoftBodySolver* m_softBodySolver;
 
 	//
 	// Enumerations
 	//
 
-	///eAeroModel 
-	struct eAeroModel { enum _ {
-		V_Point,			///Vertex normals are oriented toward velocity
-		V_TwoSided,			///Vertex normals are flipped to match velocity	
-		V_TwoSidedLiftDrag, ///Vertex normals are flipped to match velocity and lift and drag forces are applied
-		V_OneSided,			///Vertex normals are taken as it is	
-		F_TwoSided,			///Face normals are flipped to match velocity
-		F_TwoSidedLiftDrag,	///Face normals are flipped to match velocity and lift and drag forces are applied 
-		F_OneSided,			///Face normals are taken as it is		
-		END
-	};};
+	///eAeroModel
+	struct eAeroModel
+	{
+		enum _
+		{
+			V_Point,             ///Vertex normals are oriented toward velocity
+			V_TwoSided,          ///Vertex normals are flipped to match velocity
+			V_TwoSidedLiftDrag,  ///Vertex normals are flipped to match velocity and lift and drag forces are applied
+			V_OneSided,          ///Vertex normals are taken as it is
+			F_TwoSided,          ///Face normals are flipped to match velocity
+			F_TwoSidedLiftDrag,  ///Face normals are flipped to match velocity and lift and drag forces are applied
+			F_OneSided,          ///Face normals are taken as it is
+			END
+		};
+	};
 
 	///eVSolver : velocities solvers
-	struct	eVSolver { enum _ {
-		Linear,		///Linear solver
-		END
-	};};
+	struct eVSolver
+	{
+		enum _
+		{
+			Linear,  ///Linear solver
+			END
+		};
+	};
 
 	///ePSolver : positions solvers
-	struct	ePSolver { enum _ {
-		Linear,		///Linear solver
-		Anchors,	///Anchor solver
-		RContacts,	///Rigid contacts solver
-		SContacts,	///Soft contacts solver
-		END
-	};};
+	struct ePSolver
+	{
+		enum _
+		{
+			Linear,     ///Linear solver
+			Anchors,    ///Anchor solver
+			RContacts,  ///Rigid contacts solver
+			SContacts,  ///Soft contacts solver
+			END
+		};
+	};
 
 	///eSolverPresets
-	struct	eSolverPresets { enum _ {
-		Positions,
-		Velocities,
-		Default	=	Positions,
-		END
-	};};
+	struct eSolverPresets
+	{
+		enum _
+		{
+			Positions,
+			Velocities,
+			Default = Positions,
+			END
+		};
+	};
 
 	///eFeature
-	struct	eFeature { enum _ {
-		None,
-		Node,
-		Link,
-		Face,
-		Tetra,
-		END
-	};};
+	struct eFeature
+	{
+		enum _
+		{
+			None,
+			Node,
+			Link,
+			Face,
+			Tetra,
+			END
+		};
+	};
 
-	typedef btAlignedObjectArray<eVSolver::_>	tVSolverArray;
-	typedef btAlignedObjectArray<ePSolver::_>	tPSolverArray;
+	typedef btAlignedObjectArray<eVSolver::_> tVSolverArray;
+	typedef btAlignedObjectArray<ePSolver::_> tPSolverArray;
 
 	//
 	// Flags
 	//
 
 	///fCollision
-	struct fCollision { enum _ {
-		RVSmask	=	0x000f,	///Rigid versus soft mask
-		SDF_RS	=	0x0001,	///SDF based rigid vs soft
-		CL_RS	=	0x0002, ///Cluster vs convex rigid vs soft
+	struct fCollision
+	{
+		enum _
+		{
+			RVSmask = 0x000f,  ///Rigid versus soft mask
+			SDF_RS = 0x0001,   ///SDF based rigid vs soft
+			CL_RS = 0x0002,    ///Cluster vs convex rigid vs soft
 
-		SVSmask	=	0x0030,	///Rigid versus soft mask		
-		VF_SS	=	0x0010,	///Vertex vs face soft vs soft handling
-		CL_SS	=	0x0020, ///Cluster vs cluster soft vs soft handling
-		CL_SELF =	0x0040, ///Cluster soft body self collision
-		/* presets	*/ 
-		Default	=	SDF_RS,
-		END
-	};};
+			SVSmask = 0x0030,  ///Rigid versus soft mask
+			VF_SS = 0x0010,    ///Vertex vs face soft vs soft handling
+			CL_SS = 0x0020,    ///Cluster vs cluster soft vs soft handling
+			CL_SELF = 0x0040,  ///Cluster soft body self collision
+			/* presets	*/
+			Default = SDF_RS,
+			END
+		};
+	};
 
 	///fMaterial
-	struct fMaterial { enum _ {
-		DebugDraw	=	0x0001,	/// Enable debug draw
-		/* presets	*/ 
-		Default		=	DebugDraw,
-		END
-	};};
+	struct fMaterial
+	{
+		enum _
+		{
+			DebugDraw = 0x0001,  /// Enable debug draw
+			/* presets	*/
+			Default = DebugDraw,
+			END
+		};
+	};
 
 	//
 	// API Types
 	//
 
-	/* sRayCast		*/ 
+	/* sRayCast		*/
 	struct sRayCast
 	{
-		btSoftBody*	body;		/// soft body
-		eFeature::_	feature;	/// feature type
-		int			index;		/// feature index
-		btScalar	fraction;		/// time of impact fraction (rayorg+(rayto-rayfrom)*fraction)
+		btSoftBody* body;     /// soft body
+		eFeature::_ feature;  /// feature type
+		int index;            /// feature index
+		btScalar fraction;    /// time of impact fraction (rayorg+(rayto-rayfrom)*fraction)
 	};
 
-	/* ImplicitFn	*/ 
-	struct	ImplicitFn
+	/* ImplicitFn	*/
+	struct ImplicitFn
 	{
 		virtual ~ImplicitFn() {}
-		virtual btScalar	Eval(const btVector3& x)=0;
+		virtual btScalar Eval(const btVector3& x) = 0;
 	};
 
 	//
 	// Internal types
 	//
 
-	typedef btAlignedObjectArray<btScalar>	tScalarArray;
-	typedef btAlignedObjectArray<btVector3>	tVector3Array;
+	typedef btAlignedObjectArray<btScalar> tScalarArray;
+	typedef btAlignedObjectArray<btVector3> tVector3Array;
 
-	/* sCti is Softbody contact info	*/ 
-	struct	sCti
+	/* sCti is Softbody contact info	*/
+	struct sCti
 	{
-		const btCollisionObject*	m_colObj;		/* Rigid body			*/ 
-		btVector3		m_normal;	/* Outward normal		*/ 
-		btScalar		m_offset;	/* Offset from origin	*/ 
-	};	
-
-	/* sMedium		*/ 
-	struct	sMedium
-	{
-		btVector3		m_velocity;	/* Velocity				*/ 
-		btScalar		m_pressure;	/* Pressure				*/ 
-		btScalar		m_density;	/* Density				*/ 
+		const btCollisionObject* m_colObj; /* Rigid body			*/
+		btVector3 m_normal;                /* Outward normal		*/
+		btScalar m_offset;                 /* Offset from origin	*/
 	};
 
-	/* Base type	*/ 
-	struct	Element
+	/* sMedium		*/
+	struct sMedium
 	{
-		void*			m_tag;			// User data
+		btVector3 m_velocity; /* Velocity				*/
+		btScalar m_pressure;  /* Pressure				*/
+		btScalar m_density;   /* Density				*/
+	};
+
+	/* Base type	*/
+	struct Element
+	{
+		void* m_tag;  // User data
 		Element() : m_tag(0) {}
 	};
-	/* Material		*/ 
-	struct	Material : Element
+	/* Material		*/
+	struct Material : Element
 	{
-		btScalar				m_kLST;			// Linear stiffness coefficient [0,1]
-		btScalar				m_kAST;			// Area/Angular stiffness coefficient [0,1]
-		btScalar				m_kVST;			// Volume stiffness coefficient [0,1]
-		int						m_flags;		// Flags
+		btScalar m_kLST;  // Linear stiffness coefficient [0,1]
+		btScalar m_kAST;  // Area/Angular stiffness coefficient [0,1]
+		btScalar m_kVST;  // Volume stiffness coefficient [0,1]
+		int m_flags;      // Flags
 	};
 
-	/* Feature		*/ 
-	struct	Feature : Element
+	/* Feature		*/
+	struct Feature : Element
 	{
-		Material*				m_material;		// Material
+		Material* m_material;  // Material
 	};
-	/* Node			*/ 
-	struct	Node : Feature
+	/* Node			*/
+	struct Node : Feature
 	{
-		btVector3				m_x;			// Position
-		btVector3				m_q;			// Previous step position
-		btVector3				m_v;			// Velocity
-		btVector3				m_f;			// Force accumulator
-		btVector3				m_n;			// Normal
-		btScalar				m_im;			// 1/mass
-		btScalar				m_area;			// Area
-		btDbvtNode*				m_leaf;			// Leaf data
-		int						m_battach:1;	// Attached
+		btVector3 m_x;       // Position
+		btVector3 m_q;       // Previous step position
+		btVector3 m_v;       // Velocity
+		btVector3 m_f;       // Force accumulator
+		btVector3 m_n;       // Normal
+		btScalar m_im;       // 1/mass
+		btScalar m_area;     // Area
+		btDbvtNode* m_leaf;  // Leaf data
+		int m_battach : 1;   // Attached
 	};
-	/* Link			*/ 
-	ATTRIBUTE_ALIGNED16(struct)	Link : Feature
+	/* Link			*/
+	ATTRIBUTE_ALIGNED16(struct)
+	Link : Feature
 	{
-		btVector3				m_c3;			// gradient
-		Node*					m_n[2];			// Node pointers
-		btScalar				m_rl;			// Rest length		
-		int						m_bbending:1;	// Bending link
-		btScalar				m_c0;			// (ima+imb)*kLST
-		btScalar				m_c1;			// rl^2
-		btScalar				m_c2;			// |gradient|^2/c0
-	
+		btVector3 m_c3;      // gradient
+		Node* m_n[2];        // Node pointers
+		btScalar m_rl;       // Rest length
+		int m_bbending : 1;  // Bending link
+		btScalar m_c0;       // (ima+imb)*kLST
+		btScalar m_c1;       // rl^2
+		btScalar m_c2;       // |gradient|^2/c0
+
 		BT_DECLARE_ALIGNED_ALLOCATOR();
-
 	};
-	/* Face			*/ 
-	struct	Face : Feature
+	/* Face			*/
+	struct Face : Feature
 	{
-		Node*					m_n[3];			// Node pointers
-		btVector3				m_normal;		// Normal
-		btScalar				m_ra;			// Rest area
-		btDbvtNode*				m_leaf;			// Leaf data
+		Node* m_n[3];        // Node pointers
+		btVector3 m_normal;  // Normal
+		btScalar m_ra;       // Rest area
+		btDbvtNode* m_leaf;  // Leaf data
 	};
-	/* Tetra		*/ 
-	struct	Tetra : Feature
+	/* Tetra		*/
+	struct Tetra : Feature
 	{
-		Node*					m_n[4];			// Node pointers		
-		btScalar				m_rv;			// Rest volume
-		btDbvtNode*				m_leaf;			// Leaf data
-		btVector3				m_c0[4];		// gradients
-		btScalar				m_c1;			// (4*kVST)/(im0+im1+im2+im3)
-		btScalar				m_c2;			// m_c1/sum(|g0..3|^2)
+		Node* m_n[4];        // Node pointers
+		btScalar m_rv;       // Rest volume
+		btDbvtNode* m_leaf;  // Leaf data
+		btVector3 m_c0[4];   // gradients
+		btScalar m_c1;       // (4*kVST)/(im0+im1+im2+im3)
+		btScalar m_c2;       // m_c1/sum(|g0..3|^2)
 	};
-	/* RContact		*/ 
-	struct	RContact
+	/* RContact		*/
+	struct RContact
 	{
-		sCti		m_cti;			// Contact infos
-		Node*					m_node;			// Owner node
-		btMatrix3x3				m_c0;			// Impulse matrix
-		btVector3				m_c1;			// Relative anchor
-		btScalar				m_c2;			// ima*dt
-		btScalar				m_c3;			// Friction
-		btScalar				m_c4;			// Hardness
+		sCti m_cti;        // Contact infos
+		Node* m_node;      // Owner node
+		btMatrix3x3 m_c0;  // Impulse matrix
+		btVector3 m_c1;    // Relative anchor
+		btScalar m_c2;     // ima*dt
+		btScalar m_c3;     // Friction
+		btScalar m_c4;     // Hardness
 	};
-	/* SContact		*/ 
-	struct	SContact
+	/* SContact		*/
+	struct SContact
 	{
-		Node*					m_node;			// Node
-		Face*					m_face;			// Face
-		btVector3				m_weights;		// Weigths
-		btVector3				m_normal;		// Normal
-		btScalar				m_margin;		// Margin
-		btScalar				m_friction;		// Friction
-		btScalar				m_cfm[2];		// Constraint force mixing
+		Node* m_node;         // Node
+		Face* m_face;         // Face
+		btVector3 m_weights;  // Weigths
+		btVector3 m_normal;   // Normal
+		btScalar m_margin;    // Margin
+		btScalar m_friction;  // Friction
+		btScalar m_cfm[2];    // Constraint force mixing
 	};
-	/* Anchor		*/ 
-	struct	Anchor
+	/* Anchor		*/
+	struct Anchor
 	{
-		Node*					m_node;			// Node pointer
-		btVector3				m_local;		// Anchor position in body space
-		btRigidBody*			m_body;			// Body
-		btScalar				m_influence;
-		btMatrix3x3				m_c0;			// Impulse matrix
-		btVector3				m_c1;			// Relative anchor
-		btScalar				m_c2;			// ima*dt
+		Node* m_node;         // Node pointer
+		btVector3 m_local;    // Anchor position in body space
+		btRigidBody* m_body;  // Body
+		btScalar m_influence;
+		btMatrix3x3 m_c0;  // Impulse matrix
+		btVector3 m_c1;    // Relative anchor
+		btScalar m_c2;     // ima*dt
 	};
-	/* Note			*/ 
-	struct	Note : Element
+	/* Note			*/
+	struct Note : Element
 	{
-		const char*				m_text;			// Text
-		btVector3				m_offset;		// Offset
-		int						m_rank;			// Rank
-		Node*					m_nodes[4];		// Nodes
-		btScalar				m_coords[4];	// Coordinates
-	};	
-	/* Pose			*/ 
-	struct	Pose
-	{
-		bool					m_bvolume;		// Is valid
-		bool					m_bframe;		// Is frame
-		btScalar				m_volume;		// Rest volume
-		tVector3Array			m_pos;			// Reference positions
-		tScalarArray			m_wgh;			// Weights
-		btVector3				m_com;			// COM
-		btMatrix3x3				m_rot;			// Rotation
-		btMatrix3x3				m_scl;			// Scale
-		btMatrix3x3				m_aqq;			// Base scaling
+		const char* m_text;    // Text
+		btVector3 m_offset;    // Offset
+		int m_rank;            // Rank
+		Node* m_nodes[4];      // Nodes
+		btScalar m_coords[4];  // Coordinates
 	};
-	/* Cluster		*/ 
-	struct	Cluster
+	/* Pose			*/
+	struct Pose
 	{
-		tScalarArray				m_masses;
-		btAlignedObjectArray<Node*>	m_nodes;		
-		tVector3Array				m_framerefs;
-		btTransform					m_framexform;
-		btScalar					m_idmass;
-		btScalar					m_imass;
-		btMatrix3x3					m_locii;
-		btMatrix3x3					m_invwi;
-		btVector3					m_com;
-		btVector3					m_vimpulses[2];
-		btVector3					m_dimpulses[2];
-		int							m_nvimpulses;
-		int							m_ndimpulses;
-		btVector3					m_lv;
-		btVector3					m_av;
-		btDbvtNode*					m_leaf;
-		btScalar					m_ndamping;	/* Node damping		*/ 
-		btScalar					m_ldamping;	/* Linear damping	*/ 
-		btScalar					m_adamping;	/* Angular damping	*/ 
-		btScalar					m_matching;
-		btScalar					m_maxSelfCollisionImpulse;
-		btScalar					m_selfCollisionImpulseFactor;
-		bool						m_containsAnchor;
-		bool						m_collide;
-		int							m_clusterIndex;
-		Cluster() : m_leaf(0),m_ndamping(0),m_ldamping(0),m_adamping(0),m_matching(0) 
-		,m_maxSelfCollisionImpulse(100.f),
-		m_selfCollisionImpulseFactor(0.01f),
-		m_containsAnchor(false)
-		{}
+		bool m_bvolume;       // Is valid
+		bool m_bframe;        // Is frame
+		btScalar m_volume;    // Rest volume
+		tVector3Array m_pos;  // Reference positions
+		tScalarArray m_wgh;   // Weights
+		btVector3 m_com;      // COM
+		btMatrix3x3 m_rot;    // Rotation
+		btMatrix3x3 m_scl;    // Scale
+		btMatrix3x3 m_aqq;    // Base scaling
 	};
-	/* Impulse		*/ 
-	struct	Impulse
+	/* Cluster		*/
+	struct Cluster
 	{
-		btVector3					m_velocity;
-		btVector3					m_drift;
-		int							m_asVelocity:1;
-		int							m_asDrift:1;
-		Impulse() : m_velocity(0,0,0),m_drift(0,0,0),m_asVelocity(0),m_asDrift(0)	{}
-		Impulse						operator -() const
+		tScalarArray m_masses;
+		btAlignedObjectArray<Node*> m_nodes;
+		tVector3Array m_framerefs;
+		btTransform m_framexform;
+		btScalar m_idmass;
+		btScalar m_imass;
+		btMatrix3x3 m_locii;
+		btMatrix3x3 m_invwi;
+		btVector3 m_com;
+		btVector3 m_vimpulses[2];
+		btVector3 m_dimpulses[2];
+		int m_nvimpulses;
+		int m_ndimpulses;
+		btVector3 m_lv;
+		btVector3 m_av;
+		btDbvtNode* m_leaf;
+		btScalar m_ndamping; /* Node damping		*/
+		btScalar m_ldamping; /* Linear damping	*/
+		btScalar m_adamping; /* Angular damping	*/
+		btScalar m_matching;
+		btScalar m_maxSelfCollisionImpulse;
+		btScalar m_selfCollisionImpulseFactor;
+		bool m_containsAnchor;
+		bool m_collide;
+		int m_clusterIndex;
+		Cluster() : m_leaf(0), m_ndamping(0), m_ldamping(0), m_adamping(0), m_matching(0), m_maxSelfCollisionImpulse(100.f), m_selfCollisionImpulseFactor(0.01f), m_containsAnchor(false)
 		{
-			Impulse i=*this;
-			i.m_velocity=-i.m_velocity;
-			i.m_drift=-i.m_drift;
-			return(i);
-		}
-		Impulse						operator*(btScalar x) const
-		{
-			Impulse i=*this;
-			i.m_velocity*=x;
-			i.m_drift*=x;
-			return(i);
 		}
 	};
-	/* Body			*/ 
-	struct	Body
+	/* Impulse		*/
+	struct Impulse
 	{
-		Cluster*			m_soft;
-		btRigidBody*		m_rigid;
-		const btCollisionObject*	m_collisionObject;
+		btVector3 m_velocity;
+		btVector3 m_drift;
+		int m_asVelocity : 1;
+		int m_asDrift : 1;
+		Impulse() : m_velocity(0, 0, 0), m_drift(0, 0, 0), m_asVelocity(0), m_asDrift(0) {}
+		Impulse operator-() const
+		{
+			Impulse i = *this;
+			i.m_velocity = -i.m_velocity;
+			i.m_drift = -i.m_drift;
+			return (i);
+		}
+		Impulse operator*(btScalar x) const
+		{
+			Impulse i = *this;
+			i.m_velocity *= x;
+			i.m_drift *= x;
+			return (i);
+		}
+	};
+	/* Body			*/
+	struct Body
+	{
+		Cluster* m_soft;
+		btRigidBody* m_rigid;
+		const btCollisionObject* m_collisionObject;
 
-		Body() : m_soft(0),m_rigid(0),m_collisionObject(0)				{}
-		Body(Cluster* p) : m_soft(p),m_rigid(0),m_collisionObject(0)	{}
-		Body(const btCollisionObject* colObj) : m_soft(0),m_collisionObject(colObj)
+		Body() : m_soft(0), m_rigid(0), m_collisionObject(0) {}
+		Body(Cluster* p) : m_soft(p), m_rigid(0), m_collisionObject(0) {}
+		Body(const btCollisionObject* colObj) : m_soft(0), m_collisionObject(colObj)
 		{
 			m_rigid = (btRigidBody*)btRigidBody::upcast(m_collisionObject);
 		}
 
-		void						activate() const
+		void activate() const
 		{
-			if(m_rigid) 
+			if (m_rigid)
 				m_rigid->activate();
 			if (m_collisionObject)
 				m_collisionObject->activate();
-
 		}
-		const btMatrix3x3&			invWorldInertia() const
+		const btMatrix3x3& invWorldInertia() const
 		{
-			static const btMatrix3x3	iwi(0,0,0,0,0,0,0,0,0);
-			if(m_rigid) return(m_rigid->getInvInertiaTensorWorld());
-			if(m_soft)	return(m_soft->m_invwi);
-			return(iwi);
+			static const btMatrix3x3 iwi(0, 0, 0, 0, 0, 0, 0, 0, 0);
+			if (m_rigid) return (m_rigid->getInvInertiaTensorWorld());
+			if (m_soft) return (m_soft->m_invwi);
+			return (iwi);
 		}
-		btScalar					invMass() const
+		btScalar invMass() const
 		{
-			if(m_rigid) return(m_rigid->getInvMass());
-			if(m_soft)	return(m_soft->m_imass);
-			return(0);
+			if (m_rigid) return (m_rigid->getInvMass());
+			if (m_soft) return (m_soft->m_imass);
+			return (0);
 		}
-		const btTransform&			xform() const
+		const btTransform& xform() const
 		{
-			static const btTransform	identity=btTransform::getIdentity();		
-			if(m_collisionObject) return(m_collisionObject->getWorldTransform());
-			if(m_soft)	return(m_soft->m_framexform);
-			return(identity);
+			static const btTransform identity = btTransform::getIdentity();
+			if (m_collisionObject) return (m_collisionObject->getWorldTransform());
+			if (m_soft) return (m_soft->m_framexform);
+			return (identity);
 		}
-		btVector3					linearVelocity() const
+		btVector3 linearVelocity() const
 		{
-			if(m_rigid) return(m_rigid->getLinearVelocity());
-			if(m_soft)	return(m_soft->m_lv);
-			return(btVector3(0,0,0));
+			if (m_rigid) return (m_rigid->getLinearVelocity());
+			if (m_soft) return (m_soft->m_lv);
+			return (btVector3(0, 0, 0));
 		}
-		btVector3					angularVelocity(const btVector3& rpos) const
-		{			
-			if(m_rigid) return(btCross(m_rigid->getAngularVelocity(),rpos));
-			if(m_soft)	return(btCross(m_soft->m_av,rpos));
-			return(btVector3(0,0,0));
-		}
-		btVector3					angularVelocity() const
-		{			
-			if(m_rigid) return(m_rigid->getAngularVelocity());
-			if(m_soft)	return(m_soft->m_av);
-			return(btVector3(0,0,0));
-		}
-		btVector3					velocity(const btVector3& rpos) const
+		btVector3 angularVelocity(const btVector3& rpos) const
 		{
-			return(linearVelocity()+angularVelocity(rpos));
+			if (m_rigid) return (btCross(m_rigid->getAngularVelocity(), rpos));
+			if (m_soft) return (btCross(m_soft->m_av, rpos));
+			return (btVector3(0, 0, 0));
 		}
-		void						applyVImpulse(const btVector3& impulse,const btVector3& rpos) const
+		btVector3 angularVelocity() const
 		{
-			if(m_rigid)	m_rigid->applyImpulse(impulse,rpos);
-			if(m_soft)	btSoftBody::clusterVImpulse(m_soft,rpos,impulse);
+			if (m_rigid) return (m_rigid->getAngularVelocity());
+			if (m_soft) return (m_soft->m_av);
+			return (btVector3(0, 0, 0));
 		}
-		void						applyDImpulse(const btVector3& impulse,const btVector3& rpos) const
+		btVector3 velocity(const btVector3& rpos) const
 		{
-			if(m_rigid)	m_rigid->applyImpulse(impulse,rpos);
-			if(m_soft)	btSoftBody::clusterDImpulse(m_soft,rpos,impulse);
-		}		
-		void						applyImpulse(const Impulse& impulse,const btVector3& rpos) const
+			return (linearVelocity() + angularVelocity(rpos));
+		}
+		void applyVImpulse(const btVector3& impulse, const btVector3& rpos) const
 		{
-			if(impulse.m_asVelocity)	
+			if (m_rigid) m_rigid->applyImpulse(impulse, rpos);
+			if (m_soft) btSoftBody::clusterVImpulse(m_soft, rpos, impulse);
+		}
+		void applyDImpulse(const btVector3& impulse, const btVector3& rpos) const
+		{
+			if (m_rigid) m_rigid->applyImpulse(impulse, rpos);
+			if (m_soft) btSoftBody::clusterDImpulse(m_soft, rpos, impulse);
+		}
+		void applyImpulse(const Impulse& impulse, const btVector3& rpos) const
+		{
+			if (impulse.m_asVelocity)
 			{
-//				printf("impulse.m_velocity = %f,%f,%f\n",impulse.m_velocity.getX(),impulse.m_velocity.getY(),impulse.m_velocity.getZ());
-				applyVImpulse(impulse.m_velocity,rpos);
+				//				printf("impulse.m_velocity = %f,%f,%f\n",impulse.m_velocity.getX(),impulse.m_velocity.getY(),impulse.m_velocity.getZ());
+				applyVImpulse(impulse.m_velocity, rpos);
 			}
-			if(impulse.m_asDrift)		
+			if (impulse.m_asDrift)
 			{
-//				printf("impulse.m_drift = %f,%f,%f\n",impulse.m_drift.getX(),impulse.m_drift.getY(),impulse.m_drift.getZ());
-				applyDImpulse(impulse.m_drift,rpos);
+				//				printf("impulse.m_drift = %f,%f,%f\n",impulse.m_drift.getX(),impulse.m_drift.getY(),impulse.m_drift.getZ());
+				applyDImpulse(impulse.m_drift, rpos);
 			}
 		}
-		void						applyVAImpulse(const btVector3& impulse) const
+		void applyVAImpulse(const btVector3& impulse) const
 		{
-			if(m_rigid)	m_rigid->applyTorqueImpulse(impulse);
-			if(m_soft)	btSoftBody::clusterVAImpulse(m_soft,impulse);
+			if (m_rigid) m_rigid->applyTorqueImpulse(impulse);
+			if (m_soft) btSoftBody::clusterVAImpulse(m_soft, impulse);
 		}
-		void						applyDAImpulse(const btVector3& impulse) const
+		void applyDAImpulse(const btVector3& impulse) const
 		{
-			if(m_rigid)	m_rigid->applyTorqueImpulse(impulse);
-			if(m_soft)	btSoftBody::clusterDAImpulse(m_soft,impulse);
+			if (m_rigid) m_rigid->applyTorqueImpulse(impulse);
+			if (m_soft) btSoftBody::clusterDAImpulse(m_soft, impulse);
 		}
-		void						applyAImpulse(const Impulse& impulse) const
+		void applyAImpulse(const Impulse& impulse) const
 		{
-			if(impulse.m_asVelocity)	applyVAImpulse(impulse.m_velocity);
-			if(impulse.m_asDrift)		applyDAImpulse(impulse.m_drift);
+			if (impulse.m_asVelocity) applyVAImpulse(impulse.m_velocity);
+			if (impulse.m_asDrift) applyDAImpulse(impulse.m_drift);
 		}
-		void						applyDCImpulse(const btVector3& impulse) const
+		void applyDCImpulse(const btVector3& impulse) const
 		{
-			if(m_rigid)	m_rigid->applyCentralImpulse(impulse);
-			if(m_soft)	btSoftBody::clusterDCImpulse(m_soft,impulse);
+			if (m_rigid) m_rigid->applyCentralImpulse(impulse);
+			if (m_soft) btSoftBody::clusterDCImpulse(m_soft, impulse);
 		}
 	};
-	/* Joint		*/ 
-	struct	Joint
+	/* Joint		*/
+	struct Joint
 	{
-		struct eType { enum _ {
-			Linear=0,
-			Angular,
-			Contact
-		};};
+		struct eType
+		{
+			enum _
+			{
+				Linear = 0,
+				Angular,
+				Contact
+			};
+		};
 		struct Specs
 		{
-			Specs() : erp(1),cfm(1),split(1) {}
-			btScalar	erp;
-			btScalar	cfm;
-			btScalar	split;
+			Specs() : erp(1), cfm(1), split(1) {}
+			btScalar erp;
+			btScalar cfm;
+			btScalar split;
 		};
-		Body						m_bodies[2];
-		btVector3					m_refs[2];
-		btScalar					m_cfm;
-		btScalar					m_erp;
-		btScalar					m_split;
-		btVector3					m_drift;
-		btVector3					m_sdrift;
-		btMatrix3x3					m_massmatrix;
-		bool						m_delete;
-		virtual						~Joint() {}
+		Body m_bodies[2];
+		btVector3 m_refs[2];
+		btScalar m_cfm;
+		btScalar m_erp;
+		btScalar m_split;
+		btVector3 m_drift;
+		btVector3 m_sdrift;
+		btMatrix3x3 m_massmatrix;
+		bool m_delete;
+		virtual ~Joint() {}
 		Joint() : m_delete(false) {}
-		virtual void				Prepare(btScalar dt,int iterations);
-		virtual void				Solve(btScalar dt,btScalar sor)=0;
-		virtual void				Terminate(btScalar dt)=0;
-		virtual eType::_			Type() const=0;
+		virtual void Prepare(btScalar dt, int iterations);
+		virtual void Solve(btScalar dt, btScalar sor) = 0;
+		virtual void Terminate(btScalar dt) = 0;
+		virtual eType::_ Type() const = 0;
 	};
-	/* LJoint		*/ 
-	struct	LJoint : Joint
+	/* LJoint		*/
+	struct LJoint : Joint
 	{
 		struct Specs : Joint::Specs
 		{
-			btVector3	position;
-		};		
-		btVector3					m_rpos[2];
-		void						Prepare(btScalar dt,int iterations);
-		void						Solve(btScalar dt,btScalar sor);
-		void						Terminate(btScalar dt);
-		eType::_					Type() const { return(eType::Linear); }
+			btVector3 position;
+		};
+		btVector3 m_rpos[2];
+		void Prepare(btScalar dt, int iterations);
+		void Solve(btScalar dt, btScalar sor);
+		void Terminate(btScalar dt);
+		eType::_ Type() const { return (eType::Linear); }
 	};
-	/* AJoint		*/ 
-	struct	AJoint : Joint
+	/* AJoint		*/
+	struct AJoint : Joint
 	{
 		struct IControl
 		{
 			virtual ~IControl() {}
-			virtual void			Prepare(AJoint*)				{}
-			virtual btScalar		Speed(AJoint*,btScalar current) { return(current); }
-			static IControl*		Default()						{ static IControl def;return(&def); }
+			virtual void Prepare(AJoint*) {}
+			virtual btScalar Speed(AJoint*, btScalar current) { return (current); }
+			static IControl* Default()
+			{
+				static IControl def;
+				return (&def);
+			}
 		};
 		struct Specs : Joint::Specs
 		{
 			Specs() : icontrol(IControl::Default()) {}
-			btVector3	axis;
-			IControl*	icontrol;
-		};		
-		btVector3					m_axis[2];
-		IControl*					m_icontrol;
-		void						Prepare(btScalar dt,int iterations);
-		void						Solve(btScalar dt,btScalar sor);
-		void						Terminate(btScalar dt);
-		eType::_					Type() const { return(eType::Angular); }
+			btVector3 axis;
+			IControl* icontrol;
+		};
+		btVector3 m_axis[2];
+		IControl* m_icontrol;
+		void Prepare(btScalar dt, int iterations);
+		void Solve(btScalar dt, btScalar sor);
+		void Terminate(btScalar dt);
+		eType::_ Type() const { return (eType::Angular); }
 	};
-	/* CJoint		*/ 
-	struct	CJoint : Joint
-	{		
-		int							m_life;
-		int							m_maxlife;
-		btVector3					m_rpos[2];
-		btVector3					m_normal;
-		btScalar					m_friction;
-		void						Prepare(btScalar dt,int iterations);
-		void						Solve(btScalar dt,btScalar sor);
-		void						Terminate(btScalar dt);
-		eType::_					Type() const { return(eType::Contact); }
-	};
-	/* Config		*/ 
-	struct	Config
+	/* CJoint		*/
+	struct CJoint : Joint
 	{
-		eAeroModel::_			aeromodel;		// Aerodynamic model (default: V_Point)
-		btScalar				kVCF;			// Velocities correction factor (Baumgarte)
-		btScalar				kDP;			// Damping coefficient [0,1]
-		btScalar				kDG;			// Drag coefficient [0,+inf]
-		btScalar				kLF;			// Lift coefficient [0,+inf]
-		btScalar				kPR;			// Pressure coefficient [-inf,+inf]
-		btScalar				kVC;			// Volume conversation coefficient [0,+inf]
-		btScalar				kDF;			// Dynamic friction coefficient [0,1]
-		btScalar				kMT;			// Pose matching coefficient [0,1]		
-		btScalar				kCHR;			// Rigid contacts hardness [0,1]
-		btScalar				kKHR;			// Kinetic contacts hardness [0,1]
-		btScalar				kSHR;			// Soft contacts hardness [0,1]
-		btScalar				kAHR;			// Anchors hardness [0,1]
-		btScalar				kSRHR_CL;		// Soft vs rigid hardness [0,1] (cluster only)
-		btScalar				kSKHR_CL;		// Soft vs kinetic hardness [0,1] (cluster only)
-		btScalar				kSSHR_CL;		// Soft vs soft hardness [0,1] (cluster only)
-		btScalar				kSR_SPLT_CL;	// Soft vs rigid impulse split [0,1] (cluster only)
-		btScalar				kSK_SPLT_CL;	// Soft vs rigid impulse split [0,1] (cluster only)
-		btScalar				kSS_SPLT_CL;	// Soft vs rigid impulse split [0,1] (cluster only)
-		btScalar				maxvolume;		// Maximum volume ratio for pose
-		btScalar				timescale;		// Time scale
-		int						viterations;	// Velocities solver iterations
-		int						piterations;	// Positions solver iterations
-		int						diterations;	// Drift solver iterations
-		int						citerations;	// Cluster solver iterations
-		int						collisions;		// Collisions flags
-		tVSolverArray			m_vsequence;	// Velocity solvers sequence
-		tPSolverArray			m_psequence;	// Position solvers sequence
-		tPSolverArray			m_dsequence;	// Drift solvers sequence
+		int m_life;
+		int m_maxlife;
+		btVector3 m_rpos[2];
+		btVector3 m_normal;
+		btScalar m_friction;
+		void Prepare(btScalar dt, int iterations);
+		void Solve(btScalar dt, btScalar sor);
+		void Terminate(btScalar dt);
+		eType::_ Type() const { return (eType::Contact); }
 	};
-	/* SolverState	*/ 
-	struct	SolverState
+	/* Config		*/
+	struct Config
 	{
-		btScalar				sdt;			// dt*timescale
-		btScalar				isdt;			// 1/sdt
-		btScalar				velmrg;			// velocity margin
-		btScalar				radmrg;			// radial margin
-		btScalar				updmrg;			// Update margin
-	};	
+		eAeroModel::_ aeromodel;    // Aerodynamic model (default: V_Point)
+		btScalar kVCF;              // Velocities correction factor (Baumgarte)
+		btScalar kDP;               // Damping coefficient [0,1]
+		btScalar kDG;               // Drag coefficient [0,+inf]
+		btScalar kLF;               // Lift coefficient [0,+inf]
+		btScalar kPR;               // Pressure coefficient [-inf,+inf]
+		btScalar kVC;               // Volume conversation coefficient [0,+inf]
+		btScalar kDF;               // Dynamic friction coefficient [0,1]
+		btScalar kMT;               // Pose matching coefficient [0,1]
+		btScalar kCHR;              // Rigid contacts hardness [0,1]
+		btScalar kKHR;              // Kinetic contacts hardness [0,1]
+		btScalar kSHR;              // Soft contacts hardness [0,1]
+		btScalar kAHR;              // Anchors hardness [0,1]
+		btScalar kSRHR_CL;          // Soft vs rigid hardness [0,1] (cluster only)
+		btScalar kSKHR_CL;          // Soft vs kinetic hardness [0,1] (cluster only)
+		btScalar kSSHR_CL;          // Soft vs soft hardness [0,1] (cluster only)
+		btScalar kSR_SPLT_CL;       // Soft vs rigid impulse split [0,1] (cluster only)
+		btScalar kSK_SPLT_CL;       // Soft vs rigid impulse split [0,1] (cluster only)
+		btScalar kSS_SPLT_CL;       // Soft vs rigid impulse split [0,1] (cluster only)
+		btScalar maxvolume;         // Maximum volume ratio for pose
+		btScalar timescale;         // Time scale
+		int viterations;            // Velocities solver iterations
+		int piterations;            // Positions solver iterations
+		int diterations;            // Drift solver iterations
+		int citerations;            // Cluster solver iterations
+		int collisions;             // Collisions flags
+		tVSolverArray m_vsequence;  // Velocity solvers sequence
+		tPSolverArray m_psequence;  // Position solvers sequence
+		tPSolverArray m_dsequence;  // Drift solvers sequence
+	};
+	/* SolverState	*/
+	struct SolverState
+	{
+		btScalar sdt;     // dt*timescale
+		btScalar isdt;    // 1/sdt
+		btScalar velmrg;  // velocity margin
+		btScalar radmrg;  // radial margin
+		btScalar updmrg;  // Update margin
+	};
 	/// RayFromToCaster takes a ray from, ray to (instead of direction!)
-	struct	RayFromToCaster : btDbvt::ICollide
+	struct RayFromToCaster : btDbvt::ICollide
 	{
-		btVector3			m_rayFrom;
-		btVector3			m_rayTo;
-		btVector3			m_rayNormalizedDirection;
-		btScalar			m_mint;
-		Face*				m_face;
-		int					m_tests;
-		RayFromToCaster(const btVector3& rayFrom,const btVector3& rayTo,btScalar mxt);
-		void					Process(const btDbvtNode* leaf);
+		btVector3 m_rayFrom;
+		btVector3 m_rayTo;
+		btVector3 m_rayNormalizedDirection;
+		btScalar m_mint;
+		Face* m_face;
+		int m_tests;
+		RayFromToCaster(const btVector3& rayFrom, const btVector3& rayTo, btScalar mxt);
+		void Process(const btDbvtNode* leaf);
 
-		static /*inline*/ btScalar	rayFromToTriangle(const btVector3& rayFrom,
-			const btVector3& rayTo,
-			const btVector3& rayNormalizedDirection,
-			const btVector3& a,
-			const btVector3& b,
-			const btVector3& c,
-			btScalar maxt=SIMD_INFINITY);
+		static /*inline*/ btScalar rayFromToTriangle(const btVector3& rayFrom,
+													 const btVector3& rayTo,
+													 const btVector3& rayNormalizedDirection,
+													 const btVector3& a,
+													 const btVector3& b,
+													 const btVector3& c,
+													 btScalar maxt = SIMD_INFINITY);
 	};
 
 	//
 	// Typedefs
 	//
 
-	typedef void								(*psolver_t)(btSoftBody*,btScalar,btScalar);
-	typedef void								(*vsolver_t)(btSoftBody*,btScalar);
-	typedef btAlignedObjectArray<Cluster*>		tClusterArray;
-	typedef btAlignedObjectArray<Note>			tNoteArray;
-	typedef btAlignedObjectArray<Node>			tNodeArray;
-	typedef btAlignedObjectArray<btDbvtNode*>	tLeafArray;
-	typedef btAlignedObjectArray<Link>			tLinkArray;
-	typedef btAlignedObjectArray<Face>			tFaceArray;
-	typedef btAlignedObjectArray<Tetra>			tTetraArray;
-	typedef btAlignedObjectArray<Anchor>		tAnchorArray;
-	typedef btAlignedObjectArray<RContact>		tRContactArray;
-	typedef btAlignedObjectArray<SContact>		tSContactArray;
-	typedef btAlignedObjectArray<Material*>		tMaterialArray;
-	typedef btAlignedObjectArray<Joint*>		tJointArray;
-	typedef btAlignedObjectArray<btSoftBody*>	tSoftBodyArray;	
+	typedef void (*psolver_t)(btSoftBody*, btScalar, btScalar);
+	typedef void (*vsolver_t)(btSoftBody*, btScalar);
+	typedef btAlignedObjectArray<Cluster*> tClusterArray;
+	typedef btAlignedObjectArray<Note> tNoteArray;
+	typedef btAlignedObjectArray<Node> tNodeArray;
+	typedef btAlignedObjectArray<btDbvtNode*> tLeafArray;
+	typedef btAlignedObjectArray<Link> tLinkArray;
+	typedef btAlignedObjectArray<Face> tFaceArray;
+	typedef btAlignedObjectArray<Tetra> tTetraArray;
+	typedef btAlignedObjectArray<Anchor> tAnchorArray;
+	typedef btAlignedObjectArray<RContact> tRContactArray;
+	typedef btAlignedObjectArray<SContact> tSContactArray;
+	typedef btAlignedObjectArray<Material*> tMaterialArray;
+	typedef btAlignedObjectArray<Joint*> tJointArray;
+	typedef btAlignedObjectArray<btSoftBody*> tSoftBodyArray;
 
 	//
 	// Fields
 	//
 
-	Config					m_cfg;			// Configuration
-	SolverState				m_sst;			// Solver state
-	Pose					m_pose;			// Pose
-	void*					m_tag;			// User data
-	btSoftBodyWorldInfo*	m_worldInfo;	// World info
-	tNoteArray				m_notes;		// Notes
-	tNodeArray				m_nodes;		// Nodes
-	tLinkArray				m_links;		// Links
-	tFaceArray				m_faces;		// Faces
-	tTetraArray				m_tetras;		// Tetras
-	tAnchorArray			m_anchors;		// Anchors
-	tRContactArray			m_rcontacts;	// Rigid contacts
-	tSContactArray			m_scontacts;	// Soft contacts
-	tJointArray				m_joints;		// Joints
-	tMaterialArray			m_materials;	// Materials
-	btScalar				m_timeacc;		// Time accumulator
-	btVector3				m_bounds[2];	// Spatial bounds	
-	bool					m_bUpdateRtCst;	// Update runtime constants
-	btDbvt					m_ndbvt;		// Nodes tree
-	btDbvt					m_fdbvt;		// Faces tree
-	btDbvt					m_cdbvt;		// Clusters tree
-	tClusterArray			m_clusters;		// Clusters
+	Config m_cfg;                      // Configuration
+	SolverState m_sst;                 // Solver state
+	Pose m_pose;                       // Pose
+	void* m_tag;                       // User data
+	btSoftBodyWorldInfo* m_worldInfo;  // World info
+	tNoteArray m_notes;                // Notes
+	tNodeArray m_nodes;                // Nodes
+	tLinkArray m_links;                // Links
+	tFaceArray m_faces;                // Faces
+	tTetraArray m_tetras;              // Tetras
+	tAnchorArray m_anchors;            // Anchors
+	tRContactArray m_rcontacts;        // Rigid contacts
+	tSContactArray m_scontacts;        // Soft contacts
+	tJointArray m_joints;              // Joints
+	tMaterialArray m_materials;        // Materials
+	btScalar m_timeacc;                // Time accumulator
+	btVector3 m_bounds[2];             // Spatial bounds
+	bool m_bUpdateRtCst;               // Update runtime constants
+	btDbvt m_ndbvt;                    // Nodes tree
+	btDbvt m_fdbvt;                    // Faces tree
+	btDbvt m_cdbvt;                    // Clusters tree
+	tClusterArray m_clusters;          // Clusters
 
-	btAlignedObjectArray<bool>m_clusterConnectivity;//cluster connectivity, for self-collision
+	btAlignedObjectArray<bool> m_clusterConnectivity;  //cluster connectivity, for self-collision
 
-	btTransform			m_initialWorldTransform;
+	btTransform m_initialWorldTransform;
+
+	btVector3 m_windVelocity;
+
+	btScalar m_restLengthScale;
 
-	btVector3			m_windVelocity;
-	
-	btScalar        m_restLengthScale;
-	
 	//
 	// Api
 	//
 
-	/* ctor																	*/ 
-	btSoftBody(	btSoftBodyWorldInfo* worldInfo,int node_count,		const btVector3* x,		const btScalar* m);
+	/* ctor																	*/
+	btSoftBody(btSoftBodyWorldInfo* worldInfo, int node_count, const btVector3* x, const btScalar* m);
 
-	/* ctor																	*/ 
-	btSoftBody(	btSoftBodyWorldInfo* worldInfo);
+	/* ctor																	*/
+	btSoftBody(btSoftBodyWorldInfo* worldInfo);
 
-	void	initDefaults();
+	void initDefaults();
 
-	/* dtor																	*/ 
+	/* dtor																	*/
 	virtual ~btSoftBody();
-	/* Check for existing link												*/ 
+	/* Check for existing link												*/
 
-	btAlignedObjectArray<int>	m_userIndexMapping;
+	btAlignedObjectArray<int> m_userIndexMapping;
 
-	btSoftBodyWorldInfo*	getWorldInfo()
+	btSoftBodyWorldInfo* getWorldInfo()
 	{
 		return m_worldInfo;
 	}
 
 	///@todo: avoid internal softbody shape hack and move collision code to collision library
-	virtual void	setCollisionShape(btCollisionShape* collisionShape)
+	virtual void setCollisionShape(btCollisionShape* collisionShape)
 	{
-		
 	}
 
-	bool				checkLink(	int node0,
-		int node1) const;
-	bool				checkLink(	const Node* node0,
-		const Node* node1) const;
-	/* Check for existring face												*/ 
-	bool				checkFace(	int node0,
-		int node1,
-		int node2) const;
-	/* Append material														*/ 
-	Material*			appendMaterial();
-	/* Append note															*/ 
-	void				appendNote(	const char* text,
-		const btVector3& o,
-		const btVector4& c=btVector4(1,0,0,0),
-		Node* n0=0,
-		Node* n1=0,
-		Node* n2=0,
-		Node* n3=0);
-	void				appendNote(	const char* text,
-		const btVector3& o,
-		Node* feature);
-	void				appendNote(	const char* text,
-		const btVector3& o,
-		Link* feature);
-	void				appendNote(	const char* text,
-		const btVector3& o,
-		Face* feature);
-	/* Append node															*/ 
-	void				appendNode(	const btVector3& x,btScalar m);
-	/* Append link															*/ 
-	void				appendLink(int model=-1,Material* mat=0);
-	void				appendLink(	int node0,
-		int node1,
-		Material* mat=0,
-		bool bcheckexist=false);
-	void				appendLink(	Node* node0,
-		Node* node1,
-		Material* mat=0,
-		bool bcheckexist=false);
-	/* Append face															*/ 
-	void				appendFace(int model=-1,Material* mat=0);
-	void				appendFace(	int node0,
-		int node1,
-		int node2,
-		Material* mat=0);
-	void			appendTetra(int model,Material* mat);
+	bool checkLink(int node0,
+				   int node1) const;
+	bool checkLink(const Node* node0,
+				   const Node* node1) const;
+	/* Check for existring face												*/
+	bool checkFace(int node0,
+				   int node1,
+				   int node2) const;
+	/* Append material														*/
+	Material* appendMaterial();
+	/* Append note															*/
+	void appendNote(const char* text,
+					const btVector3& o,
+					const btVector4& c = btVector4(1, 0, 0, 0),
+					Node* n0 = 0,
+					Node* n1 = 0,
+					Node* n2 = 0,
+					Node* n3 = 0);
+	void appendNote(const char* text,
+					const btVector3& o,
+					Node* feature);
+	void appendNote(const char* text,
+					const btVector3& o,
+					Link* feature);
+	void appendNote(const char* text,
+					const btVector3& o,
+					Face* feature);
+	/* Append node															*/
+	void appendNode(const btVector3& x, btScalar m);
+	/* Append link															*/
+	void appendLink(int model = -1, Material* mat = 0);
+	void appendLink(int node0,
+					int node1,
+					Material* mat = 0,
+					bool bcheckexist = false);
+	void appendLink(Node* node0,
+					Node* node1,
+					Material* mat = 0,
+					bool bcheckexist = false);
+	/* Append face															*/
+	void appendFace(int model = -1, Material* mat = 0);
+	void appendFace(int node0,
+					int node1,
+					int node2,
+					Material* mat = 0);
+	void appendTetra(int model, Material* mat);
 	//
-	void			appendTetra(int node0,
-										int node1,
-										int node2,
-										int node3,
-										Material* mat=0);
+	void appendTetra(int node0,
+					 int node1,
+					 int node2,
+					 int node3,
+					 Material* mat = 0);
 
-
-	/* Append anchor														*/ 
-	void				appendAnchor(	int node,
-		btRigidBody* body, bool disableCollisionBetweenLinkedBodies=false,btScalar influence = 1);
-	void			appendAnchor(int node,btRigidBody* body, const btVector3& localPivot,bool disableCollisionBetweenLinkedBodies=false,btScalar influence = 1);
-	/* Append linear joint													*/ 
-	void				appendLinearJoint(const LJoint::Specs& specs,Cluster* body0,Body body1);
-	void				appendLinearJoint(const LJoint::Specs& specs,Body body=Body());
-	void				appendLinearJoint(const LJoint::Specs& specs,btSoftBody* body);
-	/* Append linear joint													*/ 
-	void				appendAngularJoint(const AJoint::Specs& specs,Cluster* body0,Body body1);
-	void				appendAngularJoint(const AJoint::Specs& specs,Body body=Body());
-	void				appendAngularJoint(const AJoint::Specs& specs,btSoftBody* body);
-	/* Add force (or gravity) to the entire body							*/ 
-	void				addForce(		const btVector3& force);
-	/* Add force (or gravity) to a node of the body							*/ 
-	void				addForce(		const btVector3& force,
-		int node);
+	/* Append anchor														*/
+	void appendAnchor(int node,
+					  btRigidBody* body, bool disableCollisionBetweenLinkedBodies = false, btScalar influence = 1);
+	void appendAnchor(int node, btRigidBody* body, const btVector3& localPivot, bool disableCollisionBetweenLinkedBodies = false, btScalar influence = 1);
+	/* Append linear joint													*/
+	void appendLinearJoint(const LJoint::Specs& specs, Cluster* body0, Body body1);
+	void appendLinearJoint(const LJoint::Specs& specs, Body body = Body());
+	void appendLinearJoint(const LJoint::Specs& specs, btSoftBody* body);
+	/* Append linear joint													*/
+	void appendAngularJoint(const AJoint::Specs& specs, Cluster* body0, Body body1);
+	void appendAngularJoint(const AJoint::Specs& specs, Body body = Body());
+	void appendAngularJoint(const AJoint::Specs& specs, btSoftBody* body);
+	/* Add force (or gravity) to the entire body							*/
+	void addForce(const btVector3& force);
+	/* Add force (or gravity) to a node of the body							*/
+	void addForce(const btVector3& force,
+				  int node);
 	/* Add aero force to a node of the body */
-	void			    addAeroForceToNode(const btVector3& windVelocity,int nodeIndex);
+	void addAeroForceToNode(const btVector3& windVelocity, int nodeIndex);
 
 	/* Add aero force to a face of the body */
-	void			    addAeroForceToFace(const btVector3& windVelocity,int faceIndex);
+	void addAeroForceToFace(const btVector3& windVelocity, int faceIndex);
 
-	/* Add velocity to the entire body										*/ 
-	void				addVelocity(	const btVector3& velocity);
+	/* Add velocity to the entire body										*/
+	void addVelocity(const btVector3& velocity);
 
-	/* Set velocity for the entire body										*/ 
-	void				setVelocity(	const btVector3& velocity);
+	/* Set velocity for the entire body										*/
+	void setVelocity(const btVector3& velocity);
 
-	/* Add velocity to a node of the body									*/ 
-	void				addVelocity(	const btVector3& velocity,
-		int node);
-	/* Set mass																*/ 
-	void				setMass(		int node,
-		btScalar mass);
-	/* Get mass																*/ 
-	btScalar			getMass(		int node) const;
-	/* Get total mass														*/ 
-	btScalar			getTotalMass() const;
-	/* Set total mass (weighted by previous masses)							*/ 
-	void				setTotalMass(	btScalar mass,
-		bool fromfaces=false);
-	/* Set total density													*/ 
-	void				setTotalDensity(btScalar density);
+	/* Add velocity to a node of the body									*/
+	void addVelocity(const btVector3& velocity,
+					 int node);
+	/* Set mass																*/
+	void setMass(int node,
+				 btScalar mass);
+	/* Get mass																*/
+	btScalar getMass(int node) const;
+	/* Get total mass														*/
+	btScalar getTotalMass() const;
+	/* Set total mass (weighted by previous masses)							*/
+	void setTotalMass(btScalar mass,
+					  bool fromfaces = false);
+	/* Set total density													*/
+	void setTotalDensity(btScalar density);
 	/* Set volume mass (using tetrahedrons)									*/
-	void				setVolumeMass(		btScalar mass);
+	void setVolumeMass(btScalar mass);
 	/* Set volume density (using tetrahedrons)								*/
-	void				setVolumeDensity(	btScalar density);
-	/* Transform															*/ 
-	void				transform(		const btTransform& trs);
-	/* Translate															*/ 
-	void				translate(		const btVector3& trs);
-	/* Rotate															*/ 
-	void				rotate(	const btQuaternion& rot);
-	/* Scale																*/ 
-	void				scale(	const btVector3& scl);
+	void setVolumeDensity(btScalar density);
+	/* Transform															*/
+	void transform(const btTransform& trs);
+	/* Translate															*/
+	void translate(const btVector3& trs);
+	/* Rotate															*/
+	void rotate(const btQuaternion& rot);
+	/* Scale																*/
+	void scale(const btVector3& scl);
 	/* Get link resting lengths scale										*/
-	btScalar			getRestLengthScale();
+	btScalar getRestLengthScale();
 	/* Scale resting length of all springs									*/
-	void				setRestLengthScale(btScalar restLength);
-	/* Set current state as pose											*/ 
-	void				setPose(		bool bvolume,
-		bool bframe);
-	/* Set current link lengths as resting lengths							*/ 
-	void				resetLinkRestLengths();
-	/* Return the volume													*/ 
-	btScalar			getVolume() const;
-	/* Cluster count														*/ 
-	int					clusterCount() const;
-	/* Cluster center of mass												*/ 
-	static btVector3	clusterCom(const Cluster* cluster);
-	btVector3			clusterCom(int cluster) const;
-	/* Cluster velocity at rpos												*/ 
-	static btVector3	clusterVelocity(const Cluster* cluster,const btVector3& rpos);
-	/* Cluster impulse														*/ 
-	static void			clusterVImpulse(Cluster* cluster,const btVector3& rpos,const btVector3& impulse);
-	static void			clusterDImpulse(Cluster* cluster,const btVector3& rpos,const btVector3& impulse);
-	static void			clusterImpulse(Cluster* cluster,const btVector3& rpos,const Impulse& impulse);
-	static void			clusterVAImpulse(Cluster* cluster,const btVector3& impulse);
-	static void			clusterDAImpulse(Cluster* cluster,const btVector3& impulse);
-	static void			clusterAImpulse(Cluster* cluster,const Impulse& impulse);
-	static void			clusterDCImpulse(Cluster* cluster,const btVector3& impulse);
-	/* Generate bending constraints based on distance in the adjency graph	*/ 
-	int					generateBendingConstraints(	int distance,
-		Material* mat=0);
-	/* Randomize constraints to reduce solver bias							*/ 
-	void				randomizeConstraints();
-	/* Release clusters														*/ 
-	void				releaseCluster(int index);
-	void				releaseClusters();
-	/* Generate clusters (K-mean)											*/ 
+	void setRestLengthScale(btScalar restLength);
+	/* Set current state as pose											*/
+	void setPose(bool bvolume,
+				 bool bframe);
+	/* Set current link lengths as resting lengths							*/
+	void resetLinkRestLengths();
+	/* Return the volume													*/
+	btScalar getVolume() const;
+	/* Cluster count														*/
+	int clusterCount() const;
+	/* Cluster center of mass												*/
+	static btVector3 clusterCom(const Cluster* cluster);
+	btVector3 clusterCom(int cluster) const;
+	/* Cluster velocity at rpos												*/
+	static btVector3 clusterVelocity(const Cluster* cluster, const btVector3& rpos);
+	/* Cluster impulse														*/
+	static void clusterVImpulse(Cluster* cluster, const btVector3& rpos, const btVector3& impulse);
+	static void clusterDImpulse(Cluster* cluster, const btVector3& rpos, const btVector3& impulse);
+	static void clusterImpulse(Cluster* cluster, const btVector3& rpos, const Impulse& impulse);
+	static void clusterVAImpulse(Cluster* cluster, const btVector3& impulse);
+	static void clusterDAImpulse(Cluster* cluster, const btVector3& impulse);
+	static void clusterAImpulse(Cluster* cluster, const Impulse& impulse);
+	static void clusterDCImpulse(Cluster* cluster, const btVector3& impulse);
+	/* Generate bending constraints based on distance in the adjency graph	*/
+	int generateBendingConstraints(int distance,
+								   Material* mat = 0);
+	/* Randomize constraints to reduce solver bias							*/
+	void randomizeConstraints();
+	/* Release clusters														*/
+	void releaseCluster(int index);
+	void releaseClusters();
+	/* Generate clusters (K-mean)											*/
 	///generateClusters with k=0 will create a convex cluster for each tetrahedron or triangle
 	///otherwise an approximation will be used (better performance)
-	int					generateClusters(int k,int maxiterations=8192);
-	/* Refine																*/ 
-	void				refine(ImplicitFn* ifn,btScalar accurary,bool cut);
-	/* CutLink																*/ 
-	bool				cutLink(int node0,int node1,btScalar position);
-	bool				cutLink(const Node* node0,const Node* node1,btScalar position);
+	int generateClusters(int k, int maxiterations = 8192);
+	/* Refine																*/
+	void refine(ImplicitFn* ifn, btScalar accurary, bool cut);
+	/* CutLink																*/
+	bool cutLink(int node0, int node1, btScalar position);
+	bool cutLink(const Node* node0, const Node* node1, btScalar position);
 
 	///Ray casting using rayFrom and rayTo in worldspace, (not direction!)
-	bool				rayTest(const btVector3& rayFrom,
-		const btVector3& rayTo,
-		sRayCast& results);
-	/* Solver presets														*/ 
-	void				setSolver(eSolverPresets::_ preset);
-	/* predictMotion														*/ 
-	void				predictMotion(btScalar dt);
-	/* solveConstraints														*/ 
-	void				solveConstraints();
-	/* staticSolve															*/ 
-	void				staticSolve(int iterations);
-	/* solveCommonConstraints												*/ 
-	static void			solveCommonConstraints(btSoftBody** bodies,int count,int iterations);
-	/* solveClusters														*/ 
-	static void			solveClusters(const btAlignedObjectArray<btSoftBody*>& bodies);
-	/* integrateMotion														*/ 
-	void				integrateMotion();
-	/* defaultCollisionHandlers												*/ 
-	void				defaultCollisionHandler(const btCollisionObjectWrapper* pcoWrap);
-	void				defaultCollisionHandler(btSoftBody* psb);
-
-
+	bool rayTest(const btVector3& rayFrom,
+				 const btVector3& rayTo,
+				 sRayCast& results);
+	/* Solver presets														*/
+	void setSolver(eSolverPresets::_ preset);
+	/* predictMotion														*/
+	void predictMotion(btScalar dt);
+	/* solveConstraints														*/
+	void solveConstraints();
+	/* staticSolve															*/
+	void staticSolve(int iterations);
+	/* solveCommonConstraints												*/
+	static void solveCommonConstraints(btSoftBody** bodies, int count, int iterations);
+	/* solveClusters														*/
+	static void solveClusters(const btAlignedObjectArray<btSoftBody*>& bodies);
+	/* integrateMotion														*/
+	void integrateMotion();
+	/* defaultCollisionHandlers												*/
+	void defaultCollisionHandler(const btCollisionObjectWrapper* pcoWrap);
+	void defaultCollisionHandler(btSoftBody* psb);
 
 	//
 	// Functionality to deal with new accelerated solvers.
@@ -895,8 +923,7 @@ public:
 	/**
 	 * Set a wind velocity for interaction with the air.
 	 */
-	void setWindVelocity( const btVector3 &velocity );
-
+	void setWindVelocity(const btVector3& velocity);
 
 	/**
 	 * Return the wind velocity for interaction with the air.
@@ -907,41 +934,40 @@ public:
 	// Set the solver that handles this soft body
 	// Should not be allowed to get out of sync with reality
 	// Currently called internally on addition to the world
-	void setSoftBodySolver( btSoftBodySolver *softBodySolver )
+	void setSoftBodySolver(btSoftBodySolver* softBodySolver)
 	{
 		m_softBodySolver = softBodySolver;
 	}
 
 	//
 	// Return the solver that handles this soft body
-	// 
-	btSoftBodySolver *getSoftBodySolver()
+	//
+	btSoftBodySolver* getSoftBodySolver()
 	{
 		return m_softBodySolver;
 	}
 
 	//
 	// Return the solver that handles this soft body
-	// 
-	btSoftBodySolver *getSoftBodySolver() const
+	//
+	btSoftBodySolver* getSoftBodySolver() const
 	{
 		return m_softBodySolver;
 	}
 
-
 	//
 	// Cast
 	//
 
-	static const btSoftBody*	upcast(const btCollisionObject* colObj)
+	static const btSoftBody* upcast(const btCollisionObject* colObj)
 	{
-		if (colObj->getInternalType()==CO_SOFT_BODY)
+		if (colObj->getInternalType() == CO_SOFT_BODY)
 			return (const btSoftBody*)colObj;
 		return 0;
 	}
-	static btSoftBody*			upcast(btCollisionObject* colObj)
+	static btSoftBody* upcast(btCollisionObject* colObj)
 	{
-		if (colObj->getInternalType()==CO_SOFT_BODY)
+		if (colObj->getInternalType() == CO_SOFT_BODY)
 			return (btSoftBody*)colObj;
 		return 0;
 	}
@@ -950,7 +976,7 @@ public:
 	// ::btCollisionObject
 	//
 
-	virtual void getAabb(btVector3& aabbMin,btVector3& aabbMax) const
+	virtual void getAabb(btVector3& aabbMin, btVector3& aabbMax) const
 	{
 		aabbMin = m_bounds[0];
 		aabbMax = m_bounds[1];
@@ -958,48 +984,42 @@ public:
 	//
 	// Private
 	//
-	void				pointersToIndices();
-	void				indicesToPointers(const int* map=0);
+	void pointersToIndices();
+	void indicesToPointers(const int* map = 0);
 
-	int					rayTest(const btVector3& rayFrom,const btVector3& rayTo,
-		btScalar& mint,eFeature::_& feature,int& index,bool bcountonly) const;
-	void				initializeFaceTree();
-	btVector3			evaluateCom() const;
-	bool				checkContact(const btCollisionObjectWrapper* colObjWrap,const btVector3& x,btScalar margin,btSoftBody::sCti& cti) const;
-	void				updateNormals();
-	void				updateBounds();
-	void				updatePose();
-	void				updateConstants();
-	void				updateLinkConstants();
-	void				updateArea(bool averageArea = true);
-	void				initializeClusters();
-	void				updateClusters();
-	void				cleanupClusters();
-	void				prepareClusters(int iterations);
-	void				solveClusters(btScalar sor);
-	void				applyClusters(bool drift);
-	void				dampClusters();
-	void				applyForces();	
-	static void			PSolve_Anchors(btSoftBody* psb,btScalar kst,btScalar ti);
-	static void			PSolve_RContacts(btSoftBody* psb,btScalar kst,btScalar ti);
-	static void			PSolve_SContacts(btSoftBody* psb,btScalar,btScalar ti);
-	static void			PSolve_Links(btSoftBody* psb,btScalar kst,btScalar ti);
-	static void			VSolve_Links(btSoftBody* psb,btScalar kst);
-	static psolver_t	getSolver(ePSolver::_ solver);
-	static vsolver_t	getSolver(eVSolver::_ solver);
+	int rayTest(const btVector3& rayFrom, const btVector3& rayTo,
+				btScalar& mint, eFeature::_& feature, int& index, bool bcountonly) const;
+	void initializeFaceTree();
+	btVector3 evaluateCom() const;
+	bool checkContact(const btCollisionObjectWrapper* colObjWrap, const btVector3& x, btScalar margin, btSoftBody::sCti& cti) const;
+	void updateNormals();
+	void updateBounds();
+	void updatePose();
+	void updateConstants();
+	void updateLinkConstants();
+	void updateArea(bool averageArea = true);
+	void initializeClusters();
+	void updateClusters();
+	void cleanupClusters();
+	void prepareClusters(int iterations);
+	void solveClusters(btScalar sor);
+	void applyClusters(bool drift);
+	void dampClusters();
+	void applyForces();
+	static void PSolve_Anchors(btSoftBody* psb, btScalar kst, btScalar ti);
+	static void PSolve_RContacts(btSoftBody* psb, btScalar kst, btScalar ti);
+	static void PSolve_SContacts(btSoftBody* psb, btScalar, btScalar ti);
+	static void PSolve_Links(btSoftBody* psb, btScalar kst, btScalar ti);
+	static void VSolve_Links(btSoftBody* psb, btScalar kst);
+	static psolver_t getSolver(ePSolver::_ solver);
+	static vsolver_t getSolver(eVSolver::_ solver);
 
-
-	virtual	int	calculateSerializeBufferSize()	const;
+	virtual int calculateSerializeBufferSize() const;
 
 	///fills the dataBuffer and returns the struct name (and 0 on failure)
-	virtual	const char*	serialize(void* dataBuffer,  class btSerializer* serializer) const;
+	virtual const char* serialize(void* dataBuffer, class btSerializer* serializer) const;
 
 	//virtual void serializeSingleObject(class btSerializer* serializer) const;
-
-
 };
 
-
-
-
-#endif //_BT_SOFT_BODY_H
+#endif  //_BT_SOFT_BODY_H
diff --git a/thirdparty/bullet/BulletSoftBody/btSoftBodyConcaveCollisionAlgorithm.cpp b/thirdparty/bullet/BulletSoftBody/btSoftBodyConcaveCollisionAlgorithm.cpp
index ab84bddf2a6..750718f57f8 100644
--- a/thirdparty/bullet/BulletSoftBody/btSoftBodyConcaveCollisionAlgorithm.cpp
+++ b/thirdparty/bullet/BulletSoftBody/btSoftBodyConcaveCollisionAlgorithm.cpp
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btSoftBodyConcaveCollisionAlgorithm.h"
 #include "BulletCollision/CollisionDispatch/btCollisionObject.h"
 #include "BulletCollision/CollisionShapes/btMultiSphereShape.h"
@@ -27,34 +26,28 @@ subject to the following restrictions:
 #include "BulletCollision/CollisionShapes/btConvexHullShape.h"
 #include "BulletCollision/CollisionDispatch/btCollisionObjectWrapper.h"
 
-
 #include "LinearMath/btIDebugDraw.h"
 #include "BulletCollision/NarrowPhaseCollision/btSubSimplexConvexCast.h"
 #include "BulletSoftBody/btSoftBody.h"
 
-#define BT_SOFTBODY_TRIANGLE_EXTRUSION btScalar(0.06)//make this configurable
+#define BT_SOFTBODY_TRIANGLE_EXTRUSION btScalar(0.06)  //make this configurable
 
-btSoftBodyConcaveCollisionAlgorithm::btSoftBodyConcaveCollisionAlgorithm( const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,bool isSwapped)
-: btCollisionAlgorithm(ci),
-m_isSwapped(isSwapped),
-m_btSoftBodyTriangleCallback(ci.m_dispatcher1,body0Wrap,body1Wrap,isSwapped)
+btSoftBodyConcaveCollisionAlgorithm::btSoftBodyConcaveCollisionAlgorithm(const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, bool isSwapped)
+	: btCollisionAlgorithm(ci),
+	  m_isSwapped(isSwapped),
+	  m_btSoftBodyTriangleCallback(ci.m_dispatcher1, body0Wrap, body1Wrap, isSwapped)
 {
 }
 
-
-
 btSoftBodyConcaveCollisionAlgorithm::~btSoftBodyConcaveCollisionAlgorithm()
 {
 }
 
-
-
-btSoftBodyTriangleCallback::btSoftBodyTriangleCallback(btDispatcher*  dispatcher,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,bool isSwapped):
-m_dispatcher(dispatcher),
-m_dispatchInfoPtr(0)
+btSoftBodyTriangleCallback::btSoftBodyTriangleCallback(btDispatcher* dispatcher, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, bool isSwapped) : m_dispatcher(dispatcher),
+																																														 m_dispatchInfoPtr(0)
 {
-	m_softBody = (isSwapped? (btSoftBody*)body1Wrap->getCollisionObject():(btSoftBody*)body0Wrap->getCollisionObject());
-	m_triBody = isSwapped? body0Wrap->getCollisionObject():body1Wrap->getCollisionObject();
+	m_softBody = (isSwapped ? (btSoftBody*)body1Wrap->getCollisionObject() : (btSoftBody*)body0Wrap->getCollisionObject());
+	m_triBody = isSwapped ? body0Wrap->getCollisionObject() : body1Wrap->getCollisionObject();
 
 	//
 	// create the manifold from the dispatcher 'manifold pool'
@@ -68,46 +61,42 @@ btSoftBodyTriangleCallback::~btSoftBodyTriangleCallback()
 {
 	clearCache();
 	//	m_dispatcher->releaseManifold( m_manifoldPtr );
-
 }
 
-
-void	btSoftBodyTriangleCallback::clearCache()
+void btSoftBodyTriangleCallback::clearCache()
 {
-	for (int i=0;i<m_shapeCache.size();i++)
+	for (int i = 0; i < m_shapeCache.size(); i++)
 	{
 		btTriIndex* tmp = m_shapeCache.getAtIndex(i);
 		btAssert(tmp);
 		btAssert(tmp->m_childShape);
-		m_softBody->getWorldInfo()->m_sparsesdf.RemoveReferences(tmp->m_childShape);//necessary?
+		m_softBody->getWorldInfo()->m_sparsesdf.RemoveReferences(tmp->m_childShape);  //necessary?
 		delete tmp->m_childShape;
 	}
 	m_shapeCache.clear();
 }
 
-
-void btSoftBodyTriangleCallback::processTriangle(btVector3* triangle,int partId, int triangleIndex)
+void btSoftBodyTriangleCallback::processTriangle(btVector3* triangle, int partId, int triangleIndex)
 {
 	//just for debugging purposes
 	//printf("triangle %d",m_triangleCount++);
-	
+
 	btCollisionAlgorithmConstructionInfo ci;
 	ci.m_dispatcher1 = m_dispatcher;
 
 	///debug drawing of the overlapping triangles
-	if (m_dispatchInfoPtr && m_dispatchInfoPtr->m_debugDraw && (m_dispatchInfoPtr->m_debugDraw->getDebugMode() &btIDebugDraw::DBG_DrawWireframe))
+	if (m_dispatchInfoPtr && m_dispatchInfoPtr->m_debugDraw && (m_dispatchInfoPtr->m_debugDraw->getDebugMode() & btIDebugDraw::DBG_DrawWireframe))
 	{
-		btVector3 color(1,1,0);
+		btVector3 color(1, 1, 0);
 		const btTransform& tr = m_triBody->getWorldTransform();
-		m_dispatchInfoPtr->m_debugDraw->drawLine(tr(triangle[0]),tr(triangle[1]),color);
-		m_dispatchInfoPtr->m_debugDraw->drawLine(tr(triangle[1]),tr(triangle[2]),color);
-		m_dispatchInfoPtr->m_debugDraw->drawLine(tr(triangle[2]),tr(triangle[0]),color);
+		m_dispatchInfoPtr->m_debugDraw->drawLine(tr(triangle[0]), tr(triangle[1]), color);
+		m_dispatchInfoPtr->m_debugDraw->drawLine(tr(triangle[1]), tr(triangle[2]), color);
+		m_dispatchInfoPtr->m_debugDraw->drawLine(tr(triangle[2]), tr(triangle[0]), color);
 	}
 
-	btTriIndex	triIndex(partId,triangleIndex,0);
+	btTriIndex triIndex(partId, triangleIndex, 0);
 	btHashKey<btTriIndex> triKey(triIndex.getUid());
 
-
 	btTriIndex* shapeIndex = m_shapeCache[triKey];
 	if (shapeIndex)
 	{
@@ -117,82 +106,73 @@ void btSoftBodyTriangleCallback::processTriangle(btVector3* triangle,int partId,
 		//copy over user pointers to temporary shape
 		tm->setUserPointer(m_triBody->getCollisionShape()->getUserPointer());
 
-		btCollisionObjectWrapper softBody(0,m_softBody->getCollisionShape(),m_softBody,m_softBody->getWorldTransform(),-1,-1);
+		btCollisionObjectWrapper softBody(0, m_softBody->getCollisionShape(), m_softBody, m_softBody->getWorldTransform(), -1, -1);
 		//btCollisionObjectWrapper triBody(0,tm, ob, btTransform::getIdentity());//ob->getWorldTransform());//??
-		btCollisionObjectWrapper triBody(0,tm, m_triBody, m_triBody->getWorldTransform(),partId, triangleIndex);
+		btCollisionObjectWrapper triBody(0, tm, m_triBody, m_triBody->getWorldTransform(), partId, triangleIndex);
 		ebtDispatcherQueryType algoType = m_resultOut->m_closestPointDistanceThreshold > 0 ? BT_CLOSEST_POINT_ALGORITHMS : BT_CONTACT_POINT_ALGORITHMS;
-		btCollisionAlgorithm* colAlgo = ci.m_dispatcher1->findAlgorithm(&softBody,&triBody,0, algoType);//m_manifoldPtr);
+		btCollisionAlgorithm* colAlgo = ci.m_dispatcher1->findAlgorithm(&softBody, &triBody, 0, algoType);  //m_manifoldPtr);
 
-		colAlgo->processCollision(&softBody,&triBody,*m_dispatchInfoPtr,m_resultOut);
+		colAlgo->processCollision(&softBody, &triBody, *m_dispatchInfoPtr, m_resultOut);
 		colAlgo->~btCollisionAlgorithm();
 		ci.m_dispatcher1->freeCollisionAlgorithm(colAlgo);
-		
+
 		return;
 	}
 
-	//aabb filter is already applied!	
+	//aabb filter is already applied!
 
 	//btCollisionObject* colObj = static_cast<btCollisionObject*>(m_convexProxy->m_clientObject);
 
 	//	if (m_softBody->getCollisionShape()->getShapeType()==
 	{
 		//		btVector3 other;
-		btVector3 normal = (triangle[1]-triangle[0]).cross(triangle[2]-triangle[0]);
+		btVector3 normal = (triangle[1] - triangle[0]).cross(triangle[2] - triangle[0]);
 		normal.normalize();
-		normal*= BT_SOFTBODY_TRIANGLE_EXTRUSION;
+		normal *= BT_SOFTBODY_TRIANGLE_EXTRUSION;
 		//		other=(triangle[0]+triangle[1]+triangle[2])*0.333333f;
 		//		other+=normal*22.f;
-		btVector3	pts[6] = {triangle[0]+normal,
-			triangle[1]+normal,
-			triangle[2]+normal,
-			triangle[0]-normal,
-			triangle[1]-normal,
-			triangle[2]-normal};
-
-		btConvexHullShape* tm = new btConvexHullShape(&pts[0].getX(),6);
+		btVector3 pts[6] = {triangle[0] + normal,
+							triangle[1] + normal,
+							triangle[2] + normal,
+							triangle[0] - normal,
+							triangle[1] - normal,
+							triangle[2] - normal};
 
+		btConvexHullShape* tm = new btConvexHullShape(&pts[0].getX(), 6);
 
 		//		btBU_Simplex1to4 tm(triangle[0],triangle[1],triangle[2],other);
 
-		//btTriangleShape tm(triangle[0],triangle[1],triangle[2]);	
+		//btTriangleShape tm(triangle[0],triangle[1],triangle[2]);
 		//	tm.setMargin(m_collisionMarginTriangle);
 
 		//copy over user pointers to temporary shape
 		tm->setUserPointer(m_triBody->getCollisionShape()->getUserPointer());
 
-		
-		btCollisionObjectWrapper softBody(0,m_softBody->getCollisionShape(),m_softBody,m_softBody->getWorldTransform(),-1,-1);
-		btCollisionObjectWrapper triBody(0,tm, m_triBody, m_triBody->getWorldTransform(),partId, triangleIndex);//btTransform::getIdentity());//??
+		btCollisionObjectWrapper softBody(0, m_softBody->getCollisionShape(), m_softBody, m_softBody->getWorldTransform(), -1, -1);
+		btCollisionObjectWrapper triBody(0, tm, m_triBody, m_triBody->getWorldTransform(), partId, triangleIndex);  //btTransform::getIdentity());//??
 
 		ebtDispatcherQueryType algoType = m_resultOut->m_closestPointDistanceThreshold > 0 ? BT_CLOSEST_POINT_ALGORITHMS : BT_CONTACT_POINT_ALGORITHMS;
-		btCollisionAlgorithm* colAlgo = ci.m_dispatcher1->findAlgorithm(&softBody,&triBody,0, algoType);//m_manifoldPtr);
+		btCollisionAlgorithm* colAlgo = ci.m_dispatcher1->findAlgorithm(&softBody, &triBody, 0, algoType);  //m_manifoldPtr);
 
-		colAlgo->processCollision(&softBody,&triBody,*m_dispatchInfoPtr,m_resultOut);
+		colAlgo->processCollision(&softBody, &triBody, *m_dispatchInfoPtr, m_resultOut);
 		colAlgo->~btCollisionAlgorithm();
 		ci.m_dispatcher1->freeCollisionAlgorithm(colAlgo);
 
 		triIndex.m_childShape = tm;
-		m_shapeCache.insert(triKey,triIndex);
-
+		m_shapeCache.insert(triKey, triIndex);
 	}
-
-
-
 }
 
-
-
-void	btSoftBodyTriangleCallback::setTimeStepAndCounters(btScalar collisionMarginTriangle,const btCollisionObjectWrapper* triBodyWrap, const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+void btSoftBodyTriangleCallback::setTimeStepAndCounters(btScalar collisionMarginTriangle, const btCollisionObjectWrapper* triBodyWrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	m_dispatchInfoPtr = &dispatchInfo;
-	m_collisionMarginTriangle = collisionMarginTriangle+btScalar(BT_SOFTBODY_TRIANGLE_EXTRUSION);
+	m_collisionMarginTriangle = collisionMarginTriangle + btScalar(BT_SOFTBODY_TRIANGLE_EXTRUSION);
 	m_resultOut = resultOut;
 
-
-	btVector3	aabbWorldSpaceMin,aabbWorldSpaceMax;
-	m_softBody->getAabb(aabbWorldSpaceMin,aabbWorldSpaceMax);
-	btVector3 halfExtents = (aabbWorldSpaceMax-aabbWorldSpaceMin)*btScalar(0.5);
-	btVector3 softBodyCenter = (aabbWorldSpaceMax+aabbWorldSpaceMin)*btScalar(0.5);
+	btVector3 aabbWorldSpaceMin, aabbWorldSpaceMax;
+	m_softBody->getAabb(aabbWorldSpaceMin, aabbWorldSpaceMax);
+	btVector3 halfExtents = (aabbWorldSpaceMax - aabbWorldSpaceMin) * btScalar(0.5);
+	btVector3 softBodyCenter = (aabbWorldSpaceMax + aabbWorldSpaceMin) * btScalar(0.5);
 
 	btTransform softTransform;
 	softTransform.setIdentity();
@@ -200,56 +180,45 @@ void	btSoftBodyTriangleCallback::setTimeStepAndCounters(btScalar collisionMargin
 
 	btTransform convexInTriangleSpace;
 	convexInTriangleSpace = triBodyWrap->getWorldTransform().inverse() * softTransform;
-	btTransformAabb(halfExtents,m_collisionMarginTriangle,convexInTriangleSpace,m_aabbMin,m_aabbMax);
+	btTransformAabb(halfExtents, m_collisionMarginTriangle, convexInTriangleSpace, m_aabbMin, m_aabbMax);
 }
 
 void btSoftBodyConcaveCollisionAlgorithm::clearCache()
 {
 	m_btSoftBodyTriangleCallback.clearCache();
-
 }
 
-void btSoftBodyConcaveCollisionAlgorithm::processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+void btSoftBodyConcaveCollisionAlgorithm::processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
-
-
 	//btCollisionObject* convexBody = m_isSwapped ? body1 : body0;
 	const btCollisionObjectWrapper* triBody = m_isSwapped ? body0Wrap : body1Wrap;
 
 	if (triBody->getCollisionShape()->isConcave())
 	{
-
-
-		const btCollisionObject*	triOb = triBody->getCollisionObject();
-		const btConcaveShape* concaveShape = static_cast<const btConcaveShape*>( triOb->getCollisionShape());
+		const btCollisionObject* triOb = triBody->getCollisionObject();
+		const btConcaveShape* concaveShape = static_cast<const btConcaveShape*>(triOb->getCollisionShape());
 
 		//	if (convexBody->getCollisionShape()->isConvex())
 		{
 			btScalar collisionMarginTriangle = concaveShape->getMargin();
 
 			//			resultOut->setPersistentManifold(m_btSoftBodyTriangleCallback.m_manifoldPtr);
-			m_btSoftBodyTriangleCallback.setTimeStepAndCounters(collisionMarginTriangle,triBody,dispatchInfo,resultOut);
+			m_btSoftBodyTriangleCallback.setTimeStepAndCounters(collisionMarginTriangle, triBody, dispatchInfo, resultOut);
 
-		
-			concaveShape->processAllTriangles( &m_btSoftBodyTriangleCallback,m_btSoftBodyTriangleCallback.getAabbMin(),m_btSoftBodyTriangleCallback.getAabbMax());
+			concaveShape->processAllTriangles(&m_btSoftBodyTriangleCallback, m_btSoftBodyTriangleCallback.getAabbMin(), m_btSoftBodyTriangleCallback.getAabbMax());
 
 			//	resultOut->refreshContactPoints();
-
 		}
-
 	}
-
 }
 
-
-btScalar btSoftBodyConcaveCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+btScalar btSoftBodyConcaveCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	(void)resultOut;
 	(void)dispatchInfo;
 	btCollisionObject* convexbody = m_isSwapped ? body1 : body0;
 	btCollisionObject* triBody = m_isSwapped ? body0 : body1;
 
-
 	//quick approximation using raycast, todo: hook up to the continuous collision detection (one of the btConvexCast)
 
 	//only perform CCD above a certain threshold, this prevents blocking on the long run
@@ -268,25 +237,23 @@ btScalar btSoftBodyConcaveCollisionAlgorithm::calculateTimeOfImpact(btCollisionO
 	btTransform convexFromLocal = triInv * convexbody->getWorldTransform();
 	btTransform convexToLocal = triInv * convexbody->getInterpolationWorldTransform();
 
-	struct LocalTriangleSphereCastCallback	: public btTriangleCallback
+	struct LocalTriangleSphereCastCallback : public btTriangleCallback
 	{
 		btTransform m_ccdSphereFromTrans;
 		btTransform m_ccdSphereToTrans;
-		btTransform	m_meshTransform;
+		btTransform m_meshTransform;
 
-		btScalar	m_ccdSphereRadius;
-		btScalar	m_hitFraction;
+		btScalar m_ccdSphereRadius;
+		btScalar m_hitFraction;
 
-
-		LocalTriangleSphereCastCallback(const btTransform& from,const btTransform& to,btScalar ccdSphereRadius,btScalar hitFraction)
-			:m_ccdSphereFromTrans(from),
-			m_ccdSphereToTrans(to),
-			m_ccdSphereRadius(ccdSphereRadius),
-			m_hitFraction(hitFraction)
-		{			
+		LocalTriangleSphereCastCallback(const btTransform& from, const btTransform& to, btScalar ccdSphereRadius, btScalar hitFraction)
+			: m_ccdSphereFromTrans(from),
+			  m_ccdSphereToTrans(to),
+			  m_ccdSphereRadius(ccdSphereRadius),
+			  m_hitFraction(hitFraction)
+		{
 		}
 
-
 		virtual void processTriangle(btVector3* triangle, int partId, int triangleIndex)
 		{
 			(void)partId;
@@ -296,29 +263,23 @@ btScalar btSoftBodyConcaveCollisionAlgorithm::calculateTimeOfImpact(btCollisionO
 			ident.setIdentity();
 			btConvexCast::CastResult castResult;
 			castResult.m_fraction = m_hitFraction;
-			btSphereShape	pointShape(m_ccdSphereRadius);
-			btTriangleShape	triShape(triangle[0],triangle[1],triangle[2]);
-			btVoronoiSimplexSolver	simplexSolver;
-			btSubsimplexConvexCast convexCaster(&pointShape,&triShape,&simplexSolver);
+			btSphereShape pointShape(m_ccdSphereRadius);
+			btTriangleShape triShape(triangle[0], triangle[1], triangle[2]);
+			btVoronoiSimplexSolver simplexSolver;
+			btSubsimplexConvexCast convexCaster(&pointShape, &triShape, &simplexSolver);
 			//GjkConvexCast	convexCaster(&pointShape,convexShape,&simplexSolver);
 			//ContinuousConvexCollision convexCaster(&pointShape,convexShape,&simplexSolver,0);
 			//local space?
 
-			if (convexCaster.calcTimeOfImpact(m_ccdSphereFromTrans,m_ccdSphereToTrans,
-				ident,ident,castResult))
+			if (convexCaster.calcTimeOfImpact(m_ccdSphereFromTrans, m_ccdSphereToTrans,
+											  ident, ident, castResult))
 			{
 				if (m_hitFraction > castResult.m_fraction)
 					m_hitFraction = castResult.m_fraction;
 			}
-
 		}
-
 	};
 
-
-
-
-
 	if (triBody->getCollisionShape()->isConcave())
 	{
 		btVector3 rayAabbMin = convexFromLocal.getOrigin();
@@ -326,33 +287,30 @@ btScalar btSoftBodyConcaveCollisionAlgorithm::calculateTimeOfImpact(btCollisionO
 		btVector3 rayAabbMax = convexFromLocal.getOrigin();
 		rayAabbMax.setMax(convexToLocal.getOrigin());
 		btScalar ccdRadius0 = convexbody->getCcdSweptSphereRadius();
-		rayAabbMin -= btVector3(ccdRadius0,ccdRadius0,ccdRadius0);
-		rayAabbMax += btVector3(ccdRadius0,ccdRadius0,ccdRadius0);
+		rayAabbMin -= btVector3(ccdRadius0, ccdRadius0, ccdRadius0);
+		rayAabbMax += btVector3(ccdRadius0, ccdRadius0, ccdRadius0);
 
-		btScalar curHitFraction = btScalar(1.); //is this available?
-		LocalTriangleSphereCastCallback raycastCallback(convexFromLocal,convexToLocal,
-			convexbody->getCcdSweptSphereRadius(),curHitFraction);
+		btScalar curHitFraction = btScalar(1.);  //is this available?
+		LocalTriangleSphereCastCallback raycastCallback(convexFromLocal, convexToLocal,
+														convexbody->getCcdSweptSphereRadius(), curHitFraction);
 
 		raycastCallback.m_hitFraction = convexbody->getHitFraction();
 
 		btCollisionObject* concavebody = triBody;
 
-		btConcaveShape* triangleMesh = (btConcaveShape*) concavebody->getCollisionShape();
+		btConcaveShape* triangleMesh = (btConcaveShape*)concavebody->getCollisionShape();
 
 		if (triangleMesh)
 		{
-			triangleMesh->processAllTriangles(&raycastCallback,rayAabbMin,rayAabbMax);
+			triangleMesh->processAllTriangles(&raycastCallback, rayAabbMin, rayAabbMax);
 		}
 
-
-
 		if (raycastCallback.m_hitFraction < convexbody->getHitFraction())
 		{
-			convexbody->setHitFraction( raycastCallback.m_hitFraction);
+			convexbody->setHitFraction(raycastCallback.m_hitFraction);
 			return raycastCallback.m_hitFraction;
 		}
 	}
 
 	return btScalar(1.);
-
 }
diff --git a/thirdparty/bullet/BulletSoftBody/btSoftBodyConcaveCollisionAlgorithm.h b/thirdparty/bullet/BulletSoftBody/btSoftBodyConcaveCollisionAlgorithm.h
index 11c7b88f98e..3adedbd8051 100644
--- a/thirdparty/bullet/BulletSoftBody/btSoftBodyConcaveCollisionAlgorithm.h
+++ b/thirdparty/bullet/BulletSoftBody/btSoftBodyConcaveCollisionAlgorithm.h
@@ -29,63 +29,62 @@ class btCollisionShape;
 
 #include "LinearMath/btHashMap.h"
 
-#include "BulletCollision/BroadphaseCollision/btQuantizedBvh.h" //for definition of MAX_NUM_PARTS_IN_BITS
+#include "BulletCollision/BroadphaseCollision/btQuantizedBvh.h"  //for definition of MAX_NUM_PARTS_IN_BITS
 
 struct btTriIndex
 {
 	int m_PartIdTriangleIndex;
-	class btCollisionShape*	m_childShape;
+	class btCollisionShape* m_childShape;
 
-	btTriIndex(int partId,int triangleIndex,btCollisionShape* shape)
+	btTriIndex(int partId, int triangleIndex, btCollisionShape* shape)
 	{
-		m_PartIdTriangleIndex = (partId<<(31-MAX_NUM_PARTS_IN_BITS)) | triangleIndex;
+		m_PartIdTriangleIndex = (partId << (31 - MAX_NUM_PARTS_IN_BITS)) | triangleIndex;
 		m_childShape = shape;
 	}
 
-	int	getTriangleIndex() const
+	int getTriangleIndex() const
 	{
 		// Get only the lower bits where the triangle index is stored
 		unsigned int x = 0;
-		unsigned int y = (~(x&0))<<(31-MAX_NUM_PARTS_IN_BITS);
-		return (m_PartIdTriangleIndex&~(y));
+		unsigned int y = (~(x & 0)) << (31 - MAX_NUM_PARTS_IN_BITS);
+		return (m_PartIdTriangleIndex & ~(y));
 	}
-	int	getPartId() const
+	int getPartId() const
 	{
 		// Get only the highest bits where the part index is stored
-		return (m_PartIdTriangleIndex>>(31-MAX_NUM_PARTS_IN_BITS));
+		return (m_PartIdTriangleIndex >> (31 - MAX_NUM_PARTS_IN_BITS));
 	}
-	int	getUid() const
+	int getUid() const
 	{
 		return m_PartIdTriangleIndex;
 	}
 };
 
-
 ///For each triangle in the concave mesh that overlaps with the AABB of a soft body (m_softBody), processTriangle is called.
 class btSoftBodyTriangleCallback : public btTriangleCallback
 {
 	btSoftBody* m_softBody;
 	const btCollisionObject* m_triBody;
 
-	btVector3	m_aabbMin;
-	btVector3	m_aabbMax ;
+	btVector3 m_aabbMin;
+	btVector3 m_aabbMax;
 
 	btManifoldResult* m_resultOut;
 
-	btDispatcher*	m_dispatcher;
+	btDispatcher* m_dispatcher;
 	const btDispatcherInfo* m_dispatchInfoPtr;
 	btScalar m_collisionMarginTriangle;
 
-	btHashMap<btHashKey<btTriIndex>,btTriIndex> m_shapeCache;
+	btHashMap<btHashKey<btTriIndex>, btTriIndex> m_shapeCache;
 
 public:
-	int	m_triangleCount;
+	int m_triangleCount;
 
 	//	btPersistentManifold*	m_manifoldPtr;
 
-	btSoftBodyTriangleCallback(btDispatcher* dispatcher,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,bool isSwapped);
+	btSoftBodyTriangleCallback(btDispatcher* dispatcher, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, bool isSwapped);
 
-	void	setTimeStepAndCounters(btScalar collisionMarginTriangle,const btCollisionObjectWrapper* triObjWrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	void setTimeStepAndCounters(btScalar collisionMarginTriangle, const btCollisionObjectWrapper* triObjWrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
 	virtual ~btSoftBodyTriangleCallback();
 
@@ -101,55 +100,48 @@ public:
 	{
 		return m_aabbMax;
 	}
-
 };
 
-
-
-
 /// btSoftBodyConcaveCollisionAlgorithm  supports collision between soft body shapes and (concave) trianges meshes.
-class btSoftBodyConcaveCollisionAlgorithm  : public btCollisionAlgorithm
+class btSoftBodyConcaveCollisionAlgorithm : public btCollisionAlgorithm
 {
-
-	bool	m_isSwapped;
+	bool m_isSwapped;
 
 	btSoftBodyTriangleCallback m_btSoftBodyTriangleCallback;
 
 public:
-
-	btSoftBodyConcaveCollisionAlgorithm( const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,bool isSwapped);
+	btSoftBodyConcaveCollisionAlgorithm(const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, bool isSwapped);
 
 	virtual ~btSoftBodyConcaveCollisionAlgorithm();
 
-	virtual void processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual void processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	btScalar	calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	btScalar calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual	void	getAllContactManifolds(btManifoldArray&	manifoldArray)
+	virtual void getAllContactManifolds(btManifoldArray& manifoldArray)
 	{
 		//we don't add any manifolds
 	}
 
-	void	clearCache();
+	void clearCache();
 
-	struct CreateFunc :public 	btCollisionAlgorithmCreateFunc
+	struct CreateFunc : public btCollisionAlgorithmCreateFunc
 	{
-		virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+		virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 		{
 			void* mem = ci.m_dispatcher1->allocateCollisionAlgorithm(sizeof(btSoftBodyConcaveCollisionAlgorithm));
-			return new(mem) btSoftBodyConcaveCollisionAlgorithm(ci,body0Wrap,body1Wrap,false);
+			return new (mem) btSoftBodyConcaveCollisionAlgorithm(ci, body0Wrap, body1Wrap, false);
 		}
 	};
 
-	struct SwappedCreateFunc :public 	btCollisionAlgorithmCreateFunc
+	struct SwappedCreateFunc : public btCollisionAlgorithmCreateFunc
 	{
-		virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+		virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 		{
 			void* mem = ci.m_dispatcher1->allocateCollisionAlgorithm(sizeof(btSoftBodyConcaveCollisionAlgorithm));
-			return new(mem) btSoftBodyConcaveCollisionAlgorithm(ci,body0Wrap,body1Wrap,true);
+			return new (mem) btSoftBodyConcaveCollisionAlgorithm(ci, body0Wrap, body1Wrap, true);
 		}
 	};
-
 };
 
-#endif //BT_SOFT_BODY_CONCAVE_COLLISION_ALGORITHM_H
+#endif  //BT_SOFT_BODY_CONCAVE_COLLISION_ALGORITHM_H
diff --git a/thirdparty/bullet/BulletSoftBody/btSoftBodyData.h b/thirdparty/bullet/BulletSoftBody/btSoftBodyData.h
index 87d8841cfa4..cec6f401ecc 100644
--- a/thirdparty/bullet/BulletSoftBody/btSoftBodyData.h
+++ b/thirdparty/bullet/BulletSoftBody/btSoftBodyData.h
@@ -19,199 +19,194 @@ subject to the following restrictions:
 #include "BulletCollision/CollisionDispatch/btCollisionObject.h"
 #include "BulletDynamics/Dynamics/btRigidBody.h"
 
-
-struct	SoftBodyMaterialData
+struct SoftBodyMaterialData
 {
-	float	m_linearStiffness;
-	float	m_angularStiffness;
-	float	m_volumeStiffness;
-	int		m_flags;
+	float m_linearStiffness;
+	float m_angularStiffness;
+	float m_volumeStiffness;
+	int m_flags;
 };
 
-struct	SoftBodyNodeData
+struct SoftBodyNodeData
 {
-	SoftBodyMaterialData		*m_material;
-	btVector3FloatData			m_position;
-	btVector3FloatData			m_previousPosition;
-	btVector3FloatData			m_velocity;
-	btVector3FloatData			m_accumulatedForce;
-	btVector3FloatData			m_normal;
-	float						m_inverseMass;
-	float						m_area;
-	int							m_attach;
-	int							m_pad;
+	SoftBodyMaterialData *m_material;
+	btVector3FloatData m_position;
+	btVector3FloatData m_previousPosition;
+	btVector3FloatData m_velocity;
+	btVector3FloatData m_accumulatedForce;
+	btVector3FloatData m_normal;
+	float m_inverseMass;
+	float m_area;
+	int m_attach;
+	int m_pad;
 };
 
-struct	SoftBodyLinkData
+struct SoftBodyLinkData
 {
-	SoftBodyMaterialData	*m_material;
-	int						m_nodeIndices[2];			// Node pointers
-	float					m_restLength;			// Rest length		
-	int						m_bbending;		// Bending link
+	SoftBodyMaterialData *m_material;
+	int m_nodeIndices[2];  // Node pointers
+	float m_restLength;    // Rest length
+	int m_bbending;        // Bending link
 };
 
-struct	SoftBodyFaceData
+struct SoftBodyFaceData
 {
-	btVector3FloatData		m_normal;		// Normal
-	SoftBodyMaterialData	*m_material;
-	int						m_nodeIndices[3];			// Node pointers
-	float					m_restArea;			// Rest area
-};	
-
-struct	SoftBodyTetraData
-{
-	btVector3FloatData		m_c0[4];		// gradients
-	SoftBodyMaterialData	*m_material;
-	int						m_nodeIndices[4];			// Node pointers		
-	float					m_restVolume;			// Rest volume
-	float					m_c1;			// (4*kVST)/(im0+im1+im2+im3)
-	float					m_c2;			// m_c1/sum(|g0..3|^2)
-	int						m_pad;
+	btVector3FloatData m_normal;  // Normal
+	SoftBodyMaterialData *m_material;
+	int m_nodeIndices[3];  // Node pointers
+	float m_restArea;      // Rest area
 };
 
-struct	SoftRigidAnchorData
+struct SoftBodyTetraData
 {
-	btMatrix3x3FloatData	m_c0;			// Impulse matrix
-	btVector3FloatData		m_c1;			// Relative anchor
-	btVector3FloatData		m_localFrame;		// Anchor position in body space
-	btRigidBodyData			*m_rigidBody;
-	int						m_nodeIndex;			// Node pointer
-	float					m_c2;			// ima*dt
+	btVector3FloatData m_c0[4];  // gradients
+	SoftBodyMaterialData *m_material;
+	int m_nodeIndices[4];  // Node pointers
+	float m_restVolume;    // Rest volume
+	float m_c1;            // (4*kVST)/(im0+im1+im2+im3)
+	float m_c2;            // m_c1/sum(|g0..3|^2)
+	int m_pad;
 };
 
-
-
-struct	SoftBodyConfigData
+struct SoftRigidAnchorData
 {
-	int					m_aeroModel;		// Aerodynamic model (default: V_Point)
-	float				m_baumgarte;			// Velocities correction factor (Baumgarte)
-	float				m_damping;			// Damping coefficient [0,1]
-	float				m_drag;			// Drag coefficient [0,+inf]
-	float				m_lift;			// Lift coefficient [0,+inf]
-	float				m_pressure;			// Pressure coefficient [-inf,+inf]
-	float				m_volume;			// Volume conversation coefficient [0,+inf]
-	float				m_dynamicFriction;			// Dynamic friction coefficient [0,1]
-	float				m_poseMatch;			// Pose matching coefficient [0,1]		
-	float				m_rigidContactHardness;			// Rigid contacts hardness [0,1]
-	float				m_kineticContactHardness;			// Kinetic contacts hardness [0,1]
-	float				m_softContactHardness;			// Soft contacts hardness [0,1]
-	float				m_anchorHardness;			// Anchors hardness [0,1]
-	float				m_softRigidClusterHardness;		// Soft vs rigid hardness [0,1] (cluster only)
-	float				m_softKineticClusterHardness;		// Soft vs kinetic hardness [0,1] (cluster only)
-	float				m_softSoftClusterHardness;		// Soft vs soft hardness [0,1] (cluster only)
-	float				m_softRigidClusterImpulseSplit;	// Soft vs rigid impulse split [0,1] (cluster only)
-	float				m_softKineticClusterImpulseSplit;	// Soft vs rigid impulse split [0,1] (cluster only)
-	float				m_softSoftClusterImpulseSplit;	// Soft vs rigid impulse split [0,1] (cluster only)
-	float				m_maxVolume;		// Maximum volume ratio for pose
-	float				m_timeScale;		// Time scale
-	int					m_velocityIterations;	// Velocities solver iterations
-	int					m_positionIterations;	// Positions solver iterations
-	int					m_driftIterations;	// Drift solver iterations
-	int					m_clusterIterations;	// Cluster solver iterations
-	int					m_collisionFlags;	// Collisions flags
+	btMatrix3x3FloatData m_c0;        // Impulse matrix
+	btVector3FloatData m_c1;          // Relative anchor
+	btVector3FloatData m_localFrame;  // Anchor position in body space
+	btRigidBodyData *m_rigidBody;
+	int m_nodeIndex;  // Node pointer
+	float m_c2;       // ima*dt
 };
 
-struct	SoftBodyPoseData
+struct SoftBodyConfigData
 {
-	btMatrix3x3FloatData	m_rot;			// Rotation
-	btMatrix3x3FloatData	m_scale;			// Scale
-	btMatrix3x3FloatData	m_aqq;			// Base scaling
-	btVector3FloatData		m_com;			// COM
-
-	btVector3FloatData		*m_positions;			// Reference positions
-	float					*m_weights;	// Weights
-	int						m_numPositions;
-	int						m_numWeigts;
-
-	int						m_bvolume;		// Is valid
-	int						m_bframe;		// Is frame
-	float					m_restVolume;		// Rest volume
-	int						m_pad;
+	int m_aeroModel;                         // Aerodynamic model (default: V_Point)
+	float m_baumgarte;                       // Velocities correction factor (Baumgarte)
+	float m_damping;                         // Damping coefficient [0,1]
+	float m_drag;                            // Drag coefficient [0,+inf]
+	float m_lift;                            // Lift coefficient [0,+inf]
+	float m_pressure;                        // Pressure coefficient [-inf,+inf]
+	float m_volume;                          // Volume conversation coefficient [0,+inf]
+	float m_dynamicFriction;                 // Dynamic friction coefficient [0,1]
+	float m_poseMatch;                       // Pose matching coefficient [0,1]
+	float m_rigidContactHardness;            // Rigid contacts hardness [0,1]
+	float m_kineticContactHardness;          // Kinetic contacts hardness [0,1]
+	float m_softContactHardness;             // Soft contacts hardness [0,1]
+	float m_anchorHardness;                  // Anchors hardness [0,1]
+	float m_softRigidClusterHardness;        // Soft vs rigid hardness [0,1] (cluster only)
+	float m_softKineticClusterHardness;      // Soft vs kinetic hardness [0,1] (cluster only)
+	float m_softSoftClusterHardness;         // Soft vs soft hardness [0,1] (cluster only)
+	float m_softRigidClusterImpulseSplit;    // Soft vs rigid impulse split [0,1] (cluster only)
+	float m_softKineticClusterImpulseSplit;  // Soft vs rigid impulse split [0,1] (cluster only)
+	float m_softSoftClusterImpulseSplit;     // Soft vs rigid impulse split [0,1] (cluster only)
+	float m_maxVolume;                       // Maximum volume ratio for pose
+	float m_timeScale;                       // Time scale
+	int m_velocityIterations;                // Velocities solver iterations
+	int m_positionIterations;                // Positions solver iterations
+	int m_driftIterations;                   // Drift solver iterations
+	int m_clusterIterations;                 // Cluster solver iterations
+	int m_collisionFlags;                    // Collisions flags
 };
 
-struct	SoftBodyClusterData
+struct SoftBodyPoseData
 {
-		btTransformFloatData		m_framexform;
-		btMatrix3x3FloatData		m_locii;
-		btMatrix3x3FloatData		m_invwi;
-		btVector3FloatData			m_com;
-		btVector3FloatData			m_vimpulses[2];
-		btVector3FloatData			m_dimpulses[2];
-		btVector3FloatData			m_lv;
-		btVector3FloatData			m_av;
-		
-		btVector3FloatData			*m_framerefs;
-		int							*m_nodeIndices;
-		float						*m_masses;
+	btMatrix3x3FloatData m_rot;    // Rotation
+	btMatrix3x3FloatData m_scale;  // Scale
+	btMatrix3x3FloatData m_aqq;    // Base scaling
+	btVector3FloatData m_com;      // COM
 
-		int							m_numFrameRefs;
-		int							m_numNodes;
-		int							m_numMasses;
+	btVector3FloatData *m_positions;  // Reference positions
+	float *m_weights;                 // Weights
+	int m_numPositions;
+	int m_numWeigts;
 
-		float						m_idmass;
-		float						m_imass;
-		int							m_nvimpulses;
-		int							m_ndimpulses;
-		float						m_ndamping;
-		float						m_ldamping;
-		float						m_adamping;
-		float						m_matching;
-		float						m_maxSelfCollisionImpulse;
-		float						m_selfCollisionImpulseFactor;
-		int							m_containsAnchor;
-		int							m_collide;
-		int							m_clusterIndex;
+	int m_bvolume;       // Is valid
+	int m_bframe;        // Is frame
+	float m_restVolume;  // Rest volume
+	int m_pad;
 };
 
-
-enum	btSoftJointBodyType
+struct SoftBodyClusterData
 {
-	BT_JOINT_SOFT_BODY_CLUSTER=1,
+	btTransformFloatData m_framexform;
+	btMatrix3x3FloatData m_locii;
+	btMatrix3x3FloatData m_invwi;
+	btVector3FloatData m_com;
+	btVector3FloatData m_vimpulses[2];
+	btVector3FloatData m_dimpulses[2];
+	btVector3FloatData m_lv;
+	btVector3FloatData m_av;
+
+	btVector3FloatData *m_framerefs;
+	int *m_nodeIndices;
+	float *m_masses;
+
+	int m_numFrameRefs;
+	int m_numNodes;
+	int m_numMasses;
+
+	float m_idmass;
+	float m_imass;
+	int m_nvimpulses;
+	int m_ndimpulses;
+	float m_ndamping;
+	float m_ldamping;
+	float m_adamping;
+	float m_matching;
+	float m_maxSelfCollisionImpulse;
+	float m_selfCollisionImpulseFactor;
+	int m_containsAnchor;
+	int m_collide;
+	int m_clusterIndex;
+};
+
+enum btSoftJointBodyType
+{
+	BT_JOINT_SOFT_BODY_CLUSTER = 1,
 	BT_JOINT_RIGID_BODY,
 	BT_JOINT_COLLISION_OBJECT
 };
 
-struct	btSoftBodyJointData
+struct btSoftBodyJointData
 {
-	void						*m_bodyA;
-	void						*m_bodyB;
-	btVector3FloatData			m_refs[2];
-	float						m_cfm;
-	float						m_erp;
-	float						m_split;
-	int							m_delete;
-	btVector3FloatData			m_relPosition[2];//linear
-	int							m_bodyAtype;
-	int							m_bodyBtype;
-	int							m_jointType;
-	int							m_pad;
+	void *m_bodyA;
+	void *m_bodyB;
+	btVector3FloatData m_refs[2];
+	float m_cfm;
+	float m_erp;
+	float m_split;
+	int m_delete;
+	btVector3FloatData m_relPosition[2];  //linear
+	int m_bodyAtype;
+	int m_bodyBtype;
+	int m_jointType;
+	int m_pad;
 };
 
 ///do not change those serialization structures, it requires an updated sBulletDNAstr/sBulletDNAstr64
-struct	btSoftBodyFloatData
+struct btSoftBodyFloatData
 {
-	btCollisionObjectFloatData	m_collisionObjectData;
+	btCollisionObjectFloatData m_collisionObjectData;
 
-	SoftBodyPoseData		*m_pose;
-	SoftBodyMaterialData	**m_materials;
-	SoftBodyNodeData		*m_nodes;
-	SoftBodyLinkData		*m_links;
-	SoftBodyFaceData		*m_faces;
-	SoftBodyTetraData		*m_tetrahedra;
-	SoftRigidAnchorData		*m_anchors;
-	SoftBodyClusterData		*m_clusters;
-	btSoftBodyJointData		*m_joints;
+	SoftBodyPoseData *m_pose;
+	SoftBodyMaterialData **m_materials;
+	SoftBodyNodeData *m_nodes;
+	SoftBodyLinkData *m_links;
+	SoftBodyFaceData *m_faces;
+	SoftBodyTetraData *m_tetrahedra;
+	SoftRigidAnchorData *m_anchors;
+	SoftBodyClusterData *m_clusters;
+	btSoftBodyJointData *m_joints;
 
-	int						m_numMaterials;
-	int						m_numNodes;
-	int						m_numLinks;
-	int						m_numFaces;
-	int						m_numTetrahedra;
-	int						m_numAnchors;
-	int						m_numClusters;
-	int						m_numJoints;
-	SoftBodyConfigData		m_config;
+	int m_numMaterials;
+	int m_numNodes;
+	int m_numLinks;
+	int m_numFaces;
+	int m_numTetrahedra;
+	int m_numAnchors;
+	int m_numClusters;
+	int m_numJoints;
+	SoftBodyConfigData m_config;
 };
 
-#endif //BT_SOFTBODY_FLOAT_DATA
-
+#endif  //BT_SOFTBODY_FLOAT_DATA
diff --git a/thirdparty/bullet/BulletSoftBody/btSoftBodyHelpers.cpp b/thirdparty/bullet/BulletSoftBody/btSoftBodyHelpers.cpp
index 51fcd16da40..d0a9921d899 100644
--- a/thirdparty/bullet/BulletSoftBody/btSoftBodyHelpers.cpp
+++ b/thirdparty/bullet/BulletSoftBody/btSoftBodyHelpers.cpp
@@ -21,106 +21,111 @@ subject to the following restrictions:
 #include "LinearMath/btConvexHull.h"
 #include "LinearMath/btConvexHullComputer.h"
 
-
 //
-static void				drawVertex(	btIDebugDraw* idraw,
-								   const btVector3& x,btScalar s,const btVector3& c)
+static void drawVertex(btIDebugDraw* idraw,
+					   const btVector3& x, btScalar s, const btVector3& c)
 {
-	idraw->drawLine(x-btVector3(s,0,0),x+btVector3(s,0,0),c);
-	idraw->drawLine(x-btVector3(0,s,0),x+btVector3(0,s,0),c);
-	idraw->drawLine(x-btVector3(0,0,s),x+btVector3(0,0,s),c);
+	idraw->drawLine(x - btVector3(s, 0, 0), x + btVector3(s, 0, 0), c);
+	idraw->drawLine(x - btVector3(0, s, 0), x + btVector3(0, s, 0), c);
+	idraw->drawLine(x - btVector3(0, 0, s), x + btVector3(0, 0, s), c);
 }
 
 //
-static void				drawBox(	btIDebugDraw* idraw,
-								const btVector3& mins,
-								const btVector3& maxs,
-								const btVector3& color)
+static void drawBox(btIDebugDraw* idraw,
+					const btVector3& mins,
+					const btVector3& maxs,
+					const btVector3& color)
 {
-	const btVector3	c[]={	btVector3(mins.x(),mins.y(),mins.z()),
-		btVector3(maxs.x(),mins.y(),mins.z()),
-		btVector3(maxs.x(),maxs.y(),mins.z()),
-		btVector3(mins.x(),maxs.y(),mins.z()),
-		btVector3(mins.x(),mins.y(),maxs.z()),
-		btVector3(maxs.x(),mins.y(),maxs.z()),
-		btVector3(maxs.x(),maxs.y(),maxs.z()),
-		btVector3(mins.x(),maxs.y(),maxs.z())};
-	idraw->drawLine(c[0],c[1],color);idraw->drawLine(c[1],c[2],color);
-	idraw->drawLine(c[2],c[3],color);idraw->drawLine(c[3],c[0],color);
-	idraw->drawLine(c[4],c[5],color);idraw->drawLine(c[5],c[6],color);
-	idraw->drawLine(c[6],c[7],color);idraw->drawLine(c[7],c[4],color);
-	idraw->drawLine(c[0],c[4],color);idraw->drawLine(c[1],c[5],color);
-	idraw->drawLine(c[2],c[6],color);idraw->drawLine(c[3],c[7],color);
+	const btVector3 c[] = {btVector3(mins.x(), mins.y(), mins.z()),
+						   btVector3(maxs.x(), mins.y(), mins.z()),
+						   btVector3(maxs.x(), maxs.y(), mins.z()),
+						   btVector3(mins.x(), maxs.y(), mins.z()),
+						   btVector3(mins.x(), mins.y(), maxs.z()),
+						   btVector3(maxs.x(), mins.y(), maxs.z()),
+						   btVector3(maxs.x(), maxs.y(), maxs.z()),
+						   btVector3(mins.x(), maxs.y(), maxs.z())};
+	idraw->drawLine(c[0], c[1], color);
+	idraw->drawLine(c[1], c[2], color);
+	idraw->drawLine(c[2], c[3], color);
+	idraw->drawLine(c[3], c[0], color);
+	idraw->drawLine(c[4], c[5], color);
+	idraw->drawLine(c[5], c[6], color);
+	idraw->drawLine(c[6], c[7], color);
+	idraw->drawLine(c[7], c[4], color);
+	idraw->drawLine(c[0], c[4], color);
+	idraw->drawLine(c[1], c[5], color);
+	idraw->drawLine(c[2], c[6], color);
+	idraw->drawLine(c[3], c[7], color);
 }
 
 //
-static void				drawTree(	btIDebugDraw* idraw,
-								 const btDbvtNode* node,
-								 int depth,
-								 const btVector3& ncolor,
-								 const btVector3& lcolor,
-								 int mindepth,
-								 int maxdepth)
+static void drawTree(btIDebugDraw* idraw,
+					 const btDbvtNode* node,
+					 int depth,
+					 const btVector3& ncolor,
+					 const btVector3& lcolor,
+					 int mindepth,
+					 int maxdepth)
 {
-	if(node)
+	if (node)
 	{
-		if(node->isinternal()&&((depth<maxdepth)||(maxdepth<0)))
+		if (node->isinternal() && ((depth < maxdepth) || (maxdepth < 0)))
 		{
-			drawTree(idraw,node->childs[0],depth+1,ncolor,lcolor,mindepth,maxdepth);
-			drawTree(idraw,node->childs[1],depth+1,ncolor,lcolor,mindepth,maxdepth);
+			drawTree(idraw, node->childs[0], depth + 1, ncolor, lcolor, mindepth, maxdepth);
+			drawTree(idraw, node->childs[1], depth + 1, ncolor, lcolor, mindepth, maxdepth);
 		}
-		if(depth>=mindepth)
+		if (depth >= mindepth)
 		{
-			const btScalar	scl=(btScalar)(node->isinternal()?1:1);
-			const btVector3	mi=node->volume.Center()-node->volume.Extents()*scl;
-			const btVector3	mx=node->volume.Center()+node->volume.Extents()*scl;
-			drawBox(idraw,mi,mx,node->isleaf()?lcolor:ncolor);
+			const btScalar scl = (btScalar)(node->isinternal() ? 1 : 1);
+			const btVector3 mi = node->volume.Center() - node->volume.Extents() * scl;
+			const btVector3 mx = node->volume.Center() + node->volume.Extents() * scl;
+			drawBox(idraw, mi, mx, node->isleaf() ? lcolor : ncolor);
 		}
 	}
 }
 
 //
 template <typename T>
-static inline T				sum(const btAlignedObjectArray<T>& items)
+static inline T sum(const btAlignedObjectArray<T>& items)
 {
-	T	v;
-	if(items.size())
+	T v;
+	if (items.size())
 	{
-		v=items[0];
-		for(int i=1,ni=items.size();i<ni;++i)
+		v = items[0];
+		for (int i = 1, ni = items.size(); i < ni; ++i)
 		{
-			v+=items[i];
+			v += items[i];
 		}
 	}
-	return(v);
+	return (v);
 }
 
 //
-template <typename T,typename Q>
-static inline void			add(btAlignedObjectArray<T>& items,const Q& value)
+template <typename T, typename Q>
+static inline void add(btAlignedObjectArray<T>& items, const Q& value)
 {
-	for(int i=0,ni=items.size();i<ni;++i)
+	for (int i = 0, ni = items.size(); i < ni; ++i)
 	{
-		items[i]+=value;
+		items[i] += value;
 	}
 }
 
 //
-template <typename T,typename Q>
-static inline void			mul(btAlignedObjectArray<T>& items,const Q& value)
+template <typename T, typename Q>
+static inline void mul(btAlignedObjectArray<T>& items, const Q& value)
 {
-	for(int i=0,ni=items.size();i<ni;++i)
+	for (int i = 0, ni = items.size(); i < ni; ++i)
 	{
-		items[i]*=value;
+		items[i] *= value;
 	}
 }
 
 //
 template <typename T>
-static inline T				average(const btAlignedObjectArray<T>& items)
+static inline T average(const btAlignedObjectArray<T>& items)
 {
-	const btScalar	n=(btScalar)(items.size()>0?items.size():1);
-	return(sum(items)/n);
+	const btScalar n = (btScalar)(items.size() > 0 ? items.size() : 1);
+	return (sum(items) / n);
 }
 
 #if 0
@@ -158,86 +163,84 @@ static btVector3		stresscolor(btScalar stress)
 #endif
 
 //
-void			btSoftBodyHelpers::Draw(	btSoftBody* psb,
-										btIDebugDraw* idraw,
-										int drawflags)
+void btSoftBodyHelpers::Draw(btSoftBody* psb,
+							 btIDebugDraw* idraw,
+							 int drawflags)
 {
-	const btScalar		scl=(btScalar)0.1;
-	const btScalar		nscl=scl*5;
-	const btVector3		lcolor=btVector3(0,0,0);
-	const btVector3		ncolor=btVector3(1,1,1);
-	const btVector3		ccolor=btVector3(1,0,0);
-	int i,j,nj;
+	const btScalar scl = (btScalar)0.1;
+	const btScalar nscl = scl * 5;
+	const btVector3 lcolor = btVector3(0, 0, 0);
+	const btVector3 ncolor = btVector3(1, 1, 1);
+	const btVector3 ccolor = btVector3(1, 0, 0);
+	int i, j, nj;
 
-		/* Clusters	*/ 
-	if(0!=(drawflags&fDrawFlags::Clusters))
+	/* Clusters	*/
+	if (0 != (drawflags & fDrawFlags::Clusters))
 	{
 		srand(1806);
-		for(i=0;i<psb->m_clusters.size();++i)
+		for (i = 0; i < psb->m_clusters.size(); ++i)
 		{
-			if(psb->m_clusters[i]->m_collide)
+			if (psb->m_clusters[i]->m_collide)
 			{
-				btVector3						color(	rand()/(btScalar)RAND_MAX,
-					rand()/(btScalar)RAND_MAX,
-					rand()/(btScalar)RAND_MAX);
-				color=color.normalized()*0.75;
-				btAlignedObjectArray<btVector3>	vertices;
+				btVector3 color(rand() / (btScalar)RAND_MAX,
+								rand() / (btScalar)RAND_MAX,
+								rand() / (btScalar)RAND_MAX);
+				color = color.normalized() * 0.75;
+				btAlignedObjectArray<btVector3> vertices;
 				vertices.resize(psb->m_clusters[i]->m_nodes.size());
-				for(j=0,nj=vertices.size();j<nj;++j)
-				{				
-					vertices[j]=psb->m_clusters[i]->m_nodes[j]->m_x;
+				for (j = 0, nj = vertices.size(); j < nj; ++j)
+				{
+					vertices[j] = psb->m_clusters[i]->m_nodes[j]->m_x;
 				}
 #define USE_NEW_CONVEX_HULL_COMPUTER
 #ifdef USE_NEW_CONVEX_HULL_COMPUTER
-				btConvexHullComputer	computer;
+				btConvexHullComputer computer;
 				int stride = sizeof(btVector3);
 				int count = vertices.size();
-				btScalar shrink=0.f;
-				btScalar shrinkClamp=0.f;
-				computer.compute(&vertices[0].getX(),stride,count,shrink,shrinkClamp);
-				for (int i=0;i<computer.faces.size();i++)
+				btScalar shrink = 0.f;
+				btScalar shrinkClamp = 0.f;
+				computer.compute(&vertices[0].getX(), stride, count, shrink, shrinkClamp);
+				for (int i = 0; i < computer.faces.size(); i++)
 				{
-
 					int face = computer.faces[i];
 					//printf("face=%d\n",face);
-					const btConvexHullComputer::Edge*  firstEdge = &computer.edges[face];
-					const btConvexHullComputer::Edge*  edge = firstEdge->getNextEdgeOfFace();
+					const btConvexHullComputer::Edge* firstEdge = &computer.edges[face];
+					const btConvexHullComputer::Edge* edge = firstEdge->getNextEdgeOfFace();
 
 					int v0 = firstEdge->getSourceVertex();
 					int v1 = firstEdge->getTargetVertex();
-					while (edge!=firstEdge)
+					while (edge != firstEdge)
 					{
 						int v2 = edge->getTargetVertex();
-						idraw->drawTriangle(computer.vertices[v0],computer.vertices[v1],computer.vertices[v2],color,1);
+						idraw->drawTriangle(computer.vertices[v0], computer.vertices[v1], computer.vertices[v2], color, 1);
 						edge = edge->getNextEdgeOfFace();
-						v0=v1;
-						v1=v2;
+						v0 = v1;
+						v1 = v2;
 					};
 				}
 #else
 
-				HullDesc		hdsc(QF_TRIANGLES,vertices.size(),&vertices[0]);
-				HullResult		hres;
-				HullLibrary		hlib;
-				hdsc.mMaxVertices=vertices.size();
-				hlib.CreateConvexHull(hdsc,hres);
-				const btVector3	center=average(hres.m_OutputVertices);
-				add(hres.m_OutputVertices,-center);
-				mul(hres.m_OutputVertices,(btScalar)1);
-				add(hres.m_OutputVertices,center);
-				for(j=0;j<(int)hres.mNumFaces;++j)
+				HullDesc hdsc(QF_TRIANGLES, vertices.size(), &vertices[0]);
+				HullResult hres;
+				HullLibrary hlib;
+				hdsc.mMaxVertices = vertices.size();
+				hlib.CreateConvexHull(hdsc, hres);
+				const btVector3 center = average(hres.m_OutputVertices);
+				add(hres.m_OutputVertices, -center);
+				mul(hres.m_OutputVertices, (btScalar)1);
+				add(hres.m_OutputVertices, center);
+				for (j = 0; j < (int)hres.mNumFaces; ++j)
 				{
-					const int idx[]={hres.m_Indices[j*3+0],hres.m_Indices[j*3+1],hres.m_Indices[j*3+2]};
+					const int idx[] = {hres.m_Indices[j * 3 + 0], hres.m_Indices[j * 3 + 1], hres.m_Indices[j * 3 + 2]};
 					idraw->drawTriangle(hres.m_OutputVertices[idx[0]],
-						hres.m_OutputVertices[idx[1]],
-						hres.m_OutputVertices[idx[2]],
-						color,1);
+										hres.m_OutputVertices[idx[1]],
+										hres.m_OutputVertices[idx[2]],
+										color, 1);
 				}
 				hlib.ReleaseResult(hres);
 #endif
-
 			}
-			/* Velocities	*/ 
+			/* Velocities	*/
 #if 0
 			for(int j=0;j<psb->m_clusters[i].m_nodes.size();++j)
 			{
@@ -247,273 +250,269 @@ void			btSoftBodyHelpers::Draw(	btSoftBody* psb,
 				idraw->drawLine(c.m_nodes[j]->m_x,c.m_nodes[j]->m_x+v,btVector3(1,0,0));
 			}
 #endif
-			/* Frame		*/ 
-	//		btSoftBody::Cluster& c=*psb->m_clusters[i];
-	//		idraw->drawLine(c.m_com,c.m_framexform*btVector3(10,0,0),btVector3(1,0,0));
-	//		idraw->drawLine(c.m_com,c.m_framexform*btVector3(0,10,0),btVector3(0,1,0));
-	//		idraw->drawLine(c.m_com,c.m_framexform*btVector3(0,0,10),btVector3(0,0,1));
+			/* Frame		*/
+			//		btSoftBody::Cluster& c=*psb->m_clusters[i];
+			//		idraw->drawLine(c.m_com,c.m_framexform*btVector3(10,0,0),btVector3(1,0,0));
+			//		idraw->drawLine(c.m_com,c.m_framexform*btVector3(0,10,0),btVector3(0,1,0));
+			//		idraw->drawLine(c.m_com,c.m_framexform*btVector3(0,0,10),btVector3(0,0,1));
 		}
 	}
 	else
 	{
-		/* Nodes	*/ 
-		if(0!=(drawflags&fDrawFlags::Nodes))
+		/* Nodes	*/
+		if (0 != (drawflags & fDrawFlags::Nodes))
 		{
-			for(i=0;i<psb->m_nodes.size();++i)
+			for (i = 0; i < psb->m_nodes.size(); ++i)
 			{
-				const btSoftBody::Node&	n=psb->m_nodes[i];
-				if(0==(n.m_material->m_flags&btSoftBody::fMaterial::DebugDraw)) continue;
-				idraw->drawLine(n.m_x-btVector3(scl,0,0),n.m_x+btVector3(scl,0,0),btVector3(1,0,0));
-				idraw->drawLine(n.m_x-btVector3(0,scl,0),n.m_x+btVector3(0,scl,0),btVector3(0,1,0));
-				idraw->drawLine(n.m_x-btVector3(0,0,scl),n.m_x+btVector3(0,0,scl),btVector3(0,0,1));
+				const btSoftBody::Node& n = psb->m_nodes[i];
+				if (0 == (n.m_material->m_flags & btSoftBody::fMaterial::DebugDraw)) continue;
+				idraw->drawLine(n.m_x - btVector3(scl, 0, 0), n.m_x + btVector3(scl, 0, 0), btVector3(1, 0, 0));
+				idraw->drawLine(n.m_x - btVector3(0, scl, 0), n.m_x + btVector3(0, scl, 0), btVector3(0, 1, 0));
+				idraw->drawLine(n.m_x - btVector3(0, 0, scl), n.m_x + btVector3(0, 0, scl), btVector3(0, 0, 1));
 			}
 		}
-		/* Links	*/ 
-		if(0!=(drawflags&fDrawFlags::Links))
+		/* Links	*/
+		if (0 != (drawflags & fDrawFlags::Links))
 		{
-			for(i=0;i<psb->m_links.size();++i)
+			for (i = 0; i < psb->m_links.size(); ++i)
 			{
-				const btSoftBody::Link&	l=psb->m_links[i];
-				if(0==(l.m_material->m_flags&btSoftBody::fMaterial::DebugDraw)) continue;
-				idraw->drawLine(l.m_n[0]->m_x,l.m_n[1]->m_x,lcolor);
+				const btSoftBody::Link& l = psb->m_links[i];
+				if (0 == (l.m_material->m_flags & btSoftBody::fMaterial::DebugDraw)) continue;
+				idraw->drawLine(l.m_n[0]->m_x, l.m_n[1]->m_x, lcolor);
 			}
 		}
-		/* Normals	*/ 
-		if(0!=(drawflags&fDrawFlags::Normals))
+		/* Normals	*/
+		if (0 != (drawflags & fDrawFlags::Normals))
 		{
-			for(i=0;i<psb->m_nodes.size();++i)
+			for (i = 0; i < psb->m_nodes.size(); ++i)
 			{
-				const btSoftBody::Node&	n=psb->m_nodes[i];
-				if(0==(n.m_material->m_flags&btSoftBody::fMaterial::DebugDraw)) continue;
-				const btVector3			d=n.m_n*nscl;
-				idraw->drawLine(n.m_x,n.m_x+d,ncolor);
-				idraw->drawLine(n.m_x,n.m_x-d,ncolor*0.5);
+				const btSoftBody::Node& n = psb->m_nodes[i];
+				if (0 == (n.m_material->m_flags & btSoftBody::fMaterial::DebugDraw)) continue;
+				const btVector3 d = n.m_n * nscl;
+				idraw->drawLine(n.m_x, n.m_x + d, ncolor);
+				idraw->drawLine(n.m_x, n.m_x - d, ncolor * 0.5);
 			}
 		}
-		/* Contacts	*/ 
-		if(0!=(drawflags&fDrawFlags::Contacts))
+		/* Contacts	*/
+		if (0 != (drawflags & fDrawFlags::Contacts))
 		{
-			static const btVector3		axis[]={btVector3(1,0,0),
-				btVector3(0,1,0),
-				btVector3(0,0,1)};
-			for(i=0;i<psb->m_rcontacts.size();++i)
-			{		
-				const btSoftBody::RContact&	c=psb->m_rcontacts[i];
-				const btVector3				o=	c.m_node->m_x-c.m_cti.m_normal*
-					(btDot(c.m_node->m_x,c.m_cti.m_normal)+c.m_cti.m_offset);
-				const btVector3				x=btCross(c.m_cti.m_normal,axis[c.m_cti.m_normal.minAxis()]).normalized();
-				const btVector3				y=btCross(x,c.m_cti.m_normal).normalized();
-				idraw->drawLine(o-x*nscl,o+x*nscl,ccolor);
-				idraw->drawLine(o-y*nscl,o+y*nscl,ccolor);
-				idraw->drawLine(o,o+c.m_cti.m_normal*nscl*3,btVector3(1,1,0));
+			static const btVector3 axis[] = {btVector3(1, 0, 0),
+											 btVector3(0, 1, 0),
+											 btVector3(0, 0, 1)};
+			for (i = 0; i < psb->m_rcontacts.size(); ++i)
+			{
+				const btSoftBody::RContact& c = psb->m_rcontacts[i];
+				const btVector3 o = c.m_node->m_x - c.m_cti.m_normal *
+														(btDot(c.m_node->m_x, c.m_cti.m_normal) + c.m_cti.m_offset);
+				const btVector3 x = btCross(c.m_cti.m_normal, axis[c.m_cti.m_normal.minAxis()]).normalized();
+				const btVector3 y = btCross(x, c.m_cti.m_normal).normalized();
+				idraw->drawLine(o - x * nscl, o + x * nscl, ccolor);
+				idraw->drawLine(o - y * nscl, o + y * nscl, ccolor);
+				idraw->drawLine(o, o + c.m_cti.m_normal * nscl * 3, btVector3(1, 1, 0));
 			}
 		}
-		/* Faces	*/ 
-	if(0!=(drawflags&fDrawFlags::Faces))
+		/* Faces	*/
+		if (0 != (drawflags & fDrawFlags::Faces))
+		{
+			const btScalar scl = (btScalar)0.8;
+			const btScalar alp = (btScalar)1;
+			const btVector3 col(0, (btScalar)0.7, 0);
+			for (i = 0; i < psb->m_faces.size(); ++i)
+			{
+				const btSoftBody::Face& f = psb->m_faces[i];
+				if (0 == (f.m_material->m_flags & btSoftBody::fMaterial::DebugDraw)) continue;
+				const btVector3 x[] = {f.m_n[0]->m_x, f.m_n[1]->m_x, f.m_n[2]->m_x};
+				const btVector3 c = (x[0] + x[1] + x[2]) / 3;
+				idraw->drawTriangle((x[0] - c) * scl + c,
+									(x[1] - c) * scl + c,
+									(x[2] - c) * scl + c,
+									col, alp);
+			}
+		}
+		/* Tetras	*/
+		if (0 != (drawflags & fDrawFlags::Tetras))
+		{
+			const btScalar scl = (btScalar)0.8;
+			const btScalar alp = (btScalar)1;
+			const btVector3 col((btScalar)0.3, (btScalar)0.3, (btScalar)0.7);
+			for (int i = 0; i < psb->m_tetras.size(); ++i)
+			{
+				const btSoftBody::Tetra& t = psb->m_tetras[i];
+				if (0 == (t.m_material->m_flags & btSoftBody::fMaterial::DebugDraw)) continue;
+				const btVector3 x[] = {t.m_n[0]->m_x, t.m_n[1]->m_x, t.m_n[2]->m_x, t.m_n[3]->m_x};
+				const btVector3 c = (x[0] + x[1] + x[2] + x[3]) / 4;
+				idraw->drawTriangle((x[0] - c) * scl + c, (x[1] - c) * scl + c, (x[2] - c) * scl + c, col, alp);
+				idraw->drawTriangle((x[0] - c) * scl + c, (x[1] - c) * scl + c, (x[3] - c) * scl + c, col, alp);
+				idraw->drawTriangle((x[1] - c) * scl + c, (x[2] - c) * scl + c, (x[3] - c) * scl + c, col, alp);
+				idraw->drawTriangle((x[2] - c) * scl + c, (x[0] - c) * scl + c, (x[3] - c) * scl + c, col, alp);
+			}
+		}
+	}
+	/* Anchors	*/
+	if (0 != (drawflags & fDrawFlags::Anchors))
 	{
-		const btScalar	scl=(btScalar)0.8;
-		const btScalar	alp=(btScalar)1;
-		const btVector3	col(0,(btScalar)0.7,0);
-		for(i=0;i<psb->m_faces.size();++i)
+		for (i = 0; i < psb->m_anchors.size(); ++i)
 		{
-			const btSoftBody::Face&	f=psb->m_faces[i];
-			if(0==(f.m_material->m_flags&btSoftBody::fMaterial::DebugDraw)) continue;
-			const btVector3			x[]={f.m_n[0]->m_x,f.m_n[1]->m_x,f.m_n[2]->m_x};
-			const btVector3			c=(x[0]+x[1]+x[2])/3;
-			idraw->drawTriangle((x[0]-c)*scl+c,
-				(x[1]-c)*scl+c,
-				(x[2]-c)*scl+c,
-				col,alp);
-		}	
-	}
-	/* Tetras	*/ 
-	if(0!=(drawflags&fDrawFlags::Tetras))
-	{
-		const btScalar	scl=(btScalar)0.8;
-		const btScalar	alp=(btScalar)1;
-		const btVector3	col((btScalar)0.3,(btScalar)0.3,(btScalar)0.7);
-		for(int i=0;i<psb->m_tetras.size();++i)
-		{
-			const btSoftBody::Tetra&	t=psb->m_tetras[i];
-			if(0==(t.m_material->m_flags&btSoftBody::fMaterial::DebugDraw)) continue;
-			const btVector3				x[]={t.m_n[0]->m_x,t.m_n[1]->m_x,t.m_n[2]->m_x,t.m_n[3]->m_x};
-			const btVector3				c=(x[0]+x[1]+x[2]+x[3])/4;
-			idraw->drawTriangle((x[0]-c)*scl+c,(x[1]-c)*scl+c,(x[2]-c)*scl+c,col,alp);
-			idraw->drawTriangle((x[0]-c)*scl+c,(x[1]-c)*scl+c,(x[3]-c)*scl+c,col,alp);
-			idraw->drawTriangle((x[1]-c)*scl+c,(x[2]-c)*scl+c,(x[3]-c)*scl+c,col,alp);
-			idraw->drawTriangle((x[2]-c)*scl+c,(x[0]-c)*scl+c,(x[3]-c)*scl+c,col,alp);
-		}	
-	}
-	}
-	/* Anchors	*/ 
-	if(0!=(drawflags&fDrawFlags::Anchors))
-	{
-		for(i=0;i<psb->m_anchors.size();++i)
-		{
-			const btSoftBody::Anchor&	a=psb->m_anchors[i];
-			const btVector3				q=a.m_body->getWorldTransform()*a.m_local;
-			drawVertex(idraw,a.m_node->m_x,0.25,btVector3(1,0,0));
-			drawVertex(idraw,q,0.25,btVector3(0,1,0));
-			idraw->drawLine(a.m_node->m_x,q,btVector3(1,1,1));
+			const btSoftBody::Anchor& a = psb->m_anchors[i];
+			const btVector3 q = a.m_body->getWorldTransform() * a.m_local;
+			drawVertex(idraw, a.m_node->m_x, 0.25, btVector3(1, 0, 0));
+			drawVertex(idraw, q, 0.25, btVector3(0, 1, 0));
+			idraw->drawLine(a.m_node->m_x, q, btVector3(1, 1, 1));
 		}
-		for(i=0;i<psb->m_nodes.size();++i)
+		for (i = 0; i < psb->m_nodes.size(); ++i)
 		{
-			const btSoftBody::Node&	n=psb->m_nodes[i];		
-			if(0==(n.m_material->m_flags&btSoftBody::fMaterial::DebugDraw)) continue;
-			if(n.m_im<=0)
+			const btSoftBody::Node& n = psb->m_nodes[i];
+			if (0 == (n.m_material->m_flags & btSoftBody::fMaterial::DebugDraw)) continue;
+			if (n.m_im <= 0)
 			{
-				drawVertex(idraw,n.m_x,0.25,btVector3(1,0,0));
+				drawVertex(idraw, n.m_x, 0.25, btVector3(1, 0, 0));
 			}
 		}
 	}
-	
 
-	/* Notes	*/ 
-	if(0!=(drawflags&fDrawFlags::Notes))
+	/* Notes	*/
+	if (0 != (drawflags & fDrawFlags::Notes))
 	{
-		for(i=0;i<psb->m_notes.size();++i)
+		for (i = 0; i < psb->m_notes.size(); ++i)
 		{
-			const btSoftBody::Note&	n=psb->m_notes[i];
-			btVector3				p=n.m_offset;
-			for(int j=0;j<n.m_rank;++j)
+			const btSoftBody::Note& n = psb->m_notes[i];
+			btVector3 p = n.m_offset;
+			for (int j = 0; j < n.m_rank; ++j)
 			{
-				p+=n.m_nodes[j]->m_x*n.m_coords[j];
+				p += n.m_nodes[j]->m_x * n.m_coords[j];
 			}
-			idraw->draw3dText(p,n.m_text);
+			idraw->draw3dText(p, n.m_text);
 		}
 	}
-	/* Node tree	*/ 
-	if(0!=(drawflags&fDrawFlags::NodeTree))		DrawNodeTree(psb,idraw);
-	/* Face tree	*/ 
-	if(0!=(drawflags&fDrawFlags::FaceTree))		DrawFaceTree(psb,idraw);
-	/* Cluster tree	*/ 
-	if(0!=(drawflags&fDrawFlags::ClusterTree))	DrawClusterTree(psb,idraw);
-	/* Joints		*/ 
-	if(0!=(drawflags&fDrawFlags::Joints))
+	/* Node tree	*/
+	if (0 != (drawflags & fDrawFlags::NodeTree)) DrawNodeTree(psb, idraw);
+	/* Face tree	*/
+	if (0 != (drawflags & fDrawFlags::FaceTree)) DrawFaceTree(psb, idraw);
+	/* Cluster tree	*/
+	if (0 != (drawflags & fDrawFlags::ClusterTree)) DrawClusterTree(psb, idraw);
+	/* Joints		*/
+	if (0 != (drawflags & fDrawFlags::Joints))
 	{
-		for(i=0;i<psb->m_joints.size();++i)
+		for (i = 0; i < psb->m_joints.size(); ++i)
 		{
-			const btSoftBody::Joint*	pj=psb->m_joints[i];
-			switch(pj->Type())
+			const btSoftBody::Joint* pj = psb->m_joints[i];
+			switch (pj->Type())
 			{
-			case	btSoftBody::Joint::eType::Linear:
+				case btSoftBody::Joint::eType::Linear:
 				{
-					const btSoftBody::LJoint*	pjl=(const btSoftBody::LJoint*)pj;
-					const btVector3	a0=pj->m_bodies[0].xform()*pjl->m_refs[0];
-					const btVector3	a1=pj->m_bodies[1].xform()*pjl->m_refs[1];
-					idraw->drawLine(pj->m_bodies[0].xform().getOrigin(),a0,btVector3(1,1,0));
-					idraw->drawLine(pj->m_bodies[1].xform().getOrigin(),a1,btVector3(0,1,1));
-					drawVertex(idraw,a0,0.25,btVector3(1,1,0));
-					drawVertex(idraw,a1,0.25,btVector3(0,1,1));
+					const btSoftBody::LJoint* pjl = (const btSoftBody::LJoint*)pj;
+					const btVector3 a0 = pj->m_bodies[0].xform() * pjl->m_refs[0];
+					const btVector3 a1 = pj->m_bodies[1].xform() * pjl->m_refs[1];
+					idraw->drawLine(pj->m_bodies[0].xform().getOrigin(), a0, btVector3(1, 1, 0));
+					idraw->drawLine(pj->m_bodies[1].xform().getOrigin(), a1, btVector3(0, 1, 1));
+					drawVertex(idraw, a0, 0.25, btVector3(1, 1, 0));
+					drawVertex(idraw, a1, 0.25, btVector3(0, 1, 1));
 				}
 				break;
-			case	btSoftBody::Joint::eType::Angular:
+				case btSoftBody::Joint::eType::Angular:
 				{
 					//const btSoftBody::AJoint*	pja=(const btSoftBody::AJoint*)pj;
-					const btVector3	o0=pj->m_bodies[0].xform().getOrigin();
-					const btVector3	o1=pj->m_bodies[1].xform().getOrigin();
-					const btVector3	a0=pj->m_bodies[0].xform().getBasis()*pj->m_refs[0];
-					const btVector3	a1=pj->m_bodies[1].xform().getBasis()*pj->m_refs[1];
-					idraw->drawLine(o0,o0+a0*10,btVector3(1,1,0));
-					idraw->drawLine(o0,o0+a1*10,btVector3(1,1,0));
-					idraw->drawLine(o1,o1+a0*10,btVector3(0,1,1));
-					idraw->drawLine(o1,o1+a1*10,btVector3(0,1,1));
+					const btVector3 o0 = pj->m_bodies[0].xform().getOrigin();
+					const btVector3 o1 = pj->m_bodies[1].xform().getOrigin();
+					const btVector3 a0 = pj->m_bodies[0].xform().getBasis() * pj->m_refs[0];
+					const btVector3 a1 = pj->m_bodies[1].xform().getBasis() * pj->m_refs[1];
+					idraw->drawLine(o0, o0 + a0 * 10, btVector3(1, 1, 0));
+					idraw->drawLine(o0, o0 + a1 * 10, btVector3(1, 1, 0));
+					idraw->drawLine(o1, o1 + a0 * 10, btVector3(0, 1, 1));
+					idraw->drawLine(o1, o1 + a1 * 10, btVector3(0, 1, 1));
 					break;
 				}
 				default:
 				{
 				}
-					
-			}		
+			}
 		}
 	}
 }
 
 //
-void			btSoftBodyHelpers::DrawInfos(		btSoftBody* psb,
-											 btIDebugDraw* idraw,
-											 bool masses,
-											 bool areas,
-											 bool /*stress*/)
+void btSoftBodyHelpers::DrawInfos(btSoftBody* psb,
+								  btIDebugDraw* idraw,
+								  bool masses,
+								  bool areas,
+								  bool /*stress*/)
 {
-	for(int i=0;i<psb->m_nodes.size();++i)
+	for (int i = 0; i < psb->m_nodes.size(); ++i)
 	{
-		const btSoftBody::Node&	n=psb->m_nodes[i];
-		char					text[2048]={0};
-		char					buff[1024];
-		if(masses)
+		const btSoftBody::Node& n = psb->m_nodes[i];
+		char text[2048] = {0};
+		char buff[1024];
+		if (masses)
 		{
-			sprintf(buff," M(%.2f)",1/n.m_im);
-			strcat(text,buff);
+			sprintf(buff, " M(%.2f)", 1 / n.m_im);
+			strcat(text, buff);
 		}
-		if(areas)
+		if (areas)
 		{
-			sprintf(buff," A(%.2f)",n.m_area);
-			strcat(text,buff);
+			sprintf(buff, " A(%.2f)", n.m_area);
+			strcat(text, buff);
 		}
-		if(text[0]) idraw->draw3dText(n.m_x,text);
+		if (text[0]) idraw->draw3dText(n.m_x, text);
 	}
 }
 
 //
-void			btSoftBodyHelpers::DrawNodeTree(	btSoftBody* psb,
-												btIDebugDraw* idraw,
-												int mindepth,
-												int maxdepth)
+void btSoftBodyHelpers::DrawNodeTree(btSoftBody* psb,
+									 btIDebugDraw* idraw,
+									 int mindepth,
+									 int maxdepth)
 {
-	drawTree(idraw,psb->m_ndbvt.m_root,0,btVector3(1,0,1),btVector3(1,1,1),mindepth,maxdepth);
+	drawTree(idraw, psb->m_ndbvt.m_root, 0, btVector3(1, 0, 1), btVector3(1, 1, 1), mindepth, maxdepth);
 }
 
 //
-void			btSoftBodyHelpers::DrawFaceTree(	btSoftBody* psb,
-												btIDebugDraw* idraw,
-												int mindepth,
-												int maxdepth)
+void btSoftBodyHelpers::DrawFaceTree(btSoftBody* psb,
+									 btIDebugDraw* idraw,
+									 int mindepth,
+									 int maxdepth)
 {
-	drawTree(idraw,psb->m_fdbvt.m_root,0,btVector3(0,1,0),btVector3(1,0,0),mindepth,maxdepth);
+	drawTree(idraw, psb->m_fdbvt.m_root, 0, btVector3(0, 1, 0), btVector3(1, 0, 0), mindepth, maxdepth);
 }
 
 //
-void			btSoftBodyHelpers::DrawClusterTree(	btSoftBody* psb,
-												   btIDebugDraw* idraw,
-												   int mindepth,
-												   int maxdepth)
+void btSoftBodyHelpers::DrawClusterTree(btSoftBody* psb,
+										btIDebugDraw* idraw,
+										int mindepth,
+										int maxdepth)
 {
-	drawTree(idraw,psb->m_cdbvt.m_root,0,btVector3(0,1,1),btVector3(1,0,0),mindepth,maxdepth);
+	drawTree(idraw, psb->m_cdbvt.m_root, 0, btVector3(0, 1, 1), btVector3(1, 0, 0), mindepth, maxdepth);
 }
 
-
 //The btSoftBody object from the BulletSDK includes an array of Nodes and Links. These links appear
-// to be first set up to connect a node to between 5 and 6 of its neighbors [480 links], 
-//and then to the rest of the nodes after the execution of the Floyd-Warshall graph algorithm 
-//[another 930 links]. 
+// to be first set up to connect a node to between 5 and 6 of its neighbors [480 links],
+//and then to the rest of the nodes after the execution of the Floyd-Warshall graph algorithm
+//[another 930 links].
 //The way the links are stored by default, we have a number of cases where adjacent links share a node in common
-// - this leads to the creation of a data dependency through memory. 
-//The PSolve_Links() function reads and writes nodes as it iterates over each link. 
-//So, we now have the possibility of a data dependency between iteration X 
-//that processes link L with iteration X+1 that processes link L+1 
-//because L and L+1 have one node in common, and iteration X updates the positions of that node, 
+// - this leads to the creation of a data dependency through memory.
+//The PSolve_Links() function reads and writes nodes as it iterates over each link.
+//So, we now have the possibility of a data dependency between iteration X
+//that processes link L with iteration X+1 that processes link L+1
+//because L and L+1 have one node in common, and iteration X updates the positions of that node,
 //and iteration X+1 reads in the position of that shared node.
 //
-//Such a memory dependency limits the ability of a modern CPU to speculate beyond 
-//a certain point because it has to respect a possible dependency 
-//- this prevents the CPU from making full use of its out-of-order resources. 
-//If we re-order the links such that we minimize the cases where a link L and L+1 share a common node, 
-//we create a temporal gap between when the node position is written, 
-//and when it is subsequently read. This in turn allows the CPU to continue execution without 
-//risking a dependency violation. Such a reordering would result in significant speedups on 
-//modern CPUs with lots of execution resources. 
-//In our testing, we see it have a tremendous impact not only on the A7, 
-//but also on all x86 cores that ship with modern Macs. 
-//The attached source file includes a single function (ReoptimizeLinkOrder) which can be called on a 
-//btSoftBody object in the solveConstraints() function before the actual solver is invoked, 
+//Such a memory dependency limits the ability of a modern CPU to speculate beyond
+//a certain point because it has to respect a possible dependency
+//- this prevents the CPU from making full use of its out-of-order resources.
+//If we re-order the links such that we minimize the cases where a link L and L+1 share a common node,
+//we create a temporal gap between when the node position is written,
+//and when it is subsequently read. This in turn allows the CPU to continue execution without
+//risking a dependency violation. Such a reordering would result in significant speedups on
+//modern CPUs with lots of execution resources.
+//In our testing, we see it have a tremendous impact not only on the A7,
+//but also on all x86 cores that ship with modern Macs.
+//The attached source file includes a single function (ReoptimizeLinkOrder) which can be called on a
+//btSoftBody object in the solveConstraints() function before the actual solver is invoked,
 //or right after generateBendingConstraints() once we have all 1410 links.
 
-
 //===================================================================
 //
 //
-// This function takes in a list of interdependent Links and tries 
+// This function takes in a list of interdependent Links and tries
 // to maximize the distance between calculation
 // of dependent links.  This increases the amount of parallelism that can
 // be exploited by out-of-order instruction processors with large but
@@ -522,93 +521,103 @@ void			btSoftBodyHelpers::DrawClusterTree(	btSoftBody* psb,
 //===================================================================
 
 // A small structure to track lists of dependent link calculations
-class LinkDeps_t {
-	public:
-	int value;			// A link calculation that is dependent on this one
-		// Positive values = "input A" while negative values = "input B"
-	LinkDeps_t *next;	// Next dependence in the list
+class LinkDeps_t
+{
+public:
+	int value;         // A link calculation that is dependent on this one
+					   // Positive values = "input A" while negative values = "input B"
+	LinkDeps_t* next;  // Next dependence in the list
 };
-typedef LinkDeps_t *LinkDepsPtr_t;
+typedef LinkDeps_t* LinkDepsPtr_t;
 
 // Dependency list constants
-#define REOP_NOT_DEPENDENT	-1
-#define REOP_NODE_COMPLETE	-2	// Must be less than REOP_NOT_DEPENDENT
+#define REOP_NOT_DEPENDENT -1
+#define REOP_NODE_COMPLETE -2  // Must be less than REOP_NOT_DEPENDENT
 
-
-void btSoftBodyHelpers::ReoptimizeLinkOrder(btSoftBody *psb /* This can be replaced by a btSoftBody pointer */)
+void btSoftBodyHelpers::ReoptimizeLinkOrder(btSoftBody* psb /* This can be replaced by a btSoftBody pointer */)
 {
-	int i, nLinks=psb->m_links.size(), nNodes=psb->m_nodes.size();
-	btSoftBody::Link *lr;
+	int i, nLinks = psb->m_links.size(), nNodes = psb->m_nodes.size();
+	btSoftBody::Link* lr;
 	int ar, br;
-	btSoftBody::Node *node0 = &(psb->m_nodes[0]);
-	btSoftBody::Node *node1 = &(psb->m_nodes[1]);
+	btSoftBody::Node* node0 = &(psb->m_nodes[0]);
+	btSoftBody::Node* node1 = &(psb->m_nodes[1]);
 	LinkDepsPtr_t linkDep;
 	int readyListHead, readyListTail, linkNum, linkDepFrees, depLink;
-	
+
 	// Allocate temporary buffers
-	int *nodeWrittenAt = new int[nNodes+1];	// What link calculation produced this node's current values?
-	int *linkDepA = new int[nLinks];			// Link calculation input is dependent upon prior calculation #N
-	int *linkDepB = new int[nLinks];
-	int *readyList = new int[nLinks];		// List of ready-to-process link calculations (# of links, maximum)
-	LinkDeps_t *linkDepFreeList = new LinkDeps_t[2*nLinks];		// Dependent-on-me list elements (2x# of links, maximum)
-	LinkDepsPtr_t *linkDepListStarts = new LinkDepsPtr_t[nLinks];	// Start nodes of dependent-on-me lists, one for each link
-		
+	int* nodeWrittenAt = new int[nNodes + 1];  // What link calculation produced this node's current values?
+	int* linkDepA = new int[nLinks];           // Link calculation input is dependent upon prior calculation #N
+	int* linkDepB = new int[nLinks];
+	int* readyList = new int[nLinks];                              // List of ready-to-process link calculations (# of links, maximum)
+	LinkDeps_t* linkDepFreeList = new LinkDeps_t[2 * nLinks];      // Dependent-on-me list elements (2x# of links, maximum)
+	LinkDepsPtr_t* linkDepListStarts = new LinkDepsPtr_t[nLinks];  // Start nodes of dependent-on-me lists, one for each link
+
 	// Copy the original, unsorted links to a side buffer
-	btSoftBody::Link *linkBuffer = new btSoftBody::Link[nLinks];
-	memcpy(linkBuffer, &(psb->m_links[0]), sizeof(btSoftBody::Link)*nLinks);
+	btSoftBody::Link* linkBuffer = new btSoftBody::Link[nLinks];
+	memcpy(linkBuffer, &(psb->m_links[0]), sizeof(btSoftBody::Link) * nLinks);
 
 	// Clear out the node setup and ready list
-	for (i=0; i < nNodes+1; i++) {
+	for (i = 0; i < nNodes + 1; i++)
+	{
 		nodeWrittenAt[i] = REOP_NOT_DEPENDENT;
 	}
-	for (i=0; i < nLinks; i++) {
+	for (i = 0; i < nLinks; i++)
+	{
 		linkDepListStarts[i] = NULL;
 	}
 	readyListHead = readyListTail = linkDepFrees = 0;
 
 	// Initial link analysis to set up data structures
-	for (i=0; i < nLinks; i++) {
-	
+	for (i = 0; i < nLinks; i++)
+	{
 		// Note which prior link calculations we are dependent upon & build up dependence lists
 		lr = &(psb->m_links[i]);
-		ar = (lr->m_n[0] - node0)/(node1 - node0);
-		br = (lr->m_n[1] - node0)/(node1 - node0);
-		if (nodeWrittenAt[ar] > REOP_NOT_DEPENDENT) {
+		ar = (lr->m_n[0] - node0) / (node1 - node0);
+		br = (lr->m_n[1] - node0) / (node1 - node0);
+		if (nodeWrittenAt[ar] > REOP_NOT_DEPENDENT)
+		{
 			linkDepA[i] = nodeWrittenAt[ar];
 			linkDep = &linkDepFreeList[linkDepFrees++];
 			linkDep->value = i;
 			linkDep->next = linkDepListStarts[nodeWrittenAt[ar]];
 			linkDepListStarts[nodeWrittenAt[ar]] = linkDep;
-		} else {
+		}
+		else
+		{
 			linkDepA[i] = REOP_NOT_DEPENDENT;
 		}
-		if (nodeWrittenAt[br] > REOP_NOT_DEPENDENT) {
+		if (nodeWrittenAt[br] > REOP_NOT_DEPENDENT)
+		{
 			linkDepB[i] = nodeWrittenAt[br];
 			linkDep = &linkDepFreeList[linkDepFrees++];
-			linkDep->value = -(i+1);
+			linkDep->value = -(i + 1);
 			linkDep->next = linkDepListStarts[nodeWrittenAt[br]];
 			linkDepListStarts[nodeWrittenAt[br]] = linkDep;
-		} else {
+		}
+		else
+		{
 			linkDepB[i] = REOP_NOT_DEPENDENT;
 		}
-		
+
 		// Add this link to the initial ready list, if it is not dependent on any other links
-		if ((linkDepA[i] == REOP_NOT_DEPENDENT) && (linkDepB[i] == REOP_NOT_DEPENDENT)) {
+		if ((linkDepA[i] == REOP_NOT_DEPENDENT) && (linkDepB[i] == REOP_NOT_DEPENDENT))
+		{
 			readyList[readyListTail++] = i;
-			linkDepA[i] = linkDepB[i] = REOP_NODE_COMPLETE;	// Probably not needed now
+			linkDepA[i] = linkDepB[i] = REOP_NODE_COMPLETE;  // Probably not needed now
 		}
-		
+
 		// Update the nodes to mark which ones are calculated by this link
 		nodeWrittenAt[ar] = nodeWrittenAt[br] = i;
 	}
-	
+
 	// Process the ready list and create the sorted list of links
 	// -- By treating the ready list as a queue, we maximize the distance between any
 	//    inter-dependent node calculations
 	// -- All other (non-related) nodes in the ready list will automatically be inserted
 	//    in between each set of inter-dependent link calculations by this loop
 	i = 0;
-	while (readyListHead != readyListTail) {
+	while (readyListHead != readyListTail)
+	{
 		// Use ready list to select the next link to process
 		linkNum = readyList[readyListHead++];
 		// Copy the next-to-calculate link back into the original link array
@@ -616,180 +625,183 @@ void btSoftBodyHelpers::ReoptimizeLinkOrder(btSoftBody *psb /* This can be repla
 
 		// Free up any link inputs that are dependent on this one
 		linkDep = linkDepListStarts[linkNum];
-		while (linkDep) {
+		while (linkDep)
+		{
 			depLink = linkDep->value;
-			if (depLink >= 0) {
+			if (depLink >= 0)
+			{
 				linkDepA[depLink] = REOP_NOT_DEPENDENT;
-			} else {
+			}
+			else
+			{
 				depLink = -depLink - 1;
 				linkDepB[depLink] = REOP_NOT_DEPENDENT;
 			}
 			// Add this dependent link calculation to the ready list if *both* inputs are clear
-			if ((linkDepA[depLink] == REOP_NOT_DEPENDENT) && (linkDepB[depLink] == REOP_NOT_DEPENDENT)) {
+			if ((linkDepA[depLink] == REOP_NOT_DEPENDENT) && (linkDepB[depLink] == REOP_NOT_DEPENDENT))
+			{
 				readyList[readyListTail++] = depLink;
-				linkDepA[depLink] = linkDepB[depLink] = REOP_NODE_COMPLETE;	// Probably not needed now
+				linkDepA[depLink] = linkDepB[depLink] = REOP_NODE_COMPLETE;  // Probably not needed now
 			}
 			linkDep = linkDep->next;
 		}
 	}
 
 	// Delete the temporary buffers
-	delete [] nodeWrittenAt;
-	delete [] linkDepA;
-	delete [] linkDepB;
-	delete [] readyList;
-	delete [] linkDepFreeList;
-	delete [] linkDepListStarts;
-	delete [] linkBuffer;
+	delete[] nodeWrittenAt;
+	delete[] linkDepA;
+	delete[] linkDepB;
+	delete[] readyList;
+	delete[] linkDepFreeList;
+	delete[] linkDepListStarts;
+	delete[] linkBuffer;
 }
 
-
 //
-void			btSoftBodyHelpers::DrawFrame(		btSoftBody* psb,
-											 btIDebugDraw* idraw)
+void btSoftBodyHelpers::DrawFrame(btSoftBody* psb,
+								  btIDebugDraw* idraw)
 {
-	if(psb->m_pose.m_bframe)
+	if (psb->m_pose.m_bframe)
 	{
-		static const btScalar	ascl=10;
-		static const btScalar	nscl=(btScalar)0.1;
-		const btVector3			com=psb->m_pose.m_com;
-		const btMatrix3x3		trs=psb->m_pose.m_rot*psb->m_pose.m_scl;
-		const btVector3			Xaxis=(trs*btVector3(1,0,0)).normalized();
-		const btVector3			Yaxis=(trs*btVector3(0,1,0)).normalized();
-		const btVector3			Zaxis=(trs*btVector3(0,0,1)).normalized();
-		idraw->drawLine(com,com+Xaxis*ascl,btVector3(1,0,0));
-		idraw->drawLine(com,com+Yaxis*ascl,btVector3(0,1,0));
-		idraw->drawLine(com,com+Zaxis*ascl,btVector3(0,0,1));
-		for(int i=0;i<psb->m_pose.m_pos.size();++i)
+		static const btScalar ascl = 10;
+		static const btScalar nscl = (btScalar)0.1;
+		const btVector3 com = psb->m_pose.m_com;
+		const btMatrix3x3 trs = psb->m_pose.m_rot * psb->m_pose.m_scl;
+		const btVector3 Xaxis = (trs * btVector3(1, 0, 0)).normalized();
+		const btVector3 Yaxis = (trs * btVector3(0, 1, 0)).normalized();
+		const btVector3 Zaxis = (trs * btVector3(0, 0, 1)).normalized();
+		idraw->drawLine(com, com + Xaxis * ascl, btVector3(1, 0, 0));
+		idraw->drawLine(com, com + Yaxis * ascl, btVector3(0, 1, 0));
+		idraw->drawLine(com, com + Zaxis * ascl, btVector3(0, 0, 1));
+		for (int i = 0; i < psb->m_pose.m_pos.size(); ++i)
 		{
-			const btVector3	x=com+trs*psb->m_pose.m_pos[i];
-			drawVertex(idraw,x,nscl,btVector3(1,0,1));
+			const btVector3 x = com + trs * psb->m_pose.m_pos[i];
+			drawVertex(idraw, x, nscl, btVector3(1, 0, 1));
 		}
 	}
 }
 
 //
-btSoftBody*		btSoftBodyHelpers::CreateRope(	btSoftBodyWorldInfo& worldInfo, const btVector3& from,
-											  const btVector3& to,
-											  int res,
-											  int fixeds)
+btSoftBody* btSoftBodyHelpers::CreateRope(btSoftBodyWorldInfo& worldInfo, const btVector3& from,
+										  const btVector3& to,
+										  int res,
+										  int fixeds)
 {
-	/* Create nodes	*/ 
-	const int		r=res+2;
-	btVector3*		x=new btVector3[r];
-	btScalar*		m=new btScalar[r];
+	/* Create nodes	*/
+	const int r = res + 2;
+	btVector3* x = new btVector3[r];
+	btScalar* m = new btScalar[r];
 	int i;
 
-	for(i=0;i<r;++i)
+	for (i = 0; i < r; ++i)
 	{
-		const btScalar	t=i/(btScalar)(r-1);
-		x[i]=lerp(from,to,t);
-		m[i]=1;
+		const btScalar t = i / (btScalar)(r - 1);
+		x[i] = lerp(from, to, t);
+		m[i] = 1;
 	}
-	btSoftBody*		psb= new btSoftBody(&worldInfo,r,x,m);
-	if(fixeds&1) psb->setMass(0,0);
-	if(fixeds&2) psb->setMass(r-1,0);
+	btSoftBody* psb = new btSoftBody(&worldInfo, r, x, m);
+	if (fixeds & 1) psb->setMass(0, 0);
+	if (fixeds & 2) psb->setMass(r - 1, 0);
 	delete[] x;
 	delete[] m;
-	/* Create links	*/ 
-	for(i=1;i<r;++i)
+	/* Create links	*/
+	for (i = 1; i < r; ++i)
 	{
-		psb->appendLink(i-1,i);
+		psb->appendLink(i - 1, i);
 	}
-	/* Finished		*/ 
-	return(psb);
+	/* Finished		*/
+	return (psb);
 }
 
 //
-btSoftBody*		btSoftBodyHelpers::CreatePatch(btSoftBodyWorldInfo& worldInfo,const btVector3& corner00,
-											   const btVector3& corner10,
-											   const btVector3& corner01,
-											   const btVector3& corner11,
-											   int resx,
-											   int resy,
-											   int fixeds,
-											   bool gendiags)
+btSoftBody* btSoftBodyHelpers::CreatePatch(btSoftBodyWorldInfo& worldInfo, const btVector3& corner00,
+										   const btVector3& corner10,
+										   const btVector3& corner01,
+										   const btVector3& corner11,
+										   int resx,
+										   int resy,
+										   int fixeds,
+										   bool gendiags)
 {
-#define IDX(_x_,_y_)	((_y_)*rx+(_x_))
-	/* Create nodes	*/ 
-	if((resx<2)||(resy<2)) return(0);
-	const int	rx=resx;
-	const int	ry=resy;
-	const int	tot=rx*ry;
-	btVector3*	x=new btVector3[tot];
-	btScalar*	m=new btScalar[tot];
+#define IDX(_x_, _y_) ((_y_)*rx + (_x_))
+	/* Create nodes	*/
+	if ((resx < 2) || (resy < 2)) return (0);
+	const int rx = resx;
+	const int ry = resy;
+	const int tot = rx * ry;
+	btVector3* x = new btVector3[tot];
+	btScalar* m = new btScalar[tot];
 	int iy;
 
-	for(iy=0;iy<ry;++iy)
+	for (iy = 0; iy < ry; ++iy)
 	{
-		const btScalar	ty=iy/(btScalar)(ry-1);
-		const btVector3	py0=lerp(corner00,corner01,ty);
-		const btVector3	py1=lerp(corner10,corner11,ty);
-		for(int ix=0;ix<rx;++ix)
+		const btScalar ty = iy / (btScalar)(ry - 1);
+		const btVector3 py0 = lerp(corner00, corner01, ty);
+		const btVector3 py1 = lerp(corner10, corner11, ty);
+		for (int ix = 0; ix < rx; ++ix)
 		{
-			const btScalar	tx=ix/(btScalar)(rx-1);
-			x[IDX(ix,iy)]=lerp(py0,py1,tx);
-			m[IDX(ix,iy)]=1;
+			const btScalar tx = ix / (btScalar)(rx - 1);
+			x[IDX(ix, iy)] = lerp(py0, py1, tx);
+			m[IDX(ix, iy)] = 1;
 		}
 	}
-	btSoftBody*		psb=new btSoftBody(&worldInfo,tot,x,m);
-	if(fixeds&1)	psb->setMass(IDX(0,0),0);
-	if(fixeds&2)	psb->setMass(IDX(rx-1,0),0);
-	if(fixeds&4)	psb->setMass(IDX(0,ry-1),0);
-	if(fixeds&8)	psb->setMass(IDX(rx-1,ry-1),0);
+	btSoftBody* psb = new btSoftBody(&worldInfo, tot, x, m);
+	if (fixeds & 1) psb->setMass(IDX(0, 0), 0);
+	if (fixeds & 2) psb->setMass(IDX(rx - 1, 0), 0);
+	if (fixeds & 4) psb->setMass(IDX(0, ry - 1), 0);
+	if (fixeds & 8) psb->setMass(IDX(rx - 1, ry - 1), 0);
 	delete[] x;
 	delete[] m;
-	/* Create links	and faces */ 
-	for(iy=0;iy<ry;++iy)
+	/* Create links	and faces */
+	for (iy = 0; iy < ry; ++iy)
 	{
-		for(int ix=0;ix<rx;++ix)
+		for (int ix = 0; ix < rx; ++ix)
 		{
-			const int	idx=IDX(ix,iy);
-			const bool	mdx=(ix+1)<rx;
-			const bool	mdy=(iy+1)<ry;
-			if(mdx) psb->appendLink(idx,IDX(ix+1,iy));
-			if(mdy) psb->appendLink(idx,IDX(ix,iy+1));
-			if(mdx&&mdy)
+			const int idx = IDX(ix, iy);
+			const bool mdx = (ix + 1) < rx;
+			const bool mdy = (iy + 1) < ry;
+			if (mdx) psb->appendLink(idx, IDX(ix + 1, iy));
+			if (mdy) psb->appendLink(idx, IDX(ix, iy + 1));
+			if (mdx && mdy)
 			{
-				if((ix+iy)&1)
+				if ((ix + iy) & 1)
 				{
-					psb->appendFace(IDX(ix,iy),IDX(ix+1,iy),IDX(ix+1,iy+1));
-					psb->appendFace(IDX(ix,iy),IDX(ix+1,iy+1),IDX(ix,iy+1));
-					if(gendiags)
+					psb->appendFace(IDX(ix, iy), IDX(ix + 1, iy), IDX(ix + 1, iy + 1));
+					psb->appendFace(IDX(ix, iy), IDX(ix + 1, iy + 1), IDX(ix, iy + 1));
+					if (gendiags)
 					{
-						psb->appendLink(IDX(ix,iy),IDX(ix+1,iy+1));
+						psb->appendLink(IDX(ix, iy), IDX(ix + 1, iy + 1));
 					}
 				}
 				else
 				{
-					psb->appendFace(IDX(ix,iy+1),IDX(ix,iy),IDX(ix+1,iy));
-					psb->appendFace(IDX(ix,iy+1),IDX(ix+1,iy),IDX(ix+1,iy+1));
-					if(gendiags)
+					psb->appendFace(IDX(ix, iy + 1), IDX(ix, iy), IDX(ix + 1, iy));
+					psb->appendFace(IDX(ix, iy + 1), IDX(ix + 1, iy), IDX(ix + 1, iy + 1));
+					if (gendiags)
 					{
-						psb->appendLink(IDX(ix+1,iy),IDX(ix,iy+1));
+						psb->appendLink(IDX(ix + 1, iy), IDX(ix, iy + 1));
 					}
 				}
 			}
 		}
 	}
-	/* Finished		*/ 
+	/* Finished		*/
 #undef IDX
-	return(psb);
+	return (psb);
 }
 
 //
-btSoftBody*		btSoftBodyHelpers::CreatePatchUV(btSoftBodyWorldInfo& worldInfo,
-												 const btVector3& corner00,
-												 const btVector3& corner10,
-												 const btVector3& corner01,
-												 const btVector3& corner11,
-												 int resx,
-												 int resy,
-												 int fixeds,
-												 bool gendiags,
-												 float* tex_coords)
+btSoftBody* btSoftBodyHelpers::CreatePatchUV(btSoftBodyWorldInfo& worldInfo,
+											 const btVector3& corner00,
+											 const btVector3& corner10,
+											 const btVector3& corner01,
+											 const btVector3& corner11,
+											 int resx,
+											 int resy,
+											 int fixeds,
+											 bool gendiags,
+											 float* tex_coords)
 {
-
 	/*
 	*
 	*  corners:
@@ -857,92 +869,92 @@ btSoftBody*		btSoftBodyHelpers::CreatePatchUV(btSoftBodyWorldInfo& worldInfo,
 	*
 	*/
 
-#define IDX(_x_,_y_)	((_y_)*rx+(_x_))
-	/* Create nodes		*/ 
-	if((resx<2)||(resy<2)) return(0);
-	const int	rx=resx;
-	const int	ry=resy;
-	const int	tot=rx*ry;
-	btVector3*	x=new btVector3[tot];
-	btScalar*	m=new btScalar[tot];
+#define IDX(_x_, _y_) ((_y_)*rx + (_x_))
+	/* Create nodes		*/
+	if ((resx < 2) || (resy < 2)) return (0);
+	const int rx = resx;
+	const int ry = resy;
+	const int tot = rx * ry;
+	btVector3* x = new btVector3[tot];
+	btScalar* m = new btScalar[tot];
 
 	int iy;
 
-	for(iy=0;iy<ry;++iy)
+	for (iy = 0; iy < ry; ++iy)
 	{
-		const btScalar	ty=iy/(btScalar)(ry-1);
-		const btVector3	py0=lerp(corner00,corner01,ty);
-		const btVector3	py1=lerp(corner10,corner11,ty);
-		for(int ix=0;ix<rx;++ix)
+		const btScalar ty = iy / (btScalar)(ry - 1);
+		const btVector3 py0 = lerp(corner00, corner01, ty);
+		const btVector3 py1 = lerp(corner10, corner11, ty);
+		for (int ix = 0; ix < rx; ++ix)
 		{
-			const btScalar	tx=ix/(btScalar)(rx-1);
-			x[IDX(ix,iy)]=lerp(py0,py1,tx);
-			m[IDX(ix,iy)]=1;
+			const btScalar tx = ix / (btScalar)(rx - 1);
+			x[IDX(ix, iy)] = lerp(py0, py1, tx);
+			m[IDX(ix, iy)] = 1;
 		}
 	}
-	btSoftBody*	psb=new btSoftBody(&worldInfo,tot,x,m);
-	if(fixeds&1)		psb->setMass(IDX(0,0),0);
-	if(fixeds&2)		psb->setMass(IDX(rx-1,0),0);
-	if(fixeds&4)		psb->setMass(IDX(0,ry-1),0);
-	if(fixeds&8)		psb->setMass(IDX(rx-1,ry-1),0);
-	if(fixeds&16)		psb->setMass(IDX((rx-1)/2,0),0);
-	if(fixeds&32)		psb->setMass(IDX(0,(ry-1)/2),0);
-	if(fixeds&64)		psb->setMass(IDX(rx-1,(ry-1)/2),0);
-	if(fixeds&128)		psb->setMass(IDX((rx-1)/2,ry-1),0);
-	if(fixeds&256)		psb->setMass(IDX((rx-1)/2,(ry-1)/2),0);
+	btSoftBody* psb = new btSoftBody(&worldInfo, tot, x, m);
+	if (fixeds & 1) psb->setMass(IDX(0, 0), 0);
+	if (fixeds & 2) psb->setMass(IDX(rx - 1, 0), 0);
+	if (fixeds & 4) psb->setMass(IDX(0, ry - 1), 0);
+	if (fixeds & 8) psb->setMass(IDX(rx - 1, ry - 1), 0);
+	if (fixeds & 16) psb->setMass(IDX((rx - 1) / 2, 0), 0);
+	if (fixeds & 32) psb->setMass(IDX(0, (ry - 1) / 2), 0);
+	if (fixeds & 64) psb->setMass(IDX(rx - 1, (ry - 1) / 2), 0);
+	if (fixeds & 128) psb->setMass(IDX((rx - 1) / 2, ry - 1), 0);
+	if (fixeds & 256) psb->setMass(IDX((rx - 1) / 2, (ry - 1) / 2), 0);
 	delete[] x;
 	delete[] m;
 
-
 	int z = 0;
-	/* Create links	and faces	*/ 
-	for(iy=0;iy<ry;++iy)
+	/* Create links	and faces	*/
+	for (iy = 0; iy < ry; ++iy)
 	{
-		for(int ix=0;ix<rx;++ix)
+		for (int ix = 0; ix < rx; ++ix)
 		{
-			const bool	mdx=(ix+1)<rx;
-			const bool	mdy=(iy+1)<ry;
+			const bool mdx = (ix + 1) < rx;
+			const bool mdy = (iy + 1) < ry;
 
-			int node00=IDX(ix,iy);
-			int node01=IDX(ix+1,iy);
-			int node10=IDX(ix,iy+1);
-			int node11=IDX(ix+1,iy+1);
+			int node00 = IDX(ix, iy);
+			int node01 = IDX(ix + 1, iy);
+			int node10 = IDX(ix, iy + 1);
+			int node11 = IDX(ix + 1, iy + 1);
 
-			if(mdx) psb->appendLink(node00,node01);
-			if(mdy) psb->appendLink(node00,node10);
-			if(mdx&&mdy)
+			if (mdx) psb->appendLink(node00, node01);
+			if (mdy) psb->appendLink(node00, node10);
+			if (mdx && mdy)
 			{
-				psb->appendFace(node00,node10,node11);
-				if (tex_coords) {
-					tex_coords[z+0]=CalculateUV(resx,resy,ix,iy,0);
-					tex_coords[z+1]=CalculateUV(resx,resy,ix,iy,1);
-					tex_coords[z+2]=CalculateUV(resx,resy,ix,iy,0);
-					tex_coords[z+3]=CalculateUV(resx,resy,ix,iy,2);
-					tex_coords[z+4]=CalculateUV(resx,resy,ix,iy,3);
-					tex_coords[z+5]=CalculateUV(resx,resy,ix,iy,2);
+				psb->appendFace(node00, node10, node11);
+				if (tex_coords)
+				{
+					tex_coords[z + 0] = CalculateUV(resx, resy, ix, iy, 0);
+					tex_coords[z + 1] = CalculateUV(resx, resy, ix, iy, 1);
+					tex_coords[z + 2] = CalculateUV(resx, resy, ix, iy, 0);
+					tex_coords[z + 3] = CalculateUV(resx, resy, ix, iy, 2);
+					tex_coords[z + 4] = CalculateUV(resx, resy, ix, iy, 3);
+					tex_coords[z + 5] = CalculateUV(resx, resy, ix, iy, 2);
 				}
-				psb->appendFace(node11,node01,node00);
-				if (tex_coords) {
-					tex_coords[z+6 ]=CalculateUV(resx,resy,ix,iy,3);
-					tex_coords[z+7 ]=CalculateUV(resx,resy,ix,iy,2);
-					tex_coords[z+8 ]=CalculateUV(resx,resy,ix,iy,3);
-					tex_coords[z+9 ]=CalculateUV(resx,resy,ix,iy,1);
-					tex_coords[z+10]=CalculateUV(resx,resy,ix,iy,0);
-					tex_coords[z+11]=CalculateUV(resx,resy,ix,iy,1);
+				psb->appendFace(node11, node01, node00);
+				if (tex_coords)
+				{
+					tex_coords[z + 6] = CalculateUV(resx, resy, ix, iy, 3);
+					tex_coords[z + 7] = CalculateUV(resx, resy, ix, iy, 2);
+					tex_coords[z + 8] = CalculateUV(resx, resy, ix, iy, 3);
+					tex_coords[z + 9] = CalculateUV(resx, resy, ix, iy, 1);
+					tex_coords[z + 10] = CalculateUV(resx, resy, ix, iy, 0);
+					tex_coords[z + 11] = CalculateUV(resx, resy, ix, iy, 1);
 				}
-				if (gendiags) psb->appendLink(node00,node11);
+				if (gendiags) psb->appendLink(node00, node11);
 				z += 12;
 			}
 		}
 	}
-	/* Finished	*/ 
+	/* Finished	*/
 #undef IDX
-	return(psb);
+	return (psb);
 }
 
-float   btSoftBodyHelpers::CalculateUV(int resx,int resy,int ix,int iy,int id)
+float btSoftBodyHelpers::CalculateUV(int resx, int resy, int ix, int iy, int id)
 {
-
 	/*
 	*
 	*
@@ -968,90 +980,93 @@ float   btSoftBodyHelpers::CalculateUV(int resx,int resy,int ix,int iy,int id)
 	*
 	*/
 
-	float tc=0.0f;
-	if (id == 0) {
-		tc = (1.0f/((resx-1))*ix);
+	float tc = 0.0f;
+	if (id == 0)
+	{
+		tc = (1.0f / ((resx - 1)) * ix);
 	}
-	else if (id==1) {
-		tc = (1.0f/((resy-1))*(resy-1-iy));
+	else if (id == 1)
+	{
+		tc = (1.0f / ((resy - 1)) * (resy - 1 - iy));
 	}
-	else if (id==2) {
-		tc = (1.0f/((resy-1))*(resy-1-iy-1));
+	else if (id == 2)
+	{
+		tc = (1.0f / ((resy - 1)) * (resy - 1 - iy - 1));
 	}
-	else if (id==3) {
-		tc = (1.0f/((resx-1))*(ix+1));
+	else if (id == 3)
+	{
+		tc = (1.0f / ((resx - 1)) * (ix + 1));
 	}
 	return tc;
 }
 //
-btSoftBody*		btSoftBodyHelpers::CreateEllipsoid(btSoftBodyWorldInfo& worldInfo,const btVector3& center,
-												   const btVector3& radius,
-												   int res)
+btSoftBody* btSoftBodyHelpers::CreateEllipsoid(btSoftBodyWorldInfo& worldInfo, const btVector3& center,
+											   const btVector3& radius,
+											   int res)
 {
-	struct	Hammersley
+	struct Hammersley
 	{
-		static void	Generate(btVector3* x,int n)
+		static void Generate(btVector3* x, int n)
 		{
-			for(int i=0;i<n;i++)
+			for (int i = 0; i < n; i++)
 			{
-				btScalar	p=0.5,t=0;
-				for(int j=i;j;p*=0.5,j>>=1) if(j&1) t+=p;
-				btScalar	w=2*t-1;
-				btScalar	a=(SIMD_PI+2*i*SIMD_PI)/n;
-				btScalar	s=btSqrt(1-w*w);
-				*x++=btVector3(s*btCos(a),s*btSin(a),w);
+				btScalar p = 0.5, t = 0;
+				for (int j = i; j; p *= 0.5, j >>= 1)
+					if (j & 1) t += p;
+				btScalar w = 2 * t - 1;
+				btScalar a = (SIMD_PI + 2 * i * SIMD_PI) / n;
+				btScalar s = btSqrt(1 - w * w);
+				*x++ = btVector3(s * btCos(a), s * btSin(a), w);
 			}
 		}
 	};
-	btAlignedObjectArray<btVector3>	vtx;
-	vtx.resize(3+res);
-	Hammersley::Generate(&vtx[0],vtx.size());
-	for(int i=0;i<vtx.size();++i)
+	btAlignedObjectArray<btVector3> vtx;
+	vtx.resize(3 + res);
+	Hammersley::Generate(&vtx[0], vtx.size());
+	for (int i = 0; i < vtx.size(); ++i)
 	{
-		vtx[i]=vtx[i]*radius+center;
+		vtx[i] = vtx[i] * radius + center;
 	}
-	return(CreateFromConvexHull(worldInfo,&vtx[0],vtx.size()));
+	return (CreateFromConvexHull(worldInfo, &vtx[0], vtx.size()));
 }
 
-
-
 //
-btSoftBody*		btSoftBodyHelpers::CreateFromTriMesh(btSoftBodyWorldInfo& worldInfo,const btScalar*	vertices,
-													 const int* triangles,
-													 int ntriangles, bool randomizeConstraints)
+btSoftBody* btSoftBodyHelpers::CreateFromTriMesh(btSoftBodyWorldInfo& worldInfo, const btScalar* vertices,
+												 const int* triangles,
+												 int ntriangles, bool randomizeConstraints)
 {
-	int		maxidx=0;
-	int i,j,ni;
+	int maxidx = 0;
+	int i, j, ni;
 
-	for(i=0,ni=ntriangles*3;i<ni;++i)
+	for (i = 0, ni = ntriangles * 3; i < ni; ++i)
 	{
-		maxidx=btMax(triangles[i],maxidx);
+		maxidx = btMax(triangles[i], maxidx);
 	}
 	++maxidx;
-	btAlignedObjectArray<bool>		chks;
-	btAlignedObjectArray<btVector3>	vtx;
-	chks.resize(maxidx*maxidx,false);
+	btAlignedObjectArray<bool> chks;
+	btAlignedObjectArray<btVector3> vtx;
+	chks.resize(maxidx * maxidx, false);
 	vtx.resize(maxidx);
-	for(i=0,j=0,ni=maxidx*3;i<ni;++j,i+=3)
+	for (i = 0, j = 0, ni = maxidx * 3; i < ni; ++j, i += 3)
 	{
-		vtx[j]=btVector3(vertices[i],vertices[i+1],vertices[i+2]);
+		vtx[j] = btVector3(vertices[i], vertices[i + 1], vertices[i + 2]);
 	}
-	btSoftBody*		psb=new btSoftBody(&worldInfo,vtx.size(),&vtx[0],0);
-	for( i=0,ni=ntriangles*3;i<ni;i+=3)
+	btSoftBody* psb = new btSoftBody(&worldInfo, vtx.size(), &vtx[0], 0);
+	for (i = 0, ni = ntriangles * 3; i < ni; i += 3)
 	{
-		const int idx[]={triangles[i],triangles[i+1],triangles[i+2]};
-#define IDX(_x_,_y_) ((_y_)*maxidx+(_x_))
-		for(int j=2,k=0;k<3;j=k++)
+		const int idx[] = {triangles[i], triangles[i + 1], triangles[i + 2]};
+#define IDX(_x_, _y_) ((_y_)*maxidx + (_x_))
+		for (int j = 2, k = 0; k < 3; j = k++)
 		{
-			if(!chks[IDX(idx[j],idx[k])])
+			if (!chks[IDX(idx[j], idx[k])])
 			{
-				chks[IDX(idx[j],idx[k])]=true;
-				chks[IDX(idx[k],idx[j])]=true;
-				psb->appendLink(idx[j],idx[k]);
+				chks[IDX(idx[j], idx[k])] = true;
+				chks[IDX(idx[k], idx[j])] = true;
+				psb->appendLink(idx[j], idx[k]);
 			}
 		}
 #undef IDX
-		psb->appendFace(idx[0],idx[1],idx[2]);
+		psb->appendFace(idx[0], idx[1], idx[2]);
 	}
 
 	if (randomizeConstraints)
@@ -1059,44 +1074,41 @@ btSoftBody*		btSoftBodyHelpers::CreateFromTriMesh(btSoftBodyWorldInfo& worldInfo
 		psb->randomizeConstraints();
 	}
 
-	return(psb);
+	return (psb);
 }
 
 //
-btSoftBody*		btSoftBodyHelpers::CreateFromConvexHull(btSoftBodyWorldInfo& worldInfo,	const btVector3* vertices,
-														int nvertices, bool randomizeConstraints)
+btSoftBody* btSoftBodyHelpers::CreateFromConvexHull(btSoftBodyWorldInfo& worldInfo, const btVector3* vertices,
+													int nvertices, bool randomizeConstraints)
 {
-	HullDesc		hdsc(QF_TRIANGLES,nvertices,vertices);
-	HullResult		hres;
-	HullLibrary		hlib;/*??*/ 
-	hdsc.mMaxVertices=nvertices;
-	hlib.CreateConvexHull(hdsc,hres);
-	btSoftBody*		psb=new btSoftBody(&worldInfo,(int)hres.mNumOutputVertices,
-		&hres.m_OutputVertices[0],0);
-	for(int i=0;i<(int)hres.mNumFaces;++i)
+	HullDesc hdsc(QF_TRIANGLES, nvertices, vertices);
+	HullResult hres;
+	HullLibrary hlib; /*??*/
+	hdsc.mMaxVertices = nvertices;
+	hlib.CreateConvexHull(hdsc, hres);
+	btSoftBody* psb = new btSoftBody(&worldInfo, (int)hres.mNumOutputVertices,
+									 &hres.m_OutputVertices[0], 0);
+	for (int i = 0; i < (int)hres.mNumFaces; ++i)
 	{
-		const int idx[]={	static_cast<int>(hres.m_Indices[i*3+0]),
-							static_cast<int>(hres.m_Indices[i*3+1]),
-							static_cast<int>(hres.m_Indices[i*3+2])};
-		if(idx[0]<idx[1]) psb->appendLink(	idx[0],idx[1]);
-		if(idx[1]<idx[2]) psb->appendLink(	idx[1],idx[2]);
-		if(idx[2]<idx[0]) psb->appendLink(	idx[2],idx[0]);
-		psb->appendFace(idx[0],idx[1],idx[2]);
+		const int idx[] = {static_cast<int>(hres.m_Indices[i * 3 + 0]),
+						   static_cast<int>(hres.m_Indices[i * 3 + 1]),
+						   static_cast<int>(hres.m_Indices[i * 3 + 2])};
+		if (idx[0] < idx[1]) psb->appendLink(idx[0], idx[1]);
+		if (idx[1] < idx[2]) psb->appendLink(idx[1], idx[2]);
+		if (idx[2] < idx[0]) psb->appendLink(idx[2], idx[0]);
+		psb->appendFace(idx[0], idx[1], idx[2]);
 	}
 	hlib.ReleaseResult(hres);
 	if (randomizeConstraints)
 	{
 		psb->randomizeConstraints();
 	}
-	return(psb);
+	return (psb);
 }
 
-
-
-
 static int nextLine(const char* buffer)
 {
-	int numBytesRead=0;
+	int numBytesRead = 0;
 
 	while (*buffer != '\n')
 	{
@@ -1104,8 +1116,7 @@ static int nextLine(const char* buffer)
 		numBytesRead++;
 	}
 
-	
-	if (buffer[0]==0x0a)
+	if (buffer[0] == 0x0a)
 	{
 		buffer++;
 		numBytesRead++;
@@ -1113,8 +1124,8 @@ static int nextLine(const char* buffer)
 	return numBytesRead;
 }
 
-/* Create from TetGen .ele, .face, .node data							*/ 
-btSoftBody*	btSoftBodyHelpers::CreateFromTetGenData(btSoftBodyWorldInfo& worldInfo,
+/* Create from TetGen .ele, .face, .node data							*/
+btSoftBody* btSoftBodyHelpers::CreateFromTetGenData(btSoftBodyWorldInfo& worldInfo,
 													const char* ele,
 													const char* face,
 													const char* node,
@@ -1122,38 +1133,38 @@ btSoftBody*	btSoftBodyHelpers::CreateFromTetGenData(btSoftBodyWorldInfo& worldIn
 													bool btetralinks,
 													bool bfacesfromtetras)
 {
-btAlignedObjectArray<btVector3>	pos;
-int								nnode=0;
-int								ndims=0;
-int								nattrb=0;
-int								hasbounds=0;
-int result = sscanf(node,"%d %d %d %d",&nnode,&ndims,&nattrb,&hasbounds);
-result = sscanf(node,"%d %d %d %d",&nnode,&ndims,&nattrb,&hasbounds);
-node += nextLine(node);
-
-pos.resize(nnode);
-for(int i=0;i<pos.size();++i)
-	{
-	int			index=0;
-	//int			bound=0;
-	float	x,y,z;
-	sscanf(node,"%d %f %f %f",&index,&x,&y,&z);
-
-//	sn>>index;
-//	sn>>x;sn>>y;sn>>z;
+	btAlignedObjectArray<btVector3> pos;
+	int nnode = 0;
+	int ndims = 0;
+	int nattrb = 0;
+	int hasbounds = 0;
+	int result = sscanf(node, "%d %d %d %d", &nnode, &ndims, &nattrb, &hasbounds);
+	result = sscanf(node, "%d %d %d %d", &nnode, &ndims, &nattrb, &hasbounds);
 	node += nextLine(node);
 
-	//for(int j=0;j<nattrb;++j) 
-	//	sn>>a;
+	pos.resize(nnode);
+	for (int i = 0; i < pos.size(); ++i)
+	{
+		int index = 0;
+		//int			bound=0;
+		float x, y, z;
+		sscanf(node, "%d %f %f %f", &index, &x, &y, &z);
 
-	//if(hasbounds) 
-	//	sn>>bound;
+		//	sn>>index;
+		//	sn>>x;sn>>y;sn>>z;
+		node += nextLine(node);
 
-	pos[index].setX(btScalar(x));
-	pos[index].setY(btScalar(y));
-	pos[index].setZ(btScalar(z));
+		//for(int j=0;j<nattrb;++j)
+		//	sn>>a;
+
+		//if(hasbounds)
+		//	sn>>bound;
+
+		pos[index].setX(btScalar(x));
+		pos[index].setY(btScalar(y));
+		pos[index].setZ(btScalar(z));
 	}
-btSoftBody*						psb=new btSoftBody(&worldInfo,nnode,&pos[0],0);
+	btSoftBody* psb = new btSoftBody(&worldInfo, nnode, &pos[0], 0);
 #if 0
 if(face&&face[0])
 	{
@@ -1178,42 +1189,41 @@ if(face&&face[0])
 	}
 #endif
 
-if(ele&&ele[0])
+	if (ele && ele[0])
 	{
-	int								ntetra=0;
-	int								ncorner=0;
-	int								neattrb=0;
-	sscanf(ele,"%d %d %d",&ntetra,&ncorner,&neattrb);
-	ele += nextLine(ele);
-	
-	//se>>ntetra;se>>ncorner;se>>neattrb;
-	for(int i=0;i<ntetra;++i)
-		{
-		int			index=0;
-		int			ni[4];
+		int ntetra = 0;
+		int ncorner = 0;
+		int neattrb = 0;
+		sscanf(ele, "%d %d %d", &ntetra, &ncorner, &neattrb);
+		ele += nextLine(ele);
 
-		//se>>index;
-		//se>>ni[0];se>>ni[1];se>>ni[2];se>>ni[3];
-		sscanf(ele,"%d %d %d %d %d",&index,&ni[0],&ni[1],&ni[2],&ni[3]);
-		ele+=nextLine(ele);
-		//for(int j=0;j<neattrb;++j) 
-		//	se>>a;
-		psb->appendTetra(ni[0],ni[1],ni[2],ni[3]);
-		if(btetralinks)
+		//se>>ntetra;se>>ncorner;se>>neattrb;
+		for (int i = 0; i < ntetra; ++i)
+		{
+			int index = 0;
+			int ni[4];
+
+			//se>>index;
+			//se>>ni[0];se>>ni[1];se>>ni[2];se>>ni[3];
+			sscanf(ele, "%d %d %d %d %d", &index, &ni[0], &ni[1], &ni[2], &ni[3]);
+			ele += nextLine(ele);
+			//for(int j=0;j<neattrb;++j)
+			//	se>>a;
+			psb->appendTetra(ni[0], ni[1], ni[2], ni[3]);
+			if (btetralinks)
 			{
-			psb->appendLink(ni[0],ni[1],0,true);
-			psb->appendLink(ni[1],ni[2],0,true);
-			psb->appendLink(ni[2],ni[0],0,true);
-			psb->appendLink(ni[0],ni[3],0,true);
-			psb->appendLink(ni[1],ni[3],0,true);
-			psb->appendLink(ni[2],ni[3],0,true);
+				psb->appendLink(ni[0], ni[1], 0, true);
+				psb->appendLink(ni[1], ni[2], 0, true);
+				psb->appendLink(ni[2], ni[0], 0, true);
+				psb->appendLink(ni[0], ni[3], 0, true);
+				psb->appendLink(ni[1], ni[3], 0, true);
+				psb->appendLink(ni[2], ni[3], 0, true);
 			}
 		}
 	}
-printf("Nodes:  %u\r\n",psb->m_nodes.size());
-printf("Links:  %u\r\n",psb->m_links.size());
-printf("Faces:  %u\r\n",psb->m_faces.size());
-printf("Tetras: %u\r\n",psb->m_tetras.size());
-return(psb);
+	printf("Nodes:  %u\r\n", psb->m_nodes.size());
+	printf("Links:  %u\r\n", psb->m_links.size());
+	printf("Faces:  %u\r\n", psb->m_faces.size());
+	printf("Tetras: %u\r\n", psb->m_tetras.size());
+	return (psb);
 }
-
diff --git a/thirdparty/bullet/BulletSoftBody/btSoftBodyHelpers.h b/thirdparty/bullet/BulletSoftBody/btSoftBodyHelpers.h
index 7271530109a..e433558c182 100644
--- a/thirdparty/bullet/BulletSoftBody/btSoftBodyHelpers.h
+++ b/thirdparty/bullet/BulletSoftBody/btSoftBodyHelpers.h
@@ -22,127 +22,130 @@ subject to the following restrictions:
 // Helpers
 //
 
-/* fDrawFlags															*/ 
-struct	fDrawFlags { enum _ {
-	Nodes		=	0x0001,
-	Links		=	0x0002,
-	Faces		=	0x0004,
-	Tetras		=	0x0008,
-	Normals		=	0x0010,
-	Contacts	=	0x0020,
-	Anchors		=	0x0040,
-	Notes		=	0x0080,
-	Clusters	=	0x0100,
-	NodeTree	=	0x0200,
-	FaceTree	=	0x0400,
-	ClusterTree	=	0x0800,
-	Joints		=	0x1000,
-	/* presets	*/ 
-	Std			=	Links+Faces+Tetras+Anchors+Notes+Joints,
-	StdTetra	=	Std-Faces+Tetras
-};};
-
-struct	btSoftBodyHelpers
+/* fDrawFlags															*/
+struct fDrawFlags
 {
-	/* Draw body															*/ 
-	static void				Draw(		btSoftBody* psb,
-		btIDebugDraw* idraw,
-		int drawflags=fDrawFlags::Std);
-	/* Draw body infos														*/ 
-	static	void			DrawInfos(	btSoftBody* psb,
-		btIDebugDraw* idraw,
-		bool masses,
-		bool areas,
-		bool stress);
-	/* Draw node tree														*/ 
-	static void				DrawNodeTree(	btSoftBody* psb,
-		btIDebugDraw* idraw,
-		int mindepth=0,
-		int maxdepth=-1);
-	/* Draw face tree														*/ 
-	static void				DrawFaceTree(	btSoftBody* psb,
-		btIDebugDraw* idraw,
-		int mindepth=0,
-		int maxdepth=-1);
-	/* Draw cluster tree													*/ 
-	static void				DrawClusterTree(btSoftBody* psb,
-		btIDebugDraw* idraw,
-		int mindepth=0,
-		int maxdepth=-1);
-	/* Draw rigid frame														*/ 
-	static	void			DrawFrame(		btSoftBody* psb,
-		btIDebugDraw* idraw);
-	/* Create a rope														*/ 
-	static	btSoftBody*		CreateRope( btSoftBodyWorldInfo& worldInfo,
-		const btVector3& from,
-		const btVector3& to,
-		int res,
-		int fixeds);
-	/* Create a patch														*/ 
-	static	btSoftBody*		CreatePatch(btSoftBodyWorldInfo& worldInfo,
-		const btVector3& corner00,
-		const btVector3& corner10,
-		const btVector3& corner01,
-		const btVector3& corner11,
-		int resx,
-		int resy,
-		int fixeds,
-		bool gendiags);
-	/* Create a patch with UV Texture Coordinates	*/ 
-	static	btSoftBody*		CreatePatchUV(btSoftBodyWorldInfo& worldInfo,
-		const btVector3& corner00,
-		const btVector3& corner10,
-		const btVector3& corner01,
-		const btVector3& corner11,
-		int resx,
-		int resy,
-		int fixeds,
-		bool gendiags,
-		float* tex_coords=0);
-	static	float	CalculateUV(int resx,int resy,int ix,int iy,int id);
-	/* Create an ellipsoid													*/ 
-	static	btSoftBody*		CreateEllipsoid(btSoftBodyWorldInfo& worldInfo,
-		const btVector3& center,
-		const btVector3& radius,
-		int res);	
-	/* Create from trimesh													*/ 
-	static	btSoftBody*		CreateFromTriMesh(	btSoftBodyWorldInfo& worldInfo,
-		const btScalar*	vertices,
-		const int* triangles,
-		int ntriangles,
-		bool randomizeConstraints = true);
-	/* Create from convex-hull												*/ 
-	static	btSoftBody*		CreateFromConvexHull(	btSoftBodyWorldInfo& worldInfo,
-		const btVector3* vertices,
-		int nvertices,
-		bool randomizeConstraints = true);
+	enum _
+	{
+		Nodes = 0x0001,
+		Links = 0x0002,
+		Faces = 0x0004,
+		Tetras = 0x0008,
+		Normals = 0x0010,
+		Contacts = 0x0020,
+		Anchors = 0x0040,
+		Notes = 0x0080,
+		Clusters = 0x0100,
+		NodeTree = 0x0200,
+		FaceTree = 0x0400,
+		ClusterTree = 0x0800,
+		Joints = 0x1000,
+		/* presets	*/
+		Std = Links + Faces + Tetras + Anchors + Notes + Joints,
+		StdTetra = Std - Faces + Tetras
+	};
+};
 
+struct btSoftBodyHelpers
+{
+	/* Draw body															*/
+	static void Draw(btSoftBody* psb,
+					 btIDebugDraw* idraw,
+					 int drawflags = fDrawFlags::Std);
+	/* Draw body infos														*/
+	static void DrawInfos(btSoftBody* psb,
+						  btIDebugDraw* idraw,
+						  bool masses,
+						  bool areas,
+						  bool stress);
+	/* Draw node tree														*/
+	static void DrawNodeTree(btSoftBody* psb,
+							 btIDebugDraw* idraw,
+							 int mindepth = 0,
+							 int maxdepth = -1);
+	/* Draw face tree														*/
+	static void DrawFaceTree(btSoftBody* psb,
+							 btIDebugDraw* idraw,
+							 int mindepth = 0,
+							 int maxdepth = -1);
+	/* Draw cluster tree													*/
+	static void DrawClusterTree(btSoftBody* psb,
+								btIDebugDraw* idraw,
+								int mindepth = 0,
+								int maxdepth = -1);
+	/* Draw rigid frame														*/
+	static void DrawFrame(btSoftBody* psb,
+						  btIDebugDraw* idraw);
+	/* Create a rope														*/
+	static btSoftBody* CreateRope(btSoftBodyWorldInfo& worldInfo,
+								  const btVector3& from,
+								  const btVector3& to,
+								  int res,
+								  int fixeds);
+	/* Create a patch														*/
+	static btSoftBody* CreatePatch(btSoftBodyWorldInfo& worldInfo,
+								   const btVector3& corner00,
+								   const btVector3& corner10,
+								   const btVector3& corner01,
+								   const btVector3& corner11,
+								   int resx,
+								   int resy,
+								   int fixeds,
+								   bool gendiags);
+	/* Create a patch with UV Texture Coordinates	*/
+	static btSoftBody* CreatePatchUV(btSoftBodyWorldInfo& worldInfo,
+									 const btVector3& corner00,
+									 const btVector3& corner10,
+									 const btVector3& corner01,
+									 const btVector3& corner11,
+									 int resx,
+									 int resy,
+									 int fixeds,
+									 bool gendiags,
+									 float* tex_coords = 0);
+	static float CalculateUV(int resx, int resy, int ix, int iy, int id);
+	/* Create an ellipsoid													*/
+	static btSoftBody* CreateEllipsoid(btSoftBodyWorldInfo& worldInfo,
+									   const btVector3& center,
+									   const btVector3& radius,
+									   int res);
+	/* Create from trimesh													*/
+	static btSoftBody* CreateFromTriMesh(btSoftBodyWorldInfo& worldInfo,
+										 const btScalar* vertices,
+										 const int* triangles,
+										 int ntriangles,
+										 bool randomizeConstraints = true);
+	/* Create from convex-hull												*/
+	static btSoftBody* CreateFromConvexHull(btSoftBodyWorldInfo& worldInfo,
+											const btVector3* vertices,
+											int nvertices,
+											bool randomizeConstraints = true);
 
-	/* Export TetGen compatible .smesh file									*/ 
-//	static void				ExportAsSMeshFile(	btSoftBody* psb,
-//												const char* filename);	
-	/* Create from TetGen .ele, .face, .node files							*/ 
-//	static btSoftBody*		CreateFromTetGenFile(	btSoftBodyWorldInfo& worldInfo,
-//													const char* ele,
-//													const char* face,
-//													const char* node,
-//													bool bfacelinks,
-//													bool btetralinks,
-//													bool bfacesfromtetras);
-	/* Create from TetGen .ele, .face, .node data							*/ 
-	static btSoftBody*		CreateFromTetGenData(	btSoftBodyWorldInfo& worldInfo,
-													const char* ele,
-													const char* face,
-													const char* node,
-													bool bfacelinks,
-													bool btetralinks,
-													bool bfacesfromtetras);
+	/* Export TetGen compatible .smesh file									*/
+	//	static void				ExportAsSMeshFile(	btSoftBody* psb,
+	//												const char* filename);
+	/* Create from TetGen .ele, .face, .node files							*/
+	//	static btSoftBody*		CreateFromTetGenFile(	btSoftBodyWorldInfo& worldInfo,
+	//													const char* ele,
+	//													const char* face,
+	//													const char* node,
+	//													bool bfacelinks,
+	//													bool btetralinks,
+	//													bool bfacesfromtetras);
+	/* Create from TetGen .ele, .face, .node data							*/
+	static btSoftBody* CreateFromTetGenData(btSoftBodyWorldInfo& worldInfo,
+											const char* ele,
+											const char* face,
+											const char* node,
+											bool bfacelinks,
+											bool btetralinks,
+											bool bfacesfromtetras);
 
 	/// Sort the list of links to move link calculations that are dependent upon earlier
 	/// ones as far as possible away from the calculation of those values
 	/// This tends to make adjacent loop iterations not dependent upon one another,
 	/// so out-of-order processors can execute instructions from multiple iterations at once
-	static void ReoptimizeLinkOrder(btSoftBody *psb );
+	static void ReoptimizeLinkOrder(btSoftBody* psb);
 };
 
-#endif //BT_SOFT_BODY_HELPERS_H
+#endif  //BT_SOFT_BODY_HELPERS_H
diff --git a/thirdparty/bullet/BulletSoftBody/btSoftBodyInternals.h b/thirdparty/bullet/BulletSoftBody/btSoftBodyInternals.h
index 1ad82616ea0..7efe514f38d 100644
--- a/thirdparty/bullet/BulletSoftBody/btSoftBodyInternals.h
+++ b/thirdparty/bullet/BulletSoftBody/btSoftBodyInternals.h
@@ -19,29 +19,37 @@ subject to the following restrictions:
 
 #include "btSoftBody.h"
 
-
 #include "LinearMath/btQuickprof.h"
 #include "LinearMath/btPolarDecomposition.h"
 #include "BulletCollision/BroadphaseCollision/btBroadphaseInterface.h"
 #include "BulletCollision/CollisionDispatch/btCollisionDispatcher.h"
 #include "BulletCollision/CollisionShapes/btConvexInternalShape.h"
 #include "BulletCollision/NarrowPhaseCollision/btGjkEpa2.h"
-#include <string.h> //for memset
+#include <string.h>  //for memset
 //
 // btSymMatrix
 //
 template <typename T>
 struct btSymMatrix
 {
-	btSymMatrix() : dim(0)					{}
-	btSymMatrix(int n,const T& init=T())	{ resize(n,init); }
-	void					resize(int n,const T& init=T())			{ dim=n;store.resize((n*(n+1))/2,init); }
-	int						index(int c,int r) const				{ if(c>r) btSwap(c,r);btAssert(r<dim);return((r*(r+1))/2+c); }
-	T&						operator()(int c,int r)					{ return(store[index(c,r)]); }
-	const T&				operator()(int c,int r) const			{ return(store[index(c,r)]); }
-	btAlignedObjectArray<T>	store;
-	int						dim;
-};	
+	btSymMatrix() : dim(0) {}
+	btSymMatrix(int n, const T& init = T()) { resize(n, init); }
+	void resize(int n, const T& init = T())
+	{
+		dim = n;
+		store.resize((n * (n + 1)) / 2, init);
+	}
+	int index(int c, int r) const
+	{
+		if (c > r) btSwap(c, r);
+		btAssert(r < dim);
+		return ((r * (r + 1)) / 2 + c);
+	}
+	T& operator()(int c, int r) { return (store[index(c, r)]); }
+	const T& operator()(int c, int r) const { return (store[index(c, r)]); }
+	btAlignedObjectArray<T> store;
+	int dim;
+};
 
 //
 // btSoftBodyCollisionShape
@@ -49,67 +57,64 @@ struct btSymMatrix
 class btSoftBodyCollisionShape : public btConcaveShape
 {
 public:
-	btSoftBody*						m_body;
+	btSoftBody* m_body;
 
 	btSoftBodyCollisionShape(btSoftBody* backptr)
 	{
 		m_shapeType = SOFTBODY_SHAPE_PROXYTYPE;
-		m_body=backptr;
+		m_body = backptr;
 	}
 
 	virtual ~btSoftBodyCollisionShape()
 	{
-
 	}
 
-	void	processAllTriangles(btTriangleCallback* /*callback*/,const btVector3& /*aabbMin*/,const btVector3& /*aabbMax*/) const
+	void processAllTriangles(btTriangleCallback* /*callback*/, const btVector3& /*aabbMin*/, const btVector3& /*aabbMax*/) const
 	{
 		//not yet
 		btAssert(0);
 	}
 
 	///getAabb returns the axis aligned bounding box in the coordinate frame of the given transform t.
-	virtual void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const
+	virtual void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
 	{
 		/* t is usually identity, except when colliding against btCompoundShape. See Issue 512 */
-		const btVector3	mins=m_body->m_bounds[0];
-		const btVector3	maxs=m_body->m_bounds[1];
-		const btVector3	crns[]={t*btVector3(mins.x(),mins.y(),mins.z()),
-			t*btVector3(maxs.x(),mins.y(),mins.z()),
-			t*btVector3(maxs.x(),maxs.y(),mins.z()),
-			t*btVector3(mins.x(),maxs.y(),mins.z()),
-			t*btVector3(mins.x(),mins.y(),maxs.z()),
-			t*btVector3(maxs.x(),mins.y(),maxs.z()),
-			t*btVector3(maxs.x(),maxs.y(),maxs.z()),
-			t*btVector3(mins.x(),maxs.y(),maxs.z())};
-		aabbMin=aabbMax=crns[0];
-		for(int i=1;i<8;++i)
+		const btVector3 mins = m_body->m_bounds[0];
+		const btVector3 maxs = m_body->m_bounds[1];
+		const btVector3 crns[] = {t * btVector3(mins.x(), mins.y(), mins.z()),
+								  t * btVector3(maxs.x(), mins.y(), mins.z()),
+								  t * btVector3(maxs.x(), maxs.y(), mins.z()),
+								  t * btVector3(mins.x(), maxs.y(), mins.z()),
+								  t * btVector3(mins.x(), mins.y(), maxs.z()),
+								  t * btVector3(maxs.x(), mins.y(), maxs.z()),
+								  t * btVector3(maxs.x(), maxs.y(), maxs.z()),
+								  t * btVector3(mins.x(), maxs.y(), maxs.z())};
+		aabbMin = aabbMax = crns[0];
+		for (int i = 1; i < 8; ++i)
 		{
 			aabbMin.setMin(crns[i]);
 			aabbMax.setMax(crns[i]);
 		}
 	}
 
-
-	virtual void	setLocalScaling(const btVector3& /*scaling*/)
-	{		
+	virtual void setLocalScaling(const btVector3& /*scaling*/)
+	{
 		///na
 	}
 	virtual const btVector3& getLocalScaling() const
 	{
-		static const btVector3 dummy(1,1,1);
+		static const btVector3 dummy(1, 1, 1);
 		return dummy;
 	}
-	virtual void	calculateLocalInertia(btScalar /*mass*/,btVector3& /*inertia*/) const
+	virtual void calculateLocalInertia(btScalar /*mass*/, btVector3& /*inertia*/) const
 	{
 		///not yet
 		btAssert(0);
 	}
-	virtual const char*	getName()const
+	virtual const char* getName() const
 	{
 		return "SoftBody";
 	}
-
 };
 
 //
@@ -118,48 +123,53 @@ public:
 class btSoftClusterCollisionShape : public btConvexInternalShape
 {
 public:
-	const btSoftBody::Cluster*	m_cluster;
+	const btSoftBody::Cluster* m_cluster;
 
-	btSoftClusterCollisionShape (const btSoftBody::Cluster* cluster) : m_cluster(cluster) { setMargin(0); }
+	btSoftClusterCollisionShape(const btSoftBody::Cluster* cluster) : m_cluster(cluster) { setMargin(0); }
 
-
-	virtual btVector3	localGetSupportingVertex(const btVector3& vec) const
+	virtual btVector3 localGetSupportingVertex(const btVector3& vec) const
 	{
-		btSoftBody::Node* const *						n=&m_cluster->m_nodes[0];
-		btScalar										d=btDot(vec,n[0]->m_x);
-		int												j=0;
-		for(int i=1,ni=m_cluster->m_nodes.size();i<ni;++i)
+		btSoftBody::Node* const* n = &m_cluster->m_nodes[0];
+		btScalar d = btDot(vec, n[0]->m_x);
+		int j = 0;
+		for (int i = 1, ni = m_cluster->m_nodes.size(); i < ni; ++i)
 		{
-			const btScalar	k=btDot(vec,n[i]->m_x);
-			if(k>d) { d=k;j=i; }
+			const btScalar k = btDot(vec, n[i]->m_x);
+			if (k > d)
+			{
+				d = k;
+				j = i;
+			}
 		}
-		return(n[j]->m_x);
+		return (n[j]->m_x);
 	}
-	virtual btVector3	localGetSupportingVertexWithoutMargin(const btVector3& vec)const
+	virtual btVector3 localGetSupportingVertexWithoutMargin(const btVector3& vec) const
 	{
-		return(localGetSupportingVertex(vec));
+		return (localGetSupportingVertex(vec));
 	}
 	//notice that the vectors should be unit length
-	virtual void	batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors,btVector3* supportVerticesOut,int numVectors) const
-	{}
+	virtual void batchedUnitVectorGetSupportingVertexWithoutMargin(const btVector3* vectors, btVector3* supportVerticesOut, int numVectors) const
+	{
+	}
 
+	virtual void calculateLocalInertia(btScalar mass, btVector3& inertia) const
+	{
+	}
 
-	virtual void	calculateLocalInertia(btScalar mass,btVector3& inertia) const
-	{}
+	virtual void getAabb(const btTransform& t, btVector3& aabbMin, btVector3& aabbMax) const
+	{
+	}
 
-	virtual void getAabb(const btTransform& t,btVector3& aabbMin,btVector3& aabbMax) const
-	{}
-
-	virtual int	getShapeType() const { return SOFTBODY_SHAPE_PROXYTYPE; }
+	virtual int getShapeType() const { return SOFTBODY_SHAPE_PROXYTYPE; }
 
 	//debugging
-	virtual const char*	getName()const {return "SOFTCLUSTER";}
+	virtual const char* getName() const { return "SOFTCLUSTER"; }
 
-	virtual void	setMargin(btScalar margin)
+	virtual void setMargin(btScalar margin)
 	{
 		btConvexInternalShape::setMargin(margin);
 	}
-	virtual btScalar	getMargin() const
+	virtual btScalar getMargin() const
 	{
 		return btConvexInternalShape::getMargin();
 	}
@@ -171,229 +181,247 @@ public:
 
 //
 template <typename T>
-static inline void			ZeroInitialize(T& value)
+static inline void ZeroInitialize(T& value)
 {
-	memset(&value,0,sizeof(T));
+	memset(&value, 0, sizeof(T));
 }
 //
 template <typename T>
-static inline bool			CompLess(const T& a,const T& b)
-{ return(a<b); }
-//
-template <typename T>
-static inline bool			CompGreater(const T& a,const T& b)
-{ return(a>b); }
-//
-template <typename T>
-static inline T				Lerp(const T& a,const T& b,btScalar t)
-{ return(a+(b-a)*t); }
-//
-template <typename T>
-static inline T				InvLerp(const T& a,const T& b,btScalar t)
-{ return((b+a*t-b*t)/(a*b)); }
-//
-static inline btMatrix3x3	Lerp(	const btMatrix3x3& a,
-								 const btMatrix3x3& b,
-								 btScalar t)
+static inline bool CompLess(const T& a, const T& b)
 {
-	btMatrix3x3	r;
-	r[0]=Lerp(a[0],b[0],t);
-	r[1]=Lerp(a[1],b[1],t);
-	r[2]=Lerp(a[2],b[2],t);
-	return(r);
+	return (a < b);
 }
 //
-static inline btVector3		Clamp(const btVector3& v,btScalar maxlength)
+template <typename T>
+static inline bool CompGreater(const T& a, const T& b)
 {
-	const btScalar sql=v.length2();
-	if(sql>(maxlength*maxlength))
-		return((v*maxlength)/btSqrt(sql));
+	return (a > b);
+}
+//
+template <typename T>
+static inline T Lerp(const T& a, const T& b, btScalar t)
+{
+	return (a + (b - a) * t);
+}
+//
+template <typename T>
+static inline T InvLerp(const T& a, const T& b, btScalar t)
+{
+	return ((b + a * t - b * t) / (a * b));
+}
+//
+static inline btMatrix3x3 Lerp(const btMatrix3x3& a,
+							   const btMatrix3x3& b,
+							   btScalar t)
+{
+	btMatrix3x3 r;
+	r[0] = Lerp(a[0], b[0], t);
+	r[1] = Lerp(a[1], b[1], t);
+	r[2] = Lerp(a[2], b[2], t);
+	return (r);
+}
+//
+static inline btVector3 Clamp(const btVector3& v, btScalar maxlength)
+{
+	const btScalar sql = v.length2();
+	if (sql > (maxlength * maxlength))
+		return ((v * maxlength) / btSqrt(sql));
 	else
-		return(v);
+		return (v);
 }
 //
 template <typename T>
-static inline T				Clamp(const T& x,const T& l,const T& h)
-{ return(x<l?l:x>h?h:x); }
+static inline T Clamp(const T& x, const T& l, const T& h)
+{
+	return (x < l ? l : x > h ? h : x);
+}
 //
 template <typename T>
-static inline T				Sq(const T& x)
-{ return(x*x); }
+static inline T Sq(const T& x)
+{
+	return (x * x);
+}
 //
 template <typename T>
-static inline T				Cube(const T& x)
-{ return(x*x*x); }
+static inline T Cube(const T& x)
+{
+	return (x * x * x);
+}
 //
 template <typename T>
-static inline T				Sign(const T& x)
-{ return((T)(x<0?-1:+1)); }
+static inline T Sign(const T& x)
+{
+	return ((T)(x < 0 ? -1 : +1));
+}
 //
 template <typename T>
-static inline bool			SameSign(const T& x,const T& y)
-{ return((x*y)>0); }
-//
-static inline btScalar		ClusterMetric(const btVector3& x,const btVector3& y)
+static inline bool SameSign(const T& x, const T& y)
 {
-	const btVector3	d=x-y;
-	return(btFabs(d[0])+btFabs(d[1])+btFabs(d[2]));
+	return ((x * y) > 0);
 }
 //
-static inline btMatrix3x3	ScaleAlongAxis(const btVector3& a,btScalar s)
+static inline btScalar ClusterMetric(const btVector3& x, const btVector3& y)
 {
-	const btScalar	xx=a.x()*a.x();
-	const btScalar	yy=a.y()*a.y();
-	const btScalar	zz=a.z()*a.z();
-	const btScalar	xy=a.x()*a.y();
-	const btScalar	yz=a.y()*a.z();
-	const btScalar	zx=a.z()*a.x();
-	btMatrix3x3		m;
-	m[0]=btVector3(1-xx+xx*s,xy*s-xy,zx*s-zx);
-	m[1]=btVector3(xy*s-xy,1-yy+yy*s,yz*s-yz);
-	m[2]=btVector3(zx*s-zx,yz*s-yz,1-zz+zz*s);
-	return(m);
+	const btVector3 d = x - y;
+	return (btFabs(d[0]) + btFabs(d[1]) + btFabs(d[2]));
 }
 //
-static inline btMatrix3x3	Cross(const btVector3& v)
+static inline btMatrix3x3 ScaleAlongAxis(const btVector3& a, btScalar s)
 {
-	btMatrix3x3	m;
-	m[0]=btVector3(0,-v.z(),+v.y());
-	m[1]=btVector3(+v.z(),0,-v.x());
-	m[2]=btVector3(-v.y(),+v.x(),0);
-	return(m);
+	const btScalar xx = a.x() * a.x();
+	const btScalar yy = a.y() * a.y();
+	const btScalar zz = a.z() * a.z();
+	const btScalar xy = a.x() * a.y();
+	const btScalar yz = a.y() * a.z();
+	const btScalar zx = a.z() * a.x();
+	btMatrix3x3 m;
+	m[0] = btVector3(1 - xx + xx * s, xy * s - xy, zx * s - zx);
+	m[1] = btVector3(xy * s - xy, 1 - yy + yy * s, yz * s - yz);
+	m[2] = btVector3(zx * s - zx, yz * s - yz, 1 - zz + zz * s);
+	return (m);
 }
 //
-static inline btMatrix3x3	Diagonal(btScalar x)
+static inline btMatrix3x3 Cross(const btVector3& v)
 {
-	btMatrix3x3	m;
-	m[0]=btVector3(x,0,0);
-	m[1]=btVector3(0,x,0);
-	m[2]=btVector3(0,0,x);
-	return(m);
+	btMatrix3x3 m;
+	m[0] = btVector3(0, -v.z(), +v.y());
+	m[1] = btVector3(+v.z(), 0, -v.x());
+	m[2] = btVector3(-v.y(), +v.x(), 0);
+	return (m);
 }
 //
-static inline btMatrix3x3	Add(const btMatrix3x3& a,
-								const btMatrix3x3& b)
+static inline btMatrix3x3 Diagonal(btScalar x)
 {
-	btMatrix3x3	r;
-	for(int i=0;i<3;++i) r[i]=a[i]+b[i];
-	return(r);
+	btMatrix3x3 m;
+	m[0] = btVector3(x, 0, 0);
+	m[1] = btVector3(0, x, 0);
+	m[2] = btVector3(0, 0, x);
+	return (m);
 }
 //
-static inline btMatrix3x3	Sub(const btMatrix3x3& a,
-								const btMatrix3x3& b)
+static inline btMatrix3x3 Add(const btMatrix3x3& a,
+							  const btMatrix3x3& b)
 {
-	btMatrix3x3	r;
-	for(int i=0;i<3;++i) r[i]=a[i]-b[i];
-	return(r);
+	btMatrix3x3 r;
+	for (int i = 0; i < 3; ++i) r[i] = a[i] + b[i];
+	return (r);
 }
 //
-static inline btMatrix3x3	Mul(const btMatrix3x3& a,
-								btScalar b)
+static inline btMatrix3x3 Sub(const btMatrix3x3& a,
+							  const btMatrix3x3& b)
 {
-	btMatrix3x3	r;
-	for(int i=0;i<3;++i) r[i]=a[i]*b;
-	return(r);
+	btMatrix3x3 r;
+	for (int i = 0; i < 3; ++i) r[i] = a[i] - b[i];
+	return (r);
 }
 //
-static inline void			Orthogonalize(btMatrix3x3& m)
+static inline btMatrix3x3 Mul(const btMatrix3x3& a,
+							  btScalar b)
 {
-	m[2]=btCross(m[0],m[1]).normalized();
-	m[1]=btCross(m[2],m[0]).normalized();
-	m[0]=btCross(m[1],m[2]).normalized();
+	btMatrix3x3 r;
+	for (int i = 0; i < 3; ++i) r[i] = a[i] * b;
+	return (r);
 }
 //
-static inline btMatrix3x3	MassMatrix(btScalar im,const btMatrix3x3& iwi,const btVector3& r)
+static inline void Orthogonalize(btMatrix3x3& m)
 {
-	const btMatrix3x3	cr=Cross(r);
-	return(Sub(Diagonal(im),cr*iwi*cr));
+	m[2] = btCross(m[0], m[1]).normalized();
+	m[1] = btCross(m[2], m[0]).normalized();
+	m[0] = btCross(m[1], m[2]).normalized();
+}
+//
+static inline btMatrix3x3 MassMatrix(btScalar im, const btMatrix3x3& iwi, const btVector3& r)
+{
+	const btMatrix3x3 cr = Cross(r);
+	return (Sub(Diagonal(im), cr * iwi * cr));
 }
 
 //
-static inline btMatrix3x3	ImpulseMatrix(	btScalar dt,
-										  btScalar ima,
-										  btScalar imb,
-										  const btMatrix3x3& iwi,
-										  const btVector3& r)
+static inline btMatrix3x3 ImpulseMatrix(btScalar dt,
+										btScalar ima,
+										btScalar imb,
+										const btMatrix3x3& iwi,
+										const btVector3& r)
 {
-	return(Diagonal(1/dt)*Add(Diagonal(ima),MassMatrix(imb,iwi,r)).inverse());
+	return (Diagonal(1 / dt) * Add(Diagonal(ima), MassMatrix(imb, iwi, r)).inverse());
 }
 
 //
-static inline btMatrix3x3	ImpulseMatrix(	btScalar ima,const btMatrix3x3& iia,const btVector3& ra,
-										  btScalar imb,const btMatrix3x3& iib,const btVector3& rb)	
+static inline btMatrix3x3 ImpulseMatrix(btScalar ima, const btMatrix3x3& iia, const btVector3& ra,
+										btScalar imb, const btMatrix3x3& iib, const btVector3& rb)
 {
-	return(Add(MassMatrix(ima,iia,ra),MassMatrix(imb,iib,rb)).inverse());
+	return (Add(MassMatrix(ima, iia, ra), MassMatrix(imb, iib, rb)).inverse());
 }
 
 //
-static inline btMatrix3x3	AngularImpulseMatrix(	const btMatrix3x3& iia,
-												 const btMatrix3x3& iib)
+static inline btMatrix3x3 AngularImpulseMatrix(const btMatrix3x3& iia,
+											   const btMatrix3x3& iib)
 {
-	return(Add(iia,iib).inverse());
+	return (Add(iia, iib).inverse());
 }
 
 //
-static inline btVector3		ProjectOnAxis(	const btVector3& v,
-										  const btVector3& a)
+static inline btVector3 ProjectOnAxis(const btVector3& v,
+									  const btVector3& a)
 {
-	return(a*btDot(v,a));
+	return (a * btDot(v, a));
 }
 //
-static inline btVector3		ProjectOnPlane(	const btVector3& v,
-										   const btVector3& a)
+static inline btVector3 ProjectOnPlane(const btVector3& v,
+									   const btVector3& a)
 {
-	return(v-ProjectOnAxis(v,a));
+	return (v - ProjectOnAxis(v, a));
 }
 
 //
-static inline void			ProjectOrigin(	const btVector3& a,
-										  const btVector3& b,
-										  btVector3& prj,
-										  btScalar& sqd)
+static inline void ProjectOrigin(const btVector3& a,
+								 const btVector3& b,
+								 btVector3& prj,
+								 btScalar& sqd)
 {
-	const btVector3	d=b-a;
-	const btScalar	m2=d.length2();
-	if(m2>SIMD_EPSILON)
-	{	
-		const btScalar	t=Clamp<btScalar>(-btDot(a,d)/m2,0,1);
-		const btVector3	p=a+d*t;
-		const btScalar	l2=p.length2();
-		if(l2<sqd)
+	const btVector3 d = b - a;
+	const btScalar m2 = d.length2();
+	if (m2 > SIMD_EPSILON)
+	{
+		const btScalar t = Clamp<btScalar>(-btDot(a, d) / m2, 0, 1);
+		const btVector3 p = a + d * t;
+		const btScalar l2 = p.length2();
+		if (l2 < sqd)
 		{
-			prj=p;
-			sqd=l2;
+			prj = p;
+			sqd = l2;
 		}
 	}
 }
 //
-static inline void			ProjectOrigin(	const btVector3& a,
-										  const btVector3& b,
-										  const btVector3& c,
-										  btVector3& prj,
-										  btScalar& sqd)
+static inline void ProjectOrigin(const btVector3& a,
+								 const btVector3& b,
+								 const btVector3& c,
+								 btVector3& prj,
+								 btScalar& sqd)
 {
-	const btVector3&	q=btCross(b-a,c-a);
-	const btScalar		m2=q.length2();
-	if(m2>SIMD_EPSILON)
+	const btVector3& q = btCross(b - a, c - a);
+	const btScalar m2 = q.length2();
+	if (m2 > SIMD_EPSILON)
 	{
-		const btVector3	n=q/btSqrt(m2);
-		const btScalar	k=btDot(a,n);
-		const btScalar	k2=k*k;
-		if(k2<sqd)
+		const btVector3 n = q / btSqrt(m2);
+		const btScalar k = btDot(a, n);
+		const btScalar k2 = k * k;
+		if (k2 < sqd)
 		{
-			const btVector3	p=n*k;
-			if(	(btDot(btCross(a-p,b-p),q)>0)&&
-				(btDot(btCross(b-p,c-p),q)>0)&&
-				(btDot(btCross(c-p,a-p),q)>0))
-			{			
-				prj=p;
-				sqd=k2;
+			const btVector3 p = n * k;
+			if ((btDot(btCross(a - p, b - p), q) > 0) &&
+				(btDot(btCross(b - p, c - p), q) > 0) &&
+				(btDot(btCross(c - p, a - p), q) > 0))
+			{
+				prj = p;
+				sqd = k2;
 			}
 			else
 			{
-				ProjectOrigin(a,b,prj,sqd);
-				ProjectOrigin(b,c,prj,sqd);
-				ProjectOrigin(c,a,prj,sqd);
+				ProjectOrigin(a, b, prj, sqd);
+				ProjectOrigin(b, c, prj, sqd);
+				ProjectOrigin(c, a, prj, sqd);
 			}
 		}
 	}
@@ -401,155 +429,159 @@ static inline void			ProjectOrigin(	const btVector3& a,
 
 //
 template <typename T>
-static inline T				BaryEval(		const T& a,
-									 const T& b,
-									 const T& c,
-									 const btVector3& coord)
+static inline T BaryEval(const T& a,
+						 const T& b,
+						 const T& c,
+						 const btVector3& coord)
 {
-	return(a*coord.x()+b*coord.y()+c*coord.z());
+	return (a * coord.x() + b * coord.y() + c * coord.z());
 }
 //
-static inline btVector3		BaryCoord(	const btVector3& a,
-									  const btVector3& b,
-									  const btVector3& c,
-									  const btVector3& p)
+static inline btVector3 BaryCoord(const btVector3& a,
+								  const btVector3& b,
+								  const btVector3& c,
+								  const btVector3& p)
 {
-	const btScalar	w[]={	btCross(a-p,b-p).length(),
-		btCross(b-p,c-p).length(),
-		btCross(c-p,a-p).length()};
-	const btScalar	isum=1/(w[0]+w[1]+w[2]);
-	return(btVector3(w[1]*isum,w[2]*isum,w[0]*isum));
+	const btScalar w[] = {btCross(a - p, b - p).length(),
+						  btCross(b - p, c - p).length(),
+						  btCross(c - p, a - p).length()};
+	const btScalar isum = 1 / (w[0] + w[1] + w[2]);
+	return (btVector3(w[1] * isum, w[2] * isum, w[0] * isum));
 }
 
 //
-inline static btScalar				ImplicitSolve(	btSoftBody::ImplicitFn* fn,
-										  const btVector3& a,
-										  const btVector3& b,
-										  const btScalar accuracy,
-										  const int maxiterations=256)
+inline static btScalar ImplicitSolve(btSoftBody::ImplicitFn* fn,
+									 const btVector3& a,
+									 const btVector3& b,
+									 const btScalar accuracy,
+									 const int maxiterations = 256)
 {
-	btScalar	span[2]={0,1};
-	btScalar	values[2]={fn->Eval(a),fn->Eval(b)};
-	if(values[0]>values[1])
+	btScalar span[2] = {0, 1};
+	btScalar values[2] = {fn->Eval(a), fn->Eval(b)};
+	if (values[0] > values[1])
 	{
-		btSwap(span[0],span[1]);
-		btSwap(values[0],values[1]);
+		btSwap(span[0], span[1]);
+		btSwap(values[0], values[1]);
 	}
-	if(values[0]>-accuracy) return(-1);
-	if(values[1]<+accuracy) return(-1);
-	for(int i=0;i<maxiterations;++i)
+	if (values[0] > -accuracy) return (-1);
+	if (values[1] < +accuracy) return (-1);
+	for (int i = 0; i < maxiterations; ++i)
 	{
-		const btScalar	t=Lerp(span[0],span[1],values[0]/(values[0]-values[1]));
-		const btScalar	v=fn->Eval(Lerp(a,b,t));
-		if((t<=0)||(t>=1))		break;
-		if(btFabs(v)<accuracy)	return(t);
-		if(v<0)
-		{ span[0]=t;values[0]=v; }
-		else
-		{ span[1]=t;values[1]=v; }
-	}
-	return(-1);
-}
-
-inline static void					EvaluateMedium(	const btSoftBodyWorldInfo* wfi,
-										   const btVector3& x,
-										   btSoftBody::sMedium& medium)
-{
-	medium.m_velocity	=	btVector3(0,0,0);
-	medium.m_pressure	=	0;
-	medium.m_density	=	wfi->air_density;
-	if(wfi->water_density>0)
-	{
-		const btScalar	depth=-(btDot(x,wfi->water_normal)+wfi->water_offset);
-		if(depth>0)
+		const btScalar t = Lerp(span[0], span[1], values[0] / (values[0] - values[1]));
+		const btScalar v = fn->Eval(Lerp(a, b, t));
+		if ((t <= 0) || (t >= 1)) break;
+		if (btFabs(v) < accuracy) return (t);
+		if (v < 0)
 		{
-			medium.m_density	=	wfi->water_density;
-			medium.m_pressure	=	depth*wfi->water_density*wfi->m_gravity.length();
+			span[0] = t;
+			values[0] = v;
+		}
+		else
+		{
+			span[1] = t;
+			values[1] = v;
+		}
+	}
+	return (-1);
+}
+
+inline static void EvaluateMedium(const btSoftBodyWorldInfo* wfi,
+								  const btVector3& x,
+								  btSoftBody::sMedium& medium)
+{
+	medium.m_velocity = btVector3(0, 0, 0);
+	medium.m_pressure = 0;
+	medium.m_density = wfi->air_density;
+	if (wfi->water_density > 0)
+	{
+		const btScalar depth = -(btDot(x, wfi->water_normal) + wfi->water_offset);
+		if (depth > 0)
+		{
+			medium.m_density = wfi->water_density;
+			medium.m_pressure = depth * wfi->water_density * wfi->m_gravity.length();
 		}
 	}
 }
 
-
 //
-static inline btVector3		NormalizeAny(const btVector3& v)
+static inline btVector3 NormalizeAny(const btVector3& v)
 {
-	const btScalar l=v.length();
-	if(l>SIMD_EPSILON)
-		return(v/l);
+	const btScalar l = v.length();
+	if (l > SIMD_EPSILON)
+		return (v / l);
 	else
-		return(btVector3(0,0,0));
+		return (btVector3(0, 0, 0));
 }
 
 //
-static inline btDbvtVolume	VolumeOf(	const btSoftBody::Face& f,
-									 btScalar margin)
+static inline btDbvtVolume VolumeOf(const btSoftBody::Face& f,
+									btScalar margin)
 {
-	const btVector3*	pts[]={	&f.m_n[0]->m_x,
-		&f.m_n[1]->m_x,
-		&f.m_n[2]->m_x};
-	btDbvtVolume		vol=btDbvtVolume::FromPoints(pts,3);
-	vol.Expand(btVector3(margin,margin,margin));
-	return(vol);
+	const btVector3* pts[] = {&f.m_n[0]->m_x,
+							  &f.m_n[1]->m_x,
+							  &f.m_n[2]->m_x};
+	btDbvtVolume vol = btDbvtVolume::FromPoints(pts, 3);
+	vol.Expand(btVector3(margin, margin, margin));
+	return (vol);
 }
 
 //
-static inline btVector3			CenterOf(	const btSoftBody::Face& f)
+static inline btVector3 CenterOf(const btSoftBody::Face& f)
 {
-	return((f.m_n[0]->m_x+f.m_n[1]->m_x+f.m_n[2]->m_x)/3);
+	return ((f.m_n[0]->m_x + f.m_n[1]->m_x + f.m_n[2]->m_x) / 3);
 }
 
 //
-static inline btScalar			AreaOf(		const btVector3& x0,
-									   const btVector3& x1,
-									   const btVector3& x2)
+static inline btScalar AreaOf(const btVector3& x0,
+							  const btVector3& x1,
+							  const btVector3& x2)
 {
-	const btVector3	a=x1-x0;
-	const btVector3	b=x2-x0;
-	const btVector3	cr=btCross(a,b);
-	const btScalar	area=cr.length();
-	return(area);
+	const btVector3 a = x1 - x0;
+	const btVector3 b = x2 - x0;
+	const btVector3 cr = btCross(a, b);
+	const btScalar area = cr.length();
+	return (area);
 }
 
 //
-static inline btScalar		VolumeOf(	const btVector3& x0,
-									 const btVector3& x1,
-									 const btVector3& x2,
-									 const btVector3& x3)
+static inline btScalar VolumeOf(const btVector3& x0,
+								const btVector3& x1,
+								const btVector3& x2,
+								const btVector3& x3)
 {
-	const btVector3	a=x1-x0;
-	const btVector3	b=x2-x0;
-	const btVector3	c=x3-x0;
-	return(btDot(a,btCross(b,c)));
+	const btVector3 a = x1 - x0;
+	const btVector3 b = x2 - x0;
+	const btVector3 c = x3 - x0;
+	return (btDot(a, btCross(b, c)));
 }
 
 //
 
-
 //
-static inline void			ApplyClampedForce(	btSoftBody::Node& n,
-											  const btVector3& f,
-											  btScalar dt)
+static inline void ApplyClampedForce(btSoftBody::Node& n,
+									 const btVector3& f,
+									 btScalar dt)
 {
-	const btScalar	dtim=dt*n.m_im;
-	if((f*dtim).length2()>n.m_v.length2())
-	{/* Clamp	*/ 
-		n.m_f-=ProjectOnAxis(n.m_v,f.normalized())/dtim;						
+	const btScalar dtim = dt * n.m_im;
+	if ((f * dtim).length2() > n.m_v.length2())
+	{ /* Clamp	*/
+		n.m_f -= ProjectOnAxis(n.m_v, f.normalized()) / dtim;
 	}
 	else
-	{/* Apply	*/ 
-		n.m_f+=f;
+	{ /* Apply	*/
+		n.m_f += f;
 	}
 }
 
 //
-static inline int		MatchEdge(	const btSoftBody::Node* a,
-								  const btSoftBody::Node* b,
-								  const btSoftBody::Node* ma,
-								  const btSoftBody::Node* mb)
+static inline int MatchEdge(const btSoftBody::Node* a,
+							const btSoftBody::Node* b,
+							const btSoftBody::Node* ma,
+							const btSoftBody::Node* mb)
 {
-	if((a==ma)&&(b==mb)) return(0);
-	if((a==mb)&&(b==ma)) return(1);
-	return(-1);
+	if ((a == ma) && (b == mb)) return (0);
+	if ((a == mb) && (b == ma)) return (1);
+	return (-1);
 }
 
 //
@@ -557,58 +589,72 @@ static inline int		MatchEdge(	const btSoftBody::Node* a,
 // straitforward implementation of http://math.fullerton.edu/mathews/n2003/JacobiMethodMod.html
 // outputs are NOT sorted.
 //
-struct	btEigen
+struct btEigen
 {
-	static int			system(btMatrix3x3& a,btMatrix3x3* vectors,btVector3* values=0)
+	static int system(btMatrix3x3& a, btMatrix3x3* vectors, btVector3* values = 0)
 	{
-		static const int		maxiterations=16;
-		static const btScalar	accuracy=(btScalar)0.0001;
-		btMatrix3x3&			v=*vectors;
-		int						iterations=0;
+		static const int maxiterations = 16;
+		static const btScalar accuracy = (btScalar)0.0001;
+		btMatrix3x3& v = *vectors;
+		int iterations = 0;
 		vectors->setIdentity();
-		do	{
-			int				p=0,q=1;
-			if(btFabs(a[p][q])<btFabs(a[0][2])) { p=0;q=2; }
-			if(btFabs(a[p][q])<btFabs(a[1][2])) { p=1;q=2; }
-			if(btFabs(a[p][q])>accuracy)
-			{
-				const btScalar	w=(a[q][q]-a[p][p])/(2*a[p][q]);
-				const btScalar	z=btFabs(w);
-				const btScalar	t=w/(z*(btSqrt(1+w*w)+z));
-				if(t==t)/* [WARNING] let hope that one does not get thrown aways by some compilers... */ 
-				{
-					const btScalar	c=1/btSqrt(t*t+1);
-					const btScalar	s=c*t;
-					mulPQ(a,c,s,p,q);
-					mulTPQ(a,c,s,p,q);
-					mulPQ(v,c,s,p,q);
-				} else break;
-			} else break;
-		} while((++iterations)<maxiterations);
-		if(values)
+		do
 		{
-			*values=btVector3(a[0][0],a[1][1],a[2][2]);
+			int p = 0, q = 1;
+			if (btFabs(a[p][q]) < btFabs(a[0][2]))
+			{
+				p = 0;
+				q = 2;
+			}
+			if (btFabs(a[p][q]) < btFabs(a[1][2]))
+			{
+				p = 1;
+				q = 2;
+			}
+			if (btFabs(a[p][q]) > accuracy)
+			{
+				const btScalar w = (a[q][q] - a[p][p]) / (2 * a[p][q]);
+				const btScalar z = btFabs(w);
+				const btScalar t = w / (z * (btSqrt(1 + w * w) + z));
+				if (t == t) /* [WARNING] let hope that one does not get thrown aways by some compilers... */
+				{
+					const btScalar c = 1 / btSqrt(t * t + 1);
+					const btScalar s = c * t;
+					mulPQ(a, c, s, p, q);
+					mulTPQ(a, c, s, p, q);
+					mulPQ(v, c, s, p, q);
+				}
+				else
+					break;
+			}
+			else
+				break;
+		} while ((++iterations) < maxiterations);
+		if (values)
+		{
+			*values = btVector3(a[0][0], a[1][1], a[2][2]);
 		}
-		return(iterations);
+		return (iterations);
 	}
+
 private:
-	static inline void	mulTPQ(btMatrix3x3& a,btScalar c,btScalar s,int p,int q)
+	static inline void mulTPQ(btMatrix3x3& a, btScalar c, btScalar s, int p, int q)
 	{
-		const btScalar	m[2][3]={	{a[p][0],a[p][1],a[p][2]},
-		{a[q][0],a[q][1],a[q][2]}};
+		const btScalar m[2][3] = {{a[p][0], a[p][1], a[p][2]},
+								  {a[q][0], a[q][1], a[q][2]}};
 		int i;
 
-		for(i=0;i<3;++i) a[p][i]=c*m[0][i]-s*m[1][i];
-		for(i=0;i<3;++i) a[q][i]=c*m[1][i]+s*m[0][i];
+		for (i = 0; i < 3; ++i) a[p][i] = c * m[0][i] - s * m[1][i];
+		for (i = 0; i < 3; ++i) a[q][i] = c * m[1][i] + s * m[0][i];
 	}
-	static inline void	mulPQ(btMatrix3x3& a,btScalar c,btScalar s,int p,int q)
+	static inline void mulPQ(btMatrix3x3& a, btScalar c, btScalar s, int p, int q)
 	{
-		const btScalar	m[2][3]={	{a[0][p],a[1][p],a[2][p]},
-		{a[0][q],a[1][q],a[2][q]}};
+		const btScalar m[2][3] = {{a[0][p], a[1][p], a[2][p]},
+								  {a[0][q], a[1][q], a[2][q]}};
 		int i;
 
-		for(i=0;i<3;++i) a[i][p]=c*m[0][i]-s*m[1][i];
-		for(i=0;i<3;++i) a[i][q]=c*m[1][i]+s*m[0][i];
+		for (i = 0; i < 3; ++i) a[i][p] = c * m[0][i] - s * m[1][i];
+		for (i = 0; i < 3; ++i) a[i][q] = c * m[1][i] + s * m[0][i];
 	}
 };
 
@@ -616,9 +662,9 @@ private:
 // Polar decomposition,
 // "Computing the Polar Decomposition with Applications", Nicholas J. Higham, 1986.
 //
-static inline int			PolarDecompose(	const btMatrix3x3& m,btMatrix3x3& q,btMatrix3x3& s)
+static inline int PolarDecompose(const btMatrix3x3& m, btMatrix3x3& q, btMatrix3x3& s)
 {
-	static const btPolarDecomposition polar;  
+	static const btPolarDecomposition polar;
 	return polar.decompose(m, q, s);
 }
 
@@ -630,282 +676,284 @@ struct btSoftColliders
 	//
 	// ClusterBase
 	//
-	struct	ClusterBase : btDbvt::ICollide
+	struct ClusterBase : btDbvt::ICollide
 	{
-		btScalar			erp;
-		btScalar			idt;
-		btScalar			m_margin;
-		btScalar			friction;
-		btScalar			threshold;
+		btScalar erp;
+		btScalar idt;
+		btScalar m_margin;
+		btScalar friction;
+		btScalar threshold;
 		ClusterBase()
 		{
-			erp			=(btScalar)1;
-			idt			=0;
-			m_margin		=0;
-			friction	=0;
-			threshold	=(btScalar)0;
+			erp = (btScalar)1;
+			idt = 0;
+			m_margin = 0;
+			friction = 0;
+			threshold = (btScalar)0;
 		}
-		bool				SolveContact(	const btGjkEpaSolver2::sResults& res,
-			btSoftBody::Body ba,const btSoftBody::Body bb,
-			btSoftBody::CJoint& joint)
+		bool SolveContact(const btGjkEpaSolver2::sResults& res,
+						  btSoftBody::Body ba, const btSoftBody::Body bb,
+						  btSoftBody::CJoint& joint)
 		{
-			if(res.distance<m_margin)
+			if (res.distance < m_margin)
 			{
 				btVector3 norm = res.normal;
-				norm.normalize();//is it necessary?
+				norm.normalize();  //is it necessary?
 
-				const btVector3		ra=res.witnesses[0]-ba.xform().getOrigin();
-				const btVector3		rb=res.witnesses[1]-bb.xform().getOrigin();
-				const btVector3		va=ba.velocity(ra);
-				const btVector3		vb=bb.velocity(rb);
-				const btVector3		vrel=va-vb;
-				const btScalar		rvac=btDot(vrel,norm);
-				 btScalar		depth=res.distance-m_margin;
-				
-//				printf("depth=%f\n",depth);
-				const btVector3		iv=norm*rvac;
-				const btVector3		fv=vrel-iv;
-				joint.m_bodies[0]	=	ba;
-				joint.m_bodies[1]	=	bb;
-				joint.m_refs[0]		=	ra*ba.xform().getBasis();
-				joint.m_refs[1]		=	rb*bb.xform().getBasis();
-				joint.m_rpos[0]		=	ra;
-				joint.m_rpos[1]		=	rb;
-				joint.m_cfm			=	1;
-				joint.m_erp			=	1;
-				joint.m_life		=	0;
-				joint.m_maxlife		=	0;
-				joint.m_split		=	1;
-				
-				joint.m_drift		=	depth*norm;
+				const btVector3 ra = res.witnesses[0] - ba.xform().getOrigin();
+				const btVector3 rb = res.witnesses[1] - bb.xform().getOrigin();
+				const btVector3 va = ba.velocity(ra);
+				const btVector3 vb = bb.velocity(rb);
+				const btVector3 vrel = va - vb;
+				const btScalar rvac = btDot(vrel, norm);
+				btScalar depth = res.distance - m_margin;
 
-				joint.m_normal		=	norm;
-//				printf("normal=%f,%f,%f\n",res.normal.getX(),res.normal.getY(),res.normal.getZ());
-				joint.m_delete		=	false;
-				joint.m_friction	=	fv.length2()<(rvac*friction*rvac*friction)?1:friction;
-				joint.m_massmatrix	=	ImpulseMatrix(	ba.invMass(),ba.invWorldInertia(),joint.m_rpos[0],
-					bb.invMass(),bb.invWorldInertia(),joint.m_rpos[1]);
+				//				printf("depth=%f\n",depth);
+				const btVector3 iv = norm * rvac;
+				const btVector3 fv = vrel - iv;
+				joint.m_bodies[0] = ba;
+				joint.m_bodies[1] = bb;
+				joint.m_refs[0] = ra * ba.xform().getBasis();
+				joint.m_refs[1] = rb * bb.xform().getBasis();
+				joint.m_rpos[0] = ra;
+				joint.m_rpos[1] = rb;
+				joint.m_cfm = 1;
+				joint.m_erp = 1;
+				joint.m_life = 0;
+				joint.m_maxlife = 0;
+				joint.m_split = 1;
 
-				return(true);
+				joint.m_drift = depth * norm;
+
+				joint.m_normal = norm;
+				//				printf("normal=%f,%f,%f\n",res.normal.getX(),res.normal.getY(),res.normal.getZ());
+				joint.m_delete = false;
+				joint.m_friction = fv.length2() < (rvac * friction * rvac * friction) ? 1 : friction;
+				joint.m_massmatrix = ImpulseMatrix(ba.invMass(), ba.invWorldInertia(), joint.m_rpos[0],
+												   bb.invMass(), bb.invWorldInertia(), joint.m_rpos[1]);
+
+				return (true);
 			}
-			return(false);
+			return (false);
 		}
 	};
 	//
 	// CollideCL_RS
 	//
-	struct	CollideCL_RS : ClusterBase
+	struct CollideCL_RS : ClusterBase
 	{
-		btSoftBody*		psb;
-		const btCollisionObjectWrapper*	m_colObjWrap;
+		btSoftBody* psb;
+		const btCollisionObjectWrapper* m_colObjWrap;
 
-		void		Process(const btDbvtNode* leaf)
+		void Process(const btDbvtNode* leaf)
 		{
-			btSoftBody::Cluster*		cluster=(btSoftBody::Cluster*)leaf->data;
-			btSoftClusterCollisionShape	cshape(cluster);
-			
-			const btConvexShape*		rshape=(const btConvexShape*)m_colObjWrap->getCollisionShape();
+			btSoftBody::Cluster* cluster = (btSoftBody::Cluster*)leaf->data;
+			btSoftClusterCollisionShape cshape(cluster);
+
+			const btConvexShape* rshape = (const btConvexShape*)m_colObjWrap->getCollisionShape();
 
 			///don't collide an anchored cluster with a static/kinematic object
-			if(m_colObjWrap->getCollisionObject()->isStaticOrKinematicObject() && cluster->m_containsAnchor)
+			if (m_colObjWrap->getCollisionObject()->isStaticOrKinematicObject() && cluster->m_containsAnchor)
 				return;
 
-			btGjkEpaSolver2::sResults	res;		
-			if(btGjkEpaSolver2::SignedDistance(	&cshape,btTransform::getIdentity(),
-				rshape,m_colObjWrap->getWorldTransform(),
-				btVector3(1,0,0),res))
+			btGjkEpaSolver2::sResults res;
+			if (btGjkEpaSolver2::SignedDistance(&cshape, btTransform::getIdentity(),
+												rshape, m_colObjWrap->getWorldTransform(),
+												btVector3(1, 0, 0), res))
 			{
-				btSoftBody::CJoint	joint;
-				if(SolveContact(res,cluster,m_colObjWrap->getCollisionObject(),joint))//prb,joint))
+				btSoftBody::CJoint joint;
+				if (SolveContact(res, cluster, m_colObjWrap->getCollisionObject(), joint))  //prb,joint))
 				{
-					btSoftBody::CJoint*	pj=new(btAlignedAlloc(sizeof(btSoftBody::CJoint),16)) btSoftBody::CJoint();
-					*pj=joint;psb->m_joints.push_back(pj);
-					if(m_colObjWrap->getCollisionObject()->isStaticOrKinematicObject())
+					btSoftBody::CJoint* pj = new (btAlignedAlloc(sizeof(btSoftBody::CJoint), 16)) btSoftBody::CJoint();
+					*pj = joint;
+					psb->m_joints.push_back(pj);
+					if (m_colObjWrap->getCollisionObject()->isStaticOrKinematicObject())
 					{
-						pj->m_erp	*=	psb->m_cfg.kSKHR_CL;
-						pj->m_split	*=	psb->m_cfg.kSK_SPLT_CL;
+						pj->m_erp *= psb->m_cfg.kSKHR_CL;
+						pj->m_split *= psb->m_cfg.kSK_SPLT_CL;
 					}
 					else
 					{
-						pj->m_erp	*=	psb->m_cfg.kSRHR_CL;
-						pj->m_split	*=	psb->m_cfg.kSR_SPLT_CL;
+						pj->m_erp *= psb->m_cfg.kSRHR_CL;
+						pj->m_split *= psb->m_cfg.kSR_SPLT_CL;
 					}
 				}
 			}
 		}
-		void		ProcessColObj(btSoftBody* ps,const btCollisionObjectWrapper* colObWrap)
+		void ProcessColObj(btSoftBody* ps, const btCollisionObjectWrapper* colObWrap)
 		{
-			psb			=	ps;
-			m_colObjWrap			=	colObWrap;
-			idt			=	ps->m_sst.isdt;
-			m_margin		=	m_colObjWrap->getCollisionShape()->getMargin()+psb->getCollisionShape()->getMargin();
+			psb = ps;
+			m_colObjWrap = colObWrap;
+			idt = ps->m_sst.isdt;
+			m_margin = m_colObjWrap->getCollisionShape()->getMargin() + psb->getCollisionShape()->getMargin();
 			///Bullet rigid body uses multiply instead of minimum to determine combined friction. Some customization would be useful.
-			friction	=	btMin(psb->m_cfg.kDF,m_colObjWrap->getCollisionObject()->getFriction());
-			btVector3			mins;
-			btVector3			maxs;
+			friction = btMin(psb->m_cfg.kDF, m_colObjWrap->getCollisionObject()->getFriction());
+			btVector3 mins;
+			btVector3 maxs;
 
-			ATTRIBUTE_ALIGNED16(btDbvtVolume)		volume;
-			colObWrap->getCollisionShape()->getAabb(colObWrap->getWorldTransform(),mins,maxs);
-			volume=btDbvtVolume::FromMM(mins,maxs);
-			volume.Expand(btVector3(1,1,1)*m_margin);
-			ps->m_cdbvt.collideTV(ps->m_cdbvt.m_root,volume,*this);
-		}	
+			ATTRIBUTE_ALIGNED16(btDbvtVolume)
+			volume;
+			colObWrap->getCollisionShape()->getAabb(colObWrap->getWorldTransform(), mins, maxs);
+			volume = btDbvtVolume::FromMM(mins, maxs);
+			volume.Expand(btVector3(1, 1, 1) * m_margin);
+			ps->m_cdbvt.collideTV(ps->m_cdbvt.m_root, volume, *this);
+		}
 	};
 	//
 	// CollideCL_SS
 	//
-	struct	CollideCL_SS : ClusterBase
+	struct CollideCL_SS : ClusterBase
 	{
-		btSoftBody*	bodies[2];
-		void		Process(const btDbvtNode* la,const btDbvtNode* lb)
+		btSoftBody* bodies[2];
+		void Process(const btDbvtNode* la, const btDbvtNode* lb)
 		{
-			btSoftBody::Cluster*		cla=(btSoftBody::Cluster*)la->data;
-			btSoftBody::Cluster*		clb=(btSoftBody::Cluster*)lb->data;
+			btSoftBody::Cluster* cla = (btSoftBody::Cluster*)la->data;
+			btSoftBody::Cluster* clb = (btSoftBody::Cluster*)lb->data;
 
-
-			bool connected=false;
-			if ((bodies[0]==bodies[1])&&(bodies[0]->m_clusterConnectivity.size()))
+			bool connected = false;
+			if ((bodies[0] == bodies[1]) && (bodies[0]->m_clusterConnectivity.size()))
 			{
-				connected = bodies[0]->m_clusterConnectivity[cla->m_clusterIndex+bodies[0]->m_clusters.size()*clb->m_clusterIndex];
+				connected = bodies[0]->m_clusterConnectivity[cla->m_clusterIndex + bodies[0]->m_clusters.size() * clb->m_clusterIndex];
 			}
 
 			if (!connected)
 			{
-				btSoftClusterCollisionShape	csa(cla);
-				btSoftClusterCollisionShape	csb(clb);
-				btGjkEpaSolver2::sResults	res;		
-				if(btGjkEpaSolver2::SignedDistance(	&csa,btTransform::getIdentity(),
-					&csb,btTransform::getIdentity(),
-					cla->m_com-clb->m_com,res))
+				btSoftClusterCollisionShape csa(cla);
+				btSoftClusterCollisionShape csb(clb);
+				btGjkEpaSolver2::sResults res;
+				if (btGjkEpaSolver2::SignedDistance(&csa, btTransform::getIdentity(),
+													&csb, btTransform::getIdentity(),
+													cla->m_com - clb->m_com, res))
 				{
-					btSoftBody::CJoint	joint;
-					if(SolveContact(res,cla,clb,joint))
+					btSoftBody::CJoint joint;
+					if (SolveContact(res, cla, clb, joint))
 					{
-						btSoftBody::CJoint*	pj=new(btAlignedAlloc(sizeof(btSoftBody::CJoint),16)) btSoftBody::CJoint();
-						*pj=joint;bodies[0]->m_joints.push_back(pj);
-						pj->m_erp	*=	btMax(bodies[0]->m_cfg.kSSHR_CL,bodies[1]->m_cfg.kSSHR_CL);
-						pj->m_split	*=	(bodies[0]->m_cfg.kSS_SPLT_CL+bodies[1]->m_cfg.kSS_SPLT_CL)/2;
+						btSoftBody::CJoint* pj = new (btAlignedAlloc(sizeof(btSoftBody::CJoint), 16)) btSoftBody::CJoint();
+						*pj = joint;
+						bodies[0]->m_joints.push_back(pj);
+						pj->m_erp *= btMax(bodies[0]->m_cfg.kSSHR_CL, bodies[1]->m_cfg.kSSHR_CL);
+						pj->m_split *= (bodies[0]->m_cfg.kSS_SPLT_CL + bodies[1]->m_cfg.kSS_SPLT_CL) / 2;
 					}
 				}
-			} else
+			}
+			else
 			{
-				static int count=0;
+				static int count = 0;
 				count++;
 				//printf("count=%d\n",count);
-				
 			}
 		}
-		void		ProcessSoftSoft(btSoftBody* psa,btSoftBody* psb)
+		void ProcessSoftSoft(btSoftBody* psa, btSoftBody* psb)
 		{
-			idt			=	psa->m_sst.isdt;
+			idt = psa->m_sst.isdt;
 			//m_margin		=	(psa->getCollisionShape()->getMargin()+psb->getCollisionShape()->getMargin())/2;
-			m_margin		=	(psa->getCollisionShape()->getMargin()+psb->getCollisionShape()->getMargin());
-			friction	=	btMin(psa->m_cfg.kDF,psb->m_cfg.kDF);
-			bodies[0]	=	psa;
-			bodies[1]	=	psb;
-			psa->m_cdbvt.collideTT(psa->m_cdbvt.m_root,psb->m_cdbvt.m_root,*this);
-		}	
+			m_margin = (psa->getCollisionShape()->getMargin() + psb->getCollisionShape()->getMargin());
+			friction = btMin(psa->m_cfg.kDF, psb->m_cfg.kDF);
+			bodies[0] = psa;
+			bodies[1] = psb;
+			psa->m_cdbvt.collideTT(psa->m_cdbvt.m_root, psb->m_cdbvt.m_root, *this);
+		}
 	};
 	//
 	// CollideSDF_RS
 	//
-	struct	CollideSDF_RS : btDbvt::ICollide
+	struct CollideSDF_RS : btDbvt::ICollide
 	{
-		void		Process(const btDbvtNode* leaf)
+		void Process(const btDbvtNode* leaf)
 		{
-			btSoftBody::Node*	node=(btSoftBody::Node*)leaf->data;
+			btSoftBody::Node* node = (btSoftBody::Node*)leaf->data;
 			DoNode(*node);
 		}
-		void		DoNode(btSoftBody::Node& n) const
+		void DoNode(btSoftBody::Node& n) const
 		{
-			const btScalar			m=n.m_im>0?dynmargin:stamargin;
-			btSoftBody::RContact	c;
+			const btScalar m = n.m_im > 0 ? dynmargin : stamargin;
+			btSoftBody::RContact c;
 
-			if(	(!n.m_battach)&&
-				psb->checkContact(m_colObj1Wrap,n.m_x,m,c.m_cti))
+			if ((!n.m_battach) &&
+				psb->checkContact(m_colObj1Wrap, n.m_x, m, c.m_cti))
 			{
-				const btScalar	ima=n.m_im;
-				const btScalar	imb= m_rigidBody? m_rigidBody->getInvMass() : 0.f;
-				const btScalar	ms=ima+imb;
-				if(ms>0)
+				const btScalar ima = n.m_im;
+				const btScalar imb = m_rigidBody ? m_rigidBody->getInvMass() : 0.f;
+				const btScalar ms = ima + imb;
+				if (ms > 0)
 				{
-					const btTransform&	wtr=m_rigidBody?m_rigidBody->getWorldTransform() : m_colObj1Wrap->getCollisionObject()->getWorldTransform();
-					static const btMatrix3x3	iwiStatic(0,0,0,0,0,0,0,0,0);
-					const btMatrix3x3&	iwi=m_rigidBody?m_rigidBody->getInvInertiaTensorWorld() : iwiStatic;
-					const btVector3		ra=n.m_x-wtr.getOrigin();
-					const btVector3		va=m_rigidBody ? m_rigidBody->getVelocityInLocalPoint(ra)*psb->m_sst.sdt : btVector3(0,0,0);
-					const btVector3		vb=n.m_x-n.m_q;	
-					const btVector3		vr=vb-va;
-					const btScalar		dn=btDot(vr,c.m_cti.m_normal);
-					const btVector3		fv=vr-c.m_cti.m_normal*dn;
-					const btScalar		fc=psb->m_cfg.kDF*m_colObj1Wrap->getCollisionObject()->getFriction();
-					c.m_node	=	&n;
-					c.m_c0		=	ImpulseMatrix(psb->m_sst.sdt,ima,imb,iwi,ra);
-					c.m_c1		=	ra;
-					c.m_c2		=	ima*psb->m_sst.sdt;
-			        c.m_c3		=	fv.length2()<(dn*fc*dn*fc)?0:1-fc;
-					c.m_c4		=	m_colObj1Wrap->getCollisionObject()->isStaticOrKinematicObject()?psb->m_cfg.kKHR:psb->m_cfg.kCHR;
+					const btTransform& wtr = m_rigidBody ? m_rigidBody->getWorldTransform() : m_colObj1Wrap->getCollisionObject()->getWorldTransform();
+					static const btMatrix3x3 iwiStatic(0, 0, 0, 0, 0, 0, 0, 0, 0);
+					const btMatrix3x3& iwi = m_rigidBody ? m_rigidBody->getInvInertiaTensorWorld() : iwiStatic;
+					const btVector3 ra = n.m_x - wtr.getOrigin();
+					const btVector3 va = m_rigidBody ? m_rigidBody->getVelocityInLocalPoint(ra) * psb->m_sst.sdt : btVector3(0, 0, 0);
+					const btVector3 vb = n.m_x - n.m_q;
+					const btVector3 vr = vb - va;
+					const btScalar dn = btDot(vr, c.m_cti.m_normal);
+					const btVector3 fv = vr - c.m_cti.m_normal * dn;
+					const btScalar fc = psb->m_cfg.kDF * m_colObj1Wrap->getCollisionObject()->getFriction();
+					c.m_node = &n;
+					c.m_c0 = ImpulseMatrix(psb->m_sst.sdt, ima, imb, iwi, ra);
+					c.m_c1 = ra;
+					c.m_c2 = ima * psb->m_sst.sdt;
+					c.m_c3 = fv.length2() < (dn * fc * dn * fc) ? 0 : 1 - fc;
+					c.m_c4 = m_colObj1Wrap->getCollisionObject()->isStaticOrKinematicObject() ? psb->m_cfg.kKHR : psb->m_cfg.kCHR;
 					psb->m_rcontacts.push_back(c);
 					if (m_rigidBody)
 						m_rigidBody->activate();
 				}
 			}
 		}
-		btSoftBody*		psb;
-		const btCollisionObjectWrapper*	m_colObj1Wrap;
-		btRigidBody*	m_rigidBody;
-		btScalar		dynmargin;
-		btScalar		stamargin;
+		btSoftBody* psb;
+		const btCollisionObjectWrapper* m_colObj1Wrap;
+		btRigidBody* m_rigidBody;
+		btScalar dynmargin;
+		btScalar stamargin;
 	};
 	//
 	// CollideVF_SS
 	//
-	struct	CollideVF_SS : btDbvt::ICollide
+	struct CollideVF_SS : btDbvt::ICollide
 	{
-		void		Process(const btDbvtNode* lnode,
-			const btDbvtNode* lface)
+		void Process(const btDbvtNode* lnode,
+					 const btDbvtNode* lface)
 		{
-			btSoftBody::Node*	node=(btSoftBody::Node*)lnode->data;
-			btSoftBody::Face*	face=(btSoftBody::Face*)lface->data;
-			btVector3			o=node->m_x;
-			btVector3			p;
-			btScalar			d=SIMD_INFINITY;
-			ProjectOrigin(	face->m_n[0]->m_x-o,
-				face->m_n[1]->m_x-o,
-				face->m_n[2]->m_x-o,
-				p,d);
-			const btScalar	m=mrg+(o-node->m_q).length()*2;
-			if(d<(m*m))
+			btSoftBody::Node* node = (btSoftBody::Node*)lnode->data;
+			btSoftBody::Face* face = (btSoftBody::Face*)lface->data;
+			btVector3 o = node->m_x;
+			btVector3 p;
+			btScalar d = SIMD_INFINITY;
+			ProjectOrigin(face->m_n[0]->m_x - o,
+						  face->m_n[1]->m_x - o,
+						  face->m_n[2]->m_x - o,
+						  p, d);
+			const btScalar m = mrg + (o - node->m_q).length() * 2;
+			if (d < (m * m))
 			{
-				const btSoftBody::Node*	n[]={face->m_n[0],face->m_n[1],face->m_n[2]};
-				const btVector3			w=BaryCoord(n[0]->m_x,n[1]->m_x,n[2]->m_x,p+o);
-				const btScalar			ma=node->m_im;
-				btScalar				mb=BaryEval(n[0]->m_im,n[1]->m_im,n[2]->m_im,w);
-				if(	(n[0]->m_im<=0)||
-					(n[1]->m_im<=0)||
-					(n[2]->m_im<=0))
+				const btSoftBody::Node* n[] = {face->m_n[0], face->m_n[1], face->m_n[2]};
+				const btVector3 w = BaryCoord(n[0]->m_x, n[1]->m_x, n[2]->m_x, p + o);
+				const btScalar ma = node->m_im;
+				btScalar mb = BaryEval(n[0]->m_im, n[1]->m_im, n[2]->m_im, w);
+				if ((n[0]->m_im <= 0) ||
+					(n[1]->m_im <= 0) ||
+					(n[2]->m_im <= 0))
 				{
-					mb=0;
+					mb = 0;
 				}
-				const btScalar	ms=ma+mb;
-				if(ms>0)
+				const btScalar ms = ma + mb;
+				if (ms > 0)
 				{
-					btSoftBody::SContact	c;
-					c.m_normal		=	p/-btSqrt(d);
-					c.m_margin		=	m;
-					c.m_node		=	node;
-					c.m_face		=	face;
-					c.m_weights		=	w;
-					c.m_friction	=	btMax(psb[0]->m_cfg.kDF,psb[1]->m_cfg.kDF);
-					c.m_cfm[0]		=	ma/ms*psb[0]->m_cfg.kSHR;
-					c.m_cfm[1]		=	mb/ms*psb[1]->m_cfg.kSHR;
+					btSoftBody::SContact c;
+					c.m_normal = p / -btSqrt(d);
+					c.m_margin = m;
+					c.m_node = node;
+					c.m_face = face;
+					c.m_weights = w;
+					c.m_friction = btMax(psb[0]->m_cfg.kDF, psb[1]->m_cfg.kDF);
+					c.m_cfm[0] = ma / ms * psb[0]->m_cfg.kSHR;
+					c.m_cfm[1] = mb / ms * psb[1]->m_cfg.kSHR;
 					psb[0]->m_scontacts.push_back(c);
 				}
-			}	
+			}
 		}
-		btSoftBody*		psb[2];
-		btScalar		mrg;
+		btSoftBody* psb[2];
+		btScalar mrg;
 	};
 };
 
-#endif //_BT_SOFT_BODY_INTERNALS_H
+#endif  //_BT_SOFT_BODY_INTERNALS_H
diff --git a/thirdparty/bullet/BulletSoftBody/btSoftBodyRigidBodyCollisionConfiguration.cpp b/thirdparty/bullet/BulletSoftBody/btSoftBodyRigidBodyCollisionConfiguration.cpp
index f5a67f6d895..3127369ccd5 100644
--- a/thirdparty/bullet/BulletSoftBody/btSoftBodyRigidBodyCollisionConfiguration.cpp
+++ b/thirdparty/bullet/BulletSoftBody/btSoftBodyRigidBodyCollisionConfiguration.cpp
@@ -23,27 +23,27 @@ subject to the following restrictions:
 #define ENABLE_SOFTBODY_CONCAVE_COLLISIONS 1
 
 btSoftBodyRigidBodyCollisionConfiguration::btSoftBodyRigidBodyCollisionConfiguration(const btDefaultCollisionConstructionInfo& constructionInfo)
-:btDefaultCollisionConfiguration(constructionInfo)
+	: btDefaultCollisionConfiguration(constructionInfo)
 {
 	void* mem;
 
-	mem = btAlignedAlloc(sizeof(btSoftSoftCollisionAlgorithm::CreateFunc),16);
-	m_softSoftCreateFunc = new(mem) btSoftSoftCollisionAlgorithm::CreateFunc;
+	mem = btAlignedAlloc(sizeof(btSoftSoftCollisionAlgorithm::CreateFunc), 16);
+	m_softSoftCreateFunc = new (mem) btSoftSoftCollisionAlgorithm::CreateFunc;
 
-	mem = btAlignedAlloc(sizeof(btSoftRigidCollisionAlgorithm::CreateFunc),16);
-	m_softRigidConvexCreateFunc = new(mem) btSoftRigidCollisionAlgorithm::CreateFunc;
+	mem = btAlignedAlloc(sizeof(btSoftRigidCollisionAlgorithm::CreateFunc), 16);
+	m_softRigidConvexCreateFunc = new (mem) btSoftRigidCollisionAlgorithm::CreateFunc;
 
-	mem = btAlignedAlloc(sizeof(btSoftRigidCollisionAlgorithm::CreateFunc),16);
-	m_swappedSoftRigidConvexCreateFunc = new(mem) btSoftRigidCollisionAlgorithm::CreateFunc;
-	m_swappedSoftRigidConvexCreateFunc->m_swapped=true;
+	mem = btAlignedAlloc(sizeof(btSoftRigidCollisionAlgorithm::CreateFunc), 16);
+	m_swappedSoftRigidConvexCreateFunc = new (mem) btSoftRigidCollisionAlgorithm::CreateFunc;
+	m_swappedSoftRigidConvexCreateFunc->m_swapped = true;
 
 #ifdef ENABLE_SOFTBODY_CONCAVE_COLLISIONS
-	mem = btAlignedAlloc(sizeof(btSoftBodyConcaveCollisionAlgorithm::CreateFunc),16);
-	m_softRigidConcaveCreateFunc = new(mem) btSoftBodyConcaveCollisionAlgorithm::CreateFunc;
+	mem = btAlignedAlloc(sizeof(btSoftBodyConcaveCollisionAlgorithm::CreateFunc), 16);
+	m_softRigidConcaveCreateFunc = new (mem) btSoftBodyConcaveCollisionAlgorithm::CreateFunc;
 
-	mem = btAlignedAlloc(sizeof(btSoftBodyConcaveCollisionAlgorithm::CreateFunc),16);
-	m_swappedSoftRigidConcaveCreateFunc = new(mem) btSoftBodyConcaveCollisionAlgorithm::SwappedCreateFunc;
-	m_swappedSoftRigidConcaveCreateFunc->m_swapped=true;
+	mem = btAlignedAlloc(sizeof(btSoftBodyConcaveCollisionAlgorithm::CreateFunc), 16);
+	m_swappedSoftRigidConcaveCreateFunc = new (mem) btSoftBodyConcaveCollisionAlgorithm::SwappedCreateFunc;
+	m_swappedSoftRigidConcaveCreateFunc->m_swapped = true;
 #endif
 
 	//replace pool by a new one, with potential larger size
@@ -53,82 +53,79 @@ btSoftBodyRigidBodyCollisionConfiguration::btSoftBodyRigidBodyCollisionConfigura
 		int curElemSize = m_collisionAlgorithmPool->getElementSize();
 		///calculate maximum element size, big enough to fit any collision algorithm in the memory pool
 
-
 		int maxSize0 = sizeof(btSoftSoftCollisionAlgorithm);
 		int maxSize1 = sizeof(btSoftRigidCollisionAlgorithm);
 		int maxSize2 = sizeof(btSoftBodyConcaveCollisionAlgorithm);
 
-		int	collisionAlgorithmMaxElementSize = btMax(maxSize0,maxSize1);
-		collisionAlgorithmMaxElementSize = btMax(collisionAlgorithmMaxElementSize,maxSize2);
-		
+		int collisionAlgorithmMaxElementSize = btMax(maxSize0, maxSize1);
+		collisionAlgorithmMaxElementSize = btMax(collisionAlgorithmMaxElementSize, maxSize2);
+
 		if (collisionAlgorithmMaxElementSize > curElemSize)
 		{
 			m_collisionAlgorithmPool->~btPoolAllocator();
 			btAlignedFree(m_collisionAlgorithmPool);
-			void* mem = btAlignedAlloc(sizeof(btPoolAllocator),16);
-			m_collisionAlgorithmPool = new(mem) btPoolAllocator(collisionAlgorithmMaxElementSize,constructionInfo.m_defaultMaxCollisionAlgorithmPoolSize);
+			void* mem = btAlignedAlloc(sizeof(btPoolAllocator), 16);
+			m_collisionAlgorithmPool = new (mem) btPoolAllocator(collisionAlgorithmMaxElementSize, constructionInfo.m_defaultMaxCollisionAlgorithmPoolSize);
 		}
 	}
-
 }
 
 btSoftBodyRigidBodyCollisionConfiguration::~btSoftBodyRigidBodyCollisionConfiguration()
 {
 	m_softSoftCreateFunc->~btCollisionAlgorithmCreateFunc();
-	btAlignedFree(	m_softSoftCreateFunc);
+	btAlignedFree(m_softSoftCreateFunc);
 
 	m_softRigidConvexCreateFunc->~btCollisionAlgorithmCreateFunc();
-	btAlignedFree(	m_softRigidConvexCreateFunc);
+	btAlignedFree(m_softRigidConvexCreateFunc);
 
 	m_swappedSoftRigidConvexCreateFunc->~btCollisionAlgorithmCreateFunc();
-	btAlignedFree(	m_swappedSoftRigidConvexCreateFunc);
+	btAlignedFree(m_swappedSoftRigidConvexCreateFunc);
 
 #ifdef ENABLE_SOFTBODY_CONCAVE_COLLISIONS
 	m_softRigidConcaveCreateFunc->~btCollisionAlgorithmCreateFunc();
-	btAlignedFree(	m_softRigidConcaveCreateFunc);
+	btAlignedFree(m_softRigidConcaveCreateFunc);
 
 	m_swappedSoftRigidConcaveCreateFunc->~btCollisionAlgorithmCreateFunc();
-	btAlignedFree(	m_swappedSoftRigidConcaveCreateFunc);
+	btAlignedFree(m_swappedSoftRigidConcaveCreateFunc);
 #endif
 }
 
 ///creation of soft-soft and soft-rigid, and otherwise fallback to base class implementation
-btCollisionAlgorithmCreateFunc* btSoftBodyRigidBodyCollisionConfiguration::getCollisionAlgorithmCreateFunc(int proxyType0,int proxyType1)
+btCollisionAlgorithmCreateFunc* btSoftBodyRigidBodyCollisionConfiguration::getCollisionAlgorithmCreateFunc(int proxyType0, int proxyType1)
 {
-
 	///try to handle the softbody interactions first
 
-	if ((proxyType0 == SOFTBODY_SHAPE_PROXYTYPE  ) && (proxyType1==SOFTBODY_SHAPE_PROXYTYPE))
+	if ((proxyType0 == SOFTBODY_SHAPE_PROXYTYPE) && (proxyType1 == SOFTBODY_SHAPE_PROXYTYPE))
 	{
-		return	m_softSoftCreateFunc;
+		return m_softSoftCreateFunc;
 	}
 
 	///softbody versus convex
-	if (proxyType0 == SOFTBODY_SHAPE_PROXYTYPE  && btBroadphaseProxy::isConvex(proxyType1))
+	if (proxyType0 == SOFTBODY_SHAPE_PROXYTYPE && btBroadphaseProxy::isConvex(proxyType1))
 	{
-		return	m_softRigidConvexCreateFunc;
+		return m_softRigidConvexCreateFunc;
 	}
 
 	///convex versus soft body
-	if (btBroadphaseProxy::isConvex(proxyType0) && proxyType1 == SOFTBODY_SHAPE_PROXYTYPE )
+	if (btBroadphaseProxy::isConvex(proxyType0) && proxyType1 == SOFTBODY_SHAPE_PROXYTYPE)
 	{
-		return	m_swappedSoftRigidConvexCreateFunc;
+		return m_swappedSoftRigidConvexCreateFunc;
 	}
 
 #ifdef ENABLE_SOFTBODY_CONCAVE_COLLISIONS
 	///softbody versus convex
-	if (proxyType0 == SOFTBODY_SHAPE_PROXYTYPE  && btBroadphaseProxy::isConcave(proxyType1))
+	if (proxyType0 == SOFTBODY_SHAPE_PROXYTYPE && btBroadphaseProxy::isConcave(proxyType1))
 	{
-		return	m_softRigidConcaveCreateFunc;
+		return m_softRigidConcaveCreateFunc;
 	}
 
 	///convex versus soft body
-	if (btBroadphaseProxy::isConcave(proxyType0) && proxyType1 == SOFTBODY_SHAPE_PROXYTYPE )
+	if (btBroadphaseProxy::isConcave(proxyType0) && proxyType1 == SOFTBODY_SHAPE_PROXYTYPE)
 	{
-		return	m_swappedSoftRigidConcaveCreateFunc;
+		return m_swappedSoftRigidConcaveCreateFunc;
 	}
 #endif
 
 	///fallback to the regular rigid collision shape
-	return btDefaultCollisionConfiguration::getCollisionAlgorithmCreateFunc(proxyType0,proxyType1);
+	return btDefaultCollisionConfiguration::getCollisionAlgorithmCreateFunc(proxyType0, proxyType1);
 }
diff --git a/thirdparty/bullet/BulletSoftBody/btSoftBodyRigidBodyCollisionConfiguration.h b/thirdparty/bullet/BulletSoftBody/btSoftBodyRigidBodyCollisionConfiguration.h
index 21addcfe2e1..0396a52dac7 100644
--- a/thirdparty/bullet/BulletSoftBody/btSoftBodyRigidBodyCollisionConfiguration.h
+++ b/thirdparty/bullet/BulletSoftBody/btSoftBodyRigidBodyCollisionConfiguration.h
@@ -21,28 +21,23 @@ subject to the following restrictions:
 class btVoronoiSimplexSolver;
 class btGjkEpaPenetrationDepthSolver;
 
-
 ///btSoftBodyRigidBodyCollisionConfiguration add softbody interaction on top of btDefaultCollisionConfiguration
-class	btSoftBodyRigidBodyCollisionConfiguration : public btDefaultCollisionConfiguration
+class btSoftBodyRigidBodyCollisionConfiguration : public btDefaultCollisionConfiguration
 {
-
 	//default CreationFunctions, filling the m_doubleDispatch table
-	btCollisionAlgorithmCreateFunc*	m_softSoftCreateFunc;
-	btCollisionAlgorithmCreateFunc*	m_softRigidConvexCreateFunc;
-	btCollisionAlgorithmCreateFunc*	m_swappedSoftRigidConvexCreateFunc;
-	btCollisionAlgorithmCreateFunc*	m_softRigidConcaveCreateFunc;
-	btCollisionAlgorithmCreateFunc*	m_swappedSoftRigidConcaveCreateFunc;
+	btCollisionAlgorithmCreateFunc* m_softSoftCreateFunc;
+	btCollisionAlgorithmCreateFunc* m_softRigidConvexCreateFunc;
+	btCollisionAlgorithmCreateFunc* m_swappedSoftRigidConvexCreateFunc;
+	btCollisionAlgorithmCreateFunc* m_softRigidConcaveCreateFunc;
+	btCollisionAlgorithmCreateFunc* m_swappedSoftRigidConcaveCreateFunc;
 
 public:
-
 	btSoftBodyRigidBodyCollisionConfiguration(const btDefaultCollisionConstructionInfo& constructionInfo = btDefaultCollisionConstructionInfo());
 
 	virtual ~btSoftBodyRigidBodyCollisionConfiguration();
 
 	///creation of soft-soft and soft-rigid, and otherwise fallback to base class implementation
-	virtual btCollisionAlgorithmCreateFunc* getCollisionAlgorithmCreateFunc(int proxyType0,int proxyType1);
-
+	virtual btCollisionAlgorithmCreateFunc* getCollisionAlgorithmCreateFunc(int proxyType0, int proxyType1);
 };
 
-#endif //BT_SOFTBODY_RIGIDBODY_COLLISION_CONFIGURATION
-
+#endif  //BT_SOFTBODY_RIGIDBODY_COLLISION_CONFIGURATION
diff --git a/thirdparty/bullet/BulletSoftBody/btSoftBodySolverVertexBuffer.h b/thirdparty/bullet/BulletSoftBody/btSoftBodySolverVertexBuffer.h
index c4733d64000..bc538db4a25 100644
--- a/thirdparty/bullet/BulletSoftBody/btSoftBodySolverVertexBuffer.h
+++ b/thirdparty/bullet/BulletSoftBody/btSoftBodySolverVertexBuffer.h
@@ -16,7 +16,6 @@ subject to the following restrictions:
 #ifndef BT_SOFT_BODY_SOLVER_VERTEX_BUFFER_H
 #define BT_SOFT_BODY_SOLVER_VERTEX_BUFFER_H
 
-
 class btVertexBufferDescriptor
 {
 public:
@@ -27,8 +26,7 @@ public:
 		OPENGL_BUFFER
 	};
 
-protected:	
-
+protected:
 	bool m_hasVertexPositions;
 	bool m_hasNormals;
 
@@ -51,7 +49,6 @@ public:
 
 	virtual ~btVertexBufferDescriptor()
 	{
-
 	}
 
 	virtual bool hasVertexPositions() const
@@ -102,7 +99,6 @@ public:
 	}
 };
 
-
 class btCPUVertexBufferDescriptor : public btVertexBufferDescriptor
 {
 protected:
@@ -114,7 +110,7 @@ public:
 	 * vertexOffset is the offset in floats to the first vertex.
 	 * vertexStride is the stride in floats between vertices.
 	 */
-	btCPUVertexBufferDescriptor( float *basePointer, int vertexOffset, int vertexStride )
+	btCPUVertexBufferDescriptor(float *basePointer, int vertexOffset, int vertexStride)
 	{
 		m_basePointer = basePointer;
 		m_vertexOffset = vertexOffset;
@@ -127,7 +123,7 @@ public:
 	 * vertexOffset is the offset in floats to the first vertex.
 	 * vertexStride is the stride in floats between vertices.
 	 */
-	btCPUVertexBufferDescriptor( float *basePointer, int vertexOffset, int vertexStride, int normalOffset, int normalStride )
+	btCPUVertexBufferDescriptor(float *basePointer, int vertexOffset, int vertexStride, int normalOffset, int normalStride)
 	{
 		m_basePointer = basePointer;
 
@@ -142,7 +138,6 @@ public:
 
 	virtual ~btCPUVertexBufferDescriptor()
 	{
-
 	}
 
 	/**
@@ -162,4 +157,4 @@ public:
 	}
 };
 
-#endif // #ifndef BT_SOFT_BODY_SOLVER_VERTEX_BUFFER_H
+#endif  // #ifndef BT_SOFT_BODY_SOLVER_VERTEX_BUFFER_H
diff --git a/thirdparty/bullet/BulletSoftBody/btSoftBodySolvers.h b/thirdparty/bullet/BulletSoftBody/btSoftBodySolvers.h
index 6947bc27d28..dcf50826506 100644
--- a/thirdparty/bullet/BulletSoftBody/btSoftBodySolvers.h
+++ b/thirdparty/bullet/BulletSoftBody/btSoftBodySolvers.h
@@ -18,7 +18,6 @@ subject to the following restrictions:
 
 #include "BulletCollision/CollisionShapes/btTriangleIndexVertexArray.h"
 
-
 class btSoftBodyTriangleData;
 class btSoftBodyLinkData;
 class btSoftBodyVertexData;
@@ -26,7 +25,6 @@ class btVertexBufferDescriptor;
 class btCollisionObject;
 class btSoftBody;
 
-
 class btSoftBodySolver
 {
 public:
@@ -40,17 +38,15 @@ public:
 		DX_SIMD_SOLVER
 	};
 
-
 protected:
 	int m_numberOfPositionIterations;
 	int m_numberOfVelocityIterations;
 	// Simulation timescale
 	float m_timeScale;
-	
+
 public:
-	btSoftBodySolver() :
-		m_numberOfPositionIterations( 10 ),
-		m_timeScale( 1 )
+	btSoftBodySolver() : m_numberOfPositionIterations(10),
+						 m_timeScale(1)
 	{
 		m_numberOfVelocityIterations = 0;
 		m_numberOfPositionIterations = 5;
@@ -59,39 +55,38 @@ public:
 	virtual ~btSoftBodySolver()
 	{
 	}
-	
+
 	/**
 	 * Return the type of the solver.
 	 */
 	virtual SolverTypes getSolverType() const = 0;
 
-
 	/** Ensure that this solver is initialized. */
 	virtual bool checkInitialized() = 0;
 
 	/** Optimize soft bodies in this solver. */
-	virtual void optimize( btAlignedObjectArray< btSoftBody * > &softBodies , bool forceUpdate=false) = 0;
+	virtual void optimize(btAlignedObjectArray<btSoftBody *> &softBodies, bool forceUpdate = false) = 0;
 
 	/** Copy necessary data back to the original soft body source objects. */
 	virtual void copyBackToSoftBodies(bool bMove = true) = 0;
 
 	/** Predict motion of soft bodies into next timestep */
-	virtual void predictMotion( float solverdt ) = 0;
+	virtual void predictMotion(float solverdt) = 0;
 
 	/** Solve constraints for a set of soft bodies */
-	virtual void solveConstraints( float solverdt ) = 0;
+	virtual void solveConstraints(float solverdt) = 0;
 
 	/** Perform necessary per-step updates of soft bodies such as recomputing normals and bounding boxes */
 	virtual void updateSoftBodies() = 0;
 
 	/** Process a collision between one of the world's soft bodies and another collision object */
-	virtual void processCollision( btSoftBody *, const struct btCollisionObjectWrapper* ) = 0;
+	virtual void processCollision(btSoftBody *, const struct btCollisionObjectWrapper *) = 0;
 
 	/** Process a collision between two soft bodies */
-	virtual void processCollision( btSoftBody*, btSoftBody* ) = 0;
+	virtual void processCollision(btSoftBody *, btSoftBody *) = 0;
 
 	/** Set the number of velocity constraint solver iterations this solver uses. */
-	virtual void setNumberOfPositionIterations( int iterations )
+	virtual void setNumberOfPositionIterations(int iterations)
 	{
 		m_numberOfPositionIterations = iterations;
 	}
@@ -103,7 +98,7 @@ public:
 	}
 
 	/** Set the number of velocity constraint solver iterations this solver uses. */
-	virtual void setNumberOfVelocityIterations( int iterations )
+	virtual void setNumberOfVelocityIterations(int iterations)
 	{
 		m_numberOfVelocityIterations = iterations;
 	}
@@ -135,7 +130,6 @@ public:
 class btSoftBodySolverOutput
 {
 protected:
-
 public:
 	btSoftBodySolverOutput()
 	{
@@ -145,10 +139,8 @@ public:
 	{
 	}
 
-
 	/** Output current computed vertex data to the vertex buffers for all cloths in the solver. */
-	virtual void copySoftBodyToVertexBuffer( const btSoftBody * const softBody, btVertexBufferDescriptor *vertexBuffer ) = 0;
+	virtual void copySoftBodyToVertexBuffer(const btSoftBody *const softBody, btVertexBufferDescriptor *vertexBuffer) = 0;
 };
 
-
-#endif // #ifndef BT_SOFT_BODY_SOLVERS_H
+#endif  // #ifndef BT_SOFT_BODY_SOLVERS_H
diff --git a/thirdparty/bullet/BulletSoftBody/btSoftMultiBodyDynamicsWorld.cpp b/thirdparty/bullet/BulletSoftBody/btSoftMultiBodyDynamicsWorld.cpp
index 6facce4e863..282dbf75f08 100644
--- a/thirdparty/bullet/BulletSoftBody/btSoftMultiBodyDynamicsWorld.cpp
+++ b/thirdparty/bullet/BulletSoftBody/btSoftMultiBodyDynamicsWorld.cpp
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btSoftMultiBodyDynamicsWorld.h"
 #include "LinearMath/btQuickprof.h"
 
@@ -24,42 +23,38 @@ subject to the following restrictions:
 #include "BulletSoftBody/btDefaultSoftBodySolver.h"
 #include "LinearMath/btSerializer.h"
 
-
 btSoftMultiBodyDynamicsWorld::btSoftMultiBodyDynamicsWorld(
 	btDispatcher* dispatcher,
 	btBroadphaseInterface* pairCache,
 	btMultiBodyConstraintSolver* constraintSolver,
 	btCollisionConfiguration* collisionConfiguration,
-	btSoftBodySolver *softBodySolver ) : 
-		btMultiBodyDynamicsWorld(dispatcher,pairCache,constraintSolver,collisionConfiguration),
-        m_softBodySolver( softBodySolver ),
-        m_ownsSolver(false)
+	btSoftBodySolver* softBodySolver) : btMultiBodyDynamicsWorld(dispatcher, pairCache, constraintSolver, collisionConfiguration),
+										m_softBodySolver(softBodySolver),
+										m_ownsSolver(false)
 {
-	if( !m_softBodySolver )
+	if (!m_softBodySolver)
 	{
-		void* ptr = btAlignedAlloc(sizeof(btDefaultSoftBodySolver),16);
-		m_softBodySolver = new(ptr) btDefaultSoftBodySolver();
+		void* ptr = btAlignedAlloc(sizeof(btDefaultSoftBodySolver), 16);
+		m_softBodySolver = new (ptr) btDefaultSoftBodySolver();
 		m_ownsSolver = true;
 	}
 
-	m_drawFlags			=	fDrawFlags::Std;
-	m_drawNodeTree		=	true;
-	m_drawFaceTree		=	false;
-	m_drawClusterTree	=	false;
+	m_drawFlags = fDrawFlags::Std;
+	m_drawNodeTree = true;
+	m_drawFaceTree = false;
+	m_drawClusterTree = false;
 	m_sbi.m_broadphase = pairCache;
 	m_sbi.m_dispatcher = dispatcher;
 	m_sbi.m_sparsesdf.Initialize();
 	m_sbi.m_sparsesdf.Reset();
 
-	m_sbi.air_density		=	(btScalar)1.2;
-	m_sbi.water_density	=	0;
-	m_sbi.water_offset		=	0;
-	m_sbi.water_normal		=	btVector3(0,0,0);
-	m_sbi.m_gravity.setValue(0,-10,0);
+	m_sbi.air_density = (btScalar)1.2;
+	m_sbi.water_density = 0;
+	m_sbi.water_offset = 0;
+	m_sbi.water_normal = btVector3(0, 0, 0);
+	m_sbi.m_gravity.setValue(0, -10, 0);
 
 	m_sbi.m_sparsesdf.Initialize();
-
-
 }
 
 btSoftMultiBodyDynamicsWorld::~btSoftMultiBodyDynamicsWorld()
@@ -71,82 +66,78 @@ btSoftMultiBodyDynamicsWorld::~btSoftMultiBodyDynamicsWorld()
 	}
 }
 
-void	btSoftMultiBodyDynamicsWorld::predictUnconstraintMotion(btScalar timeStep)
+void btSoftMultiBodyDynamicsWorld::predictUnconstraintMotion(btScalar timeStep)
 {
-	btDiscreteDynamicsWorld::predictUnconstraintMotion( timeStep );
+	btDiscreteDynamicsWorld::predictUnconstraintMotion(timeStep);
 	{
 		BT_PROFILE("predictUnconstraintMotionSoftBody");
-		m_softBodySolver->predictMotion( float(timeStep) );
+		m_softBodySolver->predictMotion(float(timeStep));
 	}
 }
 
-void	btSoftMultiBodyDynamicsWorld::internalSingleStepSimulation( btScalar timeStep )
+void btSoftMultiBodyDynamicsWorld::internalSingleStepSimulation(btScalar timeStep)
 {
-
 	// Let the solver grab the soft bodies and if necessary optimize for it
-	m_softBodySolver->optimize( getSoftBodyArray() );
+	m_softBodySolver->optimize(getSoftBodyArray());
 
-	if( !m_softBodySolver->checkInitialized() )
+	if (!m_softBodySolver->checkInitialized())
 	{
-		btAssert( "Solver initialization failed\n" );
+		btAssert("Solver initialization failed\n");
 	}
 
-	btDiscreteDynamicsWorld::internalSingleStepSimulation( timeStep );
+	btDiscreteDynamicsWorld::internalSingleStepSimulation(timeStep);
 
 	///solve soft bodies constraints
-	solveSoftBodiesConstraints( timeStep );
+	solveSoftBodiesConstraints(timeStep);
 
 	//self collisions
-	for ( int i=0;i<m_softBodies.size();i++)
+	for (int i = 0; i < m_softBodies.size(); i++)
 	{
-		btSoftBody*	psb=(btSoftBody*)m_softBodies[i];
+		btSoftBody* psb = (btSoftBody*)m_softBodies[i];
 		psb->defaultCollisionHandler(psb);
 	}
 
 	///update soft bodies
-	m_softBodySolver->updateSoftBodies( );
-	
+	m_softBodySolver->updateSoftBodies();
+
 	// End solver-wise simulation step
 	// ///////////////////////////////
-
 }
 
-void	btSoftMultiBodyDynamicsWorld::solveSoftBodiesConstraints( btScalar timeStep )
+void btSoftMultiBodyDynamicsWorld::solveSoftBodiesConstraints(btScalar timeStep)
 {
 	BT_PROFILE("solveSoftConstraints");
 
-	if(m_softBodies.size())
+	if (m_softBodies.size())
 	{
 		btSoftBody::solveClusters(m_softBodies);
 	}
 
 	// Solve constraints solver-wise
-	m_softBodySolver->solveConstraints( timeStep * m_softBodySolver->getTimeScale() );
-
+	m_softBodySolver->solveConstraints(timeStep * m_softBodySolver->getTimeScale());
 }
 
-void	btSoftMultiBodyDynamicsWorld::addSoftBody(btSoftBody* body, int collisionFilterGroup, int collisionFilterMask)
+void btSoftMultiBodyDynamicsWorld::addSoftBody(btSoftBody* body, int collisionFilterGroup, int collisionFilterMask)
 {
 	m_softBodies.push_back(body);
 
 	// Set the soft body solver that will deal with this body
 	// to be the world's solver
-	body->setSoftBodySolver( m_softBodySolver );
+	body->setSoftBodySolver(m_softBodySolver);
 
 	btCollisionWorld::addCollisionObject(body,
-		collisionFilterGroup,
-		collisionFilterMask);
-
+										 collisionFilterGroup,
+										 collisionFilterMask);
 }
 
-void	btSoftMultiBodyDynamicsWorld::removeSoftBody(btSoftBody* body)
+void btSoftMultiBodyDynamicsWorld::removeSoftBody(btSoftBody* body)
 {
 	m_softBodies.remove(body);
 
 	btCollisionWorld::removeCollisionObject(body);
 }
 
-void	btSoftMultiBodyDynamicsWorld::removeCollisionObject(btCollisionObject* collisionObject)
+void btSoftMultiBodyDynamicsWorld::removeCollisionObject(btCollisionObject* collisionObject)
 {
 	btSoftBody* body = btSoftBody::upcast(collisionObject);
 	if (body)
@@ -155,60 +146,57 @@ void	btSoftMultiBodyDynamicsWorld::removeCollisionObject(btCollisionObject* coll
 		btDiscreteDynamicsWorld::removeCollisionObject(collisionObject);
 }
 
-void	btSoftMultiBodyDynamicsWorld::debugDrawWorld()
+void btSoftMultiBodyDynamicsWorld::debugDrawWorld()
 {
 	btMultiBodyDynamicsWorld::debugDrawWorld();
 
 	if (getDebugDrawer())
 	{
 		int i;
-		for (  i=0;i<this->m_softBodies.size();i++)
+		for (i = 0; i < this->m_softBodies.size(); i++)
 		{
-			btSoftBody*	psb=(btSoftBody*)this->m_softBodies[i];
+			btSoftBody* psb = (btSoftBody*)this->m_softBodies[i];
 			if (getDebugDrawer() && (getDebugDrawer()->getDebugMode() & (btIDebugDraw::DBG_DrawWireframe)))
 			{
-				btSoftBodyHelpers::DrawFrame(psb,m_debugDrawer);
-				btSoftBodyHelpers::Draw(psb,m_debugDrawer,m_drawFlags);
+				btSoftBodyHelpers::DrawFrame(psb, m_debugDrawer);
+				btSoftBodyHelpers::Draw(psb, m_debugDrawer, m_drawFlags);
 			}
-			
+
 			if (m_debugDrawer && (m_debugDrawer->getDebugMode() & btIDebugDraw::DBG_DrawAabb))
 			{
-				if(m_drawNodeTree)		btSoftBodyHelpers::DrawNodeTree(psb,m_debugDrawer);
-				if(m_drawFaceTree)		btSoftBodyHelpers::DrawFaceTree(psb,m_debugDrawer);
-				if(m_drawClusterTree)	btSoftBodyHelpers::DrawClusterTree(psb,m_debugDrawer);
+				if (m_drawNodeTree) btSoftBodyHelpers::DrawNodeTree(psb, m_debugDrawer);
+				if (m_drawFaceTree) btSoftBodyHelpers::DrawFaceTree(psb, m_debugDrawer);
+				if (m_drawClusterTree) btSoftBodyHelpers::DrawClusterTree(psb, m_debugDrawer);
 			}
-		}		
-	}	
+		}
+	}
 }
 
-
-
-
 struct btSoftSingleRayCallback : public btBroadphaseRayCallback
 {
-	btVector3	m_rayFromWorld;
-	btVector3	m_rayToWorld;
-	btTransform	m_rayFromTrans;
-	btTransform	m_rayToTrans;
-	btVector3	m_hitNormal;
+	btVector3 m_rayFromWorld;
+	btVector3 m_rayToWorld;
+	btTransform m_rayFromTrans;
+	btTransform m_rayToTrans;
+	btVector3 m_hitNormal;
 
-	const btSoftMultiBodyDynamicsWorld*	m_world;
-	btCollisionWorld::RayResultCallback&	m_resultCallback;
+	const btSoftMultiBodyDynamicsWorld* m_world;
+	btCollisionWorld::RayResultCallback& m_resultCallback;
 
-	btSoftSingleRayCallback(const btVector3& rayFromWorld,const btVector3& rayToWorld,const btSoftMultiBodyDynamicsWorld* world,btCollisionWorld::RayResultCallback& resultCallback)
-	:m_rayFromWorld(rayFromWorld),
-	m_rayToWorld(rayToWorld),
-	m_world(world),
-	m_resultCallback(resultCallback)
+	btSoftSingleRayCallback(const btVector3& rayFromWorld, const btVector3& rayToWorld, const btSoftMultiBodyDynamicsWorld* world, btCollisionWorld::RayResultCallback& resultCallback)
+		: m_rayFromWorld(rayFromWorld),
+		  m_rayToWorld(rayToWorld),
+		  m_world(world),
+		  m_resultCallback(resultCallback)
 	{
 		m_rayFromTrans.setIdentity();
 		m_rayFromTrans.setOrigin(m_rayFromWorld);
 		m_rayToTrans.setIdentity();
 		m_rayToTrans.setOrigin(m_rayToWorld);
 
-		btVector3 rayDir = (rayToWorld-rayFromWorld);
+		btVector3 rayDir = (rayToWorld - rayFromWorld);
 
-		rayDir.normalize ();
+		rayDir.normalize();
 		///what about division by zero? --> just set rayDirection[i] to INF/1e30
 		m_rayDirectionInverse[0] = rayDir[0] == btScalar(0.0) ? btScalar(1e30) : btScalar(1.0) / rayDir[0];
 		m_rayDirectionInverse[1] = rayDir[1] == btScalar(0.0) ? btScalar(1e30) : btScalar(1.0) / rayDir[1];
@@ -217,22 +205,19 @@ struct btSoftSingleRayCallback : public btBroadphaseRayCallback
 		m_signs[1] = m_rayDirectionInverse[1] < 0.0;
 		m_signs[2] = m_rayDirectionInverse[2] < 0.0;
 
-		m_lambda_max = rayDir.dot(m_rayToWorld-m_rayFromWorld);
-
+		m_lambda_max = rayDir.dot(m_rayToWorld - m_rayFromWorld);
 	}
 
-	
-
-	virtual bool	process(const btBroadphaseProxy* proxy)
+	virtual bool process(const btBroadphaseProxy* proxy)
 	{
 		///terminate further ray tests, once the closestHitFraction reached zero
 		if (m_resultCallback.m_closestHitFraction == btScalar(0.f))
 			return false;
 
-		btCollisionObject*	collisionObject = (btCollisionObject*)proxy->m_clientObject;
+		btCollisionObject* collisionObject = (btCollisionObject*)proxy->m_clientObject;
 
 		//only perform raycast if filterMask matches
-		if(m_resultCallback.needsCollision(collisionObject->getBroadphaseHandle())) 
+		if (m_resultCallback.needsCollision(collisionObject->getBroadphaseHandle()))
 		{
 			//RigidcollisionObject* collisionObject = ctrl->GetRigidcollisionObject();
 			//btVector3 collisionObjectAabbMin,collisionObjectAabbMax;
@@ -250,110 +235,106 @@ struct btSoftSingleRayCallback : public btBroadphaseRayCallback
 			//culling already done by broadphase
 			//if (btRayAabb(m_rayFromWorld,m_rayToWorld,collisionObjectAabbMin,collisionObjectAabbMax,hitLambda,m_hitNormal))
 			{
-				m_world->rayTestSingle(m_rayFromTrans,m_rayToTrans,
-					collisionObject,
-						collisionObject->getCollisionShape(),
-						collisionObject->getWorldTransform(),
-						m_resultCallback);
+				m_world->rayTestSingle(m_rayFromTrans, m_rayToTrans,
+									   collisionObject,
+									   collisionObject->getCollisionShape(),
+									   collisionObject->getWorldTransform(),
+									   m_resultCallback);
 			}
 		}
 		return true;
 	}
 };
 
-void	btSoftMultiBodyDynamicsWorld::rayTest(const btVector3& rayFromWorld, const btVector3& rayToWorld, RayResultCallback& resultCallback) const
+void btSoftMultiBodyDynamicsWorld::rayTest(const btVector3& rayFromWorld, const btVector3& rayToWorld, RayResultCallback& resultCallback) const
 {
 	BT_PROFILE("rayTest");
 	/// use the broadphase to accelerate the search for objects, based on their aabb
 	/// and for each object with ray-aabb overlap, perform an exact ray test
-	btSoftSingleRayCallback rayCB(rayFromWorld,rayToWorld,this,resultCallback);
+	btSoftSingleRayCallback rayCB(rayFromWorld, rayToWorld, this, resultCallback);
 
 #ifndef USE_BRUTEFORCE_RAYBROADPHASE
-	m_broadphasePairCache->rayTest(rayFromWorld,rayToWorld,rayCB);
+	m_broadphasePairCache->rayTest(rayFromWorld, rayToWorld, rayCB);
 #else
-	for (int i=0;i<this->getNumCollisionObjects();i++)
+	for (int i = 0; i < this->getNumCollisionObjects(); i++)
 	{
 		rayCB.process(m_collisionObjects[i]->getBroadphaseHandle());
-	}	
-#endif //USE_BRUTEFORCE_RAYBROADPHASE
-
+	}
+#endif  //USE_BRUTEFORCE_RAYBROADPHASE
 }
 
-
-void	btSoftMultiBodyDynamicsWorld::rayTestSingle(const btTransform& rayFromTrans,const btTransform& rayToTrans,
-					  btCollisionObject* collisionObject,
-					  const btCollisionShape* collisionShape,
-					  const btTransform& colObjWorldTransform,
-					  RayResultCallback& resultCallback)
+void btSoftMultiBodyDynamicsWorld::rayTestSingle(const btTransform& rayFromTrans, const btTransform& rayToTrans,
+												 btCollisionObject* collisionObject,
+												 const btCollisionShape* collisionShape,
+												 const btTransform& colObjWorldTransform,
+												 RayResultCallback& resultCallback)
 {
-	if (collisionShape->isSoftBody()) {
+	if (collisionShape->isSoftBody())
+	{
 		btSoftBody* softBody = btSoftBody::upcast(collisionObject);
-		if (softBody) {
+		if (softBody)
+		{
 			btSoftBody::sRayCast softResult;
-			if (softBody->rayTest(rayFromTrans.getOrigin(), rayToTrans.getOrigin(), softResult)) 
+			if (softBody->rayTest(rayFromTrans.getOrigin(), rayToTrans.getOrigin(), softResult))
 			{
-				
-				if (softResult.fraction<= resultCallback.m_closestHitFraction)
+				if (softResult.fraction <= resultCallback.m_closestHitFraction)
 				{
-
 					btCollisionWorld::LocalShapeInfo shapeInfo;
 					shapeInfo.m_shapePart = 0;
 					shapeInfo.m_triangleIndex = softResult.index;
 					// get the normal
 					btVector3 rayDir = rayToTrans.getOrigin() - rayFromTrans.getOrigin();
-					btVector3 normal=-rayDir;
+					btVector3 normal = -rayDir;
 					normal.normalize();
 
 					if (softResult.feature == btSoftBody::eFeature::Face)
 					{
 						normal = softBody->m_faces[softResult.index].m_normal;
-						if (normal.dot(rayDir) > 0) {
+						if (normal.dot(rayDir) > 0)
+						{
 							// normal always point toward origin of the ray
 							normal = -normal;
 						}
 					}
-	
-					btCollisionWorld::LocalRayResult rayResult
-						(collisionObject,
-						 &shapeInfo,
-						 normal,
-						 softResult.fraction);
-					bool	normalInWorldSpace = true;
-					resultCallback.addSingleResult(rayResult,normalInWorldSpace);
+
+					btCollisionWorld::LocalRayResult rayResult(collisionObject,
+															   &shapeInfo,
+															   normal,
+															   softResult.fraction);
+					bool normalInWorldSpace = true;
+					resultCallback.addSingleResult(rayResult, normalInWorldSpace);
 				}
 			}
 		}
-	} 
-	else {
-		btCollisionWorld::rayTestSingle(rayFromTrans,rayToTrans,collisionObject,collisionShape,colObjWorldTransform,resultCallback);
+	}
+	else
+	{
+		btCollisionWorld::rayTestSingle(rayFromTrans, rayToTrans, collisionObject, collisionShape, colObjWorldTransform, resultCallback);
 	}
 }
 
-
-void	btSoftMultiBodyDynamicsWorld::serializeSoftBodies(btSerializer* serializer)
+void btSoftMultiBodyDynamicsWorld::serializeSoftBodies(btSerializer* serializer)
 {
 	int i;
 	//serialize all collision objects
-	for (i=0;i<m_collisionObjects.size();i++)
+	for (i = 0; i < m_collisionObjects.size(); i++)
 	{
 		btCollisionObject* colObj = m_collisionObjects[i];
 		if (colObj->getInternalType() & btCollisionObject::CO_SOFT_BODY)
 		{
 			int len = colObj->calculateSerializeBufferSize();
-			btChunk* chunk = serializer->allocate(len,1);
+			btChunk* chunk = serializer->allocate(len, 1);
 			const char* structType = colObj->serialize(chunk->m_oldPtr, serializer);
-			serializer->finalizeChunk(chunk,structType,BT_SOFTBODY_CODE,colObj);
+			serializer->finalizeChunk(chunk, structType, BT_SOFTBODY_CODE, colObj);
 		}
 	}
-
 }
 
-void	btSoftMultiBodyDynamicsWorld::serialize(btSerializer* serializer)
+void btSoftMultiBodyDynamicsWorld::serialize(btSerializer* serializer)
 {
-
 	serializer->startSerialization();
 
-	serializeDynamicsWorldInfo( serializer);
+	serializeDynamicsWorldInfo(serializer);
 
 	serializeSoftBodies(serializer);
 
@@ -367,5 +348,3 @@ void	btSoftMultiBodyDynamicsWorld::serialize(btSerializer* serializer)
 
 	serializer->finishSerialization();
 }
-
-
diff --git a/thirdparty/bullet/BulletSoftBody/btSoftMultiBodyDynamicsWorld.h b/thirdparty/bullet/BulletSoftBody/btSoftMultiBodyDynamicsWorld.h
index 6d46a21db5c..f295945a6db 100644
--- a/thirdparty/bullet/BulletSoftBody/btSoftMultiBodyDynamicsWorld.h
+++ b/thirdparty/bullet/BulletSoftBody/btSoftMultiBodyDynamicsWorld.h
@@ -21,64 +21,61 @@ subject to the following restrictions:
 #include "BulletSoftBody/btSoftBody.h"
 
 #ifndef BT_SOFT_RIGID_DYNAMICS_WORLD_H
-typedef	btAlignedObjectArray<btSoftBody*> btSoftBodyArray;
+typedef btAlignedObjectArray<btSoftBody*> btSoftBodyArray;
 #endif
 
 class btSoftBodySolver;
 
 class btSoftMultiBodyDynamicsWorld : public btMultiBodyDynamicsWorld
 {
-
-	btSoftBodyArray	m_softBodies;
-	int				m_drawFlags;
-	bool			m_drawNodeTree;
-	bool			m_drawFaceTree;
-	bool			m_drawClusterTree;
+	btSoftBodyArray m_softBodies;
+	int m_drawFlags;
+	bool m_drawNodeTree;
+	bool m_drawFaceTree;
+	bool m_drawClusterTree;
 	btSoftBodyWorldInfo m_sbi;
 	///Solver classes that encapsulate multiple soft bodies for solving
-	btSoftBodySolver *m_softBodySolver;
-	bool			m_ownsSolver;
+	btSoftBodySolver* m_softBodySolver;
+	bool m_ownsSolver;
 
 protected:
+	virtual void predictUnconstraintMotion(btScalar timeStep);
 
-	virtual void	predictUnconstraintMotion(btScalar timeStep);
+	virtual void internalSingleStepSimulation(btScalar timeStep);
 
-	virtual void	internalSingleStepSimulation( btScalar timeStep);
+	void solveSoftBodiesConstraints(btScalar timeStep);
 
-	void	solveSoftBodiesConstraints( btScalar timeStep );
-
-	void	serializeSoftBodies(btSerializer* serializer);
+	void serializeSoftBodies(btSerializer* serializer);
 
 public:
-
-	btSoftMultiBodyDynamicsWorld(btDispatcher* dispatcher,btBroadphaseInterface* pairCache,btMultiBodyConstraintSolver* constraintSolver, btCollisionConfiguration* collisionConfiguration, btSoftBodySolver *softBodySolver = 0 );
+	btSoftMultiBodyDynamicsWorld(btDispatcher* dispatcher, btBroadphaseInterface* pairCache, btMultiBodyConstraintSolver* constraintSolver, btCollisionConfiguration* collisionConfiguration, btSoftBodySolver* softBodySolver = 0);
 
 	virtual ~btSoftMultiBodyDynamicsWorld();
 
-	virtual void	debugDrawWorld();
+	virtual void debugDrawWorld();
 
-	void	addSoftBody(btSoftBody* body, int collisionFilterGroup=btBroadphaseProxy::DefaultFilter, int collisionFilterMask=btBroadphaseProxy::AllFilter);
+	void addSoftBody(btSoftBody* body, int collisionFilterGroup = btBroadphaseProxy::DefaultFilter, int collisionFilterMask = btBroadphaseProxy::AllFilter);
 
-	void	removeSoftBody(btSoftBody* body);
+	void removeSoftBody(btSoftBody* body);
 
 	///removeCollisionObject will first check if it is a rigid body, if so call removeRigidBody otherwise call btDiscreteDynamicsWorld::removeCollisionObject
-	virtual void	removeCollisionObject(btCollisionObject* collisionObject);
+	virtual void removeCollisionObject(btCollisionObject* collisionObject);
 
-	int		getDrawFlags() const { return(m_drawFlags); }
-	void	setDrawFlags(int f)	{ m_drawFlags=f; }
+	int getDrawFlags() const { return (m_drawFlags); }
+	void setDrawFlags(int f) { m_drawFlags = f; }
 
-	btSoftBodyWorldInfo&	getWorldInfo()
+	btSoftBodyWorldInfo& getWorldInfo()
 	{
 		return m_sbi;
 	}
-	const btSoftBodyWorldInfo&	getWorldInfo() const
+	const btSoftBodyWorldInfo& getWorldInfo() const
 	{
 		return m_sbi;
 	}
 
-	virtual btDynamicsWorldType	getWorldType() const
+	virtual btDynamicsWorldType getWorldType() const
 	{
-		return	BT_SOFT_MULTIBODY_DYNAMICS_WORLD;
+		return BT_SOFT_MULTIBODY_DYNAMICS_WORLD;
 	}
 
 	btSoftBodyArray& getSoftBodyArray()
@@ -91,20 +88,18 @@ public:
 		return m_softBodies;
 	}
 
-
-	virtual void rayTest(const btVector3& rayFromWorld, const btVector3& rayToWorld, RayResultCallback& resultCallback) const; 
+	virtual void rayTest(const btVector3& rayFromWorld, const btVector3& rayToWorld, RayResultCallback& resultCallback) const;
 
 	/// rayTestSingle performs a raycast call and calls the resultCallback. It is used internally by rayTest.
 	/// In a future implementation, we consider moving the ray test as a virtual method in btCollisionShape.
 	/// This allows more customization.
-	static void	rayTestSingle(const btTransform& rayFromTrans,const btTransform& rayToTrans,
-					  btCollisionObject* collisionObject,
-					  const btCollisionShape* collisionShape,
-					  const btTransform& colObjWorldTransform,
-					  RayResultCallback& resultCallback);
-
-	virtual	void	serialize(btSerializer* serializer);
+	static void rayTestSingle(const btTransform& rayFromTrans, const btTransform& rayToTrans,
+							  btCollisionObject* collisionObject,
+							  const btCollisionShape* collisionShape,
+							  const btTransform& colObjWorldTransform,
+							  RayResultCallback& resultCallback);
 
+	virtual void serialize(btSerializer* serializer);
 };
 
-#endif //BT_SOFT_MULTIBODY_DYNAMICS_WORLD_H
+#endif  //BT_SOFT_MULTIBODY_DYNAMICS_WORLD_H
diff --git a/thirdparty/bullet/BulletSoftBody/btSoftRigidCollisionAlgorithm.cpp b/thirdparty/bullet/BulletSoftBody/btSoftRigidCollisionAlgorithm.cpp
index 01c148a2ca8..56d8083f22a 100644
--- a/thirdparty/bullet/BulletSoftBody/btSoftRigidCollisionAlgorithm.cpp
+++ b/thirdparty/bullet/BulletSoftBody/btSoftRigidCollisionAlgorithm.cpp
@@ -27,18 +27,16 @@ subject to the following restrictions:
 
 //#include <stdio.h>
 
-btSoftRigidCollisionAlgorithm::btSoftRigidCollisionAlgorithm(btPersistentManifold* /*mf*/,const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* ,const btCollisionObjectWrapper* , bool isSwapped)
-: btCollisionAlgorithm(ci),
-//m_ownManifold(false),
-//m_manifoldPtr(mf),
-m_isSwapped(isSwapped)
+btSoftRigidCollisionAlgorithm::btSoftRigidCollisionAlgorithm(btPersistentManifold* /*mf*/, const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper*, const btCollisionObjectWrapper*, bool isSwapped)
+	: btCollisionAlgorithm(ci),
+	  //m_ownManifold(false),
+	  //m_manifoldPtr(mf),
+	  m_isSwapped(isSwapped)
 {
 }
 
-
 btSoftRigidCollisionAlgorithm::~btSoftRigidCollisionAlgorithm()
 {
-
 	//m_softBody->m_overlappingRigidBodies.remove(m_rigidCollisionObject);
 
 	/*if (m_ownManifold)
@@ -47,31 +45,27 @@ btSoftRigidCollisionAlgorithm::~btSoftRigidCollisionAlgorithm()
 	m_dispatcher->releaseManifold(m_manifoldPtr);
 	}
 	*/
-
 }
 
-
 #include <stdio.h>
 
-void btSoftRigidCollisionAlgorithm::processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+void btSoftRigidCollisionAlgorithm::processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	(void)dispatchInfo;
 	(void)resultOut;
 	//printf("btSoftRigidCollisionAlgorithm\n");
-//	const btCollisionObjectWrapper* softWrap = m_isSwapped?body1Wrap:body0Wrap;
-//	const btCollisionObjectWrapper* rigidWrap = m_isSwapped?body0Wrap:body1Wrap;
-	btSoftBody* softBody =  m_isSwapped? (btSoftBody*)body1Wrap->getCollisionObject() : (btSoftBody*)body0Wrap->getCollisionObject();
-	const btCollisionObjectWrapper* rigidCollisionObjectWrap = m_isSwapped? body0Wrap : body1Wrap;
-	
-	if (softBody->m_collisionDisabledObjects.findLinearSearch(rigidCollisionObjectWrap->getCollisionObject())==softBody->m_collisionDisabledObjects.size())
+	//	const btCollisionObjectWrapper* softWrap = m_isSwapped?body1Wrap:body0Wrap;
+	//	const btCollisionObjectWrapper* rigidWrap = m_isSwapped?body0Wrap:body1Wrap;
+	btSoftBody* softBody = m_isSwapped ? (btSoftBody*)body1Wrap->getCollisionObject() : (btSoftBody*)body0Wrap->getCollisionObject();
+	const btCollisionObjectWrapper* rigidCollisionObjectWrap = m_isSwapped ? body0Wrap : body1Wrap;
+
+	if (softBody->m_collisionDisabledObjects.findLinearSearch(rigidCollisionObjectWrap->getCollisionObject()) == softBody->m_collisionDisabledObjects.size())
 	{
 		softBody->getSoftBodySolver()->processCollision(softBody, rigidCollisionObjectWrap);
 	}
-
-
 }
 
-btScalar btSoftRigidCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* col0,btCollisionObject* col1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut)
+btScalar btSoftRigidCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* col0, btCollisionObject* col1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut)
 {
 	(void)resultOut;
 	(void)dispatchInfo;
@@ -81,6 +75,3 @@ btScalar btSoftRigidCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject*
 	//not yet
 	return btScalar(1.);
 }
-
-
-
diff --git a/thirdparty/bullet/BulletSoftBody/btSoftRigidCollisionAlgorithm.h b/thirdparty/bullet/BulletSoftBody/btSoftRigidCollisionAlgorithm.h
index 93fcc6065b5..9773af19a0c 100644
--- a/thirdparty/bullet/BulletSoftBody/btSoftRigidCollisionAlgorithm.h
+++ b/thirdparty/bullet/BulletSoftBody/btSoftRigidCollisionAlgorithm.h
@@ -35,41 +35,37 @@ class btSoftRigidCollisionAlgorithm : public btCollisionAlgorithm
 	//btCollisionObject*		m_rigidCollisionObject;
 
 	///for rigid versus soft (instead of soft versus rigid), we use this swapped boolean
-	bool	m_isSwapped;
+	bool m_isSwapped;
 
 public:
-
-	btSoftRigidCollisionAlgorithm(btPersistentManifold* mf,const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* col0,const btCollisionObjectWrapper* col1Wrap, bool isSwapped);
+	btSoftRigidCollisionAlgorithm(btPersistentManifold* mf, const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* col0, const btCollisionObjectWrapper* col1Wrap, bool isSwapped);
 
 	virtual ~btSoftRigidCollisionAlgorithm();
 
-	virtual void processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual void processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual	void	getAllContactManifolds(btManifoldArray&	manifoldArray)
+	virtual void getAllContactManifolds(btManifoldArray& manifoldArray)
 	{
 		//we don't add any manifolds
 	}
 
-
-	struct CreateFunc :public 	btCollisionAlgorithmCreateFunc
+	struct CreateFunc : public btCollisionAlgorithmCreateFunc
 	{
-		virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+		virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 		{
 			void* mem = ci.m_dispatcher1->allocateCollisionAlgorithm(sizeof(btSoftRigidCollisionAlgorithm));
 			if (!m_swapped)
 			{
-				return new(mem) btSoftRigidCollisionAlgorithm(0,ci,body0Wrap,body1Wrap,false);
-			} else
+				return new (mem) btSoftRigidCollisionAlgorithm(0, ci, body0Wrap, body1Wrap, false);
+			}
+			else
 			{
-				return new(mem) btSoftRigidCollisionAlgorithm(0,ci,body0Wrap,body1Wrap,true);
+				return new (mem) btSoftRigidCollisionAlgorithm(0, ci, body0Wrap, body1Wrap, true);
 			}
 		}
 	};
-
 };
 
-#endif //BT_SOFT_RIGID_COLLISION_ALGORITHM_H
-
-
+#endif  //BT_SOFT_RIGID_COLLISION_ALGORITHM_H
diff --git a/thirdparty/bullet/BulletSoftBody/btSoftRigidDynamicsWorld.cpp b/thirdparty/bullet/BulletSoftBody/btSoftRigidDynamicsWorld.cpp
index 204b4f576d5..510b731fc15 100644
--- a/thirdparty/bullet/BulletSoftBody/btSoftRigidDynamicsWorld.cpp
+++ b/thirdparty/bullet/BulletSoftBody/btSoftRigidDynamicsWorld.cpp
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btSoftRigidDynamicsWorld.h"
 #include "LinearMath/btQuickprof.h"
 
@@ -24,42 +23,38 @@ subject to the following restrictions:
 #include "btDefaultSoftBodySolver.h"
 #include "LinearMath/btSerializer.h"
 
-
 btSoftRigidDynamicsWorld::btSoftRigidDynamicsWorld(
 	btDispatcher* dispatcher,
 	btBroadphaseInterface* pairCache,
 	btConstraintSolver* constraintSolver,
 	btCollisionConfiguration* collisionConfiguration,
-	btSoftBodySolver *softBodySolver ) : 
-		btDiscreteDynamicsWorld(dispatcher,pairCache,constraintSolver,collisionConfiguration),
-		m_softBodySolver( softBodySolver ),
-		m_ownsSolver(false)
+	btSoftBodySolver* softBodySolver) : btDiscreteDynamicsWorld(dispatcher, pairCache, constraintSolver, collisionConfiguration),
+										m_softBodySolver(softBodySolver),
+										m_ownsSolver(false)
 {
-	if( !m_softBodySolver )
+	if (!m_softBodySolver)
 	{
-		void* ptr = btAlignedAlloc(sizeof(btDefaultSoftBodySolver),16);
-		m_softBodySolver = new(ptr) btDefaultSoftBodySolver();
+		void* ptr = btAlignedAlloc(sizeof(btDefaultSoftBodySolver), 16);
+		m_softBodySolver = new (ptr) btDefaultSoftBodySolver();
 		m_ownsSolver = true;
 	}
 
-	m_drawFlags			=	fDrawFlags::Std;
-	m_drawNodeTree		=	true;
-	m_drawFaceTree		=	false;
-	m_drawClusterTree	=	false;
+	m_drawFlags = fDrawFlags::Std;
+	m_drawNodeTree = true;
+	m_drawFaceTree = false;
+	m_drawClusterTree = false;
 	m_sbi.m_broadphase = pairCache;
 	m_sbi.m_dispatcher = dispatcher;
 	m_sbi.m_sparsesdf.Initialize();
 	m_sbi.m_sparsesdf.Reset();
 
-	m_sbi.air_density		=	(btScalar)1.2;
-	m_sbi.water_density	=	0;
-	m_sbi.water_offset		=	0;
-	m_sbi.water_normal		=	btVector3(0,0,0);
-	m_sbi.m_gravity.setValue(0,-10,0);
+	m_sbi.air_density = (btScalar)1.2;
+	m_sbi.water_density = 0;
+	m_sbi.water_offset = 0;
+	m_sbi.water_normal = btVector3(0, 0, 0);
+	m_sbi.m_gravity.setValue(0, -10, 0);
 
 	m_sbi.m_sparsesdf.Initialize();
-
-
 }
 
 btSoftRigidDynamicsWorld::~btSoftRigidDynamicsWorld()
@@ -71,82 +66,78 @@ btSoftRigidDynamicsWorld::~btSoftRigidDynamicsWorld()
 	}
 }
 
-void	btSoftRigidDynamicsWorld::predictUnconstraintMotion(btScalar timeStep)
+void btSoftRigidDynamicsWorld::predictUnconstraintMotion(btScalar timeStep)
 {
-	btDiscreteDynamicsWorld::predictUnconstraintMotion( timeStep );
+	btDiscreteDynamicsWorld::predictUnconstraintMotion(timeStep);
 	{
 		BT_PROFILE("predictUnconstraintMotionSoftBody");
-		m_softBodySolver->predictMotion( float(timeStep) );
+		m_softBodySolver->predictMotion(float(timeStep));
 	}
 }
 
-void	btSoftRigidDynamicsWorld::internalSingleStepSimulation( btScalar timeStep )
+void btSoftRigidDynamicsWorld::internalSingleStepSimulation(btScalar timeStep)
 {
-
 	// Let the solver grab the soft bodies and if necessary optimize for it
-	m_softBodySolver->optimize( getSoftBodyArray() );
+	m_softBodySolver->optimize(getSoftBodyArray());
 
-	if( !m_softBodySolver->checkInitialized() )
+	if (!m_softBodySolver->checkInitialized())
 	{
-		btAssert( "Solver initialization failed\n" );
+		btAssert("Solver initialization failed\n");
 	}
 
-	btDiscreteDynamicsWorld::internalSingleStepSimulation( timeStep );
+	btDiscreteDynamicsWorld::internalSingleStepSimulation(timeStep);
 
 	///solve soft bodies constraints
-	solveSoftBodiesConstraints( timeStep );
+	solveSoftBodiesConstraints(timeStep);
 
 	//self collisions
-	for ( int i=0;i<m_softBodies.size();i++)
+	for (int i = 0; i < m_softBodies.size(); i++)
 	{
-		btSoftBody*	psb=(btSoftBody*)m_softBodies[i];
+		btSoftBody* psb = (btSoftBody*)m_softBodies[i];
 		psb->defaultCollisionHandler(psb);
 	}
 
 	///update soft bodies
-	m_softBodySolver->updateSoftBodies( );
-	
+	m_softBodySolver->updateSoftBodies();
+
 	// End solver-wise simulation step
 	// ///////////////////////////////
-
 }
 
-void	btSoftRigidDynamicsWorld::solveSoftBodiesConstraints( btScalar timeStep )
+void btSoftRigidDynamicsWorld::solveSoftBodiesConstraints(btScalar timeStep)
 {
 	BT_PROFILE("solveSoftConstraints");
 
-	if(m_softBodies.size())
+	if (m_softBodies.size())
 	{
 		btSoftBody::solveClusters(m_softBodies);
 	}
 
 	// Solve constraints solver-wise
-	m_softBodySolver->solveConstraints( timeStep * m_softBodySolver->getTimeScale() );
-
+	m_softBodySolver->solveConstraints(timeStep * m_softBodySolver->getTimeScale());
 }
 
-void	btSoftRigidDynamicsWorld::addSoftBody(btSoftBody* body, int collisionFilterGroup, int collisionFilterMask)
+void btSoftRigidDynamicsWorld::addSoftBody(btSoftBody* body, int collisionFilterGroup, int collisionFilterMask)
 {
 	m_softBodies.push_back(body);
 
 	// Set the soft body solver that will deal with this body
 	// to be the world's solver
-	body->setSoftBodySolver( m_softBodySolver );
+	body->setSoftBodySolver(m_softBodySolver);
 
 	btCollisionWorld::addCollisionObject(body,
-		collisionFilterGroup,
-		collisionFilterMask);
-
+										 collisionFilterGroup,
+										 collisionFilterMask);
 }
 
-void	btSoftRigidDynamicsWorld::removeSoftBody(btSoftBody* body)
+void btSoftRigidDynamicsWorld::removeSoftBody(btSoftBody* body)
 {
 	m_softBodies.remove(body);
 
 	btCollisionWorld::removeCollisionObject(body);
 }
 
-void	btSoftRigidDynamicsWorld::removeCollisionObject(btCollisionObject* collisionObject)
+void btSoftRigidDynamicsWorld::removeCollisionObject(btCollisionObject* collisionObject)
 {
 	btSoftBody* body = btSoftBody::upcast(collisionObject);
 	if (body)
@@ -155,60 +146,57 @@ void	btSoftRigidDynamicsWorld::removeCollisionObject(btCollisionObject* collisio
 		btDiscreteDynamicsWorld::removeCollisionObject(collisionObject);
 }
 
-void	btSoftRigidDynamicsWorld::debugDrawWorld()
+void btSoftRigidDynamicsWorld::debugDrawWorld()
 {
 	btDiscreteDynamicsWorld::debugDrawWorld();
 
 	if (getDebugDrawer())
 	{
 		int i;
-		for (  i=0;i<this->m_softBodies.size();i++)
+		for (i = 0; i < this->m_softBodies.size(); i++)
 		{
-			btSoftBody*	psb=(btSoftBody*)this->m_softBodies[i];
+			btSoftBody* psb = (btSoftBody*)this->m_softBodies[i];
 			if (getDebugDrawer() && (getDebugDrawer()->getDebugMode() & (btIDebugDraw::DBG_DrawWireframe)))
 			{
-				btSoftBodyHelpers::DrawFrame(psb,m_debugDrawer);
-				btSoftBodyHelpers::Draw(psb,m_debugDrawer,m_drawFlags);
+				btSoftBodyHelpers::DrawFrame(psb, m_debugDrawer);
+				btSoftBodyHelpers::Draw(psb, m_debugDrawer, m_drawFlags);
 			}
-			
+
 			if (m_debugDrawer && (m_debugDrawer->getDebugMode() & btIDebugDraw::DBG_DrawAabb))
 			{
-				if(m_drawNodeTree)		btSoftBodyHelpers::DrawNodeTree(psb,m_debugDrawer);
-				if(m_drawFaceTree)		btSoftBodyHelpers::DrawFaceTree(psb,m_debugDrawer);
-				if(m_drawClusterTree)	btSoftBodyHelpers::DrawClusterTree(psb,m_debugDrawer);
+				if (m_drawNodeTree) btSoftBodyHelpers::DrawNodeTree(psb, m_debugDrawer);
+				if (m_drawFaceTree) btSoftBodyHelpers::DrawFaceTree(psb, m_debugDrawer);
+				if (m_drawClusterTree) btSoftBodyHelpers::DrawClusterTree(psb, m_debugDrawer);
 			}
-		}		
-	}	
+		}
+	}
 }
 
-
-
-
 struct btSoftSingleRayCallback : public btBroadphaseRayCallback
 {
-	btVector3	m_rayFromWorld;
-	btVector3	m_rayToWorld;
-	btTransform	m_rayFromTrans;
-	btTransform	m_rayToTrans;
-	btVector3	m_hitNormal;
+	btVector3 m_rayFromWorld;
+	btVector3 m_rayToWorld;
+	btTransform m_rayFromTrans;
+	btTransform m_rayToTrans;
+	btVector3 m_hitNormal;
 
-	const btSoftRigidDynamicsWorld*	m_world;
-	btCollisionWorld::RayResultCallback&	m_resultCallback;
+	const btSoftRigidDynamicsWorld* m_world;
+	btCollisionWorld::RayResultCallback& m_resultCallback;
 
-	btSoftSingleRayCallback(const btVector3& rayFromWorld,const btVector3& rayToWorld,const btSoftRigidDynamicsWorld* world,btCollisionWorld::RayResultCallback& resultCallback)
-	:m_rayFromWorld(rayFromWorld),
-	m_rayToWorld(rayToWorld),
-	m_world(world),
-	m_resultCallback(resultCallback)
+	btSoftSingleRayCallback(const btVector3& rayFromWorld, const btVector3& rayToWorld, const btSoftRigidDynamicsWorld* world, btCollisionWorld::RayResultCallback& resultCallback)
+		: m_rayFromWorld(rayFromWorld),
+		  m_rayToWorld(rayToWorld),
+		  m_world(world),
+		  m_resultCallback(resultCallback)
 	{
 		m_rayFromTrans.setIdentity();
 		m_rayFromTrans.setOrigin(m_rayFromWorld);
 		m_rayToTrans.setIdentity();
 		m_rayToTrans.setOrigin(m_rayToWorld);
 
-		btVector3 rayDir = (rayToWorld-rayFromWorld);
+		btVector3 rayDir = (rayToWorld - rayFromWorld);
 
-		rayDir.normalize ();
+		rayDir.normalize();
 		///what about division by zero? --> just set rayDirection[i] to INF/1e30
 		m_rayDirectionInverse[0] = rayDir[0] == btScalar(0.0) ? btScalar(1e30) : btScalar(1.0) / rayDir[0];
 		m_rayDirectionInverse[1] = rayDir[1] == btScalar(0.0) ? btScalar(1e30) : btScalar(1.0) / rayDir[1];
@@ -217,22 +205,19 @@ struct btSoftSingleRayCallback : public btBroadphaseRayCallback
 		m_signs[1] = m_rayDirectionInverse[1] < 0.0;
 		m_signs[2] = m_rayDirectionInverse[2] < 0.0;
 
-		m_lambda_max = rayDir.dot(m_rayToWorld-m_rayFromWorld);
-
+		m_lambda_max = rayDir.dot(m_rayToWorld - m_rayFromWorld);
 	}
 
-	
-
-	virtual bool	process(const btBroadphaseProxy* proxy)
+	virtual bool process(const btBroadphaseProxy* proxy)
 	{
 		///terminate further ray tests, once the closestHitFraction reached zero
 		if (m_resultCallback.m_closestHitFraction == btScalar(0.f))
 			return false;
 
-		btCollisionObject*	collisionObject = (btCollisionObject*)proxy->m_clientObject;
+		btCollisionObject* collisionObject = (btCollisionObject*)proxy->m_clientObject;
 
 		//only perform raycast if filterMask matches
-		if(m_resultCallback.needsCollision(collisionObject->getBroadphaseHandle())) 
+		if (m_resultCallback.needsCollision(collisionObject->getBroadphaseHandle()))
 		{
 			//RigidcollisionObject* collisionObject = ctrl->GetRigidcollisionObject();
 			//btVector3 collisionObjectAabbMin,collisionObjectAabbMax;
@@ -250,110 +235,106 @@ struct btSoftSingleRayCallback : public btBroadphaseRayCallback
 			//culling already done by broadphase
 			//if (btRayAabb(m_rayFromWorld,m_rayToWorld,collisionObjectAabbMin,collisionObjectAabbMax,hitLambda,m_hitNormal))
 			{
-				m_world->rayTestSingle(m_rayFromTrans,m_rayToTrans,
-					collisionObject,
-						collisionObject->getCollisionShape(),
-						collisionObject->getWorldTransform(),
-						m_resultCallback);
+				m_world->rayTestSingle(m_rayFromTrans, m_rayToTrans,
+									   collisionObject,
+									   collisionObject->getCollisionShape(),
+									   collisionObject->getWorldTransform(),
+									   m_resultCallback);
 			}
 		}
 		return true;
 	}
 };
 
-void	btSoftRigidDynamicsWorld::rayTest(const btVector3& rayFromWorld, const btVector3& rayToWorld, RayResultCallback& resultCallback) const
+void btSoftRigidDynamicsWorld::rayTest(const btVector3& rayFromWorld, const btVector3& rayToWorld, RayResultCallback& resultCallback) const
 {
 	BT_PROFILE("rayTest");
 	/// use the broadphase to accelerate the search for objects, based on their aabb
 	/// and for each object with ray-aabb overlap, perform an exact ray test
-	btSoftSingleRayCallback rayCB(rayFromWorld,rayToWorld,this,resultCallback);
+	btSoftSingleRayCallback rayCB(rayFromWorld, rayToWorld, this, resultCallback);
 
 #ifndef USE_BRUTEFORCE_RAYBROADPHASE
-	m_broadphasePairCache->rayTest(rayFromWorld,rayToWorld,rayCB);
+	m_broadphasePairCache->rayTest(rayFromWorld, rayToWorld, rayCB);
 #else
-	for (int i=0;i<this->getNumCollisionObjects();i++)
+	for (int i = 0; i < this->getNumCollisionObjects(); i++)
 	{
 		rayCB.process(m_collisionObjects[i]->getBroadphaseHandle());
-	}	
-#endif //USE_BRUTEFORCE_RAYBROADPHASE
-
+	}
+#endif  //USE_BRUTEFORCE_RAYBROADPHASE
 }
 
-
-void	btSoftRigidDynamicsWorld::rayTestSingle(const btTransform& rayFromTrans,const btTransform& rayToTrans,
-					  btCollisionObject* collisionObject,
-					  const btCollisionShape* collisionShape,
-					  const btTransform& colObjWorldTransform,
-					  RayResultCallback& resultCallback)
+void btSoftRigidDynamicsWorld::rayTestSingle(const btTransform& rayFromTrans, const btTransform& rayToTrans,
+											 btCollisionObject* collisionObject,
+											 const btCollisionShape* collisionShape,
+											 const btTransform& colObjWorldTransform,
+											 RayResultCallback& resultCallback)
 {
-	if (collisionShape->isSoftBody()) {
+	if (collisionShape->isSoftBody())
+	{
 		btSoftBody* softBody = btSoftBody::upcast(collisionObject);
-		if (softBody) {
+		if (softBody)
+		{
 			btSoftBody::sRayCast softResult;
-			if (softBody->rayTest(rayFromTrans.getOrigin(), rayToTrans.getOrigin(), softResult)) 
+			if (softBody->rayTest(rayFromTrans.getOrigin(), rayToTrans.getOrigin(), softResult))
 			{
-				
-				if (softResult.fraction<= resultCallback.m_closestHitFraction)
+				if (softResult.fraction <= resultCallback.m_closestHitFraction)
 				{
-
 					btCollisionWorld::LocalShapeInfo shapeInfo;
 					shapeInfo.m_shapePart = 0;
 					shapeInfo.m_triangleIndex = softResult.index;
 					// get the normal
 					btVector3 rayDir = rayToTrans.getOrigin() - rayFromTrans.getOrigin();
-					btVector3 normal=-rayDir;
+					btVector3 normal = -rayDir;
 					normal.normalize();
 
 					if (softResult.feature == btSoftBody::eFeature::Face)
 					{
 						normal = softBody->m_faces[softResult.index].m_normal;
-						if (normal.dot(rayDir) > 0) {
+						if (normal.dot(rayDir) > 0)
+						{
 							// normal always point toward origin of the ray
 							normal = -normal;
 						}
 					}
-	
-					btCollisionWorld::LocalRayResult rayResult
-						(collisionObject,
-						 &shapeInfo,
-						 normal,
-						 softResult.fraction);
-					bool	normalInWorldSpace = true;
-					resultCallback.addSingleResult(rayResult,normalInWorldSpace);
+
+					btCollisionWorld::LocalRayResult rayResult(collisionObject,
+															   &shapeInfo,
+															   normal,
+															   softResult.fraction);
+					bool normalInWorldSpace = true;
+					resultCallback.addSingleResult(rayResult, normalInWorldSpace);
 				}
 			}
 		}
-	} 
-	else {
-		btCollisionWorld::rayTestSingle(rayFromTrans,rayToTrans,collisionObject,collisionShape,colObjWorldTransform,resultCallback);
+	}
+	else
+	{
+		btCollisionWorld::rayTestSingle(rayFromTrans, rayToTrans, collisionObject, collisionShape, colObjWorldTransform, resultCallback);
 	}
 }
 
-
-void	btSoftRigidDynamicsWorld::serializeSoftBodies(btSerializer* serializer)
+void btSoftRigidDynamicsWorld::serializeSoftBodies(btSerializer* serializer)
 {
 	int i;
 	//serialize all collision objects
-	for (i=0;i<m_collisionObjects.size();i++)
+	for (i = 0; i < m_collisionObjects.size(); i++)
 	{
 		btCollisionObject* colObj = m_collisionObjects[i];
 		if (colObj->getInternalType() & btCollisionObject::CO_SOFT_BODY)
 		{
 			int len = colObj->calculateSerializeBufferSize();
-			btChunk* chunk = serializer->allocate(len,1);
+			btChunk* chunk = serializer->allocate(len, 1);
 			const char* structType = colObj->serialize(chunk->m_oldPtr, serializer);
-			serializer->finalizeChunk(chunk,structType,BT_SOFTBODY_CODE,colObj);
+			serializer->finalizeChunk(chunk, structType, BT_SOFTBODY_CODE, colObj);
 		}
 	}
-
 }
 
-void	btSoftRigidDynamicsWorld::serialize(btSerializer* serializer)
+void btSoftRigidDynamicsWorld::serialize(btSerializer* serializer)
 {
-
 	serializer->startSerialization();
 
-	serializeDynamicsWorldInfo( serializer);
+	serializeDynamicsWorldInfo(serializer);
 
 	serializeSoftBodies(serializer);
 
@@ -363,5 +344,3 @@ void	btSoftRigidDynamicsWorld::serialize(btSerializer* serializer)
 
 	serializer->finishSerialization();
 }
-
-
diff --git a/thirdparty/bullet/BulletSoftBody/btSoftRigidDynamicsWorld.h b/thirdparty/bullet/BulletSoftBody/btSoftRigidDynamicsWorld.h
index d921a6488d3..be49c444d7d 100644
--- a/thirdparty/bullet/BulletSoftBody/btSoftRigidDynamicsWorld.h
+++ b/thirdparty/bullet/BulletSoftBody/btSoftRigidDynamicsWorld.h
@@ -19,63 +19,60 @@ subject to the following restrictions:
 #include "BulletDynamics/Dynamics/btDiscreteDynamicsWorld.h"
 #include "btSoftBody.h"
 
-typedef	btAlignedObjectArray<btSoftBody*> btSoftBodyArray;
+typedef btAlignedObjectArray<btSoftBody*> btSoftBodyArray;
 
 class btSoftBodySolver;
 
 class btSoftRigidDynamicsWorld : public btDiscreteDynamicsWorld
 {
-
-	btSoftBodyArray	m_softBodies;
-	int				m_drawFlags;
-	bool			m_drawNodeTree;
-	bool			m_drawFaceTree;
-	bool			m_drawClusterTree;
+	btSoftBodyArray m_softBodies;
+	int m_drawFlags;
+	bool m_drawNodeTree;
+	bool m_drawFaceTree;
+	bool m_drawClusterTree;
 	btSoftBodyWorldInfo m_sbi;
 	///Solver classes that encapsulate multiple soft bodies for solving
-	btSoftBodySolver *m_softBodySolver;
-	bool			m_ownsSolver;
+	btSoftBodySolver* m_softBodySolver;
+	bool m_ownsSolver;
 
 protected:
+	virtual void predictUnconstraintMotion(btScalar timeStep);
 
-	virtual void	predictUnconstraintMotion(btScalar timeStep);
+	virtual void internalSingleStepSimulation(btScalar timeStep);
 
-	virtual void	internalSingleStepSimulation( btScalar timeStep);
+	void solveSoftBodiesConstraints(btScalar timeStep);
 
-	void	solveSoftBodiesConstraints( btScalar timeStep );
-
-	void	serializeSoftBodies(btSerializer* serializer);
+	void serializeSoftBodies(btSerializer* serializer);
 
 public:
-
-	btSoftRigidDynamicsWorld(btDispatcher* dispatcher,btBroadphaseInterface* pairCache,btConstraintSolver* constraintSolver, btCollisionConfiguration* collisionConfiguration, btSoftBodySolver *softBodySolver = 0 );
+	btSoftRigidDynamicsWorld(btDispatcher* dispatcher, btBroadphaseInterface* pairCache, btConstraintSolver* constraintSolver, btCollisionConfiguration* collisionConfiguration, btSoftBodySolver* softBodySolver = 0);
 
 	virtual ~btSoftRigidDynamicsWorld();
 
-	virtual void	debugDrawWorld();
+	virtual void debugDrawWorld();
 
-	void	addSoftBody(btSoftBody* body, int collisionFilterGroup=btBroadphaseProxy::DefaultFilter, int collisionFilterMask=btBroadphaseProxy::AllFilter);
+	void addSoftBody(btSoftBody* body, int collisionFilterGroup = btBroadphaseProxy::DefaultFilter, int collisionFilterMask = btBroadphaseProxy::AllFilter);
 
-	void	removeSoftBody(btSoftBody* body);
+	void removeSoftBody(btSoftBody* body);
 
 	///removeCollisionObject will first check if it is a rigid body, if so call removeRigidBody otherwise call btDiscreteDynamicsWorld::removeCollisionObject
-	virtual void	removeCollisionObject(btCollisionObject* collisionObject);
+	virtual void removeCollisionObject(btCollisionObject* collisionObject);
 
-	int		getDrawFlags() const { return(m_drawFlags); }
-	void	setDrawFlags(int f)	{ m_drawFlags=f; }
+	int getDrawFlags() const { return (m_drawFlags); }
+	void setDrawFlags(int f) { m_drawFlags = f; }
 
-	btSoftBodyWorldInfo&	getWorldInfo()
+	btSoftBodyWorldInfo& getWorldInfo()
 	{
 		return m_sbi;
 	}
-	const btSoftBodyWorldInfo&	getWorldInfo() const
+	const btSoftBodyWorldInfo& getWorldInfo() const
 	{
 		return m_sbi;
 	}
 
-	virtual btDynamicsWorldType	getWorldType() const
+	virtual btDynamicsWorldType getWorldType() const
 	{
-		return	BT_SOFT_RIGID_DYNAMICS_WORLD;
+		return BT_SOFT_RIGID_DYNAMICS_WORLD;
 	}
 
 	btSoftBodyArray& getSoftBodyArray()
@@ -88,20 +85,18 @@ public:
 		return m_softBodies;
 	}
 
-
-	virtual void rayTest(const btVector3& rayFromWorld, const btVector3& rayToWorld, RayResultCallback& resultCallback) const; 
+	virtual void rayTest(const btVector3& rayFromWorld, const btVector3& rayToWorld, RayResultCallback& resultCallback) const;
 
 	/// rayTestSingle performs a raycast call and calls the resultCallback. It is used internally by rayTest.
 	/// In a future implementation, we consider moving the ray test as a virtual method in btCollisionShape.
 	/// This allows more customization.
-	static void	rayTestSingle(const btTransform& rayFromTrans,const btTransform& rayToTrans,
-					  btCollisionObject* collisionObject,
-					  const btCollisionShape* collisionShape,
-					  const btTransform& colObjWorldTransform,
-					  RayResultCallback& resultCallback);
-
-	virtual	void	serialize(btSerializer* serializer);
+	static void rayTestSingle(const btTransform& rayFromTrans, const btTransform& rayToTrans,
+							  btCollisionObject* collisionObject,
+							  const btCollisionShape* collisionShape,
+							  const btTransform& colObjWorldTransform,
+							  RayResultCallback& resultCallback);
 
+	virtual void serialize(btSerializer* serializer);
 };
 
-#endif //BT_SOFT_RIGID_DYNAMICS_WORLD_H
+#endif  //BT_SOFT_RIGID_DYNAMICS_WORLD_H
diff --git a/thirdparty/bullet/BulletSoftBody/btSoftSoftCollisionAlgorithm.cpp b/thirdparty/bullet/BulletSoftBody/btSoftSoftCollisionAlgorithm.cpp
index 72043e69e2f..9c3e904f645 100644
--- a/thirdparty/bullet/BulletSoftBody/btSoftSoftCollisionAlgorithm.cpp
+++ b/thirdparty/bullet/BulletSoftBody/btSoftSoftCollisionAlgorithm.cpp
@@ -23,8 +23,8 @@ subject to the following restrictions:
 
 #define USE_PERSISTENT_CONTACTS 1
 
-btSoftSoftCollisionAlgorithm::btSoftSoftCollisionAlgorithm(btPersistentManifold* /*mf*/,const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* /*obj0*/,const btCollisionObjectWrapper* /*obj1*/)
-: btCollisionAlgorithm(ci)
+btSoftSoftCollisionAlgorithm::btSoftSoftCollisionAlgorithm(btPersistentManifold* /*mf*/, const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* /*obj0*/, const btCollisionObjectWrapper* /*obj1*/)
+	: btCollisionAlgorithm(ci)
 //m_ownManifold(false),
 //m_manifoldPtr(mf)
 {
@@ -34,14 +34,14 @@ btSoftSoftCollisionAlgorithm::~btSoftSoftCollisionAlgorithm()
 {
 }
 
-void btSoftSoftCollisionAlgorithm::processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& /*dispatchInfo*/,btManifoldResult* /*resultOut*/)
+void btSoftSoftCollisionAlgorithm::processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& /*dispatchInfo*/, btManifoldResult* /*resultOut*/)
 {
-	btSoftBody* soft0 =	(btSoftBody*)body0Wrap->getCollisionObject();
-	btSoftBody* soft1 =	(btSoftBody*)body1Wrap->getCollisionObject();
+	btSoftBody* soft0 = (btSoftBody*)body0Wrap->getCollisionObject();
+	btSoftBody* soft1 = (btSoftBody*)body1Wrap->getCollisionObject();
 	soft0->getSoftBodySolver()->processCollision(soft0, soft1);
 }
 
-btScalar btSoftSoftCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* /*body0*/,btCollisionObject* /*body1*/,const btDispatcherInfo& /*dispatchInfo*/,btManifoldResult* /*resultOut*/)
+btScalar btSoftSoftCollisionAlgorithm::calculateTimeOfImpact(btCollisionObject* /*body0*/, btCollisionObject* /*body1*/, const btDispatcherInfo& /*dispatchInfo*/, btManifoldResult* /*resultOut*/)
 {
 	//not yet
 	return 1.f;
diff --git a/thirdparty/bullet/BulletSoftBody/btSoftSoftCollisionAlgorithm.h b/thirdparty/bullet/BulletSoftBody/btSoftSoftCollisionAlgorithm.h
index 4eab7aea2fe..6f871f5b855 100644
--- a/thirdparty/bullet/BulletSoftBody/btSoftSoftCollisionAlgorithm.h
+++ b/thirdparty/bullet/BulletSoftBody/btSoftSoftCollisionAlgorithm.h
@@ -27,43 +27,39 @@ class btSoftBody;
 ///collision detection between two btSoftBody shapes
 class btSoftSoftCollisionAlgorithm : public btCollisionAlgorithm
 {
-	bool	m_ownManifold;
-	btPersistentManifold*	m_manifoldPtr;
-
-//	btSoftBody*	m_softBody0;
-//	btSoftBody*	m_softBody1;
+	bool m_ownManifold;
+	btPersistentManifold* m_manifoldPtr;
 
+	//	btSoftBody*	m_softBody0;
+	//	btSoftBody*	m_softBody1;
 
 public:
 	btSoftSoftCollisionAlgorithm(const btCollisionAlgorithmConstructionInfo& ci)
 		: btCollisionAlgorithm(ci) {}
 
-	virtual void processCollision (const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual void processCollision(const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0,btCollisionObject* body1,const btDispatcherInfo& dispatchInfo,btManifoldResult* resultOut);
+	virtual btScalar calculateTimeOfImpact(btCollisionObject* body0, btCollisionObject* body1, const btDispatcherInfo& dispatchInfo, btManifoldResult* resultOut);
 
-	virtual	void	getAllContactManifolds(btManifoldArray&	manifoldArray)
+	virtual void getAllContactManifolds(btManifoldArray& manifoldArray)
 	{
 		if (m_manifoldPtr && m_ownManifold)
 			manifoldArray.push_back(m_manifoldPtr);
 	}
 
-	btSoftSoftCollisionAlgorithm(btPersistentManifold* mf,const btCollisionAlgorithmConstructionInfo& ci,const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap);
+	btSoftSoftCollisionAlgorithm(btPersistentManifold* mf, const btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap);
 
 	virtual ~btSoftSoftCollisionAlgorithm();
 
-	struct CreateFunc :public 	btCollisionAlgorithmCreateFunc
+	struct CreateFunc : public btCollisionAlgorithmCreateFunc
 	{
-		virtual	btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap,const btCollisionObjectWrapper* body1Wrap)
+		virtual btCollisionAlgorithm* CreateCollisionAlgorithm(btCollisionAlgorithmConstructionInfo& ci, const btCollisionObjectWrapper* body0Wrap, const btCollisionObjectWrapper* body1Wrap)
 		{
 			int bbsize = sizeof(btSoftSoftCollisionAlgorithm);
 			void* ptr = ci.m_dispatcher1->allocateCollisionAlgorithm(bbsize);
-			return new(ptr) btSoftSoftCollisionAlgorithm(0,ci,body0Wrap,body1Wrap);
+			return new (ptr) btSoftSoftCollisionAlgorithm(0, ci, body0Wrap, body1Wrap);
 		}
 	};
-
 };
 
-#endif //BT_SOFT_SOFT_COLLISION_ALGORITHM_H
-
-
+#endif  //BT_SOFT_SOFT_COLLISION_ALGORITHM_H
diff --git a/thirdparty/bullet/BulletSoftBody/btSparseSDF.h b/thirdparty/bullet/BulletSoftBody/btSparseSDF.h
index ba437c28efc..a52b2cb1cc7 100644
--- a/thirdparty/bullet/BulletSoftBody/btSparseSDF.h
+++ b/thirdparty/bullet/BulletSoftBody/btSparseSDF.h
@@ -24,296 +24,320 @@ subject to the following restrictions:
 template <const int DWORDLEN>
 unsigned int HsiehHash(const void* pdata)
 {
-	const unsigned short*	data=(const unsigned short*)pdata;
-	unsigned				hash=DWORDLEN<<2,tmp;
-	for(int i=0;i<DWORDLEN;++i)
+	const unsigned short* data = (const unsigned short*)pdata;
+	unsigned hash = DWORDLEN << 2, tmp;
+	for (int i = 0; i < DWORDLEN; ++i)
 	{
-		hash	+=	data[0];
-		tmp		=	(data[1]<<11)^hash;
-		hash	=	(hash<<16)^tmp;
-		data	+=	2;
-		hash	+=	hash>>11;
+		hash += data[0];
+		tmp = (data[1] << 11) ^ hash;
+		hash = (hash << 16) ^ tmp;
+		data += 2;
+		hash += hash >> 11;
 	}
-	hash^=hash<<3;hash+=hash>>5;
-	hash^=hash<<4;hash+=hash>>17;
-	hash^=hash<<25;hash+=hash>>6;
-	return(hash);
+	hash ^= hash << 3;
+	hash += hash >> 5;
+	hash ^= hash << 4;
+	hash += hash >> 17;
+	hash ^= hash << 25;
+	hash += hash >> 6;
+	return (hash);
 }
 
 template <const int CELLSIZE>
-struct	btSparseSdf
+struct btSparseSdf
 {
 	//
 	// Inner types
 	//
 	struct IntFrac
 	{
-		int					b;
-		int					i;
-		btScalar			f;
+		int b;
+		int i;
+		btScalar f;
 	};
-	struct	Cell
+	struct Cell
 	{
-		btScalar			d[CELLSIZE+1][CELLSIZE+1][CELLSIZE+1];
-		int					c[3];
-		int					puid;
-		unsigned			hash;
-		const btCollisionShape*	pclient;
-		Cell*				next;
+		btScalar d[CELLSIZE + 1][CELLSIZE + 1][CELLSIZE + 1];
+		int c[3];
+		int puid;
+		unsigned hash;
+		const btCollisionShape* pclient;
+		Cell* next;
 	};
 	//
 	// Fields
 	//
 
-	btAlignedObjectArray<Cell*>		cells;	
-	btScalar						voxelsz;
-	int								puid;
-	int								ncells;
-	int								m_clampCells;
-	int								nprobes;
-	int								nqueries;	
+	btAlignedObjectArray<Cell*> cells;
+	btScalar voxelsz;
+	int puid;
+	int ncells;
+	int m_clampCells;
+	int nprobes;
+	int nqueries;
 
 	//
 	// Methods
 	//
 
 	//
-	void					Initialize(int hashsize=2383, int clampCells = 256*1024)
+	void Initialize(int hashsize = 2383, int clampCells = 256 * 1024)
 	{
 		//avoid a crash due to running out of memory, so clamp the maximum number of cells allocated
 		//if this limit is reached, the SDF is reset (at the cost of some performance during the reset)
 		m_clampCells = clampCells;
-		cells.resize(hashsize,0);
+		cells.resize(hashsize, 0);
 		Reset();
 	}
 	//
-	void					Reset()
+	void Reset()
 	{
-		for(int i=0,ni=cells.size();i<ni;++i)
+		for (int i = 0, ni = cells.size(); i < ni; ++i)
 		{
-			Cell*	pc=cells[i];
-			cells[i]=0;
-			while(pc)
+			Cell* pc = cells[i];
+			cells[i] = 0;
+			while (pc)
 			{
-				Cell*	pn=pc->next;
+				Cell* pn = pc->next;
 				delete pc;
-				pc=pn;
+				pc = pn;
 			}
 		}
-		voxelsz		=0.25;
-		puid		=0;
-		ncells		=0;
-		nprobes		=1;
-		nqueries	=1;
+		voxelsz = 0.25;
+		puid = 0;
+		ncells = 0;
+		nprobes = 1;
+		nqueries = 1;
 	}
 	//
-	void					GarbageCollect(int lifetime=256)
+	void GarbageCollect(int lifetime = 256)
 	{
-		const int life=puid-lifetime;
-		for(int i=0;i<cells.size();++i)
+		const int life = puid - lifetime;
+		for (int i = 0; i < cells.size(); ++i)
 		{
-			Cell*&	root=cells[i];
-			Cell*	pp=0;
-			Cell*	pc=root;
-			while(pc)
+			Cell*& root = cells[i];
+			Cell* pp = 0;
+			Cell* pc = root;
+			while (pc)
 			{
-				Cell*	pn=pc->next;
-				if(pc->puid<life)
+				Cell* pn = pc->next;
+				if (pc->puid < life)
 				{
-					if(pp) pp->next=pn; else root=pn;
-					delete pc;pc=pp;--ncells;
+					if (pp)
+						pp->next = pn;
+					else
+						root = pn;
+					delete pc;
+					pc = pp;
+					--ncells;
 				}
-				pp=pc;pc=pn;
+				pp = pc;
+				pc = pn;
 			}
 		}
 		//printf("GC[%d]: %d cells, PpQ: %f\r\n",puid,ncells,nprobes/(btScalar)nqueries);
-		nqueries=1;
-		nprobes=1;
-		++puid;	///@todo: Reset puid's when int range limit is reached	*/ 
-		/* else setup a priority list...						*/ 
+		nqueries = 1;
+		nprobes = 1;
+		++puid;  ///@todo: Reset puid's when int range limit is reached	*/
+				 /* else setup a priority list...						*/
 	}
 	//
-	int						RemoveReferences(btCollisionShape* pcs)
+	int RemoveReferences(btCollisionShape* pcs)
 	{
-		int	refcount=0;
-		for(int i=0;i<cells.size();++i)
+		int refcount = 0;
+		for (int i = 0; i < cells.size(); ++i)
 		{
-			Cell*&	root=cells[i];
-			Cell*	pp=0;
-			Cell*	pc=root;
-			while(pc)
+			Cell*& root = cells[i];
+			Cell* pp = 0;
+			Cell* pc = root;
+			while (pc)
 			{
-				Cell*	pn=pc->next;
-				if(pc->pclient==pcs)
+				Cell* pn = pc->next;
+				if (pc->pclient == pcs)
 				{
-					if(pp) pp->next=pn; else root=pn;
-					delete pc;pc=pp;++refcount;
+					if (pp)
+						pp->next = pn;
+					else
+						root = pn;
+					delete pc;
+					pc = pp;
+					++refcount;
 				}
-				pp=pc;pc=pn;
+				pp = pc;
+				pc = pn;
 			}
 		}
-		return(refcount);
+		return (refcount);
 	}
 	//
-	btScalar				Evaluate(	const btVector3& x,
-		const btCollisionShape* shape,
-		btVector3& normal,
-		btScalar margin)
+	btScalar Evaluate(const btVector3& x,
+					  const btCollisionShape* shape,
+					  btVector3& normal,
+					  btScalar margin)
 	{
-		/* Lookup cell			*/ 
-		const btVector3	scx=x/voxelsz;
-		const IntFrac	ix=Decompose(scx.x());
-		const IntFrac	iy=Decompose(scx.y());
-		const IntFrac	iz=Decompose(scx.z());
-		const unsigned	h=Hash(ix.b,iy.b,iz.b,shape);
-		Cell*&			root=cells[static_cast<int>(h%cells.size())];
-		Cell*			c=root;
+		/* Lookup cell			*/
+		const btVector3 scx = x / voxelsz;
+		const IntFrac ix = Decompose(scx.x());
+		const IntFrac iy = Decompose(scx.y());
+		const IntFrac iz = Decompose(scx.z());
+		const unsigned h = Hash(ix.b, iy.b, iz.b, shape);
+		Cell*& root = cells[static_cast<int>(h % cells.size())];
+		Cell* c = root;
 		++nqueries;
-		while(c)
+		while (c)
 		{
 			++nprobes;
-			if(	(c->hash==h)	&&
-				(c->c[0]==ix.b)	&&
-				(c->c[1]==iy.b)	&&
-				(c->c[2]==iz.b)	&&
-				(c->pclient==shape))
-			{ break; }
+			if ((c->hash == h) &&
+				(c->c[0] == ix.b) &&
+				(c->c[1] == iy.b) &&
+				(c->c[2] == iz.b) &&
+				(c->pclient == shape))
+			{
+				break;
+			}
 			else
-			{ c=c->next; }
+			{
+				c = c->next;
+			}
 		}
-		if(!c)
+		if (!c)
 		{
-			++nprobes;		
+			++nprobes;
 			++ncells;
 			//int sz = sizeof(Cell);
-			if (ncells>m_clampCells)
+			if (ncells > m_clampCells)
 			{
-				static int numResets=0;
+				static int numResets = 0;
 				numResets++;
-//				printf("numResets=%d\n",numResets);
+				//				printf("numResets=%d\n",numResets);
 				Reset();
 			}
 
-			c=new Cell();
-			c->next=root;root=c;
-			c->pclient=shape;
-			c->hash=h;
-			c->c[0]=ix.b;c->c[1]=iy.b;c->c[2]=iz.b;
+			c = new Cell();
+			c->next = root;
+			root = c;
+			c->pclient = shape;
+			c->hash = h;
+			c->c[0] = ix.b;
+			c->c[1] = iy.b;
+			c->c[2] = iz.b;
 			BuildCell(*c);
 		}
-		c->puid=puid;
-		/* Extract infos		*/ 
-		const int		o[]={	ix.i,iy.i,iz.i};
-		const btScalar	d[]={	c->d[o[0]+0][o[1]+0][o[2]+0],
-			c->d[o[0]+1][o[1]+0][o[2]+0],
-			c->d[o[0]+1][o[1]+1][o[2]+0],
-			c->d[o[0]+0][o[1]+1][o[2]+0],
-			c->d[o[0]+0][o[1]+0][o[2]+1],
-			c->d[o[0]+1][o[1]+0][o[2]+1],
-			c->d[o[0]+1][o[1]+1][o[2]+1],
-			c->d[o[0]+0][o[1]+1][o[2]+1]};
-		/* Normal	*/ 
+		c->puid = puid;
+		/* Extract infos		*/
+		const int o[] = {ix.i, iy.i, iz.i};
+		const btScalar d[] = {c->d[o[0] + 0][o[1] + 0][o[2] + 0],
+							  c->d[o[0] + 1][o[1] + 0][o[2] + 0],
+							  c->d[o[0] + 1][o[1] + 1][o[2] + 0],
+							  c->d[o[0] + 0][o[1] + 1][o[2] + 0],
+							  c->d[o[0] + 0][o[1] + 0][o[2] + 1],
+							  c->d[o[0] + 1][o[1] + 0][o[2] + 1],
+							  c->d[o[0] + 1][o[1] + 1][o[2] + 1],
+							  c->d[o[0] + 0][o[1] + 1][o[2] + 1]};
+		/* Normal	*/
 #if 1
-		const btScalar	gx[]={	d[1]-d[0],d[2]-d[3],
-			d[5]-d[4],d[6]-d[7]};
-		const btScalar	gy[]={	d[3]-d[0],d[2]-d[1],
-			d[7]-d[4],d[6]-d[5]};
-		const btScalar	gz[]={	d[4]-d[0],d[5]-d[1],
-			d[7]-d[3],d[6]-d[2]};
-		normal.setX(Lerp(	Lerp(gx[0],gx[1],iy.f),
-			Lerp(gx[2],gx[3],iy.f),iz.f));
-		normal.setY(Lerp(	Lerp(gy[0],gy[1],ix.f),
-			Lerp(gy[2],gy[3],ix.f),iz.f));
-		normal.setZ(Lerp(	Lerp(gz[0],gz[1],ix.f),
-			Lerp(gz[2],gz[3],ix.f),iy.f));
-		normal		=	normal.normalized();
+		const btScalar gx[] = {d[1] - d[0], d[2] - d[3],
+							   d[5] - d[4], d[6] - d[7]};
+		const btScalar gy[] = {d[3] - d[0], d[2] - d[1],
+							   d[7] - d[4], d[6] - d[5]};
+		const btScalar gz[] = {d[4] - d[0], d[5] - d[1],
+							   d[7] - d[3], d[6] - d[2]};
+		normal.setX(Lerp(Lerp(gx[0], gx[1], iy.f),
+						 Lerp(gx[2], gx[3], iy.f), iz.f));
+		normal.setY(Lerp(Lerp(gy[0], gy[1], ix.f),
+						 Lerp(gy[2], gy[3], ix.f), iz.f));
+		normal.setZ(Lerp(Lerp(gz[0], gz[1], ix.f),
+						 Lerp(gz[2], gz[3], ix.f), iy.f));
+		normal = normal.normalized();
 #else
-		normal		=	btVector3(d[1]-d[0],d[3]-d[0],d[4]-d[0]).normalized();
+		normal = btVector3(d[1] - d[0], d[3] - d[0], d[4] - d[0]).normalized();
 #endif
-		/* Distance	*/ 
-		const btScalar	d0=Lerp(Lerp(d[0],d[1],ix.f),
-			Lerp(d[3],d[2],ix.f),iy.f);
-		const btScalar	d1=Lerp(Lerp(d[4],d[5],ix.f),
-			Lerp(d[7],d[6],ix.f),iy.f);
-		return(Lerp(d0,d1,iz.f)-margin);
+		/* Distance	*/
+		const btScalar d0 = Lerp(Lerp(d[0], d[1], ix.f),
+								 Lerp(d[3], d[2], ix.f), iy.f);
+		const btScalar d1 = Lerp(Lerp(d[4], d[5], ix.f),
+								 Lerp(d[7], d[6], ix.f), iy.f);
+		return (Lerp(d0, d1, iz.f) - margin);
 	}
 	//
-	void					BuildCell(Cell& c)
+	void BuildCell(Cell& c)
 	{
-		const btVector3	org=btVector3(	(btScalar)c.c[0],
-			(btScalar)c.c[1],
-			(btScalar)c.c[2])	*
-			CELLSIZE*voxelsz;
-		for(int k=0;k<=CELLSIZE;++k)
+		const btVector3 org = btVector3((btScalar)c.c[0],
+										(btScalar)c.c[1],
+										(btScalar)c.c[2]) *
+							  CELLSIZE * voxelsz;
+		for (int k = 0; k <= CELLSIZE; ++k)
 		{
-			const btScalar	z=voxelsz*k+org.z();
-			for(int j=0;j<=CELLSIZE;++j)
+			const btScalar z = voxelsz * k + org.z();
+			for (int j = 0; j <= CELLSIZE; ++j)
 			{
-				const btScalar	y=voxelsz*j+org.y();
-				for(int i=0;i<=CELLSIZE;++i)
+				const btScalar y = voxelsz * j + org.y();
+				for (int i = 0; i <= CELLSIZE; ++i)
 				{
-					const btScalar	x=voxelsz*i+org.x();
-					c.d[i][j][k]=DistanceToShape(	btVector3(x,y,z),
-						c.pclient);
+					const btScalar x = voxelsz * i + org.x();
+					c.d[i][j][k] = DistanceToShape(btVector3(x, y, z),
+												   c.pclient);
 				}
 			}
 		}
 	}
 	//
-	static inline btScalar	DistanceToShape(const btVector3& x,
-		const btCollisionShape* shape)
+	static inline btScalar DistanceToShape(const btVector3& x,
+										   const btCollisionShape* shape)
 	{
-		btTransform	unit;
+		btTransform unit;
 		unit.setIdentity();
-		if(shape->isConvex())
+		if (shape->isConvex())
 		{
-			btGjkEpaSolver2::sResults	res;
-			const btConvexShape*				csh=static_cast<const btConvexShape*>(shape);
-			return(btGjkEpaSolver2::SignedDistance(x,0,csh,unit,res));
+			btGjkEpaSolver2::sResults res;
+			const btConvexShape* csh = static_cast<const btConvexShape*>(shape);
+			return (btGjkEpaSolver2::SignedDistance(x, 0, csh, unit, res));
 		}
-		return(0);
+		return (0);
 	}
 	//
-	static inline IntFrac	Decompose(btScalar x)
+	static inline IntFrac Decompose(btScalar x)
 	{
 		/* That one need a lot of improvements...	*/
-		/* Remove test, faster floor...				*/ 
-		IntFrac			r;
-		x/=CELLSIZE;
-		const int		o=x<0?(int)(-x+1):0;
-		x+=o;r.b=(int)x;
-		const btScalar	k=(x-r.b)*CELLSIZE;
-		r.i=(int)k;r.f=k-r.i;r.b-=o;
-		return(r);
+		/* Remove test, faster floor...				*/
+		IntFrac r;
+		x /= CELLSIZE;
+		const int o = x < 0 ? (int)(-x + 1) : 0;
+		x += o;
+		r.b = (int)x;
+		const btScalar k = (x - r.b) * CELLSIZE;
+		r.i = (int)k;
+		r.f = k - r.i;
+		r.b -= o;
+		return (r);
 	}
 	//
-	static inline btScalar	Lerp(btScalar a,btScalar b,btScalar t)
+	static inline btScalar Lerp(btScalar a, btScalar b, btScalar t)
 	{
-		return(a+(b-a)*t);
+		return (a + (b - a) * t);
 	}
 
-
-
 	//
-	static inline unsigned int	Hash(int x,int y,int z,const btCollisionShape* shape)
+	static inline unsigned int Hash(int x, int y, int z, const btCollisionShape* shape)
 	{
 		struct btS
-		{ 
-			int x,y,z;
+		{
+			int x, y, z;
 			void* p;
 		};
 
 		btS myset;
 
-		myset.x=x;myset.y=y;myset.z=z;myset.p=(void*)shape;
+		myset.x = x;
+		myset.y = y;
+		myset.z = z;
+		myset.p = (void*)shape;
 		const void* ptr = &myset;
 
-		unsigned int result = HsiehHash<sizeof(btS)/4> (ptr);
-
+		unsigned int result = HsiehHash<sizeof(btS) / 4>(ptr);
 
 		return result;
 	}
 };
 
-
-#endif //BT_SPARSE_SDF_H
+#endif  //BT_SPARSE_SDF_H
diff --git a/thirdparty/bullet/LinearMath/TaskScheduler/btTaskScheduler.cpp b/thirdparty/bullet/LinearMath/TaskScheduler/btTaskScheduler.cpp
index 49510d16601..5f1115c4027 100644
--- a/thirdparty/bullet/LinearMath/TaskScheduler/btTaskScheduler.cpp
+++ b/thirdparty/bullet/LinearMath/TaskScheduler/btTaskScheduler.cpp
@@ -6,13 +6,11 @@
 #include <stdio.h>
 #include <algorithm>
 
-
-
 #if BT_THREADSAFE
 
 #include "btThreadSupportInterface.h"
 
-#if defined( _WIN32 )
+#if defined(_WIN32)
 
 #define WIN32_LEAN_AND_MEAN
 
@@ -20,404 +18,399 @@
 
 #endif
 
-
 typedef unsigned long long btU64;
 static const int kCacheLineSize = 64;
 
 void btSpinPause()
 {
-#if defined( _WIN32 )
-    YieldProcessor();
+#if defined(_WIN32)
+	YieldProcessor();
 #endif
 }
 
-
 struct WorkerThreadStatus
 {
-    enum Type
-    {
-        kInvalid,
-        kWaitingForWork,
-        kWorking,
-        kSleeping,
-    };
+	enum Type
+	{
+		kInvalid,
+		kWaitingForWork,
+		kWorking,
+		kSleeping,
+	};
 };
 
-
-ATTRIBUTE_ALIGNED64(class) WorkerThreadDirectives
+ATTRIBUTE_ALIGNED64(class)
+WorkerThreadDirectives
 {
-    static const int kMaxThreadCount = BT_MAX_THREAD_COUNT;
-    // directives for all worker threads packed into a single cacheline
-    char m_threadDirs[kMaxThreadCount];
+	static const int kMaxThreadCount = BT_MAX_THREAD_COUNT;
+	// directives for all worker threads packed into a single cacheline
+	char m_threadDirs[kMaxThreadCount];
 
 public:
-    enum Type
-    {
-        kInvalid,
-        kGoToSleep,         // go to sleep
-        kStayAwakeButIdle,  // wait for not checking job queue
-        kScanForJobs,       // actively scan job queue for jobs
-    };
-    WorkerThreadDirectives()
-    {
-        for ( int i = 0; i < kMaxThreadCount; ++i )
-        {
-            m_threadDirs[ i ] = 0;
-        }
-    }
+	enum Type
+	{
+		kInvalid,
+		kGoToSleep,         // go to sleep
+		kStayAwakeButIdle,  // wait for not checking job queue
+		kScanForJobs,       // actively scan job queue for jobs
+	};
+	WorkerThreadDirectives()
+	{
+		for (int i = 0; i < kMaxThreadCount; ++i)
+		{
+			m_threadDirs[i] = 0;
+		}
+	}
 
-    Type getDirective(int threadId)
-    {
-        btAssert(threadId < kMaxThreadCount);
-        return static_cast<Type>(m_threadDirs[threadId]);
-    }
+	Type getDirective(int threadId)
+	{
+		btAssert(threadId < kMaxThreadCount);
+		return static_cast<Type>(m_threadDirs[threadId]);
+	}
 
-    void setDirectiveByRange(int threadBegin, int threadEnd, Type dir)
-    {
-        btAssert( threadBegin < threadEnd );
-        btAssert( threadEnd <= kMaxThreadCount );
-        char dirChar = static_cast<char>(dir);
-        for ( int i = threadBegin; i < threadEnd; ++i )
-        {
-            m_threadDirs[ i ] = dirChar;
-        }
-    }
+	void setDirectiveByRange(int threadBegin, int threadEnd, Type dir)
+	{
+		btAssert(threadBegin < threadEnd);
+		btAssert(threadEnd <= kMaxThreadCount);
+		char dirChar = static_cast<char>(dir);
+		for (int i = threadBegin; i < threadEnd; ++i)
+		{
+			m_threadDirs[i] = dirChar;
+		}
+	}
 };
 
 class JobQueue;
 
-ATTRIBUTE_ALIGNED64(struct) ThreadLocalStorage
+ATTRIBUTE_ALIGNED64(struct)
+ThreadLocalStorage
 {
-    int m_threadId;
-    WorkerThreadStatus::Type m_status;
-    int m_numJobsFinished;
-    btSpinMutex m_mutex;
-    btScalar m_sumResult;
-    WorkerThreadDirectives * m_directive;
-    JobQueue* m_queue;
-    btClock* m_clock;
-    unsigned int m_cooldownTime;
+	int m_threadId;
+	WorkerThreadStatus::Type m_status;
+	int m_numJobsFinished;
+	btSpinMutex m_mutex;
+	btScalar m_sumResult;
+	WorkerThreadDirectives* m_directive;
+	JobQueue* m_queue;
+	btClock* m_clock;
+	unsigned int m_cooldownTime;
 };
 
-
 struct IJob
 {
-    virtual void executeJob(int threadId) = 0;
+	virtual void executeJob(int threadId) = 0;
 };
 
 class ParallelForJob : public IJob
 {
-    const btIParallelForBody* m_body;
-    int m_begin;
-    int m_end;
+	const btIParallelForBody* m_body;
+	int m_begin;
+	int m_end;
 
 public:
-    ParallelForJob( int iBegin, int iEnd, const btIParallelForBody& body )
-    {
-        m_body = &body;
-        m_begin = iBegin;
-        m_end = iEnd;
-    }
-    virtual void executeJob(int threadId) BT_OVERRIDE
-    {
-        BT_PROFILE( "executeJob" );
+	ParallelForJob(int iBegin, int iEnd, const btIParallelForBody& body)
+	{
+		m_body = &body;
+		m_begin = iBegin;
+		m_end = iEnd;
+	}
+	virtual void executeJob(int threadId) BT_OVERRIDE
+	{
+		BT_PROFILE("executeJob");
 
-        // call the functor body to do the work
-        m_body->forLoop( m_begin, m_end );
-    }
+		// call the functor body to do the work
+		m_body->forLoop(m_begin, m_end);
+	}
 };
 
-
 class ParallelSumJob : public IJob
 {
-    const btIParallelSumBody* m_body;
-    ThreadLocalStorage* m_threadLocalStoreArray;
-    int m_begin;
-    int m_end;
+	const btIParallelSumBody* m_body;
+	ThreadLocalStorage* m_threadLocalStoreArray;
+	int m_begin;
+	int m_end;
 
 public:
-    ParallelSumJob( int iBegin, int iEnd, const btIParallelSumBody& body, ThreadLocalStorage* tls )
-    {
-        m_body = &body;
-        m_threadLocalStoreArray = tls;
-        m_begin = iBegin;
-        m_end = iEnd;
-    }
-    virtual void executeJob( int threadId ) BT_OVERRIDE
-    {
-        BT_PROFILE( "executeJob" );
+	ParallelSumJob(int iBegin, int iEnd, const btIParallelSumBody& body, ThreadLocalStorage* tls)
+	{
+		m_body = &body;
+		m_threadLocalStoreArray = tls;
+		m_begin = iBegin;
+		m_end = iEnd;
+	}
+	virtual void executeJob(int threadId) BT_OVERRIDE
+	{
+		BT_PROFILE("executeJob");
 
-        // call the functor body to do the work
-        btScalar val = m_body->sumLoop( m_begin, m_end );
+		// call the functor body to do the work
+		btScalar val = m_body->sumLoop(m_begin, m_end);
 #if BT_PARALLEL_SUM_DETERMINISTISM
-        // by truncating bits of the result, we can make the parallelSum deterministic (at the expense of precision)
-        const float TRUNC_SCALE = float(1<<19);
-        val = floor(val*TRUNC_SCALE+0.5f)/TRUNC_SCALE;  // truncate some bits
+		// by truncating bits of the result, we can make the parallelSum deterministic (at the expense of precision)
+		const float TRUNC_SCALE = float(1 << 19);
+		val = floor(val * TRUNC_SCALE + 0.5f) / TRUNC_SCALE;  // truncate some bits
 #endif
-        m_threadLocalStoreArray[threadId].m_sumResult += val;
-    }
+		m_threadLocalStoreArray[threadId].m_sumResult += val;
+	}
 };
 
-
-ATTRIBUTE_ALIGNED64(class) JobQueue
+ATTRIBUTE_ALIGNED64(class)
+JobQueue
 {
-    btThreadSupportInterface* m_threadSupport;
-    btCriticalSection* m_queueLock;
-    btSpinMutex m_mutex;
+	btThreadSupportInterface* m_threadSupport;
+	btCriticalSection* m_queueLock;
+	btSpinMutex m_mutex;
 
-    btAlignedObjectArray<IJob*> m_jobQueue;
-    char* m_jobMem;
-    int m_jobMemSize;
-    bool m_queueIsEmpty;
-    int m_tailIndex;
-    int m_headIndex;
-    int m_allocSize;
-    bool m_useSpinMutex;
-    btAlignedObjectArray<JobQueue*> m_neighborContexts;
-    char m_cachePadding[kCacheLineSize];  // prevent false sharing
+	btAlignedObjectArray<IJob*> m_jobQueue;
+	char* m_jobMem;
+	int m_jobMemSize;
+	bool m_queueIsEmpty;
+	int m_tailIndex;
+	int m_headIndex;
+	int m_allocSize;
+	bool m_useSpinMutex;
+	btAlignedObjectArray<JobQueue*> m_neighborContexts;
+	char m_cachePadding[kCacheLineSize];  // prevent false sharing
 
-    void freeJobMem()
-    {
-        if ( m_jobMem )
-        {
-            // free old
-            btAlignedFree(m_jobMem);
-            m_jobMem = NULL;
-        }
-    }
-    void resizeJobMem(int newSize)
-    {
-        if (newSize > m_jobMemSize)
-        {
-            freeJobMem();
-            m_jobMem = static_cast<char*>(btAlignedAlloc(newSize, kCacheLineSize));
-            m_jobMemSize = newSize;
-        }
-    }
-
-public:
-
-    JobQueue()
-    {
-        m_jobMem = NULL;
-        m_jobMemSize = 0;
-        m_threadSupport = NULL;
-        m_queueLock = NULL;
-        m_headIndex = 0;
-        m_tailIndex = 0;
-        m_useSpinMutex = false;
-    }
-    ~JobQueue()
-    {
-		exit();
-    }
-	void exit()
-    {
-		freeJobMem();
-        if (m_queueLock && m_threadSupport)
-        {
-            m_threadSupport->deleteCriticalSection(m_queueLock);
-            m_queueLock = NULL;
-			m_threadSupport = 0;
-        }
+	void freeJobMem()
+	{
+		if (m_jobMem)
+		{
+			// free old
+			btAlignedFree(m_jobMem);
+			m_jobMem = NULL;
+		}
+	}
+	void resizeJobMem(int newSize)
+	{
+		if (newSize > m_jobMemSize)
+		{
+			freeJobMem();
+			m_jobMem = static_cast<char*>(btAlignedAlloc(newSize, kCacheLineSize));
+			m_jobMemSize = newSize;
+		}
 	}
 
-    void init(btThreadSupportInterface* threadSup, btAlignedObjectArray<JobQueue>* contextArray)
-    {
-        m_threadSupport = threadSup;
-        if (threadSup)
-        {
-            m_queueLock = m_threadSupport->createCriticalSection();
-        }
-        setupJobStealing(contextArray, contextArray->size());
-    }
-    void setupJobStealing(btAlignedObjectArray<JobQueue>* contextArray, int numActiveContexts)
-    {
-        btAlignedObjectArray<JobQueue>& contexts = *contextArray;
-        int selfIndex = 0;
-        for (int i = 0; i < contexts.size(); ++i)
-        {
-            if ( this == &contexts[ i ] )
-            {
-                selfIndex = i;
-                break;
-            }
-        }
-        int numNeighbors = btMin(2, contexts.size() - 1);
-        int neighborOffsets[ ] = {-1, 1, -2, 2, -3, 3};
-        int numOffsets = sizeof(neighborOffsets)/sizeof(neighborOffsets[0]);
-        m_neighborContexts.reserve( numNeighbors );
-        m_neighborContexts.resizeNoInitialize(0);
-        for (int i = 0; i < numOffsets && m_neighborContexts.size() < numNeighbors; i++)
-        {
-            int neighborIndex = selfIndex + neighborOffsets[i];
-            if ( neighborIndex >= 0 && neighborIndex < numActiveContexts)
-            {
-                m_neighborContexts.push_back( &contexts[ neighborIndex ] );
-            }
-        }
-    }
+public:
+	JobQueue()
+	{
+		m_jobMem = NULL;
+		m_jobMemSize = 0;
+		m_threadSupport = NULL;
+		m_queueLock = NULL;
+		m_headIndex = 0;
+		m_tailIndex = 0;
+		m_useSpinMutex = false;
+	}
+	~JobQueue()
+	{
+		exit();
+	}
+	void exit()
+	{
+		freeJobMem();
+		if (m_queueLock && m_threadSupport)
+		{
+			m_threadSupport->deleteCriticalSection(m_queueLock);
+			m_queueLock = NULL;
+			m_threadSupport = 0;
+		}
+	}
 
-    bool isQueueEmpty() const {return m_queueIsEmpty;}
-    void lockQueue()
-    {
-        if ( m_useSpinMutex )
-        {
-            m_mutex.lock();
-        }
-        else
-        {
-            m_queueLock->lock();
-        }
-    }
-    void unlockQueue()
-    {
-        if ( m_useSpinMutex )
-        {
-            m_mutex.unlock();
-        }
-        else
-        {
-            m_queueLock->unlock();
-        }
-    }
-    void clearQueue(int jobCount, int jobSize)
-    {
-        lockQueue();
-        m_headIndex = 0;
-        m_tailIndex = 0;
-        m_allocSize = 0;
-        m_queueIsEmpty = true;
-        int jobBufSize = jobSize * jobCount;
-        // make sure we have enough memory allocated to store jobs
-        if ( jobBufSize > m_jobMemSize )
-        {
-            resizeJobMem( jobBufSize );
-        }
-        // make sure job queue is big enough
-        if ( jobCount > m_jobQueue.capacity() )
-        {
-            m_jobQueue.reserve( jobCount );
-        }
-        unlockQueue();
-        m_jobQueue.resizeNoInitialize( 0 );
-    }
-    void* allocJobMem(int jobSize)
-    {
-        btAssert(m_jobMemSize >= (m_allocSize + jobSize));
-        void* jobMem = &m_jobMem[m_allocSize];
-        m_allocSize += jobSize;
-        return jobMem;
-    }
-    void submitJob( IJob* job )
-    {
-        btAssert( reinterpret_cast<char*>( job ) >= &m_jobMem[ 0 ] && reinterpret_cast<char*>( job ) < &m_jobMem[ 0 ] + m_allocSize );
-        m_jobQueue.push_back( job );
-        lockQueue();
-        m_tailIndex++;
-        m_queueIsEmpty = false;
-        unlockQueue();
-    }
-    IJob* consumeJobFromOwnQueue()
-    {
-        if ( m_queueIsEmpty )
-        {
-            // lock free path. even if this is taken erroneously it isn't harmful
-            return NULL;
-        }
-        IJob* job = NULL;
-        lockQueue();
-        if ( !m_queueIsEmpty )
-        {
-            job = m_jobQueue[ m_headIndex++ ];
-            btAssert( reinterpret_cast<char*>( job ) >= &m_jobMem[ 0 ] && reinterpret_cast<char*>( job ) < &m_jobMem[ 0 ] + m_allocSize );
-            if ( m_headIndex == m_tailIndex )
-            {
-                m_queueIsEmpty = true;
-            }
-        }
-        unlockQueue();
-        return job;
-    }
-    IJob* consumeJob()
-    {
-        if (IJob* job = consumeJobFromOwnQueue())
-        {
-            return job;
-        }
-        // own queue is empty, try to steal from neighbor
-        for (int i = 0; i < m_neighborContexts.size(); ++i)
-        {
-            JobQueue* otherContext = m_neighborContexts[ i ];
-            if ( IJob* job = otherContext->consumeJobFromOwnQueue() )
-            {
-                return job;
-            }
-        }
-        return NULL;
-    }
+	void init(btThreadSupportInterface * threadSup, btAlignedObjectArray<JobQueue> * contextArray)
+	{
+		m_threadSupport = threadSup;
+		if (threadSup)
+		{
+			m_queueLock = m_threadSupport->createCriticalSection();
+		}
+		setupJobStealing(contextArray, contextArray->size());
+	}
+	void setupJobStealing(btAlignedObjectArray<JobQueue> * contextArray, int numActiveContexts)
+	{
+		btAlignedObjectArray<JobQueue>& contexts = *contextArray;
+		int selfIndex = 0;
+		for (int i = 0; i < contexts.size(); ++i)
+		{
+			if (this == &contexts[i])
+			{
+				selfIndex = i;
+				break;
+			}
+		}
+		int numNeighbors = btMin(2, contexts.size() - 1);
+		int neighborOffsets[] = {-1, 1, -2, 2, -3, 3};
+		int numOffsets = sizeof(neighborOffsets) / sizeof(neighborOffsets[0]);
+		m_neighborContexts.reserve(numNeighbors);
+		m_neighborContexts.resizeNoInitialize(0);
+		for (int i = 0; i < numOffsets && m_neighborContexts.size() < numNeighbors; i++)
+		{
+			int neighborIndex = selfIndex + neighborOffsets[i];
+			if (neighborIndex >= 0 && neighborIndex < numActiveContexts)
+			{
+				m_neighborContexts.push_back(&contexts[neighborIndex]);
+			}
+		}
+	}
+
+	bool isQueueEmpty() const { return m_queueIsEmpty; }
+	void lockQueue()
+	{
+		if (m_useSpinMutex)
+		{
+			m_mutex.lock();
+		}
+		else
+		{
+			m_queueLock->lock();
+		}
+	}
+	void unlockQueue()
+	{
+		if (m_useSpinMutex)
+		{
+			m_mutex.unlock();
+		}
+		else
+		{
+			m_queueLock->unlock();
+		}
+	}
+	void clearQueue(int jobCount, int jobSize)
+	{
+		lockQueue();
+		m_headIndex = 0;
+		m_tailIndex = 0;
+		m_allocSize = 0;
+		m_queueIsEmpty = true;
+		int jobBufSize = jobSize * jobCount;
+		// make sure we have enough memory allocated to store jobs
+		if (jobBufSize > m_jobMemSize)
+		{
+			resizeJobMem(jobBufSize);
+		}
+		// make sure job queue is big enough
+		if (jobCount > m_jobQueue.capacity())
+		{
+			m_jobQueue.reserve(jobCount);
+		}
+		unlockQueue();
+		m_jobQueue.resizeNoInitialize(0);
+	}
+	void* allocJobMem(int jobSize)
+	{
+		btAssert(m_jobMemSize >= (m_allocSize + jobSize));
+		void* jobMem = &m_jobMem[m_allocSize];
+		m_allocSize += jobSize;
+		return jobMem;
+	}
+	void submitJob(IJob * job)
+	{
+		btAssert(reinterpret_cast<char*>(job) >= &m_jobMem[0] && reinterpret_cast<char*>(job) < &m_jobMem[0] + m_allocSize);
+		m_jobQueue.push_back(job);
+		lockQueue();
+		m_tailIndex++;
+		m_queueIsEmpty = false;
+		unlockQueue();
+	}
+	IJob* consumeJobFromOwnQueue()
+	{
+		if (m_queueIsEmpty)
+		{
+			// lock free path. even if this is taken erroneously it isn't harmful
+			return NULL;
+		}
+		IJob* job = NULL;
+		lockQueue();
+		if (!m_queueIsEmpty)
+		{
+			job = m_jobQueue[m_headIndex++];
+			btAssert(reinterpret_cast<char*>(job) >= &m_jobMem[0] && reinterpret_cast<char*>(job) < &m_jobMem[0] + m_allocSize);
+			if (m_headIndex == m_tailIndex)
+			{
+				m_queueIsEmpty = true;
+			}
+		}
+		unlockQueue();
+		return job;
+	}
+	IJob* consumeJob()
+	{
+		if (IJob* job = consumeJobFromOwnQueue())
+		{
+			return job;
+		}
+		// own queue is empty, try to steal from neighbor
+		for (int i = 0; i < m_neighborContexts.size(); ++i)
+		{
+			JobQueue* otherContext = m_neighborContexts[i];
+			if (IJob* job = otherContext->consumeJobFromOwnQueue())
+			{
+				return job;
+			}
+		}
+		return NULL;
+	}
 };
 
-
-static void WorkerThreadFunc( void* userPtr )
+static void WorkerThreadFunc(void* userPtr)
 {
-    BT_PROFILE( "WorkerThreadFunc" );
-    ThreadLocalStorage* localStorage = (ThreadLocalStorage*) userPtr;
-    JobQueue* jobQueue = localStorage->m_queue;
+	BT_PROFILE("WorkerThreadFunc");
+	ThreadLocalStorage* localStorage = (ThreadLocalStorage*)userPtr;
+	JobQueue* jobQueue = localStorage->m_queue;
 
-    bool shouldSleep = false;
-    int threadId = localStorage->m_threadId;
-    while (! shouldSleep)
-    {
-        // do work
-        localStorage->m_mutex.lock();
-        while ( IJob* job = jobQueue->consumeJob() )
-        {
-            localStorage->m_status = WorkerThreadStatus::kWorking;
-            job->executeJob( threadId );
-            localStorage->m_numJobsFinished++;
-        }
-        localStorage->m_status = WorkerThreadStatus::kWaitingForWork;
-        localStorage->m_mutex.unlock();
-        btU64 clockStart = localStorage->m_clock->getTimeMicroseconds();
-        // while queue is empty,
-        while (jobQueue->isQueueEmpty())
-        {
-            // todo: spin wait a bit to avoid hammering the empty queue
-            btSpinPause();
-            if ( localStorage->m_directive->getDirective(threadId) == WorkerThreadDirectives::kGoToSleep )
-            {
-                shouldSleep = true;
-                break;
-            }
-            // if jobs are incoming,
-            if ( localStorage->m_directive->getDirective( threadId ) == WorkerThreadDirectives::kScanForJobs )
-            {
-                clockStart = localStorage->m_clock->getTimeMicroseconds(); // reset clock
-            }
-            else
-            {
-                for ( int i = 0; i < 50; ++i )
-                {
-                    btSpinPause();
-                    btSpinPause();
-                    btSpinPause();
-                    btSpinPause();
-                    if (localStorage->m_directive->getDirective( threadId ) == WorkerThreadDirectives::kScanForJobs || !jobQueue->isQueueEmpty())
-                    {
-                        break;
-                    }
-                }
-                // if no jobs incoming and queue has been empty for the cooldown time, sleep
-                btU64 timeElapsed = localStorage->m_clock->getTimeMicroseconds() - clockStart;
-                if (timeElapsed > localStorage->m_cooldownTime)
-                {
-                    shouldSleep = true;
-                    break;
-                }
-            }
-        }
-    }
+	bool shouldSleep = false;
+	int threadId = localStorage->m_threadId;
+	while (!shouldSleep)
+	{
+		// do work
+		localStorage->m_mutex.lock();
+		while (IJob* job = jobQueue->consumeJob())
+		{
+			localStorage->m_status = WorkerThreadStatus::kWorking;
+			job->executeJob(threadId);
+			localStorage->m_numJobsFinished++;
+		}
+		localStorage->m_status = WorkerThreadStatus::kWaitingForWork;
+		localStorage->m_mutex.unlock();
+		btU64 clockStart = localStorage->m_clock->getTimeMicroseconds();
+		// while queue is empty,
+		while (jobQueue->isQueueEmpty())
+		{
+			// todo: spin wait a bit to avoid hammering the empty queue
+			btSpinPause();
+			if (localStorage->m_directive->getDirective(threadId) == WorkerThreadDirectives::kGoToSleep)
+			{
+				shouldSleep = true;
+				break;
+			}
+			// if jobs are incoming,
+			if (localStorage->m_directive->getDirective(threadId) == WorkerThreadDirectives::kScanForJobs)
+			{
+				clockStart = localStorage->m_clock->getTimeMicroseconds();  // reset clock
+			}
+			else
+			{
+				for (int i = 0; i < 50; ++i)
+				{
+					btSpinPause();
+					btSpinPause();
+					btSpinPause();
+					btSpinPause();
+					if (localStorage->m_directive->getDirective(threadId) == WorkerThreadDirectives::kScanForJobs || !jobQueue->isQueueEmpty())
+					{
+						break;
+					}
+				}
+				// if no jobs incoming and queue has been empty for the cooldown time, sleep
+				btU64 timeElapsed = localStorage->m_clock->getTimeMicroseconds() - clockStart;
+				if (timeElapsed > localStorage->m_cooldownTime)
+				{
+					shouldSleep = true;
+					break;
+				}
+			}
+		}
+	}
 	{
 		BT_PROFILE("sleep");
 		// go sleep
@@ -427,376 +420,373 @@ static void WorkerThreadFunc( void* userPtr )
 	}
 }
 
-
 class btTaskSchedulerDefault : public btITaskScheduler
 {
-    btThreadSupportInterface* m_threadSupport;
-    WorkerThreadDirectives* m_workerDirective;
-    btAlignedObjectArray<JobQueue> m_jobQueues;
-    btAlignedObjectArray<JobQueue*> m_perThreadJobQueues;
-    btAlignedObjectArray<ThreadLocalStorage> m_threadLocalStorage;
-    btSpinMutex m_antiNestingLock;  // prevent nested parallel-for
-    btClock m_clock;
-    int m_numThreads;
-    int m_numWorkerThreads;
-    int m_numActiveJobQueues;
-    int m_maxNumThreads;
-    int m_numJobs;
-    static const int kFirstWorkerThreadId = 1;
+	btThreadSupportInterface* m_threadSupport;
+	WorkerThreadDirectives* m_workerDirective;
+	btAlignedObjectArray<JobQueue> m_jobQueues;
+	btAlignedObjectArray<JobQueue*> m_perThreadJobQueues;
+	btAlignedObjectArray<ThreadLocalStorage> m_threadLocalStorage;
+	btSpinMutex m_antiNestingLock;  // prevent nested parallel-for
+	btClock m_clock;
+	int m_numThreads;
+	int m_numWorkerThreads;
+	int m_numActiveJobQueues;
+	int m_maxNumThreads;
+	int m_numJobs;
+	static const int kFirstWorkerThreadId = 1;
+
 public:
+	btTaskSchedulerDefault() : btITaskScheduler("ThreadSupport")
+	{
+		m_threadSupport = NULL;
+		m_workerDirective = NULL;
+	}
 
-    btTaskSchedulerDefault() : btITaskScheduler("ThreadSupport")
-    {
-        m_threadSupport = NULL;
-        m_workerDirective = NULL;
-    }
+	virtual ~btTaskSchedulerDefault()
+	{
+		waitForWorkersToSleep();
 
-    virtual ~btTaskSchedulerDefault()
-    {
-        waitForWorkersToSleep();
+		for (int i = 0; i < m_jobQueues.size(); ++i)
+		{
+			m_jobQueues[i].exit();
+		}
 
-		for ( int i = 0; i < m_jobQueues.size(); ++i )
-        {
-            m_jobQueues[i].exit();
-        }
+		if (m_threadSupport)
+		{
+			delete m_threadSupport;
+			m_threadSupport = NULL;
+		}
+		if (m_workerDirective)
+		{
+			btAlignedFree(m_workerDirective);
+			m_workerDirective = NULL;
+		}
+	}
 
-        if (m_threadSupport)
-        {
-            delete m_threadSupport;
-            m_threadSupport = NULL;
-        }
-        if (m_workerDirective)
-        {
-            btAlignedFree(m_workerDirective);
-            m_workerDirective = NULL;
-        }
-    }
+	void init()
+	{
+		btThreadSupportInterface::ConstructionInfo constructionInfo("TaskScheduler", WorkerThreadFunc);
+		m_threadSupport = btThreadSupportInterface::create(constructionInfo);
+		m_workerDirective = static_cast<WorkerThreadDirectives*>(btAlignedAlloc(sizeof(*m_workerDirective), 64));
 
-    void init()
-    {
-        btThreadSupportInterface::ConstructionInfo constructionInfo( "TaskScheduler", WorkerThreadFunc );
-        m_threadSupport = btThreadSupportInterface::create( constructionInfo );
-        m_workerDirective = static_cast<WorkerThreadDirectives*>(btAlignedAlloc(sizeof(*m_workerDirective), 64));
+		m_numWorkerThreads = m_threadSupport->getNumWorkerThreads();
+		m_maxNumThreads = m_threadSupport->getNumWorkerThreads() + 1;
+		m_numThreads = m_maxNumThreads;
+		// ideal to have one job queue for each physical processor (except for the main thread which needs no queue)
+		int numThreadsPerQueue = m_threadSupport->getLogicalToPhysicalCoreRatio();
+		int numJobQueues = (numThreadsPerQueue == 1) ? (m_maxNumThreads - 1) : (m_maxNumThreads / numThreadsPerQueue);
+		m_jobQueues.resize(numJobQueues);
+		m_numActiveJobQueues = numJobQueues;
+		for (int i = 0; i < m_jobQueues.size(); ++i)
+		{
+			m_jobQueues[i].init(m_threadSupport, &m_jobQueues);
+		}
+		m_perThreadJobQueues.resize(m_numThreads);
+		for (int i = 0; i < m_numThreads; i++)
+		{
+			JobQueue* jq = NULL;
+			// only worker threads get a job queue
+			if (i > 0)
+			{
+				if (numThreadsPerQueue == 1)
+				{
+					// one queue per worker thread
+					jq = &m_jobQueues[i - kFirstWorkerThreadId];
+				}
+				else
+				{
+					// 2 threads share each queue
+					jq = &m_jobQueues[i / numThreadsPerQueue];
+				}
+			}
+			m_perThreadJobQueues[i] = jq;
+		}
+		m_threadLocalStorage.resize(m_numThreads);
+		for (int i = 0; i < m_numThreads; i++)
+		{
+			ThreadLocalStorage& storage = m_threadLocalStorage[i];
+			storage.m_threadId = i;
+			storage.m_directive = m_workerDirective;
+			storage.m_status = WorkerThreadStatus::kSleeping;
+			storage.m_cooldownTime = 100;  // 100 microseconds, threads go to sleep after this long if they have nothing to do
+			storage.m_clock = &m_clock;
+			storage.m_queue = m_perThreadJobQueues[i];
+		}
+		setWorkerDirectives(WorkerThreadDirectives::kGoToSleep);  // no work for them yet
+		setNumThreads(m_threadSupport->getCacheFriendlyNumThreads());
+	}
 
-        m_numWorkerThreads = m_threadSupport->getNumWorkerThreads();
-        m_maxNumThreads = m_threadSupport->getNumWorkerThreads() + 1;
-        m_numThreads = m_maxNumThreads;
-        // ideal to have one job queue for each physical processor (except for the main thread which needs no queue)
-        int numThreadsPerQueue = m_threadSupport->getLogicalToPhysicalCoreRatio();
-        int numJobQueues = (numThreadsPerQueue == 1) ? (m_maxNumThreads-1) : (m_maxNumThreads / numThreadsPerQueue);
-        m_jobQueues.resize(numJobQueues);
-        m_numActiveJobQueues = numJobQueues;
-        for ( int i = 0; i < m_jobQueues.size(); ++i )
-        {
-            m_jobQueues[i].init( m_threadSupport, &m_jobQueues );
-        }
-        m_perThreadJobQueues.resize(m_numThreads);
-        for ( int i = 0; i < m_numThreads; i++ )
-        {
-            JobQueue* jq = NULL;
-            // only worker threads get a job queue
-            if (i > 0)
-            {
-                if (numThreadsPerQueue == 1)
-                {
-                    // one queue per worker thread
-                    jq = &m_jobQueues[ i - kFirstWorkerThreadId ];
-                }
-                else
-                {
-                    // 2 threads share each queue
-                    jq = &m_jobQueues[ i / numThreadsPerQueue ];
-                }
-            }
-            m_perThreadJobQueues[i] = jq;
-        }
-        m_threadLocalStorage.resize(m_numThreads);
-        for ( int i = 0; i < m_numThreads; i++ )
-        {
-            ThreadLocalStorage& storage = m_threadLocalStorage[i];
-            storage.m_threadId = i;
-            storage.m_directive = m_workerDirective;
-            storage.m_status = WorkerThreadStatus::kSleeping;
-            storage.m_cooldownTime = 100; // 100 microseconds, threads go to sleep after this long if they have nothing to do
-            storage.m_clock = &m_clock;
-            storage.m_queue = m_perThreadJobQueues[i];
-        }
-        setWorkerDirectives( WorkerThreadDirectives::kGoToSleep ); // no work for them yet
-        setNumThreads( m_threadSupport->getCacheFriendlyNumThreads() );
-    }
+	void setWorkerDirectives(WorkerThreadDirectives::Type dir)
+	{
+		m_workerDirective->setDirectiveByRange(kFirstWorkerThreadId, m_numThreads, dir);
+	}
 
-    void setWorkerDirectives(WorkerThreadDirectives::Type dir)
-    {
-        m_workerDirective->setDirectiveByRange(kFirstWorkerThreadId, m_numThreads, dir);
-    }
+	virtual int getMaxNumThreads() const BT_OVERRIDE
+	{
+		return m_maxNumThreads;
+	}
 
-    virtual int getMaxNumThreads() const BT_OVERRIDE
-    {
-        return m_maxNumThreads;
-    }
+	virtual int getNumThreads() const BT_OVERRIDE
+	{
+		return m_numThreads;
+	}
 
-    virtual int getNumThreads() const BT_OVERRIDE
-    {
-        return m_numThreads;
-    }
+	virtual void setNumThreads(int numThreads) BT_OVERRIDE
+	{
+		m_numThreads = btMax(btMin(numThreads, int(m_maxNumThreads)), 1);
+		m_numWorkerThreads = m_numThreads - 1;
+		m_numActiveJobQueues = 0;
+		// if there is at least 1 worker,
+		if (m_numWorkerThreads > 0)
+		{
+			// re-setup job stealing between queues to avoid attempting to steal from an inactive job queue
+			JobQueue* lastActiveContext = m_perThreadJobQueues[m_numThreads - 1];
+			int iLastActiveContext = lastActiveContext - &m_jobQueues[0];
+			m_numActiveJobQueues = iLastActiveContext + 1;
+			for (int i = 0; i < m_jobQueues.size(); ++i)
+			{
+				m_jobQueues[i].setupJobStealing(&m_jobQueues, m_numActiveJobQueues);
+			}
+		}
+		m_workerDirective->setDirectiveByRange(m_numThreads, BT_MAX_THREAD_COUNT, WorkerThreadDirectives::kGoToSleep);
+	}
 
-    virtual void setNumThreads( int numThreads ) BT_OVERRIDE
-    {
-        m_numThreads = btMax( btMin(numThreads, int(m_maxNumThreads)), 1 );
-        m_numWorkerThreads = m_numThreads - 1;
-        m_numActiveJobQueues = 0;
-        // if there is at least 1 worker,
-        if ( m_numWorkerThreads > 0 )
-        {
-            // re-setup job stealing between queues to avoid attempting to steal from an inactive job queue
-            JobQueue* lastActiveContext = m_perThreadJobQueues[ m_numThreads - 1 ];
-            int iLastActiveContext = lastActiveContext - &m_jobQueues[0];
-            m_numActiveJobQueues = iLastActiveContext + 1;
-            for ( int i = 0; i < m_jobQueues.size(); ++i )
-            {
-                m_jobQueues[ i ].setupJobStealing( &m_jobQueues, m_numActiveJobQueues );
-            }
-        }
-        m_workerDirective->setDirectiveByRange(m_numThreads, BT_MAX_THREAD_COUNT, WorkerThreadDirectives::kGoToSleep);
-    }
+	void waitJobs()
+	{
+		BT_PROFILE("waitJobs");
+		// have the main thread work until the job queues are empty
+		int numMainThreadJobsFinished = 0;
+		for (int i = 0; i < m_numActiveJobQueues; ++i)
+		{
+			while (IJob* job = m_jobQueues[i].consumeJob())
+			{
+				job->executeJob(0);
+				numMainThreadJobsFinished++;
+			}
+		}
 
-    void waitJobs()
-    {
-        BT_PROFILE( "waitJobs" );
-        // have the main thread work until the job queues are empty
-        int numMainThreadJobsFinished = 0;
-        for ( int i = 0; i < m_numActiveJobQueues; ++i )
-        {
-            while ( IJob* job = m_jobQueues[i].consumeJob() )
-            {
-                job->executeJob( 0 );
-                numMainThreadJobsFinished++;
-            }
-        }
+		// done with jobs for now, tell workers to rest (but not sleep)
+		setWorkerDirectives(WorkerThreadDirectives::kStayAwakeButIdle);
 
-        // done with jobs for now, tell workers to rest (but not sleep)
-        setWorkerDirectives( WorkerThreadDirectives::kStayAwakeButIdle );
+		btU64 clockStart = m_clock.getTimeMicroseconds();
+		// wait for workers to finish any jobs in progress
+		while (true)
+		{
+			int numWorkerJobsFinished = 0;
+			for (int iThread = kFirstWorkerThreadId; iThread < m_numThreads; ++iThread)
+			{
+				ThreadLocalStorage* storage = &m_threadLocalStorage[iThread];
+				storage->m_mutex.lock();
+				numWorkerJobsFinished += storage->m_numJobsFinished;
+				storage->m_mutex.unlock();
+			}
+			if (numWorkerJobsFinished + numMainThreadJobsFinished == m_numJobs)
+			{
+				break;
+			}
+			btU64 timeElapsed = m_clock.getTimeMicroseconds() - clockStart;
+			btAssert(timeElapsed < 1000);
+			if (timeElapsed > 100000)
+			{
+				break;
+			}
+			btSpinPause();
+		}
+	}
 
-        btU64 clockStart = m_clock.getTimeMicroseconds();
-        // wait for workers to finish any jobs in progress
-        while ( true )
-        {
-            int numWorkerJobsFinished = 0;
-            for ( int iThread = kFirstWorkerThreadId; iThread < m_numThreads; ++iThread )
-            {
-                ThreadLocalStorage* storage = &m_threadLocalStorage[iThread];
-                storage->m_mutex.lock();
-                numWorkerJobsFinished += storage->m_numJobsFinished;
-                storage->m_mutex.unlock();
-            }
-            if (numWorkerJobsFinished + numMainThreadJobsFinished == m_numJobs)
-            {
-                break;
-            }
-            btU64 timeElapsed = m_clock.getTimeMicroseconds() - clockStart;
-            btAssert(timeElapsed < 1000);
-            if (timeElapsed > 100000)
-            {
-                break;
-            }
-            btSpinPause();
-        }
-    }
+	void wakeWorkers(int numWorkersToWake)
+	{
+		BT_PROFILE("wakeWorkers");
+		btAssert(m_workerDirective->getDirective(1) == WorkerThreadDirectives::kScanForJobs);
+		int numDesiredWorkers = btMin(numWorkersToWake, m_numWorkerThreads);
+		int numActiveWorkers = 0;
+		for (int iWorker = 0; iWorker < m_numWorkerThreads; ++iWorker)
+		{
+			// note this count of active workers is not necessarily totally reliable, because a worker thread could be
+			// just about to put itself to sleep. So we may on occasion fail to wake up all the workers. It should be rare.
+			ThreadLocalStorage& storage = m_threadLocalStorage[kFirstWorkerThreadId + iWorker];
+			if (storage.m_status != WorkerThreadStatus::kSleeping)
+			{
+				numActiveWorkers++;
+			}
+		}
+		for (int iWorker = 0; iWorker < m_numWorkerThreads && numActiveWorkers < numDesiredWorkers; ++iWorker)
+		{
+			ThreadLocalStorage& storage = m_threadLocalStorage[kFirstWorkerThreadId + iWorker];
+			if (storage.m_status == WorkerThreadStatus::kSleeping)
+			{
+				m_threadSupport->runTask(iWorker, &storage);
+				numActiveWorkers++;
+			}
+		}
+	}
 
-    void wakeWorkers(int numWorkersToWake)
-    {
-        BT_PROFILE( "wakeWorkers" );
-        btAssert( m_workerDirective->getDirective(1) == WorkerThreadDirectives::kScanForJobs );
-        int numDesiredWorkers = btMin(numWorkersToWake, m_numWorkerThreads);
-        int numActiveWorkers = 0;
-        for ( int iWorker = 0; iWorker < m_numWorkerThreads; ++iWorker )
-        {
-            // note this count of active workers is not necessarily totally reliable, because a worker thread could be
-            // just about to put itself to sleep. So we may on occasion fail to wake up all the workers. It should be rare.
-            ThreadLocalStorage& storage = m_threadLocalStorage[ kFirstWorkerThreadId + iWorker ];
-            if (storage.m_status != WorkerThreadStatus::kSleeping)
-            {
-                numActiveWorkers++;
-            }
-        }
-        for ( int iWorker = 0; iWorker < m_numWorkerThreads && numActiveWorkers < numDesiredWorkers; ++iWorker )
-        {
-            ThreadLocalStorage& storage = m_threadLocalStorage[ kFirstWorkerThreadId + iWorker ];
-            if (storage.m_status == WorkerThreadStatus::kSleeping)
-            {
-                m_threadSupport->runTask( iWorker, &storage );
-                numActiveWorkers++;
-            }
-        }
-    }
+	void waitForWorkersToSleep()
+	{
+		BT_PROFILE("waitForWorkersToSleep");
+		setWorkerDirectives(WorkerThreadDirectives::kGoToSleep);
+		m_threadSupport->waitForAllTasks();
+		for (int i = kFirstWorkerThreadId; i < m_numThreads; i++)
+		{
+			ThreadLocalStorage& storage = m_threadLocalStorage[i];
+			btAssert(storage.m_status == WorkerThreadStatus::kSleeping);
+		}
+	}
 
-    void waitForWorkersToSleep()
-    {
-        BT_PROFILE( "waitForWorkersToSleep" );
-        setWorkerDirectives( WorkerThreadDirectives::kGoToSleep );
-        m_threadSupport->waitForAllTasks();
-        for ( int i = kFirstWorkerThreadId; i < m_numThreads; i++ )
-        {
-            ThreadLocalStorage& storage = m_threadLocalStorage[i];
-            btAssert( storage.m_status == WorkerThreadStatus::kSleeping );
-        }
-    }
+	virtual void sleepWorkerThreadsHint() BT_OVERRIDE
+	{
+		BT_PROFILE("sleepWorkerThreadsHint");
+		// hint the task scheduler that we may not be using these threads for a little while
+		setWorkerDirectives(WorkerThreadDirectives::kGoToSleep);
+	}
 
-    virtual void sleepWorkerThreadsHint() BT_OVERRIDE
-    {
-        BT_PROFILE( "sleepWorkerThreadsHint" );
-        // hint the task scheduler that we may not be using these threads for a little while
-        setWorkerDirectives( WorkerThreadDirectives::kGoToSleep );
-    }
+	void prepareWorkerThreads()
+	{
+		for (int i = kFirstWorkerThreadId; i < m_numThreads; ++i)
+		{
+			ThreadLocalStorage& storage = m_threadLocalStorage[i];
+			storage.m_mutex.lock();
+			storage.m_numJobsFinished = 0;
+			storage.m_mutex.unlock();
+		}
+		setWorkerDirectives(WorkerThreadDirectives::kScanForJobs);
+	}
 
-    void prepareWorkerThreads()
-    {
-        for ( int i = kFirstWorkerThreadId; i < m_numThreads; ++i )
-        {
-            ThreadLocalStorage& storage = m_threadLocalStorage[i];
-            storage.m_mutex.lock();
-            storage.m_numJobsFinished = 0;
-            storage.m_mutex.unlock();
-        }
-        setWorkerDirectives( WorkerThreadDirectives::kScanForJobs );
-    }
+	virtual void parallelFor(int iBegin, int iEnd, int grainSize, const btIParallelForBody& body) BT_OVERRIDE
+	{
+		BT_PROFILE("parallelFor_ThreadSupport");
+		btAssert(iEnd >= iBegin);
+		btAssert(grainSize >= 1);
+		int iterationCount = iEnd - iBegin;
+		if (iterationCount > grainSize && m_numWorkerThreads > 0 && m_antiNestingLock.tryLock())
+		{
+			typedef ParallelForJob JobType;
+			int jobCount = (iterationCount + grainSize - 1) / grainSize;
+			m_numJobs = jobCount;
+			btAssert(jobCount >= 2);  // need more than one job for multithreading
+			int jobSize = sizeof(JobType);
 
-    virtual void parallelFor( int iBegin, int iEnd, int grainSize, const btIParallelForBody& body ) BT_OVERRIDE
-    {
-        BT_PROFILE( "parallelFor_ThreadSupport" );
-        btAssert( iEnd >= iBegin );
-        btAssert( grainSize >= 1 );
-        int iterationCount = iEnd - iBegin;
-        if ( iterationCount > grainSize && m_numWorkerThreads > 0 && m_antiNestingLock.tryLock() )
-        {
-            typedef ParallelForJob JobType;
-            int jobCount = ( iterationCount + grainSize - 1 ) / grainSize;
-            m_numJobs = jobCount;
-            btAssert( jobCount >= 2 );  // need more than one job for multithreading
-            int jobSize = sizeof( JobType );
+			for (int i = 0; i < m_numActiveJobQueues; ++i)
+			{
+				m_jobQueues[i].clearQueue(jobCount, jobSize);
+			}
+			// prepare worker threads for incoming work
+			prepareWorkerThreads();
+			// submit all of the jobs
+			int iJob = 0;
+			int iThread = kFirstWorkerThreadId;  // first worker thread
+			for (int i = iBegin; i < iEnd; i += grainSize)
+			{
+				btAssert(iJob < jobCount);
+				int iE = btMin(i + grainSize, iEnd);
+				JobQueue* jq = m_perThreadJobQueues[iThread];
+				btAssert(jq);
+				btAssert((jq - &m_jobQueues[0]) < m_numActiveJobQueues);
+				void* jobMem = jq->allocJobMem(jobSize);
+				JobType* job = new (jobMem) ParallelForJob(i, iE, body);  // placement new
+				jq->submitJob(job);
+				iJob++;
+				iThread++;
+				if (iThread >= m_numThreads)
+				{
+					iThread = kFirstWorkerThreadId;  // first worker thread
+				}
+			}
+			wakeWorkers(jobCount - 1);
 
-            for (int i = 0; i < m_numActiveJobQueues; ++i)
-            {
-                m_jobQueues[i].clearQueue( jobCount, jobSize );
-            }
-            // prepare worker threads for incoming work
-            prepareWorkerThreads();
-            // submit all of the jobs
-            int iJob = 0;
-            int iThread = kFirstWorkerThreadId;  // first worker thread
-            for ( int i = iBegin; i < iEnd; i += grainSize )
-            {
-                btAssert( iJob < jobCount );
-                int iE = btMin( i + grainSize, iEnd );
-                JobQueue* jq = m_perThreadJobQueues[ iThread ];
-                btAssert(jq);
-                btAssert((jq - &m_jobQueues[0]) < m_numActiveJobQueues);
-                void* jobMem = jq->allocJobMem(jobSize);
-                JobType* job = new ( jobMem ) ParallelForJob( i, iE, body );  // placement new
-                jq->submitJob( job );
-                iJob++;
-                iThread++;
-                if ( iThread >= m_numThreads )
-                {
-                    iThread = kFirstWorkerThreadId;  // first worker thread
-                }
-            }
-            wakeWorkers( jobCount - 1 );
+			// put the main thread to work on emptying the job queue and then wait for all workers to finish
+			waitJobs();
+			m_antiNestingLock.unlock();
+		}
+		else
+		{
+			BT_PROFILE("parallelFor_mainThread");
+			// just run on main thread
+			body.forLoop(iBegin, iEnd);
+		}
+	}
+	virtual btScalar parallelSum(int iBegin, int iEnd, int grainSize, const btIParallelSumBody& body) BT_OVERRIDE
+	{
+		BT_PROFILE("parallelSum_ThreadSupport");
+		btAssert(iEnd >= iBegin);
+		btAssert(grainSize >= 1);
+		int iterationCount = iEnd - iBegin;
+		if (iterationCount > grainSize && m_numWorkerThreads > 0 && m_antiNestingLock.tryLock())
+		{
+			typedef ParallelSumJob JobType;
+			int jobCount = (iterationCount + grainSize - 1) / grainSize;
+			m_numJobs = jobCount;
+			btAssert(jobCount >= 2);  // need more than one job for multithreading
+			int jobSize = sizeof(JobType);
+			for (int i = 0; i < m_numActiveJobQueues; ++i)
+			{
+				m_jobQueues[i].clearQueue(jobCount, jobSize);
+			}
 
-            // put the main thread to work on emptying the job queue and then wait for all workers to finish
-            waitJobs();
-            m_antiNestingLock.unlock();
-        }
-        else
-        {
-            BT_PROFILE( "parallelFor_mainThread" );
-            // just run on main thread
-            body.forLoop( iBegin, iEnd );
-        }
-    }
-    virtual btScalar parallelSum( int iBegin, int iEnd, int grainSize, const btIParallelSumBody& body ) BT_OVERRIDE
-    {
-        BT_PROFILE( "parallelSum_ThreadSupport" );
-        btAssert( iEnd >= iBegin );
-        btAssert( grainSize >= 1 );
-        int iterationCount = iEnd - iBegin;
-        if ( iterationCount > grainSize && m_numWorkerThreads > 0 && m_antiNestingLock.tryLock() )
-        {
-            typedef ParallelSumJob JobType;
-            int jobCount = ( iterationCount + grainSize - 1 ) / grainSize;
-            m_numJobs = jobCount;
-            btAssert( jobCount >= 2 );  // need more than one job for multithreading
-            int jobSize = sizeof( JobType );
-            for (int i = 0; i < m_numActiveJobQueues; ++i)
-            {
-                m_jobQueues[i].clearQueue( jobCount, jobSize );
-            }
+			// initialize summation
+			for (int iThread = 0; iThread < m_numThreads; ++iThread)
+			{
+				m_threadLocalStorage[iThread].m_sumResult = btScalar(0);
+			}
 
-            // initialize summation
-            for ( int iThread = 0; iThread < m_numThreads; ++iThread )
-            {
-                m_threadLocalStorage[iThread].m_sumResult = btScalar(0);
-            }
+			// prepare worker threads for incoming work
+			prepareWorkerThreads();
+			// submit all of the jobs
+			int iJob = 0;
+			int iThread = kFirstWorkerThreadId;  // first worker thread
+			for (int i = iBegin; i < iEnd; i += grainSize)
+			{
+				btAssert(iJob < jobCount);
+				int iE = btMin(i + grainSize, iEnd);
+				JobQueue* jq = m_perThreadJobQueues[iThread];
+				btAssert(jq);
+				btAssert((jq - &m_jobQueues[0]) < m_numActiveJobQueues);
+				void* jobMem = jq->allocJobMem(jobSize);
+				JobType* job = new (jobMem) ParallelSumJob(i, iE, body, &m_threadLocalStorage[0]);  // placement new
+				jq->submitJob(job);
+				iJob++;
+				iThread++;
+				if (iThread >= m_numThreads)
+				{
+					iThread = kFirstWorkerThreadId;  // first worker thread
+				}
+			}
+			wakeWorkers(jobCount - 1);
 
-            // prepare worker threads for incoming work
-            prepareWorkerThreads();
-            // submit all of the jobs
-            int iJob = 0;
-            int iThread = kFirstWorkerThreadId;  // first worker thread
-            for ( int i = iBegin; i < iEnd; i += grainSize )
-            {
-                btAssert( iJob < jobCount );
-                int iE = btMin( i + grainSize, iEnd );
-                JobQueue* jq = m_perThreadJobQueues[ iThread ];
-                btAssert(jq);
-                btAssert((jq - &m_jobQueues[0]) < m_numActiveJobQueues);
-                void* jobMem = jq->allocJobMem(jobSize);
-                JobType* job = new ( jobMem ) ParallelSumJob( i, iE, body, &m_threadLocalStorage[0] );  // placement new
-                jq->submitJob( job );
-                iJob++;
-                iThread++;
-                if ( iThread >= m_numThreads )
-                {
-                    iThread = kFirstWorkerThreadId;  // first worker thread
-                }
-            }
-            wakeWorkers( jobCount - 1 );
+			// put the main thread to work on emptying the job queue and then wait for all workers to finish
+			waitJobs();
 
-            // put the main thread to work on emptying the job queue and then wait for all workers to finish
-            waitJobs();
-
-            // add up all the thread sums
-            btScalar sum = btScalar(0);
-            for ( int iThread = 0; iThread < m_numThreads; ++iThread )
-            {
-                sum += m_threadLocalStorage[ iThread ].m_sumResult;
-            }
-            m_antiNestingLock.unlock();
-            return sum;
-        }
-        else
-        {
-            BT_PROFILE( "parallelSum_mainThread" );
-            // just run on main thread
-            return body.sumLoop( iBegin, iEnd );
-        }
-    }
+			// add up all the thread sums
+			btScalar sum = btScalar(0);
+			for (int iThread = 0; iThread < m_numThreads; ++iThread)
+			{
+				sum += m_threadLocalStorage[iThread].m_sumResult;
+			}
+			m_antiNestingLock.unlock();
+			return sum;
+		}
+		else
+		{
+			BT_PROFILE("parallelSum_mainThread");
+			// just run on main thread
+			return body.sumLoop(iBegin, iEnd);
+		}
+	}
 };
 
+btITaskScheduler* btCreateDefaultTaskScheduler()
+{
+	btTaskSchedulerDefault* ts = new btTaskSchedulerDefault();
+	ts->init();
+	return ts;
+}
 
+#else  // #if BT_THREADSAFE
 
 btITaskScheduler* btCreateDefaultTaskScheduler()
 {
-    btTaskSchedulerDefault* ts = new btTaskSchedulerDefault();
-    ts->init();
-    return ts;
+	return NULL;
 }
 
-#else // #if BT_THREADSAFE
-
-btITaskScheduler* btCreateDefaultTaskScheduler()
-{
-    return NULL;
-}
-
-#endif // #else // #if BT_THREADSAFE
+#endif  // #else // #if BT_THREADSAFE
diff --git a/thirdparty/bullet/LinearMath/TaskScheduler/btThreadSupportInterface.h b/thirdparty/bullet/LinearMath/TaskScheduler/btThreadSupportInterface.h
index a0ad802b1e1..1fe49335a19 100644
--- a/thirdparty/bullet/LinearMath/TaskScheduler/btThreadSupportInterface.h
+++ b/thirdparty/bullet/LinearMath/TaskScheduler/btThreadSupportInterface.h
@@ -16,55 +16,49 @@ subject to the following restrictions:
 #ifndef BT_THREAD_SUPPORT_INTERFACE_H
 #define BT_THREAD_SUPPORT_INTERFACE_H
 
-
-
 class btCriticalSection
 {
 public:
-    btCriticalSection() {}
-    virtual ~btCriticalSection() {}
+	btCriticalSection() {}
+	virtual ~btCriticalSection() {}
 
-    virtual void lock() = 0;
-    virtual void unlock() = 0;
+	virtual void lock() = 0;
+	virtual void unlock() = 0;
 };
 
-
 class btThreadSupportInterface
 {
 public:
+	virtual ~btThreadSupportInterface() {}
 
-    virtual ~btThreadSupportInterface() {}
+	virtual int getNumWorkerThreads() const = 0;            // number of worker threads (total number of logical processors - 1)
+	virtual int getCacheFriendlyNumThreads() const = 0;     // the number of logical processors sharing a single L3 cache
+	virtual int getLogicalToPhysicalCoreRatio() const = 0;  // the number of logical processors per physical processor (usually 1 or 2)
+	virtual void runTask(int threadIndex, void* userData) = 0;
+	virtual void waitForAllTasks() = 0;
 
-    virtual int getNumWorkerThreads() const = 0;  // number of worker threads (total number of logical processors - 1)
-    virtual int getCacheFriendlyNumThreads() const = 0;  // the number of logical processors sharing a single L3 cache
-    virtual int getLogicalToPhysicalCoreRatio() const = 0;  // the number of logical processors per physical processor (usually 1 or 2)
-    virtual void runTask( int threadIndex, void* userData ) = 0;
-    virtual void waitForAllTasks() = 0;
+	virtual btCriticalSection* createCriticalSection() = 0;
+	virtual void deleteCriticalSection(btCriticalSection* criticalSection) = 0;
 
-    virtual btCriticalSection* createCriticalSection() = 0;
-    virtual void deleteCriticalSection( btCriticalSection* criticalSection ) = 0;
+	typedef void (*ThreadFunc)(void* userPtr);
 
-    typedef void( *ThreadFunc )( void* userPtr );
+	struct ConstructionInfo
+	{
+		ConstructionInfo(const char* uniqueName,
+						 ThreadFunc userThreadFunc,
+						 int threadStackSize = 65535)
+			: m_uniqueName(uniqueName),
+			  m_userThreadFunc(userThreadFunc),
+			  m_threadStackSize(threadStackSize)
+		{
+		}
 
-    struct ConstructionInfo
-    {
-        ConstructionInfo( const char* uniqueName,
-            ThreadFunc userThreadFunc,
-            int threadStackSize = 65535
-        )
-            :m_uniqueName( uniqueName ),
-            m_userThreadFunc( userThreadFunc ),
-            m_threadStackSize( threadStackSize )
-        {
-        }
+		const char* m_uniqueName;
+		ThreadFunc m_userThreadFunc;
+		int m_threadStackSize;
+	};
 
-        const char*     m_uniqueName;
-        ThreadFunc      m_userThreadFunc;
-        int             m_threadStackSize;
-    };
-
-    static btThreadSupportInterface* create( const ConstructionInfo& info );
+	static btThreadSupportInterface* create(const ConstructionInfo& info);
 };
 
-#endif //BT_THREAD_SUPPORT_INTERFACE_H
-
+#endif  //BT_THREAD_SUPPORT_INTERFACE_H
diff --git a/thirdparty/bullet/LinearMath/TaskScheduler/btThreadSupportPosix.cpp b/thirdparty/bullet/LinearMath/TaskScheduler/btThreadSupportPosix.cpp
index 50ca060dfec..02f4ed16310 100644
--- a/thirdparty/bullet/LinearMath/TaskScheduler/btThreadSupportPosix.cpp
+++ b/thirdparty/bullet/LinearMath/TaskScheduler/btThreadSupportPosix.cpp
@@ -1,3 +1,4 @@
+
 /*
 Bullet Continuous Collision Detection and Physics Library
 Copyright (c) 2003-2018 Erwin Coumans  http://bulletphysics.com
@@ -13,9 +14,7 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-#if BT_THREADSAFE && !defined( _WIN32 )
-
+#if BT_THREADSAFE && !defined(_WIN32)
 
 #include "LinearMath/btScalar.h"
 #include "LinearMath/btAlignedObjectArray.h"
@@ -27,14 +26,12 @@ subject to the following restrictions:
 #include <errno.h>
 #include <unistd.h>
 
-
 #ifndef _XOPEN_SOURCE
-#define _XOPEN_SOURCE 600 //for definition of pthread_barrier_t, see http://pages.cs.wisc.edu/~travitch/pthreads_primer.html
-#endif //_XOPEN_SOURCE
+#define _XOPEN_SOURCE 600  //for definition of pthread_barrier_t, see http://pages.cs.wisc.edu/~travitch/pthreads_primer.html
+#endif                     //_XOPEN_SOURCE
 #include <pthread.h>
 #include <semaphore.h>
-#include <unistd.h>   //for sysconf
-
+#include <unistd.h>  //for sysconf
 
 ///
 /// getNumHardwareThreads()
@@ -48,318 +45,309 @@ subject to the following restrictions:
 
 int btGetNumHardwareThreads()
 {
-    return btMin<int>(BT_MAX_THREAD_COUNT, std::thread::hardware_concurrency());
+	return btMin<int>(BT_MAX_THREAD_COUNT, std::thread::hardware_concurrency());
 }
 
 #else
 
 int btGetNumHardwareThreads()
 {
-    return btMin<int>(BT_MAX_THREAD_COUNT, sysconf( _SC_NPROCESSORS_ONLN ));
+	return btMin<int>(BT_MAX_THREAD_COUNT, sysconf(_SC_NPROCESSORS_ONLN));
 }
 
 #endif
 
-
 // btThreadSupportPosix helps to initialize/shutdown libspe2, start/stop SPU tasks and communication
 class btThreadSupportPosix : public btThreadSupportInterface
 {
 public:
-    struct btThreadStatus
-    {
-        int m_taskId;
-        int m_commandId;
-        int m_status;
+	struct btThreadStatus
+	{
+		int m_taskId;
+		int m_commandId;
+		int m_status;
 
-        ThreadFunc m_userThreadFunc;
-        void* m_userPtr; //for taskDesc etc
+		ThreadFunc m_userThreadFunc;
+		void* m_userPtr;  //for taskDesc etc
 
-        pthread_t thread;
-        //each tread will wait until this signal to start its work
-        sem_t* startSemaphore;
+		pthread_t thread;
+		//each tread will wait until this signal to start its work
+		sem_t* startSemaphore;
+		btCriticalSection* m_cs;
+		// this is a copy of m_mainSemaphore,
+		//each tread will signal once it is finished with its work
+		sem_t* m_mainSemaphore;
+		unsigned long threadUsed;
+	};
 
-        // this is a copy of m_mainSemaphore, 
-        //each tread will signal once it is finished with its work
-        sem_t* m_mainSemaphore;
-        unsigned long threadUsed;
-    };
 private:
-    typedef unsigned long long UINT64;
-
-    btAlignedObjectArray<btThreadStatus> m_activeThreadStatus;
-    // m_mainSemaphoresemaphore will signal, if and how many threads are finished with their work
-    sem_t* m_mainSemaphore;
-    int m_numThreads;
-    UINT64 m_startedThreadsMask;
-    void startThreads( const ConstructionInfo& threadInfo );
-    void stopThreads();
-    int waitForResponse();
+	typedef unsigned long long UINT64;
 
+	btAlignedObjectArray<btThreadStatus> m_activeThreadStatus;
+	// m_mainSemaphoresemaphore will signal, if and how many threads are finished with their work
+	sem_t* m_mainSemaphore;
+	int m_numThreads;
+	UINT64 m_startedThreadsMask;
+	void startThreads(const ConstructionInfo& threadInfo);
+	void stopThreads();
+	int waitForResponse();
+	btCriticalSection* m_cs;
 public:
-    btThreadSupportPosix( const ConstructionInfo& threadConstructionInfo );
-    virtual ~btThreadSupportPosix();
+	btThreadSupportPosix(const ConstructionInfo& threadConstructionInfo);
+	virtual ~btThreadSupportPosix();
 
-    virtual int getNumWorkerThreads() const BT_OVERRIDE { return m_numThreads; }
-    // TODO: return the number of logical processors sharing the first L3 cache
-    virtual int getCacheFriendlyNumThreads() const BT_OVERRIDE { return m_numThreads + 1; }
-    // TODO: detect if CPU has hyperthreading enabled
-    virtual int getLogicalToPhysicalCoreRatio() const BT_OVERRIDE { return 1; }
+	virtual int getNumWorkerThreads() const BT_OVERRIDE { return m_numThreads; }
+	// TODO: return the number of logical processors sharing the first L3 cache
+	virtual int getCacheFriendlyNumThreads() const BT_OVERRIDE { return m_numThreads + 1; }
+	// TODO: detect if CPU has hyperthreading enabled
+	virtual int getLogicalToPhysicalCoreRatio() const BT_OVERRIDE { return 1; }
 
-    virtual void runTask( int threadIndex, void* userData ) BT_OVERRIDE;
-    virtual void waitForAllTasks() BT_OVERRIDE;
+	virtual void runTask(int threadIndex, void* userData) BT_OVERRIDE;
+	virtual void waitForAllTasks() BT_OVERRIDE;
 
-    virtual btCriticalSection* createCriticalSection() BT_OVERRIDE;
-    virtual void deleteCriticalSection( btCriticalSection* criticalSection ) BT_OVERRIDE;
+	virtual btCriticalSection* createCriticalSection() BT_OVERRIDE;
+	virtual void deleteCriticalSection(btCriticalSection* criticalSection) BT_OVERRIDE;
 };
 
-
-#define checkPThreadFunction(returnValue) \
-    if(0 != returnValue) { \
-        printf("PThread problem at line %i in file %s: %i %d\n", __LINE__, __FILE__, returnValue, errno); \
-    }
+#define checkPThreadFunction(returnValue)                                                                 \
+	if (0 != returnValue)                                                                                 \
+	{                                                                                                     \
+		printf("PThread problem at line %i in file %s: %i %d\n", __LINE__, __FILE__, returnValue, errno); \
+	}
 
 // The number of threads should be equal to the number of available cores
 // Todo: each worker should be linked to a single core, using SetThreadIdealProcessor.
 
-
-btThreadSupportPosix::btThreadSupportPosix( const ConstructionInfo& threadConstructionInfo )
+btThreadSupportPosix::btThreadSupportPosix(const ConstructionInfo& threadConstructionInfo)
 {
-    startThreads( threadConstructionInfo );
+	m_cs = createCriticalSection();
+	startThreads(threadConstructionInfo);
 }
 
 // cleanup/shutdown Libspe2
 btThreadSupportPosix::~btThreadSupportPosix()
 {
-    stopThreads();
+	stopThreads();
+	deleteCriticalSection(m_cs);
+	m_cs=0;
 }
 
-#if (defined (__APPLE__))
+#if (defined(__APPLE__))
 #define NAMED_SEMAPHORES
 #endif
 
-
-static sem_t* createSem( const char* baseName )
+static sem_t* createSem(const char* baseName)
 {
-    static int semCount = 0;
+	static int semCount = 0;
 #ifdef NAMED_SEMAPHORES
-    /// Named semaphore begin
-    char name[ 32 ];
-    snprintf( name, 32, "/%8.s-%4.d-%4.4d", baseName, getpid(), semCount++ );
-    sem_t* tempSem = sem_open( name, O_CREAT, 0600, 0 );
+	/// Named semaphore begin
+	char name[32];
+	snprintf(name, 32, "/%8.s-%4.d-%4.4d", baseName, getpid(), semCount++);
+	sem_t* tempSem = sem_open(name, O_CREAT, 0600, 0);
 
-    if ( tempSem != reinterpret_cast<sem_t *>( SEM_FAILED ) )
-    {
-        //        printf("Created \"%s\" Semaphore %p\n", name, tempSem);
-    }
-    else
-    {
-        //printf("Error creating Semaphore %d\n", errno);
-        exit( -1 );
-    }
-    /// Named semaphore end
+	if (tempSem != reinterpret_cast<sem_t*>(SEM_FAILED))
+	{
+		//        printf("Created \"%s\" Semaphore %p\n", name, tempSem);
+	}
+	else
+	{
+		//printf("Error creating Semaphore %d\n", errno);
+		exit(-1);
+	}
+	/// Named semaphore end
 #else
-    sem_t* tempSem = new sem_t;
-    checkPThreadFunction( sem_init( tempSem, 0, 0 ) );
+	sem_t* tempSem = new sem_t;
+	checkPThreadFunction(sem_init(tempSem, 0, 0));
 #endif
-    return tempSem;
+	return tempSem;
 }
 
-static void destroySem( sem_t* semaphore )
+static void destroySem(sem_t* semaphore)
 {
 #ifdef NAMED_SEMAPHORES
-    checkPThreadFunction( sem_close( semaphore ) );
+	checkPThreadFunction(sem_close(semaphore));
 #else
-    checkPThreadFunction( sem_destroy( semaphore ) );
-    delete semaphore;
+	checkPThreadFunction(sem_destroy(semaphore));
+	delete semaphore;
 #endif
 }
 
-static void *threadFunction( void *argument )
+static void* threadFunction(void* argument)
 {
-    btThreadSupportPosix::btThreadStatus* status = ( btThreadSupportPosix::btThreadStatus* )argument;
+	btThreadSupportPosix::btThreadStatus* status = (btThreadSupportPosix::btThreadStatus*)argument;
 
-    while ( 1 )
-    {
-        checkPThreadFunction( sem_wait( status->startSemaphore ) );
-        void* userPtr = status->m_userPtr;
+	while (1)
+	{
+		checkPThreadFunction(sem_wait(status->startSemaphore));
+		void* userPtr = status->m_userPtr;
 
-        if ( userPtr )
-        {
-            btAssert( status->m_status );
-            status->m_userThreadFunc( userPtr );
-            status->m_status = 2;
-            checkPThreadFunction( sem_post( status->m_mainSemaphore ) );
-            status->threadUsed++;
-        }
-        else
-        {
-            //exit Thread
-            status->m_status = 3;
-            checkPThreadFunction( sem_post( status->m_mainSemaphore ) );
-            printf( "Thread with taskId %i exiting\n", status->m_taskId );
-            break;
-        }
-    }
+		if (userPtr)
+		{
+			btAssert(status->m_status);
+			status->m_userThreadFunc(userPtr);
+			status->m_cs->lock();
+			status->m_status = 2;
+			status->m_cs->unlock();
+			checkPThreadFunction(sem_post(status->m_mainSemaphore));
+			status->threadUsed++;
+		}
+		else
+		{
+			//exit Thread
+			status->m_cs->lock();
+			status->m_status = 3;
+			status->m_cs->unlock();
+			checkPThreadFunction(sem_post(status->m_mainSemaphore));
+			break;
+		}
+	}
 
-    printf( "Thread TERMINATED\n" );
-    return 0;
+	return 0;
 }
 
 ///send messages to SPUs
-void btThreadSupportPosix::runTask( int threadIndex, void* userData )
+void btThreadSupportPosix::runTask(int threadIndex, void* userData)
 {
-    ///we should spawn an SPU task here, and in 'waitForResponse' it should wait for response of the (one of) the first tasks that finished
-    btThreadStatus& threadStatus = m_activeThreadStatus[ threadIndex ];
-    btAssert( threadIndex >= 0 );
-    btAssert( threadIndex < m_activeThreadStatus.size() );
+	///we should spawn an SPU task here, and in 'waitForResponse' it should wait for response of the (one of) the first tasks that finished
+	btThreadStatus& threadStatus = m_activeThreadStatus[threadIndex];
+	btAssert(threadIndex >= 0);
+	btAssert(threadIndex < m_activeThreadStatus.size());
+	threadStatus.m_cs = m_cs;
+	threadStatus.m_commandId = 1;
+	threadStatus.m_status = 1;
+	threadStatus.m_userPtr = userData;
+	m_startedThreadsMask |= UINT64(1) << threadIndex;
 
-    threadStatus.m_commandId = 1;
-    threadStatus.m_status = 1;
-    threadStatus.m_userPtr = userData;
-    m_startedThreadsMask |= UINT64( 1 ) << threadIndex;
-
-    // fire event to start new task
-    checkPThreadFunction( sem_post( threadStatus.startSemaphore ) );
+	// fire event to start new task
+	checkPThreadFunction(sem_post(threadStatus.startSemaphore));
 }
 
-
 ///check for messages from SPUs
 int btThreadSupportPosix::waitForResponse()
 {
-    ///We should wait for (one of) the first tasks to finish (or other SPU messages), and report its response
-    ///A possible response can be 'yes, SPU handled it', or 'no, please do a PPU fallback'
+	///We should wait for (one of) the first tasks to finish (or other SPU messages), and report its response
+	///A possible response can be 'yes, SPU handled it', or 'no, please do a PPU fallback'
 
-    btAssert( m_activeThreadStatus.size() );
+	btAssert(m_activeThreadStatus.size());
 
-    // wait for any of the threads to finish
-    checkPThreadFunction( sem_wait( m_mainSemaphore ) );
-    // get at least one thread which has finished
-    size_t last = -1;
+	// wait for any of the threads to finish
+	checkPThreadFunction(sem_wait(m_mainSemaphore));
+	// get at least one thread which has finished
+	size_t last = -1;
 
-    for ( size_t t = 0; t < size_t( m_activeThreadStatus.size() ); ++t )
-    {
-        if ( 2 == m_activeThreadStatus[ t ].m_status )
-        {
-            last = t;
-            break;
-        }
-    }
+	for (size_t t = 0; t < size_t(m_activeThreadStatus.size()); ++t)
+	{
+		m_cs->lock();
+		bool hasFinished = (2 == m_activeThreadStatus[t].m_status);
+		m_cs->unlock(); 
+		if (hasFinished)
+		{
+			last = t;
+			break;
+		}
+	}
 
-    btThreadStatus& threadStatus = m_activeThreadStatus[ last ];
+	btThreadStatus& threadStatus = m_activeThreadStatus[last];
 
-    btAssert( threadStatus.m_status > 1 );
-    threadStatus.m_status = 0;
+	btAssert(threadStatus.m_status > 1);
+	threadStatus.m_status = 0;
 
-    // need to find an active spu
-    btAssert( last >= 0 );
-    m_startedThreadsMask &= ~( UINT64( 1 ) << last );
+	// need to find an active spu
+	btAssert(last >= 0);
+	m_startedThreadsMask &= ~(UINT64(1) << last);
 
-    return last;
+	return last;
 }
 
-
 void btThreadSupportPosix::waitForAllTasks()
 {
-    while ( m_startedThreadsMask )
-    {
-        waitForResponse();
-    }
+	while (m_startedThreadsMask)
+	{
+		waitForResponse();
+	}
 }
 
-
-void btThreadSupportPosix::startThreads( const ConstructionInfo& threadConstructionInfo )
+void btThreadSupportPosix::startThreads(const ConstructionInfo& threadConstructionInfo)
 {
-    m_numThreads = btGetNumHardwareThreads() - 1;  // main thread exists already
-    printf( "%s creating %i threads.\n", __FUNCTION__, m_numThreads );
-    m_activeThreadStatus.resize( m_numThreads );
-    m_startedThreadsMask = 0;
+	m_numThreads = btGetNumHardwareThreads() - 1;  // main thread exists already
+	m_activeThreadStatus.resize(m_numThreads);
+	m_startedThreadsMask = 0;
 
-    m_mainSemaphore = createSem( "main" );
-    //checkPThreadFunction(sem_wait(mainSemaphore));
+	m_mainSemaphore = createSem("main");
+	//checkPThreadFunction(sem_wait(mainSemaphore));
 
-    for ( int i = 0; i < m_numThreads; i++ )
-    {
-        printf( "starting thread %d\n", i );
-        btThreadStatus& threadStatus = m_activeThreadStatus[ i ];
-        threadStatus.startSemaphore = createSem( "threadLocal" );
-        checkPThreadFunction( pthread_create( &threadStatus.thread, NULL, &threadFunction, (void*) &threadStatus ) );
+	for (int i = 0; i < m_numThreads; i++)
+	{
+		btThreadStatus& threadStatus = m_activeThreadStatus[i];
+		threadStatus.startSemaphore = createSem("threadLocal");
+		threadStatus.m_userPtr = 0;
+		threadStatus.m_cs = m_cs;
+		threadStatus.m_taskId = i;
+		threadStatus.m_commandId = 0;
+		threadStatus.m_status = 0;
+		threadStatus.m_mainSemaphore = m_mainSemaphore;
+		threadStatus.m_userThreadFunc = threadConstructionInfo.m_userThreadFunc;
+		threadStatus.threadUsed = 0;
+		checkPThreadFunction(pthread_create(&threadStatus.thread, NULL, &threadFunction, (void*)&threadStatus));
 
-        threadStatus.m_userPtr = 0;
-        threadStatus.m_taskId = i;
-        threadStatus.m_commandId = 0;
-        threadStatus.m_status = 0;
-        threadStatus.m_mainSemaphore = m_mainSemaphore;
-        threadStatus.m_userThreadFunc = threadConstructionInfo.m_userThreadFunc;
-        threadStatus.threadUsed = 0;
-
-        printf( "started thread %d \n", i );
-    }
+	}
 }
 
 ///tell the task scheduler we are done with the SPU tasks
 void btThreadSupportPosix::stopThreads()
 {
-    for ( size_t t = 0; t < size_t( m_activeThreadStatus.size() ); ++t )
-    {
-        btThreadStatus& threadStatus = m_activeThreadStatus[ t ];
-        printf( "%s: Thread %i used: %ld\n", __FUNCTION__, int( t ), threadStatus.threadUsed );
+	for (size_t t = 0; t < size_t(m_activeThreadStatus.size()); ++t)
+	{
+		btThreadStatus& threadStatus = m_activeThreadStatus[t];
 
-        threadStatus.m_userPtr = 0;
-        checkPThreadFunction( sem_post( threadStatus.startSemaphore ) );
-        checkPThreadFunction( sem_wait( m_mainSemaphore ) );
+		threadStatus.m_userPtr = 0;
+		checkPThreadFunction(sem_post(threadStatus.startSemaphore));
+		checkPThreadFunction(sem_wait(m_mainSemaphore));
 
-        printf( "destroy semaphore\n" );
-        destroySem( threadStatus.startSemaphore );
-        printf( "semaphore destroyed\n" );
-        checkPThreadFunction( pthread_join( threadStatus.thread, 0 ) );
-
-    }
-    printf( "destroy main semaphore\n" );
-    destroySem( m_mainSemaphore );
-    printf( "main semaphore destroyed\n" );
-    m_activeThreadStatus.clear();
+		destroySem(threadStatus.startSemaphore);
+		checkPThreadFunction(pthread_join(threadStatus.thread, 0));
+	}
+	destroySem(m_mainSemaphore);
+	m_activeThreadStatus.clear();
 }
 
 class btCriticalSectionPosix : public btCriticalSection
 {
-    pthread_mutex_t m_mutex;
+	pthread_mutex_t m_mutex;
 
 public:
-    btCriticalSectionPosix()
-    {
-        pthread_mutex_init( &m_mutex, NULL );
-    }
-    virtual ~btCriticalSectionPosix()
-    {
-        pthread_mutex_destroy( &m_mutex );
-    }
+	btCriticalSectionPosix()
+	{
+		pthread_mutex_init(&m_mutex, NULL);
+	}
+	virtual ~btCriticalSectionPosix()
+	{
+		pthread_mutex_destroy(&m_mutex);
+	}
 
-    virtual void lock()
-    {
-        pthread_mutex_lock( &m_mutex );
-    }
-    virtual void unlock()
-    {
-        pthread_mutex_unlock( &m_mutex );
-    }
+	virtual void lock()
+	{
+		pthread_mutex_lock(&m_mutex);
+	}
+	virtual void unlock()
+	{
+		pthread_mutex_unlock(&m_mutex);
+	}
 };
 
-
 btCriticalSection* btThreadSupportPosix::createCriticalSection()
 {
-    return new btCriticalSectionPosix();
+	return new btCriticalSectionPosix();
 }
 
-void btThreadSupportPosix::deleteCriticalSection( btCriticalSection* cs )
+void btThreadSupportPosix::deleteCriticalSection(btCriticalSection* cs)
 {
-    delete cs;
+	delete cs;
 }
 
-
-btThreadSupportInterface* btThreadSupportInterface::create( const ConstructionInfo& info )
+btThreadSupportInterface* btThreadSupportInterface::create(const ConstructionInfo& info)
 {
-    return new btThreadSupportPosix( info );
+	return new btThreadSupportPosix(info);
 }
 
-#endif // BT_THREADSAFE && !defined( _WIN32 )
-
+#endif  // BT_THREADSAFE && !defined( _WIN32 )
diff --git a/thirdparty/bullet/LinearMath/TaskScheduler/btThreadSupportWin32.cpp b/thirdparty/bullet/LinearMath/TaskScheduler/btThreadSupportWin32.cpp
index 00edac650b4..922e449cce2 100644
--- a/thirdparty/bullet/LinearMath/TaskScheduler/btThreadSupportWin32.cpp
+++ b/thirdparty/bullet/LinearMath/TaskScheduler/btThreadSupportWin32.cpp
@@ -13,7 +13,7 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-#if defined( _WIN32 ) &&  BT_THREADSAFE
+#if defined(_WIN32) && BT_THREADSAFE
 
 #include "LinearMath/btScalar.h"
 #include "LinearMath/btMinMax.h"
@@ -23,450 +23,430 @@ subject to the following restrictions:
 #include <windows.h>
 #include <stdio.h>
 
-
 struct btProcessorInfo
 {
-    int numLogicalProcessors;
-    int numCores;
-    int numNumaNodes;
-    int numL1Cache;
-    int numL2Cache;
-    int numL3Cache;
-    int numPhysicalPackages;
-    static const int maxNumTeamMasks = 32;
-    int numTeamMasks;
-    UINT64 processorTeamMasks[ maxNumTeamMasks ];
+	int numLogicalProcessors;
+	int numCores;
+	int numNumaNodes;
+	int numL1Cache;
+	int numL2Cache;
+	int numL3Cache;
+	int numPhysicalPackages;
+	static const int maxNumTeamMasks = 32;
+	int numTeamMasks;
+	UINT64 processorTeamMasks[maxNumTeamMasks];
 };
 
-UINT64 getProcessorTeamMask( const btProcessorInfo& procInfo, int procId )
+UINT64 getProcessorTeamMask(const btProcessorInfo& procInfo, int procId)
 {
-    UINT64 procMask = UINT64( 1 ) << procId;
-    for ( int i = 0; i < procInfo.numTeamMasks; ++i )
-    {
-        if ( procMask & procInfo.processorTeamMasks[ i ] )
-        {
-            return procInfo.processorTeamMasks[ i ];
-        }
-    }
-    return 0;
+	UINT64 procMask = UINT64(1) << procId;
+	for (int i = 0; i < procInfo.numTeamMasks; ++i)
+	{
+		if (procMask & procInfo.processorTeamMasks[i])
+		{
+			return procInfo.processorTeamMasks[i];
+		}
+	}
+	return 0;
 }
 
-int getProcessorTeamIndex( const btProcessorInfo& procInfo, int procId )
+int getProcessorTeamIndex(const btProcessorInfo& procInfo, int procId)
 {
-    UINT64 procMask = UINT64( 1 ) << procId;
-    for ( int i = 0; i < procInfo.numTeamMasks; ++i )
-    {
-        if ( procMask & procInfo.processorTeamMasks[ i ] )
-        {
-            return i;
-        }
-    }
-    return -1;
+	UINT64 procMask = UINT64(1) << procId;
+	for (int i = 0; i < procInfo.numTeamMasks; ++i)
+	{
+		if (procMask & procInfo.processorTeamMasks[i])
+		{
+			return i;
+		}
+	}
+	return -1;
 }
 
-int countSetBits( ULONG64 bits )
+int countSetBits(ULONG64 bits)
 {
-    int count = 0;
-    while ( bits )
-    {
-        if ( bits & 1 )
-        {
-            count++;
-        }
-        bits >>= 1;
-    }
-    return count;
+	int count = 0;
+	while (bits)
+	{
+		if (bits & 1)
+		{
+			count++;
+		}
+		bits >>= 1;
+	}
+	return count;
 }
 
+typedef BOOL(WINAPI* Pfn_GetLogicalProcessorInformation)(PSYSTEM_LOGICAL_PROCESSOR_INFORMATION, PDWORD);
 
-typedef BOOL( WINAPI *Pfn_GetLogicalProcessorInformation )( PSYSTEM_LOGICAL_PROCESSOR_INFORMATION, PDWORD );
-
-
-void getProcessorInformation( btProcessorInfo* procInfo )
+void getProcessorInformation(btProcessorInfo* procInfo)
 {
-    memset( procInfo, 0, sizeof( *procInfo ) );
-    Pfn_GetLogicalProcessorInformation getLogicalProcInfo =
-        (Pfn_GetLogicalProcessorInformation) GetProcAddress( GetModuleHandle( TEXT( "kernel32" ) ), "GetLogicalProcessorInformation" );
-    if ( getLogicalProcInfo == NULL )
-    {
-        // no info
-        return;
-    }
-    PSYSTEM_LOGICAL_PROCESSOR_INFORMATION buf = NULL;
-    DWORD bufSize = 0;
-    while ( true )
-    {
-        if ( getLogicalProcInfo( buf, &bufSize ) )
-        {
-            break;
-        }
-        else
-        {
-            if ( GetLastError() == ERROR_INSUFFICIENT_BUFFER )
-            {
-                if ( buf )
-                {
-                    free( buf );
-                }
-                buf = (PSYSTEM_LOGICAL_PROCESSOR_INFORMATION) malloc( bufSize );
-            }
-        }
-    }
+	memset(procInfo, 0, sizeof(*procInfo));
+	Pfn_GetLogicalProcessorInformation getLogicalProcInfo =
+		(Pfn_GetLogicalProcessorInformation)GetProcAddress(GetModuleHandle(TEXT("kernel32")), "GetLogicalProcessorInformation");
+	if (getLogicalProcInfo == NULL)
+	{
+		// no info
+		return;
+	}
+	PSYSTEM_LOGICAL_PROCESSOR_INFORMATION buf = NULL;
+	DWORD bufSize = 0;
+	while (true)
+	{
+		if (getLogicalProcInfo(buf, &bufSize))
+		{
+			break;
+		}
+		else
+		{
+			if (GetLastError() == ERROR_INSUFFICIENT_BUFFER)
+			{
+				if (buf)
+				{
+					free(buf);
+				}
+				buf = (PSYSTEM_LOGICAL_PROCESSOR_INFORMATION)malloc(bufSize);
+			}
+		}
+	}
 
-    int len = bufSize / sizeof( *buf );
-    for ( int i = 0; i < len; ++i )
-    {
-        PSYSTEM_LOGICAL_PROCESSOR_INFORMATION info = buf + i;
-        switch ( info->Relationship )
-        {
-        case RelationNumaNode:
-            procInfo->numNumaNodes++;
-            break;
+	int len = bufSize / sizeof(*buf);
+	for (int i = 0; i < len; ++i)
+	{
+		PSYSTEM_LOGICAL_PROCESSOR_INFORMATION info = buf + i;
+		switch (info->Relationship)
+		{
+			case RelationNumaNode:
+				procInfo->numNumaNodes++;
+				break;
 
-        case RelationProcessorCore:
-            procInfo->numCores++;
-            procInfo->numLogicalProcessors += countSetBits( info->ProcessorMask );
-            break;
+			case RelationProcessorCore:
+				procInfo->numCores++;
+				procInfo->numLogicalProcessors += countSetBits(info->ProcessorMask);
+				break;
 
-        case RelationCache:
-            if ( info->Cache.Level == 1 )
-            {
-                procInfo->numL1Cache++;
-            }
-            else if ( info->Cache.Level == 2 )
-            {
-                procInfo->numL2Cache++;
-            }
-            else if ( info->Cache.Level == 3 )
-            {
-                procInfo->numL3Cache++;
-                // processors that share L3 cache are considered to be on the same team
-                // because they can more easily work together on the same data.
-                // Large performance penalties will occur if 2 or more threads from different
-                // teams attempt to frequently read and modify the same cache lines.
-                //
-                // On the AMD Ryzen 7 CPU for example, the 8 cores on the CPU are split into
-                // 2 CCX units of 4 cores each. Each CCX has a separate L3 cache, so if both
-                // CCXs are operating on the same data, many cycles will be spent keeping the
-                // two caches coherent.
-                if ( procInfo->numTeamMasks < btProcessorInfo::maxNumTeamMasks )
-                {
-                    procInfo->processorTeamMasks[ procInfo->numTeamMasks ] = info->ProcessorMask;
-                    procInfo->numTeamMasks++;
-                }
-            }
-            break;
+			case RelationCache:
+				if (info->Cache.Level == 1)
+				{
+					procInfo->numL1Cache++;
+				}
+				else if (info->Cache.Level == 2)
+				{
+					procInfo->numL2Cache++;
+				}
+				else if (info->Cache.Level == 3)
+				{
+					procInfo->numL3Cache++;
+					// processors that share L3 cache are considered to be on the same team
+					// because they can more easily work together on the same data.
+					// Large performance penalties will occur if 2 or more threads from different
+					// teams attempt to frequently read and modify the same cache lines.
+					//
+					// On the AMD Ryzen 7 CPU for example, the 8 cores on the CPU are split into
+					// 2 CCX units of 4 cores each. Each CCX has a separate L3 cache, so if both
+					// CCXs are operating on the same data, many cycles will be spent keeping the
+					// two caches coherent.
+					if (procInfo->numTeamMasks < btProcessorInfo::maxNumTeamMasks)
+					{
+						procInfo->processorTeamMasks[procInfo->numTeamMasks] = info->ProcessorMask;
+						procInfo->numTeamMasks++;
+					}
+				}
+				break;
 
-        case RelationProcessorPackage:
-            procInfo->numPhysicalPackages++;
-            break;
-        }
-    }
-    free( buf );
+			case RelationProcessorPackage:
+				procInfo->numPhysicalPackages++;
+				break;
+		}
+	}
+	free(buf);
 }
 
-
-
 ///btThreadSupportWin32 helps to initialize/shutdown libspe2, start/stop SPU tasks and communication
 class btThreadSupportWin32 : public btThreadSupportInterface
 {
 public:
-    struct btThreadStatus
-    {
-        int m_taskId;
-        int m_commandId;
-        int m_status;
+	struct btThreadStatus
+	{
+		int m_taskId;
+		int m_commandId;
+		int m_status;
 
-        ThreadFunc m_userThreadFunc;
-        void* m_userPtr; //for taskDesc etc
+		ThreadFunc m_userThreadFunc;
+		void* m_userPtr;  //for taskDesc etc
 
-        void* m_threadHandle; //this one is calling 'Win32ThreadFunc'
+		void* m_threadHandle;  //this one is calling 'Win32ThreadFunc'
 
-        void* m_eventStartHandle;
-        char m_eventStartHandleName[ 32 ];
+		void* m_eventStartHandle;
+		char m_eventStartHandleName[32];
 
-        void* m_eventCompleteHandle;
-        char m_eventCompleteHandleName[ 32 ];
-    };
+		void* m_eventCompleteHandle;
+		char m_eventCompleteHandleName[32];
+	};
 
 private:
-    btAlignedObjectArray<btThreadStatus> m_activeThreadStatus;
-    btAlignedObjectArray<void*> m_completeHandles;
-    int m_numThreads;
-    DWORD_PTR m_startedThreadMask;
-    btProcessorInfo m_processorInfo;
+	btAlignedObjectArray<btThreadStatus> m_activeThreadStatus;
+	btAlignedObjectArray<void*> m_completeHandles;
+	int m_numThreads;
+	DWORD_PTR m_startedThreadMask;
+	btProcessorInfo m_processorInfo;
 
-    void startThreads( const ConstructionInfo& threadInfo );
-    void stopThreads();
-    int waitForResponse();
+	void startThreads(const ConstructionInfo& threadInfo);
+	void stopThreads();
+	int waitForResponse();
 
 public:
+	btThreadSupportWin32(const ConstructionInfo& threadConstructionInfo);
+	virtual ~btThreadSupportWin32();
 
-    btThreadSupportWin32( const ConstructionInfo& threadConstructionInfo );
-    virtual ~btThreadSupportWin32();
+	virtual int getNumWorkerThreads() const BT_OVERRIDE { return m_numThreads; }
+	virtual int getCacheFriendlyNumThreads() const BT_OVERRIDE { return countSetBits(m_processorInfo.processorTeamMasks[0]); }
+	virtual int getLogicalToPhysicalCoreRatio() const BT_OVERRIDE { return m_processorInfo.numLogicalProcessors / m_processorInfo.numCores; }
 
-    virtual int getNumWorkerThreads() const BT_OVERRIDE { return m_numThreads; }
-    virtual int getCacheFriendlyNumThreads() const BT_OVERRIDE { return countSetBits(m_processorInfo.processorTeamMasks[0]); }
-    virtual int getLogicalToPhysicalCoreRatio() const BT_OVERRIDE { return m_processorInfo.numLogicalProcessors / m_processorInfo.numCores; }
+	virtual void runTask(int threadIndex, void* userData) BT_OVERRIDE;
+	virtual void waitForAllTasks() BT_OVERRIDE;
 
-    virtual void runTask( int threadIndex, void* userData ) BT_OVERRIDE;
-    virtual void waitForAllTasks() BT_OVERRIDE;
-
-    virtual btCriticalSection* createCriticalSection() BT_OVERRIDE;
-    virtual void deleteCriticalSection( btCriticalSection* criticalSection ) BT_OVERRIDE;
+	virtual btCriticalSection* createCriticalSection() BT_OVERRIDE;
+	virtual void deleteCriticalSection(btCriticalSection* criticalSection) BT_OVERRIDE;
 };
 
-
-btThreadSupportWin32::btThreadSupportWin32( const ConstructionInfo & threadConstructionInfo )
+btThreadSupportWin32::btThreadSupportWin32(const ConstructionInfo& threadConstructionInfo)
 {
-    startThreads( threadConstructionInfo );
+	startThreads(threadConstructionInfo);
 }
 
-
 btThreadSupportWin32::~btThreadSupportWin32()
 {
-    stopThreads();
+	stopThreads();
 }
 
-
-DWORD WINAPI win32threadStartFunc( LPVOID lpParam )
+DWORD WINAPI win32threadStartFunc(LPVOID lpParam)
 {
-    btThreadSupportWin32::btThreadStatus* status = ( btThreadSupportWin32::btThreadStatus* )lpParam;
+	btThreadSupportWin32::btThreadStatus* status = (btThreadSupportWin32::btThreadStatus*)lpParam;
 
-    while ( 1 )
-    {
-        WaitForSingleObject( status->m_eventStartHandle, INFINITE );
-        void* userPtr = status->m_userPtr;
+	while (1)
+	{
+		WaitForSingleObject(status->m_eventStartHandle, INFINITE);
+		void* userPtr = status->m_userPtr;
 
-        if ( userPtr )
-        {
-            btAssert( status->m_status );
-            status->m_userThreadFunc( userPtr );
-            status->m_status = 2;
-            SetEvent( status->m_eventCompleteHandle );
-        }
-        else
-        {
-            //exit Thread
-            status->m_status = 3;
-            printf( "Thread with taskId %i with handle %p exiting\n", status->m_taskId, status->m_threadHandle );
-            SetEvent( status->m_eventCompleteHandle );
-            break;
-        }
-    }
-    printf( "Thread TERMINATED\n" );
-    return 0;
+		if (userPtr)
+		{
+			btAssert(status->m_status);
+			status->m_userThreadFunc(userPtr);
+			status->m_status = 2;
+			SetEvent(status->m_eventCompleteHandle);
+		}
+		else
+		{
+			//exit Thread
+			status->m_status = 3;
+			printf("Thread with taskId %i with handle %p exiting\n", status->m_taskId, status->m_threadHandle);
+			SetEvent(status->m_eventCompleteHandle);
+			break;
+		}
+	}
+	printf("Thread TERMINATED\n");
+	return 0;
 }
 
-
-void btThreadSupportWin32::runTask( int threadIndex, void* userData )
+void btThreadSupportWin32::runTask(int threadIndex, void* userData)
 {
-    btThreadStatus& threadStatus = m_activeThreadStatus[ threadIndex ];
-    btAssert( threadIndex >= 0 );
-    btAssert( int( threadIndex ) < m_activeThreadStatus.size() );
+	btThreadStatus& threadStatus = m_activeThreadStatus[threadIndex];
+	btAssert(threadIndex >= 0);
+	btAssert(int(threadIndex) < m_activeThreadStatus.size());
 
-    threadStatus.m_commandId = 1;
-    threadStatus.m_status = 1;
-    threadStatus.m_userPtr = userData;
-    m_startedThreadMask |= DWORD_PTR( 1 ) << threadIndex;
+	threadStatus.m_commandId = 1;
+	threadStatus.m_status = 1;
+	threadStatus.m_userPtr = userData;
+	m_startedThreadMask |= DWORD_PTR(1) << threadIndex;
 
-    ///fire event to start new task
-    SetEvent( threadStatus.m_eventStartHandle );
+	///fire event to start new task
+	SetEvent(threadStatus.m_eventStartHandle);
 }
 
-
 int btThreadSupportWin32::waitForResponse()
 {
-    btAssert( m_activeThreadStatus.size() );
+	btAssert(m_activeThreadStatus.size());
 
-    int last = -1;
-    DWORD res = WaitForMultipleObjects( m_completeHandles.size(), &m_completeHandles[ 0 ], FALSE, INFINITE );
-    btAssert( res != WAIT_FAILED );
-    last = res - WAIT_OBJECT_0;
+	int last = -1;
+	DWORD res = WaitForMultipleObjects(m_completeHandles.size(), &m_completeHandles[0], FALSE, INFINITE);
+	btAssert(res != WAIT_FAILED);
+	last = res - WAIT_OBJECT_0;
 
-    btThreadStatus& threadStatus = m_activeThreadStatus[ last ];
-    btAssert( threadStatus.m_threadHandle );
-    btAssert( threadStatus.m_eventCompleteHandle );
+	btThreadStatus& threadStatus = m_activeThreadStatus[last];
+	btAssert(threadStatus.m_threadHandle);
+	btAssert(threadStatus.m_eventCompleteHandle);
 
-    //WaitForSingleObject(threadStatus.m_eventCompleteHandle, INFINITE);
-    btAssert( threadStatus.m_status > 1 );
-    threadStatus.m_status = 0;
+	//WaitForSingleObject(threadStatus.m_eventCompleteHandle, INFINITE);
+	btAssert(threadStatus.m_status > 1);
+	threadStatus.m_status = 0;
 
-    ///need to find an active spu
-    btAssert( last >= 0 );
-    m_startedThreadMask &= ~( DWORD_PTR( 1 ) << last );
+	///need to find an active spu
+	btAssert(last >= 0);
+	m_startedThreadMask &= ~(DWORD_PTR(1) << last);
 
-    return last;
+	return last;
 }
 
-
 void btThreadSupportWin32::waitForAllTasks()
 {
-    while ( m_startedThreadMask )
-    {
-        waitForResponse();
-    }
+	while (m_startedThreadMask)
+	{
+		waitForResponse();
+	}
 }
 
-
-void btThreadSupportWin32::startThreads( const ConstructionInfo& threadConstructionInfo )
+void btThreadSupportWin32::startThreads(const ConstructionInfo& threadConstructionInfo)
 {
-    static int uniqueId = 0;
-    uniqueId++;
-    btProcessorInfo& procInfo = m_processorInfo;
-    getProcessorInformation( &procInfo );
-    DWORD_PTR dwProcessAffinityMask = 0;
-    DWORD_PTR dwSystemAffinityMask = 0;
-    if ( !GetProcessAffinityMask( GetCurrentProcess(), &dwProcessAffinityMask, &dwSystemAffinityMask ) )
-    {
-        dwProcessAffinityMask = 0;
-    }
-    ///The number of threads should be equal to the number of available cores - 1
-    m_numThreads = btMin(procInfo.numLogicalProcessors, int(BT_MAX_THREAD_COUNT)) - 1; // cap to max thread count (-1 because main thread already exists)
+	static int uniqueId = 0;
+	uniqueId++;
+	btProcessorInfo& procInfo = m_processorInfo;
+	getProcessorInformation(&procInfo);
+	DWORD_PTR dwProcessAffinityMask = 0;
+	DWORD_PTR dwSystemAffinityMask = 0;
+	if (!GetProcessAffinityMask(GetCurrentProcess(), &dwProcessAffinityMask, &dwSystemAffinityMask))
+	{
+		dwProcessAffinityMask = 0;
+	}
+	///The number of threads should be equal to the number of available cores - 1
+	m_numThreads = btMin(procInfo.numLogicalProcessors, int(BT_MAX_THREAD_COUNT)) - 1;  // cap to max thread count (-1 because main thread already exists)
 
-    m_activeThreadStatus.resize( m_numThreads );
-    m_completeHandles.resize( m_numThreads );
-    m_startedThreadMask = 0;
+	m_activeThreadStatus.resize(m_numThreads);
+	m_completeHandles.resize(m_numThreads);
+	m_startedThreadMask = 0;
 
-    // set main thread affinity
-    if ( DWORD_PTR mask = dwProcessAffinityMask & getProcessorTeamMask( procInfo, 0 ))
-    {
-        SetThreadAffinityMask( GetCurrentThread(), mask );
-        SetThreadIdealProcessor( GetCurrentThread(), 0 );
-    }
+	// set main thread affinity
+	if (DWORD_PTR mask = dwProcessAffinityMask & getProcessorTeamMask(procInfo, 0))
+	{
+		SetThreadAffinityMask(GetCurrentThread(), mask);
+		SetThreadIdealProcessor(GetCurrentThread(), 0);
+	}
 
-    for ( int i = 0; i < m_numThreads; i++ )
-    {
-        printf( "starting thread %d\n", i );
+	for (int i = 0; i < m_numThreads; i++)
+	{
+		printf("starting thread %d\n", i);
 
-        btThreadStatus& threadStatus = m_activeThreadStatus[ i ];
+		btThreadStatus& threadStatus = m_activeThreadStatus[i];
 
-        LPSECURITY_ATTRIBUTES lpThreadAttributes = NULL;
-        SIZE_T dwStackSize = threadConstructionInfo.m_threadStackSize;
-        LPTHREAD_START_ROUTINE lpStartAddress = &win32threadStartFunc;
-        LPVOID lpParameter = &threadStatus;
-        DWORD dwCreationFlags = 0;
-        LPDWORD lpThreadId = 0;
+		LPSECURITY_ATTRIBUTES lpThreadAttributes = NULL;
+		SIZE_T dwStackSize = threadConstructionInfo.m_threadStackSize;
+		LPTHREAD_START_ROUTINE lpStartAddress = &win32threadStartFunc;
+		LPVOID lpParameter = &threadStatus;
+		DWORD dwCreationFlags = 0;
+		LPDWORD lpThreadId = 0;
 
-        threadStatus.m_userPtr = 0;
+		threadStatus.m_userPtr = 0;
 
-        sprintf( threadStatus.m_eventStartHandleName, "es%.8s%d%d", threadConstructionInfo.m_uniqueName, uniqueId, i );
-        threadStatus.m_eventStartHandle = CreateEventA( 0, false, false, threadStatus.m_eventStartHandleName );
+		sprintf(threadStatus.m_eventStartHandleName, "es%.8s%d%d", threadConstructionInfo.m_uniqueName, uniqueId, i);
+		threadStatus.m_eventStartHandle = CreateEventA(0, false, false, threadStatus.m_eventStartHandleName);
 
-        sprintf( threadStatus.m_eventCompleteHandleName, "ec%.8s%d%d", threadConstructionInfo.m_uniqueName, uniqueId, i );
-        threadStatus.m_eventCompleteHandle = CreateEventA( 0, false, false, threadStatus.m_eventCompleteHandleName );
+		sprintf(threadStatus.m_eventCompleteHandleName, "ec%.8s%d%d", threadConstructionInfo.m_uniqueName, uniqueId, i);
+		threadStatus.m_eventCompleteHandle = CreateEventA(0, false, false, threadStatus.m_eventCompleteHandleName);
 
-        m_completeHandles[ i ] = threadStatus.m_eventCompleteHandle;
+		m_completeHandles[i] = threadStatus.m_eventCompleteHandle;
 
-        HANDLE handle = CreateThread( lpThreadAttributes, dwStackSize, lpStartAddress, lpParameter, dwCreationFlags, lpThreadId );
-        //SetThreadPriority( handle, THREAD_PRIORITY_HIGHEST );
-        // highest priority -- can cause erratic performance when numThreads > numCores
-        //                     we don't want worker threads to be higher priority than the main thread or the main thread could get
-        //                     totally shut out and unable to tell the workers to stop
-        //SetThreadPriority( handle, THREAD_PRIORITY_BELOW_NORMAL );
+		HANDLE handle = CreateThread(lpThreadAttributes, dwStackSize, lpStartAddress, lpParameter, dwCreationFlags, lpThreadId);
+		//SetThreadPriority( handle, THREAD_PRIORITY_HIGHEST );
+		// highest priority -- can cause erratic performance when numThreads > numCores
+		//                     we don't want worker threads to be higher priority than the main thread or the main thread could get
+		//                     totally shut out and unable to tell the workers to stop
+		//SetThreadPriority( handle, THREAD_PRIORITY_BELOW_NORMAL );
 
-        {
-            int processorId = i + 1;  // leave processor 0 for main thread
-            DWORD_PTR teamMask = getProcessorTeamMask( procInfo, processorId );
-            if ( teamMask )
-            {
-                // bind each thread to only execute on processors of it's assigned team
-                //  - for single-socket Intel x86 CPUs this has no effect (only a single, shared L3 cache so there is only 1 team)
-                //  - for multi-socket Intel this will keep threads from migrating from one socket to another
-                //  - for AMD Ryzen this will keep threads from migrating from one CCX to another
-                DWORD_PTR mask = teamMask & dwProcessAffinityMask;
-                if ( mask )
-                {
-                    SetThreadAffinityMask( handle, mask );
-                }
-            }
-            SetThreadIdealProcessor( handle, processorId );
-        }
+		{
+			int processorId = i + 1;  // leave processor 0 for main thread
+			DWORD_PTR teamMask = getProcessorTeamMask(procInfo, processorId);
+			if (teamMask)
+			{
+				// bind each thread to only execute on processors of it's assigned team
+				//  - for single-socket Intel x86 CPUs this has no effect (only a single, shared L3 cache so there is only 1 team)
+				//  - for multi-socket Intel this will keep threads from migrating from one socket to another
+				//  - for AMD Ryzen this will keep threads from migrating from one CCX to another
+				DWORD_PTR mask = teamMask & dwProcessAffinityMask;
+				if (mask)
+				{
+					SetThreadAffinityMask(handle, mask);
+				}
+			}
+			SetThreadIdealProcessor(handle, processorId);
+		}
 
-        threadStatus.m_taskId = i;
-        threadStatus.m_commandId = 0;
-        threadStatus.m_status = 0;
-        threadStatus.m_threadHandle = handle;
-        threadStatus.m_userThreadFunc = threadConstructionInfo.m_userThreadFunc;
+		threadStatus.m_taskId = i;
+		threadStatus.m_commandId = 0;
+		threadStatus.m_status = 0;
+		threadStatus.m_threadHandle = handle;
+		threadStatus.m_userThreadFunc = threadConstructionInfo.m_userThreadFunc;
 
-        printf( "started %s thread %d with threadHandle %p\n", threadConstructionInfo.m_uniqueName, i, handle );
-    }
+		printf("started %s thread %d with threadHandle %p\n", threadConstructionInfo.m_uniqueName, i, handle);
+	}
 }
 
 ///tell the task scheduler we are done with the SPU tasks
 void btThreadSupportWin32::stopThreads()
 {
-    for ( int i = 0; i < m_activeThreadStatus.size(); i++ )
-    {
-        btThreadStatus& threadStatus = m_activeThreadStatus[ i ];
-        if ( threadStatus.m_status > 0 )
-        {
-            WaitForSingleObject( threadStatus.m_eventCompleteHandle, INFINITE );
-        }
+	for (int i = 0; i < m_activeThreadStatus.size(); i++)
+	{
+		btThreadStatus& threadStatus = m_activeThreadStatus[i];
+		if (threadStatus.m_status > 0)
+		{
+			WaitForSingleObject(threadStatus.m_eventCompleteHandle, INFINITE);
+		}
 
-        threadStatus.m_userPtr = NULL;
-        SetEvent( threadStatus.m_eventStartHandle );
-        WaitForSingleObject( threadStatus.m_eventCompleteHandle, INFINITE );
+		threadStatus.m_userPtr = NULL;
+		SetEvent(threadStatus.m_eventStartHandle);
+		WaitForSingleObject(threadStatus.m_eventCompleteHandle, INFINITE);
 
-        CloseHandle( threadStatus.m_eventCompleteHandle );
-        CloseHandle( threadStatus.m_eventStartHandle );
-        CloseHandle( threadStatus.m_threadHandle );
+		CloseHandle(threadStatus.m_eventCompleteHandle);
+		CloseHandle(threadStatus.m_eventStartHandle);
+		CloseHandle(threadStatus.m_threadHandle);
+	}
 
-    }
-
-    m_activeThreadStatus.clear();
-    m_completeHandles.clear();
+	m_activeThreadStatus.clear();
+	m_completeHandles.clear();
 }
 
-
 class btWin32CriticalSection : public btCriticalSection
 {
 private:
-    CRITICAL_SECTION mCriticalSection;
+	CRITICAL_SECTION mCriticalSection;
 
 public:
-    btWin32CriticalSection()
-    {
-        InitializeCriticalSection( &mCriticalSection );
-    }
+	btWin32CriticalSection()
+	{
+		InitializeCriticalSection(&mCriticalSection);
+	}
 
-    ~btWin32CriticalSection()
-    {
-        DeleteCriticalSection( &mCriticalSection );
-    }
+	~btWin32CriticalSection()
+	{
+		DeleteCriticalSection(&mCriticalSection);
+	}
 
-    void lock()
-    {
-        EnterCriticalSection( &mCriticalSection );
-    }
+	void lock()
+	{
+		EnterCriticalSection(&mCriticalSection);
+	}
 
-    void unlock()
-    {
-        LeaveCriticalSection( &mCriticalSection );
-    }
+	void unlock()
+	{
+		LeaveCriticalSection(&mCriticalSection);
+	}
 };
 
-
 btCriticalSection* btThreadSupportWin32::createCriticalSection()
 {
-    unsigned char* mem = (unsigned char*) btAlignedAlloc( sizeof( btWin32CriticalSection ), 16 );
-    btWin32CriticalSection* cs = new( mem ) btWin32CriticalSection();
-    return cs;
+	unsigned char* mem = (unsigned char*)btAlignedAlloc(sizeof(btWin32CriticalSection), 16);
+	btWin32CriticalSection* cs = new (mem) btWin32CriticalSection();
+	return cs;
 }
 
-void btThreadSupportWin32::deleteCriticalSection( btCriticalSection* criticalSection )
+void btThreadSupportWin32::deleteCriticalSection(btCriticalSection* criticalSection)
 {
-    criticalSection->~btCriticalSection();
-    btAlignedFree( criticalSection );
+	criticalSection->~btCriticalSection();
+	btAlignedFree(criticalSection);
 }
 
-
-btThreadSupportInterface* btThreadSupportInterface::create( const ConstructionInfo& info )
+btThreadSupportInterface* btThreadSupportInterface::create(const ConstructionInfo& info)
 {
-    return new btThreadSupportWin32( info );
+	return new btThreadSupportWin32(info);
 }
 
-
-
-#endif //defined(_WIN32) && BT_THREADSAFE
-
+#endif  //defined(_WIN32) && BT_THREADSAFE
diff --git a/thirdparty/bullet/LinearMath/btAabbUtil2.h b/thirdparty/bullet/LinearMath/btAabbUtil2.h
index d2997b4e65c..eea49dd33f9 100644
--- a/thirdparty/bullet/LinearMath/btAabbUtil2.h
+++ b/thirdparty/bullet/LinearMath/btAabbUtil2.h
@@ -12,8 +12,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #ifndef BT_AABB_UTIL2
 #define BT_AABB_UTIL2
 
@@ -21,20 +19,18 @@ subject to the following restrictions:
 #include "btVector3.h"
 #include "btMinMax.h"
 
-
-
-SIMD_FORCE_INLINE void AabbExpand (btVector3& aabbMin,
-								   btVector3& aabbMax,
-								   const btVector3& expansionMin,
-								   const btVector3& expansionMax)
+SIMD_FORCE_INLINE void AabbExpand(btVector3& aabbMin,
+								  btVector3& aabbMax,
+								  const btVector3& expansionMin,
+								  const btVector3& expansionMax)
 {
 	aabbMin = aabbMin + expansionMin;
 	aabbMax = aabbMax + expansionMax;
 }
 
 /// conservative test for overlap between two aabbs
-SIMD_FORCE_INLINE bool TestPointAgainstAabb2(const btVector3 &aabbMin1, const btVector3 &aabbMax1,
-								const btVector3 &point)
+SIMD_FORCE_INLINE bool TestPointAgainstAabb2(const btVector3& aabbMin1, const btVector3& aabbMax1,
+											 const btVector3& point)
 {
 	bool overlap = true;
 	overlap = (aabbMin1.getX() > point.getX() || aabbMax1.getX() < point.getX()) ? false : overlap;
@@ -43,10 +39,9 @@ SIMD_FORCE_INLINE bool TestPointAgainstAabb2(const btVector3 &aabbMin1, const bt
 	return overlap;
 }
 
-
 /// conservative test for overlap between two aabbs
-SIMD_FORCE_INLINE bool TestAabbAgainstAabb2(const btVector3 &aabbMin1, const btVector3 &aabbMax1,
-								const btVector3 &aabbMin2, const btVector3 &aabbMax2)
+SIMD_FORCE_INLINE bool TestAabbAgainstAabb2(const btVector3& aabbMin1, const btVector3& aabbMax1,
+											const btVector3& aabbMin2, const btVector3& aabbMax2)
 {
 	bool overlap = true;
 	overlap = (aabbMin1.getX() > aabbMax2.getX() || aabbMax1.getX() < aabbMin2.getX()) ? false : overlap;
@@ -56,37 +51,34 @@ SIMD_FORCE_INLINE bool TestAabbAgainstAabb2(const btVector3 &aabbMin1, const btV
 }
 
 /// conservative test for overlap between triangle and aabb
-SIMD_FORCE_INLINE bool TestTriangleAgainstAabb2(const btVector3 *vertices,
-									const btVector3 &aabbMin, const btVector3 &aabbMax)
+SIMD_FORCE_INLINE bool TestTriangleAgainstAabb2(const btVector3* vertices,
+												const btVector3& aabbMin, const btVector3& aabbMax)
 {
-	const btVector3 &p1 = vertices[0];
-	const btVector3 &p2 = vertices[1];
-	const btVector3 &p3 = vertices[2];
+	const btVector3& p1 = vertices[0];
+	const btVector3& p2 = vertices[1];
+	const btVector3& p3 = vertices[2];
 
 	if (btMin(btMin(p1[0], p2[0]), p3[0]) > aabbMax[0]) return false;
 	if (btMax(btMax(p1[0], p2[0]), p3[0]) < aabbMin[0]) return false;
 
 	if (btMin(btMin(p1[2], p2[2]), p3[2]) > aabbMax[2]) return false;
 	if (btMax(btMax(p1[2], p2[2]), p3[2]) < aabbMin[2]) return false;
-  
+
 	if (btMin(btMin(p1[1], p2[1]), p3[1]) > aabbMax[1]) return false;
 	if (btMax(btMax(p1[1], p2[1]), p3[1]) < aabbMin[1]) return false;
 	return true;
 }
 
-
-SIMD_FORCE_INLINE int	btOutcode(const btVector3& p,const btVector3& halfExtent) 
+SIMD_FORCE_INLINE int btOutcode(const btVector3& p, const btVector3& halfExtent)
 {
-	return (p.getX()  < -halfExtent.getX() ? 0x01 : 0x0) |    
-		   (p.getX() >  halfExtent.getX() ? 0x08 : 0x0) |
-		   (p.getY() < -halfExtent.getY() ? 0x02 : 0x0) |    
-		   (p.getY() >  halfExtent.getY() ? 0x10 : 0x0) |
-		   (p.getZ() < -halfExtent.getZ() ? 0x4 : 0x0) |    
-		   (p.getZ() >  halfExtent.getZ() ? 0x20 : 0x0);
+	return (p.getX() < -halfExtent.getX() ? 0x01 : 0x0) |
+		   (p.getX() > halfExtent.getX() ? 0x08 : 0x0) |
+		   (p.getY() < -halfExtent.getY() ? 0x02 : 0x0) |
+		   (p.getY() > halfExtent.getY() ? 0x10 : 0x0) |
+		   (p.getZ() < -halfExtent.getZ() ? 0x4 : 0x0) |
+		   (p.getZ() > halfExtent.getZ() ? 0x20 : 0x0);
 }
 
-
-
 SIMD_FORCE_INLINE bool btRayAabb2(const btVector3& rayFrom,
 								  const btVector3& rayInvDirection,
 								  const unsigned int raySign[3],
@@ -97,11 +89,11 @@ SIMD_FORCE_INLINE bool btRayAabb2(const btVector3& rayFrom,
 {
 	btScalar tmax, tymin, tymax, tzmin, tzmax;
 	tmin = (bounds[raySign[0]].getX() - rayFrom.getX()) * rayInvDirection.getX();
-	tmax = (bounds[1-raySign[0]].getX() - rayFrom.getX()) * rayInvDirection.getX();
+	tmax = (bounds[1 - raySign[0]].getX() - rayFrom.getX()) * rayInvDirection.getX();
 	tymin = (bounds[raySign[1]].getY() - rayFrom.getY()) * rayInvDirection.getY();
-	tymax = (bounds[1-raySign[1]].getY() - rayFrom.getY()) * rayInvDirection.getY();
+	tymax = (bounds[1 - raySign[1]].getY() - rayFrom.getY()) * rayInvDirection.getY();
 
-	if ( (tmin > tymax) || (tymin > tmax) )
+	if ((tmin > tymax) || (tymin > tmax))
 		return false;
 
 	if (tymin > tmin)
@@ -111,59 +103,59 @@ SIMD_FORCE_INLINE bool btRayAabb2(const btVector3& rayFrom,
 		tmax = tymax;
 
 	tzmin = (bounds[raySign[2]].getZ() - rayFrom.getZ()) * rayInvDirection.getZ();
-	tzmax = (bounds[1-raySign[2]].getZ() - rayFrom.getZ()) * rayInvDirection.getZ();
+	tzmax = (bounds[1 - raySign[2]].getZ() - rayFrom.getZ()) * rayInvDirection.getZ();
 
-	if ( (tmin > tzmax) || (tzmin > tmax) )
+	if ((tmin > tzmax) || (tzmin > tmax))
 		return false;
 	if (tzmin > tmin)
 		tmin = tzmin;
 	if (tzmax < tmax)
 		tmax = tzmax;
-	return ( (tmin < lambda_max) && (tmax > lambda_min) );
+	return ((tmin < lambda_max) && (tmax > lambda_min));
 }
 
-SIMD_FORCE_INLINE bool btRayAabb(const btVector3& rayFrom, 
-								 const btVector3& rayTo, 
-								 const btVector3& aabbMin, 
+SIMD_FORCE_INLINE bool btRayAabb(const btVector3& rayFrom,
+								 const btVector3& rayTo,
+								 const btVector3& aabbMin,
 								 const btVector3& aabbMax,
-					  btScalar& param, btVector3& normal) 
+								 btScalar& param, btVector3& normal)
 {
-	btVector3 aabbHalfExtent = (aabbMax-aabbMin)* btScalar(0.5);
-	btVector3 aabbCenter = (aabbMax+aabbMin)* btScalar(0.5);
-	btVector3	source = rayFrom - aabbCenter;
-	btVector3	target = rayTo - aabbCenter;
-	int	sourceOutcode = btOutcode(source,aabbHalfExtent);
-	int targetOutcode = btOutcode(target,aabbHalfExtent);
+	btVector3 aabbHalfExtent = (aabbMax - aabbMin) * btScalar(0.5);
+	btVector3 aabbCenter = (aabbMax + aabbMin) * btScalar(0.5);
+	btVector3 source = rayFrom - aabbCenter;
+	btVector3 target = rayTo - aabbCenter;
+	int sourceOutcode = btOutcode(source, aabbHalfExtent);
+	int targetOutcode = btOutcode(target, aabbHalfExtent);
 	if ((sourceOutcode & targetOutcode) == 0x0)
 	{
 		btScalar lambda_enter = btScalar(0.0);
-		btScalar lambda_exit  = param;
+		btScalar lambda_exit = param;
 		btVector3 r = target - source;
 		int i;
-		btScalar	normSign = 1;
-		btVector3	hitNormal(0,0,0);
-		int bit=1;
+		btScalar normSign = 1;
+		btVector3 hitNormal(0, 0, 0);
+		int bit = 1;
 
-		for (int j=0;j<2;j++)
+		for (int j = 0; j < 2; j++)
 		{
 			for (i = 0; i != 3; ++i)
 			{
 				if (sourceOutcode & bit)
 				{
-					btScalar lambda = (-source[i] - aabbHalfExtent[i]*normSign) / r[i];
+					btScalar lambda = (-source[i] - aabbHalfExtent[i] * normSign) / r[i];
 					if (lambda_enter <= lambda)
 					{
 						lambda_enter = lambda;
-						hitNormal.setValue(0,0,0);
+						hitNormal.setValue(0, 0, 0);
 						hitNormal[i] = normSign;
 					}
 				}
-				else if (targetOutcode & bit) 
+				else if (targetOutcode & bit)
 				{
-					btScalar lambda = (-source[i] - aabbHalfExtent[i]*normSign) / r[i];
+					btScalar lambda = (-source[i] - aabbHalfExtent[i] * normSign) / r[i];
 					btSetMin(lambda_exit, lambda);
 				}
-				bit<<=1;
+				bit <<= 1;
 			}
 			normSign = btScalar(-1.);
 		}
@@ -177,56 +169,49 @@ SIMD_FORCE_INLINE bool btRayAabb(const btVector3& rayFrom,
 	return false;
 }
 
-
-
-SIMD_FORCE_INLINE	void btTransformAabb(const btVector3& halfExtents, btScalar margin,const btTransform& t,btVector3& aabbMinOut,btVector3& aabbMaxOut)
+SIMD_FORCE_INLINE void btTransformAabb(const btVector3& halfExtents, btScalar margin, const btTransform& t, btVector3& aabbMinOut, btVector3& aabbMaxOut)
 {
-	btVector3 halfExtentsWithMargin = halfExtents+btVector3(margin,margin,margin);
-	btMatrix3x3 abs_b = t.getBasis().absolute();  
+	btVector3 halfExtentsWithMargin = halfExtents + btVector3(margin, margin, margin);
+	btMatrix3x3 abs_b = t.getBasis().absolute();
 	btVector3 center = t.getOrigin();
-    btVector3 extent = halfExtentsWithMargin.dot3( abs_b[0], abs_b[1], abs_b[2] );
+	btVector3 extent = halfExtentsWithMargin.dot3(abs_b[0], abs_b[1], abs_b[2]);
 	aabbMinOut = center - extent;
 	aabbMaxOut = center + extent;
 }
 
-
-SIMD_FORCE_INLINE	void btTransformAabb(const btVector3& localAabbMin,const btVector3& localAabbMax, btScalar margin,const btTransform& trans,btVector3& aabbMinOut,btVector3& aabbMaxOut)
+SIMD_FORCE_INLINE void btTransformAabb(const btVector3& localAabbMin, const btVector3& localAabbMax, btScalar margin, const btTransform& trans, btVector3& aabbMinOut, btVector3& aabbMaxOut)
 {
-		btAssert(localAabbMin.getX() <= localAabbMax.getX());
-		btAssert(localAabbMin.getY() <= localAabbMax.getY());
-		btAssert(localAabbMin.getZ() <= localAabbMax.getZ());
-		btVector3 localHalfExtents = btScalar(0.5)*(localAabbMax-localAabbMin);
-		localHalfExtents+=btVector3(margin,margin,margin);
+	btAssert(localAabbMin.getX() <= localAabbMax.getX());
+	btAssert(localAabbMin.getY() <= localAabbMax.getY());
+	btAssert(localAabbMin.getZ() <= localAabbMax.getZ());
+	btVector3 localHalfExtents = btScalar(0.5) * (localAabbMax - localAabbMin);
+	localHalfExtents += btVector3(margin, margin, margin);
 
-		btVector3 localCenter = btScalar(0.5)*(localAabbMax+localAabbMin);
-		btMatrix3x3 abs_b = trans.getBasis().absolute();  
-		btVector3 center = trans(localCenter);
-        btVector3 extent = localHalfExtents.dot3( abs_b[0], abs_b[1], abs_b[2] );
-		aabbMinOut = center-extent;
-		aabbMaxOut = center+extent;
+	btVector3 localCenter = btScalar(0.5) * (localAabbMax + localAabbMin);
+	btMatrix3x3 abs_b = trans.getBasis().absolute();
+	btVector3 center = trans(localCenter);
+	btVector3 extent = localHalfExtents.dot3(abs_b[0], abs_b[1], abs_b[2]);
+	aabbMinOut = center - extent;
+	aabbMaxOut = center + extent;
 }
 
 #define USE_BANCHLESS 1
 #ifdef USE_BANCHLESS
-	//This block replaces the block below and uses no branches, and replaces the 8 bit return with a 32 bit return for improved performance (~3x on XBox 360)
-	SIMD_FORCE_INLINE unsigned testQuantizedAabbAgainstQuantizedAabb(const unsigned short int* aabbMin1,const unsigned short int* aabbMax1,const unsigned short int* aabbMin2,const unsigned short int* aabbMax2)
-	{		
-		return static_cast<unsigned int>(btSelect((unsigned)((aabbMin1[0] <= aabbMax2[0]) & (aabbMax1[0] >= aabbMin2[0])
-			& (aabbMin1[2] <= aabbMax2[2]) & (aabbMax1[2] >= aabbMin2[2])
-			& (aabbMin1[1] <= aabbMax2[1]) & (aabbMax1[1] >= aabbMin2[1])),
-			1, 0));
-	}
+//This block replaces the block below and uses no branches, and replaces the 8 bit return with a 32 bit return for improved performance (~3x on XBox 360)
+SIMD_FORCE_INLINE unsigned testQuantizedAabbAgainstQuantizedAabb(const unsigned short int* aabbMin1, const unsigned short int* aabbMax1, const unsigned short int* aabbMin2, const unsigned short int* aabbMax2)
+{
+	return static_cast<unsigned int>(btSelect((unsigned)((aabbMin1[0] <= aabbMax2[0]) & (aabbMax1[0] >= aabbMin2[0]) & (aabbMin1[2] <= aabbMax2[2]) & (aabbMax1[2] >= aabbMin2[2]) & (aabbMin1[1] <= aabbMax2[1]) & (aabbMax1[1] >= aabbMin2[1])),
+											  1, 0));
+}
 #else
-	SIMD_FORCE_INLINE bool testQuantizedAabbAgainstQuantizedAabb(const unsigned short int* aabbMin1,const unsigned short int* aabbMax1,const unsigned short int* aabbMin2,const unsigned short int* aabbMax2)
-	{
-		bool overlap = true;
-		overlap = (aabbMin1[0] > aabbMax2[0] || aabbMax1[0] < aabbMin2[0]) ? false : overlap;
-		overlap = (aabbMin1[2] > aabbMax2[2] || aabbMax1[2] < aabbMin2[2]) ? false : overlap;
-		overlap = (aabbMin1[1] > aabbMax2[1] || aabbMax1[1] < aabbMin2[1]) ? false : overlap;
-		return overlap;
-	}
-#endif //USE_BANCHLESS
-
-#endif //BT_AABB_UTIL2
-
+SIMD_FORCE_INLINE bool testQuantizedAabbAgainstQuantizedAabb(const unsigned short int* aabbMin1, const unsigned short int* aabbMax1, const unsigned short int* aabbMin2, const unsigned short int* aabbMax2)
+{
+	bool overlap = true;
+	overlap = (aabbMin1[0] > aabbMax2[0] || aabbMax1[0] < aabbMin2[0]) ? false : overlap;
+	overlap = (aabbMin1[2] > aabbMax2[2] || aabbMax1[2] < aabbMin2[2]) ? false : overlap;
+	overlap = (aabbMin1[1] > aabbMax2[1] || aabbMax1[1] < aabbMin2[1]) ? false : overlap;
+	return overlap;
+}
+#endif  //USE_BANCHLESS
 
+#endif  //BT_AABB_UTIL2
diff --git a/thirdparty/bullet/LinearMath/btAlignedAllocator.cpp b/thirdparty/bullet/LinearMath/btAlignedAllocator.cpp
index 0526a42283f..39b302b600c 100644
--- a/thirdparty/bullet/LinearMath/btAlignedAllocator.cpp
+++ b/thirdparty/bullet/LinearMath/btAlignedAllocator.cpp
@@ -18,8 +18,8 @@ subject to the following restrictions:
 #ifdef BT_DEBUG_MEMORY_ALLOCATIONS
 int gNumAlignedAllocs = 0;
 int gNumAlignedFree = 0;
-int gTotalBytesAlignedAllocs = 0;//detect memory leaks
-#endif //BT_DEBUG_MEMORY_ALLOCATIONST_DEBUG_ALLOCATIONS
+int gTotalBytesAlignedAllocs = 0;  //detect memory leaks
+#endif                             //BT_DEBUG_MEMORY_ALLOCATIONST_DEBUG_ALLOCATIONS
 
 static void *btAllocDefault(size_t size)
 {
@@ -34,9 +34,7 @@ static void btFreeDefault(void *ptr)
 static btAllocFunc *sAllocFunc = btAllocDefault;
 static btFreeFunc *sFreeFunc = btFreeDefault;
 
-
-
-#if defined (BT_HAS_ALIGNED_ALLOCATOR)
+#if defined(BT_HAS_ALIGNED_ALLOCATOR)
 #include <malloc.h>
 static void *btAlignedAllocDefault(size_t size, int alignment)
 {
@@ -61,49 +59,48 @@ static inline void btAlignedFreeDefault(void *ptr)
 }
 #else
 
-
-
-
-
 static inline void *btAlignedAllocDefault(size_t size, int alignment)
 {
-  void *ret;
-  char *real;
-  real = (char *)sAllocFunc(size + sizeof(void *) + (alignment-1));
-  if (real) {
-	ret = btAlignPointer(real + sizeof(void *),alignment);
-    *((void **)(ret)-1) = (void *)(real);
-  } else {
-    ret = (void *)(real);
-  }
-  return (ret);
+	void *ret;
+	char *real;
+	real = (char *)sAllocFunc(size + sizeof(void *) + (alignment - 1));
+	if (real)
+	{
+		ret = btAlignPointer(real + sizeof(void *), alignment);
+		*((void **)(ret)-1) = (void *)(real);
+	}
+	else
+	{
+		ret = (void *)(real);
+	}
+	return (ret);
 }
 
 static inline void btAlignedFreeDefault(void *ptr)
 {
-  void* real;
+	void *real;
 
-  if (ptr) {
-    real = *((void **)(ptr)-1);
-    sFreeFunc(real);
-  }
+	if (ptr)
+	{
+		real = *((void **)(ptr)-1);
+		sFreeFunc(real);
+	}
 }
 #endif
 
-
 static btAlignedAllocFunc *sAlignedAllocFunc = btAlignedAllocDefault;
 static btAlignedFreeFunc *sAlignedFreeFunc = btAlignedFreeDefault;
 
 void btAlignedAllocSetCustomAligned(btAlignedAllocFunc *allocFunc, btAlignedFreeFunc *freeFunc)
 {
-  sAlignedAllocFunc = allocFunc ? allocFunc : btAlignedAllocDefault;
-  sAlignedFreeFunc = freeFunc ? freeFunc : btAlignedFreeDefault;
+	sAlignedAllocFunc = allocFunc ? allocFunc : btAlignedAllocDefault;
+	sAlignedFreeFunc = freeFunc ? freeFunc : btAlignedFreeDefault;
 }
 
 void btAlignedAllocSetCustom(btAllocFunc *allocFunc, btFreeFunc *freeFunc)
 {
-  sAllocFunc = allocFunc ? allocFunc : btAllocDefault;
-  sFreeFunc = freeFunc ? freeFunc : btFreeDefault;
+	sAllocFunc = allocFunc ? allocFunc : btAllocDefault;
+	sFreeFunc = freeFunc ? freeFunc : btFreeDefault;
 }
 
 #ifdef BT_DEBUG_MEMORY_ALLOCATIONS
@@ -116,15 +113,15 @@ static int mynumallocs = 0;
 int btDumpMemoryLeaks()
 {
 	int totalLeak = 0;
-	
-	for (int i=0;i<mynumallocs;i++)
+
+	for (int i = 0; i < mynumallocs; i++)
 	{
 		printf("Error: leaked memory of allocation #%d (%d bytes)\n", allocations_id[i], allocations_bytes[i]);
-		totalLeak+=allocations_bytes[i];
+		totalLeak += allocations_bytes[i];
 	}
 	if (totalLeak)
 	{
-		printf("Error: memory leaks: %d allocations were not freed and leaked together %d bytes\n",mynumallocs,totalLeak);
+		printf("Error: memory leaks: %d allocations were not freed and leaked together %d bytes\n", mynumallocs, totalLeak);
 	}
 	return totalLeak;
 }
@@ -133,137 +130,134 @@ int btDumpMemoryLeaks()
 
 struct btDebugPtrMagic
 {
-	union
-	{
-		void** vptrptr;
-		void* vptr;
-		int* iptr;
-		char* cptr;
+	union {
+		void **vptrptr;
+		void *vptr;
+		int *iptr;
+		char *cptr;
 	};
 };
 
-
-void*   btAlignedAllocInternal  (size_t size, int alignment,int line,char* filename)
+void *btAlignedAllocInternal(size_t size, int alignment, int line, char *filename)
 {
-	if (size==0)
+	if (size == 0)
 	{
 		printf("Whaat? size==0");
 		return 0;
 	}
 	static int allocId = 0;
-	
- void *ret;
- char *real;
 
-// to find some particular memory leak, you could do something like this:
-//	if (allocId==172)
-//	{
-//		printf("catch me!\n");
-//	}
-//	if (size>1024*1024)
-//	{
-//		printf("big alloc!%d\n", size);
-//	}
+	void *ret;
+	char *real;
 
- gTotalBytesAlignedAllocs += size;
- gNumAlignedAllocs++;
+	// to find some particular memory leak, you could do something like this:
+	//	if (allocId==172)
+	//	{
+	//		printf("catch me!\n");
+	//	}
+	//	if (size>1024*1024)
+	//	{
+	//		printf("big alloc!%d\n", size);
+	//	}
 
- 
-int sz4prt = 4*sizeof(void *);
-	
- real = (char *)sAllocFunc(size + sz4prt + (alignment-1));
- if (real) {
-	 
-   ret = (void*) btAlignPointer(real + sz4prt, alignment);
-	 btDebugPtrMagic p;
-	 p.vptr = ret;
-	 p.cptr-=sizeof(void*);
-	 *p.vptrptr = (void*)real;
-	 p.cptr-=sizeof(void*);
-	 *p.iptr = size;
-	 p.cptr-=sizeof(void*);
-	 *p.iptr = allocId;
-	 
-	 allocations_id[mynumallocs] = allocId;
-	 allocations_bytes[mynumallocs] = size;
-	 mynumallocs++;
+	gTotalBytesAlignedAllocs += size;
+	gNumAlignedAllocs++;
 
- } else {
-   ret = (void *)(real);//??
- }
+	int sz4prt = 4 * sizeof(void *);
 
- printf("allocation %d at address %x, from %s,line %d, size %d (total allocated = %d)\n",allocId,real, filename,line,size,gTotalBytesAlignedAllocs);
+	real = (char *)sAllocFunc(size + sz4prt + (alignment - 1));
+	if (real)
+	{
+		ret = (void *)btAlignPointer(real + sz4prt, alignment);
+		btDebugPtrMagic p;
+		p.vptr = ret;
+		p.cptr -= sizeof(void *);
+		*p.vptrptr = (void *)real;
+		p.cptr -= sizeof(void *);
+		*p.iptr = size;
+		p.cptr -= sizeof(void *);
+		*p.iptr = allocId;
+
+		allocations_id[mynumallocs] = allocId;
+		allocations_bytes[mynumallocs] = size;
+		mynumallocs++;
+	}
+	else
+	{
+		ret = (void *)(real);  //??
+	}
+
+	printf("allocation %d at address %x, from %s,line %d, size %d (total allocated = %d)\n", allocId, real, filename, line, size, gTotalBytesAlignedAllocs);
 	allocId++;
-	
- int* ptr = (int*)ret;
- *ptr = 12;
- return (ret);
+
+	int *ptr = (int *)ret;
+	*ptr = 12;
+	return (ret);
 }
 
-void    btAlignedFreeInternal   (void* ptr,int line,char* filename)
+void btAlignedFreeInternal(void *ptr, int line, char *filename)
 {
+	void *real;
 
- void* real;
+	if (ptr)
+	{
+		gNumAlignedFree++;
 
- if (ptr) {
-	 gNumAlignedFree++;
+		btDebugPtrMagic p;
+		p.vptr = ptr;
+		p.cptr -= sizeof(void *);
+		real = *p.vptrptr;
+		p.cptr -= sizeof(void *);
+		int size = *p.iptr;
+		p.cptr -= sizeof(void *);
+		int allocId = *p.iptr;
 
-	 btDebugPtrMagic p;
-	 p.vptr = ptr;
-	 p.cptr-=sizeof(void*);
-	 real = *p.vptrptr;
-	 p.cptr-=sizeof(void*);
-	 int size = *p.iptr;
-	 p.cptr-=sizeof(void*);
-	 int allocId = *p.iptr;
+		bool found = false;
 
-	 bool found = false;
-	 
-	 for (int i=0;i<mynumallocs;i++)
-	 {
-		 if ( allocations_id[i] == allocId)
-		 {
-			 allocations_id[i] = allocations_id[mynumallocs-1];
-			 allocations_bytes[i] = allocations_bytes[mynumallocs-1];
-			 mynumallocs--;
-			 found = true;
-			 break;
-		 }
-	 }
-	 
-	
-	gTotalBytesAlignedAllocs -= size;
+		for (int i = 0; i < mynumallocs; i++)
+		{
+			if (allocations_id[i] == allocId)
+			{
+				allocations_id[i] = allocations_id[mynumallocs - 1];
+				allocations_bytes[i] = allocations_bytes[mynumallocs - 1];
+				mynumallocs--;
+				found = true;
+				break;
+			}
+		}
 
-	 int diff = gNumAlignedAllocs-gNumAlignedFree;
-	printf("free %d at address %x, from %s,line %d, size %d (total remain = %d in %d non-freed allocations)\n",allocId,real, filename,line,size, gTotalBytesAlignedAllocs, diff);
+		gTotalBytesAlignedAllocs -= size;
 
-   sFreeFunc(real);
- } else
- {
-	 //printf("deleting a NULL ptr, no effect\n");
- }
+		int diff = gNumAlignedAllocs - gNumAlignedFree;
+		printf("free %d at address %x, from %s,line %d, size %d (total remain = %d in %d non-freed allocations)\n", allocId, real, filename, line, size, gTotalBytesAlignedAllocs, diff);
+
+		sFreeFunc(real);
+	}
+	else
+	{
+		//printf("deleting a NULL ptr, no effect\n");
+	}
 }
 
-#else //BT_DEBUG_MEMORY_ALLOCATIONS
+#else  //BT_DEBUG_MEMORY_ALLOCATIONS
 
-void*	btAlignedAllocInternal	(size_t size, int alignment)
+void *btAlignedAllocInternal(size_t size, int alignment)
 {
-	void* ptr;
+	void *ptr;
 	ptr = sAlignedAllocFunc(size, alignment);
-//	printf("btAlignedAllocInternal %d, %x\n",size,ptr);
+	//	printf("btAlignedAllocInternal %d, %x\n",size,ptr);
 	return ptr;
 }
 
-void	btAlignedFreeInternal	(void* ptr)
+void btAlignedFreeInternal(void *ptr)
 {
 	if (!ptr)
 	{
 		return;
 	}
 
-//	printf("btAlignedFreeInternal %x\n",ptr);
+	//	printf("btAlignedFreeInternal %x\n",ptr);
 	sAlignedFreeFunc(ptr);
 }
 
-#endif //BT_DEBUG_MEMORY_ALLOCATIONS
-
+#endif  //BT_DEBUG_MEMORY_ALLOCATIONS
diff --git a/thirdparty/bullet/LinearMath/btAlignedAllocator.h b/thirdparty/bullet/LinearMath/btAlignedAllocator.h
index 9873b338d96..ce4d3585f1f 100644
--- a/thirdparty/bullet/LinearMath/btAlignedAllocator.h
+++ b/thirdparty/bullet/LinearMath/btAlignedAllocator.h
@@ -22,7 +22,6 @@ subject to the following restrictions:
 
 #include "btScalar.h"
 
-
 ///BT_DEBUG_MEMORY_ALLOCATIONS preprocessor can be set in build system
 ///for regression tests to detect memory leaks
 ///#define BT_DEBUG_MEMORY_ALLOCATIONS 1
@@ -30,84 +29,87 @@ subject to the following restrictions:
 
 int btDumpMemoryLeaks();
 
-#define btAlignedAlloc(a,b) \
-		btAlignedAllocInternal(a,b,__LINE__,__FILE__)
+#define btAlignedAlloc(a, b) \
+	btAlignedAllocInternal(a, b, __LINE__, __FILE__)
 
 #define btAlignedFree(ptr) \
-		btAlignedFreeInternal(ptr,__LINE__,__FILE__)
+	btAlignedFreeInternal(ptr, __LINE__, __FILE__)
 
-void*	btAlignedAllocInternal	(size_t size, int alignment,int line,char* filename);
+void* btAlignedAllocInternal(size_t size, int alignment, int line, char* filename);
 
-void	btAlignedFreeInternal	(void* ptr,int line,char* filename);
+void btAlignedFreeInternal(void* ptr, int line, char* filename);
 
 #else
-	void*	btAlignedAllocInternal	(size_t size, int alignment);
-	void	btAlignedFreeInternal	(void* ptr);
+void* btAlignedAllocInternal(size_t size, int alignment);
+void btAlignedFreeInternal(void* ptr);
 
-	#define btAlignedAlloc(size,alignment) btAlignedAllocInternal(size,alignment)
-	#define btAlignedFree(ptr) btAlignedFreeInternal(ptr)
+#define btAlignedAlloc(size, alignment) btAlignedAllocInternal(size, alignment)
+#define btAlignedFree(ptr) btAlignedFreeInternal(ptr)
 
 #endif
-typedef int	size_type;
+typedef int size_type;
 
-typedef void *(btAlignedAllocFunc)(size_t size, int alignment);
-typedef void (btAlignedFreeFunc)(void *memblock);
-typedef void *(btAllocFunc)(size_t size);
-typedef void (btFreeFunc)(void *memblock);
+typedef void*(btAlignedAllocFunc)(size_t size, int alignment);
+typedef void(btAlignedFreeFunc)(void* memblock);
+typedef void*(btAllocFunc)(size_t size);
+typedef void(btFreeFunc)(void* memblock);
 
 ///The developer can let all Bullet memory allocations go through a custom memory allocator, using btAlignedAllocSetCustom
-void btAlignedAllocSetCustom(btAllocFunc *allocFunc, btFreeFunc *freeFunc);
+void btAlignedAllocSetCustom(btAllocFunc* allocFunc, btFreeFunc* freeFunc);
 ///If the developer has already an custom aligned allocator, then btAlignedAllocSetCustomAligned can be used. The default aligned allocator pre-allocates extra memory using the non-aligned allocator, and instruments it.
-void btAlignedAllocSetCustomAligned(btAlignedAllocFunc *allocFunc, btAlignedFreeFunc *freeFunc);
-
+void btAlignedAllocSetCustomAligned(btAlignedAllocFunc* allocFunc, btAlignedFreeFunc* freeFunc);
 
 ///The btAlignedAllocator is a portable class for aligned memory allocations.
 ///Default implementations for unaligned and aligned allocations can be overridden by a custom allocator using btAlignedAllocSetCustom and btAlignedAllocSetCustomAligned.
-template < typename T , unsigned Alignment >
-class btAlignedAllocator {
-	
-	typedef btAlignedAllocator< T , Alignment > self_type;
-	
-public:
+template <typename T, unsigned Alignment>
+class btAlignedAllocator
+{
+	typedef btAlignedAllocator<T, Alignment> self_type;
 
+public:
 	//just going down a list:
 	btAlignedAllocator() {}
 	/*
 	btAlignedAllocator( const self_type & ) {}
 	*/
 
-	template < typename Other >
-	btAlignedAllocator( const btAlignedAllocator< Other , Alignment > & ) {}
+	template <typename Other>
+	btAlignedAllocator(const btAlignedAllocator<Other, Alignment>&)
+	{
+	}
 
-	typedef const T*         const_pointer;
-	typedef const T&         const_reference;
-	typedef T*               pointer;
-	typedef T&               reference;
-	typedef T                value_type;
+	typedef const T* const_pointer;
+	typedef const T& const_reference;
+	typedef T* pointer;
+	typedef T& reference;
+	typedef T value_type;
 
-	pointer       address   ( reference        ref ) const                           { return &ref; }
-	const_pointer address   ( const_reference  ref ) const                           { return &ref; }
-	pointer       allocate  ( size_type        n   , const_pointer *      hint = 0 ) {
+	pointer address(reference ref) const { return &ref; }
+	const_pointer address(const_reference ref) const { return &ref; }
+	pointer allocate(size_type n, const_pointer* hint = 0)
+	{
 		(void)hint;
-		return reinterpret_cast< pointer >(btAlignedAlloc( sizeof(value_type) * n , Alignment ));
+		return reinterpret_cast<pointer>(btAlignedAlloc(sizeof(value_type) * n, Alignment));
 	}
-	void          construct ( pointer          ptr , const value_type &   value    ) { new (ptr) value_type( value ); }
-	void          deallocate( pointer          ptr ) {
-		btAlignedFree( reinterpret_cast< void * >( ptr ) );
+	void construct(pointer ptr, const value_type& value) { new (ptr) value_type(value); }
+	void deallocate(pointer ptr)
+	{
+		btAlignedFree(reinterpret_cast<void*>(ptr));
 	}
-	void          destroy   ( pointer          ptr )                                 { ptr->~value_type(); }
-	
+	void destroy(pointer ptr) { ptr->~value_type(); }
 
-	template < typename O > struct rebind {
-		typedef btAlignedAllocator< O , Alignment > other;
+	template <typename O>
+	struct rebind
+	{
+		typedef btAlignedAllocator<O, Alignment> other;
 	};
-	template < typename O >
-	self_type & operator=( const btAlignedAllocator< O , Alignment > & ) { return *this; }
+	template <typename O>
+	self_type& operator=(const btAlignedAllocator<O, Alignment>&)
+	{
+		return *this;
+	}
 
-	friend bool operator==( const self_type & , const self_type & ) { return true; }
+	friend bool operator==(const self_type&, const self_type&) { return true; }
 };
 
-
-
-#endif //BT_ALIGNED_ALLOCATOR
-
+#endif  //BT_ALIGNED_ALLOCATOR
diff --git a/thirdparty/bullet/LinearMath/btAlignedObjectArray.h b/thirdparty/bullet/LinearMath/btAlignedObjectArray.h
index f0b646529a9..b4671bc19f0 100644
--- a/thirdparty/bullet/LinearMath/btAlignedObjectArray.h
+++ b/thirdparty/bullet/LinearMath/btAlignedObjectArray.h
@@ -13,11 +13,10 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef BT_OBJECT_ARRAY__
 #define BT_OBJECT_ARRAY__
 
-#include "btScalar.h" // has definitions like SIMD_FORCE_INLINE
+#include "btScalar.h"  // has definitions like SIMD_FORCE_INLINE
 #include "btAlignedAllocator.h"
 
 ///If the platform doesn't support placement new, you can disable BT_USE_PLACEMENT_NEW
@@ -28,16 +27,16 @@ subject to the following restrictions:
 
 #define BT_USE_PLACEMENT_NEW 1
 //#define BT_USE_MEMCPY 1 //disable, because it is cumbersome to find out for each platform where memcpy is defined. It can be in <memory.h> or <string.h> or otherwise...
-#define BT_ALLOW_ARRAY_COPY_OPERATOR // enabling this can accidently perform deep copies of data if you are not careful
+#define BT_ALLOW_ARRAY_COPY_OPERATOR  // enabling this can accidently perform deep copies of data if you are not careful
 
 #ifdef BT_USE_MEMCPY
 #include <memory.h>
 #include <string.h>
-#endif //BT_USE_MEMCPY
+#endif  //BT_USE_MEMCPY
 
 #ifdef BT_USE_PLACEMENT_NEW
-#include <new> //for placement new
-#endif //BT_USE_PLACEMENT_NEW
+#include <new>  //for placement new
+#endif          //BT_USE_PLACEMENT_NEW
 
 // The register keyword is deprecated in C++11 so don't use it.
 #if __cplusplus > 199711L
@@ -48,374 +47,358 @@ subject to the following restrictions:
 
 ///The btAlignedObjectArray template class uses a subset of the stl::vector interface for its methods
 ///It is developed to replace stl::vector to avoid portability issues, including STL alignment issues to add SIMD/SSE data
-template <typename T> 
-//template <class T> 
+template <typename T>
+//template <class T>
 class btAlignedObjectArray
 {
-	btAlignedAllocator<T , 16>	m_allocator;
+	btAlignedAllocator<T, 16> m_allocator;
 
-	int					m_size;
-	int					m_capacity;
-	T*					m_data;
+	int m_size;
+	int m_capacity;
+	T* m_data;
 	//PCK: added this line
-	bool				m_ownsMemory;
+	bool m_ownsMemory;
 
 #ifdef BT_ALLOW_ARRAY_COPY_OPERATOR
 public:
-	SIMD_FORCE_INLINE btAlignedObjectArray<T>& operator=(const btAlignedObjectArray<T> &other)
+	SIMD_FORCE_INLINE btAlignedObjectArray<T>& operator=(const btAlignedObjectArray<T>& other)
 	{
 		copyFromArray(other);
 		return *this;
 	}
-#else//BT_ALLOW_ARRAY_COPY_OPERATOR
+#else   //BT_ALLOW_ARRAY_COPY_OPERATOR
 private:
-		SIMD_FORCE_INLINE btAlignedObjectArray<T>& operator=(const btAlignedObjectArray<T> &other);
-#endif//BT_ALLOW_ARRAY_COPY_OPERATOR
+	SIMD_FORCE_INLINE btAlignedObjectArray<T>& operator=(const btAlignedObjectArray<T>& other);
+#endif  //BT_ALLOW_ARRAY_COPY_OPERATOR
 
 protected:
-		SIMD_FORCE_INLINE	int	allocSize(int size)
-		{
-			return (size ? size*2 : 1);
-		}
-		SIMD_FORCE_INLINE	void	copy(int start,int end, T* dest) const
-		{
-			int i;
-			for (i=start;i<end;++i)
+	SIMD_FORCE_INLINE int allocSize(int size)
+	{
+		return (size ? size * 2 : 1);
+	}
+	SIMD_FORCE_INLINE void copy(int start, int end, T* dest) const
+	{
+		int i;
+		for (i = start; i < end; ++i)
 #ifdef BT_USE_PLACEMENT_NEW
-				new (&dest[i]) T(m_data[i]);
+			new (&dest[i]) T(m_data[i]);
 #else
-				dest[i] = m_data[i];
-#endif //BT_USE_PLACEMENT_NEW
-		}
+			dest[i] = m_data[i];
+#endif  //BT_USE_PLACEMENT_NEW
+	}
 
-		SIMD_FORCE_INLINE	void	init()
+	SIMD_FORCE_INLINE void init()
+	{
+		//PCK: added this line
+		m_ownsMemory = true;
+		m_data = 0;
+		m_size = 0;
+		m_capacity = 0;
+	}
+	SIMD_FORCE_INLINE void destroy(int first, int last)
+	{
+		int i;
+		for (i = first; i < last; i++)
 		{
-			//PCK: added this line
-			m_ownsMemory = true;
-			m_data = 0;
-			m_size = 0;
-			m_capacity = 0;
+			m_data[i].~T();
 		}
-		SIMD_FORCE_INLINE	void	destroy(int first,int last)
+	}
+
+	SIMD_FORCE_INLINE void* allocate(int size)
+	{
+		if (size)
+			return m_allocator.allocate(size);
+		return 0;
+	}
+
+	SIMD_FORCE_INLINE void deallocate()
+	{
+		if (m_data)
 		{
-			int i;
-			for (i=first; i<last;i++)
+			//PCK: enclosed the deallocation in this block
+			if (m_ownsMemory)
+			{
+				m_allocator.deallocate(m_data);
+			}
+			m_data = 0;
+		}
+	}
+
+public:
+	btAlignedObjectArray()
+	{
+		init();
+	}
+
+	~btAlignedObjectArray()
+	{
+		clear();
+	}
+
+	///Generally it is best to avoid using the copy constructor of an btAlignedObjectArray, and use a (const) reference to the array instead.
+	btAlignedObjectArray(const btAlignedObjectArray& otherArray)
+	{
+		init();
+
+		int otherSize = otherArray.size();
+		resize(otherSize);
+		otherArray.copy(0, otherSize, m_data);
+	}
+
+	/// return the number of elements in the array
+	SIMD_FORCE_INLINE int size() const
+	{
+		return m_size;
+	}
+
+	SIMD_FORCE_INLINE const T& at(int n) const
+	{
+		btAssert(n >= 0);
+		btAssert(n < size());
+		return m_data[n];
+	}
+
+	SIMD_FORCE_INLINE T& at(int n)
+	{
+		btAssert(n >= 0);
+		btAssert(n < size());
+		return m_data[n];
+	}
+
+	SIMD_FORCE_INLINE const T& operator[](int n) const
+	{
+		btAssert(n >= 0);
+		btAssert(n < size());
+		return m_data[n];
+	}
+
+	SIMD_FORCE_INLINE T& operator[](int n)
+	{
+		btAssert(n >= 0);
+		btAssert(n < size());
+		return m_data[n];
+	}
+
+	///clear the array, deallocated memory. Generally it is better to use array.resize(0), to reduce performance overhead of run-time memory (de)allocations.
+	SIMD_FORCE_INLINE void clear()
+	{
+		destroy(0, size());
+
+		deallocate();
+
+		init();
+	}
+
+	SIMD_FORCE_INLINE void pop_back()
+	{
+		btAssert(m_size > 0);
+		m_size--;
+		m_data[m_size].~T();
+	}
+
+	///resize changes the number of elements in the array. If the new size is larger, the new elements will be constructed using the optional second argument.
+	///when the new number of elements is smaller, the destructor will be called, but memory will not be freed, to reduce performance overhead of run-time memory (de)allocations.
+	SIMD_FORCE_INLINE void resizeNoInitialize(int newsize)
+	{
+		if (newsize > size())
+		{
+			reserve(newsize);
+		}
+		m_size = newsize;
+	}
+
+	SIMD_FORCE_INLINE void resize(int newsize, const T& fillData = T())
+	{
+		const BT_REGISTER int curSize = size();
+
+		if (newsize < curSize)
+		{
+			for (int i = newsize; i < curSize; i++)
 			{
 				m_data[i].~T();
 			}
 		}
-
-		SIMD_FORCE_INLINE	void* allocate(int size)
+		else
 		{
-			if (size)
-				return m_allocator.allocate(size);
-			return 0;
-		}
-
-		SIMD_FORCE_INLINE	void	deallocate()
-		{
-			if(m_data)	{
-				//PCK: enclosed the deallocation in this block
-				if (m_ownsMemory)
-				{
-					m_allocator.deallocate(m_data);
-				}
-				m_data = 0;
-			}
-		}
-
-	
-
-
-	public:
-		
-		btAlignedObjectArray()
-		{
-			init();
-		}
-
-		~btAlignedObjectArray()
-		{
-			clear();
-		}
-
-		///Generally it is best to avoid using the copy constructor of an btAlignedObjectArray, and use a (const) reference to the array instead.
-		btAlignedObjectArray(const btAlignedObjectArray& otherArray)
-		{
-			init();
-
-			int otherSize = otherArray.size();
-			resize (otherSize);
-			otherArray.copy(0, otherSize, m_data);
-		}
-
-		
-		
-		/// return the number of elements in the array
-		SIMD_FORCE_INLINE	int size() const
-		{	
-			return m_size;
-		}
-		
-		SIMD_FORCE_INLINE const T& at(int n) const
-		{
-			btAssert(n>=0);
-			btAssert(n<size());
-			return m_data[n];
-		}
-
-		SIMD_FORCE_INLINE T& at(int n)
-		{
-			btAssert(n>=0);
-			btAssert(n<size());
-			return m_data[n];
-		}
-
-		SIMD_FORCE_INLINE const T& operator[](int n) const
-		{
-			btAssert(n>=0);
-			btAssert(n<size());
-			return m_data[n];
-		}
-
-		SIMD_FORCE_INLINE T& operator[](int n)
-		{
-			btAssert(n>=0);
-			btAssert(n<size());
-			return m_data[n];
-		}
-		
-
-		///clear the array, deallocated memory. Generally it is better to use array.resize(0), to reduce performance overhead of run-time memory (de)allocations.
-		SIMD_FORCE_INLINE	void	clear()
-		{
-			destroy(0,size());
-			
-			deallocate();
-			
-			init();
-		}
-
-		SIMD_FORCE_INLINE	void	pop_back()
-		{
-			btAssert(m_size>0);
-			m_size--;
-			m_data[m_size].~T();
-		}
-
-
-		///resize changes the number of elements in the array. If the new size is larger, the new elements will be constructed using the optional second argument.
-		///when the new number of elements is smaller, the destructor will be called, but memory will not be freed, to reduce performance overhead of run-time memory (de)allocations.
-		SIMD_FORCE_INLINE	void	resizeNoInitialize(int newsize)
-		{
-			if (newsize > size())
+			if (newsize > curSize)
 			{
 				reserve(newsize);
 			}
-			m_size = newsize;
+#ifdef BT_USE_PLACEMENT_NEW
+			for (int i = curSize; i < newsize; i++)
+			{
+				new (&m_data[i]) T(fillData);
+			}
+#endif  //BT_USE_PLACEMENT_NEW
 		}
-	
-		SIMD_FORCE_INLINE	void	resize(int newsize, const T& fillData=T())
+
+		m_size = newsize;
+	}
+	SIMD_FORCE_INLINE T& expandNonInitializing()
+	{
+		const BT_REGISTER int sz = size();
+		if (sz == capacity())
 		{
-			const BT_REGISTER int curSize = size();
-
-			if (newsize < curSize)
-			{
-				for(int i = newsize; i < curSize; i++)
-				{
-					m_data[i].~T();
-				}
-			} else
-			{
-				if (newsize > curSize)
-				{
-					reserve(newsize);
-				}
-#ifdef BT_USE_PLACEMENT_NEW
-				for (int i=curSize;i<newsize;i++)
-				{
-					new ( &m_data[i]) T(fillData);
-				}
-#endif //BT_USE_PLACEMENT_NEW
-
-			}
-
-			m_size = newsize;
+			reserve(allocSize(size()));
 		}
-		SIMD_FORCE_INLINE	T&  expandNonInitializing( )
-		{	
-			const BT_REGISTER int sz = size();
-			if( sz == capacity() )
-			{
-				reserve( allocSize(size()) );
-			}
-			m_size++;
+		m_size++;
 
-			return m_data[sz];		
+		return m_data[sz];
+	}
+
+	SIMD_FORCE_INLINE T& expand(const T& fillValue = T())
+	{
+		const BT_REGISTER int sz = size();
+		if (sz == capacity())
+		{
+			reserve(allocSize(size()));
 		}
-
-
-		SIMD_FORCE_INLINE	T&  expand( const T& fillValue=T())
-		{	
-			const BT_REGISTER int sz = size();
-			if( sz == capacity() )
-			{
-				reserve( allocSize(size()) );
-			}
-			m_size++;
+		m_size++;
 #ifdef BT_USE_PLACEMENT_NEW
-			new (&m_data[sz]) T(fillValue); //use the in-place new (not really allocating heap memory)
+		new (&m_data[sz]) T(fillValue);  //use the in-place new (not really allocating heap memory)
 #endif
 
-			return m_data[sz];		
+		return m_data[sz];
+	}
+
+	SIMD_FORCE_INLINE void push_back(const T& _Val)
+	{
+		const BT_REGISTER int sz = size();
+		if (sz == capacity())
+		{
+			reserve(allocSize(size()));
 		}
 
-
-		SIMD_FORCE_INLINE	void push_back(const T& _Val)
-		{	
-			const BT_REGISTER int sz = size();
-			if( sz == capacity() )
-			{
-				reserve( allocSize(size()) );
-			}
-			
 #ifdef BT_USE_PLACEMENT_NEW
-			new ( &m_data[m_size] ) T(_Val);
+		new (&m_data[m_size]) T(_Val);
 #else
-			m_data[size()] = _Val;			
-#endif //BT_USE_PLACEMENT_NEW
+		m_data[size()] = _Val;
+#endif  //BT_USE_PLACEMENT_NEW
 
-			m_size++;
+		m_size++;
+	}
+
+	/// return the pre-allocated (reserved) elements, this is at least as large as the total number of elements,see size() and reserve()
+	SIMD_FORCE_INLINE int capacity() const
+	{
+		return m_capacity;
+	}
+
+	SIMD_FORCE_INLINE void reserve(int _Count)
+	{  // determine new minimum length of allocated storage
+		if (capacity() < _Count)
+		{  // not enough room, reallocate
+			T* s = (T*)allocate(_Count);
+
+			copy(0, size(), s);
+
+			destroy(0, size());
+
+			deallocate();
+
+			//PCK: added this line
+			m_ownsMemory = true;
+
+			m_data = s;
+
+			m_capacity = _Count;
 		}
+	}
 
-	
-		/// return the pre-allocated (reserved) elements, this is at least as large as the total number of elements,see size() and reserve()
-		SIMD_FORCE_INLINE	int capacity() const
-		{	
-			return m_capacity;
-		}
-		
-		SIMD_FORCE_INLINE	void reserve(int _Count)
-		{	// determine new minimum length of allocated storage
-			if (capacity() < _Count)
-			{	// not enough room, reallocate
-				T*	s = (T*)allocate(_Count);
-
-				copy(0, size(), s);
-
-				destroy(0,size());
-
-				deallocate();
-				
-				//PCK: added this line
-				m_ownsMemory = true;
-
-				m_data = s;
-				
-				m_capacity = _Count;
-
-			}
-		}
-
-
-		class less
+	class less
+	{
+	public:
+		bool operator()(const T& a, const T& b) const
 		{
-			public:
+			return (a < b);
+		}
+	};
 
-				bool operator() ( const T& a, const T& b ) const
-				{
-					return ( a < b );
-				}
-		};
-	
-
-		template <typename L>
-		void quickSortInternal(const L& CompareFunc,int lo, int hi)
-		{
+	template <typename L>
+	void quickSortInternal(const L& CompareFunc, int lo, int hi)
+	{
 		//  lo is the lower index, hi is the upper index
 		//  of the region of array a that is to be sorted
-			int i=lo, j=hi;
-			T x=m_data[(lo+hi)/2];
+		int i = lo, j = hi;
+		T x = m_data[(lo + hi) / 2];
 
-			//  partition
-			do
-			{    
-				while (CompareFunc(m_data[i],x)) 
-					i++; 
-				while (CompareFunc(x,m_data[j])) 
-					j--;
-				if (i<=j)
-				{
-					swap(i,j);
-					i++; j--;
-				}
-			} while (i<=j);
-
-			//  recursion
-			if (lo<j) 
-				quickSortInternal( CompareFunc, lo, j);
-			if (i<hi) 
-				quickSortInternal( CompareFunc, i, hi);
-		}
-
-
-		template <typename L>
-		void quickSort(const L& CompareFunc)
+		//  partition
+		do
 		{
-			//don't sort 0 or 1 elements
-			if (size()>1)
+			while (CompareFunc(m_data[i], x))
+				i++;
+			while (CompareFunc(x, m_data[j]))
+				j--;
+			if (i <= j)
 			{
-				quickSortInternal(CompareFunc,0,size()-1);
+				swap(i, j);
+				i++;
+				j--;
+			}
+		} while (i <= j);
+
+		//  recursion
+		if (lo < j)
+			quickSortInternal(CompareFunc, lo, j);
+		if (i < hi)
+			quickSortInternal(CompareFunc, i, hi);
+	}
+
+	template <typename L>
+	void quickSort(const L& CompareFunc)
+	{
+		//don't sort 0 or 1 elements
+		if (size() > 1)
+		{
+			quickSortInternal(CompareFunc, 0, size() - 1);
+		}
+	}
+
+	///heap sort from http://www.csse.monash.edu.au/~lloyd/tildeAlgDS/Sort/Heap/
+	template <typename L>
+	void downHeap(T* pArr, int k, int n, const L& CompareFunc)
+	{
+		/*  PRE: a[k+1..N] is a heap */
+		/* POST:  a[k..N]  is a heap */
+
+		T temp = pArr[k - 1];
+		/* k has child(s) */
+		while (k <= n / 2)
+		{
+			int child = 2 * k;
+
+			if ((child < n) && CompareFunc(pArr[child - 1], pArr[child]))
+			{
+				child++;
+			}
+			/* pick larger child */
+			if (CompareFunc(temp, pArr[child - 1]))
+			{
+				/* move child up */
+				pArr[k - 1] = pArr[child - 1];
+				k = child;
+			}
+			else
+			{
+				break;
 			}
 		}
+		pArr[k - 1] = temp;
+	} /*downHeap*/
 
-
-		///heap sort from http://www.csse.monash.edu.au/~lloyd/tildeAlgDS/Sort/Heap/
-		template <typename L>
-		void downHeap(T *pArr, int k, int n, const L& CompareFunc)
-		{
-			/*  PRE: a[k+1..N] is a heap */
-			/* POST:  a[k..N]  is a heap */
-			
-			T temp = pArr[k - 1];
-			/* k has child(s) */
-			while (k <= n/2) 
-			{
-				int child = 2*k;
-				
-				if ((child < n) && CompareFunc(pArr[child - 1] , pArr[child]))
-				{
-					child++;
-				}
-				/* pick larger child */
-				if (CompareFunc(temp , pArr[child - 1]))
-				{
-					/* move child up */
-					pArr[k - 1] = pArr[child - 1];
-					k = child;
-				}
-				else
-				{
-					break;
-				}
-			}
-			pArr[k - 1] = temp;
-		} /*downHeap*/
-
-		void	swap(int index0,int index1)
-		{
+	void swap(int index0, int index1)
+	{
 #ifdef BT_USE_MEMCPY
-			char	temp[sizeof(T)];
-			memcpy(temp,&m_data[index0],sizeof(T));
-			memcpy(&m_data[index0],&m_data[index1],sizeof(T));
-			memcpy(&m_data[index1],temp,sizeof(T));
+		char temp[sizeof(T)];
+		memcpy(temp, &m_data[index0], sizeof(T));
+		memcpy(&m_data[index0], &m_data[index1], sizeof(T));
+		memcpy(&m_data[index1], temp, sizeof(T));
 #else
-			T temp = m_data[index0];
-			m_data[index0] = m_data[index1];
-			m_data[index1] = temp;
-#endif //BT_USE_PLACEMENT_NEW
-
-		}
+		T temp = m_data[index0];
+		m_data[index0] = m_data[index1];
+		m_data[index1] = temp;
+#endif  //BT_USE_PLACEMENT_NEW
+	}
 
 	template <typename L>
 	void heapSort(const L& CompareFunc)
@@ -423,49 +406,48 @@ protected:
 		/* sort a[0..N-1],  N.B. 0 to N-1 */
 		int k;
 		int n = m_size;
-		for (k = n/2; k > 0; k--) 
+		for (k = n / 2; k > 0; k--)
 		{
 			downHeap(m_data, k, n, CompareFunc);
 		}
 
 		/* a[1..N] is now a heap */
-		while ( n>=1 ) 
+		while (n >= 1)
 		{
-			swap(0,n-1); /* largest of a[0..n-1] */
-
+			swap(0, n - 1); /* largest of a[0..n-1] */
 
 			n = n - 1;
 			/* restore a[1..i-1] heap */
 			downHeap(m_data, 1, n, CompareFunc);
-		} 
+		}
 	}
 
 	///non-recursive binary search, assumes sorted array
-	int	findBinarySearch(const T& key) const
+	int findBinarySearch(const T& key) const
 	{
 		int first = 0;
-		int last = size()-1;
+		int last = size() - 1;
 
 		//assume sorted array
-		while (first <= last) {
+		while (first <= last)
+		{
 			int mid = (first + last) / 2;  // compute mid point.
-			if (key > m_data[mid]) 
+			if (key > m_data[mid])
 				first = mid + 1;  // repeat search in top half.
-			else if (key < m_data[mid]) 
-				last = mid - 1; // repeat search in bottom half.
+			else if (key < m_data[mid])
+				last = mid - 1;  // repeat search in bottom half.
 			else
-				return mid;     // found it. return position /////
+				return mid;  // found it. return position /////
 		}
-		return size();    // failed to find key
+		return size();  // failed to find key
 	}
 
-
-	int	findLinearSearch(const T& key) const
+	int findLinearSearch(const T& key) const
 	{
-		int index=size();
+		int index = size();
 		int i;
 
-		for (i=0;i<size();i++)
+		for (i = 0; i < size(); i++)
 		{
 			if (m_data[i] == key)
 			{
@@ -475,41 +457,41 @@ protected:
 		}
 		return index;
 	}
-    
-    // If the key is not in the array, return -1 instead of 0,
-    // since 0 also means the first element in the array.
-    int	findLinearSearch2(const T& key) const
-    {
-        int index=-1;
-        int i;
-        
-        for (i=0;i<size();i++)
-        {
-            if (m_data[i] == key)
-            {
-                index = i;
-                break;
-            }
-        }
-        return index;
-    }
 
-    void removeAtIndex(int index)
-    {
-        if (index<size())
-        {
-            swap( index,size()-1);
-            pop_back();
-        }
-    }
-	void	remove(const T& key)
+	// If the key is not in the array, return -1 instead of 0,
+	// since 0 also means the first element in the array.
+	int findLinearSearch2(const T& key) const
+	{
+		int index = -1;
+		int i;
+
+		for (i = 0; i < size(); i++)
+		{
+			if (m_data[i] == key)
+			{
+				index = i;
+				break;
+			}
+		}
+		return index;
+	}
+
+	void removeAtIndex(int index)
+	{
+		if (index < size())
+		{
+			swap(index, size() - 1);
+			pop_back();
+		}
+	}
+	void remove(const T& key)
 	{
 		int findIndex = findLinearSearch(key);
-        removeAtIndex(findIndex);
+		removeAtIndex(findIndex);
 	}
 
 	//PCK: whole function
-	void initializeFromBuffer(void *buffer, int size, int capacity)
+	void initializeFromBuffer(void* buffer, int size, int capacity)
 	{
 		clear();
 		m_ownsMemory = false;
@@ -521,10 +503,9 @@ protected:
 	void copyFromArray(const btAlignedObjectArray& otherArray)
 	{
 		int otherSize = otherArray.size();
-		resize (otherSize);
+		resize(otherSize);
 		otherArray.copy(0, otherSize, m_data);
 	}
-
 };
 
-#endif //BT_OBJECT_ARRAY__
+#endif  //BT_OBJECT_ARRAY__
diff --git a/thirdparty/bullet/LinearMath/btConvexHull.cpp b/thirdparty/bullet/LinearMath/btConvexHull.cpp
index f8b79a1abaa..e7de2a3694c 100644
--- a/thirdparty/bullet/LinearMath/btConvexHull.cpp
+++ b/thirdparty/bullet/LinearMath/btConvexHull.cpp
@@ -20,51 +20,52 @@ subject to the following restrictions:
 #include "btMinMax.h"
 #include "btVector3.h"
 
-
-
-
-
 //----------------------------------
 
-class int3  
+class int3
 {
 public:
-	int x,y,z;
+	int x, y, z;
 	int3(){};
-	int3(int _x,int _y, int _z){x=_x;y=_y;z=_z;}
-	const int& operator[](int i) const {return (&x)[i];}
-	int& operator[](int i) {return (&x)[i];}
+	int3(int _x, int _y, int _z)
+	{
+		x = _x;
+		y = _y;
+		z = _z;
+	}
+	const int &operator[](int i) const { return (&x)[i]; }
+	int &operator[](int i) { return (&x)[i]; }
 };
 
-
 //------- btPlane ----------
 
-
-inline btPlane PlaneFlip(const btPlane &plane){return btPlane(-plane.normal,-plane.dist);}
-inline int operator==( const btPlane &a, const btPlane &b ) { return (a.normal==b.normal && a.dist==b.dist); }
-inline int coplanar( const btPlane &a, const btPlane &b ) { return (a==b || a==PlaneFlip(b)); }
-
+inline btPlane PlaneFlip(const btPlane &plane) { return btPlane(-plane.normal, -plane.dist); }
+inline int operator==(const btPlane &a, const btPlane &b) { return (a.normal == b.normal && a.dist == b.dist); }
+inline int coplanar(const btPlane &a, const btPlane &b) { return (a == b || a == PlaneFlip(b)); }
 
 //--------- Utility Functions ------
 
-btVector3  PlaneLineIntersection(const btPlane &plane, const btVector3 &p0, const btVector3 &p1);
-btVector3  PlaneProject(const btPlane &plane, const btVector3 &point);
+btVector3 PlaneLineIntersection(const btPlane &plane, const btVector3 &p0, const btVector3 &p1);
+btVector3 PlaneProject(const btPlane &plane, const btVector3 &point);
 
-btVector3  ThreePlaneIntersection(const btPlane &p0,const btPlane &p1, const btPlane &p2);
-btVector3  ThreePlaneIntersection(const btPlane &p0,const btPlane &p1, const btPlane &p2)
+btVector3 ThreePlaneIntersection(const btPlane &p0, const btPlane &p1, const btPlane &p2);
+btVector3 ThreePlaneIntersection(const btPlane &p0, const btPlane &p1, const btPlane &p2)
 {
 	btVector3 N1 = p0.normal;
 	btVector3 N2 = p1.normal;
 	btVector3 N3 = p2.normal;
 
-	btVector3 n2n3; n2n3 = N2.cross(N3);
-	btVector3 n3n1; n3n1 = N3.cross(N1);
-	btVector3 n1n2; n1n2 = N1.cross(N2);
+	btVector3 n2n3;
+	n2n3 = N2.cross(N3);
+	btVector3 n3n1;
+	n3n1 = N3.cross(N1);
+	btVector3 n1n2;
+	n1n2 = N1.cross(N2);
 
 	btScalar quotient = (N1.dot(n2n3));
 
 	btAssert(btFabs(quotient) > btScalar(0.000001));
-	
+
 	quotient = btScalar(-1.) / quotient;
 	n2n3 *= p0.dist;
 	n3n1 *= p1.dist;
@@ -74,105 +75,96 @@ btVector3  ThreePlaneIntersection(const btPlane &p0,const btPlane &p1, const btP
 	potentialVertex += n1n2;
 	potentialVertex *= quotient;
 
-	btVector3 result(potentialVertex.getX(),potentialVertex.getY(),potentialVertex.getZ());
+	btVector3 result(potentialVertex.getX(), potentialVertex.getY(), potentialVertex.getZ());
 	return result;
-
 }
 
-btScalar   DistanceBetweenLines(const btVector3 &ustart, const btVector3 &udir, const btVector3 &vstart, const btVector3 &vdir, btVector3 *upoint=NULL, btVector3 *vpoint=NULL);
-btVector3  TriNormal(const btVector3 &v0, const btVector3 &v1, const btVector3 &v2);
-btVector3  NormalOf(const btVector3 *vert, const int n);
-
+btScalar DistanceBetweenLines(const btVector3 &ustart, const btVector3 &udir, const btVector3 &vstart, const btVector3 &vdir, btVector3 *upoint = NULL, btVector3 *vpoint = NULL);
+btVector3 TriNormal(const btVector3 &v0, const btVector3 &v1, const btVector3 &v2);
+btVector3 NormalOf(const btVector3 *vert, const int n);
 
 btVector3 PlaneLineIntersection(const btPlane &plane, const btVector3 &p0, const btVector3 &p1)
 {
 	// returns the point where the line p0-p1 intersects the plane n&d
-    btVector3 dif;
-		dif = p1-p0;
-				btScalar dn= btDot(plane.normal,dif);
-				btScalar t = -(plane.dist+btDot(plane.normal,p0) )/dn;
-				return p0 + (dif*t);
+	btVector3 dif;
+	dif = p1 - p0;
+	btScalar dn = btDot(plane.normal, dif);
+	btScalar t = -(plane.dist + btDot(plane.normal, p0)) / dn;
+	return p0 + (dif * t);
 }
 
 btVector3 PlaneProject(const btPlane &plane, const btVector3 &point)
 {
-	return point - plane.normal * (btDot(point,plane.normal)+plane.dist);
+	return point - plane.normal * (btDot(point, plane.normal) + plane.dist);
 }
 
 btVector3 TriNormal(const btVector3 &v0, const btVector3 &v1, const btVector3 &v2)
 {
 	// return the normal of the triangle
 	// inscribed by v0, v1, and v2
-	btVector3 cp=btCross(v1-v0,v2-v1);
-	btScalar m=cp.length();
-	if(m==0) return btVector3(1,0,0);
-	return cp*(btScalar(1.0)/m);
+	btVector3 cp = btCross(v1 - v0, v2 - v1);
+	btScalar m = cp.length();
+	if (m == 0) return btVector3(1, 0, 0);
+	return cp * (btScalar(1.0) / m);
 }
 
-
 btScalar DistanceBetweenLines(const btVector3 &ustart, const btVector3 &udir, const btVector3 &vstart, const btVector3 &vdir, btVector3 *upoint, btVector3 *vpoint)
 {
 	btVector3 cp;
-	cp = btCross(udir,vdir).normalized();
+	cp = btCross(udir, vdir).normalized();
 
-	btScalar distu = -btDot(cp,ustart);
-	btScalar distv = -btDot(cp,vstart);
-	btScalar dist = (btScalar)fabs(distu-distv);
-	if(upoint) 
-		{
+	btScalar distu = -btDot(cp, ustart);
+	btScalar distv = -btDot(cp, vstart);
+	btScalar dist = (btScalar)fabs(distu - distv);
+	if (upoint)
+	{
 		btPlane plane;
-		plane.normal = btCross(vdir,cp).normalized();
-		plane.dist = -btDot(plane.normal,vstart);
-		*upoint = PlaneLineIntersection(plane,ustart,ustart+udir);
+		plane.normal = btCross(vdir, cp).normalized();
+		plane.dist = -btDot(plane.normal, vstart);
+		*upoint = PlaneLineIntersection(plane, ustart, ustart + udir);
 	}
-	if(vpoint) 
-		{
+	if (vpoint)
+	{
 		btPlane plane;
-		plane.normal = btCross(udir,cp).normalized();
-		plane.dist = -btDot(plane.normal,ustart);
-		*vpoint = PlaneLineIntersection(plane,vstart,vstart+vdir);
+		plane.normal = btCross(udir, cp).normalized();
+		plane.dist = -btDot(plane.normal, ustart);
+		*vpoint = PlaneLineIntersection(plane, vstart, vstart + vdir);
 	}
 	return dist;
 }
 
-
-
-
-
-
-
-#define COPLANAR   (0)
-#define UNDER      (1)
-#define OVER       (2)
-#define SPLIT      (OVER|UNDER)
+#define COPLANAR (0)
+#define UNDER (1)
+#define OVER (2)
+#define SPLIT (OVER | UNDER)
 #define PAPERWIDTH (btScalar(0.001))
 
 btScalar planetestepsilon = PAPERWIDTH;
 
-
-
 typedef ConvexH::HalfEdge HalfEdge;
 
-ConvexH::ConvexH(int vertices_size,int edges_size,int facets_size)
+ConvexH::ConvexH(int vertices_size, int edges_size, int facets_size)
 {
 	vertices.resize(vertices_size);
 	edges.resize(edges_size);
 	facets.resize(facets_size);
 }
 
-
 int PlaneTest(const btPlane &p, const btVector3 &v);
-int PlaneTest(const btPlane &p, const btVector3 &v) {
-	btScalar a  = btDot(v,p.normal)+p.dist;
-	int   flag = (a>planetestepsilon)?OVER:((a<-planetestepsilon)?UNDER:COPLANAR);
+int PlaneTest(const btPlane &p, const btVector3 &v)
+{
+	btScalar a = btDot(v, p.normal) + p.dist;
+	int flag = (a > planetestepsilon) ? OVER : ((a < -planetestepsilon) ? UNDER : COPLANAR);
 	return flag;
 }
 
-int SplitTest(ConvexH &convex,const btPlane &plane);
-int SplitTest(ConvexH &convex,const btPlane &plane) {
-	int flag=0;
-	for(int i=0;i<convex.vertices.size();i++) {
-		flag |= PlaneTest(plane,convex.vertices[i]);
+int SplitTest(ConvexH &convex, const btPlane &plane);
+int SplitTest(ConvexH &convex, const btPlane &plane)
+{
+	int flag = 0;
+	for (int i = 0; i < convex.vertices.size(); i++)
+	{
+		flag |= PlaneTest(plane, convex.vertices[i]);
 	}
 	return flag;
 }
@@ -185,7 +177,7 @@ public:
 	unsigned char undermap;
 	unsigned char overmap;
 };
-class EdgeFlag 
+class EdgeFlag
 {
 public:
 	unsigned char planetest;
@@ -199,146 +191,135 @@ public:
 	unsigned char undermap;
 	unsigned char overmap;
 };
-class Coplanar{
+class Coplanar
+{
 public:
 	unsigned short ea;
 	unsigned char v0;
 	unsigned char v1;
 };
 
-
-
-
-
-
-
-
-template<class T>
-int maxdirfiltered(const T *p,int count,const T &dir,btAlignedObjectArray<int> &allow)
+template <class T>
+int maxdirfiltered(const T *p, int count, const T &dir, btAlignedObjectArray<int> &allow)
 {
 	btAssert(count);
-	int m=-1;
-	for(int i=0;i<count;i++) 
-		if(allow[i])
+	int m = -1;
+	for (int i = 0; i < count; i++)
+		if (allow[i])
 		{
-			if(m==-1 || btDot(p[i],dir)>btDot(p[m],dir))
-				m=i;
+			if (m == -1 || btDot(p[i], dir) > btDot(p[m], dir))
+				m = i;
 		}
-	btAssert(m!=-1);
+	btAssert(m != -1);
 	return m;
-} 
+}
 
 btVector3 orth(const btVector3 &v);
 btVector3 orth(const btVector3 &v)
 {
-	btVector3 a=btCross(v,btVector3(0,0,1));
-	btVector3 b=btCross(v,btVector3(0,1,0));
+	btVector3 a = btCross(v, btVector3(0, 0, 1));
+	btVector3 b = btCross(v, btVector3(0, 1, 0));
 	if (a.length() > b.length())
 	{
 		return a.normalized();
-	} else {
+	}
+	else
+	{
 		return b.normalized();
 	}
 }
 
-
-template<class T>
-int maxdirsterid(const T *p,int count,const T &dir,btAlignedObjectArray<int> &allow)
+template <class T>
+int maxdirsterid(const T *p, int count, const T &dir, btAlignedObjectArray<int> &allow)
 {
-	int m=-1;
-	while(m==-1)
+	int m = -1;
+	while (m == -1)
 	{
-		m = maxdirfiltered(p,count,dir,allow);
-		if(allow[m]==3) return m;
+		m = maxdirfiltered(p, count, dir, allow);
+		if (allow[m] == 3) return m;
 		T u = orth(dir);
-		T v = btCross(u,dir);
-		int ma=-1;
-		for(btScalar x = btScalar(0.0) ; x<= btScalar(360.0) ; x+= btScalar(45.0))
+		T v = btCross(u, dir);
+		int ma = -1;
+		for (btScalar x = btScalar(0.0); x <= btScalar(360.0); x += btScalar(45.0))
 		{
-			btScalar s = btSin(SIMD_RADS_PER_DEG*(x));
-			btScalar c = btCos(SIMD_RADS_PER_DEG*(x));
-			int mb = maxdirfiltered(p,count,dir+(u*s+v*c)*btScalar(0.025),allow);
-			if(ma==m && mb==m)
+			btScalar s = btSin(SIMD_RADS_PER_DEG * (x));
+			btScalar c = btCos(SIMD_RADS_PER_DEG * (x));
+			int mb = maxdirfiltered(p, count, dir + (u * s + v * c) * btScalar(0.025), allow);
+			if (ma == m && mb == m)
 			{
-				allow[m]=3;
+				allow[m] = 3;
 				return m;
 			}
-			if(ma!=-1 && ma!=mb)  // Yuck - this is really ugly
+			if (ma != -1 && ma != mb)  // Yuck - this is really ugly
 			{
 				int mc = ma;
-				for(btScalar xx = x-btScalar(40.0) ; xx <= x ; xx+= btScalar(5.0))
+				for (btScalar xx = x - btScalar(40.0); xx <= x; xx += btScalar(5.0))
 				{
-					btScalar s = btSin(SIMD_RADS_PER_DEG*(xx));
-					btScalar c = btCos(SIMD_RADS_PER_DEG*(xx));
-					int md = maxdirfiltered(p,count,dir+(u*s+v*c)*btScalar(0.025),allow);
-					if(mc==m && md==m)
+					btScalar s = btSin(SIMD_RADS_PER_DEG * (xx));
+					btScalar c = btCos(SIMD_RADS_PER_DEG * (xx));
+					int md = maxdirfiltered(p, count, dir + (u * s + v * c) * btScalar(0.025), allow);
+					if (mc == m && md == m)
 					{
-						allow[m]=3;
+						allow[m] = 3;
 						return m;
 					}
-					mc=md;
+					mc = md;
 				}
 			}
-			ma=mb;
+			ma = mb;
 		}
-		allow[m]=0;
-		m=-1;
+		allow[m] = 0;
+		m = -1;
 	}
 	btAssert(0);
 	return m;
-} 
+}
 
-
-
-
-int operator ==(const int3 &a,const int3 &b);
-int operator ==(const int3 &a,const int3 &b) 
+int operator==(const int3 &a, const int3 &b);
+int operator==(const int3 &a, const int3 &b)
 {
-	for(int i=0;i<3;i++) 
+	for (int i = 0; i < 3; i++)
 	{
-		if(a[i]!=b[i]) return 0;
+		if (a[i] != b[i]) return 0;
 	}
 	return 1;
 }
 
-
-int above(btVector3* vertices,const int3& t, const btVector3 &p, btScalar epsilon);
-int above(btVector3* vertices,const int3& t, const btVector3 &p, btScalar epsilon) 
+int above(btVector3 *vertices, const int3 &t, const btVector3 &p, btScalar epsilon);
+int above(btVector3 *vertices, const int3 &t, const btVector3 &p, btScalar epsilon)
 {
-	btVector3 n=TriNormal(vertices[t[0]],vertices[t[1]],vertices[t[2]]);
-	return (btDot(n,p-vertices[t[0]]) > epsilon); // EPSILON???
+	btVector3 n = TriNormal(vertices[t[0]], vertices[t[1]], vertices[t[2]]);
+	return (btDot(n, p - vertices[t[0]]) > epsilon);  // EPSILON???
 }
-int hasedge(const int3 &t, int a,int b);
-int hasedge(const int3 &t, int a,int b)
+int hasedge(const int3 &t, int a, int b);
+int hasedge(const int3 &t, int a, int b)
 {
-	for(int i=0;i<3;i++)
+	for (int i = 0; i < 3; i++)
 	{
-		int i1= (i+1)%3;
-		if(t[i]==a && t[i1]==b) return 1;
+		int i1 = (i + 1) % 3;
+		if (t[i] == a && t[i1] == b) return 1;
 	}
 	return 0;
 }
 int hasvert(const int3 &t, int v);
 int hasvert(const int3 &t, int v)
 {
-	return (t[0]==v || t[1]==v || t[2]==v) ;
+	return (t[0] == v || t[1] == v || t[2] == v);
 }
-int shareedge(const int3 &a,const int3 &b);
-int shareedge(const int3 &a,const int3 &b)
+int shareedge(const int3 &a, const int3 &b);
+int shareedge(const int3 &a, const int3 &b)
 {
 	int i;
-	for(i=0;i<3;i++)
+	for (i = 0; i < 3; i++)
 	{
-		int i1= (i+1)%3;
-		if(hasedge(a,b[i1],b[i])) return 1;
+		int i1 = (i + 1) % 3;
+		if (hasedge(a, b[i1], b[i])) return 1;
 	}
 	return 0;
 }
 
 class btHullTriangle;
 
-
-
 class btHullTriangle : public int3
 {
 public:
@@ -346,51 +327,50 @@ public:
 	int id;
 	int vmax;
 	btScalar rise;
-	btHullTriangle(int a,int b,int c):int3(a,b,c),n(-1,-1,-1)
+	btHullTriangle(int a, int b, int c) : int3(a, b, c), n(-1, -1, -1)
 	{
-		vmax=-1;
+		vmax = -1;
 		rise = btScalar(0.0);
 	}
 	~btHullTriangle()
 	{
 	}
-	int &neib(int a,int b);
+	int &neib(int a, int b);
 };
 
-
-int &btHullTriangle::neib(int a,int b)
+int &btHullTriangle::neib(int a, int b)
 {
-	static int er=-1;
+	static int er = -1;
 	int i;
-	for(i=0;i<3;i++) 
+	for (i = 0; i < 3; i++)
 	{
-		int i1=(i+1)%3;
-		int i2=(i+2)%3;
-		if((*this)[i]==a && (*this)[i1]==b) return n[i2];
-		if((*this)[i]==b && (*this)[i1]==a) return n[i2];
+		int i1 = (i + 1) % 3;
+		int i2 = (i + 2) % 3;
+		if ((*this)[i] == a && (*this)[i1] == b) return n[i2];
+		if ((*this)[i] == b && (*this)[i1] == a) return n[i2];
 	}
 	btAssert(0);
 	return er;
 }
-void HullLibrary::b2bfix(btHullTriangle* s,btHullTriangle*t)
+void HullLibrary::b2bfix(btHullTriangle *s, btHullTriangle *t)
 {
 	int i;
-	for(i=0;i<3;i++) 
+	for (i = 0; i < 3; i++)
 	{
-		int i1=(i+1)%3;
-		int i2=(i+2)%3;
+		int i1 = (i + 1) % 3;
+		int i2 = (i + 2) % 3;
 		int a = (*s)[i1];
 		int b = (*s)[i2];
-		btAssert(m_tris[s->neib(a,b)]->neib(b,a) == s->id);
-		btAssert(m_tris[t->neib(a,b)]->neib(b,a) == t->id);
-		m_tris[s->neib(a,b)]->neib(b,a) = t->neib(b,a);
-		m_tris[t->neib(b,a)]->neib(a,b) = s->neib(a,b);
+		btAssert(m_tris[s->neib(a, b)]->neib(b, a) == s->id);
+		btAssert(m_tris[t->neib(a, b)]->neib(b, a) == t->id);
+		m_tris[s->neib(a, b)]->neib(b, a) = t->neib(b, a);
+		m_tris[t->neib(b, a)]->neib(a, b) = s->neib(a, b);
 	}
 }
 
-void HullLibrary::removeb2b(btHullTriangle* s,btHullTriangle*t)
+void HullLibrary::removeb2b(btHullTriangle *s, btHullTriangle *t)
 {
-	b2bfix(s,t);
+	b2bfix(s, t);
 	deAllocateTriangle(s);
 
 	deAllocateTriangle(t);
@@ -401,11 +381,11 @@ void HullLibrary::checkit(btHullTriangle *t)
 	(void)t;
 
 	int i;
-	btAssert(m_tris[t->id]==t);
-	for(i=0;i<3;i++)
+	btAssert(m_tris[t->id] == t);
+	for (i = 0; i < 3; i++)
 	{
-		int i1=(i+1)%3;
-		int i2=(i+2)%3;
+		int i1 = (i + 1) % 3;
+		int i2 = (i + 2) % 3;
 		int a = (*t)[i1];
 		int b = (*t)[i2];
 
@@ -415,226 +395,233 @@ void HullLibrary::checkit(btHullTriangle *t)
 		(void)a;
 		(void)b;
 
-		btAssert(a!=b);
-		btAssert( m_tris[t->n[i]]->neib(b,a) == t->id);
+		btAssert(a != b);
+		btAssert(m_tris[t->n[i]]->neib(b, a) == t->id);
 	}
 }
 
-btHullTriangle*	HullLibrary::allocateTriangle(int a,int b,int c)
+btHullTriangle *HullLibrary::allocateTriangle(int a, int b, int c)
 {
-	void* mem = btAlignedAlloc(sizeof(btHullTriangle),16);
-	btHullTriangle* tr = new (mem)btHullTriangle(a,b,c);
+	void *mem = btAlignedAlloc(sizeof(btHullTriangle), 16);
+	btHullTriangle *tr = new (mem) btHullTriangle(a, b, c);
 	tr->id = m_tris.size();
 	m_tris.push_back(tr);
 
 	return tr;
 }
 
-void	HullLibrary::deAllocateTriangle(btHullTriangle* tri)
+void HullLibrary::deAllocateTriangle(btHullTriangle *tri)
 {
-	btAssert(m_tris[tri->id]==tri);
-	m_tris[tri->id]=NULL;
+	btAssert(m_tris[tri->id] == tri);
+	m_tris[tri->id] = NULL;
 	tri->~btHullTriangle();
 	btAlignedFree(tri);
 }
 
-
-void HullLibrary::extrude(btHullTriangle *t0,int v)
+void HullLibrary::extrude(btHullTriangle *t0, int v)
 {
-	int3 t= *t0;
+	int3 t = *t0;
 	int n = m_tris.size();
-	btHullTriangle* ta = allocateTriangle(v,t[1],t[2]);
-	ta->n = int3(t0->n[0],n+1,n+2);
-	m_tris[t0->n[0]]->neib(t[1],t[2]) = n+0;
-	btHullTriangle* tb = allocateTriangle(v,t[2],t[0]);
-	tb->n = int3(t0->n[1],n+2,n+0);
-	m_tris[t0->n[1]]->neib(t[2],t[0]) = n+1;
-	btHullTriangle* tc = allocateTriangle(v,t[0],t[1]);
-	tc->n = int3(t0->n[2],n+0,n+1);
-	m_tris[t0->n[2]]->neib(t[0],t[1]) = n+2;
+	btHullTriangle *ta = allocateTriangle(v, t[1], t[2]);
+	ta->n = int3(t0->n[0], n + 1, n + 2);
+	m_tris[t0->n[0]]->neib(t[1], t[2]) = n + 0;
+	btHullTriangle *tb = allocateTriangle(v, t[2], t[0]);
+	tb->n = int3(t0->n[1], n + 2, n + 0);
+	m_tris[t0->n[1]]->neib(t[2], t[0]) = n + 1;
+	btHullTriangle *tc = allocateTriangle(v, t[0], t[1]);
+	tc->n = int3(t0->n[2], n + 0, n + 1);
+	m_tris[t0->n[2]]->neib(t[0], t[1]) = n + 2;
 	checkit(ta);
 	checkit(tb);
 	checkit(tc);
-	if(hasvert(*m_tris[ta->n[0]],v)) removeb2b(ta,m_tris[ta->n[0]]);
-	if(hasvert(*m_tris[tb->n[0]],v)) removeb2b(tb,m_tris[tb->n[0]]);
-	if(hasvert(*m_tris[tc->n[0]],v)) removeb2b(tc,m_tris[tc->n[0]]);
+	if (hasvert(*m_tris[ta->n[0]], v)) removeb2b(ta, m_tris[ta->n[0]]);
+	if (hasvert(*m_tris[tb->n[0]], v)) removeb2b(tb, m_tris[tb->n[0]]);
+	if (hasvert(*m_tris[tc->n[0]], v)) removeb2b(tc, m_tris[tc->n[0]]);
 	deAllocateTriangle(t0);
-
 }
 
-btHullTriangle* HullLibrary::extrudable(btScalar epsilon)
+btHullTriangle *HullLibrary::extrudable(btScalar epsilon)
 {
 	int i;
-	btHullTriangle *t=NULL;
-	for(i=0;i<m_tris.size();i++)
+	btHullTriangle *t = NULL;
+	for (i = 0; i < m_tris.size(); i++)
 	{
-		if(!t || (m_tris[i] && t->rise<m_tris[i]->rise))
+		if (!t || (m_tris[i] && t->rise < m_tris[i]->rise))
 		{
 			t = m_tris[i];
 		}
 	}
-	return (t->rise >epsilon)?t:NULL ;
+	return (t->rise > epsilon) ? t : NULL;
 }
 
-
-
-
-int4 HullLibrary::FindSimplex(btVector3 *verts,int verts_count,btAlignedObjectArray<int> &allow)
+int4 HullLibrary::FindSimplex(btVector3 *verts, int verts_count, btAlignedObjectArray<int> &allow)
 {
 	btVector3 basis[3];
-	basis[0] = btVector3( btScalar(0.01), btScalar(0.02), btScalar(1.0) );      
-	int p0 = maxdirsterid(verts,verts_count, basis[0],allow);   
-	int	p1 = maxdirsterid(verts,verts_count,-basis[0],allow);
-	basis[0] = verts[p0]-verts[p1];
-	if(p0==p1 || basis[0]==btVector3(0,0,0)) 
-		return int4(-1,-1,-1,-1);
-	basis[1] = btCross(btVector3(     btScalar(1),btScalar(0.02), btScalar(0)),basis[0]);
-	basis[2] = btCross(btVector3(btScalar(-0.02),     btScalar(1), btScalar(0)),basis[0]);
+	basis[0] = btVector3(btScalar(0.01), btScalar(0.02), btScalar(1.0));
+	int p0 = maxdirsterid(verts, verts_count, basis[0], allow);
+	int p1 = maxdirsterid(verts, verts_count, -basis[0], allow);
+	basis[0] = verts[p0] - verts[p1];
+	if (p0 == p1 || basis[0] == btVector3(0, 0, 0))
+		return int4(-1, -1, -1, -1);
+	basis[1] = btCross(btVector3(btScalar(1), btScalar(0.02), btScalar(0)), basis[0]);
+	basis[2] = btCross(btVector3(btScalar(-0.02), btScalar(1), btScalar(0)), basis[0]);
 	if (basis[1].length() > basis[2].length())
 	{
 		basis[1].normalize();
-	} else {
-		basis[1] = basis[2];
-		basis[1].normalize ();
 	}
-	int p2 = maxdirsterid(verts,verts_count,basis[1],allow);
-	if(p2 == p0 || p2 == p1)
+	else
 	{
-		p2 = maxdirsterid(verts,verts_count,-basis[1],allow);
+		basis[1] = basis[2];
+		basis[1].normalize();
 	}
-	if(p2 == p0 || p2 == p1) 
-		return int4(-1,-1,-1,-1);
+	int p2 = maxdirsterid(verts, verts_count, basis[1], allow);
+	if (p2 == p0 || p2 == p1)
+	{
+		p2 = maxdirsterid(verts, verts_count, -basis[1], allow);
+	}
+	if (p2 == p0 || p2 == p1)
+		return int4(-1, -1, -1, -1);
 	basis[1] = verts[p2] - verts[p0];
-	basis[2] = btCross(basis[1],basis[0]).normalized();
-	int p3 = maxdirsterid(verts,verts_count,basis[2],allow);
-	if(p3==p0||p3==p1||p3==p2) p3 = maxdirsterid(verts,verts_count,-basis[2],allow);
-	if(p3==p0||p3==p1||p3==p2) 
-		return int4(-1,-1,-1,-1);
-	btAssert(!(p0==p1||p0==p2||p0==p3||p1==p2||p1==p3||p2==p3));
-	if(btDot(verts[p3]-verts[p0],btCross(verts[p1]-verts[p0],verts[p2]-verts[p0])) <0) {btSwap(p2,p3);}
-	return int4(p0,p1,p2,p3);
+	basis[2] = btCross(basis[1], basis[0]).normalized();
+	int p3 = maxdirsterid(verts, verts_count, basis[2], allow);
+	if (p3 == p0 || p3 == p1 || p3 == p2) p3 = maxdirsterid(verts, verts_count, -basis[2], allow);
+	if (p3 == p0 || p3 == p1 || p3 == p2)
+		return int4(-1, -1, -1, -1);
+	btAssert(!(p0 == p1 || p0 == p2 || p0 == p3 || p1 == p2 || p1 == p3 || p2 == p3));
+	if (btDot(verts[p3] - verts[p0], btCross(verts[p1] - verts[p0], verts[p2] - verts[p0])) < 0)
+	{
+		btSwap(p2, p3);
+	}
+	return int4(p0, p1, p2, p3);
 }
 
-int HullLibrary::calchullgen(btVector3 *verts,int verts_count, int vlimit)
+int HullLibrary::calchullgen(btVector3 *verts, int verts_count, int vlimit)
 {
-	if(verts_count <4) return 0;
-	if(vlimit==0) vlimit=1000000000;
+	if (verts_count < 4) return 0;
+	if (vlimit == 0) vlimit = 1000000000;
 	int j;
-	btVector3 bmin(*verts),bmax(*verts);
+	btVector3 bmin(*verts), bmax(*verts);
 	btAlignedObjectArray<int> isextreme;
 	isextreme.reserve(verts_count);
 	btAlignedObjectArray<int> allow;
 	allow.reserve(verts_count);
 
-	for(j=0;j<verts_count;j++) 
+	for (j = 0; j < verts_count; j++)
 	{
 		allow.push_back(1);
 		isextreme.push_back(0);
-		bmin.setMin (verts[j]);
-		bmax.setMax (verts[j]);
+		bmin.setMin(verts[j]);
+		bmax.setMax(verts[j]);
 	}
-	btScalar epsilon = (bmax-bmin).length() * btScalar(0.001);
-	btAssert (epsilon != 0.0);
+	btScalar epsilon = (bmax - bmin).length() * btScalar(0.001);
+	btAssert(epsilon != 0.0);
 
+	int4 p = FindSimplex(verts, verts_count, allow);
+	if (p.x == -1) return 0;  // simplex failed
 
-	int4 p = FindSimplex(verts,verts_count,allow);
-	if(p.x==-1) return 0; // simplex failed
+	btVector3 center = (verts[p[0]] + verts[p[1]] + verts[p[2]] + verts[p[3]]) / btScalar(4.0);  // a valid interior point
+	btHullTriangle *t0 = allocateTriangle(p[2], p[3], p[1]);
+	t0->n = int3(2, 3, 1);
+	btHullTriangle *t1 = allocateTriangle(p[3], p[2], p[0]);
+	t1->n = int3(3, 2, 0);
+	btHullTriangle *t2 = allocateTriangle(p[0], p[1], p[3]);
+	t2->n = int3(0, 1, 3);
+	btHullTriangle *t3 = allocateTriangle(p[1], p[0], p[2]);
+	t3->n = int3(1, 0, 2);
+	isextreme[p[0]] = isextreme[p[1]] = isextreme[p[2]] = isextreme[p[3]] = 1;
+	checkit(t0);
+	checkit(t1);
+	checkit(t2);
+	checkit(t3);
 
-
-
-	btVector3 center = (verts[p[0]]+verts[p[1]]+verts[p[2]]+verts[p[3]]) / btScalar(4.0);  // a valid interior point
-	btHullTriangle *t0 = allocateTriangle(p[2],p[3],p[1]); t0->n=int3(2,3,1);
-	btHullTriangle *t1 = allocateTriangle(p[3],p[2],p[0]); t1->n=int3(3,2,0);
-	btHullTriangle *t2 = allocateTriangle(p[0],p[1],p[3]); t2->n=int3(0,1,3);
-	btHullTriangle *t3 = allocateTriangle(p[1],p[0],p[2]); t3->n=int3(1,0,2);
-	isextreme[p[0]]=isextreme[p[1]]=isextreme[p[2]]=isextreme[p[3]]=1;
-	checkit(t0);checkit(t1);checkit(t2);checkit(t3);
-
-	for(j=0;j<m_tris.size();j++)
+	for (j = 0; j < m_tris.size(); j++)
 	{
-		btHullTriangle *t=m_tris[j];
+		btHullTriangle *t = m_tris[j];
 		btAssert(t);
-		btAssert(t->vmax<0);
-		btVector3 n=TriNormal(verts[(*t)[0]],verts[(*t)[1]],verts[(*t)[2]]);
-		t->vmax = maxdirsterid(verts,verts_count,n,allow);
-		t->rise = btDot(n,verts[t->vmax]-verts[(*t)[0]]);
+		btAssert(t->vmax < 0);
+		btVector3 n = TriNormal(verts[(*t)[0]], verts[(*t)[1]], verts[(*t)[2]]);
+		t->vmax = maxdirsterid(verts, verts_count, n, allow);
+		t->rise = btDot(n, verts[t->vmax] - verts[(*t)[0]]);
 	}
 	btHullTriangle *te;
-	vlimit-=4;
-	while(vlimit >0 && ((te=extrudable(epsilon)) != 0))
+	vlimit -= 4;
+	while (vlimit > 0 && ((te = extrudable(epsilon)) != 0))
 	{
 		//int3 ti=*te;
-		int v=te->vmax;
+		int v = te->vmax;
 		btAssert(v != -1);
 		btAssert(!isextreme[v]);  // wtf we've already done this vertex
-		isextreme[v]=1;
+		isextreme[v] = 1;
 		//if(v==p0 || v==p1 || v==p2 || v==p3) continue; // done these already
-		j=m_tris.size();
-		while(j--) {
-			if(!m_tris[j]) continue;
-			int3 t=*m_tris[j];
-			if(above(verts,t,verts[v],btScalar(0.01)*epsilon)) 
+		j = m_tris.size();
+		while (j--)
+		{
+			if (!m_tris[j]) continue;
+			int3 t = *m_tris[j];
+			if (above(verts, t, verts[v], btScalar(0.01) * epsilon))
 			{
-				extrude(m_tris[j],v);
+				extrude(m_tris[j], v);
 			}
 		}
 		// now check for those degenerate cases where we have a flipped triangle or a really skinny triangle
-		j=m_tris.size();
-		while(j--)
+		j = m_tris.size();
+		while (j--)
 		{
-			if(!m_tris[j]) continue;
-			if(!hasvert(*m_tris[j],v)) break;
-			int3 nt=*m_tris[j];
-			if(above(verts,nt,center,btScalar(0.01)*epsilon)  || btCross(verts[nt[1]]-verts[nt[0]],verts[nt[2]]-verts[nt[1]]).length()< epsilon*epsilon*btScalar(0.1) )
+			if (!m_tris[j]) continue;
+			if (!hasvert(*m_tris[j], v)) break;
+			int3 nt = *m_tris[j];
+			if (above(verts, nt, center, btScalar(0.01) * epsilon) || btCross(verts[nt[1]] - verts[nt[0]], verts[nt[2]] - verts[nt[1]]).length() < epsilon * epsilon * btScalar(0.1))
 			{
 				btHullTriangle *nb = m_tris[m_tris[j]->n[0]];
-				btAssert(nb);btAssert(!hasvert(*nb,v));btAssert(nb->id<j);
-				extrude(nb,v);
-				j=m_tris.size(); 
+				btAssert(nb);
+				btAssert(!hasvert(*nb, v));
+				btAssert(nb->id < j);
+				extrude(nb, v);
+				j = m_tris.size();
 			}
-		} 
-		j=m_tris.size();
-		while(j--)
+		}
+		j = m_tris.size();
+		while (j--)
 		{
-			btHullTriangle *t=m_tris[j];
-			if(!t) continue;
-			if(t->vmax>=0) break;
-			btVector3 n=TriNormal(verts[(*t)[0]],verts[(*t)[1]],verts[(*t)[2]]);
-			t->vmax = maxdirsterid(verts,verts_count,n,allow);
-			if(isextreme[t->vmax]) 
+			btHullTriangle *t = m_tris[j];
+			if (!t) continue;
+			if (t->vmax >= 0) break;
+			btVector3 n = TriNormal(verts[(*t)[0]], verts[(*t)[1]], verts[(*t)[2]]);
+			t->vmax = maxdirsterid(verts, verts_count, n, allow);
+			if (isextreme[t->vmax])
 			{
-				t->vmax=-1; // already done that vertex - algorithm needs to be able to terminate.
+				t->vmax = -1;  // already done that vertex - algorithm needs to be able to terminate.
 			}
 			else
 			{
-				t->rise = btDot(n,verts[t->vmax]-verts[(*t)[0]]);
+				t->rise = btDot(n, verts[t->vmax] - verts[(*t)[0]]);
 			}
 		}
-		vlimit --;
+		vlimit--;
 	}
 	return 1;
 }
 
-int HullLibrary::calchull(btVector3 *verts,int verts_count, TUIntArray& tris_out, int &tris_count,int vlimit) 
+int HullLibrary::calchull(btVector3 *verts, int verts_count, TUIntArray &tris_out, int &tris_count, int vlimit)
 {
-	int rc=calchullgen(verts,verts_count,  vlimit) ;
-	if(!rc) return 0;
+	int rc = calchullgen(verts, verts_count, vlimit);
+	if (!rc) return 0;
 	btAlignedObjectArray<int> ts;
 	int i;
 
-	for(i=0;i<m_tris.size();i++)
+	for (i = 0; i < m_tris.size(); i++)
 	{
-		if(m_tris[i])
+		if (m_tris[i])
 		{
-			for(int j=0;j<3;j++)
+			for (int j = 0; j < 3; j++)
 				ts.push_back((*m_tris[i])[j]);
 			deAllocateTriangle(m_tris[i]);
 		}
 	}
-	tris_count = ts.size()/3;
+	tris_count = ts.size() / 3;
 	tris_out.resize(ts.size());
-	
-	for (i=0;i<ts.size();i++)
+
+	for (i = 0; i < ts.size(); i++)
 	{
 		tris_out[i] = static_cast<unsigned int>(ts[i]);
 	}
@@ -643,29 +630,22 @@ int HullLibrary::calchull(btVector3 *verts,int verts_count, TUIntArray& tris_out
 	return 1;
 }
 
-
-
-
-
-bool HullLibrary::ComputeHull(unsigned int vcount,const btVector3 *vertices,PHullResult &result,unsigned int vlimit)
+bool HullLibrary::ComputeHull(unsigned int vcount, const btVector3 *vertices, PHullResult &result, unsigned int vlimit)
 {
-	
-	int    tris_count;
-	int ret = calchull( (btVector3 *) vertices, (int) vcount, result.m_Indices, tris_count, static_cast<int>(vlimit) );
-	if(!ret) return false;
-	result.mIndexCount = (unsigned int) (tris_count*3);
-	result.mFaceCount  = (unsigned int) tris_count;
-	result.mVertices   = (btVector3*) vertices;
-	result.mVcount     = (unsigned int) vcount;
+	int tris_count;
+	int ret = calchull((btVector3 *)vertices, (int)vcount, result.m_Indices, tris_count, static_cast<int>(vlimit));
+	if (!ret) return false;
+	result.mIndexCount = (unsigned int)(tris_count * 3);
+	result.mFaceCount = (unsigned int)tris_count;
+	result.mVertices = (btVector3 *)vertices;
+	result.mVcount = (unsigned int)vcount;
 	return true;
-
 }
 
-
 void ReleaseHull(PHullResult &result);
 void ReleaseHull(PHullResult &result)
 {
-	if ( result.m_Indices.size() )
+	if (result.m_Indices.size())
 	{
 		result.m_Indices.clear();
 	}
@@ -675,7 +655,6 @@ void ReleaseHull(PHullResult &result)
 	result.mVertices = 0;
 }
 
-
 //*********************************************************************
 //*********************************************************************
 //********  HullLib header
@@ -688,16 +667,15 @@ void ReleaseHull(PHullResult &result)
 //*********************************************************************
 //*********************************************************************
 
-HullError HullLibrary::CreateConvexHull(const HullDesc       &desc,           // describes the input request
-																					HullResult           &result)         // contains the resulst
+HullError HullLibrary::CreateConvexHull(const HullDesc &desc,  // describes the input request
+										HullResult &result)    // contains the resulst
 {
 	HullError ret = QE_FAIL;
 
-
 	PHullResult hr;
 
 	unsigned int vcount = desc.mVcount;
-	if ( vcount < 8 ) vcount = 8;
+	if (vcount < 8) vcount = 8;
 
 	btAlignedObjectArray<btVector3> vertexSource;
 	vertexSource.resize(static_cast<int>(vcount));
@@ -706,87 +684,82 @@ HullError HullLibrary::CreateConvexHull(const HullDesc       &desc,           //
 
 	unsigned int ovcount;
 
-	bool ok = CleanupVertices(desc.mVcount,desc.mVertices, desc.mVertexStride, ovcount, &vertexSource[0], desc.mNormalEpsilon, scale ); // normalize point cloud, remove duplicates!
+	bool ok = CleanupVertices(desc.mVcount, desc.mVertices, desc.mVertexStride, ovcount, &vertexSource[0], desc.mNormalEpsilon, scale);  // normalize point cloud, remove duplicates!
 
-	if ( ok )
+	if (ok)
 	{
-
-
-//		if ( 1 ) // scale vertices back to their original size.
+		//		if ( 1 ) // scale vertices back to their original size.
 		{
-			for (unsigned int i=0; i<ovcount; i++)
+			for (unsigned int i = 0; i < ovcount; i++)
 			{
-				btVector3& v = vertexSource[static_cast<int>(i)];
-				v[0]*=scale[0];
-				v[1]*=scale[1];
-				v[2]*=scale[2];
+				btVector3 &v = vertexSource[static_cast<int>(i)];
+				v[0] *= scale[0];
+				v[1] *= scale[1];
+				v[2] *= scale[2];
 			}
 		}
 
-		ok = ComputeHull(ovcount,&vertexSource[0],hr,desc.mMaxVertices);
+		ok = ComputeHull(ovcount, &vertexSource[0], hr, desc.mMaxVertices);
 
-		if ( ok )
+		if (ok)
 		{
-
 			// re-index triangle mesh so it refers to only used vertices, rebuild a new vertex table.
-			btAlignedObjectArray<btVector3>	vertexScratch;
+			btAlignedObjectArray<btVector3> vertexScratch;
 			vertexScratch.resize(static_cast<int>(hr.mVcount));
 
-			BringOutYourDead(hr.mVertices,hr.mVcount, &vertexScratch[0], ovcount, &hr.m_Indices[0], hr.mIndexCount );
+			BringOutYourDead(hr.mVertices, hr.mVcount, &vertexScratch[0], ovcount, &hr.m_Indices[0], hr.mIndexCount);
 
 			ret = QE_OK;
 
-			if ( desc.HasHullFlag(QF_TRIANGLES) ) // if he wants the results as triangle!
+			if (desc.HasHullFlag(QF_TRIANGLES))  // if he wants the results as triangle!
 			{
-				result.mPolygons          = false;
+				result.mPolygons = false;
 				result.mNumOutputVertices = ovcount;
 				result.m_OutputVertices.resize(static_cast<int>(ovcount));
-				result.mNumFaces          = hr.mFaceCount;
-				result.mNumIndices        = hr.mIndexCount;
+				result.mNumFaces = hr.mFaceCount;
+				result.mNumIndices = hr.mIndexCount;
 
 				result.m_Indices.resize(static_cast<int>(hr.mIndexCount));
 
-				memcpy(&result.m_OutputVertices[0], &vertexScratch[0], sizeof(btVector3)*ovcount );
+				memcpy(&result.m_OutputVertices[0], &vertexScratch[0], sizeof(btVector3) * ovcount);
 
-  			if ( desc.HasHullFlag(QF_REVERSE_ORDER) )
+				if (desc.HasHullFlag(QF_REVERSE_ORDER))
 				{
-
 					const unsigned int *source = &hr.m_Indices[0];
-					unsigned int *dest   = &result.m_Indices[0];
+					unsigned int *dest = &result.m_Indices[0];
 
-					for (unsigned int i=0; i<hr.mFaceCount; i++)
+					for (unsigned int i = 0; i < hr.mFaceCount; i++)
 					{
 						dest[0] = source[2];
 						dest[1] = source[1];
 						dest[2] = source[0];
-						dest+=3;
-						source+=3;
+						dest += 3;
+						source += 3;
 					}
-
 				}
 				else
 				{
-					memcpy(&result.m_Indices[0], &hr.m_Indices[0], sizeof(unsigned int)*hr.mIndexCount);
+					memcpy(&result.m_Indices[0], &hr.m_Indices[0], sizeof(unsigned int) * hr.mIndexCount);
 				}
 			}
 			else
 			{
-				result.mPolygons          = true;
+				result.mPolygons = true;
 				result.mNumOutputVertices = ovcount;
 				result.m_OutputVertices.resize(static_cast<int>(ovcount));
-				result.mNumFaces          = hr.mFaceCount;
-				result.mNumIndices        = hr.mIndexCount+hr.mFaceCount;
+				result.mNumFaces = hr.mFaceCount;
+				result.mNumIndices = hr.mIndexCount + hr.mFaceCount;
 				result.m_Indices.resize(static_cast<int>(result.mNumIndices));
-				memcpy(&result.m_OutputVertices[0], &vertexScratch[0], sizeof(btVector3)*ovcount );
+				memcpy(&result.m_OutputVertices[0], &vertexScratch[0], sizeof(btVector3) * ovcount);
 
-//				if ( 1 )
+				//				if ( 1 )
 				{
 					const unsigned int *source = &hr.m_Indices[0];
-					unsigned int *dest   = &result.m_Indices[0];
-					for (unsigned int i=0; i<hr.mFaceCount; i++)
+					unsigned int *dest = &result.m_Indices[0];
+					for (unsigned int i = 0; i < hr.mFaceCount; i++)
 					{
 						dest[0] = 3;
-						if ( desc.HasHullFlag(QF_REVERSE_ORDER) )
+						if (desc.HasHullFlag(QF_REVERSE_ORDER))
 						{
 							dest[1] = source[2];
 							dest[2] = source[1];
@@ -799,8 +772,8 @@ HullError HullLibrary::CreateConvexHull(const HullDesc       &desc,           //
 							dest[3] = source[2];
 						}
 
-						dest+=4;
-						source+=3;
+						dest += 4;
+						source += 3;
 					}
 				}
 			}
@@ -811,90 +784,83 @@ HullError HullLibrary::CreateConvexHull(const HullDesc       &desc,           //
 	return ret;
 }
 
-
-
-HullError HullLibrary::ReleaseResult(HullResult &result) // release memory allocated for this result, we are done with it.
+HullError HullLibrary::ReleaseResult(HullResult &result)  // release memory allocated for this result, we are done with it.
 {
-	if ( result.m_OutputVertices.size())
+	if (result.m_OutputVertices.size())
 	{
-		result.mNumOutputVertices=0;
+		result.mNumOutputVertices = 0;
 		result.m_OutputVertices.clear();
 	}
-	if ( result.m_Indices.size() )
+	if (result.m_Indices.size())
 	{
-		result.mNumIndices=0;
+		result.mNumIndices = 0;
 		result.m_Indices.clear();
 	}
 	return QE_OK;
 }
 
-
-static void addPoint(unsigned int &vcount,btVector3 *p,btScalar x,btScalar y,btScalar z)
+static void addPoint(unsigned int &vcount, btVector3 *p, btScalar x, btScalar y, btScalar z)
 {
 	// XXX, might be broken
-	btVector3& dest = p[vcount];
+	btVector3 &dest = p[vcount];
 	dest[0] = x;
 	dest[1] = y;
 	dest[2] = z;
 	vcount++;
 }
 
-btScalar GetDist(btScalar px,btScalar py,btScalar pz,const btScalar *p2);
-btScalar GetDist(btScalar px,btScalar py,btScalar pz,const btScalar *p2)
+btScalar GetDist(btScalar px, btScalar py, btScalar pz, const btScalar *p2);
+btScalar GetDist(btScalar px, btScalar py, btScalar pz, const btScalar *p2)
 {
-
 	btScalar dx = px - p2[0];
 	btScalar dy = py - p2[1];
 	btScalar dz = pz - p2[2];
 
-	return dx*dx+dy*dy+dz*dz;
+	return dx * dx + dy * dy + dz * dz;
 }
 
-
-
-bool  HullLibrary::CleanupVertices(unsigned int svcount,
-				   const btVector3 *svertices,
-				   unsigned int stride,
-				   unsigned int &vcount,       // output number of vertices
-				   btVector3 *vertices,                 // location to store the results.
-				   btScalar  normalepsilon,
-				   btVector3& scale)
+bool HullLibrary::CleanupVertices(unsigned int svcount,
+								  const btVector3 *svertices,
+								  unsigned int stride,
+								  unsigned int &vcount,  // output number of vertices
+								  btVector3 *vertices,   // location to store the results.
+								  btScalar normalepsilon,
+								  btVector3 &scale)
 {
-	if ( svcount == 0 ) return false;
+	if (svcount == 0) return false;
 
 	m_vertexIndexMapping.resize(0);
 
-
 #define EPSILON btScalar(0.000001) /* close enough to consider two btScalaring point numbers to be 'the same'. */
 
 	vcount = 0;
 
-	btScalar recip[3]={0.f,0.f,0.f};
+	btScalar recip[3] = {0.f, 0.f, 0.f};
 
-	if ( scale )
+	if (scale)
 	{
 		scale[0] = 1;
 		scale[1] = 1;
 		scale[2] = 1;
 	}
 
-	btScalar bmin[3] = {  FLT_MAX,  FLT_MAX,  FLT_MAX };
-	btScalar bmax[3] = { -FLT_MAX, -FLT_MAX, -FLT_MAX };
+	btScalar bmin[3] = {FLT_MAX, FLT_MAX, FLT_MAX};
+	btScalar bmax[3] = {-FLT_MAX, -FLT_MAX, -FLT_MAX};
 
-	const char *vtx = (const char *) svertices;
+	const char *vtx = (const char *)svertices;
 
-//	if ( 1 )
+	//	if ( 1 )
 	{
-		for (unsigned int i=0; i<svcount; i++)
+		for (unsigned int i = 0; i < svcount; i++)
 		{
-			const btScalar *p = (const btScalar *) vtx;
+			const btScalar *p = (const btScalar *)vtx;
 
-			vtx+=stride;
+			vtx += stride;
 
-			for (int j=0; j<3; j++)
+			for (int j = 0; j < 3; j++)
 			{
-				if ( p[j] < bmin[j] ) bmin[j] = p[j];
-				if ( p[j] > bmax[j] ) bmax[j] = p[j];
+				if (p[j] < bmin[j]) bmin[j] = p[j];
+				if (p[j] > bmax[j]) bmax[j] = p[j];
 			}
 		}
 	}
@@ -905,28 +871,27 @@ bool  HullLibrary::CleanupVertices(unsigned int svcount,
 
 	btVector3 center;
 
-	center[0] = dx*btScalar(0.5) + bmin[0];
-	center[1] = dy*btScalar(0.5) + bmin[1];
-	center[2] = dz*btScalar(0.5) + bmin[2];
+	center[0] = dx * btScalar(0.5) + bmin[0];
+	center[1] = dy * btScalar(0.5) + bmin[1];
+	center[2] = dz * btScalar(0.5) + bmin[2];
 
-	if ( dx < EPSILON || dy < EPSILON || dz < EPSILON || svcount < 3 )
+	if (dx < EPSILON || dy < EPSILON || dz < EPSILON || svcount < 3)
 	{
-
 		btScalar len = FLT_MAX;
 
-		if ( dx > EPSILON && dx < len ) len = dx;
-		if ( dy > EPSILON && dy < len ) len = dy;
-		if ( dz > EPSILON && dz < len ) len = dz;
+		if (dx > EPSILON && dx < len) len = dx;
+		if (dy > EPSILON && dy < len) len = dy;
+		if (dz > EPSILON && dz < len) len = dz;
 
-		if ( len == FLT_MAX )
+		if (len == FLT_MAX)
 		{
-			dx = dy = dz = btScalar(0.01); // one centimeter
+			dx = dy = dz = btScalar(0.01);  // one centimeter
 		}
 		else
 		{
-			if ( dx < EPSILON ) dx = len * btScalar(0.05); // 1/5th the shortest non-zero edge.
-			if ( dy < EPSILON ) dy = len * btScalar(0.05);
-			if ( dz < EPSILON ) dz = len * btScalar(0.05);
+			if (dx < EPSILON) dx = len * btScalar(0.05);  // 1/5th the shortest non-zero edge.
+			if (dy < EPSILON) dy = len * btScalar(0.05);
+			if (dz < EPSILON) dz = len * btScalar(0.05);
 		}
 
 		btScalar x1 = center[0] - dx;
@@ -938,22 +903,20 @@ bool  HullLibrary::CleanupVertices(unsigned int svcount,
 		btScalar z1 = center[2] - dz;
 		btScalar z2 = center[2] + dz;
 
-		addPoint(vcount,vertices,x1,y1,z1);
-		addPoint(vcount,vertices,x2,y1,z1);
-		addPoint(vcount,vertices,x2,y2,z1);
-		addPoint(vcount,vertices,x1,y2,z1);
-		addPoint(vcount,vertices,x1,y1,z2);
-		addPoint(vcount,vertices,x2,y1,z2);
-		addPoint(vcount,vertices,x2,y2,z2);
-		addPoint(vcount,vertices,x1,y2,z2);
-
-		return true; // return cube
-
+		addPoint(vcount, vertices, x1, y1, z1);
+		addPoint(vcount, vertices, x2, y1, z1);
+		addPoint(vcount, vertices, x2, y2, z1);
+		addPoint(vcount, vertices, x1, y2, z1);
+		addPoint(vcount, vertices, x1, y1, z2);
+		addPoint(vcount, vertices, x2, y1, z2);
+		addPoint(vcount, vertices, x2, y2, z2);
+		addPoint(vcount, vertices, x1, y2, z2);
 
+		return true;  // return cube
 	}
 	else
 	{
-		if ( scale )
+		if (scale)
 		{
 			scale[0] = dx;
 			scale[1] = dy;
@@ -963,75 +926,70 @@ bool  HullLibrary::CleanupVertices(unsigned int svcount,
 			recip[1] = 1 / dy;
 			recip[2] = 1 / dz;
 
-			center[0]*=recip[0];
-			center[1]*=recip[1];
-			center[2]*=recip[2];
-
+			center[0] *= recip[0];
+			center[1] *= recip[1];
+			center[2] *= recip[2];
 		}
-
 	}
 
+	vtx = (const char *)svertices;
 
-
-	vtx = (const char *) svertices;
-
-	for (unsigned int i=0; i<svcount; i++)
+	for (unsigned int i = 0; i < svcount; i++)
 	{
 		const btVector3 *p = (const btVector3 *)vtx;
-		vtx+=stride;
+		vtx += stride;
 
 		btScalar px = p->getX();
 		btScalar py = p->getY();
 		btScalar pz = p->getZ();
 
-		if ( scale )
+		if (scale)
 		{
-			px = px*recip[0]; // normalize
-			py = py*recip[1]; // normalize
-			pz = pz*recip[2]; // normalize
+			px = px * recip[0];  // normalize
+			py = py * recip[1];  // normalize
+			pz = pz * recip[2];  // normalize
 		}
 
-//		if ( 1 )
+		//		if ( 1 )
 		{
 			unsigned int j;
 
-			for (j=0; j<vcount; j++)
+			for (j = 0; j < vcount; j++)
 			{
 				/// XXX might be broken
-				btVector3& v = vertices[j];
+				btVector3 &v = vertices[j];
 
 				btScalar x = v[0];
 				btScalar y = v[1];
 				btScalar z = v[2];
 
-				btScalar dx = btFabs(x - px );
-				btScalar dy = btFabs(y - py );
-				btScalar dz = btFabs(z - pz );
+				btScalar dx = btFabs(x - px);
+				btScalar dy = btFabs(y - py);
+				btScalar dz = btFabs(z - pz);
 
-				if ( dx < normalepsilon && dy < normalepsilon && dz < normalepsilon )
+				if (dx < normalepsilon && dy < normalepsilon && dz < normalepsilon)
 				{
 					// ok, it is close enough to the old one
 					// now let us see if it is further from the center of the point cloud than the one we already recorded.
 					// in which case we keep this one instead.
 
-					btScalar dist1 = GetDist(px,py,pz,center);
-					btScalar dist2 = GetDist(v[0],v[1],v[2],center);
+					btScalar dist1 = GetDist(px, py, pz, center);
+					btScalar dist2 = GetDist(v[0], v[1], v[2], center);
 
-					if ( dist1 > dist2 )
+					if (dist1 > dist2)
 					{
 						v[0] = px;
 						v[1] = py;
 						v[2] = pz;
-						
 					}
 
 					break;
 				}
 			}
 
-			if ( j == vcount )
+			if (j == vcount)
 			{
-				btVector3& dest = vertices[vcount];
+				btVector3 &dest = vertices[vcount];
 				dest[0] = px;
 				dest[1] = py;
 				dest[2] = pz;
@@ -1042,18 +1000,18 @@ bool  HullLibrary::CleanupVertices(unsigned int svcount,
 	}
 
 	// ok..now make sure we didn't prune so many vertices it is now invalid.
-//	if ( 1 )
+	//	if ( 1 )
 	{
-		btScalar bmin[3] = {  FLT_MAX,  FLT_MAX,  FLT_MAX };
-		btScalar bmax[3] = { -FLT_MAX, -FLT_MAX, -FLT_MAX };
+		btScalar bmin[3] = {FLT_MAX, FLT_MAX, FLT_MAX};
+		btScalar bmax[3] = {-FLT_MAX, -FLT_MAX, -FLT_MAX};
 
-		for (unsigned int i=0; i<vcount; i++)
+		for (unsigned int i = 0; i < vcount; i++)
 		{
-			const btVector3& p = vertices[i];
-			for (int j=0; j<3; j++)
+			const btVector3 &p = vertices[i];
+			for (int j = 0; j < 3; j++)
 			{
-				if ( p[j] < bmin[j] ) bmin[j] = p[j];
-				if ( p[j] > bmax[j] ) bmax[j] = p[j];
+				if (p[j] < bmin[j]) bmin[j] = p[j];
+				if (p[j] > bmax[j]) bmax[j] = p[j];
 			}
 		}
 
@@ -1061,27 +1019,27 @@ bool  HullLibrary::CleanupVertices(unsigned int svcount,
 		btScalar dy = bmax[1] - bmin[1];
 		btScalar dz = bmax[2] - bmin[2];
 
-		if ( dx < EPSILON || dy < EPSILON || dz < EPSILON || vcount < 3)
+		if (dx < EPSILON || dy < EPSILON || dz < EPSILON || vcount < 3)
 		{
-			btScalar cx = dx*btScalar(0.5) + bmin[0];
-			btScalar cy = dy*btScalar(0.5) + bmin[1];
-			btScalar cz = dz*btScalar(0.5) + bmin[2];
+			btScalar cx = dx * btScalar(0.5) + bmin[0];
+			btScalar cy = dy * btScalar(0.5) + bmin[1];
+			btScalar cz = dz * btScalar(0.5) + bmin[2];
 
 			btScalar len = FLT_MAX;
 
-			if ( dx >= EPSILON && dx < len ) len = dx;
-			if ( dy >= EPSILON && dy < len ) len = dy;
-			if ( dz >= EPSILON && dz < len ) len = dz;
+			if (dx >= EPSILON && dx < len) len = dx;
+			if (dy >= EPSILON && dy < len) len = dy;
+			if (dz >= EPSILON && dz < len) len = dz;
 
-			if ( len == FLT_MAX )
+			if (len == FLT_MAX)
 			{
-				dx = dy = dz = btScalar(0.01); // one centimeter
+				dx = dy = dz = btScalar(0.01);  // one centimeter
 			}
 			else
 			{
-				if ( dx < EPSILON ) dx = len * btScalar(0.05); // 1/5th the shortest non-zero edge.
-				if ( dy < EPSILON ) dy = len * btScalar(0.05);
-				if ( dz < EPSILON ) dz = len * btScalar(0.05);
+				if (dx < EPSILON) dx = len * btScalar(0.05);  // 1/5th the shortest non-zero edge.
+				if (dy < EPSILON) dy = len * btScalar(0.05);
+				if (dz < EPSILON) dz = len * btScalar(0.05);
 			}
 
 			btScalar x1 = cx - dx;
@@ -1093,16 +1051,16 @@ bool  HullLibrary::CleanupVertices(unsigned int svcount,
 			btScalar z1 = cz - dz;
 			btScalar z2 = cz + dz;
 
-			vcount = 0; // add box
+			vcount = 0;  // add box
 
-			addPoint(vcount,vertices,x1,y1,z1);
-			addPoint(vcount,vertices,x2,y1,z1);
-			addPoint(vcount,vertices,x2,y2,z1);
-			addPoint(vcount,vertices,x1,y2,z1);
-			addPoint(vcount,vertices,x1,y1,z2);
-			addPoint(vcount,vertices,x2,y1,z2);
-			addPoint(vcount,vertices,x2,y2,z2);
-			addPoint(vcount,vertices,x1,y2,z2);
+			addPoint(vcount, vertices, x1, y1, z1);
+			addPoint(vcount, vertices, x2, y1, z1);
+			addPoint(vcount, vertices, x2, y2, z1);
+			addPoint(vcount, vertices, x1, y2, z1);
+			addPoint(vcount, vertices, x1, y1, z2);
+			addPoint(vcount, vertices, x2, y1, z2);
+			addPoint(vcount, vertices, x2, y2, z2);
+			addPoint(vcount, vertices, x1, y2, z2);
 
 			return true;
 		}
@@ -1111,57 +1069,52 @@ bool  HullLibrary::CleanupVertices(unsigned int svcount,
 	return true;
 }
 
-void HullLibrary::BringOutYourDead(const btVector3* verts,unsigned int vcount, btVector3* overts,unsigned int &ocount,unsigned int *indices,unsigned indexcount)
+void HullLibrary::BringOutYourDead(const btVector3 *verts, unsigned int vcount, btVector3 *overts, unsigned int &ocount, unsigned int *indices, unsigned indexcount)
 {
-	btAlignedObjectArray<int>tmpIndices;
+	btAlignedObjectArray<int> tmpIndices;
 	tmpIndices.resize(m_vertexIndexMapping.size());
 	int i;
 
-	for (i=0;i<m_vertexIndexMapping.size();i++)
+	for (i = 0; i < m_vertexIndexMapping.size(); i++)
 	{
 		tmpIndices[i] = m_vertexIndexMapping[i];
 	}
 
 	TUIntArray usedIndices;
 	usedIndices.resize(static_cast<int>(vcount));
-	memset(&usedIndices[0],0,sizeof(unsigned int)*vcount);
+	memset(&usedIndices[0], 0, sizeof(unsigned int) * vcount);
 
 	ocount = 0;
 
-	for (i=0; i<int (indexcount); i++)
+	for (i = 0; i < int(indexcount); i++)
 	{
-		unsigned int v = indices[i]; // original array index
+		unsigned int v = indices[i];  // original array index
 
-		btAssert( v >= 0 && v < vcount );
+		btAssert(v >= 0 && v < vcount);
 
-		if ( usedIndices[static_cast<int>(v)] ) // if already remapped
+		if (usedIndices[static_cast<int>(v)])  // if already remapped
 		{
-			indices[i] = usedIndices[static_cast<int>(v)]-1; // index to new array
+			indices[i] = usedIndices[static_cast<int>(v)] - 1;  // index to new array
 		}
 		else
 		{
+			indices[i] = ocount;  // new index mapping
 
-			indices[i] = ocount;      // new index mapping
-
-			overts[ocount][0] = verts[v][0]; // copy old vert to new vert array
+			overts[ocount][0] = verts[v][0];  // copy old vert to new vert array
 			overts[ocount][1] = verts[v][1];
 			overts[ocount][2] = verts[v][2];
 
-			for (int k=0;k<m_vertexIndexMapping.size();k++)
+			for (int k = 0; k < m_vertexIndexMapping.size(); k++)
 			{
-				if (tmpIndices[k]==int(v))
-					m_vertexIndexMapping[k]=ocount;
+				if (tmpIndices[k] == int(v))
+					m_vertexIndexMapping[k] = ocount;
 			}
 
-			ocount++; // increment output vert count
+			ocount++;  // increment output vert count
 
-			btAssert( ocount >=0 && ocount <= vcount );
+			btAssert(ocount >= 0 && ocount <= vcount);
 
-			usedIndices[static_cast<int>(v)] = ocount; // assign new index remapping
-
-		
+			usedIndices[static_cast<int>(v)] = ocount;  // assign new index remapping
 		}
 	}
-
-	
 }
diff --git a/thirdparty/bullet/LinearMath/btConvexHull.h b/thirdparty/bullet/LinearMath/btConvexHull.h
index 69c52bc6f83..f890d75ea1d 100644
--- a/thirdparty/bullet/LinearMath/btConvexHull.h
+++ b/thirdparty/bullet/LinearMath/btConvexHull.h
@@ -34,106 +34,102 @@ public:
 		mNumFaces = 0;
 		mNumIndices = 0;
 	}
-	bool                    mPolygons;                  // true if indices represents polygons, false indices are triangles
-	unsigned int            mNumOutputVertices;         // number of vertices in the output hull
-	btAlignedObjectArray<btVector3>	m_OutputVertices;            // array of vertices
-	unsigned int            mNumFaces;                  // the number of faces produced
-	unsigned int            mNumIndices;                // the total number of indices
-	btAlignedObjectArray<unsigned int>    m_Indices;                   // pointer to indices.
+	bool mPolygons;                                    // true if indices represents polygons, false indices are triangles
+	unsigned int mNumOutputVertices;                   // number of vertices in the output hull
+	btAlignedObjectArray<btVector3> m_OutputVertices;  // array of vertices
+	unsigned int mNumFaces;                            // the number of faces produced
+	unsigned int mNumIndices;                          // the total number of indices
+	btAlignedObjectArray<unsigned int> m_Indices;      // pointer to indices.
 
-// If triangles, then indices are array indexes into the vertex list.
-// If polygons, indices are in the form (number of points in face) (p1, p2, p3, ..) etc..
+	// If triangles, then indices are array indexes into the vertex list.
+	// If polygons, indices are in the form (number of points in face) (p1, p2, p3, ..) etc..
 };
 
 enum HullFlag
 {
-	QF_TRIANGLES         = (1<<0),             // report results as triangles, not polygons.
-	QF_REVERSE_ORDER     = (1<<1),             // reverse order of the triangle indices.
-	QF_DEFAULT           = QF_TRIANGLES
+	QF_TRIANGLES = (1 << 0),      // report results as triangles, not polygons.
+	QF_REVERSE_ORDER = (1 << 1),  // reverse order of the triangle indices.
+	QF_DEFAULT = QF_TRIANGLES
 };
 
-
 class HullDesc
 {
 public:
 	HullDesc(void)
 	{
-		mFlags          = QF_DEFAULT;
-		mVcount         = 0;
-		mVertices       = 0;
-		mVertexStride   = sizeof(btVector3);
-		mNormalEpsilon  = 0.001f;
-		mMaxVertices	= 4096; // maximum number of points to be considered for a convex hull.
-		mMaxFaces	= 4096;
+		mFlags = QF_DEFAULT;
+		mVcount = 0;
+		mVertices = 0;
+		mVertexStride = sizeof(btVector3);
+		mNormalEpsilon = 0.001f;
+		mMaxVertices = 4096;  // maximum number of points to be considered for a convex hull.
+		mMaxFaces = 4096;
 	};
 
 	HullDesc(HullFlag flag,
-		 unsigned int vcount,
-		 const btVector3 *vertices,
-		 unsigned int stride = sizeof(btVector3))
+			 unsigned int vcount,
+			 const btVector3* vertices,
+			 unsigned int stride = sizeof(btVector3))
 	{
-		mFlags          = flag;
-		mVcount         = vcount;
-		mVertices       = vertices;
-		mVertexStride   = stride;
-		mNormalEpsilon  = btScalar(0.001);
-		mMaxVertices    = 4096;
+		mFlags = flag;
+		mVcount = vcount;
+		mVertices = vertices;
+		mVertexStride = stride;
+		mNormalEpsilon = btScalar(0.001);
+		mMaxVertices = 4096;
 	}
 
 	bool HasHullFlag(HullFlag flag) const
 	{
-		if ( mFlags & flag ) return true;
+		if (mFlags & flag) return true;
 		return false;
 	}
 
 	void SetHullFlag(HullFlag flag)
 	{
-		mFlags|=flag;
+		mFlags |= flag;
 	}
 
 	void ClearHullFlag(HullFlag flag)
 	{
-		mFlags&=~flag;
+		mFlags &= ~flag;
 	}
 
-	unsigned int      mFlags;           // flags to use when generating the convex hull.
-	unsigned int      mVcount;          // number of vertices in the input point cloud
-	const btVector3  *mVertices;        // the array of vertices.
-	unsigned int      mVertexStride;    // the stride of each vertex, in bytes.
-	btScalar             mNormalEpsilon;   // the epsilon for removing duplicates.  This is a normalized value, if normalized bit is on.
-	unsigned int      mMaxVertices;     // maximum number of vertices to be considered for the hull!
-	unsigned int      mMaxFaces;
+	unsigned int mFlags;         // flags to use when generating the convex hull.
+	unsigned int mVcount;        // number of vertices in the input point cloud
+	const btVector3* mVertices;  // the array of vertices.
+	unsigned int mVertexStride;  // the stride of each vertex, in bytes.
+	btScalar mNormalEpsilon;     // the epsilon for removing duplicates.  This is a normalized value, if normalized bit is on.
+	unsigned int mMaxVertices;   // maximum number of vertices to be considered for the hull!
+	unsigned int mMaxFaces;
 };
 
 enum HullError
 {
-	QE_OK,            // success!
-	QE_FAIL           // failed.
+	QE_OK,   // success!
+	QE_FAIL  // failed.
 };
 
 class btPlane
 {
-	public:
-	btVector3	normal;
-	btScalar	dist;   // distance below origin - the D from plane equasion Ax+By+Cz+D=0
-			btPlane(const btVector3 &n,btScalar d):normal(n),dist(d){}
-			btPlane():normal(),dist(0){}
-	
+public:
+	btVector3 normal;
+	btScalar dist;  // distance below origin - the D from plane equasion Ax+By+Cz+D=0
+	btPlane(const btVector3& n, btScalar d) : normal(n), dist(d) {}
+	btPlane() : normal(), dist(0) {}
 };
 
-
-
-class ConvexH 
+class ConvexH
 {
-  public:
+public:
 	class HalfEdge
 	{
-	  public:
+	public:
 		short ea;         // the other half of the edge (index into edges list)
 		unsigned char v;  // the vertex at the start of this edge (index into vertices list)
 		unsigned char p;  // the facet on which this edge lies (index into facets list)
-		HalfEdge(){}
-		HalfEdge(short _ea,unsigned char _v, unsigned char _p):ea(_ea),v(_v),p(_p){}
+		HalfEdge() {}
+		HalfEdge(short _ea, unsigned char _v, unsigned char _p) : ea(_ea), v(_v), p(_p) {}
 	};
 	ConvexH()
 	{
@@ -143,25 +139,29 @@ class ConvexH
 	}
 	btAlignedObjectArray<btVector3> vertices;
 	btAlignedObjectArray<HalfEdge> edges;
-	btAlignedObjectArray<btPlane>  facets;
-	ConvexH(int vertices_size,int edges_size,int facets_size);
+	btAlignedObjectArray<btPlane> facets;
+	ConvexH(int vertices_size, int edges_size, int facets_size);
 };
 
-
 class int4
 {
 public:
-	int x,y,z,w;
+	int x, y, z, w;
 	int4(){};
-	int4(int _x,int _y, int _z,int _w){x=_x;y=_y;z=_z;w=_w;}
-	const int& operator[](int i) const {return (&x)[i];}
-	int& operator[](int i) {return (&x)[i];}
+	int4(int _x, int _y, int _z, int _w)
+	{
+		x = _x;
+		y = _y;
+		z = _z;
+		w = _w;
+	}
+	const int& operator[](int i) const { return (&x)[i]; }
+	int& operator[](int i) { return (&x)[i]; }
 };
 
 class PHullResult
 {
 public:
-
 	PHullResult(void)
 	{
 		mVcount = 0;
@@ -173,69 +173,61 @@ public:
 	unsigned int mVcount;
 	unsigned int mIndexCount;
 	unsigned int mFaceCount;
-	btVector3*   mVertices;
+	btVector3* mVertices;
 	TUIntArray m_Indices;
 };
 
-
-
 ///The HullLibrary class can create a convex hull from a collection of vertices, using the ComputeHull method.
 ///The btShapeHull class uses this HullLibrary to create a approximate convex mesh given a general (non-polyhedral) convex shape.
 class HullLibrary
 {
-
 	btAlignedObjectArray<class btHullTriangle*> m_tris;
 
 public:
-
 	btAlignedObjectArray<int> m_vertexIndexMapping;
 
-
-	HullError CreateConvexHull(const HullDesc& desc, // describes the input request
-				   HullResult&     result);        // contains the resulst
-	HullError ReleaseResult(HullResult &result); // release memory allocated for this result, we are done with it.
+	HullError CreateConvexHull(const HullDesc& desc,  // describes the input request
+							   HullResult& result);   // contains the resulst
+	HullError ReleaseResult(HullResult& result);      // release memory allocated for this result, we are done with it.
 
 private:
+	bool ComputeHull(unsigned int vcount, const btVector3* vertices, PHullResult& result, unsigned int vlimit);
 
-	bool ComputeHull(unsigned int vcount,const btVector3 *vertices,PHullResult &result,unsigned int vlimit);
+	class btHullTriangle* allocateTriangle(int a, int b, int c);
+	void deAllocateTriangle(btHullTriangle*);
+	void b2bfix(btHullTriangle* s, btHullTriangle* t);
 
-	class btHullTriangle*	allocateTriangle(int a,int b,int c);
-	void	deAllocateTriangle(btHullTriangle*);
-	void b2bfix(btHullTriangle* s,btHullTriangle*t);
+	void removeb2b(btHullTriangle* s, btHullTriangle* t);
 
-	void removeb2b(btHullTriangle* s,btHullTriangle*t);
-
-	void checkit(btHullTriangle *t);
+	void checkit(btHullTriangle* t);
 
 	btHullTriangle* extrudable(btScalar epsilon);
 
-	int calchull(btVector3 *verts,int verts_count, TUIntArray& tris_out, int &tris_count,int vlimit);
+	int calchull(btVector3* verts, int verts_count, TUIntArray& tris_out, int& tris_count, int vlimit);
 
-	int calchullgen(btVector3 *verts,int verts_count, int vlimit);
+	int calchullgen(btVector3* verts, int verts_count, int vlimit);
 
-	int4 FindSimplex(btVector3 *verts,int verts_count,btAlignedObjectArray<int> &allow);
+	int4 FindSimplex(btVector3* verts, int verts_count, btAlignedObjectArray<int>& allow);
 
-	class ConvexH* ConvexHCrop(ConvexH& convex,const btPlane& slice);
+	class ConvexH* ConvexHCrop(ConvexH& convex, const btPlane& slice);
 
-	void extrude(class btHullTriangle* t0,int v);
+	void extrude(class btHullTriangle* t0, int v);
 
 	ConvexH* test_cube();
 
-	//BringOutYourDead (John Ratcliff): When you create a convex hull you hand it a large input set of vertices forming a 'point cloud'. 
+	//BringOutYourDead (John Ratcliff): When you create a convex hull you hand it a large input set of vertices forming a 'point cloud'.
 	//After the hull is generated it give you back a set of polygon faces which index the *original* point cloud.
 	//The thing is, often times, there are many 'dead vertices' in the point cloud that are on longer referenced by the hull.
 	//The routine 'BringOutYourDead' find only the referenced vertices, copies them to an new buffer, and re-indexes the hull so that it is a minimal representation.
-	void BringOutYourDead(const btVector3* verts,unsigned int vcount, btVector3* overts,unsigned int &ocount,unsigned int* indices,unsigned indexcount);
+	void BringOutYourDead(const btVector3* verts, unsigned int vcount, btVector3* overts, unsigned int& ocount, unsigned int* indices, unsigned indexcount);
 
 	bool CleanupVertices(unsigned int svcount,
-			     const btVector3* svertices,
-			     unsigned int stride,
-			     unsigned int &vcount, // output number of vertices
-			     btVector3* vertices, // location to store the results.
-			     btScalar  normalepsilon,
-			     btVector3& scale);
+						 const btVector3* svertices,
+						 unsigned int stride,
+						 unsigned int& vcount,  // output number of vertices
+						 btVector3* vertices,   // location to store the results.
+						 btScalar normalepsilon,
+						 btVector3& scale);
 };
 
-
-#endif //BT_CD_HULL_H
-
+#endif  //BT_CD_HULL_H
diff --git a/thirdparty/bullet/LinearMath/btConvexHullComputer.cpp b/thirdparty/bullet/LinearMath/btConvexHullComputer.cpp
index 2ea22cbe3ba..8bbfdc5f25b 100644
--- a/thirdparty/bullet/LinearMath/btConvexHullComputer.cpp
+++ b/thirdparty/bullet/LinearMath/btConvexHullComputer.cpp
@@ -20,846 +20,847 @@ subject to the following restrictions:
 #include "btVector3.h"
 
 #ifdef __GNUC__
-	#include <stdint.h>
+#include <stdint.h>
 #elif defined(_MSC_VER)
-	typedef __int32 int32_t;
-	typedef __int64 int64_t;
-	typedef unsigned __int32 uint32_t;
-	typedef unsigned __int64 uint64_t;
+typedef __int32 int32_t;
+typedef __int64 int64_t;
+typedef unsigned __int32 uint32_t;
+typedef unsigned __int64 uint64_t;
 #else
-	typedef int int32_t;
-	typedef long long int int64_t;
-	typedef unsigned int uint32_t;
-	typedef unsigned long long int uint64_t;
+typedef int int32_t;
+typedef long long int int64_t;
+typedef unsigned int uint32_t;
+typedef unsigned long long int uint64_t;
 #endif
 
-
 //The definition of USE_X86_64_ASM is moved into the build system. You can enable it manually by commenting out the following lines
 //#if (defined(__GNUC__) && defined(__x86_64__) && !defined(__ICL))  // || (defined(__ICL) && defined(_M_X64))   bug in Intel compiler, disable inline assembly
 //	#define USE_X86_64_ASM
 //#endif
 
-
 //#define DEBUG_CONVEX_HULL
 //#define SHOW_ITERATIONS
 
 #if defined(DEBUG_CONVEX_HULL) || defined(SHOW_ITERATIONS)
-	#include <stdio.h>
+#include <stdio.h>
 #endif
 
 // Convex hull implementation based on Preparata and Hong
 // Ole Kniemeyer, MAXON Computer GmbH
 class btConvexHullInternal
 {
+public:
+	class Point64
+	{
 	public:
-		
-		class Point64
+		int64_t x;
+		int64_t y;
+		int64_t z;
+
+		Point64(int64_t x, int64_t y, int64_t z) : x(x), y(y), z(z)
 		{
-			public:
-				int64_t x;
-				int64_t y;
-				int64_t z;
-				
-				Point64(int64_t x, int64_t y, int64_t z): x(x), y(y), z(z)
-				{
-				}
-
-				bool isZero()
-				{
-					return (x == 0) && (y == 0) && (z == 0);
-				}
-
-				int64_t dot(const Point64& b) const
-				{
-					return x * b.x + y * b.y + z * b.z;
-				}
-		};
-		
-		class Point32
-		{
-			public:
-				int32_t x;
-				int32_t y;
-				int32_t z;
-				int index;
-				
-				Point32()
-				{
-				}
-				
-				Point32(int32_t x, int32_t y, int32_t z): x(x), y(y), z(z), index(-1)
-				{
-				}
-				
-				bool operator==(const Point32& b) const
-				{
-					return (x == b.x) && (y == b.y) && (z == b.z);
-				}
-
-				bool operator!=(const Point32& b) const
-				{
-					return (x != b.x) || (y != b.y) || (z != b.z);
-				}
-
-				bool isZero()
-				{
-					return (x == 0) && (y == 0) && (z == 0);
-				}
-
-				Point64 cross(const Point32& b) const
-				{
-					return Point64(y * b.z - z * b.y, z * b.x - x * b.z, x * b.y - y * b.x);
-				}
-
-				Point64 cross(const Point64& b) const
-				{
-					return Point64(y * b.z - z * b.y, z * b.x - x * b.z, x * b.y - y * b.x);
-				}
-
-				int64_t dot(const Point32& b) const
-				{
-					return x * b.x + y * b.y + z * b.z;
-				}
-
-				int64_t dot(const Point64& b) const
-				{
-					return x * b.x + y * b.y + z * b.z;
-				}
-
-				Point32 operator+(const Point32& b) const
-				{
-					return Point32(x + b.x, y + b.y, z + b.z);
-				}
-
-				Point32 operator-(const Point32& b) const
-				{
-					return Point32(x - b.x, y - b.y, z - b.z);
-				}
-		};
-
-		class Int128
-		{
-			public:
-				uint64_t low;
-				uint64_t high;
-
-				Int128()
-				{
-				}
-
-				Int128(uint64_t low, uint64_t high): low(low), high(high)
-				{
-				}
-
-				Int128(uint64_t low): low(low), high(0)
-				{
-				}
-
-				Int128(int64_t value): low(value), high((value >= 0) ? 0 : (uint64_t) -1LL)
-				{
-				}
-
-				static Int128 mul(int64_t a, int64_t b);
-
-				static Int128 mul(uint64_t a, uint64_t b);
-
-				Int128 operator-() const
-				{
-					return Int128((uint64_t) -(int64_t)low, ~high + (low == 0));
-				}
-
-				Int128 operator+(const Int128& b) const
-				{
-#ifdef USE_X86_64_ASM
-					Int128 result;
-					__asm__ ("addq %[bl], %[rl]\n\t"
-									 "adcq %[bh], %[rh]\n\t"
-									 : [rl] "=r" (result.low), [rh] "=r" (result.high)
-									 : "0"(low), "1"(high), [bl] "g"(b.low), [bh] "g"(b.high)
-									 : "cc" );
-					return result;
-#else
-					uint64_t lo = low + b.low;
-					return Int128(lo, high + b.high + (lo < low));
-#endif
-				}
-
-				Int128 operator-(const Int128& b) const
-				{
-#ifdef USE_X86_64_ASM
-					Int128 result;
-					__asm__ ("subq %[bl], %[rl]\n\t"
-									 "sbbq %[bh], %[rh]\n\t"
-									 : [rl] "=r" (result.low), [rh] "=r" (result.high)
-									 : "0"(low), "1"(high), [bl] "g"(b.low), [bh] "g"(b.high)
-									 : "cc" );
-					return result;
-#else
-					return *this + -b;
-#endif
-				}
-
-				Int128& operator+=(const Int128& b)
-				{
-#ifdef USE_X86_64_ASM
-					__asm__ ("addq %[bl], %[rl]\n\t"
-									 "adcq %[bh], %[rh]\n\t"
-									 : [rl] "=r" (low), [rh] "=r" (high)
-									 : "0"(low), "1"(high), [bl] "g"(b.low), [bh] "g"(b.high)
-									 : "cc" );
-#else
-					uint64_t lo = low + b.low;
-					if (lo < low)
-					{
-						++high;
-					}
-					low = lo;
-					high += b.high;
-#endif
-					return *this;
-				}
-
-				Int128& operator++()
-				{
-					if (++low == 0)
-					{
-						++high;
-					}
-					return *this;
-				}
-
-				Int128 operator*(int64_t b) const;
-
-				btScalar toScalar() const
-				{
-					return ((int64_t) high >= 0) ? btScalar(high) * (btScalar(0x100000000LL) * btScalar(0x100000000LL)) + btScalar(low)
-						: -(-*this).toScalar();
-				}
-
-				int getSign() const
-				{
-					return ((int64_t) high < 0) ? -1 : (high || low) ? 1 : 0;
-				}
-
-				bool operator<(const Int128& b) const
-				{
-					return (high < b.high) || ((high == b.high) && (low < b.low));
-				}
-
-				int ucmp(const Int128&b) const
-				{
-					if (high < b.high)
-					{
-						return -1;
-					}
-					if (high > b.high)
-					{
-						return 1;
-					}
-					if (low < b.low)
-					{
-						return -1;
-					}
-					if (low > b.low)
-					{
-						return 1;
-					}
-					return 0;
-				}
-		};
-
-
-		class Rational64
-		{
-			private:
-				uint64_t m_numerator;
-				uint64_t m_denominator;
-				int sign;
-				
-			public:
-				Rational64(int64_t numerator, int64_t denominator)
-				{
-					if (numerator > 0)
-					{
-						sign = 1;
-						m_numerator = (uint64_t) numerator;
-					}
-					else if (numerator < 0)
-					{
-						sign = -1;
-						m_numerator = (uint64_t) -numerator;
-					}
-					else
-					{
-						sign = 0;
-						m_numerator = 0;
-					}
-					if (denominator > 0)
-					{
-						m_denominator = (uint64_t) denominator;
-					}
-					else if (denominator < 0)
-					{
-						sign = -sign;
-						m_denominator = (uint64_t) -denominator;
-					}
-					else
-					{
-						m_denominator = 0;
-					}
-				}
-				
-				bool isNegativeInfinity() const
-				{
-					return (sign < 0) && (m_denominator == 0);
-				}
-				
-				bool isNaN() const
-				{
-					return (sign == 0) && (m_denominator == 0);
-				}
-				
-				int compare(const Rational64& b) const;
-				
-				btScalar toScalar() const
-				{
-					return sign * ((m_denominator == 0) ? SIMD_INFINITY : (btScalar) m_numerator / m_denominator);
-				}
-		};
-
-
-		class Rational128
-		{
-			private:
-				Int128 numerator;
-				Int128 denominator;
-				int sign;
-				bool isInt64;
-
-			public:
-				Rational128(int64_t value)
-				{
-					if (value > 0)
-					{
-						sign = 1;
-						this->numerator = value;
-					}
-					else if (value < 0)
-					{
-						sign = -1;
-						this->numerator = -value;
-					}
-					else
-					{
-						sign = 0;
-						this->numerator = (uint64_t) 0;
-					}
-					this->denominator = (uint64_t) 1;
-					isInt64 = true;
-				}
-
-				Rational128(const Int128& numerator, const Int128& denominator)
-				{
-					sign = numerator.getSign();
-					if (sign >= 0)
-					{
-						this->numerator = numerator;
-					}
-					else
-					{
-						this->numerator = -numerator;
-					}
-					int dsign = denominator.getSign();
-					if (dsign >= 0)
-					{
-						this->denominator = denominator;
-					}
-					else
-					{
-						sign = -sign;
-						this->denominator = -denominator;
-					}
-					isInt64 = false;
-				}
-
-				int compare(const Rational128& b) const;
-
-				int compare(int64_t b) const;
-
-				btScalar toScalar() const
-				{
-					return sign * ((denominator.getSign() == 0) ? SIMD_INFINITY : numerator.toScalar() / denominator.toScalar());
-				}
-		};
-
-		class PointR128
-		{
-			public:
-				Int128 x;
-				Int128 y;
-				Int128 z;
-				Int128 denominator;
-
-				PointR128()
-				{
-				}
-
-				PointR128(Int128 x, Int128 y, Int128 z, Int128 denominator): x(x), y(y), z(z), denominator(denominator)
-				{
-				}
-
-				btScalar xvalue() const
-				{
-					return x.toScalar() / denominator.toScalar();
-				}
-
-				btScalar yvalue() const
-				{
-					return y.toScalar() / denominator.toScalar();
-				}
-
-				btScalar zvalue() const
-				{
-					return z.toScalar() / denominator.toScalar();
-				}
-		};
-
-
-		class Edge;
-		class Face;
-
-		class Vertex
-		{
-			public:
-				Vertex* next;
-				Vertex* prev;
-				Edge* edges;
-				Face* firstNearbyFace;
-				Face* lastNearbyFace;
-				PointR128 point128;
-				Point32 point;
-				int copy;
-				
-				Vertex(): next(NULL), prev(NULL), edges(NULL), firstNearbyFace(NULL), lastNearbyFace(NULL), copy(-1)
-				{
-				}
-
-#ifdef DEBUG_CONVEX_HULL
-				void print()
-				{
-					printf("V%d (%d, %d, %d)", point.index, point.x, point.y, point.z);
-				}
-
-				void printGraph();
-#endif
-
-				Point32 operator-(const Vertex& b) const
-				{
-					return point - b.point;
-				}
-
-				Rational128 dot(const Point64& b) const
-				{
-					return (point.index >= 0) ? Rational128(point.dot(b))
-						: Rational128(point128.x * b.x + point128.y * b.y + point128.z * b.z, point128.denominator);
-				}
-
-				btScalar xvalue() const
-				{
-					return (point.index >= 0) ? btScalar(point.x) : point128.xvalue();
-				}
-
-				btScalar yvalue() const
-				{
-					return (point.index >= 0) ? btScalar(point.y) : point128.yvalue();
-				}
-
-				btScalar zvalue() const
-				{
-					return (point.index >= 0) ? btScalar(point.z) : point128.zvalue();
-				}
-
-				void receiveNearbyFaces(Vertex* src)
-				{
-					if (lastNearbyFace)
-					{
-						lastNearbyFace->nextWithSameNearbyVertex = src->firstNearbyFace;
-					}
-					else
-					{
-						firstNearbyFace = src->firstNearbyFace;
-					}
-					if (src->lastNearbyFace)
-					{
-						lastNearbyFace = src->lastNearbyFace;
-					}
-					for (Face* f = src->firstNearbyFace; f; f = f->nextWithSameNearbyVertex)
-					{
-						btAssert(f->nearbyVertex == src);
-						f->nearbyVertex = this;
-					}
-					src->firstNearbyFace = NULL;
-					src->lastNearbyFace = NULL;
-				}
-		};
-
-
-		class Edge
-		{
-			public:
-				Edge* next;
-				Edge* prev;
-				Edge* reverse;
-				Vertex* target;
-				Face* face;
-				int copy;
-
-				~Edge()
-				{
-					next = NULL;
-					prev = NULL;
-					reverse = NULL;
-					target = NULL;
-					face = NULL;
-				}
-
-				void link(Edge* n)
-				{
-					btAssert(reverse->target == n->reverse->target);
-					next = n;
-					n->prev = this;
-				}
-
-#ifdef DEBUG_CONVEX_HULL
-				void print()
-				{
-					printf("E%p : %d -> %d,  n=%p p=%p   (0 %d\t%d\t%d) -> (%d %d %d)", this, reverse->target->point.index, target->point.index, next, prev,
-								 reverse->target->point.x, reverse->target->point.y, reverse->target->point.z, target->point.x, target->point.y, target->point.z);
-				}
-#endif
-		};
-
-		class Face
-		{
-			public:
-				Face* next;
-				Vertex* nearbyVertex;
-				Face* nextWithSameNearbyVertex;
-				Point32 origin;
-				Point32 dir0;
-				Point32 dir1;
-
-				Face(): next(NULL), nearbyVertex(NULL), nextWithSameNearbyVertex(NULL)
-				{
-				}
-
-				void init(Vertex* a, Vertex* b, Vertex* c)
-				{
-					nearbyVertex = a;
-					origin = a->point;
-					dir0 = *b - *a;
-					dir1 = *c - *a;
-					if (a->lastNearbyFace)
-					{
-						a->lastNearbyFace->nextWithSameNearbyVertex = this;
-					}
-					else
-					{
-						a->firstNearbyFace = this;
-					}
-					a->lastNearbyFace = this;
-				}
-
-				Point64 getNormal()
-				{
-					return dir0.cross(dir1);
-				}
-		};
-
-		template<typename UWord, typename UHWord> class DMul
-		{
-			private:
-				static uint32_t high(uint64_t value)
-				{
-					return (uint32_t) (value >> 32);
-				}
-				
-				static uint32_t low(uint64_t value)
-				{
-					return (uint32_t) value;
-				}
-				
-				static uint64_t mul(uint32_t a, uint32_t b)
-				{
-					return (uint64_t) a * (uint64_t) b;
-				}
-				
-				static void shlHalf(uint64_t& value)
-				{
-					value <<= 32;
-				}
-				
-				static uint64_t high(Int128 value)
-				{
-					return value.high;
-				}
-				
-				static uint64_t low(Int128 value)
-				{
-					return value.low;
-				}
-				
-				static Int128 mul(uint64_t a, uint64_t b)
-				{
-					return Int128::mul(a, b);
-				}
-				
-				static void shlHalf(Int128& value)
-				{
-					value.high = value.low;
-					value.low = 0;
-				}
-				
-			public:
-				
-				static void mul(UWord a, UWord b, UWord& resLow, UWord& resHigh)
-				{
-					UWord p00 = mul(low(a), low(b));
-					UWord p01 = mul(low(a), high(b));
-					UWord p10 = mul(high(a), low(b));
-					UWord p11 = mul(high(a), high(b));
-					UWord p0110 = UWord(low(p01)) + UWord(low(p10));
-					p11 += high(p01);
-					p11 += high(p10);
-					p11 += high(p0110);
-					shlHalf(p0110);
-					p00 += p0110;
-					if (p00 < p0110)
-					{
-						++p11;
-					}
-					resLow = p00;
-					resHigh = p11;
-				}
-		};
-	
-	private:
-
-		class IntermediateHull
-		{
-			public:
-				Vertex* minXy;
-				Vertex* maxXy;
-				Vertex* minYx;
-				Vertex* maxYx;
-				
-				IntermediateHull(): minXy(NULL), maxXy(NULL), minYx(NULL), maxYx(NULL)
-				{
-				}
-				
-				void print();
-		};
-	
-		enum Orientation {NONE, CLOCKWISE, COUNTER_CLOCKWISE};
-
-		template <typename T> class PoolArray
-		{
-			private:
-				T* array;
-				int size;
-
-			public:
-				PoolArray<T>* next;
-
-				PoolArray(int size): size(size), next(NULL)
-				{
-					array = (T*) btAlignedAlloc(sizeof(T) * size, 16);
-				}
-
-				~PoolArray()
-				{
-					btAlignedFree(array);
-				}
-
-				T* init()
-				{
-					T* o = array;
-					for (int i = 0; i < size; i++, o++)
-					{
-						o->next = (i+1 < size) ? o + 1 : NULL;
-					}
-					return array;
-				}
-		};
-
-		template <typename T> class Pool
-		{
-			private:
-				PoolArray<T>* arrays;
-				PoolArray<T>* nextArray;
-				T* freeObjects;
-				int arraySize;
-
-			public:
-				Pool(): arrays(NULL), nextArray(NULL), freeObjects(NULL), arraySize(256)
-				{
-				}
-
-				~Pool()
-				{
-					while (arrays)
-					{
-						PoolArray<T>* p = arrays;
-						arrays = p->next;
-						p->~PoolArray<T>();
-						btAlignedFree(p);
-					}
-				}
-
-				void reset()
-				{
-					nextArray = arrays;
-					freeObjects = NULL;
-				}
-
-				void setArraySize(int arraySize)
-				{
-					this->arraySize = arraySize;
-				}
-
-				T* newObject()
-				{
-					T* o = freeObjects;
-					if (!o)
-					{
-						PoolArray<T>* p = nextArray;
-						if (p)
-						{
-							nextArray = p->next;
-						}
-						else
-						{
-							p = new(btAlignedAlloc(sizeof(PoolArray<T>), 16)) PoolArray<T>(arraySize);
-							p->next = arrays;
-							arrays = p;
-						}
-						o = p->init();
-					}
-					freeObjects = o->next;
-					return new(o) T();
-				};
-
-				void freeObject(T* object)
-				{
-					object->~T();
-					object->next = freeObjects;
-					freeObjects = object;
-				}
-		};
-
-		btVector3 scaling;
-		btVector3 center;
-		Pool<Vertex> vertexPool;
-		Pool<Edge> edgePool;
-		Pool<Face> facePool;
-		btAlignedObjectArray<Vertex*> originalVertices;
-		int mergeStamp;
-		int minAxis;
-		int medAxis;
-		int maxAxis;
-		int usedEdgePairs;
-		int maxUsedEdgePairs;
-
-		static Orientation getOrientation(const Edge* prev, const Edge* next, const Point32& s, const Point32& t);
-		Edge* findMaxAngle(bool ccw, const Vertex* start, const Point32& s, const Point64& rxs, const Point64& sxrxs, Rational64& minCot);
-		void findEdgeForCoplanarFaces(Vertex* c0, Vertex* c1, Edge*& e0, Edge*& e1, Vertex* stop0, Vertex* stop1);
-
-		Edge* newEdgePair(Vertex* from, Vertex* to);
-
-		void removeEdgePair(Edge* edge)
-		{
-			Edge* n = edge->next;
-			Edge* r = edge->reverse;
-
-			btAssert(edge->target && r->target);
-
-			if (n != edge)
-			{
-				n->prev = edge->prev;
-				edge->prev->next = n;
-				r->target->edges = n;
-			}
-			else
-			{
-				r->target->edges = NULL;
-			}
-			
-			n = r->next;
-			
-			if (n != r)
-			{
-				n->prev = r->prev;
-				r->prev->next = n;
-				edge->target->edges = n;
-			}
-			else
-			{
-				edge->target->edges = NULL;
-			}
-
-			edgePool.freeObject(edge);
-			edgePool.freeObject(r);
-			usedEdgePairs--;
 		}
-		
-		void computeInternal(int start, int end, IntermediateHull& result);
-		
-		bool mergeProjection(IntermediateHull& h0, IntermediateHull& h1, Vertex*& c0, Vertex*& c1);
-		
-		void merge(IntermediateHull& h0, IntermediateHull& h1);
 
-		btVector3 toBtVector(const Point32& v);
+		bool isZero()
+		{
+			return (x == 0) && (y == 0) && (z == 0);
+		}
 
-		btVector3 getBtNormal(Face* face);
+		int64_t dot(const Point64& b) const
+		{
+			return x * b.x + y * b.y + z * b.z;
+		}
+	};
 
-		bool shiftFace(Face* face, btScalar amount, btAlignedObjectArray<Vertex*> stack);
+	class Point32
+	{
+	public:
+		int32_t x;
+		int32_t y;
+		int32_t z;
+		int index;
+
+		Point32()
+		{
+		}
+
+		Point32(int32_t x, int32_t y, int32_t z) : x(x), y(y), z(z), index(-1)
+		{
+		}
+
+		bool operator==(const Point32& b) const
+		{
+			return (x == b.x) && (y == b.y) && (z == b.z);
+		}
+
+		bool operator!=(const Point32& b) const
+		{
+			return (x != b.x) || (y != b.y) || (z != b.z);
+		}
+
+		bool isZero()
+		{
+			return (x == 0) && (y == 0) && (z == 0);
+		}
+
+		Point64 cross(const Point32& b) const
+		{
+			return Point64(y * b.z - z * b.y, z * b.x - x * b.z, x * b.y - y * b.x);
+		}
+
+		Point64 cross(const Point64& b) const
+		{
+			return Point64(y * b.z - z * b.y, z * b.x - x * b.z, x * b.y - y * b.x);
+		}
+
+		int64_t dot(const Point32& b) const
+		{
+			return x * b.x + y * b.y + z * b.z;
+		}
+
+		int64_t dot(const Point64& b) const
+		{
+			return x * b.x + y * b.y + z * b.z;
+		}
+
+		Point32 operator+(const Point32& b) const
+		{
+			return Point32(x + b.x, y + b.y, z + b.z);
+		}
+
+		Point32 operator-(const Point32& b) const
+		{
+			return Point32(x - b.x, y - b.y, z - b.z);
+		}
+	};
+
+	class Int128
+	{
+	public:
+		uint64_t low;
+		uint64_t high;
+
+		Int128()
+		{
+		}
+
+		Int128(uint64_t low, uint64_t high) : low(low), high(high)
+		{
+		}
+
+		Int128(uint64_t low) : low(low), high(0)
+		{
+		}
+
+		Int128(int64_t value) : low(value), high((value >= 0) ? 0 : (uint64_t)-1LL)
+		{
+		}
+
+		static Int128 mul(int64_t a, int64_t b);
+
+		static Int128 mul(uint64_t a, uint64_t b);
+
+		Int128 operator-() const
+		{
+			return Int128((uint64_t) - (int64_t)low, ~high + (low == 0));
+		}
+
+		Int128 operator+(const Int128& b) const
+		{
+#ifdef USE_X86_64_ASM
+			Int128 result;
+			__asm__(
+				"addq %[bl], %[rl]\n\t"
+				"adcq %[bh], %[rh]\n\t"
+				: [rl] "=r"(result.low), [rh] "=r"(result.high)
+				: "0"(low), "1"(high), [bl] "g"(b.low), [bh] "g"(b.high)
+				: "cc");
+			return result;
+#else
+			uint64_t lo = low + b.low;
+			return Int128(lo, high + b.high + (lo < low));
+#endif
+		}
+
+		Int128 operator-(const Int128& b) const
+		{
+#ifdef USE_X86_64_ASM
+			Int128 result;
+			__asm__(
+				"subq %[bl], %[rl]\n\t"
+				"sbbq %[bh], %[rh]\n\t"
+				: [rl] "=r"(result.low), [rh] "=r"(result.high)
+				: "0"(low), "1"(high), [bl] "g"(b.low), [bh] "g"(b.high)
+				: "cc");
+			return result;
+#else
+			return *this + -b;
+#endif
+		}
+
+		Int128& operator+=(const Int128& b)
+		{
+#ifdef USE_X86_64_ASM
+			__asm__(
+				"addq %[bl], %[rl]\n\t"
+				"adcq %[bh], %[rh]\n\t"
+				: [rl] "=r"(low), [rh] "=r"(high)
+				: "0"(low), "1"(high), [bl] "g"(b.low), [bh] "g"(b.high)
+				: "cc");
+#else
+			uint64_t lo = low + b.low;
+			if (lo < low)
+			{
+				++high;
+			}
+			low = lo;
+			high += b.high;
+#endif
+			return *this;
+		}
+
+		Int128& operator++()
+		{
+			if (++low == 0)
+			{
+				++high;
+			}
+			return *this;
+		}
+
+		Int128 operator*(int64_t b) const;
+
+		btScalar toScalar() const
+		{
+			return ((int64_t)high >= 0) ? btScalar(high) * (btScalar(0x100000000LL) * btScalar(0x100000000LL)) + btScalar(low)
+										: -(-*this).toScalar();
+		}
+
+		int getSign() const
+		{
+			return ((int64_t)high < 0) ? -1 : (high || low) ? 1 : 0;
+		}
+
+		bool operator<(const Int128& b) const
+		{
+			return (high < b.high) || ((high == b.high) && (low < b.low));
+		}
+
+		int ucmp(const Int128& b) const
+		{
+			if (high < b.high)
+			{
+				return -1;
+			}
+			if (high > b.high)
+			{
+				return 1;
+			}
+			if (low < b.low)
+			{
+				return -1;
+			}
+			if (low > b.low)
+			{
+				return 1;
+			}
+			return 0;
+		}
+	};
+
+	class Rational64
+	{
+	private:
+		uint64_t m_numerator;
+		uint64_t m_denominator;
+		int sign;
 
 	public:
-		Vertex* vertexList;
+		Rational64(int64_t numerator, int64_t denominator)
+		{
+			if (numerator > 0)
+			{
+				sign = 1;
+				m_numerator = (uint64_t)numerator;
+			}
+			else if (numerator < 0)
+			{
+				sign = -1;
+				m_numerator = (uint64_t)-numerator;
+			}
+			else
+			{
+				sign = 0;
+				m_numerator = 0;
+			}
+			if (denominator > 0)
+			{
+				m_denominator = (uint64_t)denominator;
+			}
+			else if (denominator < 0)
+			{
+				sign = -sign;
+				m_denominator = (uint64_t)-denominator;
+			}
+			else
+			{
+				m_denominator = 0;
+			}
+		}
 
-		void compute(const void* coords, bool doubleCoords, int stride, int count);
+		bool isNegativeInfinity() const
+		{
+			return (sign < 0) && (m_denominator == 0);
+		}
 
-		btVector3 getCoordinates(const Vertex* v);
+		bool isNaN() const
+		{
+			return (sign == 0) && (m_denominator == 0);
+		}
 
-		btScalar shrink(btScalar amount, btScalar clampAmount);
+		int compare(const Rational64& b) const;
+
+		btScalar toScalar() const
+		{
+			return sign * ((m_denominator == 0) ? SIMD_INFINITY : (btScalar)m_numerator / m_denominator);
+		}
+	};
+
+	class Rational128
+	{
+	private:
+		Int128 numerator;
+		Int128 denominator;
+		int sign;
+		bool isInt64;
+
+	public:
+		Rational128(int64_t value)
+		{
+			if (value > 0)
+			{
+				sign = 1;
+				this->numerator = value;
+			}
+			else if (value < 0)
+			{
+				sign = -1;
+				this->numerator = -value;
+			}
+			else
+			{
+				sign = 0;
+				this->numerator = (uint64_t)0;
+			}
+			this->denominator = (uint64_t)1;
+			isInt64 = true;
+		}
+
+		Rational128(const Int128& numerator, const Int128& denominator)
+		{
+			sign = numerator.getSign();
+			if (sign >= 0)
+			{
+				this->numerator = numerator;
+			}
+			else
+			{
+				this->numerator = -numerator;
+			}
+			int dsign = denominator.getSign();
+			if (dsign >= 0)
+			{
+				this->denominator = denominator;
+			}
+			else
+			{
+				sign = -sign;
+				this->denominator = -denominator;
+			}
+			isInt64 = false;
+		}
+
+		int compare(const Rational128& b) const;
+
+		int compare(int64_t b) const;
+
+		btScalar toScalar() const
+		{
+			return sign * ((denominator.getSign() == 0) ? SIMD_INFINITY : numerator.toScalar() / denominator.toScalar());
+		}
+	};
+
+	class PointR128
+	{
+	public:
+		Int128 x;
+		Int128 y;
+		Int128 z;
+		Int128 denominator;
+
+		PointR128()
+		{
+		}
+
+		PointR128(Int128 x, Int128 y, Int128 z, Int128 denominator) : x(x), y(y), z(z), denominator(denominator)
+		{
+		}
+
+		btScalar xvalue() const
+		{
+			return x.toScalar() / denominator.toScalar();
+		}
+
+		btScalar yvalue() const
+		{
+			return y.toScalar() / denominator.toScalar();
+		}
+
+		btScalar zvalue() const
+		{
+			return z.toScalar() / denominator.toScalar();
+		}
+	};
+
+	class Edge;
+	class Face;
+
+	class Vertex
+	{
+	public:
+		Vertex* next;
+		Vertex* prev;
+		Edge* edges;
+		Face* firstNearbyFace;
+		Face* lastNearbyFace;
+		PointR128 point128;
+		Point32 point;
+		int copy;
+
+		Vertex() : next(NULL), prev(NULL), edges(NULL), firstNearbyFace(NULL), lastNearbyFace(NULL), copy(-1)
+		{
+		}
+
+#ifdef DEBUG_CONVEX_HULL
+		void print()
+		{
+			printf("V%d (%d, %d, %d)", point.index, point.x, point.y, point.z);
+		}
+
+		void printGraph();
+#endif
+
+		Point32 operator-(const Vertex& b) const
+		{
+			return point - b.point;
+		}
+
+		Rational128 dot(const Point64& b) const
+		{
+			return (point.index >= 0) ? Rational128(point.dot(b))
+									  : Rational128(point128.x * b.x + point128.y * b.y + point128.z * b.z, point128.denominator);
+		}
+
+		btScalar xvalue() const
+		{
+			return (point.index >= 0) ? btScalar(point.x) : point128.xvalue();
+		}
+
+		btScalar yvalue() const
+		{
+			return (point.index >= 0) ? btScalar(point.y) : point128.yvalue();
+		}
+
+		btScalar zvalue() const
+		{
+			return (point.index >= 0) ? btScalar(point.z) : point128.zvalue();
+		}
+
+		void receiveNearbyFaces(Vertex* src)
+		{
+			if (lastNearbyFace)
+			{
+				lastNearbyFace->nextWithSameNearbyVertex = src->firstNearbyFace;
+			}
+			else
+			{
+				firstNearbyFace = src->firstNearbyFace;
+			}
+			if (src->lastNearbyFace)
+			{
+				lastNearbyFace = src->lastNearbyFace;
+			}
+			for (Face* f = src->firstNearbyFace; f; f = f->nextWithSameNearbyVertex)
+			{
+				btAssert(f->nearbyVertex == src);
+				f->nearbyVertex = this;
+			}
+			src->firstNearbyFace = NULL;
+			src->lastNearbyFace = NULL;
+		}
+	};
+
+	class Edge
+	{
+	public:
+		Edge* next;
+		Edge* prev;
+		Edge* reverse;
+		Vertex* target;
+		Face* face;
+		int copy;
+
+		~Edge()
+		{
+			next = NULL;
+			prev = NULL;
+			reverse = NULL;
+			target = NULL;
+			face = NULL;
+		}
+
+		void link(Edge* n)
+		{
+			btAssert(reverse->target == n->reverse->target);
+			next = n;
+			n->prev = this;
+		}
+
+#ifdef DEBUG_CONVEX_HULL
+		void print()
+		{
+			printf("E%p : %d -> %d,  n=%p p=%p   (0 %d\t%d\t%d) -> (%d %d %d)", this, reverse->target->point.index, target->point.index, next, prev,
+				   reverse->target->point.x, reverse->target->point.y, reverse->target->point.z, target->point.x, target->point.y, target->point.z);
+		}
+#endif
+	};
+
+	class Face
+	{
+	public:
+		Face* next;
+		Vertex* nearbyVertex;
+		Face* nextWithSameNearbyVertex;
+		Point32 origin;
+		Point32 dir0;
+		Point32 dir1;
+
+		Face() : next(NULL), nearbyVertex(NULL), nextWithSameNearbyVertex(NULL)
+		{
+		}
+
+		void init(Vertex* a, Vertex* b, Vertex* c)
+		{
+			nearbyVertex = a;
+			origin = a->point;
+			dir0 = *b - *a;
+			dir1 = *c - *a;
+			if (a->lastNearbyFace)
+			{
+				a->lastNearbyFace->nextWithSameNearbyVertex = this;
+			}
+			else
+			{
+				a->firstNearbyFace = this;
+			}
+			a->lastNearbyFace = this;
+		}
+
+		Point64 getNormal()
+		{
+			return dir0.cross(dir1);
+		}
+	};
+
+	template <typename UWord, typename UHWord>
+	class DMul
+	{
+	private:
+		static uint32_t high(uint64_t value)
+		{
+			return (uint32_t)(value >> 32);
+		}
+
+		static uint32_t low(uint64_t value)
+		{
+			return (uint32_t)value;
+		}
+
+		static uint64_t mul(uint32_t a, uint32_t b)
+		{
+			return (uint64_t)a * (uint64_t)b;
+		}
+
+		static void shlHalf(uint64_t& value)
+		{
+			value <<= 32;
+		}
+
+		static uint64_t high(Int128 value)
+		{
+			return value.high;
+		}
+
+		static uint64_t low(Int128 value)
+		{
+			return value.low;
+		}
+
+		static Int128 mul(uint64_t a, uint64_t b)
+		{
+			return Int128::mul(a, b);
+		}
+
+		static void shlHalf(Int128& value)
+		{
+			value.high = value.low;
+			value.low = 0;
+		}
+
+	public:
+		static void mul(UWord a, UWord b, UWord& resLow, UWord& resHigh)
+		{
+			UWord p00 = mul(low(a), low(b));
+			UWord p01 = mul(low(a), high(b));
+			UWord p10 = mul(high(a), low(b));
+			UWord p11 = mul(high(a), high(b));
+			UWord p0110 = UWord(low(p01)) + UWord(low(p10));
+			p11 += high(p01);
+			p11 += high(p10);
+			p11 += high(p0110);
+			shlHalf(p0110);
+			p00 += p0110;
+			if (p00 < p0110)
+			{
+				++p11;
+			}
+			resLow = p00;
+			resHigh = p11;
+		}
+	};
+
+private:
+	class IntermediateHull
+	{
+	public:
+		Vertex* minXy;
+		Vertex* maxXy;
+		Vertex* minYx;
+		Vertex* maxYx;
+
+		IntermediateHull() : minXy(NULL), maxXy(NULL), minYx(NULL), maxYx(NULL)
+		{
+		}
+
+		void print();
+	};
+
+	enum Orientation
+	{
+		NONE,
+		CLOCKWISE,
+		COUNTER_CLOCKWISE
+	};
+
+	template <typename T>
+	class PoolArray
+	{
+	private:
+		T* array;
+		int size;
+
+	public:
+		PoolArray<T>* next;
+
+		PoolArray(int size) : size(size), next(NULL)
+		{
+			array = (T*)btAlignedAlloc(sizeof(T) * size, 16);
+		}
+
+		~PoolArray()
+		{
+			btAlignedFree(array);
+		}
+
+		T* init()
+		{
+			T* o = array;
+			for (int i = 0; i < size; i++, o++)
+			{
+				o->next = (i + 1 < size) ? o + 1 : NULL;
+			}
+			return array;
+		}
+	};
+
+	template <typename T>
+	class Pool
+	{
+	private:
+		PoolArray<T>* arrays;
+		PoolArray<T>* nextArray;
+		T* freeObjects;
+		int arraySize;
+
+	public:
+		Pool() : arrays(NULL), nextArray(NULL), freeObjects(NULL), arraySize(256)
+		{
+		}
+
+		~Pool()
+		{
+			while (arrays)
+			{
+				PoolArray<T>* p = arrays;
+				arrays = p->next;
+				p->~PoolArray<T>();
+				btAlignedFree(p);
+			}
+		}
+
+		void reset()
+		{
+			nextArray = arrays;
+			freeObjects = NULL;
+		}
+
+		void setArraySize(int arraySize)
+		{
+			this->arraySize = arraySize;
+		}
+
+		T* newObject()
+		{
+			T* o = freeObjects;
+			if (!o)
+			{
+				PoolArray<T>* p = nextArray;
+				if (p)
+				{
+					nextArray = p->next;
+				}
+				else
+				{
+					p = new (btAlignedAlloc(sizeof(PoolArray<T>), 16)) PoolArray<T>(arraySize);
+					p->next = arrays;
+					arrays = p;
+				}
+				o = p->init();
+			}
+			freeObjects = o->next;
+			return new (o) T();
+		};
+
+		void freeObject(T* object)
+		{
+			object->~T();
+			object->next = freeObjects;
+			freeObjects = object;
+		}
+	};
+
+	btVector3 scaling;
+	btVector3 center;
+	Pool<Vertex> vertexPool;
+	Pool<Edge> edgePool;
+	Pool<Face> facePool;
+	btAlignedObjectArray<Vertex*> originalVertices;
+	int mergeStamp;
+	int minAxis;
+	int medAxis;
+	int maxAxis;
+	int usedEdgePairs;
+	int maxUsedEdgePairs;
+
+	static Orientation getOrientation(const Edge* prev, const Edge* next, const Point32& s, const Point32& t);
+	Edge* findMaxAngle(bool ccw, const Vertex* start, const Point32& s, const Point64& rxs, const Point64& sxrxs, Rational64& minCot);
+	void findEdgeForCoplanarFaces(Vertex* c0, Vertex* c1, Edge*& e0, Edge*& e1, Vertex* stop0, Vertex* stop1);
+
+	Edge* newEdgePair(Vertex* from, Vertex* to);
+
+	void removeEdgePair(Edge* edge)
+	{
+		Edge* n = edge->next;
+		Edge* r = edge->reverse;
+
+		btAssert(edge->target && r->target);
+
+		if (n != edge)
+		{
+			n->prev = edge->prev;
+			edge->prev->next = n;
+			r->target->edges = n;
+		}
+		else
+		{
+			r->target->edges = NULL;
+		}
+
+		n = r->next;
+
+		if (n != r)
+		{
+			n->prev = r->prev;
+			r->prev->next = n;
+			edge->target->edges = n;
+		}
+		else
+		{
+			edge->target->edges = NULL;
+		}
+
+		edgePool.freeObject(edge);
+		edgePool.freeObject(r);
+		usedEdgePairs--;
+	}
+
+	void computeInternal(int start, int end, IntermediateHull& result);
+
+	bool mergeProjection(IntermediateHull& h0, IntermediateHull& h1, Vertex*& c0, Vertex*& c1);
+
+	void merge(IntermediateHull& h0, IntermediateHull& h1);
+
+	btVector3 toBtVector(const Point32& v);
+
+	btVector3 getBtNormal(Face* face);
+
+	bool shiftFace(Face* face, btScalar amount, btAlignedObjectArray<Vertex*> stack);
+
+public:
+	Vertex* vertexList;
+
+	void compute(const void* coords, bool doubleCoords, int stride, int count);
+
+	btVector3 getCoordinates(const Vertex* v);
+
+	btScalar shrink(btScalar amount, btScalar clampAmount);
 };
 
-
 btConvexHullInternal::Int128 btConvexHullInternal::Int128::operator*(int64_t b) const
 {
-	bool negative = (int64_t) high < 0;
+	bool negative = (int64_t)high < 0;
 	Int128 a = negative ? -*this : *this;
 	if (b < 0)
 	{
 		negative = !negative;
 		b = -b;
 	}
-	Int128 result = mul(a.low, (uint64_t) b);
-	result.high += a.high * (uint64_t) b;
+	Int128 result = mul(a.low, (uint64_t)b);
+	result.high += a.high * (uint64_t)b;
 	return negative ? -result : result;
 }
 
 btConvexHullInternal::Int128 btConvexHullInternal::Int128::mul(int64_t a, int64_t b)
 {
 	Int128 result;
-	
+
 #ifdef USE_X86_64_ASM
-	__asm__ ("imulq %[b]"
-					 : "=a" (result.low), "=d" (result.high)
-					 : "0"(a), [b] "r"(b)
-					 : "cc" );
+	__asm__("imulq %[b]"
+			: "=a"(result.low), "=d"(result.high)
+			: "0"(a), [b] "r"(b)
+			: "cc");
 	return result;
-	
+
 #else
 	bool negative = a < 0;
 	if (negative)
@@ -871,7 +872,7 @@ btConvexHullInternal::Int128 btConvexHullInternal::Int128::mul(int64_t a, int64_
 		negative = !negative;
 		b = -b;
 	}
-	DMul<uint64_t, uint32_t>::mul((uint64_t) a, (uint64_t) b, result.low, result.high);
+	DMul<uint64_t, uint32_t>::mul((uint64_t)a, (uint64_t)b, result.low, result.high);
 	return negative ? -result : result;
 #endif
 }
@@ -881,10 +882,10 @@ btConvexHullInternal::Int128 btConvexHullInternal::Int128::mul(uint64_t a, uint6
 	Int128 result;
 
 #ifdef USE_X86_64_ASM
-	__asm__ ("mulq %[b]"
-					 : "=a" (result.low), "=d" (result.high)
-					 : "0"(a), [b] "r"(b)
-					 : "cc" );
+	__asm__("mulq %[b]"
+			: "=a"(result.low), "=d"(result.high)
+			: "0"(a), [b] "r"(b)
+			: "cc");
 
 #else
 	DMul<uint64_t, uint32_t>::mul(a, b, result.low, result.high);
@@ -911,24 +912,25 @@ int btConvexHullInternal::Rational64::compare(const Rational64& b) const
 	int result;
 	int64_t tmp;
 	int64_t dummy;
-	__asm__ ("mulq %[bn]\n\t"
-					 "movq %%rax, %[tmp]\n\t"
-					 "movq %%rdx, %%rbx\n\t"
-					 "movq %[tn], %%rax\n\t"
-					 "mulq %[bd]\n\t"
-					 "subq %[tmp], %%rax\n\t"
-					 "sbbq %%rbx, %%rdx\n\t" // rdx:rax contains 128-bit-difference "numerator*b.denominator - b.numerator*denominator"
-					 "setnsb %%bh\n\t" // bh=1 if difference is non-negative, bh=0 otherwise
-					 "orq %%rdx, %%rax\n\t"
-					 "setnzb %%bl\n\t" // bl=1 if difference if non-zero, bl=0 if it is zero
-					 "decb %%bh\n\t" // now bx=0x0000 if difference is zero, 0xff01 if it is negative, 0x0001 if it is positive (i.e., same sign as difference)
-					 "shll $16, %%ebx\n\t" // ebx has same sign as difference
-					 : "=&b"(result), [tmp] "=&r"(tmp), "=a"(dummy)
-					 : "a"(denominator), [bn] "g"(b.numerator), [tn] "g"(numerator), [bd] "g"(b.denominator)
-					 : "%rdx", "cc" );
-	return result ? result ^ sign // if sign is +1, only bit 0 of result is inverted, which does not change the sign of result (and cannot result in zero)
-																// if sign is -1, all bits of result are inverted, which changes the sign of result (and again cannot result in zero)
-								: 0;
+	__asm__(
+		"mulq %[bn]\n\t"
+		"movq %%rax, %[tmp]\n\t"
+		"movq %%rdx, %%rbx\n\t"
+		"movq %[tn], %%rax\n\t"
+		"mulq %[bd]\n\t"
+		"subq %[tmp], %%rax\n\t"
+		"sbbq %%rbx, %%rdx\n\t"  // rdx:rax contains 128-bit-difference "numerator*b.denominator - b.numerator*denominator"
+		"setnsb %%bh\n\t"        // bh=1 if difference is non-negative, bh=0 otherwise
+		"orq %%rdx, %%rax\n\t"
+		"setnzb %%bl\n\t"      // bl=1 if difference if non-zero, bl=0 if it is zero
+		"decb %%bh\n\t"        // now bx=0x0000 if difference is zero, 0xff01 if it is negative, 0x0001 if it is positive (i.e., same sign as difference)
+		"shll $16, %%ebx\n\t"  // ebx has same sign as difference
+		: "=&b"(result), [tmp] "=&r"(tmp), "=a"(dummy)
+		: "a"(m_denominator), [bn] "g"(b.m_numerator), [tn] "g"(m_numerator), [bd] "g"(b.m_denominator)
+		: "%rdx", "cc");
+	return result ? result ^ sign  // if sign is +1, only bit 0 of result is inverted, which does not change the sign of result (and cannot result in zero)
+								   // if sign is -1, all bits of result are inverted, which changes the sign of result (and again cannot result in zero)
+				  : 0;
 
 #else
 
@@ -949,7 +951,7 @@ int btConvexHullInternal::Rational128::compare(const Rational128& b) const
 	}
 	if (isInt64)
 	{
-		return -b.compare(sign * (int64_t) numerator.low);
+		return -b.compare(sign * (int64_t)numerator.low);
 	}
 
 	Int128 nbdLow, nbdHigh, dbnLow, dbnHigh;
@@ -968,7 +970,7 @@ int btConvexHullInternal::Rational128::compare(int64_t b) const
 {
 	if (isInt64)
 	{
-		int64_t a = sign * (int64_t) numerator.low;
+		int64_t a = sign * (int64_t)numerator.low;
 		return (a > b) ? 1 : (a < b) ? -1 : 0;
 	}
 	if (b > 0)
@@ -994,7 +996,6 @@ int btConvexHullInternal::Rational128::compare(int64_t b) const
 	return numerator.ucmp(denominator * b) * sign;
 }
 
-
 btConvexHullInternal::Edge* btConvexHullInternal::newEdgePair(Vertex* from, Vertex* to)
 {
 	btAssert(from && to);
@@ -1062,7 +1063,7 @@ bool btConvexHullInternal::mergeProjection(IntermediateHull& h0, IntermediateHul
 			}
 		}
 	}
-	
+
 	v0 = h0.maxXy;
 	v1 = h1.maxXy;
 	Vertex* v00 = NULL;
@@ -1070,7 +1071,7 @@ bool btConvexHullInternal::mergeProjection(IntermediateHull& h0, IntermediateHul
 	int32_t sign = 1;
 
 	for (int side = 0; side <= 1; side++)
-	{		
+	{
 		int32_t dx = (v1->point.x - v0->point.x) * sign;
 		if (dx > 0)
 		{
@@ -1113,7 +1114,7 @@ bool btConvexHullInternal::mergeProjection(IntermediateHull& h0, IntermediateHul
 			while (true)
 			{
 				int32_t dy = v1->point.y - v0->point.y;
-				
+
 				Vertex* w1 = side ? v1->prev : v1->next;
 				if (w1 != v1)
 				{
@@ -1126,7 +1127,7 @@ bool btConvexHullInternal::mergeProjection(IntermediateHull& h0, IntermediateHul
 						continue;
 					}
 				}
-				
+
 				Vertex* w0 = side ? v0->prev : v0->next;
 				if (w0 != v0)
 				{
@@ -1140,7 +1141,7 @@ bool btConvexHullInternal::mergeProjection(IntermediateHull& h0, IntermediateHul
 						continue;
 					}
 				}
-				
+
 				break;
 			}
 		}
@@ -1166,7 +1167,7 @@ bool btConvexHullInternal::mergeProjection(IntermediateHull& h0, IntermediateHul
 			}
 			v1 = w1;
 		}
-		
+
 		if (side == 0)
 		{
 			v00 = v0;
@@ -1192,7 +1193,7 @@ bool btConvexHullInternal::mergeProjection(IntermediateHull& h0, IntermediateHul
 	{
 		h0.maxXy = h1.maxXy;
 	}
-	
+
 	h0.maxYx = h1.maxYx;
 
 	c0 = v00;
@@ -1279,19 +1280,19 @@ void btConvexHullInternal::computeInternal(int start, int end, IntermediateHull&
 			}
 			{
 				Vertex* v = originalVertices[start];
-							v->edges = NULL;
-							v->next = v;
-							v->prev = v;
+				v->edges = NULL;
+				v->next = v;
+				v->prev = v;
 
-							result.minXy = v;
-							result.maxXy = v;
-							result.minYx = v;
-							result.maxYx = v;
+				result.minXy = v;
+				result.maxXy = v;
+				result.minYx = v;
+				result.maxYx = v;
 			}
-			
+
 			return;
 		}
-		
+
 		case 1:
 		{
 			Vertex* v = originalVertices[start];
@@ -1309,7 +1310,7 @@ void btConvexHullInternal::computeInternal(int start, int end, IntermediateHull&
 	}
 
 	int split0 = start + n / 2;
-	Point32 p = originalVertices[split0-1]->point;
+	Point32 p = originalVertices[split0 - 1]->point;
 	int split1 = split0;
 	while ((split1 < end) && (originalVertices[split1]->point == p))
 	{
@@ -1334,7 +1335,7 @@ void btConvexHullInternal::computeInternal(int start, int end, IntermediateHull&
 void btConvexHullInternal::IntermediateHull::print()
 {
 	printf("    Hull\n");
-	for (Vertex* v = minXy; v; )
+	for (Vertex* v = minXy; v;)
 	{
 		printf("      ");
 		v->print();
@@ -1362,7 +1363,7 @@ void btConvexHullInternal::IntermediateHull::print()
 		}
 	}
 	if (minXy)
-	{		
+	{
 		minXy->copy = (minXy->copy == -1) ? -2 : -1;
 		minXy->printGraph();
 	}
@@ -1438,7 +1439,7 @@ btConvexHullInternal::Edge* btConvexHullInternal::findMaxAngle(bool ccw, const V
 				Point32 t = *e->target - *start;
 				Rational64 cot(t.dot(sxrxs), t.dot(rxs));
 #ifdef DEBUG_CONVEX_HULL
-				printf("      Angle is %f (%d) for ", (float) btAtan(cot.toScalar()), (int) cot.isNaN());
+				printf("      Angle is %f (%d) for ", (float)btAtan(cot.toScalar()), (int)cot.isNaN());
 				e->print();
 #endif
 				if (cot.isNaN())
@@ -1485,7 +1486,7 @@ void btConvexHullInternal::findEdgeForCoplanarFaces(Vertex* c0, Vertex* c1, Edge
 	btAssert(!start1 || (start1->target->point.dot(normal) == dist));
 	Point64 perp = s.cross(normal);
 	btAssert(!perp.isZero());
-	
+
 #ifdef DEBUG_CONVEX_HULL
 	printf("   Advancing %d %d  (%p %p, %d %d)\n", c0->point.index, c1->point.index, start0, start1, start0 ? start0->target->point.index : -1, start1 ? start1->target->point.index : -1);
 #endif
@@ -1515,7 +1516,7 @@ void btConvexHullInternal::findEdgeForCoplanarFaces(Vertex* c0, Vertex* c1, Edge
 			et0 = e->target->point;
 		}
 	}
-	
+
 	int64_t maxDot1 = et1.dot(perp);
 	if (e1)
 	{
@@ -1552,7 +1553,7 @@ void btConvexHullInternal::findEdgeForCoplanarFaces(Vertex* c0, Vertex* c1, Edge
 		while (true)
 		{
 			int64_t dy = (et1 - et0).dot(s);
-			
+
 			if (e0 && (e0->target != stop0))
 			{
 				Edge* f0 = e0->next->reverse;
@@ -1569,7 +1570,7 @@ void btConvexHullInternal::findEdgeForCoplanarFaces(Vertex* c0, Vertex* c1, Edge
 					}
 				}
 			}
-			
+
 			if (e1 && (e1->target != stop1))
 			{
 				Edge* f1 = e1->reverse->next;
@@ -1604,7 +1605,7 @@ void btConvexHullInternal::findEdgeForCoplanarFaces(Vertex* c0, Vertex* c1, Edge
 		while (true)
 		{
 			int64_t dy = (et1 - et0).dot(s);
-			
+
 			if (e1 && (e1->target != stop1))
 			{
 				Edge* f1 = e1->prev->reverse;
@@ -1621,7 +1622,7 @@ void btConvexHullInternal::findEdgeForCoplanarFaces(Vertex* c0, Vertex* c1, Edge
 					}
 				}
 			}
-			
+
 			if (e0 && (e0->target != stop0))
 			{
 				Edge* f0 = e0->reverse->prev;
@@ -1656,7 +1657,6 @@ void btConvexHullInternal::findEdgeForCoplanarFaces(Vertex* c0, Vertex* c1, Edge
 #endif
 }
 
-
 void btConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 {
 	if (!h1.maxXy)
@@ -1668,7 +1668,7 @@ void btConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 		h0 = h1;
 		return;
 	}
-	
+
 	mergeStamp--;
 
 	Vertex* c0 = NULL;
@@ -1708,7 +1708,7 @@ void btConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 				e = e->next;
 			} while (e != c0->edges);
 		}
-		
+
 		e = c1->edges;
 		Edge* start1 = NULL;
 		if (e)
@@ -1760,7 +1760,7 @@ void btConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 		Point32 r = prevPoint - c0->point;
 		Point64 rxs = r.cross(s);
 		Point64 sxrxs = s.cross(rxs);
-		
+
 #ifdef DEBUG_CONVEX_HULL
 		printf("\n  Checking %d %d\n", c0->point.index, c1->point.index);
 #endif
@@ -1811,7 +1811,7 @@ void btConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 				e->prev = pendingTail1;
 				pendingTail1 = e;
 			}
-			
+
 			Edge* e0 = min0;
 			Edge* e1 = min1;
 
@@ -1828,7 +1828,7 @@ void btConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 			{
 				if (toPrev1)
 				{
-					for (Edge* e = toPrev1->next, *n = NULL; e != min1; e = n)
+					for (Edge *e = toPrev1->next, *n = NULL; e != min1; e = n)
 					{
 						n = e->next;
 						removeEdgePair(e);
@@ -1864,7 +1864,7 @@ void btConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 			{
 				if (toPrev0)
 				{
-					for (Edge* e = toPrev0->prev, *n = NULL; e != min0; e = n)
+					for (Edge *e = toPrev0->prev, *n = NULL; e != min0; e = n)
 					{
 						n = e->prev;
 						removeEdgePair(e);
@@ -1906,7 +1906,7 @@ void btConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 			}
 			else
 			{
-				for (Edge* e = toPrev0->prev, *n = NULL; e != firstNew0; e = n)
+				for (Edge *e = toPrev0->prev, *n = NULL; e != firstNew0; e = n)
 				{
 					n = e->prev;
 					removeEdgePair(e);
@@ -1925,7 +1925,7 @@ void btConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 			}
 			else
 			{
-				for (Edge* e = toPrev1->next, *n = NULL; e != firstNew1; e = n)
+				for (Edge *e = toPrev1->next, *n = NULL; e != firstNew1; e = n)
 				{
 					n = e->next;
 					removeEdgePair(e);
@@ -1936,7 +1936,7 @@ void btConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 					pendingTail1->link(firstNew1);
 				}
 			}
-			
+
 			return;
 		}
 
@@ -1946,24 +1946,23 @@ void btConvexHullInternal::merge(IntermediateHull& h0, IntermediateHull& h1)
 
 class pointCmp
 {
-	public:
-
-    bool operator() ( const btConvexHullInternal::Point32& p, const btConvexHullInternal::Point32& q ) const
-		{
-			return (p.y < q.y) || ((p.y == q.y) && ((p.x < q.x) || ((p.x == q.x) && (p.z < q.z))));
-		}
+public:
+	bool operator()(const btConvexHullInternal::Point32& p, const btConvexHullInternal::Point32& q) const
+	{
+		return (p.y < q.y) || ((p.y == q.y) && ((p.x < q.x) || ((p.x == q.x) && (p.z < q.z))));
+	}
 };
 
 void btConvexHullInternal::compute(const void* coords, bool doubleCoords, int stride, int count)
 {
 	btVector3 min(btScalar(1e30), btScalar(1e30), btScalar(1e30)), max(btScalar(-1e30), btScalar(-1e30), btScalar(-1e30));
-	const char* ptr = (const char*) coords;
+	const char* ptr = (const char*)coords;
 	if (doubleCoords)
 	{
 		for (int i = 0; i < count; i++)
 		{
-			const double* v = (const double*) ptr;
-			btVector3 p((btScalar) v[0], (btScalar) v[1], (btScalar) v[2]);
+			const double* v = (const double*)ptr;
+			btVector3 p((btScalar)v[0], (btScalar)v[1], (btScalar)v[2]);
 			ptr += stride;
 			min.setMin(p);
 			max.setMax(p);
@@ -1973,7 +1972,7 @@ void btConvexHullInternal::compute(const void* coords, bool doubleCoords, int st
 	{
 		for (int i = 0; i < count; i++)
 		{
-			const float* v = (const float*) ptr;
+			const float* v = (const float*)ptr;
 			btVector3 p(v[0], v[1], v[2]);
 			ptr += stride;
 			min.setMin(p);
@@ -2014,18 +2013,18 @@ void btConvexHullInternal::compute(const void* coords, bool doubleCoords, int st
 
 	btAlignedObjectArray<Point32> points;
 	points.resize(count);
-	ptr = (const char*) coords;
+	ptr = (const char*)coords;
 	if (doubleCoords)
 	{
 		for (int i = 0; i < count; i++)
 		{
-			const double* v = (const double*) ptr;
-			btVector3 p((btScalar) v[0], (btScalar) v[1], (btScalar) v[2]);
+			const double* v = (const double*)ptr;
+			btVector3 p((btScalar)v[0], (btScalar)v[1], (btScalar)v[2]);
 			ptr += stride;
 			p = (p - center) * s;
-			points[i].x = (int32_t) p[medAxis];
-			points[i].y = (int32_t) p[maxAxis];
-			points[i].z = (int32_t) p[minAxis];
+			points[i].x = (int32_t)p[medAxis];
+			points[i].y = (int32_t)p[maxAxis];
+			points[i].z = (int32_t)p[minAxis];
 			points[i].index = i;
 		}
 	}
@@ -2033,13 +2032,13 @@ void btConvexHullInternal::compute(const void* coords, bool doubleCoords, int st
 	{
 		for (int i = 0; i < count; i++)
 		{
-			const float* v = (const float*) ptr;
+			const float* v = (const float*)ptr;
 			btVector3 p(v[0], v[1], v[2]);
 			ptr += stride;
 			p = (p - center) * s;
-			points[i].x = (int32_t) p[medAxis];
-			points[i].y = (int32_t) p[maxAxis];
-			points[i].z = (int32_t) p[minAxis];
+			points[i].x = (int32_t)p[medAxis];
+			points[i].y = (int32_t)p[maxAxis];
+			points[i].z = (int32_t)p[minAxis];
 			points[i].index = i;
 		}
 	}
@@ -2193,7 +2192,7 @@ btScalar btConvexHullInternal::shrink(btScalar amount, btScalar clampAmount)
 				minDist = dist;
 			}
 		}
-		
+
 		if (minDist <= 0)
 		{
 			return 0;
@@ -2234,7 +2233,7 @@ bool btConvexHullInternal::shiftFace(Face* face, btScalar amount, btAlignedObjec
 	{
 		origShift[2] /= scaling[2];
 	}
-	Point32 shift((int32_t) origShift[medAxis], (int32_t) origShift[maxAxis], (int32_t) origShift[minAxis]);
+	Point32 shift((int32_t)origShift[medAxis], (int32_t)origShift[maxAxis], (int32_t)origShift[minAxis]);
 	if (shift.isZero())
 	{
 		return true;
@@ -2242,7 +2241,7 @@ bool btConvexHullInternal::shiftFace(Face* face, btScalar amount, btAlignedObjec
 	Point64 normal = face->getNormal();
 #ifdef DEBUG_CONVEX_HULL
 	printf("\nShrinking face (%d %d %d) (%d %d %d) (%d %d %d) by (%d %d %d)\n",
-				 face->origin.x, face->origin.y, face->origin.z, face->dir0.x, face->dir0.y, face->dir0.z, face->dir1.x, face->dir1.y, face->dir1.z, shift.x, shift.y, shift.z);
+		   face->origin.x, face->origin.y, face->origin.z, face->dir0.x, face->dir0.y, face->dir0.z, face->dir1.x, face->dir1.y, face->dir1.z, shift.x, shift.y, shift.z);
 #endif
 	int64_t origDot = face->origin.dot(normal);
 	Point32 shiftedOrigin = face->origin + shift;
@@ -2279,7 +2278,7 @@ bool btConvexHullInternal::shiftFace(Face* face, btScalar amount, btAlignedObjec
 #ifdef DEBUG_CONVEX_HULL
 			printf("Moving downwards, edge is ");
 			e->print();
-			printf(", dot is %f (%f %lld)\n", (float) dot.toScalar(), (float) optDot.toScalar(), shiftedDot);
+			printf(", dot is %f (%f %lld)\n", (float)dot.toScalar(), (float)optDot.toScalar(), shiftedDot);
 #endif
 			if (dot.compare(optDot) < 0)
 			{
@@ -2315,7 +2314,7 @@ bool btConvexHullInternal::shiftFace(Face* face, btScalar amount, btAlignedObjec
 #ifdef DEBUG_CONVEX_HULL
 			printf("Moving upwards, edge is ");
 			e->print();
-			printf(", dot is %f (%f %lld)\n", (float) dot.toScalar(), (float) optDot.toScalar(), shiftedDot);
+			printf(", dot is %f (%f %lld)\n", (float)dot.toScalar(), (float)optDot.toScalar(), shiftedDot);
 #endif
 			if (dot.compare(optDot) > 0)
 			{
@@ -2331,7 +2330,7 @@ bool btConvexHullInternal::shiftFace(Face* face, btScalar amount, btAlignedObjec
 			}
 			e = e->prev;
 		} while (e != startEdge);
-		
+
 		if (!intersection)
 		{
 			return true;
@@ -2368,7 +2367,7 @@ bool btConvexHullInternal::shiftFace(Face* face, btScalar amount, btAlignedObjec
 		printf("Needed %d iterations to check for complete containment\n", n);
 #endif
 	}
-	
+
 	Edge* firstIntersection = NULL;
 	Edge* faceEdge = NULL;
 	Edge* firstFaceEdge = NULL;
@@ -2477,7 +2476,7 @@ bool btConvexHullInternal::shiftFace(Face* face, btScalar amount, btAlignedObjec
 #ifdef DEBUG_CONVEX_HULL
 			printf("1: Removed part contains (%d %d %d)\n", removed->point.x, removed->point.y, removed->point.z);
 #endif
-			
+
 			Point64 n0 = intersection->face->getNormal();
 			Point64 n1 = intersection->reverse->face->getNormal();
 			int64_t m00 = face->dir0.dot(n0);
@@ -2491,16 +2490,13 @@ bool btConvexHullInternal::shiftFace(Face* face, btScalar amount, btAlignedObjec
 			Vertex* v = vertexPool.newObject();
 			v->point.index = -1;
 			v->copy = -1;
-			v->point128 = PointR128(Int128::mul(face->dir0.x * r0, m11) - Int128::mul(face->dir0.x * r1, m01)
-															+ Int128::mul(face->dir1.x * r1, m00) - Int128::mul(face->dir1.x * r0, m10) + det * shiftedOrigin.x,
-															Int128::mul(face->dir0.y * r0, m11) - Int128::mul(face->dir0.y * r1, m01)
-															+ Int128::mul(face->dir1.y * r1, m00) - Int128::mul(face->dir1.y * r0, m10) + det * shiftedOrigin.y,
-															Int128::mul(face->dir0.z * r0, m11) - Int128::mul(face->dir0.z * r1, m01)
-															+ Int128::mul(face->dir1.z * r1, m00) - Int128::mul(face->dir1.z * r0, m10) + det * shiftedOrigin.z,
-															det);
-			v->point.x = (int32_t) v->point128.xvalue();
-			v->point.y = (int32_t) v->point128.yvalue();
-			v->point.z = (int32_t) v->point128.zvalue();
+			v->point128 = PointR128(Int128::mul(face->dir0.x * r0, m11) - Int128::mul(face->dir0.x * r1, m01) + Int128::mul(face->dir1.x * r1, m00) - Int128::mul(face->dir1.x * r0, m10) + det * shiftedOrigin.x,
+									Int128::mul(face->dir0.y * r0, m11) - Int128::mul(face->dir0.y * r1, m01) + Int128::mul(face->dir1.y * r1, m00) - Int128::mul(face->dir1.y * r0, m10) + det * shiftedOrigin.y,
+									Int128::mul(face->dir0.z * r0, m11) - Int128::mul(face->dir0.z * r1, m01) + Int128::mul(face->dir1.z * r1, m00) - Int128::mul(face->dir1.z * r0, m10) + det * shiftedOrigin.z,
+									det);
+			v->point.x = (int32_t)v->point128.xvalue();
+			v->point.y = (int32_t)v->point128.yvalue();
+			v->point.z = (int32_t)v->point128.zvalue();
 			intersection->target = v;
 			v->edges = e;
 
@@ -2639,7 +2635,6 @@ bool btConvexHullInternal::shiftFace(Face* face, btScalar amount, btAlignedObjec
 	return true;
 }
 
-
 static int getVertexCopy(btConvexHullInternal::Vertex* vertex, btAlignedObjectArray<btConvexHullInternal::Vertex*>& vertices)
 {
 	int index = vertex->copy;
@@ -2761,8 +2756,3 @@ btScalar btConvexHullComputer::compute(const void* coords, bool doubleCoords, in
 
 	return shift;
 }
-
-
-
-
-
diff --git a/thirdparty/bullet/LinearMath/btConvexHullComputer.h b/thirdparty/bullet/LinearMath/btConvexHullComputer.h
index 7240ac4fb52..cba684f2dc7 100644
--- a/thirdparty/bullet/LinearMath/btConvexHullComputer.h
+++ b/thirdparty/bullet/LinearMath/btConvexHullComputer.h
@@ -23,58 +23,56 @@ subject to the following restrictions:
 /// Ole Kniemeyer, MAXON Computer GmbH
 class btConvexHullComputer
 {
+private:
+	btScalar compute(const void* coords, bool doubleCoords, int stride, int count, btScalar shrink, btScalar shrinkClamp);
+
+public:
+	class Edge
+	{
 	private:
-		btScalar compute(const void* coords, bool doubleCoords, int stride, int count, btScalar shrink, btScalar shrinkClamp);
+		int next;
+		int reverse;
+		int targetVertex;
+
+		friend class btConvexHullComputer;
 
 	public:
-
-		class Edge
+		int getSourceVertex() const
 		{
-			private:
-				int next;
-				int reverse;
-				int targetVertex;
+			return (this + reverse)->targetVertex;
+		}
 
-				friend class btConvexHullComputer;
+		int getTargetVertex() const
+		{
+			return targetVertex;
+		}
 
-			public:
-				int getSourceVertex() const
-				{
-					return (this + reverse)->targetVertex;
-				}
+		const Edge* getNextEdgeOfVertex() const  // clockwise list of all edges of a vertex
+		{
+			return this + next;
+		}
 
-				int getTargetVertex() const
-				{
-					return targetVertex;
-				}
+		const Edge* getNextEdgeOfFace() const  // counter-clockwise list of all edges of a face
+		{
+			return (this + reverse)->getNextEdgeOfVertex();
+		}
 
-				const Edge* getNextEdgeOfVertex() const // clockwise list of all edges of a vertex
-				{
-					return this + next;
-				}
+		const Edge* getReverseEdge() const
+		{
+			return this + reverse;
+		}
+	};
 
-				const Edge* getNextEdgeOfFace() const // counter-clockwise list of all edges of a face
-				{
-					return (this + reverse)->getNextEdgeOfVertex();
-				}
+	// Vertices of the output hull
+	btAlignedObjectArray<btVector3> vertices;
 
-				const Edge* getReverseEdge() const
-				{
-					return this + reverse;
-				}
-		};
+	// Edges of the output hull
+	btAlignedObjectArray<Edge> edges;
 
+	// Faces of the convex hull. Each entry is an index into the "edges" array pointing to an edge of the face. Faces are planar n-gons
+	btAlignedObjectArray<int> faces;
 
-		// Vertices of the output hull
-		btAlignedObjectArray<btVector3> vertices;
-
-		// Edges of the output hull
-		btAlignedObjectArray<Edge> edges;
-
-		// Faces of the convex hull. Each entry is an index into the "edges" array pointing to an edge of the face. Faces are planar n-gons
-		btAlignedObjectArray<int> faces;
-
-		/*
+	/*
 		Compute convex hull of "count" vertices stored in "coords". "stride" is the difference in bytes
 		between the addresses of consecutive vertices. If "shrink" is positive, the convex hull is shrunken
 		by that amount (each face is moved by "shrink" length units towards the center along its normal).
@@ -86,18 +84,16 @@ class btConvexHullComputer
 
 		The output convex hull can be found in the member variables "vertices", "edges", "faces".
 		*/
-		btScalar compute(const float* coords, int stride, int count, btScalar shrink, btScalar shrinkClamp)
-		{
-			return compute(coords, false, stride, count, shrink, shrinkClamp);
-		}
+	btScalar compute(const float* coords, int stride, int count, btScalar shrink, btScalar shrinkClamp)
+	{
+		return compute(coords, false, stride, count, shrink, shrinkClamp);
+	}
 
-		// same as above, but double precision
-		btScalar compute(const double* coords, int stride, int count, btScalar shrink, btScalar shrinkClamp)
-		{
-			return compute(coords, true, stride, count, shrink, shrinkClamp);
-		}
+	// same as above, but double precision
+	btScalar compute(const double* coords, int stride, int count, btScalar shrink, btScalar shrinkClamp)
+	{
+		return compute(coords, true, stride, count, shrink, shrinkClamp);
+	}
 };
 
-
-#endif //BT_CONVEX_HULL_COMPUTER_H
-
+#endif  //BT_CONVEX_HULL_COMPUTER_H
diff --git a/thirdparty/bullet/LinearMath/btCpuFeatureUtility.h b/thirdparty/bullet/LinearMath/btCpuFeatureUtility.h
index d2cab52d488..5e4b9a313c7 100644
--- a/thirdparty/bullet/LinearMath/btCpuFeatureUtility.h
+++ b/thirdparty/bullet/LinearMath/btCpuFeatureUtility.h
@@ -4,20 +4,20 @@
 
 #include "LinearMath/btScalar.h"
 
-#include <string.h>//memset
-#ifdef  USE_SIMD
+#include <string.h>  //memset
+#ifdef USE_SIMD
 #include <emmintrin.h>
 #ifdef BT_ALLOW_SSE4
 #include <intrin.h>
-#endif //BT_ALLOW_SSE4
-#endif //USE_SIMD
+#endif  //BT_ALLOW_SSE4
+#endif  //USE_SIMD
 
 #if defined BT_USE_NEON
-#define ARM_NEON_GCC_COMPATIBILITY  1
+#define ARM_NEON_GCC_COMPATIBILITY 1
 #include <arm_neon.h>
 #include <sys/types.h>
-#include <sys/sysctl.h> //for sysctlbyname
-#endif //BT_USE_NEON
+#include <sys/sysctl.h>  //for sysctlbyname
+#endif                   //BT_USE_NEON
 
 ///Rudimentary btCpuFeatureUtility for CPU features: only report the features that Bullet actually uses (SSE4/FMA3, NEON_HPFP)
 ///We assume SSE2 in case BT_USE_SSE2 is defined in LinearMath/btScalar.h
@@ -26,14 +26,13 @@ class btCpuFeatureUtility
 public:
 	enum btCpuFeature
 	{
-		CPU_FEATURE_FMA3=1,
-		CPU_FEATURE_SSE4_1=2,
-		CPU_FEATURE_NEON_HPFP=4
+		CPU_FEATURE_FMA3 = 1,
+		CPU_FEATURE_SSE4_1 = 2,
+		CPU_FEATURE_NEON_HPFP = 4
 	};
 
 	static int getCpuFeatures()
 	{
-
 		static int capabilities = 0;
 		static bool testedCapabilities = false;
 		if (0 != testedCapabilities)
@@ -49,15 +48,15 @@ public:
 			if (0 == err && hasFeature)
 				capabilities |= CPU_FEATURE_NEON_HPFP;
 		}
-#endif //BT_USE_NEON
+#endif  //BT_USE_NEON
 
-#ifdef  BT_ALLOW_SSE4
+#ifdef BT_ALLOW_SSE4
 		{
-			int					cpuInfo[4];
+			int cpuInfo[4];
 			memset(cpuInfo, 0, sizeof(cpuInfo));
-			unsigned long long	sseExt = 0;
+			unsigned long long sseExt = 0;
 			__cpuid(cpuInfo, 1);
-			
+
 			bool osUsesXSAVE_XRSTORE = cpuInfo[2] & (1 << 27) || false;
 			bool cpuAVXSuport = cpuInfo[2] & (1 << 28) || false;
 
@@ -79,14 +78,11 @@ public:
 				capabilities |= btCpuFeatureUtility::CPU_FEATURE_SSE4_1;
 			}
 		}
-#endif//BT_ALLOW_SSE4
+#endif  //BT_ALLOW_SSE4
 
 		testedCapabilities = true;
 		return capabilities;
 	}
-
-
 };
 
-
-#endif //BT_CPU_UTILITY_H
+#endif  //BT_CPU_UTILITY_H
diff --git a/thirdparty/bullet/LinearMath/btDefaultMotionState.h b/thirdparty/bullet/LinearMath/btDefaultMotionState.h
index 01c5f8d932d..14c40d36b07 100644
--- a/thirdparty/bullet/LinearMath/btDefaultMotionState.h
+++ b/thirdparty/bullet/LinearMath/btDefaultMotionState.h
@@ -4,39 +4,37 @@
 #include "btMotionState.h"
 
 ///The btDefaultMotionState provides a common implementation to synchronize world transforms with offsets.
-ATTRIBUTE_ALIGNED16(struct)	btDefaultMotionState : public btMotionState
+ATTRIBUTE_ALIGNED16(struct)
+btDefaultMotionState : public btMotionState
 {
 	btTransform m_graphicsWorldTrans;
-	btTransform	m_centerOfMassOffset;
+	btTransform m_centerOfMassOffset;
 	btTransform m_startWorldTrans;
-	void*		m_userPointer;
+	void* m_userPointer;
 
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-	btDefaultMotionState(const btTransform& startTrans = btTransform::getIdentity(),const btTransform& centerOfMassOffset = btTransform::getIdentity())
+	btDefaultMotionState(const btTransform& startTrans = btTransform::getIdentity(), const btTransform& centerOfMassOffset = btTransform::getIdentity())
 		: m_graphicsWorldTrans(startTrans),
-		m_centerOfMassOffset(centerOfMassOffset),
-		m_startWorldTrans(startTrans),
-		m_userPointer(0)
+		  m_centerOfMassOffset(centerOfMassOffset),
+		  m_startWorldTrans(startTrans),
+		  m_userPointer(0)
 
 	{
 	}
 
 	///synchronizes world transform from user to physics
-	virtual void	getWorldTransform(btTransform& centerOfMassWorldTrans ) const 
+	virtual void getWorldTransform(btTransform & centerOfMassWorldTrans) const
 	{
-			centerOfMassWorldTrans = m_graphicsWorldTrans * m_centerOfMassOffset.inverse() ;
+		centerOfMassWorldTrans = m_graphicsWorldTrans * m_centerOfMassOffset.inverse();
 	}
 
 	///synchronizes world transform from physics to user
 	///Bullet only calls the update of worldtransform for active objects
-	virtual void	setWorldTransform(const btTransform& centerOfMassWorldTrans)
+	virtual void setWorldTransform(const btTransform& centerOfMassWorldTrans)
 	{
-			m_graphicsWorldTrans = centerOfMassWorldTrans * m_centerOfMassOffset;
+		m_graphicsWorldTrans = centerOfMassWorldTrans * m_centerOfMassOffset;
 	}
-
-	
-
 };
 
-#endif //BT_DEFAULT_MOTION_STATE_H
+#endif  //BT_DEFAULT_MOTION_STATE_H
diff --git a/thirdparty/bullet/LinearMath/btGeometryUtil.cpp b/thirdparty/bullet/LinearMath/btGeometryUtil.cpp
index 5ac230f712f..115e3eab810 100644
--- a/thirdparty/bullet/LinearMath/btGeometryUtil.cpp
+++ b/thirdparty/bullet/LinearMath/btGeometryUtil.cpp
@@ -12,49 +12,43 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #include "btGeometryUtil.h"
 
-
 /*
   Make sure this dummy function never changes so that it
   can be used by probes that are checking whether the
   library is actually installed.
 */
 extern "C"
-{	
-	void btBulletMathProbe ();
+{
+	void btBulletMathProbe();
 
-	void btBulletMathProbe () {}
+	void btBulletMathProbe() {}
 }
 
-
-bool	btGeometryUtil::isPointInsidePlanes(const btAlignedObjectArray<btVector3>& planeEquations, const btVector3& point, btScalar	margin)
+bool btGeometryUtil::isPointInsidePlanes(const btAlignedObjectArray<btVector3>& planeEquations, const btVector3& point, btScalar margin)
 {
 	int numbrushes = planeEquations.size();
-	for (int i=0;i<numbrushes;i++)
+	for (int i = 0; i < numbrushes; i++)
 	{
 		const btVector3& N1 = planeEquations[i];
-		btScalar dist = btScalar(N1.dot(point))+btScalar(N1[3])-margin;
-		if (dist>btScalar(0.))
+		btScalar dist = btScalar(N1.dot(point)) + btScalar(N1[3]) - margin;
+		if (dist > btScalar(0.))
 		{
 			return false;
 		}
 	}
 	return true;
-		
 }
 
-
-bool	btGeometryUtil::areVerticesBehindPlane(const btVector3& planeNormal, const btAlignedObjectArray<btVector3>& vertices, btScalar	margin)
+bool btGeometryUtil::areVerticesBehindPlane(const btVector3& planeNormal, const btAlignedObjectArray<btVector3>& vertices, btScalar margin)
 {
 	int numvertices = vertices.size();
-	for (int i=0;i<numvertices;i++)
+	for (int i = 0; i < numvertices; i++)
 	{
 		const btVector3& N1 = vertices[i];
-		btScalar dist = btScalar(planeNormal.dot(N1))+btScalar(planeNormal[3])-margin;
-		if (dist>btScalar(0.))
+		btScalar dist = btScalar(planeNormal.dot(N1)) + btScalar(planeNormal[3]) - margin;
+		if (dist > btScalar(0.))
 		{
 			return false;
 		}
@@ -62,102 +56,98 @@ bool	btGeometryUtil::areVerticesBehindPlane(const btVector3& planeNormal, const
 	return true;
 }
 
-bool notExist(const btVector3& planeEquation,const btAlignedObjectArray<btVector3>& planeEquations);
+bool notExist(const btVector3& planeEquation, const btAlignedObjectArray<btVector3>& planeEquations);
 
-bool notExist(const btVector3& planeEquation,const btAlignedObjectArray<btVector3>& planeEquations)
+bool notExist(const btVector3& planeEquation, const btAlignedObjectArray<btVector3>& planeEquations)
 {
 	int numbrushes = planeEquations.size();
-	for (int i=0;i<numbrushes;i++)
+	for (int i = 0; i < numbrushes; i++)
 	{
 		const btVector3& N1 = planeEquations[i];
 		if (planeEquation.dot(N1) > btScalar(0.999))
 		{
 			return false;
-		} 
+		}
 	}
 	return true;
 }
 
-void	btGeometryUtil::getPlaneEquationsFromVertices(btAlignedObjectArray<btVector3>& vertices, btAlignedObjectArray<btVector3>& planeEquationsOut )
+void btGeometryUtil::getPlaneEquationsFromVertices(btAlignedObjectArray<btVector3>& vertices, btAlignedObjectArray<btVector3>& planeEquationsOut)
 {
-		const int numvertices = vertices.size();
+	const int numvertices = vertices.size();
 	// brute force:
-	for (int i=0;i<numvertices;i++)
+	for (int i = 0; i < numvertices; i++)
 	{
 		const btVector3& N1 = vertices[i];
-		
 
-		for (int j=i+1;j<numvertices;j++)
+		for (int j = i + 1; j < numvertices; j++)
 		{
 			const btVector3& N2 = vertices[j];
-				
-			for (int k=j+1;k<numvertices;k++)
-			{
 
+			for (int k = j + 1; k < numvertices; k++)
+			{
 				const btVector3& N3 = vertices[k];
 
-				btVector3 planeEquation,edge0,edge1;
-				edge0 = N2-N1;
-				edge1 = N3-N1;
+				btVector3 planeEquation, edge0, edge1;
+				edge0 = N2 - N1;
+				edge1 = N3 - N1;
 				btScalar normalSign = btScalar(1.);
-				for (int ww=0;ww<2;ww++)
+				for (int ww = 0; ww < 2; ww++)
 				{
 					planeEquation = normalSign * edge0.cross(edge1);
 					if (planeEquation.length2() > btScalar(0.0001))
 					{
 						planeEquation.normalize();
-						if (notExist(planeEquation,planeEquationsOut))
+						if (notExist(planeEquation, planeEquationsOut))
 						{
 							planeEquation[3] = -planeEquation.dot(N1);
-							
-								//check if inside, and replace supportingVertexOut if needed
-								if (areVerticesBehindPlane(planeEquation,vertices,btScalar(0.01)))
-								{
-									planeEquationsOut.push_back(planeEquation);
-								}
+
+							//check if inside, and replace supportingVertexOut if needed
+							if (areVerticesBehindPlane(planeEquation, vertices, btScalar(0.01)))
+							{
+								planeEquationsOut.push_back(planeEquation);
+							}
 						}
 					}
 					normalSign = btScalar(-1.);
 				}
-			
 			}
 		}
 	}
-
 }
 
-void	btGeometryUtil::getVerticesFromPlaneEquations(const btAlignedObjectArray<btVector3>& planeEquations , btAlignedObjectArray<btVector3>& verticesOut )
+void btGeometryUtil::getVerticesFromPlaneEquations(const btAlignedObjectArray<btVector3>& planeEquations, btAlignedObjectArray<btVector3>& verticesOut)
 {
 	const int numbrushes = planeEquations.size();
 	// brute force:
-	for (int i=0;i<numbrushes;i++)
+	for (int i = 0; i < numbrushes; i++)
 	{
 		const btVector3& N1 = planeEquations[i];
-		
 
-		for (int j=i+1;j<numbrushes;j++)
+		for (int j = i + 1; j < numbrushes; j++)
 		{
 			const btVector3& N2 = planeEquations[j];
-				
-			for (int k=j+1;k<numbrushes;k++)
-			{
 
+			for (int k = j + 1; k < numbrushes; k++)
+			{
 				const btVector3& N3 = planeEquations[k];
 
-				btVector3 n2n3; n2n3 = N2.cross(N3);
-				btVector3 n3n1; n3n1 = N3.cross(N1);
-				btVector3 n1n2; n1n2 = N1.cross(N2);
-				
-				if ( ( n2n3.length2() > btScalar(0.0001) ) &&
-					 ( n3n1.length2() > btScalar(0.0001) ) &&
-					 ( n1n2.length2() > btScalar(0.0001) ) )
+				btVector3 n2n3;
+				n2n3 = N2.cross(N3);
+				btVector3 n3n1;
+				n3n1 = N3.cross(N1);
+				btVector3 n1n2;
+				n1n2 = N1.cross(N2);
+
+				if ((n2n3.length2() > btScalar(0.0001)) &&
+					(n3n1.length2() > btScalar(0.0001)) &&
+					(n1n2.length2() > btScalar(0.0001)))
 				{
 					//point P out of 3 plane equations:
 
-					//	d1 ( N2 * N3 ) + d2 ( N3 * N1 ) + d3 ( N1 * N2 )  
-					//P =  -------------------------------------------------------------------------  
-					//   N1 . ( N2 * N3 )  
-
+					//	d1 ( N2 * N3 ) + d2 ( N3 * N1 ) + d3 ( N1 * N2 )
+					//P =  -------------------------------------------------------------------------
+					//   N1 . ( N2 * N3 )
 
 					btScalar quotient = (N1.dot(n2n3));
 					if (btFabs(quotient) > btScalar(0.000001))
@@ -172,7 +162,7 @@ void	btGeometryUtil::getVerticesFromPlaneEquations(const btAlignedObjectArray<bt
 						potentialVertex *= quotient;
 
 						//check if inside, and replace supportingVertexOut if needed
-						if (isPointInsidePlanes(planeEquations,potentialVertex,btScalar(0.01)))
+						if (isPointInsidePlanes(planeEquations, potentialVertex, btScalar(0.01)))
 						{
 							verticesOut.push_back(potentialVertex);
 						}
@@ -182,4 +172,3 @@ void	btGeometryUtil::getVerticesFromPlaneEquations(const btAlignedObjectArray<bt
 		}
 	}
 }
-
diff --git a/thirdparty/bullet/LinearMath/btGeometryUtil.h b/thirdparty/bullet/LinearMath/btGeometryUtil.h
index a4b13b45609..0ce5b76d921 100644
--- a/thirdparty/bullet/LinearMath/btGeometryUtil.h
+++ b/thirdparty/bullet/LinearMath/btGeometryUtil.h
@@ -12,7 +12,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef BT_GEOMETRY_UTIL_H
 #define BT_GEOMETRY_UTIL_H
 
@@ -22,21 +21,16 @@ subject to the following restrictions:
 ///The btGeometryUtil helper class provides a few methods to convert between plane equations and vertices.
 class btGeometryUtil
 {
-	public:
-	
-	
-		static void	getPlaneEquationsFromVertices(btAlignedObjectArray<btVector3>& vertices, btAlignedObjectArray<btVector3>& planeEquationsOut );
+public:
+	static void getPlaneEquationsFromVertices(btAlignedObjectArray<btVector3>& vertices, btAlignedObjectArray<btVector3>& planeEquationsOut);
 
-		static void	getVerticesFromPlaneEquations(const btAlignedObjectArray<btVector3>& planeEquations , btAlignedObjectArray<btVector3>& verticesOut );
-	
-		static bool	isInside(const btAlignedObjectArray<btVector3>& vertices, const btVector3& planeNormal, btScalar	margin);
-		
-		static bool	isPointInsidePlanes(const btAlignedObjectArray<btVector3>& planeEquations, const btVector3& point, btScalar	margin);
+	static void getVerticesFromPlaneEquations(const btAlignedObjectArray<btVector3>& planeEquations, btAlignedObjectArray<btVector3>& verticesOut);
 
-		static bool	areVerticesBehindPlane(const btVector3& planeNormal, const btAlignedObjectArray<btVector3>& vertices, btScalar	margin);
+	static bool isInside(const btAlignedObjectArray<btVector3>& vertices, const btVector3& planeNormal, btScalar margin);
 
+	static bool isPointInsidePlanes(const btAlignedObjectArray<btVector3>& planeEquations, const btVector3& point, btScalar margin);
+
+	static bool areVerticesBehindPlane(const btVector3& planeNormal, const btAlignedObjectArray<btVector3>& vertices, btScalar margin);
 };
 
-
-#endif //BT_GEOMETRY_UTIL_H
-
+#endif  //BT_GEOMETRY_UTIL_H
diff --git a/thirdparty/bullet/LinearMath/btGrahamScan2dConvexHull.h b/thirdparty/bullet/LinearMath/btGrahamScan2dConvexHull.h
index 13a79aa5856..0fcb285971d 100644
--- a/thirdparty/bullet/LinearMath/btGrahamScan2dConvexHull.h
+++ b/thirdparty/bullet/LinearMath/btGrahamScan2dConvexHull.h
@@ -13,41 +13,40 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef GRAHAM_SCAN_2D_CONVEX_HULL_H
 #define GRAHAM_SCAN_2D_CONVEX_HULL_H
 
-
 #include "btVector3.h"
 #include "btAlignedObjectArray.h"
 
 struct GrahamVector3 : public btVector3
 {
 	GrahamVector3(const btVector3& org, int orgIndex)
-		:btVector3(org),
-			m_orgIndex(orgIndex)
+		: btVector3(org),
+		  m_orgIndex(orgIndex)
 	{
 	}
-	btScalar	m_angle;
+	btScalar m_angle;
 	int m_orgIndex;
 };
 
-
-struct btAngleCompareFunc {
+struct btAngleCompareFunc
+{
 	btVector3 m_anchor;
 	btAngleCompareFunc(const btVector3& anchor)
-	: m_anchor(anchor) 
+		: m_anchor(anchor)
 	{
 	}
-	bool operator()(const GrahamVector3& a, const GrahamVector3& b) const {
+	bool operator()(const GrahamVector3& a, const GrahamVector3& b) const
+	{
 		if (a.m_angle != b.m_angle)
 			return a.m_angle < b.m_angle;
 		else
 		{
-			btScalar al = (a-m_anchor).length2();
-			btScalar bl = (b-m_anchor).length2();
+			btScalar al = (a - m_anchor).length2();
+			btScalar bl = (b - m_anchor).length2();
 			if (al != bl)
-				return  al < bl;
+				return al < bl;
 			else
 			{
 				return a.m_orgIndex < b.m_orgIndex;
@@ -58,73 +57,73 @@ struct btAngleCompareFunc {
 
 inline void GrahamScanConvexHull2D(btAlignedObjectArray<GrahamVector3>& originalPoints, btAlignedObjectArray<GrahamVector3>& hull, const btVector3& normalAxis)
 {
-	btVector3 axis0,axis1;
-	btPlaneSpace1(normalAxis,axis0,axis1);
-	
+	btVector3 axis0, axis1;
+	btPlaneSpace1(normalAxis, axis0, axis1);
 
-	if (originalPoints.size()<=1)
+	if (originalPoints.size() <= 1)
 	{
-		for (int i=0;i<originalPoints.size();i++)
+		for (int i = 0; i < originalPoints.size(); i++)
 			hull.push_back(originalPoints[0]);
 		return;
 	}
 	//step1 : find anchor point with smallest projection on axis0 and move it to first location
-	for (int i=0;i<originalPoints.size();i++)
+	for (int i = 0; i < originalPoints.size(); i++)
 	{
-//		const btVector3& left = originalPoints[i];
-//		const btVector3& right = originalPoints[0];
+		//		const btVector3& left = originalPoints[i];
+		//		const btVector3& right = originalPoints[0];
 		btScalar projL = originalPoints[i].dot(axis0);
 		btScalar projR = originalPoints[0].dot(axis0);
 		if (projL < projR)
 		{
-			originalPoints.swap(0,i);
+			originalPoints.swap(0, i);
 		}
 	}
 
 	//also precompute angles
 	originalPoints[0].m_angle = -1e30f;
-	for (int i=1;i<originalPoints.size();i++)
+	for (int i = 1; i < originalPoints.size(); i++)
 	{
-	    btVector3 ar = originalPoints[i]-originalPoints[0];
-	    btScalar ar1 = axis1.dot(ar);
-	    btScalar ar0 = axis0.dot(ar);
-	    if( ar1*ar1+ar0*ar0 < FLT_EPSILON ) 
-	    {
-	      originalPoints[i].m_angle = 0.0f;
-	    }
-	    else
-	    {
-	      originalPoints[i].m_angle = btAtan2Fast(ar1, ar0);
-	    }
+		btVector3 ar = originalPoints[i] - originalPoints[0];
+		btScalar ar1 = axis1.dot(ar);
+		btScalar ar0 = axis0.dot(ar);
+		if (ar1 * ar1 + ar0 * ar0 < FLT_EPSILON)
+		{
+			originalPoints[i].m_angle = 0.0f;
+		}
+		else
+		{
+			originalPoints[i].m_angle = btAtan2Fast(ar1, ar0);
+		}
 	}
 
 	//step 2: sort all points, based on 'angle' with this anchor
 	btAngleCompareFunc comp(originalPoints[0]);
-	originalPoints.quickSortInternal(comp,1,originalPoints.size()-1);
+	originalPoints.quickSortInternal(comp, 1, originalPoints.size() - 1);
 
 	int i;
-	for (i = 0; i<2; i++) 
+	for (i = 0; i < 2; i++)
 		hull.push_back(originalPoints[i]);
 
 	//step 3: keep all 'convex' points and discard concave points (using back tracking)
-	for (; i != originalPoints.size(); i++) 
+	for (; i != originalPoints.size(); i++)
 	{
 		bool isConvex = false;
-		while (!isConvex&& hull.size()>1) {
-			btVector3& a = hull[hull.size()-2];
-			btVector3& b = hull[hull.size()-1];
-			isConvex = btCross(a-b,a-originalPoints[i]).dot(normalAxis)> 0;
+		while (!isConvex && hull.size() > 1)
+		{
+			btVector3& a = hull[hull.size() - 2];
+			btVector3& b = hull[hull.size() - 1];
+			isConvex = btCross(a - b, a - originalPoints[i]).dot(normalAxis) > 0;
 			if (!isConvex)
 				hull.pop_back();
-			else 
+			else
 				hull.push_back(originalPoints[i]);
 		}
 
-	    if( hull.size() == 1 )
-	    {
-	      hull.push_back( originalPoints[i] );
-	    }
+		if (hull.size() == 1)
+		{
+			hull.push_back(originalPoints[i]);
+		}
 	}
 }
 
-#endif //GRAHAM_SCAN_2D_CONVEX_HULL_H
+#endif  //GRAHAM_SCAN_2D_CONVEX_HULL_H
diff --git a/thirdparty/bullet/LinearMath/btHashMap.h b/thirdparty/bullet/LinearMath/btHashMap.h
index 180e7b44afa..1fca0fb73a5 100644
--- a/thirdparty/bullet/LinearMath/btHashMap.h
+++ b/thirdparty/bullet/LinearMath/btHashMap.h
@@ -13,7 +13,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef BT_HASH_MAP_H
 #define BT_HASH_MAP_H
 
@@ -24,32 +23,32 @@ subject to the following restrictions:
 struct btHashString
 {
 	std::string m_string1;
-	unsigned int	m_hash;
+	unsigned int m_hash;
 
-	SIMD_FORCE_INLINE	unsigned int getHash()const
+	SIMD_FORCE_INLINE unsigned int getHash() const
 	{
 		return m_hash;
 	}
 
 	btHashString()
 	{
-		m_string1="";
-		m_hash=0;
+		m_string1 = "";
+		m_hash = 0;
 	}
 	btHashString(const char* name)
-		:m_string1(name)
+		: m_string1(name)
 	{
 		/* magic numbers from http://www.isthe.com/chongo/tech/comp/fnv/ */
-		static const unsigned int  InitialFNV = 2166136261u;
+		static const unsigned int InitialFNV = 2166136261u;
 		static const unsigned int FNVMultiple = 16777619u;
 
 		/* Fowler / Noll / Vo (FNV) Hash */
 		unsigned int hash = InitialFNV;
-		
-		for(int i = 0; m_string1.c_str()[i]; i++)
+
+		for (int i = 0; m_string1.c_str()[i]; i++)
 		{
-			hash = hash ^ (m_string1.c_str()[i]);       /* xor  the low 8 bits */
-			hash = hash * FNVMultiple;  /* multiply by the magic number */
+			hash = hash ^ (m_string1.c_str()[i]); /* xor  the low 8 bits */
+			hash = hash * FNVMultiple;            /* multiply by the magic number */
 		}
 		m_hash = hash;
 	}
@@ -60,28 +59,27 @@ struct btHashString
 	}
 };
 
-const int BT_HASH_NULL=0xffffffff;
-
+const int BT_HASH_NULL = 0xffffffff;
 
 class btHashInt
 {
-	int	m_uid;
-public:
+	int m_uid;
 
+public:
 	btHashInt()
 	{
 	}
 
-	btHashInt(int uid)	:m_uid(uid)
+	btHashInt(int uid) : m_uid(uid)
 	{
 	}
 
-	int	getUid1() const
+	int getUid1() const
 	{
 		return m_uid;
 	}
 
-	void	setUid1(int uid)
+	void setUid1(int uid)
 	{
 		m_uid = uid;
 	}
@@ -91,35 +89,35 @@ public:
 		return getUid1() == other.getUid1();
 	}
 	//to our success
-	SIMD_FORCE_INLINE	unsigned int getHash()const
+	SIMD_FORCE_INLINE unsigned int getHash() const
 	{
 		unsigned int key = m_uid;
 		// Thomas Wang's hash
-		key += ~(key << 15);	key ^=  (key >> 10);	key +=  (key << 3);	key ^=  (key >> 6);	key += ~(key << 11);	key ^=  (key >> 16);
-		
+		key += ~(key << 15);
+		key ^= (key >> 10);
+		key += (key << 3);
+		key ^= (key >> 6);
+		key += ~(key << 11);
+		key ^= (key >> 16);
+
 		return key;
 	}
 };
 
-
-
 class btHashPtr
 {
-
-	union
-	{
-		const void*	m_pointer;
-		unsigned int	m_hashValues[2];
+	union {
+		const void* m_pointer;
+		unsigned int m_hashValues[2];
 	};
 
 public:
-
 	btHashPtr(const void* ptr)
-		:m_pointer(ptr)
+		: m_pointer(ptr)
 	{
 	}
 
-	const void*	getPointer() const
+	const void* getPointer() const
 	{
 		return m_pointer;
 	}
@@ -130,64 +128,68 @@ public:
 	}
 
 	//to our success
-	SIMD_FORCE_INLINE	unsigned int getHash()const
+	SIMD_FORCE_INLINE unsigned int getHash() const
 	{
-		const bool VOID_IS_8 = ((sizeof(void*)==8));
-		
-		unsigned int key = VOID_IS_8? m_hashValues[0]+m_hashValues[1] : m_hashValues[0];
+		const bool VOID_IS_8 = ((sizeof(void*) == 8));
+
+		unsigned int key = VOID_IS_8 ? m_hashValues[0] + m_hashValues[1] : m_hashValues[0];
 		// Thomas Wang's hash
-		key += ~(key << 15);	key ^=  (key >> 10);	key +=  (key << 3);	key ^=  (key >> 6);	key += ~(key << 11);	key ^=  (key >> 16);
+		key += ~(key << 15);
+		key ^= (key >> 10);
+		key += (key << 3);
+		key ^= (key >> 6);
+		key += ~(key << 11);
+		key ^= (key >> 16);
 		return key;
 	}
-
-	
 };
 
-
 template <class Value>
 class btHashKeyPtr
 {
-        int     m_uid;
+	int m_uid;
+
 public:
+	btHashKeyPtr(int uid) : m_uid(uid)
+	{
+	}
 
-        btHashKeyPtr(int uid)    :m_uid(uid)
-        {
-        }
+	int getUid1() const
+	{
+		return m_uid;
+	}
 
-        int     getUid1() const
-        {
-                return m_uid;
-        }
+	bool equals(const btHashKeyPtr<Value>& other) const
+	{
+		return getUid1() == other.getUid1();
+	}
 
-        bool equals(const btHashKeyPtr<Value>& other) const
-        {
-                return getUid1() == other.getUid1();
-        }
-
-        //to our success
-        SIMD_FORCE_INLINE       unsigned int getHash()const
-        {
-                unsigned int key = m_uid;
-                // Thomas Wang's hash
-                key += ~(key << 15);	key ^=  (key >> 10);	key +=  (key << 3);	key ^=  (key >> 6);	key += ~(key << 11);	key ^=  (key >> 16);
-                return key;
-        }
-
-        
+	//to our success
+	SIMD_FORCE_INLINE unsigned int getHash() const
+	{
+		unsigned int key = m_uid;
+		// Thomas Wang's hash
+		key += ~(key << 15);
+		key ^= (key >> 10);
+		key += (key << 3);
+		key ^= (key >> 6);
+		key += ~(key << 11);
+		key ^= (key >> 16);
+		return key;
+	}
 };
 
-
 template <class Value>
 class btHashKey
 {
-	int	m_uid;
-public:
+	int m_uid;
 
-	btHashKey(int uid)	:m_uid(uid)
+public:
+	btHashKey(int uid) : m_uid(uid)
 	{
 	}
 
-	int	getUid1() const
+	int getUid1() const
 	{
 		return m_uid;
 	}
@@ -197,30 +199,33 @@ public:
 		return getUid1() == other.getUid1();
 	}
 	//to our success
-	SIMD_FORCE_INLINE	unsigned int getHash()const
+	SIMD_FORCE_INLINE unsigned int getHash() const
 	{
 		unsigned int key = m_uid;
 		// Thomas Wang's hash
-		key += ~(key << 15);	key ^=  (key >> 10);	key +=  (key << 3);	key ^=  (key >> 6);	key += ~(key << 11);	key ^=  (key >> 16);
+		key += ~(key << 15);
+		key ^= (key >> 10);
+		key += (key << 3);
+		key ^= (key >> 6);
+		key += ~(key << 11);
+		key ^= (key >> 16);
 		return key;
 	}
 };
 
-
 ///The btHashMap template class implements a generic and lightweight hashmap.
 ///A basic sample of how to use btHashMap is located in Demos\BasicDemo\main.cpp
 template <class Key, class Value>
 class btHashMap
 {
-
 protected:
-	btAlignedObjectArray<int>		m_hashTable;
-	btAlignedObjectArray<int>		m_next;
-	
-	btAlignedObjectArray<Value>		m_valueArray;
-	btAlignedObjectArray<Key>		m_keyArray;
+	btAlignedObjectArray<int> m_hashTable;
+	btAlignedObjectArray<int> m_next;
 
-	void	growTables(const Key& /*key*/)
+	btAlignedObjectArray<Value> m_valueArray;
+	btAlignedObjectArray<Key> m_keyArray;
+
+	void growTables(const Key& /*key*/)
 	{
 		int newCapacity = m_valueArray.capacity();
 
@@ -234,7 +239,7 @@ protected:
 
 			int i;
 
-			for (i= 0; i < newCapacity; ++i)
+			for (i = 0; i < newCapacity; ++i)
 			{
 				m_hashTable[i] = BT_HASH_NULL;
 			}
@@ -243,30 +248,28 @@ protected:
 				m_next[i] = BT_HASH_NULL;
 			}
 
-			for(i=0;i<curHashtableSize;i++)
+			for (i = 0; i < curHashtableSize; i++)
 			{
 				//const Value& value = m_valueArray[i];
 				//const Key& key = m_keyArray[i];
 
-				int	hashValue = m_keyArray[i].getHash() & (m_valueArray.capacity()-1);	// New hash value with new mask
+				int hashValue = m_keyArray[i].getHash() & (m_valueArray.capacity() - 1);  // New hash value with new mask
 				m_next[i] = m_hashTable[hashValue];
 				m_hashTable[hashValue] = i;
 			}
-
-
 		}
 	}
 
-	public:
-
-	void insert(const Key& key, const Value& value) {
-		int hash = key.getHash() & (m_valueArray.capacity()-1);
+public:
+	void insert(const Key& key, const Value& value)
+	{
+		int hash = key.getHash() & (m_valueArray.capacity() - 1);
 
 		//replace value if the key is already there
 		int index = findIndex(key);
 		if (index != BT_HASH_NULL)
 		{
-			m_valueArray[index]=value;
+			m_valueArray[index] = value;
 			return;
 		}
 
@@ -280,19 +283,19 @@ protected:
 		{
 			growTables(key);
 			//hash with new capacity
-			hash = key.getHash() & (m_valueArray.capacity()-1);
+			hash = key.getHash() & (m_valueArray.capacity() - 1);
 		}
 		m_next[count] = m_hashTable[hash];
 		m_hashTable[hash] = count;
 	}
 
-	void remove(const Key& key) {
-
-		int hash = key.getHash() & (m_valueArray.capacity()-1);
+	void remove(const Key& key)
+	{
+		int hash = key.getHash() & (m_valueArray.capacity() - 1);
 
 		int pairIndex = findIndex(key);
-		
-		if (pairIndex ==BT_HASH_NULL)
+
+		if (pairIndex == BT_HASH_NULL)
 		{
 			return;
 		}
@@ -333,7 +336,7 @@ protected:
 		}
 
 		// Remove the last pair from the hash table.
-		int lastHash = m_keyArray[lastPairIndex].getHash() & (m_valueArray.capacity()-1);
+		int lastHash = m_keyArray[lastPairIndex].getHash() & (m_valueArray.capacity() - 1);
 
 		index = m_hashTable[lastHash];
 		btAssert(index != BT_HASH_NULL);
@@ -365,10 +368,8 @@ protected:
 
 		m_valueArray.pop_back();
 		m_keyArray.pop_back();
-
 	}
 
-
 	int size() const
 	{
 		return m_valueArray.size();
@@ -377,8 +378,8 @@ protected:
 	const Value* getAtIndex(int index) const
 	{
 		btAssert(index < m_valueArray.size());
-		btAssert(index>=0);
-		if (index>=0 && index < m_valueArray.size())
+		btAssert(index >= 0);
+		if (index >= 0 && index < m_valueArray.size())
 		{
 			return &m_valueArray[index];
 		}
@@ -388,38 +389,39 @@ protected:
 	Value* getAtIndex(int index)
 	{
 		btAssert(index < m_valueArray.size());
-		btAssert(index>=0);
-		if (index>=0 && index < m_valueArray.size())
+		btAssert(index >= 0);
+		if (index >= 0 && index < m_valueArray.size())
 		{
 			return &m_valueArray[index];
 		}
 		return 0;
 	}
 
-    Key getKeyAtIndex(int index)
-    {
-        btAssert(index < m_keyArray.size());
-		btAssert(index>=0);
+	Key getKeyAtIndex(int index)
+	{
+		btAssert(index < m_keyArray.size());
+		btAssert(index >= 0);
 		return m_keyArray[index];
-    }
-    
-    const Key getKeyAtIndex(int index) const
-    {
-        btAssert(index < m_keyArray.size());
-		btAssert(index>=0);
+	}
+
+	const Key getKeyAtIndex(int index) const
+	{
+		btAssert(index < m_keyArray.size());
+		btAssert(index >= 0);
 		return m_keyArray[index];
-   }
+	}
 
-
-	Value* operator[](const Key& key) {
+	Value* operator[](const Key& key)
+	{
 		return find(key);
 	}
 
-	const Value* operator[](const Key& key) const {
+	const Value* operator[](const Key& key) const
+	{
 		return find(key);
 	}
 
-	const Value*	find(const Key& key) const
+	const Value* find(const Key& key) const
 	{
 		int index = findIndex(key);
 		if (index == BT_HASH_NULL)
@@ -429,7 +431,7 @@ protected:
 		return &m_valueArray[index];
 	}
 
-	Value*	find(const Key& key)
+	Value* find(const Key& key)
 	{
 		int index = findIndex(key);
 		if (index == BT_HASH_NULL)
@@ -439,10 +441,9 @@ protected:
 		return &m_valueArray[index];
 	}
 
-
-	int	findIndex(const Key& key) const
+	int findIndex(const Key& key) const
 	{
-		unsigned int hash = key.getHash() & (m_valueArray.capacity()-1);
+		unsigned int hash = key.getHash() & (m_valueArray.capacity() - 1);
 
 		if (hash >= (unsigned int)m_hashTable.size())
 		{
@@ -457,14 +458,13 @@ protected:
 		return index;
 	}
 
-	void	clear()
+	void clear()
 	{
 		m_hashTable.clear();
 		m_next.clear();
 		m_valueArray.clear();
 		m_keyArray.clear();
 	}
-
 };
 
-#endif //BT_HASH_MAP_H
+#endif  //BT_HASH_MAP_H
diff --git a/thirdparty/bullet/LinearMath/btIDebugDraw.h b/thirdparty/bullet/LinearMath/btIDebugDraw.h
index b57282717d7..82ec19a69b2 100644
--- a/thirdparty/bullet/LinearMath/btIDebugDraw.h
+++ b/thirdparty/bullet/LinearMath/btIDebugDraw.h
@@ -13,86 +13,84 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef BT_IDEBUG_DRAW__H
 #define BT_IDEBUG_DRAW__H
 
 #include "btVector3.h"
 #include "btTransform.h"
 
-
-
 ///The btIDebugDraw interface class allows hooking up a debug renderer to visually debug simulations.
 ///Typical use case: create a debug drawer object, and assign it to a btCollisionWorld or btDynamicsWorld using setDebugDrawer and call debugDrawWorld.
 ///A class that implements the btIDebugDraw interface has to implement the drawLine method at a minimum.
 ///For color arguments the X,Y,Z components refer to Red, Green and Blue each in the range [0..1]
-class	btIDebugDraw
+class btIDebugDraw
 {
-	public:
-
-	ATTRIBUTE_ALIGNED16(struct) DefaultColors
+public:
+	ATTRIBUTE_ALIGNED16(struct)
+	DefaultColors
 	{
-		btVector3	m_activeObject;
-		btVector3	m_deactivatedObject;
-		btVector3	m_wantsDeactivationObject;
-		btVector3	m_disabledDeactivationObject;
-		btVector3	m_disabledSimulationObject;
-		btVector3	m_aabb;
+		btVector3 m_activeObject;
+		btVector3 m_deactivatedObject;
+		btVector3 m_wantsDeactivationObject;
+		btVector3 m_disabledDeactivationObject;
+		btVector3 m_disabledSimulationObject;
+		btVector3 m_aabb;
 		btVector3 m_contactPoint;
-		
+
 		DefaultColors()
-		:	m_activeObject(1,1,1),
-			m_deactivatedObject(0,1,0),
-			m_wantsDeactivationObject(0,1,1),
-			m_disabledDeactivationObject(1,0,0),
-			m_disabledSimulationObject(1,1,0),
-			m_aabb(1,0,0),
-			m_contactPoint(1,1,0)
+			: m_activeObject(1, 1, 1),
+			  m_deactivatedObject(0, 1, 0),
+			  m_wantsDeactivationObject(0, 1, 1),
+			  m_disabledDeactivationObject(1, 0, 0),
+			  m_disabledSimulationObject(1, 1, 0),
+			  m_aabb(1, 0, 0),
+			  m_contactPoint(1, 1, 0)
 		{
 		}
 	};
 
-	
-	enum	DebugDrawModes
+	enum DebugDrawModes
 	{
-		DBG_NoDebug=0,
+		DBG_NoDebug = 0,
 		DBG_DrawWireframe = 1,
-		DBG_DrawAabb=2,
-		DBG_DrawFeaturesText=4,
-		DBG_DrawContactPoints=8,
-		DBG_NoDeactivation=16,
+		DBG_DrawAabb = 2,
+		DBG_DrawFeaturesText = 4,
+		DBG_DrawContactPoints = 8,
+		DBG_NoDeactivation = 16,
 		DBG_NoHelpText = 32,
-		DBG_DrawText=64,
+		DBG_DrawText = 64,
 		DBG_ProfileTimings = 128,
 		DBG_EnableSatComparison = 256,
 		DBG_DisableBulletLCP = 512,
 		DBG_EnableCCD = 1024,
 		DBG_DrawConstraints = (1 << 11),
 		DBG_DrawConstraintLimits = (1 << 12),
-		DBG_FastWireframe = (1<<13),
-		DBG_DrawNormals = (1<<14),
-		DBG_DrawFrames = (1<<15),
+		DBG_FastWireframe = (1 << 13),
+		DBG_DrawNormals = (1 << 14),
+		DBG_DrawFrames = (1 << 15),
 		DBG_MAX_DEBUG_DRAW_MODE
 	};
 
-	virtual ~btIDebugDraw() {};
+	virtual ~btIDebugDraw(){};
 
-	
-	virtual DefaultColors	getDefaultColors() const	{	DefaultColors colors;	return colors;	}
+	virtual DefaultColors getDefaultColors() const
+	{
+		DefaultColors colors;
+		return colors;
+	}
 	///the default implementation for setDefaultColors has no effect. A derived class can implement it and store the colors.
 	virtual void setDefaultColors(const DefaultColors& /*colors*/) {}
-	
-	virtual void	drawLine(const btVector3& from,const btVector3& to,const btVector3& color)=0;
-		
-	virtual void    drawLine(const btVector3& from,const btVector3& to, const btVector3& fromColor, const btVector3& toColor)
+
+	virtual void drawLine(const btVector3& from, const btVector3& to, const btVector3& color) = 0;
+
+	virtual void drawLine(const btVector3& from, const btVector3& to, const btVector3& fromColor, const btVector3& toColor)
 	{
-        (void) toColor;
-		drawLine (from, to, fromColor);
+		(void)toColor;
+		drawLine(from, to, fromColor);
 	}
 
-	virtual void	drawSphere(btScalar radius, const btTransform& transform, const btVector3& color)
+	virtual void drawSphere(btScalar radius, const btTransform& transform, const btVector3& color)
 	{
-		
 		btVector3 center = transform.getOrigin();
 		btVector3 up = transform.getBasis().getColumn(1);
 		btVector3 axis = transform.getBasis().getColumn(0);
@@ -101,103 +99,102 @@ class	btIDebugDraw
 		btScalar minPs = -SIMD_HALF_PI;
 		btScalar maxPs = SIMD_HALF_PI;
 		btScalar stepDegrees = 30.f;
-		drawSpherePatch(center, up, axis, radius,minTh, maxTh, minPs, maxPs, color, stepDegrees ,false);
-		drawSpherePatch(center, up, -axis, radius,minTh, maxTh, minPs, maxPs, color, stepDegrees,false );
+		drawSpherePatch(center, up, axis, radius, minTh, maxTh, minPs, maxPs, color, stepDegrees, false);
+		drawSpherePatch(center, up, -axis, radius, minTh, maxTh, minPs, maxPs, color, stepDegrees, false);
 	}
-	
-	virtual void	drawSphere (const btVector3& p, btScalar radius, const btVector3& color)
+
+	virtual void drawSphere(const btVector3& p, btScalar radius, const btVector3& color)
 	{
 		btTransform tr;
 		tr.setIdentity();
 		tr.setOrigin(p);
-		drawSphere(radius,tr,color);
-	}
-	
-	virtual	void	drawTriangle(const btVector3& v0,const btVector3& v1,const btVector3& v2,const btVector3& /*n0*/,const btVector3& /*n1*/,const btVector3& /*n2*/,const btVector3& color, btScalar alpha)
-	{
-		drawTriangle(v0,v1,v2,color,alpha);
-	}
-	virtual	void	drawTriangle(const btVector3& v0,const btVector3& v1,const btVector3& v2,const btVector3& color, btScalar /*alpha*/)
-	{
-		drawLine(v0,v1,color);
-		drawLine(v1,v2,color);
-		drawLine(v2,v0,color);
+		drawSphere(radius, tr, color);
 	}
 
-	virtual void	drawContactPoint(const btVector3& PointOnB,const btVector3& normalOnB,btScalar distance,int lifeTime,const btVector3& color)=0;
-
-	virtual void	reportErrorWarning(const char* warningString) = 0;
-
-	virtual void	draw3dText(const btVector3& location,const char* textString) = 0;
-	
-	virtual void	setDebugMode(int debugMode) =0;
-	
-	virtual int		getDebugMode() const = 0;
-
-	virtual void drawAabb(const btVector3& from,const btVector3& to,const btVector3& color)
+	virtual void drawTriangle(const btVector3& v0, const btVector3& v1, const btVector3& v2, const btVector3& /*n0*/, const btVector3& /*n1*/, const btVector3& /*n2*/, const btVector3& color, btScalar alpha)
 	{
+		drawTriangle(v0, v1, v2, color, alpha);
+	}
+	virtual void drawTriangle(const btVector3& v0, const btVector3& v1, const btVector3& v2, const btVector3& color, btScalar /*alpha*/)
+	{
+		drawLine(v0, v1, color);
+		drawLine(v1, v2, color);
+		drawLine(v2, v0, color);
+	}
 
-		btVector3 halfExtents = (to-from)* 0.5f;
-		btVector3 center = (to+from) *0.5f;
-		int i,j;
+	virtual void drawContactPoint(const btVector3& PointOnB, const btVector3& normalOnB, btScalar distance, int lifeTime, const btVector3& color) = 0;
 
-		btVector3 edgecoord(1.f,1.f,1.f),pa,pb;
-		for (i=0;i<4;i++)
+	virtual void reportErrorWarning(const char* warningString) = 0;
+
+	virtual void draw3dText(const btVector3& location, const char* textString) = 0;
+
+	virtual void setDebugMode(int debugMode) = 0;
+
+	virtual int getDebugMode() const = 0;
+
+	virtual void drawAabb(const btVector3& from, const btVector3& to, const btVector3& color)
+	{
+		btVector3 halfExtents = (to - from) * 0.5f;
+		btVector3 center = (to + from) * 0.5f;
+		int i, j;
+
+		btVector3 edgecoord(1.f, 1.f, 1.f), pa, pb;
+		for (i = 0; i < 4; i++)
 		{
-			for (j=0;j<3;j++)
+			for (j = 0; j < 3; j++)
 			{
-				pa = btVector3(edgecoord[0]*halfExtents[0], edgecoord[1]*halfExtents[1],		
-					edgecoord[2]*halfExtents[2]);
-				pa+=center;
+				pa = btVector3(edgecoord[0] * halfExtents[0], edgecoord[1] * halfExtents[1],
+							   edgecoord[2] * halfExtents[2]);
+				pa += center;
 
-				int othercoord = j%3;
-				edgecoord[othercoord]*=-1.f;
-				pb = btVector3(edgecoord[0]*halfExtents[0], edgecoord[1]*halfExtents[1],	
-					edgecoord[2]*halfExtents[2]);
-				pb+=center;
+				int othercoord = j % 3;
+				edgecoord[othercoord] *= -1.f;
+				pb = btVector3(edgecoord[0] * halfExtents[0], edgecoord[1] * halfExtents[1],
+							   edgecoord[2] * halfExtents[2]);
+				pb += center;
 
-				drawLine(pa,pb,color);
+				drawLine(pa, pb, color);
 			}
-			edgecoord = btVector3(-1.f,-1.f,-1.f);
-			if (i<3)
-				edgecoord[i]*=-1.f;
+			edgecoord = btVector3(-1.f, -1.f, -1.f);
+			if (i < 3)
+				edgecoord[i] *= -1.f;
 		}
 	}
 	virtual void drawTransform(const btTransform& transform, btScalar orthoLen)
 	{
 		btVector3 start = transform.getOrigin();
-		drawLine(start, start+transform.getBasis() * btVector3(orthoLen, 0, 0), btVector3(btScalar(1.), btScalar(0.3), btScalar(0.3)));
-		drawLine(start, start+transform.getBasis() * btVector3(0, orthoLen, 0), btVector3(btScalar(0.3), btScalar(1.), btScalar(0.3)));
-		drawLine(start, start+transform.getBasis() * btVector3(0, 0, orthoLen), btVector3(btScalar(0.3), btScalar(0.3), btScalar(1.)));
+		drawLine(start, start + transform.getBasis() * btVector3(orthoLen, 0, 0), btVector3(btScalar(1.), btScalar(0.3), btScalar(0.3)));
+		drawLine(start, start + transform.getBasis() * btVector3(0, orthoLen, 0), btVector3(btScalar(0.3), btScalar(1.), btScalar(0.3)));
+		drawLine(start, start + transform.getBasis() * btVector3(0, 0, orthoLen), btVector3(btScalar(0.3), btScalar(0.3), btScalar(1.)));
 	}
 
-	virtual void drawArc(const btVector3& center, const btVector3& normal, const btVector3& axis, btScalar radiusA, btScalar radiusB, btScalar minAngle, btScalar maxAngle, 
-				const btVector3& color, bool drawSect, btScalar stepDegrees = btScalar(10.f))
+	virtual void drawArc(const btVector3& center, const btVector3& normal, const btVector3& axis, btScalar radiusA, btScalar radiusB, btScalar minAngle, btScalar maxAngle,
+						 const btVector3& color, bool drawSect, btScalar stepDegrees = btScalar(10.f))
 	{
 		const btVector3& vx = axis;
 		btVector3 vy = normal.cross(axis);
 		btScalar step = stepDegrees * SIMD_RADS_PER_DEG;
 		int nSteps = (int)btFabs((maxAngle - minAngle) / step);
-		if(!nSteps) nSteps = 1;
+		if (!nSteps) nSteps = 1;
 		btVector3 prev = center + radiusA * vx * btCos(minAngle) + radiusB * vy * btSin(minAngle);
-		if(drawSect)
+		if (drawSect)
 		{
 			drawLine(center, prev, color);
 		}
-		for(int i = 1; i <= nSteps; i++)
+		for (int i = 1; i <= nSteps; i++)
 		{
 			btScalar angle = minAngle + (maxAngle - minAngle) * btScalar(i) / btScalar(nSteps);
 			btVector3 next = center + radiusA * vx * btCos(angle) + radiusB * vy * btSin(angle);
 			drawLine(prev, next, color);
 			prev = next;
 		}
-		if(drawSect)
+		if (drawSect)
 		{
 			drawLine(center, prev, color);
 		}
 	}
-	virtual void drawSpherePatch(const btVector3& center, const btVector3& up, const btVector3& axis, btScalar radius, 
-		btScalar minTh, btScalar maxTh, btScalar minPs, btScalar maxPs, const btVector3& color, btScalar stepDegrees = btScalar(10.f),bool drawCenter = true)
+	virtual void drawSpherePatch(const btVector3& center, const btVector3& up, const btVector3& axis, btScalar radius,
+								 btScalar minTh, btScalar maxTh, btScalar minPs, btScalar maxPs, const btVector3& color, btScalar stepDegrees = btScalar(10.f), bool drawCenter = true)
 	{
 		btVector3 vA[74];
 		btVector3 vB[74];
@@ -211,33 +208,33 @@ class	btIDebugDraw
 		btVector3 jv = kv.cross(iv);
 		bool drawN = false;
 		bool drawS = false;
-		if(minTh <= -SIMD_HALF_PI)
+		if (minTh <= -SIMD_HALF_PI)
 		{
 			minTh = -SIMD_HALF_PI + step;
 			drawN = true;
 		}
-		if(maxTh >= SIMD_HALF_PI)
+		if (maxTh >= SIMD_HALF_PI)
 		{
 			maxTh = SIMD_HALF_PI - step;
 			drawS = true;
 		}
-		if(minTh > maxTh)
+		if (minTh > maxTh)
 		{
 			minTh = -SIMD_HALF_PI + step;
-			maxTh =  SIMD_HALF_PI - step;
+			maxTh = SIMD_HALF_PI - step;
 			drawN = drawS = true;
 		}
 		int n_hor = (int)((maxTh - minTh) / step) + 1;
-		if(n_hor < 2) n_hor = 2;
+		if (n_hor < 2) n_hor = 2;
 		btScalar step_h = (maxTh - minTh) / btScalar(n_hor - 1);
 		bool isClosed = false;
-		if(minPs > maxPs)
+		if (minPs > maxPs)
 		{
 			minPs = -SIMD_PI + step;
-			maxPs =  SIMD_PI;
+			maxPs = SIMD_PI;
 			isClosed = true;
 		}
-		else if((maxPs - minPs) >= SIMD_PI * btScalar(2.f))
+		else if ((maxPs - minPs) >= SIMD_PI * btScalar(2.f))
 		{
 			isClosed = true;
 		}
@@ -246,63 +243,64 @@ class	btIDebugDraw
 			isClosed = false;
 		}
 		int n_vert = (int)((maxPs - minPs) / step) + 1;
-		if(n_vert < 2) n_vert = 2;
+		if (n_vert < 2) n_vert = 2;
 		btScalar step_v = (maxPs - minPs) / btScalar(n_vert - 1);
-		for(int i = 0; i < n_hor; i++)
+		for (int i = 0; i < n_hor; i++)
 		{
 			btScalar th = minTh + btScalar(i) * step_h;
 			btScalar sth = radius * btSin(th);
 			btScalar cth = radius * btCos(th);
-			for(int j = 0; j < n_vert; j++)
+			for (int j = 0; j < n_vert; j++)
 			{
 				btScalar psi = minPs + btScalar(j) * step_v;
 				btScalar sps = btSin(psi);
 				btScalar cps = btCos(psi);
 				pvB[j] = center + cth * cps * iv + cth * sps * jv + sth * kv;
-				if(i)
+				if (i)
 				{
 					drawLine(pvA[j], pvB[j], color);
 				}
-				else if(drawS)
+				else if (drawS)
 				{
 					drawLine(spole, pvB[j], color);
 				}
-				if(j)
+				if (j)
 				{
-					drawLine(pvB[j-1], pvB[j], color);
+					drawLine(pvB[j - 1], pvB[j], color);
 				}
 				else
 				{
 					arcStart = pvB[j];
 				}
-				if((i == (n_hor - 1)) && drawN)
+				if ((i == (n_hor - 1)) && drawN)
 				{
 					drawLine(npole, pvB[j], color);
 				}
-				
+
 				if (drawCenter)
 				{
-					if(isClosed)
+					if (isClosed)
 					{
-						if(j == (n_vert-1))
+						if (j == (n_vert - 1))
 						{
 							drawLine(arcStart, pvB[j], color);
 						}
 					}
 					else
 					{
-						if(((!i) || (i == (n_hor-1))) && ((!j) || (j == (n_vert-1))))
+						if (((!i) || (i == (n_hor - 1))) && ((!j) || (j == (n_vert - 1))))
 						{
 							drawLine(center, pvB[j], color);
 						}
 					}
 				}
 			}
-			pT = pvA; pvA = pvB; pvB = pT;
+			pT = pvA;
+			pvA = pvB;
+			pvB = pT;
 		}
 	}
-	
-  
+
 	virtual void drawBox(const btVector3& bbMin, const btVector3& bbMax, const btVector3& color)
 	{
 		drawLine(btVector3(bbMin[0], bbMin[1], bbMin[2]), btVector3(bbMax[0], bbMin[1], bbMin[2]), color);
@@ -338,31 +336,27 @@ class	btIDebugDraw
 	{
 		int stepDegrees = 30;
 
-		btVector3 capStart(0.f,0.f,0.f);
+		btVector3 capStart(0.f, 0.f, 0.f);
 		capStart[upAxis] = -halfHeight;
 
-		btVector3 capEnd(0.f,0.f,0.f);
+		btVector3 capEnd(0.f, 0.f, 0.f);
 		capEnd[upAxis] = halfHeight;
 
 		// Draw the ends
 		{
-
 			btTransform childTransform = transform;
 			childTransform.getOrigin() = transform * capStart;
 			{
 				btVector3 center = childTransform.getOrigin();
-				btVector3 up = childTransform.getBasis().getColumn((upAxis+1)%3);
+				btVector3 up = childTransform.getBasis().getColumn((upAxis + 1) % 3);
 				btVector3 axis = -childTransform.getBasis().getColumn(upAxis);
 				btScalar minTh = -SIMD_HALF_PI;
 				btScalar maxTh = SIMD_HALF_PI;
 				btScalar minPs = -SIMD_HALF_PI;
 				btScalar maxPs = SIMD_HALF_PI;
-				
-				drawSpherePatch(center, up, axis, radius,minTh, maxTh, minPs, maxPs, color, btScalar(stepDegrees) ,false);
+
+				drawSpherePatch(center, up, axis, radius, minTh, maxTh, minPs, maxPs, color, btScalar(stepDegrees), false);
 			}
-
-
-
 		}
 
 		{
@@ -370,52 +364,51 @@ class	btIDebugDraw
 			childTransform.getOrigin() = transform * capEnd;
 			{
 				btVector3 center = childTransform.getOrigin();
-				btVector3 up = childTransform.getBasis().getColumn((upAxis+1)%3);
+				btVector3 up = childTransform.getBasis().getColumn((upAxis + 1) % 3);
 				btVector3 axis = childTransform.getBasis().getColumn(upAxis);
 				btScalar minTh = -SIMD_HALF_PI;
 				btScalar maxTh = SIMD_HALF_PI;
 				btScalar minPs = -SIMD_HALF_PI;
 				btScalar maxPs = SIMD_HALF_PI;
-				drawSpherePatch(center, up, axis, radius,minTh, maxTh, minPs, maxPs, color, btScalar(stepDegrees) ,false);
+				drawSpherePatch(center, up, axis, radius, minTh, maxTh, minPs, maxPs, color, btScalar(stepDegrees), false);
 			}
 		}
 
 		// Draw some additional lines
 		btVector3 start = transform.getOrigin();
 
-		for (int i=0;i<360;i+=stepDegrees)
+		for (int i = 0; i < 360; i += stepDegrees)
 		{
-			capEnd[(upAxis+1)%3] = capStart[(upAxis+1)%3] = btSin(btScalar(i)*SIMD_RADS_PER_DEG)*radius;
-			capEnd[(upAxis+2)%3] = capStart[(upAxis+2)%3]  = btCos(btScalar(i)*SIMD_RADS_PER_DEG)*radius;
-			drawLine(start+transform.getBasis() * capStart,start+transform.getBasis() * capEnd, color);
+			capEnd[(upAxis + 1) % 3] = capStart[(upAxis + 1) % 3] = btSin(btScalar(i) * SIMD_RADS_PER_DEG) * radius;
+			capEnd[(upAxis + 2) % 3] = capStart[(upAxis + 2) % 3] = btCos(btScalar(i) * SIMD_RADS_PER_DEG) * radius;
+			drawLine(start + transform.getBasis() * capStart, start + transform.getBasis() * capEnd, color);
 		}
-		
 	}
 
 	virtual void drawCylinder(btScalar radius, btScalar halfHeight, int upAxis, const btTransform& transform, const btVector3& color)
 	{
 		btVector3 start = transform.getOrigin();
-		btVector3	offsetHeight(0,0,0);
+		btVector3 offsetHeight(0, 0, 0);
 		offsetHeight[upAxis] = halfHeight;
-		int stepDegrees=30;
-		btVector3 capStart(0.f,0.f,0.f);
+		int stepDegrees = 30;
+		btVector3 capStart(0.f, 0.f, 0.f);
 		capStart[upAxis] = -halfHeight;
-		btVector3 capEnd(0.f,0.f,0.f);
+		btVector3 capEnd(0.f, 0.f, 0.f);
 		capEnd[upAxis] = halfHeight;
 
-		for (int i=0;i<360;i+=stepDegrees)
+		for (int i = 0; i < 360; i += stepDegrees)
 		{
-			capEnd[(upAxis+1)%3] = capStart[(upAxis+1)%3] = btSin(btScalar(i)*SIMD_RADS_PER_DEG)*radius;
-			capEnd[(upAxis+2)%3] = capStart[(upAxis+2)%3]  = btCos(btScalar(i)*SIMD_RADS_PER_DEG)*radius;
-			drawLine(start+transform.getBasis() * capStart,start+transform.getBasis() * capEnd, color);
+			capEnd[(upAxis + 1) % 3] = capStart[(upAxis + 1) % 3] = btSin(btScalar(i) * SIMD_RADS_PER_DEG) * radius;
+			capEnd[(upAxis + 2) % 3] = capStart[(upAxis + 2) % 3] = btCos(btScalar(i) * SIMD_RADS_PER_DEG) * radius;
+			drawLine(start + transform.getBasis() * capStart, start + transform.getBasis() * capEnd, color);
 		}
 		// Drawing top and bottom caps of the cylinder
-		btVector3 yaxis(0,0,0);
+		btVector3 yaxis(0, 0, 0);
 		yaxis[upAxis] = btScalar(1.0);
-		btVector3 xaxis(0,0,0);
-		xaxis[(upAxis+1)%3] = btScalar(1.0);
-		drawArc(start-transform.getBasis()*(offsetHeight),transform.getBasis()*yaxis,transform.getBasis()*xaxis,radius,radius,0,SIMD_2_PI,color,false,btScalar(10.0));
-		drawArc(start+transform.getBasis()*(offsetHeight),transform.getBasis()*yaxis,transform.getBasis()*xaxis,radius,radius,0,SIMD_2_PI,color,false,btScalar(10.0));
+		btVector3 xaxis(0, 0, 0);
+		xaxis[(upAxis + 1) % 3] = btScalar(1.0);
+		drawArc(start - transform.getBasis() * (offsetHeight), transform.getBasis() * yaxis, transform.getBasis() * xaxis, radius, radius, 0, SIMD_2_PI, color, false, btScalar(10.0));
+		drawArc(start + transform.getBasis() * (offsetHeight), transform.getBasis() * yaxis, transform.getBasis() * xaxis, radius, radius, 0, SIMD_2_PI, color, false, btScalar(10.0));
 	}
 
 	virtual void drawCone(btScalar radius, btScalar height, int upAxis, const btTransform& transform, const btVector3& color)
@@ -423,50 +416,49 @@ class	btIDebugDraw
 		int stepDegrees = 30;
 		btVector3 start = transform.getOrigin();
 
-		btVector3	offsetHeight(0,0,0);
+		btVector3 offsetHeight(0, 0, 0);
 		btScalar halfHeight = height * btScalar(0.5);
 		offsetHeight[upAxis] = halfHeight;
-		btVector3	offsetRadius(0,0,0);
-		offsetRadius[(upAxis+1)%3] = radius;
-		btVector3	offset2Radius(0,0,0);
-		offset2Radius[(upAxis+2)%3] = radius;
+		btVector3 offsetRadius(0, 0, 0);
+		offsetRadius[(upAxis + 1) % 3] = radius;
+		btVector3 offset2Radius(0, 0, 0);
+		offset2Radius[(upAxis + 2) % 3] = radius;
 
-
-		btVector3 capEnd(0.f,0.f,0.f);
+		btVector3 capEnd(0.f, 0.f, 0.f);
 		capEnd[upAxis] = -halfHeight;
 
-		for (int i=0;i<360;i+=stepDegrees)
+		for (int i = 0; i < 360; i += stepDegrees)
 		{
-			capEnd[(upAxis+1)%3] = btSin(btScalar(i)*SIMD_RADS_PER_DEG)*radius;
-			capEnd[(upAxis+2)%3] = btCos(btScalar(i)*SIMD_RADS_PER_DEG)*radius;
-			drawLine(start+transform.getBasis() * (offsetHeight),start+transform.getBasis() * capEnd, color);
+			capEnd[(upAxis + 1) % 3] = btSin(btScalar(i) * SIMD_RADS_PER_DEG) * radius;
+			capEnd[(upAxis + 2) % 3] = btCos(btScalar(i) * SIMD_RADS_PER_DEG) * radius;
+			drawLine(start + transform.getBasis() * (offsetHeight), start + transform.getBasis() * capEnd, color);
 		}
 
-		drawLine(start+transform.getBasis() * (offsetHeight),start+transform.getBasis() * (-offsetHeight+offsetRadius),color);
-		drawLine(start+transform.getBasis() * (offsetHeight),start+transform.getBasis() * (-offsetHeight-offsetRadius),color);
-		drawLine(start+transform.getBasis() * (offsetHeight),start+transform.getBasis() * (-offsetHeight+offset2Radius),color);
-		drawLine(start+transform.getBasis() * (offsetHeight),start+transform.getBasis() * (-offsetHeight-offset2Radius),color);
+		drawLine(start + transform.getBasis() * (offsetHeight), start + transform.getBasis() * (-offsetHeight + offsetRadius), color);
+		drawLine(start + transform.getBasis() * (offsetHeight), start + transform.getBasis() * (-offsetHeight - offsetRadius), color);
+		drawLine(start + transform.getBasis() * (offsetHeight), start + transform.getBasis() * (-offsetHeight + offset2Radius), color);
+		drawLine(start + transform.getBasis() * (offsetHeight), start + transform.getBasis() * (-offsetHeight - offset2Radius), color);
 
 		// Drawing the base of the cone
-		btVector3 yaxis(0,0,0);
+		btVector3 yaxis(0, 0, 0);
 		yaxis[upAxis] = btScalar(1.0);
-		btVector3 xaxis(0,0,0);
-		xaxis[(upAxis+1)%3] = btScalar(1.0);
-		drawArc(start-transform.getBasis()*(offsetHeight),transform.getBasis()*yaxis,transform.getBasis()*xaxis,radius,radius,0,SIMD_2_PI,color,false,10.0);
+		btVector3 xaxis(0, 0, 0);
+		xaxis[(upAxis + 1) % 3] = btScalar(1.0);
+		drawArc(start - transform.getBasis() * (offsetHeight), transform.getBasis() * yaxis, transform.getBasis() * xaxis, radius, radius, 0, SIMD_2_PI, color, false, 10.0);
 	}
 
 	virtual void drawPlane(const btVector3& planeNormal, btScalar planeConst, const btTransform& transform, const btVector3& color)
 	{
 		btVector3 planeOrigin = planeNormal * planeConst;
-		btVector3 vec0,vec1;
-		btPlaneSpace1(planeNormal,vec0,vec1);
+		btVector3 vec0, vec1;
+		btPlaneSpace1(planeNormal, vec0, vec1);
 		btScalar vecLen = 100.f;
-		btVector3 pt0 = planeOrigin + vec0*vecLen;
-		btVector3 pt1 = planeOrigin - vec0*vecLen;
-		btVector3 pt2 = planeOrigin + vec1*vecLen;
-		btVector3 pt3 = planeOrigin - vec1*vecLen;
-		drawLine(transform*pt0,transform*pt1,color);
-		drawLine(transform*pt2,transform*pt3,color);
+		btVector3 pt0 = planeOrigin + vec0 * vecLen;
+		btVector3 pt1 = planeOrigin - vec0 * vecLen;
+		btVector3 pt2 = planeOrigin + vec1 * vecLen;
+		btVector3 pt3 = planeOrigin - vec1 * vecLen;
+		drawLine(transform * pt0, transform * pt1, color);
+		drawLine(transform * pt2, transform * pt3, color);
 	}
 
 	virtual void clearLines()
@@ -478,6 +470,4 @@ class	btIDebugDraw
 	}
 };
 
-
-#endif //BT_IDEBUG_DRAW__H
-
+#endif  //BT_IDEBUG_DRAW__H
diff --git a/thirdparty/bullet/LinearMath/btList.h b/thirdparty/bullet/LinearMath/btList.h
index eec80a7064b..b255938c30b 100644
--- a/thirdparty/bullet/LinearMath/btList.h
+++ b/thirdparty/bullet/LinearMath/btList.h
@@ -12,62 +12,62 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #ifndef BT_GEN_LIST_H
 #define BT_GEN_LIST_H
 
-class btGEN_Link {
+class btGEN_Link
+{
 public:
-    btGEN_Link() : m_next(0), m_prev(0) {}
-    btGEN_Link(btGEN_Link *next, btGEN_Link *prev) : m_next(next), m_prev(prev) {}
-    
-    btGEN_Link *getNext() const { return m_next; }  
-    btGEN_Link *getPrev() const { return m_prev; }  
+	btGEN_Link() : m_next(0), m_prev(0) {}
+	btGEN_Link(btGEN_Link *next, btGEN_Link *prev) : m_next(next), m_prev(prev) {}
 
-    bool isHead() const { return m_prev == 0; }
-    bool isTail() const { return m_next == 0; }
+	btGEN_Link *getNext() const { return m_next; }
+	btGEN_Link *getPrev() const { return m_prev; }
 
-    void insertBefore(btGEN_Link *link) {
-        m_next         = link;
-        m_prev         = link->m_prev;
-        m_next->m_prev = this;
-        m_prev->m_next = this;
-    } 
+	bool isHead() const { return m_prev == 0; }
+	bool isTail() const { return m_next == 0; }
 
-    void insertAfter(btGEN_Link *link) {
-        m_next         = link->m_next;
-        m_prev         = link;
-        m_next->m_prev = this;
-        m_prev->m_next = this;
-    } 
+	void insertBefore(btGEN_Link *link)
+	{
+		m_next = link;
+		m_prev = link->m_prev;
+		m_next->m_prev = this;
+		m_prev->m_next = this;
+	}
 
-    void remove() { 
-        m_next->m_prev = m_prev; 
-        m_prev->m_next = m_next;
-    }
+	void insertAfter(btGEN_Link *link)
+	{
+		m_next = link->m_next;
+		m_prev = link;
+		m_next->m_prev = this;
+		m_prev->m_next = this;
+	}
 
-private:  
-    btGEN_Link  *m_next;
-    btGEN_Link  *m_prev;
-};
+	void remove()
+	{
+		m_next->m_prev = m_prev;
+		m_prev->m_next = m_next;
+	}
 
-class btGEN_List {
-public:
-    btGEN_List() : m_head(&m_tail, 0), m_tail(0, &m_head) {}
-
-    btGEN_Link *getHead() const { return m_head.getNext(); } 
-    btGEN_Link *getTail() const { return m_tail.getPrev(); } 
-
-    void addHead(btGEN_Link *link) { link->insertAfter(&m_head); }
-    void addTail(btGEN_Link *link) { link->insertBefore(&m_tail); }
-    
 private:
-    btGEN_Link m_head;
-    btGEN_Link m_tail;
+	btGEN_Link *m_next;
+	btGEN_Link *m_prev;
 };
 
-#endif //BT_GEN_LIST_H
+class btGEN_List
+{
+public:
+	btGEN_List() : m_head(&m_tail, 0), m_tail(0, &m_head) {}
 
+	btGEN_Link *getHead() const { return m_head.getNext(); }
+	btGEN_Link *getTail() const { return m_tail.getPrev(); }
 
+	void addHead(btGEN_Link *link) { link->insertAfter(&m_head); }
+	void addTail(btGEN_Link *link) { link->insertBefore(&m_tail); }
 
+private:
+	btGEN_Link m_head;
+	btGEN_Link m_tail;
+};
+
+#endif  //BT_GEN_LIST_H
diff --git a/thirdparty/bullet/LinearMath/btMatrix3x3.h b/thirdparty/bullet/LinearMath/btMatrix3x3.h
index 6cc4993da56..0a08ae409a1 100644
--- a/thirdparty/bullet/LinearMath/btMatrix3x3.h
+++ b/thirdparty/bullet/LinearMath/btMatrix3x3.h
@@ -12,8 +12,7 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-#ifndef	BT_MATRIX3x3_H
+#ifndef BT_MATRIX3x3_H
 #define BT_MATRIX3x3_H
 
 #include "btVector3.h"
@@ -23,13 +22,13 @@ subject to the following restrictions:
 #ifdef BT_USE_SSE
 //const __m128 ATTRIBUTE_ALIGNED16(v2220) = {2.0f, 2.0f, 2.0f, 0.0f};
 //const __m128 ATTRIBUTE_ALIGNED16(vMPPP) = {-0.0f, +0.0f, +0.0f, +0.0f};
-#define vMPPP (_mm_set_ps (+0.0f, +0.0f, +0.0f, -0.0f))
+#define vMPPP (_mm_set_ps(+0.0f, +0.0f, +0.0f, -0.0f))
 #endif
 
 #if defined(BT_USE_SSE)
-#define v1000 (_mm_set_ps(0.0f,0.0f,0.0f,1.0f))
-#define v0100 (_mm_set_ps(0.0f,0.0f,1.0f,0.0f))
-#define v0010 (_mm_set_ps(0.0f,1.0f,0.0f,0.0f))
+#define v1000 (_mm_set_ps(0.0f, 0.0f, 0.0f, 1.0f))
+#define v0100 (_mm_set_ps(0.0f, 0.0f, 1.0f, 0.0f))
+#define v0010 (_mm_set_ps(0.0f, 1.0f, 0.0f, 0.0f))
 #elif defined(BT_USE_NEON)
 const btSimdFloat4 ATTRIBUTE_ALIGNED16(v1000) = {1.0f, 0.0f, 0.0f, 0.0f};
 const btSimdFloat4 ATTRIBUTE_ALIGNED16(v0100) = {0.0f, 1.0f, 0.0f, 0.0f};
@@ -37,22 +36,22 @@ const btSimdFloat4 ATTRIBUTE_ALIGNED16(v0010) = {0.0f, 0.0f, 1.0f, 0.0f};
 #endif
 
 #ifdef BT_USE_DOUBLE_PRECISION
-#define btMatrix3x3Data	btMatrix3x3DoubleData 
+#define btMatrix3x3Data btMatrix3x3DoubleData
 #else
-#define btMatrix3x3Data	btMatrix3x3FloatData
-#endif //BT_USE_DOUBLE_PRECISION
-
+#define btMatrix3x3Data btMatrix3x3FloatData
+#endif  //BT_USE_DOUBLE_PRECISION
 
 /**@brief The btMatrix3x3 class implements a 3x3 rotation matrix, to perform linear algebra in combination with btQuaternion, btTransform and btVector3.
 * Make sure to only include a pure orthogonal matrix without scaling. */
-ATTRIBUTE_ALIGNED16(class) btMatrix3x3 {
-
+ATTRIBUTE_ALIGNED16(class)
+btMatrix3x3
+{
 	///Data storage for the matrix, each vector is a row of the matrix
 	btVector3 m_el[3];
 
 public:
 	/** @brief No initializaion constructor */
-	btMatrix3x3 () {}
+	btMatrix3x3() {}
 
 	//		explicit btMatrix3x3(const btScalar *m) { setFromOpenGLSubMatrix(m); }
 
@@ -67,27 +66,27 @@ public:
 	*/
 	/** @brief Constructor with row major formatting */
 	btMatrix3x3(const btScalar& xx, const btScalar& xy, const btScalar& xz,
-		const btScalar& yx, const btScalar& yy, const btScalar& yz,
-		const btScalar& zx, const btScalar& zy, const btScalar& zz)
-	{ 
-		setValue(xx, xy, xz, 
-			yx, yy, yz, 
-			zx, zy, zz);
+				const btScalar& yx, const btScalar& yy, const btScalar& yz,
+				const btScalar& zx, const btScalar& zy, const btScalar& zz)
+	{
+		setValue(xx, xy, xz,
+				 yx, yy, yz,
+				 zx, zy, zz);
 	}
 
-#if (defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE))|| defined (BT_USE_NEON)
-	SIMD_FORCE_INLINE btMatrix3x3 (const btSimdFloat4 v0, const btSimdFloat4 v1, const btSimdFloat4 v2 ) 
+#if (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)) || defined(BT_USE_NEON)
+	SIMD_FORCE_INLINE btMatrix3x3(const btSimdFloat4 v0, const btSimdFloat4 v1, const btSimdFloat4 v2)
 	{
-        m_el[0].mVec128 = v0;
-        m_el[1].mVec128 = v1;
-        m_el[2].mVec128 = v2;
+		m_el[0].mVec128 = v0;
+		m_el[1].mVec128 = v1;
+		m_el[2].mVec128 = v2;
 	}
 
-	SIMD_FORCE_INLINE btMatrix3x3 (const btVector3& v0, const btVector3& v1, const btVector3& v2 ) 
+	SIMD_FORCE_INLINE btMatrix3x3(const btVector3& v0, const btVector3& v1, const btVector3& v2)
 	{
-        m_el[0] = v0;
-        m_el[1] = v1;
-        m_el[2] = v2;
+		m_el[0] = v0;
+		m_el[1] = v1;
+		m_el[2] = v2;
 	}
 
 	// Copy constructor
@@ -99,25 +98,25 @@ public:
 	}
 
 	// Assignment Operator
-	SIMD_FORCE_INLINE btMatrix3x3& operator=(const btMatrix3x3& m) 
+	SIMD_FORCE_INLINE btMatrix3x3& operator=(const btMatrix3x3& m)
 	{
 		m_el[0].mVec128 = m.m_el[0].mVec128;
 		m_el[1].mVec128 = m.m_el[1].mVec128;
 		m_el[2].mVec128 = m.m_el[2].mVec128;
-		
+
 		return *this;
 	}
 
 #else
 
 	/** @brief Copy constructor */
-	SIMD_FORCE_INLINE btMatrix3x3 (const btMatrix3x3& other)
+	SIMD_FORCE_INLINE btMatrix3x3(const btMatrix3x3& other)
 	{
 		m_el[0] = other.m_el[0];
 		m_el[1] = other.m_el[1];
 		m_el[2] = other.m_el[2];
 	}
-    
+
 	/** @brief Assignment Operator */
 	SIMD_FORCE_INLINE btMatrix3x3& operator=(const btMatrix3x3& other)
 	{
@@ -133,10 +132,9 @@ public:
 	*  @param i Column number 0 indexed */
 	SIMD_FORCE_INLINE btVector3 getColumn(int i) const
 	{
-		return btVector3(m_el[0][i],m_el[1][i],m_el[2][i]);
+		return btVector3(m_el[0][i], m_el[1][i], m_el[2][i]);
 	}
 
-
 	/** @brief Get a row of the matrix as a vector 
 	*  @param i Row number 0 indexed */
 	SIMD_FORCE_INLINE const btVector3& getRow(int i) const
@@ -147,10 +145,10 @@ public:
 
 	/** @brief Get a mutable reference to a row of the matrix as a vector 
 	*  @param i Row number 0 indexed */
-	SIMD_FORCE_INLINE btVector3&  operator[](int i)
-	{ 
+	SIMD_FORCE_INLINE btVector3& operator[](int i)
+	{
 		btFullAssert(0 <= i && i < 3);
-		return m_el[i]; 
+		return m_el[i];
 	}
 
 	/** @brief Get a const reference to a row of the matrix as a vector 
@@ -158,32 +156,31 @@ public:
 	SIMD_FORCE_INLINE const btVector3& operator[](int i) const
 	{
 		btFullAssert(0 <= i && i < 3);
-		return m_el[i]; 
+		return m_el[i];
 	}
 
 	/** @brief Multiply by the target matrix on the right
 	*  @param m Rotation matrix to be applied 
 	* Equivilant to this = this * m */
-	btMatrix3x3& operator*=(const btMatrix3x3& m); 
+	btMatrix3x3& operator*=(const btMatrix3x3& m);
 
 	/** @brief Adds by the target matrix on the right
 	*  @param m matrix to be applied 
 	* Equivilant to this = this + m */
-	btMatrix3x3& operator+=(const btMatrix3x3& m); 
+	btMatrix3x3& operator+=(const btMatrix3x3& m);
 
 	/** @brief Substractss by the target matrix on the right
 	*  @param m matrix to be applied 
 	* Equivilant to this = this - m */
-	btMatrix3x3& operator-=(const btMatrix3x3& m); 
+	btMatrix3x3& operator-=(const btMatrix3x3& m);
 
 	/** @brief Set from the rotational part of a 4x4 OpenGL matrix
 	*  @param m A pointer to the beginning of the array of scalars*/
-	void setFromOpenGLSubMatrix(const btScalar *m)
+	void setFromOpenGLSubMatrix(const btScalar* m)
 	{
-		m_el[0].setValue(m[0],m[4],m[8]);
-		m_el[1].setValue(m[1],m[5],m[9]);
-		m_el[2].setValue(m[2],m[6],m[10]);
-
+		m_el[0].setValue(m[0], m[4], m[8]);
+		m_el[1].setValue(m[1], m[5], m[9]);
+		m_el[2].setValue(m[2], m[6], m[10]);
 	}
 	/** @brief Set the values of the matrix explicitly (row major)
 	*  @param xx Top left
@@ -195,93 +192,92 @@ public:
 	*  @param zx Bottom Left
 	*  @param zy Bottom Middle
 	*  @param zz Bottom Right*/
-	void setValue(const btScalar& xx, const btScalar& xy, const btScalar& xz, 
-		const btScalar& yx, const btScalar& yy, const btScalar& yz, 
-		const btScalar& zx, const btScalar& zy, const btScalar& zz)
+	void setValue(const btScalar& xx, const btScalar& xy, const btScalar& xz,
+				  const btScalar& yx, const btScalar& yy, const btScalar& yz,
+				  const btScalar& zx, const btScalar& zy, const btScalar& zz)
 	{
-		m_el[0].setValue(xx,xy,xz);
-		m_el[1].setValue(yx,yy,yz);
-		m_el[2].setValue(zx,zy,zz);
+		m_el[0].setValue(xx, xy, xz);
+		m_el[1].setValue(yx, yy, yz);
+		m_el[2].setValue(zx, zy, zz);
 	}
 
 	/** @brief Set the matrix from a quaternion
-	*  @param q The Quaternion to match */  
-	void setRotation(const btQuaternion& q) 
+	*  @param q The Quaternion to match */
+	void setRotation(const btQuaternion& q)
 	{
 		btScalar d = q.length2();
 		btFullAssert(d != btScalar(0.0));
 		btScalar s = btScalar(2.0) / d;
-    
-    #if defined BT_USE_SIMD_VECTOR3 && defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
-        __m128	vs, Q = q.get128();
+
+#if defined BT_USE_SIMD_VECTOR3 && defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
+		__m128 vs, Q = q.get128();
 		__m128i Qi = btCastfTo128i(Q);
-        __m128	Y, Z;
-        __m128	V1, V2, V3;
-        __m128	V11, V21, V31;
-        __m128	NQ = _mm_xor_ps(Q, btvMzeroMask);
+		__m128 Y, Z;
+		__m128 V1, V2, V3;
+		__m128 V11, V21, V31;
+		__m128 NQ = _mm_xor_ps(Q, btvMzeroMask);
 		__m128i NQi = btCastfTo128i(NQ);
-        
-        V1 = btCastiTo128f(_mm_shuffle_epi32 (Qi, BT_SHUFFLE(1,0,2,3)));	// Y X Z W
-		V2 = _mm_shuffle_ps(NQ, Q, BT_SHUFFLE(0,0,1,3));     // -X -X  Y  W
-        V3 = btCastiTo128f(_mm_shuffle_epi32 (Qi, BT_SHUFFLE(2,1,0,3)));	// Z Y X W
-        V1 = _mm_xor_ps(V1, vMPPP);	//	change the sign of the first element
-			
-        V11	= btCastiTo128f(_mm_shuffle_epi32 (Qi, BT_SHUFFLE(1,1,0,3)));	// Y Y X W
-		V21 = _mm_unpackhi_ps(Q, Q);                    //  Z  Z  W  W
-		V31 = _mm_shuffle_ps(Q, NQ, BT_SHUFFLE(0,2,0,3));	//  X  Z -X -W
 
-		V2 = V2 * V1;	//
-		V1 = V1 * V11;	//
-		V3 = V3 * V31;	//
+		V1 = btCastiTo128f(_mm_shuffle_epi32(Qi, BT_SHUFFLE(1, 0, 2, 3)));  // Y X Z W
+		V2 = _mm_shuffle_ps(NQ, Q, BT_SHUFFLE(0, 0, 1, 3));                 // -X -X  Y  W
+		V3 = btCastiTo128f(_mm_shuffle_epi32(Qi, BT_SHUFFLE(2, 1, 0, 3)));  // Z Y X W
+		V1 = _mm_xor_ps(V1, vMPPP);                                         //	change the sign of the first element
 
-        V11 = _mm_shuffle_ps(NQ, Q, BT_SHUFFLE(2,3,1,3));	//	-Z -W  Y  W
-		V11 = V11 * V21;	//
-        V21 = _mm_xor_ps(V21, vMPPP);	//	change the sign of the first element
-		V31 = _mm_shuffle_ps(Q, NQ, BT_SHUFFLE(3,3,1,3));	//	 W  W -Y -W
-        V31 = _mm_xor_ps(V31, vMPPP);	//	change the sign of the first element
-		Y = btCastiTo128f(_mm_shuffle_epi32 (NQi, BT_SHUFFLE(3,2,0,3)));	// -W -Z -X -W
-		Z = btCastiTo128f(_mm_shuffle_epi32 (Qi, BT_SHUFFLE(1,0,1,3)));	//  Y  X  Y  W
+		V11 = btCastiTo128f(_mm_shuffle_epi32(Qi, BT_SHUFFLE(1, 1, 0, 3)));  // Y Y X W
+		V21 = _mm_unpackhi_ps(Q, Q);                                         //  Z  Z  W  W
+		V31 = _mm_shuffle_ps(Q, NQ, BT_SHUFFLE(0, 2, 0, 3));                 //  X  Z -X -W
+
+		V2 = V2 * V1;   //
+		V1 = V1 * V11;  //
+		V3 = V3 * V31;  //
+
+		V11 = _mm_shuffle_ps(NQ, Q, BT_SHUFFLE(2, 3, 1, 3));                //	-Z -W  Y  W
+		V11 = V11 * V21;                                                    //
+		V21 = _mm_xor_ps(V21, vMPPP);                                       //	change the sign of the first element
+		V31 = _mm_shuffle_ps(Q, NQ, BT_SHUFFLE(3, 3, 1, 3));                //	 W  W -Y -W
+		V31 = _mm_xor_ps(V31, vMPPP);                                       //	change the sign of the first element
+		Y = btCastiTo128f(_mm_shuffle_epi32(NQi, BT_SHUFFLE(3, 2, 0, 3)));  // -W -Z -X -W
+		Z = btCastiTo128f(_mm_shuffle_epi32(Qi, BT_SHUFFLE(1, 0, 1, 3)));   //  Y  X  Y  W
 
 		vs = _mm_load_ss(&s);
 		V21 = V21 * Y;
 		V31 = V31 * Z;
 
 		V1 = V1 + V11;
-        V2 = V2 + V21;
-        V3 = V3 + V31;
+		V2 = V2 + V21;
+		V3 = V3 + V31;
 
-        vs = bt_splat3_ps(vs, 0);
-            //	s ready
-        V1 = V1 * vs;
-        V2 = V2 * vs;
-        V3 = V3 * vs;
-        
-        V1 = V1 + v1000;
-        V2 = V2 + v0100;
-        V3 = V3 + v0010;
-        
-        m_el[0] = V1; 
-        m_el[1] = V2;
-        m_el[2] = V3;
-    #else    
-		btScalar xs = q.x() * s,   ys = q.y() * s,   zs = q.z() * s;
-		btScalar wx = q.w() * xs,  wy = q.w() * ys,  wz = q.w() * zs;
-		btScalar xx = q.x() * xs,  xy = q.x() * ys,  xz = q.x() * zs;
-		btScalar yy = q.y() * ys,  yz = q.y() * zs,  zz = q.z() * zs;
+		vs = bt_splat3_ps(vs, 0);
+		//	s ready
+		V1 = V1 * vs;
+		V2 = V2 * vs;
+		V3 = V3 * vs;
+
+		V1 = V1 + v1000;
+		V2 = V2 + v0100;
+		V3 = V3 + v0010;
+
+		m_el[0] = V1;
+		m_el[1] = V2;
+		m_el[2] = V3;
+#else
+		btScalar xs = q.x() * s, ys = q.y() * s, zs = q.z() * s;
+		btScalar wx = q.w() * xs, wy = q.w() * ys, wz = q.w() * zs;
+		btScalar xx = q.x() * xs, xy = q.x() * ys, xz = q.x() * zs;
+		btScalar yy = q.y() * ys, yz = q.y() * zs, zz = q.z() * zs;
 		setValue(
-            btScalar(1.0) - (yy + zz), xy - wz, xz + wy,
+			btScalar(1.0) - (yy + zz), xy - wz, xz + wy,
 			xy + wz, btScalar(1.0) - (xx + zz), yz - wx,
 			xz - wy, yz + wx, btScalar(1.0) - (xx + yy));
-	#endif
-    }
-
+#endif
+	}
 
 	/** @brief Set the matrix from euler angles using YPR around YXZ respectively
 	*  @param yaw Yaw about Y axis
 	*  @param pitch Pitch about X axis
 	*  @param roll Roll about Z axis 
 	*/
-	void setEulerYPR(const btScalar& yaw, const btScalar& pitch, const btScalar& roll) 
+	void setEulerYPR(const btScalar& yaw, const btScalar& pitch, const btScalar& roll)
 	{
 		setEulerZYX(roll, pitch, yaw);
 	}
@@ -295,182 +291,197 @@ public:
 	* angles are applied in ZYX order. I.e a vector is first rotated 
 	* about X then Y and then Z
 	**/
-	void setEulerZYX(btScalar eulerX,btScalar eulerY,btScalar eulerZ) { 
+	void setEulerZYX(btScalar eulerX, btScalar eulerY, btScalar eulerZ)
+	{
 		///@todo proposed to reverse this since it's labeled zyx but takes arguments xyz and it will match all other parts of the code
-		btScalar ci ( btCos(eulerX)); 
-		btScalar cj ( btCos(eulerY)); 
-		btScalar ch ( btCos(eulerZ)); 
-		btScalar si ( btSin(eulerX)); 
-		btScalar sj ( btSin(eulerY)); 
-		btScalar sh ( btSin(eulerZ)); 
-		btScalar cc = ci * ch; 
-		btScalar cs = ci * sh; 
-		btScalar sc = si * ch; 
+		btScalar ci(btCos(eulerX));
+		btScalar cj(btCos(eulerY));
+		btScalar ch(btCos(eulerZ));
+		btScalar si(btSin(eulerX));
+		btScalar sj(btSin(eulerY));
+		btScalar sh(btSin(eulerZ));
+		btScalar cc = ci * ch;
+		btScalar cs = ci * sh;
+		btScalar sc = si * ch;
 		btScalar ss = si * sh;
 
 		setValue(cj * ch, sj * sc - cs, sj * cc + ss,
-			cj * sh, sj * ss + cc, sj * cs - sc, 
-			-sj,      cj * si,      cj * ci);
+				 cj * sh, sj * ss + cc, sj * cs - sc,
+				 -sj, cj * si, cj * ci);
 	}
 
 	/**@brief Set the matrix to the identity */
 	void setIdentity()
-	{ 
-#if (defined(BT_USE_SSE_IN_API)&& defined (BT_USE_SSE)) || defined(BT_USE_NEON)
-			m_el[0] = v1000; 
-			m_el[1] = v0100;
-			m_el[2] = v0010;
+	{
+#if (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)) || defined(BT_USE_NEON)
+		m_el[0] = v1000;
+		m_el[1] = v0100;
+		m_el[2] = v0010;
 #else
-		setValue(btScalar(1.0), btScalar(0.0), btScalar(0.0), 
-			btScalar(0.0), btScalar(1.0), btScalar(0.0), 
-			btScalar(0.0), btScalar(0.0), btScalar(1.0)); 
+		setValue(btScalar(1.0), btScalar(0.0), btScalar(0.0),
+				 btScalar(0.0), btScalar(1.0), btScalar(0.0),
+				 btScalar(0.0), btScalar(0.0), btScalar(1.0));
 #endif
 	}
 
-	static const btMatrix3x3&	getIdentity()
+	static const btMatrix3x3& getIdentity()
 	{
-#if (defined(BT_USE_SSE_IN_API)&& defined (BT_USE_SSE)) || defined(BT_USE_NEON)
-        static const btMatrix3x3 
-        identityMatrix(v1000, v0100, v0010);
+#if (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)) || defined(BT_USE_NEON)
+		static const btMatrix3x3
+			identityMatrix(v1000, v0100, v0010);
 #else
-		static const btMatrix3x3 
-        identityMatrix(
-            btScalar(1.0), btScalar(0.0), btScalar(0.0), 
-			btScalar(0.0), btScalar(1.0), btScalar(0.0), 
-			btScalar(0.0), btScalar(0.0), btScalar(1.0));
+		static const btMatrix3x3
+			identityMatrix(
+				btScalar(1.0), btScalar(0.0), btScalar(0.0),
+				btScalar(0.0), btScalar(1.0), btScalar(0.0),
+				btScalar(0.0), btScalar(0.0), btScalar(1.0));
 #endif
 		return identityMatrix;
 	}
 
 	/**@brief Fill the rotational part of an OpenGL matrix and clear the shear/perspective
 	* @param m The array to be filled */
-	void getOpenGLSubMatrix(btScalar *m) const 
+	void getOpenGLSubMatrix(btScalar * m) const
 	{
-#if defined BT_USE_SIMD_VECTOR3 && defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
-        __m128 v0 = m_el[0].mVec128;
-        __m128 v1 = m_el[1].mVec128;
-        __m128 v2 = m_el[2].mVec128;    //  x2 y2 z2 w2
-        __m128 *vm = (__m128 *)m;
-        __m128 vT;
-        
-        v2 = _mm_and_ps(v2, btvFFF0fMask);  //  x2 y2 z2 0
-        
-        vT = _mm_unpackhi_ps(v0, v1);	//	z0 z1 * *
-        v0 = _mm_unpacklo_ps(v0, v1);	//	x0 x1 y0 y1
+#if defined BT_USE_SIMD_VECTOR3 && defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
+		__m128 v0 = m_el[0].mVec128;
+		__m128 v1 = m_el[1].mVec128;
+		__m128 v2 = m_el[2].mVec128;  //  x2 y2 z2 w2
+		__m128* vm = (__m128*)m;
+		__m128 vT;
 
-        v1 = _mm_shuffle_ps(v0, v2, BT_SHUFFLE(2, 3, 1, 3) );	// y0 y1 y2 0
-        v0 = _mm_shuffle_ps(v0, v2, BT_SHUFFLE(0, 1, 0, 3) );	// x0 x1 x2 0
-        v2 = btCastdTo128f(_mm_move_sd(btCastfTo128d(v2), btCastfTo128d(vT)));	// z0 z1 z2 0
+		v2 = _mm_and_ps(v2, btvFFF0fMask);  //  x2 y2 z2 0
 
-        vm[0] = v0;
-        vm[1] = v1;
-        vm[2] = v2;
+		vT = _mm_unpackhi_ps(v0, v1);  //	z0 z1 * *
+		v0 = _mm_unpacklo_ps(v0, v1);  //	x0 x1 y0 y1
+
+		v1 = _mm_shuffle_ps(v0, v2, BT_SHUFFLE(2, 3, 1, 3));                    // y0 y1 y2 0
+		v0 = _mm_shuffle_ps(v0, v2, BT_SHUFFLE(0, 1, 0, 3));                    // x0 x1 x2 0
+		v2 = btCastdTo128f(_mm_move_sd(btCastfTo128d(v2), btCastfTo128d(vT)));  // z0 z1 z2 0
+
+		vm[0] = v0;
+		vm[1] = v1;
+		vm[2] = v2;
 #elif defined(BT_USE_NEON)
-        // note: zeros the w channel. We can preserve it at the cost of two more vtrn instructions.
-        static const uint32x2_t zMask = (const uint32x2_t) {static_cast<uint32_t>(-1), 0 };
-        float32x4_t *vm = (float32x4_t *)m;
-        float32x4x2_t top = vtrnq_f32( m_el[0].mVec128, m_el[1].mVec128 );  // {x0 x1 z0 z1}, {y0 y1 w0 w1}
-        float32x2x2_t bl = vtrn_f32( vget_low_f32(m_el[2].mVec128), vdup_n_f32(0.0f) );       // {x2  0 }, {y2 0}
-        float32x4_t v0 = vcombine_f32( vget_low_f32(top.val[0]), bl.val[0] );
-        float32x4_t v1 = vcombine_f32( vget_low_f32(top.val[1]), bl.val[1] );
-        float32x2_t q = (float32x2_t) vand_u32( (uint32x2_t) vget_high_f32( m_el[2].mVec128), zMask );
-        float32x4_t v2 = vcombine_f32( vget_high_f32(top.val[0]), q );       // z0 z1 z2  0
+		// note: zeros the w channel. We can preserve it at the cost of two more vtrn instructions.
+		static const uint32x2_t zMask = (const uint32x2_t){static_cast<uint32_t>(-1), 0};
+		float32x4_t* vm = (float32x4_t*)m;
+		float32x4x2_t top = vtrnq_f32(m_el[0].mVec128, m_el[1].mVec128);               // {x0 x1 z0 z1}, {y0 y1 w0 w1}
+		float32x2x2_t bl = vtrn_f32(vget_low_f32(m_el[2].mVec128), vdup_n_f32(0.0f));  // {x2  0 }, {y2 0}
+		float32x4_t v0 = vcombine_f32(vget_low_f32(top.val[0]), bl.val[0]);
+		float32x4_t v1 = vcombine_f32(vget_low_f32(top.val[1]), bl.val[1]);
+		float32x2_t q = (float32x2_t)vand_u32((uint32x2_t)vget_high_f32(m_el[2].mVec128), zMask);
+		float32x4_t v2 = vcombine_f32(vget_high_f32(top.val[0]), q);  // z0 z1 z2  0
 
-        vm[0] = v0;
-        vm[1] = v1;
-        vm[2] = v2;
+		vm[0] = v0;
+		vm[1] = v1;
+		vm[2] = v2;
 #else
-		m[0]  = btScalar(m_el[0].x()); 
-		m[1]  = btScalar(m_el[1].x());
-		m[2]  = btScalar(m_el[2].x());
-		m[3]  = btScalar(0.0); 
-		m[4]  = btScalar(m_el[0].y());
-		m[5]  = btScalar(m_el[1].y());
-		m[6]  = btScalar(m_el[2].y());
-		m[7]  = btScalar(0.0); 
-		m[8]  = btScalar(m_el[0].z()); 
-		m[9]  = btScalar(m_el[1].z());
+		m[0] = btScalar(m_el[0].x());
+		m[1] = btScalar(m_el[1].x());
+		m[2] = btScalar(m_el[2].x());
+		m[3] = btScalar(0.0);
+		m[4] = btScalar(m_el[0].y());
+		m[5] = btScalar(m_el[1].y());
+		m[6] = btScalar(m_el[2].y());
+		m[7] = btScalar(0.0);
+		m[8] = btScalar(m_el[0].z());
+		m[9] = btScalar(m_el[1].z());
 		m[10] = btScalar(m_el[2].z());
-		m[11] = btScalar(0.0); 
+		m[11] = btScalar(0.0);
 #endif
 	}
 
 	/**@brief Get the matrix represented as a quaternion 
 	* @param q The quaternion which will be set */
-	void getRotation(btQuaternion& q) const
+	void getRotation(btQuaternion & q) const
 	{
-#if (defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE))|| defined (BT_USE_NEON)
-        btScalar trace = m_el[0].x() + m_el[1].y() + m_el[2].z();
-        btScalar s, x;
-        
-        union {
-            btSimdFloat4 vec;
-            btScalar f[4];
-        } temp;
-        
-        if (trace > btScalar(0.0)) 
-        {
-            x = trace + btScalar(1.0);
+#if (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)) || defined(BT_USE_NEON)
+		btScalar trace = m_el[0].x() + m_el[1].y() + m_el[2].z();
+		btScalar s, x;
 
-            temp.f[0]=m_el[2].y() - m_el[1].z();
-            temp.f[1]=m_el[0].z() - m_el[2].x();
-            temp.f[2]=m_el[1].x() - m_el[0].y();
-            temp.f[3]=x;
-            //temp.f[3]= s * btScalar(0.5);
-        } 
-        else 
-        {
-            int i, j, k;
-            if(m_el[0].x() < m_el[1].y()) 
-            { 
-                if( m_el[1].y() < m_el[2].z() )
-                    { i = 2; j = 0; k = 1; }
-                else
-                    { i = 1; j = 2; k = 0; }
-            }
-            else
-            {
-                if( m_el[0].x() < m_el[2].z())
-                    { i = 2; j = 0; k = 1; }
-                else
-                    { i = 0; j = 1; k = 2; }
-            }
+		union {
+			btSimdFloat4 vec;
+			btScalar f[4];
+		} temp;
 
-            x = m_el[i][i] - m_el[j][j] - m_el[k][k] + btScalar(1.0);
+		if (trace > btScalar(0.0))
+		{
+			x = trace + btScalar(1.0);
 
-            temp.f[3] = (m_el[k][j] - m_el[j][k]);
-            temp.f[j] = (m_el[j][i] + m_el[i][j]);
-            temp.f[k] = (m_el[k][i] + m_el[i][k]);
-            temp.f[i] = x;
-            //temp.f[i] = s * btScalar(0.5);
-        }
+			temp.f[0] = m_el[2].y() - m_el[1].z();
+			temp.f[1] = m_el[0].z() - m_el[2].x();
+			temp.f[2] = m_el[1].x() - m_el[0].y();
+			temp.f[3] = x;
+			//temp.f[3]= s * btScalar(0.5);
+		}
+		else
+		{
+			int i, j, k;
+			if (m_el[0].x() < m_el[1].y())
+			{
+				if (m_el[1].y() < m_el[2].z())
+				{
+					i = 2;
+					j = 0;
+					k = 1;
+				}
+				else
+				{
+					i = 1;
+					j = 2;
+					k = 0;
+				}
+			}
+			else
+			{
+				if (m_el[0].x() < m_el[2].z())
+				{
+					i = 2;
+					j = 0;
+					k = 1;
+				}
+				else
+				{
+					i = 0;
+					j = 1;
+					k = 2;
+				}
+			}
 
-        s = btSqrt(x);
-        q.set128(temp.vec);
-        s = btScalar(0.5) / s;
+			x = m_el[i][i] - m_el[j][j] - m_el[k][k] + btScalar(1.0);
 
-        q *= s;
-#else    
+			temp.f[3] = (m_el[k][j] - m_el[j][k]);
+			temp.f[j] = (m_el[j][i] + m_el[i][j]);
+			temp.f[k] = (m_el[k][i] + m_el[i][k]);
+			temp.f[i] = x;
+			//temp.f[i] = s * btScalar(0.5);
+		}
+
+		s = btSqrt(x);
+		q.set128(temp.vec);
+		s = btScalar(0.5) / s;
+
+		q *= s;
+#else
 		btScalar trace = m_el[0].x() + m_el[1].y() + m_el[2].z();
 
 		btScalar temp[4];
 
-		if (trace > btScalar(0.0)) 
+		if (trace > btScalar(0.0))
 		{
 			btScalar s = btSqrt(trace + btScalar(1.0));
-			temp[3]=(s * btScalar(0.5));
+			temp[3] = (s * btScalar(0.5));
 			s = btScalar(0.5) / s;
 
-			temp[0]=((m_el[2].y() - m_el[1].z()) * s);
-			temp[1]=((m_el[0].z() - m_el[2].x()) * s);
-			temp[2]=((m_el[1].x() - m_el[0].y()) * s);
-		} 
-		else 
+			temp[0] = ((m_el[2].y() - m_el[1].z()) * s);
+			temp[1] = ((m_el[0].z() - m_el[2].x()) * s);
+			temp[2] = ((m_el[1].x() - m_el[0].y()) * s);
+		}
+		else
 		{
-			int i = m_el[0].x() < m_el[1].y() ? 
-				(m_el[1].y() < m_el[2].z() ? 2 : 1) :
-				(m_el[0].x() < m_el[2].z() ? 2 : 0); 
-			int j = (i + 1) % 3;  
+			int i = m_el[0].x() < m_el[1].y() ? (m_el[1].y() < m_el[2].z() ? 2 : 1) : (m_el[0].x() < m_el[2].z() ? 2 : 0);
+			int j = (i + 1) % 3;
 			int k = (i + 2) % 3;
 
 			btScalar s = btSqrt(m_el[i][i] - m_el[j][j] - m_el[k][k] + btScalar(1.0));
@@ -481,44 +492,42 @@ public:
 			temp[j] = (m_el[j][i] + m_el[i][j]) * s;
 			temp[k] = (m_el[k][i] + m_el[i][k]) * s;
 		}
-		q.setValue(temp[0],temp[1],temp[2],temp[3]);
+		q.setValue(temp[0], temp[1], temp[2], temp[3]);
 #endif
 	}
 
 	/**@brief Get the matrix represented as euler angles around YXZ, roundtrip with setEulerYPR
 	* @param yaw Yaw around Y axis
 	* @param pitch Pitch around X axis
-	* @param roll around Z axis */	
-	void getEulerYPR(btScalar& yaw, btScalar& pitch, btScalar& roll) const
+	* @param roll around Z axis */
+	void getEulerYPR(btScalar & yaw, btScalar & pitch, btScalar & roll) const
 	{
-
 		// first use the normal calculus
 		yaw = btScalar(btAtan2(m_el[1].x(), m_el[0].x()));
 		pitch = btScalar(btAsin(-m_el[2].x()));
 		roll = btScalar(btAtan2(m_el[2].y(), m_el[2].z()));
 
 		// on pitch = +/-HalfPI
-		if (btFabs(pitch)==SIMD_HALF_PI)
+		if (btFabs(pitch) == SIMD_HALF_PI)
 		{
-			if (yaw>0)
-				yaw-=SIMD_PI;
+			if (yaw > 0)
+				yaw -= SIMD_PI;
 			else
-				yaw+=SIMD_PI;
+				yaw += SIMD_PI;
 
-			if (roll>0)
-				roll-=SIMD_PI;
+			if (roll > 0)
+				roll -= SIMD_PI;
 			else
-				roll+=SIMD_PI;
+				roll += SIMD_PI;
 		}
 	};
 
-
 	/**@brief Get the matrix represented as euler angles around ZYX
 	* @param yaw Yaw around Z axis
 	* @param pitch Pitch around Y axis
 	* @param roll around X axis 
-	* @param solution_number Which solution of two possible solutions ( 1 or 2) are possible values*/	
-	void getEulerZYX(btScalar& yaw, btScalar& pitch, btScalar& roll, unsigned int solution_number = 1) const
+	* @param solution_number Which solution of two possible solutions ( 1 or 2) are possible values*/
+	void getEulerZYX(btScalar & yaw, btScalar & pitch, btScalar & roll, unsigned int solution_number = 1) const
 	{
 		struct Euler
 		{
@@ -528,7 +537,7 @@ public:
 		};
 
 		Euler euler_out;
-		Euler euler_out2; //second solution
+		Euler euler_out2;  //second solution
 		//get the pointer to the raw data
 
 		// Check that pitch is not at a singularity
@@ -538,7 +547,7 @@ public:
 			euler_out2.yaw = 0;
 
 			// From difference of angles formula
-			btScalar delta = btAtan2(m_el[0].x(),m_el[0].z());
+			btScalar delta = btAtan2(m_el[0].x(), m_el[0].z());
 			if (m_el[2].x() > 0)  //gimbal locked up
 			{
 				euler_out.pitch = SIMD_PI / btScalar(2.0);
@@ -546,7 +555,7 @@ public:
 				euler_out.roll = euler_out.pitch + delta;
 				euler_out2.roll = euler_out.pitch + delta;
 			}
-			else // gimbal locked down
+			else  // gimbal locked down
 			{
 				euler_out.pitch = -SIMD_PI / btScalar(2.0);
 				euler_out2.pitch = -SIMD_PI / btScalar(2.0);
@@ -556,29 +565,29 @@ public:
 		}
 		else
 		{
-			euler_out.pitch = - btAsin(m_el[2].x());
+			euler_out.pitch = -btAsin(m_el[2].x());
 			euler_out2.pitch = SIMD_PI - euler_out.pitch;
 
-			euler_out.roll = btAtan2(m_el[2].y()/btCos(euler_out.pitch), 
-				m_el[2].z()/btCos(euler_out.pitch));
-			euler_out2.roll = btAtan2(m_el[2].y()/btCos(euler_out2.pitch), 
-				m_el[2].z()/btCos(euler_out2.pitch));
+			euler_out.roll = btAtan2(m_el[2].y() / btCos(euler_out.pitch),
+									 m_el[2].z() / btCos(euler_out.pitch));
+			euler_out2.roll = btAtan2(m_el[2].y() / btCos(euler_out2.pitch),
+									  m_el[2].z() / btCos(euler_out2.pitch));
 
-			euler_out.yaw = btAtan2(m_el[1].x()/btCos(euler_out.pitch), 
-				m_el[0].x()/btCos(euler_out.pitch));
-			euler_out2.yaw = btAtan2(m_el[1].x()/btCos(euler_out2.pitch), 
-				m_el[0].x()/btCos(euler_out2.pitch));
+			euler_out.yaw = btAtan2(m_el[1].x() / btCos(euler_out.pitch),
+									m_el[0].x() / btCos(euler_out.pitch));
+			euler_out2.yaw = btAtan2(m_el[1].x() / btCos(euler_out2.pitch),
+									 m_el[0].x() / btCos(euler_out2.pitch));
 		}
 
 		if (solution_number == 1)
-		{ 
-			yaw = euler_out.yaw; 
+		{
+			yaw = euler_out.yaw;
 			pitch = euler_out.pitch;
 			roll = euler_out.roll;
 		}
 		else
-		{ 
-			yaw = euler_out2.yaw; 
+		{
+			yaw = euler_out2.yaw;
 			pitch = euler_out2.pitch;
 			roll = euler_out2.roll;
 		}
@@ -589,18 +598,18 @@ public:
 
 	btMatrix3x3 scaled(const btVector3& s) const
 	{
-#if (defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE))|| defined (BT_USE_NEON)
+#if (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)) || defined(BT_USE_NEON)
 		return btMatrix3x3(m_el[0] * s, m_el[1] * s, m_el[2] * s);
-#else		
+#else
 		return btMatrix3x3(
-            m_el[0].x() * s.x(), m_el[0].y() * s.y(), m_el[0].z() * s.z(),
+			m_el[0].x() * s.x(), m_el[0].y() * s.y(), m_el[0].z() * s.z(),
 			m_el[1].x() * s.x(), m_el[1].y() * s.y(), m_el[1].z() * s.z(),
 			m_el[2].x() * s.x(), m_el[2].y() * s.y(), m_el[2].z() * s.z());
 #endif
 	}
 
 	/**@brief Return the determinant of the matrix */
-	btScalar            determinant() const;
+	btScalar determinant() const;
 	/**@brief Return the adjoint of the matrix */
 	btMatrix3x3 adjoint() const;
 	/**@brief Return the matrix with all values non negative */
@@ -608,7 +617,7 @@ public:
 	/**@brief Return the transpose of the matrix */
 	btMatrix3x3 transpose() const;
 	/**@brief Return the inverse of the matrix */
-	btMatrix3x3 inverse() const; 
+	btMatrix3x3 inverse() const;
 
 	/// Solve A * x = b, where b is a column vector. This is more efficient
 	/// than computing the inverse in one-shot cases.
@@ -618,9 +627,9 @@ public:
 		btVector3 col1 = getColumn(0);
 		btVector3 col2 = getColumn(1);
 		btVector3 col3 = getColumn(2);
-		
+
 		btScalar det = btDot(col1, btCross(col2, col3));
-		if (btFabs(det)>SIMD_EPSILON)
+		if (btFabs(det) > SIMD_EPSILON)
 		{
 			det = 1.0f / det;
 		}
@@ -634,15 +643,15 @@ public:
 	btMatrix3x3 transposeTimes(const btMatrix3x3& m) const;
 	btMatrix3x3 timesTranspose(const btMatrix3x3& m) const;
 
-	SIMD_FORCE_INLINE btScalar tdotx(const btVector3& v) const 
+	SIMD_FORCE_INLINE btScalar tdotx(const btVector3& v) const
 	{
 		return m_el[0].x() * v.x() + m_el[1].x() * v.y() + m_el[2].x() * v.z();
 	}
-	SIMD_FORCE_INLINE btScalar tdoty(const btVector3& v) const 
+	SIMD_FORCE_INLINE btScalar tdoty(const btVector3& v) const
 	{
 		return m_el[0].y() * v.x() + m_el[1].y() * v.y() + m_el[2].y() * v.z();
 	}
-	SIMD_FORCE_INLINE btScalar tdotz(const btVector3& v) const 
+	SIMD_FORCE_INLINE btScalar tdotz(const btVector3& v) const
 	{
 		return m_el[0].z() * v.x() + m_el[1].z() * v.y() + m_el[2].z() * v.z();
 	}
@@ -653,31 +662,25 @@ public:
 	///symmetric matrix S:
 	///A = R*S.
 	///note that R can include both rotation and scaling.
-	SIMD_FORCE_INLINE void extractRotation(btQuaternion &q,btScalar tolerance = 1.0e-9, int maxIter=100)
+	SIMD_FORCE_INLINE void extractRotation(btQuaternion & q, btScalar tolerance = 1.0e-9, int maxIter = 100)
 	{
-		int iter =0;
+		int iter = 0;
 		btScalar w;
-		const btMatrix3x3& A=*this;
-		for(iter = 0; iter < maxIter; iter++)
+		const btMatrix3x3& A = *this;
+		for (iter = 0; iter < maxIter; iter++)
 		{
 			btMatrix3x3 R(q);
-			btVector3 omega = (R.getColumn(0).cross(A.getColumn(0)) + R.getColumn(1).cross(A.getColumn(1)) 
-				+ R.getColumn(2).cross(A.getColumn(2))
-				) * (btScalar(1.0) / btFabs(R.getColumn(0).dot(A.getColumn(0)) + R.getColumn
-				(1).dot(A.getColumn(1)) + R.getColumn(2).dot(A.getColumn(2))) +
-					tolerance);
+			btVector3 omega = (R.getColumn(0).cross(A.getColumn(0)) + R.getColumn(1).cross(A.getColumn(1)) + R.getColumn(2).cross(A.getColumn(2))) * (btScalar(1.0) / btFabs(R.getColumn(0).dot(A.getColumn(0)) + R.getColumn(1).dot(A.getColumn(1)) + R.getColumn(2).dot(A.getColumn(2))) +
+																																					  tolerance);
 			w = omega.norm();
-			if(w < tolerance)
+			if (w < tolerance)
 				break;
-			q = btQuaternion(btVector3((btScalar(1.0) / w) * omega),w) *
+			q = btQuaternion(btVector3((btScalar(1.0) / w) * omega), w) *
 				q;
 			q.normalize();
 		}
 	}
 
-
-
-
 	/**@brief diagonalizes this matrix by the Jacobi method.
 	* @param rot stores the rotation from the coordinate system in which the matrix is diagonal to the original
 	* coordinate system, i.e., old_this = rot * new_this * rot^T.
@@ -687,7 +690,7 @@ public:
 	*
 	* Note that this matrix is assumed to be symmetric.
 	*/
-	void diagonalize(btMatrix3x3& rot, btScalar threshold, int maxSteps)
+	void diagonalize(btMatrix3x3 & rot, btScalar threshold, int maxSteps)
 	{
 		rot.setIdentity();
 		for (int step = maxSteps; step > 0; step--)
@@ -723,7 +726,7 @@ public:
 				step = 1;
 			}
 
-			// compute Jacobi rotation J which leads to a zero for element [p][q] 
+			// compute Jacobi rotation J which leads to a zero for element [p][q]
 			btScalar mpq = m_el[p][q];
 			btScalar theta = (m_el[q][q] - m_el[p][p]) / (2 * mpq);
 			btScalar theta2 = theta * theta;
@@ -732,7 +735,7 @@ public:
 			if (theta2 * theta2 < btScalar(10 / SIMD_EPSILON))
 			{
 				t = (theta >= 0) ? 1 / (theta + btSqrt(1 + theta2))
-					: 1 / (theta - btSqrt(1 + theta2));
+								 : 1 / (theta - btSqrt(1 + theta2));
 				cos = 1 / btSqrt(1 + t * t);
 				sin = cos * t;
 			}
@@ -765,8 +768,6 @@ public:
 		}
 	}
 
-
-
 	/**@brief Calculate the matrix cofactor 
 	* @param r1 The first row to use for calculating the cofactor
 	* @param c1 The first column to use for calculating the cofactor
@@ -774,304 +775,298 @@ public:
 	* @param c1 The second column to use for calculating the cofactor
 	* See http://en.wikipedia.org/wiki/Cofactor_(linear_algebra) for more details
 	*/
-	btScalar cofac(int r1, int c1, int r2, int c2) const 
+	btScalar cofac(int r1, int c1, int r2, int c2) const
 	{
 		return m_el[r1][c1] * m_el[r2][c2] - m_el[r1][c2] * m_el[r2][c1];
 	}
 
-	void	serialize(struct	btMatrix3x3Data& dataOut) const;
+	void serialize(struct btMatrix3x3Data & dataOut) const;
 
-	void	serializeFloat(struct	btMatrix3x3FloatData& dataOut) const;
+	void serializeFloat(struct btMatrix3x3FloatData & dataOut) const;
 
-	void	deSerialize(const struct	btMatrix3x3Data& dataIn);
+	void deSerialize(const struct btMatrix3x3Data& dataIn);
 
-	void	deSerializeFloat(const struct	btMatrix3x3FloatData& dataIn);
-
-	void	deSerializeDouble(const struct	btMatrix3x3DoubleData& dataIn);
+	void deSerializeFloat(const struct btMatrix3x3FloatData& dataIn);
 
+	void deSerializeDouble(const struct btMatrix3x3DoubleData& dataIn);
 };
 
-
-SIMD_FORCE_INLINE btMatrix3x3& 
+SIMD_FORCE_INLINE btMatrix3x3&
 btMatrix3x3::operator*=(const btMatrix3x3& m)
 {
-#if defined BT_USE_SIMD_VECTOR3 && defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
-    __m128 rv00, rv01, rv02;
-    __m128 rv10, rv11, rv12;
-    __m128 rv20, rv21, rv22;
-    __m128 mv0, mv1, mv2;
+#if defined BT_USE_SIMD_VECTOR3 && defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
+	__m128 rv00, rv01, rv02;
+	__m128 rv10, rv11, rv12;
+	__m128 rv20, rv21, rv22;
+	__m128 mv0, mv1, mv2;
 
-    rv02 = m_el[0].mVec128;
-    rv12 = m_el[1].mVec128;
-    rv22 = m_el[2].mVec128;
+	rv02 = m_el[0].mVec128;
+	rv12 = m_el[1].mVec128;
+	rv22 = m_el[2].mVec128;
 
-    mv0 = _mm_and_ps(m[0].mVec128, btvFFF0fMask); 
-    mv1 = _mm_and_ps(m[1].mVec128, btvFFF0fMask); 
-    mv2 = _mm_and_ps(m[2].mVec128, btvFFF0fMask); 
-    
-    // rv0
-    rv00 = bt_splat_ps(rv02, 0);
-    rv01 = bt_splat_ps(rv02, 1);
-    rv02 = bt_splat_ps(rv02, 2);
-    
-    rv00 = _mm_mul_ps(rv00, mv0);
-    rv01 = _mm_mul_ps(rv01, mv1);
-    rv02 = _mm_mul_ps(rv02, mv2);
-    
-    // rv1
-    rv10 = bt_splat_ps(rv12, 0);
-    rv11 = bt_splat_ps(rv12, 1);
-    rv12 = bt_splat_ps(rv12, 2);
-    
-    rv10 = _mm_mul_ps(rv10, mv0);
-    rv11 = _mm_mul_ps(rv11, mv1);
-    rv12 = _mm_mul_ps(rv12, mv2);
-    
-    // rv2
-    rv20 = bt_splat_ps(rv22, 0);
-    rv21 = bt_splat_ps(rv22, 1);
-    rv22 = bt_splat_ps(rv22, 2);
-    
-    rv20 = _mm_mul_ps(rv20, mv0);
-    rv21 = _mm_mul_ps(rv21, mv1);
-    rv22 = _mm_mul_ps(rv22, mv2);
+	mv0 = _mm_and_ps(m[0].mVec128, btvFFF0fMask);
+	mv1 = _mm_and_ps(m[1].mVec128, btvFFF0fMask);
+	mv2 = _mm_and_ps(m[2].mVec128, btvFFF0fMask);
 
-    rv00 = _mm_add_ps(rv00, rv01);
-    rv10 = _mm_add_ps(rv10, rv11);
-    rv20 = _mm_add_ps(rv20, rv21);
+	// rv0
+	rv00 = bt_splat_ps(rv02, 0);
+	rv01 = bt_splat_ps(rv02, 1);
+	rv02 = bt_splat_ps(rv02, 2);
 
-    m_el[0].mVec128 = _mm_add_ps(rv00, rv02);
-    m_el[1].mVec128 = _mm_add_ps(rv10, rv12);
-    m_el[2].mVec128 = _mm_add_ps(rv20, rv22);
+	rv00 = _mm_mul_ps(rv00, mv0);
+	rv01 = _mm_mul_ps(rv01, mv1);
+	rv02 = _mm_mul_ps(rv02, mv2);
+
+	// rv1
+	rv10 = bt_splat_ps(rv12, 0);
+	rv11 = bt_splat_ps(rv12, 1);
+	rv12 = bt_splat_ps(rv12, 2);
+
+	rv10 = _mm_mul_ps(rv10, mv0);
+	rv11 = _mm_mul_ps(rv11, mv1);
+	rv12 = _mm_mul_ps(rv12, mv2);
+
+	// rv2
+	rv20 = bt_splat_ps(rv22, 0);
+	rv21 = bt_splat_ps(rv22, 1);
+	rv22 = bt_splat_ps(rv22, 2);
+
+	rv20 = _mm_mul_ps(rv20, mv0);
+	rv21 = _mm_mul_ps(rv21, mv1);
+	rv22 = _mm_mul_ps(rv22, mv2);
+
+	rv00 = _mm_add_ps(rv00, rv01);
+	rv10 = _mm_add_ps(rv10, rv11);
+	rv20 = _mm_add_ps(rv20, rv21);
+
+	m_el[0].mVec128 = _mm_add_ps(rv00, rv02);
+	m_el[1].mVec128 = _mm_add_ps(rv10, rv12);
+	m_el[2].mVec128 = _mm_add_ps(rv20, rv22);
 
 #elif defined(BT_USE_NEON)
 
-    float32x4_t rv0, rv1, rv2;
-    float32x4_t v0, v1, v2;
-    float32x4_t mv0, mv1, mv2;
+	float32x4_t rv0, rv1, rv2;
+	float32x4_t v0, v1, v2;
+	float32x4_t mv0, mv1, mv2;
 
-    v0 = m_el[0].mVec128;
-    v1 = m_el[1].mVec128;
-    v2 = m_el[2].mVec128;
+	v0 = m_el[0].mVec128;
+	v1 = m_el[1].mVec128;
+	v2 = m_el[2].mVec128;
 
-    mv0 = (float32x4_t) vandq_s32((int32x4_t)m[0].mVec128, btvFFF0Mask); 
-    mv1 = (float32x4_t) vandq_s32((int32x4_t)m[1].mVec128, btvFFF0Mask); 
-    mv2 = (float32x4_t) vandq_s32((int32x4_t)m[2].mVec128, btvFFF0Mask); 
-    
-    rv0 = vmulq_lane_f32(mv0, vget_low_f32(v0), 0);
-    rv1 = vmulq_lane_f32(mv0, vget_low_f32(v1), 0);
-    rv2 = vmulq_lane_f32(mv0, vget_low_f32(v2), 0);
-    
-    rv0 = vmlaq_lane_f32(rv0, mv1, vget_low_f32(v0), 1);
-    rv1 = vmlaq_lane_f32(rv1, mv1, vget_low_f32(v1), 1);
-    rv2 = vmlaq_lane_f32(rv2, mv1, vget_low_f32(v2), 1);
-    
-    rv0 = vmlaq_lane_f32(rv0, mv2, vget_high_f32(v0), 0);
-    rv1 = vmlaq_lane_f32(rv1, mv2, vget_high_f32(v1), 0);
-    rv2 = vmlaq_lane_f32(rv2, mv2, vget_high_f32(v2), 0);
+	mv0 = (float32x4_t)vandq_s32((int32x4_t)m[0].mVec128, btvFFF0Mask);
+	mv1 = (float32x4_t)vandq_s32((int32x4_t)m[1].mVec128, btvFFF0Mask);
+	mv2 = (float32x4_t)vandq_s32((int32x4_t)m[2].mVec128, btvFFF0Mask);
 
-    m_el[0].mVec128 = rv0;
-    m_el[1].mVec128 = rv1;
-    m_el[2].mVec128 = rv2;
-#else    
+	rv0 = vmulq_lane_f32(mv0, vget_low_f32(v0), 0);
+	rv1 = vmulq_lane_f32(mv0, vget_low_f32(v1), 0);
+	rv2 = vmulq_lane_f32(mv0, vget_low_f32(v2), 0);
+
+	rv0 = vmlaq_lane_f32(rv0, mv1, vget_low_f32(v0), 1);
+	rv1 = vmlaq_lane_f32(rv1, mv1, vget_low_f32(v1), 1);
+	rv2 = vmlaq_lane_f32(rv2, mv1, vget_low_f32(v2), 1);
+
+	rv0 = vmlaq_lane_f32(rv0, mv2, vget_high_f32(v0), 0);
+	rv1 = vmlaq_lane_f32(rv1, mv2, vget_high_f32(v1), 0);
+	rv2 = vmlaq_lane_f32(rv2, mv2, vget_high_f32(v2), 0);
+
+	m_el[0].mVec128 = rv0;
+	m_el[1].mVec128 = rv1;
+	m_el[2].mVec128 = rv2;
+#else
 	setValue(
-        m.tdotx(m_el[0]), m.tdoty(m_el[0]), m.tdotz(m_el[0]),
+		m.tdotx(m_el[0]), m.tdoty(m_el[0]), m.tdotz(m_el[0]),
 		m.tdotx(m_el[1]), m.tdoty(m_el[1]), m.tdotz(m_el[1]),
 		m.tdotx(m_el[2]), m.tdoty(m_el[2]), m.tdotz(m_el[2]));
 #endif
 	return *this;
 }
 
-SIMD_FORCE_INLINE btMatrix3x3& 
+SIMD_FORCE_INLINE btMatrix3x3&
 btMatrix3x3::operator+=(const btMatrix3x3& m)
 {
-#if (defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE))|| defined (BT_USE_NEON)
-    m_el[0].mVec128 = m_el[0].mVec128 + m.m_el[0].mVec128;
-    m_el[1].mVec128 = m_el[1].mVec128 + m.m_el[1].mVec128;
-    m_el[2].mVec128 = m_el[2].mVec128 + m.m_el[2].mVec128;
+#if (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)) || defined(BT_USE_NEON)
+	m_el[0].mVec128 = m_el[0].mVec128 + m.m_el[0].mVec128;
+	m_el[1].mVec128 = m_el[1].mVec128 + m.m_el[1].mVec128;
+	m_el[2].mVec128 = m_el[2].mVec128 + m.m_el[2].mVec128;
 #else
 	setValue(
-		m_el[0][0]+m.m_el[0][0], 
-		m_el[0][1]+m.m_el[0][1],
-		m_el[0][2]+m.m_el[0][2],
-		m_el[1][0]+m.m_el[1][0], 
-		m_el[1][1]+m.m_el[1][1],
-		m_el[1][2]+m.m_el[1][2],
-		m_el[2][0]+m.m_el[2][0], 
-		m_el[2][1]+m.m_el[2][1],
-		m_el[2][2]+m.m_el[2][2]);
+		m_el[0][0] + m.m_el[0][0],
+		m_el[0][1] + m.m_el[0][1],
+		m_el[0][2] + m.m_el[0][2],
+		m_el[1][0] + m.m_el[1][0],
+		m_el[1][1] + m.m_el[1][1],
+		m_el[1][2] + m.m_el[1][2],
+		m_el[2][0] + m.m_el[2][0],
+		m_el[2][1] + m.m_el[2][1],
+		m_el[2][2] + m.m_el[2][2]);
 #endif
 	return *this;
 }
 
 SIMD_FORCE_INLINE btMatrix3x3
-operator*(const btMatrix3x3& m, const btScalar & k)
+operator*(const btMatrix3x3& m, const btScalar& k)
 {
-#if (defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE))
-    __m128 vk = bt_splat_ps(_mm_load_ss((float *)&k), 0x80);
-    return btMatrix3x3(
-                _mm_mul_ps(m[0].mVec128, vk), 
-                _mm_mul_ps(m[1].mVec128, vk), 
-                _mm_mul_ps(m[2].mVec128, vk)); 
+#if (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE))
+	__m128 vk = bt_splat_ps(_mm_load_ss((float*)&k), 0x80);
+	return btMatrix3x3(
+		_mm_mul_ps(m[0].mVec128, vk),
+		_mm_mul_ps(m[1].mVec128, vk),
+		_mm_mul_ps(m[2].mVec128, vk));
 #elif defined(BT_USE_NEON)
-    return btMatrix3x3(
-                vmulq_n_f32(m[0].mVec128, k),
-                vmulq_n_f32(m[1].mVec128, k),
-                vmulq_n_f32(m[2].mVec128, k)); 
+	return btMatrix3x3(
+		vmulq_n_f32(m[0].mVec128, k),
+		vmulq_n_f32(m[1].mVec128, k),
+		vmulq_n_f32(m[2].mVec128, k));
 #else
 	return btMatrix3x3(
-		m[0].x()*k,m[0].y()*k,m[0].z()*k,
-		m[1].x()*k,m[1].y()*k,m[1].z()*k,
-		m[2].x()*k,m[2].y()*k,m[2].z()*k);
+		m[0].x() * k, m[0].y() * k, m[0].z() * k,
+		m[1].x() * k, m[1].y() * k, m[1].z() * k,
+		m[2].x() * k, m[2].y() * k, m[2].z() * k);
 #endif
 }
 
-SIMD_FORCE_INLINE btMatrix3x3 
+SIMD_FORCE_INLINE btMatrix3x3
 operator+(const btMatrix3x3& m1, const btMatrix3x3& m2)
 {
-#if (defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE))|| defined (BT_USE_NEON)
+#if (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)) || defined(BT_USE_NEON)
 	return btMatrix3x3(
-        m1[0].mVec128 + m2[0].mVec128,
-        m1[1].mVec128 + m2[1].mVec128,
-        m1[2].mVec128 + m2[2].mVec128);
+		m1[0].mVec128 + m2[0].mVec128,
+		m1[1].mVec128 + m2[1].mVec128,
+		m1[2].mVec128 + m2[2].mVec128);
 #else
 	return btMatrix3x3(
-        m1[0][0]+m2[0][0], 
-        m1[0][1]+m2[0][1],
-        m1[0][2]+m2[0][2],
-        
-        m1[1][0]+m2[1][0], 
-        m1[1][1]+m2[1][1],
-        m1[1][2]+m2[1][2],
-        
-        m1[2][0]+m2[2][0], 
-        m1[2][1]+m2[2][1],
-        m1[2][2]+m2[2][2]);
-#endif    
-}
+		m1[0][0] + m2[0][0],
+		m1[0][1] + m2[0][1],
+		m1[0][2] + m2[0][2],
 
-SIMD_FORCE_INLINE btMatrix3x3 
-operator-(const btMatrix3x3& m1, const btMatrix3x3& m2)
-{
-#if (defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE))|| defined (BT_USE_NEON)
-	return btMatrix3x3(
-        m1[0].mVec128 - m2[0].mVec128,
-        m1[1].mVec128 - m2[1].mVec128,
-        m1[2].mVec128 - m2[2].mVec128);
-#else
-	return btMatrix3x3(
-        m1[0][0]-m2[0][0], 
-        m1[0][1]-m2[0][1],
-        m1[0][2]-m2[0][2],
-        
-        m1[1][0]-m2[1][0], 
-        m1[1][1]-m2[1][1],
-        m1[1][2]-m2[1][2],
-        
-        m1[2][0]-m2[2][0], 
-        m1[2][1]-m2[2][1],
-        m1[2][2]-m2[2][2]);
+		m1[1][0] + m2[1][0],
+		m1[1][1] + m2[1][1],
+		m1[1][2] + m2[1][2],
+
+		m1[2][0] + m2[2][0],
+		m1[2][1] + m2[2][1],
+		m1[2][2] + m2[2][2]);
 #endif
 }
 
+SIMD_FORCE_INLINE btMatrix3x3
+operator-(const btMatrix3x3& m1, const btMatrix3x3& m2)
+{
+#if (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)) || defined(BT_USE_NEON)
+	return btMatrix3x3(
+		m1[0].mVec128 - m2[0].mVec128,
+		m1[1].mVec128 - m2[1].mVec128,
+		m1[2].mVec128 - m2[2].mVec128);
+#else
+	return btMatrix3x3(
+		m1[0][0] - m2[0][0],
+		m1[0][1] - m2[0][1],
+		m1[0][2] - m2[0][2],
 
-SIMD_FORCE_INLINE btMatrix3x3& 
+		m1[1][0] - m2[1][0],
+		m1[1][1] - m2[1][1],
+		m1[1][2] - m2[1][2],
+
+		m1[2][0] - m2[2][0],
+		m1[2][1] - m2[2][1],
+		m1[2][2] - m2[2][2]);
+#endif
+}
+
+SIMD_FORCE_INLINE btMatrix3x3&
 btMatrix3x3::operator-=(const btMatrix3x3& m)
 {
-#if (defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE))|| defined (BT_USE_NEON)
-    m_el[0].mVec128 = m_el[0].mVec128 - m.m_el[0].mVec128;
-    m_el[1].mVec128 = m_el[1].mVec128 - m.m_el[1].mVec128;
-    m_el[2].mVec128 = m_el[2].mVec128 - m.m_el[2].mVec128;
+#if (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)) || defined(BT_USE_NEON)
+	m_el[0].mVec128 = m_el[0].mVec128 - m.m_el[0].mVec128;
+	m_el[1].mVec128 = m_el[1].mVec128 - m.m_el[1].mVec128;
+	m_el[2].mVec128 = m_el[2].mVec128 - m.m_el[2].mVec128;
 #else
 	setValue(
-	m_el[0][0]-m.m_el[0][0], 
-	m_el[0][1]-m.m_el[0][1],
-	m_el[0][2]-m.m_el[0][2],
-	m_el[1][0]-m.m_el[1][0], 
-	m_el[1][1]-m.m_el[1][1],
-	m_el[1][2]-m.m_el[1][2],
-	m_el[2][0]-m.m_el[2][0], 
-	m_el[2][1]-m.m_el[2][1],
-	m_el[2][2]-m.m_el[2][2]);
+		m_el[0][0] - m.m_el[0][0],
+		m_el[0][1] - m.m_el[0][1],
+		m_el[0][2] - m.m_el[0][2],
+		m_el[1][0] - m.m_el[1][0],
+		m_el[1][1] - m.m_el[1][1],
+		m_el[1][2] - m.m_el[1][2],
+		m_el[2][0] - m.m_el[2][0],
+		m_el[2][1] - m.m_el[2][1],
+		m_el[2][2] - m.m_el[2][2]);
 #endif
 	return *this;
 }
 
-
-SIMD_FORCE_INLINE btScalar 
+SIMD_FORCE_INLINE btScalar
 btMatrix3x3::determinant() const
-{ 
+{
 	return btTriple((*this)[0], (*this)[1], (*this)[2]);
 }
 
-
-SIMD_FORCE_INLINE btMatrix3x3 
+SIMD_FORCE_INLINE btMatrix3x3
 btMatrix3x3::absolute() const
 {
-#if defined BT_USE_SIMD_VECTOR3 && (defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE))
-    return btMatrix3x3(
-            _mm_and_ps(m_el[0].mVec128, btvAbsfMask),
-            _mm_and_ps(m_el[1].mVec128, btvAbsfMask),
-            _mm_and_ps(m_el[2].mVec128, btvAbsfMask));
-#elif defined(BT_USE_NEON)
-    return btMatrix3x3(
-            (float32x4_t)vandq_s32((int32x4_t)m_el[0].mVec128, btv3AbsMask),
-            (float32x4_t)vandq_s32((int32x4_t)m_el[1].mVec128, btv3AbsMask),
-            (float32x4_t)vandq_s32((int32x4_t)m_el[2].mVec128, btv3AbsMask));
-#else	
+#if defined BT_USE_SIMD_VECTOR3 && (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE))
 	return btMatrix3x3(
-            btFabs(m_el[0].x()), btFabs(m_el[0].y()), btFabs(m_el[0].z()),
-            btFabs(m_el[1].x()), btFabs(m_el[1].y()), btFabs(m_el[1].z()),
-            btFabs(m_el[2].x()), btFabs(m_el[2].y()), btFabs(m_el[2].z()));
-#endif
-}
-
-SIMD_FORCE_INLINE btMatrix3x3 
-btMatrix3x3::transpose() const 
-{
-#if defined BT_USE_SIMD_VECTOR3 && (defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE))
-    __m128 v0 = m_el[0].mVec128;
-    __m128 v1 = m_el[1].mVec128;
-    __m128 v2 = m_el[2].mVec128;    //  x2 y2 z2 w2
-    __m128 vT;
-    
-    v2 = _mm_and_ps(v2, btvFFF0fMask);  //  x2 y2 z2 0
-    
-    vT = _mm_unpackhi_ps(v0, v1);	//	z0 z1 * *
-    v0 = _mm_unpacklo_ps(v0, v1);	//	x0 x1 y0 y1
-
-    v1 = _mm_shuffle_ps(v0, v2, BT_SHUFFLE(2, 3, 1, 3) );	// y0 y1 y2 0
-    v0 = _mm_shuffle_ps(v0, v2, BT_SHUFFLE(0, 1, 0, 3) );	// x0 x1 x2 0
-    v2 = btCastdTo128f(_mm_move_sd(btCastfTo128d(v2), btCastfTo128d(vT)));	// z0 z1 z2 0
-
-
-    return btMatrix3x3( v0, v1, v2 );
+		_mm_and_ps(m_el[0].mVec128, btvAbsfMask),
+		_mm_and_ps(m_el[1].mVec128, btvAbsfMask),
+		_mm_and_ps(m_el[2].mVec128, btvAbsfMask));
 #elif defined(BT_USE_NEON)
-    // note: zeros the w channel. We can preserve it at the cost of two more vtrn instructions.
-    static const uint32x2_t zMask = (const uint32x2_t) {static_cast<uint32_t>(-1), 0 };
-    float32x4x2_t top = vtrnq_f32( m_el[0].mVec128, m_el[1].mVec128 );  // {x0 x1 z0 z1}, {y0 y1 w0 w1}
-    float32x2x2_t bl = vtrn_f32( vget_low_f32(m_el[2].mVec128), vdup_n_f32(0.0f) );       // {x2  0 }, {y2 0}
-    float32x4_t v0 = vcombine_f32( vget_low_f32(top.val[0]), bl.val[0] );
-    float32x4_t v1 = vcombine_f32( vget_low_f32(top.val[1]), bl.val[1] );
-    float32x2_t q = (float32x2_t) vand_u32( (uint32x2_t) vget_high_f32( m_el[2].mVec128), zMask );
-    float32x4_t v2 = vcombine_f32( vget_high_f32(top.val[0]), q );       // z0 z1 z2  0
-    return btMatrix3x3( v0, v1, v2 ); 
+	return btMatrix3x3(
+		(float32x4_t)vandq_s32((int32x4_t)m_el[0].mVec128, btv3AbsMask),
+		(float32x4_t)vandq_s32((int32x4_t)m_el[1].mVec128, btv3AbsMask),
+		(float32x4_t)vandq_s32((int32x4_t)m_el[2].mVec128, btv3AbsMask));
 #else
-	return btMatrix3x3( m_el[0].x(), m_el[1].x(), m_el[2].x(),
-                        m_el[0].y(), m_el[1].y(), m_el[2].y(),
-                        m_el[0].z(), m_el[1].z(), m_el[2].z());
+	return btMatrix3x3(
+		btFabs(m_el[0].x()), btFabs(m_el[0].y()), btFabs(m_el[0].z()),
+		btFabs(m_el[1].x()), btFabs(m_el[1].y()), btFabs(m_el[1].z()),
+		btFabs(m_el[2].x()), btFabs(m_el[2].y()), btFabs(m_el[2].z()));
 #endif
 }
 
-SIMD_FORCE_INLINE btMatrix3x3 
-btMatrix3x3::adjoint() const 
+SIMD_FORCE_INLINE btMatrix3x3
+btMatrix3x3::transpose() const
+{
+#if defined BT_USE_SIMD_VECTOR3 && (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE))
+	__m128 v0 = m_el[0].mVec128;
+	__m128 v1 = m_el[1].mVec128;
+	__m128 v2 = m_el[2].mVec128;  //  x2 y2 z2 w2
+	__m128 vT;
+
+	v2 = _mm_and_ps(v2, btvFFF0fMask);  //  x2 y2 z2 0
+
+	vT = _mm_unpackhi_ps(v0, v1);  //	z0 z1 * *
+	v0 = _mm_unpacklo_ps(v0, v1);  //	x0 x1 y0 y1
+
+	v1 = _mm_shuffle_ps(v0, v2, BT_SHUFFLE(2, 3, 1, 3));                    // y0 y1 y2 0
+	v0 = _mm_shuffle_ps(v0, v2, BT_SHUFFLE(0, 1, 0, 3));                    // x0 x1 x2 0
+	v2 = btCastdTo128f(_mm_move_sd(btCastfTo128d(v2), btCastfTo128d(vT)));  // z0 z1 z2 0
+
+	return btMatrix3x3(v0, v1, v2);
+#elif defined(BT_USE_NEON)
+	// note: zeros the w channel. We can preserve it at the cost of two more vtrn instructions.
+	static const uint32x2_t zMask = (const uint32x2_t){static_cast<uint32_t>(-1), 0};
+	float32x4x2_t top = vtrnq_f32(m_el[0].mVec128, m_el[1].mVec128);               // {x0 x1 z0 z1}, {y0 y1 w0 w1}
+	float32x2x2_t bl = vtrn_f32(vget_low_f32(m_el[2].mVec128), vdup_n_f32(0.0f));  // {x2  0 }, {y2 0}
+	float32x4_t v0 = vcombine_f32(vget_low_f32(top.val[0]), bl.val[0]);
+	float32x4_t v1 = vcombine_f32(vget_low_f32(top.val[1]), bl.val[1]);
+	float32x2_t q = (float32x2_t)vand_u32((uint32x2_t)vget_high_f32(m_el[2].mVec128), zMask);
+	float32x4_t v2 = vcombine_f32(vget_high_f32(top.val[0]), q);  // z0 z1 z2  0
+	return btMatrix3x3(v0, v1, v2);
+#else
+	return btMatrix3x3(m_el[0].x(), m_el[1].x(), m_el[2].x(),
+					   m_el[0].y(), m_el[1].y(), m_el[2].y(),
+					   m_el[0].z(), m_el[1].z(), m_el[2].z());
+#endif
+}
+
+SIMD_FORCE_INLINE btMatrix3x3
+btMatrix3x3::adjoint() const
 {
 	return btMatrix3x3(cofac(1, 1, 2, 2), cofac(0, 2, 2, 1), cofac(0, 1, 1, 2),
-		cofac(1, 2, 2, 0), cofac(0, 0, 2, 2), cofac(0, 2, 1, 0),
-		cofac(1, 0, 2, 1), cofac(0, 1, 2, 0), cofac(0, 0, 1, 1));
+					   cofac(1, 2, 2, 0), cofac(0, 0, 2, 2), cofac(0, 2, 1, 0),
+					   cofac(1, 0, 2, 1), cofac(0, 1, 2, 0), cofac(0, 0, 1, 1));
 }
 
-SIMD_FORCE_INLINE btMatrix3x3 
+SIMD_FORCE_INLINE btMatrix3x3
 btMatrix3x3::inverse() const
 {
 	btVector3 co(cofac(1, 1, 2, 2), cofac(1, 2, 2, 0), cofac(1, 0, 2, 1));
@@ -1080,54 +1075,54 @@ btMatrix3x3::inverse() const
 	btAssert(det != btScalar(0.0));
 	btScalar s = btScalar(1.0) / det;
 	return btMatrix3x3(co.x() * s, cofac(0, 2, 2, 1) * s, cofac(0, 1, 1, 2) * s,
-		co.y() * s, cofac(0, 0, 2, 2) * s, cofac(0, 2, 1, 0) * s,
-		co.z() * s, cofac(0, 1, 2, 0) * s, cofac(0, 0, 1, 1) * s);
+					   co.y() * s, cofac(0, 0, 2, 2) * s, cofac(0, 2, 1, 0) * s,
+					   co.z() * s, cofac(0, 1, 2, 0) * s, cofac(0, 0, 1, 1) * s);
 }
 
-SIMD_FORCE_INLINE btMatrix3x3 
+SIMD_FORCE_INLINE btMatrix3x3
 btMatrix3x3::transposeTimes(const btMatrix3x3& m) const
 {
-#if defined BT_USE_SIMD_VECTOR3 && (defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE))
-    // zeros w
-//    static const __m128i xyzMask = (const __m128i){ -1ULL, 0xffffffffULL };
-    __m128 row = m_el[0].mVec128;
-    __m128 m0 = _mm_and_ps( m.getRow(0).mVec128, btvFFF0fMask );
-    __m128 m1 = _mm_and_ps( m.getRow(1).mVec128, btvFFF0fMask);
-    __m128 m2 = _mm_and_ps( m.getRow(2).mVec128, btvFFF0fMask );
-    __m128 r0 = _mm_mul_ps(m0, _mm_shuffle_ps(row, row, 0));
-    __m128 r1 = _mm_mul_ps(m0, _mm_shuffle_ps(row, row, 0x55));
-    __m128 r2 = _mm_mul_ps(m0, _mm_shuffle_ps(row, row, 0xaa));
-    row = m_el[1].mVec128;
-    r0 = _mm_add_ps( r0, _mm_mul_ps(m1, _mm_shuffle_ps(row, row, 0)));
-    r1 = _mm_add_ps( r1, _mm_mul_ps(m1, _mm_shuffle_ps(row, row, 0x55)));
-    r2 = _mm_add_ps( r2, _mm_mul_ps(m1, _mm_shuffle_ps(row, row, 0xaa)));
-    row = m_el[2].mVec128;
-    r0 = _mm_add_ps( r0, _mm_mul_ps(m2, _mm_shuffle_ps(row, row, 0)));
-    r1 = _mm_add_ps( r1, _mm_mul_ps(m2, _mm_shuffle_ps(row, row, 0x55)));
-    r2 = _mm_add_ps( r2, _mm_mul_ps(m2, _mm_shuffle_ps(row, row, 0xaa)));
-    return btMatrix3x3( r0, r1, r2 );
+#if defined BT_USE_SIMD_VECTOR3 && (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE))
+	// zeros w
+	//    static const __m128i xyzMask = (const __m128i){ -1ULL, 0xffffffffULL };
+	__m128 row = m_el[0].mVec128;
+	__m128 m0 = _mm_and_ps(m.getRow(0).mVec128, btvFFF0fMask);
+	__m128 m1 = _mm_and_ps(m.getRow(1).mVec128, btvFFF0fMask);
+	__m128 m2 = _mm_and_ps(m.getRow(2).mVec128, btvFFF0fMask);
+	__m128 r0 = _mm_mul_ps(m0, _mm_shuffle_ps(row, row, 0));
+	__m128 r1 = _mm_mul_ps(m0, _mm_shuffle_ps(row, row, 0x55));
+	__m128 r2 = _mm_mul_ps(m0, _mm_shuffle_ps(row, row, 0xaa));
+	row = m_el[1].mVec128;
+	r0 = _mm_add_ps(r0, _mm_mul_ps(m1, _mm_shuffle_ps(row, row, 0)));
+	r1 = _mm_add_ps(r1, _mm_mul_ps(m1, _mm_shuffle_ps(row, row, 0x55)));
+	r2 = _mm_add_ps(r2, _mm_mul_ps(m1, _mm_shuffle_ps(row, row, 0xaa)));
+	row = m_el[2].mVec128;
+	r0 = _mm_add_ps(r0, _mm_mul_ps(m2, _mm_shuffle_ps(row, row, 0)));
+	r1 = _mm_add_ps(r1, _mm_mul_ps(m2, _mm_shuffle_ps(row, row, 0x55)));
+	r2 = _mm_add_ps(r2, _mm_mul_ps(m2, _mm_shuffle_ps(row, row, 0xaa)));
+	return btMatrix3x3(r0, r1, r2);
 
 #elif defined BT_USE_NEON
-    // zeros w
-    static const uint32x4_t xyzMask = (const uint32x4_t){ static_cast<uint32_t>(-1), static_cast<uint32_t>(-1), static_cast<uint32_t>(-1), 0 };
-    float32x4_t m0 = (float32x4_t) vandq_u32( (uint32x4_t) m.getRow(0).mVec128, xyzMask );
-    float32x4_t m1 = (float32x4_t) vandq_u32( (uint32x4_t) m.getRow(1).mVec128, xyzMask );
-    float32x4_t m2 = (float32x4_t) vandq_u32( (uint32x4_t) m.getRow(2).mVec128, xyzMask );
-    float32x4_t row = m_el[0].mVec128;
-    float32x4_t r0 = vmulq_lane_f32( m0, vget_low_f32(row), 0);
-    float32x4_t r1 = vmulq_lane_f32( m0, vget_low_f32(row), 1);
-    float32x4_t r2 = vmulq_lane_f32( m0, vget_high_f32(row), 0);
-    row = m_el[1].mVec128;
-    r0 = vmlaq_lane_f32( r0, m1, vget_low_f32(row), 0);
-    r1 = vmlaq_lane_f32( r1, m1, vget_low_f32(row), 1);
-    r2 = vmlaq_lane_f32( r2, m1, vget_high_f32(row), 0);
-    row = m_el[2].mVec128;
-    r0 = vmlaq_lane_f32( r0, m2, vget_low_f32(row), 0);
-    r1 = vmlaq_lane_f32( r1, m2, vget_low_f32(row), 1);
-    r2 = vmlaq_lane_f32( r2, m2, vget_high_f32(row), 0);
-    return btMatrix3x3( r0, r1, r2 );
+	// zeros w
+	static const uint32x4_t xyzMask = (const uint32x4_t){static_cast<uint32_t>(-1), static_cast<uint32_t>(-1), static_cast<uint32_t>(-1), 0};
+	float32x4_t m0 = (float32x4_t)vandq_u32((uint32x4_t)m.getRow(0).mVec128, xyzMask);
+	float32x4_t m1 = (float32x4_t)vandq_u32((uint32x4_t)m.getRow(1).mVec128, xyzMask);
+	float32x4_t m2 = (float32x4_t)vandq_u32((uint32x4_t)m.getRow(2).mVec128, xyzMask);
+	float32x4_t row = m_el[0].mVec128;
+	float32x4_t r0 = vmulq_lane_f32(m0, vget_low_f32(row), 0);
+	float32x4_t r1 = vmulq_lane_f32(m0, vget_low_f32(row), 1);
+	float32x4_t r2 = vmulq_lane_f32(m0, vget_high_f32(row), 0);
+	row = m_el[1].mVec128;
+	r0 = vmlaq_lane_f32(r0, m1, vget_low_f32(row), 0);
+	r1 = vmlaq_lane_f32(r1, m1, vget_low_f32(row), 1);
+	r2 = vmlaq_lane_f32(r2, m1, vget_high_f32(row), 0);
+	row = m_el[2].mVec128;
+	r0 = vmlaq_lane_f32(r0, m2, vget_low_f32(row), 0);
+	r1 = vmlaq_lane_f32(r1, m2, vget_low_f32(row), 1);
+	r2 = vmlaq_lane_f32(r2, m2, vget_high_f32(row), 0);
+	return btMatrix3x3(r0, r1, r2);
 #else
-    return btMatrix3x3(
+	return btMatrix3x3(
 		m_el[0].x() * m[0].x() + m_el[1].x() * m[1].x() + m_el[2].x() * m[2].x(),
 		m_el[0].x() * m[0].y() + m_el[1].x() * m[1].y() + m_el[2].x() * m[2].y(),
 		m_el[0].x() * m[0].z() + m_el[1].x() * m[1].z() + m_el[2].x() * m[2].z(),
@@ -1140,51 +1135,51 @@ btMatrix3x3::transposeTimes(const btMatrix3x3& m) const
 #endif
 }
 
-SIMD_FORCE_INLINE btMatrix3x3 
+SIMD_FORCE_INLINE btMatrix3x3
 btMatrix3x3::timesTranspose(const btMatrix3x3& m) const
 {
-#if (defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE))
-    __m128 a0 = m_el[0].mVec128;
-    __m128 a1 = m_el[1].mVec128;
-    __m128 a2 = m_el[2].mVec128;
-    
-    btMatrix3x3 mT = m.transpose(); // we rely on transpose() zeroing w channel so that we don't have to do it here
-    __m128 mx = mT[0].mVec128;
-    __m128 my = mT[1].mVec128;
-    __m128 mz = mT[2].mVec128;
-    
-    __m128 r0 = _mm_mul_ps(mx, _mm_shuffle_ps(a0, a0, 0x00));
-    __m128 r1 = _mm_mul_ps(mx, _mm_shuffle_ps(a1, a1, 0x00));
-    __m128 r2 = _mm_mul_ps(mx, _mm_shuffle_ps(a2, a2, 0x00));
-    r0 = _mm_add_ps(r0, _mm_mul_ps(my, _mm_shuffle_ps(a0, a0, 0x55)));
-    r1 = _mm_add_ps(r1, _mm_mul_ps(my, _mm_shuffle_ps(a1, a1, 0x55)));
-    r2 = _mm_add_ps(r2, _mm_mul_ps(my, _mm_shuffle_ps(a2, a2, 0x55)));
-    r0 = _mm_add_ps(r0, _mm_mul_ps(mz, _mm_shuffle_ps(a0, a0, 0xaa)));
-    r1 = _mm_add_ps(r1, _mm_mul_ps(mz, _mm_shuffle_ps(a1, a1, 0xaa)));
-    r2 = _mm_add_ps(r2, _mm_mul_ps(mz, _mm_shuffle_ps(a2, a2, 0xaa)));
-    return btMatrix3x3( r0, r1, r2);
-            
+#if (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE))
+	__m128 a0 = m_el[0].mVec128;
+	__m128 a1 = m_el[1].mVec128;
+	__m128 a2 = m_el[2].mVec128;
+
+	btMatrix3x3 mT = m.transpose();  // we rely on transpose() zeroing w channel so that we don't have to do it here
+	__m128 mx = mT[0].mVec128;
+	__m128 my = mT[1].mVec128;
+	__m128 mz = mT[2].mVec128;
+
+	__m128 r0 = _mm_mul_ps(mx, _mm_shuffle_ps(a0, a0, 0x00));
+	__m128 r1 = _mm_mul_ps(mx, _mm_shuffle_ps(a1, a1, 0x00));
+	__m128 r2 = _mm_mul_ps(mx, _mm_shuffle_ps(a2, a2, 0x00));
+	r0 = _mm_add_ps(r0, _mm_mul_ps(my, _mm_shuffle_ps(a0, a0, 0x55)));
+	r1 = _mm_add_ps(r1, _mm_mul_ps(my, _mm_shuffle_ps(a1, a1, 0x55)));
+	r2 = _mm_add_ps(r2, _mm_mul_ps(my, _mm_shuffle_ps(a2, a2, 0x55)));
+	r0 = _mm_add_ps(r0, _mm_mul_ps(mz, _mm_shuffle_ps(a0, a0, 0xaa)));
+	r1 = _mm_add_ps(r1, _mm_mul_ps(mz, _mm_shuffle_ps(a1, a1, 0xaa)));
+	r2 = _mm_add_ps(r2, _mm_mul_ps(mz, _mm_shuffle_ps(a2, a2, 0xaa)));
+	return btMatrix3x3(r0, r1, r2);
+
 #elif defined BT_USE_NEON
-    float32x4_t a0 = m_el[0].mVec128;
-    float32x4_t a1 = m_el[1].mVec128;
-    float32x4_t a2 = m_el[2].mVec128;
-    
-    btMatrix3x3 mT = m.transpose(); // we rely on transpose() zeroing w channel so that we don't have to do it here
-    float32x4_t mx = mT[0].mVec128;
-    float32x4_t my = mT[1].mVec128;
-    float32x4_t mz = mT[2].mVec128;
-    
-    float32x4_t r0 = vmulq_lane_f32( mx, vget_low_f32(a0), 0);
-    float32x4_t r1 = vmulq_lane_f32( mx, vget_low_f32(a1), 0);
-    float32x4_t r2 = vmulq_lane_f32( mx, vget_low_f32(a2), 0);
-    r0 = vmlaq_lane_f32( r0, my, vget_low_f32(a0), 1);
-    r1 = vmlaq_lane_f32( r1, my, vget_low_f32(a1), 1);
-    r2 = vmlaq_lane_f32( r2, my, vget_low_f32(a2), 1);
-    r0 = vmlaq_lane_f32( r0, mz, vget_high_f32(a0), 0);
-    r1 = vmlaq_lane_f32( r1, mz, vget_high_f32(a1), 0);
-    r2 = vmlaq_lane_f32( r2, mz, vget_high_f32(a2), 0);
-    return btMatrix3x3( r0, r1, r2 );
-    
+	float32x4_t a0 = m_el[0].mVec128;
+	float32x4_t a1 = m_el[1].mVec128;
+	float32x4_t a2 = m_el[2].mVec128;
+
+	btMatrix3x3 mT = m.transpose();  // we rely on transpose() zeroing w channel so that we don't have to do it here
+	float32x4_t mx = mT[0].mVec128;
+	float32x4_t my = mT[1].mVec128;
+	float32x4_t mz = mT[2].mVec128;
+
+	float32x4_t r0 = vmulq_lane_f32(mx, vget_low_f32(a0), 0);
+	float32x4_t r1 = vmulq_lane_f32(mx, vget_low_f32(a1), 0);
+	float32x4_t r2 = vmulq_lane_f32(mx, vget_low_f32(a2), 0);
+	r0 = vmlaq_lane_f32(r0, my, vget_low_f32(a0), 1);
+	r1 = vmlaq_lane_f32(r1, my, vget_low_f32(a1), 1);
+	r2 = vmlaq_lane_f32(r2, my, vget_low_f32(a2), 1);
+	r0 = vmlaq_lane_f32(r0, mz, vget_high_f32(a0), 0);
+	r1 = vmlaq_lane_f32(r1, mz, vget_high_f32(a1), 0);
+	r2 = vmlaq_lane_f32(r2, mz, vget_high_f32(a2), 0);
+	return btMatrix3x3(r0, r1, r2);
+
 #else
 	return btMatrix3x3(
 		m_el[0].dot(m[0]), m_el[0].dot(m[1]), m_el[0].dot(m[2]),
@@ -1193,139 +1188,138 @@ btMatrix3x3::timesTranspose(const btMatrix3x3& m) const
 #endif
 }
 
-SIMD_FORCE_INLINE btVector3 
-operator*(const btMatrix3x3& m, const btVector3& v) 
+SIMD_FORCE_INLINE btVector3
+operator*(const btMatrix3x3& m, const btVector3& v)
 {
-#if (defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE))|| defined (BT_USE_NEON)
-    return v.dot3(m[0], m[1], m[2]);
+#if (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)) || defined(BT_USE_NEON)
+	return v.dot3(m[0], m[1], m[2]);
 #else
 	return btVector3(m[0].dot(v), m[1].dot(v), m[2].dot(v));
 #endif
 }
 
-
 SIMD_FORCE_INLINE btVector3
 operator*(const btVector3& v, const btMatrix3x3& m)
 {
-#if defined BT_USE_SIMD_VECTOR3 && (defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE))
+#if defined BT_USE_SIMD_VECTOR3 && (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE))
 
-    const __m128 vv = v.mVec128;
+	const __m128 vv = v.mVec128;
 
-    __m128 c0 = bt_splat_ps( vv, 0);
-    __m128 c1 = bt_splat_ps( vv, 1);
-    __m128 c2 = bt_splat_ps( vv, 2);
+	__m128 c0 = bt_splat_ps(vv, 0);
+	__m128 c1 = bt_splat_ps(vv, 1);
+	__m128 c2 = bt_splat_ps(vv, 2);
 
-    c0 = _mm_mul_ps(c0, _mm_and_ps(m[0].mVec128, btvFFF0fMask) );
-    c1 = _mm_mul_ps(c1, _mm_and_ps(m[1].mVec128, btvFFF0fMask) );
-    c0 = _mm_add_ps(c0, c1);
-    c2 = _mm_mul_ps(c2, _mm_and_ps(m[2].mVec128, btvFFF0fMask) );
-    
-    return btVector3(_mm_add_ps(c0, c2));
+	c0 = _mm_mul_ps(c0, _mm_and_ps(m[0].mVec128, btvFFF0fMask));
+	c1 = _mm_mul_ps(c1, _mm_and_ps(m[1].mVec128, btvFFF0fMask));
+	c0 = _mm_add_ps(c0, c1);
+	c2 = _mm_mul_ps(c2, _mm_and_ps(m[2].mVec128, btvFFF0fMask));
+
+	return btVector3(_mm_add_ps(c0, c2));
 #elif defined(BT_USE_NEON)
-    const float32x4_t vv = v.mVec128;
-    const float32x2_t vlo = vget_low_f32(vv);
-    const float32x2_t vhi = vget_high_f32(vv);
+	const float32x4_t vv = v.mVec128;
+	const float32x2_t vlo = vget_low_f32(vv);
+	const float32x2_t vhi = vget_high_f32(vv);
 
-    float32x4_t c0, c1, c2;
+	float32x4_t c0, c1, c2;
 
-    c0 = (float32x4_t) vandq_s32((int32x4_t)m[0].mVec128, btvFFF0Mask);
-    c1 = (float32x4_t) vandq_s32((int32x4_t)m[1].mVec128, btvFFF0Mask);
-    c2 = (float32x4_t) vandq_s32((int32x4_t)m[2].mVec128, btvFFF0Mask);
+	c0 = (float32x4_t)vandq_s32((int32x4_t)m[0].mVec128, btvFFF0Mask);
+	c1 = (float32x4_t)vandq_s32((int32x4_t)m[1].mVec128, btvFFF0Mask);
+	c2 = (float32x4_t)vandq_s32((int32x4_t)m[2].mVec128, btvFFF0Mask);
 
-    c0 = vmulq_lane_f32(c0, vlo, 0);
-    c1 = vmulq_lane_f32(c1, vlo, 1);
-    c2 = vmulq_lane_f32(c2, vhi, 0);
-    c0 = vaddq_f32(c0, c1);
-    c0 = vaddq_f32(c0, c2);
-    
-    return btVector3(c0);
+	c0 = vmulq_lane_f32(c0, vlo, 0);
+	c1 = vmulq_lane_f32(c1, vlo, 1);
+	c2 = vmulq_lane_f32(c2, vhi, 0);
+	c0 = vaddq_f32(c0, c1);
+	c0 = vaddq_f32(c0, c2);
+
+	return btVector3(c0);
 #else
 	return btVector3(m.tdotx(v), m.tdoty(v), m.tdotz(v));
 #endif
 }
 
-SIMD_FORCE_INLINE btMatrix3x3 
+SIMD_FORCE_INLINE btMatrix3x3
 operator*(const btMatrix3x3& m1, const btMatrix3x3& m2)
 {
-#if defined BT_USE_SIMD_VECTOR3 && (defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE))
+#if defined BT_USE_SIMD_VECTOR3 && (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE))
 
-    __m128 m10 = m1[0].mVec128;  
-    __m128 m11 = m1[1].mVec128;
-    __m128 m12 = m1[2].mVec128;
-    
-    __m128 m2v = _mm_and_ps(m2[0].mVec128, btvFFF0fMask);
-    
-    __m128 c0 = bt_splat_ps( m10, 0);
-    __m128 c1 = bt_splat_ps( m11, 0);
-    __m128 c2 = bt_splat_ps( m12, 0);
-    
-    c0 = _mm_mul_ps(c0, m2v);
-    c1 = _mm_mul_ps(c1, m2v);
-    c2 = _mm_mul_ps(c2, m2v);
-    
-    m2v = _mm_and_ps(m2[1].mVec128, btvFFF0fMask);
-    
-    __m128 c0_1 = bt_splat_ps( m10, 1);
-    __m128 c1_1 = bt_splat_ps( m11, 1);
-    __m128 c2_1 = bt_splat_ps( m12, 1);
-    
-    c0_1 = _mm_mul_ps(c0_1, m2v);
-    c1_1 = _mm_mul_ps(c1_1, m2v);
-    c2_1 = _mm_mul_ps(c2_1, m2v);
-    
-    m2v = _mm_and_ps(m2[2].mVec128, btvFFF0fMask);
-    
-    c0 = _mm_add_ps(c0, c0_1);
-    c1 = _mm_add_ps(c1, c1_1);
-    c2 = _mm_add_ps(c2, c2_1);
-    
-    m10 = bt_splat_ps( m10, 2);
-    m11 = bt_splat_ps( m11, 2);
-    m12 = bt_splat_ps( m12, 2);
-    
-    m10 = _mm_mul_ps(m10, m2v);
-    m11 = _mm_mul_ps(m11, m2v);
-    m12 = _mm_mul_ps(m12, m2v);
-    
-    c0 = _mm_add_ps(c0, m10);
-    c1 = _mm_add_ps(c1, m11);
-    c2 = _mm_add_ps(c2, m12);
-    
-    return btMatrix3x3(c0, c1, c2);
+	__m128 m10 = m1[0].mVec128;
+	__m128 m11 = m1[1].mVec128;
+	__m128 m12 = m1[2].mVec128;
+
+	__m128 m2v = _mm_and_ps(m2[0].mVec128, btvFFF0fMask);
+
+	__m128 c0 = bt_splat_ps(m10, 0);
+	__m128 c1 = bt_splat_ps(m11, 0);
+	__m128 c2 = bt_splat_ps(m12, 0);
+
+	c0 = _mm_mul_ps(c0, m2v);
+	c1 = _mm_mul_ps(c1, m2v);
+	c2 = _mm_mul_ps(c2, m2v);
+
+	m2v = _mm_and_ps(m2[1].mVec128, btvFFF0fMask);
+
+	__m128 c0_1 = bt_splat_ps(m10, 1);
+	__m128 c1_1 = bt_splat_ps(m11, 1);
+	__m128 c2_1 = bt_splat_ps(m12, 1);
+
+	c0_1 = _mm_mul_ps(c0_1, m2v);
+	c1_1 = _mm_mul_ps(c1_1, m2v);
+	c2_1 = _mm_mul_ps(c2_1, m2v);
+
+	m2v = _mm_and_ps(m2[2].mVec128, btvFFF0fMask);
+
+	c0 = _mm_add_ps(c0, c0_1);
+	c1 = _mm_add_ps(c1, c1_1);
+	c2 = _mm_add_ps(c2, c2_1);
+
+	m10 = bt_splat_ps(m10, 2);
+	m11 = bt_splat_ps(m11, 2);
+	m12 = bt_splat_ps(m12, 2);
+
+	m10 = _mm_mul_ps(m10, m2v);
+	m11 = _mm_mul_ps(m11, m2v);
+	m12 = _mm_mul_ps(m12, m2v);
+
+	c0 = _mm_add_ps(c0, m10);
+	c1 = _mm_add_ps(c1, m11);
+	c2 = _mm_add_ps(c2, m12);
+
+	return btMatrix3x3(c0, c1, c2);
 
 #elif defined(BT_USE_NEON)
 
-    float32x4_t rv0, rv1, rv2;
-    float32x4_t v0, v1, v2;
-    float32x4_t mv0, mv1, mv2;
+	float32x4_t rv0, rv1, rv2;
+	float32x4_t v0, v1, v2;
+	float32x4_t mv0, mv1, mv2;
 
-    v0 = m1[0].mVec128;
-    v1 = m1[1].mVec128;
-    v2 = m1[2].mVec128;
+	v0 = m1[0].mVec128;
+	v1 = m1[1].mVec128;
+	v2 = m1[2].mVec128;
 
-    mv0 = (float32x4_t) vandq_s32((int32x4_t)m2[0].mVec128, btvFFF0Mask); 
-    mv1 = (float32x4_t) vandq_s32((int32x4_t)m2[1].mVec128, btvFFF0Mask); 
-    mv2 = (float32x4_t) vandq_s32((int32x4_t)m2[2].mVec128, btvFFF0Mask); 
-    
-    rv0 = vmulq_lane_f32(mv0, vget_low_f32(v0), 0);
-    rv1 = vmulq_lane_f32(mv0, vget_low_f32(v1), 0);
-    rv2 = vmulq_lane_f32(mv0, vget_low_f32(v2), 0);
-    
-    rv0 = vmlaq_lane_f32(rv0, mv1, vget_low_f32(v0), 1);
-    rv1 = vmlaq_lane_f32(rv1, mv1, vget_low_f32(v1), 1);
-    rv2 = vmlaq_lane_f32(rv2, mv1, vget_low_f32(v2), 1);
-    
-    rv0 = vmlaq_lane_f32(rv0, mv2, vget_high_f32(v0), 0);
-    rv1 = vmlaq_lane_f32(rv1, mv2, vget_high_f32(v1), 0);
-    rv2 = vmlaq_lane_f32(rv2, mv2, vget_high_f32(v2), 0);
+	mv0 = (float32x4_t)vandq_s32((int32x4_t)m2[0].mVec128, btvFFF0Mask);
+	mv1 = (float32x4_t)vandq_s32((int32x4_t)m2[1].mVec128, btvFFF0Mask);
+	mv2 = (float32x4_t)vandq_s32((int32x4_t)m2[2].mVec128, btvFFF0Mask);
+
+	rv0 = vmulq_lane_f32(mv0, vget_low_f32(v0), 0);
+	rv1 = vmulq_lane_f32(mv0, vget_low_f32(v1), 0);
+	rv2 = vmulq_lane_f32(mv0, vget_low_f32(v2), 0);
+
+	rv0 = vmlaq_lane_f32(rv0, mv1, vget_low_f32(v0), 1);
+	rv1 = vmlaq_lane_f32(rv1, mv1, vget_low_f32(v1), 1);
+	rv2 = vmlaq_lane_f32(rv2, mv1, vget_low_f32(v2), 1);
+
+	rv0 = vmlaq_lane_f32(rv0, mv2, vget_high_f32(v0), 0);
+	rv1 = vmlaq_lane_f32(rv1, mv2, vget_high_f32(v1), 0);
+	rv2 = vmlaq_lane_f32(rv2, mv2, vget_high_f32(v2), 0);
 
 	return btMatrix3x3(rv0, rv1, rv2);
-        
-#else	
+
+#else
 	return btMatrix3x3(
-		m2.tdotx( m1[0]), m2.tdoty( m1[0]), m2.tdotz( m1[0]),
-		m2.tdotx( m1[1]), m2.tdoty( m1[1]), m2.tdotz( m1[1]),
-		m2.tdotx( m1[2]), m2.tdoty( m1[2]), m2.tdotz( m1[2]));
+		m2.tdotx(m1[0]), m2.tdoty(m1[0]), m2.tdotz(m1[0]),
+		m2.tdotx(m1[1]), m2.tdoty(m1[1]), m2.tdotz(m1[1]),
+		m2.tdotx(m1[2]), m2.tdoty(m1[2]), m2.tdotz(m1[2]));
 #endif
 }
 
@@ -1348,73 +1342,67 @@ m1[0][2] * m2[0][2] + m1[1][2] * m2[1][2] + m1[2][2] * m2[2][2]);
 * It will test all elements are equal.  */
 SIMD_FORCE_INLINE bool operator==(const btMatrix3x3& m1, const btMatrix3x3& m2)
 {
-#if (defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE))
+#if (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE))
 
-    __m128 c0, c1, c2;
+	__m128 c0, c1, c2;
 
-    c0 = _mm_cmpeq_ps(m1[0].mVec128, m2[0].mVec128);
-    c1 = _mm_cmpeq_ps(m1[1].mVec128, m2[1].mVec128);
-    c2 = _mm_cmpeq_ps(m1[2].mVec128, m2[2].mVec128);
-    
-    c0 = _mm_and_ps(c0, c1);
-    c0 = _mm_and_ps(c0, c2);
+	c0 = _mm_cmpeq_ps(m1[0].mVec128, m2[0].mVec128);
+	c1 = _mm_cmpeq_ps(m1[1].mVec128, m2[1].mVec128);
+	c2 = _mm_cmpeq_ps(m1[2].mVec128, m2[2].mVec128);
+
+	c0 = _mm_and_ps(c0, c1);
+	c0 = _mm_and_ps(c0, c2);
 
 	int m = _mm_movemask_ps((__m128)c0);
 	return (0x7 == (m & 0x7));
-	
-#else 
-	return 
-    (   m1[0][0] == m2[0][0] && m1[1][0] == m2[1][0] && m1[2][0] == m2[2][0] &&
-		m1[0][1] == m2[0][1] && m1[1][1] == m2[1][1] && m1[2][1] == m2[2][1] &&
-		m1[0][2] == m2[0][2] && m1[1][2] == m2[1][2] && m1[2][2] == m2[2][2] );
+
+#else
+	return (m1[0][0] == m2[0][0] && m1[1][0] == m2[1][0] && m1[2][0] == m2[2][0] &&
+			m1[0][1] == m2[0][1] && m1[1][1] == m2[1][1] && m1[2][1] == m2[2][1] &&
+			m1[0][2] == m2[0][2] && m1[1][2] == m2[1][2] && m1[2][2] == m2[2][2]);
 #endif
 }
 
 ///for serialization
-struct	btMatrix3x3FloatData
+struct btMatrix3x3FloatData
 {
 	btVector3FloatData m_el[3];
 };
 
 ///for serialization
-struct	btMatrix3x3DoubleData
+struct btMatrix3x3DoubleData
 {
 	btVector3DoubleData m_el[3];
 };
 
-
-	
-
-SIMD_FORCE_INLINE	void	btMatrix3x3::serialize(struct	btMatrix3x3Data& dataOut) const
+SIMD_FORCE_INLINE void btMatrix3x3::serialize(struct btMatrix3x3Data& dataOut) const
 {
-	for (int i=0;i<3;i++)
+	for (int i = 0; i < 3; i++)
 		m_el[i].serialize(dataOut.m_el[i]);
 }
 
-SIMD_FORCE_INLINE	void	btMatrix3x3::serializeFloat(struct	btMatrix3x3FloatData& dataOut) const
+SIMD_FORCE_INLINE void btMatrix3x3::serializeFloat(struct btMatrix3x3FloatData& dataOut) const
 {
-	for (int i=0;i<3;i++)
+	for (int i = 0; i < 3; i++)
 		m_el[i].serializeFloat(dataOut.m_el[i]);
 }
 
-
-SIMD_FORCE_INLINE	void	btMatrix3x3::deSerialize(const struct	btMatrix3x3Data& dataIn)
+SIMD_FORCE_INLINE void btMatrix3x3::deSerialize(const struct btMatrix3x3Data& dataIn)
 {
-	for (int i=0;i<3;i++)
+	for (int i = 0; i < 3; i++)
 		m_el[i].deSerialize(dataIn.m_el[i]);
 }
 
-SIMD_FORCE_INLINE	void	btMatrix3x3::deSerializeFloat(const struct	btMatrix3x3FloatData& dataIn)
+SIMD_FORCE_INLINE void btMatrix3x3::deSerializeFloat(const struct btMatrix3x3FloatData& dataIn)
 {
-	for (int i=0;i<3;i++)
+	for (int i = 0; i < 3; i++)
 		m_el[i].deSerializeFloat(dataIn.m_el[i]);
 }
 
-SIMD_FORCE_INLINE	void	btMatrix3x3::deSerializeDouble(const struct	btMatrix3x3DoubleData& dataIn)
+SIMD_FORCE_INLINE void btMatrix3x3::deSerializeDouble(const struct btMatrix3x3DoubleData& dataIn)
 {
-	for (int i=0;i<3;i++)
+	for (int i = 0; i < 3; i++)
 		m_el[i].deSerializeDouble(dataIn.m_el[i]);
 }
 
-#endif //BT_MATRIX3x3_H
-
+#endif  //BT_MATRIX3x3_H
diff --git a/thirdparty/bullet/LinearMath/btMatrixX.h b/thirdparty/bullet/LinearMath/btMatrixX.h
index 42caed42eff..9df9e49469e 100644
--- a/thirdparty/bullet/LinearMath/btMatrixX.h
+++ b/thirdparty/bullet/LinearMath/btMatrixX.h
@@ -24,24 +24,23 @@ subject to the following restrictions:
 //#define BT_DEBUG_OSTREAM
 #ifdef BT_DEBUG_OSTREAM
 #include <iostream>
-#include <iomanip>      // std::setw
-#endif //BT_DEBUG_OSTREAM
+#include <iomanip>  // std::setw
+#endif              //BT_DEBUG_OSTREAM
 
 class btIntSortPredicate
 {
-	public:
-		bool operator() ( const int& a, const int& b ) const
-		{
-			 return a < b;
-		}
+public:
+	bool operator()(const int& a, const int& b) const
+	{
+		return a < b;
+	}
 };
 
-
 template <typename T>
 struct btVectorX
 {
-	btAlignedObjectArray<T>	m_storage;
-	
+	btAlignedObjectArray<T> m_storage;
+
 	btVectorX()
 	{
 	}
@@ -49,7 +48,7 @@ struct btVectorX
 	{
 		m_storage.resize(numRows);
 	}
-	
+
 	void resize(int rows)
 	{
 		m_storage.resize(rows);
@@ -66,13 +65,13 @@ struct btVectorX
 	{
 		return rows();
 	}
-	
+
 	T nrm2() const
 	{
 		T norm = T(0);
-		
+
 		int nn = rows();
-		
+
 		{
 			if (nn == 1)
 			{
@@ -82,11 +81,11 @@ struct btVectorX
 			{
 				T scale = 0.0;
 				T ssq = 1.0;
-				
+
 				/* The following loop is equivalent to this call to the LAPACK
 				 auxiliary routine:   CALL SLASSQ( N, X, INCX, SCALE, SSQ ) */
-				
-				for (int ix=0;ix<nn;ix++)
+
+				for (int ix = 0; ix < nn; ix++)
 				{
 					if ((*this)[ix] != 0.0)
 					{
@@ -110,38 +109,36 @@ struct btVectorX
 			}
 		}
 		return norm;
-		
 	}
-	void	setZero()
+	void setZero()
 	{
 		if (m_storage.size())
 		{
 			//	for (int i=0;i<m_storage.size();i++)
 			//		m_storage[i]=0;
 			//memset(&m_storage[0],0,sizeof(T)*m_storage.size());
-			btSetZero(&m_storage[0],m_storage.size());
+			btSetZero(&m_storage[0], m_storage.size());
 		}
 	}
-	const T& operator[] (int index) const
+	const T& operator[](int index) const
 	{
 		return m_storage[index];
 	}
-	
-	T& operator[] (int index)
+
+	T& operator[](int index)
 	{
 		return m_storage[index];
 	}
-	
+
 	T* getBufferPointerWritable()
 	{
 		return m_storage.size() ? &m_storage[0] : 0;
 	}
-	
+
 	const T* getBufferPointer() const
 	{
 		return m_storage.size() ? &m_storage[0] : 0;
 	}
-	
 };
 /*
  template <typename T>
@@ -151,8 +148,7 @@ struct btVectorX
  }
  */
 
-
-template <typename T> 
+template <typename T>
 struct btMatrixX
 {
 	int m_rows;
@@ -161,10 +157,10 @@ struct btMatrixX
 	int m_resizeOperations;
 	int m_setElemOperations;
 
-	btAlignedObjectArray<T>	m_storage;
-	mutable btAlignedObjectArray< btAlignedObjectArray<int> > m_rowNonZeroElements1;
+	btAlignedObjectArray<T> m_storage;
+	mutable btAlignedObjectArray<btAlignedObjectArray<int> > m_rowNonZeroElements1;
 
-	T* getBufferPointerWritable() 
+	T* getBufferPointerWritable()
 	{
 		return m_storage.size() ? &m_storage[0] : 0;
 	}
@@ -174,21 +170,21 @@ struct btMatrixX
 		return m_storage.size() ? &m_storage[0] : 0;
 	}
 	btMatrixX()
-		:m_rows(0),
-		m_cols(0),
-		m_operations(0),
-		m_resizeOperations(0),
-		m_setElemOperations(0)
+		: m_rows(0),
+		  m_cols(0),
+		  m_operations(0),
+		  m_resizeOperations(0),
+		  m_setElemOperations(0)
 	{
 	}
-	btMatrixX(int rows,int cols)
-		:m_rows(rows),
-		m_cols(cols),
-		m_operations(0),
-		m_resizeOperations(0),
-		m_setElemOperations(0)
+	btMatrixX(int rows, int cols)
+		: m_rows(rows),
+		  m_cols(cols),
+		  m_operations(0),
+		  m_resizeOperations(0),
+		  m_setElemOperations(0)
 	{
-		resize(rows,cols);
+		resize(rows, cols);
 	}
 	void resize(int rows, int cols)
 	{
@@ -197,7 +193,7 @@ struct btMatrixX
 		m_cols = cols;
 		{
 			BT_PROFILE("m_storage.resize");
-			m_storage.resize(rows*cols);
+			m_storage.resize(rows * cols);
 		}
 	}
 	int cols() const
@@ -215,108 +211,99 @@ struct btMatrixX
 	}
 	*/
 
-	void addElem(int row,int col, T val)
+	void addElem(int row, int col, T val)
 	{
 		if (val)
 		{
-			if (m_storage[col+row*m_cols]==0.f)
+			if (m_storage[col + row * m_cols] == 0.f)
 			{
-				setElem(row,col,val);
-			} else
+				setElem(row, col, val);
+			}
+			else
 			{
-				m_storage[row*m_cols+col] += val;
+				m_storage[row * m_cols + col] += val;
 			}
 		}
 	}
-	
-	
-	void setElem(int row,int col, T val)
+
+	void setElem(int row, int col, T val)
 	{
 		m_setElemOperations++;
-		m_storage[row*m_cols+col] = val;
+		m_storage[row * m_cols + col] = val;
 	}
-	
-	void mulElem(int row,int col, T val)
+
+	void mulElem(int row, int col, T val)
 	{
 		m_setElemOperations++;
 		//mul doesn't change sparsity info
 
-		m_storage[row*m_cols+col] *= val;
+		m_storage[row * m_cols + col] *= val;
 	}
-	
-	
-	
-	
+
 	void copyLowerToUpperTriangle()
 	{
-		int count=0;
-		for (int row=0;row<rows();row++)
+		int count = 0;
+		for (int row = 0; row < rows(); row++)
 		{
-			for (int col=0;col<row;col++)
+			for (int col = 0; col < row; col++)
 			{
-				setElem(col,row, (*this)(row,col));
+				setElem(col, row, (*this)(row, col));
 				count++;
-				
 			}
 		}
 		//printf("copyLowerToUpperTriangle copied %d elements out of %dx%d=%d\n", count,rows(),cols(),cols()*rows());
 	}
-	
-	const T& operator() (int row,int col) const
-	{
-		return m_storage[col+row*m_cols];
-	}
 
+	const T& operator()(int row, int col) const
+	{
+		return m_storage[col + row * m_cols];
+	}
 
 	void setZero()
 	{
 		{
 			BT_PROFILE("storage=0");
-			btSetZero(&m_storage[0],m_storage.size());
+			btSetZero(&m_storage[0], m_storage.size());
 			//memset(&m_storage[0],0,sizeof(T)*m_storage.size());
 			//for (int i=0;i<m_storage.size();i++)
-	//			m_storage[i]=0;
+			//			m_storage[i]=0;
 		}
 	}
-	
+
 	void setIdentity()
 	{
 		btAssert(rows() == cols());
-		
+
 		setZero();
-		for (int row=0;row<rows();row++)
+		for (int row = 0; row < rows(); row++)
 		{
-			setElem(row,row,1);
+			setElem(row, row, 1);
 		}
 	}
 
-	
-
-	void	printMatrix(const char* msg)
+	void printMatrix(const char* msg)
 	{
-		printf("%s ---------------------\n",msg);
-		for (int i=0;i<rows();i++)
+		printf("%s ---------------------\n", msg);
+		for (int i = 0; i < rows(); i++)
 		{
 			printf("\n");
-			for (int j=0;j<cols();j++)
+			for (int j = 0; j < cols(); j++)
 			{
-				printf("%2.1f\t",(*this)(i,j));
+				printf("%2.1f\t", (*this)(i, j));
 			}
 		}
 		printf("\n---------------------\n");
-
 	}
 
-
 	void rowComputeNonZeroElements() const
 	{
 		m_rowNonZeroElements1.resize(rows());
-		for (int i=0;i<rows();i++)
+		for (int i = 0; i < rows(); i++)
 		{
 			m_rowNonZeroElements1[i].resize(0);
-			for (int j=0;j<cols();j++)
+			for (int j = 0; j < cols(); j++)
 			{
-				if ((*this)(i,j)!=0.f)
+				if ((*this)(i, j) != 0.f)
 				{
 					m_rowNonZeroElements1[i].push_back(j);
 				}
@@ -326,54 +313,52 @@ struct btMatrixX
 	btMatrixX transpose() const
 	{
 		//transpose is optimized for sparse matrices
-		btMatrixX tr(m_cols,m_rows);
+		btMatrixX tr(m_cols, m_rows);
 		tr.setZero();
-		for (int i=0;i<m_cols;i++)
-			for (int j=0;j<m_rows;j++)
+		for (int i = 0; i < m_cols; i++)
+			for (int j = 0; j < m_rows; j++)
 			{
-				T v = (*this)(j,i);
+				T v = (*this)(j, i);
 				if (v)
 				{
-					tr.setElem(i,j,v);
+					tr.setElem(i, j, v);
 				}
 			}
 		return tr;
 	}
 
-
 	btMatrixX operator*(const btMatrixX& other)
 	{
 		//btMatrixX*btMatrixX implementation, brute force
 		btAssert(cols() == other.rows());
 
-		btMatrixX res(rows(),other.cols());
+		btMatrixX res(rows(), other.cols());
 		res.setZero();
-//		BT_PROFILE("btMatrixX mul");
-		for (int j=0; j < res.cols(); ++j)
+		//		BT_PROFILE("btMatrixX mul");
+		for (int j = 0; j < res.cols(); ++j)
 		{
 			{
-				for (int i=0; i < res.rows(); ++i)
+				for (int i = 0; i < res.rows(); ++i)
 				{
-					T dotProd=0;
-//					T dotProd2=0;
+					T dotProd = 0;
+					//					T dotProd2=0;
 					//int waste=0,waste2=0;
 
 					{
-//						bool useOtherCol = true;
+						//						bool useOtherCol = true;
 						{
-							for (int v=0;v<rows();v++)
+							for (int v = 0; v < rows(); v++)
 							{
-								T w = (*this)(i,v);
-								if (other(v,j)!=0.f)
+								T w = (*this)(i, v);
+								if (other(v, j) != 0.f)
 								{
-									dotProd+=w*other(v,j);	
+									dotProd += w * other(v, j);
 								}
-						
 							}
 						}
 					}
 					if (dotProd)
-						res.setElem(i,j,dotProd);
+						res.setElem(i, j, dotProd);
 				}
 			}
 		}
@@ -381,174 +366,166 @@ struct btMatrixX
 	}
 
 	// this assumes the 4th and 8th rows of B and C are zero.
-	void multiplyAdd2_p8r (const btScalar *B, const btScalar *C,  int numRows,  int numRowsOther ,int row, int col)
+	void multiplyAdd2_p8r(const btScalar* B, const btScalar* C, int numRows, int numRowsOther, int row, int col)
 	{
-		const btScalar *bb = B;
-		for ( int i = 0;i<numRows;i++)
+		const btScalar* bb = B;
+		for (int i = 0; i < numRows; i++)
 		{
-			const btScalar *cc = C;
-			for ( int j = 0;j<numRowsOther;j++)
+			const btScalar* cc = C;
+			for (int j = 0; j < numRowsOther; j++)
 			{
 				btScalar sum;
-				sum  = bb[0]*cc[0];
-				sum += bb[1]*cc[1];
-				sum += bb[2]*cc[2];
-				sum += bb[4]*cc[4];
-				sum += bb[5]*cc[5];
-				sum += bb[6]*cc[6];
-				addElem(row+i,col+j,sum);
+				sum = bb[0] * cc[0];
+				sum += bb[1] * cc[1];
+				sum += bb[2] * cc[2];
+				sum += bb[4] * cc[4];
+				sum += bb[5] * cc[5];
+				sum += bb[6] * cc[6];
+				addElem(row + i, col + j, sum);
 				cc += 8;
 			}
 			bb += 8;
 		}
 	}
 
-	void multiply2_p8r (const btScalar *B, const btScalar *C,  int numRows,  int numRowsOther, int row, int col)
+	void multiply2_p8r(const btScalar* B, const btScalar* C, int numRows, int numRowsOther, int row, int col)
 	{
-		btAssert (numRows>0 && numRowsOther>0 && B && C);
-		const btScalar *bb = B;
-		for ( int i = 0;i<numRows;i++)
+		btAssert(numRows > 0 && numRowsOther > 0 && B && C);
+		const btScalar* bb = B;
+		for (int i = 0; i < numRows; i++)
 		{
-			const btScalar *cc = C;
-			for ( int j = 0;j<numRowsOther;j++)
+			const btScalar* cc = C;
+			for (int j = 0; j < numRowsOther; j++)
 			{
 				btScalar sum;
-				sum  = bb[0]*cc[0];
-				sum += bb[1]*cc[1];
-				sum += bb[2]*cc[2];
-				sum += bb[4]*cc[4];
-				sum += bb[5]*cc[5];
-				sum += bb[6]*cc[6];
-				setElem(row+i,col+j,sum);
+				sum = bb[0] * cc[0];
+				sum += bb[1] * cc[1];
+				sum += bb[2] * cc[2];
+				sum += bb[4] * cc[4];
+				sum += bb[5] * cc[5];
+				sum += bb[6] * cc[6];
+				setElem(row + i, col + j, sum);
 				cc += 8;
 			}
 			bb += 8;
 		}
 	}
-	
-	void setSubMatrix(int rowstart,int colstart,int rowend,int colend,const T value)
+
+	void setSubMatrix(int rowstart, int colstart, int rowend, int colend, const T value)
 	{
-		int numRows = rowend+1-rowstart;
-		int numCols = colend+1-colstart;
-		
-		for (int row=0;row<numRows;row++)
+		int numRows = rowend + 1 - rowstart;
+		int numCols = colend + 1 - colstart;
+
+		for (int row = 0; row < numRows; row++)
 		{
-			for (int col=0;col<numCols;col++)
+			for (int col = 0; col < numCols; col++)
 			{
-				setElem(rowstart+row,colstart+col,value);
+				setElem(rowstart + row, colstart + col, value);
 			}
 		}
 	}
-	
-	void setSubMatrix(int rowstart,int colstart,int rowend,int colend,const btMatrixX& block)
+
+	void setSubMatrix(int rowstart, int colstart, int rowend, int colend, const btMatrixX& block)
 	{
-		btAssert(rowend+1-rowstart == block.rows());
-		btAssert(colend+1-colstart == block.cols());
-		for (int row=0;row<block.rows();row++)
+		btAssert(rowend + 1 - rowstart == block.rows());
+		btAssert(colend + 1 - colstart == block.cols());
+		for (int row = 0; row < block.rows(); row++)
 		{
-			for (int col=0;col<block.cols();col++)
+			for (int col = 0; col < block.cols(); col++)
 			{
-				setElem(rowstart+row,colstart+col,block(row,col));
+				setElem(rowstart + row, colstart + col, block(row, col));
 			}
 		}
 	}
-	void setSubMatrix(int rowstart,int colstart,int rowend,int colend,const btVectorX<T>& block)
+	void setSubMatrix(int rowstart, int colstart, int rowend, int colend, const btVectorX<T>& block)
 	{
-		btAssert(rowend+1-rowstart == block.rows());
-		btAssert(colend+1-colstart == block.cols());
-		for (int row=0;row<block.rows();row++)
+		btAssert(rowend + 1 - rowstart == block.rows());
+		btAssert(colend + 1 - colstart == block.cols());
+		for (int row = 0; row < block.rows(); row++)
 		{
-			for (int col=0;col<block.cols();col++)
+			for (int col = 0; col < block.cols(); col++)
 			{
-				setElem(rowstart+row,colstart+col,block[row]);
+				setElem(rowstart + row, colstart + col, block[row]);
 			}
 		}
 	}
-	
-	
+
 	btMatrixX negative()
 	{
-		btMatrixX neg(rows(),cols());
-		for (int i=0;i<rows();i++)
-			for (int j=0;j<cols();j++)
+		btMatrixX neg(rows(), cols());
+		for (int i = 0; i < rows(); i++)
+			for (int j = 0; j < cols(); j++)
 			{
-				T v = (*this)(i,j);
-				neg.setElem(i,j,-v);
+				T v = (*this)(i, j);
+				neg.setElem(i, j, -v);
 			}
 		return neg;
 	}
-
 };
 
-
-
 typedef btMatrixX<float> btMatrixXf;
 typedef btVectorX<float> btVectorXf;
 
 typedef btMatrixX<double> btMatrixXd;
 typedef btVectorX<double> btVectorXd;
 
-
 #ifdef BT_DEBUG_OSTREAM
-template <typename T> 
-std::ostream& operator<< (std::ostream& os, const btMatrixX<T>& mat)
+template <typename T>
+std::ostream& operator<<(std::ostream& os, const btMatrixX<T>& mat)
+{
+	os << " [";
+	//printf("%s ---------------------\n",msg);
+	for (int i = 0; i < mat.rows(); i++)
 	{
-		
-		os << " [";
-		//printf("%s ---------------------\n",msg);
-		for (int i=0;i<mat.rows();i++)
+		for (int j = 0; j < mat.cols(); j++)
 		{
-			for (int j=0;j<mat.cols();j++)
-			{
-				os << std::setw(12) << mat(i,j);
-			}
-			if (i!=mat.rows()-1)
-				os << std::endl << "  ";
+			os << std::setw(12) << mat(i, j);
 		}
-		os << " ]";
-		//printf("\n---------------------\n");
-
-		return os;
+		if (i != mat.rows() - 1)
+			os << std::endl
+			   << "  ";
 	}
-template <typename T> 
-std::ostream& operator<< (std::ostream& os, const btVectorX<T>& mat)
+	os << " ]";
+	//printf("\n---------------------\n");
+
+	return os;
+}
+template <typename T>
+std::ostream& operator<<(std::ostream& os, const btVectorX<T>& mat)
+{
+	os << " [";
+	//printf("%s ---------------------\n",msg);
+	for (int i = 0; i < mat.rows(); i++)
 	{
-		
-		os << " [";
-		//printf("%s ---------------------\n",msg);
-		for (int i=0;i<mat.rows();i++)
-		{
-				os << std::setw(12) << mat[i];
-			if (i!=mat.rows()-1)
-				os << std::endl << "  ";
-		}
-		os << " ]";
-		//printf("\n---------------------\n");
-
-		return os;
+		os << std::setw(12) << mat[i];
+		if (i != mat.rows() - 1)
+			os << std::endl
+			   << "  ";
 	}
+	os << " ]";
+	//printf("\n---------------------\n");
 
-#endif //BT_DEBUG_OSTREAM
+	return os;
+}
 
+#endif  //BT_DEBUG_OSTREAM
 
 inline void setElem(btMatrixXd& mat, int row, int col, double val)
 {
-	mat.setElem(row,col,val);
+	mat.setElem(row, col, val);
 }
 
 inline void setElem(btMatrixXf& mat, int row, int col, float val)
 {
-	mat.setElem(row,col,val);
+	mat.setElem(row, col, val);
 }
 
 #ifdef BT_USE_DOUBLE_PRECISION
-	#define btVectorXu btVectorXd
-	#define btMatrixXu btMatrixXd
+#define btVectorXu btVectorXd
+#define btMatrixXu btMatrixXd
 #else
-	#define btVectorXu btVectorXf
-	#define btMatrixXu btMatrixXf
-#endif //BT_USE_DOUBLE_PRECISION
+#define btVectorXu btVectorXf
+#define btMatrixXu btMatrixXf
+#endif  //BT_USE_DOUBLE_PRECISION
 
-
-
-#endif//BT_MATRIX_H_H
+#endif  //BT_MATRIX_H_H
diff --git a/thirdparty/bullet/LinearMath/btMinMax.h b/thirdparty/bullet/LinearMath/btMinMax.h
index 5b436e9ba4e..92fea0275ab 100644
--- a/thirdparty/bullet/LinearMath/btMinMax.h
+++ b/thirdparty/bullet/LinearMath/btMinMax.h
@@ -12,60 +12,58 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #ifndef BT_GEN_MINMAX_H
 #define BT_GEN_MINMAX_H
 
 #include "btScalar.h"
 
 template <class T>
-SIMD_FORCE_INLINE const T& btMin(const T& a, const T& b) 
+SIMD_FORCE_INLINE const T& btMin(const T& a, const T& b)
 {
-  return a < b ? a : b ;
+	return a < b ? a : b;
 }
 
 template <class T>
-SIMD_FORCE_INLINE const T& btMax(const T& a, const T& b) 
+SIMD_FORCE_INLINE const T& btMax(const T& a, const T& b)
 {
-  return  a > b ? a : b;
+	return a > b ? a : b;
 }
 
 template <class T>
-SIMD_FORCE_INLINE const T& btClamped(const T& a, const T& lb, const T& ub) 
+SIMD_FORCE_INLINE const T& btClamped(const T& a, const T& lb, const T& ub)
 {
-	return a < lb ? lb : (ub < a ? ub : a); 
+	return a < lb ? lb : (ub < a ? ub : a);
 }
 
 template <class T>
-SIMD_FORCE_INLINE void btSetMin(T& a, const T& b) 
+SIMD_FORCE_INLINE void btSetMin(T& a, const T& b)
 {
-    if (b < a) 
+	if (b < a)
 	{
 		a = b;
 	}
 }
 
 template <class T>
-SIMD_FORCE_INLINE void btSetMax(T& a, const T& b) 
+SIMD_FORCE_INLINE void btSetMax(T& a, const T& b)
 {
-    if (a < b) 
+	if (a < b)
 	{
 		a = b;
 	}
 }
 
 template <class T>
-SIMD_FORCE_INLINE void btClamp(T& a, const T& lb, const T& ub) 
+SIMD_FORCE_INLINE void btClamp(T& a, const T& lb, const T& ub)
 {
-	if (a < lb) 
+	if (a < lb)
 	{
-		a = lb; 
+		a = lb;
 	}
-	else if (ub < a) 
+	else if (ub < a)
 	{
 		a = ub;
 	}
 }
 
-#endif //BT_GEN_MINMAX_H
+#endif  //BT_GEN_MINMAX_H
diff --git a/thirdparty/bullet/LinearMath/btMotionState.h b/thirdparty/bullet/LinearMath/btMotionState.h
index 94318140902..ae6a51611d8 100644
--- a/thirdparty/bullet/LinearMath/btMotionState.h
+++ b/thirdparty/bullet/LinearMath/btMotionState.h
@@ -20,21 +20,17 @@ subject to the following restrictions:
 
 ///The btMotionState interface class allows the dynamics world to synchronize and interpolate the updated world transforms with graphics
 ///For optimizations, potentially only moving objects get synchronized (using setWorldPosition/setWorldOrientation)
-class	btMotionState
+class btMotionState
 {
-	public:
-		
-		virtual ~btMotionState()
-		{
-			
-		}
-		
-		virtual void	getWorldTransform(btTransform& worldTrans ) const =0;
+public:
+	virtual ~btMotionState()
+	{
+	}
 
-		//Bullet only calls the update of worldtransform for active objects
-		virtual void	setWorldTransform(const btTransform& worldTrans)=0;
-		
-	
+	virtual void getWorldTransform(btTransform& worldTrans) const = 0;
+
+	//Bullet only calls the update of worldtransform for active objects
+	virtual void setWorldTransform(const btTransform& worldTrans) = 0;
 };
 
-#endif //BT_MOTIONSTATE_H
+#endif  //BT_MOTIONSTATE_H
diff --git a/thirdparty/bullet/LinearMath/btPolarDecomposition.cpp b/thirdparty/bullet/LinearMath/btPolarDecomposition.cpp
index b3664faa4ea..d9c72a8014a 100644
--- a/thirdparty/bullet/LinearMath/btPolarDecomposition.cpp
+++ b/thirdparty/bullet/LinearMath/btPolarDecomposition.cpp
@@ -3,96 +3,92 @@
 
 namespace
 {
-  btScalar abs_column_sum(const btMatrix3x3& a, int i)
-  {
-    return btFabs(a[0][i]) + btFabs(a[1][i]) + btFabs(a[2][i]);
-  }
-
-  btScalar abs_row_sum(const btMatrix3x3& a, int i)
-  {
-    return btFabs(a[i][0]) + btFabs(a[i][1]) + btFabs(a[i][2]);
-  }
-
-  btScalar p1_norm(const btMatrix3x3& a)
-  {
-    const btScalar sum0 = abs_column_sum(a,0);
-    const btScalar sum1 = abs_column_sum(a,1);
-    const btScalar sum2 = abs_column_sum(a,2);
-    return btMax(btMax(sum0, sum1), sum2);
-  }
-
-  btScalar pinf_norm(const btMatrix3x3& a)
-  {
-    const btScalar sum0 = abs_row_sum(a,0);
-    const btScalar sum1 = abs_row_sum(a,1);
-    const btScalar sum2 = abs_row_sum(a,2);
-    return btMax(btMax(sum0, sum1), sum2);
-  }
+btScalar abs_column_sum(const btMatrix3x3& a, int i)
+{
+	return btFabs(a[0][i]) + btFabs(a[1][i]) + btFabs(a[2][i]);
 }
 
+btScalar abs_row_sum(const btMatrix3x3& a, int i)
+{
+	return btFabs(a[i][0]) + btFabs(a[i][1]) + btFabs(a[i][2]);
+}
 
+btScalar p1_norm(const btMatrix3x3& a)
+{
+	const btScalar sum0 = abs_column_sum(a, 0);
+	const btScalar sum1 = abs_column_sum(a, 1);
+	const btScalar sum2 = abs_column_sum(a, 2);
+	return btMax(btMax(sum0, sum1), sum2);
+}
+
+btScalar pinf_norm(const btMatrix3x3& a)
+{
+	const btScalar sum0 = abs_row_sum(a, 0);
+	const btScalar sum1 = abs_row_sum(a, 1);
+	const btScalar sum2 = abs_row_sum(a, 2);
+	return btMax(btMax(sum0, sum1), sum2);
+}
+}  // namespace
 
 btPolarDecomposition::btPolarDecomposition(btScalar tolerance, unsigned int maxIterations)
-: m_tolerance(tolerance)
-, m_maxIterations(maxIterations)
+	: m_tolerance(tolerance), m_maxIterations(maxIterations)
 {
 }
 
 unsigned int btPolarDecomposition::decompose(const btMatrix3x3& a, btMatrix3x3& u, btMatrix3x3& h) const
 {
-  // Use the 'u' and 'h' matrices for intermediate calculations
-  u = a;
-  h = a.inverse();
+	// Use the 'u' and 'h' matrices for intermediate calculations
+	u = a;
+	h = a.inverse();
 
-  for (unsigned int i = 0; i < m_maxIterations; ++i)
-  {
-    const btScalar h_1 = p1_norm(h);
-    const btScalar h_inf = pinf_norm(h);
-    const btScalar u_1 = p1_norm(u);
-    const btScalar u_inf = pinf_norm(u);
+	for (unsigned int i = 0; i < m_maxIterations; ++i)
+	{
+		const btScalar h_1 = p1_norm(h);
+		const btScalar h_inf = pinf_norm(h);
+		const btScalar u_1 = p1_norm(u);
+		const btScalar u_inf = pinf_norm(u);
 
-    const btScalar h_norm = h_1 * h_inf;
-    const btScalar u_norm = u_1 * u_inf;
+		const btScalar h_norm = h_1 * h_inf;
+		const btScalar u_norm = u_1 * u_inf;
 
-    // The matrix is effectively singular so we cannot invert it
-    if (btFuzzyZero(h_norm) || btFuzzyZero(u_norm))
-      break;
+		// The matrix is effectively singular so we cannot invert it
+		if (btFuzzyZero(h_norm) || btFuzzyZero(u_norm))
+			break;
 
-    const btScalar gamma = btPow(h_norm / u_norm, 0.25f);
-    const btScalar inv_gamma = btScalar(1.0) / gamma;
+		const btScalar gamma = btPow(h_norm / u_norm, 0.25f);
+		const btScalar inv_gamma = btScalar(1.0) / gamma;
 
-    // Determine the delta to 'u'
-    const btMatrix3x3 delta = (u * (gamma - btScalar(2.0)) + h.transpose() * inv_gamma) * btScalar(0.5);
+		// Determine the delta to 'u'
+		const btMatrix3x3 delta = (u * (gamma - btScalar(2.0)) + h.transpose() * inv_gamma) * btScalar(0.5);
 
-    // Update the matrices
-    u += delta;
-    h = u.inverse();
+		// Update the matrices
+		u += delta;
+		h = u.inverse();
 
-    // Check for convergence
-    if (p1_norm(delta) <= m_tolerance * u_1)
-    {
-      h = u.transpose() * a;
-      h = (h + h.transpose()) * 0.5;
-      return i;
-    }
-  }
+		// Check for convergence
+		if (p1_norm(delta) <= m_tolerance * u_1)
+		{
+			h = u.transpose() * a;
+			h = (h + h.transpose()) * 0.5;
+			return i;
+		}
+	}
 
-  // The algorithm has failed to converge to the specified tolerance, but we
-  // want to make sure that the matrices returned are in the right form.
-  h = u.transpose() * a;
-  h = (h + h.transpose()) * 0.5;
+	// The algorithm has failed to converge to the specified tolerance, but we
+	// want to make sure that the matrices returned are in the right form.
+	h = u.transpose() * a;
+	h = (h + h.transpose()) * 0.5;
 
-  return m_maxIterations;
+	return m_maxIterations;
 }
 
 unsigned int btPolarDecomposition::maxIterations() const
 {
-  return m_maxIterations;
+	return m_maxIterations;
 }
 
 unsigned int polarDecompose(const btMatrix3x3& a, btMatrix3x3& u, btMatrix3x3& h)
 {
-  static btPolarDecomposition polar;
-  return polar.decompose(a, u, h);
+	static btPolarDecomposition polar;
+	return polar.decompose(a, u, h);
 }
-
diff --git a/thirdparty/bullet/LinearMath/btPolarDecomposition.h b/thirdparty/bullet/LinearMath/btPolarDecomposition.h
index 1feea0f78e8..bf29140a143 100644
--- a/thirdparty/bullet/LinearMath/btPolarDecomposition.h
+++ b/thirdparty/bullet/LinearMath/btPolarDecomposition.h
@@ -13,10 +13,8 @@
  */
 class btPolarDecomposition
 {
-  public:
-   
-
-    /**
+public:
+	/**
      * Creates an instance with optional parameters.
      *
      * @param tolerance     - the tolerance used to determine convergence of the
@@ -24,10 +22,10 @@ class btPolarDecomposition
      * @param maxIterations - the maximum number of iterations used to achieve
      *                        convergence
      */
-    btPolarDecomposition(btScalar tolerance = btScalar(0.0001),
-      unsigned int maxIterations = 16);
+	btPolarDecomposition(btScalar tolerance = btScalar(0.0001),
+						 unsigned int maxIterations = 16);
 
-    /**
+	/**
      * Decomposes a matrix into orthogonal and symmetric, positive-definite
      * parts. If the number of iterations returned by this function is equal to
      * the maximum number of iterations, the algorithm has failed to converge.
@@ -38,19 +36,19 @@ class btPolarDecomposition
      *
      * @return the number of iterations performed by the algorithm.
      */
-    unsigned int decompose(const btMatrix3x3& a, btMatrix3x3& u, btMatrix3x3& h) const; 
+	unsigned int decompose(const btMatrix3x3& a, btMatrix3x3& u, btMatrix3x3& h) const;
 
-    /**
+	/**
      * Returns the maximum number of iterations that this algorithm will perform
      * to achieve convergence.
      *
      * @return maximum number of iterations
      */
-    unsigned int maxIterations() const;
+	unsigned int maxIterations() const;
 
-  private:
-    btScalar m_tolerance;
-    unsigned int m_maxIterations;
+private:
+	btScalar m_tolerance;
+	unsigned int m_maxIterations;
 };
 
 /**
@@ -66,7 +64,6 @@ class btPolarDecomposition
  *
  * @return the number of iterations performed by the algorithm.
  */
-unsigned int polarDecompose(const btMatrix3x3& a, btMatrix3x3& u, btMatrix3x3& h); 
-
-#endif // POLARDECOMPOSITION_H
+unsigned int polarDecompose(const btMatrix3x3& a, btMatrix3x3& u, btMatrix3x3& h);
 
+#endif  // POLARDECOMPOSITION_H
diff --git a/thirdparty/bullet/LinearMath/btPoolAllocator.h b/thirdparty/bullet/LinearMath/btPoolAllocator.h
index efdeda8ffc4..4e7b49660a9 100644
--- a/thirdparty/bullet/LinearMath/btPoolAllocator.h
+++ b/thirdparty/bullet/LinearMath/btPoolAllocator.h
@@ -12,7 +12,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef _BT_POOL_ALLOCATOR_H
 #define _BT_POOL_ALLOCATOR_H
 
@@ -23,38 +22,38 @@ subject to the following restrictions:
 ///The btPoolAllocator class allows to efficiently allocate a large pool of objects, instead of dynamically allocating them separately.
 class btPoolAllocator
 {
-	int				m_elemSize;
-	int				m_maxElements;
-	int				m_freeCount;
-	void*			m_firstFree;
-	unsigned char*	m_pool;
-    btSpinMutex     m_mutex;  // only used if BT_THREADSAFE
+	int m_elemSize;
+	int m_maxElements;
+	int m_freeCount;
+	void* m_firstFree;
+	unsigned char* m_pool;
+	btSpinMutex m_mutex;  // only used if BT_THREADSAFE
 
 public:
-
 	btPoolAllocator(int elemSize, int maxElements)
-		:m_elemSize(elemSize),
-		m_maxElements(maxElements)
+		: m_elemSize(elemSize),
+		  m_maxElements(maxElements)
 	{
-		m_pool = (unsigned char*) btAlignedAlloc( static_cast<unsigned int>(m_elemSize*m_maxElements),16);
+		m_pool = (unsigned char*)btAlignedAlloc(static_cast<unsigned int>(m_elemSize * m_maxElements), 16);
 
 		unsigned char* p = m_pool;
-        m_firstFree = p;
-        m_freeCount = m_maxElements;
-        int count = m_maxElements;
-        while (--count) {
-            *(void**)p = (p + m_elemSize);
-            p += m_elemSize;
-        }
-        *(void**)p = 0;
-    }
+		m_firstFree = p;
+		m_freeCount = m_maxElements;
+		int count = m_maxElements;
+		while (--count)
+		{
+			*(void**)p = (p + m_elemSize);
+			p += m_elemSize;
+		}
+		*(void**)p = 0;
+	}
 
 	~btPoolAllocator()
 	{
-		btAlignedFree( m_pool);
+		btAlignedFree(m_pool);
 	}
 
-	int	getFreeCount() const
+	int getFreeCount() const
 	{
 		return m_freeCount;
 	}
@@ -69,26 +68,27 @@ public:
 		return m_maxElements;
 	}
 
-	void*	allocate(int size)
+	void* allocate(int size)
 	{
 		// release mode fix
 		(void)size;
-        btMutexLock(&m_mutex);
-		btAssert(!size || size<=m_elemSize);
+		btMutexLock(&m_mutex);
+		btAssert(!size || size <= m_elemSize);
 		//btAssert(m_freeCount>0);  // should return null if all full
-        void* result = m_firstFree;
-        if (NULL != m_firstFree)
-        {
-            m_firstFree = *(void**)m_firstFree;
-            --m_freeCount;
-        }
-        btMutexUnlock(&m_mutex);
-        return result;
+		void* result = m_firstFree;
+		if (NULL != m_firstFree)
+		{
+			m_firstFree = *(void**)m_firstFree;
+			--m_freeCount;
+		}
+		btMutexUnlock(&m_mutex);
+		return result;
 	}
 
 	bool validPtr(void* ptr)
 	{
-		if (ptr) {
+		if (ptr)
+		{
 			if (((unsigned char*)ptr >= m_pool && (unsigned char*)ptr < m_pool + m_maxElements * m_elemSize))
 			{
 				return true;
@@ -97,34 +97,34 @@ public:
 		return false;
 	}
 
-	void	freeMemory(void* ptr)
+	void freeMemory(void* ptr)
 	{
-		 if (ptr) {
-            btAssert((unsigned char*)ptr >= m_pool && (unsigned char*)ptr < m_pool + m_maxElements * m_elemSize);
+		if (ptr)
+		{
+			btAssert((unsigned char*)ptr >= m_pool && (unsigned char*)ptr < m_pool + m_maxElements * m_elemSize);
 
-            btMutexLock(&m_mutex);
-            *(void**)ptr = m_firstFree;
-            m_firstFree = ptr;
-            ++m_freeCount;
-            btMutexUnlock(&m_mutex);
-        }
+			btMutexLock(&m_mutex);
+			*(void**)ptr = m_firstFree;
+			m_firstFree = ptr;
+			++m_freeCount;
+			btMutexUnlock(&m_mutex);
+		}
 	}
 
-	int	getElementSize() const
+	int getElementSize() const
 	{
 		return m_elemSize;
 	}
 
-	unsigned char*	getPoolAddress()
+	unsigned char* getPoolAddress()
 	{
 		return m_pool;
 	}
 
-	const unsigned char*	getPoolAddress() const
+	const unsigned char* getPoolAddress() const
 	{
 		return m_pool;
 	}
-
 };
 
-#endif //_BT_POOL_ALLOCATOR_H
+#endif  //_BT_POOL_ALLOCATOR_H
diff --git a/thirdparty/bullet/LinearMath/btQuadWord.h b/thirdparty/bullet/LinearMath/btQuadWord.h
index fcfb3be4447..ab2d3175ad6 100644
--- a/thirdparty/bullet/LinearMath/btQuadWord.h
+++ b/thirdparty/bullet/LinearMath/btQuadWord.h
@@ -12,18 +12,13 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef BT_SIMD_QUADWORD_H
 #define BT_SIMD_QUADWORD_H
 
 #include "btScalar.h"
 #include "btMinMax.h"
 
-
-
-
-
-#if defined (__CELLOS_LV2) && defined (__SPU__)
+#if defined(__CELLOS_LV2) && defined(__SPU__)
 #include <altivec.h>
 #endif
 
@@ -31,51 +26,53 @@ subject to the following restrictions:
  * Some issues under PS3 Linux with IBM 2.1 SDK, gcc compiler prevent from using aligned quadword.
  */
 #ifndef USE_LIBSPE2
-ATTRIBUTE_ALIGNED16(class) btQuadWord
+ATTRIBUTE_ALIGNED16(class)
+btQuadWord
 #else
 class btQuadWord
 #endif
 {
 protected:
-
-#if defined (__SPU__) && defined (__CELLOS_LV2__)
+#if defined(__SPU__) && defined(__CELLOS_LV2__)
 	union {
 		vec_float4 mVec128;
-		btScalar	m_floats[4];
+		btScalar m_floats[4];
 	};
+
 public:
-	vec_float4	get128() const
+	vec_float4 get128() const
 	{
 		return mVec128;
 	}
-protected:
-#else //__CELLOS_LV2__ __SPU__
 
-#if defined(BT_USE_SSE) || defined(BT_USE_NEON) 
+protected:
+#else  //__CELLOS_LV2__ __SPU__
+
+#if defined(BT_USE_SSE) || defined(BT_USE_NEON)
 	union {
 		btSimdFloat4 mVec128;
-		btScalar	m_floats[4];
+		btScalar m_floats[4];
 	};
+
 public:
-	SIMD_FORCE_INLINE	btSimdFloat4	get128() const
+	SIMD_FORCE_INLINE btSimdFloat4 get128() const
 	{
 		return mVec128;
 	}
-	SIMD_FORCE_INLINE	void	set128(btSimdFloat4 v128)
+	SIMD_FORCE_INLINE void set128(btSimdFloat4 v128)
 	{
 		mVec128 = v128;
 	}
 #else
-	btScalar	m_floats[4];
-#endif // BT_USE_SSE
+	btScalar m_floats[4];
+#endif  // BT_USE_SSE
 
-#endif //__CELLOS_LV2__ __SPU__
+#endif  //__CELLOS_LV2__ __SPU__
 
-	public:
-  
+public:
 #if (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)) || defined(BT_USE_NEON)
 
-	// Set Vector 
+	// Set Vector
 	SIMD_FORCE_INLINE btQuadWord(const btSimdFloat4 vec)
 	{
 		mVec128 = vec;
@@ -88,157 +85,154 @@ public:
 	}
 
 	// Assignment Operator
-	SIMD_FORCE_INLINE btQuadWord& 
-	operator=(const btQuadWord& v) 
+	SIMD_FORCE_INLINE btQuadWord&
+	operator=(const btQuadWord& v)
 	{
 		mVec128 = v.mVec128;
-		
+
 		return *this;
 	}
-	
+
 #endif
 
-  /**@brief Return the x value */
-		SIMD_FORCE_INLINE const btScalar& getX() const { return m_floats[0]; }
-  /**@brief Return the y value */
-		SIMD_FORCE_INLINE const btScalar& getY() const { return m_floats[1]; }
-  /**@brief Return the z value */
-		SIMD_FORCE_INLINE const btScalar& getZ() const { return m_floats[2]; }
-  /**@brief Set the x value */
-		SIMD_FORCE_INLINE void	setX(btScalar _x) { m_floats[0] = _x;};
-  /**@brief Set the y value */
-		SIMD_FORCE_INLINE void	setY(btScalar _y) { m_floats[1] = _y;};
-  /**@brief Set the z value */
-		SIMD_FORCE_INLINE void	setZ(btScalar _z) { m_floats[2] = _z;};
-  /**@brief Set the w value */
-		SIMD_FORCE_INLINE void	setW(btScalar _w) { m_floats[3] = _w;};
-  /**@brief Return the x value */
-		SIMD_FORCE_INLINE const btScalar& x() const { return m_floats[0]; }
-  /**@brief Return the y value */
-		SIMD_FORCE_INLINE const btScalar& y() const { return m_floats[1]; }
-  /**@brief Return the z value */
-		SIMD_FORCE_INLINE const btScalar& z() const { return m_floats[2]; }
-  /**@brief Return the w value */
-		SIMD_FORCE_INLINE const btScalar& w() const { return m_floats[3]; }
+	/**@brief Return the x value */
+	SIMD_FORCE_INLINE const btScalar& getX() const { return m_floats[0]; }
+	/**@brief Return the y value */
+	SIMD_FORCE_INLINE const btScalar& getY() const { return m_floats[1]; }
+	/**@brief Return the z value */
+	SIMD_FORCE_INLINE const btScalar& getZ() const { return m_floats[2]; }
+	/**@brief Set the x value */
+	SIMD_FORCE_INLINE void setX(btScalar _x) { m_floats[0] = _x; };
+	/**@brief Set the y value */
+	SIMD_FORCE_INLINE void setY(btScalar _y) { m_floats[1] = _y; };
+	/**@brief Set the z value */
+	SIMD_FORCE_INLINE void setZ(btScalar _z) { m_floats[2] = _z; };
+	/**@brief Set the w value */
+	SIMD_FORCE_INLINE void setW(btScalar _w) { m_floats[3] = _w; };
+	/**@brief Return the x value */
+	SIMD_FORCE_INLINE const btScalar& x() const { return m_floats[0]; }
+	/**@brief Return the y value */
+	SIMD_FORCE_INLINE const btScalar& y() const { return m_floats[1]; }
+	/**@brief Return the z value */
+	SIMD_FORCE_INLINE const btScalar& z() const { return m_floats[2]; }
+	/**@brief Return the w value */
+	SIMD_FORCE_INLINE const btScalar& w() const { return m_floats[3]; }
 
-	//SIMD_FORCE_INLINE btScalar&       operator[](int i)       { return (&m_floats[0])[i];	}      
+	//SIMD_FORCE_INLINE btScalar&       operator[](int i)       { return (&m_floats[0])[i];	}
 	//SIMD_FORCE_INLINE const btScalar& operator[](int i) const { return (&m_floats[0])[i]; }
 	///operator btScalar*() replaces operator[], using implicit conversion. We added operator != and operator == to avoid pointer comparisons.
-	SIMD_FORCE_INLINE	operator       btScalar *()       { return &m_floats[0]; }
-	SIMD_FORCE_INLINE	operator const btScalar *() const { return &m_floats[0]; }
+	SIMD_FORCE_INLINE operator btScalar*() { return &m_floats[0]; }
+	SIMD_FORCE_INLINE operator const btScalar*() const { return &m_floats[0]; }
 
-	SIMD_FORCE_INLINE	bool	operator==(const btQuadWord& other) const
+	SIMD_FORCE_INLINE bool operator==(const btQuadWord& other) const
 	{
 #ifdef BT_USE_SSE
-        return (0xf == _mm_movemask_ps((__m128)_mm_cmpeq_ps(mVec128, other.mVec128)));
-#else 
-		return ((m_floats[3]==other.m_floats[3]) && 
-                (m_floats[2]==other.m_floats[2]) && 
-                (m_floats[1]==other.m_floats[1]) && 
-                (m_floats[0]==other.m_floats[0]));
+		return (0xf == _mm_movemask_ps((__m128)_mm_cmpeq_ps(mVec128, other.mVec128)));
+#else
+		return ((m_floats[3] == other.m_floats[3]) &&
+				(m_floats[2] == other.m_floats[2]) &&
+				(m_floats[1] == other.m_floats[1]) &&
+				(m_floats[0] == other.m_floats[0]));
 #endif
 	}
 
-	SIMD_FORCE_INLINE	bool	operator!=(const btQuadWord& other) const
+	SIMD_FORCE_INLINE bool operator!=(const btQuadWord& other) const
 	{
 		return !(*this == other);
 	}
 
-  /**@brief Set x,y,z and zero w 
+	/**@brief Set x,y,z and zero w 
    * @param x Value of x
    * @param y Value of y
    * @param z Value of z
    */
-		SIMD_FORCE_INLINE void 	setValue(const btScalar& _x, const btScalar& _y, const btScalar& _z)
-		{
-			m_floats[0]=_x;
-			m_floats[1]=_y;
-			m_floats[2]=_z;
-			m_floats[3] = 0.f;
-		}
+	SIMD_FORCE_INLINE void setValue(const btScalar& _x, const btScalar& _y, const btScalar& _z)
+	{
+		m_floats[0] = _x;
+		m_floats[1] = _y;
+		m_floats[2] = _z;
+		m_floats[3] = 0.f;
+	}
 
-/*		void getValue(btScalar *m) const 
+	/*		void getValue(btScalar *m) const 
 		{
 			m[0] = m_floats[0];
 			m[1] = m_floats[1];
 			m[2] = m_floats[2];
 		}
 */
-/**@brief Set the values 
+	/**@brief Set the values 
    * @param x Value of x
    * @param y Value of y
    * @param z Value of z
    * @param w Value of w
    */
-		SIMD_FORCE_INLINE void	setValue(const btScalar& _x, const btScalar& _y, const btScalar& _z,const btScalar& _w)
-		{
-			m_floats[0]=_x;
-			m_floats[1]=_y;
-			m_floats[2]=_z;
-			m_floats[3]=_w;
-		}
-  /**@brief No initialization constructor */
-		SIMD_FORCE_INLINE btQuadWord()
-		//	:m_floats[0](btScalar(0.)),m_floats[1](btScalar(0.)),m_floats[2](btScalar(0.)),m_floats[3](btScalar(0.))
-		{
-		}
- 
-  /**@brief Three argument constructor (zeros w)
+	SIMD_FORCE_INLINE void setValue(const btScalar& _x, const btScalar& _y, const btScalar& _z, const btScalar& _w)
+	{
+		m_floats[0] = _x;
+		m_floats[1] = _y;
+		m_floats[2] = _z;
+		m_floats[3] = _w;
+	}
+	/**@brief No initialization constructor */
+	SIMD_FORCE_INLINE btQuadWord()
+	//	:m_floats[0](btScalar(0.)),m_floats[1](btScalar(0.)),m_floats[2](btScalar(0.)),m_floats[3](btScalar(0.))
+	{
+	}
+
+	/**@brief Three argument constructor (zeros w)
    * @param x Value of x
    * @param y Value of y
    * @param z Value of z
    */
-		SIMD_FORCE_INLINE btQuadWord(const btScalar& _x, const btScalar& _y, const btScalar& _z)		
-		{
-			m_floats[0] = _x, m_floats[1] = _y, m_floats[2] = _z, m_floats[3] = 0.0f;
-		}
+	SIMD_FORCE_INLINE btQuadWord(const btScalar& _x, const btScalar& _y, const btScalar& _z)
+	{
+		m_floats[0] = _x, m_floats[1] = _y, m_floats[2] = _z, m_floats[3] = 0.0f;
+	}
 
-/**@brief Initializing constructor
+	/**@brief Initializing constructor
    * @param x Value of x
    * @param y Value of y
    * @param z Value of z
    * @param w Value of w
    */
-		SIMD_FORCE_INLINE btQuadWord(const btScalar& _x, const btScalar& _y, const btScalar& _z,const btScalar& _w) 
-		{
-			m_floats[0] = _x, m_floats[1] = _y, m_floats[2] = _z, m_floats[3] = _w;
-		}
+	SIMD_FORCE_INLINE btQuadWord(const btScalar& _x, const btScalar& _y, const btScalar& _z, const btScalar& _w)
+	{
+		m_floats[0] = _x, m_floats[1] = _y, m_floats[2] = _z, m_floats[3] = _w;
+	}
 
-  /**@brief Set each element to the max of the current values and the values of another btQuadWord
+	/**@brief Set each element to the max of the current values and the values of another btQuadWord
    * @param other The other btQuadWord to compare with 
    */
-		SIMD_FORCE_INLINE void	setMax(const btQuadWord& other)
-		{
-        #ifdef BT_USE_SSE
-            mVec128 = _mm_max_ps(mVec128, other.mVec128);
-        #elif defined(BT_USE_NEON)
-            mVec128 = vmaxq_f32(mVec128, other.mVec128);
-        #else
-        	btSetMax(m_floats[0], other.m_floats[0]);
-			btSetMax(m_floats[1], other.m_floats[1]);
-			btSetMax(m_floats[2], other.m_floats[2]);
-			btSetMax(m_floats[3], other.m_floats[3]);
-		#endif
-        }
-  /**@brief Set each element to the min of the current values and the values of another btQuadWord
+	SIMD_FORCE_INLINE void setMax(const btQuadWord& other)
+	{
+#ifdef BT_USE_SSE
+		mVec128 = _mm_max_ps(mVec128, other.mVec128);
+#elif defined(BT_USE_NEON)
+		mVec128 = vmaxq_f32(mVec128, other.mVec128);
+#else
+		btSetMax(m_floats[0], other.m_floats[0]);
+		btSetMax(m_floats[1], other.m_floats[1]);
+		btSetMax(m_floats[2], other.m_floats[2]);
+		btSetMax(m_floats[3], other.m_floats[3]);
+#endif
+	}
+	/**@brief Set each element to the min of the current values and the values of another btQuadWord
    * @param other The other btQuadWord to compare with 
    */
-		SIMD_FORCE_INLINE void	setMin(const btQuadWord& other)
-		{
-        #ifdef BT_USE_SSE
-            mVec128 = _mm_min_ps(mVec128, other.mVec128);
-        #elif defined(BT_USE_NEON)
-            mVec128 = vminq_f32(mVec128, other.mVec128);
-        #else
-        	btSetMin(m_floats[0], other.m_floats[0]);
-			btSetMin(m_floats[1], other.m_floats[1]);
-			btSetMin(m_floats[2], other.m_floats[2]);
-			btSetMin(m_floats[3], other.m_floats[3]);
-		#endif
-        }
-
-
-
+	SIMD_FORCE_INLINE void setMin(const btQuadWord& other)
+	{
+#ifdef BT_USE_SSE
+		mVec128 = _mm_min_ps(mVec128, other.mVec128);
+#elif defined(BT_USE_NEON)
+		mVec128 = vminq_f32(mVec128, other.mVec128);
+#else
+		btSetMin(m_floats[0], other.m_floats[0]);
+		btSetMin(m_floats[1], other.m_floats[1]);
+		btSetMin(m_floats[2], other.m_floats[2]);
+		btSetMin(m_floats[3], other.m_floats[3]);
+#endif
+	}
 };
 
-#endif //BT_SIMD_QUADWORD_H
+#endif  //BT_SIMD_QUADWORD_H
diff --git a/thirdparty/bullet/LinearMath/btQuaternion.h b/thirdparty/bullet/LinearMath/btQuaternion.h
index a98fec7bc47..53e8169b805 100644
--- a/thirdparty/bullet/LinearMath/btQuaternion.h
+++ b/thirdparty/bullet/LinearMath/btQuaternion.h
@@ -12,25 +12,19 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #ifndef BT_SIMD__QUATERNION_H_
 #define BT_SIMD__QUATERNION_H_
 
-
 #include "btVector3.h"
 #include "btQuadWord.h"
 
-
 #ifdef BT_USE_DOUBLE_PRECISION
 #define btQuaternionData btQuaternionDoubleData
 #define btQuaternionDataName "btQuaternionDoubleData"
 #else
 #define btQuaternionData btQuaternionFloatData
 #define btQuaternionDataName "btQuaternionFloatData"
-#endif //BT_USE_DOUBLE_PRECISION
-
-
+#endif  //BT_USE_DOUBLE_PRECISION
 
 #ifdef BT_USE_SSE
 
@@ -39,7 +33,7 @@ subject to the following restrictions:
 
 #endif
 
-#if defined(BT_USE_SSE) 
+#if defined(BT_USE_SSE)
 
 #define vQInv (_mm_set_ps(+0.0f, -0.0f, -0.0f, -0.0f))
 #define vPPPM (_mm_set_ps(-0.0f, +0.0f, +0.0f, +0.0f))
@@ -52,13 +46,14 @@ const btSimdFloat4 ATTRIBUTE_ALIGNED16(vPPPM) = {+0.0f, +0.0f, +0.0f, -0.0f};
 #endif
 
 /**@brief The btQuaternion implements quaternion to perform linear algebra rotations in combination with btMatrix3x3, btVector3 and btTransform. */
-class btQuaternion : public btQuadWord {
+class btQuaternion : public btQuadWord
+{
 public:
-  /**@brief No initialization constructor */
+	/**@brief No initialization constructor */
 	btQuaternion() {}
 
-#if (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE))|| defined(BT_USE_NEON) 
-	// Set Vector 
+#if (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)) || defined(BT_USE_NEON)
+	// Set Vector
 	SIMD_FORCE_INLINE btQuaternion(const btSimdFloat4 vec)
 	{
 		mVec128 = vec;
@@ -71,42 +66,43 @@ public:
 	}
 
 	// Assignment Operator
-	SIMD_FORCE_INLINE btQuaternion& 
-	operator=(const btQuaternion& v) 
+	SIMD_FORCE_INLINE btQuaternion&
+	operator=(const btQuaternion& v)
 	{
 		mVec128 = v.mVec128;
-		
+
 		return *this;
 	}
-	
+
 #endif
 
 	//		template <typename btScalar>
 	//		explicit Quaternion(const btScalar *v) : Tuple4<btScalar>(v) {}
-  /**@brief Constructor from scalars */
-	btQuaternion(const btScalar& _x, const btScalar& _y, const btScalar& _z, const btScalar& _w) 
-		: btQuadWord(_x, _y, _z, _w) 
-	{}
-  /**@brief Axis angle Constructor
+	/**@brief Constructor from scalars */
+	btQuaternion(const btScalar& _x, const btScalar& _y, const btScalar& _z, const btScalar& _w)
+		: btQuadWord(_x, _y, _z, _w)
+	{
+	}
+	/**@brief Axis angle Constructor
    * @param axis The axis which the rotation is around
    * @param angle The magnitude of the rotation around the angle (Radians) */
-	btQuaternion(const btVector3& _axis, const btScalar& _angle) 
-	{ 
-		setRotation(_axis, _angle); 
+	btQuaternion(const btVector3& _axis, const btScalar& _angle)
+	{
+		setRotation(_axis, _angle);
 	}
-  /**@brief Constructor from Euler angles
+	/**@brief Constructor from Euler angles
    * @param yaw Angle around Y unless BT_EULER_DEFAULT_ZYX defined then Z
    * @param pitch Angle around X unless BT_EULER_DEFAULT_ZYX defined then Y
    * @param roll Angle around Z unless BT_EULER_DEFAULT_ZYX defined then X */
 	btQuaternion(const btScalar& yaw, const btScalar& pitch, const btScalar& roll)
-	{ 
+	{
 #ifndef BT_EULER_DEFAULT_ZYX
-		setEuler(yaw, pitch, roll); 
+		setEuler(yaw, pitch, roll);
 #else
-		setEulerZYX(yaw, pitch, roll); 
-#endif 
+		setEulerZYX(yaw, pitch, roll);
+#endif
 	}
-  /**@brief Set the rotation using axis angle notation 
+	/**@brief Set the rotation using axis angle notation 
    * @param axis The axis around which to rotate
    * @param angle The magnitude of the rotation in Radians */
 	void setRotation(const btVector3& axis, const btScalar& _angle)
@@ -114,18 +110,18 @@ public:
 		btScalar d = axis.length();
 		btAssert(d != btScalar(0.0));
 		btScalar s = btSin(_angle * btScalar(0.5)) / d;
-		setValue(axis.x() * s, axis.y() * s, axis.z() * s, 
-			btCos(_angle * btScalar(0.5)));
+		setValue(axis.x() * s, axis.y() * s, axis.z() * s,
+				 btCos(_angle * btScalar(0.5)));
 	}
-  /**@brief Set the quaternion using Euler angles
+	/**@brief Set the quaternion using Euler angles
    * @param yaw Angle around Y
    * @param pitch Angle around X
    * @param roll Angle around Z */
 	void setEuler(const btScalar& yaw, const btScalar& pitch, const btScalar& roll)
 	{
-		btScalar halfYaw = btScalar(yaw) * btScalar(0.5);  
-		btScalar halfPitch = btScalar(pitch) * btScalar(0.5);  
-		btScalar halfRoll = btScalar(roll) * btScalar(0.5);  
+		btScalar halfYaw = btScalar(yaw) * btScalar(0.5);
+		btScalar halfPitch = btScalar(pitch) * btScalar(0.5);
+		btScalar halfRoll = btScalar(roll) * btScalar(0.5);
 		btScalar cosYaw = btCos(halfYaw);
 		btScalar sinYaw = btSin(halfYaw);
 		btScalar cosPitch = btCos(halfPitch);
@@ -133,32 +129,32 @@ public:
 		btScalar cosRoll = btCos(halfRoll);
 		btScalar sinRoll = btSin(halfRoll);
 		setValue(cosRoll * sinPitch * cosYaw + sinRoll * cosPitch * sinYaw,
-			cosRoll * cosPitch * sinYaw - sinRoll * sinPitch * cosYaw,
-			sinRoll * cosPitch * cosYaw - cosRoll * sinPitch * sinYaw,
-			cosRoll * cosPitch * cosYaw + sinRoll * sinPitch * sinYaw);
+				 cosRoll * cosPitch * sinYaw - sinRoll * sinPitch * cosYaw,
+				 sinRoll * cosPitch * cosYaw - cosRoll * sinPitch * sinYaw,
+				 cosRoll * cosPitch * cosYaw + sinRoll * sinPitch * sinYaw);
 	}
-  /**@brief Set the quaternion using euler angles 
+	/**@brief Set the quaternion using euler angles 
    * @param yaw Angle around Z
    * @param pitch Angle around Y
    * @param roll Angle around X */
 	void setEulerZYX(const btScalar& yawZ, const btScalar& pitchY, const btScalar& rollX)
 	{
-		btScalar halfYaw = btScalar(yawZ) * btScalar(0.5);  
-		btScalar halfPitch = btScalar(pitchY) * btScalar(0.5);  
-		btScalar halfRoll = btScalar(rollX) * btScalar(0.5);  
+		btScalar halfYaw = btScalar(yawZ) * btScalar(0.5);
+		btScalar halfPitch = btScalar(pitchY) * btScalar(0.5);
+		btScalar halfRoll = btScalar(rollX) * btScalar(0.5);
 		btScalar cosYaw = btCos(halfYaw);
 		btScalar sinYaw = btSin(halfYaw);
 		btScalar cosPitch = btCos(halfPitch);
 		btScalar sinPitch = btSin(halfPitch);
 		btScalar cosRoll = btCos(halfRoll);
 		btScalar sinRoll = btSin(halfRoll);
-		setValue(sinRoll * cosPitch * cosYaw - cosRoll * sinPitch * sinYaw, //x
-                         cosRoll * sinPitch * cosYaw + sinRoll * cosPitch * sinYaw, //y
-                         cosRoll * cosPitch * sinYaw - sinRoll * sinPitch * cosYaw, //z
-                         cosRoll * cosPitch * cosYaw + sinRoll * sinPitch * sinYaw); //formerly yzx
+		setValue(sinRoll * cosPitch * cosYaw - cosRoll * sinPitch * sinYaw,   //x
+				 cosRoll * sinPitch * cosYaw + sinRoll * cosPitch * sinYaw,   //y
+				 cosRoll * cosPitch * sinYaw - sinRoll * sinPitch * cosYaw,   //z
+				 cosRoll * cosPitch * cosYaw + sinRoll * sinPitch * sinYaw);  //formerly yzx
 	}
 
-	  /**@brief Get the euler angles from this quaternion
+	/**@brief Get the euler angles from this quaternion
 	   * @param yaw Angle around Z
 	   * @param pitch Angle around Y
 	   * @param roll Angle around X */
@@ -173,23 +169,25 @@ public:
 		sqy = m_floats[1] * m_floats[1];
 		sqz = m_floats[2] * m_floats[2];
 		squ = m_floats[3] * m_floats[3];
-	    sarg = btScalar(-2.) * (m_floats[0] * m_floats[2] - m_floats[3] * m_floats[1]);
-		
+		sarg = btScalar(-2.) * (m_floats[0] * m_floats[2] - m_floats[3] * m_floats[1]);
+
 		// If the pitch angle is PI/2 or -PI/2, we can only compute
 		// the sum roll + yaw.  However, any combination that gives
 		// the right sum will produce the correct orientation, so we
 		// set rollX = 0 and compute yawZ.
 		if (sarg <= -btScalar(0.99999))
 		{
-			pitchY = btScalar(-0.5)*SIMD_PI;
-			rollX  = 0;
-			yawZ   = btScalar(2) * btAtan2(m_floats[0],-m_floats[1]);
-		} else if (sarg >= btScalar(0.99999))
+			pitchY = btScalar(-0.5) * SIMD_PI;
+			rollX = 0;
+			yawZ = btScalar(2) * btAtan2(m_floats[0], -m_floats[1]);
+		}
+		else if (sarg >= btScalar(0.99999))
 		{
-			pitchY = btScalar(0.5)*SIMD_PI;
-			rollX  = 0;
-			yawZ   = btScalar(2) * btAtan2(-m_floats[0], m_floats[1]);
-		} else
+			pitchY = btScalar(0.5) * SIMD_PI;
+			rollX = 0;
+			yawZ = btScalar(2) * btAtan2(-m_floats[0], m_floats[1]);
+		}
+		else
 		{
 			pitchY = btAsin(sarg);
 			rollX = btAtan2(2 * (m_floats[1] * m_floats[2] + m_floats[3] * m_floats[0]), squ - sqx - sqy + sqz);
@@ -197,178 +195,178 @@ public:
 		}
 	}
 
-  /**@brief Add two quaternions
+	/**@brief Add two quaternions
    * @param q The quaternion to add to this one */
-	SIMD_FORCE_INLINE	btQuaternion& operator+=(const btQuaternion& q)
+	SIMD_FORCE_INLINE btQuaternion& operator+=(const btQuaternion& q)
 	{
-#if defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 		mVec128 = _mm_add_ps(mVec128, q.mVec128);
 #elif defined(BT_USE_NEON)
 		mVec128 = vaddq_f32(mVec128, q.mVec128);
-#else	
-		m_floats[0] += q.x(); 
-        m_floats[1] += q.y(); 
-        m_floats[2] += q.z(); 
-        m_floats[3] += q.m_floats[3];
+#else
+		m_floats[0] += q.x();
+		m_floats[1] += q.y();
+		m_floats[2] += q.z();
+		m_floats[3] += q.m_floats[3];
 #endif
 		return *this;
 	}
 
-  /**@brief Subtract out a quaternion
+	/**@brief Subtract out a quaternion
    * @param q The quaternion to subtract from this one */
-	btQuaternion& operator-=(const btQuaternion& q) 
+	btQuaternion& operator-=(const btQuaternion& q)
 	{
-#if defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 		mVec128 = _mm_sub_ps(mVec128, q.mVec128);
 #elif defined(BT_USE_NEON)
 		mVec128 = vsubq_f32(mVec128, q.mVec128);
-#else	
-		m_floats[0] -= q.x(); 
-        m_floats[1] -= q.y(); 
-        m_floats[2] -= q.z(); 
-        m_floats[3] -= q.m_floats[3];
+#else
+		m_floats[0] -= q.x();
+		m_floats[1] -= q.y();
+		m_floats[2] -= q.z();
+		m_floats[3] -= q.m_floats[3];
 #endif
-        return *this;
+		return *this;
 	}
 
-  /**@brief Scale this quaternion
+	/**@brief Scale this quaternion
    * @param s The scalar to scale by */
 	btQuaternion& operator*=(const btScalar& s)
 	{
-#if defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
-		__m128	vs = _mm_load_ss(&s);	//	(S 0 0 0)
-		vs = bt_pshufd_ps(vs, 0);	//	(S S S S)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
+		__m128 vs = _mm_load_ss(&s);  //	(S 0 0 0)
+		vs = bt_pshufd_ps(vs, 0);     //	(S S S S)
 		mVec128 = _mm_mul_ps(mVec128, vs);
 #elif defined(BT_USE_NEON)
 		mVec128 = vmulq_n_f32(mVec128, s);
 #else
-		m_floats[0] *= s; 
-        m_floats[1] *= s; 
-        m_floats[2] *= s; 
-        m_floats[3] *= s;
+		m_floats[0] *= s;
+		m_floats[1] *= s;
+		m_floats[2] *= s;
+		m_floats[3] *= s;
 #endif
 		return *this;
 	}
 
-  /**@brief Multiply this quaternion by q on the right
+	/**@brief Multiply this quaternion by q on the right
    * @param q The other quaternion 
    * Equivilant to this = this * q */
 	btQuaternion& operator*=(const btQuaternion& q)
 	{
-#if defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 		__m128 vQ2 = q.get128();
-		
-		__m128 A1 = bt_pshufd_ps(mVec128, BT_SHUFFLE(0,1,2,0));
-		__m128 B1 = bt_pshufd_ps(vQ2, BT_SHUFFLE(3,3,3,0));
-		
+
+		__m128 A1 = bt_pshufd_ps(mVec128, BT_SHUFFLE(0, 1, 2, 0));
+		__m128 B1 = bt_pshufd_ps(vQ2, BT_SHUFFLE(3, 3, 3, 0));
+
 		A1 = A1 * B1;
-		
-		__m128 A2 = bt_pshufd_ps(mVec128, BT_SHUFFLE(1,2,0,1));
-		__m128 B2 = bt_pshufd_ps(vQ2, BT_SHUFFLE(2,0,1,1));
-		
+
+		__m128 A2 = bt_pshufd_ps(mVec128, BT_SHUFFLE(1, 2, 0, 1));
+		__m128 B2 = bt_pshufd_ps(vQ2, BT_SHUFFLE(2, 0, 1, 1));
+
 		A2 = A2 * B2;
-		
-		B1 = bt_pshufd_ps(mVec128, BT_SHUFFLE(2,0,1,2));
-		B2 = bt_pshufd_ps(vQ2, BT_SHUFFLE(1,2,0,2));
-		
-		B1 = B1 * B2;	//	A3 *= B3
-		
-		mVec128 = bt_splat_ps(mVec128, 3);	//	A0
-		mVec128 = mVec128 * vQ2;	//	A0 * B0
-		
-		A1 = A1 + A2;	//	AB12
-		mVec128 = mVec128 - B1;	//	AB03 = AB0 - AB3 
-		A1 = _mm_xor_ps(A1, vPPPM);	//	change sign of the last element
-		mVec128 = mVec128+ A1;	//	AB03 + AB12
 
-#elif defined(BT_USE_NEON)     
+		B1 = bt_pshufd_ps(mVec128, BT_SHUFFLE(2, 0, 1, 2));
+		B2 = bt_pshufd_ps(vQ2, BT_SHUFFLE(1, 2, 0, 2));
 
-        float32x4_t vQ1 = mVec128;
-        float32x4_t vQ2 = q.get128();
-        float32x4_t A0, A1, B1, A2, B2, A3, B3;
-        float32x2_t vQ1zx, vQ2wx, vQ1yz, vQ2zx, vQ2yz, vQ2xz;
-        
-        {
-        float32x2x2_t tmp;
-        tmp = vtrn_f32( vget_high_f32(vQ1), vget_low_f32(vQ1) );       // {z x}, {w y}
-        vQ1zx = tmp.val[0];
+		B1 = B1 * B2;  //	A3 *= B3
 
-        tmp = vtrn_f32( vget_high_f32(vQ2), vget_low_f32(vQ2) );       // {z x}, {w y}
-        vQ2zx = tmp.val[0];
-        }
-        vQ2wx = vext_f32(vget_high_f32(vQ2), vget_low_f32(vQ2), 1); 
+		mVec128 = bt_splat_ps(mVec128, 3);  //	A0
+		mVec128 = mVec128 * vQ2;            //	A0 * B0
 
-        vQ1yz = vext_f32(vget_low_f32(vQ1), vget_high_f32(vQ1), 1);
+		A1 = A1 + A2;                //	AB12
+		mVec128 = mVec128 - B1;      //	AB03 = AB0 - AB3
+		A1 = _mm_xor_ps(A1, vPPPM);  //	change sign of the last element
+		mVec128 = mVec128 + A1;      //	AB03 + AB12
 
-        vQ2yz = vext_f32(vget_low_f32(vQ2), vget_high_f32(vQ2), 1);
-        vQ2xz = vext_f32(vQ2zx, vQ2zx, 1);
+#elif defined(BT_USE_NEON)
 
-        A1 = vcombine_f32(vget_low_f32(vQ1), vQ1zx);                    // X Y  z x 
-        B1 = vcombine_f32(vdup_lane_f32(vget_high_f32(vQ2), 1), vQ2wx); // W W  W X 
+		float32x4_t vQ1 = mVec128;
+		float32x4_t vQ2 = q.get128();
+		float32x4_t A0, A1, B1, A2, B2, A3, B3;
+		float32x2_t vQ1zx, vQ2wx, vQ1yz, vQ2zx, vQ2yz, vQ2xz;
 
-        A2 = vcombine_f32(vQ1yz, vget_low_f32(vQ1));
-        B2 = vcombine_f32(vQ2zx, vdup_lane_f32(vget_low_f32(vQ2), 1));
+		{
+			float32x2x2_t tmp;
+			tmp = vtrn_f32(vget_high_f32(vQ1), vget_low_f32(vQ1));  // {z x}, {w y}
+			vQ1zx = tmp.val[0];
 
-        A3 = vcombine_f32(vQ1zx, vQ1yz);        // Z X Y Z
-        B3 = vcombine_f32(vQ2yz, vQ2xz);        // Y Z x z
+			tmp = vtrn_f32(vget_high_f32(vQ2), vget_low_f32(vQ2));  // {z x}, {w y}
+			vQ2zx = tmp.val[0];
+		}
+		vQ2wx = vext_f32(vget_high_f32(vQ2), vget_low_f32(vQ2), 1);
 
-        A1 = vmulq_f32(A1, B1);
-        A2 = vmulq_f32(A2, B2);
-        A3 = vmulq_f32(A3, B3);	//	A3 *= B3
-        A0 = vmulq_lane_f32(vQ2, vget_high_f32(vQ1), 1); //	A0 * B0
+		vQ1yz = vext_f32(vget_low_f32(vQ1), vget_high_f32(vQ1), 1);
 
-        A1 = vaddq_f32(A1, A2);	//	AB12 = AB1 + AB2
-        A0 = vsubq_f32(A0, A3);	//	AB03 = AB0 - AB3 
-        
-        //	change the sign of the last element
-        A1 = (btSimdFloat4)veorq_s32((int32x4_t)A1, (int32x4_t)vPPPM);	
-        A0 = vaddq_f32(A0, A1);	//	AB03 + AB12
-        
-        mVec128 = A0;
+		vQ2yz = vext_f32(vget_low_f32(vQ2), vget_high_f32(vQ2), 1);
+		vQ2xz = vext_f32(vQ2zx, vQ2zx, 1);
+
+		A1 = vcombine_f32(vget_low_f32(vQ1), vQ1zx);                     // X Y  z x
+		B1 = vcombine_f32(vdup_lane_f32(vget_high_f32(vQ2), 1), vQ2wx);  // W W  W X
+
+		A2 = vcombine_f32(vQ1yz, vget_low_f32(vQ1));
+		B2 = vcombine_f32(vQ2zx, vdup_lane_f32(vget_low_f32(vQ2), 1));
+
+		A3 = vcombine_f32(vQ1zx, vQ1yz);  // Z X Y Z
+		B3 = vcombine_f32(vQ2yz, vQ2xz);  // Y Z x z
+
+		A1 = vmulq_f32(A1, B1);
+		A2 = vmulq_f32(A2, B2);
+		A3 = vmulq_f32(A3, B3);                           //	A3 *= B3
+		A0 = vmulq_lane_f32(vQ2, vget_high_f32(vQ1), 1);  //	A0 * B0
+
+		A1 = vaddq_f32(A1, A2);  //	AB12 = AB1 + AB2
+		A0 = vsubq_f32(A0, A3);  //	AB03 = AB0 - AB3
+
+		//	change the sign of the last element
+		A1 = (btSimdFloat4)veorq_s32((int32x4_t)A1, (int32x4_t)vPPPM);
+		A0 = vaddq_f32(A0, A1);  //	AB03 + AB12
+
+		mVec128 = A0;
 #else
 		setValue(
-            m_floats[3] * q.x() + m_floats[0] * q.m_floats[3] + m_floats[1] * q.z() - m_floats[2] * q.y(),
+			m_floats[3] * q.x() + m_floats[0] * q.m_floats[3] + m_floats[1] * q.z() - m_floats[2] * q.y(),
 			m_floats[3] * q.y() + m_floats[1] * q.m_floats[3] + m_floats[2] * q.x() - m_floats[0] * q.z(),
 			m_floats[3] * q.z() + m_floats[2] * q.m_floats[3] + m_floats[0] * q.y() - m_floats[1] * q.x(),
 			m_floats[3] * q.m_floats[3] - m_floats[0] * q.x() - m_floats[1] * q.y() - m_floats[2] * q.z());
 #endif
 		return *this;
 	}
-  /**@brief Return the dot product between this quaternion and another
+	/**@brief Return the dot product between this quaternion and another
    * @param q The other quaternion */
 	btScalar dot(const btQuaternion& q) const
 	{
-#if defined BT_USE_SIMD_VECTOR3 && defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
-		__m128	vd;
-		
+#if defined BT_USE_SIMD_VECTOR3 && defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
+		__m128 vd;
+
 		vd = _mm_mul_ps(mVec128, q.mVec128);
-		
-        __m128 t = _mm_movehl_ps(vd, vd);
+
+		__m128 t = _mm_movehl_ps(vd, vd);
 		vd = _mm_add_ps(vd, t);
 		t = _mm_shuffle_ps(vd, vd, 0x55);
 		vd = _mm_add_ss(vd, t);
-		
-        return _mm_cvtss_f32(vd);
+
+		return _mm_cvtss_f32(vd);
 #elif defined(BT_USE_NEON)
 		float32x4_t vd = vmulq_f32(mVec128, q.mVec128);
-		float32x2_t x = vpadd_f32(vget_low_f32(vd), vget_high_f32(vd));  
+		float32x2_t x = vpadd_f32(vget_low_f32(vd), vget_high_f32(vd));
 		x = vpadd_f32(x, x);
 		return vget_lane_f32(x, 0);
-#else    
-		return  m_floats[0] * q.x() + 
-                m_floats[1] * q.y() + 
-                m_floats[2] * q.z() + 
-                m_floats[3] * q.m_floats[3];
+#else
+		return m_floats[0] * q.x() +
+			   m_floats[1] * q.y() +
+			   m_floats[2] * q.z() +
+			   m_floats[3] * q.m_floats[3];
 #endif
 	}
 
-  /**@brief Return the length squared of the quaternion */
+	/**@brief Return the length squared of the quaternion */
 	btScalar length2() const
 	{
 		return dot(*this);
 	}
 
-  /**@brief Return the length of the quaternion */
+	/**@brief Return the length of the quaternion */
 	btScalar length() const
 	{
 		return btSqrt(length2());
@@ -376,46 +374,46 @@ public:
 	btQuaternion& safeNormalize()
 	{
 		btScalar l2 = length2();
-		if (l2>SIMD_EPSILON)
+		if (l2 > SIMD_EPSILON)
 		{
 			normalize();
 		}
 		return *this;
 	}
-  /**@brief Normalize the quaternion 
+	/**@brief Normalize the quaternion 
    * Such that x^2 + y^2 + z^2 +w^2 = 1 */
-	btQuaternion& normalize() 
+	btQuaternion& normalize()
 	{
-#if defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
-		__m128	vd;
-		
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
+		__m128 vd;
+
 		vd = _mm_mul_ps(mVec128, mVec128);
-		
-        __m128 t = _mm_movehl_ps(vd, vd);
+
+		__m128 t = _mm_movehl_ps(vd, vd);
 		vd = _mm_add_ps(vd, t);
 		t = _mm_shuffle_ps(vd, vd, 0x55);
 		vd = _mm_add_ss(vd, t);
 
 		vd = _mm_sqrt_ss(vd);
 		vd = _mm_div_ss(vOnes, vd);
-        vd = bt_pshufd_ps(vd, 0); // splat
+		vd = bt_pshufd_ps(vd, 0);  // splat
 		mVec128 = _mm_mul_ps(mVec128, vd);
-    
+
 		return *this;
-#else    
+#else
 		return *this /= length();
 #endif
 	}
 
-  /**@brief Return a scaled version of this quaternion
+	/**@brief Return a scaled version of this quaternion
    * @param s The scale factor */
 	SIMD_FORCE_INLINE btQuaternion
 	operator*(const btScalar& s) const
 	{
-#if defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
-		__m128	vs = _mm_load_ss(&s);	//	(S 0 0 0)
-		vs = bt_pshufd_ps(vs, 0x00);	//	(S S S S)
-		
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
+		__m128 vs = _mm_load_ss(&s);  //	(S 0 0 0)
+		vs = bt_pshufd_ps(vs, 0x00);  //	(S S S S)
+
 		return btQuaternion(_mm_mul_ps(mVec128, vs));
 #elif defined(BT_USE_NEON)
 		return btQuaternion(vmulq_n_f32(mVec128, s));
@@ -424,7 +422,7 @@ public:
 #endif
 	}
 
-  /**@brief Return an inversely scaled versionof this quaternion
+	/**@brief Return an inversely scaled versionof this quaternion
    * @param s The inverse scale factor */
 	btQuaternion operator/(const btScalar& s) const
 	{
@@ -432,49 +430,49 @@ public:
 		return *this * (btScalar(1.0) / s);
 	}
 
-  /**@brief Inversely scale this quaternion
+	/**@brief Inversely scale this quaternion
    * @param s The scale factor */
-	btQuaternion& operator/=(const btScalar& s) 
+	btQuaternion& operator/=(const btScalar& s)
 	{
 		btAssert(s != btScalar(0.0));
 		return *this *= btScalar(1.0) / s;
 	}
 
-  /**@brief Return a normalized version of this quaternion */
-	btQuaternion normalized() const 
+	/**@brief Return a normalized version of this quaternion */
+	btQuaternion normalized() const
 	{
 		return *this / length();
-	} 
+	}
 	/**@brief Return the ***half*** angle between this quaternion and the other
    * @param q The other quaternion */
-	btScalar angle(const btQuaternion& q) const 
+	btScalar angle(const btQuaternion& q) const
 	{
 		btScalar s = btSqrt(length2() * q.length2());
 		btAssert(s != btScalar(0.0));
 		return btAcos(dot(q) / s);
 	}
-	
+
 	/**@brief Return the angle between this quaternion and the other along the shortest path
 	* @param q The other quaternion */
-	btScalar angleShortestPath(const btQuaternion& q) const 
+	btScalar angleShortestPath(const btQuaternion& q) const
 	{
 		btScalar s = btSqrt(length2() * q.length2());
 		btAssert(s != btScalar(0.0));
-		if (dot(q) < 0) // Take care of long angle case see http://en.wikipedia.org/wiki/Slerp
+		if (dot(q) < 0)  // Take care of long angle case see http://en.wikipedia.org/wiki/Slerp
 			return btAcos(dot(-q) / s) * btScalar(2.0);
-		else 
+		else
 			return btAcos(dot(q) / s) * btScalar(2.0);
 	}
 
 	/**@brief Return the angle [0, 2Pi] of rotation represented by this quaternion */
-	btScalar getAngle() const 
+	btScalar getAngle() const
 	{
 		btScalar s = btScalar(2.) * btAcos(m_floats[3]);
 		return s;
 	}
 
 	/**@brief Return the angle [0, Pi] of rotation represented by this quaternion along the shortest path */
-	btScalar getAngleShortestPath() const 
+	btScalar getAngleShortestPath() const
 	{
 		btScalar s;
 		if (m_floats[3] >= 0)
@@ -484,120 +482,117 @@ public:
 		return s;
 	}
 
-
 	/**@brief Return the axis of the rotation represented by this quaternion */
 	btVector3 getAxis() const
 	{
-		btScalar s_squared = 1.f-m_floats[3]*m_floats[3];
-		
-		if (s_squared < btScalar(10.) * SIMD_EPSILON) //Check for divide by zero
-			return btVector3(1.0, 0.0, 0.0);  // Arbitrary
-		btScalar s = 1.f/btSqrt(s_squared);
+		btScalar s_squared = 1.f - m_floats[3] * m_floats[3];
+
+		if (s_squared < btScalar(10.) * SIMD_EPSILON)  //Check for divide by zero
+			return btVector3(1.0, 0.0, 0.0);           // Arbitrary
+		btScalar s = 1.f / btSqrt(s_squared);
 		return btVector3(m_floats[0] * s, m_floats[1] * s, m_floats[2] * s);
 	}
 
 	/**@brief Return the inverse of this quaternion */
 	btQuaternion inverse() const
 	{
-#if defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 		return btQuaternion(_mm_xor_ps(mVec128, vQInv));
 #elif defined(BT_USE_NEON)
-        return btQuaternion((btSimdFloat4)veorq_s32((int32x4_t)mVec128, (int32x4_t)vQInv));
-#else	
+		return btQuaternion((btSimdFloat4)veorq_s32((int32x4_t)mVec128, (int32x4_t)vQInv));
+#else
 		return btQuaternion(-m_floats[0], -m_floats[1], -m_floats[2], m_floats[3]);
 #endif
 	}
 
-  /**@brief Return the sum of this quaternion and the other 
+	/**@brief Return the sum of this quaternion and the other 
    * @param q2 The other quaternion */
 	SIMD_FORCE_INLINE btQuaternion
 	operator+(const btQuaternion& q2) const
 	{
-#if defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 		return btQuaternion(_mm_add_ps(mVec128, q2.mVec128));
 #elif defined(BT_USE_NEON)
-        return btQuaternion(vaddq_f32(mVec128, q2.mVec128));
-#else	
+		return btQuaternion(vaddq_f32(mVec128, q2.mVec128));
+#else
 		const btQuaternion& q1 = *this;
 		return btQuaternion(q1.x() + q2.x(), q1.y() + q2.y(), q1.z() + q2.z(), q1.m_floats[3] + q2.m_floats[3]);
 #endif
 	}
 
-  /**@brief Return the difference between this quaternion and the other 
+	/**@brief Return the difference between this quaternion and the other 
    * @param q2 The other quaternion */
 	SIMD_FORCE_INLINE btQuaternion
 	operator-(const btQuaternion& q2) const
 	{
-#if defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 		return btQuaternion(_mm_sub_ps(mVec128, q2.mVec128));
 #elif defined(BT_USE_NEON)
-        return btQuaternion(vsubq_f32(mVec128, q2.mVec128));
-#else	
+		return btQuaternion(vsubq_f32(mVec128, q2.mVec128));
+#else
 		const btQuaternion& q1 = *this;
 		return btQuaternion(q1.x() - q2.x(), q1.y() - q2.y(), q1.z() - q2.z(), q1.m_floats[3] - q2.m_floats[3]);
 #endif
 	}
 
-  /**@brief Return the negative of this quaternion 
+	/**@brief Return the negative of this quaternion 
    * This simply negates each element */
 	SIMD_FORCE_INLINE btQuaternion operator-() const
 	{
-#if defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 		return btQuaternion(_mm_xor_ps(mVec128, btvMzeroMask));
 #elif defined(BT_USE_NEON)
-		return btQuaternion((btSimdFloat4)veorq_s32((int32x4_t)mVec128, (int32x4_t)btvMzeroMask) );
-#else	
+		return btQuaternion((btSimdFloat4)veorq_s32((int32x4_t)mVec128, (int32x4_t)btvMzeroMask));
+#else
 		const btQuaternion& q2 = *this;
-		return btQuaternion( - q2.x(), - q2.y(),  - q2.z(),  - q2.m_floats[3]);
+		return btQuaternion(-q2.x(), -q2.y(), -q2.z(), -q2.m_floats[3]);
 #endif
 	}
-  /**@todo document this and it's use */
-	SIMD_FORCE_INLINE btQuaternion farthest( const btQuaternion& qd) const 
+	/**@todo document this and it's use */
+	SIMD_FORCE_INLINE btQuaternion farthest(const btQuaternion& qd) const
 	{
-		btQuaternion diff,sum;
+		btQuaternion diff, sum;
 		diff = *this - qd;
 		sum = *this + qd;
-		if( diff.dot(diff) > sum.dot(sum) )
+		if (diff.dot(diff) > sum.dot(sum))
 			return qd;
 		return (-qd);
 	}
 
 	/**@todo document this and it's use */
-	SIMD_FORCE_INLINE btQuaternion nearest( const btQuaternion& qd) const 
+	SIMD_FORCE_INLINE btQuaternion nearest(const btQuaternion& qd) const
 	{
-		btQuaternion diff,sum;
+		btQuaternion diff, sum;
 		diff = *this - qd;
 		sum = *this + qd;
-		if( diff.dot(diff) < sum.dot(sum) )
+		if (diff.dot(diff) < sum.dot(sum))
 			return qd;
 		return (-qd);
 	}
 
-
-  /**@brief Return the quaternion which is the result of Spherical Linear Interpolation between this and the other quaternion
+	/**@brief Return the quaternion which is the result of Spherical Linear Interpolation between this and the other quaternion
    * @param q The other quaternion to interpolate with 
    * @param t The ratio between this and q to interpolate.  If t = 0 the result is this, if t=1 the result is q.
    * Slerp interpolates assuming constant velocity.  */
 	btQuaternion slerp(const btQuaternion& q, const btScalar& t) const
 	{
-
 		const btScalar magnitude = btSqrt(length2() * q.length2());
 		btAssert(magnitude > btScalar(0));
-		
+
 		const btScalar product = dot(q) / magnitude;
 		const btScalar absproduct = btFabs(product);
-		
-		if(absproduct < btScalar(1.0 - SIMD_EPSILON))
+
+		if (absproduct < btScalar(1.0 - SIMD_EPSILON))
 		{
 			// Take care of long angle case see http://en.wikipedia.org/wiki/Slerp
 			const btScalar theta = btAcos(absproduct);
 			const btScalar d = btSin(theta);
 			btAssert(d > btScalar(0));
-			
+
 			const btScalar sign = (product < 0) ? btScalar(-1) : btScalar(1);
 			const btScalar s0 = btSin((btScalar(1.0) - t) * theta) / d;
 			const btScalar s1 = btSin(sign * t * theta) / d;
-			
+
 			return btQuaternion(
 				(m_floats[0] * s0 + q.x() * s1),
 				(m_floats[1] * s0 + q.y() * s1),
@@ -610,314 +605,308 @@ public:
 		}
 	}
 
-	static const btQuaternion&	getIdentity()
+	static const btQuaternion& getIdentity()
 	{
-		static const btQuaternion identityQuat(btScalar(0.),btScalar(0.),btScalar(0.),btScalar(1.));
+		static const btQuaternion identityQuat(btScalar(0.), btScalar(0.), btScalar(0.), btScalar(1.));
 		return identityQuat;
 	}
 
 	SIMD_FORCE_INLINE const btScalar& getW() const { return m_floats[3]; }
 
-	SIMD_FORCE_INLINE	void	serialize(struct	btQuaternionData& dataOut) const;
+	SIMD_FORCE_INLINE void serialize(struct btQuaternionData& dataOut) const;
 
-	SIMD_FORCE_INLINE	void	deSerialize(const struct	btQuaternionFloatData& dataIn);
+	SIMD_FORCE_INLINE void deSerialize(const struct btQuaternionFloatData& dataIn);
 
-	SIMD_FORCE_INLINE	void	deSerialize(const struct	btQuaternionDoubleData& dataIn);
+	SIMD_FORCE_INLINE void deSerialize(const struct btQuaternionDoubleData& dataIn);
 
-	SIMD_FORCE_INLINE	void	serializeFloat(struct	btQuaternionFloatData& dataOut) const;
+	SIMD_FORCE_INLINE void serializeFloat(struct btQuaternionFloatData& dataOut) const;
 
-	SIMD_FORCE_INLINE	void	deSerializeFloat(const struct	btQuaternionFloatData& dataIn);
+	SIMD_FORCE_INLINE void deSerializeFloat(const struct btQuaternionFloatData& dataIn);
 
-	SIMD_FORCE_INLINE	void	serializeDouble(struct	btQuaternionDoubleData& dataOut) const;
-
-	SIMD_FORCE_INLINE	void	deSerializeDouble(const struct	btQuaternionDoubleData& dataIn);
+	SIMD_FORCE_INLINE void serializeDouble(struct btQuaternionDoubleData& dataOut) const;
 
+	SIMD_FORCE_INLINE void deSerializeDouble(const struct btQuaternionDoubleData& dataIn);
 };
 
-
-
-
-
 /**@brief Return the product of two quaternions */
 SIMD_FORCE_INLINE btQuaternion
-operator*(const btQuaternion& q1, const btQuaternion& q2) 
+operator*(const btQuaternion& q1, const btQuaternion& q2)
 {
-#if defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 	__m128 vQ1 = q1.get128();
 	__m128 vQ2 = q2.get128();
 	__m128 A0, A1, B1, A2, B2;
-    
-	A1 = bt_pshufd_ps(vQ1, BT_SHUFFLE(0,1,2,0)); // X Y  z x     //      vtrn
-	B1 = bt_pshufd_ps(vQ2, BT_SHUFFLE(3,3,3,0)); // W W  W X     // vdup vext
+
+	A1 = bt_pshufd_ps(vQ1, BT_SHUFFLE(0, 1, 2, 0));  // X Y  z x     //      vtrn
+	B1 = bt_pshufd_ps(vQ2, BT_SHUFFLE(3, 3, 3, 0));  // W W  W X     // vdup vext
 
 	A1 = A1 * B1;
-	
-	A2 = bt_pshufd_ps(vQ1, BT_SHUFFLE(1,2,0,1)); // Y Z  X Y     // vext 
-	B2 = bt_pshufd_ps(vQ2, BT_SHUFFLE(2,0,1,1)); // z x  Y Y     // vtrn vdup
+
+	A2 = bt_pshufd_ps(vQ1, BT_SHUFFLE(1, 2, 0, 1));  // Y Z  X Y     // vext
+	B2 = bt_pshufd_ps(vQ2, BT_SHUFFLE(2, 0, 1, 1));  // z x  Y Y     // vtrn vdup
 
 	A2 = A2 * B2;
 
-	B1 = bt_pshufd_ps(vQ1, BT_SHUFFLE(2,0,1,2)); // z x Y Z      // vtrn vext
-	B2 = bt_pshufd_ps(vQ2, BT_SHUFFLE(1,2,0,2)); // Y Z x z      // vext vtrn
-	
-	B1 = B1 * B2;	//	A3 *= B3
+	B1 = bt_pshufd_ps(vQ1, BT_SHUFFLE(2, 0, 1, 2));  // z x Y Z      // vtrn vext
+	B2 = bt_pshufd_ps(vQ2, BT_SHUFFLE(1, 2, 0, 2));  // Y Z x z      // vext vtrn
 
-	A0 = bt_splat_ps(vQ1, 3);	//	A0
-	A0 = A0 * vQ2;	//	A0 * B0
+	B1 = B1 * B2;  //	A3 *= B3
+
+	A0 = bt_splat_ps(vQ1, 3);  //	A0
+	A0 = A0 * vQ2;             //	A0 * B0
+
+	A1 = A1 + A2;  //	AB12
+	A0 = A0 - B1;  //	AB03 = AB0 - AB3
+
+	A1 = _mm_xor_ps(A1, vPPPM);  //	change sign of the last element
+	A0 = A0 + A1;                //	AB03 + AB12
 
-	A1 = A1 + A2;	//	AB12
-	A0 =  A0 - B1;	//	AB03 = AB0 - AB3 
-	
-    A1 = _mm_xor_ps(A1, vPPPM);	//	change sign of the last element
-	A0 = A0 + A1;	//	AB03 + AB12
-	
 	return btQuaternion(A0);
 
-#elif defined(BT_USE_NEON)     
+#elif defined(BT_USE_NEON)
 
 	float32x4_t vQ1 = q1.get128();
 	float32x4_t vQ2 = q2.get128();
 	float32x4_t A0, A1, B1, A2, B2, A3, B3;
-    float32x2_t vQ1zx, vQ2wx, vQ1yz, vQ2zx, vQ2yz, vQ2xz;
-    
-    {
-    float32x2x2_t tmp;
-    tmp = vtrn_f32( vget_high_f32(vQ1), vget_low_f32(vQ1) );       // {z x}, {w y}
-    vQ1zx = tmp.val[0];
+	float32x2_t vQ1zx, vQ2wx, vQ1yz, vQ2zx, vQ2yz, vQ2xz;
 
-    tmp = vtrn_f32( vget_high_f32(vQ2), vget_low_f32(vQ2) );       // {z x}, {w y}
-    vQ2zx = tmp.val[0];
-    }
-    vQ2wx = vext_f32(vget_high_f32(vQ2), vget_low_f32(vQ2), 1); 
+	{
+		float32x2x2_t tmp;
+		tmp = vtrn_f32(vget_high_f32(vQ1), vget_low_f32(vQ1));  // {z x}, {w y}
+		vQ1zx = tmp.val[0];
 
-    vQ1yz = vext_f32(vget_low_f32(vQ1), vget_high_f32(vQ1), 1);
+		tmp = vtrn_f32(vget_high_f32(vQ2), vget_low_f32(vQ2));  // {z x}, {w y}
+		vQ2zx = tmp.val[0];
+	}
+	vQ2wx = vext_f32(vget_high_f32(vQ2), vget_low_f32(vQ2), 1);
 
-    vQ2yz = vext_f32(vget_low_f32(vQ2), vget_high_f32(vQ2), 1);
-    vQ2xz = vext_f32(vQ2zx, vQ2zx, 1);
+	vQ1yz = vext_f32(vget_low_f32(vQ1), vget_high_f32(vQ1), 1);
 
-    A1 = vcombine_f32(vget_low_f32(vQ1), vQ1zx);                    // X Y  z x 
-    B1 = vcombine_f32(vdup_lane_f32(vget_high_f32(vQ2), 1), vQ2wx); // W W  W X 
+	vQ2yz = vext_f32(vget_low_f32(vQ2), vget_high_f32(vQ2), 1);
+	vQ2xz = vext_f32(vQ2zx, vQ2zx, 1);
+
+	A1 = vcombine_f32(vget_low_f32(vQ1), vQ1zx);                     // X Y  z x
+	B1 = vcombine_f32(vdup_lane_f32(vget_high_f32(vQ2), 1), vQ2wx);  // W W  W X
 
 	A2 = vcombine_f32(vQ1yz, vget_low_f32(vQ1));
-    B2 = vcombine_f32(vQ2zx, vdup_lane_f32(vget_low_f32(vQ2), 1));
+	B2 = vcombine_f32(vQ2zx, vdup_lane_f32(vget_low_f32(vQ2), 1));
 
-    A3 = vcombine_f32(vQ1zx, vQ1yz);        // Z X Y Z
-    B3 = vcombine_f32(vQ2yz, vQ2xz);        // Y Z x z
+	A3 = vcombine_f32(vQ1zx, vQ1yz);  // Z X Y Z
+	B3 = vcombine_f32(vQ2yz, vQ2xz);  // Y Z x z
 
 	A1 = vmulq_f32(A1, B1);
 	A2 = vmulq_f32(A2, B2);
-	A3 = vmulq_f32(A3, B3);	//	A3 *= B3
-	A0 = vmulq_lane_f32(vQ2, vget_high_f32(vQ1), 1); //	A0 * B0
+	A3 = vmulq_f32(A3, B3);                           //	A3 *= B3
+	A0 = vmulq_lane_f32(vQ2, vget_high_f32(vQ1), 1);  //	A0 * B0
+
+	A1 = vaddq_f32(A1, A2);  //	AB12 = AB1 + AB2
+	A0 = vsubq_f32(A0, A3);  //	AB03 = AB0 - AB3
+
+	//	change the sign of the last element
+	A1 = (btSimdFloat4)veorq_s32((int32x4_t)A1, (int32x4_t)vPPPM);
+	A0 = vaddq_f32(A0, A1);  //	AB03 + AB12
 
-	A1 = vaddq_f32(A1, A2);	//	AB12 = AB1 + AB2
-	A0 = vsubq_f32(A0, A3);	//	AB03 = AB0 - AB3 
-	
-    //	change the sign of the last element
-    A1 = (btSimdFloat4)veorq_s32((int32x4_t)A1, (int32x4_t)vPPPM);	
-	A0 = vaddq_f32(A0, A1);	//	AB03 + AB12
-	
 	return btQuaternion(A0);
 
 #else
 	return btQuaternion(
-        q1.w() * q2.x() + q1.x() * q2.w() + q1.y() * q2.z() - q1.z() * q2.y(),
+		q1.w() * q2.x() + q1.x() * q2.w() + q1.y() * q2.z() - q1.z() * q2.y(),
 		q1.w() * q2.y() + q1.y() * q2.w() + q1.z() * q2.x() - q1.x() * q2.z(),
 		q1.w() * q2.z() + q1.z() * q2.w() + q1.x() * q2.y() - q1.y() * q2.x(),
-		q1.w() * q2.w() - q1.x() * q2.x() - q1.y() * q2.y() - q1.z() * q2.z()); 
+		q1.w() * q2.w() - q1.x() * q2.x() - q1.y() * q2.y() - q1.z() * q2.z());
 #endif
 }
 
 SIMD_FORCE_INLINE btQuaternion
 operator*(const btQuaternion& q, const btVector3& w)
 {
-#if defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 	__m128 vQ1 = q.get128();
 	__m128 vQ2 = w.get128();
 	__m128 A1, B1, A2, B2, A3, B3;
-	
-	A1 = bt_pshufd_ps(vQ1, BT_SHUFFLE(3,3,3,0));
-	B1 = bt_pshufd_ps(vQ2, BT_SHUFFLE(0,1,2,0));
+
+	A1 = bt_pshufd_ps(vQ1, BT_SHUFFLE(3, 3, 3, 0));
+	B1 = bt_pshufd_ps(vQ2, BT_SHUFFLE(0, 1, 2, 0));
 
 	A1 = A1 * B1;
-	
-	A2 = bt_pshufd_ps(vQ1, BT_SHUFFLE(1,2,0,1));
-	B2 = bt_pshufd_ps(vQ2, BT_SHUFFLE(2,0,1,1));
+
+	A2 = bt_pshufd_ps(vQ1, BT_SHUFFLE(1, 2, 0, 1));
+	B2 = bt_pshufd_ps(vQ2, BT_SHUFFLE(2, 0, 1, 1));
 
 	A2 = A2 * B2;
 
-	A3 = bt_pshufd_ps(vQ1, BT_SHUFFLE(2,0,1,2));
-	B3 = bt_pshufd_ps(vQ2, BT_SHUFFLE(1,2,0,2));
-	
-	A3 = A3 * B3;	//	A3 *= B3
+	A3 = bt_pshufd_ps(vQ1, BT_SHUFFLE(2, 0, 1, 2));
+	B3 = bt_pshufd_ps(vQ2, BT_SHUFFLE(1, 2, 0, 2));
+
+	A3 = A3 * B3;  //	A3 *= B3
+
+	A1 = A1 + A2;                //	AB12
+	A1 = _mm_xor_ps(A1, vPPPM);  //	change sign of the last element
+	A1 = A1 - A3;                //	AB123 = AB12 - AB3
 
-	A1 = A1 + A2;	//	AB12
-	A1 = _mm_xor_ps(A1, vPPPM);	//	change sign of the last element
-    A1 = A1 - A3;	//	AB123 = AB12 - AB3 
-	
 	return btQuaternion(A1);
-    
-#elif defined(BT_USE_NEON)     
+
+#elif defined(BT_USE_NEON)
 
 	float32x4_t vQ1 = q.get128();
 	float32x4_t vQ2 = w.get128();
 	float32x4_t A1, B1, A2, B2, A3, B3;
-    float32x2_t vQ1wx, vQ2zx, vQ1yz, vQ2yz, vQ1zx, vQ2xz;
-    
-    vQ1wx = vext_f32(vget_high_f32(vQ1), vget_low_f32(vQ1), 1); 
-    {
-    float32x2x2_t tmp;
+	float32x2_t vQ1wx, vQ2zx, vQ1yz, vQ2yz, vQ1zx, vQ2xz;
 
-    tmp = vtrn_f32( vget_high_f32(vQ2), vget_low_f32(vQ2) );       // {z x}, {w y}
-    vQ2zx = tmp.val[0];
+	vQ1wx = vext_f32(vget_high_f32(vQ1), vget_low_f32(vQ1), 1);
+	{
+		float32x2x2_t tmp;
 
-    tmp = vtrn_f32( vget_high_f32(vQ1), vget_low_f32(vQ1) );       // {z x}, {w y}
-    vQ1zx = tmp.val[0];
-    }
+		tmp = vtrn_f32(vget_high_f32(vQ2), vget_low_f32(vQ2));  // {z x}, {w y}
+		vQ2zx = tmp.val[0];
 
-    vQ1yz = vext_f32(vget_low_f32(vQ1), vget_high_f32(vQ1), 1);
+		tmp = vtrn_f32(vget_high_f32(vQ1), vget_low_f32(vQ1));  // {z x}, {w y}
+		vQ1zx = tmp.val[0];
+	}
 
-    vQ2yz = vext_f32(vget_low_f32(vQ2), vget_high_f32(vQ2), 1);
-    vQ2xz = vext_f32(vQ2zx, vQ2zx, 1);
+	vQ1yz = vext_f32(vget_low_f32(vQ1), vget_high_f32(vQ1), 1);
 
-    A1 = vcombine_f32(vdup_lane_f32(vget_high_f32(vQ1), 1), vQ1wx); // W W  W X 
-    B1 = vcombine_f32(vget_low_f32(vQ2), vQ2zx);                    // X Y  z x 
+	vQ2yz = vext_f32(vget_low_f32(vQ2), vget_high_f32(vQ2), 1);
+	vQ2xz = vext_f32(vQ2zx, vQ2zx, 1);
+
+	A1 = vcombine_f32(vdup_lane_f32(vget_high_f32(vQ1), 1), vQ1wx);  // W W  W X
+	B1 = vcombine_f32(vget_low_f32(vQ2), vQ2zx);                     // X Y  z x
 
 	A2 = vcombine_f32(vQ1yz, vget_low_f32(vQ1));
-    B2 = vcombine_f32(vQ2zx, vdup_lane_f32(vget_low_f32(vQ2), 1));
+	B2 = vcombine_f32(vQ2zx, vdup_lane_f32(vget_low_f32(vQ2), 1));
 
-    A3 = vcombine_f32(vQ1zx, vQ1yz);        // Z X Y Z
-    B3 = vcombine_f32(vQ2yz, vQ2xz);        // Y Z x z
+	A3 = vcombine_f32(vQ1zx, vQ1yz);  // Z X Y Z
+	B3 = vcombine_f32(vQ2yz, vQ2xz);  // Y Z x z
 
 	A1 = vmulq_f32(A1, B1);
 	A2 = vmulq_f32(A2, B2);
-	A3 = vmulq_f32(A3, B3);	//	A3 *= B3
+	A3 = vmulq_f32(A3, B3);  //	A3 *= B3
+
+	A1 = vaddq_f32(A1, A2);  //	AB12 = AB1 + AB2
+
+	//	change the sign of the last element
+	A1 = (btSimdFloat4)veorq_s32((int32x4_t)A1, (int32x4_t)vPPPM);
+
+	A1 = vsubq_f32(A1, A3);  //	AB123 = AB12 - AB3
 
-	A1 = vaddq_f32(A1, A2);	//	AB12 = AB1 + AB2
-	
-    //	change the sign of the last element
-    A1 = (btSimdFloat4)veorq_s32((int32x4_t)A1, (int32x4_t)vPPPM);	
-	
-    A1 = vsubq_f32(A1, A3);	//	AB123 = AB12 - AB3
-	
 	return btQuaternion(A1);
-    
+
 #else
-	return btQuaternion( 
-         q.w() * w.x() + q.y() * w.z() - q.z() * w.y(),
-		 q.w() * w.y() + q.z() * w.x() - q.x() * w.z(),
-		 q.w() * w.z() + q.x() * w.y() - q.y() * w.x(),
-		-q.x() * w.x() - q.y() * w.y() - q.z() * w.z()); 
+	return btQuaternion(
+		q.w() * w.x() + q.y() * w.z() - q.z() * w.y(),
+		q.w() * w.y() + q.z() * w.x() - q.x() * w.z(),
+		q.w() * w.z() + q.x() * w.y() - q.y() * w.x(),
+		-q.x() * w.x() - q.y() * w.y() - q.z() * w.z());
 #endif
 }
 
 SIMD_FORCE_INLINE btQuaternion
 operator*(const btVector3& w, const btQuaternion& q)
 {
-#if defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 	__m128 vQ1 = w.get128();
 	__m128 vQ2 = q.get128();
 	__m128 A1, B1, A2, B2, A3, B3;
-	
-	A1 = bt_pshufd_ps(vQ1, BT_SHUFFLE(0,1,2,0));  // X Y  z x
-	B1 = bt_pshufd_ps(vQ2, BT_SHUFFLE(3,3,3,0));  // W W  W X 
+
+	A1 = bt_pshufd_ps(vQ1, BT_SHUFFLE(0, 1, 2, 0));  // X Y  z x
+	B1 = bt_pshufd_ps(vQ2, BT_SHUFFLE(3, 3, 3, 0));  // W W  W X
 
 	A1 = A1 * B1;
-	
-	A2 = bt_pshufd_ps(vQ1, BT_SHUFFLE(1,2,0,1));
-	B2 = bt_pshufd_ps(vQ2, BT_SHUFFLE(2,0,1,1));
 
-	A2 = A2 *B2;
+	A2 = bt_pshufd_ps(vQ1, BT_SHUFFLE(1, 2, 0, 1));
+	B2 = bt_pshufd_ps(vQ2, BT_SHUFFLE(2, 0, 1, 1));
 
-	A3 = bt_pshufd_ps(vQ1, BT_SHUFFLE(2,0,1,2));
-	B3 = bt_pshufd_ps(vQ2, BT_SHUFFLE(1,2,0,2));
-	
-	A3 = A3 * B3;	//	A3 *= B3
+	A2 = A2 * B2;
+
+	A3 = bt_pshufd_ps(vQ1, BT_SHUFFLE(2, 0, 1, 2));
+	B3 = bt_pshufd_ps(vQ2, BT_SHUFFLE(1, 2, 0, 2));
+
+	A3 = A3 * B3;  //	A3 *= B3
+
+	A1 = A1 + A2;                //	AB12
+	A1 = _mm_xor_ps(A1, vPPPM);  //	change sign of the last element
+	A1 = A1 - A3;                //	AB123 = AB12 - AB3
 
-	A1 = A1 + A2;	//	AB12
-	A1 = _mm_xor_ps(A1, vPPPM);	//	change sign of the last element
-	A1 = A1 - A3;	//	AB123 = AB12 - AB3 
-	
 	return btQuaternion(A1);
 
-#elif defined(BT_USE_NEON)     
+#elif defined(BT_USE_NEON)
 
 	float32x4_t vQ1 = w.get128();
 	float32x4_t vQ2 = q.get128();
-	float32x4_t  A1, B1, A2, B2, A3, B3;
-    float32x2_t vQ1zx, vQ2wx, vQ1yz, vQ2zx, vQ2yz, vQ2xz;
-    
-    {
-    float32x2x2_t tmp;
-   
-    tmp = vtrn_f32( vget_high_f32(vQ1), vget_low_f32(vQ1) );       // {z x}, {w y}
-    vQ1zx = tmp.val[0];
+	float32x4_t A1, B1, A2, B2, A3, B3;
+	float32x2_t vQ1zx, vQ2wx, vQ1yz, vQ2zx, vQ2yz, vQ2xz;
 
-    tmp = vtrn_f32( vget_high_f32(vQ2), vget_low_f32(vQ2) );       // {z x}, {w y}
-    vQ2zx = tmp.val[0];
-    }
-    vQ2wx = vext_f32(vget_high_f32(vQ2), vget_low_f32(vQ2), 1); 
+	{
+		float32x2x2_t tmp;
 
-    vQ1yz = vext_f32(vget_low_f32(vQ1), vget_high_f32(vQ1), 1);
+		tmp = vtrn_f32(vget_high_f32(vQ1), vget_low_f32(vQ1));  // {z x}, {w y}
+		vQ1zx = tmp.val[0];
 
-    vQ2yz = vext_f32(vget_low_f32(vQ2), vget_high_f32(vQ2), 1);
-    vQ2xz = vext_f32(vQ2zx, vQ2zx, 1);
+		tmp = vtrn_f32(vget_high_f32(vQ2), vget_low_f32(vQ2));  // {z x}, {w y}
+		vQ2zx = tmp.val[0];
+	}
+	vQ2wx = vext_f32(vget_high_f32(vQ2), vget_low_f32(vQ2), 1);
 
-    A1 = vcombine_f32(vget_low_f32(vQ1), vQ1zx);                    // X Y  z x 
-    B1 = vcombine_f32(vdup_lane_f32(vget_high_f32(vQ2), 1), vQ2wx); // W W  W X 
+	vQ1yz = vext_f32(vget_low_f32(vQ1), vget_high_f32(vQ1), 1);
+
+	vQ2yz = vext_f32(vget_low_f32(vQ2), vget_high_f32(vQ2), 1);
+	vQ2xz = vext_f32(vQ2zx, vQ2zx, 1);
+
+	A1 = vcombine_f32(vget_low_f32(vQ1), vQ1zx);                     // X Y  z x
+	B1 = vcombine_f32(vdup_lane_f32(vget_high_f32(vQ2), 1), vQ2wx);  // W W  W X
 
 	A2 = vcombine_f32(vQ1yz, vget_low_f32(vQ1));
-    B2 = vcombine_f32(vQ2zx, vdup_lane_f32(vget_low_f32(vQ2), 1));
+	B2 = vcombine_f32(vQ2zx, vdup_lane_f32(vget_low_f32(vQ2), 1));
 
-    A3 = vcombine_f32(vQ1zx, vQ1yz);        // Z X Y Z
-    B3 = vcombine_f32(vQ2yz, vQ2xz);        // Y Z x z
+	A3 = vcombine_f32(vQ1zx, vQ1yz);  // Z X Y Z
+	B3 = vcombine_f32(vQ2yz, vQ2xz);  // Y Z x z
 
 	A1 = vmulq_f32(A1, B1);
 	A2 = vmulq_f32(A2, B2);
-	A3 = vmulq_f32(A3, B3);	//	A3 *= B3
+	A3 = vmulq_f32(A3, B3);  //	A3 *= B3
+
+	A1 = vaddq_f32(A1, A2);  //	AB12 = AB1 + AB2
+
+	//	change the sign of the last element
+	A1 = (btSimdFloat4)veorq_s32((int32x4_t)A1, (int32x4_t)vPPPM);
+
+	A1 = vsubq_f32(A1, A3);  //	AB123 = AB12 - AB3
 
-	A1 = vaddq_f32(A1, A2);	//	AB12 = AB1 + AB2
-	
-    //	change the sign of the last element
-    A1 = (btSimdFloat4)veorq_s32((int32x4_t)A1, (int32x4_t)vPPPM);	
-	
-    A1 = vsubq_f32(A1, A3);	//	AB123 = AB12 - AB3
-	
 	return btQuaternion(A1);
-    
+
 #else
-	return btQuaternion( 
-        +w.x() * q.w() + w.y() * q.z() - w.z() * q.y(),
+	return btQuaternion(
+		+w.x() * q.w() + w.y() * q.z() - w.z() * q.y(),
 		+w.y() * q.w() + w.z() * q.x() - w.x() * q.z(),
 		+w.z() * q.w() + w.x() * q.y() - w.y() * q.x(),
-		-w.x() * q.x() - w.y() * q.y() - w.z() * q.z()); 
+		-w.x() * q.x() - w.y() * q.y() - w.z() * q.z());
 #endif
 }
 
 /**@brief Calculate the dot product between two quaternions */
-SIMD_FORCE_INLINE btScalar 
-dot(const btQuaternion& q1, const btQuaternion& q2) 
-{ 
-	return q1.dot(q2); 
+SIMD_FORCE_INLINE btScalar
+dot(const btQuaternion& q1, const btQuaternion& q2)
+{
+	return q1.dot(q2);
 }
 
-
 /**@brief Return the length of a quaternion */
 SIMD_FORCE_INLINE btScalar
-length(const btQuaternion& q) 
-{ 
-	return q.length(); 
+length(const btQuaternion& q)
+{
+	return q.length();
 }
 
 /**@brief Return the angle between two quaternions*/
 SIMD_FORCE_INLINE btScalar
-btAngle(const btQuaternion& q1, const btQuaternion& q2) 
-{ 
-	return q1.angle(q2); 
+btAngle(const btQuaternion& q1, const btQuaternion& q2)
+{
+	return q1.angle(q2);
 }
 
 /**@brief Return the inverse of a quaternion*/
 SIMD_FORCE_INLINE btQuaternion
-inverse(const btQuaternion& q) 
+inverse(const btQuaternion& q)
 {
 	return q.inverse();
 }
@@ -928,115 +917,105 @@ inverse(const btQuaternion& q)
  * @param t The ration between q1 and q2.  t = 0 return q1, t=1 returns q2 
  * Slerp assumes constant velocity between positions. */
 SIMD_FORCE_INLINE btQuaternion
-slerp(const btQuaternion& q1, const btQuaternion& q2, const btScalar& t) 
+slerp(const btQuaternion& q1, const btQuaternion& q2, const btScalar& t)
 {
 	return q1.slerp(q2, t);
 }
 
-SIMD_FORCE_INLINE btVector3 
-quatRotate(const btQuaternion& rotation, const btVector3& v) 
+SIMD_FORCE_INLINE btVector3
+quatRotate(const btQuaternion& rotation, const btVector3& v)
 {
 	btQuaternion q = rotation * v;
 	q *= rotation.inverse();
-#if defined BT_USE_SIMD_VECTOR3 && defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined BT_USE_SIMD_VECTOR3 && defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 	return btVector3(_mm_and_ps(q.get128(), btvFFF0fMask));
 #elif defined(BT_USE_NEON)
-    return btVector3((float32x4_t)vandq_s32((int32x4_t)q.get128(), btvFFF0Mask));
-#else	
-	return btVector3(q.getX(),q.getY(),q.getZ());
+	return btVector3((float32x4_t)vandq_s32((int32x4_t)q.get128(), btvFFF0Mask));
+#else
+	return btVector3(q.getX(), q.getY(), q.getZ());
 #endif
 }
 
-SIMD_FORCE_INLINE btQuaternion 
-shortestArcQuat(const btVector3& v0, const btVector3& v1) // Game Programming Gems 2.10. make sure v0,v1 are normalized
+SIMD_FORCE_INLINE btQuaternion
+shortestArcQuat(const btVector3& v0, const btVector3& v1)  // Game Programming Gems 2.10. make sure v0,v1 are normalized
 {
 	btVector3 c = v0.cross(v1);
-	btScalar  d = v0.dot(v1);
+	btScalar d = v0.dot(v1);
 
 	if (d < -1.0 + SIMD_EPSILON)
 	{
-		btVector3 n,unused;
-		btPlaneSpace1(v0,n,unused);
-		return btQuaternion(n.x(),n.y(),n.z(),0.0f); // just pick any vector that is orthogonal to v0
+		btVector3 n, unused;
+		btPlaneSpace1(v0, n, unused);
+		return btQuaternion(n.x(), n.y(), n.z(), 0.0f);  // just pick any vector that is orthogonal to v0
 	}
 
-	btScalar  s = btSqrt((1.0f + d) * 2.0f);
+	btScalar s = btSqrt((1.0f + d) * 2.0f);
 	btScalar rs = 1.0f / s;
 
-	return btQuaternion(c.getX()*rs,c.getY()*rs,c.getZ()*rs,s * 0.5f);
+	return btQuaternion(c.getX() * rs, c.getY() * rs, c.getZ() * rs, s * 0.5f);
 }
 
-SIMD_FORCE_INLINE btQuaternion 
-shortestArcQuatNormalize2(btVector3& v0,btVector3& v1)
+SIMD_FORCE_INLINE btQuaternion
+shortestArcQuatNormalize2(btVector3& v0, btVector3& v1)
 {
 	v0.normalize();
 	v1.normalize();
-	return shortestArcQuat(v0,v1);
+	return shortestArcQuat(v0, v1);
 }
 
-
-
-
-struct	btQuaternionFloatData
+struct btQuaternionFloatData
 {
-	float	m_floats[4];
+	float m_floats[4];
 };
 
-struct	btQuaternionDoubleData
+struct btQuaternionDoubleData
 {
-	double	m_floats[4];
-
+	double m_floats[4];
 };
 
-SIMD_FORCE_INLINE	void	btQuaternion::serializeFloat(struct	btQuaternionFloatData& dataOut) const
+SIMD_FORCE_INLINE void btQuaternion::serializeFloat(struct btQuaternionFloatData& dataOut) const
 {
 	///could also do a memcpy, check if it is worth it
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 		dataOut.m_floats[i] = float(m_floats[i]);
 }
 
-SIMD_FORCE_INLINE void	btQuaternion::deSerializeFloat(const struct	btQuaternionFloatData& dataIn)
+SIMD_FORCE_INLINE void btQuaternion::deSerializeFloat(const struct btQuaternionFloatData& dataIn)
 {
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 		m_floats[i] = btScalar(dataIn.m_floats[i]);
 }
 
-
-SIMD_FORCE_INLINE	void	btQuaternion::serializeDouble(struct	btQuaternionDoubleData& dataOut) const
+SIMD_FORCE_INLINE void btQuaternion::serializeDouble(struct btQuaternionDoubleData& dataOut) const
 {
 	///could also do a memcpy, check if it is worth it
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 		dataOut.m_floats[i] = double(m_floats[i]);
 }
 
-SIMD_FORCE_INLINE void	btQuaternion::deSerializeDouble(const struct	btQuaternionDoubleData& dataIn)
+SIMD_FORCE_INLINE void btQuaternion::deSerializeDouble(const struct btQuaternionDoubleData& dataIn)
 {
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 		m_floats[i] = btScalar(dataIn.m_floats[i]);
 }
 
-
-SIMD_FORCE_INLINE	void	btQuaternion::serialize(struct	btQuaternionData& dataOut) const
+SIMD_FORCE_INLINE void btQuaternion::serialize(struct btQuaternionData& dataOut) const
 {
 	///could also do a memcpy, check if it is worth it
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 		dataOut.m_floats[i] = m_floats[i];
 }
 
-SIMD_FORCE_INLINE void	btQuaternion::deSerialize(const struct	btQuaternionFloatData& dataIn)
+SIMD_FORCE_INLINE void btQuaternion::deSerialize(const struct btQuaternionFloatData& dataIn)
 {
-	for (int i = 0; i<4; i++)
+	for (int i = 0; i < 4; i++)
 		m_floats[i] = (btScalar)dataIn.m_floats[i];
 }
 
-SIMD_FORCE_INLINE void	btQuaternion::deSerialize(const struct	btQuaternionDoubleData& dataIn)
+SIMD_FORCE_INLINE void btQuaternion::deSerialize(const struct btQuaternionDoubleData& dataIn)
 {
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 		m_floats[i] = (btScalar)dataIn.m_floats[i];
 }
 
-
-#endif //BT_SIMD__QUATERNION_H_
-
-
-
+#endif  //BT_SIMD__QUATERNION_H_
diff --git a/thirdparty/bullet/LinearMath/btQuickprof.cpp b/thirdparty/bullet/LinearMath/btQuickprof.cpp
index 1572b962622..86fd1d78122 100644
--- a/thirdparty/bullet/LinearMath/btQuickprof.cpp
+++ b/thirdparty/bullet/LinearMath/btQuickprof.cpp
@@ -16,16 +16,13 @@
 #include "btQuickprof.h"
 #include "btThreads.h"
 
-
-
-
 #ifdef __CELLOS_LV2__
 #include <sys/sys_time.h>
 #include <sys/time_util.h>
 #include <stdio.h>
 #endif
 
-#if defined (SUNOS) || defined (__SUNOS__)
+#if defined(SUNOS) || defined(__SUNOS__)
 #include <stdio.h>
 #endif
 #ifdef __APPLE__
@@ -42,49 +39,46 @@
 #define NOIME
 
 #ifdef _XBOX
-	#include <Xtl.h>
-#else //_XBOX
-	#include <windows.h>
+#include <Xtl.h>
+#else  //_XBOX
+#include <windows.h>
 
-#if WINVER <0x0602
+#if WINVER < 0x0602
 #define GetTickCount64 GetTickCount
 #endif
 
-#endif //_XBOX
+#endif  //_XBOX
 
 #include <time.h>
 
-
-#else //_WIN32
+#else  //_WIN32
 #include <sys/time.h>
 
 #ifdef BT_LINUX_REALTIME
 //required linking against rt (librt)
 #include <time.h>
-#endif //BT_LINUX_REALTIME
+#endif  //BT_LINUX_REALTIME
 
-#endif //_WIN32
+#endif  //_WIN32
 
-#define mymin(a,b) (a > b ? a : b)
+#define mymin(a, b) (a > b ? a : b)
 
 struct btClockData
 {
-
 #ifdef BT_USE_WINDOWS_TIMERS
 	LARGE_INTEGER mClockFrequency;
 	LONGLONG mStartTick;
 	LARGE_INTEGER mStartTime;
 #else
 #ifdef __CELLOS_LV2__
-	uint64_t	mStartTime;
+	uint64_t mStartTime;
 #else
 #ifdef __APPLE__
-    uint64_t mStartTimeNano;
+	uint64_t mStartTimeNano;
 #endif
 	struct timeval mStartTime;
 #endif
-#endif //__CELLOS_LV2__
-
+#endif  //__CELLOS_LV2__
 };
 
 ///The btClock is a portable basic clock that measures accurate time in seconds, use for profiling.
@@ -114,8 +108,7 @@ btClock& btClock::operator=(const btClock& other)
 	return *this;
 }
 
-
-	/// Resets the initial reference time.
+/// Resets the initial reference time.
 void btClock::reset()
 {
 #ifdef BT_USE_WINDOWS_TIMERS
@@ -124,14 +117,14 @@ void btClock::reset()
 #else
 #ifdef __CELLOS_LV2__
 
-	typedef uint64_t  ClockSize;
+	typedef uint64_t ClockSize;
 	ClockSize newTime;
 	//__asm __volatile__( "mftb %0" : "=r" (newTime) : : "memory");
-	SYS_TIMEBASE_GET( newTime );
+	SYS_TIMEBASE_GET(newTime);
 	m_data->mStartTime = newTime;
 #else
 #ifdef __APPLE__
-    m_data->mStartTimeNano = mach_absolute_time();
+	m_data->mStartTimeNano = mach_absolute_time();
 #endif
 	gettimeofday(&m_data->mStartTime, 0);
 #endif
@@ -146,66 +139,66 @@ unsigned long long int btClock::getTimeMilliseconds()
 	LARGE_INTEGER currentTime;
 	QueryPerformanceCounter(&currentTime);
 	LONGLONG elapsedTime = currentTime.QuadPart -
-		m_data->mStartTime.QuadPart;
-		// Compute the number of millisecond ticks elapsed.
+						   m_data->mStartTime.QuadPart;
+	// Compute the number of millisecond ticks elapsed.
 	unsigned long msecTicks = (unsigned long)(1000 * elapsedTime /
-		m_data->mClockFrequency.QuadPart);
+											  m_data->mClockFrequency.QuadPart);
 
-		return msecTicks;
+	return msecTicks;
 #else
 
 #ifdef __CELLOS_LV2__
-		uint64_t freq=sys_time_get_timebase_frequency();
-		double dFreq=((double) freq) / 1000.0;
-		typedef uint64_t  ClockSize;
-		ClockSize newTime;
-		SYS_TIMEBASE_GET( newTime );
-		//__asm __volatile__( "mftb %0" : "=r" (newTime) : : "memory");
+	uint64_t freq = sys_time_get_timebase_frequency();
+	double dFreq = ((double)freq) / 1000.0;
+	typedef uint64_t ClockSize;
+	ClockSize newTime;
+	SYS_TIMEBASE_GET(newTime);
+	//__asm __volatile__( "mftb %0" : "=r" (newTime) : : "memory");
 
-		return (unsigned long int)((double(newTime-m_data->mStartTime)) / dFreq);
+	return (unsigned long int)((double(newTime - m_data->mStartTime)) / dFreq);
 #else
 
-		struct timeval currentTime;
-		gettimeofday(&currentTime, 0);
-		return (currentTime.tv_sec - m_data->mStartTime.tv_sec) * 1000 +
-			(currentTime.tv_usec - m_data->mStartTime.tv_usec) / 1000;
-#endif //__CELLOS_LV2__
+	struct timeval currentTime;
+	gettimeofday(&currentTime, 0);
+	return (currentTime.tv_sec - m_data->mStartTime.tv_sec) * 1000 +
+		   (currentTime.tv_usec - m_data->mStartTime.tv_usec) / 1000;
+#endif  //__CELLOS_LV2__
 #endif
 }
 
-	/// Returns the time in us since the last call to reset or since
-	/// the Clock was created.
+/// Returns the time in us since the last call to reset or since
+/// the Clock was created.
 unsigned long long int btClock::getTimeMicroseconds()
 {
 #ifdef BT_USE_WINDOWS_TIMERS
-	//see https://msdn.microsoft.com/en-us/library/windows/desktop/dn553408(v=vs.85).aspx	
-		LARGE_INTEGER currentTime, elapsedTime;
-		
-		QueryPerformanceCounter(&currentTime);
-		elapsedTime.QuadPart = currentTime.QuadPart - 
-			m_data->mStartTime.QuadPart;
-		elapsedTime.QuadPart *= 1000000;
-		elapsedTime.QuadPart /= m_data->mClockFrequency.QuadPart;
+	//see https://msdn.microsoft.com/en-us/library/windows/desktop/dn553408(v=vs.85).aspx
+	LARGE_INTEGER currentTime, elapsedTime;
 
-		return (unsigned long long) elapsedTime.QuadPart;
+	QueryPerformanceCounter(&currentTime);
+	elapsedTime.QuadPart = currentTime.QuadPart -
+						   m_data->mStartTime.QuadPart;
+	elapsedTime.QuadPart *= 1000000;
+	elapsedTime.QuadPart /= m_data->mClockFrequency.QuadPart;
+
+	return (unsigned long long)elapsedTime.QuadPart;
 #else
 
 #ifdef __CELLOS_LV2__
-		uint64_t freq=sys_time_get_timebase_frequency();
-		double dFreq=((double) freq)/ 1000000.0;
-		typedef uint64_t  ClockSize;
-		ClockSize newTime;
-		//__asm __volatile__( "mftb %0" : "=r" (newTime) : : "memory");
-		SYS_TIMEBASE_GET( newTime );
+	uint64_t freq = sys_time_get_timebase_frequency();
+	double dFreq = ((double)freq) / 1000000.0;
+	typedef uint64_t ClockSize;
+	ClockSize newTime;
+	//__asm __volatile__( "mftb %0" : "=r" (newTime) : : "memory");
+	SYS_TIMEBASE_GET(newTime);
 
-		return (unsigned long int)((double(newTime-m_data->mStartTime)) / dFreq);
+	return (unsigned long int)((double(newTime - m_data->mStartTime)) / dFreq);
 #else
 
-		struct timeval currentTime;
-		gettimeofday(&currentTime, 0);
-		return (currentTime.tv_sec - m_data->mStartTime.tv_sec) * 1000000 +
-			(currentTime.tv_usec - m_data->mStartTime.tv_usec);
-#endif//__CELLOS_LV2__
+	struct timeval currentTime;
+	gettimeofday(&currentTime, 0);
+	return (currentTime.tv_sec - m_data->mStartTime.tv_sec) * 1000000 +
+		   (currentTime.tv_usec - m_data->mStartTime.tv_usec);
+#endif  //__CELLOS_LV2__
 #endif
 }
 
@@ -213,65 +206,63 @@ unsigned long long int btClock::getTimeNanoseconds()
 {
 #ifdef BT_USE_WINDOWS_TIMERS
 	//see https://msdn.microsoft.com/en-us/library/windows/desktop/dn553408(v=vs.85).aspx
-		LARGE_INTEGER currentTime, elapsedTime;
-		
-		QueryPerformanceCounter(&currentTime);
-		elapsedTime.QuadPart = currentTime.QuadPart - 
-			m_data->mStartTime.QuadPart;
-		elapsedTime.QuadPart *= 1000000000;
-		elapsedTime.QuadPart /= m_data->mClockFrequency.QuadPart;
+	LARGE_INTEGER currentTime, elapsedTime;
 
-		return (unsigned long long) elapsedTime.QuadPart;
+	QueryPerformanceCounter(&currentTime);
+	elapsedTime.QuadPart = currentTime.QuadPart -
+						   m_data->mStartTime.QuadPart;
+	elapsedTime.QuadPart *= 1000000000;
+	elapsedTime.QuadPart /= m_data->mClockFrequency.QuadPart;
+
+	return (unsigned long long)elapsedTime.QuadPart;
 #else
 
 #ifdef __CELLOS_LV2__
-		uint64_t freq=sys_time_get_timebase_frequency();
-		double dFreq=((double) freq)/ 1e9;
-		typedef uint64_t  ClockSize;
-		ClockSize newTime;
-		//__asm __volatile__( "mftb %0" : "=r" (newTime) : : "memory");
-		SYS_TIMEBASE_GET( newTime );
+	uint64_t freq = sys_time_get_timebase_frequency();
+	double dFreq = ((double)freq) / 1e9;
+	typedef uint64_t ClockSize;
+	ClockSize newTime;
+	//__asm __volatile__( "mftb %0" : "=r" (newTime) : : "memory");
+	SYS_TIMEBASE_GET(newTime);
 
-		return (unsigned long int)((double(newTime-m_data->mStartTime)) / dFreq);
+	return (unsigned long int)((double(newTime - m_data->mStartTime)) / dFreq);
 #else
 #ifdef __APPLE__
-    uint64_t ticks = mach_absolute_time() - m_data->mStartTimeNano;
-    static long double conversion = 0.0L;
-    if( 0.0L == conversion )
-    {
-        // attempt to get conversion to nanoseconds
-        mach_timebase_info_data_t info;
-        int err = mach_timebase_info( &info );
-        if( err )
-        {
-            btAssert(0);
-            conversion = 1.;
-        }
-        conversion = info.numer / info.denom;
-    }
-    return (ticks * conversion);
+	uint64_t ticks = mach_absolute_time() - m_data->mStartTimeNano;
+	static long double conversion = 0.0L;
+	if (0.0L == conversion)
+	{
+		// attempt to get conversion to nanoseconds
+		mach_timebase_info_data_t info;
+		int err = mach_timebase_info(&info);
+		if (err)
+		{
+			btAssert(0);
+			conversion = 1.;
+		}
+		conversion = info.numer / info.denom;
+	}
+	return (ticks * conversion);
+
+#else  //__APPLE__
 
-    
-#else//__APPLE__
-    
 #ifdef BT_LINUX_REALTIME
-    timespec ts;
-    clock_gettime(CLOCK_REALTIME,&ts);
-    return 1000000000*ts.tv_sec + ts.tv_nsec;
+	timespec ts;
+	clock_gettime(CLOCK_REALTIME, &ts);
+	return 1000000000 * ts.tv_sec + ts.tv_nsec;
 #else
-    	struct timeval currentTime;
-		gettimeofday(&currentTime, 0);
-		return (currentTime.tv_sec - m_data->mStartTime.tv_sec) * 1e9 +
-			(currentTime.tv_usec - m_data->mStartTime.tv_usec)*1000;
-#endif //BT_LINUX_REALTIME
+	struct timeval currentTime;
+	gettimeofday(&currentTime, 0);
+	return (currentTime.tv_sec - m_data->mStartTime.tv_sec) * 1e9 +
+		   (currentTime.tv_usec - m_data->mStartTime.tv_usec) * 1000;
+#endif  //BT_LINUX_REALTIME
 
-#endif//__APPLE__
-#endif//__CELLOS_LV2__
-#endif 
+#endif  //__APPLE__
+#endif  //__CELLOS_LV2__
+#endif
 }
 
-
-/// Returns the time in s since the last call to reset or since 
+/// Returns the time in s since the last call to reset or since
 /// the Clock was created.
 btScalar btClock::getTimeSeconds()
 {
@@ -281,23 +272,19 @@ btScalar btClock::getTimeSeconds()
 
 #ifndef BT_NO_PROFILE
 
-
 static btClock gProfileClock;
 
-
-inline void Profile_Get_Ticks(unsigned long int * ticks)
+inline void Profile_Get_Ticks(unsigned long int* ticks)
 {
 	*ticks = (unsigned long int)gProfileClock.getTimeMicroseconds();
 }
 
 inline float Profile_Get_Tick_Rate(void)
 {
-//	return 1000000.f;
+	//	return 1000000.f;
 	return 1000.f;
-
 }
 
-
 /***************************************************************************************************
 **
 ** CProfileNode
@@ -313,35 +300,32 @@ inline float Profile_Get_Tick_Rate(void)
  * The name is assumed to be a static pointer, only the pointer is stored and compared for     *
  * efficiency reasons.                                                                         *
  *=============================================================================================*/
-CProfileNode::CProfileNode( const char * name, CProfileNode * parent ) :
-	Name( name ),
-	TotalCalls( 0 ),
-	TotalTime( 0 ),
-	StartTime( 0 ),
-	RecursionCounter( 0 ),
-	Parent( parent ),
-	Child( NULL ),
-	Sibling( NULL ),
-	m_userPtr(0)
+CProfileNode::CProfileNode(const char* name, CProfileNode* parent) : Name(name),
+																	 TotalCalls(0),
+																	 TotalTime(0),
+																	 StartTime(0),
+																	 RecursionCounter(0),
+																	 Parent(parent),
+																	 Child(NULL),
+																	 Sibling(NULL),
+																	 m_userPtr(0)
 {
 	Reset();
 }
 
-
-void	CProfileNode::CleanupMemory()
+void CProfileNode::CleanupMemory()
 {
-	delete ( Child);
+	delete (Child);
 	Child = NULL;
-	delete ( Sibling);
+	delete (Sibling);
 	Sibling = NULL;
 }
 
-CProfileNode::~CProfileNode( void )
+CProfileNode::~CProfileNode(void)
 {
 	CleanupMemory();
 }
 
-
 /***********************************************************************************************
  * INPUT:                                                                                      *
  * name - static string pointer to the name of the node we are searching for                   *
@@ -350,12 +334,14 @@ CProfileNode::~CProfileNode( void )
  * All profile names are assumed to be static strings so this function uses pointer compares   *
  * to find the named node.                                                                     *
  *=============================================================================================*/
-CProfileNode * CProfileNode::Get_Sub_Node( const char * name )
+CProfileNode* CProfileNode::Get_Sub_Node(const char* name)
 {
 	// Try to find this sub node
-	CProfileNode * child = Child;
-	while ( child ) {
-		if ( child->Name == name ) {
+	CProfileNode* child = Child;
+	while (child)
+	{
+		if (child->Name == name)
+		{
 			return child;
 		}
 		child = child->Sibling;
@@ -363,176 +349,212 @@ CProfileNode * CProfileNode::Get_Sub_Node( const char * name )
 
 	// We didn't find it, so add it
 
-	CProfileNode * node = new CProfileNode( name, this );
+	CProfileNode* node = new CProfileNode(name, this);
 	node->Sibling = Child;
 	Child = node;
 	return node;
 }
 
-
-void	CProfileNode::Reset( void )
+void CProfileNode::Reset(void)
 {
 	TotalCalls = 0;
 	TotalTime = 0.0f;
 
-
-	if ( Child ) {
+	if (Child)
+	{
 		Child->Reset();
 	}
-	if ( Sibling ) {
+	if (Sibling)
+	{
 		Sibling->Reset();
 	}
 }
 
-
-void	CProfileNode::Call( void )
+void CProfileNode::Call(void)
 {
 	TotalCalls++;
-	if (RecursionCounter++ == 0) {
+	if (RecursionCounter++ == 0)
+	{
 		Profile_Get_Ticks(&StartTime);
 	}
 }
 
-
-bool	CProfileNode::Return( void )
+bool CProfileNode::Return(void)
 {
-	if ( --RecursionCounter == 0 && TotalCalls != 0 ) {
+	if (--RecursionCounter == 0 && TotalCalls != 0)
+	{
 		unsigned long int time;
 		Profile_Get_Ticks(&time);
 
-		time-=StartTime;
+		time -= StartTime;
 		TotalTime += (float)time / Profile_Get_Tick_Rate();
 	}
-	return ( RecursionCounter == 0 );
+	return (RecursionCounter == 0);
 }
 
-
 /***************************************************************************************************
 **
 ** CProfileIterator
 **
 ***************************************************************************************************/
-CProfileIterator::CProfileIterator( CProfileNode * start )
+CProfileIterator::CProfileIterator(CProfileNode* start)
 {
 	CurrentParent = start;
 	CurrentChild = CurrentParent->Get_Child();
 }
 
-
-void	CProfileIterator::First(void)
+void CProfileIterator::First(void)
 {
 	CurrentChild = CurrentParent->Get_Child();
 }
 
-
-void	CProfileIterator::Next(void)
+void CProfileIterator::Next(void)
 {
 	CurrentChild = CurrentChild->Get_Sibling();
 }
 
-
-bool	CProfileIterator::Is_Done(void)
+bool CProfileIterator::Is_Done(void)
 {
 	return CurrentChild == NULL;
 }
 
-
-void	CProfileIterator::Enter_Child( int index )
+void CProfileIterator::Enter_Child(int index)
 {
 	CurrentChild = CurrentParent->Get_Child();
-	while ( (CurrentChild != NULL) && (index != 0) ) {
+	while ((CurrentChild != NULL) && (index != 0))
+	{
 		index--;
 		CurrentChild = CurrentChild->Get_Sibling();
 	}
 
-	if ( CurrentChild != NULL ) {
+	if (CurrentChild != NULL)
+	{
 		CurrentParent = CurrentChild;
 		CurrentChild = CurrentParent->Get_Child();
 	}
 }
 
-
-void	CProfileIterator::Enter_Parent( void )
+void CProfileIterator::Enter_Parent(void)
 {
-	if ( CurrentParent->Get_Parent() != NULL ) {
+	if (CurrentParent->Get_Parent() != NULL)
+	{
 		CurrentParent = CurrentParent->Get_Parent();
 	}
 	CurrentChild = CurrentParent->Get_Child();
 }
 
-
 /***************************************************************************************************
 **
 ** CProfileManager
 **
 ***************************************************************************************************/
 
+CProfileNode gRoots[BT_QUICKPROF_MAX_THREAD_COUNT] = {
+	CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL),
+	CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL),
+	CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL),
+	CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL),
+	CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL),
+	CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL),
+	CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL),
+	CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL),
+	CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL),
+	CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL),
+	CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL),
+	CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL),
+	CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL),
+	CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL),
+	CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL),
+	CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL), CProfileNode("Root", NULL)};
 
-
-
-CProfileNode	gRoots[BT_QUICKPROF_MAX_THREAD_COUNT]={
-	CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),
-	CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),
-	CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),
-	CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),
-	CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),
-	CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),
-	CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),
-	CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),
-	CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),
-	CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),
-	CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),
-	CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),
-	CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),
-	CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),
-	CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),
-	CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL),CProfileNode("Root",NULL)
+CProfileNode* gCurrentNodes[BT_QUICKPROF_MAX_THREAD_COUNT] =
+	{
+		&gRoots[0],
+		&gRoots[1],
+		&gRoots[2],
+		&gRoots[3],
+		&gRoots[4],
+		&gRoots[5],
+		&gRoots[6],
+		&gRoots[7],
+		&gRoots[8],
+		&gRoots[9],
+		&gRoots[10],
+		&gRoots[11],
+		&gRoots[12],
+		&gRoots[13],
+		&gRoots[14],
+		&gRoots[15],
+		&gRoots[16],
+		&gRoots[17],
+		&gRoots[18],
+		&gRoots[19],
+		&gRoots[20],
+		&gRoots[21],
+		&gRoots[22],
+		&gRoots[23],
+		&gRoots[24],
+		&gRoots[25],
+		&gRoots[26],
+		&gRoots[27],
+		&gRoots[28],
+		&gRoots[29],
+		&gRoots[30],
+		&gRoots[31],
+		&gRoots[32],
+		&gRoots[33],
+		&gRoots[34],
+		&gRoots[35],
+		&gRoots[36],
+		&gRoots[37],
+		&gRoots[38],
+		&gRoots[39],
+		&gRoots[40],
+		&gRoots[41],
+		&gRoots[42],
+		&gRoots[43],
+		&gRoots[44],
+		&gRoots[45],
+		&gRoots[46],
+		&gRoots[47],
+		&gRoots[48],
+		&gRoots[49],
+		&gRoots[50],
+		&gRoots[51],
+		&gRoots[52],
+		&gRoots[53],
+		&gRoots[54],
+		&gRoots[55],
+		&gRoots[56],
+		&gRoots[57],
+		&gRoots[58],
+		&gRoots[59],
+		&gRoots[60],
+		&gRoots[61],
+		&gRoots[62],
+		&gRoots[63],
 };
 
+int CProfileManager::FrameCounter = 0;
+unsigned long int CProfileManager::ResetTime = 0;
 
-CProfileNode* gCurrentNodes[BT_QUICKPROF_MAX_THREAD_COUNT]=
+CProfileIterator* CProfileManager::Get_Iterator(void)
 {
-	&gRoots[ 0],	&gRoots[ 1],	&gRoots[ 2],	&gRoots[ 3],
-	&gRoots[ 4],	&gRoots[ 5],	&gRoots[ 6],	&gRoots[ 7],
-	&gRoots[ 8],	&gRoots[ 9],	&gRoots[10],	&gRoots[11],
-	&gRoots[12],	&gRoots[13],	&gRoots[14],	&gRoots[15],
-	&gRoots[16],	&gRoots[17],	&gRoots[18],	&gRoots[19],
-	&gRoots[20],	&gRoots[21],	&gRoots[22],	&gRoots[23],
-	&gRoots[24],	&gRoots[25],	&gRoots[26],	&gRoots[27],
-	&gRoots[28],	&gRoots[29],	&gRoots[30],	&gRoots[31],
-	&gRoots[32],	&gRoots[33],	&gRoots[34],	&gRoots[35],
-	&gRoots[36],	&gRoots[37],	&gRoots[38],	&gRoots[39],
-	&gRoots[40],	&gRoots[41],	&gRoots[42],	&gRoots[43],
-	&gRoots[44],	&gRoots[45],	&gRoots[46],	&gRoots[47],
-	&gRoots[48],	&gRoots[49],	&gRoots[50],	&gRoots[51],
-	&gRoots[52],	&gRoots[53],	&gRoots[54],	&gRoots[55],
-	&gRoots[56],	&gRoots[57],	&gRoots[58],	&gRoots[59],
-	&gRoots[60],	&gRoots[61],	&gRoots[62],	&gRoots[63],
-};
+	int threadIndex = btQuickprofGetCurrentThreadIndex2();
+	if ((threadIndex < 0) || threadIndex >= BT_QUICKPROF_MAX_THREAD_COUNT)
+		return 0;
 
-
-int				CProfileManager::FrameCounter = 0;
-unsigned long int			CProfileManager::ResetTime = 0;
-
-CProfileIterator *	CProfileManager::Get_Iterator( void )
-{ 
-
-		int threadIndex = btQuickprofGetCurrentThreadIndex2();
-		if ((threadIndex<0) || threadIndex >= BT_QUICKPROF_MAX_THREAD_COUNT)
-			return 0;
-
-		return new CProfileIterator( &gRoots[threadIndex]); 
+	return new CProfileIterator(&gRoots[threadIndex]);
 }
 
-void						CProfileManager::CleanupMemory(void)
+void CProfileManager::CleanupMemory(void)
 {
-	for (int i=0;i<BT_QUICKPROF_MAX_THREAD_COUNT;i++)
+	for (int i = 0; i < BT_QUICKPROF_MAX_THREAD_COUNT; i++)
 	{
 		gRoots[i].CleanupMemory();
 	}
 }
 
-
 /***********************************************************************************************
  * CProfileManager::Start_Profile -- Begin a named profile                                    *
  *                                                                                             *
@@ -546,51 +568,47 @@ void						CProfileManager::CleanupMemory(void)
  * The string used is assumed to be a static string; pointer compares are used throughout      *
  * the profiling code for efficiency.                                                          *
  *=============================================================================================*/
-void	CProfileManager::Start_Profile( const char * name )
+void CProfileManager::Start_Profile(const char* name)
 {
 	int threadIndex = btQuickprofGetCurrentThreadIndex2();
-	if ((threadIndex<0) || threadIndex >= BT_QUICKPROF_MAX_THREAD_COUNT)
+	if ((threadIndex < 0) || threadIndex >= BT_QUICKPROF_MAX_THREAD_COUNT)
 		return;
 
-	if (name != gCurrentNodes[threadIndex]->Get_Name()) {
-		gCurrentNodes[threadIndex] = gCurrentNodes[threadIndex]->Get_Sub_Node( name );
+	if (name != gCurrentNodes[threadIndex]->Get_Name())
+	{
+		gCurrentNodes[threadIndex] = gCurrentNodes[threadIndex]->Get_Sub_Node(name);
 	}
 
 	gCurrentNodes[threadIndex]->Call();
 }
 
-
 /***********************************************************************************************
  * CProfileManager::Stop_Profile -- Stop timing and record the results.                       *
  *=============================================================================================*/
-void	CProfileManager::Stop_Profile( void )
+void CProfileManager::Stop_Profile(void)
 {
 	int threadIndex = btQuickprofGetCurrentThreadIndex2();
-	if ((threadIndex<0) || threadIndex >= BT_QUICKPROF_MAX_THREAD_COUNT)
+	if ((threadIndex < 0) || threadIndex >= BT_QUICKPROF_MAX_THREAD_COUNT)
 		return;
 
 	// Return will indicate whether we should back up to our parent (we may
 	// be profiling a recursive function)
-	if (gCurrentNodes[threadIndex]->Return()) {
+	if (gCurrentNodes[threadIndex]->Return())
+	{
 		gCurrentNodes[threadIndex] = gCurrentNodes[threadIndex]->Get_Parent();
 	}
 }
 
-
-
-
-
-
 /***********************************************************************************************
  * CProfileManager::Reset -- Reset the contents of the profiling system                       *
  *                                                                                             *
  *    This resets everything except for the tree structure.  All of the timing data is reset.  *
  *=============================================================================================*/
-void	CProfileManager::Reset( void )
+void CProfileManager::Reset(void)
 {
 	gProfileClock.reset();
 	int threadIndex = btQuickprofGetCurrentThreadIndex2();
-	if ((threadIndex<0) || threadIndex >= BT_QUICKPROF_MAX_THREAD_COUNT)
+	if ((threadIndex < 0) || threadIndex >= BT_QUICKPROF_MAX_THREAD_COUNT)
 		return;
 	gRoots[threadIndex].Reset();
 	gRoots[threadIndex].Call();
@@ -598,20 +616,18 @@ void	CProfileManager::Reset( void )
 	Profile_Get_Ticks(&ResetTime);
 }
 
-
 /***********************************************************************************************
  * CProfileManager::Increment_Frame_Counter -- Increment the frame counter                    *
  *=============================================================================================*/
-void CProfileManager::Increment_Frame_Counter( void )
+void CProfileManager::Increment_Frame_Counter(void)
 {
 	FrameCounter++;
 }
 
-
 /***********************************************************************************************
  * CProfileManager::Get_Time_Since_Reset -- returns the elapsed time since last reset         *
  *=============================================================================================*/
-float CProfileManager::Get_Time_Since_Reset( void )
+float CProfileManager::Get_Time_Since_Reset(void)
 {
 	unsigned long int time;
 	Profile_Get_Ticks(&time);
@@ -621,34 +637,34 @@ float CProfileManager::Get_Time_Since_Reset( void )
 
 #include <stdio.h>
 
-void	CProfileManager::dumpRecursive(CProfileIterator* profileIterator, int spacing)
+void CProfileManager::dumpRecursive(CProfileIterator* profileIterator, int spacing)
 {
 	profileIterator->First();
 	if (profileIterator->Is_Done())
 		return;
 
-	float accumulated_time=0,parent_time = profileIterator->Is_Root() ? CProfileManager::Get_Time_Since_Reset() : profileIterator->Get_Current_Parent_Total_Time();
+	float accumulated_time = 0, parent_time = profileIterator->Is_Root() ? CProfileManager::Get_Time_Since_Reset() : profileIterator->Get_Current_Parent_Total_Time();
 	int i;
 	int frames_since_reset = CProfileManager::Get_Frame_Count_Since_Reset();
-	for (i=0;i<spacing;i++)	printf(".");
+	for (i = 0; i < spacing; i++) printf(".");
 	printf("----------------------------------\n");
-	for (i=0;i<spacing;i++)	printf(".");
-	printf("Profiling: %s (total running time: %.3f ms) ---\n",	profileIterator->Get_Current_Parent_Name(), parent_time );
+	for (i = 0; i < spacing; i++) printf(".");
+	printf("Profiling: %s (total running time: %.3f ms) ---\n", profileIterator->Get_Current_Parent_Name(), parent_time);
 	float totalTime = 0.f;
 
-
 	int numChildren = 0;
 
-	for (i = 0; !profileIterator->Is_Done(); i++,profileIterator->Next())
+	for (i = 0; !profileIterator->Is_Done(); i++, profileIterator->Next())
 	{
 		numChildren++;
 		float current_total_time = profileIterator->Get_Current_Total_Time();
 		accumulated_time += current_total_time;
 		float fraction = parent_time > SIMD_EPSILON ? (current_total_time / parent_time) * 100 : 0.f;
 		{
-			int i;	for (i=0;i<spacing;i++)	printf(".");
+			int i;
+			for (i = 0; i < spacing; i++) printf(".");
 		}
-		printf("%d -- %s (%.2f %%) :: %.3f ms / frame (%d calls)\n",i, profileIterator->Get_Current_Name(), fraction,(current_total_time / (double)frames_since_reset),profileIterator->Get_Current_Total_Calls());
+		printf("%d -- %s (%.2f %%) :: %.3f ms / frame (%d calls)\n", i, profileIterator->Get_Current_Name(), fraction, (current_total_time / (double)frames_since_reset), profileIterator->Get_Current_Total_Calls());
 		totalTime += current_total_time;
 		//recurse into children
 	}
@@ -657,29 +673,45 @@ void	CProfileManager::dumpRecursive(CProfileIterator* profileIterator, int spaci
 	{
 		//printf("what's wrong\n");
 	}
-	for (i=0;i<spacing;i++)	printf(".");
-	printf("%s (%.3f %%) :: %.3f ms\n", "Unaccounted:",parent_time > SIMD_EPSILON ? ((parent_time - accumulated_time) / parent_time) * 100 : 0.f, parent_time - accumulated_time);
+	for (i = 0; i < spacing; i++) printf(".");
+	printf("%s (%.3f %%) :: %.3f ms\n", "Unaccounted:", parent_time > SIMD_EPSILON ? ((parent_time - accumulated_time) / parent_time) * 100 : 0.f, parent_time - accumulated_time);
 
-	for (i=0;i<numChildren;i++)
+	for (i = 0; i < numChildren; i++)
 	{
 		profileIterator->Enter_Child(i);
-		dumpRecursive(profileIterator,spacing+3);
+		dumpRecursive(profileIterator, spacing + 3);
 		profileIterator->Enter_Parent();
 	}
 }
 
-
-
-void	CProfileManager::dumpAll()
+void CProfileManager::dumpAll()
 {
 	CProfileIterator* profileIterator = 0;
 	profileIterator = CProfileManager::Get_Iterator();
 
-	dumpRecursive(profileIterator,0);
+	dumpRecursive(profileIterator, 0);
 
 	CProfileManager::Release_Iterator(profileIterator);
 }
 
+
+void btEnterProfileZoneDefault(const char* name)
+{
+}
+void btLeaveProfileZoneDefault()
+{
+}
+
+#else
+void btEnterProfileZoneDefault(const char* name)
+{
+}
+void btLeaveProfileZoneDefault()
+{
+}
+#endif  //BT_NO_PROFILE
+
+
 // clang-format off
 #if defined(_WIN32) && (defined(__MINGW32__) || defined(__MINGW64__))
   #define BT_HAVE_TLS 1
@@ -703,51 +735,32 @@ void	CProfileManager::dumpAll()
 #endif  // defined(__ANDROID__) && defined(__clang__)
 // clang-format on
 
-unsigned int btQuickprofGetCurrentThreadIndex2() {
-  const unsigned int kNullIndex = ~0U;
+unsigned int btQuickprofGetCurrentThreadIndex2()
+{
+	const unsigned int kNullIndex = ~0U;
 
 #if BT_THREADSAFE
-  return btGetCurrentThreadIndex();
+	return btGetCurrentThreadIndex();
 #else
 #if defined(BT_HAVE_TLS)
-  static __thread unsigned int sThreadIndex = kNullIndex;
+	static __thread unsigned int sThreadIndex = kNullIndex;
 #elif defined(_WIN32)
-  __declspec(thread) static unsigned int sThreadIndex = kNullIndex;
+	__declspec(thread) static unsigned int sThreadIndex = kNullIndex;
 #else
-  unsigned int sThreadIndex = 0;
-  return -1;
+	unsigned int sThreadIndex = 0;
+	return -1;
 #endif
 
-  static int gThreadCounter = 0;
+	static int gThreadCounter = 0;
 
-  if (sThreadIndex == kNullIndex) {
-    sThreadIndex = gThreadCounter++;
-  }
-  return sThreadIndex;
-#endif //BT_THREADSAFE
+	if (sThreadIndex == kNullIndex)
+	{
+		sThreadIndex = gThreadCounter++;
+	}
+	return sThreadIndex;
+#endif  //BT_THREADSAFE
 }
 
-void	btEnterProfileZoneDefault(const char* name)
-{
-}
-void	btLeaveProfileZoneDefault()
-{
-}
-
-
-#else
-void	btEnterProfileZoneDefault(const char* name)
-{
-}
-void	btLeaveProfileZoneDefault()
-{
-}
-#endif //BT_NO_PROFILE
-
-
-
-
-
 static btEnterProfileZoneFunc* bts_enterFunc = btEnterProfileZoneDefault;
 static btLeaveProfileZoneFunc* bts_leaveFunc = btLeaveProfileZoneDefault;
 
@@ -762,14 +775,13 @@ void btLeaveProfileZone()
 
 btEnterProfileZoneFunc* btGetCurrentEnterProfileZoneFunc()
 {
-	return bts_enterFunc ;
+	return bts_enterFunc;
 }
 btLeaveProfileZoneFunc* btGetCurrentLeaveProfileZoneFunc()
 {
 	return bts_leaveFunc;
 }
 
-
 void btSetCustomEnterProfileZoneFunc(btEnterProfileZoneFunc* enterFunc)
 {
 	bts_enterFunc = enterFunc;
@@ -779,13 +791,12 @@ void btSetCustomLeaveProfileZoneFunc(btLeaveProfileZoneFunc* leaveFunc)
 	bts_leaveFunc = leaveFunc;
 }
 
-CProfileSample::CProfileSample( const char * name )
-{ 
+CProfileSample::CProfileSample(const char* name)
+{
 	btEnterProfileZone(name);
 }
 
-CProfileSample::~CProfileSample( void )					
-{ 
+CProfileSample::~CProfileSample(void)
+{
 	btLeaveProfileZone();
 }
-
diff --git a/thirdparty/bullet/LinearMath/btQuickprof.h b/thirdparty/bullet/LinearMath/btQuickprof.h
index 98a26757714..990d401d507 100644
--- a/thirdparty/bullet/LinearMath/btQuickprof.h
+++ b/thirdparty/bullet/LinearMath/btQuickprof.h
@@ -7,11 +7,9 @@
 **
 ***************************************************************************************************/
 
-// Credits: The Clock class was inspired by the Timer classes in 
+// Credits: The Clock class was inspired by the Timer classes in
 // Ogre (www.ogre3d.org).
 
-
-
 #ifndef BT_QUICK_PROF_H
 #define BT_QUICK_PROF_H
 
@@ -34,97 +32,88 @@ public:
 	/// Resets the initial reference time.
 	void reset();
 
-	/// Returns the time in ms since the last call to reset or since 
+	/// Returns the time in ms since the last call to reset or since
 	/// the btClock was created.
 	unsigned long long int getTimeMilliseconds();
 
-	/// Returns the time in us since the last call to reset or since 
+	/// Returns the time in us since the last call to reset or since
 	/// the Clock was created.
 	unsigned long long int getTimeMicroseconds();
-	
+
 	unsigned long long int getTimeNanoseconds();
 
-	/// Returns the time in s since the last call to reset or since 
+	/// Returns the time in s since the last call to reset or since
 	/// the Clock was created.
 	btScalar getTimeSeconds();
-	
+
 private:
 	struct btClockData* m_data;
 };
 
-#endif //USE_BT_CLOCK
+#endif  //USE_BT_CLOCK
 
-typedef void (btEnterProfileZoneFunc)(const char* msg);
-typedef void (btLeaveProfileZoneFunc)();
+typedef void(btEnterProfileZoneFunc)(const char* msg);
+typedef void(btLeaveProfileZoneFunc)();
 
 btEnterProfileZoneFunc* btGetCurrentEnterProfileZoneFunc();
 btLeaveProfileZoneFunc* btGetCurrentLeaveProfileZoneFunc();
 
-
-
 void btSetCustomEnterProfileZoneFunc(btEnterProfileZoneFunc* enterFunc);
 void btSetCustomLeaveProfileZoneFunc(btLeaveProfileZoneFunc* leaveFunc);
 
-#ifndef BT_NO_PROFILE // FIX redefinition
-//To disable built-in profiling, please comment out next line
-//#define BT_NO_PROFILE 1
-#endif //BT_NO_PROFILE
+#ifndef BT_ENABLE_PROFILE
+#define BT_NO_PROFILE 1
+#endif  //BT_NO_PROFILE
 
 const unsigned int BT_QUICKPROF_MAX_THREAD_COUNT = 64;
 
-#ifndef BT_NO_PROFILE
-//btQuickprofGetCurrentThreadIndex will return -1 if thread index cannot be determined, 
+//btQuickprofGetCurrentThreadIndex will return -1 if thread index cannot be determined,
 //otherwise returns thread index in range [0..maxThreads]
 unsigned int btQuickprofGetCurrentThreadIndex2();
 
-#include <stdio.h>//@todo remove this, backwards compatibility
+#ifndef BT_NO_PROFILE
+
+
+#include <stdio.h>  //@todo remove this, backwards compatibility
 
 #include "btAlignedAllocator.h"
 #include <new>
 
-
-
-
-
-
-
-
-
 ///A node in the Profile Hierarchy Tree
-class	CProfileNode {
-
+class CProfileNode
+{
 public:
-	CProfileNode( const char * name, CProfileNode * parent );
-	~CProfileNode( void );
+	CProfileNode(const char* name, CProfileNode* parent);
+	~CProfileNode(void);
 
-	CProfileNode * Get_Sub_Node( const char * name );
+	CProfileNode* Get_Sub_Node(const char* name);
 
-	CProfileNode * Get_Parent( void )		{ return Parent; }
-	CProfileNode * Get_Sibling( void )		{ return Sibling; }
-	CProfileNode * Get_Child( void )			{ return Child; }
+	CProfileNode* Get_Parent(void) { return Parent; }
+	CProfileNode* Get_Sibling(void) { return Sibling; }
+	CProfileNode* Get_Child(void) { return Child; }
 
-	void				CleanupMemory();
-	void				Reset( void );
-	void				Call( void );
-	bool				Return( void );
+	void CleanupMemory();
+	void Reset(void);
+	void Call(void);
+	bool Return(void);
+
+	const char* Get_Name(void) { return Name; }
+	int Get_Total_Calls(void) { return TotalCalls; }
+	float Get_Total_Time(void) { return TotalTime; }
+	void* GetUserPointer() const { return m_userPtr; }
+	void SetUserPointer(void* ptr) { m_userPtr = ptr; }
 
-	const char *	Get_Name( void )				{ return Name; }
-	int				Get_Total_Calls( void )		{ return TotalCalls; }
-	float				Get_Total_Time( void )		{ return TotalTime; }
-	void*			GetUserPointer() const {return m_userPtr;}
-	void			SetUserPointer(void* ptr) { m_userPtr = ptr;}
 protected:
+	const char* Name;
+	int TotalCalls;
+	float TotalTime;
+	unsigned long int StartTime;
+	int RecursionCounter;
 
-	const char *	Name;
-	int				TotalCalls;
-	float				TotalTime;
-	unsigned long int			StartTime;
-	int				RecursionCounter;
-
-	CProfileNode *	Parent;
-	CProfileNode *	Child;
-	CProfileNode *	Sibling;
-	void*	m_userPtr;
+	CProfileNode* Parent;
+	CProfileNode* Child;
+	CProfileNode* Sibling;
+	void* m_userPtr;
 };
 
 ///An iterator to navigate through the tree
@@ -132,91 +121,80 @@ class CProfileIterator
 {
 public:
 	// Access all the children of the current parent
-	void				First(void);
-	void				Next(void);
-	bool				Is_Done(void);
-	bool                Is_Root(void) { return (CurrentParent->Get_Parent() == 0); }
+	void First(void);
+	void Next(void);
+	bool Is_Done(void);
+	bool Is_Root(void) { return (CurrentParent->Get_Parent() == 0); }
 
-	void				Enter_Child( int index );		// Make the given child the new parent
-	void				Enter_Largest_Child( void );	// Make the largest child the new parent
-	void				Enter_Parent( void );			// Make the current parent's parent the new parent
+	void Enter_Child(int index);     // Make the given child the new parent
+	void Enter_Largest_Child(void);  // Make the largest child the new parent
+	void Enter_Parent(void);         // Make the current parent's parent the new parent
 
 	// Access the current child
-	const char *	Get_Current_Name( void )			{ return CurrentChild->Get_Name(); }
-	int				Get_Current_Total_Calls( void )	{ return CurrentChild->Get_Total_Calls(); }
-	float				Get_Current_Total_Time( void )	{ return CurrentChild->Get_Total_Time(); }
+	const char* Get_Current_Name(void) { return CurrentChild->Get_Name(); }
+	int Get_Current_Total_Calls(void) { return CurrentChild->Get_Total_Calls(); }
+	float Get_Current_Total_Time(void) { return CurrentChild->Get_Total_Time(); }
 
-	void*	Get_Current_UserPointer( void )			{ return CurrentChild->GetUserPointer(); }
-	void	Set_Current_UserPointer(void* ptr) {CurrentChild->SetUserPointer(ptr);}
+	void* Get_Current_UserPointer(void) { return CurrentChild->GetUserPointer(); }
+	void Set_Current_UserPointer(void* ptr) { CurrentChild->SetUserPointer(ptr); }
 	// Access the current parent
-	const char *	Get_Current_Parent_Name( void )			{ return CurrentParent->Get_Name(); }
-	int				Get_Current_Parent_Total_Calls( void )	{ return CurrentParent->Get_Total_Calls(); }
-	float				Get_Current_Parent_Total_Time( void )	{ return CurrentParent->Get_Total_Time(); }
-
-	
+	const char* Get_Current_Parent_Name(void) { return CurrentParent->Get_Name(); }
+	int Get_Current_Parent_Total_Calls(void) { return CurrentParent->Get_Total_Calls(); }
+	float Get_Current_Parent_Total_Time(void) { return CurrentParent->Get_Total_Time(); }
 
 protected:
+	CProfileNode* CurrentParent;
+	CProfileNode* CurrentChild;
 
-	CProfileNode *	CurrentParent;
-	CProfileNode *	CurrentChild;
-	
-
-	CProfileIterator( CProfileNode * start );
-	friend	class		CProfileManager;
+	CProfileIterator(CProfileNode* start);
+	friend class CProfileManager;
 };
 
-
 ///The Manager for the Profile system
-class	CProfileManager {
+class CProfileManager
+{
 public:
-	static	void						Start_Profile( const char * name );
-	static	void						Stop_Profile( void );
+	static void Start_Profile(const char* name);
+	static void Stop_Profile(void);
 
-	static	void						CleanupMemory(void);
-//	{
-//		Root.CleanupMemory();
-//	}
+	static void CleanupMemory(void);
+	//	{
+	//		Root.CleanupMemory();
+	//	}
 
-	static	void						Reset( void );
-	static	void						Increment_Frame_Counter( void );
-	static	int						Get_Frame_Count_Since_Reset( void )		{ return FrameCounter; }
-	static	float						Get_Time_Since_Reset( void );
+	static void Reset(void);
+	static void Increment_Frame_Counter(void);
+	static int Get_Frame_Count_Since_Reset(void) { return FrameCounter; }
+	static float Get_Time_Since_Reset(void);
 
-	static	CProfileIterator *	Get_Iterator( void );	
-//	{ 
-//		
-//		return new CProfileIterator( &Root ); 
-//	}
-	static	void						Release_Iterator( CProfileIterator * iterator ) { delete ( iterator); }
+	static CProfileIterator* Get_Iterator(void);
+	//	{
+	//
+	//		return new CProfileIterator( &Root );
+	//	}
+	static void Release_Iterator(CProfileIterator* iterator) { delete (iterator); }
 
-	static void	dumpRecursive(CProfileIterator* profileIterator, int spacing);
+	static void dumpRecursive(CProfileIterator* profileIterator, int spacing);
 
-	static void	dumpAll();
+	static void dumpAll();
 
 private:
-
-	static	int						FrameCounter;
-	static	unsigned long int					ResetTime;
+	static int FrameCounter;
+	static unsigned long int ResetTime;
 };
 
-
-
-
-#endif //#ifndef BT_NO_PROFILE
+#endif  //#ifndef BT_NO_PROFILE
 
 ///ProfileSampleClass is a simple way to profile a function's scope
 ///Use the BT_PROFILE macro at the start of scope to time
-class	CProfileSample {
+class CProfileSample
+{
 public:
-	CProfileSample( const char * name );
+	CProfileSample(const char* name);
 
-	~CProfileSample( void );
+	~CProfileSample(void);
 };
 
-#define	BT_PROFILE( name )			CProfileSample __profile( name )
-
-
-
-#endif //BT_QUICK_PROF_H
-
+#define BT_PROFILE(name) CProfileSample __profile(name)
 
+#endif  //BT_QUICK_PROF_H
diff --git a/thirdparty/bullet/LinearMath/btRandom.h b/thirdparty/bullet/LinearMath/btRandom.h
index 4cbfc6bfe9f..e659af86055 100644
--- a/thirdparty/bullet/LinearMath/btRandom.h
+++ b/thirdparty/bullet/LinearMath/btRandom.h
@@ -12,8 +12,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #ifndef BT_GEN_RANDOM_H
 #define BT_GEN_RANDOM_H
 
@@ -24,8 +22,8 @@ subject to the following restrictions:
 
 #define GEN_RAND_MAX UINT_MAX
 
-SIMD_FORCE_INLINE void         GEN_srand(unsigned int seed) { init_genrand(seed); }
-SIMD_FORCE_INLINE unsigned int GEN_rand()                   { return genrand_int32(); }
+SIMD_FORCE_INLINE void GEN_srand(unsigned int seed) { init_genrand(seed); }
+SIMD_FORCE_INLINE unsigned int GEN_rand() { return genrand_int32(); }
 
 #else
 
@@ -33,10 +31,9 @@ SIMD_FORCE_INLINE unsigned int GEN_rand()                   { return genrand_int
 
 #define GEN_RAND_MAX RAND_MAX
 
-SIMD_FORCE_INLINE void         GEN_srand(unsigned int seed) { srand(seed); } 
-SIMD_FORCE_INLINE unsigned int GEN_rand()                   { return rand(); }
+SIMD_FORCE_INLINE void GEN_srand(unsigned int seed) { srand(seed); }
+SIMD_FORCE_INLINE unsigned int GEN_rand() { return rand(); }
 
 #endif
 
-#endif //BT_GEN_RANDOM_H
-
+#endif  //BT_GEN_RANDOM_H
diff --git a/thirdparty/bullet/LinearMath/btScalar.h b/thirdparty/bullet/LinearMath/btScalar.h
index 24e8454c1f2..c198bd4b350 100644
--- a/thirdparty/bullet/LinearMath/btScalar.h
+++ b/thirdparty/bullet/LinearMath/btScalar.h
@@ -32,7 +32,6 @@ inline int btGetVersion()
 	return BT_BULLET_VERSION;
 }
 
-
 // The following macro "BT_NOT_EMPTY_FILE" can be put into a file
 // in order suppress the MS Visual C++ Linker warning 4221
 //
@@ -44,16 +43,19 @@ inline int btGetVersion()
 //
 // see more https://stackoverflow.com/questions/1822887/what-is-the-best-way-to-eliminate-ms-visual-c-linker-warning-warning-lnk422
 
-#if defined (_MSC_VER)
-	#define BT_NOT_EMPTY_FILE_CAT_II(p, res) res
-	#define BT_NOT_EMPTY_FILE_CAT_I(a, b) BT_NOT_EMPTY_FILE_CAT_II(~, a ## b)
-	#define BT_NOT_EMPTY_FILE_CAT(a, b) BT_NOT_EMPTY_FILE_CAT_I(a, b)
-	#define BT_NOT_EMPTY_FILE namespace { char BT_NOT_EMPTY_FILE_CAT(NoEmptyFileDummy, __COUNTER__); }
+#if defined(_MSC_VER)
+#define BT_NOT_EMPTY_FILE_CAT_II(p, res) res
+#define BT_NOT_EMPTY_FILE_CAT_I(a, b) BT_NOT_EMPTY_FILE_CAT_II(~, a##b)
+#define BT_NOT_EMPTY_FILE_CAT(a, b) BT_NOT_EMPTY_FILE_CAT_I(a, b)
+#define BT_NOT_EMPTY_FILE                                      \
+	namespace                                                  \
+	{                                                          \
+	char BT_NOT_EMPTY_FILE_CAT(NoEmptyFileDummy, __COUNTER__); \
+	}
 #else
-	#define BT_NOT_EMPTY_FILE 
+#define BT_NOT_EMPTY_FILE
 #endif
 
-
 // clang and most formatting tools don't support indentation of preprocessor guards, so turn it off
 // clang-format off
 #if defined(DEBUG) || defined (_DEBUG)
diff --git a/thirdparty/bullet/LinearMath/btSerializer.cpp b/thirdparty/bullet/LinearMath/btSerializer.cpp
index 4faa8f536bf..18683c8fa7f 100644
--- a/thirdparty/bullet/LinearMath/btSerializer.cpp
+++ b/thirdparty/bullet/LinearMath/btSerializer.cpp
@@ -1,3 +1,4 @@
+// clang-format off
 char sBulletDNAstr[]= {
 char(83),char(68),char(78),char(65),char(78),char(65),char(77),char(69),char(-76),char(1),char(0),char(0),char(109),char(95),char(115),char(105),char(122),char(101),char(0),char(109),
 char(95),char(99),char(97),char(112),char(97),char(99),char(105),char(116),char(121),char(0),char(42),char(109),char(95),char(100),char(97),char(116),char(97),char(0),char(109),char(95),
@@ -687,3 +688,5 @@ char(97),char(0),char(4),char(0),char(50),char(0),char(-79),char(1),char(96),cha
 char(98),char(0),char(4),char(0),char(48),char(0),char(-79),char(1),char(95),char(0),char(-78),char(1),char(4),char(0),char(-77),char(1),char(0),char(0),char(37),char(0),
 };
 int sBulletDNAlen= sizeof(sBulletDNAstr);
+
+// clang-format on
diff --git a/thirdparty/bullet/LinearMath/btSerializer.h b/thirdparty/bullet/LinearMath/btSerializer.h
index 39be3f810e4..ba344416154 100644
--- a/thirdparty/bullet/LinearMath/btSerializer.h
+++ b/thirdparty/bullet/LinearMath/btSerializer.h
@@ -16,49 +16,45 @@ subject to the following restrictions:
 #ifndef BT_SERIALIZER_H
 #define BT_SERIALIZER_H
 
-#include "btScalar.h" // has definitions like SIMD_FORCE_INLINE
+#include "btScalar.h"  // has definitions like SIMD_FORCE_INLINE
 #include "btHashMap.h"
 
-#if !defined( __CELLOS_LV2__) && !defined(__MWERKS__)
+#if !defined(__CELLOS_LV2__) && !defined(__MWERKS__)
 #include <memory.h>
 #endif
 #include <string.h>
 
-
-
-
 extern char sBulletDNAstr[];
 extern int sBulletDNAlen;
 extern char sBulletDNAstr64[];
 extern int sBulletDNAlen64;
 
-SIMD_FORCE_INLINE	int btStrLen(const char* str)
+SIMD_FORCE_INLINE int btStrLen(const char* str)
 {
-    if (!str)
-		return(0);
+	if (!str)
+		return (0);
 	int len = 0;
 
 	while (*str != 0)
 	{
-        str++;
-        len++;
-    }
+		str++;
+		len++;
+	}
 
-    return len;
+	return len;
 }
 
-
 class btChunk
 {
 public:
-	int		m_chunkCode;
-	int		m_length;
-	void	*m_oldPtr;
-	int		m_dna_nr;
-	int		m_number;
+	int m_chunkCode;
+	int m_length;
+	void* m_oldPtr;
+	int m_dna_nr;
+	int m_number;
 };
 
-enum	btSerializationFlags
+enum btSerializationFlags
 {
 	BT_SERIALIZE_NO_BVH = 1,
 	BT_SERIALIZE_NO_TRIANGLEINFOMAP = 2,
@@ -66,78 +62,71 @@ enum	btSerializationFlags
 	BT_SERIALIZE_CONTACT_MANIFOLDS = 8,
 };
 
-class	btSerializer
+class btSerializer
 {
-
 public:
-
 	virtual ~btSerializer() {}
 
-	virtual	const unsigned char*		getBufferPointer() const = 0;
+	virtual const unsigned char* getBufferPointer() const = 0;
 
-	virtual	int		getCurrentBufferSize() const = 0;
+	virtual int getCurrentBufferSize() const = 0;
 
-	virtual	btChunk*	allocate(size_t size, int numElements) = 0;
+	virtual btChunk* allocate(size_t size, int numElements) = 0;
 
-	virtual	void	finalizeChunk(btChunk* chunk, const char* structType, int chunkCode,void* oldPtr)= 0;
+	virtual void finalizeChunk(btChunk* chunk, const char* structType, int chunkCode, void* oldPtr) = 0;
 
-	virtual	 void*	findPointer(void* oldPtr)  = 0;
+	virtual void* findPointer(void* oldPtr) = 0;
 
-	virtual	void*	getUniquePointer(void*oldPtr) = 0;
+	virtual void* getUniquePointer(void* oldPtr) = 0;
 
-	virtual	void	startSerialization() = 0;
+	virtual void startSerialization() = 0;
 
-	virtual	void	finishSerialization() = 0;
+	virtual void finishSerialization() = 0;
 
-	virtual	const char*	findNameForPointer(const void* ptr) const = 0;
+	virtual const char* findNameForPointer(const void* ptr) const = 0;
 
-	virtual	void	registerNameForPointer(const void* ptr, const char* name) = 0;
+	virtual void registerNameForPointer(const void* ptr, const char* name) = 0;
 
-	virtual void	serializeName(const char* ptr) = 0;
+	virtual void serializeName(const char* ptr) = 0;
 
-	virtual int		getSerializationFlags() const = 0;
+	virtual int getSerializationFlags() const = 0;
 
-	virtual void	setSerializationFlags(int flags) = 0;
+	virtual void setSerializationFlags(int flags) = 0;
 
 	virtual int getNumChunks() const = 0;
 
 	virtual const btChunk* getChunk(int chunkIndex) const = 0;
-
 };
 
-
-
 #define BT_HEADER_LENGTH 12
-#if defined(__sgi) || defined (__sparc) || defined (__sparc__) || defined (__PPC__) || defined (__ppc__) || defined (__BIG_ENDIAN__)
-#	define BT_MAKE_ID(a,b,c,d) ( (int)(a)<<24 | (int)(b)<<16 | (c)<<8 | (d) )
+#if defined(__sgi) || defined(__sparc) || defined(__sparc__) || defined(__PPC__) || defined(__ppc__) || defined(__BIG_ENDIAN__)
+#define BT_MAKE_ID(a, b, c, d) ((int)(a) << 24 | (int)(b) << 16 | (c) << 8 | (d))
 #else
-#	define BT_MAKE_ID(a,b,c,d) ( (int)(d)<<24 | (int)(c)<<16 | (b)<<8 | (a) )
+#define BT_MAKE_ID(a, b, c, d) ((int)(d) << 24 | (int)(c) << 16 | (b) << 8 | (a))
 #endif
 
+#define BT_MULTIBODY_CODE BT_MAKE_ID('M', 'B', 'D', 'Y')
+#define BT_MB_LINKCOLLIDER_CODE BT_MAKE_ID('M', 'B', 'L', 'C')
+#define BT_SOFTBODY_CODE BT_MAKE_ID('S', 'B', 'D', 'Y')
+#define BT_COLLISIONOBJECT_CODE BT_MAKE_ID('C', 'O', 'B', 'J')
+#define BT_RIGIDBODY_CODE BT_MAKE_ID('R', 'B', 'D', 'Y')
+#define BT_CONSTRAINT_CODE BT_MAKE_ID('C', 'O', 'N', 'S')
+#define BT_BOXSHAPE_CODE BT_MAKE_ID('B', 'O', 'X', 'S')
+#define BT_QUANTIZED_BVH_CODE BT_MAKE_ID('Q', 'B', 'V', 'H')
+#define BT_TRIANLGE_INFO_MAP BT_MAKE_ID('T', 'M', 'A', 'P')
+#define BT_SHAPE_CODE BT_MAKE_ID('S', 'H', 'A', 'P')
+#define BT_ARRAY_CODE BT_MAKE_ID('A', 'R', 'A', 'Y')
+#define BT_SBMATERIAL_CODE BT_MAKE_ID('S', 'B', 'M', 'T')
+#define BT_SBNODE_CODE BT_MAKE_ID('S', 'B', 'N', 'D')
+#define BT_DYNAMICSWORLD_CODE BT_MAKE_ID('D', 'W', 'L', 'D')
+#define BT_CONTACTMANIFOLD_CODE BT_MAKE_ID('C', 'O', 'N', 'T')
+#define BT_DNA_CODE BT_MAKE_ID('D', 'N', 'A', '1')
 
-#define BT_MULTIBODY_CODE       BT_MAKE_ID('M','B','D','Y')
-#define BT_MB_LINKCOLLIDER_CODE BT_MAKE_ID('M','B','L','C')
-#define BT_SOFTBODY_CODE		BT_MAKE_ID('S','B','D','Y')
-#define BT_COLLISIONOBJECT_CODE BT_MAKE_ID('C','O','B','J')
-#define BT_RIGIDBODY_CODE		BT_MAKE_ID('R','B','D','Y')
-#define BT_CONSTRAINT_CODE		BT_MAKE_ID('C','O','N','S')
-#define BT_BOXSHAPE_CODE		BT_MAKE_ID('B','O','X','S')
-#define BT_QUANTIZED_BVH_CODE	BT_MAKE_ID('Q','B','V','H')
-#define BT_TRIANLGE_INFO_MAP	BT_MAKE_ID('T','M','A','P')
-#define BT_SHAPE_CODE			BT_MAKE_ID('S','H','A','P')
-#define BT_ARRAY_CODE			BT_MAKE_ID('A','R','A','Y')
-#define BT_SBMATERIAL_CODE		BT_MAKE_ID('S','B','M','T')
-#define BT_SBNODE_CODE			BT_MAKE_ID('S','B','N','D')
-#define BT_DYNAMICSWORLD_CODE	BT_MAKE_ID('D','W','L','D')
-#define BT_CONTACTMANIFOLD_CODE BT_MAKE_ID('C','O','N','T')
-#define BT_DNA_CODE				BT_MAKE_ID('D','N','A','1')
-
-struct	btPointerUid
+struct btPointerUid
 {
-	union
-	{
-		void*	m_ptr;
-		int		m_uniqueIds[2];
+	union {
+		void* m_ptr;
+		int m_uniqueIds[2];
 	};
 };
 
@@ -146,8 +135,8 @@ struct btBulletSerializedArrays
 	btBulletSerializedArrays()
 	{
 	}
-	btAlignedObjectArray<struct btQuantizedBvhDoubleData*>	m_bvhsDouble;
-	btAlignedObjectArray<struct btQuantizedBvhFloatData*>	m_bvhsFloat;
+	btAlignedObjectArray<struct btQuantizedBvhDoubleData*> m_bvhsDouble;
+	btAlignedObjectArray<struct btQuantizedBvhFloatData*> m_bvhsFloat;
 	btAlignedObjectArray<struct btCollisionShapeData*> m_colShapeData;
 	btAlignedObjectArray<struct btDynamicsWorldDoubleData*> m_dynamicWorldInfoDataDouble;
 	btAlignedObjectArray<struct btDynamicsWorldFloatData*> m_dynamicWorldInfoDataFloat;
@@ -157,51 +146,42 @@ struct btBulletSerializedArrays
 	btAlignedObjectArray<struct btCollisionObjectFloatData*> m_collisionObjectDataFloat;
 	btAlignedObjectArray<struct btTypedConstraintFloatData*> m_constraintDataFloat;
 	btAlignedObjectArray<struct btTypedConstraintDoubleData*> m_constraintDataDouble;
-	btAlignedObjectArray<struct btTypedConstraintData*> m_constraintData;//for backwards compatibility
+	btAlignedObjectArray<struct btTypedConstraintData*> m_constraintData;  //for backwards compatibility
 	btAlignedObjectArray<struct btSoftBodyFloatData*> m_softBodyFloatData;
 	btAlignedObjectArray<struct btSoftBodyDoubleData*> m_softBodyDoubleData;
-
 };
 
-
 ///The btDefaultSerializer is the main Bullet serialization class.
 ///The constructor takes an optional argument for backwards compatibility, it is recommended to leave this empty/zero.
-class btDefaultSerializer	:	public btSerializer
+class btDefaultSerializer : public btSerializer
 {
+protected:
+	btAlignedObjectArray<char*> mTypes;
+	btAlignedObjectArray<short*> mStructs;
+	btAlignedObjectArray<short> mTlens;
+	btHashMap<btHashInt, int> mStructReverse;
+	btHashMap<btHashString, int> mTypeLookup;
+
+	btHashMap<btHashPtr, void*> m_chunkP;
+
+	btHashMap<btHashPtr, const char*> m_nameMap;
+
+	btHashMap<btHashPtr, btPointerUid> m_uniquePointers;
+	int m_uniqueIdGenerator;
+
+	int m_totalSize;
+	unsigned char* m_buffer;
+	bool m_ownsBuffer;
+	int m_currentSize;
+	void* m_dna;
+	int m_dnaLength;
+
+	int m_serializationFlags;
+
+	btAlignedObjectArray<btChunk*> m_chunkPtrs;
 
 protected:
-
-	btAlignedObjectArray<char*>			mTypes;
-	btAlignedObjectArray<short*>			mStructs;
-	btAlignedObjectArray<short>			mTlens;
-	btHashMap<btHashInt, int>			mStructReverse;
-	btHashMap<btHashString,int>	mTypeLookup;
-	
-
-
-	btHashMap<btHashPtr,void*>	m_chunkP;
-
-	btHashMap<btHashPtr,const char*>	m_nameMap;
-
-	btHashMap<btHashPtr,btPointerUid>	m_uniquePointers;
-	int	m_uniqueIdGenerator;
-
-	int					m_totalSize;
-	unsigned char*		m_buffer;
-	bool                m_ownsBuffer;
-	int					m_currentSize;
-	void*				m_dna;
-	int					m_dnaLength;
-
-	int					m_serializationFlags;
-
-
-	btAlignedObjectArray<btChunk*>	m_chunkPtrs;
-
-protected:
-
-	
-	virtual	void*	findPointer(void* oldPtr)
+	virtual void* findPointer(void* oldPtr)
 	{
 		void** ptr = m_chunkP.find(oldPtr);
 		if (ptr && *ptr)
@@ -209,48 +189,43 @@ protected:
 		return 0;
 	}
 
+	virtual void writeDNA()
+	{
+		btChunk* dnaChunk = allocate(m_dnaLength, 1);
+		memcpy(dnaChunk->m_oldPtr, m_dna, m_dnaLength);
+		finalizeChunk(dnaChunk, "DNA1", BT_DNA_CODE, m_dna);
+	}
 
+	int getReverseType(const char* type) const
+	{
+		btHashString key(type);
+		const int* valuePtr = mTypeLookup.find(key);
+		if (valuePtr)
+			return *valuePtr;
 
+		return -1;
+	}
 
+	void initDNA(const char* bdnaOrg, int dnalen)
+	{
+		///was already initialized
+		if (m_dna)
+			return;
 
-		virtual void	writeDNA()
-		{
-			btChunk* dnaChunk = allocate(m_dnaLength,1);
-			memcpy(dnaChunk->m_oldPtr,m_dna,m_dnaLength);
-			finalizeChunk(dnaChunk,"DNA1",BT_DNA_CODE, m_dna);
-		}
+		int littleEndian = 1;
+		littleEndian = ((char*)&littleEndian)[0];
 
-		int getReverseType(const char *type) const
-		{
+		m_dna = btAlignedAlloc(dnalen, 16);
+		memcpy(m_dna, bdnaOrg, dnalen);
+		m_dnaLength = dnalen;
 
-			btHashString key(type);
-			const int* valuePtr = mTypeLookup.find(key);
-			if (valuePtr)
-				return *valuePtr;
+		int* intPtr = 0;
+		short* shtPtr = 0;
+		char* cp = 0;
+		int dataLen = 0;
+		intPtr = (int*)m_dna;
 
-			return -1;
-		}
-
-		void initDNA(const char* bdnaOrg,int dnalen)
-		{
-			///was already initialized
-			if (m_dna)
-				return;
-
-			int littleEndian= 1;
-			littleEndian= ((char*)&littleEndian)[0];
-
-
-			m_dna = btAlignedAlloc(dnalen,16);
-			memcpy(m_dna,bdnaOrg,dnalen);
-			m_dnaLength = dnalen;
-
-			int *intPtr=0;
-			short *shtPtr=0;
-			char *cp = 0;int dataLen =0;
-			intPtr = (int*)m_dna;
-
-			/*
+		/*
 				SDNA (4 bytes) (magic number)
 				NAME (4 bytes)
 				<nr> (4 bytes) amount of names (int)
@@ -258,81 +233,81 @@ protected:
 				<string>
 			*/
 
-			if (strncmp((const char*)m_dna, "SDNA", 4)==0)
-			{
-				// skip ++ NAME
-				intPtr++; intPtr++;
-			}
-
-			// Parse names
-			if (!littleEndian)
-				*intPtr = btSwapEndian(*intPtr);
-
-			dataLen = *intPtr;
-
+		if (strncmp((const char*)m_dna, "SDNA", 4) == 0)
+		{
+			// skip ++ NAME
 			intPtr++;
+			intPtr++;
+		}
 
-			cp = (char*)intPtr;
-			int i;
-			for ( i=0; i<dataLen; i++)
-			{
+		// Parse names
+		if (!littleEndian)
+			*intPtr = btSwapEndian(*intPtr);
 
-				while (*cp)cp++;
-				cp++;
-			}
-			cp = btAlignPointer(cp,4);
+		dataLen = *intPtr;
 
-			/*
+		intPtr++;
+
+		cp = (char*)intPtr;
+		int i;
+		for (i = 0; i < dataLen; i++)
+		{
+			while (*cp) cp++;
+			cp++;
+		}
+		cp = btAlignPointer(cp, 4);
+
+		/*
 				TYPE (4 bytes)
 				<nr> amount of types (int)
 				<string>
 				<string>
 			*/
 
-			intPtr = (int*)cp;
-			btAssert(strncmp(cp, "TYPE", 4)==0); intPtr++;
+		intPtr = (int*)cp;
+		btAssert(strncmp(cp, "TYPE", 4) == 0);
+		intPtr++;
 
-			if (!littleEndian)
-				*intPtr =  btSwapEndian(*intPtr);
+		if (!littleEndian)
+			*intPtr = btSwapEndian(*intPtr);
 
-			dataLen = *intPtr;
-			intPtr++;
+		dataLen = *intPtr;
+		intPtr++;
 
+		cp = (char*)intPtr;
+		for (i = 0; i < dataLen; i++)
+		{
+			mTypes.push_back(cp);
+			while (*cp) cp++;
+			cp++;
+		}
 
-			cp = (char*)intPtr;
-			for (i=0; i<dataLen; i++)
-			{
-				mTypes.push_back(cp);
-				while (*cp)cp++;
-				cp++;
-			}
+		cp = btAlignPointer(cp, 4);
 
-			cp = btAlignPointer(cp,4);
-
-
-			/*
+		/*
 				TLEN (4 bytes)
 				<len> (short) the lengths of types
 				<len>
 			*/
 
-			// Parse type lens
-			intPtr = (int*)cp;
-			btAssert(strncmp(cp, "TLEN", 4)==0); intPtr++;
+		// Parse type lens
+		intPtr = (int*)cp;
+		btAssert(strncmp(cp, "TLEN", 4) == 0);
+		intPtr++;
 
-			dataLen = (int)mTypes.size();
+		dataLen = (int)mTypes.size();
 
-			shtPtr = (short*)intPtr;
-			for (i=0; i<dataLen; i++, shtPtr++)
-			{
-				if (!littleEndian)
-					shtPtr[0] = btSwapEndian(shtPtr[0]);
-				mTlens.push_back(shtPtr[0]);
-			}
+		shtPtr = (short*)intPtr;
+		for (i = 0; i < dataLen; i++, shtPtr++)
+		{
+			if (!littleEndian)
+				shtPtr[0] = btSwapEndian(shtPtr[0]);
+			mTlens.push_back(shtPtr[0]);
+		}
 
-			if (dataLen & 1) shtPtr++;
+		if (dataLen & 1) shtPtr++;
 
-			/*
+		/*
 				STRC (4 bytes)
 				<nr> amount of structs (int)
 				<typenr>
@@ -343,384 +318,372 @@ protected:
 				<namenr>
 			*/
 
-			intPtr = (int*)shtPtr;
-			cp = (char*)intPtr;
-			btAssert(strncmp(cp, "STRC", 4)==0); intPtr++;
+		intPtr = (int*)shtPtr;
+		cp = (char*)intPtr;
+		btAssert(strncmp(cp, "STRC", 4) == 0);
+		intPtr++;
+
+		if (!littleEndian)
+			*intPtr = btSwapEndian(*intPtr);
+		dataLen = *intPtr;
+		intPtr++;
+
+		shtPtr = (short*)intPtr;
+		for (i = 0; i < dataLen; i++)
+		{
+			mStructs.push_back(shtPtr);
 
 			if (!littleEndian)
-				*intPtr = btSwapEndian(*intPtr);
-			dataLen = *intPtr ;
-			intPtr++;
-
-
-			shtPtr = (short*)intPtr;
-			for (i=0; i<dataLen; i++)
 			{
-				mStructs.push_back (shtPtr);
+				shtPtr[0] = btSwapEndian(shtPtr[0]);
+				shtPtr[1] = btSwapEndian(shtPtr[1]);
 
-				if (!littleEndian)
+				int len = shtPtr[1];
+				shtPtr += 2;
+
+				for (int a = 0; a < len; a++, shtPtr += 2)
 				{
-					shtPtr[0]= btSwapEndian(shtPtr[0]);
-					shtPtr[1]= btSwapEndian(shtPtr[1]);
-
-					int len = shtPtr[1];
-					shtPtr+= 2;
-
-					for (int a=0; a<len; a++, shtPtr+=2)
-					{
-							shtPtr[0]= btSwapEndian(shtPtr[0]);
-							shtPtr[1]= btSwapEndian(shtPtr[1]);
-					}
-
-				} else
-				{
-					shtPtr+= (2*shtPtr[1])+2;
+					shtPtr[0] = btSwapEndian(shtPtr[0]);
+					shtPtr[1] = btSwapEndian(shtPtr[1]);
 				}
 			}
-
-			// build reverse lookups
-			for (i=0; i<(int)mStructs.size(); i++)
+			else
 			{
-				short *strc = mStructs.at(i);
-				mStructReverse.insert(strc[0], i);
-				mTypeLookup.insert(btHashString(mTypes[strc[0]]),i);
+				shtPtr += (2 * shtPtr[1]) + 2;
 			}
 		}
 
+		// build reverse lookups
+		for (i = 0; i < (int)mStructs.size(); i++)
+		{
+			short* strc = mStructs.at(i);
+			mStructReverse.insert(strc[0], i);
+			mTypeLookup.insert(btHashString(mTypes[strc[0]]), i);
+		}
+	}
+
 public:
+	btHashMap<btHashPtr, void*> m_skipPointers;
 
-	btHashMap<btHashPtr,void*> m_skipPointers;
-
-
-		btDefaultSerializer(int totalSize=0, unsigned char*	buffer=0)
-			:m_uniqueIdGenerator(0),
-			m_totalSize(totalSize),
-			m_currentSize(0),
-			m_dna(0),
-			m_dnaLength(0),
-			m_serializationFlags(0)
+	btDefaultSerializer(int totalSize = 0, unsigned char* buffer = 0)
+		: m_uniqueIdGenerator(0),
+		  m_totalSize(totalSize),
+		  m_currentSize(0),
+		  m_dna(0),
+		  m_dnaLength(0),
+		  m_serializationFlags(0)
+	{
+		if (buffer == 0)
 		{
-		    if (buffer==0)
-            {
-                m_buffer = m_totalSize?(unsigned char*)btAlignedAlloc(totalSize,16):0;
-                m_ownsBuffer = true;
-            } else
-            {
-                m_buffer = buffer;
-                m_ownsBuffer = false;
-            }
+			m_buffer = m_totalSize ? (unsigned char*)btAlignedAlloc(totalSize, 16) : 0;
+			m_ownsBuffer = true;
+		}
+		else
+		{
+			m_buffer = buffer;
+			m_ownsBuffer = false;
+		}
 
-			const bool VOID_IS_8 = ((sizeof(void*)==8));
+		const bool VOID_IS_8 = ((sizeof(void*) == 8));
 
 #ifdef BT_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
-			if (VOID_IS_8)
-			{
+		if (VOID_IS_8)
+		{
 #if _WIN64
-				initDNA((const char*)sBulletDNAstr64,sBulletDNAlen64);
+			initDNA((const char*)sBulletDNAstr64, sBulletDNAlen64);
 #else
-				btAssert(0);
+			btAssert(0);
 #endif
-			} else
-			{
+		}
+		else
+		{
 #ifndef _WIN64
-				initDNA((const char*)sBulletDNAstr,sBulletDNAlen);
+			initDNA((const char*)sBulletDNAstr, sBulletDNAlen);
 #else
-				btAssert(0);
+			btAssert(0);
 #endif
-			}
-
-#else //BT_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
-			if (VOID_IS_8)
-			{
-				initDNA((const char*)sBulletDNAstr64,sBulletDNAlen64);
-			} else
-			{
-				initDNA((const char*)sBulletDNAstr,sBulletDNAlen);
-			}
-#endif //BT_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
-
 		}
 
-		virtual ~btDefaultSerializer()
+#else   //BT_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
+		if (VOID_IS_8)
 		{
-			if (m_buffer && m_ownsBuffer)
-				btAlignedFree(m_buffer);
-			if (m_dna)
-				btAlignedFree(m_dna);
+			initDNA((const char*)sBulletDNAstr64, sBulletDNAlen64);
 		}
-
-		static int getMemoryDnaSizeInBytes()
+		else
 		{
-			const bool VOID_IS_8 = ((sizeof(void*) == 8));
-
-			if (VOID_IS_8)
-			{
-				return sBulletDNAlen64;
-			}
-			return sBulletDNAlen;
+			initDNA((const char*)sBulletDNAstr, sBulletDNAlen);
 		}
-		static const char* getMemoryDna()
+#endif  //BT_INTERNAL_UPDATE_SERIALIZATION_STRUCTURES
+	}
+
+	virtual ~btDefaultSerializer()
+	{
+		if (m_buffer && m_ownsBuffer)
+			btAlignedFree(m_buffer);
+		if (m_dna)
+			btAlignedFree(m_dna);
+	}
+
+	static int getMemoryDnaSizeInBytes()
+	{
+		const bool VOID_IS_8 = ((sizeof(void*) == 8));
+
+		if (VOID_IS_8)
 		{
-			const bool VOID_IS_8 = ((sizeof(void*) == 8));
-			if (VOID_IS_8)
-			{
-				return (const char*)sBulletDNAstr64;
-			}
-			return (const char*)sBulletDNAstr;
+			return sBulletDNAlen64;
 		}
-
-		void	insertHeader()
+		return sBulletDNAlen;
+	}
+	static const char* getMemoryDna()
+	{
+		const bool VOID_IS_8 = ((sizeof(void*) == 8));
+		if (VOID_IS_8)
 		{
-			writeHeader(m_buffer);
-			m_currentSize += BT_HEADER_LENGTH;
+			return (const char*)sBulletDNAstr64;
 		}
+		return (const char*)sBulletDNAstr;
+	}
 
-		void	writeHeader(unsigned char* buffer) const
-		{
+	void insertHeader()
+	{
+		writeHeader(m_buffer);
+		m_currentSize += BT_HEADER_LENGTH;
+	}
 
-
-#ifdef  BT_USE_DOUBLE_PRECISION
-			memcpy(buffer, "BULLETd", 7);
+	void writeHeader(unsigned char* buffer) const
+	{
+#ifdef BT_USE_DOUBLE_PRECISION
+		memcpy(buffer, "BULLETd", 7);
 #else
-			memcpy(buffer, "BULLETf", 7);
-#endif //BT_USE_DOUBLE_PRECISION
+		memcpy(buffer, "BULLETf", 7);
+#endif  //BT_USE_DOUBLE_PRECISION
 
-			int littleEndian= 1;
-			littleEndian= ((char*)&littleEndian)[0];
+		int littleEndian = 1;
+		littleEndian = ((char*)&littleEndian)[0];
 
-			if (sizeof(void*)==8)
+		if (sizeof(void*) == 8)
+		{
+			buffer[7] = '-';
+		}
+		else
+		{
+			buffer[7] = '_';
+		}
+
+		if (littleEndian)
+		{
+			buffer[8] = 'v';
+		}
+		else
+		{
+			buffer[8] = 'V';
+		}
+
+		buffer[9] = '2';
+		buffer[10] = '8';
+		buffer[11] = '8';
+	}
+
+	virtual void startSerialization()
+	{
+		m_uniqueIdGenerator = 1;
+		if (m_totalSize)
+		{
+			unsigned char* buffer = internalAlloc(BT_HEADER_LENGTH);
+			writeHeader(buffer);
+		}
+	}
+
+	virtual void finishSerialization()
+	{
+		writeDNA();
+
+		//if we didn't pre-allocate a buffer, we need to create a contiguous buffer now
+		int mysize = 0;
+		if (!m_totalSize)
+		{
+			if (m_buffer)
+				btAlignedFree(m_buffer);
+
+			m_currentSize += BT_HEADER_LENGTH;
+			m_buffer = (unsigned char*)btAlignedAlloc(m_currentSize, 16);
+
+			unsigned char* currentPtr = m_buffer;
+			writeHeader(m_buffer);
+			currentPtr += BT_HEADER_LENGTH;
+			mysize += BT_HEADER_LENGTH;
+			for (int i = 0; i < m_chunkPtrs.size(); i++)
 			{
-				buffer[7] = '-';
-			} else
-			{
-				buffer[7] = '_';
+				int curLength = sizeof(btChunk) + m_chunkPtrs[i]->m_length;
+				memcpy(currentPtr, m_chunkPtrs[i], curLength);
+				btAlignedFree(m_chunkPtrs[i]);
+				currentPtr += curLength;
+				mysize += curLength;
 			}
-
-			if (littleEndian)
-			{
-				buffer[8]='v';
-			} else
-			{
-				buffer[8]='V';
-			}
-
-
-			buffer[9] = '2';
-			buffer[10] = '8';
-			buffer[11] = '8';
-
 		}
 
-		virtual	void	startSerialization()
-		{
-			m_uniqueIdGenerator= 1;
-			if (m_totalSize)
-			{
-				unsigned char* buffer = internalAlloc(BT_HEADER_LENGTH);
-				writeHeader(buffer);
-			}
+		mTypes.clear();
+		mStructs.clear();
+		mTlens.clear();
+		mStructReverse.clear();
+		mTypeLookup.clear();
+		m_skipPointers.clear();
+		m_chunkP.clear();
+		m_nameMap.clear();
+		m_uniquePointers.clear();
+		m_chunkPtrs.clear();
+	}
 
-		}
-
-		virtual	void	finishSerialization()
-		{
-			writeDNA();
-
-			//if we didn't pre-allocate a buffer, we need to create a contiguous buffer now
-			int mysize = 0;
-			if (!m_totalSize)
-			{
-				if (m_buffer)
-					btAlignedFree(m_buffer);
-
-				m_currentSize += BT_HEADER_LENGTH;
-				m_buffer = (unsigned char*)btAlignedAlloc(m_currentSize,16);
-
-				unsigned char* currentPtr = m_buffer;
-				writeHeader(m_buffer);
-				currentPtr += BT_HEADER_LENGTH;
-				mysize+=BT_HEADER_LENGTH;
-				for (int i=0;i<	m_chunkPtrs.size();i++)
-				{
-					int curLength = sizeof(btChunk)+m_chunkPtrs[i]->m_length;
-					memcpy(currentPtr,m_chunkPtrs[i], curLength);
-					btAlignedFree(m_chunkPtrs[i]);
-					currentPtr+=curLength;
-					mysize+=curLength;
-				}
-			}
-
-			mTypes.clear();
-			mStructs.clear();
-			mTlens.clear();
-			mStructReverse.clear();
-			mTypeLookup.clear();
-			m_skipPointers.clear();
-			m_chunkP.clear();
-			m_nameMap.clear();
-			m_uniquePointers.clear();
-			m_chunkPtrs.clear();
-		}
-
-		virtual	void*	getUniquePointer(void*oldPtr)
-		{
-			btAssert(m_uniqueIdGenerator >= 0);
-			if (!oldPtr)
-				return 0;
-
-			btPointerUid* uptr = (btPointerUid*)m_uniquePointers.find(oldPtr);
-			if (uptr)
-			{
-				return uptr->m_ptr;
-			}
-
-			void** ptr2 = m_skipPointers[oldPtr];
-            if (ptr2)
-			{
-				return 0;
-			}
-
-			m_uniqueIdGenerator++;
-
-			btPointerUid uid;
-			uid.m_uniqueIds[0] = m_uniqueIdGenerator;
-			uid.m_uniqueIds[1] = m_uniqueIdGenerator;
-			m_uniquePointers.insert(oldPtr,uid);
-			return uid.m_ptr;
-
-		}
-
-		virtual	const unsigned char*		getBufferPointer() const
-		{
-			return m_buffer;
-		}
-
-		virtual	int					getCurrentBufferSize() const
-		{
-			return	m_currentSize;
-		}
-
-		virtual	void	finalizeChunk(btChunk* chunk, const char* structType, int chunkCode,void* oldPtr)
-		{
-			if (!(m_serializationFlags&BT_SERIALIZE_NO_DUPLICATE_ASSERT))
-			{
-				btAssert(!findPointer(oldPtr));
-			}
-
-			chunk->m_dna_nr = getReverseType(structType);
-
-			chunk->m_chunkCode = chunkCode;
-
-			void* uniquePtr = getUniquePointer(oldPtr);
-
-			m_chunkP.insert(oldPtr,uniquePtr);//chunk->m_oldPtr);
-			chunk->m_oldPtr = uniquePtr;//oldPtr;
-
-		}
-
-
-		virtual unsigned char* internalAlloc(size_t size)
-		{
-			unsigned char* ptr = 0;
-
-			if (m_totalSize)
-			{
-				ptr = m_buffer+m_currentSize;
-				m_currentSize += int(size);
-				btAssert(m_currentSize<m_totalSize);
-			} else
-			{
-				ptr = (unsigned char*)btAlignedAlloc(size,16);
-				m_currentSize += int(size);
-			}
-			return ptr;
-		}
-
-
-
-		virtual	btChunk*	allocate(size_t size, int numElements)
-		{
-
-			unsigned char* ptr = internalAlloc(int(size)*numElements+sizeof(btChunk));
-
-			unsigned char* data = ptr + sizeof(btChunk);
-
-			btChunk* chunk = (btChunk*)ptr;
-			chunk->m_chunkCode = 0;
-			chunk->m_oldPtr = data;
-			chunk->m_length = int(size)*numElements;
-			chunk->m_number = numElements;
-
-			m_chunkPtrs.push_back(chunk);
-
-
-			return chunk;
-		}
-
-		virtual	const char*	findNameForPointer(const void* ptr) const
-		{
-			const char*const * namePtr = m_nameMap.find(ptr);
-			if (namePtr && *namePtr)
-				return *namePtr;
+	virtual void* getUniquePointer(void* oldPtr)
+	{
+		btAssert(m_uniqueIdGenerator >= 0);
+		if (!oldPtr)
 			return 0;
 
+		btPointerUid* uptr = (btPointerUid*)m_uniquePointers.find(oldPtr);
+		if (uptr)
+		{
+			return uptr->m_ptr;
 		}
 
-		virtual	void	registerNameForPointer(const void* ptr, const char* name)
+		void** ptr2 = m_skipPointers[oldPtr];
+		if (ptr2)
 		{
-			m_nameMap.insert(ptr,name);
+			return 0;
 		}
 
-		virtual void	serializeName(const char* name)
+		m_uniqueIdGenerator++;
+
+		btPointerUid uid;
+		uid.m_uniqueIds[0] = m_uniqueIdGenerator;
+		uid.m_uniqueIds[1] = m_uniqueIdGenerator;
+		m_uniquePointers.insert(oldPtr, uid);
+		return uid.m_ptr;
+	}
+
+	virtual const unsigned char* getBufferPointer() const
+	{
+		return m_buffer;
+	}
+
+	virtual int getCurrentBufferSize() const
+	{
+		return m_currentSize;
+	}
+
+	virtual void finalizeChunk(btChunk* chunk, const char* structType, int chunkCode, void* oldPtr)
+	{
+		if (!(m_serializationFlags & BT_SERIALIZE_NO_DUPLICATE_ASSERT))
 		{
-			if (name)
+			btAssert(!findPointer(oldPtr));
+		}
+
+		chunk->m_dna_nr = getReverseType(structType);
+
+		chunk->m_chunkCode = chunkCode;
+
+		void* uniquePtr = getUniquePointer(oldPtr);
+
+		m_chunkP.insert(oldPtr, uniquePtr);  //chunk->m_oldPtr);
+		chunk->m_oldPtr = uniquePtr;         //oldPtr;
+	}
+
+	virtual unsigned char* internalAlloc(size_t size)
+	{
+		unsigned char* ptr = 0;
+
+		if (m_totalSize)
+		{
+			ptr = m_buffer + m_currentSize;
+			m_currentSize += int(size);
+			btAssert(m_currentSize < m_totalSize);
+		}
+		else
+		{
+			ptr = (unsigned char*)btAlignedAlloc(size, 16);
+			m_currentSize += int(size);
+		}
+		return ptr;
+	}
+
+	virtual btChunk* allocate(size_t size, int numElements)
+	{
+		unsigned char* ptr = internalAlloc(int(size) * numElements + sizeof(btChunk));
+
+		unsigned char* data = ptr + sizeof(btChunk);
+
+		btChunk* chunk = (btChunk*)ptr;
+		chunk->m_chunkCode = 0;
+		chunk->m_oldPtr = data;
+		chunk->m_length = int(size) * numElements;
+		chunk->m_number = numElements;
+
+		m_chunkPtrs.push_back(chunk);
+
+		return chunk;
+	}
+
+	virtual const char* findNameForPointer(const void* ptr) const
+	{
+		const char* const* namePtr = m_nameMap.find(ptr);
+		if (namePtr && *namePtr)
+			return *namePtr;
+		return 0;
+	}
+
+	virtual void registerNameForPointer(const void* ptr, const char* name)
+	{
+		m_nameMap.insert(ptr, name);
+	}
+
+	virtual void serializeName(const char* name)
+	{
+		if (name)
+		{
+			//don't serialize name twice
+			if (findPointer((void*)name))
+				return;
+
+			int len = btStrLen(name);
+			if (len)
 			{
-				//don't serialize name twice
-				if (findPointer((void*)name))
-					return;
+				int newLen = len + 1;
+				int padding = ((newLen + 3) & ~3) - newLen;
+				newLen += padding;
 
-				int len = btStrLen(name);
-				if (len)
+				//serialize name string now
+				btChunk* chunk = allocate(sizeof(char), newLen);
+				char* destinationName = (char*)chunk->m_oldPtr;
+				for (int i = 0; i < len; i++)
 				{
-
-					int newLen = len+1;
-					int padding = ((newLen+3)&~3)-newLen;
-					newLen += padding;
-
-					//serialize name string now
-					btChunk* chunk = allocate(sizeof(char),newLen);
-					char* destinationName = (char*)chunk->m_oldPtr;
-					for (int i=0;i<len;i++)
-					{
-						destinationName[i] = name[i];
-					}
-					destinationName[len] = 0;
-					finalizeChunk(chunk,"char",BT_ARRAY_CODE,(void*)name);
+					destinationName[i] = name[i];
 				}
+				destinationName[len] = 0;
+				finalizeChunk(chunk, "char", BT_ARRAY_CODE, (void*)name);
 			}
 		}
+	}
 
-		virtual int		getSerializationFlags() const
-		{
-			return m_serializationFlags;
-		}
+	virtual int getSerializationFlags() const
+	{
+		return m_serializationFlags;
+	}
 
-		virtual void	setSerializationFlags(int flags)
-		{
-			m_serializationFlags = flags;
-		}
-		int getNumChunks() const
-		{
-			return m_chunkPtrs.size();
-		}
+	virtual void setSerializationFlags(int flags)
+	{
+		m_serializationFlags = flags;
+	}
+	int getNumChunks() const
+	{
+		return m_chunkPtrs.size();
+	}
 
-		const btChunk* getChunk(int chunkIndex) const
-		{
-			return m_chunkPtrs[chunkIndex];
-		}
+	const btChunk* getChunk(int chunkIndex) const
+	{
+		return m_chunkPtrs[chunkIndex];
+	}
 };
 
-
 ///In general it is best to use btDefaultSerializer,
 ///in particular when writing the data to disk or sending it over the network.
 ///The btInMemorySerializer is experimental and only suitable in a few cases.
@@ -730,181 +693,177 @@ public:
 
 struct btInMemorySerializer : public btDefaultSerializer
 {
-    btHashMap<btHashPtr,btChunk*> m_uid2ChunkPtr;
-    btHashMap<btHashPtr,void*> m_orgPtr2UniqueDataPtr;
-    btHashMap<btHashString,const void*> m_names2Ptr;
-    
+	btHashMap<btHashPtr, btChunk*> m_uid2ChunkPtr;
+	btHashMap<btHashPtr, void*> m_orgPtr2UniqueDataPtr;
+	btHashMap<btHashString, const void*> m_names2Ptr;
 
-    btBulletSerializedArrays    m_arrays;
+	btBulletSerializedArrays m_arrays;
 
-    btInMemorySerializer(int totalSize=0, unsigned char*	buffer=0)
-    :btDefaultSerializer(totalSize,buffer)
-    {
-        
-    }
+	btInMemorySerializer(int totalSize = 0, unsigned char* buffer = 0)
+		: btDefaultSerializer(totalSize, buffer)
+	{
+	}
 
-    virtual void startSerialization()
-    {
-        m_uid2ChunkPtr.clear();
-        //todo: m_arrays.clear();
-        btDefaultSerializer::startSerialization();
-    }
+	virtual void startSerialization()
+	{
+		m_uid2ChunkPtr.clear();
+		//todo: m_arrays.clear();
+		btDefaultSerializer::startSerialization();
+	}
 
-    
+	btChunk* findChunkFromUniquePointer(void* uniquePointer)
+	{
+		btChunk** chkPtr = m_uid2ChunkPtr[uniquePointer];
+		if (chkPtr)
+		{
+			return *chkPtr;
+		}
+		return 0;
+	}
 
-    btChunk* findChunkFromUniquePointer(void* uniquePointer)
-    {
-        btChunk** chkPtr = m_uid2ChunkPtr[uniquePointer];
-        if (chkPtr)
-        {
-            return *chkPtr;
-        }
-        return 0;
-    }
+	virtual void registerNameForPointer(const void* ptr, const char* name)
+	{
+		btDefaultSerializer::registerNameForPointer(ptr, name);
+		m_names2Ptr.insert(name, ptr);
+	}
 
-	virtual	void	registerNameForPointer(const void* ptr, const char* name)
-    {
-       btDefaultSerializer::registerNameForPointer(ptr,name);
-       m_names2Ptr.insert(name,ptr);
-    }
+	virtual void finishSerialization()
+	{
+	}
 
-    virtual void finishSerialization()
-    {
-    }
+	virtual void* getUniquePointer(void* oldPtr)
+	{
+		if (oldPtr == 0)
+			return 0;
 
-    virtual void* getUniquePointer(void*oldPtr)
-    {
-        if (oldPtr==0)
-            return 0;
-
-        // void* uniquePtr = getUniquePointer(oldPtr);
-        btChunk* chunk = findChunkFromUniquePointer(oldPtr);
-        if (chunk)
-        {
-            return chunk->m_oldPtr;
-        } else
-        {
-            const char* n = (const char*) oldPtr;
-            const void** ptr = m_names2Ptr[n];
-            if (ptr)
-            {
-                return oldPtr;
-            } else
-            {
-            		void** ptr2 = m_skipPointers[oldPtr];
-            		if (ptr2)
-								{
-									return 0;
-								} else
-								{
-									//If this assert hit, serialization happened in the wrong order
-									// 'getUniquePointer'
-									btAssert(0);
-								}
-
-            }
-            return 0;
-        }
-				return oldPtr;
-    }
-
-    virtual void finalizeChunk(btChunk* chunk, const char* structType, int chunkCode,void* oldPtr)
-    {
-        if (!(m_serializationFlags&BT_SERIALIZE_NO_DUPLICATE_ASSERT))
-        {
-            btAssert(!findPointer(oldPtr));
-        }
-
-        chunk->m_dna_nr = getReverseType(structType);
-        chunk->m_chunkCode = chunkCode;
-        //void* uniquePtr = getUniquePointer(oldPtr);
-        m_chunkP.insert(oldPtr,oldPtr);//chunk->m_oldPtr);
-        // chunk->m_oldPtr = uniquePtr;//oldPtr;
-
-        void* uid = findPointer(oldPtr);
-        m_uid2ChunkPtr.insert(uid,chunk);
-
-        switch (chunk->m_chunkCode)
+		// void* uniquePtr = getUniquePointer(oldPtr);
+		btChunk* chunk = findChunkFromUniquePointer(oldPtr);
+		if (chunk)
+		{
+			return chunk->m_oldPtr;
+		}
+		else
+		{
+			const char* n = (const char*)oldPtr;
+			const void** ptr = m_names2Ptr[n];
+			if (ptr)
 			{
+				return oldPtr;
+			}
+			else
+			{
+				void** ptr2 = m_skipPointers[oldPtr];
+				if (ptr2)
+				{
+					return 0;
+				}
+				else
+				{
+					//If this assert hit, serialization happened in the wrong order
+					// 'getUniquePointer'
+					btAssert(0);
+				}
+			}
+			return 0;
+		}
+		return oldPtr;
+	}
+
+	virtual void finalizeChunk(btChunk* chunk, const char* structType, int chunkCode, void* oldPtr)
+	{
+		if (!(m_serializationFlags & BT_SERIALIZE_NO_DUPLICATE_ASSERT))
+		{
+			btAssert(!findPointer(oldPtr));
+		}
+
+		chunk->m_dna_nr = getReverseType(structType);
+		chunk->m_chunkCode = chunkCode;
+		//void* uniquePtr = getUniquePointer(oldPtr);
+		m_chunkP.insert(oldPtr, oldPtr);  //chunk->m_oldPtr);
+		// chunk->m_oldPtr = uniquePtr;//oldPtr;
+
+		void* uid = findPointer(oldPtr);
+		m_uid2ChunkPtr.insert(uid, chunk);
+
+		switch (chunk->m_chunkCode)
+		{
 			case BT_SOFTBODY_CODE:
 			{
-	#ifdef BT_USE_DOUBLE_PRECISION
-					m_arrays.m_softBodyDoubleData.push_back((btSoftBodyDoubleData*) chunk->m_oldPtr);
-	#else
-					m_arrays.m_softBodyFloatData.push_back((btSoftBodyFloatData*) chunk->m_oldPtr);
-	#endif
-					break;
-				}
+#ifdef BT_USE_DOUBLE_PRECISION
+				m_arrays.m_softBodyDoubleData.push_back((btSoftBodyDoubleData*)chunk->m_oldPtr);
+#else
+				m_arrays.m_softBodyFloatData.push_back((btSoftBodyFloatData*)chunk->m_oldPtr);
+#endif
+				break;
+			}
 			case BT_COLLISIONOBJECT_CODE:
-				{
-	#ifdef BT_USE_DOUBLE_PRECISION
-					m_arrays.m_collisionObjectDataDouble.push_back((btCollisionObjectDoubleData*)chunk->m_oldPtr);
-	#else//BT_USE_DOUBLE_PRECISION
-					m_arrays.m_collisionObjectDataFloat.push_back((btCollisionObjectFloatData*)chunk->m_oldPtr);
-	#endif //BT_USE_DOUBLE_PRECISION
-					break;
-				}
+			{
+#ifdef BT_USE_DOUBLE_PRECISION
+				m_arrays.m_collisionObjectDataDouble.push_back((btCollisionObjectDoubleData*)chunk->m_oldPtr);
+#else   //BT_USE_DOUBLE_PRECISION
+				m_arrays.m_collisionObjectDataFloat.push_back((btCollisionObjectFloatData*)chunk->m_oldPtr);
+#endif  //BT_USE_DOUBLE_PRECISION
+				break;
+			}
 			case BT_RIGIDBODY_CODE:
-				{
-	#ifdef BT_USE_DOUBLE_PRECISION
-					m_arrays.m_rigidBodyDataDouble.push_back((btRigidBodyDoubleData*)chunk->m_oldPtr);
-	#else
-					m_arrays.m_rigidBodyDataFloat.push_back((btRigidBodyFloatData*)chunk->m_oldPtr);
-	#endif//BT_USE_DOUBLE_PRECISION
-					break;
-				};
+			{
+#ifdef BT_USE_DOUBLE_PRECISION
+				m_arrays.m_rigidBodyDataDouble.push_back((btRigidBodyDoubleData*)chunk->m_oldPtr);
+#else
+				m_arrays.m_rigidBodyDataFloat.push_back((btRigidBodyFloatData*)chunk->m_oldPtr);
+#endif  //BT_USE_DOUBLE_PRECISION
+				break;
+			};
 			case BT_CONSTRAINT_CODE:
-				{
-	#ifdef BT_USE_DOUBLE_PRECISION
-					m_arrays.m_constraintDataDouble.push_back((btTypedConstraintDoubleData*)chunk->m_oldPtr);
-	#else
-					m_arrays.m_constraintDataFloat.push_back((btTypedConstraintFloatData*)chunk->m_oldPtr);
-	#endif
-					break;
-				}
+			{
+#ifdef BT_USE_DOUBLE_PRECISION
+				m_arrays.m_constraintDataDouble.push_back((btTypedConstraintDoubleData*)chunk->m_oldPtr);
+#else
+				m_arrays.m_constraintDataFloat.push_back((btTypedConstraintFloatData*)chunk->m_oldPtr);
+#endif
+				break;
+			}
 			case BT_QUANTIZED_BVH_CODE:
-				{
-	#ifdef BT_USE_DOUBLE_PRECISION
-					m_arrays.m_bvhsDouble.push_back((btQuantizedBvhDoubleData*) chunk->m_oldPtr);
-	#else
-					m_arrays.m_bvhsFloat.push_back((btQuantizedBvhFloatData*) chunk->m_oldPtr);
-	#endif
-					break;
-				}
+			{
+#ifdef BT_USE_DOUBLE_PRECISION
+				m_arrays.m_bvhsDouble.push_back((btQuantizedBvhDoubleData*)chunk->m_oldPtr);
+#else
+				m_arrays.m_bvhsFloat.push_back((btQuantizedBvhFloatData*)chunk->m_oldPtr);
+#endif
+				break;
+			}
 
 			case BT_SHAPE_CODE:
-				{
-					btCollisionShapeData* shapeData = (btCollisionShapeData*) chunk->m_oldPtr;
-					m_arrays.m_colShapeData.push_back(shapeData);
-					break;
-				}
+			{
+				btCollisionShapeData* shapeData = (btCollisionShapeData*)chunk->m_oldPtr;
+				m_arrays.m_colShapeData.push_back(shapeData);
+				break;
+			}
 			case BT_TRIANLGE_INFO_MAP:
 			case BT_ARRAY_CODE:
 			case BT_SBMATERIAL_CODE:
 			case BT_SBNODE_CODE:
 			case BT_DYNAMICSWORLD_CODE:
 			case BT_DNA_CODE:
-				{
-					break;
-				}
+			{
+				break;
+			}
 			default:
-				{
-				}
-			};
-    }
+			{
+			}
+		};
+	}
 
-    int getNumChunks() const
-    {
-        return m_uid2ChunkPtr.size();
-    }
-
-    const btChunk* getChunk(int chunkIndex) const
-    {
-        return *m_uid2ChunkPtr.getAtIndex(chunkIndex);
-    }
+	int getNumChunks() const
+	{
+		return m_uid2ChunkPtr.size();
+	}
 
+	const btChunk* getChunk(int chunkIndex) const
+	{
+		return *m_uid2ChunkPtr.getAtIndex(chunkIndex);
+	}
 };
-#endif //ENABLE_INMEMORY_SERIALIZER
-
-#endif //BT_SERIALIZER_H
+#endif  //ENABLE_INMEMORY_SERIALIZER
 
+#endif  //BT_SERIALIZER_H
diff --git a/thirdparty/bullet/LinearMath/btSerializer64.cpp b/thirdparty/bullet/LinearMath/btSerializer64.cpp
index 0aa5cbf30e0..cf281cdb362 100644
--- a/thirdparty/bullet/LinearMath/btSerializer64.cpp
+++ b/thirdparty/bullet/LinearMath/btSerializer64.cpp
@@ -1,3 +1,4 @@
+// clang-format off
 char sBulletDNAstr64[]= {
 char(83),char(68),char(78),char(65),char(78),char(65),char(77),char(69),char(-76),char(1),char(0),char(0),char(109),char(95),char(115),char(105),char(122),char(101),char(0),char(109),
 char(95),char(99),char(97),char(112),char(97),char(99),char(105),char(116),char(121),char(0),char(42),char(109),char(95),char(100),char(97),char(116),char(97),char(0),char(109),char(95),
@@ -687,3 +688,4 @@ char(97),char(0),char(4),char(0),char(50),char(0),char(-79),char(1),char(96),cha
 char(98),char(0),char(4),char(0),char(48),char(0),char(-79),char(1),char(95),char(0),char(-78),char(1),char(4),char(0),char(-77),char(1),char(0),char(0),char(37),char(0),
 };
 int sBulletDNAlen64= sizeof(sBulletDNAstr64);
+// clang-format on
diff --git a/thirdparty/bullet/LinearMath/btSpatialAlgebra.h b/thirdparty/bullet/LinearMath/btSpatialAlgebra.h
index 8e59658bca7..6ad67a10812 100644
--- a/thirdparty/bullet/LinearMath/btSpatialAlgebra.h
+++ b/thirdparty/bullet/LinearMath/btSpatialAlgebra.h
@@ -12,18 +12,17 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-///These spatial algebra classes are used for btMultiBody, 
+///These spatial algebra classes are used for btMultiBody,
 ///see BulletDynamics/Featherstone
 
 #ifndef BT_SPATIAL_ALGEBRA_H
 #define BT_SPATIAL_ALGEBRA_H
 
-
 #include "btMatrix3x3.h"
 
 struct btSpatialForceVector
-{	
-	btVector3 m_topVec, m_bottomVec;	
+{
+	btVector3 m_topVec, m_bottomVec;
 	//
 	btSpatialForceVector() { setZero(); }
 	btSpatialForceVector(const btVector3 &angular, const btVector3 &linear) : m_topVec(linear), m_bottomVec(angular) {}
@@ -32,21 +31,34 @@ struct btSpatialForceVector
 		setValue(ax, ay, az, lx, ly, lz);
 	}
 	//
-	void setVector(const btVector3 &angular, const btVector3 &linear) { m_topVec = linear; m_bottomVec = angular; }
+	void setVector(const btVector3 &angular, const btVector3 &linear)
+	{
+		m_topVec = linear;
+		m_bottomVec = angular;
+	}
 	void setValue(const btScalar &ax, const btScalar &ay, const btScalar &az, const btScalar &lx, const btScalar &ly, const btScalar &lz)
 	{
-		m_bottomVec.setValue(ax, ay, az); m_topVec.setValue(lx, ly, lz);
+		m_bottomVec.setValue(ax, ay, az);
+		m_topVec.setValue(lx, ly, lz);
 	}
 	//
-	void addVector(const btVector3 &angular, const btVector3 &linear) { m_topVec += linear; m_bottomVec += angular; }
+	void addVector(const btVector3 &angular, const btVector3 &linear)
+	{
+		m_topVec += linear;
+		m_bottomVec += angular;
+	}
 	void addValue(const btScalar &ax, const btScalar &ay, const btScalar &az, const btScalar &lx, const btScalar &ly, const btScalar &lz)
 	{
-		m_bottomVec[0] += ax; m_bottomVec[1] += ay; m_bottomVec[2] += az;
-		m_topVec[0] += lx; m_topVec[1] += ly; m_topVec[2] += lz;			
+		m_bottomVec[0] += ax;
+		m_bottomVec[1] += ay;
+		m_bottomVec[2] += az;
+		m_topVec[0] += lx;
+		m_topVec[1] += ly;
+		m_topVec[2] += lz;
 	}
 	//
-	const btVector3 & getLinear()  const { return m_topVec; }
-	const btVector3 & getAngular() const { return m_bottomVec; }
+	const btVector3 &getLinear() const { return m_topVec; }
+	const btVector3 &getAngular() const { return m_bottomVec; }
 	//
 	void setLinear(const btVector3 &linear) { m_topVec = linear; }
 	void setAngular(const btVector3 &angular) { m_bottomVec = angular; }
@@ -54,14 +66,28 @@ struct btSpatialForceVector
 	void addAngular(const btVector3 &angular) { m_bottomVec += angular; }
 	void addLinear(const btVector3 &linear) { m_topVec += linear; }
 	//
-	void setZero() { m_topVec.setZero(); m_bottomVec.setZero(); }
+	void setZero()
+	{
+		m_topVec.setZero();
+		m_bottomVec.setZero();
+	}
 	//
-	btSpatialForceVector & operator += (const btSpatialForceVector &vec) { m_topVec += vec.m_topVec; m_bottomVec += vec.m_bottomVec; return *this; }
-	btSpatialForceVector & operator -= (const btSpatialForceVector &vec) { m_topVec -= vec.m_topVec; m_bottomVec -= vec.m_bottomVec; return *this; }
-	btSpatialForceVector operator - (const btSpatialForceVector &vec) const { return btSpatialForceVector(m_bottomVec - vec.m_bottomVec, m_topVec - vec.m_topVec); }
-	btSpatialForceVector operator + (const btSpatialForceVector &vec) const { return btSpatialForceVector(m_bottomVec + vec.m_bottomVec, m_topVec + vec.m_topVec); }
-	btSpatialForceVector operator - () const { return btSpatialForceVector(-m_bottomVec, -m_topVec); }
-	btSpatialForceVector operator * (const btScalar &s) const { return btSpatialForceVector(s * m_bottomVec, s * m_topVec); }		
+	btSpatialForceVector &operator+=(const btSpatialForceVector &vec)
+	{
+		m_topVec += vec.m_topVec;
+		m_bottomVec += vec.m_bottomVec;
+		return *this;
+	}
+	btSpatialForceVector &operator-=(const btSpatialForceVector &vec)
+	{
+		m_topVec -= vec.m_topVec;
+		m_bottomVec -= vec.m_bottomVec;
+		return *this;
+	}
+	btSpatialForceVector operator-(const btSpatialForceVector &vec) const { return btSpatialForceVector(m_bottomVec - vec.m_bottomVec, m_topVec - vec.m_topVec); }
+	btSpatialForceVector operator+(const btSpatialForceVector &vec) const { return btSpatialForceVector(m_bottomVec + vec.m_bottomVec, m_topVec + vec.m_topVec); }
+	btSpatialForceVector operator-() const { return btSpatialForceVector(-m_bottomVec, -m_topVec); }
+	btSpatialForceVector operator*(const btScalar &s) const { return btSpatialForceVector(s * m_bottomVec, s * m_topVec); }
 	//btSpatialForceVector & operator = (const btSpatialForceVector &vec) { m_topVec = vec.m_topVec; m_bottomVec = vec.m_bottomVec; return *this; }
 };
 
@@ -70,23 +96,36 @@ struct btSpatialMotionVector
 	btVector3 m_topVec, m_bottomVec;
 	//
 	btSpatialMotionVector() { setZero(); }
-	btSpatialMotionVector(const btVector3 &angular, const btVector3 &linear) : m_topVec(angular), m_bottomVec(linear) {}		
+	btSpatialMotionVector(const btVector3 &angular, const btVector3 &linear) : m_topVec(angular), m_bottomVec(linear) {}
 	//
-	void setVector(const btVector3 &angular, const btVector3 &linear) { m_topVec = angular; m_bottomVec = linear; }
+	void setVector(const btVector3 &angular, const btVector3 &linear)
+	{
+		m_topVec = angular;
+		m_bottomVec = linear;
+	}
 	void setValue(const btScalar &ax, const btScalar &ay, const btScalar &az, const btScalar &lx, const btScalar &ly, const btScalar &lz)
 	{
-		m_topVec.setValue(ax, ay, az); m_bottomVec.setValue(lx, ly, lz);
+		m_topVec.setValue(ax, ay, az);
+		m_bottomVec.setValue(lx, ly, lz);
 	}
 	//
-	void addVector(const btVector3 &angular, const btVector3 &linear) { m_topVec += linear; m_bottomVec += angular; }
+	void addVector(const btVector3 &angular, const btVector3 &linear)
+	{
+		m_topVec += linear;
+		m_bottomVec += angular;
+	}
 	void addValue(const btScalar &ax, const btScalar &ay, const btScalar &az, const btScalar &lx, const btScalar &ly, const btScalar &lz)
 	{
-		m_topVec[0] += ax; m_topVec[1] += ay; m_topVec[2] += az;
-		m_bottomVec[0] += lx; m_bottomVec[1] += ly; m_bottomVec[2] += lz;			
+		m_topVec[0] += ax;
+		m_topVec[1] += ay;
+		m_topVec[2] += az;
+		m_bottomVec[0] += lx;
+		m_bottomVec[1] += ly;
+		m_bottomVec[2] += lz;
 	}
-	//	
-	const btVector3 & getAngular() const { return m_topVec; }
-	const btVector3 & getLinear() const { return m_bottomVec; }
+	//
+	const btVector3 &getAngular() const { return m_topVec; }
+	const btVector3 &getLinear() const { return m_bottomVec; }
 	//
 	void setAngular(const btVector3 &angular) { m_topVec = angular; }
 	void setLinear(const btVector3 &linear) { m_bottomVec = linear; }
@@ -94,20 +133,24 @@ struct btSpatialMotionVector
 	void addAngular(const btVector3 &angular) { m_topVec += angular; }
 	void addLinear(const btVector3 &linear) { m_bottomVec += linear; }
 	//
-	void setZero() { m_topVec.setZero(); m_bottomVec.setZero(); }
+	void setZero()
+	{
+		m_topVec.setZero();
+		m_bottomVec.setZero();
+	}
 	//
 	btScalar dot(const btSpatialForceVector &b) const
 	{
 		return m_bottomVec.dot(b.m_topVec) + m_topVec.dot(b.m_bottomVec);
 	}
 	//
-	template<typename SpatialVectorType>
+	template <typename SpatialVectorType>
 	void cross(const SpatialVectorType &b, SpatialVectorType &out) const
 	{
 		out.m_topVec = m_topVec.cross(b.m_topVec);
 		out.m_bottomVec = m_bottomVec.cross(b.m_topVec) + m_topVec.cross(b.m_bottomVec);
 	}
-	template<typename SpatialVectorType>
+	template <typename SpatialVectorType>
 	SpatialVectorType cross(const SpatialVectorType &b) const
 	{
 		SpatialVectorType out;
@@ -116,21 +159,36 @@ struct btSpatialMotionVector
 		return out;
 	}
 	//
-	btSpatialMotionVector & operator += (const btSpatialMotionVector &vec) { m_topVec += vec.m_topVec; m_bottomVec += vec.m_bottomVec; return *this; }
-	btSpatialMotionVector & operator -= (const btSpatialMotionVector &vec) { m_topVec -= vec.m_topVec; m_bottomVec -= vec.m_bottomVec; return *this; }
-	btSpatialMotionVector & operator *= (const btScalar &s) { m_topVec *= s; m_bottomVec *= s; return *this; }
-	btSpatialMotionVector operator - (const btSpatialMotionVector &vec) const { return btSpatialMotionVector(m_topVec - vec.m_topVec, m_bottomVec - vec.m_bottomVec); }
-	btSpatialMotionVector operator + (const btSpatialMotionVector &vec) const { return btSpatialMotionVector(m_topVec + vec.m_topVec, m_bottomVec + vec.m_bottomVec); }
-	btSpatialMotionVector operator - () const { return btSpatialMotionVector(-m_topVec, -m_bottomVec); }
-	btSpatialMotionVector operator * (const btScalar &s) const { return btSpatialMotionVector(s * m_topVec, s * m_bottomVec); }
+	btSpatialMotionVector &operator+=(const btSpatialMotionVector &vec)
+	{
+		m_topVec += vec.m_topVec;
+		m_bottomVec += vec.m_bottomVec;
+		return *this;
+	}
+	btSpatialMotionVector &operator-=(const btSpatialMotionVector &vec)
+	{
+		m_topVec -= vec.m_topVec;
+		m_bottomVec -= vec.m_bottomVec;
+		return *this;
+	}
+	btSpatialMotionVector &operator*=(const btScalar &s)
+	{
+		m_topVec *= s;
+		m_bottomVec *= s;
+		return *this;
+	}
+	btSpatialMotionVector operator-(const btSpatialMotionVector &vec) const { return btSpatialMotionVector(m_topVec - vec.m_topVec, m_bottomVec - vec.m_bottomVec); }
+	btSpatialMotionVector operator+(const btSpatialMotionVector &vec) const { return btSpatialMotionVector(m_topVec + vec.m_topVec, m_bottomVec + vec.m_bottomVec); }
+	btSpatialMotionVector operator-() const { return btSpatialMotionVector(-m_topVec, -m_bottomVec); }
+	btSpatialMotionVector operator*(const btScalar &s) const { return btSpatialMotionVector(s * m_topVec, s * m_bottomVec); }
 };
 
 struct btSymmetricSpatialDyad
 {
 	btMatrix3x3 m_topLeftMat, m_topRightMat, m_bottomLeftMat;
-	//		
+	//
 	btSymmetricSpatialDyad() { setIdentity(); }
-	btSymmetricSpatialDyad(const btMatrix3x3 &topLeftMat, const btMatrix3x3 &topRightMat, const btMatrix3x3 &bottomLeftMat) { setMatrix(topLeftMat, topRightMat, bottomLeftMat); }			
+	btSymmetricSpatialDyad(const btMatrix3x3 &topLeftMat, const btMatrix3x3 &topRightMat, const btMatrix3x3 &bottomLeftMat) { setMatrix(topLeftMat, topRightMat, bottomLeftMat); }
 	//
 	void setMatrix(const btMatrix3x3 &topLeftMat, const btMatrix3x3 &topRightMat, const btMatrix3x3 &bottomLeftMat)
 	{
@@ -146,17 +204,22 @@ struct btSymmetricSpatialDyad
 		m_bottomLeftMat += bottomLeftMat;
 	}
 	//
-	void setIdentity() { m_topLeftMat.setIdentity(); m_topRightMat.setIdentity(); m_bottomLeftMat.setIdentity();  }
+	void setIdentity()
+	{
+		m_topLeftMat.setIdentity();
+		m_topRightMat.setIdentity();
+		m_bottomLeftMat.setIdentity();
+	}
 	//
-	btSymmetricSpatialDyad & operator -= (const btSymmetricSpatialDyad &mat)
+	btSymmetricSpatialDyad &operator-=(const btSymmetricSpatialDyad &mat)
 	{
 		m_topLeftMat -= mat.m_topLeftMat;
 		m_topRightMat -= mat.m_topRightMat;
 		m_bottomLeftMat -= mat.m_bottomLeftMat;
-		return *this; 
+		return *this;
 	}
 	//
-	btSpatialForceVector operator * (const btSpatialMotionVector &vec)
+	btSpatialForceVector operator*(const btSpatialMotionVector &vec)
 	{
 		return btSpatialForceVector(m_bottomLeftMat * vec.m_topVec + m_topLeftMat.transpose() * vec.m_bottomVec, m_topLeftMat * vec.m_topVec + m_topRightMat * vec.m_bottomVec);
 	}
@@ -164,7 +227,7 @@ struct btSymmetricSpatialDyad
 
 struct btSpatialTransformationMatrix
 {
-	btMatrix3x3 m_rotMat; //btMatrix3x3 m_trnCrossMat;
+	btMatrix3x3 m_rotMat;  //btMatrix3x3 m_trnCrossMat;
 	btVector3 m_trnVec;
 	//
 	enum eOutputOperation
@@ -174,128 +237,124 @@ struct btSpatialTransformationMatrix
 		Subtract = 2
 	};
 	//
-	template<typename SpatialVectorType>
-	void transform(	const SpatialVectorType &inVec,
-                      SpatialVectorType &outVec,
-					eOutputOperation outOp = None)
+	template <typename SpatialVectorType>
+	void transform(const SpatialVectorType &inVec,
+				   SpatialVectorType &outVec,
+				   eOutputOperation outOp = None)
 	{
-		if(outOp == None)
+		if (outOp == None)
 		{
 			outVec.m_topVec = m_rotMat * inVec.m_topVec;
 			outVec.m_bottomVec = -m_trnVec.cross(outVec.m_topVec) + m_rotMat * inVec.m_bottomVec;
 		}
-		else if(outOp == Add)
+		else if (outOp == Add)
 		{
 			outVec.m_topVec += m_rotMat * inVec.m_topVec;
 			outVec.m_bottomVec += -m_trnVec.cross(outVec.m_topVec) + m_rotMat * inVec.m_bottomVec;
 		}
-		else if(outOp == Subtract)
+		else if (outOp == Subtract)
 		{
 			outVec.m_topVec -= m_rotMat * inVec.m_topVec;
 			outVec.m_bottomVec -= -m_trnVec.cross(outVec.m_topVec) + m_rotMat * inVec.m_bottomVec;
 		}
-		
 	}
 
-	template<typename SpatialVectorType>
-	void transformRotationOnly(	const SpatialVectorType &inVec,
-								SpatialVectorType &outVec,
-								eOutputOperation outOp = None)
+	template <typename SpatialVectorType>
+	void transformRotationOnly(const SpatialVectorType &inVec,
+							   SpatialVectorType &outVec,
+							   eOutputOperation outOp = None)
 	{
-		if(outOp == None)
+		if (outOp == None)
 		{
 			outVec.m_topVec = m_rotMat * inVec.m_topVec;
 			outVec.m_bottomVec = m_rotMat * inVec.m_bottomVec;
 		}
-		else if(outOp == Add)
+		else if (outOp == Add)
 		{
 			outVec.m_topVec += m_rotMat * inVec.m_topVec;
 			outVec.m_bottomVec += m_rotMat * inVec.m_bottomVec;
 		}
-		else if(outOp == Subtract)
+		else if (outOp == Subtract)
 		{
 			outVec.m_topVec -= m_rotMat * inVec.m_topVec;
 			outVec.m_bottomVec -= m_rotMat * inVec.m_bottomVec;
 		}
-		
 	}
 
-	template<typename SpatialVectorType>
-	void transformInverse(	const SpatialVectorType &inVec,
-							SpatialVectorType &outVec,
-							eOutputOperation outOp = None)
+	template <typename SpatialVectorType>
+	void transformInverse(const SpatialVectorType &inVec,
+						  SpatialVectorType &outVec,
+						  eOutputOperation outOp = None)
 	{
-		if(outOp == None)
+		if (outOp == None)
 		{
 			outVec.m_topVec = m_rotMat.transpose() * inVec.m_topVec;
 			outVec.m_bottomVec = m_rotMat.transpose() * (inVec.m_bottomVec + m_trnVec.cross(inVec.m_topVec));
 		}
-		else if(outOp == Add)
+		else if (outOp == Add)
 		{
 			outVec.m_topVec += m_rotMat.transpose() * inVec.m_topVec;
 			outVec.m_bottomVec += m_rotMat.transpose() * (inVec.m_bottomVec + m_trnVec.cross(inVec.m_topVec));
 		}
-		else if(outOp == Subtract)
+		else if (outOp == Subtract)
 		{
 			outVec.m_topVec -= m_rotMat.transpose() * inVec.m_topVec;
 			outVec.m_bottomVec -= m_rotMat.transpose() * (inVec.m_bottomVec + m_trnVec.cross(inVec.m_topVec));
-		}			
+		}
 	}
 
-	template<typename SpatialVectorType>
-	void transformInverseRotationOnly(	const SpatialVectorType &inVec,
-										SpatialVectorType &outVec,
-										eOutputOperation outOp = None)
+	template <typename SpatialVectorType>
+	void transformInverseRotationOnly(const SpatialVectorType &inVec,
+									  SpatialVectorType &outVec,
+									  eOutputOperation outOp = None)
 	{
-		if(outOp == None)
+		if (outOp == None)
 		{
 			outVec.m_topVec = m_rotMat.transpose() * inVec.m_topVec;
 			outVec.m_bottomVec = m_rotMat.transpose() * inVec.m_bottomVec;
 		}
-		else if(outOp == Add)
+		else if (outOp == Add)
 		{
 			outVec.m_topVec += m_rotMat.transpose() * inVec.m_topVec;
 			outVec.m_bottomVec += m_rotMat.transpose() * inVec.m_bottomVec;
 		}
-		else if(outOp == Subtract)
+		else if (outOp == Subtract)
 		{
 			outVec.m_topVec -= m_rotMat.transpose() * inVec.m_topVec;
 			outVec.m_bottomVec -= m_rotMat.transpose() * inVec.m_bottomVec;
 		}
-		
 	}
 
-	void transformInverse(	const btSymmetricSpatialDyad &inMat,
-							btSymmetricSpatialDyad &outMat,
-							eOutputOperation outOp = None)
+	void transformInverse(const btSymmetricSpatialDyad &inMat,
+						  btSymmetricSpatialDyad &outMat,
+						  eOutputOperation outOp = None)
 	{
-		const btMatrix3x3 r_cross(	0, -m_trnVec[2], m_trnVec[1],
-								m_trnVec[2], 0, -m_trnVec[0],
-								-m_trnVec[1], m_trnVec[0], 0);
+		const btMatrix3x3 r_cross(0, -m_trnVec[2], m_trnVec[1],
+								  m_trnVec[2], 0, -m_trnVec[0],
+								  -m_trnVec[1], m_trnVec[0], 0);
 
-
-		if(outOp == None)
+		if (outOp == None)
 		{
-			outMat.m_topLeftMat = m_rotMat.transpose() * ( inMat.m_topLeftMat - inMat.m_topRightMat * r_cross ) * m_rotMat;
+			outMat.m_topLeftMat = m_rotMat.transpose() * (inMat.m_topLeftMat - inMat.m_topRightMat * r_cross) * m_rotMat;
 			outMat.m_topRightMat = m_rotMat.transpose() * inMat.m_topRightMat * m_rotMat;
 			outMat.m_bottomLeftMat = m_rotMat.transpose() * (r_cross * (inMat.m_topLeftMat - inMat.m_topRightMat * r_cross) + inMat.m_bottomLeftMat - inMat.m_topLeftMat.transpose() * r_cross) * m_rotMat;
 		}
-		else if(outOp == Add)
+		else if (outOp == Add)
 		{
-			outMat.m_topLeftMat += m_rotMat.transpose() * ( inMat.m_topLeftMat - inMat.m_topRightMat * r_cross ) * m_rotMat;
+			outMat.m_topLeftMat += m_rotMat.transpose() * (inMat.m_topLeftMat - inMat.m_topRightMat * r_cross) * m_rotMat;
 			outMat.m_topRightMat += m_rotMat.transpose() * inMat.m_topRightMat * m_rotMat;
 			outMat.m_bottomLeftMat += m_rotMat.transpose() * (r_cross * (inMat.m_topLeftMat - inMat.m_topRightMat * r_cross) + inMat.m_bottomLeftMat - inMat.m_topLeftMat.transpose() * r_cross) * m_rotMat;
 		}
-		else if(outOp == Subtract)
+		else if (outOp == Subtract)
 		{
-			outMat.m_topLeftMat -= m_rotMat.transpose() * ( inMat.m_topLeftMat - inMat.m_topRightMat * r_cross ) * m_rotMat;
+			outMat.m_topLeftMat -= m_rotMat.transpose() * (inMat.m_topLeftMat - inMat.m_topRightMat * r_cross) * m_rotMat;
 			outMat.m_topRightMat -= m_rotMat.transpose() * inMat.m_topRightMat * m_rotMat;
 			outMat.m_bottomLeftMat -= m_rotMat.transpose() * (r_cross * (inMat.m_topLeftMat - inMat.m_topRightMat * r_cross) + inMat.m_bottomLeftMat - inMat.m_topLeftMat.transpose() * r_cross) * m_rotMat;
 		}
 	}
 
-	template<typename SpatialVectorType>
-	SpatialVectorType operator * (const SpatialVectorType &vec)
+	template <typename SpatialVectorType>
+	SpatialVectorType operator*(const SpatialVectorType &vec)
 	{
 		SpatialVectorType out;
 		transform(vec, out);
@@ -303,7 +362,7 @@ struct btSpatialTransformationMatrix
 	}
 };
 
-template<typename SpatialVectorType>
+template <typename SpatialVectorType>
 void symmetricSpatialOuterProduct(const SpatialVectorType &a, const SpatialVectorType &b, btSymmetricSpatialDyad &out)
 {
 	//output op maybe?
@@ -314,7 +373,7 @@ void symmetricSpatialOuterProduct(const SpatialVectorType &a, const SpatialVecto
 	//maybe simple a*spatTranspose(a) would be nicer?
 }
 
-template<typename SpatialVectorType>
+template <typename SpatialVectorType>
 btSymmetricSpatialDyad symmetricSpatialOuterProduct(const SpatialVectorType &a, const SpatialVectorType &b)
 {
 	btSymmetricSpatialDyad out;
@@ -327,5 +386,4 @@ btSymmetricSpatialDyad symmetricSpatialOuterProduct(const SpatialVectorType &a,
 	//maybe simple a*spatTranspose(a) would be nicer?
 }
 
-#endif //BT_SPATIAL_ALGEBRA_H
-
+#endif  //BT_SPATIAL_ALGEBRA_H
diff --git a/thirdparty/bullet/LinearMath/btStackAlloc.h b/thirdparty/bullet/LinearMath/btStackAlloc.h
index 397b084877f..3fc20849765 100644
--- a/thirdparty/bullet/LinearMath/btStackAlloc.h
+++ b/thirdparty/bullet/LinearMath/btStackAlloc.h
@@ -20,97 +20,99 @@ Nov.2006
 #ifndef BT_STACK_ALLOC
 #define BT_STACK_ALLOC
 
-#include "btScalar.h" //for btAssert
+#include "btScalar.h"  //for btAssert
 #include "btAlignedAllocator.h"
 
 ///The btBlock class is an internal structure for the btStackAlloc memory allocator.
 struct btBlock
 {
-	btBlock*			previous;
-	unsigned char*		address;
+	btBlock* previous;
+	unsigned char* address;
 };
 
 ///The StackAlloc class provides some fast stack-based memory allocator (LIFO last-in first-out)
 class btStackAlloc
 {
 public:
+	btStackAlloc(unsigned int size)
+	{
+		ctor();
+		create(size);
+	}
+	~btStackAlloc() { destroy(); }
 
-	btStackAlloc(unsigned int size)	{ ctor();create(size); }
-	~btStackAlloc()		{ destroy(); }
-	
-	inline void		create(unsigned int size)
+	inline void create(unsigned int size)
 	{
 		destroy();
-		data		=  (unsigned char*) btAlignedAlloc(size,16);
-		totalsize	=	size;
+		data = (unsigned char*)btAlignedAlloc(size, 16);
+		totalsize = size;
 	}
-	inline void		destroy()
+	inline void destroy()
 	{
-		btAssert(usedsize==0);
+		btAssert(usedsize == 0);
 		//Raise(L"StackAlloc is still in use");
 
-		if(usedsize==0)
+		if (usedsize == 0)
 		{
-			if(!ischild && data)		
+			if (!ischild && data)
 				btAlignedFree(data);
 
-			data				=	0;
-			usedsize			=	0;
+			data = 0;
+			usedsize = 0;
 		}
-		
 	}
 
-	int	getAvailableMemory() const
+	int getAvailableMemory() const
 	{
 		return static_cast<int>(totalsize - usedsize);
 	}
 
-	unsigned char*			allocate(unsigned int size)
+	unsigned char* allocate(unsigned int size)
 	{
-		const unsigned int	nus(usedsize+size);
-		if(nus<totalsize)
+		const unsigned int nus(usedsize + size);
+		if (nus < totalsize)
 		{
-			usedsize=nus;
-			return(data+(usedsize-size));
+			usedsize = nus;
+			return (data + (usedsize - size));
 		}
 		btAssert(0);
 		//&& (L"Not enough memory"));
-		
-		return(0);
+
+		return (0);
 	}
-	SIMD_FORCE_INLINE btBlock*		beginBlock()
+	SIMD_FORCE_INLINE btBlock* beginBlock()
 	{
-		btBlock*	pb = (btBlock*)allocate(sizeof(btBlock));
-		pb->previous	=	current;
-		pb->address		=	data+usedsize;
-		current			=	pb;
-		return(pb);
+		btBlock* pb = (btBlock*)allocate(sizeof(btBlock));
+		pb->previous = current;
+		pb->address = data + usedsize;
+		current = pb;
+		return (pb);
 	}
-	SIMD_FORCE_INLINE void		endBlock(btBlock* block)
+	SIMD_FORCE_INLINE void endBlock(btBlock* block)
 	{
-		btAssert(block==current);
+		btAssert(block == current);
 		//Raise(L"Unmatched blocks");
-		if(block==current)
+		if (block == current)
 		{
-			current		=	block->previous;
-			usedsize	=	(unsigned int)((block->address-data)-sizeof(btBlock));
+			current = block->previous;
+			usedsize = (unsigned int)((block->address - data) - sizeof(btBlock));
 		}
 	}
 
 private:
-	void		ctor()
+	void ctor()
 	{
-		data		=	0;
-		totalsize	=	0;
-		usedsize	=	0;
-		current		=	0;
-		ischild		=	false;
+		data = 0;
+		totalsize = 0;
+		usedsize = 0;
+		current = 0;
+		ischild = false;
 	}
-	unsigned char*		data;
-	unsigned int		totalsize;
-	unsigned int		usedsize;
-	btBlock*	current;
-	bool		ischild;
+	unsigned char* data;
+	unsigned int totalsize;
+	unsigned int usedsize;
+	btBlock* current;
+	bool ischild;
 };
 
-#endif //BT_STACK_ALLOC
+#endif  //BT_STACK_ALLOC
diff --git a/thirdparty/bullet/LinearMath/btThreads.cpp b/thirdparty/bullet/LinearMath/btThreads.cpp
index c037626ffb0..69a86799fad 100644
--- a/thirdparty/bullet/LinearMath/btThreads.cpp
+++ b/thirdparty/bullet/LinearMath/btThreads.cpp
@@ -12,18 +12,15 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #include "btThreads.h"
 #include "btQuickprof.h"
 #include <algorithm>  // for min and max
 
-
 #if BT_USE_OPENMP && BT_THREADSAFE
 
 #include <omp.h>
 
-#endif // #if BT_USE_OPENMP && BT_THREADSAFE
-
+#endif  // #if BT_USE_OPENMP && BT_THREADSAFE
 
 #if BT_USE_PPL && BT_THREADSAFE
 
@@ -32,8 +29,7 @@ subject to the following restrictions:
 // Visual Studio 2010 and later should come with it
 #include <concrtrm.h>  // for GetProcessorCount()
 
-#endif // #if BT_USE_PPL && BT_THREADSAFE
-
+#endif  // #if BT_USE_PPL && BT_THREADSAFE
 
 #if BT_USE_TBB && BT_THREADSAFE
 
@@ -44,8 +40,7 @@ subject to the following restrictions:
 #include <tbb/parallel_for.h>
 #include <tbb/blocked_range.h>
 
-#endif // #if BT_USE_TBB && BT_THREADSAFE
-
+#endif  // #if BT_USE_TBB && BT_THREADSAFE
 
 #if BT_THREADSAFE
 //
@@ -53,7 +48,7 @@ subject to the following restrictions:
 // Using ordinary system-provided mutexes like Windows critical sections was noticeably slower
 // presumably because when it fails to lock at first it would sleep the thread and trigger costly
 // context switching.
-// 
+//
 
 #if __cplusplus >= 201103L
 
@@ -61,25 +56,24 @@ subject to the following restrictions:
 // on GCC or Clang you need to compile with -std=c++11
 #define USE_CPP11_ATOMICS 1
 
-#elif defined( _MSC_VER )
+#elif defined(_MSC_VER)
 
 // on MSVC, use intrinsics instead
 #define USE_MSVC_INTRINSICS 1
 
-#elif defined( __GNUC__ ) && (__GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 7))
+#elif defined(__GNUC__) && (__GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 7))
 
 // available since GCC 4.7 and some versions of clang
 // todo: check for clang
 #define USE_GCC_BUILTIN_ATOMICS 1
 
-#elif defined( __GNUC__ ) && (__GNUC__ == 4 && __GNUC_MINOR__ >= 1)
+#elif defined(__GNUC__) && (__GNUC__ == 4 && __GNUC_MINOR__ >= 1)
 
 // available since GCC 4.1
 #define USE_GCC_BUILTIN_ATOMICS_OLD 1
 
 #endif
 
-
 #if USE_CPP11_ATOMICS
 
 #include <atomic>
@@ -89,27 +83,26 @@ subject to the following restrictions:
 
 bool btSpinMutex::tryLock()
 {
-    std::atomic<int>* aDest = reinterpret_cast<std::atomic<int>*>(&mLock);
-    int expected = 0;
-    return std::atomic_compare_exchange_weak_explicit( aDest, &expected, int(1), std::memory_order_acq_rel, std::memory_order_acquire );
+	std::atomic<int>* aDest = reinterpret_cast<std::atomic<int>*>(&mLock);
+	int expected = 0;
+	return std::atomic_compare_exchange_weak_explicit(aDest, &expected, int(1), std::memory_order_acq_rel, std::memory_order_acquire);
 }
 
 void btSpinMutex::lock()
 {
-    // note: this lock does not sleep the thread.
-    while (! tryLock())
-    {
-        // spin
-    }
+	// note: this lock does not sleep the thread.
+	while (!tryLock())
+	{
+		// spin
+	}
 }
 
 void btSpinMutex::unlock()
 {
-    std::atomic<int>* aDest = reinterpret_cast<std::atomic<int>*>(&mLock);
-    std::atomic_store_explicit( aDest, int(0), std::memory_order_release );
+	std::atomic<int>* aDest = reinterpret_cast<std::atomic<int>*>(&mLock);
+	std::atomic_store_explicit(aDest, int(0), std::memory_order_release);
 }
 
-
 #elif USE_MSVC_INTRINSICS
 
 #define WIN32_LEAN_AND_MEAN
@@ -117,148 +110,142 @@ void btSpinMutex::unlock()
 #include <windows.h>
 #include <intrin.h>
 
-#define THREAD_LOCAL_STATIC __declspec( thread ) static
-
+#define THREAD_LOCAL_STATIC __declspec(thread) static
 
 bool btSpinMutex::tryLock()
 {
-    volatile long* aDest = reinterpret_cast<long*>(&mLock);
-    return ( 0 == _InterlockedCompareExchange( aDest, 1, 0) );
+	volatile long* aDest = reinterpret_cast<long*>(&mLock);
+	return (0 == _InterlockedCompareExchange(aDest, 1, 0));
 }
 
 void btSpinMutex::lock()
 {
-    // note: this lock does not sleep the thread
-    while (! tryLock())
-    {
-        // spin
-    }
+	// note: this lock does not sleep the thread
+	while (!tryLock())
+	{
+		// spin
+	}
 }
 
 void btSpinMutex::unlock()
 {
-    volatile long* aDest = reinterpret_cast<long*>( &mLock );
-    _InterlockedExchange( aDest, 0 );
+	volatile long* aDest = reinterpret_cast<long*>(&mLock);
+	_InterlockedExchange(aDest, 0);
 }
 
 #elif USE_GCC_BUILTIN_ATOMICS
 
 #define THREAD_LOCAL_STATIC static __thread
 
-
 bool btSpinMutex::tryLock()
 {
-    int expected = 0;
-    bool weak = false;
-    const int memOrderSuccess = __ATOMIC_ACQ_REL;
-    const int memOrderFail = __ATOMIC_ACQUIRE;
-    return __atomic_compare_exchange_n(&mLock, &expected, int(1), weak, memOrderSuccess, memOrderFail);
+	int expected = 0;
+	bool weak = false;
+	const int memOrderSuccess = __ATOMIC_ACQ_REL;
+	const int memOrderFail = __ATOMIC_ACQUIRE;
+	return __atomic_compare_exchange_n(&mLock, &expected, int(1), weak, memOrderSuccess, memOrderFail);
 }
 
 void btSpinMutex::lock()
 {
-    // note: this lock does not sleep the thread
-    while (! tryLock())
-    {
-        // spin
-    }
+	// note: this lock does not sleep the thread
+	while (!tryLock())
+	{
+		// spin
+	}
 }
 
 void btSpinMutex::unlock()
 {
-    __atomic_store_n(&mLock, int(0), __ATOMIC_RELEASE);
+	__atomic_store_n(&mLock, int(0), __ATOMIC_RELEASE);
 }
 
 #elif USE_GCC_BUILTIN_ATOMICS_OLD
 
-
 #define THREAD_LOCAL_STATIC static __thread
 
 bool btSpinMutex::tryLock()
 {
-    return __sync_bool_compare_and_swap(&mLock, int(0), int(1));
+	return __sync_bool_compare_and_swap(&mLock, int(0), int(1));
 }
 
 void btSpinMutex::lock()
 {
-    // note: this lock does not sleep the thread
-    while (! tryLock())
-    {
-        // spin
-    }
+	// note: this lock does not sleep the thread
+	while (!tryLock())
+	{
+		// spin
+	}
 }
 
 void btSpinMutex::unlock()
 {
-    // write 0
-    __sync_fetch_and_and(&mLock, int(0));
+	// write 0
+	__sync_fetch_and_and(&mLock, int(0));
 }
 
-#else //#elif USE_MSVC_INTRINSICS
+#else  //#elif USE_MSVC_INTRINSICS
 
 #error "no threading primitives defined -- unknown platform"
 
 #endif  //#else //#elif USE_MSVC_INTRINSICS
 
-#else //#if BT_THREADSAFE
+#else  //#if BT_THREADSAFE
 
 // These should not be called ever
 void btSpinMutex::lock()
 {
-    btAssert( !"unimplemented btSpinMutex::lock() called" );
+	btAssert(!"unimplemented btSpinMutex::lock() called");
 }
 
 void btSpinMutex::unlock()
 {
-    btAssert( !"unimplemented btSpinMutex::unlock() called" );
+	btAssert(!"unimplemented btSpinMutex::unlock() called");
 }
 
 bool btSpinMutex::tryLock()
 {
-    btAssert( !"unimplemented btSpinMutex::tryLock() called" );
-    return true;
+	btAssert(!"unimplemented btSpinMutex::tryLock() called");
+	return true;
 }
 
 #define THREAD_LOCAL_STATIC static
 
-#endif // #else //#if BT_THREADSAFE
-
+#endif  // #else //#if BT_THREADSAFE
 
 struct ThreadsafeCounter
 {
-    unsigned int mCounter;
-    btSpinMutex mMutex;
+	unsigned int mCounter;
+	btSpinMutex mMutex;
 
-    ThreadsafeCounter()
-    {
-        mCounter = 0;
-        --mCounter; // first count should come back 0
-    }
+	ThreadsafeCounter()
+	{
+		mCounter = 0;
+		--mCounter;  // first count should come back 0
+	}
 
-    unsigned int getNext()
-    {
-        // no need to optimize this with atomics, it is only called ONCE per thread!
-        mMutex.lock();
-        mCounter++;
-        if ( mCounter >= BT_MAX_THREAD_COUNT )
-        {
-            btAssert( !"thread counter exceeded" );
-            // wrap back to the first worker index
-            mCounter = 1;
-        }
-        unsigned int val = mCounter;
-        mMutex.unlock();
-        return val;
-    }
+	unsigned int getNext()
+	{
+		// no need to optimize this with atomics, it is only called ONCE per thread!
+		mMutex.lock();
+		mCounter++;
+		if (mCounter >= BT_MAX_THREAD_COUNT)
+		{
+			btAssert(!"thread counter exceeded");
+			// wrap back to the first worker index
+			mCounter = 1;
+		}
+		unsigned int val = mCounter;
+		mMutex.unlock();
+		return val;
+	}
 };
 
-
-static btITaskScheduler* gBtTaskScheduler;
+static btITaskScheduler* gBtTaskScheduler=0;
 static int gThreadsRunningCounter = 0;  // useful for detecting if we are trying to do nested parallel-for calls
 static btSpinMutex gThreadsRunningCounterMutex;
 static ThreadsafeCounter gThreadCounter;
 
-
 //
 // BT_DETECT_BAD_THREAD_INDEX tries to detect when there are multiple threads assigned the same thread index.
 //
@@ -276,7 +263,7 @@ static ThreadsafeCounter gThreadCounter;
 // We allocate thread-indexes as needed with a sequential global thread counter.
 //
 // Our simple thread-counting scheme falls apart if the task scheduler destroys some threads but
-// continues to re-use other threads and the application repeatedly resizes the thread pool of the 
+// continues to re-use other threads and the application repeatedly resizes the thread pool of the
 // task scheduler.
 // In order to prevent the thread-counter from exceeding the global max (BT_MAX_THREAD_COUNT), we
 // wrap the thread counter back to 1. This should only happen if the worker threads have all been
@@ -290,197 +277,192 @@ static ThreadsafeCounter gThreadCounter;
 
 typedef DWORD ThreadId_t;
 const static ThreadId_t kInvalidThreadId = 0;
-ThreadId_t gDebugThreadIds[ BT_MAX_THREAD_COUNT ];
+ThreadId_t gDebugThreadIds[BT_MAX_THREAD_COUNT];
 
 static ThreadId_t getDebugThreadId()
 {
-    return GetCurrentThreadId();
+	return GetCurrentThreadId();
 }
 
-#endif // #if BT_DETECT_BAD_THREAD_INDEX
-
+#endif  // #if BT_DETECT_BAD_THREAD_INDEX
 
 // return a unique index per thread, main thread is 0, worker threads are in [1, BT_MAX_THREAD_COUNT)
 unsigned int btGetCurrentThreadIndex()
 {
-    const unsigned int kNullIndex = ~0U;
-    THREAD_LOCAL_STATIC unsigned int sThreadIndex = kNullIndex;
-    if ( sThreadIndex == kNullIndex )
-    {
-        sThreadIndex = gThreadCounter.getNext();
-        btAssert( sThreadIndex < BT_MAX_THREAD_COUNT );
-    }
+	const unsigned int kNullIndex = ~0U;
+	THREAD_LOCAL_STATIC unsigned int sThreadIndex = kNullIndex;
+	if (sThreadIndex == kNullIndex)
+	{
+		sThreadIndex = gThreadCounter.getNext();
+		btAssert(sThreadIndex < BT_MAX_THREAD_COUNT);
+	}
 #if BT_DETECT_BAD_THREAD_INDEX
-    if ( gBtTaskScheduler && sThreadIndex > 0 )
-    {
-        ThreadId_t tid = getDebugThreadId();
-        // if not set
-        if ( gDebugThreadIds[ sThreadIndex ] == kInvalidThreadId )
-        {
-            // set it
-            gDebugThreadIds[ sThreadIndex ] = tid;
-        }
-        else
-        {
-            if ( gDebugThreadIds[ sThreadIndex ] != tid )
-            {
-                // this could indicate the task scheduler is breaking our assumptions about
-                // how threads are managed when threadpool is resized
-                btAssert( !"there are 2 or more threads with the same thread-index!" );
-                __debugbreak();
-            }
-        }
-    }
-#endif // #if BT_DETECT_BAD_THREAD_INDEX
-    return sThreadIndex;
+	if (gBtTaskScheduler && sThreadIndex > 0)
+	{
+		ThreadId_t tid = getDebugThreadId();
+		// if not set
+		if (gDebugThreadIds[sThreadIndex] == kInvalidThreadId)
+		{
+			// set it
+			gDebugThreadIds[sThreadIndex] = tid;
+		}
+		else
+		{
+			if (gDebugThreadIds[sThreadIndex] != tid)
+			{
+				// this could indicate the task scheduler is breaking our assumptions about
+				// how threads are managed when threadpool is resized
+				btAssert(!"there are 2 or more threads with the same thread-index!");
+				__debugbreak();
+			}
+		}
+	}
+#endif  // #if BT_DETECT_BAD_THREAD_INDEX
+	return sThreadIndex;
 }
 
 bool btIsMainThread()
 {
-    return btGetCurrentThreadIndex() == 0;
+	return btGetCurrentThreadIndex() == 0;
 }
 
 void btResetThreadIndexCounter()
 {
-    // for when all current worker threads are destroyed
-    btAssert( btIsMainThread() );
-    gThreadCounter.mCounter = 0;
+	// for when all current worker threads are destroyed
+	btAssert(btIsMainThread());
+	gThreadCounter.mCounter = 0;
 }
 
-btITaskScheduler::btITaskScheduler( const char* name )
+btITaskScheduler::btITaskScheduler(const char* name)
 {
-    m_name = name;
-    m_savedThreadCounter = 0;
-    m_isActive = false;
+	m_name = name;
+	m_savedThreadCounter = 0;
+	m_isActive = false;
 }
 
 void btITaskScheduler::activate()
 {
-    // gThreadCounter is used to assign a thread-index to each worker thread in a task scheduler.
-    // The main thread is always thread-index 0, and worker threads are numbered from 1 to 63 (BT_MAX_THREAD_COUNT-1)
-    // The thread-indexes need to be unique amongst the threads that can be running simultaneously.
-    // Since only one task scheduler can be used at a time, it is OK for a pair of threads that belong to different
-    // task schedulers to share the same thread index because they can't be running at the same time.
-    // So each task scheduler needs to keep its own thread counter value
-    if ( !m_isActive )
-    {
-        gThreadCounter.mCounter = m_savedThreadCounter;  // restore saved thread counter
-        m_isActive = true;
-    }
+	// gThreadCounter is used to assign a thread-index to each worker thread in a task scheduler.
+	// The main thread is always thread-index 0, and worker threads are numbered from 1 to 63 (BT_MAX_THREAD_COUNT-1)
+	// The thread-indexes need to be unique amongst the threads that can be running simultaneously.
+	// Since only one task scheduler can be used at a time, it is OK for a pair of threads that belong to different
+	// task schedulers to share the same thread index because they can't be running at the same time.
+	// So each task scheduler needs to keep its own thread counter value
+	if (!m_isActive)
+	{
+		gThreadCounter.mCounter = m_savedThreadCounter;  // restore saved thread counter
+		m_isActive = true;
+	}
 }
 
 void btITaskScheduler::deactivate()
 {
-    if ( m_isActive )
-    {
-        m_savedThreadCounter = gThreadCounter.mCounter;  // save thread counter
-        m_isActive = false;
-    }
+	if (m_isActive)
+	{
+		m_savedThreadCounter = gThreadCounter.mCounter;  // save thread counter
+		m_isActive = false;
+	}
 }
 
 void btPushThreadsAreRunning()
 {
-    gThreadsRunningCounterMutex.lock();
-    gThreadsRunningCounter++;
-    gThreadsRunningCounterMutex.unlock();
+	gThreadsRunningCounterMutex.lock();
+	gThreadsRunningCounter++;
+	gThreadsRunningCounterMutex.unlock();
 }
 
 void btPopThreadsAreRunning()
 {
-    gThreadsRunningCounterMutex.lock();
-    gThreadsRunningCounter--;
-    gThreadsRunningCounterMutex.unlock();
+	gThreadsRunningCounterMutex.lock();
+	gThreadsRunningCounter--;
+	gThreadsRunningCounterMutex.unlock();
 }
 
 bool btThreadsAreRunning()
 {
-    return gThreadsRunningCounter != 0;
+	return gThreadsRunningCounter != 0;
 }
 
-
-void btSetTaskScheduler( btITaskScheduler* ts )
+void btSetTaskScheduler(btITaskScheduler* ts)
 {
-    int threadId = btGetCurrentThreadIndex();  // make sure we call this on main thread at least once before any workers run
-    if ( threadId != 0 )
-    {
-        btAssert( !"btSetTaskScheduler must be called from the main thread!" );
-        return;
-    }
-    if ( gBtTaskScheduler )
-    {
-        // deactivate old task scheduler
-        gBtTaskScheduler->deactivate();
-    }
-    gBtTaskScheduler = ts;
-    if ( ts )
-    {
-        // activate new task scheduler
-        ts->activate();
-    }
+	int threadId = btGetCurrentThreadIndex();  // make sure we call this on main thread at least once before any workers run
+	if (threadId != 0)
+	{
+		btAssert(!"btSetTaskScheduler must be called from the main thread!");
+		return;
+	}
+	if (gBtTaskScheduler)
+	{
+		// deactivate old task scheduler
+		gBtTaskScheduler->deactivate();
+	}
+	gBtTaskScheduler = ts;
+	if (ts)
+	{
+		// activate new task scheduler
+		ts->activate();
+	}
 }
 
-
 btITaskScheduler* btGetTaskScheduler()
 {
-    return gBtTaskScheduler;
+	return gBtTaskScheduler;
 }
 
-
-void btParallelFor( int iBegin, int iEnd, int grainSize, const btIParallelForBody& body )
+void btParallelFor(int iBegin, int iEnd, int grainSize, const btIParallelForBody& body)
 {
 #if BT_THREADSAFE
 
 #if BT_DETECT_BAD_THREAD_INDEX
-    if ( !btThreadsAreRunning() )
-    {
-        // clear out thread ids
-        for ( int i = 0; i < BT_MAX_THREAD_COUNT; ++i )
-        {
-            gDebugThreadIds[ i ] = kInvalidThreadId;
-        }
-    }
-#endif // #if BT_DETECT_BAD_THREAD_INDEX
+	if (!btThreadsAreRunning())
+	{
+		// clear out thread ids
+		for (int i = 0; i < BT_MAX_THREAD_COUNT; ++i)
+		{
+			gDebugThreadIds[i] = kInvalidThreadId;
+		}
+	}
+#endif  // #if BT_DETECT_BAD_THREAD_INDEX
 
-    btAssert( gBtTaskScheduler != NULL );  // call btSetTaskScheduler() with a valid task scheduler first!
-    gBtTaskScheduler->parallelFor( iBegin, iEnd, grainSize, body );
+	btAssert(gBtTaskScheduler != NULL);  // call btSetTaskScheduler() with a valid task scheduler first!
+	gBtTaskScheduler->parallelFor(iBegin, iEnd, grainSize, body);
 
-#else // #if BT_THREADSAFE
+#else  // #if BT_THREADSAFE
 
-    // non-parallel version of btParallelFor
-    btAssert( !"called btParallelFor in non-threadsafe build. enable BT_THREADSAFE" );
-    body.forLoop( iBegin, iEnd );
+	// non-parallel version of btParallelFor
+	btAssert(!"called btParallelFor in non-threadsafe build. enable BT_THREADSAFE");
+	body.forLoop(iBegin, iEnd);
 
-#endif// #if BT_THREADSAFE
+#endif  // #if BT_THREADSAFE
 }
 
-btScalar btParallelSum( int iBegin, int iEnd, int grainSize, const btIParallelSumBody& body )
+btScalar btParallelSum(int iBegin, int iEnd, int grainSize, const btIParallelSumBody& body)
 {
 #if BT_THREADSAFE
 
 #if BT_DETECT_BAD_THREAD_INDEX
-    if ( !btThreadsAreRunning() )
-    {
-        // clear out thread ids
-        for ( int i = 0; i < BT_MAX_THREAD_COUNT; ++i )
-        {
-            gDebugThreadIds[ i ] = kInvalidThreadId;
-        }
-    }
-#endif // #if BT_DETECT_BAD_THREAD_INDEX
+	if (!btThreadsAreRunning())
+	{
+		// clear out thread ids
+		for (int i = 0; i < BT_MAX_THREAD_COUNT; ++i)
+		{
+			gDebugThreadIds[i] = kInvalidThreadId;
+		}
+	}
+#endif  // #if BT_DETECT_BAD_THREAD_INDEX
 
-    btAssert( gBtTaskScheduler != NULL );  // call btSetTaskScheduler() with a valid task scheduler first!
-    return gBtTaskScheduler->parallelSum( iBegin, iEnd, grainSize, body );
+	btAssert(gBtTaskScheduler != NULL);  // call btSetTaskScheduler() with a valid task scheduler first!
+	return gBtTaskScheduler->parallelSum(iBegin, iEnd, grainSize, body);
 
-#else // #if BT_THREADSAFE
+#else  // #if BT_THREADSAFE
 
-    // non-parallel version of btParallelSum
-    btAssert( !"called btParallelFor in non-threadsafe build. enable BT_THREADSAFE" );
-    return body.sumLoop( iBegin, iEnd );
+	// non-parallel version of btParallelSum
+	btAssert(!"called btParallelFor in non-threadsafe build. enable BT_THREADSAFE");
+	return body.sumLoop(iBegin, iEnd);
 
-#endif //#else // #if BT_THREADSAFE
+#endif  //#else // #if BT_THREADSAFE
 }
 
-
 ///
 /// btTaskSchedulerSequential -- non-threaded implementation of task scheduler
 ///                              (really just useful for testing performance of single threaded vs multi)
@@ -488,86 +470,86 @@ btScalar btParallelSum( int iBegin, int iEnd, int grainSize, const btIParallelSu
 class btTaskSchedulerSequential : public btITaskScheduler
 {
 public:
-    btTaskSchedulerSequential() : btITaskScheduler( "Sequential" ) {}
-    virtual int getMaxNumThreads() const BT_OVERRIDE { return 1; }
-    virtual int getNumThreads() const BT_OVERRIDE { return 1; }
-    virtual void setNumThreads( int numThreads ) BT_OVERRIDE {}
-    virtual void parallelFor( int iBegin, int iEnd, int grainSize, const btIParallelForBody& body ) BT_OVERRIDE
-    {
-        BT_PROFILE( "parallelFor_sequential" );
-        body.forLoop( iBegin, iEnd );
-    }
-    virtual btScalar parallelSum( int iBegin, int iEnd, int grainSize, const btIParallelSumBody& body ) BT_OVERRIDE
-    {
-        BT_PROFILE( "parallelSum_sequential" );
-        return body.sumLoop( iBegin, iEnd );
-    }
+	btTaskSchedulerSequential() : btITaskScheduler("Sequential") {}
+	virtual int getMaxNumThreads() const BT_OVERRIDE { return 1; }
+	virtual int getNumThreads() const BT_OVERRIDE { return 1; }
+	virtual void setNumThreads(int numThreads) BT_OVERRIDE {}
+	virtual void parallelFor(int iBegin, int iEnd, int grainSize, const btIParallelForBody& body) BT_OVERRIDE
+	{
+		BT_PROFILE("parallelFor_sequential");
+		body.forLoop(iBegin, iEnd);
+	}
+	virtual btScalar parallelSum(int iBegin, int iEnd, int grainSize, const btIParallelSumBody& body) BT_OVERRIDE
+	{
+		BT_PROFILE("parallelSum_sequential");
+		return body.sumLoop(iBegin, iEnd);
+	}
 };
 
-
 #if BT_USE_OPENMP && BT_THREADSAFE
 ///
 /// btTaskSchedulerOpenMP -- wrapper around OpenMP task scheduler
 ///
 class btTaskSchedulerOpenMP : public btITaskScheduler
 {
-    int m_numThreads;
-public:
-    btTaskSchedulerOpenMP() : btITaskScheduler( "OpenMP" )
-    {
-        m_numThreads = 0;
-    }
-    virtual int getMaxNumThreads() const BT_OVERRIDE
-    {
-        return omp_get_max_threads();
-    }
-    virtual int getNumThreads() const BT_OVERRIDE
-    {
-        return m_numThreads;
-    }
-    virtual void setNumThreads( int numThreads ) BT_OVERRIDE
-    {
-        // With OpenMP, because it is a standard with various implementations, we can't
-        // know for sure if every implementation has the same behavior of destroying all
-        // previous threads when resizing the threadpool
-        m_numThreads = ( std::max )( 1, ( std::min )( int( BT_MAX_THREAD_COUNT ), numThreads ) );
-        omp_set_num_threads( 1 );  // hopefully, all previous threads get destroyed here
-        omp_set_num_threads( m_numThreads );
-        m_savedThreadCounter = 0;
-        if ( m_isActive )
-        {
-            btResetThreadIndexCounter();
-        }
-    }
-    virtual void parallelFor( int iBegin, int iEnd, int grainSize, const btIParallelForBody& body ) BT_OVERRIDE
-    {
-        BT_PROFILE( "parallelFor_OpenMP" );
-        btPushThreadsAreRunning();
-#pragma omp parallel for schedule( static, 1 )
-        for ( int i = iBegin; i < iEnd; i += grainSize )
-        {
-            BT_PROFILE( "OpenMP_forJob" );
-            body.forLoop( i, ( std::min )( i + grainSize, iEnd ) );
-        }
-        btPopThreadsAreRunning();
-    }
-    virtual btScalar parallelSum( int iBegin, int iEnd, int grainSize, const btIParallelSumBody& body ) BT_OVERRIDE
-    {
-        BT_PROFILE( "parallelFor_OpenMP" );
-        btPushThreadsAreRunning();
-        btScalar sum = btScalar( 0 );
-#pragma omp parallel for schedule( static, 1 ) reduction(+:sum)
-        for ( int i = iBegin; i < iEnd; i += grainSize )
-        {
-            BT_PROFILE( "OpenMP_sumJob" );
-            sum += body.sumLoop( i, ( std::min )( i + grainSize, iEnd ) );
-        }
-        btPopThreadsAreRunning();
-        return sum;
-    }
-};
-#endif // #if BT_USE_OPENMP && BT_THREADSAFE
+	int m_numThreads;
 
+public:
+	btTaskSchedulerOpenMP() : btITaskScheduler("OpenMP")
+	{
+		m_numThreads = 0;
+	}
+	virtual int getMaxNumThreads() const BT_OVERRIDE
+	{
+		return omp_get_max_threads();
+	}
+	virtual int getNumThreads() const BT_OVERRIDE
+	{
+		return m_numThreads;
+	}
+	virtual void setNumThreads(int numThreads) BT_OVERRIDE
+	{
+		// With OpenMP, because it is a standard with various implementations, we can't
+		// know for sure if every implementation has the same behavior of destroying all
+		// previous threads when resizing the threadpool
+		m_numThreads = (std::max)(1, (std::min)(int(BT_MAX_THREAD_COUNT), numThreads));
+		omp_set_num_threads(1);  // hopefully, all previous threads get destroyed here
+		omp_set_num_threads(m_numThreads);
+		m_savedThreadCounter = 0;
+		if (m_isActive)
+		{
+			btResetThreadIndexCounter();
+		}
+	}
+	virtual void parallelFor(int iBegin, int iEnd, int grainSize, const btIParallelForBody& body) BT_OVERRIDE
+	{
+		BT_PROFILE("parallelFor_OpenMP");
+		btPushThreadsAreRunning();
+#pragma omp parallel for schedule(static, 1)
+		for (int i = iBegin; i < iEnd; i += grainSize)
+		{
+			BT_PROFILE("OpenMP_forJob");
+			body.forLoop(i, (std::min)(i + grainSize, iEnd));
+		}
+		btPopThreadsAreRunning();
+	}
+	virtual btScalar parallelSum(int iBegin, int iEnd, int grainSize, const btIParallelSumBody& body) BT_OVERRIDE
+	{
+		BT_PROFILE("parallelFor_OpenMP");
+		btPushThreadsAreRunning();
+		btScalar sum = btScalar(0);
+#pragma omp parallel for schedule(static, 1) reduction(+ \
+													   : sum)
+		for (int i = iBegin; i < iEnd; i += grainSize)
+		{
+			BT_PROFILE("OpenMP_sumJob");
+			sum += body.sumLoop(i, (std::min)(i + grainSize, iEnd));
+		}
+		btPopThreadsAreRunning();
+		return sum;
+	}
+};
+#endif  // #if BT_USE_OPENMP && BT_THREADSAFE
 
 #if BT_USE_TBB && BT_THREADSAFE
 ///
@@ -575,96 +557,94 @@ public:
 ///
 class btTaskSchedulerTBB : public btITaskScheduler
 {
-    int m_numThreads;
-    tbb::task_scheduler_init* m_tbbSchedulerInit;
+	int m_numThreads;
+	tbb::task_scheduler_init* m_tbbSchedulerInit;
 
 public:
-    btTaskSchedulerTBB() : btITaskScheduler( "IntelTBB" )
-    {
-        m_numThreads = 0;
-        m_tbbSchedulerInit = NULL;
-    }
-    ~btTaskSchedulerTBB()
-    {
-        if ( m_tbbSchedulerInit )
-        {
-            delete m_tbbSchedulerInit;
-            m_tbbSchedulerInit = NULL;
-        }
-    }
+	btTaskSchedulerTBB() : btITaskScheduler("IntelTBB")
+	{
+		m_numThreads = 0;
+		m_tbbSchedulerInit = NULL;
+	}
+	~btTaskSchedulerTBB()
+	{
+		if (m_tbbSchedulerInit)
+		{
+			delete m_tbbSchedulerInit;
+			m_tbbSchedulerInit = NULL;
+		}
+	}
 
-    virtual int getMaxNumThreads() const BT_OVERRIDE
-    {
-        return tbb::task_scheduler_init::default_num_threads();
-    }
-    virtual int getNumThreads() const BT_OVERRIDE
-    {
-        return m_numThreads;
-    }
-    virtual void setNumThreads( int numThreads ) BT_OVERRIDE
-    {
-        m_numThreads = ( std::max )( 1, ( std::min )( int(BT_MAX_THREAD_COUNT), numThreads ) );
-        if ( m_tbbSchedulerInit )
-        {
-            // destroys all previous threads
-            delete m_tbbSchedulerInit;
-            m_tbbSchedulerInit = NULL;
-        }
-        m_tbbSchedulerInit = new tbb::task_scheduler_init( m_numThreads );
-        m_savedThreadCounter = 0;
-        if ( m_isActive )
-        {
-            btResetThreadIndexCounter();
-        }
-    }
-    struct ForBodyAdapter
-    {
-        const btIParallelForBody* mBody;
+	virtual int getMaxNumThreads() const BT_OVERRIDE
+	{
+		return tbb::task_scheduler_init::default_num_threads();
+	}
+	virtual int getNumThreads() const BT_OVERRIDE
+	{
+		return m_numThreads;
+	}
+	virtual void setNumThreads(int numThreads) BT_OVERRIDE
+	{
+		m_numThreads = (std::max)(1, (std::min)(int(BT_MAX_THREAD_COUNT), numThreads));
+		if (m_tbbSchedulerInit)
+		{
+			// destroys all previous threads
+			delete m_tbbSchedulerInit;
+			m_tbbSchedulerInit = NULL;
+		}
+		m_tbbSchedulerInit = new tbb::task_scheduler_init(m_numThreads);
+		m_savedThreadCounter = 0;
+		if (m_isActive)
+		{
+			btResetThreadIndexCounter();
+		}
+	}
+	struct ForBodyAdapter
+	{
+		const btIParallelForBody* mBody;
 
-        ForBodyAdapter( const btIParallelForBody* body ) : mBody( body ) {}
-        void operator()( const tbb::blocked_range<int>& range ) const
-        {
-            BT_PROFILE( "TBB_forJob" );
-            mBody->forLoop( range.begin(), range.end() );
-        }
-    };
-    virtual void parallelFor( int iBegin, int iEnd, int grainSize, const btIParallelForBody& body ) BT_OVERRIDE
-    {
-        BT_PROFILE( "parallelFor_TBB" );
-        ForBodyAdapter tbbBody( &body );
-        btPushThreadsAreRunning();
-        tbb::parallel_for( tbb::blocked_range<int>( iBegin, iEnd, grainSize ),
-            tbbBody,
-            tbb::simple_partitioner()
-        );
-        btPopThreadsAreRunning();
-    }
-    struct SumBodyAdapter
-    {
-        const btIParallelSumBody* mBody;
-        btScalar mSum;
+		ForBodyAdapter(const btIParallelForBody* body) : mBody(body) {}
+		void operator()(const tbb::blocked_range<int>& range) const
+		{
+			BT_PROFILE("TBB_forJob");
+			mBody->forLoop(range.begin(), range.end());
+		}
+	};
+	virtual void parallelFor(int iBegin, int iEnd, int grainSize, const btIParallelForBody& body) BT_OVERRIDE
+	{
+		BT_PROFILE("parallelFor_TBB");
+		ForBodyAdapter tbbBody(&body);
+		btPushThreadsAreRunning();
+		tbb::parallel_for(tbb::blocked_range<int>(iBegin, iEnd, grainSize),
+						  tbbBody,
+						  tbb::simple_partitioner());
+		btPopThreadsAreRunning();
+	}
+	struct SumBodyAdapter
+	{
+		const btIParallelSumBody* mBody;
+		btScalar mSum;
 
-        SumBodyAdapter( const btIParallelSumBody* body ) : mBody( body ), mSum( btScalar( 0 ) ) {}
-        SumBodyAdapter( const SumBodyAdapter& src, tbb::split ) : mBody( src.mBody ), mSum( btScalar( 0 ) ) {}
-        void join( const SumBodyAdapter& src ) { mSum += src.mSum; }
-        void operator()( const tbb::blocked_range<int>& range )
-        {
-            BT_PROFILE( "TBB_sumJob" );
-            mSum += mBody->sumLoop( range.begin(), range.end() );
-        }
-    };
-    virtual btScalar parallelSum( int iBegin, int iEnd, int grainSize, const btIParallelSumBody& body ) BT_OVERRIDE
-    {
-        BT_PROFILE( "parallelSum_TBB" );
-        SumBodyAdapter tbbBody( &body );
-        btPushThreadsAreRunning();
-        tbb::parallel_deterministic_reduce( tbb::blocked_range<int>( iBegin, iEnd, grainSize ), tbbBody );
-        btPopThreadsAreRunning();
-        return tbbBody.mSum;
-    }
+		SumBodyAdapter(const btIParallelSumBody* body) : mBody(body), mSum(btScalar(0)) {}
+		SumBodyAdapter(const SumBodyAdapter& src, tbb::split) : mBody(src.mBody), mSum(btScalar(0)) {}
+		void join(const SumBodyAdapter& src) { mSum += src.mSum; }
+		void operator()(const tbb::blocked_range<int>& range)
+		{
+			BT_PROFILE("TBB_sumJob");
+			mSum += mBody->sumLoop(range.begin(), range.end());
+		}
+	};
+	virtual btScalar parallelSum(int iBegin, int iEnd, int grainSize, const btIParallelSumBody& body) BT_OVERRIDE
+	{
+		BT_PROFILE("parallelSum_TBB");
+		SumBodyAdapter tbbBody(&body);
+		btPushThreadsAreRunning();
+		tbb::parallel_deterministic_reduce(tbb::blocked_range<int>(iBegin, iEnd, grainSize), tbbBody);
+		btPopThreadsAreRunning();
+		return tbbBody.mSum;
+	}
 };
-#endif // #if BT_USE_TBB && BT_THREADSAFE
-
+#endif  // #if BT_USE_TBB && BT_THREADSAFE
 
 #if BT_USE_PPL && BT_THREADSAFE
 ///
@@ -672,148 +652,141 @@ public:
 ///
 class btTaskSchedulerPPL : public btITaskScheduler
 {
-    int m_numThreads;
-    concurrency::combinable<btScalar> m_sum;  // for parallelSum
+	int m_numThreads;
+	concurrency::combinable<btScalar> m_sum;  // for parallelSum
 public:
-    btTaskSchedulerPPL() : btITaskScheduler( "PPL" )
-    {
-        m_numThreads = 0;
-    }
-    virtual int getMaxNumThreads() const BT_OVERRIDE
-    {
-        return concurrency::GetProcessorCount();
-    }
-    virtual int getNumThreads() const BT_OVERRIDE
-    {
-        return m_numThreads;
-    }
-    virtual void setNumThreads( int numThreads ) BT_OVERRIDE
-    {
-        // capping the thread count for PPL due to a thread-index issue
-        const int maxThreadCount = (std::min)(int(BT_MAX_THREAD_COUNT), 31);
-        m_numThreads = ( std::max )( 1, ( std::min )( maxThreadCount, numThreads ) );
-        using namespace concurrency;
-        if ( CurrentScheduler::Id() != -1 )
-        {
-            CurrentScheduler::Detach();
-        }
-        SchedulerPolicy policy;
-        {
-            // PPL seems to destroy threads when threadpool is shrunk, but keeps reusing old threads
-            // force it to destroy old threads
-            policy.SetConcurrencyLimits( 1, 1 );
-            CurrentScheduler::Create( policy );
-            CurrentScheduler::Detach();
-        }
-        policy.SetConcurrencyLimits( m_numThreads, m_numThreads );
-        CurrentScheduler::Create( policy );
-        m_savedThreadCounter = 0;
-        if ( m_isActive )
-        {
-            btResetThreadIndexCounter();
-        }
-    }
-    struct ForBodyAdapter
-    {
-        const btIParallelForBody* mBody;
-        int mGrainSize;
-        int mIndexEnd;
+	btTaskSchedulerPPL() : btITaskScheduler("PPL")
+	{
+		m_numThreads = 0;
+	}
+	virtual int getMaxNumThreads() const BT_OVERRIDE
+	{
+		return concurrency::GetProcessorCount();
+	}
+	virtual int getNumThreads() const BT_OVERRIDE
+	{
+		return m_numThreads;
+	}
+	virtual void setNumThreads(int numThreads) BT_OVERRIDE
+	{
+		// capping the thread count for PPL due to a thread-index issue
+		const int maxThreadCount = (std::min)(int(BT_MAX_THREAD_COUNT), 31);
+		m_numThreads = (std::max)(1, (std::min)(maxThreadCount, numThreads));
+		using namespace concurrency;
+		if (CurrentScheduler::Id() != -1)
+		{
+			CurrentScheduler::Detach();
+		}
+		SchedulerPolicy policy;
+		{
+			// PPL seems to destroy threads when threadpool is shrunk, but keeps reusing old threads
+			// force it to destroy old threads
+			policy.SetConcurrencyLimits(1, 1);
+			CurrentScheduler::Create(policy);
+			CurrentScheduler::Detach();
+		}
+		policy.SetConcurrencyLimits(m_numThreads, m_numThreads);
+		CurrentScheduler::Create(policy);
+		m_savedThreadCounter = 0;
+		if (m_isActive)
+		{
+			btResetThreadIndexCounter();
+		}
+	}
+	struct ForBodyAdapter
+	{
+		const btIParallelForBody* mBody;
+		int mGrainSize;
+		int mIndexEnd;
 
-        ForBodyAdapter( const btIParallelForBody* body, int grainSize, int end ) : mBody( body ), mGrainSize( grainSize ), mIndexEnd( end ) {}
-        void operator()( int i ) const
-        {
-            BT_PROFILE( "PPL_forJob" );
-            mBody->forLoop( i, ( std::min )( i + mGrainSize, mIndexEnd ) );
-        }
-    };
-    virtual void parallelFor( int iBegin, int iEnd, int grainSize, const btIParallelForBody& body ) BT_OVERRIDE
-    {
-        BT_PROFILE( "parallelFor_PPL" );
-        // PPL dispatch
-        ForBodyAdapter pplBody( &body, grainSize, iEnd );
-        btPushThreadsAreRunning();
-        // note: MSVC 2010 doesn't support partitioner args, so avoid them
-        concurrency::parallel_for( iBegin,
-            iEnd,
-            grainSize,
-            pplBody
-        );
-        btPopThreadsAreRunning();
-    }
-    struct SumBodyAdapter
-    {
-        const btIParallelSumBody* mBody;
-        concurrency::combinable<btScalar>* mSum;
-        int mGrainSize;
-        int mIndexEnd;
+		ForBodyAdapter(const btIParallelForBody* body, int grainSize, int end) : mBody(body), mGrainSize(grainSize), mIndexEnd(end) {}
+		void operator()(int i) const
+		{
+			BT_PROFILE("PPL_forJob");
+			mBody->forLoop(i, (std::min)(i + mGrainSize, mIndexEnd));
+		}
+	};
+	virtual void parallelFor(int iBegin, int iEnd, int grainSize, const btIParallelForBody& body) BT_OVERRIDE
+	{
+		BT_PROFILE("parallelFor_PPL");
+		// PPL dispatch
+		ForBodyAdapter pplBody(&body, grainSize, iEnd);
+		btPushThreadsAreRunning();
+		// note: MSVC 2010 doesn't support partitioner args, so avoid them
+		concurrency::parallel_for(iBegin,
+								  iEnd,
+								  grainSize,
+								  pplBody);
+		btPopThreadsAreRunning();
+	}
+	struct SumBodyAdapter
+	{
+		const btIParallelSumBody* mBody;
+		concurrency::combinable<btScalar>* mSum;
+		int mGrainSize;
+		int mIndexEnd;
 
-        SumBodyAdapter( const btIParallelSumBody* body, concurrency::combinable<btScalar>* sum, int grainSize, int end ) : mBody( body ), mSum(sum), mGrainSize( grainSize ), mIndexEnd( end ) {}
-        void operator()( int i ) const
-        {
-            BT_PROFILE( "PPL_sumJob" );
-            mSum->local() += mBody->sumLoop( i, ( std::min )( i + mGrainSize, mIndexEnd ) );
-        }
-    };
-    static btScalar sumFunc( btScalar a, btScalar b ) { return a + b; }
-    virtual btScalar parallelSum( int iBegin, int iEnd, int grainSize, const btIParallelSumBody& body ) BT_OVERRIDE
-    {
-        BT_PROFILE( "parallelSum_PPL" );
-        m_sum.clear();
-        SumBodyAdapter pplBody( &body, &m_sum, grainSize, iEnd );
-        btPushThreadsAreRunning();
-        // note: MSVC 2010 doesn't support partitioner args, so avoid them
-        concurrency::parallel_for( iBegin,
-            iEnd,
-            grainSize,
-            pplBody
-        );
-        btPopThreadsAreRunning();
-        return m_sum.combine( sumFunc );
-    }
+		SumBodyAdapter(const btIParallelSumBody* body, concurrency::combinable<btScalar>* sum, int grainSize, int end) : mBody(body), mSum(sum), mGrainSize(grainSize), mIndexEnd(end) {}
+		void operator()(int i) const
+		{
+			BT_PROFILE("PPL_sumJob");
+			mSum->local() += mBody->sumLoop(i, (std::min)(i + mGrainSize, mIndexEnd));
+		}
+	};
+	static btScalar sumFunc(btScalar a, btScalar b) { return a + b; }
+	virtual btScalar parallelSum(int iBegin, int iEnd, int grainSize, const btIParallelSumBody& body) BT_OVERRIDE
+	{
+		BT_PROFILE("parallelSum_PPL");
+		m_sum.clear();
+		SumBodyAdapter pplBody(&body, &m_sum, grainSize, iEnd);
+		btPushThreadsAreRunning();
+		// note: MSVC 2010 doesn't support partitioner args, so avoid them
+		concurrency::parallel_for(iBegin,
+								  iEnd,
+								  grainSize,
+								  pplBody);
+		btPopThreadsAreRunning();
+		return m_sum.combine(sumFunc);
+	}
 };
-#endif // #if BT_USE_PPL && BT_THREADSAFE
-
+#endif  // #if BT_USE_PPL && BT_THREADSAFE
 
 // create a non-threaded task scheduler (always available)
 btITaskScheduler* btGetSequentialTaskScheduler()
 {
-    static btTaskSchedulerSequential sTaskScheduler;
-    return &sTaskScheduler;
+	static btTaskSchedulerSequential sTaskScheduler;
+	return &sTaskScheduler;
 }
 
-
 // create an OpenMP task scheduler (if available, otherwise returns null)
 btITaskScheduler* btGetOpenMPTaskScheduler()
 {
 #if BT_USE_OPENMP && BT_THREADSAFE
-    static btTaskSchedulerOpenMP sTaskScheduler;
-    return &sTaskScheduler;
+	static btTaskSchedulerOpenMP sTaskScheduler;
+	return &sTaskScheduler;
 #else
-    return NULL;
+	return NULL;
 #endif
 }
 
-
 // create an Intel TBB task scheduler (if available, otherwise returns null)
 btITaskScheduler* btGetTBBTaskScheduler()
 {
 #if BT_USE_TBB && BT_THREADSAFE
-    static btTaskSchedulerTBB sTaskScheduler;
-    return &sTaskScheduler;
+	static btTaskSchedulerTBB sTaskScheduler;
+	return &sTaskScheduler;
 #else
-    return NULL;
+	return NULL;
 #endif
 }
 
-
 // create a PPL task scheduler (if available, otherwise returns null)
 btITaskScheduler* btGetPPLTaskScheduler()
 {
 #if BT_USE_PPL && BT_THREADSAFE
-    static btTaskSchedulerPPL sTaskScheduler;
-    return &sTaskScheduler;
+	static btTaskSchedulerPPL sTaskScheduler;
+	return &sTaskScheduler;
 #else
-    return NULL;
+	return NULL;
 #endif
 }
-
diff --git a/thirdparty/bullet/LinearMath/btThreads.h b/thirdparty/bullet/LinearMath/btThreads.h
index 921fd088c0b..b2227e17247 100644
--- a/thirdparty/bullet/LinearMath/btThreads.h
+++ b/thirdparty/bullet/LinearMath/btThreads.h
@@ -12,14 +12,12 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #ifndef BT_THREADS_H
 #define BT_THREADS_H
 
-#include "btScalar.h" // has definitions like SIMD_FORCE_INLINE
+#include "btScalar.h"  // has definitions like SIMD_FORCE_INLINE
 
-#if defined (_MSC_VER) && _MSC_VER >= 1600
+#if defined(_MSC_VER) && _MSC_VER >= 1600
 // give us a compile error if any signatures of overriden methods is changed
 #define BT_OVERRIDE override
 #endif
@@ -36,7 +34,7 @@ const unsigned int BT_MAX_THREAD_COUNT = 64;  // only if BT_THREADSAFE is 1
 bool btIsMainThread();
 bool btThreadsAreRunning();
 unsigned int btGetCurrentThreadIndex();
-void btResetThreadIndexCounter(); // notify that all worker threads have been destroyed
+void btResetThreadIndexCounter();  // notify that all worker threads have been destroyed
 
 ///
 /// btSpinMutex -- lightweight spin-mutex implemented with atomic ops, never puts
@@ -46,19 +44,18 @@ void btResetThreadIndexCounter(); // notify that all worker threads have been de
 ///
 class btSpinMutex
 {
-    int mLock;
+	int mLock;
 
 public:
-    btSpinMutex()
-    {
-        mLock = 0;
-    }
-    void lock();
-    void unlock();
-    bool tryLock();
+	btSpinMutex()
+	{
+		mLock = 0;
+	}
+	void lock();
+	void unlock();
+	bool tryLock();
 };
 
-
 //
 // NOTE: btMutex* is for internal Bullet use only
 //
@@ -70,43 +67,42 @@ public:
 // of bad because if you call any of these functions from external code
 // (where BT_THREADSAFE is undefined) you will get unexpected race conditions.
 //
-SIMD_FORCE_INLINE void btMutexLock( btSpinMutex* mutex )
+SIMD_FORCE_INLINE void btMutexLock(btSpinMutex* mutex)
 {
 #if BT_THREADSAFE
-    mutex->lock();
+	mutex->lock();
 #else
-    (void)mutex;
-#endif // #if BT_THREADSAFE
+	(void)mutex;
+#endif  // #if BT_THREADSAFE
 }
 
-SIMD_FORCE_INLINE void btMutexUnlock( btSpinMutex* mutex )
+SIMD_FORCE_INLINE void btMutexUnlock(btSpinMutex* mutex)
 {
 #if BT_THREADSAFE
-    mutex->unlock();
+	mutex->unlock();
 #else
-    (void)mutex;
-#endif // #if BT_THREADSAFE
+	(void)mutex;
+#endif  // #if BT_THREADSAFE
 }
 
-SIMD_FORCE_INLINE bool btMutexTryLock( btSpinMutex* mutex )
+SIMD_FORCE_INLINE bool btMutexTryLock(btSpinMutex* mutex)
 {
 #if BT_THREADSAFE
-    return mutex->tryLock();
+	return mutex->tryLock();
 #else
-    (void)mutex;
-    return true;
-#endif // #if BT_THREADSAFE
+	(void)mutex;
+	return true;
+#endif  // #if BT_THREADSAFE
 }
 
-
 //
 // btIParallelForBody -- subclass this to express work that can be done in parallel
 //
 class btIParallelForBody
 {
 public:
-    virtual ~btIParallelForBody() {}
-    virtual void forLoop( int iBegin, int iEnd ) const = 0;
+	virtual ~btIParallelForBody() {}
+	virtual void forLoop(int iBegin, int iEnd) const = 0;
 };
 
 //
@@ -116,8 +112,8 @@ public:
 class btIParallelSumBody
 {
 public:
-    virtual ~btIParallelSumBody() {}
-    virtual btScalar sumLoop( int iBegin, int iEnd ) const = 0;
+	virtual ~btIParallelSumBody() {}
+	virtual btScalar sumLoop(int iBegin, int iEnd) const = 0;
 };
 
 //
@@ -127,30 +123,30 @@ public:
 class btITaskScheduler
 {
 public:
-    btITaskScheduler( const char* name );
-    virtual ~btITaskScheduler() {}
-    const char* getName() const { return m_name; }
+	btITaskScheduler(const char* name);
+	virtual ~btITaskScheduler() {}
+	const char* getName() const { return m_name; }
 
-    virtual int getMaxNumThreads() const = 0;
-    virtual int getNumThreads() const = 0;
-    virtual void setNumThreads( int numThreads ) = 0;
-    virtual void parallelFor( int iBegin, int iEnd, int grainSize, const btIParallelForBody& body ) = 0;
-    virtual btScalar parallelSum( int iBegin, int iEnd, int grainSize, const btIParallelSumBody& body ) = 0;
-    virtual void sleepWorkerThreadsHint() {}  // hint the task scheduler that we may not be using these threads for a little while
+	virtual int getMaxNumThreads() const = 0;
+	virtual int getNumThreads() const = 0;
+	virtual void setNumThreads(int numThreads) = 0;
+	virtual void parallelFor(int iBegin, int iEnd, int grainSize, const btIParallelForBody& body) = 0;
+	virtual btScalar parallelSum(int iBegin, int iEnd, int grainSize, const btIParallelSumBody& body) = 0;
+	virtual void sleepWorkerThreadsHint() {}  // hint the task scheduler that we may not be using these threads for a little while
 
-    // internal use only
-    virtual void activate();
-    virtual void deactivate();
+	// internal use only
+	virtual void activate();
+	virtual void deactivate();
 
 protected:
-    const char* m_name;
-    unsigned int m_savedThreadCounter;
-    bool m_isActive;
+	const char* m_name;
+	unsigned int m_savedThreadCounter;
+	bool m_isActive;
 };
 
 // set the task scheduler to use for all calls to btParallelFor()
 // NOTE: you must set this prior to using any of the multi-threaded "Mt" classes
-void btSetTaskScheduler( btITaskScheduler* ts );
+void btSetTaskScheduler(btITaskScheduler* ts);
 
 // get the current task scheduler
 btITaskScheduler* btGetTaskScheduler();
@@ -172,11 +168,10 @@ btITaskScheduler* btGetPPLTaskScheduler();
 
 // btParallelFor -- call this to dispatch work like a for-loop
 //                 (iterations may be done out of order, so no dependencies are allowed)
-void btParallelFor( int iBegin, int iEnd, int grainSize, const btIParallelForBody& body );
+void btParallelFor(int iBegin, int iEnd, int grainSize, const btIParallelForBody& body);
 
 // btParallelSum -- call this to dispatch work like a for-loop, returns the sum of all iterations
 //                 (iterations may be done out of order, so no dependencies are allowed)
-btScalar btParallelSum( int iBegin, int iEnd, int grainSize, const btIParallelSumBody& body );
-
+btScalar btParallelSum(int iBegin, int iEnd, int grainSize, const btIParallelSumBody& body);
 
 #endif
diff --git a/thirdparty/bullet/LinearMath/btTransform.h b/thirdparty/bullet/LinearMath/btTransform.h
index d4f939a5d99..6f2f99818cf 100644
--- a/thirdparty/bullet/LinearMath/btTransform.h
+++ b/thirdparty/bullet/LinearMath/btTransform.h
@@ -12,12 +12,9 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #ifndef BT_TRANSFORM_H
 #define BT_TRANSFORM_H
 
-
 #include "btMatrix3x3.h"
 
 #ifdef BT_USE_DOUBLE_PRECISION
@@ -26,46 +23,45 @@ subject to the following restrictions:
 #define btTransformData btTransformFloatData
 #endif
 
-
-
-
 /**@brief The btTransform class supports rigid transforms with only translation and rotation and no scaling/shear.
  *It can be used in combination with btVector3, btQuaternion and btMatrix3x3 linear algebra classes. */
-ATTRIBUTE_ALIGNED16(class) btTransform {
-	
-  ///Storage for the rotation
+ATTRIBUTE_ALIGNED16(class)
+btTransform
+{
+	///Storage for the rotation
 	btMatrix3x3 m_basis;
-  ///Storage for the translation
-	btVector3   m_origin;
+	///Storage for the translation
+	btVector3 m_origin;
 
 public:
-	
-  /**@brief No initialization constructor */
+	/**@brief No initialization constructor */
 	btTransform() {}
-  /**@brief Constructor from btQuaternion (optional btVector3 )
+	/**@brief Constructor from btQuaternion (optional btVector3 )
    * @param q Rotation from quaternion 
    * @param c Translation from Vector (default 0,0,0) */
-	explicit SIMD_FORCE_INLINE btTransform(const btQuaternion& q, 
-		const btVector3& c = btVector3(btScalar(0), btScalar(0), btScalar(0))) 
+	explicit SIMD_FORCE_INLINE btTransform(const btQuaternion& q,
+										   const btVector3& c = btVector3(btScalar(0), btScalar(0), btScalar(0)))
 		: m_basis(q),
-		m_origin(c)
-	{}
-
-  /**@brief Constructor from btMatrix3x3 (optional btVector3)
-   * @param b Rotation from Matrix 
-   * @param c Translation from Vector default (0,0,0)*/
-	explicit SIMD_FORCE_INLINE btTransform(const btMatrix3x3& b, 
-		const btVector3& c = btVector3(btScalar(0), btScalar(0), btScalar(0)))
-		: m_basis(b),
-		m_origin(c)
-	{}
-  /**@brief Copy constructor */
-	SIMD_FORCE_INLINE btTransform (const btTransform& other)
-		: m_basis(other.m_basis),
-		m_origin(other.m_origin)
+		  m_origin(c)
 	{
 	}
-  /**@brief Assignment Operator */
+
+	/**@brief Constructor from btMatrix3x3 (optional btVector3)
+   * @param b Rotation from Matrix 
+   * @param c Translation from Vector default (0,0,0)*/
+	explicit SIMD_FORCE_INLINE btTransform(const btMatrix3x3& b,
+										   const btVector3& c = btVector3(btScalar(0), btScalar(0), btScalar(0)))
+		: m_basis(b),
+		  m_origin(c)
+	{
+	}
+	/**@brief Copy constructor */
+	SIMD_FORCE_INLINE btTransform(const btTransform& other)
+		: m_basis(other.m_basis),
+		  m_origin(other.m_origin)
+	{
+	}
+	/**@brief Assignment Operator */
 	SIMD_FORCE_INLINE btTransform& operator=(const btTransform& other)
 	{
 		m_basis = other.m_basis;
@@ -73,70 +69,70 @@ public:
 		return *this;
 	}
 
-
-  /**@brief Set the current transform as the value of the product of two transforms
+	/**@brief Set the current transform as the value of the product of two transforms
    * @param t1 Transform 1
    * @param t2 Transform 2
    * This = Transform1 * Transform2 */
-		SIMD_FORCE_INLINE void mult(const btTransform& t1, const btTransform& t2) {
-			m_basis = t1.m_basis * t2.m_basis;
-			m_origin = t1(t2.m_origin);
-		}
+	SIMD_FORCE_INLINE void mult(const btTransform& t1, const btTransform& t2)
+	{
+		m_basis = t1.m_basis * t2.m_basis;
+		m_origin = t1(t2.m_origin);
+	}
 
-/*		void multInverseLeft(const btTransform& t1, const btTransform& t2) {
+	/*		void multInverseLeft(const btTransform& t1, const btTransform& t2) {
 			btVector3 v = t2.m_origin - t1.m_origin;
 			m_basis = btMultTransposeLeft(t1.m_basis, t2.m_basis);
 			m_origin = v * t1.m_basis;
 		}
 		*/
 
-/**@brief Return the transform of the vector */
+	/**@brief Return the transform of the vector */
 	SIMD_FORCE_INLINE btVector3 operator()(const btVector3& x) const
 	{
-        return x.dot3(m_basis[0], m_basis[1], m_basis[2]) + m_origin;
+		return x.dot3(m_basis[0], m_basis[1], m_basis[2]) + m_origin;
 	}
 
-  /**@brief Return the transform of the vector */
+	/**@brief Return the transform of the vector */
 	SIMD_FORCE_INLINE btVector3 operator*(const btVector3& x) const
 	{
 		return (*this)(x);
 	}
 
-  /**@brief Return the transform of the btQuaternion */
+	/**@brief Return the transform of the btQuaternion */
 	SIMD_FORCE_INLINE btQuaternion operator*(const btQuaternion& q) const
 	{
 		return getRotation() * q;
 	}
 
-  /**@brief Return the basis matrix for the rotation */
-	SIMD_FORCE_INLINE btMatrix3x3&       getBasis()          { return m_basis; }
-  /**@brief Return the basis matrix for the rotation */
-	SIMD_FORCE_INLINE const btMatrix3x3& getBasis()    const { return m_basis; }
+	/**@brief Return the basis matrix for the rotation */
+	SIMD_FORCE_INLINE btMatrix3x3& getBasis() { return m_basis; }
+	/**@brief Return the basis matrix for the rotation */
+	SIMD_FORCE_INLINE const btMatrix3x3& getBasis() const { return m_basis; }
 
-  /**@brief Return the origin vector translation */
-	SIMD_FORCE_INLINE btVector3&         getOrigin()         { return m_origin; }
-  /**@brief Return the origin vector translation */
-	SIMD_FORCE_INLINE const btVector3&   getOrigin()   const { return m_origin; }
+	/**@brief Return the origin vector translation */
+	SIMD_FORCE_INLINE btVector3& getOrigin() { return m_origin; }
+	/**@brief Return the origin vector translation */
+	SIMD_FORCE_INLINE const btVector3& getOrigin() const { return m_origin; }
 
-  /**@brief Return a quaternion representing the rotation */
-	btQuaternion getRotation() const { 
+	/**@brief Return a quaternion representing the rotation */
+	btQuaternion getRotation() const
+	{
 		btQuaternion q;
 		m_basis.getRotation(q);
 		return q;
 	}
-	
-	
-  /**@brief Set from an array 
+
+	/**@brief Set from an array 
    * @param m A pointer to a 16 element array (12 rotation(row major padded on the right by 1), and 3 translation */
-	void setFromOpenGLMatrix(const btScalar *m)
+	void setFromOpenGLMatrix(const btScalar* m)
 	{
 		m_basis.setFromOpenGLSubMatrix(m);
-		m_origin.setValue(m[12],m[13],m[14]);
+		m_origin.setValue(m[12], m[13], m[14]);
 	}
 
-  /**@brief Fill an array representation
+	/**@brief Fill an array representation
    * @param m A pointer to a 16 element array (12 rotation(row major padded on the right by 1), and 3 translation */
-	void getOpenGLMatrix(btScalar *m) const 
+	void getOpenGLMatrix(btScalar * m) const
 	{
 		m_basis.getOpenGLSubMatrix(m);
 		m[12] = m_origin.x();
@@ -145,80 +141,76 @@ public:
 		m[15] = btScalar(1.0);
 	}
 
-  /**@brief Set the translational element
+	/**@brief Set the translational element
    * @param origin The vector to set the translation to */
-	SIMD_FORCE_INLINE void setOrigin(const btVector3& origin) 
-	{ 
+	SIMD_FORCE_INLINE void setOrigin(const btVector3& origin)
+	{
 		m_origin = origin;
 	}
 
 	SIMD_FORCE_INLINE btVector3 invXform(const btVector3& inVec) const;
 
-
-  /**@brief Set the rotational element by btMatrix3x3 */
+	/**@brief Set the rotational element by btMatrix3x3 */
 	SIMD_FORCE_INLINE void setBasis(const btMatrix3x3& basis)
-	{ 
+	{
 		m_basis = basis;
 	}
 
-  /**@brief Set the rotational element by btQuaternion */
+	/**@brief Set the rotational element by btQuaternion */
 	SIMD_FORCE_INLINE void setRotation(const btQuaternion& q)
 	{
 		m_basis.setRotation(q);
 	}
 
-
-  /**@brief Set this transformation to the identity */
+	/**@brief Set this transformation to the identity */
 	void setIdentity()
 	{
 		m_basis.setIdentity();
 		m_origin.setValue(btScalar(0.0), btScalar(0.0), btScalar(0.0));
 	}
 
-  /**@brief Multiply this Transform by another(this = this * another) 
+	/**@brief Multiply this Transform by another(this = this * another) 
    * @param t The other transform */
-	btTransform& operator*=(const btTransform& t) 
+	btTransform& operator*=(const btTransform& t)
 	{
 		m_origin += m_basis * t.m_origin;
 		m_basis *= t.m_basis;
 		return *this;
 	}
 
-  /**@brief Return the inverse of this transform */
+	/**@brief Return the inverse of this transform */
 	btTransform inverse() const
-	{ 
+	{
 		btMatrix3x3 inv = m_basis.transpose();
 		return btTransform(inv, inv * -m_origin);
 	}
 
-  /**@brief Return the inverse of this transform times the other transform
+	/**@brief Return the inverse of this transform times the other transform
    * @param t The other transform 
    * return this.inverse() * the other */
-	btTransform inverseTimes(const btTransform& t) const;  
+	btTransform inverseTimes(const btTransform& t) const;
 
-  /**@brief Return the product of this transform and the other */
+	/**@brief Return the product of this transform and the other */
 	btTransform operator*(const btTransform& t) const;
 
-  /**@brief Return an identity transform */
-	static const btTransform&	getIdentity()
+	/**@brief Return an identity transform */
+	static const btTransform& getIdentity()
 	{
 		static const btTransform identityTransform(btMatrix3x3::getIdentity());
 		return identityTransform;
 	}
 
-	void	serialize(struct	btTransformData& dataOut) const;
+	void serialize(struct btTransformData & dataOut) const;
 
-	void	serializeFloat(struct	btTransformFloatData& dataOut) const;
+	void serializeFloat(struct btTransformFloatData & dataOut) const;
 
-	void	deSerialize(const struct	btTransformData& dataIn);
+	void deSerialize(const struct btTransformData& dataIn);
 
-	void	deSerializeDouble(const struct	btTransformDoubleData& dataIn);
-
-	void	deSerializeFloat(const struct	btTransformFloatData& dataIn);
+	void deSerializeDouble(const struct btTransformDoubleData& dataIn);
 
+	void deSerializeFloat(const struct btTransformFloatData& dataIn);
 };
 
-
 SIMD_FORCE_INLINE btVector3
 btTransform::invXform(const btVector3& inVec) const
 {
@@ -226,80 +218,69 @@ btTransform::invXform(const btVector3& inVec) const
 	return (m_basis.transpose() * v);
 }
 
-SIMD_FORCE_INLINE btTransform 
-btTransform::inverseTimes(const btTransform& t) const  
+SIMD_FORCE_INLINE btTransform
+btTransform::inverseTimes(const btTransform& t) const
 {
 	btVector3 v = t.getOrigin() - m_origin;
-		return btTransform(m_basis.transposeTimes(t.m_basis),
-			v * m_basis);
+	return btTransform(m_basis.transposeTimes(t.m_basis),
+					   v * m_basis);
 }
 
-SIMD_FORCE_INLINE btTransform 
-btTransform::operator*(const btTransform& t) const
+SIMD_FORCE_INLINE btTransform
+	btTransform::operator*(const btTransform& t) const
 {
-	return btTransform(m_basis * t.m_basis, 
-		(*this)(t.m_origin));
+	return btTransform(m_basis * t.m_basis,
+					   (*this)(t.m_origin));
 }
 
 /**@brief Test if two transforms have all elements equal */
 SIMD_FORCE_INLINE bool operator==(const btTransform& t1, const btTransform& t2)
 {
-   return ( t1.getBasis()  == t2.getBasis() &&
-            t1.getOrigin() == t2.getOrigin() );
+	return (t1.getBasis() == t2.getBasis() &&
+			t1.getOrigin() == t2.getOrigin());
 }
 
-
 ///for serialization
-struct	btTransformFloatData
+struct btTransformFloatData
 {
-	btMatrix3x3FloatData	m_basis;
-	btVector3FloatData	m_origin;
+	btMatrix3x3FloatData m_basis;
+	btVector3FloatData m_origin;
 };
 
-struct	btTransformDoubleData
+struct btTransformDoubleData
 {
-	btMatrix3x3DoubleData	m_basis;
-	btVector3DoubleData	m_origin;
+	btMatrix3x3DoubleData m_basis;
+	btVector3DoubleData m_origin;
 };
 
-
-
-SIMD_FORCE_INLINE	void	btTransform::serialize(btTransformData& dataOut) const
+SIMD_FORCE_INLINE void btTransform::serialize(btTransformData& dataOut) const
 {
 	m_basis.serialize(dataOut.m_basis);
 	m_origin.serialize(dataOut.m_origin);
 }
 
-SIMD_FORCE_INLINE	void	btTransform::serializeFloat(btTransformFloatData& dataOut) const
+SIMD_FORCE_INLINE void btTransform::serializeFloat(btTransformFloatData& dataOut) const
 {
 	m_basis.serializeFloat(dataOut.m_basis);
 	m_origin.serializeFloat(dataOut.m_origin);
 }
 
-
-SIMD_FORCE_INLINE	void	btTransform::deSerialize(const btTransformData& dataIn)
+SIMD_FORCE_INLINE void btTransform::deSerialize(const btTransformData& dataIn)
 {
 	m_basis.deSerialize(dataIn.m_basis);
 	m_origin.deSerialize(dataIn.m_origin);
 }
 
-SIMD_FORCE_INLINE	void	btTransform::deSerializeFloat(const btTransformFloatData& dataIn)
+SIMD_FORCE_INLINE void btTransform::deSerializeFloat(const btTransformFloatData& dataIn)
 {
 	m_basis.deSerializeFloat(dataIn.m_basis);
 	m_origin.deSerializeFloat(dataIn.m_origin);
 }
 
-SIMD_FORCE_INLINE	void	btTransform::deSerializeDouble(const btTransformDoubleData& dataIn)
+SIMD_FORCE_INLINE void btTransform::deSerializeDouble(const btTransformDoubleData& dataIn)
 {
 	m_basis.deSerializeDouble(dataIn.m_basis);
 	m_origin.deSerializeDouble(dataIn.m_origin);
 }
 
-
-#endif //BT_TRANSFORM_H
-
-
-
-
-
-
+#endif  //BT_TRANSFORM_H
diff --git a/thirdparty/bullet/LinearMath/btTransformUtil.h b/thirdparty/bullet/LinearMath/btTransformUtil.h
index 182cc43fab9..b874dd6807d 100644
--- a/thirdparty/bullet/LinearMath/btTransformUtil.h
+++ b/thirdparty/bullet/LinearMath/btTransformUtil.h
@@ -12,77 +12,66 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
 #ifndef BT_TRANSFORM_UTIL_H
 #define BT_TRANSFORM_UTIL_H
 
 #include "btTransform.h"
-#define ANGULAR_MOTION_THRESHOLD btScalar(0.5)*SIMD_HALF_PI
+#define ANGULAR_MOTION_THRESHOLD btScalar(0.5) * SIMD_HALF_PI
 
-
-
-
-SIMD_FORCE_INLINE btVector3 btAabbSupport(const btVector3& halfExtents,const btVector3& supportDir)
+SIMD_FORCE_INLINE btVector3 btAabbSupport(const btVector3& halfExtents, const btVector3& supportDir)
 {
 	return btVector3(supportDir.x() < btScalar(0.0) ? -halfExtents.x() : halfExtents.x(),
-      supportDir.y() < btScalar(0.0) ? -halfExtents.y() : halfExtents.y(),
-      supportDir.z() < btScalar(0.0) ? -halfExtents.z() : halfExtents.z()); 
+					 supportDir.y() < btScalar(0.0) ? -halfExtents.y() : halfExtents.y(),
+					 supportDir.z() < btScalar(0.0) ? -halfExtents.z() : halfExtents.z());
 }
 
-
-
-
-
-
 /// Utils related to temporal transforms
 class btTransformUtil
 {
-
 public:
-
-	static void integrateTransform(const btTransform& curTrans,const btVector3& linvel,const btVector3& angvel,btScalar timeStep,btTransform& predictedTransform)
+	static void integrateTransform(const btTransform& curTrans, const btVector3& linvel, const btVector3& angvel, btScalar timeStep, btTransform& predictedTransform)
 	{
 		predictedTransform.setOrigin(curTrans.getOrigin() + linvel * timeStep);
-//	#define QUATERNION_DERIVATIVE
-	#ifdef QUATERNION_DERIVATIVE
+		//	#define QUATERNION_DERIVATIVE
+#ifdef QUATERNION_DERIVATIVE
 		btQuaternion predictedOrn = curTrans.getRotation();
 		predictedOrn += (angvel * predictedOrn) * (timeStep * btScalar(0.5));
 		predictedOrn.safeNormalize();
-	#else
+#else
 		//Exponential map
 		//google for "Practical Parameterization of Rotations Using the Exponential Map", F. Sebastian Grassia
 
 		btVector3 axis;
-		btScalar	fAngle2 = angvel.length2();
-        	btScalar    fAngle = 0;
-        	if (fAngle2>SIMD_EPSILON)
-        	{
-            		fAngle = btSqrt(fAngle2);
-        	}
+		btScalar fAngle2 = angvel.length2();
+		btScalar fAngle = 0;
+		if (fAngle2 > SIMD_EPSILON)
+		{
+			fAngle = btSqrt(fAngle2);
+		}
 
 		//limit the angular motion
-		if (fAngle*timeStep > ANGULAR_MOTION_THRESHOLD)
+		if (fAngle * timeStep > ANGULAR_MOTION_THRESHOLD)
 		{
 			fAngle = ANGULAR_MOTION_THRESHOLD / timeStep;
 		}
 
-		if ( fAngle < btScalar(0.001) )
+		if (fAngle < btScalar(0.001))
 		{
 			// use Taylor's expansions of sync function
-			axis   = angvel*( btScalar(0.5)*timeStep-(timeStep*timeStep*timeStep)*(btScalar(0.020833333333))*fAngle*fAngle );
+			axis = angvel * (btScalar(0.5) * timeStep - (timeStep * timeStep * timeStep) * (btScalar(0.020833333333)) * fAngle * fAngle);
 		}
 		else
 		{
 			// sync(fAngle) = sin(c*fAngle)/t
-			axis   = angvel*( btSin(btScalar(0.5)*fAngle*timeStep)/fAngle );
+			axis = angvel * (btSin(btScalar(0.5) * fAngle * timeStep) / fAngle);
 		}
-		btQuaternion dorn (axis.x(),axis.y(),axis.z(),btCos( fAngle*timeStep*btScalar(0.5) ));
+		btQuaternion dorn(axis.x(), axis.y(), axis.z(), btCos(fAngle * timeStep * btScalar(0.5)));
 		btQuaternion orn0 = curTrans.getRotation();
 
 		btQuaternion predictedOrn = dorn * orn0;
 		predictedOrn.safeNormalize();
-	#endif
-		if (predictedOrn.length2()>SIMD_EPSILON)
+#endif
+		if (predictedOrn.length2() > SIMD_EPSILON)
 		{
 			predictedTransform.setRotation(predictedOrn);
 		}
@@ -92,137 +81,133 @@ public:
 		}
 	}
 
-	static void	calculateVelocityQuaternion(const btVector3& pos0,const btVector3& pos1,const btQuaternion& orn0,const btQuaternion& orn1,btScalar timeStep,btVector3& linVel,btVector3& angVel)
+	static void calculateVelocityQuaternion(const btVector3& pos0, const btVector3& pos1, const btQuaternion& orn0, const btQuaternion& orn1, btScalar timeStep, btVector3& linVel, btVector3& angVel)
 	{
 		linVel = (pos1 - pos0) / timeStep;
 		btVector3 axis;
-		btScalar  angle;
+		btScalar angle;
 		if (orn0 != orn1)
 		{
-			calculateDiffAxisAngleQuaternion(orn0,orn1,axis,angle);
+			calculateDiffAxisAngleQuaternion(orn0, orn1, axis, angle);
 			angVel = axis * angle / timeStep;
-		} else
+		}
+		else
 		{
-			angVel.setValue(0,0,0);
+			angVel.setValue(0, 0, 0);
 		}
 	}
 
-	static void calculateDiffAxisAngleQuaternion(const btQuaternion& orn0,const btQuaternion& orn1a,btVector3& axis,btScalar& angle)
+	static void calculateDiffAxisAngleQuaternion(const btQuaternion& orn0, const btQuaternion& orn1a, btVector3& axis, btScalar& angle)
 	{
 		btQuaternion orn1 = orn0.nearest(orn1a);
 		btQuaternion dorn = orn1 * orn0.inverse();
 		angle = dorn.getAngle();
-		axis = btVector3(dorn.x(),dorn.y(),dorn.z());
+		axis = btVector3(dorn.x(), dorn.y(), dorn.z());
 		axis[3] = btScalar(0.);
 		//check for axis length
 		btScalar len = axis.length2();
-		if (len < SIMD_EPSILON*SIMD_EPSILON)
-			axis = btVector3(btScalar(1.),btScalar(0.),btScalar(0.));
+		if (len < SIMD_EPSILON * SIMD_EPSILON)
+			axis = btVector3(btScalar(1.), btScalar(0.), btScalar(0.));
 		else
 			axis /= btSqrt(len);
 	}
 
-	static void	calculateVelocity(const btTransform& transform0,const btTransform& transform1,btScalar timeStep,btVector3& linVel,btVector3& angVel)
+	static void calculateVelocity(const btTransform& transform0, const btTransform& transform1, btScalar timeStep, btVector3& linVel, btVector3& angVel)
 	{
 		linVel = (transform1.getOrigin() - transform0.getOrigin()) / timeStep;
 		btVector3 axis;
-		btScalar  angle;
-		calculateDiffAxisAngle(transform0,transform1,axis,angle);
+		btScalar angle;
+		calculateDiffAxisAngle(transform0, transform1, axis, angle);
 		angVel = axis * angle / timeStep;
 	}
 
-	static void calculateDiffAxisAngle(const btTransform& transform0,const btTransform& transform1,btVector3& axis,btScalar& angle)
+	static void calculateDiffAxisAngle(const btTransform& transform0, const btTransform& transform1, btVector3& axis, btScalar& angle)
 	{
 		btMatrix3x3 dmat = transform1.getBasis() * transform0.getBasis().inverse();
 		btQuaternion dorn;
 		dmat.getRotation(dorn);
 
-		///floating point inaccuracy can lead to w component > 1..., which breaks 
+		///floating point inaccuracy can lead to w component > 1..., which breaks
 		dorn.normalize();
-		
+
 		angle = dorn.getAngle();
-		axis = btVector3(dorn.x(),dorn.y(),dorn.z());
+		axis = btVector3(dorn.x(), dorn.y(), dorn.z());
 		axis[3] = btScalar(0.);
 		//check for axis length
 		btScalar len = axis.length2();
-		if (len < SIMD_EPSILON*SIMD_EPSILON)
-			axis = btVector3(btScalar(1.),btScalar(0.),btScalar(0.));
+		if (len < SIMD_EPSILON * SIMD_EPSILON)
+			axis = btVector3(btScalar(1.), btScalar(0.), btScalar(0.));
 		else
 			axis /= btSqrt(len);
 	}
-
 };
 
-
-///The btConvexSeparatingDistanceUtil can help speed up convex collision detection 
+///The btConvexSeparatingDistanceUtil can help speed up convex collision detection
 ///by conservatively updating a cached separating distance/vector instead of re-calculating the closest distance
-class	btConvexSeparatingDistanceUtil
+class btConvexSeparatingDistanceUtil
 {
-	btQuaternion	m_ornA;
-	btQuaternion	m_ornB;
-	btVector3	m_posA;
-	btVector3	m_posB;
-	
-	btVector3	m_separatingNormal;
+	btQuaternion m_ornA;
+	btQuaternion m_ornB;
+	btVector3 m_posA;
+	btVector3 m_posB;
 
-	btScalar	m_boundingRadiusA;
-	btScalar	m_boundingRadiusB;
-	btScalar	m_separatingDistance;
+	btVector3 m_separatingNormal;
+
+	btScalar m_boundingRadiusA;
+	btScalar m_boundingRadiusB;
+	btScalar m_separatingDistance;
 
 public:
-
-	btConvexSeparatingDistanceUtil(btScalar	boundingRadiusA,btScalar	boundingRadiusB)
-		:m_boundingRadiusA(boundingRadiusA),
-		m_boundingRadiusB(boundingRadiusB),
-		m_separatingDistance(0.f)
+	btConvexSeparatingDistanceUtil(btScalar boundingRadiusA, btScalar boundingRadiusB)
+		: m_boundingRadiusA(boundingRadiusA),
+		  m_boundingRadiusB(boundingRadiusB),
+		  m_separatingDistance(0.f)
 	{
 	}
 
-	btScalar	getConservativeSeparatingDistance()
+	btScalar getConservativeSeparatingDistance()
 	{
 		return m_separatingDistance;
 	}
 
-	void	updateSeparatingDistance(const btTransform& transA,const btTransform& transB)
+	void updateSeparatingDistance(const btTransform& transA, const btTransform& transB)
 	{
 		const btVector3& toPosA = transA.getOrigin();
 		const btVector3& toPosB = transB.getOrigin();
 		btQuaternion toOrnA = transA.getRotation();
 		btQuaternion toOrnB = transB.getRotation();
 
-		if (m_separatingDistance>0.f)
+		if (m_separatingDistance > 0.f)
 		{
-			
-
-			btVector3 linVelA,angVelA,linVelB,angVelB;
-			btTransformUtil::calculateVelocityQuaternion(m_posA,toPosA,m_ornA,toOrnA,btScalar(1.),linVelA,angVelA);
-			btTransformUtil::calculateVelocityQuaternion(m_posB,toPosB,m_ornB,toOrnB,btScalar(1.),linVelB,angVelB);
+			btVector3 linVelA, angVelA, linVelB, angVelB;
+			btTransformUtil::calculateVelocityQuaternion(m_posA, toPosA, m_ornA, toOrnA, btScalar(1.), linVelA, angVelA);
+			btTransformUtil::calculateVelocityQuaternion(m_posB, toPosB, m_ornB, toOrnB, btScalar(1.), linVelB, angVelB);
 			btScalar maxAngularProjectedVelocity = angVelA.length() * m_boundingRadiusA + angVelB.length() * m_boundingRadiusB;
-			btVector3 relLinVel = (linVelB-linVelA);
+			btVector3 relLinVel = (linVelB - linVelA);
 			btScalar relLinVelocLength = relLinVel.dot(m_separatingNormal);
-			if (relLinVelocLength<0.f)
+			if (relLinVelocLength < 0.f)
 			{
 				relLinVelocLength = 0.f;
 			}
-	
-			btScalar	projectedMotion = maxAngularProjectedVelocity +relLinVelocLength;
+
+			btScalar projectedMotion = maxAngularProjectedVelocity + relLinVelocLength;
 			m_separatingDistance -= projectedMotion;
 		}
-	
+
 		m_posA = toPosA;
 		m_posB = toPosB;
 		m_ornA = toOrnA;
 		m_ornB = toOrnB;
 	}
 
-	void	initSeparatingDistance(const btVector3& separatingVector,btScalar separatingDistance,const btTransform& transA,const btTransform& transB)
+	void initSeparatingDistance(const btVector3& separatingVector, btScalar separatingDistance, const btTransform& transA, const btTransform& transB)
 	{
 		m_separatingDistance = separatingDistance;
 
-		if (m_separatingDistance>0.f)
+		if (m_separatingDistance > 0.f)
 		{
 			m_separatingNormal = separatingVector;
-			
+
 			const btVector3& toPosA = transA.getOrigin();
 			const btVector3& toPosB = transB.getOrigin();
 			btQuaternion toOrnA = transA.getRotation();
@@ -233,9 +218,6 @@ public:
 			m_ornB = toOrnB;
 		}
 	}
-
 };
 
-
-#endif //BT_TRANSFORM_UTIL_H
-
+#endif  //BT_TRANSFORM_UTIL_H
diff --git a/thirdparty/bullet/LinearMath/btVector3.cpp b/thirdparty/bullet/LinearMath/btVector3.cpp
index e05bdccd67e..13111157af4 100644
--- a/thirdparty/bullet/LinearMath/btVector3.cpp
+++ b/thirdparty/bullet/LinearMath/btVector3.cpp
@@ -15,282 +15,285 @@
  This source version has been altered.
  */
 
-#if defined (_WIN32) || defined (__i386__)
+#if defined(_WIN32) || defined(__i386__)
 #define BT_USE_SSE_IN_API
 #endif
 
-
 #include "btVector3.h"
 
-
-
 #if defined BT_USE_SIMD_VECTOR3
 
 #if DEBUG
-#include <string.h>//for memset
+#include <string.h>  //for memset
 #endif
 
-
 #ifdef __APPLE__
 #include <stdint.h>
-typedef  float float4 __attribute__ ((vector_size(16)));
+typedef float float4 __attribute__((vector_size(16)));
 #else
 #define float4 __m128
 #endif
 //typedef  uint32_t uint4 __attribute__ ((vector_size(16)));
 
-
 #if defined BT_USE_SSE || defined _WIN32
 
-#define LOG2_ARRAY_SIZE     6
-#define STACK_ARRAY_COUNT   (1UL << LOG2_ARRAY_SIZE)
+#define LOG2_ARRAY_SIZE 6
+#define STACK_ARRAY_COUNT (1UL << LOG2_ARRAY_SIZE)
 
 #include <emmintrin.h>
 
-long _maxdot_large( const float *vv, const float *vec, unsigned long count, float *dotResult );
-long _maxdot_large( const float *vv, const float *vec, unsigned long count, float *dotResult )
+long _maxdot_large(const float *vv, const float *vec, unsigned long count, float *dotResult);
+long _maxdot_large(const float *vv, const float *vec, unsigned long count, float *dotResult)
 {
-    const float4 *vertices = (const float4*) vv;
-    static const unsigned char indexTable[16] = {(unsigned char)-1, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0 };
-    float4 dotMax = btAssign128( -BT_INFINITY,  -BT_INFINITY,  -BT_INFINITY,  -BT_INFINITY );
-    float4 vvec = _mm_loadu_ps( vec );
-    float4 vHi = btCastiTo128f(_mm_shuffle_epi32( btCastfTo128i( vvec), 0xaa ));          /// zzzz
-    float4 vLo = _mm_movelh_ps( vvec, vvec );                               /// xyxy
-    
-    long maxIndex = -1L;
-    
-    size_t segment = 0;
-    float4 stack_array[ STACK_ARRAY_COUNT ];
-    
+	const float4 *vertices = (const float4 *)vv;
+	static const unsigned char indexTable[16] = {(unsigned char)-1, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0};
+	float4 dotMax = btAssign128(-BT_INFINITY, -BT_INFINITY, -BT_INFINITY, -BT_INFINITY);
+	float4 vvec = _mm_loadu_ps(vec);
+	float4 vHi = btCastiTo128f(_mm_shuffle_epi32(btCastfTo128i(vvec), 0xaa));  /// zzzz
+	float4 vLo = _mm_movelh_ps(vvec, vvec);                                    /// xyxy
+
+	long maxIndex = -1L;
+
+	size_t segment = 0;
+	float4 stack_array[STACK_ARRAY_COUNT];
+
 #if DEBUG
-    //memset( stack_array, -1, STACK_ARRAY_COUNT * sizeof(stack_array[0]) );
+	//memset( stack_array, -1, STACK_ARRAY_COUNT * sizeof(stack_array[0]) );
 #endif
-    
-    size_t index;
-    float4 max;
-    // Faster loop without cleanup code for full tiles
-    for ( segment = 0; segment + STACK_ARRAY_COUNT*4 <= count; segment += STACK_ARRAY_COUNT*4 ) 
-    {
-        max = dotMax;
-        
-        for( index = 0; index < STACK_ARRAY_COUNT; index+= 4 )   
-        { // do four dot products at a time. Carefully avoid touching the w element.
-            float4 v0 = vertices[0];
-            float4 v1 = vertices[1];
-            float4 v2 = vertices[2];
-            float4 v3 = vertices[3];            vertices += 4;
-            
-            float4 lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            float4 hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            float4 lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            float4 hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
-            
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index] = x;
-            max = _mm_max_ps( x, max );         // control the order here so that max is never NaN even if x is nan
-            
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
-            
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
-            
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+1] = x;
-            max = _mm_max_ps( x, max );         // control the order here so that max is never NaN even if x is nan
-            
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
-            
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
-            
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+2] = x;
-            max = _mm_max_ps( x, max );         // control the order here so that max is never NaN even if x is nan
-            
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
-            
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
-            
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+3] = x;
-            max = _mm_max_ps( x, max );         // control the order here so that max is never NaN even if x is nan
-            
-            // It is too costly to keep the index of the max here. We will look for it again later.  We save a lot of work this way.
-        }
-        
-        // If we found a new max
-        if( 0xf != _mm_movemask_ps( (float4) _mm_cmpeq_ps(max, dotMax)))
-        { 
-            // copy the new max across all lanes of our max accumulator
-            max = _mm_max_ps(max, (float4) _mm_shuffle_ps( max, max, 0x4e));
-            max = _mm_max_ps(max, (float4) _mm_shuffle_ps( max, max, 0xb1));
-            
-            dotMax = max;
-            
-            // find first occurrence of that max  
-            size_t test;
-            for( index = 0; 0 == (test=_mm_movemask_ps( _mm_cmpeq_ps( stack_array[index], max))); index++ )   // local_count must be a multiple of 4
-            {}
-            // record where it is.
-            maxIndex = 4*index + segment + indexTable[test];
-        }
-    }
-    
-    // account for work we've already done
-    count -= segment;
-    
-    // Deal with the last < STACK_ARRAY_COUNT vectors
-    max = dotMax;
-    index = 0;
-    
-    
-    if( btUnlikely( count > 16) )
-    {
-        for( ; index + 4 <= count / 4; index+=4 )   
-        { // do four dot products at a time. Carefully avoid touching the w element.
-            float4 v0 = vertices[0];
-            float4 v1 = vertices[1];
-            float4 v2 = vertices[2];
-            float4 v3 = vertices[3];            vertices += 4;
-            
-            float4 lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            float4 hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            float4 lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            float4 hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
-            
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index] = x;
-            max = _mm_max_ps( x, max );         // control the order here so that max is never NaN even if x is nan
-            
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
-            
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
-            
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+1] = x;
-            max = _mm_max_ps( x, max );         // control the order here so that max is never NaN even if x is nan
-            
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
-            
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
-            
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+2] = x;
-            max = _mm_max_ps( x, max );         // control the order here so that max is never NaN even if x is nan
-            
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
-            
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
-            
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+3] = x;
-            max = _mm_max_ps( x, max );         // control the order here so that max is never NaN even if x is nan
-            
-            // It is too costly to keep the index of the max here. We will look for it again later.  We save a lot of work this way.
-        }
-    }
-    
-    size_t localCount = (count & -4L) - 4*index;
-    if( localCount )
-    {
+
+	size_t index;
+	float4 max;
+	// Faster loop without cleanup code for full tiles
+	for (segment = 0; segment + STACK_ARRAY_COUNT * 4 <= count; segment += STACK_ARRAY_COUNT * 4)
+	{
+		max = dotMax;
+
+		for (index = 0; index < STACK_ARRAY_COUNT; index += 4)
+		{  // do four dot products at a time. Carefully avoid touching the w element.
+			float4 v0 = vertices[0];
+			float4 v1 = vertices[1];
+			float4 v2 = vertices[2];
+			float4 v3 = vertices[3];
+			vertices += 4;
+
+			float4 lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			float4 hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			float4 lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			float4 hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
+
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index] = x;
+			max = _mm_max_ps(x, max);  // control the order here so that max is never NaN even if x is nan
+
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
+
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
+
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 1] = x;
+			max = _mm_max_ps(x, max);  // control the order here so that max is never NaN even if x is nan
+
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
+
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
+
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 2] = x;
+			max = _mm_max_ps(x, max);  // control the order here so that max is never NaN even if x is nan
+
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
+
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
+
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 3] = x;
+			max = _mm_max_ps(x, max);  // control the order here so that max is never NaN even if x is nan
+
+			// It is too costly to keep the index of the max here. We will look for it again later.  We save a lot of work this way.
+		}
+
+		// If we found a new max
+		if (0xf != _mm_movemask_ps((float4)_mm_cmpeq_ps(max, dotMax)))
+		{
+			// copy the new max across all lanes of our max accumulator
+			max = _mm_max_ps(max, (float4)_mm_shuffle_ps(max, max, 0x4e));
+			max = _mm_max_ps(max, (float4)_mm_shuffle_ps(max, max, 0xb1));
+
+			dotMax = max;
+
+			// find first occurrence of that max
+			size_t test;
+			for (index = 0; 0 == (test = _mm_movemask_ps(_mm_cmpeq_ps(stack_array[index], max))); index++)  // local_count must be a multiple of 4
+			{
+			}
+			// record where it is.
+			maxIndex = 4 * index + segment + indexTable[test];
+		}
+	}
+
+	// account for work we've already done
+	count -= segment;
+
+	// Deal with the last < STACK_ARRAY_COUNT vectors
+	max = dotMax;
+	index = 0;
+
+	if (btUnlikely(count > 16))
+	{
+		for (; index + 4 <= count / 4; index += 4)
+		{  // do four dot products at a time. Carefully avoid touching the w element.
+			float4 v0 = vertices[0];
+			float4 v1 = vertices[1];
+			float4 v2 = vertices[2];
+			float4 v3 = vertices[3];
+			vertices += 4;
+
+			float4 lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			float4 hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			float4 lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			float4 hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
+
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index] = x;
+			max = _mm_max_ps(x, max);  // control the order here so that max is never NaN even if x is nan
+
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
+
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
+
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 1] = x;
+			max = _mm_max_ps(x, max);  // control the order here so that max is never NaN even if x is nan
+
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
+
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
+
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 2] = x;
+			max = _mm_max_ps(x, max);  // control the order here so that max is never NaN even if x is nan
+
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
+
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
+
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 3] = x;
+			max = _mm_max_ps(x, max);  // control the order here so that max is never NaN even if x is nan
+
+			// It is too costly to keep the index of the max here. We will look for it again later.  We save a lot of work this way.
+		}
+	}
+
+	size_t localCount = (count & -4L) - 4 * index;
+	if (localCount)
+	{
 #ifdef __APPLE__
-        float4 t0, t1, t2, t3, t4;
-        float4 * sap = &stack_array[index + localCount / 4];
-          vertices += localCount;      // counter the offset
-         size_t byteIndex = -(localCount) * sizeof(float);
-        //AT&T Code style assembly
-        asm volatile
-        (   ".align 4                                                                   \n\
+		float4 t0, t1, t2, t3, t4;
+		float4 *sap = &stack_array[index + localCount / 4];
+		vertices += localCount;  // counter the offset
+		size_t byteIndex = -(localCount) * sizeof(float);
+		//AT&T Code style assembly
+		asm volatile(
+			".align 4                                                                   \n\
              0: movaps  %[max], %[t2]                            // move max out of the way to avoid propagating NaNs in max \n\
           movaps  (%[vertices], %[byteIndex], 4),    %[t0]    // vertices[0]      \n\
           movaps  16(%[vertices], %[byteIndex], 4),  %[t1]    // vertices[1]      \n\
@@ -316,368 +319,374 @@ long _maxdot_large( const float *vv, const float *vec, unsigned long count, floa
          add     $16, %[byteIndex]                           // advance loop counter\n\
          jnz     0b                                          \n\
      "
-         : [max] "+x" (max), [t0] "=&x" (t0), [t1] "=&x" (t1), [t2] "=&x" (t2), [t3] "=&x" (t3), [t4] "=&x" (t4), [byteIndex] "+r" (byteIndex)
-         : [vLo] "x" (vLo), [vHi] "x" (vHi), [vertices] "r" (vertices), [sap] "r" (sap)
-         : "memory", "cc"
-         );
-        index += localCount/4;
+			: [max] "+x"(max), [t0] "=&x"(t0), [t1] "=&x"(t1), [t2] "=&x"(t2), [t3] "=&x"(t3), [t4] "=&x"(t4), [byteIndex] "+r"(byteIndex)
+			: [vLo] "x"(vLo), [vHi] "x"(vHi), [vertices] "r"(vertices), [sap] "r"(sap)
+			: "memory", "cc");
+		index += localCount / 4;
 #else
-        {
-            for( unsigned int i=0; i<localCount/4; i++,index++)   
-            { // do four dot products at a time. Carefully avoid touching the w element.
-                float4 v0 = vertices[0];
-                float4 v1 = vertices[1];
-                float4 v2 = vertices[2];
-                float4 v3 = vertices[3];            
-                vertices += 4;
-                
-                float4 lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-                float4 hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-                float4 lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-                float4 hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
-                
-                lo0 = lo0*vLo;
-                lo1 = lo1*vLo;
-                float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
-                float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
-                float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-                z = z*vHi;
-                x = x+y;
-                x = x+z;
-                stack_array[index] = x;
-                max = _mm_max_ps( x, max );         // control the order here so that max is never NaN even if x is nan
-            }
-        }
-#endif //__APPLE__
-    }
+		{
+			for (unsigned int i = 0; i < localCount / 4; i++, index++)
+			{  // do four dot products at a time. Carefully avoid touching the w element.
+				float4 v0 = vertices[0];
+				float4 v1 = vertices[1];
+				float4 v2 = vertices[2];
+				float4 v3 = vertices[3];
+				vertices += 4;
 
-    // process the last few points
-    if( count & 3 )
-    {
-        float4 v0, v1, v2, x, y, z;
-        switch( count & 3 )
-        {
-            case 3:
-            {
-                v0 = vertices[0];
-                v1 = vertices[1];
-                v2 = vertices[2];
-                
-                // Calculate 3 dot products, transpose, duplicate v2
-                float4 lo0 = _mm_movelh_ps( v0, v1);        // xyxy.lo
-                float4 hi0 = _mm_movehl_ps( v1, v0);        // z?z?.lo
-                lo0 = lo0*vLo;
-                z = _mm_shuffle_ps(hi0, v2,  0xa8 );           // z0z1z2z2
-                z = z*vHi;
-                float4 lo1 = _mm_movelh_ps(v2, v2);          // xyxy
-                lo1 = lo1*vLo;
-                x = _mm_shuffle_ps(lo0, lo1, 0x88);
-                y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            }
-                break;
-            case 2:
-            {
-                v0 = vertices[0];
-                v1 = vertices[1];
-                float4 xy = _mm_movelh_ps(v0, v1);
-                z = _mm_movehl_ps(v1, v0);
-                xy = xy*vLo;
-                z = _mm_shuffle_ps( z, z,  0xa8);
-                x = _mm_shuffle_ps( xy, xy, 0xa8);
-                y = _mm_shuffle_ps( xy, xy, 0xfd);
-                z = z*vHi;
-            }
-                break;
-            case 1:
-            {
-                float4 xy = vertices[0];
-                z =  _mm_shuffle_ps( xy, xy, 0xaa);
-                xy = xy*vLo;
-                z = z*vHi;
-                x = _mm_shuffle_ps(xy, xy, 0);
-                y = _mm_shuffle_ps(xy, xy, 0x55);
-            }
-                break;
-        }
-        x = x+y;
-        x = x+z;
-        stack_array[index] = x;
-        max = _mm_max_ps( x, max );         // control the order here so that max is never NaN even if x is nan
-        index++;
-    }
-    
-    // if we found a new max. 
-    if( 0 == segment || 0xf != _mm_movemask_ps( (float4) _mm_cmpeq_ps(max, dotMax)))
-    { // we found a new max. Search for it
-      // find max across the max vector, place in all elements of max -- big latency hit here
-        max = _mm_max_ps(max, (float4) _mm_shuffle_ps( max, max, 0x4e));
-        max = _mm_max_ps(max, (float4) _mm_shuffle_ps( max, max, 0xb1));
-        
-        // It is slightly faster to do this part in scalar code when count < 8. However, the common case for
-        // this where it actually makes a difference is handled in the early out at the top of the function, 
-        // so it is less than a 1% difference here. I opted for improved code size, fewer branches and reduced 
-        // complexity, and removed it.
-        
-        dotMax = max;
-        
-        // scan for the first occurence of max in the array  
-        size_t test;
-        for( index = 0; 0 == (test=_mm_movemask_ps( _mm_cmpeq_ps( stack_array[index], max))); index++ )   // local_count must be a multiple of 4
-        {}
-        maxIndex = 4*index + segment + indexTable[test];
-    }
-    
-    _mm_store_ss( dotResult, dotMax);
-    return maxIndex;
+				float4 lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+				float4 hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+				float4 lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+				float4 hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
+
+				lo0 = lo0 * vLo;
+				lo1 = lo1 * vLo;
+				float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
+				float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
+				float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+				z = z * vHi;
+				x = x + y;
+				x = x + z;
+				stack_array[index] = x;
+				max = _mm_max_ps(x, max);  // control the order here so that max is never NaN even if x is nan
+			}
+		}
+#endif  //__APPLE__
+	}
+
+	// process the last few points
+	if (count & 3)
+	{
+		float4 v0, v1, v2, x, y, z;
+		switch (count & 3)
+		{
+			case 3:
+			{
+				v0 = vertices[0];
+				v1 = vertices[1];
+				v2 = vertices[2];
+
+				// Calculate 3 dot products, transpose, duplicate v2
+				float4 lo0 = _mm_movelh_ps(v0, v1);  // xyxy.lo
+				float4 hi0 = _mm_movehl_ps(v1, v0);  // z?z?.lo
+				lo0 = lo0 * vLo;
+				z = _mm_shuffle_ps(hi0, v2, 0xa8);  // z0z1z2z2
+				z = z * vHi;
+				float4 lo1 = _mm_movelh_ps(v2, v2);  // xyxy
+				lo1 = lo1 * vLo;
+				x = _mm_shuffle_ps(lo0, lo1, 0x88);
+				y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			}
+			break;
+			case 2:
+			{
+				v0 = vertices[0];
+				v1 = vertices[1];
+				float4 xy = _mm_movelh_ps(v0, v1);
+				z = _mm_movehl_ps(v1, v0);
+				xy = xy * vLo;
+				z = _mm_shuffle_ps(z, z, 0xa8);
+				x = _mm_shuffle_ps(xy, xy, 0xa8);
+				y = _mm_shuffle_ps(xy, xy, 0xfd);
+				z = z * vHi;
+			}
+			break;
+			case 1:
+			{
+				float4 xy = vertices[0];
+				z = _mm_shuffle_ps(xy, xy, 0xaa);
+				xy = xy * vLo;
+				z = z * vHi;
+				x = _mm_shuffle_ps(xy, xy, 0);
+				y = _mm_shuffle_ps(xy, xy, 0x55);
+			}
+			break;
+		}
+		x = x + y;
+		x = x + z;
+		stack_array[index] = x;
+		max = _mm_max_ps(x, max);  // control the order here so that max is never NaN even if x is nan
+		index++;
+	}
+
+	// if we found a new max.
+	if (0 == segment || 0xf != _mm_movemask_ps((float4)_mm_cmpeq_ps(max, dotMax)))
+	{  // we found a new max. Search for it
+		// find max across the max vector, place in all elements of max -- big latency hit here
+		max = _mm_max_ps(max, (float4)_mm_shuffle_ps(max, max, 0x4e));
+		max = _mm_max_ps(max, (float4)_mm_shuffle_ps(max, max, 0xb1));
+
+		// It is slightly faster to do this part in scalar code when count < 8. However, the common case for
+		// this where it actually makes a difference is handled in the early out at the top of the function,
+		// so it is less than a 1% difference here. I opted for improved code size, fewer branches and reduced
+		// complexity, and removed it.
+
+		dotMax = max;
+
+		// scan for the first occurence of max in the array
+		size_t test;
+		for (index = 0; 0 == (test = _mm_movemask_ps(_mm_cmpeq_ps(stack_array[index], max))); index++)  // local_count must be a multiple of 4
+		{
+		}
+		maxIndex = 4 * index + segment + indexTable[test];
+	}
+
+	_mm_store_ss(dotResult, dotMax);
+	return maxIndex;
 }
 
-long _mindot_large( const float *vv, const float *vec, unsigned long count, float *dotResult );
+long _mindot_large(const float *vv, const float *vec, unsigned long count, float *dotResult);
 
-long _mindot_large( const float *vv, const float *vec, unsigned long count, float *dotResult )
+long _mindot_large(const float *vv, const float *vec, unsigned long count, float *dotResult)
 {
-    const float4 *vertices = (const float4*) vv;
-    static const unsigned char indexTable[16] = {(unsigned char)-1, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0 };
-    float4 dotmin = btAssign128( BT_INFINITY,  BT_INFINITY,  BT_INFINITY,  BT_INFINITY );
-    float4 vvec = _mm_loadu_ps( vec );
-    float4 vHi = btCastiTo128f(_mm_shuffle_epi32( btCastfTo128i( vvec), 0xaa ));          /// zzzz
-    float4 vLo = _mm_movelh_ps( vvec, vvec );                               /// xyxy
-    
-    long minIndex = -1L;
+	const float4 *vertices = (const float4 *)vv;
+	static const unsigned char indexTable[16] = {(unsigned char)-1, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0};
+	float4 dotmin = btAssign128(BT_INFINITY, BT_INFINITY, BT_INFINITY, BT_INFINITY);
+	float4 vvec = _mm_loadu_ps(vec);
+	float4 vHi = btCastiTo128f(_mm_shuffle_epi32(btCastfTo128i(vvec), 0xaa));  /// zzzz
+	float4 vLo = _mm_movelh_ps(vvec, vvec);                                    /// xyxy
+
+	long minIndex = -1L;
+
+	size_t segment = 0;
+	float4 stack_array[STACK_ARRAY_COUNT];
 
-    size_t segment = 0;
-    float4 stack_array[ STACK_ARRAY_COUNT ];
-    
 #if DEBUG
-    //memset( stack_array, -1, STACK_ARRAY_COUNT * sizeof(stack_array[0]) );
+	//memset( stack_array, -1, STACK_ARRAY_COUNT * sizeof(stack_array[0]) );
 #endif
-    
-    size_t index;
-    float4 min;
-    // Faster loop without cleanup code for full tiles
-    for ( segment = 0; segment + STACK_ARRAY_COUNT*4 <= count; segment += STACK_ARRAY_COUNT*4 ) 
-    {
-        min = dotmin;
-        
-        for( index = 0; index < STACK_ARRAY_COUNT; index+= 4 )   
-        { // do four dot products at a time. Carefully avoid touching the w element.
-            float4 v0 = vertices[0];
-            float4 v1 = vertices[1];
-            float4 v2 = vertices[2];
-            float4 v3 = vertices[3];            vertices += 4;
-            
-            float4 lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            float4 hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            float4 lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            float4 hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
-            
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index] = x;
-            min = _mm_min_ps( x, min );         // control the order here so that min is never NaN even if x is nan
-            
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
-            
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
-            
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+1] = x;
-            min = _mm_min_ps( x, min );         // control the order here so that min is never NaN even if x is nan
-            
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
-            
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
-            
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+2] = x;
-            min = _mm_min_ps( x, min );         // control the order here so that min is never NaN even if x is nan
-            
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
-            
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
-            
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+3] = x;
-            min = _mm_min_ps( x, min );         // control the order here so that min is never NaN even if x is nan
-            
-            // It is too costly to keep the index of the min here. We will look for it again later.  We save a lot of work this way.
-        }
-        
-        // If we found a new min
-        if( 0xf != _mm_movemask_ps( (float4) _mm_cmpeq_ps(min, dotmin)))
-        { 
-            // copy the new min across all lanes of our min accumulator
-            min = _mm_min_ps(min, (float4) _mm_shuffle_ps( min, min, 0x4e));
-            min = _mm_min_ps(min, (float4) _mm_shuffle_ps( min, min, 0xb1));
-            
-            dotmin = min;
-            
-            // find first occurrence of that min  
-            size_t test;
-            for( index = 0; 0 == (test=_mm_movemask_ps( _mm_cmpeq_ps( stack_array[index], min))); index++ )   // local_count must be a multiple of 4
-            {}
-            // record where it is.
-            minIndex = 4*index + segment + indexTable[test];
-        }
-    }
-    
-    // account for work we've already done
-    count -= segment;
-    
-    // Deal with the last < STACK_ARRAY_COUNT vectors
-    min = dotmin;
-    index = 0;
-    
-    
-    if(btUnlikely( count > 16) )
-    {
-        for( ; index + 4 <= count / 4; index+=4 )   
-        { // do four dot products at a time. Carefully avoid touching the w element.
-            float4 v0 = vertices[0];
-            float4 v1 = vertices[1];
-            float4 v2 = vertices[2];
-            float4 v3 = vertices[3];            vertices += 4;
-            
-            float4 lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            float4 hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            float4 lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            float4 hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
-            
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index] = x;
-            min = _mm_min_ps( x, min );         // control the order here so that min is never NaN even if x is nan
-            
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
-            
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
-            
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+1] = x;
-            min = _mm_min_ps( x, min );         // control the order here so that min is never NaN even if x is nan
-            
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
-            
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
-            
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+2] = x;
-            min = _mm_min_ps( x, min );         // control the order here so that min is never NaN even if x is nan
-            
-            v0 = vertices[0];
-            v1 = vertices[1];
-            v2 = vertices[2];
-            v3 = vertices[3];            vertices += 4;
-            
-            lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-            hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-            lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-            hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
-            
-            lo0 = lo0*vLo;
-            lo1 = lo1*vLo;
-            z = _mm_shuffle_ps(hi0, hi1, 0x88);
-            x = _mm_shuffle_ps(lo0, lo1, 0x88);
-            y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            z = z*vHi;
-            x = x+y;
-            x = x+z;
-            stack_array[index+3] = x;
-            min = _mm_min_ps( x, min );         // control the order here so that min is never NaN even if x is nan
-            
-            // It is too costly to keep the index of the min here. We will look for it again later.  We save a lot of work this way.
-        }
-    }
-    
-    size_t localCount = (count & -4L) - 4*index;
-    if( localCount )
-    {
-        
-        
+
+	size_t index;
+	float4 min;
+	// Faster loop without cleanup code for full tiles
+	for (segment = 0; segment + STACK_ARRAY_COUNT * 4 <= count; segment += STACK_ARRAY_COUNT * 4)
+	{
+		min = dotmin;
+
+		for (index = 0; index < STACK_ARRAY_COUNT; index += 4)
+		{  // do four dot products at a time. Carefully avoid touching the w element.
+			float4 v0 = vertices[0];
+			float4 v1 = vertices[1];
+			float4 v2 = vertices[2];
+			float4 v3 = vertices[3];
+			vertices += 4;
+
+			float4 lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			float4 hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			float4 lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			float4 hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
+
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index] = x;
+			min = _mm_min_ps(x, min);  // control the order here so that min is never NaN even if x is nan
+
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
+
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
+
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 1] = x;
+			min = _mm_min_ps(x, min);  // control the order here so that min is never NaN even if x is nan
+
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
+
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
+
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 2] = x;
+			min = _mm_min_ps(x, min);  // control the order here so that min is never NaN even if x is nan
+
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
+
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
+
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 3] = x;
+			min = _mm_min_ps(x, min);  // control the order here so that min is never NaN even if x is nan
+
+			// It is too costly to keep the index of the min here. We will look for it again later.  We save a lot of work this way.
+		}
+
+		// If we found a new min
+		if (0xf != _mm_movemask_ps((float4)_mm_cmpeq_ps(min, dotmin)))
+		{
+			// copy the new min across all lanes of our min accumulator
+			min = _mm_min_ps(min, (float4)_mm_shuffle_ps(min, min, 0x4e));
+			min = _mm_min_ps(min, (float4)_mm_shuffle_ps(min, min, 0xb1));
+
+			dotmin = min;
+
+			// find first occurrence of that min
+			size_t test;
+			for (index = 0; 0 == (test = _mm_movemask_ps(_mm_cmpeq_ps(stack_array[index], min))); index++)  // local_count must be a multiple of 4
+			{
+			}
+			// record where it is.
+			minIndex = 4 * index + segment + indexTable[test];
+		}
+	}
+
+	// account for work we've already done
+	count -= segment;
+
+	// Deal with the last < STACK_ARRAY_COUNT vectors
+	min = dotmin;
+	index = 0;
+
+	if (btUnlikely(count > 16))
+	{
+		for (; index + 4 <= count / 4; index += 4)
+		{  // do four dot products at a time. Carefully avoid touching the w element.
+			float4 v0 = vertices[0];
+			float4 v1 = vertices[1];
+			float4 v2 = vertices[2];
+			float4 v3 = vertices[3];
+			vertices += 4;
+
+			float4 lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			float4 hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			float4 lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			float4 hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
+
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index] = x;
+			min = _mm_min_ps(x, min);  // control the order here so that min is never NaN even if x is nan
+
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
+
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
+
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 1] = x;
+			min = _mm_min_ps(x, min);  // control the order here so that min is never NaN even if x is nan
+
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
+
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
+
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 2] = x;
+			min = _mm_min_ps(x, min);  // control the order here so that min is never NaN even if x is nan
+
+			v0 = vertices[0];
+			v1 = vertices[1];
+			v2 = vertices[2];
+			v3 = vertices[3];
+			vertices += 4;
+
+			lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+			hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+			lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+			hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
+
+			lo0 = lo0 * vLo;
+			lo1 = lo1 * vLo;
+			z = _mm_shuffle_ps(hi0, hi1, 0x88);
+			x = _mm_shuffle_ps(lo0, lo1, 0x88);
+			y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			z = z * vHi;
+			x = x + y;
+			x = x + z;
+			stack_array[index + 3] = x;
+			min = _mm_min_ps(x, min);  // control the order here so that min is never NaN even if x is nan
+
+			// It is too costly to keep the index of the min here. We will look for it again later.  We save a lot of work this way.
+		}
+	}
+
+	size_t localCount = (count & -4L) - 4 * index;
+	if (localCount)
+	{
 #ifdef __APPLE__
-        vertices += localCount;      // counter the offset
-        float4 t0, t1, t2, t3, t4;
-        size_t byteIndex = -(localCount) * sizeof(float);
-        float4 * sap = &stack_array[index + localCount / 4];
-        
-        asm volatile
-        (   ".align 4                                                                   \n\
+		vertices += localCount;  // counter the offset
+		float4 t0, t1, t2, t3, t4;
+		size_t byteIndex = -(localCount) * sizeof(float);
+		float4 *sap = &stack_array[index + localCount / 4];
+
+		asm volatile(
+			".align 4                                                                   \n\
              0: movaps  %[min], %[t2]                            // move min out of the way to avoid propagating NaNs in min \n\
              movaps  (%[vertices], %[byteIndex], 4),    %[t0]    // vertices[0]      \n\
              movaps  16(%[vertices], %[byteIndex], 4),  %[t1]    // vertices[1]      \n\
@@ -703,968 +712,953 @@ long _mindot_large( const float *vv, const float *vec, unsigned long count, floa
              add     $16, %[byteIndex]                           // advance loop counter\n\
              jnz     0b                                          \n\
              "
-         : [min] "+x" (min), [t0] "=&x" (t0), [t1] "=&x" (t1), [t2] "=&x" (t2), [t3] "=&x" (t3), [t4] "=&x" (t4), [byteIndex] "+r" (byteIndex)
-         : [vLo] "x" (vLo), [vHi] "x" (vHi), [vertices] "r" (vertices), [sap] "r" (sap)
-         : "memory", "cc"
-         );
-        index += localCount/4;
+			: [min] "+x"(min), [t0] "=&x"(t0), [t1] "=&x"(t1), [t2] "=&x"(t2), [t3] "=&x"(t3), [t4] "=&x"(t4), [byteIndex] "+r"(byteIndex)
+			: [vLo] "x"(vLo), [vHi] "x"(vHi), [vertices] "r"(vertices), [sap] "r"(sap)
+			: "memory", "cc");
+		index += localCount / 4;
 #else
-        {
-            for( unsigned int i=0; i<localCount/4; i++,index++)   
-            { // do four dot products at a time. Carefully avoid touching the w element.
-                float4 v0 = vertices[0];
-                float4 v1 = vertices[1];
-                float4 v2 = vertices[2];
-                float4 v3 = vertices[3];            
-                vertices += 4;
-                
-                float4 lo0 = _mm_movelh_ps( v0, v1);    // x0y0x1y1
-                float4 hi0 = _mm_movehl_ps( v1, v0);    // z0?0z1?1
-                float4 lo1 = _mm_movelh_ps( v2, v3);    // x2y2x3y3
-                float4 hi1 = _mm_movehl_ps( v3, v2);    // z2?2z3?3
-                
-                lo0 = lo0*vLo;
-                lo1 = lo1*vLo;
-                float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
-                float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
-                float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-                z = z*vHi;
-                x = x+y;
-                x = x+z;
-                stack_array[index] = x;
-                min = _mm_min_ps( x, min );         // control the order here so that max is never NaN even if x is nan
-            }
-        }
+		{
+			for (unsigned int i = 0; i < localCount / 4; i++, index++)
+			{  // do four dot products at a time. Carefully avoid touching the w element.
+				float4 v0 = vertices[0];
+				float4 v1 = vertices[1];
+				float4 v2 = vertices[2];
+				float4 v3 = vertices[3];
+				vertices += 4;
+
+				float4 lo0 = _mm_movelh_ps(v0, v1);  // x0y0x1y1
+				float4 hi0 = _mm_movehl_ps(v1, v0);  // z0?0z1?1
+				float4 lo1 = _mm_movelh_ps(v2, v3);  // x2y2x3y3
+				float4 hi1 = _mm_movehl_ps(v3, v2);  // z2?2z3?3
+
+				lo0 = lo0 * vLo;
+				lo1 = lo1 * vLo;
+				float4 z = _mm_shuffle_ps(hi0, hi1, 0x88);
+				float4 x = _mm_shuffle_ps(lo0, lo1, 0x88);
+				float4 y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+				z = z * vHi;
+				x = x + y;
+				x = x + z;
+				stack_array[index] = x;
+				min = _mm_min_ps(x, min);  // control the order here so that max is never NaN even if x is nan
+			}
+		}
 
 #endif
-    }
-    
-    // process the last few points
-    if( count & 3 )
-    {
-        float4 v0, v1, v2, x, y, z;
-        switch( count & 3 )
-        {
-            case 3:
-            {
-                v0 = vertices[0];
-                v1 = vertices[1];
-                v2 = vertices[2];
-                
-                // Calculate 3 dot products, transpose, duplicate v2
-                float4 lo0 = _mm_movelh_ps( v0, v1);        // xyxy.lo
-                float4 hi0 = _mm_movehl_ps( v1, v0);        // z?z?.lo
-                lo0 = lo0*vLo;
-                z = _mm_shuffle_ps(hi0, v2,  0xa8 );           // z0z1z2z2
-                z = z*vHi;
-                float4 lo1 = _mm_movelh_ps(v2, v2);          // xyxy
-                lo1 = lo1*vLo;
-                x = _mm_shuffle_ps(lo0, lo1, 0x88);
-                y = _mm_shuffle_ps(lo0, lo1, 0xdd);
-            }
-                break;
-            case 2:
-            {
-                v0 = vertices[0];
-                v1 = vertices[1];
-                float4 xy = _mm_movelh_ps(v0, v1);
-                z = _mm_movehl_ps(v1, v0);
-                xy = xy*vLo;
-                z = _mm_shuffle_ps( z, z,  0xa8);
-                x = _mm_shuffle_ps( xy, xy, 0xa8);
-                y = _mm_shuffle_ps( xy, xy, 0xfd);
-                z = z*vHi;
-            }
-                break;
-            case 1:
-            {
-                float4 xy = vertices[0];
-                z =  _mm_shuffle_ps( xy, xy, 0xaa);
-                xy = xy*vLo;
-                z = z*vHi;
-                x = _mm_shuffle_ps(xy, xy, 0);
-                y = _mm_shuffle_ps(xy, xy, 0x55);
-            }
-                break;
-        }
-        x = x+y;
-        x = x+z;
-        stack_array[index] = x;
-        min = _mm_min_ps( x, min );         // control the order here so that min is never NaN even if x is nan
-        index++;
-    }
-    
-    // if we found a new min. 
-    if( 0 == segment || 0xf != _mm_movemask_ps( (float4) _mm_cmpeq_ps(min, dotmin)))
-    { // we found a new min. Search for it
-      // find min across the min vector, place in all elements of min -- big latency hit here
-        min = _mm_min_ps(min, (float4) _mm_shuffle_ps( min, min, 0x4e));
-        min = _mm_min_ps(min, (float4) _mm_shuffle_ps( min, min, 0xb1));
-        
-        // It is slightly faster to do this part in scalar code when count < 8. However, the common case for
-        // this where it actually makes a difference is handled in the early out at the top of the function, 
-        // so it is less than a 1% difference here. I opted for improved code size, fewer branches and reduced 
-        // complexity, and removed it.
-        
-        dotmin = min;
-        
-        // scan for the first occurence of min in the array  
-        size_t test;
-        for( index = 0; 0 == (test=_mm_movemask_ps( _mm_cmpeq_ps( stack_array[index], min))); index++ )   // local_count must be a multiple of 4
-        {}
-        minIndex = 4*index + segment + indexTable[test];
-    }
-    
-    _mm_store_ss( dotResult, dotmin);
-    return minIndex;
-}
+	}
 
+	// process the last few points
+	if (count & 3)
+	{
+		float4 v0, v1, v2, x, y, z;
+		switch (count & 3)
+		{
+			case 3:
+			{
+				v0 = vertices[0];
+				v1 = vertices[1];
+				v2 = vertices[2];
+
+				// Calculate 3 dot products, transpose, duplicate v2
+				float4 lo0 = _mm_movelh_ps(v0, v1);  // xyxy.lo
+				float4 hi0 = _mm_movehl_ps(v1, v0);  // z?z?.lo
+				lo0 = lo0 * vLo;
+				z = _mm_shuffle_ps(hi0, v2, 0xa8);  // z0z1z2z2
+				z = z * vHi;
+				float4 lo1 = _mm_movelh_ps(v2, v2);  // xyxy
+				lo1 = lo1 * vLo;
+				x = _mm_shuffle_ps(lo0, lo1, 0x88);
+				y = _mm_shuffle_ps(lo0, lo1, 0xdd);
+			}
+			break;
+			case 2:
+			{
+				v0 = vertices[0];
+				v1 = vertices[1];
+				float4 xy = _mm_movelh_ps(v0, v1);
+				z = _mm_movehl_ps(v1, v0);
+				xy = xy * vLo;
+				z = _mm_shuffle_ps(z, z, 0xa8);
+				x = _mm_shuffle_ps(xy, xy, 0xa8);
+				y = _mm_shuffle_ps(xy, xy, 0xfd);
+				z = z * vHi;
+			}
+			break;
+			case 1:
+			{
+				float4 xy = vertices[0];
+				z = _mm_shuffle_ps(xy, xy, 0xaa);
+				xy = xy * vLo;
+				z = z * vHi;
+				x = _mm_shuffle_ps(xy, xy, 0);
+				y = _mm_shuffle_ps(xy, xy, 0x55);
+			}
+			break;
+		}
+		x = x + y;
+		x = x + z;
+		stack_array[index] = x;
+		min = _mm_min_ps(x, min);  // control the order here so that min is never NaN even if x is nan
+		index++;
+	}
+
+	// if we found a new min.
+	if (0 == segment || 0xf != _mm_movemask_ps((float4)_mm_cmpeq_ps(min, dotmin)))
+	{  // we found a new min. Search for it
+		// find min across the min vector, place in all elements of min -- big latency hit here
+		min = _mm_min_ps(min, (float4)_mm_shuffle_ps(min, min, 0x4e));
+		min = _mm_min_ps(min, (float4)_mm_shuffle_ps(min, min, 0xb1));
+
+		// It is slightly faster to do this part in scalar code when count < 8. However, the common case for
+		// this where it actually makes a difference is handled in the early out at the top of the function,
+		// so it is less than a 1% difference here. I opted for improved code size, fewer branches and reduced
+		// complexity, and removed it.
+
+		dotmin = min;
+
+		// scan for the first occurence of min in the array
+		size_t test;
+		for (index = 0; 0 == (test = _mm_movemask_ps(_mm_cmpeq_ps(stack_array[index], min))); index++)  // local_count must be a multiple of 4
+		{
+		}
+		minIndex = 4 * index + segment + indexTable[test];
+	}
+
+	_mm_store_ss(dotResult, dotmin);
+	return minIndex;
+}
 
 #elif defined BT_USE_NEON
 
-#define ARM_NEON_GCC_COMPATIBILITY  1
+#define ARM_NEON_GCC_COMPATIBILITY 1
 #include <arm_neon.h>
 #include <sys/types.h>
-#include <sys/sysctl.h> //for sysctlbyname
+#include <sys/sysctl.h>  //for sysctlbyname
 
-static long _maxdot_large_v0( const float *vv, const float *vec, unsigned long count, float *dotResult );
-static long _maxdot_large_v1( const float *vv, const float *vec, unsigned long count, float *dotResult );
-static long _maxdot_large_sel( const float *vv, const float *vec, unsigned long count, float *dotResult );
-static long _mindot_large_v0( const float *vv, const float *vec, unsigned long count, float *dotResult );
-static long _mindot_large_v1( const float *vv, const float *vec, unsigned long count, float *dotResult );
-static long _mindot_large_sel( const float *vv, const float *vec, unsigned long count, float *dotResult );
+static long _maxdot_large_v0(const float *vv, const float *vec, unsigned long count, float *dotResult);
+static long _maxdot_large_v1(const float *vv, const float *vec, unsigned long count, float *dotResult);
+static long _maxdot_large_sel(const float *vv, const float *vec, unsigned long count, float *dotResult);
+static long _mindot_large_v0(const float *vv, const float *vec, unsigned long count, float *dotResult);
+static long _mindot_large_v1(const float *vv, const float *vec, unsigned long count, float *dotResult);
+static long _mindot_large_sel(const float *vv, const float *vec, unsigned long count, float *dotResult);
 
-long (*_maxdot_large)( const float *vv, const float *vec, unsigned long count, float *dotResult ) = _maxdot_large_sel;
-long (*_mindot_large)( const float *vv, const float *vec, unsigned long count, float *dotResult ) = _mindot_large_sel;
+long (*_maxdot_large)(const float *vv, const float *vec, unsigned long count, float *dotResult) = _maxdot_large_sel;
+long (*_mindot_large)(const float *vv, const float *vec, unsigned long count, float *dotResult) = _mindot_large_sel;
 
-
-static inline uint32_t btGetCpuCapabilities( void )
+static inline uint32_t btGetCpuCapabilities(void)
 {
-    static uint32_t capabilities = 0;
-    static bool testedCapabilities = false;
+	static uint32_t capabilities = 0;
+	static bool testedCapabilities = false;
 
-    if( 0 == testedCapabilities)
-    {
-        uint32_t hasFeature = 0;
-        size_t featureSize = sizeof( hasFeature );
-        int err = sysctlbyname( "hw.optional.neon_hpfp", &hasFeature, &featureSize, NULL, 0 );
+	if (0 == testedCapabilities)
+	{
+		uint32_t hasFeature = 0;
+		size_t featureSize = sizeof(hasFeature);
+		int err = sysctlbyname("hw.optional.neon_hpfp", &hasFeature, &featureSize, NULL, 0);
 
-        if( 0 == err && hasFeature)
-            capabilities |= 0x2000;
+		if (0 == err && hasFeature)
+			capabilities |= 0x2000;
 
 		testedCapabilities = true;
-    }
-    
-    return capabilities;
+	}
+
+	return capabilities;
 }
 
-
-
-
-static long _maxdot_large_sel( const float *vv, const float *vec, unsigned long count, float *dotResult )
+static long _maxdot_large_sel(const float *vv, const float *vec, unsigned long count, float *dotResult)
 {
+	if (btGetCpuCapabilities() & 0x2000)
+		_maxdot_large = _maxdot_large_v1;
+	else
+		_maxdot_large = _maxdot_large_v0;
 
-    if( btGetCpuCapabilities() & 0x2000 )
-        _maxdot_large = _maxdot_large_v1;
-    else
-        _maxdot_large = _maxdot_large_v0;
-    
-    return _maxdot_large(vv, vec, count, dotResult);
+	return _maxdot_large(vv, vec, count, dotResult);
 }
 
-static long _mindot_large_sel( const float *vv, const float *vec, unsigned long count, float *dotResult )
+static long _mindot_large_sel(const float *vv, const float *vec, unsigned long count, float *dotResult)
 {
+	if (btGetCpuCapabilities() & 0x2000)
+		_mindot_large = _mindot_large_v1;
+	else
+		_mindot_large = _mindot_large_v0;
 
-    if( btGetCpuCapabilities() & 0x2000 )
-        _mindot_large = _mindot_large_v1;
-    else
-        _mindot_large = _mindot_large_v0;
-    
-    return _mindot_large(vv, vec, count, dotResult);
+	return _mindot_large(vv, vec, count, dotResult);
 }
 
-
-
 #if defined __arm__
-# define vld1q_f32_aligned_postincrement( _ptr ) ({ float32x4_t _r; asm( "vld1.f32 {%0}, [%1, :128]!\n" : "=w" (_r), "+r" (_ptr) ); /*return*/ _r; })
+#define vld1q_f32_aligned_postincrement(_ptr) ({ float32x4_t _r; asm( "vld1.f32 {%0}, [%1, :128]!\n" : "=w" (_r), "+r" (_ptr) ); /*return*/ _r; })
 #else
 //support 64bit arm
-# define vld1q_f32_aligned_postincrement( _ptr) ({ float32x4_t _r = ((float32x4_t*)(_ptr))[0]; (_ptr) = (const float*) ((const char*)(_ptr) + 16L); /*return*/ _r; })
+#define vld1q_f32_aligned_postincrement(_ptr) ({ float32x4_t _r = ((float32x4_t*)(_ptr))[0]; (_ptr) = (const float*) ((const char*)(_ptr) + 16L); /*return*/ _r; })
 #endif
 
-
-long _maxdot_large_v0( const float *vv, const float *vec, unsigned long count, float *dotResult )
+long _maxdot_large_v0(const float *vv, const float *vec, unsigned long count, float *dotResult)
 {
-    unsigned long i = 0;
-    float32x4_t vvec = vld1q_f32_aligned_postincrement( vec );
-    float32x2_t vLo = vget_low_f32(vvec);
-    float32x2_t vHi = vdup_lane_f32(vget_high_f32(vvec), 0);
-    float32x2_t dotMaxLo = (float32x2_t) { -BT_INFINITY, -BT_INFINITY };
-    float32x2_t dotMaxHi = (float32x2_t) { -BT_INFINITY, -BT_INFINITY };
-    uint32x2_t indexLo = (uint32x2_t) {0, 1};
-    uint32x2_t indexHi = (uint32x2_t) {2, 3};
-    uint32x2_t iLo = (uint32x2_t) {static_cast<uint32_t>(-1), static_cast<uint32_t>(-1)};
-    uint32x2_t iHi = (uint32x2_t) {static_cast<uint32_t>(-1), static_cast<uint32_t>(-1)};
-    const uint32x2_t four = (uint32x2_t) {4,4};
+	unsigned long i = 0;
+	float32x4_t vvec = vld1q_f32_aligned_postincrement(vec);
+	float32x2_t vLo = vget_low_f32(vvec);
+	float32x2_t vHi = vdup_lane_f32(vget_high_f32(vvec), 0);
+	float32x2_t dotMaxLo = (float32x2_t){-BT_INFINITY, -BT_INFINITY};
+	float32x2_t dotMaxHi = (float32x2_t){-BT_INFINITY, -BT_INFINITY};
+	uint32x2_t indexLo = (uint32x2_t){0, 1};
+	uint32x2_t indexHi = (uint32x2_t){2, 3};
+	uint32x2_t iLo = (uint32x2_t){static_cast<uint32_t>(-1), static_cast<uint32_t>(-1)};
+	uint32x2_t iHi = (uint32x2_t){static_cast<uint32_t>(-1), static_cast<uint32_t>(-1)};
+	const uint32x2_t four = (uint32x2_t){4, 4};
 
-    for( ; i+8 <= count; i+= 8 )
-    {
-        float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v3 = vld1q_f32_aligned_postincrement( vv );
-        
-        float32x2_t xy0 = vmul_f32( vget_low_f32(v0), vLo);
-        float32x2_t xy1 = vmul_f32( vget_low_f32(v1), vLo);
-        float32x2_t xy2 = vmul_f32( vget_low_f32(v2), vLo);
-        float32x2_t xy3 = vmul_f32( vget_low_f32(v3), vLo);
-        
-        float32x2x2_t z0 = vtrn_f32( vget_high_f32(v0), vget_high_f32(v1));
-        float32x2x2_t z1 = vtrn_f32( vget_high_f32(v2), vget_high_f32(v3));
-        float32x2_t zLo = vmul_f32( z0.val[0], vHi);
-        float32x2_t zHi = vmul_f32( z1.val[0], vHi);
-        
-        float32x2_t rLo = vpadd_f32( xy0, xy1);
-        float32x2_t rHi = vpadd_f32( xy2, xy3);
-        rLo = vadd_f32(rLo, zLo);
-        rHi = vadd_f32(rHi, zHi);
-        
-        uint32x2_t maskLo = vcgt_f32( rLo, dotMaxLo );
-        uint32x2_t maskHi = vcgt_f32( rHi, dotMaxHi );
-        dotMaxLo = vbsl_f32( maskLo, rLo, dotMaxLo);
-        dotMaxHi = vbsl_f32( maskHi, rHi, dotMaxHi);
-        iLo = vbsl_u32(maskLo, indexLo, iLo);
-        iHi = vbsl_u32(maskHi, indexHi, iHi);
-        indexLo = vadd_u32(indexLo, four); 
-        indexHi = vadd_u32(indexHi, four);
+	for (; i + 8 <= count; i += 8)
+	{
+		float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v3 = vld1q_f32_aligned_postincrement(vv);
 
-        v0 = vld1q_f32_aligned_postincrement( vv );
-        v1 = vld1q_f32_aligned_postincrement( vv );
-        v2 = vld1q_f32_aligned_postincrement( vv );
-        v3 = vld1q_f32_aligned_postincrement( vv );
-        
-        xy0 = vmul_f32( vget_low_f32(v0), vLo);
-        xy1 = vmul_f32( vget_low_f32(v1), vLo);
-        xy2 = vmul_f32( vget_low_f32(v2), vLo);
-        xy3 = vmul_f32( vget_low_f32(v3), vLo);
-        
-        z0 = vtrn_f32( vget_high_f32(v0), vget_high_f32(v1));
-        z1 = vtrn_f32( vget_high_f32(v2), vget_high_f32(v3));
-        zLo = vmul_f32( z0.val[0], vHi);
-        zHi = vmul_f32( z1.val[0], vHi);
-        
-        rLo = vpadd_f32( xy0, xy1);
-        rHi = vpadd_f32( xy2, xy3);
-        rLo = vadd_f32(rLo, zLo);
-        rHi = vadd_f32(rHi, zHi);
-        
-        maskLo = vcgt_f32( rLo, dotMaxLo );
-        maskHi = vcgt_f32( rHi, dotMaxHi );
-        dotMaxLo = vbsl_f32( maskLo, rLo, dotMaxLo);
-        dotMaxHi = vbsl_f32( maskHi, rHi, dotMaxHi);
-        iLo = vbsl_u32(maskLo, indexLo, iLo);
-        iHi = vbsl_u32(maskHi, indexHi, iHi);
-        indexLo = vadd_u32(indexLo, four);
-        indexHi = vadd_u32(indexHi, four);
-    }
+		float32x2_t xy0 = vmul_f32(vget_low_f32(v0), vLo);
+		float32x2_t xy1 = vmul_f32(vget_low_f32(v1), vLo);
+		float32x2_t xy2 = vmul_f32(vget_low_f32(v2), vLo);
+		float32x2_t xy3 = vmul_f32(vget_low_f32(v3), vLo);
 
-    for( ; i+4 <= count; i+= 4 )
-    {
-        float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v3 = vld1q_f32_aligned_postincrement( vv );
-        
-        float32x2_t xy0 = vmul_f32( vget_low_f32(v0), vLo);
-        float32x2_t xy1 = vmul_f32( vget_low_f32(v1), vLo);
-        float32x2_t xy2 = vmul_f32( vget_low_f32(v2), vLo);
-        float32x2_t xy3 = vmul_f32( vget_low_f32(v3), vLo);
-        
-        float32x2x2_t z0 = vtrn_f32( vget_high_f32(v0), vget_high_f32(v1));
-        float32x2x2_t z1 = vtrn_f32( vget_high_f32(v2), vget_high_f32(v3));
-        float32x2_t zLo = vmul_f32( z0.val[0], vHi);
-        float32x2_t zHi = vmul_f32( z1.val[0], vHi);
-        
-        float32x2_t rLo = vpadd_f32( xy0, xy1);
-        float32x2_t rHi = vpadd_f32( xy2, xy3);
-        rLo = vadd_f32(rLo, zLo);
-        rHi = vadd_f32(rHi, zHi);
-        
-        uint32x2_t maskLo = vcgt_f32( rLo, dotMaxLo );
-        uint32x2_t maskHi = vcgt_f32( rHi, dotMaxHi );
-        dotMaxLo = vbsl_f32( maskLo, rLo, dotMaxLo);
-        dotMaxHi = vbsl_f32( maskHi, rHi, dotMaxHi);
-        iLo = vbsl_u32(maskLo, indexLo, iLo);
-        iHi = vbsl_u32(maskHi, indexHi, iHi);
-        indexLo = vadd_u32(indexLo, four);
-        indexHi = vadd_u32(indexHi, four);
-    }
-    
-    switch( count & 3 )
-    {
-        case 3:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-            
-            float32x2_t xy0 = vmul_f32( vget_low_f32(v0), vLo);
-            float32x2_t xy1 = vmul_f32( vget_low_f32(v1), vLo);
-            float32x2_t xy2 = vmul_f32( vget_low_f32(v2), vLo);
-            
-            float32x2x2_t z0 = vtrn_f32( vget_high_f32(v0), vget_high_f32(v1));
-            float32x2_t zLo = vmul_f32( z0.val[0], vHi);
-            float32x2_t zHi = vmul_f32( vdup_lane_f32(vget_high_f32(v2), 0), vHi);
-            
-            float32x2_t rLo = vpadd_f32( xy0, xy1);
-            float32x2_t rHi = vpadd_f32( xy2, xy2);
-            rLo = vadd_f32(rLo, zLo);
-            rHi = vadd_f32(rHi, zHi);
-            
-            uint32x2_t maskLo = vcgt_f32( rLo, dotMaxLo );
-            uint32x2_t maskHi = vcgt_f32( rHi, dotMaxHi );
-            dotMaxLo = vbsl_f32( maskLo, rLo, dotMaxLo);
-            dotMaxHi = vbsl_f32( maskHi, rHi, dotMaxHi);
-            iLo = vbsl_u32(maskLo, indexLo, iLo);
-            iHi = vbsl_u32(maskHi, indexHi, iHi);
-        }
-            break;
-        case 2:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-            
-            float32x2_t xy0 = vmul_f32( vget_low_f32(v0), vLo);
-            float32x2_t xy1 = vmul_f32( vget_low_f32(v1), vLo);
-            
-            float32x2x2_t z0 = vtrn_f32( vget_high_f32(v0), vget_high_f32(v1));
-            float32x2_t zLo = vmul_f32( z0.val[0], vHi);
-            
-            float32x2_t rLo = vpadd_f32( xy0, xy1);
-            rLo = vadd_f32(rLo, zLo);
-            
-            uint32x2_t maskLo = vcgt_f32( rLo, dotMaxLo );
-            dotMaxLo = vbsl_f32( maskLo, rLo, dotMaxLo);
-            iLo = vbsl_u32(maskLo, indexLo, iLo);
-        }
-            break;
-        case 1:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            float32x2_t xy0 = vmul_f32( vget_low_f32(v0), vLo);
-            float32x2_t z0 = vdup_lane_f32(vget_high_f32(v0), 0);
-            float32x2_t zLo = vmul_f32( z0, vHi);
-            float32x2_t rLo = vpadd_f32( xy0, xy0);
-            rLo = vadd_f32(rLo, zLo);
-            uint32x2_t maskLo = vcgt_f32( rLo, dotMaxLo );
-            dotMaxLo = vbsl_f32( maskLo, rLo, dotMaxLo);
-            iLo = vbsl_u32(maskLo, indexLo, iLo);
-        }
-            break;
-        
-        default:
-            break;
-    }
-    
-    // select best answer between hi and lo results
-    uint32x2_t mask = vcgt_f32( dotMaxHi, dotMaxLo );
-    dotMaxLo = vbsl_f32(mask, dotMaxHi, dotMaxLo);
-    iLo = vbsl_u32(mask, iHi, iLo);
-    
-    // select best answer between even and odd results
-    dotMaxHi = vdup_lane_f32(dotMaxLo, 1);
-    iHi = vdup_lane_u32(iLo, 1);
-    mask = vcgt_f32( dotMaxHi, dotMaxLo );
-    dotMaxLo = vbsl_f32(mask, dotMaxHi, dotMaxLo);
-    iLo = vbsl_u32(mask, iHi, iLo);
-    
-    *dotResult = vget_lane_f32( dotMaxLo, 0);
-    return vget_lane_u32(iLo, 0);
+		float32x2x2_t z0 = vtrn_f32(vget_high_f32(v0), vget_high_f32(v1));
+		float32x2x2_t z1 = vtrn_f32(vget_high_f32(v2), vget_high_f32(v3));
+		float32x2_t zLo = vmul_f32(z0.val[0], vHi);
+		float32x2_t zHi = vmul_f32(z1.val[0], vHi);
+
+		float32x2_t rLo = vpadd_f32(xy0, xy1);
+		float32x2_t rHi = vpadd_f32(xy2, xy3);
+		rLo = vadd_f32(rLo, zLo);
+		rHi = vadd_f32(rHi, zHi);
+
+		uint32x2_t maskLo = vcgt_f32(rLo, dotMaxLo);
+		uint32x2_t maskHi = vcgt_f32(rHi, dotMaxHi);
+		dotMaxLo = vbsl_f32(maskLo, rLo, dotMaxLo);
+		dotMaxHi = vbsl_f32(maskHi, rHi, dotMaxHi);
+		iLo = vbsl_u32(maskLo, indexLo, iLo);
+		iHi = vbsl_u32(maskHi, indexHi, iHi);
+		indexLo = vadd_u32(indexLo, four);
+		indexHi = vadd_u32(indexHi, four);
+
+		v0 = vld1q_f32_aligned_postincrement(vv);
+		v1 = vld1q_f32_aligned_postincrement(vv);
+		v2 = vld1q_f32_aligned_postincrement(vv);
+		v3 = vld1q_f32_aligned_postincrement(vv);
+
+		xy0 = vmul_f32(vget_low_f32(v0), vLo);
+		xy1 = vmul_f32(vget_low_f32(v1), vLo);
+		xy2 = vmul_f32(vget_low_f32(v2), vLo);
+		xy3 = vmul_f32(vget_low_f32(v3), vLo);
+
+		z0 = vtrn_f32(vget_high_f32(v0), vget_high_f32(v1));
+		z1 = vtrn_f32(vget_high_f32(v2), vget_high_f32(v3));
+		zLo = vmul_f32(z0.val[0], vHi);
+		zHi = vmul_f32(z1.val[0], vHi);
+
+		rLo = vpadd_f32(xy0, xy1);
+		rHi = vpadd_f32(xy2, xy3);
+		rLo = vadd_f32(rLo, zLo);
+		rHi = vadd_f32(rHi, zHi);
+
+		maskLo = vcgt_f32(rLo, dotMaxLo);
+		maskHi = vcgt_f32(rHi, dotMaxHi);
+		dotMaxLo = vbsl_f32(maskLo, rLo, dotMaxLo);
+		dotMaxHi = vbsl_f32(maskHi, rHi, dotMaxHi);
+		iLo = vbsl_u32(maskLo, indexLo, iLo);
+		iHi = vbsl_u32(maskHi, indexHi, iHi);
+		indexLo = vadd_u32(indexLo, four);
+		indexHi = vadd_u32(indexHi, four);
+	}
+
+	for (; i + 4 <= count; i += 4)
+	{
+		float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v3 = vld1q_f32_aligned_postincrement(vv);
+
+		float32x2_t xy0 = vmul_f32(vget_low_f32(v0), vLo);
+		float32x2_t xy1 = vmul_f32(vget_low_f32(v1), vLo);
+		float32x2_t xy2 = vmul_f32(vget_low_f32(v2), vLo);
+		float32x2_t xy3 = vmul_f32(vget_low_f32(v3), vLo);
+
+		float32x2x2_t z0 = vtrn_f32(vget_high_f32(v0), vget_high_f32(v1));
+		float32x2x2_t z1 = vtrn_f32(vget_high_f32(v2), vget_high_f32(v3));
+		float32x2_t zLo = vmul_f32(z0.val[0], vHi);
+		float32x2_t zHi = vmul_f32(z1.val[0], vHi);
+
+		float32x2_t rLo = vpadd_f32(xy0, xy1);
+		float32x2_t rHi = vpadd_f32(xy2, xy3);
+		rLo = vadd_f32(rLo, zLo);
+		rHi = vadd_f32(rHi, zHi);
+
+		uint32x2_t maskLo = vcgt_f32(rLo, dotMaxLo);
+		uint32x2_t maskHi = vcgt_f32(rHi, dotMaxHi);
+		dotMaxLo = vbsl_f32(maskLo, rLo, dotMaxLo);
+		dotMaxHi = vbsl_f32(maskHi, rHi, dotMaxHi);
+		iLo = vbsl_u32(maskLo, indexLo, iLo);
+		iHi = vbsl_u32(maskHi, indexHi, iHi);
+		indexLo = vadd_u32(indexLo, four);
+		indexHi = vadd_u32(indexHi, four);
+	}
+
+	switch (count & 3)
+	{
+		case 3:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+
+			float32x2_t xy0 = vmul_f32(vget_low_f32(v0), vLo);
+			float32x2_t xy1 = vmul_f32(vget_low_f32(v1), vLo);
+			float32x2_t xy2 = vmul_f32(vget_low_f32(v2), vLo);
+
+			float32x2x2_t z0 = vtrn_f32(vget_high_f32(v0), vget_high_f32(v1));
+			float32x2_t zLo = vmul_f32(z0.val[0], vHi);
+			float32x2_t zHi = vmul_f32(vdup_lane_f32(vget_high_f32(v2), 0), vHi);
+
+			float32x2_t rLo = vpadd_f32(xy0, xy1);
+			float32x2_t rHi = vpadd_f32(xy2, xy2);
+			rLo = vadd_f32(rLo, zLo);
+			rHi = vadd_f32(rHi, zHi);
+
+			uint32x2_t maskLo = vcgt_f32(rLo, dotMaxLo);
+			uint32x2_t maskHi = vcgt_f32(rHi, dotMaxHi);
+			dotMaxLo = vbsl_f32(maskLo, rLo, dotMaxLo);
+			dotMaxHi = vbsl_f32(maskHi, rHi, dotMaxHi);
+			iLo = vbsl_u32(maskLo, indexLo, iLo);
+			iHi = vbsl_u32(maskHi, indexHi, iHi);
+		}
+		break;
+		case 2:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+
+			float32x2_t xy0 = vmul_f32(vget_low_f32(v0), vLo);
+			float32x2_t xy1 = vmul_f32(vget_low_f32(v1), vLo);
+
+			float32x2x2_t z0 = vtrn_f32(vget_high_f32(v0), vget_high_f32(v1));
+			float32x2_t zLo = vmul_f32(z0.val[0], vHi);
+
+			float32x2_t rLo = vpadd_f32(xy0, xy1);
+			rLo = vadd_f32(rLo, zLo);
+
+			uint32x2_t maskLo = vcgt_f32(rLo, dotMaxLo);
+			dotMaxLo = vbsl_f32(maskLo, rLo, dotMaxLo);
+			iLo = vbsl_u32(maskLo, indexLo, iLo);
+		}
+		break;
+		case 1:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+			float32x2_t xy0 = vmul_f32(vget_low_f32(v0), vLo);
+			float32x2_t z0 = vdup_lane_f32(vget_high_f32(v0), 0);
+			float32x2_t zLo = vmul_f32(z0, vHi);
+			float32x2_t rLo = vpadd_f32(xy0, xy0);
+			rLo = vadd_f32(rLo, zLo);
+			uint32x2_t maskLo = vcgt_f32(rLo, dotMaxLo);
+			dotMaxLo = vbsl_f32(maskLo, rLo, dotMaxLo);
+			iLo = vbsl_u32(maskLo, indexLo, iLo);
+		}
+		break;
+
+		default:
+			break;
+	}
+
+	// select best answer between hi and lo results
+	uint32x2_t mask = vcgt_f32(dotMaxHi, dotMaxLo);
+	dotMaxLo = vbsl_f32(mask, dotMaxHi, dotMaxLo);
+	iLo = vbsl_u32(mask, iHi, iLo);
+
+	// select best answer between even and odd results
+	dotMaxHi = vdup_lane_f32(dotMaxLo, 1);
+	iHi = vdup_lane_u32(iLo, 1);
+	mask = vcgt_f32(dotMaxHi, dotMaxLo);
+	dotMaxLo = vbsl_f32(mask, dotMaxHi, dotMaxLo);
+	iLo = vbsl_u32(mask, iHi, iLo);
+
+	*dotResult = vget_lane_f32(dotMaxLo, 0);
+	return vget_lane_u32(iLo, 0);
 }
 
-
-long _maxdot_large_v1( const float *vv, const float *vec, unsigned long count, float *dotResult )
+long _maxdot_large_v1(const float *vv, const float *vec, unsigned long count, float *dotResult)
 {
-    float32x4_t vvec = vld1q_f32_aligned_postincrement( vec );
-    float32x4_t vLo = vcombine_f32(vget_low_f32(vvec), vget_low_f32(vvec));
-    float32x4_t vHi = vdupq_lane_f32(vget_high_f32(vvec), 0);
-    const uint32x4_t four = (uint32x4_t){ 4, 4, 4, 4 };
-    uint32x4_t local_index = (uint32x4_t) {0, 1, 2, 3};
-    uint32x4_t index = (uint32x4_t) { static_cast<uint32_t>(-1), static_cast<uint32_t>(-1), static_cast<uint32_t>(-1), static_cast<uint32_t>(-1) };
-    float32x4_t maxDot = (float32x4_t) { -BT_INFINITY, -BT_INFINITY, -BT_INFINITY, -BT_INFINITY };
-    
-    unsigned long i = 0;
-    for( ; i + 8 <= count; i += 8 )
-    {
-        float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v3 = vld1q_f32_aligned_postincrement( vv );
-        
-        // the next two lines should resolve to a single vswp d, d
-        float32x4_t xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v1));
-        float32x4_t xy1 = vcombine_f32( vget_low_f32(v2), vget_low_f32(v3));
-        // the next two lines should resolve to a single vswp d, d
-        float32x4_t z0 = vcombine_f32( vget_high_f32(v0), vget_high_f32(v1));
-        float32x4_t z1 = vcombine_f32( vget_high_f32(v2), vget_high_f32(v3));
-        
-        xy0 = vmulq_f32(xy0, vLo);
-        xy1 = vmulq_f32(xy1, vLo);
-        
-        float32x4x2_t zb = vuzpq_f32( z0, z1);
-        float32x4_t z = vmulq_f32( zb.val[0], vHi);
-        float32x4x2_t xy = vuzpq_f32( xy0, xy1);
-        float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
-        x = vaddq_f32(x, z);
-        
-        uint32x4_t mask = vcgtq_f32(x, maxDot);
-        maxDot = vbslq_f32( mask, x, maxDot);
-        index = vbslq_u32(mask, local_index, index);
-        local_index = vaddq_u32(local_index, four);
+	float32x4_t vvec = vld1q_f32_aligned_postincrement(vec);
+	float32x4_t vLo = vcombine_f32(vget_low_f32(vvec), vget_low_f32(vvec));
+	float32x4_t vHi = vdupq_lane_f32(vget_high_f32(vvec), 0);
+	const uint32x4_t four = (uint32x4_t){4, 4, 4, 4};
+	uint32x4_t local_index = (uint32x4_t){0, 1, 2, 3};
+	uint32x4_t index = (uint32x4_t){static_cast<uint32_t>(-1), static_cast<uint32_t>(-1), static_cast<uint32_t>(-1), static_cast<uint32_t>(-1)};
+	float32x4_t maxDot = (float32x4_t){-BT_INFINITY, -BT_INFINITY, -BT_INFINITY, -BT_INFINITY};
 
-        v0 = vld1q_f32_aligned_postincrement( vv );
-        v1 = vld1q_f32_aligned_postincrement( vv );
-        v2 = vld1q_f32_aligned_postincrement( vv );
-        v3 = vld1q_f32_aligned_postincrement( vv );
-        
-        // the next two lines should resolve to a single vswp d, d
-        xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v1));
-        xy1 = vcombine_f32( vget_low_f32(v2), vget_low_f32(v3));
-        // the next two lines should resolve to a single vswp d, d
-        z0 = vcombine_f32( vget_high_f32(v0), vget_high_f32(v1));
-        z1 = vcombine_f32( vget_high_f32(v2), vget_high_f32(v3));
-        
-        xy0 = vmulq_f32(xy0, vLo);
-        xy1 = vmulq_f32(xy1, vLo);
-        
-        zb = vuzpq_f32( z0, z1);
-        z = vmulq_f32( zb.val[0], vHi);
-        xy = vuzpq_f32( xy0, xy1);
-        x = vaddq_f32(xy.val[0], xy.val[1]);
-        x = vaddq_f32(x, z);
-        
-        mask = vcgtq_f32(x, maxDot);
-        maxDot = vbslq_f32( mask, x, maxDot);
-        index = vbslq_u32(mask, local_index, index);
-        local_index = vaddq_u32(local_index, four);
-    }
+	unsigned long i = 0;
+	for (; i + 8 <= count; i += 8)
+	{
+		float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v3 = vld1q_f32_aligned_postincrement(vv);
 
-    for( ; i + 4 <= count; i += 4 )
-    {
-        float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v3 = vld1q_f32_aligned_postincrement( vv );
+		// the next two lines should resolve to a single vswp d, d
+		float32x4_t xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v1));
+		float32x4_t xy1 = vcombine_f32(vget_low_f32(v2), vget_low_f32(v3));
+		// the next two lines should resolve to a single vswp d, d
+		float32x4_t z0 = vcombine_f32(vget_high_f32(v0), vget_high_f32(v1));
+		float32x4_t z1 = vcombine_f32(vget_high_f32(v2), vget_high_f32(v3));
 
-        // the next two lines should resolve to a single vswp d, d
-        float32x4_t xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v1));
-        float32x4_t xy1 = vcombine_f32( vget_low_f32(v2), vget_low_f32(v3));
-        // the next two lines should resolve to a single vswp d, d
-        float32x4_t z0 = vcombine_f32( vget_high_f32(v0), vget_high_f32(v1));
-        float32x4_t z1 = vcombine_f32( vget_high_f32(v2), vget_high_f32(v3));
-        
-        xy0 = vmulq_f32(xy0, vLo);
-        xy1 = vmulq_f32(xy1, vLo);
-        
-        float32x4x2_t zb = vuzpq_f32( z0, z1);
-        float32x4_t z = vmulq_f32( zb.val[0], vHi);
-        float32x4x2_t xy = vuzpq_f32( xy0, xy1);
-        float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
-        x = vaddq_f32(x, z);
-        
-        uint32x4_t mask = vcgtq_f32(x, maxDot);
-        maxDot = vbslq_f32( mask, x, maxDot);
-        index = vbslq_u32(mask, local_index, index);
-        local_index = vaddq_u32(local_index, four);
-    }
-    
-    switch (count & 3) {
-        case 3:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-            
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v1));
-            float32x4_t xy1 = vcombine_f32( vget_low_f32(v2), vget_low_f32(v2));
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t z0 = vcombine_f32( vget_high_f32(v0), vget_high_f32(v1));
-            float32x4_t z1 = vcombine_f32( vget_high_f32(v2), vget_high_f32(v2));
-            
-            xy0 = vmulq_f32(xy0, vLo);
-            xy1 = vmulq_f32(xy1, vLo);
-            
-            float32x4x2_t zb = vuzpq_f32( z0, z1);
-            float32x4_t z = vmulq_f32( zb.val[0], vHi);
-            float32x4x2_t xy = vuzpq_f32( xy0, xy1);
-            float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
-            x = vaddq_f32(x, z);
-            
-            uint32x4_t mask = vcgtq_f32(x, maxDot);
-            maxDot = vbslq_f32( mask, x, maxDot);
-            index = vbslq_u32(mask, local_index, index);
-            local_index = vaddq_u32(local_index, four);
-        }
-            break;
+		xy0 = vmulq_f32(xy0, vLo);
+		xy1 = vmulq_f32(xy1, vLo);
 
-        case 2:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-            
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v1));
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t z0 = vcombine_f32( vget_high_f32(v0), vget_high_f32(v1));
-            
-            xy0 = vmulq_f32(xy0, vLo);
-            
-            float32x4x2_t zb = vuzpq_f32( z0, z0);
-            float32x4_t z = vmulq_f32( zb.val[0], vHi);
-            float32x4x2_t xy = vuzpq_f32( xy0, xy0);
-            float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
-            x = vaddq_f32(x, z);
-            
-            uint32x4_t mask = vcgtq_f32(x, maxDot);
-            maxDot = vbslq_f32( mask, x, maxDot);
-            index = vbslq_u32(mask, local_index, index);
-            local_index = vaddq_u32(local_index, four);
-        }
-            break;
+		float32x4x2_t zb = vuzpq_f32(z0, z1);
+		float32x4_t z = vmulq_f32(zb.val[0], vHi);
+		float32x4x2_t xy = vuzpq_f32(xy0, xy1);
+		float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
+		x = vaddq_f32(x, z);
 
-        case 1:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v0));
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t z = vdupq_lane_f32(vget_high_f32(v0), 0); 
-            
-            xy0 = vmulq_f32(xy0, vLo);
-            
-            z = vmulq_f32( z, vHi);
-            float32x4x2_t xy = vuzpq_f32( xy0, xy0);
-            float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
-            x = vaddq_f32(x, z);
-            
-            uint32x4_t mask = vcgtq_f32(x, maxDot);
-            maxDot = vbslq_f32( mask, x, maxDot);
-            index = vbslq_u32(mask, local_index, index);
-            local_index = vaddq_u32(local_index, four);
-        }
-            break;
+		uint32x4_t mask = vcgtq_f32(x, maxDot);
+		maxDot = vbslq_f32(mask, x, maxDot);
+		index = vbslq_u32(mask, local_index, index);
+		local_index = vaddq_u32(local_index, four);
 
-        default:
-            break;
-    }
-    
-    
-    // select best answer between hi and lo results
-    uint32x2_t mask = vcgt_f32( vget_high_f32(maxDot), vget_low_f32(maxDot));
-    float32x2_t maxDot2 = vbsl_f32(mask, vget_high_f32(maxDot), vget_low_f32(maxDot));
-    uint32x2_t index2 = vbsl_u32(mask, vget_high_u32(index), vget_low_u32(index));
-    
-    // select best answer between even and odd results
-    float32x2_t maxDotO = vdup_lane_f32(maxDot2, 1);
-    uint32x2_t indexHi = vdup_lane_u32(index2, 1);
-    mask = vcgt_f32( maxDotO, maxDot2 );
-    maxDot2 = vbsl_f32(mask, maxDotO, maxDot2);
-    index2 = vbsl_u32(mask, indexHi, index2);
-    
-    *dotResult = vget_lane_f32( maxDot2, 0);
-    return vget_lane_u32(index2, 0);
-    
+		v0 = vld1q_f32_aligned_postincrement(vv);
+		v1 = vld1q_f32_aligned_postincrement(vv);
+		v2 = vld1q_f32_aligned_postincrement(vv);
+		v3 = vld1q_f32_aligned_postincrement(vv);
+
+		// the next two lines should resolve to a single vswp d, d
+		xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v1));
+		xy1 = vcombine_f32(vget_low_f32(v2), vget_low_f32(v3));
+		// the next two lines should resolve to a single vswp d, d
+		z0 = vcombine_f32(vget_high_f32(v0), vget_high_f32(v1));
+		z1 = vcombine_f32(vget_high_f32(v2), vget_high_f32(v3));
+
+		xy0 = vmulq_f32(xy0, vLo);
+		xy1 = vmulq_f32(xy1, vLo);
+
+		zb = vuzpq_f32(z0, z1);
+		z = vmulq_f32(zb.val[0], vHi);
+		xy = vuzpq_f32(xy0, xy1);
+		x = vaddq_f32(xy.val[0], xy.val[1]);
+		x = vaddq_f32(x, z);
+
+		mask = vcgtq_f32(x, maxDot);
+		maxDot = vbslq_f32(mask, x, maxDot);
+		index = vbslq_u32(mask, local_index, index);
+		local_index = vaddq_u32(local_index, four);
+	}
+
+	for (; i + 4 <= count; i += 4)
+	{
+		float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v3 = vld1q_f32_aligned_postincrement(vv);
+
+		// the next two lines should resolve to a single vswp d, d
+		float32x4_t xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v1));
+		float32x4_t xy1 = vcombine_f32(vget_low_f32(v2), vget_low_f32(v3));
+		// the next two lines should resolve to a single vswp d, d
+		float32x4_t z0 = vcombine_f32(vget_high_f32(v0), vget_high_f32(v1));
+		float32x4_t z1 = vcombine_f32(vget_high_f32(v2), vget_high_f32(v3));
+
+		xy0 = vmulq_f32(xy0, vLo);
+		xy1 = vmulq_f32(xy1, vLo);
+
+		float32x4x2_t zb = vuzpq_f32(z0, z1);
+		float32x4_t z = vmulq_f32(zb.val[0], vHi);
+		float32x4x2_t xy = vuzpq_f32(xy0, xy1);
+		float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
+		x = vaddq_f32(x, z);
+
+		uint32x4_t mask = vcgtq_f32(x, maxDot);
+		maxDot = vbslq_f32(mask, x, maxDot);
+		index = vbslq_u32(mask, local_index, index);
+		local_index = vaddq_u32(local_index, four);
+	}
+
+	switch (count & 3)
+	{
+		case 3:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v1));
+			float32x4_t xy1 = vcombine_f32(vget_low_f32(v2), vget_low_f32(v2));
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t z0 = vcombine_f32(vget_high_f32(v0), vget_high_f32(v1));
+			float32x4_t z1 = vcombine_f32(vget_high_f32(v2), vget_high_f32(v2));
+
+			xy0 = vmulq_f32(xy0, vLo);
+			xy1 = vmulq_f32(xy1, vLo);
+
+			float32x4x2_t zb = vuzpq_f32(z0, z1);
+			float32x4_t z = vmulq_f32(zb.val[0], vHi);
+			float32x4x2_t xy = vuzpq_f32(xy0, xy1);
+			float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
+			x = vaddq_f32(x, z);
+
+			uint32x4_t mask = vcgtq_f32(x, maxDot);
+			maxDot = vbslq_f32(mask, x, maxDot);
+			index = vbslq_u32(mask, local_index, index);
+			local_index = vaddq_u32(local_index, four);
+		}
+		break;
+
+		case 2:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v1));
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t z0 = vcombine_f32(vget_high_f32(v0), vget_high_f32(v1));
+
+			xy0 = vmulq_f32(xy0, vLo);
+
+			float32x4x2_t zb = vuzpq_f32(z0, z0);
+			float32x4_t z = vmulq_f32(zb.val[0], vHi);
+			float32x4x2_t xy = vuzpq_f32(xy0, xy0);
+			float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
+			x = vaddq_f32(x, z);
+
+			uint32x4_t mask = vcgtq_f32(x, maxDot);
+			maxDot = vbslq_f32(mask, x, maxDot);
+			index = vbslq_u32(mask, local_index, index);
+			local_index = vaddq_u32(local_index, four);
+		}
+		break;
+
+		case 1:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v0));
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t z = vdupq_lane_f32(vget_high_f32(v0), 0);
+
+			xy0 = vmulq_f32(xy0, vLo);
+
+			z = vmulq_f32(z, vHi);
+			float32x4x2_t xy = vuzpq_f32(xy0, xy0);
+			float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
+			x = vaddq_f32(x, z);
+
+			uint32x4_t mask = vcgtq_f32(x, maxDot);
+			maxDot = vbslq_f32(mask, x, maxDot);
+			index = vbslq_u32(mask, local_index, index);
+			local_index = vaddq_u32(local_index, four);
+		}
+		break;
+
+		default:
+			break;
+	}
+
+	// select best answer between hi and lo results
+	uint32x2_t mask = vcgt_f32(vget_high_f32(maxDot), vget_low_f32(maxDot));
+	float32x2_t maxDot2 = vbsl_f32(mask, vget_high_f32(maxDot), vget_low_f32(maxDot));
+	uint32x2_t index2 = vbsl_u32(mask, vget_high_u32(index), vget_low_u32(index));
+
+	// select best answer between even and odd results
+	float32x2_t maxDotO = vdup_lane_f32(maxDot2, 1);
+	uint32x2_t indexHi = vdup_lane_u32(index2, 1);
+	mask = vcgt_f32(maxDotO, maxDot2);
+	maxDot2 = vbsl_f32(mask, maxDotO, maxDot2);
+	index2 = vbsl_u32(mask, indexHi, index2);
+
+	*dotResult = vget_lane_f32(maxDot2, 0);
+	return vget_lane_u32(index2, 0);
 }
 
-long _mindot_large_v0( const float *vv, const float *vec, unsigned long count, float *dotResult )
+long _mindot_large_v0(const float *vv, const float *vec, unsigned long count, float *dotResult)
 {
-    unsigned long i = 0;
-    float32x4_t vvec = vld1q_f32_aligned_postincrement( vec );
-    float32x2_t vLo = vget_low_f32(vvec);
-    float32x2_t vHi = vdup_lane_f32(vget_high_f32(vvec), 0);
-    float32x2_t dotMinLo = (float32x2_t) { BT_INFINITY, BT_INFINITY };
-    float32x2_t dotMinHi = (float32x2_t) { BT_INFINITY, BT_INFINITY };
-    uint32x2_t indexLo = (uint32x2_t) {0, 1};
-    uint32x2_t indexHi = (uint32x2_t) {2, 3};
-    uint32x2_t iLo = (uint32x2_t) {static_cast<uint32_t>(-1), static_cast<uint32_t>(-1)};
-    uint32x2_t iHi = (uint32x2_t) {static_cast<uint32_t>(-1), static_cast<uint32_t>(-1)};
-    const uint32x2_t four = (uint32x2_t) {4,4};
-    
-    for( ; i+8 <= count; i+= 8 )
-    {
-        float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v3 = vld1q_f32_aligned_postincrement( vv );
-        
-        float32x2_t xy0 = vmul_f32( vget_low_f32(v0), vLo);
-        float32x2_t xy1 = vmul_f32( vget_low_f32(v1), vLo);
-        float32x2_t xy2 = vmul_f32( vget_low_f32(v2), vLo);
-        float32x2_t xy3 = vmul_f32( vget_low_f32(v3), vLo);
-        
-        float32x2x2_t z0 = vtrn_f32( vget_high_f32(v0), vget_high_f32(v1));
-        float32x2x2_t z1 = vtrn_f32( vget_high_f32(v2), vget_high_f32(v3));
-        float32x2_t zLo = vmul_f32( z0.val[0], vHi);
-        float32x2_t zHi = vmul_f32( z1.val[0], vHi);
-        
-        float32x2_t rLo = vpadd_f32( xy0, xy1);
-        float32x2_t rHi = vpadd_f32( xy2, xy3);
-        rLo = vadd_f32(rLo, zLo);
-        rHi = vadd_f32(rHi, zHi);
-        
-        uint32x2_t maskLo = vclt_f32( rLo, dotMinLo );
-        uint32x2_t maskHi = vclt_f32( rHi, dotMinHi );
-        dotMinLo = vbsl_f32( maskLo, rLo, dotMinLo);
-        dotMinHi = vbsl_f32( maskHi, rHi, dotMinHi);
-        iLo = vbsl_u32(maskLo, indexLo, iLo);
-        iHi = vbsl_u32(maskHi, indexHi, iHi);
-        indexLo = vadd_u32(indexLo, four);
-        indexHi = vadd_u32(indexHi, four);
-        
-        v0 = vld1q_f32_aligned_postincrement( vv );
-        v1 = vld1q_f32_aligned_postincrement( vv );
-        v2 = vld1q_f32_aligned_postincrement( vv );
-        v3 = vld1q_f32_aligned_postincrement( vv );
-        
-        xy0 = vmul_f32( vget_low_f32(v0), vLo);
-        xy1 = vmul_f32( vget_low_f32(v1), vLo);
-        xy2 = vmul_f32( vget_low_f32(v2), vLo);
-        xy3 = vmul_f32( vget_low_f32(v3), vLo);
-        
-        z0 = vtrn_f32( vget_high_f32(v0), vget_high_f32(v1));
-        z1 = vtrn_f32( vget_high_f32(v2), vget_high_f32(v3));
-        zLo = vmul_f32( z0.val[0], vHi);
-        zHi = vmul_f32( z1.val[0], vHi);
-        
-        rLo = vpadd_f32( xy0, xy1);
-        rHi = vpadd_f32( xy2, xy3);
-        rLo = vadd_f32(rLo, zLo);
-        rHi = vadd_f32(rHi, zHi);
-        
-        maskLo = vclt_f32( rLo, dotMinLo );
-        maskHi = vclt_f32( rHi, dotMinHi );
-        dotMinLo = vbsl_f32( maskLo, rLo, dotMinLo);
-        dotMinHi = vbsl_f32( maskHi, rHi, dotMinHi);
-        iLo = vbsl_u32(maskLo, indexLo, iLo);
-        iHi = vbsl_u32(maskHi, indexHi, iHi);
-        indexLo = vadd_u32(indexLo, four);
-        indexHi = vadd_u32(indexHi, four);
-    }
+	unsigned long i = 0;
+	float32x4_t vvec = vld1q_f32_aligned_postincrement(vec);
+	float32x2_t vLo = vget_low_f32(vvec);
+	float32x2_t vHi = vdup_lane_f32(vget_high_f32(vvec), 0);
+	float32x2_t dotMinLo = (float32x2_t){BT_INFINITY, BT_INFINITY};
+	float32x2_t dotMinHi = (float32x2_t){BT_INFINITY, BT_INFINITY};
+	uint32x2_t indexLo = (uint32x2_t){0, 1};
+	uint32x2_t indexHi = (uint32x2_t){2, 3};
+	uint32x2_t iLo = (uint32x2_t){static_cast<uint32_t>(-1), static_cast<uint32_t>(-1)};
+	uint32x2_t iHi = (uint32x2_t){static_cast<uint32_t>(-1), static_cast<uint32_t>(-1)};
+	const uint32x2_t four = (uint32x2_t){4, 4};
 
-    for( ; i+4 <= count; i+= 4 )
-    {
-        float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v3 = vld1q_f32_aligned_postincrement( vv );
-        
-        float32x2_t xy0 = vmul_f32( vget_low_f32(v0), vLo);
-        float32x2_t xy1 = vmul_f32( vget_low_f32(v1), vLo);
-        float32x2_t xy2 = vmul_f32( vget_low_f32(v2), vLo);
-        float32x2_t xy3 = vmul_f32( vget_low_f32(v3), vLo);
-        
-        float32x2x2_t z0 = vtrn_f32( vget_high_f32(v0), vget_high_f32(v1));
-        float32x2x2_t z1 = vtrn_f32( vget_high_f32(v2), vget_high_f32(v3));
-        float32x2_t zLo = vmul_f32( z0.val[0], vHi);
-        float32x2_t zHi = vmul_f32( z1.val[0], vHi);
-        
-        float32x2_t rLo = vpadd_f32( xy0, xy1);
-        float32x2_t rHi = vpadd_f32( xy2, xy3);
-        rLo = vadd_f32(rLo, zLo);
-        rHi = vadd_f32(rHi, zHi);
-        
-        uint32x2_t maskLo = vclt_f32( rLo, dotMinLo );
-        uint32x2_t maskHi = vclt_f32( rHi, dotMinHi );
-        dotMinLo = vbsl_f32( maskLo, rLo, dotMinLo);
-        dotMinHi = vbsl_f32( maskHi, rHi, dotMinHi);
-        iLo = vbsl_u32(maskLo, indexLo, iLo);
-        iHi = vbsl_u32(maskHi, indexHi, iHi);
-        indexLo = vadd_u32(indexLo, four);
-        indexHi = vadd_u32(indexHi, four);
-    }
-    switch( count & 3 )
-    {
-        case 3:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-            
-            float32x2_t xy0 = vmul_f32( vget_low_f32(v0), vLo);
-            float32x2_t xy1 = vmul_f32( vget_low_f32(v1), vLo);
-            float32x2_t xy2 = vmul_f32( vget_low_f32(v2), vLo);
-            
-            float32x2x2_t z0 = vtrn_f32( vget_high_f32(v0), vget_high_f32(v1));
-            float32x2_t zLo = vmul_f32( z0.val[0], vHi);
-            float32x2_t zHi = vmul_f32( vdup_lane_f32(vget_high_f32(v2), 0), vHi);
-            
-            float32x2_t rLo = vpadd_f32( xy0, xy1);
-            float32x2_t rHi = vpadd_f32( xy2, xy2);
-            rLo = vadd_f32(rLo, zLo);
-            rHi = vadd_f32(rHi, zHi);
-            
-            uint32x2_t maskLo = vclt_f32( rLo, dotMinLo );
-            uint32x2_t maskHi = vclt_f32( rHi, dotMinHi );
-            dotMinLo = vbsl_f32( maskLo, rLo, dotMinLo);
-            dotMinHi = vbsl_f32( maskHi, rHi, dotMinHi);
-            iLo = vbsl_u32(maskLo, indexLo, iLo);
-            iHi = vbsl_u32(maskHi, indexHi, iHi);
-        }
-            break;
-        case 2:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-            
-            float32x2_t xy0 = vmul_f32( vget_low_f32(v0), vLo);
-            float32x2_t xy1 = vmul_f32( vget_low_f32(v1), vLo);
-            
-            float32x2x2_t z0 = vtrn_f32( vget_high_f32(v0), vget_high_f32(v1));
-            float32x2_t zLo = vmul_f32( z0.val[0], vHi);
-            
-            float32x2_t rLo = vpadd_f32( xy0, xy1);
-            rLo = vadd_f32(rLo, zLo);
-            
-            uint32x2_t maskLo = vclt_f32( rLo, dotMinLo );
-            dotMinLo = vbsl_f32( maskLo, rLo, dotMinLo);
-            iLo = vbsl_u32(maskLo, indexLo, iLo);
-        }
-            break;
-        case 1:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            float32x2_t xy0 = vmul_f32( vget_low_f32(v0), vLo);
-            float32x2_t z0 = vdup_lane_f32(vget_high_f32(v0), 0);
-            float32x2_t zLo = vmul_f32( z0, vHi);
-            float32x2_t rLo = vpadd_f32( xy0, xy0);
-            rLo = vadd_f32(rLo, zLo);
-            uint32x2_t maskLo = vclt_f32( rLo, dotMinLo );
-            dotMinLo = vbsl_f32( maskLo, rLo, dotMinLo);
-            iLo = vbsl_u32(maskLo, indexLo, iLo);
-        }
-            break;
-            
-        default:
-            break;
-    }
-    
-    // select best answer between hi and lo results
-    uint32x2_t mask = vclt_f32( dotMinHi, dotMinLo );
-    dotMinLo = vbsl_f32(mask, dotMinHi, dotMinLo);
-    iLo = vbsl_u32(mask, iHi, iLo);
-    
-    // select best answer between even and odd results
-    dotMinHi = vdup_lane_f32(dotMinLo, 1);
-    iHi = vdup_lane_u32(iLo, 1);
-    mask = vclt_f32( dotMinHi, dotMinLo );
-    dotMinLo = vbsl_f32(mask, dotMinHi, dotMinLo);
-    iLo = vbsl_u32(mask, iHi, iLo);
-    
-    *dotResult = vget_lane_f32( dotMinLo, 0);
-    return vget_lane_u32(iLo, 0);
+	for (; i + 8 <= count; i += 8)
+	{
+		float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v3 = vld1q_f32_aligned_postincrement(vv);
+
+		float32x2_t xy0 = vmul_f32(vget_low_f32(v0), vLo);
+		float32x2_t xy1 = vmul_f32(vget_low_f32(v1), vLo);
+		float32x2_t xy2 = vmul_f32(vget_low_f32(v2), vLo);
+		float32x2_t xy3 = vmul_f32(vget_low_f32(v3), vLo);
+
+		float32x2x2_t z0 = vtrn_f32(vget_high_f32(v0), vget_high_f32(v1));
+		float32x2x2_t z1 = vtrn_f32(vget_high_f32(v2), vget_high_f32(v3));
+		float32x2_t zLo = vmul_f32(z0.val[0], vHi);
+		float32x2_t zHi = vmul_f32(z1.val[0], vHi);
+
+		float32x2_t rLo = vpadd_f32(xy0, xy1);
+		float32x2_t rHi = vpadd_f32(xy2, xy3);
+		rLo = vadd_f32(rLo, zLo);
+		rHi = vadd_f32(rHi, zHi);
+
+		uint32x2_t maskLo = vclt_f32(rLo, dotMinLo);
+		uint32x2_t maskHi = vclt_f32(rHi, dotMinHi);
+		dotMinLo = vbsl_f32(maskLo, rLo, dotMinLo);
+		dotMinHi = vbsl_f32(maskHi, rHi, dotMinHi);
+		iLo = vbsl_u32(maskLo, indexLo, iLo);
+		iHi = vbsl_u32(maskHi, indexHi, iHi);
+		indexLo = vadd_u32(indexLo, four);
+		indexHi = vadd_u32(indexHi, four);
+
+		v0 = vld1q_f32_aligned_postincrement(vv);
+		v1 = vld1q_f32_aligned_postincrement(vv);
+		v2 = vld1q_f32_aligned_postincrement(vv);
+		v3 = vld1q_f32_aligned_postincrement(vv);
+
+		xy0 = vmul_f32(vget_low_f32(v0), vLo);
+		xy1 = vmul_f32(vget_low_f32(v1), vLo);
+		xy2 = vmul_f32(vget_low_f32(v2), vLo);
+		xy3 = vmul_f32(vget_low_f32(v3), vLo);
+
+		z0 = vtrn_f32(vget_high_f32(v0), vget_high_f32(v1));
+		z1 = vtrn_f32(vget_high_f32(v2), vget_high_f32(v3));
+		zLo = vmul_f32(z0.val[0], vHi);
+		zHi = vmul_f32(z1.val[0], vHi);
+
+		rLo = vpadd_f32(xy0, xy1);
+		rHi = vpadd_f32(xy2, xy3);
+		rLo = vadd_f32(rLo, zLo);
+		rHi = vadd_f32(rHi, zHi);
+
+		maskLo = vclt_f32(rLo, dotMinLo);
+		maskHi = vclt_f32(rHi, dotMinHi);
+		dotMinLo = vbsl_f32(maskLo, rLo, dotMinLo);
+		dotMinHi = vbsl_f32(maskHi, rHi, dotMinHi);
+		iLo = vbsl_u32(maskLo, indexLo, iLo);
+		iHi = vbsl_u32(maskHi, indexHi, iHi);
+		indexLo = vadd_u32(indexLo, four);
+		indexHi = vadd_u32(indexHi, four);
+	}
+
+	for (; i + 4 <= count; i += 4)
+	{
+		float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v3 = vld1q_f32_aligned_postincrement(vv);
+
+		float32x2_t xy0 = vmul_f32(vget_low_f32(v0), vLo);
+		float32x2_t xy1 = vmul_f32(vget_low_f32(v1), vLo);
+		float32x2_t xy2 = vmul_f32(vget_low_f32(v2), vLo);
+		float32x2_t xy3 = vmul_f32(vget_low_f32(v3), vLo);
+
+		float32x2x2_t z0 = vtrn_f32(vget_high_f32(v0), vget_high_f32(v1));
+		float32x2x2_t z1 = vtrn_f32(vget_high_f32(v2), vget_high_f32(v3));
+		float32x2_t zLo = vmul_f32(z0.val[0], vHi);
+		float32x2_t zHi = vmul_f32(z1.val[0], vHi);
+
+		float32x2_t rLo = vpadd_f32(xy0, xy1);
+		float32x2_t rHi = vpadd_f32(xy2, xy3);
+		rLo = vadd_f32(rLo, zLo);
+		rHi = vadd_f32(rHi, zHi);
+
+		uint32x2_t maskLo = vclt_f32(rLo, dotMinLo);
+		uint32x2_t maskHi = vclt_f32(rHi, dotMinHi);
+		dotMinLo = vbsl_f32(maskLo, rLo, dotMinLo);
+		dotMinHi = vbsl_f32(maskHi, rHi, dotMinHi);
+		iLo = vbsl_u32(maskLo, indexLo, iLo);
+		iHi = vbsl_u32(maskHi, indexHi, iHi);
+		indexLo = vadd_u32(indexLo, four);
+		indexHi = vadd_u32(indexHi, four);
+	}
+	switch (count & 3)
+	{
+		case 3:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+
+			float32x2_t xy0 = vmul_f32(vget_low_f32(v0), vLo);
+			float32x2_t xy1 = vmul_f32(vget_low_f32(v1), vLo);
+			float32x2_t xy2 = vmul_f32(vget_low_f32(v2), vLo);
+
+			float32x2x2_t z0 = vtrn_f32(vget_high_f32(v0), vget_high_f32(v1));
+			float32x2_t zLo = vmul_f32(z0.val[0], vHi);
+			float32x2_t zHi = vmul_f32(vdup_lane_f32(vget_high_f32(v2), 0), vHi);
+
+			float32x2_t rLo = vpadd_f32(xy0, xy1);
+			float32x2_t rHi = vpadd_f32(xy2, xy2);
+			rLo = vadd_f32(rLo, zLo);
+			rHi = vadd_f32(rHi, zHi);
+
+			uint32x2_t maskLo = vclt_f32(rLo, dotMinLo);
+			uint32x2_t maskHi = vclt_f32(rHi, dotMinHi);
+			dotMinLo = vbsl_f32(maskLo, rLo, dotMinLo);
+			dotMinHi = vbsl_f32(maskHi, rHi, dotMinHi);
+			iLo = vbsl_u32(maskLo, indexLo, iLo);
+			iHi = vbsl_u32(maskHi, indexHi, iHi);
+		}
+		break;
+		case 2:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+
+			float32x2_t xy0 = vmul_f32(vget_low_f32(v0), vLo);
+			float32x2_t xy1 = vmul_f32(vget_low_f32(v1), vLo);
+
+			float32x2x2_t z0 = vtrn_f32(vget_high_f32(v0), vget_high_f32(v1));
+			float32x2_t zLo = vmul_f32(z0.val[0], vHi);
+
+			float32x2_t rLo = vpadd_f32(xy0, xy1);
+			rLo = vadd_f32(rLo, zLo);
+
+			uint32x2_t maskLo = vclt_f32(rLo, dotMinLo);
+			dotMinLo = vbsl_f32(maskLo, rLo, dotMinLo);
+			iLo = vbsl_u32(maskLo, indexLo, iLo);
+		}
+		break;
+		case 1:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+			float32x2_t xy0 = vmul_f32(vget_low_f32(v0), vLo);
+			float32x2_t z0 = vdup_lane_f32(vget_high_f32(v0), 0);
+			float32x2_t zLo = vmul_f32(z0, vHi);
+			float32x2_t rLo = vpadd_f32(xy0, xy0);
+			rLo = vadd_f32(rLo, zLo);
+			uint32x2_t maskLo = vclt_f32(rLo, dotMinLo);
+			dotMinLo = vbsl_f32(maskLo, rLo, dotMinLo);
+			iLo = vbsl_u32(maskLo, indexLo, iLo);
+		}
+		break;
+
+		default:
+			break;
+	}
+
+	// select best answer between hi and lo results
+	uint32x2_t mask = vclt_f32(dotMinHi, dotMinLo);
+	dotMinLo = vbsl_f32(mask, dotMinHi, dotMinLo);
+	iLo = vbsl_u32(mask, iHi, iLo);
+
+	// select best answer between even and odd results
+	dotMinHi = vdup_lane_f32(dotMinLo, 1);
+	iHi = vdup_lane_u32(iLo, 1);
+	mask = vclt_f32(dotMinHi, dotMinLo);
+	dotMinLo = vbsl_f32(mask, dotMinHi, dotMinLo);
+	iLo = vbsl_u32(mask, iHi, iLo);
+
+	*dotResult = vget_lane_f32(dotMinLo, 0);
+	return vget_lane_u32(iLo, 0);
 }
 
-long _mindot_large_v1( const float *vv, const float *vec, unsigned long count, float *dotResult )
+long _mindot_large_v1(const float *vv, const float *vec, unsigned long count, float *dotResult)
 {
-    float32x4_t vvec = vld1q_f32_aligned_postincrement( vec );
-    float32x4_t vLo = vcombine_f32(vget_low_f32(vvec), vget_low_f32(vvec));
-    float32x4_t vHi = vdupq_lane_f32(vget_high_f32(vvec), 0);
-    const uint32x4_t four = (uint32x4_t){ 4, 4, 4, 4 };
-    uint32x4_t local_index = (uint32x4_t) {0, 1, 2, 3};
-    uint32x4_t index = (uint32x4_t) { static_cast<uint32_t>(-1), static_cast<uint32_t>(-1), static_cast<uint32_t>(-1), static_cast<uint32_t>(-1) };
-    float32x4_t minDot = (float32x4_t) { BT_INFINITY, BT_INFINITY, BT_INFINITY, BT_INFINITY };
-    
-    unsigned long i = 0;
-    for( ; i + 8 <= count; i += 8 )
-    {
-        float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v3 = vld1q_f32_aligned_postincrement( vv );
-        
-        // the next two lines should resolve to a single vswp d, d
-        float32x4_t xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v1));
-        float32x4_t xy1 = vcombine_f32( vget_low_f32(v2), vget_low_f32(v3));
-        // the next two lines should resolve to a single vswp d, d
-        float32x4_t z0 = vcombine_f32( vget_high_f32(v0), vget_high_f32(v1));
-        float32x4_t z1 = vcombine_f32( vget_high_f32(v2), vget_high_f32(v3));
-        
-        xy0 = vmulq_f32(xy0, vLo);
-        xy1 = vmulq_f32(xy1, vLo);
-        
-        float32x4x2_t zb = vuzpq_f32( z0, z1);
-        float32x4_t z = vmulq_f32( zb.val[0], vHi);
-        float32x4x2_t xy = vuzpq_f32( xy0, xy1);
-        float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
-        x = vaddq_f32(x, z);
-        
-        uint32x4_t mask = vcltq_f32(x, minDot);
-        minDot = vbslq_f32( mask, x, minDot);
-        index = vbslq_u32(mask, local_index, index);
-        local_index = vaddq_u32(local_index, four);
-        
-        v0 = vld1q_f32_aligned_postincrement( vv );
-        v1 = vld1q_f32_aligned_postincrement( vv );
-        v2 = vld1q_f32_aligned_postincrement( vv );
-        v3 = vld1q_f32_aligned_postincrement( vv );
-        
-        // the next two lines should resolve to a single vswp d, d
-        xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v1));
-        xy1 = vcombine_f32( vget_low_f32(v2), vget_low_f32(v3));
-        // the next two lines should resolve to a single vswp d, d
-        z0 = vcombine_f32( vget_high_f32(v0), vget_high_f32(v1));
-        z1 = vcombine_f32( vget_high_f32(v2), vget_high_f32(v3));
-        
-        xy0 = vmulq_f32(xy0, vLo);
-        xy1 = vmulq_f32(xy1, vLo);
-        
-        zb = vuzpq_f32( z0, z1);
-        z = vmulq_f32( zb.val[0], vHi);
-        xy = vuzpq_f32( xy0, xy1);
-        x = vaddq_f32(xy.val[0], xy.val[1]);
-        x = vaddq_f32(x, z);
-        
-        mask = vcltq_f32(x, minDot);
-        minDot = vbslq_f32( mask, x, minDot);
-        index = vbslq_u32(mask, local_index, index);
-        local_index = vaddq_u32(local_index, four);
-    }
-    
-    for( ; i + 4 <= count; i += 4 )
-    {
-        float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-        float32x4_t v3 = vld1q_f32_aligned_postincrement( vv );
-        
-        // the next two lines should resolve to a single vswp d, d
-        float32x4_t xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v1));
-        float32x4_t xy1 = vcombine_f32( vget_low_f32(v2), vget_low_f32(v3));
-        // the next two lines should resolve to a single vswp d, d
-        float32x4_t z0 = vcombine_f32( vget_high_f32(v0), vget_high_f32(v1));
-        float32x4_t z1 = vcombine_f32( vget_high_f32(v2), vget_high_f32(v3));
-        
-        xy0 = vmulq_f32(xy0, vLo);
-        xy1 = vmulq_f32(xy1, vLo);
-        
-        float32x4x2_t zb = vuzpq_f32( z0, z1);
-        float32x4_t z = vmulq_f32( zb.val[0], vHi);
-        float32x4x2_t xy = vuzpq_f32( xy0, xy1);
-        float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
-        x = vaddq_f32(x, z);
-        
-        uint32x4_t mask = vcltq_f32(x, minDot);
-        minDot = vbslq_f32( mask, x, minDot);
-        index = vbslq_u32(mask, local_index, index);
-        local_index = vaddq_u32(local_index, four);
-    }
-    
-    switch (count & 3) {
-        case 3:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v2 = vld1q_f32_aligned_postincrement( vv );
-            
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v1));
-            float32x4_t xy1 = vcombine_f32( vget_low_f32(v2), vget_low_f32(v2));
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t z0 = vcombine_f32( vget_high_f32(v0), vget_high_f32(v1));
-            float32x4_t z1 = vcombine_f32( vget_high_f32(v2), vget_high_f32(v2));
-            
-            xy0 = vmulq_f32(xy0, vLo);
-            xy1 = vmulq_f32(xy1, vLo);
-            
-            float32x4x2_t zb = vuzpq_f32( z0, z1);
-            float32x4_t z = vmulq_f32( zb.val[0], vHi);
-            float32x4x2_t xy = vuzpq_f32( xy0, xy1);
-            float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
-            x = vaddq_f32(x, z);
-            
-            uint32x4_t mask = vcltq_f32(x, minDot);
-            minDot = vbslq_f32( mask, x, minDot);
-            index = vbslq_u32(mask, local_index, index);
-            local_index = vaddq_u32(local_index, four);
-        }
-            break;
-            
-        case 2:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            float32x4_t v1 = vld1q_f32_aligned_postincrement( vv );
-            
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v1));
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t z0 = vcombine_f32( vget_high_f32(v0), vget_high_f32(v1));
-            
-            xy0 = vmulq_f32(xy0, vLo);
-            
-            float32x4x2_t zb = vuzpq_f32( z0, z0);
-            float32x4_t z = vmulq_f32( zb.val[0], vHi);
-            float32x4x2_t xy = vuzpq_f32( xy0, xy0);
-            float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
-            x = vaddq_f32(x, z);
-            
-            uint32x4_t mask = vcltq_f32(x, minDot);
-            minDot = vbslq_f32( mask, x, minDot);
-            index = vbslq_u32(mask, local_index, index);
-            local_index = vaddq_u32(local_index, four);
-        }
-            break;
-            
-        case 1:
-        {
-            float32x4_t v0 = vld1q_f32_aligned_postincrement( vv );
-            
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t xy0 = vcombine_f32( vget_low_f32(v0), vget_low_f32(v0));
-            // the next two lines should resolve to a single vswp d, d
-            float32x4_t z = vdupq_lane_f32(vget_high_f32(v0), 0); 
-            
-            xy0 = vmulq_f32(xy0, vLo);
-            
-            z = vmulq_f32( z, vHi);
-            float32x4x2_t xy = vuzpq_f32( xy0, xy0);
-            float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
-            x = vaddq_f32(x, z);
-            
-            uint32x4_t mask = vcltq_f32(x, minDot);
-            minDot = vbslq_f32( mask, x, minDot);
-            index = vbslq_u32(mask, local_index, index);
-            local_index = vaddq_u32(local_index, four);
-        }
-            break;
-            
-        default:
-            break;
-    }
-    
-    
-    // select best answer between hi and lo results
-    uint32x2_t mask = vclt_f32( vget_high_f32(minDot), vget_low_f32(minDot));
-    float32x2_t minDot2 = vbsl_f32(mask, vget_high_f32(minDot), vget_low_f32(minDot));
-    uint32x2_t index2 = vbsl_u32(mask, vget_high_u32(index), vget_low_u32(index));
-    
-    // select best answer between even and odd results
-    float32x2_t minDotO = vdup_lane_f32(minDot2, 1);
-    uint32x2_t indexHi = vdup_lane_u32(index2, 1);
-    mask = vclt_f32( minDotO, minDot2 );
-    minDot2 = vbsl_f32(mask, minDotO, minDot2);
-    index2 = vbsl_u32(mask, indexHi, index2);
-    
-    *dotResult = vget_lane_f32( minDot2, 0);
-    return vget_lane_u32(index2, 0);
-    
+	float32x4_t vvec = vld1q_f32_aligned_postincrement(vec);
+	float32x4_t vLo = vcombine_f32(vget_low_f32(vvec), vget_low_f32(vvec));
+	float32x4_t vHi = vdupq_lane_f32(vget_high_f32(vvec), 0);
+	const uint32x4_t four = (uint32x4_t){4, 4, 4, 4};
+	uint32x4_t local_index = (uint32x4_t){0, 1, 2, 3};
+	uint32x4_t index = (uint32x4_t){static_cast<uint32_t>(-1), static_cast<uint32_t>(-1), static_cast<uint32_t>(-1), static_cast<uint32_t>(-1)};
+	float32x4_t minDot = (float32x4_t){BT_INFINITY, BT_INFINITY, BT_INFINITY, BT_INFINITY};
+
+	unsigned long i = 0;
+	for (; i + 8 <= count; i += 8)
+	{
+		float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v3 = vld1q_f32_aligned_postincrement(vv);
+
+		// the next two lines should resolve to a single vswp d, d
+		float32x4_t xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v1));
+		float32x4_t xy1 = vcombine_f32(vget_low_f32(v2), vget_low_f32(v3));
+		// the next two lines should resolve to a single vswp d, d
+		float32x4_t z0 = vcombine_f32(vget_high_f32(v0), vget_high_f32(v1));
+		float32x4_t z1 = vcombine_f32(vget_high_f32(v2), vget_high_f32(v3));
+
+		xy0 = vmulq_f32(xy0, vLo);
+		xy1 = vmulq_f32(xy1, vLo);
+
+		float32x4x2_t zb = vuzpq_f32(z0, z1);
+		float32x4_t z = vmulq_f32(zb.val[0], vHi);
+		float32x4x2_t xy = vuzpq_f32(xy0, xy1);
+		float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
+		x = vaddq_f32(x, z);
+
+		uint32x4_t mask = vcltq_f32(x, minDot);
+		minDot = vbslq_f32(mask, x, minDot);
+		index = vbslq_u32(mask, local_index, index);
+		local_index = vaddq_u32(local_index, four);
+
+		v0 = vld1q_f32_aligned_postincrement(vv);
+		v1 = vld1q_f32_aligned_postincrement(vv);
+		v2 = vld1q_f32_aligned_postincrement(vv);
+		v3 = vld1q_f32_aligned_postincrement(vv);
+
+		// the next two lines should resolve to a single vswp d, d
+		xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v1));
+		xy1 = vcombine_f32(vget_low_f32(v2), vget_low_f32(v3));
+		// the next two lines should resolve to a single vswp d, d
+		z0 = vcombine_f32(vget_high_f32(v0), vget_high_f32(v1));
+		z1 = vcombine_f32(vget_high_f32(v2), vget_high_f32(v3));
+
+		xy0 = vmulq_f32(xy0, vLo);
+		xy1 = vmulq_f32(xy1, vLo);
+
+		zb = vuzpq_f32(z0, z1);
+		z = vmulq_f32(zb.val[0], vHi);
+		xy = vuzpq_f32(xy0, xy1);
+		x = vaddq_f32(xy.val[0], xy.val[1]);
+		x = vaddq_f32(x, z);
+
+		mask = vcltq_f32(x, minDot);
+		minDot = vbslq_f32(mask, x, minDot);
+		index = vbslq_u32(mask, local_index, index);
+		local_index = vaddq_u32(local_index, four);
+	}
+
+	for (; i + 4 <= count; i += 4)
+	{
+		float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+		float32x4_t v3 = vld1q_f32_aligned_postincrement(vv);
+
+		// the next two lines should resolve to a single vswp d, d
+		float32x4_t xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v1));
+		float32x4_t xy1 = vcombine_f32(vget_low_f32(v2), vget_low_f32(v3));
+		// the next two lines should resolve to a single vswp d, d
+		float32x4_t z0 = vcombine_f32(vget_high_f32(v0), vget_high_f32(v1));
+		float32x4_t z1 = vcombine_f32(vget_high_f32(v2), vget_high_f32(v3));
+
+		xy0 = vmulq_f32(xy0, vLo);
+		xy1 = vmulq_f32(xy1, vLo);
+
+		float32x4x2_t zb = vuzpq_f32(z0, z1);
+		float32x4_t z = vmulq_f32(zb.val[0], vHi);
+		float32x4x2_t xy = vuzpq_f32(xy0, xy1);
+		float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
+		x = vaddq_f32(x, z);
+
+		uint32x4_t mask = vcltq_f32(x, minDot);
+		minDot = vbslq_f32(mask, x, minDot);
+		index = vbslq_u32(mask, local_index, index);
+		local_index = vaddq_u32(local_index, four);
+	}
+
+	switch (count & 3)
+	{
+		case 3:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v2 = vld1q_f32_aligned_postincrement(vv);
+
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v1));
+			float32x4_t xy1 = vcombine_f32(vget_low_f32(v2), vget_low_f32(v2));
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t z0 = vcombine_f32(vget_high_f32(v0), vget_high_f32(v1));
+			float32x4_t z1 = vcombine_f32(vget_high_f32(v2), vget_high_f32(v2));
+
+			xy0 = vmulq_f32(xy0, vLo);
+			xy1 = vmulq_f32(xy1, vLo);
+
+			float32x4x2_t zb = vuzpq_f32(z0, z1);
+			float32x4_t z = vmulq_f32(zb.val[0], vHi);
+			float32x4x2_t xy = vuzpq_f32(xy0, xy1);
+			float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
+			x = vaddq_f32(x, z);
+
+			uint32x4_t mask = vcltq_f32(x, minDot);
+			minDot = vbslq_f32(mask, x, minDot);
+			index = vbslq_u32(mask, local_index, index);
+			local_index = vaddq_u32(local_index, four);
+		}
+		break;
+
+		case 2:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+			float32x4_t v1 = vld1q_f32_aligned_postincrement(vv);
+
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v1));
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t z0 = vcombine_f32(vget_high_f32(v0), vget_high_f32(v1));
+
+			xy0 = vmulq_f32(xy0, vLo);
+
+			float32x4x2_t zb = vuzpq_f32(z0, z0);
+			float32x4_t z = vmulq_f32(zb.val[0], vHi);
+			float32x4x2_t xy = vuzpq_f32(xy0, xy0);
+			float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
+			x = vaddq_f32(x, z);
+
+			uint32x4_t mask = vcltq_f32(x, minDot);
+			minDot = vbslq_f32(mask, x, minDot);
+			index = vbslq_u32(mask, local_index, index);
+			local_index = vaddq_u32(local_index, four);
+		}
+		break;
+
+		case 1:
+		{
+			float32x4_t v0 = vld1q_f32_aligned_postincrement(vv);
+
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t xy0 = vcombine_f32(vget_low_f32(v0), vget_low_f32(v0));
+			// the next two lines should resolve to a single vswp d, d
+			float32x4_t z = vdupq_lane_f32(vget_high_f32(v0), 0);
+
+			xy0 = vmulq_f32(xy0, vLo);
+
+			z = vmulq_f32(z, vHi);
+			float32x4x2_t xy = vuzpq_f32(xy0, xy0);
+			float32x4_t x = vaddq_f32(xy.val[0], xy.val[1]);
+			x = vaddq_f32(x, z);
+
+			uint32x4_t mask = vcltq_f32(x, minDot);
+			minDot = vbslq_f32(mask, x, minDot);
+			index = vbslq_u32(mask, local_index, index);
+			local_index = vaddq_u32(local_index, four);
+		}
+		break;
+
+		default:
+			break;
+	}
+
+	// select best answer between hi and lo results
+	uint32x2_t mask = vclt_f32(vget_high_f32(minDot), vget_low_f32(minDot));
+	float32x2_t minDot2 = vbsl_f32(mask, vget_high_f32(minDot), vget_low_f32(minDot));
+	uint32x2_t index2 = vbsl_u32(mask, vget_high_u32(index), vget_low_u32(index));
+
+	// select best answer between even and odd results
+	float32x2_t minDotO = vdup_lane_f32(minDot2, 1);
+	uint32x2_t indexHi = vdup_lane_u32(index2, 1);
+	mask = vclt_f32(minDotO, minDot2);
+	minDot2 = vbsl_f32(mask, minDotO, minDot2);
+	index2 = vbsl_u32(mask, indexHi, index2);
+
+	*dotResult = vget_lane_f32(minDot2, 0);
+	return vget_lane_u32(index2, 0);
 }
 
 #else
-    #error Unhandled __APPLE__ arch
+#error Unhandled __APPLE__ arch
 #endif
 
-#endif  /* __APPLE__ */
-
-
+#endif /* __APPLE__ */
diff --git a/thirdparty/bullet/LinearMath/btVector3.h b/thirdparty/bullet/LinearMath/btVector3.h
index 76024f1236c..61fd8d1e46d 100644
--- a/thirdparty/bullet/LinearMath/btVector3.h
+++ b/thirdparty/bullet/LinearMath/btVector3.h
@@ -12,8 +12,6 @@ subject to the following restrictions:
 3. This notice may not be removed or altered from any source distribution.
 */
 
-
-
 #ifndef BT_VECTOR3_H
 #define BT_VECTOR3_H
 
@@ -28,25 +26,24 @@ subject to the following restrictions:
 #else
 #define btVector3Data btVector3FloatData
 #define btVector3DataName "btVector3FloatData"
-#endif //BT_USE_DOUBLE_PRECISION
+#endif  //BT_USE_DOUBLE_PRECISION
 
 #if defined BT_USE_SSE
 
 //typedef  uint32_t __m128i __attribute__ ((vector_size(16)));
 
 #ifdef _MSC_VER
-#pragma warning(disable: 4556) // value of intrinsic immediate argument '4294967239' is out of range '0 - 255'
+#pragma warning(disable : 4556)  // value of intrinsic immediate argument '4294967239' is out of range '0 - 255'
 #endif
 
-
-#define BT_SHUFFLE(x,y,z,w) ((w)<<6 | (z)<<4 | (y)<<2 | (x))
+#define BT_SHUFFLE(x, y, z, w) ((w) << 6 | (z) << 4 | (y) << 2 | (x))
 //#define bt_pshufd_ps( _a, _mask ) (__m128) _mm_shuffle_epi32((__m128i)(_a), (_mask) )
-#define bt_pshufd_ps( _a, _mask ) _mm_shuffle_ps((_a), (_a), (_mask) )
-#define bt_splat3_ps( _a, _i ) bt_pshufd_ps((_a), BT_SHUFFLE(_i,_i,_i, 3) )
-#define bt_splat_ps( _a, _i )  bt_pshufd_ps((_a), BT_SHUFFLE(_i,_i,_i,_i) )
+#define bt_pshufd_ps(_a, _mask) _mm_shuffle_ps((_a), (_a), (_mask))
+#define bt_splat3_ps(_a, _i) bt_pshufd_ps((_a), BT_SHUFFLE(_i, _i, _i, 3))
+#define bt_splat_ps(_a, _i) bt_pshufd_ps((_a), BT_SHUFFLE(_i, _i, _i, _i))
 
 #define btv3AbsiMask (_mm_set_epi32(0x00000000, 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF))
-#define btvAbsMask (_mm_set_epi32( 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF))
+#define btvAbsMask (_mm_set_epi32(0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF))
 #define btvFFF0Mask (_mm_set_epi32(0x00000000, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF))
 #define btv3AbsfMask btCastiTo128f(btv3AbsiMask)
 #define btvFFF0fMask btCastiTo128f(btvFFF0Mask)
@@ -55,9 +52,9 @@ subject to the following restrictions:
 
 //there is an issue with XCode 3.2 (LCx errors)
 #define btvMzeroMask (_mm_set_ps(-0.0f, -0.0f, -0.0f, -0.0f))
-#define v1110		 (_mm_set_ps(0.0f, 1.0f, 1.0f, 1.0f))
-#define vHalf		 (_mm_set_ps(0.5f, 0.5f, 0.5f, 0.5f))
-#define v1_5		 (_mm_set_ps(1.5f, 1.5f, 1.5f, 1.5f))
+#define v1110 (_mm_set_ps(0.0f, 1.0f, 1.0f, 1.0f))
+#define vHalf (_mm_set_ps(0.5f, 0.5f, 0.5f, 0.5f))
+#define v1_5 (_mm_set_ps(1.5f, 1.5f, 1.5f, 1.5f))
 
 //const __m128 ATTRIBUTE_ALIGNED16(btvMzeroMask) = {-0.0f, -0.0f, -0.0f, -0.0f};
 //const __m128 ATTRIBUTE_ALIGNED16(v1110) = {1.0f, 1.0f, 1.0f, 0.0f};
@@ -70,7 +67,7 @@ subject to the following restrictions:
 
 const float32x4_t ATTRIBUTE_ALIGNED16(btvMzeroMask) = (float32x4_t){-0.0f, -0.0f, -0.0f, -0.0f};
 const int32x4_t ATTRIBUTE_ALIGNED16(btvFFF0Mask) = (int32x4_t){static_cast<int32_t>(0xFFFFFFFF),
-	static_cast<int32_t>(0xFFFFFFFF), static_cast<int32_t>(0xFFFFFFFF), 0x0};
+															   static_cast<int32_t>(0xFFFFFFFF), static_cast<int32_t>(0xFFFFFFFF), 0x0};
 const int32x4_t ATTRIBUTE_ALIGNED16(btvAbsMask) = (int32x4_t){0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF};
 const int32x4_t ATTRIBUTE_ALIGNED16(btv3AbsMask) = (int32x4_t){0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF, 0x0};
 
@@ -80,50 +77,48 @@ const int32x4_t ATTRIBUTE_ALIGNED16(btv3AbsMask) = (int32x4_t){0x7FFFFFFF, 0x7FF
  * It has an un-used w component to suit 16-byte alignment when btVector3 is stored in containers. This extra component can be used by derived classes (Quaternion?) or by user
  * Ideally, this class should be replaced by a platform optimized SIMD version that keeps the data in registers
  */
-ATTRIBUTE_ALIGNED16(class) btVector3
+ATTRIBUTE_ALIGNED16(class)
+btVector3
 {
 public:
-
 	BT_DECLARE_ALIGNED_ALLOCATOR();
 
-#if defined (__SPU__) && defined (__CELLOS_LV2__)
-		btScalar	m_floats[4];
+#if defined(__SPU__) && defined(__CELLOS_LV2__)
+	btScalar m_floats[4];
+
 public:
-	SIMD_FORCE_INLINE const vec_float4&	get128() const
+	SIMD_FORCE_INLINE const vec_float4& get128() const
 	{
 		return *((const vec_float4*)&m_floats[0]);
 	}
+
 public:
-#else //__CELLOS_LV2__ __SPU__
-    #if defined (BT_USE_SSE) || defined(BT_USE_NEON) // _WIN32 || ARM
-        union {
-            btSimdFloat4      mVec128;
-            btScalar	m_floats[4];
-        };
-        SIMD_FORCE_INLINE	btSimdFloat4	get128() const
-        {
-            return mVec128;
-        }
-        SIMD_FORCE_INLINE	void	set128(btSimdFloat4 v128)
-        {
-            mVec128 = v128;
-        }
-    #else
-        btScalar	m_floats[4];
-    #endif
-#endif //__CELLOS_LV2__ __SPU__
-
-	public:
-
-  /**@brief No initialization constructor */
-	SIMD_FORCE_INLINE btVector3() 
+#else                                            //__CELLOS_LV2__ __SPU__
+#if defined(BT_USE_SSE) || defined(BT_USE_NEON)  // _WIN32 || ARM
+	union {
+		btSimdFloat4 mVec128;
+		btScalar m_floats[4];
+	};
+	SIMD_FORCE_INLINE btSimdFloat4 get128() const
 	{
+		return mVec128;
+	}
+	SIMD_FORCE_INLINE void set128(btSimdFloat4 v128)
+	{
+		mVec128 = v128;
+	}
+#else
+	btScalar m_floats[4];
+#endif
+#endif  //__CELLOS_LV2__ __SPU__
 
+public:
+	/**@brief No initialization constructor */
+	SIMD_FORCE_INLINE btVector3()
+	{
 	}
 
- 
-	
-  /**@brief Constructor from scalars 
+	/**@brief Constructor from scalars 
    * @param x X value
    * @param y Y value 
    * @param z Z value 
@@ -136,9 +131,9 @@ public:
 		m_floats[3] = btScalar(0.f);
 	}
 
-#if (defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE) )|| defined (BT_USE_NEON)
-	// Set Vector 
-	SIMD_FORCE_INLINE btVector3( btSimdFloat4 v)
+#if (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)) || defined(BT_USE_NEON)
+	// Set Vector
+	SIMD_FORCE_INLINE btVector3(btSimdFloat4 v)
 	{
 		mVec128 = v;
 	}
@@ -150,73 +145,72 @@ public:
 	}
 
 	// Assignment Operator
-	SIMD_FORCE_INLINE btVector3& 
-	operator=(const btVector3& v) 
+	SIMD_FORCE_INLINE btVector3&
+	operator=(const btVector3& v)
 	{
 		mVec128 = v.mVec128;
-		
+
 		return *this;
 	}
-#endif // #if defined (BT_USE_SSE_IN_API) || defined (BT_USE_NEON) 
-    
-/**@brief Add a vector to this one 
+#endif  // #if defined (BT_USE_SSE_IN_API) || defined (BT_USE_NEON)
+
+	/**@brief Add a vector to this one 
  * @param The vector to add to this one */
 	SIMD_FORCE_INLINE btVector3& operator+=(const btVector3& v)
 	{
-#if defined(BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 		mVec128 = _mm_add_ps(mVec128, v.mVec128);
 #elif defined(BT_USE_NEON)
 		mVec128 = vaddq_f32(mVec128, v.mVec128);
 #else
-		m_floats[0] += v.m_floats[0]; 
+		m_floats[0] += v.m_floats[0];
 		m_floats[1] += v.m_floats[1];
 		m_floats[2] += v.m_floats[2];
 #endif
 		return *this;
 	}
 
-
-  /**@brief Subtract a vector from this one
+	/**@brief Subtract a vector from this one
    * @param The vector to subtract */
-	SIMD_FORCE_INLINE btVector3& operator-=(const btVector3& v) 
+	SIMD_FORCE_INLINE btVector3& operator-=(const btVector3& v)
 	{
-#if defined(BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 		mVec128 = _mm_sub_ps(mVec128, v.mVec128);
 #elif defined(BT_USE_NEON)
 		mVec128 = vsubq_f32(mVec128, v.mVec128);
 #else
-		m_floats[0] -= v.m_floats[0]; 
+		m_floats[0] -= v.m_floats[0];
 		m_floats[1] -= v.m_floats[1];
 		m_floats[2] -= v.m_floats[2];
 #endif
 		return *this;
 	}
-	
-  /**@brief Scale the vector
+
+	/**@brief Scale the vector
    * @param s Scale factor */
 	SIMD_FORCE_INLINE btVector3& operator*=(const btScalar& s)
 	{
-#if defined(BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
-		__m128	vs = _mm_load_ss(&s);	//	(S 0 0 0)
-		vs = bt_pshufd_ps(vs, 0x80);	//	(S S S 0.0)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
+		__m128 vs = _mm_load_ss(&s);  //	(S 0 0 0)
+		vs = bt_pshufd_ps(vs, 0x80);  //	(S S S 0.0)
 		mVec128 = _mm_mul_ps(mVec128, vs);
 #elif defined(BT_USE_NEON)
 		mVec128 = vmulq_n_f32(mVec128, s);
 #else
-		m_floats[0] *= s; 
+		m_floats[0] *= s;
 		m_floats[1] *= s;
 		m_floats[2] *= s;
 #endif
 		return *this;
 	}
 
-  /**@brief Inversely scale the vector 
+	/**@brief Inversely scale the vector 
    * @param s Scale factor to divide by */
-	SIMD_FORCE_INLINE btVector3& operator/=(const btScalar& s) 
+	SIMD_FORCE_INLINE btVector3& operator/=(const btScalar& s)
 	{
 		btFullAssert(s != btScalar(0.0));
 
-#if 0 //defined(BT_USE_SSE_IN_API)
+#if 0  //defined(BT_USE_SSE_IN_API)
 // this code is not faster !
 		__m128 vs = _mm_load_ss(&s);
 		vs = _mm_div_ss(v1110, vs);
@@ -230,11 +224,11 @@ public:
 #endif
 	}
 
-  /**@brief Return the dot product
+	/**@brief Return the dot product
    * @param v The other vector in the dot product */
 	SIMD_FORCE_INLINE btScalar dot(const btVector3& v) const
 	{
-#if defined BT_USE_SIMD_VECTOR3 && defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined BT_USE_SIMD_VECTOR3 && defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 		__m128 vd = _mm_mul_ps(mVec128, v.mVec128);
 		__m128 z = _mm_movehl_ps(vd, vd);
 		__m128 y = _mm_shuffle_ps(vd, vd, 0x55);
@@ -243,23 +237,23 @@ public:
 		return _mm_cvtss_f32(vd);
 #elif defined(BT_USE_NEON)
 		float32x4_t vd = vmulq_f32(mVec128, v.mVec128);
-		float32x2_t x = vpadd_f32(vget_low_f32(vd), vget_low_f32(vd));  
+		float32x2_t x = vpadd_f32(vget_low_f32(vd), vget_low_f32(vd));
 		x = vadd_f32(x, vget_high_f32(vd));
 		return vget_lane_f32(x, 0);
-#else	
-		return	m_floats[0] * v.m_floats[0] + 
-				m_floats[1] * v.m_floats[1] + 
-				m_floats[2] * v.m_floats[2];
+#else
+		return m_floats[0] * v.m_floats[0] +
+			   m_floats[1] * v.m_floats[1] +
+			   m_floats[2] * v.m_floats[2];
 #endif
 	}
 
-  /**@brief Return the length of the vector squared */
+	/**@brief Return the length of the vector squared */
 	SIMD_FORCE_INLINE btScalar length2() const
 	{
 		return dot(*this);
 	}
 
-  /**@brief Return the length of the vector */
+	/**@brief Return the length of the vector */
 	SIMD_FORCE_INLINE btScalar length() const
 	{
 		return btSqrt(length2());
@@ -267,7 +261,7 @@ public:
 
 	/**@brief Return the norm (length) of the vector */
 	SIMD_FORCE_INLINE btScalar norm() const
-	{		
+	{
 		return length();
 	}
 
@@ -276,24 +270,24 @@ public:
 	{
 		btScalar d = length2();
 		//workaround for some clang/gcc issue of sqrtf(tiny number) = -INF
-		if (d>SIMD_EPSILON)
+		if (d > SIMD_EPSILON)
 			return btSqrt(d);
 		return btScalar(0);
 	}
 
-  /**@brief Return the distance squared between the ends of this and another vector
+	/**@brief Return the distance squared between the ends of this and another vector
    * This is symantically treating the vector like a point */
 	SIMD_FORCE_INLINE btScalar distance2(const btVector3& v) const;
 
-  /**@brief Return the distance between the ends of this and another vector
+	/**@brief Return the distance between the ends of this and another vector
    * This is symantically treating the vector like a point */
 	SIMD_FORCE_INLINE btScalar distance(const btVector3& v) const;
 
-	SIMD_FORCE_INLINE btVector3& safeNormalize() 
+	SIMD_FORCE_INLINE btVector3& safeNormalize()
 	{
 		btScalar l2 = length2();
 		//triNormal.normalize();
-		if (l2 >= SIMD_EPSILON*SIMD_EPSILON)
+		if (l2 >= SIMD_EPSILON * SIMD_EPSILON)
 		{
 			(*this) /= btSqrt(l2);
 		}
@@ -304,100 +298,97 @@ public:
 		return *this;
 	}
 
-  /**@brief Normalize this vector 
+	/**@brief Normalize this vector 
    * x^2 + y^2 + z^2 = 1 */
-	SIMD_FORCE_INLINE btVector3& normalize() 
+	SIMD_FORCE_INLINE btVector3& normalize()
 	{
-		
 		btAssert(!fuzzyZero());
 
-#if defined(BT_USE_SSE_IN_API) && defined (BT_USE_SSE)		
-        // dot product first
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
+		// dot product first
 		__m128 vd = _mm_mul_ps(mVec128, mVec128);
 		__m128 z = _mm_movehl_ps(vd, vd);
 		__m128 y = _mm_shuffle_ps(vd, vd, 0x55);
 		vd = _mm_add_ss(vd, y);
 		vd = _mm_add_ss(vd, z);
-		
-        #if 0
+
+#if 0
         vd = _mm_sqrt_ss(vd);
 		vd = _mm_div_ss(v1110, vd);
 		vd = bt_splat_ps(vd, 0x80);
 		mVec128 = _mm_mul_ps(mVec128, vd);
-        #else
-        
-        // NR step 1/sqrt(x) - vd is x, y is output 
-        y = _mm_rsqrt_ss(vd); // estimate 
-        
-        //  one step NR 
-        z = v1_5;
-        vd = _mm_mul_ss(vd, vHalf); // vd * 0.5	
-        //x2 = vd;
-        vd = _mm_mul_ss(vd, y); // vd * 0.5 * y0
-        vd = _mm_mul_ss(vd, y); // vd * 0.5 * y0 * y0
-        z = _mm_sub_ss(z, vd);  // 1.5 - vd * 0.5 * y0 * y0 
+#else
 
-        y = _mm_mul_ss(y, z);   // y0 * (1.5 - vd * 0.5 * y0 * y0)
+		// NR step 1/sqrt(x) - vd is x, y is output
+		y = _mm_rsqrt_ss(vd);  // estimate
+
+		//  one step NR
+		z = v1_5;
+		vd = _mm_mul_ss(vd, vHalf);  // vd * 0.5
+		//x2 = vd;
+		vd = _mm_mul_ss(vd, y);  // vd * 0.5 * y0
+		vd = _mm_mul_ss(vd, y);  // vd * 0.5 * y0 * y0
+		z = _mm_sub_ss(z, vd);   // 1.5 - vd * 0.5 * y0 * y0
+
+		y = _mm_mul_ss(y, z);  // y0 * (1.5 - vd * 0.5 * y0 * y0)
 
 		y = bt_splat_ps(y, 0x80);
 		mVec128 = _mm_mul_ps(mVec128, y);
 
-        #endif
+#endif
 
-		
 		return *this;
-#else	
+#else
 		return *this /= length();
 #endif
 	}
 
-  /**@brief Return a normalized version of this vector */
+	/**@brief Return a normalized version of this vector */
 	SIMD_FORCE_INLINE btVector3 normalized() const;
 
-  /**@brief Return a rotated version of this vector
+	/**@brief Return a rotated version of this vector
    * @param wAxis The axis to rotate about 
    * @param angle The angle to rotate by */
-	SIMD_FORCE_INLINE btVector3 rotate( const btVector3& wAxis, const btScalar angle ) const;
+	SIMD_FORCE_INLINE btVector3 rotate(const btVector3& wAxis, const btScalar angle) const;
 
-  /**@brief Return the angle between this and another vector
+	/**@brief Return the angle between this and another vector
    * @param v The other vector */
-	SIMD_FORCE_INLINE btScalar angle(const btVector3& v) const 
+	SIMD_FORCE_INLINE btScalar angle(const btVector3& v) const
 	{
 		btScalar s = btSqrt(length2() * v.length2());
 		btFullAssert(s != btScalar(0.0));
 		return btAcos(dot(v) / s);
 	}
-	
-  /**@brief Return a vector with the absolute values of each element */
-	SIMD_FORCE_INLINE btVector3 absolute() const 
-	{
 
-#if defined BT_USE_SIMD_VECTOR3 && defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE) 
+	/**@brief Return a vector with the absolute values of each element */
+	SIMD_FORCE_INLINE btVector3 absolute() const
+	{
+#if defined BT_USE_SIMD_VECTOR3 && defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 		return btVector3(_mm_and_ps(mVec128, btv3AbsfMask));
 #elif defined(BT_USE_NEON)
 		return btVector3(vabsq_f32(mVec128));
-#else	
+#else
 		return btVector3(
-			btFabs(m_floats[0]), 
-			btFabs(m_floats[1]), 
+			btFabs(m_floats[0]),
+			btFabs(m_floats[1]),
 			btFabs(m_floats[2]));
 #endif
 	}
-	
-  /**@brief Return the cross product between this and another vector 
+
+	/**@brief Return the cross product between this and another vector 
    * @param v The other vector */
 	SIMD_FORCE_INLINE btVector3 cross(const btVector3& v) const
 	{
-#if defined(BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
-		__m128	T, V;
-		
-		T = bt_pshufd_ps(mVec128, BT_SHUFFLE(1, 2, 0, 3));	//	(Y Z X 0)
-		V = bt_pshufd_ps(v.mVec128, BT_SHUFFLE(1, 2, 0, 3));	//	(Y Z X 0)
-		
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
+		__m128 T, V;
+
+		T = bt_pshufd_ps(mVec128, BT_SHUFFLE(1, 2, 0, 3));    //	(Y Z X 0)
+		V = bt_pshufd_ps(v.mVec128, BT_SHUFFLE(1, 2, 0, 3));  //	(Y Z X 0)
+
 		V = _mm_mul_ps(V, mVec128);
 		T = _mm_mul_ps(T, v.mVec128);
 		V = _mm_sub_ps(V, T);
-		
+
 		V = bt_pshufd_ps(V, BT_SHUFFLE(1, 2, 0, 3));
 		return btVector3(V);
 #elif defined(BT_USE_NEON)
@@ -407,7 +398,7 @@ public:
 		float32x2_t Vlow = vget_low_f32(v.mVec128);
 		T = vcombine_f32(vext_f32(Tlow, vget_high_f32(mVec128), 1), Tlow);
 		V = vcombine_f32(vext_f32(Vlow, vget_high_f32(v.mVec128), 1), Vlow);
-		
+
 		V = vmulq_f32(V, mVec128);
 		T = vmulq_f32(T, v.mVec128);
 		V = vsubq_f32(V, T);
@@ -415,7 +406,7 @@ public:
 		// form (Y, Z, X, _);
 		V = vcombine_f32(vext_f32(Vlow, vget_high_f32(V), 1), Vlow);
 		V = (float32x4_t)vandq_s32((int32x4_t)V, btvFFF0Mask);
-		
+
 		return btVector3(V);
 #else
 		return btVector3(
@@ -427,18 +418,18 @@ public:
 
 	SIMD_FORCE_INLINE btScalar triple(const btVector3& v1, const btVector3& v2) const
 	{
-#if defined BT_USE_SIMD_VECTOR3 && defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined BT_USE_SIMD_VECTOR3 && defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 		// cross:
-		__m128 T = _mm_shuffle_ps(v1.mVec128, v1.mVec128, BT_SHUFFLE(1, 2, 0, 3));	//	(Y Z X 0)
-		__m128 V = _mm_shuffle_ps(v2.mVec128, v2.mVec128, BT_SHUFFLE(1, 2, 0, 3));	//	(Y Z X 0)
-		
+		__m128 T = _mm_shuffle_ps(v1.mVec128, v1.mVec128, BT_SHUFFLE(1, 2, 0, 3));  //	(Y Z X 0)
+		__m128 V = _mm_shuffle_ps(v2.mVec128, v2.mVec128, BT_SHUFFLE(1, 2, 0, 3));  //	(Y Z X 0)
+
 		V = _mm_mul_ps(V, v1.mVec128);
 		T = _mm_mul_ps(T, v2.mVec128);
 		V = _mm_sub_ps(V, T);
-		
+
 		V = _mm_shuffle_ps(V, V, BT_SHUFFLE(1, 2, 0, 3));
 
-		// dot: 
+		// dot:
 		V = _mm_mul_ps(V, mVec128);
 		__m128 z = _mm_movehl_ps(V, V);
 		__m128 y = _mm_shuffle_ps(V, V, 0x55);
@@ -454,7 +445,7 @@ public:
 		float32x2_t Vlow = vget_low_f32(v2.mVec128);
 		T = vcombine_f32(vext_f32(Tlow, vget_high_f32(v1.mVec128), 1), Tlow);
 		V = vcombine_f32(vext_f32(Vlow, vget_high_f32(v2.mVec128), 1), Vlow);
-		
+
 		V = vmulq_f32(V, v1.mVec128);
 		T = vmulq_f32(T, v2.mVec128);
 		V = vsubq_f32(V, T);
@@ -462,31 +453,30 @@ public:
 		// form (Y, Z, X, _);
 		V = vcombine_f32(vext_f32(Vlow, vget_high_f32(V), 1), Vlow);
 
-		// dot: 
+		// dot:
 		V = vmulq_f32(mVec128, V);
-		float32x2_t x = vpadd_f32(vget_low_f32(V), vget_low_f32(V));  
+		float32x2_t x = vpadd_f32(vget_low_f32(V), vget_low_f32(V));
 		x = vadd_f32(x, vget_high_f32(V));
 		return vget_lane_f32(x, 0);
 #else
-		return 
-			m_floats[0] * (v1.m_floats[1] * v2.m_floats[2] - v1.m_floats[2] * v2.m_floats[1]) + 
-			m_floats[1] * (v1.m_floats[2] * v2.m_floats[0] - v1.m_floats[0] * v2.m_floats[2]) + 
-			m_floats[2] * (v1.m_floats[0] * v2.m_floats[1] - v1.m_floats[1] * v2.m_floats[0]);
+		return m_floats[0] * (v1.m_floats[1] * v2.m_floats[2] - v1.m_floats[2] * v2.m_floats[1]) +
+			   m_floats[1] * (v1.m_floats[2] * v2.m_floats[0] - v1.m_floats[0] * v2.m_floats[2]) +
+			   m_floats[2] * (v1.m_floats[0] * v2.m_floats[1] - v1.m_floats[1] * v2.m_floats[0]);
 #endif
 	}
 
-  /**@brief Return the axis with the smallest value 
+	/**@brief Return the axis with the smallest value 
    * Note return values are 0,1,2 for x, y, or z */
 	SIMD_FORCE_INLINE int minAxis() const
 	{
-		return m_floats[0] < m_floats[1] ? (m_floats[0] <m_floats[2] ? 0 : 2) : (m_floats[1] <m_floats[2] ? 1 : 2);
+		return m_floats[0] < m_floats[1] ? (m_floats[0] < m_floats[2] ? 0 : 2) : (m_floats[1] < m_floats[2] ? 1 : 2);
 	}
 
-  /**@brief Return the axis with the largest value 
+	/**@brief Return the axis with the largest value 
    * Note return values are 0,1,2 for x, y, or z */
-	SIMD_FORCE_INLINE int maxAxis() const 
+	SIMD_FORCE_INLINE int maxAxis() const
 	{
-		return m_floats[0] < m_floats[1] ? (m_floats[1] <m_floats[2] ? 2 : 1) : (m_floats[0] <m_floats[2] ? 2 : 0);
+		return m_floats[0] < m_floats[1] ? (m_floats[1] < m_floats[2] ? 2 : 1) : (m_floats[0] < m_floats[2] ? 2 : 0);
 	}
 
 	SIMD_FORCE_INLINE int furthestAxis() const
@@ -494,23 +484,22 @@ public:
 		return absolute().minAxis();
 	}
 
-	SIMD_FORCE_INLINE int closestAxis() const 
+	SIMD_FORCE_INLINE int closestAxis() const
 	{
 		return absolute().maxAxis();
 	}
 
-	
 	SIMD_FORCE_INLINE void setInterpolate3(const btVector3& v0, const btVector3& v1, btScalar rt)
 	{
-#if defined(BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
-		__m128	vrt = _mm_load_ss(&rt);	//	(rt 0 0 0)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
+		__m128 vrt = _mm_load_ss(&rt);  //	(rt 0 0 0)
 		btScalar s = btScalar(1.0) - rt;
-		__m128	vs = _mm_load_ss(&s);	//	(S 0 0 0)
-		vs = bt_pshufd_ps(vs, 0x80);	//	(S S S 0.0)
+		__m128 vs = _mm_load_ss(&s);  //	(S 0 0 0)
+		vs = bt_pshufd_ps(vs, 0x80);  //	(S S S 0.0)
 		__m128 r0 = _mm_mul_ps(v0.mVec128, vs);
-		vrt = bt_pshufd_ps(vrt, 0x80);	//	(rt rt rt 0.0)
+		vrt = bt_pshufd_ps(vrt, 0x80);  //	(rt rt rt 0.0)
 		__m128 r1 = _mm_mul_ps(v1.mVec128, vrt);
-		__m128 tmp3 = _mm_add_ps(r0,r1);
+		__m128 tmp3 = _mm_add_ps(r0, r1);
 		mVec128 = tmp3;
 #elif defined(BT_USE_NEON)
 		float32x4_t vl = vsubq_f32(v1.mVec128, v0.mVec128);
@@ -526,101 +515,100 @@ public:
 #endif
 	}
 
-  /**@brief Return the linear interpolation between this and another vector 
+	/**@brief Return the linear interpolation between this and another vector 
    * @param v The other vector 
    * @param t The ration of this to v (t = 0 => return this, t=1 => return other) */
-	SIMD_FORCE_INLINE btVector3 lerp(const btVector3& v, const btScalar& t) const 
+	SIMD_FORCE_INLINE btVector3 lerp(const btVector3& v, const btScalar& t) const
 	{
-#if defined(BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
-		__m128	vt = _mm_load_ss(&t);	//	(t 0 0 0)
-		vt = bt_pshufd_ps(vt, 0x80);	//	(rt rt rt 0.0)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
+		__m128 vt = _mm_load_ss(&t);  //	(t 0 0 0)
+		vt = bt_pshufd_ps(vt, 0x80);  //	(rt rt rt 0.0)
 		__m128 vl = _mm_sub_ps(v.mVec128, mVec128);
 		vl = _mm_mul_ps(vl, vt);
 		vl = _mm_add_ps(vl, mVec128);
-		
+
 		return btVector3(vl);
 #elif defined(BT_USE_NEON)
 		float32x4_t vl = vsubq_f32(v.mVec128, mVec128);
 		vl = vmulq_n_f32(vl, t);
 		vl = vaddq_f32(vl, mVec128);
-		
+
 		return btVector3(vl);
-#else	
-		return 
-			btVector3(	m_floats[0] + (v.m_floats[0] - m_floats[0]) * t,
-						m_floats[1] + (v.m_floats[1] - m_floats[1]) * t,
-						m_floats[2] + (v.m_floats[2] - m_floats[2]) * t);
+#else
+		return btVector3(m_floats[0] + (v.m_floats[0] - m_floats[0]) * t,
+						 m_floats[1] + (v.m_floats[1] - m_floats[1]) * t,
+						 m_floats[2] + (v.m_floats[2] - m_floats[2]) * t);
 #endif
 	}
 
-  /**@brief Elementwise multiply this vector by the other 
+	/**@brief Elementwise multiply this vector by the other 
    * @param v The other vector */
 	SIMD_FORCE_INLINE btVector3& operator*=(const btVector3& v)
 	{
-#if defined(BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 		mVec128 = _mm_mul_ps(mVec128, v.mVec128);
 #elif defined(BT_USE_NEON)
 		mVec128 = vmulq_f32(mVec128, v.mVec128);
-#else	
-		m_floats[0] *= v.m_floats[0]; 
+#else
+		m_floats[0] *= v.m_floats[0];
 		m_floats[1] *= v.m_floats[1];
 		m_floats[2] *= v.m_floats[2];
 #endif
 		return *this;
 	}
 
-	 /**@brief Return the x value */
-		SIMD_FORCE_INLINE const btScalar& getX() const { return m_floats[0]; }
-  /**@brief Return the y value */
-		SIMD_FORCE_INLINE const btScalar& getY() const { return m_floats[1]; }
-  /**@brief Return the z value */
-		SIMD_FORCE_INLINE const btScalar& getZ() const { return m_floats[2]; }
-  /**@brief Set the x value */
-		SIMD_FORCE_INLINE void	setX(btScalar _x) { m_floats[0] = _x;};
-  /**@brief Set the y value */
-		SIMD_FORCE_INLINE void	setY(btScalar _y) { m_floats[1] = _y;};
-  /**@brief Set the z value */
-		SIMD_FORCE_INLINE void	setZ(btScalar _z) { m_floats[2] = _z;};
-  /**@brief Set the w value */
-		SIMD_FORCE_INLINE void	setW(btScalar _w) { m_floats[3] = _w;};
-  /**@brief Return the x value */
-		SIMD_FORCE_INLINE const btScalar& x() const { return m_floats[0]; }
-  /**@brief Return the y value */
-		SIMD_FORCE_INLINE const btScalar& y() const { return m_floats[1]; }
-  /**@brief Return the z value */
-		SIMD_FORCE_INLINE const btScalar& z() const { return m_floats[2]; }
-  /**@brief Return the w value */
-		SIMD_FORCE_INLINE const btScalar& w() const { return m_floats[3]; }
+	/**@brief Return the x value */
+	SIMD_FORCE_INLINE const btScalar& getX() const { return m_floats[0]; }
+	/**@brief Return the y value */
+	SIMD_FORCE_INLINE const btScalar& getY() const { return m_floats[1]; }
+	/**@brief Return the z value */
+	SIMD_FORCE_INLINE const btScalar& getZ() const { return m_floats[2]; }
+	/**@brief Set the x value */
+	SIMD_FORCE_INLINE void setX(btScalar _x) { m_floats[0] = _x; };
+	/**@brief Set the y value */
+	SIMD_FORCE_INLINE void setY(btScalar _y) { m_floats[1] = _y; };
+	/**@brief Set the z value */
+	SIMD_FORCE_INLINE void setZ(btScalar _z) { m_floats[2] = _z; };
+	/**@brief Set the w value */
+	SIMD_FORCE_INLINE void setW(btScalar _w) { m_floats[3] = _w; };
+	/**@brief Return the x value */
+	SIMD_FORCE_INLINE const btScalar& x() const { return m_floats[0]; }
+	/**@brief Return the y value */
+	SIMD_FORCE_INLINE const btScalar& y() const { return m_floats[1]; }
+	/**@brief Return the z value */
+	SIMD_FORCE_INLINE const btScalar& z() const { return m_floats[2]; }
+	/**@brief Return the w value */
+	SIMD_FORCE_INLINE const btScalar& w() const { return m_floats[3]; }
 
-	//SIMD_FORCE_INLINE btScalar&       operator[](int i)       { return (&m_floats[0])[i];	}      
+	//SIMD_FORCE_INLINE btScalar&       operator[](int i)       { return (&m_floats[0])[i];	}
 	//SIMD_FORCE_INLINE const btScalar& operator[](int i) const { return (&m_floats[0])[i]; }
 	///operator btScalar*() replaces operator[], using implicit conversion. We added operator != and operator == to avoid pointer comparisons.
-	SIMD_FORCE_INLINE	operator       btScalar *()       { return &m_floats[0]; }
-	SIMD_FORCE_INLINE	operator const btScalar *() const { return &m_floats[0]; }
+	SIMD_FORCE_INLINE operator btScalar*() { return &m_floats[0]; }
+	SIMD_FORCE_INLINE operator const btScalar*() const { return &m_floats[0]; }
 
-	SIMD_FORCE_INLINE	bool	operator==(const btVector3& other) const
+	SIMD_FORCE_INLINE bool operator==(const btVector3& other) const
 	{
-#if defined(BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
-        return (0xf == _mm_movemask_ps((__m128)_mm_cmpeq_ps(mVec128, other.mVec128)));
-#else 
-		return ((m_floats[3]==other.m_floats[3]) && 
-                (m_floats[2]==other.m_floats[2]) && 
-                (m_floats[1]==other.m_floats[1]) && 
-                (m_floats[0]==other.m_floats[0]));
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
+		return (0xf == _mm_movemask_ps((__m128)_mm_cmpeq_ps(mVec128, other.mVec128)));
+#else
+		return ((m_floats[3] == other.m_floats[3]) &&
+				(m_floats[2] == other.m_floats[2]) &&
+				(m_floats[1] == other.m_floats[1]) &&
+				(m_floats[0] == other.m_floats[0]));
 #endif
 	}
 
-	SIMD_FORCE_INLINE	bool	operator!=(const btVector3& other) const
+	SIMD_FORCE_INLINE bool operator!=(const btVector3& other) const
 	{
 		return !(*this == other);
 	}
 
-  /**@brief Set each element to the max of the current values and the values of another btVector3
+	/**@brief Set each element to the max of the current values and the values of another btVector3
    * @param other The other btVector3 to compare with 
    */
-	SIMD_FORCE_INLINE void	setMax(const btVector3& other)
+	SIMD_FORCE_INLINE void setMax(const btVector3& other)
 	{
-#if defined(BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 		mVec128 = _mm_max_ps(mVec128, other.mVec128);
 #elif defined(BT_USE_NEON)
 		mVec128 = vmaxq_f32(mVec128, other.mVec128);
@@ -632,12 +620,12 @@ public:
 #endif
 	}
 
-  /**@brief Set each element to the min of the current values and the values of another btVector3
+	/**@brief Set each element to the min of the current values and the values of another btVector3
    * @param other The other btVector3 to compare with 
    */
-	SIMD_FORCE_INLINE void	setMin(const btVector3& other)
+	SIMD_FORCE_INLINE void setMin(const btVector3& other)
 	{
-#if defined(BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 		mVec128 = _mm_min_ps(mVec128, other.mVec128);
 #elif defined(BT_USE_NEON)
 		mVec128 = vminq_f32(mVec128, other.mVec128);
@@ -649,156 +637,155 @@ public:
 #endif
 	}
 
-	SIMD_FORCE_INLINE void 	setValue(const btScalar& _x, const btScalar& _y, const btScalar& _z)
+	SIMD_FORCE_INLINE void setValue(const btScalar& _x, const btScalar& _y, const btScalar& _z)
 	{
-		m_floats[0]=_x;
-		m_floats[1]=_y;
-		m_floats[2]=_z;
+		m_floats[0] = _x;
+		m_floats[1] = _y;
+		m_floats[2] = _z;
 		m_floats[3] = btScalar(0.f);
 	}
 
-	void	getSkewSymmetricMatrix(btVector3* v0,btVector3* v1,btVector3* v2) const
+	void getSkewSymmetricMatrix(btVector3 * v0, btVector3 * v1, btVector3 * v2) const
 	{
-#if defined BT_USE_SIMD_VECTOR3 && defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
- 
-		__m128 V  = _mm_and_ps(mVec128, btvFFF0fMask);
+#if defined BT_USE_SIMD_VECTOR3 && defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
+
+		__m128 V = _mm_and_ps(mVec128, btvFFF0fMask);
 		__m128 V0 = _mm_xor_ps(btvMzeroMask, V);
 		__m128 V2 = _mm_movelh_ps(V0, V);
-		
+
 		__m128 V1 = _mm_shuffle_ps(V, V0, 0xCE);
-		
-        V0 = _mm_shuffle_ps(V0, V, 0xDB);
+
+		V0 = _mm_shuffle_ps(V0, V, 0xDB);
 		V2 = _mm_shuffle_ps(V2, V, 0xF9);
-		
+
 		v0->mVec128 = V0;
 		v1->mVec128 = V1;
 		v2->mVec128 = V2;
 #else
-		v0->setValue(0.		,-z()		,y());
-		v1->setValue(z()	,0.			,-x());
-		v2->setValue(-y()	,x()	,0.);
+		v0->setValue(0., -z(), y());
+		v1->setValue(z(), 0., -x());
+		v2->setValue(-y(), x(), 0.);
 #endif
 	}
 
 	void setZero()
 	{
-#if defined(BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 		mVec128 = (__m128)_mm_xor_ps(mVec128, mVec128);
 #elif defined(BT_USE_NEON)
-		int32x4_t vi = vdupq_n_s32(0); 
+		int32x4_t vi = vdupq_n_s32(0);
 		mVec128 = vreinterpretq_f32_s32(vi);
-#else	
-		setValue(btScalar(0.),btScalar(0.),btScalar(0.));
+#else
+		setValue(btScalar(0.), btScalar(0.), btScalar(0.));
 #endif
 	}
 
-	SIMD_FORCE_INLINE bool isZero() const 
+	SIMD_FORCE_INLINE bool isZero() const
 	{
 		return m_floats[0] == btScalar(0) && m_floats[1] == btScalar(0) && m_floats[2] == btScalar(0);
 	}
 
-
-	SIMD_FORCE_INLINE bool fuzzyZero() const 
+	SIMD_FORCE_INLINE bool fuzzyZero() const
 	{
-		return length2() < SIMD_EPSILON*SIMD_EPSILON;
+		return length2() < SIMD_EPSILON * SIMD_EPSILON;
 	}
 
-	SIMD_FORCE_INLINE	void	serialize(struct	btVector3Data& dataOut) const;
+	SIMD_FORCE_INLINE void serialize(struct btVector3Data & dataOut) const;
 
-	SIMD_FORCE_INLINE	void	deSerialize(const struct	btVector3DoubleData& dataIn);
+	SIMD_FORCE_INLINE void deSerialize(const struct btVector3DoubleData& dataIn);
 
-	SIMD_FORCE_INLINE	void	deSerialize(const struct	btVector3FloatData& dataIn);
+	SIMD_FORCE_INLINE void deSerialize(const struct btVector3FloatData& dataIn);
 
-	SIMD_FORCE_INLINE	void	serializeFloat(struct	btVector3FloatData& dataOut) const;
+	SIMD_FORCE_INLINE void serializeFloat(struct btVector3FloatData & dataOut) const;
 
-	SIMD_FORCE_INLINE	void	deSerializeFloat(const struct	btVector3FloatData& dataIn);
+	SIMD_FORCE_INLINE void deSerializeFloat(const struct btVector3FloatData& dataIn);
 
-	SIMD_FORCE_INLINE	void	serializeDouble(struct	btVector3DoubleData& dataOut) const;
+	SIMD_FORCE_INLINE void serializeDouble(struct btVector3DoubleData & dataOut) const;
 
-	SIMD_FORCE_INLINE	void	deSerializeDouble(const struct	btVector3DoubleData& dataIn);
-    
-        /**@brief returns index of maximum dot product between this and vectors in array[]
+	SIMD_FORCE_INLINE void deSerializeDouble(const struct btVector3DoubleData& dataIn);
+
+	/**@brief returns index of maximum dot product between this and vectors in array[]
          * @param array The other vectors 
          * @param array_count The number of other vectors 
          * @param dotOut The maximum dot product */
-        SIMD_FORCE_INLINE   long    maxDot( const btVector3 *array, long array_count, btScalar &dotOut ) const; 
+	SIMD_FORCE_INLINE long maxDot(const btVector3* array, long array_count, btScalar& dotOut) const;
 
-        /**@brief returns index of minimum dot product between this and vectors in array[]
+	/**@brief returns index of minimum dot product between this and vectors in array[]
          * @param array The other vectors 
          * @param array_count The number of other vectors 
-         * @param dotOut The minimum dot product */    
-        SIMD_FORCE_INLINE   long    minDot( const btVector3 *array, long array_count, btScalar &dotOut ) const; 
+         * @param dotOut The minimum dot product */
+	SIMD_FORCE_INLINE long minDot(const btVector3* array, long array_count, btScalar& dotOut) const;
 
-    /* create a vector as  btVector3( this->dot( btVector3 v0 ), this->dot( btVector3 v1), this->dot( btVector3 v2 ))  */
-    SIMD_FORCE_INLINE btVector3  dot3( const btVector3 &v0, const btVector3 &v1, const btVector3 &v2 ) const
-    {
-#if defined BT_USE_SIMD_VECTOR3 && defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+	/* create a vector as  btVector3( this->dot( btVector3 v0 ), this->dot( btVector3 v1), this->dot( btVector3 v2 ))  */
+	SIMD_FORCE_INLINE btVector3 dot3(const btVector3& v0, const btVector3& v1, const btVector3& v2) const
+	{
+#if defined BT_USE_SIMD_VECTOR3 && defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
+
+		__m128 a0 = _mm_mul_ps(v0.mVec128, this->mVec128);
+		__m128 a1 = _mm_mul_ps(v1.mVec128, this->mVec128);
+		__m128 a2 = _mm_mul_ps(v2.mVec128, this->mVec128);
+		__m128 b0 = _mm_unpacklo_ps(a0, a1);
+		__m128 b1 = _mm_unpackhi_ps(a0, a1);
+		__m128 b2 = _mm_unpacklo_ps(a2, _mm_setzero_ps());
+		__m128 r = _mm_movelh_ps(b0, b2);
+		r = _mm_add_ps(r, _mm_movehl_ps(b2, b0));
+		a2 = _mm_and_ps(a2, btvxyzMaskf);
+		r = _mm_add_ps(r, btCastdTo128f(_mm_move_sd(btCastfTo128d(a2), btCastfTo128d(b1))));
+		return btVector3(r);
 
-        __m128 a0 = _mm_mul_ps( v0.mVec128, this->mVec128 );
-        __m128 a1 = _mm_mul_ps( v1.mVec128, this->mVec128 );
-        __m128 a2 = _mm_mul_ps( v2.mVec128, this->mVec128 );
-        __m128 b0 = _mm_unpacklo_ps( a0, a1 );
-        __m128 b1 = _mm_unpackhi_ps( a0, a1 );
-        __m128 b2 = _mm_unpacklo_ps( a2, _mm_setzero_ps() );
-        __m128 r = _mm_movelh_ps( b0, b2 );
-        r = _mm_add_ps( r, _mm_movehl_ps( b2, b0 ));
-        a2 = _mm_and_ps( a2, btvxyzMaskf);
-        r = _mm_add_ps( r, btCastdTo128f (_mm_move_sd( btCastfTo128d(a2), btCastfTo128d(b1) )));
-        return btVector3(r);
-        
 #elif defined(BT_USE_NEON)
-        static const uint32x4_t xyzMask = (const uint32x4_t){ static_cast<uint32_t>(-1), static_cast<uint32_t>(-1), static_cast<uint32_t>(-1), 0 };
-        float32x4_t a0 = vmulq_f32( v0.mVec128, this->mVec128);
-        float32x4_t a1 = vmulq_f32( v1.mVec128, this->mVec128);
-        float32x4_t a2 = vmulq_f32( v2.mVec128, this->mVec128);
-        float32x2x2_t zLo = vtrn_f32( vget_high_f32(a0), vget_high_f32(a1));
-        a2 = (float32x4_t) vandq_u32((uint32x4_t) a2, xyzMask );
-        float32x2_t b0 = vadd_f32( vpadd_f32( vget_low_f32(a0), vget_low_f32(a1)), zLo.val[0] );
-        float32x2_t b1 = vpadd_f32( vpadd_f32( vget_low_f32(a2), vget_high_f32(a2)), vdup_n_f32(0.0f));
-        return btVector3( vcombine_f32(b0, b1) );
-#else	
-		return btVector3( dot(v0), dot(v1), dot(v2));
+		static const uint32x4_t xyzMask = (const uint32x4_t){static_cast<uint32_t>(-1), static_cast<uint32_t>(-1), static_cast<uint32_t>(-1), 0};
+		float32x4_t a0 = vmulq_f32(v0.mVec128, this->mVec128);
+		float32x4_t a1 = vmulq_f32(v1.mVec128, this->mVec128);
+		float32x4_t a2 = vmulq_f32(v2.mVec128, this->mVec128);
+		float32x2x2_t zLo = vtrn_f32(vget_high_f32(a0), vget_high_f32(a1));
+		a2 = (float32x4_t)vandq_u32((uint32x4_t)a2, xyzMask);
+		float32x2_t b0 = vadd_f32(vpadd_f32(vget_low_f32(a0), vget_low_f32(a1)), zLo.val[0]);
+		float32x2_t b1 = vpadd_f32(vpadd_f32(vget_low_f32(a2), vget_high_f32(a2)), vdup_n_f32(0.0f));
+		return btVector3(vcombine_f32(b0, b1));
+#else
+		return btVector3(dot(v0), dot(v1), dot(v2));
 #endif
-    }
+	}
 };
 
 /**@brief Return the sum of two vectors (Point symantics)*/
-SIMD_FORCE_INLINE btVector3 
-operator+(const btVector3& v1, const btVector3& v2) 
+SIMD_FORCE_INLINE btVector3
+operator+(const btVector3& v1, const btVector3& v2)
 {
-#if defined(BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 	return btVector3(_mm_add_ps(v1.mVec128, v2.mVec128));
 #elif defined(BT_USE_NEON)
 	return btVector3(vaddq_f32(v1.mVec128, v2.mVec128));
 #else
 	return btVector3(
-			v1.m_floats[0] + v2.m_floats[0], 
-			v1.m_floats[1] + v2.m_floats[1], 
-			v1.m_floats[2] + v2.m_floats[2]);
+		v1.m_floats[0] + v2.m_floats[0],
+		v1.m_floats[1] + v2.m_floats[1],
+		v1.m_floats[2] + v2.m_floats[2]);
 #endif
 }
 
 /**@brief Return the elementwise product of two vectors */
-SIMD_FORCE_INLINE btVector3 
-operator*(const btVector3& v1, const btVector3& v2) 
+SIMD_FORCE_INLINE btVector3
+operator*(const btVector3& v1, const btVector3& v2)
 {
-#if defined(BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 	return btVector3(_mm_mul_ps(v1.mVec128, v2.mVec128));
 #elif defined(BT_USE_NEON)
 	return btVector3(vmulq_f32(v1.mVec128, v2.mVec128));
 #else
 	return btVector3(
-			v1.m_floats[0] * v2.m_floats[0], 
-			v1.m_floats[1] * v2.m_floats[1], 
-			v1.m_floats[2] * v2.m_floats[2]);
+		v1.m_floats[0] * v2.m_floats[0],
+		v1.m_floats[1] * v2.m_floats[1],
+		v1.m_floats[2] * v2.m_floats[2]);
 #endif
 }
 
 /**@brief Return the difference between two vectors */
-SIMD_FORCE_INLINE btVector3 
+SIMD_FORCE_INLINE btVector3
 operator-(const btVector3& v1, const btVector3& v2)
 {
-#if defined BT_USE_SIMD_VECTOR3 && (defined(BT_USE_SSE_IN_API)  && defined(BT_USE_SSE))
+#if defined BT_USE_SIMD_VECTOR3 && (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE))
 
 	//	without _mm_and_ps this code causes slowdown in Concave moving
 	__m128 r = _mm_sub_ps(v1.mVec128, v2.mVec128);
@@ -808,33 +795,33 @@ operator-(const btVector3& v1, const btVector3& v2)
 	return btVector3((float32x4_t)vandq_s32((int32x4_t)r, btvFFF0Mask));
 #else
 	return btVector3(
-			v1.m_floats[0] - v2.m_floats[0], 
-			v1.m_floats[1] - v2.m_floats[1], 
-			v1.m_floats[2] - v2.m_floats[2]);
+		v1.m_floats[0] - v2.m_floats[0],
+		v1.m_floats[1] - v2.m_floats[1],
+		v1.m_floats[2] - v2.m_floats[2]);
 #endif
 }
 
 /**@brief Return the negative of the vector */
-SIMD_FORCE_INLINE btVector3 
+SIMD_FORCE_INLINE btVector3
 operator-(const btVector3& v)
 {
-#if defined BT_USE_SIMD_VECTOR3 && (defined(BT_USE_SSE_IN_API) && defined (BT_USE_SSE))
+#if defined BT_USE_SIMD_VECTOR3 && (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE))
 	__m128 r = _mm_xor_ps(v.mVec128, btvMzeroMask);
-	return btVector3(_mm_and_ps(r, btvFFF0fMask)); 
+	return btVector3(_mm_and_ps(r, btvFFF0fMask));
 #elif defined(BT_USE_NEON)
 	return btVector3((btSimdFloat4)veorq_s32((int32x4_t)v.mVec128, (int32x4_t)btvMzeroMask));
-#else	
+#else
 	return btVector3(-v.m_floats[0], -v.m_floats[1], -v.m_floats[2]);
 #endif
 }
 
 /**@brief Return the vector scaled by s */
-SIMD_FORCE_INLINE btVector3 
+SIMD_FORCE_INLINE btVector3
 operator*(const btVector3& v, const btScalar& s)
 {
-#if defined(BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
-	__m128	vs = _mm_load_ss(&s);	//	(S 0 0 0)
-	vs = bt_pshufd_ps(vs, 0x80);	//	(S S S 0.0)
+#if defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
+	__m128 vs = _mm_load_ss(&s);  //	(S 0 0 0)
+	vs = bt_pshufd_ps(vs, 0x80);  //	(S S S 0.0)
 	return btVector3(_mm_mul_ps(v.mVec128, vs));
 #elif defined(BT_USE_NEON)
 	float32x4_t r = vmulq_n_f32(v.mVec128, s);
@@ -845,10 +832,10 @@ operator*(const btVector3& v, const btScalar& s)
 }
 
 /**@brief Return the vector scaled by s */
-SIMD_FORCE_INLINE btVector3 
+SIMD_FORCE_INLINE btVector3
 operator*(const btScalar& s, const btVector3& v)
-{ 
-	return v * s; 
+{
+	return v * s;
 }
 
 /**@brief Return the vector inversely scaled by s */
@@ -856,7 +843,7 @@ SIMD_FORCE_INLINE btVector3
 operator/(const btVector3& v, const btScalar& s)
 {
 	btFullAssert(s != btScalar(0.0));
-#if 0 //defined(BT_USE_SSE_IN_API)
+#if 0  //defined(BT_USE_SSE_IN_API)
 // this code is not faster !
 	__m128 vs = _mm_load_ss(&s);
     vs = _mm_div_ss(v1110, vs);
@@ -872,67 +859,65 @@ operator/(const btVector3& v, const btScalar& s)
 SIMD_FORCE_INLINE btVector3
 operator/(const btVector3& v1, const btVector3& v2)
 {
-#if defined BT_USE_SIMD_VECTOR3 && (defined(BT_USE_SSE_IN_API)&& defined (BT_USE_SSE))
+#if defined BT_USE_SIMD_VECTOR3 && (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE))
 	__m128 vec = _mm_div_ps(v1.mVec128, v2.mVec128);
 	vec = _mm_and_ps(vec, btvFFF0fMask);
-	return btVector3(vec); 
+	return btVector3(vec);
 #elif defined(BT_USE_NEON)
 	float32x4_t x, y, v, m;
 
 	x = v1.mVec128;
 	y = v2.mVec128;
-	
-	v = vrecpeq_f32(y);			// v ~ 1/y
-	m = vrecpsq_f32(y, v);		// m = (2-v*y)
-	v = vmulq_f32(v, m);		// vv = v*m ~~ 1/y
-	m = vrecpsq_f32(y, v);		// mm = (2-vv*y)
-	v = vmulq_f32(v, x);		// x*vv
-	v = vmulq_f32(v, m);		// (x*vv)*(2-vv*y) = x*(vv(2-vv*y)) ~~~ x/y
+
+	v = vrecpeq_f32(y);     // v ~ 1/y
+	m = vrecpsq_f32(y, v);  // m = (2-v*y)
+	v = vmulq_f32(v, m);    // vv = v*m ~~ 1/y
+	m = vrecpsq_f32(y, v);  // mm = (2-vv*y)
+	v = vmulq_f32(v, x);    // x*vv
+	v = vmulq_f32(v, m);    // (x*vv)*(2-vv*y) = x*(vv(2-vv*y)) ~~~ x/y
 
 	return btVector3(v);
 #else
 	return btVector3(
-			v1.m_floats[0] / v2.m_floats[0], 
-			v1.m_floats[1] / v2.m_floats[1],
-			v1.m_floats[2] / v2.m_floats[2]);
+		v1.m_floats[0] / v2.m_floats[0],
+		v1.m_floats[1] / v2.m_floats[1],
+		v1.m_floats[2] / v2.m_floats[2]);
 #endif
 }
 
 /**@brief Return the dot product between two vectors */
-SIMD_FORCE_INLINE btScalar 
-btDot(const btVector3& v1, const btVector3& v2) 
-{ 
-	return v1.dot(v2); 
+SIMD_FORCE_INLINE btScalar
+btDot(const btVector3& v1, const btVector3& v2)
+{
+	return v1.dot(v2);
 }
 
-
 /**@brief Return the distance squared between two vectors */
 SIMD_FORCE_INLINE btScalar
-btDistance2(const btVector3& v1, const btVector3& v2) 
-{ 
-	return v1.distance2(v2); 
+btDistance2(const btVector3& v1, const btVector3& v2)
+{
+	return v1.distance2(v2);
 }
 
-
 /**@brief Return the distance between two vectors */
 SIMD_FORCE_INLINE btScalar
-btDistance(const btVector3& v1, const btVector3& v2) 
-{ 
-	return v1.distance(v2); 
+btDistance(const btVector3& v1, const btVector3& v2)
+{
+	return v1.distance(v2);
 }
 
 /**@brief Return the angle between two vectors */
 SIMD_FORCE_INLINE btScalar
-btAngle(const btVector3& v1, const btVector3& v2) 
-{ 
-	return v1.angle(v2); 
+btAngle(const btVector3& v1, const btVector3& v2)
+{
+	return v1.angle(v2);
 }
 
 /**@brief Return the cross product of two vectors */
-SIMD_FORCE_INLINE btVector3 
-btCross(const btVector3& v1, const btVector3& v2) 
-{ 
-	return v1.cross(v2); 
+SIMD_FORCE_INLINE btVector3
+btCross(const btVector3& v1, const btVector3& v2)
+{
+	return v1.cross(v2);
 }
 
 SIMD_FORCE_INLINE btScalar
@@ -945,14 +930,12 @@ btTriple(const btVector3& v1, const btVector3& v2, const btVector3& v3)
  * @param v1 One vector 
  * @param v2 The other vector 
  * @param t The ration of this to v (t = 0 => return v1, t=1 => return v2) */
-SIMD_FORCE_INLINE btVector3 
+SIMD_FORCE_INLINE btVector3
 lerp(const btVector3& v1, const btVector3& v2, const btScalar& t)
 {
 	return v1.lerp(v2, t);
 }
 
-
-
 SIMD_FORCE_INLINE btScalar btVector3::distance2(const btVector3& v) const
 {
 	return (v - *this).length2();
@@ -968,140 +951,137 @@ SIMD_FORCE_INLINE btVector3 btVector3::normalized() const
 	btVector3 nrm = *this;
 
 	return nrm.normalize();
-} 
+}
 
-SIMD_FORCE_INLINE btVector3 btVector3::rotate( const btVector3& wAxis, const btScalar _angle ) const
+SIMD_FORCE_INLINE btVector3 btVector3::rotate(const btVector3& wAxis, const btScalar _angle) const
 {
 	// wAxis must be a unit lenght vector
 
-#if defined BT_USE_SIMD_VECTOR3 && defined (BT_USE_SSE_IN_API) && defined (BT_USE_SSE)
+#if defined BT_USE_SIMD_VECTOR3 && defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 
-    __m128 O = _mm_mul_ps(wAxis.mVec128, mVec128);
-	btScalar ssin = btSin( _angle );
-    __m128 C = wAxis.cross( mVec128 ).mVec128;
+	__m128 O = _mm_mul_ps(wAxis.mVec128, mVec128);
+	btScalar ssin = btSin(_angle);
+	__m128 C = wAxis.cross(mVec128).mVec128;
 	O = _mm_and_ps(O, btvFFF0fMask);
-    btScalar scos = btCos( _angle );
-	
-	__m128 vsin = _mm_load_ss(&ssin);	//	(S 0 0 0)
-    __m128 vcos = _mm_load_ss(&scos);	//	(S 0 0 0)
-	
-	__m128 Y = bt_pshufd_ps(O, 0xC9);	//	(Y Z X 0)
-	__m128 Z = bt_pshufd_ps(O, 0xD2);	//	(Z X Y 0)
+	btScalar scos = btCos(_angle);
+
+	__m128 vsin = _mm_load_ss(&ssin);  //	(S 0 0 0)
+	__m128 vcos = _mm_load_ss(&scos);  //	(S 0 0 0)
+
+	__m128 Y = bt_pshufd_ps(O, 0xC9);  //	(Y Z X 0)
+	__m128 Z = bt_pshufd_ps(O, 0xD2);  //	(Z X Y 0)
 	O = _mm_add_ps(O, Y);
-	vsin = bt_pshufd_ps(vsin, 0x80);	//	(S S S 0)
+	vsin = bt_pshufd_ps(vsin, 0x80);  //	(S S S 0)
 	O = _mm_add_ps(O, Z);
-    vcos = bt_pshufd_ps(vcos, 0x80);	//	(S S S 0)
-	
-    vsin = vsin * C; 
-	O = O * wAxis.mVec128; 
-	__m128 X = mVec128 - O; 
-	
-    O = O + vsin;
+	vcos = bt_pshufd_ps(vcos, 0x80);  //	(S S S 0)
+
+	vsin = vsin * C;
+	O = O * wAxis.mVec128;
+	__m128 X = mVec128 - O;
+
+	O = O + vsin;
 	vcos = vcos * X;
-	O = O + vcos;	
-	
+	O = O + vcos;
+
 	return btVector3(O);
 #else
-	btVector3 o = wAxis * wAxis.dot( *this );
+	btVector3 o = wAxis * wAxis.dot(*this);
 	btVector3 _x = *this - o;
 	btVector3 _y;
 
-	_y = wAxis.cross( *this );
+	_y = wAxis.cross(*this);
 
-	return ( o + _x * btCos( _angle ) + _y * btSin( _angle ) );
+	return (o + _x * btCos(_angle) + _y * btSin(_angle));
 #endif
 }
 
-SIMD_FORCE_INLINE   long    btVector3::maxDot( const btVector3 *array, long array_count, btScalar &dotOut ) const
+SIMD_FORCE_INLINE long btVector3::maxDot(const btVector3* array, long array_count, btScalar& dotOut) const
 {
-#if (defined BT_USE_SSE && defined BT_USE_SIMD_VECTOR3 && defined BT_USE_SSE_IN_API) || defined (BT_USE_NEON)
-    #if defined _WIN32 || defined (BT_USE_SSE)
-        const long scalar_cutoff = 10;
-        long _maxdot_large( const float *array, const float *vec, unsigned long array_count, float *dotOut );
-    #elif defined BT_USE_NEON
-        const long scalar_cutoff = 4;
-        extern long (*_maxdot_large)( const float *array, const float *vec, unsigned long array_count, float *dotOut );
-    #endif
-    if( array_count < scalar_cutoff )	
+#if (defined BT_USE_SSE && defined BT_USE_SIMD_VECTOR3 && defined BT_USE_SSE_IN_API) || defined(BT_USE_NEON)
+#if defined _WIN32 || defined(BT_USE_SSE)
+	const long scalar_cutoff = 10;
+	long _maxdot_large(const float* array, const float* vec, unsigned long array_count, float* dotOut);
+#elif defined BT_USE_NEON
+	const long scalar_cutoff = 4;
+	extern long (*_maxdot_large)(const float* array, const float* vec, unsigned long array_count, float* dotOut);
 #endif
-    {
-        btScalar maxDot1 = -SIMD_INFINITY;
-        int i = 0;
-        int ptIndex = -1;
-        for( i = 0; i < array_count; i++ )
-        {
-            btScalar dot = array[i].dot(*this);
-            
-            if( dot > maxDot1 )
-            {
-                maxDot1 = dot;
-                ptIndex = i;
-            }
-        }
-        
-        dotOut = maxDot1;
-        return ptIndex;
-    }
-#if (defined BT_USE_SSE && defined BT_USE_SIMD_VECTOR3 && defined BT_USE_SSE_IN_API) || defined (BT_USE_NEON)
-    return _maxdot_large( (float*) array, (float*) &m_floats[0], array_count, &dotOut );
+	if (array_count < scalar_cutoff)
+#endif
+	{
+		btScalar maxDot1 = -SIMD_INFINITY;
+		int i = 0;
+		int ptIndex = -1;
+		for (i = 0; i < array_count; i++)
+		{
+			btScalar dot = array[i].dot(*this);
+
+			if (dot > maxDot1)
+			{
+				maxDot1 = dot;
+				ptIndex = i;
+			}
+		}
+
+		dotOut = maxDot1;
+		return ptIndex;
+	}
+#if (defined BT_USE_SSE && defined BT_USE_SIMD_VECTOR3 && defined BT_USE_SSE_IN_API) || defined(BT_USE_NEON)
+	return _maxdot_large((float*)array, (float*)&m_floats[0], array_count, &dotOut);
 #endif
 }
 
-SIMD_FORCE_INLINE   long    btVector3::minDot( const btVector3 *array, long array_count, btScalar &dotOut ) const
+SIMD_FORCE_INLINE long btVector3::minDot(const btVector3* array, long array_count, btScalar& dotOut) const
 {
-#if (defined BT_USE_SSE && defined BT_USE_SIMD_VECTOR3 && defined BT_USE_SSE_IN_API) || defined (BT_USE_NEON)
-    #if defined BT_USE_SSE
-        const long scalar_cutoff = 10;
-        long _mindot_large( const float *array, const float *vec, unsigned long array_count, float *dotOut );
-    #elif defined BT_USE_NEON
-        const long scalar_cutoff = 4;
-        extern long (*_mindot_large)( const float *array, const float *vec, unsigned long array_count, float *dotOut );
-    #else
-        #error unhandled arch!
-    #endif
-    
-    if( array_count < scalar_cutoff )
+#if (defined BT_USE_SSE && defined BT_USE_SIMD_VECTOR3 && defined BT_USE_SSE_IN_API) || defined(BT_USE_NEON)
+#if defined BT_USE_SSE
+	const long scalar_cutoff = 10;
+	long _mindot_large(const float* array, const float* vec, unsigned long array_count, float* dotOut);
+#elif defined BT_USE_NEON
+	const long scalar_cutoff = 4;
+	extern long (*_mindot_large)(const float* array, const float* vec, unsigned long array_count, float* dotOut);
+#else
+#error unhandled arch!
 #endif
-    {
-        btScalar  minDot = SIMD_INFINITY;
-        int i = 0;
-        int ptIndex = -1;
-        
-        for( i = 0; i < array_count; i++ )
-        {
-            btScalar dot = array[i].dot(*this);
-            
-            if( dot < minDot )
-            {
-                minDot = dot;
-                ptIndex = i;
-            }
-        }
-        
-        dotOut = minDot;
-        
-        return ptIndex;
-    }
-#if (defined BT_USE_SSE && defined BT_USE_SIMD_VECTOR3 && defined BT_USE_SSE_IN_API) || defined (BT_USE_NEON)
-    return _mindot_large( (float*) array, (float*) &m_floats[0], array_count, &dotOut );
-#endif//BT_USE_SIMD_VECTOR3
-}
 
+	if (array_count < scalar_cutoff)
+#endif
+	{
+		btScalar minDot = SIMD_INFINITY;
+		int i = 0;
+		int ptIndex = -1;
+
+		for (i = 0; i < array_count; i++)
+		{
+			btScalar dot = array[i].dot(*this);
+
+			if (dot < minDot)
+			{
+				minDot = dot;
+				ptIndex = i;
+			}
+		}
+
+		dotOut = minDot;
+
+		return ptIndex;
+	}
+#if (defined BT_USE_SSE && defined BT_USE_SIMD_VECTOR3 && defined BT_USE_SSE_IN_API) || defined(BT_USE_NEON)
+	return _mindot_large((float*)array, (float*)&m_floats[0], array_count, &dotOut);
+#endif  //BT_USE_SIMD_VECTOR3
+}
 
 class btVector4 : public btVector3
 {
 public:
-
 	SIMD_FORCE_INLINE btVector4() {}
 
-
-	SIMD_FORCE_INLINE btVector4(const btScalar& _x, const btScalar& _y, const btScalar& _z,const btScalar& _w) 
-		: btVector3(_x,_y,_z)
+	SIMD_FORCE_INLINE btVector4(const btScalar& _x, const btScalar& _y, const btScalar& _z, const btScalar& _w)
+		: btVector3(_x, _y, _z)
 	{
 		m_floats[3] = _w;
 	}
 
-#if (defined (BT_USE_SSE_IN_API)&& defined (BT_USE_SSE)) || defined (BT_USE_NEON) 
+#if (defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)) || defined(BT_USE_NEON)
 	SIMD_FORCE_INLINE btVector4(const btSimdFloat4 vec)
 	{
 		mVec128 = vec;
@@ -1112,34 +1092,32 @@ public:
 		mVec128 = rhs.mVec128;
 	}
 
-	SIMD_FORCE_INLINE btVector4& 
-	operator=(const btVector4& v) 
+	SIMD_FORCE_INLINE btVector4&
+	operator=(const btVector4& v)
 	{
 		mVec128 = v.mVec128;
 		return *this;
 	}
-#endif // #if defined (BT_USE_SSE_IN_API) || defined (BT_USE_NEON) 
+#endif  // #if defined (BT_USE_SSE_IN_API) || defined (BT_USE_NEON)
 
-	SIMD_FORCE_INLINE btVector4 absolute4() const 
+	SIMD_FORCE_INLINE btVector4 absolute4() const
 	{
-#if defined BT_USE_SIMD_VECTOR3 && defined(BT_USE_SSE_IN_API) && defined (BT_USE_SSE) 
+#if defined BT_USE_SIMD_VECTOR3 && defined(BT_USE_SSE_IN_API) && defined(BT_USE_SSE)
 		return btVector4(_mm_and_ps(mVec128, btvAbsfMask));
 #elif defined(BT_USE_NEON)
 		return btVector4(vabsq_f32(mVec128));
-#else	
+#else
 		return btVector4(
-			btFabs(m_floats[0]), 
-			btFabs(m_floats[1]), 
+			btFabs(m_floats[0]),
+			btFabs(m_floats[1]),
 			btFabs(m_floats[2]),
 			btFabs(m_floats[3]));
 #endif
 	}
 
+	btScalar getW() const { return m_floats[3]; }
 
-	btScalar	getW() const { return m_floats[3];}
-
-
-		SIMD_FORCE_INLINE int maxAxis4() const
+	SIMD_FORCE_INLINE int maxAxis4() const
 	{
 		int maxIndex = -1;
 		btScalar maxVal = btScalar(-BT_LARGE_FLOAT);
@@ -1156,7 +1134,7 @@ public:
 		if (m_floats[2] > maxVal)
 		{
 			maxIndex = 2;
-			maxVal =m_floats[2];
+			maxVal = m_floats[2];
 		}
 		if (m_floats[3] > maxVal)
 		{
@@ -1166,7 +1144,6 @@ public:
 		return maxIndex;
 	}
 
-
 	SIMD_FORCE_INLINE int minAxis4() const
 	{
 		int minIndex = -1;
@@ -1184,190 +1161,176 @@ public:
 		if (m_floats[2] < minVal)
 		{
 			minIndex = 2;
-			minVal =m_floats[2];
+			minVal = m_floats[2];
 		}
 		if (m_floats[3] < minVal)
 		{
 			minIndex = 3;
 		}
-		
+
 		return minIndex;
 	}
 
-
-	SIMD_FORCE_INLINE int closestAxis4() const 
+	SIMD_FORCE_INLINE int closestAxis4() const
 	{
 		return absolute4().maxAxis4();
 	}
 
-	
- 
-
-  /**@brief Set x,y,z and zero w 
+	/**@brief Set x,y,z and zero w 
    * @param x Value of x
    * @param y Value of y
    * @param z Value of z
    */
-		
 
-/*		void getValue(btScalar *m) const 
+	/*		void getValue(btScalar *m) const 
 		{
 			m[0] = m_floats[0];
 			m[1] = m_floats[1];
 			m[2] =m_floats[2];
 		}
 */
-/**@brief Set the values 
+	/**@brief Set the values 
    * @param x Value of x
    * @param y Value of y
    * @param z Value of z
    * @param w Value of w
    */
-		SIMD_FORCE_INLINE void	setValue(const btScalar& _x, const btScalar& _y, const btScalar& _z,const btScalar& _w)
-		{
-			m_floats[0]=_x;
-			m_floats[1]=_y;
-			m_floats[2]=_z;
-			m_floats[3]=_w;
-		}
-
-
+	SIMD_FORCE_INLINE void setValue(const btScalar& _x, const btScalar& _y, const btScalar& _z, const btScalar& _w)
+	{
+		m_floats[0] = _x;
+		m_floats[1] = _y;
+		m_floats[2] = _z;
+		m_floats[3] = _w;
+	}
 };
 
-
 ///btSwapVector3Endian swaps vector endianness, useful for network and cross-platform serialization
-SIMD_FORCE_INLINE void	btSwapScalarEndian(const btScalar& sourceVal, btScalar& destVal)
+SIMD_FORCE_INLINE void btSwapScalarEndian(const btScalar& sourceVal, btScalar& destVal)
 {
 #ifdef BT_USE_DOUBLE_PRECISION
-	unsigned char* dest = (unsigned char*) &destVal;
-	const unsigned char* src  = (const unsigned char*) &sourceVal;
+	unsigned char* dest = (unsigned char*)&destVal;
+	const unsigned char* src = (const unsigned char*)&sourceVal;
 	dest[0] = src[7];
-    dest[1] = src[6];
-    dest[2] = src[5];
-    dest[3] = src[4];
-    dest[4] = src[3];
-    dest[5] = src[2];
-    dest[6] = src[1];
-    dest[7] = src[0];
+	dest[1] = src[6];
+	dest[2] = src[5];
+	dest[3] = src[4];
+	dest[4] = src[3];
+	dest[5] = src[2];
+	dest[6] = src[1];
+	dest[7] = src[0];
 #else
-	unsigned char* dest = (unsigned char*) &destVal;
-	const unsigned char* src  = (const unsigned char*) &sourceVal;
+	unsigned char* dest = (unsigned char*)&destVal;
+	const unsigned char* src = (const unsigned char*)&sourceVal;
 	dest[0] = src[3];
-    dest[1] = src[2];
-    dest[2] = src[1];
-    dest[3] = src[0];
-#endif //BT_USE_DOUBLE_PRECISION
+	dest[1] = src[2];
+	dest[2] = src[1];
+	dest[3] = src[0];
+#endif  //BT_USE_DOUBLE_PRECISION
 }
 ///btSwapVector3Endian swaps vector endianness, useful for network and cross-platform serialization
-SIMD_FORCE_INLINE void	btSwapVector3Endian(const btVector3& sourceVec, btVector3& destVec)
+SIMD_FORCE_INLINE void btSwapVector3Endian(const btVector3& sourceVec, btVector3& destVec)
 {
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 	{
-		btSwapScalarEndian(sourceVec[i],destVec[i]);
+		btSwapScalarEndian(sourceVec[i], destVec[i]);
 	}
-
 }
 
 ///btUnSwapVector3Endian swaps vector endianness, useful for network and cross-platform serialization
-SIMD_FORCE_INLINE void	btUnSwapVector3Endian(btVector3& vector)
+SIMD_FORCE_INLINE void btUnSwapVector3Endian(btVector3& vector)
 {
-
-	btVector3	swappedVec;
-	for (int i=0;i<4;i++)
+	btVector3 swappedVec;
+	for (int i = 0; i < 4; i++)
 	{
-		btSwapScalarEndian(vector[i],swappedVec[i]);
+		btSwapScalarEndian(vector[i], swappedVec[i]);
 	}
 	vector = swappedVec;
 }
 
 template <class T>
-SIMD_FORCE_INLINE void btPlaneSpace1 (const T& n, T& p, T& q)
+SIMD_FORCE_INLINE void btPlaneSpace1(const T& n, T& p, T& q)
 {
-  if (btFabs(n[2]) > SIMDSQRT12) {
-    // choose p in y-z plane
-    btScalar a = n[1]*n[1] + n[2]*n[2];
-    btScalar k = btRecipSqrt (a);
-    p[0] = 0;
-	p[1] = -n[2]*k;
-	p[2] = n[1]*k;
-    // set q = n x p
-    q[0] = a*k;
-	q[1] = -n[0]*p[2];
-	q[2] = n[0]*p[1];
-  }
-  else {
-    // choose p in x-y plane
-    btScalar a = n[0]*n[0] + n[1]*n[1];
-    btScalar k = btRecipSqrt (a);
-    p[0] = -n[1]*k;
-	p[1] = n[0]*k;
-	p[2] = 0;
-    // set q = n x p
-    q[0] = -n[2]*p[1];
-	q[1] = n[2]*p[0];
-	q[2] = a*k;
-  }
+	if (btFabs(n[2]) > SIMDSQRT12)
+	{
+		// choose p in y-z plane
+		btScalar a = n[1] * n[1] + n[2] * n[2];
+		btScalar k = btRecipSqrt(a);
+		p[0] = 0;
+		p[1] = -n[2] * k;
+		p[2] = n[1] * k;
+		// set q = n x p
+		q[0] = a * k;
+		q[1] = -n[0] * p[2];
+		q[2] = n[0] * p[1];
+	}
+	else
+	{
+		// choose p in x-y plane
+		btScalar a = n[0] * n[0] + n[1] * n[1];
+		btScalar k = btRecipSqrt(a);
+		p[0] = -n[1] * k;
+		p[1] = n[0] * k;
+		p[2] = 0;
+		// set q = n x p
+		q[0] = -n[2] * p[1];
+		q[1] = n[2] * p[0];
+		q[2] = a * k;
+	}
 }
 
-
-struct	btVector3FloatData
+struct btVector3FloatData
 {
-	float	m_floats[4];
+	float m_floats[4];
 };
 
-struct	btVector3DoubleData
+struct btVector3DoubleData
 {
-	double	m_floats[4];
-
+	double m_floats[4];
 };
 
-SIMD_FORCE_INLINE	void	btVector3::serializeFloat(struct	btVector3FloatData& dataOut) const
+SIMD_FORCE_INLINE void btVector3::serializeFloat(struct btVector3FloatData& dataOut) const
 {
 	///could also do a memcpy, check if it is worth it
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 		dataOut.m_floats[i] = float(m_floats[i]);
 }
 
-SIMD_FORCE_INLINE void	btVector3::deSerializeFloat(const struct	btVector3FloatData& dataIn)
+SIMD_FORCE_INLINE void btVector3::deSerializeFloat(const struct btVector3FloatData& dataIn)
 {
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 		m_floats[i] = btScalar(dataIn.m_floats[i]);
 }
 
-
-SIMD_FORCE_INLINE	void	btVector3::serializeDouble(struct	btVector3DoubleData& dataOut) const
+SIMD_FORCE_INLINE void btVector3::serializeDouble(struct btVector3DoubleData& dataOut) const
 {
 	///could also do a memcpy, check if it is worth it
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 		dataOut.m_floats[i] = double(m_floats[i]);
 }
 
-SIMD_FORCE_INLINE void	btVector3::deSerializeDouble(const struct	btVector3DoubleData& dataIn)
+SIMD_FORCE_INLINE void btVector3::deSerializeDouble(const struct btVector3DoubleData& dataIn)
 {
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 		m_floats[i] = btScalar(dataIn.m_floats[i]);
 }
 
-
-SIMD_FORCE_INLINE	void	btVector3::serialize(struct	btVector3Data& dataOut) const
+SIMD_FORCE_INLINE void btVector3::serialize(struct btVector3Data& dataOut) const
 {
 	///could also do a memcpy, check if it is worth it
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 		dataOut.m_floats[i] = m_floats[i];
 }
 
-
-SIMD_FORCE_INLINE void	btVector3::deSerialize(const struct	btVector3FloatData& dataIn)
+SIMD_FORCE_INLINE void btVector3::deSerialize(const struct btVector3FloatData& dataIn)
 {
-	for (int i = 0; i<4; i++)
+	for (int i = 0; i < 4; i++)
 		m_floats[i] = (btScalar)dataIn.m_floats[i];
 }
 
-
-SIMD_FORCE_INLINE void	btVector3::deSerialize(const struct	btVector3DoubleData& dataIn)
+SIMD_FORCE_INLINE void btVector3::deSerialize(const struct btVector3DoubleData& dataIn)
 {
-	for (int i=0;i<4;i++)
+	for (int i = 0; i < 4; i++)
 		m_floats[i] = (btScalar)dataIn.m_floats[i];
 }
 
-#endif //BT_VECTOR3_H
+#endif  //BT_VECTOR3_H
diff --git a/thirdparty/bullet/btBulletCollisionCommon.h b/thirdparty/bullet/btBulletCollisionCommon.h
index 948e02eb4cd..4f523756a74 100644
--- a/thirdparty/bullet/btBulletCollisionCommon.h
+++ b/thirdparty/bullet/btBulletCollisionCommon.h
@@ -62,6 +62,4 @@ subject to the following restrictions:
 #include "LinearMath/btIDebugDraw.h"
 #include "LinearMath/btSerializer.h"
 
-
-#endif //BULLET_COLLISION_COMMON_H
-
+#endif  //BULLET_COLLISION_COMMON_H
diff --git a/thirdparty/bullet/btBulletDynamicsCommon.h b/thirdparty/bullet/btBulletDynamicsCommon.h
index 50282bf2105..a421fa4461f 100644
--- a/thirdparty/bullet/btBulletDynamicsCommon.h
+++ b/thirdparty/bullet/btBulletDynamicsCommon.h
@@ -35,17 +35,9 @@ subject to the following restrictions:
 #include "BulletDynamics/ConstraintSolver/btGearConstraint.h"
 #include "BulletDynamics/ConstraintSolver/btFixedConstraint.h"
 
-
 #include "BulletDynamics/ConstraintSolver/btSequentialImpulseConstraintSolver.h"
 
-
 ///Vehicle simulation, with wheel contact simulated by raycasts
 #include "BulletDynamics/Vehicle/btRaycastVehicle.h"
 
-
-
-
-
-
-#endif //BULLET_DYNAMICS_COMMON_H
-
+#endif  //BULLET_DYNAMICS_COMMON_H
diff --git a/thirdparty/bullet/clew/clew.c b/thirdparty/bullet/clew/clew.c
index 5afc42a4855..90caced535a 100644
--- a/thirdparty/bullet/clew/clew.c
+++ b/thirdparty/bullet/clew/clew.c
@@ -9,23 +9,23 @@
 #include "clew.h"
 
 #ifdef _WIN32
-    #define WIN32_LEAN_AND_MEAN
-    #define VC_EXTRALEAN
-    #include <windows.h>
+#define WIN32_LEAN_AND_MEAN
+#define VC_EXTRALEAN
+#include <windows.h>
 
-    typedef HMODULE             CLEW_DYNLIB_HANDLE;
+typedef HMODULE CLEW_DYNLIB_HANDLE;
 
-    #define CLEW_DYNLIB_OPEN    LoadLibraryA
-    #define CLEW_DYNLIB_CLOSE   FreeLibrary
-    #define CLEW_DYNLIB_IMPORT  GetProcAddress
+#define CLEW_DYNLIB_OPEN LoadLibraryA
+#define CLEW_DYNLIB_CLOSE FreeLibrary
+#define CLEW_DYNLIB_IMPORT GetProcAddress
 #else
-    #include <dlfcn.h>
-    
-    typedef void*                   CLEW_DYNLIB_HANDLE;
+#include <dlfcn.h>
 
-    #define CLEW_DYNLIB_OPEN(path)  dlopen(path, RTLD_NOW | RTLD_GLOBAL)
-    #define CLEW_DYNLIB_CLOSE       dlclose
-    #define CLEW_DYNLIB_IMPORT      dlsym
+typedef void* CLEW_DYNLIB_HANDLE;
+
+#define CLEW_DYNLIB_OPEN(path) dlopen(path, RTLD_NOW | RTLD_GLOBAL)
+#define CLEW_DYNLIB_CLOSE dlclose
+#define CLEW_DYNLIB_IMPORT dlsym
 #endif
 
 #include <stdlib.h>
@@ -34,279 +34,341 @@
 static CLEW_DYNLIB_HANDLE module = NULL;
 
 //  Variables holding function entry points
-PFNCLGETPLATFORMIDS                 __clewGetPlatformIDs                = NULL;
-PFNCLGETPLATFORMINFO                __clewGetPlatformInfo               = NULL;
-PFNCLGETDEVICEIDS                   __clewGetDeviceIDs                  = NULL;
-PFNCLGETDEVICEINFO                  __clewGetDeviceInfo                 = NULL;
-PFNCLCREATECONTEXT                  __clewCreateContext                 = NULL;
-PFNCLCREATECONTEXTFROMTYPE          __clewCreateContextFromType         = NULL;
-PFNCLRETAINCONTEXT                  __clewRetainContext                 = NULL;
-PFNCLRELEASECONTEXT                 __clewReleaseContext                = NULL;
-PFNCLGETCONTEXTINFO                 __clewGetContextInfo                = NULL;
-PFNCLCREATECOMMANDQUEUE             __clewCreateCommandQueue            = NULL;
-PFNCLRETAINCOMMANDQUEUE             __clewRetainCommandQueue            = NULL;
-PFNCLRELEASECOMMANDQUEUE            __clewReleaseCommandQueue           = NULL;
-PFNCLGETCOMMANDQUEUEINFO            __clewGetCommandQueueInfo           = NULL;
+PFNCLGETPLATFORMIDS __clewGetPlatformIDs = NULL;
+PFNCLGETPLATFORMINFO __clewGetPlatformInfo = NULL;
+PFNCLGETDEVICEIDS __clewGetDeviceIDs = NULL;
+PFNCLGETDEVICEINFO __clewGetDeviceInfo = NULL;
+PFNCLCREATECONTEXT __clewCreateContext = NULL;
+PFNCLCREATECONTEXTFROMTYPE __clewCreateContextFromType = NULL;
+PFNCLRETAINCONTEXT __clewRetainContext = NULL;
+PFNCLRELEASECONTEXT __clewReleaseContext = NULL;
+PFNCLGETCONTEXTINFO __clewGetContextInfo = NULL;
+PFNCLCREATECOMMANDQUEUE __clewCreateCommandQueue = NULL;
+PFNCLRETAINCOMMANDQUEUE __clewRetainCommandQueue = NULL;
+PFNCLRELEASECOMMANDQUEUE __clewReleaseCommandQueue = NULL;
+PFNCLGETCOMMANDQUEUEINFO __clewGetCommandQueueInfo = NULL;
 #ifdef CL_USE_DEPRECATED_OPENCL_1_0_APIS
-PFNCLSETCOMMANDQUEUEPROPERTY        __clewSetCommandQueueProperty       = NULL;
+PFNCLSETCOMMANDQUEUEPROPERTY __clewSetCommandQueueProperty = NULL;
 #endif
-PFNCLCREATEBUFFER                   __clewCreateBuffer                  = NULL;
-PFNCLCREATESUBBUFFER                __clewCreateSubBuffer               = NULL;
-PFNCLCREATEIMAGE2D                  __clewCreateImage2D                 = NULL;
-PFNCLCREATEIMAGE3D                  __clewCreateImage3D                 = NULL;
-PFNCLRETAINMEMOBJECT                __clewRetainMemObject               = NULL;
-PFNCLRELEASEMEMOBJECT               __clewReleaseMemObject              = NULL;
-PFNCLGETSUPPORTEDIMAGEFORMATS       __clewGetSupportedImageFormats      = NULL;
-PFNCLGETMEMOBJECTINFO               __clewGetMemObjectInfo              = NULL;
-PFNCLGETIMAGEINFO                   __clewGetImageInfo                  = NULL;
+PFNCLCREATEBUFFER __clewCreateBuffer = NULL;
+PFNCLCREATESUBBUFFER __clewCreateSubBuffer = NULL;
+PFNCLCREATEIMAGE2D __clewCreateImage2D = NULL;
+PFNCLCREATEIMAGE3D __clewCreateImage3D = NULL;
+PFNCLRETAINMEMOBJECT __clewRetainMemObject = NULL;
+PFNCLRELEASEMEMOBJECT __clewReleaseMemObject = NULL;
+PFNCLGETSUPPORTEDIMAGEFORMATS __clewGetSupportedImageFormats = NULL;
+PFNCLGETMEMOBJECTINFO __clewGetMemObjectInfo = NULL;
+PFNCLGETIMAGEINFO __clewGetImageInfo = NULL;
 PFNCLSETMEMOBJECTDESTRUCTORCALLBACK __clewSetMemObjectDestructorCallback = NULL;
-PFNCLCREATESAMPLER                  __clewCreateSampler                 = NULL;
-PFNCLRETAINSAMPLER                  __clewRetainSampler                 = NULL;
-PFNCLRELEASESAMPLER                 __clewReleaseSampler                = NULL;
-PFNCLGETSAMPLERINFO                 __clewGetSamplerInfo                = NULL;
-PFNCLCREATEPROGRAMWITHSOURCE        __clewCreateProgramWithSource       = NULL;
-PFNCLCREATEPROGRAMWITHBINARY        __clewCreateProgramWithBinary       = NULL;
-PFNCLRETAINPROGRAM                  __clewRetainProgram                 = NULL;
-PFNCLRELEASEPROGRAM                 __clewReleaseProgram                = NULL;
-PFNCLBUILDPROGRAM                   __clewBuildProgram                  = NULL;
-PFNCLUNLOADCOMPILER                 __clewUnloadCompiler                = NULL;
-PFNCLGETPROGRAMINFO                 __clewGetProgramInfo                = NULL;
-PFNCLGETPROGRAMBUILDINFO            __clewGetProgramBuildInfo           = NULL;
-PFNCLCREATEKERNEL                   __clewCreateKernel                  = NULL;
-PFNCLCREATEKERNELSINPROGRAM         __clewCreateKernelsInProgram        = NULL;
-PFNCLRETAINKERNEL                   __clewRetainKernel                  = NULL;
-PFNCLRELEASEKERNEL                  __clewReleaseKernel                 = NULL;
-PFNCLSETKERNELARG                   __clewSetKernelArg                  = NULL;
-PFNCLGETKERNELINFO                  __clewGetKernelInfo                 = NULL;
-PFNCLGETKERNELWORKGROUPINFO         __clewGetKernelWorkGroupInfo        = NULL;
-PFNCLWAITFOREVENTS                  __clewWaitForEvents                 = NULL;
-PFNCLGETEVENTINFO                   __clewGetEventInfo                  = NULL;
-PFNCLCREATEUSEREVENT                __clewCreateUserEvent               = NULL;
-PFNCLRETAINEVENT                    __clewRetainEvent                   = NULL;
-PFNCLRELEASEEVENT                   __clewReleaseEvent                  = NULL;
-PFNCLSETUSEREVENTSTATUS             __clewSetUserEventStatus            = NULL;
-PFNCLSETEVENTCALLBACK               __clewSetEventCallback              = NULL;
-PFNCLGETEVENTPROFILINGINFO          __clewGetEventProfilingInfo         = NULL;
-PFNCLFLUSH                          __clewFlush                         = NULL;
-PFNCLFINISH                         __clewFinish                        = NULL;
-PFNCLENQUEUEREADBUFFER              __clewEnqueueReadBuffer             = NULL;
-PFNCLENQUEUEREADBUFFERRECT          __clewEnqueueReadBufferRect         = NULL;
-PFNCLENQUEUEWRITEBUFFER             __clewEnqueueWriteBuffer            = NULL;
-PFNCLENQUEUEWRITEBUFFERRECT         __clewEnqueueWriteBufferRect        = NULL;
-PFNCLENQUEUECOPYBUFFER              __clewEnqueueCopyBuffer             = NULL;
-PFNCLENQUEUEREADIMAGE               __clewEnqueueReadImage              = NULL;
-PFNCLENQUEUEWRITEIMAGE              __clewEnqueueWriteImage             = NULL;
-PFNCLENQUEUECOPYIMAGE               __clewEnqueueCopyImage              = NULL;
-PFNCLENQUEUECOPYBUFFERRECT          __clewEnqueueCopyBufferRect         = NULL;
-PFNCLENQUEUECOPYIMAGETOBUFFER       __clewEnqueueCopyImageToBuffer      = NULL;
-PFNCLENQUEUECOPYBUFFERTOIMAGE       __clewEnqueueCopyBufferToImage      = NULL;
-PFNCLENQUEUEMAPBUFFER               __clewEnqueueMapBuffer              = NULL;
-PFNCLENQUEUEMAPIMAGE                __clewEnqueueMapImage               = NULL;
-PFNCLENQUEUEUNMAPMEMOBJECT          __clewEnqueueUnmapMemObject         = NULL;
-PFNCLENQUEUENDRANGEKERNEL           __clewEnqueueNDRangeKernel          = NULL;
-PFNCLENQUEUETASK                    __clewEnqueueTask                   = NULL;
-PFNCLENQUEUENATIVEKERNEL            __clewEnqueueNativeKernel           = NULL;
-PFNCLENQUEUEMARKER                  __clewEnqueueMarker                 = NULL;
-PFNCLENQUEUEWAITFOREVENTS           __clewEnqueueWaitForEvents          = NULL;
-PFNCLENQUEUEBARRIER                 __clewEnqueueBarrier                = NULL;
-PFNCLGETEXTENSIONFUNCTIONADDRESS    __clewGetExtensionFunctionAddress   = NULL;
-
+PFNCLCREATESAMPLER __clewCreateSampler = NULL;
+PFNCLRETAINSAMPLER __clewRetainSampler = NULL;
+PFNCLRELEASESAMPLER __clewReleaseSampler = NULL;
+PFNCLGETSAMPLERINFO __clewGetSamplerInfo = NULL;
+PFNCLCREATEPROGRAMWITHSOURCE __clewCreateProgramWithSource = NULL;
+PFNCLCREATEPROGRAMWITHBINARY __clewCreateProgramWithBinary = NULL;
+PFNCLRETAINPROGRAM __clewRetainProgram = NULL;
+PFNCLRELEASEPROGRAM __clewReleaseProgram = NULL;
+PFNCLBUILDPROGRAM __clewBuildProgram = NULL;
+PFNCLUNLOADCOMPILER __clewUnloadCompiler = NULL;
+PFNCLGETPROGRAMINFO __clewGetProgramInfo = NULL;
+PFNCLGETPROGRAMBUILDINFO __clewGetProgramBuildInfo = NULL;
+PFNCLCREATEKERNEL __clewCreateKernel = NULL;
+PFNCLCREATEKERNELSINPROGRAM __clewCreateKernelsInProgram = NULL;
+PFNCLRETAINKERNEL __clewRetainKernel = NULL;
+PFNCLRELEASEKERNEL __clewReleaseKernel = NULL;
+PFNCLSETKERNELARG __clewSetKernelArg = NULL;
+PFNCLGETKERNELINFO __clewGetKernelInfo = NULL;
+PFNCLGETKERNELWORKGROUPINFO __clewGetKernelWorkGroupInfo = NULL;
+PFNCLWAITFOREVENTS __clewWaitForEvents = NULL;
+PFNCLGETEVENTINFO __clewGetEventInfo = NULL;
+PFNCLCREATEUSEREVENT __clewCreateUserEvent = NULL;
+PFNCLRETAINEVENT __clewRetainEvent = NULL;
+PFNCLRELEASEEVENT __clewReleaseEvent = NULL;
+PFNCLSETUSEREVENTSTATUS __clewSetUserEventStatus = NULL;
+PFNCLSETEVENTCALLBACK __clewSetEventCallback = NULL;
+PFNCLGETEVENTPROFILINGINFO __clewGetEventProfilingInfo = NULL;
+PFNCLFLUSH __clewFlush = NULL;
+PFNCLFINISH __clewFinish = NULL;
+PFNCLENQUEUEREADBUFFER __clewEnqueueReadBuffer = NULL;
+PFNCLENQUEUEREADBUFFERRECT __clewEnqueueReadBufferRect = NULL;
+PFNCLENQUEUEWRITEBUFFER __clewEnqueueWriteBuffer = NULL;
+PFNCLENQUEUEWRITEBUFFERRECT __clewEnqueueWriteBufferRect = NULL;
+PFNCLENQUEUECOPYBUFFER __clewEnqueueCopyBuffer = NULL;
+PFNCLENQUEUEREADIMAGE __clewEnqueueReadImage = NULL;
+PFNCLENQUEUEWRITEIMAGE __clewEnqueueWriteImage = NULL;
+PFNCLENQUEUECOPYIMAGE __clewEnqueueCopyImage = NULL;
+PFNCLENQUEUECOPYBUFFERRECT __clewEnqueueCopyBufferRect = NULL;
+PFNCLENQUEUECOPYIMAGETOBUFFER __clewEnqueueCopyImageToBuffer = NULL;
+PFNCLENQUEUECOPYBUFFERTOIMAGE __clewEnqueueCopyBufferToImage = NULL;
+PFNCLENQUEUEMAPBUFFER __clewEnqueueMapBuffer = NULL;
+PFNCLENQUEUEMAPIMAGE __clewEnqueueMapImage = NULL;
+PFNCLENQUEUEUNMAPMEMOBJECT __clewEnqueueUnmapMemObject = NULL;
+PFNCLENQUEUENDRANGEKERNEL __clewEnqueueNDRangeKernel = NULL;
+PFNCLENQUEUETASK __clewEnqueueTask = NULL;
+PFNCLENQUEUENATIVEKERNEL __clewEnqueueNativeKernel = NULL;
+PFNCLENQUEUEMARKER __clewEnqueueMarker = NULL;
+PFNCLENQUEUEWAITFOREVENTS __clewEnqueueWaitForEvents = NULL;
+PFNCLENQUEUEBARRIER __clewEnqueueBarrier = NULL;
+PFNCLGETEXTENSIONFUNCTIONADDRESS __clewGetExtensionFunctionAddress = NULL;
 
 void clewExit(void)
 {
-    if (module != NULL)
-    {
-        //  Ignore errors
-        CLEW_DYNLIB_CLOSE(module);
-        module = NULL;
-    }
+	if (module != NULL)
+	{
+		//  Ignore errors
+		CLEW_DYNLIB_CLOSE(module);
+		module = NULL;
+	}
 }
 
 int clewInit(const char* path)
 {
-    int error = 0;
+	int error = 0;
 
-    //  Check if already initialized
-    if (module != NULL)
-    {
-        return CLEW_SUCCESS;
-    }
+	//  Check if already initialized
+	if (module != NULL)
+	{
+		return CLEW_SUCCESS;
+	}
 
-    //  Load library
-    module = CLEW_DYNLIB_OPEN(path);
+	//  Load library
+	module = CLEW_DYNLIB_OPEN(path);
 
-    //  Check for errors
-    if (module == NULL)
-    {
-        return CLEW_ERROR_OPEN_FAILED;
-    }
+	//  Check for errors
+	if (module == NULL)
+	{
+		return CLEW_ERROR_OPEN_FAILED;
+	}
 
-    //  Set unloading
-    error = atexit(clewExit);
+	//  Set unloading
+	error = atexit(clewExit);
 
-    if (error)
-    {
-        //  Failure queuing atexit, shutdown with error
-        CLEW_DYNLIB_CLOSE(module);
-        module = NULL;
+	if (error)
+	{
+		//  Failure queuing atexit, shutdown with error
+		CLEW_DYNLIB_CLOSE(module);
+		module = NULL;
 
-        return CLEW_ERROR_ATEXIT_FAILED;
-    }
+		return CLEW_ERROR_ATEXIT_FAILED;
+	}
 
-    //  Determine function entry-points
-    __clewGetPlatformIDs                = (PFNCLGETPLATFORMIDS              )CLEW_DYNLIB_IMPORT(module, "clGetPlatformIDs");
-    __clewGetPlatformInfo               = (PFNCLGETPLATFORMINFO             )CLEW_DYNLIB_IMPORT(module, "clGetPlatformInfo");
-    __clewGetDeviceIDs                  = (PFNCLGETDEVICEIDS                )CLEW_DYNLIB_IMPORT(module, "clGetDeviceIDs");
-    __clewGetDeviceInfo                 = (PFNCLGETDEVICEINFO               )CLEW_DYNLIB_IMPORT(module, "clGetDeviceInfo");
-    __clewCreateContext                 = (PFNCLCREATECONTEXT               )CLEW_DYNLIB_IMPORT(module, "clCreateContext");
-    __clewCreateContextFromType         = (PFNCLCREATECONTEXTFROMTYPE       )CLEW_DYNLIB_IMPORT(module, "clCreateContextFromType");
-    __clewRetainContext                 = (PFNCLRETAINCONTEXT               )CLEW_DYNLIB_IMPORT(module, "clRetainContext");
-    __clewReleaseContext                = (PFNCLRELEASECONTEXT              )CLEW_DYNLIB_IMPORT(module, "clReleaseContext");
-    __clewGetContextInfo                = (PFNCLGETCONTEXTINFO              )CLEW_DYNLIB_IMPORT(module, "clGetContextInfo");
-    __clewCreateCommandQueue            = (PFNCLCREATECOMMANDQUEUE          )CLEW_DYNLIB_IMPORT(module, "clCreateCommandQueue");
-    __clewRetainCommandQueue            = (PFNCLRETAINCOMMANDQUEUE          )CLEW_DYNLIB_IMPORT(module, "clRetainCommandQueue");
-    __clewReleaseCommandQueue           = (PFNCLRELEASECOMMANDQUEUE         )CLEW_DYNLIB_IMPORT(module, "clReleaseCommandQueue");
-    __clewGetCommandQueueInfo           = (PFNCLGETCOMMANDQUEUEINFO         )CLEW_DYNLIB_IMPORT(module, "clGetCommandQueueInfo");
+	//  Determine function entry-points
+	__clewGetPlatformIDs = (PFNCLGETPLATFORMIDS)CLEW_DYNLIB_IMPORT(module, "clGetPlatformIDs");
+	__clewGetPlatformInfo = (PFNCLGETPLATFORMINFO)CLEW_DYNLIB_IMPORT(module, "clGetPlatformInfo");
+	__clewGetDeviceIDs = (PFNCLGETDEVICEIDS)CLEW_DYNLIB_IMPORT(module, "clGetDeviceIDs");
+	__clewGetDeviceInfo = (PFNCLGETDEVICEINFO)CLEW_DYNLIB_IMPORT(module, "clGetDeviceInfo");
+	__clewCreateContext = (PFNCLCREATECONTEXT)CLEW_DYNLIB_IMPORT(module, "clCreateContext");
+	__clewCreateContextFromType = (PFNCLCREATECONTEXTFROMTYPE)CLEW_DYNLIB_IMPORT(module, "clCreateContextFromType");
+	__clewRetainContext = (PFNCLRETAINCONTEXT)CLEW_DYNLIB_IMPORT(module, "clRetainContext");
+	__clewReleaseContext = (PFNCLRELEASECONTEXT)CLEW_DYNLIB_IMPORT(module, "clReleaseContext");
+	__clewGetContextInfo = (PFNCLGETCONTEXTINFO)CLEW_DYNLIB_IMPORT(module, "clGetContextInfo");
+	__clewCreateCommandQueue = (PFNCLCREATECOMMANDQUEUE)CLEW_DYNLIB_IMPORT(module, "clCreateCommandQueue");
+	__clewRetainCommandQueue = (PFNCLRETAINCOMMANDQUEUE)CLEW_DYNLIB_IMPORT(module, "clRetainCommandQueue");
+	__clewReleaseCommandQueue = (PFNCLRELEASECOMMANDQUEUE)CLEW_DYNLIB_IMPORT(module, "clReleaseCommandQueue");
+	__clewGetCommandQueueInfo = (PFNCLGETCOMMANDQUEUEINFO)CLEW_DYNLIB_IMPORT(module, "clGetCommandQueueInfo");
 #ifdef CL_USE_DEPRECATED_OPENCL_1_0_APIS
-    __clewSetCommandQueueProperty       = (PFNCLSETCOMMANDQUEUEPROPERTY     )CLEW_DYNLIB_IMPORT(module, "clSetCommandQueueProperty");
+	__clewSetCommandQueueProperty = (PFNCLSETCOMMANDQUEUEPROPERTY)CLEW_DYNLIB_IMPORT(module, "clSetCommandQueueProperty");
 #endif
-    __clewCreateBuffer                  = (PFNCLCREATEBUFFER                )CLEW_DYNLIB_IMPORT(module, "clCreateBuffer");
-    __clewCreateSubBuffer               = (PFNCLCREATESUBBUFFER                )CLEW_DYNLIB_IMPORT(module, "clCreateBuffer");
-    __clewCreateImage2D                 = (PFNCLCREATEIMAGE2D               )CLEW_DYNLIB_IMPORT(module, "clCreateImage2D");
-    __clewCreateImage3D                 = (PFNCLCREATEIMAGE3D               )CLEW_DYNLIB_IMPORT(module, "clCreateImage3D");
-    __clewRetainMemObject               = (PFNCLRETAINMEMOBJECT             )CLEW_DYNLIB_IMPORT(module, "clRetainMemObject");
-    __clewReleaseMemObject              = (PFNCLRELEASEMEMOBJECT            )CLEW_DYNLIB_IMPORT(module, "clReleaseMemObject");
-    __clewGetSupportedImageFormats      = (PFNCLGETSUPPORTEDIMAGEFORMATS    )CLEW_DYNLIB_IMPORT(module, "clGetSupportedImageFormats");
-    __clewGetMemObjectInfo              = (PFNCLGETMEMOBJECTINFO            )CLEW_DYNLIB_IMPORT(module, "clGetMemObjectInfo");
-    __clewGetImageInfo                  = (PFNCLGETIMAGEINFO                )CLEW_DYNLIB_IMPORT(module, "clGetImageInfo");
-    __clewSetMemObjectDestructorCallback = (PFNCLSETMEMOBJECTDESTRUCTORCALLBACK)CLEW_DYNLIB_IMPORT(module, "clSetMemObjectDestructorCallback");
-    __clewCreateSampler                 = (PFNCLCREATESAMPLER               )CLEW_DYNLIB_IMPORT(module, "clCreateSampler");
-    __clewRetainSampler                 = (PFNCLRETAINSAMPLER               )CLEW_DYNLIB_IMPORT(module, "clRetainSampler");
-    __clewReleaseSampler                = (PFNCLRELEASESAMPLER              )CLEW_DYNLIB_IMPORT(module, "clReleaseSampler");
-    __clewGetSamplerInfo                = (PFNCLGETSAMPLERINFO              )CLEW_DYNLIB_IMPORT(module, "clGetSamplerInfo");
-    __clewCreateProgramWithSource       = (PFNCLCREATEPROGRAMWITHSOURCE     )CLEW_DYNLIB_IMPORT(module, "clCreateProgramWithSource");
-    __clewCreateProgramWithBinary       = (PFNCLCREATEPROGRAMWITHBINARY     )CLEW_DYNLIB_IMPORT(module, "clCreateProgramWithBinary");
-    __clewRetainProgram                 = (PFNCLRETAINPROGRAM               )CLEW_DYNLIB_IMPORT(module, "clRetainProgram");
-    __clewReleaseProgram                = (PFNCLRELEASEPROGRAM              )CLEW_DYNLIB_IMPORT(module, "clReleaseProgram");
-    __clewBuildProgram                  = (PFNCLBUILDPROGRAM                )CLEW_DYNLIB_IMPORT(module, "clBuildProgram");
-    __clewUnloadCompiler                = (PFNCLUNLOADCOMPILER              )CLEW_DYNLIB_IMPORT(module, "clUnloadCompiler");
-    __clewGetProgramInfo                = (PFNCLGETPROGRAMINFO              )CLEW_DYNLIB_IMPORT(module, "clGetProgramInfo");
-    __clewGetProgramBuildInfo           = (PFNCLGETPROGRAMBUILDINFO         )CLEW_DYNLIB_IMPORT(module, "clGetProgramBuildInfo");
-    __clewCreateKernel                  = (PFNCLCREATEKERNEL                )CLEW_DYNLIB_IMPORT(module, "clCreateKernel");
-    __clewCreateKernelsInProgram        = (PFNCLCREATEKERNELSINPROGRAM      )CLEW_DYNLIB_IMPORT(module, "clCreateKernelsInProgram");
-    __clewRetainKernel                  = (PFNCLRETAINKERNEL                )CLEW_DYNLIB_IMPORT(module, "clRetainKernel");
-    __clewReleaseKernel                 = (PFNCLRELEASEKERNEL               )CLEW_DYNLIB_IMPORT(module, "clReleaseKernel");
-    __clewSetKernelArg                  = (PFNCLSETKERNELARG                )CLEW_DYNLIB_IMPORT(module, "clSetKernelArg");
-    __clewGetKernelInfo                 = (PFNCLGETKERNELINFO               )CLEW_DYNLIB_IMPORT(module, "clGetKernelInfo");
-    __clewGetKernelWorkGroupInfo        = (PFNCLGETKERNELWORKGROUPINFO      )CLEW_DYNLIB_IMPORT(module, "clGetKernelWorkGroupInfo");
-    __clewWaitForEvents                 = (PFNCLWAITFOREVENTS               )CLEW_DYNLIB_IMPORT(module, "clWaitForEvents");
-    __clewGetEventInfo                  = (PFNCLGETEVENTINFO                )CLEW_DYNLIB_IMPORT(module, "clGetEventInfo");
-    __clewCreateUserEvent               = (PFNCLCREATEUSEREVENT             )CLEW_DYNLIB_IMPORT(module, "clCreateUserEvent");
-    __clewRetainEvent                   = (PFNCLRETAINEVENT                 )CLEW_DYNLIB_IMPORT(module, "clRetainEvent");
-    __clewReleaseEvent                  = (PFNCLRELEASEEVENT                )CLEW_DYNLIB_IMPORT(module, "clReleaseEvent");
-    __clewSetUserEventStatus            = (PFNCLSETUSEREVENTSTATUS          )CLEW_DYNLIB_IMPORT(module, "clSetUserEventStatus");
-    __clewSetEventCallback              = (PFNCLSETEVENTCALLBACK            )CLEW_DYNLIB_IMPORT(module, "clSetEventCallback");
-    __clewGetEventProfilingInfo         = (PFNCLGETEVENTPROFILINGINFO       )CLEW_DYNLIB_IMPORT(module, "clGetEventProfilingInfo");
-    __clewFlush                         = (PFNCLFLUSH                       )CLEW_DYNLIB_IMPORT(module, "clFlush");
-    __clewFinish                        = (PFNCLFINISH                      )CLEW_DYNLIB_IMPORT(module, "clFinish");
-    __clewEnqueueReadBuffer             = (PFNCLENQUEUEREADBUFFER           )CLEW_DYNLIB_IMPORT(module, "clEnqueueReadBuffer");
-    __clewEnqueueReadBufferRect         = (PFNCLENQUEUEREADBUFFERRECT       )CLEW_DYNLIB_IMPORT(module, "clEnqueueReadBufferRect");
-    __clewEnqueueWriteBuffer            = (PFNCLENQUEUEWRITEBUFFER          )CLEW_DYNLIB_IMPORT(module, "clEnqueueWriteBuffer");
-    __clewEnqueueWriteBufferRect        = (PFNCLENQUEUEWRITEBUFFERRECT      )CLEW_DYNLIB_IMPORT(module, "clEnqueueWriteBufferRect");
-    __clewEnqueueCopyBuffer             = (PFNCLENQUEUECOPYBUFFER           )CLEW_DYNLIB_IMPORT(module, "clEnqueueCopyBuffer");
-    __clewEnqueueCopyBufferRect         = (PFNCLENQUEUECOPYBUFFERRECT       )CLEW_DYNLIB_IMPORT(module, "clEnqueueCopyBufferRect");
-    __clewEnqueueReadImage              = (PFNCLENQUEUEREADIMAGE            )CLEW_DYNLIB_IMPORT(module, "clEnqueueReadImage");
-    __clewEnqueueWriteImage             = (PFNCLENQUEUEWRITEIMAGE           )CLEW_DYNLIB_IMPORT(module, "clEnqueueWriteImage");
-    __clewEnqueueCopyImage              = (PFNCLENQUEUECOPYIMAGE            )CLEW_DYNLIB_IMPORT(module, "clEnqueueCopyImage");
-    __clewEnqueueCopyImageToBuffer      = (PFNCLENQUEUECOPYIMAGETOBUFFER    )CLEW_DYNLIB_IMPORT(module, "clEnqueueCopyImageToBuffer");
-    __clewEnqueueCopyBufferToImage      = (PFNCLENQUEUECOPYBUFFERTOIMAGE    )CLEW_DYNLIB_IMPORT(module, "clEnqueueCopyBufferToImage");
-    __clewEnqueueMapBuffer              = (PFNCLENQUEUEMAPBUFFER            )CLEW_DYNLIB_IMPORT(module, "clEnqueueMapBuffer");
-    __clewEnqueueMapImage               = (PFNCLENQUEUEMAPIMAGE             )CLEW_DYNLIB_IMPORT(module, "clEnqueueMapImage");
-    __clewEnqueueUnmapMemObject         = (PFNCLENQUEUEUNMAPMEMOBJECT       )CLEW_DYNLIB_IMPORT(module, "clEnqueueUnmapMemObject");
-    __clewEnqueueNDRangeKernel          = (PFNCLENQUEUENDRANGEKERNEL        )CLEW_DYNLIB_IMPORT(module, "clEnqueueNDRangeKernel");
-    __clewEnqueueTask                   = (PFNCLENQUEUETASK                 )CLEW_DYNLIB_IMPORT(module, "clEnqueueTask");
-    __clewEnqueueNativeKernel           = (PFNCLENQUEUENATIVEKERNEL         )CLEW_DYNLIB_IMPORT(module, "clEnqueueNativeKernel");
-    __clewEnqueueMarker                 = (PFNCLENQUEUEMARKER               )CLEW_DYNLIB_IMPORT(module, "clEnqueueMarker");
-    __clewEnqueueWaitForEvents          = (PFNCLENQUEUEWAITFOREVENTS        )CLEW_DYNLIB_IMPORT(module, "clEnqueueWaitForEvents");
-    __clewEnqueueBarrier                = (PFNCLENQUEUEBARRIER              )CLEW_DYNLIB_IMPORT(module, "clEnqueueBarrier");
-    __clewGetExtensionFunctionAddress   = (PFNCLGETEXTENSIONFUNCTIONADDRESS )CLEW_DYNLIB_IMPORT(module, "clGetExtensionFunctionAddress");
+	__clewCreateBuffer = (PFNCLCREATEBUFFER)CLEW_DYNLIB_IMPORT(module, "clCreateBuffer");
+	__clewCreateSubBuffer = (PFNCLCREATESUBBUFFER)CLEW_DYNLIB_IMPORT(module, "clCreateBuffer");
+	__clewCreateImage2D = (PFNCLCREATEIMAGE2D)CLEW_DYNLIB_IMPORT(module, "clCreateImage2D");
+	__clewCreateImage3D = (PFNCLCREATEIMAGE3D)CLEW_DYNLIB_IMPORT(module, "clCreateImage3D");
+	__clewRetainMemObject = (PFNCLRETAINMEMOBJECT)CLEW_DYNLIB_IMPORT(module, "clRetainMemObject");
+	__clewReleaseMemObject = (PFNCLRELEASEMEMOBJECT)CLEW_DYNLIB_IMPORT(module, "clReleaseMemObject");
+	__clewGetSupportedImageFormats = (PFNCLGETSUPPORTEDIMAGEFORMATS)CLEW_DYNLIB_IMPORT(module, "clGetSupportedImageFormats");
+	__clewGetMemObjectInfo = (PFNCLGETMEMOBJECTINFO)CLEW_DYNLIB_IMPORT(module, "clGetMemObjectInfo");
+	__clewGetImageInfo = (PFNCLGETIMAGEINFO)CLEW_DYNLIB_IMPORT(module, "clGetImageInfo");
+	__clewSetMemObjectDestructorCallback = (PFNCLSETMEMOBJECTDESTRUCTORCALLBACK)CLEW_DYNLIB_IMPORT(module, "clSetMemObjectDestructorCallback");
+	__clewCreateSampler = (PFNCLCREATESAMPLER)CLEW_DYNLIB_IMPORT(module, "clCreateSampler");
+	__clewRetainSampler = (PFNCLRETAINSAMPLER)CLEW_DYNLIB_IMPORT(module, "clRetainSampler");
+	__clewReleaseSampler = (PFNCLRELEASESAMPLER)CLEW_DYNLIB_IMPORT(module, "clReleaseSampler");
+	__clewGetSamplerInfo = (PFNCLGETSAMPLERINFO)CLEW_DYNLIB_IMPORT(module, "clGetSamplerInfo");
+	__clewCreateProgramWithSource = (PFNCLCREATEPROGRAMWITHSOURCE)CLEW_DYNLIB_IMPORT(module, "clCreateProgramWithSource");
+	__clewCreateProgramWithBinary = (PFNCLCREATEPROGRAMWITHBINARY)CLEW_DYNLIB_IMPORT(module, "clCreateProgramWithBinary");
+	__clewRetainProgram = (PFNCLRETAINPROGRAM)CLEW_DYNLIB_IMPORT(module, "clRetainProgram");
+	__clewReleaseProgram = (PFNCLRELEASEPROGRAM)CLEW_DYNLIB_IMPORT(module, "clReleaseProgram");
+	__clewBuildProgram = (PFNCLBUILDPROGRAM)CLEW_DYNLIB_IMPORT(module, "clBuildProgram");
+	__clewUnloadCompiler = (PFNCLUNLOADCOMPILER)CLEW_DYNLIB_IMPORT(module, "clUnloadCompiler");
+	__clewGetProgramInfo = (PFNCLGETPROGRAMINFO)CLEW_DYNLIB_IMPORT(module, "clGetProgramInfo");
+	__clewGetProgramBuildInfo = (PFNCLGETPROGRAMBUILDINFO)CLEW_DYNLIB_IMPORT(module, "clGetProgramBuildInfo");
+	__clewCreateKernel = (PFNCLCREATEKERNEL)CLEW_DYNLIB_IMPORT(module, "clCreateKernel");
+	__clewCreateKernelsInProgram = (PFNCLCREATEKERNELSINPROGRAM)CLEW_DYNLIB_IMPORT(module, "clCreateKernelsInProgram");
+	__clewRetainKernel = (PFNCLRETAINKERNEL)CLEW_DYNLIB_IMPORT(module, "clRetainKernel");
+	__clewReleaseKernel = (PFNCLRELEASEKERNEL)CLEW_DYNLIB_IMPORT(module, "clReleaseKernel");
+	__clewSetKernelArg = (PFNCLSETKERNELARG)CLEW_DYNLIB_IMPORT(module, "clSetKernelArg");
+	__clewGetKernelInfo = (PFNCLGETKERNELINFO)CLEW_DYNLIB_IMPORT(module, "clGetKernelInfo");
+	__clewGetKernelWorkGroupInfo = (PFNCLGETKERNELWORKGROUPINFO)CLEW_DYNLIB_IMPORT(module, "clGetKernelWorkGroupInfo");
+	__clewWaitForEvents = (PFNCLWAITFOREVENTS)CLEW_DYNLIB_IMPORT(module, "clWaitForEvents");
+	__clewGetEventInfo = (PFNCLGETEVENTINFO)CLEW_DYNLIB_IMPORT(module, "clGetEventInfo");
+	__clewCreateUserEvent = (PFNCLCREATEUSEREVENT)CLEW_DYNLIB_IMPORT(module, "clCreateUserEvent");
+	__clewRetainEvent = (PFNCLRETAINEVENT)CLEW_DYNLIB_IMPORT(module, "clRetainEvent");
+	__clewReleaseEvent = (PFNCLRELEASEEVENT)CLEW_DYNLIB_IMPORT(module, "clReleaseEvent");
+	__clewSetUserEventStatus = (PFNCLSETUSEREVENTSTATUS)CLEW_DYNLIB_IMPORT(module, "clSetUserEventStatus");
+	__clewSetEventCallback = (PFNCLSETEVENTCALLBACK)CLEW_DYNLIB_IMPORT(module, "clSetEventCallback");
+	__clewGetEventProfilingInfo = (PFNCLGETEVENTPROFILINGINFO)CLEW_DYNLIB_IMPORT(module, "clGetEventProfilingInfo");
+	__clewFlush = (PFNCLFLUSH)CLEW_DYNLIB_IMPORT(module, "clFlush");
+	__clewFinish = (PFNCLFINISH)CLEW_DYNLIB_IMPORT(module, "clFinish");
+	__clewEnqueueReadBuffer = (PFNCLENQUEUEREADBUFFER)CLEW_DYNLIB_IMPORT(module, "clEnqueueReadBuffer");
+	__clewEnqueueReadBufferRect = (PFNCLENQUEUEREADBUFFERRECT)CLEW_DYNLIB_IMPORT(module, "clEnqueueReadBufferRect");
+	__clewEnqueueWriteBuffer = (PFNCLENQUEUEWRITEBUFFER)CLEW_DYNLIB_IMPORT(module, "clEnqueueWriteBuffer");
+	__clewEnqueueWriteBufferRect = (PFNCLENQUEUEWRITEBUFFERRECT)CLEW_DYNLIB_IMPORT(module, "clEnqueueWriteBufferRect");
+	__clewEnqueueCopyBuffer = (PFNCLENQUEUECOPYBUFFER)CLEW_DYNLIB_IMPORT(module, "clEnqueueCopyBuffer");
+	__clewEnqueueCopyBufferRect = (PFNCLENQUEUECOPYBUFFERRECT)CLEW_DYNLIB_IMPORT(module, "clEnqueueCopyBufferRect");
+	__clewEnqueueReadImage = (PFNCLENQUEUEREADIMAGE)CLEW_DYNLIB_IMPORT(module, "clEnqueueReadImage");
+	__clewEnqueueWriteImage = (PFNCLENQUEUEWRITEIMAGE)CLEW_DYNLIB_IMPORT(module, "clEnqueueWriteImage");
+	__clewEnqueueCopyImage = (PFNCLENQUEUECOPYIMAGE)CLEW_DYNLIB_IMPORT(module, "clEnqueueCopyImage");
+	__clewEnqueueCopyImageToBuffer = (PFNCLENQUEUECOPYIMAGETOBUFFER)CLEW_DYNLIB_IMPORT(module, "clEnqueueCopyImageToBuffer");
+	__clewEnqueueCopyBufferToImage = (PFNCLENQUEUECOPYBUFFERTOIMAGE)CLEW_DYNLIB_IMPORT(module, "clEnqueueCopyBufferToImage");
+	__clewEnqueueMapBuffer = (PFNCLENQUEUEMAPBUFFER)CLEW_DYNLIB_IMPORT(module, "clEnqueueMapBuffer");
+	__clewEnqueueMapImage = (PFNCLENQUEUEMAPIMAGE)CLEW_DYNLIB_IMPORT(module, "clEnqueueMapImage");
+	__clewEnqueueUnmapMemObject = (PFNCLENQUEUEUNMAPMEMOBJECT)CLEW_DYNLIB_IMPORT(module, "clEnqueueUnmapMemObject");
+	__clewEnqueueNDRangeKernel = (PFNCLENQUEUENDRANGEKERNEL)CLEW_DYNLIB_IMPORT(module, "clEnqueueNDRangeKernel");
+	__clewEnqueueTask = (PFNCLENQUEUETASK)CLEW_DYNLIB_IMPORT(module, "clEnqueueTask");
+	__clewEnqueueNativeKernel = (PFNCLENQUEUENATIVEKERNEL)CLEW_DYNLIB_IMPORT(module, "clEnqueueNativeKernel");
+	__clewEnqueueMarker = (PFNCLENQUEUEMARKER)CLEW_DYNLIB_IMPORT(module, "clEnqueueMarker");
+	__clewEnqueueWaitForEvents = (PFNCLENQUEUEWAITFOREVENTS)CLEW_DYNLIB_IMPORT(module, "clEnqueueWaitForEvents");
+	__clewEnqueueBarrier = (PFNCLENQUEUEBARRIER)CLEW_DYNLIB_IMPORT(module, "clEnqueueBarrier");
+	__clewGetExtensionFunctionAddress = (PFNCLGETEXTENSIONFUNCTIONADDRESS)CLEW_DYNLIB_IMPORT(module, "clGetExtensionFunctionAddress");
 
-    return CLEW_SUCCESS;
+	return CLEW_SUCCESS;
 }
 
 const char* clewErrorString(cl_int error)
 {
-    static const char* strings[] =
-    {
-        // Error Codes
-          "CL_SUCCESS"                                  //   0
-        , "CL_DEVICE_NOT_FOUND"                         //  -1
-        , "CL_DEVICE_NOT_AVAILABLE"                     //  -2
-        , "CL_COMPILER_NOT_AVAILABLE"                   //  -3
-        , "CL_MEM_OBJECT_ALLOCATION_FAILURE"            //  -4
-        , "CL_OUT_OF_RESOURCES"                         //  -5
-        , "CL_OUT_OF_HOST_MEMORY"                       //  -6
-        , "CL_PROFILING_INFO_NOT_AVAILABLE"             //  -7
-        , "CL_MEM_COPY_OVERLAP"                         //  -8
-        , "CL_IMAGE_FORMAT_MISMATCH"                    //  -9
-        , "CL_IMAGE_FORMAT_NOT_SUPPORTED"               //  -10
-        , "CL_BUILD_PROGRAM_FAILURE"                    //  -11
-        , "CL_MAP_FAILURE"                              //  -12
+	static const char* strings[] =
+		{
+			// Error Codes
+			"CL_SUCCESS"  //   0
+			,
+			"CL_DEVICE_NOT_FOUND"  //  -1
+			,
+			"CL_DEVICE_NOT_AVAILABLE"  //  -2
+			,
+			"CL_COMPILER_NOT_AVAILABLE"  //  -3
+			,
+			"CL_MEM_OBJECT_ALLOCATION_FAILURE"  //  -4
+			,
+			"CL_OUT_OF_RESOURCES"  //  -5
+			,
+			"CL_OUT_OF_HOST_MEMORY"  //  -6
+			,
+			"CL_PROFILING_INFO_NOT_AVAILABLE"  //  -7
+			,
+			"CL_MEM_COPY_OVERLAP"  //  -8
+			,
+			"CL_IMAGE_FORMAT_MISMATCH"  //  -9
+			,
+			"CL_IMAGE_FORMAT_NOT_SUPPORTED"  //  -10
+			,
+			"CL_BUILD_PROGRAM_FAILURE"  //  -11
+			,
+			"CL_MAP_FAILURE"  //  -12
 
-        , ""    //  -13
-        , ""    //  -14
-        , ""    //  -15
-        , ""    //  -16
-        , ""    //  -17
-        , ""    //  -18
-        , ""    //  -19
+			,
+			""  //  -13
+			,
+			""  //  -14
+			,
+			""  //  -15
+			,
+			""  //  -16
+			,
+			""  //  -17
+			,
+			""  //  -18
+			,
+			""  //  -19
 
-        , ""    //  -20
-        , ""    //  -21
-        , ""    //  -22
-        , ""    //  -23
-        , ""    //  -24
-        , ""    //  -25
-        , ""    //  -26
-        , ""    //  -27
-        , ""    //  -28
-        , ""    //  -29
+			,
+			""  //  -20
+			,
+			""  //  -21
+			,
+			""  //  -22
+			,
+			""  //  -23
+			,
+			""  //  -24
+			,
+			""  //  -25
+			,
+			""  //  -26
+			,
+			""  //  -27
+			,
+			""  //  -28
+			,
+			""  //  -29
 
-        , "CL_INVALID_VALUE"                            //  -30
-        , "CL_INVALID_DEVICE_TYPE"                      //  -31
-        , "CL_INVALID_PLATFORM"                         //  -32
-        , "CL_INVALID_DEVICE"                           //  -33
-        , "CL_INVALID_CONTEXT"                          //  -34
-        , "CL_INVALID_QUEUE_PROPERTIES"                 //  -35
-        , "CL_INVALID_COMMAND_QUEUE"                    //  -36
-        , "CL_INVALID_HOST_PTR"                         //  -37
-        , "CL_INVALID_MEM_OBJECT"                       //  -38
-        , "CL_INVALID_IMAGE_FORMAT_DESCRIPTOR"          //  -39
-        , "CL_INVALID_IMAGE_SIZE"                       //  -40
-        , "CL_INVALID_SAMPLER"                          //  -41
-        , "CL_INVALID_BINARY"                           //  -42
-        , "CL_INVALID_BUILD_OPTIONS"                    //  -43
-        , "CL_INVALID_PROGRAM"                          //  -44
-        , "CL_INVALID_PROGRAM_EXECUTABLE"               //  -45
-        , "CL_INVALID_KERNEL_NAME"                      //  -46
-        , "CL_INVALID_KERNEL_DEFINITION"                //  -47
-        , "CL_INVALID_KERNEL"                           //  -48
-        , "CL_INVALID_ARG_INDEX"                        //  -49
-        , "CL_INVALID_ARG_VALUE"                        //  -50
-        , "CL_INVALID_ARG_SIZE"                         //  -51
-        , "CL_INVALID_KERNEL_ARGS"                      //  -52
-        , "CL_INVALID_WORK_DIMENSION"                   //  -53
-        , "CL_INVALID_WORK_GROUP_SIZE"                  //  -54
-        , "CL_INVALID_WORK_ITEM_SIZE"                   //  -55
-        , "CL_INVALID_GLOBAL_OFFSET"                    //  -56
-        , "CL_INVALID_EVENT_WAIT_LIST"                  //  -57
-        , "CL_INVALID_EVENT"                            //  -58
-        , "CL_INVALID_OPERATION"                        //  -59
-        , "CL_INVALID_GL_OBJECT"                        //  -60
-        , "CL_INVALID_BUFFER_SIZE"                      //  -61
-        , "CL_INVALID_MIP_LEVEL"                        //  -62
-        , "CL_INVALID_GLOBAL_WORK_SIZE"                 //  -63
-    };
+			,
+			"CL_INVALID_VALUE"  //  -30
+			,
+			"CL_INVALID_DEVICE_TYPE"  //  -31
+			,
+			"CL_INVALID_PLATFORM"  //  -32
+			,
+			"CL_INVALID_DEVICE"  //  -33
+			,
+			"CL_INVALID_CONTEXT"  //  -34
+			,
+			"CL_INVALID_QUEUE_PROPERTIES"  //  -35
+			,
+			"CL_INVALID_COMMAND_QUEUE"  //  -36
+			,
+			"CL_INVALID_HOST_PTR"  //  -37
+			,
+			"CL_INVALID_MEM_OBJECT"  //  -38
+			,
+			"CL_INVALID_IMAGE_FORMAT_DESCRIPTOR"  //  -39
+			,
+			"CL_INVALID_IMAGE_SIZE"  //  -40
+			,
+			"CL_INVALID_SAMPLER"  //  -41
+			,
+			"CL_INVALID_BINARY"  //  -42
+			,
+			"CL_INVALID_BUILD_OPTIONS"  //  -43
+			,
+			"CL_INVALID_PROGRAM"  //  -44
+			,
+			"CL_INVALID_PROGRAM_EXECUTABLE"  //  -45
+			,
+			"CL_INVALID_KERNEL_NAME"  //  -46
+			,
+			"CL_INVALID_KERNEL_DEFINITION"  //  -47
+			,
+			"CL_INVALID_KERNEL"  //  -48
+			,
+			"CL_INVALID_ARG_INDEX"  //  -49
+			,
+			"CL_INVALID_ARG_VALUE"  //  -50
+			,
+			"CL_INVALID_ARG_SIZE"  //  -51
+			,
+			"CL_INVALID_KERNEL_ARGS"  //  -52
+			,
+			"CL_INVALID_WORK_DIMENSION"  //  -53
+			,
+			"CL_INVALID_WORK_GROUP_SIZE"  //  -54
+			,
+			"CL_INVALID_WORK_ITEM_SIZE"  //  -55
+			,
+			"CL_INVALID_GLOBAL_OFFSET"  //  -56
+			,
+			"CL_INVALID_EVENT_WAIT_LIST"  //  -57
+			,
+			"CL_INVALID_EVENT"  //  -58
+			,
+			"CL_INVALID_OPERATION"  //  -59
+			,
+			"CL_INVALID_GL_OBJECT"  //  -60
+			,
+			"CL_INVALID_BUFFER_SIZE"  //  -61
+			,
+			"CL_INVALID_MIP_LEVEL"  //  -62
+			,
+			"CL_INVALID_GLOBAL_WORK_SIZE"  //  -63
+		};
 
-    return strings[-error];
+	return strings[-error];
 }
diff --git a/thirdparty/bullet/clew/clew.h b/thirdparty/bullet/clew/clew.h
index ee0fef18b41..cba85852339 100644
--- a/thirdparty/bullet/clew/clew.h
+++ b/thirdparty/bullet/clew/clew.h
@@ -11,7 +11,7 @@
 //! \file clew.h
 //! \brief OpenCL run-time loader header
 //!
-//! This file contains a copy of the contents of CL.H and CL_PLATFORM.H from the 
+//! This file contains a copy of the contents of CL.H and CL_PLATFORM.H from the
 //! official OpenCL spec. The purpose of this code is to load the OpenCL dynamic
 //! library at run-time and thus allow the executable to function on many
 //! platforms regardless of the vendor of the OpenCL driver actually installed.
@@ -19,11 +19,11 @@
 //! library (http://glew.sourceforge.net/)
 
 //  Run-time dynamic linking functionality based on concepts used in GLEW
-#ifdef  __OPENCL_CL_H
+#ifdef __OPENCL_CL_H
 #error cl.h included before clew.h
 #endif
 
-#ifdef  __OPENCL_CL_PLATFORM_H
+#ifdef __OPENCL_CL_PLATFORM_H
 #error cl_platform.h included before clew.h
 #endif
 
@@ -55,238 +55,239 @@
 * MATERIALS OR THE USE OR OTHER DEALINGS IN THE MATERIALS.
 ******************************************************************************/
 #ifdef __APPLE__
-    /* Contains #defines for AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER below */
-    #include <AvailabilityMacros.h>
+/* Contains #defines for AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER below */
+#include <AvailabilityMacros.h>
 #endif
 
 #ifdef __cplusplus
-extern "C" {
+extern "C"
+{
 #endif
 
 #if defined(_WIN32)
-    #define CL_API_ENTRY
-    #define CL_API_CALL     __stdcall
-    #define CL_CALLBACK     __stdcall
+#define CL_API_ENTRY
+#define CL_API_CALL __stdcall
+#define CL_CALLBACK __stdcall
 #else
-    #define CL_API_ENTRY
-    #define CL_API_CALL
-    #define CL_CALLBACK
+#define CL_API_ENTRY
+#define CL_API_CALL
+#define CL_CALLBACK
 #endif
-//disabled the APPLE thing, don't know why it is there, is just causes tons of warnings
+	//disabled the APPLE thing, don't know why it is there, is just causes tons of warnings
 
 #ifdef __APPLE1__
-    #define CL_EXTENSION_WEAK_LINK                  __attribute__((weak_import))       
-    #define CL_API_SUFFIX__VERSION_1_0              AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER
-    #define CL_EXT_SUFFIX__VERSION_1_0              CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER
-    #define CL_API_SUFFIX__VERSION_1_1              CL_EXTENSION_WEAK_LINK
-    #define CL_EXT_SUFFIX__VERSION_1_1              CL_EXTENSION_WEAK_LINK
-    #define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED   CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER
+#define CL_EXTENSION_WEAK_LINK __attribute__((weak_import))
+#define CL_API_SUFFIX__VERSION_1_0 AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER
+#define CL_EXT_SUFFIX__VERSION_1_0 CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER
+#define CL_API_SUFFIX__VERSION_1_1 CL_EXTENSION_WEAK_LINK
+#define CL_EXT_SUFFIX__VERSION_1_1 CL_EXTENSION_WEAK_LINK
+#define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED CL_EXTENSION_WEAK_LINK AVAILABLE_MAC_OS_X_VERSION_10_6_AND_LATER
 #else
-    #define CL_EXTENSION_WEAK_LINK                         
-    #define CL_API_SUFFIX__VERSION_1_0
-    #define CL_EXT_SUFFIX__VERSION_1_0
-    #define CL_API_SUFFIX__VERSION_1_1
-    #define CL_EXT_SUFFIX__VERSION_1_1
-    #define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED
+#define CL_EXTENSION_WEAK_LINK
+#define CL_API_SUFFIX__VERSION_1_0
+#define CL_EXT_SUFFIX__VERSION_1_0
+#define CL_API_SUFFIX__VERSION_1_1
+#define CL_EXT_SUFFIX__VERSION_1_1
+#define CL_EXT_SUFFIX__VERSION_1_0_DEPRECATED
 #endif
 
-#if (defined (_WIN32) && defined(_MSC_VER))
+#if (defined(_WIN32) && defined(_MSC_VER))
 
-/* scalar types  */
-typedef signed   __int8         cl_char;
-typedef unsigned __int8         cl_uchar;
-typedef signed   __int16        cl_short;
-typedef unsigned __int16        cl_ushort;
-typedef signed   __int32        cl_int;
-typedef unsigned __int32        cl_uint;
-typedef signed   __int64        cl_long;
-typedef unsigned __int64        cl_ulong;
+	/* scalar types  */
+	typedef signed __int8 cl_char;
+	typedef unsigned __int8 cl_uchar;
+	typedef signed __int16 cl_short;
+	typedef unsigned __int16 cl_ushort;
+	typedef signed __int32 cl_int;
+	typedef unsigned __int32 cl_uint;
+	typedef signed __int64 cl_long;
+	typedef unsigned __int64 cl_ulong;
 
-typedef unsigned __int16        cl_half;
-typedef float                   cl_float;
-typedef double                  cl_double;
+	typedef unsigned __int16 cl_half;
+	typedef float cl_float;
+	typedef double cl_double;
 
 /* Macro names and corresponding values defined by OpenCL */
-#define CL_CHAR_BIT         8
-#define CL_SCHAR_MAX        127
-#define CL_SCHAR_MIN        (-127-1)
-#define CL_CHAR_MAX         CL_SCHAR_MAX
-#define CL_CHAR_MIN         CL_SCHAR_MIN
-#define CL_UCHAR_MAX        255
-#define CL_SHRT_MAX         32767
-#define CL_SHRT_MIN         (-32767-1)
-#define CL_USHRT_MAX        65535
-#define CL_INT_MAX          2147483647
-#define CL_INT_MIN          (-2147483647-1)
-#define CL_UINT_MAX         0xffffffffU
-#define CL_LONG_MAX         ((cl_long) 0x7FFFFFFFFFFFFFFFLL)
-#define CL_LONG_MIN         ((cl_long) -0x7FFFFFFFFFFFFFFFLL - 1LL)
-#define CL_ULONG_MAX        ((cl_ulong) 0xFFFFFFFFFFFFFFFFULL)
+#define CL_CHAR_BIT 8
+#define CL_SCHAR_MAX 127
+#define CL_SCHAR_MIN (-127 - 1)
+#define CL_CHAR_MAX CL_SCHAR_MAX
+#define CL_CHAR_MIN CL_SCHAR_MIN
+#define CL_UCHAR_MAX 255
+#define CL_SHRT_MAX 32767
+#define CL_SHRT_MIN (-32767 - 1)
+#define CL_USHRT_MAX 65535
+#define CL_INT_MAX 2147483647
+#define CL_INT_MIN (-2147483647 - 1)
+#define CL_UINT_MAX 0xffffffffU
+#define CL_LONG_MAX ((cl_long)0x7FFFFFFFFFFFFFFFLL)
+#define CL_LONG_MIN ((cl_long)-0x7FFFFFFFFFFFFFFFLL - 1LL)
+#define CL_ULONG_MAX ((cl_ulong)0xFFFFFFFFFFFFFFFFULL)
 
-#define CL_FLT_DIG          6
-#define CL_FLT_MANT_DIG     24
-#define CL_FLT_MAX_10_EXP   +38
-#define CL_FLT_MAX_EXP      +128
-#define CL_FLT_MIN_10_EXP   -37
-#define CL_FLT_MIN_EXP      -125
-#define CL_FLT_RADIX        2
-#define CL_FLT_MAX          340282346638528859811704183484516925440.0f
-#define CL_FLT_MIN          1.175494350822287507969e-38f
-#define CL_FLT_EPSILON      0x1.0p-23f
+#define CL_FLT_DIG 6
+#define CL_FLT_MANT_DIG 24
+#define CL_FLT_MAX_10_EXP +38
+#define CL_FLT_MAX_EXP +128
+#define CL_FLT_MIN_10_EXP -37
+#define CL_FLT_MIN_EXP -125
+#define CL_FLT_RADIX 2
+#define CL_FLT_MAX 340282346638528859811704183484516925440.0f
+#define CL_FLT_MIN 1.175494350822287507969e-38f
+#define CL_FLT_EPSILON 0x1.0p-23f
 
-#define CL_DBL_DIG          15
-#define CL_DBL_MANT_DIG     53
-#define CL_DBL_MAX_10_EXP   +308
-#define CL_DBL_MAX_EXP      +1024
-#define CL_DBL_MIN_10_EXP   -307
-#define CL_DBL_MIN_EXP      -1021
-#define CL_DBL_RADIX        2
-#define CL_DBL_MAX          179769313486231570814527423731704356798070567525844996598917476803157260780028538760589558632766878171540458953514382464234321326889464182768467546703537516986049910576551282076245490090389328944075868508455133942304583236903222948165808559332123348274797826204144723168738177180919299881250404026184124858368.0
-#define CL_DBL_MIN          2.225073858507201383090e-308
-#define CL_DBL_EPSILON      2.220446049250313080847e-16
+#define CL_DBL_DIG 15
+#define CL_DBL_MANT_DIG 53
+#define CL_DBL_MAX_10_EXP +308
+#define CL_DBL_MAX_EXP +1024
+#define CL_DBL_MIN_10_EXP -307
+#define CL_DBL_MIN_EXP -1021
+#define CL_DBL_RADIX 2
+#define CL_DBL_MAX 179769313486231570814527423731704356798070567525844996598917476803157260780028538760589558632766878171540458953514382464234321326889464182768467546703537516986049910576551282076245490090389328944075868508455133942304583236903222948165808559332123348274797826204144723168738177180919299881250404026184124858368.0
+#define CL_DBL_MIN 2.225073858507201383090e-308
+#define CL_DBL_EPSILON 2.220446049250313080847e-16
 
-#define  CL_M_E             2.718281828459045090796
-#define  CL_M_LOG2E         1.442695040888963387005
-#define  CL_M_LOG10E        0.434294481903251816668
-#define  CL_M_LN2           0.693147180559945286227
-#define  CL_M_LN10          2.302585092994045901094
-#define  CL_M_PI            3.141592653589793115998
-#define  CL_M_PI_2          1.570796326794896557999
-#define  CL_M_PI_4          0.785398163397448278999
-#define  CL_M_1_PI          0.318309886183790691216
-#define  CL_M_2_PI          0.636619772367581382433
-#define  CL_M_2_SQRTPI      1.128379167095512558561
-#define  CL_M_SQRT2         1.414213562373095145475
-#define  CL_M_SQRT1_2       0.707106781186547572737
+#define CL_M_E 2.718281828459045090796
+#define CL_M_LOG2E 1.442695040888963387005
+#define CL_M_LOG10E 0.434294481903251816668
+#define CL_M_LN2 0.693147180559945286227
+#define CL_M_LN10 2.302585092994045901094
+#define CL_M_PI 3.141592653589793115998
+#define CL_M_PI_2 1.570796326794896557999
+#define CL_M_PI_4 0.785398163397448278999
+#define CL_M_1_PI 0.318309886183790691216
+#define CL_M_2_PI 0.636619772367581382433
+#define CL_M_2_SQRTPI 1.128379167095512558561
+#define CL_M_SQRT2 1.414213562373095145475
+#define CL_M_SQRT1_2 0.707106781186547572737
 
-#define  CL_M_E_F           2.71828174591064f
-#define  CL_M_LOG2E_F       1.44269502162933f
-#define  CL_M_LOG10E_F      0.43429449200630f
-#define  CL_M_LN2_F         0.69314718246460f
-#define  CL_M_LN10_F        2.30258512496948f
-#define  CL_M_PI_F          3.14159274101257f
-#define  CL_M_PI_2_F        1.57079637050629f
-#define  CL_M_PI_4_F        0.78539818525314f
-#define  CL_M_1_PI_F        0.31830987334251f
-#define  CL_M_2_PI_F        0.63661974668503f
-#define  CL_M_2_SQRTPI_F    1.12837922573090f
-#define  CL_M_SQRT2_F       1.41421353816986f
-#define  CL_M_SQRT1_2_F     0.70710676908493f
+#define CL_M_E_F 2.71828174591064f
+#define CL_M_LOG2E_F 1.44269502162933f
+#define CL_M_LOG10E_F 0.43429449200630f
+#define CL_M_LN2_F 0.69314718246460f
+#define CL_M_LN10_F 2.30258512496948f
+#define CL_M_PI_F 3.14159274101257f
+#define CL_M_PI_2_F 1.57079637050629f
+#define CL_M_PI_4_F 0.78539818525314f
+#define CL_M_1_PI_F 0.31830987334251f
+#define CL_M_2_PI_F 0.63661974668503f
+#define CL_M_2_SQRTPI_F 1.12837922573090f
+#define CL_M_SQRT2_F 1.41421353816986f
+#define CL_M_SQRT1_2_F 0.70710676908493f
 
-#define CL_NAN              (CL_INFINITY - CL_INFINITY)
-#define CL_HUGE_VALF        ((cl_float) 1e50)
-#define CL_HUGE_VAL         ((cl_double) 1e500)
-#define CL_MAXFLOAT         CL_FLT_MAX
-#define CL_INFINITY         CL_HUGE_VALF
+#define CL_NAN (CL_INFINITY - CL_INFINITY)
+#define CL_HUGE_VALF ((cl_float)1e50)
+#define CL_HUGE_VAL ((cl_double)1e500)
+#define CL_MAXFLOAT CL_FLT_MAX
+#define CL_INFINITY CL_HUGE_VALF
 
 #else
 
 #include <stdint.h>
 
 /* scalar types  */
-typedef int8_t          cl_char;
-typedef uint8_t         cl_uchar;
-typedef int16_t         cl_short    __attribute__((aligned(2)));
-typedef uint16_t        cl_ushort   __attribute__((aligned(2)));
-typedef int32_t         cl_int      __attribute__((aligned(4)));
-typedef uint32_t        cl_uint     __attribute__((aligned(4)));
-typedef int64_t         cl_long     __attribute__((aligned(8)));
-typedef uint64_t        cl_ulong    __attribute__((aligned(8)));
+typedef int8_t cl_char;
+typedef uint8_t cl_uchar;
+typedef int16_t cl_short __attribute__((aligned(2)));
+typedef uint16_t cl_ushort __attribute__((aligned(2)));
+typedef int32_t cl_int __attribute__((aligned(4)));
+typedef uint32_t cl_uint __attribute__((aligned(4)));
+typedef int64_t cl_long __attribute__((aligned(8)));
+typedef uint64_t cl_ulong __attribute__((aligned(8)));
 
-typedef uint16_t        cl_half     __attribute__((aligned(2)));
-typedef float           cl_float    __attribute__((aligned(4)));
-typedef double          cl_double   __attribute__((aligned(8)));
+typedef uint16_t cl_half __attribute__((aligned(2)));
+typedef float cl_float __attribute__((aligned(4)));
+typedef double cl_double __attribute__((aligned(8)));
 
 /* Macro names and corresponding values defined by OpenCL */
-#define CL_CHAR_BIT         8
-#define CL_SCHAR_MAX        127
-#define CL_SCHAR_MIN        (-127-1)
-#define CL_CHAR_MAX         CL_SCHAR_MAX
-#define CL_CHAR_MIN         CL_SCHAR_MIN
-#define CL_UCHAR_MAX        255
-#define CL_SHRT_MAX         32767
-#define CL_SHRT_MIN         (-32767-1)
-#define CL_USHRT_MAX        65535
-#define CL_INT_MAX          2147483647
-#define CL_INT_MIN          (-2147483647-1)
-#define CL_UINT_MAX         0xffffffffU
-#define CL_LONG_MAX         ((cl_long) 0x7FFFFFFFFFFFFFFFLL)
-#define CL_LONG_MIN         ((cl_long) -0x7FFFFFFFFFFFFFFFLL - 1LL)
-#define CL_ULONG_MAX        ((cl_ulong) 0xFFFFFFFFFFFFFFFFULL)
+#define CL_CHAR_BIT 8
+#define CL_SCHAR_MAX 127
+#define CL_SCHAR_MIN (-127 - 1)
+#define CL_CHAR_MAX CL_SCHAR_MAX
+#define CL_CHAR_MIN CL_SCHAR_MIN
+#define CL_UCHAR_MAX 255
+#define CL_SHRT_MAX 32767
+#define CL_SHRT_MIN (-32767 - 1)
+#define CL_USHRT_MAX 65535
+#define CL_INT_MAX 2147483647
+#define CL_INT_MIN (-2147483647 - 1)
+#define CL_UINT_MAX 0xffffffffU
+#define CL_LONG_MAX ((cl_long)0x7FFFFFFFFFFFFFFFLL)
+#define CL_LONG_MIN ((cl_long)-0x7FFFFFFFFFFFFFFFLL - 1LL)
+#define CL_ULONG_MAX ((cl_ulong)0xFFFFFFFFFFFFFFFFULL)
 
-#define CL_FLT_DIG          6
-#define CL_FLT_MANT_DIG     24
-#define CL_FLT_MAX_10_EXP   +38
-#define CL_FLT_MAX_EXP      +128
-#define CL_FLT_MIN_10_EXP   -37
-#define CL_FLT_MIN_EXP      -125
-#define CL_FLT_RADIX        2
-#define CL_FLT_MAX          0x1.fffffep127f
-#define CL_FLT_MIN          0x1.0p-126f
-#define CL_FLT_EPSILON      0x1.0p-23f
+#define CL_FLT_DIG 6
+#define CL_FLT_MANT_DIG 24
+#define CL_FLT_MAX_10_EXP +38
+#define CL_FLT_MAX_EXP +128
+#define CL_FLT_MIN_10_EXP -37
+#define CL_FLT_MIN_EXP -125
+#define CL_FLT_RADIX 2
+#define CL_FLT_MAX 0x1.fffffep127f
+#define CL_FLT_MIN 0x1.0p-126f
+#define CL_FLT_EPSILON 0x1.0p-23f
 
-#define CL_DBL_DIG          15
-#define CL_DBL_MANT_DIG     53
-#define CL_DBL_MAX_10_EXP   +308
-#define CL_DBL_MAX_EXP      +1024
-#define CL_DBL_MIN_10_EXP   -307
-#define CL_DBL_MIN_EXP      -1021
-#define CL_DBL_RADIX        2
-#define CL_DBL_MAX          0x1.fffffffffffffp1023
-#define CL_DBL_MIN          0x1.0p-1022
-#define CL_DBL_EPSILON      0x1.0p-52
+#define CL_DBL_DIG 15
+#define CL_DBL_MANT_DIG 53
+#define CL_DBL_MAX_10_EXP +308
+#define CL_DBL_MAX_EXP +1024
+#define CL_DBL_MIN_10_EXP -307
+#define CL_DBL_MIN_EXP -1021
+#define CL_DBL_RADIX 2
+#define CL_DBL_MAX 0x1.fffffffffffffp1023
+#define CL_DBL_MIN 0x1.0p-1022
+#define CL_DBL_EPSILON 0x1.0p-52
 
-#define  CL_M_E             2.718281828459045090796
-#define  CL_M_LOG2E         1.442695040888963387005
-#define  CL_M_LOG10E        0.434294481903251816668
-#define  CL_M_LN2           0.693147180559945286227
-#define  CL_M_LN10          2.302585092994045901094
-#define  CL_M_PI            3.141592653589793115998
-#define  CL_M_PI_2          1.570796326794896557999
-#define  CL_M_PI_4          0.785398163397448278999
-#define  CL_M_1_PI          0.318309886183790691216
-#define  CL_M_2_PI          0.636619772367581382433
-#define  CL_M_2_SQRTPI      1.128379167095512558561
-#define  CL_M_SQRT2         1.414213562373095145475
-#define  CL_M_SQRT1_2       0.707106781186547572737
+#define CL_M_E 2.718281828459045090796
+#define CL_M_LOG2E 1.442695040888963387005
+#define CL_M_LOG10E 0.434294481903251816668
+#define CL_M_LN2 0.693147180559945286227
+#define CL_M_LN10 2.302585092994045901094
+#define CL_M_PI 3.141592653589793115998
+#define CL_M_PI_2 1.570796326794896557999
+#define CL_M_PI_4 0.785398163397448278999
+#define CL_M_1_PI 0.318309886183790691216
+#define CL_M_2_PI 0.636619772367581382433
+#define CL_M_2_SQRTPI 1.128379167095512558561
+#define CL_M_SQRT2 1.414213562373095145475
+#define CL_M_SQRT1_2 0.707106781186547572737
 
-#define  CL_M_E_F           2.71828174591064f
-#define  CL_M_LOG2E_F       1.44269502162933f
-#define  CL_M_LOG10E_F      0.43429449200630f
-#define  CL_M_LN2_F         0.69314718246460f
-#define  CL_M_LN10_F        2.30258512496948f
-#define  CL_M_PI_F          3.14159274101257f
-#define  CL_M_PI_2_F        1.57079637050629f
-#define  CL_M_PI_4_F        0.78539818525314f
-#define  CL_M_1_PI_F        0.31830987334251f
-#define  CL_M_2_PI_F        0.63661974668503f
-#define  CL_M_2_SQRTPI_F    1.12837922573090f
-#define  CL_M_SQRT2_F       1.41421353816986f
-#define  CL_M_SQRT1_2_F     0.70710676908493f
+#define CL_M_E_F 2.71828174591064f
+#define CL_M_LOG2E_F 1.44269502162933f
+#define CL_M_LOG10E_F 0.43429449200630f
+#define CL_M_LN2_F 0.69314718246460f
+#define CL_M_LN10_F 2.30258512496948f
+#define CL_M_PI_F 3.14159274101257f
+#define CL_M_PI_2_F 1.57079637050629f
+#define CL_M_PI_4_F 0.78539818525314f
+#define CL_M_1_PI_F 0.31830987334251f
+#define CL_M_2_PI_F 0.63661974668503f
+#define CL_M_2_SQRTPI_F 1.12837922573090f
+#define CL_M_SQRT2_F 1.41421353816986f
+#define CL_M_SQRT1_2_F 0.70710676908493f
 
-#if defined( __GNUC__ )
-   #define CL_HUGE_VALF     __builtin_huge_valf()
-   #define CL_HUGE_VAL      __builtin_huge_val()
-   #define CL_NAN           __builtin_nanf( "" )
+#if defined(__GNUC__)
+#define CL_HUGE_VALF __builtin_huge_valf()
+#define CL_HUGE_VAL __builtin_huge_val()
+#define CL_NAN __builtin_nanf("")
 #else
-   #define CL_HUGE_VALF     ((cl_float) 1e50)
-   #define CL_HUGE_VAL      ((cl_double) 1e500)
-   float nanf( const char * );
-   #define CL_NAN           nanf( "" )  
+#define CL_HUGE_VALF ((cl_float)1e50)
+#define CL_HUGE_VAL ((cl_double)1e500)
+float nanf(const char *);
+#define CL_NAN nanf("")
 #endif
-#define CL_MAXFLOAT         CL_FLT_MAX
-#define CL_INFINITY         CL_HUGE_VALF
+#define CL_MAXFLOAT CL_FLT_MAX
+#define CL_INFINITY CL_HUGE_VALF
 
 #endif
 
 #include <stddef.h>
 
-/* Mirror types to GL types. Mirror types allow us to avoid deciding which headers to load based on whether we are using GL or GLES here. */
-typedef unsigned int cl_GLuint;
-typedef int          cl_GLint;
-typedef unsigned int cl_GLenum;
+	/* Mirror types to GL types. Mirror types allow us to avoid deciding which headers to load based on whether we are using GL or GLES here. */
+	typedef unsigned int cl_GLuint;
+	typedef int cl_GLint;
+	typedef unsigned int cl_GLenum;
 
-/*
+	/*
  * Vector types 
  *
  *  Note:   OpenCL requires that all types be naturally aligned. 
@@ -302,7 +303,6 @@ typedef unsigned int cl_GLenum;
  *          Maintaining proper alignment is the user's responsibility.
  */
 
-
 #ifdef _MSC_VER
 #if defined(_M_IX86)
 #if _M_IX86_FP >= 0
@@ -318,904 +318,1218 @@ typedef unsigned int cl_GLenum;
 #endif
 
 /* Define basic vector types */
-#if defined( __VEC__ )
-   #include <altivec.h>   /* may be omitted depending on compiler. AltiVec spec provides no way to detect whether the header is required. */
-   typedef vector unsigned char     __cl_uchar16;
-   typedef vector signed char       __cl_char16;
-   typedef vector unsigned short    __cl_ushort8;
-   typedef vector signed short      __cl_short8;
-   typedef vector unsigned int      __cl_uint4;
-   typedef vector signed int        __cl_int4;
-   typedef vector float             __cl_float4;
-   #define  __CL_UCHAR16__  1
-   #define  __CL_CHAR16__   1
-   #define  __CL_USHORT8__  1
-   #define  __CL_SHORT8__   1
-   #define  __CL_UINT4__    1
-   #define  __CL_INT4__     1
-   #define  __CL_FLOAT4__   1
+#if defined(__VEC__)
+#include <altivec.h> /* may be omitted depending on compiler. AltiVec spec provides no way to detect whether the header is required. */
+	typedef vector unsigned char __cl_uchar16;
+	typedef vector signed char __cl_char16;
+	typedef vector unsigned short __cl_ushort8;
+	typedef vector signed short __cl_short8;
+	typedef vector unsigned int __cl_uint4;
+	typedef vector signed int __cl_int4;
+	typedef vector float __cl_float4;
+#define __CL_UCHAR16__ 1
+#define __CL_CHAR16__ 1
+#define __CL_USHORT8__ 1
+#define __CL_SHORT8__ 1
+#define __CL_UINT4__ 1
+#define __CL_INT4__ 1
+#define __CL_FLOAT4__ 1
 #endif
 
-#if defined( __SSE__ )
-    #if defined( __MINGW64__ )
-        #include <intrin.h>
-    #else
-        #include <xmmintrin.h>
-    #endif
-    #if defined( __GNUC__ ) && !defined( __ICC )
-        typedef float __cl_float4   __attribute__((vector_size(16)));
-    #else
-        typedef __m128 __cl_float4;
-    #endif
-    #define __CL_FLOAT4__   1
+#if defined(__SSE__)
+#if defined(__MINGW64__)
+#include <intrin.h>
+#else
+#include <xmmintrin.h>
+#endif
+#if defined(__GNUC__) && !defined(__ICC)
+	typedef float __cl_float4 __attribute__((vector_size(16)));
+#else
+	typedef __m128 __cl_float4;
+#endif
+#define __CL_FLOAT4__ 1
 #endif
 
-#if defined( __SSE2__ )
-    #if defined( __MINGW64__ )
-        #include <intrin.h>
-    #else
-        #include <emmintrin.h>
-    #endif
-    #if defined( __GNUC__ ) && !defined( __ICC )
-        typedef cl_uchar    __cl_uchar16    __attribute__((vector_size(16)));
-        typedef cl_char     __cl_char16     __attribute__((vector_size(16)));
-        typedef cl_ushort   __cl_ushort8    __attribute__((vector_size(16)));
-        typedef cl_short    __cl_short8     __attribute__((vector_size(16)));
-        typedef cl_uint     __cl_uint4      __attribute__((vector_size(16)));
-        typedef cl_int      __cl_int4       __attribute__((vector_size(16)));
-        typedef cl_ulong    __cl_ulong2     __attribute__((vector_size(16)));
-        typedef cl_long     __cl_long2      __attribute__((vector_size(16)));
-        typedef cl_double   __cl_double2    __attribute__((vector_size(16)));
-    #else
-        typedef __m128i __cl_uchar16;
-        typedef __m128i __cl_char16;
-        typedef __m128i __cl_ushort8;
-        typedef __m128i __cl_short8;
-        typedef __m128i __cl_uint4;
-        typedef __m128i __cl_int4;
-        typedef __m128i __cl_ulong2;
-        typedef __m128i __cl_long2;
-        typedef __m128d __cl_double2;
-    #endif
-    #define __CL_UCHAR16__  1
-    #define __CL_CHAR16__   1
-    #define __CL_USHORT8__  1
-    #define __CL_SHORT8__   1
-    #define __CL_INT4__     1
-    #define __CL_UINT4__    1
-    #define __CL_ULONG2__   1
-    #define __CL_LONG2__    1
-    #define __CL_DOUBLE2__  1
+#if defined(__SSE2__)
+#if defined(__MINGW64__)
+#include <intrin.h>
+#else
+#include <emmintrin.h>
+#endif
+#if defined(__GNUC__) && !defined(__ICC)
+	typedef cl_uchar __cl_uchar16 __attribute__((vector_size(16)));
+	typedef cl_char __cl_char16 __attribute__((vector_size(16)));
+	typedef cl_ushort __cl_ushort8 __attribute__((vector_size(16)));
+	typedef cl_short __cl_short8 __attribute__((vector_size(16)));
+	typedef cl_uint __cl_uint4 __attribute__((vector_size(16)));
+	typedef cl_int __cl_int4 __attribute__((vector_size(16)));
+	typedef cl_ulong __cl_ulong2 __attribute__((vector_size(16)));
+	typedef cl_long __cl_long2 __attribute__((vector_size(16)));
+	typedef cl_double __cl_double2 __attribute__((vector_size(16)));
+#else
+	typedef __m128i __cl_uchar16;
+	typedef __m128i __cl_char16;
+	typedef __m128i __cl_ushort8;
+	typedef __m128i __cl_short8;
+	typedef __m128i __cl_uint4;
+	typedef __m128i __cl_int4;
+	typedef __m128i __cl_ulong2;
+	typedef __m128i __cl_long2;
+	typedef __m128d __cl_double2;
+#endif
+#define __CL_UCHAR16__ 1
+#define __CL_CHAR16__ 1
+#define __CL_USHORT8__ 1
+#define __CL_SHORT8__ 1
+#define __CL_INT4__ 1
+#define __CL_UINT4__ 1
+#define __CL_ULONG2__ 1
+#define __CL_LONG2__ 1
+#define __CL_DOUBLE2__ 1
 #endif
 
-#if defined( __MMX__ )
-    #include <mmintrin.h>
-    #if defined( __GNUC__ ) && !defined( __ICC )
-        typedef cl_uchar    __cl_uchar8     __attribute__((vector_size(8)));
-        typedef cl_char     __cl_char8      __attribute__((vector_size(8)));
-        typedef cl_ushort   __cl_ushort4    __attribute__((vector_size(8)));
-        typedef cl_short    __cl_short4     __attribute__((vector_size(8)));
-        typedef cl_uint     __cl_uint2      __attribute__((vector_size(8)));
-        typedef cl_int      __cl_int2       __attribute__((vector_size(8)));
-        typedef cl_ulong    __cl_ulong1     __attribute__((vector_size(8)));
-        typedef cl_long     __cl_long1      __attribute__((vector_size(8)));
-        typedef cl_float    __cl_float2     __attribute__((vector_size(8)));
-    #else
-        typedef __m64       __cl_uchar8;
-        typedef __m64       __cl_char8;
-        typedef __m64       __cl_ushort4;
-        typedef __m64       __cl_short4;
-        typedef __m64       __cl_uint2;
-        typedef __m64       __cl_int2;
-        typedef __m64       __cl_ulong1;
-        typedef __m64       __cl_long1;
-        typedef __m64       __cl_float2;
-    #endif
-    #define __CL_UCHAR8__   1
-    #define __CL_CHAR8__    1
-    #define __CL_USHORT4__  1
-    #define __CL_SHORT4__   1
-    #define __CL_INT2__     1
-    #define __CL_UINT2__    1
-    #define __CL_ULONG1__   1
-    #define __CL_LONG1__    1
-    #define __CL_FLOAT2__   1
+#if defined(__MMX__)
+#include <mmintrin.h>
+#if defined(__GNUC__) && !defined(__ICC)
+	typedef cl_uchar __cl_uchar8 __attribute__((vector_size(8)));
+	typedef cl_char __cl_char8 __attribute__((vector_size(8)));
+	typedef cl_ushort __cl_ushort4 __attribute__((vector_size(8)));
+	typedef cl_short __cl_short4 __attribute__((vector_size(8)));
+	typedef cl_uint __cl_uint2 __attribute__((vector_size(8)));
+	typedef cl_int __cl_int2 __attribute__((vector_size(8)));
+	typedef cl_ulong __cl_ulong1 __attribute__((vector_size(8)));
+	typedef cl_long __cl_long1 __attribute__((vector_size(8)));
+	typedef cl_float __cl_float2 __attribute__((vector_size(8)));
+#else
+	typedef __m64 __cl_uchar8;
+	typedef __m64 __cl_char8;
+	typedef __m64 __cl_ushort4;
+	typedef __m64 __cl_short4;
+	typedef __m64 __cl_uint2;
+	typedef __m64 __cl_int2;
+	typedef __m64 __cl_ulong1;
+	typedef __m64 __cl_long1;
+	typedef __m64 __cl_float2;
+#endif
+#define __CL_UCHAR8__ 1
+#define __CL_CHAR8__ 1
+#define __CL_USHORT4__ 1
+#define __CL_SHORT4__ 1
+#define __CL_INT2__ 1
+#define __CL_UINT2__ 1
+#define __CL_ULONG1__ 1
+#define __CL_LONG1__ 1
+#define __CL_FLOAT2__ 1
 #endif
 
-#if defined( __AVX__ )
-    #if defined( __MINGW64__ )
-        #include <intrin.h>
-    #else
-        #include <immintrin.h> 
-    #endif
-    #if defined( __GNUC__ ) && !defined( __ICC )
-        typedef cl_float    __cl_float8     __attribute__((vector_size(32)));
-        typedef cl_double   __cl_double4    __attribute__((vector_size(32)));
-    #else
-        typedef __m256      __cl_float8;
-        typedef __m256d     __cl_double4;
-    #endif
-    #define __CL_FLOAT8__   1
-    #define __CL_DOUBLE4__  1
+#if defined(__AVX__)
+#if defined(__MINGW64__)
+#include <intrin.h>
+#else
+#include <immintrin.h>
+#endif
+#if defined(__GNUC__) && !defined(__ICC)
+	typedef cl_float __cl_float8 __attribute__((vector_size(32)));
+	typedef cl_double __cl_double4 __attribute__((vector_size(32)));
+#else
+	typedef __m256 __cl_float8;
+	typedef __m256d __cl_double4;
+#endif
+#define __CL_FLOAT8__ 1
+#define __CL_DOUBLE4__ 1
 #endif
 
 /* Define alignment keys */
-#if defined( __GNUC__ )
-    #define CL_ALIGNED(_x)          __attribute__ ((aligned(_x)))
-#elif defined( _WIN32) && (_MSC_VER)
-    /* Alignment keys neutered on windows because MSVC can't swallow function arguments with alignment requirements     */
-    /* http://msdn.microsoft.com/en-us/library/373ak2y1%28VS.71%29.aspx                                                 */
-    /* #include <crtdefs.h>                                                                                             */
-    /* #define CL_ALIGNED(_x)          _CRT_ALIGN(_x)                                                                   */
-    #define CL_ALIGNED(_x)
+#if defined(__GNUC__)
+#define CL_ALIGNED(_x) __attribute__((aligned(_x)))
+#elif defined(_WIN32) && (_MSC_VER)
+/* Alignment keys neutered on windows because MSVC can't swallow function arguments with alignment requirements     */
+/* http://msdn.microsoft.com/en-us/library/373ak2y1%28VS.71%29.aspx                                                 */
+/* #include <crtdefs.h>                                                                                             */
+/* #define CL_ALIGNED(_x)          _CRT_ALIGN(_x)                                                                   */
+#define CL_ALIGNED(_x)
 #else
-   #warning  Need to implement some method to align data here
-   #define  CL_ALIGNED(_x)
+#warning Need to implement some method to align data here
+#define CL_ALIGNED(_x)
 #endif
 
 /* Indicate whether .xyzw, .s0123 and .hi.lo are supported */
-#if (defined( __GNUC__) && ! defined( __STRICT_ANSI__ )) || (defined( _MSC_VER ) && ! defined( __STDC__ ))
-    /* .xyzw and .s0123...{f|F} are supported */
-    #define CL_HAS_NAMED_VECTOR_FIELDS 1
-    /* .hi and .lo are supported */
-    #define CL_HAS_HI_LO_VECTOR_FIELDS 1
+#if (defined(__GNUC__) && !defined(__STRICT_ANSI__)) || (defined(_MSC_VER) && !defined(__STDC__))
+/* .xyzw and .s0123...{f|F} are supported */
+#define CL_HAS_NAMED_VECTOR_FIELDS 1
+/* .hi and .lo are supported */
+#define CL_HAS_HI_LO_VECTOR_FIELDS 1
 
-    #define CL_NAMED_STRUCT_SUPPORTED
+#define CL_NAMED_STRUCT_SUPPORTED
 #endif
 
-#if defined( CL_NAMED_STRUCT_SUPPORTED) && defined( _MSC_VER )
-#define __extension__ __pragma(warning(suppress:4201))
+#if defined(CL_NAMED_STRUCT_SUPPORTED) && defined(_MSC_VER)
+#define __extension__ __pragma(warning(suppress : 4201))
 #endif
 
-/* Define cl_vector types */
+	/* Define cl_vector types */
 
-/* ---- cl_charn ---- */
-typedef union
-{
-    cl_char  CL_ALIGNED(2) s[2];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_char  x, y; };
-   __extension__ struct{ cl_char  s0, s1; };
-   __extension__ struct{ cl_char  lo, hi; };
+	/* ---- cl_charn ---- */
+	typedef union {
+		cl_char CL_ALIGNED(2) s[2];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_char x, y;
+		};
+		__extension__ struct
+		{
+			cl_char s0, s1;
+		};
+		__extension__ struct
+		{
+			cl_char lo, hi;
+		};
 #endif
-#if defined( __CL_CHAR2__) 
-    __cl_char2     v2;
+#if defined(__CL_CHAR2__)
+		__cl_char2 v2;
 #endif
-}cl_char2;
+	} cl_char2;
 
-typedef union
-{
-    cl_char  CL_ALIGNED(4) s[4];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_char  x, y, z, w; };
-   __extension__ struct{ cl_char  s0, s1, s2, s3; };
-   __extension__ struct{ cl_char2 lo, hi; };
+	typedef union {
+		cl_char CL_ALIGNED(4) s[4];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_char x, y, z, w;
+		};
+		__extension__ struct
+		{
+			cl_char s0, s1, s2, s3;
+		};
+		__extension__ struct
+		{
+			cl_char2 lo, hi;
+		};
 #endif
-#if defined( __CL_CHAR2__) 
-    __cl_char2     v2[2];
+#if defined(__CL_CHAR2__)
+		__cl_char2 v2[2];
 #endif
-#if defined( __CL_CHAR4__) 
-    __cl_char4     v4;
+#if defined(__CL_CHAR4__)
+		__cl_char4 v4;
 #endif
-}cl_char4;
+	} cl_char4;
 
-/* cl_char3 is identical in size, alignment and behavior to cl_char4. See section 6.1.5. */
-typedef  cl_char4  cl_char3;
+	/* cl_char3 is identical in size, alignment and behavior to cl_char4. See section 6.1.5. */
+	typedef cl_char4 cl_char3;
 
-typedef union
-{
-    cl_char   CL_ALIGNED(8) s[8];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_char  x, y, z, w; };
-   __extension__ struct{ cl_char  s0, s1, s2, s3, s4, s5, s6, s7; };
-   __extension__ struct{ cl_char4 lo, hi; };
+	typedef union {
+		cl_char CL_ALIGNED(8) s[8];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_char x, y, z, w;
+		};
+		__extension__ struct
+		{
+			cl_char s0, s1, s2, s3, s4, s5, s6, s7;
+		};
+		__extension__ struct
+		{
+			cl_char4 lo, hi;
+		};
 #endif
-#if defined( __CL_CHAR2__) 
-    __cl_char2     v2[4];
+#if defined(__CL_CHAR2__)
+		__cl_char2 v2[4];
 #endif
-#if defined( __CL_CHAR4__) 
-    __cl_char4     v4[2];
+#if defined(__CL_CHAR4__)
+		__cl_char4 v4[2];
 #endif
-#if defined( __CL_CHAR8__ )
-    __cl_char8     v8;
+#if defined(__CL_CHAR8__)
+		__cl_char8 v8;
 #endif
-}cl_char8;
+	} cl_char8;
 
-typedef union
-{
-    cl_char  CL_ALIGNED(16) s[16];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_char  x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; };
-   __extension__ struct{ cl_char  s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
-   __extension__ struct{ cl_char8 lo, hi; };
+	typedef union {
+		cl_char CL_ALIGNED(16) s[16];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_char x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf;
+		};
+		__extension__ struct
+		{
+			cl_char s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF;
+		};
+		__extension__ struct
+		{
+			cl_char8 lo, hi;
+		};
 #endif
-#if defined( __CL_CHAR2__) 
-    __cl_char2     v2[8];
+#if defined(__CL_CHAR2__)
+		__cl_char2 v2[8];
 #endif
-#if defined( __CL_CHAR4__) 
-    __cl_char4     v4[4];
+#if defined(__CL_CHAR4__)
+		__cl_char4 v4[4];
 #endif
-#if defined( __CL_CHAR8__ )
-    __cl_char8     v8[2];
+#if defined(__CL_CHAR8__)
+		__cl_char8 v8[2];
 #endif
-#if defined( __CL_CHAR16__ )
-    __cl_char16    v16;
+#if defined(__CL_CHAR16__)
+		__cl_char16 v16;
 #endif
-}cl_char16;
+	} cl_char16;
 
+	/* ---- cl_ucharn ---- */
+	typedef union {
+		cl_uchar CL_ALIGNED(2) s[2];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_uchar x, y;
+		};
+		__extension__ struct
+		{
+			cl_uchar s0, s1;
+		};
+		__extension__ struct
+		{
+			cl_uchar lo, hi;
+		};
+#endif
+#if defined(__cl_uchar2__)
+		__cl_uchar2 v2;
+#endif
+	} cl_uchar2;
 
-/* ---- cl_ucharn ---- */
-typedef union
-{
-    cl_uchar  CL_ALIGNED(2) s[2];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_uchar  x, y; };
-   __extension__ struct{ cl_uchar  s0, s1; };
-   __extension__ struct{ cl_uchar  lo, hi; };
+	typedef union {
+		cl_uchar CL_ALIGNED(4) s[4];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_uchar x, y, z, w;
+		};
+		__extension__ struct
+		{
+			cl_uchar s0, s1, s2, s3;
+		};
+		__extension__ struct
+		{
+			cl_uchar2 lo, hi;
+		};
 #endif
-#if defined( __cl_uchar2__) 
-    __cl_uchar2     v2;
+#if defined(__CL_UCHAR2__)
+		__cl_uchar2 v2[2];
 #endif
-}cl_uchar2;
+#if defined(__CL_UCHAR4__)
+		__cl_uchar4 v4;
+#endif
+	} cl_uchar4;
 
-typedef union
-{
-    cl_uchar  CL_ALIGNED(4) s[4];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_uchar  x, y, z, w; };
-   __extension__ struct{ cl_uchar  s0, s1, s2, s3; };
-   __extension__ struct{ cl_uchar2 lo, hi; };
-#endif
-#if defined( __CL_UCHAR2__) 
-    __cl_uchar2     v2[2];
-#endif
-#if defined( __CL_UCHAR4__) 
-    __cl_uchar4     v4;
-#endif
-}cl_uchar4;
+	/* cl_uchar3 is identical in size, alignment and behavior to cl_uchar4. See section 6.1.5. */
+	typedef cl_uchar4 cl_uchar3;
 
-/* cl_uchar3 is identical in size, alignment and behavior to cl_uchar4. See section 6.1.5. */
-typedef  cl_uchar4  cl_uchar3;
+	typedef union {
+		cl_uchar CL_ALIGNED(8) s[8];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_uchar x, y, z, w;
+		};
+		__extension__ struct
+		{
+			cl_uchar s0, s1, s2, s3, s4, s5, s6, s7;
+		};
+		__extension__ struct
+		{
+			cl_uchar4 lo, hi;
+		};
+#endif
+#if defined(__CL_UCHAR2__)
+		__cl_uchar2 v2[4];
+#endif
+#if defined(__CL_UCHAR4__)
+		__cl_uchar4 v4[2];
+#endif
+#if defined(__CL_UCHAR8__)
+		__cl_uchar8 v8;
+#endif
+	} cl_uchar8;
 
-typedef union
-{
-    cl_uchar   CL_ALIGNED(8) s[8];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_uchar  x, y, z, w; };
-   __extension__ struct{ cl_uchar  s0, s1, s2, s3, s4, s5, s6, s7; };
-   __extension__ struct{ cl_uchar4 lo, hi; };
+	typedef union {
+		cl_uchar CL_ALIGNED(16) s[16];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_uchar x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf;
+		};
+		__extension__ struct
+		{
+			cl_uchar s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF;
+		};
+		__extension__ struct
+		{
+			cl_uchar8 lo, hi;
+		};
 #endif
-#if defined( __CL_UCHAR2__) 
-    __cl_uchar2     v2[4];
+#if defined(__CL_UCHAR2__)
+		__cl_uchar2 v2[8];
 #endif
-#if defined( __CL_UCHAR4__) 
-    __cl_uchar4     v4[2];
+#if defined(__CL_UCHAR4__)
+		__cl_uchar4 v4[4];
 #endif
-#if defined( __CL_UCHAR8__ )
-    __cl_uchar8     v8;
+#if defined(__CL_UCHAR8__)
+		__cl_uchar8 v8[2];
 #endif
-}cl_uchar8;
+#if defined(__CL_UCHAR16__)
+		__cl_uchar16 v16;
+#endif
+	} cl_uchar16;
 
-typedef union
-{
-    cl_uchar  CL_ALIGNED(16) s[16];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_uchar  x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; };
-   __extension__ struct{ cl_uchar  s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
-   __extension__ struct{ cl_uchar8 lo, hi; };
+	/* ---- cl_shortn ---- */
+	typedef union {
+		cl_short CL_ALIGNED(4) s[2];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_short x, y;
+		};
+		__extension__ struct
+		{
+			cl_short s0, s1;
+		};
+		__extension__ struct
+		{
+			cl_short lo, hi;
+		};
 #endif
-#if defined( __CL_UCHAR2__) 
-    __cl_uchar2     v2[8];
+#if defined(__CL_SHORT2__)
+		__cl_short2 v2;
 #endif
-#if defined( __CL_UCHAR4__) 
-    __cl_uchar4     v4[4];
-#endif
-#if defined( __CL_UCHAR8__ )
-    __cl_uchar8     v8[2];
-#endif
-#if defined( __CL_UCHAR16__ )
-    __cl_uchar16    v16;
-#endif
-}cl_uchar16;
+	} cl_short2;
 
+	typedef union {
+		cl_short CL_ALIGNED(8) s[4];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_short x, y, z, w;
+		};
+		__extension__ struct
+		{
+			cl_short s0, s1, s2, s3;
+		};
+		__extension__ struct
+		{
+			cl_short2 lo, hi;
+		};
+#endif
+#if defined(__CL_SHORT2__)
+		__cl_short2 v2[2];
+#endif
+#if defined(__CL_SHORT4__)
+		__cl_short4 v4;
+#endif
+	} cl_short4;
 
-/* ---- cl_shortn ---- */
-typedef union
-{
-    cl_short  CL_ALIGNED(4) s[2];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_short  x, y; };
-   __extension__ struct{ cl_short  s0, s1; };
-   __extension__ struct{ cl_short  lo, hi; };
-#endif
-#if defined( __CL_SHORT2__) 
-    __cl_short2     v2;
-#endif
-}cl_short2;
+	/* cl_short3 is identical in size, alignment and behavior to cl_short4. See section 6.1.5. */
+	typedef cl_short4 cl_short3;
 
-typedef union
-{
-    cl_short  CL_ALIGNED(8) s[4];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_short  x, y, z, w; };
-   __extension__ struct{ cl_short  s0, s1, s2, s3; };
-   __extension__ struct{ cl_short2 lo, hi; };
+	typedef union {
+		cl_short CL_ALIGNED(16) s[8];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_short x, y, z, w;
+		};
+		__extension__ struct
+		{
+			cl_short s0, s1, s2, s3, s4, s5, s6, s7;
+		};
+		__extension__ struct
+		{
+			cl_short4 lo, hi;
+		};
 #endif
-#if defined( __CL_SHORT2__) 
-    __cl_short2     v2[2];
+#if defined(__CL_SHORT2__)
+		__cl_short2 v2[4];
 #endif
-#if defined( __CL_SHORT4__) 
-    __cl_short4     v4;
+#if defined(__CL_SHORT4__)
+		__cl_short4 v4[2];
 #endif
-}cl_short4;
+#if defined(__CL_SHORT8__)
+		__cl_short8 v8;
+#endif
+	} cl_short8;
 
-/* cl_short3 is identical in size, alignment and behavior to cl_short4. See section 6.1.5. */
-typedef  cl_short4  cl_short3;
+	typedef union {
+		cl_short CL_ALIGNED(32) s[16];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_short x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf;
+		};
+		__extension__ struct
+		{
+			cl_short s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF;
+		};
+		__extension__ struct
+		{
+			cl_short8 lo, hi;
+		};
+#endif
+#if defined(__CL_SHORT2__)
+		__cl_short2 v2[8];
+#endif
+#if defined(__CL_SHORT4__)
+		__cl_short4 v4[4];
+#endif
+#if defined(__CL_SHORT8__)
+		__cl_short8 v8[2];
+#endif
+#if defined(__CL_SHORT16__)
+		__cl_short16 v16;
+#endif
+	} cl_short16;
 
-typedef union
-{
-    cl_short   CL_ALIGNED(16) s[8];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_short  x, y, z, w; };
-   __extension__ struct{ cl_short  s0, s1, s2, s3, s4, s5, s6, s7; };
-   __extension__ struct{ cl_short4 lo, hi; };
+	/* ---- cl_ushortn ---- */
+	typedef union {
+		cl_ushort CL_ALIGNED(4) s[2];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_ushort x, y;
+		};
+		__extension__ struct
+		{
+			cl_ushort s0, s1;
+		};
+		__extension__ struct
+		{
+			cl_ushort lo, hi;
+		};
 #endif
-#if defined( __CL_SHORT2__) 
-    __cl_short2     v2[4];
+#if defined(__CL_USHORT2__)
+		__cl_ushort2 v2;
 #endif
-#if defined( __CL_SHORT4__) 
-    __cl_short4     v4[2];
-#endif
-#if defined( __CL_SHORT8__ )
-    __cl_short8     v8;
-#endif
-}cl_short8;
+	} cl_ushort2;
 
-typedef union
-{
-    cl_short  CL_ALIGNED(32) s[16];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_short  x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; };
-   __extension__ struct{ cl_short  s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
-   __extension__ struct{ cl_short8 lo, hi; };
+	typedef union {
+		cl_ushort CL_ALIGNED(8) s[4];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_ushort x, y, z, w;
+		};
+		__extension__ struct
+		{
+			cl_ushort s0, s1, s2, s3;
+		};
+		__extension__ struct
+		{
+			cl_ushort2 lo, hi;
+		};
 #endif
-#if defined( __CL_SHORT2__) 
-    __cl_short2     v2[8];
+#if defined(__CL_USHORT2__)
+		__cl_ushort2 v2[2];
 #endif
-#if defined( __CL_SHORT4__) 
-    __cl_short4     v4[4];
+#if defined(__CL_USHORT4__)
+		__cl_ushort4 v4;
 #endif
-#if defined( __CL_SHORT8__ )
-    __cl_short8     v8[2];
-#endif
-#if defined( __CL_SHORT16__ )
-    __cl_short16    v16;
-#endif
-}cl_short16;
+	} cl_ushort4;
 
+	/* cl_ushort3 is identical in size, alignment and behavior to cl_ushort4. See section 6.1.5. */
+	typedef cl_ushort4 cl_ushort3;
 
-/* ---- cl_ushortn ---- */
-typedef union
-{
-    cl_ushort  CL_ALIGNED(4) s[2];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_ushort  x, y; };
-   __extension__ struct{ cl_ushort  s0, s1; };
-   __extension__ struct{ cl_ushort  lo, hi; };
+	typedef union {
+		cl_ushort CL_ALIGNED(16) s[8];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_ushort x, y, z, w;
+		};
+		__extension__ struct
+		{
+			cl_ushort s0, s1, s2, s3, s4, s5, s6, s7;
+		};
+		__extension__ struct
+		{
+			cl_ushort4 lo, hi;
+		};
 #endif
-#if defined( __CL_USHORT2__) 
-    __cl_ushort2     v2;
+#if defined(__CL_USHORT2__)
+		__cl_ushort2 v2[4];
 #endif
-}cl_ushort2;
+#if defined(__CL_USHORT4__)
+		__cl_ushort4 v4[2];
+#endif
+#if defined(__CL_USHORT8__)
+		__cl_ushort8 v8;
+#endif
+	} cl_ushort8;
 
-typedef union
-{
-    cl_ushort  CL_ALIGNED(8) s[4];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_ushort  x, y, z, w; };
-   __extension__ struct{ cl_ushort  s0, s1, s2, s3; };
-   __extension__ struct{ cl_ushort2 lo, hi; };
+	typedef union {
+		cl_ushort CL_ALIGNED(32) s[16];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_ushort x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf;
+		};
+		__extension__ struct
+		{
+			cl_ushort s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF;
+		};
+		__extension__ struct
+		{
+			cl_ushort8 lo, hi;
+		};
 #endif
-#if defined( __CL_USHORT2__) 
-    __cl_ushort2     v2[2];
+#if defined(__CL_USHORT2__)
+		__cl_ushort2 v2[8];
 #endif
-#if defined( __CL_USHORT4__) 
-    __cl_ushort4     v4;
+#if defined(__CL_USHORT4__)
+		__cl_ushort4 v4[4];
 #endif
-}cl_ushort4;
+#if defined(__CL_USHORT8__)
+		__cl_ushort8 v8[2];
+#endif
+#if defined(__CL_USHORT16__)
+		__cl_ushort16 v16;
+#endif
+	} cl_ushort16;
 
-/* cl_ushort3 is identical in size, alignment and behavior to cl_ushort4. See section 6.1.5. */
-typedef  cl_ushort4  cl_ushort3;
+	/* ---- cl_intn ---- */
+	typedef union {
+		cl_int CL_ALIGNED(8) s[2];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_int x, y;
+		};
+		__extension__ struct
+		{
+			cl_int s0, s1;
+		};
+		__extension__ struct
+		{
+			cl_int lo, hi;
+		};
+#endif
+#if defined(__CL_INT2__)
+		__cl_int2 v2;
+#endif
+	} cl_int2;
 
-typedef union
-{
-    cl_ushort   CL_ALIGNED(16) s[8];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_ushort  x, y, z, w; };
-   __extension__ struct{ cl_ushort  s0, s1, s2, s3, s4, s5, s6, s7; };
-   __extension__ struct{ cl_ushort4 lo, hi; };
+	typedef union {
+		cl_int CL_ALIGNED(16) s[4];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_int x, y, z, w;
+		};
+		__extension__ struct
+		{
+			cl_int s0, s1, s2, s3;
+		};
+		__extension__ struct
+		{
+			cl_int2 lo, hi;
+		};
 #endif
-#if defined( __CL_USHORT2__) 
-    __cl_ushort2     v2[4];
+#if defined(__CL_INT2__)
+		__cl_int2 v2[2];
 #endif
-#if defined( __CL_USHORT4__) 
-    __cl_ushort4     v4[2];
+#if defined(__CL_INT4__)
+		__cl_int4 v4;
 #endif
-#if defined( __CL_USHORT8__ )
-    __cl_ushort8     v8;
-#endif
-}cl_ushort8;
+	} cl_int4;
 
-typedef union
-{
-    cl_ushort  CL_ALIGNED(32) s[16];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_ushort  x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; };
-   __extension__ struct{ cl_ushort  s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
-   __extension__ struct{ cl_ushort8 lo, hi; };
-#endif
-#if defined( __CL_USHORT2__) 
-    __cl_ushort2     v2[8];
-#endif
-#if defined( __CL_USHORT4__) 
-    __cl_ushort4     v4[4];
-#endif
-#if defined( __CL_USHORT8__ )
-    __cl_ushort8     v8[2];
-#endif
-#if defined( __CL_USHORT16__ )
-    __cl_ushort16    v16;
-#endif
-}cl_ushort16;
+	/* cl_int3 is identical in size, alignment and behavior to cl_int4. See section 6.1.5. */
+	typedef cl_int4 cl_int3;
 
-/* ---- cl_intn ---- */
-typedef union
-{
-    cl_int  CL_ALIGNED(8) s[2];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_int  x, y; };
-   __extension__ struct{ cl_int  s0, s1; };
-   __extension__ struct{ cl_int  lo, hi; };
+	typedef union {
+		cl_int CL_ALIGNED(32) s[8];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_int x, y, z, w;
+		};
+		__extension__ struct
+		{
+			cl_int s0, s1, s2, s3, s4, s5, s6, s7;
+		};
+		__extension__ struct
+		{
+			cl_int4 lo, hi;
+		};
 #endif
-#if defined( __CL_INT2__) 
-    __cl_int2     v2;
+#if defined(__CL_INT2__)
+		__cl_int2 v2[4];
 #endif
-}cl_int2;
+#if defined(__CL_INT4__)
+		__cl_int4 v4[2];
+#endif
+#if defined(__CL_INT8__)
+		__cl_int8 v8;
+#endif
+	} cl_int8;
 
-typedef union
-{
-    cl_int  CL_ALIGNED(16) s[4];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_int  x, y, z, w; };
-   __extension__ struct{ cl_int  s0, s1, s2, s3; };
-   __extension__ struct{ cl_int2 lo, hi; };
+	typedef union {
+		cl_int CL_ALIGNED(64) s[16];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_int x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf;
+		};
+		__extension__ struct
+		{
+			cl_int s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF;
+		};
+		__extension__ struct
+		{
+			cl_int8 lo, hi;
+		};
 #endif
-#if defined( __CL_INT2__) 
-    __cl_int2     v2[2];
+#if defined(__CL_INT2__)
+		__cl_int2 v2[8];
 #endif
-#if defined( __CL_INT4__) 
-    __cl_int4     v4;
+#if defined(__CL_INT4__)
+		__cl_int4 v4[4];
 #endif
-}cl_int4;
+#if defined(__CL_INT8__)
+		__cl_int8 v8[2];
+#endif
+#if defined(__CL_INT16__)
+		__cl_int16 v16;
+#endif
+	} cl_int16;
 
-/* cl_int3 is identical in size, alignment and behavior to cl_int4. See section 6.1.5. */
-typedef  cl_int4  cl_int3;
+	/* ---- cl_uintn ---- */
+	typedef union {
+		cl_uint CL_ALIGNED(8) s[2];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_uint x, y;
+		};
+		__extension__ struct
+		{
+			cl_uint s0, s1;
+		};
+		__extension__ struct
+		{
+			cl_uint lo, hi;
+		};
+#endif
+#if defined(__CL_UINT2__)
+		__cl_uint2 v2;
+#endif
+	} cl_uint2;
 
-typedef union
-{
-    cl_int   CL_ALIGNED(32) s[8];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_int  x, y, z, w; };
-   __extension__ struct{ cl_int  s0, s1, s2, s3, s4, s5, s6, s7; };
-   __extension__ struct{ cl_int4 lo, hi; };
+	typedef union {
+		cl_uint CL_ALIGNED(16) s[4];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_uint x, y, z, w;
+		};
+		__extension__ struct
+		{
+			cl_uint s0, s1, s2, s3;
+		};
+		__extension__ struct
+		{
+			cl_uint2 lo, hi;
+		};
 #endif
-#if defined( __CL_INT2__) 
-    __cl_int2     v2[4];
+#if defined(__CL_UINT2__)
+		__cl_uint2 v2[2];
 #endif
-#if defined( __CL_INT4__) 
-    __cl_int4     v4[2];
+#if defined(__CL_UINT4__)
+		__cl_uint4 v4;
 #endif
-#if defined( __CL_INT8__ )
-    __cl_int8     v8;
-#endif
-}cl_int8;
+	} cl_uint4;
 
-typedef union
-{
-    cl_int  CL_ALIGNED(64) s[16];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_int  x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; };
-   __extension__ struct{ cl_int  s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
-   __extension__ struct{ cl_int8 lo, hi; };
-#endif
-#if defined( __CL_INT2__) 
-    __cl_int2     v2[8];
-#endif
-#if defined( __CL_INT4__) 
-    __cl_int4     v4[4];
-#endif
-#if defined( __CL_INT8__ )
-    __cl_int8     v8[2];
-#endif
-#if defined( __CL_INT16__ )
-    __cl_int16    v16;
-#endif
-}cl_int16;
+	/* cl_uint3 is identical in size, alignment and behavior to cl_uint4. See section 6.1.5. */
+	typedef cl_uint4 cl_uint3;
 
+	typedef union {
+		cl_uint CL_ALIGNED(32) s[8];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_uint x, y, z, w;
+		};
+		__extension__ struct
+		{
+			cl_uint s0, s1, s2, s3, s4, s5, s6, s7;
+		};
+		__extension__ struct
+		{
+			cl_uint4 lo, hi;
+		};
+#endif
+#if defined(__CL_UINT2__)
+		__cl_uint2 v2[4];
+#endif
+#if defined(__CL_UINT4__)
+		__cl_uint4 v4[2];
+#endif
+#if defined(__CL_UINT8__)
+		__cl_uint8 v8;
+#endif
+	} cl_uint8;
 
-/* ---- cl_uintn ---- */
-typedef union
-{
-    cl_uint  CL_ALIGNED(8) s[2];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_uint  x, y; };
-   __extension__ struct{ cl_uint  s0, s1; };
-   __extension__ struct{ cl_uint  lo, hi; };
+	typedef union {
+		cl_uint CL_ALIGNED(64) s[16];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_uint x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf;
+		};
+		__extension__ struct
+		{
+			cl_uint s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF;
+		};
+		__extension__ struct
+		{
+			cl_uint8 lo, hi;
+		};
 #endif
-#if defined( __CL_UINT2__) 
-    __cl_uint2     v2;
+#if defined(__CL_UINT2__)
+		__cl_uint2 v2[8];
 #endif
-}cl_uint2;
+#if defined(__CL_UINT4__)
+		__cl_uint4 v4[4];
+#endif
+#if defined(__CL_UINT8__)
+		__cl_uint8 v8[2];
+#endif
+#if defined(__CL_UINT16__)
+		__cl_uint16 v16;
+#endif
+	} cl_uint16;
 
-typedef union
-{
-    cl_uint  CL_ALIGNED(16) s[4];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_uint  x, y, z, w; };
-   __extension__ struct{ cl_uint  s0, s1, s2, s3; };
-   __extension__ struct{ cl_uint2 lo, hi; };
+	/* ---- cl_longn ---- */
+	typedef union {
+		cl_long CL_ALIGNED(16) s[2];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_long x, y;
+		};
+		__extension__ struct
+		{
+			cl_long s0, s1;
+		};
+		__extension__ struct
+		{
+			cl_long lo, hi;
+		};
 #endif
-#if defined( __CL_UINT2__) 
-    __cl_uint2     v2[2];
+#if defined(__CL_LONG2__)
+		__cl_long2 v2;
 #endif
-#if defined( __CL_UINT4__) 
-    __cl_uint4     v4;
-#endif
-}cl_uint4;
+	} cl_long2;
 
-/* cl_uint3 is identical in size, alignment and behavior to cl_uint4. See section 6.1.5. */
-typedef  cl_uint4  cl_uint3;
+	typedef union {
+		cl_long CL_ALIGNED(32) s[4];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_long x, y, z, w;
+		};
+		__extension__ struct
+		{
+			cl_long s0, s1, s2, s3;
+		};
+		__extension__ struct
+		{
+			cl_long2 lo, hi;
+		};
+#endif
+#if defined(__CL_LONG2__)
+		__cl_long2 v2[2];
+#endif
+#if defined(__CL_LONG4__)
+		__cl_long4 v4;
+#endif
+	} cl_long4;
 
-typedef union
-{
-    cl_uint   CL_ALIGNED(32) s[8];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_uint  x, y, z, w; };
-   __extension__ struct{ cl_uint  s0, s1, s2, s3, s4, s5, s6, s7; };
-   __extension__ struct{ cl_uint4 lo, hi; };
-#endif
-#if defined( __CL_UINT2__) 
-    __cl_uint2     v2[4];
-#endif
-#if defined( __CL_UINT4__) 
-    __cl_uint4     v4[2];
-#endif
-#if defined( __CL_UINT8__ )
-    __cl_uint8     v8;
-#endif
-}cl_uint8;
+	/* cl_long3 is identical in size, alignment and behavior to cl_long4. See section 6.1.5. */
+	typedef cl_long4 cl_long3;
 
-typedef union
-{
-    cl_uint  CL_ALIGNED(64) s[16];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_uint  x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; };
-   __extension__ struct{ cl_uint  s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
-   __extension__ struct{ cl_uint8 lo, hi; };
+	typedef union {
+		cl_long CL_ALIGNED(64) s[8];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_long x, y, z, w;
+		};
+		__extension__ struct
+		{
+			cl_long s0, s1, s2, s3, s4, s5, s6, s7;
+		};
+		__extension__ struct
+		{
+			cl_long4 lo, hi;
+		};
 #endif
-#if defined( __CL_UINT2__) 
-    __cl_uint2     v2[8];
+#if defined(__CL_LONG2__)
+		__cl_long2 v2[4];
 #endif
-#if defined( __CL_UINT4__) 
-    __cl_uint4     v4[4];
+#if defined(__CL_LONG4__)
+		__cl_long4 v4[2];
 #endif
-#if defined( __CL_UINT8__ )
-    __cl_uint8     v8[2];
+#if defined(__CL_LONG8__)
+		__cl_long8 v8;
 #endif
-#if defined( __CL_UINT16__ )
-    __cl_uint16    v16;
-#endif
-}cl_uint16;
+	} cl_long8;
 
-/* ---- cl_longn ---- */
-typedef union
-{
-    cl_long  CL_ALIGNED(16) s[2];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_long  x, y; };
-   __extension__ struct{ cl_long  s0, s1; };
-   __extension__ struct{ cl_long  lo, hi; };
+	typedef union {
+		cl_long CL_ALIGNED(128) s[16];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_long x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf;
+		};
+		__extension__ struct
+		{
+			cl_long s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF;
+		};
+		__extension__ struct
+		{
+			cl_long8 lo, hi;
+		};
 #endif
-#if defined( __CL_LONG2__) 
-    __cl_long2     v2;
+#if defined(__CL_LONG2__)
+		__cl_long2 v2[8];
 #endif
-}cl_long2;
+#if defined(__CL_LONG4__)
+		__cl_long4 v4[4];
+#endif
+#if defined(__CL_LONG8__)
+		__cl_long8 v8[2];
+#endif
+#if defined(__CL_LONG16__)
+		__cl_long16 v16;
+#endif
+	} cl_long16;
 
-typedef union
-{
-    cl_long  CL_ALIGNED(32) s[4];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_long  x, y, z, w; };
-   __extension__ struct{ cl_long  s0, s1, s2, s3; };
-   __extension__ struct{ cl_long2 lo, hi; };
+	/* ---- cl_ulongn ---- */
+	typedef union {
+		cl_ulong CL_ALIGNED(16) s[2];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_ulong x, y;
+		};
+		__extension__ struct
+		{
+			cl_ulong s0, s1;
+		};
+		__extension__ struct
+		{
+			cl_ulong lo, hi;
+		};
 #endif
-#if defined( __CL_LONG2__) 
-    __cl_long2     v2[2];
+#if defined(__CL_ULONG2__)
+		__cl_ulong2 v2;
 #endif
-#if defined( __CL_LONG4__) 
-    __cl_long4     v4;
-#endif
-}cl_long4;
+	} cl_ulong2;
 
-/* cl_long3 is identical in size, alignment and behavior to cl_long4. See section 6.1.5. */
-typedef  cl_long4  cl_long3;
+	typedef union {
+		cl_ulong CL_ALIGNED(32) s[4];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_ulong x, y, z, w;
+		};
+		__extension__ struct
+		{
+			cl_ulong s0, s1, s2, s3;
+		};
+		__extension__ struct
+		{
+			cl_ulong2 lo, hi;
+		};
+#endif
+#if defined(__CL_ULONG2__)
+		__cl_ulong2 v2[2];
+#endif
+#if defined(__CL_ULONG4__)
+		__cl_ulong4 v4;
+#endif
+	} cl_ulong4;
 
-typedef union
-{
-    cl_long   CL_ALIGNED(64) s[8];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_long  x, y, z, w; };
-   __extension__ struct{ cl_long  s0, s1, s2, s3, s4, s5, s6, s7; };
-   __extension__ struct{ cl_long4 lo, hi; };
-#endif
-#if defined( __CL_LONG2__) 
-    __cl_long2     v2[4];
-#endif
-#if defined( __CL_LONG4__) 
-    __cl_long4     v4[2];
-#endif
-#if defined( __CL_LONG8__ )
-    __cl_long8     v8;
-#endif
-}cl_long8;
+	/* cl_ulong3 is identical in size, alignment and behavior to cl_ulong4. See section 6.1.5. */
+	typedef cl_ulong4 cl_ulong3;
 
-typedef union
-{
-    cl_long  CL_ALIGNED(128) s[16];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_long  x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; };
-   __extension__ struct{ cl_long  s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
-   __extension__ struct{ cl_long8 lo, hi; };
+	typedef union {
+		cl_ulong CL_ALIGNED(64) s[8];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_ulong x, y, z, w;
+		};
+		__extension__ struct
+		{
+			cl_ulong s0, s1, s2, s3, s4, s5, s6, s7;
+		};
+		__extension__ struct
+		{
+			cl_ulong4 lo, hi;
+		};
 #endif
-#if defined( __CL_LONG2__) 
-    __cl_long2     v2[8];
+#if defined(__CL_ULONG2__)
+		__cl_ulong2 v2[4];
 #endif
-#if defined( __CL_LONG4__) 
-    __cl_long4     v4[4];
+#if defined(__CL_ULONG4__)
+		__cl_ulong4 v4[2];
 #endif
-#if defined( __CL_LONG8__ )
-    __cl_long8     v8[2];
+#if defined(__CL_ULONG8__)
+		__cl_ulong8 v8;
 #endif
-#if defined( __CL_LONG16__ )
-    __cl_long16    v16;
-#endif
-}cl_long16;
+	} cl_ulong8;
 
+	typedef union {
+		cl_ulong CL_ALIGNED(128) s[16];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_ulong x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf;
+		};
+		__extension__ struct
+		{
+			cl_ulong s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF;
+		};
+		__extension__ struct
+		{
+			cl_ulong8 lo, hi;
+		};
+#endif
+#if defined(__CL_ULONG2__)
+		__cl_ulong2 v2[8];
+#endif
+#if defined(__CL_ULONG4__)
+		__cl_ulong4 v4[4];
+#endif
+#if defined(__CL_ULONG8__)
+		__cl_ulong8 v8[2];
+#endif
+#if defined(__CL_ULONG16__)
+		__cl_ulong16 v16;
+#endif
+	} cl_ulong16;
 
-/* ---- cl_ulongn ---- */
-typedef union
-{
-    cl_ulong  CL_ALIGNED(16) s[2];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_ulong  x, y; };
-   __extension__ struct{ cl_ulong  s0, s1; };
-   __extension__ struct{ cl_ulong  lo, hi; };
-#endif
-#if defined( __CL_ULONG2__) 
-    __cl_ulong2     v2;
-#endif
-}cl_ulong2;
+	/* --- cl_floatn ---- */
 
-typedef union
-{
-    cl_ulong  CL_ALIGNED(32) s[4];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_ulong  x, y, z, w; };
-   __extension__ struct{ cl_ulong  s0, s1, s2, s3; };
-   __extension__ struct{ cl_ulong2 lo, hi; };
+	typedef union {
+		cl_float CL_ALIGNED(8) s[2];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_float x, y;
+		};
+		__extension__ struct
+		{
+			cl_float s0, s1;
+		};
+		__extension__ struct
+		{
+			cl_float lo, hi;
+		};
 #endif
-#if defined( __CL_ULONG2__) 
-    __cl_ulong2     v2[2];
+#if defined(__CL_FLOAT2__)
+		__cl_float2 v2;
 #endif
-#if defined( __CL_ULONG4__) 
-    __cl_ulong4     v4;
-#endif
-}cl_ulong4;
+	} cl_float2;
 
-/* cl_ulong3 is identical in size, alignment and behavior to cl_ulong4. See section 6.1.5. */
-typedef  cl_ulong4  cl_ulong3;
+	typedef union {
+		cl_float CL_ALIGNED(16) s[4];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_float x, y, z, w;
+		};
+		__extension__ struct
+		{
+			cl_float s0, s1, s2, s3;
+		};
+		__extension__ struct
+		{
+			cl_float2 lo, hi;
+		};
+#endif
+#if defined(__CL_FLOAT2__)
+		__cl_float2 v2[2];
+#endif
+#if defined(__CL_FLOAT4__)
+		__cl_float4 v4;
+#endif
+	} cl_float4;
 
-typedef union
-{
-    cl_ulong   CL_ALIGNED(64) s[8];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_ulong  x, y, z, w; };
-   __extension__ struct{ cl_ulong  s0, s1, s2, s3, s4, s5, s6, s7; };
-   __extension__ struct{ cl_ulong4 lo, hi; };
-#endif
-#if defined( __CL_ULONG2__) 
-    __cl_ulong2     v2[4];
-#endif
-#if defined( __CL_ULONG4__) 
-    __cl_ulong4     v4[2];
-#endif
-#if defined( __CL_ULONG8__ )
-    __cl_ulong8     v8;
-#endif
-}cl_ulong8;
+	/* cl_float3 is identical in size, alignment and behavior to cl_float4. See section 6.1.5. */
+	typedef cl_float4 cl_float3;
 
-typedef union
-{
-    cl_ulong  CL_ALIGNED(128) s[16];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_ulong  x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; };
-   __extension__ struct{ cl_ulong  s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
-   __extension__ struct{ cl_ulong8 lo, hi; };
+	typedef union {
+		cl_float CL_ALIGNED(32) s[8];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_float x, y, z, w;
+		};
+		__extension__ struct
+		{
+			cl_float s0, s1, s2, s3, s4, s5, s6, s7;
+		};
+		__extension__ struct
+		{
+			cl_float4 lo, hi;
+		};
 #endif
-#if defined( __CL_ULONG2__) 
-    __cl_ulong2     v2[8];
+#if defined(__CL_FLOAT2__)
+		__cl_float2 v2[4];
 #endif
-#if defined( __CL_ULONG4__) 
-    __cl_ulong4     v4[4];
+#if defined(__CL_FLOAT4__)
+		__cl_float4 v4[2];
 #endif
-#if defined( __CL_ULONG8__ )
-    __cl_ulong8     v8[2];
+#if defined(__CL_FLOAT8__)
+		__cl_float8 v8;
 #endif
-#if defined( __CL_ULONG16__ )
-    __cl_ulong16    v16;
-#endif
-}cl_ulong16;
+	} cl_float8;
 
+	typedef union {
+		cl_float CL_ALIGNED(64) s[16];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_float x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf;
+		};
+		__extension__ struct
+		{
+			cl_float s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF;
+		};
+		__extension__ struct
+		{
+			cl_float8 lo, hi;
+		};
+#endif
+#if defined(__CL_FLOAT2__)
+		__cl_float2 v2[8];
+#endif
+#if defined(__CL_FLOAT4__)
+		__cl_float4 v4[4];
+#endif
+#if defined(__CL_FLOAT8__)
+		__cl_float8 v8[2];
+#endif
+#if defined(__CL_FLOAT16__)
+		__cl_float16 v16;
+#endif
+	} cl_float16;
 
-/* --- cl_floatn ---- */
+	/* --- cl_doublen ---- */
 
-typedef union
-{
-    cl_float  CL_ALIGNED(8) s[2];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_float  x, y; };
-   __extension__ struct{ cl_float  s0, s1; };
-   __extension__ struct{ cl_float  lo, hi; };
+	typedef union {
+		cl_double CL_ALIGNED(16) s[2];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_double x, y;
+		};
+		__extension__ struct
+		{
+			cl_double s0, s1;
+		};
+		__extension__ struct
+		{
+			cl_double lo, hi;
+		};
 #endif
-#if defined( __CL_FLOAT2__) 
-    __cl_float2     v2;
+#if defined(__CL_DOUBLE2__)
+		__cl_double2 v2;
 #endif
-}cl_float2;
+	} cl_double2;
 
-typedef union
-{
-    cl_float  CL_ALIGNED(16) s[4];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_float   x, y, z, w; };
-   __extension__ struct{ cl_float   s0, s1, s2, s3; };
-   __extension__ struct{ cl_float2  lo, hi; };
+	typedef union {
+		cl_double CL_ALIGNED(32) s[4];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_double x, y, z, w;
+		};
+		__extension__ struct
+		{
+			cl_double s0, s1, s2, s3;
+		};
+		__extension__ struct
+		{
+			cl_double2 lo, hi;
+		};
 #endif
-#if defined( __CL_FLOAT2__) 
-    __cl_float2     v2[2];
+#if defined(__CL_DOUBLE2__)
+		__cl_double2 v2[2];
 #endif
-#if defined( __CL_FLOAT4__) 
-    __cl_float4     v4;
+#if defined(__CL_DOUBLE4__)
+		__cl_double4 v4;
 #endif
-}cl_float4;
+	} cl_double4;
 
-/* cl_float3 is identical in size, alignment and behavior to cl_float4. See section 6.1.5. */
-typedef  cl_float4  cl_float3;
+	/* cl_double3 is identical in size, alignment and behavior to cl_double4. See section 6.1.5. */
+	typedef cl_double4 cl_double3;
 
-typedef union
-{
-    cl_float   CL_ALIGNED(32) s[8];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_float   x, y, z, w; };
-   __extension__ struct{ cl_float   s0, s1, s2, s3, s4, s5, s6, s7; };
-   __extension__ struct{ cl_float4  lo, hi; };
+	typedef union {
+		cl_double CL_ALIGNED(64) s[8];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_double x, y, z, w;
+		};
+		__extension__ struct
+		{
+			cl_double s0, s1, s2, s3, s4, s5, s6, s7;
+		};
+		__extension__ struct
+		{
+			cl_double4 lo, hi;
+		};
 #endif
-#if defined( __CL_FLOAT2__) 
-    __cl_float2     v2[4];
+#if defined(__CL_DOUBLE2__)
+		__cl_double2 v2[4];
 #endif
-#if defined( __CL_FLOAT4__) 
-    __cl_float4     v4[2];
+#if defined(__CL_DOUBLE4__)
+		__cl_double4 v4[2];
 #endif
-#if defined( __CL_FLOAT8__ )
-    __cl_float8     v8;
+#if defined(__CL_DOUBLE8__)
+		__cl_double8 v8;
 #endif
-}cl_float8;
+	} cl_double8;
 
-typedef union
-{
-    cl_float  CL_ALIGNED(64) s[16];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_float  x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; };
-   __extension__ struct{ cl_float  s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
-   __extension__ struct{ cl_float8 lo, hi; };
+	typedef union {
+		cl_double CL_ALIGNED(128) s[16];
+#if defined(CL_NAMED_STRUCT_SUPPORTED)
+		__extension__ struct
+		{
+			cl_double x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf;
+		};
+		__extension__ struct
+		{
+			cl_double s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF;
+		};
+		__extension__ struct
+		{
+			cl_double8 lo, hi;
+		};
 #endif
-#if defined( __CL_FLOAT2__) 
-    __cl_float2     v2[8];
+#if defined(__CL_DOUBLE2__)
+		__cl_double2 v2[8];
 #endif
-#if defined( __CL_FLOAT4__) 
-    __cl_float4     v4[4];
+#if defined(__CL_DOUBLE4__)
+		__cl_double4 v4[4];
 #endif
-#if defined( __CL_FLOAT8__ )
-    __cl_float8     v8[2];
+#if defined(__CL_DOUBLE8__)
+		__cl_double8 v8[2];
 #endif
-#if defined( __CL_FLOAT16__ )
-    __cl_float16    v16;
+#if defined(__CL_DOUBLE16__)
+		__cl_double16 v16;
 #endif
-}cl_float16;
-
-/* --- cl_doublen ---- */
-
-typedef union
-{
-    cl_double  CL_ALIGNED(16) s[2];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_double  x, y; };
-   __extension__ struct{ cl_double s0, s1; };
-   __extension__ struct{ cl_double lo, hi; };
-#endif
-#if defined( __CL_DOUBLE2__) 
-    __cl_double2     v2;
-#endif
-}cl_double2;
-
-typedef union
-{
-    cl_double  CL_ALIGNED(32) s[4];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_double  x, y, z, w; };
-   __extension__ struct{ cl_double  s0, s1, s2, s3; };
-   __extension__ struct{ cl_double2 lo, hi; };
-#endif
-#if defined( __CL_DOUBLE2__) 
-    __cl_double2     v2[2];
-#endif
-#if defined( __CL_DOUBLE4__) 
-    __cl_double4     v4;
-#endif
-}cl_double4;
-
-/* cl_double3 is identical in size, alignment and behavior to cl_double4. See section 6.1.5. */
-typedef  cl_double4  cl_double3;
-
-typedef union
-{
-    cl_double   CL_ALIGNED(64) s[8];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_double  x, y, z, w; };
-   __extension__ struct{ cl_double  s0, s1, s2, s3, s4, s5, s6, s7; };
-   __extension__ struct{ cl_double4 lo, hi; };
-#endif
-#if defined( __CL_DOUBLE2__) 
-    __cl_double2     v2[4];
-#endif
-#if defined( __CL_DOUBLE4__) 
-    __cl_double4     v4[2];
-#endif
-#if defined( __CL_DOUBLE8__ )
-    __cl_double8     v8;
-#endif
-}cl_double8;
-
-typedef union
-{
-    cl_double  CL_ALIGNED(128) s[16];
-#if defined( CL_NAMED_STRUCT_SUPPORTED )
-   __extension__ struct{ cl_double  x, y, z, w, __spacer4, __spacer5, __spacer6, __spacer7, __spacer8, __spacer9, sa, sb, sc, sd, se, sf; };
-   __extension__ struct{ cl_double  s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, sA, sB, sC, sD, sE, sF; };
-   __extension__ struct{ cl_double8 lo, hi; };
-#endif
-#if defined( __CL_DOUBLE2__) 
-    __cl_double2     v2[8];
-#endif
-#if defined( __CL_DOUBLE4__) 
-    __cl_double4     v4[4];
-#endif
-#if defined( __CL_DOUBLE8__ )
-    __cl_double8     v8[2];
-#endif
-#if defined( __CL_DOUBLE16__ )
-    __cl_double16    v16;
-#endif
-}cl_double16;
+	} cl_double16;
 
 /* Macro to facilitate debugging 
  * Usage:
@@ -1237,967 +1551,967 @@ typedef union
  * This should correctly set up the line, (column) and file information for your source 
  * string so you can do source level debugging.
  */
-#define  __CL_STRINGIFY( _x )               # _x
-#define  _CL_STRINGIFY( _x )                __CL_STRINGIFY( _x )
-#define  CL_PROGRAM_STRING_DEBUG_INFO       "#line "  _CL_STRINGIFY(__LINE__) " \"" __FILE__ "\" \n\n" 
+#define __CL_STRINGIFY(_x) #_x
+#define _CL_STRINGIFY(_x) __CL_STRINGIFY(_x)
+#define CL_PROGRAM_STRING_DEBUG_INFO "#line " _CL_STRINGIFY(__LINE__) " \"" __FILE__ "\" \n\n"
 
-//  CL.h contents
-/******************************************************************************/
+	//  CL.h contents
+	/******************************************************************************/
 
-typedef struct _cl_platform_id *    cl_platform_id;
-typedef struct _cl_device_id *      cl_device_id;
-typedef struct _cl_context *        cl_context;
-typedef struct _cl_command_queue *  cl_command_queue;
-typedef struct _cl_mem *            cl_mem;
-typedef struct _cl_program *        cl_program;
-typedef struct _cl_kernel *         cl_kernel;
-typedef struct _cl_event *          cl_event;
-typedef struct _cl_sampler *        cl_sampler;
+	typedef struct _cl_platform_id *cl_platform_id;
+	typedef struct _cl_device_id *cl_device_id;
+	typedef struct _cl_context *cl_context;
+	typedef struct _cl_command_queue *cl_command_queue;
+	typedef struct _cl_mem *cl_mem;
+	typedef struct _cl_program *cl_program;
+	typedef struct _cl_kernel *cl_kernel;
+	typedef struct _cl_event *cl_event;
+	typedef struct _cl_sampler *cl_sampler;
 
-typedef cl_uint             cl_bool;                     /* WARNING!  Unlike cl_ types in cl_platform.h, cl_bool is not guaranteed to be the same size as the bool in kernels. */ 
-typedef cl_ulong            cl_bitfield;
-typedef cl_bitfield         cl_device_type;
-typedef cl_uint             cl_platform_info;
-typedef cl_uint             cl_device_info;
-typedef cl_bitfield         cl_device_fp_config;
-typedef cl_uint             cl_device_mem_cache_type;
-typedef cl_uint             cl_device_local_mem_type;
-typedef cl_bitfield         cl_device_exec_capabilities;
-typedef cl_bitfield         cl_command_queue_properties;
+	typedef cl_uint cl_bool; /* WARNING!  Unlike cl_ types in cl_platform.h, cl_bool is not guaranteed to be the same size as the bool in kernels. */
+	typedef cl_ulong cl_bitfield;
+	typedef cl_bitfield cl_device_type;
+	typedef cl_uint cl_platform_info;
+	typedef cl_uint cl_device_info;
+	typedef cl_bitfield cl_device_fp_config;
+	typedef cl_uint cl_device_mem_cache_type;
+	typedef cl_uint cl_device_local_mem_type;
+	typedef cl_bitfield cl_device_exec_capabilities;
+	typedef cl_bitfield cl_command_queue_properties;
 
-typedef intptr_t			cl_context_properties;
-typedef cl_uint             cl_context_info;
-typedef cl_uint             cl_command_queue_info;
-typedef cl_uint             cl_channel_order;
-typedef cl_uint             cl_channel_type;
-typedef cl_bitfield         cl_mem_flags;
-typedef cl_uint             cl_mem_object_type;
-typedef cl_uint             cl_mem_info;
-typedef cl_uint             cl_image_info;
-typedef cl_uint             cl_buffer_create_type;
-typedef cl_uint             cl_addressing_mode;
-typedef cl_uint             cl_filter_mode;
-typedef cl_uint             cl_sampler_info;
-typedef cl_bitfield         cl_map_flags;
-typedef cl_uint             cl_program_info;
-typedef cl_uint             cl_program_build_info;
-typedef cl_int              cl_build_status;
-typedef cl_uint             cl_kernel_info;
-typedef cl_uint             cl_kernel_work_group_info;
-typedef cl_uint             cl_event_info;
-typedef cl_uint             cl_command_type;
-typedef cl_uint             cl_profiling_info;
+	typedef intptr_t cl_context_properties;
+	typedef cl_uint cl_context_info;
+	typedef cl_uint cl_command_queue_info;
+	typedef cl_uint cl_channel_order;
+	typedef cl_uint cl_channel_type;
+	typedef cl_bitfield cl_mem_flags;
+	typedef cl_uint cl_mem_object_type;
+	typedef cl_uint cl_mem_info;
+	typedef cl_uint cl_image_info;
+	typedef cl_uint cl_buffer_create_type;
+	typedef cl_uint cl_addressing_mode;
+	typedef cl_uint cl_filter_mode;
+	typedef cl_uint cl_sampler_info;
+	typedef cl_bitfield cl_map_flags;
+	typedef cl_uint cl_program_info;
+	typedef cl_uint cl_program_build_info;
+	typedef cl_int cl_build_status;
+	typedef cl_uint cl_kernel_info;
+	typedef cl_uint cl_kernel_work_group_info;
+	typedef cl_uint cl_event_info;
+	typedef cl_uint cl_command_type;
+	typedef cl_uint cl_profiling_info;
 
-typedef struct _cl_image_format {
-    cl_channel_order        image_channel_order;
-    cl_channel_type         image_channel_data_type;
-} cl_image_format;
+	typedef struct _cl_image_format
+	{
+		cl_channel_order image_channel_order;
+		cl_channel_type image_channel_data_type;
+	} cl_image_format;
 
-
-typedef struct _cl_buffer_region {
-    size_t                  origin;
-    size_t                  size;
-} cl_buffer_region;
+	typedef struct _cl_buffer_region
+	{
+		size_t origin;
+		size_t size;
+	} cl_buffer_region;
 
 /******************************************************************************/
 
 /* Error Codes */
-#define CL_SUCCESS                                  0
-#define CL_DEVICE_NOT_FOUND                         -1
-#define CL_DEVICE_NOT_AVAILABLE                     -2
-#define CL_COMPILER_NOT_AVAILABLE                   -3
-#define CL_MEM_OBJECT_ALLOCATION_FAILURE            -4
-#define CL_OUT_OF_RESOURCES                         -5
-#define CL_OUT_OF_HOST_MEMORY                       -6
-#define CL_PROFILING_INFO_NOT_AVAILABLE             -7
-#define CL_MEM_COPY_OVERLAP                         -8
-#define CL_IMAGE_FORMAT_MISMATCH                    -9
-#define CL_IMAGE_FORMAT_NOT_SUPPORTED               -10
-#define CL_BUILD_PROGRAM_FAILURE                    -11
-#define CL_MAP_FAILURE                              -12
-#define CL_MISALIGNED_SUB_BUFFER_OFFSET             -13
+#define CL_SUCCESS 0
+#define CL_DEVICE_NOT_FOUND -1
+#define CL_DEVICE_NOT_AVAILABLE -2
+#define CL_COMPILER_NOT_AVAILABLE -3
+#define CL_MEM_OBJECT_ALLOCATION_FAILURE -4
+#define CL_OUT_OF_RESOURCES -5
+#define CL_OUT_OF_HOST_MEMORY -6
+#define CL_PROFILING_INFO_NOT_AVAILABLE -7
+#define CL_MEM_COPY_OVERLAP -8
+#define CL_IMAGE_FORMAT_MISMATCH -9
+#define CL_IMAGE_FORMAT_NOT_SUPPORTED -10
+#define CL_BUILD_PROGRAM_FAILURE -11
+#define CL_MAP_FAILURE -12
+#define CL_MISALIGNED_SUB_BUFFER_OFFSET -13
 #define CL_EXEC_STATUS_ERROR_FOR_EVENTS_IN_WAIT_LIST -14
 
-#define CL_INVALID_VALUE                            -30
-#define CL_INVALID_DEVICE_TYPE                      -31
-#define CL_INVALID_PLATFORM                         -32
-#define CL_INVALID_DEVICE                           -33
-#define CL_INVALID_CONTEXT                          -34
-#define CL_INVALID_QUEUE_PROPERTIES                 -35
-#define CL_INVALID_COMMAND_QUEUE                    -36
-#define CL_INVALID_HOST_PTR                         -37
-#define CL_INVALID_MEM_OBJECT                       -38
-#define CL_INVALID_IMAGE_FORMAT_DESCRIPTOR          -39
-#define CL_INVALID_IMAGE_SIZE                       -40
-#define CL_INVALID_SAMPLER                          -41
-#define CL_INVALID_BINARY                           -42
-#define CL_INVALID_BUILD_OPTIONS                    -43
-#define CL_INVALID_PROGRAM                          -44
-#define CL_INVALID_PROGRAM_EXECUTABLE               -45
-#define CL_INVALID_KERNEL_NAME                      -46
-#define CL_INVALID_KERNEL_DEFINITION                -47
-#define CL_INVALID_KERNEL                           -48
-#define CL_INVALID_ARG_INDEX                        -49
-#define CL_INVALID_ARG_VALUE                        -50
-#define CL_INVALID_ARG_SIZE                         -51
-#define CL_INVALID_KERNEL_ARGS                      -52
-#define CL_INVALID_WORK_DIMENSION                   -53
-#define CL_INVALID_WORK_GROUP_SIZE                  -54
-#define CL_INVALID_WORK_ITEM_SIZE                   -55
-#define CL_INVALID_GLOBAL_OFFSET                    -56
-#define CL_INVALID_EVENT_WAIT_LIST                  -57
-#define CL_INVALID_EVENT                            -58
-#define CL_INVALID_OPERATION                        -59
-#define CL_INVALID_GL_OBJECT                        -60
-#define CL_INVALID_BUFFER_SIZE                      -61
-#define CL_INVALID_MIP_LEVEL                        -62
-#define CL_INVALID_GLOBAL_WORK_SIZE                 -63
-#define CL_INVALID_PROPERTY                         -64
+#define CL_INVALID_VALUE -30
+#define CL_INVALID_DEVICE_TYPE -31
+#define CL_INVALID_PLATFORM -32
+#define CL_INVALID_DEVICE -33
+#define CL_INVALID_CONTEXT -34
+#define CL_INVALID_QUEUE_PROPERTIES -35
+#define CL_INVALID_COMMAND_QUEUE -36
+#define CL_INVALID_HOST_PTR -37
+#define CL_INVALID_MEM_OBJECT -38
+#define CL_INVALID_IMAGE_FORMAT_DESCRIPTOR -39
+#define CL_INVALID_IMAGE_SIZE -40
+#define CL_INVALID_SAMPLER -41
+#define CL_INVALID_BINARY -42
+#define CL_INVALID_BUILD_OPTIONS -43
+#define CL_INVALID_PROGRAM -44
+#define CL_INVALID_PROGRAM_EXECUTABLE -45
+#define CL_INVALID_KERNEL_NAME -46
+#define CL_INVALID_KERNEL_DEFINITION -47
+#define CL_INVALID_KERNEL -48
+#define CL_INVALID_ARG_INDEX -49
+#define CL_INVALID_ARG_VALUE -50
+#define CL_INVALID_ARG_SIZE -51
+#define CL_INVALID_KERNEL_ARGS -52
+#define CL_INVALID_WORK_DIMENSION -53
+#define CL_INVALID_WORK_GROUP_SIZE -54
+#define CL_INVALID_WORK_ITEM_SIZE -55
+#define CL_INVALID_GLOBAL_OFFSET -56
+#define CL_INVALID_EVENT_WAIT_LIST -57
+#define CL_INVALID_EVENT -58
+#define CL_INVALID_OPERATION -59
+#define CL_INVALID_GL_OBJECT -60
+#define CL_INVALID_BUFFER_SIZE -61
+#define CL_INVALID_MIP_LEVEL -62
+#define CL_INVALID_GLOBAL_WORK_SIZE -63
+#define CL_INVALID_PROPERTY -64
 
 /* OpenCL Version */
-#define CL_VERSION_1_0                              1
-#define CL_VERSION_1_1                              1
+#define CL_VERSION_1_0 1
+#define CL_VERSION_1_1 1
 
 /* cl_bool */
-#define CL_FALSE                                    0
-#define CL_TRUE                                     1
+#define CL_FALSE 0
+#define CL_TRUE 1
 
 /* cl_platform_info */
-#define CL_PLATFORM_PROFILE                         0x0900
-#define CL_PLATFORM_VERSION                         0x0901
-#define CL_PLATFORM_NAME                            0x0902
-#define CL_PLATFORM_VENDOR                          0x0903
-#define CL_PLATFORM_EXTENSIONS                      0x0904
+#define CL_PLATFORM_PROFILE 0x0900
+#define CL_PLATFORM_VERSION 0x0901
+#define CL_PLATFORM_NAME 0x0902
+#define CL_PLATFORM_VENDOR 0x0903
+#define CL_PLATFORM_EXTENSIONS 0x0904
 
 /* cl_device_type - bitfield */
-#define CL_DEVICE_TYPE_DEFAULT                      (1 << 0)
-#define CL_DEVICE_TYPE_CPU                          (1 << 1)
-#define CL_DEVICE_TYPE_GPU                          (1 << 2)
-#define CL_DEVICE_TYPE_ACCELERATOR                  (1 << 3)
-#define CL_DEVICE_TYPE_ALL                          0xFFFFFFFF
+#define CL_DEVICE_TYPE_DEFAULT (1 << 0)
+#define CL_DEVICE_TYPE_CPU (1 << 1)
+#define CL_DEVICE_TYPE_GPU (1 << 2)
+#define CL_DEVICE_TYPE_ACCELERATOR (1 << 3)
+#define CL_DEVICE_TYPE_ALL 0xFFFFFFFF
 
 /* cl_device_info */
-#define CL_DEVICE_TYPE                              0x1000
-#define CL_DEVICE_VENDOR_ID                         0x1001
-#define CL_DEVICE_MAX_COMPUTE_UNITS                 0x1002
-#define CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS          0x1003
-#define CL_DEVICE_MAX_WORK_GROUP_SIZE               0x1004
-#define CL_DEVICE_MAX_WORK_ITEM_SIZES               0x1005
-#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_CHAR       0x1006
-#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_SHORT      0x1007
-#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_INT        0x1008
-#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_LONG       0x1009
-#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_FLOAT      0x100A
-#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_DOUBLE     0x100B
-#define CL_DEVICE_MAX_CLOCK_FREQUENCY               0x100C
-#define CL_DEVICE_ADDRESS_BITS                      0x100D
-#define CL_DEVICE_MAX_READ_IMAGE_ARGS               0x100E
-#define CL_DEVICE_MAX_WRITE_IMAGE_ARGS              0x100F
-#define CL_DEVICE_MAX_MEM_ALLOC_SIZE                0x1010
-#define CL_DEVICE_IMAGE2D_MAX_WIDTH                 0x1011
-#define CL_DEVICE_IMAGE2D_MAX_HEIGHT                0x1012
-#define CL_DEVICE_IMAGE3D_MAX_WIDTH                 0x1013
-#define CL_DEVICE_IMAGE3D_MAX_HEIGHT                0x1014
-#define CL_DEVICE_IMAGE3D_MAX_DEPTH                 0x1015
-#define CL_DEVICE_IMAGE_SUPPORT                     0x1016
-#define CL_DEVICE_MAX_PARAMETER_SIZE                0x1017
-#define CL_DEVICE_MAX_SAMPLERS                      0x1018
-#define CL_DEVICE_MEM_BASE_ADDR_ALIGN               0x1019
-#define CL_DEVICE_MIN_DATA_TYPE_ALIGN_SIZE          0x101A
-#define CL_DEVICE_SINGLE_FP_CONFIG                  0x101B
-#define CL_DEVICE_GLOBAL_MEM_CACHE_TYPE             0x101C
-#define CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE         0x101D
-#define CL_DEVICE_GLOBAL_MEM_CACHE_SIZE             0x101E
-#define CL_DEVICE_GLOBAL_MEM_SIZE                   0x101F
-#define CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE          0x1020
-#define CL_DEVICE_MAX_CONSTANT_ARGS                 0x1021
-#define CL_DEVICE_LOCAL_MEM_TYPE                    0x1022
-#define CL_DEVICE_LOCAL_MEM_SIZE                    0x1023
-#define CL_DEVICE_ERROR_CORRECTION_SUPPORT          0x1024
-#define CL_DEVICE_PROFILING_TIMER_RESOLUTION        0x1025
-#define CL_DEVICE_ENDIAN_LITTLE                     0x1026
-#define CL_DEVICE_AVAILABLE                         0x1027
-#define CL_DEVICE_COMPILER_AVAILABLE                0x1028
-#define CL_DEVICE_EXECUTION_CAPABILITIES            0x1029
-#define CL_DEVICE_QUEUE_PROPERTIES                  0x102A
-#define CL_DEVICE_NAME                              0x102B
-#define CL_DEVICE_VENDOR                            0x102C
-#define CL_DRIVER_VERSION                           0x102D
-#define CL_DEVICE_PROFILE                           0x102E
-#define CL_DEVICE_VERSION                           0x102F
-#define CL_DEVICE_EXTENSIONS                        0x1030
-#define CL_DEVICE_PLATFORM                          0x1031
+#define CL_DEVICE_TYPE 0x1000
+#define CL_DEVICE_VENDOR_ID 0x1001
+#define CL_DEVICE_MAX_COMPUTE_UNITS 0x1002
+#define CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS 0x1003
+#define CL_DEVICE_MAX_WORK_GROUP_SIZE 0x1004
+#define CL_DEVICE_MAX_WORK_ITEM_SIZES 0x1005
+#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_CHAR 0x1006
+#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_SHORT 0x1007
+#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_INT 0x1008
+#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_LONG 0x1009
+#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_FLOAT 0x100A
+#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_DOUBLE 0x100B
+#define CL_DEVICE_MAX_CLOCK_FREQUENCY 0x100C
+#define CL_DEVICE_ADDRESS_BITS 0x100D
+#define CL_DEVICE_MAX_READ_IMAGE_ARGS 0x100E
+#define CL_DEVICE_MAX_WRITE_IMAGE_ARGS 0x100F
+#define CL_DEVICE_MAX_MEM_ALLOC_SIZE 0x1010
+#define CL_DEVICE_IMAGE2D_MAX_WIDTH 0x1011
+#define CL_DEVICE_IMAGE2D_MAX_HEIGHT 0x1012
+#define CL_DEVICE_IMAGE3D_MAX_WIDTH 0x1013
+#define CL_DEVICE_IMAGE3D_MAX_HEIGHT 0x1014
+#define CL_DEVICE_IMAGE3D_MAX_DEPTH 0x1015
+#define CL_DEVICE_IMAGE_SUPPORT 0x1016
+#define CL_DEVICE_MAX_PARAMETER_SIZE 0x1017
+#define CL_DEVICE_MAX_SAMPLERS 0x1018
+#define CL_DEVICE_MEM_BASE_ADDR_ALIGN 0x1019
+#define CL_DEVICE_MIN_DATA_TYPE_ALIGN_SIZE 0x101A
+#define CL_DEVICE_SINGLE_FP_CONFIG 0x101B
+#define CL_DEVICE_GLOBAL_MEM_CACHE_TYPE 0x101C
+#define CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE 0x101D
+#define CL_DEVICE_GLOBAL_MEM_CACHE_SIZE 0x101E
+#define CL_DEVICE_GLOBAL_MEM_SIZE 0x101F
+#define CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE 0x1020
+#define CL_DEVICE_MAX_CONSTANT_ARGS 0x1021
+#define CL_DEVICE_LOCAL_MEM_TYPE 0x1022
+#define CL_DEVICE_LOCAL_MEM_SIZE 0x1023
+#define CL_DEVICE_ERROR_CORRECTION_SUPPORT 0x1024
+#define CL_DEVICE_PROFILING_TIMER_RESOLUTION 0x1025
+#define CL_DEVICE_ENDIAN_LITTLE 0x1026
+#define CL_DEVICE_AVAILABLE 0x1027
+#define CL_DEVICE_COMPILER_AVAILABLE 0x1028
+#define CL_DEVICE_EXECUTION_CAPABILITIES 0x1029
+#define CL_DEVICE_QUEUE_PROPERTIES 0x102A
+#define CL_DEVICE_NAME 0x102B
+#define CL_DEVICE_VENDOR 0x102C
+#define CL_DRIVER_VERSION 0x102D
+#define CL_DEVICE_PROFILE 0x102E
+#define CL_DEVICE_VERSION 0x102F
+#define CL_DEVICE_EXTENSIONS 0x1030
+#define CL_DEVICE_PLATFORM 0x1031
 /* 0x1032 reserved for CL_DEVICE_DOUBLE_FP_CONFIG */
 /* 0x1033 reserved for CL_DEVICE_HALF_FP_CONFIG */
-#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_HALF       0x1034
-#define CL_DEVICE_HOST_UNIFIED_MEMORY               0x1035
-#define CL_DEVICE_NATIVE_VECTOR_WIDTH_CHAR          0x1036
-#define CL_DEVICE_NATIVE_VECTOR_WIDTH_SHORT         0x1037
-#define CL_DEVICE_NATIVE_VECTOR_WIDTH_INT           0x1038
-#define CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG          0x1039
-#define CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT         0x103A
-#define CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE        0x103B
-#define CL_DEVICE_NATIVE_VECTOR_WIDTH_HALF          0x103C
-#define CL_DEVICE_OPENCL_C_VERSION                  0x103D
+#define CL_DEVICE_PREFERRED_VECTOR_WIDTH_HALF 0x1034
+#define CL_DEVICE_HOST_UNIFIED_MEMORY 0x1035
+#define CL_DEVICE_NATIVE_VECTOR_WIDTH_CHAR 0x1036
+#define CL_DEVICE_NATIVE_VECTOR_WIDTH_SHORT 0x1037
+#define CL_DEVICE_NATIVE_VECTOR_WIDTH_INT 0x1038
+#define CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG 0x1039
+#define CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT 0x103A
+#define CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE 0x103B
+#define CL_DEVICE_NATIVE_VECTOR_WIDTH_HALF 0x103C
+#define CL_DEVICE_OPENCL_C_VERSION 0x103D
 
 /* cl_device_fp_config - bitfield */
-#define CL_FP_DENORM                                (1 << 0)
-#define CL_FP_INF_NAN                               (1 << 1)
-#define CL_FP_ROUND_TO_NEAREST                      (1 << 2)
-#define CL_FP_ROUND_TO_ZERO                         (1 << 3)
-#define CL_FP_ROUND_TO_INF                          (1 << 4)
-#define CL_FP_FMA                                   (1 << 5)
-#define CL_FP_SOFT_FLOAT                            (1 << 6)
+#define CL_FP_DENORM (1 << 0)
+#define CL_FP_INF_NAN (1 << 1)
+#define CL_FP_ROUND_TO_NEAREST (1 << 2)
+#define CL_FP_ROUND_TO_ZERO (1 << 3)
+#define CL_FP_ROUND_TO_INF (1 << 4)
+#define CL_FP_FMA (1 << 5)
+#define CL_FP_SOFT_FLOAT (1 << 6)
 
 /* cl_device_mem_cache_type */
-#define CL_NONE                                     0x0
-#define CL_READ_ONLY_CACHE                          0x1
-#define CL_READ_WRITE_CACHE                         0x2
+#define CL_NONE 0x0
+#define CL_READ_ONLY_CACHE 0x1
+#define CL_READ_WRITE_CACHE 0x2
 
 /* cl_device_local_mem_type */
-#define CL_LOCAL                                    0x1
-#define CL_GLOBAL                                   0x2
+#define CL_LOCAL 0x1
+#define CL_GLOBAL 0x2
 
 /* cl_device_exec_capabilities - bitfield */
-#define CL_EXEC_KERNEL                              (1 << 0)
-#define CL_EXEC_NATIVE_KERNEL                       (1 << 1)
+#define CL_EXEC_KERNEL (1 << 0)
+#define CL_EXEC_NATIVE_KERNEL (1 << 1)
 
 /* cl_command_queue_properties - bitfield */
-#define CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE      (1 << 0)
-#define CL_QUEUE_PROFILING_ENABLE                   (1 << 1)
+#define CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE (1 << 0)
+#define CL_QUEUE_PROFILING_ENABLE (1 << 1)
 
 /* cl_context_info  */
-#define CL_CONTEXT_REFERENCE_COUNT                  0x1080
-#define CL_CONTEXT_DEVICES                          0x1081
-#define CL_CONTEXT_PROPERTIES                       0x1082
-#define CL_CONTEXT_NUM_DEVICES                      0x1083
+#define CL_CONTEXT_REFERENCE_COUNT 0x1080
+#define CL_CONTEXT_DEVICES 0x1081
+#define CL_CONTEXT_PROPERTIES 0x1082
+#define CL_CONTEXT_NUM_DEVICES 0x1083
 
 /* cl_context_info + cl_context_properties */
-#define CL_CONTEXT_PLATFORM                         0x1084
+#define CL_CONTEXT_PLATFORM 0x1084
 
 /* cl_command_queue_info */
-#define CL_QUEUE_CONTEXT                            0x1090
-#define CL_QUEUE_DEVICE                             0x1091
-#define CL_QUEUE_REFERENCE_COUNT                    0x1092
-#define CL_QUEUE_PROPERTIES                         0x1093
+#define CL_QUEUE_CONTEXT 0x1090
+#define CL_QUEUE_DEVICE 0x1091
+#define CL_QUEUE_REFERENCE_COUNT 0x1092
+#define CL_QUEUE_PROPERTIES 0x1093
 
 /* cl_mem_flags - bitfield */
-#define CL_MEM_READ_WRITE                           (1 << 0)
-#define CL_MEM_WRITE_ONLY                           (1 << 1)
-#define CL_MEM_READ_ONLY                            (1 << 2)
-#define CL_MEM_USE_HOST_PTR                         (1 << 3)
-#define CL_MEM_ALLOC_HOST_PTR                       (1 << 4)
-#define CL_MEM_COPY_HOST_PTR                        (1 << 5)
+#define CL_MEM_READ_WRITE (1 << 0)
+#define CL_MEM_WRITE_ONLY (1 << 1)
+#define CL_MEM_READ_ONLY (1 << 2)
+#define CL_MEM_USE_HOST_PTR (1 << 3)
+#define CL_MEM_ALLOC_HOST_PTR (1 << 4)
+#define CL_MEM_COPY_HOST_PTR (1 << 5)
 
 /* cl_channel_order */
-#define CL_R                                        0x10B0
-#define CL_A                                        0x10B1
-#define CL_RG                                       0x10B2
-#define CL_RA                                       0x10B3
-#define CL_RGB                                      0x10B4
-#define CL_RGBA                                     0x10B5
-#define CL_BGRA                                     0x10B6
-#define CL_ARGB                                     0x10B7
-#define CL_INTENSITY                                0x10B8
-#define CL_LUMINANCE                                0x10B9
-#define CL_Rx                                       0x10BA
-#define CL_RGx                                      0x10BB
-#define CL_RGBx                                     0x10BC
+#define CL_R 0x10B0
+#define CL_A 0x10B1
+#define CL_RG 0x10B2
+#define CL_RA 0x10B3
+#define CL_RGB 0x10B4
+#define CL_RGBA 0x10B5
+#define CL_BGRA 0x10B6
+#define CL_ARGB 0x10B7
+#define CL_INTENSITY 0x10B8
+#define CL_LUMINANCE 0x10B9
+#define CL_Rx 0x10BA
+#define CL_RGx 0x10BB
+#define CL_RGBx 0x10BC
 
 /* cl_channel_type */
-#define CL_SNORM_INT8                               0x10D0
-#define CL_SNORM_INT16                              0x10D1
-#define CL_UNORM_INT8                               0x10D2
-#define CL_UNORM_INT16                              0x10D3
-#define CL_UNORM_SHORT_565                          0x10D4
-#define CL_UNORM_SHORT_555                          0x10D5
-#define CL_UNORM_INT_101010                         0x10D6
-#define CL_SIGNED_INT8                              0x10D7
-#define CL_SIGNED_INT16                             0x10D8
-#define CL_SIGNED_INT32                             0x10D9
-#define CL_UNSIGNED_INT8                            0x10DA
-#define CL_UNSIGNED_INT16                           0x10DB
-#define CL_UNSIGNED_INT32                           0x10DC
-#define CL_HALF_FLOAT                               0x10DD
-#define CL_FLOAT                                    0x10DE
+#define CL_SNORM_INT8 0x10D0
+#define CL_SNORM_INT16 0x10D1
+#define CL_UNORM_INT8 0x10D2
+#define CL_UNORM_INT16 0x10D3
+#define CL_UNORM_SHORT_565 0x10D4
+#define CL_UNORM_SHORT_555 0x10D5
+#define CL_UNORM_INT_101010 0x10D6
+#define CL_SIGNED_INT8 0x10D7
+#define CL_SIGNED_INT16 0x10D8
+#define CL_SIGNED_INT32 0x10D9
+#define CL_UNSIGNED_INT8 0x10DA
+#define CL_UNSIGNED_INT16 0x10DB
+#define CL_UNSIGNED_INT32 0x10DC
+#define CL_HALF_FLOAT 0x10DD
+#define CL_FLOAT 0x10DE
 
 /* cl_mem_object_type */
-#define CL_MEM_OBJECT_BUFFER                        0x10F0
-#define CL_MEM_OBJECT_IMAGE2D                       0x10F1
-#define CL_MEM_OBJECT_IMAGE3D                       0x10F2
+#define CL_MEM_OBJECT_BUFFER 0x10F0
+#define CL_MEM_OBJECT_IMAGE2D 0x10F1
+#define CL_MEM_OBJECT_IMAGE3D 0x10F2
 
 /* cl_mem_info */
-#define CL_MEM_TYPE                                 0x1100
-#define CL_MEM_FLAGS                                0x1101
-#define CL_MEM_SIZE                                 0x1102
-#define CL_MEM_HOST_PTR                             0x1103
-#define CL_MEM_MAP_COUNT                            0x1104
-#define CL_MEM_REFERENCE_COUNT                      0x1105
-#define CL_MEM_CONTEXT                              0x1106
-#define CL_MEM_ASSOCIATED_MEMOBJECT                 0x1107
-#define CL_MEM_OFFSET                               0x1108
+#define CL_MEM_TYPE 0x1100
+#define CL_MEM_FLAGS 0x1101
+#define CL_MEM_SIZE 0x1102
+#define CL_MEM_HOST_PTR 0x1103
+#define CL_MEM_MAP_COUNT 0x1104
+#define CL_MEM_REFERENCE_COUNT 0x1105
+#define CL_MEM_CONTEXT 0x1106
+#define CL_MEM_ASSOCIATED_MEMOBJECT 0x1107
+#define CL_MEM_OFFSET 0x1108
 
 /* cl_image_info */
-#define CL_IMAGE_FORMAT                             0x1110
-#define CL_IMAGE_ELEMENT_SIZE                       0x1111
-#define CL_IMAGE_ROW_PITCH                          0x1112
-#define CL_IMAGE_SLICE_PITCH                        0x1113
-#define CL_IMAGE_WIDTH                              0x1114
-#define CL_IMAGE_HEIGHT                             0x1115
-#define CL_IMAGE_DEPTH                              0x1116
+#define CL_IMAGE_FORMAT 0x1110
+#define CL_IMAGE_ELEMENT_SIZE 0x1111
+#define CL_IMAGE_ROW_PITCH 0x1112
+#define CL_IMAGE_SLICE_PITCH 0x1113
+#define CL_IMAGE_WIDTH 0x1114
+#define CL_IMAGE_HEIGHT 0x1115
+#define CL_IMAGE_DEPTH 0x1116
 
 /* cl_addressing_mode */
-#define CL_ADDRESS_NONE                             0x1130
-#define CL_ADDRESS_CLAMP_TO_EDGE                    0x1131
-#define CL_ADDRESS_CLAMP                            0x1132
-#define CL_ADDRESS_REPEAT                           0x1133
-#define CL_ADDRESS_MIRRORED_REPEAT                  0x1134
+#define CL_ADDRESS_NONE 0x1130
+#define CL_ADDRESS_CLAMP_TO_EDGE 0x1131
+#define CL_ADDRESS_CLAMP 0x1132
+#define CL_ADDRESS_REPEAT 0x1133
+#define CL_ADDRESS_MIRRORED_REPEAT 0x1134
 
 /* cl_filter_mode */
-#define CL_FILTER_NEAREST                           0x1140
-#define CL_FILTER_LINEAR                            0x1141
+#define CL_FILTER_NEAREST 0x1140
+#define CL_FILTER_LINEAR 0x1141
 
 /* cl_sampler_info */
-#define CL_SAMPLER_REFERENCE_COUNT                  0x1150
-#define CL_SAMPLER_CONTEXT                          0x1151
-#define CL_SAMPLER_NORMALIZED_COORDS                0x1152
-#define CL_SAMPLER_ADDRESSING_MODE                  0x1153
-#define CL_SAMPLER_FILTER_MODE                      0x1154
+#define CL_SAMPLER_REFERENCE_COUNT 0x1150
+#define CL_SAMPLER_CONTEXT 0x1151
+#define CL_SAMPLER_NORMALIZED_COORDS 0x1152
+#define CL_SAMPLER_ADDRESSING_MODE 0x1153
+#define CL_SAMPLER_FILTER_MODE 0x1154
 
 /* cl_map_flags - bitfield */
-#define CL_MAP_READ                                 (1 << 0)
-#define CL_MAP_WRITE                                (1 << 1)
+#define CL_MAP_READ (1 << 0)
+#define CL_MAP_WRITE (1 << 1)
 
 /* cl_program_info */
-#define CL_PROGRAM_REFERENCE_COUNT                  0x1160
-#define CL_PROGRAM_CONTEXT                          0x1161
-#define CL_PROGRAM_NUM_DEVICES                      0x1162
-#define CL_PROGRAM_DEVICES                          0x1163
-#define CL_PROGRAM_SOURCE                           0x1164
-#define CL_PROGRAM_BINARY_SIZES                     0x1165
-#define CL_PROGRAM_BINARIES                         0x1166
+#define CL_PROGRAM_REFERENCE_COUNT 0x1160
+#define CL_PROGRAM_CONTEXT 0x1161
+#define CL_PROGRAM_NUM_DEVICES 0x1162
+#define CL_PROGRAM_DEVICES 0x1163
+#define CL_PROGRAM_SOURCE 0x1164
+#define CL_PROGRAM_BINARY_SIZES 0x1165
+#define CL_PROGRAM_BINARIES 0x1166
 
 /* cl_program_build_info */
-#define CL_PROGRAM_BUILD_STATUS                     0x1181
-#define CL_PROGRAM_BUILD_OPTIONS                    0x1182
-#define CL_PROGRAM_BUILD_LOG                        0x1183
+#define CL_PROGRAM_BUILD_STATUS 0x1181
+#define CL_PROGRAM_BUILD_OPTIONS 0x1182
+#define CL_PROGRAM_BUILD_LOG 0x1183
 
 /* cl_build_status */
-#define CL_BUILD_SUCCESS                            0
-#define CL_BUILD_NONE                               -1
-#define CL_BUILD_ERROR                              -2
-#define CL_BUILD_IN_PROGRESS                        -3
+#define CL_BUILD_SUCCESS 0
+#define CL_BUILD_NONE -1
+#define CL_BUILD_ERROR -2
+#define CL_BUILD_IN_PROGRESS -3
 
 /* cl_kernel_info */
-#define CL_KERNEL_FUNCTION_NAME                     0x1190
-#define CL_KERNEL_NUM_ARGS                          0x1191
-#define CL_KERNEL_REFERENCE_COUNT                   0x1192
-#define CL_KERNEL_CONTEXT                           0x1193
-#define CL_KERNEL_PROGRAM                           0x1194
+#define CL_KERNEL_FUNCTION_NAME 0x1190
+#define CL_KERNEL_NUM_ARGS 0x1191
+#define CL_KERNEL_REFERENCE_COUNT 0x1192
+#define CL_KERNEL_CONTEXT 0x1193
+#define CL_KERNEL_PROGRAM 0x1194
 
 /* cl_kernel_work_group_info */
-#define CL_KERNEL_WORK_GROUP_SIZE                   0x11B0
-#define CL_KERNEL_COMPILE_WORK_GROUP_SIZE           0x11B1
-#define CL_KERNEL_LOCAL_MEM_SIZE                    0x11B2
+#define CL_KERNEL_WORK_GROUP_SIZE 0x11B0
+#define CL_KERNEL_COMPILE_WORK_GROUP_SIZE 0x11B1
+#define CL_KERNEL_LOCAL_MEM_SIZE 0x11B2
 #define CL_KERNEL_PREFERRED_WORK_GROUP_SIZE_MULTIPLE 0x11B3
-#define CL_KERNEL_PRIVATE_MEM_SIZE                  0x11B4
+#define CL_KERNEL_PRIVATE_MEM_SIZE 0x11B4
 
 /* cl_event_info  */
-#define CL_EVENT_COMMAND_QUEUE                      0x11D0
-#define CL_EVENT_COMMAND_TYPE                       0x11D1
-#define CL_EVENT_REFERENCE_COUNT                    0x11D2
-#define CL_EVENT_COMMAND_EXECUTION_STATUS           0x11D3
-#define CL_EVENT_CONTEXT                            0x11D4
+#define CL_EVENT_COMMAND_QUEUE 0x11D0
+#define CL_EVENT_COMMAND_TYPE 0x11D1
+#define CL_EVENT_REFERENCE_COUNT 0x11D2
+#define CL_EVENT_COMMAND_EXECUTION_STATUS 0x11D3
+#define CL_EVENT_CONTEXT 0x11D4
 
 /* cl_command_type */
-#define CL_COMMAND_NDRANGE_KERNEL                   0x11F0
-#define CL_COMMAND_TASK                             0x11F1
-#define CL_COMMAND_NATIVE_KERNEL                    0x11F2
-#define CL_COMMAND_READ_BUFFER                      0x11F3
-#define CL_COMMAND_WRITE_BUFFER                     0x11F4
-#define CL_COMMAND_COPY_BUFFER                      0x11F5
-#define CL_COMMAND_READ_IMAGE                       0x11F6
-#define CL_COMMAND_WRITE_IMAGE                      0x11F7
-#define CL_COMMAND_COPY_IMAGE                       0x11F8
-#define CL_COMMAND_COPY_IMAGE_TO_BUFFER             0x11F9
-#define CL_COMMAND_COPY_BUFFER_TO_IMAGE             0x11FA
-#define CL_COMMAND_MAP_BUFFER                       0x11FB
-#define CL_COMMAND_MAP_IMAGE                        0x11FC
-#define CL_COMMAND_UNMAP_MEM_OBJECT                 0x11FD
-#define CL_COMMAND_MARKER                           0x11FE
-#define CL_COMMAND_ACQUIRE_GL_OBJECTS               0x11FF
-#define CL_COMMAND_RELEASE_GL_OBJECTS               0x1200
-#define CL_COMMAND_READ_BUFFER_RECT                 0x1201
-#define CL_COMMAND_WRITE_BUFFER_RECT                0x1202
-#define CL_COMMAND_COPY_BUFFER_RECT                 0x1203
-#define CL_COMMAND_USER                             0x1204
+#define CL_COMMAND_NDRANGE_KERNEL 0x11F0
+#define CL_COMMAND_TASK 0x11F1
+#define CL_COMMAND_NATIVE_KERNEL 0x11F2
+#define CL_COMMAND_READ_BUFFER 0x11F3
+#define CL_COMMAND_WRITE_BUFFER 0x11F4
+#define CL_COMMAND_COPY_BUFFER 0x11F5
+#define CL_COMMAND_READ_IMAGE 0x11F6
+#define CL_COMMAND_WRITE_IMAGE 0x11F7
+#define CL_COMMAND_COPY_IMAGE 0x11F8
+#define CL_COMMAND_COPY_IMAGE_TO_BUFFER 0x11F9
+#define CL_COMMAND_COPY_BUFFER_TO_IMAGE 0x11FA
+#define CL_COMMAND_MAP_BUFFER 0x11FB
+#define CL_COMMAND_MAP_IMAGE 0x11FC
+#define CL_COMMAND_UNMAP_MEM_OBJECT 0x11FD
+#define CL_COMMAND_MARKER 0x11FE
+#define CL_COMMAND_ACQUIRE_GL_OBJECTS 0x11FF
+#define CL_COMMAND_RELEASE_GL_OBJECTS 0x1200
+#define CL_COMMAND_READ_BUFFER_RECT 0x1201
+#define CL_COMMAND_WRITE_BUFFER_RECT 0x1202
+#define CL_COMMAND_COPY_BUFFER_RECT 0x1203
+#define CL_COMMAND_USER 0x1204
 
 /* command execution status */
-#define CL_COMPLETE                                 0x0
-#define CL_RUNNING                                  0x1
-#define CL_SUBMITTED                                0x2
-#define CL_QUEUED                                   0x3
-  
+#define CL_COMPLETE 0x0
+#define CL_RUNNING 0x1
+#define CL_SUBMITTED 0x2
+#define CL_QUEUED 0x3
+
 /* cl_buffer_create_type  */
-#define CL_BUFFER_CREATE_TYPE_REGION                0x1220
+#define CL_BUFFER_CREATE_TYPE_REGION 0x1220
 
 /* cl_profiling_info  */
-#define CL_PROFILING_COMMAND_QUEUED                 0x1280
-#define CL_PROFILING_COMMAND_SUBMIT                 0x1281
-#define CL_PROFILING_COMMAND_START                  0x1282
-#define CL_PROFILING_COMMAND_END                    0x1283
-
-/********************************************************************************************************/
-
-/********************************************************************************************************/
-
-/*  Function signature typedef's */
-
-/* Platform API */
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLGETPLATFORMIDS)(cl_uint          /* num_entries */,
-                 cl_platform_id * /* platforms */,
-                 cl_uint *        /* num_platforms */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL * 
-PFNCLGETPLATFORMINFO)(cl_platform_id   /* platform */, 
-                  cl_platform_info /* param_name */,
-                  size_t           /* param_value_size */, 
-                  void *           /* param_value */,
-                  size_t *         /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-/* Device APIs */
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLGETDEVICEIDS)(cl_platform_id   /* platform */,
-               cl_device_type   /* device_type */, 
-               cl_uint          /* num_entries */, 
-               cl_device_id *   /* devices */, 
-               cl_uint *        /* num_devices */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLGETDEVICEINFO)(cl_device_id    /* device */,
-                cl_device_info  /* param_name */, 
-                size_t          /* param_value_size */, 
-                void *          /* param_value */,
-                size_t *        /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-// Context APIs  
-typedef CL_API_ENTRY cl_context (CL_API_CALL *
-PFNCLCREATECONTEXT)(const cl_context_properties * /* properties */,
-                cl_uint                       /* num_devices */,
-                const cl_device_id *          /* devices */,
-                void (CL_CALLBACK * /* pfn_notify */)(const char *, const void *, size_t, void *),
-                void *                        /* user_data */,
-                cl_int *                      /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_context (CL_API_CALL *
-PFNCLCREATECONTEXTFROMTYPE)(const cl_context_properties * /* properties */,
-                        cl_device_type                /* device_type */,
-                        void (CL_CALLBACK *     /* pfn_notify*/ )(const char *, const void *, size_t, void *),
-                        void *                        /* user_data */,
-                        cl_int *                      /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLRETAINCONTEXT)(cl_context /* context */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLRELEASECONTEXT)(cl_context /* context */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLGETCONTEXTINFO)(cl_context         /* context */, 
-                 cl_context_info    /* param_name */, 
-                 size_t             /* param_value_size */, 
-                 void *             /* param_value */, 
-                 size_t *           /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-/* Command Queue APIs */
-typedef CL_API_ENTRY cl_command_queue (CL_API_CALL *
-PFNCLCREATECOMMANDQUEUE)(cl_context                     /* context */, 
-                     cl_device_id                   /* device */, 
-                     cl_command_queue_properties    /* properties */,
-                     cl_int *                       /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLRETAINCOMMANDQUEUE)(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLRELEASECOMMANDQUEUE)(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLGETCOMMANDQUEUEINFO)(cl_command_queue      /* command_queue */,
-                      cl_command_queue_info /* param_name */,
-                      size_t                /* param_value_size */,
-                      void *                /* param_value */,
-                      size_t *              /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLSETCOMMANDQUEUEPROPERTY)(cl_command_queue              /* command_queue */,
-                          cl_command_queue_properties   /* properties */, 
-                          cl_bool                        /* enable */,
-                          cl_command_queue_properties * /* old_properties */) CL_API_SUFFIX__VERSION_1_0;
-
-/* Memory Object APIs */
-typedef CL_API_ENTRY cl_mem (CL_API_CALL *
-PFNCLCREATEBUFFER)(cl_context   /* context */,
-               cl_mem_flags /* flags */,
-               size_t       /* size */,
-               void *       /* host_ptr */,
-               cl_int *     /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_mem (CL_API_CALL *
-PFNCLCREATESUBBUFFER)(cl_mem   			/* buffer */,
-               cl_mem_flags 			/* flags */,
-               cl_buffer_create_type    /* buffer_create_type */,
-               const void *       		/* buffer_create_info */,
-               cl_int *     			/* errcode_ret */) CL_API_SUFFIX__VERSION_1_1;
-
-typedef CL_API_ENTRY cl_mem (CL_API_CALL *
-PFNCLCREATEIMAGE2D)(cl_context              /* context */,
-                cl_mem_flags            /* flags */,
-                const cl_image_format * /* image_format */,
-                size_t                  /* image_width */,
-                size_t                  /* image_height */,
-                size_t                  /* image_row_pitch */, 
-                void *                  /* host_ptr */,
-                cl_int *                /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_mem (CL_API_CALL *
-PFNCLCREATEIMAGE3D)(cl_context              /* context */,
-                cl_mem_flags            /* flags */,
-                const cl_image_format * /* image_format */,
-                size_t                  /* image_width */, 
-                size_t                  /* image_height */,
-                size_t                  /* image_depth */, 
-                size_t                  /* image_row_pitch */, 
-                size_t                  /* image_slice_pitch */, 
-                void *                  /* host_ptr */,
-                cl_int *                /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLRETAINMEMOBJECT)(cl_mem /* memobj */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLRELEASEMEMOBJECT)(cl_mem /* memobj */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLGETSUPPORTEDIMAGEFORMATS)(cl_context           /* context */,
-                           cl_mem_flags         /* flags */,
-                           cl_mem_object_type   /* image_type */,
-                           cl_uint              /* num_entries */,
-                           cl_image_format *    /* image_formats */,
-                           cl_uint *            /* num_image_formats */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLGETMEMOBJECTINFO)(cl_mem           /* memobj */,
-                   cl_mem_info      /* param_name */, 
-                   size_t           /* param_value_size */,
-                   void *           /* param_value */,
-                   size_t *         /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLGETIMAGEINFO)(cl_mem           /* image */,
-               cl_image_info    /* param_name */, 
-               size_t           /* param_value_size */,
-               void *           /* param_value */,
-               size_t *         /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLSETMEMOBJECTDESTRUCTORCALLBACK)(  cl_mem /* memobj */, 
-                                    void (CL_CALLBACK * /*pfn_notify*/)( cl_mem /* memobj */, void* /*user_data*/), 
-                                    void * /*user_data */ )             CL_API_SUFFIX__VERSION_1_1;  
-
-/* Sampler APIs  */
-typedef CL_API_ENTRY cl_sampler (CL_API_CALL *
-PFNCLCREATESAMPLER)(cl_context          /* context */,
-                cl_bool             /* normalized_coords */, 
-                cl_addressing_mode  /* addressing_mode */, 
-                cl_filter_mode      /* filter_mode */,
-                cl_int *            /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLRETAINSAMPLER)(cl_sampler /* sampler */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLRELEASESAMPLER)(cl_sampler /* sampler */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLGETSAMPLERINFO)(cl_sampler         /* sampler */,
-                 cl_sampler_info    /* param_name */,
-                 size_t             /* param_value_size */,
-                 void *             /* param_value */,
-                 size_t *           /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
- 
-/* Program Object APIs  */
-typedef CL_API_ENTRY cl_program (CL_API_CALL *
-PFNCLCREATEPROGRAMWITHSOURCE)(cl_context        /* context */,
-                          cl_uint           /* count */,
-                          const char **     /* strings */,
-                          const size_t *    /* lengths */,
-                          cl_int *          /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_program (CL_API_CALL *
-PFNCLCREATEPROGRAMWITHBINARY)(cl_context                     /* context */,
-                          cl_uint                        /* num_devices */,
-                          const cl_device_id *           /* device_list */,
-                          const size_t *                 /* lengths */,
-                          const unsigned char **         /* binaries */,
-                          cl_int *                       /* binary_status */,
-                          cl_int *                       /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLRETAINPROGRAM)(cl_program /* program */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLRELEASEPROGRAM)(cl_program /* program */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLBUILDPROGRAM)(cl_program           /* program */,
-               cl_uint              /* num_devices */,
-               const cl_device_id * /* device_list */,
-               const char *         /* options */, 
-               void (CL_CALLBACK *  /* pfn_notify */)(cl_program /* program */, void * /* user_data */),
-               void *               /* user_data */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLUNLOADCOMPILER)(void) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLGETPROGRAMINFO)(cl_program         /* program */,
-                 cl_program_info    /* param_name */,
-                 size_t             /* param_value_size */,
-                 void *             /* param_value */,
-                 size_t *           /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLGETPROGRAMBUILDINFO)(cl_program            /* program */,
-                      cl_device_id          /* device */,
-                      cl_program_build_info /* param_name */,
-                      size_t                /* param_value_size */,
-                      void *                /* param_value */,
-                      size_t *              /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-/* Kernel Object APIs */
-typedef CL_API_ENTRY cl_kernel (CL_API_CALL *
-PFNCLCREATEKERNEL)(cl_program      /* program */,
-               const char *    /* kernel_name */,
-               cl_int *        /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLCREATEKERNELSINPROGRAM)(cl_program     /* program */,
-                         cl_uint        /* num_kernels */,
-                         cl_kernel *    /* kernels */,
-                         cl_uint *      /* num_kernels_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLRETAINKERNEL)(cl_kernel    /* kernel */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLRELEASEKERNEL)(cl_kernel   /* kernel */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLSETKERNELARG)(cl_kernel    /* kernel */,
-               cl_uint      /* arg_index */,
-               size_t       /* arg_size */,
-               const void * /* arg_value */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLGETKERNELINFO)(cl_kernel       /* kernel */,
-                cl_kernel_info  /* param_name */,
-                size_t          /* param_value_size */,
-                void *          /* param_value */,
-                size_t *        /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLGETKERNELWORKGROUPINFO)(cl_kernel                  /* kernel */,
-                         cl_device_id               /* device */,
-                         cl_kernel_work_group_info  /* param_name */,
-                         size_t                     /* param_value_size */,
-                         void *                     /* param_value */,
-                         size_t *                   /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-// Event Object APIs
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLWAITFOREVENTS)(cl_uint             /* num_events */,
-                const cl_event *    /* event_list */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLGETEVENTINFO)(cl_event         /* event */,
-               cl_event_info    /* param_name */,
-               size_t           /* param_value_size */,
-               void *           /* param_value */,
-               size_t *         /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_event (CL_API_CALL *
-PFNCLCREATEUSEREVENT)(cl_context    /* context */,
-                  	  cl_int *      /* errcode_ret */) CL_API_SUFFIX__VERSION_1_1;               
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLRETAINEVENT)(cl_event /* event */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLRELEASEEVENT)(cl_event /* event */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLSETUSEREVENTSTATUS)(cl_event   /* event */,
-                     cl_int     /* execution_status */) CL_API_SUFFIX__VERSION_1_1;
-                     
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLSETEVENTCALLBACK)( cl_event    /* event */,
-                    cl_int      /* command_exec_callback_type */,
-                    void (CL_CALLBACK * /* pfn_notify */)(cl_event, cl_int, void *),
-                    void *      /* user_data */) CL_API_SUFFIX__VERSION_1_1;
-
-/* Profiling APIs  */
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLGETEVENTPROFILINGINFO)(cl_event            /* event */,
-                        cl_profiling_info   /* param_name */,
-                        size_t              /* param_value_size */,
-                        void *              /* param_value */,
-                        size_t *            /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-// Flush and Finish APIs
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLFLUSH)(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLFINISH)(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0;
-
-/* Enqueued Commands APIs */
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLENQUEUEREADBUFFER)(cl_command_queue    /* command_queue */,
-                    cl_mem              /* buffer */,
-                    cl_bool             /* blocking_read */,
-                    size_t              /* offset */,
-                    size_t              /* cb */, 
-                    void *              /* ptr */,
-                    cl_uint             /* num_events_in_wait_list */,
-                    const cl_event *    /* event_wait_list */,
-                    cl_event *          /* event */) CL_API_SUFFIX__VERSION_1_0;
-                            
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLENQUEUEREADBUFFERRECT)(cl_command_queue    /* command_queue */,
-                        cl_mem              /* buffer */,
-                        cl_bool             /* blocking_read */,
-                        const size_t *      /* buffer_origin */,
-                        const size_t *      /* host_origin */, 
-                        const size_t *      /* region */,
-                        size_t              /* buffer_row_pitch */,
-                        size_t              /* buffer_slice_pitch */,
-                        size_t              /* host_row_pitch */,
-                        size_t              /* host_slice_pitch */,                        
-                        void *              /* ptr */,
-                        cl_uint             /* num_events_in_wait_list */,
-                        const cl_event *    /* event_wait_list */,
-                        cl_event *          /* event */) CL_API_SUFFIX__VERSION_1_1;
-                            
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLENQUEUEWRITEBUFFER)(cl_command_queue   /* command_queue */, 
-                     cl_mem             /* buffer */, 
-                     cl_bool            /* blocking_write */, 
-                     size_t             /* offset */, 
-                     size_t             /* cb */, 
-                     const void *       /* ptr */, 
-                     cl_uint            /* num_events_in_wait_list */, 
-                     const cl_event *   /* event_wait_list */, 
-                     cl_event *         /* event */) CL_API_SUFFIX__VERSION_1_0;
-                            
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLENQUEUEWRITEBUFFERRECT)(cl_command_queue    /* command_queue */,
-                         cl_mem              /* buffer */,
-                         cl_bool             /* blocking_write */,
-                         const size_t *      /* buffer_origin */,
-                         const size_t *      /* host_origin */, 
-                         const size_t *      /* region */,
-                         size_t              /* buffer_row_pitch */,
-                         size_t              /* buffer_slice_pitch */,
-                         size_t              /* host_row_pitch */,
-                         size_t              /* host_slice_pitch */,                        
-                         const void *        /* ptr */,
-                         cl_uint             /* num_events_in_wait_list */,
-                         const cl_event *    /* event_wait_list */,
-                         cl_event *          /* event */) CL_API_SUFFIX__VERSION_1_1;
-                            
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLENQUEUECOPYBUFFER)(cl_command_queue    /* command_queue */, 
-                    cl_mem              /* src_buffer */,
-                    cl_mem              /* dst_buffer */, 
-                    size_t              /* src_offset */,
-                    size_t              /* dst_offset */,
-                    size_t              /* cb */, 
-                    cl_uint             /* num_events_in_wait_list */,
-                    const cl_event *    /* event_wait_list */,
-                    cl_event *          /* event */) CL_API_SUFFIX__VERSION_1_0;
-                            
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLENQUEUECOPYBUFFERRECT)(cl_command_queue    /* command_queue */, 
-                        cl_mem              /* src_buffer */,
-                        cl_mem              /* dst_buffer */, 
-                        const size_t *      /* src_origin */,
-                        const size_t *      /* dst_origin */,
-                        const size_t *      /* region */, 
-                        size_t              /* src_row_pitch */,
-                        size_t              /* src_slice_pitch */,
-                        size_t              /* dst_row_pitch */,
-                        size_t              /* dst_slice_pitch */,
-                        cl_uint             /* num_events_in_wait_list */,
-                        const cl_event *    /* event_wait_list */,
-                        cl_event *          /* event */) CL_API_SUFFIX__VERSION_1_1;
-                            
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLENQUEUEREADIMAGE)(cl_command_queue     /* command_queue */,
-                   cl_mem               /* image */,
-                   cl_bool              /* blocking_read */, 
-                   const size_t *       /* origin[3] */,
-                   const size_t *       /* region[3] */,
-                   size_t               /* row_pitch */,
-                   size_t               /* slice_pitch */, 
-                   void *               /* ptr */,
-                   cl_uint              /* num_events_in_wait_list */,
-                   const cl_event *     /* event_wait_list */,
-                   cl_event *           /* event */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLENQUEUEWRITEIMAGE)(cl_command_queue    /* command_queue */,
-                    cl_mem              /* image */,
-                    cl_bool             /* blocking_write */, 
-                    const size_t *      /* origin[3] */,
-                    const size_t *      /* region[3] */,
-                    size_t              /* input_row_pitch */,
-                    size_t              /* input_slice_pitch */, 
-                    const void *        /* ptr */,
-                    cl_uint             /* num_events_in_wait_list */,
-                    const cl_event *    /* event_wait_list */,
-                    cl_event *          /* event */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLENQUEUECOPYIMAGE)(cl_command_queue     /* command_queue */,
-                   cl_mem               /* src_image */,
-                   cl_mem               /* dst_image */, 
-                   const size_t *       /* src_origin[3] */,
-                   const size_t *       /* dst_origin[3] */,
-                   const size_t *       /* region[3] */, 
-                   cl_uint              /* num_events_in_wait_list */,
-                   const cl_event *     /* event_wait_list */,
-                   cl_event *           /* event */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLENQUEUECOPYIMAGETOBUFFER)(cl_command_queue /* command_queue */,
-                           cl_mem           /* src_image */,
-                           cl_mem           /* dst_buffer */, 
-                           const size_t *   /* src_origin[3] */,
-                           const size_t *   /* region[3] */, 
-                           size_t           /* dst_offset */,
-                           cl_uint          /* num_events_in_wait_list */,
-                           const cl_event * /* event_wait_list */,
-                           cl_event *       /* event */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLENQUEUECOPYBUFFERTOIMAGE)(cl_command_queue /* command_queue */,
-                           cl_mem           /* src_buffer */,
-                           cl_mem           /* dst_image */, 
-                           size_t           /* src_offset */,
-                           const size_t *   /* dst_origin[3] */,
-                           const size_t *   /* region[3] */, 
-                           cl_uint          /* num_events_in_wait_list */,
-                           const cl_event * /* event_wait_list */,
-                           cl_event *       /* event */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY void * (CL_API_CALL *
-PFNCLENQUEUEMAPBUFFER)(cl_command_queue /* command_queue */,
-                   cl_mem           /* buffer */,
-                   cl_bool          /* blocking_map */, 
-                   cl_map_flags     /* map_flags */,
-                   size_t           /* offset */,
-                   size_t           /* cb */,
-                   cl_uint          /* num_events_in_wait_list */,
-                   const cl_event * /* event_wait_list */,
-                   cl_event *       /* event */,
-                   cl_int *         /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY void * (CL_API_CALL *
-PFNCLENQUEUEMAPIMAGE)(cl_command_queue  /* command_queue */,
-                  cl_mem            /* image */, 
-                  cl_bool           /* blocking_map */, 
-                  cl_map_flags      /* map_flags */, 
-                  const size_t *    /* origin[3] */,
-                  const size_t *    /* region[3] */,
-                  size_t *          /* image_row_pitch */,
-                  size_t *          /* image_slice_pitch */,
-                  cl_uint           /* num_events_in_wait_list */,
-                  const cl_event *  /* event_wait_list */,
-                  cl_event *        /* event */,
-                  cl_int *          /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLENQUEUEUNMAPMEMOBJECT)(cl_command_queue /* command_queue */,
-                        cl_mem           /* memobj */,
-                        void *           /* mapped_ptr */,
-                        cl_uint          /* num_events_in_wait_list */,
-                        const cl_event *  /* event_wait_list */,
-                        cl_event *        /* event */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLENQUEUENDRANGEKERNEL)(cl_command_queue /* command_queue */,
-                       cl_kernel        /* kernel */,
-                       cl_uint          /* work_dim */,
-                       const size_t *   /* global_work_offset */,
-                       const size_t *   /* global_work_size */,
-                       const size_t *   /* local_work_size */,
-                       cl_uint          /* num_events_in_wait_list */,
-                       const cl_event * /* event_wait_list */,
-                       cl_event *       /* event */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLENQUEUETASK)(cl_command_queue  /* command_queue */,
-              cl_kernel         /* kernel */,
-              cl_uint           /* num_events_in_wait_list */,
-              const cl_event *  /* event_wait_list */,
-              cl_event *        /* event */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLENQUEUENATIVEKERNEL)(cl_command_queue  /* command_queue */,
-                      void (*user_func)(void *), 
-                      void *            /* args */,
-                      size_t            /* cb_args */, 
-                      cl_uint           /* num_mem_objects */,
-                      const cl_mem *    /* mem_list */,
-                      const void **     /* args_mem_loc */,
-                      cl_uint           /* num_events_in_wait_list */,
-                      const cl_event *  /* event_wait_list */,
-                      cl_event *        /* event */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLENQUEUEMARKER)(cl_command_queue    /* command_queue */,
-                cl_event *          /* event */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLENQUEUEWAITFOREVENTS)(cl_command_queue /* command_queue */,
-                       cl_uint          /* num_events */,
-                       const cl_event * /* event_list */) CL_API_SUFFIX__VERSION_1_0;
-
-typedef CL_API_ENTRY cl_int (CL_API_CALL *
-PFNCLENQUEUEBARRIER)(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0;
-
-// Extension function access
-//
-// Returns the extension function address for the given function name,
-// or NULL if a valid function can not be found.  The client must
-// check to make sure the address is not NULL, before using or 
-// calling the returned function address.
-//
-typedef CL_API_ENTRY void * (CL_API_CALL * PFNCLGETEXTENSIONFUNCTIONADDRESS)(const char * /* func_name */) CL_API_SUFFIX__VERSION_1_0;
-
+#define CL_PROFILING_COMMAND_QUEUED 0x1280
+#define CL_PROFILING_COMMAND_SUBMIT 0x1281
+#define CL_PROFILING_COMMAND_START 0x1282
+#define CL_PROFILING_COMMAND_END 0x1283
+
+	/********************************************************************************************************/
+
+	/********************************************************************************************************/
+
+	/*  Function signature typedef's */
+
+	/* Platform API */
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLGETPLATFORMIDS)(cl_uint /* num_entries */,
+														 cl_platform_id * /* platforms */,
+														 cl_uint * /* num_platforms */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLGETPLATFORMINFO)(cl_platform_id /* platform */,
+														  cl_platform_info /* param_name */,
+														  size_t /* param_value_size */,
+														  void * /* param_value */,
+														  size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	/* Device APIs */
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLGETDEVICEIDS)(cl_platform_id /* platform */,
+													   cl_device_type /* device_type */,
+													   cl_uint /* num_entries */,
+													   cl_device_id * /* devices */,
+													   cl_uint * /* num_devices */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLGETDEVICEINFO)(cl_device_id /* device */,
+														cl_device_info /* param_name */,
+														size_t /* param_value_size */,
+														void * /* param_value */,
+														size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	// Context APIs
+	typedef CL_API_ENTRY cl_context(CL_API_CALL *
+										PFNCLCREATECONTEXT)(const cl_context_properties * /* properties */,
+															cl_uint /* num_devices */,
+															const cl_device_id * /* devices */,
+															void(CL_CALLBACK * /* pfn_notify */)(const char *, const void *, size_t, void *),
+															void * /* user_data */,
+															cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_context(CL_API_CALL *
+										PFNCLCREATECONTEXTFROMTYPE)(const cl_context_properties * /* properties */,
+																	cl_device_type /* device_type */,
+																	void(CL_CALLBACK * /* pfn_notify*/)(const char *, const void *, size_t, void *),
+																	void * /* user_data */,
+																	cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLRETAINCONTEXT)(cl_context /* context */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLRELEASECONTEXT)(cl_context /* context */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLGETCONTEXTINFO)(cl_context /* context */,
+														 cl_context_info /* param_name */,
+														 size_t /* param_value_size */,
+														 void * /* param_value */,
+														 size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	/* Command Queue APIs */
+	typedef CL_API_ENTRY cl_command_queue(CL_API_CALL *
+											  PFNCLCREATECOMMANDQUEUE)(cl_context /* context */,
+																	   cl_device_id /* device */,
+																	   cl_command_queue_properties /* properties */,
+																	   cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLRETAINCOMMANDQUEUE)(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLRELEASECOMMANDQUEUE)(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLGETCOMMANDQUEUEINFO)(cl_command_queue /* command_queue */,
+															  cl_command_queue_info /* param_name */,
+															  size_t /* param_value_size */,
+															  void * /* param_value */,
+															  size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLSETCOMMANDQUEUEPROPERTY)(cl_command_queue /* command_queue */,
+																  cl_command_queue_properties /* properties */,
+																  cl_bool /* enable */,
+																  cl_command_queue_properties * /* old_properties */) CL_API_SUFFIX__VERSION_1_0;
+
+	/* Memory Object APIs */
+	typedef CL_API_ENTRY cl_mem(CL_API_CALL *
+									PFNCLCREATEBUFFER)(cl_context /* context */,
+													   cl_mem_flags /* flags */,
+													   size_t /* size */,
+													   void * /* host_ptr */,
+													   cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_mem(CL_API_CALL *
+									PFNCLCREATESUBBUFFER)(cl_mem /* buffer */,
+														  cl_mem_flags /* flags */,
+														  cl_buffer_create_type /* buffer_create_type */,
+														  const void * /* buffer_create_info */,
+														  cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_1;
+
+	typedef CL_API_ENTRY cl_mem(CL_API_CALL *
+									PFNCLCREATEIMAGE2D)(cl_context /* context */,
+														cl_mem_flags /* flags */,
+														const cl_image_format * /* image_format */,
+														size_t /* image_width */,
+														size_t /* image_height */,
+														size_t /* image_row_pitch */,
+														void * /* host_ptr */,
+														cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_mem(CL_API_CALL *
+									PFNCLCREATEIMAGE3D)(cl_context /* context */,
+														cl_mem_flags /* flags */,
+														const cl_image_format * /* image_format */,
+														size_t /* image_width */,
+														size_t /* image_height */,
+														size_t /* image_depth */,
+														size_t /* image_row_pitch */,
+														size_t /* image_slice_pitch */,
+														void * /* host_ptr */,
+														cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLRETAINMEMOBJECT)(cl_mem /* memobj */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLRELEASEMEMOBJECT)(cl_mem /* memobj */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLGETSUPPORTEDIMAGEFORMATS)(cl_context /* context */,
+																   cl_mem_flags /* flags */,
+																   cl_mem_object_type /* image_type */,
+																   cl_uint /* num_entries */,
+																   cl_image_format * /* image_formats */,
+																   cl_uint * /* num_image_formats */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLGETMEMOBJECTINFO)(cl_mem /* memobj */,
+														   cl_mem_info /* param_name */,
+														   size_t /* param_value_size */,
+														   void * /* param_value */,
+														   size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLGETIMAGEINFO)(cl_mem /* image */,
+													   cl_image_info /* param_name */,
+													   size_t /* param_value_size */,
+													   void * /* param_value */,
+													   size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLSETMEMOBJECTDESTRUCTORCALLBACK)(cl_mem /* memobj */,
+																		 void(CL_CALLBACK * /*pfn_notify*/)(cl_mem /* memobj */, void * /*user_data*/),
+																		 void * /*user_data */) CL_API_SUFFIX__VERSION_1_1;
+
+	/* Sampler APIs  */
+	typedef CL_API_ENTRY cl_sampler(CL_API_CALL *
+										PFNCLCREATESAMPLER)(cl_context /* context */,
+															cl_bool /* normalized_coords */,
+															cl_addressing_mode /* addressing_mode */,
+															cl_filter_mode /* filter_mode */,
+															cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLRETAINSAMPLER)(cl_sampler /* sampler */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLRELEASESAMPLER)(cl_sampler /* sampler */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLGETSAMPLERINFO)(cl_sampler /* sampler */,
+														 cl_sampler_info /* param_name */,
+														 size_t /* param_value_size */,
+														 void * /* param_value */,
+														 size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	/* Program Object APIs  */
+	typedef CL_API_ENTRY cl_program(CL_API_CALL *
+										PFNCLCREATEPROGRAMWITHSOURCE)(cl_context /* context */,
+																	  cl_uint /* count */,
+																	  const char ** /* strings */,
+																	  const size_t * /* lengths */,
+																	  cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_program(CL_API_CALL *
+										PFNCLCREATEPROGRAMWITHBINARY)(cl_context /* context */,
+																	  cl_uint /* num_devices */,
+																	  const cl_device_id * /* device_list */,
+																	  const size_t * /* lengths */,
+																	  const unsigned char ** /* binaries */,
+																	  cl_int * /* binary_status */,
+																	  cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLRETAINPROGRAM)(cl_program /* program */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLRELEASEPROGRAM)(cl_program /* program */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLBUILDPROGRAM)(cl_program /* program */,
+													   cl_uint /* num_devices */,
+													   const cl_device_id * /* device_list */,
+													   const char * /* options */,
+													   void(CL_CALLBACK * /* pfn_notify */)(cl_program /* program */, void * /* user_data */),
+													   void * /* user_data */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLUNLOADCOMPILER)(void) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLGETPROGRAMINFO)(cl_program /* program */,
+														 cl_program_info /* param_name */,
+														 size_t /* param_value_size */,
+														 void * /* param_value */,
+														 size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLGETPROGRAMBUILDINFO)(cl_program /* program */,
+															  cl_device_id /* device */,
+															  cl_program_build_info /* param_name */,
+															  size_t /* param_value_size */,
+															  void * /* param_value */,
+															  size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	/* Kernel Object APIs */
+	typedef CL_API_ENTRY cl_kernel(CL_API_CALL *
+									   PFNCLCREATEKERNEL)(cl_program /* program */,
+														  const char * /* kernel_name */,
+														  cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLCREATEKERNELSINPROGRAM)(cl_program /* program */,
+																 cl_uint /* num_kernels */,
+																 cl_kernel * /* kernels */,
+																 cl_uint * /* num_kernels_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLRETAINKERNEL)(cl_kernel /* kernel */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLRELEASEKERNEL)(cl_kernel /* kernel */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLSETKERNELARG)(cl_kernel /* kernel */,
+													   cl_uint /* arg_index */,
+													   size_t /* arg_size */,
+													   const void * /* arg_value */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLGETKERNELINFO)(cl_kernel /* kernel */,
+														cl_kernel_info /* param_name */,
+														size_t /* param_value_size */,
+														void * /* param_value */,
+														size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLGETKERNELWORKGROUPINFO)(cl_kernel /* kernel */,
+																 cl_device_id /* device */,
+																 cl_kernel_work_group_info /* param_name */,
+																 size_t /* param_value_size */,
+																 void * /* param_value */,
+																 size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	// Event Object APIs
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLWAITFOREVENTS)(cl_uint /* num_events */,
+														const cl_event * /* event_list */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLGETEVENTINFO)(cl_event /* event */,
+													   cl_event_info /* param_name */,
+													   size_t /* param_value_size */,
+													   void * /* param_value */,
+													   size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_event(CL_API_CALL *
+									  PFNCLCREATEUSEREVENT)(cl_context /* context */,
+															cl_int * /* errcode_ret */) CL_API_SUFFIX__VERSION_1_1;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLRETAINEVENT)(cl_event /* event */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLRELEASEEVENT)(cl_event /* event */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLSETUSEREVENTSTATUS)(cl_event /* event */,
+															 cl_int /* execution_status */) CL_API_SUFFIX__VERSION_1_1;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLSETEVENTCALLBACK)(cl_event /* event */,
+														   cl_int /* command_exec_callback_type */,
+														   void(CL_CALLBACK * /* pfn_notify */)(cl_event, cl_int, void *),
+														   void * /* user_data */) CL_API_SUFFIX__VERSION_1_1;
+
+	/* Profiling APIs  */
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLGETEVENTPROFILINGINFO)(cl_event /* event */,
+																cl_profiling_info /* param_name */,
+																size_t /* param_value_size */,
+																void * /* param_value */,
+																size_t * /* param_value_size_ret */) CL_API_SUFFIX__VERSION_1_0;
+
+	// Flush and Finish APIs
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLFLUSH)(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLFINISH)(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0;
+
+	/* Enqueued Commands APIs */
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLENQUEUEREADBUFFER)(cl_command_queue /* command_queue */,
+															cl_mem /* buffer */,
+															cl_bool /* blocking_read */,
+															size_t /* offset */,
+															size_t /* cb */,
+															void * /* ptr */,
+															cl_uint /* num_events_in_wait_list */,
+															const cl_event * /* event_wait_list */,
+															cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLENQUEUEREADBUFFERRECT)(cl_command_queue /* command_queue */,
+																cl_mem /* buffer */,
+																cl_bool /* blocking_read */,
+																const size_t * /* buffer_origin */,
+																const size_t * /* host_origin */,
+																const size_t * /* region */,
+																size_t /* buffer_row_pitch */,
+																size_t /* buffer_slice_pitch */,
+																size_t /* host_row_pitch */,
+																size_t /* host_slice_pitch */,
+																void * /* ptr */,
+																cl_uint /* num_events_in_wait_list */,
+																const cl_event * /* event_wait_list */,
+																cl_event * /* event */) CL_API_SUFFIX__VERSION_1_1;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLENQUEUEWRITEBUFFER)(cl_command_queue /* command_queue */,
+															 cl_mem /* buffer */,
+															 cl_bool /* blocking_write */,
+															 size_t /* offset */,
+															 size_t /* cb */,
+															 const void * /* ptr */,
+															 cl_uint /* num_events_in_wait_list */,
+															 const cl_event * /* event_wait_list */,
+															 cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLENQUEUEWRITEBUFFERRECT)(cl_command_queue /* command_queue */,
+																 cl_mem /* buffer */,
+																 cl_bool /* blocking_write */,
+																 const size_t * /* buffer_origin */,
+																 const size_t * /* host_origin */,
+																 const size_t * /* region */,
+																 size_t /* buffer_row_pitch */,
+																 size_t /* buffer_slice_pitch */,
+																 size_t /* host_row_pitch */,
+																 size_t /* host_slice_pitch */,
+																 const void * /* ptr */,
+																 cl_uint /* num_events_in_wait_list */,
+																 const cl_event * /* event_wait_list */,
+																 cl_event * /* event */) CL_API_SUFFIX__VERSION_1_1;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLENQUEUECOPYBUFFER)(cl_command_queue /* command_queue */,
+															cl_mem /* src_buffer */,
+															cl_mem /* dst_buffer */,
+															size_t /* src_offset */,
+															size_t /* dst_offset */,
+															size_t /* cb */,
+															cl_uint /* num_events_in_wait_list */,
+															const cl_event * /* event_wait_list */,
+															cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLENQUEUECOPYBUFFERRECT)(cl_command_queue /* command_queue */,
+																cl_mem /* src_buffer */,
+																cl_mem /* dst_buffer */,
+																const size_t * /* src_origin */,
+																const size_t * /* dst_origin */,
+																const size_t * /* region */,
+																size_t /* src_row_pitch */,
+																size_t /* src_slice_pitch */,
+																size_t /* dst_row_pitch */,
+																size_t /* dst_slice_pitch */,
+																cl_uint /* num_events_in_wait_list */,
+																const cl_event * /* event_wait_list */,
+																cl_event * /* event */) CL_API_SUFFIX__VERSION_1_1;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLENQUEUEREADIMAGE)(cl_command_queue /* command_queue */,
+														   cl_mem /* image */,
+														   cl_bool /* blocking_read */,
+														   const size_t * /* origin[3] */,
+														   const size_t * /* region[3] */,
+														   size_t /* row_pitch */,
+														   size_t /* slice_pitch */,
+														   void * /* ptr */,
+														   cl_uint /* num_events_in_wait_list */,
+														   const cl_event * /* event_wait_list */,
+														   cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLENQUEUEWRITEIMAGE)(cl_command_queue /* command_queue */,
+															cl_mem /* image */,
+															cl_bool /* blocking_write */,
+															const size_t * /* origin[3] */,
+															const size_t * /* region[3] */,
+															size_t /* input_row_pitch */,
+															size_t /* input_slice_pitch */,
+															const void * /* ptr */,
+															cl_uint /* num_events_in_wait_list */,
+															const cl_event * /* event_wait_list */,
+															cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLENQUEUECOPYIMAGE)(cl_command_queue /* command_queue */,
+														   cl_mem /* src_image */,
+														   cl_mem /* dst_image */,
+														   const size_t * /* src_origin[3] */,
+														   const size_t * /* dst_origin[3] */,
+														   const size_t * /* region[3] */,
+														   cl_uint /* num_events_in_wait_list */,
+														   const cl_event * /* event_wait_list */,
+														   cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLENQUEUECOPYIMAGETOBUFFER)(cl_command_queue /* command_queue */,
+																   cl_mem /* src_image */,
+																   cl_mem /* dst_buffer */,
+																   const size_t * /* src_origin[3] */,
+																   const size_t * /* region[3] */,
+																   size_t /* dst_offset */,
+																   cl_uint /* num_events_in_wait_list */,
+																   const cl_event * /* event_wait_list */,
+																   cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLENQUEUECOPYBUFFERTOIMAGE)(cl_command_queue /* command_queue */,
+																   cl_mem /* src_buffer */,
+																   cl_mem /* dst_image */,
+																   size_t /* src_offset */,
+																   const size_t * /* dst_origin[3] */,
+																   const size_t * /* region[3] */,
+																   cl_uint /* num_events_in_wait_list */,
+																   const cl_event * /* event_wait_list */,
+																   cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY void *(CL_API_CALL *
+									PFNCLENQUEUEMAPBUFFER)(cl_command_queue /* command_queue */,
+														   cl_mem /* buffer */,
+														   cl_bool /* blocking_map */,
+														   cl_map_flags /* map_flags */,
+														   size_t /* offset */,
+														   size_t /* cb */,
+														   cl_uint /* num_events_in_wait_list */,
+														   const cl_event * /* event_wait_list */,
+														   cl_event * /* event */,
+														   cl_int * /* errcode_ret */)CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY void *(CL_API_CALL *
+									PFNCLENQUEUEMAPIMAGE)(cl_command_queue /* command_queue */,
+														  cl_mem /* image */,
+														  cl_bool /* blocking_map */,
+														  cl_map_flags /* map_flags */,
+														  const size_t * /* origin[3] */,
+														  const size_t * /* region[3] */,
+														  size_t * /* image_row_pitch */,
+														  size_t * /* image_slice_pitch */,
+														  cl_uint /* num_events_in_wait_list */,
+														  const cl_event * /* event_wait_list */,
+														  cl_event * /* event */,
+														  cl_int * /* errcode_ret */)CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLENQUEUEUNMAPMEMOBJECT)(cl_command_queue /* command_queue */,
+																cl_mem /* memobj */,
+																void * /* mapped_ptr */,
+																cl_uint /* num_events_in_wait_list */,
+																const cl_event * /* event_wait_list */,
+																cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLENQUEUENDRANGEKERNEL)(cl_command_queue /* command_queue */,
+															   cl_kernel /* kernel */,
+															   cl_uint /* work_dim */,
+															   const size_t * /* global_work_offset */,
+															   const size_t * /* global_work_size */,
+															   const size_t * /* local_work_size */,
+															   cl_uint /* num_events_in_wait_list */,
+															   const cl_event * /* event_wait_list */,
+															   cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLENQUEUETASK)(cl_command_queue /* command_queue */,
+													  cl_kernel /* kernel */,
+													  cl_uint /* num_events_in_wait_list */,
+													  const cl_event * /* event_wait_list */,
+													  cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLENQUEUENATIVEKERNEL)(cl_command_queue /* command_queue */,
+															  void (*user_func)(void *),
+															  void * /* args */,
+															  size_t /* cb_args */,
+															  cl_uint /* num_mem_objects */,
+															  const cl_mem * /* mem_list */,
+															  const void ** /* args_mem_loc */,
+															  cl_uint /* num_events_in_wait_list */,
+															  const cl_event * /* event_wait_list */,
+															  cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLENQUEUEMARKER)(cl_command_queue /* command_queue */,
+														cl_event * /* event */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLENQUEUEWAITFOREVENTS)(cl_command_queue /* command_queue */,
+															   cl_uint /* num_events */,
+															   const cl_event * /* event_list */) CL_API_SUFFIX__VERSION_1_0;
+
+	typedef CL_API_ENTRY cl_int(CL_API_CALL *
+									PFNCLENQUEUEBARRIER)(cl_command_queue /* command_queue */) CL_API_SUFFIX__VERSION_1_0;
+
+	// Extension function access
+	//
+	// Returns the extension function address for the given function name,
+	// or NULL if a valid function can not be found.  The client must
+	// check to make sure the address is not NULL, before using or
+	// calling the returned function address.
+	//
+	typedef CL_API_ENTRY void *(CL_API_CALL *PFNCLGETEXTENSIONFUNCTIONADDRESS)(const char * /* func_name */)CL_API_SUFFIX__VERSION_1_0;
 
 #define CLEW_STATIC
 
 #ifdef CLEW_STATIC
-#  define CLEWAPI extern
+#define CLEWAPI extern
 #else
-#  ifdef CLEW_BUILD
-#    define CLEWAPI extern __declspec(dllexport)
-#  else
-#    define CLEWAPI extern __declspec(dllimport)
-#  endif
+#ifdef CLEW_BUILD
+#define CLEWAPI extern __declspec(dllexport)
+#else
+#define CLEWAPI extern __declspec(dllimport)
+#endif
 #endif
 
 #if defined(_WIN32)
@@ -2208,99 +2522,97 @@ typedef CL_API_ENTRY void * (CL_API_CALL * PFNCLGETEXTENSIONFUNCTIONADDRESS)(con
 
 #define CLEW_GET_FUN(x) x
 
-
-//  Variables holding function entry points
-CLEW_FUN_EXPORT     PFNCLGETPLATFORMIDS                 __clewGetPlatformIDs                ;
-CLEW_FUN_EXPORT     PFNCLGETPLATFORMINFO                __clewGetPlatformInfo               ;
-CLEW_FUN_EXPORT     PFNCLGETDEVICEIDS                   __clewGetDeviceIDs                  ;
-CLEW_FUN_EXPORT     PFNCLGETDEVICEINFO                  __clewGetDeviceInfo                 ;
-CLEW_FUN_EXPORT     PFNCLCREATECONTEXT                  __clewCreateContext                 ;
-CLEW_FUN_EXPORT     PFNCLCREATECONTEXTFROMTYPE          __clewCreateContextFromType         ;
-CLEW_FUN_EXPORT     PFNCLRETAINCONTEXT                  __clewRetainContext                 ;
-CLEW_FUN_EXPORT     PFNCLRELEASECONTEXT                 __clewReleaseContext                ;
-CLEW_FUN_EXPORT     PFNCLGETCONTEXTINFO                 __clewGetContextInfo                ;
-CLEW_FUN_EXPORT     PFNCLCREATECOMMANDQUEUE             __clewCreateCommandQueue            ;
-CLEW_FUN_EXPORT     PFNCLRETAINCOMMANDQUEUE             __clewRetainCommandQueue            ;
-CLEW_FUN_EXPORT     PFNCLRELEASECOMMANDQUEUE            __clewReleaseCommandQueue           ;
-CLEW_FUN_EXPORT     PFNCLGETCOMMANDQUEUEINFO            __clewGetCommandQueueInfo           ;
+	//  Variables holding function entry points
+	CLEW_FUN_EXPORT PFNCLGETPLATFORMIDS __clewGetPlatformIDs;
+	CLEW_FUN_EXPORT PFNCLGETPLATFORMINFO __clewGetPlatformInfo;
+	CLEW_FUN_EXPORT PFNCLGETDEVICEIDS __clewGetDeviceIDs;
+	CLEW_FUN_EXPORT PFNCLGETDEVICEINFO __clewGetDeviceInfo;
+	CLEW_FUN_EXPORT PFNCLCREATECONTEXT __clewCreateContext;
+	CLEW_FUN_EXPORT PFNCLCREATECONTEXTFROMTYPE __clewCreateContextFromType;
+	CLEW_FUN_EXPORT PFNCLRETAINCONTEXT __clewRetainContext;
+	CLEW_FUN_EXPORT PFNCLRELEASECONTEXT __clewReleaseContext;
+	CLEW_FUN_EXPORT PFNCLGETCONTEXTINFO __clewGetContextInfo;
+	CLEW_FUN_EXPORT PFNCLCREATECOMMANDQUEUE __clewCreateCommandQueue;
+	CLEW_FUN_EXPORT PFNCLRETAINCOMMANDQUEUE __clewRetainCommandQueue;
+	CLEW_FUN_EXPORT PFNCLRELEASECOMMANDQUEUE __clewReleaseCommandQueue;
+	CLEW_FUN_EXPORT PFNCLGETCOMMANDQUEUEINFO __clewGetCommandQueueInfo;
 #ifdef CL_USE_DEPRECATED_OPENCL_1_0_APIS
-CLEW_FUN_EXPORT     PFNCLSETCOMMANDQUEUEPROPERTY        __clewSetCommandQueueProperty       ;
+	CLEW_FUN_EXPORT PFNCLSETCOMMANDQUEUEPROPERTY __clewSetCommandQueueProperty;
 #endif
-CLEW_FUN_EXPORT     PFNCLCREATEBUFFER                   __clewCreateBuffer                  ;
-CLEW_FUN_EXPORT     PFNCLCREATESUBBUFFER                __clewCreateSubBuffer               ;
-CLEW_FUN_EXPORT     PFNCLCREATEIMAGE2D                  __clewCreateImage2D                 ;
-CLEW_FUN_EXPORT     PFNCLCREATEIMAGE3D                  __clewCreateImage3D                 ;
-CLEW_FUN_EXPORT     PFNCLRETAINMEMOBJECT                __clewRetainMemObject               ;
-CLEW_FUN_EXPORT     PFNCLRELEASEMEMOBJECT               __clewReleaseMemObject              ;
-CLEW_FUN_EXPORT     PFNCLGETSUPPORTEDIMAGEFORMATS       __clewGetSupportedImageFormats      ;
-CLEW_FUN_EXPORT     PFNCLGETMEMOBJECTINFO               __clewGetMemObjectInfo              ;
-CLEW_FUN_EXPORT     PFNCLGETIMAGEINFO                   __clewGetImageInfo                  ;
-CLEW_FUN_EXPORT     PFNCLSETMEMOBJECTDESTRUCTORCALLBACK __clewSetMemObjectDestructorCallback;
-CLEW_FUN_EXPORT     PFNCLCREATESAMPLER                  __clewCreateSampler                 ;
-CLEW_FUN_EXPORT     PFNCLRETAINSAMPLER                  __clewRetainSampler                 ;
-CLEW_FUN_EXPORT     PFNCLRELEASESAMPLER                 __clewReleaseSampler                ;
-CLEW_FUN_EXPORT     PFNCLGETSAMPLERINFO                 __clewGetSamplerInfo                ;
-CLEW_FUN_EXPORT     PFNCLCREATEPROGRAMWITHSOURCE        __clewCreateProgramWithSource       ;
-CLEW_FUN_EXPORT     PFNCLCREATEPROGRAMWITHBINARY        __clewCreateProgramWithBinary       ;
-CLEW_FUN_EXPORT     PFNCLRETAINPROGRAM                  __clewRetainProgram                 ;
-CLEW_FUN_EXPORT     PFNCLRELEASEPROGRAM                 __clewReleaseProgram                ;
-CLEW_FUN_EXPORT     PFNCLBUILDPROGRAM                   __clewBuildProgram                  ;
-CLEW_FUN_EXPORT     PFNCLUNLOADCOMPILER                 __clewUnloadCompiler                ;
-CLEW_FUN_EXPORT     PFNCLGETPROGRAMINFO                 __clewGetProgramInfo                ;
-CLEW_FUN_EXPORT     PFNCLGETPROGRAMBUILDINFO            __clewGetProgramBuildInfo           ;
-CLEW_FUN_EXPORT     PFNCLCREATEKERNEL                   __clewCreateKernel                  ;
-CLEW_FUN_EXPORT     PFNCLCREATEKERNELSINPROGRAM         __clewCreateKernelsInProgram        ;
-CLEW_FUN_EXPORT     PFNCLRETAINKERNEL                   __clewRetainKernel                  ;
-CLEW_FUN_EXPORT     PFNCLRELEASEKERNEL                  __clewReleaseKernel                 ;
-CLEW_FUN_EXPORT     PFNCLSETKERNELARG                   __clewSetKernelArg                  ;
-CLEW_FUN_EXPORT     PFNCLGETKERNELINFO                  __clewGetKernelInfo                 ;
-CLEW_FUN_EXPORT     PFNCLGETKERNELWORKGROUPINFO         __clewGetKernelWorkGroupInfo        ;
-CLEW_FUN_EXPORT     PFNCLWAITFOREVENTS                  __clewWaitForEvents                 ;
-CLEW_FUN_EXPORT     PFNCLGETEVENTINFO                   __clewGetEventInfo                  ;
-CLEW_FUN_EXPORT     PFNCLCREATEUSEREVENT                __clewCreateUserEvent               ;
-CLEW_FUN_EXPORT     PFNCLRETAINEVENT                    __clewRetainEvent                   ;
-CLEW_FUN_EXPORT     PFNCLRELEASEEVENT                   __clewReleaseEvent                  ;
-CLEW_FUN_EXPORT     PFNCLSETUSEREVENTSTATUS             __clewSetUserEventStatus            ;
-CLEW_FUN_EXPORT     PFNCLSETEVENTCALLBACK               __clewSetEventCallback              ;
-CLEW_FUN_EXPORT     PFNCLGETEVENTPROFILINGINFO          __clewGetEventProfilingInfo         ;
-CLEW_FUN_EXPORT     PFNCLFLUSH                          __clewFlush                         ;
-CLEW_FUN_EXPORT     PFNCLFINISH                         __clewFinish                        ;
-CLEW_FUN_EXPORT     PFNCLENQUEUEREADBUFFER              __clewEnqueueReadBuffer             ;
-CLEW_FUN_EXPORT     PFNCLENQUEUEREADBUFFERRECT          __clewEnqueueReadBufferRect         ;
-CLEW_FUN_EXPORT     PFNCLENQUEUEWRITEBUFFER             __clewEnqueueWriteBuffer            ;
-CLEW_FUN_EXPORT     PFNCLENQUEUEWRITEBUFFERRECT         __clewEnqueueWriteBufferRect        ;
-CLEW_FUN_EXPORT     PFNCLENQUEUECOPYBUFFER              __clewEnqueueCopyBuffer             ;
-CLEW_FUN_EXPORT     PFNCLENQUEUECOPYBUFFERRECT          __clewEnqueueCopyBufferRect         ;
-CLEW_FUN_EXPORT     PFNCLENQUEUEREADIMAGE               __clewEnqueueReadImage              ;
-CLEW_FUN_EXPORT     PFNCLENQUEUEWRITEIMAGE              __clewEnqueueWriteImage             ;
-CLEW_FUN_EXPORT     PFNCLENQUEUECOPYIMAGE               __clewEnqueueCopyImage              ;
-CLEW_FUN_EXPORT     PFNCLENQUEUECOPYIMAGETOBUFFER       __clewEnqueueCopyImageToBuffer      ;
-CLEW_FUN_EXPORT     PFNCLENQUEUECOPYBUFFERTOIMAGE       __clewEnqueueCopyBufferToImage      ;
-CLEW_FUN_EXPORT     PFNCLENQUEUEMAPBUFFER               __clewEnqueueMapBuffer              ;
-CLEW_FUN_EXPORT     PFNCLENQUEUEMAPIMAGE                __clewEnqueueMapImage               ;
-CLEW_FUN_EXPORT     PFNCLENQUEUEUNMAPMEMOBJECT          __clewEnqueueUnmapMemObject         ;
-CLEW_FUN_EXPORT     PFNCLENQUEUENDRANGEKERNEL           __clewEnqueueNDRangeKernel          ;
-CLEW_FUN_EXPORT     PFNCLENQUEUETASK                    __clewEnqueueTask                   ;
-CLEW_FUN_EXPORT     PFNCLENQUEUENATIVEKERNEL            __clewEnqueueNativeKernel           ;
-CLEW_FUN_EXPORT     PFNCLENQUEUEMARKER                  __clewEnqueueMarker                 ;
-CLEW_FUN_EXPORT     PFNCLENQUEUEWAITFOREVENTS           __clewEnqueueWaitForEvents          ;
-CLEW_FUN_EXPORT     PFNCLENQUEUEBARRIER                 __clewEnqueueBarrier                ;
-CLEW_FUN_EXPORT     PFNCLGETEXTENSIONFUNCTIONADDRESS    __clewGetExtensionFunctionAddress   ;
+	CLEW_FUN_EXPORT PFNCLCREATEBUFFER __clewCreateBuffer;
+	CLEW_FUN_EXPORT PFNCLCREATESUBBUFFER __clewCreateSubBuffer;
+	CLEW_FUN_EXPORT PFNCLCREATEIMAGE2D __clewCreateImage2D;
+	CLEW_FUN_EXPORT PFNCLCREATEIMAGE3D __clewCreateImage3D;
+	CLEW_FUN_EXPORT PFNCLRETAINMEMOBJECT __clewRetainMemObject;
+	CLEW_FUN_EXPORT PFNCLRELEASEMEMOBJECT __clewReleaseMemObject;
+	CLEW_FUN_EXPORT PFNCLGETSUPPORTEDIMAGEFORMATS __clewGetSupportedImageFormats;
+	CLEW_FUN_EXPORT PFNCLGETMEMOBJECTINFO __clewGetMemObjectInfo;
+	CLEW_FUN_EXPORT PFNCLGETIMAGEINFO __clewGetImageInfo;
+	CLEW_FUN_EXPORT PFNCLSETMEMOBJECTDESTRUCTORCALLBACK __clewSetMemObjectDestructorCallback;
+	CLEW_FUN_EXPORT PFNCLCREATESAMPLER __clewCreateSampler;
+	CLEW_FUN_EXPORT PFNCLRETAINSAMPLER __clewRetainSampler;
+	CLEW_FUN_EXPORT PFNCLRELEASESAMPLER __clewReleaseSampler;
+	CLEW_FUN_EXPORT PFNCLGETSAMPLERINFO __clewGetSamplerInfo;
+	CLEW_FUN_EXPORT PFNCLCREATEPROGRAMWITHSOURCE __clewCreateProgramWithSource;
+	CLEW_FUN_EXPORT PFNCLCREATEPROGRAMWITHBINARY __clewCreateProgramWithBinary;
+	CLEW_FUN_EXPORT PFNCLRETAINPROGRAM __clewRetainProgram;
+	CLEW_FUN_EXPORT PFNCLRELEASEPROGRAM __clewReleaseProgram;
+	CLEW_FUN_EXPORT PFNCLBUILDPROGRAM __clewBuildProgram;
+	CLEW_FUN_EXPORT PFNCLUNLOADCOMPILER __clewUnloadCompiler;
+	CLEW_FUN_EXPORT PFNCLGETPROGRAMINFO __clewGetProgramInfo;
+	CLEW_FUN_EXPORT PFNCLGETPROGRAMBUILDINFO __clewGetProgramBuildInfo;
+	CLEW_FUN_EXPORT PFNCLCREATEKERNEL __clewCreateKernel;
+	CLEW_FUN_EXPORT PFNCLCREATEKERNELSINPROGRAM __clewCreateKernelsInProgram;
+	CLEW_FUN_EXPORT PFNCLRETAINKERNEL __clewRetainKernel;
+	CLEW_FUN_EXPORT PFNCLRELEASEKERNEL __clewReleaseKernel;
+	CLEW_FUN_EXPORT PFNCLSETKERNELARG __clewSetKernelArg;
+	CLEW_FUN_EXPORT PFNCLGETKERNELINFO __clewGetKernelInfo;
+	CLEW_FUN_EXPORT PFNCLGETKERNELWORKGROUPINFO __clewGetKernelWorkGroupInfo;
+	CLEW_FUN_EXPORT PFNCLWAITFOREVENTS __clewWaitForEvents;
+	CLEW_FUN_EXPORT PFNCLGETEVENTINFO __clewGetEventInfo;
+	CLEW_FUN_EXPORT PFNCLCREATEUSEREVENT __clewCreateUserEvent;
+	CLEW_FUN_EXPORT PFNCLRETAINEVENT __clewRetainEvent;
+	CLEW_FUN_EXPORT PFNCLRELEASEEVENT __clewReleaseEvent;
+	CLEW_FUN_EXPORT PFNCLSETUSEREVENTSTATUS __clewSetUserEventStatus;
+	CLEW_FUN_EXPORT PFNCLSETEVENTCALLBACK __clewSetEventCallback;
+	CLEW_FUN_EXPORT PFNCLGETEVENTPROFILINGINFO __clewGetEventProfilingInfo;
+	CLEW_FUN_EXPORT PFNCLFLUSH __clewFlush;
+	CLEW_FUN_EXPORT PFNCLFINISH __clewFinish;
+	CLEW_FUN_EXPORT PFNCLENQUEUEREADBUFFER __clewEnqueueReadBuffer;
+	CLEW_FUN_EXPORT PFNCLENQUEUEREADBUFFERRECT __clewEnqueueReadBufferRect;
+	CLEW_FUN_EXPORT PFNCLENQUEUEWRITEBUFFER __clewEnqueueWriteBuffer;
+	CLEW_FUN_EXPORT PFNCLENQUEUEWRITEBUFFERRECT __clewEnqueueWriteBufferRect;
+	CLEW_FUN_EXPORT PFNCLENQUEUECOPYBUFFER __clewEnqueueCopyBuffer;
+	CLEW_FUN_EXPORT PFNCLENQUEUECOPYBUFFERRECT __clewEnqueueCopyBufferRect;
+	CLEW_FUN_EXPORT PFNCLENQUEUEREADIMAGE __clewEnqueueReadImage;
+	CLEW_FUN_EXPORT PFNCLENQUEUEWRITEIMAGE __clewEnqueueWriteImage;
+	CLEW_FUN_EXPORT PFNCLENQUEUECOPYIMAGE __clewEnqueueCopyImage;
+	CLEW_FUN_EXPORT PFNCLENQUEUECOPYIMAGETOBUFFER __clewEnqueueCopyImageToBuffer;
+	CLEW_FUN_EXPORT PFNCLENQUEUECOPYBUFFERTOIMAGE __clewEnqueueCopyBufferToImage;
+	CLEW_FUN_EXPORT PFNCLENQUEUEMAPBUFFER __clewEnqueueMapBuffer;
+	CLEW_FUN_EXPORT PFNCLENQUEUEMAPIMAGE __clewEnqueueMapImage;
+	CLEW_FUN_EXPORT PFNCLENQUEUEUNMAPMEMOBJECT __clewEnqueueUnmapMemObject;
+	CLEW_FUN_EXPORT PFNCLENQUEUENDRANGEKERNEL __clewEnqueueNDRangeKernel;
+	CLEW_FUN_EXPORT PFNCLENQUEUETASK __clewEnqueueTask;
+	CLEW_FUN_EXPORT PFNCLENQUEUENATIVEKERNEL __clewEnqueueNativeKernel;
+	CLEW_FUN_EXPORT PFNCLENQUEUEMARKER __clewEnqueueMarker;
+	CLEW_FUN_EXPORT PFNCLENQUEUEWAITFOREVENTS __clewEnqueueWaitForEvents;
+	CLEW_FUN_EXPORT PFNCLENQUEUEBARRIER __clewEnqueueBarrier;
+	CLEW_FUN_EXPORT PFNCLGETEXTENSIONFUNCTIONADDRESS __clewGetExtensionFunctionAddress;
 
-
-#define	clGetPlatformIDs                CLEW_GET_FUN(__clewGetPlatformIDs                )
-#define	clGetPlatformInfo               CLEW_GET_FUN(__clewGetPlatformInfo               )
-#define	clGetDeviceIDs                  CLEW_GET_FUN(__clewGetDeviceIDs                  )
-#define	clGetDeviceInfo                 CLEW_GET_FUN(__clewGetDeviceInfo                 )
-#define	clCreateContext                 CLEW_GET_FUN(__clewCreateContext                 )
-#define	clCreateContextFromType         CLEW_GET_FUN(__clewCreateContextFromType         )
-#define	clRetainContext                 CLEW_GET_FUN(__clewRetainContext                 )
-#define	clReleaseContext                CLEW_GET_FUN(__clewReleaseContext                )
-#define	clGetContextInfo                CLEW_GET_FUN(__clewGetContextInfo                )
-#define	clCreateCommandQueue            CLEW_GET_FUN(__clewCreateCommandQueue            )
-#define	clRetainCommandQueue            CLEW_GET_FUN(__clewRetainCommandQueue            )
-#define	clReleaseCommandQueue           CLEW_GET_FUN(__clewReleaseCommandQueue           )
-#define	clGetCommandQueueInfo           CLEW_GET_FUN(__clewGetCommandQueueInfo           )
+#define clGetPlatformIDs CLEW_GET_FUN(__clewGetPlatformIDs)
+#define clGetPlatformInfo CLEW_GET_FUN(__clewGetPlatformInfo)
+#define clGetDeviceIDs CLEW_GET_FUN(__clewGetDeviceIDs)
+#define clGetDeviceInfo CLEW_GET_FUN(__clewGetDeviceInfo)
+#define clCreateContext CLEW_GET_FUN(__clewCreateContext)
+#define clCreateContextFromType CLEW_GET_FUN(__clewCreateContextFromType)
+#define clRetainContext CLEW_GET_FUN(__clewRetainContext)
+#define clReleaseContext CLEW_GET_FUN(__clewReleaseContext)
+#define clGetContextInfo CLEW_GET_FUN(__clewGetContextInfo)
+#define clCreateCommandQueue CLEW_GET_FUN(__clewCreateCommandQueue)
+#define clRetainCommandQueue CLEW_GET_FUN(__clewRetainCommandQueue)
+#define clReleaseCommandQueue CLEW_GET_FUN(__clewReleaseCommandQueue)
+#define clGetCommandQueueInfo CLEW_GET_FUN(__clewGetCommandQueueInfo)
 #ifdef CL_USE_DEPRECATED_OPENCL_1_0_APIS
 #warning CL_USE_DEPRECATED_OPENCL_1_0_APIS is defined. These APIs are unsupported and untested in OpenCL 1.1!
 /* 
@@ -2313,82 +2625,81 @@ CLEW_FUN_EXPORT     PFNCLGETEXTENSIONFUNCTIONADDRESS    __clewGetExtensionFuncti
  *  Software developers previously relying on this API are instructed to set the command queue 
  *  properties when creating the queue, instead. 
  */
-#define	clSetCommandQueueProperty       CLEW_GET_FUN(__clewSetCommandQueueProperty       )
+#define clSetCommandQueueProperty CLEW_GET_FUN(__clewSetCommandQueueProperty)
 #endif /* CL_USE_DEPRECATED_OPENCL_1_0_APIS */
-#define	clCreateBuffer                  CLEW_GET_FUN(__clewCreateBuffer                  )
-#define	clCreateSubBuffer               CLEW_GET_FUN(__clewCreateSubBuffer               )
-#define	clCreateImage2D                 CLEW_GET_FUN(__clewCreateImage2D                 )
-#define	clCreateImage3D                 CLEW_GET_FUN(__clewCreateImage3D                 )
-#define	clRetainMemObject               CLEW_GET_FUN(__clewRetainMemObject               )
-#define	clReleaseMemObject              CLEW_GET_FUN(__clewReleaseMemObject              )
-#define	clGetSupportedImageFormats      CLEW_GET_FUN(__clewGetSupportedImageFormats      )
-#define	clGetMemObjectInfo              CLEW_GET_FUN(__clewGetMemObjectInfo              )
-#define	clGetImageInfo                  CLEW_GET_FUN(__clewGetImageInfo                  )
-#define	clSetMemObjectDestructorCallback CLEW_GET_FUN(__clewSetMemObjectDestructorCallback)
-#define	clCreateSampler                 CLEW_GET_FUN(__clewCreateSampler                 )
-#define	clRetainSampler                 CLEW_GET_FUN(__clewRetainSampler                 )
-#define	clReleaseSampler                CLEW_GET_FUN(__clewReleaseSampler                )
-#define	clGetSamplerInfo                CLEW_GET_FUN(__clewGetSamplerInfo                )
-#define	clCreateProgramWithSource       CLEW_GET_FUN(__clewCreateProgramWithSource       )
-#define	clCreateProgramWithBinary       CLEW_GET_FUN(__clewCreateProgramWithBinary       )
-#define	clRetainProgram                 CLEW_GET_FUN(__clewRetainProgram                 )
-#define	clReleaseProgram                CLEW_GET_FUN(__clewReleaseProgram                )
-#define	clBuildProgram                  CLEW_GET_FUN(__clewBuildProgram                  )
-#define	clUnloadCompiler                CLEW_GET_FUN(__clewUnloadCompiler                )
-#define	clGetProgramInfo                CLEW_GET_FUN(__clewGetProgramInfo                )
-#define	clGetProgramBuildInfo           CLEW_GET_FUN(__clewGetProgramBuildInfo           )
-#define	clCreateKernel                  CLEW_GET_FUN(__clewCreateKernel                  )
-#define	clCreateKernelsInProgram        CLEW_GET_FUN(__clewCreateKernelsInProgram        )
-#define	clRetainKernel                  CLEW_GET_FUN(__clewRetainKernel                  )
-#define	clReleaseKernel                 CLEW_GET_FUN(__clewReleaseKernel                 )
-#define	clSetKernelArg                  CLEW_GET_FUN(__clewSetKernelArg                  )
-#define	clGetKernelInfo                 CLEW_GET_FUN(__clewGetKernelInfo                 )
-#define	clGetKernelWorkGroupInfo        CLEW_GET_FUN(__clewGetKernelWorkGroupInfo        )
-#define	clWaitForEvents                 CLEW_GET_FUN(__clewWaitForEvents                 )
-#define	clGetEventInfo                  CLEW_GET_FUN(__clewGetEventInfo                  )
-#define	clCreateUserEvent               CLEW_GET_FUN(__clewCreateUserEvent               )
-#define	clRetainEvent                   CLEW_GET_FUN(__clewRetainEvent                   )
-#define	clReleaseEvent                  CLEW_GET_FUN(__clewReleaseEvent                  )
-#define	clSetUserEventStatus            CLEW_GET_FUN(__clewSetUserEventStatus            )
-#define	clSetEventCallback              CLEW_GET_FUN(__clewSetEventCallback              )
-#define	clGetEventProfilingInfo         CLEW_GET_FUN(__clewGetEventProfilingInfo         )
-#define	clFlush                         CLEW_GET_FUN(__clewFlush                         )
-#define	clFinish                        CLEW_GET_FUN(__clewFinish                        )
-#define	clEnqueueReadBuffer             CLEW_GET_FUN(__clewEnqueueReadBuffer             )
-#define	clEnqueueReadBufferRect         CLEW_GET_FUN(__clewEnqueueReadBufferRect         )
-#define	clEnqueueWriteBuffer            CLEW_GET_FUN(__clewEnqueueWriteBuffer            )
-#define	clEnqueueWriteBufferRect        CLEW_GET_FUN(__clewEnqueueWriteBufferRect        )
-#define	clEnqueueCopyBuffer             CLEW_GET_FUN(__clewEnqueueCopyBuffer             )
-#define	clEnqueueCopyBufferRect         CLEW_GET_FUN(__clewEnqueueCopyBufferRect         )
-#define	clEnqueueReadImage              CLEW_GET_FUN(__clewEnqueueReadImage              )
-#define	clEnqueueWriteImage             CLEW_GET_FUN(__clewEnqueueWriteImage             )
-#define	clEnqueueCopyImage              CLEW_GET_FUN(__clewEnqueueCopyImage              )
-#define	clEnqueueCopyImageToBuffer      CLEW_GET_FUN(__clewEnqueueCopyImageToBuffer      )
-#define	clEnqueueCopyBufferToImage      CLEW_GET_FUN(__clewEnqueueCopyBufferToImage      )
-#define	clEnqueueMapBuffer              CLEW_GET_FUN(__clewEnqueueMapBuffer              )
-#define	clEnqueueMapImage               CLEW_GET_FUN(__clewEnqueueMapImage               )
-#define	clEnqueueUnmapMemObject         CLEW_GET_FUN(__clewEnqueueUnmapMemObject         )
-#define	clEnqueueNDRangeKernel          CLEW_GET_FUN(__clewEnqueueNDRangeKernel          )
-#define	clEnqueueTask                   CLEW_GET_FUN(__clewEnqueueTask                   )
-#define	clEnqueueNativeKernel           CLEW_GET_FUN(__clewEnqueueNativeKernel           )
-#define	clEnqueueMarker                 CLEW_GET_FUN(__clewEnqueueMarker                 )
-#define	clEnqueueWaitForEvents          CLEW_GET_FUN(__clewEnqueueWaitForEvents          )
-#define	clEnqueueBarrier                CLEW_GET_FUN(__clewEnqueueBarrier                )
-#define	clGetExtensionFunctionAddress   CLEW_GET_FUN(__clewGetExtensionFunctionAddress   )
+#define clCreateBuffer CLEW_GET_FUN(__clewCreateBuffer)
+#define clCreateSubBuffer CLEW_GET_FUN(__clewCreateSubBuffer)
+#define clCreateImage2D CLEW_GET_FUN(__clewCreateImage2D)
+#define clCreateImage3D CLEW_GET_FUN(__clewCreateImage3D)
+#define clRetainMemObject CLEW_GET_FUN(__clewRetainMemObject)
+#define clReleaseMemObject CLEW_GET_FUN(__clewReleaseMemObject)
+#define clGetSupportedImageFormats CLEW_GET_FUN(__clewGetSupportedImageFormats)
+#define clGetMemObjectInfo CLEW_GET_FUN(__clewGetMemObjectInfo)
+#define clGetImageInfo CLEW_GET_FUN(__clewGetImageInfo)
+#define clSetMemObjectDestructorCallback CLEW_GET_FUN(__clewSetMemObjectDestructorCallback)
+#define clCreateSampler CLEW_GET_FUN(__clewCreateSampler)
+#define clRetainSampler CLEW_GET_FUN(__clewRetainSampler)
+#define clReleaseSampler CLEW_GET_FUN(__clewReleaseSampler)
+#define clGetSamplerInfo CLEW_GET_FUN(__clewGetSamplerInfo)
+#define clCreateProgramWithSource CLEW_GET_FUN(__clewCreateProgramWithSource)
+#define clCreateProgramWithBinary CLEW_GET_FUN(__clewCreateProgramWithBinary)
+#define clRetainProgram CLEW_GET_FUN(__clewRetainProgram)
+#define clReleaseProgram CLEW_GET_FUN(__clewReleaseProgram)
+#define clBuildProgram CLEW_GET_FUN(__clewBuildProgram)
+#define clUnloadCompiler CLEW_GET_FUN(__clewUnloadCompiler)
+#define clGetProgramInfo CLEW_GET_FUN(__clewGetProgramInfo)
+#define clGetProgramBuildInfo CLEW_GET_FUN(__clewGetProgramBuildInfo)
+#define clCreateKernel CLEW_GET_FUN(__clewCreateKernel)
+#define clCreateKernelsInProgram CLEW_GET_FUN(__clewCreateKernelsInProgram)
+#define clRetainKernel CLEW_GET_FUN(__clewRetainKernel)
+#define clReleaseKernel CLEW_GET_FUN(__clewReleaseKernel)
+#define clSetKernelArg CLEW_GET_FUN(__clewSetKernelArg)
+#define clGetKernelInfo CLEW_GET_FUN(__clewGetKernelInfo)
+#define clGetKernelWorkGroupInfo CLEW_GET_FUN(__clewGetKernelWorkGroupInfo)
+#define clWaitForEvents CLEW_GET_FUN(__clewWaitForEvents)
+#define clGetEventInfo CLEW_GET_FUN(__clewGetEventInfo)
+#define clCreateUserEvent CLEW_GET_FUN(__clewCreateUserEvent)
+#define clRetainEvent CLEW_GET_FUN(__clewRetainEvent)
+#define clReleaseEvent CLEW_GET_FUN(__clewReleaseEvent)
+#define clSetUserEventStatus CLEW_GET_FUN(__clewSetUserEventStatus)
+#define clSetEventCallback CLEW_GET_FUN(__clewSetEventCallback)
+#define clGetEventProfilingInfo CLEW_GET_FUN(__clewGetEventProfilingInfo)
+#define clFlush CLEW_GET_FUN(__clewFlush)
+#define clFinish CLEW_GET_FUN(__clewFinish)
+#define clEnqueueReadBuffer CLEW_GET_FUN(__clewEnqueueReadBuffer)
+#define clEnqueueReadBufferRect CLEW_GET_FUN(__clewEnqueueReadBufferRect)
+#define clEnqueueWriteBuffer CLEW_GET_FUN(__clewEnqueueWriteBuffer)
+#define clEnqueueWriteBufferRect CLEW_GET_FUN(__clewEnqueueWriteBufferRect)
+#define clEnqueueCopyBuffer CLEW_GET_FUN(__clewEnqueueCopyBuffer)
+#define clEnqueueCopyBufferRect CLEW_GET_FUN(__clewEnqueueCopyBufferRect)
+#define clEnqueueReadImage CLEW_GET_FUN(__clewEnqueueReadImage)
+#define clEnqueueWriteImage CLEW_GET_FUN(__clewEnqueueWriteImage)
+#define clEnqueueCopyImage CLEW_GET_FUN(__clewEnqueueCopyImage)
+#define clEnqueueCopyImageToBuffer CLEW_GET_FUN(__clewEnqueueCopyImageToBuffer)
+#define clEnqueueCopyBufferToImage CLEW_GET_FUN(__clewEnqueueCopyBufferToImage)
+#define clEnqueueMapBuffer CLEW_GET_FUN(__clewEnqueueMapBuffer)
+#define clEnqueueMapImage CLEW_GET_FUN(__clewEnqueueMapImage)
+#define clEnqueueUnmapMemObject CLEW_GET_FUN(__clewEnqueueUnmapMemObject)
+#define clEnqueueNDRangeKernel CLEW_GET_FUN(__clewEnqueueNDRangeKernel)
+#define clEnqueueTask CLEW_GET_FUN(__clewEnqueueTask)
+#define clEnqueueNativeKernel CLEW_GET_FUN(__clewEnqueueNativeKernel)
+#define clEnqueueMarker CLEW_GET_FUN(__clewEnqueueMarker)
+#define clEnqueueWaitForEvents CLEW_GET_FUN(__clewEnqueueWaitForEvents)
+#define clEnqueueBarrier CLEW_GET_FUN(__clewEnqueueBarrier)
+#define clGetExtensionFunctionAddress CLEW_GET_FUN(__clewGetExtensionFunctionAddress)
 
+#define CLEW_SUCCESS 0               //!<    Success error code
+#define CLEW_ERROR_OPEN_FAILED -1    //!<    Error code for failing to open the dynamic library
+#define CLEW_ERROR_ATEXIT_FAILED -2  //!<    Error code for failing to queue the closing of the dynamic library to atexit()
 
-#define CLEW_SUCCESS                0       //!<    Success error code
-#define CLEW_ERROR_OPEN_FAILED      -1      //!<    Error code for failing to open the dynamic library
-#define CLEW_ERROR_ATEXIT_FAILED    -2      //!<    Error code for failing to queue the closing of the dynamic library to atexit()
+	//! \brief Load OpenCL dynamic library and set function entry points
+	int clewInit(const char *);
 
-//! \brief Load OpenCL dynamic library and set function entry points
-int         clewInit        (const char*);
+	//! \brief Exit clew and unload OpenCL dynamic library
+	void clewExit();
 
-//! \brief Exit clew and unload OpenCL dynamic library
-void		clewExit();
-
-//! \brief Convert an OpenCL error code to its string equivalent
-const char* clewErrorString (cl_int error);
+	//! \brief Convert an OpenCL error code to its string equivalent
+	const char *clewErrorString(cl_int error);
 
 #ifdef __cplusplus
 }